From 345191d8b592e9a65c0febde09569c34fd3b66e8 Mon Sep 17 00:00:00 2001
From: Packit <packit>
Date: Sep 10 2020 08:54:56 +0000
Subject: memkind-1.10.0 base


---

diff --git a/.codecov.yml b/.codecov.yml
new file mode 100644
index 0000000..d42f217
--- /dev/null
+++ b/.codecov.yml
@@ -0,0 +1,10 @@
+ignore:
+  - "autohbw/.*"
+  - "examples/autohbw*"
+  - "examples/*example*"
+  - "examples/memkind*"
+  - "test/.*"
+comment:
+  layout: "diff"
+  behaviour: default
+  require_changes: true
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..b581c84
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,126 @@
+*.a
+*.la
+*.lo
+*.o
+*.so
+.cproject
+.deps/
+.dirstamp
+.libs/
+.project
+/Makefile
+/Makefile.in
+/VERSION
+/aclocal.m4
+/ar-lib
+/compile
+/config.guess
+/config.h
+/config.h.in
+/config.log
+/config.status
+/config.sub
+/config_tls.h
+/configure
+/depcomp
+/examples/*.la
+/examples/*.lo
+/examples/*.o
+/examples/.deps/
+/examples/.dirstamp
+/examples/.libs/
+/examples/autohbw_candidates
+/examples/filter_memkind
+/examples/hello_hbw
+/examples/hello_memkind
+/examples/hello_memkind_debug
+/examples/memkind_allocated
+/examples/memkind_cpp_allocator
+/examples/memkind_get_stat
+/examples/pmem_alignment
+/examples/pmem_and_dax_kmem_kind
+/examples/pmem_and_default_kind
+/examples/pmem_config
+/examples/pmem_cpp_allocator
+/examples/pmem_detect_kind
+/examples/pmem_free_with_unknown_kind
+/examples/pmem_kinds
+/examples/pmem_malloc
+/examples/pmem_malloc_unlimited
+/examples/pmem_multithreads
+/examples/pmem_multithreads_onekind
+/examples/pmem_usable_size
+/install-sh
+/jemalloc/configure
+/jemalloc/obj/
+/libtool
+/ltmain.sh
+/m4/
+/memkind-*.spec
+/memkind-*.tar.gz
+/memkind-auto-dax-kmem-nodes
+/memkind-hbw-nodes
+/missing
+/src/memkind_defines.h.in
+/stamp-h1
+/stamp-h2
+/test-driver
+/test-suite.log
+/test/*.o
+/test/*.pyc
+/test/*.so
+/test/.deps/
+/test/.dirstamp
+/test/.libs/
+/test/all_tests
+/test/alloc_benchmark_glibc
+/test/alloc_benchmark_hbw
+/test/alloc_benchmark_pmem
+/test/alloc_benchmark_tbb
+/test/allocator_perf_tool_tests
+/test/autohbw_candidates
+/test/autohbw_test_helper
+/test/check.sh.log
+/test/check.sh.trs
+/test/dax_kmem_test
+/test/decorator_test
+/test/environ_err_dax_kmem_malloc_positive_test
+/test/environ_err_dax_kmem_malloc_test
+/test/environ_err_hbw_malloc_test
+/test/environerr_hbw_malloc_test
+/test/environerr_test
+/test/filter_memkind
+/test/fragmentation_benchmark_pmem
+/test/freeing_memory_segfault_test
+/test/gb_page_tests_bind_policy
+/test/hello_hbw
+/test/hello_memkind
+/test/hello_memkind_debug
+/test/locality_test
+/test/mallocerr_test
+/test/memkind_allocated
+/test/memkind_cpp_allocator
+/test/memkind_get_stat
+/test/memkind_stat_test
+/test/perf_tool
+/test/performance_test
+/test/performance/.deps/
+/test/pmem_alignment
+/test/pmem_and_dax_kmem_kind
+/test/pmem_and_default_kind
+/test/pmem_config
+/test/pmem_cpp_allocator
+/test/pmem_detect_kind
+/test/pmem_free_with_unknown_kind
+/test/pmem_kinds
+/test/pmem_malloc
+/test/pmem_malloc_unlimited
+/test/pmem_multithreads
+/test/pmem_multithreads_onekind
+/test/pmem_test
+/test/pmem_usable_size
+/test/trace_mechanism_test_helper
+/test/defrag_reallocate
+/vs
+autom4te.cache/
+memkind.pc
diff --git a/.travis.yml b/.travis.yml
new file mode 100644
index 0000000..3899848
--- /dev/null
+++ b/.travis.yml
@@ -0,0 +1,65 @@
+dist: bionic
+sudo: required
+language: cpp
+cache: ccache
+git:
+  depth: false
+matrix:
+  fast_finish: true
+  include:
+
+    - name: "Astyle"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+          packages:
+            - astyle
+      script:
+        - ./astyle.sh
+
+    - name: "GCC 4.8"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+          packages:
+            - g++-4.8
+            - libnuma-dev
+      env:
+         - MATRIX_EVAL="CC=gcc-4.8 && CXX=g++-4.8"
+      install:
+      script:
+        - ./build.sh
+
+    - name: "GCC 8.0"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+          packages:
+            - g++-8
+            - libnuma-dev
+      env:
+         - MATRIX_EVAL="CC=gcc-8 && CXX=g++-8"
+      install:
+      script:
+        - ./build.sh
+
+    - name: "Clang 8"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+            - llvm-toolchain-bionic-8
+          packages:
+            - clang-8
+            - libnuma-dev
+      env:
+        - MATRIX_EVAL="CC=clang-8 && CXX=clang++-8"
+      install:
+      script:
+        - ./build.sh
+
+before_install:
+  - eval "${MATRIX_EVAL}"
diff --git a/AUTHORS b/AUTHORS
new file mode 100644
index 0000000..ea4ca98
--- /dev/null
+++ b/AUTHORS
@@ -0,0 +1,26 @@
+Christopher Cantalupo <christopher.m.cantalupo@intel.com>
+Vishwanath Venkatesan <vishwanath.venkatesan@intel.com>
+Steven Hampson <steven.t.hampson@intel.com>
+Krzysztof Czurylo <krzysztof.czurylo@intel.com>
+Leobardo Rountree <leobardo.r.rountree@intel.com>
+Hector Barajas Villalobos <hector.a.barajas.villalobos@intel.com>
+Ruchira Sasanka <ruchira.sasanka@intel.com>
+Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>
+Rafael Aquini <aquini@redhat.com>
+Alexandr Konovalov <alexandr.konovalov@intel.com>
+Grzegorz Ozanski <grzegorz.ozanski@intel.com>
+Artur Koziej <artur.koziej@intel.com>
+Jakub Dlugolecki <jakub.dlugolecki@intel.com>
+Lukasz Anaczkowski <lukasz.anaczkowski@intel.com>
+Agata Wozniak <agata.wozniak@intel.com>
+Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
+Kamil Patelczyk <kamil.patelczyk@intel.com>
+Jeff Hammond <jeff.r.hammond@intel.com>
+Aleksandra Hernik <aleksandra.hernik@intel.com>
+Pawel Kochanek <pawel.b.kochanek@intel.com>
+Katarzyna Wasiuta <katarzyna.wasiuta@intel.com>
+Krzysztof Filipek <krzysztof.filipek@intel.com>
+Michal Biesek <michal.biesek@intel.com>
+Adam Borowski <adam.borowski@intel.com>
+Yuan Zhou <yuan.zhou@intel.com>
+Harald Servat <harald.servat@intel.com>
diff --git a/CONTRIBUTING b/CONTRIBUTING
new file mode 100644
index 0000000..ca0039b
--- /dev/null
+++ b/CONTRIBUTING
@@ -0,0 +1,134 @@
+CONTRIBUTING
+============
+
+This file is intended to help those interested in contributing to the
+memkind library.
+
+
+COMMUNICATION
+=============
+
+Please participate in the memkind mailing list:
+
+    https://lists.01.org/mailman/listinfo/memkind
+
+There is also the option of opening an issue through github:
+
+    https://github.com/memkind/memkind/issues
+
+The mailing list is intended for discussion and the issues are useful
+for tracking progress on tasks.  The TODO file lists out a number of
+topics, and in order to prioritize one of them please open a github
+issue with a related subject.
+
+
+TESTING
+=======
+
+The tests require a Linux kernel newer than 3.11 (the details are
+documented in the memkind README), and the reservation of 3000 huge
+pages.  The huge pages can be reserved with the following command:
+
+    $ sudo echo 3000 > /proc/sys/vm/nr_hugepages
+
+Only in the case where gigabyte pages have been reserved will the
+tests associated with gigabyte pages be executed.  Reserving gigabyte
+pages may require a modification to the kernel command line unless the
+kernel is quite recent.
+
+To test memkind simply execute the "make check" target after building.
+This target calls memkind/test/test.sh with parameters
+depending on the environment.
+
+Most of the tests are written within the gtest framework, however, as
+part of testing the example programs are also executed and the return
+code of the executable determines pass or fail.  The autotools test
+infrastructure is used as a high level executor, but it does not track
+individual tests.  The consequence of this is that the autotools
+output records only one high level test which passes in the case where
+every underlying test was successful and fails if any underlying test
+fails.  The individual test results are recorded in the directory
+called "gtest_output." Here you will find the log of the tests in
+gtest_output/test.out and a set of junit style xml results: one for
+each test.  Note that a side effect of having only one autotools test
+is that autotools parallel testing is disabled.  We have
+multi-threaded tests that use the OpenMP run-time which enables more
+purposeful and deterministic testing of threading issues.  Note that
+the OpenMP run-time is only required for testing, it is not used by
+the memkind library internally.
+
+
+CODING STYLE
+============
+
+Before submitting a patch for inclusion, please run the modified
+source code files through astyle with the following options
+
+    $ astyle --style=linux --indent=spaces=4 -S --max-continuation-indent=80 \
+             --max-code-length=80 --break-after-logical --indent-namespaces -z2 \
+             --align-pointer=name
+
+More information about astyle can be found here:
+
+    http://astyle.sourceforge.net/
+
+In C, source code constants are in all caps and everything else is in
+all lower case.  Underscores are used to separate words within a
+symbol name. No camel case shall be used in C code.  The test code is
+largely written in C++.  Here camel-case should be used for class
+names and should always have a capitalized first letter.  Other
+symbols in C++ should generally follow the C style.
+
+Most symbols with global scope shall be prefixed with "memkind_" or
+"hbw_" depending on which interface they are a part of.
+
+Any global variable shall have _g appended to the variable name and in
+most cases shall be statically scoped within a single compilation
+unit.  The exception to that rule are static memory kinds that are
+declared as extern within the associated interface header and are
+constant after initialization.  Global variables should be used in a
+very narrow set of circumstances and in most cases modifications
+should be guarded with pthread_once(3).
+
+Functions not used outside of the compilation unit shall be declared
+as static.  All functions which are not declared as static shall be
+documented in a man page and have an associated interface header file.
+
+Preprocessor mark-up is discouraged when other options are possible.
+Please use enum in place of #define when value control at configure or
+build time is not required.
+
+
+TESTS
+=====
+
+The current state of the tests is not nearly as well organized as it
+could be.  That being said, it is quite simple to add a new test.
+Most C++ files in the test directory are associated with a single
+gtest testing::Test class.  These classes usually have several
+associated test fixtures in the same file.  If a new test can be added
+as a fixture to an existing class, simply add the fixture to the file
+and the test will be incorporated into the test infrastructure.
+If a new class is required, create a new file and add it to the list
+of "test_all_tests_SOURCES" in memkind/test/Makfile.mk and it will
+be incorporated into the test infrastructure.
+
+There are a few files which define classes which are not google test
+classes.  These are check.cpp, trial_generator.cpp and main.cpp.  The
+check.cpp file defines a class Check that can be used to validate
+fundamental memkind features like NUMA node location, and page size.
+The trial_generator.cpp file is used to abstract a sequence of
+allocation and deallocation calls while performing checks on the
+results of each call; this can be used to apply similar tests to all
+of the different allocation APIs.  The main.cpp file is a simple
+wrapper around testing::InitGoogleTest and RUN_ALL_TESTS().
+
+
+SUBMITTING A PATCH
+==================
+
+Please be sure that all tests pass before submission and that the
+style conforms to the specifications given here.  If a new feature is
+implemented in the patch, please also include unit tests and an
+example which exercises this feature.  Once these requirements have
+been met, please submit a pull request through github.
diff --git a/COPYING b/COPYING
new file mode 100644
index 0000000..cc1c9cc
--- /dev/null
+++ b/COPYING
@@ -0,0 +1,25 @@
+Unless otherwise specified, files in the memkind source distribution are
+subject to the following license:
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+      this list of conditions and the following disclaimer.
+    * Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+    * Neither the name of Intel Corporation nor the names of its contributors
+      may be used to endorse or promote products derived from this software
+      without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/ChangeLog b/ChangeLog
new file mode 100644
index 0000000..d887e18
--- /dev/null
+++ b/ChangeLog
@@ -0,0 +1,242 @@
+* Mon Dec 16 2019 Michal Biesek <michal.biesek@intel.com> v1.10.0
+- Provided another way to use Persistent Memory in memkind (MEMKIND_DAX_KMEM_* kinds)
+- Added C++ allocator for static kinds (including MEMKIND_DAX_KMEM_* kinds)
+- Provided an interface to get memkind allocation statistics
+- Provided a method to reduce fragmentation - memkind_defrag_reallocate()
+- Added support for background thread
+- Updated internally used jemalloc to upstream version 5.2.1
+- Extended hbw interface with hbw_malloc_usable_size()
+- Simplified building process
+
+* Wed Apr 3 2019 Michal Biesek <michal.biesek@intel.com> v1.9.0
+- Provided a second function to create PMEM kind memkind_create_pmem_with_config()
+- Provided the memkind_detect_kind() function to recognize a kind from memory allocation
+- Provided memkind_config_set_memory_usage_policy() function
+  that modifies the default memory usage policy behavior
+- Provided the ability to pass a NULL kind to memkind_malloc_usable_size() and memkind_realloc()
+- Unified the memkind_realloc() behavior for all kinds
+- Added the support for the pool_msize() function to TBB
+- Extended build arch to ppc64, ppc64le, s390x
+- Removed disable-stats from used jemalloc
+- Fixed compilation under the Clang and rpm build configuration
+- Provided the support for Dockerfile
+
+* Mon Dec 10 2018 Michal Biesek <michal.biesek@intel.com> v1.8.0
+- Fixed error with realloc/free method with passing thread-specific cache flag
+- Fixed error with memkind_create_pmem(), if other PMEM kind was destroyed before
+- Fixed error with zeroing large allocations in PMEM kind
+- Added support to create kind without maximum size limit of PMEM kind (max_size=0)
+- Extended memkind API with memkind_malloc_usable_size()
+- Removed EXPERIMENTAL from most methods in memkind API
+- Added MEMKIND_ERROR_ARENAS_CREATE code
+- Added C++ allocator for PMEM kind
+- Extended PMEM examples
+- Fixed integration with Travis CI
+- Extended Travis CI configuration with Astyle and Coverage
+- Added PMEM kind tests
+
+* Wed Jan 24 2018 Pawel Kochanek <pawel.b.kochanek@intel.com> v1.7.0
+- Updated internally used jemalloc to upstream version 5.0.
+- Fixed error that has been occurring while memkind was dynamically loaded.
+- Fixed MEMKIND_HBW_NODES behavior for single NUMA node system.
+- Removed licenses other than BSD 3-clause from COPYING.
+- Changed build instructions.
+- Added configurable jemalloc prefix in build scripts.
+- Upgraded gtest to version 1.8.0.
+- Added memory footprint tests.
+- Added locality test for MEMKIND_PREFERRED.
+- Applied test parametrization in BATests.
+- Fixed problems with pytest testing in Travis.
+- Added huge page configuration in several tests.
+- Removed several symbols that are no longer exposed in public API from man pages.
+- Fixed HBW_POLICY_BIND_ALL documentation.
+
+* Tue Jun 13 2017 Artur Koziej <artur.koziej@intel.com> v1.6.0
+- Introduced new policy HBW_POLICY_BIND_ALL.
+- Introduced MEMKIND_HBW_ALL, MEMKIND_HBW_ALL_HUGETLB and MEMKIND_REGULAR kinds.
+- Fixed hbw_posix_memalign_psize() return value in case of unsupported policy and
+  page size combination.
+- Documented supported policy and page size combination for
+  hbw_posix_memalign_psize().
+- Merged libmemkind.a and libjemalloc_pic.a into a single static library.
+- Fixed OOM during tcache acquisition.
+- Tuned huge pages configuration in memkind tests.
+- Added stress tests for allocations with large random sizes.
+- Added stress tests for allocations with random kinds.
+- Extended policy and NUMA node binding tests.
+- Fixed several Travis build problems.
+- Removed gb_page_tests_preferred_policy tests.
+
+* Tue Feb 21 2017 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.5.0
+- Intel(R) TBB scalable_allocator can be used as heap management alternative
+  to jemalloc (requires installed Intel (R) TBB package and environment variable
+  MEMKIND_HEAP_MANAGER=TBB)
+- Updated internally used jemalloc to upstream version 4.3.1 (previously modified 3.5.1)
+- Introduced new environment option MEMKIND_HOG_MEMORY. Setting it to "1" will prevent
+  memkind from releasing memory to OS.
+- Introduced support for KNM processor from Intel(R) Xeon Phi(TM) family
+- General cleanup of build scripts (e.g. fixed issues found on Ubuntu,
+  introduced simple build.sh script)
+- Deprecated support for gigabyte-pages
+- Removed number of previously deprecated symbols
+
+* Thu Nov 17 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.4.0
+- Introduced hbw_verify_memory_region() for verifying if provided memory
+  region fully falls into high-bandwidth memory (for details please take a look
+  at man/hbwmalloc.3).
+- New API for creating kinds using set of predefined attributes:
+  memkind_create_kind(), memkind_destroy_kind() (for details please take a look
+  at man/memkind.3).
+- Fixed issue in autohbw where it was reporting limit error regardless of
+  value provided.
+- Added logging on error paths.
+- Test base fixes and improvements.
+
+* Tue Sep 27 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.3.0
+- Introduced logging mechanism (for details please take a look at
+  man/memkind.3).
+- Deprecated symbols: memkind_finalize(), memkind_get_num_kind(),
+  memkind_get_kind_by_partition(), memkind_get_kind_by_name(),
+  memkind_partition_mmap(), memkind_get_size(), MEMKIND_ERROR_MEMALIGN,
+  MEMKIND_ERROR_MALLCTL,  MEMKIND_ERROR_GETCPU, MEMKIND_ERROR_PMTT,
+  MEMKIND_ERROR_TIEDISTANCE, MEMKIND_ERROR_ALIGNMENT, MEMKIND_ERROR_MALLOCX,
+  MEMKIND_ERROR_REPNAME, MEMKIND_ERROR_PTHREAD, MEMKIND_ERROR_BADPOLICY,
+  MEMKIND_ERROR_REPPOLICY.
+- Added integration with hwloc (turned on with --with-hwloc).
+- Cleanup of symbols exposed by libmemkind.so (e.g. no longer exposing libnuma
+  and jemalloc symbols).
+- AutoHBW files have been moved to "memkind/autohbw" directory, code has
+  been refactored and tests have been added to appropriate scenarios.
+- Library is now built with flags improving security (can be turned off
+  with --disable-secure configure time option).
+- Changed configuration of jemalloc to turn off unused features.
+
+* Thu Aug 18 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.2.0
+- memkind_create() and memkind_ops have been deprecated (moved to
+  memkind_deprecated.h).
+- Deprecated the headers from memkind/internal and added compile-time
+  warnings to them (plan is to remove those from rpm packages).
+- autohbw files have been moved to memkind/autohbw directory.
+- API decorators (memkind_malloc_pre, memkind_malloc_post etc.) now
+  need to be enabled with configure-time option --enable-decorators.
+- Allocation time optimizations with up to 20% improvement.
+
+* Wed Jul 13 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.1.1
+- Performance improvement for memkind_free() in scenario where NULL
+  was passed as kind (reduced by 60%).
+- Introduced integration with Travis CI.
+- Fixed issue where memory returned from calloc was not filled with zeroes
+  when using memkind_pmem kinds.
+- Fixed issue where interleave kinds was failing on systems without
+  Transparent Huge Pages module configured.
+- Resolved several issues that was causing compilation errors on some systems.
+- test/test.sh is now able to detect system configuration, and run only those
+  tests which requirements are meet.
+- Added gtest to repo to avoid downloading it during build proecess.
+
+* Mon May 23 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.1.0
+- Implemented algorithm for detecting high-bandwidth memory on Intel Xeon Phi X200
+  without parsing PMTT.
+- PMTT-parsing service has been removed.
+- memkind struct definition from memkind.h has been replaced by forward
+  declaration. Definition has been moved to /internal/memkind_private.h and
+  shouldn't be considered as part of library interface anymore.
+- Performance improvement for hugetlb kinds (check_available() is no longer
+  parsing sysfs every time).
+- Performance fix by removing --enable-safe from jemmaloc build flags.
+- New tests for validating high-bandwidth memory detection has been added.
+- AFTS has been split into two groups: memkind-afts.ts and memkind-afts-ext.ts,
+  based on amount of required memory (for details view test/README).
+
+* Thu Mar 03 2016 Artur Koziej <artur.koziej@intel.com> v1.0.0
+- hbwmalloc.h released as stable API.
+- Introduce API standards: STANDARD API, NON-STANDARD API, EXPERIMENTAL API.
+- Introduce memkind versioning API.
+- Fix type names in hbwmalloc API.
+- Change error codes to POSIX standard ones in hbwmalloc API.
+- Move API headers to include directory.
+- Move NON-STANDARD API, or EXPERIMENTAL API to include/memkind/internal directory.
+- New man page (memkind_interleave.3) describing interleave kind.
+- Significant documentation improvement.
+- Performance fix for jemalloc - significantly decrease of library initialization time.
+- Enforce 2MB alignment in jemalloc due to the Linux kernel bug (munmap() fails for
+  huge pages, when the size is not aligned).
+- Move to systemd defined service (memkind.service), drop init.d script.
+- New rpm layout (memkind and memkind-devel).
+- Remove false dependency on OpenMP.
+- Extend test base for stress, longevity and initialization performance tests.
+- Fix memory leak in ctl_growk().
+- Rename namespace: hbwmalloc -> hbw.
+- Rename class: hbwmalloc_allocator -> allocator.
+
+* Thu Oct 15 2015 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v0.3.0
+- Added two new kinds: MEMKIND_INTERLEAVE and MEMKIND_HBW_INTERLEAVE
+  (for details please take a look at man/memkind.3)
+- Added support for file-backed memory heaps
+  (for details please take a look at man/memkind_pmem.3)
+- Added autohbw library, that intercepts the standard heap allocations,
+  which let use high-bandwidth memory without modifying existing codebase
+  (for details please take a look at examples/autohbw_README)
+- jemalloc is now, statically linked, internal component of memkind RPM
+  instead of rpm dependency.
+- Added memkind_allocated example which demonstrates usage of memkind
+  with alignas() alignment specifier introduced in C++11
+  (for details please take a look at examples/README)
+- Added support for using thread local storage to improve performance in
+  multithreaded applications (enabled by configure time option: --enable-tls)
+- Added possibility to set number of jemalloc arenas per kind by
+  configure time parameter or environmental variable
+  (for details please take a look at MEMKIND_ARENA_NUM_PER_KIND section
+  of man/memkind.3).
+- Added decorators to the memkind allocation APIs.
+  These are weak symbols (pre and post for each API) which can modify
+  the input and output of each of the calls.
+- Significantly extended test base (new groups of tests: performance,
+  multithreaded, PMTT, nagtive, stress).
+
+* Fri Jan 9 2015 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.2.0
+- Bumped memkind ABI version to 0:1:0.
+- Removed memkind_get_kind_for_free() from externally facing API's.  Instead if
+  memkind_free() is called with zero passed for the kind then
+  memkind_get_kind_for_free() is called internally.
+- Moved to single callback memkind_partition_mmap() to simplify
+  jemalloc modifications.
+- Added hooks for setting file descriptor and offset for mmap enabling
+  file-backed memory kinds.
+- Added a void pointer called "priv" to memkind structure for storing data for
+  user-defined kinds.
+- Removed call to sched_getcpu(), now thread id is hashed to determine the arena
+  index.
+- Added weak symbol hooks for decorating the heap management functions.
+- Fixed several issues with init.d/memkind script and spec file scriptlets that
+  are exposed by SLES-12.
+- Introduced an example library called numakind.
+  The numakind library will allocate from the closest NUMA node to a thread as
+  measured when that thread makes its first allocation call.
+- Fixed error handling in memkind_gbtlb_mmap() that could cause a segfault when
+  gigabyte pages are not available.
+- Added tests for PMTT parser.
+- Removed binary mock PMTT table from source code, replaced it with a hexdump.
+- Fixed a number of issues in test scripts which were suppressing errors.
+- Removed unnecessary includes from header files.
+- Better error checking in example code.
+- Documentation update and clean up.
+
+* Thu Nov 13 2014 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.1.0
+- Increased test code coverage significantly.
+- Fixed bug in memkind_error_message() for MEMKIND_ERROR_TOOMANY.
+- Removed memkind_arena_free() API since it was redundant with memkind_default_free().
+- Static memkind structs are now declared as extern in the headers and defined in the
+  source files files rather than being statically defined in the headers.
+
+* Thu Oct 30 2014 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.0.9
+- Now building with autotools.
+- Updated documentation.
+- Fixed typo in copyright.
+- Fixed test scripts to properly handle return code of each test.
+- Added C++03 standard allocator that uses hbw_malloc and hbw_free.
+* Tue Sep 30 2014 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.0.8
+- Added GBTLB functionality, code clean up, documentation updates,
+  examples directory.  Examples includes stream modified to use
+  memkind interface.  Code coverage still lacking, and documentation
+  incomplete.
diff --git a/MANIFEST b/MANIFEST
new file mode 100644
index 0000000..f5aa3e4
--- /dev/null
+++ b/MANIFEST
@@ -0,0 +1,552 @@
+.codecov.yml
+.travis.yml
+AUTHORS
+CONTRIBUTING
+COPYING
+ChangeLog
+Makefile.am
+NEWS
+PULL_REQUEST_TEMPLATE.md
+README
+astyle.sh
+autogen.sh
+autohbw/Makefile.mk
+autohbw/autohbw.c
+autohbw/autohbw_README
+autohbw/autohbw_get_src_lines.pl
+autohbw/autohbw_test.sh
+build.sh
+configure.ac
+examples/Makefile.mk
+examples/README
+examples/autohbw_candidates.c
+examples/filter_example.c
+examples/hello_hbw_example.c
+examples/hello_memkind_example.c
+examples/memkind_allocated.hpp
+examples/memkind_allocated_example.cpp
+examples/memkind_cpp_allocator.cpp
+examples/memkind_decorator_debug.c
+examples/memkind_get_stat.c
+examples/pmem_alignment.c
+examples/pmem_and_default_kind.c
+examples/pmem_and_dax_kmem.c
+examples/pmem_config.c
+examples/pmem_cpp_allocator.cpp
+examples/pmem_detect_kind.c
+examples/pmem_free_with_unknown_kind.c
+examples/pmem_kinds.c
+examples/pmem_malloc.c
+examples/pmem_malloc_unlimited.c
+examples/pmem_multithreads.c
+examples/pmem_multithreads_onekind.c
+examples/pmem_usable_size.c
+include/hbw_allocator.h
+include/hbwmalloc.h
+include/memkind.h
+include/memkind/internal/heap_manager.h
+include/memkind/internal/memkind_arena.h
+include/memkind/internal/memkind_bandwidth.h
+include/memkind/internal/memkind_dax_kmem.h
+include/memkind/internal/memkind_default.h
+include/memkind/internal/memkind_gbtlb.h
+include/memkind/internal/memkind_hbw.h
+include/memkind/internal/memkind_hugetlb.h
+include/memkind/internal/memkind_interleave.h
+include/memkind/internal/memkind_log.h
+include/memkind/internal/memkind_pmem.h
+include/memkind/internal/memkind_private.h
+include/memkind/internal/memkind_regular.h
+include/memkind/internal/tbb_mem_pool_policy.h
+include/memkind/internal/tbb_wrapper.h
+include/memkind/internal/vec.h
+include/memkind_allocator.h
+include/memkind_deprecated.h
+include/pmem_allocator.h
+install_astyle.sh
+jemalloc/.appveyor.yml
+jemalloc/.autom4te.cfg
+jemalloc/.gitattributes
+jemalloc/.gitignore
+jemalloc/.travis.yml
+jemalloc/autogen.sh
+jemalloc/bin/jemalloc-config.in
+jemalloc/bin/jemalloc.sh.in
+jemalloc/bin/jeprof.in
+jemalloc/build-aux/config.guess
+jemalloc/build-aux/config.sub
+jemalloc/build-aux/install-sh
+jemalloc/ChangeLog
+jemalloc/config.stamp.in
+jemalloc/configure.ac
+jemalloc/COPYING
+jemalloc/doc/html.xsl.in
+jemalloc/doc/jemalloc.xml.in
+jemalloc/doc/manpages.xsl.in
+jemalloc/doc/stylesheet.xsl
+jemalloc/include/jemalloc/internal/arena_externs.h
+jemalloc/include/jemalloc/internal/arena_inlines_a.h
+jemalloc/include/jemalloc/internal/arena_inlines_b.h
+jemalloc/include/jemalloc/internal/arena_stats.h
+jemalloc/include/jemalloc/internal/arena_structs_a.h
+jemalloc/include/jemalloc/internal/arena_structs_b.h
+jemalloc/include/jemalloc/internal/arena_types.h
+jemalloc/include/jemalloc/internal/assert.h
+jemalloc/include/jemalloc/internal/atomic.h
+jemalloc/include/jemalloc/internal/atomic_c11.h
+jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
+jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
+jemalloc/include/jemalloc/internal/atomic_msvc.h
+jemalloc/include/jemalloc/internal/background_thread_externs.h
+jemalloc/include/jemalloc/internal/background_thread_inlines.h
+jemalloc/include/jemalloc/internal/background_thread_structs.h
+jemalloc/include/jemalloc/internal/base_externs.h
+jemalloc/include/jemalloc/internal/base_inlines.h
+jemalloc/include/jemalloc/internal/base_structs.h
+jemalloc/include/jemalloc/internal/base_types.h
+jemalloc/include/jemalloc/internal/bin.h
+jemalloc/include/jemalloc/internal/bin_stats.h
+jemalloc/include/jemalloc/internal/bin_types.h
+jemalloc/include/jemalloc/internal/bitmap.h
+jemalloc/include/jemalloc/internal/bit_util.h
+jemalloc/include/jemalloc/internal/cache_bin.h
+jemalloc/include/jemalloc/internal/ckh.h
+jemalloc/include/jemalloc/internal/ctl.h
+jemalloc/include/jemalloc/internal/div.h
+jemalloc/include/jemalloc/internal/emitter.h
+jemalloc/include/jemalloc/internal/extent_dss.h
+jemalloc/include/jemalloc/internal/extent_externs.h
+jemalloc/include/jemalloc/internal/extent_inlines.h
+jemalloc/include/jemalloc/internal/extent_mmap.h
+jemalloc/include/jemalloc/internal/extent_structs.h
+jemalloc/include/jemalloc/internal/extent_types.h
+jemalloc/include/jemalloc/internal/hash.h
+jemalloc/include/jemalloc/internal/hook.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
+jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
+jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
+jemalloc/include/jemalloc/internal/large_externs.h
+jemalloc/include/jemalloc/internal/log.h
+jemalloc/include/jemalloc/internal/malloc_io.h
+jemalloc/include/jemalloc/internal/mutex.h
+jemalloc/include/jemalloc/internal/mutex_pool.h
+jemalloc/include/jemalloc/internal/mutex_prof.h
+jemalloc/include/jemalloc/internal/nstime.h
+jemalloc/include/jemalloc/internal/pages.h
+jemalloc/include/jemalloc/internal/ph.h
+jemalloc/include/jemalloc/internal/private_namespace.sh
+jemalloc/include/jemalloc/internal/private_symbols.sh
+jemalloc/include/jemalloc/internal/prng.h
+jemalloc/include/jemalloc/internal/prof_externs.h
+jemalloc/include/jemalloc/internal/prof_inlines_a.h
+jemalloc/include/jemalloc/internal/prof_inlines_b.h
+jemalloc/include/jemalloc/internal/prof_structs.h
+jemalloc/include/jemalloc/internal/prof_types.h
+jemalloc/include/jemalloc/internal/public_namespace.sh
+jemalloc/include/jemalloc/internal/public_unnamespace.sh
+jemalloc/include/jemalloc/internal/ql.h
+jemalloc/include/jemalloc/internal/qr.h
+jemalloc/include/jemalloc/internal/quantum.h
+jemalloc/include/jemalloc/internal/rb.h
+jemalloc/include/jemalloc/internal/rtree.h
+jemalloc/include/jemalloc/internal/rtree_tsd.h
+jemalloc/include/jemalloc/internal/safety_check.h
+jemalloc/include/jemalloc/internal/sc.h
+jemalloc/include/jemalloc/internal/seq.h
+jemalloc/include/jemalloc/internal/smoothstep.h
+jemalloc/include/jemalloc/internal/smoothstep.sh
+jemalloc/include/jemalloc/internal/spin.h
+jemalloc/include/jemalloc/internal/stats.h
+jemalloc/include/jemalloc/internal/sz.h
+jemalloc/include/jemalloc/internal/tcache_externs.h
+jemalloc/include/jemalloc/internal/tcache_inlines.h
+jemalloc/include/jemalloc/internal/tcache_structs.h
+jemalloc/include/jemalloc/internal/tcache_types.h
+jemalloc/include/jemalloc/internal/test_hooks.h
+jemalloc/include/jemalloc/internal/ticker.h
+jemalloc/include/jemalloc/internal/tsd.h
+jemalloc/include/jemalloc/internal/tsd_generic.h
+jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
+jemalloc/include/jemalloc/internal/tsd_tls.h
+jemalloc/include/jemalloc/internal/tsd_types.h
+jemalloc/include/jemalloc/internal/tsd_win.h
+jemalloc/include/jemalloc/internal/util.h
+jemalloc/include/jemalloc/internal/witness.h
+jemalloc/include/jemalloc/jemalloc.sh
+jemalloc/include/jemalloc/jemalloc_defs.h.in
+jemalloc/include/jemalloc/jemalloc_macros.h.in
+jemalloc/include/jemalloc/jemalloc_mangle.sh
+jemalloc/include/jemalloc/jemalloc_protos.h.in
+jemalloc/include/jemalloc/jemalloc_rename.sh
+jemalloc/include/jemalloc/jemalloc_typedefs.h.in
+jemalloc/include/msvc_compat/C99/stdbool.h
+jemalloc/include/msvc_compat/C99/stdint.h
+jemalloc/include/msvc_compat/strings.h
+jemalloc/include/msvc_compat/windows_extra.h
+jemalloc/INSTALL.md
+jemalloc/jemalloc.pc.in
+jemalloc/m4/ax_cxx_compile_stdcxx.m4
+jemalloc/Makefile.in
+jemalloc/msvc/jemalloc_vc2015.sln
+jemalloc/msvc/jemalloc_vc2017.sln
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/ReadMe.txt
+jemalloc/msvc/test_threads/test_threads.cpp
+jemalloc/msvc/test_threads/test_threads.h
+jemalloc/msvc/test_threads/test_threads_main.cpp
+jemalloc/README
+jemalloc/run_tests.sh
+jemalloc/scripts/gen_run_tests.py
+jemalloc/scripts/gen_travis.py
+jemalloc/src/arena.c
+jemalloc/src/background_thread.c
+jemalloc/src/base.c
+jemalloc/src/bin.c
+jemalloc/src/bitmap.c
+jemalloc/src/ckh.c
+jemalloc/src/ctl.c
+jemalloc/src/div.c
+jemalloc/src/extent.c
+jemalloc/src/extent_dss.c
+jemalloc/src/extent_mmap.c
+jemalloc/src/hash.c
+jemalloc/src/hook.c
+jemalloc/src/jemalloc.c
+jemalloc/src/jemalloc_cpp.cpp
+jemalloc/src/large.c
+jemalloc/src/log.c
+jemalloc/src/malloc_io.c
+jemalloc/src/mutex.c
+jemalloc/src/mutex_pool.c
+jemalloc/src/nstime.c
+jemalloc/src/pages.c
+jemalloc/src/prng.c
+jemalloc/src/prof.c
+jemalloc/src/rtree.c
+jemalloc/src/safety_check.c
+jemalloc/src/sc.c
+jemalloc/src/stats.c
+jemalloc/src/sz.c
+jemalloc/src/tcache.c
+jemalloc/src/test_hooks.c
+jemalloc/src/ticker.c
+jemalloc/src/tsd.c
+jemalloc/src/witness.c
+jemalloc/src/zone.c
+jemalloc/test/include/test/btalloc.h
+jemalloc/test/include/test/extent_hooks.h
+jemalloc/test/include/test/jemalloc_test.h.in
+jemalloc/test/include/test/jemalloc_test_defs.h.in
+jemalloc/test/include/test/math.h
+jemalloc/test/include/test/mq.h
+jemalloc/test/include/test/mtx.h
+jemalloc/test/include/test/SFMT-alti.h
+jemalloc/test/include/test/SFMT-params.h
+jemalloc/test/include/test/SFMT-params607.h
+jemalloc/test/include/test/SFMT-params1279.h
+jemalloc/test/include/test/SFMT-params2281.h
+jemalloc/test/include/test/SFMT-params4253.h
+jemalloc/test/include/test/SFMT-params11213.h
+jemalloc/test/include/test/SFMT-params19937.h
+jemalloc/test/include/test/SFMT-params44497.h
+jemalloc/test/include/test/SFMT-params86243.h
+jemalloc/test/include/test/SFMT-params132049.h
+jemalloc/test/include/test/SFMT-params216091.h
+jemalloc/test/include/test/SFMT-sse2.h
+jemalloc/test/include/test/SFMT.h
+jemalloc/test/include/test/test.h
+jemalloc/test/include/test/thd.h
+jemalloc/test/include/test/timer.h
+jemalloc/test/integration/aligned_alloc.c
+jemalloc/test/integration/allocated.c
+jemalloc/test/integration/extent.c
+jemalloc/test/integration/extent.sh
+jemalloc/test/integration/malloc.c
+jemalloc/test/integration/mallocx.c
+jemalloc/test/integration/mallocx.sh
+jemalloc/test/integration/MALLOCX_ARENA.c
+jemalloc/test/integration/overflow.c
+jemalloc/test/integration/posix_memalign.c
+jemalloc/test/integration/rallocx.c
+jemalloc/test/integration/sdallocx.c
+jemalloc/test/integration/slab_sizes.c
+jemalloc/test/integration/slab_sizes.sh
+jemalloc/test/integration/smallocx.c
+jemalloc/test/integration/smallocx.sh
+jemalloc/test/integration/thread_arena.c
+jemalloc/test/integration/thread_tcache_enabled.c
+jemalloc/test/integration/xallocx.c
+jemalloc/test/integration/xallocx.sh
+jemalloc/test/src/btalloc.c
+jemalloc/test/src/btalloc_0.c
+jemalloc/test/src/btalloc_1.c
+jemalloc/test/src/math.c
+jemalloc/test/src/mq.c
+jemalloc/test/src/mtx.c
+jemalloc/test/src/SFMT.c
+jemalloc/test/src/test.c
+jemalloc/test/src/thd.c
+jemalloc/test/src/timer.c
+jemalloc/test/stress/hookbench.c
+jemalloc/test/stress/microbench.c
+jemalloc/test/test.sh.in
+jemalloc/test/unit/a0.c
+jemalloc/test/unit/arena_reset.c
+jemalloc/test/unit/arena_reset_prof.c
+jemalloc/test/unit/arena_reset_prof.sh
+jemalloc/test/unit/atomic.c
+jemalloc/test/unit/background_thread.c
+jemalloc/test/unit/background_thread_enable.c
+jemalloc/test/unit/base.c
+jemalloc/test/unit/binshard.c
+jemalloc/test/unit/binshard.sh
+jemalloc/test/unit/bitmap.c
+jemalloc/test/unit/bit_util.c
+jemalloc/test/unit/ckh.c
+jemalloc/test/unit/decay.c
+jemalloc/test/unit/decay.sh
+jemalloc/test/unit/div.c
+jemalloc/test/unit/emitter.c
+jemalloc/test/unit/extent_quantize.c
+jemalloc/test/unit/extent_util.c
+jemalloc/test/unit/fork.c
+jemalloc/test/unit/hash.c
+jemalloc/test/unit/hook.c
+jemalloc/test/unit/huge.c
+jemalloc/test/unit/junk.c
+jemalloc/test/unit/junk.sh
+jemalloc/test/unit/junk_alloc.c
+jemalloc/test/unit/junk_alloc.sh
+jemalloc/test/unit/junk_free.c
+jemalloc/test/unit/junk_free.sh
+jemalloc/test/unit/log.c
+jemalloc/test/unit/mallctl.c
+jemalloc/test/unit/malloc_io.c
+jemalloc/test/unit/math.c
+jemalloc/test/unit/mq.c
+jemalloc/test/unit/mtx.c
+jemalloc/test/unit/nstime.c
+jemalloc/test/unit/pack.c
+jemalloc/test/unit/pack.sh
+jemalloc/test/unit/pages.c
+jemalloc/test/unit/ph.c
+jemalloc/test/unit/prng.c
+jemalloc/test/unit/prof_accum.c
+jemalloc/test/unit/prof_accum.sh
+jemalloc/test/unit/prof_active.c
+jemalloc/test/unit/prof_active.sh
+jemalloc/test/unit/prof_gdump.c
+jemalloc/test/unit/prof_gdump.sh
+jemalloc/test/unit/prof_idump.c
+jemalloc/test/unit/prof_idump.sh
+jemalloc/test/unit/prof_log.c
+jemalloc/test/unit/prof_log.sh
+jemalloc/test/unit/prof_reset.c
+jemalloc/test/unit/prof_reset.sh
+jemalloc/test/unit/prof_tctx.c
+jemalloc/test/unit/prof_tctx.sh
+jemalloc/test/unit/prof_thread_name.c
+jemalloc/test/unit/prof_thread_name.sh
+jemalloc/test/unit/ql.c
+jemalloc/test/unit/qr.c
+jemalloc/test/unit/rb.c
+jemalloc/test/unit/retained.c
+jemalloc/test/unit/rtree.c
+jemalloc/test/unit/safety_check.c
+jemalloc/test/unit/safety_check.sh
+jemalloc/test/unit/sc.c
+jemalloc/test/unit/seq.c
+jemalloc/test/unit/SFMT.c
+jemalloc/test/unit/size_classes.c
+jemalloc/test/unit/slab.c
+jemalloc/test/unit/smoothstep.c
+jemalloc/test/unit/spin.c
+jemalloc/test/unit/stats.c
+jemalloc/test/unit/stats_print.c
+jemalloc/test/unit/test_hooks.c
+jemalloc/test/unit/ticker.c
+jemalloc/test/unit/tsd.c
+jemalloc/test/unit/witness.c
+jemalloc/test/unit/zero.c
+jemalloc/test/unit/zero.sh
+jemalloc/TUNING.md
+m4/ax_cxx_compile_stdcxx.m4
+m4/ax_cxx_compile_stdcxx_11.m4
+m4/ax_pthread.m4
+man/autohbw.7
+man/hbwallocator.3
+man/hbwmalloc.3
+man/memkind-auto-dax-kmem-nodes.1
+man/memkind-hbw-nodes.1
+man/memkind.3
+man/memkind_arena.3
+man/memkind_dax_kmem.3
+man/memkind_default.3
+man/memkind_hbw.3
+man/memkind_hugetlb.3
+man/memkind_pmem.3
+man/memkindallocator.3
+man/pmemallocator.3
+memkind.pc.in
+memkind.spec.mk
+src/Makefile.mk
+src/hbwmalloc.c
+src/heap_manager.c
+man/memkind-auto-dax-kmem-nodes.c
+src/memkind-hbw-nodes.c
+src/memkind.c
+src/memkind_arena.c
+src/memkind_bandwidth.c
+src/memkind_dax_kmem.c
+src/memkind_default.c
+src/memkind_gbtlb.c
+src/memkind_hbw.c
+src/memkind_hugetlb.c
+src/memkind_interleave.c
+src/memkind_log.c
+src/memkind_pmem.c
+src/memkind_regular.c
+src/tbb_wrapper.c
+test/Allocator.hpp
+test/Makefile.mk
+test/TestPolicy.hpp
+test/alloc_benchmark.c
+test/alloc_performance_tests.cpp
+test/allocate_to_max_stress_test.cpp
+test/allocator_perf_tool/AllocationSizes.hpp
+test/allocator_perf_tool/Allocation_info.cpp
+test/allocator_perf_tool/Allocation_info.hpp
+test/allocator_perf_tool/Allocator.hpp
+test/allocator_perf_tool/AllocatorFactory.hpp
+test/allocator_perf_tool/CSVLogger.hpp
+test/allocator_perf_tool/CommandLine.hpp
+test/allocator_perf_tool/Configuration.hpp
+test/allocator_perf_tool/ConsoleLog.hpp
+test/allocator_perf_tool/FunctionCalls.hpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.h
+test/allocator_perf_tool/GTestAdapter.hpp
+test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/HugePageOrganizer.hpp
+test/allocator_perf_tool/HugePageUnmap.hpp
+test/allocator_perf_tool/Iterator.hpp
+test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/Makefile
+test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
+test/allocator_perf_tool/Numastat.hpp
+test/allocator_perf_tool/PmemMockup.cpp
+test/allocator_perf_tool/PmemMockup.hpp
+test/allocator_perf_tool/Runnable.hpp
+test/allocator_perf_tool/ScenarioWorkload.cpp
+test/allocator_perf_tool/ScenarioWorkload.h
+test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
+test/allocator_perf_tool/Stats.hpp
+test/allocator_perf_tool/StressIncreaseToMax.cpp
+test/allocator_perf_tool/StressIncreaseToMax.h
+test/allocator_perf_tool/Task.hpp
+test/allocator_perf_tool/TaskFactory.hpp
+test/allocator_perf_tool/Tests.hpp
+test/allocator_perf_tool/Thread.hpp
+test/allocator_perf_tool/TimerSysTime.hpp
+test/allocator_perf_tool/VectorIterator.hpp
+test/allocator_perf_tool/Workload.hpp
+test/allocator_perf_tool/WrappersMacros.hpp
+test/allocator_perf_tool/main.cpp
+test/autohbw_test.py
+test/autohbw_test_helper.c
+test/bat_tests.cpp
+test/check.cpp
+test/check.h
+test/common.h
+test/dax_kmem_nodes.cpp
+test/dax_kmem_nodes.h
+test/dax_kmem_env_var_test.py
+test/decorator_test.cpp
+test/decorator_test.h
+test/dlopen_test.cpp
+test/draw_plots.py
+test/environ_err_dax_kmem_malloc_positive_test.cpp
+test/environ_err_dax_kmem_malloc_test.cpp
+test/environ_err_hbw_malloc_test.cpp
+test/error_message_tests.cpp
+test/fragmentation_benchmark_pmem.cpp
+test/freeing_memory_segfault_test.cpp
+test/gb_page_tests_bind_policy.cpp
+test/get_arena_test.cpp
+test/gtest_fused/gtest/gtest-all.cc
+test/gtest_fused/gtest/gtest.h
+test/hbw_allocator_performance_tests.cpp
+test/hbw_allocator_tests.cpp
+test/hbw_detection_test.py
+test/hbw_verify_function_test.cpp
+test/heap_manager_init_perf_test.cpp
+test/huge_page_test.cpp
+test/load_tbbmalloc_symbols.c
+test/locality_test.cpp
+test/main.cpp
+test/memkind_allocator_tests.cpp
+test/memkind-afts-ext.ts
+test/memkind-afts.ts
+test/memkind-perf-ext.ts
+test/memkind-perf.ts
+test/memkind-pytests.ts
+test/memkind-slts.ts
+test/memkind_dax_kmem_test.cpp
+test/memkind_detect_kind_tests.cpp
+test/memkind_null_kind_test.cpp
+test/memkind_pmem_config_tests.cpp
+test/memkind_pmem_long_time_tests.cpp
+test/memkind_pmem_tests.cpp
+test/memkind_stat_test.cpp
+test/memkind_defrag_reallocate.cpp
+test/memkind_versioning_tests.cpp
+test/memory_footprint_test.cpp
+test/memory_manager.h
+test/multithreaded_tests.cpp
+test/negative_tests.cpp
+test/performance/framework.cpp
+test/performance/framework.hpp
+test/performance/operations.hpp
+test/performance/perf_tests.cpp
+test/performance/perf_tests.hpp
+test/pmem_alloc_performance_tests.cpp
+test/pmem_allocator_tests.cpp
+test/proc_stat.h
+test/python_framework/__init__.py
+test/python_framework/cmd_helper.py
+test/python_framework/huge_page_organizer.py
+test/random_sizes_allocator.h
+test/run_alloc_benchmark.sh
+test/static_kinds_list.h
+test/static_kinds_tests.cpp
+test/tbbmalloc.h
+test/test.sh
+test/test_dax_kmem.sh
+test/trace_mechanism_test.py
+test/trace_mechanism_test_helper.c
+test/trial_generator.cpp
+test/trial_generator.h
+utils/docker/Dockerfile.fedora-31
+utils/docker/Dockerfile.ubuntu-18.04
+utils/docker/README.md
+utils/docker/docker_install_ndctl.sh
+utils/docker/docker_install_tbb.sh
+utils/docker/docker_run_build.sh
+utils/docker/docker_run_coverage.sh
+utils/docker/docker_run_test.sh
+utils/docker/run_local.sh
+utils/docker/set_host_configuration.sh
diff --git a/Makefile.am b/Makefile.am
new file mode 100644
index 0000000..d3b2039
--- /dev/null
+++ b/Makefile.am
@@ -0,0 +1,526 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+ACLOCAL_AMFLAGS = -I m4
+AM_CPPFLAGS = -I$(top_srcdir)/include -I$(top_srcdir)/jemalloc/include
+MAN2HTML = groff -mandoc -Thtml
+
+lib_LTLIBRARIES = libmemkind.la
+
+libmemkind_la_SOURCES = src/hbwmalloc.c \
+                        src/heap_manager.c \
+                        src/memkind.c \
+                        src/memkind_arena.c \
+                        src/memkind_bandwidth.c \
+                        src/memkind_dax_kmem.c \
+                        src/memkind_default.c \
+                        src/memkind_gbtlb.c \
+                        src/memkind_hbw.c \
+                        src/memkind_hugetlb.c \
+                        src/memkind_interleave.c \
+                        src/memkind_log.c \
+                        src/memkind_pmem.c \
+                        src/memkind_regular.c \
+                        src/tbb_wrapper.c \
+                        # end
+
+
+libmemkind_la_LIBADD = jemalloc/lib/libjemalloc_pic.a
+libmemkind_la_LDFLAGS = -version-info 0:1:0 -ldl -lrt
+include_HEADERS = include/hbw_allocator.h \
+                  include/hbwmalloc.h \
+                  include/memkind.h \
+                  include/memkind_allocator.h \
+                  include/memkind_deprecated.h \
+                  include/pmem_allocator.h \
+                  # end
+
+noinst_HEADERS =  include/memkind/internal/heap_manager.h \
+                  include/memkind/internal/memkind_arena.h \
+                  include/memkind/internal/memkind_bandwidth.h \
+                  include/memkind/internal/memkind_dax_kmem.h \
+                  include/memkind/internal/memkind_default.h \
+                  include/memkind/internal/memkind_gbtlb.h \
+                  include/memkind/internal/memkind_hbw.h \
+                  include/memkind/internal/memkind_hugetlb.h \
+                  include/memkind/internal/memkind_interleave.h \
+                  include/memkind/internal/memkind_log.h \
+                  include/memkind/internal/memkind_pmem.h \
+                  include/memkind/internal/memkind_private.h \
+                  include/memkind/internal/memkind_regular.h \
+                  include/memkind/internal/tbb_mem_pool_policy.h \
+                  include/memkind/internal/tbb_wrapper.h \
+                  include/memkind/internal/vec.h \
+                  # end
+
+EXTRA_DIST = CONTRIBUTING \
+             VERSION \
+             astyle.sh \
+             autogen.sh \
+             build.sh \
+             examples/README \
+             jemalloc/.appveyor.yml \
+             jemalloc/.autom4te.cfg \
+             jemalloc/.gitattributes\
+             jemalloc/.gitignore \
+             jemalloc/.travis.yml \
+             jemalloc/autogen.sh \
+             jemalloc/bin/jemalloc-config.in \
+             jemalloc/bin/jemalloc.sh.in \
+             jemalloc/bin/jeprof.in \
+             jemalloc/build-aux/config.guess \
+             jemalloc/build-aux/config.sub \
+             jemalloc/build-aux/install-sh \
+             jemalloc/ChangeLog \
+             jemalloc/config.stamp.in \
+             jemalloc/configure.ac \
+             jemalloc/COPYING \
+             jemalloc/doc/html.xsl.in \
+             jemalloc/doc/jemalloc.xml.in \
+             jemalloc/doc/manpages.xsl.in \
+             jemalloc/doc/stylesheet.xsl \
+             jemalloc/include/jemalloc/internal/arena_externs.h \
+             jemalloc/include/jemalloc/internal/arena_inlines_a.h \
+             jemalloc/include/jemalloc/internal/arena_inlines_b.h \
+             jemalloc/include/jemalloc/internal/arena_stats.h \
+             jemalloc/include/jemalloc/internal/arena_structs_a.h \
+             jemalloc/include/jemalloc/internal/arena_structs_b.h \
+             jemalloc/include/jemalloc/internal/arena_types.h \
+             jemalloc/include/jemalloc/internal/assert.h \
+             jemalloc/include/jemalloc/internal/atomic.h \
+             jemalloc/include/jemalloc/internal/atomic_c11.h \
+             jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h \
+             jemalloc/include/jemalloc/internal/atomic_gcc_sync.h \
+             jemalloc/include/jemalloc/internal/atomic_msvc.h \
+             jemalloc/include/jemalloc/internal/background_thread_externs.h \
+             jemalloc/include/jemalloc/internal/background_thread_inlines.h \
+             jemalloc/include/jemalloc/internal/background_thread_structs.h \
+             jemalloc/include/jemalloc/internal/base_externs.h \
+             jemalloc/include/jemalloc/internal/base_inlines.h \
+             jemalloc/include/jemalloc/internal/base_structs.h \
+             jemalloc/include/jemalloc/internal/base_types.h \
+             jemalloc/include/jemalloc/internal/bin.h \
+             jemalloc/include/jemalloc/internal/bin_stats.h \
+             jemalloc/include/jemalloc/internal/bin_types.h \
+             jemalloc/include/jemalloc/internal/bitmap.h \
+             jemalloc/include/jemalloc/internal/bit_util.h \
+             jemalloc/include/jemalloc/internal/cache_bin.h \
+             jemalloc/include/jemalloc/internal/ckh.h \
+             jemalloc/include/jemalloc/internal/ctl.h \
+             jemalloc/include/jemalloc/internal/div.h \
+             jemalloc/include/jemalloc/internal/emitter.h \
+             jemalloc/include/jemalloc/internal/extent_dss.h \
+             jemalloc/include/jemalloc/internal/extent_externs.h \
+             jemalloc/include/jemalloc/internal/extent_inlines.h \
+             jemalloc/include/jemalloc/internal/extent_mmap.h \
+             jemalloc/include/jemalloc/internal/extent_structs.h \
+             jemalloc/include/jemalloc/internal/extent_types.h \
+             jemalloc/include/jemalloc/internal/hash.h \
+             jemalloc/include/jemalloc/internal/hook.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_types.h \
+             jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in \
+             jemalloc/include/jemalloc/internal/large_externs.h \
+             jemalloc/include/jemalloc/internal/log.h \
+             jemalloc/include/jemalloc/internal/malloc_io.h \
+             jemalloc/include/jemalloc/internal/mutex.h \
+             jemalloc/include/jemalloc/internal/mutex_pool.h \
+             jemalloc/include/jemalloc/internal/mutex_prof.h \
+             jemalloc/include/jemalloc/internal/nstime.h \
+             jemalloc/include/jemalloc/internal/pages.h \
+             jemalloc/include/jemalloc/internal/ph.h \
+             jemalloc/include/jemalloc/internal/private_namespace.sh \
+             jemalloc/include/jemalloc/internal/private_symbols.sh \
+             jemalloc/include/jemalloc/internal/prng.h \
+             jemalloc/include/jemalloc/internal/prof_externs.h \
+             jemalloc/include/jemalloc/internal/prof_inlines_a.h \
+             jemalloc/include/jemalloc/internal/prof_inlines_b.h \
+             jemalloc/include/jemalloc/internal/prof_structs.h \
+             jemalloc/include/jemalloc/internal/prof_types.h \
+             jemalloc/include/jemalloc/internal/public_namespace.sh \
+             jemalloc/include/jemalloc/internal/public_unnamespace.sh \
+             jemalloc/include/jemalloc/internal/ql.h \
+             jemalloc/include/jemalloc/internal/qr.h \
+             jemalloc/include/jemalloc/internal/quantum.h \
+             jemalloc/include/jemalloc/internal/rb.h \
+             jemalloc/include/jemalloc/internal/rtree.h \
+             jemalloc/include/jemalloc/internal/rtree_tsd.h \
+             jemalloc/include/jemalloc/internal/safety_check.h \
+             jemalloc/include/jemalloc/internal/sc.h \
+             jemalloc/include/jemalloc/internal/seq.h \
+             jemalloc/include/jemalloc/internal/smoothstep.h \
+             jemalloc/include/jemalloc/internal/smoothstep.sh \
+             jemalloc/include/jemalloc/internal/spin.h \
+             jemalloc/include/jemalloc/internal/stats.h \
+             jemalloc/include/jemalloc/internal/sz.h \
+             jemalloc/include/jemalloc/internal/tcache_externs.h \
+             jemalloc/include/jemalloc/internal/tcache_inlines.h \
+             jemalloc/include/jemalloc/internal/tcache_structs.h \
+             jemalloc/include/jemalloc/internal/tcache_types.h \
+             jemalloc/include/jemalloc/internal/test_hooks.h \
+             jemalloc/include/jemalloc/internal/ticker.h \
+             jemalloc/include/jemalloc/internal/tsd.h \
+             jemalloc/include/jemalloc/internal/tsd_generic.h \
+             jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h \
+             jemalloc/include/jemalloc/internal/tsd_tls.h \
+             jemalloc/include/jemalloc/internal/tsd_types.h \
+             jemalloc/include/jemalloc/internal/tsd_win.h \
+             jemalloc/include/jemalloc/internal/util.h \
+             jemalloc/include/jemalloc/internal/witness.h \
+             jemalloc/include/jemalloc/jemalloc.sh \
+             jemalloc/include/jemalloc/jemalloc_defs.h.in \
+             jemalloc/include/jemalloc/jemalloc_macros.h.in \
+             jemalloc/include/jemalloc/jemalloc_mangle.sh \
+             jemalloc/include/jemalloc/jemalloc_protos.h.in \
+             jemalloc/include/jemalloc/jemalloc_rename.sh \
+             jemalloc/include/jemalloc/jemalloc_typedefs.h.in \
+             jemalloc/include/msvc_compat/C99/stdbool.h \
+             jemalloc/include/msvc_compat/C99/stdint.h \
+             jemalloc/include/msvc_compat/strings.h \
+             jemalloc/include/msvc_compat/windows_extra.h \
+             jemalloc/INSTALL.md \
+             jemalloc/jemalloc.pc.in \
+             jemalloc/m4/ax_cxx_compile_stdcxx.m4 \
+             jemalloc/Makefile.in \
+             jemalloc/msvc/jemalloc_vc2015.sln \
+             jemalloc/msvc/jemalloc_vc2017.sln \
+             jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj \
+             jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters \
+             jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj \
+             jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters \
+             jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj \
+             jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters \
+             jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj \
+             jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters \
+             jemalloc/msvc/ReadMe.txt \
+             jemalloc/msvc/test_threads/test_threads.cpp \
+             jemalloc/msvc/test_threads/test_threads.h \
+             jemalloc/msvc/test_threads/test_threads_main.cpp \
+             jemalloc/README \
+             jemalloc/run_tests.sh \
+             jemalloc/scripts/gen_run_tests.py \
+             jemalloc/scripts/gen_travis.py \
+             jemalloc/src/arena.c \
+             jemalloc/src/background_thread.c \
+             jemalloc/src/base.c \
+             jemalloc/src/bin.c \
+             jemalloc/src/bitmap.c \
+             jemalloc/src/ckh.c \
+             jemalloc/src/ctl.c \
+             jemalloc/src/div.c \
+             jemalloc/src/extent.c \
+             jemalloc/src/extent_dss.c \
+             jemalloc/src/extent_mmap.c \
+             jemalloc/src/hash.c \
+             jemalloc/src/hook.c \
+             jemalloc/src/jemalloc.c \
+             jemalloc/src/jemalloc_cpp.cpp \
+             jemalloc/src/large.c \
+             jemalloc/src/log.c \
+             jemalloc/src/malloc_io.c \
+             jemalloc/src/mutex.c \
+             jemalloc/src/mutex_pool.c \
+             jemalloc/src/nstime.c \
+             jemalloc/src/pages.c \
+             jemalloc/src/prng.c \
+             jemalloc/src/prof.c \
+             jemalloc/src/rtree.c \
+             jemalloc/src/safety_check.c \
+             jemalloc/src/sc.c \
+             jemalloc/src/stats.c \
+             jemalloc/src/sz.c \
+             jemalloc/src/tcache.c \
+             jemalloc/src/test_hooks.c \
+             jemalloc/src/ticker.c \
+             jemalloc/src/tsd.c \
+             jemalloc/src/witness.c \
+             jemalloc/src/zone.c \
+             jemalloc/test/include/test/btalloc.h \
+             jemalloc/test/include/test/extent_hooks.h \
+             jemalloc/test/include/test/jemalloc_test.h.in \
+             jemalloc/test/include/test/jemalloc_test_defs.h.in \
+             jemalloc/test/include/test/math.h \
+             jemalloc/test/include/test/mq.h \
+             jemalloc/test/include/test/mtx.h \
+             jemalloc/test/include/test/SFMT-alti.h \
+             jemalloc/test/include/test/SFMT-params.h \
+             jemalloc/test/include/test/SFMT-params607.h \
+             jemalloc/test/include/test/SFMT-params1279.h \
+             jemalloc/test/include/test/SFMT-params2281.h \
+             jemalloc/test/include/test/SFMT-params4253.h \
+             jemalloc/test/include/test/SFMT-params11213.h \
+             jemalloc/test/include/test/SFMT-params19937.h \
+             jemalloc/test/include/test/SFMT-params44497.h \
+             jemalloc/test/include/test/SFMT-params86243.h \
+             jemalloc/test/include/test/SFMT-params132049.h \
+             jemalloc/test/include/test/SFMT-params216091.h \
+             jemalloc/test/include/test/SFMT-sse2.h \
+             jemalloc/test/include/test/SFMT.h \
+             jemalloc/test/include/test/test.h \
+             jemalloc/test/include/test/thd.h \
+             jemalloc/test/include/test/timer.h \
+             jemalloc/test/integration/aligned_alloc.c \
+             jemalloc/test/integration/allocated.c \
+             jemalloc/test/integration/extent.c \
+             jemalloc/test/integration/extent.sh \
+             jemalloc/test/integration/malloc.c \
+             jemalloc/test/integration/mallocx.c \
+             jemalloc/test/integration/mallocx.sh \
+             jemalloc/test/integration/MALLOCX_ARENA.c \
+             jemalloc/test/integration/overflow.c \
+             jemalloc/test/integration/posix_memalign.c \
+             jemalloc/test/integration/rallocx.c \
+             jemalloc/test/integration/sdallocx.c \
+             jemalloc/test/integration/slab_sizes.c \
+             jemalloc/test/integration/slab_sizes.sh \
+             jemalloc/test/integration/smallocx.c \
+             jemalloc/test/integration/smallocx.sh \
+             jemalloc/test/integration/thread_arena.c \
+             jemalloc/test/integration/thread_tcache_enabled.c \
+             jemalloc/test/integration/xallocx.c \
+             jemalloc/test/integration/xallocx.sh \
+             jemalloc/test/src/btalloc.c \
+             jemalloc/test/src/btalloc_0.c \
+             jemalloc/test/src/btalloc_1.c \
+             jemalloc/test/src/math.c \
+             jemalloc/test/src/mq.c \
+             jemalloc/test/src/mtx.c \
+             jemalloc/test/src/SFMT.c \
+             jemalloc/test/src/test.c \
+             jemalloc/test/src/thd.c  \
+             jemalloc/test/src/timer.c \
+             jemalloc/test/stress/hookbench.c \
+             jemalloc/test/stress/microbench.c \
+             jemalloc/test/test.sh.in \
+             jemalloc/test/unit/a0.c \
+             jemalloc/test/unit/arena_reset.c \
+             jemalloc/test/unit/arena_reset_prof.c \
+             jemalloc/test/unit/arena_reset_prof.sh \
+             jemalloc/test/unit/atomic.c \
+             jemalloc/test/unit/background_thread.c \
+             jemalloc/test/unit/background_thread_enable.c \
+             jemalloc/test/unit/base.c \
+             jemalloc/test/unit/binshard.c \
+             jemalloc/test/unit/binshard.sh \
+             jemalloc/test/unit/bitmap.c \
+             jemalloc/test/unit/bit_util.c \
+             jemalloc/test/unit/ckh.c \
+             jemalloc/test/unit/decay.c \
+             jemalloc/test/unit/decay.sh \
+             jemalloc/test/unit/div.c \
+             jemalloc/test/unit/emitter.c \
+             jemalloc/test/unit/extent_quantize.c\
+             jemalloc/test/unit/extent_util.c \
+             jemalloc/test/unit/fork.c \
+             jemalloc/test/unit/hash.c \
+             jemalloc/test/unit/hook.c \
+             jemalloc/test/unit/huge.c \
+             jemalloc/test/unit/junk.c \
+             jemalloc/test/unit/junk.sh \
+             jemalloc/test/unit/junk_alloc.c \
+             jemalloc/test/unit/junk_alloc.sh \
+             jemalloc/test/unit/junk_free.c \
+             jemalloc/test/unit/junk_free.sh \
+             jemalloc/test/unit/log.c \
+             jemalloc/test/unit/mallctl.c \
+             jemalloc/test/unit/malloc_io.c \
+             jemalloc/test/unit/math.c \
+             jemalloc/test/unit/mq.c \
+             jemalloc/test/unit/mtx.c \
+             jemalloc/test/unit/nstime.c \
+             jemalloc/test/unit/pack.c \
+             jemalloc/test/unit/pack.sh \
+             jemalloc/test/unit/pages.c \
+             jemalloc/test/unit/ph.c \
+             jemalloc/test/unit/prng.c \
+             jemalloc/test/unit/prof_accum.c \
+             jemalloc/test/unit/prof_accum.sh \
+             jemalloc/test/unit/prof_active.c \
+             jemalloc/test/unit/prof_active.sh \
+             jemalloc/test/unit/prof_gdump.c \
+             jemalloc/test/unit/prof_gdump.sh \
+             jemalloc/test/unit/prof_idump.c \
+             jemalloc/test/unit/prof_idump.sh \
+             jemalloc/test/unit/prof_log.c \
+             jemalloc/test/unit/prof_log.sh \
+             jemalloc/test/unit/prof_reset.c \
+             jemalloc/test/unit/prof_reset.sh \
+             jemalloc/test/unit/prof_tctx.c \
+             jemalloc/test/unit/prof_tctx.sh \
+             jemalloc/test/unit/prof_thread_name.c \
+             jemalloc/test/unit/prof_thread_name.sh\
+             jemalloc/test/unit/ql.c \
+             jemalloc/test/unit/qr.c \
+             jemalloc/test/unit/rb.c \
+             jemalloc/test/unit/retained.c \
+             jemalloc/test/unit/rtree.c \
+             jemalloc/test/unit/safety_check.c \
+             jemalloc/test/unit/safety_check.sh \
+             jemalloc/test/unit/sc.c \
+             jemalloc/test/unit/seq.c \
+             jemalloc/test/unit/SFMT.c \
+             jemalloc/test/unit/size_classes.c \
+             jemalloc/test/unit/slab.c \
+             jemalloc/test/unit/smoothstep.c \
+             jemalloc/test/unit/spin.c \
+             jemalloc/test/unit/stats.c \
+             jemalloc/test/unit/stats_print.c \
+             jemalloc/test/unit/test_hooks.c \
+             jemalloc/test/unit/ticker.c \
+             jemalloc/test/unit/tsd.c \
+             jemalloc/test/unit/witness.c \
+             jemalloc/test/unit/zero.c \
+             jemalloc/test/unit/zero.sh \
+             jemalloc/TUNING.md \
+             memkind-$(VERSION).spec \
+             memkind.pc.in \
+             memkind.spec.mk \
+             test/test.sh \
+             test/test_dax_kmem.sh \
+             utils/docker/Dockerfile.fedora-31 \
+             utils/docker/Dockerfile.ubuntu-18.04 \
+             utils/docker/README.md \
+             utils/docker/docker_install_ndctl.sh \
+             utils/docker/docker_install_tbb.sh \
+             utils/docker/docker_run_build.sh \
+             utils/docker/docker_run_coverage.sh \
+             utils/docker/docker_run_test.sh \
+             utils/docker/run_local.sh \
+             utils/docker/set_host_configuration.sh \
+             # end
+
+dist_doc_DATA = COPYING README VERSION
+dist_man_MANS =  man/autohbw.7 \
+                 man/hbwallocator.3 \
+                 man/hbwmalloc.3 \
+                 man/memkind-auto-dax-kmem-nodes.1 \
+                 man/memkind-hbw-nodes.1 \
+                 man/memkind.3 \
+                 man/memkind_arena.3 \
+                 man/memkind_dax_kmem.3 \
+                 man/memkind_default.3 \
+                 man/memkind_hbw.3 \
+                 man/memkind_hugetlb.3 \
+                 man/memkind_pmem.3 \
+                 man/memkindallocator.3 \
+                 man/pmemallocator.3 \
+                 # end
+
+pkgconfigdir = $(libdir)/pkgconfig
+pkgconfig_DATA = memkind.pc
+
+CLEANFILES = memkind-$(VERSION).spec $(libmemkind_la_LIBADD)
+DISTCLEANFILES = VERSION configure pkgconfig_DATA
+
+bin_PROGRAMS = memkind-auto-dax-kmem-nodes memkind-hbw-nodes
+
+memkind_auto_dax_kmem_nodes_SOURCES = src/memkind-auto-dax-kmem-nodes.c
+memkind_auto_dax_kmem_nodes_LDADD = libmemkind.la
+memkind_auto_dax_kmem_nodes_LDFLAGS = $(DAXCTL_LIBS)
+
+memkind_hbw_nodes_SOURCES = src/memkind-hbw-nodes.c
+memkind_hbw_nodes_LDADD = libmemkind.la
+
+bin_SCRIPTS =
+check_PROGRAMS =
+noinst_PROGRAMS =
+noinst_LTLIBRARIES =
+TESTS =
+
+req_flags = -fvisibility=hidden -Wall -Werror -D_GNU_SOURCE -DJE_PREFIX=@memkind_prefix@
+
+AM_CFLAGS = $(req_flags)
+AM_CXXFLAGS = $(req_flags)
+
+clean-local: clean-jemalloc_deps
+
+html-local:
+	$(MAN2HTML) man/hbwmalloc.3 > man/hbwmalloc.html
+	$(MAN2HTML) man/memkind.3 > man/memkind.html
+	$(MAN2HTML) man/memkindallocator.3 > man/memkindallocator.html
+	$(MAN2HTML) man/pmemallocator.3 > man/pmemallocator.html
+
+clean-jemalloc_deps:
+	[ -e jemalloc/configure ] && (cd jemalloc && $(MAKE) clean) || exit 0
+	[ -e jemalloc/configure ] && (cd jemalloc && rm configure) || exit 0
+
+.PHONY: jemalloc_deps checkprogs clean-jemalloc_deps html-local
+.NOTPARALLEL: static_lib
+
+# build check programs without running tests
+checkprogs: libmemkind.la $(check_PROGRAMS)
+
+memkind-$(VERSION).spec:
+	$(MAKE) version="$(VERSION)" -f memkind.spec.mk $@
+
+rpm: dist
+	$(MAKE) version="$(VERSION)" -f memkind.spec.mk $@
+
+all: jemalloc_deps static_lib
+
+# the script merge memkind and jemalloc libraries into one static library.
+define ar_prog
+create libmemkind.a\n\
+addlib .libs/libmemkind.a\n\
+addlib $(libmemkind_la_LIBADD)\n\
+save\n\
+end
+endef
+
+static_lib: libmemkind.la
+	bash -c "ar -M < <(echo -e '$(ar_prog)')"
+	cp libmemkind.a .libs/
+	rm libmemkind.a
+
+JEMALLOC_CONFIG = --enable-autogen --without-export --with-version=5.2.1-0-gea6b3e973b477b8061e0076bb257dbd7f3faa756 \
+			--disable-fill --disable-initial-exec-tls --with-jemalloc-prefix=@memkind_prefix@ \
+			--with-malloc-conf="narenas:@auto_arenas@,lg_tcache_max:@tcache_max_size_class@" \
+			#end
+
+jemalloc_deps:
+	[ ! -e jemalloc/configure ] && (cd jemalloc && autoconf) || exit 0
+	[ ! -e $(libmemkind_la_LIBADD) ] && (cd jemalloc && ./configure $(JEMALLOC_CONFIG) && $(MAKE)) || exit 0
+
+unit_tests: checkprogs
+	test/all_tests --gtest_filter=*free_with_NULL_kind_4096_bytes:NegativeTest.*:ErrorMessage.*:GetArenaTest.*:MemkindVersioningTests.*
+	test/decorator_test
+
+PMEM_PATH=/tmp/
+unit_tests_pmem: checkprogs
+	test/pmem_test -d $(PMEM_PATH)
+
+unit_tests_dax_kmem: checkprogs
+	test/test_dax_kmem.sh
+
+include autohbw/Makefile.mk
+include test/Makefile.mk
+include examples/Makefile.mk
+include src/Makefile.mk
diff --git a/NEWS b/NEWS
new file mode 100644
index 0000000..c72d191
--- /dev/null
+++ b/NEWS
@@ -0,0 +1,42 @@
+OPEN DEVELOPMENT
+================
+
+We are moving to a more open development model with memkind.
+
+- The memkind github organization:
+  https://github.com/memkind
+
+- Rather than posting bulk patches at time of tag, we are posting more
+  incremental patches with finer grain commit details.
+
+- There is now a "dev" branch in the memkind repository and a
+  "memkind-dev" branch in the memkind fork of the jemalloc repository.
+  These will be used for open development work.
+
+- There is a memkind web page here:
+  http://memkind.github.io/memkind
+
+- We have posted an architecture document here:
+  http://memkind.github.io/memkind/memkind_arch_20150318.pdf
+
+- We invite you to join the memkind mailing list:
+  https://lists.01.org/mailman/listinfo/memkind
+
+- Feel free to open a github issue, especially if there is a TODO item
+  that you would like to prioritize.
+
+- We have posted source and binary RPMs for several distributions here:
+  http://download.opensuse.org/repositories/home:/cmcantalupo/
+
+- memkind is also available in Fedora package repository (devel, EPEL7):
+  https://admin.fedoraproject.org/pkgdb/package/memkind/
+
+- Pull requests are welcome.
+
+- White space formatting generally conforms to
+  $ astyle --style=linux --indent=spaces=4 -S --max-continuation-indent=80 \
+           --max-code-length=80 --break-after-logical --indent-namespaces -z2 \
+           --align-pointer=name
+
+- For more information about astyle:
+  http://astyle.sourceforge.net/
diff --git a/PULL_REQUEST_TEMPLATE.md b/PULL_REQUEST_TEMPLATE.md
new file mode 100644
index 0000000..41b7945
--- /dev/null
+++ b/PULL_REQUEST_TEMPLATE.md
@@ -0,0 +1,29 @@
+<!--- Provide a general summary of your changes in the Title above -->
+
+### Description
+<!--- Describe your changes in detail -->
+<!--- If it fixes an open issue, please link to the issue here. -->
+
+### Types of changes
+<!--- Put an `x` in the box(es) that apply -->
+
+- [ ] Bugfix (non-breaking change which fixes issue linked in Description above)
+- [ ] New feature (non-breaking change which adds functionality)
+- [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
+- [ ] Documentation (correction or otherwise)
+- [ ] Cosmetics (whitespace, appearance)
+- [ ] Other
+
+### Checklist
+<!--- Put an `x` in the box(es) that apply -->
+
+- [ ] Code compiles without errors
+- [ ] All tests pass locally with my changes (see TESTING section in CONTRIBUTING file)
+- [ ] Created tests which will fail without the change (if possible)
+- [ ] Extended the README/documentation (if necessary)
+- [ ] All newly added files have proprietary license (if necessary)
+- [ ] All newly added files are referenced in MANIFEST files (if necessary)
+
+## Further comments
+<!--- If this is a relatively large or complex change, kick off the discussion by explaining why you -->
+<!--- choose the solution you did and what alternatives you considered, etc... -->
diff --git a/README b/README
new file mode 100644
index 0000000..f76195a
--- /dev/null
+++ b/README
@@ -0,0 +1,390 @@
+# **MEMKIND**
+
+[![Build Status](https://travis-ci.org/memkind/memkind.svg)](https://travis-ci.org/memkind/memkind)
+[![MEMKIND version](https://img.shields.io/github/tag/memkind/memkind.svg)](https://github.com/memkind/memkind/releases/latest)
+[![Coverage Status](http://codecov.io/github/memkind/memkind/coverage.svg?branch=master)](http://codecov.io/gh/memkind/memkind?branch=master)
+[![Packaging status](https://repology.org/badge/tiny-repos/memkind.svg)](https://repology.org/project/memkind/versions)
+
+The memkind library is a user extensible heap manager built on top of
+jemalloc which enables control of memory characteristics and a
+partitioning of the heap between kinds of memory.  The kinds of memory
+are defined by operating system memory policies that have been applied
+to virtual address ranges.  Memory characteristics supported by
+memkind without user extension include control of NUMA and page size
+features.  The jemalloc non-standard interface has been extended to
+enable specialized arenas to make requests for virtual memory from the
+operating system through the memkind partition interface.  Through the
+other memkind interfaces the user can control and extend memory
+partition features and allocate memory while selecting enabled
+features.  Memkind interface allows to create and control file-backed memory
+(PMEM kind) on specified device.
+
+## Contents
+
+1. [Interfaces](#interfaces)
+2. [Dependencies](#dependencies)
+3. [Building and Installing](#building-and-installing)
+	* [jemalloc](#jemalloc)
+4. [Run Requirements](#run-requirements)
+5. [Kind Requirements](#kind-requirements)
+6. [Kernel](#kernel)
+7. [NVDIMM volatile usage](#nvdimm-volatile-usage)
+	* [DAX device](#dax-device)
+	* [DAX filesystem](#dax-filesystem)
+8. [The Detection Mechanism of the Kind](#the-detection-mechanism-of-the-kind)
+9. [Setting Logging Mechanism](#setting-logging-mechanism)
+10. [Setting Heap Manager](#setting-heap-manager)
+11. [Testing](#testing)
+12. [Simulating High Bandwidth Memory](#simulating-high-bandwidth-memory)
+13. [Notes](#notes)
+14. [Status](#status)
+15. [Disclaimer](#disclaimer)
+
+## Interfaces
+The memkind library delivers four interfaces:
+ * hbwmalloc.h - recommended for high-bandwidth memory use cases (stable)
+ * memkind.h - generic interface for more complex use cases (stable)
+ * pmem_allocator.h - the C++ allocator that satisfies the C++ Allocator
+   [requirements](https://en.cppreference.com/w/cpp/named_req/Allocator)
+   used for PMEM memory use cases (stable)
+ * memkind_allocator.h - the C++ allocator that satisfies the C++ Allocator
+   [requirements](https://en.cppreference.com/w/cpp/named_req/Allocator)
+   used for static kinds (stable)
+
+For more detailed information about those interfaces see
+corresponding manpages (located in man/ subdir):
+
+    man memkind
+
+    man hbwmalloc
+
+    man pmemallocator
+
+    man memkindallocator
+
+## Dependencies
+
+You will need to install required packages on the build system:
+
+* **autoconf**
+* **automake**
+* **gcc-c++**
+* **libnuma-devel**
+* **libtool**
+* **numactl-devel**
+* **unzip**
+
+For using automatic recognition of PMEM NUMA in MEMKIND_DAX_KMEM:
+
+* **libdaxctl-devel** (v66 or later)
+
+## Building and Installing
+
+The memkind library has a dependency on a related fork of jemalloc.
+The configure scripts and gtest source code are distributed with the
+source tarball included in the source RPM, and this tarball is created
+with the memkind "make dist" target. In contrast to the distributed source
+tarball, the git repository does not include any generated files.
+For this reason some additional steps are required when building
+from a checkout of the git repo. Those steps include running
+the bash script called "autogen.sh" prior to configure.  This script
+will populate a VERSION file based on "git describe", and use
+autoreconf to generate a configure script.
+
+Building and installing memkind in standard system location can be as simple as
+typing the following while in the root directory of the source tree:
+
+    ./autogen.sh
+    ./configure
+    make
+    make install
+
+To install this library into **other locations**, you can use the prefix variable, e.g.:
+
+    ./autogen.sh
+    ./configure --prefix=/usr
+    make
+    make install
+
+This will install files to /usr/lib, /usr/include, /usr/share/doc/, usr/share/man.
+
+See the output of:
+
+    ./configure --help
+
+for more information about either the memkind or the jemalloc configuration options.
+
+## jemalloc
+The jemalloc source was forked from jemalloc version 5.2.1.  This source tree
+is located within the jemalloc subdirectory of the memkind source.  The jemalloc
+source code has been kept close to the original form, and in particular
+the build system has been lightly modified.
+
+## Run Requirements
+
+You will need to install required packages for applications,
+which are using the memkind library for dynamic linking at run time:
+
+* **libnuma**
+* **numactl**
+* **pthread**
+
+## Kind Requirements
+
+| Memory kind                   | NUMA  |  HBW Memory | Hugepages | Device DAX | Filesystem supporting hole punching |
+| ----------------------------- |:-----:|:-----------:|:---------:|:----------:|:-----------------------------------:|
+| MEMKIND_DEFAULT               |       |             |           |            |                                     |
+| MEMKIND_HUGETLB               | X     | X           | X         |            |                                     |
+| MEMKIND_HBW                   | X     | X           |           |            |                                     |
+| MEMKIND_HBW_ALL               | X     | X           |           |            |                                     |
+| MEMKIND_HBW_HUGETLB           | X     | X           | X         |            |                                     |
+| MEMKIND_HBW_ALL_HUGETLB       | X     | X           | X         |            |                                     |
+| MEMKIND_HBW_PREFERRED         | X     | X           |           |            |                                     |
+| MEMKIND_HBW_PREFERRED_HUGETLB | X     | X           | X         |            |                                     |
+| MEMKIND_HBW_INTERLEAVE        | X     | X           |           |            |                                     |
+| MEMKIND_REGULAR               | X     |             |           |            |                                     |
+| MEMKIND_DAX_KMEM              | X     |             |           | X          |                                     |
+| MEMKIND_DAX_KMEM_ALL          | X     |             |           | X          |                                     |
+| MEMKIND_DAX_KMEM_PREFERRED    | X     |             |           | X          |                                     |
+| PMEM kind                     |       |             |           |            | X                                   |
+
+## Kernel
+To correctly control of NUMA, huge pages and file-backed memory following
+requirements regarding Linux kernel must be satisfied:
+
+* **NUMA**
+
+Requires kernel
+[patch](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=3964acd0dbec123aa0a621973a2a0580034b4788)
+introduced in Linux v3.11 that impacts
+functionality of the NUMA system calls.  Red Hat has back-ported this patch to the
+v3.10 kernel in the RHEL 7.0 GA release, so RHEL 7.0 onward supports
+memkind even though this kernel version predates v3.11.
+
+* **Hugepages**
+
+Functionality related to hugepages allocation require patches
+[patch1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e0ec90ee7e6f6cbaa6d59ffb48d2a7af5e80e61d) and
+[patch2](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=099730d67417dfee273e9b10ac2560ca7fac7eb9)
+Without them physical memory may end up being located on incorrect NUMA node.
+
+* **2MB Pages**
+
+To use the interfaces for obtaining 2MB pages please be sure to follow the instructions
+[here](https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt) and pay particular
+attention to the use of the procfs files:
+
+    /proc/sys/vm/nr_hugepages
+    /proc/sys/vm/nr_overcommit_hugepages
+
+for enabling the kernel's huge page pool.
+
+* **Filesystem supporting hole punching**
+
+To use the PMEM kind, please be sure that filesystem which is used for PMEM creation supports
+[FALLOC_FL_PUNCH_HOLE](http://man7.org/linux/man-pages/man2/fallocate.2.html) flag.
+
+* **Device DAX**
+
+To use MEMKIND_DAX_KMEM_* kinds you need at least Linux Kernel 5.1 (with enabled DEV_DAX_KMEM Kernel option) and created DAX device.
+If you have loaded dax_pmem_compat module instead of dax_pmem, please read
+[this](https://pmem.io/ndctl/daxctl-migrate-device-model.html) article and migrate device model
+to use alternative device dax drivers, e.g. kmem.
+
+If you want to migrate device model type:
+
+    daxctl migrate-device-model
+
+To create a new DAX device you can type:
+
+    ndctl create-namespace --mode=devdax --map=mem
+
+To display a list of created devices:
+
+    ndctl list
+
+If you have already created device in another mode, you can reconfigure the device using:
+
+    ndctl create-namespace --mode=devdax --map=mem --force -e namespaceX.Y
+
+Where namespaceX.Y means namespace you want to reconfigure.
+
+For more details about creating new namespace read [this](https://pmem.io/ndctl/ndctl-create-namespace.html).
+
+Conversion from device dax to NUMA node can be performed using following commands:
+
+    daxctl reconfigure-device daxX.Y --mode=system-ram
+
+Where daxX.Y is DAX device you want to reconfigure.
+
+This will migrate device from device_dax to kmem driver. After this step daxctl should be able to
+see devices in system-ram mode:
+
+    daxctl list
+
+Example output after reconfigure dax2.0 as system-ram:
+
+    [
+      {
+        "chardev":"dax1.0",
+        "size":263182090240,
+        "target_node":3,
+        "mode":"devdax"
+      },
+      {
+        "chardev":"dax2.0",
+        "size":263182090240,
+        "target_node":4,
+        "mode":"system-ram"
+      }
+    ]
+
+Also you should be able to check new NUMA node configuration by:
+
+    numactl -H
+
+Example output, where NUMA node 4 is NUMA node created from persistent memory:
+
+    available: 3 nodes (0-1,4)
+    node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83
+    node 0 size: 192112 MB
+    node 0 free: 185575 MB
+    node 1 cpus: 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
+    node 1 size: 193522 MB
+    node 1 free: 193107 MB
+    node 4 cpus:
+    node 4 size: 250880 MB
+    node 4 free: 250879 MB
+    node distances:
+    node   0   1   4
+      0:  10  21  17
+      1:  21  10  28
+      4:  17  28  10
+
+## NVDIMM volatile usage
+
+Memkind supports using persistent memory as an extension of DRAM.
+This volatile memory solution is provided by the library with two separate ways described below.
+
+## DAX device
+
+NVDIMM memory as DAX device is supported by MEMKIND_DAX_KMEM_* kinds. With this solution
+persistent memory will be seen in OS as separate NUMA nodes.
+
+Memkind allows two ways to use this kind:
+
+- first implicitly, by allowing memkind library for automatic recognition of NUMA nodes
+created from persistent memory using [libdaxctl-devel](#dependencies)
+- secondary explicitly, by using MEMKIND_DAX_KMEM_NODES environment variable set to
+comma separated list of NUMA nodes which will be treated as NUMA nodes created from persistent memory,
+this solution overrides the first one
+
+## DAX filesystem
+PMEM kind supports the traditional malloc/free interfaces on a memory mapped file.
+This allows the use of [persistent memory](https://pmem.io/)
+as volatile memory, for cases where the region of persistent memory
+is useful to an application, but when the application doesn't need it to be persistent.
+PMEM kind is most useful when used with Direct Access storage
+[DAX](https://www.kernel.org/doc/Documentation/filesystems/dax.txt), which is
+memory-addressable persistent storage that supports load/store access without being paged via the system page cache.
+
+Application using memkind library supports managing a data placement:
+
+| Data placement | Memory kind               |
+|----------------|---------------------------|
+| PMEM (fsdax)   | PMEM kind                 |
+| PMEM (devdax)  | MEMKIND_DAX_KMEM kind     |
+| DRAM           | e.g. MEMKIND_DEFAULT kind |
+
+Currently, the PMEM kind is supported only by the jemalloc heap manager.
+
+## The Detection Mechanism of the Kind
+One of the notable features of the memkind is to detect the correct kind of previously allocated memory.
+
+| Operations                                  | Memkind API function                   |
+|---------------------------------------------|----------------------------------------|
+| Freeing memory                              | memkind_free(kind, ptr)                |
+| Reallocating memory                         | memkind_realloc(kind, ptr, size)       |
+| Obtaining the size of allocated memory      | memkind_malloc_usable_size(kind, ptr)  |
+| Reallocating memory to reduce fragmentation | memkind_defrag_reallocate(kind, ptr)   |
+
+Operations above could be unified for all used memory kinds
+by passing a NULL value as a kind to the functions mentioned above.
+
+For more details, please see the following
+[example](https://github.com/memkind/memkind/blob/master/examples/pmem_free_with_unknown_kind.c).
+
+**Important Notes:**
+The look up for correct kind could result in serious performance penalty,
+which can be avoided by specifying a correct kind explicitly.
+
+## Setting Logging Mechanism
+In memkind library logging mechanism could be enabled by setting MEMKIND_DEBUG
+environment variable. Setting MEMKIND_DEBUG to "1" enables printing messages
+like errors and general information about environment to stderr.
+
+## Setting Heap Manager
+In memkind library heap management can be adjusted with MEMKIND_HEAP_MANAGER
+environment variable, which allows for switching to one of the available
+heap managers.
+Values:
+- JEMALLOC - sets the jemalloc heap manager
+- TBB - sets Intel Threading Building Blocks heap manager. This option requires installed
+Intel Threading Building Blocks library.
+
+If the MEMKIND_HEAP_MANAGER is not set then the jemalloc heap manager will be used by default.
+
+## Testing
+All existing tests pass. For more information on how to execute tests
+see the [CONTRIBUTING](CONTRIBUTING) file.
+
+When tests are run on a NUMA platform without high bandwidth memory
+the MEMKIND_HBW_NODES environment variable is used in conjunction with
+"numactl --membind" to force standard allocations to one NUMA node and
+high bandwidth allocations through a different NUMA node.  See next
+section for more details.
+
+
+## Simulating High Bandwidth Memory
+A method for testing for the benefit of high bandwidth memory on a
+dual socket Intel(R) Xeon(TM) system is to use the QPI bus to simulate
+slow memory.  This is not an accurate model of the bandwidth and
+latency characteristics of the Intel's 2nd generation Intel(R) Xeon Phi(TM)
+Product Family on package memory, but is a reasonable way to determine
+which data structures rely critically on bandwidth.
+
+If the application a.out has been modified to use high bandwidth
+memory with the memkind library then this can be done with numactl as
+follows with the bash shell:
+
+    export MEMKIND_HBW_NODES=0
+    numactl --membind=1 --cpunodebind=0 a.out
+
+or with csh:
+
+    setenv MEMKIND_HBW_NODES 0
+    numactl --membind=1 --cpunodebind=0 a.out
+
+The MEMKIND_HBW_NODES environment variable set to zero will bind high
+bandwidth allocations to NUMA node 0.  The --membind=1 flag to numactl
+will bind standard allocations, static and stack variables to NUMA
+node 1.  The --cpunodebind=0 option to numactl will bind the process
+threads to CPUs associated with NUMA node 0.  With this configuration
+standard allocations will be fetched across the QPI bus, and high
+bandwidth allocations will be local to the process CPU.
+
+## Notes
+* Using memkind with Transparent Huge Pages enabled may result in
+ undesirably high memory footprint. To avoid that disable THP using following
+ [instruction](https://www.kernel.org/doc/Documentation/vm/transhuge.txt)
+
+## Status
+Different interfaces can represent different maturity level
+(as described in corresponding man pages).
+Feedback on design and implementation is greatly appreciated.
+
+## Disclaimer
+SEE [COPYING](COPYING) FILE FOR LICENSE INFORMATION.
+
+THIS SOFTWARE IS PROVIDED AS A DEVELOPMENT SNAPSHOT TO AID
+COLLABORATION AND WAS NOT ISSUED AS A RELEASED PRODUCT BY INTEL.
diff --git a/README.md b/README.md
new file mode 120000
index 0000000..100b938
--- /dev/null
+++ b/README.md
@@ -0,0 +1 @@
+README
\ No newline at end of file
diff --git a/astyle.sh b/astyle.sh
new file mode 100755
index 0000000..96b801b
--- /dev/null
+++ b/astyle.sh
@@ -0,0 +1,51 @@
+#!/bin/bash
+#
+#  Copyright (C) 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ASTYLE_MIN_VER="3.1"
+ASTYLE_OPT="--style=linux --indent=spaces=4 -S -r --max-continuation-indent=80 "
+ASTYLE_OPT+="--max-code-length=80 --break-after-logical --indent-namespaces -z2 "
+ASTYLE_OPT+="--align-pointer=name"
+if ! ASTYLE=$(which astyle)
+then
+    echo "Package astyle was not found. Unable to check source files format policy." >&2
+    exit 1
+fi
+ASTYLE_VER=$(astyle --version 2>&1 | awk '{print $NF}')
+if (( $(echo "$ASTYLE_VER < $ASTYLE_MIN_VER" | bc -l) ));
+then
+    echo "Minimal required version of astyle is $ASTYLE_MIN_VER. Detected version is $ASTYLE_VER" >&2
+    echo "Unable to check source files format policy." >&2
+    exit 1
+fi
+$ASTYLE $ASTYLE_OPT ./*.c,*.cpp,*.h,*.hpp --exclude=jemalloc > astyle.out
+if TEST=$(cat astyle.out | grep -c Formatted)
+then
+    cat astyle.out
+    git --no-pager diff
+    echo "Please fix style issues as shown above"
+    exit 1
+else
+    echo "Source code is compliant with format policy. No style issues were found."
+    exit 0
+fi
diff --git a/autogen.sh b/autogen.sh
new file mode 100755
index 0000000..8949097
--- /dev/null
+++ b/autogen.sh
@@ -0,0 +1,54 @@
+#!/bin/bash
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+set -e
+
+# If the VERSION file does not exist, then create it based on git
+# describe or if not in a git repo just set VERSION to 0.0.0.
+# Determine VERSION scheme, based on git:
+# -if HEAD is tag annotated - it is official release e.g. setting VERSION to "1.9.0"
+# -if HEAD is not tag annotated - it is snapshot e.g. setting VERSION to "1.8.0+dev19+g1c93b2d"
+
+if [ ! -f VERSION ]; then
+    if [ -f .git/config ]; then
+        sha=$(git describe --long | awk -F- '{print $(NF)}')
+        release=$(git describe --long | awk -F- '{print $(NF-1)}')
+        version=$(git describe --long | sed -e "s|\(.*\)-$release-$sha|\1|" -e "s|-|+|g" -e "s|^v||")
+        if [ ${release} != "0" ]; then
+            echo "WARNING: No annotated tag referring to this commit was found, setting version to development build " 2>&1
+            version=${version}+dev${release}+${sha}
+        else
+            echo "Annotated tag referring to this commit was found, setting version as an official release" 2>&1
+            version=${version}
+        fi
+    else
+        echo "WARNING: VERSION file does not exist and working directory is not a git repository, setting version to 0.0.0" 2>&1
+        version=0.0.0
+    fi
+    echo $version > VERSION
+fi
+
+autoreconf --install
+
diff --git a/autohbw/Makefile.mk b/autohbw/Makefile.mk
new file mode 100644
index 0000000..9d2fd12
--- /dev/null
+++ b/autohbw/Makefile.mk
@@ -0,0 +1,37 @@
+#
+#  Copyright (C) 2014 - 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+lib_LTLIBRARIES += autohbw/libautohbw.la \
+                   # end
+
+EXTRA_DIST += autohbw/autohbw_get_src_lines.pl
+
+autohbw_libautohbw_la_LIBADD = libmemkind.la
+
+autohbw_libautohbw_la_SOURCES = autohbw/autohbw.c
+
+clean-local: autohbw-clean
+
+autohbw-clean:
+	rm -f autohbw/*.gcno
diff --git a/autohbw/autohbw.c b/autohbw/autohbw.c
new file mode 100644
index 0000000..4bff27b
--- /dev/null
+++ b/autohbw/autohbw.c
@@ -0,0 +1,457 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////
+// File   : autohbw.c
+// Purpose: Library to automatically allocate HBW (MCDRAM)
+// Author : Ruchira Sasanka (ruchira DOT sasanka AT intel DOT com)
+// Date   : Jan 30, 2015
+///////////////////////////////////////////////////////////////////////////
+
+#include <memkind.h>
+
+#include <assert.h>
+#include <ctype.h>
+#include <errno.h>
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#define AUTOHBW_EXPORT __attribute__((visibility("default")))
+#define AUTOHBW_INIT __attribute__((constructor))
+
+//-2 = nothing is printed
+//-1 = critical messages are printed
+// 0 = no log messages for allocations are printed but INFO messages are printed
+// 1 = a log message is printed for each allocation (Default)
+// 2 = a log message is printed for each allocation with a backtrace
+enum {
+    ALWAYS = -1,
+    INFO,
+    ALLOC,
+    VERBOSE
+};
+
+// Default is to print allocations
+static int LogLevel = ALLOC;
+
+// Allocations of size greater than low limit promoted to HBW memory.
+// If there is a high limit specified, allocations larger than this limit
+// will not be allocated in HBW.
+static size_t HBWLowLimit = 1 * 1024 * 1024;
+static size_t HBWHighLimit = -1ull;
+
+// Whether we have initialized HBW arena of memkind library -- by making
+// a dummy call to it. HBW arena (and hence any memkind_* call with kind
+// HBW) must NOT be used until this flag is set true.
+static bool MemkindInitDone = false;
+
+// Following is the type of HBW memory that is allocated using memkind.
+// By changing this type, this library can be used to allocate other
+// types of memory types (e.g., MEMKIND_HUGETLB, MEMKIND_GBTLB,
+// MEMKIND_HBW_HUGETLB etc.)
+static memkind_t hbw_kind;
+
+// API control for HBW allocations.
+static bool isAutoHBWEnabled = true;
+
+#define LOG(level, ...)                                                                            \
+    do {                                                                                           \
+        if (LogLevel >= level) {                                                                   \
+            fprintf(stderr, __VA_ARGS__);                                                          \
+        }                                                                                          \
+    } while (0)
+
+static bool isAllocInHBW(size_t size)
+{
+    if (!MemkindInitDone)
+        return false;
+
+    if (!isAutoHBWEnabled)
+        return false;
+
+    if (size < HBWLowLimit)
+        return false;
+
+    if (size > HBWHighLimit)
+        return false;
+
+    return true;
+}
+
+// Returns the limit in bytes using a limit value and a multiplier
+// character like K, M, G
+static size_t getLimit(size_t limit, char lchar)
+{
+    // Now read the trailing character (e.g., K, M, G)
+    // Based on the character, determine the multiplier
+    if ((limit > 0) && isalpha(lchar)) {
+        long mult = 1;
+
+        switch (toupper(lchar)) {
+            case 'G':
+                mult *= 1024;
+            case 'M':
+                mult *= 1024;
+            case 'K':
+                mult *= 1024;
+        }
+
+        // check for overflow, saturate at max
+        if (limit >= -1ull / mult)
+            return -1ull;
+
+        return limit * mult;
+    }
+
+    return limit;
+}
+
+// Once HBWLowLimit (and HBWHighLimit) are set, call this method to
+// inform the user about the size range of arrays that will be allocated
+// in HBW
+static void printLimits()
+{
+    // Inform according to the limits set
+    if ((HBWLowLimit > 0) && (HBWHighLimit < -1ull)) {
+        // if both high and low limits are specified, we use a range
+        LOG(INFO, "INFO: Allocations between %ldK - %ldK will be allocated in "
+            "HBW. Set AUTO_HBW_SIZE=X:Y to change this limit.\n",
+            HBWLowLimit / 1024, HBWHighLimit / 1024);
+    } else if (HBWLowLimit > 0) {
+        // if only a low limit is provided, use that
+        LOG(INFO, "INFO: Allocations greater than %ldK will be allocated in HBW."
+            " Set AUTO_HBW_SIZE=X:Y to change this limit.\n",
+            HBWLowLimit / 1024);
+    } else if (HBWHighLimit < -1ull) {
+        // if only a high limit is provided, use that
+        LOG(INFO, "INFO: Allocations smaller than %ldK will be allocated in HBW. "
+            "Set AUTO_HBW_SIZE=X:Y to change this limit.\n",
+            HBWHighLimit / 1024);
+    } else {
+        // none of limits is set to non-edge value, everything goes to HBW
+        LOG(INFO, "INFO: All allocation will be done in HBW.");
+    }
+}
+
+struct kind_name_t {
+    memkind_t *kind;
+    const char *name;
+};
+
+static struct kind_name_t named_kinds[] = {
+    { &MEMKIND_DEFAULT, "memkind_default" },
+    { &MEMKIND_HUGETLB, "memkind_hugetlb" },
+    { &MEMKIND_INTERLEAVE, "memkind_interleave" },
+    { &MEMKIND_HBW, "memkind_hbw" },
+    { &MEMKIND_HBW_PREFERRED, "memkind_hbw_preferred" },
+    { &MEMKIND_HBW_HUGETLB, "memkind_hbw_hugetlb" },
+    { &MEMKIND_HBW_PREFERRED_HUGETLB, "memkind_hbw_preferred_hugetlb" },
+    { &MEMKIND_HBW_GBTLB, "memkind_hbw_gbtlb" },
+    { &MEMKIND_HBW_PREFERRED_GBTLB, "memkind_hbw_preferred_gbtlb" },
+    { &MEMKIND_GBTLB, "memkind_gbtlb" },
+    { &MEMKIND_HBW_INTERLEAVE, "memkind_hbw_interleave" },
+};
+
+static memkind_t get_kind_by_name(const char *name)
+{
+    int i;
+    for (i = 0; i < sizeof(named_kinds) / sizeof(named_kinds[0]); ++i)
+        if (strcasecmp(named_kinds[i].name, name) == 0)
+            return *named_kinds[i].kind;
+
+    return 0;
+}
+
+// Read from the environment and sets global variables
+// Env variables are:
+//   AUTO_HBW_SIZE = gives the size for auto HBW allocation
+//   AUTO_HBW_LOG = gives logging level
+static void setEnvValues()
+{
+    // STEP: Read the log level from the env variable. Do this early because
+    //       printing depends on this
+    char *log_str = secure_getenv("AUTO_HBW_LOG");
+    if (log_str && strlen(log_str)) {
+        int level = atoi(log_str);
+        LogLevel = level;
+        LOG(ALWAYS, "INFO: Setting log level to %d\n", LogLevel);
+    }
+
+    if (LogLevel == INFO) {
+        LOG(INFO, "INFO: HBW allocation stats will not be printed. "
+            "Set AUTO_HBW_LOG to enable.\n");
+    } else if (LogLevel == ALLOC) {
+        LOG(INFO, "INFO: Only HBW allocations will be printed. "
+            "Set AUTO_HBW_LOG to disable/enable.\n");
+    } else if (LogLevel == VERBOSE) {
+        LOG(INFO, "INFO: HBW allocation with backtrace info will be printed. "
+            "Set AUTO_HBW_LOG to disable.\n");
+    }
+
+    // Set the memory type allocated by this library. By default, it is
+    // MEMKIND_HBW, but we can use this library to allocate other memory
+    // types
+    const char *memtype_str = secure_getenv("AUTO_HBW_MEM_TYPE");
+    if (memtype_str && strlen(memtype_str)) {
+        // Find the memkind_t using the name the user has provided in the env variable
+        memkind_t mty = get_kind_by_name(memtype_str);
+        if (mty != 0) {
+            hbw_kind = mty;
+            LOG(INFO, "INFO: Setting HBW memory type to %s\n", memtype_str);
+        } else {
+            LOG(ALWAYS, "WARN: Memory type %s not recognized. Using default type\n",
+                memtype_str);
+        }
+    }
+
+    // STEP: Set the size limits (thresholds) for HBW allocation
+    //
+    // Reads the environment variable
+    const char *size_str = secure_getenv("AUTO_HBW_SIZE");
+    if (size_str) {
+        size_t lowlim = HBWLowLimit / 1024;
+        size_t highlim = HBWHighLimit / 1024;
+        char lowC = 'K', highC = 'K';
+
+        if (size_str) {
+            char *ptr = (char *)size_str;
+            lowlim = strtoll(ptr, &ptr, 10);
+            if (*ptr != 0 && *ptr != ':')
+                lowC = *ptr++;
+            else
+                lowC = ' ';
+
+            if (*ptr++ == ':') {
+                highlim = strtoll(ptr, &ptr, 10);
+                if (*ptr)
+                    highC = *ptr;
+                else
+                    highC = ' ';
+            }
+
+            LOG(INFO, "INFO: lowlim=%zu(%c), highlim=%zu(%c)\n", lowlim, lowC, highlim,
+                highC);
+        }
+
+        HBWLowLimit = getLimit(lowlim, lowC);
+        HBWHighLimit = getLimit(highlim, highC);
+
+        if (HBWLowLimit >= HBWHighLimit) {
+            LOG(ALWAYS, "WARN: In AUTO_HBW_SIZE=X:Y, X cannot be greater or equal to Y. "
+                "None of allocations will use HBW memory.\n");
+        }
+    } else {
+        // if the user did not specify any limits, inform that we are using
+        // default limits
+        LOG(INFO, "INFO: Using default values for array size thresholds. "
+            "Set AUTO_HBW_SIZE=X:Y to change.\n");
+    }
+
+    // inform the user about limits
+    printLimits();
+}
+
+// This function is executed at library load time.
+// Initialize HBW arena by making a dummy allocation/free at library load
+// time. Until HBW initialization is complete, we must not call any
+// allocation routines with HBW as kind.
+static void AUTOHBW_INIT autohbw_load(void)
+{
+    // First set the default memory type this library allocates. This can
+    // be overridden by env variable
+    // Note: 'memkind_hbw_preferred' will allow falling back to DDR but
+    //       'memkind_hbw will not'
+    // Note: If HBM is not installed on a system, memkind_hbw_preferred call
+    //       would fail. Therefore, we need to check for availability first.
+    if (memkind_check_available(MEMKIND_HBW) != 0) {
+        LOG(ALWAYS, "WARN: *** No HBM found in system. Will use default (DDR) "
+            "OR user specified type ***\n");
+        hbw_kind = MEMKIND_DEFAULT;
+    } else {
+        hbw_kind = MEMKIND_HBW_PREFERRED;
+    }
+
+    // Read any env variables. This has to be done first because DbgLevel
+    // is set using env variables and debug printing is used below
+    setEnvValues(); // read any env variables
+
+    LOG(INFO, "INFO: autohbw.so loaded!\n");
+
+    // dummy HBW call to initialize HBW arena
+    void *pp = memkind_malloc(hbw_kind, 16);
+    if (pp == 0) {
+        LOG(ALWAYS, "\t-HBW init call FAILED. "
+            "Is required memory type present on your system?\n");
+        abort();
+    }
+
+    LOG(ALWAYS, "\t-HBW int call succeeded\n");
+    memkind_free(hbw_kind, pp);
+
+    MemkindInitDone = true; // enable HBW allocation
+}
+
+static void *MemkindMalloc(size_t size)
+{
+    LOG(VERBOSE, "In my memkind malloc sz:%ld ... ", size);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    void *ptr = memkind_malloc(kind, size);
+
+    LOG(VERBOSE, "\tptr:%p\n", ptr);
+    return ptr;
+}
+
+static void *MemkindCalloc(size_t nmemb, size_t size)
+{
+    LOG(VERBOSE, "In my memkind calloc sz:%ld ..", size * nmemb);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    void *ptr = memkind_calloc(kind, nmemb, size);
+
+    LOG(VERBOSE, "\tptr:%p\n", ptr);
+    return ptr;
+}
+
+static void *MemkindRealloc(void *ptr, size_t size)
+{
+    LOG(VERBOSE, "In my memkind realloc sz:%ld, p1:%p ..", size, ptr);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    void *nptr = memkind_realloc(kind, ptr, size);
+
+    LOG(VERBOSE, "\tptr=%p\n", nptr);
+    return nptr;
+}
+
+static int MemkindAlign(void **memptr, size_t alignment, size_t size)
+{
+    LOG(VERBOSE, "In my memkind align sz:%ld .. ", size);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    int ret = memkind_posix_memalign(kind, memptr, alignment, size);
+
+    LOG(VERBOSE, "\tptr:%p\n", *memptr);
+    return ret;
+}
+
+// memkind_free does not need the exact kind, if kind is 0. Then
+// the library can figure out the proper kind itself.
+static void MemkindFree(void *ptr)
+{
+    // avoid to many useless logs
+    if (ptr)
+        LOG(VERBOSE, "In my memkind free, ptr:%p\n", ptr);
+    memkind_free(0, ptr);
+}
+
+//--------------------------------------------------------------------------
+// ------------------ Public API of autohbw           ----------------------
+//--------------------------------------------------------------------------
+
+AUTOHBW_EXPORT void enableAutoHBW()
+{
+    isAutoHBWEnabled = true;
+    LOG(INFO, "INFO: HBW allocations enabled by application (for this rank)\n");
+}
+
+AUTOHBW_EXPORT void disableAutoHBW()
+{
+    isAutoHBWEnabled = false;
+    LOG(INFO, "INFO: HBW allocations disabled by application (for this rank)\n");
+}
+
+AUTOHBW_EXPORT void *malloc(size_t size)
+{
+    return MemkindMalloc(size);
+}
+
+AUTOHBW_EXPORT void *calloc(size_t nmemb, size_t size)
+{
+    return MemkindCalloc(nmemb, size);
+}
+
+AUTOHBW_EXPORT void *realloc(void *ptr, size_t size)
+{
+    return MemkindRealloc(ptr, size);
+}
+
+AUTOHBW_EXPORT int posix_memalign(void **memptr, size_t alignment, size_t size)
+{
+    return MemkindAlign(memptr, alignment, size);
+}
+
+// Warn about deprecated function usage.
+AUTOHBW_EXPORT void *valloc(size_t size)
+{
+    LOG(ALWAYS, "use of deprecated valloc. Use posix_memalign instead\n");
+    void *memptr = 0;
+    size_t boundary = sysconf(_SC_PAGESIZE);
+    int status = MemkindAlign(&memptr, boundary, size);
+    if (status == 0 && memptr != 0)
+        return memptr;
+
+    return 0;
+}
+
+// Warn about deprecated function usage.
+AUTOHBW_EXPORT void *memalign(size_t boundary, size_t size)
+{
+    LOG(ALWAYS, "use of deprecated memalign. Use posix_memalign instead\n");
+    void *memptr = 0;
+    int status = MemkindAlign(&memptr, boundary, size);
+    if (status == 0 && memptr != 0)
+        return memptr;
+
+    return 0;
+}
+
+AUTOHBW_EXPORT void free(void *ptr)
+{
+    return MemkindFree(ptr);
+}
diff --git a/autohbw/autohbw_README b/autohbw/autohbw_README
new file mode 100644
index 0000000..cf23a0f
--- /dev/null
+++ b/autohbw/autohbw_README
@@ -0,0 +1,92 @@
+README for auothbw library
+Ruchira Sasanka (ruchira DOT sasanka AT intel DOT com)
+2015 April 8
+
+PURPOSE
+-------
+The autohbw library is used to automatically allocate high-bandwidth (HBW)
+memory without any modifications to source code of your application.
+This library intercepts the standard heap allocations (e.g., malloc, calloc)
+in your program so that it can serve those allocations out of HBW memory.
+
+There are two libraries installed at the same location as memkind library:
+  1) libautohbw.so -- dynamic library
+  2) libautohbw.a  -- static library (provided for static linking)
+
+USAGE
+-----
+To use the dynamic library, insert LD_PRELOAD=libautohbw.so and libmekind.so
+before your commandline. For instance,
+
+   LD_PRELOAD=libautohbw.so /bin/ls
+
+executes Unix ls utility with autohbw library. You can execute the
+above command on the prompt or put it in a shell script. An example of
+the latter approach is given in ./autohbw_test.sh. Make sure that
+LD_LIBRARY_PATH environment variable contains the path to libautohbw.so
+and libmekind.so before you exeute the above command.
+
+To use the statically linked library, insert -lautohbw -lmemkind
+on your link line. You will have to provide the library path with -L to
+point to the above libraries. If you are linking your program statically
+(e.g., with -static or -fast flags in ifort/icc), you must use the static
+version of autohbw library (libautohbw.a).
+
+It is possible to temporarily disable/enable automatic HBM allocations by
+calling disableAutoHBW() and enableAutoHBW() in your source code. To call
+these routines, please include autohbw_api.h header file and link with
+-lautohbw.
+
+
+Usage with MPI programs
+-----------------------
+To use with MPI programs, you need to create a shell script to do the
+LD_PRELOAD. E.g.,
+   mpirun -n 2 ./autohbw_test.sh
+
+ENVIRONMENT VARIABLES
+---------------------
+The following environment variables control the behavior of the autohbw
+library:
+
+  AUTO_HBW_SIZE=x[:y]
+  Indicates that any allocation larger than x and smaller than y should be
+  allocated in HBW memory. x and y can be followed by a K, M, or G to
+  indicate the size in Kilo/Memga/Giga bytes (e.g., 4K, 3M, 2G).
+  Examples:
+    AUTO_HBW_SIZE=4K     # all allocation larger than 4K allocated in HBW
+    AUTO_HBW_SIZE=1M:5M  # allocations between 1M and 5M allocated in HBW
+
+
+  AUTO_HBW_LOG=level
+  Sets the value of logging (printing) level. If level is:
+   -1 = no messages are printed
+    0 = no log messages for allocations are printed but INFO messages
+        are printed
+    1 = a log message is printed for each allocation (Default)
+    2 = a log message is printed for each allocation with a backtrace
+        Redirect this output and use autohbw_get_src_lines.pl to find
+        source lines for each allocation. Your application must be compiled
+        with -g to see source lines.
+  Notes:
+    1. Logging adds extra overhead. Therefore, performance critical runs,
+       logging level should be 0
+    2. The amount of free memory printed with log messages is only
+       approximate -- e.g. pages that are not touched yet are excluded
+  Examples:
+    AUTO_HBW_LOG=1
+
+
+  AUTO_HBW_MEM_TYPE=memory_type
+  Sets the type of memory type that should be automatically allocated. By
+  default, this type is MEMKIND_HBW_PREFERRED, if MCDRAM is found in your
+  system; otherwise, the default is MEMKIND_DEFAULT. The names of memory
+  types are definedin memkind(3) man page. If you are requesting any huge
+  TLB pages, pleasemake sure that the requested type is currently enabled
+  in your OS.
+  Examples:
+    AUTO_HBW_MEM_TYPE=MEMKIND_HBW_PREFERRED   (default, if MCDRAM present)
+    AUTO_HBW_MEM_TYPE=MEMKIND_DEFAULT         (default, if MCDRAM absent)
+    AUTO_HBW_MEM_TYPE=MEMKIND_HBW_HUGETLB
+    AUTO_HBW_MEM_TYPE=MEMKIND_HUGETLB
+
diff --git a/autohbw/autohbw_api.h b/autohbw/autohbw_api.h
new file mode 100644
index 0000000..45f666a
--- /dev/null
+++ b/autohbw/autohbw_api.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (C) 2015 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////
+// File   : autohbw_api.h
+// Purpose: Header file to include, to use API calls to AutoHBW
+// Author : Ruchira Sasanka (ruchira.sasanka AT intel.com)
+// Date   : Jan 30, 2015
+///////////////////////////////////////////////////////////////////////////
+
+#ifndef AUTOHBW_API_H
+#define AUTOHBW_API_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// Temporarily enable HBM allocations
+void enableAutoHBW();
+
+// Temporarily disable HBM allocations
+void disableAutoHBW();
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/autohbw/autohbw_get_src_lines.pl b/autohbw/autohbw_get_src_lines.pl
new file mode 100644
index 0000000..81ccb67
--- /dev/null
+++ b/autohbw/autohbw_get_src_lines.pl
@@ -0,0 +1,106 @@
+#!/usr/bin/perl
+#
+#  Copyright (C) 2015 - 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+use strict;
+
+my $usage = "Usage: get_autohbw_srclines.pl output_log_of_AutoHBW executable";
+
+# Check for 2 arguments
+#
+if (@ARGV != 2) {
+    print $usage, "\n";
+    exit;
+}
+
+# Read the command line arguments
+#
+my $LogF = shift @ARGV;
+my $BinaryF = shift @ARGV;
+
+&main();
+
+sub main {
+
+
+    print("Info: Reading AutoHBW log from: $LogF\n");
+    print("Info: Binary file: $BinaryF\n");
+
+    # open the log file produced by AutoHBM and look at lines starting
+    # with Log
+    open LOGF, "grep Log $LogF |" or die "Can't open log file $LogF";
+
+    my $line;
+
+    # Read each log line
+    #
+    while ($line = <LOGF>) {
+
+        # if the line contain 3 backtrace addresses, try to find the source
+        # lines for them
+        #
+        if ($line =~ /^(Log:.*)Backtrace:.*0x([0-9a-f]+).*0x([0-9a-f]+).*0x([0-9a-f]+)/ ) {
+
+            #  Read the pointers
+            #
+            my @ptrs;
+
+            my $pre = $1;
+
+            $ptrs[0] = $2;
+            $ptrs[1] = $3;
+            $ptrs[2] = $4;
+
+            # prints the first portion of the line
+            #
+            print $pre, "\n";
+
+            # for each of the pointers, lookup its source line using
+            # addr2line and print the src line(s) if found
+            #
+            my $i=0;
+            for ($i=0; $i < @ptrs; $i++) {
+
+                my $addr = $ptrs[$i];
+
+                open SRCL, "addr2line -e $BinaryF 0x$addr |"
+                    or die "addr2line fail";
+
+
+                my $srcl = <SRCL>;
+
+                if ($srcl =~ /^\?/) {
+
+                } else {
+
+                    print "\t- Src: $srcl";
+                }
+
+            }
+
+        }
+
+    }
+
+}
diff --git a/autohbw/autohbw_test.sh b/autohbw/autohbw_test.sh
new file mode 100755
index 0000000..52cf313
--- /dev/null
+++ b/autohbw/autohbw_test.sh
@@ -0,0 +1,37 @@
+#!/bin/bash
+
+#  Copyright (C) 2015 - 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# The following is needed if your system does not really have HBW nodes.
+# Please see 'man memkind' for a description of this variable
+#
+# export MEMKIND_HBW_NODES=0
+
+# Set any autohbw env variables
+#
+export AUTO_HBW_SIZE=4K
+
+# Make sure you have set LD_LIBRARY_PATH to lib directory with libautohbw.so
+# and libmemkind.so
+#
+LD_PRELOAD=libautohbw.so:libmemkind.so /bin/ls
diff --git a/build.sh b/build.sh
new file mode 100755
index 0000000..6dfd578
--- /dev/null
+++ b/build.sh
@@ -0,0 +1,35 @@
+#!/bin/bash
+#
+#  Copyright (C) 2016 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+set -e
+
+cd $(dirname $0)
+EXTRA_CONF=$@
+
+./autogen.sh
+./configure $EXTRA_CONF
+
+#use V=1 for full cmdlines of build
+make all -j`nproc`
+make checkprogs -j`nproc`
diff --git a/configure.ac b/configure.ac
new file mode 100644
index 0000000..95116a1
--- /dev/null
+++ b/configure.ac
@@ -0,0 +1,234 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+#                                               -*- Autoconf -*-
+# Process this file with autoconf to produce a configure script.
+
+AC_PREREQ([2.63])
+AC_INIT([memkind],m4_esyscmd([tr -d '\n' < VERSION]))
+
+AC_CONFIG_MACRO_DIR([m4])
+AC_CONFIG_HEADERS([config.h])
+AC_CONFIG_SRCDIR([memkind.spec.mk])
+
+AM_INIT_AUTOMAKE([-Wall -Werror foreign 1.11 silent-rules subdir-objects parallel-tests tar-pax])
+AM_SILENT_RULES([yes])
+
+# Checks for programs and libraries.
+m4_ifdef([AM_PROG_AR], [AM_PROG_AR])
+AC_PROG_CXX
+AC_PROG_CC
+AC_OPENMP
+AC_CHECK_LIB(numa, numa_available, [], [AC_MSG_ERROR([libnuma is required dependency])])
+AX_PTHREAD([LIBS="$PTHREAD_LIBS $LIBS" CFLAGS="$CFLAGS $PTHREAD_CFLAGS" CC="$PTHREAD_CC"],
+           [AC_MSG_ERROR([pthreads are required dependency])])
+
+AM_PROG_CC_C_O
+
+#============================tls===============================================
+# Check for thread local storage support
+AC_ARG_ENABLE([tls],
+  [AS_HELP_STRING([--enable-tls], [Enable thread-local storage (__thread keyword)])],
+[if test "x$enable_tls" = "xyes" ; then
+  enable_tls="1"
+else
+  enable_tls="0"
+fi
+],
+[enable_tls="0"]
+)
+if test "x${enable_tls}" = "x1" ; then
+AC_MSG_CHECKING([for TLS support])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+    __thread int x;
+]], [[
+    x = 1234;
+]])],
+    AC_MSG_RESULT([yes]),
+    AC_MSG_RESULT([no])
+    enable_tls="0")
+fi
+if test "x${enable_tls}" = "x1" ; then
+  AC_DEFINE([MEMKIND_TLS], [ ], [Enables TLS usage for mapping arenas to threads])
+fi
+AC_SUBST([enable_tls])
+
+#============================decorators========================================
+AC_ARG_ENABLE([decorators],
+  [AS_HELP_STRING([--enable-decorators], [Enable decorators])],
+[if test "x$enable_decorators" = "xyes" ; then
+  enable_decorators="1"
+else
+  enable_decorators="0"
+fi
+],
+[enable_decorators="0"]
+)
+if test "x${enable_decorators}" = "x1" ; then
+  AC_DEFINE([MEMKIND_DECORATION_ENABLED], [ ], [Enables decorators])
+fi
+AC_SUBST([enable_decorators])
+
+#============================debug=============================================
+AC_ARG_ENABLE([debug],
+  [AS_HELP_STRING([--enable-debug], [Build debugging code and compile with -O0 -g])],
+[if test "x$enable_debug" = "xno" ; then
+  enable_debug="0"
+else
+  enable_debug="1"
+fi
+],
+[enable_debug="0"]
+)
+if test "x$enable_debug" = "x1" ; then
+  AC_DEFINE([MEMKIND_DEBUG], [ ], [Enables code for debugging])
+  CFLAGS="$CFLAGS -O0 -g"
+  CXXFLAGS="$CXXFLAGS -O0 -g"
+fi
+AC_SUBST([enable_debug])
+
+#============================gcov==============================================
+AC_ARG_ENABLE([gcov],
+  [AS_HELP_STRING([--enable-gcov], [Build code with gcov instructions])],
+[if test "x$enable_gcov" = "xno" ; then
+  enable_gcov="0"
+else
+  enable_gcov="1"
+fi
+],
+[enable_gcov="0"]
+)
+if test "x$enable_gcov" = "x1" ; then
+  CFLAGS="$CFLAGS -O0 -fprofile-arcs -ftest-coverage"
+  CXXFLAGS="$CXXFLAGS -O0 -fprofile-arcs -ftest-coverage"
+fi
+AC_SUBST([enable_gcov])
+
+#============================secure_flags======================================
+AC_ARG_ENABLE([secure],
+  [AS_HELP_STRING([--enable-secure], [Build library with security enchantments])],
+[if test "x$enable_secure" = "xno" ; then
+  enable_secure="0"
+else
+  enable_secure="1"
+fi
+],
+[enable_secure="1"]
+)
+if test "x$enable_secure" = "x1" ; then
+  CFLAGS="$CFLAGS -fstack-protector"
+  LDFLAGS="$LDFLAGS -Wl,-z,relro,-z,now"
+
+  if test "$CFLAGS" != "${CFLAGS%-O0*}" ; then # if CFLAGS contains -O0
+      echo "WARNING: Could not apply FORTIFY_SOURCE=2 due to lack of optimization (-O0)"
+  else
+      CFLAGS="$CFLAGS -D_FORTIFY_SOURCE=2" #FORTITFY_SOURCE does not work with -O0 (ex. if enable_debug=1 or enable_gcov=1)
+  fi
+fi
+
+AC_SUBST([enable_secure])
+
+#============================tcache_max_size_class======================
+
+tcache_max_size_class=12;
+AC_DEFINE_UNQUOTED([JEMALLOC_TCACHE_CLASS], $tcache_max_size_class,
+  [Maximum size class (log base 2) to cache in jemalloc tcache])
+
+AC_SUBST(tcache_max_size_class)
+
+#============================auto_arenas================================
+auto_arenas=256;
+AC_DEFINE_UNQUOTED([ARENA_LIMIT_DEFAULT_KIND], $auto_arenas,
+  [Maximum number of automatic managed arenas by jemalloc (used by MEMKIND_DEFAULT)])
+
+AC_SUBST(auto_arenas)
+
+#============================arena_limit=======================================
+arena_limit=256;
+AC_ARG_VAR(ARENA_LIMIT,
+  [Upper bound for number of arenas per kind, if set to 0 then no limit]
+)
+if test "$ARENA_LIMIT" != "" ; then
+  arena_limit=$ARENA_LIMIT;
+fi
+
+AC_DEFINE_UNQUOTED([ARENA_LIMIT_PER_KIND], $arena_limit, [Upper bound for number of arenas per kind])
+
+#============================memkind_prefix=======================================
+memkind_prefix=jemk_;
+AC_ARG_VAR(MEMKIND_PREFIX,
+  [Prefix used for all jemalloc public API under memkind, default value jemk_]
+)
+if test "$MEMKIND_PREFIX" != "" ; then
+  memkind_prefix=$MEMKIND_PREFIX;
+fi
+
+AC_SUBST(memkind_prefix)
+
+#===============================daxctl=========================================
+AC_ARG_ENABLE([daxctl],
+  [AS_HELP_STRING([--enable-daxctl], [Build library with daxctl support])],
+[if test "x$enable_daxctl" = "xno" ; then
+  enable_daxctl="0"
+else
+  enable_daxctl="1"
+fi
+],
+[enable_daxctl="1"]
+)
+if test "x$enable_daxctl" = "x1" ; then
+    AC_CHECK_LIB(daxctl, daxctl_dev_get_memory, [daxctl_kmem=yes], [daxctl_kmem=no])
+fi
+if test $daxctl_kmem = "yes" ; then
+    AC_DEFINE([MEMKIND_DAXCTL_KMEM], [1], [Automatic recognition of PMEM as system memory in MEMKIND_DAX_KMEM])
+    AC_SUBST(DAXCTL_LIBS, [-ldaxctl])
+    LDFLAGS="$LDFLAGS -ldaxctl"
+fi
+AC_SUBST([enable_daxctl])
+
+#============================cxx11=============================================
+
+AX_CXX_COMPILE_STDCXX_11([noext], [optional])
+AM_CONDITIONAL([HAVE_CXX11], [test "x$HAVE_CXX11" = x1])
+
+LT_PREREQ([2.2])
+LT_INIT
+
+AC_CONFIG_FILES([Makefile memkind.pc])
+
+AC_OUTPUT
+AC_MSG_RESULT([=================================================================================])
+AC_MSG_RESULT([
+               Memkind version $VERSION])
+if test $daxctl_kmem = "yes" ; then
+AC_MSG_RESULT([
+               Automatic recognition of PMEM NUMA in MEMKIND_DAX_KMEM is enabled])
+else
+AC_MSG_RESULT([
+               Automatic recognition of PMEM NUMA in MEMKIND_DAX_KMEM is disabled
+               To enable this feature please install libdaxctl-devel v66 or later
+               and run configure again])
+fi
+AC_MSG_RESULT([=================================================================================])
diff --git a/copying_headers/MANIFEST.EXEMPT b/copying_headers/MANIFEST.EXEMPT
new file mode 100644
index 0000000..0e6326d
--- /dev/null
+++ b/copying_headers/MANIFEST.EXEMPT
@@ -0,0 +1,347 @@
+.codecov.yml
+.gitignore
+.travis.yml
+AUTHORS
+CONTRIBUTING
+COPYING
+ChangeLog
+NEWS
+PULL_REQUEST_TEMPLATE.md
+README
+VERSION
+autohbw/autohbw_README
+examples/README
+include/memkind/internal/tbb_mem_pool_policy.h
+jemalloc/.appveyor.yml
+jemalloc/.autom4te.cfg
+jemalloc/.gitattributes
+jemalloc/.gitignore
+jemalloc/.travis.yml
+jemalloc/autogen.sh
+jemalloc/bin/jemalloc-config.in
+jemalloc/bin/jemalloc.sh.in
+jemalloc/bin/jeprof.in
+jemalloc/build-aux/config.guess
+jemalloc/build-aux/config.sub
+jemalloc/build-aux/install-sh
+jemalloc/ChangeLog
+jemalloc/config.stamp.in
+jemalloc/configure.ac
+jemalloc/COPYING
+jemalloc/doc/html.xsl.in
+jemalloc/doc/jemalloc.xml.in
+jemalloc/doc/manpages.xsl.in
+jemalloc/doc/stylesheet.xsl
+jemalloc/include/jemalloc/internal/arena_externs.h
+jemalloc/include/jemalloc/internal/arena_inlines_a.h
+jemalloc/include/jemalloc/internal/arena_inlines_b.h
+jemalloc/include/jemalloc/internal/arena_stats.h
+jemalloc/include/jemalloc/internal/arena_structs_a.h
+jemalloc/include/jemalloc/internal/arena_structs_b.h
+jemalloc/include/jemalloc/internal/arena_types.h
+jemalloc/include/jemalloc/internal/assert.h
+jemalloc/include/jemalloc/internal/atomic.h
+jemalloc/include/jemalloc/internal/atomic_c11.h
+jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
+jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
+jemalloc/include/jemalloc/internal/atomic_msvc.h
+jemalloc/include/jemalloc/internal/background_thread_externs.h
+jemalloc/include/jemalloc/internal/background_thread_inlines.h
+jemalloc/include/jemalloc/internal/background_thread_structs.h
+jemalloc/include/jemalloc/internal/base_externs.h
+jemalloc/include/jemalloc/internal/base_inlines.h
+jemalloc/include/jemalloc/internal/base_structs.h
+jemalloc/include/jemalloc/internal/base_types.h
+jemalloc/include/jemalloc/internal/bin.h
+jemalloc/include/jemalloc/internal/bin_stats.h
+jemalloc/include/jemalloc/internal/bin_types.h
+jemalloc/include/jemalloc/internal/bitmap.h
+jemalloc/include/jemalloc/internal/bit_util.h
+jemalloc/include/jemalloc/internal/cache_bin.h
+jemalloc/include/jemalloc/internal/ckh.h
+jemalloc/include/jemalloc/internal/ctl.h
+jemalloc/include/jemalloc/internal/div.h
+jemalloc/include/jemalloc/internal/emitter.h
+jemalloc/include/jemalloc/internal/extent_dss.h
+jemalloc/include/jemalloc/internal/extent_externs.h
+jemalloc/include/jemalloc/internal/extent_inlines.h
+jemalloc/include/jemalloc/internal/extent_mmap.h
+jemalloc/include/jemalloc/internal/extent_structs.h
+jemalloc/include/jemalloc/internal/extent_types.h
+jemalloc/include/jemalloc/internal/hash.h
+jemalloc/include/jemalloc/internal/hook.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
+jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
+jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
+jemalloc/include/jemalloc/internal/large_externs.h
+jemalloc/include/jemalloc/internal/log.h
+jemalloc/include/jemalloc/internal/malloc_io.h
+jemalloc/include/jemalloc/internal/mutex.h
+jemalloc/include/jemalloc/internal/mutex_pool.h
+jemalloc/include/jemalloc/internal/mutex_prof.h
+jemalloc/include/jemalloc/internal/nstime.h
+jemalloc/include/jemalloc/internal/pages.h
+jemalloc/include/jemalloc/internal/ph.h
+jemalloc/include/jemalloc/internal/private_namespace.sh
+jemalloc/include/jemalloc/internal/private_symbols.sh
+jemalloc/include/jemalloc/internal/prng.h
+jemalloc/include/jemalloc/internal/prof_externs.h
+jemalloc/include/jemalloc/internal/prof_inlines_a.h
+jemalloc/include/jemalloc/internal/prof_inlines_b.h
+jemalloc/include/jemalloc/internal/prof_structs.h
+jemalloc/include/jemalloc/internal/prof_types.h
+jemalloc/include/jemalloc/internal/public_namespace.sh
+jemalloc/include/jemalloc/internal/public_unnamespace.sh
+jemalloc/include/jemalloc/internal/ql.h
+jemalloc/include/jemalloc/internal/qr.h
+jemalloc/include/jemalloc/internal/quantum.h
+jemalloc/include/jemalloc/internal/rb.h
+jemalloc/include/jemalloc/internal/rtree.h
+jemalloc/include/jemalloc/internal/rtree_tsd.h
+jemalloc/include/jemalloc/internal/safety_check.h
+jemalloc/include/jemalloc/internal/sc.h
+jemalloc/include/jemalloc/internal/seq.h
+jemalloc/include/jemalloc/internal/smoothstep.h
+jemalloc/include/jemalloc/internal/smoothstep.sh
+jemalloc/include/jemalloc/internal/spin.h
+jemalloc/include/jemalloc/internal/stats.h
+jemalloc/include/jemalloc/internal/sz.h
+jemalloc/include/jemalloc/internal/tcache_externs.h
+jemalloc/include/jemalloc/internal/tcache_inlines.h
+jemalloc/include/jemalloc/internal/tcache_structs.h
+jemalloc/include/jemalloc/internal/tcache_types.h
+jemalloc/include/jemalloc/internal/test_hooks.h
+jemalloc/include/jemalloc/internal/ticker.h
+jemalloc/include/jemalloc/internal/tsd.h
+jemalloc/include/jemalloc/internal/tsd_generic.h
+jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
+jemalloc/include/jemalloc/internal/tsd_tls.h
+jemalloc/include/jemalloc/internal/tsd_types.h
+jemalloc/include/jemalloc/internal/tsd_win.h
+jemalloc/include/jemalloc/internal/util.h
+jemalloc/include/jemalloc/internal/witness.h
+jemalloc/include/jemalloc/jemalloc.sh
+jemalloc/include/jemalloc/jemalloc_defs.h.in
+jemalloc/include/jemalloc/jemalloc_macros.h.in
+jemalloc/include/jemalloc/jemalloc_mangle.sh
+jemalloc/include/jemalloc/jemalloc_protos.h.in
+jemalloc/include/jemalloc/jemalloc_rename.sh
+jemalloc/include/jemalloc/jemalloc_typedefs.h.in
+jemalloc/include/msvc_compat/C99/stdbool.h
+jemalloc/include/msvc_compat/C99/stdint.h
+jemalloc/include/msvc_compat/strings.h
+jemalloc/include/msvc_compat/windows_extra.h
+jemalloc/INSTALL.md
+jemalloc/jemalloc.pc.in
+jemalloc/m4/ax_cxx_compile_stdcxx.m4
+jemalloc/Makefile.in
+jemalloc/msvc/jemalloc_vc2015.sln
+jemalloc/msvc/jemalloc_vc2017.sln
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/ReadMe.txt
+jemalloc/msvc/test_threads/test_threads.cpp
+jemalloc/msvc/test_threads/test_threads.h
+jemalloc/msvc/test_threads/test_threads_main.cpp
+jemalloc/README
+jemalloc/run_tests.sh
+jemalloc/scripts/gen_run_tests.py
+jemalloc/scripts/gen_travis.py
+jemalloc/src/arena.c
+jemalloc/src/background_thread.c
+jemalloc/src/base.c
+jemalloc/src/bin.c
+jemalloc/src/bitmap.c
+jemalloc/src/ckh.c
+jemalloc/src/ctl.c
+jemalloc/src/div.c
+jemalloc/src/extent.c
+jemalloc/src/extent_dss.c
+jemalloc/src/extent_mmap.c
+jemalloc/src/hash.c
+jemalloc/src/hook.c
+jemalloc/src/jemalloc.c
+jemalloc/src/jemalloc_cpp.cpp
+jemalloc/src/large.c
+jemalloc/src/log.c
+jemalloc/src/malloc_io.c
+jemalloc/src/mutex.c
+jemalloc/src/mutex_pool.c
+jemalloc/src/nstime.c
+jemalloc/src/pages.c
+jemalloc/src/prng.c
+jemalloc/src/prof.c
+jemalloc/src/rtree.c
+jemalloc/src/safety_check.c
+jemalloc/src/sc.c
+jemalloc/src/stats.c
+jemalloc/src/sz.c
+jemalloc/src/tcache.c
+jemalloc/src/test_hooks.c
+jemalloc/src/ticker.c
+jemalloc/src/tsd.c
+jemalloc/src/witness.c
+jemalloc/src/zone.c
+jemalloc/test/include/test/btalloc.h
+jemalloc/test/include/test/extent_hooks.h
+jemalloc/test/include/test/jemalloc_test.h.in
+jemalloc/test/include/test/jemalloc_test_defs.h.in
+jemalloc/test/include/test/math.h
+jemalloc/test/include/test/mq.h
+jemalloc/test/include/test/mtx.h
+jemalloc/test/include/test/SFMT-alti.h
+jemalloc/test/include/test/SFMT-params.h
+jemalloc/test/include/test/SFMT-params607.h
+jemalloc/test/include/test/SFMT-params1279.h
+jemalloc/test/include/test/SFMT-params2281.h
+jemalloc/test/include/test/SFMT-params4253.h
+jemalloc/test/include/test/SFMT-params11213.h
+jemalloc/test/include/test/SFMT-params19937.h
+jemalloc/test/include/test/SFMT-params44497.h
+jemalloc/test/include/test/SFMT-params86243.h
+jemalloc/test/include/test/SFMT-params132049.h
+jemalloc/test/include/test/SFMT-params216091.h
+jemalloc/test/include/test/SFMT-sse2.h
+jemalloc/test/include/test/SFMT.h
+jemalloc/test/include/test/test.h
+jemalloc/test/include/test/thd.h
+jemalloc/test/include/test/timer.h
+jemalloc/test/integration/aligned_alloc.c
+jemalloc/test/integration/allocated.c
+jemalloc/test/integration/extent.c
+jemalloc/test/integration/extent.sh
+jemalloc/test/integration/malloc.c
+jemalloc/test/integration/mallocx.c
+jemalloc/test/integration/mallocx.sh
+jemalloc/test/integration/MALLOCX_ARENA.c
+jemalloc/test/integration/overflow.c
+jemalloc/test/integration/posix_memalign.c
+jemalloc/test/integration/rallocx.c
+jemalloc/test/integration/sdallocx.c
+jemalloc/test/integration/slab_sizes.c
+jemalloc/test/integration/slab_sizes.sh
+jemalloc/test/integration/smallocx.c
+jemalloc/test/integration/smallocx.sh
+jemalloc/test/integration/thread_arena.c
+jemalloc/test/integration/thread_tcache_enabled.c
+jemalloc/test/integration/xallocx.c
+jemalloc/test/integration/xallocx.sh
+jemalloc/test/src/btalloc.c
+jemalloc/test/src/btalloc_0.c
+jemalloc/test/src/btalloc_1.c
+jemalloc/test/src/math.c
+jemalloc/test/src/mq.c
+jemalloc/test/src/mtx.c
+jemalloc/test/src/SFMT.c
+jemalloc/test/src/test.c
+jemalloc/test/src/thd.c
+jemalloc/test/src/timer.c
+jemalloc/test/stress/hookbench.c
+jemalloc/test/stress/microbench.c
+jemalloc/test/test.sh.in
+jemalloc/test/unit/a0.c
+jemalloc/test/unit/arena_reset.c
+jemalloc/test/unit/arena_reset_prof.c
+jemalloc/test/unit/arena_reset_prof.sh
+jemalloc/test/unit/atomic.c
+jemalloc/test/unit/background_thread.c
+jemalloc/test/unit/background_thread_enable.c
+jemalloc/test/unit/base.c
+jemalloc/test/unit/binshard.c
+jemalloc/test/unit/binshard.sh
+jemalloc/test/unit/bitmap.c
+jemalloc/test/unit/bit_util.c
+jemalloc/test/unit/ckh.c
+jemalloc/test/unit/decay.c
+jemalloc/test/unit/decay.sh
+jemalloc/test/unit/div.c
+jemalloc/test/unit/emitter.c
+jemalloc/test/unit/extent_quantize.c
+jemalloc/test/unit/extent_util.c
+jemalloc/test/unit/fork.c
+jemalloc/test/unit/hash.c
+jemalloc/test/unit/hook.c
+jemalloc/test/unit/huge.c
+jemalloc/test/unit/junk.c
+jemalloc/test/unit/junk.sh
+jemalloc/test/unit/junk_alloc.c
+jemalloc/test/unit/junk_alloc.sh
+jemalloc/test/unit/junk_free.c
+jemalloc/test/unit/junk_free.sh
+jemalloc/test/unit/log.c
+jemalloc/test/unit/mallctl.c
+jemalloc/test/unit/malloc_io.c
+jemalloc/test/unit/math.c
+jemalloc/test/unit/mq.c
+jemalloc/test/unit/mtx.c
+jemalloc/test/unit/nstime.c
+jemalloc/test/unit/pack.c
+jemalloc/test/unit/pack.sh
+jemalloc/test/unit/pages.c
+jemalloc/test/unit/ph.c
+jemalloc/test/unit/prng.c
+jemalloc/test/unit/prof_accum.c
+jemalloc/test/unit/prof_accum.sh
+jemalloc/test/unit/prof_active.c
+jemalloc/test/unit/prof_active.sh
+jemalloc/test/unit/prof_gdump.c
+jemalloc/test/unit/prof_gdump.sh
+jemalloc/test/unit/prof_idump.c
+jemalloc/test/unit/prof_idump.sh
+jemalloc/test/unit/prof_log.c
+jemalloc/test/unit/prof_log.sh
+jemalloc/test/unit/prof_reset.c
+jemalloc/test/unit/prof_reset.sh
+jemalloc/test/unit/prof_tctx.c
+jemalloc/test/unit/prof_tctx.sh
+jemalloc/test/unit/prof_thread_name.c
+jemalloc/test/unit/prof_thread_name.sh
+jemalloc/test/unit/ql.c
+jemalloc/test/unit/qr.c
+jemalloc/test/unit/rb.c
+jemalloc/test/unit/retained.c
+jemalloc/test/unit/rtree.c
+jemalloc/test/unit/safety_check.c
+jemalloc/test/unit/safety_check.sh
+jemalloc/test/unit/sc.c
+jemalloc/test/unit/seq.c
+jemalloc/test/unit/SFMT.c
+jemalloc/test/unit/size_classes.c
+jemalloc/test/unit/slab.c
+jemalloc/test/unit/smoothstep.c
+jemalloc/test/unit/spin.c
+jemalloc/test/unit/stats.c
+jemalloc/test/unit/stats_print.c
+jemalloc/test/unit/test_hooks.c
+jemalloc/test/unit/ticker.c
+jemalloc/test/unit/tsd.c
+jemalloc/test/unit/witness.c
+jemalloc/test/unit/zero.c
+jemalloc/test/unit/zero.sh
+jemalloc/TUNING.md
+m4/ax_cxx_compile_stdcxx.m4
+m4/ax_cxx_compile_stdcxx_11.m4
+m4/ax_pthread.m4
+memkind.pc.in
+test/gtest_fused/gtest/gtest-all.cc
+test/gtest_fused/gtest/gtest.h
+test/memkind-afts-ext.ts
+test/memkind-afts.ts
+test/memkind-perf-ext.ts
+test/memkind-perf.ts
+test/memkind-pytests.ts
+test/memkind-slts.ts
+utils/docker/README.md
diff --git a/copying_headers/MANIFEST.freeBSD b/copying_headers/MANIFEST.freeBSD
new file mode 100644
index 0000000..56798ed
--- /dev/null
+++ b/copying_headers/MANIFEST.freeBSD
@@ -0,0 +1,193 @@
+Makefile.am
+astyle.sh
+autogen.sh
+autohbw/Makefile.mk
+autohbw/autohbw.c
+autohbw/autohbw_get_src_lines.pl
+autohbw/autohbw_test.sh
+build.sh
+configure.ac
+examples/Makefile.mk
+examples/autohbw_candidates.c
+examples/filter_example.c
+examples/hello_hbw_example.c
+examples/hello_memkind_example.c
+examples/memkind_allocated.hpp
+examples/memkind_allocated_example.cpp
+examples/memkind_decorator_debug.c
+examples/memkind_get_stat.c
+include/hbw_allocator.h
+include/hbwmalloc.h
+include/memkind.h
+include/memkind/internal/heap_manager.h
+include/memkind/internal/memkind_arena.h
+include/memkind/internal/memkind_bandwidth.h
+include/memkind/internal/memkind_dax_kmem.h
+include/memkind/internal/memkind_default.h
+include/memkind/internal/memkind_gbtlb.h
+include/memkind/internal/memkind_hbw.h
+include/memkind/internal/memkind_hugetlb.h
+include/memkind/internal/memkind_interleave.h
+include/memkind/internal/memkind_log.h
+include/memkind/internal/memkind_pmem.h
+include/memkind/internal/memkind_private.h
+include/memkind/internal/memkind_regular.h
+include/memkind/internal/tbb_wrapper.h
+include/memkind/internal/vec.h
+include/memkind_allocator.h
+include/memkind_deprecated.h
+include/pmem_allocator.h
+install_astyle.sh
+man/autohbw.7
+man/hbwallocator.3
+man/hbwmalloc.3
+mna/memkind-auto-dax-kmem-nodes.1
+man/memkind-hbw-nodes.1
+man/memkind.3
+man/memkind_arena.3
+man/memkind_dax_kmem.3
+man/memkind_default.3
+man/memkind_hbw.3
+man/memkind_hugetlb.3
+man/memkind_pmem.3
+man/memkindallocator.3
+man/pmemallocator.3
+memkind.spec.mk
+src/Makefile.mk
+src/hbwmalloc.c
+src/heap_manager.c
+src/memkind-auto-dax-kmem-nodes.c
+src/memkind-hbw-nodes.c
+src/memkind.c
+src/memkind_arena.c
+src/memkind_bandwidth.c
+src/memkind_dax_kmem.c
+src/memkind_default.c
+src/memkind_gbtlb.c
+src/memkind_hbw.c
+src/memkind_hugetlb.c
+src/memkind_interleave.c
+src/memkind_log.c
+src/memkind_pmem.c
+src/memkind_regular.c
+src/tbb_wrapper.c
+test/Allocator.hpp
+test/Makefile.mk
+test/TestPolicy.hpp
+test/alloc_benchmark.c
+test/alloc_performance_tests.cpp
+test/allocate_to_max_stress_test.cpp
+test/allocator_perf_tool/AllocationSizes.hpp
+test/allocator_perf_tool/Allocation_info.cpp
+test/allocator_perf_tool/Allocation_info.hpp
+test/allocator_perf_tool/Allocator.hpp
+test/allocator_perf_tool/AllocatorFactory.hpp
+test/allocator_perf_tool/CSVLogger.hpp
+test/allocator_perf_tool/CommandLine.hpp
+test/allocator_perf_tool/Configuration.hpp
+test/allocator_perf_tool/ConsoleLog.hpp
+test/allocator_perf_tool/FunctionCalls.hpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.h
+test/allocator_perf_tool/GTestAdapter.hpp
+test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/HugePageOrganizer.hpp
+test/allocator_perf_tool/HugePageUnmap.hpp
+test/allocator_perf_tool/Iterator.hpp
+test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/Makefile
+test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
+test/allocator_perf_tool/Numastat.hpp
+test/allocator_perf_tool/PmemMockup.cpp
+test/allocator_perf_tool/PmemMockup.hpp
+test/allocator_perf_tool/Runnable.hpp
+test/allocator_perf_tool/ScenarioWorkload.cpp
+test/allocator_perf_tool/ScenarioWorkload.h
+test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
+test/allocator_perf_tool/Stats.hpp
+test/allocator_perf_tool/StressIncreaseToMax.cpp
+test/allocator_perf_tool/StressIncreaseToMax.h
+test/allocator_perf_tool/Task.hpp
+test/allocator_perf_tool/TaskFactory.hpp
+test/allocator_perf_tool/Tests.hpp
+test/allocator_perf_tool/Thread.hpp
+test/allocator_perf_tool/TimerSysTime.hpp
+test/allocator_perf_tool/VectorIterator.hpp
+test/allocator_perf_tool/Workload.hpp
+test/allocator_perf_tool/WrappersMacros.hpp
+test/allocator_perf_tool/main.cpp
+test/autohbw_test.py
+test/autohbw_test_helper.c
+test/bat_tests.cpp
+test/check.cpp
+test/check.h
+test/common.h
+test/dax_kmem_nodes.cpp
+test/dax_kmem_nodes.h
+test/dax_kmem_env_var_test.py
+test/decorator_test.cpp
+test/decorator_test.h
+test/dlopen_test.cpp
+test/draw_plots.py
+test/environ_err_dax_kmem_malloc_positive_test.cpp
+test/environ_err_dax_kmem_malloc_test.cpp
+test/environ_err_hbw_malloc_test.cpp
+test/error_message_tests.cpp
+test/fragmentation_benchmark_pmem.cpp
+test/freeing_memory_segfault_test.cpp
+test/gb_page_tests_bind_policy.cpp
+test/get_arena_test.cpp
+test/hbw_allocator_performance_tests.cpp
+test/hbw_allocator_tests.cpp
+test/hbw_detection_test.py
+test/hbw_verify_function_test.cpp
+test/heap_manager_init_perf_test.cpp
+test/huge_page_test.cpp
+test/load_tbbmalloc_symbols.c
+test/locality_test.cpp
+test/main.cpp
+test/memkind_allocator_tests.cpp
+test/memkind_dax_kmem_test.cpp
+test/memkind_detect_kind_tests.cpp
+test/memkind_null_kind_test.cpp
+test/memkind_pmem_config_tests.cpp
+test/memkind_pmem_long_time_tests.cpp
+test/memkind_pmem_tests.cpp
+test/memkind_stat_test.cpp
+test/memkind_defrag_reallocate.cpp
+test/memkind_versioning_tests.cpp
+test/memory_footprint_test.cpp
+test/memory_manager.h
+test/multithreaded_tests.cpp
+test/negative_tests.cpp
+test/performance/framework.cpp
+test/performance/framework.hpp
+test/performance/operations.hpp
+test/performance/perf_tests.cpp
+test/performance/perf_tests.hpp
+test/pmem_alloc_performance_tests.cpp
+test/pmem_allocator_tests.cpp
+test/proc_stat.h
+test/python_framework/__init__.py
+test/python_framework/cmd_helper.py
+test/python_framework/huge_page_organizer.py
+test/random_sizes_allocator.h
+test/run_alloc_benchmark.sh
+test/static_kinds_list.h
+test/static_kinds_tests.cpp
+test/tbbmalloc.h
+test/test.sh
+test/test_dax_kmem.sh
+test/trace_mechanism_test.py
+test/trace_mechanism_test_helper.c
+test/trial_generator.cpp
+test/trial_generator.h
+utils/docker/Dockerfile.fedora-31
+utils/docker/Dockerfile.ubuntu-18.04
+utils/docker/docker_install_ndctl.sh
+utils/docker/docker_install_tbb.sh
+utils/docker/docker_run_build.sh
+utils/docker/docker_run_coverage.sh
+utils/docker/docker_run_test.sh
+utils/docker/run_local.sh
+utils/docker/set_host_configuration.sh
diff --git a/copying_headers/MANIFEST.freeBSD3 b/copying_headers/MANIFEST.freeBSD3
new file mode 100644
index 0000000..553a45e
--- /dev/null
+++ b/copying_headers/MANIFEST.freeBSD3
@@ -0,0 +1,14 @@
+examples/memkind_cpp_allocator.cpp
+examples/pmem_alignment.c
+examples/pmem_and_dax_kmem_kind.c
+examples/pmem_and_default_kind.c
+examples/pmem_config.c
+examples/pmem_cpp_allocator.cpp
+examples/pmem_detect_kind.c
+examples/pmem_free_with_unknown_kind.c
+examples/pmem_kinds.c
+examples/pmem_malloc.c
+examples/pmem_malloc_unlimited.c
+examples/pmem_multithreads.c
+examples/pmem_multithreads_onekind.c
+examples/pmem_usable_size.c
diff --git a/copying_headers/header.freeBSD b/copying_headers/header.freeBSD
new file mode 100644
index 0000000..db5a089
--- /dev/null
+++ b/copying_headers/header.freeBSD
@@ -0,0 +1,23 @@
+/*
+ * Copyright (C) 2014 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
\ No newline at end of file
diff --git a/copying_headers/header.freeBSD3 b/copying_headers/header.freeBSD3
new file mode 100644
index 0000000..9119412
--- /dev/null
+++ b/copying_headers/header.freeBSD3
@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2015, Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
diff --git a/copying_headers/header.intel-acpi b/copying_headers/header.intel-acpi
new file mode 100644
index 0000000..c4603f8
--- /dev/null
+++ b/copying_headers/header.intel-acpi
@@ -0,0 +1,36 @@
+/*
+ * Copyright (C) 2000 - 2014, Intel Corp.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions, and the following disclaimer,
+ *    without modification.
+ * 2. Redistributions in binary form must reproduce at minimum a disclaimer
+ *    substantially similar to the "NO WARRANTY" disclaimer below
+ *    ("Disclaimer") and any redistribution must be conditioned upon
+ *    including a substantially similar Disclaimer requirement for further
+ *    binary redistribution.
+ * 3. Neither the names of the above-listed copyright holders nor the names
+ *    of any contributors may be used to endorse or promote products derived
+ *    from this software without specific prior written permission.
+ *
+ * Alternatively, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") version 2 as published by the Free
+ * Software Foundation.
+ *
+ * NO WARRANTY
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * HOLDERS OR CONTRIBUTORS BE LIABLE FOR SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
+ * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGES.
+ */
\ No newline at end of file
diff --git a/debian/changelog b/debian/changelog
new file mode 100644
index 0000000..8d118a4
--- /dev/null
+++ b/debian/changelog
@@ -0,0 +1,5 @@
+memkind (0.0.0) UNRELEASED; urgency=low
+
+  * See ChangeLog for more details
+
+ -- Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>  Mon, 12 Dec 2016 15:53:31 +0000
diff --git a/debian/compat b/debian/compat
new file mode 100644
index 0000000..ec63514
--- /dev/null
+++ b/debian/compat
@@ -0,0 +1 @@
+9
diff --git a/debian/control b/debian/control
new file mode 100644
index 0000000..5a13b16
--- /dev/null
+++ b/debian/control
@@ -0,0 +1,25 @@
+Source: memkind
+Section: utils
+Priority: optional
+Maintainer: Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>
+Build-Depends: debhelper (>= 8.0.0)
+Standards-Version: 3.9.4
+Homepage: https://github.com/memkind/memkind
+
+Package: memkind
+Architecture: any
+Depends: ${shlibs:Depends}, ${misc:Depends}
+Description:
+ The memkind library is an user extensible heap manager built on top of
+ jemalloc which enables control of memory characteristics and a
+ partitioning of the heap between kinds of memory. The kinds of memory
+ are defined by operating system memory policies that have been applied
+ to virtual address ranges. Memory characteristics supported by
+ memkind without user extension include control of NUMA and page size
+ features. The jemalloc non-standard interface has been extended to
+ enable specialized arenas to make requests for virtual memory from the
+ operating system through the memkind partition interface. Through the
+ other memkind interfaces the user can control and extend memory
+ partition features and allocate memory while selecting enabled
+ features. This software is being made available for early evaluation.
+ Feedback on design or implementation is greatly appreciated.
diff --git a/debian/rules b/debian/rules
new file mode 100755
index 0000000..5e7f167
--- /dev/null
+++ b/debian/rules
@@ -0,0 +1,21 @@
+#!/usr/bin/make -f
+# -*- makefile -*-
+
+# Uncomment this to turn on verbose mode.
+#export DH_VERBOSE=1
+
+# ignore tests during build
+override_dh_auto_test:
+
+override_dh_auto_configure:
+	./build_jemalloc.sh
+	./autogen.sh
+	dh_auto_configure -- --prefix=/usr --libdir=/usr/lib \
+           --includedir=/usr/include --sbindir=/usr/sbin --enable-cxx11 \
+           --mandir=/usr/share/man --docdir=/usr/share/doc/memkind
+
+override_dh_auto_build:
+	dh_auto_build -- checkprogs
+
+%:
+	dh $@
diff --git a/examples/Makefile.mk b/examples/Makefile.mk
new file mode 100644
index 0000000..bcc8ba2
--- /dev/null
+++ b/examples/Makefile.mk
@@ -0,0 +1,99 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+noinst_PROGRAMS += examples/autohbw_candidates \
+                   examples/filter_memkind \
+                   examples/hello_hbw \
+                   examples/hello_memkind \
+                   examples/hello_memkind_debug \
+                   examples/memkind_get_stat \
+                   examples/pmem_alignment \
+                   examples/pmem_and_dax_kmem_kind \
+                   examples/pmem_and_default_kind \
+                   examples/pmem_config \
+                   examples/pmem_detect_kind \
+                   examples/pmem_free_with_unknown_kind \
+                   examples/pmem_kinds \
+                   examples/pmem_malloc \
+                   examples/pmem_malloc_unlimited \
+                   examples/pmem_multithreads \
+                   examples/pmem_multithreads_onekind \
+                   examples/pmem_usable_size \
+                   # end
+if HAVE_CXX11
+noinst_PROGRAMS += examples/memkind_allocated
+noinst_PROGRAMS += examples/memkind_cpp_allocator
+noinst_PROGRAMS += examples/pmem_cpp_allocator
+endif
+
+examples_autohbw_candidates_LDADD = libmemkind.la
+examples_filter_memkind_LDADD = libmemkind.la
+examples_hello_hbw_LDADD = libmemkind.la
+examples_hello_memkind_LDADD = libmemkind.la
+examples_hello_memkind_debug_LDADD = libmemkind.la
+examples_memkind_get_stat_LDADD = libmemkind.la
+examples_pmem_alignment_LDADD = libmemkind.la
+examples_pmem_and_dax_kmem_kind_LDADD = libmemkind.la
+examples_pmem_and_default_kind_LDADD = libmemkind.la
+examples_pmem_config_LDADD = libmemkind.la
+examples_pmem_detect_kind_LDADD = libmemkind.la
+examples_pmem_free_with_unknown_kind_LDADD = libmemkind.la
+examples_pmem_kinds_LDADD = libmemkind.la
+examples_pmem_malloc_LDADD = libmemkind.la
+examples_pmem_malloc_unlimited_LDADD = libmemkind.la
+examples_pmem_multithreads_LDADD = libmemkind.la
+examples_pmem_multithreads_onekind_LDADD = libmemkind.la
+examples_pmem_usable_size_LDADD = libmemkind.la
+
+if HAVE_CXX11
+examples_memkind_allocated_LDADD = libmemkind.la
+examples_memkind_cpp_allocator_LDADD  = libmemkind.la
+examples_pmem_cpp_allocator_LDADD = libmemkind.la
+endif
+
+examples_autohbw_candidates_SOURCES = examples/autohbw_candidates.c
+examples_filter_memkind_SOURCES = examples/filter_example.c
+examples_hello_hbw_SOURCES = examples/hello_hbw_example.c
+examples_hello_memkind_SOURCES = examples/hello_memkind_example.c
+examples_hello_memkind_debug_SOURCES = examples/hello_memkind_example.c examples/memkind_decorator_debug.c
+examples_pmem_alignment_SOURCES = examples/pmem_alignment.c
+examples_pmem_and_dax_kmem_kind_SOURCES = examples/pmem_and_dax_kmem_kind.c
+examples_pmem_and_default_kind_SOURCES = examples/pmem_and_default_kind.c
+examples_pmem_config_SOURCES = examples/pmem_config.c
+examples_pmem_detect_kind_SOURCES = examples/pmem_detect_kind.c
+examples_pmem_free_with_unknown_kind_SOURCES = examples/pmem_free_with_unknown_kind.c
+examples_pmem_kinds_SOURCES = examples/pmem_kinds.c
+examples_pmem_malloc_SOURCES = examples/pmem_malloc.c
+examples_pmem_malloc_unlimited_SOURCES = examples/pmem_malloc_unlimited.c
+examples_pmem_multithreads_SOURCES = examples/pmem_multithreads.c
+examples_pmem_multithreads_onekind_SOURCES = examples/pmem_multithreads_onekind.c
+examples_pmem_usable_size_SOURCES = examples/pmem_usable_size.c
+if HAVE_CXX11
+examples_memkind_allocated_SOURCES = examples/memkind_allocated_example.cpp examples/memkind_allocated.hpp
+examples_memkind_cpp_allocator_SOURCES = examples/memkind_cpp_allocator.cpp
+examples_pmem_cpp_allocator_SOURCES = examples/pmem_cpp_allocator.cpp
+endif
+
+clean-local:
+	rm -f examples/*.gcno examples/*.gcda
diff --git a/examples/README b/examples/README
new file mode 100644
index 0000000..79e4408
--- /dev/null
+++ b/examples/README
@@ -0,0 +1,104 @@
+# Memkind examples
+
+The example directory contains example codes that use the memkind
+interface.
+
+## PMEM
+
+The pmem_*.c(pp) demonstrates how to create and use a file-backed memory kind.
+The default pmem path is "/tmp/".
+Custom directory is pass as first argument to all of PMEM example programs,
+e.g. to execute pmem_malloc example in /mnt/pmem location, call:
+
+    ./pmem_malloc /mnt/pmem/
+
+### pmem_kinds.c
+
+This example shows how to create and destroy pmem kind with defined or unlimited size.
+
+### pmem_malloc.c
+
+This example shows how to allocate memory and possibility to exceed pmem kind size.
+
+### pmem_malloc_unlimited.c
+
+This example shows how to allocate memory with unlimited kind size.
+
+### pmem_usable_size.c
+
+This example shows difference between the expected and the actual allocation size.
+
+### pmem_alignment.c
+
+This example shows how to use memkind alignment and how it affects allocations.
+
+### pmem_multithreads.c
+
+This example shows how to use multithreading with independent pmem kinds.
+
+### pmem_multithreads_onekind.c
+
+This example shows how to use multithreading with one main pmem kind.
+
+### pmem_and_dax_kmem_kind.c
+
+This example shows how to allocate to PMEM memory using file-backed memory (pmem kind)
+and persistent memory NUMA node (MEMKIND_DAX_KMEM).
+
+### pmem_and_default_kind.c
+
+This example shows how to allocate in standard memory and file-backed memory (pmem kind).
+
+### pmem_detect_kind.c
+
+This example shows how to distinguish allocation from different kinds using detect kind function.
+
+### pmem_config.c
+
+This example shows how to use custom configuration to create pmem kind.
+
+### pmem_free_with_unknown_kind.c
+
+This example shows how to allocate in standard memory and file-backed memory (pmem kind)
+and free memory without a need to remember which kind it belongs to.
+
+### pmem_cpp_allocator.cpp
+
+This example shows usage of C++ allocator mechanism designed for file-backed memory
+kind with different data structures like: vector, list and map.
+
+## Other memkind examples
+
+### memkind_get_stat.c
+
+This example shows how to get information about allocator and kind statistics.
+
+### other
+
+The simplest example is the hello_example.c which is a hello world
+variant.  The filter_example.c shows how you would use high bandwidth
+memory to store a reduction of a larger data set stored in DDR. There is
+also an example of how to create user defined kinds.  This example
+creates kinds which isolate allocations to a single NUMA node each
+backed by a single arena.
+
+### memkind_cpp_allocator.cpp
+
+This example shows usage of C++ allocator mechanism designed for static
+kinds with different data structures like: forward list, deque and multiset.
+
+The memkind_allocated example is simple usage of memkind in C++11 which
+shows how memkind can be used to allocate objects, and consists of two files:
+memkind_allocated.hpp - which is definition of template class that should be
+inherited and parametrized by derived type (curiously recurring template
+pattern), to let deriving class allocate their objects using specified kind.
+memkind_allocated_example.cpp - which is usage example of this approach.
+Logic of memkind_allocated is based on overriding operator new() in template,
+and allocating memory on kind specified in new() parameter, or by overridable
+static method getClassKind(). This implementation also supports alignment
+specifier's (alignas() - new feature in C++11).
+The downside of this approach is that it will work properly only if
+memkind_allocated template is inherited once in inheritance chain (which
+probably makes that not very useful for most scenarios). Other thing is that it
+overriding class new() operator which can cause various problems if used
+unwisely.
diff --git a/examples/autohbw_candidates.c b/examples/autohbw_candidates.c
new file mode 100644
index 0000000..f4844ce
--- /dev/null
+++ b/examples/autohbw_candidates.c
@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////
+// File   : autohbw_candidates.c
+// Purpose: Shows which functions are interposed by AutoHBW library.
+//        : These functions can be used for testing purposes
+// Author : Ruchira Sasanka (ruchira.sasanka AT intel.com)
+// Date   : Sept 10, 2015
+///////////////////////////////////////////////////////////////////////////
+
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+
+
+///////////////////////////////////////////////////////////////////////////
+// This function contains an example case for each heap allocation function
+// intercepted by the AutoHBW library
+///////////////////////////////////////////////////////////////////////////
+
+//volatile is needed to prevent optimizing out below hooks
+volatile int memkind_called_g;
+
+void memkind_malloc_post(struct memkind *kind, size_t size, void **result)
+{
+    memkind_called_g = 1;
+}
+void memkind_calloc_post(struct memkind *kind, size_t nmemb, size_t size,
+                         void **result)
+{
+    memkind_called_g = 1;
+}
+void memkind_posix_memalign_post(struct memkind *kind, void **memptr,
+                                 size_t alignment, size_t size, int *err)
+{
+    memkind_called_g = 1;
+}
+void memkind_realloc_post(struct memkind *kind, void *ptr, size_t size,
+                          void **result)
+{
+    memkind_called_g = 1;
+}
+void memkind_free_pre(struct memkind **kind, void **ptr)
+{
+    memkind_called_g = 1;
+}
+
+void finish_testcase(int fail_condition, const char *fail_message, int *err)
+{
+
+    if(memkind_called_g != 1 || fail_condition) {
+        printf("%s\n", fail_message);
+        *err= -1;
+    }
+    memkind_called_g = 0;
+}
+
+int main()
+{
+    int err = 0;
+    const size_t size = 1024 * 1024;   // 1M of data
+
+    void *buf = NULL;
+    memkind_called_g = 0;
+
+    // Test 1: Test malloc and free
+    buf = malloc(size);
+    finish_testcase(buf==NULL, "Malloc failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after malloc failed!", &err);
+
+    // Test 2: Test calloc and free
+    buf = calloc(size, 1);
+    finish_testcase(buf==NULL, "Calloc failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after calloc failed!", &err);
+
+    // Test 3: Test realloc and free
+    buf = malloc(size);
+    finish_testcase(buf==NULL, "Malloc before realloc failed!", &err);
+
+    buf = realloc(buf,  size * 2);
+    finish_testcase(buf==NULL, "Realloc failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after realloc failed!", &err);
+    buf = NULL;
+
+    // Test 4: Test posix_memalign and free
+    int ret = posix_memalign(&buf,  64, size);
+    finish_testcase(ret, "Posix_memalign failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after posix_memalign failed!", &err);
+
+    return err;
+}
diff --git a/examples/filter_example.c b/examples/filter_example.c
new file mode 100644
index 0000000..f128d71
--- /dev/null
+++ b/examples/filter_example.c
@@ -0,0 +1,87 @@
+/*
+ * Copyright (C) 2014 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <errno.h>
+
+int main(int argc, char **argv)
+{
+    const size_t stream_len = 1024 * 1024;
+    const size_t filter_len = 1024;
+    const size_t num_filter = stream_len / filter_len;
+    size_t i, j;
+    double *stream = NULL;
+    double *filter = NULL;
+    double *result = NULL;
+
+    srandom(0);
+
+    stream = (double *)memkind_malloc(MEMKIND_DEFAULT, stream_len * sizeof(double));
+    if (stream == NULL) {
+        perror("<memkind>");
+        fprintf(stderr, "Unable to allocate stream\n");
+        return errno ? -errno : 1;
+    }
+
+    filter = (double *)memkind_malloc(MEMKIND_HBW, filter_len * sizeof(double));
+    if (filter == NULL) {
+        perror("<memkind>");
+        fprintf(stderr, "Unable to allocate filter\n");
+        return errno ? -errno : 1;
+    }
+
+    result = (double *)memkind_calloc(MEMKIND_HBW, filter_len, sizeof(double));
+    if (result == NULL) {
+        perror("<memkind>");
+        fprintf(stderr, "Unable to allocate result\n");
+        return errno ? -errno : 1;
+    }
+
+    for (i = 0; i < stream_len; i++) {
+        stream[i] = (double)(random())/(double)(RAND_MAX);
+    }
+
+    for (i = 0; i < filter_len; i++) {
+        filter[i] = (double)(i)/(double)(filter_len);
+    }
+
+    for (i = 0; i < num_filter; i++) {
+        for (j = 0; j < filter_len; j++) {
+            result[j] += stream[i * filter_len + j] * filter[j];
+        }
+    }
+
+    for (i = 0; i < filter_len; i++) {
+        fprintf(stdout, "%.6e\n", result[i]);
+    }
+
+    memkind_free(MEMKIND_HBW, result);
+    memkind_free(MEMKIND_HBW, filter);
+    memkind_free(MEMKIND_DEFAULT, stream);
+
+    return 0;
+}
diff --git a/examples/hello_hbw_example.c b/examples/hello_hbw_example.c
new file mode 100644
index 0000000..7b044fa
--- /dev/null
+++ b/examples/hello_hbw_example.c
@@ -0,0 +1,81 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbwmalloc.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <errno.h>
+
+int main(int argc, char **argv)
+{
+    const size_t size = 512;
+    char *default_str = NULL;
+    char *hbw_str = NULL;
+    char *hbw_hugetlb_str = NULL;
+    int err = 0;
+
+    default_str = (char *)malloc(size);
+    if (default_str == NULL) {
+        perror("malloc()");
+        fprintf(stderr, "Unable to allocate default string\n");
+        err = errno ? -errno : 1;
+        goto exit;
+    }
+    hbw_str = (char *)hbw_malloc(size);
+    if (hbw_str == NULL) {
+        perror("hbw_malloc()");
+        fprintf(stderr, "Unable to allocate hbw string\n");
+        err = errno ? -errno : 1;
+        goto exit;
+    }
+    err = hbw_posix_memalign_psize((void **)&hbw_hugetlb_str, 2097152, size,
+                                   HBW_PAGESIZE_2MB);
+    if (err) {
+        perror("hbw_posix_memalign()");
+        fprintf(stderr, "Unable to allocate hbw hugetlb string\n");
+        err = errno ? -errno : 1;
+        goto exit;
+    }
+
+    sprintf(default_str, "Hello world from standard memory\n");
+    sprintf(hbw_str, "Hello world from high bandwidth memory\n");
+    sprintf(hbw_hugetlb_str, "Hello world from high bandwidth 2 MB paged memory\n");
+
+    fprintf(stdout, "%s", default_str);
+    fprintf(stdout, "%s", hbw_str);
+    fprintf(stdout, "%s", hbw_hugetlb_str);
+
+exit:
+    if (hbw_hugetlb_str) {
+        hbw_free(hbw_hugetlb_str);
+    }
+    if (hbw_str) {
+        hbw_free(hbw_str);
+    }
+    if (default_str) {
+        free(default_str);
+    }
+    return err;
+}
diff --git a/examples/hello_memkind_example.c b/examples/hello_memkind_example.c
new file mode 100644
index 0000000..e85e30c
--- /dev/null
+++ b/examples/hello_memkind_example.c
@@ -0,0 +1,115 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <errno.h>
+
+int main(int argc, char **argv)
+{
+    const size_t size = 512;
+    char *default_str = NULL;
+    char *hugetlb_str = NULL;
+    char *hbw_str = NULL;
+    char *hbw_hugetlb_str = NULL;
+    char *hbw_preferred_str = NULL;
+    char *hbw_preferred_hugetlb_str = NULL;
+    char *hbw_interleave_str = NULL;
+
+    default_str = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    if (default_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate default string\n");
+        return errno ? -errno : 1;
+    }
+    hugetlb_str = (char *)memkind_malloc(MEMKIND_HUGETLB, size);
+    if (hugetlb_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hugetlb string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_str = (char *)memkind_malloc(MEMKIND_HBW, size);
+    if (hbw_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_hugetlb_str = (char *)memkind_malloc(MEMKIND_HBW_HUGETLB, size);
+    if (hbw_hugetlb_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw_hugetlb string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_preferred_str = (char *)memkind_malloc(MEMKIND_HBW_PREFERRED, size);
+    if (hbw_preferred_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw_preferred string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_preferred_hugetlb_str = (char *)memkind_malloc(
+                                    MEMKIND_HBW_PREFERRED_HUGETLB, size);
+    if (hbw_preferred_hugetlb_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw_preferred_hugetlb string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_interleave_str = (char *)memkind_malloc(MEMKIND_HBW_INTERLEAVE, size);
+    if (hbw_interleave_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr,"Unable to allocate hbw_interleave string\n");
+        return errno ? -errno : 1;
+    }
+
+    sprintf(default_str, "Hello world from standard memory\n");
+    sprintf(hugetlb_str, "Hello world from standard memory with 2 MB pages\n");
+    sprintf(hbw_str, "Hello world from high bandwidth memory\n");
+    sprintf(hbw_hugetlb_str, "Hello world from high bandwidth 2 MB paged memory\n");
+    sprintf(hbw_preferred_str,
+            "Hello world from high bandwidth memory if sufficient resources exist\n");
+    sprintf(hbw_preferred_hugetlb_str,
+            "Hello world from high bandwidth 2 MB paged memory if sufficient resources exist\n");
+
+    sprintf(hbw_interleave_str,
+            "Hello world from high bandwidth interleaved memory\n");
+
+    fprintf(stdout, "%s", default_str);
+    fprintf(stdout, "%s", hugetlb_str);
+    fprintf(stdout, "%s", hbw_str);
+    fprintf(stdout, "%s", hbw_hugetlb_str);
+    fprintf(stdout, "%s", hbw_preferred_str);
+    fprintf(stdout, "%s", hbw_preferred_hugetlb_str);
+    fprintf(stdout, "%s", hbw_interleave_str);
+
+    memkind_free(MEMKIND_HBW_INTERLEAVE, hbw_interleave_str);
+    memkind_free(MEMKIND_HBW_PREFERRED_HUGETLB, hbw_preferred_hugetlb_str);
+    memkind_free(MEMKIND_HBW_PREFERRED, hbw_preferred_str);
+    memkind_free(MEMKIND_HBW_HUGETLB, hbw_hugetlb_str);
+    memkind_free(MEMKIND_HBW, hbw_str);
+    memkind_free(MEMKIND_HUGETLB, hugetlb_str);
+    memkind_free(MEMKIND_DEFAULT, default_str);
+
+    return 0;
+}
diff --git a/examples/memkind_allocated.hpp b/examples/memkind_allocated.hpp
new file mode 100644
index 0000000..dd75bef
--- /dev/null
+++ b/examples/memkind_allocated.hpp
@@ -0,0 +1,107 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+#include <memkind.h>
+
+#include <cstdlib>
+#include <new>
+
+template <class deriving_class>
+class memkind_allocated
+{
+public:
+    static memkind_t getClassKind()
+    {
+        return MEMKIND_DEFAULT;
+    }
+
+    void *operator new(std::size_t size)
+    {
+        return deriving_class::operator new(size, deriving_class::getClassKind());
+    }
+
+    void *operator new[](std::size_t size)
+    {
+        return deriving_class::operator new(size, deriving_class::getClassKind());
+    }
+
+    void *operator new(std::size_t size, memkind_t memory_kind)
+    {
+        void *result_ptr = NULL;
+        int allocation_result = 0;
+
+        //This check if deriving_class has specified alignment, which is suitable
+        //to be used with posix_memalign()
+        if(alignof(deriving_class) <  sizeof(void *)) {
+            result_ptr = memkind_malloc(memory_kind, size);
+            allocation_result = result_ptr ? 1 : 0;
+        } else {
+            allocation_result = memkind_posix_memalign(memory_kind, &result_ptr,
+                                                       alignof(deriving_class), size);
+        }
+
+        if(allocation_result) {
+            throw std::bad_alloc();
+        }
+
+        return result_ptr;
+    }
+
+    void *operator new[](std::size_t size, memkind_t memory_kind)
+    {
+        return deriving_class::operator new(size, memory_kind);
+    }
+
+    void operator delete(void *ptr, memkind_t memory_kind)
+    {
+        memkind_free(memory_kind, ptr);
+    }
+
+    void operator delete(void *ptr)
+    {
+        memkind_free(0, ptr);
+    }
+
+    void operator delete[](void *ptr)
+    {
+        deriving_class::operator delete(ptr);
+    }
+
+    void operator delete[](void *ptr, memkind_t memory_kind)
+    {
+        deriving_class::operator delete(ptr, memory_kind);
+    }
+
+protected:
+    memkind_allocated()
+    {
+    }
+
+    ~memkind_allocated()
+    {
+    }
+
+};
diff --git a/examples/memkind_allocated_example.cpp b/examples/memkind_allocated_example.cpp
new file mode 100644
index 0000000..52f916b
--- /dev/null
+++ b/examples/memkind_allocated_example.cpp
@@ -0,0 +1,105 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+#include <memkind.h>
+#include "memkind_allocated.hpp"
+
+// This code is example usage of C++11 features with custom allocator,
+// support for C++11 is required.
+#include <iostream>
+
+#if __cplusplus > 199711L
+
+#include <cstdlib>
+#include <new>
+#include <string>
+
+//example class definition, which derive from  memkind_allocated template to
+//have objects allocated with memkind, and have alignment specified by alignas()
+class alignas(128) memkind_allocated_example : public
+    memkind_allocated<memkind_allocated_example>
+{
+    std::string message;
+
+public:
+    //Override method for returning class default kind to make objects be by default allocated on High-Bandwidth Memory
+    static memkind_t getClassKind()
+    {
+        return MEMKIND_HBW;
+    }
+
+    memkind_allocated_example(std::string my_message)
+    {
+        this->message = my_message;
+    }
+
+    memkind_allocated_example()
+    {
+    }
+
+    void print_message()
+    {
+        std::cout << message << std::endl;
+        std::cout << "Memory address of this object is: " <<  (void *)this << std::endl
+                  << std::endl;
+    }
+};
+
+int main()
+{
+    memkind_t specified_kind = MEMKIND_HBW_HUGETLB;
+
+    memkind_allocated_example *default_kind_example = new memkind_allocated_example(
+        std::string("This object has been allocated using class default kind, which is: MEMKIND_DEFAULT") );
+    default_kind_example->print_message();
+    delete default_kind_example;
+
+    memkind_allocated_example *specified_kind_example = new(
+        specified_kind) memkind_allocated_example(
+        std::string("This object has been allocated using specified kind, which is: MEMKIND_HBW_HUGETLB") );
+    specified_kind_example->print_message();
+    delete specified_kind_example;
+
+    //examples for using same approach for allocating arrays of objects, note that objects created that way can be initialized only with default (unparameterized) constructor
+    memkind_allocated_example *default_kind_array_example = new
+    memkind_allocated_example[5]();
+    delete[] default_kind_array_example;
+
+    memkind_allocated_example *specified_kind_array_example = new(
+        specified_kind) memkind_allocated_example[5]();
+    delete[] specified_kind_array_example;
+
+    return 0;
+}
+#else //If C++11 is not available - do nothing.
+int main()
+{
+    std::cout << "WARNING: because your compiler does not support C++11 standard,"
+              << std::endl;
+    std::cout << "this example is only as a dummy placeholder." << std::endl;
+    return 0;
+}
+#endif
diff --git a/examples/memkind_cpp_allocator.cpp b/examples/memkind_cpp_allocator.cpp
new file mode 100644
index 0000000..98b165f
--- /dev/null
+++ b/examples/memkind_cpp_allocator.cpp
@@ -0,0 +1,126 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#include "memkind_allocator.h"
+
+#include <cassert>
+#include <forward_list>
+#include <iostream>
+#include <scoped_allocator>
+#include <set>
+#include <string>
+#include <deque>
+
+#define STL_FWLIST_TEST
+#define STL_DEQUE_TEST
+#if _GLIBCXX_USE_CXX11_ABI
+#define STL_MULTISET_TEST
+#endif
+
+int main(int argc, char *argv[])
+{
+    std::cout << "TEST SCOPE: HELLO" << std::endl;
+
+#ifdef STL_FWLIST_TEST
+    {
+        std::cout << "FORWARD LIST OPEN" << std::endl;
+        libmemkind::static_kind::allocator<int> alc{ libmemkind::kinds::REGULAR };
+        std::forward_list<int, libmemkind::static_kind::allocator<int>> fwlist {alc};
+        int i;
+
+        for (i = 0; i <= 20; ++i) {
+            fwlist.push_front(i);
+        }
+
+        fwlist.sort(std::greater<int>());
+
+        for(auto &el: fwlist) {
+            i--;
+            assert(el == i);
+        }
+
+        fwlist.clear();
+
+        std::cout << "FORWARD LIST CLOSE" << std::endl;
+    }
+#endif
+
+#ifdef STL_DEQUE_TEST
+    std::cout << "DEQUE OPEN" << std::endl;
+    libmemkind::static_kind::allocator<int> int_alc{ libmemkind::kinds::REGULAR };
+    std::deque<int, libmemkind::static_kind::allocator<int>> deque {int_alc};
+
+    for (int i=0; i<10; i++) {
+        deque.push_back(i);
+    }
+
+    assert(deque.size() == 10);
+
+    while(!deque.empty()) {
+        deque.pop_front();
+    }
+
+    assert(deque.size() == 0);
+
+    std::cout << "DEQUE CLOSE" << std::endl;
+#endif
+
+#ifdef STL_MULTISET_TEST
+    {
+        std::cout << "MULTISET OPEN" << std::endl;
+        typedef libmemkind::static_kind::allocator<std::string> multiset_alloc_t;
+        multiset_alloc_t multiset_alloc { libmemkind::kinds::DEFAULT };
+        std::multiset<std::string, std::less<std::string>, std::scoped_allocator_adaptor<multiset_alloc_t> >
+        multiset{ std::scoped_allocator_adaptor<multiset_alloc_t>(multiset_alloc) };
+
+        multiset.insert("S");
+        multiset.insert("A");
+        multiset.insert("C");
+        multiset.insert("S");
+        multiset.insert("C");
+        multiset.insert("E");
+
+        std::string test;
+        for (auto &n : multiset) {
+            test.append(n);
+        }
+        assert(test == "ACCESS");
+
+        std::cout << "MULTISET CLOSE" << std::endl;
+    }
+#endif
+
+    std::cout << "TEST SCOPE: GOODBYE" << std::endl;
+
+    return 0;
+}
diff --git a/examples/memkind_decorator_debug.c b/examples/memkind_decorator_debug.c
new file mode 100644
index 0000000..51183d3
--- /dev/null
+++ b/examples/memkind_decorator_debug.c
@@ -0,0 +1,64 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+
+/* This is an example that enables debug printing on every alloction call */
+
+static void memkind_debug(const char *func, memkind_t kind, size_t size,
+                          void *ptr)
+{
+    fprintf(stderr, "[ DEBUG ] func=%s kind=%p size=%zu ptr=0x%lx\n", func, kind,
+            size, (size_t)ptr);
+}
+
+void memkind_malloc_post(memkind_t kind, size_t size, void **result)
+{
+    memkind_debug("memkind_malloc", kind, size, *result);
+}
+
+void memkind_calloc_post(memkind_t kind, size_t nmemb, size_t size,
+                         void **result)
+{
+    memkind_debug("memkind_calloc", kind, nmemb * size, *result);
+}
+
+void memkind_posix_memalign_post(memkind_t kind, void **memptr,
+                                 size_t alignment, size_t size, int *err)
+{
+    memkind_debug("memkind_posix_memalign", kind, size, *memptr);
+}
+
+void memkind_realloc_post(memkind_t kind, void *ptr, size_t size, void **result)
+{
+    memkind_debug("memkind_realloc", kind, size, *result);
+}
+
+void memkind_free_pre(memkind_t kind, void **ptr)
+{
+    memkind_debug("memkind_free", kind, 0, *ptr);
+}
diff --git a/examples/memkind_get_stat.c b/examples/memkind_get_stat.c
new file mode 100644
index 0000000..7e8d0d5
--- /dev/null
+++ b/examples/memkind_get_stat.c
@@ -0,0 +1,69 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include <stdio.h>
+
+int main()
+{
+    const size_t size = 1024;
+
+    size_t stats_active;
+    size_t stats_resident;
+    size_t stats_allocated;
+    size_t stats_kind_regular_active;
+    size_t stats_kind_regular_resident;
+    size_t stats_kind_regular_allocated;
+
+    char *ptr_regular = (char *)memkind_malloc(MEMKIND_REGULAR, size);
+
+    fprintf(stdout,
+            "This example shows how to use memkind API to retrieve information about allocation stats.\n");
+
+    snprintf(ptr_regular, size,
+             "Hello world from regular kind memory - ptr_regular.\n");
+
+    memkind_update_cached_stats();
+    memkind_get_stat(NULL, MEMKIND_STAT_TYPE_RESIDENT, &stats_resident);
+    memkind_get_stat(NULL, MEMKIND_STAT_TYPE_ACTIVE, &stats_active);
+    memkind_get_stat(NULL, MEMKIND_STAT_TYPE_ALLOCATED, &stats_allocated);
+    memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_RESIDENT,
+                     &stats_kind_regular_resident);
+    memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_ACTIVE,
+                     &stats_kind_regular_active);
+    memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_ALLOCATED,
+                     &stats_kind_regular_allocated);
+
+    fprintf(stdout,
+            "\n Global stats \nresident %zu \nactive %zu, \nallocated %zu \n",
+            stats_resident, stats_active, stats_allocated);
+    fprintf(stdout,
+            "\n MEMKIND_REGULAR stats \nresident %zu \nactive %zu, \nallocated %zu \n",
+            stats_kind_regular_resident, stats_kind_regular_active,
+            stats_kind_regular_allocated);
+
+    memkind_free(NULL, ptr_regular);
+
+    return 0;
+}
diff --git a/examples/pmem_alignment.c b/examples/pmem_alignment.c
new file mode 100644
index 0000000..3c314fc
--- /dev/null
+++ b/examples/pmem_alignment.c
@@ -0,0 +1,132 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+
+static char path[PATH_MAX] = "/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use memkind alignment and how it affects allocations.\nPMEM kind directory: %s\n",
+            path);
+
+    err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str10 = NULL;
+    char *pmem_str11 = NULL;
+
+    // Perform two allocations - 32 bytes
+    pmem_str10 = (char *)memkind_malloc(pmem_kind, 32);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    pmem_str11 = (char *)memkind_malloc(pmem_kind, 32);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11)\n");
+        return 1;
+    }
+
+    // The allocations should be very close to each other in memory
+    if (pmem_str11 - pmem_str10 != 32) {
+        fprintf(stderr, "Something went wrong with allocations.\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind, pmem_str10);
+    memkind_free(pmem_kind, pmem_str11);
+
+    // Perform two allocations - 32 bytes with alignment 64
+    err = memkind_posix_memalign(pmem_kind, (void **)&pmem_str10, 64, 32);
+    if (err) {
+        fprintf(stderr,
+                "Unable to allocate pmem string (pmem_str10) with alignment.\n");
+        return 1;
+    }
+
+    err = memkind_posix_memalign(pmem_kind, (void **)&pmem_str11, 64, 32);
+    if (err) {
+        fprintf(stderr,
+                "Unable to allocate pmem string (pmem_str11) with alignment.\n");
+        return 1;
+    }
+
+    // The addresses of allocations are not close to each other in memory, they are aligned to 64
+    if (pmem_str11 - pmem_str10 != 64) {
+        fprintf(stderr, "Something went wrong with alignment allocations.\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind, pmem_str10);
+    memkind_free(pmem_kind, pmem_str11);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "The memory has been successfully allocated using memkind alignment.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_and_dax_kmem_kind.c b/examples/pmem_and_dax_kmem_kind.c
new file mode 100644
index 0000000..45a544f
--- /dev/null
+++ b/examples/pmem_and_dax_kmem_kind.c
@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    const size_t size = 512;
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2) {
+        if (realpath(argv[1], path) == NULL) {
+            fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+            return 1;
+        }
+    }
+
+    char *ptr_dax_kmem = NULL;
+    char *ptr_pmem = NULL;
+
+    fprintf(stdout,
+            "This example shows how to allocate to PMEM memory using file-backed memory (pmem kind) "
+            "and persistent memory NUMA node (MEMKIND_DAX_KMEM).\nPMEM kind directory: %s\n",
+            path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    ptr_dax_kmem = (char *)memkind_malloc(MEMKIND_DAX_KMEM, size);
+    if (!ptr_dax_kmem) {
+        fprintf(stderr, "Unable allocate 512 bytes in persistent memory NUMA node.\n");
+        return 1;
+    }
+
+    ptr_pmem = (char *)memkind_malloc(pmem_kind, size);
+    if (!ptr_pmem) {
+        fprintf(stderr, "Unable allocate 512 bytes in file-backed memory.\n");
+        return 1;
+    }
+
+    snprintf(ptr_dax_kmem, size,
+             "Hello world from persistent memory NUMA node - ptr_dax_kmem.\n");
+    snprintf(ptr_pmem, size, "Hello world from file-backed memory - ptr_pmem.\n");
+
+    fprintf(stdout, "%s", ptr_dax_kmem);
+    fprintf(stdout, "%s", ptr_pmem);
+
+    memkind_free(MEMKIND_DAX_KMEM, ptr_dax_kmem);
+    memkind_free(pmem_kind, ptr_pmem);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_and_default_kind.c b/examples/pmem_and_default_kind.c
new file mode 100644
index 0000000..4b0ec38
--- /dev/null
+++ b/examples/pmem_and_default_kind.c
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+#include <sys/resource.h>
+
+#define MB (1024 * 1024)
+#define HEAP_LIMIT_SIMULATE (1024 * MB)
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    const size_t size = 512;
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+    errno = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2) {
+        if (realpath(argv[1], path) == NULL) {
+            fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+            return 1;
+        }
+    }
+
+    // Operation below limit the current size of heap
+    // to show different place of allocation
+    const struct rlimit heap_limit = { HEAP_LIMIT_SIMULATE, HEAP_LIMIT_SIMULATE };
+    err = setrlimit(RLIMIT_DATA, &heap_limit);
+    if (err) {
+        fprintf(stderr, "Unable to set heap limit.\n");
+        return 1;
+    }
+
+    char *ptr_default = NULL;
+    char *ptr_default_not_possible = NULL;
+    char *ptr_pmem = NULL;
+
+    fprintf(stdout,
+            "This example shows how to allocate memory using standard memory (MEMKIND_DEFAULT) "
+            "and file-backed kind of memory (PMEM).\nPMEM kind directory: %s\n", path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    ptr_default = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    if (!ptr_default) {
+        fprintf(stderr, "Unable allocate 512 bytes in standard memory.\n");
+        return 1;
+    }
+
+    errno = 0;
+    ptr_default_not_possible = (char *)memkind_malloc(MEMKIND_DEFAULT,
+                                                      HEAP_LIMIT_SIMULATE);
+    if (ptr_default_not_possible) {
+        fprintf(stderr,
+                "Failure, this allocation should not be possible "
+                "(expected result was NULL), because of setlimit function.\n");
+        return 1;
+    }
+    if (errno != ENOMEM) {
+        fprintf(stderr,
+                "Failure, this allocation should set errno to ENOMEM value, because of setlimit function.\n");
+        return 1;
+    }
+
+    errno = 0;
+    ptr_pmem = (char *)memkind_malloc(pmem_kind, HEAP_LIMIT_SIMULATE);
+    if (!ptr_pmem) {
+        fprintf(stderr, "Unable allocate HEAP_LIMIT_SIMULATE in file-backed memory.\n");
+        return 1;
+    }
+    if (errno != 0) {
+        fprintf(stderr, "Failure, this allocation should not set errno value.\n");
+        return 1;
+    }
+
+    sprintf(ptr_default, "Hello world from standard memory - ptr_default.\n");
+    sprintf(ptr_pmem, "Hello world from file-backed memory - ptr_pmem.\n");
+
+    fprintf(stdout, "%s", ptr_default);
+    fprintf(stdout, "%s", ptr_pmem);
+
+    memkind_free(MEMKIND_DEFAULT, ptr_default);
+    memkind_free(pmem_kind, ptr_pmem);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_config.c b/examples/pmem_config.c
new file mode 100644
index 0000000..cd7cc00
--- /dev/null
+++ b/examples/pmem_config.c
@@ -0,0 +1,98 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+
+static char path[PATH_MAX] = "/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use custom configuration to create pmem kind."
+            "\nPMEM kind directory: %s\n", path);
+
+    struct memkind_config *test_cfg = memkind_config_new();
+    if (!test_cfg) {
+        fprintf(stderr, "Unable to create memkind cfg.\n");
+        return 1;
+    }
+
+    memkind_config_set_path(test_cfg, path);
+    memkind_config_set_size(test_cfg, PMEM_MAX_SIZE);
+    memkind_config_set_memory_usage_policy(test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+
+
+    // Create PMEM partition with specific configuration
+    err =  memkind_create_pmem_with_config(test_cfg, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    memkind_config_delete(test_cfg);
+
+    fprintf(stdout,
+            "PMEM kind and configuration was successfully created and destroyed.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_cpp_allocator.cpp b/examples/pmem_cpp_allocator.cpp
new file mode 100644
index 0000000..e3decd1
--- /dev/null
+++ b/examples/pmem_cpp_allocator.cpp
@@ -0,0 +1,167 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#include "pmem_allocator.h"
+
+#include <sys/stat.h>
+#include <iostream>
+#include <vector>
+#include <list>
+#include <map>
+#include <string>
+#include <scoped_allocator>
+#include <cassert>
+
+#define STL_VECTOR_TEST
+#define STL_LIST_TEST
+#if _GLIBCXX_USE_CXX11_ABI
+#define STL_VEC_STRING_TEST
+#define STL_MAP_INT_STRING_TEST
+#endif
+
+void cpp_allocator_test(const char *pmem_directory)
+{
+    std::cout << "TEST SCOPE: HELLO" << std::endl;
+
+    size_t pmem_max_size = 1024*1024*1024;
+
+#ifdef STL_VECTOR_TEST
+    {
+        std::cout << "VECTOR OPEN" << std::endl;
+        libmemkind::pmem::allocator<int> alc{ pmem_directory, pmem_max_size };
+        std::vector<int, libmemkind::pmem::allocator<int>> vector{ alc };
+
+        for (int i = 0; i < 20; ++i) {
+            vector.push_back(0xDEAD + i);
+            assert(vector.back() == 0xDEAD + i);
+        }
+
+        std::cout << "VECTOR CLOSE" << std::endl;
+    }
+#endif
+
+#ifdef STL_LIST_TEST
+    {
+        std::cout << "LIST OPEN" << std::endl;
+        libmemkind::pmem::allocator<int> alc{ pmem_directory, pmem_max_size };
+        std::list<int, libmemkind::pmem::allocator<int>> list{ alc };
+
+        const int nx2 = 4;
+        for (int i = 0; i < nx2; ++i) {
+            list.emplace_back(0xBEAC011 + i);
+            assert(list.back() == 0xBEAC011 + i);
+        }
+
+        for (int i = 0; i < nx2; ++i) {
+            list.pop_back();
+        }
+
+        std::cout << "LIST CLOSE" << std::endl;
+    }
+#endif
+
+#ifdef STL_VEC_STRING_TEST
+    {
+        std::cout << "STRINGED VECTOR OPEN" << std::endl;
+        typedef libmemkind::pmem::allocator<char> str_alloc_t;
+        typedef std::basic_string<char, std::char_traits<char>, str_alloc_t>
+        pmem_string;
+        typedef libmemkind::pmem::allocator<pmem_string> vec_alloc_t;
+
+        vec_alloc_t vec_alloc{ pmem_directory, pmem_max_size };
+        str_alloc_t str_alloc{ pmem_directory, pmem_max_size };
+
+        std::vector<pmem_string, std::scoped_allocator_adaptor<vec_alloc_t> >
+        vec{ std::scoped_allocator_adaptor<vec_alloc_t>(vec_alloc) };
+
+        pmem_string arg{ "Very very loooong striiiing", str_alloc };
+
+        vec.push_back(arg);
+        assert(vec.back() == arg);
+
+        std::cout << "STRINGED VECTOR CLOSE" << std::endl;
+    }
+
+#endif
+
+#ifdef STL_MAP_INT_STRING_TEST
+    {
+        std::cout << "INT_STRING MAP OPEN" << std::endl;
+        typedef std::basic_string<char, std::char_traits<char>, libmemkind::pmem::allocator<char>>
+                pmem_string;
+        typedef int key_t;
+        typedef pmem_string value_t;
+        typedef libmemkind::pmem::allocator<std::pair<const key_t, value_t>>
+                                                                          allocator_t;
+        typedef std::map<key_t, value_t, std::less<key_t>, std::scoped_allocator_adaptor<allocator_t>>
+                map_t;
+
+        allocator_t allocator( pmem_directory, pmem_max_size );
+
+        value_t source_str1("Lorem ipsum dolor ", allocator);
+        value_t source_str2("sit amet consectetuer adipiscing elit", allocator );
+
+        map_t target_map{ std::scoped_allocator_adaptor<allocator_t>(allocator) };
+
+        target_map[key_t(165)] = source_str1;
+        assert(target_map[key_t(165)] == source_str1);
+        target_map[key_t(165)] = source_str2;
+        assert(target_map[key_t(165)] == source_str2);
+
+        std::cout << "INT_STRING MAP CLOSE" << std::endl;
+    }
+#endif
+    std::cout << "TEST SCOPE: GOODBYE" << std::endl;
+}
+
+int main(int argc, char *argv[])
+{
+    const char *pmem_directory = "/tmp/";
+
+    if (argc > 2) {
+        std::cerr << "Usage: pmem_cpp_allocator [directory path]\n"
+                  << "\t[directory path] - directory where temporary file is created (default = \"/tmp/\")"
+                  << std::endl;
+        return 0;
+    } else if (argc == 2) {
+        struct stat st;
+        if (stat(argv[1], &st) != 0 || !S_ISDIR(st.st_mode)) {
+            fprintf(stderr,"%s : Invalid path to pmem kind directory\n", argv[1]);
+            return 1;
+        }
+        pmem_directory = argv[1];
+    }
+
+    cpp_allocator_test(pmem_directory);
+    return 0;
+}
diff --git a/examples/pmem_detect_kind.c b/examples/pmem_detect_kind.c
new file mode 100644
index 0000000..ba4c609
--- /dev/null
+++ b/examples/pmem_detect_kind.c
@@ -0,0 +1,165 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+
+#define MALLOC_SIZE 512U
+#define REALLOC_SIZE 2048U
+#define ALLOC_LIMIT  1000U
+
+static void *alloc_buffer[ALLOC_LIMIT];
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+static int allocate_pmem_and_default_kind(struct memkind *pmem_kind)
+{
+    unsigned i;
+    for(i = 0; i < ALLOC_LIMIT; i++) {
+        if (i%2)
+            alloc_buffer[i] = memkind_malloc(pmem_kind, MALLOC_SIZE);
+        else
+            alloc_buffer[i] = memkind_malloc(MEMKIND_DEFAULT, MALLOC_SIZE);
+
+        if (!alloc_buffer[i]) {
+            return 1;
+        }
+    }
+
+    return 0;
+}
+
+static int realloc_using_get_kind_only_on_pmem()
+{
+    unsigned i;
+    for(i = 0; i < ALLOC_LIMIT; i++) {
+        if (memkind_detect_kind(alloc_buffer[i]) != MEMKIND_DEFAULT) {
+            void *temp = memkind_realloc(NULL, alloc_buffer[i], REALLOC_SIZE);
+            if (!temp) {
+                return 1;
+            }
+            alloc_buffer[i] = temp;
+        }
+    }
+
+    return 0;
+}
+
+
+static int verify_allocation_size(struct memkind *pmem_kind, size_t pmem_size)
+{
+    unsigned i;
+    for(i = 0; i < ALLOC_LIMIT; i++) {
+        void *val = alloc_buffer[i];
+        if (i%2) {
+            if (memkind_malloc_usable_size(pmem_kind, val) != pmem_size ) {
+                return 1;
+            }
+        } else {
+            if (memkind_malloc_usable_size(MEMKIND_DEFAULT, val) != MALLOC_SIZE) {
+                return 1;
+            }
+        }
+    }
+
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to distinguish allocation from different kinds using detect kind function"
+            "\nPMEM kind directory: %s\n", path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Allocate to PMEM and DEFAULT kind.\n");
+
+    if (allocate_pmem_and_default_kind(pmem_kind)) {
+        fprintf(stderr, "allocate_pmem_and_default_kind().\n");
+        return 1;
+    }
+
+    if (verify_allocation_size(pmem_kind, MALLOC_SIZE)) {
+        fprintf(stderr, "verify_allocation_size() before resize.\n");
+        return 1;
+    }
+
+    fprintf(stdout,
+            "Reallocate memory only on PMEM kind using memkind_detect_kind().\n");
+
+    if (realloc_using_get_kind_only_on_pmem()) {
+        fprintf(stderr, "realloc_using_get_kind_only_on_pmem().\n");
+        return 1;
+    }
+
+    if (verify_allocation_size(pmem_kind, REALLOC_SIZE)) {
+        fprintf(stderr, "verify_allocation_size() after resize.\n");
+        return 1;
+    }
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory from PMEM kind was successfully reallocated.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_free_with_unknown_kind.c b/examples/pmem_free_with_unknown_kind.c
new file mode 100644
index 0000000..cedd18a
--- /dev/null
+++ b/examples/pmem_free_with_unknown_kind.c
@@ -0,0 +1,115 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+static const size_t PMEM_PART_SIZE = MEMKIND_PMEM_MIN_SIZE + 4 * 1024;
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char **argv)
+{
+    const size_t size = 512;
+    struct memkind *pmem_kind = NULL;
+    const int arraySize = 100;
+    char *ptr[100] = { NULL };
+    int i = 0;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use memkind_free with unknown kind as a parameter.\n");
+
+    err = memkind_create_pmem(path, PMEM_PART_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    for (i = 0; i < arraySize; ++i) {
+        if (i < 50) {
+            ptr[i] = memkind_malloc(MEMKIND_DEFAULT, size);
+            if (ptr[i] == NULL) {
+                fprintf(stderr, "Unable to allocate memkind default.\n");
+                return 1;
+            }
+        } else {
+            ptr[i] = memkind_malloc(pmem_kind, size);
+            if (ptr[i] == NULL) {
+                fprintf(stderr, "Unable to allocate pmem.\n");
+                return 1;
+            }
+        }
+    }
+    fprintf(stdout,
+            "Memory was successfully allocated in default kind and pmem kind.\n");
+
+    sprintf(ptr[10], "Hello world from standard memory - ptr[10].\n");
+    sprintf(ptr[40], "Hello world from standard memory - ptr[40].\n");
+    sprintf(ptr[80], "Hello world from persistent memory - ptr[80].\n");
+
+    fprintf(stdout, "%s", ptr[10]);
+    fprintf(stdout, "%s", ptr[40]);
+    fprintf(stdout, "%s", ptr[80]);
+
+    fprintf(stdout, "Free memory without specifying kind.\n");
+    for (i = 0; i < arraySize; ++i) {
+        memkind_free(NULL, ptr[i]);
+    }
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully released.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_kinds.c b/examples/pmem_kinds.c
new file mode 100644
index 0000000..0571428
--- /dev/null
+++ b/examples/pmem_kinds.c
@@ -0,0 +1,119 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+#define NUM_KINDS 10
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kinds[NUM_KINDS] = {NULL};
+    struct memkind *pmem_kind = NULL;
+    struct memkind *pmem_kind_unlimited = NULL;
+
+    int err = 0, i = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to create and destroy pmem kind with defined or unlimited size."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create first PMEM partition with specific size
+    err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Create second PMEM partition with unlimited size
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Destroy both PMEM partitions
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    err = memkind_destroy_kind(pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Create many PMEM kinds with same specific size
+    for (i = 0; i < NUM_KINDS; i++) {
+        err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kinds[i]);
+        if (err) {
+            print_err_message(err);
+            return 1;
+        }
+    }
+
+    // Destroy all of them
+    for (i = 0; i < NUM_KINDS; i++) {
+        err = memkind_destroy_kind(pmem_kinds[i]);
+        if (err) {
+            print_err_message(err);
+            return 1;
+        }
+    }
+
+    fprintf(stdout, "PMEM kinds have been successfully created and destroyed.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_malloc.c b/examples/pmem_malloc.c
new file mode 100644
index 0000000..61e201d
--- /dev/null
+++ b/examples/pmem_malloc.c
@@ -0,0 +1,129 @@
+/*
+ * Copyright (c) 2015 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to allocate memory and possibility to exceed pmem kind size."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create PMEM partition with specific size
+    err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str1 = NULL;
+    char *pmem_str2 = NULL;
+    char *pmem_str3 = NULL;
+    char *pmem_str4 = NULL;
+
+    // Allocate 512 Bytes of 32 MB available
+    pmem_str1 = (char *)memkind_malloc(pmem_kind, 512);
+    if (pmem_str1 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str1).\n");
+        return 1;
+    }
+
+    // Allocate 8 MB of 31.9 MB available
+    pmem_str2 = (char *)memkind_malloc(pmem_kind, 8 * 1024 * 1024);
+    if (pmem_str2 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str2).\n");
+        return 1;
+    }
+
+    // Allocate 16 MB of 23.9 MB available
+    pmem_str3 = (char *)memkind_malloc(pmem_kind, 16 * 1024 * 1024);
+    if (pmem_str3 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str3).\n");
+        return 1;
+    }
+
+    // Allocate 16 MB of 7.9 MB available - Out Of Memory expected
+    pmem_str4 = (char *)memkind_malloc(pmem_kind, 16 * 1024 * 1024);
+    if (pmem_str4 != NULL) {
+        fprintf(stderr,
+                "Failure, this allocation should not be possible (expected result was NULL).\n");
+        return 1;
+    }
+
+    sprintf(pmem_str1, "Hello world from pmem - pmem_str1.\n");
+    sprintf(pmem_str2, "Hello world from pmem - pmem_str2.\n");
+    sprintf(pmem_str3, "Hello world from persistent memory - pmem_str3.\n");
+
+    fprintf(stdout, "%s", pmem_str1);
+    fprintf(stdout, "%s", pmem_str2);
+    fprintf(stdout, "%s", pmem_str3);
+
+    memkind_free(pmem_kind, pmem_str1);
+    memkind_free(pmem_kind, pmem_str2);
+    memkind_free(pmem_kind, pmem_str3);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_malloc_unlimited.c b/examples/pmem_malloc_unlimited.c
new file mode 100644
index 0000000..a267069
--- /dev/null
+++ b/examples/pmem_malloc_unlimited.c
@@ -0,0 +1,101 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind_unlimited = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to allocate memory with unlimited kind size."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create PMEM partition with unlimited size
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str10 = NULL;
+    char *pmem_str11 = NULL;
+
+    // Huge allocation
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_unlimited, 32 * 1024 * 1024);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    // Another huge allocation, kind size is only limited by OS resources
+    pmem_str11 = (char *)memkind_malloc(pmem_kind_unlimited, 32 * 1024 * 1024);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11).\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind_unlimited, pmem_str10);
+    memkind_free(pmem_kind_unlimited, pmem_str11);
+
+    err = memkind_destroy_kind(pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/examples/pmem_multithreads.c b/examples/pmem_multithreads.c
new file mode 100644
index 0000000..fe72a43
--- /dev/null
+++ b/examples/pmem_multithreads.c
@@ -0,0 +1,138 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <pthread.h>
+#include <unistd.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+#define NUM_THREADS 10
+
+static char path[PATH_MAX]="/tmp/";
+
+void *thread_ind(void *arg);
+
+static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
+static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use multithreading with independent pmem kinds."
+            "\nPMEM kind directory: %s\n", path);
+
+    pthread_t pmem_threads[NUM_THREADS];
+    int t;
+
+    // Create many independent threads
+    for (t = 0; t < NUM_THREADS; t++) {
+        if (pthread_create(&pmem_threads[t], NULL, thread_ind, NULL) != 0) {
+            fprintf(stderr, "Unable to create a thread.\n");
+            return 1;
+        }
+    }
+
+    sleep(1);
+    if (pthread_cond_broadcast(&cond) != 0) {
+        fprintf(stderr, "Unable to broadcast a condition.\n");
+        return 1;
+    }
+
+    for (t = 0; t < NUM_THREADS; t++) {
+        if (pthread_join(pmem_threads[t], NULL) != 0) {
+            fprintf(stderr, "Thread join failed.\n");
+            return 1;
+        }
+    }
+
+    fprintf(stdout, "Threads successfully allocated memory in the PMEM kinds.\n");
+
+    return 0;
+}
+
+void *thread_ind(void *arg)
+{
+    struct memkind *pmem_kind;
+
+    if (pthread_mutex_lock(&mutex) != 0) {
+        fprintf(stderr, "Failed to acquire mutex.\n");
+        return NULL;
+    }
+    if (pthread_cond_wait(&cond, &mutex) != 0) {
+        fprintf(stderr, "Failed to block mutex on condition.\n");
+        return NULL;
+    }
+    if (pthread_mutex_unlock(&mutex) != 0) {
+        fprintf(stderr, "Failed to release mutex.\n");
+        return NULL;
+    }
+
+    int err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return NULL;
+    }
+
+    void *test = memkind_malloc(pmem_kind, 32);
+    if (test == NULL) {
+        fprintf(stderr, "Unable to allocate pmem (test) in thread.\n");
+        return NULL;
+    }
+
+    memkind_free(pmem_kind, test);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+    }
+
+    return NULL;
+}
diff --git a/examples/pmem_multithreads_onekind.c b/examples/pmem_multithreads_onekind.c
new file mode 100644
index 0000000..ed7bc07
--- /dev/null
+++ b/examples/pmem_multithreads_onekind.c
@@ -0,0 +1,167 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <pthread.h>
+#include <unistd.h>
+
+#define NUM_THREADS 10
+#define NUM_ALLOCS 100
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+struct arg_struct {
+    int id;
+    struct memkind *kind;
+    int **ptr;
+};
+
+void *thread_onekind(void *arg);
+
+static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
+static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind_unlimited = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use multithreading with one main pmem kind."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create PMEM partition with unlimited size
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Create a few threads which will access to our main pmem_kind
+    pthread_t pmem_threads[NUM_THREADS];
+    int *pmem_tint[NUM_THREADS][NUM_ALLOCS];
+    int t = 0, i = 0;
+
+    struct arg_struct *args[NUM_THREADS];
+
+    for (t = 0; t<NUM_THREADS; t++) {
+        args[t] = malloc(sizeof(struct arg_struct));
+        args[t]->id = t;
+        args[t]->ptr = &pmem_tint[t][0];
+        args[t]->kind = pmem_kind_unlimited;
+
+        if (pthread_create(&pmem_threads[t], NULL, thread_onekind,
+                           (void *)args[t])!= 0) {
+            fprintf(stderr, "Unable to create a thread.\n");
+            return 1;
+        }
+    }
+
+    sleep(1);
+    if (pthread_cond_broadcast(&cond) != 0) {
+        fprintf(stderr, "Unable to broadcast a condition.\n");
+        return 1;
+    }
+
+    for (t = 0; t < NUM_THREADS; t++) {
+        if (pthread_join(pmem_threads[t], NULL) != 0) {
+            fprintf(stderr, "Thread join failed.\n");
+            return 1;
+        }
+    }
+
+    // Check if we can read the values outside of threads and free resources
+    for (t = 0; t < NUM_THREADS; t++) {
+        for (i = 0; i < NUM_ALLOCS; i++) {
+            if(*pmem_tint[t][i] != t) {
+                fprintf(stderr,
+                        "pmem_tint value has not been saved correctly in the thread.\n");
+                return 1;
+            }
+            memkind_free(args[t]->kind, *(args[t]->ptr+i));
+        }
+        free(args[t]);
+    }
+
+    fprintf(stdout, "Threads successfully allocated memory in the PMEM kind.\n");
+
+    return 0;
+}
+
+void *thread_onekind(void *arg)
+{
+    struct arg_struct *args = (struct arg_struct *)arg;
+    int i;
+
+    if (pthread_mutex_lock(&mutex) != 0) {
+        fprintf(stderr, "Failed to acquire mutex.\n");
+        return NULL;
+    }
+    if (pthread_cond_wait(&cond, &mutex) != 0) {
+        fprintf(stderr, "Failed to block mutex on condition.\n");
+        return NULL;
+    }
+    if (pthread_mutex_unlock(&mutex) != 0) {
+        fprintf(stderr, "Failed to release mutex.\n");
+        return NULL;
+    }
+
+    // Lets alloc int and put there thread ID
+    for (i = 0; i < NUM_ALLOCS; i++) {
+        *(args->ptr+i) = (int *)memkind_malloc(args->kind, sizeof(int));
+        if (*(args->ptr+i) == NULL) {
+            fprintf(stderr, "Unable to allocate pmem int.\n");
+            return NULL;
+        }
+        **(args->ptr+i) = args->id;
+    }
+
+    return NULL;
+}
diff --git a/examples/pmem_usable_size.c b/examples/pmem_usable_size.c
new file mode 100644
index 0000000..92dd078
--- /dev/null
+++ b/examples/pmem_usable_size.c
@@ -0,0 +1,159 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define MB (1024 * 1024)
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind_unlimited = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows difference between the expected and the actual allocation size."
+            "\nPMEM kind directory: %s\n", path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str10 = NULL;
+    char *pmem_str11 = NULL;
+    char *pmem_str12 = NULL;
+
+    // 32 bytes allocation
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_unlimited, 32);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str10) != 32) {
+        fprintf(stderr, "Wrong usable size for small allocation (pmem_str10).\n");
+        return 1;
+    }
+
+    // 31 bytes allocation
+    pmem_str11 = (char *)memkind_malloc(pmem_kind_unlimited, 31);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation, its 32 again
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str11) != 32) {
+        fprintf(stderr, "Wrong usable size for small allocation (pmem_str11).\n");
+        return 1;
+    }
+
+    // 33 bytes allocation
+    pmem_str12 = (char *)memkind_malloc(pmem_kind_unlimited, 33);
+    if (pmem_str12 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str12).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation, its 48 now
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str12) != 48) {
+        fprintf(stderr, "Wrong usable size for small allocation (pmem_str12).\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind_unlimited, pmem_str10);
+    memkind_free(pmem_kind_unlimited, pmem_str11);
+    memkind_free(pmem_kind_unlimited, pmem_str12);
+
+    // 5MB allocation
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_unlimited, 5 * MB);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str10) != 5 * MB) {
+        fprintf(stderr, "Wrong usable size for large allocation (pmem_str10).\n");
+        return 1;
+    }
+
+    // 5MB + 1B allocation
+    pmem_str11 = (char *)memkind_malloc(pmem_kind_unlimited, 5 * MB + 1);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation, its 6MB now
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str11) != 6 * MB) {
+        fprintf(stderr, "Wrong usable size for large allocation (pmem_str11).\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind_unlimited, pmem_str10);
+    memkind_free(pmem_kind_unlimited, pmem_str11);
+
+    err = memkind_destroy_kind(pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "The real size of the allocation has been successfully read.\n");
+
+    return 0;
+}
diff --git a/include/hbw_allocator.h b/include/hbw_allocator.h
new file mode 100644
index 0000000..ec9c1c0
--- /dev/null
+++ b/include/hbw_allocator.h
@@ -0,0 +1,134 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <hbwmalloc.h>
+
+#include <stddef.h>
+#include <new>
+/*
+ * Header file for the C++ allocator compatible with the C++ standard library allocator concepts.
+ * More details in hbwallocator(3) man page.
+ * Note: memory heap management is based on hbwmalloc, refer to the hbwmalloc man page for more information.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+namespace hbw
+{
+
+    template <class T>
+    class allocator
+    {
+    public:
+        /*
+         *  Public member types required and defined by the standard library allocator concepts.
+         */
+        typedef size_t size_type;
+        typedef ptrdiff_t difference_type;
+        typedef T *pointer;
+        typedef const T *const_pointer;
+        typedef T &reference;
+        typedef const T &const_reference;
+        typedef T value_type;
+
+        template <class U>
+        struct rebind {
+            typedef hbw::allocator<U> other;
+        };
+
+        /*
+         *  Public member functions required and defined by the standard library allocator concepts.
+         */
+        allocator() throw() { }
+
+        template <class U>
+        allocator(const allocator<U> &) throw() { }
+
+        ~allocator() throw() { }
+
+        pointer address(reference x) const
+        {
+            return &x;
+        }
+
+        const_pointer address(const_reference x) const
+        {
+            return &x;
+        }
+
+        /*
+         *  Allocates n*sizeof(T) bytes of high bandwidth memory using hbw_malloc().
+         *  Throws std::bad_alloc when cannot allocate memory.
+         */
+        pointer allocate(size_type n, const void * = 0)
+        {
+            if (n > this->max_size()) {
+                throw std::bad_alloc();
+            }
+            pointer result = static_cast<pointer>(hbw_malloc(n * sizeof(T)));
+            if (!result) {
+                throw std::bad_alloc();
+            }
+            return result;
+        }
+
+        /*
+         *  Deallocates memory associated with pointer returned by allocate() using hbw_free().
+         */
+        void deallocate(pointer p, size_type n)
+        {
+            hbw_free(static_cast<void *>(p));
+        }
+
+        size_type max_size() const throw()
+        {
+            return size_t(-1) / sizeof(T);
+        }
+
+        void construct(pointer p, const_reference val)
+        {
+            ::new(p) value_type(val);
+        }
+
+        void destroy(pointer p)
+        {
+            p->~T();
+        }
+    };
+
+    template <class T, class U>
+    bool operator==(const allocator<T> &, const allocator<U> &)
+    {
+        return true;
+    }
+
+    template <class T, class U>
+    bool operator!=(const allocator<T> &, const allocator<U> &)
+    {
+        return false;
+    }
+
+}
diff --git a/include/hbwmalloc.h b/include/hbwmalloc.h
new file mode 100644
index 0000000..3e8eac8
--- /dev/null
+++ b/include/hbwmalloc.h
@@ -0,0 +1,245 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stdlib.h>
+/*
+ *  Header file for the high bandwidth memory interface.
+ *
+ *  This file defines the external API's and enumerations for the
+ *  hbwmalloc library.  These interfaces define a heap manager that
+ *  targets the high bandwidth memory numa nodes.
+ *
+ *  hbwmalloc.h functionality is considered as stable API (STANDARD API).
+ *
+ *  Please read hbwmalloc(3) man page for or more details.
+ */
+
+/*
+ *  Fallback policy.
+ *
+ *  Policy that determines behavior when there is not enough free high
+ *  bandwidth memory to satisfy a user request.  This enum is used with
+ *  hbw_get_policy() and hbw_set_policy().
+ */
+typedef enum {
+    /*
+     *  If insufficient high bandwidth memory pages on nearest NUMA node are
+     *  available then OOM killer will be triggered.
+     */
+    HBW_POLICY_BIND = 1,
+    /*
+     *  If insufficient high bandwidth memory pages are available fall
+     *  back on standard memory pages.
+     */
+    HBW_POLICY_PREFERRED = 2,
+    /*
+     *  Interleave pages across high bandwidth nodes. If insufficient memory
+     *  pages are available then OOM killer will be triggered.
+     */
+    HBW_POLICY_INTERLEAVE = 3,
+    /*
+     *  If insufficient high bandwidth memory pages are available then
+     *  OOM killer will be triggered.
+     */
+    HBW_POLICY_BIND_ALL = 4,
+} hbw_policy_t;
+
+/*
+ *  Page size selection.
+ *
+ *  The hbw_posix_memalign_psize() API gives the user the option to
+ *  select the page size from this enumerated list.
+ */
+typedef enum {
+
+    /*
+     * The four kilobyte page size option. Note that with transparent huge
+     * pages enabled these allocations may be promoted by the operating system
+     * to two megabyte pages.
+     */
+    HBW_PAGESIZE_4KB           = 1,
+
+    /*
+     * The two megabyte page size option.
+     */
+    HBW_PAGESIZE_2MB           = 2,
+    /*
+     * This option is deprecated.
+     * Allocate high bandwidth memory using 1GB chunks backed by huge pages.
+     */
+    HBW_PAGESIZE_1GB_STRICT    = 3,
+
+    /*
+     * This option is deprecated.
+     * Allocate high bandwidth memory using 1GB chunks backed by huge pages.
+     */
+    HBW_PAGESIZE_1GB           = 4,
+
+    /*
+     * Helper representing value of the last enum element incremented by 1.
+     * Shall not be treated as a valid value for functions taking hbw_pagesize_t
+     * as parameter.
+     */
+    HBW_PAGESIZE_MAX_VALUE
+} hbw_pagesize_t;
+
+/*
+ * Flags for hbw_verify_ptr function
+ */
+enum {
+
+    /*
+     * This option touches first byte of all pages in address range starting from "addr" to "addr" + "size"
+     * by read and write (so the content will be overwritten by the same data as it was read).
+     */
+    HBW_TOUCH_PAGES      = (1 << 0)
+};
+
+/*
+ * Returns the current fallback policy when insufficient high bandwidth memory
+ * is available.
+ */
+hbw_policy_t hbw_get_policy(void);
+
+/*
+ * Sets the current fallback policy. The policy can be modified only once in
+ * the lifetime of an application and before calling hbw_*alloc() or
+ * hbw_posix_memalign*() function.
+ * Note: If the policy is not set, than HBW_POLICY_PREFERRED will be used by
+ * default.
+ *
+ * Returns:
+ *   0: on success
+ *   EPERM: if hbw_set_policy () was called more than once
+ *   EINVAL: if mode argument was neither HBW_POLICY_PREFERRED, HBW_POLICY_BIND, HBW_POLICY_BIND_ALL nor HBW_POLICY_INTERLEAVE
+ */
+int hbw_set_policy(hbw_policy_t mode);
+
+/*
+ * Verifies high bandwidth memory availability.
+ * Returns:
+ *   0: if high bandwidth memory is available
+ *   ENODEV: if high-bandwidth memory is unavailable.
+ */
+int hbw_check_available(void);
+
+/*
+ * Verifies if allocated memory fully fall into high bandwidth memory.
+ * Returns:
+ *   0: if memory in address range from "addr" to "addr" + "size" is allocated in high bandwidth memory
+ *   -1: if any region of memory was not allocated in high bandwidth memory
+ *   EINVAL: if addr is NULL, size equals 0 or flags contained unsupported bit set
+ *   EFAULT: could not verify memory
+ */
+int hbw_verify_memory_region(void *addr, size_t size, int flags);
+
+/*
+ * Allocates size bytes of uninitialized high bandwidth memory.
+ * The allocated space is  suitably  aligned (after  possible  pointer
+ * coercion) for storage of any type of object. If size is zero then
+ * hbw_malloc() returns NULL.
+ */
+void *hbw_malloc(size_t size);
+
+/*
+ * Allocates space for num objects in high bandwidth memory, each size bytes
+ * in length.
+ * The result is identical to calling hbw_malloc() with an argument of
+ * num*size, with the exception that the allocated memory is explicitly
+ * initialized to zero bytes.
+ * If num or size is 0, then hbw_calloc() returns NULL.
+ */
+void *hbw_calloc(size_t num, size_t size);
+
+/*
+ * Allocates size bytes of high bandwidth memory such that the allocation's
+ * base address is an even multiple of alignment, and returns the allocation
+ * in the value pointed to by memptr.  The requested alignment must be a power
+ * of 2 at least as large as sizeof(void *).
+ * Returns:
+ *   0: on success
+ *   ENOMEM: if there was insufficient memory to satisfy the request
+ *   EINVAL: if the alignment parameter was not a power of two, or was less than sizeof(void *)
+ */
+int hbw_posix_memalign(void **memptr, size_t alignment, size_t size);
+
+/*
+ * Allocates size bytes of high bandwidth memory such that the allocation's
+ * base address is an even multiple of alignment, and returns the allocation
+ * in the value pointed to by memptr. The requested alignment must be a power
+ * of 2 at least as large as sizeof(void  *). The memory will be allocated
+ * using pages determined by the pagesize variable.
+ * Returns:
+ *   0: on success
+ *   ENOMEM: if there was insufficient memory to satisfy the request
+ *   EINVAL: if the alignment parameter was not a power of two, or was less than sizeof(void *)
+ */
+int hbw_posix_memalign_psize(void **memptr, size_t alignment, size_t size,
+                             hbw_pagesize_t pagesize);
+
+/*
+ * Changes the size of the previously allocated memory referenced by ptr to
+ * size bytes of the specified kind. The contents of the memory are unchanged
+ * up to the lesser of the new and old size.
+ * If the new size is larger, the contents of the newly allocated portion
+ * of the memory are undefined.
+ * Upon success, the memory referenced by ptr is freed and a pointer to the
+ * newly allocated high bandwidth memory is returned.
+ * Note: memkind_realloc() may move the memory allocation, resulting in a
+ * different return value than ptr.
+ * If ptr is NULL, the hbw_realloc() function behaves identically to
+ * hbw_malloc() for the specified size. The address ptr, if not NULL,
+ * was returned by a previous call to hbw_malloc(), hbw_calloc(),
+ * hbw_realloc(), or hbw_posix_memalign(). Otherwise, or if hbw_free(ptr)
+ * was called before, undefined behavior occurs.
+ * Note: hbw_realloc() cannot be used with a pointer returned by
+ * hbw_posix_memalign_psize().
+ */
+void *hbw_realloc(void *ptr, size_t size);
+
+/*
+ * Causes the allocated memory referenced by ptr to be made
+ * available for future allocations. If ptr is NULL, no action occurs.
+ * The address ptr, if not NULL, must have been returned by a previous call
+ * to hbw_malloc(), hbw_calloc(), hbw_realloc(), hbw_posix_memalign(), or
+ * hbw_posix_memalign_psize(). Otherwise, if hbw_free(ptr) was called before,
+ * undefined behavior occurs.
+ */
+void hbw_free(void *ptr);
+
+/*
+ * Returns the number of usable bytes in the block pointed to by ptr, a pointer
+ * to a block of memory allocated by hbw_malloc(), hbw_calloc(), hbw_realloc(),
+ * hbw_posix_memalign(), or hbw_posix_memalign_psize().
+ */
+size_t hbw_malloc_usable_size(void *ptr);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind.h b/include/memkind.h
new file mode 100644
index 0000000..0be640e
--- /dev/null
+++ b/include/memkind.h
@@ -0,0 +1,433 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <sys/types.h>
+
+/**
+ * Header file for the memkind heap manager.
+ * More details in memkind(3) man page.
+ *
+ * API standards are described in memkind(3) man page.
+ */
+
+/// \warning EXPERIMENTAL API
+#define _MEMKIND_BIT(N) (1ull << N)
+
+/// \brief Memkind memory types
+/// \warning EXPERIMENTAL API
+typedef enum memkind_memtype_t {
+
+    /**
+     * Select standard memory, the same as process use.
+     */
+    MEMKIND_MEMTYPE_DEFAULT = _MEMKIND_BIT(0),
+
+    /**
+     * Select high bandwidth memory (HBM).
+     * There must be at least two memories with different bandwidth to
+     * determine the HBM.
+     */
+    MEMKIND_MEMTYPE_HIGH_BANDWIDTH = _MEMKIND_BIT(1)
+
+} memkind_memtype_t;
+
+#undef _MEMKIND_BIT
+
+/// \brief Memkind policy
+/// \warning EXPERIMENTAL API
+typedef enum memkind_policy_t {
+
+    /**
+     * Allocate local memory.
+     * If there is not enough memory to satisfy the request errno is set to
+     * ENOMEM and the allocated pointer is set to NULL.
+     */
+    MEMKIND_POLICY_BIND_LOCAL = 0,
+
+    /**
+     * Memory locality is ignored.
+     * If there is not enough memory to satisfy the request errno is set to
+     * ENOMEM and the allocated pointer is set to NULL.
+     */
+    MEMKIND_POLICY_BIND_ALL,
+
+    /**
+     * Allocate preferred memory that is local.
+     * If there is not enough preferred memory to satisfy the request or
+     * preferred memory is not available, the allocation will fall back on any
+     * other memory.
+     */
+    MEMKIND_POLICY_PREFERRED_LOCAL,
+
+    /**
+     * Interleave allocation across local memory.
+     * For n memory types the allocation will be interleaved across all of
+     * them.
+     */
+    MEMKIND_POLICY_INTERLEAVE_LOCAL,
+
+    /**
+     * Interleave allocation. Locality is ignored.
+     * For n memory types the allocation will be interleaved across all of
+     * them.
+     */
+    MEMKIND_POLICY_INTERLEAVE_ALL,
+
+    /**
+     * Max policy value.
+     */
+    MEMKIND_POLICY_MAX_VALUE
+
+} memkind_policy_t;
+
+/// \brief Memkind bits definition
+/// \warning EXPERIMENTAL API
+/// \note The bits specify flags and masks. Bits <0,1,2,...,7> are reserved for page size, where page sizes are encoded
+///       by base-2 logarithm. If the page size bits are set to zero value, than default page size will be used.
+typedef enum memkind_bits_t {
+    MEMKIND_MASK_PAGE_SIZE_2MB = 21ull,  /**<  Allocations backed by 2 MB page size (2^21 = 2MB) */
+} memkind_bits_t;
+
+/// \brief Memkind type definition
+/// \warning EXPERIMENTAL API
+typedef struct memkind *memkind_t;
+
+
+/// \brief Memkind constant values
+/// \warning EXPERIMENTAL API
+enum memkind_const {
+    MEMKIND_MAX_KIND = 512,                     /**<  Maximum number of kinds */
+    MEMKIND_ERROR_MESSAGE_SIZE = 128,           /**<  Error message size */
+    MEMKIND_PMEM_MIN_SIZE = (1024 * 1024 * 16)  /**<  The minimum size which allows to limit the file-backed memory partition */
+};
+
+/// \brief Memkind operation statuses
+/// \warning EXPERIMENTAL API
+enum {
+    MEMKIND_SUCCESS = 0,                        /**<  Operation success */
+    MEMKIND_ERROR_UNAVAILABLE = -1,             /**<  Error: Memory kind is not available */
+    MEMKIND_ERROR_MBIND = -2,                   /**<  Error: Call to mbind() failed */
+    MEMKIND_ERROR_MMAP  = -3,                   /**<  Error: Call to mmap() failed */
+    MEMKIND_ERROR_MALLOC = -6,                  /**<  Error: Call to malloc() failed */
+    MEMKIND_ERROR_ENVIRON = -12,                /**<  Error: Unable to parse environment variable */
+    MEMKIND_ERROR_INVALID = -13,                /**<  Error: Invalid argument */
+    MEMKIND_ERROR_TOOMANY = -15,                /**<  Error: Attempt to initialize more than MEMKIND_MAX_KIND number of kinds */
+    MEMKIND_ERROR_BADOPS = -17,                 /**<  Error: Invalid memkind_ops structure */
+    MEMKIND_ERROR_HUGETLB = -18,                /**<  Error: Unable to allocate huge pages */
+    MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE = -20,  /**<  Error: Requested memory type is not available */
+    MEMKIND_ERROR_OPERATION_FAILED = -21,       /**<  Error: Operation failed */
+    MEMKIND_ERROR_ARENAS_CREATE = -22,          /**<  Error: Call to jemalloc's arenas.create failed */
+    MEMKIND_ERROR_RUNTIME = -255                /**<  Error: Unspecified run-time error */
+};
+
+/* KIND CONFIGURATION MANAGEMENT INTERFACE */
+
+/// \brief Memkind memory usage policy
+typedef enum memkind_mem_usage_policy {
+    MEMKIND_MEM_USAGE_POLICY_DEFAULT      = 0,        /**<  Default memory usage */
+    MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE = 1,        /**<  Minimize memory usage at all costs, */
+    MEMKIND_MEM_USAGE_POLICY_MAX_VALUE
+} memkind_mem_usage_policy;
+
+/// \brief Memkind memory statistics type
+typedef enum memkind_stat_type {
+    MEMKIND_STAT_TYPE_RESIDENT      = 0,        /**<  Maximum number of bytes in physically resident data pages mapped */
+    MEMKIND_STAT_TYPE_ACTIVE        = 1,        /**<  Total number of bytes in active pages */
+    MEMKIND_STAT_TYPE_ALLOCATED     = 2,        /**<  Total number of allocated bytes */
+    MEMKIND_STAT_TYPE_MAX_VALUE
+} memkind_stat_type;
+
+/// \brief Forward declaration of memkind configuration
+struct memkind_config;
+
+///
+/// \brief Create a memkind configuration
+/// \note STANDARD API
+/// \return Memkind configuration, NULL on failure
+///
+struct memkind_config *memkind_config_new(void);
+
+///
+/// \brief Delete memkind configuration
+/// \note STANDARD API
+/// \param cfg memkind configuration
+///
+void memkind_config_delete(struct memkind_config *cfg);
+
+///
+/// \brief Update memkind configuration with path to specified directory parameter
+/// \note STANDARD API
+/// \param cfg memkind configuration
+/// \param pmem_dir path to specified directory for PMEM kind
+///
+void memkind_config_set_path(struct memkind_config *cfg, const char *pmem_dir);
+
+///
+/// \brief Update memkind configuration with PMEM kind size
+/// \note STANDARD API
+/// \param cfg memkind configuration
+/// \param pmem_size size limit for PMEM kind
+///
+void memkind_config_set_size(struct memkind_config *cfg, size_t pmem_size);
+
+///
+/// \brief Update memkind configuration with memory usage policy parameter
+/// \note STANDARD API
+/// \param cfg memkind configuration
+/// \param policy memkind memory usage policy
+///
+void memkind_config_set_memory_usage_policy(struct memkind_config *cfg,
+                                            memkind_mem_usage_policy policy);
+
+///
+/// \brief Create kind that allocates memory with specific memory type, memory binding policy and flags.
+/// \warning EXPERIMENTAL API
+/// \note Currently implemented memory type and policy configurations:
+///.      {MEMKIND_MEMTYPE_DEFAULT, MEMKIND_POLICY_PREFERRED_LOCAL},
+///.      {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_BIND_LOCAL},
+///       {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_PREFERRED_LOCAL},
+///       {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_INTERLEAVE_ALL},
+///       {MEMKIND_MEMTYPE_DEFAULT | MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_INTERLEAVE_ALL}.
+/// \param memtype_flags determine the memory types to allocate from by combination of memkind_memtype_t values.
+///        This field cannot have zero value.
+/// \param policy specify policy for page binding to memory types selected by  memtype_flags.
+///        This field must be set to memkind_policy_t value. If policy is set to MEMKIND_POLICY_PREFERRED_LOCAL then only one memory
+///        type must be selected. Note: the value cannot be set to MEMKIND_POLICY_MAX_VALUE.
+/// \param flags the field must be set to a combination of memkind_bits_t values.
+/// \param kind pointer to kind which will be created
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE or MEMKIND_ERROR_INVALID on failure
+///
+int memkind_create_kind(memkind_memtype_t memtype_flags,
+                        memkind_policy_t policy,
+                        memkind_bits_t flags,
+                        memkind_t *kind);
+
+///
+/// \brief Destroy previously created kind object, which must have been returned
+///        by a call to memkind_create_kind() or memkind_create_pmem().
+///        The function has undefined behavior when the handle is invalid or
+///        memkind_destroy_kind(kind) was already called before
+/// \warning EXPERIMENTAL API
+/// \note if the kind was returned by memkind_create_kind() all allocated memory must be freed
+///       before kind is destroyed, otherwise this will cause memory leak.
+/// \param kind specified memory kind
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, MEMKIND_ERROR_OPERATION_FAILED on failure
+///
+int memkind_destroy_kind(memkind_t kind);
+
+///
+/// \brief Get kind associated with allocated memory referenced by ptr
+/// \note STANDARD API
+/// \note This function has non-trivial performance overhead
+/// \param ptr pointer to the allocated memory
+/// \return Kind associated with allocated memory, NULL on failure
+///
+memkind_t memkind_detect_kind(void *ptr);
+
+#include "memkind_deprecated.h"
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_REGULAR;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_DEFAULT;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_ALL;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_PREFERRED;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_ALL_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_PREFERRED_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_INTERLEAVE;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_INTERLEAVE;
+
+/// \note STANDARD API
+extern memkind_t MEMKIND_DAX_KMEM;
+
+/// \note STANDARD API
+extern memkind_t MEMKIND_DAX_KMEM_ALL;
+
+/// \note STANDARD API
+extern memkind_t MEMKIND_DAX_KMEM_PREFERRED;
+
+///
+/// \brief Get Memkind API version
+/// \note STANDARD API
+/// \return Version number represented by a single integer number(major * 1000000 + minor * 1000 + patch)
+///
+int memkind_get_version();
+
+///
+/// \brief Convert error number into an error message
+/// \note STANDARD API
+/// \param err error number
+/// \param msg error message
+/// \param size size of message
+///
+void memkind_error_message(int err, char *msg, size_t size);
+
+///
+/// \brief Create a new PMEM (file-backed) kind of given size on top of a temporary file
+///        in the given directory dir
+/// \note STANDARD API
+/// \param dir path to specified directory to temporary file
+/// \param max_size size limit for kind
+/// \param kind pointer to kind which will be created
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_create_pmem(const char *dir, size_t max_size, memkind_t *kind);
+
+///
+/// \brief Create a new PMEM kind with given memkind configuration
+/// \note STANDARD API
+/// \param cfg memkind configuration for specifying PMEM parameters
+/// \param kind pointer to kind which will be created
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_create_pmem_with_config(struct memkind_config *cfg,
+                                    memkind_t *kind);
+
+///
+/// \brief Check if kind is available
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_check_available(memkind_t kind);
+
+///
+/// \brief Update memkind cached statistics
+/// \note STANDARD API
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_update_cached_stats(void);
+
+///
+/// \brief Get memkind statistic
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param stat specified type of memory statistic
+/// \param value reference to value of memory statistic
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_get_stat(memkind_t kind, memkind_stat_type stat, size_t *value);
+
+/* HEAP MANAGEMENT INTERFACE */
+
+///
+/// \brief Allocates size bytes of uninitialized storage of the specified kind
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param size number of bytes to allocate
+/// \return Pointer to the allocated memory
+///
+void *memkind_malloc(memkind_t kind, size_t size);
+
+///
+/// \brief Obtain size of block of memory allocated with the memkind API
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the allocated memory
+/// \return Number of usable bytes
+///
+size_t memkind_malloc_usable_size(memkind_t kind, void *ptr);
+
+///
+/// \brief Allocates memory of the specified kind for an array of num elements
+///        of size bytes each and initializes all bytes in the allocated storage to zero
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param num number of objects
+/// \param size specified size of each element
+/// \return Pointer to the allocated memory
+///
+void *memkind_calloc(memkind_t kind, size_t num, size_t size);
+
+///
+/// \brief Allocates size bytes of the specified kind and places the address of the allocated memory
+///        in *memptr. The address of the allocated memory will be a multiple of alignment,
+///        which must be a power of two and a multiple of sizeof(void *)
+/// \note EXPERIMENTAL API
+/// \param kind specified memory kind
+/// \param memptr address of the allocated memory
+/// \param alignment specified alignment of bytes
+/// \param size specified size of bytes
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, EINVAL or ENOMEM on failure
+///
+int memkind_posix_memalign(memkind_t kind, void **memptr, size_t alignment,
+                           size_t size);
+
+///
+/// \brief Reallocates memory of the specified kind
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the memory block to be reallocated
+/// \param size new size for the memory block in bytes
+/// \return Pointer to the allocated memory
+///
+void *memkind_realloc(memkind_t kind, void *ptr, size_t size);
+
+///
+/// \brief Free the memory space of the specified kind pointed by ptr
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the allocated memory
+///
+void memkind_free(memkind_t kind, void *ptr);
+
+///
+/// \brief Try to reallocate allocation to reduce fragmentation
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the allocated memory
+/// \return Pointer to newly transferred allocated memory
+///
+void *memkind_defrag_reallocate(memkind_t kind, void *ptr);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/heap_manager.h b/include/memkind/internal/heap_manager.h
new file mode 100644
index 0000000..2686f1f
--- /dev/null
+++ b/include/memkind/internal/heap_manager.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <memkind.h>
+
+void heap_manager_init(struct memkind *kind);
+void heap_manager_free(void *ptr);
+size_t heap_manager_malloc_usable_size(void *ptr);
+void *heap_manager_realloc(void *ptr, size_t size);
+struct memkind *heap_manager_detect_kind(void *ptr);
+int heap_manager_update_cached_stats(void);
+int heap_manager_get_stat(memkind_stat_type stat, size_t *value);
+void *heap_manager_defrag_reallocate(void *ptr);
diff --git a/include/memkind/internal/memkind_arena.h b/include/memkind/internal/memkind_arena.h
new file mode 100644
index 0000000..210c3b2
--- /dev/null
+++ b/include/memkind/internal/memkind_arena.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+#include <jemalloc/jemalloc.h>
+#include <memkind/internal/memkind_private.h>
+
+/*
+ * Header file for the jemalloc arena allocation memkind operations.
+ * More details in memkind_arena(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+struct memkind *get_kind_by_arena(unsigned arena_ind);
+struct memkind *memkind_arena_detect_kind(void *ptr);
+int memkind_arena_create(struct memkind *kind, struct memkind_ops *ops,
+                         const char *name);
+int memkind_arena_create_map(struct memkind *kind, extent_hooks_t *hooks);
+int memkind_arena_destroy(struct memkind *kind);
+void *memkind_arena_malloc(struct memkind *kind, size_t size);
+void *memkind_arena_calloc(struct memkind *kind, size_t num, size_t size);
+int memkind_arena_posix_memalign(struct memkind *kind, void **memptr,
+                                 size_t alignment, size_t size);
+void *memkind_arena_realloc(struct memkind *kind, void *ptr, size_t size);
+void *memkind_arena_realloc_with_kind_detect(void *ptr, size_t size);
+int memkind_bijective_get_arena(struct memkind *kind, unsigned int *arena,
+                                size_t size);
+int memkind_thread_get_arena(struct memkind *kind, unsigned int *arena,
+                             size_t size);
+int memkind_arena_finalize(struct memkind *kind);
+void memkind_arena_init(struct memkind *kind);
+void memkind_arena_free(struct memkind *kind, void *ptr);
+void memkind_arena_free_with_kind_detect(void *ptr);
+size_t memkind_arena_malloc_usable_size(void *ptr);
+int memkind_arena_update_memory_usage_policy(struct memkind *kind,
+                                             memkind_mem_usage_policy policy);
+int memkind_arena_enable_background_threads(size_t threads_limit);
+int memkind_arena_update_cached_stats(void);
+int memkind_arena_get_kind_stat(struct memkind *kind,
+                                memkind_stat_type stat_type,
+                                size_t *stat);
+int memkind_arena_get_stat_with_check_init(struct memkind *kind,
+                                           memkind_stat_type stat, bool check_init, size_t *value);
+int memkind_arena_get_global_stat(memkind_stat_type stat_type, size_t *stat);
+void *memkind_arena_defrag_reallocate(struct memkind *kind, void *ptr);
+void *memkind_arena_defrag_reallocate_with_kind_detect(void *ptr);
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_bandwidth.h b/include/memkind/internal/memkind_bandwidth.h
new file mode 100644
index 0000000..e741a0b
--- /dev/null
+++ b/include/memkind/internal/memkind_bandwidth.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind/internal/vec.h>
+#include <numa.h>
+#include <stdbool.h>
+
+struct vec_cpu_node;
+
+typedef int (*get_node_bitmask)(struct bitmask *);
+typedef int (*fill_bandwidth_values)(int *);
+
+int bandwidth_fill(int *bandwidth, get_node_bitmask get_bitmask);
+int set_closest_numanode(fill_bandwidth_values fill, const char *env,
+                         struct vec_cpu_node **closest_numanode, int num_cpu, bool is_single_node);
+void set_bitmask_for_all_closest_numanodes(unsigned long *nodemask,
+                                           unsigned long maxnode, const struct vec_cpu_node *closest_numanode,
+                                           int num_cpu);
+int set_bitmask_for_current_closest_numanode(unsigned long *nodemask,
+                                             unsigned long maxnode, const struct vec_cpu_node *closest_numanode,
+                                             int num_cpu);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_dax_kmem.h b/include/memkind/internal/memkind_dax_kmem.h
new file mode 100644
index 0000000..2bbc193
--- /dev/null
+++ b/include/memkind/internal/memkind_dax_kmem.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the DAX KMEM memory memkind operations.
+ * More details in memkind_dax_kmem(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_dax_kmem_all_get_mbind_nodemask(struct memkind *kind,
+                                            unsigned long *nodemask, unsigned long maxnode);
+
+extern struct memkind_ops MEMKIND_DAX_KMEM_OPS;
+extern struct memkind_ops MEMKIND_DAX_KMEM_ALL_OPS;
+extern struct memkind_ops MEMKIND_DAX_KMEM_PREFERRED_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_default.h b/include/memkind/internal/memkind_default.h
new file mode 100644
index 0000000..51e24d5
--- /dev/null
+++ b/include/memkind/internal/memkind_default.h
@@ -0,0 +1,71 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+#include <memkind/internal/memkind_private.h>
+
+/*
+ * Header file for the default implementations for memkind operations.
+ * More details in memkind_default(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_default_create(struct memkind *kind, struct memkind_ops *ops,
+                           const char *name);
+int memkind_default_destroy(struct memkind *kind);
+void *memkind_default_malloc(struct memkind *kind, size_t size);
+void *memkind_default_calloc(struct memkind *kind, size_t num, size_t size);
+int memkind_default_posix_memalign(struct memkind *kind, void **memptr,
+                                   size_t alignment, size_t size);
+void *memkind_default_realloc(struct memkind *kind, void *ptr, size_t size);
+void memkind_default_free(struct memkind *kind, void *ptr);
+void *memkind_default_mmap(struct memkind *kind, void *addr, size_t size);
+int memkind_default_mbind(struct memkind *kind, void *ptr, size_t size);
+int memkind_default_get_mmap_flags(struct memkind *kind, int *flags);
+int memkind_default_get_mbind_mode(struct memkind *kind, int *mode);
+int memkind_default_get_mbind_nodemask(struct memkind *kind,
+                                       unsigned long *nodemask, unsigned long maxnode);
+int memkind_preferred_get_mbind_mode(struct memkind *kind, int *mode);
+int memkind_interleave_get_mbind_mode(struct memkind *kind, int *mode);
+int memkind_nohugepage_madvise(struct memkind *kind, void *addr, size_t size);
+int memkind_posix_check_alignment(struct memkind *kind, size_t alignment);
+void memkind_default_init_once(void);
+size_t memkind_default_malloc_usable_size(struct memkind *kind, void *ptr);
+static inline bool size_out_of_bounds(size_t size)
+{
+    return !size;
+}
+
+extern struct memkind_ops MEMKIND_DEFAULT_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_gbtlb.h b/include/memkind/internal/memkind_gbtlb.h
new file mode 100755
index 0000000..97f0796
--- /dev/null
+++ b/include/memkind/internal/memkind_gbtlb.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the gigabyte TLB memkind operations.
+ *
+ * All function declarations has been moved to memkind_deprecated.h
+ * because of end of GBTLB support.
+ *
+ * API standards are described in memkind(3) man page.
+ */
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_hbw.h b/include/memkind/internal/memkind_hbw.h
new file mode 100755
index 0000000..7eb039d
--- /dev/null
+++ b/include/memkind/internal/memkind_hbw.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the high bandwidth memory memkind operations.
+ * More details in memkind_hbw(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_hbw_check_available(struct memkind *kind);
+int memkind_hbw_hugetlb_check_available(struct memkind *kind);
+int memkind_hbw_get_mbind_nodemask(struct memkind *kind,
+                                   unsigned long *nodemask,
+                                   unsigned long maxnode);
+int memkind_hbw_all_get_mbind_nodemask(struct memkind *kind,
+                                       unsigned long *nodemask,
+                                       unsigned long maxnode);
+void memkind_hbw_init_once(void);
+void memkind_hbw_all_init_once(void);
+void memkind_hbw_hugetlb_init_once(void);
+void memkind_hbw_all_hugetlb_init_once(void);
+void memkind_hbw_preferred_init_once(void);
+void memkind_hbw_preferred_hugetlb_init_once(void);
+void memkind_hbw_interleave_init_once(void);
+
+extern struct memkind_ops MEMKIND_HBW_OPS;
+extern struct memkind_ops MEMKIND_HBW_ALL_OPS;
+extern struct memkind_ops MEMKIND_HBW_HUGETLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_ALL_HUGETLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_PREFERRED_OPS;
+extern struct memkind_ops MEMKIND_HBW_PREFERRED_HUGETLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_INTERLEAVE_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_hugetlb.h b/include/memkind/internal/memkind_hugetlb.h
new file mode 100755
index 0000000..d76915f
--- /dev/null
+++ b/include/memkind/internal/memkind_hugetlb.h
@@ -0,0 +1,48 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the hugetlb memory memkind operations.
+ * More details in memkind_hugetlb(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_hugetlb_get_mmap_flags(struct memkind *kind, int *flags);
+void memkind_hugetlb_init_once(void);
+int memkind_hugetlb_check_available_2mb(struct memkind *kind);
+
+extern struct memkind_ops MEMKIND_HUGETLB_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_interleave.h b/include/memkind/internal/memkind_interleave.h
new file mode 100755
index 0000000..1913d6a
--- /dev/null
+++ b/include/memkind/internal/memkind_interleave.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+ * Header file for the interleave memory memkind operations.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+void memkind_interleave_init_once(void);
+
+extern struct memkind_ops MEMKIND_INTERLEAVE_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_log.h b/include/memkind/internal/memkind_log.h
new file mode 100644
index 0000000..7e9b7e0
--- /dev/null
+++ b/include/memkind/internal/memkind_log.h
@@ -0,0 +1,52 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define PRINTF_FORMAT __attribute__ ((format (printf, 1, 2)))
+
+/*
+ * For printing informational messages
+ * Requires environment variable MEMKIND_DEBUG to be set to appropriate value
+ */
+void log_info(const char *format, ...) PRINTF_FORMAT;
+
+/*
+ * For printing messages regarding errors and failures
+ * Requires environment variable MEMKIND_DEBUG to be set to appropriate value
+ */
+void log_err(const char *format, ...) PRINTF_FORMAT;
+
+/*
+ * For printing messages regarding fatal errors before calling abort()
+ * Works *no matter* of MEMKIND_DEBUG state
+ */
+void log_fatal(const char *format, ...)PRINTF_FORMAT;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_pmem.h b/include/memkind/internal/memkind_pmem.h
new file mode 100644
index 0000000..b6098ba
--- /dev/null
+++ b/include/memkind/internal/memkind_pmem.h
@@ -0,0 +1,64 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+#include "memkind_default.h"
+#include "memkind_arena.h"
+
+#include <pthread.h>
+
+/*
+ * Header file for the file-backed memory memkind operations.
+ * More details in memkind_pmem(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+#define MEMKIND_PMEM_CHUNK_SIZE (1ull << 21ull) // 2MB
+
+int memkind_pmem_create(struct memkind *kind, struct memkind_ops *ops,
+                        const char *name);
+int memkind_pmem_destroy(struct memkind *kind);
+void *memkind_pmem_mmap(struct memkind *kind, void *addr, size_t size);
+int memkind_pmem_get_mmap_flags(struct memkind *kind, int *flags);
+
+struct memkind_pmem {
+    int fd;
+    off_t offset;
+    size_t max_size;
+    pthread_mutex_t pmem_lock;
+    size_t current_size;
+};
+
+extern struct memkind_ops MEMKIND_PMEM_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/memkind_private.h b/include/memkind/internal/memkind_private.h
new file mode 100644
index 0000000..af01ba5
--- /dev/null
+++ b/include/memkind/internal/memkind_private.h
@@ -0,0 +1,125 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "memkind.h"
+
+#include <stdbool.h>
+#include <pthread.h>
+
+#ifdef __GNUC__
+#   define MEMKIND_LIKELY(x)       __builtin_expect(!!(x), 1)
+#   define MEMKIND_UNLIKELY(x)     __builtin_expect(!!(x), 0)
+#else
+#   define MEMKIND_LIKELY(x)       (x)
+#   define MEMKIND_UNLIKELY(x)     (x)
+#endif
+
+#ifndef MEMKIND_EXPORT
+#   define MEMKIND_EXPORT __attribute__((visibility("default")))
+#endif
+
+// This ladder call is required due to meanders of C's preprocessor logic.
+// Without it, JE_PREFIX would be used directly (i.e. 'JE_PREFIX') and not
+// substituted with defined value.
+#define JE_SYMBOL2(a, b) a ## b
+#define JE_SYMBOL1(a, b) JE_SYMBOL2(a, b)
+#define JE_SYMBOL(b)     JE_SYMBOL1(JE_PREFIX, b)
+
+// Redefine symbols
+#define jemk_malloc                 JE_SYMBOL(malloc)
+#define jemk_mallocx                JE_SYMBOL(mallocx)
+#define jemk_calloc                 JE_SYMBOL(calloc)
+#define jemk_rallocx                JE_SYMBOL(rallocx)
+#define jemk_realloc                JE_SYMBOL(realloc)
+#define jemk_mallctl                JE_SYMBOL(mallctl)
+#define jemk_memalign               JE_SYMBOL(memalign)
+#define jemk_posix_memalign         JE_SYMBOL(posix_memalign)
+#define jemk_free                   JE_SYMBOL(free)
+#define jemk_dallocx                JE_SYMBOL(dallocx)
+#define jemk_malloc_usable_size     JE_SYMBOL(malloc_usable_size)
+
+enum memkind_const_private {
+    MEMKIND_NAME_LENGTH_PRIV = 64
+};
+
+struct memkind_ops {
+    int (* create)(struct memkind *kind, struct memkind_ops *ops, const char *name);
+    int (* destroy)(struct memkind *kind);
+    void *(* malloc)(struct memkind *kind, size_t size);
+    void *(* calloc)(struct memkind *kind, size_t num, size_t size);
+    int (* posix_memalign)(struct memkind *kind, void **memptr, size_t alignment,
+                           size_t size);
+    void *(* realloc)(struct memkind *kind, void *ptr, size_t size);
+    void (* free)(struct memkind *kind, void *ptr);
+    void *(* mmap)(struct memkind *kind, void *addr, size_t size);
+    int (* mbind)(struct memkind *kind, void *ptr, size_t size);
+    int (* madvise)(struct memkind *kind, void *addr, size_t size);
+    int (* get_mmap_flags)(struct memkind *kind, int *flags);
+    int (* get_mbind_mode)(struct memkind *kind, int *mode);
+    int (* get_mbind_nodemask)(struct memkind *kind, unsigned long *nodemask,
+                               unsigned long maxnode);
+    int (* get_arena)(struct memkind *kind, unsigned int *arena, size_t size);
+    int (* check_available)(struct memkind *kind);
+    void (* init_once)(void);
+    int (* finalize)(struct memkind *kind);
+    size_t (* malloc_usable_size)(struct memkind *kind, void *ptr);
+    int (* update_memory_usage_policy)(struct memkind *kind,
+                                       memkind_mem_usage_policy policy);
+    int (* get_stat)(memkind_t kind, memkind_stat_type stat, size_t *value);
+    void *(* defrag_reallocate)(struct memkind *kind, void *ptr);
+};
+
+struct memkind {
+    struct memkind_ops *ops;
+    unsigned int partition;
+    char name[MEMKIND_NAME_LENGTH_PRIV];
+    pthread_once_t init_once;
+    unsigned int arena_map_len; // is power of 2
+    unsigned int *arena_map; // To be deleted beyond 1.2.0+
+    pthread_key_t arena_key;
+    void *priv;
+    unsigned int
+    arena_map_mask; // arena_map_len - 1 to optimize modulo operation on arena_map_len
+    unsigned int arena_zero; // index first jemalloc arena of this kind
+};
+
+struct memkind_config {
+    const char *pmem_dir;            //PMEM kind path
+    size_t pmem_size;                //PMEM kind size
+    memkind_mem_usage_policy policy; //kind memory usage policy
+};
+
+void memkind_init(memkind_t kind, bool check_numa);
+
+void *kind_mmap(struct memkind *kind, void *addr, size_t size);
+
+#ifdef __cplusplus
+}
+#endif
+
diff --git a/include/memkind/internal/memkind_regular.h b/include/memkind/internal/memkind_regular.h
new file mode 100755
index 0000000..067ea04
--- /dev/null
+++ b/include/memkind/internal/memkind_regular.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the regular memory memkind operations.
+ * More details in memkind_regular(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+extern struct memkind_ops MEMKIND_REGULAR_OPS;
+int memkind_regular_all_get_mbind_nodemask(struct memkind *kind,
+                                           unsigned long *nodemask,
+                                           unsigned long maxnode);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/tbb_mem_pool_policy.h b/include/memkind/internal/tbb_mem_pool_policy.h
new file mode 100644
index 0000000..085cb29
--- /dev/null
+++ b/include/memkind/internal/tbb_mem_pool_policy.h
@@ -0,0 +1,34 @@
+/*
+    Copyright (c) 2005-2018 Intel Corporation
+
+    Licensed under the Apache License, Version 2.0 (the "License");
+    you may not use this file except in compliance with the License.
+    You may obtain a copy of the License at
+
+        http://www.apache.org/licenses/LICENSE-2.0
+
+    Unless required by applicable law or agreed to in writing, software
+    distributed under the License is distributed on an "AS IS" BASIS,
+    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+    See the License for the specific language governing permissions and
+    limitations under the License.
+
+
+
+
+*/
+
+#include <stdint.h>
+
+typedef void *(*rawAllocType)(intptr_t pool_id, size_t *bytes);
+typedef int   (*rawFreeType)(intptr_t pool_id, void *raw_ptr, size_t raw_bytes);
+
+struct MemPoolPolicy {
+    rawAllocType pAlloc;
+    rawFreeType  pFree;
+    size_t       granularity;
+    int          version;
+    unsigned     fixedPool : 1,
+                 keepAllMemory : 1,
+                 reserved : 30;
+};
diff --git a/include/memkind/internal/tbb_wrapper.h b/include/memkind/internal/tbb_wrapper.h
new file mode 100644
index 0000000..11615d8
--- /dev/null
+++ b/include/memkind/internal/tbb_wrapper.h
@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include <memkind.h>
+#include <memkind_deprecated.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* dynamically load TBB symbols */
+void load_tbb_symbols(void);
+
+/* ops callbacks are replaced by TBB callbacks. */
+void tbb_initialize(struct memkind *kind);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+void tbb_pool_free_with_kind_detect(void *ptr);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+void *tbb_pool_realloc_with_kind_detect(void *ptr, size_t size);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+size_t tbb_pool_malloc_usable_size_with_kind_detect(void *ptr);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+struct memkind *tbb_detect_kind(void *ptr);
+
+/* update cached stats for TBB (unsupported) */
+int tbb_update_cached_stats(void);
+
+/* get allocator stat for TBB (unsupported) */
+int tbb_get_global_stat(memkind_stat_type stat, size_t *value);
+
+/* defrag reallocate for TBB (unsupported) */
+void *tbb_pool_defrag_reallocate_with_kind_detect(void *ptr);
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind/internal/vec.h b/include/memkind/internal/vec.h
new file mode 100644
index 0000000..982618f
--- /dev/null
+++ b/include/memkind/internal/vec.h
@@ -0,0 +1,102 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stddef.h>
+#include <stdlib.h>
+
+#define VEC_INIT_SIZE (8)
+
+#define VEC(name, type)\
+struct name {\
+    type *buffer;\
+    size_t size;\
+    size_t capacity;\
+}
+
+static inline int
+vec_reserve(void *vec, size_t ncapacity, size_t s)
+{
+    size_t ncap = ncapacity == 0 ? VEC_INIT_SIZE : ncapacity;
+    VEC(vvec, void) *vecp = (struct vvec *)vec;
+    void *tbuf = realloc(vecp->buffer, s * ncap);
+    if (tbuf == NULL) {
+        return -1;
+    }
+    vecp->buffer = tbuf;
+    vecp->capacity = ncap;
+    return 0;
+}
+
+#define VEC_RESERVE(vec, ncapacity)\
+(((vec)->size == 0 || (ncapacity) > (vec)->size) ?\
+    vec_reserve((void *)vec, ncapacity, sizeof(*(vec)->buffer)) :\
+    0)
+
+#define VEC_INSERT(vec, element)\
+((vec)->buffer[(vec)->size - 1] = (element), 0)
+
+#define VEC_INC_SIZE(vec)\
+(((vec)->size++), 0)
+
+#define VEC_INC_BACK(vec)\
+((vec)->capacity == (vec)->size ?\
+    (VEC_RESERVE((vec), ((vec)->capacity * 2)) == 0 ?\
+        VEC_INC_SIZE(vec) : -1) :\
+    VEC_INC_SIZE(vec))
+
+#define VEC_PUSH_BACK(vec, element)\
+(VEC_INC_BACK(vec) == 0? VEC_INSERT(vec, element) : -1)
+
+#define VEC_FOREACH(el, vec)\
+size_t _vec_i; \
+for (_vec_i = 0;\
+    _vec_i < (vec)->size && (((el) = (vec)->buffer[_vec_i]), 1);\
+    ++_vec_i)
+
+#define VEC_SIZE(vec)\
+((vec)->size)
+
+#define VEC_GET(vec, id)\
+(&(vec)->buffer[id])
+
+#define VEC_CLEAR(vec) do {\
+    (vec)->size = 0;\
+} while (0)
+
+#define VEC_DELETE(vec) do {\
+    free((vec)->buffer);\
+    (vec)->buffer = NULL;\
+    (vec)->size = 0;\
+    (vec)->capacity = 0;\
+} while (0)
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/memkind_allocator.h b/include/memkind_allocator.h
new file mode 100644
index 0000000..45bde75
--- /dev/null
+++ b/include/memkind_allocator.h
@@ -0,0 +1,231 @@
+/*
+ * Copyright (C) 2019 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <cmath>
+#include <map>
+#include <memory>
+#include <string>
+#include <exception>
+#include <type_traits>
+#include <cstddef>
+#include <stdexcept>
+
+#include "memkind.h"
+
+/*
+ * Header file for the C++ allocator compatible with the C++ standard library allocator concepts.
+ * More details in memkind(3) man page.
+ * Note: memory heap management is based on memkind_malloc, refer to the memkind(3) man page for more
+ * information.
+ *
+ * Functionality defined in this header is considered as stable API (STANDARD API).
+ * API standards are described in memkind(3) man page.
+ */
+namespace libmemkind
+{
+    enum class kinds {
+        DEFAULT = 0,
+        HUGETLB = 1,
+        INTERLEAVE = 2,
+        HBW = 3,
+        HBW_ALL = 4,
+        HBW_HUGETLB = 5,
+        HBW_ALL_HUGETLB = 6,
+        HBW_PREFERRED = 7,
+        HBW_PREFERRED_HUGETLB = 8,
+        HBW_INTERLEAVE = 9,
+        REGULAR = 10,
+        DAX_KMEM = 11,
+        DAX_KMEM_ALL = 12,
+        DAX_KMEM_PREFERRED = 13,
+    };
+
+    namespace static_kind
+    {
+        template<typename T>
+        class allocator
+        {
+        public:
+            using value_type = T;
+            using pointer = value_type*;
+            using const_pointer = const value_type*;
+            using reference = value_type&;
+            using const_reference = const value_type&;
+            using size_type = size_t;
+            using difference_type = ptrdiff_t;
+
+            template<class U>
+            struct rebind {
+                using other = allocator<U>;
+            };
+
+            template<typename U>
+            friend class allocator;
+
+#if !_GLIBCXX_USE_CXX11_ABI
+            /* This is a workaround for compilers (e.g GCC 4.8) that uses C++11 standard,
+             * but use old - non C++11 ABI */
+            template<typename V = void>
+            explicit allocator()
+            {
+                static_assert(std::is_same<V, void>::value,
+                              "libmemkind::static_kind::allocator cannot be compiled without CXX11 ABI");
+            }
+#endif
+
+            explicit allocator(libmemkind::kinds kind)
+            {
+                switch (kind) {
+                    case libmemkind::kinds::DEFAULT:
+                        _kind = MEMKIND_DEFAULT;
+                        break;
+                    case libmemkind::kinds::HUGETLB:
+                        _kind = MEMKIND_HUGETLB;
+                        break;
+                    case libmemkind::kinds::INTERLEAVE:
+                        _kind = MEMKIND_INTERLEAVE;
+                        break;
+                    case libmemkind::kinds::HBW:
+                        _kind = MEMKIND_HBW;
+                        break;
+                    case libmemkind::kinds::HBW_ALL:
+                        _kind = MEMKIND_HBW_ALL;
+                        break;
+                    case libmemkind::kinds::HBW_HUGETLB:
+                        _kind = MEMKIND_HBW_HUGETLB;
+                        break;
+                    case libmemkind::kinds::HBW_ALL_HUGETLB:
+                        _kind = MEMKIND_HBW_ALL_HUGETLB;
+                        break;
+                    case libmemkind::kinds::HBW_PREFERRED:
+                        _kind = MEMKIND_HBW_PREFERRED;
+                        break;
+                    case libmemkind::kinds::HBW_PREFERRED_HUGETLB:
+                        _kind = MEMKIND_HBW_PREFERRED_HUGETLB;
+                        break;
+                    case libmemkind::kinds::HBW_INTERLEAVE:
+                        _kind = MEMKIND_HBW_INTERLEAVE;
+                        break;
+                    case libmemkind::kinds::REGULAR:
+                        _kind = MEMKIND_REGULAR;
+                        break;
+                    case libmemkind::kinds::DAX_KMEM:
+                        _kind = MEMKIND_DAX_KMEM;
+                        break;
+                    case libmemkind::kinds::DAX_KMEM_ALL:
+                        _kind = MEMKIND_DAX_KMEM_ALL;
+                        break;
+                    case libmemkind::kinds::DAX_KMEM_PREFERRED:
+                        _kind = MEMKIND_DAX_KMEM_PREFERRED;
+                        break;
+                    default:
+                        throw std::runtime_error("Unknown libmemkind::kinds");
+                        break;
+                }
+            }
+
+            allocator(const allocator &other) = default;
+
+            template <typename U>
+            allocator(const allocator<U> &other) noexcept
+            {
+                _kind = other._kind;
+            }
+
+            allocator(allocator &&other) = default;
+
+            template <typename U>
+            allocator(const allocator<U> &&other) noexcept
+            {
+                _kind = std::move(other._kind);
+            }
+
+            allocator<T> &operator = (const allocator &other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (const allocator<U> &other) noexcept
+            {
+                _kind = other._kind;
+                return *this;
+            }
+
+            allocator<T> &operator = (allocator &&other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (allocator<U> &&other) noexcept
+            {
+                _kind = std::move(other._kind);
+                return *this;
+            }
+
+            pointer allocate(size_type n) const
+            {
+                pointer result = static_cast<pointer>(memkind_malloc(_kind, n*sizeof(T)));
+                if (!result) {
+                    throw std::bad_alloc();
+                }
+                return result;
+            }
+
+            void deallocate(pointer p, size_type n) const
+            {
+                memkind_free(_kind, static_cast<void *>(p));
+            }
+
+            template <class U, class... Args>
+            void construct(U *p, Args &&... args) const
+            {
+                ::new((void *)p) U(std::forward<Args>(args)...);
+            }
+
+            void destroy(pointer p) const
+            {
+                p->~value_type();
+            }
+
+            template <typename U, typename V>
+            friend bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs);
+
+            template <typename U, typename V>
+            friend bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs);
+
+        private:
+            memkind_t _kind;
+        };
+
+        template <typename U, typename V>
+        bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return lhs._kind == rhs._kind;
+        }
+
+        template <typename U, typename V>
+        bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return !(lhs._kind == rhs._kind);
+        }
+    }  // namespace static_kind
+}  // namespace libmemkind
diff --git a/include/memkind_deprecated.h b/include/memkind_deprecated.h
new file mode 100644
index 0000000..3e2c3c5
--- /dev/null
+++ b/include/memkind_deprecated.h
@@ -0,0 +1,88 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/*
+ * !!!!!!!!!!!!!!!!!!!
+ * !!!   WARNING   !!!
+ * !!! PLEASE READ !!!
+ * !!!!!!!!!!!!!!!!!!!
+ *
+ * This header file contains all memkind deprecated symbols.
+ *
+ * Please avoid usage of this API in newly developed code, as
+ * eventually this code is subject for removal.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef MEMKIND_DEPRECATED
+
+#ifdef __GNUC__
+#define MEMKIND_DEPRECATED(func) func __attribute__ ((deprecated))
+#elif defined(_MSC_VER)
+#define MEMKIND_DEPRECATED(func) __declspec(deprecated) func
+#else
+#pragma message("WARNING: You need to implement MEMKIND_DEPRECATED for this compiler")
+#define MEMKIND_DEPRECATED(func) func
+#endif
+
+#endif
+
+/*
+ * Symbols related to GBTLB that are no longer supported
+ */
+extern memkind_t MEMKIND_HBW_GBTLB;
+extern memkind_t MEMKIND_HBW_PREFERRED_GBTLB;
+extern memkind_t MEMKIND_GBTLB;
+
+int MEMKIND_DEPRECATED(memkind_get_kind_by_partition(int partition,
+                                                     memkind_t *kind));
+
+enum memkind_base_partition {
+    MEMKIND_PARTITION_DEFAULT = 0,
+    MEMKIND_PARTITION_HBW = 1,
+    MEMKIND_PARTITION_HBW_HUGETLB = 2,
+    MEMKIND_PARTITION_HBW_PREFERRED = 3,
+    MEMKIND_PARTITION_HBW_PREFERRED_HUGETLB = 4,
+    MEMKIND_PARTITION_HUGETLB = 5,
+    MEMKIND_PARTITION_HBW_GBTLB = 6,
+    MEMKIND_PARTITION_HBW_PREFERRED_GBTLB = 7,
+    MEMKIND_PARTITION_GBTLB = 8,
+    MEMKIND_PARTITION_HBW_INTERLEAVE = 9,
+    MEMKIND_PARTITION_INTERLEAVE = 10,
+    MEMKIND_PARTITION_REGULAR = 11,
+    MEMKIND_PARTITION_HBW_ALL = 12,
+    MEMKIND_PARTITION_HBW_ALL_HUGETLB = 13,
+    MEMKIND_PARTITION_DAX_KMEM = 14,
+    MEMKIND_PARTITION_DAX_KMEM_ALL = 15,
+    MEMKIND_PARTITION_DAX_KMEM_PREFERRED = 16,
+    MEMKIND_NUM_BASE_KIND
+};
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/include/pmem_allocator.h b/include/pmem_allocator.h
new file mode 100644
index 0000000..d373e3c
--- /dev/null
+++ b/include/pmem_allocator.h
@@ -0,0 +1,236 @@
+/*
+ * Copyright (C) 2018 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <memory>
+#include <string>
+#include <exception>
+#include <type_traits>
+#include <atomic>
+#include <cstddef>
+#include <stdexcept>
+
+#include "memkind.h"
+
+/*
+ * Header file for the C++ allocator compatible with the C++ standard library allocator concepts.
+ * More details in pmemallocator(3) man page.
+ * Note: memory heap management is based on memkind_malloc, refer to the memkind(3) man page for more
+ * information.
+ *
+ * Functionality defined in this header is considered as stable API (STANDARD API).
+ * API standards are described in memkind(3) man page.
+ */
+namespace libmemkind
+{
+    enum class allocation_policy {
+        DEFAULT = MEMKIND_MEM_USAGE_POLICY_DEFAULT,
+        CONSERVATIVE = MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE,
+        MAX
+    };
+
+    namespace pmem
+    {
+        namespace internal
+        {
+            class kind_wrapper_t
+            {
+            public:
+                kind_wrapper_t(const char *dir, std::size_t max_size,
+                               libmemkind::allocation_policy alloc_policy=
+                                   libmemkind::allocation_policy::DEFAULT)
+                {
+                    cfg = memkind_config_new();
+                    if (!cfg) {
+                        throw std::runtime_error(
+                            std::string("An error occurred while creating pmem config"));
+                    }
+                    memkind_config_set_path(cfg, dir);
+                    memkind_config_set_size(cfg, max_size);
+                    memkind_config_set_memory_usage_policy(cfg,
+                                                           static_cast<memkind_mem_usage_policy>(alloc_policy));
+                    int err_c  = memkind_create_pmem_with_config(cfg, &kind);
+                    memkind_config_delete(cfg);
+                    if (err_c) {
+                        throw std::runtime_error(
+                            std::string("An error occurred while creating pmem kind; error code: ") +
+                            std::to_string(err_c));
+                    }
+                }
+
+                kind_wrapper_t(const kind_wrapper_t &) = delete;
+                void operator=(const kind_wrapper_t &) = delete;
+
+                ~kind_wrapper_t()
+                {
+                    memkind_destroy_kind(kind);
+                }
+
+                memkind_t get() const
+                {
+                    return kind;
+                }
+
+            private:
+                memkind_t kind;
+                struct memkind_config *cfg;
+            };
+        }
+
+        template<typename T>
+        class allocator
+        {
+            using kind_wrapper_t = internal::kind_wrapper_t;
+            std::shared_ptr<kind_wrapper_t> kind_wrapper_ptr;
+        public:
+            using value_type = T;
+            using pointer = value_type*;
+            using const_pointer = const value_type*;
+            using reference = value_type&;
+            using const_reference = const value_type&;
+            using size_type = size_t;
+            using difference_type = ptrdiff_t;
+
+            template<class U>
+            struct rebind {
+                using other = allocator<U>;
+            };
+
+            template<typename U>
+            friend class allocator;
+
+#if !_GLIBCXX_USE_CXX11_ABI
+            /* This is a workaround for compilers (e.g GCC 4.8) that uses C++11 standard,
+             * but use old - non C++11 ABI */
+            template<typename V = void>
+            explicit allocator()
+            {
+                static_assert(std::is_same<V, void>::value,
+                              "libmemkind::pmem::allocator cannot be compiled without CXX11 ABI");
+            }
+#endif
+
+            explicit allocator(const char *dir, size_t max_size) :
+                kind_wrapper_ptr(std::make_shared<kind_wrapper_t>(dir, max_size))
+            {
+            }
+
+            explicit allocator(const std::string &dir, size_t max_size) :
+                allocator(dir.c_str(), max_size)
+            {
+            }
+
+            explicit allocator(const char *dir, size_t max_size,
+                               libmemkind::allocation_policy alloc_policy) :
+                kind_wrapper_ptr(std::make_shared<kind_wrapper_t>(dir, max_size, alloc_policy))
+            {
+            }
+
+            explicit allocator(const std::string &dir, size_t max_size,
+                               libmemkind::allocation_policy alloc_policy) :
+                allocator(dir.c_str(), max_size, alloc_policy)
+            {
+            }
+
+            allocator(const allocator &other) = default;
+
+            template <typename U>
+            allocator(const allocator<U> &other) noexcept : kind_wrapper_ptr(
+                    other.kind_wrapper_ptr)
+            {
+            }
+
+            allocator(allocator &&other) = default;
+
+            template <typename U>
+            allocator(allocator<U> &&other) noexcept :
+                kind_wrapper_ptr(std::move(other.kind_wrapper_ptr))
+            {
+            }
+
+            allocator<T> &operator = (const allocator &other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (const allocator<U> &other) noexcept
+            {
+                kind_wrapper_ptr = other.kind_wrapper_ptr;
+                return *this;
+            }
+
+            allocator<T> &operator = (allocator &&other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (allocator<U> &&other) noexcept
+            {
+                kind_wrapper_ptr = std::move(other.kind_wrapper_ptr);
+                return *this;
+            }
+
+            pointer allocate(size_type n) const
+            {
+                pointer result = static_cast<pointer>(memkind_malloc(kind_wrapper_ptr->get(),
+                                                                     n*sizeof(T)));
+                if (!result) {
+                    throw std::bad_alloc();
+                }
+                return result;
+            }
+
+            void deallocate(pointer p, size_type n) const
+            {
+                memkind_free(kind_wrapper_ptr->get(), static_cast<void *>(p));
+            }
+
+            template <class U, class... Args>
+            void construct(U *p, Args &&... args) const
+            {
+                ::new((void *)p) U(std::forward<Args>(args)...);
+            }
+
+            void destroy(pointer p) const
+            {
+                p->~value_type();
+            }
+
+            template <typename U, typename V>
+            friend bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs);
+
+            template <typename U, typename V>
+            friend bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs);
+        };
+
+        template <typename U, typename V>
+        bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return lhs.kind_wrapper_ptr->get() == rhs.kind_wrapper_ptr->get();
+        }
+
+        template <typename U, typename V>
+        bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return !(lhs == rhs);
+        }
+    }  // namespace pmem
+}  // namespace libmemkind
diff --git a/install_astyle.sh b/install_astyle.sh
new file mode 100755
index 0000000..58615fc
--- /dev/null
+++ b/install_astyle.sh
@@ -0,0 +1,31 @@
+#!/bin/bash
+#
+#  Copyright (C) 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+set -ex
+curl -SL https://sourceforge.net/projects/astyle/files/astyle/astyle%203.1/astyle_3.1_linux.tar.gz -o /tmp/astyle.tar.gz
+tar -xzvf /tmp/astyle.tar.gz -C /tmp/
+echo "Install astyle"
+cd /tmp/astyle/build/gcc
+make
+sudo make install
diff --git a/jemalloc/.appveyor.yml b/jemalloc/.appveyor.yml
new file mode 100644
index 0000000..90b0368
--- /dev/null
+++ b/jemalloc/.appveyor.yml
@@ -0,0 +1,42 @@
+version: '{build}'
+
+environment:
+  matrix:
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    MSVC: amd64
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW32
+    CPU: i686
+    MSVC: x86
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW32
+    CPU: i686
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    MSVC: amd64
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+  - MSYSTEM: MINGW32
+    CPU: i686
+    MSVC: x86
+  - MSYSTEM: MINGW32
+    CPU: i686
+
+install:
+  - set PATH=c:\msys64\%MSYSTEM%\bin;c:\msys64\usr\bin;%PATH%
+  - if defined MSVC call "c:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\vcvarsall.bat" %MSVC%
+  - if defined MSVC pacman --noconfirm -Rsc mingw-w64-%CPU%-gcc gcc
+  - pacman --noconfirm -Suy mingw-w64-%CPU%-make
+
+build_script:
+  - bash -c "autoconf"
+  - bash -c "./configure $CONFIG_FLAGS"
+  - mingw32-make
+  - file lib/jemalloc.dll
+  - mingw32-make tests
+  - mingw32-make -k check
diff --git a/jemalloc/.autom4te.cfg b/jemalloc/.autom4te.cfg
new file mode 100644
index 0000000..fe2424d
--- /dev/null
+++ b/jemalloc/.autom4te.cfg
@@ -0,0 +1,3 @@
+begin-language: "Autoconf-without-aclocal-m4"
+args: --no-cache
+end-language: "Autoconf-without-aclocal-m4"
diff --git a/jemalloc/.gitattributes b/jemalloc/.gitattributes
new file mode 100644
index 0000000..6313b56
--- /dev/null
+++ b/jemalloc/.gitattributes
@@ -0,0 +1 @@
+* text=auto eol=lf
diff --git a/jemalloc/.gitignore b/jemalloc/.gitignore
new file mode 100644
index 0000000..5ca0ad1
--- /dev/null
+++ b/jemalloc/.gitignore
@@ -0,0 +1,92 @@
+/bin/jemalloc-config
+/bin/jemalloc.sh
+/bin/jeprof
+
+/config.stamp
+/config.log
+/config.status
+/configure
+
+/doc/html.xsl
+/doc/manpages.xsl
+/doc/jemalloc.xml
+/doc/jemalloc.html
+/doc/jemalloc.3
+
+/jemalloc.pc
+
+/lib/
+
+/Makefile
+
+/include/jemalloc/internal/jemalloc_preamble.h
+/include/jemalloc/internal/jemalloc_internal_defs.h
+/include/jemalloc/internal/private_namespace.gen.h
+/include/jemalloc/internal/private_namespace.h
+/include/jemalloc/internal/private_namespace_jet.gen.h
+/include/jemalloc/internal/private_namespace_jet.h
+/include/jemalloc/internal/private_symbols.awk
+/include/jemalloc/internal/private_symbols_jet.awk
+/include/jemalloc/internal/public_namespace.h
+/include/jemalloc/internal/public_symbols.txt
+/include/jemalloc/internal/public_unnamespace.h
+/include/jemalloc/jemalloc.h
+/include/jemalloc/jemalloc_defs.h
+/include/jemalloc/jemalloc_macros.h
+/include/jemalloc/jemalloc_mangle.h
+/include/jemalloc/jemalloc_mangle_jet.h
+/include/jemalloc/jemalloc_protos.h
+/include/jemalloc/jemalloc_protos_jet.h
+/include/jemalloc/jemalloc_rename.h
+/include/jemalloc/jemalloc_typedefs.h
+
+/src/*.[od]
+/src/*.sym
+
+/run_tests.out/
+
+/test/test.sh
+test/include/test/jemalloc_test.h
+test/include/test/jemalloc_test_defs.h
+
+/test/integration/[A-Za-z]*
+!/test/integration/[A-Za-z]*.*
+/test/integration/*.[od]
+/test/integration/*.out
+
+/test/integration/cpp/[A-Za-z]*
+!/test/integration/cpp/[A-Za-z]*.*
+/test/integration/cpp/*.[od]
+/test/integration/cpp/*.out
+
+/test/src/*.[od]
+
+/test/stress/[A-Za-z]*
+!/test/stress/[A-Za-z]*.*
+/test/stress/*.[od]
+/test/stress/*.out
+
+/test/unit/[A-Za-z]*
+!/test/unit/[A-Za-z]*.*
+/test/unit/*.[od]
+/test/unit/*.out
+
+/VERSION
+
+*.pdb
+*.sdf
+*.opendb
+*.VC.db
+*.opensdf
+*.cachefile
+*.suo
+*.user
+*.sln.docstates
+*.tmp
+.vs/
+/msvc/Win32/
+/msvc/x64/
+/msvc/projects/*/*/Debug*/
+/msvc/projects/*/*/Release*/
+/msvc/projects/*/*/Win32/
+/msvc/projects/*/*/x64/
diff --git a/jemalloc/.travis.yml b/jemalloc/.travis.yml
new file mode 100644
index 0000000..2da5da8
--- /dev/null
+++ b/jemalloc/.travis.yml
@@ -0,0 +1,195 @@
+language: generic
+dist: precise
+
+matrix:
+  include:
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: &gcc_multilib
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary,percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    # Development build
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-cache-oblivious --enable-stats --enable-log --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    # --enable-expermental-smallocx:
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-experimental-smallocx --enable-stats --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+
+    # Valgrind
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds" JEMALLOC_TEST_PREFIX="valgrind"
+      addons:
+        apt:
+          packages:
+            - valgrind
+
+
+before_script:
+  - autoconf
+  - scripts/gen_travis.py > travis_script && diff .travis.yml travis_script
+  - ./configure ${COMPILER_FLAGS:+       CC="$CC $COMPILER_FLAGS"       CXX="$CXX $COMPILER_FLAGS" }       $CONFIGURE_FLAGS
+  - make -j3
+  - make -j3 tests
+
+script:
+  - make check
+
diff --git a/jemalloc/COPYING b/jemalloc/COPYING
new file mode 100644
index 0000000..3b7fd35
--- /dev/null
+++ b/jemalloc/COPYING
@@ -0,0 +1,27 @@
+Unless otherwise specified, files in the jemalloc source distribution are
+subject to the following license:
+--------------------------------------------------------------------------------
+Copyright (C) 2002-present Jason Evans <jasone@canonware.com>.
+All rights reserved.
+Copyright (C) 2007-2012 Mozilla Foundation.  All rights reserved.
+Copyright (C) 2009-present Facebook, Inc.  All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+1. Redistributions of source code must retain the above copyright notice(s),
+   this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright notice(s),
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+--------------------------------------------------------------------------------
diff --git a/jemalloc/ChangeLog b/jemalloc/ChangeLog
new file mode 100644
index 0000000..e55813b
--- /dev/null
+++ b/jemalloc/ChangeLog
@@ -0,0 +1,1518 @@
+Following are change highlights associated with official releases.  Important
+bug fixes are all mentioned, but some internal enhancements are omitted here for
+brevity.  Much more detail can be found in the git revision history:
+
+    https://github.com/jemalloc/jemalloc
+
+* 5.2.1 (August 5, 2019)
+
+  This release is primarily about Windows.  A critical virtual memory leak is
+  resolved on all Windows platforms.  The regression was present in all releases
+  since 5.0.0.
+
+  Bug fixes:
+  - Fix a severe virtual memory leak on Windows.  This regression was first
+    released in 5.0.0.  (@Ignition, @j0t, @frederik-h, @davidtgoldblatt,
+    @interwq)
+  - Fix size 0 handling in posix_memalign().  This regression was first released
+    in 5.2.0.  (@interwq)
+  - Fix the prof_log unit test which may observe unexpected backtraces from
+    compiler optimizations.  The test was first added in 5.2.0.  (@marxin,
+    @gnzlbg, @interwq)
+  - Fix the declaration of the extent_avail tree.  This regression was first
+    released in 5.1.0.  (@zoulasc)
+  - Fix an incorrect reference in jeprof.  This functionality was first released
+    in 3.0.0.  (@prehistoric-penguin)
+  - Fix an assertion on the deallocation fast-path.  This regression was first
+    released in 5.2.0.  (@yinan1048576)
+  - Fix the TLS_MODEL attribute in headers.  This regression was first released
+    in 5.0.0.  (@zoulasc, @interwq)
+
+  Optimizations and refactors:
+  - Implement opt.retain on Windows and enable by default on 64-bit.  (@interwq,
+    @davidtgoldblatt)
+  - Optimize away a branch on the operator delete[] path.  (@mgrice)
+  - Add format annotation to the format generator function.  (@zoulasc)
+  - Refactor and improve the size class header generation.  (@yinan1048576)
+  - Remove best fit.  (@djwatson)
+  - Avoid blocking on background thread locks for stats.  (@oranagra, @interwq)
+
+* 5.2.0 (April 2, 2019)
+
+  This release includes a few notable improvements, which are summarized below:
+  1) improved fast-path performance from the optimizations by @djwatson; 2)
+  reduced virtual memory fragmentation and metadata usage; and 3) bug fixes on
+  setting the number of background threads.  In addition, peak / spike memory
+  usage is improved with certain allocation patterns.  As usual, the release and
+  prior dev versions have gone through large-scale production testing.
+
+  New features:
+  - Implement oversize_threshold, which uses a dedicated arena for allocations
+    crossing the specified threshold to reduce fragmentation.  (@interwq)
+  - Add extents usage information to stats.  (@tyleretzel)
+  - Log time information for sampled allocations.  (@tyleretzel)
+  - Support 0 size in sdallocx.  (@djwatson)
+  - Output rate for certain counters in malloc_stats.  (@zinoale)
+  - Add configure option --enable-readlinkat, which allows the use of readlinkat
+    over readlink.  (@davidtgoldblatt)
+  - Add configure options --{enable,disable}-{static,shared} to allow not
+    building unwanted libraries.  (@Ericson2314)
+  - Add configure option --disable-libdl to enable fully static builds.
+    (@interwq)
+  - Add mallctl interfaces:
+	+ opt.oversize_threshold (@interwq)
+	+ stats.arenas.<i>.extent_avail (@tyleretzel)
+	+ stats.arenas.<i>.extents.<j>.n{dirty,muzzy,retained} (@tyleretzel)
+	+ stats.arenas.<i>.extents.<j>.{dirty,muzzy,retained}_bytes
+	  (@tyleretzel)
+
+  Portability improvements:
+  - Update MSVC builds.  (@maksqwe, @rustyx)
+  - Workaround a compiler optimizer bug on s390x.  (@rkmisra)
+  - Make use of pthread_set_name_np(3) on FreeBSD.  (@trasz)
+  - Implement malloc_getcpu() to enable percpu_arena for windows.  (@santagada)
+  - Link against -pthread instead of -lpthread.  (@paravoid)
+  - Make background_thread not dependent on libdl.  (@interwq)
+  - Add stringify to fix a linker directive issue on MSVC.  (@daverigby)
+  - Detect and fall back when 8-bit atomics are unavailable.  (@interwq)
+  - Fall back to the default pthread_create if dlsym(3) fails.  (@interwq)
+
+  Optimizations and refactors:
+  - Refactor the TSD module.  (@davidtgoldblatt)
+  - Avoid taking extents_muzzy mutex when muzzy is disabled.  (@interwq)
+  - Avoid taking large_mtx for auto arenas on the tcache flush path.  (@interwq)
+  - Optimize ixalloc by avoiding a size lookup.  (@interwq)
+  - Implement opt.oversize_threshold which uses a dedicated arena for requests
+    crossing the threshold, also eagerly purges the oversize extents.  Default
+    the threshold to 8 MiB.  (@interwq)
+  - Clean compilation with -Wextra.  (@gnzlbg, @jasone)
+  - Refactor the size class module.  (@davidtgoldblatt)
+  - Refactor the stats emitter.  (@tyleretzel)
+  - Optimize pow2_ceil.  (@rkmisra)
+  - Avoid runtime detection of lazy purging on FreeBSD.  (@trasz)
+  - Optimize mmap(2) alignment handling on FreeBSD.  (@trasz)
+  - Improve error handling for THP state initialization.  (@jsteemann)
+  - Rework the malloc() fast path.  (@djwatson)
+  - Rework the free() fast path.  (@djwatson)
+  - Refactor and optimize the tcache fill / flush paths.  (@djwatson)
+  - Optimize sync / lwsync on PowerPC.  (@chmeeedalf)
+  - Bypass extent_dalloc() when retain is enabled.  (@interwq)
+  - Optimize the locking on large deallocation.  (@interwq)
+  - Reduce the number of pages committed from sanity checking in debug build.
+    (@trasz, @interwq)
+  - Deprecate OSSpinLock.  (@interwq)
+  - Lower the default number of background threads to 4 (when the feature
+    is enabled).  (@interwq)
+  - Optimize the trylock spin wait.  (@djwatson)
+  - Use arena index for arena-matching checks.  (@interwq)
+  - Avoid forced decay on thread termination when using background threads.
+    (@interwq)
+  - Disable muzzy decay by default.  (@djwatson, @interwq)
+  - Only initialize libgcc unwinder when profiling is enabled.  (@paravoid,
+    @interwq)
+
+  Bug fixes (all only relevant to jemalloc 5.x):
+  - Fix background thread index issues with max_background_threads.  (@djwatson,
+    @interwq)
+  - Fix stats output for opt.lg_extent_max_active_fit.  (@interwq)
+  - Fix opt.prof_prefix initialization.  (@davidtgoldblatt)
+  - Properly trigger decay on tcache destroy.  (@interwq, @amosbird)
+  - Fix tcache.flush.  (@interwq)
+  - Detect whether explicit extent zero out is necessary with huge pages or
+    custom extent hooks, which may change the purge semantics.  (@interwq)
+  - Fix a side effect caused by extent_max_active_fit combined with decay-based
+    purging, where freed extents can accumulate and not be reused for an
+    extended period of time.  (@interwq, @mpghf)
+  - Fix a missing unlock on extent register error handling.  (@zoulasc)
+
+  Testing:
+  - Simplify the Travis script output.  (@gnzlbg)
+  - Update the test scripts for FreeBSD.  (@devnexen)
+  - Add unit tests for the producer-consumer pattern.  (@interwq)
+  - Add Cirrus-CI config for FreeBSD builds.  (@jasone)
+  - Add size-matching sanity checks on tcache flush.  (@davidtgoldblatt,
+    @interwq)
+
+  Incompatible changes:
+  - Remove --with-lg-page-sizes.  (@davidtgoldblatt)
+
+  Documentation:
+  - Attempt to build docs by default, however skip doc building when xsltproc
+    is missing. (@interwq, @cmuellner)
+
+* 5.1.0 (May 4, 2018)
+
+  This release is primarily about fine-tuning, ranging from several new features
+  to numerous notable performance and portability enhancements.  The release and
+  prior dev versions have been running in multiple large scale applications for
+  months, and the cumulative improvements are substantial in many cases.
+
+  Given the long and successful production runs, this release is likely a good
+  candidate for applications to upgrade, from both jemalloc 5.0 and before.  For
+  performance-critical applications, the newly added TUNING.md provides
+  guidelines on jemalloc tuning.
+
+  New features:
+  - Implement transparent huge page support for internal metadata.  (@interwq)
+  - Add opt.thp to allow enabling / disabling transparent huge pages for all
+    mappings.  (@interwq)
+  - Add maximum background thread count option.  (@djwatson)
+  - Allow prof_active to control opt.lg_prof_interval and prof.gdump.
+    (@interwq)
+  - Allow arena index lookup based on allocation addresses via mallctl.
+    (@lionkov)
+  - Allow disabling initial-exec TLS model.  (@davidtgoldblatt, @KenMacD)
+  - Add opt.lg_extent_max_active_fit to set the max ratio between the size of
+    the active extent selected (to split off from) and the size of the requested
+    allocation.  (@interwq, @davidtgoldblatt)
+  - Add retain_grow_limit to set the max size when growing virtual address
+    space.  (@interwq)
+  - Add mallctl interfaces:
+    + arena.<i>.retain_grow_limit  (@interwq)
+    + arenas.lookup  (@lionkov)
+    + max_background_threads  (@djwatson)
+    + opt.lg_extent_max_active_fit  (@interwq)
+    + opt.max_background_threads  (@djwatson)
+    + opt.metadata_thp  (@interwq)
+    + opt.thp  (@interwq)
+    + stats.metadata_thp  (@interwq)
+
+  Portability improvements:
+  - Support GNU/kFreeBSD configuration.  (@paravoid)
+  - Support m68k, nios2 and SH3 architectures.  (@paravoid)
+  - Fall back to FD_CLOEXEC when O_CLOEXEC is unavailable.  (@zonyitoo)
+  - Fix symbol listing for cross-compiling.  (@tamird)
+  - Fix high bits computation on ARM.  (@davidtgoldblatt, @paravoid)
+  - Disable the CPU_SPINWAIT macro for Power.  (@davidtgoldblatt, @marxin)
+  - Fix MSVC 2015 & 2017 builds.  (@rustyx)
+  - Improve RISC-V support.  (@EdSchouten)
+  - Set name mangling script in strict mode.  (@nicolov)
+  - Avoid MADV_HUGEPAGE on ARM.  (@marxin)
+  - Modify configure to determine return value of strerror_r.
+    (@davidtgoldblatt, @cferris1000)
+  - Make sure CXXFLAGS is tested with CPP compiler.  (@nehaljwani)
+  - Fix 32-bit build on MSVC.  (@rustyx)
+  - Fix external symbol on MSVC.  (@maksqwe)
+  - Avoid a printf format specifier warning.  (@jasone)
+  - Add configure option --disable-initial-exec-tls which can allow jemalloc to
+    be dynamically loaded after program startup.  (@davidtgoldblatt, @KenMacD)
+  - AArch64: Add ILP32 support.  (@cmuellner)
+  - Add --with-lg-vaddr configure option to support cross compiling.
+    (@cmuellner, @davidtgoldblatt)
+
+  Optimizations and refactors:
+  - Improve active extent fit with extent_max_active_fit.  This considerably
+    reduces fragmentation over time and improves virtual memory and metadata
+    usage.  (@davidtgoldblatt, @interwq)
+  - Eagerly coalesce large extents to reduce fragmentation.  (@interwq)
+  - sdallocx: only read size info when page aligned (i.e. possibly sampled),
+    which speeds up the sized deallocation path significantly.  (@interwq)
+  - Avoid attempting new mappings for in place expansion with retain, since
+    it rarely succeeds in practice and causes high overhead.  (@interwq)
+  - Refactor OOM handling in newImpl.  (@wqfish)
+  - Add internal fine-grained logging functionality for debugging use.
+    (@davidtgoldblatt)
+  - Refactor arena / tcache interactions.  (@davidtgoldblatt)
+  - Refactor extent management with dumpable flag.  (@davidtgoldblatt)
+  - Add runtime detection of lazy purging.  (@interwq)
+  - Use pairing heap instead of red-black tree for extents_avail.  (@djwatson)
+  - Use sysctl on startup in FreeBSD.  (@trasz)
+  - Use thread local prng state instead of atomic.  (@djwatson)
+  - Make decay to always purge one more extent than before, because in
+    practice large extents are usually the ones that cross the decay threshold.
+    Purging the additional extent helps save memory as well as reduce VM
+    fragmentation.  (@interwq)
+  - Fast division by dynamic values.  (@davidtgoldblatt)
+  - Improve the fit for aligned allocation.  (@interwq, @edwinsmith)
+  - Refactor extent_t bitpacking.  (@rkmisra)
+  - Optimize the generated assembly for ticker operations.  (@davidtgoldblatt)
+  - Convert stats printing to use a structured text emitter.  (@davidtgoldblatt)
+  - Remove preserve_lru feature for extents management.  (@djwatson)
+  - Consolidate two memory loads into one on the fast deallocation path.
+    (@davidtgoldblatt, @interwq)
+
+  Bug fixes (most of the issues are only relevant to jemalloc 5.0):
+  - Fix deadlock with multithreaded fork in OS X.  (@davidtgoldblatt)
+  - Validate returned file descriptor before use.  (@zonyitoo)
+  - Fix a few background thread initialization and shutdown issues.  (@interwq)
+  - Fix an extent coalesce + decay race by taking both coalescing extents off
+    the LRU list.  (@interwq)
+  - Fix potentially unbound increase during decay, caused by one thread keep
+    stashing memory to purge while other threads generating new pages.  The
+    number of pages to purge is checked to prevent this.  (@interwq)
+  - Fix a FreeBSD bootstrap assertion.  (@strejda, @interwq)
+  - Handle 32 bit mutex counters.  (@rkmisra)
+  - Fix a indexing bug when creating background threads.  (@davidtgoldblatt,
+    @binliu19)
+  - Fix arguments passed to extent_init.  (@yuleniwo, @interwq)
+  - Fix addresses used for ordering mutexes.  (@rkmisra)
+  - Fix abort_conf processing during bootstrap.  (@interwq)
+  - Fix include path order for out-of-tree builds.  (@cmuellner)
+
+  Incompatible changes:
+  - Remove --disable-thp.  (@interwq)
+  - Remove mallctl interfaces:
+    + config.thp  (@interwq)
+
+  Documentation:
+  - Add TUNING.md.  (@interwq, @davidtgoldblatt, @djwatson)
+
+* 5.0.1 (July 1, 2017)
+
+  This bugfix release fixes several issues, most of which are obscure enough
+  that typical applications are not impacted.
+
+  Bug fixes:
+  - Update decay->nunpurged before purging, in order to avoid potential update
+    races and subsequent incorrect purging volume.  (@interwq)
+  - Only abort on dlsym(3) error if the failure impacts an enabled feature (lazy
+    locking and/or background threads).  This mitigates an initialization
+    failure bug for which we still do not have a clear reproduction test case.
+    (@interwq)
+  - Modify tsd management so that it neither crashes nor leaks if a thread's
+    only allocation activity is to call free() after TLS destructors have been
+    executed.  This behavior was observed when operating with GNU libc, and is
+    unlikely to be an issue with other libc implementations.  (@interwq)
+  - Mask signals during background thread creation.  This prevents signals from
+    being inadvertently delivered to background threads.  (@jasone,
+    @davidtgoldblatt, @interwq)
+  - Avoid inactivity checks within background threads, in order to prevent
+    recursive mutex acquisition.  (@interwq)
+  - Fix extent_grow_retained() to use the specified hooks when the
+    arena.<i>.extent_hooks mallctl is used to override the default hooks.
+    (@interwq)
+  - Add missing reentrancy support for custom extent hooks which allocate.
+    (@interwq)
+  - Post-fork(2), re-initialize the list of tcaches associated with each arena
+    to contain no tcaches except the forking thread's.  (@interwq)
+  - Add missing post-fork(2) mutex reinitialization for extent_grow_mtx.  This
+    fixes potential deadlocks after fork(2).  (@interwq)
+  - Enforce minimum autoconf version (currently 2.68), since 2.63 is known to
+    generate corrupt configure scripts.  (@jasone)
+  - Ensure that the configured page size (--with-lg-page) is no larger than the
+    configured huge page size (--with-lg-hugepage).  (@jasone)
+
+* 5.0.0 (June 13, 2017)
+
+  Unlike all previous jemalloc releases, this release does not use naturally
+  aligned "chunks" for virtual memory management, and instead uses page-aligned
+  "extents".  This change has few externally visible effects, but the internal
+  impacts are... extensive.  Many other internal changes combine to make this
+  the most cohesively designed version of jemalloc so far, with ample
+  opportunity for further enhancements.
+
+  Continuous integration is now an integral aspect of development thanks to the
+  efforts of @davidtgoldblatt, and the dev branch tends to remain reasonably
+  stable on the tested platforms (Linux, FreeBSD, macOS, and Windows).  As a
+  side effect the official release frequency may decrease over time.
+
+  New features:
+  - Implement optional per-CPU arena support; threads choose which arena to use
+    based on current CPU rather than on fixed thread-->arena associations.
+    (@interwq)
+  - Implement two-phase decay of unused dirty pages.  Pages transition from
+    dirty-->muzzy-->clean, where the first phase transition relies on
+    madvise(... MADV_FREE) semantics, and the second phase transition discards
+    pages such that they are replaced with demand-zeroed pages on next access.
+    (@jasone)
+  - Increase decay time resolution from seconds to milliseconds.  (@jasone)
+  - Implement opt-in per CPU background threads, and use them for asynchronous
+    decay-driven unused dirty page purging.  (@interwq)
+  - Add mutex profiling, which collects a variety of statistics useful for
+    diagnosing overhead/contention issues.  (@interwq)
+  - Add C++ new/delete operator bindings.  (@djwatson)
+  - Support manually created arena destruction, such that all data and metadata
+    are discarded.  Add MALLCTL_ARENAS_DESTROYED for accessing merged stats
+    associated with destroyed arenas.  (@jasone)
+  - Add MALLCTL_ARENAS_ALL as a fixed index for use in accessing
+    merged/destroyed arena statistics via mallctl.  (@jasone)
+  - Add opt.abort_conf to optionally abort if invalid configuration options are
+    detected during initialization.  (@interwq)
+  - Add opt.stats_print_opts, so that e.g. JSON output can be selected for the
+    stats dumped during exit if opt.stats_print is true.  (@jasone)
+  - Add --with-version=VERSION for use when embedding jemalloc into another
+    project's git repository.  (@jasone)
+  - Add --disable-thp to support cross compiling.  (@jasone)
+  - Add --with-lg-hugepage to support cross compiling.  (@jasone)
+  - Add mallctl interfaces (various authors):
+    + background_thread
+    + opt.abort_conf
+    + opt.retain
+    + opt.percpu_arena
+    + opt.background_thread
+    + opt.{dirty,muzzy}_decay_ms
+    + opt.stats_print_opts
+    + arena.<i>.initialized
+    + arena.<i>.destroy
+    + arena.<i>.{dirty,muzzy}_decay_ms
+    + arena.<i>.extent_hooks
+    + arenas.{dirty,muzzy}_decay_ms
+    + arenas.bin.<i>.slab_size
+    + arenas.nlextents
+    + arenas.lextent.<i>.size
+    + arenas.create
+    + stats.background_thread.{num_threads,num_runs,run_interval}
+    + stats.mutexes.{ctl,background_thread,prof,reset}.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+    + stats.arenas.<i>.{dirty,muzzy}_decay_ms
+    + stats.arenas.<i>.uptime
+    + stats.arenas.<i>.{pmuzzy,base,internal,resident}
+    + stats.arenas.<i>.{dirty,muzzy}_{npurge,nmadvise,purged}
+    + stats.arenas.<i>.bins.<j>.{nslabs,reslabs,curslabs}
+    + stats.arenas.<i>.bins.<j>.mutex.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+    + stats.arenas.<i>.lextents.<j>.{nmalloc,ndalloc,nrequests,curlextents}
+    + stats.arenas.i.mutexes.{large,extent_avail,extents_dirty,extents_muzzy,
+      extents_retained,decay_dirty,decay_muzzy,base,tcache_list}.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+
+  Portability improvements:
+  - Improve reentrant allocation support, such that deadlock is less likely if
+    e.g. a system library call in turn allocates memory.  (@davidtgoldblatt,
+    @interwq)
+  - Support static linking of jemalloc with glibc.  (@djwatson)
+
+  Optimizations and refactors:
+  - Organize virtual memory as "extents" of virtual memory pages, rather than as
+    naturally aligned "chunks", and store all metadata in arbitrarily distant
+    locations.  This reduces virtual memory external fragmentation, and will
+    interact better with huge pages (not yet explicitly supported).  (@jasone)
+  - Fold large and huge size classes together; only small and large size classes
+    remain.  (@jasone)
+  - Unify the allocation paths, and merge most fast-path branching decisions.
+    (@davidtgoldblatt, @interwq)
+  - Embed per thread automatic tcache into thread-specific data, which reduces
+    conditional branches and dereferences.  Also reorganize tcache to increase
+    fast-path data locality.  (@interwq)
+  - Rewrite atomics to closely model the C11 API, convert various
+    synchronization from mutex-based to atomic, and use the explicit memory
+    ordering control to resolve various hypothetical races without increasing
+    synchronization overhead.  (@davidtgoldblatt)
+  - Extensively optimize rtree via various methods:
+    + Add multiple layers of rtree lookup caching, since rtree lookups are now
+      part of fast-path deallocation.  (@interwq)
+    + Determine rtree layout at compile time.  (@jasone)
+    + Make the tree shallower for common configurations.  (@jasone)
+    + Embed the root node in the top-level rtree data structure, thus avoiding
+      one level of indirection.  (@jasone)
+    + Further specialize leaf elements as compared to internal node elements,
+      and directly embed extent metadata needed for fast-path deallocation.
+      (@jasone)
+    + Ignore leading always-zero address bits (architecture-specific).
+      (@jasone)
+  - Reorganize headers (ongoing work) to make them hermetic, and disentangle
+    various module dependencies.  (@davidtgoldblatt)
+  - Convert various internal data structures such as size class metadata from
+    boot-time-initialized to compile-time-initialized.  Propagate resulting data
+    structure simplifications, such as making arena metadata fixed-size.
+    (@jasone)
+  - Simplify size class lookups when constrained to size classes that are
+    multiples of the page size.  This speeds lookups, but the primary benefit is
+    complexity reduction in code that was the source of numerous regressions.
+    (@jasone)
+  - Lock individual extents when possible for localized extent operations,
+    rather than relying on a top-level arena lock.  (@davidtgoldblatt, @jasone)
+  - Use first fit layout policy instead of best fit, in order to improve
+    packing.  (@jasone)
+  - If munmap(2) is not in use, use an exponential series to grow each arena's
+    virtual memory, so that the number of disjoint virtual memory mappings
+    remains low.  (@jasone)
+  - Implement per arena base allocators, so that arenas never share any virtual
+    memory pages.  (@jasone)
+  - Automatically generate private symbol name mangling macros.  (@jasone)
+
+  Incompatible changes:
+  - Replace chunk hooks with an expanded/normalized set of extent hooks.
+    (@jasone)
+  - Remove ratio-based purging.  (@jasone)
+  - Remove --disable-tcache.  (@jasone)
+  - Remove --disable-tls.  (@jasone)
+  - Remove --enable-ivsalloc.  (@jasone)
+  - Remove --with-lg-size-class-group.  (@jasone)
+  - Remove --with-lg-tiny-min.  (@jasone)
+  - Remove --disable-cc-silence.  (@jasone)
+  - Remove --enable-code-coverage.  (@jasone)
+  - Remove --disable-munmap (replaced by opt.retain).  (@jasone)
+  - Remove Valgrind support.  (@jasone)
+  - Remove quarantine support.  (@jasone)
+  - Remove redzone support.  (@jasone)
+  - Remove mallctl interfaces (various authors):
+    + config.munmap
+    + config.tcache
+    + config.tls
+    + config.valgrind
+    + opt.lg_chunk
+    + opt.purge
+    + opt.lg_dirty_mult
+    + opt.decay_time
+    + opt.quarantine
+    + opt.redzone
+    + opt.thp
+    + arena.<i>.lg_dirty_mult
+    + arena.<i>.decay_time
+    + arena.<i>.chunk_hooks
+    + arenas.initialized
+    + arenas.lg_dirty_mult
+    + arenas.decay_time
+    + arenas.bin.<i>.run_size
+    + arenas.nlruns
+    + arenas.lrun.<i>.size
+    + arenas.nhchunks
+    + arenas.hchunk.<i>.size
+    + arenas.extend
+    + stats.cactive
+    + stats.arenas.<i>.lg_dirty_mult
+    + stats.arenas.<i>.decay_time
+    + stats.arenas.<i>.metadata.{mapped,allocated}
+    + stats.arenas.<i>.{npurge,nmadvise,purged}
+    + stats.arenas.<i>.huge.{allocated,nmalloc,ndalloc,nrequests}
+    + stats.arenas.<i>.bins.<j>.{nruns,reruns,curruns}
+    + stats.arenas.<i>.lruns.<j>.{nmalloc,ndalloc,nrequests,curruns}
+    + stats.arenas.<i>.hchunks.<j>.{nmalloc,ndalloc,nrequests,curhchunks}
+
+  Bug fixes:
+  - Improve interval-based profile dump triggering to dump only one profile when
+    a single allocation's size exceeds the interval.  (@jasone)
+  - Use prefixed function names (as controlled by --with-jemalloc-prefix) when
+    pruning backtrace frames in jeprof.  (@jasone)
+
+* 4.5.0 (February 28, 2017)
+
+  This is the first release to benefit from much broader continuous integration
+  testing, thanks to @davidtgoldblatt.  Had we had this testing infrastructure
+  in place for prior releases, it would have caught all of the most serious
+  regressions fixed by this release.
+
+  New features:
+  - Add --disable-thp and the opt.thp mallctl to provide opt-out mechanisms for
+    transparent huge page integration.  (@jasone)
+  - Update zone allocator integration to work with macOS 10.12.  (@glandium)
+  - Restructure *CFLAGS configuration, so that CFLAGS behaves typically, and
+    EXTRA_CFLAGS provides a way to specify e.g. -Werror during building, but not
+    during configuration.  (@jasone, @ronawho)
+
+  Bug fixes:
+  - Fix DSS (sbrk(2)-based) allocation.  This regression was first released in
+    4.3.0.  (@jasone)
+  - Handle race in per size class utilization computation.  This functionality
+    was first released in 4.0.0.  (@interwq)
+  - Fix lock order reversal during gdump.  (@jasone)
+  - Fix/refactor tcache synchronization.  This regression was first released in
+    4.0.0.  (@jasone)
+  - Fix various JSON-formatted malloc_stats_print() bugs.  This functionality
+    was first released in 4.3.0.  (@jasone)
+  - Fix huge-aligned allocation.  This regression was first released in 4.4.0.
+    (@jasone)
+  - When transparent huge page integration is enabled, detect what state pages
+    start in according to the kernel's current operating mode, and only convert
+    arena chunks to non-huge during purging if that is not their initial state.
+    This functionality was first released in 4.4.0.  (@jasone)
+  - Fix lg_chunk clamping for the --enable-cache-oblivious --disable-fill case.
+    This regression was first released in 4.0.0.  (@jasone, @428desmo)
+  - Properly detect sparc64 when building for Linux.  (@glaubitz)
+
+* 4.4.0 (December 3, 2016)
+
+  New features:
+  - Add configure support for *-*-linux-android.  (@cferris1000, @jasone)
+  - Add the --disable-syscall configure option, for use on systems that place
+    security-motivated limitations on syscall(2).  (@jasone)
+  - Add support for Debian GNU/kFreeBSD.  (@thesam)
+
+  Optimizations:
+  - Add extent serial numbers and use them where appropriate as a sort key that
+    is higher priority than address, so that the allocation policy prefers older
+    extents.  This tends to improve locality (decrease fragmentation) when
+    memory grows downward.  (@jasone)
+  - Refactor madvise(2) configuration so that MADV_FREE is detected and utilized
+    on Linux 4.5 and newer.  (@jasone)
+  - Mark partially purged arena chunks as non-huge-page.  This improves
+    interaction with Linux's transparent huge page functionality.  (@jasone)
+
+  Bug fixes:
+  - Fix size class computations for edge conditions involving extremely large
+    allocations.  This regression was first released in 4.0.0.  (@jasone,
+    @ingvarha)
+  - Remove overly restrictive assertions related to the cactive statistic.  This
+    regression was first released in 4.1.0.  (@jasone)
+  - Implement a more reliable detection scheme for os_unfair_lock on macOS.
+    (@jszakmeister)
+
+* 4.3.1 (November 7, 2016)
+
+  Bug fixes:
+  - Fix a severe virtual memory leak.  This regression was first released in
+    4.3.0.  (@interwq, @jasone)
+  - Refactor atomic and prng APIs to restore support for 32-bit platforms that
+    use pre-C11 toolchains, e.g. FreeBSD's mips.  (@jasone)
+
+* 4.3.0 (November 4, 2016)
+
+  This is the first release that passes the test suite for multiple Windows
+  configurations, thanks in large part to @glandium setting up continuous
+  integration via AppVeyor (and Travis CI for Linux and OS X).
+
+  New features:
+  - Add "J" (JSON) support to malloc_stats_print().  (@jasone)
+  - Add Cray compiler support.  (@ronawho)
+
+  Optimizations:
+  - Add/use adaptive spinning for bootstrapping and radix tree node
+    initialization.  (@jasone)
+
+  Bug fixes:
+  - Fix large allocation to search starting in the optimal size class heap,
+    which can substantially reduce virtual memory churn and fragmentation.  This
+    regression was first released in 4.0.0.  (@mjp41, @jasone)
+  - Fix stats.arenas.<i>.nthreads accounting.  (@interwq)
+  - Fix and simplify decay-based purging.  (@jasone)
+  - Make DSS (sbrk(2)-related) operations lockless, which resolves potential
+    deadlocks during thread exit.  (@jasone)
+  - Fix over-sized allocation of radix tree leaf nodes.  (@mjp41, @ogaun,
+    @jasone)
+  - Fix over-sized allocation of arena_t (plus associated stats) data
+    structures.  (@jasone, @interwq)
+  - Fix EXTRA_CFLAGS to not affect configuration.  (@jasone)
+  - Fix a Valgrind integration bug.  (@ronawho)
+  - Disallow 0x5a junk filling when running in Valgrind.  (@jasone)
+  - Fix a file descriptor leak on Linux.  This regression was first released in
+    4.2.0.  (@vsarunas, @jasone)
+  - Fix static linking of jemalloc with glibc.  (@djwatson)
+  - Use syscall(2) rather than {open,read,close}(2) during boot on Linux.  This
+    works around other libraries' system call wrappers performing reentrant
+    allocation.  (@kspinka, @Whissi, @jasone)
+  - Fix OS X default zone replacement to work with OS X 10.12.  (@glandium,
+    @jasone)
+  - Fix cached memory management to avoid needless commit/decommit operations
+    during purging, which resolves permanent virtual memory map fragmentation
+    issues on Windows.  (@mjp41, @jasone)
+  - Fix TSD fetches to avoid (recursive) allocation.  This is relevant to
+    non-TLS and Windows configurations.  (@jasone)
+  - Fix malloc_conf overriding to work on Windows.  (@jasone)
+  - Forcibly disable lazy-lock on Windows (was forcibly *enabled*).  (@jasone)
+
+* 4.2.1 (June 8, 2016)
+
+  Bug fixes:
+  - Fix bootstrapping issues for configurations that require allocation during
+    tsd initialization (e.g. --disable-tls).  (@cferris1000, @jasone)
+  - Fix gettimeofday() version of nstime_update().  (@ronawho)
+  - Fix Valgrind regressions in calloc() and chunk_alloc_wrapper().  (@ronawho)
+  - Fix potential VM map fragmentation regression.  (@jasone)
+  - Fix opt_zero-triggered in-place huge reallocation zeroing.  (@jasone)
+  - Fix heap profiling context leaks in reallocation edge cases.  (@jasone)
+
+* 4.2.0 (May 12, 2016)
+
+  New features:
+  - Add the arena.<i>.reset mallctl, which makes it possible to discard all of
+    an arena's allocations in a single operation.  (@jasone)
+  - Add the stats.retained and stats.arenas.<i>.retained statistics.  (@jasone)
+  - Add the --with-version configure option.  (@jasone)
+  - Support --with-lg-page values larger than actual page size.  (@jasone)
+
+  Optimizations:
+  - Use pairing heaps rather than red-black trees for various hot data
+    structures.  (@djwatson, @jasone)
+  - Streamline fast paths of rtree operations.  (@jasone)
+  - Optimize the fast paths of calloc() and [m,d,sd]allocx().  (@jasone)
+  - Decommit unused virtual memory if the OS does not overcommit.  (@jasone)
+  - Specify MAP_NORESERVE on Linux if [heuristic] overcommit is active, in order
+    to avoid unfortunate interactions during fork(2).  (@jasone)
+
+  Bug fixes:
+  - Fix chunk accounting related to triggering gdump profiles.  (@jasone)
+  - Link against librt for clock_gettime(2) if glibc < 2.17.  (@jasone)
+  - Scale leak report summary according to sampling probability.  (@jasone)
+
+* 4.1.1 (May 3, 2016)
+
+  This bugfix release resolves a variety of mostly minor issues, though the
+  bitmap fix is critical for 64-bit Windows.
+
+  Bug fixes:
+  - Fix the linear scan version of bitmap_sfu() to shift by the proper amount
+    even when sizeof(long) is not the same as sizeof(void *), as on 64-bit
+    Windows.  (@jasone)
+  - Fix hashing functions to avoid unaligned memory accesses (and resulting
+    crashes).  This is relevant at least to some ARM-based platforms.
+    (@rkmisra)
+  - Fix fork()-related lock rank ordering reversals.  These reversals were
+    unlikely to cause deadlocks in practice except when heap profiling was
+    enabled and active.  (@jasone)
+  - Fix various chunk leaks in OOM code paths.  (@jasone)
+  - Fix malloc_stats_print() to print opt.narenas correctly.  (@jasone)
+  - Fix MSVC-specific build/test issues.  (@rustyx, @yuslepukhin)
+  - Fix a variety of test failures that were due to test fragility rather than
+    core bugs.  (@jasone)
+
+* 4.1.0 (February 28, 2016)
+
+  This release is primarily about optimizations, but it also incorporates a lot
+  of portability-motivated refactoring and enhancements.  Many people worked on
+  this release, to an extent that even with the omission here of minor changes
+  (see git revision history), and of the people who reported and diagnosed
+  issues, so much of the work was contributed that starting with this release,
+  changes are annotated with author credits to help reflect the collaborative
+  effort involved.
+
+  New features:
+  - Implement decay-based unused dirty page purging, a major optimization with
+    mallctl API impact.  This is an alternative to the existing ratio-based
+    unused dirty page purging, and is intended to eventually become the sole
+    purging mechanism.  New mallctls:
+    + opt.purge
+    + opt.decay_time
+    + arena.<i>.decay
+    + arena.<i>.decay_time
+    + arenas.decay_time
+    + stats.arenas.<i>.decay_time
+    (@jasone, @cevans87)
+  - Add --with-malloc-conf, which makes it possible to embed a default
+    options string during configuration.  This was motivated by the desire to
+    specify --with-malloc-conf=purge:decay , since the default must remain
+    purge:ratio until the 5.0.0 release.  (@jasone)
+  - Add MS Visual Studio 2015 support.  (@rustyx, @yuslepukhin)
+  - Make *allocx() size class overflow behavior defined.  The maximum
+    size class is now less than PTRDIFF_MAX to protect applications against
+    numerical overflow, and all allocation functions are guaranteed to indicate
+    errors rather than potentially crashing if the request size exceeds the
+    maximum size class.  (@jasone)
+  - jeprof:
+    + Add raw heap profile support.  (@jasone)
+    + Add --retain and --exclude for backtrace symbol filtering.  (@jasone)
+
+  Optimizations:
+  - Optimize the fast path to combine various bootstrapping and configuration
+    checks and execute more streamlined code in the common case.  (@interwq)
+  - Use linear scan for small bitmaps (used for small object tracking).  In
+    addition to speeding up bitmap operations on 64-bit systems, this reduces
+    allocator metadata overhead by approximately 0.2%.  (@djwatson)
+  - Separate arena_avail trees, which substantially speeds up run tree
+    operations.  (@djwatson)
+  - Use memoization (boot-time-computed table) for run quantization.  Separate
+    arena_avail trees reduced the importance of this optimization.  (@jasone)
+  - Attempt mmap-based in-place huge reallocation.  This can dramatically speed
+    up incremental huge reallocation.  (@jasone)
+
+  Incompatible changes:
+  - Make opt.narenas unsigned rather than size_t.  (@jasone)
+
+  Bug fixes:
+  - Fix stats.cactive accounting regression.  (@rustyx, @jasone)
+  - Handle unaligned keys in hash().  This caused problems for some ARM systems.
+    (@jasone, @cferris1000)
+  - Refactor arenas array.  In addition to fixing a fork-related deadlock, this
+    makes arena lookups faster and simpler.  (@jasone)
+  - Move retained memory allocation out of the default chunk allocation
+    function, to a location that gets executed even if the application installs
+    a custom chunk allocation function.  This resolves a virtual memory leak.
+    (@buchgr)
+  - Fix a potential tsd cleanup leak.  (@cferris1000, @jasone)
+  - Fix run quantization.  In practice this bug had no impact unless
+    applications requested memory with alignment exceeding one page.
+    (@jasone, @djwatson)
+  - Fix LinuxThreads-specific bootstrapping deadlock.  (Cosmin Paraschiv)
+  - jeprof:
+    + Don't discard curl options if timeout is not defined.  (@djwatson)
+    + Detect failed profile fetches.  (@djwatson)
+  - Fix stats.arenas.<i>.{dss,lg_dirty_mult,decay_time,pactive,pdirty} for
+    --disable-stats case.  (@jasone)
+
+* 4.0.4 (October 24, 2015)
+
+  This bugfix release fixes another xallocx() regression.  No other regressions
+  have come to light in over a month, so this is likely a good starting point
+  for people who prefer to wait for "dot one" releases with all the major issues
+  shaken out.
+
+  Bug fixes:
+  - Fix xallocx(..., MALLOCX_ZERO to zero the last full trailing page of large
+    allocations that have been randomly assigned an offset of 0 when
+    --enable-cache-oblivious configure option is enabled.
+
+* 4.0.3 (September 24, 2015)
+
+  This bugfix release continues the trend of xallocx() and heap profiling fixes.
+
+  Bug fixes:
+  - Fix xallocx(..., MALLOCX_ZERO) to zero all trailing bytes of large
+    allocations when --enable-cache-oblivious configure option is enabled.
+  - Fix xallocx(..., MALLOCX_ZERO) to zero trailing bytes of huge allocations
+    when resizing from/to a size class that is not a multiple of the chunk size.
+  - Fix prof_tctx_dump_iter() to filter out nodes that were created after heap
+    profile dumping started.
+  - Work around a potentially bad thread-specific data initialization
+    interaction with NPTL (glibc's pthreads implementation).
+
+* 4.0.2 (September 21, 2015)
+
+  This bugfix release addresses a few bugs specific to heap profiling.
+
+  Bug fixes:
+  - Fix ixallocx_prof_sample() to never modify nor create sampled small
+    allocations.  xallocx() is in general incapable of moving small allocations,
+    so this fix removes buggy code without loss of generality.
+  - Fix irallocx_prof_sample() to always allocate large regions, even when
+    alignment is non-zero.
+  - Fix prof_alloc_rollback() to read tdata from thread-specific data rather
+    than dereferencing a potentially invalid tctx.
+
+* 4.0.1 (September 15, 2015)
+
+  This is a bugfix release that is somewhat high risk due to the amount of
+  refactoring required to address deep xallocx() problems.  As a side effect of
+  these fixes, xallocx() now tries harder to partially fulfill requests for
+  optional extra space.  Note that a couple of minor heap profiling
+  optimizations are included, but these are better thought of as performance
+  fixes that were integral to discovering most of the other bugs.
+
+  Optimizations:
+  - Avoid a chunk metadata read in arena_prof_tctx_set(), since it is in the
+    fast path when heap profiling is enabled.  Additionally, split a special
+    case out into arena_prof_tctx_reset(), which also avoids chunk metadata
+    reads.
+  - Optimize irallocx_prof() to optimistically update the sampler state.  The
+    prior implementation appears to have been a holdover from when
+    rallocx()/xallocx() functionality was combined as rallocm().
+
+  Bug fixes:
+  - Fix TLS configuration such that it is enabled by default for platforms on
+    which it works correctly.
+  - Fix arenas_cache_cleanup() and arena_get_hard() to handle
+    allocation/deallocation within the application's thread-specific data
+    cleanup functions even after arenas_cache is torn down.
+  - Fix xallocx() bugs related to size+extra exceeding HUGE_MAXCLASS.
+  - Fix chunk purge hook calls for in-place huge shrinking reallocation to
+    specify the old chunk size rather than the new chunk size.  This bug caused
+    no correctness issues for the default chunk purge function, but was
+    visible to custom functions set via the "arena.<i>.chunk_hooks" mallctl.
+  - Fix heap profiling bugs:
+    + Fix heap profiling to distinguish among otherwise identical sample sites
+      with interposed resets (triggered via the "prof.reset" mallctl).  This bug
+      could cause data structure corruption that would most likely result in a
+      segfault.
+    + Fix irealloc_prof() to prof_alloc_rollback() on OOM.
+    + Make one call to prof_active_get_unlocked() per allocation event, and use
+      the result throughout the relevant functions that handle an allocation
+      event.  Also add a missing check in prof_realloc().  These fixes protect
+      allocation events against concurrent prof_active changes.
+    + Fix ixallocx_prof() to pass usize_max and zero to ixallocx_prof_sample()
+      in the correct order.
+    + Fix prof_realloc() to call prof_free_sampled_object() after calling
+      prof_malloc_sample_object().  Prior to this fix, if tctx and old_tctx were
+      the same, the tctx could have been prematurely destroyed.
+  - Fix portability bugs:
+    + Don't bitshift by negative amounts when encoding/decoding run sizes in
+      chunk header maps.  This affected systems with page sizes greater than 8
+      KiB.
+    + Rename index_t to szind_t to avoid an existing type on Solaris.
+    + Add JEMALLOC_CXX_THROW to the memalign() function prototype, in order to
+      match glibc and avoid compilation errors when including both
+      jemalloc/jemalloc.h and malloc.h in C++ code.
+    + Don't assume that /bin/sh is appropriate when running size_classes.sh
+      during configuration.
+    + Consider __sparcv9 a synonym for __sparc64__ when defining LG_QUANTUM.
+    + Link tests to librt if it contains clock_gettime(2).
+
+* 4.0.0 (August 17, 2015)
+
+  This version contains many speed and space optimizations, both minor and
+  major.  The major themes are generalization, unification, and simplification.
+  Although many of these optimizations cause no visible behavior change, their
+  cumulative effect is substantial.
+
+  New features:
+  - Normalize size class spacing to be consistent across the complete size
+    range.  By default there are four size classes per size doubling, but this
+    is now configurable via the --with-lg-size-class-group option.  Also add the
+    --with-lg-page, --with-lg-page-sizes, --with-lg-quantum, and
+    --with-lg-tiny-min options, which can be used to tweak page and size class
+    settings.  Impacts:
+    + Worst case performance for incrementally growing/shrinking reallocation
+      is improved because there are far fewer size classes, and therefore
+      copying happens less often.
+    + Internal fragmentation is limited to 20% for all but the smallest size
+      classes (those less than four times the quantum).  (1B + 4 KiB)
+      and (1B + 4 MiB) previously suffered nearly 50% internal fragmentation.
+    + Chunk fragmentation tends to be lower because there are fewer distinct run
+      sizes to pack.
+  - Add support for explicit tcaches.  The "tcache.create", "tcache.flush", and
+    "tcache.destroy" mallctls control tcache lifetime and flushing, and the
+    MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to the *allocx() API
+    control which tcache is used for each operation.
+  - Implement per thread heap profiling, as well as the ability to
+    enable/disable heap profiling on a per thread basis.  Add the "prof.reset",
+    "prof.lg_sample", "thread.prof.name", "thread.prof.active",
+    "opt.prof_thread_active_init", "prof.thread_active_init", and
+    "thread.prof.active" mallctls.
+  - Add support for per arena application-specified chunk allocators, configured
+    via the "arena.<i>.chunk_hooks" mallctl.
+  - Refactor huge allocation to be managed by arenas, so that arenas now
+    function as general purpose independent allocators.  This is important in
+    the context of user-specified chunk allocators, aside from the scalability
+    benefits.  Related new statistics:
+    + The "stats.arenas.<i>.huge.allocated", "stats.arenas.<i>.huge.nmalloc",
+      "stats.arenas.<i>.huge.ndalloc", and "stats.arenas.<i>.huge.nrequests"
+      mallctls provide high level per arena huge allocation statistics.
+    + The "arenas.nhchunks", "arenas.hchunk.<i>.size",
+      "stats.arenas.<i>.hchunks.<j>.nmalloc",
+      "stats.arenas.<i>.hchunks.<j>.ndalloc",
+      "stats.arenas.<i>.hchunks.<j>.nrequests", and
+      "stats.arenas.<i>.hchunks.<j>.curhchunks" mallctls provide per size class
+      statistics.
+  - Add the 'util' column to malloc_stats_print() output, which reports the
+    proportion of available regions that are currently in use for each small
+    size class.
+  - Add "alloc" and "free" modes for for junk filling (see the "opt.junk"
+    mallctl), so that it is possible to separately enable junk filling for
+    allocation versus deallocation.
+  - Add the jemalloc-config script, which provides information about how
+    jemalloc was configured, and how to integrate it into application builds.
+  - Add metadata statistics, which are accessible via the "stats.metadata",
+    "stats.arenas.<i>.metadata.mapped", and
+    "stats.arenas.<i>.metadata.allocated" mallctls.
+  - Add the "stats.resident" mallctl, which reports the upper limit of
+    physically resident memory mapped by the allocator.
+  - Add per arena control over unused dirty page purging, via the
+    "arenas.lg_dirty_mult", "arena.<i>.lg_dirty_mult", and
+    "stats.arenas.<i>.lg_dirty_mult" mallctls.
+  - Add the "prof.gdump" mallctl, which makes it possible to toggle the gdump
+    feature on/off during program execution.
+  - Add sdallocx(), which implements sized deallocation.  The primary
+    optimization over dallocx() is the removal of a metadata read, which often
+    suffers an L1 cache miss.
+  - Add missing header includes in jemalloc/jemalloc.h, so that applications
+    only have to #include <jemalloc/jemalloc.h>.
+  - Add support for additional platforms:
+    + Bitrig
+    + Cygwin
+    + DragonFlyBSD
+    + iOS
+    + OpenBSD
+    + OpenRISC/or1k
+
+  Optimizations:
+  - Maintain dirty runs in per arena LRUs rather than in per arena trees of
+    dirty-run-containing chunks.  In practice this change significantly reduces
+    dirty page purging volume.
+  - Integrate whole chunks into the unused dirty page purging machinery.  This
+    reduces the cost of repeated huge allocation/deallocation, because it
+    effectively introduces a cache of chunks.
+  - Split the arena chunk map into two separate arrays, in order to increase
+    cache locality for the frequently accessed bits.
+  - Move small run metadata out of runs, into arena chunk headers.  This reduces
+    run fragmentation, smaller runs reduce external fragmentation for small size
+    classes, and packed (less uniformly aligned) metadata layout improves CPU
+    cache set distribution.
+  - Randomly distribute large allocation base pointer alignment relative to page
+    boundaries in order to more uniformly utilize CPU cache sets.  This can be
+    disabled via the --disable-cache-oblivious configure option, and queried via
+    the "config.cache_oblivious" mallctl.
+  - Micro-optimize the fast paths for the public API functions.
+  - Refactor thread-specific data to reside in a single structure.  This assures
+    that only a single TLS read is necessary per call into the public API.
+  - Implement in-place huge allocation growing and shrinking.
+  - Refactor rtree (radix tree for chunk lookups) to be lock-free, and make
+    additional optimizations that reduce maximum lookup depth to one or two
+    levels.  This resolves what was a concurrency bottleneck for per arena huge
+    allocation, because a global data structure is critical for determining
+    which arenas own which huge allocations.
+
+  Incompatible changes:
+  - Replace --enable-cc-silence with --disable-cc-silence to suppress spurious
+    warnings by default.
+  - Assure that the constness of malloc_usable_size()'s return type matches that
+    of the system implementation.
+  - Change the heap profile dump format to support per thread heap profiling,
+    rename pprof to jeprof, and enhance it with the --thread=<n> option.  As a
+    result, the bundled jeprof must now be used rather than the upstream
+    (gperftools) pprof.
+  - Disable "opt.prof_final" by default, in order to avoid atexit(3), which can
+    internally deadlock on some platforms.
+  - Change the "arenas.nlruns" mallctl type from size_t to unsigned.
+  - Replace the "stats.arenas.<i>.bins.<j>.allocated" mallctl with
+    "stats.arenas.<i>.bins.<j>.curregs".
+  - Ignore MALLOC_CONF in set{uid,gid,cap} binaries.
+  - Ignore MALLOCX_ARENA(a) in dallocx(), in favor of using the
+    MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to control tcache usage.
+
+  Removed features:
+  - Remove the *allocm() API, which is superseded by the *allocx() API.
+  - Remove the --enable-dss options, and make dss non-optional on all platforms
+    which support sbrk(2).
+  - Remove the "arenas.purge" mallctl, which was obsoleted by the
+    "arena.<i>.purge" mallctl in 3.1.0.
+  - Remove the unnecessary "opt.valgrind" mallctl; jemalloc automatically
+    detects whether it is running inside Valgrind.
+  - Remove the "stats.huge.allocated", "stats.huge.nmalloc", and
+    "stats.huge.ndalloc" mallctls.
+  - Remove the --enable-mremap option.
+  - Remove the "stats.chunks.current", "stats.chunks.total", and
+    "stats.chunks.high" mallctls.
+
+  Bug fixes:
+  - Fix the cactive statistic to decrease (rather than increase) when active
+    memory decreases.  This regression was first released in 3.5.0.
+  - Fix OOM handling in memalign() and valloc().  A variant of this bug existed
+    in all releases since 2.0.0, which introduced these functions.
+  - Fix an OOM-related regression in arena_tcache_fill_small(), which could
+    cause cache corruption on OOM.  This regression was present in all releases
+    from 2.2.0 through 3.6.0.
+  - Fix size class overflow handling for malloc(), posix_memalign(), memalign(),
+    calloc(), and realloc() when profiling is enabled.
+  - Fix the "arena.<i>.dss" mallctl to return an error if "primary" or
+    "secondary" precedence is specified, but sbrk(2) is not supported.
+  - Fix fallback lg_floor() implementations to handle extremely large inputs.
+  - Ensure the default purgeable zone is after the default zone on OS X.
+  - Fix latent bugs in atomic_*().
+  - Fix the "arena.<i>.dss" mallctl to handle read-only calls.
+  - Fix tls_model configuration to enable the initial-exec model when possible.
+  - Mark malloc_conf as a weak symbol so that the application can override it.
+  - Correctly detect glibc's adaptive pthread mutexes.
+  - Fix the --without-export configure option.
+
+* 3.6.0 (March 31, 2014)
+
+  This version contains a critical bug fix for a regression present in 3.5.0 and
+  3.5.1.
+
+  Bug fixes:
+  - Fix a regression in arena_chunk_alloc() that caused crashes during
+    small/large allocation if chunk allocation failed.  In the absence of this
+    bug, chunk allocation failure would result in allocation failure, e.g.  NULL
+    return from malloc().  This regression was introduced in 3.5.0.
+  - Fix backtracing for gcc intrinsics-based backtracing by specifying
+    -fno-omit-frame-pointer to gcc.  Note that the application (and all the
+    libraries it links to) must also be compiled with this option for
+    backtracing to be reliable.
+  - Use dss allocation precedence for huge allocations as well as small/large
+    allocations.
+  - Fix test assertion failure message formatting.  This bug did not manifest on
+    x86_64 systems because of implementation subtleties in va_list.
+  - Fix inconsequential test failures for hash and SFMT code.
+
+  New features:
+  - Support heap profiling on FreeBSD.  This feature depends on the proc
+    filesystem being mounted during heap profile dumping.
+
+* 3.5.1 (February 25, 2014)
+
+  This version primarily addresses minor bugs in test code.
+
+  Bug fixes:
+  - Configure Solaris/Illumos to use MADV_FREE.
+  - Fix junk filling for mremap(2)-based huge reallocation.  This is only
+    relevant if configuring with the --enable-mremap option specified.
+  - Avoid compilation failure if 'restrict' C99 keyword is not supported by the
+    compiler.
+  - Add a configure test for SSE2 rather than assuming it is usable on i686
+    systems.  This fixes test compilation errors, especially on 32-bit Linux
+    systems.
+  - Fix mallctl argument size mismatches (size_t vs. uint64_t) in the stats unit
+    test.
+  - Fix/remove flawed alignment-related overflow tests.
+  - Prevent compiler optimizations that could change backtraces in the
+    prof_accum unit test.
+
+* 3.5.0 (January 22, 2014)
+
+  This version focuses on refactoring and automated testing, though it also
+  includes some non-trivial heap profiling optimizations not mentioned below.
+
+  New features:
+  - Add the *allocx() API, which is a successor to the experimental *allocm()
+    API.  The *allocx() functions are slightly simpler to use because they have
+    fewer parameters, they directly return the results of primary interest, and
+    mallocx()/rallocx() avoid the strict aliasing pitfall that
+    allocm()/rallocm() share with posix_memalign().  Note that *allocm() is
+    slated for removal in the next non-bugfix release.
+  - Add support for LinuxThreads.
+
+  Bug fixes:
+  - Unless heap profiling is enabled, disable floating point code and don't link
+    with libm.  This, in combination with e.g. EXTRA_CFLAGS=-mno-sse on x64
+    systems, makes it possible to completely disable floating point register
+    use.  Some versions of glibc neglect to save/restore caller-saved floating
+    point registers during dynamic lazy symbol loading, and the symbol loading
+    code uses whatever malloc the application happens to have linked/loaded
+    with, the result being potential floating point register corruption.
+  - Report ENOMEM rather than EINVAL if an OOM occurs during heap profiling
+    backtrace creation in imemalign().  This bug impacted posix_memalign() and
+    aligned_alloc().
+  - Fix a file descriptor leak in a prof_dump_maps() error path.
+  - Fix prof_dump() to close the dump file descriptor for all relevant error
+    paths.
+  - Fix rallocm() to use the arena specified by the ALLOCM_ARENA(s) flag for
+    allocation, not just deallocation.
+  - Fix a data race for large allocation stats counters.
+  - Fix a potential infinite loop during thread exit.  This bug occurred on
+    Solaris, and could affect other platforms with similar pthreads TSD
+    implementations.
+  - Don't junk-fill reallocations unless usable size changes.  This fixes a
+    violation of the *allocx()/*allocm() semantics.
+  - Fix growing large reallocation to junk fill new space.
+  - Fix huge deallocation to junk fill when munmap is disabled.
+  - Change the default private namespace prefix from empty to je_, and change
+    --with-private-namespace-prefix so that it prepends an additional prefix
+    rather than replacing je_.  This reduces the likelihood of applications
+    which statically link jemalloc experiencing symbol name collisions.
+  - Add missing private namespace mangling (relevant when
+    --with-private-namespace is specified).
+  - Add and use JEMALLOC_INLINE_C so that static inline functions are marked as
+    static even for debug builds.
+  - Add a missing mutex unlock in a malloc_init_hard() error path.  In practice
+    this error path is never executed.
+  - Fix numerous bugs in malloc_strotumax() error handling/reporting.  These
+    bugs had no impact except for malformed inputs.
+  - Fix numerous bugs in malloc_snprintf().  These bugs were not exercised by
+    existing calls, so they had no impact.
+
+* 3.4.1 (October 20, 2013)
+
+  Bug fixes:
+  - Fix a race in the "arenas.extend" mallctl that could cause memory corruption
+    of internal data structures and subsequent crashes.
+  - Fix Valgrind integration flaws that caused Valgrind warnings about reads of
+    uninitialized memory in:
+    + arena chunk headers
+    + internal zero-initialized data structures (relevant to tcache and prof
+      code)
+  - Preserve errno during the first allocation.  A readlink(2) call during
+    initialization fails unless /etc/malloc.conf exists, so errno was typically
+    set during the first allocation prior to this fix.
+  - Fix compilation warnings reported by gcc 4.8.1.
+
+* 3.4.0 (June 2, 2013)
+
+  This version is essentially a small bugfix release, but the addition of
+  aarch64 support requires that the minor version be incremented.
+
+  Bug fixes:
+  - Fix race-triggered deadlocks in chunk_record().  These deadlocks were
+    typically triggered by multiple threads concurrently deallocating huge
+    objects.
+
+  New features:
+  - Add support for the aarch64 architecture.
+
+* 3.3.1 (March 6, 2013)
+
+  This version fixes bugs that are typically encountered only when utilizing
+  custom run-time options.
+
+  Bug fixes:
+  - Fix a locking order bug that could cause deadlock during fork if heap
+    profiling were enabled.
+  - Fix a chunk recycling bug that could cause the allocator to lose track of
+    whether a chunk was zeroed.  On FreeBSD, NetBSD, and OS X, it could cause
+    corruption if allocating via sbrk(2) (unlikely unless running with the
+    "dss:primary" option specified).  This was completely harmless on Linux
+    unless using mlockall(2) (and unlikely even then, unless the
+    --disable-munmap configure option or the "dss:primary" option was
+    specified).  This regression was introduced in 3.1.0 by the
+    mlockall(2)/madvise(2) interaction fix.
+  - Fix TLS-related memory corruption that could occur during thread exit if the
+    thread never allocated memory.  Only the quarantine and prof facilities were
+    susceptible.
+  - Fix two quarantine bugs:
+    + Internal reallocation of the quarantined object array leaked the old
+      array.
+    + Reallocation failure for internal reallocation of the quarantined object
+      array (very unlikely) resulted in memory corruption.
+  - Fix Valgrind integration to annotate all internally allocated memory in a
+    way that keeps Valgrind happy about internal data structure access.
+  - Fix building for s390 systems.
+
+* 3.3.0 (January 23, 2013)
+
+  This version includes a few minor performance improvements in addition to the
+  listed new features and bug fixes.
+
+  New features:
+  - Add clipping support to lg_chunk option processing.
+  - Add the --enable-ivsalloc option.
+  - Add the --without-export option.
+  - Add the --disable-zone-allocator option.
+
+  Bug fixes:
+  - Fix "arenas.extend" mallctl to output the number of arenas.
+  - Fix chunk_recycle() to unconditionally inform Valgrind that returned memory
+    is undefined.
+  - Fix build break on FreeBSD related to alloca.h.
+
+* 3.2.0 (November 9, 2012)
+
+  In addition to a couple of bug fixes, this version modifies page run
+  allocation and dirty page purging algorithms in order to better control
+  page-level virtual memory fragmentation.
+
+  Incompatible changes:
+  - Change the "opt.lg_dirty_mult" default from 5 to 3 (32:1 to 8:1).
+
+  Bug fixes:
+  - Fix dss/mmap allocation precedence code to use recyclable mmap memory only
+    after primary dss allocation fails.
+  - Fix deadlock in the "arenas.purge" mallctl.  This regression was introduced
+    in 3.1.0 by the addition of the "arena.<i>.purge" mallctl.
+
+* 3.1.0 (October 16, 2012)
+
+  New features:
+  - Auto-detect whether running inside Valgrind, thus removing the need to
+    manually specify MALLOC_CONF=valgrind:true.
+  - Add the "arenas.extend" mallctl, which allows applications to create
+    manually managed arenas.
+  - Add the ALLOCM_ARENA() flag for {,r,d}allocm().
+  - Add the "opt.dss", "arena.<i>.dss", and "stats.arenas.<i>.dss" mallctls,
+    which provide control over dss/mmap precedence.
+  - Add the "arena.<i>.purge" mallctl, which obsoletes "arenas.purge".
+  - Define LG_QUANTUM for hppa.
+
+  Incompatible changes:
+  - Disable tcache by default if running inside Valgrind, in order to avoid
+    making unallocated objects appear reachable to Valgrind.
+  - Drop const from malloc_usable_size() argument on Linux.
+
+  Bug fixes:
+  - Fix heap profiling crash if sampled object is freed via realloc(p, 0).
+  - Remove const from __*_hook variable declarations, so that glibc can modify
+    them during process forking.
+  - Fix mlockall(2)/madvise(2) interaction.
+  - Fix fork(2)-related deadlocks.
+  - Fix error return value for "thread.tcache.enabled" mallctl.
+
+* 3.0.0 (May 11, 2012)
+
+  Although this version adds some major new features, the primary focus is on
+  internal code cleanup that facilitates maintainability and portability, most
+  of which is not reflected in the ChangeLog.  This is the first release to
+  incorporate substantial contributions from numerous other developers, and the
+  result is a more broadly useful allocator (see the git revision history for
+  contribution details).  Note that the license has been unified, thanks to
+  Facebook granting a license under the same terms as the other copyright
+  holders (see COPYING).
+
+  New features:
+  - Implement Valgrind support, redzones, and quarantine.
+  - Add support for additional platforms:
+    + FreeBSD
+    + Mac OS X Lion
+    + MinGW
+    + Windows (no support yet for replacing the system malloc)
+  - Add support for additional architectures:
+    + MIPS
+    + SH4
+    + Tilera
+  - Add support for cross compiling.
+  - Add nallocm(), which rounds a request size up to the nearest size class
+    without actually allocating.
+  - Implement aligned_alloc() (blame C11).
+  - Add the "thread.tcache.enabled" mallctl.
+  - Add the "opt.prof_final" mallctl.
+  - Update pprof (from gperftools 2.0).
+  - Add the --with-mangling option.
+  - Add the --disable-experimental option.
+  - Add the --disable-munmap option, and make it the default on Linux.
+  - Add the --enable-mremap option, which disables use of mremap(2) by default.
+
+  Incompatible changes:
+  - Enable stats by default.
+  - Enable fill by default.
+  - Disable lazy locking by default.
+  - Rename the "tcache.flush" mallctl to "thread.tcache.flush".
+  - Rename the "arenas.pagesize" mallctl to "arenas.page".
+  - Change the "opt.lg_prof_sample" default from 0 to 19 (1 B to 512 KiB).
+  - Change the "opt.prof_accum" default from true to false.
+
+  Removed features:
+  - Remove the swap feature, including the "config.swap", "swap.avail",
+    "swap.prezeroed", "swap.nfds", and "swap.fds" mallctls.
+  - Remove highruns statistics, including the
+    "stats.arenas.<i>.bins.<j>.highruns" and
+    "stats.arenas.<i>.lruns.<j>.highruns" mallctls.
+  - As part of small size class refactoring, remove the "opt.lg_[qc]space_max",
+    "arenas.cacheline", "arenas.subpage", "arenas.[tqcs]space_{min,max}", and
+    "arenas.[tqcs]bins" mallctls.
+  - Remove the "arenas.chunksize" mallctl.
+  - Remove the "opt.lg_prof_tcmax" option.
+  - Remove the "opt.lg_prof_bt_max" option.
+  - Remove the "opt.lg_tcache_gc_sweep" option.
+  - Remove the --disable-tiny option, including the "config.tiny" mallctl.
+  - Remove the --enable-dynamic-page-shift configure option.
+  - Remove the --enable-sysv configure option.
+
+  Bug fixes:
+  - Fix a statistics-related bug in the "thread.arena" mallctl that could cause
+    invalid statistics and crashes.
+  - Work around TLS deallocation via free() on Linux.  This bug could cause
+    write-after-free memory corruption.
+  - Fix a potential deadlock that could occur during interval- and
+    growth-triggered heap profile dumps.
+  - Fix large calloc() zeroing bugs due to dropping chunk map unzeroed flags.
+  - Fix chunk_alloc_dss() to stop claiming memory is zeroed.  This bug could
+    cause memory corruption and crashes with --enable-dss specified.
+  - Fix fork-related bugs that could cause deadlock in children between fork
+    and exec.
+  - Fix malloc_stats_print() to honor 'b' and 'l' in the opts parameter.
+  - Fix realloc(p, 0) to act like free(p).
+  - Do not enforce minimum alignment in memalign().
+  - Check for NULL pointer in malloc_usable_size().
+  - Fix an off-by-one heap profile statistics bug that could be observed in
+    interval- and growth-triggered heap profiles.
+  - Fix the "epoch" mallctl to update cached stats even if the passed in epoch
+    is 0.
+  - Fix bin->runcur management to fix a layout policy bug.  This bug did not
+    affect correctness.
+  - Fix a bug in choose_arena_hard() that potentially caused more arenas to be
+    initialized than necessary.
+  - Add missing "opt.lg_tcache_max" mallctl implementation.
+  - Use glibc allocator hooks to make mixed allocator usage less likely.
+  - Fix build issues for --disable-tcache.
+  - Don't mangle pthread_create() when --with-private-namespace is specified.
+
+* 2.2.5 (November 14, 2011)
+
+  Bug fixes:
+  - Fix huge_ralloc() race when using mremap(2).  This is a serious bug that
+    could cause memory corruption and/or crashes.
+  - Fix huge_ralloc() to maintain chunk statistics.
+  - Fix malloc_stats_print(..., "a") output.
+
+* 2.2.4 (November 5, 2011)
+
+  Bug fixes:
+  - Initialize arenas_tsd before using it.  This bug existed for 2.2.[0-3], as
+    well as for --disable-tls builds in earlier releases.
+  - Do not assume a 4 KiB page size in test/rallocm.c.
+
+* 2.2.3 (August 31, 2011)
+
+  This version fixes numerous bugs related to heap profiling.
+
+  Bug fixes:
+  - Fix a prof-related race condition.  This bug could cause memory corruption,
+    but only occurred in non-default configurations (prof_accum:false).
+  - Fix off-by-one backtracing issues (make sure that prof_alloc_prep() is
+    excluded from backtraces).
+  - Fix a prof-related bug in realloc() (only triggered by OOM errors).
+  - Fix prof-related bugs in allocm() and rallocm().
+  - Fix prof_tdata_cleanup() for --disable-tls builds.
+  - Fix a relative include path, to fix objdir builds.
+
+* 2.2.2 (July 30, 2011)
+
+  Bug fixes:
+  - Fix a build error for --disable-tcache.
+  - Fix assertions in arena_purge() (for real this time).
+  - Add the --with-private-namespace option.  This is a workaround for symbol
+    conflicts that can inadvertently arise when using static libraries.
+
+* 2.2.1 (March 30, 2011)
+
+  Bug fixes:
+  - Implement atomic operations for x86/x64.  This fixes compilation failures
+    for versions of gcc that are still in wide use.
+  - Fix an assertion in arena_purge().
+
+* 2.2.0 (March 22, 2011)
+
+  This version incorporates several improvements to algorithms and data
+  structures that tend to reduce fragmentation and increase speed.
+
+  New features:
+  - Add the "stats.cactive" mallctl.
+  - Update pprof (from google-perftools 1.7).
+  - Improve backtracing-related configuration logic, and add the
+    --disable-prof-libgcc option.
+
+  Bug fixes:
+  - Change default symbol visibility from "internal", to "hidden", which
+    decreases the overhead of library-internal function calls.
+  - Fix symbol visibility so that it is also set on OS X.
+  - Fix a build dependency regression caused by the introduction of the .pic.o
+    suffix for PIC object files.
+  - Add missing checks for mutex initialization failures.
+  - Don't use libgcc-based backtracing except on x64, where it is known to work.
+  - Fix deadlocks on OS X that were due to memory allocation in
+    pthread_mutex_lock().
+  - Heap profiling-specific fixes:
+    + Fix memory corruption due to integer overflow in small region index
+      computation, when using a small enough sample interval that profiling
+      context pointers are stored in small run headers.
+    + Fix a bootstrap ordering bug that only occurred with TLS disabled.
+    + Fix a rallocm() rsize bug.
+    + Fix error detection bugs for aligned memory allocation.
+
+* 2.1.3 (March 14, 2011)
+
+  Bug fixes:
+  - Fix a cpp logic regression (due to the "thread.{de,}allocatedp" mallctl fix
+    for OS X in 2.1.2).
+  - Fix a "thread.arena" mallctl bug.
+  - Fix a thread cache stats merging bug.
+
+* 2.1.2 (March 2, 2011)
+
+  Bug fixes:
+  - Fix "thread.{de,}allocatedp" mallctl for OS X.
+  - Add missing jemalloc.a to build system.
+
+* 2.1.1 (January 31, 2011)
+
+  Bug fixes:
+  - Fix aligned huge reallocation (affected allocm()).
+  - Fix the ALLOCM_LG_ALIGN macro definition.
+  - Fix a heap dumping deadlock.
+  - Fix a "thread.arena" mallctl bug.
+
+* 2.1.0 (December 3, 2010)
+
+  This version incorporates some optimizations that can't quite be considered
+  bug fixes.
+
+  New features:
+  - Use Linux's mremap(2) for huge object reallocation when possible.
+  - Avoid locking in mallctl*() when possible.
+  - Add the "thread.[de]allocatedp" mallctl's.
+  - Convert the manual page source from roff to DocBook, and generate both roff
+    and HTML manuals.
+
+  Bug fixes:
+  - Fix a crash due to incorrect bootstrap ordering.  This only impacted
+    --enable-debug --enable-dss configurations.
+  - Fix a minor statistics bug for mallctl("swap.avail", ...).
+
+* 2.0.1 (October 29, 2010)
+
+  Bug fixes:
+  - Fix a race condition in heap profiling that could cause undefined behavior
+    if "opt.prof_accum" were disabled.
+  - Add missing mutex unlocks for some OOM error paths in the heap profiling
+    code.
+  - Fix a compilation error for non-C99 builds.
+
+* 2.0.0 (October 24, 2010)
+
+  This version focuses on the experimental *allocm() API, and on improved
+  run-time configuration/introspection.  Nonetheless, numerous performance
+  improvements are also included.
+
+  New features:
+  - Implement the experimental {,r,s,d}allocm() API, which provides a superset
+    of the functionality available via malloc(), calloc(), posix_memalign(),
+    realloc(), malloc_usable_size(), and free().  These functions can be used to
+    allocate/reallocate aligned zeroed memory, ask for optional extra memory
+    during reallocation, prevent object movement during reallocation, etc.
+  - Replace JEMALLOC_OPTIONS/JEMALLOC_PROF_PREFIX with MALLOC_CONF, which is
+    more human-readable, and more flexible.  For example:
+      JEMALLOC_OPTIONS=AJP
+    is now:
+      MALLOC_CONF=abort:true,fill:true,stats_print:true
+  - Port to Apple OS X.  Sponsored by Mozilla.
+  - Make it possible for the application to control thread-->arena mappings via
+    the "thread.arena" mallctl.
+  - Add compile-time support for all TLS-related functionality via pthreads TSD.
+    This is mainly of interest for OS X, which does not support TLS, but has a
+    TSD implementation with similar performance.
+  - Override memalign() and valloc() if they are provided by the system.
+  - Add the "arenas.purge" mallctl, which can be used to synchronously purge all
+    dirty unused pages.
+  - Make cumulative heap profiling data optional, so that it is possible to
+    limit the amount of memory consumed by heap profiling data structures.
+  - Add per thread allocation counters that can be accessed via the
+    "thread.allocated" and "thread.deallocated" mallctls.
+
+  Incompatible changes:
+  - Remove JEMALLOC_OPTIONS and malloc_options (see MALLOC_CONF above).
+  - Increase default backtrace depth from 4 to 128 for heap profiling.
+  - Disable interval-based profile dumps by default.
+
+  Bug fixes:
+  - Remove bad assertions in fork handler functions.  These assertions could
+    cause aborts for some combinations of configure settings.
+  - Fix strerror_r() usage to deal with non-standard semantics in GNU libc.
+  - Fix leak context reporting.  This bug tended to cause the number of contexts
+    to be underreported (though the reported number of objects and bytes were
+    correct).
+  - Fix a realloc() bug for large in-place growing reallocation.  This bug could
+    cause memory corruption, but it was hard to trigger.
+  - Fix an allocation bug for small allocations that could be triggered if
+    multiple threads raced to create a new run of backing pages.
+  - Enhance the heap profiler to trigger samples based on usable size, rather
+    than request size.
+  - Fix a heap profiling bug due to sometimes losing track of requested object
+    size for sampled objects.
+
+* 1.0.3 (August 12, 2010)
+
+  Bug fixes:
+  - Fix the libunwind-based implementation of stack backtracing (used for heap
+    profiling).  This bug could cause zero-length backtraces to be reported.
+  - Add a missing mutex unlock in library initialization code.  If multiple
+    threads raced to initialize malloc, some of them could end up permanently
+    blocked.
+
+* 1.0.2 (May 11, 2010)
+
+  Bug fixes:
+  - Fix junk filling of large objects, which could cause memory corruption.
+  - Add MAP_NORESERVE support for chunk mapping, because otherwise virtual
+    memory limits could cause swap file configuration to fail.  Contributed by
+    Jordan DeLong.
+
+* 1.0.1 (April 14, 2010)
+
+  Bug fixes:
+  - Fix compilation when --enable-fill is specified.
+  - Fix threads-related profiling bugs that affected accuracy and caused memory
+    to be leaked during thread exit.
+  - Fix dirty page purging race conditions that could cause crashes.
+  - Fix crash in tcache flushing code during thread destruction.
+
+* 1.0.0 (April 11, 2010)
+
+  This release focuses on speed and run-time introspection.  Numerous
+  algorithmic improvements make this release substantially faster than its
+  predecessors.
+
+  New features:
+  - Implement autoconf-based configuration system.
+  - Add mallctl*(), for the purposes of introspection and run-time
+    configuration.
+  - Make it possible for the application to manually flush a thread's cache, via
+    the "tcache.flush" mallctl.
+  - Base maximum dirty page count on proportion of active memory.
+  - Compute various additional run-time statistics, including per size class
+    statistics for large objects.
+  - Expose malloc_stats_print(), which can be called repeatedly by the
+    application.
+  - Simplify the malloc_message() signature to only take one string argument,
+    and incorporate an opaque data pointer argument for use by the application
+    in combination with malloc_stats_print().
+  - Add support for allocation backed by one or more swap files, and allow the
+    application to disable over-commit if swap files are in use.
+  - Implement allocation profiling and leak checking.
+
+  Removed features:
+  - Remove the dynamic arena rebalancing code, since thread-specific caching
+    reduces its utility.
+
+  Bug fixes:
+  - Modify chunk allocation to work when address space layout randomization
+    (ASLR) is in use.
+  - Fix thread cleanup bugs related to TLS destruction.
+  - Handle 0-size allocation requests in posix_memalign().
+  - Fix a chunk leak.  The leaked chunks were never touched, so this impacted
+    virtual memory usage, but not physical memory usage.
+
+* linux_2008082[78]a (August 27/28, 2008)
+
+  These snapshot releases are the simple result of incorporating Linux-specific
+  support into the FreeBSD malloc sources.
+
+--------------------------------------------------------------------------------
+vim:filetype=text:textwidth=80
diff --git a/jemalloc/INSTALL.md b/jemalloc/INSTALL.md
new file mode 100644
index 0000000..b8f729b
--- /dev/null
+++ b/jemalloc/INSTALL.md
@@ -0,0 +1,421 @@
+Building and installing a packaged release of jemalloc can be as simple as
+typing the following while in the root directory of the source tree:
+
+    ./configure
+    make
+    make install
+
+If building from unpackaged developer sources, the simplest command sequence
+that might work is:
+
+    ./autogen.sh
+    make dist
+    make
+    make install
+
+Note that documentation is not built by the default target because doing so
+would create a dependency on xsltproc in packaged releases, hence the
+requirement to either run 'make dist' or avoid installing docs via the various
+install_* targets documented below.
+
+
+## Advanced configuration
+
+The 'configure' script supports numerous options that allow control of which
+functionality is enabled, where jemalloc is installed, etc.  Optionally, pass
+any of the following arguments (not a definitive list) to 'configure':
+
+* `--help`
+
+    Print a definitive list of options.
+
+* `--prefix=<install-root-dir>`
+
+    Set the base directory in which to install.  For example:
+
+        ./configure --prefix=/usr/local
+
+    will cause files to be installed into /usr/local/include, /usr/local/lib,
+    and /usr/local/man.
+
+* `--with-version=(<major>.<minor>.<bugfix>-<nrev>-g<gid>|VERSION)`
+
+    The VERSION file is mandatory for successful configuration, and the
+    following steps are taken to assure its presence:
+    1) If --with-version=<major>.<minor>.<bugfix>-<nrev>-g<gid> is specified,
+       generate VERSION using the specified value.
+    2) If --with-version is not specified in either form and the source
+       directory is inside a git repository, try to generate VERSION via 'git
+       describe' invocations that pattern-match release tags.
+    3) If VERSION is missing, generate it with a bogus version:
+       0.0.0-0-g0000000000000000000000000000000000000000
+
+    Note that --with-version=VERSION bypasses (1) and (2), which simplifies
+    VERSION configuration when embedding a jemalloc release into another
+    project's git repository.
+
+* `--with-rpath=<colon-separated-rpath>`
+
+    Embed one or more library paths, so that libjemalloc can find the libraries
+    it is linked to.  This works only on ELF-based systems.
+
+* `--with-mangling=<map>`
+
+    Mangle public symbols specified in <map> which is a comma-separated list of
+    name:mangled pairs.
+
+    For example, to use ld's --wrap option as an alternative method for
+    overriding libc's malloc implementation, specify something like:
+
+      --with-mangling=malloc:__wrap_malloc,free:__wrap_free[...]
+
+    Note that mangling happens prior to application of the prefix specified by
+    --with-jemalloc-prefix, and mangled symbols are then ignored when applying
+    the prefix.
+
+* `--with-jemalloc-prefix=<prefix>`
+
+    Prefix all public APIs with <prefix>.  For example, if <prefix> is
+    "prefix_", API changes like the following occur:
+
+      malloc()         --> prefix_malloc()
+      malloc_conf      --> prefix_malloc_conf
+      /etc/malloc.conf --> /etc/prefix_malloc.conf
+      MALLOC_CONF      --> PREFIX_MALLOC_CONF
+
+    This makes it possible to use jemalloc at the same time as the system
+    allocator, or even to use multiple copies of jemalloc simultaneously.
+
+    By default, the prefix is "", except on OS X, where it is "je_".  On OS X,
+    jemalloc overlays the default malloc zone, but makes no attempt to actually
+    replace the "malloc", "calloc", etc. symbols.
+
+* `--without-export`
+
+    Don't export public APIs.  This can be useful when building jemalloc as a
+    static library, or to avoid exporting public APIs when using the zone
+    allocator on OSX.
+
+* `--with-private-namespace=<prefix>`
+
+    Prefix all library-private APIs with <prefix>je_.  For shared libraries,
+    symbol visibility mechanisms prevent these symbols from being exported, but
+    for static libraries, naming collisions are a real possibility.  By
+    default, <prefix> is empty, which results in a symbol prefix of je_ .
+
+* `--with-install-suffix=<suffix>`
+
+    Append <suffix> to the base name of all installed files, such that multiple
+    versions of jemalloc can coexist in the same installation directory.  For
+    example, libjemalloc.so.0 becomes libjemalloc<suffix>.so.0.
+
+* `--with-malloc-conf=<malloc_conf>`
+
+    Embed `<malloc_conf>` as a run-time options string that is processed prior to
+    the malloc_conf global variable, the /etc/malloc.conf symlink, and the
+    MALLOC_CONF environment variable.  For example, to change the default decay
+    time to 30 seconds:
+
+      --with-malloc-conf=decay_ms:30000
+
+* `--enable-debug`
+
+    Enable assertions and validation code.  This incurs a substantial
+    performance hit, but is very useful during application development.
+
+* `--disable-stats`
+
+    Disable statistics gathering functionality.  See the "opt.stats_print"
+    option documentation for usage details.
+
+* `--enable-prof`
+
+    Enable heap profiling and leak detection functionality.  See the "opt.prof"
+    option documentation for usage details.  When enabled, there are several
+    approaches to backtracing, and the configure script chooses the first one
+    in the following list that appears to function correctly:
+
+    + libunwind      (requires --enable-prof-libunwind)
+    + libgcc         (unless --disable-prof-libgcc)
+    + gcc intrinsics (unless --disable-prof-gcc)
+
+* `--enable-prof-libunwind`
+
+    Use the libunwind library (http://www.nongnu.org/libunwind/) for stack
+    backtracing.
+
+* `--disable-prof-libgcc`
+
+    Disable the use of libgcc's backtracing functionality.
+
+* `--disable-prof-gcc`
+
+    Disable the use of gcc intrinsics for backtracing.
+
+* `--with-static-libunwind=<libunwind.a>`
+
+    Statically link against the specified libunwind.a rather than dynamically
+    linking with -lunwind.
+
+* `--disable-fill`
+
+    Disable support for junk/zero filling of memory.  See the "opt.junk" and
+    "opt.zero" option documentation for usage details.
+
+* `--disable-zone-allocator`
+
+    Disable zone allocator for Darwin.  This means jemalloc won't be hooked as
+    the default allocator on OSX/iOS.
+
+* `--enable-utrace`
+
+    Enable utrace(2)-based allocation tracing.  This feature is not broadly
+    portable (FreeBSD has it, but Linux and OS X do not).
+
+* `--enable-xmalloc`
+
+    Enable support for optional immediate termination due to out-of-memory
+    errors, as is commonly implemented by "xmalloc" wrapper function for malloc.
+    See the "opt.xmalloc" option documentation for usage details.
+
+* `--enable-lazy-lock`
+
+    Enable code that wraps pthread_create() to detect when an application
+    switches from single-threaded to multi-threaded mode, so that it can avoid
+    mutex locking/unlocking operations while in single-threaded mode.  In
+    practice, this feature usually has little impact on performance unless
+    thread-specific caching is disabled.
+
+* `--disable-cache-oblivious`
+
+    Disable cache-oblivious large allocation alignment for large allocation
+    requests with no alignment constraints.  If this feature is disabled, all
+    large allocations are page-aligned as an implementation artifact, which can
+    severely harm CPU cache utilization.  However, the cache-oblivious layout
+    comes at the cost of one extra page per large allocation, which in the
+    most extreme case increases physical memory usage for the 16 KiB size class
+    to 20 KiB.
+
+* `--disable-syscall`
+
+    Disable use of syscall(2) rather than {open,read,write,close}(2).  This is
+    intended as a workaround for systems that place security limitations on
+    syscall(2).
+
+* `--disable-cxx`
+
+    Disable C++ integration.  This will cause new and delete operator
+    implementations to be omitted.
+
+* `--with-xslroot=<path>`
+
+    Specify where to find DocBook XSL stylesheets when building the
+    documentation.
+
+* `--with-lg-page=<lg-page>`
+
+    Specify the base 2 log of the allocator page size, which must in turn be at
+    least as large as the system page size.  By default the configure script
+    determines the host's page size and sets the allocator page size equal to
+    the system page size, so this option need not be specified unless the
+    system page size may change between configuration and execution, e.g. when
+    cross compiling.
+
+* `--with-lg-hugepage=<lg-hugepage>`
+
+    Specify the base 2 log of the system huge page size.  This option is useful
+    when cross compiling, or when overriding the default for systems that do
+    not explicitly support huge pages.
+
+* `--with-lg-quantum=<lg-quantum>`
+
+    Specify the base 2 log of the minimum allocation alignment.  jemalloc needs
+    to know the minimum alignment that meets the following C standard
+    requirement (quoted from the April 12, 2011 draft of the C11 standard):
+
+    >  The pointer returned if the allocation succeeds is suitably aligned so
+      that it may be assigned to a pointer to any type of object with a
+      fundamental alignment requirement and then used to access such an object
+      or an array of such objects in the space allocated [...]
+
+    This setting is architecture-specific, and although jemalloc includes known
+    safe values for the most commonly used modern architectures, there is a
+    wrinkle related to GNU libc (glibc) that may impact your choice of
+    <lg-quantum>.  On most modern architectures, this mandates 16-byte
+    alignment (<lg-quantum>=4), but the glibc developers chose not to meet this
+    requirement for performance reasons.  An old discussion can be found at
+    <https://sourceware.org/bugzilla/show_bug.cgi?id=206> .  Unlike glibc,
+    jemalloc does follow the C standard by default (caveat: jemalloc
+    technically cheats for size classes smaller than the quantum), but the fact
+    that Linux systems already work around this allocator noncompliance means
+    that it is generally safe in practice to let jemalloc's minimum alignment
+    follow glibc's lead.  If you specify `--with-lg-quantum=3` during
+    configuration, jemalloc will provide additional size classes that are not
+    16-byte-aligned (24, 40, and 56).
+
+* `--with-lg-vaddr=<lg-vaddr>`
+
+    Specify the number of significant virtual address bits.  By default, the
+    configure script attempts to detect virtual address size on those platforms
+    where it knows how, and picks a default otherwise.  This option may be
+    useful when cross-compiling.
+
+* `--disable-initial-exec-tls`
+
+    Disable the initial-exec TLS model for jemalloc's internal thread-local
+    storage (on those platforms that support explicit settings).  This can allow
+    jemalloc to be dynamically loaded after program startup (e.g. using dlopen).
+    Note that in this case, there will be two malloc implementations operating
+    in the same process, which will almost certainly result in confusing runtime
+    crashes if pointers leak from one implementation to the other.
+
+* `--disable-libdl`
+
+    Disable the usage of libdl, namely dlsym(3) which is required by the lazy
+    lock option.  This can allow building static binaries.
+
+The following environment variables (not a definitive list) impact configure's
+behavior:
+
+* `CFLAGS="?"`
+* `CXXFLAGS="?"`
+
+    Pass these flags to the C/C++ compiler.  Any flags set by the configure
+    script are prepended, which means explicitly set flags generally take
+    precedence.  Take care when specifying flags such as -Werror, because
+    configure tests may be affected in undesirable ways.
+
+* `EXTRA_CFLAGS="?"`
+* `EXTRA_CXXFLAGS="?"`
+
+    Append these flags to CFLAGS/CXXFLAGS, without passing them to the
+    compiler(s) during configuration.  This makes it possible to add flags such
+    as -Werror, while allowing the configure script to determine what other
+    flags are appropriate for the specified configuration.
+
+* `CPPFLAGS="?"`
+
+    Pass these flags to the C preprocessor.  Note that CFLAGS is not passed to
+    'cpp' when 'configure' is looking for include files, so you must use
+    CPPFLAGS instead if you need to help 'configure' find header files.
+
+* `LD_LIBRARY_PATH="?"`
+
+    'ld' uses this colon-separated list to find libraries.
+
+* `LDFLAGS="?"`
+
+    Pass these flags when linking.
+
+* `PATH="?"`
+
+    'configure' uses this to find programs.
+
+In some cases it may be necessary to work around configuration results that do
+not match reality.  For example, Linux 4.5 added support for the MADV_FREE flag
+to madvise(2), which can cause problems if building on a host with MADV_FREE
+support and deploying to a target without.  To work around this, use a cache
+file to override the relevant configuration variable defined in configure.ac,
+e.g.:
+
+    echo "je_cv_madv_free=no" > config.cache && ./configure -C
+
+
+## Advanced compilation
+
+To build only parts of jemalloc, use the following targets:
+
+    build_lib_shared
+    build_lib_static
+    build_lib
+    build_doc_html
+    build_doc_man
+    build_doc
+
+To install only parts of jemalloc, use the following targets:
+
+    install_bin
+    install_include
+    install_lib_shared
+    install_lib_static
+    install_lib_pc
+    install_lib
+    install_doc_html
+    install_doc_man
+    install_doc
+
+To clean up build results to varying degrees, use the following make targets:
+
+    clean
+    distclean
+    relclean
+
+
+## Advanced installation
+
+Optionally, define make variables when invoking make, including (not
+exclusively):
+
+* `INCLUDEDIR="?"`
+
+    Use this as the installation prefix for header files.
+
+* `LIBDIR="?"`
+
+    Use this as the installation prefix for libraries.
+
+* `MANDIR="?"`
+
+    Use this as the installation prefix for man pages.
+
+* `DESTDIR="?"`
+
+    Prepend DESTDIR to INCLUDEDIR, LIBDIR, DATADIR, and MANDIR.  This is useful
+    when installing to a different path than was specified via --prefix.
+
+* `CC="?"`
+
+    Use this to invoke the C compiler.
+
+* `CFLAGS="?"`
+
+    Pass these flags to the compiler.
+
+* `CPPFLAGS="?"`
+
+    Pass these flags to the C preprocessor.
+
+* `LDFLAGS="?"`
+
+    Pass these flags when linking.
+
+* `PATH="?"`
+
+    Use this to search for programs used during configuration and building.
+
+
+## Development
+
+If you intend to make non-trivial changes to jemalloc, use the 'autogen.sh'
+script rather than 'configure'.  This re-generates 'configure', enables
+configuration dependency rules, and enables re-generation of automatically
+generated source files.
+
+The build system supports using an object directory separate from the source
+tree.  For example, you can create an 'obj' directory, and from within that
+directory, issue configuration and build commands:
+
+    autoconf
+    mkdir obj
+    cd obj
+    ../configure --enable-autogen
+    make
+
+
+## Documentation
+
+The manual page is generated in both html and roff formats.  Any web browser
+can be used to view the html manual.  The roff manual page can be formatted
+prior to installation via the following command:
+
+    nroff -man -t doc/jemalloc.3
diff --git a/jemalloc/Makefile.in b/jemalloc/Makefile.in
new file mode 100644
index 0000000..7128b00
--- /dev/null
+++ b/jemalloc/Makefile.in
@@ -0,0 +1,623 @@
+# Clear out all vpaths, then set just one (default vpath) for the main build
+# directory.
+vpath
+vpath % .
+
+# Clear the default suffixes, so that built-in rules are not used.
+.SUFFIXES :
+
+SHELL := /bin/sh
+
+CC := @CC@
+CXX := @CXX@
+
+# Configuration parameters.
+DESTDIR =
+BINDIR := $(DESTDIR)@BINDIR@
+INCLUDEDIR := $(DESTDIR)@INCLUDEDIR@
+LIBDIR := $(DESTDIR)@LIBDIR@
+DATADIR := $(DESTDIR)@DATADIR@
+MANDIR := $(DESTDIR)@MANDIR@
+srcroot := @srcroot@
+objroot := @objroot@
+abs_srcroot := @abs_srcroot@
+abs_objroot := @abs_objroot@
+
+# Build parameters.
+CPPFLAGS := @CPPFLAGS@ -I$(objroot)include -I$(srcroot)include
+CONFIGURE_CFLAGS := @CONFIGURE_CFLAGS@
+SPECIFIED_CFLAGS := @SPECIFIED_CFLAGS@
+EXTRA_CFLAGS := @EXTRA_CFLAGS@
+CFLAGS := $(strip $(CONFIGURE_CFLAGS) $(SPECIFIED_CFLAGS) $(EXTRA_CFLAGS))
+CONFIGURE_CXXFLAGS := @CONFIGURE_CXXFLAGS@
+SPECIFIED_CXXFLAGS := @SPECIFIED_CXXFLAGS@
+EXTRA_CXXFLAGS := @EXTRA_CXXFLAGS@
+CXXFLAGS := $(strip $(CONFIGURE_CXXFLAGS) $(SPECIFIED_CXXFLAGS) $(EXTRA_CXXFLAGS))
+LDFLAGS := @LDFLAGS@
+EXTRA_LDFLAGS := @EXTRA_LDFLAGS@
+LIBS := @LIBS@
+RPATH_EXTRA := @RPATH_EXTRA@
+SO := @so@
+IMPORTLIB := @importlib@
+O := @o@
+A := @a@
+EXE := @exe@
+LIBPREFIX := @libprefix@
+REV := @rev@
+install_suffix := @install_suffix@
+ABI := @abi@
+XSLTPROC := @XSLTPROC@
+XSLROOT := @XSLROOT@
+AUTOCONF := @AUTOCONF@
+_RPATH = @RPATH@
+RPATH = $(if $(1),$(call _RPATH,$(1)))
+cfghdrs_in := $(addprefix $(srcroot),@cfghdrs_in@)
+cfghdrs_out := @cfghdrs_out@
+cfgoutputs_in := $(addprefix $(srcroot),@cfgoutputs_in@)
+cfgoutputs_out := @cfgoutputs_out@
+enable_autogen := @enable_autogen@
+enable_doc := @enable_doc@
+enable_shared := @enable_shared@
+enable_static := @enable_static@
+enable_prof := @enable_prof@
+enable_zone_allocator := @enable_zone_allocator@
+enable_experimental_smallocx := @enable_experimental_smallocx@
+MALLOC_CONF := @JEMALLOC_CPREFIX@MALLOC_CONF
+link_whole_archive := @link_whole_archive@
+DSO_LDFLAGS = @DSO_LDFLAGS@
+SOREV = @SOREV@
+PIC_CFLAGS = @PIC_CFLAGS@
+CTARGET = @CTARGET@
+LDTARGET = @LDTARGET@
+TEST_LD_MODE = @TEST_LD_MODE@
+MKLIB = @MKLIB@
+AR = @AR@
+ARFLAGS = @ARFLAGS@
+DUMP_SYMS = @DUMP_SYMS@
+AWK := @AWK@
+CC_MM = @CC_MM@
+LM := @LM@
+INSTALL = @INSTALL@
+
+ifeq (macho, $(ABI))
+TEST_LIBRARY_PATH := DYLD_FALLBACK_LIBRARY_PATH="$(objroot)lib"
+else
+ifeq (pecoff, $(ABI))
+TEST_LIBRARY_PATH := PATH="$(PATH):$(objroot)lib"
+else
+TEST_LIBRARY_PATH :=
+endif
+endif
+
+LIBJEMALLOC := $(LIBPREFIX)jemalloc$(install_suffix)
+
+# Lists of files.
+BINS := $(objroot)bin/jemalloc-config $(objroot)bin/jemalloc.sh $(objroot)bin/jeprof
+C_HDRS := $(objroot)include/jemalloc/jemalloc$(install_suffix).h
+C_SRCS := $(srcroot)src/jemalloc.c \
+	$(srcroot)src/arena.c \
+	$(srcroot)src/background_thread.c \
+	$(srcroot)src/base.c \
+	$(srcroot)src/bin.c \
+	$(srcroot)src/bitmap.c \
+	$(srcroot)src/ckh.c \
+	$(srcroot)src/ctl.c \
+	$(srcroot)src/div.c \
+	$(srcroot)src/extent.c \
+	$(srcroot)src/extent_dss.c \
+	$(srcroot)src/extent_mmap.c \
+	$(srcroot)src/hash.c \
+	$(srcroot)src/hook.c \
+	$(srcroot)src/large.c \
+	$(srcroot)src/log.c \
+	$(srcroot)src/malloc_io.c \
+	$(srcroot)src/mutex.c \
+	$(srcroot)src/mutex_pool.c \
+	$(srcroot)src/nstime.c \
+	$(srcroot)src/pages.c \
+	$(srcroot)src/prng.c \
+	$(srcroot)src/prof.c \
+	$(srcroot)src/rtree.c \
+	$(srcroot)src/safety_check.c \
+	$(srcroot)src/stats.c \
+	$(srcroot)src/sc.c \
+	$(srcroot)src/sz.c \
+	$(srcroot)src/tcache.c \
+	$(srcroot)src/test_hooks.c \
+	$(srcroot)src/ticker.c \
+	$(srcroot)src/tsd.c \
+	$(srcroot)src/witness.c
+ifeq ($(enable_zone_allocator), 1)
+C_SRCS += $(srcroot)src/zone.c
+endif
+ifeq ($(IMPORTLIB),$(SO))
+STATIC_LIBS := $(objroot)lib/$(LIBJEMALLOC).$(A)
+endif
+ifdef PIC_CFLAGS
+STATIC_LIBS += $(objroot)lib/$(LIBJEMALLOC)_pic.$(A)
+else
+STATIC_LIBS += $(objroot)lib/$(LIBJEMALLOC)_s.$(A)
+endif
+DSOS := $(objroot)lib/$(LIBJEMALLOC).$(SOREV)
+ifneq ($(SOREV),$(SO))
+DSOS += $(objroot)lib/$(LIBJEMALLOC).$(SO)
+endif
+ifeq (1, $(link_whole_archive))
+LJEMALLOC := -Wl,--whole-archive -L$(objroot)lib -l$(LIBJEMALLOC) -Wl,--no-whole-archive
+else
+LJEMALLOC := $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+endif
+PC := $(objroot)jemalloc.pc
+MAN3 := $(objroot)doc/jemalloc$(install_suffix).3
+DOCS_XML := $(objroot)doc/jemalloc$(install_suffix).xml
+DOCS_HTML := $(DOCS_XML:$(objroot)%.xml=$(objroot)%.html)
+DOCS_MAN3 := $(DOCS_XML:$(objroot)%.xml=$(objroot)%.3)
+DOCS := $(DOCS_HTML) $(DOCS_MAN3)
+C_TESTLIB_SRCS := $(srcroot)test/src/btalloc.c $(srcroot)test/src/btalloc_0.c \
+	$(srcroot)test/src/btalloc_1.c $(srcroot)test/src/math.c \
+	$(srcroot)test/src/mtx.c $(srcroot)test/src/mq.c \
+	$(srcroot)test/src/SFMT.c $(srcroot)test/src/test.c \
+	$(srcroot)test/src/thd.c $(srcroot)test/src/timer.c
+ifeq (1, $(link_whole_archive))
+C_UTIL_INTEGRATION_SRCS :=
+C_UTIL_CPP_SRCS :=
+else
+C_UTIL_INTEGRATION_SRCS := $(srcroot)src/nstime.c $(srcroot)src/malloc_io.c
+C_UTIL_CPP_SRCS := $(srcroot)src/nstime.c $(srcroot)src/malloc_io.c
+endif
+TESTS_UNIT := \
+	$(srcroot)test/unit/a0.c \
+	$(srcroot)test/unit/arena_reset.c \
+	$(srcroot)test/unit/atomic.c \
+	$(srcroot)test/unit/background_thread.c \
+	$(srcroot)test/unit/background_thread_enable.c \
+	$(srcroot)test/unit/base.c \
+	$(srcroot)test/unit/bitmap.c \
+	$(srcroot)test/unit/bit_util.c \
+	$(srcroot)test/unit/binshard.c \
+	$(srcroot)test/unit/ckh.c \
+	$(srcroot)test/unit/decay.c \
+	$(srcroot)test/unit/div.c \
+	$(srcroot)test/unit/emitter.c \
+	$(srcroot)test/unit/extent_quantize.c \
+	$(srcroot)test/unit/extent_util.c \
+	$(srcroot)test/unit/fork.c \
+	$(srcroot)test/unit/hash.c \
+	$(srcroot)test/unit/hook.c \
+	$(srcroot)test/unit/huge.c \
+	$(srcroot)test/unit/junk.c \
+	$(srcroot)test/unit/junk_alloc.c \
+	$(srcroot)test/unit/junk_free.c \
+	$(srcroot)test/unit/log.c \
+	$(srcroot)test/unit/mallctl.c \
+	$(srcroot)test/unit/malloc_io.c \
+	$(srcroot)test/unit/math.c \
+	$(srcroot)test/unit/mq.c \
+	$(srcroot)test/unit/mtx.c \
+	$(srcroot)test/unit/pack.c \
+	$(srcroot)test/unit/pages.c \
+	$(srcroot)test/unit/ph.c \
+	$(srcroot)test/unit/prng.c \
+	$(srcroot)test/unit/prof_accum.c \
+	$(srcroot)test/unit/prof_active.c \
+	$(srcroot)test/unit/prof_gdump.c \
+	$(srcroot)test/unit/prof_idump.c \
+	$(srcroot)test/unit/prof_log.c \
+	$(srcroot)test/unit/prof_reset.c \
+	$(srcroot)test/unit/prof_tctx.c \
+	$(srcroot)test/unit/prof_thread_name.c \
+	$(srcroot)test/unit/ql.c \
+	$(srcroot)test/unit/qr.c \
+	$(srcroot)test/unit/rb.c \
+	$(srcroot)test/unit/retained.c \
+	$(srcroot)test/unit/rtree.c \
+	$(srcroot)test/unit/safety_check.c \
+	$(srcroot)test/unit/seq.c \
+	$(srcroot)test/unit/SFMT.c \
+	$(srcroot)test/unit/sc.c \
+	$(srcroot)test/unit/size_classes.c \
+	$(srcroot)test/unit/slab.c \
+	$(srcroot)test/unit/smoothstep.c \
+	$(srcroot)test/unit/spin.c \
+	$(srcroot)test/unit/stats.c \
+	$(srcroot)test/unit/stats_print.c \
+	$(srcroot)test/unit/test_hooks.c \
+	$(srcroot)test/unit/ticker.c \
+	$(srcroot)test/unit/nstime.c \
+	$(srcroot)test/unit/tsd.c \
+	$(srcroot)test/unit/witness.c \
+	$(srcroot)test/unit/zero.c
+ifeq (@enable_prof@, 1)
+TESTS_UNIT += \
+	$(srcroot)test/unit/arena_reset_prof.c
+endif
+TESTS_INTEGRATION := $(srcroot)test/integration/aligned_alloc.c \
+	$(srcroot)test/integration/allocated.c \
+	$(srcroot)test/integration/extent.c \
+	$(srcroot)test/integration/malloc.c \
+	$(srcroot)test/integration/mallocx.c \
+	$(srcroot)test/integration/MALLOCX_ARENA.c \
+	$(srcroot)test/integration/overflow.c \
+	$(srcroot)test/integration/posix_memalign.c \
+	$(srcroot)test/integration/rallocx.c \
+	$(srcroot)test/integration/sdallocx.c \
+	$(srcroot)test/integration/slab_sizes.c \
+	$(srcroot)test/integration/thread_arena.c \
+	$(srcroot)test/integration/thread_tcache_enabled.c \
+	$(srcroot)test/integration/xallocx.c
+ifeq (@enable_experimental_smallocx@, 1)
+TESTS_INTEGRATION += \
+  $(srcroot)test/integration/smallocx.c
+endif
+ifeq (@enable_cxx@, 1)
+CPP_SRCS := $(srcroot)src/jemalloc_cpp.cpp
+TESTS_INTEGRATION_CPP := $(srcroot)test/integration/cpp/basic.cpp
+else
+CPP_SRCS :=
+TESTS_INTEGRATION_CPP :=
+endif
+TESTS_STRESS := $(srcroot)test/stress/microbench.c \
+	$(srcroot)test/stress/hookbench.c
+
+
+TESTS := $(TESTS_UNIT) $(TESTS_INTEGRATION) $(TESTS_INTEGRATION_CPP) $(TESTS_STRESS)
+
+PRIVATE_NAMESPACE_HDRS := $(objroot)include/jemalloc/internal/private_namespace.h $(objroot)include/jemalloc/internal/private_namespace_jet.h
+PRIVATE_NAMESPACE_GEN_HDRS := $(PRIVATE_NAMESPACE_HDRS:%.h=%.gen.h)
+C_SYM_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.sym.$(O))
+C_SYMS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.sym)
+C_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.$(O))
+CPP_OBJS := $(CPP_SRCS:$(srcroot)%.cpp=$(objroot)%.$(O))
+C_PIC_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.pic.$(O))
+CPP_PIC_OBJS := $(CPP_SRCS:$(srcroot)%.cpp=$(objroot)%.pic.$(O))
+C_JET_SYM_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.sym.$(O))
+C_JET_SYMS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.sym)
+C_JET_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.$(O))
+C_TESTLIB_UNIT_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.unit.$(O))
+C_TESTLIB_INTEGRATION_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.integration.$(O))
+C_UTIL_INTEGRATION_OBJS := $(C_UTIL_INTEGRATION_SRCS:$(srcroot)%.c=$(objroot)%.integration.$(O))
+C_TESTLIB_STRESS_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.stress.$(O))
+C_TESTLIB_OBJS := $(C_TESTLIB_UNIT_OBJS) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(C_TESTLIB_STRESS_OBJS)
+
+TESTS_UNIT_OBJS := $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_INTEGRATION_OBJS := $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_INTEGRATION_CPP_OBJS := $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%.$(O))
+TESTS_STRESS_OBJS := $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_OBJS := $(TESTS_UNIT_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_STRESS_OBJS)
+TESTS_CPP_OBJS := $(TESTS_INTEGRATION_CPP_OBJS)
+
+.PHONY: all dist build_doc_html build_doc_man build_doc
+.PHONY: install_bin install_include install_lib
+.PHONY: install_doc_html install_doc_man install_doc install
+.PHONY: tests check clean distclean relclean
+
+.SECONDARY : $(PRIVATE_NAMESPACE_GEN_HDRS) $(TESTS_OBJS) $(TESTS_CPP_OBJS)
+
+# Default target.
+all: build_lib
+
+dist: build_doc
+
+$(objroot)doc/%.html : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/html.xsl
+ifneq ($(XSLROOT),)
+	$(XSLTPROC) -o $@ $(objroot)doc/html.xsl $<
+else
+ifeq ($(wildcard $(DOCS_HTML)),)
+	@echo "<p>Missing xsltproc.  Doc not built.</p>" > $@
+endif
+	@echo "Missing xsltproc.  "$@" not (re)built."
+endif
+
+$(objroot)doc/%.3 : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/manpages.xsl
+ifneq ($(XSLROOT),)
+	$(XSLTPROC) -o $@ $(objroot)doc/manpages.xsl $<
+else
+ifeq ($(wildcard $(DOCS_MAN3)),)
+	@echo "Missing xsltproc.  Doc not built." > $@
+endif
+	@echo "Missing xsltproc.  "$@" not (re)built."
+endif
+
+build_doc_html: $(DOCS_HTML)
+build_doc_man: $(DOCS_MAN3)
+build_doc: $(DOCS)
+
+#
+# Include generated dependency files.
+#
+ifdef CC_MM
+-include $(C_SYM_OBJS:%.$(O)=%.d)
+-include $(C_OBJS:%.$(O)=%.d)
+-include $(CPP_OBJS:%.$(O)=%.d)
+-include $(C_PIC_OBJS:%.$(O)=%.d)
+-include $(CPP_PIC_OBJS:%.$(O)=%.d)
+-include $(C_JET_SYM_OBJS:%.$(O)=%.d)
+-include $(C_JET_OBJS:%.$(O)=%.d)
+-include $(C_TESTLIB_OBJS:%.$(O)=%.d)
+-include $(TESTS_OBJS:%.$(O)=%.d)
+-include $(TESTS_CPP_OBJS:%.$(O)=%.d)
+endif
+
+$(C_SYM_OBJS): $(objroot)src/%.sym.$(O): $(srcroot)src/%.c
+$(C_SYM_OBJS): CPPFLAGS += -DJEMALLOC_NO_PRIVATE_NAMESPACE
+$(C_SYMS): $(objroot)src/%.sym: $(objroot)src/%.sym.$(O)
+$(C_OBJS): $(objroot)src/%.$(O): $(srcroot)src/%.c
+$(CPP_OBJS): $(objroot)src/%.$(O): $(srcroot)src/%.cpp
+$(C_PIC_OBJS): $(objroot)src/%.pic.$(O): $(srcroot)src/%.c
+$(C_PIC_OBJS): CFLAGS += $(PIC_CFLAGS)
+$(CPP_PIC_OBJS): $(objroot)src/%.pic.$(O): $(srcroot)src/%.cpp
+$(CPP_PIC_OBJS): CXXFLAGS += $(PIC_CFLAGS)
+$(C_JET_SYM_OBJS): $(objroot)src/%.jet.sym.$(O): $(srcroot)src/%.c
+$(C_JET_SYM_OBJS): CPPFLAGS += -DJEMALLOC_JET -DJEMALLOC_NO_PRIVATE_NAMESPACE
+$(C_JET_SYMS): $(objroot)src/%.jet.sym: $(objroot)src/%.jet.sym.$(O)
+$(C_JET_OBJS): $(objroot)src/%.jet.$(O): $(srcroot)src/%.c
+$(C_JET_OBJS): CPPFLAGS += -DJEMALLOC_JET
+$(C_TESTLIB_UNIT_OBJS): $(objroot)test/src/%.unit.$(O): $(srcroot)test/src/%.c
+$(C_TESTLIB_UNIT_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
+$(C_TESTLIB_INTEGRATION_OBJS): $(objroot)test/src/%.integration.$(O): $(srcroot)test/src/%.c
+$(C_TESTLIB_INTEGRATION_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_TEST
+$(C_UTIL_INTEGRATION_OBJS): $(objroot)src/%.integration.$(O): $(srcroot)src/%.c
+$(C_TESTLIB_STRESS_OBJS): $(objroot)test/src/%.stress.$(O): $(srcroot)test/src/%.c
+$(C_TESTLIB_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST -DJEMALLOC_STRESS_TESTLIB
+$(C_TESTLIB_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
+$(TESTS_UNIT_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
+$(TESTS_INTEGRATION_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_TEST
+$(TESTS_INTEGRATION_CPP_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_CPP_TEST
+$(TESTS_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST
+$(TESTS_OBJS): $(objroot)test/%.$(O): $(srcroot)test/%.c
+$(TESTS_CPP_OBJS): $(objroot)test/%.$(O): $(srcroot)test/%.cpp
+$(TESTS_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
+$(TESTS_CPP_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
+ifneq ($(IMPORTLIB),$(SO))
+$(CPP_OBJS) $(C_SYM_OBJS) $(C_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS): CPPFLAGS += -DDLLEXPORT
+endif
+
+# Dependencies.
+ifndef CC_MM
+HEADER_DIRS = $(srcroot)include/jemalloc/internal \
+	$(objroot)include/jemalloc $(objroot)include/jemalloc/internal
+HEADERS = $(filter-out $(PRIVATE_NAMESPACE_HDRS),$(wildcard $(foreach dir,$(HEADER_DIRS),$(dir)/*.h)))
+$(C_SYM_OBJS) $(C_OBJS) $(CPP_OBJS) $(C_PIC_OBJS) $(CPP_PIC_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS) $(TESTS_CPP_OBJS): $(HEADERS)
+$(TESTS_OBJS) $(TESTS_CPP_OBJS): $(objroot)test/include/test/jemalloc_test.h
+endif
+
+$(C_OBJS) $(CPP_OBJS) $(C_PIC_OBJS) $(CPP_PIC_OBJS) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_INTEGRATION_CPP_OBJS): $(objroot)include/jemalloc/internal/private_namespace.h
+$(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS) $(C_TESTLIB_STRESS_OBJS) $(TESTS_UNIT_OBJS) $(TESTS_STRESS_OBJS): $(objroot)include/jemalloc/internal/private_namespace_jet.h
+
+$(C_SYM_OBJS) $(C_OBJS) $(C_PIC_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS): %.$(O):
+	@mkdir -p $(@D)
+	$(CC) $(CFLAGS) -c $(CPPFLAGS) $(CTARGET) $<
+ifdef CC_MM
+	@$(CC) -MM $(CPPFLAGS) -MT $@ -o $(@:%.$(O)=%.d) $<
+endif
+
+$(C_SYMS): %.sym:
+	@mkdir -p $(@D)
+	$(DUMP_SYMS) $< | $(AWK) -f $(objroot)include/jemalloc/internal/private_symbols.awk > $@
+
+$(C_JET_SYMS): %.sym:
+	@mkdir -p $(@D)
+	$(DUMP_SYMS) $< | $(AWK) -f $(objroot)include/jemalloc/internal/private_symbols_jet.awk > $@
+
+$(objroot)include/jemalloc/internal/private_namespace.gen.h: $(C_SYMS)
+	$(SHELL) $(srcroot)include/jemalloc/internal/private_namespace.sh $^ > $@
+
+$(objroot)include/jemalloc/internal/private_namespace_jet.gen.h: $(C_JET_SYMS)
+	$(SHELL) $(srcroot)include/jemalloc/internal/private_namespace.sh $^ > $@
+
+%.h: %.gen.h
+	@if ! `cmp -s $< $@` ; then echo "cp $< $<"; cp $< $@ ; fi
+
+$(CPP_OBJS) $(CPP_PIC_OBJS) $(TESTS_CPP_OBJS): %.$(O):
+	@mkdir -p $(@D)
+	$(CXX) $(CXXFLAGS) -c $(CPPFLAGS) $(CTARGET) $<
+ifdef CC_MM
+	@$(CXX) -MM $(CPPFLAGS) -MT $@ -o $(@:%.$(O)=%.d) $<
+endif
+
+ifneq ($(SOREV),$(SO))
+%.$(SO) : %.$(SOREV)
+	@mkdir -p $(@D)
+	ln -sf $(<F) $@
+endif
+
+$(objroot)lib/$(LIBJEMALLOC).$(SOREV) : $(if $(PIC_CFLAGS),$(C_PIC_OBJS),$(C_OBJS)) $(if $(PIC_CFLAGS),$(CPP_PIC_OBJS),$(CPP_OBJS))
+	@mkdir -p $(@D)
+	$(CC) $(DSO_LDFLAGS) $(call RPATH,$(RPATH_EXTRA)) $(LDTARGET) $+ $(LDFLAGS) $(LIBS) $(EXTRA_LDFLAGS)
+
+$(objroot)lib/$(LIBJEMALLOC)_pic.$(A) : $(C_PIC_OBJS) $(CPP_PIC_OBJS)
+$(objroot)lib/$(LIBJEMALLOC).$(A) : $(C_OBJS) $(CPP_OBJS)
+$(objroot)lib/$(LIBJEMALLOC)_s.$(A) : $(C_OBJS) $(CPP_OBJS)
+
+$(STATIC_LIBS):
+	@mkdir -p $(@D)
+	$(AR) $(ARFLAGS)@AROUT@ $+
+
+$(objroot)test/unit/%$(EXE): $(objroot)test/unit/%.$(O) $(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS)
+	@mkdir -p $(@D)
+	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LDFLAGS) $(filter-out -lm,$(LIBS)) $(LM) $(EXTRA_LDFLAGS)
+
+$(objroot)test/integration/%$(EXE): $(objroot)test/integration/%.$(O) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+	@mkdir -p $(@D)
+	$(CC) $(TEST_LD_MODE) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LJEMALLOC) $(LDFLAGS) $(filter-out -lm,$(filter -lrt -pthread -lstdc++,$(LIBS))) $(LM) $(EXTRA_LDFLAGS)
+
+$(objroot)test/integration/cpp/%$(EXE): $(objroot)test/integration/cpp/%.$(O) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+	@mkdir -p $(@D)
+	$(CXX) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(LIBS)) -lm $(EXTRA_LDFLAGS)
+
+$(objroot)test/stress/%$(EXE): $(objroot)test/stress/%.$(O) $(C_JET_OBJS) $(C_TESTLIB_STRESS_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+	@mkdir -p $(@D)
+	$(CC) $(TEST_LD_MODE) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(LIBS)) $(LM) $(EXTRA_LDFLAGS)
+
+build_lib_shared: $(DSOS)
+build_lib_static: $(STATIC_LIBS)
+ifeq ($(enable_shared), 1)
+build_lib: build_lib_shared
+endif
+ifeq ($(enable_static), 1)
+build_lib: build_lib_static
+endif
+
+install_bin:
+	$(INSTALL) -d $(BINDIR)
+	@for b in $(BINS); do \
+	echo "$(INSTALL) -m 755 $$b $(BINDIR)"; \
+	$(INSTALL) -m 755 $$b $(BINDIR); \
+done
+
+install_include:
+	$(INSTALL) -d $(INCLUDEDIR)/jemalloc
+	@for h in $(C_HDRS); do \
+	echo "$(INSTALL) -m 644 $$h $(INCLUDEDIR)/jemalloc"; \
+	$(INSTALL) -m 644 $$h $(INCLUDEDIR)/jemalloc; \
+done
+
+install_lib_shared: $(DSOS)
+	$(INSTALL) -d $(LIBDIR)
+	$(INSTALL) -m 755 $(objroot)lib/$(LIBJEMALLOC).$(SOREV) $(LIBDIR)
+ifneq ($(SOREV),$(SO))
+	ln -sf $(LIBJEMALLOC).$(SOREV) $(LIBDIR)/$(LIBJEMALLOC).$(SO)
+endif
+
+install_lib_static: $(STATIC_LIBS)
+	$(INSTALL) -d $(LIBDIR)
+	@for l in $(STATIC_LIBS); do \
+	echo "$(INSTALL) -m 755 $$l $(LIBDIR)"; \
+	$(INSTALL) -m 755 $$l $(LIBDIR); \
+done
+
+install_lib_pc: $(PC)
+	$(INSTALL) -d $(LIBDIR)/pkgconfig
+	@for l in $(PC); do \
+	echo "$(INSTALL) -m 644 $$l $(LIBDIR)/pkgconfig"; \
+	$(INSTALL) -m 644 $$l $(LIBDIR)/pkgconfig; \
+done
+
+ifeq ($(enable_shared), 1)
+install_lib: install_lib_shared
+endif
+ifeq ($(enable_static), 1)
+install_lib: install_lib_static
+endif
+install_lib: install_lib_pc
+
+install_doc_html:
+	$(INSTALL) -d $(DATADIR)/doc/jemalloc$(install_suffix)
+	@for d in $(DOCS_HTML); do \
+	echo "$(INSTALL) -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix)"; \
+	$(INSTALL) -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix); \
+done
+
+install_doc_man:
+	$(INSTALL) -d $(MANDIR)/man3
+	@for d in $(DOCS_MAN3); do \
+	echo "$(INSTALL) -m 644 $$d $(MANDIR)/man3"; \
+	$(INSTALL) -m 644 $$d $(MANDIR)/man3; \
+done
+
+install_doc: build_doc install_doc_html install_doc_man
+
+install: install_bin install_include install_lib
+
+ifeq ($(enable_doc), 1)
+install: install_doc
+endif
+
+tests_unit: $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%$(EXE))
+tests_integration: $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%$(EXE)) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%$(EXE))
+tests_stress: $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%$(EXE))
+tests: tests_unit tests_integration tests_stress
+
+check_unit_dir:
+	@mkdir -p $(objroot)test/unit
+check_integration_dir:
+	@mkdir -p $(objroot)test/integration
+stress_dir:
+	@mkdir -p $(objroot)test/stress
+check_dir: check_unit_dir check_integration_dir
+
+check_unit: tests_unit check_unit_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%)
+check_integration_prof: tests_integration check_integration_dir
+ifeq ($(enable_prof), 1)
+	$(MALLOC_CONF)="prof:true" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+	$(MALLOC_CONF)="prof:true,prof_active:false" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+endif
+check_integration_decay: tests_integration check_integration_dir
+	$(MALLOC_CONF)="dirty_decay_ms:-1,muzzy_decay_ms:-1" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+	$(MALLOC_CONF)="dirty_decay_ms:0,muzzy_decay_ms:0" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+check_integration: tests_integration check_integration_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+stress: tests_stress stress_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%)
+check: check_unit check_integration check_integration_decay check_integration_prof
+
+clean:
+	rm -f $(PRIVATE_NAMESPACE_HDRS)
+	rm -f $(PRIVATE_NAMESPACE_GEN_HDRS)
+	rm -f $(C_SYM_OBJS)
+	rm -f $(C_SYMS)
+	rm -f $(C_OBJS)
+	rm -f $(CPP_OBJS)
+	rm -f $(C_PIC_OBJS)
+	rm -f $(CPP_PIC_OBJS)
+	rm -f $(C_JET_SYM_OBJS)
+	rm -f $(C_JET_SYMS)
+	rm -f $(C_JET_OBJS)
+	rm -f $(C_TESTLIB_OBJS)
+	rm -f $(C_SYM_OBJS:%.$(O)=%.d)
+	rm -f $(C_OBJS:%.$(O)=%.d)
+	rm -f $(CPP_OBJS:%.$(O)=%.d)
+	rm -f $(C_PIC_OBJS:%.$(O)=%.d)
+	rm -f $(CPP_PIC_OBJS:%.$(O)=%.d)
+	rm -f $(C_JET_SYM_OBJS:%.$(O)=%.d)
+	rm -f $(C_JET_OBJS:%.$(O)=%.d)
+	rm -f $(C_TESTLIB_OBJS:%.$(O)=%.d)
+	rm -f $(TESTS_OBJS:%.$(O)=%$(EXE))
+	rm -f $(TESTS_OBJS)
+	rm -f $(TESTS_OBJS:%.$(O)=%.d)
+	rm -f $(TESTS_OBJS:%.$(O)=%.out)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%$(EXE))
+	rm -f $(TESTS_CPP_OBJS)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%.d)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%.out)
+	rm -f $(DSOS) $(STATIC_LIBS)
+
+distclean: clean
+	rm -f $(objroot)bin/jemalloc-config
+	rm -f $(objroot)bin/jemalloc.sh
+	rm -f $(objroot)bin/jeprof
+	rm -f $(objroot)config.log
+	rm -f $(objroot)config.status
+	rm -f $(objroot)config.stamp
+	rm -f $(cfghdrs_out)
+	rm -f $(cfgoutputs_out)
+
+relclean: distclean
+	rm -f $(objroot)configure
+	rm -f $(objroot)VERSION
+	rm -f $(DOCS_HTML)
+	rm -f $(DOCS_MAN3)
+
+#===============================================================================
+# Re-configuration rules.
+
+ifeq ($(enable_autogen), 1)
+$(srcroot)configure : $(srcroot)configure.ac
+	cd ./$(srcroot) && $(AUTOCONF)
+
+$(objroot)config.status : $(srcroot)configure
+	./$(objroot)config.status --recheck
+
+$(srcroot)config.stamp.in : $(srcroot)configure.ac
+	echo stamp > $(srcroot)config.stamp.in
+
+$(objroot)config.stamp : $(cfgoutputs_in) $(cfghdrs_in) $(srcroot)configure
+	./$(objroot)config.status
+	@touch $@
+
+# There must be some action in order for make to re-read Makefile when it is
+# out of date.
+$(cfgoutputs_out) $(cfghdrs_out) : $(objroot)config.stamp
+	@true
+endif
diff --git a/jemalloc/README b/jemalloc/README
new file mode 100644
index 0000000..3a6e0d2
--- /dev/null
+++ b/jemalloc/README
@@ -0,0 +1,20 @@
+jemalloc is a general purpose malloc(3) implementation that emphasizes
+fragmentation avoidance and scalable concurrency support.  jemalloc first came
+into use as the FreeBSD libc allocator in 2005, and since then it has found its
+way into numerous applications that rely on its predictable behavior.  In 2010
+jemalloc development efforts broadened to include developer support features
+such as heap profiling and extensive monitoring/tuning hooks.  Modern jemalloc
+releases continue to be integrated back into FreeBSD, and therefore versatility
+remains critical.  Ongoing development efforts trend toward making jemalloc
+among the best allocators for a broad range of demanding applications, and
+eliminating/mitigating weaknesses that have practical repercussions for real
+world applications.
+
+The COPYING file contains copyright and licensing information.
+
+The INSTALL file contains information on how to configure, build, and install
+jemalloc.
+
+The ChangeLog file contains a brief summary of changes for each release.
+
+URL: http://jemalloc.net/
diff --git a/jemalloc/TUNING.md b/jemalloc/TUNING.md
new file mode 100644
index 0000000..7d094f3
--- /dev/null
+++ b/jemalloc/TUNING.md
@@ -0,0 +1,129 @@
+This document summarizes the common approaches for performance fine tuning with
+jemalloc (as of 5.1.0).  The default configuration of jemalloc tends to work
+reasonably well in practice, and most applications should not have to tune any
+options. However, in order to cover a wide range of applications and avoid
+pathological cases, the default setting is sometimes kept conservative and
+suboptimal, even for many common workloads.  When jemalloc is properly tuned for
+a specific application / workload, it is common to improve system level metrics
+by a few percent, or make favorable trade-offs.
+
+
+## Notable runtime options for performance tuning
+
+Runtime options can be set via
+[malloc_conf](http://jemalloc.net/jemalloc.3.html#tuning).
+
+* [background_thread](http://jemalloc.net/jemalloc.3.html#background_thread)
+
+    Enabling jemalloc background threads generally improves the tail latency for
+    application threads, since unused memory purging is shifted to the dedicated
+    background threads.  In addition, unintended purging delay caused by
+    application inactivity is avoided with background threads.
+
+    Suggested: `background_thread:true` when jemalloc managed threads can be
+    allowed.
+
+* [metadata_thp](http://jemalloc.net/jemalloc.3.html#opt.metadata_thp)
+
+    Allowing jemalloc to utilize transparent huge pages for its internal
+    metadata usually reduces TLB misses significantly, especially for programs
+    with large memory footprint and frequent allocation / deallocation
+    activities.  Metadata memory usage may increase due to the use of huge
+    pages.
+
+    Suggested for allocation intensive programs: `metadata_thp:auto` or
+    `metadata_thp:always`, which is expected to improve CPU utilization at a
+    small memory cost.
+
+* [dirty_decay_ms](http://jemalloc.net/jemalloc.3.html#opt.dirty_decay_ms) and
+  [muzzy_decay_ms](http://jemalloc.net/jemalloc.3.html#opt.muzzy_decay_ms)
+
+    Decay time determines how fast jemalloc returns unused pages back to the
+    operating system, and therefore provides a fairly straightforward trade-off
+    between CPU and memory usage.  Shorter decay time purges unused pages faster
+    to reduces memory usage (usually at the cost of more CPU cycles spent on
+    purging), and vice versa.
+
+    Suggested: tune the values based on the desired trade-offs.
+
+* [narenas](http://jemalloc.net/jemalloc.3.html#opt.narenas)
+
+    By default jemalloc uses multiple arenas to reduce internal lock contention.
+    However high arena count may also increase overall memory fragmentation,
+    since arenas manage memory independently.  When high degree of parallelism
+    is not expected at the allocator level, lower number of arenas often
+    improves memory usage.
+
+    Suggested: if low parallelism is expected, try lower arena count while
+    monitoring CPU and memory usage.
+
+* [percpu_arena](http://jemalloc.net/jemalloc.3.html#opt.percpu_arena)
+
+    Enable dynamic thread to arena association based on running CPU.  This has
+    the potential to improve locality, e.g. when thread to CPU affinity is
+    present.
+
+    Suggested: try `percpu_arena:percpu` or `percpu_arena:phycpu` if
+    thread migration between processors is expected to be infrequent.
+
+Examples:
+
+* High resource consumption application, prioritizing CPU utilization:
+
+    `background_thread:true,metadata_thp:auto` combined with relaxed decay time
+    (increased `dirty_decay_ms` and / or `muzzy_decay_ms`,
+    e.g. `dirty_decay_ms:30000,muzzy_decay_ms:30000`).
+
+* High resource consumption application, prioritizing memory usage:
+
+    `background_thread:true` combined with shorter decay time (decreased
+    `dirty_decay_ms` and / or `muzzy_decay_ms`,
+    e.g. `dirty_decay_ms:5000,muzzy_decay_ms:5000`), and lower arena count
+    (e.g. number of CPUs).
+
+* Low resource consumption application:
+
+    `narenas:1,lg_tcache_max:13` combined with shorter decay time (decreased
+    `dirty_decay_ms` and / or `muzzy_decay_ms`,e.g.
+    `dirty_decay_ms:1000,muzzy_decay_ms:0`).
+
+* Extremely conservative -- minimize memory usage at all costs, only suitable when
+allocation activity is very rare:
+
+    `narenas:1,tcache:false,dirty_decay_ms:0,muzzy_decay_ms:0`
+
+Note that it is recommended to combine the options with `abort_conf:true` which
+aborts immediately on illegal options.
+
+## Beyond runtime options
+
+In addition to the runtime options, there are a number of programmatic ways to
+improve application performance with jemalloc.
+
+* [Explicit arenas](http://jemalloc.net/jemalloc.3.html#arenas.create)
+
+    Manually created arenas can help performance in various ways, e.g. by
+    managing locality and contention for specific usages.  For example,
+    applications can explicitly allocate frequently accessed objects from a
+    dedicated arena with
+    [mallocx()](http://jemalloc.net/jemalloc.3.html#MALLOCX_ARENA) to improve
+    locality.  In addition, explicit arenas often benefit from individually
+    tuned options, e.g. relaxed [decay
+    time](http://jemalloc.net/jemalloc.3.html#arena.i.dirty_decay_ms) if
+    frequent reuse is expected.
+
+* [Extent hooks](http://jemalloc.net/jemalloc.3.html#arena.i.extent_hooks)
+
+    Extent hooks allow customization for managing underlying memory.  One use
+    case for performance purpose is to utilize huge pages -- for example,
+    [HHVM](https://github.com/facebook/hhvm/blob/master/hphp/util/alloc.cpp)
+    uses explicit arenas with customized extent hooks to manage 1GB huge pages
+    for frequently accessed data, which reduces TLB misses significantly.
+
+* [Explicit thread-to-arena
+  binding](http://jemalloc.net/jemalloc.3.html#thread.arena)
+
+    It is common for some threads in an application to have different memory
+    access / allocation patterns.  Threads with heavy workloads often benefit
+    from explicit binding, e.g. binding very active threads to dedicated arenas
+    may reduce contention at the allocator level.
diff --git a/jemalloc/autogen.sh b/jemalloc/autogen.sh
new file mode 100755
index 0000000..75f32da
--- /dev/null
+++ b/jemalloc/autogen.sh
@@ -0,0 +1,17 @@
+#!/bin/sh
+
+for i in autoconf; do
+    echo "$i"
+    $i
+    if [ $? -ne 0 ]; then
+	echo "Error $? in $i"
+	exit 1
+    fi
+done
+
+echo "./configure --enable-autogen $@"
+./configure --enable-autogen $@
+if [ $? -ne 0 ]; then
+    echo "Error $? in ./configure"
+    exit 1
+fi
diff --git a/jemalloc/bin/jemalloc-config.in b/jemalloc/bin/jemalloc-config.in
new file mode 100644
index 0000000..80eca2e
--- /dev/null
+++ b/jemalloc/bin/jemalloc-config.in
@@ -0,0 +1,83 @@
+#!/bin/sh
+
+usage() {
+	cat <<EOF
+Usage:
+  @BINDIR@/jemalloc-config <option>
+Options:
+  --help | -h  : Print usage.
+  --version    : Print jemalloc version.
+  --revision   : Print shared library revision number.
+  --config     : Print configure options used to build jemalloc.
+  --prefix     : Print installation directory prefix.
+  --bindir     : Print binary installation directory.
+  --datadir    : Print data installation directory.
+  --includedir : Print include installation directory.
+  --libdir     : Print library installation directory.
+  --mandir     : Print manual page installation directory.
+  --cc         : Print compiler used to build jemalloc.
+  --cflags     : Print compiler flags used to build jemalloc.
+  --cppflags   : Print preprocessor flags used to build jemalloc.
+  --cxxflags   : Print C++ compiler flags used to build jemalloc.
+  --ldflags    : Print library flags used to build jemalloc.
+  --libs       : Print libraries jemalloc was linked against.
+EOF
+}
+
+prefix="@prefix@"
+exec_prefix="@exec_prefix@"
+
+case "$1" in
+--help | -h)
+	usage
+	exit 0
+	;;
+--version)
+	echo "@jemalloc_version@"
+	;;
+--revision)
+	echo "@rev@"
+	;;
+--config)
+	echo "@CONFIG@"
+	;;
+--prefix)
+	echo "@PREFIX@"
+	;;
+--bindir)
+	echo "@BINDIR@"
+	;;
+--datadir)
+	echo "@DATADIR@"
+	;;
+--includedir)
+	echo "@INCLUDEDIR@"
+	;;
+--libdir)
+	echo "@LIBDIR@"
+	;;
+--mandir)
+	echo "@MANDIR@"
+	;;
+--cc)
+	echo "@CC@"
+	;;
+--cflags)
+	echo "@CFLAGS@"
+	;;
+--cppflags)
+	echo "@CPPFLAGS@"
+	;;
+--cxxflags)
+	echo "@CXXFLAGS@"
+	;;
+--ldflags)
+	echo "@LDFLAGS@ @EXTRA_LDFLAGS@"
+	;;
+--libs)
+	echo "@LIBS@"
+	;;
+*)
+	usage
+	exit 1
+esac
diff --git a/jemalloc/bin/jemalloc.sh.in b/jemalloc/bin/jemalloc.sh.in
new file mode 100644
index 0000000..cdf3673
--- /dev/null
+++ b/jemalloc/bin/jemalloc.sh.in
@@ -0,0 +1,9 @@
+#!/bin/sh
+
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+
+@LD_PRELOAD_VAR@=${libdir}/libjemalloc.@SOREV@
+export @LD_PRELOAD_VAR@
+exec "$@"
diff --git a/jemalloc/bin/jeprof.in b/jemalloc/bin/jeprof.in
new file mode 100644
index 0000000..3ed408c
--- /dev/null
+++ b/jemalloc/bin/jeprof.in
@@ -0,0 +1,5625 @@
+#! /usr/bin/env perl
+
+# Copyright (c) 1998-2007, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+#     * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+#     * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+#     * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# ---
+# Program for printing the profile generated by common/profiler.cc,
+# or by the heap profiler (common/debugallocation.cc)
+#
+# The profile contains a sequence of entries of the form:
+#       <count> <stack trace>
+# This program parses the profile, and generates user-readable
+# output.
+#
+# Examples:
+#
+# % tools/jeprof "program" "profile"
+#   Enters "interactive" mode
+#
+# % tools/jeprof --text "program" "profile"
+#   Generates one line per procedure
+#
+# % tools/jeprof --gv "program" "profile"
+#   Generates annotated call-graph and displays via "gv"
+#
+# % tools/jeprof --gv --focus=Mutex "program" "profile"
+#   Restrict to code paths that involve an entry that matches "Mutex"
+#
+# % tools/jeprof --gv --focus=Mutex --ignore=string "program" "profile"
+#   Restrict to code paths that involve an entry that matches "Mutex"
+#   and does not match "string"
+#
+# % tools/jeprof --list=IBF_CheckDocid "program" "profile"
+#   Generates disassembly listing of all routines with at least one
+#   sample that match the --list=<regexp> pattern.  The listing is
+#   annotated with the flat and cumulative sample counts at each line.
+#
+# % tools/jeprof --disasm=IBF_CheckDocid "program" "profile"
+#   Generates disassembly listing of all routines with at least one
+#   sample that match the --disasm=<regexp> pattern.  The listing is
+#   annotated with the flat and cumulative sample counts at each PC value.
+#
+# TODO: Use color to indicate files?
+
+use strict;
+use warnings;
+use Getopt::Long;
+use Cwd;
+
+my $JEPROF_VERSION = "@jemalloc_version@";
+my $PPROF_VERSION = "2.0";
+
+# These are the object tools we use which can come from a
+# user-specified location using --tools, from the JEPROF_TOOLS
+# environment variable, or from the environment.
+my %obj_tool_map = (
+  "objdump" => "objdump",
+  "nm" => "nm",
+  "addr2line" => "addr2line",
+  "c++filt" => "c++filt",
+  ## ConfigureObjTools may add architecture-specific entries:
+  #"nm_pdb" => "nm-pdb",       # for reading windows (PDB-format) executables
+  #"addr2line_pdb" => "addr2line-pdb",                                # ditto
+  #"otool" => "otool",         # equivalent of objdump on OS X
+);
+# NOTE: these are lists, so you can put in commandline flags if you want.
+my @DOT = ("dot");          # leave non-absolute, since it may be in /usr/local
+my @GV = ("gv");
+my @EVINCE = ("evince");    # could also be xpdf or perhaps acroread
+my @KCACHEGRIND = ("kcachegrind");
+my @PS2PDF = ("ps2pdf");
+# These are used for dynamic profiles
+my @URL_FETCHER = ("curl", "-s", "--fail");
+
+# These are the web pages that servers need to support for dynamic profiles
+my $HEAP_PAGE = "/pprof/heap";
+my $PROFILE_PAGE = "/pprof/profile";   # must support cgi-param "?seconds=#"
+my $PMUPROFILE_PAGE = "/pprof/pmuprofile(?:\\?.*)?"; # must support cgi-param
+                                                # ?seconds=#&event=x&period=n
+my $GROWTH_PAGE = "/pprof/growth";
+my $CONTENTION_PAGE = "/pprof/contention";
+my $WALL_PAGE = "/pprof/wall(?:\\?.*)?";  # accepts options like namefilter
+my $FILTEREDPROFILE_PAGE = "/pprof/filteredprofile(?:\\?.*)?";
+my $CENSUSPROFILE_PAGE = "/pprof/censusprofile(?:\\?.*)?"; # must support cgi-param
+                                                       # "?seconds=#",
+                                                       # "?tags_regexp=#" and
+                                                       # "?type=#".
+my $SYMBOL_PAGE = "/pprof/symbol";     # must support symbol lookup via POST
+my $PROGRAM_NAME_PAGE = "/pprof/cmdline";
+
+# These are the web pages that can be named on the command line.
+# All the alternatives must begin with /.
+my $PROFILES = "($HEAP_PAGE|$PROFILE_PAGE|$PMUPROFILE_PAGE|" .
+               "$GROWTH_PAGE|$CONTENTION_PAGE|$WALL_PAGE|" .
+               "$FILTEREDPROFILE_PAGE|$CENSUSPROFILE_PAGE)";
+
+# default binary name
+my $UNKNOWN_BINARY = "(unknown)";
+
+# There is a pervasive dependency on the length (in hex characters,
+# i.e., nibbles) of an address, distinguishing between 32-bit and
+# 64-bit profiles.  To err on the safe size, default to 64-bit here:
+my $address_length = 16;
+
+my $dev_null = "/dev/null";
+if (! -e $dev_null && $^O =~ /MSWin/) {    # $^O is the OS perl was built for
+  $dev_null = "nul";
+}
+
+# A list of paths to search for shared object files
+my @prefix_list = ();
+
+# Special routine name that should not have any symbols.
+# Used as separator to parse "addr2line -i" output.
+my $sep_symbol = '_fini';
+my $sep_address = undef;
+
+##### Argument parsing #####
+
+sub usage_string {
+  return <<EOF;
+Usage:
+jeprof [options] <program> <profiles>
+   <profiles> is a space separated list of profile names.
+jeprof [options] <symbolized-profiles>
+   <symbolized-profiles> is a list of profile files where each file contains
+   the necessary symbol mappings  as well as profile data (likely generated
+   with --raw).
+jeprof [options] <profile>
+   <profile> is a remote form.  Symbols are obtained from host:port$SYMBOL_PAGE
+
+   Each name can be:
+   /path/to/profile        - a path to a profile file
+   host:port[/<service>]   - a location of a service to get profile from
+
+   The /<service> can be $HEAP_PAGE, $PROFILE_PAGE, /pprof/pmuprofile,
+                         $GROWTH_PAGE, $CONTENTION_PAGE, /pprof/wall,
+                         $CENSUSPROFILE_PAGE, or /pprof/filteredprofile.
+   For instance:
+     jeprof http://myserver.com:80$HEAP_PAGE
+   If /<service> is omitted, the service defaults to $PROFILE_PAGE (cpu profiling).
+jeprof --symbols <program>
+   Maps addresses to symbol names.  In this mode, stdin should be a
+   list of library mappings, in the same format as is found in the heap-
+   and cpu-profile files (this loosely matches that of /proc/self/maps
+   on linux), followed by a list of hex addresses to map, one per line.
+
+   For more help with querying remote servers, including how to add the
+   necessary server-side support code, see this filename (or one like it):
+
+   /usr/doc/gperftools-$PPROF_VERSION/pprof_remote_servers.html
+
+Options:
+   --cum               Sort by cumulative data
+   --base=<base>       Subtract <base> from <profile> before display
+   --interactive       Run in interactive mode (interactive "help" gives help) [default]
+   --seconds=<n>       Length of time for dynamic profiles [default=30 secs]
+   --add_lib=<file>    Read additional symbols and line info from the given library
+   --lib_prefix=<dir>  Comma separated list of library path prefixes
+
+Reporting Granularity:
+   --addresses         Report at address level
+   --lines             Report at source line level
+   --functions         Report at function level [default]
+   --files             Report at source file level
+
+Output type:
+   --text              Generate text report
+   --callgrind         Generate callgrind format to stdout
+   --gv                Generate Postscript and display
+   --evince            Generate PDF and display
+   --web               Generate SVG and display
+   --list=<regexp>     Generate source listing of matching routines
+   --disasm=<regexp>   Generate disassembly of matching routines
+   --symbols           Print demangled symbol names found at given addresses
+   --dot               Generate DOT file to stdout
+   --ps                Generate Postcript to stdout
+   --pdf               Generate PDF to stdout
+   --svg               Generate SVG to stdout
+   --gif               Generate GIF to stdout
+   --raw               Generate symbolized jeprof data (useful with remote fetch)
+
+Heap-Profile Options:
+   --inuse_space       Display in-use (mega)bytes [default]
+   --inuse_objects     Display in-use objects
+   --alloc_space       Display allocated (mega)bytes
+   --alloc_objects     Display allocated objects
+   --show_bytes        Display space in bytes
+   --drop_negative     Ignore negative differences
+
+Contention-profile options:
+   --total_delay       Display total delay at each region [default]
+   --contentions       Display number of delays at each region
+   --mean_delay        Display mean delay at each region
+
+Call-graph Options:
+   --nodecount=<n>     Show at most so many nodes [default=80]
+   --nodefraction=<f>  Hide nodes below <f>*total [default=.005]
+   --edgefraction=<f>  Hide edges below <f>*total [default=.001]
+   --maxdegree=<n>     Max incoming/outgoing edges per node [default=8]
+   --focus=<regexp>    Focus on backtraces with nodes matching <regexp>
+   --thread=<n>        Show profile for thread <n>
+   --ignore=<regexp>   Ignore backtraces with nodes matching <regexp>
+   --scale=<n>         Set GV scaling [default=0]
+   --heapcheck         Make nodes with non-0 object counts
+                       (i.e. direct leak generators) more visible
+   --retain=<regexp>   Retain only nodes that match <regexp>
+   --exclude=<regexp>  Exclude all nodes that match <regexp>
+
+Miscellaneous:
+   --tools=<prefix or binary:fullpath>[,...]   \$PATH for object tool pathnames
+   --test              Run unit tests
+   --help              This message
+   --version           Version information
+
+Environment Variables:
+   JEPROF_TMPDIR        Profiles directory. Defaults to \$HOME/jeprof
+   JEPROF_TOOLS         Prefix for object tools pathnames
+
+Examples:
+
+jeprof /bin/ls ls.prof
+                       Enters "interactive" mode
+jeprof --text /bin/ls ls.prof
+                       Outputs one line per procedure
+jeprof --web /bin/ls ls.prof
+                       Displays annotated call-graph in web browser
+jeprof --gv /bin/ls ls.prof
+                       Displays annotated call-graph via 'gv'
+jeprof --gv --focus=Mutex /bin/ls ls.prof
+                       Restricts to code paths including a .*Mutex.* entry
+jeprof --gv --focus=Mutex --ignore=string /bin/ls ls.prof
+                       Code paths including Mutex but not string
+jeprof --list=getdir /bin/ls ls.prof
+                       (Per-line) annotated source listing for getdir()
+jeprof --disasm=getdir /bin/ls ls.prof
+                       (Per-PC) annotated disassembly for getdir()
+
+jeprof http://localhost:1234/
+                       Enters "interactive" mode
+jeprof --text localhost:1234
+                       Outputs one line per procedure for localhost:1234
+jeprof --raw localhost:1234 > ./local.raw
+jeprof --text ./local.raw
+                       Fetches a remote profile for later analysis and then
+                       analyzes it in text mode.
+EOF
+}
+
+sub version_string {
+  return <<EOF
+jeprof (part of jemalloc $JEPROF_VERSION)
+based on pprof (part of gperftools $PPROF_VERSION)
+
+Copyright 1998-2007 Google Inc.
+
+This is BSD licensed software; see the source for copying conditions
+and license information.
+There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
+PARTICULAR PURPOSE.
+EOF
+}
+
+sub usage {
+  my $msg = shift;
+  print STDERR "$msg\n\n";
+  print STDERR usage_string();
+  print STDERR "\nFATAL ERROR: $msg\n";    # just as a reminder
+  exit(1);
+}
+
+sub Init() {
+  # Setup tmp-file name and handler to clean it up.
+  # We do this in the very beginning so that we can use
+  # error() and cleanup() function anytime here after.
+  $main::tmpfile_sym = "/tmp/jeprof$$.sym";
+  $main::tmpfile_ps = "/tmp/jeprof$$";
+  $main::next_tmpfile = 0;
+  $SIG{'INT'} = \&sighandler;
+
+  # Cache from filename/linenumber to source code
+  $main::source_cache = ();
+
+  $main::opt_help = 0;
+  $main::opt_version = 0;
+
+  $main::opt_cum = 0;
+  $main::opt_base = '';
+  $main::opt_addresses = 0;
+  $main::opt_lines = 0;
+  $main::opt_functions = 0;
+  $main::opt_files = 0;
+  $main::opt_lib_prefix = "";
+
+  $main::opt_text = 0;
+  $main::opt_callgrind = 0;
+  $main::opt_list = "";
+  $main::opt_disasm = "";
+  $main::opt_symbols = 0;
+  $main::opt_gv = 0;
+  $main::opt_evince = 0;
+  $main::opt_web = 0;
+  $main::opt_dot = 0;
+  $main::opt_ps = 0;
+  $main::opt_pdf = 0;
+  $main::opt_gif = 0;
+  $main::opt_svg = 0;
+  $main::opt_raw = 0;
+
+  $main::opt_nodecount = 80;
+  $main::opt_nodefraction = 0.005;
+  $main::opt_edgefraction = 0.001;
+  $main::opt_maxdegree = 8;
+  $main::opt_focus = '';
+  $main::opt_thread = undef;
+  $main::opt_ignore = '';
+  $main::opt_scale = 0;
+  $main::opt_heapcheck = 0;
+  $main::opt_retain = '';
+  $main::opt_exclude = '';
+  $main::opt_seconds = 30;
+  $main::opt_lib = "";
+
+  $main::opt_inuse_space   = 0;
+  $main::opt_inuse_objects = 0;
+  $main::opt_alloc_space   = 0;
+  $main::opt_alloc_objects = 0;
+  $main::opt_show_bytes    = 0;
+  $main::opt_drop_negative = 0;
+  $main::opt_interactive   = 0;
+
+  $main::opt_total_delay = 0;
+  $main::opt_contentions = 0;
+  $main::opt_mean_delay = 0;
+
+  $main::opt_tools   = "";
+  $main::opt_debug   = 0;
+  $main::opt_test    = 0;
+
+  # These are undocumented flags used only by unittests.
+  $main::opt_test_stride = 0;
+
+  # Are we using $SYMBOL_PAGE?
+  $main::use_symbol_page = 0;
+
+  # Files returned by TempName.
+  %main::tempnames = ();
+
+  # Type of profile we are dealing with
+  # Supported types:
+  #     cpu
+  #     heap
+  #     growth
+  #     contention
+  $main::profile_type = '';     # Empty type means "unknown"
+
+  GetOptions("help!"          => \$main::opt_help,
+             "version!"       => \$main::opt_version,
+             "cum!"           => \$main::opt_cum,
+             "base=s"         => \$main::opt_base,
+             "seconds=i"      => \$main::opt_seconds,
+             "add_lib=s"      => \$main::opt_lib,
+             "lib_prefix=s"   => \$main::opt_lib_prefix,
+             "functions!"     => \$main::opt_functions,
+             "lines!"         => \$main::opt_lines,
+             "addresses!"     => \$main::opt_addresses,
+             "files!"         => \$main::opt_files,
+             "text!"          => \$main::opt_text,
+             "callgrind!"     => \$main::opt_callgrind,
+             "list=s"         => \$main::opt_list,
+             "disasm=s"       => \$main::opt_disasm,
+             "symbols!"       => \$main::opt_symbols,
+             "gv!"            => \$main::opt_gv,
+             "evince!"        => \$main::opt_evince,
+             "web!"           => \$main::opt_web,
+             "dot!"           => \$main::opt_dot,
+             "ps!"            => \$main::opt_ps,
+             "pdf!"           => \$main::opt_pdf,
+             "svg!"           => \$main::opt_svg,
+             "gif!"           => \$main::opt_gif,
+             "raw!"           => \$main::opt_raw,
+             "interactive!"   => \$main::opt_interactive,
+             "nodecount=i"    => \$main::opt_nodecount,
+             "nodefraction=f" => \$main::opt_nodefraction,
+             "edgefraction=f" => \$main::opt_edgefraction,
+             "maxdegree=i"    => \$main::opt_maxdegree,
+             "focus=s"        => \$main::opt_focus,
+             "thread=s"       => \$main::opt_thread,
+             "ignore=s"       => \$main::opt_ignore,
+             "scale=i"        => \$main::opt_scale,
+             "heapcheck"      => \$main::opt_heapcheck,
+             "retain=s"       => \$main::opt_retain,
+             "exclude=s"      => \$main::opt_exclude,
+             "inuse_space!"   => \$main::opt_inuse_space,
+             "inuse_objects!" => \$main::opt_inuse_objects,
+             "alloc_space!"   => \$main::opt_alloc_space,
+             "alloc_objects!" => \$main::opt_alloc_objects,
+             "show_bytes!"    => \$main::opt_show_bytes,
+             "drop_negative!" => \$main::opt_drop_negative,
+             "total_delay!"   => \$main::opt_total_delay,
+             "contentions!"   => \$main::opt_contentions,
+             "mean_delay!"    => \$main::opt_mean_delay,
+             "tools=s"        => \$main::opt_tools,
+             "test!"          => \$main::opt_test,
+             "debug!"         => \$main::opt_debug,
+             # Undocumented flags used only by unittests:
+             "test_stride=i"  => \$main::opt_test_stride,
+      ) || usage("Invalid option(s)");
+
+  # Deal with the standard --help and --version
+  if ($main::opt_help) {
+    print usage_string();
+    exit(0);
+  }
+
+  if ($main::opt_version) {
+    print version_string();
+    exit(0);
+  }
+
+  # Disassembly/listing/symbols mode requires address-level info
+  if ($main::opt_disasm || $main::opt_list || $main::opt_symbols) {
+    $main::opt_functions = 0;
+    $main::opt_lines = 0;
+    $main::opt_addresses = 1;
+    $main::opt_files = 0;
+  }
+
+  # Check heap-profiling flags
+  if ($main::opt_inuse_space +
+      $main::opt_inuse_objects +
+      $main::opt_alloc_space +
+      $main::opt_alloc_objects > 1) {
+    usage("Specify at most on of --inuse/--alloc options");
+  }
+
+  # Check output granularities
+  my $grains =
+      $main::opt_functions +
+      $main::opt_lines +
+      $main::opt_addresses +
+      $main::opt_files +
+      0;
+  if ($grains > 1) {
+    usage("Only specify one output granularity option");
+  }
+  if ($grains == 0) {
+    $main::opt_functions = 1;
+  }
+
+  # Check output modes
+  my $modes =
+      $main::opt_text +
+      $main::opt_callgrind +
+      ($main::opt_list eq '' ? 0 : 1) +
+      ($main::opt_disasm eq '' ? 0 : 1) +
+      ($main::opt_symbols == 0 ? 0 : 1) +
+      $main::opt_gv +
+      $main::opt_evince +
+      $main::opt_web +
+      $main::opt_dot +
+      $main::opt_ps +
+      $main::opt_pdf +
+      $main::opt_svg +
+      $main::opt_gif +
+      $main::opt_raw +
+      $main::opt_interactive +
+      0;
+  if ($modes > 1) {
+    usage("Only specify one output mode");
+  }
+  if ($modes == 0) {
+    if (-t STDOUT) {  # If STDOUT is a tty, activate interactive mode
+      $main::opt_interactive = 1;
+    } else {
+      $main::opt_text = 1;
+    }
+  }
+
+  if ($main::opt_test) {
+    RunUnitTests();
+    # Should not return
+    exit(1);
+  }
+
+  # Binary name and profile arguments list
+  $main::prog = "";
+  @main::pfile_args = ();
+
+  # Remote profiling without a binary (using $SYMBOL_PAGE instead)
+  if (@ARGV > 0) {
+    if (IsProfileURL($ARGV[0])) {
+      $main::use_symbol_page = 1;
+    } elsif (IsSymbolizedProfileFile($ARGV[0])) {
+      $main::use_symbolized_profile = 1;
+      $main::prog = $UNKNOWN_BINARY;  # will be set later from the profile file
+    }
+  }
+
+  if ($main::use_symbol_page || $main::use_symbolized_profile) {
+    # We don't need a binary!
+    my %disabled = ('--lines' => $main::opt_lines,
+                    '--disasm' => $main::opt_disasm);
+    for my $option (keys %disabled) {
+      usage("$option cannot be used without a binary") if $disabled{$option};
+    }
+    # Set $main::prog later...
+    scalar(@ARGV) || usage("Did not specify profile file");
+  } elsif ($main::opt_symbols) {
+    # --symbols needs a binary-name (to run nm on, etc) but not profiles
+    $main::prog = shift(@ARGV) || usage("Did not specify program");
+  } else {
+    $main::prog = shift(@ARGV) || usage("Did not specify program");
+    scalar(@ARGV) || usage("Did not specify profile file");
+  }
+
+  # Parse profile file/location arguments
+  foreach my $farg (@ARGV) {
+    if ($farg =~ m/(.*)\@([0-9]+)(|\/.*)$/ ) {
+      my $machine = $1;
+      my $num_machines = $2;
+      my $path = $3;
+      for (my $i = 0; $i < $num_machines; $i++) {
+        unshift(@main::pfile_args, "$i.$machine$path");
+      }
+    } else {
+      unshift(@main::pfile_args, $farg);
+    }
+  }
+
+  if ($main::use_symbol_page) {
+    unless (IsProfileURL($main::pfile_args[0])) {
+      error("The first profile should be a remote form to use $SYMBOL_PAGE\n");
+    }
+    CheckSymbolPage();
+    $main::prog = FetchProgramName();
+  } elsif (!$main::use_symbolized_profile) {  # may not need objtools!
+    ConfigureObjTools($main::prog)
+  }
+
+  # Break the opt_lib_prefix into the prefix_list array
+  @prefix_list = split (',', $main::opt_lib_prefix);
+
+  # Remove trailing / from the prefixes, in the list to prevent
+  # searching things like /my/path//lib/mylib.so
+  foreach (@prefix_list) {
+    s|/+$||;
+  }
+}
+
+sub FilterAndPrint {
+  my ($profile, $symbols, $libs, $thread) = @_;
+
+  # Get total data in profile
+  my $total = TotalProfile($profile);
+
+  # Remove uniniteresting stack items
+  $profile = RemoveUninterestingFrames($symbols, $profile);
+
+  # Focus?
+  if ($main::opt_focus ne '') {
+    $profile = FocusProfile($symbols, $profile, $main::opt_focus);
+  }
+
+  # Ignore?
+  if ($main::opt_ignore ne '') {
+    $profile = IgnoreProfile($symbols, $profile, $main::opt_ignore);
+  }
+
+  my $calls = ExtractCalls($symbols, $profile);
+
+  # Reduce profiles to required output granularity, and also clean
+  # each stack trace so a given entry exists at most once.
+  my $reduced = ReduceProfile($symbols, $profile);
+
+  # Get derived profiles
+  my $flat = FlatProfile($reduced);
+  my $cumulative = CumulativeProfile($reduced);
+
+  # Print
+  if (!$main::opt_interactive) {
+    if ($main::opt_disasm) {
+      PrintDisassembly($libs, $flat, $cumulative, $main::opt_disasm);
+    } elsif ($main::opt_list) {
+      PrintListing($total, $libs, $flat, $cumulative, $main::opt_list, 0);
+    } elsif ($main::opt_text) {
+      # Make sure the output is empty when have nothing to report
+      # (only matters when --heapcheck is given but we must be
+      # compatible with old branches that did not pass --heapcheck always):
+      if ($total != 0) {
+        printf("Total%s: %s %s\n",
+               (defined($thread) ? " (t$thread)" : ""),
+               Unparse($total), Units());
+      }
+      PrintText($symbols, $flat, $cumulative, -1);
+    } elsif ($main::opt_raw) {
+      PrintSymbolizedProfile($symbols, $profile, $main::prog);
+    } elsif ($main::opt_callgrind) {
+      PrintCallgrind($calls);
+    } else {
+      if (PrintDot($main::prog, $symbols, $profile, $flat, $cumulative, $total)) {
+        if ($main::opt_gv) {
+          RunGV(TempName($main::next_tmpfile, "ps"), "");
+        } elsif ($main::opt_evince) {
+          RunEvince(TempName($main::next_tmpfile, "pdf"), "");
+        } elsif ($main::opt_web) {
+          my $tmp = TempName($main::next_tmpfile, "svg");
+          RunWeb($tmp);
+          # The command we run might hand the file name off
+          # to an already running browser instance and then exit.
+          # Normally, we'd remove $tmp on exit (right now),
+          # but fork a child to remove $tmp a little later, so that the
+          # browser has time to load it first.
+          delete $main::tempnames{$tmp};
+          if (fork() == 0) {
+            sleep 5;
+            unlink($tmp);
+            exit(0);
+          }
+        }
+      } else {
+        cleanup();
+        exit(1);
+      }
+    }
+  } else {
+    InteractiveMode($profile, $symbols, $libs, $total);
+  }
+}
+
+sub Main() {
+  Init();
+  $main::collected_profile = undef;
+  @main::profile_files = ();
+  $main::op_time = time();
+
+  # Printing symbols is special and requires a lot less info that most.
+  if ($main::opt_symbols) {
+    PrintSymbols(*STDIN);   # Get /proc/maps and symbols output from stdin
+    return;
+  }
+
+  # Fetch all profile data
+  FetchDynamicProfiles();
+
+  # this will hold symbols that we read from the profile files
+  my $symbol_map = {};
+
+  # Read one profile, pick the last item on the list
+  my $data = ReadProfile($main::prog, pop(@main::profile_files));
+  my $profile = $data->{profile};
+  my $pcs = $data->{pcs};
+  my $libs = $data->{libs};   # Info about main program and shared libraries
+  $symbol_map = MergeSymbols($symbol_map, $data->{symbols});
+
+  # Add additional profiles, if available.
+  if (scalar(@main::profile_files) > 0) {
+    foreach my $pname (@main::profile_files) {
+      my $data2 = ReadProfile($main::prog, $pname);
+      $profile = AddProfile($profile, $data2->{profile});
+      $pcs = AddPcs($pcs, $data2->{pcs});
+      $symbol_map = MergeSymbols($symbol_map, $data2->{symbols});
+    }
+  }
+
+  # Subtract base from profile, if specified
+  if ($main::opt_base ne '') {
+    my $base = ReadProfile($main::prog, $main::opt_base);
+    $profile = SubtractProfile($profile, $base->{profile});
+    $pcs = AddPcs($pcs, $base->{pcs});
+    $symbol_map = MergeSymbols($symbol_map, $base->{symbols});
+  }
+
+  # Collect symbols
+  my $symbols;
+  if ($main::use_symbolized_profile) {
+    $symbols = FetchSymbols($pcs, $symbol_map);
+  } elsif ($main::use_symbol_page) {
+    $symbols = FetchSymbols($pcs);
+  } else {
+    # TODO(csilvers): $libs uses the /proc/self/maps data from profile1,
+    # which may differ from the data from subsequent profiles, especially
+    # if they were run on different machines.  Use appropriate libs for
+    # each pc somehow.
+    $symbols = ExtractSymbols($libs, $pcs);
+  }
+
+  if (!defined($main::opt_thread)) {
+    FilterAndPrint($profile, $symbols, $libs);
+  }
+  if (defined($data->{threads})) {
+    foreach my $thread (sort { $a <=> $b } keys(%{$data->{threads}})) {
+      if (defined($main::opt_thread) &&
+          ($main::opt_thread eq '*' || $main::opt_thread == $thread)) {
+        my $thread_profile = $data->{threads}{$thread};
+        FilterAndPrint($thread_profile, $symbols, $libs, $thread);
+      }
+    }
+  }
+
+  cleanup();
+  exit(0);
+}
+
+##### Entry Point #####
+
+Main();
+
+# Temporary code to detect if we're running on a Goobuntu system.
+# These systems don't have the right stuff installed for the special
+# Readline libraries to work, so as a temporary workaround, we default
+# to using the normal stdio code, rather than the fancier readline-based
+# code
+sub ReadlineMightFail {
+  if (-e '/lib/libtermcap.so.2') {
+    return 0;  # libtermcap exists, so readline should be okay
+  } else {
+    return 1;
+  }
+}
+
+sub RunGV {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  if (!system(ShellEscape(@GV, "--version") . " >$dev_null 2>&1")) {
+    # Options using double dash are supported by this gv version.
+    # Also, turn on noantialias to better handle bug in gv for
+    # postscript files with large dimensions.
+    # TODO: Maybe we should not pass the --noantialias flag
+    # if the gv version is known to work properly without the flag.
+    system(ShellEscape(@GV, "--scale=$main::opt_scale", "--noantialias", $fname)
+           . $bg);
+  } else {
+    # Old gv version - only supports options that use single dash.
+    print STDERR ShellEscape(@GV, "-scale", $main::opt_scale) . "\n";
+    system(ShellEscape(@GV, "-scale", "$main::opt_scale", $fname) . $bg);
+  }
+}
+
+sub RunEvince {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  system(ShellEscape(@EVINCE, $fname) . $bg);
+}
+
+sub RunWeb {
+  my $fname = shift;
+  print STDERR "Loading web page file:///$fname\n";
+
+  if (`uname` =~ /Darwin/) {
+    # OS X: open will use standard preference for SVG files.
+    system("/usr/bin/open", $fname);
+    return;
+  }
+
+  # Some kind of Unix; try generic symlinks, then specific browsers.
+  # (Stop once we find one.)
+  # Works best if the browser is already running.
+  my @alt = (
+    "/etc/alternatives/gnome-www-browser",
+    "/etc/alternatives/x-www-browser",
+    "google-chrome",
+    "firefox",
+  );
+  foreach my $b (@alt) {
+    if (system($b, $fname) == 0) {
+      return;
+    }
+  }
+
+  print STDERR "Could not load web browser.\n";
+}
+
+sub RunKcachegrind {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  print STDERR "Starting '@KCACHEGRIND " . $fname . $bg . "'\n";
+  system(ShellEscape(@KCACHEGRIND, $fname) . $bg);
+}
+
+
+##### Interactive helper routines #####
+
+sub InteractiveMode {
+  $| = 1;  # Make output unbuffered for interactive mode
+  my ($orig_profile, $symbols, $libs, $total) = @_;
+
+  print STDERR "Welcome to jeprof!  For help, type 'help'.\n";
+
+  # Use ReadLine if it's installed and input comes from a console.
+  if ( -t STDIN &&
+       !ReadlineMightFail() &&
+       defined(eval {require Term::ReadLine}) ) {
+    my $term = new Term::ReadLine 'jeprof';
+    while ( defined ($_ = $term->readline('(jeprof) '))) {
+      $term->addhistory($_) if /\S/;
+      if (!InteractiveCommand($orig_profile, $symbols, $libs, $total, $_)) {
+        last;    # exit when we get an interactive command to quit
+      }
+    }
+  } else {       # don't have readline
+    while (1) {
+      print STDERR "(jeprof) ";
+      $_ = <STDIN>;
+      last if ! defined $_ ;
+      s/\r//g;         # turn windows-looking lines into unix-looking lines
+
+      # Save some flags that might be reset by InteractiveCommand()
+      my $save_opt_lines = $main::opt_lines;
+
+      if (!InteractiveCommand($orig_profile, $symbols, $libs, $total, $_)) {
+        last;    # exit when we get an interactive command to quit
+      }
+
+      # Restore flags
+      $main::opt_lines = $save_opt_lines;
+    }
+  }
+}
+
+# Takes two args: orig profile, and command to run.
+# Returns 1 if we should keep going, or 0 if we were asked to quit
+sub InteractiveCommand {
+  my($orig_profile, $symbols, $libs, $total, $command) = @_;
+  $_ = $command;                # just to make future m//'s easier
+  if (!defined($_)) {
+    print STDERR "\n";
+    return 0;
+  }
+  if (m/^\s*quit/) {
+    return 0;
+  }
+  if (m/^\s*help/) {
+    InteractiveHelpMessage();
+    return 1;
+  }
+  # Clear all the mode options -- mode is controlled by "$command"
+  $main::opt_text = 0;
+  $main::opt_callgrind = 0;
+  $main::opt_disasm = 0;
+  $main::opt_list = 0;
+  $main::opt_gv = 0;
+  $main::opt_evince = 0;
+  $main::opt_cum = 0;
+
+  if (m/^\s*(text|top)(\d*)\s*(.*)/) {
+    $main::opt_text = 1;
+
+    my $line_limit = ($2 ne "") ? int($2) : 10;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($3);
+
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintText($symbols, $flat, $cumulative, $line_limit);
+    return 1;
+  }
+  if (m/^\s*callgrind\s*([^ \n]*)/) {
+    $main::opt_callgrind = 1;
+
+    # Get derived profiles
+    my $calls = ExtractCalls($symbols, $orig_profile);
+    my $filename = $1;
+    if ( $1 eq '' ) {
+      $filename = TempName($main::next_tmpfile, "callgrind");
+    }
+    PrintCallgrind($calls, $filename);
+    if ( $1 eq '' ) {
+      RunKcachegrind($filename, " & ");
+      $main::next_tmpfile++;
+    }
+
+    return 1;
+  }
+  if (m/^\s*(web)?list\s*(.+)/) {
+    my $html = (defined($1) && ($1 eq "web"));
+    $main::opt_list = 1;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($2);
+
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintListing($total, $libs, $flat, $cumulative, $routine, $html);
+    return 1;
+  }
+  if (m/^\s*disasm\s*(.+)/) {
+    $main::opt_disasm = 1;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($1);
+
+    # Process current profile to account for various settings
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintDisassembly($libs, $flat, $cumulative, $routine);
+    return 1;
+  }
+  if (m/^\s*(gv|web|evince)\s*(.*)/) {
+    $main::opt_gv = 0;
+    $main::opt_evince = 0;
+    $main::opt_web = 0;
+    if ($1 eq "gv") {
+      $main::opt_gv = 1;
+    } elsif ($1 eq "evince") {
+      $main::opt_evince = 1;
+    } elsif ($1 eq "web") {
+      $main::opt_web = 1;
+    }
+
+    my $focus;
+    my $ignore;
+    ($focus, $ignore) = ParseInteractiveArgs($2);
+
+    # Process current profile to account for various settings
+    my $profile = ProcessProfile($total, $orig_profile, $symbols,
+                                 $focus, $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    if (PrintDot($main::prog, $symbols, $profile, $flat, $cumulative, $total)) {
+      if ($main::opt_gv) {
+        RunGV(TempName($main::next_tmpfile, "ps"), " &");
+      } elsif ($main::opt_evince) {
+        RunEvince(TempName($main::next_tmpfile, "pdf"), " &");
+      } elsif ($main::opt_web) {
+        RunWeb(TempName($main::next_tmpfile, "svg"));
+      }
+      $main::next_tmpfile++;
+    }
+    return 1;
+  }
+  if (m/^\s*$/) {
+    return 1;
+  }
+  print STDERR "Unknown command: try 'help'.\n";
+  return 1;
+}
+
+
+sub ProcessProfile {
+  my $total_count = shift;
+  my $orig_profile = shift;
+  my $symbols = shift;
+  my $focus = shift;
+  my $ignore = shift;
+
+  # Process current profile to account for various settings
+  my $profile = $orig_profile;
+  printf("Total: %s %s\n", Unparse($total_count), Units());
+  if ($focus ne '') {
+    $profile = FocusProfile($symbols, $profile, $focus);
+    my $focus_count = TotalProfile($profile);
+    printf("After focusing on '%s': %s %s of %s (%0.1f%%)\n",
+           $focus,
+           Unparse($focus_count), Units(),
+           Unparse($total_count), ($focus_count*100.0) / $total_count);
+  }
+  if ($ignore ne '') {
+    $profile = IgnoreProfile($symbols, $profile, $ignore);
+    my $ignore_count = TotalProfile($profile);
+    printf("After ignoring '%s': %s %s of %s (%0.1f%%)\n",
+           $ignore,
+           Unparse($ignore_count), Units(),
+           Unparse($total_count),
+           ($ignore_count*100.0) / $total_count);
+  }
+
+  return $profile;
+}
+
+sub InteractiveHelpMessage {
+  print STDERR <<ENDOFHELP;
+Interactive jeprof mode
+
+Commands:
+  gv
+  gv [focus] [-ignore1] [-ignore2]
+      Show graphical hierarchical display of current profile.  Without
+      any arguments, shows all samples in the profile.  With the optional
+      "focus" argument, restricts the samples shown to just those where
+      the "focus" regular expression matches a routine name on the stack
+      trace.
+
+  web
+  web [focus] [-ignore1] [-ignore2]
+      Like GV, but displays profile in your web browser instead of using
+      Ghostview. Works best if your web browser is already running.
+      To change the browser that gets used:
+      On Linux, set the /etc/alternatives/gnome-www-browser symlink.
+      On OS X, change the Finder association for SVG files.
+
+  list [routine_regexp] [-ignore1] [-ignore2]
+      Show source listing of routines whose names match "routine_regexp"
+
+  weblist [routine_regexp] [-ignore1] [-ignore2]
+     Displays a source listing of routines whose names match "routine_regexp"
+     in a web browser.  You can click on source lines to view the
+     corresponding disassembly.
+
+  top [--cum] [-ignore1] [-ignore2]
+  top20 [--cum] [-ignore1] [-ignore2]
+  top37 [--cum] [-ignore1] [-ignore2]
+      Show top lines ordered by flat profile count, or cumulative count
+      if --cum is specified.  If a number is present after 'top', the
+      top K routines will be shown (defaults to showing the top 10)
+
+  disasm [routine_regexp] [-ignore1] [-ignore2]
+      Show disassembly of routines whose names match "routine_regexp",
+      annotated with sample counts.
+
+  callgrind
+  callgrind [filename]
+      Generates callgrind file. If no filename is given, kcachegrind is called.
+
+  help - This listing
+  quit or ^D - End jeprof
+
+For commands that accept optional -ignore tags, samples where any routine in
+the stack trace matches the regular expression in any of the -ignore
+parameters will be ignored.
+
+Further pprof details are available at this location (or one similar):
+
+ /usr/doc/gperftools-$PPROF_VERSION/cpu_profiler.html
+ /usr/doc/gperftools-$PPROF_VERSION/heap_profiler.html
+
+ENDOFHELP
+}
+sub ParseInteractiveArgs {
+  my $args = shift;
+  my $focus = "";
+  my $ignore = "";
+  my @x = split(/ +/, $args);
+  foreach $a (@x) {
+    if ($a =~ m/^(--|-)lines$/) {
+      $main::opt_lines = 1;
+    } elsif ($a =~ m/^(--|-)cum$/) {
+      $main::opt_cum = 1;
+    } elsif ($a =~ m/^-(.*)/) {
+      $ignore .= (($ignore ne "") ? "|" : "" ) . $1;
+    } else {
+      $focus .= (($focus ne "") ? "|" : "" ) . $a;
+    }
+  }
+  if ($ignore ne "") {
+    print STDERR "Ignoring samples in call stacks that match '$ignore'\n";
+  }
+  return ($focus, $ignore);
+}
+
+##### Output code #####
+
+sub TempName {
+  my $fnum = shift;
+  my $ext = shift;
+  my $file = "$main::tmpfile_ps.$fnum.$ext";
+  $main::tempnames{$file} = 1;
+  return $file;
+}
+
+# Print profile data in packed binary format (64-bit) to standard out
+sub PrintProfileData {
+  my $profile = shift;
+
+  # print header (64-bit style)
+  # (zero) (header-size) (version) (sample-period) (zero)
+  print pack('L*', 0, 0, 3, 0, 0, 0, 1, 0, 0, 0);
+
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    if ($#addrs >= 0) {
+      my $depth = $#addrs + 1;
+      # int(foo / 2**32) is the only reliable way to get rid of bottom
+      # 32 bits on both 32- and 64-bit systems.
+      print pack('L*', $count & 0xFFFFFFFF, int($count / 2**32));
+      print pack('L*', $depth & 0xFFFFFFFF, int($depth / 2**32));
+
+      foreach my $full_addr (@addrs) {
+        my $addr = $full_addr;
+        $addr =~ s/0x0*//;  # strip off leading 0x, zeroes
+        if (length($addr) > 16) {
+          print STDERR "Invalid address in profile: $full_addr\n";
+          next;
+        }
+        my $low_addr = substr($addr, -8);       # get last 8 hex chars
+        my $high_addr = substr($addr, -16, 8);  # get up to 8 more hex chars
+        print pack('L*', hex('0x' . $low_addr), hex('0x' . $high_addr));
+      }
+    }
+  }
+}
+
+# Print symbols and profile data
+sub PrintSymbolizedProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $prog = shift;
+
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+
+  print '--- ', $symbol_marker, "\n";
+  if (defined($prog)) {
+    print 'binary=', $prog, "\n";
+  }
+  while (my ($pc, $name) = each(%{$symbols})) {
+    my $sep = ' ';
+    print '0x', $pc;
+    # We have a list of function names, which include the inlined
+    # calls.  They are separated (and terminated) by --, which is
+    # illegal in function names.
+    for (my $j = 2; $j <= $#{$name}; $j += 3) {
+      print $sep, $name->[$j];
+      $sep = '--';
+    }
+    print "\n";
+  }
+  print '---', "\n";
+
+  my $profile_marker;
+  if ($main::profile_type eq 'heap') {
+    $HEAP_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  } elsif ($main::profile_type eq 'growth') {
+    $GROWTH_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  } elsif ($main::profile_type eq 'contention') {
+    $CONTENTION_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  } else { # elsif ($main::profile_type eq 'cpu')
+    $PROFILE_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  }
+
+  print '--- ', $profile_marker, "\n";
+  if (defined($main::collected_profile)) {
+    # if used with remote fetch, simply dump the collected profile to output.
+    open(SRC, "<$main::collected_profile");
+    while (<SRC>) {
+      print $_;
+    }
+    close(SRC);
+  } else {
+    # --raw/http: For everything to work correctly for non-remote profiles, we
+    # would need to extend PrintProfileData() to handle all possible profile
+    # types, re-enable the code that is currently disabled in ReadCPUProfile()
+    # and FixCallerAddresses(), and remove the remote profile dumping code in
+    # the block above.
+    die "--raw/http: jeprof can only dump remote profiles for --raw\n";
+    # dump a cpu-format profile to standard out
+    PrintProfileData($profile);
+  }
+}
+
+# Print text output
+sub PrintText {
+  my $symbols = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $line_limit = shift;
+
+  my $total = TotalProfile($flat);
+
+  # Which profile to sort by?
+  my $s = $main::opt_cum ? $cumulative : $flat;
+
+  my $running_sum = 0;
+  my $lines = 0;
+  foreach my $k (sort { GetEntry($s, $b) <=> GetEntry($s, $a) || $a cmp $b }
+                 keys(%{$cumulative})) {
+    my $f = GetEntry($flat, $k);
+    my $c = GetEntry($cumulative, $k);
+    $running_sum += $f;
+
+    my $sym = $k;
+    if (exists($symbols->{$k})) {
+      $sym = $symbols->{$k}->[0] . " " . $symbols->{$k}->[1];
+      if ($main::opt_addresses) {
+        $sym = $k . " " . $sym;
+      }
+    }
+
+    if ($f != 0 || $c != 0) {
+      printf("%8s %6s %6s %8s %6s %s\n",
+             Unparse($f),
+             Percent($f, $total),
+             Percent($running_sum, $total),
+             Unparse($c),
+             Percent($c, $total),
+             $sym);
+    }
+    $lines++;
+    last if ($line_limit >= 0 && $lines >= $line_limit);
+  }
+}
+
+# Callgrind format has a compression for repeated function and file
+# names.  You show the name the first time, and just use its number
+# subsequently.  This can cut down the file to about a third or a
+# quarter of its uncompressed size.  $key and $val are the key/value
+# pair that would normally be printed by callgrind; $map is a map from
+# value to number.
+sub CompressedCGName {
+  my($key, $val, $map) = @_;
+  my $idx = $map->{$val};
+  # For very short keys, providing an index hurts rather than helps.
+  if (length($val) <= 3) {
+    return "$key=$val\n";
+  } elsif (defined($idx)) {
+    return "$key=($idx)\n";
+  } else {
+    # scalar(keys $map) gives the number of items in the map.
+    $idx = scalar(keys(%{$map})) + 1;
+    $map->{$val} = $idx;
+    return "$key=($idx) $val\n";
+  }
+}
+
+# Print the call graph in a way that's suiteable for callgrind.
+sub PrintCallgrind {
+  my $calls = shift;
+  my $filename;
+  my %filename_to_index_map;
+  my %fnname_to_index_map;
+
+  if ($main::opt_interactive) {
+    $filename = shift;
+    print STDERR "Writing callgrind file to '$filename'.\n"
+  } else {
+    $filename = "&STDOUT";
+  }
+  open(CG, ">$filename");
+  printf CG ("events: Hits\n\n");
+  foreach my $call ( map { $_->[0] }
+                     sort { $a->[1] cmp $b ->[1] ||
+                            $a->[2] <=> $b->[2] }
+                     map { /([^:]+):(\d+):([^ ]+)( -> ([^:]+):(\d+):(.+))?/;
+                           [$_, $1, $2] }
+                     keys %$calls ) {
+    my $count = int($calls->{$call});
+    $call =~ /([^:]+):(\d+):([^ ]+)( -> ([^:]+):(\d+):(.+))?/;
+    my ( $caller_file, $caller_line, $caller_function,
+         $callee_file, $callee_line, $callee_function ) =
+       ( $1, $2, $3, $5, $6, $7 );
+
+    # TODO(csilvers): for better compression, collect all the
+    # caller/callee_files and functions first, before printing
+    # anything, and only compress those referenced more than once.
+    printf CG CompressedCGName("fl", $caller_file, \%filename_to_index_map);
+    printf CG CompressedCGName("fn", $caller_function, \%fnname_to_index_map);
+    if (defined $6) {
+      printf CG CompressedCGName("cfl", $callee_file, \%filename_to_index_map);
+      printf CG CompressedCGName("cfn", $callee_function, \%fnname_to_index_map);
+      printf CG ("calls=$count $callee_line\n");
+    }
+    printf CG ("$caller_line $count\n\n");
+  }
+}
+
+# Print disassembly for all all routines that match $main::opt_disasm
+sub PrintDisassembly {
+  my $libs = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $disasm_opts = shift;
+
+  my $total = TotalProfile($flat);
+
+  foreach my $lib (@{$libs}) {
+    my $symbol_table = GetProcedureBoundaries($lib->[0], $disasm_opts);
+    my $offset = AddressSub($lib->[1], $lib->[3]);
+    foreach my $routine (sort ByName keys(%{$symbol_table})) {
+      my $start_addr = $symbol_table->{$routine}->[0];
+      my $end_addr = $symbol_table->{$routine}->[1];
+      # See if there are any samples in this routine
+      my $length = hex(AddressSub($end_addr, $start_addr));
+      my $addr = AddressAdd($start_addr, $offset);
+      for (my $i = 0; $i < $length; $i++) {
+        if (defined($cumulative->{$addr})) {
+          PrintDisassembledFunction($lib->[0], $offset,
+                                    $routine, $flat, $cumulative,
+                                    $start_addr, $end_addr, $total);
+          last;
+        }
+        $addr = AddressInc($addr);
+      }
+    }
+  }
+}
+
+# Return reference to array of tuples of the form:
+#       [start_address, filename, linenumber, instruction, limit_address]
+# E.g.,
+#       ["0x806c43d", "/foo/bar.cc", 131, "ret", "0x806c440"]
+sub Disassemble {
+  my $prog = shift;
+  my $offset = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+
+  my $objdump = $obj_tool_map{"objdump"};
+  my $cmd = ShellEscape($objdump, "-C", "-d", "-l", "--no-show-raw-insn",
+                        "--start-address=0x$start_addr",
+                        "--stop-address=0x$end_addr", $prog);
+  open(OBJDUMP, "$cmd |") || error("$cmd: $!\n");
+  my @result = ();
+  my $filename = "";
+  my $linenumber = -1;
+  my $last = ["", "", "", ""];
+  while (<OBJDUMP>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    chop;
+    if (m|\s*([^:\s]+):(\d+)\s*$|) {
+      # Location line of the form:
+      #   <filename>:<linenumber>
+      $filename = $1;
+      $linenumber = $2;
+    } elsif (m/^ +([0-9a-f]+):\s*(.*)/) {
+      # Disassembly line -- zero-extend address to full length
+      my $addr = HexExtend($1);
+      my $k = AddressAdd($addr, $offset);
+      $last->[4] = $k;   # Store ending address for previous instruction
+      $last = [$k, $filename, $linenumber, $2, $end_addr];
+      push(@result, $last);
+    }
+  }
+  close(OBJDUMP);
+  return @result;
+}
+
+# The input file should contain lines of the form /proc/maps-like
+# output (same format as expected from the profiles) or that looks
+# like hex addresses (like "0xDEADBEEF").  We will parse all
+# /proc/maps output, and for all the hex addresses, we will output
+# "short" symbol names, one per line, in the same order as the input.
+sub PrintSymbols {
+  my $maps_and_symbols_file = shift;
+
+  # ParseLibraries expects pcs to be in a set.  Fine by us...
+  my @pclist = ();   # pcs in sorted order
+  my $pcs = {};
+  my $map = "";
+  foreach my $line (<$maps_and_symbols_file>) {
+    $line =~ s/\r//g;    # turn windows-looking lines into unix-looking lines
+    if ($line =~ /\b(0x[0-9a-f]+)\b/i) {
+      push(@pclist, HexExtend($1));
+      $pcs->{$pclist[-1]} = 1;
+    } else {
+      $map .= $line;
+    }
+  }
+
+  my $libs = ParseLibraries($main::prog, $map, $pcs);
+  my $symbols = ExtractSymbols($libs, $pcs);
+
+  foreach my $pc (@pclist) {
+    # ->[0] is the shortname, ->[2] is the full name
+    print(($symbols->{$pc}->[0] || "??") . "\n");
+  }
+}
+
+
+# For sorting functions by name
+sub ByName {
+  return ShortFunctionName($a) cmp ShortFunctionName($b);
+}
+
+# Print source-listing for all all routines that match $list_opts
+sub PrintListing {
+  my $total = shift;
+  my $libs = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $list_opts = shift;
+  my $html = shift;
+
+  my $output = \*STDOUT;
+  my $fname = "";
+
+  if ($html) {
+    # Arrange to write the output to a temporary file
+    $fname = TempName($main::next_tmpfile, "html");
+    $main::next_tmpfile++;
+    if (!open(TEMP, ">$fname")) {
+      print STDERR "$fname: $!\n";
+      return;
+    }
+    $output = \*TEMP;
+    print $output HtmlListingHeader();
+    printf $output ("<div class=\"legend\">%s<br>Total: %s %s</div>\n",
+                    $main::prog, Unparse($total), Units());
+  }
+
+  my $listed = 0;
+  foreach my $lib (@{$libs}) {
+    my $symbol_table = GetProcedureBoundaries($lib->[0], $list_opts);
+    my $offset = AddressSub($lib->[1], $lib->[3]);
+    foreach my $routine (sort ByName keys(%{$symbol_table})) {
+      # Print if there are any samples in this routine
+      my $start_addr = $symbol_table->{$routine}->[0];
+      my $end_addr = $symbol_table->{$routine}->[1];
+      my $length = hex(AddressSub($end_addr, $start_addr));
+      my $addr = AddressAdd($start_addr, $offset);
+      for (my $i = 0; $i < $length; $i++) {
+        if (defined($cumulative->{$addr})) {
+          $listed += PrintSource(
+            $lib->[0], $offset,
+            $routine, $flat, $cumulative,
+            $start_addr, $end_addr,
+            $html,
+            $output);
+          last;
+        }
+        $addr = AddressInc($addr);
+      }
+    }
+  }
+
+  if ($html) {
+    if ($listed > 0) {
+      print $output HtmlListingFooter();
+      close($output);
+      RunWeb($fname);
+    } else {
+      close($output);
+      unlink($fname);
+    }
+  }
+}
+
+sub HtmlListingHeader {
+  return <<'EOF';
+<DOCTYPE html>
+<html>
+<head>
+<title>Pprof listing</title>
+<style type="text/css">
+body {
+  font-family: sans-serif;
+}
+h1 {
+  font-size: 1.5em;
+  margin-bottom: 4px;
+}
+.legend {
+  font-size: 1.25em;
+}
+.line {
+  color: #aaaaaa;
+}
+.nop {
+  color: #aaaaaa;
+}
+.unimportant {
+  color: #cccccc;
+}
+.disasmloc {
+  color: #000000;
+}
+.deadsrc {
+  cursor: pointer;
+}
+.deadsrc:hover {
+  background-color: #eeeeee;
+}
+.livesrc {
+  color: #0000ff;
+  cursor: pointer;
+}
+.livesrc:hover {
+  background-color: #eeeeee;
+}
+.asm {
+  color: #008800;
+  display: none;
+}
+</style>
+<script type="text/javascript">
+function jeprof_toggle_asm(e) {
+  var target;
+  if (!e) e = window.event;
+  if (e.target) target = e.target;
+  else if (e.srcElement) target = e.srcElement;
+
+  if (target) {
+    var asm = target.nextSibling;
+    if (asm && asm.className == "asm") {
+      asm.style.display = (asm.style.display == "block" ? "" : "block");
+      e.preventDefault();
+      return false;
+    }
+  }
+}
+</script>
+</head>
+<body>
+EOF
+}
+
+sub HtmlListingFooter {
+  return <<'EOF';
+</body>
+</html>
+EOF
+}
+
+sub HtmlEscape {
+  my $text = shift;
+  $text =~ s/&/&amp;/g;
+  $text =~ s/</&lt;/g;
+  $text =~ s/>/&gt;/g;
+  return $text;
+}
+
+# Returns the indentation of the line, if it has any non-whitespace
+# characters.  Otherwise, returns -1.
+sub Indentation {
+  my $line = shift;
+  if (m/^(\s*)\S/) {
+    return length($1);
+  } else {
+    return -1;
+  }
+}
+
+# If the symbol table contains inlining info, Disassemble() may tag an
+# instruction with a location inside an inlined function.  But for
+# source listings, we prefer to use the location in the function we
+# are listing.  So use MapToSymbols() to fetch full location
+# information for each instruction and then pick out the first
+# location from a location list (location list contains callers before
+# callees in case of inlining).
+#
+# After this routine has run, each entry in $instructions contains:
+#   [0] start address
+#   [1] filename for function we are listing
+#   [2] line number for function we are listing
+#   [3] disassembly
+#   [4] limit address
+#   [5] most specific filename (may be different from [1] due to inlining)
+#   [6] most specific line number (may be different from [2] due to inlining)
+sub GetTopLevelLineNumbers {
+  my ($lib, $offset, $instructions) = @_;
+  my $pcs = [];
+  for (my $i = 0; $i <= $#{$instructions}; $i++) {
+    push(@{$pcs}, $instructions->[$i]->[0]);
+  }
+  my $symbols = {};
+  MapToSymbols($lib, $offset, $pcs, $symbols);
+  for (my $i = 0; $i <= $#{$instructions}; $i++) {
+    my $e = $instructions->[$i];
+    push(@{$e}, $e->[1]);
+    push(@{$e}, $e->[2]);
+    my $addr = $e->[0];
+    my $sym = $symbols->{$addr};
+    if (defined($sym)) {
+      if ($#{$sym} >= 2 && $sym->[1] =~ m/^(.*):(\d+)$/) {
+        $e->[1] = $1;  # File name
+        $e->[2] = $2;  # Line number
+      }
+    }
+  }
+}
+
+# Print source-listing for one routine
+sub PrintSource {
+  my $prog = shift;
+  my $offset = shift;
+  my $routine = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+  my $html = shift;
+  my $output = shift;
+
+  # Disassemble all instructions (just to get line numbers)
+  my @instructions = Disassemble($prog, $offset, $start_addr, $end_addr);
+  GetTopLevelLineNumbers($prog, $offset, \@instructions);
+
+  # Hack 1: assume that the first source file encountered in the
+  # disassembly contains the routine
+  my $filename = undef;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    if ($instructions[$i]->[2] >= 0) {
+      $filename = $instructions[$i]->[1];
+      last;
+    }
+  }
+  if (!defined($filename)) {
+    print STDERR "no filename found in $routine\n";
+    return 0;
+  }
+
+  # Hack 2: assume that the largest line number from $filename is the
+  # end of the procedure.  This is typically safe since if P1 contains
+  # an inlined call to P2, then P2 usually occurs earlier in the
+  # source file.  If this does not work, we might have to compute a
+  # density profile or just print all regions we find.
+  my $lastline = 0;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    my $f = $instructions[$i]->[1];
+    my $l = $instructions[$i]->[2];
+    if (($f eq $filename) && ($l > $lastline)) {
+      $lastline = $l;
+    }
+  }
+
+  # Hack 3: assume the first source location from "filename" is the start of
+  # the source code.
+  my $firstline = 1;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    if ($instructions[$i]->[1] eq $filename) {
+      $firstline = $instructions[$i]->[2];
+      last;
+    }
+  }
+
+  # Hack 4: Extend last line forward until its indentation is less than
+  # the indentation we saw on $firstline
+  my $oldlastline = $lastline;
+  {
+    if (!open(FILE, "<$filename")) {
+      print STDERR "$filename: $!\n";
+      return 0;
+    }
+    my $l = 0;
+    my $first_indentation = -1;
+    while (<FILE>) {
+      s/\r//g;         # turn windows-looking lines into unix-looking lines
+      $l++;
+      my $indent = Indentation($_);
+      if ($l >= $firstline) {
+        if ($first_indentation < 0 && $indent >= 0) {
+          $first_indentation = $indent;
+          last if ($first_indentation == 0);
+        }
+      }
+      if ($l >= $lastline && $indent >= 0) {
+        if ($indent >= $first_indentation) {
+          $lastline = $l+1;
+        } else {
+          last;
+        }
+      }
+    }
+    close(FILE);
+  }
+
+  # Assign all samples to the range $firstline,$lastline,
+  # Hack 4: If an instruction does not occur in the range, its samples
+  # are moved to the next instruction that occurs in the range.
+  my $samples1 = {};        # Map from line number to flat count
+  my $samples2 = {};        # Map from line number to cumulative count
+  my $running1 = 0;         # Unassigned flat counts
+  my $running2 = 0;         # Unassigned cumulative counts
+  my $total1 = 0;           # Total flat counts
+  my $total2 = 0;           # Total cumulative counts
+  my %disasm = ();          # Map from line number to disassembly
+  my $running_disasm = "";  # Unassigned disassembly
+  my $skip_marker = "---\n";
+  if ($html) {
+    $skip_marker = "";
+    for (my $l = $firstline; $l <= $lastline; $l++) {
+      $disasm{$l} = "";
+    }
+  }
+  my $last_dis_filename = '';
+  my $last_dis_linenum = -1;
+  my $last_touched_line = -1;  # To detect gaps in disassembly for a line
+  foreach my $e (@instructions) {
+    # Add up counts for all address that fall inside this instruction
+    my $c1 = 0;
+    my $c2 = 0;
+    for (my $a = $e->[0]; $a lt $e->[4]; $a = AddressInc($a)) {
+      $c1 += GetEntry($flat, $a);
+      $c2 += GetEntry($cumulative, $a);
+    }
+
+    if ($html) {
+      my $dis = sprintf("      %6s %6s \t\t%8s: %s ",
+                        HtmlPrintNumber($c1),
+                        HtmlPrintNumber($c2),
+                        UnparseAddress($offset, $e->[0]),
+                        CleanDisassembly($e->[3]));
+
+      # Append the most specific source line associated with this instruction
+      if (length($dis) < 80) { $dis .= (' ' x (80 - length($dis))) };
+      $dis = HtmlEscape($dis);
+      my $f = $e->[5];
+      my $l = $e->[6];
+      if ($f ne $last_dis_filename) {
+        $dis .= sprintf("<span class=disasmloc>%s:%d</span>",
+                        HtmlEscape(CleanFileName($f)), $l);
+      } elsif ($l ne $last_dis_linenum) {
+        # De-emphasize the unchanged file name portion
+        $dis .= sprintf("<span class=unimportant>%s</span>" .
+                        "<span class=disasmloc>:%d</span>",
+                        HtmlEscape(CleanFileName($f)), $l);
+      } else {
+        # De-emphasize the entire location
+        $dis .= sprintf("<span class=unimportant>%s:%d</span>",
+                        HtmlEscape(CleanFileName($f)), $l);
+      }
+      $last_dis_filename = $f;
+      $last_dis_linenum = $l;
+      $running_disasm .= $dis;
+      $running_disasm .= "\n";
+    }
+
+    $running1 += $c1;
+    $running2 += $c2;
+    $total1 += $c1;
+    $total2 += $c2;
+    my $file = $e->[1];
+    my $line = $e->[2];
+    if (($file eq $filename) &&
+        ($line >= $firstline) &&
+        ($line <= $lastline)) {
+      # Assign all accumulated samples to this line
+      AddEntry($samples1, $line, $running1);
+      AddEntry($samples2, $line, $running2);
+      $running1 = 0;
+      $running2 = 0;
+      if ($html) {
+        if ($line != $last_touched_line && $disasm{$line} ne '') {
+          $disasm{$line} .= "\n";
+        }
+        $disasm{$line} .= $running_disasm;
+        $running_disasm = '';
+        $last_touched_line = $line;
+      }
+    }
+  }
+
+  # Assign any leftover samples to $lastline
+  AddEntry($samples1, $lastline, $running1);
+  AddEntry($samples2, $lastline, $running2);
+  if ($html) {
+    if ($lastline != $last_touched_line && $disasm{$lastline} ne '') {
+      $disasm{$lastline} .= "\n";
+    }
+    $disasm{$lastline} .= $running_disasm;
+  }
+
+  if ($html) {
+    printf $output (
+      "<h1>%s</h1>%s\n<pre onClick=\"jeprof_toggle_asm()\">\n" .
+      "Total:%6s %6s (flat / cumulative %s)\n",
+      HtmlEscape(ShortFunctionName($routine)),
+      HtmlEscape(CleanFileName($filename)),
+      Unparse($total1),
+      Unparse($total2),
+      Units());
+  } else {
+    printf $output (
+      "ROUTINE ====================== %s in %s\n" .
+      "%6s %6s Total %s (flat / cumulative)\n",
+      ShortFunctionName($routine),
+      CleanFileName($filename),
+      Unparse($total1),
+      Unparse($total2),
+      Units());
+  }
+  if (!open(FILE, "<$filename")) {
+    print STDERR "$filename: $!\n";
+    return 0;
+  }
+  my $l = 0;
+  while (<FILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    $l++;
+    if ($l >= $firstline - 5 &&
+        (($l <= $oldlastline + 5) || ($l <= $lastline))) {
+      chop;
+      my $text = $_;
+      if ($l == $firstline) { print $output $skip_marker; }
+      my $n1 = GetEntry($samples1, $l);
+      my $n2 = GetEntry($samples2, $l);
+      if ($html) {
+        # Emit a span that has one of the following classes:
+        #    livesrc -- has samples
+        #    deadsrc -- has disassembly, but with no samples
+        #    nop     -- has no matching disasembly
+        # Also emit an optional span containing disassembly.
+        my $dis = $disasm{$l};
+        my $asm = "";
+        if (defined($dis) && $dis ne '') {
+          $asm = "<span class=\"asm\">" . $dis . "</span>";
+        }
+        my $source_class = (($n1 + $n2 > 0)
+                            ? "livesrc"
+                            : (($asm ne "") ? "deadsrc" : "nop"));
+        printf $output (
+          "<span class=\"line\">%5d</span> " .
+          "<span class=\"%s\">%6s %6s %s</span>%s\n",
+          $l, $source_class,
+          HtmlPrintNumber($n1),
+          HtmlPrintNumber($n2),
+          HtmlEscape($text),
+          $asm);
+      } else {
+        printf $output(
+          "%6s %6s %4d: %s\n",
+          UnparseAlt($n1),
+          UnparseAlt($n2),
+          $l,
+          $text);
+      }
+      if ($l == $lastline)  { print $output $skip_marker; }
+    };
+  }
+  close(FILE);
+  if ($html) {
+    print $output "</pre>\n";
+  }
+  return 1;
+}
+
+# Return the source line for the specified file/linenumber.
+# Returns undef if not found.
+sub SourceLine {
+  my $file = shift;
+  my $line = shift;
+
+  # Look in cache
+  if (!defined($main::source_cache{$file})) {
+    if (100 < scalar keys(%main::source_cache)) {
+      # Clear the cache when it gets too big
+      $main::source_cache = ();
+    }
+
+    # Read all lines from the file
+    if (!open(FILE, "<$file")) {
+      print STDERR "$file: $!\n";
+      $main::source_cache{$file} = [];  # Cache the negative result
+      return undef;
+    }
+    my $lines = [];
+    push(@{$lines}, "");        # So we can use 1-based line numbers as indices
+    while (<FILE>) {
+      push(@{$lines}, $_);
+    }
+    close(FILE);
+
+    # Save the lines in the cache
+    $main::source_cache{$file} = $lines;
+  }
+
+  my $lines = $main::source_cache{$file};
+  if (($line < 0) || ($line > $#{$lines})) {
+    return undef;
+  } else {
+    return $lines->[$line];
+  }
+}
+
+# Print disassembly for one routine with interspersed source if available
+sub PrintDisassembledFunction {
+  my $prog = shift;
+  my $offset = shift;
+  my $routine = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+  my $total = shift;
+
+  # Disassemble all instructions
+  my @instructions = Disassemble($prog, $offset, $start_addr, $end_addr);
+
+  # Make array of counts per instruction
+  my @flat_count = ();
+  my @cum_count = ();
+  my $flat_total = 0;
+  my $cum_total = 0;
+  foreach my $e (@instructions) {
+    # Add up counts for all address that fall inside this instruction
+    my $c1 = 0;
+    my $c2 = 0;
+    for (my $a = $e->[0]; $a lt $e->[4]; $a = AddressInc($a)) {
+      $c1 += GetEntry($flat, $a);
+      $c2 += GetEntry($cumulative, $a);
+    }
+    push(@flat_count, $c1);
+    push(@cum_count, $c2);
+    $flat_total += $c1;
+    $cum_total += $c2;
+  }
+
+  # Print header with total counts
+  printf("ROUTINE ====================== %s\n" .
+         "%6s %6s %s (flat, cumulative) %.1f%% of total\n",
+         ShortFunctionName($routine),
+         Unparse($flat_total),
+         Unparse($cum_total),
+         Units(),
+         ($cum_total * 100.0) / $total);
+
+  # Process instructions in order
+  my $current_file = "";
+  for (my $i = 0; $i <= $#instructions; ) {
+    my $e = $instructions[$i];
+
+    # Print the new file name whenever we switch files
+    if ($e->[1] ne $current_file) {
+      $current_file = $e->[1];
+      my $fname = $current_file;
+      $fname =~ s|^\./||;   # Trim leading "./"
+
+      # Shorten long file names
+      if (length($fname) >= 58) {
+        $fname = "..." . substr($fname, -55);
+      }
+      printf("-------------------- %s\n", $fname);
+    }
+
+    # TODO: Compute range of lines to print together to deal with
+    # small reorderings.
+    my $first_line = $e->[2];
+    my $last_line = $first_line;
+    my %flat_sum = ();
+    my %cum_sum = ();
+    for (my $l = $first_line; $l <= $last_line; $l++) {
+      $flat_sum{$l} = 0;
+      $cum_sum{$l} = 0;
+    }
+
+    # Find run of instructions for this range of source lines
+    my $first_inst = $i;
+    while (($i <= $#instructions) &&
+           ($instructions[$i]->[2] >= $first_line) &&
+           ($instructions[$i]->[2] <= $last_line)) {
+      $e = $instructions[$i];
+      $flat_sum{$e->[2]} += $flat_count[$i];
+      $cum_sum{$e->[2]} += $cum_count[$i];
+      $i++;
+    }
+    my $last_inst = $i - 1;
+
+    # Print source lines
+    for (my $l = $first_line; $l <= $last_line; $l++) {
+      my $line = SourceLine($current_file, $l);
+      if (!defined($line)) {
+        $line = "?\n";
+        next;
+      } else {
+        $line =~ s/^\s+//;
+      }
+      printf("%6s %6s %5d: %s",
+             UnparseAlt($flat_sum{$l}),
+             UnparseAlt($cum_sum{$l}),
+             $l,
+             $line);
+    }
+
+    # Print disassembly
+    for (my $x = $first_inst; $x <= $last_inst; $x++) {
+      my $e = $instructions[$x];
+      printf("%6s %6s    %8s: %6s\n",
+             UnparseAlt($flat_count[$x]),
+             UnparseAlt($cum_count[$x]),
+             UnparseAddress($offset, $e->[0]),
+             CleanDisassembly($e->[3]));
+    }
+  }
+}
+
+# Print DOT graph
+sub PrintDot {
+  my $prog = shift;
+  my $symbols = shift;
+  my $raw = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $overall_total = shift;
+
+  # Get total
+  my $local_total = TotalProfile($flat);
+  my $nodelimit = int($main::opt_nodefraction * $local_total);
+  my $edgelimit = int($main::opt_edgefraction * $local_total);
+  my $nodecount = $main::opt_nodecount;
+
+  # Find nodes to include
+  my @list = (sort { abs(GetEntry($cumulative, $b)) <=>
+                     abs(GetEntry($cumulative, $a))
+                     || $a cmp $b }
+              keys(%{$cumulative}));
+  my $last = $nodecount - 1;
+  if ($last > $#list) {
+    $last = $#list;
+  }
+  while (($last >= 0) &&
+         (abs(GetEntry($cumulative, $list[$last])) <= $nodelimit)) {
+    $last--;
+  }
+  if ($last < 0) {
+    print STDERR "No nodes to print\n";
+    return 0;
+  }
+
+  if ($nodelimit > 0 || $edgelimit > 0) {
+    printf STDERR ("Dropping nodes with <= %s %s; edges with <= %s abs(%s)\n",
+                   Unparse($nodelimit), Units(),
+                   Unparse($edgelimit), Units());
+  }
+
+  # Open DOT output file
+  my $output;
+  my $escaped_dot = ShellEscape(@DOT);
+  my $escaped_ps2pdf = ShellEscape(@PS2PDF);
+  if ($main::opt_gv) {
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "ps"));
+    $output = "| $escaped_dot -Tps2 >$escaped_outfile";
+  } elsif ($main::opt_evince) {
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "pdf"));
+    $output = "| $escaped_dot -Tps2 | $escaped_ps2pdf - $escaped_outfile";
+  } elsif ($main::opt_ps) {
+    $output = "| $escaped_dot -Tps2";
+  } elsif ($main::opt_pdf) {
+    $output = "| $escaped_dot -Tps2 | $escaped_ps2pdf - -";
+  } elsif ($main::opt_web || $main::opt_svg) {
+    # We need to post-process the SVG, so write to a temporary file always.
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "svg"));
+    $output = "| $escaped_dot -Tsvg >$escaped_outfile";
+  } elsif ($main::opt_gif) {
+    $output = "| $escaped_dot -Tgif";
+  } else {
+    $output = ">&STDOUT";
+  }
+  open(DOT, $output) || error("$output: $!\n");
+
+  # Title
+  printf DOT ("digraph \"%s; %s %s\" {\n",
+              $prog,
+              Unparse($overall_total),
+              Units());
+  if ($main::opt_pdf) {
+    # The output is more printable if we set the page size for dot.
+    printf DOT ("size=\"8,11\"\n");
+  }
+  printf DOT ("node [width=0.375,height=0.25];\n");
+
+  # Print legend
+  printf DOT ("Legend [shape=box,fontsize=24,shape=plaintext," .
+              "label=\"%s\\l%s\\l%s\\l%s\\l%s\\l\"];\n",
+              $prog,
+              sprintf("Total %s: %s", Units(), Unparse($overall_total)),
+              sprintf("Focusing on: %s", Unparse($local_total)),
+              sprintf("Dropped nodes with <= %s abs(%s)",
+                      Unparse($nodelimit), Units()),
+              sprintf("Dropped edges with <= %s %s",
+                      Unparse($edgelimit), Units())
+              );
+
+  # Print nodes
+  my %node = ();
+  my $nextnode = 1;
+  foreach my $a (@list[0..$last]) {
+    # Pick font size
+    my $f = GetEntry($flat, $a);
+    my $c = GetEntry($cumulative, $a);
+
+    my $fs = 8;
+    if ($local_total > 0) {
+      $fs = 8 + (50.0 * sqrt(abs($f * 1.0 / $local_total)));
+    }
+
+    $node{$a} = $nextnode++;
+    my $sym = $a;
+    $sym =~ s/\s+/\\n/g;
+    $sym =~ s/::/\\n/g;
+
+    # Extra cumulative info to print for non-leaves
+    my $extra = "";
+    if ($f != $c) {
+      $extra = sprintf("\\rof %s (%s)",
+                       Unparse($c),
+                       Percent($c, $local_total));
+    }
+    my $style = "";
+    if ($main::opt_heapcheck) {
+      if ($f > 0) {
+        # make leak-causing nodes more visible (add a background)
+        $style = ",style=filled,fillcolor=gray"
+      } elsif ($f < 0) {
+        # make anti-leak-causing nodes (which almost never occur)
+        # stand out as well (triple border)
+        $style = ",peripheries=3"
+      }
+    }
+
+    printf DOT ("N%d [label=\"%s\\n%s (%s)%s\\r" .
+                "\",shape=box,fontsize=%.1f%s];\n",
+                $node{$a},
+                $sym,
+                Unparse($f),
+                Percent($f, $local_total),
+                $extra,
+                $fs,
+                $style,
+               );
+  }
+
+  # Get edges and counts per edge
+  my %edge = ();
+  my $n;
+  my $fullname_to_shortname_map = {};
+  FillFullnameToShortnameMap($symbols, $fullname_to_shortname_map);
+  foreach my $k (keys(%{$raw})) {
+    # TODO: omit low %age edges
+    $n = $raw->{$k};
+    my @translated = TranslateStack($symbols, $fullname_to_shortname_map, $k);
+    for (my $i = 1; $i <= $#translated; $i++) {
+      my $src = $translated[$i];
+      my $dst = $translated[$i-1];
+      #next if ($src eq $dst);  # Avoid self-edges?
+      if (exists($node{$src}) && exists($node{$dst})) {
+        my $edge_label = "$src\001$dst";
+        if (!exists($edge{$edge_label})) {
+          $edge{$edge_label} = 0;
+        }
+        $edge{$edge_label} += $n;
+      }
+    }
+  }
+
+  # Print edges (process in order of decreasing counts)
+  my %indegree = ();   # Number of incoming edges added per node so far
+  my %outdegree = ();  # Number of outgoing edges added per node so far
+  foreach my $e (sort { $edge{$b} <=> $edge{$a} } keys(%edge)) {
+    my @x = split(/\001/, $e);
+    $n = $edge{$e};
+
+    # Initialize degree of kept incoming and outgoing edges if necessary
+    my $src = $x[0];
+    my $dst = $x[1];
+    if (!exists($outdegree{$src})) { $outdegree{$src} = 0; }
+    if (!exists($indegree{$dst})) { $indegree{$dst} = 0; }
+
+    my $keep;
+    if ($indegree{$dst} == 0) {
+      # Keep edge if needed for reachability
+      $keep = 1;
+    } elsif (abs($n) <= $edgelimit) {
+      # Drop if we are below --edgefraction
+      $keep = 0;
+    } elsif ($outdegree{$src} >= $main::opt_maxdegree ||
+             $indegree{$dst} >= $main::opt_maxdegree) {
+      # Keep limited number of in/out edges per node
+      $keep = 0;
+    } else {
+      $keep = 1;
+    }
+
+    if ($keep) {
+      $outdegree{$src}++;
+      $indegree{$dst}++;
+
+      # Compute line width based on edge count
+      my $fraction = abs($local_total ? (3 * ($n / $local_total)) : 0);
+      if ($fraction > 1) { $fraction = 1; }
+      my $w = $fraction * 2;
+      if ($w < 1 && ($main::opt_web || $main::opt_svg)) {
+        # SVG output treats line widths < 1 poorly.
+        $w = 1;
+      }
+
+      # Dot sometimes segfaults if given edge weights that are too large, so
+      # we cap the weights at a large value
+      my $edgeweight = abs($n) ** 0.7;
+      if ($edgeweight > 100000) { $edgeweight = 100000; }
+      $edgeweight = int($edgeweight);
+
+      my $style = sprintf("setlinewidth(%f)", $w);
+      if ($x[1] =~ m/\(inline\)/) {
+        $style .= ",dashed";
+      }
+
+      # Use a slightly squashed function of the edge count as the weight
+      printf DOT ("N%s -> N%s [label=%s, weight=%d, style=\"%s\"];\n",
+                  $node{$x[0]},
+                  $node{$x[1]},
+                  Unparse($n),
+                  $edgeweight,
+                  $style);
+    }
+  }
+
+  print DOT ("}\n");
+  close(DOT);
+
+  if ($main::opt_web || $main::opt_svg) {
+    # Rewrite SVG to be more usable inside web browser.
+    RewriteSvg(TempName($main::next_tmpfile, "svg"));
+  }
+
+  return 1;
+}
+
+sub RewriteSvg {
+  my $svgfile = shift;
+
+  open(SVG, $svgfile) || die "open temp svg: $!";
+  my @svg = <SVG>;
+  close(SVG);
+  unlink $svgfile;
+  my $svg = join('', @svg);
+
+  # Dot's SVG output is
+  #
+  #    <svg width="___" height="___"
+  #     viewBox="___" xmlns=...>
+  #    <g id="graph0" transform="...">
+  #    ...
+  #    </g>
+  #    </svg>
+  #
+  # Change it to
+  #
+  #    <svg width="100%" height="100%"
+  #     xmlns=...>
+  #    $svg_javascript
+  #    <g id="viewport" transform="translate(0,0)">
+  #    <g id="graph0" transform="...">
+  #    ...
+  #    </g>
+  #    </g>
+  #    </svg>
+
+  # Fix width, height; drop viewBox.
+  $svg =~ s/(?s)<svg width="[^"]+" height="[^"]+"(.*?)viewBox="[^"]+"/<svg width="100%" height="100%"$1/;
+
+  # Insert script, viewport <g> above first <g>
+  my $svg_javascript = SvgJavascript();
+  my $viewport = "<g id=\"viewport\" transform=\"translate(0,0)\">\n";
+  $svg =~ s/<g id="graph\d"/$svg_javascript$viewport$&/;
+
+  # Insert final </g> above </svg>.
+  $svg =~ s/(.*)(<\/svg>)/$1<\/g>$2/;
+  $svg =~ s/<g id="graph\d"(.*?)/<g id="viewport"$1/;
+
+  if ($main::opt_svg) {
+    # --svg: write to standard output.
+    print $svg;
+  } else {
+    # Write back to temporary file.
+    open(SVG, ">$svgfile") || die "open $svgfile: $!";
+    print SVG $svg;
+    close(SVG);
+  }
+}
+
+sub SvgJavascript {
+  return <<'EOF';
+<script type="text/ecmascript"><![CDATA[
+// SVGPan
+// http://www.cyberz.org/blog/2009/12/08/svgpan-a-javascript-svg-panzoomdrag-library/
+// Local modification: if(true || ...) below to force panning, never moving.
+
+/**
+ *  SVGPan library 1.2
+ * ====================
+ *
+ * Given an unique existing element with id "viewport", including the
+ * the library into any SVG adds the following capabilities:
+ *
+ *  - Mouse panning
+ *  - Mouse zooming (using the wheel)
+ *  - Object dargging
+ *
+ * Known issues:
+ *
+ *  - Zooming (while panning) on Safari has still some issues
+ *
+ * Releases:
+ *
+ * 1.2, Sat Mar 20 08:42:50 GMT 2010, Zeng Xiaohui
+ *	Fixed a bug with browser mouse handler interaction
+ *
+ * 1.1, Wed Feb  3 17:39:33 GMT 2010, Zeng Xiaohui
+ *	Updated the zoom code to support the mouse wheel on Safari/Chrome
+ *
+ * 1.0, Andrea Leofreddi
+ *	First release
+ *
+ * This code is licensed under the following BSD license:
+ *
+ * Copyright 2009-2010 Andrea Leofreddi <a.leofreddi@itcharm.com>. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without modification, are
+ * permitted provided that the following conditions are met:
+ *
+ *    1. Redistributions of source code must retain the above copyright notice, this list of
+ *       conditions and the following disclaimer.
+ *
+ *    2. Redistributions in binary form must reproduce the above copyright notice, this list
+ *       of conditions and the following disclaimer in the documentation and/or other materials
+ *       provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Andrea Leofreddi ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL Andrea Leofreddi OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * The views and conclusions contained in the software and documentation are those of the
+ * authors and should not be interpreted as representing official policies, either expressed
+ * or implied, of Andrea Leofreddi.
+ */
+
+var root = document.documentElement;
+
+var state = 'none', stateTarget, stateOrigin, stateTf;
+
+setupHandlers(root);
+
+/**
+ * Register handlers
+ */
+function setupHandlers(root){
+	setAttributes(root, {
+		"onmouseup" : "add(evt)",
+		"onmousedown" : "handleMouseDown(evt)",
+		"onmousemove" : "handleMouseMove(evt)",
+		"onmouseup" : "handleMouseUp(evt)",
+		//"onmouseout" : "handleMouseUp(evt)", // Decomment this to stop the pan functionality when dragging out of the SVG element
+	});
+
+	if(navigator.userAgent.toLowerCase().indexOf('webkit') >= 0)
+		window.addEventListener('mousewheel', handleMouseWheel, false); // Chrome/Safari
+	else
+		window.addEventListener('DOMMouseScroll', handleMouseWheel, false); // Others
+
+	var g = svgDoc.getElementById("svg");
+	g.width = "100%";
+	g.height = "100%";
+}
+
+/**
+ * Instance an SVGPoint object with given event coordinates.
+ */
+function getEventPoint(evt) {
+	var p = root.createSVGPoint();
+
+	p.x = evt.clientX;
+	p.y = evt.clientY;
+
+	return p;
+}
+
+/**
+ * Sets the current transform matrix of an element.
+ */
+function setCTM(element, matrix) {
+	var s = "matrix(" + matrix.a + "," + matrix.b + "," + matrix.c + "," + matrix.d + "," + matrix.e + "," + matrix.f + ")";
+
+	element.setAttribute("transform", s);
+}
+
+/**
+ * Dumps a matrix to a string (useful for debug).
+ */
+function dumpMatrix(matrix) {
+	var s = "[ " + matrix.a + ", " + matrix.c + ", " + matrix.e + "\n  " + matrix.b + ", " + matrix.d + ", " + matrix.f + "\n  0, 0, 1 ]";
+
+	return s;
+}
+
+/**
+ * Sets attributes of an element.
+ */
+function setAttributes(element, attributes){
+	for (i in attributes)
+		element.setAttributeNS(null, i, attributes[i]);
+}
+
+/**
+ * Handle mouse move event.
+ */
+function handleMouseWheel(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var delta;
+
+	if(evt.wheelDelta)
+		delta = evt.wheelDelta / 3600; // Chrome/Safari
+	else
+		delta = evt.detail / -90; // Mozilla
+
+	var z = 1 + delta; // Zoom factor: 0.9/1.1
+
+	var g = svgDoc.getElementById("viewport");
+
+	var p = getEventPoint(evt);
+
+	p = p.matrixTransform(g.getCTM().inverse());
+
+	// Compute new scale matrix in current mouse position
+	var k = root.createSVGMatrix().translate(p.x, p.y).scale(z).translate(-p.x, -p.y);
+
+        setCTM(g, g.getCTM().multiply(k));
+
+	stateTf = stateTf.multiply(k.inverse());
+}
+
+/**
+ * Handle mouse move event.
+ */
+function handleMouseMove(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var g = svgDoc.getElementById("viewport");
+
+	if(state == 'pan') {
+		// Pan mode
+		var p = getEventPoint(evt).matrixTransform(stateTf);
+
+		setCTM(g, stateTf.inverse().translate(p.x - stateOrigin.x, p.y - stateOrigin.y));
+	} else if(state == 'move') {
+		// Move mode
+		var p = getEventPoint(evt).matrixTransform(g.getCTM().inverse());
+
+		setCTM(stateTarget, root.createSVGMatrix().translate(p.x - stateOrigin.x, p.y - stateOrigin.y).multiply(g.getCTM().inverse()).multiply(stateTarget.getCTM()));
+
+		stateOrigin = p;
+	}
+}
+
+/**
+ * Handle click event.
+ */
+function handleMouseDown(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var g = svgDoc.getElementById("viewport");
+
+	if(true || evt.target.tagName == "svg") {
+		// Pan mode
+		state = 'pan';
+
+		stateTf = g.getCTM().inverse();
+
+		stateOrigin = getEventPoint(evt).matrixTransform(stateTf);
+	} else {
+		// Move mode
+		state = 'move';
+
+		stateTarget = evt.target;
+
+		stateTf = g.getCTM().inverse();
+
+		stateOrigin = getEventPoint(evt).matrixTransform(stateTf);
+	}
+}
+
+/**
+ * Handle mouse button release event.
+ */
+function handleMouseUp(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	if(state == 'pan' || state == 'move') {
+		// Quit pan mode
+		state = '';
+	}
+}
+
+]]></script>
+EOF
+}
+
+# Provides a map from fullname to shortname for cases where the
+# shortname is ambiguous.  The symlist has both the fullname and
+# shortname for all symbols, which is usually fine, but sometimes --
+# such as overloaded functions -- two different fullnames can map to
+# the same shortname.  In that case, we use the address of the
+# function to disambiguate the two.  This function fills in a map that
+# maps fullnames to modified shortnames in such cases.  If a fullname
+# is not present in the map, the 'normal' shortname provided by the
+# symlist is the appropriate one to use.
+sub FillFullnameToShortnameMap {
+  my $symbols = shift;
+  my $fullname_to_shortname_map = shift;
+  my $shortnames_seen_once = {};
+  my $shortnames_seen_more_than_once = {};
+
+  foreach my $symlist (values(%{$symbols})) {
+    # TODO(csilvers): deal with inlined symbols too.
+    my $shortname = $symlist->[0];
+    my $fullname = $symlist->[2];
+    if ($fullname !~ /<[0-9a-fA-F]+>$/) {  # fullname doesn't end in an address
+      next;       # the only collisions we care about are when addresses differ
+    }
+    if (defined($shortnames_seen_once->{$shortname}) &&
+        $shortnames_seen_once->{$shortname} ne $fullname) {
+      $shortnames_seen_more_than_once->{$shortname} = 1;
+    } else {
+      $shortnames_seen_once->{$shortname} = $fullname;
+    }
+  }
+
+  foreach my $symlist (values(%{$symbols})) {
+    my $shortname = $symlist->[0];
+    my $fullname = $symlist->[2];
+    # TODO(csilvers): take in a list of addresses we care about, and only
+    # store in the map if $symlist->[1] is in that list.  Saves space.
+    next if defined($fullname_to_shortname_map->{$fullname});
+    if (defined($shortnames_seen_more_than_once->{$shortname})) {
+      if ($fullname =~ /<0*([^>]*)>$/) {   # fullname has address at end of it
+        $fullname_to_shortname_map->{$fullname} = "$shortname\@$1";
+      }
+    }
+  }
+}
+
+# Return a small number that identifies the argument.
+# Multiple calls with the same argument will return the same number.
+# Calls with different arguments will return different numbers.
+sub ShortIdFor {
+  my $key = shift;
+  my $id = $main::uniqueid{$key};
+  if (!defined($id)) {
+    $id = keys(%main::uniqueid) + 1;
+    $main::uniqueid{$key} = $id;
+  }
+  return $id;
+}
+
+# Translate a stack of addresses into a stack of symbols
+sub TranslateStack {
+  my $symbols = shift;
+  my $fullname_to_shortname_map = shift;
+  my $k = shift;
+
+  my @addrs = split(/\n/, $k);
+  my @result = ();
+  for (my $i = 0; $i <= $#addrs; $i++) {
+    my $a = $addrs[$i];
+
+    # Skip large addresses since they sometimes show up as fake entries on RH9
+    if (length($a) > 8 && $a gt "7fffffffffffffff") {
+      next;
+    }
+
+    if ($main::opt_disasm || $main::opt_list) {
+      # We want just the address for the key
+      push(@result, $a);
+      next;
+    }
+
+    my $symlist = $symbols->{$a};
+    if (!defined($symlist)) {
+      $symlist = [$a, "", $a];
+    }
+
+    # We can have a sequence of symbols for a particular entry
+    # (more than one symbol in the case of inlining).  Callers
+    # come before callees in symlist, so walk backwards since
+    # the translated stack should contain callees before callers.
+    for (my $j = $#{$symlist}; $j >= 2; $j -= 3) {
+      my $func = $symlist->[$j-2];
+      my $fileline = $symlist->[$j-1];
+      my $fullfunc = $symlist->[$j];
+      if (defined($fullname_to_shortname_map->{$fullfunc})) {
+        $func = $fullname_to_shortname_map->{$fullfunc};
+      }
+      if ($j > 2) {
+        $func = "$func (inline)";
+      }
+
+      # Do not merge nodes corresponding to Callback::Run since that
+      # causes confusing cycles in dot display.  Instead, we synthesize
+      # a unique name for this frame per caller.
+      if ($func =~ m/Callback.*::Run$/) {
+        my $caller = ($i > 0) ? $addrs[$i-1] : 0;
+        $func = "Run#" . ShortIdFor($caller);
+      }
+
+      if ($main::opt_addresses) {
+        push(@result, "$a $func $fileline");
+      } elsif ($main::opt_lines) {
+        if ($func eq '??' && $fileline eq '??:0') {
+          push(@result, "$a");
+        } else {
+          push(@result, "$func $fileline");
+        }
+      } elsif ($main::opt_functions) {
+        if ($func eq '??') {
+          push(@result, "$a");
+        } else {
+          push(@result, $func);
+        }
+      } elsif ($main::opt_files) {
+        if ($fileline eq '??:0' || $fileline eq '') {
+          push(@result, "$a");
+        } else {
+          my $f = $fileline;
+          $f =~ s/:\d+$//;
+          push(@result, $f);
+        }
+      } else {
+        push(@result, $a);
+        last;  # Do not print inlined info
+      }
+    }
+  }
+
+  # print join(",", @addrs), " => ", join(",", @result), "\n";
+  return @result;
+}
+
+# Generate percent string for a number and a total
+sub Percent {
+  my $num = shift;
+  my $tot = shift;
+  if ($tot != 0) {
+    return sprintf("%.1f%%", $num * 100.0 / $tot);
+  } else {
+    return ($num == 0) ? "nan" : (($num > 0) ? "+inf" : "-inf");
+  }
+}
+
+# Generate pretty-printed form of number
+sub Unparse {
+  my $num = shift;
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    if ($main::opt_inuse_objects || $main::opt_alloc_objects) {
+      return sprintf("%d", $num);
+    } else {
+      if ($main::opt_show_bytes) {
+        return sprintf("%d", $num);
+      } else {
+        return sprintf("%.1f", $num / 1048576.0);
+      }
+    }
+  } elsif ($main::profile_type eq 'contention' && !$main::opt_contentions) {
+    return sprintf("%.3f", $num / 1e9); # Convert nanoseconds to seconds
+  } else {
+    return sprintf("%d", $num);
+  }
+}
+
+# Alternate pretty-printed form: 0 maps to "."
+sub UnparseAlt {
+  my $num = shift;
+  if ($num == 0) {
+    return ".";
+  } else {
+    return Unparse($num);
+  }
+}
+
+# Alternate pretty-printed form: 0 maps to ""
+sub HtmlPrintNumber {
+  my $num = shift;
+  if ($num == 0) {
+    return "";
+  } else {
+    return Unparse($num);
+  }
+}
+
+# Return output units
+sub Units {
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    if ($main::opt_inuse_objects || $main::opt_alloc_objects) {
+      return "objects";
+    } else {
+      if ($main::opt_show_bytes) {
+        return "B";
+      } else {
+        return "MB";
+      }
+    }
+  } elsif ($main::profile_type eq 'contention' && !$main::opt_contentions) {
+    return "seconds";
+  } else {
+    return "samples";
+  }
+}
+
+##### Profile manipulation code #####
+
+# Generate flattened profile:
+# If count is charged to stack [a,b,c,d], in generated profile,
+# it will be charged to [a]
+sub FlatProfile {
+  my $profile = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    if ($#addrs >= 0) {
+      AddEntry($result, $addrs[0], $count);
+    }
+  }
+  return $result;
+}
+
+# Generate cumulative profile:
+# If count is charged to stack [a,b,c,d], in generated profile,
+# it will be charged to [a], [b], [c], [d]
+sub CumulativeProfile {
+  my $profile = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    foreach my $a (@addrs) {
+      AddEntry($result, $a, $count);
+    }
+  }
+  return $result;
+}
+
+# If the second-youngest PC on the stack is always the same, returns
+# that pc.  Otherwise, returns undef.
+sub IsSecondPcAlwaysTheSame {
+  my $profile = shift;
+
+  my $second_pc = undef;
+  foreach my $k (keys(%{$profile})) {
+    my @addrs = split(/\n/, $k);
+    if ($#addrs < 1) {
+      return undef;
+    }
+    if (not defined $second_pc) {
+      $second_pc = $addrs[1];
+    } else {
+      if ($second_pc ne $addrs[1]) {
+        return undef;
+      }
+    }
+  }
+  return $second_pc;
+}
+
+sub ExtractSymbolLocation {
+  my $symbols = shift;
+  my $address = shift;
+  # 'addr2line' outputs "??:0" for unknown locations; we do the
+  # same to be consistent.
+  my $location = "??:0:unknown";
+  if (exists $symbols->{$address}) {
+    my $file = $symbols->{$address}->[1];
+    if ($file eq "?") {
+      $file = "??:0"
+    }
+    $location = $file . ":" . $symbols->{$address}->[0];
+  }
+  return $location;
+}
+
+# Extracts a graph of calls.
+sub ExtractCalls {
+  my $symbols = shift;
+  my $profile = shift;
+
+  my $calls = {};
+  while( my ($stack_trace, $count) = each %$profile ) {
+    my @address = split(/\n/, $stack_trace);
+    my $destination = ExtractSymbolLocation($symbols, $address[0]);
+    AddEntry($calls, $destination, $count);
+    for (my $i = 1; $i <= $#address; $i++) {
+      my $source = ExtractSymbolLocation($symbols, $address[$i]);
+      my $call = "$source -> $destination";
+      AddEntry($calls, $call, $count);
+      $destination = $source;
+    }
+  }
+
+  return $calls;
+}
+
+sub FilterFrames {
+  my $symbols = shift;
+  my $profile = shift;
+
+  if ($main::opt_retain eq '' && $main::opt_exclude eq '') {
+    return $profile;
+  }
+
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my @path = ();
+    foreach my $a (@addrs) {
+      my $sym;
+      if (exists($symbols->{$a})) {
+        $sym = $symbols->{$a}->[0];
+      } else {
+        $sym = $a;
+      }
+      if ($main::opt_retain ne '' && $sym !~ m/$main::opt_retain/) {
+        next;
+      }
+      if ($main::opt_exclude ne '' && $sym =~ m/$main::opt_exclude/) {
+        next;
+      }
+      push(@path, $a);
+    }
+    if (scalar(@path) > 0) {
+      my $reduced_path = join("\n", @path);
+      AddEntry($result, $reduced_path, $count);
+    }
+  }
+
+  return $result;
+}
+
+sub RemoveUninterestingFrames {
+  my $symbols = shift;
+  my $profile = shift;
+
+  # List of function names to skip
+  my %skip = ();
+  my $skip_regexp = 'NOMATCH';
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    foreach my $name ('@JEMALLOC_PREFIX@calloc',
+                      'cfree',
+                      '@JEMALLOC_PREFIX@malloc',
+                      'newImpl',
+                      'void* newImpl',
+                      '@JEMALLOC_PREFIX@free',
+                      '@JEMALLOC_PREFIX@memalign',
+                      '@JEMALLOC_PREFIX@posix_memalign',
+                      '@JEMALLOC_PREFIX@aligned_alloc',
+                      'pvalloc',
+                      '@JEMALLOC_PREFIX@valloc',
+                      '@JEMALLOC_PREFIX@realloc',
+                      '@JEMALLOC_PREFIX@mallocx',
+                      '@JEMALLOC_PREFIX@rallocx',
+                      '@JEMALLOC_PREFIX@xallocx',
+                      '@JEMALLOC_PREFIX@dallocx',
+                      '@JEMALLOC_PREFIX@sdallocx',
+                      '@JEMALLOC_PREFIX@sdallocx_noflags',
+                      'tc_calloc',
+                      'tc_cfree',
+                      'tc_malloc',
+                      'tc_free',
+                      'tc_memalign',
+                      'tc_posix_memalign',
+                      'tc_pvalloc',
+                      'tc_valloc',
+                      'tc_realloc',
+                      'tc_new',
+                      'tc_delete',
+                      'tc_newarray',
+                      'tc_deletearray',
+                      'tc_new_nothrow',
+                      'tc_newarray_nothrow',
+                      'do_malloc',
+                      '::do_malloc',   # new name -- got moved to an unnamed ns
+                      '::do_malloc_or_cpp_alloc',
+                      'DoSampledAllocation',
+                      'simple_alloc::allocate',
+                      '__malloc_alloc_template::allocate',
+                      '__builtin_delete',
+                      '__builtin_new',
+                      '__builtin_vec_delete',
+                      '__builtin_vec_new',
+                      'operator new',
+                      'operator new[]',
+                      # The entry to our memory-allocation routines on OS X
+                      'malloc_zone_malloc',
+                      'malloc_zone_calloc',
+                      'malloc_zone_valloc',
+                      'malloc_zone_realloc',
+                      'malloc_zone_memalign',
+                      'malloc_zone_free',
+                      # These mark the beginning/end of our custom sections
+                      '__start_google_malloc',
+                      '__stop_google_malloc',
+                      '__start_malloc_hook',
+                      '__stop_malloc_hook') {
+      $skip{$name} = 1;
+      $skip{"_" . $name} = 1;   # Mach (OS X) adds a _ prefix to everything
+    }
+    # TODO: Remove TCMalloc once everything has been
+    # moved into the tcmalloc:: namespace and we have flushed
+    # old code out of the system.
+    $skip_regexp = "TCMalloc|^tcmalloc::";
+  } elsif ($main::profile_type eq 'contention') {
+    foreach my $vname ('base::RecordLockProfileData',
+                       'base::SubmitMutexProfileData',
+                       'base::SubmitSpinLockProfileData',
+                       'Mutex::Unlock',
+                       'Mutex::UnlockSlow',
+                       'Mutex::ReaderUnlock',
+                       'MutexLock::~MutexLock',
+                       'SpinLock::Unlock',
+                       'SpinLock::SlowUnlock',
+                       'SpinLockHolder::~SpinLockHolder') {
+      $skip{$vname} = 1;
+    }
+  } elsif ($main::profile_type eq 'cpu') {
+    # Drop signal handlers used for CPU profile collection
+    # TODO(dpeng): this should not be necessary; it's taken
+    # care of by the general 2nd-pc mechanism below.
+    foreach my $name ('ProfileData::Add',           # historical
+                      'ProfileData::prof_handler',  # historical
+                      'CpuProfiler::prof_handler',
+                      '__FRAME_END__',
+                      '__pthread_sighandler',
+                      '__restore') {
+      $skip{$name} = 1;
+    }
+  } else {
+    # Nothing skipped for unknown types
+  }
+
+  if ($main::profile_type eq 'cpu') {
+    # If all the second-youngest program counters are the same,
+    # this STRONGLY suggests that it is an artifact of measurement,
+    # i.e., stack frames pushed by the CPU profiler signal handler.
+    # Hence, we delete them.
+    # (The topmost PC is read from the signal structure, not from
+    # the stack, so it does not get involved.)
+    while (my $second_pc = IsSecondPcAlwaysTheSame($profile)) {
+      my $result = {};
+      my $func = '';
+      if (exists($symbols->{$second_pc})) {
+        $second_pc = $symbols->{$second_pc}->[0];
+      }
+      print STDERR "Removing $second_pc from all stack traces.\n";
+      foreach my $k (keys(%{$profile})) {
+        my $count = $profile->{$k};
+        my @addrs = split(/\n/, $k);
+        splice @addrs, 1, 1;
+        my $reduced_path = join("\n", @addrs);
+        AddEntry($result, $reduced_path, $count);
+      }
+      $profile = $result;
+    }
+  }
+
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my @path = ();
+    foreach my $a (@addrs) {
+      if (exists($symbols->{$a})) {
+        my $func = $symbols->{$a}->[0];
+        if ($skip{$func} || ($func =~ m/$skip_regexp/)) {
+          # Throw away the portion of the backtrace seen so far, under the
+          # assumption that previous frames were for functions internal to the
+          # allocator.
+          @path = ();
+          next;
+        }
+      }
+      push(@path, $a);
+    }
+    my $reduced_path = join("\n", @path);
+    AddEntry($result, $reduced_path, $count);
+  }
+
+  $result = FilterFrames($symbols, $result);
+
+  return $result;
+}
+
+# Reduce profile to granularity given by user
+sub ReduceProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $result = {};
+  my $fullname_to_shortname_map = {};
+  FillFullnameToShortnameMap($symbols, $fullname_to_shortname_map);
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @translated = TranslateStack($symbols, $fullname_to_shortname_map, $k);
+    my @path = ();
+    my %seen = ();
+    $seen{''} = 1;      # So that empty keys are skipped
+    foreach my $e (@translated) {
+      # To avoid double-counting due to recursion, skip a stack-trace
+      # entry if it has already been seen
+      if (!$seen{$e}) {
+        $seen{$e} = 1;
+        push(@path, $e);
+      }
+    }
+    my $reduced_path = join("\n", @path);
+    AddEntry($result, $reduced_path, $count);
+  }
+  return $result;
+}
+
+# Does the specified symbol array match the regexp?
+sub SymbolMatches {
+  my $sym = shift;
+  my $re = shift;
+  if (defined($sym)) {
+    for (my $i = 0; $i < $#{$sym}; $i += 3) {
+      if ($sym->[$i] =~ m/$re/ || $sym->[$i+1] =~ m/$re/) {
+        return 1;
+      }
+    }
+  }
+  return 0;
+}
+
+# Focus only on paths involving specified regexps
+sub FocusProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $focus = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    foreach my $a (@addrs) {
+      # Reply if it matches either the address/shortname/fileline
+      if (($a =~ m/$focus/) || SymbolMatches($symbols->{$a}, $focus)) {
+        AddEntry($result, $k, $count);
+        last;
+      }
+    }
+  }
+  return $result;
+}
+
+# Focus only on paths not involving specified regexps
+sub IgnoreProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $ignore = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my $matched = 0;
+    foreach my $a (@addrs) {
+      # Reply if it matches either the address/shortname/fileline
+      if (($a =~ m/$ignore/) || SymbolMatches($symbols->{$a}, $ignore)) {
+        $matched = 1;
+        last;
+      }
+    }
+    if (!$matched) {
+      AddEntry($result, $k, $count);
+    }
+  }
+  return $result;
+}
+
+# Get total count in profile
+sub TotalProfile {
+  my $profile = shift;
+  my $result = 0;
+  foreach my $k (keys(%{$profile})) {
+    $result += $profile->{$k};
+  }
+  return $result;
+}
+
+# Add A to B
+sub AddProfile {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  # add all keys in A
+  foreach my $k (keys(%{$A})) {
+    my $v = $A->{$k};
+    AddEntry($R, $k, $v);
+  }
+  # add all keys in B
+  foreach my $k (keys(%{$B})) {
+    my $v = $B->{$k};
+    AddEntry($R, $k, $v);
+  }
+  return $R;
+}
+
+# Merges symbol maps
+sub MergeSymbols {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  foreach my $k (keys(%{$A})) {
+    $R->{$k} = $A->{$k};
+  }
+  if (defined($B)) {
+    foreach my $k (keys(%{$B})) {
+      $R->{$k} = $B->{$k};
+    }
+  }
+  return $R;
+}
+
+
+# Add A to B
+sub AddPcs {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  # add all keys in A
+  foreach my $k (keys(%{$A})) {
+    $R->{$k} = 1
+  }
+  # add all keys in B
+  foreach my $k (keys(%{$B})) {
+    $R->{$k} = 1
+  }
+  return $R;
+}
+
+# Subtract B from A
+sub SubtractProfile {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  foreach my $k (keys(%{$A})) {
+    my $v = $A->{$k} - GetEntry($B, $k);
+    if ($v < 0 && $main::opt_drop_negative) {
+      $v = 0;
+    }
+    AddEntry($R, $k, $v);
+  }
+  if (!$main::opt_drop_negative) {
+    # Take care of when subtracted profile has more entries
+    foreach my $k (keys(%{$B})) {
+      if (!exists($A->{$k})) {
+        AddEntry($R, $k, 0 - $B->{$k});
+      }
+    }
+  }
+  return $R;
+}
+
+# Get entry from profile; zero if not present
+sub GetEntry {
+  my $profile = shift;
+  my $k = shift;
+  if (exists($profile->{$k})) {
+    return $profile->{$k};
+  } else {
+    return 0;
+  }
+}
+
+# Add entry to specified profile
+sub AddEntry {
+  my $profile = shift;
+  my $k = shift;
+  my $n = shift;
+  if (!exists($profile->{$k})) {
+    $profile->{$k} = 0;
+  }
+  $profile->{$k} += $n;
+}
+
+# Add a stack of entries to specified profile, and add them to the $pcs
+# list.
+sub AddEntries {
+  my $profile = shift;
+  my $pcs = shift;
+  my $stack = shift;
+  my $count = shift;
+  my @k = ();
+
+  foreach my $e (split(/\s+/, $stack)) {
+    my $pc = HexExtend($e);
+    $pcs->{$pc} = 1;
+    push @k, $pc;
+  }
+  AddEntry($profile, (join "\n", @k), $count);
+}
+
+##### Code to profile a server dynamically #####
+
+sub CheckSymbolPage {
+  my $url = SymbolPageURL();
+  my $command = ShellEscape(@URL_FETCHER, $url);
+  open(SYMBOL, "$command |") or error($command);
+  my $line = <SYMBOL>;
+  $line =~ s/\r//g;         # turn windows-looking lines into unix-looking lines
+  close(SYMBOL);
+  unless (defined($line)) {
+    error("$url doesn't exist\n");
+  }
+
+  if ($line =~ /^num_symbols:\s+(\d+)$/) {
+    if ($1 == 0) {
+      error("Stripped binary. No symbols available.\n");
+    }
+  } else {
+    error("Failed to get the number of symbols from $url\n");
+  }
+}
+
+sub IsProfileURL {
+  my $profile_name = shift;
+  if (-f $profile_name) {
+    printf STDERR "Using local file $profile_name.\n";
+    return 0;
+  }
+  return 1;
+}
+
+sub ParseProfileURL {
+  my $profile_name = shift;
+
+  if (!defined($profile_name) || $profile_name eq "") {
+    return ();
+  }
+
+  # Split profile URL - matches all non-empty strings, so no test.
+  $profile_name =~ m,^(https?://)?([^/]+)(.*?)(/|$PROFILES)?$,;
+
+  my $proto = $1 || "http://";
+  my $hostport = $2;
+  my $prefix = $3;
+  my $profile = $4 || "/";
+
+  my $host = $hostport;
+  $host =~ s/:.*//;
+
+  my $baseurl = "$proto$hostport$prefix";
+  return ($host, $baseurl, $profile);
+}
+
+# We fetch symbols from the first profile argument.
+sub SymbolPageURL {
+  my ($host, $baseURL, $path) = ParseProfileURL($main::pfile_args[0]);
+  return "$baseURL$SYMBOL_PAGE";
+}
+
+sub FetchProgramName() {
+  my ($host, $baseURL, $path) = ParseProfileURL($main::pfile_args[0]);
+  my $url = "$baseURL$PROGRAM_NAME_PAGE";
+  my $command_line = ShellEscape(@URL_FETCHER, $url);
+  open(CMDLINE, "$command_line |") or error($command_line);
+  my $cmdline = <CMDLINE>;
+  $cmdline =~ s/\r//g;   # turn windows-looking lines into unix-looking lines
+  close(CMDLINE);
+  error("Failed to get program name from $url\n") unless defined($cmdline);
+  $cmdline =~ s/\x00.+//;  # Remove argv[1] and latters.
+  $cmdline =~ s!\n!!g;  # Remove LFs.
+  return $cmdline;
+}
+
+# Gee, curl's -L (--location) option isn't reliable at least
+# with its 7.12.3 version.  Curl will forget to post data if
+# there is a redirection.  This function is a workaround for
+# curl.  Redirection happens on borg hosts.
+sub ResolveRedirectionForCurl {
+  my $url = shift;
+  my $command_line = ShellEscape(@URL_FETCHER, "--head", $url);
+  open(CMDLINE, "$command_line |") or error($command_line);
+  while (<CMDLINE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (/^Location: (.*)/) {
+      $url = $1;
+    }
+  }
+  close(CMDLINE);
+  return $url;
+}
+
+# Add a timeout flat to URL_FETCHER.  Returns a new list.
+sub AddFetchTimeout {
+  my $timeout = shift;
+  my @fetcher = @_;
+  if (defined($timeout)) {
+    if (join(" ", @fetcher) =~ m/\bcurl -s/) {
+      push(@fetcher, "--max-time", sprintf("%d", $timeout));
+    } elsif (join(" ", @fetcher) =~ m/\brpcget\b/) {
+      push(@fetcher, sprintf("--deadline=%d", $timeout));
+    }
+  }
+  return @fetcher;
+}
+
+# Reads a symbol map from the file handle name given as $1, returning
+# the resulting symbol map.  Also processes variables relating to symbols.
+# Currently, the only variable processed is 'binary=<value>' which updates
+# $main::prog to have the correct program name.
+sub ReadSymbols {
+  my $in = shift;
+  my $map = {};
+  while (<$in>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Removes all the leading zeroes from the symbols, see comment below.
+    if (m/^0x0*([0-9a-f]+)\s+(.+)/) {
+      $map->{$1} = $2;
+    } elsif (m/^---/) {
+      last;
+    } elsif (m/^([a-z][^=]*)=(.*)$/ ) {
+      my ($variable, $value) = ($1, $2);
+      for ($variable, $value) {
+        s/^\s+//;
+        s/\s+$//;
+      }
+      if ($variable eq "binary") {
+        if ($main::prog ne $UNKNOWN_BINARY && $main::prog ne $value) {
+          printf STDERR ("Warning: Mismatched binary name '%s', using '%s'.\n",
+                         $main::prog, $value);
+        }
+        $main::prog = $value;
+      } else {
+        printf STDERR ("Ignoring unknown variable in symbols list: " .
+            "'%s' = '%s'\n", $variable, $value);
+      }
+    }
+  }
+  return $map;
+}
+
+sub URLEncode {
+  my $str = shift;
+  $str =~ s/([^A-Za-z0-9\-_.!~*'()])/ sprintf "%%%02x", ord $1 /eg;
+  return $str;
+}
+
+sub AppendSymbolFilterParams {
+  my $url = shift;
+  my @params = ();
+  if ($main::opt_retain ne '') {
+    push(@params, sprintf("retain=%s", URLEncode($main::opt_retain)));
+  }
+  if ($main::opt_exclude ne '') {
+    push(@params, sprintf("exclude=%s", URLEncode($main::opt_exclude)));
+  }
+  if (scalar @params > 0) {
+    $url = sprintf("%s?%s", $url, join("&", @params));
+  }
+  return $url;
+}
+
+# Fetches and processes symbols to prepare them for use in the profile output
+# code.  If the optional 'symbol_map' arg is not given, fetches symbols from
+# $SYMBOL_PAGE for all PC values found in profile.  Otherwise, the raw symbols
+# are assumed to have already been fetched into 'symbol_map' and are simply
+# extracted and processed.
+sub FetchSymbols {
+  my $pcset = shift;
+  my $symbol_map = shift;
+
+  my %seen = ();
+  my @pcs = grep { !$seen{$_}++ } keys(%$pcset);  # uniq
+
+  if (!defined($symbol_map)) {
+    my $post_data = join("+", sort((map {"0x" . "$_"} @pcs)));
+
+    open(POSTFILE, ">$main::tmpfile_sym");
+    print POSTFILE $post_data;
+    close(POSTFILE);
+
+    my $url = SymbolPageURL();
+
+    my $command_line;
+    if (join(" ", @URL_FETCHER) =~ m/\bcurl -s/) {
+      $url = ResolveRedirectionForCurl($url);
+      $url = AppendSymbolFilterParams($url);
+      $command_line = ShellEscape(@URL_FETCHER, "-d", "\@$main::tmpfile_sym",
+                                  $url);
+    } else {
+      $url = AppendSymbolFilterParams($url);
+      $command_line = (ShellEscape(@URL_FETCHER, "--post", $url)
+                       . " < " . ShellEscape($main::tmpfile_sym));
+    }
+    # We use c++filt in case $SYMBOL_PAGE gives us mangled symbols.
+    my $escaped_cppfilt = ShellEscape($obj_tool_map{"c++filt"});
+    open(SYMBOL, "$command_line | $escaped_cppfilt |") or error($command_line);
+    $symbol_map = ReadSymbols(*SYMBOL{IO});
+    close(SYMBOL);
+  }
+
+  my $symbols = {};
+  foreach my $pc (@pcs) {
+    my $fullname;
+    # For 64 bits binaries, symbols are extracted with 8 leading zeroes.
+    # Then /symbol reads the long symbols in as uint64, and outputs
+    # the result with a "0x%08llx" format which get rid of the zeroes.
+    # By removing all the leading zeroes in both $pc and the symbols from
+    # /symbol, the symbols match and are retrievable from the map.
+    my $shortpc = $pc;
+    $shortpc =~ s/^0*//;
+    # Each line may have a list of names, which includes the function
+    # and also other functions it has inlined.  They are separated (in
+    # PrintSymbolizedProfile), by --, which is illegal in function names.
+    my $fullnames;
+    if (defined($symbol_map->{$shortpc})) {
+      $fullnames = $symbol_map->{$shortpc};
+    } else {
+      $fullnames = "0x" . $pc;  # Just use addresses
+    }
+    my $sym = [];
+    $symbols->{$pc} = $sym;
+    foreach my $fullname (split("--", $fullnames)) {
+      my $name = ShortFunctionName($fullname);
+      push(@{$sym}, $name, "?", $fullname);
+    }
+  }
+  return $symbols;
+}
+
+sub BaseName {
+  my $file_name = shift;
+  $file_name =~ s!^.*/!!;  # Remove directory name
+  return $file_name;
+}
+
+sub MakeProfileBaseName {
+  my ($binary_name, $profile_name) = @_;
+  my ($host, $baseURL, $path) = ParseProfileURL($profile_name);
+  my $binary_shortname = BaseName($binary_name);
+  return sprintf("%s.%s.%s",
+                 $binary_shortname, $main::op_time, $host);
+}
+
+sub FetchDynamicProfile {
+  my $binary_name = shift;
+  my $profile_name = shift;
+  my $fetch_name_only = shift;
+  my $encourage_patience = shift;
+
+  if (!IsProfileURL($profile_name)) {
+    return $profile_name;
+  } else {
+    my ($host, $baseURL, $path) = ParseProfileURL($profile_name);
+    if ($path eq "" || $path eq "/") {
+      # Missing type specifier defaults to cpu-profile
+      $path = $PROFILE_PAGE;
+    }
+
+    my $profile_file = MakeProfileBaseName($binary_name, $profile_name);
+
+    my $url = "$baseURL$path";
+    my $fetch_timeout = undef;
+    if ($path =~ m/$PROFILE_PAGE|$PMUPROFILE_PAGE/) {
+      if ($path =~ m/[?]/) {
+        $url .= "&";
+      } else {
+        $url .= "?";
+      }
+      $url .= sprintf("seconds=%d", $main::opt_seconds);
+      $fetch_timeout = $main::opt_seconds * 1.01 + 60;
+      # Set $profile_type for consumption by PrintSymbolizedProfile.
+      $main::profile_type = 'cpu';
+    } else {
+      # For non-CPU profiles, we add a type-extension to
+      # the target profile file name.
+      my $suffix = $path;
+      $suffix =~ s,/,.,g;
+      $profile_file .= $suffix;
+      # Set $profile_type for consumption by PrintSymbolizedProfile.
+      if ($path =~ m/$HEAP_PAGE/) {
+        $main::profile_type = 'heap';
+      } elsif ($path =~ m/$GROWTH_PAGE/) {
+        $main::profile_type = 'growth';
+      } elsif ($path =~ m/$CONTENTION_PAGE/) {
+        $main::profile_type = 'contention';
+      }
+    }
+
+    my $profile_dir = $ENV{"JEPROF_TMPDIR"} || ($ENV{HOME} . "/jeprof");
+    if (! -d $profile_dir) {
+      mkdir($profile_dir)
+          || die("Unable to create profile directory $profile_dir: $!\n");
+    }
+    my $tmp_profile = "$profile_dir/.tmp.$profile_file";
+    my $real_profile = "$profile_dir/$profile_file";
+
+    if ($fetch_name_only > 0) {
+      return $real_profile;
+    }
+
+    my @fetcher = AddFetchTimeout($fetch_timeout, @URL_FETCHER);
+    my $cmd = ShellEscape(@fetcher, $url) . " > " . ShellEscape($tmp_profile);
+    if ($path =~ m/$PROFILE_PAGE|$PMUPROFILE_PAGE|$CENSUSPROFILE_PAGE/){
+      print STDERR "Gathering CPU profile from $url for $main::opt_seconds seconds to\n  ${real_profile}\n";
+      if ($encourage_patience) {
+        print STDERR "Be patient...\n";
+      }
+    } else {
+      print STDERR "Fetching $path profile from $url to\n  ${real_profile}\n";
+    }
+
+    (system($cmd) == 0) || error("Failed to get profile: $cmd: $!\n");
+    (system("mv", $tmp_profile, $real_profile) == 0) || error("Unable to rename profile\n");
+    print STDERR "Wrote profile to $real_profile\n";
+    $main::collected_profile = $real_profile;
+    return $main::collected_profile;
+  }
+}
+
+# Collect profiles in parallel
+sub FetchDynamicProfiles {
+  my $items = scalar(@main::pfile_args);
+  my $levels = log($items) / log(2);
+
+  if ($items == 1) {
+    $main::profile_files[0] = FetchDynamicProfile($main::prog, $main::pfile_args[0], 0, 1);
+  } else {
+    # math rounding issues
+    if ((2 ** $levels) < $items) {
+     $levels++;
+    }
+    my $count = scalar(@main::pfile_args);
+    for (my $i = 0; $i < $count; $i++) {
+      $main::profile_files[$i] = FetchDynamicProfile($main::prog, $main::pfile_args[$i], 1, 0);
+    }
+    print STDERR "Fetching $count profiles, Be patient...\n";
+    FetchDynamicProfilesRecurse($levels, 0, 0);
+    $main::collected_profile = join(" \\\n    ", @main::profile_files);
+  }
+}
+
+# Recursively fork a process to get enough processes
+# collecting profiles
+sub FetchDynamicProfilesRecurse {
+  my $maxlevel = shift;
+  my $level = shift;
+  my $position = shift;
+
+  if (my $pid = fork()) {
+    $position = 0 | ($position << 1);
+    TryCollectProfile($maxlevel, $level, $position);
+    wait;
+  } else {
+    $position = 1 | ($position << 1);
+    TryCollectProfile($maxlevel, $level, $position);
+    cleanup();
+    exit(0);
+  }
+}
+
+# Collect a single profile
+sub TryCollectProfile {
+  my $maxlevel = shift;
+  my $level = shift;
+  my $position = shift;
+
+  if ($level >= ($maxlevel - 1)) {
+    if ($position < scalar(@main::pfile_args)) {
+      FetchDynamicProfile($main::prog, $main::pfile_args[$position], 0, 0);
+    }
+  } else {
+    FetchDynamicProfilesRecurse($maxlevel, $level+1, $position);
+  }
+}
+
+##### Parsing code #####
+
+# Provide a small streaming-read module to handle very large
+# cpu-profile files.  Stream in chunks along a sliding window.
+# Provides an interface to get one 'slot', correctly handling
+# endian-ness differences.  A slot is one 32-bit or 64-bit word
+# (depending on the input profile).  We tell endianness and bit-size
+# for the profile by looking at the first 8 bytes: in cpu profiles,
+# the second slot is always 3 (we'll accept anything that's not 0).
+BEGIN {
+  package CpuProfileStream;
+
+  sub new {
+    my ($class, $file, $fname) = @_;
+    my $self = { file        => $file,
+                 base        => 0,
+                 stride      => 512 * 1024,   # must be a multiple of bitsize/8
+                 slots       => [],
+                 unpack_code => "",           # N for big-endian, V for little
+                 perl_is_64bit => 1,          # matters if profile is 64-bit
+    };
+    bless $self, $class;
+    # Let unittests adjust the stride
+    if ($main::opt_test_stride > 0) {
+      $self->{stride} = $main::opt_test_stride;
+    }
+    # Read the first two slots to figure out bitsize and endianness.
+    my $slots = $self->{slots};
+    my $str;
+    read($self->{file}, $str, 8);
+    # Set the global $address_length based on what we see here.
+    # 8 is 32-bit (8 hexadecimal chars); 16 is 64-bit (16 hexadecimal chars).
+    $address_length = ($str eq (chr(0)x8)) ? 16 : 8;
+    if ($address_length == 8) {
+      if (substr($str, 6, 2) eq chr(0)x2) {
+        $self->{unpack_code} = 'V';  # Little-endian.
+      } elsif (substr($str, 4, 2) eq chr(0)x2) {
+        $self->{unpack_code} = 'N';  # Big-endian
+      } else {
+        ::error("$fname: header size >= 2**16\n");
+      }
+      @$slots = unpack($self->{unpack_code} . "*", $str);
+    } else {
+      # If we're a 64-bit profile, check if we're a 64-bit-capable
+      # perl.  Otherwise, each slot will be represented as a float
+      # instead of an int64, losing precision and making all the
+      # 64-bit addresses wrong.  We won't complain yet, but will
+      # later if we ever see a value that doesn't fit in 32 bits.
+      my $has_q = 0;
+      eval { $has_q = pack("Q", "1") ? 1 : 1; };
+      if (!$has_q) {
+        $self->{perl_is_64bit} = 0;
+      }
+      read($self->{file}, $str, 8);
+      if (substr($str, 4, 4) eq chr(0)x4) {
+        # We'd love to use 'Q', but it's a) not universal, b) not endian-proof.
+        $self->{unpack_code} = 'V';  # Little-endian.
+      } elsif (substr($str, 0, 4) eq chr(0)x4) {
+        $self->{unpack_code} = 'N';  # Big-endian
+      } else {
+        ::error("$fname: header size >= 2**32\n");
+      }
+      my @pair = unpack($self->{unpack_code} . "*", $str);
+      # Since we know one of the pair is 0, it's fine to just add them.
+      @$slots = (0, $pair[0] + $pair[1]);
+    }
+    return $self;
+  }
+
+  # Load more data when we access slots->get(X) which is not yet in memory.
+  sub overflow {
+    my ($self) = @_;
+    my $slots = $self->{slots};
+    $self->{base} += $#$slots + 1;   # skip over data we're replacing
+    my $str;
+    read($self->{file}, $str, $self->{stride});
+    if ($address_length == 8) {      # the 32-bit case
+      # This is the easy case: unpack provides 32-bit unpacking primitives.
+      @$slots = unpack($self->{unpack_code} . "*", $str);
+    } else {
+      # We need to unpack 32 bits at a time and combine.
+      my @b32_values = unpack($self->{unpack_code} . "*", $str);
+      my @b64_values = ();
+      for (my $i = 0; $i < $#b32_values; $i += 2) {
+        # TODO(csilvers): if this is a 32-bit perl, the math below
+        #    could end up in a too-large int, which perl will promote
+        #    to a double, losing necessary precision.  Deal with that.
+        #    Right now, we just die.
+        my ($lo, $hi) = ($b32_values[$i], $b32_values[$i+1]);
+        if ($self->{unpack_code} eq 'N') {    # big-endian
+          ($lo, $hi) = ($hi, $lo);
+        }
+        my $value = $lo + $hi * (2**32);
+        if (!$self->{perl_is_64bit} &&   # check value is exactly represented
+            (($value % (2**32)) != $lo || int($value / (2**32)) != $hi)) {
+          ::error("Need a 64-bit perl to process this 64-bit profile.\n");
+        }
+        push(@b64_values, $value);
+      }
+      @$slots = @b64_values;
+    }
+  }
+
+  # Access the i-th long in the file (logically), or -1 at EOF.
+  sub get {
+    my ($self, $idx) = @_;
+    my $slots = $self->{slots};
+    while ($#$slots >= 0) {
+      if ($idx < $self->{base}) {
+        # The only time we expect a reference to $slots[$i - something]
+        # after referencing $slots[$i] is reading the very first header.
+        # Since $stride > |header|, that shouldn't cause any lookback
+        # errors.  And everything after the header is sequential.
+        print STDERR "Unexpected look-back reading CPU profile";
+        return -1;   # shrug, don't know what better to return
+      } elsif ($idx > $self->{base} + $#$slots) {
+        $self->overflow();
+      } else {
+        return $slots->[$idx - $self->{base}];
+      }
+    }
+    # If we get here, $slots is [], which means we've reached EOF
+    return -1;  # unique since slots is supposed to hold unsigned numbers
+  }
+}
+
+# Reads the top, 'header' section of a profile, and returns the last
+# line of the header, commonly called a 'header line'.  The header
+# section of a profile consists of zero or more 'command' lines that
+# are instructions to jeprof, which jeprof executes when reading the
+# header.  All 'command' lines start with a %.  After the command
+# lines is the 'header line', which is a profile-specific line that
+# indicates what type of profile it is, and perhaps other global
+# information about the profile.  For instance, here's a header line
+# for a heap profile:
+#   heap profile:     53:    38236 [  5525:  1284029] @ heapprofile
+# For historical reasons, the CPU profile does not contain a text-
+# readable header line.  If the profile looks like a CPU profile,
+# this function returns "".  If no header line could be found, this
+# function returns undef.
+#
+# The following commands are recognized:
+#   %warn -- emit the rest of this line to stderr, prefixed by 'WARNING:'
+#
+# The input file should be in binmode.
+sub ReadProfileHeader {
+  local *PROFILE = shift;
+  my $firstchar = "";
+  my $line = "";
+  read(PROFILE, $firstchar, 1);
+  seek(PROFILE, -1, 1);                    # unread the firstchar
+  if ($firstchar !~ /[[:print:]]/) {       # is not a text character
+    return "";
+  }
+  while (defined($line = <PROFILE>)) {
+    $line =~ s/\r//g;   # turn windows-looking lines into unix-looking lines
+    if ($line =~ /^%warn\s+(.*)/) {        # 'warn' command
+      # Note this matches both '%warn blah\n' and '%warn\n'.
+      print STDERR "WARNING: $1\n";        # print the rest of the line
+    } elsif ($line =~ /^%/) {
+      print STDERR "Ignoring unknown command from profile header: $line";
+    } else {
+      # End of commands, must be the header line.
+      return $line;
+    }
+  }
+  return undef;     # got to EOF without seeing a header line
+}
+
+sub IsSymbolizedProfileFile {
+  my $file_name = shift;
+  if (!(-e $file_name) || !(-r $file_name)) {
+    return 0;
+  }
+  # Check if the file contains a symbol-section marker.
+  open(TFILE, "<$file_name");
+  binmode TFILE;
+  my $firstline = ReadProfileHeader(*TFILE);
+  close(TFILE);
+  if (!$firstline) {
+    return 0;
+  }
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+  return $firstline =~ /^--- *$symbol_marker/;
+}
+
+# Parse profile generated by common/profiler.cc and return a reference
+# to a map:
+#      $result->{version}     Version number of profile file
+#      $result->{period}      Sampling period (in microseconds)
+#      $result->{profile}     Profile object
+#      $result->{threads}     Map of thread IDs to profile objects
+#      $result->{map}         Memory map info from profile
+#      $result->{pcs}         Hash of all PC values seen, key is hex address
+sub ReadProfile {
+  my $prog = shift;
+  my $fname = shift;
+  my $result;            # return value
+
+  $CONTENTION_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $contention_marker = $&;
+  $GROWTH_PAGE  =~ m,[^/]+$,;    # matches everything after the last slash
+  my $growth_marker = $&;
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+  $PROFILE_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $profile_marker = $&;
+  $HEAP_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $heap_marker = $&;
+
+  # Look at first line to see if it is a heap or a CPU profile.
+  # CPU profile may start with no header at all, and just binary data
+  # (starting with \0\0\0\0) -- in that case, don't try to read the
+  # whole firstline, since it may be gigabytes(!) of data.
+  open(PROFILE, "<$fname") || error("$fname: $!\n");
+  binmode PROFILE;      # New perls do UTF-8 processing
+  my $header = ReadProfileHeader(*PROFILE);
+  if (!defined($header)) {   # means "at EOF"
+    error("Profile is empty.\n");
+  }
+
+  my $symbols;
+  if ($header =~ m/^--- *$symbol_marker/o) {
+    # Verify that the user asked for a symbolized profile
+    if (!$main::use_symbolized_profile) {
+      # we have both a binary and symbolized profiles, abort
+      error("FATAL ERROR: Symbolized profile\n   $fname\ncannot be used with " .
+            "a binary arg. Try again without passing\n   $prog\n");
+    }
+    # Read the symbol section of the symbolized profile file.
+    $symbols = ReadSymbols(*PROFILE{IO});
+    # Read the next line to get the header for the remaining profile.
+    $header = ReadProfileHeader(*PROFILE) || "";
+  }
+
+  if ($header =~ m/^--- *($heap_marker|$growth_marker)/o) {
+    # Skip "--- ..." line for profile types that have their own headers.
+    $header = ReadProfileHeader(*PROFILE) || "";
+  }
+
+  $main::profile_type = '';
+
+  if ($header =~ m/^heap profile:.*$growth_marker/o) {
+    $main::profile_type = 'growth';
+    $result =  ReadHeapProfile($prog, *PROFILE, $header);
+  } elsif ($header =~ m/^heap profile:/) {
+    $main::profile_type = 'heap';
+    $result =  ReadHeapProfile($prog, *PROFILE, $header);
+  } elsif ($header =~ m/^heap/) {
+    $main::profile_type = 'heap';
+    $result = ReadThreadedHeapProfile($prog, $fname, $header);
+  } elsif ($header =~ m/^--- *$contention_marker/o) {
+    $main::profile_type = 'contention';
+    $result = ReadSynchProfile($prog, *PROFILE);
+  } elsif ($header =~ m/^--- *Stacks:/) {
+    print STDERR
+      "Old format contention profile: mistakenly reports " .
+      "condition variable signals as lock contentions.\n";
+    $main::profile_type = 'contention';
+    $result = ReadSynchProfile($prog, *PROFILE);
+  } elsif ($header =~ m/^--- *$profile_marker/) {
+    # the binary cpu profile data starts immediately after this line
+    $main::profile_type = 'cpu';
+    $result = ReadCPUProfile($prog, $fname, *PROFILE);
+  } else {
+    if (defined($symbols)) {
+      # a symbolized profile contains a format we don't recognize, bail out
+      error("$fname: Cannot recognize profile section after symbols.\n");
+    }
+    # no ascii header present -- must be a CPU profile
+    $main::profile_type = 'cpu';
+    $result = ReadCPUProfile($prog, $fname, *PROFILE);
+  }
+
+  close(PROFILE);
+
+  # if we got symbols along with the profile, return those as well
+  if (defined($symbols)) {
+    $result->{symbols} = $symbols;
+  }
+
+  return $result;
+}
+
+# Subtract one from caller pc so we map back to call instr.
+# However, don't do this if we're reading a symbolized profile
+# file, in which case the subtract-one was done when the file
+# was written.
+#
+# We apply the same logic to all readers, though ReadCPUProfile uses an
+# independent implementation.
+sub FixCallerAddresses {
+  my $stack = shift;
+  # --raw/http: Always subtract one from pc's, because PrintSymbolizedProfile()
+  # dumps unadjusted profiles.
+  {
+    $stack =~ /(\s)/;
+    my $delimiter = $1;
+    my @addrs = split(' ', $stack);
+    my @fixedaddrs;
+    $#fixedaddrs = $#addrs;
+    if ($#addrs >= 0) {
+      $fixedaddrs[0] = $addrs[0];
+    }
+    for (my $i = 1; $i <= $#addrs; $i++) {
+      $fixedaddrs[$i] = AddressSub($addrs[$i], "0x1");
+    }
+    return join $delimiter, @fixedaddrs;
+  }
+}
+
+# CPU profile reader
+sub ReadCPUProfile {
+  my $prog = shift;
+  my $fname = shift;       # just used for logging
+  local *PROFILE = shift;
+  my $version;
+  my $period;
+  my $i;
+  my $profile = {};
+  my $pcs = {};
+
+  # Parse string into array of slots.
+  my $slots = CpuProfileStream->new(*PROFILE, $fname);
+
+  # Read header.  The current header version is a 5-element structure
+  # containing:
+  #   0: header count (always 0)
+  #   1: header "words" (after this one: 3)
+  #   2: format version (0)
+  #   3: sampling period (usec)
+  #   4: unused padding (always 0)
+  if ($slots->get(0) != 0 ) {
+    error("$fname: not a profile file, or old format profile file\n");
+  }
+  $i = 2 + $slots->get(1);
+  $version = $slots->get(2);
+  $period = $slots->get(3);
+  # Do some sanity checking on these header values.
+  if ($version > (2**32) || $period > (2**32) || $i > (2**32) || $i < 5) {
+    error("$fname: not a profile file, or corrupted profile file\n");
+  }
+
+  # Parse profile
+  while ($slots->get($i) != -1) {
+    my $n = $slots->get($i++);
+    my $d = $slots->get($i++);
+    if ($d > (2**16)) {  # TODO(csilvers): what's a reasonable max-stack-depth?
+      my $addr = sprintf("0%o", $i * ($address_length == 8 ? 4 : 8));
+      print STDERR "At index $i (address $addr):\n";
+      error("$fname: stack trace depth >= 2**32\n");
+    }
+    if ($slots->get($i) == 0) {
+      # End of profile data marker
+      $i += $d;
+      last;
+    }
+
+    # Make key out of the stack entries
+    my @k = ();
+    for (my $j = 0; $j < $d; $j++) {
+      my $pc = $slots->get($i+$j);
+      # Subtract one from caller pc so we map back to call instr.
+      $pc--;
+      $pc = sprintf("%0*x", $address_length, $pc);
+      $pcs->{$pc} = 1;
+      push @k, $pc;
+    }
+
+    AddEntry($profile, (join "\n", @k), $n);
+    $i += $d;
+  }
+
+  # Parse map
+  my $map = '';
+  seek(PROFILE, $i * 4, 0);
+  read(PROFILE, $map, (stat PROFILE)[7]);
+
+  my $r = {};
+  $r->{version} = $version;
+  $r->{period} = $period;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+
+  return $r;
+}
+
+sub HeapProfileIndex {
+  my $index = 1;
+  if ($main::opt_inuse_space) {
+    $index = 1;
+  } elsif ($main::opt_inuse_objects) {
+    $index = 0;
+  } elsif ($main::opt_alloc_space) {
+    $index = 3;
+  } elsif ($main::opt_alloc_objects) {
+    $index = 2;
+  }
+  return $index;
+}
+
+sub ReadMappedLibraries {
+  my $fh = shift;
+  my $map = "";
+  # Read the /proc/self/maps data
+  while (<$fh>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    $map .= $_;
+  }
+  return $map;
+}
+
+sub ReadMemoryMap {
+  my $fh = shift;
+  my $map = "";
+  # Read /proc/self/maps data as formatted by DumpAddressMap()
+  my $buildvar = "";
+  while (<PROFILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Parse "build=<dir>" specification if supplied
+    if (m/^\s*build=(.*)\n/) {
+      $buildvar = $1;
+    }
+
+    # Expand "$build" variable if available
+    $_ =~ s/\$build\b/$buildvar/g;
+
+    $map .= $_;
+  }
+  return $map;
+}
+
+sub AdjustSamples {
+  my ($sample_adjustment, $sampling_algorithm, $n1, $s1, $n2, $s2) = @_;
+  if ($sample_adjustment) {
+    if ($sampling_algorithm == 2) {
+      # Remote-heap version 2
+      # The sampling frequency is the rate of a Poisson process.
+      # This means that the probability of sampling an allocation of
+      # size X with sampling rate Y is 1 - exp(-X/Y)
+      if ($n1 != 0) {
+        my $ratio = (($s1*1.0)/$n1)/($sample_adjustment);
+        my $scale_factor = 1/(1 - exp(-$ratio));
+        $n1 *= $scale_factor;
+        $s1 *= $scale_factor;
+      }
+      if ($n2 != 0) {
+        my $ratio = (($s2*1.0)/$n2)/($sample_adjustment);
+        my $scale_factor = 1/(1 - exp(-$ratio));
+        $n2 *= $scale_factor;
+        $s2 *= $scale_factor;
+      }
+    } else {
+      # Remote-heap version 1
+      my $ratio;
+      $ratio = (($s1*1.0)/$n1)/($sample_adjustment);
+      if ($ratio < 1) {
+        $n1 /= $ratio;
+        $s1 /= $ratio;
+      }
+      $ratio = (($s2*1.0)/$n2)/($sample_adjustment);
+      if ($ratio < 1) {
+        $n2 /= $ratio;
+        $s2 /= $ratio;
+      }
+    }
+  }
+  return ($n1, $s1, $n2, $s2);
+}
+
+sub ReadHeapProfile {
+  my $prog = shift;
+  local *PROFILE = shift;
+  my $header = shift;
+
+  my $index = HeapProfileIndex();
+
+  # Find the type of this profile.  The header line looks like:
+  #    heap profile:   1246:  8800744 [  1246:  8800744] @ <heap-url>/266053
+  # There are two pairs <count: size>, the first inuse objects/space, and the
+  # second allocated objects/space.  This is followed optionally by a profile
+  # type, and if that is present, optionally by a sampling frequency.
+  # For remote heap profiles (v1):
+  # The interpretation of the sampling frequency is that the profiler, for
+  # each sample, calculates a uniformly distributed random integer less than
+  # the given value, and records the next sample after that many bytes have
+  # been allocated.  Therefore, the expected sample interval is half of the
+  # given frequency.  By default, if not specified, the expected sample
+  # interval is 128KB.  Only remote-heap-page profiles are adjusted for
+  # sample size.
+  # For remote heap profiles (v2):
+  # The sampling frequency is the rate of a Poisson process. This means that
+  # the probability of sampling an allocation of size X with sampling rate Y
+  # is 1 - exp(-X/Y)
+  # For version 2, a typical header line might look like this:
+  # heap profile:   1922: 127792360 [  1922: 127792360] @ <heap-url>_v2/524288
+  # the trailing number (524288) is the sampling rate. (Version 1 showed
+  # double the 'rate' here)
+  my $sampling_algorithm = 0;
+  my $sample_adjustment = 0;
+  chomp($header);
+  my $type = "unknown";
+  if ($header =~ m"^heap profile:\s*(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\](\s*@\s*([^/]*)(/(\d+))?)?") {
+    if (defined($6) && ($6 ne '')) {
+      $type = $6;
+      my $sample_period = $8;
+      # $type is "heapprofile" for profiles generated by the
+      # heap-profiler, and either "heap" or "heap_v2" for profiles
+      # generated by sampling directly within tcmalloc.  It can also
+      # be "growth" for heap-growth profiles.  The first is typically
+      # found for profiles generated locally, and the others for
+      # remote profiles.
+      if (($type eq "heapprofile") || ($type !~ /heap/) ) {
+        # No need to adjust for the sampling rate with heap-profiler-derived data
+        $sampling_algorithm = 0;
+      } elsif ($type =~ /_v2/) {
+        $sampling_algorithm = 2;     # version 2 sampling
+        if (defined($sample_period) && ($sample_period ne '')) {
+          $sample_adjustment = int($sample_period);
+        }
+      } else {
+        $sampling_algorithm = 1;     # version 1 sampling
+        if (defined($sample_period) && ($sample_period ne '')) {
+          $sample_adjustment = int($sample_period)/2;
+        }
+      }
+    } else {
+      # We detect whether or not this is a remote-heap profile by checking
+      # that the total-allocated stats ($n2,$s2) are exactly the
+      # same as the in-use stats ($n1,$s1).  It is remotely conceivable
+      # that a non-remote-heap profile may pass this check, but it is hard
+      # to imagine how that could happen.
+      # In this case it's so old it's guaranteed to be remote-heap version 1.
+      my ($n1, $s1, $n2, $s2) = ($1, $2, $3, $4);
+      if (($n1 == $n2) && ($s1 == $s2)) {
+        # This is likely to be a remote-heap based sample profile
+        $sampling_algorithm = 1;
+      }
+    }
+  }
+
+  if ($sampling_algorithm > 0) {
+    # For remote-heap generated profiles, adjust the counts and sizes to
+    # account for the sample rate (we sample once every 128KB by default).
+    if ($sample_adjustment == 0) {
+      # Turn on profile adjustment.
+      $sample_adjustment = 128*1024;
+      print STDERR "Adjusting heap profiles for 1-in-128KB sampling rate\n";
+    } else {
+      printf STDERR ("Adjusting heap profiles for 1-in-%d sampling rate\n",
+                     $sample_adjustment);
+    }
+    if ($sampling_algorithm > 1) {
+      # We don't bother printing anything for the original version (version 1)
+      printf STDERR "Heap version $sampling_algorithm\n";
+    }
+  }
+
+  my $profile = {};
+  my $pcs = {};
+  my $map = "";
+
+  while (<PROFILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (/^MAPPED_LIBRARIES:/) {
+      $map .= ReadMappedLibraries(*PROFILE);
+      last;
+    }
+
+    if (/^--- Memory map:/) {
+      $map .= ReadMemoryMap(*PROFILE);
+      last;
+    }
+
+    # Read entry of the form:
+    #  <count1>: <bytes1> [<count2>: <bytes2>] @ a1 a2 a3 ... an
+    s/^\s*//;
+    s/\s*$//;
+    if (m/^\s*(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\]\s+@\s+(.*)$/) {
+      my $stack = $5;
+      my ($n1, $s1, $n2, $s2) = ($1, $2, $3, $4);
+      my @counts = AdjustSamples($sample_adjustment, $sampling_algorithm,
+                                 $n1, $s1, $n2, $s2);
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $counts[$index]);
+    }
+  }
+
+  my $r = {};
+  $r->{version} = "heap";
+  $r->{period} = 1;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+sub ReadThreadedHeapProfile {
+  my ($prog, $fname, $header) = @_;
+
+  my $index = HeapProfileIndex();
+  my $sampling_algorithm = 0;
+  my $sample_adjustment = 0;
+  chomp($header);
+  my $type = "unknown";
+  # Assuming a very specific type of header for now.
+  if ($header =~ m"^heap_v2/(\d+)") {
+    $type = "_v2";
+    $sampling_algorithm = 2;
+    $sample_adjustment = int($1);
+  }
+  if ($type ne "_v2" || !defined($sample_adjustment)) {
+    die "Threaded heap profiles require v2 sampling with a sample rate\n";
+  }
+
+  my $profile = {};
+  my $thread_profiles = {};
+  my $pcs = {};
+  my $map = "";
+  my $stack = "";
+
+  while (<PROFILE>) {
+    s/\r//g;
+    if (/^MAPPED_LIBRARIES:/) {
+      $map .= ReadMappedLibraries(*PROFILE);
+      last;
+    }
+
+    if (/^--- Memory map:/) {
+      $map .= ReadMemoryMap(*PROFILE);
+      last;
+    }
+
+    # Read entry of the form:
+    # @ a1 a2 ... an
+    #   t*: <count1>: <bytes1> [<count2>: <bytes2>]
+    #   t1: <count1>: <bytes1> [<count2>: <bytes2>]
+    #     ...
+    #   tn: <count1>: <bytes1> [<count2>: <bytes2>]
+    s/^\s*//;
+    s/\s*$//;
+    if (m/^@\s+(.*)$/) {
+      $stack = $1;
+    } elsif (m/^\s*(t(\*|\d+)):\s+(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\]$/) {
+      if ($stack eq "") {
+        # Still in the header, so this is just a per-thread summary.
+        next;
+      }
+      my $thread = $2;
+      my ($n1, $s1, $n2, $s2) = ($3, $4, $5, $6);
+      my @counts = AdjustSamples($sample_adjustment, $sampling_algorithm,
+                                 $n1, $s1, $n2, $s2);
+      if ($thread eq "*") {
+        AddEntries($profile, $pcs, FixCallerAddresses($stack), $counts[$index]);
+      } else {
+        if (!exists($thread_profiles->{$thread})) {
+          $thread_profiles->{$thread} = {};
+        }
+        AddEntries($thread_profiles->{$thread}, $pcs,
+                   FixCallerAddresses($stack), $counts[$index]);
+      }
+    }
+  }
+
+  my $r = {};
+  $r->{version} = "heap";
+  $r->{period} = 1;
+  $r->{profile} = $profile;
+  $r->{threads} = $thread_profiles;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+sub ReadSynchProfile {
+  my $prog = shift;
+  local *PROFILE = shift;
+  my $header = shift;
+
+  my $map = '';
+  my $profile = {};
+  my $pcs = {};
+  my $sampling_period = 1;
+  my $cyclespernanosec = 2.8;   # Default assumption for old binaries
+  my $seen_clockrate = 0;
+  my $line;
+
+  my $index = 0;
+  if ($main::opt_total_delay) {
+    $index = 0;
+  } elsif ($main::opt_contentions) {
+    $index = 1;
+  } elsif ($main::opt_mean_delay) {
+    $index = 2;
+  }
+
+  while ( $line = <PROFILE> ) {
+    $line =~ s/\r//g;      # turn windows-looking lines into unix-looking lines
+    if ( $line =~ /^\s*(\d+)\s+(\d+) \@\s*(.*?)\s*$/ ) {
+      my ($cycles, $count, $stack) = ($1, $2, $3);
+
+      # Convert cycles to nanoseconds
+      $cycles /= $cyclespernanosec;
+
+      # Adjust for sampling done by application
+      $cycles *= $sampling_period;
+      $count *= $sampling_period;
+
+      my @values = ($cycles, $count, $cycles / $count);
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $values[$index]);
+
+    } elsif ( $line =~ /^(slow release).*thread \d+  \@\s*(.*?)\s*$/ ||
+              $line =~ /^\s*(\d+) \@\s*(.*?)\s*$/ ) {
+      my ($cycles, $stack) = ($1, $2);
+      if ($cycles !~ /^\d+$/) {
+        next;
+      }
+
+      # Convert cycles to nanoseconds
+      $cycles /= $cyclespernanosec;
+
+      # Adjust for sampling done by application
+      $cycles *= $sampling_period;
+
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $cycles);
+
+    } elsif ( $line =~ m/^([a-z][^=]*)=(.*)$/ ) {
+      my ($variable, $value) = ($1,$2);
+      for ($variable, $value) {
+        s/^\s+//;
+        s/\s+$//;
+      }
+      if ($variable eq "cycles/second") {
+        $cyclespernanosec = $value / 1e9;
+        $seen_clockrate = 1;
+      } elsif ($variable eq "sampling period") {
+        $sampling_period = $value;
+      } elsif ($variable eq "ms since reset") {
+        # Currently nothing is done with this value in jeprof
+        # So we just silently ignore it for now
+      } elsif ($variable eq "discarded samples") {
+        # Currently nothing is done with this value in jeprof
+        # So we just silently ignore it for now
+      } else {
+        printf STDERR ("Ignoring unnknown variable in /contention output: " .
+                       "'%s' = '%s'\n",$variable,$value);
+      }
+    } else {
+      # Memory map entry
+      $map .= $line;
+    }
+  }
+
+  if (!$seen_clockrate) {
+    printf STDERR ("No cycles/second entry in profile; Guessing %.1f GHz\n",
+                   $cyclespernanosec);
+  }
+
+  my $r = {};
+  $r->{version} = 0;
+  $r->{period} = $sampling_period;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+# Given a hex value in the form "0x1abcd" or "1abcd", return either
+# "0001abcd" or "000000000001abcd", depending on the current (global)
+# address length.
+sub HexExtend {
+  my $addr = shift;
+
+  $addr =~ s/^(0x)?0*//;
+  my $zeros_needed = $address_length - length($addr);
+  if ($zeros_needed < 0) {
+    printf STDERR "Warning: address $addr is longer than address length $address_length\n";
+    return $addr;
+  }
+  return ("0" x $zeros_needed) . $addr;
+}
+
+##### Symbol extraction #####
+
+# Aggressively search the lib_prefix values for the given library
+# If all else fails, just return the name of the library unmodified.
+# If the lib_prefix is "/my/path,/other/path" and $file is "/lib/dir/mylib.so"
+# it will search the following locations in this order, until it finds a file:
+#   /my/path/lib/dir/mylib.so
+#   /other/path/lib/dir/mylib.so
+#   /my/path/dir/mylib.so
+#   /other/path/dir/mylib.so
+#   /my/path/mylib.so
+#   /other/path/mylib.so
+#   /lib/dir/mylib.so              (returned as last resort)
+sub FindLibrary {
+  my $file = shift;
+  my $suffix = $file;
+
+  # Search for the library as described above
+  do {
+    foreach my $prefix (@prefix_list) {
+      my $fullpath = $prefix . $suffix;
+      if (-e $fullpath) {
+        return $fullpath;
+      }
+    }
+  } while ($suffix =~ s|^/[^/]+/|/|);
+  return $file;
+}
+
+# Return path to library with debugging symbols.
+# For libc libraries, the copy in /usr/lib/debug contains debugging symbols
+sub DebuggingLibrary {
+  my $file = shift;
+  if ($file =~ m|^/|) {
+      if (-f "/usr/lib/debug$file") {
+        return "/usr/lib/debug$file";
+      } elsif (-f "/usr/lib/debug$file.debug") {
+        return "/usr/lib/debug$file.debug";
+      }
+  }
+  return undef;
+}
+
+# Parse text section header of a library using objdump
+sub ParseTextSectionHeaderFromObjdump {
+  my $lib = shift;
+
+  my $size = undef;
+  my $vma;
+  my $file_offset;
+  # Get objdump output from the library file to figure out how to
+  # map between mapped addresses and addresses in the library.
+  my $cmd = ShellEscape($obj_tool_map{"objdump"}, "-h", $lib);
+  open(OBJDUMP, "$cmd |") || error("$cmd: $!\n");
+  while (<OBJDUMP>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Idx Name          Size      VMA       LMA       File off  Algn
+    #  10 .text         00104b2c  420156f0  420156f0  000156f0  2**4
+    # For 64-bit objects, VMA and LMA will be 16 hex digits, size and file
+    # offset may still be 8.  But AddressSub below will still handle that.
+    my @x = split;
+    if (($#x >= 6) && ($x[1] eq '.text')) {
+      $size = $x[2];
+      $vma = $x[3];
+      $file_offset = $x[5];
+      last;
+    }
+  }
+  close(OBJDUMP);
+
+  if (!defined($size)) {
+    return undef;
+  }
+
+  my $r = {};
+  $r->{size} = $size;
+  $r->{vma} = $vma;
+  $r->{file_offset} = $file_offset;
+
+  return $r;
+}
+
+# Parse text section header of a library using otool (on OS X)
+sub ParseTextSectionHeaderFromOtool {
+  my $lib = shift;
+
+  my $size = undef;
+  my $vma = undef;
+  my $file_offset = undef;
+  # Get otool output from the library file to figure out how to
+  # map between mapped addresses and addresses in the library.
+  my $command = ShellEscape($obj_tool_map{"otool"}, "-l", $lib);
+  open(OTOOL, "$command |") || error("$command: $!\n");
+  my $cmd = "";
+  my $sectname = "";
+  my $segname = "";
+  foreach my $line (<OTOOL>) {
+    $line =~ s/\r//g;      # turn windows-looking lines into unix-looking lines
+    # Load command <#>
+    #       cmd LC_SEGMENT
+    # [...]
+    # Section
+    #   sectname __text
+    #    segname __TEXT
+    #       addr 0x000009f8
+    #       size 0x00018b9e
+    #     offset 2552
+    #      align 2^2 (4)
+    # We will need to strip off the leading 0x from the hex addresses,
+    # and convert the offset into hex.
+    if ($line =~ /Load command/) {
+      $cmd = "";
+      $sectname = "";
+      $segname = "";
+    } elsif ($line =~ /Section/) {
+      $sectname = "";
+      $segname = "";
+    } elsif ($line =~ /cmd (\w+)/) {
+      $cmd = $1;
+    } elsif ($line =~ /sectname (\w+)/) {
+      $sectname = $1;
+    } elsif ($line =~ /segname (\w+)/) {
+      $segname = $1;
+    } elsif (!(($cmd eq "LC_SEGMENT" || $cmd eq "LC_SEGMENT_64") &&
+               $sectname eq "__text" &&
+               $segname eq "__TEXT")) {
+      next;
+    } elsif ($line =~ /\baddr 0x([0-9a-fA-F]+)/) {
+      $vma = $1;
+    } elsif ($line =~ /\bsize 0x([0-9a-fA-F]+)/) {
+      $size = $1;
+    } elsif ($line =~ /\boffset ([0-9]+)/) {
+      $file_offset = sprintf("%016x", $1);
+    }
+    if (defined($vma) && defined($size) && defined($file_offset)) {
+      last;
+    }
+  }
+  close(OTOOL);
+
+  if (!defined($vma) || !defined($size) || !defined($file_offset)) {
+     return undef;
+  }
+
+  my $r = {};
+  $r->{size} = $size;
+  $r->{vma} = $vma;
+  $r->{file_offset} = $file_offset;
+
+  return $r;
+}
+
+sub ParseTextSectionHeader {
+  # obj_tool_map("otool") is only defined if we're in a Mach-O environment
+  if (defined($obj_tool_map{"otool"})) {
+    my $r = ParseTextSectionHeaderFromOtool(@_);
+    if (defined($r)){
+      return $r;
+    }
+  }
+  # If otool doesn't work, or we don't have it, fall back to objdump
+  return ParseTextSectionHeaderFromObjdump(@_);
+}
+
+# Split /proc/pid/maps dump into a list of libraries
+sub ParseLibraries {
+  return if $main::use_symbol_page;  # We don't need libraries info.
+  my $prog = Cwd::abs_path(shift);
+  my $map = shift;
+  my $pcs = shift;
+
+  my $result = [];
+  my $h = "[a-f0-9]+";
+  my $zero_offset = HexExtend("0");
+
+  my $buildvar = "";
+  foreach my $l (split("\n", $map)) {
+    if ($l =~ m/^\s*build=(.*)$/) {
+      $buildvar = $1;
+    }
+
+    my $start;
+    my $finish;
+    my $offset;
+    my $lib;
+    if ($l =~ /^($h)-($h)\s+..x.\s+($h)\s+\S+:\S+\s+\d+\s+(\S+\.(so|dll|dylib|bundle)((\.\d+)+\w*(\.\d+){0,3})?)$/i) {
+      # Full line from /proc/self/maps.  Example:
+      #   40000000-40015000 r-xp 00000000 03:01 12845071   /lib/ld-2.3.2.so
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = HexExtend($3);
+      $lib = $4;
+      $lib =~ s|\\|/|g;     # turn windows-style paths into unix-style paths
+    } elsif ($l =~ /^\s*($h)-($h):\s*(\S+\.so(\.\d+)*)/) {
+      # Cooked line from DumpAddressMap.  Example:
+      #   40000000-40015000: /lib/ld-2.3.2.so
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = $zero_offset;
+      $lib = $3;
+    } elsif (($l =~ /^($h)-($h)\s+..x.\s+($h)\s+\S+:\S+\s+\d+\s+(\S+)$/i) && ($4 eq $prog)) {
+      # PIEs and address space randomization do not play well with our
+      # default assumption that main executable is at lowest
+      # addresses. So we're detecting main executable in
+      # /proc/self/maps as well.
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = HexExtend($3);
+      $lib = $4;
+      $lib =~ s|\\|/|g;     # turn windows-style paths into unix-style paths
+    }
+    # FreeBSD 10.0 virtual memory map /proc/curproc/map as defined in
+    # function procfs_doprocmap (sys/fs/procfs/procfs_map.c)
+    #
+    # Example:
+    # 0x800600000 0x80061a000 26 0 0xfffff800035a0000 r-x 75 33 0x1004 COW NC vnode /libexec/ld-elf.s
+    # o.1 NCH -1
+    elsif ($l =~ /^(0x$h)\s(0x$h)\s\d+\s\d+\s0x$h\sr-x\s\d+\s\d+\s0x\d+\s(COW|NCO)\s(NC|NNC)\svnode\s(\S+\.so(\.\d+)*)/) {
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = $zero_offset;
+      $lib = FindLibrary($5);
+
+    } else {
+      next;
+    }
+
+    # Expand "$build" variable if available
+    $lib =~ s/\$build\b/$buildvar/g;
+
+    $lib = FindLibrary($lib);
+
+    # Check for pre-relocated libraries, which use pre-relocated symbol tables
+    # and thus require adjusting the offset that we'll use to translate
+    # VM addresses into symbol table addresses.
+    # Only do this if we're not going to fetch the symbol table from a
+    # debugging copy of the library.
+    if (!DebuggingLibrary($lib)) {
+      my $text = ParseTextSectionHeader($lib);
+      if (defined($text)) {
+         my $vma_offset = AddressSub($text->{vma}, $text->{file_offset});
+         $offset = AddressAdd($offset, $vma_offset);
+      }
+    }
+
+    if($main::opt_debug) { printf STDERR "$start:$finish ($offset) $lib\n"; }
+    push(@{$result}, [$lib, $start, $finish, $offset]);
+  }
+
+  # Append special entry for additional library (not relocated)
+  if ($main::opt_lib ne "") {
+    my $text = ParseTextSectionHeader($main::opt_lib);
+    if (defined($text)) {
+       my $start = $text->{vma};
+       my $finish = AddressAdd($start, $text->{size});
+
+       push(@{$result}, [$main::opt_lib, $start, $finish, $start]);
+    }
+  }
+
+  # Append special entry for the main program.  This covers
+  # 0..max_pc_value_seen, so that we assume pc values not found in one
+  # of the library ranges will be treated as coming from the main
+  # program binary.
+  my $min_pc = HexExtend("0");
+  my $max_pc = $min_pc;          # find the maximal PC value in any sample
+  foreach my $pc (keys(%{$pcs})) {
+    if (HexExtend($pc) gt $max_pc) { $max_pc = HexExtend($pc); }
+  }
+  push(@{$result}, [$prog, $min_pc, $max_pc, $zero_offset]);
+
+  return $result;
+}
+
+# Add two hex addresses of length $address_length.
+# Run jeprof --test for unit test if this is changed.
+sub AddressAdd {
+  my $addr1 = shift;
+  my $addr2 = shift;
+  my $sum;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $sum = (hex($addr1)+hex($addr2)) % (0x10000000 * 16);
+    return sprintf("%08x", $sum);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize carry handling.
+
+    if ($main::opt_debug and $main::opt_test) {
+      print STDERR "AddressAdd $addr1 + $addr2 = ";
+    }
+
+    my $a1 = substr($addr1,-7);
+    $addr1 = substr($addr1,0,-7);
+    my $a2 = substr($addr2,-7);
+    $addr2 = substr($addr2,0,-7);
+    $sum = hex($a1) + hex($a2);
+    my $c = 0;
+    if ($sum > 0xfffffff) {
+      $c = 1;
+      $sum -= 0x10000000;
+    }
+    my $r = sprintf("%07x", $sum);
+
+    $a1 = substr($addr1,-7);
+    $addr1 = substr($addr1,0,-7);
+    $a2 = substr($addr2,-7);
+    $addr2 = substr($addr2,0,-7);
+    $sum = hex($a1) + hex($a2) + $c;
+    $c = 0;
+    if ($sum > 0xfffffff) {
+      $c = 1;
+      $sum -= 0x10000000;
+    }
+    $r = sprintf("%07x", $sum) . $r;
+
+    $sum = hex($addr1) + hex($addr2) + $c;
+    if ($sum > 0xff) { $sum -= 0x100; }
+    $r = sprintf("%02x", $sum) . $r;
+
+    if ($main::opt_debug and $main::opt_test) { print STDERR "$r\n"; }
+
+    return $r;
+  }
+}
+
+
+# Subtract two hex addresses of length $address_length.
+# Run jeprof --test for unit test if this is changed.
+sub AddressSub {
+  my $addr1 = shift;
+  my $addr2 = shift;
+  my $diff;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $diff = (hex($addr1)-hex($addr2)) % (0x10000000 * 16);
+    return sprintf("%08x", $diff);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize borrow handling.
+    # if ($main::opt_debug) { print STDERR "AddressSub $addr1 - $addr2 = "; }
+
+    my $a1 = hex(substr($addr1,-7));
+    $addr1 = substr($addr1,0,-7);
+    my $a2 = hex(substr($addr2,-7));
+    $addr2 = substr($addr2,0,-7);
+    my $b = 0;
+    if ($a2 > $a1) {
+      $b = 1;
+      $a1 += 0x10000000;
+    }
+    $diff = $a1 - $a2;
+    my $r = sprintf("%07x", $diff);
+
+    $a1 = hex(substr($addr1,-7));
+    $addr1 = substr($addr1,0,-7);
+    $a2 = hex(substr($addr2,-7)) + $b;
+    $addr2 = substr($addr2,0,-7);
+    $b = 0;
+    if ($a2 > $a1) {
+      $b = 1;
+      $a1 += 0x10000000;
+    }
+    $diff = $a1 - $a2;
+    $r = sprintf("%07x", $diff) . $r;
+
+    $a1 = hex($addr1);
+    $a2 = hex($addr2) + $b;
+    if ($a2 > $a1) { $a1 += 0x100; }
+    $diff = $a1 - $a2;
+    $r = sprintf("%02x", $diff) . $r;
+
+    # if ($main::opt_debug) { print STDERR "$r\n"; }
+
+    return $r;
+  }
+}
+
+# Increment a hex addresses of length $address_length.
+# Run jeprof --test for unit test if this is changed.
+sub AddressInc {
+  my $addr = shift;
+  my $sum;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $sum = (hex($addr)+1) % (0x10000000 * 16);
+    return sprintf("%08x", $sum);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize carry handling.
+    # We are always doing this to step through the addresses in a function,
+    # and will almost never overflow the first chunk, so we check for this
+    # case and exit early.
+
+    # if ($main::opt_debug) { print STDERR "AddressInc $addr1 = "; }
+
+    my $a1 = substr($addr,-7);
+    $addr = substr($addr,0,-7);
+    $sum = hex($a1) + 1;
+    my $r = sprintf("%07x", $sum);
+    if ($sum <= 0xfffffff) {
+      $r = $addr . $r;
+      # if ($main::opt_debug) { print STDERR "$r\n"; }
+      return HexExtend($r);
+    } else {
+      $r = "0000000";
+    }
+
+    $a1 = substr($addr,-7);
+    $addr = substr($addr,0,-7);
+    $sum = hex($a1) + 1;
+    $r = sprintf("%07x", $sum) . $r;
+    if ($sum <= 0xfffffff) {
+      $r = $addr . $r;
+      # if ($main::opt_debug) { print STDERR "$r\n"; }
+      return HexExtend($r);
+    } else {
+      $r = "00000000000000";
+    }
+
+    $sum = hex($addr) + 1;
+    if ($sum > 0xff) { $sum -= 0x100; }
+    $r = sprintf("%02x", $sum) . $r;
+
+    # if ($main::opt_debug) { print STDERR "$r\n"; }
+    return $r;
+  }
+}
+
+# Extract symbols for all PC values found in profile
+sub ExtractSymbols {
+  my $libs = shift;
+  my $pcset = shift;
+
+  my $symbols = {};
+
+  # Map each PC value to the containing library.  To make this faster,
+  # we sort libraries by their starting pc value (highest first), and
+  # advance through the libraries as we advance the pc.  Sometimes the
+  # addresses of libraries may overlap with the addresses of the main
+  # binary, so to make sure the libraries 'win', we iterate over the
+  # libraries in reverse order (which assumes the binary doesn't start
+  # in the middle of a library, which seems a fair assumption).
+  my @pcs = (sort { $a cmp $b } keys(%{$pcset}));  # pcset is 0-extended strings
+  foreach my $lib (sort {$b->[1] cmp $a->[1]} @{$libs}) {
+    my $libname = $lib->[0];
+    my $start = $lib->[1];
+    my $finish = $lib->[2];
+    my $offset = $lib->[3];
+
+    # Use debug library if it exists
+    my $debug_libname = DebuggingLibrary($libname);
+    if ($debug_libname) {
+        $libname = $debug_libname;
+    }
+
+    # Get list of pcs that belong in this library.
+    my $contained = [];
+    my ($start_pc_index, $finish_pc_index);
+    # Find smallest finish_pc_index such that $finish < $pc[$finish_pc_index].
+    for ($finish_pc_index = $#pcs + 1; $finish_pc_index > 0;
+         $finish_pc_index--) {
+      last if $pcs[$finish_pc_index - 1] le $finish;
+    }
+    # Find smallest start_pc_index such that $start <= $pc[$start_pc_index].
+    for ($start_pc_index = $finish_pc_index; $start_pc_index > 0;
+         $start_pc_index--) {
+      last if $pcs[$start_pc_index - 1] lt $start;
+    }
+    # This keeps PC values higher than $pc[$finish_pc_index] in @pcs,
+    # in case there are overlaps in libraries and the main binary.
+    @{$contained} = splice(@pcs, $start_pc_index,
+                           $finish_pc_index - $start_pc_index);
+    # Map to symbols
+    MapToSymbols($libname, AddressSub($start, $offset), $contained, $symbols);
+  }
+
+  return $symbols;
+}
+
+# Map list of PC values to symbols for a given image
+sub MapToSymbols {
+  my $image = shift;
+  my $offset = shift;
+  my $pclist = shift;
+  my $symbols = shift;
+
+  my $debug = 0;
+
+  # Ignore empty binaries
+  if ($#{$pclist} < 0) { return; }
+
+  # Figure out the addr2line command to use
+  my $addr2line = $obj_tool_map{"addr2line"};
+  my $cmd = ShellEscape($addr2line, "-f", "-C", "-e", $image);
+  if (exists $obj_tool_map{"addr2line_pdb"}) {
+    $addr2line = $obj_tool_map{"addr2line_pdb"};
+    $cmd = ShellEscape($addr2line, "--demangle", "-f", "-C", "-e", $image);
+  }
+
+  # If "addr2line" isn't installed on the system at all, just use
+  # nm to get what info we can (function names, but not line numbers).
+  if (system(ShellEscape($addr2line, "--help") . " >$dev_null 2>&1") != 0) {
+    MapSymbolsWithNM($image, $offset, $pclist, $symbols);
+    return;
+  }
+
+  # "addr2line -i" can produce a variable number of lines per input
+  # address, with no separator that allows us to tell when data for
+  # the next address starts.  So we find the address for a special
+  # symbol (_fini) and interleave this address between all real
+  # addresses passed to addr2line.  The name of this special symbol
+  # can then be used as a separator.
+  $sep_address = undef;  # May be filled in by MapSymbolsWithNM()
+  my $nm_symbols = {};
+  MapSymbolsWithNM($image, $offset, $pclist, $nm_symbols);
+  if (defined($sep_address)) {
+    # Only add " -i" to addr2line if the binary supports it.
+    # addr2line --help returns 0, but not if it sees an unknown flag first.
+    if (system("$cmd -i --help >$dev_null 2>&1") == 0) {
+      $cmd .= " -i";
+    } else {
+      $sep_address = undef;   # no need for sep_address if we don't support -i
+    }
+  }
+
+  # Make file with all PC values with intervening 'sep_address' so
+  # that we can reliably detect the end of inlined function list
+  open(ADDRESSES, ">$main::tmpfile_sym") || error("$main::tmpfile_sym: $!\n");
+  if ($debug) { print("---- $image ---\n"); }
+  for (my $i = 0; $i <= $#{$pclist}; $i++) {
+    # addr2line always reads hex addresses, and does not need '0x' prefix.
+    if ($debug) { printf STDERR ("%s\n", $pclist->[$i]); }
+    printf ADDRESSES ("%s\n", AddressSub($pclist->[$i], $offset));
+    if (defined($sep_address)) {
+      printf ADDRESSES ("%s\n", $sep_address);
+    }
+  }
+  close(ADDRESSES);
+  if ($debug) {
+    print("----\n");
+    system("cat", $main::tmpfile_sym);
+    print("----\n");
+    system("$cmd < " . ShellEscape($main::tmpfile_sym));
+    print("----\n");
+  }
+
+  open(SYMBOLS, "$cmd <" . ShellEscape($main::tmpfile_sym) . " |")
+      || error("$cmd: $!\n");
+  my $count = 0;   # Index in pclist
+  while (<SYMBOLS>) {
+    # Read fullfunction and filelineinfo from next pair of lines
+    s/\r?\n$//g;
+    my $fullfunction = $_;
+    $_ = <SYMBOLS>;
+    s/\r?\n$//g;
+    my $filelinenum = $_;
+
+    if (defined($sep_address) && $fullfunction eq $sep_symbol) {
+      # Terminating marker for data for this address
+      $count++;
+      next;
+    }
+
+    $filelinenum =~ s|\\|/|g; # turn windows-style paths into unix-style paths
+
+    my $pcstr = $pclist->[$count];
+    my $function = ShortFunctionName($fullfunction);
+    my $nms = $nm_symbols->{$pcstr};
+    if (defined($nms)) {
+      if ($fullfunction eq '??') {
+        # nm found a symbol for us.
+        $function = $nms->[0];
+        $fullfunction = $nms->[2];
+      } else {
+	# MapSymbolsWithNM tags each routine with its starting address,
+	# useful in case the image has multiple occurrences of this
+	# routine.  (It uses a syntax that resembles template paramters,
+	# that are automatically stripped out by ShortFunctionName().)
+	# addr2line does not provide the same information.  So we check
+	# if nm disambiguated our symbol, and if so take the annotated
+	# (nm) version of the routine-name.  TODO(csilvers): this won't
+	# catch overloaded, inlined symbols, which nm doesn't see.
+	# Better would be to do a check similar to nm's, in this fn.
+	if ($nms->[2] =~ m/^\Q$function\E/) {  # sanity check it's the right fn
+	  $function = $nms->[0];
+	  $fullfunction = $nms->[2];
+	}
+      }
+    }
+
+    # Prepend to accumulated symbols for pcstr
+    # (so that caller comes before callee)
+    my $sym = $symbols->{$pcstr};
+    if (!defined($sym)) {
+      $sym = [];
+      $symbols->{$pcstr} = $sym;
+    }
+    unshift(@{$sym}, $function, $filelinenum, $fullfunction);
+    if ($debug) { printf STDERR ("%s => [%s]\n", $pcstr, join(" ", @{$sym})); }
+    if (!defined($sep_address)) {
+      # Inlining is off, so this entry ends immediately
+      $count++;
+    }
+  }
+  close(SYMBOLS);
+}
+
+# Use nm to map the list of referenced PCs to symbols.  Return true iff we
+# are able to read procedure information via nm.
+sub MapSymbolsWithNM {
+  my $image = shift;
+  my $offset = shift;
+  my $pclist = shift;
+  my $symbols = shift;
+
+  # Get nm output sorted by increasing address
+  my $symbol_table = GetProcedureBoundaries($image, ".");
+  if (!%{$symbol_table}) {
+    return 0;
+  }
+  # Start addresses are already the right length (8 or 16 hex digits).
+  my @names = sort { $symbol_table->{$a}->[0] cmp $symbol_table->{$b}->[0] }
+    keys(%{$symbol_table});
+
+  if ($#names < 0) {
+    # No symbols: just use addresses
+    foreach my $pc (@{$pclist}) {
+      my $pcstr = "0x" . $pc;
+      $symbols->{$pc} = [$pcstr, "?", $pcstr];
+    }
+    return 0;
+  }
+
+  # Sort addresses so we can do a join against nm output
+  my $index = 0;
+  my $fullname = $names[0];
+  my $name = ShortFunctionName($fullname);
+  foreach my $pc (sort { $a cmp $b } @{$pclist}) {
+    # Adjust for mapped offset
+    my $mpc = AddressSub($pc, $offset);
+    while (($index < $#names) && ($mpc ge $symbol_table->{$fullname}->[1])){
+      $index++;
+      $fullname = $names[$index];
+      $name = ShortFunctionName($fullname);
+    }
+    if ($mpc lt $symbol_table->{$fullname}->[1]) {
+      $symbols->{$pc} = [$name, "?", $fullname];
+    } else {
+      my $pcstr = "0x" . $pc;
+      $symbols->{$pc} = [$pcstr, "?", $pcstr];
+    }
+  }
+  return 1;
+}
+
+sub ShortFunctionName {
+  my $function = shift;
+  while ($function =~ s/\([^()]*\)(\s*const)?//g) { }   # Argument types
+  while ($function =~ s/<[^<>]*>//g)  { }    # Remove template arguments
+  $function =~ s/^.*\s+(\w+::)/$1/;          # Remove leading type
+  return $function;
+}
+
+# Trim overly long symbols found in disassembler output
+sub CleanDisassembly {
+  my $d = shift;
+  while ($d =~ s/\([^()%]*\)(\s*const)?//g) { } # Argument types, not (%rax)
+  while ($d =~ s/(\w+)<[^<>]*>/$1/g)  { }       # Remove template arguments
+  return $d;
+}
+
+# Clean file name for display
+sub CleanFileName {
+  my ($f) = @_;
+  $f =~ s|^/proc/self/cwd/||;
+  $f =~ s|^\./||;
+  return $f;
+}
+
+# Make address relative to section and clean up for display
+sub UnparseAddress {
+  my ($offset, $address) = @_;
+  $address = AddressSub($address, $offset);
+  $address =~ s/^0x//;
+  $address =~ s/^0*//;
+  return $address;
+}
+
+##### Miscellaneous #####
+
+# Find the right versions of the above object tools to use.  The
+# argument is the program file being analyzed, and should be an ELF
+# 32-bit or ELF 64-bit executable file.  The location of the tools
+# is determined by considering the following options in this order:
+#   1) --tools option, if set
+#   2) JEPROF_TOOLS environment variable, if set
+#   3) the environment
+sub ConfigureObjTools {
+  my $prog_file = shift;
+
+  # Check for the existence of $prog_file because /usr/bin/file does not
+  # predictably return error status in prod.
+  (-e $prog_file)  || error("$prog_file does not exist.\n");
+
+  my $file_type = undef;
+  if (-e "/usr/bin/file") {
+    # Follow symlinks (at least for systems where "file" supports that).
+    my $escaped_prog_file = ShellEscape($prog_file);
+    $file_type = `/usr/bin/file -L $escaped_prog_file 2>$dev_null ||
+                  /usr/bin/file $escaped_prog_file`;
+  } elsif ($^O == "MSWin32") {
+    $file_type = "MS Windows";
+  } else {
+    print STDERR "WARNING: Can't determine the file type of $prog_file";
+  }
+
+  if ($file_type =~ /64-bit/) {
+    # Change $address_length to 16 if the program file is ELF 64-bit.
+    # We can't detect this from many (most?) heap or lock contention
+    # profiles, since the actual addresses referenced are generally in low
+    # memory even for 64-bit programs.
+    $address_length = 16;
+  }
+
+  if ($file_type =~ /MS Windows/) {
+    # For windows, we provide a version of nm and addr2line as part of
+    # the opensource release, which is capable of parsing
+    # Windows-style PDB executables.  It should live in the path, or
+    # in the same directory as jeprof.
+    $obj_tool_map{"nm_pdb"} = "nm-pdb";
+    $obj_tool_map{"addr2line_pdb"} = "addr2line-pdb";
+  }
+
+  if ($file_type =~ /Mach-O/) {
+    # OS X uses otool to examine Mach-O files, rather than objdump.
+    $obj_tool_map{"otool"} = "otool";
+    $obj_tool_map{"addr2line"} = "false";  # no addr2line
+    $obj_tool_map{"objdump"} = "false";  # no objdump
+  }
+
+  # Go fill in %obj_tool_map with the pathnames to use:
+  foreach my $tool (keys %obj_tool_map) {
+    $obj_tool_map{$tool} = ConfigureTool($obj_tool_map{$tool});
+  }
+}
+
+# Returns the path of a caller-specified object tool.  If --tools or
+# JEPROF_TOOLS are specified, then returns the full path to the tool
+# with that prefix.  Otherwise, returns the path unmodified (which
+# means we will look for it on PATH).
+sub ConfigureTool {
+  my $tool = shift;
+  my $path;
+
+  # --tools (or $JEPROF_TOOLS) is a comma separated list, where each
+  # item is either a) a pathname prefix, or b) a map of the form
+  # <tool>:<path>.  First we look for an entry of type (b) for our
+  # tool.  If one is found, we use it.  Otherwise, we consider all the
+  # pathname prefixes in turn, until one yields an existing file.  If
+  # none does, we use a default path.
+  my $tools = $main::opt_tools || $ENV{"JEPROF_TOOLS"} || "";
+  if ($tools =~ m/(,|^)\Q$tool\E:([^,]*)/) {
+    $path = $2;
+    # TODO(csilvers): sanity-check that $path exists?  Hard if it's relative.
+  } elsif ($tools ne '') {
+    foreach my $prefix (split(',', $tools)) {
+      next if ($prefix =~ /:/);    # ignore "tool:fullpath" entries in the list
+      if (-x $prefix . $tool) {
+        $path = $prefix . $tool;
+        last;
+      }
+    }
+    if (!$path) {
+      error("No '$tool' found with prefix specified by " .
+            "--tools (or \$JEPROF_TOOLS) '$tools'\n");
+    }
+  } else {
+    # ... otherwise use the version that exists in the same directory as
+    # jeprof.  If there's nothing there, use $PATH.
+    $0 =~ m,[^/]*$,;     # this is everything after the last slash
+    my $dirname = $`;    # this is everything up to and including the last slash
+    if (-x "$dirname$tool") {
+      $path = "$dirname$tool";
+    } else {
+      $path = $tool;
+    }
+  }
+  if ($main::opt_debug) { print STDERR "Using '$path' for '$tool'.\n"; }
+  return $path;
+}
+
+sub ShellEscape {
+  my @escaped_words = ();
+  foreach my $word (@_) {
+    my $escaped_word = $word;
+    if ($word =~ m![^a-zA-Z0-9/.,_=-]!) {  # check for anything not in whitelist
+      $escaped_word =~ s/'/'\\''/;
+      $escaped_word = "'$escaped_word'";
+    }
+    push(@escaped_words, $escaped_word);
+  }
+  return join(" ", @escaped_words);
+}
+
+sub cleanup {
+  unlink($main::tmpfile_sym);
+  unlink(keys %main::tempnames);
+
+  # We leave any collected profiles in $HOME/jeprof in case the user wants
+  # to look at them later.  We print a message informing them of this.
+  if ((scalar(@main::profile_files) > 0) &&
+      defined($main::collected_profile)) {
+    if (scalar(@main::profile_files) == 1) {
+      print STDERR "Dynamically gathered profile is in $main::collected_profile\n";
+    }
+    print STDERR "If you want to investigate this profile further, you can do:\n";
+    print STDERR "\n";
+    print STDERR "  jeprof \\\n";
+    print STDERR "    $main::prog \\\n";
+    print STDERR "    $main::collected_profile\n";
+    print STDERR "\n";
+  }
+}
+
+sub sighandler {
+  cleanup();
+  exit(1);
+}
+
+sub error {
+  my $msg = shift;
+  print STDERR $msg;
+  cleanup();
+  exit(1);
+}
+
+
+# Run $nm_command and get all the resulting procedure boundaries whose
+# names match "$regexp" and returns them in a hashtable mapping from
+# procedure name to a two-element vector of [start address, end address]
+sub GetProcedureBoundariesViaNm {
+  my $escaped_nm_command = shift;    # shell-escaped
+  my $regexp = shift;
+
+  my $symbol_table = {};
+  open(NM, "$escaped_nm_command |") || error("$escaped_nm_command: $!\n");
+  my $last_start = "0";
+  my $routine = "";
+  while (<NM>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (m/^\s*([0-9a-f]+) (.) (..*)/) {
+      my $start_val = $1;
+      my $type = $2;
+      my $this_routine = $3;
+
+      # It's possible for two symbols to share the same address, if
+      # one is a zero-length variable (like __start_google_malloc) or
+      # one symbol is a weak alias to another (like __libc_malloc).
+      # In such cases, we want to ignore all values except for the
+      # actual symbol, which in nm-speak has type "T".  The logic
+      # below does this, though it's a bit tricky: what happens when
+      # we have a series of lines with the same address, is the first
+      # one gets queued up to be processed.  However, it won't
+      # *actually* be processed until later, when we read a line with
+      # a different address.  That means that as long as we're reading
+      # lines with the same address, we have a chance to replace that
+      # item in the queue, which we do whenever we see a 'T' entry --
+      # that is, a line with type 'T'.  If we never see a 'T' entry,
+      # we'll just go ahead and process the first entry (which never
+      # got touched in the queue), and ignore the others.
+      if ($start_val eq $last_start && $type =~ /t/i) {
+        # We are the 'T' symbol at this address, replace previous symbol.
+        $routine = $this_routine;
+        next;
+      } elsif ($start_val eq $last_start) {
+        # We're not the 'T' symbol at this address, so ignore us.
+        next;
+      }
+
+      if ($this_routine eq $sep_symbol) {
+        $sep_address = HexExtend($start_val);
+      }
+
+      # Tag this routine with the starting address in case the image
+      # has multiple occurrences of this routine.  We use a syntax
+      # that resembles template parameters that are automatically
+      # stripped out by ShortFunctionName()
+      $this_routine .= "<$start_val>";
+
+      if (defined($routine) && $routine =~ m/$regexp/) {
+        $symbol_table->{$routine} = [HexExtend($last_start),
+                                     HexExtend($start_val)];
+      }
+      $last_start = $start_val;
+      $routine = $this_routine;
+    } elsif (m/^Loaded image name: (.+)/) {
+      # The win32 nm workalike emits information about the binary it is using.
+      if ($main::opt_debug) { print STDERR "Using Image $1\n"; }
+    } elsif (m/^PDB file name: (.+)/) {
+      # The win32 nm workalike emits information about the pdb it is using.
+      if ($main::opt_debug) { print STDERR "Using PDB $1\n"; }
+    }
+  }
+  close(NM);
+  # Handle the last line in the nm output.  Unfortunately, we don't know
+  # how big this last symbol is, because we don't know how big the file
+  # is.  For now, we just give it a size of 0.
+  # TODO(csilvers): do better here.
+  if (defined($routine) && $routine =~ m/$regexp/) {
+    $symbol_table->{$routine} = [HexExtend($last_start),
+                                 HexExtend($last_start)];
+  }
+  return $symbol_table;
+}
+
+# Gets the procedure boundaries for all routines in "$image" whose names
+# match "$regexp" and returns them in a hashtable mapping from procedure
+# name to a two-element vector of [start address, end address].
+# Will return an empty map if nm is not installed or not working properly.
+sub GetProcedureBoundaries {
+  my $image = shift;
+  my $regexp = shift;
+
+  # If $image doesn't start with /, then put ./ in front of it.  This works
+  # around an obnoxious bug in our probing of nm -f behavior.
+  # "nm -f $image" is supposed to fail on GNU nm, but if:
+  #
+  # a. $image starts with [BbSsPp] (for example, bin/foo/bar), AND
+  # b. you have a.out in your current directory (a not uncommon occurence)
+  #
+  # then "nm -f $image" succeeds because -f only looks at the first letter of
+  # the argument, which looks valid because it's [BbSsPp], and then since
+  # there's no image provided, it looks for a.out and finds it.
+  #
+  # This regex makes sure that $image starts with . or /, forcing the -f
+  # parsing to fail since . and / are not valid formats.
+  $image =~ s#^[^/]#./$&#;
+
+  # For libc libraries, the copy in /usr/lib/debug contains debugging symbols
+  my $debugging = DebuggingLibrary($image);
+  if ($debugging) {
+    $image = $debugging;
+  }
+
+  my $nm = $obj_tool_map{"nm"};
+  my $cppfilt = $obj_tool_map{"c++filt"};
+
+  # nm can fail for two reasons: 1) $image isn't a debug library; 2) nm
+  # binary doesn't support --demangle.  In addition, for OS X we need
+  # to use the -f flag to get 'flat' nm output (otherwise we don't sort
+  # properly and get incorrect results).  Unfortunately, GNU nm uses -f
+  # in an incompatible way.  So first we test whether our nm supports
+  # --demangle and -f.
+  my $demangle_flag = "";
+  my $cppfilt_flag = "";
+  my $to_devnull = ">$dev_null 2>&1";
+  if (system(ShellEscape($nm, "--demangle", $image) . $to_devnull) == 0) {
+    # In this mode, we do "nm --demangle <foo>"
+    $demangle_flag = "--demangle";
+    $cppfilt_flag = "";
+  } elsif (system(ShellEscape($cppfilt, $image) . $to_devnull) == 0) {
+    # In this mode, we do "nm <foo> | c++filt"
+    $cppfilt_flag = " | " . ShellEscape($cppfilt);
+  };
+  my $flatten_flag = "";
+  if (system(ShellEscape($nm, "-f", $image) . $to_devnull) == 0) {
+    $flatten_flag = "-f";
+  }
+
+  # Finally, in the case $imagie isn't a debug library, we try again with
+  # -D to at least get *exported* symbols.  If we can't use --demangle,
+  # we use c++filt instead, if it exists on this system.
+  my @nm_commands = (ShellEscape($nm, "-n", $flatten_flag, $demangle_flag,
+                                 $image) . " 2>$dev_null $cppfilt_flag",
+                     ShellEscape($nm, "-D", "-n", $flatten_flag, $demangle_flag,
+                                 $image) . " 2>$dev_null $cppfilt_flag",
+                     # 6nm is for Go binaries
+                     ShellEscape("6nm", "$image") . " 2>$dev_null | sort",
+                     );
+
+  # If the executable is an MS Windows PDB-format executable, we'll
+  # have set up obj_tool_map("nm_pdb").  In this case, we actually
+  # want to use both unix nm and windows-specific nm_pdb, since
+  # PDB-format executables can apparently include dwarf .o files.
+  if (exists $obj_tool_map{"nm_pdb"}) {
+    push(@nm_commands,
+         ShellEscape($obj_tool_map{"nm_pdb"}, "--demangle", $image)
+         . " 2>$dev_null");
+  }
+
+  foreach my $nm_command (@nm_commands) {
+    my $symbol_table = GetProcedureBoundariesViaNm($nm_command, $regexp);
+    return $symbol_table if (%{$symbol_table});
+  }
+  my $symbol_table = {};
+  return $symbol_table;
+}
+
+
+# The test vectors for AddressAdd/Sub/Inc are 8-16-nibble hex strings.
+# To make them more readable, we add underscores at interesting places.
+# This routine removes the underscores, producing the canonical representation
+# used by jeprof to represent addresses, particularly in the tested routines.
+sub CanonicalHex {
+  my $arg = shift;
+  return join '', (split '_',$arg);
+}
+
+
+# Unit test for AddressAdd:
+sub AddressAddUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressAddUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressAdd ($row->[0], $row->[1]);
+    if ($sum ne $row->[2]) {
+      printf STDERR "ERROR: %s != %s + %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[2];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressAdd 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressAdd (CanonicalHex($row->[0]), CanonicalHex($row->[1]));
+    my $expected = join '', (split '_',$row->[2]);
+    if ($sum ne CanonicalHex($row->[2])) {
+      printf STDERR "ERROR: %s != %s + %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[2];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressAdd 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Unit test for AddressSub:
+sub AddressSubUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressSubUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressSub ($row->[0], $row->[1]);
+    if ($sum ne $row->[3]) {
+      printf STDERR "ERROR: %s != %s - %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[3];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressSub 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressSub (CanonicalHex($row->[0]), CanonicalHex($row->[1]));
+    if ($sum ne CanonicalHex($row->[3])) {
+      printf STDERR "ERROR: %s != %s - %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[3];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressSub 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Unit test for AddressInc:
+sub AddressIncUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressIncUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressInc ($row->[0]);
+    if ($sum ne $row->[4]) {
+      printf STDERR "ERROR: %s != %s + 1 = %s\n", $sum,
+             $row->[0], $row->[4];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressInc 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressInc (CanonicalHex($row->[0]));
+    if ($sum ne CanonicalHex($row->[4])) {
+      printf STDERR "ERROR: %s != %s + 1 = %s\n", $sum,
+             $row->[0], $row->[4];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressInc 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Driver for unit tests.
+# Currently just the address add/subtract/increment routines for 64-bit.
+sub RunUnitTests {
+  my $error_count = 0;
+
+  # This is a list of tuples [a, b, a+b, a-b, a+1]
+  my $unit_test_data_8 = [
+    [qw(aaaaaaaa 50505050 fafafafa 5a5a5a5a aaaaaaab)],
+    [qw(50505050 aaaaaaaa fafafafa a5a5a5a6 50505051)],
+    [qw(ffffffff aaaaaaaa aaaaaaa9 55555555 00000000)],
+    [qw(00000001 ffffffff 00000000 00000002 00000002)],
+    [qw(00000001 fffffff0 fffffff1 00000011 00000002)],
+  ];
+  my $unit_test_data_16 = [
+    # The implementation handles data in 7-nibble chunks, so those are the
+    # interesting boundaries.
+    [qw(aaaaaaaa 50505050
+        00_000000f_afafafa 00_0000005_a5a5a5a 00_000000a_aaaaaab)],
+    [qw(50505050 aaaaaaaa
+        00_000000f_afafafa ff_ffffffa_5a5a5a6 00_0000005_0505051)],
+    [qw(ffffffff aaaaaaaa
+        00_000001a_aaaaaa9 00_0000005_5555555 00_0000010_0000000)],
+    [qw(00000001 ffffffff
+        00_0000010_0000000 ff_ffffff0_0000002 00_0000000_0000002)],
+    [qw(00000001 fffffff0
+        00_000000f_ffffff1 ff_ffffff0_0000011 00_0000000_0000002)],
+
+    [qw(00_a00000a_aaaaaaa 50505050
+        00_a00000f_afafafa 00_a000005_a5a5a5a 00_a00000a_aaaaaab)],
+    [qw(0f_fff0005_0505050 aaaaaaaa
+        0f_fff000f_afafafa 0f_ffefffa_5a5a5a6 0f_fff0005_0505051)],
+    [qw(00_000000f_fffffff 01_800000a_aaaaaaa
+        01_800001a_aaaaaa9 fe_8000005_5555555 00_0000010_0000000)],
+    [qw(00_0000000_0000001 ff_fffffff_fffffff
+        00_0000000_0000000 00_0000000_0000002 00_0000000_0000002)],
+    [qw(00_0000000_0000001 ff_fffffff_ffffff0
+        ff_fffffff_ffffff1 00_0000000_0000011 00_0000000_0000002)],
+  ];
+
+  $error_count += AddressAddUnitTest($unit_test_data_8, $unit_test_data_16);
+  $error_count += AddressSubUnitTest($unit_test_data_8, $unit_test_data_16);
+  $error_count += AddressIncUnitTest($unit_test_data_8, $unit_test_data_16);
+  if ($error_count > 0) {
+    print STDERR $error_count, " errors: FAILED\n";
+  } else {
+    print STDERR "PASS\n";
+  }
+  exit ($error_count);
+}
diff --git a/jemalloc/build-aux/config.guess b/jemalloc/build-aux/config.guess
new file mode 100755
index 0000000..2e9ad7f
--- /dev/null
+++ b/jemalloc/build-aux/config.guess
@@ -0,0 +1,1462 @@
+#! /bin/sh
+# Attempt to guess a canonical system name.
+#   Copyright 1992-2016 Free Software Foundation, Inc.
+
+timestamp='2016-10-02'
+
+# This file is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+#
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that
+# program.  This Exception is an additional permission under section 7
+# of the GNU General Public License, version 3 ("GPLv3").
+#
+# Originally written by Per Bothner; maintained since 2000 by Ben Elliston.
+#
+# You can get the latest version of this script from:
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess
+#
+# Please send patches to <config-patches@gnu.org>.
+
+
+me=`echo "$0" | sed -e 's,.*/,,'`
+
+usage="\
+Usage: $0 [OPTION]
+
+Output the configuration name of the system \`$me' is run on.
+
+Operation modes:
+  -h, --help         print this help, then exit
+  -t, --time-stamp   print date of last modification, then exit
+  -v, --version      print version number, then exit
+
+Report bugs and patches to <config-patches@gnu.org>."
+
+version="\
+GNU config.guess ($timestamp)
+
+Originally written by Per Bothner.
+Copyright 1992-2016 Free Software Foundation, Inc.
+
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
+
+help="
+Try \`$me --help' for more information."
+
+# Parse command line
+while test $# -gt 0 ; do
+  case $1 in
+    --time-stamp | --time* | -t )
+       echo "$timestamp" ; exit ;;
+    --version | -v )
+       echo "$version" ; exit ;;
+    --help | --h* | -h )
+       echo "$usage"; exit ;;
+    -- )     # Stop option processing
+       shift; break ;;
+    - )	# Use stdin as input.
+       break ;;
+    -* )
+       echo "$me: invalid option $1$help" >&2
+       exit 1 ;;
+    * )
+       break ;;
+  esac
+done
+
+if test $# != 0; then
+  echo "$me: too many arguments$help" >&2
+  exit 1
+fi
+
+trap 'exit 1' 1 2 15
+
+# CC_FOR_BUILD -- compiler used by this script. Note that the use of a
+# compiler to aid in system detection is discouraged as it requires
+# temporary files to be created and, as you can see below, it is a
+# headache to deal with in a portable fashion.
+
+# Historically, `CC_FOR_BUILD' used to be named `HOST_CC'. We still
+# use `HOST_CC' if defined, but it is deprecated.
+
+# Portable tmp directory creation inspired by the Autoconf team.
+
+set_cc_for_build='
+trap "exitcode=\$?; (rm -f \$tmpfiles 2>/dev/null; rmdir \$tmp 2>/dev/null) && exit \$exitcode" 0 ;
+trap "rm -f \$tmpfiles 2>/dev/null; rmdir \$tmp 2>/dev/null; exit 1" 1 2 13 15 ;
+: ${TMPDIR=/tmp} ;
+ { tmp=`(umask 077 && mktemp -d "$TMPDIR/cgXXXXXX") 2>/dev/null` && test -n "$tmp" && test -d "$tmp" ; } ||
+ { test -n "$RANDOM" && tmp=$TMPDIR/cg$$-$RANDOM && (umask 077 && mkdir $tmp) ; } ||
+ { tmp=$TMPDIR/cg-$$ && (umask 077 && mkdir $tmp) && echo "Warning: creating insecure temp directory" >&2 ; } ||
+ { echo "$me: cannot create a temporary directory in $TMPDIR" >&2 ; exit 1 ; } ;
+dummy=$tmp/dummy ;
+tmpfiles="$dummy.c $dummy.o $dummy.rel $dummy" ;
+case $CC_FOR_BUILD,$HOST_CC,$CC in
+ ,,)    echo "int x;" > $dummy.c ;
+	for c in cc gcc c89 c99 ; do
+	  if ($c -c -o $dummy.o $dummy.c) >/dev/null 2>&1 ; then
+	     CC_FOR_BUILD="$c"; break ;
+	  fi ;
+	done ;
+	if test x"$CC_FOR_BUILD" = x ; then
+	  CC_FOR_BUILD=no_compiler_found ;
+	fi
+	;;
+ ,,*)   CC_FOR_BUILD=$CC ;;
+ ,*,*)  CC_FOR_BUILD=$HOST_CC ;;
+esac ; set_cc_for_build= ;'
+
+# This is needed to find uname on a Pyramid OSx when run in the BSD universe.
+# (ghazi@noc.rutgers.edu 1994-08-24)
+if (test -f /.attbin/uname) >/dev/null 2>&1 ; then
+	PATH=$PATH:/.attbin ; export PATH
+fi
+
+UNAME_MACHINE=`(uname -m) 2>/dev/null` || UNAME_MACHINE=unknown
+UNAME_RELEASE=`(uname -r) 2>/dev/null` || UNAME_RELEASE=unknown
+UNAME_SYSTEM=`(uname -s) 2>/dev/null`  || UNAME_SYSTEM=unknown
+UNAME_VERSION=`(uname -v) 2>/dev/null` || UNAME_VERSION=unknown
+
+case "${UNAME_SYSTEM}" in
+Linux|GNU|GNU/*)
+	# If the system lacks a compiler, then just pick glibc.
+	# We could probably try harder.
+	LIBC=gnu
+
+	eval $set_cc_for_build
+	cat <<-EOF > $dummy.c
+	#include <features.h>
+	#if defined(__UCLIBC__)
+	LIBC=uclibc
+	#elif defined(__dietlibc__)
+	LIBC=dietlibc
+	#else
+	LIBC=gnu
+	#endif
+	EOF
+	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^LIBC' | sed 's, ,,g'`
+	;;
+esac
+
+# Note: order is significant - the case branches are not exclusive.
+
+case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
+    *:NetBSD:*:*)
+	# NetBSD (nbsd) targets should (where applicable) match one or
+	# more of the tuples: *-*-netbsdelf*, *-*-netbsdaout*,
+	# *-*-netbsdecoff* and *-*-netbsd*.  For targets that recently
+	# switched to ELF, *-*-netbsd* would select the old
+	# object file format.  This provides both forward
+	# compatibility and a consistent mechanism for selecting the
+	# object file format.
+	#
+	# Note: NetBSD doesn't particularly care about the vendor
+	# portion of the name.  We always set it to "unknown".
+	sysctl="sysctl -n hw.machine_arch"
+	UNAME_MACHINE_ARCH=`(uname -p 2>/dev/null || \
+	    /sbin/$sysctl 2>/dev/null || \
+	    /usr/sbin/$sysctl 2>/dev/null || \
+	    echo unknown)`
+	case "${UNAME_MACHINE_ARCH}" in
+	    armeb) machine=armeb-unknown ;;
+	    arm*) machine=arm-unknown ;;
+	    sh3el) machine=shl-unknown ;;
+	    sh3eb) machine=sh-unknown ;;
+	    sh5el) machine=sh5le-unknown ;;
+	    earmv*)
+		arch=`echo ${UNAME_MACHINE_ARCH} | sed -e 's,^e\(armv[0-9]\).*$,\1,'`
+		endian=`echo ${UNAME_MACHINE_ARCH} | sed -ne 's,^.*\(eb\)$,\1,p'`
+		machine=${arch}${endian}-unknown
+		;;
+	    *) machine=${UNAME_MACHINE_ARCH}-unknown ;;
+	esac
+	# The Operating System including object format, if it has switched
+	# to ELF recently (or will in the future) and ABI.
+	case "${UNAME_MACHINE_ARCH}" in
+	    earm*)
+		os=netbsdelf
+		;;
+	    arm*|i386|m68k|ns32k|sh3*|sparc|vax)
+		eval $set_cc_for_build
+		if echo __ELF__ | $CC_FOR_BUILD -E - 2>/dev/null \
+			| grep -q __ELF__
+		then
+		    # Once all utilities can be ECOFF (netbsdecoff) or a.out (netbsdaout).
+		    # Return netbsd for either.  FIX?
+		    os=netbsd
+		else
+		    os=netbsdelf
+		fi
+		;;
+	    *)
+		os=netbsd
+		;;
+	esac
+	# Determine ABI tags.
+	case "${UNAME_MACHINE_ARCH}" in
+	    earm*)
+		expr='s/^earmv[0-9]/-eabi/;s/eb$//'
+		abi=`echo ${UNAME_MACHINE_ARCH} | sed -e "$expr"`
+		;;
+	esac
+	# The OS release
+	# Debian GNU/NetBSD machines have a different userland, and
+	# thus, need a distinct triplet. However, they do not need
+	# kernel version information, so it can be replaced with a
+	# suitable tag, in the style of linux-gnu.
+	case "${UNAME_VERSION}" in
+	    Debian*)
+		release='-gnu'
+		;;
+	    *)
+		release=`echo ${UNAME_RELEASE} | sed -e 's/[-_].*//' | cut -d. -f1,2`
+		;;
+	esac
+	# Since CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM:
+	# contains redundant information, the shorter form:
+	# CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM is used.
+	echo "${machine}-${os}${release}${abi}"
+	exit ;;
+    *:Bitrig:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/Bitrig.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-bitrig${UNAME_RELEASE}
+	exit ;;
+    *:OpenBSD:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/OpenBSD.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-openbsd${UNAME_RELEASE}
+	exit ;;
+    *:LibertyBSD:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/^.*BSD\.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-libertybsd${UNAME_RELEASE}
+	exit ;;
+    *:ekkoBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-ekkobsd${UNAME_RELEASE}
+	exit ;;
+    *:SolidBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-solidbsd${UNAME_RELEASE}
+	exit ;;
+    macppc:MirBSD:*:*)
+	echo powerpc-unknown-mirbsd${UNAME_RELEASE}
+	exit ;;
+    *:MirBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-mirbsd${UNAME_RELEASE}
+	exit ;;
+    *:Sortix:*:*)
+	echo ${UNAME_MACHINE}-unknown-sortix
+	exit ;;
+    alpha:OSF1:*:*)
+	case $UNAME_RELEASE in
+	*4.0)
+		UNAME_RELEASE=`/usr/sbin/sizer -v | awk '{print $3}'`
+		;;
+	*5.*)
+		UNAME_RELEASE=`/usr/sbin/sizer -v | awk '{print $4}'`
+		;;
+	esac
+	# According to Compaq, /usr/sbin/psrinfo has been available on
+	# OSF/1 and Tru64 systems produced since 1995.  I hope that
+	# covers most systems running today.  This code pipes the CPU
+	# types through head -n 1, so we only detect the type of CPU 0.
+	ALPHA_CPU_TYPE=`/usr/sbin/psrinfo -v | sed -n -e 's/^  The alpha \(.*\) processor.*$/\1/p' | head -n 1`
+	case "$ALPHA_CPU_TYPE" in
+	    "EV4 (21064)")
+		UNAME_MACHINE=alpha ;;
+	    "EV4.5 (21064)")
+		UNAME_MACHINE=alpha ;;
+	    "LCA4 (21066/21068)")
+		UNAME_MACHINE=alpha ;;
+	    "EV5 (21164)")
+		UNAME_MACHINE=alphaev5 ;;
+	    "EV5.6 (21164A)")
+		UNAME_MACHINE=alphaev56 ;;
+	    "EV5.6 (21164PC)")
+		UNAME_MACHINE=alphapca56 ;;
+	    "EV5.7 (21164PC)")
+		UNAME_MACHINE=alphapca57 ;;
+	    "EV6 (21264)")
+		UNAME_MACHINE=alphaev6 ;;
+	    "EV6.7 (21264A)")
+		UNAME_MACHINE=alphaev67 ;;
+	    "EV6.8CB (21264C)")
+		UNAME_MACHINE=alphaev68 ;;
+	    "EV6.8AL (21264B)")
+		UNAME_MACHINE=alphaev68 ;;
+	    "EV6.8CX (21264D)")
+		UNAME_MACHINE=alphaev68 ;;
+	    "EV6.9A (21264/EV69A)")
+		UNAME_MACHINE=alphaev69 ;;
+	    "EV7 (21364)")
+		UNAME_MACHINE=alphaev7 ;;
+	    "EV7.9 (21364A)")
+		UNAME_MACHINE=alphaev79 ;;
+	esac
+	# A Pn.n version is a patched version.
+	# A Vn.n version is a released version.
+	# A Tn.n version is a released field test version.
+	# A Xn.n version is an unreleased experimental baselevel.
+	# 1.2 uses "1.2" for uname -r.
+	echo ${UNAME_MACHINE}-dec-osf`echo ${UNAME_RELEASE} | sed -e 's/^[PVTX]//' | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz`
+	# Reset EXIT trap before exiting to avoid spurious non-zero exit code.
+	exitcode=$?
+	trap '' 0
+	exit $exitcode ;;
+    Alpha\ *:Windows_NT*:*)
+	# How do we know it's Interix rather than the generic POSIX subsystem?
+	# Should we change UNAME_MACHINE based on the output of uname instead
+	# of the specific Alpha model?
+	echo alpha-pc-interix
+	exit ;;
+    21064:Windows_NT:50:3)
+	echo alpha-dec-winnt3.5
+	exit ;;
+    Amiga*:UNIX_System_V:4.0:*)
+	echo m68k-unknown-sysv4
+	exit ;;
+    *:[Aa]miga[Oo][Ss]:*:*)
+	echo ${UNAME_MACHINE}-unknown-amigaos
+	exit ;;
+    *:[Mm]orph[Oo][Ss]:*:*)
+	echo ${UNAME_MACHINE}-unknown-morphos
+	exit ;;
+    *:OS/390:*:*)
+	echo i370-ibm-openedition
+	exit ;;
+    *:z/VM:*:*)
+	echo s390-ibm-zvmoe
+	exit ;;
+    *:OS400:*:*)
+	echo powerpc-ibm-os400
+	exit ;;
+    arm:RISC*:1.[012]*:*|arm:riscix:1.[012]*:*)
+	echo arm-acorn-riscix${UNAME_RELEASE}
+	exit ;;
+    arm*:riscos:*:*|arm*:RISCOS:*:*)
+	echo arm-unknown-riscos
+	exit ;;
+    SR2?01:HI-UX/MPP:*:* | SR8000:HI-UX/MPP:*:*)
+	echo hppa1.1-hitachi-hiuxmpp
+	exit ;;
+    Pyramid*:OSx*:*:* | MIS*:OSx*:*:* | MIS*:SMP_DC-OSx*:*:*)
+	# akee@wpdis03.wpafb.af.mil (Earle F. Ake) contributed MIS and NILE.
+	if test "`(/bin/universe) 2>/dev/null`" = att ; then
+		echo pyramid-pyramid-sysv3
+	else
+		echo pyramid-pyramid-bsd
+	fi
+	exit ;;
+    NILE*:*:*:dcosx)
+	echo pyramid-pyramid-svr4
+	exit ;;
+    DRS?6000:unix:4.0:6*)
+	echo sparc-icl-nx6
+	exit ;;
+    DRS?6000:UNIX_SV:4.2*:7* | DRS?6000:isis:4.2*:7*)
+	case `/usr/bin/uname -p` in
+	    sparc) echo sparc-icl-nx7; exit ;;
+	esac ;;
+    s390x:SunOS:*:*)
+	echo ${UNAME_MACHINE}-ibm-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4H:SunOS:5.*:*)
+	echo sparc-hal-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:5.*:* | tadpole*:SunOS:5.*:*)
+	echo sparc-sun-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    i86pc:AuroraUX:5.*:* | i86xen:AuroraUX:5.*:*)
+	echo i386-pc-auroraux${UNAME_RELEASE}
+	exit ;;
+    i86pc:SunOS:5.*:* | i86xen:SunOS:5.*:*)
+	eval $set_cc_for_build
+	SUN_ARCH=i386
+	# If there is a compiler, see if it is configured for 64-bit objects.
+	# Note that the Sun cc does not turn __LP64__ into 1 like gcc does.
+	# This test works for both compilers.
+	if [ "$CC_FOR_BUILD" != no_compiler_found ]; then
+	    if (echo '#ifdef __amd64'; echo IS_64BIT_ARCH; echo '#endif') | \
+		(CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) | \
+		grep IS_64BIT_ARCH >/dev/null
+	    then
+		SUN_ARCH=x86_64
+	    fi
+	fi
+	echo ${SUN_ARCH}-pc-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:6*:*)
+	# According to config.sub, this is the proper way to canonicalize
+	# SunOS6.  Hard to guess exactly what SunOS6 will be like, but
+	# it's likely to be more like Solaris than SunOS4.
+	echo sparc-sun-solaris3`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:*:*)
+	case "`/usr/bin/arch -k`" in
+	    Series*|S4*)
+		UNAME_RELEASE=`uname -v`
+		;;
+	esac
+	# Japanese Language versions have a version number like `4.1.3-JL'.
+	echo sparc-sun-sunos`echo ${UNAME_RELEASE}|sed -e 's/-/_/'`
+	exit ;;
+    sun3*:SunOS:*:*)
+	echo m68k-sun-sunos${UNAME_RELEASE}
+	exit ;;
+    sun*:*:4.2BSD:*)
+	UNAME_RELEASE=`(sed 1q /etc/motd | awk '{print substr($5,1,3)}') 2>/dev/null`
+	test "x${UNAME_RELEASE}" = x && UNAME_RELEASE=3
+	case "`/bin/arch`" in
+	    sun3)
+		echo m68k-sun-sunos${UNAME_RELEASE}
+		;;
+	    sun4)
+		echo sparc-sun-sunos${UNAME_RELEASE}
+		;;
+	esac
+	exit ;;
+    aushp:SunOS:*:*)
+	echo sparc-auspex-sunos${UNAME_RELEASE}
+	exit ;;
+    # The situation for MiNT is a little confusing.  The machine name
+    # can be virtually everything (everything which is not
+    # "atarist" or "atariste" at least should have a processor
+    # > m68000).  The system name ranges from "MiNT" over "FreeMiNT"
+    # to the lowercase version "mint" (or "freemint").  Finally
+    # the system name "TOS" denotes a system which is actually not
+    # MiNT.  But MiNT is downward compatible to TOS, so this should
+    # be no problem.
+    atarist[e]:*MiNT:*:* | atarist[e]:*mint:*:* | atarist[e]:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    atari*:*MiNT:*:* | atari*:*mint:*:* | atarist[e]:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    *falcon*:*MiNT:*:* | *falcon*:*mint:*:* | *falcon*:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    milan*:*MiNT:*:* | milan*:*mint:*:* | *milan*:*TOS:*:*)
+	echo m68k-milan-mint${UNAME_RELEASE}
+	exit ;;
+    hades*:*MiNT:*:* | hades*:*mint:*:* | *hades*:*TOS:*:*)
+	echo m68k-hades-mint${UNAME_RELEASE}
+	exit ;;
+    *:*MiNT:*:* | *:*mint:*:* | *:*TOS:*:*)
+	echo m68k-unknown-mint${UNAME_RELEASE}
+	exit ;;
+    m68k:machten:*:*)
+	echo m68k-apple-machten${UNAME_RELEASE}
+	exit ;;
+    powerpc:machten:*:*)
+	echo powerpc-apple-machten${UNAME_RELEASE}
+	exit ;;
+    RISC*:Mach:*:*)
+	echo mips-dec-mach_bsd4.3
+	exit ;;
+    RISC*:ULTRIX:*:*)
+	echo mips-dec-ultrix${UNAME_RELEASE}
+	exit ;;
+    VAX*:ULTRIX*:*:*)
+	echo vax-dec-ultrix${UNAME_RELEASE}
+	exit ;;
+    2020:CLIX:*:* | 2430:CLIX:*:*)
+	echo clipper-intergraph-clix${UNAME_RELEASE}
+	exit ;;
+    mips:*:*:UMIPS | mips:*:*:RISCos)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+#ifdef __cplusplus
+#include <stdio.h>  /* for printf() prototype */
+	int main (int argc, char *argv[]) {
+#else
+	int main (argc, argv) int argc; char *argv[]; {
+#endif
+	#if defined (host_mips) && defined (MIPSEB)
+	#if defined (SYSTYPE_SYSV)
+	  printf ("mips-mips-riscos%ssysv\n", argv[1]); exit (0);
+	#endif
+	#if defined (SYSTYPE_SVR4)
+	  printf ("mips-mips-riscos%ssvr4\n", argv[1]); exit (0);
+	#endif
+	#if defined (SYSTYPE_BSD43) || defined(SYSTYPE_BSD)
+	  printf ("mips-mips-riscos%sbsd\n", argv[1]); exit (0);
+	#endif
+	#endif
+	  exit (-1);
+	}
+EOF
+	$CC_FOR_BUILD -o $dummy $dummy.c &&
+	  dummyarg=`echo "${UNAME_RELEASE}" | sed -n 's/\([0-9]*\).*/\1/p'` &&
+	  SYSTEM_NAME=`$dummy $dummyarg` &&
+	    { echo "$SYSTEM_NAME"; exit; }
+	echo mips-mips-riscos${UNAME_RELEASE}
+	exit ;;
+    Motorola:PowerMAX_OS:*:*)
+	echo powerpc-motorola-powermax
+	exit ;;
+    Motorola:*:4.3:PL8-*)
+	echo powerpc-harris-powermax
+	exit ;;
+    Night_Hawk:*:*:PowerMAX_OS | Synergy:PowerMAX_OS:*:*)
+	echo powerpc-harris-powermax
+	exit ;;
+    Night_Hawk:Power_UNIX:*:*)
+	echo powerpc-harris-powerunix
+	exit ;;
+    m88k:CX/UX:7*:*)
+	echo m88k-harris-cxux7
+	exit ;;
+    m88k:*:4*:R4*)
+	echo m88k-motorola-sysv4
+	exit ;;
+    m88k:*:3*:R3*)
+	echo m88k-motorola-sysv3
+	exit ;;
+    AViiON:dgux:*:*)
+	# DG/UX returns AViiON for all architectures
+	UNAME_PROCESSOR=`/usr/bin/uname -p`
+	if [ $UNAME_PROCESSOR = mc88100 ] || [ $UNAME_PROCESSOR = mc88110 ]
+	then
+	    if [ ${TARGET_BINARY_INTERFACE}x = m88kdguxelfx ] || \
+	       [ ${TARGET_BINARY_INTERFACE}x = x ]
+	    then
+		echo m88k-dg-dgux${UNAME_RELEASE}
+	    else
+		echo m88k-dg-dguxbcs${UNAME_RELEASE}
+	    fi
+	else
+	    echo i586-dg-dgux${UNAME_RELEASE}
+	fi
+	exit ;;
+    M88*:DolphinOS:*:*)	# DolphinOS (SVR3)
+	echo m88k-dolphin-sysv3
+	exit ;;
+    M88*:*:R3*:*)
+	# Delta 88k system running SVR3
+	echo m88k-motorola-sysv3
+	exit ;;
+    XD88*:*:*:*) # Tektronix XD88 system running UTekV (SVR3)
+	echo m88k-tektronix-sysv3
+	exit ;;
+    Tek43[0-9][0-9]:UTek:*:*) # Tektronix 4300 system running UTek (BSD)
+	echo m68k-tektronix-bsd
+	exit ;;
+    *:IRIX*:*:*)
+	echo mips-sgi-irix`echo ${UNAME_RELEASE}|sed -e 's/-/_/g'`
+	exit ;;
+    ????????:AIX?:[12].1:2)   # AIX 2.2.1 or AIX 2.1.1 is RT/PC AIX.
+	echo romp-ibm-aix     # uname -m gives an 8 hex-code CPU id
+	exit ;;               # Note that: echo "'`uname -s`'" gives 'AIX '
+    i*86:AIX:*:*)
+	echo i386-ibm-aix
+	exit ;;
+    ia64:AIX:*:*)
+	if [ -x /usr/bin/oslevel ] ; then
+		IBM_REV=`/usr/bin/oslevel`
+	else
+		IBM_REV=${UNAME_VERSION}.${UNAME_RELEASE}
+	fi
+	echo ${UNAME_MACHINE}-ibm-aix${IBM_REV}
+	exit ;;
+    *:AIX:2:3)
+	if grep bos325 /usr/include/stdio.h >/dev/null 2>&1; then
+		eval $set_cc_for_build
+		sed 's/^		//' << EOF >$dummy.c
+		#include <sys/systemcfg.h>
+
+		main()
+			{
+			if (!__power_pc())
+				exit(1);
+			puts("powerpc-ibm-aix3.2.5");
+			exit(0);
+			}
+EOF
+		if $CC_FOR_BUILD -o $dummy $dummy.c && SYSTEM_NAME=`$dummy`
+		then
+			echo "$SYSTEM_NAME"
+		else
+			echo rs6000-ibm-aix3.2.5
+		fi
+	elif grep bos324 /usr/include/stdio.h >/dev/null 2>&1; then
+		echo rs6000-ibm-aix3.2.4
+	else
+		echo rs6000-ibm-aix3.2
+	fi
+	exit ;;
+    *:AIX:*:[4567])
+	IBM_CPU_ID=`/usr/sbin/lsdev -C -c processor -S available | sed 1q | awk '{ print $1 }'`
+	if /usr/sbin/lsattr -El ${IBM_CPU_ID} | grep ' POWER' >/dev/null 2>&1; then
+		IBM_ARCH=rs6000
+	else
+		IBM_ARCH=powerpc
+	fi
+	if [ -x /usr/bin/lslpp ] ; then
+		IBM_REV=`/usr/bin/lslpp -Lqc bos.rte.libc |
+			   awk -F: '{ print $3 }' | sed s/[0-9]*$/0/`
+	else
+		IBM_REV=${UNAME_VERSION}.${UNAME_RELEASE}
+	fi
+	echo ${IBM_ARCH}-ibm-aix${IBM_REV}
+	exit ;;
+    *:AIX:*:*)
+	echo rs6000-ibm-aix
+	exit ;;
+    ibmrt:4.4BSD:*|romp-ibm:BSD:*)
+	echo romp-ibm-bsd4.4
+	exit ;;
+    ibmrt:*BSD:*|romp-ibm:BSD:*)            # covers RT/PC BSD and
+	echo romp-ibm-bsd${UNAME_RELEASE}   # 4.3 with uname added to
+	exit ;;                             # report: romp-ibm BSD 4.3
+    *:BOSX:*:*)
+	echo rs6000-bull-bosx
+	exit ;;
+    DPX/2?00:B.O.S.:*:*)
+	echo m68k-bull-sysv3
+	exit ;;
+    9000/[34]??:4.3bsd:1.*:*)
+	echo m68k-hp-bsd
+	exit ;;
+    hp300:4.4BSD:*:* | 9000/[34]??:4.3bsd:2.*:*)
+	echo m68k-hp-bsd4.4
+	exit ;;
+    9000/[34678]??:HP-UX:*:*)
+	HPUX_REV=`echo ${UNAME_RELEASE}|sed -e 's/[^.]*.[0B]*//'`
+	case "${UNAME_MACHINE}" in
+	    9000/31? )            HP_ARCH=m68000 ;;
+	    9000/[34]?? )         HP_ARCH=m68k ;;
+	    9000/[678][0-9][0-9])
+		if [ -x /usr/bin/getconf ]; then
+		    sc_cpu_version=`/usr/bin/getconf SC_CPU_VERSION 2>/dev/null`
+		    sc_kernel_bits=`/usr/bin/getconf SC_KERNEL_BITS 2>/dev/null`
+		    case "${sc_cpu_version}" in
+		      523) HP_ARCH=hppa1.0 ;; # CPU_PA_RISC1_0
+		      528) HP_ARCH=hppa1.1 ;; # CPU_PA_RISC1_1
+		      532)                      # CPU_PA_RISC2_0
+			case "${sc_kernel_bits}" in
+			  32) HP_ARCH=hppa2.0n ;;
+			  64) HP_ARCH=hppa2.0w ;;
+			  '') HP_ARCH=hppa2.0 ;;   # HP-UX 10.20
+			esac ;;
+		    esac
+		fi
+		if [ "${HP_ARCH}" = "" ]; then
+		    eval $set_cc_for_build
+		    sed 's/^		//' << EOF >$dummy.c
+
+		#define _HPUX_SOURCE
+		#include <stdlib.h>
+		#include <unistd.h>
+
+		int main ()
+		{
+		#if defined(_SC_KERNEL_BITS)
+		    long bits = sysconf(_SC_KERNEL_BITS);
+		#endif
+		    long cpu  = sysconf (_SC_CPU_VERSION);
+
+		    switch (cpu)
+			{
+			case CPU_PA_RISC1_0: puts ("hppa1.0"); break;
+			case CPU_PA_RISC1_1: puts ("hppa1.1"); break;
+			case CPU_PA_RISC2_0:
+		#if defined(_SC_KERNEL_BITS)
+			    switch (bits)
+				{
+				case 64: puts ("hppa2.0w"); break;
+				case 32: puts ("hppa2.0n"); break;
+				default: puts ("hppa2.0"); break;
+				} break;
+		#else  /* !defined(_SC_KERNEL_BITS) */
+			    puts ("hppa2.0"); break;
+		#endif
+			default: puts ("hppa1.0"); break;
+			}
+		    exit (0);
+		}
+EOF
+		    (CCOPTS="" $CC_FOR_BUILD -o $dummy $dummy.c 2>/dev/null) && HP_ARCH=`$dummy`
+		    test -z "$HP_ARCH" && HP_ARCH=hppa
+		fi ;;
+	esac
+	if [ ${HP_ARCH} = hppa2.0w ]
+	then
+	    eval $set_cc_for_build
+
+	    # hppa2.0w-hp-hpux* has a 64-bit kernel and a compiler generating
+	    # 32-bit code.  hppa64-hp-hpux* has the same kernel and a compiler
+	    # generating 64-bit code.  GNU and HP use different nomenclature:
+	    #
+	    # $ CC_FOR_BUILD=cc ./config.guess
+	    # => hppa2.0w-hp-hpux11.23
+	    # $ CC_FOR_BUILD="cc +DA2.0w" ./config.guess
+	    # => hppa64-hp-hpux11.23
+
+	    if echo __LP64__ | (CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) |
+		grep -q __LP64__
+	    then
+		HP_ARCH=hppa2.0w
+	    else
+		HP_ARCH=hppa64
+	    fi
+	fi
+	echo ${HP_ARCH}-hp-hpux${HPUX_REV}
+	exit ;;
+    ia64:HP-UX:*:*)
+	HPUX_REV=`echo ${UNAME_RELEASE}|sed -e 's/[^.]*.[0B]*//'`
+	echo ia64-hp-hpux${HPUX_REV}
+	exit ;;
+    3050*:HI-UX:*:*)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+	#include <unistd.h>
+	int
+	main ()
+	{
+	  long cpu = sysconf (_SC_CPU_VERSION);
+	  /* The order matters, because CPU_IS_HP_MC68K erroneously returns
+	     true for CPU_PA_RISC1_0.  CPU_IS_PA_RISC returns correct
+	     results, however.  */
+	  if (CPU_IS_PA_RISC (cpu))
+	    {
+	      switch (cpu)
+		{
+		  case CPU_PA_RISC1_0: puts ("hppa1.0-hitachi-hiuxwe2"); break;
+		  case CPU_PA_RISC1_1: puts ("hppa1.1-hitachi-hiuxwe2"); break;
+		  case CPU_PA_RISC2_0: puts ("hppa2.0-hitachi-hiuxwe2"); break;
+		  default: puts ("hppa-hitachi-hiuxwe2"); break;
+		}
+	    }
+	  else if (CPU_IS_HP_MC68K (cpu))
+	    puts ("m68k-hitachi-hiuxwe2");
+	  else puts ("unknown-hitachi-hiuxwe2");
+	  exit (0);
+	}
+EOF
+	$CC_FOR_BUILD -o $dummy $dummy.c && SYSTEM_NAME=`$dummy` &&
+		{ echo "$SYSTEM_NAME"; exit; }
+	echo unknown-hitachi-hiuxwe2
+	exit ;;
+    9000/7??:4.3bsd:*:* | 9000/8?[79]:4.3bsd:*:* )
+	echo hppa1.1-hp-bsd
+	exit ;;
+    9000/8??:4.3bsd:*:*)
+	echo hppa1.0-hp-bsd
+	exit ;;
+    *9??*:MPE/iX:*:* | *3000*:MPE/iX:*:*)
+	echo hppa1.0-hp-mpeix
+	exit ;;
+    hp7??:OSF1:*:* | hp8?[79]:OSF1:*:* )
+	echo hppa1.1-hp-osf
+	exit ;;
+    hp8??:OSF1:*:*)
+	echo hppa1.0-hp-osf
+	exit ;;
+    i*86:OSF1:*:*)
+	if [ -x /usr/sbin/sysversion ] ; then
+	    echo ${UNAME_MACHINE}-unknown-osf1mk
+	else
+	    echo ${UNAME_MACHINE}-unknown-osf1
+	fi
+	exit ;;
+    parisc*:Lites*:*:*)
+	echo hppa1.1-hp-lites
+	exit ;;
+    C1*:ConvexOS:*:* | convex:ConvexOS:C1*:*)
+	echo c1-convex-bsd
+	exit ;;
+    C2*:ConvexOS:*:* | convex:ConvexOS:C2*:*)
+	if getsysinfo -f scalar_acc
+	then echo c32-convex-bsd
+	else echo c2-convex-bsd
+	fi
+	exit ;;
+    C34*:ConvexOS:*:* | convex:ConvexOS:C34*:*)
+	echo c34-convex-bsd
+	exit ;;
+    C38*:ConvexOS:*:* | convex:ConvexOS:C38*:*)
+	echo c38-convex-bsd
+	exit ;;
+    C4*:ConvexOS:*:* | convex:ConvexOS:C4*:*)
+	echo c4-convex-bsd
+	exit ;;
+    CRAY*Y-MP:*:*:*)
+	echo ymp-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*[A-Z]90:*:*:*)
+	echo ${UNAME_MACHINE}-cray-unicos${UNAME_RELEASE} \
+	| sed -e 's/CRAY.*\([A-Z]90\)/\1/' \
+	      -e y/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/ \
+	      -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*TS:*:*:*)
+	echo t90-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*T3E:*:*:*)
+	echo alphaev5-cray-unicosmk${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*SV1:*:*:*)
+	echo sv1-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    *:UNICOS/mp:*:*)
+	echo craynv-cray-unicosmp${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    F30[01]:UNIX_System_V:*:* | F700:UNIX_System_V:*:*)
+	FUJITSU_PROC=`uname -m | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz`
+	FUJITSU_SYS=`uname -p | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/\///'`
+	FUJITSU_REL=`echo ${UNAME_RELEASE} | sed -e 's/ /_/'`
+	echo "${FUJITSU_PROC}-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
+	exit ;;
+    5000:UNIX_System_V:4.*:*)
+	FUJITSU_SYS=`uname -p | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/\///'`
+	FUJITSU_REL=`echo ${UNAME_RELEASE} | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/ /_/'`
+	echo "sparc-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
+	exit ;;
+    i*86:BSD/386:*:* | i*86:BSD/OS:*:* | *:Ascend\ Embedded/OS:*:*)
+	echo ${UNAME_MACHINE}-pc-bsdi${UNAME_RELEASE}
+	exit ;;
+    sparc*:BSD/OS:*:*)
+	echo sparc-unknown-bsdi${UNAME_RELEASE}
+	exit ;;
+    *:BSD/OS:*:*)
+	echo ${UNAME_MACHINE}-unknown-bsdi${UNAME_RELEASE}
+	exit ;;
+    *:FreeBSD:*:*)
+	UNAME_PROCESSOR=`/usr/bin/uname -p`
+	case ${UNAME_PROCESSOR} in
+	    amd64)
+		echo x86_64-unknown-freebsd`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'` ;;
+	    *)
+		echo ${UNAME_PROCESSOR}-unknown-freebsd`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'` ;;
+	esac
+	exit ;;
+    i*:CYGWIN*:*)
+	echo ${UNAME_MACHINE}-pc-cygwin
+	exit ;;
+    *:MINGW64*:*)
+	echo ${UNAME_MACHINE}-pc-mingw64
+	exit ;;
+    *:MINGW*:*)
+	echo ${UNAME_MACHINE}-pc-mingw32
+	exit ;;
+    *:MSYS*:*)
+	echo ${UNAME_MACHINE}-pc-msys
+	exit ;;
+    i*:windows32*:*)
+	# uname -m includes "-pc" on this system.
+	echo ${UNAME_MACHINE}-mingw32
+	exit ;;
+    i*:PW*:*)
+	echo ${UNAME_MACHINE}-pc-pw32
+	exit ;;
+    *:Interix*:*)
+	case ${UNAME_MACHINE} in
+	    x86)
+		echo i586-pc-interix${UNAME_RELEASE}
+		exit ;;
+	    authenticamd | genuineintel | EM64T)
+		echo x86_64-unknown-interix${UNAME_RELEASE}
+		exit ;;
+	    IA64)
+		echo ia64-unknown-interix${UNAME_RELEASE}
+		exit ;;
+	esac ;;
+    [345]86:Windows_95:* | [345]86:Windows_98:* | [345]86:Windows_NT:*)
+	echo i${UNAME_MACHINE}-pc-mks
+	exit ;;
+    8664:Windows_NT:*)
+	echo x86_64-pc-mks
+	exit ;;
+    i*:Windows_NT*:* | Pentium*:Windows_NT*:*)
+	# How do we know it's Interix rather than the generic POSIX subsystem?
+	# It also conflicts with pre-2.0 versions of AT&T UWIN. Should we
+	# UNAME_MACHINE based on the output of uname instead of i386?
+	echo i586-pc-interix
+	exit ;;
+    i*:UWIN*:*)
+	echo ${UNAME_MACHINE}-pc-uwin
+	exit ;;
+    amd64:CYGWIN*:*:* | x86_64:CYGWIN*:*:*)
+	echo x86_64-unknown-cygwin
+	exit ;;
+    p*:CYGWIN*:*)
+	echo powerpcle-unknown-cygwin
+	exit ;;
+    prep*:SunOS:5.*:*)
+	echo powerpcle-unknown-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    *:GNU:*:*)
+	# the GNU system
+	echo `echo ${UNAME_MACHINE}|sed -e 's,[-/].*$,,'`-unknown-${LIBC}`echo ${UNAME_RELEASE}|sed -e 's,/.*$,,'`
+	exit ;;
+    *:GNU/*:*:*)
+	# other systems with GNU libc and userland
+	echo ${UNAME_MACHINE}-unknown-`echo ${UNAME_SYSTEM} | sed 's,^[^/]*/,,' | tr "[:upper:]" "[:lower:]"``echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`-${LIBC}
+	exit ;;
+    i*86:Minix:*:*)
+	echo ${UNAME_MACHINE}-pc-minix
+	exit ;;
+    aarch64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    aarch64_be:Linux:*:*)
+	UNAME_MACHINE=aarch64_be
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    alpha:Linux:*:*)
+	case `sed -n '/^cpu model/s/^.*: \(.*\)/\1/p' < /proc/cpuinfo` in
+	  EV5)   UNAME_MACHINE=alphaev5 ;;
+	  EV56)  UNAME_MACHINE=alphaev56 ;;
+	  PCA56) UNAME_MACHINE=alphapca56 ;;
+	  PCA57) UNAME_MACHINE=alphapca56 ;;
+	  EV6)   UNAME_MACHINE=alphaev6 ;;
+	  EV67)  UNAME_MACHINE=alphaev67 ;;
+	  EV68*) UNAME_MACHINE=alphaev68 ;;
+	esac
+	objdump --private-headers /bin/sh | grep -q ld.so.1
+	if test "$?" = 0 ; then LIBC=gnulibc1 ; fi
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    arc:Linux:*:* | arceb:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    arm*:Linux:*:*)
+	eval $set_cc_for_build
+	if echo __ARM_EABI__ | $CC_FOR_BUILD -E - 2>/dev/null \
+	    | grep -q __ARM_EABI__
+	then
+	    echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	else
+	    if echo __ARM_PCS_VFP | $CC_FOR_BUILD -E - 2>/dev/null \
+		| grep -q __ARM_PCS_VFP
+	    then
+		echo ${UNAME_MACHINE}-unknown-linux-${LIBC}eabi
+	    else
+		echo ${UNAME_MACHINE}-unknown-linux-${LIBC}eabihf
+	    fi
+	fi
+	exit ;;
+    avr32*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    cris:Linux:*:*)
+	echo ${UNAME_MACHINE}-axis-linux-${LIBC}
+	exit ;;
+    crisv32:Linux:*:*)
+	echo ${UNAME_MACHINE}-axis-linux-${LIBC}
+	exit ;;
+    e2k:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    frv:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    hexagon:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    i*86:Linux:*:*)
+	echo ${UNAME_MACHINE}-pc-linux-${LIBC}
+	exit ;;
+    ia64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    k1om:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    m32r*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    m68*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    mips:Linux:*:* | mips64:Linux:*:*)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+	#undef CPU
+	#undef ${UNAME_MACHINE}
+	#undef ${UNAME_MACHINE}el
+	#if defined(__MIPSEL__) || defined(__MIPSEL) || defined(_MIPSEL) || defined(MIPSEL)
+	CPU=${UNAME_MACHINE}el
+	#else
+	#if defined(__MIPSEB__) || defined(__MIPSEB) || defined(_MIPSEB) || defined(MIPSEB)
+	CPU=${UNAME_MACHINE}
+	#else
+	CPU=
+	#endif
+	#endif
+EOF
+	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^CPU'`
+	test x"${CPU}" != x && { echo "${CPU}-unknown-linux-${LIBC}"; exit; }
+	;;
+    mips64el:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    openrisc*:Linux:*:*)
+	echo or1k-unknown-linux-${LIBC}
+	exit ;;
+    or32:Linux:*:* | or1k*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    padre:Linux:*:*)
+	echo sparc-unknown-linux-${LIBC}
+	exit ;;
+    parisc64:Linux:*:* | hppa64:Linux:*:*)
+	echo hppa64-unknown-linux-${LIBC}
+	exit ;;
+    parisc:Linux:*:* | hppa:Linux:*:*)
+	# Look for CPU level
+	case `grep '^cpu[^a-z]*:' /proc/cpuinfo 2>/dev/null | cut -d' ' -f2` in
+	  PA7*) echo hppa1.1-unknown-linux-${LIBC} ;;
+	  PA8*) echo hppa2.0-unknown-linux-${LIBC} ;;
+	  *)    echo hppa-unknown-linux-${LIBC} ;;
+	esac
+	exit ;;
+    ppc64:Linux:*:*)
+	echo powerpc64-unknown-linux-${LIBC}
+	exit ;;
+    ppc:Linux:*:*)
+	echo powerpc-unknown-linux-${LIBC}
+	exit ;;
+    ppc64le:Linux:*:*)
+	echo powerpc64le-unknown-linux-${LIBC}
+	exit ;;
+    ppcle:Linux:*:*)
+	echo powerpcle-unknown-linux-${LIBC}
+	exit ;;
+    riscv32:Linux:*:* | riscv64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    s390:Linux:*:* | s390x:Linux:*:*)
+	echo ${UNAME_MACHINE}-ibm-linux-${LIBC}
+	exit ;;
+    sh64*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    sh*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    sparc:Linux:*:* | sparc64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    tile*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    vax:Linux:*:*)
+	echo ${UNAME_MACHINE}-dec-linux-${LIBC}
+	exit ;;
+    x86_64:Linux:*:*)
+	echo ${UNAME_MACHINE}-pc-linux-${LIBC}
+	exit ;;
+    xtensa*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    i*86:DYNIX/ptx:4*:*)
+	# ptx 4.0 does uname -s correctly, with DYNIX/ptx in there.
+	# earlier versions are messed up and put the nodename in both
+	# sysname and nodename.
+	echo i386-sequent-sysv4
+	exit ;;
+    i*86:UNIX_SV:4.2MP:2.*)
+	# Unixware is an offshoot of SVR4, but it has its own version
+	# number series starting with 2...
+	# I am not positive that other SVR4 systems won't match this,
+	# I just have to hope.  -- rms.
+	# Use sysv4.2uw... so that sysv4* matches it.
+	echo ${UNAME_MACHINE}-pc-sysv4.2uw${UNAME_VERSION}
+	exit ;;
+    i*86:OS/2:*:*)
+	# If we were able to find `uname', then EMX Unix compatibility
+	# is probably installed.
+	echo ${UNAME_MACHINE}-pc-os2-emx
+	exit ;;
+    i*86:XTS-300:*:STOP)
+	echo ${UNAME_MACHINE}-unknown-stop
+	exit ;;
+    i*86:atheos:*:*)
+	echo ${UNAME_MACHINE}-unknown-atheos
+	exit ;;
+    i*86:syllable:*:*)
+	echo ${UNAME_MACHINE}-pc-syllable
+	exit ;;
+    i*86:LynxOS:2.*:* | i*86:LynxOS:3.[01]*:* | i*86:LynxOS:4.[02]*:*)
+	echo i386-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    i*86:*DOS:*:*)
+	echo ${UNAME_MACHINE}-pc-msdosdjgpp
+	exit ;;
+    i*86:*:4.*:* | i*86:SYSTEM_V:4.*:*)
+	UNAME_REL=`echo ${UNAME_RELEASE} | sed 's/\/MP$//'`
+	if grep Novell /usr/include/link.h >/dev/null 2>/dev/null; then
+		echo ${UNAME_MACHINE}-univel-sysv${UNAME_REL}
+	else
+		echo ${UNAME_MACHINE}-pc-sysv${UNAME_REL}
+	fi
+	exit ;;
+    i*86:*:5:[678]*)
+	# UnixWare 7.x, OpenUNIX and OpenServer 6.
+	case `/bin/uname -X | grep "^Machine"` in
+	    *486*)	     UNAME_MACHINE=i486 ;;
+	    *Pentium)	     UNAME_MACHINE=i586 ;;
+	    *Pent*|*Celeron) UNAME_MACHINE=i686 ;;
+	esac
+	echo ${UNAME_MACHINE}-unknown-sysv${UNAME_RELEASE}${UNAME_SYSTEM}${UNAME_VERSION}
+	exit ;;
+    i*86:*:3.2:*)
+	if test -f /usr/options/cb.name; then
+		UNAME_REL=`sed -n 's/.*Version //p' </usr/options/cb.name`
+		echo ${UNAME_MACHINE}-pc-isc$UNAME_REL
+	elif /bin/uname -X 2>/dev/null >/dev/null ; then
+		UNAME_REL=`(/bin/uname -X|grep Release|sed -e 's/.*= //')`
+		(/bin/uname -X|grep i80486 >/dev/null) && UNAME_MACHINE=i486
+		(/bin/uname -X|grep '^Machine.*Pentium' >/dev/null) \
+			&& UNAME_MACHINE=i586
+		(/bin/uname -X|grep '^Machine.*Pent *II' >/dev/null) \
+			&& UNAME_MACHINE=i686
+		(/bin/uname -X|grep '^Machine.*Pentium Pro' >/dev/null) \
+			&& UNAME_MACHINE=i686
+		echo ${UNAME_MACHINE}-pc-sco$UNAME_REL
+	else
+		echo ${UNAME_MACHINE}-pc-sysv32
+	fi
+	exit ;;
+    pc:*:*:*)
+	# Left here for compatibility:
+	# uname -m prints for DJGPP always 'pc', but it prints nothing about
+	# the processor, so we play safe by assuming i586.
+	# Note: whatever this is, it MUST be the same as what config.sub
+	# prints for the "djgpp" host, or else GDB configure will decide that
+	# this is a cross-build.
+	echo i586-pc-msdosdjgpp
+	exit ;;
+    Intel:Mach:3*:*)
+	echo i386-pc-mach3
+	exit ;;
+    paragon:*:*:*)
+	echo i860-intel-osf1
+	exit ;;
+    i860:*:4.*:*) # i860-SVR4
+	if grep Stardent /usr/include/sys/uadmin.h >/dev/null 2>&1 ; then
+	  echo i860-stardent-sysv${UNAME_RELEASE} # Stardent Vistra i860-SVR4
+	else # Add other i860-SVR4 vendors below as they are discovered.
+	  echo i860-unknown-sysv${UNAME_RELEASE}  # Unknown i860-SVR4
+	fi
+	exit ;;
+    mini*:CTIX:SYS*5:*)
+	# "miniframe"
+	echo m68010-convergent-sysv
+	exit ;;
+    mc68k:UNIX:SYSTEM5:3.51m)
+	echo m68k-convergent-sysv
+	exit ;;
+    M680?0:D-NIX:5.3:*)
+	echo m68k-diab-dnix
+	exit ;;
+    M68*:*:R3V[5678]*:*)
+	test -r /sysV68 && { echo 'm68k-motorola-sysv'; exit; } ;;
+    3[345]??:*:4.0:3.0 | 3[34]??A:*:4.0:3.0 | 3[34]??,*:*:4.0:3.0 | 3[34]??/*:*:4.0:3.0 | 4400:*:4.0:3.0 | 4850:*:4.0:3.0 | SKA40:*:4.0:3.0 | SDS2:*:4.0:3.0 | SHG2:*:4.0:3.0 | S7501*:*:4.0:3.0)
+	OS_REL=''
+	test -r /etc/.relid \
+	&& OS_REL=.`sed -n 's/[^ ]* [^ ]* \([0-9][0-9]\).*/\1/p' < /etc/.relid`
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	  && { echo i486-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep entium >/dev/null \
+	  && { echo i586-ncr-sysv4.3${OS_REL}; exit; } ;;
+    3[34]??:*:4.0:* | 3[34]??,*:*:4.0:*)
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	  && { echo i486-ncr-sysv4; exit; } ;;
+    NCR*:*:4.2:* | MPRAS*:*:4.2:*)
+	OS_REL='.3'
+	test -r /etc/.relid \
+	    && OS_REL=.`sed -n 's/[^ ]* [^ ]* \([0-9][0-9]\).*/\1/p' < /etc/.relid`
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	    && { echo i486-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep entium >/dev/null \
+	    && { echo i586-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep pteron >/dev/null \
+	    && { echo i586-ncr-sysv4.3${OS_REL}; exit; } ;;
+    m68*:LynxOS:2.*:* | m68*:LynxOS:3.0*:*)
+	echo m68k-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    mc68030:UNIX_System_V:4.*:*)
+	echo m68k-atari-sysv4
+	exit ;;
+    TSUNAMI:LynxOS:2.*:*)
+	echo sparc-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    rs6000:LynxOS:2.*:*)
+	echo rs6000-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    PowerPC:LynxOS:2.*:* | PowerPC:LynxOS:3.[01]*:* | PowerPC:LynxOS:4.[02]*:*)
+	echo powerpc-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    SM[BE]S:UNIX_SV:*:*)
+	echo mips-dde-sysv${UNAME_RELEASE}
+	exit ;;
+    RM*:ReliantUNIX-*:*:*)
+	echo mips-sni-sysv4
+	exit ;;
+    RM*:SINIX-*:*:*)
+	echo mips-sni-sysv4
+	exit ;;
+    *:SINIX-*:*:*)
+	if uname -p 2>/dev/null >/dev/null ; then
+		UNAME_MACHINE=`(uname -p) 2>/dev/null`
+		echo ${UNAME_MACHINE}-sni-sysv4
+	else
+		echo ns32k-sni-sysv
+	fi
+	exit ;;
+    PENTIUM:*:4.0*:*)	# Unisys `ClearPath HMP IX 4000' SVR4/MP effort
+			# says <Richard.M.Bartel@ccMail.Census.GOV>
+	echo i586-unisys-sysv4
+	exit ;;
+    *:UNIX_System_V:4*:FTX*)
+	# From Gerald Hewes <hewes@openmarket.com>.
+	# How about differentiating between stratus architectures? -djm
+	echo hppa1.1-stratus-sysv4
+	exit ;;
+    *:*:*:FTX*)
+	# From seanf@swdc.stratus.com.
+	echo i860-stratus-sysv4
+	exit ;;
+    i*86:VOS:*:*)
+	# From Paul.Green@stratus.com.
+	echo ${UNAME_MACHINE}-stratus-vos
+	exit ;;
+    *:VOS:*:*)
+	# From Paul.Green@stratus.com.
+	echo hppa1.1-stratus-vos
+	exit ;;
+    mc68*:A/UX:*:*)
+	echo m68k-apple-aux${UNAME_RELEASE}
+	exit ;;
+    news*:NEWS-OS:6*:*)
+	echo mips-sony-newsos6
+	exit ;;
+    R[34]000:*System_V*:*:* | R4000:UNIX_SYSV:*:* | R*000:UNIX_SV:*:*)
+	if [ -d /usr/nec ]; then
+		echo mips-nec-sysv${UNAME_RELEASE}
+	else
+		echo mips-unknown-sysv${UNAME_RELEASE}
+	fi
+	exit ;;
+    BeBox:BeOS:*:*)	# BeOS running on hardware made by Be, PPC only.
+	echo powerpc-be-beos
+	exit ;;
+    BeMac:BeOS:*:*)	# BeOS running on Mac or Mac clone, PPC only.
+	echo powerpc-apple-beos
+	exit ;;
+    BePC:BeOS:*:*)	# BeOS running on Intel PC compatible.
+	echo i586-pc-beos
+	exit ;;
+    BePC:Haiku:*:*)	# Haiku running on Intel PC compatible.
+	echo i586-pc-haiku
+	exit ;;
+    x86_64:Haiku:*:*)
+	echo x86_64-unknown-haiku
+	exit ;;
+    SX-4:SUPER-UX:*:*)
+	echo sx4-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-5:SUPER-UX:*:*)
+	echo sx5-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-6:SUPER-UX:*:*)
+	echo sx6-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-7:SUPER-UX:*:*)
+	echo sx7-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-8:SUPER-UX:*:*)
+	echo sx8-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-8R:SUPER-UX:*:*)
+	echo sx8r-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-ACE:SUPER-UX:*:*)
+	echo sxace-nec-superux${UNAME_RELEASE}
+	exit ;;
+    Power*:Rhapsody:*:*)
+	echo powerpc-apple-rhapsody${UNAME_RELEASE}
+	exit ;;
+    *:Rhapsody:*:*)
+	echo ${UNAME_MACHINE}-apple-rhapsody${UNAME_RELEASE}
+	exit ;;
+    *:Darwin:*:*)
+	UNAME_PROCESSOR=`uname -p` || UNAME_PROCESSOR=unknown
+	eval $set_cc_for_build
+	if test "$UNAME_PROCESSOR" = unknown ; then
+	    UNAME_PROCESSOR=powerpc
+	fi
+	if test `echo "$UNAME_RELEASE" | sed -e 's/\..*//'` -le 10 ; then
+	    if [ "$CC_FOR_BUILD" != no_compiler_found ]; then
+		if (echo '#ifdef __LP64__'; echo IS_64BIT_ARCH; echo '#endif') | \
+		    (CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) | \
+		    grep IS_64BIT_ARCH >/dev/null
+		then
+		    case $UNAME_PROCESSOR in
+			i386) UNAME_PROCESSOR=x86_64 ;;
+			powerpc) UNAME_PROCESSOR=powerpc64 ;;
+		    esac
+		fi
+	    fi
+	elif test "$UNAME_PROCESSOR" = i386 ; then
+	    # Avoid executing cc on OS X 10.9, as it ships with a stub
+	    # that puts up a graphical alert prompting to install
+	    # developer tools.  Any system running Mac OS X 10.7 or
+	    # later (Darwin 11 and later) is required to have a 64-bit
+	    # processor. This is not true of the ARM version of Darwin
+	    # that Apple uses in portable devices.
+	    UNAME_PROCESSOR=x86_64
+	fi
+	echo ${UNAME_PROCESSOR}-apple-darwin${UNAME_RELEASE}
+	exit ;;
+    *:procnto*:*:* | *:QNX:[0123456789]*:*)
+	UNAME_PROCESSOR=`uname -p`
+	if test "$UNAME_PROCESSOR" = x86; then
+		UNAME_PROCESSOR=i386
+		UNAME_MACHINE=pc
+	fi
+	echo ${UNAME_PROCESSOR}-${UNAME_MACHINE}-nto-qnx${UNAME_RELEASE}
+	exit ;;
+    *:QNX:*:4*)
+	echo i386-pc-qnx
+	exit ;;
+    NEO-?:NONSTOP_KERNEL:*:*)
+	echo neo-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    NSE-*:NONSTOP_KERNEL:*:*)
+	echo nse-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    NSR-?:NONSTOP_KERNEL:*:*)
+	echo nsr-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    *:NonStop-UX:*:*)
+	echo mips-compaq-nonstopux
+	exit ;;
+    BS2000:POSIX*:*:*)
+	echo bs2000-siemens-sysv
+	exit ;;
+    DS/*:UNIX_System_V:*:*)
+	echo ${UNAME_MACHINE}-${UNAME_SYSTEM}-${UNAME_RELEASE}
+	exit ;;
+    *:Plan9:*:*)
+	# "uname -m" is not consistent, so use $cputype instead. 386
+	# is converted to i386 for consistency with other x86
+	# operating systems.
+	if test "$cputype" = 386; then
+	    UNAME_MACHINE=i386
+	else
+	    UNAME_MACHINE="$cputype"
+	fi
+	echo ${UNAME_MACHINE}-unknown-plan9
+	exit ;;
+    *:TOPS-10:*:*)
+	echo pdp10-unknown-tops10
+	exit ;;
+    *:TENEX:*:*)
+	echo pdp10-unknown-tenex
+	exit ;;
+    KS10:TOPS-20:*:* | KL10:TOPS-20:*:* | TYPE4:TOPS-20:*:*)
+	echo pdp10-dec-tops20
+	exit ;;
+    XKL-1:TOPS-20:*:* | TYPE5:TOPS-20:*:*)
+	echo pdp10-xkl-tops20
+	exit ;;
+    *:TOPS-20:*:*)
+	echo pdp10-unknown-tops20
+	exit ;;
+    *:ITS:*:*)
+	echo pdp10-unknown-its
+	exit ;;
+    SEI:*:*:SEIUX)
+	echo mips-sei-seiux${UNAME_RELEASE}
+	exit ;;
+    *:DragonFly:*:*)
+	echo ${UNAME_MACHINE}-unknown-dragonfly`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`
+	exit ;;
+    *:*VMS:*:*)
+	UNAME_MACHINE=`(uname -p) 2>/dev/null`
+	case "${UNAME_MACHINE}" in
+	    A*) echo alpha-dec-vms ; exit ;;
+	    I*) echo ia64-dec-vms ; exit ;;
+	    V*) echo vax-dec-vms ; exit ;;
+	esac ;;
+    *:XENIX:*:SysV)
+	echo i386-pc-xenix
+	exit ;;
+    i*86:skyos:*:*)
+	echo ${UNAME_MACHINE}-pc-skyos`echo ${UNAME_RELEASE} | sed -e 's/ .*$//'`
+	exit ;;
+    i*86:rdos:*:*)
+	echo ${UNAME_MACHINE}-pc-rdos
+	exit ;;
+    i*86:AROS:*:*)
+	echo ${UNAME_MACHINE}-pc-aros
+	exit ;;
+    x86_64:VMkernel:*:*)
+	echo ${UNAME_MACHINE}-unknown-esx
+	exit ;;
+    amd64:Isilon\ OneFS:*:*)
+	echo x86_64-unknown-onefs
+	exit ;;
+esac
+
+cat >&2 <<EOF
+$0: unable to guess system type
+
+This script (version $timestamp), has failed to recognize the
+operating system you are using. If your script is old, overwrite
+config.guess and config.sub with the latest versions from:
+
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess
+and
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub
+
+If $0 has already been updated, send the following data and any
+information you think might be pertinent to config-patches@gnu.org to
+provide the necessary information to handle your system.
+
+config.guess timestamp = $timestamp
+
+uname -m = `(uname -m) 2>/dev/null || echo unknown`
+uname -r = `(uname -r) 2>/dev/null || echo unknown`
+uname -s = `(uname -s) 2>/dev/null || echo unknown`
+uname -v = `(uname -v) 2>/dev/null || echo unknown`
+
+/usr/bin/uname -p = `(/usr/bin/uname -p) 2>/dev/null`
+/bin/uname -X     = `(/bin/uname -X) 2>/dev/null`
+
+hostinfo               = `(hostinfo) 2>/dev/null`
+/bin/universe          = `(/bin/universe) 2>/dev/null`
+/usr/bin/arch -k       = `(/usr/bin/arch -k) 2>/dev/null`
+/bin/arch              = `(/bin/arch) 2>/dev/null`
+/usr/bin/oslevel       = `(/usr/bin/oslevel) 2>/dev/null`
+/usr/convex/getsysinfo = `(/usr/convex/getsysinfo) 2>/dev/null`
+
+UNAME_MACHINE = ${UNAME_MACHINE}
+UNAME_RELEASE = ${UNAME_RELEASE}
+UNAME_SYSTEM  = ${UNAME_SYSTEM}
+UNAME_VERSION = ${UNAME_VERSION}
+EOF
+
+exit 1
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "timestamp='"
+# time-stamp-format: "%:y-%02m-%02d"
+# time-stamp-end: "'"
+# End:
diff --git a/jemalloc/build-aux/config.sub b/jemalloc/build-aux/config.sub
new file mode 100755
index 0000000..dd2ca93
--- /dev/null
+++ b/jemalloc/build-aux/config.sub
@@ -0,0 +1,1825 @@
+#! /bin/sh
+# Configuration validation subroutine script.
+#   Copyright 1992-2016 Free Software Foundation, Inc.
+
+timestamp='2016-11-04'
+
+# This file is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+#
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that
+# program.  This Exception is an additional permission under section 7
+# of the GNU General Public License, version 3 ("GPLv3").
+
+
+# Please send patches to <config-patches@gnu.org>.
+#
+# Configuration subroutine to validate and canonicalize a configuration type.
+# Supply the specified configuration type as an argument.
+# If it is invalid, we print an error message on stderr and exit with code 1.
+# Otherwise, we print the canonical config type on stdout and succeed.
+
+# You can get the latest version of this script from:
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub
+
+# This file is supposed to be the same for all GNU packages
+# and recognize all the CPU types, system types and aliases
+# that are meaningful with *any* GNU software.
+# Each package is responsible for reporting which valid configurations
+# it does not support.  The user should be able to distinguish
+# a failure to support a valid configuration from a meaningless
+# configuration.
+
+# The goal of this file is to map all the various variations of a given
+# machine specification into a single specification in the form:
+#	CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM
+# or in some cases, the newer four-part form:
+#	CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM
+# It is wrong to echo any other type of specification.
+
+me=`echo "$0" | sed -e 's,.*/,,'`
+
+usage="\
+Usage: $0 [OPTION] CPU-MFR-OPSYS or ALIAS
+
+Canonicalize a configuration name.
+
+Operation modes:
+  -h, --help         print this help, then exit
+  -t, --time-stamp   print date of last modification, then exit
+  -v, --version      print version number, then exit
+
+Report bugs and patches to <config-patches@gnu.org>."
+
+version="\
+GNU config.sub ($timestamp)
+
+Copyright 1992-2016 Free Software Foundation, Inc.
+
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
+
+help="
+Try \`$me --help' for more information."
+
+# Parse command line
+while test $# -gt 0 ; do
+  case $1 in
+    --time-stamp | --time* | -t )
+       echo "$timestamp" ; exit ;;
+    --version | -v )
+       echo "$version" ; exit ;;
+    --help | --h* | -h )
+       echo "$usage"; exit ;;
+    -- )     # Stop option processing
+       shift; break ;;
+    - )	# Use stdin as input.
+       break ;;
+    -* )
+       echo "$me: invalid option $1$help"
+       exit 1 ;;
+
+    *local*)
+       # First pass through any local machine types.
+       echo $1
+       exit ;;
+
+    * )
+       break ;;
+  esac
+done
+
+case $# in
+ 0) echo "$me: missing argument$help" >&2
+    exit 1;;
+ 1) ;;
+ *) echo "$me: too many arguments$help" >&2
+    exit 1;;
+esac
+
+# Separate what the user gave into CPU-COMPANY and OS or KERNEL-OS (if any).
+# Here we must recognize all the valid KERNEL-OS combinations.
+maybe_os=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\2/'`
+case $maybe_os in
+  nto-qnx* | linux-gnu* | linux-android* | linux-dietlibc | linux-newlib* | \
+  linux-musl* | linux-uclibc* | uclinux-uclibc* | uclinux-gnu* | kfreebsd*-gnu* | \
+  knetbsd*-gnu* | netbsd*-gnu* | netbsd*-eabi* | \
+  kopensolaris*-gnu* | cloudabi*-eabi* | \
+  storm-chaos* | os2-emx* | rtmk-nova*)
+    os=-$maybe_os
+    basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`
+    ;;
+  android-linux)
+    os=-linux-android
+    basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`-unknown
+    ;;
+  *)
+    basic_machine=`echo $1 | sed 's/-[^-]*$//'`
+    if [ $basic_machine != $1 ]
+    then os=`echo $1 | sed 's/.*-/-/'`
+    else os=; fi
+    ;;
+esac
+
+### Let's recognize common machines as not being operating systems so
+### that things like config.sub decstation-3100 work.  We also
+### recognize some manufacturers as not being operating systems, so we
+### can provide default operating systems below.
+case $os in
+	-sun*os*)
+		# Prevent following clause from handling this invalid input.
+		;;
+	-dec* | -mips* | -sequent* | -encore* | -pc532* | -sgi* | -sony* | \
+	-att* | -7300* | -3300* | -delta* | -motorola* | -sun[234]* | \
+	-unicom* | -ibm* | -next | -hp | -isi* | -apollo | -altos* | \
+	-convergent* | -ncr* | -news | -32* | -3600* | -3100* | -hitachi* |\
+	-c[123]* | -convex* | -sun | -crds | -omron* | -dg | -ultra | -tti* | \
+	-harris | -dolphin | -highlevel | -gould | -cbm | -ns | -masscomp | \
+	-apple | -axis | -knuth | -cray | -microblaze*)
+		os=
+		basic_machine=$1
+		;;
+	-bluegene*)
+		os=-cnk
+		;;
+	-sim | -cisco | -oki | -wec | -winbond)
+		os=
+		basic_machine=$1
+		;;
+	-scout)
+		;;
+	-wrs)
+		os=-vxworks
+		basic_machine=$1
+		;;
+	-chorusos*)
+		os=-chorusos
+		basic_machine=$1
+		;;
+	-chorusrdb)
+		os=-chorusrdb
+		basic_machine=$1
+		;;
+	-hiux*)
+		os=-hiuxwe2
+		;;
+	-sco6)
+		os=-sco5v6
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco5)
+		os=-sco3.2v5
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco4)
+		os=-sco3.2v4
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco3.2.[4-9]*)
+		os=`echo $os | sed -e 's/sco3.2./sco3.2v/'`
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco3.2v[4-9]*)
+		# Don't forget version if it is 3.2v4 or newer.
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco5v6*)
+		# Don't forget version if it is 3.2v4 or newer.
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco*)
+		os=-sco3.2v2
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-udk*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-isc)
+		os=-isc2.2
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-clix*)
+		basic_machine=clipper-intergraph
+		;;
+	-isc*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-lynx*178)
+		os=-lynxos178
+		;;
+	-lynx*5)
+		os=-lynxos5
+		;;
+	-lynx*)
+		os=-lynxos
+		;;
+	-ptx*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-sequent/'`
+		;;
+	-windowsnt*)
+		os=`echo $os | sed -e 's/windowsnt/winnt/'`
+		;;
+	-psos*)
+		os=-psos
+		;;
+	-mint | -mint[0-9]*)
+		basic_machine=m68k-atari
+		os=-mint
+		;;
+esac
+
+# Decode aliases for certain CPU-COMPANY combinations.
+case $basic_machine in
+	# Recognize the basic CPU types without company name.
+	# Some are omitted here because they have special meanings below.
+	1750a | 580 \
+	| a29k \
+	| aarch64 | aarch64_be \
+	| alpha | alphaev[4-8] | alphaev56 | alphaev6[78] | alphapca5[67] \
+	| alpha64 | alpha64ev[4-8] | alpha64ev56 | alpha64ev6[78] | alpha64pca5[67] \
+	| am33_2.0 \
+	| arc | arceb \
+	| arm | arm[bl]e | arme[lb] | armv[2-8] | armv[3-8][lb] | armv7[arm] \
+	| avr | avr32 \
+	| ba \
+	| be32 | be64 \
+	| bfin \
+	| c4x | c8051 | clipper \
+	| d10v | d30v | dlx | dsp16xx \
+	| e2k | epiphany \
+	| fido | fr30 | frv | ft32 \
+	| h8300 | h8500 | hppa | hppa1.[01] | hppa2.0 | hppa2.0[nw] | hppa64 \
+	| hexagon \
+	| i370 | i860 | i960 | ia64 \
+	| ip2k | iq2000 \
+	| k1om \
+	| le32 | le64 \
+	| lm32 \
+	| m32c | m32r | m32rle | m68000 | m68k | m88k \
+	| maxq | mb | microblaze | microblazeel | mcore | mep | metag \
+	| mips | mipsbe | mipseb | mipsel | mipsle \
+	| mips16 \
+	| mips64 | mips64el \
+	| mips64octeon | mips64octeonel \
+	| mips64orion | mips64orionel \
+	| mips64r5900 | mips64r5900el \
+	| mips64vr | mips64vrel \
+	| mips64vr4100 | mips64vr4100el \
+	| mips64vr4300 | mips64vr4300el \
+	| mips64vr5000 | mips64vr5000el \
+	| mips64vr5900 | mips64vr5900el \
+	| mipsisa32 | mipsisa32el \
+	| mipsisa32r2 | mipsisa32r2el \
+	| mipsisa32r6 | mipsisa32r6el \
+	| mipsisa64 | mipsisa64el \
+	| mipsisa64r2 | mipsisa64r2el \
+	| mipsisa64r6 | mipsisa64r6el \
+	| mipsisa64sb1 | mipsisa64sb1el \
+	| mipsisa64sr71k | mipsisa64sr71kel \
+	| mipsr5900 | mipsr5900el \
+	| mipstx39 | mipstx39el \
+	| mn10200 | mn10300 \
+	| moxie \
+	| mt \
+	| msp430 \
+	| nds32 | nds32le | nds32be \
+	| nios | nios2 | nios2eb | nios2el \
+	| ns16k | ns32k \
+	| open8 | or1k | or1knd | or32 \
+	| pdp10 | pdp11 | pj | pjl \
+	| powerpc | powerpc64 | powerpc64le | powerpcle \
+	| pru \
+	| pyramid \
+	| riscv32 | riscv64 \
+	| rl78 | rx \
+	| score \
+	| sh | sh[1234] | sh[24]a | sh[24]aeb | sh[23]e | sh[234]eb | sheb | shbe | shle | sh[1234]le | sh3ele \
+	| sh64 | sh64le \
+	| sparc | sparc64 | sparc64b | sparc64v | sparc86x | sparclet | sparclite \
+	| sparcv8 | sparcv9 | sparcv9b | sparcv9v \
+	| spu \
+	| tahoe | tic4x | tic54x | tic55x | tic6x | tic80 | tron \
+	| ubicom32 \
+	| v850 | v850e | v850e1 | v850e2 | v850es | v850e2v3 \
+	| visium \
+	| we32k \
+	| x86 | xc16x | xstormy16 | xtensa \
+	| z8k | z80)
+		basic_machine=$basic_machine-unknown
+		;;
+	c54x)
+		basic_machine=tic54x-unknown
+		;;
+	c55x)
+		basic_machine=tic55x-unknown
+		;;
+	c6x)
+		basic_machine=tic6x-unknown
+		;;
+	leon|leon[3-9])
+		basic_machine=sparc-$basic_machine
+		;;
+	m6811 | m68hc11 | m6812 | m68hc12 | m68hcs12x | nvptx | picochip)
+		basic_machine=$basic_machine-unknown
+		os=-none
+		;;
+	m88110 | m680[12346]0 | m683?2 | m68360 | m5200 | v70 | w65 | z8k)
+		;;
+	ms1)
+		basic_machine=mt-unknown
+		;;
+
+	strongarm | thumb | xscale)
+		basic_machine=arm-unknown
+		;;
+	xgate)
+		basic_machine=$basic_machine-unknown
+		os=-none
+		;;
+	xscaleeb)
+		basic_machine=armeb-unknown
+		;;
+
+	xscaleel)
+		basic_machine=armel-unknown
+		;;
+
+	# We use `pc' rather than `unknown'
+	# because (1) that's what they normally are, and
+	# (2) the word "unknown" tends to confuse beginning users.
+	i*86 | x86_64)
+	  basic_machine=$basic_machine-pc
+	  ;;
+	# Object if more than one company name word.
+	*-*-*)
+		echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2
+		exit 1
+		;;
+	# Recognize the basic CPU types with company name.
+	580-* \
+	| a29k-* \
+	| aarch64-* | aarch64_be-* \
+	| alpha-* | alphaev[4-8]-* | alphaev56-* | alphaev6[78]-* \
+	| alpha64-* | alpha64ev[4-8]-* | alpha64ev56-* | alpha64ev6[78]-* \
+	| alphapca5[67]-* | alpha64pca5[67]-* | arc-* | arceb-* \
+	| arm-*  | armbe-* | armle-* | armeb-* | armv*-* \
+	| avr-* | avr32-* \
+	| ba-* \
+	| be32-* | be64-* \
+	| bfin-* | bs2000-* \
+	| c[123]* | c30-* | [cjt]90-* | c4x-* \
+	| c8051-* | clipper-* | craynv-* | cydra-* \
+	| d10v-* | d30v-* | dlx-* \
+	| e2k-* | elxsi-* \
+	| f30[01]-* | f700-* | fido-* | fr30-* | frv-* | fx80-* \
+	| h8300-* | h8500-* \
+	| hppa-* | hppa1.[01]-* | hppa2.0-* | hppa2.0[nw]-* | hppa64-* \
+	| hexagon-* \
+	| i*86-* | i860-* | i960-* | ia64-* \
+	| ip2k-* | iq2000-* \
+	| k1om-* \
+	| le32-* | le64-* \
+	| lm32-* \
+	| m32c-* | m32r-* | m32rle-* \
+	| m68000-* | m680[012346]0-* | m68360-* | m683?2-* | m68k-* \
+	| m88110-* | m88k-* | maxq-* | mcore-* | metag-* \
+	| microblaze-* | microblazeel-* \
+	| mips-* | mipsbe-* | mipseb-* | mipsel-* | mipsle-* \
+	| mips16-* \
+	| mips64-* | mips64el-* \
+	| mips64octeon-* | mips64octeonel-* \
+	| mips64orion-* | mips64orionel-* \
+	| mips64r5900-* | mips64r5900el-* \
+	| mips64vr-* | mips64vrel-* \
+	| mips64vr4100-* | mips64vr4100el-* \
+	| mips64vr4300-* | mips64vr4300el-* \
+	| mips64vr5000-* | mips64vr5000el-* \
+	| mips64vr5900-* | mips64vr5900el-* \
+	| mipsisa32-* | mipsisa32el-* \
+	| mipsisa32r2-* | mipsisa32r2el-* \
+	| mipsisa32r6-* | mipsisa32r6el-* \
+	| mipsisa64-* | mipsisa64el-* \
+	| mipsisa64r2-* | mipsisa64r2el-* \
+	| mipsisa64r6-* | mipsisa64r6el-* \
+	| mipsisa64sb1-* | mipsisa64sb1el-* \
+	| mipsisa64sr71k-* | mipsisa64sr71kel-* \
+	| mipsr5900-* | mipsr5900el-* \
+	| mipstx39-* | mipstx39el-* \
+	| mmix-* \
+	| mt-* \
+	| msp430-* \
+	| nds32-* | nds32le-* | nds32be-* \
+	| nios-* | nios2-* | nios2eb-* | nios2el-* \
+	| none-* | np1-* | ns16k-* | ns32k-* \
+	| open8-* \
+	| or1k*-* \
+	| orion-* \
+	| pdp10-* | pdp11-* | pj-* | pjl-* | pn-* | power-* \
+	| powerpc-* | powerpc64-* | powerpc64le-* | powerpcle-* \
+	| pru-* \
+	| pyramid-* \
+	| riscv32-* | riscv64-* \
+	| rl78-* | romp-* | rs6000-* | rx-* \
+	| sh-* | sh[1234]-* | sh[24]a-* | sh[24]aeb-* | sh[23]e-* | sh[34]eb-* | sheb-* | shbe-* \
+	| shle-* | sh[1234]le-* | sh3ele-* | sh64-* | sh64le-* \
+	| sparc-* | sparc64-* | sparc64b-* | sparc64v-* | sparc86x-* | sparclet-* \
+	| sparclite-* \
+	| sparcv8-* | sparcv9-* | sparcv9b-* | sparcv9v-* | sv1-* | sx*-* \
+	| tahoe-* \
+	| tic30-* | tic4x-* | tic54x-* | tic55x-* | tic6x-* | tic80-* \
+	| tile*-* \
+	| tron-* \
+	| ubicom32-* \
+	| v850-* | v850e-* | v850e1-* | v850es-* | v850e2-* | v850e2v3-* \
+	| vax-* \
+	| visium-* \
+	| we32k-* \
+	| x86-* | x86_64-* | xc16x-* | xps100-* \
+	| xstormy16-* | xtensa*-* \
+	| ymp-* \
+	| z8k-* | z80-*)
+		;;
+	# Recognize the basic CPU types without company name, with glob match.
+	xtensa*)
+		basic_machine=$basic_machine-unknown
+		;;
+	# Recognize the various machine names and aliases which stand
+	# for a CPU type and a company and sometimes even an OS.
+	386bsd)
+		basic_machine=i386-unknown
+		os=-bsd
+		;;
+	3b1 | 7300 | 7300-att | att-7300 | pc7300 | safari | unixpc)
+		basic_machine=m68000-att
+		;;
+	3b*)
+		basic_machine=we32k-att
+		;;
+	a29khif)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	abacus)
+		basic_machine=abacus-unknown
+		;;
+	adobe68k)
+		basic_machine=m68010-adobe
+		os=-scout
+		;;
+	alliant | fx80)
+		basic_machine=fx80-alliant
+		;;
+	altos | altos3068)
+		basic_machine=m68k-altos
+		;;
+	am29k)
+		basic_machine=a29k-none
+		os=-bsd
+		;;
+	amd64)
+		basic_machine=x86_64-pc
+		;;
+	amd64-*)
+		basic_machine=x86_64-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	amdahl)
+		basic_machine=580-amdahl
+		os=-sysv
+		;;
+	amiga | amiga-*)
+		basic_machine=m68k-unknown
+		;;
+	amigaos | amigados)
+		basic_machine=m68k-unknown
+		os=-amigaos
+		;;
+	amigaunix | amix)
+		basic_machine=m68k-unknown
+		os=-sysv4
+		;;
+	apollo68)
+		basic_machine=m68k-apollo
+		os=-sysv
+		;;
+	apollo68bsd)
+		basic_machine=m68k-apollo
+		os=-bsd
+		;;
+	aros)
+		basic_machine=i386-pc
+		os=-aros
+		;;
+	asmjs)
+		basic_machine=asmjs-unknown
+		;;
+	aux)
+		basic_machine=m68k-apple
+		os=-aux
+		;;
+	balance)
+		basic_machine=ns32k-sequent
+		os=-dynix
+		;;
+	blackfin)
+		basic_machine=bfin-unknown
+		os=-linux
+		;;
+	blackfin-*)
+		basic_machine=bfin-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	bluegene*)
+		basic_machine=powerpc-ibm
+		os=-cnk
+		;;
+	c54x-*)
+		basic_machine=tic54x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c55x-*)
+		basic_machine=tic55x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c6x-*)
+		basic_machine=tic6x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c90)
+		basic_machine=c90-cray
+		os=-unicos
+		;;
+	cegcc)
+		basic_machine=arm-unknown
+		os=-cegcc
+		;;
+	convex-c1)
+		basic_machine=c1-convex
+		os=-bsd
+		;;
+	convex-c2)
+		basic_machine=c2-convex
+		os=-bsd
+		;;
+	convex-c32)
+		basic_machine=c32-convex
+		os=-bsd
+		;;
+	convex-c34)
+		basic_machine=c34-convex
+		os=-bsd
+		;;
+	convex-c38)
+		basic_machine=c38-convex
+		os=-bsd
+		;;
+	cray | j90)
+		basic_machine=j90-cray
+		os=-unicos
+		;;
+	craynv)
+		basic_machine=craynv-cray
+		os=-unicosmp
+		;;
+	cr16 | cr16-*)
+		basic_machine=cr16-unknown
+		os=-elf
+		;;
+	crds | unos)
+		basic_machine=m68k-crds
+		;;
+	crisv32 | crisv32-* | etraxfs*)
+		basic_machine=crisv32-axis
+		;;
+	cris | cris-* | etrax*)
+		basic_machine=cris-axis
+		;;
+	crx)
+		basic_machine=crx-unknown
+		os=-elf
+		;;
+	da30 | da30-*)
+		basic_machine=m68k-da30
+		;;
+	decstation | decstation-3100 | pmax | pmax-* | pmin | dec3100 | decstatn)
+		basic_machine=mips-dec
+		;;
+	decsystem10* | dec10*)
+		basic_machine=pdp10-dec
+		os=-tops10
+		;;
+	decsystem20* | dec20*)
+		basic_machine=pdp10-dec
+		os=-tops20
+		;;
+	delta | 3300 | motorola-3300 | motorola-delta \
+	      | 3300-motorola | delta-motorola)
+		basic_machine=m68k-motorola
+		;;
+	delta88)
+		basic_machine=m88k-motorola
+		os=-sysv3
+		;;
+	dicos)
+		basic_machine=i686-pc
+		os=-dicos
+		;;
+	djgpp)
+		basic_machine=i586-pc
+		os=-msdosdjgpp
+		;;
+	dpx20 | dpx20-*)
+		basic_machine=rs6000-bull
+		os=-bosx
+		;;
+	dpx2* | dpx2*-bull)
+		basic_machine=m68k-bull
+		os=-sysv3
+		;;
+	e500v[12])
+		basic_machine=powerpc-unknown
+		os=$os"spe"
+		;;
+	e500v[12]-*)
+		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=$os"spe"
+		;;
+	ebmon29k)
+		basic_machine=a29k-amd
+		os=-ebmon
+		;;
+	elxsi)
+		basic_machine=elxsi-elxsi
+		os=-bsd
+		;;
+	encore | umax | mmax)
+		basic_machine=ns32k-encore
+		;;
+	es1800 | OSE68k | ose68k | ose | OSE)
+		basic_machine=m68k-ericsson
+		os=-ose
+		;;
+	fx2800)
+		basic_machine=i860-alliant
+		;;
+	genix)
+		basic_machine=ns32k-ns
+		;;
+	gmicro)
+		basic_machine=tron-gmicro
+		os=-sysv
+		;;
+	go32)
+		basic_machine=i386-pc
+		os=-go32
+		;;
+	h3050r* | hiux*)
+		basic_machine=hppa1.1-hitachi
+		os=-hiuxwe2
+		;;
+	h8300hms)
+		basic_machine=h8300-hitachi
+		os=-hms
+		;;
+	h8300xray)
+		basic_machine=h8300-hitachi
+		os=-xray
+		;;
+	h8500hms)
+		basic_machine=h8500-hitachi
+		os=-hms
+		;;
+	harris)
+		basic_machine=m88k-harris
+		os=-sysv3
+		;;
+	hp300-*)
+		basic_machine=m68k-hp
+		;;
+	hp300bsd)
+		basic_machine=m68k-hp
+		os=-bsd
+		;;
+	hp300hpux)
+		basic_machine=m68k-hp
+		os=-hpux
+		;;
+	hp3k9[0-9][0-9] | hp9[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hp9k2[0-9][0-9] | hp9k31[0-9])
+		basic_machine=m68000-hp
+		;;
+	hp9k3[2-9][0-9])
+		basic_machine=m68k-hp
+		;;
+	hp9k6[0-9][0-9] | hp6[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hp9k7[0-79][0-9] | hp7[0-79][0-9])
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k78[0-9] | hp78[0-9])
+		# FIXME: really hppa2.0-hp
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[67]1 | hp8[67]1 | hp9k80[24] | hp80[24] | hp9k8[78]9 | hp8[78]9 | hp9k893 | hp893)
+		# FIXME: really hppa2.0-hp
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[0-9][13679] | hp8[0-9][13679])
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[0-9][0-9] | hp8[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hppa-next)
+		os=-nextstep3
+		;;
+	hppaosf)
+		basic_machine=hppa1.1-hp
+		os=-osf
+		;;
+	hppro)
+		basic_machine=hppa1.1-hp
+		os=-proelf
+		;;
+	i370-ibm* | ibm*)
+		basic_machine=i370-ibm
+		;;
+	i*86v32)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv32
+		;;
+	i*86v4*)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv4
+		;;
+	i*86v)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv
+		;;
+	i*86sol2)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-solaris2
+		;;
+	i386mach)
+		basic_machine=i386-mach
+		os=-mach
+		;;
+	i386-vsta | vsta)
+		basic_machine=i386-unknown
+		os=-vsta
+		;;
+	iris | iris4d)
+		basic_machine=mips-sgi
+		case $os in
+		    -irix*)
+			;;
+		    *)
+			os=-irix4
+			;;
+		esac
+		;;
+	isi68 | isi)
+		basic_machine=m68k-isi
+		os=-sysv
+		;;
+	leon-*|leon[3-9]-*)
+		basic_machine=sparc-`echo $basic_machine | sed 's/-.*//'`
+		;;
+	m68knommu)
+		basic_machine=m68k-unknown
+		os=-linux
+		;;
+	m68knommu-*)
+		basic_machine=m68k-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	m88k-omron*)
+		basic_machine=m88k-omron
+		;;
+	magnum | m3230)
+		basic_machine=mips-mips
+		os=-sysv
+		;;
+	merlin)
+		basic_machine=ns32k-utek
+		os=-sysv
+		;;
+	microblaze*)
+		basic_machine=microblaze-xilinx
+		;;
+	mingw64)
+		basic_machine=x86_64-pc
+		os=-mingw64
+		;;
+	mingw32)
+		basic_machine=i686-pc
+		os=-mingw32
+		;;
+	mingw32ce)
+		basic_machine=arm-unknown
+		os=-mingw32ce
+		;;
+	miniframe)
+		basic_machine=m68000-convergent
+		;;
+	*mint | -mint[0-9]* | *MiNT | *MiNT[0-9]*)
+		basic_machine=m68k-atari
+		os=-mint
+		;;
+	mips3*-*)
+		basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'`
+		;;
+	mips3*)
+		basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'`-unknown
+		;;
+	monitor)
+		basic_machine=m68k-rom68k
+		os=-coff
+		;;
+	morphos)
+		basic_machine=powerpc-unknown
+		os=-morphos
+		;;
+	moxiebox)
+		basic_machine=moxie-unknown
+		os=-moxiebox
+		;;
+	msdos)
+		basic_machine=i386-pc
+		os=-msdos
+		;;
+	ms1-*)
+		basic_machine=`echo $basic_machine | sed -e 's/ms1-/mt-/'`
+		;;
+	msys)
+		basic_machine=i686-pc
+		os=-msys
+		;;
+	mvs)
+		basic_machine=i370-ibm
+		os=-mvs
+		;;
+	nacl)
+		basic_machine=le32-unknown
+		os=-nacl
+		;;
+	ncr3000)
+		basic_machine=i486-ncr
+		os=-sysv4
+		;;
+	netbsd386)
+		basic_machine=i386-unknown
+		os=-netbsd
+		;;
+	netwinder)
+		basic_machine=armv4l-rebel
+		os=-linux
+		;;
+	news | news700 | news800 | news900)
+		basic_machine=m68k-sony
+		os=-newsos
+		;;
+	news1000)
+		basic_machine=m68030-sony
+		os=-newsos
+		;;
+	news-3600 | risc-news)
+		basic_machine=mips-sony
+		os=-newsos
+		;;
+	necv70)
+		basic_machine=v70-nec
+		os=-sysv
+		;;
+	next | m*-next )
+		basic_machine=m68k-next
+		case $os in
+		    -nextstep* )
+			;;
+		    -ns2*)
+		      os=-nextstep2
+			;;
+		    *)
+		      os=-nextstep3
+			;;
+		esac
+		;;
+	nh3000)
+		basic_machine=m68k-harris
+		os=-cxux
+		;;
+	nh[45]000)
+		basic_machine=m88k-harris
+		os=-cxux
+		;;
+	nindy960)
+		basic_machine=i960-intel
+		os=-nindy
+		;;
+	mon960)
+		basic_machine=i960-intel
+		os=-mon960
+		;;
+	nonstopux)
+		basic_machine=mips-compaq
+		os=-nonstopux
+		;;
+	np1)
+		basic_machine=np1-gould
+		;;
+	neo-tandem)
+		basic_machine=neo-tandem
+		;;
+	nse-tandem)
+		basic_machine=nse-tandem
+		;;
+	nsr-tandem)
+		basic_machine=nsr-tandem
+		;;
+	op50n-* | op60c-*)
+		basic_machine=hppa1.1-oki
+		os=-proelf
+		;;
+	openrisc | openrisc-*)
+		basic_machine=or32-unknown
+		;;
+	os400)
+		basic_machine=powerpc-ibm
+		os=-os400
+		;;
+	OSE68000 | ose68000)
+		basic_machine=m68000-ericsson
+		os=-ose
+		;;
+	os68k)
+		basic_machine=m68k-none
+		os=-os68k
+		;;
+	pa-hitachi)
+		basic_machine=hppa1.1-hitachi
+		os=-hiuxwe2
+		;;
+	paragon)
+		basic_machine=i860-intel
+		os=-osf
+		;;
+	parisc)
+		basic_machine=hppa-unknown
+		os=-linux
+		;;
+	parisc-*)
+		basic_machine=hppa-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	pbd)
+		basic_machine=sparc-tti
+		;;
+	pbb)
+		basic_machine=m68k-tti
+		;;
+	pc532 | pc532-*)
+		basic_machine=ns32k-pc532
+		;;
+	pc98)
+		basic_machine=i386-pc
+		;;
+	pc98-*)
+		basic_machine=i386-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentium | p5 | k5 | k6 | nexgen | viac3)
+		basic_machine=i586-pc
+		;;
+	pentiumpro | p6 | 6x86 | athlon | athlon_*)
+		basic_machine=i686-pc
+		;;
+	pentiumii | pentium2 | pentiumiii | pentium3)
+		basic_machine=i686-pc
+		;;
+	pentium4)
+		basic_machine=i786-pc
+		;;
+	pentium-* | p5-* | k5-* | k6-* | nexgen-* | viac3-*)
+		basic_machine=i586-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentiumpro-* | p6-* | 6x86-* | athlon-*)
+		basic_machine=i686-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentiumii-* | pentium2-* | pentiumiii-* | pentium3-*)
+		basic_machine=i686-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentium4-*)
+		basic_machine=i786-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pn)
+		basic_machine=pn-gould
+		;;
+	power)	basic_machine=power-ibm
+		;;
+	ppc | ppcbe)	basic_machine=powerpc-unknown
+		;;
+	ppc-* | ppcbe-*)
+		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppcle | powerpclittle)
+		basic_machine=powerpcle-unknown
+		;;
+	ppcle-* | powerpclittle-*)
+		basic_machine=powerpcle-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppc64)	basic_machine=powerpc64-unknown
+		;;
+	ppc64-*) basic_machine=powerpc64-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppc64le | powerpc64little)
+		basic_machine=powerpc64le-unknown
+		;;
+	ppc64le-* | powerpc64little-*)
+		basic_machine=powerpc64le-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ps2)
+		basic_machine=i386-ibm
+		;;
+	pw32)
+		basic_machine=i586-unknown
+		os=-pw32
+		;;
+	rdos | rdos64)
+		basic_machine=x86_64-pc
+		os=-rdos
+		;;
+	rdos32)
+		basic_machine=i386-pc
+		os=-rdos
+		;;
+	rom68k)
+		basic_machine=m68k-rom68k
+		os=-coff
+		;;
+	rm[46]00)
+		basic_machine=mips-siemens
+		;;
+	rtpc | rtpc-*)
+		basic_machine=romp-ibm
+		;;
+	s390 | s390-*)
+		basic_machine=s390-ibm
+		;;
+	s390x | s390x-*)
+		basic_machine=s390x-ibm
+		;;
+	sa29200)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	sb1)
+		basic_machine=mipsisa64sb1-unknown
+		;;
+	sb1el)
+		basic_machine=mipsisa64sb1el-unknown
+		;;
+	sde)
+		basic_machine=mipsisa32-sde
+		os=-elf
+		;;
+	sei)
+		basic_machine=mips-sei
+		os=-seiux
+		;;
+	sequent)
+		basic_machine=i386-sequent
+		;;
+	sh)
+		basic_machine=sh-hitachi
+		os=-hms
+		;;
+	sh5el)
+		basic_machine=sh5le-unknown
+		;;
+	sh64)
+		basic_machine=sh64-unknown
+		;;
+	sparclite-wrs | simso-wrs)
+		basic_machine=sparclite-wrs
+		os=-vxworks
+		;;
+	sps7)
+		basic_machine=m68k-bull
+		os=-sysv2
+		;;
+	spur)
+		basic_machine=spur-unknown
+		;;
+	st2000)
+		basic_machine=m68k-tandem
+		;;
+	stratus)
+		basic_machine=i860-stratus
+		os=-sysv4
+		;;
+	strongarm-* | thumb-*)
+		basic_machine=arm-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	sun2)
+		basic_machine=m68000-sun
+		;;
+	sun2os3)
+		basic_machine=m68000-sun
+		os=-sunos3
+		;;
+	sun2os4)
+		basic_machine=m68000-sun
+		os=-sunos4
+		;;
+	sun3os3)
+		basic_machine=m68k-sun
+		os=-sunos3
+		;;
+	sun3os4)
+		basic_machine=m68k-sun
+		os=-sunos4
+		;;
+	sun4os3)
+		basic_machine=sparc-sun
+		os=-sunos3
+		;;
+	sun4os4)
+		basic_machine=sparc-sun
+		os=-sunos4
+		;;
+	sun4sol2)
+		basic_machine=sparc-sun
+		os=-solaris2
+		;;
+	sun3 | sun3-*)
+		basic_machine=m68k-sun
+		;;
+	sun4)
+		basic_machine=sparc-sun
+		;;
+	sun386 | sun386i | roadrunner)
+		basic_machine=i386-sun
+		;;
+	sv1)
+		basic_machine=sv1-cray
+		os=-unicos
+		;;
+	symmetry)
+		basic_machine=i386-sequent
+		os=-dynix
+		;;
+	t3e)
+		basic_machine=alphaev5-cray
+		os=-unicos
+		;;
+	t90)
+		basic_machine=t90-cray
+		os=-unicos
+		;;
+	tile*)
+		basic_machine=$basic_machine-unknown
+		os=-linux-gnu
+		;;
+	tx39)
+		basic_machine=mipstx39-unknown
+		;;
+	tx39el)
+		basic_machine=mipstx39el-unknown
+		;;
+	toad1)
+		basic_machine=pdp10-xkl
+		os=-tops20
+		;;
+	tower | tower-32)
+		basic_machine=m68k-ncr
+		;;
+	tpf)
+		basic_machine=s390x-ibm
+		os=-tpf
+		;;
+	udi29k)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	ultra3)
+		basic_machine=a29k-nyu
+		os=-sym1
+		;;
+	v810 | necv810)
+		basic_machine=v810-nec
+		os=-none
+		;;
+	vaxv)
+		basic_machine=vax-dec
+		os=-sysv
+		;;
+	vms)
+		basic_machine=vax-dec
+		os=-vms
+		;;
+	vpp*|vx|vx-*)
+		basic_machine=f301-fujitsu
+		;;
+	vxworks960)
+		basic_machine=i960-wrs
+		os=-vxworks
+		;;
+	vxworks68)
+		basic_machine=m68k-wrs
+		os=-vxworks
+		;;
+	vxworks29k)
+		basic_machine=a29k-wrs
+		os=-vxworks
+		;;
+	w65*)
+		basic_machine=w65-wdc
+		os=-none
+		;;
+	w89k-*)
+		basic_machine=hppa1.1-winbond
+		os=-proelf
+		;;
+	xbox)
+		basic_machine=i686-pc
+		os=-mingw32
+		;;
+	xps | xps100)
+		basic_machine=xps100-honeywell
+		;;
+	xscale-* | xscalee[bl]-*)
+		basic_machine=`echo $basic_machine | sed 's/^xscale/arm/'`
+		;;
+	ymp)
+		basic_machine=ymp-cray
+		os=-unicos
+		;;
+	z8k-*-coff)
+		basic_machine=z8k-unknown
+		os=-sim
+		;;
+	z80-*-coff)
+		basic_machine=z80-unknown
+		os=-sim
+		;;
+	none)
+		basic_machine=none-none
+		os=-none
+		;;
+
+# Here we handle the default manufacturer of certain CPU types.  It is in
+# some cases the only manufacturer, in others, it is the most popular.
+	w89k)
+		basic_machine=hppa1.1-winbond
+		;;
+	op50n)
+		basic_machine=hppa1.1-oki
+		;;
+	op60c)
+		basic_machine=hppa1.1-oki
+		;;
+	romp)
+		basic_machine=romp-ibm
+		;;
+	mmix)
+		basic_machine=mmix-knuth
+		;;
+	rs6000)
+		basic_machine=rs6000-ibm
+		;;
+	vax)
+		basic_machine=vax-dec
+		;;
+	pdp10)
+		# there are many clones, so DEC is not a safe bet
+		basic_machine=pdp10-unknown
+		;;
+	pdp11)
+		basic_machine=pdp11-dec
+		;;
+	we32k)
+		basic_machine=we32k-att
+		;;
+	sh[1234] | sh[24]a | sh[24]aeb | sh[34]eb | sh[1234]le | sh[23]ele)
+		basic_machine=sh-unknown
+		;;
+	sparc | sparcv8 | sparcv9 | sparcv9b | sparcv9v)
+		basic_machine=sparc-sun
+		;;
+	cydra)
+		basic_machine=cydra-cydrome
+		;;
+	orion)
+		basic_machine=orion-highlevel
+		;;
+	orion105)
+		basic_machine=clipper-highlevel
+		;;
+	mac | mpw | mac-mpw)
+		basic_machine=m68k-apple
+		;;
+	pmac | pmac-mpw)
+		basic_machine=powerpc-apple
+		;;
+	*-unknown)
+		# Make sure to match an already-canonicalized machine name.
+		;;
+	*)
+		echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2
+		exit 1
+		;;
+esac
+
+# Here we canonicalize certain aliases for manufacturers.
+case $basic_machine in
+	*-digital*)
+		basic_machine=`echo $basic_machine | sed 's/digital.*/dec/'`
+		;;
+	*-commodore*)
+		basic_machine=`echo $basic_machine | sed 's/commodore.*/cbm/'`
+		;;
+	*)
+		;;
+esac
+
+# Decode manufacturer-specific aliases for certain operating systems.
+
+if [ x"$os" != x"" ]
+then
+case $os in
+	# First match some system type aliases
+	# that might get confused with valid system types.
+	# -solaris* is a basic system type, with this one exception.
+	-auroraux)
+		os=-auroraux
+		;;
+	-solaris1 | -solaris1.*)
+		os=`echo $os | sed -e 's|solaris1|sunos4|'`
+		;;
+	-solaris)
+		os=-solaris2
+		;;
+	-svr4*)
+		os=-sysv4
+		;;
+	-unixware*)
+		os=-sysv4.2uw
+		;;
+	-gnu/linux*)
+		os=`echo $os | sed -e 's|gnu/linux|linux-gnu|'`
+		;;
+	# First accept the basic system types.
+	# The portable systems comes first.
+	# Each alternative MUST END IN A *, to match a version number.
+	# -sysv* is not here because it comes later, after sysvr4.
+	-gnu* | -bsd* | -mach* | -minix* | -genix* | -ultrix* | -irix* \
+	      | -*vms* | -sco* | -esix* | -isc* | -aix* | -cnk* | -sunos | -sunos[34]*\
+	      | -hpux* | -unos* | -osf* | -luna* | -dgux* | -auroraux* | -solaris* \
+	      | -sym* | -kopensolaris* | -plan9* \
+	      | -amigaos* | -amigados* | -msdos* | -newsos* | -unicos* | -aof* \
+	      | -aos* | -aros* | -cloudabi* | -sortix* \
+	      | -nindy* | -vxsim* | -vxworks* | -ebmon* | -hms* | -mvs* \
+	      | -clix* | -riscos* | -uniplus* | -iris* | -rtu* | -xenix* \
+	      | -hiux* | -386bsd* | -knetbsd* | -mirbsd* | -netbsd* \
+	      | -bitrig* | -openbsd* | -solidbsd* | -libertybsd* \
+	      | -ekkobsd* | -kfreebsd* | -freebsd* | -riscix* | -lynxos* \
+	      | -bosx* | -nextstep* | -cxux* | -aout* | -elf* | -oabi* \
+	      | -ptx* | -coff* | -ecoff* | -winnt* | -domain* | -vsta* \
+	      | -udi* | -eabi* | -lites* | -ieee* | -go32* | -aux* \
+	      | -chorusos* | -chorusrdb* | -cegcc* \
+	      | -cygwin* | -msys* | -pe* | -psos* | -moss* | -proelf* | -rtems* \
+	      | -midipix* | -mingw32* | -mingw64* | -linux-gnu* | -linux-android* \
+	      | -linux-newlib* | -linux-musl* | -linux-uclibc* \
+	      | -uxpv* | -beos* | -mpeix* | -udk* | -moxiebox* \
+	      | -interix* | -uwin* | -mks* | -rhapsody* | -darwin* | -opened* \
+	      | -openstep* | -oskit* | -conix* | -pw32* | -nonstopux* \
+	      | -storm-chaos* | -tops10* | -tenex* | -tops20* | -its* \
+	      | -os2* | -vos* | -palmos* | -uclinux* | -nucleus* \
+	      | -morphos* | -superux* | -rtmk* | -rtmk-nova* | -windiss* \
+	      | -powermax* | -dnix* | -nx6 | -nx7 | -sei* | -dragonfly* \
+	      | -skyos* | -haiku* | -rdos* | -toppers* | -drops* | -es* \
+	      | -onefs* | -tirtos* | -phoenix* | -fuchsia*)
+	# Remember, each alternative MUST END IN *, to match a version number.
+		;;
+	-qnx*)
+		case $basic_machine in
+		    x86-* | i*86-*)
+			;;
+		    *)
+			os=-nto$os
+			;;
+		esac
+		;;
+	-nto-qnx*)
+		;;
+	-nto*)
+		os=`echo $os | sed -e 's|nto|nto-qnx|'`
+		;;
+	-sim | -es1800* | -hms* | -xray | -os68k* | -none* | -v88r* \
+	      | -windows* | -osx | -abug | -netware* | -os9* | -beos* | -haiku* \
+	      | -macos* | -mpw* | -magic* | -mmixware* | -mon960* | -lnews*)
+		;;
+	-mac*)
+		os=`echo $os | sed -e 's|mac|macos|'`
+		;;
+	-linux-dietlibc)
+		os=-linux-dietlibc
+		;;
+	-linux*)
+		os=`echo $os | sed -e 's|linux|linux-gnu|'`
+		;;
+	-sunos5*)
+		os=`echo $os | sed -e 's|sunos5|solaris2|'`
+		;;
+	-sunos6*)
+		os=`echo $os | sed -e 's|sunos6|solaris3|'`
+		;;
+	-opened*)
+		os=-openedition
+		;;
+	-os400*)
+		os=-os400
+		;;
+	-wince*)
+		os=-wince
+		;;
+	-osfrose*)
+		os=-osfrose
+		;;
+	-osf*)
+		os=-osf
+		;;
+	-utek*)
+		os=-bsd
+		;;
+	-dynix*)
+		os=-bsd
+		;;
+	-acis*)
+		os=-aos
+		;;
+	-atheos*)
+		os=-atheos
+		;;
+	-syllable*)
+		os=-syllable
+		;;
+	-386bsd)
+		os=-bsd
+		;;
+	-ctix* | -uts*)
+		os=-sysv
+		;;
+	-nova*)
+		os=-rtmk-nova
+		;;
+	-ns2 )
+		os=-nextstep2
+		;;
+	-nsk*)
+		os=-nsk
+		;;
+	# Preserve the version number of sinix5.
+	-sinix5.*)
+		os=`echo $os | sed -e 's|sinix|sysv|'`
+		;;
+	-sinix*)
+		os=-sysv4
+		;;
+	-tpf*)
+		os=-tpf
+		;;
+	-triton*)
+		os=-sysv3
+		;;
+	-oss*)
+		os=-sysv3
+		;;
+	-svr4)
+		os=-sysv4
+		;;
+	-svr3)
+		os=-sysv3
+		;;
+	-sysvr4)
+		os=-sysv4
+		;;
+	# This must come after -sysvr4.
+	-sysv*)
+		;;
+	-ose*)
+		os=-ose
+		;;
+	-es1800*)
+		os=-ose
+		;;
+	-xenix)
+		os=-xenix
+		;;
+	-*mint | -mint[0-9]* | -*MiNT | -MiNT[0-9]*)
+		os=-mint
+		;;
+	-aros*)
+		os=-aros
+		;;
+	-zvmoe)
+		os=-zvmoe
+		;;
+	-dicos*)
+		os=-dicos
+		;;
+	-nacl*)
+		;;
+	-ios)
+		;;
+	-none)
+		;;
+	*)
+		# Get rid of the `-' at the beginning of $os.
+		os=`echo $os | sed 's/[^-]*-//'`
+		echo Invalid configuration \`$1\': system \`$os\' not recognized 1>&2
+		exit 1
+		;;
+esac
+else
+
+# Here we handle the default operating systems that come with various machines.
+# The value should be what the vendor currently ships out the door with their
+# machine or put another way, the most popular os provided with the machine.
+
+# Note that if you're going to try to match "-MANUFACTURER" here (say,
+# "-sun"), then you have to tell the case statement up towards the top
+# that MANUFACTURER isn't an operating system.  Otherwise, code above
+# will signal an error saying that MANUFACTURER isn't an operating
+# system, and we'll never get to this point.
+
+case $basic_machine in
+	score-*)
+		os=-elf
+		;;
+	spu-*)
+		os=-elf
+		;;
+	*-acorn)
+		os=-riscix1.2
+		;;
+	arm*-rebel)
+		os=-linux
+		;;
+	arm*-semi)
+		os=-aout
+		;;
+	c4x-* | tic4x-*)
+		os=-coff
+		;;
+	c8051-*)
+		os=-elf
+		;;
+	hexagon-*)
+		os=-elf
+		;;
+	tic54x-*)
+		os=-coff
+		;;
+	tic55x-*)
+		os=-coff
+		;;
+	tic6x-*)
+		os=-coff
+		;;
+	# This must come before the *-dec entry.
+	pdp10-*)
+		os=-tops20
+		;;
+	pdp11-*)
+		os=-none
+		;;
+	*-dec | vax-*)
+		os=-ultrix4.2
+		;;
+	m68*-apollo)
+		os=-domain
+		;;
+	i386-sun)
+		os=-sunos4.0.2
+		;;
+	m68000-sun)
+		os=-sunos3
+		;;
+	m68*-cisco)
+		os=-aout
+		;;
+	mep-*)
+		os=-elf
+		;;
+	mips*-cisco)
+		os=-elf
+		;;
+	mips*-*)
+		os=-elf
+		;;
+	or32-*)
+		os=-coff
+		;;
+	*-tti)	# must be before sparc entry or we get the wrong os.
+		os=-sysv3
+		;;
+	sparc-* | *-sun)
+		os=-sunos4.1.1
+		;;
+	*-be)
+		os=-beos
+		;;
+	*-haiku)
+		os=-haiku
+		;;
+	*-ibm)
+		os=-aix
+		;;
+	*-knuth)
+		os=-mmixware
+		;;
+	*-wec)
+		os=-proelf
+		;;
+	*-winbond)
+		os=-proelf
+		;;
+	*-oki)
+		os=-proelf
+		;;
+	*-hp)
+		os=-hpux
+		;;
+	*-hitachi)
+		os=-hiux
+		;;
+	i860-* | *-att | *-ncr | *-altos | *-motorola | *-convergent)
+		os=-sysv
+		;;
+	*-cbm)
+		os=-amigaos
+		;;
+	*-dg)
+		os=-dgux
+		;;
+	*-dolphin)
+		os=-sysv3
+		;;
+	m68k-ccur)
+		os=-rtu
+		;;
+	m88k-omron*)
+		os=-luna
+		;;
+	*-next )
+		os=-nextstep
+		;;
+	*-sequent)
+		os=-ptx
+		;;
+	*-crds)
+		os=-unos
+		;;
+	*-ns)
+		os=-genix
+		;;
+	i370-*)
+		os=-mvs
+		;;
+	*-next)
+		os=-nextstep3
+		;;
+	*-gould)
+		os=-sysv
+		;;
+	*-highlevel)
+		os=-bsd
+		;;
+	*-encore)
+		os=-bsd
+		;;
+	*-sgi)
+		os=-irix
+		;;
+	*-siemens)
+		os=-sysv4
+		;;
+	*-masscomp)
+		os=-rtu
+		;;
+	f30[01]-fujitsu | f700-fujitsu)
+		os=-uxpv
+		;;
+	*-rom68k)
+		os=-coff
+		;;
+	*-*bug)
+		os=-coff
+		;;
+	*-apple)
+		os=-macos
+		;;
+	*-atari*)
+		os=-mint
+		;;
+	*)
+		os=-none
+		;;
+esac
+fi
+
+# Here we handle the case where we know the os, and the CPU type, but not the
+# manufacturer.  We pick the logical manufacturer.
+vendor=unknown
+case $basic_machine in
+	*-unknown)
+		case $os in
+			-riscix*)
+				vendor=acorn
+				;;
+			-sunos*)
+				vendor=sun
+				;;
+			-cnk*|-aix*)
+				vendor=ibm
+				;;
+			-beos*)
+				vendor=be
+				;;
+			-hpux*)
+				vendor=hp
+				;;
+			-mpeix*)
+				vendor=hp
+				;;
+			-hiux*)
+				vendor=hitachi
+				;;
+			-unos*)
+				vendor=crds
+				;;
+			-dgux*)
+				vendor=dg
+				;;
+			-luna*)
+				vendor=omron
+				;;
+			-genix*)
+				vendor=ns
+				;;
+			-mvs* | -opened*)
+				vendor=ibm
+				;;
+			-os400*)
+				vendor=ibm
+				;;
+			-ptx*)
+				vendor=sequent
+				;;
+			-tpf*)
+				vendor=ibm
+				;;
+			-vxsim* | -vxworks* | -windiss*)
+				vendor=wrs
+				;;
+			-aux*)
+				vendor=apple
+				;;
+			-hms*)
+				vendor=hitachi
+				;;
+			-mpw* | -macos*)
+				vendor=apple
+				;;
+			-*mint | -mint[0-9]* | -*MiNT | -MiNT[0-9]*)
+				vendor=atari
+				;;
+			-vos*)
+				vendor=stratus
+				;;
+		esac
+		basic_machine=`echo $basic_machine | sed "s/unknown/$vendor/"`
+		;;
+esac
+
+echo $basic_machine$os
+exit
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "timestamp='"
+# time-stamp-format: "%:y-%02m-%02d"
+# time-stamp-end: "'"
+# End:
diff --git a/jemalloc/build-aux/install-sh b/jemalloc/build-aux/install-sh
new file mode 100755
index 0000000..ebc6691
--- /dev/null
+++ b/jemalloc/build-aux/install-sh
@@ -0,0 +1,250 @@
+#! /bin/sh
+#
+# install - install a program, script, or datafile
+# This comes from X11R5 (mit/util/scripts/install.sh).
+#
+# Copyright 1991 by the Massachusetts Institute of Technology
+#
+# Permission to use, copy, modify, distribute, and sell this software and its
+# documentation for any purpose is hereby granted without fee, provided that
+# the above copyright notice appear in all copies and that both that
+# copyright notice and this permission notice appear in supporting
+# documentation, and that the name of M.I.T. not be used in advertising or
+# publicity pertaining to distribution of the software without specific,
+# written prior permission.  M.I.T. makes no representations about the
+# suitability of this software for any purpose.  It is provided "as is"
+# without express or implied warranty.
+#
+# Calling this script install-sh is preferred over install.sh, to prevent
+# `make' implicit rules from creating a file called install from it
+# when there is no Makefile.
+#
+# This script is compatible with the BSD install script, but was written
+# from scratch.  It can only install one file at a time, a restriction
+# shared with many OS's install programs.
+
+
+# set DOITPROG to echo to test this script
+
+# Don't use :- since 4.3BSD and earlier shells don't like it.
+doit="${DOITPROG-}"
+
+
+# put in absolute paths if you don't have them in your path; or use env. vars.
+
+mvprog="${MVPROG-mv}"
+cpprog="${CPPROG-cp}"
+chmodprog="${CHMODPROG-chmod}"
+chownprog="${CHOWNPROG-chown}"
+chgrpprog="${CHGRPPROG-chgrp}"
+stripprog="${STRIPPROG-strip}"
+rmprog="${RMPROG-rm}"
+mkdirprog="${MKDIRPROG-mkdir}"
+
+transformbasename=""
+transform_arg=""
+instcmd="$mvprog"
+chmodcmd="$chmodprog 0755"
+chowncmd=""
+chgrpcmd=""
+stripcmd=""
+rmcmd="$rmprog -f"
+mvcmd="$mvprog"
+src=""
+dst=""
+dir_arg=""
+
+while [ x"$1" != x ]; do
+    case $1 in
+	-c) instcmd="$cpprog"
+	    shift
+	    continue;;
+
+	-d) dir_arg=true
+	    shift
+	    continue;;
+
+	-m) chmodcmd="$chmodprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-o) chowncmd="$chownprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-g) chgrpcmd="$chgrpprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-s) stripcmd="$stripprog"
+	    shift
+	    continue;;
+
+	-t=*) transformarg=`echo $1 | sed 's/-t=//'`
+	    shift
+	    continue;;
+
+	-b=*) transformbasename=`echo $1 | sed 's/-b=//'`
+	    shift
+	    continue;;
+
+	*)  if [ x"$src" = x ]
+	    then
+		src=$1
+	    else
+		# this colon is to work around a 386BSD /bin/sh bug
+		:
+		dst=$1
+	    fi
+	    shift
+	    continue;;
+    esac
+done
+
+if [ x"$src" = x ]
+then
+	echo "install:	no input file specified"
+	exit 1
+else
+	true
+fi
+
+if [ x"$dir_arg" != x ]; then
+	dst=$src
+	src=""
+	
+	if [ -d $dst ]; then
+		instcmd=:
+	else
+		instcmd=mkdir
+	fi
+else
+
+# Waiting for this to be detected by the "$instcmd $src $dsttmp" command
+# might cause directories to be created, which would be especially bad 
+# if $src (and thus $dsttmp) contains '*'.
+
+	if [ -f $src -o -d $src ]
+	then
+		true
+	else
+		echo "install:  $src does not exist"
+		exit 1
+	fi
+	
+	if [ x"$dst" = x ]
+	then
+		echo "install:	no destination specified"
+		exit 1
+	else
+		true
+	fi
+
+# If destination is a directory, append the input filename; if your system
+# does not like double slashes in filenames, you may need to add some logic
+
+	if [ -d $dst ]
+	then
+		dst="$dst"/`basename $src`
+	else
+		true
+	fi
+fi
+
+## this sed command emulates the dirname command
+dstdir=`echo $dst | sed -e 's,[^/]*$,,;s,/$,,;s,^$,.,'`
+
+# Make sure that the destination directory exists.
+#  this part is taken from Noah Friedman's mkinstalldirs script
+
+# Skip lots of stat calls in the usual case.
+if [ ! -d "$dstdir" ]; then
+defaultIFS='	
+'
+IFS="${IFS-${defaultIFS}}"
+
+oIFS="${IFS}"
+# Some sh's can't handle IFS=/ for some reason.
+IFS='%'
+set - `echo ${dstdir} | sed -e 's@/@%@g' -e 's@^%@/@'`
+IFS="${oIFS}"
+
+pathcomp=''
+
+while [ $# -ne 0 ] ; do
+	pathcomp="${pathcomp}${1}"
+	shift
+
+	if [ ! -d "${pathcomp}" ] ;
+        then
+		$mkdirprog "${pathcomp}"
+	else
+		true
+	fi
+
+	pathcomp="${pathcomp}/"
+done
+fi
+
+if [ x"$dir_arg" != x ]
+then
+	$doit $instcmd $dst &&
+
+	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dst; else true ; fi &&
+	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dst; else true ; fi &&
+	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dst; else true ; fi &&
+	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dst; else true ; fi
+else
+
+# If we're going to rename the final executable, determine the name now.
+
+	if [ x"$transformarg" = x ] 
+	then
+		dstfile=`basename $dst`
+	else
+		dstfile=`basename $dst $transformbasename | 
+			sed $transformarg`$transformbasename
+	fi
+
+# don't allow the sed command to completely eliminate the filename
+
+	if [ x"$dstfile" = x ] 
+	then
+		dstfile=`basename $dst`
+	else
+		true
+	fi
+
+# Make a temp file name in the proper directory.
+
+	dsttmp=$dstdir/#inst.$$#
+
+# Move or copy the file name to the temp name
+
+	$doit $instcmd $src $dsttmp &&
+
+	trap "rm -f ${dsttmp}" 0 &&
+
+# and set any options; do chmod last to preserve setuid bits
+
+# If any of these fail, we abort the whole thing.  If we want to
+# ignore errors from any of these, just make sure not to ignore
+# errors from the above "$doit $instcmd $src $dsttmp" command.
+
+	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dsttmp; else true;fi &&
+	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dsttmp; else true;fi &&
+	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dsttmp; else true;fi &&
+	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dsttmp; else true;fi &&
+
+# Now rename the file to the real destination.
+
+	$doit $rmcmd -f $dstdir/$dstfile &&
+	$doit $mvcmd $dsttmp $dstdir/$dstfile 
+
+fi &&
+
+
+exit 0
diff --git a/jemalloc/config.stamp.in b/jemalloc/config.stamp.in
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/jemalloc/config.stamp.in
diff --git a/jemalloc/configure.ac b/jemalloc/configure.ac
new file mode 100644
index 0000000..ea6b756
--- /dev/null
+++ b/jemalloc/configure.ac
@@ -0,0 +1,2410 @@
+dnl Process this file with autoconf to produce a configure script.
+AC_PREREQ(2.63)
+AC_INIT([Makefile.in])
+
+AC_CONFIG_AUX_DIR([build-aux])
+
+dnl ============================================================================
+dnl Custom macro definitions.
+
+dnl JE_CONCAT_VVV(r, a, b)
+dnl
+dnl Set $r to the concatenation of $a and $b, with a space separating them iff
+dnl both $a and $b are non-empty.
+AC_DEFUN([JE_CONCAT_VVV],
+if test "x[$]{$2}" = "x" -o "x[$]{$3}" = "x" ; then
+  $1="[$]{$2}[$]{$3}"
+else
+  $1="[$]{$2} [$]{$3}"
+fi
+)
+
+dnl JE_APPEND_VS(a, b)
+dnl
+dnl Set $a to the concatenation of $a and b, with a space separating them iff
+dnl both $a and b are non-empty.
+AC_DEFUN([JE_APPEND_VS],
+  T_APPEND_V=$2
+  JE_CONCAT_VVV($1, $1, T_APPEND_V)
+)
+
+CONFIGURE_CFLAGS=
+SPECIFIED_CFLAGS="${CFLAGS}"
+dnl JE_CFLAGS_ADD(cflag)
+dnl
+dnl CFLAGS is the concatenation of CONFIGURE_CFLAGS and SPECIFIED_CFLAGS
+dnl (ignoring EXTRA_CFLAGS, which does not impact configure tests.  This macro
+dnl appends to CONFIGURE_CFLAGS and regenerates CFLAGS.
+AC_DEFUN([JE_CFLAGS_ADD],
+[
+AC_MSG_CHECKING([whether compiler supports $1])
+T_CONFIGURE_CFLAGS="${CONFIGURE_CFLAGS}"
+JE_APPEND_VS(CONFIGURE_CFLAGS, $1)
+JE_CONCAT_VVV(CFLAGS, CONFIGURE_CFLAGS, SPECIFIED_CFLAGS)
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+]], [[
+    return 0;
+]])],
+              [je_cv_cflags_added=$1]
+              AC_MSG_RESULT([yes]),
+              [je_cv_cflags_added=]
+              AC_MSG_RESULT([no])
+              [CONFIGURE_CFLAGS="${T_CONFIGURE_CFLAGS}"]
+)
+JE_CONCAT_VVV(CFLAGS, CONFIGURE_CFLAGS, SPECIFIED_CFLAGS)
+])
+
+dnl JE_CFLAGS_SAVE()
+dnl JE_CFLAGS_RESTORE()
+dnl
+dnl Save/restore CFLAGS.  Nesting is not supported.
+AC_DEFUN([JE_CFLAGS_SAVE],
+SAVED_CONFIGURE_CFLAGS="${CONFIGURE_CFLAGS}"
+)
+AC_DEFUN([JE_CFLAGS_RESTORE],
+CONFIGURE_CFLAGS="${SAVED_CONFIGURE_CFLAGS}"
+JE_CONCAT_VVV(CFLAGS, CONFIGURE_CFLAGS, SPECIFIED_CFLAGS)
+)
+
+CONFIGURE_CXXFLAGS=
+SPECIFIED_CXXFLAGS="${CXXFLAGS}"
+dnl JE_CXXFLAGS_ADD(cxxflag)
+AC_DEFUN([JE_CXXFLAGS_ADD],
+[
+AC_MSG_CHECKING([whether compiler supports $1])
+T_CONFIGURE_CXXFLAGS="${CONFIGURE_CXXFLAGS}"
+JE_APPEND_VS(CONFIGURE_CXXFLAGS, $1)
+JE_CONCAT_VVV(CXXFLAGS, CONFIGURE_CXXFLAGS, SPECIFIED_CXXFLAGS)
+AC_LANG_PUSH([C++])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+]], [[
+    return 0;
+]])],
+              [je_cv_cxxflags_added=$1]
+              AC_MSG_RESULT([yes]),
+              [je_cv_cxxflags_added=]
+              AC_MSG_RESULT([no])
+              [CONFIGURE_CXXFLAGS="${T_CONFIGURE_CXXFLAGS}"]
+)
+AC_LANG_POP([C++])
+JE_CONCAT_VVV(CXXFLAGS, CONFIGURE_CXXFLAGS, SPECIFIED_CXXFLAGS)
+])
+
+dnl JE_COMPILABLE(label, hcode, mcode, rvar)
+dnl
+dnl Use AC_LINK_IFELSE() rather than AC_COMPILE_IFELSE() so that linker errors
+dnl cause failure.
+AC_DEFUN([JE_COMPILABLE],
+[
+AC_CACHE_CHECK([whether $1 is compilable],
+               [$4],
+               [AC_LINK_IFELSE([AC_LANG_PROGRAM([$2],
+                                                [$3])],
+                               [$4=yes],
+                               [$4=no])])
+])
+
+dnl ============================================================================
+
+CONFIG=`echo ${ac_configure_args} | sed -e 's#'"'"'\([^ ]*\)'"'"'#\1#g'`
+AC_SUBST([CONFIG])
+
+dnl Library revision.
+rev=2
+AC_SUBST([rev])
+
+srcroot=$srcdir
+if test "x${srcroot}" = "x." ; then
+  srcroot=""
+else
+  srcroot="${srcroot}/"
+fi
+AC_SUBST([srcroot])
+abs_srcroot="`cd \"${srcdir}\"; pwd`/"
+AC_SUBST([abs_srcroot])
+
+objroot=""
+AC_SUBST([objroot])
+abs_objroot="`pwd`/"
+AC_SUBST([abs_objroot])
+
+dnl Munge install path variables.
+if test "x$prefix" = "xNONE" ; then
+  prefix="/usr/local"
+fi
+if test "x$exec_prefix" = "xNONE" ; then
+  exec_prefix=$prefix
+fi
+PREFIX=$prefix
+AC_SUBST([PREFIX])
+BINDIR=`eval echo $bindir`
+BINDIR=`eval echo $BINDIR`
+AC_SUBST([BINDIR])
+INCLUDEDIR=`eval echo $includedir`
+INCLUDEDIR=`eval echo $INCLUDEDIR`
+AC_SUBST([INCLUDEDIR])
+LIBDIR=`eval echo $libdir`
+LIBDIR=`eval echo $LIBDIR`
+AC_SUBST([LIBDIR])
+DATADIR=`eval echo $datadir`
+DATADIR=`eval echo $DATADIR`
+AC_SUBST([DATADIR])
+MANDIR=`eval echo $mandir`
+MANDIR=`eval echo $MANDIR`
+AC_SUBST([MANDIR])
+
+dnl Support for building documentation.
+AC_PATH_PROG([XSLTPROC], [xsltproc], [false], [$PATH])
+if test -d "/usr/share/xml/docbook/stylesheet/docbook-xsl" ; then
+  DEFAULT_XSLROOT="/usr/share/xml/docbook/stylesheet/docbook-xsl"
+elif test -d "/usr/share/sgml/docbook/xsl-stylesheets" ; then
+  DEFAULT_XSLROOT="/usr/share/sgml/docbook/xsl-stylesheets"
+else
+  dnl Documentation building will fail if this default gets used.
+  DEFAULT_XSLROOT=""
+fi
+AC_ARG_WITH([xslroot],
+  [AS_HELP_STRING([--with-xslroot=<path>], [XSL stylesheet root path])], [
+if test "x$with_xslroot" = "xno" ; then
+  XSLROOT="${DEFAULT_XSLROOT}"
+else
+  XSLROOT="${with_xslroot}"
+fi
+],
+  XSLROOT="${DEFAULT_XSLROOT}"
+)
+if test "x$XSLTPROC" = "xfalse" ; then
+  XSLROOT=""
+fi
+AC_SUBST([XSLROOT])
+
+dnl If CFLAGS isn't defined, set CFLAGS to something reasonable.  Otherwise,
+dnl just prevent autoconf from molesting CFLAGS.
+CFLAGS=$CFLAGS
+AC_PROG_CC
+
+if test "x$GCC" != "xyes" ; then
+  AC_CACHE_CHECK([whether compiler is MSVC],
+                 [je_cv_msvc],
+                 [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([],
+                                                     [
+#ifndef _MSC_VER
+  int fail[-1];
+#endif
+])],
+                               [je_cv_msvc=yes],
+                               [je_cv_msvc=no])])
+fi
+
+dnl check if a cray prgenv wrapper compiler is being used
+je_cv_cray_prgenv_wrapper=""
+if test "x${PE_ENV}" != "x" ; then
+  case "${CC}" in
+    CC|cc)
+	je_cv_cray_prgenv_wrapper="yes"
+	;;
+    *)
+       ;;
+  esac
+fi
+
+AC_CACHE_CHECK([whether compiler is cray],
+              [je_cv_cray],
+              [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([],
+                                                  [
+#ifndef _CRAYC
+  int fail[-1];
+#endif
+])],
+                            [je_cv_cray=yes],
+                            [je_cv_cray=no])])
+
+if test "x${je_cv_cray}" = "xyes" ; then
+  AC_CACHE_CHECK([whether cray compiler version is 8.4],
+                [je_cv_cray_84],
+                [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([],
+                                                      [
+#if !(_RELEASE_MAJOR == 8 && _RELEASE_MINOR == 4)
+  int fail[-1];
+#endif
+])],
+                              [je_cv_cray_84=yes],
+                              [je_cv_cray_84=no])])
+fi
+
+if test "x$GCC" = "xyes" ; then
+  JE_CFLAGS_ADD([-std=gnu11])
+  if test "x$je_cv_cflags_added" = "x-std=gnu11" ; then
+    AC_DEFINE_UNQUOTED([JEMALLOC_HAS_RESTRICT])
+  else
+    JE_CFLAGS_ADD([-std=gnu99])
+    if test "x$je_cv_cflags_added" = "x-std=gnu99" ; then
+      AC_DEFINE_UNQUOTED([JEMALLOC_HAS_RESTRICT])
+    fi
+  fi
+  JE_CFLAGS_ADD([-Wall])
+  JE_CFLAGS_ADD([-Wextra])
+  JE_CFLAGS_ADD([-Wshorten-64-to-32])
+  JE_CFLAGS_ADD([-Wsign-compare])
+  JE_CFLAGS_ADD([-Wundef])
+  JE_CFLAGS_ADD([-Wno-format-zero-length])
+  JE_CFLAGS_ADD([-pipe])
+  JE_CFLAGS_ADD([-g3])
+elif test "x$je_cv_msvc" = "xyes" ; then
+  CC="$CC -nologo"
+  JE_CFLAGS_ADD([-Zi])
+  JE_CFLAGS_ADD([-MT])
+  JE_CFLAGS_ADD([-W3])
+  JE_CFLAGS_ADD([-FS])
+  JE_APPEND_VS(CPPFLAGS, -I${srcdir}/include/msvc_compat)
+fi
+if test "x$je_cv_cray" = "xyes" ; then
+  dnl cray compiler 8.4 has an inlining bug
+  if test "x$je_cv_cray_84" = "xyes" ; then
+    JE_CFLAGS_ADD([-hipa2])
+    JE_CFLAGS_ADD([-hnognu])
+  fi
+  dnl ignore unreachable code warning
+  JE_CFLAGS_ADD([-hnomessage=128])
+  dnl ignore redefinition of "malloc", "free", etc warning
+  JE_CFLAGS_ADD([-hnomessage=1357])
+fi
+AC_SUBST([CONFIGURE_CFLAGS])
+AC_SUBST([SPECIFIED_CFLAGS])
+AC_SUBST([EXTRA_CFLAGS])
+AC_PROG_CPP
+
+AC_ARG_ENABLE([cxx],
+  [AS_HELP_STRING([--disable-cxx], [Disable C++ integration])],
+if test "x$enable_cxx" = "xno" ; then
+  enable_cxx="0"
+else
+  enable_cxx="1"
+fi
+,
+enable_cxx="1"
+)
+if test "x$enable_cxx" = "x1" ; then
+  dnl Require at least c++14, which is the first version to support sized
+  dnl deallocation.  C++ support is not compiled otherwise.
+  m4_include([m4/ax_cxx_compile_stdcxx.m4])
+  AX_CXX_COMPILE_STDCXX([14], [noext], [optional])
+  if test "x${HAVE_CXX14}" = "x1" ; then
+    JE_CXXFLAGS_ADD([-Wall])
+    JE_CXXFLAGS_ADD([-Wextra])
+    JE_CXXFLAGS_ADD([-g3])
+
+    SAVED_LIBS="${LIBS}"
+    JE_APPEND_VS(LIBS, -lstdc++)
+    JE_COMPILABLE([libstdc++ linkage], [
+#include <stdlib.h>
+], [[
+	int *arr = (int *)malloc(sizeof(int) * 42);
+	if (arr == NULL)
+		return 1;
+]], [je_cv_libstdcxx])
+    if test "x${je_cv_libstdcxx}" = "xno" ; then
+      LIBS="${SAVED_LIBS}"
+    fi
+  else
+    enable_cxx="0"
+  fi
+fi
+AC_SUBST([enable_cxx])
+AC_SUBST([CONFIGURE_CXXFLAGS])
+AC_SUBST([SPECIFIED_CXXFLAGS])
+AC_SUBST([EXTRA_CXXFLAGS])
+
+AC_C_BIGENDIAN([ac_cv_big_endian=1], [ac_cv_big_endian=0])
+if test "x${ac_cv_big_endian}" = "x1" ; then
+  AC_DEFINE_UNQUOTED([JEMALLOC_BIG_ENDIAN], [ ])
+fi
+
+if test "x${je_cv_msvc}" = "xyes" -a "x${ac_cv_header_inttypes_h}" = "xno"; then
+  JE_APPEND_VS(CPPFLAGS, -I${srcdir}/include/msvc_compat/C99)
+fi
+
+if test "x${je_cv_msvc}" = "xyes" ; then
+  LG_SIZEOF_PTR=LG_SIZEOF_PTR_WIN
+  AC_MSG_RESULT([Using a predefined value for sizeof(void *): 4 for 32-bit, 8 for 64-bit])
+else
+  AC_CHECK_SIZEOF([void *])
+  if test "x${ac_cv_sizeof_void_p}" = "x8" ; then
+    LG_SIZEOF_PTR=3
+  elif test "x${ac_cv_sizeof_void_p}" = "x4" ; then
+    LG_SIZEOF_PTR=2
+  else
+    AC_MSG_ERROR([Unsupported pointer size: ${ac_cv_sizeof_void_p}])
+  fi
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_PTR], [$LG_SIZEOF_PTR])
+
+AC_CHECK_SIZEOF([int])
+if test "x${ac_cv_sizeof_int}" = "x8" ; then
+  LG_SIZEOF_INT=3
+elif test "x${ac_cv_sizeof_int}" = "x4" ; then
+  LG_SIZEOF_INT=2
+else
+  AC_MSG_ERROR([Unsupported int size: ${ac_cv_sizeof_int}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_INT], [$LG_SIZEOF_INT])
+
+AC_CHECK_SIZEOF([long])
+if test "x${ac_cv_sizeof_long}" = "x8" ; then
+  LG_SIZEOF_LONG=3
+elif test "x${ac_cv_sizeof_long}" = "x4" ; then
+  LG_SIZEOF_LONG=2
+else
+  AC_MSG_ERROR([Unsupported long size: ${ac_cv_sizeof_long}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_LONG], [$LG_SIZEOF_LONG])
+
+AC_CHECK_SIZEOF([long long])
+if test "x${ac_cv_sizeof_long_long}" = "x8" ; then
+  LG_SIZEOF_LONG_LONG=3
+elif test "x${ac_cv_sizeof_long_long}" = "x4" ; then
+  LG_SIZEOF_LONG_LONG=2
+else
+  AC_MSG_ERROR([Unsupported long long size: ${ac_cv_sizeof_long_long}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_LONG_LONG], [$LG_SIZEOF_LONG_LONG])
+
+AC_CHECK_SIZEOF([intmax_t])
+if test "x${ac_cv_sizeof_intmax_t}" = "x16" ; then
+  LG_SIZEOF_INTMAX_T=4
+elif test "x${ac_cv_sizeof_intmax_t}" = "x8" ; then
+  LG_SIZEOF_INTMAX_T=3
+elif test "x${ac_cv_sizeof_intmax_t}" = "x4" ; then
+  LG_SIZEOF_INTMAX_T=2
+else
+  AC_MSG_ERROR([Unsupported intmax_t size: ${ac_cv_sizeof_intmax_t}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_INTMAX_T], [$LG_SIZEOF_INTMAX_T])
+
+AC_CANONICAL_HOST
+dnl CPU-specific settings.
+CPU_SPINWAIT=""
+case "${host_cpu}" in
+  i686|x86_64)
+	HAVE_CPU_SPINWAIT=1
+	if test "x${je_cv_msvc}" = "xyes" ; then
+	    AC_CACHE_VAL([je_cv_pause_msvc],
+	      [JE_COMPILABLE([pause instruction MSVC], [],
+					[[_mm_pause(); return 0;]],
+					[je_cv_pause_msvc])])
+	    if test "x${je_cv_pause_msvc}" = "xyes" ; then
+		CPU_SPINWAIT='_mm_pause()'
+	    fi
+	else
+	    AC_CACHE_VAL([je_cv_pause],
+	      [JE_COMPILABLE([pause instruction], [],
+					[[__asm__ volatile("pause"); return 0;]],
+					[je_cv_pause])])
+	    if test "x${je_cv_pause}" = "xyes" ; then
+		CPU_SPINWAIT='__asm__ volatile("pause")'
+	    fi
+	fi
+	;;
+  *)
+	HAVE_CPU_SPINWAIT=0
+	;;
+esac
+AC_DEFINE_UNQUOTED([HAVE_CPU_SPINWAIT], [$HAVE_CPU_SPINWAIT])
+AC_DEFINE_UNQUOTED([CPU_SPINWAIT], [$CPU_SPINWAIT])
+
+AC_ARG_WITH([lg_vaddr],
+  [AS_HELP_STRING([--with-lg-vaddr=<lg-vaddr>], [Number of significant virtual address bits])],
+  [LG_VADDR="$with_lg_vaddr"], [LG_VADDR="detect"])
+
+case "${host_cpu}" in
+  aarch64)
+    if test "x$LG_VADDR" = "xdetect"; then
+      AC_MSG_CHECKING([number of significant virtual address bits])
+      if test "x${LG_SIZEOF_PTR}" = "x2" ; then
+        #aarch64 ILP32
+        LG_VADDR=32
+      else
+        #aarch64 LP64
+        LG_VADDR=48
+      fi
+      AC_MSG_RESULT([$LG_VADDR])
+    fi
+    ;;
+  x86_64)
+    if test "x$LG_VADDR" = "xdetect"; then
+      AC_CACHE_CHECK([number of significant virtual address bits],
+                     [je_cv_lg_vaddr],
+                     AC_RUN_IFELSE([AC_LANG_PROGRAM(
+[[
+#include <stdio.h>
+#ifdef _WIN32
+#include <limits.h>
+#include <intrin.h>
+typedef unsigned __int32 uint32_t;
+#else
+#include <stdint.h>
+#endif
+]], [[
+	uint32_t r[[4]];
+	uint32_t eax_in = 0x80000008U;
+#ifdef _WIN32
+	__cpuid((int *)r, (int)eax_in);
+#else
+	asm volatile ("cpuid"
+	    : "=a" (r[[0]]), "=b" (r[[1]]), "=c" (r[[2]]), "=d" (r[[3]])
+	    : "a" (eax_in), "c" (0)
+	);
+#endif
+	uint32_t eax_out = r[[0]];
+	uint32_t vaddr = ((eax_out & 0x0000ff00U) >> 8);
+	FILE *f = fopen("conftest.out", "w");
+	if (f == NULL) {
+		return 1;
+	}
+	if (vaddr > (sizeof(void *) << 3)) {
+		vaddr = sizeof(void *) << 3;
+	}
+	fprintf(f, "%u", vaddr);
+	fclose(f);
+	return 0;
+]])],
+                   [je_cv_lg_vaddr=`cat conftest.out`],
+                   [je_cv_lg_vaddr=error],
+                   [je_cv_lg_vaddr=57]))
+      if test "x${je_cv_lg_vaddr}" != "x" ; then
+        LG_VADDR="${je_cv_lg_vaddr}"
+      fi
+      if test "x${LG_VADDR}" != "xerror" ; then
+        AC_DEFINE_UNQUOTED([LG_VADDR], [$LG_VADDR])
+      else
+        AC_MSG_ERROR([cannot determine number of significant virtual address bits])
+      fi
+    fi
+    ;;
+  *)
+    if test "x$LG_VADDR" = "xdetect"; then
+      AC_MSG_CHECKING([number of significant virtual address bits])
+      if test "x${LG_SIZEOF_PTR}" = "x3" ; then
+        LG_VADDR=64
+      elif test "x${LG_SIZEOF_PTR}" = "x2" ; then
+        LG_VADDR=32
+      elif test "x${LG_SIZEOF_PTR}" = "xLG_SIZEOF_PTR_WIN" ; then
+        LG_VADDR="(1U << (LG_SIZEOF_PTR_WIN+3))"
+      else
+        AC_MSG_ERROR([Unsupported lg(pointer size): ${LG_SIZEOF_PTR}])
+      fi
+      AC_MSG_RESULT([$LG_VADDR])
+    fi
+    ;;
+esac
+AC_DEFINE_UNQUOTED([LG_VADDR], [$LG_VADDR])
+
+LD_PRELOAD_VAR="LD_PRELOAD"
+so="so"
+importlib="${so}"
+o="$ac_objext"
+a="a"
+exe="$ac_exeext"
+libprefix="lib"
+link_whole_archive="0"
+DSO_LDFLAGS='-shared -Wl,-soname,$(@F)'
+RPATH='-Wl,-rpath,$(1)'
+SOREV="${so}.${rev}"
+PIC_CFLAGS='-fPIC -DPIC'
+CTARGET='-o $@'
+LDTARGET='-o $@'
+TEST_LD_MODE=
+EXTRA_LDFLAGS=
+ARFLAGS='crus'
+AROUT=' $@'
+CC_MM=1
+
+if test "x$je_cv_cray_prgenv_wrapper" = "xyes" ; then
+  TEST_LD_MODE='-dynamic'
+fi
+
+if test "x${je_cv_cray}" = "xyes" ; then
+  CC_MM=
+fi
+
+AN_MAKEVAR([AR], [AC_PROG_AR])
+AN_PROGRAM([ar], [AC_PROG_AR])
+AC_DEFUN([AC_PROG_AR], [AC_CHECK_TOOL(AR, ar, :)])
+AC_PROG_AR
+
+AN_MAKEVAR([NM], [AC_PROG_NM])
+AN_PROGRAM([nm], [AC_PROG_NM])
+AC_DEFUN([AC_PROG_NM], [AC_CHECK_TOOL(NM, nm, :)])
+AC_PROG_NM
+
+AC_PROG_AWK
+
+dnl ============================================================================
+dnl jemalloc version.
+dnl
+
+AC_ARG_WITH([version],
+  [AS_HELP_STRING([--with-version=<major>.<minor>.<bugfix>-<nrev>-g<gid>],
+   [Version string])],
+  [
+    echo "${with_version}" | grep ['^[0-9]\+\.[0-9]\+\.[0-9]\+-[0-9]\+-g[0-9a-f]\+$'] 2>&1 1>/dev/null
+    if test $? -eq 0 ; then
+      echo "$with_version" > "${objroot}VERSION"
+    else
+      echo "${with_version}" | grep ['^VERSION$'] 2>&1 1>/dev/null
+      if test $? -ne 0 ; then
+        AC_MSG_ERROR([${with_version} does not match <major>.<minor>.<bugfix>-<nrev>-g<gid> or VERSION])
+      fi
+    fi
+  ], [
+    dnl Set VERSION if source directory is inside a git repository.
+    if test "x`test ! \"${srcroot}\" && cd \"${srcroot}\"; git rev-parse --is-inside-work-tree 2>/dev/null`" = "xtrue" ; then
+      dnl Pattern globs aren't powerful enough to match both single- and
+      dnl double-digit version numbers, so iterate over patterns to support up
+      dnl to version 99.99.99 without any accidental matches.
+      for pattern in ['[0-9].[0-9].[0-9]' '[0-9].[0-9].[0-9][0-9]' \
+                     '[0-9].[0-9][0-9].[0-9]' '[0-9].[0-9][0-9].[0-9][0-9]' \
+                     '[0-9][0-9].[0-9].[0-9]' '[0-9][0-9].[0-9].[0-9][0-9]' \
+                     '[0-9][0-9].[0-9][0-9].[0-9]' \
+                     '[0-9][0-9].[0-9][0-9].[0-9][0-9]']; do
+        (test ! "${srcroot}" && cd "${srcroot}"; git describe --long --abbrev=40 --match="${pattern}") > "${objroot}VERSION.tmp" 2>/dev/null
+        if test $? -eq 0 ; then
+          mv "${objroot}VERSION.tmp" "${objroot}VERSION"
+          break
+        fi
+      done
+    fi
+    rm -f "${objroot}VERSION.tmp"
+  ])
+
+if test ! -e "${objroot}VERSION" ; then
+  if test ! -e "${srcroot}VERSION" ; then
+    AC_MSG_RESULT(
+      [Missing VERSION file, and unable to generate it; creating bogus VERSION])
+    echo "0.0.0-0-g0000000000000000000000000000000000000000" > "${objroot}VERSION"
+  else
+    cp ${srcroot}VERSION ${objroot}VERSION
+  fi
+fi
+jemalloc_version=`cat "${objroot}VERSION"`
+jemalloc_version_major=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]1}'`
+jemalloc_version_minor=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]2}'`
+jemalloc_version_bugfix=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]3}'`
+jemalloc_version_nrev=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]4}'`
+jemalloc_version_gid=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]5}'`
+AC_SUBST([jemalloc_version])
+AC_SUBST([jemalloc_version_major])
+AC_SUBST([jemalloc_version_minor])
+AC_SUBST([jemalloc_version_bugfix])
+AC_SUBST([jemalloc_version_nrev])
+AC_SUBST([jemalloc_version_gid])
+
+dnl Platform-specific settings.  abi and RPATH can probably be determined
+dnl programmatically, but doing so is error-prone, which makes it generally
+dnl not worth the trouble.
+dnl
+dnl Define cpp macros in CPPFLAGS, rather than doing AC_DEFINE(macro), since the
+dnl definitions need to be seen before any headers are included, which is a pain
+dnl to make happen otherwise.
+default_retain="0"
+maps_coalesce="1"
+DUMP_SYMS="${NM} -a"
+SYM_PREFIX=""
+case "${host}" in
+  *-*-darwin* | *-*-ios*)
+	abi="macho"
+	RPATH=""
+	LD_PRELOAD_VAR="DYLD_INSERT_LIBRARIES"
+	so="dylib"
+	importlib="${so}"
+	force_tls="0"
+	DSO_LDFLAGS='-shared -Wl,-install_name,$(LIBDIR)/$(@F)'
+	SOREV="${rev}.${so}"
+	sbrk_deprecated="1"
+	SYM_PREFIX="_"
+	;;
+  *-*-freebsd*)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_SYSCTL_VM_OVERCOMMIT], [ ])
+	force_lazy_lock="1"
+	;;
+  *-*-dragonfly*)
+	abi="elf"
+	;;
+  *-*-openbsd*)
+	abi="elf"
+	force_tls="0"
+	;;
+  *-*-bitrig*)
+	abi="elf"
+	;;
+  *-*-linux-android)
+	dnl syscall(2) and secure_getenv(3) are exposed by _GNU_SOURCE.
+	JE_APPEND_VS(CPPFLAGS, -D_GNU_SOURCE)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS], [ ])
+	AC_DEFINE([JEMALLOC_HAS_ALLOCA_H])
+	AC_DEFINE([JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY], [ ])
+	AC_DEFINE([JEMALLOC_THREADED_INIT], [ ])
+	AC_DEFINE([JEMALLOC_C11_ATOMICS])
+	force_tls="0"
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  default_retain="1"
+	fi
+	;;
+  *-*-linux*)
+	dnl syscall(2) and secure_getenv(3) are exposed by _GNU_SOURCE.
+	JE_APPEND_VS(CPPFLAGS, -D_GNU_SOURCE)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS], [ ])
+	AC_DEFINE([JEMALLOC_HAS_ALLOCA_H])
+	AC_DEFINE([JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY], [ ])
+	AC_DEFINE([JEMALLOC_THREADED_INIT], [ ])
+	AC_DEFINE([JEMALLOC_USE_CXX_THROW], [ ])
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  default_retain="1"
+	fi
+	;;
+  *-*-kfreebsd*)
+	dnl syscall(2) and secure_getenv(3) are exposed by _GNU_SOURCE.
+	JE_APPEND_VS(CPPFLAGS, -D_GNU_SOURCE)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_HAS_ALLOCA_H])
+	AC_DEFINE([JEMALLOC_SYSCTL_VM_OVERCOMMIT], [ ])
+	AC_DEFINE([JEMALLOC_THREADED_INIT], [ ])
+	AC_DEFINE([JEMALLOC_USE_CXX_THROW], [ ])
+	;;
+  *-*-netbsd*)
+	AC_MSG_CHECKING([ABI])
+        AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[#ifdef __ELF__
+/* ELF */
+#else
+#error aout
+#endif
+]])],
+                          [abi="elf"],
+                          [abi="aout"])
+	AC_MSG_RESULT([$abi])
+	;;
+  *-*-solaris2*)
+	abi="elf"
+	RPATH='-Wl,-R,$(1)'
+	dnl Solaris needs this for sigwait().
+	JE_APPEND_VS(CPPFLAGS, -D_POSIX_PTHREAD_SEMANTICS)
+	JE_APPEND_VS(LIBS, -lposix4 -lsocket -lnsl)
+	;;
+  *-ibm-aix*)
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  dnl 64bit AIX
+	  LD_PRELOAD_VAR="LDR_PRELOAD64"
+	else
+	  dnl 32bit AIX
+	  LD_PRELOAD_VAR="LDR_PRELOAD"
+	fi
+	abi="xcoff"
+	;;
+  *-*-mingw* | *-*-cygwin*)
+	abi="pecoff"
+	force_tls="0"
+	maps_coalesce="0"
+	RPATH=""
+	so="dll"
+	if test "x$je_cv_msvc" = "xyes" ; then
+	  importlib="lib"
+	  DSO_LDFLAGS="-LD"
+	  EXTRA_LDFLAGS="-link -DEBUG"
+	  CTARGET='-Fo$@'
+	  LDTARGET='-Fe$@'
+	  AR='lib'
+	  ARFLAGS='-nologo -out:'
+	  AROUT='$@'
+	  CC_MM=
+        else
+	  importlib="${so}"
+	  DSO_LDFLAGS="-shared"
+	  link_whole_archive="1"
+	fi
+	case "${host}" in
+	  *-*-cygwin*)
+	    DUMP_SYMS="dumpbin /SYMBOLS"
+	    ;;
+	  *)
+	    ;;
+	esac
+	a="lib"
+	libprefix=""
+	SOREV="${so}"
+	PIC_CFLAGS=""
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  default_retain="1"
+	fi
+	;;
+  *)
+	AC_MSG_RESULT([Unsupported operating system: ${host}])
+	abi="elf"
+	;;
+esac
+
+JEMALLOC_USABLE_SIZE_CONST=const
+AC_CHECK_HEADERS([malloc.h], [
+  AC_MSG_CHECKING([whether malloc_usable_size definition can use const argument])
+  AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+    [#include <malloc.h>
+     #include <stddef.h>
+    size_t malloc_usable_size(const void *ptr);
+    ],
+    [])],[
+                AC_MSG_RESULT([yes])
+         ],[
+                JEMALLOC_USABLE_SIZE_CONST=
+                AC_MSG_RESULT([no])
+         ])
+])
+AC_DEFINE_UNQUOTED([JEMALLOC_USABLE_SIZE_CONST], [$JEMALLOC_USABLE_SIZE_CONST])
+AC_SUBST([abi])
+AC_SUBST([RPATH])
+AC_SUBST([LD_PRELOAD_VAR])
+AC_SUBST([so])
+AC_SUBST([importlib])
+AC_SUBST([o])
+AC_SUBST([a])
+AC_SUBST([exe])
+AC_SUBST([libprefix])
+AC_SUBST([link_whole_archive])
+AC_SUBST([DSO_LDFLAGS])
+AC_SUBST([EXTRA_LDFLAGS])
+AC_SUBST([SOREV])
+AC_SUBST([PIC_CFLAGS])
+AC_SUBST([CTARGET])
+AC_SUBST([LDTARGET])
+AC_SUBST([TEST_LD_MODE])
+AC_SUBST([MKLIB])
+AC_SUBST([ARFLAGS])
+AC_SUBST([AROUT])
+AC_SUBST([DUMP_SYMS])
+AC_SUBST([CC_MM])
+
+dnl Determine whether libm must be linked to use e.g. log(3).
+AC_SEARCH_LIBS([log], [m], , [AC_MSG_ERROR([Missing math functions])])
+if test "x$ac_cv_search_log" != "xnone required" ; then
+  LM="$ac_cv_search_log"
+else
+  LM=
+fi
+AC_SUBST(LM)
+
+JE_COMPILABLE([__attribute__ syntax],
+              [static __attribute__((unused)) void foo(void){}],
+              [],
+              [je_cv_attribute])
+if test "x${je_cv_attribute}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR], [ ])
+  if test "x${GCC}" = "xyes" -a "x${abi}" = "xelf"; then
+    JE_CFLAGS_ADD([-fvisibility=hidden])
+    JE_CXXFLAGS_ADD([-fvisibility=hidden])
+  fi
+fi
+dnl Check for tls_model attribute support (clang 3.0 still lacks support).
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([tls_model attribute], [],
+              [static __thread int
+               __attribute__((tls_model("initial-exec"), unused)) foo;
+               foo = 0;],
+              [je_cv_tls_model])
+JE_CFLAGS_RESTORE()
+dnl (Setting of JEMALLOC_TLS_MODEL is done later, after we've checked for
+dnl --disable-initial-exec-tls)
+
+dnl Check for alloc_size attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([alloc_size attribute], [#include <stdlib.h>],
+              [void *foo(size_t size) __attribute__((alloc_size(1)));],
+              [je_cv_alloc_size])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_alloc_size}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_ALLOC_SIZE], [ ])
+fi
+dnl Check for format(gnu_printf, ...) attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([format(gnu_printf, ...) attribute], [#include <stdlib.h>],
+              [void *foo(const char *format, ...) __attribute__((format(gnu_printf, 1, 2)));],
+              [je_cv_format_gnu_printf])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_format_gnu_printf}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF], [ ])
+fi
+dnl Check for format(printf, ...) attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([format(printf, ...) attribute], [#include <stdlib.h>],
+              [void *foo(const char *format, ...) __attribute__((format(printf, 1, 2)));],
+              [je_cv_format_printf])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_format_printf}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_FORMAT_PRINTF], [ ])
+fi
+
+dnl Check for format_arg(...) attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([format(printf, ...) attribute], [#include <stdlib.h>],
+              [const char * __attribute__((__format_arg__(1))) foo(const char *format);],
+              [je_cv_format_arg])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_format_arg}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_FORMAT_ARG], [ ])
+fi
+
+dnl Support optional additions to rpath.
+AC_ARG_WITH([rpath],
+  [AS_HELP_STRING([--with-rpath=<rpath>], [Colon-separated rpath (ELF systems only)])],
+if test "x$with_rpath" = "xno" ; then
+  RPATH_EXTRA=
+else
+  RPATH_EXTRA="`echo $with_rpath | tr \":\" \" \"`"
+fi,
+  RPATH_EXTRA=
+)
+AC_SUBST([RPATH_EXTRA])
+
+dnl Disable rules that do automatic regeneration of configure output by default.
+AC_ARG_ENABLE([autogen],
+  [AS_HELP_STRING([--enable-autogen], [Automatically regenerate configure output])],
+if test "x$enable_autogen" = "xno" ; then
+  enable_autogen="0"
+else
+  enable_autogen="1"
+fi
+,
+enable_autogen="0"
+)
+AC_SUBST([enable_autogen])
+
+AC_PROG_INSTALL
+AC_PROG_RANLIB
+AC_PATH_PROG([LD], [ld], [false], [$PATH])
+AC_PATH_PROG([AUTOCONF], [autoconf], [false], [$PATH])
+
+dnl Enable documentation
+AC_ARG_ENABLE([doc],
+	      [AS_HELP_STRING([--enable-documentation], [Build documentation])],
+if test "x$enable_doc" = "xno" ; then
+  enable_doc="0"
+else
+  enable_doc="1"
+fi
+,
+enable_doc="1"
+)
+AC_SUBST([enable_doc])
+
+dnl Enable shared libs
+AC_ARG_ENABLE([shared],
+  [AS_HELP_STRING([--enable-shared], [Build shared libaries])],
+if test "x$enable_shared" = "xno" ; then
+  enable_shared="0"
+else
+  enable_shared="1"
+fi
+,
+enable_shared="1"
+)
+AC_SUBST([enable_shared])
+
+dnl Enable static libs
+AC_ARG_ENABLE([static],
+  [AS_HELP_STRING([--enable-static], [Build static libaries])],
+if test "x$enable_static" = "xno" ; then
+  enable_static="0"
+else
+  enable_static="1"
+fi
+,
+enable_static="1"
+)
+AC_SUBST([enable_static])
+
+if test "$enable_shared$enable_static" = "00" ; then
+  AC_MSG_ERROR([Please enable one of shared or static builds])
+fi
+
+dnl Perform no name mangling by default.
+AC_ARG_WITH([mangling],
+  [AS_HELP_STRING([--with-mangling=<map>], [Mangle symbols in <map>])],
+  [mangling_map="$with_mangling"], [mangling_map=""])
+
+dnl Do not prefix public APIs by default.
+AC_ARG_WITH([jemalloc_prefix],
+  [AS_HELP_STRING([--with-jemalloc-prefix=<prefix>], [Prefix to prepend to all public APIs])],
+  [JEMALLOC_PREFIX="$with_jemalloc_prefix"],
+  [if test "x$abi" != "xmacho" -a "x$abi" != "xpecoff"; then
+  JEMALLOC_PREFIX=""
+else
+  JEMALLOC_PREFIX="je_"
+fi]
+)
+if test "x$JEMALLOC_PREFIX" = "x" ; then
+  AC_DEFINE([JEMALLOC_IS_MALLOC])
+else
+  JEMALLOC_CPREFIX=`echo ${JEMALLOC_PREFIX} | tr "a-z" "A-Z"`
+  AC_DEFINE_UNQUOTED([JEMALLOC_PREFIX], ["$JEMALLOC_PREFIX"])
+  AC_DEFINE_UNQUOTED([JEMALLOC_CPREFIX], ["$JEMALLOC_CPREFIX"])
+fi
+AC_SUBST([JEMALLOC_PREFIX])
+AC_SUBST([JEMALLOC_CPREFIX])
+
+AC_ARG_WITH([export],
+  [AS_HELP_STRING([--without-export], [disable exporting jemalloc public APIs])],
+  [if test "x$with_export" = "xno"; then
+  AC_DEFINE([JEMALLOC_EXPORT],[])
+fi]
+)
+
+public_syms="aligned_alloc calloc dallocx free mallctl mallctlbymib mallctlnametomib malloc malloc_conf malloc_message malloc_stats_print malloc_usable_size mallocx smallocx_${jemalloc_version_gid} nallocx posix_memalign rallocx realloc sallocx sdallocx xallocx"
+dnl Check for additional platform-specific public API functions.
+AC_CHECK_FUNC([memalign],
+	      [AC_DEFINE([JEMALLOC_OVERRIDE_MEMALIGN], [ ])
+	       public_syms="${public_syms} memalign"])
+AC_CHECK_FUNC([valloc],
+	      [AC_DEFINE([JEMALLOC_OVERRIDE_VALLOC], [ ])
+	       public_syms="${public_syms} valloc"])
+
+dnl Check for allocator-related functions that should be wrapped.
+wrap_syms=
+if test "x${JEMALLOC_PREFIX}" = "x" ; then
+  AC_CHECK_FUNC([__libc_calloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_CALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_calloc"])
+  AC_CHECK_FUNC([__libc_free],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_FREE], [ ])
+		 wrap_syms="${wrap_syms} __libc_free"])
+  AC_CHECK_FUNC([__libc_malloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_MALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_malloc"])
+  AC_CHECK_FUNC([__libc_memalign],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_MEMALIGN], [ ])
+		 wrap_syms="${wrap_syms} __libc_memalign"])
+  AC_CHECK_FUNC([__libc_realloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_REALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_realloc"])
+  AC_CHECK_FUNC([__libc_valloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_VALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_valloc"])
+  AC_CHECK_FUNC([__posix_memalign],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___POSIX_MEMALIGN], [ ])
+		 wrap_syms="${wrap_syms} __posix_memalign"])
+fi
+
+case "${host}" in
+  *-*-mingw* | *-*-cygwin*)
+    wrap_syms="${wrap_syms} tls_callback"
+    ;;
+  *)
+    ;;
+esac
+
+dnl Mangle library-private APIs.
+AC_ARG_WITH([private_namespace],
+  [AS_HELP_STRING([--with-private-namespace=<prefix>], [Prefix to prepend to all library-private APIs])],
+  [JEMALLOC_PRIVATE_NAMESPACE="${with_private_namespace}je_"],
+  [JEMALLOC_PRIVATE_NAMESPACE="je_"]
+)
+AC_DEFINE_UNQUOTED([JEMALLOC_PRIVATE_NAMESPACE], [$JEMALLOC_PRIVATE_NAMESPACE])
+private_namespace="$JEMALLOC_PRIVATE_NAMESPACE"
+AC_SUBST([private_namespace])
+
+dnl Do not add suffix to installed files by default.
+AC_ARG_WITH([install_suffix],
+  [AS_HELP_STRING([--with-install-suffix=<suffix>], [Suffix to append to all installed files])],
+  [INSTALL_SUFFIX="$with_install_suffix"],
+  [INSTALL_SUFFIX=]
+)
+install_suffix="$INSTALL_SUFFIX"
+AC_SUBST([install_suffix])
+
+dnl Specify default malloc_conf.
+AC_ARG_WITH([malloc_conf],
+  [AS_HELP_STRING([--with-malloc-conf=<malloc_conf>], [config.malloc_conf options string])],
+  [JEMALLOC_CONFIG_MALLOC_CONF="$with_malloc_conf"],
+  [JEMALLOC_CONFIG_MALLOC_CONF=""]
+)
+config_malloc_conf="$JEMALLOC_CONFIG_MALLOC_CONF"
+AC_DEFINE_UNQUOTED([JEMALLOC_CONFIG_MALLOC_CONF], ["$config_malloc_conf"])
+
+dnl Substitute @je_@ in jemalloc_protos.h.in, primarily to make generation of
+dnl jemalloc_protos_jet.h easy.
+je_="je_"
+AC_SUBST([je_])
+
+cfgoutputs_in="Makefile.in"
+cfgoutputs_in="${cfgoutputs_in} jemalloc.pc.in"
+cfgoutputs_in="${cfgoutputs_in} doc/html.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} doc/manpages.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} doc/jemalloc.xml.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/jemalloc_macros.h.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/jemalloc_protos.h.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/jemalloc_typedefs.h.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/internal/jemalloc_preamble.h.in"
+cfgoutputs_in="${cfgoutputs_in} test/test.sh.in"
+cfgoutputs_in="${cfgoutputs_in} test/include/test/jemalloc_test.h.in"
+
+cfgoutputs_out="Makefile"
+cfgoutputs_out="${cfgoutputs_out} jemalloc.pc"
+cfgoutputs_out="${cfgoutputs_out} doc/html.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/manpages.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/jemalloc.xml"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc_macros.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc_protos.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc_typedefs.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/internal/jemalloc_preamble.h"
+cfgoutputs_out="${cfgoutputs_out} test/test.sh"
+cfgoutputs_out="${cfgoutputs_out} test/include/test/jemalloc_test.h"
+
+cfgoutputs_tup="Makefile"
+cfgoutputs_tup="${cfgoutputs_tup} jemalloc.pc:jemalloc.pc.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/html.xsl:doc/html.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/manpages.xsl:doc/manpages.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/jemalloc.xml:doc/jemalloc.xml.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc_macros.h:include/jemalloc/jemalloc_macros.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc_protos.h:include/jemalloc/jemalloc_protos.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc_typedefs.h:include/jemalloc/jemalloc_typedefs.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/internal/jemalloc_preamble.h"
+cfgoutputs_tup="${cfgoutputs_tup} test/test.sh:test/test.sh.in"
+cfgoutputs_tup="${cfgoutputs_tup} test/include/test/jemalloc_test.h:test/include/test/jemalloc_test.h.in"
+
+cfghdrs_in="include/jemalloc/jemalloc_defs.h.in"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/jemalloc_internal_defs.h.in"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/private_symbols.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/private_namespace.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/public_namespace.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/public_unnamespace.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/jemalloc_rename.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/jemalloc_mangle.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/jemalloc.sh"
+cfghdrs_in="${cfghdrs_in} test/include/test/jemalloc_test_defs.h.in"
+
+cfghdrs_out="include/jemalloc/jemalloc_defs.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc${install_suffix}.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/private_symbols.awk"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/private_symbols_jet.awk"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/public_symbols.txt"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/public_namespace.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/public_unnamespace.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_protos_jet.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_rename.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_mangle.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_mangle_jet.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/jemalloc_internal_defs.h"
+cfghdrs_out="${cfghdrs_out} test/include/test/jemalloc_test_defs.h"
+
+cfghdrs_tup="include/jemalloc/jemalloc_defs.h:include/jemalloc/jemalloc_defs.h.in"
+cfghdrs_tup="${cfghdrs_tup} include/jemalloc/internal/jemalloc_internal_defs.h:include/jemalloc/internal/jemalloc_internal_defs.h.in"
+cfghdrs_tup="${cfghdrs_tup} test/include/test/jemalloc_test_defs.h:test/include/test/jemalloc_test_defs.h.in"
+
+dnl ============================================================================
+dnl jemalloc build options.
+dnl
+
+dnl Do not compile with debugging by default.
+AC_ARG_ENABLE([debug],
+  [AS_HELP_STRING([--enable-debug],
+                  [Build debugging code])],
+[if test "x$enable_debug" = "xno" ; then
+  enable_debug="0"
+else
+  enable_debug="1"
+fi
+],
+[enable_debug="0"]
+)
+if test "x$enable_debug" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DEBUG], [ ])
+fi
+if test "x$enable_debug" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DEBUG], [ ])
+fi
+AC_SUBST([enable_debug])
+
+dnl Only optimize if not debugging.
+if test "x$enable_debug" = "x0" ; then
+  if test "x$GCC" = "xyes" ; then
+    JE_CFLAGS_ADD([-O3])
+    JE_CXXFLAGS_ADD([-O3])
+    JE_CFLAGS_ADD([-funroll-loops])
+  elif test "x$je_cv_msvc" = "xyes" ; then
+    JE_CFLAGS_ADD([-O2])
+    JE_CXXFLAGS_ADD([-O2])
+  else
+    JE_CFLAGS_ADD([-O])
+    JE_CXXFLAGS_ADD([-O])
+  fi
+fi
+
+dnl Enable statistics calculation by default.
+AC_ARG_ENABLE([stats],
+  [AS_HELP_STRING([--disable-stats],
+                  [Disable statistics calculation/reporting])],
+[if test "x$enable_stats" = "xno" ; then
+  enable_stats="0"
+else
+  enable_stats="1"
+fi
+],
+[enable_stats="1"]
+)
+if test "x$enable_stats" = "x1" ; then
+  AC_DEFINE([JEMALLOC_STATS], [ ])
+fi
+AC_SUBST([enable_stats])
+
+dnl Do not enable smallocx by default.
+AC_ARG_ENABLE([experimental_smallocx],
+  [AS_HELP_STRING([--enable-experimental-smallocx], [Enable experimental smallocx API])],
+[if test "x$enable_experimental_smallocx" = "xno" ; then
+enable_experimental_smallocx="0"
+else
+enable_experimental_smallocx="1"
+fi
+],
+[enable_experimental_smallocx="0"]
+)
+if test "x$enable_experimental_smallocx" = "x1" ; then
+  AC_DEFINE([JEMALLOC_EXPERIMENTAL_SMALLOCX_API])
+fi
+AC_SUBST([enable_experimental_smallocx])
+
+dnl Do not enable profiling by default.
+AC_ARG_ENABLE([prof],
+  [AS_HELP_STRING([--enable-prof], [Enable allocation profiling])],
+[if test "x$enable_prof" = "xno" ; then
+  enable_prof="0"
+else
+  enable_prof="1"
+fi
+],
+[enable_prof="0"]
+)
+if test "x$enable_prof" = "x1" ; then
+  backtrace_method=""
+else
+  backtrace_method="N/A"
+fi
+
+AC_ARG_ENABLE([prof-libunwind],
+  [AS_HELP_STRING([--enable-prof-libunwind], [Use libunwind for backtracing])],
+[if test "x$enable_prof_libunwind" = "xno" ; then
+  enable_prof_libunwind="0"
+else
+  enable_prof_libunwind="1"
+fi
+],
+[enable_prof_libunwind="0"]
+)
+AC_ARG_WITH([static_libunwind],
+  [AS_HELP_STRING([--with-static-libunwind=<libunwind.a>],
+  [Path to static libunwind library; use rather than dynamically linking])],
+if test "x$with_static_libunwind" = "xno" ; then
+  LUNWIND="-lunwind"
+else
+  if test ! -f "$with_static_libunwind" ; then
+    AC_MSG_ERROR([Static libunwind not found: $with_static_libunwind])
+  fi
+  LUNWIND="$with_static_libunwind"
+fi,
+  LUNWIND="-lunwind"
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libunwind" = "x1" ; then
+  AC_CHECK_HEADERS([libunwind.h], , [enable_prof_libunwind="0"])
+  if test "x$LUNWIND" = "x-lunwind" ; then
+    AC_CHECK_LIB([unwind], [unw_backtrace], [JE_APPEND_VS(LIBS, $LUNWIND)],
+                 [enable_prof_libunwind="0"])
+  else
+    JE_APPEND_VS(LIBS, $LUNWIND)
+  fi
+  if test "x${enable_prof_libunwind}" = "x1" ; then
+    backtrace_method="libunwind"
+    AC_DEFINE([JEMALLOC_PROF_LIBUNWIND], [ ])
+  fi
+fi
+
+AC_ARG_ENABLE([prof-libgcc],
+  [AS_HELP_STRING([--disable-prof-libgcc],
+  [Do not use libgcc for backtracing])],
+[if test "x$enable_prof_libgcc" = "xno" ; then
+  enable_prof_libgcc="0"
+else
+  enable_prof_libgcc="1"
+fi
+],
+[enable_prof_libgcc="1"]
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libgcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  AC_CHECK_HEADERS([unwind.h], , [enable_prof_libgcc="0"])
+  if test "x${enable_prof_libgcc}" = "x1" ; then
+    AC_CHECK_LIB([gcc], [_Unwind_Backtrace], [JE_APPEND_VS(LIBS, -lgcc)], [enable_prof_libgcc="0"])
+  fi
+  if test "x${enable_prof_libgcc}" = "x1" ; then
+    backtrace_method="libgcc"
+    AC_DEFINE([JEMALLOC_PROF_LIBGCC], [ ])
+  fi
+else
+  enable_prof_libgcc="0"
+fi
+
+AC_ARG_ENABLE([prof-gcc],
+  [AS_HELP_STRING([--disable-prof-gcc],
+  [Do not use gcc intrinsics for backtracing])],
+[if test "x$enable_prof_gcc" = "xno" ; then
+  enable_prof_gcc="0"
+else
+  enable_prof_gcc="1"
+fi
+],
+[enable_prof_gcc="1"]
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_gcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  JE_CFLAGS_ADD([-fno-omit-frame-pointer])
+  backtrace_method="gcc intrinsics"
+  AC_DEFINE([JEMALLOC_PROF_GCC], [ ])
+else
+  enable_prof_gcc="0"
+fi
+
+if test "x$backtrace_method" = "x" ; then
+  backtrace_method="none (disabling profiling)"
+  enable_prof="0"
+fi
+AC_MSG_CHECKING([configured backtracing method])
+AC_MSG_RESULT([$backtrace_method])
+if test "x$enable_prof" = "x1" ; then
+  dnl Heap profiling uses the log(3) function.
+  JE_APPEND_VS(LIBS, $LM)
+
+  AC_DEFINE([JEMALLOC_PROF], [ ])
+fi
+AC_SUBST([enable_prof])
+
+dnl Indicate whether adjacent virtual memory mappings automatically coalesce
+dnl (and fragment on demand).
+if test "x${maps_coalesce}" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MAPS_COALESCE], [ ])
+fi
+
+dnl Indicate whether to retain memory (rather than using munmap()) by default.
+if test "x$default_retain" = "x1" ; then
+  AC_DEFINE([JEMALLOC_RETAIN], [ ])
+fi
+
+dnl Enable allocation from DSS if supported by the OS.
+have_dss="1"
+dnl Check whether the BSD/SUSv1 sbrk() exists.  If not, disable DSS support.
+AC_CHECK_FUNC([sbrk], [have_sbrk="1"], [have_sbrk="0"])
+if test "x$have_sbrk" = "x1" ; then
+  if test "x$sbrk_deprecated" = "x1" ; then
+    AC_MSG_RESULT([Disabling dss allocation because sbrk is deprecated])
+    have_dss="0"
+  fi
+else
+  have_dss="0"
+fi
+
+if test "x$have_dss" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DSS], [ ])
+fi
+
+dnl Support the junk/zero filling option by default.
+AC_ARG_ENABLE([fill],
+  [AS_HELP_STRING([--disable-fill], [Disable support for junk/zero filling])],
+[if test "x$enable_fill" = "xno" ; then
+  enable_fill="0"
+else
+  enable_fill="1"
+fi
+],
+[enable_fill="1"]
+)
+if test "x$enable_fill" = "x1" ; then
+  AC_DEFINE([JEMALLOC_FILL], [ ])
+fi
+AC_SUBST([enable_fill])
+
+dnl Disable utrace(2)-based tracing by default.
+AC_ARG_ENABLE([utrace],
+  [AS_HELP_STRING([--enable-utrace], [Enable utrace(2)-based tracing])],
+[if test "x$enable_utrace" = "xno" ; then
+  enable_utrace="0"
+else
+  enable_utrace="1"
+fi
+],
+[enable_utrace="0"]
+)
+JE_COMPILABLE([utrace(2)], [
+#include <sys/types.h>
+#include <sys/param.h>
+#include <sys/time.h>
+#include <sys/uio.h>
+#include <sys/ktrace.h>
+], [
+	utrace((void *)0, 0);
+], [je_cv_utrace])
+if test "x${je_cv_utrace}" = "xno" ; then
+  enable_utrace="0"
+fi
+if test "x$enable_utrace" = "x1" ; then
+  AC_DEFINE([JEMALLOC_UTRACE], [ ])
+fi
+AC_SUBST([enable_utrace])
+
+dnl Do not support the xmalloc option by default.
+AC_ARG_ENABLE([xmalloc],
+  [AS_HELP_STRING([--enable-xmalloc], [Support xmalloc option])],
+[if test "x$enable_xmalloc" = "xno" ; then
+  enable_xmalloc="0"
+else
+  enable_xmalloc="1"
+fi
+],
+[enable_xmalloc="0"]
+)
+if test "x$enable_xmalloc" = "x1" ; then
+  AC_DEFINE([JEMALLOC_XMALLOC], [ ])
+fi
+AC_SUBST([enable_xmalloc])
+
+dnl Support cache-oblivious allocation alignment by default.
+AC_ARG_ENABLE([cache-oblivious],
+  [AS_HELP_STRING([--disable-cache-oblivious],
+                  [Disable support for cache-oblivious allocation alignment])],
+[if test "x$enable_cache_oblivious" = "xno" ; then
+  enable_cache_oblivious="0"
+else
+  enable_cache_oblivious="1"
+fi
+],
+[enable_cache_oblivious="1"]
+)
+if test "x$enable_cache_oblivious" = "x1" ; then
+  AC_DEFINE([JEMALLOC_CACHE_OBLIVIOUS], [ ])
+fi
+AC_SUBST([enable_cache_oblivious])
+
+dnl Do not log by default.
+AC_ARG_ENABLE([log],
+  [AS_HELP_STRING([--enable-log], [Support debug logging])],
+[if test "x$enable_log" = "xno" ; then
+  enable_log="0"
+else
+  enable_log="1"
+fi
+],
+[enable_log="0"]
+)
+if test "x$enable_log" = "x1" ; then
+  AC_DEFINE([JEMALLOC_LOG], [ ])
+fi
+AC_SUBST([enable_log])
+
+dnl Do not use readlinkat by default
+AC_ARG_ENABLE([readlinkat],
+  [AS_HELP_STRING([--enable-readlinkat], [Use readlinkat over readlink])],
+[if test "x$enable_readlinkat" = "xno" ; then
+  enable_readlinkat="0"
+else
+  enable_readlinkat="1"
+fi
+],
+[enable_readlinkat="0"]
+)
+if test "x$enable_readlinkat" = "x1" ; then
+  AC_DEFINE([JEMALLOC_READLINKAT], [ ])
+fi
+AC_SUBST([enable_readlinkat])
+
+dnl Avoid extra safety checks by default
+AC_ARG_ENABLE([opt-safety-checks],
+  [AS_HELP_STRING([--enable-opt-safety-checks],
+  [Perform certain low-overhead checks, even in opt mode])],
+[if test "x$enable_opt_safety_checks" = "xno" ; then
+  enable_opt_safety_checks="0"
+else
+  enable_opt_safety_checks="1"
+fi
+],
+[enable_opt_safety_checks="0"]
+)
+if test "x$enable_opt_safety_checks" = "x1" ; then
+  AC_DEFINE([JEMALLOC_OPT_SAFETY_CHECKS], [ ])
+fi
+AC_SUBST([enable_opt_safety_checks])
+
+JE_COMPILABLE([a program using __builtin_unreachable], [
+void foo (void) {
+  __builtin_unreachable();
+}
+], [
+	{
+		foo();
+	}
+], [je_cv_gcc_builtin_unreachable])
+if test "x${je_cv_gcc_builtin_unreachable}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_INTERNAL_UNREACHABLE], [__builtin_unreachable])
+else
+  AC_DEFINE([JEMALLOC_INTERNAL_UNREACHABLE], [abort])
+fi
+
+dnl ============================================================================
+dnl Check for  __builtin_ffsl(), then ffsl(3), and fail if neither are found.
+dnl One of those two functions should (theoretically) exist on all platforms
+dnl that jemalloc currently has a chance of functioning on without modification.
+dnl We additionally assume ffs[ll]() or __builtin_ffs[ll]() are defined if
+dnl ffsl() or __builtin_ffsl() are defined, respectively.
+JE_COMPILABLE([a program using __builtin_ffsl], [
+#include <stdio.h>
+#include <strings.h>
+#include <string.h>
+], [
+	{
+		int rv = __builtin_ffsl(0x08);
+		printf("%d\n", rv);
+	}
+], [je_cv_gcc_builtin_ffsl])
+if test "x${je_cv_gcc_builtin_ffsl}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_INTERNAL_FFSLL], [__builtin_ffsll])
+  AC_DEFINE([JEMALLOC_INTERNAL_FFSL], [__builtin_ffsl])
+  AC_DEFINE([JEMALLOC_INTERNAL_FFS], [__builtin_ffs])
+else
+  JE_COMPILABLE([a program using ffsl], [
+  #include <stdio.h>
+  #include <strings.h>
+  #include <string.h>
+  ], [
+	{
+		int rv = ffsl(0x08);
+		printf("%d\n", rv);
+	}
+  ], [je_cv_function_ffsl])
+  if test "x${je_cv_function_ffsl}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_INTERNAL_FFSLL], [ffsll])
+    AC_DEFINE([JEMALLOC_INTERNAL_FFSL], [ffsl])
+    AC_DEFINE([JEMALLOC_INTERNAL_FFS], [ffs])
+  else
+    AC_MSG_ERROR([Cannot build without ffsl(3) or __builtin_ffsl()])
+  fi
+fi
+
+JE_COMPILABLE([a program using __builtin_popcountl], [
+#include <stdio.h>
+#include <strings.h>
+#include <string.h>
+], [
+	{
+		int rv = __builtin_popcountl(0x08);
+		printf("%d\n", rv);
+	}
+], [je_cv_gcc_builtin_popcountl])
+if test "x${je_cv_gcc_builtin_popcountl}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_INTERNAL_POPCOUNT], [__builtin_popcount])
+  AC_DEFINE([JEMALLOC_INTERNAL_POPCOUNTL], [__builtin_popcountl])
+fi
+
+AC_ARG_WITH([lg_quantum],
+  [AS_HELP_STRING([--with-lg-quantum=<lg-quantum>],
+   [Base 2 log of minimum allocation alignment])],
+  [LG_QUANTA="$with_lg_quantum"],
+  [LG_QUANTA="3 4"])
+if test "x$with_lg_quantum" != "x" ; then
+  AC_DEFINE_UNQUOTED([LG_QUANTUM], [$with_lg_quantum])
+fi
+
+AC_ARG_WITH([lg_page],
+  [AS_HELP_STRING([--with-lg-page=<lg-page>], [Base 2 log of system page size])],
+  [LG_PAGE="$with_lg_page"], [LG_PAGE="detect"])
+if test "x$LG_PAGE" = "xdetect"; then
+  AC_CACHE_CHECK([LG_PAGE],
+               [je_cv_lg_page],
+               AC_RUN_IFELSE([AC_LANG_PROGRAM(
+[[
+#include <strings.h>
+#ifdef _WIN32
+#include <windows.h>
+#else
+#include <unistd.h>
+#endif
+#include <stdio.h>
+]],
+[[
+    int result;
+    FILE *f;
+
+#ifdef _WIN32
+    SYSTEM_INFO si;
+    GetSystemInfo(&si);
+    result = si.dwPageSize;
+#else
+    result = sysconf(_SC_PAGESIZE);
+#endif
+    if (result == -1) {
+	return 1;
+    }
+    result = JEMALLOC_INTERNAL_FFSL(result) - 1;
+
+    f = fopen("conftest.out", "w");
+    if (f == NULL) {
+	return 1;
+    }
+    fprintf(f, "%d", result);
+    fclose(f);
+
+    return 0;
+]])],
+                             [je_cv_lg_page=`cat conftest.out`],
+                             [je_cv_lg_page=undefined],
+                             [je_cv_lg_page=12]))
+fi
+if test "x${je_cv_lg_page}" != "x" ; then
+  LG_PAGE="${je_cv_lg_page}"
+fi
+if test "x${LG_PAGE}" != "xundefined" ; then
+   AC_DEFINE_UNQUOTED([LG_PAGE], [$LG_PAGE])
+else
+   AC_MSG_ERROR([cannot determine value for LG_PAGE])
+fi
+
+AC_ARG_WITH([lg_hugepage],
+  [AS_HELP_STRING([--with-lg-hugepage=<lg-hugepage>],
+   [Base 2 log of system huge page size])],
+  [je_cv_lg_hugepage="${with_lg_hugepage}"],
+  [je_cv_lg_hugepage=""])
+if test "x${je_cv_lg_hugepage}" = "x" ; then
+  dnl Look in /proc/meminfo (Linux-specific) for information on the default huge
+  dnl page size, if any.  The relevant line looks like:
+  dnl
+  dnl   Hugepagesize:       2048 kB
+  if test -e "/proc/meminfo" ; then
+    hpsk=[`cat /proc/meminfo 2>/dev/null | \
+          grep -e '^Hugepagesize:[[:space:]]\+[0-9]\+[[:space:]]kB$' | \
+          awk '{print $2}'`]
+    if test "x${hpsk}" != "x" ; then
+      je_cv_lg_hugepage=10
+      while test "${hpsk}" -gt 1 ; do
+        hpsk="$((hpsk / 2))"
+        je_cv_lg_hugepage="$((je_cv_lg_hugepage + 1))"
+      done
+    fi
+  fi
+
+  dnl Set default if unable to automatically configure.
+  if test "x${je_cv_lg_hugepage}" = "x" ; then
+    je_cv_lg_hugepage=21
+  fi
+fi
+if test "x${LG_PAGE}" != "xundefined" -a \
+        "${je_cv_lg_hugepage}" -lt "${LG_PAGE}" ; then
+  AC_MSG_ERROR([Huge page size (2^${je_cv_lg_hugepage}) must be at least page size (2^${LG_PAGE})])
+fi
+AC_DEFINE_UNQUOTED([LG_HUGEPAGE], [${je_cv_lg_hugepage}])
+
+dnl ============================================================================
+dnl Enable libdl by default.
+AC_ARG_ENABLE([libdl],
+  [AS_HELP_STRING([--disable-libdl],
+  [Do not use libdl])],
+[if test "x$enable_libdl" = "xno" ; then
+  enable_libdl="0"
+else
+  enable_libdl="1"
+fi
+],
+[enable_libdl="1"]
+)
+AC_SUBST([libdl])
+
+dnl ============================================================================
+dnl Configure pthreads.
+
+if test "x$abi" != "xpecoff" ; then
+  AC_DEFINE([JEMALLOC_HAVE_PTHREAD], [ ])
+  AC_CHECK_HEADERS([pthread.h], , [AC_MSG_ERROR([pthread.h is missing])])
+  dnl Some systems may embed pthreads functionality in libc; check for libpthread
+  dnl first, but try libc too before failing.
+  AC_CHECK_LIB([pthread], [pthread_create], [JE_APPEND_VS(LIBS, -pthread)],
+               [AC_SEARCH_LIBS([pthread_create], , ,
+                               AC_MSG_ERROR([libpthread is missing]))])
+  wrap_syms="${wrap_syms} pthread_create"
+  have_pthread="1"
+
+dnl Check if we have dlsym support.
+  if test "x$enable_libdl" = "x1" ; then
+    have_dlsym="1"
+    AC_CHECK_HEADERS([dlfcn.h], , [have_dlsym="0"])
+    check_dlsym_in_libdl="0"
+    if test "x$have_dlsym" = "x1" ; then
+      AC_CHECK_FUNC([dlsym], [], [check_dlsym_in_libdl="1"])
+    fi
+    if test "x$check_dlsym_in_libdl" = "x1" ; then
+      AC_CHECK_LIB([dl], [dlsym], [LIBS="$LIBS -ldl"], [have_dlsym="0"])
+    fi
+    if test "x$have_dlsym" = "x1" ; then
+      AC_DEFINE([JEMALLOC_HAVE_DLSYM], [ ])
+    fi
+  else
+    have_dlsym="0"
+  fi
+
+  JE_COMPILABLE([pthread_atfork(3)], [
+#include <pthread.h>
+], [
+  pthread_atfork((void *)0, (void *)0, (void *)0);
+], [je_cv_pthread_atfork])
+  if test "x${je_cv_pthread_atfork}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_HAVE_PTHREAD_ATFORK], [ ])
+  fi
+  dnl Check if pthread_setname_np is available with the expected API.
+  JE_COMPILABLE([pthread_setname_np(3)], [
+#include <pthread.h>
+], [
+  pthread_setname_np(pthread_self(), "setname_test");
+], [je_cv_pthread_setname_np])
+  if test "x${je_cv_pthread_setname_np}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_HAVE_PTHREAD_SETNAME_NP], [ ])
+  fi
+fi
+
+JE_APPEND_VS(CPPFLAGS, -D_REENTRANT)
+
+dnl Check whether clock_gettime(2) is in libc or librt.
+AC_SEARCH_LIBS([clock_gettime], [rt])
+
+dnl Cray wrapper compiler often adds `-lrt` when using `-static`. Check with
+dnl `-dynamic` as well in case a user tries to dynamically link in jemalloc
+if test "x$je_cv_cray_prgenv_wrapper" = "xyes" ; then
+  if test "$ac_cv_search_clock_gettime" != "-lrt"; then
+    JE_CFLAGS_SAVE()
+
+    unset ac_cv_search_clock_gettime
+    JE_CFLAGS_ADD([-dynamic])
+    AC_SEARCH_LIBS([clock_gettime], [rt])
+
+    JE_CFLAGS_RESTORE()
+  fi
+fi
+
+dnl check for CLOCK_MONOTONIC_COARSE (Linux-specific).
+JE_COMPILABLE([clock_gettime(CLOCK_MONOTONIC_COARSE, ...)], [
+#include <time.h>
+], [
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC_COARSE, &ts);
+], [je_cv_clock_monotonic_coarse])
+if test "x${je_cv_clock_monotonic_coarse}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_CLOCK_MONOTONIC_COARSE])
+fi
+
+dnl check for CLOCK_MONOTONIC.
+JE_COMPILABLE([clock_gettime(CLOCK_MONOTONIC, ...)], [
+#include <unistd.h>
+#include <time.h>
+], [
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts);
+#if !defined(_POSIX_MONOTONIC_CLOCK) || _POSIX_MONOTONIC_CLOCK < 0
+#  error _POSIX_MONOTONIC_CLOCK missing/invalid
+#endif
+], [je_cv_clock_monotonic])
+if test "x${je_cv_clock_monotonic}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_CLOCK_MONOTONIC])
+fi
+
+dnl Check for mach_absolute_time().
+JE_COMPILABLE([mach_absolute_time()], [
+#include <mach/mach_time.h>
+], [
+	mach_absolute_time();
+], [je_cv_mach_absolute_time])
+if test "x${je_cv_mach_absolute_time}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_MACH_ABSOLUTE_TIME])
+fi
+
+dnl Use syscall(2) (if available) by default.
+AC_ARG_ENABLE([syscall],
+  [AS_HELP_STRING([--disable-syscall], [Disable use of syscall(2)])],
+[if test "x$enable_syscall" = "xno" ; then
+  enable_syscall="0"
+else
+  enable_syscall="1"
+fi
+],
+[enable_syscall="1"]
+)
+if test "x$enable_syscall" = "x1" ; then
+  dnl Check if syscall(2) is usable.  Treat warnings as errors, so that e.g. OS
+  dnl X 10.12's deprecation warning prevents use.
+  JE_CFLAGS_SAVE()
+  JE_CFLAGS_ADD([-Werror])
+  JE_COMPILABLE([syscall(2)], [
+#include <sys/syscall.h>
+#include <unistd.h>
+], [
+	syscall(SYS_write, 2, "hello", 5);
+],
+                [je_cv_syscall])
+  JE_CFLAGS_RESTORE()
+  if test "x$je_cv_syscall" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_USE_SYSCALL], [ ])
+  fi
+fi
+
+dnl Check if the GNU-specific secure_getenv function exists.
+AC_CHECK_FUNC([secure_getenv],
+              [have_secure_getenv="1"],
+              [have_secure_getenv="0"]
+             )
+if test "x$have_secure_getenv" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_SECURE_GETENV], [ ])
+fi
+
+dnl Check if the GNU-specific sched_getcpu function exists.
+AC_CHECK_FUNC([sched_getcpu],
+              [have_sched_getcpu="1"],
+              [have_sched_getcpu="0"]
+             )
+if test "x$have_sched_getcpu" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_SCHED_GETCPU], [ ])
+fi
+
+dnl Check if the GNU-specific sched_setaffinity function exists.
+AC_CHECK_FUNC([sched_setaffinity],
+              [have_sched_setaffinity="1"],
+              [have_sched_setaffinity="0"]
+             )
+if test "x$have_sched_setaffinity" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_SCHED_SETAFFINITY], [ ])
+fi
+
+dnl Check if the Solaris/BSD issetugid function exists.
+AC_CHECK_FUNC([issetugid],
+              [have_issetugid="1"],
+              [have_issetugid="0"]
+             )
+if test "x$have_issetugid" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ISSETUGID], [ ])
+fi
+
+dnl Check whether the BSD-specific _malloc_thread_cleanup() exists.  If so, use
+dnl it rather than pthreads TSD cleanup functions to support cleanup during
+dnl thread exit, in order to avoid pthreads library recursion during
+dnl bootstrapping.
+AC_CHECK_FUNC([_malloc_thread_cleanup],
+              [have__malloc_thread_cleanup="1"],
+              [have__malloc_thread_cleanup="0"]
+             )
+if test "x$have__malloc_thread_cleanup" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MALLOC_THREAD_CLEANUP], [ ])
+  wrap_syms="${wrap_syms} _malloc_thread_cleanup"
+  force_tls="1"
+fi
+
+dnl Check whether the BSD-specific _pthread_mutex_init_calloc_cb() exists.  If
+dnl so, mutex initialization causes allocation, and we need to implement this
+dnl callback function in order to prevent recursive allocation.
+AC_CHECK_FUNC([_pthread_mutex_init_calloc_cb],
+              [have__pthread_mutex_init_calloc_cb="1"],
+              [have__pthread_mutex_init_calloc_cb="0"]
+             )
+if test "x$have__pthread_mutex_init_calloc_cb" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MUTEX_INIT_CB])
+  wrap_syms="${wrap_syms} _malloc_prefork _malloc_postfork"
+fi
+
+dnl Disable lazy locking by default.
+AC_ARG_ENABLE([lazy_lock],
+  [AS_HELP_STRING([--enable-lazy-lock],
+  [Enable lazy locking (only lock when multi-threaded)])],
+[if test "x$enable_lazy_lock" = "xno" ; then
+  enable_lazy_lock="0"
+else
+  enable_lazy_lock="1"
+fi
+],
+[enable_lazy_lock=""]
+)
+if test "x${enable_lazy_lock}" = "x" ; then
+  if test "x${force_lazy_lock}" = "x1" ; then
+    AC_MSG_RESULT([Forcing lazy-lock to avoid allocator/threading bootstrap issues])
+    enable_lazy_lock="1"
+  else
+    enable_lazy_lock="0"
+  fi
+fi
+if test "x${enable_lazy_lock}" = "x1" -a "x${abi}" = "xpecoff" ; then
+  AC_MSG_RESULT([Forcing no lazy-lock because thread creation monitoring is unimplemented])
+  enable_lazy_lock="0"
+fi
+if test "x$enable_lazy_lock" = "x1" ; then
+  if test "x$have_dlsym" = "x1" ; then
+    AC_DEFINE([JEMALLOC_LAZY_LOCK], [ ])
+  else
+    AC_MSG_ERROR([Missing dlsym support: lazy-lock cannot be enabled.])
+  fi
+fi
+AC_SUBST([enable_lazy_lock])
+
+dnl Automatically configure TLS.
+if test "x${force_tls}" = "x1" ; then
+  enable_tls="1"
+elif test "x${force_tls}" = "x0" ; then
+  enable_tls="0"
+else
+  enable_tls="1"
+fi
+if test "x${enable_tls}" = "x1" ; then
+AC_MSG_CHECKING([for TLS])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+    __thread int x;
+]], [[
+    x = 42;
+
+    return 0;
+]])],
+              AC_MSG_RESULT([yes]),
+              AC_MSG_RESULT([no])
+              enable_tls="0")
+else
+  enable_tls="0"
+fi
+AC_SUBST([enable_tls])
+if test "x${enable_tls}" = "x1" ; then
+  AC_DEFINE_UNQUOTED([JEMALLOC_TLS], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for C11 atomics.
+
+JE_COMPILABLE([C11 atomics], [
+#include <stdint.h>
+#if (__STDC_VERSION__ >= 201112L) && !defined(__STDC_NO_ATOMICS__)
+#include <stdatomic.h>
+#else
+#error Atomics not available
+#endif
+], [
+    uint64_t *p = (uint64_t *)0;
+    uint64_t x = 1;
+    volatile atomic_uint_least64_t *a = (volatile atomic_uint_least64_t *)p;
+    uint64_t r = atomic_fetch_add(a, x) + x;
+    return r == 0;
+], [je_cv_c11_atomics])
+if test "x${je_cv_c11_atomics}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_C11_ATOMICS])
+fi
+
+dnl ============================================================================
+dnl Check for GCC-style __atomic atomics.
+
+JE_COMPILABLE([GCC __atomic atomics], [
+], [
+    int x = 0;
+    int val = 1;
+    int y = __atomic_fetch_add(&x, val, __ATOMIC_RELAXED);
+    int after_add = x;
+    return after_add == 1;
+], [je_cv_gcc_atomic_atomics])
+if test "x${je_cv_gcc_atomic_atomics}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_GCC_ATOMIC_ATOMICS])
+
+  dnl check for 8-bit atomic support
+  JE_COMPILABLE([GCC 8-bit __atomic atomics], [
+  ], [
+      unsigned char x = 0;
+      int val = 1;
+      int y = __atomic_fetch_add(&x, val, __ATOMIC_RELAXED);
+      int after_add = (int)x;
+      return after_add == 1;
+  ], [je_cv_gcc_u8_atomic_atomics])
+  if test "x${je_cv_gcc_u8_atomic_atomics}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_GCC_U8_ATOMIC_ATOMICS])
+  fi
+fi
+
+dnl ============================================================================
+dnl Check for GCC-style __sync atomics.
+
+JE_COMPILABLE([GCC __sync atomics], [
+], [
+    int x = 0;
+    int before_add = __sync_fetch_and_add(&x, 1);
+    int after_add = x;
+    return (before_add == 0) && (after_add == 1);
+], [je_cv_gcc_sync_atomics])
+if test "x${je_cv_gcc_sync_atomics}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_GCC_SYNC_ATOMICS])
+
+  dnl check for 8-bit atomic support
+  JE_COMPILABLE([GCC 8-bit __sync atomics], [
+  ], [
+      unsigned char x = 0;
+      int before_add = __sync_fetch_and_add(&x, 1);
+      int after_add = (int)x;
+      return (before_add == 0) && (after_add == 1);
+  ], [je_cv_gcc_u8_sync_atomics])
+  if test "x${je_cv_gcc_u8_sync_atomics}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_GCC_U8_SYNC_ATOMICS])
+  fi
+fi
+
+dnl ============================================================================
+dnl Check for atomic(3) operations as provided on Darwin.
+dnl We need this not for the atomic operations (which are provided above), but
+dnl rather for the OS_unfair_lock type it exposes.
+
+JE_COMPILABLE([Darwin OSAtomic*()], [
+#include <libkern/OSAtomic.h>
+#include <inttypes.h>
+], [
+	{
+		int32_t x32 = 0;
+		volatile int32_t *x32p = &x32;
+		OSAtomicAdd32(1, x32p);
+	}
+	{
+		int64_t x64 = 0;
+		volatile int64_t *x64p = &x64;
+		OSAtomicAdd64(1, x64p);
+	}
+], [je_cv_osatomic])
+if test "x${je_cv_osatomic}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_OSATOMIC], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for madvise(2).
+
+JE_COMPILABLE([madvise(2)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, 0);
+], [je_cv_madvise])
+if test "x${je_cv_madvise}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_MADVISE], [ ])
+
+  dnl Check for madvise(..., MADV_FREE).
+  JE_COMPILABLE([madvise(..., MADV_FREE)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_FREE);
+], [je_cv_madv_free])
+  if test "x${je_cv_madv_free}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
+  elif test "x${je_cv_madvise}" = "xyes" ; then
+    case "${host_cpu}" in i686|x86_64)
+        case "${host}" in *-*-linux*)
+            AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
+            AC_DEFINE([JEMALLOC_DEFINE_MADVISE_FREE], [ ])
+	    ;;
+        esac
+        ;;
+    esac
+  fi
+
+  dnl Check for madvise(..., MADV_DONTNEED).
+  JE_COMPILABLE([madvise(..., MADV_DONTNEED)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_DONTNEED);
+], [je_cv_madv_dontneed])
+  if test "x${je_cv_madv_dontneed}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_PURGE_MADVISE_DONTNEED], [ ])
+  fi
+
+  dnl Check for madvise(..., MADV_DO[NT]DUMP).
+  JE_COMPILABLE([madvise(..., MADV_DO[[NT]]DUMP)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_DONTDUMP);
+	madvise((void *)0, 0, MADV_DODUMP);
+], [je_cv_madv_dontdump])
+  if test "x${je_cv_madv_dontdump}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_MADVISE_DONTDUMP], [ ])
+  fi
+
+  dnl Check for madvise(..., MADV_[NO]HUGEPAGE).
+  JE_COMPILABLE([madvise(..., MADV_[[NO]]HUGEPAGE)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_HUGEPAGE);
+	madvise((void *)0, 0, MADV_NOHUGEPAGE);
+], [je_cv_thp])
+case "${host_cpu}" in
+  arm*)
+    ;;
+  *)
+  if test "x${je_cv_thp}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_HAVE_MADVISE_HUGE], [ ])
+  fi
+  ;;
+esac
+fi
+
+dnl ============================================================================
+dnl Check for __builtin_clz() and __builtin_clzl().
+
+AC_CACHE_CHECK([for __builtin_clz],
+               [je_cv_builtin_clz],
+               [AC_LINK_IFELSE([AC_LANG_PROGRAM([],
+                                                [
+                                                {
+                                                        unsigned x = 0;
+                                                        int y = __builtin_clz(x);
+                                                }
+                                                {
+                                                        unsigned long x = 0;
+                                                        int y = __builtin_clzl(x);
+                                                }
+                                                ])],
+                               [je_cv_builtin_clz=yes],
+                               [je_cv_builtin_clz=no])])
+
+if test "x${je_cv_builtin_clz}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_BUILTIN_CLZ], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for os_unfair_lock operations as provided on Darwin.
+
+JE_COMPILABLE([Darwin os_unfair_lock_*()], [
+#include <os/lock.h>
+#include <AvailabilityMacros.h>
+], [
+	#if MAC_OS_X_VERSION_MIN_REQUIRED < 101200
+	#error "os_unfair_lock is not supported"
+	#else
+	os_unfair_lock lock = OS_UNFAIR_LOCK_INIT;
+	os_unfair_lock_lock(&lock);
+	os_unfair_lock_unlock(&lock);
+	#endif
+], [je_cv_os_unfair_lock])
+if test "x${je_cv_os_unfair_lock}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_OS_UNFAIR_LOCK], [ ])
+fi
+
+dnl ============================================================================
+dnl Darwin-related configuration.
+
+AC_ARG_ENABLE([zone-allocator],
+  [AS_HELP_STRING([--disable-zone-allocator],
+                  [Disable zone allocator for Darwin])],
+[if test "x$enable_zone_allocator" = "xno" ; then
+  enable_zone_allocator="0"
+else
+  enable_zone_allocator="1"
+fi
+],
+[if test "x${abi}" = "xmacho"; then
+  enable_zone_allocator="1"
+fi
+]
+)
+AC_SUBST([enable_zone_allocator])
+
+if test "x${enable_zone_allocator}" = "x1" ; then
+  if test "x${abi}" != "xmacho"; then
+    AC_MSG_ERROR([--enable-zone-allocator is only supported on Darwin])
+  fi
+  AC_DEFINE([JEMALLOC_ZONE], [ ])
+fi
+
+dnl ============================================================================
+dnl Use initial-exec TLS by default.
+AC_ARG_ENABLE([initial-exec-tls],
+  [AS_HELP_STRING([--disable-initial-exec-tls],
+                  [Disable the initial-exec tls model])],
+[if test "x$enable_initial_exec_tls" = "xno" ; then
+  enable_initial_exec_tls="0"
+else
+  enable_initial_exec_tls="1"
+fi
+],
+[enable_initial_exec_tls="1"]
+)
+AC_SUBST([enable_initial_exec_tls])
+
+if test "x${je_cv_tls_model}" = "xyes" -a \
+       "x${enable_initial_exec_tls}" = "x1" ; then
+  AC_DEFINE([JEMALLOC_TLS_MODEL],
+            [__attribute__((tls_model("initial-exec")))])
+else
+  AC_DEFINE([JEMALLOC_TLS_MODEL], [ ])
+fi
+
+dnl ============================================================================
+dnl Enable background threads if possible.
+
+if test "x${have_pthread}" = "x1" -a "x${je_cv_os_unfair_lock}" != "xyes" ; then
+  AC_DEFINE([JEMALLOC_BACKGROUND_THREAD])
+fi
+
+dnl ============================================================================
+dnl Check for glibc malloc hooks
+
+JE_COMPILABLE([glibc malloc hook], [
+#include <stddef.h>
+
+extern void (* __free_hook)(void *ptr);
+extern void *(* __malloc_hook)(size_t size);
+extern void *(* __realloc_hook)(void *ptr, size_t size);
+], [
+  void *ptr = 0L;
+  if (__malloc_hook) ptr = __malloc_hook(1);
+  if (__realloc_hook) ptr = __realloc_hook(ptr, 2);
+  if (__free_hook && ptr) __free_hook(ptr);
+], [je_cv_glibc_malloc_hook])
+if test "x${je_cv_glibc_malloc_hook}" = "xyes" ; then
+  if test "x${JEMALLOC_PREFIX}" = "x" ; then
+    AC_DEFINE([JEMALLOC_GLIBC_MALLOC_HOOK], [ ])
+    wrap_syms="${wrap_syms} __free_hook __malloc_hook __realloc_hook"
+  fi
+fi
+
+JE_COMPILABLE([glibc memalign hook], [
+#include <stddef.h>
+
+extern void *(* __memalign_hook)(size_t alignment, size_t size);
+], [
+  void *ptr = 0L;
+  if (__memalign_hook) ptr = __memalign_hook(16, 7);
+], [je_cv_glibc_memalign_hook])
+if test "x${je_cv_glibc_memalign_hook}" = "xyes" ; then
+  if test "x${JEMALLOC_PREFIX}" = "x" ; then
+    AC_DEFINE([JEMALLOC_GLIBC_MEMALIGN_HOOK], [ ])
+    wrap_syms="${wrap_syms} __memalign_hook"
+  fi
+fi
+
+JE_COMPILABLE([pthreads adaptive mutexes], [
+#include <pthread.h>
+], [
+  pthread_mutexattr_t attr;
+  pthread_mutexattr_init(&attr);
+  pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_ADAPTIVE_NP);
+  pthread_mutexattr_destroy(&attr);
+], [je_cv_pthread_mutex_adaptive_np])
+if test "x${je_cv_pthread_mutex_adaptive_np}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_PTHREAD_MUTEX_ADAPTIVE_NP], [ ])
+fi
+
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-D_GNU_SOURCE])
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([strerror_r returns char with gnu source], [
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+], [
+  char *buffer = (char *) malloc(100);
+  char *error = strerror_r(EINVAL, buffer, 100);
+  printf("%s\n", error);
+], [je_cv_strerror_r_returns_char_with_gnu_source])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_strerror_r_returns_char_with_gnu_source}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for typedefs, structures, and compiler characteristics.
+AC_HEADER_STDBOOL
+
+dnl ============================================================================
+dnl Define commands that generate output files.
+
+AC_CONFIG_COMMANDS([include/jemalloc/internal/public_symbols.txt], [
+  f="${objroot}include/jemalloc/internal/public_symbols.txt"
+  mkdir -p "${objroot}include/jemalloc/internal"
+  cp /dev/null "${f}"
+  for nm in `echo ${mangling_map} |tr ',' ' '` ; do
+    n=`echo ${nm} |tr ':' ' ' |awk '{print $[]1}'`
+    m=`echo ${nm} |tr ':' ' ' |awk '{print $[]2}'`
+    echo "${n}:${m}" >> "${f}"
+    dnl Remove name from public_syms so that it isn't redefined later.
+    public_syms=`for sym in ${public_syms}; do echo "${sym}"; done |grep -v "^${n}\$" |tr '\n' ' '`
+  done
+  for sym in ${public_syms} ; do
+    n="${sym}"
+    m="${JEMALLOC_PREFIX}${sym}"
+    echo "${n}:${m}" >> "${f}"
+  done
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  mangling_map="${mangling_map}"
+  public_syms="${public_syms}"
+  JEMALLOC_PREFIX="${JEMALLOC_PREFIX}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/private_symbols.awk], [
+  f="${objroot}include/jemalloc/internal/private_symbols.awk"
+  mkdir -p "${objroot}include/jemalloc/internal"
+  export_syms=`for sym in ${public_syms}; do echo "${JEMALLOC_PREFIX}${sym}"; done; for sym in ${wrap_syms}; do echo "${sym}"; done;`
+  "${srcdir}/include/jemalloc/internal/private_symbols.sh" "${SYM_PREFIX}" ${export_syms} > "${objroot}include/jemalloc/internal/private_symbols.awk"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  public_syms="${public_syms}"
+  wrap_syms="${wrap_syms}"
+  SYM_PREFIX="${SYM_PREFIX}"
+  JEMALLOC_PREFIX="${JEMALLOC_PREFIX}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/private_symbols_jet.awk], [
+  f="${objroot}include/jemalloc/internal/private_symbols_jet.awk"
+  mkdir -p "${objroot}include/jemalloc/internal"
+  export_syms=`for sym in ${public_syms}; do echo "jet_${sym}"; done; for sym in ${wrap_syms}; do echo "${sym}"; done;`
+  "${srcdir}/include/jemalloc/internal/private_symbols.sh" "${SYM_PREFIX}" ${export_syms} > "${objroot}include/jemalloc/internal/private_symbols_jet.awk"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  public_syms="${public_syms}"
+  wrap_syms="${wrap_syms}"
+  SYM_PREFIX="${SYM_PREFIX}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/public_namespace.h], [
+  mkdir -p "${objroot}include/jemalloc/internal"
+  "${srcdir}/include/jemalloc/internal/public_namespace.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" > "${objroot}include/jemalloc/internal/public_namespace.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/public_unnamespace.h], [
+  mkdir -p "${objroot}include/jemalloc/internal"
+  "${srcdir}/include/jemalloc/internal/public_unnamespace.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" > "${objroot}include/jemalloc/internal/public_unnamespace.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_protos_jet.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  cat "${srcdir}/include/jemalloc/jemalloc_protos.h.in" | sed -e 's/@je_@/jet_/g' > "${objroot}include/jemalloc/jemalloc_protos_jet.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_rename.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc_rename.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" > "${objroot}include/jemalloc/jemalloc_rename.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_mangle.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc_mangle.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" je_ > "${objroot}include/jemalloc/jemalloc_mangle.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_mangle_jet.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc_mangle.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" jet_ > "${objroot}include/jemalloc/jemalloc_mangle_jet.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc.sh" "${objroot}" > "${objroot}include/jemalloc/jemalloc${install_suffix}.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  install_suffix="${install_suffix}"
+])
+
+dnl Process .in files.
+AC_SUBST([cfghdrs_in])
+AC_SUBST([cfghdrs_out])
+AC_CONFIG_HEADERS([$cfghdrs_tup])
+
+dnl ============================================================================
+dnl Generate outputs.
+
+AC_CONFIG_FILES([$cfgoutputs_tup config.stamp bin/jemalloc-config bin/jemalloc.sh bin/jeprof])
+AC_SUBST([cfgoutputs_in])
+AC_SUBST([cfgoutputs_out])
+AC_OUTPUT
+
+dnl ============================================================================
+dnl Print out the results of configuration.
+AC_MSG_RESULT([===============================================================================])
+AC_MSG_RESULT([jemalloc version   : ${jemalloc_version}])
+AC_MSG_RESULT([library revision   : ${rev}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([CONFIG             : ${CONFIG}])
+AC_MSG_RESULT([CC                 : ${CC}])
+AC_MSG_RESULT([CONFIGURE_CFLAGS   : ${CONFIGURE_CFLAGS}])
+AC_MSG_RESULT([SPECIFIED_CFLAGS   : ${SPECIFIED_CFLAGS}])
+AC_MSG_RESULT([EXTRA_CFLAGS       : ${EXTRA_CFLAGS}])
+AC_MSG_RESULT([CPPFLAGS           : ${CPPFLAGS}])
+AC_MSG_RESULT([CXX                : ${CXX}])
+AC_MSG_RESULT([CONFIGURE_CXXFLAGS : ${CONFIGURE_CXXFLAGS}])
+AC_MSG_RESULT([SPECIFIED_CXXFLAGS : ${SPECIFIED_CXXFLAGS}])
+AC_MSG_RESULT([EXTRA_CXXFLAGS     : ${EXTRA_CXXFLAGS}])
+AC_MSG_RESULT([LDFLAGS            : ${LDFLAGS}])
+AC_MSG_RESULT([EXTRA_LDFLAGS      : ${EXTRA_LDFLAGS}])
+AC_MSG_RESULT([DSO_LDFLAGS        : ${DSO_LDFLAGS}])
+AC_MSG_RESULT([LIBS               : ${LIBS}])
+AC_MSG_RESULT([RPATH_EXTRA        : ${RPATH_EXTRA}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([XSLTPROC           : ${XSLTPROC}])
+AC_MSG_RESULT([XSLROOT            : ${XSLROOT}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([PREFIX             : ${PREFIX}])
+AC_MSG_RESULT([BINDIR             : ${BINDIR}])
+AC_MSG_RESULT([DATADIR            : ${DATADIR}])
+AC_MSG_RESULT([INCLUDEDIR         : ${INCLUDEDIR}])
+AC_MSG_RESULT([LIBDIR             : ${LIBDIR}])
+AC_MSG_RESULT([MANDIR             : ${MANDIR}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([srcroot            : ${srcroot}])
+AC_MSG_RESULT([abs_srcroot        : ${abs_srcroot}])
+AC_MSG_RESULT([objroot            : ${objroot}])
+AC_MSG_RESULT([abs_objroot        : ${abs_objroot}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([JEMALLOC_PREFIX    : ${JEMALLOC_PREFIX}])
+AC_MSG_RESULT([JEMALLOC_PRIVATE_NAMESPACE])
+AC_MSG_RESULT([                   : ${JEMALLOC_PRIVATE_NAMESPACE}])
+AC_MSG_RESULT([install_suffix     : ${install_suffix}])
+AC_MSG_RESULT([malloc_conf        : ${config_malloc_conf}])
+AC_MSG_RESULT([documentation      : ${enable_doc}])
+AC_MSG_RESULT([shared libs        : ${enable_shared}])
+AC_MSG_RESULT([static libs        : ${enable_static}])
+AC_MSG_RESULT([autogen            : ${enable_autogen}])
+AC_MSG_RESULT([debug              : ${enable_debug}])
+AC_MSG_RESULT([stats              : ${enable_stats}])
+AC_MSG_RESULT([experimetal_smallocx : ${enable_experimental_smallocx}])
+AC_MSG_RESULT([prof               : ${enable_prof}])
+AC_MSG_RESULT([prof-libunwind     : ${enable_prof_libunwind}])
+AC_MSG_RESULT([prof-libgcc        : ${enable_prof_libgcc}])
+AC_MSG_RESULT([prof-gcc           : ${enable_prof_gcc}])
+AC_MSG_RESULT([fill               : ${enable_fill}])
+AC_MSG_RESULT([utrace             : ${enable_utrace}])
+AC_MSG_RESULT([xmalloc            : ${enable_xmalloc}])
+AC_MSG_RESULT([log                : ${enable_log}])
+AC_MSG_RESULT([lazy_lock          : ${enable_lazy_lock}])
+AC_MSG_RESULT([cache-oblivious    : ${enable_cache_oblivious}])
+AC_MSG_RESULT([cxx                : ${enable_cxx}])
+AC_MSG_RESULT([===============================================================================])
diff --git a/jemalloc/doc/html.xsl.in b/jemalloc/doc/html.xsl.in
new file mode 100644
index 0000000..ec4fa65
--- /dev/null
+++ b/jemalloc/doc/html.xsl.in
@@ -0,0 +1,5 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:import href="@XSLROOT@/html/docbook.xsl"/>
+  <xsl:import href="@abs_srcroot@doc/stylesheet.xsl"/>
+  <xsl:output method="xml" encoding="utf-8"/>
+</xsl:stylesheet>
diff --git a/jemalloc/doc/jemalloc.xml.in b/jemalloc/doc/jemalloc.xml.in
new file mode 100644
index 0000000..7fecda7
--- /dev/null
+++ b/jemalloc/doc/jemalloc.xml.in
@@ -0,0 +1,3513 @@
+<?xml version='1.0' encoding='UTF-8'?>
+<?xml-stylesheet type="text/xsl"
+        href="http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl"?>
+<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN"
+        "http://www.oasis-open.org/docbook/xml/4.4/docbookx.dtd" [
+]>
+
+<refentry>
+  <refentryinfo>
+    <title>User Manual</title>
+    <productname>jemalloc</productname>
+    <releaseinfo role="version">@jemalloc_version@</releaseinfo>
+    <authorgroup>
+      <author>
+        <firstname>Jason</firstname>
+        <surname>Evans</surname>
+        <personblurb>Author</personblurb>
+      </author>
+    </authorgroup>
+  </refentryinfo>
+  <refmeta>
+    <refentrytitle>JEMALLOC</refentrytitle>
+    <manvolnum>3</manvolnum>
+  </refmeta>
+  <refnamediv>
+    <refdescriptor>jemalloc</refdescriptor>
+    <refname>jemalloc</refname>
+    <!-- Each refname causes a man page file to be created.  Only if this were
+         the system malloc(3) implementation would these files be appropriate.
+    <refname>malloc</refname>
+    <refname>calloc</refname>
+    <refname>posix_memalign</refname>
+    <refname>aligned_alloc</refname>
+    <refname>realloc</refname>
+    <refname>free</refname>
+    <refname>mallocx</refname>
+    <refname>rallocx</refname>
+    <refname>xallocx</refname>
+    <refname>sallocx</refname>
+    <refname>dallocx</refname>
+    <refname>sdallocx</refname>
+    <refname>nallocx</refname>
+    <refname>mallctl</refname>
+    <refname>mallctlnametomib</refname>
+    <refname>mallctlbymib</refname>
+    <refname>malloc_stats_print</refname>
+    <refname>malloc_usable_size</refname>
+    -->
+    <refpurpose>general purpose memory allocation functions</refpurpose>
+  </refnamediv>
+  <refsect1 id="library">
+    <title>LIBRARY</title>
+    <para>This manual describes jemalloc @jemalloc_version@.  More information
+    can be found at the <ulink
+    url="http://jemalloc.net/">jemalloc website</ulink>.</para>
+  </refsect1>
+  <refsynopsisdiv>
+    <title>SYNOPSIS</title>
+    <funcsynopsis>
+      <funcsynopsisinfo>#include &lt;<filename class="headerfile">jemalloc/jemalloc.h</filename>&gt;</funcsynopsisinfo>
+      <refsect2>
+        <title>Standard API</title>
+        <funcprototype>
+          <funcdef>void *<function>malloc</function></funcdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>calloc</function></funcdef>
+          <paramdef>size_t <parameter>number</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>posix_memalign</function></funcdef>
+          <paramdef>void **<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>aligned_alloc</function></funcdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>realloc</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>free</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+        </funcprototype>
+      </refsect2>
+      <refsect2>
+        <title>Non-standard API</title>
+        <funcprototype>
+          <funcdef>void *<function>mallocx</function></funcdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>rallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>xallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>extra</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>sallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>dallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>sdallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>nallocx</function></funcdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctl</function></funcdef>
+          <paramdef>const char *<parameter>name</parameter></paramdef>
+          <paramdef>void *<parameter>oldp</parameter></paramdef>
+          <paramdef>size_t *<parameter>oldlenp</parameter></paramdef>
+          <paramdef>void *<parameter>newp</parameter></paramdef>
+          <paramdef>size_t <parameter>newlen</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctlnametomib</function></funcdef>
+          <paramdef>const char *<parameter>name</parameter></paramdef>
+          <paramdef>size_t *<parameter>mibp</parameter></paramdef>
+          <paramdef>size_t *<parameter>miblenp</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctlbymib</function></funcdef>
+          <paramdef>const size_t *<parameter>mib</parameter></paramdef>
+          <paramdef>size_t <parameter>miblen</parameter></paramdef>
+          <paramdef>void *<parameter>oldp</parameter></paramdef>
+          <paramdef>size_t *<parameter>oldlenp</parameter></paramdef>
+          <paramdef>void *<parameter>newp</parameter></paramdef>
+          <paramdef>size_t <parameter>newlen</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>malloc_stats_print</function></funcdef>
+          <paramdef>void <parameter>(*write_cb)</parameter>
+            <funcparams>void *, const char *</funcparams>
+          </paramdef>
+          <paramdef>void *<parameter>cbopaque</parameter></paramdef>
+          <paramdef>const char *<parameter>opts</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>malloc_usable_size</function></funcdef>
+          <paramdef>const void *<parameter>ptr</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>(*malloc_message)</function></funcdef>
+          <paramdef>void *<parameter>cbopaque</parameter></paramdef>
+          <paramdef>const char *<parameter>s</parameter></paramdef>
+        </funcprototype>
+        <para><type>const char *</type><varname>malloc_conf</varname>;</para>
+      </refsect2>
+    </funcsynopsis>
+  </refsynopsisdiv>
+  <refsect1 id="description">
+    <title>DESCRIPTION</title>
+    <refsect2>
+      <title>Standard API</title>
+
+      <para>The <function>malloc()</function> function allocates
+      <parameter>size</parameter> bytes of uninitialized memory.  The allocated
+      space is suitably aligned (after possible pointer coercion) for storage
+      of any type of object.</para>
+
+      <para>The <function>calloc()</function> function allocates
+      space for <parameter>number</parameter> objects, each
+      <parameter>size</parameter> bytes in length.  The result is identical to
+      calling <function>malloc()</function> with an argument of
+      <parameter>number</parameter> * <parameter>size</parameter>, with the
+      exception that the allocated memory is explicitly initialized to zero
+      bytes.</para>
+
+      <para>The <function>posix_memalign()</function> function
+      allocates <parameter>size</parameter> bytes of memory such that the
+      allocation's base address is a multiple of
+      <parameter>alignment</parameter>, and returns the allocation in the value
+      pointed to by <parameter>ptr</parameter>.  The requested
+      <parameter>alignment</parameter> must be a power of 2 at least as large as
+      <code language="C">sizeof(<type>void *</type>)</code>.</para>
+
+      <para>The <function>aligned_alloc()</function> function
+      allocates <parameter>size</parameter> bytes of memory such that the
+      allocation's base address is a multiple of
+      <parameter>alignment</parameter>.  The requested
+      <parameter>alignment</parameter> must be a power of 2.  Behavior is
+      undefined if <parameter>size</parameter> is not an integral multiple of
+      <parameter>alignment</parameter>.</para>
+
+      <para>The <function>realloc()</function> function changes the
+      size of the previously allocated memory referenced by
+      <parameter>ptr</parameter> to <parameter>size</parameter> bytes.  The
+      contents of the memory are unchanged up to the lesser of the new and old
+      sizes.  If the new size is larger, the contents of the newly allocated
+      portion of the memory are undefined.  Upon success, the memory referenced
+      by <parameter>ptr</parameter> is freed and a pointer to the newly
+      allocated memory is returned.  Note that
+      <function>realloc()</function> may move the memory allocation,
+      resulting in a different return value than <parameter>ptr</parameter>.
+      If <parameter>ptr</parameter> is <constant>NULL</constant>, the
+      <function>realloc()</function> function behaves identically to
+      <function>malloc()</function> for the specified size.</para>
+
+      <para>The <function>free()</function> function causes the
+      allocated memory referenced by <parameter>ptr</parameter> to be made
+      available for future allocations.  If <parameter>ptr</parameter> is
+      <constant>NULL</constant>, no action occurs.</para>
+    </refsect2>
+    <refsect2>
+      <title>Non-standard API</title>
+      <para>The <function>mallocx()</function>,
+      <function>rallocx()</function>,
+      <function>xallocx()</function>,
+      <function>sallocx()</function>,
+      <function>dallocx()</function>,
+      <function>sdallocx()</function>, and
+      <function>nallocx()</function> functions all have a
+      <parameter>flags</parameter> argument that can be used to specify
+      options.  The functions only check the options that are contextually
+      relevant.  Use bitwise or (<code language="C">|</code>) operations to
+      specify one or more of the following:
+        <variablelist>
+          <varlistentry id="MALLOCX_LG_ALIGN">
+            <term><constant>MALLOCX_LG_ALIGN(<parameter>la</parameter>)
+            </constant></term>
+
+            <listitem><para>Align the memory allocation to start at an address
+            that is a multiple of <code language="C">(1 &lt;&lt;
+            <parameter>la</parameter>)</code>.  This macro does not validate
+            that <parameter>la</parameter> is within the valid
+            range.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_ALIGN">
+            <term><constant>MALLOCX_ALIGN(<parameter>a</parameter>)
+            </constant></term>
+
+            <listitem><para>Align the memory allocation to start at an address
+            that is a multiple of <parameter>a</parameter>, where
+            <parameter>a</parameter> is a power of two.  This macro does not
+            validate that <parameter>a</parameter> is a power of 2.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_ZERO">
+            <term><constant>MALLOCX_ZERO</constant></term>
+
+            <listitem><para>Initialize newly allocated memory to contain zero
+            bytes.  In the growing reallocation case, the real size prior to
+            reallocation defines the boundary between untouched bytes and those
+            that are initialized to contain zero bytes.  If this macro is
+            absent, newly allocated memory is uninitialized.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_TCACHE">
+            <term><constant>MALLOCX_TCACHE(<parameter>tc</parameter>)
+            </constant></term>
+
+            <listitem><para>Use the thread-specific cache (tcache) specified by
+            the identifier <parameter>tc</parameter>, which must have been
+            acquired via the <link
+            linkend="tcache.create"><mallctl>tcache.create</mallctl></link>
+            mallctl.  This macro does not validate that
+            <parameter>tc</parameter> specifies a valid
+            identifier.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOC_TCACHE_NONE">
+            <term><constant>MALLOCX_TCACHE_NONE</constant></term>
+
+            <listitem><para>Do not use a thread-specific cache (tcache).  Unless
+            <constant>MALLOCX_TCACHE(<parameter>tc</parameter>)</constant> or
+            <constant>MALLOCX_TCACHE_NONE</constant> is specified, an
+            automatically managed tcache will be used under many circumstances.
+            This macro cannot be used in the same <parameter>flags</parameter>
+            argument as
+            <constant>MALLOCX_TCACHE(<parameter>tc</parameter>)</constant>.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_ARENA">
+            <term><constant>MALLOCX_ARENA(<parameter>a</parameter>)
+            </constant></term>
+
+            <listitem><para>Use the arena specified by the index
+            <parameter>a</parameter>.  This macro has no effect for regions that
+            were allocated via an arena other than the one specified.  This
+            macro does not validate that <parameter>a</parameter> specifies an
+            arena index in the valid range.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>mallocx()</function> function allocates at
+      least <parameter>size</parameter> bytes of memory, and returns a pointer
+      to the base address of the allocation.  Behavior is undefined if
+      <parameter>size</parameter> is <constant>0</constant>.</para>
+
+      <para>The <function>rallocx()</function> function resizes the
+      allocation at <parameter>ptr</parameter> to be at least
+      <parameter>size</parameter> bytes, and returns a pointer to the base
+      address of the resulting allocation, which may or may not have moved from
+      its original location.  Behavior is undefined if
+      <parameter>size</parameter> is <constant>0</constant>.</para>
+
+      <para>The <function>xallocx()</function> function resizes the
+      allocation at <parameter>ptr</parameter> in place to be at least
+      <parameter>size</parameter> bytes, and returns the real size of the
+      allocation.  If <parameter>extra</parameter> is non-zero, an attempt is
+      made to resize the allocation to be at least <code
+      language="C">(<parameter>size</parameter> +
+      <parameter>extra</parameter>)</code> bytes, though inability to allocate
+      the extra byte(s) will not by itself result in failure to resize.
+      Behavior is undefined if <parameter>size</parameter> is
+      <constant>0</constant>, or if <code
+      language="C">(<parameter>size</parameter> + <parameter>extra</parameter>
+      &gt; <constant>SIZE_T_MAX</constant>)</code>.</para>
+
+      <para>The <function>sallocx()</function> function returns the
+      real size of the allocation at <parameter>ptr</parameter>.</para>
+
+      <para>The <function>dallocx()</function> function causes the
+      memory referenced by <parameter>ptr</parameter> to be made available for
+      future allocations.</para>
+
+      <para>The <function>sdallocx()</function> function is an
+      extension of <function>dallocx()</function> with a
+      <parameter>size</parameter> parameter to allow the caller to pass in the
+      allocation size as an optimization.  The minimum valid input size is the
+      original requested size of the allocation, and the maximum valid input
+      size is the corresponding value returned by
+      <function>nallocx()</function> or
+      <function>sallocx()</function>.</para>
+
+      <para>The <function>nallocx()</function> function allocates no
+      memory, but it performs the same size computation as the
+      <function>mallocx()</function> function, and returns the real
+      size of the allocation that would result from the equivalent
+      <function>mallocx()</function> function call, or
+      <constant>0</constant> if the inputs exceed the maximum supported size
+      class and/or alignment.  Behavior is undefined if
+      <parameter>size</parameter> is <constant>0</constant>.</para>
+
+      <para>The <function>mallctl()</function> function provides a
+      general interface for introspecting the memory allocator, as well as
+      setting modifiable parameters and triggering actions.  The
+      period-separated <parameter>name</parameter> argument specifies a
+      location in a tree-structured namespace; see the <xref
+      linkend="mallctl_namespace" xrefstyle="template:%t"/> section for
+      documentation on the tree contents.  To read a value, pass a pointer via
+      <parameter>oldp</parameter> to adequate space to contain the value, and a
+      pointer to its length via <parameter>oldlenp</parameter>; otherwise pass
+      <constant>NULL</constant> and <constant>NULL</constant>.  Similarly, to
+      write a value, pass a pointer to the value via
+      <parameter>newp</parameter>, and its length via
+      <parameter>newlen</parameter>; otherwise pass <constant>NULL</constant>
+      and <constant>0</constant>.</para>
+
+      <para>The <function>mallctlnametomib()</function> function
+      provides a way to avoid repeated name lookups for applications that
+      repeatedly query the same portion of the namespace, by translating a name
+      to a <quote>Management Information Base</quote> (MIB) that can be passed
+      repeatedly to <function>mallctlbymib()</function>.  Upon
+      successful return from <function>mallctlnametomib()</function>,
+      <parameter>mibp</parameter> contains an array of
+      <parameter>*miblenp</parameter> integers, where
+      <parameter>*miblenp</parameter> is the lesser of the number of components
+      in <parameter>name</parameter> and the input value of
+      <parameter>*miblenp</parameter>.  Thus it is possible to pass a
+      <parameter>*miblenp</parameter> that is smaller than the number of
+      period-separated name components, which results in a partial MIB that can
+      be used as the basis for constructing a complete MIB.  For name
+      components that are integers (e.g. the 2 in
+      <link
+      linkend="arenas.bin.i.size"><mallctl>arenas.bin.2.size</mallctl></link>),
+      the corresponding MIB component will always be that integer.  Therefore,
+      it is legitimate to construct code like the following: <programlisting
+      language="C"><![CDATA[
+unsigned nbins, i;
+size_t mib[4];
+size_t len, miblen;
+
+len = sizeof(nbins);
+mallctl("arenas.nbins", &nbins, &len, NULL, 0);
+
+miblen = 4;
+mallctlnametomib("arenas.bin.0.size", mib, &miblen);
+for (i = 0; i < nbins; i++) {
+	size_t bin_size;
+
+	mib[2] = i;
+	len = sizeof(bin_size);
+	mallctlbymib(mib, miblen, (void *)&bin_size, &len, NULL, 0);
+	/* Do something with bin_size... */
+}]]></programlisting></para>
+
+      <varlistentry id="malloc_stats_print_opts">
+      </varlistentry>
+      <para>The <function>malloc_stats_print()</function> function writes
+      summary statistics via the <parameter>write_cb</parameter> callback
+      function pointer and <parameter>cbopaque</parameter> data passed to
+      <parameter>write_cb</parameter>, or <function>malloc_message()</function>
+      if <parameter>write_cb</parameter> is <constant>NULL</constant>.  The
+      statistics are presented in human-readable form unless <quote>J</quote> is
+      specified as a character within the <parameter>opts</parameter> string, in
+      which case the statistics are presented in <ulink
+      url="http://www.json.org/">JSON format</ulink>.  This function can be
+      called repeatedly.  General information that never changes during
+      execution can be omitted by specifying <quote>g</quote> as a character
+      within the <parameter>opts</parameter> string.  Note that
+      <function>malloc_stats_print()</function> uses the
+      <function>mallctl*()</function> functions internally, so inconsistent
+      statistics can be reported if multiple threads use these functions
+      simultaneously.  If <option>--enable-stats</option> is specified during
+      configuration, <quote>m</quote>, <quote>d</quote>, and <quote>a</quote>
+      can be specified to omit merged arena, destroyed merged arena, and per
+      arena statistics, respectively; <quote>b</quote> and <quote>l</quote> can
+      be specified to omit per size class statistics for bins and large objects,
+      respectively; <quote>x</quote> can be specified to omit all mutex
+      statistics; <quote>e</quote> can be used to omit extent statistics.
+      Unrecognized characters are silently ignored.  Note that thread caching
+      may prevent some statistics from being completely up to date, since extra
+      locking would be required to merge counters that track thread cache
+      operations.</para>
+
+      <para>The <function>malloc_usable_size()</function> function
+      returns the usable size of the allocation pointed to by
+      <parameter>ptr</parameter>.  The return value may be larger than the size
+      that was requested during allocation.  The
+      <function>malloc_usable_size()</function> function is not a
+      mechanism for in-place <function>realloc()</function>; rather
+      it is provided solely as a tool for introspection purposes.  Any
+      discrepancy between the requested allocation size and the size reported
+      by <function>malloc_usable_size()</function> should not be
+      depended on, since such behavior is entirely implementation-dependent.
+      </para>
+    </refsect2>
+  </refsect1>
+  <refsect1 id="tuning">
+    <title>TUNING</title>
+    <para>Once, when the first call is made to one of the memory allocation
+    routines, the allocator initializes its internals based in part on various
+    options that can be specified at compile- or run-time.</para>
+
+    <para>The string specified via <option>--with-malloc-conf</option>, the
+    string pointed to by the global variable <varname>malloc_conf</varname>, the
+    <quote>name</quote> of the file referenced by the symbolic link named
+    <filename class="symlink">/etc/malloc.conf</filename>, and the value of the
+    environment variable <envar>MALLOC_CONF</envar>, will be interpreted, in
+    that order, from left to right as options.  Note that
+    <varname>malloc_conf</varname> may be read before
+    <function>main()</function> is entered, so the declaration of
+    <varname>malloc_conf</varname> should specify an initializer that contains
+    the final value to be read by jemalloc.  <option>--with-malloc-conf</option>
+    and <varname>malloc_conf</varname> are compile-time mechanisms, whereas
+    <filename class="symlink">/etc/malloc.conf</filename> and
+    <envar>MALLOC_CONF</envar> can be safely set any time prior to program
+    invocation.</para>
+
+    <para>An options string is a comma-separated list of option:value pairs.
+    There is one key corresponding to each <link
+    linkend="opt.abort"><mallctl>opt.*</mallctl></link> mallctl (see the <xref
+    linkend="mallctl_namespace" xrefstyle="template:%t"/> section for options
+    documentation).  For example, <literal>abort:true,narenas:1</literal> sets
+    the <link linkend="opt.abort"><mallctl>opt.abort</mallctl></link> and <link
+    linkend="opt.narenas"><mallctl>opt.narenas</mallctl></link> options.  Some
+    options have boolean values (true/false), others have integer values (base
+    8, 10, or 16, depending on prefix), and yet others have raw string
+    values.</para>
+  </refsect1>
+  <refsect1 id="implementation_notes">
+    <title>IMPLEMENTATION NOTES</title>
+    <para>Traditionally, allocators have used
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> to obtain memory, which is
+    suboptimal for several reasons, including race conditions, increased
+    fragmentation, and artificial limitations on maximum usable memory.  If
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> is supported by the operating
+    system, this allocator uses both
+    <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> and
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>, in that order of preference;
+    otherwise only <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> is used.</para>
+
+    <para>This allocator uses multiple arenas in order to reduce lock
+    contention for threaded programs on multi-processor systems.  This works
+    well with regard to threading scalability, but incurs some costs.  There is
+    a small fixed per-arena overhead, and additionally, arenas manage memory
+    completely independently of each other, which means a small fixed increase
+    in overall memory fragmentation.  These overheads are not generally an
+    issue, given the number of arenas normally used.  Note that using
+    substantially more arenas than the default is not likely to improve
+    performance, mainly due to reduced cache performance.  However, it may make
+    sense to reduce the number of arenas if an application does not make much
+    use of the allocation functions.</para>
+
+    <para>In addition to multiple arenas, this allocator supports
+    thread-specific caching, in order to make it possible to completely avoid
+    synchronization for most allocation requests.  Such caching allows very fast
+    allocation in the common case, but it increases memory usage and
+    fragmentation, since a bounded number of objects can remain allocated in
+    each thread cache.</para>
+
+    <para>Memory is conceptually broken into extents.  Extents are always
+    aligned to multiples of the page size.  This alignment makes it possible to
+    find metadata for user objects quickly.  User objects are broken into two
+    categories according to size: small and large.  Contiguous small objects
+    comprise a slab, which resides within a single extent, whereas large objects
+    each have their own extents backing them.</para>
+
+    <para>Small objects are managed in groups by slabs.  Each slab maintains
+    a bitmap to track which regions are in use.  Allocation requests that are no
+    more than half the quantum (8 or 16, depending on architecture) are rounded
+    up to the nearest power of two that is at least <code
+    language="C">sizeof(<type>double</type>)</code>.  All other object size
+    classes are multiples of the quantum, spaced such that there are four size
+    classes for each doubling in size, which limits internal fragmentation to
+    approximately 20% for all but the smallest size classes.  Small size classes
+    are smaller than four times the page size, and large size classes extend
+    from four times the page size up to the largest size class that does not
+    exceed <constant>PTRDIFF_MAX</constant>.</para>
+
+    <para>Allocations are packed tightly together, which can be an issue for
+    multi-threaded applications.  If you need to assure that allocations do not
+    suffer from cacheline sharing, round your allocation requests up to the
+    nearest multiple of the cacheline size, or specify cacheline alignment when
+    allocating.</para>
+
+    <para>The <function>realloc()</function>,
+    <function>rallocx()</function>, and
+    <function>xallocx()</function> functions may resize allocations
+    without moving them under limited circumstances.  Unlike the
+    <function>*allocx()</function> API, the standard API does not
+    officially round up the usable size of an allocation to the nearest size
+    class, so technically it is necessary to call
+    <function>realloc()</function> to grow e.g. a 9-byte allocation to
+    16 bytes, or shrink a 16-byte allocation to 9 bytes.  Growth and shrinkage
+    trivially succeeds in place as long as the pre-size and post-size both round
+    up to the same size class.  No other API guarantees are made regarding
+    in-place resizing, but the current implementation also tries to resize large
+    allocations in place, as long as the pre-size and post-size are both large.
+    For shrinkage to succeed, the extent allocator must support splitting (see
+    <link
+    linkend="arena.i.extent_hooks"><mallctl>arena.&lt;i&gt;.extent_hooks</mallctl></link>).
+    Growth only succeeds if the trailing memory is currently available, and the
+    extent allocator supports merging.</para>
+
+    <para>Assuming 4 KiB pages and a 16-byte quantum on a 64-bit system, the
+    size classes in each category are as shown in <xref linkend="size_classes"
+    xrefstyle="template:Table %n"/>.</para>
+
+    <table xml:id="size_classes" frame="all">
+      <title>Size classes</title>
+      <tgroup cols="3" colsep="1" rowsep="1">
+      <colspec colname="c1" align="left"/>
+      <colspec colname="c2" align="right"/>
+      <colspec colname="c3" align="left"/>
+      <thead>
+        <row>
+          <entry>Category</entry>
+          <entry>Spacing</entry>
+          <entry>Size</entry>
+        </row>
+      </thead>
+      <tbody>
+        <row>
+          <entry morerows="8">Small</entry>
+          <entry>lg</entry>
+          <entry>[8]</entry>
+        </row>
+        <row>
+          <entry>16</entry>
+          <entry>[16, 32, 48, 64, 80, 96, 112, 128]</entry>
+        </row>
+        <row>
+          <entry>32</entry>
+          <entry>[160, 192, 224, 256]</entry>
+        </row>
+        <row>
+          <entry>64</entry>
+          <entry>[320, 384, 448, 512]</entry>
+        </row>
+        <row>
+          <entry>128</entry>
+          <entry>[640, 768, 896, 1024]</entry>
+        </row>
+        <row>
+          <entry>256</entry>
+          <entry>[1280, 1536, 1792, 2048]</entry>
+        </row>
+        <row>
+          <entry>512</entry>
+          <entry>[2560, 3072, 3584, 4096]</entry>
+        </row>
+        <row>
+          <entry>1 KiB</entry>
+          <entry>[5 KiB, 6 KiB, 7 KiB, 8 KiB]</entry>
+        </row>
+        <row>
+          <entry>2 KiB</entry>
+          <entry>[10 KiB, 12 KiB, 14 KiB]</entry>
+        </row>
+        <row>
+          <entry morerows="15">Large</entry>
+          <entry>2 KiB</entry>
+          <entry>[16 KiB]</entry>
+        </row>
+        <row>
+          <entry>4 KiB</entry>
+          <entry>[20 KiB, 24 KiB, 28 KiB, 32 KiB]</entry>
+        </row>
+        <row>
+          <entry>8 KiB</entry>
+          <entry>[40 KiB, 48 KiB, 54 KiB, 64 KiB]</entry>
+        </row>
+        <row>
+          <entry>16 KiB</entry>
+          <entry>[80 KiB, 96 KiB, 112 KiB, 128 KiB]</entry>
+        </row>
+        <row>
+          <entry>32 KiB</entry>
+          <entry>[160 KiB, 192 KiB, 224 KiB, 256 KiB]</entry>
+        </row>
+        <row>
+          <entry>64 KiB</entry>
+          <entry>[320 KiB, 384 KiB, 448 KiB, 512 KiB]</entry>
+        </row>
+        <row>
+          <entry>128 KiB</entry>
+          <entry>[640 KiB, 768 KiB, 896 KiB, 1 MiB]</entry>
+        </row>
+        <row>
+          <entry>256 KiB</entry>
+          <entry>[1280 KiB, 1536 KiB, 1792 KiB, 2 MiB]</entry>
+        </row>
+        <row>
+          <entry>512 KiB</entry>
+          <entry>[2560 KiB, 3 MiB, 3584 KiB, 4 MiB]</entry>
+        </row>
+        <row>
+          <entry>1 MiB</entry>
+          <entry>[5 MiB, 6 MiB, 7 MiB, 8 MiB]</entry>
+        </row>
+        <row>
+          <entry>2 MiB</entry>
+          <entry>[10 MiB, 12 MiB, 14 MiB, 16 MiB]</entry>
+        </row>
+        <row>
+          <entry>4 MiB</entry>
+          <entry>[20 MiB, 24 MiB, 28 MiB, 32 MiB]</entry>
+        </row>
+        <row>
+          <entry>8 MiB</entry>
+          <entry>[40 MiB, 48 MiB, 56 MiB, 64 MiB]</entry>
+        </row>
+        <row>
+          <entry>...</entry>
+          <entry>...</entry>
+        </row>
+        <row>
+          <entry>512 PiB</entry>
+          <entry>[2560 PiB, 3 EiB, 3584 PiB, 4 EiB]</entry>
+        </row>
+        <row>
+          <entry>1 EiB</entry>
+          <entry>[5 EiB, 6 EiB, 7 EiB]</entry>
+        </row>
+      </tbody>
+      </tgroup>
+    </table>
+  </refsect1>
+  <refsect1 id="mallctl_namespace">
+    <title>MALLCTL NAMESPACE</title>
+    <para>The following names are defined in the namespace accessible via the
+    <function>mallctl*()</function> functions.  Value types are specified in
+    parentheses, their readable/writable statuses are encoded as
+    <literal>rw</literal>, <literal>r-</literal>, <literal>-w</literal>, or
+    <literal>--</literal>, and required build configuration flags follow, if
+    any.  A name element encoded as <literal>&lt;i&gt;</literal> or
+    <literal>&lt;j&gt;</literal> indicates an integer component, where the
+    integer varies from 0 to some upper value that must be determined via
+    introspection.  In the case of <mallctl>stats.arenas.&lt;i&gt;.*</mallctl>
+    and <mallctl>arena.&lt;i&gt;.{initialized,purge,decay,dss}</mallctl>,
+    <literal>&lt;i&gt;</literal> equal to
+    <constant>MALLCTL_ARENAS_ALL</constant> can be used to operate on all arenas
+    or access the summation of statistics from all arenas; similarly
+    <literal>&lt;i&gt;</literal> equal to
+    <constant>MALLCTL_ARENAS_DESTROYED</constant> can be used to access the
+    summation of statistics from all destroyed arenas.  These constants can be
+    utilized either via <function>mallctlnametomib()</function> followed by
+    <function>mallctlbymib()</function>, or via code such as the following:
+    <programlisting language="C"><![CDATA[
+#define STRINGIFY_HELPER(x) #x
+#define STRINGIFY(x) STRINGIFY_HELPER(x)
+
+mallctl("arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".decay",
+    NULL, NULL, NULL, 0);]]></programlisting>
+    Take special note of the <link
+    linkend="epoch"><mallctl>epoch</mallctl></link> mallctl, which controls
+    refreshing of cached dynamic statistics.</para>
+
+    <variablelist>
+      <varlistentry id="version">
+        <term>
+          <mallctl>version</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Return the jemalloc version string.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="epoch">
+        <term>
+          <mallctl>epoch</mallctl>
+          (<type>uint64_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>If a value is passed in, refresh the data from which
+        the <function>mallctl*()</function> functions report values,
+        and increment the epoch.  Return the current epoch.  This is useful for
+        detecting whether another thread caused a refresh.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="background_thread">
+        <term>
+          <mallctl>background_thread</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Enable/disable internal background worker threads.  When
+        set to true, background threads are created on demand (the number of
+        background threads will be no more than the number of CPUs or active
+        arenas).  Threads run periodically, and handle <link
+        linkend="arena.i.decay">purging</link> asynchronously.  When switching
+        off, background threads are terminated synchronously.  Note that after
+        <citerefentry><refentrytitle>fork</refentrytitle><manvolnum>2</manvolnum></citerefentry>
+        function, the state in the child process will be disabled regardless
+        the state in parent process. See <link
+        linkend="stats.background_thread.num_threads"><mallctl>stats.background_thread</mallctl></link>
+        for related stats.  <link
+        linkend="opt.background_thread"><mallctl>opt.background_thread</mallctl></link>
+        can be used to set the default option.  This option is only available on
+        selected pthread-based platforms.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="max_background_threads">
+        <term>
+          <mallctl>max_background_threads</mallctl>
+          (<type>size_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Maximum number of background worker threads that will
+        be created.  This value is capped at <link
+        linkend="opt.max_background_threads"><mallctl>opt.max_background_threads</mallctl></link> at
+        startup.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.cache_oblivious">
+        <term>
+          <mallctl>config.cache_oblivious</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-cache-oblivious</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.debug">
+        <term>
+          <mallctl>config.debug</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-debug</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.fill">
+        <term>
+          <mallctl>config.fill</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-fill</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.lazy_lock">
+        <term>
+          <mallctl>config.lazy_lock</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-lazy-lock</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.malloc_conf">
+        <term>
+          <mallctl>config.malloc_conf</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Embedded configure-time-specified run-time options
+        string, empty unless <option>--with-malloc-conf</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.prof">
+        <term>
+          <mallctl>config.prof</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-prof</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.prof_libgcc">
+        <term>
+          <mallctl>config.prof_libgcc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--disable-prof-libgcc</option> was not
+        specified during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.prof_libunwind">
+        <term>
+          <mallctl>config.prof_libunwind</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-prof-libunwind</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.stats">
+        <term>
+          <mallctl>config.stats</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-stats</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+
+      <varlistentry id="config.utrace">
+        <term>
+          <mallctl>config.utrace</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-utrace</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.xmalloc">
+        <term>
+          <mallctl>config.xmalloc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-xmalloc</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.abort">
+        <term>
+          <mallctl>opt.abort</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Abort-on-warning enabled/disabled.  If true, most
+        warnings are fatal.  Note that runtime option warnings are not included
+        (see <link
+        linkend="opt.abort_conf"><mallctl>opt.abort_conf</mallctl></link> for
+        that). The process will call
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> in these cases.  This option is
+        disabled by default unless <option>--enable-debug</option> is
+        specified during configuration, in which case it is enabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.confirm_conf">
+        <term>
+          <mallctl>opt.confirm_conf</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+	<listitem><para>Confirm-runtime-options-when-program-starts
+	enabled/disabled.  If true, the string specified via
+	<option>--with-malloc-conf</option>, the string pointed to by the
+	global variable <varname>malloc_conf</varname>, the <quote>name</quote>
+	of the file referenced by the symbolic link named
+	<filename class="symlink">/etc/malloc.conf</filename>, and the value of
+	the environment variable <envar>MALLOC_CONF</envar>, will be printed in
+	order.  Then, each option being set will be individually printed.  This
+	option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.abort_conf">
+        <term>
+          <mallctl>opt.abort_conf</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Abort-on-invalid-configuration enabled/disabled.  If
+        true, invalid runtime options are fatal.  The process will call
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> in these cases.  This option is
+        disabled by default unless <option>--enable-debug</option> is
+        specified during configuration, in which case it is enabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.metadata_thp">
+        <term>
+          <mallctl>opt.metadata_thp</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Controls whether to allow jemalloc to use transparent
+        huge page (THP) for internal metadata (see <link
+        linkend="stats.metadata">stats.metadata</link>).  <quote>always</quote>
+        allows such usage.  <quote>auto</quote> uses no THP initially, but may
+        begin to do so when metadata usage reaches certain level.  The default
+        is <quote>disabled</quote>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.retain">
+        <term>
+          <mallctl>opt.retain</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>If true, retain unused virtual memory for later reuse
+        rather than discarding it by calling
+        <citerefentry><refentrytitle>munmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> or equivalent (see <link
+        linkend="stats.retained">stats.retained</link> for related details).
+        It also makes jemalloc use <citerefentry>
+        <refentrytitle>mmap</refentrytitle><manvolnum>2</manvolnum>
+        </citerefentry> or equivalent in a more greedy way, mapping larger
+        chunks in one go.  This option is disabled by default unless discarding
+        virtual memory is known to trigger platform-specific performance
+        problems, namely 1) for [64-bit] Linux, which has a quirk in its virtual
+        memory allocation algorithm that causes semi-permanent VM map holes
+        under normal jemalloc operation; and 2) for [64-bit] Windows, which
+        disallows split / merged regions with
+        <parameter><constant>MEM_RELEASE</constant></parameter>.  Although the
+        same issues may present on 32-bit platforms as well, retaining virtual
+        memory for 32-bit Linux and Windows is disabled by default due to the
+        practical possibility of address space exhaustion.  </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.dss">
+        <term>
+          <mallctl>opt.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>dss (<citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry>) allocation precedence as
+        related to <citerefentry><refentrytitle>mmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> allocation.  The following
+        settings are supported if
+        <citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> is supported by the operating
+        system: <quote>disabled</quote>, <quote>primary</quote>, and
+        <quote>secondary</quote>; otherwise only <quote>disabled</quote> is
+        supported.  The default is <quote>secondary</quote> if
+        <citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> is supported by the operating
+        system; <quote>disabled</quote> otherwise.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.narenas">
+        <term>
+          <mallctl>opt.narenas</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum number of arenas to use for automatic
+        multiplexing of threads and arenas.  The default is four times the
+        number of CPUs, or one if there is a single CPU.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.oversize_threshold">
+        <term>
+          <mallctl>opt.oversize_threshold</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>The threshold in bytes of which requests are considered
+        oversize.  Allocation requests with greater sizes are fulfilled from a
+        dedicated arena (automatically managed, however not within
+        <literal>narenas</literal>), in order to reduce fragmentation by not
+        mixing huge allocations with small ones.  In addition, the decay API
+        guarantees on the extents greater than the specified threshold may be
+        overridden.  Note that requests with arena index specified via
+        <constant>MALLOCX_ARENA</constant>, or threads associated with explicit
+        arenas will not be considered.  The default threshold is 8MiB.  Values
+        not within large size classes disables this feature.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.percpu_arena">
+        <term>
+          <mallctl>opt.percpu_arena</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Per CPU arena mode.  Use the <quote>percpu</quote>
+        setting to enable this feature, which uses number of CPUs to determine
+        number of arenas, and bind threads to arenas dynamically based on the
+        CPU the thread runs on currently.  <quote>phycpu</quote> setting uses
+        one arena per physical CPU, which means the two hyper threads on the
+        same CPU share one arena.  Note that no runtime checking regarding the
+        availability of hyper threading is done at the moment.  When set to
+        <quote>disabled</quote>, narenas and thread to arena association will
+        not be impacted by this option.  The default is <quote>disabled</quote>.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.background_thread">
+        <term>
+          <mallctl>opt.background_thread</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Internal background worker threads enabled/disabled.
+        Because of potential circular dependencies, enabling background thread
+        using this option may cause crash or deadlock during initialization. For
+        a reliable way to use this feature, see <link
+        linkend="background_thread">background_thread</link> for dynamic control
+        options and details.  This option is disabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.max_background_threads">
+        <term>
+          <mallctl>opt.max_background_threads</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum number of background threads that will be created
+        if <link linkend="background_thread">background_thread</link> is set.
+        Defaults to number of cpus.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.dirty_decay_ms">
+        <term>
+          <mallctl>opt.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused dirty pages until an equivalent set of unused dirty pages
+        is purged (i.e. converted to muzzy via e.g.
+        <function>madvise(<parameter>...</parameter><parameter><constant>MADV_FREE</constant></parameter>)</function>
+        if supported by the operating system, or converted to clean otherwise)
+        and/or reused.  Dirty pages are defined as previously having been
+        potentially written to by the application, and therefore consuming
+        physical memory, yet having no current use.  The pages are incrementally
+        purged according to a sigmoidal decay curve that starts and ends with
+        zero purge rate.  A decay time of 0 causes all unused dirty pages to be
+        purged immediately upon creation.  A decay time of -1 disables purging.
+        The default decay time is 10 seconds.  See <link
+        linkend="arenas.dirty_decay_ms"><mallctl>arenas.dirty_decay_ms</mallctl></link>
+        and <link
+        linkend="arena.i.dirty_decay_ms"><mallctl>arena.&lt;i&gt;.dirty_decay_ms</mallctl></link>
+        for related dynamic control options.  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for a description of muzzy pages.for a description of muzzy pages.  Note
+        that when the <link
+        linkend="opt.oversize_threshold"><mallctl>oversize_threshold</mallctl></link>
+        feature is enabled, the arenas reserved for oversize requests may have
+        its own default decay settings.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.muzzy_decay_ms">
+        <term>
+          <mallctl>opt.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused muzzy pages until an equivalent set of unused muzzy pages
+        is purged (i.e. converted to clean) and/or reused.  Muzzy pages are
+        defined as previously having been unused dirty pages that were
+        subsequently purged in a manner that left them subject to the
+        reclamation whims of the operating system (e.g.
+        <function>madvise(<parameter>...</parameter><parameter><constant>MADV_FREE</constant></parameter>)</function>),
+        and therefore in an indeterminate state.  The pages are incrementally
+        purged according to a sigmoidal decay curve that starts and ends with
+        zero purge rate.  A decay time of 0 causes all unused muzzy pages to be
+        purged immediately upon creation.  A decay time of -1 disables purging.
+        The default decay time is 10 seconds.  See <link
+        linkend="arenas.muzzy_decay_ms"><mallctl>arenas.muzzy_decay_ms</mallctl></link>
+        and <link
+        linkend="arena.i.muzzy_decay_ms"><mallctl>arena.&lt;i&gt;.muzzy_decay_ms</mallctl></link>
+        for related dynamic control options.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_extent_max_active_fit">
+        <term>
+          <mallctl>opt.lg_extent_max_active_fit</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>When reusing dirty extents, this determines the (log
+        base 2 of the) maximum ratio between the size of the active extent
+        selected (to split off from) and the size of the requested allocation.
+        This prevents the splitting of large active extents for smaller
+        allocations, which can reduce fragmentation over the long run
+        (especially for non-active extents).  Lower value may reduce
+        fragmentation, at the cost of extra active extents.  The default value
+        is 6, which gives a maximum ratio of 64 (2^6).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.stats_print">
+        <term>
+          <mallctl>opt.stats_print</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Enable/disable statistics printing at exit.  If
+        enabled, the <function>malloc_stats_print()</function>
+        function is called at program exit via an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function.  <link
+        linkend="opt.stats_print_opts"><mallctl>opt.stats_print_opts</mallctl></link>
+        can be combined to specify output options. If
+        <option>--enable-stats</option> is specified during configuration, this
+        has the potential to cause deadlock for a multi-threaded process that
+        exits while one or more threads are executing in the memory allocation
+        functions.  Furthermore, <function>atexit()</function> may
+        allocate memory during application initialization and then deadlock
+        internally when jemalloc in turn calls
+        <function>atexit()</function>, so this option is not
+        universally usable (though the application can register its own
+        <function>atexit()</function> function with equivalent
+        functionality).  Therefore, this option should only be used with care;
+        it is primarily intended as a performance tuning aid during application
+        development.  This option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.stats_print_opts">
+        <term>
+          <mallctl>opt.stats_print_opts</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Options (the <parameter>opts</parameter> string) to pass
+        to the <function>malloc_stats_print()</function> at exit (enabled
+        through <link
+        linkend="opt.stats_print"><mallctl>opt.stats_print</mallctl></link>). See
+        available options in <link
+        linkend="malloc_stats_print_opts"><function>malloc_stats_print()</function></link>.
+        Has no effect unless <link
+        linkend="opt.stats_print"><mallctl>opt.stats_print</mallctl></link> is
+        enabled.  The default is <quote></quote>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.junk">
+        <term>
+          <mallctl>opt.junk</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Junk filling.  If set to <quote>alloc</quote>, each byte
+        of uninitialized allocated memory will be initialized to
+        <literal>0xa5</literal>.  If set to <quote>free</quote>, all deallocated
+        memory will be initialized to <literal>0x5a</literal>.  If set to
+        <quote>true</quote>, both allocated and deallocated memory will be
+        initialized, and if set to <quote>false</quote>, junk filling be
+        disabled entirely.  This is intended for debugging and will impact
+        performance negatively.  This option is <quote>false</quote> by default
+        unless <option>--enable-debug</option> is specified during
+        configuration, in which case it is <quote>true</quote> by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.zero">
+        <term>
+          <mallctl>opt.zero</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Zero filling enabled/disabled.  If enabled, each byte
+        of uninitialized allocated memory will be initialized to 0.  Note that
+        this initialization only happens once for each byte, so
+        <function>realloc()</function> and
+        <function>rallocx()</function> calls do not zero memory that
+        was previously allocated.  This is intended for debugging and will
+        impact performance negatively.  This option is disabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.utrace">
+        <term>
+          <mallctl>opt.utrace</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-utrace</option>]
+        </term>
+        <listitem><para>Allocation tracing based on
+        <citerefentry><refentrytitle>utrace</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> enabled/disabled.  This option
+        is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.xmalloc">
+        <term>
+          <mallctl>opt.xmalloc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-xmalloc</option>]
+        </term>
+        <listitem><para>Abort-on-out-of-memory enabled/disabled.  If enabled,
+        rather than returning failure for any allocation function, display a
+        diagnostic message on <constant>STDERR_FILENO</constant> and cause the
+        program to drop core (using
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry>).  If an application is
+        designed to depend on this behavior, set the option at compile time by
+        including the following in the source code:
+        <programlisting language="C"><![CDATA[
+malloc_conf = "xmalloc:true";]]></programlisting>
+        This option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.tcache">
+        <term>
+          <mallctl>opt.tcache</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Thread-specific caching (tcache) enabled/disabled.  When
+        there are multiple threads, each thread uses a tcache for objects up to
+        a certain size.  Thread-specific caching allows many allocations to be
+        satisfied without performing any thread synchronization, at the cost of
+        increased memory use.  See the <link
+        linkend="opt.lg_tcache_max"><mallctl>opt.lg_tcache_max</mallctl></link>
+        option for related tuning information.  This option is enabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_tcache_max">
+        <term>
+          <mallctl>opt.lg_tcache_max</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size class (log base 2) to cache in the
+        thread-specific cache (tcache).  At a minimum, all small size classes
+        are cached, and at a maximum all large size classes are cached.  The
+        default maximum is 32 KiB (2^15).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.thp">
+        <term>
+          <mallctl>opt.thp</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Transparent hugepage (THP) mode. Settings "always",
+        "never" and "default" are available if THP is supported by the operating
+        system.  The "always" setting enables transparent hugepage for all user
+        memory mappings with
+        <parameter><constant>MADV_HUGEPAGE</constant></parameter>; "never"
+        ensures no transparent hugepage with
+        <parameter><constant>MADV_NOHUGEPAGE</constant></parameter>; the default
+        setting "default" makes no changes.  Note that: this option does not
+        affect THP for jemalloc internal metadata (see <link
+        linkend="opt.metadata_thp"><mallctl>opt.metadata_thp</mallctl></link>);
+        in addition, for arenas with customized <link
+        linkend="arena.i.extent_hooks"><mallctl>extent_hooks</mallctl></link>,
+        this option is bypassed as it is implemented as part of the default
+        extent hooks.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof">
+        <term>
+          <mallctl>opt.prof</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Memory profiling enabled/disabled.  If enabled, profile
+        memory allocation activity.  See the <link
+        linkend="opt.prof_active"><mallctl>opt.prof_active</mallctl></link>
+        option for on-the-fly activation/deactivation.  See the <link
+        linkend="opt.lg_prof_sample"><mallctl>opt.lg_prof_sample</mallctl></link>
+        option for probabilistic sampling control.  See the <link
+        linkend="opt.prof_accum"><mallctl>opt.prof_accum</mallctl></link>
+        option for control of cumulative sample reporting.  See the <link
+        linkend="opt.lg_prof_interval"><mallctl>opt.lg_prof_interval</mallctl></link>
+        option for information on interval-triggered profile dumping, the <link
+        linkend="opt.prof_gdump"><mallctl>opt.prof_gdump</mallctl></link>
+        option for information on high-water-triggered profile dumping, and the
+        <link linkend="opt.prof_final"><mallctl>opt.prof_final</mallctl></link>
+        option for final profile dumping.  Profile output is compatible with
+        the <command>jeprof</command> command, which is based on the
+        <command>pprof</command> that is developed as part of the <ulink
+        url="http://code.google.com/p/gperftools/">gperftools
+        package</ulink>.  See <link linkend="heap_profile_format">HEAP PROFILE
+        FORMAT</link> for heap profile format documentation.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_prefix">
+        <term>
+          <mallctl>opt.prof_prefix</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Filename prefix for profile dumps.  If the prefix is
+        set to the empty string, no automatic dumps will occur; this is
+        primarily useful for disabling the automatic final heap dump (which
+        also disables leak reporting, if enabled).  The default prefix is
+        <filename>jeprof</filename>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_active">
+        <term>
+          <mallctl>opt.prof_active</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Profiling activated/deactivated.  This is a secondary
+        control mechanism that makes it possible to start the application with
+        profiling enabled (see the <link
+        linkend="opt.prof"><mallctl>opt.prof</mallctl></link> option) but
+        inactive, then toggle profiling at any time during program execution
+        with the <link
+        linkend="prof.active"><mallctl>prof.active</mallctl></link> mallctl.
+        This option is enabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_thread_active_init">
+        <term>
+          <mallctl>opt.prof_thread_active_init</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Initial setting for <link
+        linkend="thread.prof.active"><mallctl>thread.prof.active</mallctl></link>
+        in newly created threads.  The initial setting for newly created threads
+        can also be changed during execution via the <link
+        linkend="prof.thread_active_init"><mallctl>prof.thread_active_init</mallctl></link>
+        mallctl.  This option is enabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_prof_sample">
+        <term>
+          <mallctl>opt.lg_prof_sample</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average interval (log base 2) between allocation
+        samples, as measured in bytes of allocation activity.  Increasing the
+        sampling interval decreases profile fidelity, but also decreases the
+        computational overhead.  The default sample interval is 512 KiB (2^19
+        B).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_accum">
+        <term>
+          <mallctl>opt.prof_accum</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Reporting of cumulative object/byte counts in profile
+        dumps enabled/disabled.  If this option is enabled, every unique
+        backtrace must be stored for the duration of execution.  Depending on
+        the application, this can impose a large memory overhead, and the
+        cumulative counts are not always of interest.  This option is disabled
+        by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_prof_interval">
+        <term>
+          <mallctl>opt.lg_prof_interval</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average interval (log base 2) between memory profile
+        dumps, as measured in bytes of allocation activity.  The actual
+        interval between dumps may be sporadic because decentralized allocation
+        counters are used to avoid synchronization bottlenecks.  Profiles are
+        dumped to files named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.i&lt;iseq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the
+        <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.  By default, interval-triggered profile dumping is disabled
+        (encoded as -1).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_gdump">
+        <term>
+          <mallctl>opt.prof_gdump</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Set the initial state of <link
+        linkend="prof.gdump"><mallctl>prof.gdump</mallctl></link>, which when
+        enabled triggers a memory profile dump every time the total virtual
+        memory exceeds the previous maximum.  This option is disabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_final">
+        <term>
+          <mallctl>opt.prof_final</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Use an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function to dump final memory
+        usage to a file named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.f.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.  Note that <function>atexit()</function> may allocate
+        memory during application initialization and then deadlock internally
+        when jemalloc in turn calls <function>atexit()</function>, so
+        this option is not universally usable (though the application can
+        register its own <function>atexit()</function> function with
+        equivalent functionality).  This option is disabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_leak">
+        <term>
+          <mallctl>opt.prof_leak</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Leak reporting enabled/disabled.  If enabled, use an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function to report memory leaks
+        detected by allocation sampling.  See the
+        <link linkend="opt.prof"><mallctl>opt.prof</mallctl></link> option for
+        information on analyzing heap profile output.  This option is disabled
+        by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.arena">
+        <term>
+          <mallctl>thread.arena</mallctl>
+          (<type>unsigned</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Get or set the arena associated with the calling
+        thread.  If the specified arena was not initialized beforehand (see the
+        <link
+        linkend="arena.i.initialized"><mallctl>arena.i.initialized</mallctl></link>
+        mallctl), it will be automatically initialized as a side effect of
+        calling this interface.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.allocated">
+        <term>
+          <mallctl>thread.allocated</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get the total number of bytes ever allocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.allocatedp">
+        <term>
+          <mallctl>thread.allocatedp</mallctl>
+          (<type>uint64_t *</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get a pointer to the the value that is returned by the
+        <link
+        linkend="thread.allocated"><mallctl>thread.allocated</mallctl></link>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <function>mallctl*()</function> calls.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.deallocated">
+        <term>
+          <mallctl>thread.deallocated</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get the total number of bytes ever deallocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.deallocatedp">
+        <term>
+          <mallctl>thread.deallocatedp</mallctl>
+          (<type>uint64_t *</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get a pointer to the the value that is returned by the
+        <link
+        linkend="thread.deallocated"><mallctl>thread.deallocated</mallctl></link>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <function>mallctl*()</function> calls.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.tcache.enabled">
+        <term>
+          <mallctl>thread.tcache.enabled</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Enable/disable calling thread's tcache.  The tcache is
+        implicitly flushed as a side effect of becoming
+        disabled (see <link
+        linkend="thread.tcache.flush"><mallctl>thread.tcache.flush</mallctl></link>).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.tcache.flush">
+        <term>
+          <mallctl>thread.tcache.flush</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Flush calling thread's thread-specific cache (tcache).
+        This interface releases all cached objects and internal data structures
+        associated with the calling thread's tcache.  Ordinarily, this interface
+        need not be called, since automatic periodic incremental garbage
+        collection occurs, and the thread cache is automatically discarded when
+        a thread exits.  However, garbage collection is triggered by allocation
+        activity, so it is possible for a thread that stops
+        allocating/deallocating to retain its cache indefinitely, in which case
+        the developer may find manual flushing useful.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.prof.name">
+        <term>
+          <mallctl>thread.prof.name</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal> or
+          <literal>-w</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Get/set the descriptive name associated with the calling
+        thread in memory profile dumps.  An internal copy of the name string is
+        created, so the input string need not be maintained after this interface
+        completes execution.  The output string of this interface should be
+        copied for non-ephemeral uses, because multiple implementation details
+        can cause asynchronous string deallocation.  Furthermore, each
+        invocation of this interface can only read or write; simultaneous
+        read/write is not supported due to string lifetime limitations.  The
+        name string must be nil-terminated and comprised only of characters in
+        the sets recognized
+        by <citerefentry><refentrytitle>isgraph</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> and
+        <citerefentry><refentrytitle>isblank</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.prof.active">
+        <term>
+          <mallctl>thread.prof.active</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Control whether sampling is currently active for the
+        calling thread.  This is an activation mechanism in addition to <link
+        linkend="prof.active"><mallctl>prof.active</mallctl></link>; both must
+        be active for the calling thread to sample.  This flag is enabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="tcache.create">
+        <term>
+          <mallctl>tcache.create</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Create an explicit thread-specific cache (tcache) and
+        return an identifier that can be passed to the <link
+        linkend="MALLOCX_TCACHE"><constant>MALLOCX_TCACHE(<parameter>tc</parameter>)</constant></link>
+        macro to explicitly use the specified cache rather than the
+        automatically managed one that is used by default.  Each explicit cache
+        can be used by only one thread at a time; the application must assure
+        that this constraint holds.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="tcache.flush">
+        <term>
+          <mallctl>tcache.flush</mallctl>
+          (<type>unsigned</type>)
+          <literal>-w</literal>
+        </term>
+        <listitem><para>Flush the specified thread-specific cache (tcache).  The
+        same considerations apply to this interface as to <link
+        linkend="thread.tcache.flush"><mallctl>thread.tcache.flush</mallctl></link>,
+        except that the tcache will never be automatically discarded.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="tcache.destroy">
+        <term>
+          <mallctl>tcache.destroy</mallctl>
+          (<type>unsigned</type>)
+          <literal>-w</literal>
+        </term>
+        <listitem><para>Flush the specified thread-specific cache (tcache) and
+        make the identifier available for use during a future tcache creation.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.initialized">
+        <term>
+          <mallctl>arena.&lt;i&gt;.initialized</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Get whether the specified arena's statistics are
+        initialized (i.e. the arena was initialized prior to the current epoch).
+        This interface can also be nominally used to query whether the merged
+        statistics corresponding to <constant>MALLCTL_ARENAS_ALL</constant> are
+        initialized (always true).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.decay">
+        <term>
+          <mallctl>arena.&lt;i&gt;.decay</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Trigger decay-based purging of unused dirty/muzzy pages
+        for arena &lt;i&gt;, or for all arenas if &lt;i&gt; equals
+        <constant>MALLCTL_ARENAS_ALL</constant>.  The proportion of unused
+        dirty/muzzy pages to be purged depends on the current time; see <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        and <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzy_decay_ms</mallctl></link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.purge">
+        <term>
+          <mallctl>arena.&lt;i&gt;.purge</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Purge all unused dirty pages for arena &lt;i&gt;, or for
+        all arenas if &lt;i&gt; equals <constant>MALLCTL_ARENAS_ALL</constant>.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.reset">
+        <term>
+          <mallctl>arena.&lt;i&gt;.reset</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Discard all of the arena's extant allocations.  This
+        interface can only be used with arenas explicitly created via <link
+        linkend="arenas.create"><mallctl>arenas.create</mallctl></link>.  None
+        of the arena's discarded/cached allocations may accessed afterward.  As
+        part of this requirement, all thread caches which were used to
+        allocate/deallocate in conjunction with the arena must be flushed
+        beforehand.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.destroy">
+        <term>
+          <mallctl>arena.&lt;i&gt;.destroy</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Destroy the arena.  Discard all of the arena's extant
+        allocations using the same mechanism as for <link
+        linkend="arena.i.reset"><mallctl>arena.&lt;i&gt;.reset</mallctl></link>
+        (with all the same constraints and side effects), merge the arena stats
+        into those accessible at arena index
+        <constant>MALLCTL_ARENAS_DESTROYED</constant>, and then completely
+        discard all metadata associated with the arena.  Future calls to <link
+        linkend="arenas.create"><mallctl>arenas.create</mallctl></link> may
+        recycle the arena index.  Destruction will fail if any threads are
+        currently associated with the arena as a result of calls to <link
+        linkend="thread.arena"><mallctl>thread.arena</mallctl></link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.dss">
+        <term>
+          <mallctl>arena.&lt;i&gt;.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Set the precedence of dss allocation as related to mmap
+        allocation for arena &lt;i&gt;, or for all arenas if &lt;i&gt; equals
+        <constant>MALLCTL_ARENAS_ALL</constant>.  See <link
+        linkend="opt.dss"><mallctl>opt.dss</mallctl></link> for supported
+        settings.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.dirty_decay_ms">
+        <term>
+          <mallctl>arena.&lt;i&gt;.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current per-arena approximate time in milliseconds from
+        the creation of a set of unused dirty pages until an equivalent set of
+        unused dirty pages is purged and/or reused.  Each time this interface is
+        set, all currently unused dirty pages are considered to have fully
+        decayed, which causes immediate purging of all unused dirty pages unless
+        the decay time is set to -1 (i.e. purging disabled).  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.muzzy_decay_ms">
+        <term>
+          <mallctl>arena.&lt;i&gt;.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current per-arena approximate time in milliseconds from
+        the creation of a set of unused muzzy pages until an equivalent set of
+        unused muzzy pages is purged and/or reused.  Each time this interface is
+        set, all currently unused muzzy pages are considered to have fully
+        decayed, which causes immediate purging of all unused muzzy pages unless
+        the decay time is set to -1 (i.e. purging disabled).  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.retain_grow_limit">
+        <term>
+          <mallctl>arena.&lt;i&gt;.retain_grow_limit</mallctl>
+          (<type>size_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Maximum size to grow retained region (only relevant when
+        <link linkend="opt.retain"><mallctl>opt.retain</mallctl></link> is
+        enabled).  This controls the maximum increment to expand virtual memory,
+        or allocation through <link
+        linkend="arena.i.extent_hooks"><mallctl>arena.&lt;i&gt;extent_hooks</mallctl></link>.
+        In particular, if customized extent hooks reserve physical memory
+        (e.g. 1G huge pages), this is useful to control the allocation hook's
+        input size.  The default is no limit.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.extent_hooks">
+        <term>
+          <mallctl>arena.&lt;i&gt;.extent_hooks</mallctl>
+          (<type>extent_hooks_t *</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Get or set the extent management hook functions for
+        arena &lt;i&gt;.  The functions must be capable of operating on all
+        extant extents associated with arena &lt;i&gt;, usually by passing
+        unknown extents to the replaced functions.  In practice, it is feasible
+        to control allocation for arenas explicitly created via <link
+        linkend="arenas.create"><mallctl>arenas.create</mallctl></link> such
+        that all extents originate from an application-supplied extent allocator
+        (by specifying the custom extent hook functions during arena creation).
+        However, the API guarantees for the automatically created arenas may be
+        relaxed -- hooks set there may be called in a "best effort" fashion; in
+        addition there may be extents created prior to the application having an
+        opportunity to take over extent allocation.</para>
+
+        <programlisting language="C"><![CDATA[
+typedef extent_hooks_s extent_hooks_t;
+struct extent_hooks_s {
+	extent_alloc_t		*alloc;
+	extent_dalloc_t		*dalloc;
+	extent_destroy_t	*destroy;
+	extent_commit_t		*commit;
+	extent_decommit_t	*decommit;
+	extent_purge_t		*purge_lazy;
+	extent_purge_t		*purge_forced;
+	extent_split_t		*split;
+	extent_merge_t		*merge;
+};]]></programlisting>
+        <para>The <type>extent_hooks_t</type> structure comprises function
+        pointers which are described individually below.  jemalloc uses these
+        functions to manage extent lifetime, which starts off with allocation of
+        mapped committed memory, in the simplest case followed by deallocation.
+        However, there are performance and platform reasons to retain extents
+        for later reuse.  Cleanup attempts cascade from deallocation to decommit
+        to forced purging to lazy purging, which gives the extent management
+        functions opportunities to reject the most permanent cleanup operations
+        in favor of less permanent (and often less costly) operations.  All
+        operations except allocation can be universally opted out of by setting
+        the hook pointers to <constant>NULL</constant>, or selectively opted out
+        of by returning failure.  Note that once the extent hook is set, the
+        structure is accessed directly by the associated arenas, so it must
+        remain valid for the entire lifetime of the arenas.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef void *<function>(extent_alloc_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>new_addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>bool *<parameter>zero</parameter></paramdef>
+          <paramdef>bool *<parameter>commit</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent allocation function conforms to the
+        <type>extent_alloc_t</type> type and upon success returns a pointer to
+        <parameter>size</parameter> bytes of mapped memory on behalf of arena
+        <parameter>arena_ind</parameter> such that the extent's base address is
+        a multiple of <parameter>alignment</parameter>, as well as setting
+        <parameter>*zero</parameter> to indicate whether the extent is zeroed
+        and <parameter>*commit</parameter> to indicate whether the extent is
+        committed.  Upon error the function returns <constant>NULL</constant>
+        and leaves <parameter>*zero</parameter> and
+        <parameter>*commit</parameter> unmodified.  The
+        <parameter>size</parameter> parameter is always a multiple of the page
+        size.  The <parameter>alignment</parameter> parameter is always a power
+        of two at least as large as the page size.  Zeroing is mandatory if
+        <parameter>*zero</parameter> is true upon function entry.  Committing is
+        mandatory if <parameter>*commit</parameter> is true upon function entry.
+        If <parameter>new_addr</parameter> is not <constant>NULL</constant>, the
+        returned pointer must be <parameter>new_addr</parameter> on success or
+        <constant>NULL</constant> on error.  Committed memory may be committed
+        in absolute terms as on a system that does not overcommit, or in
+        implicit terms as on a system that overcommits and satisfies physical
+        memory needs on demand via soft page faults.  Note that replacing the
+        default extent allocation function makes the arena's <link
+        linkend="arena.i.dss"><mallctl>arena.&lt;i&gt;.dss</mallctl></link>
+        setting irrelevant.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_dalloc_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>
+        An extent deallocation function conforms to the
+        <type>extent_dalloc_t</type> type and deallocates an extent at given
+        <parameter>addr</parameter> and <parameter>size</parameter> with
+        <parameter>committed</parameter>/decommited memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>, returning false upon
+        success.  If the function returns true, this indicates opt-out from
+        deallocation; the virtual memory mapping associated with the extent
+        remains mapped, in the same commit state, and available for future use,
+        in which case it will be automatically retained for later reuse.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef void <function>(extent_destroy_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>
+        An extent destruction function conforms to the
+        <type>extent_destroy_t</type> type and unconditionally destroys an
+        extent at given <parameter>addr</parameter> and
+        <parameter>size</parameter> with
+        <parameter>committed</parameter>/decommited memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>.  This function may be
+        called to destroy retained extents during arena destruction (see <link
+        linkend="arena.i.destroy"><mallctl>arena.&lt;i&gt;.destroy</mallctl></link>).</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_commit_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>offset</parameter></paramdef>
+          <paramdef>size_t <parameter>length</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent commit function conforms to the
+        <type>extent_commit_t</type> type and commits zeroed physical memory to
+        back pages within an extent at given <parameter>addr</parameter> and
+        <parameter>size</parameter> at <parameter>offset</parameter> bytes,
+        extending for <parameter>length</parameter> on behalf of arena
+        <parameter>arena_ind</parameter>, returning false upon success.
+        Committed memory may be committed in absolute terms as on a system that
+        does not overcommit, or in implicit terms as on a system that
+        overcommits and satisfies physical memory needs on demand via soft page
+        faults. If the function returns true, this indicates insufficient
+        physical memory to satisfy the request.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_decommit_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>offset</parameter></paramdef>
+          <paramdef>size_t <parameter>length</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent decommit function conforms to the
+        <type>extent_decommit_t</type> type and decommits any physical memory
+        that is backing pages within an extent at given
+        <parameter>addr</parameter> and <parameter>size</parameter> at
+        <parameter>offset</parameter> bytes, extending for
+        <parameter>length</parameter> on behalf of arena
+        <parameter>arena_ind</parameter>, returning false upon success, in which
+        case the pages will be committed via the extent commit function before
+        being reused.  If the function returns true, this indicates opt-out from
+        decommit; the memory remains committed and available for future use, in
+        which case it will be automatically retained for later reuse.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_purge_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>offset</parameter></paramdef>
+          <paramdef>size_t <parameter>length</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent purge function conforms to the
+        <type>extent_purge_t</type> type and discards physical pages
+        within the virtual memory mapping associated with an extent at given
+        <parameter>addr</parameter> and <parameter>size</parameter> at
+        <parameter>offset</parameter> bytes, extending for
+        <parameter>length</parameter> on behalf of arena
+        <parameter>arena_ind</parameter>.  A lazy extent purge function (e.g.
+        implemented via
+        <function>madvise(<parameter>...</parameter><parameter><constant>MADV_FREE</constant></parameter>)</function>)
+        can delay purging indefinitely and leave the pages within the purged
+        virtual memory range in an indeterminite state, whereas a forced extent
+        purge function immediately purges, and the pages within the virtual
+        memory range will be zero-filled the next time they are accessed.  If
+        the function returns true, this indicates failure to purge.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_split_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>size_a</parameter></paramdef>
+          <paramdef>size_t <parameter>size_b</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent split function conforms to the
+        <type>extent_split_t</type> type and optionally splits an extent at
+        given <parameter>addr</parameter> and <parameter>size</parameter> into
+        two adjacent extents, the first of <parameter>size_a</parameter> bytes,
+        and the second of <parameter>size_b</parameter> bytes, operating on
+        <parameter>committed</parameter>/decommitted memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>, returning false upon
+        success.  If the function returns true, this indicates that the extent
+        remains unsplit and therefore should continue to be operated on as a
+        whole.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_merge_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr_a</parameter></paramdef>
+          <paramdef>size_t <parameter>size_a</parameter></paramdef>
+          <paramdef>void *<parameter>addr_b</parameter></paramdef>
+          <paramdef>size_t <parameter>size_b</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent merge function conforms to the
+        <type>extent_merge_t</type> type and optionally merges adjacent extents,
+        at given <parameter>addr_a</parameter> and <parameter>size_a</parameter>
+        with given <parameter>addr_b</parameter> and
+        <parameter>size_b</parameter> into one contiguous extent, operating on
+        <parameter>committed</parameter>/decommitted memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>, returning false upon
+        success.  If the function returns true, this indicates that the extents
+        remain distinct mappings and therefore should continue to be operated on
+        independently.</para>
+        </listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.narenas">
+        <term>
+          <mallctl>arenas.narenas</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Current limit on number of arenas.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.dirty_decay_ms">
+        <term>
+          <mallctl>arenas.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current default per-arena approximate time in
+        milliseconds from the creation of a set of unused dirty pages until an
+        equivalent set of unused dirty pages is purged and/or reused, used to
+        initialize <link
+        linkend="arena.i.dirty_decay_ms"><mallctl>arena.&lt;i&gt;.dirty_decay_ms</mallctl></link>
+        during arena creation.  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.muzzy_decay_ms">
+        <term>
+          <mallctl>arenas.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current default per-arena approximate time in
+        milliseconds from the creation of a set of unused muzzy pages until an
+        equivalent set of unused muzzy pages is purged and/or reused, used to
+        initialize <link
+        linkend="arena.i.muzzy_decay_ms"><mallctl>arena.&lt;i&gt;.muzzy_decay_ms</mallctl></link>
+        during arena creation.  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.quantum">
+        <term>
+          <mallctl>arenas.quantum</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Quantum size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.page">
+        <term>
+          <mallctl>arenas.page</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Page size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.tcache_max">
+        <term>
+          <mallctl>arenas.tcache_max</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum thread-cached size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.nbins">
+        <term>
+          <mallctl>arenas.nbins</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of bin size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.nhbins">
+        <term>
+          <mallctl>arenas.nhbins</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Total number of thread cache bin size
+        classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.bin.i.size">
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size supported by size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.bin.i.nregs">
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.nregs</mallctl>
+          (<type>uint32_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of regions per slab.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.bin.i.slab_size">
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.slab_size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of bytes per slab.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.nlextents">
+        <term>
+          <mallctl>arenas.nlextents</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Total number of large size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.lextent.i.size">
+        <term>
+          <mallctl>arenas.lextent.&lt;i&gt;.size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size supported by this large size
+        class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.create">
+        <term>
+          <mallctl>arenas.create</mallctl>
+          (<type>unsigned</type>, <type>extent_hooks_t *</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Explicitly create a new arena outside the range of
+        automatically managed arenas, with optionally specified extent hooks,
+        and return the new arena index.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.lookup">
+        <term>
+          <mallctl>arenas.lookup</mallctl>
+          (<type>unsigned</type>, <type>void*</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Index of the arena to which an allocation belongs to.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.thread_active_init">
+        <term>
+          <mallctl>prof.thread_active_init</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Control the initial setting for <link
+        linkend="thread.prof.active"><mallctl>thread.prof.active</mallctl></link>
+        in newly created threads.  See the <link
+        linkend="opt.prof_thread_active_init"><mallctl>opt.prof_thread_active_init</mallctl></link>
+        option for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.active">
+        <term>
+          <mallctl>prof.active</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Control whether sampling is currently active.  See the
+        <link
+        linkend="opt.prof_active"><mallctl>opt.prof_active</mallctl></link>
+        option for additional information, as well as the interrelated <link
+        linkend="thread.prof.active"><mallctl>thread.prof.active</mallctl></link>
+        mallctl.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.dump">
+        <term>
+          <mallctl>prof.dump</mallctl>
+          (<type>const char *</type>)
+          <literal>-w</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Dump a memory profile to the specified file, or if NULL
+        is specified, to a file according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.m&lt;mseq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the
+        <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.gdump">
+        <term>
+          <mallctl>prof.gdump</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>When enabled, trigger a memory profile dump every time
+        the total virtual memory exceeds the previous maximum.  Profiles are
+        dumped to files named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.u&lt;useq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.reset">
+        <term>
+          <mallctl>prof.reset</mallctl>
+          (<type>size_t</type>)
+          <literal>-w</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Reset all memory profile statistics, and optionally
+        update the sample rate (see <link
+        linkend="opt.lg_prof_sample"><mallctl>opt.lg_prof_sample</mallctl></link>
+        and <link
+        linkend="prof.lg_sample"><mallctl>prof.lg_sample</mallctl></link>).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.lg_sample">
+        <term>
+          <mallctl>prof.lg_sample</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Get the current sample rate (see <link
+        linkend="opt.lg_prof_sample"><mallctl>opt.lg_prof_sample</mallctl></link>).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.interval">
+        <term>
+          <mallctl>prof.interval</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average number of bytes allocated between
+        interval-based profile dumps.  See the
+        <link
+        linkend="opt.lg_prof_interval"><mallctl>opt.lg_prof_interval</mallctl></link>
+        option for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.allocated">
+        <term>
+          <mallctl>stats.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes allocated by the
+        application.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.active">
+        <term>
+          <mallctl>stats.active</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in active pages allocated by the
+        application.  This is a multiple of the page size, and greater than or
+        equal to <link
+        linkend="stats.allocated"><mallctl>stats.allocated</mallctl></link>.
+        This does not include <link linkend="stats.arenas.i.pdirty">
+        <mallctl>stats.arenas.&lt;i&gt;.pdirty</mallctl></link>,
+        <link linkend="stats.arenas.i.pmuzzy">
+        <mallctl>stats.arenas.&lt;i&gt;.pmuzzy</mallctl></link>, nor pages
+        entirely devoted to allocator metadata.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.metadata">
+        <term>
+          <mallctl>stats.metadata</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes dedicated to metadata, which
+        comprise base allocations used for bootstrap-sensitive allocator
+        metadata structures (see <link
+        linkend="stats.arenas.i.base"><mallctl>stats.arenas.&lt;i&gt;.base</mallctl></link>)
+        and internal allocations (see <link
+        linkend="stats.arenas.i.internal"><mallctl>stats.arenas.&lt;i&gt;.internal</mallctl></link>).
+        Transparent huge page (enabled with <link
+        linkend="opt.metadata_thp">opt.metadata_thp</link>) usage is not
+        considered.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.metadata_thp">
+        <term>
+          <mallctl>stats.metadata_thp</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of transparent huge pages (THP) used for
+        metadata.  See <link
+        linkend="stats.metadata"><mallctl>stats.metadata</mallctl></link> and
+        <link linkend="opt.metadata_thp">opt.metadata_thp</link>) for
+        details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.resident">
+        <term>
+          <mallctl>stats.resident</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Maximum number of bytes in physically resident data
+        pages mapped by the allocator, comprising all pages dedicated to
+        allocator metadata, pages backing active allocations, and unused dirty
+        pages.  This is a maximum rather than precise because pages may not
+        actually be physically resident if they correspond to demand-zeroed
+        virtual memory that has not yet been touched.  This is a multiple of the
+        page size, and is larger than <link
+        linkend="stats.active"><mallctl>stats.active</mallctl></link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mapped">
+        <term>
+          <mallctl>stats.mapped</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in active extents mapped by the
+        allocator.  This is larger than <link
+        linkend="stats.active"><mallctl>stats.active</mallctl></link>.  This
+        does not include inactive extents, even those that contain unused dirty
+        pages, which means that there is no strict ordering between this and
+        <link
+        linkend="stats.resident"><mallctl>stats.resident</mallctl></link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.retained">
+        <term>
+          <mallctl>stats.retained</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in virtual memory mappings that
+        were retained rather than being returned to the operating system via
+        e.g. <citerefentry><refentrytitle>munmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> or similar.  Retained virtual
+        memory is typically untouched, decommitted, or purged, so it has no
+        strongly associated physical memory (see <link
+        linkend="arena.i.extent_hooks">extent hooks</link> for details).
+        Retained memory is excluded from mapped memory statistics, e.g. <link
+        linkend="stats.mapped"><mallctl>stats.mapped</mallctl></link>.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.background_thread.num_threads">
+        <term>
+          <mallctl>stats.background_thread.num_threads</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Number of <link linkend="background_thread">background
+        threads</link> running currently.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.background_thread.num_runs">
+        <term>
+          <mallctl>stats.background_thread.num_runs</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Total number of runs from all <link
+        linkend="background_thread">background threads</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.background_thread.run_interval">
+        <term>
+          <mallctl>stats.background_thread.run_interval</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Average run interval in nanoseconds of <link
+        linkend="background_thread">background threads</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.ctl">
+        <term>
+          <mallctl>stats.mutexes.ctl.{counter};</mallctl>
+          (<type>counter specific type</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>ctl</varname> mutex (global
+        scope; mallctl related).  <mallctl>{counter}</mallctl> is one of the
+        counters below:</para>
+        <varlistentry id="mutex_counters">
+          <listitem><para><varname>num_ops</varname> (<type>uint64_t</type>):
+          Total number of lock acquisition operations on this mutex.</para>
+
+	  <para><varname>num_spin_acq</varname> (<type>uint64_t</type>): Number
+	  of times the mutex was spin-acquired.  When the mutex is currently
+	  locked and cannot be acquired immediately, a short period of
+	  spin-retry within jemalloc will be performed.  Acquired through spin
+	  generally means the contention was lightweight and not causing context
+	  switches.</para>
+
+	  <para><varname>num_wait</varname> (<type>uint64_t</type>): Number of
+	  times the mutex was wait-acquired, which means the mutex contention
+	  was not solved by spin-retry, and blocking operation was likely
+	  involved in order to acquire the mutex.  This event generally implies
+	  higher cost / longer delay, and should be investigated if it happens
+	  often.</para>
+
+	  <para><varname>max_wait_time</varname> (<type>uint64_t</type>):
+	  Maximum length of time in nanoseconds spent on a single wait-acquired
+	  lock operation.  Note that to avoid profiling overhead on the common
+	  path, this does not consider spin-acquired cases.</para>
+
+	  <para><varname>total_wait_time</varname> (<type>uint64_t</type>):
+	  Cumulative time in nanoseconds spent on wait-acquired lock operations.
+	  Similarly, spin-acquired cases are not considered.</para>
+
+	  <para><varname>max_num_thds</varname> (<type>uint32_t</type>): Maximum
+	  number of threads waiting on this mutex simultaneously.  Similarly,
+	  spin-acquired cases are not considered.</para>
+
+	  <para><varname>num_owner_switch</varname> (<type>uint64_t</type>):
+	  Number of times the current mutex owner is different from the previous
+	  one.  This event does not generally imply an issue; rather it is an
+	  indicator of how often the protected data are accessed by different
+	  threads.
+	  </para>
+	  </listitem>
+	</varlistentry>
+	</listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.background_thread">
+        <term>
+          <mallctl>stats.mutexes.background_thread.{counter}</mallctl>
+	  (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>background_thread</varname> mutex
+        (global scope; <link
+        linkend="background_thread"><mallctl>background_thread</mallctl></link>
+        related).  <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.prof">
+        <term>
+          <mallctl>stats.mutexes.prof.{counter}</mallctl>
+	  (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>prof</varname> mutex (global
+        scope; profiling related).  <mallctl>{counter}</mallctl> is one of the
+        counters in <link linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.reset">
+        <term>
+          <mallctl>stats.mutexes.reset</mallctl>
+	  (<type>void</type>) <literal>--</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Reset all mutex profile statistics, including global
+        mutexes, arena mutexes and bin mutexes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dss">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>dss (<citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry>) allocation precedence as
+        related to <citerefentry><refentrytitle>mmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> allocation.  See <link
+        linkend="opt.dss"><mallctl>opt.dss</mallctl></link> for details.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_decay_ms">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused dirty pages until an equivalent set of unused dirty pages
+        is purged and/or reused.  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_decay_ms">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused muzzy pages until an equivalent set of unused muzzy pages
+        is purged and/or reused.  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.nthreads">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.nthreads</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of threads currently assigned to
+        arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.uptime">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.uptime</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Time elapsed (in nanoseconds) since the arena was
+        created.  If &lt;i&gt; equals <constant>0</constant> or
+        <constant>MALLCTL_ARENAS_ALL</constant>, this is the uptime since malloc
+        initialization.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.pactive">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pactive</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages in active extents.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.pdirty">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pdirty</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages within unused extents that are
+        potentially dirty, and for which <function>madvise()</function> or
+        similar has not been called.  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for a description of dirty pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.pmuzzy">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pmuzzy</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages within unused extents that are muzzy.
+        See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for a description of muzzy pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mapped">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mapped</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of mapped bytes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.retained">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.retained</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of retained bytes.  See <link
+        linkend="stats.retained"><mallctl>stats.retained</mallctl></link> for
+        details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.extent_avail">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.extent_avail</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of allocated (but unused) extent structs in this
+	arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.base">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.base</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>
+        Number of bytes dedicated to bootstrap-sensitive allocator metadata
+        structures.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.internal">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.internal</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes dedicated to internal allocations.
+        Internal allocations differ from application-originated allocations in
+        that they are for internal use, and that they are omitted from heap
+        profiles.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.metadata_thp">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.metadata_thp</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of transparent huge pages (THP) used for
+        metadata.  See <link linkend="opt.metadata_thp">opt.metadata_thp</link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.resident">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.resident</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Maximum number of bytes in physically resident data
+        pages mapped by the arena, comprising all pages dedicated to allocator
+        metadata, pages backing active allocations, and unused dirty pages.
+        This is a maximum rather than precise because pages may not actually be
+        physically resident if they correspond to demand-zeroed virtual memory
+        that has not yet been touched.  This is a multiple of the page
+        size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_npurge">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_npurge</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of dirty page purge sweeps performed.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_nmadvise">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_nmadvise</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of <function>madvise()</function> or similar
+        calls made to purge dirty pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_purged">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_purged</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of dirty pages purged.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_npurge">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_npurge</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of muzzy page purge sweeps performed.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_nmadvise">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_nmadvise</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of <function>madvise()</function> or similar
+        calls made to purge muzzy pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_purged">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_purged</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of muzzy pages purged.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.allocated">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes currently allocated by small objects.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a small allocation was
+        requested from the arena's bins, whether to fill the relevant tcache if
+        <link linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is
+        enabled, or to directly satisfy an allocation request
+        otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a small allocation was
+        returned to the arena's bins, whether to flush the relevant tcache if
+        <link linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is
+        enabled, or to directly deallocate an allocation
+        otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        all bin size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nfills">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nfills</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache fills by all small size
+	classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nflushes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nflushes</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache flushes by all small size
+        classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.allocated">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes currently allocated by large objects.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent was allocated
+        from the arena, whether to fill the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly satisfy
+        an allocation request otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent was returned
+        to the arena, whether to flush the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly
+        deallocate an allocation otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        all large size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nfills">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nfills</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache fills by all large size
+	classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nflushes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nflushes</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache flushes by all large size
+        classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a bin region of the
+        corresponding size class was allocated from the arena, whether to fill
+        the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled, or
+        to directly satisfy an allocation request otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a bin region of the
+        corresponding size class was returned to the arena, whether to flush the
+        relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled, or
+        to directly deallocate an allocation otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        bin regions of the corresponding size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.curregs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curregs</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of regions for this size
+        class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nfills">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nfills</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Cumulative number of tcache fills.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nflushes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nflushes</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Cumulative number of tcache flushes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nslabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nslabs</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of slabs created.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nreslabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nreslabs</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times the current slab from which
+        to allocate changed.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.curslabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curslabs</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of slabs.</para></listitem>
+      </varlistentry>
+
+
+      <varlistentry id="stats.arenas.i.bins.j.nonfull_slabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nonfull_slabs</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of nonfull slabs.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.mutex">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.mutex.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on
+        <varname>arena.&lt;i&gt;.bins.&lt;j&gt;</varname> mutex (arena bin
+        scope; bin operation related).  <mallctl>{counter}</mallctl> is one of
+        the counters in <link linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.extents.n">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.extents.&lt;j&gt;.n{extent_type}</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Number of extents of the given type in this arena in
+	the bucket corresponding to page size index &lt;j&gt;. The extent type
+	is one of dirty, muzzy, or retained.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.extents.bytes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.extents.&lt;j&gt;.{extent_type}_bytes</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+	<listitem><para> Sum of the bytes managed by extents of the given type
+	in this arena in the bucket corresponding to page size index &lt;j&gt;.
+	The extent type is one of dirty, muzzy, or retained.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent of the
+        corresponding size class was allocated from the arena, whether to fill
+        the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly satisfy
+        an allocation request otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent of the
+        corresponding size class was returned to the arena, whether to flush the
+        relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly
+        deallocate an allocation otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        large extents of the corresponding size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.curlextents">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.curlextents</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of large allocations for this size class.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.large">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.large.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.large</varname>
+        mutex (arena scope; large allocation related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extent_avail">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extent_avail.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extent_avail
+        </varname> mutex (arena scope; extent avail related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extents_dirty">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extents_dirty.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extents_dirty
+        </varname> mutex (arena scope; dirty extents related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extents_muzzy">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extents_muzzy.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extents_muzzy
+        </varname> mutex (arena scope; muzzy extents related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extents_retained">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extents_retained.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extents_retained
+        </varname> mutex (arena scope; retained extents related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.decay_dirty">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.decay_dirty.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.decay_dirty
+        </varname> mutex (arena scope; decay for dirty pages related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.decay_muzzy">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.decay_muzzy.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.decay_muzzy
+        </varname> mutex (arena scope; decay for muzzy pages related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.base">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.base.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.base</varname>
+        mutex (arena scope; base allocator related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.tcache_list">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.tcache_list.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on
+        <varname>arena.&lt;i&gt;.tcache_list</varname> mutex (arena scope;
+        tcache to arena association related).  This mutex is expected to be
+        accessed less often.  <mallctl>{counter}</mallctl> is one of the
+        counters in <link linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+    </variablelist>
+  </refsect1>
+  <refsect1 id="heap_profile_format">
+    <title>HEAP PROFILE FORMAT</title>
+    <para>Although the heap profiling functionality was originally designed to
+    be compatible with the
+    <command>pprof</command> command that is developed as part of the <ulink
+    url="http://code.google.com/p/gperftools/">gperftools
+    package</ulink>, the addition of per thread heap profiling functionality
+    required a different heap profile format.  The <command>jeprof</command>
+    command is derived from <command>pprof</command>, with enhancements to
+    support the heap profile format described here.</para>
+
+    <para>In the following hypothetical heap profile, <constant>[...]</constant>
+    indicates elision for the sake of compactness.  <programlisting><![CDATA[
+heap_v2/524288
+  t*: 28106: 56637512 [0: 0]
+  [...]
+  t3: 352: 16777344 [0: 0]
+  [...]
+  t99: 17754: 29341640 [0: 0]
+  [...]
+@ 0x5f86da8 0x5f5a1dc [...] 0x29e4d4e 0xa200316 0xabb2988 [...]
+  t*: 13: 6688 [0: 0]
+  t3: 12: 6496 [0: ]
+  t99: 1: 192 [0: 0]
+[...]
+
+MAPPED_LIBRARIES:
+[...]]]></programlisting> The following matches the above heap profile, but most
+tokens are replaced with <constant>&lt;description&gt;</constant> to indicate
+descriptions of the corresponding fields.  <programlisting><![CDATA[
+<heap_profile_format_version>/<mean_sample_interval>
+  <aggregate>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+  [...]
+  <thread_3_aggregate>: <curobjs>: <curbytes>[<cumobjs>: <cumbytes>]
+  [...]
+  <thread_99_aggregate>: <curobjs>: <curbytes>[<cumobjs>: <cumbytes>]
+  [...]
+@ <top_frame> <frame> [...] <frame> <frame> <frame> [...]
+  <backtrace_aggregate>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+  <backtrace_thread_3>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+  <backtrace_thread_99>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+[...]
+
+MAPPED_LIBRARIES:
+</proc/<pid>/maps>]]></programlisting></para>
+  </refsect1>
+
+  <refsect1 id="debugging_malloc_problems">
+    <title>DEBUGGING MALLOC PROBLEMS</title>
+    <para>When debugging, it is a good idea to configure/build jemalloc with
+    the <option>--enable-debug</option> and <option>--enable-fill</option>
+    options, and recompile the program with suitable options and symbols for
+    debugger support.  When so configured, jemalloc incorporates a wide variety
+    of run-time assertions that catch application errors such as double-free,
+    write-after-free, etc.</para>
+
+    <para>Programs often accidentally depend on <quote>uninitialized</quote>
+    memory actually being filled with zero bytes.  Junk filling
+    (see the <link linkend="opt.junk"><mallctl>opt.junk</mallctl></link>
+    option) tends to expose such bugs in the form of obviously incorrect
+    results and/or coredumps.  Conversely, zero
+    filling (see the <link
+    linkend="opt.zero"><mallctl>opt.zero</mallctl></link> option) eliminates
+    the symptoms of such bugs.  Between these two options, it is usually
+    possible to quickly detect, diagnose, and eliminate such bugs.</para>
+
+    <para>This implementation does not provide much detail about the problems
+    it detects, because the performance impact for storing such information
+    would be prohibitive.</para>
+  </refsect1>
+  <refsect1 id="diagnostic_messages">
+    <title>DIAGNOSTIC MESSAGES</title>
+    <para>If any of the memory allocation/deallocation functions detect an
+    error or warning condition, a message will be printed to file descriptor
+    <constant>STDERR_FILENO</constant>.  Errors will result in the process
+    dumping core.  If the <link
+    linkend="opt.abort"><mallctl>opt.abort</mallctl></link> option is set, most
+    warnings are treated as errors.</para>
+
+    <para>The <varname>malloc_message</varname> variable allows the programmer
+    to override the function which emits the text strings forming the errors
+    and warnings if for some reason the <constant>STDERR_FILENO</constant> file
+    descriptor is not suitable for this.
+    <function>malloc_message()</function> takes the
+    <parameter>cbopaque</parameter> pointer argument that is
+    <constant>NULL</constant> unless overridden by the arguments in a call to
+    <function>malloc_stats_print()</function>, followed by a string
+    pointer.  Please note that doing anything which tries to allocate memory in
+    this function is likely to result in a crash or deadlock.</para>
+
+    <para>All messages are prefixed by
+    <quote><computeroutput>&lt;jemalloc&gt;: </computeroutput></quote>.</para>
+  </refsect1>
+  <refsect1 id="return_values">
+    <title>RETURN VALUES</title>
+    <refsect2>
+      <title>Standard API</title>
+      <para>The <function>malloc()</function> and
+      <function>calloc()</function> functions return a pointer to the
+      allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned and <varname>errno</varname> is set to
+      <errorname>ENOMEM</errorname>.</para>
+
+      <para>The <function>posix_memalign()</function> function
+      returns the value 0 if successful; otherwise it returns an error value.
+      The <function>posix_memalign()</function> function will fail
+      if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para>The <parameter>alignment</parameter> parameter is
+            not a power of 2 at least as large as
+            <code language="C">sizeof(<type>void *</type>)</code>.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOMEM</errorname></term>
+
+            <listitem><para>Memory allocation error.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>aligned_alloc()</function> function returns
+      a pointer to the allocated memory if successful; otherwise a
+      <constant>NULL</constant> pointer is returned and
+      <varname>errno</varname> is set.  The
+      <function>aligned_alloc()</function> function will fail if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para>The <parameter>alignment</parameter> parameter is
+            not a power of 2.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOMEM</errorname></term>
+
+            <listitem><para>Memory allocation error.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>realloc()</function> function returns a
+      pointer, possibly identical to <parameter>ptr</parameter>, to the
+      allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned, and <varname>errno</varname> is set to
+      <errorname>ENOMEM</errorname> if the error was the result of an
+      allocation failure.  The <function>realloc()</function>
+      function always leaves the original buffer intact when an error occurs.
+      </para>
+
+      <para>The <function>free()</function> function returns no
+      value.</para>
+    </refsect2>
+    <refsect2>
+      <title>Non-standard API</title>
+      <para>The <function>mallocx()</function> and
+      <function>rallocx()</function> functions return a pointer to
+      the allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned to indicate insufficient contiguous memory was
+      available to service the allocation request.  </para>
+
+      <para>The <function>xallocx()</function> function returns the
+      real size of the resulting resized allocation pointed to by
+      <parameter>ptr</parameter>, which is a value less than
+      <parameter>size</parameter> if the allocation could not be adequately
+      grown in place.  </para>
+
+      <para>The <function>sallocx()</function> function returns the
+      real size of the allocation pointed to by <parameter>ptr</parameter>.
+      </para>
+
+      <para>The <function>nallocx()</function> returns the real size
+      that would result from a successful equivalent
+      <function>mallocx()</function> function call, or zero if
+      insufficient memory is available to perform the size computation.  </para>
+
+      <para>The <function>mallctl()</function>,
+      <function>mallctlnametomib()</function>, and
+      <function>mallctlbymib()</function> functions return 0 on
+      success; otherwise they return an error value.  The functions will fail
+      if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para><parameter>newp</parameter> is not
+            <constant>NULL</constant>, and <parameter>newlen</parameter> is too
+            large or too small.  Alternatively, <parameter>*oldlenp</parameter>
+            is too large or too small; in this case as much data as possible
+            are read despite the error.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOENT</errorname></term>
+
+            <listitem><para><parameter>name</parameter> or
+            <parameter>mib</parameter> specifies an unknown/invalid
+            value.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EPERM</errorname></term>
+
+            <listitem><para>Attempt to read or write void value, or attempt to
+            write read-only value.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EAGAIN</errorname></term>
+
+            <listitem><para>A memory allocation failure
+            occurred.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EFAULT</errorname></term>
+
+            <listitem><para>An interface with side effects failed in some way
+            not directly related to <function>mallctl*()</function>
+            read/write processing.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>malloc_usable_size()</function> function
+      returns the usable size of the allocation pointed to by
+      <parameter>ptr</parameter>.  </para>
+    </refsect2>
+  </refsect1>
+  <refsect1 id="environment">
+    <title>ENVIRONMENT</title>
+    <para>The following environment variable affects the execution of the
+    allocation functions:
+      <variablelist>
+        <varlistentry>
+          <term><envar>MALLOC_CONF</envar></term>
+
+          <listitem><para>If the environment variable
+          <envar>MALLOC_CONF</envar> is set, the characters it contains
+          will be interpreted as options.</para></listitem>
+        </varlistentry>
+      </variablelist>
+    </para>
+  </refsect1>
+  <refsect1 id="examples">
+    <title>EXAMPLES</title>
+    <para>To dump core whenever a problem occurs:
+      <screen>ln -s 'abort:true' /etc/malloc.conf</screen>
+    </para>
+    <para>To specify in the source that only one arena should be automatically
+    created:
+      <programlisting language="C"><![CDATA[
+malloc_conf = "narenas:1";]]></programlisting></para>
+  </refsect1>
+  <refsect1 id="see_also">
+    <title>SEE ALSO</title>
+    <para><citerefentry><refentrytitle>madvise</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>utrace</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>alloca</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>atexit</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>getpagesize</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry></para>
+  </refsect1>
+  <refsect1 id="standards">
+    <title>STANDARDS</title>
+    <para>The <function>malloc()</function>,
+    <function>calloc()</function>,
+    <function>realloc()</function>, and
+    <function>free()</function> functions conform to ISO/IEC
+    9899:1990 (<quote>ISO C90</quote>).</para>
+
+    <para>The <function>posix_memalign()</function> function conforms
+    to IEEE Std 1003.1-2001 (<quote>POSIX.1</quote>).</para>
+  </refsect1>
+</refentry>
diff --git a/jemalloc/doc/manpages.xsl.in b/jemalloc/doc/manpages.xsl.in
new file mode 100644
index 0000000..88b2626
--- /dev/null
+++ b/jemalloc/doc/manpages.xsl.in
@@ -0,0 +1,4 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:import href="@XSLROOT@/manpages/docbook.xsl"/>
+  <xsl:import href="@abs_srcroot@doc/stylesheet.xsl"/>
+</xsl:stylesheet>
diff --git a/jemalloc/doc/stylesheet.xsl b/jemalloc/doc/stylesheet.xsl
new file mode 100644
index 0000000..619365d
--- /dev/null
+++ b/jemalloc/doc/stylesheet.xsl
@@ -0,0 +1,10 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:param name="funcsynopsis.style">ansi</xsl:param>
+  <xsl:param name="function.parens" select="0"/>
+  <xsl:template match="function">
+    <xsl:call-template name="inline.monoseq"/>
+  </xsl:template>
+  <xsl:template match="mallctl">
+    <quote><xsl:call-template name="inline.monoseq"/></quote>
+  </xsl:template>
+</xsl:stylesheet>
diff --git a/jemalloc/include/jemalloc/internal/arena_externs.h b/jemalloc/include/jemalloc/internal/arena_externs.h
new file mode 100644
index 0000000..a4523ae
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/arena_externs.h
@@ -0,0 +1,104 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_EXTERNS_H
+#define JEMALLOC_INTERNAL_ARENA_EXTERNS_H
+
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/stats.h"
+
+extern ssize_t opt_dirty_decay_ms;
+extern ssize_t opt_muzzy_decay_ms;
+
+extern percpu_arena_mode_t opt_percpu_arena;
+extern const char *percpu_arena_mode_names[];
+
+extern const uint64_t h_steps[SMOOTHSTEP_NSTEPS];
+extern malloc_mutex_t arenas_lock;
+
+extern size_t opt_oversize_threshold;
+extern size_t oversize_threshold;
+
+void arena_basic_stats_merge(tsdn_t *tsdn, arena_t *arena,
+    unsigned *nthreads, const char **dss, ssize_t *dirty_decay_ms,
+    ssize_t *muzzy_decay_ms, size_t *nactive, size_t *ndirty, size_t *nmuzzy);
+void arena_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
+    const char **dss, ssize_t *dirty_decay_ms, ssize_t *muzzy_decay_ms,
+    size_t *nactive, size_t *ndirty, size_t *nmuzzy, arena_stats_t *astats,
+    bin_stats_t *bstats, arena_stats_large_t *lstats,
+    arena_stats_extents_t *estats);
+void arena_extents_dirty_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+#ifdef JEMALLOC_JET
+size_t arena_slab_regind(extent_t *slab, szind_t binind, const void *ptr);
+#endif
+extent_t *arena_extent_alloc_large(tsdn_t *tsdn, arena_t *arena,
+    size_t usize, size_t alignment, bool *zero);
+void arena_extent_dalloc_large_prep(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent);
+void arena_extent_ralloc_large_shrink(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent, size_t oldsize);
+void arena_extent_ralloc_large_expand(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent, size_t oldsize);
+ssize_t arena_dirty_decay_ms_get(arena_t *arena);
+bool arena_dirty_decay_ms_set(tsdn_t *tsdn, arena_t *arena, ssize_t decay_ms);
+ssize_t arena_muzzy_decay_ms_get(arena_t *arena);
+bool arena_muzzy_decay_ms_set(tsdn_t *tsdn, arena_t *arena, ssize_t decay_ms);
+void arena_decay(tsdn_t *tsdn, arena_t *arena, bool is_background_thread,
+    bool all);
+void arena_reset(tsd_t *tsd, arena_t *arena);
+void arena_destroy(tsd_t *tsd, arena_t *arena);
+void arena_tcache_fill_small(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, uint64_t prof_accumbytes);
+void arena_alloc_junk_small(void *ptr, const bin_info_t *bin_info,
+    bool zero);
+
+typedef void (arena_dalloc_junk_small_t)(void *, const bin_info_t *);
+extern arena_dalloc_junk_small_t *JET_MUTABLE arena_dalloc_junk_small;
+
+void *arena_malloc_hard(tsdn_t *tsdn, arena_t *arena, size_t size,
+    szind_t ind, bool zero);
+void *arena_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache);
+void arena_prof_promote(tsdn_t *tsdn, void *ptr, size_t usize);
+void arena_dalloc_promoted(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    bool slow_path);
+void arena_dalloc_bin_junked_locked(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, extent_t *extent, void *ptr);
+void arena_dalloc_small(tsdn_t *tsdn, void *ptr);
+bool arena_ralloc_no_move(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero, size_t *newsize);
+void *arena_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t oldsize,
+    size_t size, size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args);
+dss_prec_t arena_dss_prec_get(arena_t *arena);
+bool arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec);
+ssize_t arena_dirty_decay_ms_default_get(void);
+bool arena_dirty_decay_ms_default_set(ssize_t decay_ms);
+ssize_t arena_muzzy_decay_ms_default_get(void);
+bool arena_muzzy_decay_ms_default_set(ssize_t decay_ms);
+bool arena_retain_grow_limit_get_set(tsd_t *tsd, arena_t *arena,
+    size_t *old_limit, size_t *new_limit);
+unsigned arena_nthreads_get(arena_t *arena, bool internal);
+void arena_nthreads_inc(arena_t *arena, bool internal);
+void arena_nthreads_dec(arena_t *arena, bool internal);
+size_t arena_extent_sn_next(arena_t *arena);
+arena_t *arena_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+bool arena_init_huge(void);
+bool arena_is_huge(unsigned arena_ind);
+arena_t *arena_choose_huge(tsd_t *tsd);
+bin_t *arena_bin_choose_lock(tsdn_t *tsdn, arena_t *arena, szind_t binind,
+    unsigned *binshard);
+void arena_boot(sc_data_t *sc_data);
+void arena_prefork0(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork1(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork2(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork3(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork4(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork5(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork6(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork7(tsdn_t *tsdn, arena_t *arena);
+void arena_postfork_parent(tsdn_t *tsdn, arena_t *arena);
+void arena_postfork_child(tsdn_t *tsdn, arena_t *arena);
+
+#endif /* JEMALLOC_INTERNAL_ARENA_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/arena_inlines_a.h b/jemalloc/include/jemalloc/internal/arena_inlines_a.h
new file mode 100644
index 0000000..9abf7f6
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/arena_inlines_a.h
@@ -0,0 +1,57 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_INLINES_A_H
+#define JEMALLOC_INTERNAL_ARENA_INLINES_A_H
+
+static inline unsigned
+arena_ind_get(const arena_t *arena) {
+	return base_ind_get(arena->base);
+}
+
+static inline void
+arena_internal_add(arena_t *arena, size_t size) {
+	atomic_fetch_add_zu(&arena->stats.internal, size, ATOMIC_RELAXED);
+}
+
+static inline void
+arena_internal_sub(arena_t *arena, size_t size) {
+	atomic_fetch_sub_zu(&arena->stats.internal, size, ATOMIC_RELAXED);
+}
+
+static inline size_t
+arena_internal_get(arena_t *arena) {
+	return atomic_load_zu(&arena->stats.internal, ATOMIC_RELAXED);
+}
+
+static inline bool
+arena_prof_accum(tsdn_t *tsdn, arena_t *arena, uint64_t accumbytes) {
+	cassert(config_prof);
+
+	if (likely(prof_interval == 0 || !prof_active_get_unlocked())) {
+		return false;
+	}
+
+	return prof_accum_add(tsdn, &arena->prof_accum, accumbytes);
+}
+
+static inline void
+percpu_arena_update(tsd_t *tsd, unsigned cpu) {
+	assert(have_percpu_arena);
+	arena_t *oldarena = tsd_arena_get(tsd);
+	assert(oldarena != NULL);
+	unsigned oldind = arena_ind_get(oldarena);
+
+	if (oldind != cpu) {
+		unsigned newind = cpu;
+		arena_t *newarena = arena_get(tsd_tsdn(tsd), newind, true);
+		assert(newarena != NULL);
+
+		/* Set new arena/tcache associations. */
+		arena_migrate(tsd, oldind, newind);
+		tcache_t *tcache = tcache_get(tsd);
+		if (tcache != NULL) {
+			tcache_arena_reassociate(tsd_tsdn(tsd), tcache,
+			    newarena);
+		}
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_INLINES_A_H */
diff --git a/jemalloc/include/jemalloc/internal/arena_inlines_b.h b/jemalloc/include/jemalloc/internal/arena_inlines_b.h
new file mode 100644
index 0000000..dd92657
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/arena_inlines_b.h
@@ -0,0 +1,427 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_INLINES_B_H
+#define JEMALLOC_INTERNAL_ARENA_INLINES_B_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/ticker.h"
+
+JEMALLOC_ALWAYS_INLINE bool
+arena_has_default_hooks(arena_t *arena) {
+	return (extent_hooks_get(arena) == &extent_hooks_default);
+}
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+arena_choose_maybe_huge(tsd_t *tsd, arena_t *arena, size_t size) {
+	if (arena != NULL) {
+		return arena;
+	}
+
+	/*
+	 * For huge allocations, use the dedicated huge arena if both are true:
+	 * 1) is using auto arena selection (i.e. arena == NULL), and 2) the
+	 * thread is not assigned to a manual arena.
+	 */
+	if (unlikely(size >= oversize_threshold)) {
+		arena_t *tsd_arena = tsd_arena_get(tsd);
+		if (tsd_arena == NULL || arena_is_auto(tsd_arena)) {
+			return arena_choose_huge(tsd);
+		}
+	}
+
+	return arena_choose(tsd, NULL);
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+arena_prof_tctx_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	/* Static check. */
+	if (alloc_ctx == NULL) {
+		const extent_t *extent = iealloc(tsdn, ptr);
+		if (unlikely(!extent_slab_get(extent))) {
+			return large_prof_tctx_get(tsdn, extent);
+		}
+	} else {
+		if (unlikely(!alloc_ctx->slab)) {
+			return large_prof_tctx_get(tsdn, iealloc(tsdn, ptr));
+		}
+	}
+	return (prof_tctx_t *)(uintptr_t)1U;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_prof_tctx_set(tsdn_t *tsdn, const void *ptr, size_t usize,
+    alloc_ctx_t *alloc_ctx, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	/* Static check. */
+	if (alloc_ctx == NULL) {
+		extent_t *extent = iealloc(tsdn, ptr);
+		if (unlikely(!extent_slab_get(extent))) {
+			large_prof_tctx_set(tsdn, extent, tctx);
+		}
+	} else {
+		if (unlikely(!alloc_ctx->slab)) {
+			large_prof_tctx_set(tsdn, iealloc(tsdn, ptr), tctx);
+		}
+	}
+}
+
+static inline void
+arena_prof_tctx_reset(tsdn_t *tsdn, const void *ptr, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	assert(!extent_slab_get(extent));
+
+	large_prof_tctx_reset(tsdn, extent);
+}
+
+JEMALLOC_ALWAYS_INLINE nstime_t
+arena_prof_alloc_time_get(tsdn_t *tsdn, const void *ptr,
+    alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	/*
+	 * Unlike arena_prof_prof_tctx_{get, set}, we only call this once we're
+	 * sure we have a sampled allocation.
+	 */
+	assert(!extent_slab_get(extent));
+	return large_prof_alloc_time_get(extent);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_prof_alloc_time_set(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx,
+    nstime_t t) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	assert(!extent_slab_get(extent));
+	large_prof_alloc_time_set(extent, t);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_ticks(tsdn_t *tsdn, arena_t *arena, unsigned nticks) {
+	tsd_t *tsd;
+	ticker_t *decay_ticker;
+
+	if (unlikely(tsdn_null(tsdn))) {
+		return;
+	}
+	tsd = tsdn_tsd(tsdn);
+	decay_ticker = decay_ticker_get(tsd, arena_ind_get(arena));
+	if (unlikely(decay_ticker == NULL)) {
+		return;
+	}
+	if (unlikely(ticker_ticks(decay_ticker, nticks))) {
+		arena_decay(tsdn, arena, false, false);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_tick(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_assert_not_owner(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_assert_not_owner(tsdn, &arena->decay_muzzy.mtx);
+
+	arena_decay_ticks(tsdn, arena, 1);
+}
+
+/* Purge a single extent to retained / unmapped directly. */
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_extent(tsdn_t *tsdn,arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extent_t *extent) {
+	size_t extent_size = extent_size_get(extent);
+	extent_dalloc_wrapper(tsdn, arena,
+	    r_extent_hooks, extent);
+	if (config_stats) {
+		/* Update stats accordingly. */
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_stats_add_u64(tsdn, &arena->stats,
+		    &arena->decay_dirty.stats->nmadvise, 1);
+		arena_stats_add_u64(tsdn, &arena->stats,
+		    &arena->decay_dirty.stats->purged, extent_size >> LG_PAGE);
+		arena_stats_sub_zu(tsdn, &arena->stats, &arena->stats.mapped,
+		    extent_size);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+arena_malloc(tsdn_t *tsdn, arena_t *arena, size_t size, szind_t ind, bool zero,
+    tcache_t *tcache, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+
+	if (likely(tcache != NULL)) {
+		if (likely(size <= SC_SMALL_MAXCLASS)) {
+			return tcache_alloc_small(tsdn_tsd(tsdn), arena,
+			    tcache, size, ind, zero, slow_path);
+		}
+		if (likely(size <= tcache_maxclass)) {
+			return tcache_alloc_large(tsdn_tsd(tsdn), arena,
+			    tcache, size, ind, zero, slow_path);
+		}
+		/* (size > tcache_maxclass) case falls through. */
+		assert(size > tcache_maxclass);
+	}
+
+	return arena_malloc_hard(tsdn, arena, size, ind, zero);
+}
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+arena_aalloc(tsdn_t *tsdn, const void *ptr) {
+	return extent_arena_get(iealloc(tsdn, ptr));
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_salloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	szind_t szind = rtree_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+	assert(szind != SC_NSIZES);
+
+	return sz_index2size(szind);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_vsalloc(tsdn_t *tsdn, const void *ptr) {
+	/*
+	 * Return 0 if ptr is not within an extent managed by jemalloc.  This
+	 * function has two extra costs relative to isalloc():
+	 * - The rtree calls cannot claim to be dependent lookups, which induces
+	 *   rtree lookup load dependencies.
+	 * - The lookup may fail, so there is an extra branch to check for
+	 *   failure.
+	 */
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent;
+	szind_t szind;
+	if (rtree_extent_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, false, &extent, &szind)) {
+		return 0;
+	}
+
+	if (extent == NULL) {
+		return 0;
+	}
+	assert(extent_state_get(extent) == extent_state_active);
+	/* Only slab members should be looked up via interior pointers. */
+	assert(extent_addr_get(extent) == ptr || extent_slab_get(extent));
+
+	assert(szind != SC_NSIZES);
+
+	return sz_index2size(szind);
+}
+
+static inline void
+arena_dalloc_large_no_tcache(tsdn_t *tsdn, void *ptr, szind_t szind) {
+	if (config_prof && unlikely(szind < SC_NBINS)) {
+		arena_dalloc_promoted(tsdn, ptr, NULL, true);
+	} else {
+		extent_t *extent = iealloc(tsdn, ptr);
+		large_dalloc(tsdn, extent);
+	}
+}
+
+static inline void
+arena_dalloc_no_tcache(tsdn_t *tsdn, void *ptr) {
+	assert(ptr != NULL);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	szind_t szind;
+	bool slab;
+	rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
+	    true, &szind, &slab);
+
+	if (config_debug) {
+		extent_t *extent = rtree_extent_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(szind < SC_NSIZES);
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		arena_dalloc_small(tsdn, ptr);
+	} else {
+		arena_dalloc_large_no_tcache(tsdn, ptr, szind);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_dalloc_large(tsdn_t *tsdn, void *ptr, tcache_t *tcache, szind_t szind,
+    bool slow_path) {
+	if (szind < nhbins) {
+		if (config_prof && unlikely(szind < SC_NBINS)) {
+			arena_dalloc_promoted(tsdn, ptr, tcache, slow_path);
+		} else {
+			tcache_dalloc_large(tsdn_tsd(tsdn), tcache, ptr, szind,
+			    slow_path);
+		}
+	} else {
+		extent_t *extent = iealloc(tsdn, ptr);
+		large_dalloc(tsdn, extent);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_dalloc(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+	assert(ptr != NULL);
+
+	if (unlikely(tcache == NULL)) {
+		arena_dalloc_no_tcache(tsdn, ptr);
+		return;
+	}
+
+	szind_t szind;
+	bool slab;
+	rtree_ctx_t *rtree_ctx;
+	if (alloc_ctx != NULL) {
+		szind = alloc_ctx->szind;
+		slab = alloc_ctx->slab;
+		assert(szind != SC_NSIZES);
+	} else {
+		rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+	}
+
+	if (config_debug) {
+		rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		extent_t *extent = rtree_extent_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(szind < SC_NSIZES);
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		tcache_dalloc_small(tsdn_tsd(tsdn), tcache, ptr, szind,
+		    slow_path);
+	} else {
+		arena_dalloc_large(tsdn, ptr, tcache, szind, slow_path);
+	}
+}
+
+static inline void
+arena_sdalloc_no_tcache(tsdn_t *tsdn, void *ptr, size_t size) {
+	assert(ptr != NULL);
+	assert(size <= SC_LARGE_MAXCLASS);
+
+	szind_t szind;
+	bool slab;
+	if (!config_prof || !opt_prof) {
+		/*
+		 * There is no risk of being confused by a promoted sampled
+		 * object, so base szind and slab on the given size.
+		 */
+		szind = sz_size2index(size);
+		slab = (szind < SC_NBINS);
+	}
+
+	if ((config_prof && opt_prof) || config_debug) {
+		rtree_ctx_t rtree_ctx_fallback;
+		rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+		    &rtree_ctx_fallback);
+
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+
+		assert(szind == sz_size2index(size));
+		assert((config_prof && opt_prof) || slab == (szind < SC_NBINS));
+
+		if (config_debug) {
+			extent_t *extent = rtree_extent_read(tsdn,
+			    &extents_rtree, rtree_ctx, (uintptr_t)ptr, true);
+			assert(szind == extent_szind_get(extent));
+			assert(slab == extent_slab_get(extent));
+		}
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		arena_dalloc_small(tsdn, ptr);
+	} else {
+		arena_dalloc_large_no_tcache(tsdn, ptr, szind);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_sdalloc(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+	assert(ptr != NULL);
+	assert(size <= SC_LARGE_MAXCLASS);
+
+	if (unlikely(tcache == NULL)) {
+		arena_sdalloc_no_tcache(tsdn, ptr, size);
+		return;
+	}
+
+	szind_t szind;
+	bool slab;
+	alloc_ctx_t local_ctx;
+	if (config_prof && opt_prof) {
+		if (alloc_ctx == NULL) {
+			/* Uncommon case and should be a static check. */
+			rtree_ctx_t rtree_ctx_fallback;
+			rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+			    &rtree_ctx_fallback);
+			rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+			    (uintptr_t)ptr, true, &local_ctx.szind,
+			    &local_ctx.slab);
+			assert(local_ctx.szind == sz_size2index(size));
+			alloc_ctx = &local_ctx;
+		}
+		slab = alloc_ctx->slab;
+		szind = alloc_ctx->szind;
+	} else {
+		/*
+		 * There is no risk of being confused by a promoted sampled
+		 * object, so base szind and slab on the given size.
+		 */
+		szind = sz_size2index(size);
+		slab = (szind < SC_NBINS);
+	}
+
+	if (config_debug) {
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+		extent_t *extent = rtree_extent_read(tsdn,
+		    &extents_rtree, rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		tcache_dalloc_small(tsdn_tsd(tsdn), tcache, ptr, szind,
+		    slow_path);
+	} else {
+		arena_dalloc_large(tsdn, ptr, tcache, szind, slow_path);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_INLINES_B_H */
diff --git a/jemalloc/include/jemalloc/internal/arena_stats.h b/jemalloc/include/jemalloc/internal/arena_stats.h
new file mode 100644
index 0000000..23949ed
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/arena_stats.h
@@ -0,0 +1,271 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STATS_H
+#define JEMALLOC_INTERNAL_ARENA_STATS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/sc.h"
+
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+/*
+ * In those architectures that support 64-bit atomics, we use atomic updates for
+ * our 64-bit values.  Otherwise, we use a plain uint64_t and synchronize
+ * externally.
+ */
+#ifdef JEMALLOC_ATOMIC_U64
+typedef atomic_u64_t arena_stats_u64_t;
+#else
+/* Must hold the arena stats mutex while reading atomically. */
+typedef uint64_t arena_stats_u64_t;
+#endif
+
+typedef struct arena_stats_large_s arena_stats_large_t;
+struct arena_stats_large_s {
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the arena.
+	 */
+	arena_stats_u64_t	nmalloc;
+	arena_stats_u64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to this size class.
+	 * This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	arena_stats_u64_t	nrequests; /* Partially derived. */
+	/*
+	 * Number of tcache fills / flushes for large (similarly, periodically
+	 * merged).  Note that there is no large tcache batch-fill currently
+	 * (i.e. only fill 1 at a time); however flush may be batched.
+	 */
+	arena_stats_u64_t	nfills; /* Partially derived. */
+	arena_stats_u64_t	nflushes; /* Partially derived. */
+
+	/* Current number of allocations of this size class. */
+	size_t		curlextents; /* Derived. */
+};
+
+typedef struct arena_stats_decay_s arena_stats_decay_t;
+struct arena_stats_decay_s {
+	/* Total number of purge sweeps. */
+	arena_stats_u64_t	npurge;
+	/* Total number of madvise calls made. */
+	arena_stats_u64_t	nmadvise;
+	/* Total number of pages purged. */
+	arena_stats_u64_t	purged;
+};
+
+typedef struct arena_stats_extents_s arena_stats_extents_t;
+struct arena_stats_extents_s {
+	/*
+	 * Stats for a given index in the range [0, SC_NPSIZES] in an extents_t.
+	 * We track both bytes and # of extents: two extents in the same bucket
+	 * may have different sizes if adjacent size classes differ by more than
+	 * a page, so bytes cannot always be derived from # of extents.
+	 */
+	atomic_zu_t ndirty;
+	atomic_zu_t dirty_bytes;
+	atomic_zu_t nmuzzy;
+	atomic_zu_t muzzy_bytes;
+	atomic_zu_t nretained;
+	atomic_zu_t retained_bytes;
+};
+
+/*
+ * Arena stats.  Note that fields marked "derived" are not directly maintained
+ * within the arena code; rather their values are derived during stats merge
+ * requests.
+ */
+typedef struct arena_stats_s arena_stats_t;
+struct arena_stats_s {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_t		mtx;
+#endif
+
+	/* Number of bytes currently mapped, excluding retained memory. */
+	atomic_zu_t		mapped; /* Partially derived. */
+
+	/*
+	 * Number of unused virtual memory bytes currently retained.  Retained
+	 * bytes are technically mapped (though always decommitted or purged),
+	 * but they are excluded from the mapped statistic (above).
+	 */
+	atomic_zu_t		retained; /* Derived. */
+
+	/* Number of extent_t structs allocated by base, but not being used. */
+	atomic_zu_t		extent_avail;
+
+	arena_stats_decay_t	decay_dirty;
+	arena_stats_decay_t	decay_muzzy;
+
+	atomic_zu_t		base; /* Derived. */
+	atomic_zu_t		internal;
+	atomic_zu_t		resident; /* Derived. */
+	atomic_zu_t		metadata_thp;
+
+	atomic_zu_t		allocated_large; /* Derived. */
+	arena_stats_u64_t	nmalloc_large; /* Derived. */
+	arena_stats_u64_t	ndalloc_large; /* Derived. */
+	arena_stats_u64_t	nfills_large; /* Derived. */
+	arena_stats_u64_t	nflushes_large; /* Derived. */
+	arena_stats_u64_t	nrequests_large; /* Derived. */
+
+	/* VM space had to be leaked (undocumented).  Normally 0. */
+	atomic_zu_t		abandoned_vm;
+
+	/* Number of bytes cached in tcache associated with this arena. */
+	atomic_zu_t		tcache_bytes; /* Derived. */
+
+	mutex_prof_data_t mutex_prof_data[mutex_prof_num_arena_mutexes];
+
+	/* One element for each large size class. */
+	arena_stats_large_t	lstats[SC_NSIZES - SC_NBINS];
+
+	/* Arena uptime. */
+	nstime_t		uptime;
+};
+
+static inline bool
+arena_stats_init(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+	if (config_debug) {
+		for (size_t i = 0; i < sizeof(arena_stats_t); i++) {
+			assert(((char *)arena_stats)[i] == 0);
+		}
+	}
+#ifndef JEMALLOC_ATOMIC_U64
+	if (malloc_mutex_init(&arena_stats->mtx, "arena_stats",
+	    WITNESS_RANK_ARENA_STATS, malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+#endif
+	/* Memory is zeroed, so there is no need to clear stats. */
+	return false;
+}
+
+static inline void
+arena_stats_lock(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_lock(tsdn, &arena_stats->mtx);
+#endif
+}
+
+static inline void
+arena_stats_unlock(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_unlock(tsdn, &arena_stats->mtx);
+#endif
+}
+
+static inline uint64_t
+arena_stats_read_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_u64(p, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	return *p;
+#endif
+}
+
+static inline void
+arena_stats_add_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p, uint64_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	atomic_fetch_add_u64(p, x, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	*p += x;
+#endif
+}
+
+static inline void
+arena_stats_sub_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p, uint64_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	uint64_t r = atomic_fetch_sub_u64(p, x, ATOMIC_RELAXED);
+	assert(r - x <= r);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	*p -= x;
+	assert(*p + x >= *p);
+#endif
+}
+
+/*
+ * Non-atomically sets *dst += src.  *dst needs external synchronization.
+ * This lets us avoid the cost of a fetch_add when its unnecessary (note that
+ * the types here are atomic).
+ */
+static inline void
+arena_stats_accum_u64(arena_stats_u64_t *dst, uint64_t src) {
+#ifdef JEMALLOC_ATOMIC_U64
+	uint64_t cur_dst = atomic_load_u64(dst, ATOMIC_RELAXED);
+	atomic_store_u64(dst, src + cur_dst, ATOMIC_RELAXED);
+#else
+	*dst += src;
+#endif
+}
+
+static inline size_t
+arena_stats_read_zu(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    atomic_zu_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_zu(p, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	return atomic_load_zu(p, ATOMIC_RELAXED);
+#endif
+}
+
+static inline void
+arena_stats_add_zu(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    atomic_zu_t *p, size_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	atomic_fetch_add_zu(p, x, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	size_t cur = atomic_load_zu(p, ATOMIC_RELAXED);
+	atomic_store_zu(p, cur + x, ATOMIC_RELAXED);
+#endif
+}
+
+static inline void
+arena_stats_sub_zu(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    atomic_zu_t *p, size_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	size_t r = atomic_fetch_sub_zu(p, x, ATOMIC_RELAXED);
+	assert(r - x <= r);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	size_t cur = atomic_load_zu(p, ATOMIC_RELAXED);
+	atomic_store_zu(p, cur - x, ATOMIC_RELAXED);
+#endif
+}
+
+/* Like the _u64 variant, needs an externally synchronized *dst. */
+static inline void
+arena_stats_accum_zu(atomic_zu_t *dst, size_t src) {
+	size_t cur_dst = atomic_load_zu(dst, ATOMIC_RELAXED);
+	atomic_store_zu(dst, src + cur_dst, ATOMIC_RELAXED);
+}
+
+static inline void
+arena_stats_large_flush_nrequests_add(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    szind_t szind, uint64_t nrequests) {
+	arena_stats_lock(tsdn, arena_stats);
+	arena_stats_large_t *lstats = &arena_stats->lstats[szind - SC_NBINS];
+	arena_stats_add_u64(tsdn, arena_stats, &lstats->nrequests, nrequests);
+	arena_stats_add_u64(tsdn, arena_stats, &lstats->nflushes, 1);
+	arena_stats_unlock(tsdn, arena_stats);
+}
+
+static inline void
+arena_stats_mapped_add(tsdn_t *tsdn, arena_stats_t *arena_stats, size_t size) {
+	arena_stats_lock(tsdn, arena_stats);
+	arena_stats_add_zu(tsdn, arena_stats, &arena_stats->mapped, size);
+	arena_stats_unlock(tsdn, arena_stats);
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STATS_H */
diff --git a/jemalloc/include/jemalloc/internal/arena_structs_a.h b/jemalloc/include/jemalloc/internal/arena_structs_a.h
new file mode 100644
index 0000000..46aa77c
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/arena_structs_a.h
@@ -0,0 +1,11 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H
+#define JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H
+
+#include "jemalloc/internal/bitmap.h"
+
+struct arena_slab_data_s {
+	/* Per region allocated/deallocated bitmap. */
+	bitmap_t	bitmap[BITMAP_GROUPS_MAX];
+};
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H */
diff --git a/jemalloc/include/jemalloc/internal/arena_structs_b.h b/jemalloc/include/jemalloc/internal/arena_structs_b.h
new file mode 100644
index 0000000..eeab57f
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/arena_structs_b.h
@@ -0,0 +1,232 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H
+#define JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H
+
+#include "jemalloc/internal/arena_stats.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/smoothstep.h"
+#include "jemalloc/internal/ticker.h"
+
+struct arena_decay_s {
+	/* Synchronizes all non-atomic fields. */
+	malloc_mutex_t		mtx;
+	/*
+	 * True if a thread is currently purging the extents associated with
+	 * this decay structure.
+	 */
+	bool			purging;
+	/*
+	 * Approximate time in milliseconds from the creation of a set of unused
+	 * dirty pages until an equivalent set of unused dirty pages is purged
+	 * and/or reused.
+	 */
+	atomic_zd_t		time_ms;
+	/* time / SMOOTHSTEP_NSTEPS. */
+	nstime_t		interval;
+	/*
+	 * Time at which the current decay interval logically started.  We do
+	 * not actually advance to a new epoch until sometime after it starts
+	 * because of scheduling and computation delays, and it is even possible
+	 * to completely skip epochs.  In all cases, during epoch advancement we
+	 * merge all relevant activity into the most recently recorded epoch.
+	 */
+	nstime_t		epoch;
+	/* Deadline randomness generator. */
+	uint64_t		jitter_state;
+	/*
+	 * Deadline for current epoch.  This is the sum of interval and per
+	 * epoch jitter which is a uniform random variable in [0..interval).
+	 * Epochs always advance by precise multiples of interval, but we
+	 * randomize the deadline to reduce the likelihood of arenas purging in
+	 * lockstep.
+	 */
+	nstime_t		deadline;
+	/*
+	 * Number of unpurged pages at beginning of current epoch.  During epoch
+	 * advancement we use the delta between arena->decay_*.nunpurged and
+	 * extents_npages_get(&arena->extents_*) to determine how many dirty
+	 * pages, if any, were generated.
+	 */
+	size_t			nunpurged;
+	/*
+	 * Trailing log of how many unused dirty pages were generated during
+	 * each of the past SMOOTHSTEP_NSTEPS decay epochs, where the last
+	 * element is the most recent epoch.  Corresponding epoch times are
+	 * relative to epoch.
+	 */
+	size_t			backlog[SMOOTHSTEP_NSTEPS];
+
+	/*
+	 * Pointer to associated stats.  These stats are embedded directly in
+	 * the arena's stats due to how stats structures are shared between the
+	 * arena and ctl code.
+	 *
+	 * Synchronization: Same as associated arena's stats field. */
+	arena_stats_decay_t	*stats;
+	/* Peak number of pages in associated extents.  Used for debug only. */
+	uint64_t		ceil_npages;
+};
+
+struct arena_s {
+	/*
+	 * Number of threads currently assigned to this arena.  Each thread has
+	 * two distinct assignments, one for application-serving allocation, and
+	 * the other for internal metadata allocation.  Internal metadata must
+	 * not be allocated from arenas explicitly created via the arenas.create
+	 * mallctl, because the arena.<i>.reset mallctl indiscriminately
+	 * discards all allocations for the affected arena.
+	 *
+	 *   0: Application allocation.
+	 *   1: Internal metadata allocation.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_u_t		nthreads[2];
+
+	/* Next bin shard for binding new threads. Synchronization: atomic. */
+	atomic_u_t		binshard_next;
+
+	/*
+	 * When percpu_arena is enabled, to amortize the cost of reading /
+	 * updating the current CPU id, track the most recent thread accessing
+	 * this arena, and only read CPU if there is a mismatch.
+	 */
+	tsdn_t		*last_thd;
+
+	/* Synchronization: internal. */
+	arena_stats_t		stats;
+
+	/*
+	 * Lists of tcaches and cache_bin_array_descriptors for extant threads
+	 * associated with this arena.  Stats from these are merged
+	 * incrementally, and at exit if opt_stats_print is enabled.
+	 *
+	 * Synchronization: tcache_ql_mtx.
+	 */
+	ql_head(tcache_t)			tcache_ql;
+	ql_head(cache_bin_array_descriptor_t)	cache_bin_array_descriptor_ql;
+	malloc_mutex_t				tcache_ql_mtx;
+
+	/* Synchronization: internal. */
+	prof_accum_t		prof_accum;
+
+	/*
+	 * PRNG state for cache index randomization of large allocation base
+	 * pointers.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		offset_state;
+
+	/*
+	 * Extent serial number generator state.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		extent_sn_next;
+
+	/*
+	 * Represents a dss_prec_t, but atomically.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_u_t		dss_prec;
+
+	/*
+	 * Number of pages in active extents.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		nactive;
+
+	/*
+	 * Extant large allocations.
+	 *
+	 * Synchronization: large_mtx.
+	 */
+	extent_list_t		large;
+	/* Synchronizes all large allocation/update/deallocation. */
+	malloc_mutex_t		large_mtx;
+
+	/*
+	 * Collections of extents that were previously allocated.  These are
+	 * used when allocating extents, in an attempt to re-use address space.
+	 *
+	 * Synchronization: internal.
+	 */
+	extents_t		extents_dirty;
+	extents_t		extents_muzzy;
+	extents_t		extents_retained;
+
+	/*
+	 * Decay-based purging state, responsible for scheduling extent state
+	 * transitions.
+	 *
+	 * Synchronization: internal.
+	 */
+	arena_decay_t		decay_dirty; /* dirty --> muzzy */
+	arena_decay_t		decay_muzzy; /* muzzy --> retained */
+
+	/*
+	 * Next extent size class in a growing series to use when satisfying a
+	 * request via the extent hooks (only if opt_retain).  This limits the
+	 * number of disjoint virtual memory ranges so that extent merging can
+	 * be effective even if multiple arenas' extent allocation requests are
+	 * highly interleaved.
+	 *
+	 * retain_grow_limit is the max allowed size ind to expand (unless the
+	 * required size is greater).  Default is no limit, and controlled
+	 * through mallctl only.
+	 *
+	 * Synchronization: extent_grow_mtx
+	 */
+	pszind_t		extent_grow_next;
+	pszind_t		retain_grow_limit;
+	malloc_mutex_t		extent_grow_mtx;
+
+	/*
+	 * Available extent structures that were allocated via
+	 * base_alloc_extent().
+	 *
+	 * Synchronization: extent_avail_mtx.
+	 */
+	extent_tree_t		extent_avail;
+	atomic_zu_t		extent_avail_cnt;
+	malloc_mutex_t		extent_avail_mtx;
+
+	/*
+	 * bins is used to store heaps of free regions.
+	 *
+	 * Synchronization: internal.
+	 */
+	bins_t			bins[SC_NBINS];
+
+	/*
+	 * Base allocator, from which arena metadata are allocated.
+	 *
+	 * Synchronization: internal.
+	 */
+	base_t			*base;
+	/* Used to determine uptime.  Read-only after initialization. */
+	nstime_t		create_time;
+};
+
+/* Used in conjunction with tsd for fast arena-related context lookup. */
+struct arena_tdata_s {
+	ticker_t		decay_ticker;
+};
+
+/* Used to pass rtree lookup context down the path. */
+struct alloc_ctx_s {
+	szind_t szind;
+	bool slab;
+};
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H */
diff --git a/jemalloc/include/jemalloc/internal/arena_types.h b/jemalloc/include/jemalloc/internal/arena_types.h
new file mode 100644
index 0000000..624937e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/arena_types.h
@@ -0,0 +1,51 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_TYPES_H
+#define JEMALLOC_INTERNAL_ARENA_TYPES_H
+
+#include "jemalloc/internal/sc.h"
+
+/* Maximum number of regions in one slab. */
+#define LG_SLAB_MAXREGS		(LG_PAGE - SC_LG_TINY_MIN)
+#define SLAB_MAXREGS		(1U << LG_SLAB_MAXREGS)
+
+/* Default decay times in milliseconds. */
+#define DIRTY_DECAY_MS_DEFAULT	ZD(10 * 1000)
+#define MUZZY_DECAY_MS_DEFAULT	(0)
+/* Number of event ticks between time checks. */
+#define DECAY_NTICKS_PER_UPDATE	1000
+
+typedef struct arena_slab_data_s arena_slab_data_t;
+typedef struct arena_decay_s arena_decay_t;
+typedef struct arena_s arena_t;
+typedef struct arena_tdata_s arena_tdata_t;
+typedef struct alloc_ctx_s alloc_ctx_t;
+
+typedef enum {
+	percpu_arena_mode_names_base   = 0, /* Used for options processing. */
+
+	/*
+	 * *_uninit are used only during bootstrapping, and must correspond
+	 * to initialized variant plus percpu_arena_mode_enabled_base.
+	 */
+	percpu_arena_uninit            = 0,
+	per_phycpu_arena_uninit        = 1,
+
+	/* All non-disabled modes must come after percpu_arena_disabled. */
+	percpu_arena_disabled          = 2,
+
+	percpu_arena_mode_names_limit  = 3, /* Used for options processing. */
+	percpu_arena_mode_enabled_base = 3,
+
+	percpu_arena                   = 3,
+	per_phycpu_arena               = 4  /* Hyper threads share arena. */
+} percpu_arena_mode_t;
+
+#define PERCPU_ARENA_ENABLED(m)	((m) >= percpu_arena_mode_enabled_base)
+#define PERCPU_ARENA_DEFAULT	percpu_arena_disabled
+
+/*
+ * When allocation_size >= oversize_threshold, use the dedicated huge arena
+ * (unless have explicitly spicified arena index).  0 disables the feature.
+ */
+#define OVERSIZE_THRESHOLD_DEFAULT (8 << 20)
+
+#endif /* JEMALLOC_INTERNAL_ARENA_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/assert.h b/jemalloc/include/jemalloc/internal/assert.h
new file mode 100644
index 0000000..be4d45b
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/assert.h
@@ -0,0 +1,56 @@
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/util.h"
+
+/*
+ * Define a custom assert() in order to reduce the chances of deadlock during
+ * assertion failure.
+ */
+#ifndef assert
+#define assert(e) do {							\
+	if (unlikely(config_debug && !(e))) {				\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Failed assertion: \"%s\"\n",	\
+		    __FILE__, __LINE__, #e);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#ifndef not_reached
+#define not_reached() do {						\
+	if (config_debug) {						\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Unreachable code reached\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+	unreachable();							\
+} while (0)
+#endif
+
+#ifndef not_implemented
+#define not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_printf("<jemalloc>: %s:%d: Not implemented\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#ifndef assert_not_implemented
+#define assert_not_implemented(e) do {					\
+	if (unlikely(config_debug && !(e))) {				\
+		not_implemented();					\
+	}								\
+} while (0)
+#endif
+
+/* Use to assert a particular configuration, e.g., cassert(config_debug). */
+#ifndef cassert
+#define cassert(c) do {							\
+	if (unlikely(!(c))) {						\
+		not_reached();						\
+	}								\
+} while (0)
+#endif
diff --git a/jemalloc/include/jemalloc/internal/atomic.h b/jemalloc/include/jemalloc/internal/atomic.h
new file mode 100644
index 0000000..a76f54c
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/atomic.h
@@ -0,0 +1,86 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_H
+#define JEMALLOC_INTERNAL_ATOMIC_H
+
+#define ATOMIC_INLINE JEMALLOC_ALWAYS_INLINE
+
+#define JEMALLOC_U8_ATOMICS
+#if defined(JEMALLOC_GCC_ATOMIC_ATOMICS)
+#  include "jemalloc/internal/atomic_gcc_atomic.h"
+#  if !defined(JEMALLOC_GCC_U8_ATOMIC_ATOMICS)
+#    undef JEMALLOC_U8_ATOMICS
+#  endif
+#elif defined(JEMALLOC_GCC_SYNC_ATOMICS)
+#  include "jemalloc/internal/atomic_gcc_sync.h"
+#  if !defined(JEMALLOC_GCC_U8_SYNC_ATOMICS)
+#    undef JEMALLOC_U8_ATOMICS
+#  endif
+#elif defined(_MSC_VER)
+#  include "jemalloc/internal/atomic_msvc.h"
+#elif defined(JEMALLOC_C11_ATOMICS)
+#  include "jemalloc/internal/atomic_c11.h"
+#else
+#  error "Don't have atomics implemented on this platform."
+#endif
+
+/*
+ * This header gives more or less a backport of C11 atomics. The user can write
+ * JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_sizeof_type); to generate
+ * counterparts of the C11 atomic functions for type, as so:
+ *   JEMALLOC_GENERATE_ATOMICS(int *, pi, 3);
+ * and then write things like:
+ *   int *some_ptr;
+ *   atomic_pi_t atomic_ptr_to_int;
+ *   atomic_store_pi(&atomic_ptr_to_int, some_ptr, ATOMIC_RELAXED);
+ *   int *prev_value = atomic_exchange_pi(&ptr_to_int, NULL, ATOMIC_ACQ_REL);
+ *   assert(some_ptr == prev_value);
+ * and expect things to work in the obvious way.
+ *
+ * Also included (with naming differences to avoid conflicts with the standard
+ * library):
+ *   atomic_fence(atomic_memory_order_t) (mimics C11's atomic_thread_fence).
+ *   ATOMIC_INIT (mimics C11's ATOMIC_VAR_INIT).
+ */
+
+/*
+ * Pure convenience, so that we don't have to type "atomic_memory_order_"
+ * quite so often.
+ */
+#define ATOMIC_RELAXED atomic_memory_order_relaxed
+#define ATOMIC_ACQUIRE atomic_memory_order_acquire
+#define ATOMIC_RELEASE atomic_memory_order_release
+#define ATOMIC_ACQ_REL atomic_memory_order_acq_rel
+#define ATOMIC_SEQ_CST atomic_memory_order_seq_cst
+
+/*
+ * Not all platforms have 64-bit atomics.  If we do, this #define exposes that
+ * fact.
+ */
+#if (LG_SIZEOF_PTR == 3 || LG_SIZEOF_INT == 3)
+#  define JEMALLOC_ATOMIC_U64
+#endif
+
+JEMALLOC_GENERATE_ATOMICS(void *, p, LG_SIZEOF_PTR)
+
+/*
+ * There's no actual guarantee that sizeof(bool) == 1, but it's true on the only
+ * platform that actually needs to know the size, MSVC.
+ */
+JEMALLOC_GENERATE_ATOMICS(bool, b, 0)
+
+JEMALLOC_GENERATE_INT_ATOMICS(unsigned, u, LG_SIZEOF_INT)
+
+JEMALLOC_GENERATE_INT_ATOMICS(size_t, zu, LG_SIZEOF_PTR)
+
+JEMALLOC_GENERATE_INT_ATOMICS(ssize_t, zd, LG_SIZEOF_PTR)
+
+JEMALLOC_GENERATE_INT_ATOMICS(uint8_t, u8, 0)
+
+JEMALLOC_GENERATE_INT_ATOMICS(uint32_t, u32, 2)
+
+#ifdef JEMALLOC_ATOMIC_U64
+JEMALLOC_GENERATE_INT_ATOMICS(uint64_t, u64, 3)
+#endif
+
+#undef ATOMIC_INLINE
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_H */
diff --git a/jemalloc/include/jemalloc/internal/atomic_c11.h b/jemalloc/include/jemalloc/internal/atomic_c11.h
new file mode 100644
index 0000000..a5f9313
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/atomic_c11.h
@@ -0,0 +1,97 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_C11_H
+#define JEMALLOC_INTERNAL_ATOMIC_C11_H
+
+#include <stdatomic.h>
+
+#define ATOMIC_INIT(...) ATOMIC_VAR_INIT(__VA_ARGS__)
+
+#define atomic_memory_order_t memory_order
+#define atomic_memory_order_relaxed memory_order_relaxed
+#define atomic_memory_order_acquire memory_order_acquire
+#define atomic_memory_order_release memory_order_release
+#define atomic_memory_order_acq_rel memory_order_acq_rel
+#define atomic_memory_order_seq_cst memory_order_seq_cst
+
+#define atomic_fence atomic_thread_fence
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef _Atomic(type) atomic_##short_type##_t;				\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	/*								\
+	 * A strict interpretation of the C standard prevents		\
+	 * atomic_load from taking a const argument, but it's		\
+	 * convenient for our purposes. This cast is a workaround.	\
+	 */								\
+	atomic_##short_type##_t* a_nonconst =				\
+	    (atomic_##short_type##_t*)a;				\
+	return atomic_load_explicit(a_nonconst, mo);			\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	atomic_store_explicit(a, val, mo);				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return atomic_exchange_explicit(a, val, mo);			\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return atomic_compare_exchange_weak_explicit(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return atomic_compare_exchange_strong_explicit(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}
+
+/*
+ * Integral types have some special operations available that non-integral ones
+ * lack.
+ */
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type, 		\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_add_explicit(a, val, mo);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_sub_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_and_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_or_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_xor_explicit(a, val, mo);			\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_C11_H */
diff --git a/jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h b/jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
new file mode 100644
index 0000000..471515e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
@@ -0,0 +1,129 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H
+#define JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H
+
+#include "jemalloc/internal/assert.h"
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+ATOMIC_INLINE int
+atomic_enum_to_builtin(atomic_memory_order_t mo) {
+	switch (mo) {
+	case atomic_memory_order_relaxed:
+		return __ATOMIC_RELAXED;
+	case atomic_memory_order_acquire:
+		return __ATOMIC_ACQUIRE;
+	case atomic_memory_order_release:
+		return __ATOMIC_RELEASE;
+	case atomic_memory_order_acq_rel:
+		return __ATOMIC_ACQ_REL;
+	case atomic_memory_order_seq_cst:
+		return __ATOMIC_SEQ_CST;
+	}
+	/* Can't happen; the switch is exhaustive. */
+	not_reached();
+}
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	__atomic_thread_fence(atomic_enum_to_builtin(mo));
+}
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef struct {							\
+	type repr;							\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	type result;							\
+	__atomic_load(&a->repr, &result, atomic_enum_to_builtin(mo));	\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a, type val,		\
+    atomic_memory_order_t mo) {						\
+	__atomic_store(&a->repr, &val, atomic_enum_to_builtin(mo));	\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	type result;							\
+	__atomic_exchange(&a->repr, &val, &result,			\
+	    atomic_enum_to_builtin(mo));				\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    UNUSED type *expected, type desired,				\
+    atomic_memory_order_t success_mo,					\
+    atomic_memory_order_t failure_mo) {					\
+	return __atomic_compare_exchange(&a->repr, expected, &desired,	\
+	    true, atomic_enum_to_builtin(success_mo),			\
+	    atomic_enum_to_builtin(failure_mo));			\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    UNUSED type *expected, type desired,				\
+    atomic_memory_order_t success_mo,					\
+    atomic_memory_order_t failure_mo) {					\
+	return __atomic_compare_exchange(&a->repr, expected, &desired,	\
+	    false,							\
+	    atomic_enum_to_builtin(success_mo),				\
+	    atomic_enum_to_builtin(failure_mo));			\
+}
+
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_add(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_sub(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_and(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_or(&a->repr, val,				\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_xor(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H */
diff --git a/jemalloc/include/jemalloc/internal/atomic_gcc_sync.h b/jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
new file mode 100644
index 0000000..e02b7cb
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
@@ -0,0 +1,195 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H
+#define JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	/* Easy cases first: no barrier, and full barrier. */
+	if (mo == atomic_memory_order_relaxed) {
+		asm volatile("" ::: "memory");
+		return;
+	}
+	if (mo == atomic_memory_order_seq_cst) {
+		asm volatile("" ::: "memory");
+		__sync_synchronize();
+		asm volatile("" ::: "memory");
+		return;
+	}
+	asm volatile("" ::: "memory");
+#  if defined(__i386__) || defined(__x86_64__)
+	/* This is implicit on x86. */
+#  elif defined(__ppc64__)
+	asm volatile("lwsync");
+#  elif defined(__ppc__)
+	asm volatile("sync");
+#  elif defined(__sparc__) && defined(__arch64__)
+	if (mo == atomic_memory_order_acquire) {
+		asm volatile("membar #LoadLoad | #LoadStore");
+	} else if (mo == atomic_memory_order_release) {
+		asm volatile("membar #LoadStore | #StoreStore");
+	} else {
+		asm volatile("membar #LoadLoad | #LoadStore | #StoreStore");
+	}
+#  else
+	__sync_synchronize();
+#  endif
+	asm volatile("" ::: "memory");
+}
+
+/*
+ * A correct implementation of seq_cst loads and stores on weakly ordered
+ * architectures could do either of the following:
+ *   1. store() is weak-fence -> store -> strong fence, load() is load ->
+ *      strong-fence.
+ *   2. store() is strong-fence -> store, load() is strong-fence -> load ->
+ *      weak-fence.
+ * The tricky thing is, load() and store() above can be the load or store
+ * portions of a gcc __sync builtin, so we have to follow GCC's lead, which
+ * means going with strategy 2.
+ * On strongly ordered architectures, the natural strategy is to stick a strong
+ * fence after seq_cst stores, and have naked loads.  So we want the strong
+ * fences in different places on different architectures.
+ * atomic_pre_sc_load_fence and atomic_post_sc_store_fence allow us to
+ * accomplish this.
+ */
+
+ATOMIC_INLINE void
+atomic_pre_sc_load_fence() {
+#  if defined(__i386__) || defined(__x86_64__) ||			\
+    (defined(__sparc__) && defined(__arch64__))
+	atomic_fence(atomic_memory_order_relaxed);
+#  else
+	atomic_fence(atomic_memory_order_seq_cst);
+#  endif
+}
+
+ATOMIC_INLINE void
+atomic_post_sc_store_fence() {
+#  if defined(__i386__) || defined(__x86_64__) ||			\
+    (defined(__sparc__) && defined(__arch64__))
+	atomic_fence(atomic_memory_order_seq_cst);
+#  else
+	atomic_fence(atomic_memory_order_relaxed);
+#  endif
+
+}
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef struct {							\
+	type volatile repr;						\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_pre_sc_load_fence();				\
+	}								\
+	type result = a->repr;						\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_acquire);		\
+	}								\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_release);		\
+	}								\
+	a->repr = val;							\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_post_sc_store_fence();				\
+	}								\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val, \
+    atomic_memory_order_t mo) {                  					 \
+	/*								\
+	 * Because of FreeBSD, we care about gcc 4.2, which doesn't have\
+	 * an atomic exchange builtin.  We fake it with a CAS loop.	\
+	 */								\
+	while (true) {							\
+		type old = a->repr;					\
+		if (__sync_bool_compare_and_swap(&a->repr, old, val)) {	\
+			return old;					\
+		}							\
+	}								\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired,                                     \
+    atomic_memory_order_t success_mo,                          \
+    atomic_memory_order_t failure_mo) {				                \
+	type prev = __sync_val_compare_and_swap(&a->repr, *expected,	\
+	    desired);							\
+	if (prev == *expected) {					\
+		return true;						\
+	} else {							\
+		*expected = prev;					\
+		return false;						\
+	}								\
+}									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired,                                       \
+    atomic_memory_order_t success_mo,                            \
+    atomic_memory_order_t failure_mo) {                          \
+	type prev = __sync_val_compare_and_swap(&a->repr, *expected,	\
+	    desired);							\
+	if (prev == *expected) {					\
+		return true;						\
+	} else {							\
+		*expected = prev;					\
+		return false;						\
+	}								\
+}
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_add(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_sub(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_and(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_or(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_xor(&a->repr, val);			\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H */
diff --git a/jemalloc/include/jemalloc/internal/atomic_msvc.h b/jemalloc/include/jemalloc/internal/atomic_msvc.h
new file mode 100644
index 0000000..67057ce
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/atomic_msvc.h
@@ -0,0 +1,158 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_MSVC_H
+#define JEMALLOC_INTERNAL_ATOMIC_MSVC_H
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+typedef char atomic_repr_0_t;
+typedef short atomic_repr_1_t;
+typedef long atomic_repr_2_t;
+typedef __int64 atomic_repr_3_t;
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	_ReadWriteBarrier();
+#  if defined(_M_ARM) || defined(_M_ARM64)
+	/* ARM needs a barrier for everything but relaxed. */
+	if (mo != atomic_memory_order_relaxed) {
+		MemoryBarrier();
+	}
+#  elif defined(_M_IX86) || defined (_M_X64)
+	/* x86 needs a barrier only for seq_cst. */
+	if (mo == atomic_memory_order_seq_cst) {
+		MemoryBarrier();
+	}
+#  else
+#  error "Don't know how to create atomics for this platform for MSVC."
+#  endif
+	_ReadWriteBarrier();
+}
+
+#define ATOMIC_INTERLOCKED_REPR(lg_size) atomic_repr_ ## lg_size ## _t
+
+#define ATOMIC_CONCAT(a, b) ATOMIC_RAW_CONCAT(a, b)
+#define ATOMIC_RAW_CONCAT(a, b) a ## b
+
+#define ATOMIC_INTERLOCKED_NAME(base_name, lg_size) ATOMIC_CONCAT(	\
+    base_name, ATOMIC_INTERLOCKED_SUFFIX(lg_size))
+
+#define ATOMIC_INTERLOCKED_SUFFIX(lg_size)				\
+    ATOMIC_CONCAT(ATOMIC_INTERLOCKED_SUFFIX_, lg_size)
+
+#define ATOMIC_INTERLOCKED_SUFFIX_0 8
+#define ATOMIC_INTERLOCKED_SUFFIX_1 16
+#define ATOMIC_INTERLOCKED_SUFFIX_2
+#define ATOMIC_INTERLOCKED_SUFFIX_3 64
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_size)		\
+typedef struct {							\
+	ATOMIC_INTERLOCKED_REPR(lg_size) repr;				\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	ATOMIC_INTERLOCKED_REPR(lg_size) ret = a->repr;			\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_acquire);		\
+	}								\
+	return (type) ret;						\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_release);		\
+	}								\
+	a->repr = (ATOMIC_INTERLOCKED_REPR(lg_size)) val;		\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_fence(atomic_memory_order_seq_cst);		\
+	}								\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedExchange,	\
+	    lg_size)(&a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);	\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	ATOMIC_INTERLOCKED_REPR(lg_size) e =				\
+	    (ATOMIC_INTERLOCKED_REPR(lg_size))*expected;		\
+	ATOMIC_INTERLOCKED_REPR(lg_size) d =				\
+	    (ATOMIC_INTERLOCKED_REPR(lg_size))desired;			\
+	ATOMIC_INTERLOCKED_REPR(lg_size) old =				\
+	    ATOMIC_INTERLOCKED_NAME(_InterlockedCompareExchange, 	\
+		lg_size)(&a->repr, d, e);				\
+	if (old == e) {							\
+		return true;						\
+	} else {							\
+		*expected = (type)old;					\
+		return false;						\
+	}								\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	/* We implement the weak version with strong semantics. */	\
+	return atomic_compare_exchange_weak_##short_type(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}
+
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type, lg_size)	\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_size)			\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedExchangeAdd,	\
+	    lg_size)(&a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);	\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	/*								\
+	 * MSVC warns on negation of unsigned operands, but for us it	\
+	 * gives exactly the right semantics (MAX_TYPE + 1 - operand).	\
+	 */								\
+	__pragma(warning(push))						\
+	__pragma(warning(disable: 4146))				\
+	return atomic_fetch_add_##short_type(a, -val, mo);		\
+	__pragma(warning(pop))						\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedAnd, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedOr, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedXor, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_MSVC_H */
diff --git a/jemalloc/include/jemalloc/internal/background_thread_externs.h b/jemalloc/include/jemalloc/internal/background_thread_externs.h
new file mode 100644
index 0000000..0f997e1
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/background_thread_externs.h
@@ -0,0 +1,32 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H
+
+extern bool opt_background_thread;
+extern size_t opt_max_background_threads;
+extern malloc_mutex_t background_thread_lock;
+extern atomic_b_t background_thread_enabled_state;
+extern size_t n_background_threads;
+extern size_t max_background_threads;
+extern background_thread_info_t *background_thread_info;
+
+bool background_thread_create(tsd_t *tsd, unsigned arena_ind);
+bool background_threads_enable(tsd_t *tsd);
+bool background_threads_disable(tsd_t *tsd);
+void background_thread_interval_check(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, size_t npages_new);
+void background_thread_prefork0(tsdn_t *tsdn);
+void background_thread_prefork1(tsdn_t *tsdn);
+void background_thread_postfork_parent(tsdn_t *tsdn);
+void background_thread_postfork_child(tsdn_t *tsdn);
+bool background_thread_stats_read(tsdn_t *tsdn,
+    background_thread_stats_t *stats);
+void background_thread_ctl_init(tsdn_t *tsdn);
+
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+extern int pthread_create_wrapper(pthread_t *__restrict, const pthread_attr_t *,
+    void *(*)(void *), void *__restrict);
+#endif
+bool background_thread_boot0(void);
+bool background_thread_boot1(tsdn_t *tsdn);
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/background_thread_inlines.h b/jemalloc/include/jemalloc/internal/background_thread_inlines.h
new file mode 100644
index 0000000..f85e86f
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/background_thread_inlines.h
@@ -0,0 +1,62 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H
+
+JEMALLOC_ALWAYS_INLINE bool
+background_thread_enabled(void) {
+	return atomic_load_b(&background_thread_enabled_state, ATOMIC_RELAXED);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+background_thread_enabled_set(tsdn_t *tsdn, bool state) {
+	malloc_mutex_assert_owner(tsdn, &background_thread_lock);
+	atomic_store_b(&background_thread_enabled_state, state, ATOMIC_RELAXED);
+}
+
+JEMALLOC_ALWAYS_INLINE background_thread_info_t *
+arena_background_thread_info_get(arena_t *arena) {
+	unsigned arena_ind = arena_ind_get(arena);
+	return &background_thread_info[arena_ind % max_background_threads];
+}
+
+JEMALLOC_ALWAYS_INLINE background_thread_info_t *
+background_thread_info_get(size_t ind) {
+	return &background_thread_info[ind % max_background_threads];
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+background_thread_wakeup_time_get(background_thread_info_t *info) {
+	uint64_t next_wakeup = nstime_ns(&info->next_wakeup);
+	assert(atomic_load_b(&info->indefinite_sleep, ATOMIC_ACQUIRE) ==
+	    (next_wakeup == BACKGROUND_THREAD_INDEFINITE_SLEEP));
+	return next_wakeup;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+background_thread_wakeup_time_set(tsdn_t *tsdn, background_thread_info_t *info,
+    uint64_t wakeup_time) {
+	malloc_mutex_assert_owner(tsdn, &info->mtx);
+	atomic_store_b(&info->indefinite_sleep,
+	    wakeup_time == BACKGROUND_THREAD_INDEFINITE_SLEEP, ATOMIC_RELEASE);
+	nstime_init(&info->next_wakeup, wakeup_time);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+background_thread_indefinite_sleep(background_thread_info_t *info) {
+	return atomic_load_b(&info->indefinite_sleep, ATOMIC_ACQUIRE);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_background_thread_inactivity_check(tsdn_t *tsdn, arena_t *arena,
+    bool is_background_thread) {
+	if (!background_thread_enabled() || is_background_thread) {
+		return;
+	}
+	background_thread_info_t *info =
+	    arena_background_thread_info_get(arena);
+	if (background_thread_indefinite_sleep(info)) {
+		background_thread_interval_check(tsdn, arena,
+		    &arena->decay_dirty, 0);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H */
diff --git a/jemalloc/include/jemalloc/internal/background_thread_structs.h b/jemalloc/include/jemalloc/internal/background_thread_structs.h
new file mode 100644
index 0000000..c02aa43
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/background_thread_structs.h
@@ -0,0 +1,54 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H
+
+/* This file really combines "structs" and "types", but only transitionally. */
+
+#if defined(JEMALLOC_BACKGROUND_THREAD) || defined(JEMALLOC_LAZY_LOCK)
+#  define JEMALLOC_PTHREAD_CREATE_WRAPPER
+#endif
+
+#define BACKGROUND_THREAD_INDEFINITE_SLEEP UINT64_MAX
+#define MAX_BACKGROUND_THREAD_LIMIT MALLOCX_ARENA_LIMIT
+#define DEFAULT_NUM_BACKGROUND_THREAD 4
+
+typedef enum {
+	background_thread_stopped,
+	background_thread_started,
+	/* Thread waits on the global lock when paused (for arena_reset). */
+	background_thread_paused,
+} background_thread_state_t;
+
+struct background_thread_info_s {
+#ifdef JEMALLOC_BACKGROUND_THREAD
+	/* Background thread is pthread specific. */
+	pthread_t		thread;
+	pthread_cond_t		cond;
+#endif
+	malloc_mutex_t		mtx;
+	background_thread_state_t	state;
+	/* When true, it means no wakeup scheduled. */
+	atomic_b_t		indefinite_sleep;
+	/* Next scheduled wakeup time (absolute time in ns). */
+	nstime_t		next_wakeup;
+	/*
+	 *  Since the last background thread run, newly added number of pages
+	 *  that need to be purged by the next wakeup.  This is adjusted on
+	 *  epoch advance, and is used to determine whether we should signal the
+	 *  background thread to wake up earlier.
+	 */
+	size_t			npages_to_purge_new;
+	/* Stats: total number of runs since started. */
+	uint64_t		tot_n_runs;
+	/* Stats: total sleep time since started. */
+	nstime_t		tot_sleep_time;
+};
+typedef struct background_thread_info_s background_thread_info_t;
+
+struct background_thread_stats_s {
+	size_t num_threads;
+	uint64_t num_runs;
+	nstime_t run_interval;
+};
+typedef struct background_thread_stats_s background_thread_stats_t;
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H */
diff --git a/jemalloc/include/jemalloc/internal/base_externs.h b/jemalloc/include/jemalloc/internal/base_externs.h
new file mode 100644
index 0000000..7b705c9
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/base_externs.h
@@ -0,0 +1,22 @@
+#ifndef JEMALLOC_INTERNAL_BASE_EXTERNS_H
+#define JEMALLOC_INTERNAL_BASE_EXTERNS_H
+
+extern metadata_thp_mode_t opt_metadata_thp;
+extern const char *metadata_thp_mode_names[];
+
+base_t *b0get(void);
+base_t *base_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+void base_delete(tsdn_t *tsdn, base_t *base);
+extent_hooks_t *base_extent_hooks_get(base_t *base);
+extent_hooks_t *base_extent_hooks_set(base_t *base,
+    extent_hooks_t *extent_hooks);
+void *base_alloc(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment);
+extent_t *base_alloc_extent(tsdn_t *tsdn, base_t *base);
+void base_stats_get(tsdn_t *tsdn, base_t *base, size_t *allocated,
+    size_t *resident, size_t *mapped, size_t *n_thp);
+void base_prefork(tsdn_t *tsdn, base_t *base);
+void base_postfork_parent(tsdn_t *tsdn, base_t *base);
+void base_postfork_child(tsdn_t *tsdn, base_t *base);
+bool base_boot(tsdn_t *tsdn);
+
+#endif /* JEMALLOC_INTERNAL_BASE_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/base_inlines.h b/jemalloc/include/jemalloc/internal/base_inlines.h
new file mode 100644
index 0000000..aec0e2e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/base_inlines.h
@@ -0,0 +1,13 @@
+#ifndef JEMALLOC_INTERNAL_BASE_INLINES_H
+#define JEMALLOC_INTERNAL_BASE_INLINES_H
+
+static inline unsigned
+base_ind_get(const base_t *base) {
+	return base->ind;
+}
+
+static inline bool
+metadata_thp_enabled(void) {
+	return (opt_metadata_thp != metadata_thp_disabled);
+}
+#endif /* JEMALLOC_INTERNAL_BASE_INLINES_H */
diff --git a/jemalloc/include/jemalloc/internal/base_structs.h b/jemalloc/include/jemalloc/internal/base_structs.h
new file mode 100644
index 0000000..07f214e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/base_structs.h
@@ -0,0 +1,59 @@
+#ifndef JEMALLOC_INTERNAL_BASE_STRUCTS_H
+#define JEMALLOC_INTERNAL_BASE_STRUCTS_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/sc.h"
+
+/* Embedded at the beginning of every block of base-managed virtual memory. */
+struct base_block_s {
+	/* Total size of block's virtual memory mapping. */
+	size_t		size;
+
+	/* Next block in list of base's blocks. */
+	base_block_t	*next;
+
+	/* Tracks unused trailing space. */
+	extent_t	extent;
+};
+
+struct base_s {
+	/* Associated arena's index within the arenas array. */
+	unsigned	ind;
+
+	/*
+	 * User-configurable extent hook functions.  Points to an
+	 * extent_hooks_t.
+	 */
+	atomic_p_t	extent_hooks;
+
+	/* Protects base_alloc() and base_stats_get() operations. */
+	malloc_mutex_t	mtx;
+
+	/* Using THP when true (metadata_thp auto mode). */
+	bool		auto_thp_switched;
+	/*
+	 * Most recent size class in the series of increasingly large base
+	 * extents.  Logarithmic spacing between subsequent allocations ensures
+	 * that the total number of distinct mappings remains small.
+	 */
+	pszind_t	pind_last;
+
+	/* Serial number generation state. */
+	size_t		extent_sn_next;
+
+	/* Chain of all blocks associated with base. */
+	base_block_t	*blocks;
+
+	/* Heap of extents that track unused trailing space within blocks. */
+	extent_heap_t	avail[SC_NSIZES];
+
+	/* Stats, only maintained if config_stats. */
+	size_t		allocated;
+	size_t		resident;
+	size_t		mapped;
+	/* Number of THP regions touched. */
+	size_t		n_thp;
+};
+
+#endif /* JEMALLOC_INTERNAL_BASE_STRUCTS_H */
diff --git a/jemalloc/include/jemalloc/internal/base_types.h b/jemalloc/include/jemalloc/internal/base_types.h
new file mode 100644
index 0000000..b6db77d
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/base_types.h
@@ -0,0 +1,33 @@
+#ifndef JEMALLOC_INTERNAL_BASE_TYPES_H
+#define JEMALLOC_INTERNAL_BASE_TYPES_H
+
+typedef struct base_block_s base_block_t;
+typedef struct base_s base_t;
+
+#define METADATA_THP_DEFAULT metadata_thp_disabled
+
+/*
+ * In auto mode, arenas switch to huge pages for the base allocator on the
+ * second base block.  a0 switches to thp on the 5th block (after 20 megabytes
+ * of metadata), since more metadata (e.g. rtree nodes) come from a0's base.
+ */
+
+#define BASE_AUTO_THP_THRESHOLD    2
+#define BASE_AUTO_THP_THRESHOLD_A0 5
+
+typedef enum {
+	metadata_thp_disabled   = 0,
+	/*
+	 * Lazily enable hugepage for metadata. To avoid high RSS caused by THP
+	 * + low usage arena (i.e. THP becomes a significant percentage), the
+	 * "auto" option only starts using THP after a base allocator used up
+	 * the first THP region.  Starting from the second hugepage (in a single
+	 * arena), "auto" behaves the same as "always", i.e. madvise hugepage
+	 * right away.
+	 */
+	metadata_thp_auto       = 1,
+	metadata_thp_always     = 2,
+	metadata_thp_mode_limit = 3
+} metadata_thp_mode_t;
+
+#endif /* JEMALLOC_INTERNAL_BASE_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/bin.h b/jemalloc/include/jemalloc/internal/bin.h
new file mode 100644
index 0000000..8547e89
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/bin.h
@@ -0,0 +1,123 @@
+#ifndef JEMALLOC_INTERNAL_BIN_H
+#define JEMALLOC_INTERNAL_BIN_H
+
+#include "jemalloc/internal/bin_stats.h"
+#include "jemalloc/internal/bin_types.h"
+#include "jemalloc/internal/extent_types.h"
+#include "jemalloc/internal/extent_structs.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/sc.h"
+
+/*
+ * A bin contains a set of extents that are currently being used for slab
+ * allocations.
+ */
+
+/*
+ * Read-only information associated with each element of arena_t's bins array
+ * is stored separately, partly to reduce memory usage (only one copy, rather
+ * than one per arena), but mainly to avoid false cacheline sharing.
+ *
+ * Each slab has the following layout:
+ *
+ *   /--------------------\
+ *   | region 0           |
+ *   |--------------------|
+ *   | region 1           |
+ *   |--------------------|
+ *   | ...                |
+ *   | ...                |
+ *   | ...                |
+ *   |--------------------|
+ *   | region nregs-1     |
+ *   \--------------------/
+ */
+typedef struct bin_info_s bin_info_t;
+struct bin_info_s {
+	/* Size of regions in a slab for this bin's size class. */
+	size_t			reg_size;
+
+	/* Total size of a slab for this bin's size class. */
+	size_t			slab_size;
+
+	/* Total number of regions in a slab for this bin's size class. */
+	uint32_t		nregs;
+
+	/* Number of sharded bins in each arena for this size class. */
+	uint32_t		n_shards;
+
+	/*
+	 * Metadata used to manipulate bitmaps for slabs associated with this
+	 * bin.
+	 */
+	bitmap_info_t		bitmap_info;
+};
+
+extern bin_info_t bin_infos[SC_NBINS];
+
+typedef struct bin_s bin_t;
+struct bin_s {
+	/* All operations on bin_t fields require lock ownership. */
+	malloc_mutex_t		lock;
+
+	/*
+	 * Current slab being used to service allocations of this bin's size
+	 * class.  slabcur is independent of slabs_{nonfull,full}; whenever
+	 * slabcur is reassigned, the previous slab must be deallocated or
+	 * inserted into slabs_{nonfull,full}.
+	 */
+	extent_t		*slabcur;
+
+	/*
+	 * Heap of non-full slabs.  This heap is used to assure that new
+	 * allocations come from the non-full slab that is oldest/lowest in
+	 * memory.
+	 */
+	extent_heap_t		slabs_nonfull;
+
+	/* List used to track full slabs. */
+	extent_list_t		slabs_full;
+
+	/* Bin statistics. */
+	bin_stats_t	stats;
+};
+
+/* A set of sharded bins of the same size class. */
+typedef struct bins_s bins_t;
+struct bins_s {
+	/* Sharded bins.  Dynamically sized. */
+	bin_t *bin_shards;
+};
+
+void bin_shard_sizes_boot(unsigned bin_shards[SC_NBINS]);
+bool bin_update_shard_size(unsigned bin_shards[SC_NBINS], size_t start_size,
+    size_t end_size, size_t nshards);
+void bin_boot(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS]);
+
+/* Initializes a bin to empty.  Returns true on error. */
+bool bin_init(bin_t *bin);
+
+/* Forking. */
+void bin_prefork(tsdn_t *tsdn, bin_t *bin);
+void bin_postfork_parent(tsdn_t *tsdn, bin_t *bin);
+void bin_postfork_child(tsdn_t *tsdn, bin_t *bin);
+
+/* Stats. */
+static inline void
+bin_stats_merge(tsdn_t *tsdn, bin_stats_t *dst_bin_stats, bin_t *bin) {
+	malloc_mutex_lock(tsdn, &bin->lock);
+	malloc_mutex_prof_accum(tsdn, &dst_bin_stats->mutex_data, &bin->lock);
+	dst_bin_stats->nmalloc += bin->stats.nmalloc;
+	dst_bin_stats->ndalloc += bin->stats.ndalloc;
+	dst_bin_stats->nrequests += bin->stats.nrequests;
+	dst_bin_stats->curregs += bin->stats.curregs;
+	dst_bin_stats->nfills += bin->stats.nfills;
+	dst_bin_stats->nflushes += bin->stats.nflushes;
+	dst_bin_stats->nslabs += bin->stats.nslabs;
+	dst_bin_stats->reslabs += bin->stats.reslabs;
+	dst_bin_stats->curslabs += bin->stats.curslabs;
+	dst_bin_stats->nonfull_slabs += bin->stats.nonfull_slabs;
+	malloc_mutex_unlock(tsdn, &bin->lock);
+}
+
+#endif /* JEMALLOC_INTERNAL_BIN_H */
diff --git a/jemalloc/include/jemalloc/internal/bin_stats.h b/jemalloc/include/jemalloc/internal/bin_stats.h
new file mode 100644
index 0000000..d04519c
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/bin_stats.h
@@ -0,0 +1,54 @@
+#ifndef JEMALLOC_INTERNAL_BIN_STATS_H
+#define JEMALLOC_INTERNAL_BIN_STATS_H
+
+#include "jemalloc/internal/mutex_prof.h"
+
+typedef struct bin_stats_s bin_stats_t;
+struct bin_stats_s {
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the bin.  Note that tcache may allocate an object, then recycle it
+	 * many times, resulting many increments to nrequests, but only one
+	 * each to nmalloc and ndalloc.
+	 */
+	uint64_t	nmalloc;
+	uint64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to the size of this
+	 * bin.  This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	uint64_t	nrequests;
+
+	/*
+	 * Current number of regions of this size class, including regions
+	 * currently cached by tcache.
+	 */
+	size_t		curregs;
+
+	/* Number of tcache fills from this bin. */
+	uint64_t	nfills;
+
+	/* Number of tcache flushes to this bin. */
+	uint64_t	nflushes;
+
+	/* Total number of slabs created for this bin's size class. */
+	uint64_t	nslabs;
+
+	/*
+	 * Total number of slabs reused by extracting them from the slabs heap
+	 * for this bin's size class.
+	 */
+	uint64_t	reslabs;
+
+	/* Current number of slabs in this bin. */
+	size_t		curslabs;
+
+	/* Current size of nonfull slabs heap in this bin. */
+	size_t		nonfull_slabs;
+
+	mutex_prof_data_t mutex_data;
+};
+
+#endif /* JEMALLOC_INTERNAL_BIN_STATS_H */
diff --git a/jemalloc/include/jemalloc/internal/bin_types.h b/jemalloc/include/jemalloc/internal/bin_types.h
new file mode 100644
index 0000000..3533606
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/bin_types.h
@@ -0,0 +1,17 @@
+#ifndef JEMALLOC_INTERNAL_BIN_TYPES_H
+#define JEMALLOC_INTERNAL_BIN_TYPES_H
+
+#include "jemalloc/internal/sc.h"
+
+#define BIN_SHARDS_MAX (1 << EXTENT_BITS_BINSHARD_WIDTH)
+#define N_BIN_SHARDS_DEFAULT 1
+
+/* Used in TSD static initializer only. Real init in arena_bind(). */
+#define TSD_BINSHARDS_ZERO_INITIALIZER {{UINT8_MAX}}
+
+typedef struct tsd_binshards_s tsd_binshards_t;
+struct tsd_binshards_s {
+	uint8_t binshard[SC_NBINS];
+};
+
+#endif /* JEMALLOC_INTERNAL_BIN_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/bit_util.h b/jemalloc/include/jemalloc/internal/bit_util.h
new file mode 100644
index 0000000..c045eb8
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/bit_util.h
@@ -0,0 +1,239 @@
+#ifndef JEMALLOC_INTERNAL_BIT_UTIL_H
+#define JEMALLOC_INTERNAL_BIT_UTIL_H
+
+#include "jemalloc/internal/assert.h"
+
+#define BIT_UTIL_INLINE static inline
+
+/* Sanity check. */
+#if !defined(JEMALLOC_INTERNAL_FFSLL) || !defined(JEMALLOC_INTERNAL_FFSL) \
+    || !defined(JEMALLOC_INTERNAL_FFS)
+#  error JEMALLOC_INTERNAL_FFS{,L,LL} should have been defined by configure
+#endif
+
+
+BIT_UTIL_INLINE unsigned
+ffs_llu(unsigned long long bitmap) {
+	return JEMALLOC_INTERNAL_FFSLL(bitmap);
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_lu(unsigned long bitmap) {
+	return JEMALLOC_INTERNAL_FFSL(bitmap);
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u(unsigned bitmap) {
+	return JEMALLOC_INTERNAL_FFS(bitmap);
+}
+
+#ifdef JEMALLOC_INTERNAL_POPCOUNTL
+BIT_UTIL_INLINE unsigned
+popcount_lu(unsigned long bitmap) {
+  return JEMALLOC_INTERNAL_POPCOUNTL(bitmap);
+}
+#endif
+
+/*
+ * Clears first unset bit in bitmap, and returns
+ * place of bit.  bitmap *must not* be 0.
+ */
+
+BIT_UTIL_INLINE size_t
+cfs_lu(unsigned long* bitmap) {
+	size_t bit = ffs_lu(*bitmap) - 1;
+	*bitmap ^= ZU(1) << bit;
+	return bit;
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_zu(size_t bitmap) {
+#if LG_SIZEOF_PTR == LG_SIZEOF_INT
+	return ffs_u(bitmap);
+#elif LG_SIZEOF_PTR == LG_SIZEOF_LONG
+	return ffs_lu(bitmap);
+#elif LG_SIZEOF_PTR == LG_SIZEOF_LONG_LONG
+	return ffs_llu(bitmap);
+#else
+#error No implementation for size_t ffs()
+#endif
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u64(uint64_t bitmap) {
+#if LG_SIZEOF_LONG == 3
+	return ffs_lu(bitmap);
+#elif LG_SIZEOF_LONG_LONG == 3
+	return ffs_llu(bitmap);
+#else
+#error No implementation for 64-bit ffs()
+#endif
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u32(uint32_t bitmap) {
+#if LG_SIZEOF_INT == 2
+	return ffs_u(bitmap);
+#else
+#error No implementation for 32-bit ffs()
+#endif
+	return ffs_u(bitmap);
+}
+
+BIT_UTIL_INLINE uint64_t
+pow2_ceil_u64(uint64_t x) {
+#if (defined(__amd64__) || defined(__x86_64__) || defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+	if(unlikely(x <= 1)) {
+		return x;
+	}
+	size_t msb_on_index;
+#if (defined(__amd64__) || defined(__x86_64__))
+	asm ("bsrq %1, %0"
+			: "=r"(msb_on_index) // Outputs.
+			: "r"(x-1)           // Inputs.
+		);
+#elif (defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+	msb_on_index = (63 ^ __builtin_clzll(x - 1));
+#endif
+	assert(msb_on_index < 63);
+	return 1ULL << (msb_on_index + 1);
+#else
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x |= x >> 32;
+	x++;
+	return x;
+#endif
+}
+
+BIT_UTIL_INLINE uint32_t
+pow2_ceil_u32(uint32_t x) {
+#if ((defined(__i386__) || defined(JEMALLOC_HAVE_BUILTIN_CLZ)) && (!defined(__s390__)))
+	if(unlikely(x <= 1)) {
+		return x;
+	}
+	size_t msb_on_index;
+#if (defined(__i386__))
+	asm ("bsr %1, %0"
+			: "=r"(msb_on_index) // Outputs.
+			: "r"(x-1)           // Inputs.
+		);
+#elif (defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+	msb_on_index = (31 ^ __builtin_clz(x - 1));
+#endif
+	assert(msb_on_index < 31);
+	return 1U << (msb_on_index + 1);
+#else
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x++;
+	return x;
+#endif
+}
+
+/* Compute the smallest power of 2 that is >= x. */
+BIT_UTIL_INLINE size_t
+pow2_ceil_zu(size_t x) {
+#if (LG_SIZEOF_PTR == 3)
+	return pow2_ceil_u64(x);
+#else
+	return pow2_ceil_u32(x);
+#endif
+}
+
+#if (defined(__i386__) || defined(__amd64__) || defined(__x86_64__))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	size_t ret;
+	assert(x != 0);
+
+	asm ("bsr %1, %0"
+	    : "=r"(ret) // Outputs.
+	    : "r"(x)    // Inputs.
+	    );
+	assert(ret < UINT_MAX);
+	return (unsigned)ret;
+}
+#elif (defined(_MSC_VER))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	unsigned long ret;
+
+	assert(x != 0);
+
+#if (LG_SIZEOF_PTR == 3)
+	_BitScanReverse64(&ret, x);
+#elif (LG_SIZEOF_PTR == 2)
+	_BitScanReverse(&ret, x);
+#else
+#  error "Unsupported type size for lg_floor()"
+#endif
+	assert(ret < UINT_MAX);
+	return (unsigned)ret;
+}
+#elif (defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	assert(x != 0);
+
+#if (LG_SIZEOF_PTR == LG_SIZEOF_INT)
+	return ((8 << LG_SIZEOF_PTR) - 1) - __builtin_clz(x);
+#elif (LG_SIZEOF_PTR == LG_SIZEOF_LONG)
+	return ((8 << LG_SIZEOF_PTR) - 1) - __builtin_clzl(x);
+#else
+#  error "Unsupported type size for lg_floor()"
+#endif
+}
+#else
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	assert(x != 0);
+
+	x |= (x >> 1);
+	x |= (x >> 2);
+	x |= (x >> 4);
+	x |= (x >> 8);
+	x |= (x >> 16);
+#if (LG_SIZEOF_PTR == 3)
+	x |= (x >> 32);
+#endif
+	if (x == SIZE_T_MAX) {
+		return (8 << LG_SIZEOF_PTR) - 1;
+	}
+	x++;
+	return ffs_zu(x) - 2;
+}
+#endif
+
+BIT_UTIL_INLINE unsigned
+lg_ceil(size_t x) {
+	return lg_floor(x) + ((x & (x - 1)) == 0 ? 0 : 1);
+}
+
+#undef BIT_UTIL_INLINE
+
+/* A compile-time version of lg_floor and lg_ceil. */
+#define LG_FLOOR_1(x) 0
+#define LG_FLOOR_2(x) (x < (1ULL << 1) ? LG_FLOOR_1(x) : 1 + LG_FLOOR_1(x >> 1))
+#define LG_FLOOR_4(x) (x < (1ULL << 2) ? LG_FLOOR_2(x) : 2 + LG_FLOOR_2(x >> 2))
+#define LG_FLOOR_8(x) (x < (1ULL << 4) ? LG_FLOOR_4(x) : 4 + LG_FLOOR_4(x >> 4))
+#define LG_FLOOR_16(x) (x < (1ULL << 8) ? LG_FLOOR_8(x) : 8 + LG_FLOOR_8(x >> 8))
+#define LG_FLOOR_32(x) (x < (1ULL << 16) ? LG_FLOOR_16(x) : 16 + LG_FLOOR_16(x >> 16))
+#define LG_FLOOR_64(x) (x < (1ULL << 32) ? LG_FLOOR_32(x) : 32 + LG_FLOOR_32(x >> 32))
+#if LG_SIZEOF_PTR == 2
+#  define LG_FLOOR(x) LG_FLOOR_32((x))
+#else
+#  define LG_FLOOR(x) LG_FLOOR_64((x))
+#endif
+
+#define LG_CEIL(x) (LG_FLOOR(x) + (((x) & ((x) - 1)) == 0 ? 0 : 1))
+
+#endif /* JEMALLOC_INTERNAL_BIT_UTIL_H */
diff --git a/jemalloc/include/jemalloc/internal/bitmap.h b/jemalloc/include/jemalloc/internal/bitmap.h
new file mode 100644
index 0000000..c3f9cb4
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/bitmap.h
@@ -0,0 +1,369 @@
+#ifndef JEMALLOC_INTERNAL_BITMAP_H
+#define JEMALLOC_INTERNAL_BITMAP_H
+
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/sc.h"
+
+typedef unsigned long bitmap_t;
+#define LG_SIZEOF_BITMAP	LG_SIZEOF_LONG
+
+/* Maximum bitmap bit count is 2^LG_BITMAP_MAXBITS. */
+#if LG_SLAB_MAXREGS > LG_CEIL(SC_NSIZES)
+/* Maximum bitmap bit count is determined by maximum regions per slab. */
+#  define LG_BITMAP_MAXBITS	LG_SLAB_MAXREGS
+#else
+/* Maximum bitmap bit count is determined by number of extent size classes. */
+#  define LG_BITMAP_MAXBITS	LG_CEIL(SC_NSIZES)
+#endif
+#define BITMAP_MAXBITS		(ZU(1) << LG_BITMAP_MAXBITS)
+
+/* Number of bits per group. */
+#define LG_BITMAP_GROUP_NBITS		(LG_SIZEOF_BITMAP + 3)
+#define BITMAP_GROUP_NBITS		(1U << LG_BITMAP_GROUP_NBITS)
+#define BITMAP_GROUP_NBITS_MASK		(BITMAP_GROUP_NBITS-1)
+
+/*
+ * Do some analysis on how big the bitmap is before we use a tree.  For a brute
+ * force linear search, if we would have to call ffs_lu() more than 2^3 times,
+ * use a tree instead.
+ */
+#if LG_BITMAP_MAXBITS - LG_BITMAP_GROUP_NBITS > 3
+#  define BITMAP_USE_TREE
+#endif
+
+/* Number of groups required to store a given number of bits. */
+#define BITMAP_BITS2GROUPS(nbits)					\
+    (((nbits) + BITMAP_GROUP_NBITS_MASK) >> LG_BITMAP_GROUP_NBITS)
+
+/*
+ * Number of groups required at a particular level for a given number of bits.
+ */
+#define BITMAP_GROUPS_L0(nbits)						\
+    BITMAP_BITS2GROUPS(nbits)
+#define BITMAP_GROUPS_L1(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(nbits))
+#define BITMAP_GROUPS_L2(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS((nbits))))
+#define BITMAP_GROUPS_L3(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(		\
+	BITMAP_BITS2GROUPS((nbits)))))
+#define BITMAP_GROUPS_L4(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(		\
+	BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS((nbits))))))
+
+/*
+ * Assuming the number of levels, number of groups required for a given number
+ * of bits.
+ */
+#define BITMAP_GROUPS_1_LEVEL(nbits)					\
+    BITMAP_GROUPS_L0(nbits)
+#define BITMAP_GROUPS_2_LEVEL(nbits)					\
+    (BITMAP_GROUPS_1_LEVEL(nbits) + BITMAP_GROUPS_L1(nbits))
+#define BITMAP_GROUPS_3_LEVEL(nbits)					\
+    (BITMAP_GROUPS_2_LEVEL(nbits) + BITMAP_GROUPS_L2(nbits))
+#define BITMAP_GROUPS_4_LEVEL(nbits)					\
+    (BITMAP_GROUPS_3_LEVEL(nbits) + BITMAP_GROUPS_L3(nbits))
+#define BITMAP_GROUPS_5_LEVEL(nbits)					\
+    (BITMAP_GROUPS_4_LEVEL(nbits) + BITMAP_GROUPS_L4(nbits))
+
+/*
+ * Maximum number of groups required to support LG_BITMAP_MAXBITS.
+ */
+#ifdef BITMAP_USE_TREE
+
+#if LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_1_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_1_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 2
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_2_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_2_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 3
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_3_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_3_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 4
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_4_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_4_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 5
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_5_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_5_LEVEL(BITMAP_MAXBITS)
+#else
+#  error "Unsupported bitmap size"
+#endif
+
+/*
+ * Maximum number of levels possible.  This could be statically computed based
+ * on LG_BITMAP_MAXBITS:
+ *
+ * #define BITMAP_MAX_LEVELS \
+ *     (LG_BITMAP_MAXBITS / LG_SIZEOF_BITMAP) \
+ *     + !!(LG_BITMAP_MAXBITS % LG_SIZEOF_BITMAP)
+ *
+ * However, that would not allow the generic BITMAP_INFO_INITIALIZER() macro, so
+ * instead hardcode BITMAP_MAX_LEVELS to the largest number supported by the
+ * various cascading macros.  The only additional cost this incurs is some
+ * unused trailing entries in bitmap_info_t structures; the bitmaps themselves
+ * are not impacted.
+ */
+#define BITMAP_MAX_LEVELS	5
+
+#define BITMAP_INFO_INITIALIZER(nbits) {				\
+	/* nbits. */							\
+	nbits,								\
+	/* nlevels. */							\
+	(BITMAP_GROUPS_L0(nbits) > BITMAP_GROUPS_L1(nbits)) +		\
+	    (BITMAP_GROUPS_L1(nbits) > BITMAP_GROUPS_L2(nbits)) +	\
+	    (BITMAP_GROUPS_L2(nbits) > BITMAP_GROUPS_L3(nbits)) +	\
+	    (BITMAP_GROUPS_L3(nbits) > BITMAP_GROUPS_L4(nbits)) + 1,	\
+	/* levels. */							\
+	{								\
+		{0},							\
+		{BITMAP_GROUPS_L0(nbits)},				\
+		{BITMAP_GROUPS_L1(nbits) + BITMAP_GROUPS_L0(nbits)},	\
+		{BITMAP_GROUPS_L2(nbits) + BITMAP_GROUPS_L1(nbits) +	\
+		    BITMAP_GROUPS_L0(nbits)},				\
+		{BITMAP_GROUPS_L3(nbits) + BITMAP_GROUPS_L2(nbits) +	\
+		    BITMAP_GROUPS_L1(nbits) + BITMAP_GROUPS_L0(nbits)},	\
+		{BITMAP_GROUPS_L4(nbits) + BITMAP_GROUPS_L3(nbits) +	\
+		     BITMAP_GROUPS_L2(nbits) + BITMAP_GROUPS_L1(nbits)	\
+		     + BITMAP_GROUPS_L0(nbits)}				\
+	}								\
+}
+
+#else /* BITMAP_USE_TREE */
+
+#define BITMAP_GROUPS(nbits)	BITMAP_BITS2GROUPS(nbits)
+#define BITMAP_GROUPS_MAX	BITMAP_BITS2GROUPS(BITMAP_MAXBITS)
+
+#define BITMAP_INFO_INITIALIZER(nbits) {				\
+	/* nbits. */							\
+	nbits,								\
+	/* ngroups. */							\
+	BITMAP_BITS2GROUPS(nbits)					\
+}
+
+#endif /* BITMAP_USE_TREE */
+
+typedef struct bitmap_level_s {
+	/* Offset of this level's groups within the array of groups. */
+	size_t group_offset;
+} bitmap_level_t;
+
+typedef struct bitmap_info_s {
+	/* Logical number of bits in bitmap (stored at bottom level). */
+	size_t nbits;
+
+#ifdef BITMAP_USE_TREE
+	/* Number of levels necessary for nbits. */
+	unsigned nlevels;
+
+	/*
+	 * Only the first (nlevels+1) elements are used, and levels are ordered
+	 * bottom to top (e.g. the bottom level is stored in levels[0]).
+	 */
+	bitmap_level_t levels[BITMAP_MAX_LEVELS+1];
+#else /* BITMAP_USE_TREE */
+	/* Number of groups necessary for nbits. */
+	size_t ngroups;
+#endif /* BITMAP_USE_TREE */
+} bitmap_info_t;
+
+void bitmap_info_init(bitmap_info_t *binfo, size_t nbits);
+void bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo, bool fill);
+size_t bitmap_size(const bitmap_info_t *binfo);
+
+static inline bool
+bitmap_full(bitmap_t *bitmap, const bitmap_info_t *binfo) {
+#ifdef BITMAP_USE_TREE
+	size_t rgoff = binfo->levels[binfo->nlevels].group_offset - 1;
+	bitmap_t rg = bitmap[rgoff];
+	/* The bitmap is full iff the root group is 0. */
+	return (rg == 0);
+#else
+	size_t i;
+
+	for (i = 0; i < binfo->ngroups; i++) {
+		if (bitmap[i] != 0) {
+			return false;
+		}
+	}
+	return true;
+#endif
+}
+
+static inline bool
+bitmap_get(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit) {
+	size_t goff;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	g = bitmap[goff];
+	return !(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+}
+
+static inline void
+bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit) {
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	assert(!bitmap_get(bitmap, binfo, bit));
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	assert(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+	g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(bitmap_get(bitmap, binfo, bit));
+#ifdef BITMAP_USE_TREE
+	/* Propagate group state transitions up the tree. */
+	if (g == 0) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			assert(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+			g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (g != 0) {
+				break;
+			}
+		}
+	}
+#endif
+}
+
+/* ffu: find first unset >= bit. */
+static inline size_t
+bitmap_ffu(const bitmap_t *bitmap, const bitmap_info_t *binfo, size_t min_bit) {
+	assert(min_bit < binfo->nbits);
+
+#ifdef BITMAP_USE_TREE
+	size_t bit = 0;
+	for (unsigned level = binfo->nlevels; level--;) {
+		size_t lg_bits_per_group = (LG_BITMAP_GROUP_NBITS * (level +
+		    1));
+		bitmap_t group = bitmap[binfo->levels[level].group_offset + (bit
+		    >> lg_bits_per_group)];
+		unsigned group_nmask = (unsigned)(((min_bit > bit) ? (min_bit -
+		    bit) : 0) >> (lg_bits_per_group - LG_BITMAP_GROUP_NBITS));
+		assert(group_nmask <= BITMAP_GROUP_NBITS);
+		bitmap_t group_mask = ~((1LU << group_nmask) - 1);
+		bitmap_t group_masked = group & group_mask;
+		if (group_masked == 0LU) {
+			if (group == 0LU) {
+				return binfo->nbits;
+			}
+			/*
+			 * min_bit was preceded by one or more unset bits in
+			 * this group, but there are no other unset bits in this
+			 * group.  Try again starting at the first bit of the
+			 * next sibling.  This will recurse at most once per
+			 * non-root level.
+			 */
+			size_t sib_base = bit + (ZU(1) << lg_bits_per_group);
+			assert(sib_base > min_bit);
+			assert(sib_base > bit);
+			if (sib_base >= binfo->nbits) {
+				return binfo->nbits;
+			}
+			return bitmap_ffu(bitmap, binfo, sib_base);
+		}
+		bit += ((size_t)(ffs_lu(group_masked) - 1)) <<
+		    (lg_bits_per_group - LG_BITMAP_GROUP_NBITS);
+	}
+	assert(bit >= min_bit);
+	assert(bit < binfo->nbits);
+	return bit;
+#else
+	size_t i = min_bit >> LG_BITMAP_GROUP_NBITS;
+	bitmap_t g = bitmap[i] & ~((1LU << (min_bit & BITMAP_GROUP_NBITS_MASK))
+	    - 1);
+	size_t bit;
+	do {
+		bit = ffs_lu(g);
+		if (bit != 0) {
+			return (i << LG_BITMAP_GROUP_NBITS) + (bit - 1);
+		}
+		i++;
+		g = bitmap[i];
+	} while (i < binfo->ngroups);
+	return binfo->nbits;
+#endif
+}
+
+/* sfu: set first unset. */
+static inline size_t
+bitmap_sfu(bitmap_t *bitmap, const bitmap_info_t *binfo) {
+	size_t bit;
+	bitmap_t g;
+	unsigned i;
+
+	assert(!bitmap_full(bitmap, binfo));
+
+#ifdef BITMAP_USE_TREE
+	i = binfo->nlevels - 1;
+	g = bitmap[binfo->levels[i].group_offset];
+	bit = ffs_lu(g) - 1;
+	while (i > 0) {
+		i--;
+		g = bitmap[binfo->levels[i].group_offset + bit];
+		bit = (bit << LG_BITMAP_GROUP_NBITS) + (ffs_lu(g) - 1);
+	}
+#else
+	i = 0;
+	g = bitmap[0];
+	while ((bit = ffs_lu(g)) == 0) {
+		i++;
+		g = bitmap[i];
+	}
+	bit = (i << LG_BITMAP_GROUP_NBITS) + (bit - 1);
+#endif
+	bitmap_set(bitmap, binfo, bit);
+	return bit;
+}
+
+static inline void
+bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit) {
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+	UNUSED bool propagate;
+
+	assert(bit < binfo->nbits);
+	assert(bitmap_get(bitmap, binfo, bit));
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	propagate = (g == 0);
+	assert((g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK))) == 0);
+	g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(!bitmap_get(bitmap, binfo, bit));
+#ifdef BITMAP_USE_TREE
+	/* Propagate group state transitions up the tree. */
+	if (propagate) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			propagate = (g == 0);
+			assert((g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)))
+			    == 0);
+			g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (!propagate) {
+				break;
+			}
+		}
+	}
+#endif /* BITMAP_USE_TREE */
+}
+
+#endif /* JEMALLOC_INTERNAL_BITMAP_H */
diff --git a/jemalloc/include/jemalloc/internal/cache_bin.h b/jemalloc/include/jemalloc/internal/cache_bin.h
new file mode 100644
index 0000000..d14556a
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/cache_bin.h
@@ -0,0 +1,131 @@
+#ifndef JEMALLOC_INTERNAL_CACHE_BIN_H
+#define JEMALLOC_INTERNAL_CACHE_BIN_H
+
+#include "jemalloc/internal/ql.h"
+
+/*
+ * The cache_bins are the mechanism that the tcache and the arena use to
+ * communicate.  The tcache fills from and flushes to the arena by passing a
+ * cache_bin_t to fill/flush.  When the arena needs to pull stats from the
+ * tcaches associated with it, it does so by iterating over its
+ * cache_bin_array_descriptor_t objects and reading out per-bin stats it
+ * contains.  This makes it so that the arena need not know about the existence
+ * of the tcache at all.
+ */
+
+
+/*
+ * The count of the number of cached allocations in a bin.  We make this signed
+ * so that negative numbers can encode "invalid" states (e.g. a low water mark
+ * of -1 for a cache that has been depleted).
+ */
+typedef int32_t cache_bin_sz_t;
+
+typedef struct cache_bin_stats_s cache_bin_stats_t;
+struct cache_bin_stats_s {
+	/*
+	 * Number of allocation requests that corresponded to the size of this
+	 * bin.
+	 */
+	uint64_t nrequests;
+};
+
+/*
+ * Read-only information associated with each element of tcache_t's tbins array
+ * is stored separately, mainly to reduce memory usage.
+ */
+typedef struct cache_bin_info_s cache_bin_info_t;
+struct cache_bin_info_s {
+	/* Upper limit on ncached. */
+	cache_bin_sz_t ncached_max;
+};
+
+typedef struct cache_bin_s cache_bin_t;
+struct cache_bin_s {
+	/* Min # cached since last GC. */
+	cache_bin_sz_t low_water;
+	/* # of cached objects. */
+	cache_bin_sz_t ncached;
+	/*
+	 * ncached and stats are both modified frequently.  Let's keep them
+	 * close so that they have a higher chance of being on the same
+	 * cacheline, thus less write-backs.
+	 */
+	cache_bin_stats_t tstats;
+	/*
+	 * Stack of available objects.
+	 *
+	 * To make use of adjacent cacheline prefetch, the items in the avail
+	 * stack goes to higher address for newer allocations.  avail points
+	 * just above the available space, which means that
+	 * avail[-ncached, ... -1] are available items and the lowest item will
+	 * be allocated first.
+	 */
+	void **avail;
+};
+
+typedef struct cache_bin_array_descriptor_s cache_bin_array_descriptor_t;
+struct cache_bin_array_descriptor_s {
+	/*
+	 * The arena keeps a list of the cache bins associated with it, for
+	 * stats collection.
+	 */
+	ql_elm(cache_bin_array_descriptor_t) link;
+	/* Pointers to the tcache bins. */
+	cache_bin_t *bins_small;
+	cache_bin_t *bins_large;
+};
+
+static inline void
+cache_bin_array_descriptor_init(cache_bin_array_descriptor_t *descriptor,
+    cache_bin_t *bins_small, cache_bin_t *bins_large) {
+	ql_elm_new(descriptor, link);
+	descriptor->bins_small = bins_small;
+	descriptor->bins_large = bins_large;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+cache_bin_alloc_easy(cache_bin_t *bin, bool *success) {
+	void *ret;
+
+	bin->ncached--;
+
+	/*
+	 * Check for both bin->ncached == 0 and ncached < low_water
+	 * in a single branch.
+	 */
+	if (unlikely(bin->ncached <= bin->low_water)) {
+		bin->low_water = bin->ncached;
+		if (bin->ncached == -1) {
+			bin->ncached = 0;
+			*success = false;
+			return NULL;
+		}
+	}
+
+	/*
+	 * success (instead of ret) should be checked upon the return of this
+	 * function.  We avoid checking (ret == NULL) because there is never a
+	 * null stored on the avail stack (which is unknown to the compiler),
+	 * and eagerly checking ret would cause pipeline stall (waiting for the
+	 * cacheline).
+	 */
+	*success = true;
+	ret = *(bin->avail - (bin->ncached + 1));
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+cache_bin_dalloc_easy(cache_bin_t *bin, cache_bin_info_t *bin_info, void *ptr) {
+	if (unlikely(bin->ncached == bin_info->ncached_max)) {
+		return false;
+	}
+	assert(bin->ncached < bin_info->ncached_max);
+	bin->ncached++;
+	*(bin->avail - bin->ncached) = ptr;
+
+	return true;
+}
+
+#endif /* JEMALLOC_INTERNAL_CACHE_BIN_H */
diff --git a/jemalloc/include/jemalloc/internal/ckh.h b/jemalloc/include/jemalloc/internal/ckh.h
new file mode 100644
index 0000000..7b3850b
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/ckh.h
@@ -0,0 +1,101 @@
+#ifndef JEMALLOC_INTERNAL_CKH_H
+#define JEMALLOC_INTERNAL_CKH_H
+
+#include "jemalloc/internal/tsd.h"
+
+/* Cuckoo hashing implementation.  Skip to the end for the interface. */
+
+/******************************************************************************/
+/* INTERNAL DEFINITIONS -- IGNORE */
+/******************************************************************************/
+
+/* Maintain counters used to get an idea of performance. */
+/* #define CKH_COUNT */
+/* Print counter values in ckh_delete() (requires CKH_COUNT). */
+/* #define CKH_VERBOSE */
+
+/*
+ * There are 2^LG_CKH_BUCKET_CELLS cells in each hash table bucket.  Try to fit
+ * one bucket per L1 cache line.
+ */
+#define LG_CKH_BUCKET_CELLS (LG_CACHELINE - LG_SIZEOF_PTR - 1)
+
+/* Typedefs to allow easy function pointer passing. */
+typedef void ckh_hash_t (const void *, size_t[2]);
+typedef bool ckh_keycomp_t (const void *, const void *);
+
+/* Hash table cell. */
+typedef struct {
+	const void *key;
+	const void *data;
+} ckhc_t;
+
+/* The hash table itself. */
+typedef struct {
+#ifdef CKH_COUNT
+	/* Counters used to get an idea of performance. */
+	uint64_t ngrows;
+	uint64_t nshrinks;
+	uint64_t nshrinkfails;
+	uint64_t ninserts;
+	uint64_t nrelocs;
+#endif
+
+	/* Used for pseudo-random number generation. */
+	uint64_t prng_state;
+
+	/* Total number of items. */
+	size_t count;
+
+	/*
+	 * Minimum and current number of hash table buckets.  There are
+	 * 2^LG_CKH_BUCKET_CELLS cells per bucket.
+	 */
+	unsigned lg_minbuckets;
+	unsigned lg_curbuckets;
+
+	/* Hash and comparison functions. */
+	ckh_hash_t *hash;
+	ckh_keycomp_t *keycomp;
+
+	/* Hash table with 2^lg_curbuckets buckets. */
+	ckhc_t *tab;
+} ckh_t;
+
+/******************************************************************************/
+/* BEGIN PUBLIC API */
+/******************************************************************************/
+
+/* Lifetime management.  Minitems is the initial capacity. */
+bool ckh_new(tsd_t *tsd, ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
+    ckh_keycomp_t *keycomp);
+void ckh_delete(tsd_t *tsd, ckh_t *ckh);
+
+/* Get the number of elements in the set. */
+size_t ckh_count(ckh_t *ckh);
+
+/*
+ * To iterate over the elements in the table, initialize *tabind to 0 and call
+ * this function until it returns true.  Each call that returns false will
+ * update *key and *data to the next element in the table, assuming the pointers
+ * are non-NULL.
+ */
+bool ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data);
+
+/*
+ * Basic hash table operations -- insert, removal, lookup.  For ckh_remove and
+ * ckh_search, key or data can be NULL.  The hash-table only stores pointers to
+ * the key and value, and doesn't do any lifetime management.
+ */
+bool ckh_insert(tsd_t *tsd, ckh_t *ckh, const void *key, const void *data);
+bool ckh_remove(tsd_t *tsd, ckh_t *ckh, const void *searchkey, void **key,
+    void **data);
+bool ckh_search(ckh_t *ckh, const void *searchkey, void **key, void **data);
+
+/* Some useful hash and comparison functions for strings and pointers. */
+void ckh_string_hash(const void *key, size_t r_hash[2]);
+bool ckh_string_keycomp(const void *k1, const void *k2);
+void ckh_pointer_hash(const void *key, size_t r_hash[2]);
+bool ckh_pointer_keycomp(const void *k1, const void *k2);
+
+#endif /* JEMALLOC_INTERNAL_CKH_H */
diff --git a/jemalloc/include/jemalloc/internal/ctl.h b/jemalloc/include/jemalloc/internal/ctl.h
new file mode 100644
index 0000000..1d1aacc
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/ctl.h
@@ -0,0 +1,134 @@
+#ifndef JEMALLOC_INTERNAL_CTL_H
+#define JEMALLOC_INTERNAL_CTL_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/stats.h"
+
+/* Maximum ctl tree depth. */
+#define CTL_MAX_DEPTH	7
+
+typedef struct ctl_node_s {
+	bool named;
+} ctl_node_t;
+
+typedef struct ctl_named_node_s {
+	ctl_node_t node;
+	const char *name;
+	/* If (nchildren == 0), this is a terminal node. */
+	size_t nchildren;
+	const ctl_node_t *children;
+	int (*ctl)(tsd_t *, const size_t *, size_t, void *, size_t *, void *,
+	    size_t);
+} ctl_named_node_t;
+
+typedef struct ctl_indexed_node_s {
+	struct ctl_node_s node;
+	const ctl_named_node_t *(*index)(tsdn_t *, const size_t *, size_t,
+	    size_t);
+} ctl_indexed_node_t;
+
+typedef struct ctl_arena_stats_s {
+	arena_stats_t astats;
+
+	/* Aggregate stats for small size classes, based on bin stats. */
+	size_t allocated_small;
+	uint64_t nmalloc_small;
+	uint64_t ndalloc_small;
+	uint64_t nrequests_small;
+	uint64_t nfills_small;
+	uint64_t nflushes_small;
+
+	bin_stats_t bstats[SC_NBINS];
+	arena_stats_large_t lstats[SC_NSIZES - SC_NBINS];
+	arena_stats_extents_t estats[SC_NPSIZES];
+} ctl_arena_stats_t;
+
+typedef struct ctl_stats_s {
+	size_t allocated;
+	size_t active;
+	size_t metadata;
+	size_t metadata_thp;
+	size_t resident;
+	size_t mapped;
+	size_t retained;
+
+	background_thread_stats_t background_thread;
+	mutex_prof_data_t mutex_prof_data[mutex_prof_num_global_mutexes];
+} ctl_stats_t;
+
+typedef struct ctl_arena_s ctl_arena_t;
+struct ctl_arena_s {
+	unsigned arena_ind;
+	bool initialized;
+	ql_elm(ctl_arena_t) destroyed_link;
+
+	/* Basic stats, supported even if !config_stats. */
+	unsigned nthreads;
+	const char *dss;
+	ssize_t dirty_decay_ms;
+	ssize_t muzzy_decay_ms;
+	size_t pactive;
+	size_t pdirty;
+	size_t pmuzzy;
+
+	/* NULL if !config_stats. */
+	ctl_arena_stats_t *astats;
+};
+
+typedef struct ctl_arenas_s {
+	uint64_t epoch;
+	unsigned narenas;
+	ql_head(ctl_arena_t) destroyed;
+
+	/*
+	 * Element 0 corresponds to merged stats for extant arenas (accessed via
+	 * MALLCTL_ARENAS_ALL), element 1 corresponds to merged stats for
+	 * destroyed arenas (accessed via MALLCTL_ARENAS_DESTROYED), and the
+	 * remaining MALLOCX_ARENA_LIMIT elements correspond to arenas.
+	 */
+	ctl_arena_t *arenas[2 + MALLOCX_ARENA_LIMIT];
+} ctl_arenas_t;
+
+int ctl_byname(tsd_t *tsd, const char *name, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen);
+int ctl_nametomib(tsd_t *tsd, const char *name, size_t *mibp, size_t *miblenp);
+
+int ctl_bymib(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen);
+bool ctl_boot(void);
+void ctl_prefork(tsdn_t *tsdn);
+void ctl_postfork_parent(tsdn_t *tsdn);
+void ctl_postfork_child(tsdn_t *tsdn);
+
+#define xmallctl(name, oldp, oldlenp, newp, newlen) do {		\
+	if (je_mallctl(name, oldp, oldlenp, newp, newlen)		\
+	    != 0) {							\
+		malloc_printf(						\
+		    "<jemalloc>: Failure in xmallctl(\"%s\", ...)\n",	\
+		    name);						\
+		abort();						\
+	}								\
+} while (0)
+
+#define xmallctlnametomib(name, mibp, miblenp) do {			\
+	if (je_mallctlnametomib(name, mibp, miblenp) != 0) {		\
+		malloc_printf("<jemalloc>: Failure in "			\
+		    "xmallctlnametomib(\"%s\", ...)\n", name);		\
+		abort();						\
+	}								\
+} while (0)
+
+#define xmallctlbymib(mib, miblen, oldp, oldlenp, newp, newlen) do {	\
+	if (je_mallctlbymib(mib, miblen, oldp, oldlenp, newp,		\
+	    newlen) != 0) {						\
+		malloc_write(						\
+		    "<jemalloc>: Failure in xmallctlbymib()\n");	\
+		abort();						\
+	}								\
+} while (0)
+
+#endif /* JEMALLOC_INTERNAL_CTL_H */
diff --git a/jemalloc/include/jemalloc/internal/div.h b/jemalloc/include/jemalloc/internal/div.h
new file mode 100644
index 0000000..aebae93
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/div.h
@@ -0,0 +1,41 @@
+#ifndef JEMALLOC_INTERNAL_DIV_H
+#define JEMALLOC_INTERNAL_DIV_H
+
+#include "jemalloc/internal/assert.h"
+
+/*
+ * This module does the division that computes the index of a region in a slab,
+ * given its offset relative to the base.
+ * That is, given a divisor d, an n = i * d (all integers), we'll return i.
+ * We do some pre-computation to do this more quickly than a CPU division
+ * instruction.
+ * We bound n < 2^32, and don't support dividing by one.
+ */
+
+typedef struct div_info_s div_info_t;
+struct div_info_s {
+	uint32_t magic;
+#ifdef JEMALLOC_DEBUG
+	size_t d;
+#endif
+};
+
+void div_init(div_info_t *div_info, size_t divisor);
+
+static inline size_t
+div_compute(div_info_t *div_info, size_t n) {
+	assert(n <= (uint32_t)-1);
+	/*
+	 * This generates, e.g. mov; imul; shr on x86-64. On a 32-bit machine,
+	 * the compilers I tried were all smart enough to turn this into the
+	 * appropriate "get the high 32 bits of the result of a multiply" (e.g.
+	 * mul; mov edx eax; on x86, umull on arm, etc.).
+	 */
+	size_t i = ((uint64_t)n * (uint64_t)div_info->magic) >> 32;
+#ifdef JEMALLOC_DEBUG
+	assert(i * div_info->d == n);
+#endif
+	return i;
+}
+
+#endif /* JEMALLOC_INTERNAL_DIV_H */
diff --git a/jemalloc/include/jemalloc/internal/emitter.h b/jemalloc/include/jemalloc/internal/emitter.h
new file mode 100644
index 0000000..d140d38
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/emitter.h
@@ -0,0 +1,486 @@
+#ifndef JEMALLOC_INTERNAL_EMITTER_H
+#define JEMALLOC_INTERNAL_EMITTER_H
+
+#include "jemalloc/internal/ql.h"
+
+typedef enum emitter_output_e emitter_output_t;
+enum emitter_output_e {
+	emitter_output_json,
+	emitter_output_table
+};
+
+typedef enum emitter_justify_e emitter_justify_t;
+enum emitter_justify_e {
+	emitter_justify_left,
+	emitter_justify_right,
+	/* Not for users; just to pass to internal functions. */
+	emitter_justify_none
+};
+
+typedef enum emitter_type_e emitter_type_t;
+enum emitter_type_e {
+	emitter_type_bool,
+	emitter_type_int,
+	emitter_type_unsigned,
+	emitter_type_uint32,
+	emitter_type_uint64,
+	emitter_type_size,
+	emitter_type_ssize,
+	emitter_type_string,
+	/*
+	 * A title is a column title in a table; it's just a string, but it's
+	 * not quoted.
+	 */
+	emitter_type_title,
+};
+
+typedef struct emitter_col_s emitter_col_t;
+struct emitter_col_s {
+	/* Filled in by the user. */
+	emitter_justify_t justify;
+	int width;
+	emitter_type_t type;
+	union {
+		bool bool_val;
+		int int_val;
+		unsigned unsigned_val;
+		uint32_t uint32_val;
+		uint32_t uint32_t_val;
+		uint64_t uint64_val;
+		uint64_t uint64_t_val;
+		size_t size_val;
+		ssize_t ssize_val;
+		const char *str_val;
+	};
+
+	/* Filled in by initialization. */
+	ql_elm(emitter_col_t) link;
+};
+
+typedef struct emitter_row_s emitter_row_t;
+struct emitter_row_s {
+	ql_head(emitter_col_t) cols;
+};
+
+typedef struct emitter_s emitter_t;
+struct emitter_s {
+	emitter_output_t output;
+	/* The output information. */
+	void (*write_cb)(void *, const char *);
+	void *cbopaque;
+	int nesting_depth;
+	/* True if we've already emitted a value at the given depth. */
+	bool item_at_depth;
+	/* True if we emitted a key and will emit corresponding value next. */
+	bool emitted_key;
+};
+
+/* Internal convenience function.  Write to the emitter the given string. */
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static inline void
+emitter_printf(emitter_t *emitter, const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(emitter->write_cb, emitter->cbopaque, format, ap);
+	va_end(ap);
+}
+
+static inline const char * JEMALLOC_FORMAT_ARG(3)
+emitter_gen_fmt(char *out_fmt, size_t out_size, const char *fmt_specifier,
+    emitter_justify_t justify, int width) {
+	size_t written;
+	fmt_specifier++;
+	if (justify == emitter_justify_none) {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%%s", fmt_specifier);
+	} else if (justify == emitter_justify_left) {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%-%d%s", width, fmt_specifier);
+	} else {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%%d%s", width, fmt_specifier);
+	}
+	/* Only happens in case of bad format string, which *we* choose. */
+	assert(written <  out_size);
+	return out_fmt;
+}
+
+/*
+ * Internal.  Emit the given value type in the relevant encoding (so that the
+ * bool true gets mapped to json "true", but the string "true" gets mapped to
+ * json "\"true\"", for instance.
+ *
+ * Width is ignored if justify is emitter_justify_none.
+ */
+static inline void
+emitter_print_value(emitter_t *emitter, emitter_justify_t justify, int width,
+    emitter_type_t value_type, const void *value) {
+	size_t str_written;
+#define BUF_SIZE 256
+#define FMT_SIZE 10
+	/*
+	 * We dynamically generate a format string to emit, to let us use the
+	 * snprintf machinery.  This is kinda hacky, but gets the job done
+	 * quickly without having to think about the various snprintf edge
+	 * cases.
+	 */
+	char fmt[FMT_SIZE];
+	char buf[BUF_SIZE];
+
+#define EMIT_SIMPLE(type, format)					\
+	emitter_printf(emitter,						\
+	    emitter_gen_fmt(fmt, FMT_SIZE, format, justify, width),	\
+	    *(const type *)value);
+
+	switch (value_type) {
+	case emitter_type_bool:
+		emitter_printf(emitter,
+		    emitter_gen_fmt(fmt, FMT_SIZE, "%s", justify, width),
+		    *(const bool *)value ?  "true" : "false");
+		break;
+	case emitter_type_int:
+		EMIT_SIMPLE(int, "%d")
+		break;
+	case emitter_type_unsigned:
+		EMIT_SIMPLE(unsigned, "%u")
+		break;
+	case emitter_type_ssize:
+		EMIT_SIMPLE(ssize_t, "%zd")
+		break;
+	case emitter_type_size:
+		EMIT_SIMPLE(size_t, "%zu")
+		break;
+	case emitter_type_string:
+		str_written = malloc_snprintf(buf, BUF_SIZE, "\"%s\"",
+		    *(const char *const *)value);
+		/*
+		 * We control the strings we output; we shouldn't get anything
+		 * anywhere near the fmt size.
+		 */
+		assert(str_written < BUF_SIZE);
+		emitter_printf(emitter,
+		    emitter_gen_fmt(fmt, FMT_SIZE, "%s", justify, width), buf);
+		break;
+	case emitter_type_uint32:
+		EMIT_SIMPLE(uint32_t, "%" FMTu32)
+		break;
+	case emitter_type_uint64:
+		EMIT_SIMPLE(uint64_t, "%" FMTu64)
+		break;
+	case emitter_type_title:
+		EMIT_SIMPLE(char *const, "%s");
+		break;
+	default:
+		unreachable();
+	}
+#undef BUF_SIZE
+#undef FMT_SIZE
+}
+
+
+/* Internal functions.  In json mode, tracks nesting state. */
+static inline void
+emitter_nest_inc(emitter_t *emitter) {
+	emitter->nesting_depth++;
+	emitter->item_at_depth = false;
+}
+
+static inline void
+emitter_nest_dec(emitter_t *emitter) {
+	emitter->nesting_depth--;
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_indent(emitter_t *emitter) {
+	int amount = emitter->nesting_depth;
+	const char *indent_str;
+	if (emitter->output == emitter_output_json) {
+		indent_str = "\t";
+	} else {
+		amount *= 2;
+		indent_str = " ";
+	}
+	for (int i = 0; i < amount; i++) {
+		emitter_printf(emitter, "%s", indent_str);
+	}
+}
+
+static inline void
+emitter_json_key_prefix(emitter_t *emitter) {
+	if (emitter->emitted_key) {
+		emitter->emitted_key = false;
+		return;
+	}
+	emitter_printf(emitter, "%s\n", emitter->item_at_depth ? "," : "");
+	emitter_indent(emitter);
+}
+
+/******************************************************************************/
+/* Public functions for emitter_t. */
+
+static inline void
+emitter_init(emitter_t *emitter, emitter_output_t emitter_output,
+    void (*write_cb)(void *, const char *), void *cbopaque) {
+	emitter->output = emitter_output;
+	emitter->write_cb = write_cb;
+	emitter->cbopaque = cbopaque;
+	emitter->item_at_depth = false;
+	emitter->emitted_key = false;
+	emitter->nesting_depth = 0;
+}
+
+/******************************************************************************/
+/* JSON public API. */
+
+/*
+ * Emits a key (e.g. as appears in an object). The next json entity emitted will
+ * be the corresponding value.
+ */
+static inline void
+emitter_json_key(emitter_t *emitter, const char *json_key) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "\"%s\": ", json_key);
+		emitter->emitted_key = true;
+	}
+}
+
+static inline void
+emitter_json_value(emitter_t *emitter, emitter_type_t value_type,
+    const void *value) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_print_value(emitter, emitter_justify_none, -1,
+		    value_type, value);
+		emitter->item_at_depth = true;
+	}
+}
+
+/* Shorthand for calling emitter_json_key and then emitter_json_value. */
+static inline void
+emitter_json_kv(emitter_t *emitter, const char *json_key,
+    emitter_type_t value_type, const void *value) {
+	emitter_json_key(emitter, json_key);
+	emitter_json_value(emitter, value_type, value);
+}
+
+static inline void
+emitter_json_array_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "[");
+		emitter_nest_inc(emitter);
+	}
+}
+
+/* Shorthand for calling emitter_json_key and then emitter_json_array_begin. */
+static inline void
+emitter_json_array_kv_begin(emitter_t *emitter, const char *json_key) {
+	emitter_json_key(emitter, json_key);
+	emitter_json_array_begin(emitter);
+}
+
+static inline void
+emitter_json_array_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n");
+		emitter_indent(emitter);
+		emitter_printf(emitter, "]");
+	}
+}
+
+static inline void
+emitter_json_object_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "{");
+		emitter_nest_inc(emitter);
+	}
+}
+
+/* Shorthand for calling emitter_json_key and then emitter_json_object_begin. */
+static inline void
+emitter_json_object_kv_begin(emitter_t *emitter, const char *json_key) {
+	emitter_json_key(emitter, json_key);
+	emitter_json_object_begin(emitter);
+}
+
+static inline void
+emitter_json_object_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n");
+		emitter_indent(emitter);
+		emitter_printf(emitter, "}");
+	}
+}
+
+
+/******************************************************************************/
+/* Table public API. */
+
+static inline void
+emitter_table_dict_begin(emitter_t *emitter, const char *table_key) {
+	if (emitter->output == emitter_output_table) {
+		emitter_indent(emitter);
+		emitter_printf(emitter, "%s\n", table_key);
+		emitter_nest_inc(emitter);
+	}
+}
+
+static inline void
+emitter_table_dict_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_table) {
+		emitter_nest_dec(emitter);
+	}
+}
+
+static inline void
+emitter_table_kv_note(emitter_t *emitter, const char *table_key,
+    emitter_type_t value_type, const void *value,
+    const char *table_note_key, emitter_type_t table_note_value_type,
+    const void *table_note_value) {
+	if (emitter->output == emitter_output_table) {
+		emitter_indent(emitter);
+		emitter_printf(emitter, "%s: ", table_key);
+		emitter_print_value(emitter, emitter_justify_none, -1,
+		    value_type, value);
+		if (table_note_key != NULL) {
+			emitter_printf(emitter, " (%s: ", table_note_key);
+			emitter_print_value(emitter, emitter_justify_none, -1,
+			    table_note_value_type, table_note_value);
+			emitter_printf(emitter, ")");
+		}
+		emitter_printf(emitter, "\n");
+	}
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_table_kv(emitter_t *emitter, const char *table_key,
+    emitter_type_t value_type, const void *value) {
+	emitter_table_kv_note(emitter, table_key, value_type, value, NULL,
+	    emitter_type_bool, NULL);
+}
+
+
+/* Write to the emitter the given string, but only in table mode. */
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static inline void
+emitter_table_printf(emitter_t *emitter, const char *format, ...) {
+	if (emitter->output == emitter_output_table) {
+		va_list ap;
+		va_start(ap, format);
+		malloc_vcprintf(emitter->write_cb, emitter->cbopaque, format, ap);
+		va_end(ap);
+	}
+}
+
+static inline void
+emitter_table_row(emitter_t *emitter, emitter_row_t *row) {
+	if (emitter->output != emitter_output_table) {
+		return;
+	}
+	emitter_col_t *col;
+	ql_foreach(col, &row->cols, link) {
+		emitter_print_value(emitter, col->justify, col->width,
+		    col->type, (const void *)&col->bool_val);
+	}
+	emitter_table_printf(emitter, "\n");
+}
+
+static inline void
+emitter_row_init(emitter_row_t *row) {
+	ql_new(&row->cols);
+}
+
+static inline void
+emitter_col_init(emitter_col_t *col, emitter_row_t *row) {
+	ql_elm_new(col, link);
+	ql_tail_insert(&row->cols, col, link);
+}
+
+
+/******************************************************************************/
+/*
+ * Generalized public API. Emits using either JSON or table, according to
+ * settings in the emitter_t. */
+
+/*
+ * Note emits a different kv pair as well, but only in table mode.  Omits the
+ * note if table_note_key is NULL.
+ */
+static inline void
+emitter_kv_note(emitter_t *emitter, const char *json_key, const char *table_key,
+    emitter_type_t value_type, const void *value,
+    const char *table_note_key, emitter_type_t table_note_value_type,
+    const void *table_note_value) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key(emitter, json_key);
+		emitter_json_value(emitter, value_type, value);
+	} else {
+		emitter_table_kv_note(emitter, table_key, value_type, value,
+		    table_note_key, table_note_value_type, table_note_value);
+	}
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_kv(emitter_t *emitter, const char *json_key, const char *table_key,
+    emitter_type_t value_type, const void *value) {
+	emitter_kv_note(emitter, json_key, table_key, value_type, value, NULL,
+	    emitter_type_bool, NULL);
+}
+
+static inline void
+emitter_dict_begin(emitter_t *emitter, const char *json_key,
+    const char *table_header) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key(emitter, json_key);
+		emitter_json_object_begin(emitter);
+	} else {
+		emitter_table_dict_begin(emitter, table_header);
+	}
+}
+
+static inline void
+emitter_dict_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_object_end(emitter);
+	} else {
+		emitter_table_dict_end(emitter);
+	}
+}
+
+static inline void
+emitter_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth == 0);
+		emitter_printf(emitter, "{");
+		emitter_nest_inc(emitter);
+	} else {
+		/*
+		 * This guarantees that we always call write_cb at least once.
+		 * This is useful if some invariant is established by each call
+		 * to write_cb, but doesn't hold initially: e.g., some buffer
+		 * holds a null-terminated string.
+		 */
+		emitter_printf(emitter, "%s", "");
+	}
+}
+
+static inline void
+emitter_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth == 1);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n}\n");
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_EMITTER_H */
diff --git a/jemalloc/include/jemalloc/internal/extent_dss.h b/jemalloc/include/jemalloc/internal/extent_dss.h
new file mode 100644
index 0000000..e8f02ce
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/extent_dss.h
@@ -0,0 +1,26 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_DSS_H
+#define JEMALLOC_INTERNAL_EXTENT_DSS_H
+
+typedef enum {
+	dss_prec_disabled  = 0,
+	dss_prec_primary   = 1,
+	dss_prec_secondary = 2,
+
+	dss_prec_limit     = 3
+} dss_prec_t;
+#define DSS_PREC_DEFAULT dss_prec_secondary
+#define DSS_DEFAULT "secondary"
+
+extern const char *dss_prec_names[];
+
+extern const char *opt_dss;
+
+dss_prec_t extent_dss_prec_get(void);
+bool extent_dss_prec_set(dss_prec_t dss_prec);
+void *extent_alloc_dss(tsdn_t *tsdn, arena_t *arena, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit);
+bool extent_in_dss(void *addr);
+bool extent_dss_mergeable(void *addr_a, void *addr_b);
+void extent_dss_boot(void);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_DSS_H */
diff --git a/jemalloc/include/jemalloc/internal/extent_externs.h b/jemalloc/include/jemalloc/internal/extent_externs.h
new file mode 100644
index 0000000..8aba576
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/extent_externs.h
@@ -0,0 +1,83 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTENT_EXTERNS_H
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/rtree.h"
+
+extern size_t opt_lg_extent_max_active_fit;
+
+extern rtree_t extents_rtree;
+extern const extent_hooks_t extent_hooks_default;
+extern mutex_pool_t extent_mutex_pool;
+
+extent_t *extent_alloc(tsdn_t *tsdn, arena_t *arena);
+void extent_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *extent);
+
+extent_hooks_t *extent_hooks_get(arena_t *arena);
+extent_hooks_t *extent_hooks_set(tsd_t *tsd, arena_t *arena,
+    extent_hooks_t *extent_hooks);
+
+#ifdef JEMALLOC_JET
+size_t extent_size_quantize_floor(size_t size);
+size_t extent_size_quantize_ceil(size_t size);
+#endif
+
+ph_proto(, extent_avail_, extent_tree_t, extent_t)
+ph_proto(, extent_heap_, extent_heap_t, extent_t)
+
+bool extents_init(tsdn_t *tsdn, extents_t *extents, extent_state_t state,
+    bool delay_coalesce);
+extent_state_t extents_state_get(const extents_t *extents);
+size_t extents_npages_get(extents_t *extents);
+/* Get the number of extents in the given page size index. */
+size_t extents_nextents_get(extents_t *extents, pszind_t ind);
+/* Get the sum total bytes of the extents in the given page size index. */
+size_t extents_nbytes_get(extents_t *extents, pszind_t ind);
+extent_t *extents_alloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, void *new_addr,
+    size_t size, size_t pad, size_t alignment, bool slab, szind_t szind,
+    bool *zero, bool *commit);
+void extents_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, extent_t *extent);
+extent_t *extents_evict(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, size_t npages_min);
+void extents_prefork(tsdn_t *tsdn, extents_t *extents);
+void extents_postfork_parent(tsdn_t *tsdn, extents_t *extents);
+void extents_postfork_child(tsdn_t *tsdn, extents_t *extents);
+extent_t *extent_alloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit);
+void extent_dalloc_gap(tsdn_t *tsdn, arena_t *arena, extent_t *extent);
+void extent_dalloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+void extent_destroy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+bool extent_commit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_decommit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_purge_lazy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_purge_forced_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+extent_t *extent_split_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b);
+bool extent_merge_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b);
+
+bool extent_boot(void);
+
+void extent_util_stats_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size);
+void extent_util_stats_verbose_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size,
+    size_t *bin_nfree, size_t *bin_nregs, void **slabcur_addr);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/extent_inlines.h b/jemalloc/include/jemalloc/internal/extent_inlines.h
new file mode 100644
index 0000000..77fa4c4
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/extent_inlines.h
@@ -0,0 +1,501 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_INLINES_H
+#define JEMALLOC_INTERNAL_EXTENT_INLINES_H
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/sz.h"
+
+static inline void
+extent_lock(tsdn_t *tsdn, extent_t *extent) {
+	assert(extent != NULL);
+	mutex_pool_lock(tsdn, &extent_mutex_pool, (uintptr_t)extent);
+}
+
+static inline void
+extent_unlock(tsdn_t *tsdn, extent_t *extent) {
+	assert(extent != NULL);
+	mutex_pool_unlock(tsdn, &extent_mutex_pool, (uintptr_t)extent);
+}
+
+static inline void
+extent_lock2(tsdn_t *tsdn, extent_t *extent1, extent_t *extent2) {
+	assert(extent1 != NULL && extent2 != NULL);
+	mutex_pool_lock2(tsdn, &extent_mutex_pool, (uintptr_t)extent1,
+	    (uintptr_t)extent2);
+}
+
+static inline void
+extent_unlock2(tsdn_t *tsdn, extent_t *extent1, extent_t *extent2) {
+	assert(extent1 != NULL && extent2 != NULL);
+	mutex_pool_unlock2(tsdn, &extent_mutex_pool, (uintptr_t)extent1,
+	    (uintptr_t)extent2);
+}
+
+static inline unsigned
+extent_arena_ind_get(const extent_t *extent) {
+	unsigned arena_ind = (unsigned)((extent->e_bits &
+	    EXTENT_BITS_ARENA_MASK) >> EXTENT_BITS_ARENA_SHIFT);
+	assert(arena_ind < MALLOCX_ARENA_LIMIT);
+
+	return arena_ind;
+}
+
+static inline arena_t *
+extent_arena_get(const extent_t *extent) {
+	unsigned arena_ind = extent_arena_ind_get(extent);
+
+	return (arena_t *)atomic_load_p(&arenas[arena_ind], ATOMIC_ACQUIRE);
+}
+
+static inline szind_t
+extent_szind_get_maybe_invalid(const extent_t *extent) {
+	szind_t szind = (szind_t)((extent->e_bits & EXTENT_BITS_SZIND_MASK) >>
+	    EXTENT_BITS_SZIND_SHIFT);
+	assert(szind <= SC_NSIZES);
+	return szind;
+}
+
+static inline szind_t
+extent_szind_get(const extent_t *extent) {
+	szind_t szind = extent_szind_get_maybe_invalid(extent);
+	assert(szind < SC_NSIZES); /* Never call when "invalid". */
+	return szind;
+}
+
+static inline size_t
+extent_usize_get(const extent_t *extent) {
+	return sz_index2size(extent_szind_get(extent));
+}
+
+static inline unsigned
+extent_binshard_get(const extent_t *extent) {
+	unsigned binshard = (unsigned)((extent->e_bits &
+	    EXTENT_BITS_BINSHARD_MASK) >> EXTENT_BITS_BINSHARD_SHIFT);
+	assert(binshard < bin_infos[extent_szind_get(extent)].n_shards);
+	return binshard;
+}
+
+static inline size_t
+extent_sn_get(const extent_t *extent) {
+	return (size_t)((extent->e_bits & EXTENT_BITS_SN_MASK) >>
+	    EXTENT_BITS_SN_SHIFT);
+}
+
+static inline extent_state_t
+extent_state_get(const extent_t *extent) {
+	return (extent_state_t)((extent->e_bits & EXTENT_BITS_STATE_MASK) >>
+	    EXTENT_BITS_STATE_SHIFT);
+}
+
+static inline bool
+extent_zeroed_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_ZEROED_MASK) >>
+	    EXTENT_BITS_ZEROED_SHIFT);
+}
+
+static inline bool
+extent_committed_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_COMMITTED_MASK) >>
+	    EXTENT_BITS_COMMITTED_SHIFT);
+}
+
+static inline bool
+extent_dumpable_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_DUMPABLE_MASK) >>
+	    EXTENT_BITS_DUMPABLE_SHIFT);
+}
+
+static inline bool
+extent_slab_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_SLAB_MASK) >>
+	    EXTENT_BITS_SLAB_SHIFT);
+}
+
+static inline unsigned
+extent_nfree_get(const extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return (unsigned)((extent->e_bits & EXTENT_BITS_NFREE_MASK) >>
+	    EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void *
+extent_base_get(const extent_t *extent) {
+	assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
+	    !extent_slab_get(extent));
+	return PAGE_ADDR2BASE(extent->e_addr);
+}
+
+static inline void *
+extent_addr_get(const extent_t *extent) {
+	assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
+	    !extent_slab_get(extent));
+	return extent->e_addr;
+}
+
+static inline size_t
+extent_size_get(const extent_t *extent) {
+	return (extent->e_size_esn & EXTENT_SIZE_MASK);
+}
+
+static inline size_t
+extent_esn_get(const extent_t *extent) {
+	return (extent->e_size_esn & EXTENT_ESN_MASK);
+}
+
+static inline size_t
+extent_bsize_get(const extent_t *extent) {
+	return extent->e_bsize;
+}
+
+static inline void *
+extent_before_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) - PAGE);
+}
+
+static inline void *
+extent_last_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) +
+	    extent_size_get(extent) - PAGE);
+}
+
+static inline void *
+extent_past_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) +
+	    extent_size_get(extent));
+}
+
+static inline arena_slab_data_t *
+extent_slab_data_get(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return &extent->e_slab_data;
+}
+
+static inline const arena_slab_data_t *
+extent_slab_data_get_const(const extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return &extent->e_slab_data;
+}
+
+static inline prof_tctx_t *
+extent_prof_tctx_get(const extent_t *extent) {
+	return (prof_tctx_t *)atomic_load_p(&extent->e_prof_tctx,
+	    ATOMIC_ACQUIRE);
+}
+
+static inline nstime_t
+extent_prof_alloc_time_get(const extent_t *extent) {
+	return extent->e_alloc_time;
+}
+
+static inline void
+extent_arena_set(extent_t *extent, arena_t *arena) {
+	unsigned arena_ind = (arena != NULL) ? arena_ind_get(arena) : ((1U <<
+	    MALLOCX_ARENA_BITS) - 1);
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_ARENA_MASK) |
+	    ((uint64_t)arena_ind << EXTENT_BITS_ARENA_SHIFT);
+}
+
+static inline void
+extent_binshard_set(extent_t *extent, unsigned binshard) {
+	/* The assertion assumes szind is set already. */
+	assert(binshard < bin_infos[extent_szind_get(extent)].n_shards);
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_BINSHARD_MASK) |
+	    ((uint64_t)binshard << EXTENT_BITS_BINSHARD_SHIFT);
+}
+
+static inline void
+extent_addr_set(extent_t *extent, void *addr) {
+	extent->e_addr = addr;
+}
+
+static inline void
+extent_addr_randomize(tsdn_t *tsdn, extent_t *extent, size_t alignment) {
+	assert(extent_base_get(extent) == extent_addr_get(extent));
+
+	if (alignment < PAGE) {
+		unsigned lg_range = LG_PAGE -
+		    lg_floor(CACHELINE_CEILING(alignment));
+		size_t r;
+		if (!tsdn_null(tsdn)) {
+			tsd_t *tsd = tsdn_tsd(tsdn);
+			r = (size_t)prng_lg_range_u64(
+			    tsd_offset_statep_get(tsd), lg_range);
+		} else {
+			r = prng_lg_range_zu(
+			    &extent_arena_get(extent)->offset_state,
+			    lg_range, true);
+		}
+		uintptr_t random_offset = ((uintptr_t)r) << (LG_PAGE -
+		    lg_range);
+		extent->e_addr = (void *)((uintptr_t)extent->e_addr +
+		    random_offset);
+		assert(ALIGNMENT_ADDR2BASE(extent->e_addr, alignment) ==
+		    extent->e_addr);
+	}
+}
+
+static inline void
+extent_size_set(extent_t *extent, size_t size) {
+	assert((size & ~EXTENT_SIZE_MASK) == 0);
+	extent->e_size_esn = size | (extent->e_size_esn & ~EXTENT_SIZE_MASK);
+}
+
+static inline void
+extent_esn_set(extent_t *extent, size_t esn) {
+	extent->e_size_esn = (extent->e_size_esn & ~EXTENT_ESN_MASK) | (esn &
+	    EXTENT_ESN_MASK);
+}
+
+static inline void
+extent_bsize_set(extent_t *extent, size_t bsize) {
+	extent->e_bsize = bsize;
+}
+
+static inline void
+extent_szind_set(extent_t *extent, szind_t szind) {
+	assert(szind <= SC_NSIZES); /* SC_NSIZES means "invalid". */
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SZIND_MASK) |
+	    ((uint64_t)szind << EXTENT_BITS_SZIND_SHIFT);
+}
+
+static inline void
+extent_nfree_set(extent_t *extent, unsigned nfree) {
+	assert(extent_slab_get(extent));
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_NFREE_MASK) |
+	    ((uint64_t)nfree << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_binshard_set(extent_t *extent, unsigned nfree, unsigned binshard) {
+	/* The assertion assumes szind is set already. */
+	assert(binshard < bin_infos[extent_szind_get(extent)].n_shards);
+	extent->e_bits = (extent->e_bits &
+	    (~EXTENT_BITS_NFREE_MASK & ~EXTENT_BITS_BINSHARD_MASK)) |
+	    ((uint64_t)binshard << EXTENT_BITS_BINSHARD_SHIFT) |
+	    ((uint64_t)nfree << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_inc(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	extent->e_bits += ((uint64_t)1U << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_dec(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	extent->e_bits -= ((uint64_t)1U << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_sub(extent_t *extent, uint64_t n) {
+	assert(extent_slab_get(extent));
+	extent->e_bits -= (n << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_sn_set(extent_t *extent, size_t sn) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SN_MASK) |
+	    ((uint64_t)sn << EXTENT_BITS_SN_SHIFT);
+}
+
+static inline void
+extent_state_set(extent_t *extent, extent_state_t state) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_STATE_MASK) |
+	    ((uint64_t)state << EXTENT_BITS_STATE_SHIFT);
+}
+
+static inline void
+extent_zeroed_set(extent_t *extent, bool zeroed) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_ZEROED_MASK) |
+	    ((uint64_t)zeroed << EXTENT_BITS_ZEROED_SHIFT);
+}
+
+static inline void
+extent_committed_set(extent_t *extent, bool committed) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_COMMITTED_MASK) |
+	    ((uint64_t)committed << EXTENT_BITS_COMMITTED_SHIFT);
+}
+
+static inline void
+extent_dumpable_set(extent_t *extent, bool dumpable) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_DUMPABLE_MASK) |
+	    ((uint64_t)dumpable << EXTENT_BITS_DUMPABLE_SHIFT);
+}
+
+static inline void
+extent_slab_set(extent_t *extent, bool slab) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SLAB_MASK) |
+	    ((uint64_t)slab << EXTENT_BITS_SLAB_SHIFT);
+}
+
+static inline void
+extent_prof_tctx_set(extent_t *extent, prof_tctx_t *tctx) {
+	atomic_store_p(&extent->e_prof_tctx, tctx, ATOMIC_RELEASE);
+}
+
+static inline void
+extent_prof_alloc_time_set(extent_t *extent, nstime_t t) {
+	nstime_copy(&extent->e_alloc_time, &t);
+}
+
+static inline bool
+extent_is_head_get(extent_t *extent) {
+	if (maps_coalesce) {
+		not_reached();
+	}
+
+	return (bool)((extent->e_bits & EXTENT_BITS_IS_HEAD_MASK) >>
+	    EXTENT_BITS_IS_HEAD_SHIFT);
+}
+
+static inline void
+extent_is_head_set(extent_t *extent, bool is_head) {
+	if (maps_coalesce) {
+		not_reached();
+	}
+
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_IS_HEAD_MASK) |
+	    ((uint64_t)is_head << EXTENT_BITS_IS_HEAD_SHIFT);
+}
+
+static inline void
+extent_init(extent_t *extent, arena_t *arena, void *addr, size_t size,
+    bool slab, szind_t szind, size_t sn, extent_state_t state, bool zeroed,
+    bool committed, bool dumpable, extent_head_state_t is_head) {
+	assert(addr == PAGE_ADDR2BASE(addr) || !slab);
+
+	extent_arena_set(extent, arena);
+	extent_addr_set(extent, addr);
+	extent_size_set(extent, size);
+	extent_slab_set(extent, slab);
+	extent_szind_set(extent, szind);
+	extent_sn_set(extent, sn);
+	extent_state_set(extent, state);
+	extent_zeroed_set(extent, zeroed);
+	extent_committed_set(extent, committed);
+	extent_dumpable_set(extent, dumpable);
+	ql_elm_new(extent, ql_link);
+	if (!maps_coalesce) {
+		extent_is_head_set(extent, (is_head == EXTENT_IS_HEAD) ? true :
+		    false);
+	}
+	if (config_prof) {
+		extent_prof_tctx_set(extent, NULL);
+	}
+}
+
+static inline void
+extent_binit(extent_t *extent, void *addr, size_t bsize, size_t sn) {
+	extent_arena_set(extent, NULL);
+	extent_addr_set(extent, addr);
+	extent_bsize_set(extent, bsize);
+	extent_slab_set(extent, false);
+	extent_szind_set(extent, SC_NSIZES);
+	extent_sn_set(extent, sn);
+	extent_state_set(extent, extent_state_active);
+	extent_zeroed_set(extent, true);
+	extent_committed_set(extent, true);
+	extent_dumpable_set(extent, true);
+}
+
+static inline void
+extent_list_init(extent_list_t *list) {
+	ql_new(list);
+}
+
+static inline extent_t *
+extent_list_first(const extent_list_t *list) {
+	return ql_first(list);
+}
+
+static inline extent_t *
+extent_list_last(const extent_list_t *list) {
+	return ql_last(list, ql_link);
+}
+
+static inline void
+extent_list_append(extent_list_t *list, extent_t *extent) {
+	ql_tail_insert(list, extent, ql_link);
+}
+
+static inline void
+extent_list_prepend(extent_list_t *list, extent_t *extent) {
+	ql_head_insert(list, extent, ql_link);
+}
+
+static inline void
+extent_list_replace(extent_list_t *list, extent_t *to_remove,
+    extent_t *to_insert) {
+	ql_after_insert(to_remove, to_insert, ql_link);
+	ql_remove(list, to_remove, ql_link);
+}
+
+static inline void
+extent_list_remove(extent_list_t *list, extent_t *extent) {
+	ql_remove(list, extent, ql_link);
+}
+
+static inline int
+extent_sn_comp(const extent_t *a, const extent_t *b) {
+	size_t a_sn = extent_sn_get(a);
+	size_t b_sn = extent_sn_get(b);
+
+	return (a_sn > b_sn) - (a_sn < b_sn);
+}
+
+static inline int
+extent_esn_comp(const extent_t *a, const extent_t *b) {
+	size_t a_esn = extent_esn_get(a);
+	size_t b_esn = extent_esn_get(b);
+
+	return (a_esn > b_esn) - (a_esn < b_esn);
+}
+
+static inline int
+extent_ad_comp(const extent_t *a, const extent_t *b) {
+	uintptr_t a_addr = (uintptr_t)extent_addr_get(a);
+	uintptr_t b_addr = (uintptr_t)extent_addr_get(b);
+
+	return (a_addr > b_addr) - (a_addr < b_addr);
+}
+
+static inline int
+extent_ead_comp(const extent_t *a, const extent_t *b) {
+	uintptr_t a_eaddr = (uintptr_t)a;
+	uintptr_t b_eaddr = (uintptr_t)b;
+
+	return (a_eaddr > b_eaddr) - (a_eaddr < b_eaddr);
+}
+
+static inline int
+extent_snad_comp(const extent_t *a, const extent_t *b) {
+	int ret;
+
+	ret = extent_sn_comp(a, b);
+	if (ret != 0) {
+		return ret;
+	}
+
+	ret = extent_ad_comp(a, b);
+	return ret;
+}
+
+static inline int
+extent_esnead_comp(const extent_t *a, const extent_t *b) {
+	int ret;
+
+	ret = extent_esn_comp(a, b);
+	if (ret != 0) {
+		return ret;
+	}
+
+	ret = extent_ead_comp(a, b);
+	return ret;
+}
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_INLINES_H */
diff --git a/jemalloc/include/jemalloc/internal/extent_mmap.h b/jemalloc/include/jemalloc/internal/extent_mmap.h
new file mode 100644
index 0000000..55f17ee
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/extent_mmap.h
@@ -0,0 +1,10 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H
+
+extern bool opt_retain;
+
+void *extent_alloc_mmap(void *new_addr, size_t size, size_t alignment,
+    bool *zero, bool *commit);
+bool extent_dalloc_mmap(void *addr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/extent_structs.h b/jemalloc/include/jemalloc/internal/extent_structs.h
new file mode 100644
index 0000000..767cd89
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/extent_structs.h
@@ -0,0 +1,256 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_STRUCTS_H
+#define JEMALLOC_INTERNAL_EXTENT_STRUCTS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/sc.h"
+
+typedef enum {
+	extent_state_active   = 0,
+	extent_state_dirty    = 1,
+	extent_state_muzzy    = 2,
+	extent_state_retained = 3
+} extent_state_t;
+
+/* Extent (span of pages).  Use accessor functions for e_* fields. */
+struct extent_s {
+	/*
+	 * Bitfield containing several fields:
+	 *
+	 * a: arena_ind
+	 * b: slab
+	 * c: committed
+	 * d: dumpable
+	 * z: zeroed
+	 * t: state
+	 * i: szind
+	 * f: nfree
+	 * s: bin_shard
+	 * n: sn
+	 *
+	 * nnnnnnnn ... nnnnnnss ssssffff ffffffii iiiiiitt zdcbaaaa aaaaaaaa
+	 *
+	 * arena_ind: Arena from which this extent came, or all 1 bits if
+	 *            unassociated.
+	 *
+	 * slab: The slab flag indicates whether the extent is used for a slab
+	 *       of small regions.  This helps differentiate small size classes,
+	 *       and it indicates whether interior pointers can be looked up via
+	 *       iealloc().
+	 *
+	 * committed: The committed flag indicates whether physical memory is
+	 *            committed to the extent, whether explicitly or implicitly
+	 *            as on a system that overcommits and satisfies physical
+	 *            memory needs on demand via soft page faults.
+	 *
+	 * dumpable: The dumpable flag indicates whether or not we've set the
+	 *           memory in question to be dumpable.  Note that this
+	 *           interacts somewhat subtly with user-specified extent hooks,
+	 *           since we don't know if *they* are fiddling with
+	 *           dumpability (in which case, we don't want to undo whatever
+	 *           they're doing).  To deal with this scenario, we:
+	 *             - Make dumpable false only for memory allocated with the
+	 *               default hooks.
+	 *             - Only allow memory to go from non-dumpable to dumpable,
+	 *               and only once.
+	 *             - Never make the OS call to allow dumping when the
+	 *               dumpable bit is already set.
+	 *           These three constraints mean that we will never
+	 *           accidentally dump user memory that the user meant to set
+	 *           nondumpable with their extent hooks.
+	 *
+	 *
+	 * zeroed: The zeroed flag is used by extent recycling code to track
+	 *         whether memory is zero-filled.
+	 *
+	 * state: The state flag is an extent_state_t.
+	 *
+	 * szind: The szind flag indicates usable size class index for
+	 *        allocations residing in this extent, regardless of whether the
+	 *        extent is a slab.  Extent size and usable size often differ
+	 *        even for non-slabs, either due to sz_large_pad or promotion of
+	 *        sampled small regions.
+	 *
+	 * nfree: Number of free regions in slab.
+	 *
+	 * bin_shard: the shard of the bin from which this extent came.
+	 *
+	 * sn: Serial number (potentially non-unique).
+	 *
+	 *     Serial numbers may wrap around if !opt_retain, but as long as
+	 *     comparison functions fall back on address comparison for equal
+	 *     serial numbers, stable (if imperfect) ordering is maintained.
+	 *
+	 *     Serial numbers may not be unique even in the absence of
+	 *     wrap-around, e.g. when splitting an extent and assigning the same
+	 *     serial number to both resulting adjacent extents.
+	 */
+	uint64_t		e_bits;
+#define MASK(CURRENT_FIELD_WIDTH, CURRENT_FIELD_SHIFT) ((((((uint64_t)0x1U) << (CURRENT_FIELD_WIDTH)) - 1)) << (CURRENT_FIELD_SHIFT))
+
+#define EXTENT_BITS_ARENA_WIDTH  MALLOCX_ARENA_BITS
+#define EXTENT_BITS_ARENA_SHIFT  0
+#define EXTENT_BITS_ARENA_MASK  MASK(EXTENT_BITS_ARENA_WIDTH, EXTENT_BITS_ARENA_SHIFT)
+
+#define EXTENT_BITS_SLAB_WIDTH  1
+#define EXTENT_BITS_SLAB_SHIFT  (EXTENT_BITS_ARENA_WIDTH + EXTENT_BITS_ARENA_SHIFT)
+#define EXTENT_BITS_SLAB_MASK  MASK(EXTENT_BITS_SLAB_WIDTH, EXTENT_BITS_SLAB_SHIFT)
+
+#define EXTENT_BITS_COMMITTED_WIDTH  1
+#define EXTENT_BITS_COMMITTED_SHIFT  (EXTENT_BITS_SLAB_WIDTH + EXTENT_BITS_SLAB_SHIFT)
+#define EXTENT_BITS_COMMITTED_MASK  MASK(EXTENT_BITS_COMMITTED_WIDTH, EXTENT_BITS_COMMITTED_SHIFT)
+
+#define EXTENT_BITS_DUMPABLE_WIDTH  1
+#define EXTENT_BITS_DUMPABLE_SHIFT  (EXTENT_BITS_COMMITTED_WIDTH + EXTENT_BITS_COMMITTED_SHIFT)
+#define EXTENT_BITS_DUMPABLE_MASK  MASK(EXTENT_BITS_DUMPABLE_WIDTH, EXTENT_BITS_DUMPABLE_SHIFT)
+
+#define EXTENT_BITS_ZEROED_WIDTH  1
+#define EXTENT_BITS_ZEROED_SHIFT  (EXTENT_BITS_DUMPABLE_WIDTH + EXTENT_BITS_DUMPABLE_SHIFT)
+#define EXTENT_BITS_ZEROED_MASK  MASK(EXTENT_BITS_ZEROED_WIDTH, EXTENT_BITS_ZEROED_SHIFT)
+
+#define EXTENT_BITS_STATE_WIDTH  2
+#define EXTENT_BITS_STATE_SHIFT  (EXTENT_BITS_ZEROED_WIDTH + EXTENT_BITS_ZEROED_SHIFT)
+#define EXTENT_BITS_STATE_MASK  MASK(EXTENT_BITS_STATE_WIDTH, EXTENT_BITS_STATE_SHIFT)
+
+#define EXTENT_BITS_SZIND_WIDTH  LG_CEIL(SC_NSIZES)
+#define EXTENT_BITS_SZIND_SHIFT  (EXTENT_BITS_STATE_WIDTH + EXTENT_BITS_STATE_SHIFT)
+#define EXTENT_BITS_SZIND_MASK  MASK(EXTENT_BITS_SZIND_WIDTH, EXTENT_BITS_SZIND_SHIFT)
+
+#define EXTENT_BITS_NFREE_WIDTH  (LG_SLAB_MAXREGS + 1)
+#define EXTENT_BITS_NFREE_SHIFT  (EXTENT_BITS_SZIND_WIDTH + EXTENT_BITS_SZIND_SHIFT)
+#define EXTENT_BITS_NFREE_MASK  MASK(EXTENT_BITS_NFREE_WIDTH, EXTENT_BITS_NFREE_SHIFT)
+
+#define EXTENT_BITS_BINSHARD_WIDTH  6
+#define EXTENT_BITS_BINSHARD_SHIFT  (EXTENT_BITS_NFREE_WIDTH + EXTENT_BITS_NFREE_SHIFT)
+#define EXTENT_BITS_BINSHARD_MASK  MASK(EXTENT_BITS_BINSHARD_WIDTH, EXTENT_BITS_BINSHARD_SHIFT)
+
+#define EXTENT_BITS_IS_HEAD_WIDTH 1
+#define EXTENT_BITS_IS_HEAD_SHIFT  (EXTENT_BITS_BINSHARD_WIDTH + EXTENT_BITS_BINSHARD_SHIFT)
+#define EXTENT_BITS_IS_HEAD_MASK  MASK(EXTENT_BITS_IS_HEAD_WIDTH, EXTENT_BITS_IS_HEAD_SHIFT)
+
+#define EXTENT_BITS_SN_SHIFT   (EXTENT_BITS_IS_HEAD_WIDTH + EXTENT_BITS_IS_HEAD_SHIFT)
+#define EXTENT_BITS_SN_MASK  (UINT64_MAX << EXTENT_BITS_SN_SHIFT)
+
+	/* Pointer to the extent that this structure is responsible for. */
+	void			*e_addr;
+
+	union {
+		/*
+		 * Extent size and serial number associated with the extent
+		 * structure (different than the serial number for the extent at
+		 * e_addr).
+		 *
+		 * ssssssss [...] ssssssss ssssnnnn nnnnnnnn
+		 */
+		size_t			e_size_esn;
+	#define EXTENT_SIZE_MASK	((size_t)~(PAGE-1))
+	#define EXTENT_ESN_MASK		((size_t)PAGE-1)
+		/* Base extent size, which may not be a multiple of PAGE. */
+		size_t			e_bsize;
+	};
+
+	/*
+	 * List linkage, used by a variety of lists:
+	 * - bin_t's slabs_full
+	 * - extents_t's LRU
+	 * - stashed dirty extents
+	 * - arena's large allocations
+	 */
+	ql_elm(extent_t)	ql_link;
+
+	/*
+	 * Linkage for per size class sn/address-ordered heaps, and
+	 * for extent_avail
+	 */
+	phn(extent_t)		ph_link;
+
+	union {
+		/* Small region slab metadata. */
+		arena_slab_data_t	e_slab_data;
+
+		/* Profiling data, used for large objects. */
+		struct {
+			/* Time when this was allocated. */
+			nstime_t		e_alloc_time;
+			/* Points to a prof_tctx_t. */
+			atomic_p_t		e_prof_tctx;
+		};
+	};
+};
+typedef ql_head(extent_t) extent_list_t;
+typedef ph(extent_t) extent_tree_t;
+typedef ph(extent_t) extent_heap_t;
+
+/* Quantized collection of extents, with built-in LRU queue. */
+struct extents_s {
+	malloc_mutex_t		mtx;
+
+	/*
+	 * Quantized per size class heaps of extents.
+	 *
+	 * Synchronization: mtx.
+	 */
+	extent_heap_t		heaps[SC_NPSIZES + 1];
+	atomic_zu_t		nextents[SC_NPSIZES + 1];
+	atomic_zu_t		nbytes[SC_NPSIZES + 1];
+
+	/*
+	 * Bitmap for which set bits correspond to non-empty heaps.
+	 *
+	 * Synchronization: mtx.
+	 */
+	bitmap_t		bitmap[BITMAP_GROUPS(SC_NPSIZES + 1)];
+
+	/*
+	 * LRU of all extents in heaps.
+	 *
+	 * Synchronization: mtx.
+	 */
+	extent_list_t		lru;
+
+	/*
+	 * Page sum for all extents in heaps.
+	 *
+	 * The synchronization here is a little tricky.  Modifications to npages
+	 * must hold mtx, but reads need not (though, a reader who sees npages
+	 * without holding the mutex can't assume anything about the rest of the
+	 * state of the extents_t).
+	 */
+	atomic_zu_t		npages;
+
+	/* All stored extents must be in the same state. */
+	extent_state_t		state;
+
+	/*
+	 * If true, delay coalescing until eviction; otherwise coalesce during
+	 * deallocation.
+	 */
+	bool			delay_coalesce;
+};
+
+/*
+ * The following two structs are for experimental purposes. See
+ * experimental_utilization_query_ctl and
+ * experimental_utilization_batch_query_ctl in src/ctl.c.
+ */
+
+struct extent_util_stats_s {
+	size_t nfree;
+	size_t nregs;
+	size_t size;
+};
+
+struct extent_util_stats_verbose_s {
+	void *slabcur_addr;
+	size_t nfree;
+	size_t nregs;
+	size_t size;
+	size_t bin_nfree;
+	size_t bin_nregs;
+};
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_STRUCTS_H */
diff --git a/jemalloc/include/jemalloc/internal/extent_types.h b/jemalloc/include/jemalloc/internal/extent_types.h
new file mode 100644
index 0000000..96925cf
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/extent_types.h
@@ -0,0 +1,23 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_TYPES_H
+#define JEMALLOC_INTERNAL_EXTENT_TYPES_H
+
+typedef struct extent_s extent_t;
+typedef struct extents_s extents_t;
+
+typedef struct extent_util_stats_s extent_util_stats_t;
+typedef struct extent_util_stats_verbose_s extent_util_stats_verbose_t;
+
+#define EXTENT_HOOKS_INITIALIZER	NULL
+
+/*
+ * When reuse (and split) an active extent, (1U << opt_lg_extent_max_active_fit)
+ * is the max ratio between the size of the active extent and the new extent.
+ */
+#define LG_EXTENT_MAX_ACTIVE_FIT_DEFAULT 6
+
+typedef enum {
+	EXTENT_NOT_HEAD,
+	EXTENT_IS_HEAD   /* Only relevant for Windows && opt.retain. */
+} extent_head_state_t;
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/hash.h b/jemalloc/include/jemalloc/internal/hash.h
new file mode 100644
index 0000000..0270034
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/hash.h
@@ -0,0 +1,319 @@
+#ifndef JEMALLOC_INTERNAL_HASH_H
+#define JEMALLOC_INTERNAL_HASH_H
+
+#include "jemalloc/internal/assert.h"
+
+/*
+ * The following hash function is based on MurmurHash3, placed into the public
+ * domain by Austin Appleby.  See https://github.com/aappleby/smhasher for
+ * details.
+ */
+
+/******************************************************************************/
+/* Internal implementation. */
+static inline uint32_t
+hash_rotl_32(uint32_t x, int8_t r) {
+	return ((x << r) | (x >> (32 - r)));
+}
+
+static inline uint64_t
+hash_rotl_64(uint64_t x, int8_t r) {
+	return ((x << r) | (x >> (64 - r)));
+}
+
+static inline uint32_t
+hash_get_block_32(const uint32_t *p, int i) {
+	/* Handle unaligned read. */
+	if (unlikely((uintptr_t)p & (sizeof(uint32_t)-1)) != 0) {
+		uint32_t ret;
+
+		memcpy(&ret, (uint8_t *)(p + i), sizeof(uint32_t));
+		return ret;
+	}
+
+	return p[i];
+}
+
+static inline uint64_t
+hash_get_block_64(const uint64_t *p, int i) {
+	/* Handle unaligned read. */
+	if (unlikely((uintptr_t)p & (sizeof(uint64_t)-1)) != 0) {
+		uint64_t ret;
+
+		memcpy(&ret, (uint8_t *)(p + i), sizeof(uint64_t));
+		return ret;
+	}
+
+	return p[i];
+}
+
+static inline uint32_t
+hash_fmix_32(uint32_t h) {
+	h ^= h >> 16;
+	h *= 0x85ebca6b;
+	h ^= h >> 13;
+	h *= 0xc2b2ae35;
+	h ^= h >> 16;
+
+	return h;
+}
+
+static inline uint64_t
+hash_fmix_64(uint64_t k) {
+	k ^= k >> 33;
+	k *= KQU(0xff51afd7ed558ccd);
+	k ^= k >> 33;
+	k *= KQU(0xc4ceb9fe1a85ec53);
+	k ^= k >> 33;
+
+	return k;
+}
+
+static inline uint32_t
+hash_x86_32(const void *key, int len, uint32_t seed) {
+	const uint8_t *data = (const uint8_t *) key;
+	const int nblocks = len / 4;
+
+	uint32_t h1 = seed;
+
+	const uint32_t c1 = 0xcc9e2d51;
+	const uint32_t c2 = 0x1b873593;
+
+	/* body */
+	{
+		const uint32_t *blocks = (const uint32_t *) (data + nblocks*4);
+		int i;
+
+		for (i = -nblocks; i; i++) {
+			uint32_t k1 = hash_get_block_32(blocks, i);
+
+			k1 *= c1;
+			k1 = hash_rotl_32(k1, 15);
+			k1 *= c2;
+
+			h1 ^= k1;
+			h1 = hash_rotl_32(h1, 13);
+			h1 = h1*5 + 0xe6546b64;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t *) (data + nblocks*4);
+
+		uint32_t k1 = 0;
+
+		switch (len & 3) {
+		case 3: k1 ^= tail[2] << 16; JEMALLOC_FALLTHROUGH
+		case 2: k1 ^= tail[1] << 8; JEMALLOC_FALLTHROUGH
+		case 1: k1 ^= tail[0]; k1 *= c1; k1 = hash_rotl_32(k1, 15);
+			k1 *= c2; h1 ^= k1;
+		}
+	}
+
+	/* finalization */
+	h1 ^= len;
+
+	h1 = hash_fmix_32(h1);
+
+	return h1;
+}
+
+static inline void
+hash_x86_128(const void *key, const int len, uint32_t seed,
+    uint64_t r_out[2]) {
+	const uint8_t * data = (const uint8_t *) key;
+	const int nblocks = len / 16;
+
+	uint32_t h1 = seed;
+	uint32_t h2 = seed;
+	uint32_t h3 = seed;
+	uint32_t h4 = seed;
+
+	const uint32_t c1 = 0x239b961b;
+	const uint32_t c2 = 0xab0e9789;
+	const uint32_t c3 = 0x38b34ae5;
+	const uint32_t c4 = 0xa1e38b93;
+
+	/* body */
+	{
+		const uint32_t *blocks = (const uint32_t *) (data + nblocks*16);
+		int i;
+
+		for (i = -nblocks; i; i++) {
+			uint32_t k1 = hash_get_block_32(blocks, i*4 + 0);
+			uint32_t k2 = hash_get_block_32(blocks, i*4 + 1);
+			uint32_t k3 = hash_get_block_32(blocks, i*4 + 2);
+			uint32_t k4 = hash_get_block_32(blocks, i*4 + 3);
+
+			k1 *= c1; k1 = hash_rotl_32(k1, 15); k1 *= c2; h1 ^= k1;
+
+			h1 = hash_rotl_32(h1, 19); h1 += h2;
+			h1 = h1*5 + 0x561ccd1b;
+
+			k2 *= c2; k2 = hash_rotl_32(k2, 16); k2 *= c3; h2 ^= k2;
+
+			h2 = hash_rotl_32(h2, 17); h2 += h3;
+			h2 = h2*5 + 0x0bcaa747;
+
+			k3 *= c3; k3 = hash_rotl_32(k3, 17); k3 *= c4; h3 ^= k3;
+
+			h3 = hash_rotl_32(h3, 15); h3 += h4;
+			h3 = h3*5 + 0x96cd1c35;
+
+			k4 *= c4; k4 = hash_rotl_32(k4, 18); k4 *= c1; h4 ^= k4;
+
+			h4 = hash_rotl_32(h4, 13); h4 += h1;
+			h4 = h4*5 + 0x32ac3b17;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t *) (data + nblocks*16);
+		uint32_t k1 = 0;
+		uint32_t k2 = 0;
+		uint32_t k3 = 0;
+		uint32_t k4 = 0;
+
+		switch (len & 15) {
+		case 15: k4 ^= tail[14] << 16; JEMALLOC_FALLTHROUGH
+		case 14: k4 ^= tail[13] << 8; JEMALLOC_FALLTHROUGH
+		case 13: k4 ^= tail[12] << 0;
+			k4 *= c4; k4 = hash_rotl_32(k4, 18); k4 *= c1; h4 ^= k4;
+      JEMALLOC_FALLTHROUGH
+		case 12: k3 ^= tail[11] << 24; JEMALLOC_FALLTHROUGH
+		case 11: k3 ^= tail[10] << 16; JEMALLOC_FALLTHROUGH
+		case 10: k3 ^= tail[ 9] << 8; JEMALLOC_FALLTHROUGH
+		case  9: k3 ^= tail[ 8] << 0;
+		     k3 *= c3; k3 = hash_rotl_32(k3, 17); k3 *= c4; h3 ^= k3;
+         JEMALLOC_FALLTHROUGH
+		case  8: k2 ^= tail[ 7] << 24; JEMALLOC_FALLTHROUGH
+		case  7: k2 ^= tail[ 6] << 16; JEMALLOC_FALLTHROUGH
+		case  6: k2 ^= tail[ 5] << 8; JEMALLOC_FALLTHROUGH
+		case  5: k2 ^= tail[ 4] << 0;
+			k2 *= c2; k2 = hash_rotl_32(k2, 16); k2 *= c3; h2 ^= k2;
+      JEMALLOC_FALLTHROUGH
+		case  4: k1 ^= tail[ 3] << 24; JEMALLOC_FALLTHROUGH
+		case  3: k1 ^= tail[ 2] << 16; JEMALLOC_FALLTHROUGH
+		case  2: k1 ^= tail[ 1] << 8; JEMALLOC_FALLTHROUGH
+		case  1: k1 ^= tail[ 0] << 0;
+			k1 *= c1; k1 = hash_rotl_32(k1, 15); k1 *= c2; h1 ^= k1;
+      JEMALLOC_FALLTHROUGH
+		}
+	}
+
+	/* finalization */
+	h1 ^= len; h2 ^= len; h3 ^= len; h4 ^= len;
+
+	h1 += h2; h1 += h3; h1 += h4;
+	h2 += h1; h3 += h1; h4 += h1;
+
+	h1 = hash_fmix_32(h1);
+	h2 = hash_fmix_32(h2);
+	h3 = hash_fmix_32(h3);
+	h4 = hash_fmix_32(h4);
+
+	h1 += h2; h1 += h3; h1 += h4;
+	h2 += h1; h3 += h1; h4 += h1;
+
+	r_out[0] = (((uint64_t) h2) << 32) | h1;
+	r_out[1] = (((uint64_t) h4) << 32) | h3;
+}
+
+static inline void
+hash_x64_128(const void *key, const int len, const uint32_t seed,
+    uint64_t r_out[2]) {
+	const uint8_t *data = (const uint8_t *) key;
+	const int nblocks = len / 16;
+
+	uint64_t h1 = seed;
+	uint64_t h2 = seed;
+
+	const uint64_t c1 = KQU(0x87c37b91114253d5);
+	const uint64_t c2 = KQU(0x4cf5ad432745937f);
+
+	/* body */
+	{
+		const uint64_t *blocks = (const uint64_t *) (data);
+		int i;
+
+		for (i = 0; i < nblocks; i++) {
+			uint64_t k1 = hash_get_block_64(blocks, i*2 + 0);
+			uint64_t k2 = hash_get_block_64(blocks, i*2 + 1);
+
+			k1 *= c1; k1 = hash_rotl_64(k1, 31); k1 *= c2; h1 ^= k1;
+
+			h1 = hash_rotl_64(h1, 27); h1 += h2;
+			h1 = h1*5 + 0x52dce729;
+
+			k2 *= c2; k2 = hash_rotl_64(k2, 33); k2 *= c1; h2 ^= k2;
+
+			h2 = hash_rotl_64(h2, 31); h2 += h1;
+			h2 = h2*5 + 0x38495ab5;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t*)(data + nblocks*16);
+		uint64_t k1 = 0;
+		uint64_t k2 = 0;
+
+		switch (len & 15) {
+		case 15: k2 ^= ((uint64_t)(tail[14])) << 48; JEMALLOC_FALLTHROUGH
+		case 14: k2 ^= ((uint64_t)(tail[13])) << 40; JEMALLOC_FALLTHROUGH
+		case 13: k2 ^= ((uint64_t)(tail[12])) << 32; JEMALLOC_FALLTHROUGH
+		case 12: k2 ^= ((uint64_t)(tail[11])) << 24; JEMALLOC_FALLTHROUGH
+		case 11: k2 ^= ((uint64_t)(tail[10])) << 16; JEMALLOC_FALLTHROUGH
+		case 10: k2 ^= ((uint64_t)(tail[ 9])) << 8;  JEMALLOC_FALLTHROUGH
+		case  9: k2 ^= ((uint64_t)(tail[ 8])) << 0;
+			k2 *= c2; k2 = hash_rotl_64(k2, 33); k2 *= c1; h2 ^= k2;
+			JEMALLOC_FALLTHROUGH
+		case  8: k1 ^= ((uint64_t)(tail[ 7])) << 56; JEMALLOC_FALLTHROUGH
+		case  7: k1 ^= ((uint64_t)(tail[ 6])) << 48; JEMALLOC_FALLTHROUGH
+		case  6: k1 ^= ((uint64_t)(tail[ 5])) << 40; JEMALLOC_FALLTHROUGH
+		case  5: k1 ^= ((uint64_t)(tail[ 4])) << 32; JEMALLOC_FALLTHROUGH
+		case  4: k1 ^= ((uint64_t)(tail[ 3])) << 24; JEMALLOC_FALLTHROUGH
+		case  3: k1 ^= ((uint64_t)(tail[ 2])) << 16; JEMALLOC_FALLTHROUGH
+		case  2: k1 ^= ((uint64_t)(tail[ 1])) << 8;  JEMALLOC_FALLTHROUGH
+		case  1: k1 ^= ((uint64_t)(tail[ 0])) << 0;
+			k1 *= c1; k1 = hash_rotl_64(k1, 31); k1 *= c2; h1 ^= k1;
+		}
+	}
+
+	/* finalization */
+	h1 ^= len; h2 ^= len;
+
+	h1 += h2;
+	h2 += h1;
+
+	h1 = hash_fmix_64(h1);
+	h2 = hash_fmix_64(h2);
+
+	h1 += h2;
+	h2 += h1;
+
+	r_out[0] = h1;
+	r_out[1] = h2;
+}
+
+/******************************************************************************/
+/* API. */
+static inline void
+hash(const void *key, size_t len, const uint32_t seed, size_t r_hash[2]) {
+	assert(len <= INT_MAX); /* Unfortunate implementation limitation. */
+
+#if (LG_SIZEOF_PTR == 3 && !defined(JEMALLOC_BIG_ENDIAN))
+	hash_x64_128(key, (int)len, seed, (uint64_t *)r_hash);
+#else
+	{
+		uint64_t hashes[2];
+		hash_x86_128(key, (int)len, seed, hashes);
+		r_hash[0] = (size_t)hashes[0];
+		r_hash[1] = (size_t)hashes[1];
+	}
+#endif
+}
+
+#endif /* JEMALLOC_INTERNAL_HASH_H */
diff --git a/jemalloc/include/jemalloc/internal/hook.h b/jemalloc/include/jemalloc/internal/hook.h
new file mode 100644
index 0000000..ee246b1
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/hook.h
@@ -0,0 +1,163 @@
+#ifndef JEMALLOC_INTERNAL_HOOK_H
+#define JEMALLOC_INTERNAL_HOOK_H
+
+#include "jemalloc/internal/tsd.h"
+
+/*
+ * This API is *extremely* experimental, and may get ripped out, changed in API-
+ * and ABI-incompatible ways, be insufficiently or incorrectly documented, etc.
+ *
+ * It allows hooking the stateful parts of the API to see changes as they
+ * happen.
+ *
+ * Allocation hooks are called after the allocation is done, free hooks are
+ * called before the free is done, and expand hooks are called after the
+ * allocation is expanded.
+ *
+ * For realloc and rallocx, if the expansion happens in place, the expansion
+ * hook is called.  If it is moved, then the alloc hook is called on the new
+ * location, and then the free hook is called on the old location (i.e. both
+ * hooks are invoked in between the alloc and the dalloc).
+ *
+ * If we return NULL from OOM, then usize might not be trustworthy.  Calling
+ * realloc(NULL, size) only calls the alloc hook, and calling realloc(ptr, 0)
+ * only calls the free hook.  (Calling realloc(NULL, 0) is treated as malloc(0),
+ * and only calls the alloc hook).
+ *
+ * Reentrancy:
+ *   Reentrancy is guarded against from within the hook implementation.  If you
+ *   call allocator functions from within a hook, the hooks will not be invoked
+ *   again.
+ * Threading:
+ *   The installation of a hook synchronizes with all its uses.  If you can
+ *   prove the installation of a hook happens-before a jemalloc entry point,
+ *   then the hook will get invoked (unless there's a racing removal).
+ *
+ *   Hook insertion appears to be atomic at a per-thread level (i.e. if a thread
+ *   allocates and has the alloc hook invoked, then a subsequent free on the
+ *   same thread will also have the free hook invoked).
+ *
+ *   The *removal* of a hook does *not* block until all threads are done with
+ *   the hook.  Hook authors have to be resilient to this, and need some
+ *   out-of-band mechanism for cleaning up any dynamically allocated memory
+ *   associated with their hook.
+ * Ordering:
+ *   Order of hook execution is unspecified, and may be different than insertion
+ *   order.
+ */
+
+#define HOOK_MAX 4
+
+enum hook_alloc_e {
+	hook_alloc_malloc,
+	hook_alloc_posix_memalign,
+	hook_alloc_aligned_alloc,
+	hook_alloc_calloc,
+	hook_alloc_memalign,
+	hook_alloc_valloc,
+	hook_alloc_mallocx,
+
+	/* The reallocating functions have both alloc and dalloc variants */
+	hook_alloc_realloc,
+	hook_alloc_rallocx,
+};
+/*
+ * We put the enum typedef after the enum, since this file may get included by
+ * jemalloc_cpp.cpp, and C++ disallows enum forward declarations.
+ */
+typedef enum hook_alloc_e hook_alloc_t;
+
+enum hook_dalloc_e {
+	hook_dalloc_free,
+	hook_dalloc_dallocx,
+	hook_dalloc_sdallocx,
+
+	/*
+	 * The dalloc halves of reallocation (not called if in-place expansion
+	 * happens).
+	 */
+	hook_dalloc_realloc,
+	hook_dalloc_rallocx,
+};
+typedef enum hook_dalloc_e hook_dalloc_t;
+
+
+enum hook_expand_e {
+	hook_expand_realloc,
+	hook_expand_rallocx,
+	hook_expand_xallocx,
+};
+typedef enum hook_expand_e hook_expand_t;
+
+typedef void (*hook_alloc)(
+    void *extra, hook_alloc_t type, void *result, uintptr_t result_raw,
+    uintptr_t args_raw[3]);
+
+typedef void (*hook_dalloc)(
+    void *extra, hook_dalloc_t type, void *address, uintptr_t args_raw[3]);
+
+typedef void (*hook_expand)(
+    void *extra, hook_expand_t type, void *address, size_t old_usize,
+    size_t new_usize, uintptr_t result_raw, uintptr_t args_raw[4]);
+
+typedef struct hooks_s hooks_t;
+struct hooks_s {
+	hook_alloc alloc_hook;
+	hook_dalloc dalloc_hook;
+	hook_expand expand_hook;
+	void *extra;
+};
+
+/*
+ * Begin implementation details; everything above this point might one day live
+ * in a public API.  Everything below this point never will.
+ */
+
+/*
+ * The realloc pathways haven't gotten any refactoring love in a while, and it's
+ * fairly difficult to pass information from the entry point to the hooks.  We
+ * put the informaiton the hooks will need into a struct to encapsulate
+ * everything.
+ *
+ * Much of these pathways are force-inlined, so that the compiler can avoid
+ * materializing this struct until we hit an extern arena function.  For fairly
+ * goofy reasons, *many* of the realloc paths hit an extern arena function.
+ * These paths are cold enough that it doesn't matter; eventually, we should
+ * rewrite the realloc code to make the expand-in-place and the
+ * free-then-realloc paths more orthogonal, at which point we don't need to
+ * spread the hook logic all over the place.
+ */
+typedef struct hook_ralloc_args_s hook_ralloc_args_t;
+struct hook_ralloc_args_s {
+	/* I.e. as opposed to rallocx. */
+	bool is_realloc;
+	/*
+	 * The expand hook takes 4 arguments, even if only 3 are actually used;
+	 * we add an extra one in case the user decides to memcpy without
+	 * looking too closely at the hooked function.
+	 */
+	uintptr_t args[4];
+};
+
+/*
+ * Returns an opaque handle to be used when removing the hook.  NULL means that
+ * we couldn't install the hook.
+ */
+bool hook_boot();
+
+void *hook_install(tsdn_t *tsdn, hooks_t *hooks);
+/* Uninstalls the hook with the handle previously returned from hook_install. */
+void hook_remove(tsdn_t *tsdn, void *opaque);
+
+/* Hooks */
+
+void hook_invoke_alloc(hook_alloc_t type, void *result, uintptr_t result_raw,
+    uintptr_t args_raw[3]);
+
+void hook_invoke_dalloc(hook_dalloc_t type, void *address,
+    uintptr_t args_raw[3]);
+
+void hook_invoke_expand(hook_expand_t type, void *address, size_t old_usize,
+    size_t new_usize, uintptr_t result_raw, uintptr_t args_raw[4]);
+
+#endif /* JEMALLOC_INTERNAL_HOOK_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
new file mode 100644
index 0000000..7d6053e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_DECLS_H
+#define JEMALLOC_INTERNAL_DECLS_H
+
+#include <math.h>
+#ifdef _WIN32
+#  include <windows.h>
+#  include "msvc_compat/windows_extra.h"
+#  ifdef _WIN64
+#    if LG_VADDR <= 32
+#      error Generate the headers using x64 vcargs
+#    endif
+#  else
+#    if LG_VADDR > 32
+#      undef LG_VADDR
+#      define LG_VADDR 32
+#    endif
+#  endif
+#else
+#  include <sys/param.h>
+#  include <sys/mman.h>
+#  if !defined(__pnacl__) && !defined(__native_client__)
+#    include <sys/syscall.h>
+#    if !defined(SYS_write) && defined(__NR_write)
+#      define SYS_write __NR_write
+#    endif
+#    if defined(SYS_open) && defined(__aarch64__)
+       /* Android headers may define SYS_open to __NR_open even though
+        * __NR_open may not exist on AArch64 (superseded by __NR_openat). */
+#      undef SYS_open
+#    endif
+#    include <sys/uio.h>
+#  endif
+#  include <pthread.h>
+#  ifdef __FreeBSD__
+#  include <pthread_np.h>
+#  endif
+#  include <signal.h>
+#  ifdef JEMALLOC_OS_UNFAIR_LOCK
+#    include <os/lock.h>
+#  endif
+#  ifdef JEMALLOC_GLIBC_MALLOC_HOOK
+#    include <sched.h>
+#  endif
+#  include <errno.h>
+#  include <sys/time.h>
+#  include <time.h>
+#  ifdef JEMALLOC_HAVE_MACH_ABSOLUTE_TIME
+#    include <mach/mach_time.h>
+#  endif
+#endif
+#include <sys/types.h>
+
+#include <limits.h>
+#ifndef SIZE_T_MAX
+#  define SIZE_T_MAX	SIZE_MAX
+#endif
+#ifndef SSIZE_MAX
+#  define SSIZE_MAX	((ssize_t)(SIZE_T_MAX >> 1))
+#endif
+#include <stdarg.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <stddef.h>
+#ifndef offsetof
+#  define offsetof(type, member)	((size_t)&(((type *)NULL)->member))
+#endif
+#include <string.h>
+#include <strings.h>
+#include <ctype.h>
+#ifdef _MSC_VER
+#  include <io.h>
+typedef intptr_t ssize_t;
+#  define PATH_MAX 1024
+#  define STDERR_FILENO 2
+#  define __func__ __FUNCTION__
+#  ifdef JEMALLOC_HAS_RESTRICT
+#    define restrict __restrict
+#  endif
+/* Disable warnings about deprecated system functions. */
+#  pragma warning(disable: 4996)
+#if _MSC_VER < 1800
+static int
+isblank(int c) {
+	return (c == '\t' || c == ' ');
+}
+#endif
+#else
+#  include <unistd.h>
+#endif
+#include <fcntl.h>
+
+#endif /* JEMALLOC_INTERNAL_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in b/jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
new file mode 100644
index 0000000..c442a21
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
@@ -0,0 +1,366 @@
+#ifndef JEMALLOC_INTERNAL_DEFS_H_
+#define JEMALLOC_INTERNAL_DEFS_H_
+/*
+ * If JEMALLOC_PREFIX is defined via --with-jemalloc-prefix, it will cause all
+ * public APIs to be prefixed.  This makes it possible, with some care, to use
+ * multiple allocators simultaneously.
+ */
+#undef JEMALLOC_PREFIX
+#undef JEMALLOC_CPREFIX
+
+/*
+ * Define overrides for non-standard allocator-related functions if they are
+ * present on the system.
+ */
+#undef JEMALLOC_OVERRIDE___LIBC_CALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_FREE
+#undef JEMALLOC_OVERRIDE___LIBC_MALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_MEMALIGN
+#undef JEMALLOC_OVERRIDE___LIBC_REALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_VALLOC
+#undef JEMALLOC_OVERRIDE___POSIX_MEMALIGN
+
+/*
+ * JEMALLOC_PRIVATE_NAMESPACE is used as a prefix for all library-private APIs.
+ * For shared libraries, symbol visibility mechanisms prevent these symbols
+ * from being exported, but for static libraries, naming collisions are a real
+ * possibility.
+ */
+#undef JEMALLOC_PRIVATE_NAMESPACE
+
+/*
+ * Hyper-threaded CPUs may need a special instruction inside spin loops in
+ * order to yield to another virtual CPU.
+ */
+#undef CPU_SPINWAIT
+/* 1 if CPU_SPINWAIT is defined, 0 otherwise. */
+#undef HAVE_CPU_SPINWAIT
+
+/*
+ * Number of significant bits in virtual addresses.  This may be less than the
+ * total number of bits in a pointer, e.g. on x64, for which the uppermost 16
+ * bits are the same as bit 47.
+ */
+#undef LG_VADDR
+
+/* Defined if C11 atomics are available. */
+#undef JEMALLOC_C11_ATOMICS
+
+/* Defined if GCC __atomic atomics are available. */
+#undef JEMALLOC_GCC_ATOMIC_ATOMICS
+/* and the 8-bit variant support. */
+#undef JEMALLOC_GCC_U8_ATOMIC_ATOMICS
+
+/* Defined if GCC __sync atomics are available. */
+#undef JEMALLOC_GCC_SYNC_ATOMICS
+/* and the 8-bit variant support. */
+#undef JEMALLOC_GCC_U8_SYNC_ATOMICS
+
+/*
+ * Defined if __builtin_clz() and __builtin_clzl() are available.
+ */
+#undef JEMALLOC_HAVE_BUILTIN_CLZ
+
+/*
+ * Defined if os_unfair_lock_*() functions are available, as provided by Darwin.
+ */
+#undef JEMALLOC_OS_UNFAIR_LOCK
+
+/* Defined if syscall(2) is usable. */
+#undef JEMALLOC_USE_SYSCALL
+
+/*
+ * Defined if secure_getenv(3) is available.
+ */
+#undef JEMALLOC_HAVE_SECURE_GETENV
+
+/*
+ * Defined if issetugid(2) is available.
+ */
+#undef JEMALLOC_HAVE_ISSETUGID
+
+/* Defined if pthread_atfork(3) is available. */
+#undef JEMALLOC_HAVE_PTHREAD_ATFORK
+
+/* Defined if pthread_setname_np(3) is available. */
+#undef JEMALLOC_HAVE_PTHREAD_SETNAME_NP
+
+/*
+ * Defined if clock_gettime(CLOCK_MONOTONIC_COARSE, ...) is available.
+ */
+#undef JEMALLOC_HAVE_CLOCK_MONOTONIC_COARSE
+
+/*
+ * Defined if clock_gettime(CLOCK_MONOTONIC, ...) is available.
+ */
+#undef JEMALLOC_HAVE_CLOCK_MONOTONIC
+
+/*
+ * Defined if mach_absolute_time() is available.
+ */
+#undef JEMALLOC_HAVE_MACH_ABSOLUTE_TIME
+
+/*
+ * Defined if _malloc_thread_cleanup() exists.  At least in the case of
+ * FreeBSD, pthread_key_create() allocates, which if used during malloc
+ * bootstrapping will cause recursion into the pthreads library.  Therefore, if
+ * _malloc_thread_cleanup() exists, use it as the basis for thread cleanup in
+ * malloc_tsd.
+ */
+#undef JEMALLOC_MALLOC_THREAD_CLEANUP
+
+/*
+ * Defined if threaded initialization is known to be safe on this platform.
+ * Among other things, it must be possible to initialize a mutex without
+ * triggering allocation in order for threaded allocation to be safe.
+ */
+#undef JEMALLOC_THREADED_INIT
+
+/*
+ * Defined if the pthreads implementation defines
+ * _pthread_mutex_init_calloc_cb(), in which case the function is used in order
+ * to avoid recursive allocation during mutex initialization.
+ */
+#undef JEMALLOC_MUTEX_INIT_CB
+
+/* Non-empty if the tls_model attribute is supported. */
+#undef JEMALLOC_TLS_MODEL
+
+/*
+ * JEMALLOC_DEBUG enables assertions and other sanity checks, and disables
+ * inline functions.
+ */
+#undef JEMALLOC_DEBUG
+
+/* JEMALLOC_STATS enables statistics calculation. */
+#undef JEMALLOC_STATS
+
+/* JEMALLOC_EXPERIMENTAL_SMALLOCX_API enables experimental smallocx API. */
+#undef JEMALLOC_EXPERIMENTAL_SMALLOCX_API
+
+/* JEMALLOC_PROF enables allocation profiling. */
+#undef JEMALLOC_PROF
+
+/* Use libunwind for profile backtracing if defined. */
+#undef JEMALLOC_PROF_LIBUNWIND
+
+/* Use libgcc for profile backtracing if defined. */
+#undef JEMALLOC_PROF_LIBGCC
+
+/* Use gcc intrinsics for profile backtracing if defined. */
+#undef JEMALLOC_PROF_GCC
+
+/*
+ * JEMALLOC_DSS enables use of sbrk(2) to allocate extents from the data storage
+ * segment (DSS).
+ */
+#undef JEMALLOC_DSS
+
+/* Support memory filling (junk/zero). */
+#undef JEMALLOC_FILL
+
+/* Support utrace(2)-based tracing. */
+#undef JEMALLOC_UTRACE
+
+/* Support optional abort() on OOM. */
+#undef JEMALLOC_XMALLOC
+
+/* Support lazy locking (avoid locking unless a second thread is launched). */
+#undef JEMALLOC_LAZY_LOCK
+
+/*
+ * Minimum allocation alignment is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#undef LG_QUANTUM
+
+/* One page is 2^LG_PAGE bytes. */
+#undef LG_PAGE
+
+/*
+ * One huge page is 2^LG_HUGEPAGE bytes.  Note that this is defined even if the
+ * system does not explicitly support huge pages; system calls that require
+ * explicit huge page support are separately configured.
+ */
+#undef LG_HUGEPAGE
+
+/*
+ * If defined, adjacent virtual memory mappings with identical attributes
+ * automatically coalesce, and they fragment when changes are made to subranges.
+ * This is the normal order of things for mmap()/munmap(), but on Windows
+ * VirtualAlloc()/VirtualFree() operations must be precisely matched, i.e.
+ * mappings do *not* coalesce/fragment.
+ */
+#undef JEMALLOC_MAPS_COALESCE
+
+/*
+ * If defined, retain memory for later reuse by default rather than using e.g.
+ * munmap() to unmap freed extents.  This is enabled on 64-bit Linux because
+ * common sequences of mmap()/munmap() calls will cause virtual memory map
+ * holes.
+ */
+#undef JEMALLOC_RETAIN
+
+/* TLS is used to map arenas and magazine caches to threads. */
+#undef JEMALLOC_TLS
+
+/*
+ * Used to mark unreachable code to quiet "end of non-void" compiler warnings.
+ * Don't use this directly; instead use unreachable() from util.h
+ */
+#undef JEMALLOC_INTERNAL_UNREACHABLE
+
+/*
+ * ffs*() functions to use for bitmapping.  Don't use these directly; instead,
+ * use ffs_*() from util.h.
+ */
+#undef JEMALLOC_INTERNAL_FFSLL
+#undef JEMALLOC_INTERNAL_FFSL
+#undef JEMALLOC_INTERNAL_FFS
+
+/*
+ * popcount*() functions to use for bitmapping.
+ */
+#undef JEMALLOC_INTERNAL_POPCOUNTL
+#undef JEMALLOC_INTERNAL_POPCOUNT
+
+/*
+ * If defined, explicitly attempt to more uniformly distribute large allocation
+ * pointer alignments across all cache indices.
+ */
+#undef JEMALLOC_CACHE_OBLIVIOUS
+
+/*
+ * If defined, enable logging facilities.  We make this a configure option to
+ * avoid taking extra branches everywhere.
+ */
+#undef JEMALLOC_LOG
+
+/*
+ * If defined, use readlinkat() (instead of readlink()) to follow
+ * /etc/malloc_conf.
+ */
+#undef JEMALLOC_READLINKAT
+
+/*
+ * Darwin (OS X) uses zones to work around Mach-O symbol override shortcomings.
+ */
+#undef JEMALLOC_ZONE
+
+/*
+ * Methods for determining whether the OS overcommits.
+ * JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY: Linux's
+ *                                         /proc/sys/vm.overcommit_memory file.
+ * JEMALLOC_SYSCTL_VM_OVERCOMMIT: FreeBSD's vm.overcommit sysctl.
+ */
+#undef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+#undef JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY
+
+/* Defined if madvise(2) is available. */
+#undef JEMALLOC_HAVE_MADVISE
+
+/*
+ * Defined if transparent huge pages are supported via the MADV_[NO]HUGEPAGE
+ * arguments to madvise(2).
+ */
+#undef JEMALLOC_HAVE_MADVISE_HUGE
+
+/*
+ * Methods for purging unused pages differ between operating systems.
+ *
+ *   madvise(..., MADV_FREE) : This marks pages as being unused, such that they
+ *                             will be discarded rather than swapped out.
+ *   madvise(..., MADV_DONTNEED) : If JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS is
+ *                                 defined, this immediately discards pages,
+ *                                 such that new pages will be demand-zeroed if
+ *                                 the address region is later touched;
+ *                                 otherwise this behaves similarly to
+ *                                 MADV_FREE, though typically with higher
+ *                                 system overhead.
+ */
+#undef JEMALLOC_PURGE_MADVISE_FREE
+#undef JEMALLOC_PURGE_MADVISE_DONTNEED
+#undef JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS
+
+/* Defined if madvise(2) is available but MADV_FREE is not (x86 Linux only). */
+#undef JEMALLOC_DEFINE_MADVISE_FREE
+
+/*
+ * Defined if MADV_DO[NT]DUMP is supported as an argument to madvise.
+ */
+#undef JEMALLOC_MADVISE_DONTDUMP
+
+/*
+ * Defined if transparent huge pages (THPs) are supported via the
+ * MADV_[NO]HUGEPAGE arguments to madvise(2), and THP support is enabled.
+ */
+#undef JEMALLOC_THP
+
+/* Define if operating system has alloca.h header. */
+#undef JEMALLOC_HAS_ALLOCA_H
+
+/* C99 restrict keyword supported. */
+#undef JEMALLOC_HAS_RESTRICT
+
+/* For use by hash code. */
+#undef JEMALLOC_BIG_ENDIAN
+
+/* sizeof(int) == 2^LG_SIZEOF_INT. */
+#undef LG_SIZEOF_INT
+
+/* sizeof(long) == 2^LG_SIZEOF_LONG. */
+#undef LG_SIZEOF_LONG
+
+/* sizeof(long long) == 2^LG_SIZEOF_LONG_LONG. */
+#undef LG_SIZEOF_LONG_LONG
+
+/* sizeof(intmax_t) == 2^LG_SIZEOF_INTMAX_T. */
+#undef LG_SIZEOF_INTMAX_T
+
+/* glibc malloc hooks (__malloc_hook, __realloc_hook, __free_hook). */
+#undef JEMALLOC_GLIBC_MALLOC_HOOK
+
+/* glibc memalign hook. */
+#undef JEMALLOC_GLIBC_MEMALIGN_HOOK
+
+/* pthread support */
+#undef JEMALLOC_HAVE_PTHREAD
+
+/* dlsym() support */
+#undef JEMALLOC_HAVE_DLSYM
+
+/* Adaptive mutex support in pthreads. */
+#undef JEMALLOC_HAVE_PTHREAD_MUTEX_ADAPTIVE_NP
+
+/* GNU specific sched_getcpu support */
+#undef JEMALLOC_HAVE_SCHED_GETCPU
+
+/* GNU specific sched_setaffinity support */
+#undef JEMALLOC_HAVE_SCHED_SETAFFINITY
+
+/*
+ * If defined, all the features necessary for background threads are present.
+ */
+#undef JEMALLOC_BACKGROUND_THREAD
+
+/*
+ * If defined, jemalloc symbols are not exported (doesn't work when
+ * JEMALLOC_PREFIX is not defined).
+ */
+#undef JEMALLOC_EXPORT
+
+/* config.malloc_conf options string. */
+#undef JEMALLOC_CONFIG_MALLOC_CONF
+
+/* If defined, jemalloc takes the malloc/free/etc. symbol names. */
+#undef JEMALLOC_IS_MALLOC
+
+/*
+ * Defined if strerror_r returns char * if _GNU_SOURCE is defined.
+ */
+#undef JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE
+
+/* Performs additional safety checks when defined. */
+#undef JEMALLOC_OPT_SAFETY_CHECKS
+
+#endif /* JEMALLOC_INTERNAL_DEFS_H_ */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
new file mode 100644
index 0000000..d291170
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
@@ -0,0 +1,57 @@
+#ifndef JEMALLOC_INTERNAL_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTERNS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/tsd_types.h"
+
+/* TSD checks this to set thread local slow state accordingly. */
+extern bool malloc_slow;
+
+/* Run-time options. */
+extern bool opt_abort;
+extern bool opt_abort_conf;
+extern bool opt_confirm_conf;
+extern const char *opt_junk;
+extern bool opt_junk_alloc;
+extern bool opt_junk_free;
+extern bool opt_utrace;
+extern bool opt_xmalloc;
+extern bool opt_zero;
+extern unsigned opt_narenas;
+
+/* Number of CPUs. */
+extern unsigned ncpus;
+
+/* Number of arenas used for automatic multiplexing of threads and arenas. */
+extern unsigned narenas_auto;
+
+/* Base index for manual arenas. */
+extern unsigned manual_arena_base;
+
+/*
+ * Arenas that are used to service external requests.  Not all elements of the
+ * arenas array are necessarily used; arenas are created lazily as needed.
+ */
+extern atomic_p_t arenas[];
+
+void *a0malloc(size_t size);
+void a0dalloc(void *ptr);
+void *bootstrap_malloc(size_t size);
+void *bootstrap_calloc(size_t num, size_t size);
+void bootstrap_free(void *ptr);
+void arena_set(unsigned ind, arena_t *arena);
+unsigned narenas_total_get(void);
+arena_t *arena_init(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+arena_tdata_t *arena_tdata_get_hard(tsd_t *tsd, unsigned ind);
+arena_t *arena_choose_hard(tsd_t *tsd, bool internal);
+void arena_migrate(tsd_t *tsd, unsigned oldind, unsigned newind);
+void iarena_cleanup(tsd_t *tsd);
+void arena_cleanup(tsd_t *tsd);
+void arenas_tdata_cleanup(tsd_t *tsd);
+void jemalloc_prefork(void);
+void jemalloc_postfork_parent(void);
+void jemalloc_postfork_child(void);
+bool malloc_initialized(void);
+void je_sdallocx_noflags(void *ptr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
new file mode 100644
index 0000000..437eaa4
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_INCLUDES_H
+#define JEMALLOC_INTERNAL_INCLUDES_H
+
+/*
+ * jemalloc can conceptually be broken into components (arena, tcache, etc.),
+ * but there are circular dependencies that cannot be broken without
+ * substantial performance degradation.
+ *
+ * Historically, we dealt with this by each header into four sections (types,
+ * structs, externs, and inlines), and included each header file multiple times
+ * in this file, picking out the portion we want on each pass using the
+ * following #defines:
+ *   JEMALLOC_H_TYPES   : Preprocessor-defined constants and psuedo-opaque data
+ *                        types.
+ *   JEMALLOC_H_STRUCTS : Data structures.
+ *   JEMALLOC_H_EXTERNS : Extern data declarations and function prototypes.
+ *   JEMALLOC_H_INLINES : Inline functions.
+ *
+ * We're moving toward a world in which the dependencies are explicit; each file
+ * will #include the headers it depends on (rather than relying on them being
+ * implicitly available via this file including every header file in the
+ * project).
+ *
+ * We're now in an intermediate state: we've broken up the header files to avoid
+ * having to include each one multiple times, but have not yet moved the
+ * dependency information into the header files (i.e. we still rely on the
+ * ordering in this file to ensure all a header's dependencies are available in
+ * its translation unit).  Each component is now broken up into multiple header
+ * files, corresponding to the sections above (e.g. instead of "foo.h", we now
+ * have "foo_types.h", "foo_structs.h", "foo_externs.h", "foo_inlines.h").
+ *
+ * Those files which have been converted to explicitly include their
+ * inter-component dependencies are now in the initial HERMETIC HEADERS
+ * section.  All headers may still rely on jemalloc_preamble.h (which, by fiat,
+ * must be included first in every translation unit) for system headers and
+ * global jemalloc definitions, however.
+ */
+
+/******************************************************************************/
+/* TYPES */
+/******************************************************************************/
+
+#include "jemalloc/internal/extent_types.h"
+#include "jemalloc/internal/base_types.h"
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/tcache_types.h"
+#include "jemalloc/internal/prof_types.h"
+
+/******************************************************************************/
+/* STRUCTS */
+/******************************************************************************/
+
+#include "jemalloc/internal/arena_structs_a.h"
+#include "jemalloc/internal/extent_structs.h"
+#include "jemalloc/internal/base_structs.h"
+#include "jemalloc/internal/prof_structs.h"
+#include "jemalloc/internal/arena_structs_b.h"
+#include "jemalloc/internal/tcache_structs.h"
+#include "jemalloc/internal/background_thread_structs.h"
+
+/******************************************************************************/
+/* EXTERNS */
+/******************************************************************************/
+
+#include "jemalloc/internal/jemalloc_internal_externs.h"
+#include "jemalloc/internal/extent_externs.h"
+#include "jemalloc/internal/base_externs.h"
+#include "jemalloc/internal/arena_externs.h"
+#include "jemalloc/internal/large_externs.h"
+#include "jemalloc/internal/tcache_externs.h"
+#include "jemalloc/internal/prof_externs.h"
+#include "jemalloc/internal/background_thread_externs.h"
+
+/******************************************************************************/
+/* INLINES */
+/******************************************************************************/
+
+#include "jemalloc/internal/jemalloc_internal_inlines_a.h"
+#include "jemalloc/internal/base_inlines.h"
+/*
+ * Include portions of arena code interleaved with tcache code in order to
+ * resolve circular dependencies.
+ */
+#include "jemalloc/internal/prof_inlines_a.h"
+#include "jemalloc/internal/arena_inlines_a.h"
+#include "jemalloc/internal/extent_inlines.h"
+#include "jemalloc/internal/jemalloc_internal_inlines_b.h"
+#include "jemalloc/internal/tcache_inlines.h"
+#include "jemalloc/internal/arena_inlines_b.h"
+#include "jemalloc/internal/jemalloc_internal_inlines_c.h"
+#include "jemalloc/internal/prof_inlines_b.h"
+#include "jemalloc/internal/background_thread_inlines.h"
+
+#endif /* JEMALLOC_INTERNAL_INCLUDES_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
new file mode 100644
index 0000000..ddde9b4
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
@@ -0,0 +1,174 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_A_H
+#define JEMALLOC_INTERNAL_INLINES_A_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/ticker.h"
+
+JEMALLOC_ALWAYS_INLINE malloc_cpuid_t
+malloc_getcpu(void) {
+	assert(have_percpu_arena);
+#if defined(_WIN32)
+	return GetCurrentProcessorNumber();
+#elif defined(JEMALLOC_HAVE_SCHED_GETCPU)
+	return (malloc_cpuid_t)sched_getcpu();
+#else
+	not_reached();
+	return -1;
+#endif
+}
+
+/* Return the chosen arena index based on current cpu. */
+JEMALLOC_ALWAYS_INLINE unsigned
+percpu_arena_choose(void) {
+	assert(have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena));
+
+	malloc_cpuid_t cpuid = malloc_getcpu();
+	assert(cpuid >= 0);
+
+	unsigned arena_ind;
+	if ((opt_percpu_arena == percpu_arena) || ((unsigned)cpuid < ncpus /
+	    2)) {
+		arena_ind = cpuid;
+	} else {
+		assert(opt_percpu_arena == per_phycpu_arena);
+		/* Hyper threads on the same physical CPU share arena. */
+		arena_ind = cpuid - ncpus / 2;
+	}
+
+	return arena_ind;
+}
+
+/* Return the limit of percpu auto arena range, i.e. arenas[0...ind_limit). */
+JEMALLOC_ALWAYS_INLINE unsigned
+percpu_arena_ind_limit(percpu_arena_mode_t mode) {
+	assert(have_percpu_arena && PERCPU_ARENA_ENABLED(mode));
+	if (mode == per_phycpu_arena && ncpus > 1) {
+		if (ncpus % 2) {
+			/* This likely means a misconfig. */
+			return ncpus / 2 + 1;
+		}
+		return ncpus / 2;
+	} else {
+		return ncpus;
+	}
+}
+
+static inline arena_tdata_t *
+arena_tdata_get(tsd_t *tsd, unsigned ind, bool refresh_if_missing) {
+	arena_tdata_t *tdata;
+	arena_tdata_t *arenas_tdata = tsd_arenas_tdata_get(tsd);
+
+	if (unlikely(arenas_tdata == NULL)) {
+		/* arenas_tdata hasn't been initialized yet. */
+		return arena_tdata_get_hard(tsd, ind);
+	}
+	if (unlikely(ind >= tsd_narenas_tdata_get(tsd))) {
+		/*
+		 * ind is invalid, cache is old (too small), or tdata to be
+		 * initialized.
+		 */
+		return (refresh_if_missing ? arena_tdata_get_hard(tsd, ind) :
+		    NULL);
+	}
+
+	tdata = &arenas_tdata[ind];
+	if (likely(tdata != NULL) || !refresh_if_missing) {
+		return tdata;
+	}
+	return arena_tdata_get_hard(tsd, ind);
+}
+
+static inline arena_t *
+arena_get(tsdn_t *tsdn, unsigned ind, bool init_if_missing) {
+	arena_t *ret;
+
+	assert(ind < MALLOCX_ARENA_LIMIT);
+
+	ret = (arena_t *)atomic_load_p(&arenas[ind], ATOMIC_ACQUIRE);
+	if (unlikely(ret == NULL)) {
+		if (init_if_missing) {
+			ret = arena_init(tsdn, ind,
+			    (extent_hooks_t *)&extent_hooks_default);
+		}
+	}
+	return ret;
+}
+
+static inline ticker_t *
+decay_ticker_get(tsd_t *tsd, unsigned ind) {
+	arena_tdata_t *tdata;
+
+	tdata = arena_tdata_get(tsd, ind, true);
+	if (unlikely(tdata == NULL)) {
+		return NULL;
+	}
+	return &tdata->decay_ticker;
+}
+
+JEMALLOC_ALWAYS_INLINE cache_bin_t *
+tcache_small_bin_get(tcache_t *tcache, szind_t binind) {
+	assert(binind < SC_NBINS);
+	return &tcache->bins_small[binind];
+}
+
+JEMALLOC_ALWAYS_INLINE cache_bin_t *
+tcache_large_bin_get(tcache_t *tcache, szind_t binind) {
+	assert(binind >= SC_NBINS &&binind < nhbins);
+	return &tcache->bins_large[binind - SC_NBINS];
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tcache_available(tsd_t *tsd) {
+	/*
+	 * Thread specific auto tcache might be unavailable if: 1) during tcache
+	 * initialization, or 2) disabled through thread.tcache.enabled mallctl
+	 * or config options.  This check covers all cases.
+	 */
+	if (likely(tsd_tcache_enabled_get(tsd))) {
+		/* Associated arena == NULL implies tcache init in progress. */
+		assert(tsd_tcachep_get(tsd)->arena == NULL ||
+		    tcache_small_bin_get(tsd_tcachep_get(tsd), 0)->avail !=
+		    NULL);
+		return true;
+	}
+
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE tcache_t *
+tcache_get(tsd_t *tsd) {
+	if (!tcache_available(tsd)) {
+		return NULL;
+	}
+
+	return tsd_tcachep_get(tsd);
+}
+
+static inline void
+pre_reentrancy(tsd_t *tsd, arena_t *arena) {
+	/* arena is the current context.  Reentry from a0 is not allowed. */
+	assert(arena != arena_get(tsd_tsdn(tsd), 0, false));
+
+	bool fast = tsd_fast(tsd);
+	assert(tsd_reentrancy_level_get(tsd) < INT8_MAX);
+	++*tsd_reentrancy_levelp_get(tsd);
+	if (fast) {
+		/* Prepare slow path for reentrancy. */
+		tsd_slow_update(tsd);
+		assert(tsd_state_get(tsd) == tsd_state_nominal_slow);
+	}
+}
+
+static inline void
+post_reentrancy(tsd_t *tsd) {
+	int8_t *reentrancy_level = tsd_reentrancy_levelp_get(tsd);
+	assert(*reentrancy_level > 0);
+	if (--*reentrancy_level == 0) {
+		tsd_slow_update(tsd);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_A_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
new file mode 100644
index 0000000..70d6e57
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
@@ -0,0 +1,87 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_B_H
+#define JEMALLOC_INTERNAL_INLINES_B_H
+
+#include "jemalloc/internal/rtree.h"
+
+/* Choose an arena based on a per-thread value. */
+static inline arena_t *
+arena_choose_impl(tsd_t *tsd, arena_t *arena, bool internal) {
+	arena_t *ret;
+
+	if (arena != NULL) {
+		return arena;
+	}
+
+	/* During reentrancy, arena 0 is the safest bet. */
+	if (unlikely(tsd_reentrancy_level_get(tsd) > 0)) {
+		return arena_get(tsd_tsdn(tsd), 0, true);
+	}
+
+	ret = internal ? tsd_iarena_get(tsd) : tsd_arena_get(tsd);
+	if (unlikely(ret == NULL)) {
+		ret = arena_choose_hard(tsd, internal);
+		assert(ret);
+		if (tcache_available(tsd)) {
+			tcache_t *tcache = tcache_get(tsd);
+			if (tcache->arena != NULL) {
+				/* See comments in tcache_data_init().*/
+				assert(tcache->arena ==
+				    arena_get(tsd_tsdn(tsd), 0, false));
+				if (tcache->arena != ret) {
+					tcache_arena_reassociate(tsd_tsdn(tsd),
+					    tcache, ret);
+				}
+			} else {
+				tcache_arena_associate(tsd_tsdn(tsd), tcache,
+				    ret);
+			}
+		}
+	}
+
+	/*
+	 * Note that for percpu arena, if the current arena is outside of the
+	 * auto percpu arena range, (i.e. thread is assigned to a manually
+	 * managed arena), then percpu arena is skipped.
+	 */
+	if (have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena) &&
+	    !internal && (arena_ind_get(ret) <
+	    percpu_arena_ind_limit(opt_percpu_arena)) && (ret->last_thd !=
+	    tsd_tsdn(tsd))) {
+		unsigned ind = percpu_arena_choose();
+		if (arena_ind_get(ret) != ind) {
+			percpu_arena_update(tsd, ind);
+			ret = tsd_arena_get(tsd);
+		}
+		ret->last_thd = tsd_tsdn(tsd);
+	}
+
+	return ret;
+}
+
+static inline arena_t *
+arena_choose(tsd_t *tsd, arena_t *arena) {
+	return arena_choose_impl(tsd, arena, false);
+}
+
+static inline arena_t *
+arena_ichoose(tsd_t *tsd, arena_t *arena) {
+	return arena_choose_impl(tsd, arena, true);
+}
+
+static inline bool
+arena_is_auto(arena_t *arena) {
+	assert(narenas_auto > 0);
+
+	return (arena_ind_get(arena) < manual_arena_base);
+}
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+iealloc(tsdn_t *tsdn, const void *ptr) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	return rtree_extent_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_B_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
new file mode 100644
index 0000000..cdb10eb
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
@@ -0,0 +1,222 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_C_H
+#define JEMALLOC_INTERNAL_INLINES_C_H
+
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/witness.h"
+
+/*
+ * Translating the names of the 'i' functions:
+ *   Abbreviations used in the first part of the function name (before
+ *   alloc/dalloc) describe what that function accomplishes:
+ *     a: arena (query)
+ *     s: size (query, or sized deallocation)
+ *     e: extent (query)
+ *     p: aligned (allocates)
+ *     vs: size (query, without knowing that the pointer is into the heap)
+ *     r: rallocx implementation
+ *     x: xallocx implementation
+ *   Abbreviations used in the second part of the function name (after
+ *   alloc/dalloc) describe the arguments it takes
+ *     z: whether to return zeroed memory
+ *     t: accepts a tcache_t * parameter
+ *     m: accepts an arena_t * parameter
+ */
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+iaalloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	return arena_aalloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+isalloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	return arena_salloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iallocztm(tsdn_t *tsdn, size_t size, szind_t ind, bool zero, tcache_t *tcache,
+    bool is_internal, arena_t *arena, bool slow_path) {
+	void *ret;
+
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena == NULL || arena_is_auto(arena));
+	if (!tsdn_null(tsdn) && tsd_reentrancy_level_get(tsdn_tsd(tsdn)) == 0) {
+		witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+		    WITNESS_RANK_CORE, 0);
+	}
+
+	ret = arena_malloc(tsdn, arena, size, ind, zero, tcache, slow_path);
+	if (config_stats && is_internal && likely(ret != NULL)) {
+		arena_internal_add(iaalloc(tsdn, ret), isalloc(tsdn, ret));
+	}
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ialloc(tsd_t *tsd, size_t size, szind_t ind, bool zero, bool slow_path) {
+	return iallocztm(tsd_tsdn(tsd), size, ind, zero, tcache_get(tsd), false,
+	    NULL, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipallocztm(tsdn_t *tsdn, size_t usize, size_t alignment, bool zero,
+    tcache_t *tcache, bool is_internal, arena_t *arena) {
+	void *ret;
+
+	assert(usize != 0);
+	assert(usize == sz_sa2u(usize, alignment));
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena == NULL || arena_is_auto(arena));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	ret = arena_palloc(tsdn, arena, usize, alignment, zero, tcache);
+	assert(ALIGNMENT_ADDR2BASE(ret, alignment) == ret);
+	if (config_stats && is_internal && likely(ret != NULL)) {
+		arena_internal_add(iaalloc(tsdn, ret), isalloc(tsdn, ret));
+	}
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipalloct(tsdn_t *tsdn, size_t usize, size_t alignment, bool zero,
+    tcache_t *tcache, arena_t *arena) {
+	return ipallocztm(tsdn, usize, alignment, zero, tcache, false, arena);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipalloc(tsd_t *tsd, size_t usize, size_t alignment, bool zero) {
+	return ipallocztm(tsd_tsdn(tsd), usize, alignment, zero,
+	    tcache_get(tsd), false, NULL);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ivsalloc(tsdn_t *tsdn, const void *ptr) {
+	return arena_vsalloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idalloctm(tsdn_t *tsdn, void *ptr, tcache_t *tcache, alloc_ctx_t *alloc_ctx,
+    bool is_internal, bool slow_path) {
+	assert(ptr != NULL);
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena_is_auto(iaalloc(tsdn, ptr)));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	if (config_stats && is_internal) {
+		arena_internal_sub(iaalloc(tsdn, ptr), isalloc(tsdn, ptr));
+	}
+	if (!is_internal && !tsdn_null(tsdn) &&
+	    tsd_reentrancy_level_get(tsdn_tsd(tsdn)) != 0) {
+		assert(tcache == NULL);
+	}
+	arena_dalloc(tsdn, ptr, tcache, alloc_ctx, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idalloc(tsd_t *tsd, void *ptr) {
+	idalloctm(tsd_tsdn(tsd), ptr, tcache_get(tsd), NULL, false, true);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+isdalloct(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	arena_sdalloc(tsdn, ptr, size, tcache, alloc_ctx, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloct_realign(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t alignment, bool zero, tcache_t *tcache, arena_t *arena,
+    hook_ralloc_args_t *hook_args) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	void *p;
+	size_t usize, copysize;
+
+	usize = sz_sa2u(size, alignment);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+	p = ipalloct(tsdn, usize, alignment, zero, tcache, arena);
+	if (p == NULL) {
+		return NULL;
+	}
+	/*
+	 * Copy at most size bytes (not size+extra), since the caller has no
+	 * expectation that the extra bytes will be reliably preserved.
+	 */
+	copysize = (size < oldsize) ? size : oldsize;
+	memcpy(p, ptr, copysize);
+	hook_invoke_alloc(hook_args->is_realloc
+	    ? hook_alloc_realloc : hook_alloc_rallocx, p, (uintptr_t)p,
+	    hook_args->args);
+	hook_invoke_dalloc(hook_args->is_realloc
+	    ? hook_dalloc_realloc : hook_dalloc_rallocx, ptr, hook_args->args);
+	isdalloct(tsdn, ptr, oldsize, tcache, NULL, true);
+	return p;
+}
+
+/*
+ * is_realloc threads through the knowledge of whether or not this call comes
+ * from je_realloc (as opposed to je_rallocx); this ensures that we pass the
+ * correct entry point into any hooks.
+ * Note that these functions are all force-inlined, so no actual bool gets
+ * passed-around anywhere.
+ */
+JEMALLOC_ALWAYS_INLINE void *
+iralloct(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size, size_t alignment,
+    bool zero, tcache_t *tcache, arena_t *arena, hook_ralloc_args_t *hook_args)
+{
+	assert(ptr != NULL);
+	assert(size != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		/*
+		 * Existing object alignment is inadequate; allocate new space
+		 * and copy.
+		 */
+		return iralloct_realign(tsdn, ptr, oldsize, size, alignment,
+		    zero, tcache, arena, hook_args);
+	}
+
+	return arena_ralloc(tsdn, arena, ptr, oldsize, size, alignment, zero,
+	    tcache, hook_args);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloc(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t alignment,
+    bool zero, hook_ralloc_args_t *hook_args) {
+	return iralloct(tsd_tsdn(tsd), ptr, oldsize, size, alignment, zero,
+	    tcache_get(tsd), NULL, hook_args);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+ixalloc(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero, size_t *newsize) {
+	assert(ptr != NULL);
+	assert(size != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		/* Existing object alignment is inadequate. */
+		*newsize = oldsize;
+		return true;
+	}
+
+	return arena_ralloc_no_move(tsdn, ptr, oldsize, size, extra, zero,
+	    newsize);
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_C_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
new file mode 100644
index 0000000..d8ea06f
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
@@ -0,0 +1,114 @@
+#ifndef JEMALLOC_INTERNAL_MACROS_H
+#define JEMALLOC_INTERNAL_MACROS_H
+
+#ifdef JEMALLOC_DEBUG
+#  define JEMALLOC_ALWAYS_INLINE static inline
+#else
+#  define JEMALLOC_ALWAYS_INLINE JEMALLOC_ATTR(always_inline) static inline
+#endif
+#ifdef _MSC_VER
+#  define inline _inline
+#endif
+
+#define UNUSED JEMALLOC_ATTR(unused)
+
+#define ZU(z)	((size_t)z)
+#define ZD(z)	((ssize_t)z)
+#define QU(q)	((uint64_t)q)
+#define QD(q)	((int64_t)q)
+
+#define KZU(z)	ZU(z##ULL)
+#define KZD(z)	ZD(z##LL)
+#define KQU(q)	QU(q##ULL)
+#define KQD(q)	QI(q##LL)
+
+#ifndef __DECONST
+#  define	__DECONST(type, var)	((type)(uintptr_t)(const void *)(var))
+#endif
+
+#if !defined(JEMALLOC_HAS_RESTRICT) || defined(__cplusplus)
+#  define restrict
+#endif
+
+/* Various function pointers are static and immutable except during testing. */
+#ifdef JEMALLOC_JET
+#  define JET_MUTABLE
+#else
+#  define JET_MUTABLE const
+#endif
+
+#define JEMALLOC_VA_ARGS_HEAD(head, ...) head
+#define JEMALLOC_VA_ARGS_TAIL(head, ...) __VA_ARGS__
+
+#if (defined(__GNUC__) || defined(__GNUG__)) && !defined(__clang__) \
+  && defined(JEMALLOC_HAVE_ATTR) && (__GNUC__ >= 7)
+#define JEMALLOC_FALLTHROUGH JEMALLOC_ATTR(fallthrough);
+#else
+#define JEMALLOC_FALLTHROUGH /* falls through */
+#endif
+
+/* Diagnostic suppression macros */
+#if defined(_MSC_VER) && !defined(__clang__)
+#  define JEMALLOC_DIAGNOSTIC_PUSH __pragma(warning(push))
+#  define JEMALLOC_DIAGNOSTIC_POP __pragma(warning(pop))
+#  define JEMALLOC_DIAGNOSTIC_IGNORE(W) __pragma(warning(disable:W))
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+#  define JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+/* #pragma GCC diagnostic first appeared in gcc 4.6. */
+#elif (defined(__GNUC__) && ((__GNUC__ > 4) || ((__GNUC__ == 4) && \
+  (__GNUC_MINOR__ > 5)))) || defined(__clang__)
+/*
+ * The JEMALLOC_PRAGMA__ macro is an implementation detail of the GCC and Clang
+ * diagnostic suppression macros and should not be used anywhere else.
+ */
+#  define JEMALLOC_PRAGMA__(X) _Pragma(#X)
+#  define JEMALLOC_DIAGNOSTIC_PUSH JEMALLOC_PRAGMA__(GCC diagnostic push)
+#  define JEMALLOC_DIAGNOSTIC_POP JEMALLOC_PRAGMA__(GCC diagnostic pop)
+#  define JEMALLOC_DIAGNOSTIC_IGNORE(W) \
+     JEMALLOC_PRAGMA__(GCC diagnostic ignored W)
+
+/*
+ * The -Wmissing-field-initializers warning is buggy in GCC versions < 5.1 and
+ * all clang versions up to version 7 (currently trunk, unreleased).  This macro
+ * suppresses the warning for the affected compiler versions only.
+ */
+#  if ((defined(__GNUC__) && !defined(__clang__)) && (__GNUC__ < 5)) || \
+     defined(__clang__)
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS  \
+          JEMALLOC_DIAGNOSTIC_IGNORE("-Wmissing-field-initializers")
+#  else
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+#  endif
+
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS  \
+     JEMALLOC_DIAGNOSTIC_IGNORE("-Wtype-limits")
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_UNUSED_PARAMETER \
+     JEMALLOC_DIAGNOSTIC_IGNORE("-Wunused-parameter")
+#  if defined(__GNUC__) && !defined(__clang__) && (__GNUC__ >= 7)
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN \
+       JEMALLOC_DIAGNOSTIC_IGNORE("-Walloc-size-larger-than=")
+#  else
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+#  endif
+#  define JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS \
+  JEMALLOC_DIAGNOSTIC_PUSH \
+  JEMALLOC_DIAGNOSTIC_IGNORE_UNUSED_PARAMETER
+#else
+#  define JEMALLOC_DIAGNOSTIC_PUSH
+#  define JEMALLOC_DIAGNOSTIC_POP
+#  define JEMALLOC_DIAGNOSTIC_IGNORE(W)
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+#  define JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+#endif
+
+/*
+ * Disables spurious diagnostics for all headers.  Since these headers are not
+ * included by users directly, it does not affect their diagnostic settings.
+ */
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+#endif /* JEMALLOC_INTERNAL_MACROS_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_internal_types.h b/jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
new file mode 100644
index 0000000..e296c5a
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
@@ -0,0 +1,114 @@
+#ifndef JEMALLOC_INTERNAL_TYPES_H
+#define JEMALLOC_INTERNAL_TYPES_H
+
+#include "jemalloc/internal/quantum.h"
+
+/* Page size index type. */
+typedef unsigned pszind_t;
+
+/* Size class index type. */
+typedef unsigned szind_t;
+
+/* Processor / core id type. */
+typedef int malloc_cpuid_t;
+
+/*
+ * Flags bits:
+ *
+ * a: arena
+ * t: tcache
+ * 0: unused
+ * z: zero
+ * n: alignment
+ *
+ * aaaaaaaa aaaatttt tttttttt 0znnnnnn
+ */
+#define MALLOCX_ARENA_BITS	12
+#define MALLOCX_TCACHE_BITS	12
+#define MALLOCX_LG_ALIGN_BITS	6
+#define MALLOCX_ARENA_SHIFT	20
+#define MALLOCX_TCACHE_SHIFT	8
+#define MALLOCX_ARENA_MASK \
+    (((1 << MALLOCX_ARENA_BITS) - 1) << MALLOCX_ARENA_SHIFT)
+/* NB: Arena index bias decreases the maximum number of arenas by 1. */
+#define MALLOCX_ARENA_LIMIT	((1 << MALLOCX_ARENA_BITS) - 1)
+#define MALLOCX_TCACHE_MASK \
+    (((1 << MALLOCX_TCACHE_BITS) - 1) << MALLOCX_TCACHE_SHIFT)
+#define MALLOCX_TCACHE_MAX	((1 << MALLOCX_TCACHE_BITS) - 3)
+#define MALLOCX_LG_ALIGN_MASK	((1 << MALLOCX_LG_ALIGN_BITS) - 1)
+/* Use MALLOCX_ALIGN_GET() if alignment may not be specified in flags. */
+#define MALLOCX_ALIGN_GET_SPECIFIED(flags)				\
+    (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK))
+#define MALLOCX_ALIGN_GET(flags)					\
+    (MALLOCX_ALIGN_GET_SPECIFIED(flags) & (SIZE_T_MAX-1))
+#define MALLOCX_ZERO_GET(flags)						\
+    ((bool)(flags & MALLOCX_ZERO))
+
+#define MALLOCX_TCACHE_GET(flags)					\
+    (((unsigned)((flags & MALLOCX_TCACHE_MASK) >> MALLOCX_TCACHE_SHIFT)) - 2)
+#define MALLOCX_ARENA_GET(flags)					\
+    (((unsigned)(((unsigned)flags) >> MALLOCX_ARENA_SHIFT)) - 1)
+
+/* Smallest size class to support. */
+#define TINY_MIN		(1U << LG_TINY_MIN)
+
+#define LONG			((size_t)(1U << LG_SIZEOF_LONG))
+#define LONG_MASK		(LONG - 1)
+
+/* Return the smallest long multiple that is >= a. */
+#define LONG_CEILING(a)							\
+	(((a) + LONG_MASK) & ~LONG_MASK)
+
+#define SIZEOF_PTR		(1U << LG_SIZEOF_PTR)
+#define PTR_MASK		(SIZEOF_PTR - 1)
+
+/* Return the smallest (void *) multiple that is >= a. */
+#define PTR_CEILING(a)							\
+	(((a) + PTR_MASK) & ~PTR_MASK)
+
+/*
+ * Maximum size of L1 cache line.  This is used to avoid cache line aliasing.
+ * In addition, this controls the spacing of cacheline-spaced size classes.
+ *
+ * CACHELINE cannot be based on LG_CACHELINE because __declspec(align()) can
+ * only handle raw constants.
+ */
+#define LG_CACHELINE		6
+#define CACHELINE		64
+#define CACHELINE_MASK		(CACHELINE - 1)
+
+/* Return the smallest cacheline multiple that is >= s. */
+#define CACHELINE_CEILING(s)						\
+	(((s) + CACHELINE_MASK) & ~CACHELINE_MASK)
+
+/* Return the nearest aligned address at or below a. */
+#define ALIGNMENT_ADDR2BASE(a, alignment)				\
+	((void *)((uintptr_t)(a) & ((~(alignment)) + 1)))
+
+/* Return the offset between a and the nearest aligned address at or below a. */
+#define ALIGNMENT_ADDR2OFFSET(a, alignment)				\
+	((size_t)((uintptr_t)(a) & (alignment - 1)))
+
+/* Return the smallest alignment multiple that is >= s. */
+#define ALIGNMENT_CEILING(s, alignment)					\
+	(((s) + (alignment - 1)) & ((~(alignment)) + 1))
+
+/* Declare a variable-length array. */
+#if __STDC_VERSION__ < 199901L
+#  ifdef _MSC_VER
+#    include <malloc.h>
+#    define alloca _alloca
+#  else
+#    ifdef JEMALLOC_HAS_ALLOCA_H
+#      include <alloca.h>
+#    else
+#      include <stdlib.h>
+#    endif
+#  endif
+#  define VARIABLE_ARRAY(type, name, count) \
+	type *name = alloca(sizeof(type) * (count))
+#else
+#  define VARIABLE_ARRAY(type, name, count) type name[(count)]
+#endif
+
+#endif /* JEMALLOC_INTERNAL_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in b/jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
new file mode 100644
index 0000000..3418cbf
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
@@ -0,0 +1,213 @@
+#ifndef JEMALLOC_PREAMBLE_H
+#define JEMALLOC_PREAMBLE_H
+
+#include "jemalloc_internal_defs.h"
+#include "jemalloc/internal/jemalloc_internal_decls.h"
+
+#ifdef JEMALLOC_UTRACE
+#include <sys/ktrace.h>
+#endif
+
+#define JEMALLOC_NO_DEMANGLE
+#ifdef JEMALLOC_JET
+#  undef JEMALLOC_IS_MALLOC
+#  define JEMALLOC_N(n) jet_##n
+#  include "jemalloc/internal/public_namespace.h"
+#  define JEMALLOC_NO_RENAME
+#  include "../jemalloc@install_suffix@.h"
+#  undef JEMALLOC_NO_RENAME
+#else
+#  define JEMALLOC_N(n) @private_namespace@##n
+#  include "../jemalloc@install_suffix@.h"
+#endif
+
+#if defined(JEMALLOC_OSATOMIC)
+#include <libkern/OSAtomic.h>
+#endif
+
+#ifdef JEMALLOC_ZONE
+#include <mach/mach_error.h>
+#include <mach/mach_init.h>
+#include <mach/vm_map.h>
+#endif
+
+#include "jemalloc/internal/jemalloc_internal_macros.h"
+
+/*
+ * Note that the ordering matters here; the hook itself is name-mangled.  We
+ * want the inclusion of hooks to happen early, so that we hook as much as
+ * possible.
+ */
+#ifndef JEMALLOC_NO_PRIVATE_NAMESPACE
+#  ifndef JEMALLOC_JET
+#    include "jemalloc/internal/private_namespace.h"
+#  else
+#    include "jemalloc/internal/private_namespace_jet.h"
+#  endif
+#endif
+#include "jemalloc/internal/test_hooks.h"
+
+#ifdef JEMALLOC_DEFINE_MADVISE_FREE
+#  define JEMALLOC_MADV_FREE 8
+#endif
+
+static const bool config_debug =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_dss =
+#ifdef JEMALLOC_DSS
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_madvise_huge =
+#ifdef JEMALLOC_HAVE_MADVISE_HUGE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_fill =
+#ifdef JEMALLOC_FILL
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_lazy_lock =
+#ifdef JEMALLOC_LAZY_LOCK
+    true
+#else
+    false
+#endif
+    ;
+static const char * const config_malloc_conf = JEMALLOC_CONFIG_MALLOC_CONF;
+static const bool config_prof =
+#ifdef JEMALLOC_PROF
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libgcc =
+#ifdef JEMALLOC_PROF_LIBGCC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libunwind =
+#ifdef JEMALLOC_PROF_LIBUNWIND
+    true
+#else
+    false
+#endif
+    ;
+static const bool maps_coalesce =
+#ifdef JEMALLOC_MAPS_COALESCE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_stats =
+#ifdef JEMALLOC_STATS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_tls =
+#ifdef JEMALLOC_TLS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_utrace =
+#ifdef JEMALLOC_UTRACE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_xmalloc =
+#ifdef JEMALLOC_XMALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_cache_oblivious =
+#ifdef JEMALLOC_CACHE_OBLIVIOUS
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Undocumented, for jemalloc development use only at the moment.  See the note
+ * in jemalloc/internal/log.h.
+ */
+static const bool config_log =
+#ifdef JEMALLOC_LOG
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Are extra safety checks enabled; things like checking the size of sized
+ * deallocations, double-frees, etc.
+ */
+static const bool config_opt_safety_checks =
+#ifdef JEMALLOC_OPT_SAFETY_CHECKS
+    true
+#elif defined(JEMALLOC_DEBUG)
+    /*
+     * This lets us only guard safety checks by one flag instead of two; fast
+     * checks can guard solely by config_opt_safety_checks and run in debug mode
+     * too.
+     */
+    true
+#else
+    false
+#endif
+    ;
+
+#if defined(_WIN32) || defined(JEMALLOC_HAVE_SCHED_GETCPU)
+/* Currently percpu_arena depends on sched_getcpu. */
+#define JEMALLOC_PERCPU_ARENA
+#endif
+static const bool have_percpu_arena =
+#ifdef JEMALLOC_PERCPU_ARENA
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Undocumented, and not recommended; the application should take full
+ * responsibility for tracking provenance.
+ */
+static const bool force_ivsalloc =
+#ifdef JEMALLOC_FORCE_IVSALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_background_thread =
+#ifdef JEMALLOC_BACKGROUND_THREAD
+    true
+#else
+    false
+#endif
+    ;
+
+#endif /* JEMALLOC_PREAMBLE_H */
diff --git a/jemalloc/include/jemalloc/internal/large_externs.h b/jemalloc/include/jemalloc/internal/large_externs.h
new file mode 100644
index 0000000..a05019e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/large_externs.h
@@ -0,0 +1,32 @@
+#ifndef JEMALLOC_INTERNAL_LARGE_EXTERNS_H
+#define JEMALLOC_INTERNAL_LARGE_EXTERNS_H
+
+#include "jemalloc/internal/hook.h"
+
+void *large_malloc(tsdn_t *tsdn, arena_t *arena, size_t usize, bool zero);
+void *large_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
+    bool zero);
+bool large_ralloc_no_move(tsdn_t *tsdn, extent_t *extent, size_t usize_min,
+    size_t usize_max, bool zero);
+void *large_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args);
+
+typedef void (large_dalloc_junk_t)(void *, size_t);
+extern large_dalloc_junk_t *JET_MUTABLE large_dalloc_junk;
+
+typedef void (large_dalloc_maybe_junk_t)(void *, size_t);
+extern large_dalloc_maybe_junk_t *JET_MUTABLE large_dalloc_maybe_junk;
+
+void large_dalloc_prep_junked_locked(tsdn_t *tsdn, extent_t *extent);
+void large_dalloc_finish(tsdn_t *tsdn, extent_t *extent);
+void large_dalloc(tsdn_t *tsdn, extent_t *extent);
+size_t large_salloc(tsdn_t *tsdn, const extent_t *extent);
+prof_tctx_t *large_prof_tctx_get(tsdn_t *tsdn, const extent_t *extent);
+void large_prof_tctx_set(tsdn_t *tsdn, extent_t *extent, prof_tctx_t *tctx);
+void large_prof_tctx_reset(tsdn_t *tsdn, extent_t *extent);
+
+nstime_t large_prof_alloc_time_get(const extent_t *extent);
+void large_prof_alloc_time_set(extent_t *extent, nstime_t time);
+
+#endif /* JEMALLOC_INTERNAL_LARGE_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/log.h b/jemalloc/include/jemalloc/internal/log.h
new file mode 100644
index 0000000..6420858
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/log.h
@@ -0,0 +1,115 @@
+#ifndef JEMALLOC_INTERNAL_LOG_H
+#define JEMALLOC_INTERNAL_LOG_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex.h"
+
+#ifdef JEMALLOC_LOG
+#  define JEMALLOC_LOG_VAR_BUFSIZE 1000
+#else
+#  define JEMALLOC_LOG_VAR_BUFSIZE 1
+#endif
+
+#define JEMALLOC_LOG_BUFSIZE 4096
+
+/*
+ * The log malloc_conf option is a '|'-delimited list of log_var name segments
+ * which should be logged.  The names are themselves hierarchical, with '.' as
+ * the delimiter (a "segment" is just a prefix in the log namespace).  So, if
+ * you have:
+ *
+ * log("arena", "log msg for arena"); // 1
+ * log("arena.a", "log msg for arena.a"); // 2
+ * log("arena.b", "log msg for arena.b"); // 3
+ * log("arena.a.a", "log msg for arena.a.a"); // 4
+ * log("extent.a", "log msg for extent.a"); // 5
+ * log("extent.b", "log msg for extent.b"); // 6
+ *
+ * And your malloc_conf option is "log=arena.a|extent", then lines 2, 4, 5, and
+ * 6 will print at runtime.  You can enable logging from all log vars by
+ * writing "log=.".
+ *
+ * None of this should be regarded as a stable API for right now.  It's intended
+ * as a debugging interface, to let us keep around some of our printf-debugging
+ * statements.
+ */
+
+extern char log_var_names[JEMALLOC_LOG_VAR_BUFSIZE];
+extern atomic_b_t log_init_done;
+
+typedef struct log_var_s log_var_t;
+struct log_var_s {
+	/*
+	 * Lowest bit is "inited", second lowest is "enabled".  Putting them in
+	 * a single word lets us avoid any fences on weak architectures.
+	 */
+	atomic_u_t state;
+	const char *name;
+};
+
+#define LOG_NOT_INITIALIZED 0U
+#define LOG_INITIALIZED_NOT_ENABLED 1U
+#define LOG_ENABLED 2U
+
+#define LOG_VAR_INIT(name_str) {ATOMIC_INIT(LOG_NOT_INITIALIZED), name_str}
+
+/*
+ * Returns the value we should assume for state (which is not necessarily
+ * accurate; if logging is done before logging has finished initializing, then
+ * we default to doing the safe thing by logging everything).
+ */
+unsigned log_var_update_state(log_var_t *log_var);
+
+/* We factor out the metadata management to allow us to test more easily. */
+#define log_do_begin(log_var)						\
+if (config_log) {							\
+	unsigned log_state = atomic_load_u(&(log_var).state,		\
+	    ATOMIC_RELAXED);						\
+	if (unlikely(log_state == LOG_NOT_INITIALIZED)) {		\
+		log_state = log_var_update_state(&(log_var));		\
+		assert(log_state != LOG_NOT_INITIALIZED);		\
+	}								\
+	if (log_state == LOG_ENABLED) {					\
+		{
+			/* User code executes here. */
+#define log_do_end(log_var)						\
+		}							\
+	}								\
+}
+
+/*
+ * MSVC has some preprocessor bugs in its expansion of __VA_ARGS__ during
+ * preprocessing.  To work around this, we take all potential extra arguments in
+ * a var-args functions.  Since a varargs macro needs at least one argument in
+ * the "...", we accept the format string there, and require that the first
+ * argument in this "..." is a const char *.
+ */
+static inline void
+log_impl_varargs(const char *name, ...) {
+	char buf[JEMALLOC_LOG_BUFSIZE];
+	va_list ap;
+
+	va_start(ap, name);
+	const char *format = va_arg(ap, const char *);
+	size_t dst_offset = 0;
+	dst_offset += malloc_snprintf(buf, JEMALLOC_LOG_BUFSIZE, "%s: ", name);
+	dst_offset += malloc_vsnprintf(buf + dst_offset,
+	    JEMALLOC_LOG_BUFSIZE - dst_offset, format, ap);
+	dst_offset += malloc_snprintf(buf + dst_offset,
+	    JEMALLOC_LOG_BUFSIZE - dst_offset, "\n");
+	va_end(ap);
+
+	malloc_write(buf);
+}
+
+/* Call as log("log.var.str", "format_string %d", arg_for_format_string); */
+#define LOG(log_var_str, ...)						\
+do {									\
+	static log_var_t log_var = LOG_VAR_INIT(log_var_str);		\
+	log_do_begin(log_var)						\
+		log_impl_varargs((log_var).name, __VA_ARGS__);		\
+	log_do_end(log_var)						\
+} while (0)
+
+#endif /* JEMALLOC_INTERNAL_LOG_H */
diff --git a/jemalloc/include/jemalloc/internal/malloc_io.h b/jemalloc/include/jemalloc/internal/malloc_io.h
new file mode 100644
index 0000000..1d1a414
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/malloc_io.h
@@ -0,0 +1,102 @@
+#ifndef JEMALLOC_INTERNAL_MALLOC_IO_H
+#define JEMALLOC_INTERNAL_MALLOC_IO_H
+
+#ifdef _WIN32
+#  ifdef _WIN64
+#    define FMT64_PREFIX "ll"
+#    define FMTPTR_PREFIX "ll"
+#  else
+#    define FMT64_PREFIX "ll"
+#    define FMTPTR_PREFIX ""
+#  endif
+#  define FMTd32 "d"
+#  define FMTu32 "u"
+#  define FMTx32 "x"
+#  define FMTd64 FMT64_PREFIX "d"
+#  define FMTu64 FMT64_PREFIX "u"
+#  define FMTx64 FMT64_PREFIX "x"
+#  define FMTdPTR FMTPTR_PREFIX "d"
+#  define FMTuPTR FMTPTR_PREFIX "u"
+#  define FMTxPTR FMTPTR_PREFIX "x"
+#else
+#  include <inttypes.h>
+#  define FMTd32 PRId32
+#  define FMTu32 PRIu32
+#  define FMTx32 PRIx32
+#  define FMTd64 PRId64
+#  define FMTu64 PRIu64
+#  define FMTx64 PRIx64
+#  define FMTdPTR PRIdPTR
+#  define FMTuPTR PRIuPTR
+#  define FMTxPTR PRIxPTR
+#endif
+
+/* Size of stack-allocated buffer passed to buferror(). */
+#define BUFERROR_BUF		64
+
+/*
+ * Size of stack-allocated buffer used by malloc_{,v,vc}printf().  This must be
+ * large enough for all possible uses within jemalloc.
+ */
+#define MALLOC_PRINTF_BUFSIZE	4096
+
+int buferror(int err, char *buf, size_t buflen);
+uintmax_t malloc_strtoumax(const char *restrict nptr, char **restrict endptr,
+    int base);
+void malloc_write(const char *s);
+
+/*
+ * malloc_vsnprintf() supports a subset of snprintf(3) that avoids floating
+ * point math.
+ */
+size_t malloc_vsnprintf(char *str, size_t size, const char *format,
+    va_list ap);
+size_t malloc_snprintf(char *str, size_t size, const char *format, ...)
+    JEMALLOC_FORMAT_PRINTF(3, 4);
+/*
+ * The caller can set write_cb to null to choose to print with the
+ * je_malloc_message hook.
+ */
+void malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap);
+void malloc_cprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, ...) JEMALLOC_FORMAT_PRINTF(3, 4);
+void malloc_printf(const char *format, ...) JEMALLOC_FORMAT_PRINTF(1, 2);
+
+static inline ssize_t
+malloc_write_fd(int fd, const void *buf, size_t count) {
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_write)
+	/*
+	 * Use syscall(2) rather than write(2) when possible in order to avoid
+	 * the possibility of memory allocation within libc.  This is necessary
+	 * on FreeBSD; most operating systems do not have this problem though.
+	 *
+	 * syscall() returns long or int, depending on platform, so capture the
+	 * result in the widest plausible type to avoid compiler warnings.
+	 */
+	long result = syscall(SYS_write, fd, buf, count);
+#else
+	ssize_t result = (ssize_t)write(fd, buf,
+#ifdef _WIN32
+	    (unsigned int)
+#endif
+	    count);
+#endif
+	return (ssize_t)result;
+}
+
+static inline ssize_t
+malloc_read_fd(int fd, void *buf, size_t count) {
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_read)
+	long result = syscall(SYS_read, fd, buf, count);
+#else
+	ssize_t result = read(fd, buf,
+#ifdef _WIN32
+	    (unsigned int)
+#endif
+	    count);
+#endif
+	return (ssize_t)result;
+}
+
+#endif /* JEMALLOC_INTERNAL_MALLOC_IO_H */
diff --git a/jemalloc/include/jemalloc/internal/mutex.h b/jemalloc/include/jemalloc/internal/mutex.h
new file mode 100644
index 0000000..18b170b
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/mutex.h
@@ -0,0 +1,288 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_H
+#define JEMALLOC_INTERNAL_MUTEX_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/witness.h"
+
+typedef enum {
+	/* Can only acquire one mutex of a given witness rank at a time. */
+	malloc_mutex_rank_exclusive,
+	/*
+	 * Can acquire multiple mutexes of the same witness rank, but in
+	 * address-ascending order only.
+	 */
+	malloc_mutex_address_ordered
+} malloc_mutex_lock_order_t;
+
+typedef struct malloc_mutex_s malloc_mutex_t;
+struct malloc_mutex_s {
+	union {
+		struct {
+			/*
+			 * prof_data is defined first to reduce cacheline
+			 * bouncing: the data is not touched by the mutex holder
+			 * during unlocking, while might be modified by
+			 * contenders.  Having it before the mutex itself could
+			 * avoid prefetching a modified cacheline (for the
+			 * unlocking thread).
+			 */
+			mutex_prof_data_t	prof_data;
+#ifdef _WIN32
+#  if _WIN32_WINNT >= 0x0600
+			SRWLOCK         	lock;
+#  else
+			CRITICAL_SECTION	lock;
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+			os_unfair_lock		lock;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+			pthread_mutex_t		lock;
+			malloc_mutex_t		*postponed_next;
+#else
+			pthread_mutex_t		lock;
+#endif
+			/*
+			 * Hint flag to avoid exclusive cache line contention
+			 * during spin waiting
+			 */
+			atomic_b_t		locked;
+		};
+		/*
+		 * We only touch witness when configured w/ debug.  However we
+		 * keep the field in a union when !debug so that we don't have
+		 * to pollute the code base with #ifdefs, while avoid paying the
+		 * memory cost.
+		 */
+#if !defined(JEMALLOC_DEBUG)
+		witness_t			witness;
+		malloc_mutex_lock_order_t	lock_order;
+#endif
+	};
+
+#if defined(JEMALLOC_DEBUG)
+	witness_t			witness;
+	malloc_mutex_lock_order_t	lock_order;
+#endif
+};
+
+/*
+ * Based on benchmark results, a fixed spin with this amount of retries works
+ * well for our critical sections.
+ */
+#define MALLOC_MUTEX_MAX_SPIN 250
+
+#ifdef _WIN32
+#  if _WIN32_WINNT >= 0x0600
+#    define MALLOC_MUTEX_LOCK(m)    AcquireSRWLockExclusive(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  ReleaseSRWLockExclusive(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!TryAcquireSRWLockExclusive(&(m)->lock))
+#  else
+#    define MALLOC_MUTEX_LOCK(m)    EnterCriticalSection(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  LeaveCriticalSection(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!TryEnterCriticalSection(&(m)->lock))
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#    define MALLOC_MUTEX_LOCK(m)    os_unfair_lock_lock(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  os_unfair_lock_unlock(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!os_unfair_lock_trylock(&(m)->lock))
+#else
+#    define MALLOC_MUTEX_LOCK(m)    pthread_mutex_lock(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  pthread_mutex_unlock(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (pthread_mutex_trylock(&(m)->lock) != 0)
+#endif
+
+#define LOCK_PROF_DATA_INITIALIZER					\
+    {NSTIME_ZERO_INITIALIZER, NSTIME_ZERO_INITIALIZER, 0, 0, 0,		\
+	    ATOMIC_INIT(0), 0, NULL, 0}
+
+#ifdef _WIN32
+#  define MALLOC_MUTEX_INITIALIZER
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#  if defined(JEMALLOC_DEBUG)
+#    define MALLOC_MUTEX_INITIALIZER					\
+  {{{LOCK_PROF_DATA_INITIALIZER, OS_UNFAIR_LOCK_INIT, ATOMIC_INIT(false)}}, \
+         WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT), 0}
+#  else
+#    define MALLOC_MUTEX_INITIALIZER                      \
+  {{{LOCK_PROF_DATA_INITIALIZER, OS_UNFAIR_LOCK_INIT, ATOMIC_INIT(false)}},  \
+      WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#  endif
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+#  if (defined(JEMALLOC_DEBUG))
+#     define MALLOC_MUTEX_INITIALIZER					\
+      {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, NULL, ATOMIC_INIT(false)}},	\
+           WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT), 0}
+#  else
+#     define MALLOC_MUTEX_INITIALIZER					\
+      {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, NULL, ATOMIC_INIT(false)}},	\
+           WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#  endif
+
+#else
+#    define MALLOC_MUTEX_TYPE PTHREAD_MUTEX_DEFAULT
+#  if defined(JEMALLOC_DEBUG)
+#    define MALLOC_MUTEX_INITIALIZER					\
+     {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, ATOMIC_INIT(false)}}, \
+           WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT), 0}
+#  else
+#    define MALLOC_MUTEX_INITIALIZER                          \
+     {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, ATOMIC_INIT(false)}},	\
+      WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#  endif
+#endif
+
+#ifdef JEMALLOC_LAZY_LOCK
+extern bool isthreaded;
+#else
+#  undef isthreaded /* Undo private_namespace.h definition. */
+#  define isthreaded true
+#endif
+
+bool malloc_mutex_init(malloc_mutex_t *mutex, const char *name,
+    witness_rank_t rank, malloc_mutex_lock_order_t lock_order);
+void malloc_mutex_prefork(tsdn_t *tsdn, malloc_mutex_t *mutex);
+void malloc_mutex_postfork_parent(tsdn_t *tsdn, malloc_mutex_t *mutex);
+void malloc_mutex_postfork_child(tsdn_t *tsdn, malloc_mutex_t *mutex);
+bool malloc_mutex_boot(void);
+void malloc_mutex_prof_data_reset(tsdn_t *tsdn, malloc_mutex_t *mutex);
+
+void malloc_mutex_lock_slow(malloc_mutex_t *mutex);
+
+static inline void
+malloc_mutex_lock_final(malloc_mutex_t *mutex) {
+	MALLOC_MUTEX_LOCK(mutex);
+	atomic_store_b(&mutex->locked, true, ATOMIC_RELAXED);
+}
+
+static inline bool
+malloc_mutex_trylock_final(malloc_mutex_t *mutex) {
+	return MALLOC_MUTEX_TRYLOCK(mutex);
+}
+
+static inline void
+mutex_owner_stats_update(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	if (config_stats) {
+		mutex_prof_data_t *data = &mutex->prof_data;
+		data->n_lock_ops++;
+		if (data->prev_owner != tsdn) {
+			data->prev_owner = tsdn;
+			data->n_owner_switches++;
+		}
+	}
+}
+
+/* Trylock: return false if the lock is successfully acquired. */
+static inline bool
+malloc_mutex_trylock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		if (malloc_mutex_trylock_final(mutex)) {
+			atomic_store_b(&mutex->locked, true, ATOMIC_RELAXED);
+			return true;
+		}
+		mutex_owner_stats_update(tsdn, mutex);
+	}
+	witness_lock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+
+	return false;
+}
+
+/* Aggregate lock prof data. */
+static inline void
+malloc_mutex_prof_merge(mutex_prof_data_t *sum, mutex_prof_data_t *data) {
+	nstime_add(&sum->tot_wait_time, &data->tot_wait_time);
+	if (nstime_compare(&sum->max_wait_time, &data->max_wait_time) < 0) {
+		nstime_copy(&sum->max_wait_time, &data->max_wait_time);
+	}
+
+	sum->n_wait_times += data->n_wait_times;
+	sum->n_spin_acquired += data->n_spin_acquired;
+
+	if (sum->max_n_thds < data->max_n_thds) {
+		sum->max_n_thds = data->max_n_thds;
+	}
+	uint32_t cur_n_waiting_thds = atomic_load_u32(&sum->n_waiting_thds,
+	    ATOMIC_RELAXED);
+	uint32_t new_n_waiting_thds = cur_n_waiting_thds + atomic_load_u32(
+	    &data->n_waiting_thds, ATOMIC_RELAXED);
+	atomic_store_u32(&sum->n_waiting_thds, new_n_waiting_thds,
+	    ATOMIC_RELAXED);
+	sum->n_owner_switches += data->n_owner_switches;
+	sum->n_lock_ops += data->n_lock_ops;
+}
+
+static inline void
+malloc_mutex_lock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		if (malloc_mutex_trylock_final(mutex)) {
+			malloc_mutex_lock_slow(mutex);
+			atomic_store_b(&mutex->locked, true, ATOMIC_RELAXED);
+		}
+		mutex_owner_stats_update(tsdn, mutex);
+	}
+	witness_lock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+static inline void
+malloc_mutex_unlock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	atomic_store_b(&mutex->locked, false, ATOMIC_RELAXED);
+	witness_unlock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		MALLOC_MUTEX_UNLOCK(mutex);
+	}
+}
+
+static inline void
+malloc_mutex_assert_owner(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+static inline void
+malloc_mutex_assert_not_owner(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+/* Copy the prof data from mutex for processing. */
+static inline void
+malloc_mutex_prof_read(tsdn_t *tsdn, mutex_prof_data_t *data,
+    malloc_mutex_t *mutex) {
+	mutex_prof_data_t *source = &mutex->prof_data;
+	/* Can only read holding the mutex. */
+	malloc_mutex_assert_owner(tsdn, mutex);
+
+	/*
+	 * Not *really* allowed (we shouldn't be doing non-atomic loads of
+	 * atomic data), but the mutex protection makes this safe, and writing
+	 * a member-for-member copy is tedious for this situation.
+	 */
+	*data = *source;
+	/* n_wait_thds is not reported (modified w/o locking). */
+	atomic_store_u32(&data->n_waiting_thds, 0, ATOMIC_RELAXED);
+}
+
+static inline void
+malloc_mutex_prof_accum(tsdn_t *tsdn, mutex_prof_data_t *data,
+    malloc_mutex_t *mutex) {
+	mutex_prof_data_t *source = &mutex->prof_data;
+	/* Can only read holding the mutex. */
+	malloc_mutex_assert_owner(tsdn, mutex);
+
+	nstime_add(&data->tot_wait_time, &source->tot_wait_time);
+	if (nstime_compare(&source->max_wait_time, &data->max_wait_time) > 0) {
+		nstime_copy(&data->max_wait_time, &source->max_wait_time);
+	}
+	data->n_wait_times += source->n_wait_times;
+	data->n_spin_acquired += source->n_spin_acquired;
+	if (data->max_n_thds < source->max_n_thds) {
+		data->max_n_thds = source->max_n_thds;
+	}
+	/* n_wait_thds is not reported. */
+	atomic_store_u32(&data->n_waiting_thds, 0, ATOMIC_RELAXED);
+	data->n_owner_switches += source->n_owner_switches;
+	data->n_lock_ops += source->n_lock_ops;
+}
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_H */
diff --git a/jemalloc/include/jemalloc/internal/mutex_pool.h b/jemalloc/include/jemalloc/internal/mutex_pool.h
new file mode 100644
index 0000000..726cece
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/mutex_pool.h
@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_POOL_H
+#define JEMALLOC_INTERNAL_MUTEX_POOL_H
+
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/witness.h"
+
+/* We do mod reductions by this value, so it should be kept a power of 2. */
+#define MUTEX_POOL_SIZE 256
+
+typedef struct mutex_pool_s mutex_pool_t;
+struct mutex_pool_s {
+	malloc_mutex_t mutexes[MUTEX_POOL_SIZE];
+};
+
+bool mutex_pool_init(mutex_pool_t *pool, const char *name, witness_rank_t rank);
+
+/* Internal helper - not meant to be called outside this module. */
+static inline malloc_mutex_t *
+mutex_pool_mutex(mutex_pool_t *pool, uintptr_t key) {
+	size_t hash_result[2];
+	hash(&key, sizeof(key), 0xd50dcc1b, hash_result);
+	return &pool->mutexes[hash_result[0] % MUTEX_POOL_SIZE];
+}
+
+static inline void
+mutex_pool_assert_not_held(tsdn_t *tsdn, mutex_pool_t *pool) {
+	for (int i = 0; i < MUTEX_POOL_SIZE; i++) {
+		malloc_mutex_assert_not_owner(tsdn, &pool->mutexes[i]);
+	}
+}
+
+/*
+ * Note that a mutex pool doesn't work exactly the way an embdedded mutex would.
+ * You're not allowed to acquire mutexes in the pool one at a time.  You have to
+ * acquire all the mutexes you'll need in a single function call, and then
+ * release them all in a single function call.
+ */
+
+static inline void
+mutex_pool_lock(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	mutex_pool_assert_not_held(tsdn, pool);
+
+	malloc_mutex_t *mutex = mutex_pool_mutex(pool, key);
+	malloc_mutex_lock(tsdn, mutex);
+}
+
+static inline void
+mutex_pool_unlock(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	malloc_mutex_t *mutex = mutex_pool_mutex(pool, key);
+	malloc_mutex_unlock(tsdn, mutex);
+
+	mutex_pool_assert_not_held(tsdn, pool);
+}
+
+static inline void
+mutex_pool_lock2(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key1,
+    uintptr_t key2) {
+	mutex_pool_assert_not_held(tsdn, pool);
+
+	malloc_mutex_t *mutex1 = mutex_pool_mutex(pool, key1);
+	malloc_mutex_t *mutex2 = mutex_pool_mutex(pool, key2);
+	if ((uintptr_t)mutex1 < (uintptr_t)mutex2) {
+		malloc_mutex_lock(tsdn, mutex1);
+		malloc_mutex_lock(tsdn, mutex2);
+	} else if ((uintptr_t)mutex1 == (uintptr_t)mutex2) {
+		malloc_mutex_lock(tsdn, mutex1);
+	} else {
+		malloc_mutex_lock(tsdn, mutex2);
+		malloc_mutex_lock(tsdn, mutex1);
+	}
+}
+
+static inline void
+mutex_pool_unlock2(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key1,
+    uintptr_t key2) {
+	malloc_mutex_t *mutex1 = mutex_pool_mutex(pool, key1);
+	malloc_mutex_t *mutex2 = mutex_pool_mutex(pool, key2);
+	if (mutex1 == mutex2) {
+		malloc_mutex_unlock(tsdn, mutex1);
+	} else {
+		malloc_mutex_unlock(tsdn, mutex1);
+		malloc_mutex_unlock(tsdn, mutex2);
+	}
+
+	mutex_pool_assert_not_held(tsdn, pool);
+}
+
+static inline void
+mutex_pool_assert_owner(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	malloc_mutex_assert_owner(tsdn, mutex_pool_mutex(pool, key));
+}
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_POOL_H */
diff --git a/jemalloc/include/jemalloc/internal/mutex_prof.h b/jemalloc/include/jemalloc/internal/mutex_prof.h
new file mode 100644
index 0000000..2cb8fb0
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/mutex_prof.h
@@ -0,0 +1,108 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_PROF_H
+#define JEMALLOC_INTERNAL_MUTEX_PROF_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/tsd_types.h"
+
+#define MUTEX_PROF_GLOBAL_MUTEXES					\
+    OP(background_thread)						\
+    OP(ctl)								\
+    OP(prof)
+
+typedef enum {
+#define OP(mtx) global_prof_mutex_##mtx,
+	MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+	mutex_prof_num_global_mutexes
+} mutex_prof_global_ind_t;
+
+#define MUTEX_PROF_ARENA_MUTEXES					\
+    OP(large)								\
+    OP(extent_avail)							\
+    OP(extents_dirty)							\
+    OP(extents_muzzy)							\
+    OP(extents_retained)						\
+    OP(decay_dirty)							\
+    OP(decay_muzzy)							\
+    OP(base)								\
+    OP(tcache_list)
+
+typedef enum {
+#define OP(mtx) arena_prof_mutex_##mtx,
+	MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+	mutex_prof_num_arena_mutexes
+} mutex_prof_arena_ind_t;
+
+/*
+ * The forth parameter is a boolean value that is true for derived rate counters
+ * and false for real ones.
+ */
+#define MUTEX_PROF_UINT64_COUNTERS					\
+    OP(num_ops, uint64_t, "n_lock_ops", false, num_ops)					\
+    OP(num_ops_ps, uint64_t, "(#/sec)", true, num_ops)				\
+    OP(num_wait, uint64_t, "n_waiting", false, num_wait)				\
+    OP(num_wait_ps, uint64_t, "(#/sec)", true, num_wait)				\
+    OP(num_spin_acq, uint64_t, "n_spin_acq", false, num_spin_acq)			\
+    OP(num_spin_acq_ps, uint64_t, "(#/sec)", true, num_spin_acq)			\
+    OP(num_owner_switch, uint64_t, "n_owner_switch", false, num_owner_switch)		\
+    OP(num_owner_switch_ps, uint64_t, "(#/sec)", true, num_owner_switch)	\
+    OP(total_wait_time, uint64_t, "total_wait_ns", false, total_wait_time)		\
+    OP(total_wait_time_ps, uint64_t, "(#/sec)", true, total_wait_time)		\
+    OP(max_wait_time, uint64_t, "max_wait_ns", false, max_wait_time)
+
+#define MUTEX_PROF_UINT32_COUNTERS					\
+    OP(max_num_thds, uint32_t, "max_n_thds", false, max_num_thds)
+
+#define MUTEX_PROF_COUNTERS						\
+		MUTEX_PROF_UINT64_COUNTERS				\
+		MUTEX_PROF_UINT32_COUNTERS
+
+#define OP(counter, type, human, derived, base_counter) mutex_counter_##counter,
+
+#define COUNTER_ENUM(counter_list, t)					\
+		typedef enum {						\
+			counter_list					\
+			mutex_prof_num_##t##_counters			\
+		} mutex_prof_##t##_counter_ind_t;
+
+COUNTER_ENUM(MUTEX_PROF_UINT64_COUNTERS, uint64_t)
+COUNTER_ENUM(MUTEX_PROF_UINT32_COUNTERS, uint32_t)
+
+#undef COUNTER_ENUM
+#undef OP
+
+typedef struct {
+	/*
+	 * Counters touched on the slow path, i.e. when there is lock
+	 * contention.  We update them once we have the lock.
+	 */
+	/* Total time (in nano seconds) spent waiting on this mutex. */
+	nstime_t		tot_wait_time;
+	/* Max time (in nano seconds) spent on a single lock operation. */
+	nstime_t		max_wait_time;
+	/* # of times have to wait for this mutex (after spinning). */
+	uint64_t		n_wait_times;
+	/* # of times acquired the mutex through local spinning. */
+	uint64_t		n_spin_acquired;
+	/* Max # of threads waiting for the mutex at the same time. */
+	uint32_t		max_n_thds;
+	/* Current # of threads waiting on the lock.  Atomic synced. */
+	atomic_u32_t		n_waiting_thds;
+
+	/*
+	 * Data touched on the fast path.  These are modified right after we
+	 * grab the lock, so it's placed closest to the end (i.e. right before
+	 * the lock) so that we have a higher chance of them being on the same
+	 * cacheline.
+	 */
+	/* # of times the mutex holder is different than the previous one. */
+	uint64_t		n_owner_switches;
+	/* Previous mutex holder, to facilitate n_owner_switches. */
+	tsdn_t			*prev_owner;
+	/* # of lock() operations in total. */
+	uint64_t		n_lock_ops;
+} mutex_prof_data_t;
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_PROF_H */
diff --git a/jemalloc/include/jemalloc/internal/nstime.h b/jemalloc/include/jemalloc/internal/nstime.h
new file mode 100644
index 0000000..17c177c
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/nstime.h
@@ -0,0 +1,34 @@
+#ifndef JEMALLOC_INTERNAL_NSTIME_H
+#define JEMALLOC_INTERNAL_NSTIME_H
+
+/* Maximum supported number of seconds (~584 years). */
+#define NSTIME_SEC_MAX KQU(18446744072)
+#define NSTIME_ZERO_INITIALIZER {0}
+
+typedef struct {
+	uint64_t ns;
+} nstime_t;
+
+void nstime_init(nstime_t *time, uint64_t ns);
+void nstime_init2(nstime_t *time, uint64_t sec, uint64_t nsec);
+uint64_t nstime_ns(const nstime_t *time);
+uint64_t nstime_sec(const nstime_t *time);
+uint64_t nstime_msec(const nstime_t *time);
+uint64_t nstime_nsec(const nstime_t *time);
+void nstime_copy(nstime_t *time, const nstime_t *source);
+int nstime_compare(const nstime_t *a, const nstime_t *b);
+void nstime_add(nstime_t *time, const nstime_t *addend);
+void nstime_iadd(nstime_t *time, uint64_t addend);
+void nstime_subtract(nstime_t *time, const nstime_t *subtrahend);
+void nstime_isubtract(nstime_t *time, uint64_t subtrahend);
+void nstime_imultiply(nstime_t *time, uint64_t multiplier);
+void nstime_idivide(nstime_t *time, uint64_t divisor);
+uint64_t nstime_divide(const nstime_t *time, const nstime_t *divisor);
+
+typedef bool (nstime_monotonic_t)(void);
+extern nstime_monotonic_t *JET_MUTABLE nstime_monotonic;
+
+typedef bool (nstime_update_t)(nstime_t *);
+extern nstime_update_t *JET_MUTABLE nstime_update;
+
+#endif /* JEMALLOC_INTERNAL_NSTIME_H */
diff --git a/jemalloc/include/jemalloc/internal/pages.h b/jemalloc/include/jemalloc/internal/pages.h
new file mode 100644
index 0000000..7dae633
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/pages.h
@@ -0,0 +1,88 @@
+#ifndef JEMALLOC_INTERNAL_PAGES_EXTERNS_H
+#define JEMALLOC_INTERNAL_PAGES_EXTERNS_H
+
+/* Page size.  LG_PAGE is determined by the configure script. */
+#ifdef PAGE_MASK
+#  undef PAGE_MASK
+#endif
+#define PAGE		((size_t)(1U << LG_PAGE))
+#define PAGE_MASK	((size_t)(PAGE - 1))
+/* Return the page base address for the page containing address a. */
+#define PAGE_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~PAGE_MASK))
+/* Return the smallest pagesize multiple that is >= s. */
+#define PAGE_CEILING(s)							\
+	(((s) + PAGE_MASK) & ~PAGE_MASK)
+
+/* Huge page size.  LG_HUGEPAGE is determined by the configure script. */
+#define HUGEPAGE	((size_t)(1U << LG_HUGEPAGE))
+#define HUGEPAGE_MASK	((size_t)(HUGEPAGE - 1))
+/* Return the huge page base address for the huge page containing address a. */
+#define HUGEPAGE_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~HUGEPAGE_MASK))
+/* Return the smallest pagesize multiple that is >= s. */
+#define HUGEPAGE_CEILING(s)						\
+	(((s) + HUGEPAGE_MASK) & ~HUGEPAGE_MASK)
+
+/* PAGES_CAN_PURGE_LAZY is defined if lazy purging is supported. */
+#if defined(_WIN32) || defined(JEMALLOC_PURGE_MADVISE_FREE)
+#  define PAGES_CAN_PURGE_LAZY
+#endif
+/*
+ * PAGES_CAN_PURGE_FORCED is defined if forced purging is supported.
+ *
+ * The only supported way to hard-purge on Windows is to decommit and then
+ * re-commit, but doing so is racy, and if re-commit fails it's a pain to
+ * propagate the "poisoned" memory state.  Since we typically decommit as the
+ * next step after purging on Windows anyway, there's no point in adding such
+ * complexity.
+ */
+#if !defined(_WIN32) && ((defined(JEMALLOC_PURGE_MADVISE_DONTNEED) && \
+    defined(JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS)) || \
+    defined(JEMALLOC_MAPS_COALESCE))
+#  define PAGES_CAN_PURGE_FORCED
+#endif
+
+static const bool pages_can_purge_lazy =
+#ifdef PAGES_CAN_PURGE_LAZY
+    true
+#else
+    false
+#endif
+    ;
+static const bool pages_can_purge_forced =
+#ifdef PAGES_CAN_PURGE_FORCED
+    true
+#else
+    false
+#endif
+    ;
+
+typedef enum {
+	thp_mode_default       = 0, /* Do not change hugepage settings. */
+	thp_mode_always        = 1, /* Always set MADV_HUGEPAGE. */
+	thp_mode_never         = 2, /* Always set MADV_NOHUGEPAGE. */
+
+	thp_mode_names_limit   = 3, /* Used for option processing. */
+	thp_mode_not_supported = 3  /* No THP support detected. */
+} thp_mode_t;
+
+#define THP_MODE_DEFAULT thp_mode_default
+extern thp_mode_t opt_thp;
+extern thp_mode_t init_system_thp_mode; /* Initial system wide state. */
+extern const char *thp_mode_names[];
+
+void *pages_map(void *addr, size_t size, size_t alignment, bool *commit);
+void pages_unmap(void *addr, size_t size);
+bool pages_commit(void *addr, size_t size);
+bool pages_decommit(void *addr, size_t size);
+bool pages_purge_lazy(void *addr, size_t size);
+bool pages_purge_forced(void *addr, size_t size);
+bool pages_huge(void *addr, size_t size);
+bool pages_nohuge(void *addr, size_t size);
+bool pages_dontdump(void *addr, size_t size);
+bool pages_dodump(void *addr, size_t size);
+bool pages_boot(void);
+void pages_set_thp_state (void *ptr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_PAGES_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/ph.h b/jemalloc/include/jemalloc/internal/ph.h
new file mode 100644
index 0000000..84d6778
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/ph.h
@@ -0,0 +1,391 @@
+/*
+ * A Pairing Heap implementation.
+ *
+ * "The Pairing Heap: A New Form of Self-Adjusting Heap"
+ * https://www.cs.cmu.edu/~sleator/papers/pairing-heaps.pdf
+ *
+ * With auxiliary twopass list, described in a follow on paper.
+ *
+ * "Pairing Heaps: Experiments and Analysis"
+ * http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.106.2988&rep=rep1&type=pdf
+ *
+ *******************************************************************************
+ */
+
+#ifndef PH_H_
+#define PH_H_
+
+/* Node structure. */
+#define phn(a_type)							\
+struct {								\
+	a_type	*phn_prev;						\
+	a_type	*phn_next;						\
+	a_type	*phn_lchild;						\
+}
+
+/* Root structure. */
+#define ph(a_type)							\
+struct {								\
+	a_type	*ph_root;						\
+}
+
+/* Internal utility macros. */
+#define phn_lchild_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_lchild)
+#define phn_lchild_set(a_type, a_field, a_phn, a_lchild) do {		\
+	a_phn->a_field.phn_lchild = a_lchild;				\
+} while (0)
+
+#define phn_next_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_next)
+#define phn_prev_set(a_type, a_field, a_phn, a_prev) do {		\
+	a_phn->a_field.phn_prev = a_prev;				\
+} while (0)
+
+#define phn_prev_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_prev)
+#define phn_next_set(a_type, a_field, a_phn, a_next) do {		\
+	a_phn->a_field.phn_next = a_next;				\
+} while (0)
+
+#define phn_merge_ordered(a_type, a_field, a_phn0, a_phn1, a_cmp) do {	\
+	a_type *phn0child;						\
+									\
+	assert(a_phn0 != NULL);						\
+	assert(a_phn1 != NULL);						\
+	assert(a_cmp(a_phn0, a_phn1) <= 0);				\
+									\
+	phn_prev_set(a_type, a_field, a_phn1, a_phn0);			\
+	phn0child = phn_lchild_get(a_type, a_field, a_phn0);		\
+	phn_next_set(a_type, a_field, a_phn1, phn0child);		\
+	if (phn0child != NULL) {					\
+		phn_prev_set(a_type, a_field, phn0child, a_phn1);	\
+	}								\
+	phn_lchild_set(a_type, a_field, a_phn0, a_phn1);		\
+} while (0)
+
+#define phn_merge(a_type, a_field, a_phn0, a_phn1, a_cmp, r_phn) do {	\
+	if (a_phn0 == NULL) {						\
+		r_phn = a_phn1;						\
+	} else if (a_phn1 == NULL) {					\
+		r_phn = a_phn0;						\
+	} else if (a_cmp(a_phn0, a_phn1) < 0) {				\
+		phn_merge_ordered(a_type, a_field, a_phn0, a_phn1,	\
+		    a_cmp);						\
+		r_phn = a_phn0;						\
+	} else {							\
+		phn_merge_ordered(a_type, a_field, a_phn1, a_phn0,	\
+		    a_cmp);						\
+		r_phn = a_phn1;						\
+	}								\
+} while (0)
+
+#define ph_merge_siblings(a_type, a_field, a_phn, a_cmp, r_phn) do {	\
+	a_type *head = NULL;						\
+	a_type *tail = NULL;						\
+	a_type *phn0 = a_phn;						\
+	a_type *phn1 = phn_next_get(a_type, a_field, phn0);		\
+									\
+	/*								\
+	 * Multipass merge, wherein the first two elements of a FIFO	\
+	 * are repeatedly merged, and each result is appended to the	\
+	 * singly linked FIFO, until the FIFO contains only a single	\
+	 * element.  We start with a sibling list but no reference to	\
+	 * its tail, so we do a single pass over the sibling list to	\
+	 * populate the FIFO.						\
+	 */								\
+	if (phn1 != NULL) {						\
+		a_type *phnrest = phn_next_get(a_type, a_field, phn1);	\
+		if (phnrest != NULL) {					\
+			phn_prev_set(a_type, a_field, phnrest, NULL);	\
+		}							\
+		phn_prev_set(a_type, a_field, phn0, NULL);		\
+		phn_next_set(a_type, a_field, phn0, NULL);		\
+		phn_prev_set(a_type, a_field, phn1, NULL);		\
+		phn_next_set(a_type, a_field, phn1, NULL);		\
+		phn_merge(a_type, a_field, phn0, phn1, a_cmp, phn0);	\
+		head = tail = phn0;					\
+		phn0 = phnrest;						\
+		while (phn0 != NULL) {					\
+			phn1 = phn_next_get(a_type, a_field, phn0);	\
+			if (phn1 != NULL) {				\
+				phnrest = phn_next_get(a_type, a_field,	\
+				    phn1);				\
+				if (phnrest != NULL) {			\
+					phn_prev_set(a_type, a_field,	\
+					    phnrest, NULL);		\
+				}					\
+				phn_prev_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				phn_next_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				phn_prev_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_next_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_merge(a_type, a_field, phn0, phn1,	\
+				    a_cmp, phn0);			\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = phnrest;				\
+			} else {					\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = NULL;				\
+			}						\
+		}							\
+		phn0 = head;						\
+		phn1 = phn_next_get(a_type, a_field, phn0);		\
+		if (phn1 != NULL) {					\
+			while (true) {					\
+				head = phn_next_get(a_type, a_field,	\
+				    phn1);				\
+				assert(phn_prev_get(a_type, a_field,	\
+				    phn0) == NULL);			\
+				phn_next_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				assert(phn_prev_get(a_type, a_field,	\
+				    phn1) == NULL);			\
+				phn_next_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_merge(a_type, a_field, phn0, phn1,	\
+				    a_cmp, phn0);			\
+				if (head == NULL) {			\
+					break;				\
+				}					\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = head;				\
+				phn1 = phn_next_get(a_type, a_field,	\
+				    phn0);				\
+			}						\
+		}							\
+	}								\
+	r_phn = phn0;							\
+} while (0)
+
+#define ph_merge_aux(a_type, a_field, a_ph, a_cmp) do {			\
+	a_type *phn = phn_next_get(a_type, a_field, a_ph->ph_root);	\
+	if (phn != NULL) {						\
+		phn_prev_set(a_type, a_field, a_ph->ph_root, NULL);	\
+		phn_next_set(a_type, a_field, a_ph->ph_root, NULL);	\
+		phn_prev_set(a_type, a_field, phn, NULL);		\
+		ph_merge_siblings(a_type, a_field, phn, a_cmp, phn);	\
+		assert(phn_next_get(a_type, a_field, phn) == NULL);	\
+		phn_merge(a_type, a_field, a_ph->ph_root, phn, a_cmp,	\
+		    a_ph->ph_root);					\
+	}								\
+} while (0)
+
+#define ph_merge_children(a_type, a_field, a_phn, a_cmp, r_phn) do {	\
+	a_type *lchild = phn_lchild_get(a_type, a_field, a_phn);	\
+	if (lchild == NULL) {						\
+		r_phn = NULL;						\
+	} else {							\
+		ph_merge_siblings(a_type, a_field, lchild, a_cmp,	\
+		    r_phn);						\
+	}								\
+} while (0)
+
+/*
+ * The ph_proto() macro generates function prototypes that correspond to the
+ * functions generated by an equivalently parameterized call to ph_gen().
+ */
+#define ph_proto(a_attr, a_prefix, a_ph_type, a_type)			\
+a_attr void	a_prefix##new(a_ph_type *ph);				\
+a_attr bool	a_prefix##empty(a_ph_type *ph);				\
+a_attr a_type	*a_prefix##first(a_ph_type *ph);			\
+a_attr a_type	*a_prefix##any(a_ph_type *ph);				\
+a_attr void	a_prefix##insert(a_ph_type *ph, a_type *phn);		\
+a_attr a_type	*a_prefix##remove_first(a_ph_type *ph);			\
+a_attr a_type	*a_prefix##remove_any(a_ph_type *ph);			\
+a_attr void	a_prefix##remove(a_ph_type *ph, a_type *phn);
+
+/*
+ * The ph_gen() macro generates a type-specific pairing heap implementation,
+ * based on the above cpp macros.
+ */
+#define ph_gen(a_attr, a_prefix, a_ph_type, a_type, a_field, a_cmp)	\
+a_attr void								\
+a_prefix##new(a_ph_type *ph) {						\
+	memset(ph, 0, sizeof(ph(a_type)));				\
+}									\
+a_attr bool								\
+a_prefix##empty(a_ph_type *ph) {					\
+	return (ph->ph_root == NULL);					\
+}									\
+a_attr a_type *								\
+a_prefix##first(a_ph_type *ph) {					\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	ph_merge_aux(a_type, a_field, ph, a_cmp);			\
+	return ph->ph_root;						\
+}									\
+a_attr a_type *								\
+a_prefix##any(a_ph_type *ph) {						\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	a_type *aux = phn_next_get(a_type, a_field, ph->ph_root);	\
+	if (aux != NULL) {						\
+		return aux;						\
+	}								\
+	return ph->ph_root;						\
+}									\
+a_attr void								\
+a_prefix##insert(a_ph_type *ph, a_type *phn) {				\
+	memset(&phn->a_field, 0, sizeof(phn(a_type)));			\
+									\
+	/*								\
+	 * Treat the root as an aux list during insertion, and lazily	\
+	 * merge during a_prefix##remove_first().  For elements that	\
+	 * are inserted, then removed via a_prefix##remove() before the	\
+	 * aux list is ever processed, this makes insert/remove		\
+	 * constant-time, whereas eager merging would make insert	\
+	 * O(log n).							\
+	 */								\
+	if (ph->ph_root == NULL) {					\
+		ph->ph_root = phn;					\
+	} else {							\
+		phn_next_set(a_type, a_field, phn, phn_next_get(a_type,	\
+		    a_field, ph->ph_root));				\
+		if (phn_next_get(a_type, a_field, ph->ph_root) !=	\
+		    NULL) {						\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, ph->ph_root),	\
+			    phn);					\
+		}							\
+		phn_prev_set(a_type, a_field, phn, ph->ph_root);	\
+		phn_next_set(a_type, a_field, ph->ph_root, phn);	\
+	}								\
+}									\
+a_attr a_type *								\
+a_prefix##remove_first(a_ph_type *ph) {					\
+	a_type *ret;							\
+									\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	ph_merge_aux(a_type, a_field, ph, a_cmp);			\
+									\
+	ret = ph->ph_root;						\
+									\
+	ph_merge_children(a_type, a_field, ph->ph_root, a_cmp,		\
+	    ph->ph_root);						\
+									\
+	return ret;							\
+}									\
+a_attr a_type *								\
+a_prefix##remove_any(a_ph_type *ph) {					\
+	/*								\
+	 * Remove the most recently inserted aux list element, or the	\
+	 * root if the aux list is empty.  This has the effect of	\
+	 * behaving as a LIFO (and insertion/removal is therefore	\
+	 * constant-time) if a_prefix##[remove_]first() are never	\
+	 * called.							\
+	 */								\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	a_type *ret = phn_next_get(a_type, a_field, ph->ph_root);	\
+	if (ret != NULL) {						\
+		a_type *aux = phn_next_get(a_type, a_field, ret);	\
+		phn_next_set(a_type, a_field, ph->ph_root, aux);	\
+		if (aux != NULL) {					\
+			phn_prev_set(a_type, a_field, aux,		\
+			    ph->ph_root);				\
+		}							\
+		return ret;						\
+	}								\
+	ret = ph->ph_root;						\
+	ph_merge_children(a_type, a_field, ph->ph_root, a_cmp,		\
+	    ph->ph_root);						\
+	return ret;							\
+}									\
+a_attr void								\
+a_prefix##remove(a_ph_type *ph, a_type *phn) {				\
+	a_type *replace, *parent;					\
+									\
+	if (ph->ph_root == phn) {					\
+		/*							\
+		 * We can delete from aux list without merging it, but	\
+		 * we need to merge if we are dealing with the root	\
+		 * node and it has children.				\
+		 */							\
+		if (phn_lchild_get(a_type, a_field, phn) == NULL) {	\
+			ph->ph_root = phn_next_get(a_type, a_field,	\
+			    phn);					\
+			if (ph->ph_root != NULL) {			\
+				phn_prev_set(a_type, a_field,		\
+				    ph->ph_root, NULL);			\
+			}						\
+			return;						\
+		}							\
+		ph_merge_aux(a_type, a_field, ph, a_cmp);		\
+		if (ph->ph_root == phn) {				\
+			ph_merge_children(a_type, a_field, ph->ph_root,	\
+			    a_cmp, ph->ph_root);			\
+			return;						\
+		}							\
+	}								\
+									\
+	/* Get parent (if phn is leftmost child) before mutating. */	\
+	if ((parent = phn_prev_get(a_type, a_field, phn)) != NULL) {	\
+		if (phn_lchild_get(a_type, a_field, parent) != phn) {	\
+			parent = NULL;					\
+		}							\
+	}								\
+	/* Find a possible replacement node, and link to parent. */	\
+	ph_merge_children(a_type, a_field, phn, a_cmp, replace);	\
+	/* Set next/prev for sibling linked list. */			\
+	if (replace != NULL) {						\
+		if (parent != NULL) {					\
+			phn_prev_set(a_type, a_field, replace, parent);	\
+			phn_lchild_set(a_type, a_field, parent,		\
+			    replace);					\
+		} else {						\
+			phn_prev_set(a_type, a_field, replace,		\
+			    phn_prev_get(a_type, a_field, phn));	\
+			if (phn_prev_get(a_type, a_field, phn) !=	\
+			    NULL) {					\
+				phn_next_set(a_type, a_field,		\
+				    phn_prev_get(a_type, a_field, phn),	\
+				    replace);				\
+			}						\
+		}							\
+		phn_next_set(a_type, a_field, replace,			\
+		    phn_next_get(a_type, a_field, phn));		\
+		if (phn_next_get(a_type, a_field, phn) != NULL) {	\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, phn),		\
+			    replace);					\
+		}							\
+	} else {							\
+		if (parent != NULL) {					\
+			a_type *next = phn_next_get(a_type, a_field,	\
+			    phn);					\
+			phn_lchild_set(a_type, a_field, parent, next);	\
+			if (next != NULL) {				\
+				phn_prev_set(a_type, a_field, next,	\
+				    parent);				\
+			}						\
+		} else {						\
+			assert(phn_prev_get(a_type, a_field, phn) !=	\
+			    NULL);					\
+			phn_next_set(a_type, a_field,			\
+			    phn_prev_get(a_type, a_field, phn),		\
+			    phn_next_get(a_type, a_field, phn));	\
+		}							\
+		if (phn_next_get(a_type, a_field, phn) != NULL) {	\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, phn),		\
+			    phn_prev_get(a_type, a_field, phn));	\
+		}							\
+	}								\
+}
+
+#endif /* PH_H_ */
diff --git a/jemalloc/include/jemalloc/internal/private_namespace.sh b/jemalloc/include/jemalloc/internal/private_namespace.sh
new file mode 100755
index 0000000..6ef1346
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/private_namespace.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+for symbol in `cat "$@"` ; do
+  echo "#define ${symbol} JEMALLOC_N(${symbol})"
+done
diff --git a/jemalloc/include/jemalloc/internal/private_symbols.sh b/jemalloc/include/jemalloc/internal/private_symbols.sh
new file mode 100755
index 0000000..442a259
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/private_symbols.sh
@@ -0,0 +1,51 @@
+#!/bin/sh
+#
+# Generate private_symbols[_jet].awk.
+#
+# Usage: private_symbols.sh <sym_prefix> <sym>*
+#
+# <sym_prefix> is typically "" or "_".
+
+sym_prefix=$1
+shift
+
+cat <<EOF
+#!/usr/bin/env awk -f
+
+BEGIN {
+  sym_prefix = "${sym_prefix}"
+  split("\\
+EOF
+
+for public_sym in "$@" ; do
+  cat <<EOF
+        ${sym_prefix}${public_sym} \\
+EOF
+done
+
+cat <<"EOF"
+        ", exported_symbol_names)
+  # Store exported symbol names as keys in exported_symbols.
+  for (i in exported_symbol_names) {
+    exported_symbols[exported_symbol_names[i]] = 1
+  }
+}
+
+# Process 'nm -a <c_source.o>' output.
+#
+# Handle lines like:
+#   0000000000000008 D opt_junk
+#   0000000000007574 T malloc_initialized
+(NF == 3 && $2 ~ /^[ABCDGRSTVW]$/ && !($3 in exported_symbols) && $3 ~ /^[A-Za-z0-9_]+$/) {
+  print substr($3, 1+length(sym_prefix), length($3)-length(sym_prefix))
+}
+
+# Process 'dumpbin /SYMBOLS <c_source.obj>' output.
+#
+# Handle lines like:
+#   353 00008098 SECT4  notype       External     | opt_junk
+#   3F1 00000000 SECT7  notype ()    External     | malloc_initialized
+($3 ~ /^SECT[0-9]+/ && $(NF-2) == "External" && !($NF in exported_symbols)) {
+  print $NF
+}
+EOF
diff --git a/jemalloc/include/jemalloc/internal/prng.h b/jemalloc/include/jemalloc/internal/prng.h
new file mode 100644
index 0000000..15cc2d1
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/prng.h
@@ -0,0 +1,185 @@
+#ifndef JEMALLOC_INTERNAL_PRNG_H
+#define JEMALLOC_INTERNAL_PRNG_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
+
+/*
+ * Simple linear congruential pseudo-random number generator:
+ *
+ *   prng(y) = (a*x + c) % m
+ *
+ * where the following constants ensure maximal period:
+ *
+ *   a == Odd number (relatively prime to 2^n), and (a-1) is a multiple of 4.
+ *   c == Odd number (relatively prime to 2^n).
+ *   m == 2^32
+ *
+ * See Knuth's TAOCP 3rd Ed., Vol. 2, pg. 17 for details on these constraints.
+ *
+ * This choice of m has the disadvantage that the quality of the bits is
+ * proportional to bit position.  For example, the lowest bit has a cycle of 2,
+ * the next has a cycle of 4, etc.  For this reason, we prefer to use the upper
+ * bits.
+ */
+
+/******************************************************************************/
+/* INTERNAL DEFINITIONS -- IGNORE */
+/******************************************************************************/
+#define PRNG_A_32	UINT32_C(1103515241)
+#define PRNG_C_32	UINT32_C(12347)
+
+#define PRNG_A_64	UINT64_C(6364136223846793005)
+#define PRNG_C_64	UINT64_C(1442695040888963407)
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_state_next_u32(uint32_t state) {
+	return (state * PRNG_A_32) + PRNG_C_32;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_state_next_u64(uint64_t state) {
+	return (state * PRNG_A_64) + PRNG_C_64;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_state_next_zu(size_t state) {
+#if LG_SIZEOF_PTR == 2
+	return (state * PRNG_A_32) + PRNG_C_32;
+#elif LG_SIZEOF_PTR == 3
+	return (state * PRNG_A_64) + PRNG_C_64;
+#else
+#error Unsupported pointer size
+#endif
+}
+
+/******************************************************************************/
+/* BEGIN PUBLIC API */
+/******************************************************************************/
+
+/*
+ * The prng_lg_range functions give a uniform int in the half-open range [0,
+ * 2**lg_range).  If atomic is true, they do so safely from multiple threads.
+ * Multithreaded 64-bit prngs aren't supported.
+ */
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_lg_range_u32(atomic_u32_t *state, unsigned lg_range, bool atomic) {
+	uint32_t ret, state0, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= 32);
+
+	state0 = atomic_load_u32(state, ATOMIC_RELAXED);
+
+	if (atomic) {
+		do {
+			state1 = prng_state_next_u32(state0);
+		} while (!atomic_compare_exchange_weak_u32(state, &state0,
+		    state1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+	} else {
+		state1 = prng_state_next_u32(state0);
+		atomic_store_u32(state, state1, ATOMIC_RELAXED);
+	}
+	ret = state1 >> (32 - lg_range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_lg_range_u64(uint64_t *state, unsigned lg_range) {
+	uint64_t ret, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= 64);
+
+	state1 = prng_state_next_u64(*state);
+	*state = state1;
+	ret = state1 >> (64 - lg_range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_lg_range_zu(atomic_zu_t *state, unsigned lg_range, bool atomic) {
+	size_t ret, state0, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= ZU(1) << (3 + LG_SIZEOF_PTR));
+
+	state0 = atomic_load_zu(state, ATOMIC_RELAXED);
+
+	if (atomic) {
+		do {
+			state1 = prng_state_next_zu(state0);
+		} while (atomic_compare_exchange_weak_zu(state, &state0,
+		    state1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+	} else {
+		state1 = prng_state_next_zu(state0);
+		atomic_store_zu(state, state1, ATOMIC_RELAXED);
+	}
+	ret = state1 >> ((ZU(1) << (3 + LG_SIZEOF_PTR)) - lg_range);
+
+	return ret;
+}
+
+/*
+ * The prng_range functions behave like the prng_lg_range, but return a result
+ * in [0, range) instead of [0, 2**lg_range).
+ */
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_range_u32(atomic_u32_t *state, uint32_t range, bool atomic) {
+	uint32_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u32(pow2_ceil_u32(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_u32(state, lg_range, atomic);
+	} while (ret >= range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_range_u64(uint64_t *state, uint64_t range) {
+	uint64_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u64(pow2_ceil_u64(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_u64(state, lg_range);
+	} while (ret >= range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_range_zu(atomic_zu_t *state, size_t range, bool atomic) {
+	size_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u64(pow2_ceil_u64(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_zu(state, lg_range, atomic);
+	} while (ret >= range);
+
+	return ret;
+}
+
+#endif /* JEMALLOC_INTERNAL_PRNG_H */
diff --git a/jemalloc/include/jemalloc/internal/prof_externs.h b/jemalloc/include/jemalloc/internal/prof_externs.h
new file mode 100644
index 0000000..094f3e1
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/prof_externs.h
@@ -0,0 +1,105 @@
+#ifndef JEMALLOC_INTERNAL_PROF_EXTERNS_H
+#define JEMALLOC_INTERNAL_PROF_EXTERNS_H
+
+#include "jemalloc/internal/mutex.h"
+
+extern malloc_mutex_t	bt2gctx_mtx;
+
+extern bool	opt_prof;
+extern bool	opt_prof_active;
+extern bool	opt_prof_thread_active_init;
+extern size_t	opt_lg_prof_sample;   /* Mean bytes between samples. */
+extern ssize_t	opt_lg_prof_interval; /* lg(prof_interval). */
+extern bool	opt_prof_gdump;       /* High-water memory dumping. */
+extern bool	opt_prof_final;       /* Final profile dumping. */
+extern bool	opt_prof_leak;        /* Dump leak summary at exit. */
+extern bool	opt_prof_accum;       /* Report cumulative bytes. */
+extern bool	opt_prof_log;	      /* Turn logging on at boot. */
+extern char	opt_prof_prefix[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PATH_MAX +
+#endif
+    1];
+
+/* Accessed via prof_active_[gs]et{_unlocked,}(). */
+extern bool	prof_active;
+
+/* Accessed via prof_gdump_[gs]et{_unlocked,}(). */
+extern bool	prof_gdump_val;
+
+/*
+ * Profile dump interval, measured in bytes allocated.  Each arena triggers a
+ * profile dump when it reaches this threshold.  The effect is that the
+ * interval between profile dumps averages prof_interval, though the actual
+ * interval between dumps will tend to be sporadic, and the interval will be a
+ * maximum of approximately (prof_interval * narenas).
+ */
+extern uint64_t	prof_interval;
+
+/*
+ * Initialized as opt_lg_prof_sample, and potentially modified during profiling
+ * resets.
+ */
+extern size_t	lg_prof_sample;
+
+void prof_alloc_rollback(tsd_t *tsd, prof_tctx_t *tctx, bool updated);
+void prof_malloc_sample_object(tsdn_t *tsdn, const void *ptr, size_t usize,
+    prof_tctx_t *tctx);
+void prof_free_sampled_object(tsd_t *tsd, const void *ptr, size_t usize,
+    prof_tctx_t *tctx);
+void bt_init(prof_bt_t *bt, void **vec);
+void prof_backtrace(prof_bt_t *bt);
+prof_tctx_t *prof_lookup(tsd_t *tsd, prof_bt_t *bt);
+#ifdef JEMALLOC_JET
+size_t prof_tdata_count(void);
+size_t prof_bt_count(void);
+#endif
+typedef int (prof_dump_open_t)(bool, const char *);
+extern prof_dump_open_t *JET_MUTABLE prof_dump_open;
+
+typedef bool (prof_dump_header_t)(tsdn_t *, bool, const prof_cnt_t *);
+extern prof_dump_header_t *JET_MUTABLE prof_dump_header;
+#ifdef JEMALLOC_JET
+void prof_cnt_all(uint64_t *curobjs, uint64_t *curbytes, uint64_t *accumobjs,
+    uint64_t *accumbytes);
+#endif
+bool prof_accum_init(tsdn_t *tsdn, prof_accum_t *prof_accum);
+void prof_idump(tsdn_t *tsdn);
+bool prof_mdump(tsd_t *tsd, const char *filename);
+void prof_gdump(tsdn_t *tsdn);
+prof_tdata_t *prof_tdata_init(tsd_t *tsd);
+prof_tdata_t *prof_tdata_reinit(tsd_t *tsd, prof_tdata_t *tdata);
+void prof_reset(tsd_t *tsd, size_t lg_sample);
+void prof_tdata_cleanup(tsd_t *tsd);
+bool prof_active_get(tsdn_t *tsdn);
+bool prof_active_set(tsdn_t *tsdn, bool active);
+const char *prof_thread_name_get(tsd_t *tsd);
+int prof_thread_name_set(tsd_t *tsd, const char *thread_name);
+bool prof_thread_active_get(tsd_t *tsd);
+bool prof_thread_active_set(tsd_t *tsd, bool active);
+bool prof_thread_active_init_get(tsdn_t *tsdn);
+bool prof_thread_active_init_set(tsdn_t *tsdn, bool active_init);
+bool prof_gdump_get(tsdn_t *tsdn);
+bool prof_gdump_set(tsdn_t *tsdn, bool active);
+void prof_boot0(void);
+void prof_boot1(void);
+bool prof_boot2(tsd_t *tsd);
+void prof_prefork0(tsdn_t *tsdn);
+void prof_prefork1(tsdn_t *tsdn);
+void prof_postfork_parent(tsdn_t *tsdn);
+void prof_postfork_child(tsdn_t *tsdn);
+void prof_sample_threshold_update(prof_tdata_t *tdata);
+
+bool prof_log_start(tsdn_t *tsdn, const char *filename);
+bool prof_log_stop(tsdn_t *tsdn);
+#ifdef JEMALLOC_JET
+size_t prof_log_bt_count(void);
+size_t prof_log_alloc_count(void);
+size_t prof_log_thr_count(void);
+bool prof_log_is_logging(void);
+bool prof_log_rep_check(void);
+void prof_log_dummy_set(bool new_value);
+#endif
+
+#endif /* JEMALLOC_INTERNAL_PROF_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/prof_inlines_a.h b/jemalloc/include/jemalloc/internal/prof_inlines_a.h
new file mode 100644
index 0000000..471d985
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/prof_inlines_a.h
@@ -0,0 +1,85 @@
+#ifndef JEMALLOC_INTERNAL_PROF_INLINES_A_H
+#define JEMALLOC_INTERNAL_PROF_INLINES_A_H
+
+#include "jemalloc/internal/mutex.h"
+
+static inline bool
+prof_accum_add(tsdn_t *tsdn, prof_accum_t *prof_accum,
+    uint64_t accumbytes) {
+	cassert(config_prof);
+
+	bool overflow;
+	uint64_t a0, a1;
+
+	/*
+	 * If the application allocates fast enough (and/or if idump is slow
+	 * enough), extreme overflow here (a1 >= prof_interval * 2) can cause
+	 * idump trigger coalescing.  This is an intentional mechanism that
+	 * avoids rate-limiting allocation.
+	 */
+#ifdef JEMALLOC_ATOMIC_U64
+	a0 = atomic_load_u64(&prof_accum->accumbytes, ATOMIC_RELAXED);
+	do {
+		a1 = a0 + accumbytes;
+		assert(a1 >= a0);
+		overflow = (a1 >= prof_interval);
+		if (overflow) {
+			a1 %= prof_interval;
+		}
+	} while (!atomic_compare_exchange_weak_u64(&prof_accum->accumbytes, &a0,
+	    a1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+#else
+	malloc_mutex_lock(tsdn, &prof_accum->mtx);
+	a0 = prof_accum->accumbytes;
+	a1 = a0 + accumbytes;
+	overflow = (a1 >= prof_interval);
+	if (overflow) {
+		a1 %= prof_interval;
+	}
+	prof_accum->accumbytes = a1;
+	malloc_mutex_unlock(tsdn, &prof_accum->mtx);
+#endif
+	return overflow;
+}
+
+static inline void
+prof_accum_cancel(tsdn_t *tsdn, prof_accum_t *prof_accum,
+    size_t usize) {
+	cassert(config_prof);
+
+	/*
+	 * Cancel out as much of the excessive prof_accumbytes increase as
+	 * possible without underflowing.  Interval-triggered dumps occur
+	 * slightly more often than intended as a result of incomplete
+	 * canceling.
+	 */
+	uint64_t a0, a1;
+#ifdef JEMALLOC_ATOMIC_U64
+	a0 = atomic_load_u64(&prof_accum->accumbytes, ATOMIC_RELAXED);
+	do {
+		a1 = (a0 >= SC_LARGE_MINCLASS - usize)
+		    ? a0 - (SC_LARGE_MINCLASS - usize) : 0;
+	} while (!atomic_compare_exchange_weak_u64(&prof_accum->accumbytes, &a0,
+	    a1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+#else
+	malloc_mutex_lock(tsdn, &prof_accum->mtx);
+	a0 = prof_accum->accumbytes;
+	a1 = (a0 >= SC_LARGE_MINCLASS - usize)
+	    ?  a0 - (SC_LARGE_MINCLASS - usize) : 0;
+	prof_accum->accumbytes = a1;
+	malloc_mutex_unlock(tsdn, &prof_accum->mtx);
+#endif
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_active_get_unlocked(void) {
+	/*
+	 * Even if opt_prof is true, sampling can be temporarily disabled by
+	 * setting prof_active to false.  No locking is used when reading
+	 * prof_active in the fast path, so there are no guarantees regarding
+	 * how long it will take for all threads to notice state changes.
+	 */
+	return prof_active;
+}
+
+#endif /* JEMALLOC_INTERNAL_PROF_INLINES_A_H */
diff --git a/jemalloc/include/jemalloc/internal/prof_inlines_b.h b/jemalloc/include/jemalloc/internal/prof_inlines_b.h
new file mode 100644
index 0000000..8ba8a1e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/prof_inlines_b.h
@@ -0,0 +1,250 @@
+#ifndef JEMALLOC_INTERNAL_PROF_INLINES_B_H
+#define JEMALLOC_INTERNAL_PROF_INLINES_B_H
+
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/sz.h"
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_gdump_get_unlocked(void) {
+	/*
+	 * No locking is used when reading prof_gdump_val in the fast path, so
+	 * there are no guarantees regarding how long it will take for all
+	 * threads to notice state changes.
+	 */
+	return prof_gdump_val;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tdata_t *
+prof_tdata_get(tsd_t *tsd, bool create) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	tdata = tsd_prof_tdata_get(tsd);
+	if (create) {
+		if (unlikely(tdata == NULL)) {
+			if (tsd_nominal(tsd)) {
+				tdata = prof_tdata_init(tsd);
+				tsd_prof_tdata_set(tsd, tdata);
+			}
+		} else if (unlikely(tdata->expired)) {
+			tdata = prof_tdata_reinit(tsd, tdata);
+			tsd_prof_tdata_set(tsd, tdata);
+		}
+		assert(tdata == NULL || tdata->attached);
+	}
+
+	return tdata;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+prof_tctx_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	return arena_prof_tctx_get(tsdn, ptr, alloc_ctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_tctx_set(tsdn_t *tsdn, const void *ptr, size_t usize,
+    alloc_ctx_t *alloc_ctx, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_tctx_set(tsdn, ptr, usize, alloc_ctx, tctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_tctx_reset(tsdn_t *tsdn, const void *ptr, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_tctx_reset(tsdn, ptr, tctx);
+}
+
+JEMALLOC_ALWAYS_INLINE nstime_t
+prof_alloc_time_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	return arena_prof_alloc_time_get(tsdn, ptr, alloc_ctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_alloc_time_set(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx,
+    nstime_t t) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_alloc_time_set(tsdn, ptr, alloc_ctx, t);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_sample_check(tsd_t *tsd, size_t usize, bool update) {
+	ssize_t check = update ? 0 : usize;
+
+	int64_t bytes_until_sample = tsd_bytes_until_sample_get(tsd);
+	if (update) {
+		bytes_until_sample -= usize;
+		if (tsd_nominal(tsd)) {
+			tsd_bytes_until_sample_set(tsd, bytes_until_sample);
+		}
+	}
+	if (likely(bytes_until_sample >= check)) {
+		return true;
+	}
+
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_sample_accum_update(tsd_t *tsd, size_t usize, bool update,
+			 prof_tdata_t **tdata_out) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	/* Fastpath: no need to load tdata */
+	if (likely(prof_sample_check(tsd, usize, update))) {
+		return true;
+	}
+
+	bool booted = tsd_prof_tdata_get(tsd);
+	tdata = prof_tdata_get(tsd, true);
+	if (unlikely((uintptr_t)tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)) {
+		tdata = NULL;
+	}
+
+	if (tdata_out != NULL) {
+		*tdata_out = tdata;
+	}
+
+	if (unlikely(tdata == NULL)) {
+		return true;
+	}
+
+	/*
+	 * If this was the first creation of tdata, then
+	 * prof_tdata_get() reset bytes_until_sample, so decrement and
+	 * check it again
+	 */
+	if (!booted && prof_sample_check(tsd, usize, update)) {
+		return true;
+	}
+
+	if (tsd_reentrancy_level_get(tsd) > 0) {
+		return true;
+	}
+	/* Compute new sample threshold. */
+	if (update) {
+		prof_sample_threshold_update(tdata);
+	}
+	return !tdata->active;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+prof_alloc_prep(tsd_t *tsd, size_t usize, bool prof_active, bool update) {
+	prof_tctx_t *ret;
+	prof_tdata_t *tdata;
+	prof_bt_t bt;
+
+	assert(usize == sz_s2u(usize));
+
+	if (!prof_active || likely(prof_sample_accum_update(tsd, usize, update,
+	    &tdata))) {
+		ret = (prof_tctx_t *)(uintptr_t)1U;
+	} else {
+		bt_init(&bt, tdata->vec);
+		prof_backtrace(&bt);
+		ret = prof_lookup(tsd, &bt);
+	}
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_malloc(tsdn_t *tsdn, const void *ptr, size_t usize, alloc_ctx_t *alloc_ctx,
+    prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(usize == isalloc(tsdn, ptr));
+
+	if (unlikely((uintptr_t)tctx > (uintptr_t)1U)) {
+		prof_malloc_sample_object(tsdn, ptr, usize, tctx);
+	} else {
+		prof_tctx_set(tsdn, ptr, usize, alloc_ctx,
+		    (prof_tctx_t *)(uintptr_t)1U);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_realloc(tsd_t *tsd, const void *ptr, size_t usize, prof_tctx_t *tctx,
+    bool prof_active, bool updated, const void *old_ptr, size_t old_usize,
+    prof_tctx_t *old_tctx) {
+	bool sampled, old_sampled, moved;
+
+	cassert(config_prof);
+	assert(ptr != NULL || (uintptr_t)tctx <= (uintptr_t)1U);
+
+	if (prof_active && !updated && ptr != NULL) {
+		assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+		if (prof_sample_accum_update(tsd, usize, true, NULL)) {
+			/*
+			 * Don't sample.  The usize passed to prof_alloc_prep()
+			 * was larger than what actually got allocated, so a
+			 * backtrace was captured for this allocation, even
+			 * though its actual usize was insufficient to cross the
+			 * sample threshold.
+			 */
+			prof_alloc_rollback(tsd, tctx, true);
+			tctx = (prof_tctx_t *)(uintptr_t)1U;
+		}
+	}
+
+	sampled = ((uintptr_t)tctx > (uintptr_t)1U);
+	old_sampled = ((uintptr_t)old_tctx > (uintptr_t)1U);
+	moved = (ptr != old_ptr);
+
+	if (unlikely(sampled)) {
+		prof_malloc_sample_object(tsd_tsdn(tsd), ptr, usize, tctx);
+	} else if (moved) {
+		prof_tctx_set(tsd_tsdn(tsd), ptr, usize, NULL,
+		    (prof_tctx_t *)(uintptr_t)1U);
+	} else if (unlikely(old_sampled)) {
+		/*
+		 * prof_tctx_set() would work for the !moved case as well, but
+		 * prof_tctx_reset() is slightly cheaper, and the proper thing
+		 * to do here in the presence of explicit knowledge re: moved
+		 * state.
+		 */
+		prof_tctx_reset(tsd_tsdn(tsd), ptr, tctx);
+	} else {
+		assert((uintptr_t)prof_tctx_get(tsd_tsdn(tsd), ptr, NULL) ==
+		    (uintptr_t)1U);
+	}
+
+	/*
+	 * The prof_free_sampled_object() call must come after the
+	 * prof_malloc_sample_object() call, because tctx and old_tctx may be
+	 * the same, in which case reversing the call order could cause the tctx
+	 * to be prematurely destroyed as a side effect of momentarily zeroed
+	 * counters.
+	 */
+	if (unlikely(old_sampled)) {
+		prof_free_sampled_object(tsd, ptr, old_usize, old_tctx);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_free(tsd_t *tsd, const void *ptr, size_t usize, alloc_ctx_t *alloc_ctx) {
+	prof_tctx_t *tctx = prof_tctx_get(tsd_tsdn(tsd), ptr, alloc_ctx);
+
+	cassert(config_prof);
+	assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+
+	if (unlikely((uintptr_t)tctx > (uintptr_t)1U)) {
+		prof_free_sampled_object(tsd, ptr, usize, tctx);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_PROF_INLINES_B_H */
diff --git a/jemalloc/include/jemalloc/internal/prof_structs.h b/jemalloc/include/jemalloc/internal/prof_structs.h
new file mode 100644
index 0000000..34ed482
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/prof_structs.h
@@ -0,0 +1,200 @@
+#ifndef JEMALLOC_INTERNAL_PROF_STRUCTS_H
+#define JEMALLOC_INTERNAL_PROF_STRUCTS_H
+
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/rb.h"
+
+struct prof_bt_s {
+	/* Backtrace, stored as len program counters. */
+	void		**vec;
+	unsigned	len;
+};
+
+#ifdef JEMALLOC_PROF_LIBGCC
+/* Data structure passed to libgcc _Unwind_Backtrace() callback functions. */
+typedef struct {
+	prof_bt_t	*bt;
+	unsigned	max;
+} prof_unwind_data_t;
+#endif
+
+struct prof_accum_s {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_t	mtx;
+	uint64_t	accumbytes;
+#else
+	atomic_u64_t	accumbytes;
+#endif
+};
+
+struct prof_cnt_s {
+	/* Profiling counters. */
+	uint64_t	curobjs;
+	uint64_t	curbytes;
+	uint64_t	accumobjs;
+	uint64_t	accumbytes;
+};
+
+typedef enum {
+	prof_tctx_state_initializing,
+	prof_tctx_state_nominal,
+	prof_tctx_state_dumping,
+	prof_tctx_state_purgatory /* Dumper must finish destroying. */
+} prof_tctx_state_t;
+
+struct prof_tctx_s {
+	/* Thread data for thread that performed the allocation. */
+	prof_tdata_t		*tdata;
+
+	/*
+	 * Copy of tdata->thr_{uid,discrim}, necessary because tdata may be
+	 * defunct during teardown.
+	 */
+	uint64_t		thr_uid;
+	uint64_t		thr_discrim;
+
+	/* Profiling counters, protected by tdata->lock. */
+	prof_cnt_t		cnts;
+
+	/* Associated global context. */
+	prof_gctx_t		*gctx;
+
+	/*
+	 * UID that distinguishes multiple tctx's created by the same thread,
+	 * but coexisting in gctx->tctxs.  There are two ways that such
+	 * coexistence can occur:
+	 * - A dumper thread can cause a tctx to be retained in the purgatory
+	 *   state.
+	 * - Although a single "producer" thread must create all tctx's which
+	 *   share the same thr_uid, multiple "consumers" can each concurrently
+	 *   execute portions of prof_tctx_destroy().  prof_tctx_destroy() only
+	 *   gets called once each time cnts.cur{objs,bytes} drop to 0, but this
+	 *   threshold can be hit again before the first consumer finishes
+	 *   executing prof_tctx_destroy().
+	 */
+	uint64_t		tctx_uid;
+
+	/* Linkage into gctx's tctxs. */
+	rb_node(prof_tctx_t)	tctx_link;
+
+	/*
+	 * True during prof_alloc_prep()..prof_malloc_sample_object(), prevents
+	 * sample vs destroy race.
+	 */
+	bool			prepared;
+
+	/* Current dump-related state, protected by gctx->lock. */
+	prof_tctx_state_t	state;
+
+	/*
+	 * Copy of cnts snapshotted during early dump phase, protected by
+	 * dump_mtx.
+	 */
+	prof_cnt_t		dump_cnts;
+};
+typedef rb_tree(prof_tctx_t) prof_tctx_tree_t;
+
+struct prof_gctx_s {
+	/* Protects nlimbo, cnt_summed, and tctxs. */
+	malloc_mutex_t		*lock;
+
+	/*
+	 * Number of threads that currently cause this gctx to be in a state of
+	 * limbo due to one of:
+	 *   - Initializing this gctx.
+	 *   - Initializing per thread counters associated with this gctx.
+	 *   - Preparing to destroy this gctx.
+	 *   - Dumping a heap profile that includes this gctx.
+	 * nlimbo must be 1 (single destroyer) in order to safely destroy the
+	 * gctx.
+	 */
+	unsigned		nlimbo;
+
+	/*
+	 * Tree of profile counters, one for each thread that has allocated in
+	 * this context.
+	 */
+	prof_tctx_tree_t	tctxs;
+
+	/* Linkage for tree of contexts to be dumped. */
+	rb_node(prof_gctx_t)	dump_link;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* Associated backtrace. */
+	prof_bt_t		bt;
+
+	/* Backtrace vector, variable size, referred to by bt. */
+	void			*vec[1];
+};
+typedef rb_tree(prof_gctx_t) prof_gctx_tree_t;
+
+struct prof_tdata_s {
+	malloc_mutex_t		*lock;
+
+	/* Monotonically increasing unique thread identifier. */
+	uint64_t		thr_uid;
+
+	/*
+	 * Monotonically increasing discriminator among tdata structures
+	 * associated with the same thr_uid.
+	 */
+	uint64_t		thr_discrim;
+
+	/* Included in heap profile dumps if non-NULL. */
+	char			*thread_name;
+
+	bool			attached;
+	bool			expired;
+
+	rb_node(prof_tdata_t)	tdata_link;
+
+	/*
+	 * Counter used to initialize prof_tctx_t's tctx_uid.  No locking is
+	 * necessary when incrementing this field, because only one thread ever
+	 * does so.
+	 */
+	uint64_t		tctx_uid_next;
+
+	/*
+	 * Hash of (prof_bt_t *)-->(prof_tctx_t *).  Each thread tracks
+	 * backtraces for which it has non-zero allocation/deallocation counters
+	 * associated with thread-specific prof_tctx_t objects.  Other threads
+	 * may write to prof_tctx_t contents when freeing associated objects.
+	 */
+	ckh_t			bt2tctx;
+
+	/* Sampling state. */
+	uint64_t		prng_state;
+
+	/* State used to avoid dumping while operating on prof internals. */
+	bool			enq;
+	bool			enq_idump;
+	bool			enq_gdump;
+
+	/*
+	 * Set to true during an early dump phase for tdata's which are
+	 * currently being dumped.  New threads' tdata's have this initialized
+	 * to false so that they aren't accidentally included in later dump
+	 * phases.
+	 */
+	bool			dumping;
+
+	/*
+	 * True if profiling is active for this tdata's thread
+	 * (thread.prof.active mallctl).
+	 */
+	bool			active;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* Backtrace vector, used for calls to prof_backtrace(). */
+	void			*vec[PROF_BT_MAX];
+};
+typedef rb_tree(prof_tdata_t) prof_tdata_tree_t;
+
+#endif /* JEMALLOC_INTERNAL_PROF_STRUCTS_H */
diff --git a/jemalloc/include/jemalloc/internal/prof_types.h b/jemalloc/include/jemalloc/internal/prof_types.h
new file mode 100644
index 0000000..1eff995
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/prof_types.h
@@ -0,0 +1,56 @@
+#ifndef JEMALLOC_INTERNAL_PROF_TYPES_H
+#define JEMALLOC_INTERNAL_PROF_TYPES_H
+
+typedef struct prof_bt_s prof_bt_t;
+typedef struct prof_accum_s prof_accum_t;
+typedef struct prof_cnt_s prof_cnt_t;
+typedef struct prof_tctx_s prof_tctx_t;
+typedef struct prof_gctx_s prof_gctx_t;
+typedef struct prof_tdata_s prof_tdata_t;
+
+/* Option defaults. */
+#ifdef JEMALLOC_PROF
+#  define PROF_PREFIX_DEFAULT		"jeprof"
+#else
+#  define PROF_PREFIX_DEFAULT		""
+#endif
+#define LG_PROF_SAMPLE_DEFAULT		19
+#define LG_PROF_INTERVAL_DEFAULT	-1
+
+/*
+ * Hard limit on stack backtrace depth.  The version of prof_backtrace() that
+ * is based on __builtin_return_address() necessarily has a hard-coded number
+ * of backtrace frame handlers, and should be kept in sync with this setting.
+ */
+#define PROF_BT_MAX			128
+
+/* Initial hash table size. */
+#define PROF_CKH_MINITEMS		64
+
+/* Size of memory buffer to use when writing dump files. */
+#define PROF_DUMP_BUFSIZE		65536
+
+/* Size of stack-allocated buffer used by prof_printf(). */
+#define PROF_PRINTF_BUFSIZE		128
+
+/*
+ * Number of mutexes shared among all gctx's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define PROF_NCTX_LOCKS			1024
+
+/*
+ * Number of mutexes shared among all tdata's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define PROF_NTDATA_LOCKS		256
+
+/*
+ * prof_tdata pointers close to NULL are used to encode state information that
+ * is used for cleaning up during thread shutdown.
+ */
+#define PROF_TDATA_STATE_REINCARNATED	((prof_tdata_t *)(uintptr_t)1)
+#define PROF_TDATA_STATE_PURGATORY	((prof_tdata_t *)(uintptr_t)2)
+#define PROF_TDATA_STATE_MAX		PROF_TDATA_STATE_PURGATORY
+
+#endif /* JEMALLOC_INTERNAL_PROF_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/public_namespace.sh b/jemalloc/include/jemalloc/internal/public_namespace.sh
new file mode 100755
index 0000000..4d415ba
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/public_namespace.sh
@@ -0,0 +1,6 @@
+#!/bin/sh
+
+for nm in `cat $1` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#define je_${n} JEMALLOC_N(${n})"
+done
diff --git a/jemalloc/include/jemalloc/internal/public_unnamespace.sh b/jemalloc/include/jemalloc/internal/public_unnamespace.sh
new file mode 100755
index 0000000..4239d17
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/public_unnamespace.sh
@@ -0,0 +1,6 @@
+#!/bin/sh
+
+for nm in `cat $1` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#undef je_${n}"
+done
diff --git a/jemalloc/include/jemalloc/internal/ql.h b/jemalloc/include/jemalloc/internal/ql.h
new file mode 100644
index 0000000..8029040
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/ql.h
@@ -0,0 +1,88 @@
+#ifndef JEMALLOC_INTERNAL_QL_H
+#define JEMALLOC_INTERNAL_QL_H
+
+#include "jemalloc/internal/qr.h"
+
+/* List definitions. */
+#define ql_head(a_type)							\
+struct {								\
+	a_type *qlh_first;						\
+}
+
+#define ql_head_initializer(a_head) {NULL}
+
+#define ql_elm(a_type)	qr(a_type)
+
+/* List functions. */
+#define ql_new(a_head) do {						\
+	(a_head)->qlh_first = NULL;					\
+} while (0)
+
+#define ql_elm_new(a_elm, a_field) qr_new((a_elm), a_field)
+
+#define ql_first(a_head) ((a_head)->qlh_first)
+
+#define ql_last(a_head, a_field)					\
+	((ql_first(a_head) != NULL)					\
+	    ? qr_prev(ql_first(a_head), a_field) : NULL)
+
+#define ql_next(a_head, a_elm, a_field)					\
+	((ql_last(a_head, a_field) != (a_elm))				\
+	    ? qr_next((a_elm), a_field)	: NULL)
+
+#define ql_prev(a_head, a_elm, a_field)					\
+	((ql_first(a_head) != (a_elm)) ? qr_prev((a_elm), a_field)	\
+				       : NULL)
+
+#define ql_before_insert(a_head, a_qlelm, a_elm, a_field) do {		\
+	qr_before_insert((a_qlelm), (a_elm), a_field);			\
+	if (ql_first(a_head) == (a_qlelm)) {				\
+		ql_first(a_head) = (a_elm);				\
+	}								\
+} while (0)
+
+#define ql_after_insert(a_qlelm, a_elm, a_field)			\
+	qr_after_insert((a_qlelm), (a_elm), a_field)
+
+#define ql_head_insert(a_head, a_elm, a_field) do {			\
+	if (ql_first(a_head) != NULL) {					\
+		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
+	}								\
+	ql_first(a_head) = (a_elm);					\
+} while (0)
+
+#define ql_tail_insert(a_head, a_elm, a_field) do {			\
+	if (ql_first(a_head) != NULL) {					\
+		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
+	}								\
+	ql_first(a_head) = qr_next((a_elm), a_field);			\
+} while (0)
+
+#define ql_remove(a_head, a_elm, a_field) do {				\
+	if (ql_first(a_head) == (a_elm)) {				\
+		ql_first(a_head) = qr_next(ql_first(a_head), a_field);	\
+	}								\
+	if (ql_first(a_head) != (a_elm)) {				\
+		qr_remove((a_elm), a_field);				\
+	} else {							\
+		ql_first(a_head) = NULL;				\
+	}								\
+} while (0)
+
+#define ql_head_remove(a_head, a_type, a_field) do {			\
+	a_type *t = ql_first(a_head);					\
+	ql_remove((a_head), t, a_field);				\
+} while (0)
+
+#define ql_tail_remove(a_head, a_type, a_field) do {			\
+	a_type *t = ql_last(a_head, a_field);				\
+	ql_remove((a_head), t, a_field);				\
+} while (0)
+
+#define ql_foreach(a_var, a_head, a_field)				\
+	qr_foreach((a_var), ql_first(a_head), a_field)
+
+#define ql_reverse_foreach(a_var, a_head, a_field)			\
+	qr_reverse_foreach((a_var), ql_first(a_head), a_field)
+
+#endif /* JEMALLOC_INTERNAL_QL_H */
diff --git a/jemalloc/include/jemalloc/internal/qr.h b/jemalloc/include/jemalloc/internal/qr.h
new file mode 100644
index 0000000..1e1056b
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/qr.h
@@ -0,0 +1,72 @@
+#ifndef JEMALLOC_INTERNAL_QR_H
+#define JEMALLOC_INTERNAL_QR_H
+
+/* Ring definitions. */
+#define qr(a_type)							\
+struct {								\
+	a_type	*qre_next;						\
+	a_type	*qre_prev;						\
+}
+
+/* Ring functions. */
+#define qr_new(a_qr, a_field) do {					\
+	(a_qr)->a_field.qre_next = (a_qr);				\
+	(a_qr)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_next(a_qr, a_field) ((a_qr)->a_field.qre_next)
+
+#define qr_prev(a_qr, a_field) ((a_qr)->a_field.qre_prev)
+
+#define qr_before_insert(a_qrelm, a_qr, a_field) do {			\
+	(a_qr)->a_field.qre_prev = (a_qrelm)->a_field.qre_prev;		\
+	(a_qr)->a_field.qre_next = (a_qrelm);				\
+	(a_qr)->a_field.qre_prev->a_field.qre_next = (a_qr);		\
+	(a_qrelm)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_after_insert(a_qrelm, a_qr, a_field) do {			\
+	(a_qr)->a_field.qre_next = (a_qrelm)->a_field.qre_next;		\
+	(a_qr)->a_field.qre_prev = (a_qrelm);				\
+	(a_qr)->a_field.qre_next->a_field.qre_prev = (a_qr);		\
+	(a_qrelm)->a_field.qre_next = (a_qr);				\
+} while (0)
+
+#define qr_meld(a_qr_a, a_qr_b, a_type, a_field) do {			\
+	a_type *t;							\
+	(a_qr_a)->a_field.qre_prev->a_field.qre_next = (a_qr_b);	\
+	(a_qr_b)->a_field.qre_prev->a_field.qre_next = (a_qr_a);	\
+	t = (a_qr_a)->a_field.qre_prev;					\
+	(a_qr_a)->a_field.qre_prev = (a_qr_b)->a_field.qre_prev;	\
+	(a_qr_b)->a_field.qre_prev = t;					\
+} while (0)
+
+/*
+ * qr_meld() and qr_split() are functionally equivalent, so there's no need to
+ * have two copies of the code.
+ */
+#define qr_split(a_qr_a, a_qr_b, a_type, a_field)			\
+	qr_meld((a_qr_a), (a_qr_b), a_type, a_field)
+
+#define qr_remove(a_qr, a_field) do {					\
+	(a_qr)->a_field.qre_prev->a_field.qre_next			\
+	    = (a_qr)->a_field.qre_next;					\
+	(a_qr)->a_field.qre_next->a_field.qre_prev			\
+	    = (a_qr)->a_field.qre_prev;					\
+	(a_qr)->a_field.qre_next = (a_qr);				\
+	(a_qr)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_foreach(var, a_qr, a_field)					\
+	for ((var) = (a_qr);						\
+	    (var) != NULL;						\
+	    (var) = (((var)->a_field.qre_next != (a_qr))		\
+	    ? (var)->a_field.qre_next : NULL))
+
+#define qr_reverse_foreach(var, a_qr, a_field)				\
+	for ((var) = ((a_qr) != NULL) ? qr_prev(a_qr, a_field) : NULL;	\
+	    (var) != NULL;						\
+	    (var) = (((var) != (a_qr))					\
+	    ? (var)->a_field.qre_prev : NULL))
+
+#endif /* JEMALLOC_INTERNAL_QR_H */
diff --git a/jemalloc/include/jemalloc/internal/quantum.h b/jemalloc/include/jemalloc/internal/quantum.h
new file mode 100644
index 0000000..821086e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/quantum.h
@@ -0,0 +1,77 @@
+#ifndef JEMALLOC_INTERNAL_QUANTUM_H
+#define JEMALLOC_INTERNAL_QUANTUM_H
+
+/*
+ * Minimum allocation alignment is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#ifndef LG_QUANTUM
+#  if (defined(__i386__) || defined(_M_IX86))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __ia64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __alpha__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__sparc64__) || defined(__sparcv9) || defined(__sparc_v9__))
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__amd64__) || defined(__x86_64__) || defined(_M_X64))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __arm__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __aarch64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __hppa__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __m68k__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __mips__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __nios2__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __or1k__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __powerpc__
+#    define LG_QUANTUM		4
+#  endif
+#  if defined(__riscv) || defined(__riscv__)
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __s390__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined (__SH3E__) || defined(__SH4_SINGLE__) || defined(__SH4__) || \
+	defined(__SH4_SINGLE_ONLY__))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __tile__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __le32__
+#    define LG_QUANTUM		4
+#  endif
+#  ifndef LG_QUANTUM
+#    error "Unknown minimum alignment for architecture; specify via "
+	 "--with-lg-quantum"
+#  endif
+#endif
+
+#define QUANTUM			((size_t)(1U << LG_QUANTUM))
+#define QUANTUM_MASK		(QUANTUM - 1)
+
+/* Return the smallest quantum multiple that is >= a. */
+#define QUANTUM_CEILING(a)						\
+	(((a) + QUANTUM_MASK) & ~QUANTUM_MASK)
+
+#endif /* JEMALLOC_INTERNAL_QUANTUM_H */
diff --git a/jemalloc/include/jemalloc/internal/rb.h b/jemalloc/include/jemalloc/internal/rb.h
new file mode 100644
index 0000000..47fa5ca
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/rb.h
@@ -0,0 +1,1006 @@
+/*-
+ *******************************************************************************
+ *
+ * cpp macro implementation of left-leaning 2-3 red-black trees.  Parent
+ * pointers are not used, and color bits are stored in the least significant
+ * bit of right-child pointers (if RB_COMPACT is defined), thus making node
+ * linkage as compact as is possible for red-black trees.
+ *
+ * Usage:
+ *
+ *   #include <stdint.h>
+ *   #include <stdbool.h>
+ *   #define NDEBUG // (Optional, see assert(3).)
+ *   #include <assert.h>
+ *   #define RB_COMPACT // (Optional, embed color bits in right-child pointers.)
+ *   #include <rb.h>
+ *   ...
+ *
+ *******************************************************************************
+ */
+
+#ifndef RB_H_
+#define RB_H_
+
+#ifndef __PGI
+#define RB_COMPACT
+#endif
+
+#ifdef RB_COMPACT
+/* Node structure. */
+#define rb_node(a_type)							\
+struct {								\
+    a_type *rbn_left;							\
+    a_type *rbn_right_red;						\
+}
+#else
+#define rb_node(a_type)							\
+struct {								\
+    a_type *rbn_left;							\
+    a_type *rbn_right;							\
+    bool rbn_red;							\
+}
+#endif
+
+/* Root structure. */
+#define rb_tree(a_type)							\
+struct {								\
+    a_type *rbt_root;							\
+}
+
+/* Left accessors. */
+#define rbtn_left_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_left)
+#define rbtn_left_set(a_type, a_field, a_node, a_left) do {		\
+    (a_node)->a_field.rbn_left = a_left;				\
+} while (0)
+
+#ifdef RB_COMPACT
+/* Right accessors. */
+#define rbtn_right_get(a_type, a_field, a_node)				\
+    ((a_type *) (((intptr_t) (a_node)->a_field.rbn_right_red)		\
+      & ((ssize_t)-2)))
+#define rbtn_right_set(a_type, a_field, a_node, a_right) do {		\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((uintptr_t) a_right)	\
+      | (((uintptr_t) (a_node)->a_field.rbn_right_red) & ((size_t)1)));	\
+} while (0)
+
+/* Color accessors. */
+#define rbtn_red_get(a_type, a_field, a_node)				\
+    ((bool) (((uintptr_t) (a_node)->a_field.rbn_right_red)		\
+      & ((size_t)1)))
+#define rbtn_color_set(a_type, a_field, a_node, a_red) do {		\
+    (a_node)->a_field.rbn_right_red = (a_type *) ((((intptr_t)		\
+      (a_node)->a_field.rbn_right_red) & ((ssize_t)-2))			\
+      | ((ssize_t)a_red));						\
+} while (0)
+#define rbtn_red_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((uintptr_t)		\
+      (a_node)->a_field.rbn_right_red) | ((size_t)1));			\
+} while (0)
+#define rbtn_black_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((intptr_t)		\
+      (a_node)->a_field.rbn_right_red) & ((ssize_t)-2));		\
+} while (0)
+
+/* Node initializer. */
+#define rbt_node_new(a_type, a_field, a_rbt, a_node) do {		\
+    /* Bookkeeping bit cannot be used by node pointer. */		\
+    assert(((uintptr_t)(a_node) & 0x1) == 0);				\
+    rbtn_left_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_right_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_red_set(a_type, a_field, (a_node));				\
+} while (0)
+#else
+/* Right accessors. */
+#define rbtn_right_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_right)
+#define rbtn_right_set(a_type, a_field, a_node, a_right) do {		\
+    (a_node)->a_field.rbn_right = a_right;				\
+} while (0)
+
+/* Color accessors. */
+#define rbtn_red_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_red)
+#define rbtn_color_set(a_type, a_field, a_node, a_red) do {		\
+    (a_node)->a_field.rbn_red = (a_red);				\
+} while (0)
+#define rbtn_red_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_red = true;					\
+} while (0)
+#define rbtn_black_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_red = false;					\
+} while (0)
+
+/* Node initializer. */
+#define rbt_node_new(a_type, a_field, a_rbt, a_node) do {		\
+    rbtn_left_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_right_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_red_set(a_type, a_field, (a_node));				\
+} while (0)
+#endif
+
+/* Tree initializer. */
+#define rb_new(a_type, a_field, a_rbt) do {				\
+    (a_rbt)->rbt_root = NULL;						\
+} while (0)
+
+/* Internal utility macros. */
+#define rbtn_first(a_type, a_field, a_rbt, a_root, r_node) do {		\
+    (r_node) = (a_root);						\
+    if ((r_node) != NULL) {						\
+	for (;								\
+	  rbtn_left_get(a_type, a_field, (r_node)) != NULL;		\
+	  (r_node) = rbtn_left_get(a_type, a_field, (r_node))) {	\
+	}								\
+    }									\
+} while (0)
+
+#define rbtn_last(a_type, a_field, a_rbt, a_root, r_node) do {		\
+    (r_node) = (a_root);						\
+    if ((r_node) != NULL) {						\
+	for (; rbtn_right_get(a_type, a_field, (r_node)) != NULL;	\
+	  (r_node) = rbtn_right_get(a_type, a_field, (r_node))) {	\
+	}								\
+    }									\
+} while (0)
+
+#define rbtn_rotate_left(a_type, a_field, a_node, r_node) do {		\
+    (r_node) = rbtn_right_get(a_type, a_field, (a_node));		\
+    rbtn_right_set(a_type, a_field, (a_node),				\
+      rbtn_left_get(a_type, a_field, (r_node)));			\
+    rbtn_left_set(a_type, a_field, (r_node), (a_node));			\
+} while (0)
+
+#define rbtn_rotate_right(a_type, a_field, a_node, r_node) do {		\
+    (r_node) = rbtn_left_get(a_type, a_field, (a_node));		\
+    rbtn_left_set(a_type, a_field, (a_node),				\
+      rbtn_right_get(a_type, a_field, (r_node)));			\
+    rbtn_right_set(a_type, a_field, (r_node), (a_node));		\
+} while (0)
+
+/*
+ * The rb_proto() macro generates function prototypes that correspond to the
+ * functions generated by an equivalently parameterized call to rb_gen().
+ */
+
+#define rb_proto(a_attr, a_prefix, a_rbt_type, a_type)			\
+a_attr void								\
+a_prefix##new(a_rbt_type *rbtree);					\
+a_attr bool								\
+a_prefix##empty(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##first(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##last(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##next(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##prev(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##search(a_rbt_type *rbtree, const a_type *key);		\
+a_attr a_type *								\
+a_prefix##nsearch(a_rbt_type *rbtree, const a_type *key);		\
+a_attr a_type *								\
+a_prefix##psearch(a_rbt_type *rbtree, const a_type *key);		\
+a_attr void								\
+a_prefix##insert(a_rbt_type *rbtree, a_type *node);			\
+a_attr void								\
+a_prefix##remove(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##iter(a_rbt_type *rbtree, a_type *start, a_type *(*cb)(	\
+  a_rbt_type *, a_type *, void *), void *arg);				\
+a_attr a_type *								\
+a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg);		\
+a_attr void								\
+a_prefix##destroy(a_rbt_type *rbtree, void (*cb)(a_type *, void *),	\
+  void *arg);
+
+/*
+ * The rb_gen() macro generates a type-specific red-black tree implementation,
+ * based on the above cpp macros.
+ *
+ * Arguments:
+ *
+ *   a_attr    : Function attribute for generated functions (ex: static).
+ *   a_prefix  : Prefix for generated functions (ex: ex_).
+ *   a_rb_type : Type for red-black tree data structure (ex: ex_t).
+ *   a_type    : Type for red-black tree node data structure (ex: ex_node_t).
+ *   a_field   : Name of red-black tree node linkage (ex: ex_link).
+ *   a_cmp     : Node comparison function name, with the following prototype:
+ *                 int (a_cmp *)(a_type *a_node, a_type *a_other);
+ *                                       ^^^^^^
+ *                                    or a_key
+ *               Interpretation of comparison function return values:
+ *                 -1 : a_node <  a_other
+ *                  0 : a_node == a_other
+ *                  1 : a_node >  a_other
+ *               In all cases, the a_node or a_key macro argument is the first
+ *               argument to the comparison function, which makes it possible
+ *               to write comparison functions that treat the first argument
+ *               specially.
+ *
+ * Assuming the following setup:
+ *
+ *   typedef struct ex_node_s ex_node_t;
+ *   struct ex_node_s {
+ *       rb_node(ex_node_t) ex_link;
+ *   };
+ *   typedef rb_tree(ex_node_t) ex_t;
+ *   rb_gen(static, ex_, ex_t, ex_node_t, ex_link, ex_cmp)
+ *
+ * The following API is generated:
+ *
+ *   static void
+ *   ex_new(ex_t *tree);
+ *       Description: Initialize a red-black tree structure.
+ *       Args:
+ *         tree: Pointer to an uninitialized red-black tree object.
+ *
+ *   static bool
+ *   ex_empty(ex_t *tree);
+ *       Description: Determine whether tree is empty.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *       Ret: True if tree is empty, false otherwise.
+ *
+ *   static ex_node_t *
+ *   ex_first(ex_t *tree);
+ *   static ex_node_t *
+ *   ex_last(ex_t *tree);
+ *       Description: Get the first/last node in tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *       Ret: First/last node in tree, or NULL if tree is empty.
+ *
+ *   static ex_node_t *
+ *   ex_next(ex_t *tree, ex_node_t *node);
+ *   static ex_node_t *
+ *   ex_prev(ex_t *tree, ex_node_t *node);
+ *       Description: Get node's successor/predecessor.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: A node in tree.
+ *       Ret: node's successor/predecessor in tree, or NULL if node is
+ *            last/first.
+ *
+ *   static ex_node_t *
+ *   ex_search(ex_t *tree, const ex_node_t *key);
+ *       Description: Search for node that matches key.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or NULL if no match.
+ *
+ *   static ex_node_t *
+ *   ex_nsearch(ex_t *tree, const ex_node_t *key);
+ *   static ex_node_t *
+ *   ex_psearch(ex_t *tree, const ex_node_t *key);
+ *       Description: Search for node that matches key.  If no match is found,
+ *                    return what would be key's successor/predecessor, were
+ *                    key in tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or if no match, hypothetical node's
+ *            successor/predecessor (NULL if no successor/predecessor).
+ *
+ *   static void
+ *   ex_insert(ex_t *tree, ex_node_t *node);
+ *       Description: Insert node into tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node to be inserted into tree.
+ *
+ *   static void
+ *   ex_remove(ex_t *tree, ex_node_t *node);
+ *       Description: Remove node from tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node in tree to be removed.
+ *
+ *   static ex_node_t *
+ *   ex_iter(ex_t *tree, ex_node_t *start, ex_node_t *(*cb)(ex_t *,
+ *     ex_node_t *, void *), void *arg);
+ *   static ex_node_t *
+ *   ex_reverse_iter(ex_t *tree, ex_node_t *start, ex_node *(*cb)(ex_t *,
+ *     ex_node_t *, void *), void *arg);
+ *       Description: Iterate forward/backward over tree, starting at node.  If
+ *                    tree is modified, iteration must be immediately
+ *                    terminated by the callback function that causes the
+ *                    modification.
+ *       Args:
+ *         tree : Pointer to an initialized red-black tree object.
+ *         start: Node at which to start iteration, or NULL to start at
+ *                first/last node.
+ *         cb   : Callback function, which is called for each node during
+ *                iteration.  Under normal circumstances the callback function
+ *                should return NULL, which causes iteration to continue.  If a
+ *                callback function returns non-NULL, iteration is immediately
+ *                terminated and the non-NULL return value is returned by the
+ *                iterator.  This is useful for re-starting iteration after
+ *                modifying tree.
+ *         arg  : Opaque pointer passed to cb().
+ *       Ret: NULL if iteration completed, or the non-NULL callback return value
+ *            that caused termination of the iteration.
+ *
+ *   static void
+ *   ex_destroy(ex_t *tree, void (*cb)(ex_node_t *, void *), void *arg);
+ *       Description: Iterate over the tree with post-order traversal, remove
+ *                    each node, and run the callback if non-null.  This is
+ *                    used for destroying a tree without paying the cost to
+ *                    rebalance it.  The tree must not be otherwise altered
+ *                    during traversal.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         cb  : Callback function, which, if non-null, is called for each node
+ *               during iteration.  There is no way to stop iteration once it
+ *               has begun.
+ *         arg : Opaque pointer passed to cb().
+ */
+#define rb_gen(a_attr, a_prefix, a_rbt_type, a_type, a_field, a_cmp)	\
+a_attr void								\
+a_prefix##new(a_rbt_type *rbtree) {					\
+    rb_new(a_type, a_field, rbtree);					\
+}									\
+a_attr bool								\
+a_prefix##empty(a_rbt_type *rbtree) {					\
+    return (rbtree->rbt_root == NULL);					\
+}									\
+a_attr a_type *								\
+a_prefix##first(a_rbt_type *rbtree) {					\
+    a_type *ret;							\
+    rbtn_first(a_type, a_field, rbtree, rbtree->rbt_root, ret);		\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##last(a_rbt_type *rbtree) {					\
+    a_type *ret;							\
+    rbtn_last(a_type, a_field, rbtree, rbtree->rbt_root, ret);		\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##next(a_rbt_type *rbtree, a_type *node) {			\
+    a_type *ret;							\
+    if (rbtn_right_get(a_type, a_field, node) != NULL) {		\
+	rbtn_first(a_type, a_field, rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), ret);						\
+    } else {								\
+	a_type *tnode = rbtree->rbt_root;				\
+	assert(tnode != NULL);						\
+	ret = NULL;							\
+	while (true) {							\
+	    int cmp = (a_cmp)(node, tnode);				\
+	    if (cmp < 0) {						\
+		ret = tnode;						\
+		tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	    } else if (cmp > 0) {					\
+		tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	    } else {							\
+		break;							\
+	    }								\
+	    assert(tnode != NULL);					\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##prev(a_rbt_type *rbtree, a_type *node) {			\
+    a_type *ret;							\
+    if (rbtn_left_get(a_type, a_field, node) != NULL) {			\
+	rbtn_last(a_type, a_field, rbtree, rbtn_left_get(a_type,	\
+	  a_field, node), ret);						\
+    } else {								\
+	a_type *tnode = rbtree->rbt_root;				\
+	assert(tnode != NULL);						\
+	ret = NULL;							\
+	while (true) {							\
+	    int cmp = (a_cmp)(node, tnode);				\
+	    if (cmp < 0) {						\
+		tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	    } else if (cmp > 0) {					\
+		ret = tnode;						\
+		tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	    } else {							\
+		break;							\
+	    }								\
+	    assert(tnode != NULL);					\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##search(a_rbt_type *rbtree, const a_type *key) {		\
+    a_type *ret;							\
+    int cmp;								\
+    ret = rbtree->rbt_root;						\
+    while (ret != NULL							\
+      && (cmp = (a_cmp)(key, ret)) != 0) {				\
+	if (cmp < 0) {							\
+	    ret = rbtn_left_get(a_type, a_field, ret);			\
+	} else {							\
+	    ret = rbtn_right_get(a_type, a_field, ret);			\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##nsearch(a_rbt_type *rbtree, const a_type *key) {		\
+    a_type *ret;							\
+    a_type *tnode = rbtree->rbt_root;					\
+    ret = NULL;								\
+    while (tnode != NULL) {						\
+	int cmp = (a_cmp)(key, tnode);					\
+	if (cmp < 0) {							\
+	    ret = tnode;						\
+	    tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	} else if (cmp > 0) {						\
+	    tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	} else {							\
+	    ret = tnode;						\
+	    break;							\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##psearch(a_rbt_type *rbtree, const a_type *key) {		\
+    a_type *ret;							\
+    a_type *tnode = rbtree->rbt_root;					\
+    ret = NULL;								\
+    while (tnode != NULL) {						\
+	int cmp = (a_cmp)(key, tnode);					\
+	if (cmp < 0) {							\
+	    tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	} else if (cmp > 0) {						\
+	    ret = tnode;						\
+	    tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	} else {							\
+	    ret = tnode;						\
+	    break;							\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr void								\
+a_prefix##insert(a_rbt_type *rbtree, a_type *node) {			\
+    struct {								\
+	a_type *node;							\
+	int cmp;							\
+    } path[sizeof(void *) << 4], *pathp;				\
+    rbt_node_new(a_type, a_field, rbtree, node);			\
+    /* Wind. */								\
+    path->node = rbtree->rbt_root;					\
+    for (pathp = path; pathp->node != NULL; pathp++) {			\
+	int cmp = pathp->cmp = a_cmp(node, pathp->node);		\
+	assert(cmp != 0);						\
+	if (cmp < 0) {							\
+	    pathp[1].node = rbtn_left_get(a_type, a_field,		\
+	      pathp->node);						\
+	} else {							\
+	    pathp[1].node = rbtn_right_get(a_type, a_field,		\
+	      pathp->node);						\
+	}								\
+    }									\
+    pathp->node = node;							\
+    /* Unwind. */							\
+    for (pathp--; (uintptr_t)pathp >= (uintptr_t)path; pathp--) {	\
+	a_type *cnode = pathp->node;					\
+	if (pathp->cmp < 0) {						\
+	    a_type *left = pathp[1].node;				\
+	    rbtn_left_set(a_type, a_field, cnode, left);		\
+	    if (rbtn_red_get(a_type, a_field, left)) {			\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (leftleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  leftleft)) {						\
+		    /* Fix up 4-node. */				\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, cnode, tnode);	\
+		    cnode = tnode;					\
+		}							\
+	    } else {							\
+		return;							\
+	    }								\
+	} else {							\
+	    a_type *right = pathp[1].node;				\
+	    rbtn_right_set(a_type, a_field, cnode, right);		\
+	    if (rbtn_red_get(a_type, a_field, right)) {			\
+		a_type *left = rbtn_left_get(a_type, a_field, cnode);	\
+		if (left != NULL && rbtn_red_get(a_type, a_field,	\
+		  left)) {						\
+		    /* Split 4-node. */					\
+		    rbtn_black_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, right);		\
+		    rbtn_red_set(a_type, a_field, cnode);		\
+		} else {						\
+		    /* Lean left. */					\
+		    a_type *tnode;					\
+		    bool tred = rbtn_red_get(a_type, a_field, cnode);	\
+		    rbtn_rotate_left(a_type, a_field, cnode, tnode);	\
+		    rbtn_color_set(a_type, a_field, tnode, tred);	\
+		    rbtn_red_set(a_type, a_field, cnode);		\
+		    cnode = tnode;					\
+		}							\
+	    } else {							\
+		return;							\
+	    }								\
+	}								\
+	pathp->node = cnode;						\
+    }									\
+    /* Set root, and make it black. */					\
+    rbtree->rbt_root = path->node;					\
+    rbtn_black_set(a_type, a_field, rbtree->rbt_root);			\
+}									\
+a_attr void								\
+a_prefix##remove(a_rbt_type *rbtree, a_type *node) {			\
+    struct {								\
+	a_type *node;							\
+	int cmp;							\
+    } *pathp, *nodep, path[sizeof(void *) << 4];			\
+    /* Wind. */								\
+    nodep = NULL; /* Silence compiler warning. */			\
+    path->node = rbtree->rbt_root;					\
+    for (pathp = path; pathp->node != NULL; pathp++) {			\
+	int cmp = pathp->cmp = a_cmp(node, pathp->node);		\
+	if (cmp < 0) {							\
+	    pathp[1].node = rbtn_left_get(a_type, a_field,		\
+	      pathp->node);						\
+	} else {							\
+	    pathp[1].node = rbtn_right_get(a_type, a_field,		\
+	      pathp->node);						\
+	    if (cmp == 0) {						\
+	        /* Find node's successor, in preparation for swap. */	\
+		pathp->cmp = 1;						\
+		nodep = pathp;						\
+		for (pathp++; pathp->node != NULL; pathp++) {		\
+		    pathp->cmp = -1;					\
+		    pathp[1].node = rbtn_left_get(a_type, a_field,	\
+		      pathp->node);					\
+		}							\
+		break;							\
+	    }								\
+	}								\
+    }									\
+    assert(nodep->node == node);					\
+    pathp--;								\
+    if (pathp->node != node) {						\
+	/* Swap node with its successor. */				\
+	bool tred = rbtn_red_get(a_type, a_field, pathp->node);		\
+	rbtn_color_set(a_type, a_field, pathp->node,			\
+	  rbtn_red_get(a_type, a_field, node));				\
+	rbtn_left_set(a_type, a_field, pathp->node,			\
+	  rbtn_left_get(a_type, a_field, node));			\
+	/* If node's successor is its right child, the following code */\
+	/* will do the wrong thing for the right child pointer.       */\
+	/* However, it doesn't matter, because the pointer will be    */\
+	/* properly set when the successor is pruned.                 */\
+	rbtn_right_set(a_type, a_field, pathp->node,			\
+	  rbtn_right_get(a_type, a_field, node));			\
+	rbtn_color_set(a_type, a_field, node, tred);			\
+	/* The pruned leaf node's child pointers are never accessed   */\
+	/* again, so don't bother setting them to nil.                */\
+	nodep->node = pathp->node;					\
+	pathp->node = node;						\
+	if (nodep == path) {						\
+	    rbtree->rbt_root = nodep->node;				\
+	} else {							\
+	    if (nodep[-1].cmp < 0) {					\
+		rbtn_left_set(a_type, a_field, nodep[-1].node,		\
+		  nodep->node);						\
+	    } else {							\
+		rbtn_right_set(a_type, a_field, nodep[-1].node,		\
+		  nodep->node);						\
+	    }								\
+	}								\
+    } else {								\
+	a_type *left = rbtn_left_get(a_type, a_field, node);		\
+	if (left != NULL) {						\
+	    /* node has no successor, but it has a left child.        */\
+	    /* Splice node out, without losing the left child.        */\
+	    assert(!rbtn_red_get(a_type, a_field, node));		\
+	    assert(rbtn_red_get(a_type, a_field, left));		\
+	    rbtn_black_set(a_type, a_field, left);			\
+	    if (pathp == path) {					\
+		rbtree->rbt_root = left;				\
+	    } else {							\
+		if (pathp[-1].cmp < 0) {				\
+		    rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+		      left);						\
+		} else {						\
+		    rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+		      left);						\
+		}							\
+	    }								\
+	    return;							\
+	} else if (pathp == path) {					\
+	    /* The tree only contained one node. */			\
+	    rbtree->rbt_root = NULL;					\
+	    return;							\
+	}								\
+    }									\
+    if (rbtn_red_get(a_type, a_field, pathp->node)) {			\
+	/* Prune red node, which requires no fixup. */			\
+	assert(pathp[-1].cmp < 0);					\
+	rbtn_left_set(a_type, a_field, pathp[-1].node, NULL);		\
+	return;								\
+    }									\
+    /* The node to be pruned is black, so unwind until balance is     */\
+    /* restored.                                                      */\
+    pathp->node = NULL;							\
+    for (pathp--; (uintptr_t)pathp >= (uintptr_t)path; pathp--) {	\
+	assert(pathp->cmp != 0);					\
+	if (pathp->cmp < 0) {						\
+	    rbtn_left_set(a_type, a_field, pathp->node,			\
+	      pathp[1].node);						\
+	    if (rbtn_red_get(a_type, a_field, pathp->node)) {		\
+		a_type *right = rbtn_right_get(a_type, a_field,		\
+		  pathp->node);						\
+		a_type *rightleft = rbtn_left_get(a_type, a_field,	\
+		  right);						\
+		a_type *tnode;						\
+		if (rightleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  rightleft)) {						\
+		    /* In the following diagrams, ||, //, and \\      */\
+		    /* indicate the path to the removed node.         */\
+		    /*                                                */\
+		    /*      ||                                        */\
+		    /*    pathp(r)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (r)                                   */\
+		    /*                                                */\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    rbtn_rotate_right(a_type, a_field, right, tnode);	\
+		    rbtn_right_set(a_type, a_field, pathp->node, tnode);\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(r)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (b)                                   */\
+		    /*                                                */\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		}							\
+		/* Balance restored, but rotation modified subtree    */\
+		/* root.                                              */\
+		assert((uintptr_t)pathp > (uintptr_t)path);		\
+		if (pathp[-1].cmp < 0) {				\
+		    rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+		      tnode);						\
+		} else {						\
+		    rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+		      tnode);						\
+		}							\
+		return;							\
+	    } else {							\
+		a_type *right = rbtn_right_get(a_type, a_field,		\
+		  pathp->node);						\
+		a_type *rightleft = rbtn_left_get(a_type, a_field,	\
+		  right);						\
+		if (rightleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  rightleft)) {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (r)                                   */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, rightleft);		\
+		    rbtn_rotate_right(a_type, a_field, right, tnode);	\
+		    rbtn_right_set(a_type, a_field, pathp->node, tnode);\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root, which may actually be the tree   */\
+		    /* root.                                          */\
+		    if (pathp == path) {				\
+			/* Set root. */					\
+			rbtree->rbt_root = tnode;			\
+		    } else {						\
+			if (pathp[-1].cmp < 0) {			\
+			    rbtn_left_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			} else {					\
+			    rbtn_right_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			}						\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (b)                                   */\
+		    a_type *tnode;					\
+		    rbtn_red_set(a_type, a_field, pathp->node);		\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    pathp->node = tnode;				\
+		}							\
+	    }								\
+	} else {							\
+	    a_type *left;						\
+	    rbtn_right_set(a_type, a_field, pathp->node,		\
+	      pathp[1].node);						\
+	    left = rbtn_left_get(a_type, a_field, pathp->node);		\
+	    if (rbtn_red_get(a_type, a_field, left)) {			\
+		a_type *tnode;						\
+		a_type *leftright = rbtn_right_get(a_type, a_field,	\
+		  left);						\
+		a_type *leftrightleft = rbtn_left_get(a_type, a_field,	\
+		  leftright);						\
+		if (leftrightleft != NULL && rbtn_red_get(a_type,	\
+		  a_field, leftrightleft)) {				\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*   /        \\                                  */\
+		    /* (r)        (b)                                 */\
+		    /*   \                                            */\
+		    /*   (b)                                          */\
+		    /*   /                                            */\
+		    /* (r)                                            */\
+		    a_type *unode;					\
+		    rbtn_black_set(a_type, a_field, leftrightleft);	\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      unode);						\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    rbtn_right_set(a_type, a_field, unode, tnode);	\
+		    rbtn_rotate_left(a_type, a_field, unode, tnode);	\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*   /        \\                                  */\
+		    /* (r)        (b)                                 */\
+		    /*   \                                            */\
+		    /*   (b)                                          */\
+		    /*   /                                            */\
+		    /* (b)                                            */\
+		    assert(leftright != NULL);				\
+		    rbtn_red_set(a_type, a_field, leftright);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    rbtn_black_set(a_type, a_field, tnode);		\
+		}							\
+		/* Balance restored, but rotation modified subtree    */\
+		/* root, which may actually be the tree root.         */\
+		if (pathp == path) {					\
+		    /* Set root. */					\
+		    rbtree->rbt_root = tnode;				\
+		} else {						\
+		    if (pathp[-1].cmp < 0) {				\
+			rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    } else {						\
+			rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    }							\
+		}							\
+		return;							\
+	    } else if (rbtn_red_get(a_type, a_field, pathp->node)) {	\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (leftleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  leftleft)) {						\
+		    /*        ||                                      */\
+		    /*      pathp(r)                                  */\
+		    /*     /        \\                                */\
+		    /*   (b)        (b)                               */\
+		    /*   /                                            */\
+		    /* (r)                                            */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    rbtn_red_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root.                                  */\
+		    assert((uintptr_t)pathp > (uintptr_t)path);		\
+		    if (pathp[-1].cmp < 0) {				\
+			rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    } else {						\
+			rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*        ||                                      */\
+		    /*      pathp(r)                                  */\
+		    /*     /        \\                                */\
+		    /*   (b)        (b)                               */\
+		    /*   /                                            */\
+		    /* (b)                                            */\
+		    rbtn_red_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    /* Balance restored. */				\
+		    return;						\
+		}							\
+	    } else {							\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (leftleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  leftleft)) {						\
+		    /*               ||                               */\
+		    /*             pathp(b)                           */\
+		    /*            /        \\                         */\
+		    /*          (b)        (b)                        */\
+		    /*          /                                     */\
+		    /*        (r)                                     */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root, which may actually be the tree   */\
+		    /* root.                                          */\
+		    if (pathp == path) {				\
+			/* Set root. */					\
+			rbtree->rbt_root = tnode;			\
+		    } else {						\
+			if (pathp[-1].cmp < 0) {			\
+			    rbtn_left_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			} else {					\
+			    rbtn_right_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			}						\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*               ||                               */\
+		    /*             pathp(b)                           */\
+		    /*            /        \\                         */\
+		    /*          (b)        (b)                        */\
+		    /*          /                                     */\
+		    /*        (b)                                     */\
+		    rbtn_red_set(a_type, a_field, left);		\
+		}							\
+	    }								\
+	}								\
+    }									\
+    /* Set root. */							\
+    rbtree->rbt_root = path->node;					\
+    assert(!rbtn_red_get(a_type, a_field, rbtree->rbt_root));		\
+}									\
+a_attr a_type *								\
+a_prefix##iter_recurse(a_rbt_type *rbtree, a_type *node,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    if (node == NULL) {							\
+	return NULL;							\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = a_prefix##iter_recurse(rbtree, rbtn_left_get(a_type,	\
+	  a_field, node), cb, arg)) != NULL || (ret = cb(rbtree, node,	\
+	  arg)) != NULL) {						\
+	    return ret;							\
+	}								\
+	return a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg);					\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##iter_start(a_rbt_type *rbtree, a_type *start, a_type *node,	\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    int cmp = a_cmp(start, node);					\
+    if (cmp < 0) {							\
+	a_type *ret;							\
+	if ((ret = a_prefix##iter_start(rbtree, start,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg)) != NULL ||	\
+	  (ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg);					\
+    } else if (cmp > 0) {						\
+	return a_prefix##iter_start(rbtree, start,			\
+	  rbtn_right_get(a_type, a_field, node), cb, arg);		\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg);					\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##iter(a_rbt_type *rbtree, a_type *start, a_type *(*cb)(	\
+  a_rbt_type *, a_type *, void *), void *arg) {				\
+    a_type *ret;							\
+    if (start != NULL) {						\
+	ret = a_prefix##iter_start(rbtree, start, rbtree->rbt_root,	\
+	  cb, arg);							\
+    } else {								\
+	ret = a_prefix##iter_recurse(rbtree, rbtree->rbt_root, cb, arg);\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter_recurse(a_rbt_type *rbtree, a_type *node,	\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    if (node == NULL) {							\
+	return NULL;							\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = a_prefix##reverse_iter_recurse(rbtree,		\
+	  rbtn_right_get(a_type, a_field, node), cb, arg)) != NULL ||	\
+	  (ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter_start(a_rbt_type *rbtree, a_type *start,		\
+  a_type *node, a_type *(*cb)(a_rbt_type *, a_type *, void *),		\
+  void *arg) {								\
+    int cmp = a_cmp(start, node);					\
+    if (cmp > 0) {							\
+	a_type *ret;							\
+	if ((ret = a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtn_right_get(a_type, a_field, node), cb, arg)) != NULL ||	\
+	  (ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    } else if (cmp < 0) {						\
+	return a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    a_type *ret;							\
+    if (start != NULL) {						\
+	ret = a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtree->rbt_root, cb, arg);					\
+    } else {								\
+	ret = a_prefix##reverse_iter_recurse(rbtree, rbtree->rbt_root,	\
+	  cb, arg);							\
+    }									\
+    return ret;								\
+}									\
+a_attr void								\
+a_prefix##destroy_recurse(a_rbt_type *rbtree, a_type *node, void (*cb)(	\
+  a_type *, void *), void *arg) {					\
+    if (node == NULL) {							\
+	return;								\
+    }									\
+    a_prefix##destroy_recurse(rbtree, rbtn_left_get(a_type, a_field,	\
+      node), cb, arg);							\
+    rbtn_left_set(a_type, a_field, (node), NULL);			\
+    a_prefix##destroy_recurse(rbtree, rbtn_right_get(a_type, a_field,	\
+      node), cb, arg);							\
+    rbtn_right_set(a_type, a_field, (node), NULL);			\
+    if (cb) {								\
+	cb(node, arg);							\
+    }									\
+}									\
+a_attr void								\
+a_prefix##destroy(a_rbt_type *rbtree, void (*cb)(a_type *, void *),	\
+  void *arg) {								\
+    a_prefix##destroy_recurse(rbtree, rbtree->rbt_root, cb, arg);	\
+    rbtree->rbt_root = NULL;						\
+}
+
+#endif /* RB_H_ */
diff --git a/jemalloc/include/jemalloc/internal/rtree.h b/jemalloc/include/jemalloc/internal/rtree.h
new file mode 100644
index 0000000..16ccbeb
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/rtree.h
@@ -0,0 +1,528 @@
+#ifndef JEMALLOC_INTERNAL_RTREE_H
+#define JEMALLOC_INTERNAL_RTREE_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree_tsd.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/tsd.h"
+
+/*
+ * This radix tree implementation is tailored to the singular purpose of
+ * associating metadata with extents that are currently owned by jemalloc.
+ *
+ *******************************************************************************
+ */
+
+/* Number of high insignificant bits. */
+#define RTREE_NHIB ((1U << (LG_SIZEOF_PTR+3)) - LG_VADDR)
+/* Number of low insigificant bits. */
+#define RTREE_NLIB LG_PAGE
+/* Number of significant bits. */
+#define RTREE_NSB (LG_VADDR - RTREE_NLIB)
+/* Number of levels in radix tree. */
+#if RTREE_NSB <= 10
+#  define RTREE_HEIGHT 1
+#elif RTREE_NSB <= 36
+#  define RTREE_HEIGHT 2
+#elif RTREE_NSB <= 52
+#  define RTREE_HEIGHT 3
+#else
+#  error Unsupported number of significant virtual address bits
+#endif
+/* Use compact leaf representation if virtual address encoding allows. */
+#if RTREE_NHIB >= LG_CEIL(SC_NSIZES)
+#  define RTREE_LEAF_COMPACT
+#endif
+
+/* Needed for initialization only. */
+#define RTREE_LEAFKEY_INVALID ((uintptr_t)1)
+
+typedef struct rtree_node_elm_s rtree_node_elm_t;
+struct rtree_node_elm_s {
+	atomic_p_t	child; /* (rtree_{node,leaf}_elm_t *) */
+};
+
+struct rtree_leaf_elm_s {
+#ifdef RTREE_LEAF_COMPACT
+	/*
+	 * Single pointer-width field containing all three leaf element fields.
+	 * For example, on a 64-bit x64 system with 48 significant virtual
+	 * memory address bits, the index, extent, and slab fields are packed as
+	 * such:
+	 *
+	 * x: index
+	 * e: extent
+	 * b: slab
+	 *
+	 *   00000000 xxxxxxxx eeeeeeee [...] eeeeeeee eeee000b
+	 */
+	atomic_p_t	le_bits;
+#else
+	atomic_p_t	le_extent; /* (extent_t *) */
+	atomic_u_t	le_szind; /* (szind_t) */
+	atomic_b_t	le_slab; /* (bool) */
+#endif
+};
+
+typedef struct rtree_level_s rtree_level_t;
+struct rtree_level_s {
+	/* Number of key bits distinguished by this level. */
+	unsigned		bits;
+	/*
+	 * Cumulative number of key bits distinguished by traversing to
+	 * corresponding tree level.
+	 */
+	unsigned		cumbits;
+};
+
+typedef struct rtree_s rtree_t;
+struct rtree_s {
+	malloc_mutex_t		init_lock;
+	/* Number of elements based on rtree_levels[0].bits. */
+#if RTREE_HEIGHT > 1
+	rtree_node_elm_t	root[1U << (RTREE_NSB/RTREE_HEIGHT)];
+#else
+	rtree_leaf_elm_t	root[1U << (RTREE_NSB/RTREE_HEIGHT)];
+#endif
+};
+
+/*
+ * Split the bits into one to three partitions depending on number of
+ * significant bits.  It the number of bits does not divide evenly into the
+ * number of levels, place one remainder bit per level starting at the leaf
+ * level.
+ */
+static const rtree_level_t rtree_levels[] = {
+#if RTREE_HEIGHT == 1
+	{RTREE_NSB, RTREE_NHIB + RTREE_NSB}
+#elif RTREE_HEIGHT == 2
+	{RTREE_NSB/2, RTREE_NHIB + RTREE_NSB/2},
+	{RTREE_NSB/2 + RTREE_NSB%2, RTREE_NHIB + RTREE_NSB}
+#elif RTREE_HEIGHT == 3
+	{RTREE_NSB/3, RTREE_NHIB + RTREE_NSB/3},
+	{RTREE_NSB/3 + RTREE_NSB%3/2,
+	    RTREE_NHIB + RTREE_NSB/3*2 + RTREE_NSB%3/2},
+	{RTREE_NSB/3 + RTREE_NSB%3 - RTREE_NSB%3/2, RTREE_NHIB + RTREE_NSB}
+#else
+#  error Unsupported rtree height
+#endif
+};
+
+bool rtree_new(rtree_t *rtree, bool zeroed);
+
+typedef rtree_node_elm_t *(rtree_node_alloc_t)(tsdn_t *, rtree_t *, size_t);
+extern rtree_node_alloc_t *JET_MUTABLE rtree_node_alloc;
+
+typedef rtree_leaf_elm_t *(rtree_leaf_alloc_t)(tsdn_t *, rtree_t *, size_t);
+extern rtree_leaf_alloc_t *JET_MUTABLE rtree_leaf_alloc;
+
+typedef void (rtree_node_dalloc_t)(tsdn_t *, rtree_t *, rtree_node_elm_t *);
+extern rtree_node_dalloc_t *JET_MUTABLE rtree_node_dalloc;
+
+typedef void (rtree_leaf_dalloc_t)(tsdn_t *, rtree_t *, rtree_leaf_elm_t *);
+extern rtree_leaf_dalloc_t *JET_MUTABLE rtree_leaf_dalloc;
+#ifdef JEMALLOC_JET
+void rtree_delete(tsdn_t *tsdn, rtree_t *rtree);
+#endif
+rtree_leaf_elm_t *rtree_leaf_elm_lookup_hard(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_ctx_t *rtree_ctx, uintptr_t key, bool dependent, bool init_missing);
+
+JEMALLOC_ALWAYS_INLINE uintptr_t
+rtree_leafkey(uintptr_t key) {
+	unsigned ptrbits = ZU(1) << (LG_SIZEOF_PTR+3);
+	unsigned cumbits = (rtree_levels[RTREE_HEIGHT-1].cumbits -
+	    rtree_levels[RTREE_HEIGHT-1].bits);
+	unsigned maskbits = ptrbits - cumbits;
+	uintptr_t mask = ~((ZU(1) << maskbits) - 1);
+	return (key & mask);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+rtree_cache_direct_map(uintptr_t key) {
+	unsigned ptrbits = ZU(1) << (LG_SIZEOF_PTR+3);
+	unsigned cumbits = (rtree_levels[RTREE_HEIGHT-1].cumbits -
+	    rtree_levels[RTREE_HEIGHT-1].bits);
+	unsigned maskbits = ptrbits - cumbits;
+	return (size_t)((key >> maskbits) & (RTREE_CTX_NCACHE - 1));
+}
+
+JEMALLOC_ALWAYS_INLINE uintptr_t
+rtree_subkey(uintptr_t key, unsigned level) {
+	unsigned ptrbits = ZU(1) << (LG_SIZEOF_PTR+3);
+	unsigned cumbits = rtree_levels[level].cumbits;
+	unsigned shiftbits = ptrbits - cumbits;
+	unsigned maskbits = rtree_levels[level].bits;
+	uintptr_t mask = (ZU(1) << maskbits) - 1;
+	return ((key >> shiftbits) & mask);
+}
+
+/*
+ * Atomic getters.
+ *
+ * dependent: Reading a value on behalf of a pointer to a valid allocation
+ *            is guaranteed to be a clean read even without synchronization,
+ *            because the rtree update became visible in memory before the
+ *            pointer came into existence.
+ * !dependent: An arbitrary read, e.g. on behalf of ivsalloc(), may not be
+ *             dependent on a previous rtree write, which means a stale read
+ *             could result if synchronization were omitted here.
+ */
+#  ifdef RTREE_LEAF_COMPACT
+JEMALLOC_ALWAYS_INLINE uintptr_t
+rtree_leaf_elm_bits_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+	return (uintptr_t)atomic_load_p(&elm->le_bits, dependent
+	    ? ATOMIC_RELAXED : ATOMIC_ACQUIRE);
+}
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+rtree_leaf_elm_bits_extent_get(uintptr_t bits) {
+#    ifdef __aarch64__
+	/*
+	 * aarch64 doesn't sign extend the highest virtual address bit to set
+	 * the higher ones.  Instead, the high bits gets zeroed.
+	 */
+	uintptr_t high_bit_mask = ((uintptr_t)1 << LG_VADDR) - 1;
+	/* Mask off the slab bit. */
+	uintptr_t low_bit_mask = ~(uintptr_t)1;
+	uintptr_t mask = high_bit_mask & low_bit_mask;
+	return (extent_t *)(bits & mask);
+#    else
+	/* Restore sign-extended high bits, mask slab bit. */
+	return (extent_t *)((uintptr_t)((intptr_t)(bits << RTREE_NHIB) >>
+	    RTREE_NHIB) & ~((uintptr_t)0x1));
+#    endif
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+rtree_leaf_elm_bits_szind_get(uintptr_t bits) {
+	return (szind_t)(bits >> LG_VADDR);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+rtree_leaf_elm_bits_slab_get(uintptr_t bits) {
+	return (bool)(bits & (uintptr_t)0x1);
+}
+
+#  endif
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+rtree_leaf_elm_extent_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	return rtree_leaf_elm_bits_extent_get(bits);
+#else
+	extent_t *extent = (extent_t *)atomic_load_p(&elm->le_extent, dependent
+	    ? ATOMIC_RELAXED : ATOMIC_ACQUIRE);
+	return extent;
+#endif
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+rtree_leaf_elm_szind_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	return rtree_leaf_elm_bits_szind_get(bits);
+#else
+	return (szind_t)atomic_load_u(&elm->le_szind, dependent ? ATOMIC_RELAXED
+	    : ATOMIC_ACQUIRE);
+#endif
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+rtree_leaf_elm_slab_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	return rtree_leaf_elm_bits_slab_get(bits);
+#else
+	return atomic_load_b(&elm->le_slab, dependent ? ATOMIC_RELAXED :
+	    ATOMIC_ACQUIRE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_extent_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, extent_t *extent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t old_bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, true);
+	uintptr_t bits = ((uintptr_t)rtree_leaf_elm_bits_szind_get(old_bits) <<
+	    LG_VADDR) | ((uintptr_t)extent & (((uintptr_t)0x1 << LG_VADDR) - 1))
+	    | ((uintptr_t)rtree_leaf_elm_bits_slab_get(old_bits));
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	atomic_store_p(&elm->le_extent, extent, ATOMIC_RELEASE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_szind_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, szind_t szind) {
+	assert(szind <= SC_NSIZES);
+
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t old_bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm,
+	    true);
+	uintptr_t bits = ((uintptr_t)szind << LG_VADDR) |
+	    ((uintptr_t)rtree_leaf_elm_bits_extent_get(old_bits) &
+	    (((uintptr_t)0x1 << LG_VADDR) - 1)) |
+	    ((uintptr_t)rtree_leaf_elm_bits_slab_get(old_bits));
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	atomic_store_u(&elm->le_szind, szind, ATOMIC_RELEASE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_slab_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool slab) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t old_bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm,
+	    true);
+	uintptr_t bits = ((uintptr_t)rtree_leaf_elm_bits_szind_get(old_bits) <<
+	    LG_VADDR) | ((uintptr_t)rtree_leaf_elm_bits_extent_get(old_bits) &
+	    (((uintptr_t)0x1 << LG_VADDR) - 1)) | ((uintptr_t)slab);
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	atomic_store_b(&elm->le_slab, slab, ATOMIC_RELEASE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, extent_t *extent, szind_t szind, bool slab) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = ((uintptr_t)szind << LG_VADDR) |
+	    ((uintptr_t)extent & (((uintptr_t)0x1 << LG_VADDR) - 1)) |
+	    ((uintptr_t)slab);
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	rtree_leaf_elm_slab_write(tsdn, rtree, elm, slab);
+	rtree_leaf_elm_szind_write(tsdn, rtree, elm, szind);
+	/*
+	 * Write extent last, since the element is atomically considered valid
+	 * as soon as the extent field is non-NULL.
+	 */
+	rtree_leaf_elm_extent_write(tsdn, rtree, elm, extent);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_szind_slab_update(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, szind_t szind, bool slab) {
+	assert(!slab || szind < SC_NBINS);
+
+	/*
+	 * The caller implicitly assures that it is the only writer to the szind
+	 * and slab fields, and that the extent field cannot currently change.
+	 */
+	rtree_leaf_elm_slab_write(tsdn, rtree, elm, slab);
+	rtree_leaf_elm_szind_write(tsdn, rtree, elm, szind);
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_leaf_elm_t *
+rtree_leaf_elm_lookup(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, bool init_missing) {
+	assert(key != 0);
+	assert(!dependent || !init_missing);
+
+	size_t slot = rtree_cache_direct_map(key);
+	uintptr_t leafkey = rtree_leafkey(key);
+	assert(leafkey != RTREE_LEAFKEY_INVALID);
+
+	/* Fast path: L1 direct mapped cache. */
+	if (likely(rtree_ctx->cache[slot].leafkey == leafkey)) {
+		rtree_leaf_elm_t *leaf = rtree_ctx->cache[slot].leaf;
+		assert(leaf != NULL);
+		uintptr_t subkey = rtree_subkey(key, RTREE_HEIGHT-1);
+		return &leaf[subkey];
+	}
+	/*
+	 * Search the L2 LRU cache.  On hit, swap the matching element into the
+	 * slot in L1 cache, and move the position in L2 up by 1.
+	 */
+#define RTREE_CACHE_CHECK_L2(i) do {					\
+	if (likely(rtree_ctx->l2_cache[i].leafkey == leafkey)) {	\
+		rtree_leaf_elm_t *leaf = rtree_ctx->l2_cache[i].leaf;	\
+		assert(leaf != NULL);					\
+		if (i > 0) {						\
+			/* Bubble up by one. */				\
+			rtree_ctx->l2_cache[i].leafkey =		\
+				rtree_ctx->l2_cache[i - 1].leafkey;	\
+			rtree_ctx->l2_cache[i].leaf =			\
+				rtree_ctx->l2_cache[i - 1].leaf;	\
+			rtree_ctx->l2_cache[i - 1].leafkey =		\
+			    rtree_ctx->cache[slot].leafkey;		\
+			rtree_ctx->l2_cache[i - 1].leaf =		\
+			    rtree_ctx->cache[slot].leaf;		\
+		} else {						\
+			rtree_ctx->l2_cache[0].leafkey =		\
+			    rtree_ctx->cache[slot].leafkey;		\
+			rtree_ctx->l2_cache[0].leaf =			\
+			    rtree_ctx->cache[slot].leaf;		\
+		}							\
+		rtree_ctx->cache[slot].leafkey = leafkey;		\
+		rtree_ctx->cache[slot].leaf = leaf;			\
+		uintptr_t subkey = rtree_subkey(key, RTREE_HEIGHT-1);	\
+		return &leaf[subkey];					\
+	}								\
+} while (0)
+	/* Check the first cache entry. */
+	RTREE_CACHE_CHECK_L2(0);
+	/* Search the remaining cache elements. */
+	for (unsigned i = 1; i < RTREE_CTX_NCACHE_L2; i++) {
+		RTREE_CACHE_CHECK_L2(i);
+	}
+#undef RTREE_CACHE_CHECK_L2
+
+	return rtree_leaf_elm_lookup_hard(tsdn, rtree, rtree_ctx, key,
+	    dependent, init_missing);
+}
+
+static inline bool
+rtree_write(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx, uintptr_t key,
+    extent_t *extent, szind_t szind, bool slab) {
+	/* Use rtree_clear() to set the extent to NULL. */
+	assert(extent != NULL);
+
+	rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, rtree, rtree_ctx,
+	    key, false, true);
+	if (elm == NULL) {
+		return true;
+	}
+
+	assert(rtree_leaf_elm_extent_read(tsdn, rtree, elm, false) == NULL);
+	rtree_leaf_elm_write(tsdn, rtree, elm, extent, szind, slab);
+
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_leaf_elm_t *
+rtree_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx, uintptr_t key,
+    bool dependent) {
+	rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, rtree, rtree_ctx,
+	    key, dependent, false);
+	if (!dependent && elm == NULL) {
+		return NULL;
+	}
+	assert(elm != NULL);
+	return elm;
+}
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+rtree_extent_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return NULL;
+	}
+	return rtree_leaf_elm_extent_read(tsdn, rtree, elm, dependent);
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+rtree_szind_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return SC_NSIZES;
+	}
+	return rtree_leaf_elm_szind_read(tsdn, rtree, elm, dependent);
+}
+
+/*
+ * rtree_slab_read() is intentionally omitted because slab is always read in
+ * conjunction with szind, which makes rtree_szind_slab_read() a better choice.
+ */
+
+JEMALLOC_ALWAYS_INLINE bool
+rtree_extent_szind_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, extent_t **r_extent, szind_t *r_szind) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return true;
+	}
+	*r_extent = rtree_leaf_elm_extent_read(tsdn, rtree, elm, dependent);
+	*r_szind = rtree_leaf_elm_szind_read(tsdn, rtree, elm, dependent);
+	return false;
+}
+
+/*
+ * Try to read szind_slab from the L1 cache.  Returns true on a hit,
+ * and fills in r_szind and r_slab.  Otherwise returns false.
+ *
+ * Key is allowed to be NULL in order to save an extra branch on the
+ * fastpath.  returns false in this case.
+ */
+JEMALLOC_ALWAYS_INLINE bool
+rtree_szind_slab_read_fast(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+			    uintptr_t key, szind_t *r_szind, bool *r_slab) {
+	rtree_leaf_elm_t *elm;
+
+	size_t slot = rtree_cache_direct_map(key);
+	uintptr_t leafkey = rtree_leafkey(key);
+	assert(leafkey != RTREE_LEAFKEY_INVALID);
+
+	if (likely(rtree_ctx->cache[slot].leafkey == leafkey)) {
+		rtree_leaf_elm_t *leaf = rtree_ctx->cache[slot].leaf;
+		assert(leaf != NULL);
+		uintptr_t subkey = rtree_subkey(key, RTREE_HEIGHT-1);
+		elm = &leaf[subkey];
+
+#ifdef RTREE_LEAF_COMPACT
+		uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree,
+							  elm, true);
+		*r_szind = rtree_leaf_elm_bits_szind_get(bits);
+		*r_slab = rtree_leaf_elm_bits_slab_get(bits);
+#else
+		*r_szind = rtree_leaf_elm_szind_read(tsdn, rtree, elm, true);
+		*r_slab = rtree_leaf_elm_slab_read(tsdn, rtree, elm, true);
+#endif
+		return true;
+	} else {
+		return false;
+	}
+}
+JEMALLOC_ALWAYS_INLINE bool
+rtree_szind_slab_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, szind_t *r_szind, bool *r_slab) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return true;
+	}
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	*r_szind = rtree_leaf_elm_bits_szind_get(bits);
+	*r_slab = rtree_leaf_elm_bits_slab_get(bits);
+#else
+	*r_szind = rtree_leaf_elm_szind_read(tsdn, rtree, elm, dependent);
+	*r_slab = rtree_leaf_elm_slab_read(tsdn, rtree, elm, dependent);
+#endif
+	return false;
+}
+
+static inline void
+rtree_szind_slab_update(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, szind_t szind, bool slab) {
+	assert(!slab || szind < SC_NBINS);
+
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key, true);
+	rtree_leaf_elm_szind_slab_update(tsdn, rtree, elm, szind, slab);
+}
+
+static inline void
+rtree_clear(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key, true);
+	assert(rtree_leaf_elm_extent_read(tsdn, rtree, elm, false) !=
+	    NULL);
+	rtree_leaf_elm_write(tsdn, rtree, elm, NULL, SC_NSIZES, false);
+}
+
+#endif /* JEMALLOC_INTERNAL_RTREE_H */
diff --git a/jemalloc/include/jemalloc/internal/rtree_tsd.h b/jemalloc/include/jemalloc/internal/rtree_tsd.h
new file mode 100644
index 0000000..562e292
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/rtree_tsd.h
@@ -0,0 +1,50 @@
+#ifndef JEMALLOC_INTERNAL_RTREE_CTX_H
+#define JEMALLOC_INTERNAL_RTREE_CTX_H
+
+/*
+ * Number of leafkey/leaf pairs to cache in L1 and L2 level respectively.  Each
+ * entry supports an entire leaf, so the cache hit rate is typically high even
+ * with a small number of entries.  In rare cases extent activity will straddle
+ * the boundary between two leaf nodes.  Furthermore, an arena may use a
+ * combination of dss and mmap.  Note that as memory usage grows past the amount
+ * that this cache can directly cover, the cache will become less effective if
+ * locality of reference is low, but the consequence is merely cache misses
+ * while traversing the tree nodes.
+ *
+ * The L1 direct mapped cache offers consistent and low cost on cache hit.
+ * However collision could affect hit rate negatively.  This is resolved by
+ * combining with a L2 LRU cache, which requires linear search and re-ordering
+ * on access but suffers no collision.  Note that, the cache will itself suffer
+ * cache misses if made overly large, plus the cost of linear search in the LRU
+ * cache.
+ */
+#define RTREE_CTX_LG_NCACHE 4
+#define RTREE_CTX_NCACHE (1 << RTREE_CTX_LG_NCACHE)
+#define RTREE_CTX_NCACHE_L2 8
+
+/*
+ * Zero initializer required for tsd initialization only.  Proper initialization
+ * done via rtree_ctx_data_init().
+ */
+#define RTREE_CTX_ZERO_INITIALIZER {{{0, 0}}, {{0, 0}}}
+
+
+typedef struct rtree_leaf_elm_s rtree_leaf_elm_t;
+
+typedef struct rtree_ctx_cache_elm_s rtree_ctx_cache_elm_t;
+struct rtree_ctx_cache_elm_s {
+	uintptr_t		leafkey;
+	rtree_leaf_elm_t	*leaf;
+};
+
+typedef struct rtree_ctx_s rtree_ctx_t;
+struct rtree_ctx_s {
+	/* Direct mapped cache. */
+	rtree_ctx_cache_elm_t	cache[RTREE_CTX_NCACHE];
+	/* L2 LRU cache. */
+	rtree_ctx_cache_elm_t	l2_cache[RTREE_CTX_NCACHE_L2];
+};
+
+void rtree_ctx_data_init(rtree_ctx_t *ctx);
+
+#endif /* JEMALLOC_INTERNAL_RTREE_CTX_H */
diff --git a/jemalloc/include/jemalloc/internal/safety_check.h b/jemalloc/include/jemalloc/internal/safety_check.h
new file mode 100644
index 0000000..53339ac
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/safety_check.h
@@ -0,0 +1,26 @@
+#ifndef JEMALLOC_INTERNAL_SAFETY_CHECK_H
+#define JEMALLOC_INTERNAL_SAFETY_CHECK_H
+
+void safety_check_fail(const char *format, ...);
+/* Can set to NULL for a default. */
+void safety_check_set_abort(void (*abort_fn)());
+
+JEMALLOC_ALWAYS_INLINE void
+safety_check_set_redzone(void *ptr, size_t usize, size_t bumped_usize) {
+	assert(usize < bumped_usize);
+	for (size_t i = usize; i < bumped_usize && i < usize + 32; ++i) {
+		*((unsigned char *)ptr + i) = 0xBC;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+safety_check_verify_redzone(const void *ptr, size_t usize, size_t bumped_usize)
+{
+	for (size_t i = usize; i < bumped_usize && i < usize + 32; ++i) {
+		if (unlikely(*((unsigned char *)ptr + i) != 0xBC)) {
+			safety_check_fail("Use after free error\n");
+		}
+	}
+}
+
+#endif /*JEMALLOC_INTERNAL_SAFETY_CHECK_H */
diff --git a/jemalloc/include/jemalloc/internal/sc.h b/jemalloc/include/jemalloc/internal/sc.h
new file mode 100644
index 0000000..9a099d8
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/sc.h
@@ -0,0 +1,333 @@
+#ifndef JEMALLOC_INTERNAL_SC_H
+#define JEMALLOC_INTERNAL_SC_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+
+/*
+ * Size class computations:
+ *
+ * These are a little tricky; we'll first start by describing how things
+ * generally work, and then describe some of the details.
+ *
+ * Ignore the first few size classes for a moment. We can then split all the
+ * remaining size classes into groups. The size classes in a group are spaced
+ * such that they cover allocation request sizes in a power-of-2 range. The
+ * power of two is called the base of the group, and the size classes in it
+ * satisfy allocations in the half-open range (base, base * 2]. There are
+ * SC_NGROUP size classes in each group, equally spaced in the range, so that
+ * each one covers allocations for base / SC_NGROUP possible allocation sizes.
+ * We call that value (base / SC_NGROUP) the delta of the group. Each size class
+ * is delta larger than the one before it (including the initial size class in a
+ * group, which is delta larger than base, the largest size class in the
+ * previous group).
+ * To make the math all work out nicely, we require that SC_NGROUP is a power of
+ * two, and define it in terms of SC_LG_NGROUP. We'll often talk in terms of
+ * lg_base and lg_delta. For each of these groups then, we have that
+ * lg_delta == lg_base - SC_LG_NGROUP.
+ * The size classes in a group with a given lg_base and lg_delta (which, recall,
+ * can be computed from lg_base for these groups) are therefore:
+ *   base + 1 * delta
+ *     which covers allocations in (base, base + 1 * delta]
+ *   base + 2 * delta
+ *     which covers allocations in (base + 1 * delta, base + 2 * delta].
+ *   base + 3 * delta
+ *     which covers allocations in (base + 2 * delta, base + 3 * delta].
+ *   ...
+ *   base + SC_NGROUP * delta ( == 2 * base)
+ *     which covers allocations in (base + (SC_NGROUP - 1) * delta, 2 * base].
+ * (Note that currently SC_NGROUP is always 4, so the "..." is empty in
+ * practice.)
+ * Note that the last size class in the group is the next power of two (after
+ * base), so that we've set up the induction correctly for the next group's
+ * selection of delta.
+ *
+ * Now, let's start considering the first few size classes. Two extra constants
+ * come into play here: LG_QUANTUM and SC_LG_TINY_MIN. LG_QUANTUM ensures
+ * correct platform alignment; all objects of size (1 << LG_QUANTUM) or larger
+ * are at least (1 << LG_QUANTUM) aligned; this can be used to ensure that we
+ * never return improperly aligned memory, by making (1 << LG_QUANTUM) equal the
+ * highest required alignment of a platform. For allocation sizes smaller than
+ * (1 << LG_QUANTUM) though, we can be more relaxed (since we don't support
+ * platforms with types with alignment larger than their size). To allow such
+ * allocations (without wasting space unnecessarily), we introduce tiny size
+ * classes; one per power of two, up until we hit the quantum size. There are
+ * therefore LG_QUANTUM - SC_LG_TINY_MIN such size classes.
+ *
+ * Next, we have a size class of size (1 << LG_QUANTUM).  This can't be the
+ * start of a group in the sense we described above (covering a power of two
+ * range) since, if we divided into it to pick a value of delta, we'd get a
+ * delta smaller than (1 << LG_QUANTUM) for sizes >= (1 << LG_QUANTUM), which
+ * is against the rules.
+ *
+ * The first base we can divide by SC_NGROUP while still being at least
+ * (1 << LG_QUANTUM) is SC_NGROUP * (1 << LG_QUANTUM). We can get there by
+ * having SC_NGROUP size classes, spaced (1 << LG_QUANTUM) apart. These size
+ * classes are:
+ *   1 * (1 << LG_QUANTUM)
+ *   2 * (1 << LG_QUANTUM)
+ *   3 * (1 << LG_QUANTUM)
+ *   ... (although, as above, this "..." is empty in practice)
+ *   SC_NGROUP * (1 << LG_QUANTUM).
+ *
+ * There are SC_NGROUP of these size classes, so we can regard it as a sort of
+ * pseudo-group, even though it spans multiple powers of 2, is divided
+ * differently, and both starts and ends on a power of 2 (as opposed to just
+ * ending). SC_NGROUP is itself a power of two, so the first group after the
+ * pseudo-group has the power-of-two base SC_NGROUP * (1 << LG_QUANTUM), for a
+ * lg_base of LG_QUANTUM + SC_LG_NGROUP. We can divide this base into SC_NGROUP
+ * sizes without violating our LG_QUANTUM requirements, so we can safely set
+ * lg_delta = lg_base - SC_LG_GROUP (== LG_QUANTUM).
+ *
+ * So, in order, the size classes are:
+ *
+ * Tiny size classes:
+ * - Count: LG_QUANTUM - SC_LG_TINY_MIN.
+ * - Sizes:
+ *     1 << SC_LG_TINY_MIN
+ *     1 << (SC_LG_TINY_MIN + 1)
+ *     1 << (SC_LG_TINY_MIN + 2)
+ *     ...
+ *     1 << (LG_QUANTUM - 1)
+ *
+ * Initial pseudo-group:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *     1 * (1 << LG_QUANTUM)
+ *     2 * (1 << LG_QUANTUM)
+ *     3 * (1 << LG_QUANTUM)
+ *     ...
+ *     SC_NGROUP * (1 << LG_QUANTUM)
+ *
+ * Regular group 0:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *   (relative to lg_base of LG_QUANTUM + SC_LG_NGROUP and lg_delta of
+ *   lg_base - SC_LG_NGROUP)
+ *     (1 << lg_base) + 1 * (1 << lg_delta)
+ *     (1 << lg_base) + 2 * (1 << lg_delta)
+ *     (1 << lg_base) + 3 * (1 << lg_delta)
+ *     ...
+ *     (1 << lg_base) + SC_NGROUP * (1 << lg_delta) [ == (1 << (lg_base + 1)) ]
+ *
+ * Regular group 1:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *   (relative to lg_base of LG_QUANTUM + SC_LG_NGROUP + 1 and lg_delta of
+ *   lg_base - SC_LG_NGROUP)
+ *     (1 << lg_base) + 1 * (1 << lg_delta)
+ *     (1 << lg_base) + 2 * (1 << lg_delta)
+ *     (1 << lg_base) + 3 * (1 << lg_delta)
+ *     ...
+ *     (1 << lg_base) + SC_NGROUP * (1 << lg_delta) [ == (1 << (lg_base + 1)) ]
+ *
+ * ...
+ *
+ * Regular group N:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *   (relative to lg_base of LG_QUANTUM + SC_LG_NGROUP + N and lg_delta of
+ *   lg_base - SC_LG_NGROUP)
+ *     (1 << lg_base) + 1 * (1 << lg_delta)
+ *     (1 << lg_base) + 2 * (1 << lg_delta)
+ *     (1 << lg_base) + 3 * (1 << lg_delta)
+ *     ...
+ *     (1 << lg_base) + SC_NGROUP * (1 << lg_delta) [ == (1 << (lg_base + 1)) ]
+ *
+ *
+ * Representation of metadata:
+ * To make the math easy, we'll mostly work in lg quantities. We record lg_base,
+ * lg_delta, and ndelta (i.e. number of deltas above the base) on a
+ * per-size-class basis, and maintain the invariant that, across all size
+ * classes, size == (1 << lg_base) + ndelta * (1 << lg_delta).
+ *
+ * For regular groups (i.e. those with lg_base >= LG_QUANTUM + SC_LG_NGROUP),
+ * lg_delta is lg_base - SC_LG_NGROUP, and ndelta goes from 1 to SC_NGROUP.
+ *
+ * For the initial tiny size classes (if any), lg_base is lg(size class size).
+ * lg_delta is lg_base for the first size class, and lg_base - 1 for all
+ * subsequent ones. ndelta is always 0.
+ *
+ * For the pseudo-group, if there are no tiny size classes, then we set
+ * lg_base == LG_QUANTUM, lg_delta == LG_QUANTUM, and have ndelta range from 0
+ * to SC_NGROUP - 1. (Note that delta == base, so base + (SC_NGROUP - 1) * delta
+ * is just SC_NGROUP * base, or (1 << (SC_LG_NGROUP + LG_QUANTUM)), so we do
+ * indeed get a power of two that way). If there *are* tiny size classes, then
+ * the first size class needs to have lg_delta relative to the largest tiny size
+ * class. We therefore set lg_base == LG_QUANTUM - 1,
+ * lg_delta == LG_QUANTUM - 1, and ndelta == 1, keeping the rest of the
+ * pseudo-group the same.
+ *
+ *
+ * Other terminology:
+ * "Small" size classes mean those that are allocated out of bins, which is the
+ * same as those that are slab allocated.
+ * "Large" size classes are those that are not small. The cutoff for counting as
+ * large is page size * group size.
+ */
+
+/*
+ * Size class N + (1 << SC_LG_NGROUP) twice the size of size class N.
+ */
+#define SC_LG_NGROUP 2
+#define SC_LG_TINY_MIN 3
+
+#if SC_LG_TINY_MIN == 0
+/* The div module doesn't support division by 1, which this would require. */
+#error "Unsupported LG_TINY_MIN"
+#endif
+
+/*
+ * The definitions below are all determined by the above settings and system
+ * characteristics.
+ */
+#define SC_NGROUP (1ULL << SC_LG_NGROUP)
+#define SC_PTR_BITS ((1ULL << LG_SIZEOF_PTR) * 8)
+#define SC_NTINY (LG_QUANTUM - SC_LG_TINY_MIN)
+#define SC_LG_TINY_MAXCLASS (LG_QUANTUM > SC_LG_TINY_MIN ? LG_QUANTUM - 1 : -1)
+#define SC_NPSEUDO SC_NGROUP
+#define SC_LG_FIRST_REGULAR_BASE (LG_QUANTUM + SC_LG_NGROUP)
+/*
+ * We cap allocations to be less than 2 ** (ptr_bits - 1), so the highest base
+ * we need is 2 ** (ptr_bits - 2). (This also means that the last group is 1
+ * size class shorter than the others).
+ * We could probably save some space in arenas by capping this at LG_VADDR size.
+ */
+#define SC_LG_BASE_MAX (SC_PTR_BITS - 2)
+#define SC_NREGULAR (SC_NGROUP * 					\
+    (SC_LG_BASE_MAX - SC_LG_FIRST_REGULAR_BASE + 1) - 1)
+#define SC_NSIZES (SC_NTINY + SC_NPSEUDO + SC_NREGULAR)
+
+/* The number of size classes that are a multiple of the page size. */
+#define SC_NPSIZES (							\
+    /* Start with all the size classes. */				\
+    SC_NSIZES								\
+    /* Subtract out those groups with too small a base. */		\
+    - (LG_PAGE - 1 - SC_LG_FIRST_REGULAR_BASE) * SC_NGROUP		\
+    /* And the pseudo-group. */						\
+    - SC_NPSEUDO							\
+    /* And the tiny group. */						\
+    - SC_NTINY								\
+    /* Sizes where ndelta*delta is not a multiple of the page size. */	\
+    - (SC_LG_NGROUP * SC_NGROUP))
+/*
+ * Note that the last line is computed as the sum of the second column in the
+ * following table:
+ *                      lg(base) | count of sizes to exclude
+ * ------------------------------|-----------------------------
+ *                   LG_PAGE - 1 | SC_NGROUP - 1
+ *                       LG_PAGE | SC_NGROUP - 1
+ *                   LG_PAGE + 1 | SC_NGROUP - 2
+ *                   LG_PAGE + 2 | SC_NGROUP - 4
+ *                           ... | ...
+ *  LG_PAGE + (SC_LG_NGROUP - 1) | SC_NGROUP - (SC_NGROUP / 2)
+ */
+
+/*
+ * We declare a size class is binnable if size < page size * group. Or, in other
+ * words, lg(size) < lg(page size) + lg(group size).
+ */
+#define SC_NBINS (							\
+    /* Sub-regular size classes. */					\
+    SC_NTINY + SC_NPSEUDO						\
+    /* Groups with lg_regular_min_base <= lg_base <= lg_base_max */	\
+    + SC_NGROUP * (LG_PAGE + SC_LG_NGROUP - SC_LG_FIRST_REGULAR_BASE)	\
+    /* Last SC of the last group hits the bound exactly; exclude it. */	\
+    - 1)
+
+/*
+ * The size2index_tab lookup table uses uint8_t to encode each bin index, so we
+ * cannot support more than 256 small size classes.
+ */
+#if (SC_NBINS > 256)
+#  error "Too many small size classes"
+#endif
+
+/* The largest size class in the lookup table. */
+#define SC_LOOKUP_MAXCLASS ((size_t)1 << 12)
+
+/* Internal, only used for the definition of SC_SMALL_MAXCLASS. */
+#define SC_SMALL_MAX_BASE ((size_t)1 << (LG_PAGE + SC_LG_NGROUP - 1))
+#define SC_SMALL_MAX_DELTA ((size_t)1 << (LG_PAGE - 1))
+
+/* The largest size class allocated out of a slab. */
+#define SC_SMALL_MAXCLASS (SC_SMALL_MAX_BASE				\
+    + (SC_NGROUP - 1) * SC_SMALL_MAX_DELTA)
+
+/* The smallest size class not allocated out of a slab. */
+#define SC_LARGE_MINCLASS ((size_t)1ULL << (LG_PAGE + SC_LG_NGROUP))
+#define SC_LG_LARGE_MINCLASS (LG_PAGE + SC_LG_NGROUP)
+
+/* Internal; only used for the definition of SC_LARGE_MAXCLASS. */
+#define SC_MAX_BASE ((size_t)1 << (SC_PTR_BITS - 2))
+#define SC_MAX_DELTA ((size_t)1 << (SC_PTR_BITS - 2 - SC_LG_NGROUP))
+
+/* The largest size class supported. */
+#define SC_LARGE_MAXCLASS (SC_MAX_BASE + (SC_NGROUP - 1) * SC_MAX_DELTA)
+
+typedef struct sc_s sc_t;
+struct sc_s {
+	/* Size class index, or -1 if not a valid size class. */
+	int index;
+	/* Lg group base size (no deltas added). */
+	int lg_base;
+	/* Lg delta to previous size class. */
+	int lg_delta;
+	/* Delta multiplier.  size == 1<<lg_base + ndelta<<lg_delta */
+	int ndelta;
+	/*
+	 * True if the size class is a multiple of the page size, false
+	 * otherwise.
+	 */
+	bool psz;
+	/*
+	 * True if the size class is a small, bin, size class. False otherwise.
+	 */
+	bool bin;
+	/* The slab page count if a small bin size class, 0 otherwise. */
+	int pgs;
+	/* Same as lg_delta if a lookup table size class, 0 otherwise. */
+	int lg_delta_lookup;
+};
+
+typedef struct sc_data_s sc_data_t;
+struct sc_data_s {
+	/* Number of tiny size classes. */
+	unsigned ntiny;
+	/* Number of bins supported by the lookup table. */
+	int nlbins;
+	/* Number of small size class bins. */
+	int nbins;
+	/* Number of size classes. */
+	int nsizes;
+	/* Number of bits required to store NSIZES. */
+	int lg_ceil_nsizes;
+	/* Number of size classes that are a multiple of (1U << LG_PAGE). */
+	unsigned npsizes;
+	/* Lg of maximum tiny size class (or -1, if none). */
+	int lg_tiny_maxclass;
+	/* Maximum size class included in lookup table. */
+	size_t lookup_maxclass;
+	/* Maximum small size class. */
+	size_t small_maxclass;
+	/* Lg of minimum large size class. */
+	int lg_large_minclass;
+	/* The minimum large size class. */
+	size_t large_minclass;
+	/* Maximum (large) size class. */
+	size_t large_maxclass;
+	/* True if the sc_data_t has been initialized (for debugging only). */
+	bool initialized;
+
+	sc_t sc[SC_NSIZES];
+};
+
+void sc_data_init(sc_data_t *data);
+/*
+ * Updates slab sizes in [begin, end] to be pgs pages in length, if possible.
+ * Otherwise, does its best to accomodate the request.
+ */
+void sc_data_update_slab_size(sc_data_t *data, size_t begin, size_t end,
+    int pgs);
+void sc_boot(sc_data_t *data);
+
+#endif /* JEMALLOC_INTERNAL_SC_H */
diff --git a/jemalloc/include/jemalloc/internal/seq.h b/jemalloc/include/jemalloc/internal/seq.h
new file mode 100644
index 0000000..ef2df4c
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/seq.h
@@ -0,0 +1,55 @@
+#ifndef JEMALLOC_INTERNAL_SEQ_H
+#define JEMALLOC_INTERNAL_SEQ_H
+
+#include "jemalloc/internal/atomic.h"
+
+/*
+ * A simple seqlock implementation.
+ */
+
+#define seq_define(type, short_type)					\
+typedef struct {							\
+	atomic_zu_t seq;						\
+	atomic_zu_t data[						\
+	    (sizeof(type) + sizeof(size_t) - 1) / sizeof(size_t)];	\
+} seq_##short_type##_t;							\
+									\
+/*									\
+ * No internal synchronization -- the caller must ensure that there's	\
+ * only a single writer at a time.					\
+ */									\
+static inline void							\
+seq_store_##short_type(seq_##short_type##_t *dst, type *src) {		\
+	size_t buf[sizeof(dst->data) / sizeof(size_t)];			\
+	buf[sizeof(buf) / sizeof(size_t) - 1] = 0;			\
+	memcpy(buf, src, sizeof(type));					\
+	size_t old_seq = atomic_load_zu(&dst->seq, ATOMIC_RELAXED);	\
+	atomic_store_zu(&dst->seq, old_seq + 1, ATOMIC_RELAXED);	\
+	atomic_fence(ATOMIC_RELEASE);					\
+	for (size_t i = 0; i < sizeof(buf) / sizeof(size_t); i++) {	\
+		atomic_store_zu(&dst->data[i], buf[i], ATOMIC_RELAXED);	\
+	}								\
+	atomic_store_zu(&dst->seq, old_seq + 2, ATOMIC_RELEASE);	\
+}									\
+									\
+/* Returns whether or not the read was consistent. */			\
+static inline bool							\
+seq_try_load_##short_type(type *dst, seq_##short_type##_t *src) {	\
+	size_t buf[sizeof(src->data) / sizeof(size_t)];			\
+	size_t seq1 = atomic_load_zu(&src->seq, ATOMIC_ACQUIRE);	\
+	if (seq1 % 2 != 0) {						\
+		return false;						\
+	}								\
+	for (size_t i = 0; i < sizeof(buf) / sizeof(size_t); i++) {	\
+		buf[i] = atomic_load_zu(&src->data[i], ATOMIC_RELAXED);	\
+	}								\
+	atomic_fence(ATOMIC_ACQUIRE);					\
+	size_t seq2 = atomic_load_zu(&src->seq, ATOMIC_RELAXED);	\
+	if (seq1 != seq2) {						\
+		return false;						\
+	}								\
+	memcpy(dst, buf, sizeof(type));					\
+	return true;							\
+}
+
+#endif /* JEMALLOC_INTERNAL_SEQ_H */
diff --git a/jemalloc/include/jemalloc/internal/smoothstep.h b/jemalloc/include/jemalloc/internal/smoothstep.h
new file mode 100644
index 0000000..2e14430
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/smoothstep.h
@@ -0,0 +1,232 @@
+#ifndef JEMALLOC_INTERNAL_SMOOTHSTEP_H
+#define JEMALLOC_INTERNAL_SMOOTHSTEP_H
+
+/*
+ * This file was generated by the following command:
+ *   sh smoothstep.sh smoother 200 24 3 15
+ */
+/******************************************************************************/
+
+/*
+ * This header defines a precomputed table based on the smoothstep family of
+ * sigmoidal curves (https://en.wikipedia.org/wiki/Smoothstep) that grow from 0
+ * to 1 in 0 <= x <= 1.  The table is stored as integer fixed point values so
+ * that floating point math can be avoided.
+ *
+ *                      3     2
+ *   smoothstep(x) = -2x  + 3x
+ *
+ *                       5      4      3
+ *   smootherstep(x) = 6x  - 15x  + 10x
+ *
+ *                          7      6      5      4
+ *   smootheststep(x) = -20x  + 70x  - 84x  + 35x
+ */
+
+#define SMOOTHSTEP_VARIANT	"smoother"
+#define SMOOTHSTEP_NSTEPS	200
+#define SMOOTHSTEP_BFP		24
+#define SMOOTHSTEP \
+ /* STEP(step, h,                            x,     y) */ \
+    STEP(   1, UINT64_C(0x0000000000000014), 0.005, 0.000001240643750) \
+    STEP(   2, UINT64_C(0x00000000000000a5), 0.010, 0.000009850600000) \
+    STEP(   3, UINT64_C(0x0000000000000229), 0.015, 0.000032995181250) \
+    STEP(   4, UINT64_C(0x0000000000000516), 0.020, 0.000077619200000) \
+    STEP(   5, UINT64_C(0x00000000000009dc), 0.025, 0.000150449218750) \
+    STEP(   6, UINT64_C(0x00000000000010e8), 0.030, 0.000257995800000) \
+    STEP(   7, UINT64_C(0x0000000000001aa4), 0.035, 0.000406555756250) \
+    STEP(   8, UINT64_C(0x0000000000002777), 0.040, 0.000602214400000) \
+    STEP(   9, UINT64_C(0x00000000000037c2), 0.045, 0.000850847793750) \
+    STEP(  10, UINT64_C(0x0000000000004be6), 0.050, 0.001158125000000) \
+    STEP(  11, UINT64_C(0x000000000000643c), 0.055, 0.001529510331250) \
+    STEP(  12, UINT64_C(0x000000000000811f), 0.060, 0.001970265600000) \
+    STEP(  13, UINT64_C(0x000000000000a2e2), 0.065, 0.002485452368750) \
+    STEP(  14, UINT64_C(0x000000000000c9d8), 0.070, 0.003079934200000) \
+    STEP(  15, UINT64_C(0x000000000000f64f), 0.075, 0.003758378906250) \
+    STEP(  16, UINT64_C(0x0000000000012891), 0.080, 0.004525260800000) \
+    STEP(  17, UINT64_C(0x00000000000160e7), 0.085, 0.005384862943750) \
+    STEP(  18, UINT64_C(0x0000000000019f95), 0.090, 0.006341279400000) \
+    STEP(  19, UINT64_C(0x000000000001e4dc), 0.095, 0.007398417481250) \
+    STEP(  20, UINT64_C(0x00000000000230fc), 0.100, 0.008560000000000) \
+    STEP(  21, UINT64_C(0x0000000000028430), 0.105, 0.009829567518750) \
+    STEP(  22, UINT64_C(0x000000000002deb0), 0.110, 0.011210480600000) \
+    STEP(  23, UINT64_C(0x00000000000340b1), 0.115, 0.012705922056250) \
+    STEP(  24, UINT64_C(0x000000000003aa67), 0.120, 0.014318899200000) \
+    STEP(  25, UINT64_C(0x0000000000041c00), 0.125, 0.016052246093750) \
+    STEP(  26, UINT64_C(0x00000000000495a8), 0.130, 0.017908625800000) \
+    STEP(  27, UINT64_C(0x000000000005178b), 0.135, 0.019890532631250) \
+    STEP(  28, UINT64_C(0x000000000005a1cf), 0.140, 0.022000294400000) \
+    STEP(  29, UINT64_C(0x0000000000063498), 0.145, 0.024240074668750) \
+    STEP(  30, UINT64_C(0x000000000006d009), 0.150, 0.026611875000000) \
+    STEP(  31, UINT64_C(0x000000000007743f), 0.155, 0.029117537206250) \
+    STEP(  32, UINT64_C(0x0000000000082157), 0.160, 0.031758745600000) \
+    STEP(  33, UINT64_C(0x000000000008d76b), 0.165, 0.034537029243750) \
+    STEP(  34, UINT64_C(0x0000000000099691), 0.170, 0.037453764200000) \
+    STEP(  35, UINT64_C(0x00000000000a5edf), 0.175, 0.040510175781250) \
+    STEP(  36, UINT64_C(0x00000000000b3067), 0.180, 0.043707340800000) \
+    STEP(  37, UINT64_C(0x00000000000c0b38), 0.185, 0.047046189818750) \
+    STEP(  38, UINT64_C(0x00000000000cef5e), 0.190, 0.050527509400000) \
+    STEP(  39, UINT64_C(0x00000000000ddce6), 0.195, 0.054151944356250) \
+    STEP(  40, UINT64_C(0x00000000000ed3d8), 0.200, 0.057920000000000) \
+    STEP(  41, UINT64_C(0x00000000000fd439), 0.205, 0.061832044393750) \
+    STEP(  42, UINT64_C(0x000000000010de0e), 0.210, 0.065888310600000) \
+    STEP(  43, UINT64_C(0x000000000011f158), 0.215, 0.070088898931250) \
+    STEP(  44, UINT64_C(0x0000000000130e17), 0.220, 0.074433779200000) \
+    STEP(  45, UINT64_C(0x0000000000143448), 0.225, 0.078922792968750) \
+    STEP(  46, UINT64_C(0x00000000001563e7), 0.230, 0.083555655800000) \
+    STEP(  47, UINT64_C(0x0000000000169cec), 0.235, 0.088331959506250) \
+    STEP(  48, UINT64_C(0x000000000017df4f), 0.240, 0.093251174400000) \
+    STEP(  49, UINT64_C(0x0000000000192b04), 0.245, 0.098312651543750) \
+    STEP(  50, UINT64_C(0x00000000001a8000), 0.250, 0.103515625000000) \
+    STEP(  51, UINT64_C(0x00000000001bde32), 0.255, 0.108859214081250) \
+    STEP(  52, UINT64_C(0x00000000001d458b), 0.260, 0.114342425600000) \
+    STEP(  53, UINT64_C(0x00000000001eb5f8), 0.265, 0.119964156118750) \
+    STEP(  54, UINT64_C(0x0000000000202f65), 0.270, 0.125723194200000) \
+    STEP(  55, UINT64_C(0x000000000021b1bb), 0.275, 0.131618222656250) \
+    STEP(  56, UINT64_C(0x0000000000233ce3), 0.280, 0.137647820800000) \
+    STEP(  57, UINT64_C(0x000000000024d0c3), 0.285, 0.143810466693750) \
+    STEP(  58, UINT64_C(0x0000000000266d40), 0.290, 0.150104539400000) \
+    STEP(  59, UINT64_C(0x000000000028123d), 0.295, 0.156528321231250) \
+    STEP(  60, UINT64_C(0x000000000029bf9c), 0.300, 0.163080000000000) \
+    STEP(  61, UINT64_C(0x00000000002b753d), 0.305, 0.169757671268750) \
+    STEP(  62, UINT64_C(0x00000000002d32fe), 0.310, 0.176559340600000) \
+    STEP(  63, UINT64_C(0x00000000002ef8bc), 0.315, 0.183482925806250) \
+    STEP(  64, UINT64_C(0x000000000030c654), 0.320, 0.190526259200000) \
+    STEP(  65, UINT64_C(0x0000000000329b9f), 0.325, 0.197687089843750) \
+    STEP(  66, UINT64_C(0x0000000000347875), 0.330, 0.204963085800000) \
+    STEP(  67, UINT64_C(0x0000000000365cb0), 0.335, 0.212351836381250) \
+    STEP(  68, UINT64_C(0x0000000000384825), 0.340, 0.219850854400000) \
+    STEP(  69, UINT64_C(0x00000000003a3aa8), 0.345, 0.227457578418750) \
+    STEP(  70, UINT64_C(0x00000000003c340f), 0.350, 0.235169375000000) \
+    STEP(  71, UINT64_C(0x00000000003e342b), 0.355, 0.242983540956250) \
+    STEP(  72, UINT64_C(0x0000000000403ace), 0.360, 0.250897305600000) \
+    STEP(  73, UINT64_C(0x00000000004247c8), 0.365, 0.258907832993750) \
+    STEP(  74, UINT64_C(0x0000000000445ae9), 0.370, 0.267012224200000) \
+    STEP(  75, UINT64_C(0x0000000000467400), 0.375, 0.275207519531250) \
+    STEP(  76, UINT64_C(0x00000000004892d8), 0.380, 0.283490700800000) \
+    STEP(  77, UINT64_C(0x00000000004ab740), 0.385, 0.291858693568750) \
+    STEP(  78, UINT64_C(0x00000000004ce102), 0.390, 0.300308369400000) \
+    STEP(  79, UINT64_C(0x00000000004f0fe9), 0.395, 0.308836548106250) \
+    STEP(  80, UINT64_C(0x00000000005143bf), 0.400, 0.317440000000000) \
+    STEP(  81, UINT64_C(0x0000000000537c4d), 0.405, 0.326115448143750) \
+    STEP(  82, UINT64_C(0x000000000055b95b), 0.410, 0.334859570600000) \
+    STEP(  83, UINT64_C(0x000000000057fab1), 0.415, 0.343669002681250) \
+    STEP(  84, UINT64_C(0x00000000005a4015), 0.420, 0.352540339200000) \
+    STEP(  85, UINT64_C(0x00000000005c894e), 0.425, 0.361470136718750) \
+    STEP(  86, UINT64_C(0x00000000005ed622), 0.430, 0.370454915800000) \
+    STEP(  87, UINT64_C(0x0000000000612655), 0.435, 0.379491163256250) \
+    STEP(  88, UINT64_C(0x00000000006379ac), 0.440, 0.388575334400000) \
+    STEP(  89, UINT64_C(0x000000000065cfeb), 0.445, 0.397703855293750) \
+    STEP(  90, UINT64_C(0x00000000006828d6), 0.450, 0.406873125000000) \
+    STEP(  91, UINT64_C(0x00000000006a842f), 0.455, 0.416079517831250) \
+    STEP(  92, UINT64_C(0x00000000006ce1bb), 0.460, 0.425319385600000) \
+    STEP(  93, UINT64_C(0x00000000006f413a), 0.465, 0.434589059868750) \
+    STEP(  94, UINT64_C(0x000000000071a270), 0.470, 0.443884854200000) \
+    STEP(  95, UINT64_C(0x000000000074051d), 0.475, 0.453203066406250) \
+    STEP(  96, UINT64_C(0x0000000000766905), 0.480, 0.462539980800000) \
+    STEP(  97, UINT64_C(0x000000000078cde7), 0.485, 0.471891870443750) \
+    STEP(  98, UINT64_C(0x00000000007b3387), 0.490, 0.481254999400000) \
+    STEP(  99, UINT64_C(0x00000000007d99a4), 0.495, 0.490625624981250) \
+    STEP( 100, UINT64_C(0x0000000000800000), 0.500, 0.500000000000000) \
+    STEP( 101, UINT64_C(0x000000000082665b), 0.505, 0.509374375018750) \
+    STEP( 102, UINT64_C(0x000000000084cc78), 0.510, 0.518745000600000) \
+    STEP( 103, UINT64_C(0x0000000000873218), 0.515, 0.528108129556250) \
+    STEP( 104, UINT64_C(0x00000000008996fa), 0.520, 0.537460019200000) \
+    STEP( 105, UINT64_C(0x00000000008bfae2), 0.525, 0.546796933593750) \
+    STEP( 106, UINT64_C(0x00000000008e5d8f), 0.530, 0.556115145800000) \
+    STEP( 107, UINT64_C(0x000000000090bec5), 0.535, 0.565410940131250) \
+    STEP( 108, UINT64_C(0x0000000000931e44), 0.540, 0.574680614400000) \
+    STEP( 109, UINT64_C(0x0000000000957bd0), 0.545, 0.583920482168750) \
+    STEP( 110, UINT64_C(0x000000000097d729), 0.550, 0.593126875000000) \
+    STEP( 111, UINT64_C(0x00000000009a3014), 0.555, 0.602296144706250) \
+    STEP( 112, UINT64_C(0x00000000009c8653), 0.560, 0.611424665600000) \
+    STEP( 113, UINT64_C(0x00000000009ed9aa), 0.565, 0.620508836743750) \
+    STEP( 114, UINT64_C(0x0000000000a129dd), 0.570, 0.629545084200000) \
+    STEP( 115, UINT64_C(0x0000000000a376b1), 0.575, 0.638529863281250) \
+    STEP( 116, UINT64_C(0x0000000000a5bfea), 0.580, 0.647459660800000) \
+    STEP( 117, UINT64_C(0x0000000000a8054e), 0.585, 0.656330997318750) \
+    STEP( 118, UINT64_C(0x0000000000aa46a4), 0.590, 0.665140429400000) \
+    STEP( 119, UINT64_C(0x0000000000ac83b2), 0.595, 0.673884551856250) \
+    STEP( 120, UINT64_C(0x0000000000aebc40), 0.600, 0.682560000000000) \
+    STEP( 121, UINT64_C(0x0000000000b0f016), 0.605, 0.691163451893750) \
+    STEP( 122, UINT64_C(0x0000000000b31efd), 0.610, 0.699691630600000) \
+    STEP( 123, UINT64_C(0x0000000000b548bf), 0.615, 0.708141306431250) \
+    STEP( 124, UINT64_C(0x0000000000b76d27), 0.620, 0.716509299200000) \
+    STEP( 125, UINT64_C(0x0000000000b98c00), 0.625, 0.724792480468750) \
+    STEP( 126, UINT64_C(0x0000000000bba516), 0.630, 0.732987775800000) \
+    STEP( 127, UINT64_C(0x0000000000bdb837), 0.635, 0.741092167006250) \
+    STEP( 128, UINT64_C(0x0000000000bfc531), 0.640, 0.749102694400000) \
+    STEP( 129, UINT64_C(0x0000000000c1cbd4), 0.645, 0.757016459043750) \
+    STEP( 130, UINT64_C(0x0000000000c3cbf0), 0.650, 0.764830625000000) \
+    STEP( 131, UINT64_C(0x0000000000c5c557), 0.655, 0.772542421581250) \
+    STEP( 132, UINT64_C(0x0000000000c7b7da), 0.660, 0.780149145600000) \
+    STEP( 133, UINT64_C(0x0000000000c9a34f), 0.665, 0.787648163618750) \
+    STEP( 134, UINT64_C(0x0000000000cb878a), 0.670, 0.795036914200000) \
+    STEP( 135, UINT64_C(0x0000000000cd6460), 0.675, 0.802312910156250) \
+    STEP( 136, UINT64_C(0x0000000000cf39ab), 0.680, 0.809473740800000) \
+    STEP( 137, UINT64_C(0x0000000000d10743), 0.685, 0.816517074193750) \
+    STEP( 138, UINT64_C(0x0000000000d2cd01), 0.690, 0.823440659400000) \
+    STEP( 139, UINT64_C(0x0000000000d48ac2), 0.695, 0.830242328731250) \
+    STEP( 140, UINT64_C(0x0000000000d64063), 0.700, 0.836920000000000) \
+    STEP( 141, UINT64_C(0x0000000000d7edc2), 0.705, 0.843471678768750) \
+    STEP( 142, UINT64_C(0x0000000000d992bf), 0.710, 0.849895460600000) \
+    STEP( 143, UINT64_C(0x0000000000db2f3c), 0.715, 0.856189533306250) \
+    STEP( 144, UINT64_C(0x0000000000dcc31c), 0.720, 0.862352179200000) \
+    STEP( 145, UINT64_C(0x0000000000de4e44), 0.725, 0.868381777343750) \
+    STEP( 146, UINT64_C(0x0000000000dfd09a), 0.730, 0.874276805800000) \
+    STEP( 147, UINT64_C(0x0000000000e14a07), 0.735, 0.880035843881250) \
+    STEP( 148, UINT64_C(0x0000000000e2ba74), 0.740, 0.885657574400000) \
+    STEP( 149, UINT64_C(0x0000000000e421cd), 0.745, 0.891140785918750) \
+    STEP( 150, UINT64_C(0x0000000000e58000), 0.750, 0.896484375000000) \
+    STEP( 151, UINT64_C(0x0000000000e6d4fb), 0.755, 0.901687348456250) \
+    STEP( 152, UINT64_C(0x0000000000e820b0), 0.760, 0.906748825600000) \
+    STEP( 153, UINT64_C(0x0000000000e96313), 0.765, 0.911668040493750) \
+    STEP( 154, UINT64_C(0x0000000000ea9c18), 0.770, 0.916444344200000) \
+    STEP( 155, UINT64_C(0x0000000000ebcbb7), 0.775, 0.921077207031250) \
+    STEP( 156, UINT64_C(0x0000000000ecf1e8), 0.780, 0.925566220800000) \
+    STEP( 157, UINT64_C(0x0000000000ee0ea7), 0.785, 0.929911101068750) \
+    STEP( 158, UINT64_C(0x0000000000ef21f1), 0.790, 0.934111689400000) \
+    STEP( 159, UINT64_C(0x0000000000f02bc6), 0.795, 0.938167955606250) \
+    STEP( 160, UINT64_C(0x0000000000f12c27), 0.800, 0.942080000000000) \
+    STEP( 161, UINT64_C(0x0000000000f22319), 0.805, 0.945848055643750) \
+    STEP( 162, UINT64_C(0x0000000000f310a1), 0.810, 0.949472490600000) \
+    STEP( 163, UINT64_C(0x0000000000f3f4c7), 0.815, 0.952953810181250) \
+    STEP( 164, UINT64_C(0x0000000000f4cf98), 0.820, 0.956292659200000) \
+    STEP( 165, UINT64_C(0x0000000000f5a120), 0.825, 0.959489824218750) \
+    STEP( 166, UINT64_C(0x0000000000f6696e), 0.830, 0.962546235800000) \
+    STEP( 167, UINT64_C(0x0000000000f72894), 0.835, 0.965462970756250) \
+    STEP( 168, UINT64_C(0x0000000000f7dea8), 0.840, 0.968241254400000) \
+    STEP( 169, UINT64_C(0x0000000000f88bc0), 0.845, 0.970882462793750) \
+    STEP( 170, UINT64_C(0x0000000000f92ff6), 0.850, 0.973388125000000) \
+    STEP( 171, UINT64_C(0x0000000000f9cb67), 0.855, 0.975759925331250) \
+    STEP( 172, UINT64_C(0x0000000000fa5e30), 0.860, 0.977999705600000) \
+    STEP( 173, UINT64_C(0x0000000000fae874), 0.865, 0.980109467368750) \
+    STEP( 174, UINT64_C(0x0000000000fb6a57), 0.870, 0.982091374200000) \
+    STEP( 175, UINT64_C(0x0000000000fbe400), 0.875, 0.983947753906250) \
+    STEP( 176, UINT64_C(0x0000000000fc5598), 0.880, 0.985681100800000) \
+    STEP( 177, UINT64_C(0x0000000000fcbf4e), 0.885, 0.987294077943750) \
+    STEP( 178, UINT64_C(0x0000000000fd214f), 0.890, 0.988789519400000) \
+    STEP( 179, UINT64_C(0x0000000000fd7bcf), 0.895, 0.990170432481250) \
+    STEP( 180, UINT64_C(0x0000000000fdcf03), 0.900, 0.991440000000000) \
+    STEP( 181, UINT64_C(0x0000000000fe1b23), 0.905, 0.992601582518750) \
+    STEP( 182, UINT64_C(0x0000000000fe606a), 0.910, 0.993658720600000) \
+    STEP( 183, UINT64_C(0x0000000000fe9f18), 0.915, 0.994615137056250) \
+    STEP( 184, UINT64_C(0x0000000000fed76e), 0.920, 0.995474739200000) \
+    STEP( 185, UINT64_C(0x0000000000ff09b0), 0.925, 0.996241621093750) \
+    STEP( 186, UINT64_C(0x0000000000ff3627), 0.930, 0.996920065800000) \
+    STEP( 187, UINT64_C(0x0000000000ff5d1d), 0.935, 0.997514547631250) \
+    STEP( 188, UINT64_C(0x0000000000ff7ee0), 0.940, 0.998029734400000) \
+    STEP( 189, UINT64_C(0x0000000000ff9bc3), 0.945, 0.998470489668750) \
+    STEP( 190, UINT64_C(0x0000000000ffb419), 0.950, 0.998841875000000) \
+    STEP( 191, UINT64_C(0x0000000000ffc83d), 0.955, 0.999149152206250) \
+    STEP( 192, UINT64_C(0x0000000000ffd888), 0.960, 0.999397785600000) \
+    STEP( 193, UINT64_C(0x0000000000ffe55b), 0.965, 0.999593444243750) \
+    STEP( 194, UINT64_C(0x0000000000ffef17), 0.970, 0.999742004200000) \
+    STEP( 195, UINT64_C(0x0000000000fff623), 0.975, 0.999849550781250) \
+    STEP( 196, UINT64_C(0x0000000000fffae9), 0.980, 0.999922380800000) \
+    STEP( 197, UINT64_C(0x0000000000fffdd6), 0.985, 0.999967004818750) \
+    STEP( 198, UINT64_C(0x0000000000ffff5a), 0.990, 0.999990149400000) \
+    STEP( 199, UINT64_C(0x0000000000ffffeb), 0.995, 0.999998759356250) \
+    STEP( 200, UINT64_C(0x0000000001000000), 1.000, 1.000000000000000) \
+
+#endif /* JEMALLOC_INTERNAL_SMOOTHSTEP_H */
diff --git a/jemalloc/include/jemalloc/internal/smoothstep.sh b/jemalloc/include/jemalloc/internal/smoothstep.sh
new file mode 100755
index 0000000..65de97b
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/smoothstep.sh
@@ -0,0 +1,101 @@
+#!/bin/sh
+#
+# Generate a discrete lookup table for a sigmoid function in the smoothstep
+# family (https://en.wikipedia.org/wiki/Smoothstep), where the lookup table
+# entries correspond to x in [1/nsteps, 2/nsteps, ..., nsteps/nsteps].  Encode
+# the entries using a binary fixed point representation.
+#
+# Usage: smoothstep.sh <variant> <nsteps> <bfp> <xprec> <yprec>
+#
+#        <variant> is in {smooth, smoother, smoothest}.
+#        <nsteps> must be greater than zero.
+#        <bfp> must be in [0..62]; reasonable values are roughly [10..30].
+#        <xprec> is x decimal precision.
+#        <yprec> is y decimal precision.
+
+#set -x
+
+cmd="sh smoothstep.sh $*"
+variant=$1
+nsteps=$2
+bfp=$3
+xprec=$4
+yprec=$5
+
+case "${variant}" in
+  smooth)
+    ;;
+  smoother)
+    ;;
+  smoothest)
+    ;;
+  *)
+    echo "Unsupported variant"
+    exit 1
+    ;;
+esac
+
+smooth() {
+  step=$1
+  y=`echo ${yprec} k ${step} ${nsteps} / sx _2 lx 3 ^ '*' 3 lx 2 ^ '*' + p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  h=`echo ${yprec} k 2 ${bfp} ^ ${y} '*' p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g' | tr '.' ' ' | awk '{print $1}' `
+}
+
+smoother() {
+  step=$1
+  y=`echo ${yprec} k ${step} ${nsteps} / sx 6 lx 5 ^ '*' _15 lx 4 ^ '*' + 10 lx 3 ^ '*' + p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  h=`echo ${yprec} k 2 ${bfp} ^ ${y} '*' p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g' | tr '.' ' ' | awk '{print $1}' `
+}
+
+smoothest() {
+  step=$1
+  y=`echo ${yprec} k ${step} ${nsteps} / sx _20 lx 7 ^ '*' 70 lx 6 ^ '*' + _84 lx 5 ^ '*' + 35 lx 4 ^ '*' + p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  h=`echo ${yprec} k 2 ${bfp} ^ ${y} '*' p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g' | tr '.' ' ' | awk '{print $1}' `
+}
+
+cat <<EOF
+#ifndef JEMALLOC_INTERNAL_SMOOTHSTEP_H
+#define JEMALLOC_INTERNAL_SMOOTHSTEP_H
+
+/*
+ * This file was generated by the following command:
+ *   $cmd
+ */
+/******************************************************************************/
+
+/*
+ * This header defines a precomputed table based on the smoothstep family of
+ * sigmoidal curves (https://en.wikipedia.org/wiki/Smoothstep) that grow from 0
+ * to 1 in 0 <= x <= 1.  The table is stored as integer fixed point values so
+ * that floating point math can be avoided.
+ *
+ *                      3     2
+ *   smoothstep(x) = -2x  + 3x
+ *
+ *                       5      4      3
+ *   smootherstep(x) = 6x  - 15x  + 10x
+ *
+ *                          7      6      5      4
+ *   smootheststep(x) = -20x  + 70x  - 84x  + 35x
+ */
+
+#define SMOOTHSTEP_VARIANT	"${variant}"
+#define SMOOTHSTEP_NSTEPS	${nsteps}
+#define SMOOTHSTEP_BFP		${bfp}
+#define SMOOTHSTEP \\
+ /* STEP(step, h,                            x,     y) */ \\
+EOF
+
+s=1
+while [ $s -le $nsteps ] ; do
+  $variant ${s}
+  x=`echo ${xprec} k ${s} ${nsteps} / p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  printf '    STEP(%4d, UINT64_C(0x%016x), %s, %s) \\\n' ${s} ${h} ${x} ${y}
+
+  s=$((s+1))
+done
+echo
+
+cat <<EOF
+#endif /* JEMALLOC_INTERNAL_SMOOTHSTEP_H */
+EOF
diff --git a/jemalloc/include/jemalloc/internal/spin.h b/jemalloc/include/jemalloc/internal/spin.h
new file mode 100644
index 0000000..22804c6
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/spin.h
@@ -0,0 +1,40 @@
+#ifndef JEMALLOC_INTERNAL_SPIN_H
+#define JEMALLOC_INTERNAL_SPIN_H
+
+#define SPIN_INITIALIZER {0U}
+
+typedef struct {
+	unsigned iteration;
+} spin_t;
+
+static inline void
+spin_cpu_spinwait() {
+#  if HAVE_CPU_SPINWAIT
+	CPU_SPINWAIT;
+#  else
+	volatile int x = 0;
+	x = x;
+#  endif
+}
+
+static inline void
+spin_adaptive(spin_t *spin) {
+	volatile uint32_t i;
+
+	if (spin->iteration < 5) {
+		for (i = 0; i < (1U << spin->iteration); i++) {
+			spin_cpu_spinwait();
+		}
+		spin->iteration++;
+	} else {
+#ifdef _WIN32
+		SwitchToThread();
+#else
+		sched_yield();
+#endif
+	}
+}
+
+#undef SPIN_INLINE
+
+#endif /* JEMALLOC_INTERNAL_SPIN_H */
diff --git a/jemalloc/include/jemalloc/internal/stats.h b/jemalloc/include/jemalloc/internal/stats.h
new file mode 100644
index 0000000..3b9e0ea
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/stats.h
@@ -0,0 +1,31 @@
+#ifndef JEMALLOC_INTERNAL_STATS_H
+#define JEMALLOC_INTERNAL_STATS_H
+
+/*  OPTION(opt,		var_name,	default,	set_value_to) */
+#define STATS_PRINT_OPTIONS						\
+    OPTION('J',		json,		false,		true)		\
+    OPTION('g',		general,	true,		false)		\
+    OPTION('m',		merged,		config_stats,	false)		\
+    OPTION('d',		destroyed,	config_stats,	false)		\
+    OPTION('a',		unmerged,	config_stats,	false)		\
+    OPTION('b',		bins,		true,		false)		\
+    OPTION('l',		large,		true,		false)		\
+    OPTION('x',		mutex,		true,		false)		\
+    OPTION('e',		extents,	true,		false)
+
+enum {
+#define OPTION(o, v, d, s) stats_print_option_num_##v,
+    STATS_PRINT_OPTIONS
+#undef OPTION
+    stats_print_tot_num_options
+};
+
+/* Options for stats_print. */
+extern bool opt_stats_print;
+extern char opt_stats_print_opts[stats_print_tot_num_options+1];
+
+/* Implements je_malloc_stats_print. */
+void stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts);
+
+#endif /* JEMALLOC_INTERNAL_STATS_H */
diff --git a/jemalloc/include/jemalloc/internal/sz.h b/jemalloc/include/jemalloc/internal/sz.h
new file mode 100644
index 0000000..68e558a
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/sz.h
@@ -0,0 +1,318 @@
+#ifndef JEMALLOC_INTERNAL_SIZE_H
+#define JEMALLOC_INTERNAL_SIZE_H
+
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/util.h"
+
+/*
+ * sz module: Size computations.
+ *
+ * Some abbreviations used here:
+ *   p: Page
+ *   ind: Index
+ *   s, sz: Size
+ *   u: Usable size
+ *   a: Aligned
+ *
+ * These are not always used completely consistently, but should be enough to
+ * interpret function names.  E.g. sz_psz2ind converts page size to page size
+ * index; sz_sa2u converts a (size, alignment) allocation request to the usable
+ * size that would result from such an allocation.
+ */
+
+/*
+ * sz_pind2sz_tab encodes the same information as could be computed by
+ * sz_pind2sz_compute().
+ */
+extern size_t sz_pind2sz_tab[SC_NPSIZES + 1];
+/*
+ * sz_index2size_tab encodes the same information as could be computed (at
+ * unacceptable cost in some code paths) by sz_index2size_compute().
+ */
+extern size_t sz_index2size_tab[SC_NSIZES];
+/*
+ * sz_size2index_tab is a compact lookup table that rounds request sizes up to
+ * size classes.  In order to reduce cache footprint, the table is compressed,
+ * and all accesses are via sz_size2index().
+ */
+extern uint8_t sz_size2index_tab[];
+
+static const size_t sz_large_pad =
+#ifdef JEMALLOC_CACHE_OBLIVIOUS
+    PAGE
+#else
+    0
+#endif
+    ;
+
+extern void sz_boot(const sc_data_t *sc_data);
+
+JEMALLOC_ALWAYS_INLINE pszind_t
+sz_psz2ind(size_t psz) {
+	if (unlikely(psz > SC_LARGE_MAXCLASS)) {
+		return SC_NPSIZES;
+	}
+	pszind_t x = lg_floor((psz<<1)-1);
+	pszind_t shift = (x < SC_LG_NGROUP + LG_PAGE) ?
+	    0 : x - (SC_LG_NGROUP + LG_PAGE);
+	pszind_t grp = shift << SC_LG_NGROUP;
+
+	pszind_t lg_delta = (x < SC_LG_NGROUP + LG_PAGE + 1) ?
+	    LG_PAGE : x - SC_LG_NGROUP - 1;
+
+	size_t delta_inverse_mask = ZU(-1) << lg_delta;
+	pszind_t mod = ((((psz-1) & delta_inverse_mask) >> lg_delta)) &
+	    ((ZU(1) << SC_LG_NGROUP) - 1);
+
+	pszind_t ind = grp + mod;
+	return ind;
+}
+
+static inline size_t
+sz_pind2sz_compute(pszind_t pind) {
+	if (unlikely(pind == SC_NPSIZES)) {
+		return SC_LARGE_MAXCLASS + PAGE;
+	}
+	size_t grp = pind >> SC_LG_NGROUP;
+	size_t mod = pind & ((ZU(1) << SC_LG_NGROUP) - 1);
+
+	size_t grp_size_mask = ~((!!grp)-1);
+	size_t grp_size = ((ZU(1) << (LG_PAGE + (SC_LG_NGROUP-1))) << grp)
+	    & grp_size_mask;
+
+	size_t shift = (grp == 0) ? 1 : grp;
+	size_t lg_delta = shift + (LG_PAGE-1);
+	size_t mod_size = (mod+1) << lg_delta;
+
+	size_t sz = grp_size + mod_size;
+	return sz;
+}
+
+static inline size_t
+sz_pind2sz_lookup(pszind_t pind) {
+	size_t ret = (size_t)sz_pind2sz_tab[pind];
+	assert(ret == sz_pind2sz_compute(pind));
+	return ret;
+}
+
+static inline size_t
+sz_pind2sz(pszind_t pind) {
+	assert(pind < SC_NPSIZES + 1);
+	return sz_pind2sz_lookup(pind);
+}
+
+static inline size_t
+sz_psz2u(size_t psz) {
+	if (unlikely(psz > SC_LARGE_MAXCLASS)) {
+		return SC_LARGE_MAXCLASS + PAGE;
+	}
+	size_t x = lg_floor((psz<<1)-1);
+	size_t lg_delta = (x < SC_LG_NGROUP + LG_PAGE + 1) ?
+	    LG_PAGE : x - SC_LG_NGROUP - 1;
+	size_t delta = ZU(1) << lg_delta;
+	size_t delta_mask = delta - 1;
+	size_t usize = (psz + delta_mask) & ~delta_mask;
+	return usize;
+}
+
+static inline szind_t
+sz_size2index_compute(size_t size) {
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		return SC_NSIZES;
+	}
+
+	if (size == 0) {
+		return 0;
+	}
+#if (SC_NTINY != 0)
+	if (size <= (ZU(1) << SC_LG_TINY_MAXCLASS)) {
+		szind_t lg_tmin = SC_LG_TINY_MAXCLASS - SC_NTINY + 1;
+		szind_t lg_ceil = lg_floor(pow2_ceil_zu(size));
+		return (lg_ceil < lg_tmin ? 0 : lg_ceil - lg_tmin);
+	}
+#endif
+	{
+		szind_t x = lg_floor((size<<1)-1);
+		szind_t shift = (x < SC_LG_NGROUP + LG_QUANTUM) ? 0 :
+		    x - (SC_LG_NGROUP + LG_QUANTUM);
+		szind_t grp = shift << SC_LG_NGROUP;
+
+		szind_t lg_delta = (x < SC_LG_NGROUP + LG_QUANTUM + 1)
+		    ? LG_QUANTUM : x - SC_LG_NGROUP - 1;
+
+		size_t delta_inverse_mask = ZU(-1) << lg_delta;
+		szind_t mod = ((((size-1) & delta_inverse_mask) >> lg_delta)) &
+		    ((ZU(1) << SC_LG_NGROUP) - 1);
+
+		szind_t index = SC_NTINY + grp + mod;
+		return index;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+sz_size2index_lookup(size_t size) {
+	assert(size <= SC_LOOKUP_MAXCLASS);
+	szind_t ret = (sz_size2index_tab[(size + (ZU(1) << SC_LG_TINY_MIN) - 1)
+					 >> SC_LG_TINY_MIN]);
+	assert(ret == sz_size2index_compute(size));
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+sz_size2index(size_t size) {
+	if (likely(size <= SC_LOOKUP_MAXCLASS)) {
+		return sz_size2index_lookup(size);
+	}
+	return sz_size2index_compute(size);
+}
+
+static inline size_t
+sz_index2size_compute(szind_t index) {
+#if (SC_NTINY > 0)
+	if (index < SC_NTINY) {
+		return (ZU(1) << (SC_LG_TINY_MAXCLASS - SC_NTINY + 1 + index));
+	}
+#endif
+	{
+		size_t reduced_index = index - SC_NTINY;
+		size_t grp = reduced_index >> SC_LG_NGROUP;
+		size_t mod = reduced_index & ((ZU(1) << SC_LG_NGROUP) -
+		    1);
+
+		size_t grp_size_mask = ~((!!grp)-1);
+		size_t grp_size = ((ZU(1) << (LG_QUANTUM +
+		    (SC_LG_NGROUP-1))) << grp) & grp_size_mask;
+
+		size_t shift = (grp == 0) ? 1 : grp;
+		size_t lg_delta = shift + (LG_QUANTUM-1);
+		size_t mod_size = (mod+1) << lg_delta;
+
+		size_t usize = grp_size + mod_size;
+		return usize;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_index2size_lookup(szind_t index) {
+	size_t ret = (size_t)sz_index2size_tab[index];
+	assert(ret == sz_index2size_compute(index));
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_index2size(szind_t index) {
+	assert(index < SC_NSIZES);
+	return sz_index2size_lookup(index);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u_compute(size_t size) {
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		return 0;
+	}
+
+	if (size == 0) {
+		size++;
+	}
+#if (SC_NTINY > 0)
+	if (size <= (ZU(1) << SC_LG_TINY_MAXCLASS)) {
+		size_t lg_tmin = SC_LG_TINY_MAXCLASS - SC_NTINY + 1;
+		size_t lg_ceil = lg_floor(pow2_ceil_zu(size));
+		return (lg_ceil < lg_tmin ? (ZU(1) << lg_tmin) :
+		    (ZU(1) << lg_ceil));
+	}
+#endif
+	{
+		size_t x = lg_floor((size<<1)-1);
+		size_t lg_delta = (x < SC_LG_NGROUP + LG_QUANTUM + 1)
+		    ?  LG_QUANTUM : x - SC_LG_NGROUP - 1;
+		size_t delta = ZU(1) << lg_delta;
+		size_t delta_mask = delta - 1;
+		size_t usize = (size + delta_mask) & ~delta_mask;
+		return usize;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u_lookup(size_t size) {
+	size_t ret = sz_index2size_lookup(sz_size2index_lookup(size));
+
+	assert(ret == sz_s2u_compute(size));
+	return ret;
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u(size_t size) {
+	if (likely(size <= SC_LOOKUP_MAXCLASS)) {
+		return sz_s2u_lookup(size);
+	}
+	return sz_s2u_compute(size);
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size and alignment.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+sz_sa2u(size_t size, size_t alignment) {
+	size_t usize;
+
+	assert(alignment != 0 && ((alignment - 1) & alignment) == 0);
+
+	/* Try for a small size class. */
+	if (size <= SC_SMALL_MAXCLASS && alignment < PAGE) {
+		/*
+		 * Round size up to the nearest multiple of alignment.
+		 *
+		 * This done, we can take advantage of the fact that for each
+		 * small size class, every object is aligned at the smallest
+		 * power of two that is non-zero in the base two representation
+		 * of the size.  For example:
+		 *
+		 *   Size |   Base 2 | Minimum alignment
+		 *   -----+----------+------------------
+		 *     96 |  1100000 |  32
+		 *    144 | 10100000 |  32
+		 *    192 | 11000000 |  64
+		 */
+		usize = sz_s2u(ALIGNMENT_CEILING(size, alignment));
+		if (usize < SC_LARGE_MINCLASS) {
+			return usize;
+		}
+	}
+
+	/* Large size class.  Beware of overflow. */
+
+	if (unlikely(alignment > SC_LARGE_MAXCLASS)) {
+		return 0;
+	}
+
+	/* Make sure result is a large size class. */
+	if (size <= SC_LARGE_MINCLASS) {
+		usize = SC_LARGE_MINCLASS;
+	} else {
+		usize = sz_s2u(size);
+		if (usize < size) {
+			/* size_t overflow. */
+			return 0;
+		}
+	}
+
+	/*
+	 * Calculate the multi-page mapping that large_palloc() would need in
+	 * order to guarantee the alignment.
+	 */
+	if (usize + sz_large_pad + PAGE_CEILING(alignment) - PAGE < usize) {
+		/* size_t overflow. */
+		return 0;
+	}
+	return usize;
+}
+
+#endif /* JEMALLOC_INTERNAL_SIZE_H */
diff --git a/jemalloc/include/jemalloc/internal/tcache_externs.h b/jemalloc/include/jemalloc/internal/tcache_externs.h
new file mode 100644
index 0000000..d63eafd
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tcache_externs.h
@@ -0,0 +1,53 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_EXTERNS_H
+#define JEMALLOC_INTERNAL_TCACHE_EXTERNS_H
+
+extern bool	opt_tcache;
+extern ssize_t	opt_lg_tcache_max;
+
+extern cache_bin_info_t	*tcache_bin_info;
+
+/*
+ * Number of tcache bins.  There are SC_NBINS small-object bins, plus 0 or more
+ * large-object bins.
+ */
+extern unsigned	nhbins;
+
+/* Maximum cached size class. */
+extern size_t	tcache_maxclass;
+
+/*
+ * Explicit tcaches, managed via the tcache.{create,flush,destroy} mallctls and
+ * usable via the MALLOCX_TCACHE() flag.  The automatic per thread tcaches are
+ * completely disjoint from this data structure.  tcaches starts off as a sparse
+ * array, so it has no physical memory footprint until individual pages are
+ * touched.  This allows the entire array to be allocated the first time an
+ * explicit tcache is created without a disproportionate impact on memory usage.
+ */
+extern tcaches_t	*tcaches;
+
+size_t	tcache_salloc(tsdn_t *tsdn, const void *ptr);
+void	tcache_event_hard(tsd_t *tsd, tcache_t *tcache);
+void	*tcache_alloc_small_hard(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, bool *tcache_success);
+void	tcache_bin_flush_small(tsd_t *tsd, tcache_t *tcache, cache_bin_t *tbin,
+    szind_t binind, unsigned rem);
+void	tcache_bin_flush_large(tsd_t *tsd, cache_bin_t *tbin, szind_t binind,
+    unsigned rem, tcache_t *tcache);
+void	tcache_arena_reassociate(tsdn_t *tsdn, tcache_t *tcache,
+    arena_t *arena);
+tcache_t *tcache_create_explicit(tsd_t *tsd);
+void	tcache_cleanup(tsd_t *tsd);
+void	tcache_stats_merge(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena);
+bool	tcaches_create(tsd_t *tsd, unsigned *r_ind);
+void	tcaches_flush(tsd_t *tsd, unsigned ind);
+void	tcaches_destroy(tsd_t *tsd, unsigned ind);
+bool	tcache_boot(tsdn_t *tsdn);
+void tcache_arena_associate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena);
+void tcache_prefork(tsdn_t *tsdn);
+void tcache_postfork_parent(tsdn_t *tsdn);
+void tcache_postfork_child(tsdn_t *tsdn);
+void tcache_flush(tsd_t *tsd);
+bool tsd_tcache_data_init(tsd_t *tsd);
+bool tsd_tcache_enabled_data_init(tsd_t *tsd);
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_EXTERNS_H */
diff --git a/jemalloc/include/jemalloc/internal/tcache_inlines.h b/jemalloc/include/jemalloc/internal/tcache_inlines.h
new file mode 100644
index 0000000..5eca20e
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tcache_inlines.h
@@ -0,0 +1,227 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_INLINES_H
+#define JEMALLOC_INTERNAL_TCACHE_INLINES_H
+
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/ticker.h"
+#include "jemalloc/internal/util.h"
+
+static inline bool
+tcache_enabled_get(tsd_t *tsd) {
+	return tsd_tcache_enabled_get(tsd);
+}
+
+static inline void
+tcache_enabled_set(tsd_t *tsd, bool enabled) {
+	bool was_enabled = tsd_tcache_enabled_get(tsd);
+
+	if (!was_enabled && enabled) {
+		tsd_tcache_data_init(tsd);
+	} else if (was_enabled && !enabled) {
+		tcache_cleanup(tsd);
+	}
+	/* Commit the state last.  Above calls check current state. */
+	tsd_tcache_enabled_set(tsd, enabled);
+	tsd_slow_update(tsd);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_event(tsd_t *tsd, tcache_t *tcache) {
+	if (TCACHE_GC_INCR == 0) {
+		return;
+	}
+
+	if (unlikely(ticker_tick(&tcache->gc_ticker))) {
+		tcache_event_hard(tsd, tcache);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+tcache_alloc_small(tsd_t *tsd, arena_t *arena, tcache_t *tcache,
+    size_t size, szind_t binind, bool zero, bool slow_path) {
+	void *ret;
+	cache_bin_t *bin;
+	bool tcache_success;
+	size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+
+	assert(binind < SC_NBINS);
+	bin = tcache_small_bin_get(tcache, binind);
+	ret = cache_bin_alloc_easy(bin, &tcache_success);
+	assert(tcache_success == (ret != NULL));
+	if (unlikely(!tcache_success)) {
+		bool tcache_hard_success;
+		arena = arena_choose(tsd, arena);
+		if (unlikely(arena == NULL)) {
+			return NULL;
+		}
+
+		ret = tcache_alloc_small_hard(tsd_tsdn(tsd), arena, tcache,
+		    bin, binind, &tcache_hard_success);
+		if (tcache_hard_success == false) {
+			return NULL;
+		}
+	}
+
+	assert(ret);
+	/*
+	 * Only compute usize if required.  The checks in the following if
+	 * statement are all static.
+	 */
+	if (config_prof || (slow_path && config_fill) || unlikely(zero)) {
+		usize = sz_index2size(binind);
+		assert(tcache_salloc(tsd_tsdn(tsd), ret) == usize);
+	}
+
+	if (likely(!zero)) {
+		if (slow_path && config_fill) {
+			if (unlikely(opt_junk_alloc)) {
+				arena_alloc_junk_small(ret, &bin_infos[binind],
+				    false);
+			} else if (unlikely(opt_zero)) {
+				memset(ret, 0, usize);
+			}
+		}
+	} else {
+		if (slow_path && config_fill && unlikely(opt_junk_alloc)) {
+			arena_alloc_junk_small(ret, &bin_infos[binind], true);
+		}
+		memset(ret, 0, usize);
+	}
+
+	if (config_stats) {
+		bin->tstats.nrequests++;
+	}
+	if (config_prof) {
+		tcache->prof_accumbytes += usize;
+	}
+	tcache_event(tsd, tcache);
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+tcache_alloc_large(tsd_t *tsd, arena_t *arena, tcache_t *tcache, size_t size,
+    szind_t binind, bool zero, bool slow_path) {
+	void *ret;
+	cache_bin_t *bin;
+	bool tcache_success;
+
+	assert(binind >= SC_NBINS &&binind < nhbins);
+	bin = tcache_large_bin_get(tcache, binind);
+	ret = cache_bin_alloc_easy(bin, &tcache_success);
+	assert(tcache_success == (ret != NULL));
+	if (unlikely(!tcache_success)) {
+		/*
+		 * Only allocate one large object at a time, because it's quite
+		 * expensive to create one and not use it.
+		 */
+		arena = arena_choose(tsd, arena);
+		if (unlikely(arena == NULL)) {
+			return NULL;
+		}
+
+		ret = large_malloc(tsd_tsdn(tsd), arena, sz_s2u(size), zero);
+		if (ret == NULL) {
+			return NULL;
+		}
+	} else {
+		size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+
+		/* Only compute usize on demand */
+		if (config_prof || (slow_path && config_fill) ||
+		    unlikely(zero)) {
+			usize = sz_index2size(binind);
+			assert(usize <= tcache_maxclass);
+		}
+
+		if (likely(!zero)) {
+			if (slow_path && config_fill) {
+				if (unlikely(opt_junk_alloc)) {
+					memset(ret, JEMALLOC_ALLOC_JUNK,
+					    usize);
+				} else if (unlikely(opt_zero)) {
+					memset(ret, 0, usize);
+				}
+			}
+		} else {
+			memset(ret, 0, usize);
+		}
+
+		if (config_stats) {
+			bin->tstats.nrequests++;
+		}
+		if (config_prof) {
+			tcache->prof_accumbytes += usize;
+		}
+	}
+
+	tcache_event(tsd, tcache);
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_dalloc_small(tsd_t *tsd, tcache_t *tcache, void *ptr, szind_t binind,
+    bool slow_path) {
+	cache_bin_t *bin;
+	cache_bin_info_t *bin_info;
+
+	assert(tcache_salloc(tsd_tsdn(tsd), ptr)
+	    <= SC_SMALL_MAXCLASS);
+
+	if (slow_path && config_fill && unlikely(opt_junk_free)) {
+		arena_dalloc_junk_small(ptr, &bin_infos[binind]);
+	}
+
+	bin = tcache_small_bin_get(tcache, binind);
+	bin_info = &tcache_bin_info[binind];
+	if (unlikely(!cache_bin_dalloc_easy(bin, bin_info, ptr))) {
+		tcache_bin_flush_small(tsd, tcache, bin, binind,
+		    (bin_info->ncached_max >> 1));
+		bool ret = cache_bin_dalloc_easy(bin, bin_info, ptr);
+		assert(ret);
+	}
+
+	tcache_event(tsd, tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_dalloc_large(tsd_t *tsd, tcache_t *tcache, void *ptr, szind_t binind,
+    bool slow_path) {
+	cache_bin_t *bin;
+	cache_bin_info_t *bin_info;
+
+	assert(tcache_salloc(tsd_tsdn(tsd), ptr)
+	    > SC_SMALL_MAXCLASS);
+	assert(tcache_salloc(tsd_tsdn(tsd), ptr) <= tcache_maxclass);
+
+	if (slow_path && config_fill && unlikely(opt_junk_free)) {
+		large_dalloc_junk(ptr, sz_index2size(binind));
+	}
+
+	bin = tcache_large_bin_get(tcache, binind);
+	bin_info = &tcache_bin_info[binind];
+	if (unlikely(bin->ncached == bin_info->ncached_max)) {
+		tcache_bin_flush_large(tsd, bin, binind,
+		    (bin_info->ncached_max >> 1), tcache);
+	}
+	assert(bin->ncached < bin_info->ncached_max);
+	bin->ncached++;
+	*(bin->avail - bin->ncached) = ptr;
+
+	tcache_event(tsd, tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE tcache_t *
+tcaches_get(tsd_t *tsd, unsigned ind) {
+	tcaches_t *elm = &tcaches[ind];
+	if (unlikely(elm->tcache == NULL)) {
+		malloc_printf("<jemalloc>: invalid tcache id (%u).\n", ind);
+		abort();
+	} else if (unlikely(elm->tcache == TCACHES_ELM_NEED_REINIT)) {
+		elm->tcache = tcache_create_explicit(tsd);
+	}
+	return elm->tcache;
+}
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_INLINES_H */
diff --git a/jemalloc/include/jemalloc/internal/tcache_structs.h b/jemalloc/include/jemalloc/internal/tcache_structs.h
new file mode 100644
index 0000000..172ef90
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tcache_structs.h
@@ -0,0 +1,70 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_STRUCTS_H
+#define JEMALLOC_INTERNAL_TCACHE_STRUCTS_H
+
+#include "jemalloc/internal/cache_bin.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/ticker.h"
+#include "jemalloc/internal/tsd_types.h"
+
+/* Various uses of this struct need it to be a named type. */
+typedef ql_elm(tsd_t) tsd_link_t;
+
+struct tcache_s {
+	/*
+	 * To minimize our cache-footprint, we put the frequently accessed data
+	 * together at the start of this struct.
+	 */
+
+	/* Cleared after arena_prof_accum(). */
+	uint64_t	prof_accumbytes;
+	/* Drives incremental GC. */
+	ticker_t	gc_ticker;
+	/*
+	 * The pointer stacks associated with bins follow as a contiguous array.
+	 * During tcache initialization, the avail pointer in each element of
+	 * tbins is initialized to point to the proper offset within this array.
+	 */
+	cache_bin_t	bins_small[SC_NBINS];
+
+	/*
+	 * This data is less hot; we can be a little less careful with our
+	 * footprint here.
+	 */
+	/* Lets us track all the tcaches in an arena. */
+	ql_elm(tcache_t) link;
+
+	/* Logically scoped to tsd, but put here for cache layout reasons. */
+	ql_elm(tsd_t) tsd_link;
+	bool in_hook;
+
+	/*
+	 * The descriptor lets the arena find our cache bins without seeing the
+	 * tcache definition.  This enables arenas to aggregate stats across
+	 * tcaches without having a tcache dependency.
+	 */
+	cache_bin_array_descriptor_t cache_bin_array_descriptor;
+
+	/* The arena this tcache is associated with. */
+	arena_t		*arena;
+	/* Next bin to GC. */
+	szind_t		next_gc_bin;
+	/* For small bins, fill (ncached_max >> lg_fill_div). */
+	uint8_t		lg_fill_div[SC_NBINS];
+	/*
+	 * We put the cache bins for large size classes at the end of the
+	 * struct, since some of them might not get used.  This might end up
+	 * letting us avoid touching an extra page if we don't have to.
+	 */
+	cache_bin_t	bins_large[SC_NSIZES-SC_NBINS];
+};
+
+/* Linkage for list of available (previously used) explicit tcache IDs. */
+struct tcaches_s {
+	union {
+		tcache_t	*tcache;
+		tcaches_t	*next;
+	};
+};
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_STRUCTS_H */
diff --git a/jemalloc/include/jemalloc/internal/tcache_types.h b/jemalloc/include/jemalloc/internal/tcache_types.h
new file mode 100644
index 0000000..dce6938
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tcache_types.h
@@ -0,0 +1,59 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_TYPES_H
+#define JEMALLOC_INTERNAL_TCACHE_TYPES_H
+
+#include "jemalloc/internal/sc.h"
+
+typedef struct tcache_s tcache_t;
+typedef struct tcaches_s tcaches_t;
+
+/*
+ * tcache pointers close to NULL are used to encode state information that is
+ * used for two purposes: preventing thread caching on a per thread basis and
+ * cleaning up during thread shutdown.
+ */
+#define TCACHE_STATE_DISABLED		((tcache_t *)(uintptr_t)1)
+#define TCACHE_STATE_REINCARNATED	((tcache_t *)(uintptr_t)2)
+#define TCACHE_STATE_PURGATORY		((tcache_t *)(uintptr_t)3)
+#define TCACHE_STATE_MAX		TCACHE_STATE_PURGATORY
+
+/*
+ * Absolute minimum number of cache slots for each small bin.
+ */
+#define TCACHE_NSLOTS_SMALL_MIN		20
+
+/*
+ * Absolute maximum number of cache slots for each small bin in the thread
+ * cache.  This is an additional constraint beyond that imposed as: twice the
+ * number of regions per slab for this size class.
+ *
+ * This constant must be an even number.
+ */
+#define TCACHE_NSLOTS_SMALL_MAX		200
+
+/* Number of cache slots for large size classes. */
+#define TCACHE_NSLOTS_LARGE		20
+
+/* (1U << opt_lg_tcache_max) is used to compute tcache_maxclass. */
+#define LG_TCACHE_MAXCLASS_DEFAULT	15
+
+/*
+ * TCACHE_GC_SWEEP is the approximate number of allocation events between
+ * full GC sweeps.  Integer rounding may cause the actual number to be
+ * slightly higher, since GC is performed incrementally.
+ */
+#define TCACHE_GC_SWEEP			8192
+
+/* Number of tcache allocation/deallocation events between incremental GCs. */
+#define TCACHE_GC_INCR							\
+    ((TCACHE_GC_SWEEP / SC_NBINS) + ((TCACHE_GC_SWEEP / SC_NBINS == 0) ? 0 : 1))
+
+/* Used in TSD static initializer only. Real init in tcache_data_init(). */
+#define TCACHE_ZERO_INITIALIZER {0}
+
+/* Used in TSD static initializer only. Will be initialized to opt_tcache. */
+#define TCACHE_ENABLED_ZERO_INITIALIZER false
+
+/* Used for explicit tcache only. Means flushed but not destroyed. */
+#define TCACHES_ELM_NEED_REINIT ((tcache_t *)(uintptr_t)1)
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/test_hooks.h b/jemalloc/include/jemalloc/internal/test_hooks.h
new file mode 100644
index 0000000..a6351e5
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/test_hooks.h
@@ -0,0 +1,19 @@
+#ifndef JEMALLOC_INTERNAL_TEST_HOOKS_H
+#define JEMALLOC_INTERNAL_TEST_HOOKS_H
+
+extern JEMALLOC_EXPORT void (*test_hooks_arena_new_hook)();
+extern JEMALLOC_EXPORT void (*test_hooks_libc_hook)();
+
+#define JEMALLOC_HOOK(fn, hook) ((void)(hook != NULL && (hook(), 0)), fn)
+
+#define open JEMALLOC_HOOK(open, test_hooks_libc_hook)
+#define read JEMALLOC_HOOK(read, test_hooks_libc_hook)
+#define write JEMALLOC_HOOK(write, test_hooks_libc_hook)
+#define readlink JEMALLOC_HOOK(readlink, test_hooks_libc_hook)
+#define close JEMALLOC_HOOK(close, test_hooks_libc_hook)
+#define creat JEMALLOC_HOOK(creat, test_hooks_libc_hook)
+#define secure_getenv JEMALLOC_HOOK(secure_getenv, test_hooks_libc_hook)
+/* Note that this is undef'd and re-define'd in src/prof.c. */
+#define _Unwind_Backtrace JEMALLOC_HOOK(_Unwind_Backtrace, test_hooks_libc_hook)
+
+#endif /* JEMALLOC_INTERNAL_TEST_HOOKS_H */
diff --git a/jemalloc/include/jemalloc/internal/ticker.h b/jemalloc/include/jemalloc/internal/ticker.h
new file mode 100644
index 0000000..0357800
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/ticker.h
@@ -0,0 +1,91 @@
+#ifndef JEMALLOC_INTERNAL_TICKER_H
+#define JEMALLOC_INTERNAL_TICKER_H
+
+#include "jemalloc/internal/util.h"
+
+/**
+ * A ticker makes it easy to count-down events until some limit.  You
+ * ticker_init the ticker to trigger every nticks events.  You then notify it
+ * that an event has occurred with calls to ticker_tick (or that nticks events
+ * have occurred with a call to ticker_ticks), which will return true (and reset
+ * the counter) if the countdown hit zero.
+ */
+
+typedef struct {
+	int32_t tick;
+	int32_t nticks;
+} ticker_t;
+
+static inline void
+ticker_init(ticker_t *ticker, int32_t nticks) {
+	ticker->tick = nticks;
+	ticker->nticks = nticks;
+}
+
+static inline void
+ticker_copy(ticker_t *ticker, const ticker_t *other) {
+	*ticker = *other;
+}
+
+static inline int32_t
+ticker_read(const ticker_t *ticker) {
+	return ticker->tick;
+}
+
+/*
+ * Not intended to be a public API.  Unfortunately, on x86, neither gcc nor
+ * clang seems smart enough to turn
+ *   ticker->tick -= nticks;
+ *   if (unlikely(ticker->tick < 0)) {
+ *     fixup ticker
+ *     return true;
+ *   }
+ *   return false;
+ * into
+ *   subq %nticks_reg, (%ticker_reg)
+ *   js fixup ticker
+ *
+ * unless we force "fixup ticker" out of line.  In that case, gcc gets it right,
+ * but clang now does worse than before.  So, on x86 with gcc, we force it out
+ * of line, but otherwise let the inlining occur.  Ordinarily this wouldn't be
+ * worth the hassle, but this is on the fast path of both malloc and free (via
+ * tcache_event).
+ */
+#if defined(__GNUC__) && !defined(__clang__)				\
+    && (defined(__x86_64__) || defined(__i386__))
+JEMALLOC_NOINLINE
+#endif
+static bool
+ticker_fixup(ticker_t *ticker) {
+	ticker->tick = ticker->nticks;
+	return true;
+}
+
+static inline bool
+ticker_ticks(ticker_t *ticker, int32_t nticks) {
+	ticker->tick -= nticks;
+	if (unlikely(ticker->tick < 0)) {
+		return ticker_fixup(ticker);
+	}
+	return false;
+}
+
+static inline bool
+ticker_tick(ticker_t *ticker) {
+	return ticker_ticks(ticker, 1);
+}
+
+/*
+ * Try to tick.  If ticker would fire, return true, but rely on
+ * slowpath to reset ticker.
+ */
+static inline bool
+ticker_trytick(ticker_t *ticker) {
+	--ticker->tick;
+	if (unlikely(ticker->tick < 0)) {
+		return true;
+	}
+	return false;
+}
+
+#endif /* JEMALLOC_INTERNAL_TICKER_H */
diff --git a/jemalloc/include/jemalloc/internal/tsd.h b/jemalloc/include/jemalloc/internal/tsd.h
new file mode 100644
index 0000000..9ba2600
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tsd.h
@@ -0,0 +1,415 @@
+#ifndef JEMALLOC_INTERNAL_TSD_H
+#define JEMALLOC_INTERNAL_TSD_H
+
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/bin_types.h"
+#include "jemalloc/internal/jemalloc_internal_externs.h"
+#include "jemalloc/internal/prof_types.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/rtree_tsd.h"
+#include "jemalloc/internal/tcache_types.h"
+#include "jemalloc/internal/tcache_structs.h"
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/witness.h"
+
+/*
+ * Thread-Specific-Data layout
+ * --- data accessed on tcache fast path: state, rtree_ctx, stats, prof ---
+ * s: state
+ * e: tcache_enabled
+ * m: thread_allocated (config_stats)
+ * f: thread_deallocated (config_stats)
+ * p: prof_tdata (config_prof)
+ * c: rtree_ctx (rtree cache accessed on deallocation)
+ * t: tcache
+ * --- data not accessed on tcache fast path: arena-related fields ---
+ * d: arenas_tdata_bypass
+ * r: reentrancy_level
+ * x: narenas_tdata
+ * i: iarena
+ * a: arena
+ * o: arenas_tdata
+ * Loading TSD data is on the critical path of basically all malloc operations.
+ * In particular, tcache and rtree_ctx rely on hot CPU cache to be effective.
+ * Use a compact layout to reduce cache footprint.
+ * +--- 64-bit and 64B cacheline; 1B each letter; First byte on the left. ---+
+ * |----------------------------  1st cacheline  ----------------------------|
+ * | sedrxxxx mmmmmmmm ffffffff pppppppp [c * 32  ........ ........ .......] |
+ * |----------------------------  2nd cacheline  ----------------------------|
+ * | [c * 64  ........ ........ ........ ........ ........ ........ .......] |
+ * |----------------------------  3nd cacheline  ----------------------------|
+ * | [c * 32  ........ ........ .......] iiiiiiii aaaaaaaa oooooooo [t...... |
+ * +-------------------------------------------------------------------------+
+ * Note: the entire tcache is embedded into TSD and spans multiple cachelines.
+ *
+ * The last 3 members (i, a and o) before tcache isn't really needed on tcache
+ * fast path.  However we have a number of unused tcache bins and witnesses
+ * (never touched unless config_debug) at the end of tcache, so we place them
+ * there to avoid breaking the cachelines and possibly paging in an extra page.
+ */
+#ifdef JEMALLOC_JET
+typedef void (*test_callback_t)(int *);
+#  define MALLOC_TSD_TEST_DATA_INIT 0x72b65c10
+#  define MALLOC_TEST_TSD \
+    O(test_data,		int,			int)		\
+    O(test_callback,		test_callback_t,	int)
+#  define MALLOC_TEST_TSD_INITIALIZER , MALLOC_TSD_TEST_DATA_INIT, NULL
+#else
+#  define MALLOC_TEST_TSD
+#  define MALLOC_TEST_TSD_INITIALIZER
+#endif
+
+/*  O(name,			type,			nullable type */
+#define MALLOC_TSD							\
+    O(tcache_enabled,		bool,			bool)		\
+    O(arenas_tdata_bypass,	bool,			bool)		\
+    O(reentrancy_level,		int8_t,			int8_t)		\
+    O(narenas_tdata,		uint32_t,		uint32_t)	\
+    O(offset_state,		uint64_t,		uint64_t)	\
+    O(thread_allocated,		uint64_t,		uint64_t)	\
+    O(thread_deallocated,	uint64_t,		uint64_t)	\
+    O(bytes_until_sample,	int64_t,		int64_t)	\
+    O(prof_tdata,		prof_tdata_t *,		prof_tdata_t *)	\
+    O(rtree_ctx,		rtree_ctx_t,		rtree_ctx_t)	\
+    O(iarena,			arena_t *,		arena_t *)	\
+    O(arena,			arena_t *,		arena_t *)	\
+    O(arenas_tdata,		arena_tdata_t *,	arena_tdata_t *)\
+    O(binshards,		tsd_binshards_t,	tsd_binshards_t)\
+    O(tcache,			tcache_t,		tcache_t)	\
+    O(witness_tsd,              witness_tsd_t,		witness_tsdn_t)	\
+    MALLOC_TEST_TSD
+
+#define TSD_INITIALIZER {						\
+    ATOMIC_INIT(tsd_state_uninitialized),				\
+    TCACHE_ENABLED_ZERO_INITIALIZER,					\
+    false,								\
+    0,									\
+    0,									\
+    0,									\
+    0,									\
+    0,									\
+    0,									\
+    NULL,								\
+    RTREE_CTX_ZERO_INITIALIZER,						\
+    NULL,								\
+    NULL,								\
+    NULL,								\
+    TSD_BINSHARDS_ZERO_INITIALIZER,					\
+    TCACHE_ZERO_INITIALIZER,						\
+    WITNESS_TSD_INITIALIZER						\
+    MALLOC_TEST_TSD_INITIALIZER						\
+}
+
+void *malloc_tsd_malloc(size_t size);
+void malloc_tsd_dalloc(void *wrapper);
+void malloc_tsd_cleanup_register(bool (*f)(void));
+tsd_t *malloc_tsd_boot0(void);
+void malloc_tsd_boot1(void);
+void tsd_cleanup(void *arg);
+tsd_t *tsd_fetch_slow(tsd_t *tsd, bool internal);
+void tsd_state_set(tsd_t *tsd, uint8_t new_state);
+void tsd_slow_update(tsd_t *tsd);
+void tsd_prefork(tsd_t *tsd);
+void tsd_postfork_parent(tsd_t *tsd);
+void tsd_postfork_child(tsd_t *tsd);
+
+/*
+ * Call ..._inc when your module wants to take all threads down the slow paths,
+ * and ..._dec when it no longer needs to.
+ */
+void tsd_global_slow_inc(tsdn_t *tsdn);
+void tsd_global_slow_dec(tsdn_t *tsdn);
+bool tsd_global_slow();
+
+enum {
+	/* Common case --> jnz. */
+	tsd_state_nominal = 0,
+	/* Initialized but on slow path. */
+	tsd_state_nominal_slow = 1,
+	/*
+	 * Some thread has changed global state in such a way that all nominal
+	 * threads need to recompute their fast / slow status the next time they
+	 * get a chance.
+	 *
+	 * Any thread can change another thread's status *to* recompute, but
+	 * threads are the only ones who can change their status *from*
+	 * recompute.
+	 */
+	tsd_state_nominal_recompute = 2,
+	/*
+	 * The above nominal states should be lower values.  We use
+	 * tsd_nominal_max to separate nominal states from threads in the
+	 * process of being born / dying.
+	 */
+	tsd_state_nominal_max = 2,
+
+	/*
+	 * A thread might free() during its death as its only allocator action;
+	 * in such scenarios, we need tsd, but set up in such a way that no
+	 * cleanup is necessary.
+	 */
+	tsd_state_minimal_initialized = 3,
+	/* States during which we know we're in thread death. */
+	tsd_state_purgatory = 4,
+	tsd_state_reincarnated = 5,
+	/*
+	 * What it says on the tin; tsd that hasn't been initialized.  Note
+	 * that even when the tsd struct lives in TLS, when need to keep track
+	 * of stuff like whether or not our pthread destructors have been
+	 * scheduled, so this really truly is different than the nominal state.
+	 */
+	tsd_state_uninitialized = 6
+};
+
+/*
+ * Some TSD accesses can only be done in a nominal state.  To enforce this, we
+ * wrap TSD member access in a function that asserts on TSD state, and mangle
+ * field names to prevent touching them accidentally.
+ */
+#define TSD_MANGLE(n) cant_access_tsd_items_directly_use_a_getter_or_setter_##n
+
+#ifdef JEMALLOC_U8_ATOMICS
+#  define tsd_state_t atomic_u8_t
+#  define tsd_atomic_load atomic_load_u8
+#  define tsd_atomic_store atomic_store_u8
+#  define tsd_atomic_exchange atomic_exchange_u8
+#else
+#  define tsd_state_t atomic_u32_t
+#  define tsd_atomic_load atomic_load_u32
+#  define tsd_atomic_store atomic_store_u32
+#  define tsd_atomic_exchange atomic_exchange_u32
+#endif
+
+/* The actual tsd. */
+struct tsd_s {
+	/*
+	 * The contents should be treated as totally opaque outside the tsd
+	 * module.  Access any thread-local state through the getters and
+	 * setters below.
+	 */
+
+	/*
+	 * We manually limit the state to just a single byte.  Unless the 8-bit
+	 * atomics are unavailable (which is rare).
+	 */
+	tsd_state_t state;
+#define O(n, t, nt)							\
+	t TSD_MANGLE(n);
+MALLOC_TSD
+#undef O
+};
+
+JEMALLOC_ALWAYS_INLINE uint8_t
+tsd_state_get(tsd_t *tsd) {
+	/*
+	 * This should be atomic.  Unfortunately, compilers right now can't tell
+	 * that this can be done as a memory comparison, and forces a load into
+	 * a register that hurts fast-path performance.
+	 */
+	/* return atomic_load_u8(&tsd->state, ATOMIC_RELAXED); */
+	return *(uint8_t *)&tsd->state;
+}
+
+/*
+ * Wrapper around tsd_t that makes it possible to avoid implicit conversion
+ * between tsd_t and tsdn_t, where tsdn_t is "nullable" and has to be
+ * explicitly converted to tsd_t, which is non-nullable.
+ */
+struct tsdn_s {
+	tsd_t tsd;
+};
+#define TSDN_NULL ((tsdn_t *)0)
+JEMALLOC_ALWAYS_INLINE tsdn_t *
+tsd_tsdn(tsd_t *tsd) {
+	return (tsdn_t *)tsd;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsdn_null(const tsdn_t *tsdn) {
+	return tsdn == NULL;
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsdn_tsd(tsdn_t *tsdn) {
+	assert(!tsdn_null(tsdn));
+
+	return &tsdn->tsd;
+}
+
+/*
+ * We put the platform-specific data declarations and inlines into their own
+ * header files to avoid cluttering this file.  They define tsd_boot0,
+ * tsd_boot1, tsd_boot, tsd_booted_get, tsd_get_allocates, tsd_get, and tsd_set.
+ */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#include "jemalloc/internal/tsd_malloc_thread_cleanup.h"
+#elif (defined(JEMALLOC_TLS))
+#include "jemalloc/internal/tsd_tls.h"
+#elif (defined(_WIN32))
+#include "jemalloc/internal/tsd_win.h"
+#else
+#include "jemalloc/internal/tsd_generic.h"
+#endif
+
+/*
+ * tsd_foop_get_unsafe(tsd) returns a pointer to the thread-local instance of
+ * foo.  This omits some safety checks, and so can be used during tsd
+ * initialization and cleanup.
+ */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE t *						\
+tsd_##n##p_get_unsafe(tsd_t *tsd) {					\
+	return &tsd->TSD_MANGLE(n);					\
+}
+MALLOC_TSD
+#undef O
+
+/* tsd_foop_get(tsd) returns a pointer to the thread-local instance of foo. */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE t *						\
+tsd_##n##p_get(tsd_t *tsd) {						\
+	/*								\
+	 * Because the state might change asynchronously if it's	\
+	 * nominal, we need to make sure that we only read it once.	\
+	 */								\
+	uint8_t state = tsd_state_get(tsd);				\
+	assert(state == tsd_state_nominal ||				\
+	    state == tsd_state_nominal_slow ||				\
+	    state == tsd_state_nominal_recompute ||			\
+	    state == tsd_state_reincarnated ||				\
+	    state == tsd_state_minimal_initialized);			\
+	return tsd_##n##p_get_unsafe(tsd);				\
+}
+MALLOC_TSD
+#undef O
+
+/*
+ * tsdn_foop_get(tsdn) returns either the thread-local instance of foo (if tsdn
+ * isn't NULL), or NULL (if tsdn is NULL), cast to the nullable pointer type.
+ */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE nt *						\
+tsdn_##n##p_get(tsdn_t *tsdn) {						\
+	if (tsdn_null(tsdn)) {						\
+		return NULL;						\
+	}								\
+	tsd_t *tsd = tsdn_tsd(tsdn);					\
+	return (nt *)tsd_##n##p_get(tsd);				\
+}
+MALLOC_TSD
+#undef O
+
+/* tsd_foo_get(tsd) returns the value of the thread-local instance of foo. */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE t						\
+tsd_##n##_get(tsd_t *tsd) {						\
+	return *tsd_##n##p_get(tsd);					\
+}
+MALLOC_TSD
+#undef O
+
+/* tsd_foo_set(tsd, val) updates the thread-local instance of foo to be val. */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE void						\
+tsd_##n##_set(tsd_t *tsd, t val) {					\
+	assert(tsd_state_get(tsd) != tsd_state_reincarnated &&		\
+	    tsd_state_get(tsd) != tsd_state_minimal_initialized);	\
+	*tsd_##n##p_get(tsd) = val;					\
+}
+MALLOC_TSD
+#undef O
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_assert_fast(tsd_t *tsd) {
+	/*
+	 * Note that our fastness assertion does *not* include global slowness
+	 * counters; it's not in general possible to ensure that they won't
+	 * change asynchronously from underneath us.
+	 */
+	assert(!malloc_slow && tsd_tcache_enabled_get(tsd) &&
+	    tsd_reentrancy_level_get(tsd) == 0);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_fast(tsd_t *tsd) {
+	bool fast = (tsd_state_get(tsd) == tsd_state_nominal);
+	if (fast) {
+		tsd_assert_fast(tsd);
+	}
+
+	return fast;
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_fetch_impl(bool init, bool minimal) {
+	tsd_t *tsd = tsd_get(init);
+
+	if (!init && tsd_get_allocates() && tsd == NULL) {
+		return NULL;
+	}
+	assert(tsd != NULL);
+
+	if (unlikely(tsd_state_get(tsd) != tsd_state_nominal)) {
+		return tsd_fetch_slow(tsd, minimal);
+	}
+	assert(tsd_fast(tsd));
+	tsd_assert_fast(tsd);
+
+	return tsd;
+}
+
+/* Get a minimal TSD that requires no cleanup.  See comments in free(). */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_fetch_min(void) {
+	return tsd_fetch_impl(true, true);
+}
+
+/* For internal background threads use only. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_internal_fetch(void) {
+	tsd_t *tsd = tsd_fetch_min();
+	/* Use reincarnated state to prevent full initialization. */
+	tsd_state_set(tsd, tsd_state_reincarnated);
+
+	return tsd;
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_fetch(void) {
+	return tsd_fetch_impl(true, false);
+}
+
+static inline bool
+tsd_nominal(tsd_t *tsd) {
+	return (tsd_state_get(tsd) <= tsd_state_nominal_max);
+}
+
+JEMALLOC_ALWAYS_INLINE tsdn_t *
+tsdn_fetch(void) {
+	if (!tsd_booted_get()) {
+		return NULL;
+	}
+
+	return tsd_tsdn(tsd_fetch_impl(false, false));
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_ctx_t *
+tsd_rtree_ctx(tsd_t *tsd) {
+	return tsd_rtree_ctxp_get(tsd);
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_ctx_t *
+tsdn_rtree_ctx(tsdn_t *tsdn, rtree_ctx_t *fallback) {
+	/*
+	 * If tsd cannot be accessed, initialize the fallback rtree_ctx and
+	 * return a pointer to it.
+	 */
+	if (unlikely(tsdn_null(tsdn))) {
+		rtree_ctx_data_init(fallback);
+		return fallback;
+	}
+	return tsd_rtree_ctx(tsdn_tsd(tsdn));
+}
+
+#endif /* JEMALLOC_INTERNAL_TSD_H */
diff --git a/jemalloc/include/jemalloc/internal/tsd_generic.h b/jemalloc/include/jemalloc/internal/tsd_generic.h
new file mode 100644
index 0000000..cf73c0c
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tsd_generic.h
@@ -0,0 +1,163 @@
+#ifdef JEMALLOC_INTERNAL_TSD_GENERIC_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_GENERIC_H
+
+typedef struct tsd_init_block_s tsd_init_block_t;
+struct tsd_init_block_s {
+	ql_elm(tsd_init_block_t) link;
+	pthread_t thread;
+	void *data;
+};
+
+/* Defined in tsd.c, to allow the mutex headers to have tsd dependencies. */
+typedef struct tsd_init_head_s tsd_init_head_t;
+
+typedef struct {
+	bool initialized;
+	tsd_t val;
+} tsd_wrapper_t;
+
+void *tsd_init_check_recursion(tsd_init_head_t *head,
+    tsd_init_block_t *block);
+void tsd_init_finish(tsd_init_head_t *head, tsd_init_block_t *block);
+
+extern pthread_key_t tsd_tsd;
+extern tsd_init_head_t tsd_init_head;
+extern tsd_wrapper_t tsd_boot_wrapper;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE void
+tsd_cleanup_wrapper(void *arg) {
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *)arg;
+
+	if (wrapper->initialized) {
+		wrapper->initialized = false;
+		tsd_cleanup(&wrapper->val);
+		if (wrapper->initialized) {
+			/* Trigger another cleanup round. */
+			if (pthread_setspecific(tsd_tsd, (void *)wrapper) != 0)
+			{
+				malloc_write("<jemalloc>: Error setting TSD\n");
+				if (opt_abort) {
+					abort();
+				}
+			}
+			return;
+		}
+	}
+	malloc_tsd_dalloc(wrapper);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_wrapper_set(tsd_wrapper_t *wrapper) {
+	if (pthread_setspecific(tsd_tsd, (void *)wrapper) != 0) {
+		malloc_write("<jemalloc>: Error setting TSD\n");
+		abort();
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_wrapper_t *
+tsd_wrapper_get(bool init) {
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *)pthread_getspecific(tsd_tsd);
+
+	if (init && unlikely(wrapper == NULL)) {
+		tsd_init_block_t block;
+		wrapper = (tsd_wrapper_t *)
+		    tsd_init_check_recursion(&tsd_init_head, &block);
+		if (wrapper) {
+			return wrapper;
+		}
+		wrapper = (tsd_wrapper_t *)
+		    malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+		block.data = (void *)wrapper;
+		if (wrapper == NULL) {
+			malloc_write("<jemalloc>: Error allocating TSD\n");
+			abort();
+		} else {
+			wrapper->initialized = false;
+      JEMALLOC_DIAGNOSTIC_PUSH
+      JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+			tsd_t initializer = TSD_INITIALIZER;
+      JEMALLOC_DIAGNOSTIC_POP
+			wrapper->val = initializer;
+		}
+		tsd_wrapper_set(wrapper);
+		tsd_init_finish(&tsd_init_head, &block);
+	}
+	return wrapper;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	if (pthread_key_create(&tsd_tsd, tsd_cleanup_wrapper) != 0) {
+		return true;
+	}
+	tsd_wrapper_set(&tsd_boot_wrapper);
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	tsd_wrapper_t *wrapper;
+	wrapper = (tsd_wrapper_t *)malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+	if (wrapper == NULL) {
+		malloc_write("<jemalloc>: Error allocating TSD\n");
+		abort();
+	}
+	tsd_boot_wrapper.initialized = false;
+	tsd_cleanup(&tsd_boot_wrapper.val);
+	wrapper->initialized = false;
+  JEMALLOC_DIAGNOSTIC_PUSH
+  JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+	tsd_t initializer = TSD_INITIALIZER;
+  JEMALLOC_DIAGNOSTIC_POP
+	wrapper->val = initializer;
+	tsd_wrapper_set(wrapper);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	if (tsd_boot0()) {
+		return true;
+	}
+	tsd_boot1();
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return true;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(init);
+	if (tsd_get_allocates() && !init && wrapper == NULL) {
+		return NULL;
+	}
+	return &wrapper->val;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(true);
+	if (likely(&wrapper->val != val)) {
+		wrapper->val = *(val);
+	}
+	wrapper->initialized = true;
+}
diff --git a/jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h b/jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
new file mode 100644
index 0000000..65852d5
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
@@ -0,0 +1,61 @@
+#ifdef JEMALLOC_INTERNAL_TSD_MALLOC_THREAD_CLEANUP_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_MALLOC_THREAD_CLEANUP_H
+
+#define JEMALLOC_TSD_TYPE_ATTR(type) __thread type JEMALLOC_TLS_MODEL
+
+extern JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls;
+extern JEMALLOC_TSD_TYPE_ATTR(bool) tsd_initialized;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE bool
+tsd_cleanup_wrapper(void) {
+	if (tsd_initialized) {
+		tsd_initialized = false;
+		tsd_cleanup(&tsd_tls);
+	}
+	return tsd_initialized;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	malloc_tsd_cleanup_register(&tsd_cleanup_wrapper);
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	/* Do nothing. */
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	return tsd_boot0();
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return false;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	return &tsd_tls;
+}
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	assert(tsd_booted);
+	if (likely(&tsd_tls != val)) {
+		tsd_tls = (*val);
+	}
+	tsd_initialized = true;
+}
diff --git a/jemalloc/include/jemalloc/internal/tsd_tls.h b/jemalloc/include/jemalloc/internal/tsd_tls.h
new file mode 100644
index 0000000..7d6c805
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tsd_tls.h
@@ -0,0 +1,60 @@
+#ifdef JEMALLOC_INTERNAL_TSD_TLS_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_TLS_H
+
+#define JEMALLOC_TSD_TYPE_ATTR(type) __thread type JEMALLOC_TLS_MODEL
+
+extern JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls;
+extern pthread_key_t tsd_tsd;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	if (pthread_key_create(&tsd_tsd, &tsd_cleanup) != 0) {
+		return true;
+	}
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	/* Do nothing. */
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	return tsd_boot0();
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return false;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	return &tsd_tls;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	assert(tsd_booted);
+	if (likely(&tsd_tls != val)) {
+		tsd_tls = (*val);
+	}
+	if (pthread_setspecific(tsd_tsd, (void *)(&tsd_tls)) != 0) {
+		malloc_write("<jemalloc>: Error setting tsd.\n");
+		if (opt_abort) {
+			abort();
+		}
+	}
+}
diff --git a/jemalloc/include/jemalloc/internal/tsd_types.h b/jemalloc/include/jemalloc/internal/tsd_types.h
new file mode 100644
index 0000000..6200af6
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tsd_types.h
@@ -0,0 +1,10 @@
+#ifndef JEMALLOC_INTERNAL_TSD_TYPES_H
+#define JEMALLOC_INTERNAL_TSD_TYPES_H
+
+#define MALLOC_TSD_CLEANUPS_MAX	2
+
+typedef struct tsd_s tsd_t;
+typedef struct tsdn_s tsdn_t;
+typedef bool (*malloc_tsd_cleanup_t)(void);
+
+#endif /* JEMALLOC_INTERNAL_TSD_TYPES_H */
diff --git a/jemalloc/include/jemalloc/internal/tsd_win.h b/jemalloc/include/jemalloc/internal/tsd_win.h
new file mode 100644
index 0000000..cf30d18
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/tsd_win.h
@@ -0,0 +1,139 @@
+#ifdef JEMALLOC_INTERNAL_TSD_WIN_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_WIN_H
+
+typedef struct {
+	bool initialized;
+	tsd_t val;
+} tsd_wrapper_t;
+
+extern DWORD tsd_tsd;
+extern tsd_wrapper_t tsd_boot_wrapper;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE bool
+tsd_cleanup_wrapper(void) {
+	DWORD error = GetLastError();
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *)TlsGetValue(tsd_tsd);
+	SetLastError(error);
+
+	if (wrapper == NULL) {
+		return false;
+	}
+
+	if (wrapper->initialized) {
+		wrapper->initialized = false;
+		tsd_cleanup(&wrapper->val);
+		if (wrapper->initialized) {
+			/* Trigger another cleanup round. */
+			return true;
+		}
+	}
+	malloc_tsd_dalloc(wrapper);
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_wrapper_set(tsd_wrapper_t *wrapper) {
+	if (!TlsSetValue(tsd_tsd, (void *)wrapper)) {
+		malloc_write("<jemalloc>: Error setting TSD\n");
+		abort();
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_wrapper_t *
+tsd_wrapper_get(bool init) {
+	DWORD error = GetLastError();
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *) TlsGetValue(tsd_tsd);
+	SetLastError(error);
+
+	if (init && unlikely(wrapper == NULL)) {
+		wrapper = (tsd_wrapper_t *)
+		    malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+		if (wrapper == NULL) {
+			malloc_write("<jemalloc>: Error allocating TSD\n");
+			abort();
+		} else {
+			wrapper->initialized = false;
+			/* MSVC is finicky about aggregate initialization. */
+			tsd_t tsd_initializer = TSD_INITIALIZER;
+			wrapper->val = tsd_initializer;
+		}
+		tsd_wrapper_set(wrapper);
+	}
+	return wrapper;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	tsd_tsd = TlsAlloc();
+	if (tsd_tsd == TLS_OUT_OF_INDEXES) {
+		return true;
+	}
+	malloc_tsd_cleanup_register(&tsd_cleanup_wrapper);
+	tsd_wrapper_set(&tsd_boot_wrapper);
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	tsd_wrapper_t *wrapper;
+	wrapper = (tsd_wrapper_t *)
+	    malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+	if (wrapper == NULL) {
+		malloc_write("<jemalloc>: Error allocating TSD\n");
+		abort();
+	}
+	tsd_boot_wrapper.initialized = false;
+	tsd_cleanup(&tsd_boot_wrapper.val);
+	wrapper->initialized = false;
+	tsd_t initializer = TSD_INITIALIZER;
+	wrapper->val = initializer;
+	tsd_wrapper_set(wrapper);
+}
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	if (tsd_boot0()) {
+		return true;
+	}
+	tsd_boot1();
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return true;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(init);
+	if (tsd_get_allocates() && !init && wrapper == NULL) {
+		return NULL;
+	}
+	return &wrapper->val;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(true);
+	if (likely(&wrapper->val != val)) {
+		wrapper->val = *(val);
+	}
+	wrapper->initialized = true;
+}
diff --git a/jemalloc/include/jemalloc/internal/util.h b/jemalloc/include/jemalloc/internal/util.h
new file mode 100644
index 0000000..304cb54
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/util.h
@@ -0,0 +1,67 @@
+#ifndef JEMALLOC_INTERNAL_UTIL_H
+#define JEMALLOC_INTERNAL_UTIL_H
+
+#define UTIL_INLINE static inline
+
+/* Junk fill patterns. */
+#ifndef JEMALLOC_ALLOC_JUNK
+#  define JEMALLOC_ALLOC_JUNK	((uint8_t)0xa5)
+#endif
+#ifndef JEMALLOC_FREE_JUNK
+#  define JEMALLOC_FREE_JUNK	((uint8_t)0x5a)
+#endif
+
+/*
+ * Wrap a cpp argument that contains commas such that it isn't broken up into
+ * multiple arguments.
+ */
+#define JEMALLOC_ARG_CONCAT(...) __VA_ARGS__
+
+/* cpp macro definition stringification. */
+#define STRINGIFY_HELPER(x) #x
+#define STRINGIFY(x) STRINGIFY_HELPER(x)
+
+/*
+ * Silence compiler warnings due to uninitialized values.  This is used
+ * wherever the compiler fails to recognize that the variable is never used
+ * uninitialized.
+ */
+#define JEMALLOC_CC_SILENCE_INIT(v) = v
+
+#ifdef __GNUC__
+#  define likely(x)   __builtin_expect(!!(x), 1)
+#  define unlikely(x) __builtin_expect(!!(x), 0)
+#else
+#  define likely(x)   !!(x)
+#  define unlikely(x) !!(x)
+#endif
+
+#if !defined(JEMALLOC_INTERNAL_UNREACHABLE)
+#  error JEMALLOC_INTERNAL_UNREACHABLE should have been defined by configure
+#endif
+
+#define unreachable() JEMALLOC_INTERNAL_UNREACHABLE()
+
+/* Set error code. */
+UTIL_INLINE void
+set_errno(int errnum) {
+#ifdef _WIN32
+	SetLastError(errnum);
+#else
+	errno = errnum;
+#endif
+}
+
+/* Get last error code. */
+UTIL_INLINE int
+get_errno(void) {
+#ifdef _WIN32
+	return GetLastError();
+#else
+	return errno;
+#endif
+}
+
+#undef UTIL_INLINE
+
+#endif /* JEMALLOC_INTERNAL_UTIL_H */
diff --git a/jemalloc/include/jemalloc/internal/witness.h b/jemalloc/include/jemalloc/internal/witness.h
new file mode 100644
index 0000000..fff9e98
--- /dev/null
+++ b/jemalloc/include/jemalloc/internal/witness.h
@@ -0,0 +1,347 @@
+#ifndef JEMALLOC_INTERNAL_WITNESS_H
+#define JEMALLOC_INTERNAL_WITNESS_H
+
+#include "jemalloc/internal/ql.h"
+
+/******************************************************************************/
+/* LOCK RANKS */
+/******************************************************************************/
+
+/*
+ * Witnesses with rank WITNESS_RANK_OMIT are completely ignored by the witness
+ * machinery.
+ */
+
+#define WITNESS_RANK_OMIT		0U
+
+#define WITNESS_RANK_MIN		1U
+
+#define WITNESS_RANK_INIT		1U
+#define WITNESS_RANK_CTL		1U
+#define WITNESS_RANK_TCACHES		2U
+#define WITNESS_RANK_ARENAS		3U
+
+#define WITNESS_RANK_BACKGROUND_THREAD_GLOBAL	4U
+
+#define WITNESS_RANK_PROF_DUMP		5U
+#define WITNESS_RANK_PROF_BT2GCTX	6U
+#define WITNESS_RANK_PROF_TDATAS	7U
+#define WITNESS_RANK_PROF_TDATA		8U
+#define WITNESS_RANK_PROF_LOG		9U
+#define WITNESS_RANK_PROF_GCTX		10U
+#define WITNESS_RANK_BACKGROUND_THREAD	11U
+
+/*
+ * Used as an argument to witness_assert_depth_to_rank() in order to validate
+ * depth excluding non-core locks with lower ranks.  Since the rank argument to
+ * witness_assert_depth_to_rank() is inclusive rather than exclusive, this
+ * definition can have the same value as the minimally ranked core lock.
+ */
+#define WITNESS_RANK_CORE		12U
+
+#define WITNESS_RANK_DECAY		12U
+#define WITNESS_RANK_TCACHE_QL		13U
+#define WITNESS_RANK_EXTENT_GROW	14U
+#define WITNESS_RANK_EXTENTS		15U
+#define WITNESS_RANK_EXTENT_AVAIL	16U
+
+#define WITNESS_RANK_EXTENT_POOL	17U
+#define WITNESS_RANK_RTREE		18U
+#define WITNESS_RANK_BASE		19U
+#define WITNESS_RANK_ARENA_LARGE	20U
+#define WITNESS_RANK_HOOK		21U
+
+#define WITNESS_RANK_LEAF		0xffffffffU
+#define WITNESS_RANK_BIN		WITNESS_RANK_LEAF
+#define WITNESS_RANK_ARENA_STATS	WITNESS_RANK_LEAF
+#define WITNESS_RANK_DSS		WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_ACTIVE	WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_ACCUM		WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_DUMP_SEQ	WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_GDUMP		WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_NEXT_THR_UID	WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_THREAD_ACTIVE_INIT	WITNESS_RANK_LEAF
+
+/******************************************************************************/
+/* PER-WITNESS DATA */
+/******************************************************************************/
+#if defined(JEMALLOC_DEBUG)
+#  define WITNESS_INITIALIZER(name, rank) {name, rank, NULL, NULL, {NULL, NULL}}
+#else
+#  define WITNESS_INITIALIZER(name, rank)
+#endif
+
+typedef struct witness_s witness_t;
+typedef unsigned witness_rank_t;
+typedef ql_head(witness_t) witness_list_t;
+typedef int witness_comp_t (const witness_t *, void *, const witness_t *,
+    void *);
+
+struct witness_s {
+	/* Name, used for printing lock order reversal messages. */
+	const char		*name;
+
+	/*
+	 * Witness rank, where 0 is lowest and UINT_MAX is highest.  Witnesses
+	 * must be acquired in order of increasing rank.
+	 */
+	witness_rank_t		rank;
+
+	/*
+	 * If two witnesses are of equal rank and they have the samp comp
+	 * function pointer, it is called as a last attempt to differentiate
+	 * between witnesses of equal rank.
+	 */
+	witness_comp_t		*comp;
+
+	/* Opaque data, passed to comp(). */
+	void			*opaque;
+
+	/* Linkage for thread's currently owned locks. */
+	ql_elm(witness_t)	link;
+};
+
+/******************************************************************************/
+/* PER-THREAD DATA */
+/******************************************************************************/
+typedef struct witness_tsd_s witness_tsd_t;
+struct witness_tsd_s {
+	witness_list_t witnesses;
+	bool forking;
+};
+
+#define WITNESS_TSD_INITIALIZER { ql_head_initializer(witnesses), false }
+#define WITNESS_TSDN_NULL ((witness_tsdn_t *)0)
+
+/******************************************************************************/
+/* (PER-THREAD) NULLABILITY HELPERS */
+/******************************************************************************/
+typedef struct witness_tsdn_s witness_tsdn_t;
+struct witness_tsdn_s {
+	witness_tsd_t witness_tsd;
+};
+
+JEMALLOC_ALWAYS_INLINE witness_tsdn_t *
+witness_tsd_tsdn(witness_tsd_t *witness_tsd) {
+	return (witness_tsdn_t *)witness_tsd;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+witness_tsdn_null(witness_tsdn_t *witness_tsdn) {
+	return witness_tsdn == NULL;
+}
+
+JEMALLOC_ALWAYS_INLINE witness_tsd_t *
+witness_tsdn_tsd(witness_tsdn_t *witness_tsdn) {
+	assert(!witness_tsdn_null(witness_tsdn));
+	return &witness_tsdn->witness_tsd;
+}
+
+/******************************************************************************/
+/* API */
+/******************************************************************************/
+void witness_init(witness_t *witness, const char *name, witness_rank_t rank,
+    witness_comp_t *comp, void *opaque);
+
+typedef void (witness_lock_error_t)(const witness_list_t *, const witness_t *);
+extern witness_lock_error_t *JET_MUTABLE witness_lock_error;
+
+typedef void (witness_owner_error_t)(const witness_t *);
+extern witness_owner_error_t *JET_MUTABLE witness_owner_error;
+
+typedef void (witness_not_owner_error_t)(const witness_t *);
+extern witness_not_owner_error_t *JET_MUTABLE witness_not_owner_error;
+
+typedef void (witness_depth_error_t)(const witness_list_t *,
+    witness_rank_t rank_inclusive, unsigned depth);
+extern witness_depth_error_t *JET_MUTABLE witness_depth_error;
+
+void witnesses_cleanup(witness_tsd_t *witness_tsd);
+void witness_prefork(witness_tsd_t *witness_tsd);
+void witness_postfork_parent(witness_tsd_t *witness_tsd);
+void witness_postfork_child(witness_tsd_t *witness_tsd);
+
+/* Helper, not intended for direct use. */
+static inline bool
+witness_owner(witness_tsd_t *witness_tsd, const witness_t *witness) {
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	cassert(config_debug);
+
+	witnesses = &witness_tsd->witnesses;
+	ql_foreach(w, witnesses, link) {
+		if (w == witness) {
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static inline void
+witness_assert_owner(witness_tsdn_t *witness_tsdn, const witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	if (witness_owner(witness_tsd, witness)) {
+		return;
+	}
+	witness_owner_error(witness);
+}
+
+static inline void
+witness_assert_not_owner(witness_tsdn_t *witness_tsdn,
+    const witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	witnesses = &witness_tsd->witnesses;
+	ql_foreach(w, witnesses, link) {
+		if (w == witness) {
+			witness_not_owner_error(witness);
+		}
+	}
+}
+
+static inline void
+witness_assert_depth_to_rank(witness_tsdn_t *witness_tsdn,
+    witness_rank_t rank_inclusive, unsigned depth) {
+	witness_tsd_t *witness_tsd;
+	unsigned d;
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+
+	d = 0;
+	witnesses = &witness_tsd->witnesses;
+	w = ql_last(witnesses, link);
+	if (w != NULL) {
+		ql_reverse_foreach(w, witnesses, link) {
+			if (w->rank < rank_inclusive) {
+				break;
+			}
+			d++;
+		}
+	}
+	if (d != depth) {
+		witness_depth_error(witnesses, rank_inclusive, depth);
+	}
+}
+
+static inline void
+witness_assert_depth(witness_tsdn_t *witness_tsdn, unsigned depth) {
+	witness_assert_depth_to_rank(witness_tsdn, WITNESS_RANK_MIN, depth);
+}
+
+static inline void
+witness_assert_lockless(witness_tsdn_t *witness_tsdn) {
+	witness_assert_depth(witness_tsdn, 0);
+}
+
+static inline void
+witness_lock(witness_tsdn_t *witness_tsdn, witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	witness_assert_not_owner(witness_tsdn, witness);
+
+	witnesses = &witness_tsd->witnesses;
+	w = ql_last(witnesses, link);
+	if (w == NULL) {
+		/* No other locks; do nothing. */
+	} else if (witness_tsd->forking && w->rank <= witness->rank) {
+		/* Forking, and relaxed ranking satisfied. */
+	} else if (w->rank > witness->rank) {
+		/* Not forking, rank order reversal. */
+		witness_lock_error(witnesses, witness);
+	} else if (w->rank == witness->rank && (w->comp == NULL || w->comp !=
+	    witness->comp || w->comp(w, w->opaque, witness, witness->opaque) >
+	    0)) {
+		/*
+		 * Missing/incompatible comparison function, or comparison
+		 * function indicates rank order reversal.
+		 */
+		witness_lock_error(witnesses, witness);
+	}
+
+	ql_elm_new(witness, link);
+	ql_tail_insert(witnesses, witness, link);
+}
+
+static inline void
+witness_unlock(witness_tsdn_t *witness_tsdn, witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+	witness_list_t *witnesses;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	/*
+	 * Check whether owner before removal, rather than relying on
+	 * witness_assert_owner() to abort, so that unit tests can test this
+	 * function's failure mode without causing undefined behavior.
+	 */
+	if (witness_owner(witness_tsd, witness)) {
+		witnesses = &witness_tsd->witnesses;
+		ql_remove(witnesses, witness, link);
+	} else {
+		witness_assert_owner(witness_tsdn, witness);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_WITNESS_H */
diff --git a/jemalloc/include/jemalloc/jemalloc.sh b/jemalloc/include/jemalloc/jemalloc.sh
new file mode 100755
index 0000000..b19b154
--- /dev/null
+++ b/jemalloc/include/jemalloc/jemalloc.sh
@@ -0,0 +1,27 @@
+#!/bin/sh
+
+objroot=$1
+
+cat <<EOF
+#ifndef JEMALLOC_H_
+#define JEMALLOC_H_
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+EOF
+
+for hdr in jemalloc_defs.h jemalloc_rename.h jemalloc_macros.h \
+           jemalloc_protos.h jemalloc_typedefs.h jemalloc_mangle.h ; do
+  cat "${objroot}include/jemalloc/${hdr}" \
+      | grep -v 'Generated from .* by configure\.' \
+      | sed -e 's/ $//g'
+  echo
+done
+
+cat <<EOF
+#ifdef __cplusplus
+}
+#endif
+#endif /* JEMALLOC_H_ */
+EOF
diff --git a/jemalloc/include/jemalloc/jemalloc_defs.h.in b/jemalloc/include/jemalloc/jemalloc_defs.h.in
new file mode 100644
index 0000000..11c3918
--- /dev/null
+++ b/jemalloc/include/jemalloc/jemalloc_defs.h.in
@@ -0,0 +1,48 @@
+/* Defined if __attribute__((...)) syntax is supported. */
+#undef JEMALLOC_HAVE_ATTR
+
+/* Defined if alloc_size attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_ALLOC_SIZE
+
+/* Defined if format_arg(...) attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_FORMAT_ARG
+
+/* Defined if format(gnu_printf, ...) attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF
+
+/* Defined if format(printf, ...) attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_FORMAT_PRINTF
+
+/*
+ * Define overrides for non-standard allocator-related functions if they are
+ * present on the system.
+ */
+#undef JEMALLOC_OVERRIDE_MEMALIGN
+#undef JEMALLOC_OVERRIDE_VALLOC
+
+/*
+ * At least Linux omits the "const" in:
+ *
+ *   size_t malloc_usable_size(const void *ptr);
+ *
+ * Match the operating system's prototype.
+ */
+#undef JEMALLOC_USABLE_SIZE_CONST
+
+/*
+ * If defined, specify throw() for the public function prototypes when compiling
+ * with C++.  The only justification for this is to match the prototypes that
+ * glibc defines.
+ */
+#undef JEMALLOC_USE_CXX_THROW
+
+#ifdef _MSC_VER
+#  ifdef _WIN64
+#    define LG_SIZEOF_PTR_WIN 3
+#  else
+#    define LG_SIZEOF_PTR_WIN 2
+#  endif
+#endif
+
+/* sizeof(void *) == 2^LG_SIZEOF_PTR. */
+#undef LG_SIZEOF_PTR
diff --git a/jemalloc/include/jemalloc/jemalloc_macros.h.in b/jemalloc/include/jemalloc/jemalloc_macros.h.in
new file mode 100644
index 0000000..59e2955
--- /dev/null
+++ b/jemalloc/include/jemalloc/jemalloc_macros.h.in
@@ -0,0 +1,129 @@
+#include <stdlib.h>
+#include <stdbool.h>
+#include <stdint.h>
+#include <limits.h>
+#include <strings.h>
+
+#define JEMALLOC_VERSION "@jemalloc_version@"
+#define JEMALLOC_VERSION_MAJOR @jemalloc_version_major@
+#define JEMALLOC_VERSION_MINOR @jemalloc_version_minor@
+#define JEMALLOC_VERSION_BUGFIX @jemalloc_version_bugfix@
+#define JEMALLOC_VERSION_NREV @jemalloc_version_nrev@
+#define JEMALLOC_VERSION_GID "@jemalloc_version_gid@"
+#define JEMALLOC_VERSION_GID_IDENT @jemalloc_version_gid@
+
+#define MALLOCX_LG_ALIGN(la)	((int)(la))
+#if LG_SIZEOF_PTR == 2
+#  define MALLOCX_ALIGN(a)	((int)(ffs((int)(a))-1))
+#else
+#  define MALLOCX_ALIGN(a)						\
+     ((int)(((size_t)(a) < (size_t)INT_MAX) ? ffs((int)(a))-1 :	\
+     ffs((int)(((size_t)(a))>>32))+31))
+#endif
+#define MALLOCX_ZERO	((int)0x40)
+/*
+ * Bias tcache index bits so that 0 encodes "automatic tcache management", and 1
+ * encodes MALLOCX_TCACHE_NONE.
+ */
+#define MALLOCX_TCACHE(tc)	((int)(((tc)+2) << 8))
+#define MALLOCX_TCACHE_NONE	MALLOCX_TCACHE(-1)
+/*
+ * Bias arena index bits so that 0 encodes "use an automatically chosen arena".
+ */
+#define MALLOCX_ARENA(a)	((((int)(a))+1) << 20)
+
+/*
+ * Use as arena index in "arena.<i>.{purge,decay,dss}" and
+ * "stats.arenas.<i>.*" mallctl interfaces to select all arenas.  This
+ * definition is intentionally specified in raw decimal format to support
+ * cpp-based string concatenation, e.g.
+ *
+ *   #define STRINGIFY_HELPER(x) #x
+ *   #define STRINGIFY(x) STRINGIFY_HELPER(x)
+ *
+ *   mallctl("arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".purge", NULL, NULL, NULL,
+ *       0);
+ */
+#define MALLCTL_ARENAS_ALL	4096
+/*
+ * Use as arena index in "stats.arenas.<i>.*" mallctl interfaces to select
+ * destroyed arenas.
+ */
+#define MALLCTL_ARENAS_DESTROYED	4097
+
+#if defined(__cplusplus) && defined(JEMALLOC_USE_CXX_THROW)
+#  define JEMALLOC_CXX_THROW throw()
+#else
+#  define JEMALLOC_CXX_THROW
+#endif
+
+#if defined(_MSC_VER)
+#  define JEMALLOC_ATTR(s)
+#  define JEMALLOC_ALIGNED(s) __declspec(align(s))
+#  define JEMALLOC_ALLOC_SIZE(s)
+#  define JEMALLOC_ALLOC_SIZE2(s1, s2)
+#  ifndef JEMALLOC_EXPORT
+#    ifdef DLLEXPORT
+#      define JEMALLOC_EXPORT __declspec(dllexport)
+#    else
+#      define JEMALLOC_EXPORT __declspec(dllimport)
+#    endif
+#  endif
+#  define JEMALLOC_FORMAT_ARG(i)
+#  define JEMALLOC_FORMAT_PRINTF(s, i)
+#  define JEMALLOC_NOINLINE __declspec(noinline)
+#  ifdef __cplusplus
+#    define JEMALLOC_NOTHROW __declspec(nothrow)
+#  else
+#    define JEMALLOC_NOTHROW
+#  endif
+#  define JEMALLOC_SECTION(s) __declspec(allocate(s))
+#  define JEMALLOC_RESTRICT_RETURN __declspec(restrict)
+#  if _MSC_VER >= 1900 && !defined(__EDG__)
+#    define JEMALLOC_ALLOCATOR __declspec(allocator)
+#  else
+#    define JEMALLOC_ALLOCATOR
+#  endif
+#elif defined(JEMALLOC_HAVE_ATTR)
+#  define JEMALLOC_ATTR(s) __attribute__((s))
+#  define JEMALLOC_ALIGNED(s) JEMALLOC_ATTR(aligned(s))
+#  ifdef JEMALLOC_HAVE_ATTR_ALLOC_SIZE
+#    define JEMALLOC_ALLOC_SIZE(s) JEMALLOC_ATTR(alloc_size(s))
+#    define JEMALLOC_ALLOC_SIZE2(s1, s2) JEMALLOC_ATTR(alloc_size(s1, s2))
+#  else
+#    define JEMALLOC_ALLOC_SIZE(s)
+#    define JEMALLOC_ALLOC_SIZE2(s1, s2)
+#  endif
+#  ifndef JEMALLOC_EXPORT
+#    define JEMALLOC_EXPORT JEMALLOC_ATTR(visibility("default"))
+#  endif
+#  ifdef JEMALLOC_HAVE_ATTR_FORMAT_ARG
+#    define JEMALLOC_FORMAT_ARG(i) JEMALLOC_ATTR(__format_arg__(3))
+#  else
+#    define JEMALLOC_FORMAT_ARG(i)
+#  endif
+#  ifdef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF
+#    define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(gnu_printf, s, i))
+#  elif defined(JEMALLOC_HAVE_ATTR_FORMAT_PRINTF)
+#    define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(printf, s, i))
+#  else
+#    define JEMALLOC_FORMAT_PRINTF(s, i)
+#  endif
+#  define JEMALLOC_NOINLINE JEMALLOC_ATTR(noinline)
+#  define JEMALLOC_NOTHROW JEMALLOC_ATTR(nothrow)
+#  define JEMALLOC_SECTION(s) JEMALLOC_ATTR(section(s))
+#  define JEMALLOC_RESTRICT_RETURN
+#  define JEMALLOC_ALLOCATOR
+#else
+#  define JEMALLOC_ATTR(s)
+#  define JEMALLOC_ALIGNED(s)
+#  define JEMALLOC_ALLOC_SIZE(s)
+#  define JEMALLOC_ALLOC_SIZE2(s1, s2)
+#  define JEMALLOC_EXPORT
+#  define JEMALLOC_FORMAT_PRINTF(s, i)
+#  define JEMALLOC_NOINLINE
+#  define JEMALLOC_NOTHROW
+#  define JEMALLOC_SECTION(s)
+#  define JEMALLOC_RESTRICT_RETURN
+#  define JEMALLOC_ALLOCATOR
+#endif
diff --git a/jemalloc/include/jemalloc/jemalloc_mangle.sh b/jemalloc/include/jemalloc/jemalloc_mangle.sh
new file mode 100755
index 0000000..c675bb4
--- /dev/null
+++ b/jemalloc/include/jemalloc/jemalloc_mangle.sh
@@ -0,0 +1,45 @@
+#!/bin/sh -eu
+
+public_symbols_txt=$1
+symbol_prefix=$2
+
+cat <<EOF
+/*
+ * By default application code must explicitly refer to mangled symbol names,
+ * so that it is possible to use jemalloc in conjunction with another allocator
+ * in the same application.  Define JEMALLOC_MANGLE in order to cause automatic
+ * name mangling that matches the API prefixing that happened as a result of
+ * --with-mangling and/or --with-jemalloc-prefix configuration settings.
+ */
+#ifdef JEMALLOC_MANGLE
+#  ifndef JEMALLOC_NO_DEMANGLE
+#    define JEMALLOC_NO_DEMANGLE
+#  endif
+EOF
+
+for nm in `cat ${public_symbols_txt}` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#  define ${n} ${symbol_prefix}${n}"
+done
+
+cat <<EOF
+#endif
+
+/*
+ * The ${symbol_prefix}* macros can be used as stable alternative names for the
+ * public jemalloc API if JEMALLOC_NO_DEMANGLE is defined.  This is primarily
+ * meant for use in jemalloc itself, but it can be used by application code to
+ * provide isolation from the name mangling specified via --with-mangling
+ * and/or --with-jemalloc-prefix.
+ */
+#ifndef JEMALLOC_NO_DEMANGLE
+EOF
+
+for nm in `cat ${public_symbols_txt}` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#  undef ${symbol_prefix}${n}"
+done
+
+cat <<EOF
+#endif
+EOF
diff --git a/jemalloc/include/jemalloc/jemalloc_protos.h.in b/jemalloc/include/jemalloc/jemalloc_protos.h.in
new file mode 100644
index 0000000..a78414b
--- /dev/null
+++ b/jemalloc/include/jemalloc/jemalloc_protos.h.in
@@ -0,0 +1,66 @@
+/*
+ * The @je_@ prefix on the following public symbol declarations is an artifact
+ * of namespace management, and should be omitted in application code unless
+ * JEMALLOC_NO_DEMANGLE is defined (see jemalloc_mangle@install_suffix@.h).
+ */
+extern JEMALLOC_EXPORT const char	*@je_@malloc_conf;
+extern JEMALLOC_EXPORT void		(*@je_@malloc_message)(void *cbopaque,
+    const char *s);
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@malloc(size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1);
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@calloc(size_t num, size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE2(1, 2);
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@posix_memalign(void **memptr,
+    size_t alignment, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(nonnull(1));
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@aligned_alloc(size_t alignment,
+    size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc)
+    JEMALLOC_ALLOC_SIZE(2);
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@realloc(void *ptr, size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ALLOC_SIZE(2);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@free(void *ptr)
+    JEMALLOC_CXX_THROW;
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@mallocx(size_t size, int flags)
+    JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1);
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@rallocx(void *ptr, size_t size,
+    int flags) JEMALLOC_ALLOC_SIZE(2);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@xallocx(void *ptr, size_t size,
+    size_t extra, int flags);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@sallocx(const void *ptr,
+    int flags) JEMALLOC_ATTR(pure);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@dallocx(void *ptr, int flags);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@sdallocx(void *ptr, size_t size,
+    int flags);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@nallocx(size_t size, int flags)
+    JEMALLOC_ATTR(pure);
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@mallctl(const char *name,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen);
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@mallctlnametomib(const char *name,
+    size_t *mibp, size_t *miblenp);
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@mallctlbymib(const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@malloc_stats_print(
+    void (*write_cb)(void *, const char *), void *@je_@cbopaque,
+    const char *opts);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@malloc_usable_size(
+    JEMALLOC_USABLE_SIZE_CONST void *ptr) JEMALLOC_CXX_THROW;
+
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@memalign(size_t alignment, size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc);
+#endif
+
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@valloc(size_t size) JEMALLOC_CXX_THROW
+    JEMALLOC_ATTR(malloc);
+#endif
diff --git a/jemalloc/include/jemalloc/jemalloc_rename.sh b/jemalloc/include/jemalloc/jemalloc_rename.sh
new file mode 100755
index 0000000..f943891
--- /dev/null
+++ b/jemalloc/include/jemalloc/jemalloc_rename.sh
@@ -0,0 +1,22 @@
+#!/bin/sh
+
+public_symbols_txt=$1
+
+cat <<EOF
+/*
+ * Name mangling for public symbols is controlled by --with-mangling and
+ * --with-jemalloc-prefix.  With default settings the je_ prefix is stripped by
+ * these macro definitions.
+ */
+#ifndef JEMALLOC_NO_RENAME
+EOF
+
+for nm in `cat ${public_symbols_txt}` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  m=`echo ${nm} |tr ':' ' ' |awk '{print $2}'`
+  echo "#  define je_${n} ${m}"
+done
+
+cat <<EOF
+#endif
+EOF
diff --git a/jemalloc/include/jemalloc/jemalloc_typedefs.h.in b/jemalloc/include/jemalloc/jemalloc_typedefs.h.in
new file mode 100644
index 0000000..1a58874
--- /dev/null
+++ b/jemalloc/include/jemalloc/jemalloc_typedefs.h.in
@@ -0,0 +1,77 @@
+typedef struct extent_hooks_s extent_hooks_t;
+
+/*
+ * void *
+ * extent_alloc(extent_hooks_t *extent_hooks, void *new_addr, size_t size,
+ *     size_t alignment, bool *zero, bool *commit, unsigned arena_ind);
+ */
+typedef void *(extent_alloc_t)(extent_hooks_t *, void *, size_t, size_t, bool *,
+    bool *, unsigned);
+
+/*
+ * bool
+ * extent_dalloc(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     bool committed, unsigned arena_ind);
+ */
+typedef bool (extent_dalloc_t)(extent_hooks_t *, void *, size_t, bool,
+    unsigned);
+
+/*
+ * void
+ * extent_destroy(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     bool committed, unsigned arena_ind);
+ */
+typedef void (extent_destroy_t)(extent_hooks_t *, void *, size_t, bool,
+    unsigned);
+
+/*
+ * bool
+ * extent_commit(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t offset, size_t length, unsigned arena_ind);
+ */
+typedef bool (extent_commit_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
+    unsigned);
+
+/*
+ * bool
+ * extent_decommit(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t offset, size_t length, unsigned arena_ind);
+ */
+typedef bool (extent_decommit_t)(extent_hooks_t *, void *, size_t, size_t,
+    size_t, unsigned);
+
+/*
+ * bool
+ * extent_purge(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t offset, size_t length, unsigned arena_ind);
+ */
+typedef bool (extent_purge_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
+    unsigned);
+
+/*
+ * bool
+ * extent_split(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t size_a, size_t size_b, bool committed, unsigned arena_ind);
+ */
+typedef bool (extent_split_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
+    bool, unsigned);
+
+/*
+ * bool
+ * extent_merge(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a,
+ *     void *addr_b, size_t size_b, bool committed, unsigned arena_ind);
+ */
+typedef bool (extent_merge_t)(extent_hooks_t *, void *, size_t, void *, size_t,
+    bool, unsigned);
+
+struct extent_hooks_s {
+	extent_alloc_t		*alloc;
+	extent_dalloc_t		*dalloc;
+	extent_destroy_t	*destroy;
+	extent_commit_t		*commit;
+	extent_decommit_t	*decommit;
+	extent_purge_t		*purge_lazy;
+	extent_purge_t		*purge_forced;
+	extent_split_t		*split;
+	extent_merge_t		*merge;
+};
diff --git a/jemalloc/include/msvc_compat/C99/stdbool.h b/jemalloc/include/msvc_compat/C99/stdbool.h
new file mode 100644
index 0000000..d92160e
--- /dev/null
+++ b/jemalloc/include/msvc_compat/C99/stdbool.h
@@ -0,0 +1,20 @@
+#ifndef stdbool_h
+#define stdbool_h
+
+#include <wtypes.h>
+
+/* MSVC doesn't define _Bool or bool in C, but does have BOOL */
+/* Note this doesn't pass autoconf's test because (bool) 0.5 != true */
+/* Clang-cl uses MSVC headers, so needs msvc_compat, but has _Bool as
+ * a built-in type. */
+#ifndef __clang__
+typedef BOOL _Bool;
+#endif
+
+#define bool _Bool
+#define true 1
+#define false 0
+
+#define __bool_true_false_are_defined 1
+
+#endif /* stdbool_h */
diff --git a/jemalloc/include/msvc_compat/C99/stdint.h b/jemalloc/include/msvc_compat/C99/stdint.h
new file mode 100644
index 0000000..d02608a
--- /dev/null
+++ b/jemalloc/include/msvc_compat/C99/stdint.h
@@ -0,0 +1,247 @@
+// ISO C9x  compliant stdint.h for Microsoft Visual Studio
+// Based on ISO/IEC 9899:TC2 Committee draft (May 6, 2005) WG14/N1124 
+// 
+//  Copyright (c) 2006-2008 Alexander Chemeris
+// 
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are met:
+// 
+//   1. Redistributions of source code must retain the above copyright notice,
+//      this list of conditions and the following disclaimer.
+// 
+//   2. Redistributions in binary form must reproduce the above copyright
+//      notice, this list of conditions and the following disclaimer in the
+//      documentation and/or other materials provided with the distribution.
+// 
+//   3. The name of the author may be used to endorse or promote products
+//      derived from this software without specific prior written permission.
+// 
+// THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
+// WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
+// EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, 
+// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+// 
+///////////////////////////////////////////////////////////////////////////////
+
+#ifndef _MSC_VER // [
+#error "Use this header only with Microsoft Visual C++ compilers!"
+#endif // _MSC_VER ]
+
+#ifndef _MSC_STDINT_H_ // [
+#define _MSC_STDINT_H_
+
+#if _MSC_VER > 1000
+#pragma once
+#endif
+
+#include <limits.h>
+
+// For Visual Studio 6 in C++ mode and for many Visual Studio versions when
+// compiling for ARM we should wrap <wchar.h> include with 'extern "C++" {}'
+// or compiler give many errors like this:
+//   error C2733: second C linkage of overloaded function 'wmemchr' not allowed
+#ifdef __cplusplus
+extern "C" {
+#endif
+#  include <wchar.h>
+#ifdef __cplusplus
+}
+#endif
+
+// Define _W64 macros to mark types changing their size, like intptr_t.
+#ifndef _W64
+#  if !defined(__midl) && (defined(_X86_) || defined(_M_IX86)) && _MSC_VER >= 1300
+#     define _W64 __w64
+#  else
+#     define _W64
+#  endif
+#endif
+
+
+// 7.18.1 Integer types
+
+// 7.18.1.1 Exact-width integer types
+
+// Visual Studio 6 and Embedded Visual C++ 4 doesn't
+// realize that, e.g. char has the same size as __int8
+// so we give up on __intX for them.
+#if (_MSC_VER < 1300)
+   typedef signed char       int8_t;
+   typedef signed short      int16_t;
+   typedef signed int        int32_t;
+   typedef unsigned char     uint8_t;
+   typedef unsigned short    uint16_t;
+   typedef unsigned int      uint32_t;
+#else
+   typedef signed __int8     int8_t;
+   typedef signed __int16    int16_t;
+   typedef signed __int32    int32_t;
+   typedef unsigned __int8   uint8_t;
+   typedef unsigned __int16  uint16_t;
+   typedef unsigned __int32  uint32_t;
+#endif
+typedef signed __int64       int64_t;
+typedef unsigned __int64     uint64_t;
+
+
+// 7.18.1.2 Minimum-width integer types
+typedef int8_t    int_least8_t;
+typedef int16_t   int_least16_t;
+typedef int32_t   int_least32_t;
+typedef int64_t   int_least64_t;
+typedef uint8_t   uint_least8_t;
+typedef uint16_t  uint_least16_t;
+typedef uint32_t  uint_least32_t;
+typedef uint64_t  uint_least64_t;
+
+// 7.18.1.3 Fastest minimum-width integer types
+typedef int8_t    int_fast8_t;
+typedef int16_t   int_fast16_t;
+typedef int32_t   int_fast32_t;
+typedef int64_t   int_fast64_t;
+typedef uint8_t   uint_fast8_t;
+typedef uint16_t  uint_fast16_t;
+typedef uint32_t  uint_fast32_t;
+typedef uint64_t  uint_fast64_t;
+
+// 7.18.1.4 Integer types capable of holding object pointers
+#ifdef _WIN64 // [
+   typedef signed __int64    intptr_t;
+   typedef unsigned __int64  uintptr_t;
+#else // _WIN64 ][
+   typedef _W64 signed int   intptr_t;
+   typedef _W64 unsigned int uintptr_t;
+#endif // _WIN64 ]
+
+// 7.18.1.5 Greatest-width integer types
+typedef int64_t   intmax_t;
+typedef uint64_t  uintmax_t;
+
+
+// 7.18.2 Limits of specified-width integer types
+
+#if !defined(__cplusplus) || defined(__STDC_LIMIT_MACROS) // [   See footnote 220 at page 257 and footnote 221 at page 259
+
+// 7.18.2.1 Limits of exact-width integer types
+#define INT8_MIN     ((int8_t)_I8_MIN)
+#define INT8_MAX     _I8_MAX
+#define INT16_MIN    ((int16_t)_I16_MIN)
+#define INT16_MAX    _I16_MAX
+#define INT32_MIN    ((int32_t)_I32_MIN)
+#define INT32_MAX    _I32_MAX
+#define INT64_MIN    ((int64_t)_I64_MIN)
+#define INT64_MAX    _I64_MAX
+#define UINT8_MAX    _UI8_MAX
+#define UINT16_MAX   _UI16_MAX
+#define UINT32_MAX   _UI32_MAX
+#define UINT64_MAX   _UI64_MAX
+
+// 7.18.2.2 Limits of minimum-width integer types
+#define INT_LEAST8_MIN    INT8_MIN
+#define INT_LEAST8_MAX    INT8_MAX
+#define INT_LEAST16_MIN   INT16_MIN
+#define INT_LEAST16_MAX   INT16_MAX
+#define INT_LEAST32_MIN   INT32_MIN
+#define INT_LEAST32_MAX   INT32_MAX
+#define INT_LEAST64_MIN   INT64_MIN
+#define INT_LEAST64_MAX   INT64_MAX
+#define UINT_LEAST8_MAX   UINT8_MAX
+#define UINT_LEAST16_MAX  UINT16_MAX
+#define UINT_LEAST32_MAX  UINT32_MAX
+#define UINT_LEAST64_MAX  UINT64_MAX
+
+// 7.18.2.3 Limits of fastest minimum-width integer types
+#define INT_FAST8_MIN    INT8_MIN
+#define INT_FAST8_MAX    INT8_MAX
+#define INT_FAST16_MIN   INT16_MIN
+#define INT_FAST16_MAX   INT16_MAX
+#define INT_FAST32_MIN   INT32_MIN
+#define INT_FAST32_MAX   INT32_MAX
+#define INT_FAST64_MIN   INT64_MIN
+#define INT_FAST64_MAX   INT64_MAX
+#define UINT_FAST8_MAX   UINT8_MAX
+#define UINT_FAST16_MAX  UINT16_MAX
+#define UINT_FAST32_MAX  UINT32_MAX
+#define UINT_FAST64_MAX  UINT64_MAX
+
+// 7.18.2.4 Limits of integer types capable of holding object pointers
+#ifdef _WIN64 // [
+#  define INTPTR_MIN   INT64_MIN
+#  define INTPTR_MAX   INT64_MAX
+#  define UINTPTR_MAX  UINT64_MAX
+#else // _WIN64 ][
+#  define INTPTR_MIN   INT32_MIN
+#  define INTPTR_MAX   INT32_MAX
+#  define UINTPTR_MAX  UINT32_MAX
+#endif // _WIN64 ]
+
+// 7.18.2.5 Limits of greatest-width integer types
+#define INTMAX_MIN   INT64_MIN
+#define INTMAX_MAX   INT64_MAX
+#define UINTMAX_MAX  UINT64_MAX
+
+// 7.18.3 Limits of other integer types
+
+#ifdef _WIN64 // [
+#  define PTRDIFF_MIN  _I64_MIN
+#  define PTRDIFF_MAX  _I64_MAX
+#else  // _WIN64 ][
+#  define PTRDIFF_MIN  _I32_MIN
+#  define PTRDIFF_MAX  _I32_MAX
+#endif  // _WIN64 ]
+
+#define SIG_ATOMIC_MIN  INT_MIN
+#define SIG_ATOMIC_MAX  INT_MAX
+
+#ifndef SIZE_MAX // [
+#  ifdef _WIN64 // [
+#     define SIZE_MAX  _UI64_MAX
+#  else // _WIN64 ][
+#     define SIZE_MAX  _UI32_MAX
+#  endif // _WIN64 ]
+#endif // SIZE_MAX ]
+
+// WCHAR_MIN and WCHAR_MAX are also defined in <wchar.h>
+#ifndef WCHAR_MIN // [
+#  define WCHAR_MIN  0
+#endif  // WCHAR_MIN ]
+#ifndef WCHAR_MAX // [
+#  define WCHAR_MAX  _UI16_MAX
+#endif  // WCHAR_MAX ]
+
+#define WINT_MIN  0
+#define WINT_MAX  _UI16_MAX
+
+#endif // __STDC_LIMIT_MACROS ]
+
+
+// 7.18.4 Limits of other integer types
+
+#if !defined(__cplusplus) || defined(__STDC_CONSTANT_MACROS) // [   See footnote 224 at page 260
+
+// 7.18.4.1 Macros for minimum-width integer constants
+
+#define INT8_C(val)  val##i8
+#define INT16_C(val) val##i16
+#define INT32_C(val) val##i32
+#define INT64_C(val) val##i64
+
+#define UINT8_C(val)  val##ui8
+#define UINT16_C(val) val##ui16
+#define UINT32_C(val) val##ui32
+#define UINT64_C(val) val##ui64
+
+// 7.18.4.2 Macros for greatest-width integer constants
+#define INTMAX_C   INT64_C
+#define UINTMAX_C  UINT64_C
+
+#endif // __STDC_CONSTANT_MACROS ]
+
+
+#endif // _MSC_STDINT_H_ ]
diff --git a/jemalloc/include/msvc_compat/strings.h b/jemalloc/include/msvc_compat/strings.h
new file mode 100644
index 0000000..996f256
--- /dev/null
+++ b/jemalloc/include/msvc_compat/strings.h
@@ -0,0 +1,58 @@
+#ifndef strings_h
+#define strings_h
+
+/* MSVC doesn't define ffs/ffsl. This dummy strings.h header is provided
+ * for both */
+#ifdef _MSC_VER
+#  include <intrin.h>
+#  pragma intrinsic(_BitScanForward)
+static __forceinline int ffsl(long x) {
+	unsigned long i;
+
+	if (_BitScanForward(&i, x)) {
+		return i + 1;
+	}
+	return 0;
+}
+
+static __forceinline int ffs(int x) {
+	return ffsl(x);
+}
+
+#  ifdef  _M_X64
+#    pragma intrinsic(_BitScanForward64)
+#  endif
+
+static __forceinline int ffsll(unsigned __int64 x) {
+	unsigned long i;
+#ifdef  _M_X64
+	if (_BitScanForward64(&i, x)) {
+		return i + 1;
+	}
+	return 0;
+#else
+// Fallback for 32-bit build where 64-bit version not available
+// assuming little endian
+	union {
+		unsigned __int64 ll;
+		unsigned   long l[2];
+	} s;
+
+	s.ll = x;
+
+	if (_BitScanForward(&i, s.l[0])) {
+		return i + 1;
+	} else if(_BitScanForward(&i, s.l[1])) {
+		return i + 33;
+	}
+	return 0;
+#endif
+}
+
+#else
+#  define ffsll(x) __builtin_ffsll(x)
+#  define ffsl(x) __builtin_ffsl(x)
+#  define ffs(x) __builtin_ffs(x)
+#endif
+
+#endif /* strings_h */
diff --git a/jemalloc/include/msvc_compat/windows_extra.h b/jemalloc/include/msvc_compat/windows_extra.h
new file mode 100644
index 0000000..a6ebb93
--- /dev/null
+++ b/jemalloc/include/msvc_compat/windows_extra.h
@@ -0,0 +1,6 @@
+#ifndef MSVC_COMPAT_WINDOWS_EXTRA_H
+#define MSVC_COMPAT_WINDOWS_EXTRA_H
+
+#include <errno.h>
+
+#endif /* MSVC_COMPAT_WINDOWS_EXTRA_H */
diff --git a/jemalloc/jemalloc.pc.in b/jemalloc/jemalloc.pc.in
new file mode 100644
index 0000000..c428a86
--- /dev/null
+++ b/jemalloc/jemalloc.pc.in
@@ -0,0 +1,12 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+install_suffix=@install_suffix@
+
+Name: jemalloc
+Description: A general purpose malloc(3) implementation that emphasizes fragmentation avoidance and scalable concurrency support.
+URL: http://jemalloc.net/
+Version: @jemalloc_version_major@.@jemalloc_version_minor@.@jemalloc_version_bugfix@_@jemalloc_version_nrev@
+Cflags: -I${includedir}
+Libs: -L${libdir} -ljemalloc${install_suffix}
diff --git a/jemalloc/m4/ax_cxx_compile_stdcxx.m4 b/jemalloc/m4/ax_cxx_compile_stdcxx.m4
new file mode 100644
index 0000000..2c18e49
--- /dev/null
+++ b/jemalloc/m4/ax_cxx_compile_stdcxx.m4
@@ -0,0 +1,562 @@
+# ===========================================================================
+#   http://www.gnu.org/software/autoconf-archive/ax_cxx_compile_stdcxx.html
+# ===========================================================================
+#
+# SYNOPSIS
+#
+#   AX_CXX_COMPILE_STDCXX(VERSION, [ext|noext], [mandatory|optional])
+#
+# DESCRIPTION
+#
+#   Check for baseline language coverage in the compiler for the specified
+#   version of the C++ standard.  If necessary, add switches to CXX and
+#   CXXCPP to enable support.  VERSION may be '11' (for the C++11 standard)
+#   or '14' (for the C++14 standard).
+#
+#   The second argument, if specified, indicates whether you insist on an
+#   extended mode (e.g. -std=gnu++11) or a strict conformance mode (e.g.
+#   -std=c++11).  If neither is specified, you get whatever works, with
+#   preference for an extended mode.
+#
+#   The third argument, if specified 'mandatory' or if left unspecified,
+#   indicates that baseline support for the specified C++ standard is
+#   required and that the macro should error out if no mode with that
+#   support is found.  If specified 'optional', then configuration proceeds
+#   regardless, after defining HAVE_CXX${VERSION} if and only if a
+#   supporting mode is found.
+#
+# LICENSE
+#
+#   Copyright (c) 2008 Benjamin Kosnik <bkoz@redhat.com>
+#   Copyright (c) 2012 Zack Weinberg <zackw@panix.com>
+#   Copyright (c) 2013 Roy Stogner <roystgnr@ices.utexas.edu>
+#   Copyright (c) 2014, 2015 Google Inc.; contributed by Alexey Sokolov <sokolov@google.com>
+#   Copyright (c) 2015 Paul Norman <penorman@mac.com>
+#   Copyright (c) 2015 Moritz Klammler <moritz@klammler.eu>
+#
+#   Copying and distribution of this file, with or without modification, are
+#   permitted in any medium without royalty provided the copyright notice
+#   and this notice are preserved.  This file is offered as-is, without any
+#   warranty.
+
+#serial 4
+
+dnl  This macro is based on the code from the AX_CXX_COMPILE_STDCXX_11 macro
+dnl  (serial version number 13).
+
+AC_DEFUN([AX_CXX_COMPILE_STDCXX], [dnl
+  m4_if([$1], [11], [],
+        [$1], [14], [],
+        [$1], [17], [m4_fatal([support for C++17 not yet implemented in AX_CXX_COMPILE_STDCXX])],
+        [m4_fatal([invalid first argument `$1' to AX_CXX_COMPILE_STDCXX])])dnl
+  m4_if([$2], [], [],
+        [$2], [ext], [],
+        [$2], [noext], [],
+        [m4_fatal([invalid second argument `$2' to AX_CXX_COMPILE_STDCXX])])dnl
+  m4_if([$3], [], [ax_cxx_compile_cxx$1_required=true],
+        [$3], [mandatory], [ax_cxx_compile_cxx$1_required=true],
+        [$3], [optional], [ax_cxx_compile_cxx$1_required=false],
+        [m4_fatal([invalid third argument `$3' to AX_CXX_COMPILE_STDCXX])])
+  AC_LANG_PUSH([C++])dnl
+  ac_success=no
+  AC_CACHE_CHECK(whether $CXX supports C++$1 features by default,
+  ax_cv_cxx_compile_cxx$1,
+  [AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_testbody_$1])],
+    [ax_cv_cxx_compile_cxx$1=yes],
+    [ax_cv_cxx_compile_cxx$1=no])])
+  if test x$ax_cv_cxx_compile_cxx$1 = xyes; then
+    ac_success=yes
+  fi
+
+  m4_if([$2], [noext], [], [dnl
+  if test x$ac_success = xno; then
+    for switch in -std=gnu++$1 -std=gnu++0x; do
+      cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx$1_$switch])
+      AC_CACHE_CHECK(whether $CXX supports C++$1 features with $switch,
+                     $cachevar,
+        [ac_save_CXX="$CXX"
+         CXX="$CXX $switch"
+         AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_testbody_$1])],
+          [eval $cachevar=yes],
+          [eval $cachevar=no])
+         CXX="$ac_save_CXX"])
+      if eval test x\$$cachevar = xyes; then
+        CXX="$CXX $switch"
+        if test -n "$CXXCPP" ; then
+          CXXCPP="$CXXCPP $switch"
+        fi
+        ac_success=yes
+        break
+      fi
+    done
+  fi])
+
+  m4_if([$2], [ext], [], [dnl
+  if test x$ac_success = xno; then
+    dnl HP's aCC needs +std=c++11 according to:
+    dnl http://h21007.www2.hp.com/portal/download/files/unprot/aCxx/PDF_Release_Notes/769149-001.pdf
+    dnl Cray's crayCC needs "-h std=c++11"
+    for switch in -std=c++$1 -std=c++0x +std=c++$1 "-h std=c++$1"; do
+      cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx$1_$switch])
+      AC_CACHE_CHECK(whether $CXX supports C++$1 features with $switch,
+                     $cachevar,
+        [ac_save_CXX="$CXX"
+         CXX="$CXX $switch"
+         AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_testbody_$1])],
+          [eval $cachevar=yes],
+          [eval $cachevar=no])
+         CXX="$ac_save_CXX"])
+      if eval test x\$$cachevar = xyes; then
+        CXX="$CXX $switch"
+        if test -n "$CXXCPP" ; then
+          CXXCPP="$CXXCPP $switch"
+        fi
+        ac_success=yes
+        break
+      fi
+    done
+  fi])
+  AC_LANG_POP([C++])
+  if test x$ax_cxx_compile_cxx$1_required = xtrue; then
+    if test x$ac_success = xno; then
+      AC_MSG_ERROR([*** A compiler with support for C++$1 language features is required.])
+    fi
+  fi
+  if test x$ac_success = xno; then
+    HAVE_CXX$1=0
+    AC_MSG_NOTICE([No compiler with C++$1 support was found])
+  else
+    HAVE_CXX$1=1
+    AC_DEFINE(HAVE_CXX$1,1,
+              [define if the compiler supports basic C++$1 syntax])
+  fi
+  AC_SUBST(HAVE_CXX$1)
+])
+
+
+dnl  Test body for checking C++11 support
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_11],
+  _AX_CXX_COMPILE_STDCXX_testbody_new_in_11
+)
+
+
+dnl  Test body for checking C++14 support
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_14],
+  _AX_CXX_COMPILE_STDCXX_testbody_new_in_11
+  _AX_CXX_COMPILE_STDCXX_testbody_new_in_14
+)
+
+
+dnl  Tests for new features in C++11
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_new_in_11], [[
+
+// If the compiler admits that it is not ready for C++11, why torture it?
+// Hopefully, this will speed up the test.
+
+#ifndef __cplusplus
+
+#error "This is not a C++ compiler"
+
+#elif __cplusplus < 201103L
+
+#error "This is not a C++11 compiler"
+
+#else
+
+namespace cxx11
+{
+
+  namespace test_static_assert
+  {
+
+    template <typename T>
+    struct check
+    {
+      static_assert(sizeof(int) <= sizeof(T), "not big enough");
+    };
+
+  }
+
+  namespace test_final_override
+  {
+
+    struct Base
+    {
+      virtual void f() {}
+    };
+
+    struct Derived : public Base
+    {
+      virtual void f() override {}
+    };
+
+  }
+
+  namespace test_double_right_angle_brackets
+  {
+
+    template < typename T >
+    struct check {};
+
+    typedef check<void> single_type;
+    typedef check<check<void>> double_type;
+    typedef check<check<check<void>>> triple_type;
+    typedef check<check<check<check<void>>>> quadruple_type;
+
+  }
+
+  namespace test_decltype
+  {
+
+    int
+    f()
+    {
+      int a = 1;
+      decltype(a) b = 2;
+      return a + b;
+    }
+
+  }
+
+  namespace test_type_deduction
+  {
+
+    template < typename T1, typename T2 >
+    struct is_same
+    {
+      static const bool value = false;
+    };
+
+    template < typename T >
+    struct is_same<T, T>
+    {
+      static const bool value = true;
+    };
+
+    template < typename T1, typename T2 >
+    auto
+    add(T1 a1, T2 a2) -> decltype(a1 + a2)
+    {
+      return a1 + a2;
+    }
+
+    int
+    test(const int c, volatile int v)
+    {
+      static_assert(is_same<int, decltype(0)>::value == true, "");
+      static_assert(is_same<int, decltype(c)>::value == false, "");
+      static_assert(is_same<int, decltype(v)>::value == false, "");
+      auto ac = c;
+      auto av = v;
+      auto sumi = ac + av + 'x';
+      auto sumf = ac + av + 1.0;
+      static_assert(is_same<int, decltype(ac)>::value == true, "");
+      static_assert(is_same<int, decltype(av)>::value == true, "");
+      static_assert(is_same<int, decltype(sumi)>::value == true, "");
+      static_assert(is_same<int, decltype(sumf)>::value == false, "");
+      static_assert(is_same<int, decltype(add(c, v))>::value == true, "");
+      return (sumf > 0.0) ? sumi : add(c, v);
+    }
+
+  }
+
+  namespace test_noexcept
+  {
+
+    int f() { return 0; }
+    int g() noexcept { return 0; }
+
+    static_assert(noexcept(f()) == false, "");
+    static_assert(noexcept(g()) == true, "");
+
+  }
+
+  namespace test_constexpr
+  {
+
+    template < typename CharT >
+    unsigned long constexpr
+    strlen_c_r(const CharT *const s, const unsigned long acc) noexcept
+    {
+      return *s ? strlen_c_r(s + 1, acc + 1) : acc;
+    }
+
+    template < typename CharT >
+    unsigned long constexpr
+    strlen_c(const CharT *const s) noexcept
+    {
+      return strlen_c_r(s, 0UL);
+    }
+
+    static_assert(strlen_c("") == 0UL, "");
+    static_assert(strlen_c("1") == 1UL, "");
+    static_assert(strlen_c("example") == 7UL, "");
+    static_assert(strlen_c("another\0example") == 7UL, "");
+
+  }
+
+  namespace test_rvalue_references
+  {
+
+    template < int N >
+    struct answer
+    {
+      static constexpr int value = N;
+    };
+
+    answer<1> f(int&)       { return answer<1>(); }
+    answer<2> f(const int&) { return answer<2>(); }
+    answer<3> f(int&&)      { return answer<3>(); }
+
+    void
+    test()
+    {
+      int i = 0;
+      const int c = 0;
+      static_assert(decltype(f(i))::value == 1, "");
+      static_assert(decltype(f(c))::value == 2, "");
+      static_assert(decltype(f(0))::value == 3, "");
+    }
+
+  }
+
+  namespace test_uniform_initialization
+  {
+
+    struct test
+    {
+      static const int zero {};
+      static const int one {1};
+    };
+
+    static_assert(test::zero == 0, "");
+    static_assert(test::one == 1, "");
+
+  }
+
+  namespace test_lambdas
+  {
+
+    void
+    test1()
+    {
+      auto lambda1 = [](){};
+      auto lambda2 = lambda1;
+      lambda1();
+      lambda2();
+    }
+
+    int
+    test2()
+    {
+      auto a = [](int i, int j){ return i + j; }(1, 2);
+      auto b = []() -> int { return '0'; }();
+      auto c = [=](){ return a + b; }();
+      auto d = [&](){ return c; }();
+      auto e = [a, &b](int x) mutable {
+        const auto identity = [](int y){ return y; };
+        for (auto i = 0; i < a; ++i)
+          a += b--;
+        return x + identity(a + b);
+      }(0);
+      return a + b + c + d + e;
+    }
+
+    int
+    test3()
+    {
+      const auto nullary = [](){ return 0; };
+      const auto unary = [](int x){ return x; };
+      using nullary_t = decltype(nullary);
+      using unary_t = decltype(unary);
+      const auto higher1st = [](nullary_t f){ return f(); };
+      const auto higher2nd = [unary](nullary_t f1){
+        return [unary, f1](unary_t f2){ return f2(unary(f1())); };
+      };
+      return higher1st(nullary) + higher2nd(nullary)(unary);
+    }
+
+  }
+
+  namespace test_variadic_templates
+  {
+
+    template <int...>
+    struct sum;
+
+    template <int N0, int... N1toN>
+    struct sum<N0, N1toN...>
+    {
+      static constexpr auto value = N0 + sum<N1toN...>::value;
+    };
+
+    template <>
+    struct sum<>
+    {
+      static constexpr auto value = 0;
+    };
+
+    static_assert(sum<>::value == 0, "");
+    static_assert(sum<1>::value == 1, "");
+    static_assert(sum<23>::value == 23, "");
+    static_assert(sum<1, 2>::value == 3, "");
+    static_assert(sum<5, 5, 11>::value == 21, "");
+    static_assert(sum<2, 3, 5, 7, 11, 13>::value == 41, "");
+
+  }
+
+  // http://stackoverflow.com/questions/13728184/template-aliases-and-sfinae
+  // Clang 3.1 fails with headers of libstd++ 4.8.3 when using std::function
+  // because of this.
+  namespace test_template_alias_sfinae
+  {
+
+    struct foo {};
+
+    template<typename T>
+    using member = typename T::member_type;
+
+    template<typename T>
+    void func(...) {}
+
+    template<typename T>
+    void func(member<T>*) {}
+
+    void test();
+
+    void test() { func<foo>(0); }
+
+  }
+
+}  // namespace cxx11
+
+#endif  // __cplusplus >= 201103L
+
+]])
+
+
+dnl  Tests for new features in C++14
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_new_in_14], [[
+
+// If the compiler admits that it is not ready for C++14, why torture it?
+// Hopefully, this will speed up the test.
+
+#ifndef __cplusplus
+
+#error "This is not a C++ compiler"
+
+#elif __cplusplus < 201402L
+
+#error "This is not a C++14 compiler"
+
+#else
+
+namespace cxx14
+{
+
+  namespace test_polymorphic_lambdas
+  {
+
+    int
+    test()
+    {
+      const auto lambda = [](auto&&... args){
+        const auto istiny = [](auto x){
+          return (sizeof(x) == 1UL) ? 1 : 0;
+        };
+        const int aretiny[] = { istiny(args)... };
+        return aretiny[0];
+      };
+      return lambda(1, 1L, 1.0f, '1');
+    }
+
+  }
+
+  namespace test_binary_literals
+  {
+
+    constexpr auto ivii = 0b0000000000101010;
+    static_assert(ivii == 42, "wrong value");
+
+  }
+
+  namespace test_generalized_constexpr
+  {
+
+    template < typename CharT >
+    constexpr unsigned long
+    strlen_c(const CharT *const s) noexcept
+    {
+      auto length = 0UL;
+      for (auto p = s; *p; ++p)
+        ++length;
+      return length;
+    }
+
+    static_assert(strlen_c("") == 0UL, "");
+    static_assert(strlen_c("x") == 1UL, "");
+    static_assert(strlen_c("test") == 4UL, "");
+    static_assert(strlen_c("another\0test") == 7UL, "");
+
+  }
+
+  namespace test_lambda_init_capture
+  {
+
+    int
+    test()
+    {
+      auto x = 0;
+      const auto lambda1 = [a = x](int b){ return a + b; };
+      const auto lambda2 = [a = lambda1(x)](){ return a; };
+      return lambda2();
+    }
+
+  }
+
+  namespace test_digit_seperators
+  {
+
+    constexpr auto ten_million = 100'000'000;
+    static_assert(ten_million == 100000000, "");
+
+  }
+
+  namespace test_return_type_deduction
+  {
+
+    auto f(int& x) { return x; }
+    decltype(auto) g(int& x) { return x; }
+
+    template < typename T1, typename T2 >
+    struct is_same
+    {
+      static constexpr auto value = false;
+    };
+
+    template < typename T >
+    struct is_same<T, T>
+    {
+      static constexpr auto value = true;
+    };
+
+    int
+    test()
+    {
+      auto x = 0;
+      static_assert(is_same<int, decltype(f(x))>::value, "");
+      static_assert(is_same<int&, decltype(g(x))>::value, "");
+      return x;
+    }
+
+  }
+
+}  // namespace cxx14
+
+#endif  // __cplusplus >= 201402L
+
+]])
diff --git a/jemalloc/msvc/ReadMe.txt b/jemalloc/msvc/ReadMe.txt
new file mode 100644
index 0000000..633a7d4
--- /dev/null
+++ b/jemalloc/msvc/ReadMe.txt
@@ -0,0 +1,23 @@
+
+How to build jemalloc for Windows
+=================================
+
+1. Install Cygwin with at least the following packages:
+   * autoconf
+   * autogen
+   * gawk
+   * grep
+   * sed
+
+2. Install Visual Studio 2015 or 2017 with Visual C++
+
+3. Add Cygwin\bin to the PATH environment variable
+
+4. Open "x64 Native Tools Command Prompt for VS 2017"
+   (note: x86/x64 doesn't matter at this point)
+
+5. Generate header files:
+   sh -c "CC=cl ./autogen.sh"
+
+6. Now the project can be opened and built in Visual Studio:
+   msvc\jemalloc_vc2017.sln
diff --git a/jemalloc/msvc/jemalloc_vc2015.sln b/jemalloc/msvc/jemalloc_vc2015.sln
new file mode 100644
index 0000000..aedd5e5
--- /dev/null
+++ b/jemalloc/msvc/jemalloc_vc2015.sln
@@ -0,0 +1,63 @@
+
+Microsoft Visual Studio Solution File, Format Version 12.00
+# Visual Studio 14
+VisualStudioVersion = 14.0.24720.0
+MinimumVisualStudioVersion = 10.0.40219.1
+Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "Solution Items", "Solution Items", "{70A99006-6DE9-472B-8F83-4CEE6C616DF3}"
+	ProjectSection(SolutionItems) = preProject
+		ReadMe.txt = ReadMe.txt
+	EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "jemalloc", "projects\vc2015\jemalloc\jemalloc.vcxproj", "{8D6BB292-9E1C-413D-9F98-4864BDC1514A}"
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "test_threads", "projects\vc2015\test_threads\test_threads.vcxproj", "{09028CFD-4EB7-491D-869C-0708DB97ED44}"
+EndProject
+Global
+	GlobalSection(SolutionConfigurationPlatforms) = preSolution
+		Debug|x64 = Debug|x64
+		Debug|x86 = Debug|x86
+		Debug-static|x64 = Debug-static|x64
+		Debug-static|x86 = Debug-static|x86
+		Release|x64 = Release|x64
+		Release|x86 = Release|x86
+		Release-static|x64 = Release-static|x64
+		Release-static|x86 = Release-static|x86
+	EndGlobalSection
+	GlobalSection(ProjectConfigurationPlatforms) = postSolution
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.ActiveCfg = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.Build.0 = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.ActiveCfg = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.Build.0 = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.Build.0 = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.ActiveCfg = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.Build.0 = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.ActiveCfg = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.Build.0 = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.ActiveCfg = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.Build.0 = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.Build.0 = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.ActiveCfg = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.Build.0 = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.ActiveCfg = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.Build.0 = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.Build.0 = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.ActiveCfg = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.Build.0 = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.ActiveCfg = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.Build.0 = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.ActiveCfg = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.Build.0 = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.Build.0 = Release-static|Win32
+	EndGlobalSection
+	GlobalSection(SolutionProperties) = preSolution
+		HideSolutionNode = FALSE
+	EndGlobalSection
+EndGlobal
diff --git a/jemalloc/msvc/jemalloc_vc2017.sln b/jemalloc/msvc/jemalloc_vc2017.sln
new file mode 100644
index 0000000..c22fcb4
--- /dev/null
+++ b/jemalloc/msvc/jemalloc_vc2017.sln
@@ -0,0 +1,63 @@
+
+Microsoft Visual Studio Solution File, Format Version 12.00
+# Visual Studio 14
+VisualStudioVersion = 14.0.24720.0
+MinimumVisualStudioVersion = 10.0.40219.1
+Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "Solution Items", "Solution Items", "{70A99006-6DE9-472B-8F83-4CEE6C616DF3}"
+	ProjectSection(SolutionItems) = preProject
+		ReadMe.txt = ReadMe.txt
+	EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "jemalloc", "projects\vc2017\jemalloc\jemalloc.vcxproj", "{8D6BB292-9E1C-413D-9F98-4864BDC1514A}"
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "test_threads", "projects\vc2017\test_threads\test_threads.vcxproj", "{09028CFD-4EB7-491D-869C-0708DB97ED44}"
+EndProject
+Global
+	GlobalSection(SolutionConfigurationPlatforms) = preSolution
+		Debug|x64 = Debug|x64
+		Debug|x86 = Debug|x86
+		Debug-static|x64 = Debug-static|x64
+		Debug-static|x86 = Debug-static|x86
+		Release|x64 = Release|x64
+		Release|x86 = Release|x86
+		Release-static|x64 = Release-static|x64
+		Release-static|x86 = Release-static|x86
+	EndGlobalSection
+	GlobalSection(ProjectConfigurationPlatforms) = postSolution
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.ActiveCfg = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.Build.0 = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.ActiveCfg = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.Build.0 = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.Build.0 = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.ActiveCfg = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.Build.0 = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.ActiveCfg = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.Build.0 = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.ActiveCfg = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.Build.0 = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.Build.0 = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.ActiveCfg = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.Build.0 = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.ActiveCfg = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.Build.0 = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.Build.0 = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.ActiveCfg = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.Build.0 = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.ActiveCfg = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.Build.0 = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.ActiveCfg = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.Build.0 = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.Build.0 = Release-static|Win32
+	EndGlobalSection
+	GlobalSection(SolutionProperties) = preSolution
+		HideSolutionNode = FALSE
+	EndGlobalSection
+EndGlobal
diff --git a/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj b/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
new file mode 100644
index 0000000..228e8be
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
@@ -0,0 +1,350 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="14.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c" />
+    <ClCompile Include="..\..\..\..\src\background_thread.c" />
+    <ClCompile Include="..\..\..\..\src\base.c" />
+    <ClCompile Include="..\..\..\..\src\bin.c" />
+    <ClCompile Include="..\..\..\..\src\bitmap.c" />
+    <ClCompile Include="..\..\..\..\src\ckh.c" />
+    <ClCompile Include="..\..\..\..\src\ctl.c" />
+    <ClCompile Include="..\..\..\..\src\div.c" />
+    <ClCompile Include="..\..\..\..\src\extent.c" />
+    <ClCompile Include="..\..\..\..\src\extent_dss.c" />
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c" />
+    <ClCompile Include="..\..\..\..\src\hash.c" />
+    <ClCompile Include="..\..\..\..\src\hook.c" />
+    <ClCompile Include="..\..\..\..\src\jemalloc.c" />
+    <ClCompile Include="..\..\..\..\src\large.c" />
+    <ClCompile Include="..\..\..\..\src\log.c" />
+    <ClCompile Include="..\..\..\..\src\malloc_io.c" />
+    <ClCompile Include="..\..\..\..\src\mutex.c" />
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c" />
+    <ClCompile Include="..\..\..\..\src\nstime.c" />
+    <ClCompile Include="..\..\..\..\src\pages.c" />
+    <ClCompile Include="..\..\..\..\src\prng.c" />
+    <ClCompile Include="..\..\..\..\src\prof.c" />
+    <ClCompile Include="..\..\..\..\src\rtree.c" />
+    <ClCompile Include="..\..\..\..\src\sc.c" />
+    <ClCompile Include="..\..\..\..\src\stats.c" />
+    <ClCompile Include="..\..\..\..\src\sz.c" />
+    <ClCompile Include="..\..\..\..\src\tcache.c" />
+    <ClCompile Include="..\..\..\..\src\ticker.c" />
+    <ClCompile Include="..\..\..\..\src\tsd.c" />
+    <ClCompile Include="..\..\..\..\src\witness.c" />
+    <ClCompile Include="..\..\..\..\src\safety_check.c" />
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{8D6BB292-9E1C-413D-9F98-4864BDC1514A}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>jemalloc</RootNamespace>
+    <WindowsTargetPlatformVersion>8.1</WindowsTargetPlatformVersion>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+      <MinimalRebuild>false</MinimalRebuild>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
diff --git a/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters b/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
new file mode 100644
index 0000000..d839515
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
@@ -0,0 +1,107 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\background_thread.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\base.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bitmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ckh.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ctl.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_dss.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hash.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hook.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\jemalloc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\large.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\malloc_io.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\nstime.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\pages.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prng.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prof.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\rtree.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\stats.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sz.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tcache.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ticker.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tsd.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\witness.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\log.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bin.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\div.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\safety_check.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+</Project>
diff --git a/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj b/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
new file mode 100644
index 0000000..325876d
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
@@ -0,0 +1,327 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="14.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{09028CFD-4EB7-491D-869C-0708DB97ED44}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>test_threads</RootNamespace>
+    <WindowsTargetPlatformVersion>8.1</WindowsTargetPlatformVersion>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp" />
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="..\jemalloc\jemalloc.vcxproj">
+      <Project>{8d6bb292-9e1c-413d-9f98-4864bdc1514a}</Project>
+    </ProjectReference>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h" />
+  </ItemGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
\ No newline at end of file
diff --git a/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters b/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
new file mode 100644
index 0000000..fa4588f
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
@@ -0,0 +1,26 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+    <Filter Include="Header Files">
+      <UniqueIdentifier>{93995380-89BD-4b04-88EB-625FBE52EBFB}</UniqueIdentifier>
+      <Extensions>h;hh;hpp;hxx;hm;inl;inc;xsd</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+  </ItemGroup>
+</Project>
\ No newline at end of file
diff --git a/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj b/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
new file mode 100644
index 0000000..edcceed
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
@@ -0,0 +1,350 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="15.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c" />
+    <ClCompile Include="..\..\..\..\src\background_thread.c" />
+    <ClCompile Include="..\..\..\..\src\base.c" />
+    <ClCompile Include="..\..\..\..\src\bin.c" />
+    <ClCompile Include="..\..\..\..\src\bitmap.c" />
+    <ClCompile Include="..\..\..\..\src\ckh.c" />
+    <ClCompile Include="..\..\..\..\src\ctl.c" />
+    <ClCompile Include="..\..\..\..\src\div.c" />
+    <ClCompile Include="..\..\..\..\src\extent.c" />
+    <ClCompile Include="..\..\..\..\src\extent_dss.c" />
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c" />
+    <ClCompile Include="..\..\..\..\src\hash.c" />
+    <ClCompile Include="..\..\..\..\src\hook.c" />
+    <ClCompile Include="..\..\..\..\src\jemalloc.c" />
+    <ClCompile Include="..\..\..\..\src\large.c" />
+    <ClCompile Include="..\..\..\..\src\log.c" />
+    <ClCompile Include="..\..\..\..\src\malloc_io.c" />
+    <ClCompile Include="..\..\..\..\src\mutex.c" />
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c" />
+    <ClCompile Include="..\..\..\..\src\nstime.c" />
+    <ClCompile Include="..\..\..\..\src\pages.c" />
+    <ClCompile Include="..\..\..\..\src\prng.c" />
+    <ClCompile Include="..\..\..\..\src\prof.c" />
+    <ClCompile Include="..\..\..\..\src\rtree.c" />
+    <ClCompile Include="..\..\..\..\src\sc.c" />
+    <ClCompile Include="..\..\..\..\src\stats.c" />
+    <ClCompile Include="..\..\..\..\src\sz.c" />
+    <ClCompile Include="..\..\..\..\src\tcache.c" />
+    <ClCompile Include="..\..\..\..\src\test_hooks.c" />
+    <ClCompile Include="..\..\..\..\src\ticker.c" />
+    <ClCompile Include="..\..\..\..\src\tsd.c" />
+    <ClCompile Include="..\..\..\..\src\witness.c" />
+    <ClCompile Include="..\..\..\..\src\safety_check.c" />
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{8D6BB292-9E1C-413D-9F98-4864BDC1514A}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>jemalloc</RootNamespace>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+      <MinimalRebuild>false</MinimalRebuild>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
diff --git a/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters b/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
new file mode 100644
index 0000000..6df7260
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
@@ -0,0 +1,110 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\background_thread.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\base.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bitmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ckh.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ctl.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_dss.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hash.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hook.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\jemalloc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\large.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\malloc_io.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\nstime.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\pages.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prng.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prof.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\rtree.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\stats.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sz.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tcache.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ticker.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tsd.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\witness.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\log.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bin.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\div.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\test_hooks.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\safety_check.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+</Project>
diff --git a/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj b/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
new file mode 100644
index 0000000..c35b0f5
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
@@ -0,0 +1,326 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="15.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{09028CFD-4EB7-491D-869C-0708DB97ED44}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>test_threads</RootNamespace>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp" />
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="..\jemalloc\jemalloc.vcxproj">
+      <Project>{8d6bb292-9e1c-413d-9f98-4864bdc1514a}</Project>
+    </ProjectReference>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h" />
+  </ItemGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
\ No newline at end of file
diff --git a/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters b/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
new file mode 100644
index 0000000..fa4588f
--- /dev/null
+++ b/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
@@ -0,0 +1,26 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+    <Filter Include="Header Files">
+      <UniqueIdentifier>{93995380-89BD-4b04-88EB-625FBE52EBFB}</UniqueIdentifier>
+      <Extensions>h;hh;hpp;hxx;hm;inl;inc;xsd</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+  </ItemGroup>
+</Project>
\ No newline at end of file
diff --git a/jemalloc/msvc/test_threads/test_threads.cpp b/jemalloc/msvc/test_threads/test_threads.cpp
new file mode 100644
index 0000000..92e3162
--- /dev/null
+++ b/jemalloc/msvc/test_threads/test_threads.cpp
@@ -0,0 +1,88 @@
+// jemalloc C++ threaded test
+// Author: Rustam Abdullaev
+// Public Domain
+
+#include <atomic>
+#include <functional>
+#include <future>
+#include <random>
+#include <thread>
+#include <vector>
+#include <stdio.h>
+#include <jemalloc/jemalloc.h>
+
+using std::vector;
+using std::thread;
+using std::uniform_int_distribution;
+using std::minstd_rand;
+
+int test_threads() {
+  je_malloc_conf = "narenas:3";
+  int narenas = 0;
+  size_t sz = sizeof(narenas);
+  je_mallctl("opt.narenas", (void *)&narenas, &sz, NULL, 0);
+  if (narenas != 3) {
+    printf("Error: unexpected number of arenas: %d\n", narenas);
+    return 1;
+  }
+  static const int sizes[] = { 7, 16, 32, 60, 91, 100, 120, 144, 169, 199, 255, 400, 670, 900, 917, 1025, 3333, 5190, 13131, 49192, 99999, 123123, 255265, 2333111 };
+  static const int numSizes = (int)(sizeof(sizes) / sizeof(sizes[0]));
+  vector<thread> workers;
+  static const int numThreads = narenas + 1, numAllocsMax = 25, numIter1 = 50, numIter2 = 50;
+  je_malloc_stats_print(NULL, NULL, NULL);
+  size_t allocated1;
+  size_t sz1 = sizeof(allocated1);
+  je_mallctl("stats.active", (void *)&allocated1, &sz1, NULL, 0);
+  printf("\nPress Enter to start threads...\n");
+  getchar();
+  printf("Starting %d threads x %d x %d iterations...\n", numThreads, numIter1, numIter2);
+  for (int i = 0; i < numThreads; i++) {
+    workers.emplace_back([tid=i]() {
+      uniform_int_distribution<int> sizeDist(0, numSizes - 1);
+      minstd_rand rnd(tid * 17);
+      uint8_t* ptrs[numAllocsMax];
+      int ptrsz[numAllocsMax];
+      for (int i = 0; i < numIter1; ++i) {
+        thread t([&]() {
+          for (int i = 0; i < numIter2; ++i) {
+            const int numAllocs = numAllocsMax - sizeDist(rnd);
+            for (int j = 0; j < numAllocs; j += 64) {
+              const int x = sizeDist(rnd);
+              const int sz = sizes[x];
+              ptrsz[j] = sz;
+              ptrs[j] = (uint8_t*)je_malloc(sz);
+              if (!ptrs[j]) {
+                printf("Unable to allocate %d bytes in thread %d, iter %d, alloc %d. %d\n", sz, tid, i, j, x);
+                exit(1);
+              }
+              for (int k = 0; k < sz; k++)
+                ptrs[j][k] = tid + k;
+            }
+            for (int j = 0; j < numAllocs; j += 64) {
+              for (int k = 0, sz = ptrsz[j]; k < sz; k++)
+                if (ptrs[j][k] != (uint8_t)(tid + k)) {
+                  printf("Memory error in thread %d, iter %d, alloc %d @ %d : %02X!=%02X\n", tid, i, j, k, ptrs[j][k], (uint8_t)(tid + k));
+                  exit(1);
+                }
+              je_free(ptrs[j]);
+            }
+          }
+        });
+        t.join();
+      }
+    });
+  }
+  for (thread& t : workers) {
+    t.join();
+  }
+  je_malloc_stats_print(NULL, NULL, NULL);
+  size_t allocated2;
+  je_mallctl("stats.active", (void *)&allocated2, &sz1, NULL, 0);
+  size_t leaked = allocated2 - allocated1;
+  printf("\nDone. Leaked: %zd bytes\n", leaked);
+  bool failed = leaked > 65536; // in case C++ runtime allocated something (e.g. iostream locale or facet)
+  printf("\nTest %s!\n", (failed ? "FAILED" : "successful"));
+  printf("\nPress Enter to continue...\n");
+  getchar();
+  return failed ? 1 : 0;
+}
diff --git a/jemalloc/msvc/test_threads/test_threads.h b/jemalloc/msvc/test_threads/test_threads.h
new file mode 100644
index 0000000..64d0cdb
--- /dev/null
+++ b/jemalloc/msvc/test_threads/test_threads.h
@@ -0,0 +1,3 @@
+#pragma once
+
+int test_threads();
diff --git a/jemalloc/msvc/test_threads/test_threads_main.cpp b/jemalloc/msvc/test_threads/test_threads_main.cpp
new file mode 100644
index 0000000..0a022fb
--- /dev/null
+++ b/jemalloc/msvc/test_threads/test_threads_main.cpp
@@ -0,0 +1,11 @@
+#include "test_threads.h"
+#include <future>
+#include <functional>
+#include <chrono>
+
+using namespace std::chrono_literals;
+
+int main(int argc, char** argv) {
+  int rc = test_threads();
+  return rc;
+}
diff --git a/jemalloc/run_tests.sh b/jemalloc/run_tests.sh
new file mode 100755
index 0000000..b434f15
--- /dev/null
+++ b/jemalloc/run_tests.sh
@@ -0,0 +1 @@
+$(dirname "$)")/scripts/gen_run_tests.py | bash
diff --git a/jemalloc/scripts/gen_run_tests.py b/jemalloc/scripts/gen_run_tests.py
new file mode 100755
index 0000000..a414f81
--- /dev/null
+++ b/jemalloc/scripts/gen_run_tests.py
@@ -0,0 +1,126 @@
+#!/usr/bin/env python
+
+import sys
+from itertools import combinations
+from os import uname
+from multiprocessing import cpu_count
+from subprocess import call
+
+# Later, we want to test extended vaddr support.  Apparently, the "real" way of
+# checking this is flaky on OS X.
+bits_64 = sys.maxsize > 2**32
+
+nparallel = cpu_count() * 2
+
+uname = uname()[0]
+
+if "BSD" in uname:
+    make_cmd = 'gmake'
+else:
+    make_cmd = 'make'
+
+def powerset(items):
+    result = []
+    for i in xrange(len(items) + 1):
+        result += combinations(items, i)
+    return result
+
+possible_compilers = []
+for cc, cxx in (['gcc', 'g++'], ['clang', 'clang++']):
+    try:
+        cmd_ret = call([cc, "-v"])
+        if cmd_ret == 0:
+            possible_compilers.append((cc, cxx))
+    except:
+        pass
+possible_compiler_opts = [
+    '-m32',
+]
+possible_config_opts = [
+    '--enable-debug',
+    '--enable-prof',
+    '--disable-stats',
+    '--enable-opt-safety-checks',
+]
+if bits_64:
+    possible_config_opts.append('--with-lg-vaddr=56')
+
+possible_malloc_conf_opts = [
+    'tcache:false',
+    'dss:primary',
+    'percpu_arena:percpu',
+    'background_thread:true',
+]
+
+print 'set -e'
+print 'if [ -f Makefile ] ; then %(make_cmd)s relclean ; fi' % {'make_cmd': make_cmd}
+print 'autoconf'
+print 'rm -rf run_tests.out'
+print 'mkdir run_tests.out'
+print 'cd run_tests.out'
+
+ind = 0
+for cc, cxx in possible_compilers:
+    for compiler_opts in powerset(possible_compiler_opts):
+        for config_opts in powerset(possible_config_opts):
+            for malloc_conf_opts in powerset(possible_malloc_conf_opts):
+                if cc is 'clang' \
+                  and '-m32' in possible_compiler_opts \
+                  and '--enable-prof' in config_opts:
+                    continue
+                config_line = (
+                    'EXTRA_CFLAGS=-Werror EXTRA_CXXFLAGS=-Werror '
+                    + 'CC="{} {}" '.format(cc, " ".join(compiler_opts))
+                    + 'CXX="{} {}" '.format(cxx, " ".join(compiler_opts))
+                    + '../../configure '
+                    + " ".join(config_opts) + (' --with-malloc-conf=' +
+                    ",".join(malloc_conf_opts) if len(malloc_conf_opts) > 0
+                    else '')
+                )
+
+                # We don't want to test large vaddr spaces in 32-bit mode.
+		if ('-m32' in compiler_opts and '--with-lg-vaddr=56' in
+                  config_opts):
+		    continue
+
+                # Per CPU arenas are only supported on Linux.
+                linux_supported = ('percpu_arena:percpu' in malloc_conf_opts \
+                  or 'background_thread:true' in malloc_conf_opts)
+                # Heap profiling and dss are not supported on OS X.
+                darwin_unsupported = ('--enable-prof' in config_opts or \
+                  'dss:primary' in malloc_conf_opts)
+                if (uname == 'Linux' and linux_supported) \
+                  or (not linux_supported and (uname != 'Darwin' or \
+                  not darwin_unsupported)):
+                    print """cat <<EOF > run_test_%(ind)d.sh
+#!/bin/sh
+
+set -e
+
+abort() {
+    echo "==> Error" >> run_test.log
+    echo "Error; see run_tests.out/run_test_%(ind)d.out/run_test.log"
+    exit 255 # Special exit code tells xargs to terminate.
+}
+
+# Environment variables are not supported.
+run_cmd() {
+    echo "==> \$@" >> run_test.log
+    \$@ >> run_test.log 2>&1 || abort
+}
+
+echo "=> run_test_%(ind)d: %(config_line)s"
+mkdir run_test_%(ind)d.out
+cd run_test_%(ind)d.out
+
+echo "==> %(config_line)s" >> run_test.log
+%(config_line)s >> run_test.log 2>&1 || abort
+
+run_cmd %(make_cmd)s all tests
+run_cmd %(make_cmd)s check
+run_cmd %(make_cmd)s distclean
+EOF
+chmod 755 run_test_%(ind)d.sh""" % {'ind': ind, 'config_line': config_line, 'make_cmd': make_cmd}
+                    ind += 1
+
+print 'for i in `seq 0 %(last_ind)d` ; do echo run_test_${i}.sh ; done | xargs -P %(nparallel)d -n 1 sh' % {'last_ind': ind-1, 'nparallel': nparallel}
diff --git a/jemalloc/scripts/gen_travis.py b/jemalloc/scripts/gen_travis.py
new file mode 100755
index 0000000..f1478c6
--- /dev/null
+++ b/jemalloc/scripts/gen_travis.py
@@ -0,0 +1,149 @@
+#!/usr/bin/env python
+
+from itertools import combinations
+
+travis_template = """\
+language: generic
+dist: precise
+
+matrix:
+  include:
+%s
+
+before_script:
+  - autoconf
+  - scripts/gen_travis.py > travis_script && diff .travis.yml travis_script
+  - ./configure ${COMPILER_FLAGS:+ \
+      CC="$CC $COMPILER_FLAGS" \
+      CXX="$CXX $COMPILER_FLAGS" } \
+      $CONFIGURE_FLAGS
+  - make -j3
+  - make -j3 tests
+
+script:
+  - make check
+"""
+
+# The 'default' configuration is gcc, on linux, with no compiler or configure
+# flags.  We also test with clang, -m32, --enable-debug, --enable-prof,
+# --disable-stats, and --with-malloc-conf=tcache:false.  To avoid abusing
+# travis though, we don't test all 2**7 = 128 possible combinations of these;
+# instead, we only test combinations of up to 2 'unusual' settings, under the
+# hope that bugs involving interactions of such settings are rare.
+# Things at once, for C(7, 0) + C(7, 1) + C(7, 2) = 29
+MAX_UNUSUAL_OPTIONS = 2
+
+os_default = 'linux'
+os_unusual = 'osx'
+
+compilers_default = 'CC=gcc CXX=g++'
+compilers_unusual = 'CC=clang CXX=clang++'
+
+compiler_flag_unusuals = ['-m32']
+
+configure_flag_unusuals = [
+    '--enable-debug',
+    '--enable-prof',
+    '--disable-stats',
+    '--disable-libdl',
+    '--enable-opt-safety-checks',
+]
+
+malloc_conf_unusuals = [
+    'tcache:false',
+    'dss:primary',
+    'percpu_arena:percpu',
+    'background_thread:true',
+]
+
+all_unusuals = (
+    [os_unusual] + [compilers_unusual] + compiler_flag_unusuals
+    + configure_flag_unusuals + malloc_conf_unusuals
+)
+
+unusual_combinations_to_test = []
+for i in xrange(MAX_UNUSUAL_OPTIONS + 1):
+    unusual_combinations_to_test += combinations(all_unusuals, i)
+
+gcc_multilib_set = False
+# Formats a job from a combination of flags
+def format_job(combination):
+    global gcc_multilib_set
+
+    os = os_unusual if os_unusual in combination else os_default
+    compilers = compilers_unusual if compilers_unusual in combination else compilers_default
+
+    compiler_flags = [x for x in combination if x in compiler_flag_unusuals]
+    configure_flags = [x for x in combination if x in configure_flag_unusuals]
+    malloc_conf = [x for x in combination if x in malloc_conf_unusuals]
+
+    # Filter out unsupported configurations on OS X.
+    if os == 'osx' and ('dss:primary' in malloc_conf or \
+      'percpu_arena:percpu' in malloc_conf or 'background_thread:true' \
+      in malloc_conf):
+        return ""
+    if len(malloc_conf) > 0:
+        configure_flags.append('--with-malloc-conf=' + ",".join(malloc_conf))
+
+    # Filter out an unsupported configuration - heap profiling on OS X.
+    if os == 'osx' and '--enable-prof' in configure_flags:
+        return ""
+
+    # We get some spurious errors when -Warray-bounds is enabled.
+    env_string = ('{} COMPILER_FLAGS="{}" CONFIGURE_FLAGS="{}" '
+	'EXTRA_CFLAGS="-Werror -Wno-array-bounds"').format(
+        compilers, " ".join(compiler_flags), " ".join(configure_flags))
+
+    job = ""
+    job += '    - os: %s\n' % os
+    job += '      env: %s\n' % env_string
+    if '-m32' in combination and os == 'linux':
+        job += '      addons:'
+        if gcc_multilib_set:
+            job += ' *gcc_multilib\n'
+        else:
+            job += ' &gcc_multilib\n'
+            job += '        apt:\n'
+            job += '          packages:\n'
+            job += '            - gcc-multilib\n'
+            gcc_multilib_set = True
+    return job
+
+include_rows = ""
+for combination in unusual_combinations_to_test:
+    include_rows += format_job(combination)
+
+# Development build
+include_rows += '''\
+    # Development build
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-cache-oblivious --enable-stats --enable-log --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+'''
+
+# Enable-expermental-smallocx
+include_rows += '''\
+    # --enable-expermental-smallocx:
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-experimental-smallocx --enable-stats --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+'''
+
+# Valgrind build bots
+include_rows += '''
+    # Valgrind
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds" JEMALLOC_TEST_PREFIX="valgrind"
+      addons:
+        apt:
+          packages:
+            - valgrind
+'''
+
+# To enable valgrind on macosx add:
+#
+#  - os: osx
+#    env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds" JEMALLOC_TEST_PREFIX="valgrind"
+#    install: brew install valgrind
+#
+# It currently fails due to: https://github.com/jemalloc/jemalloc/issues/1274
+
+print travis_template % include_rows
diff --git a/jemalloc/src/arena.c b/jemalloc/src/arena.c
new file mode 100644
index 0000000..ba50e41
--- /dev/null
+++ b/jemalloc/src/arena.c
@@ -0,0 +1,2298 @@
+#define JEMALLOC_ARENA_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/div.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/util.h"
+
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+/******************************************************************************/
+/* Data. */
+
+/*
+ * Define names for both unininitialized and initialized phases, so that
+ * options and mallctl processing are straightforward.
+ */
+const char *percpu_arena_mode_names[] = {
+	"percpu",
+	"phycpu",
+	"disabled",
+	"percpu",
+	"phycpu"
+};
+percpu_arena_mode_t opt_percpu_arena = PERCPU_ARENA_DEFAULT;
+
+ssize_t opt_dirty_decay_ms = DIRTY_DECAY_MS_DEFAULT;
+ssize_t opt_muzzy_decay_ms = MUZZY_DECAY_MS_DEFAULT;
+
+static atomic_zd_t dirty_decay_ms_default;
+static atomic_zd_t muzzy_decay_ms_default;
+
+const uint64_t h_steps[SMOOTHSTEP_NSTEPS] = {
+#define STEP(step, h, x, y)			\
+		h,
+		SMOOTHSTEP
+#undef STEP
+};
+
+static div_info_t arena_binind_div_info[SC_NBINS];
+
+size_t opt_oversize_threshold = OVERSIZE_THRESHOLD_DEFAULT;
+size_t oversize_threshold = OVERSIZE_THRESHOLD_DEFAULT;
+static unsigned huge_arena_ind;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static void arena_decay_to_limit(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, extents_t *extents, bool all, size_t npages_limit,
+    size_t npages_decay_max, bool is_background_thread);
+static bool arena_decay_dirty(tsdn_t *tsdn, arena_t *arena,
+    bool is_background_thread, bool all);
+static void arena_dalloc_bin_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin);
+static void arena_bin_lower_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin);
+
+/******************************************************************************/
+
+void
+arena_basic_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
+    const char **dss, ssize_t *dirty_decay_ms, ssize_t *muzzy_decay_ms,
+    size_t *nactive, size_t *ndirty, size_t *nmuzzy) {
+	*nthreads += arena_nthreads_get(arena, false);
+	*dss = dss_prec_names[arena_dss_prec_get(arena)];
+	*dirty_decay_ms = arena_dirty_decay_ms_get(arena);
+	*muzzy_decay_ms = arena_muzzy_decay_ms_get(arena);
+	*nactive += atomic_load_zu(&arena->nactive, ATOMIC_RELAXED);
+	*ndirty += extents_npages_get(&arena->extents_dirty);
+	*nmuzzy += extents_npages_get(&arena->extents_muzzy);
+}
+
+void
+arena_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
+    const char **dss, ssize_t *dirty_decay_ms, ssize_t *muzzy_decay_ms,
+    size_t *nactive, size_t *ndirty, size_t *nmuzzy, arena_stats_t *astats,
+    bin_stats_t *bstats, arena_stats_large_t *lstats,
+    arena_stats_extents_t *estats) {
+	cassert(config_stats);
+
+	arena_basic_stats_merge(tsdn, arena, nthreads, dss, dirty_decay_ms,
+	    muzzy_decay_ms, nactive, ndirty, nmuzzy);
+
+	size_t base_allocated, base_resident, base_mapped, metadata_thp;
+	base_stats_get(tsdn, arena->base, &base_allocated, &base_resident,
+	    &base_mapped, &metadata_thp);
+
+	arena_stats_lock(tsdn, &arena->stats);
+
+	arena_stats_accum_zu(&astats->mapped, base_mapped
+	    + arena_stats_read_zu(tsdn, &arena->stats, &arena->stats.mapped));
+	arena_stats_accum_zu(&astats->retained,
+	    extents_npages_get(&arena->extents_retained) << LG_PAGE);
+
+	atomic_store_zu(&astats->extent_avail,
+	    atomic_load_zu(&arena->extent_avail_cnt, ATOMIC_RELAXED),
+	    ATOMIC_RELAXED);
+
+	arena_stats_accum_u64(&astats->decay_dirty.npurge,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_dirty.npurge));
+	arena_stats_accum_u64(&astats->decay_dirty.nmadvise,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_dirty.nmadvise));
+	arena_stats_accum_u64(&astats->decay_dirty.purged,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_dirty.purged));
+
+	arena_stats_accum_u64(&astats->decay_muzzy.npurge,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_muzzy.npurge));
+	arena_stats_accum_u64(&astats->decay_muzzy.nmadvise,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_muzzy.nmadvise));
+	arena_stats_accum_u64(&astats->decay_muzzy.purged,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_muzzy.purged));
+
+	arena_stats_accum_zu(&astats->base, base_allocated);
+	arena_stats_accum_zu(&astats->internal, arena_internal_get(arena));
+	arena_stats_accum_zu(&astats->metadata_thp, metadata_thp);
+	arena_stats_accum_zu(&astats->resident, base_resident +
+	    (((atomic_load_zu(&arena->nactive, ATOMIC_RELAXED) +
+	    extents_npages_get(&arena->extents_dirty) +
+	    extents_npages_get(&arena->extents_muzzy)) << LG_PAGE)));
+	arena_stats_accum_zu(&astats->abandoned_vm, atomic_load_zu(
+	    &arena->stats.abandoned_vm, ATOMIC_RELAXED));
+
+	for (szind_t i = 0; i < SC_NSIZES - SC_NBINS; i++) {
+		uint64_t nmalloc = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].nmalloc);
+		arena_stats_accum_u64(&lstats[i].nmalloc, nmalloc);
+		arena_stats_accum_u64(&astats->nmalloc_large, nmalloc);
+
+		uint64_t ndalloc = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].ndalloc);
+		arena_stats_accum_u64(&lstats[i].ndalloc, ndalloc);
+		arena_stats_accum_u64(&astats->ndalloc_large, ndalloc);
+
+		uint64_t nrequests = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].nrequests);
+		arena_stats_accum_u64(&lstats[i].nrequests,
+		    nmalloc + nrequests);
+		arena_stats_accum_u64(&astats->nrequests_large,
+		    nmalloc + nrequests);
+
+		/* nfill == nmalloc for large currently. */
+		arena_stats_accum_u64(&lstats[i].nfills, nmalloc);
+		arena_stats_accum_u64(&astats->nfills_large, nmalloc);
+
+		uint64_t nflush = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].nflushes);
+		arena_stats_accum_u64(&lstats[i].nflushes, nflush);
+		arena_stats_accum_u64(&astats->nflushes_large, nflush);
+
+		assert(nmalloc >= ndalloc);
+		assert(nmalloc - ndalloc <= SIZE_T_MAX);
+		size_t curlextents = (size_t)(nmalloc - ndalloc);
+		lstats[i].curlextents += curlextents;
+		arena_stats_accum_zu(&astats->allocated_large,
+		    curlextents * sz_index2size(SC_NBINS + i));
+	}
+
+	for (pszind_t i = 0; i < SC_NPSIZES; i++) {
+		size_t dirty, muzzy, retained, dirty_bytes, muzzy_bytes,
+		    retained_bytes;
+		dirty = extents_nextents_get(&arena->extents_dirty, i);
+		muzzy = extents_nextents_get(&arena->extents_muzzy, i);
+		retained = extents_nextents_get(&arena->extents_retained, i);
+		dirty_bytes = extents_nbytes_get(&arena->extents_dirty, i);
+		muzzy_bytes = extents_nbytes_get(&arena->extents_muzzy, i);
+		retained_bytes =
+		    extents_nbytes_get(&arena->extents_retained, i);
+
+		atomic_store_zu(&estats[i].ndirty, dirty, ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].nmuzzy, muzzy, ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].nretained, retained, ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].dirty_bytes, dirty_bytes,
+		    ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].muzzy_bytes, muzzy_bytes,
+		    ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].retained_bytes, retained_bytes,
+		    ATOMIC_RELAXED);
+	}
+
+	arena_stats_unlock(tsdn, &arena->stats);
+
+	/* tcache_bytes counts currently cached bytes. */
+	atomic_store_zu(&astats->tcache_bytes, 0, ATOMIC_RELAXED);
+	malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+	cache_bin_array_descriptor_t *descriptor;
+	ql_foreach(descriptor, &arena->cache_bin_array_descriptor_ql, link) {
+		szind_t i = 0;
+		for (; i < SC_NBINS; i++) {
+			cache_bin_t *tbin = &descriptor->bins_small[i];
+			arena_stats_accum_zu(&astats->tcache_bytes,
+			    tbin->ncached * sz_index2size(i));
+		}
+		for (; i < nhbins; i++) {
+			cache_bin_t *tbin = &descriptor->bins_large[i];
+			arena_stats_accum_zu(&astats->tcache_bytes,
+			    tbin->ncached * sz_index2size(i));
+		}
+	}
+	malloc_mutex_prof_read(tsdn,
+	    &astats->mutex_prof_data[arena_prof_mutex_tcache_list],
+	    &arena->tcache_ql_mtx);
+	malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx);
+
+#define READ_ARENA_MUTEX_PROF_DATA(mtx, ind)				\
+    malloc_mutex_lock(tsdn, &arena->mtx);				\
+    malloc_mutex_prof_read(tsdn, &astats->mutex_prof_data[ind],		\
+        &arena->mtx);							\
+    malloc_mutex_unlock(tsdn, &arena->mtx);
+
+	/* Gather per arena mutex profiling data. */
+	READ_ARENA_MUTEX_PROF_DATA(large_mtx, arena_prof_mutex_large);
+	READ_ARENA_MUTEX_PROF_DATA(extent_avail_mtx,
+	    arena_prof_mutex_extent_avail)
+	READ_ARENA_MUTEX_PROF_DATA(extents_dirty.mtx,
+	    arena_prof_mutex_extents_dirty)
+	READ_ARENA_MUTEX_PROF_DATA(extents_muzzy.mtx,
+	    arena_prof_mutex_extents_muzzy)
+	READ_ARENA_MUTEX_PROF_DATA(extents_retained.mtx,
+	    arena_prof_mutex_extents_retained)
+	READ_ARENA_MUTEX_PROF_DATA(decay_dirty.mtx,
+	    arena_prof_mutex_decay_dirty)
+	READ_ARENA_MUTEX_PROF_DATA(decay_muzzy.mtx,
+	    arena_prof_mutex_decay_muzzy)
+	READ_ARENA_MUTEX_PROF_DATA(base->mtx,
+	    arena_prof_mutex_base)
+#undef READ_ARENA_MUTEX_PROF_DATA
+
+	nstime_copy(&astats->uptime, &arena->create_time);
+	nstime_update(&astats->uptime);
+	nstime_subtract(&astats->uptime, &arena->create_time);
+
+	for (szind_t i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_stats_merge(tsdn, &bstats[i],
+			    &arena->bins[i].bin_shards[j]);
+		}
+	}
+}
+
+void
+arena_extents_dirty_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extents_dalloc(tsdn, arena, r_extent_hooks, &arena->extents_dirty,
+	    extent);
+	if (arena_dirty_decay_ms_get(arena) == 0) {
+		arena_decay_dirty(tsdn, arena, false, true);
+	} else {
+		arena_background_thread_inactivity_check(tsdn, arena, false);
+	}
+}
+
+static void *
+arena_slab_reg_alloc(extent_t *slab, const bin_info_t *bin_info) {
+	void *ret;
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+	size_t regind;
+
+	assert(extent_nfree_get(slab) > 0);
+	assert(!bitmap_full(slab_data->bitmap, &bin_info->bitmap_info));
+
+	regind = bitmap_sfu(slab_data->bitmap, &bin_info->bitmap_info);
+	ret = (void *)((uintptr_t)extent_addr_get(slab) +
+	    (uintptr_t)(bin_info->reg_size * regind));
+	extent_nfree_dec(slab);
+	return ret;
+}
+
+static void
+arena_slab_reg_alloc_batch(extent_t *slab, const bin_info_t *bin_info,
+			   unsigned cnt, void** ptrs) {
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+
+	assert(extent_nfree_get(slab) >= cnt);
+	assert(!bitmap_full(slab_data->bitmap, &bin_info->bitmap_info));
+
+#if (! defined JEMALLOC_INTERNAL_POPCOUNTL) || (defined BITMAP_USE_TREE)
+	for (unsigned i = 0; i < cnt; i++) {
+		size_t regind = bitmap_sfu(slab_data->bitmap,
+					   &bin_info->bitmap_info);
+		*(ptrs + i) = (void *)((uintptr_t)extent_addr_get(slab) +
+		    (uintptr_t)(bin_info->reg_size * regind));
+	}
+#else
+	unsigned group = 0;
+	bitmap_t g = slab_data->bitmap[group];
+	unsigned i = 0;
+	while (i < cnt) {
+		while (g == 0) {
+			g = slab_data->bitmap[++group];
+		}
+		size_t shift = group << LG_BITMAP_GROUP_NBITS;
+		size_t pop = popcount_lu(g);
+		if (pop > (cnt - i)) {
+			pop = cnt - i;
+		}
+
+		/*
+		 * Load from memory locations only once, outside the
+		 * hot loop below.
+		 */
+		uintptr_t base = (uintptr_t)extent_addr_get(slab);
+		uintptr_t regsize = (uintptr_t)bin_info->reg_size;
+		while (pop--) {
+			size_t bit = cfs_lu(&g);
+			size_t regind = shift + bit;
+			*(ptrs + i) = (void *)(base + regsize * regind);
+
+			i++;
+		}
+		slab_data->bitmap[group] = g;
+	}
+#endif
+	extent_nfree_sub(slab, cnt);
+}
+
+#ifndef JEMALLOC_JET
+static
+#endif
+size_t
+arena_slab_regind(extent_t *slab, szind_t binind, const void *ptr) {
+	size_t diff, regind;
+
+	/* Freeing a pointer outside the slab can cause assertion failure. */
+	assert((uintptr_t)ptr >= (uintptr_t)extent_addr_get(slab));
+	assert((uintptr_t)ptr < (uintptr_t)extent_past_get(slab));
+	/* Freeing an interior pointer can cause assertion failure. */
+	assert(((uintptr_t)ptr - (uintptr_t)extent_addr_get(slab)) %
+	    (uintptr_t)bin_infos[binind].reg_size == 0);
+
+	diff = (size_t)((uintptr_t)ptr - (uintptr_t)extent_addr_get(slab));
+
+	/* Avoid doing division with a variable divisor. */
+	regind = div_compute(&arena_binind_div_info[binind], diff);
+
+	assert(regind < bin_infos[binind].nregs);
+
+	return regind;
+}
+
+static void
+arena_slab_reg_dalloc(extent_t *slab, arena_slab_data_t *slab_data, void *ptr) {
+	szind_t binind = extent_szind_get(slab);
+	const bin_info_t *bin_info = &bin_infos[binind];
+	size_t regind = arena_slab_regind(slab, binind, ptr);
+
+	assert(extent_nfree_get(slab) < bin_info->nregs);
+	/* Freeing an unallocated pointer can cause assertion failure. */
+	assert(bitmap_get(slab_data->bitmap, &bin_info->bitmap_info, regind));
+
+	bitmap_unset(slab_data->bitmap, &bin_info->bitmap_info, regind);
+	extent_nfree_inc(slab);
+}
+
+static void
+arena_nactive_add(arena_t *arena, size_t add_pages) {
+	atomic_fetch_add_zu(&arena->nactive, add_pages, ATOMIC_RELAXED);
+}
+
+static void
+arena_nactive_sub(arena_t *arena, size_t sub_pages) {
+	assert(atomic_load_zu(&arena->nactive, ATOMIC_RELAXED) >= sub_pages);
+	atomic_fetch_sub_zu(&arena->nactive, sub_pages, ATOMIC_RELAXED);
+}
+
+static void
+arena_large_malloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t usize) {
+	szind_t index, hindex;
+
+	cassert(config_stats);
+
+	if (usize < SC_LARGE_MINCLASS) {
+		usize = SC_LARGE_MINCLASS;
+	}
+	index = sz_size2index(usize);
+	hindex = (index >= SC_NBINS) ? index - SC_NBINS : 0;
+
+	arena_stats_add_u64(tsdn, &arena->stats,
+	    &arena->stats.lstats[hindex].nmalloc, 1);
+}
+
+static void
+arena_large_dalloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t usize) {
+	szind_t index, hindex;
+
+	cassert(config_stats);
+
+	if (usize < SC_LARGE_MINCLASS) {
+		usize = SC_LARGE_MINCLASS;
+	}
+	index = sz_size2index(usize);
+	hindex = (index >= SC_NBINS) ? index - SC_NBINS : 0;
+
+	arena_stats_add_u64(tsdn, &arena->stats,
+	    &arena->stats.lstats[hindex].ndalloc, 1);
+}
+
+static void
+arena_large_ralloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t oldusize,
+    size_t usize) {
+	arena_large_dalloc_stats_update(tsdn, arena, oldusize);
+	arena_large_malloc_stats_update(tsdn, arena, usize);
+}
+
+static bool
+arena_may_have_muzzy(arena_t *arena) {
+	return (pages_can_purge_lazy && (arena_muzzy_decay_ms_get(arena) != 0));
+}
+
+extent_t *
+arena_extent_alloc_large(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool *zero) {
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	szind_t szind = sz_size2index(usize);
+	size_t mapped_add;
+	bool commit = true;
+	extent_t *extent = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_dirty, NULL, usize, sz_large_pad, alignment, false,
+	    szind, zero, &commit);
+	if (extent == NULL && arena_may_have_muzzy(arena)) {
+		extent = extents_alloc(tsdn, arena, &extent_hooks,
+		    &arena->extents_muzzy, NULL, usize, sz_large_pad, alignment,
+		    false, szind, zero, &commit);
+	}
+	size_t size = usize + sz_large_pad;
+	if (extent == NULL) {
+		extent = extent_alloc_wrapper(tsdn, arena, &extent_hooks, NULL,
+		    usize, sz_large_pad, alignment, false, szind, zero,
+		    &commit);
+		if (config_stats) {
+			/*
+			 * extent may be NULL on OOM, but in that case
+			 * mapped_add isn't used below, so there's no need to
+			 * conditionlly set it to 0 here.
+			 */
+			mapped_add = size;
+		}
+	} else if (config_stats) {
+		mapped_add = 0;
+	}
+
+	if (extent != NULL) {
+		if (config_stats) {
+			arena_stats_lock(tsdn, &arena->stats);
+			arena_large_malloc_stats_update(tsdn, arena, usize);
+			if (mapped_add != 0) {
+				arena_stats_add_zu(tsdn, &arena->stats,
+				    &arena->stats.mapped, mapped_add);
+			}
+			arena_stats_unlock(tsdn, &arena->stats);
+		}
+		arena_nactive_add(arena, size >> LG_PAGE);
+	}
+
+	return extent;
+}
+
+void
+arena_extent_dalloc_large_prep(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_large_dalloc_stats_update(tsdn, arena,
+		    extent_usize_get(extent));
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+	arena_nactive_sub(arena, extent_size_get(extent) >> LG_PAGE);
+}
+
+void
+arena_extent_ralloc_large_shrink(tsdn_t *tsdn, arena_t *arena, extent_t *extent,
+    size_t oldusize) {
+	size_t usize = extent_usize_get(extent);
+	size_t udiff = oldusize - usize;
+
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_large_ralloc_stats_update(tsdn, arena, oldusize, usize);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+	arena_nactive_sub(arena, udiff >> LG_PAGE);
+}
+
+void
+arena_extent_ralloc_large_expand(tsdn_t *tsdn, arena_t *arena, extent_t *extent,
+    size_t oldusize) {
+	size_t usize = extent_usize_get(extent);
+	size_t udiff = usize - oldusize;
+
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_large_ralloc_stats_update(tsdn, arena, oldusize, usize);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+	arena_nactive_add(arena, udiff >> LG_PAGE);
+}
+
+static ssize_t
+arena_decay_ms_read(arena_decay_t *decay) {
+	return atomic_load_zd(&decay->time_ms, ATOMIC_RELAXED);
+}
+
+static void
+arena_decay_ms_write(arena_decay_t *decay, ssize_t decay_ms) {
+	atomic_store_zd(&decay->time_ms, decay_ms, ATOMIC_RELAXED);
+}
+
+static void
+arena_decay_deadline_init(arena_decay_t *decay) {
+	/*
+	 * Generate a new deadline that is uniformly random within the next
+	 * epoch after the current one.
+	 */
+	nstime_copy(&decay->deadline, &decay->epoch);
+	nstime_add(&decay->deadline, &decay->interval);
+	if (arena_decay_ms_read(decay) > 0) {
+		nstime_t jitter;
+
+		nstime_init(&jitter, prng_range_u64(&decay->jitter_state,
+		    nstime_ns(&decay->interval)));
+		nstime_add(&decay->deadline, &jitter);
+	}
+}
+
+static bool
+arena_decay_deadline_reached(const arena_decay_t *decay, const nstime_t *time) {
+	return (nstime_compare(&decay->deadline, time) <= 0);
+}
+
+static size_t
+arena_decay_backlog_npages_limit(const arena_decay_t *decay) {
+	uint64_t sum;
+	size_t npages_limit_backlog;
+	unsigned i;
+
+	/*
+	 * For each element of decay_backlog, multiply by the corresponding
+	 * fixed-point smoothstep decay factor.  Sum the products, then divide
+	 * to round down to the nearest whole number of pages.
+	 */
+	sum = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+		sum += decay->backlog[i] * h_steps[i];
+	}
+	npages_limit_backlog = (size_t)(sum >> SMOOTHSTEP_BFP);
+
+	return npages_limit_backlog;
+}
+
+static void
+arena_decay_backlog_update_last(arena_decay_t *decay, size_t current_npages) {
+	size_t npages_delta = (current_npages > decay->nunpurged) ?
+	    current_npages - decay->nunpurged : 0;
+	decay->backlog[SMOOTHSTEP_NSTEPS-1] = npages_delta;
+
+	if (config_debug) {
+		if (current_npages > decay->ceil_npages) {
+			decay->ceil_npages = current_npages;
+		}
+		size_t npages_limit = arena_decay_backlog_npages_limit(decay);
+		assert(decay->ceil_npages >= npages_limit);
+		if (decay->ceil_npages > npages_limit) {
+			decay->ceil_npages = npages_limit;
+		}
+	}
+}
+
+static void
+arena_decay_backlog_update(arena_decay_t *decay, uint64_t nadvance_u64,
+    size_t current_npages) {
+	if (nadvance_u64 >= SMOOTHSTEP_NSTEPS) {
+		memset(decay->backlog, 0, (SMOOTHSTEP_NSTEPS-1) *
+		    sizeof(size_t));
+	} else {
+		size_t nadvance_z = (size_t)nadvance_u64;
+
+		assert((uint64_t)nadvance_z == nadvance_u64);
+
+		memmove(decay->backlog, &decay->backlog[nadvance_z],
+		    (SMOOTHSTEP_NSTEPS - nadvance_z) * sizeof(size_t));
+		if (nadvance_z > 1) {
+			memset(&decay->backlog[SMOOTHSTEP_NSTEPS -
+			    nadvance_z], 0, (nadvance_z-1) * sizeof(size_t));
+		}
+	}
+
+	arena_decay_backlog_update_last(decay, current_npages);
+}
+
+static void
+arena_decay_try_purge(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, size_t current_npages, size_t npages_limit,
+    bool is_background_thread) {
+	if (current_npages > npages_limit) {
+		arena_decay_to_limit(tsdn, arena, decay, extents, false,
+		    npages_limit, current_npages - npages_limit,
+		    is_background_thread);
+	}
+}
+
+static void
+arena_decay_epoch_advance_helper(arena_decay_t *decay, const nstime_t *time,
+    size_t current_npages) {
+	assert(arena_decay_deadline_reached(decay, time));
+
+	nstime_t delta;
+	nstime_copy(&delta, time);
+	nstime_subtract(&delta, &decay->epoch);
+
+	uint64_t nadvance_u64 = nstime_divide(&delta, &decay->interval);
+	assert(nadvance_u64 > 0);
+
+	/* Add nadvance_u64 decay intervals to epoch. */
+	nstime_copy(&delta, &decay->interval);
+	nstime_imultiply(&delta, nadvance_u64);
+	nstime_add(&decay->epoch, &delta);
+
+	/* Set a new deadline. */
+	arena_decay_deadline_init(decay);
+
+	/* Update the backlog. */
+	arena_decay_backlog_update(decay, nadvance_u64, current_npages);
+}
+
+static void
+arena_decay_epoch_advance(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, const nstime_t *time, bool is_background_thread) {
+	size_t current_npages = extents_npages_get(extents);
+	arena_decay_epoch_advance_helper(decay, time, current_npages);
+
+	size_t npages_limit = arena_decay_backlog_npages_limit(decay);
+	/* We may unlock decay->mtx when try_purge(). Finish logging first. */
+	decay->nunpurged = (npages_limit > current_npages) ? npages_limit :
+	    current_npages;
+
+	if (!background_thread_enabled() || is_background_thread) {
+		arena_decay_try_purge(tsdn, arena, decay, extents,
+		    current_npages, npages_limit, is_background_thread);
+	}
+}
+
+static void
+arena_decay_reinit(arena_decay_t *decay, ssize_t decay_ms) {
+	arena_decay_ms_write(decay, decay_ms);
+	if (decay_ms > 0) {
+		nstime_init(&decay->interval, (uint64_t)decay_ms *
+		    KQU(1000000));
+		nstime_idivide(&decay->interval, SMOOTHSTEP_NSTEPS);
+	}
+
+	nstime_init(&decay->epoch, 0);
+	nstime_update(&decay->epoch);
+	decay->jitter_state = (uint64_t)(uintptr_t)decay;
+	arena_decay_deadline_init(decay);
+	decay->nunpurged = 0;
+	memset(decay->backlog, 0, SMOOTHSTEP_NSTEPS * sizeof(size_t));
+}
+
+static bool
+arena_decay_init(arena_decay_t *decay, ssize_t decay_ms,
+    arena_stats_decay_t *stats) {
+	if (config_debug) {
+		for (size_t i = 0; i < sizeof(arena_decay_t); i++) {
+			assert(((char *)decay)[i] == 0);
+		}
+		decay->ceil_npages = 0;
+	}
+	if (malloc_mutex_init(&decay->mtx, "decay", WITNESS_RANK_DECAY,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	decay->purging = false;
+	arena_decay_reinit(decay, decay_ms);
+	/* Memory is zeroed, so there is no need to clear stats. */
+	if (config_stats) {
+		decay->stats = stats;
+	}
+	return false;
+}
+
+static bool
+arena_decay_ms_valid(ssize_t decay_ms) {
+	if (decay_ms < -1) {
+		return false;
+	}
+	if (decay_ms == -1 || (uint64_t)decay_ms <= NSTIME_SEC_MAX *
+	    KQU(1000)) {
+		return true;
+	}
+	return false;
+}
+
+static bool
+arena_maybe_decay(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, bool is_background_thread) {
+	malloc_mutex_assert_owner(tsdn, &decay->mtx);
+
+	/* Purge all or nothing if the option is disabled. */
+	ssize_t decay_ms = arena_decay_ms_read(decay);
+	if (decay_ms <= 0) {
+		if (decay_ms == 0) {
+			arena_decay_to_limit(tsdn, arena, decay, extents, false,
+			    0, extents_npages_get(extents),
+			    is_background_thread);
+		}
+		return false;
+	}
+
+	nstime_t time;
+	nstime_init(&time, 0);
+	nstime_update(&time);
+	if (unlikely(!nstime_monotonic() && nstime_compare(&decay->epoch, &time)
+	    > 0)) {
+		/*
+		 * Time went backwards.  Move the epoch back in time and
+		 * generate a new deadline, with the expectation that time
+		 * typically flows forward for long enough periods of time that
+		 * epochs complete.  Unfortunately, this strategy is susceptible
+		 * to clock jitter triggering premature epoch advances, but
+		 * clock jitter estimation and compensation isn't feasible here
+		 * because calls into this code are event-driven.
+		 */
+		nstime_copy(&decay->epoch, &time);
+		arena_decay_deadline_init(decay);
+	} else {
+		/* Verify that time does not go backwards. */
+		assert(nstime_compare(&decay->epoch, &time) <= 0);
+	}
+
+	/*
+	 * If the deadline has been reached, advance to the current epoch and
+	 * purge to the new limit if necessary.  Note that dirty pages created
+	 * during the current epoch are not subject to purge until a future
+	 * epoch, so as a result purging only happens during epoch advances, or
+	 * being triggered by background threads (scheduled event).
+	 */
+	bool advance_epoch = arena_decay_deadline_reached(decay, &time);
+	if (advance_epoch) {
+		arena_decay_epoch_advance(tsdn, arena, decay, extents, &time,
+		    is_background_thread);
+	} else if (is_background_thread) {
+		arena_decay_try_purge(tsdn, arena, decay, extents,
+		    extents_npages_get(extents),
+		    arena_decay_backlog_npages_limit(decay),
+		    is_background_thread);
+	}
+
+	return advance_epoch;
+}
+
+static ssize_t
+arena_decay_ms_get(arena_decay_t *decay) {
+	return arena_decay_ms_read(decay);
+}
+
+ssize_t
+arena_dirty_decay_ms_get(arena_t *arena) {
+	return arena_decay_ms_get(&arena->decay_dirty);
+}
+
+ssize_t
+arena_muzzy_decay_ms_get(arena_t *arena) {
+	return arena_decay_ms_get(&arena->decay_muzzy);
+}
+
+static bool
+arena_decay_ms_set(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, ssize_t decay_ms) {
+	if (!arena_decay_ms_valid(decay_ms)) {
+		return true;
+	}
+
+	malloc_mutex_lock(tsdn, &decay->mtx);
+	/*
+	 * Restart decay backlog from scratch, which may cause many dirty pages
+	 * to be immediately purged.  It would conceptually be possible to map
+	 * the old backlog onto the new backlog, but there is no justification
+	 * for such complexity since decay_ms changes are intended to be
+	 * infrequent, either between the {-1, 0, >0} states, or a one-time
+	 * arbitrary change during initial arena configuration.
+	 */
+	arena_decay_reinit(decay, decay_ms);
+	arena_maybe_decay(tsdn, arena, decay, extents, false);
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	return false;
+}
+
+bool
+arena_dirty_decay_ms_set(tsdn_t *tsdn, arena_t *arena,
+    ssize_t decay_ms) {
+	return arena_decay_ms_set(tsdn, arena, &arena->decay_dirty,
+	    &arena->extents_dirty, decay_ms);
+}
+
+bool
+arena_muzzy_decay_ms_set(tsdn_t *tsdn, arena_t *arena,
+    ssize_t decay_ms) {
+	return arena_decay_ms_set(tsdn, arena, &arena->decay_muzzy,
+	    &arena->extents_muzzy, decay_ms);
+}
+
+static size_t
+arena_stash_decayed(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, size_t npages_limit,
+	size_t npages_decay_max, extent_list_t *decay_extents) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	/* Stash extents according to npages_limit. */
+	size_t nstashed = 0;
+	extent_t *extent;
+	while (nstashed < npages_decay_max &&
+	    (extent = extents_evict(tsdn, arena, r_extent_hooks, extents,
+	    npages_limit)) != NULL) {
+		extent_list_append(decay_extents, extent);
+		nstashed += extent_size_get(extent) >> LG_PAGE;
+	}
+	return nstashed;
+}
+
+static size_t
+arena_decay_stashed(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, arena_decay_t *decay, extents_t *extents,
+    bool all, extent_list_t *decay_extents, bool is_background_thread) {
+	size_t nmadvise, nunmapped;
+	size_t npurged;
+
+	if (config_stats) {
+		nmadvise = 0;
+		nunmapped = 0;
+	}
+	npurged = 0;
+
+	ssize_t muzzy_decay_ms = arena_muzzy_decay_ms_get(arena);
+	for (extent_t *extent = extent_list_first(decay_extents); extent !=
+	    NULL; extent = extent_list_first(decay_extents)) {
+		if (config_stats) {
+			nmadvise++;
+		}
+		size_t npages = extent_size_get(extent) >> LG_PAGE;
+		npurged += npages;
+		extent_list_remove(decay_extents, extent);
+		switch (extents_state_get(extents)) {
+		case extent_state_active:
+			not_reached();
+		case extent_state_dirty:
+			if (!all && muzzy_decay_ms != 0 &&
+			    !extent_purge_lazy_wrapper(tsdn, arena,
+			    r_extent_hooks, extent, 0,
+			    extent_size_get(extent))) {
+				extents_dalloc(tsdn, arena, r_extent_hooks,
+				    &arena->extents_muzzy, extent);
+				arena_background_thread_inactivity_check(tsdn,
+				    arena, is_background_thread);
+				break;
+			}
+			/* Fall through. */
+		case extent_state_muzzy:
+			extent_dalloc_wrapper(tsdn, arena, r_extent_hooks,
+			    extent);
+			if (config_stats) {
+				nunmapped += npages;
+			}
+			break;
+		case extent_state_retained:
+		default:
+			not_reached();
+		}
+	}
+
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_stats_add_u64(tsdn, &arena->stats, &decay->stats->npurge,
+		    1);
+		arena_stats_add_u64(tsdn, &arena->stats,
+		    &decay->stats->nmadvise, nmadvise);
+		arena_stats_add_u64(tsdn, &arena->stats, &decay->stats->purged,
+		    npurged);
+		arena_stats_sub_zu(tsdn, &arena->stats, &arena->stats.mapped,
+		    nunmapped << LG_PAGE);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+
+	return npurged;
+}
+
+/*
+ * npages_limit: Decay at most npages_decay_max pages without violating the
+ * invariant: (extents_npages_get(extents) >= npages_limit).  We need an upper
+ * bound on number of pages in order to prevent unbounded growth (namely in
+ * stashed), otherwise unbounded new pages could be added to extents during the
+ * current decay run, so that the purging thread never finishes.
+ */
+static void
+arena_decay_to_limit(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, bool all, size_t npages_limit, size_t npages_decay_max,
+    bool is_background_thread) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 1);
+	malloc_mutex_assert_owner(tsdn, &decay->mtx);
+
+	if (decay->purging) {
+		return;
+	}
+	decay->purging = true;
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+
+	extent_list_t decay_extents;
+	extent_list_init(&decay_extents);
+
+	size_t npurge = arena_stash_decayed(tsdn, arena, &extent_hooks, extents,
+	    npages_limit, npages_decay_max, &decay_extents);
+	if (npurge != 0) {
+		size_t npurged = arena_decay_stashed(tsdn, arena,
+		    &extent_hooks, decay, extents, all, &decay_extents,
+		    is_background_thread);
+		assert(npurged == npurge);
+	}
+
+	malloc_mutex_lock(tsdn, &decay->mtx);
+	decay->purging = false;
+}
+
+static bool
+arena_decay_impl(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, bool is_background_thread, bool all) {
+	if (all) {
+		malloc_mutex_lock(tsdn, &decay->mtx);
+		arena_decay_to_limit(tsdn, arena, decay, extents, all, 0,
+		    extents_npages_get(extents), is_background_thread);
+		malloc_mutex_unlock(tsdn, &decay->mtx);
+
+		return false;
+	}
+
+	if (malloc_mutex_trylock(tsdn, &decay->mtx)) {
+		/* No need to wait if another thread is in progress. */
+		return true;
+	}
+
+	bool epoch_advanced = arena_maybe_decay(tsdn, arena, decay, extents,
+	    is_background_thread);
+	size_t npages_new;
+	if (epoch_advanced) {
+		/* Backlog is updated on epoch advance. */
+		npages_new = decay->backlog[SMOOTHSTEP_NSTEPS-1];
+	}
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	if (have_background_thread && background_thread_enabled() &&
+	    epoch_advanced && !is_background_thread) {
+		background_thread_interval_check(tsdn, arena, decay,
+		    npages_new);
+	}
+
+	return false;
+}
+
+static bool
+arena_decay_dirty(tsdn_t *tsdn, arena_t *arena, bool is_background_thread,
+    bool all) {
+	return arena_decay_impl(tsdn, arena, &arena->decay_dirty,
+	    &arena->extents_dirty, is_background_thread, all);
+}
+
+static bool
+arena_decay_muzzy(tsdn_t *tsdn, arena_t *arena, bool is_background_thread,
+    bool all) {
+	return arena_decay_impl(tsdn, arena, &arena->decay_muzzy,
+	    &arena->extents_muzzy, is_background_thread, all);
+}
+
+void
+arena_decay(tsdn_t *tsdn, arena_t *arena, bool is_background_thread, bool all) {
+	if (arena_decay_dirty(tsdn, arena, is_background_thread, all)) {
+		return;
+	}
+	arena_decay_muzzy(tsdn, arena, is_background_thread, all);
+}
+
+static void
+arena_slab_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *slab) {
+	arena_nactive_sub(arena, extent_size_get(slab) >> LG_PAGE);
+
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+	arena_extents_dirty_dalloc(tsdn, arena, &extent_hooks, slab);
+}
+
+static void
+arena_bin_slabs_nonfull_insert(bin_t *bin, extent_t *slab) {
+	assert(extent_nfree_get(slab) > 0);
+	extent_heap_insert(&bin->slabs_nonfull, slab);
+	if (config_stats) {
+		bin->stats.nonfull_slabs++;
+	}
+}
+
+static void
+arena_bin_slabs_nonfull_remove(bin_t *bin, extent_t *slab) {
+	extent_heap_remove(&bin->slabs_nonfull, slab);
+	if (config_stats) {
+		bin->stats.nonfull_slabs--;
+	}
+}
+
+static extent_t *
+arena_bin_slabs_nonfull_tryget(bin_t *bin) {
+	extent_t *slab = extent_heap_remove_first(&bin->slabs_nonfull);
+	if (slab == NULL) {
+		return NULL;
+	}
+	if (config_stats) {
+		bin->stats.reslabs++;
+		bin->stats.nonfull_slabs--;
+	}
+	return slab;
+}
+
+static void
+arena_bin_slabs_full_insert(arena_t *arena, bin_t *bin, extent_t *slab) {
+	assert(extent_nfree_get(slab) == 0);
+	/*
+	 *  Tracking extents is required by arena_reset, which is not allowed
+	 *  for auto arenas.  Bypass this step to avoid touching the extent
+	 *  linkage (often results in cache misses) for auto arenas.
+	 */
+	if (arena_is_auto(arena)) {
+		return;
+	}
+	extent_list_append(&bin->slabs_full, slab);
+}
+
+static void
+arena_bin_slabs_full_remove(arena_t *arena, bin_t *bin, extent_t *slab) {
+	if (arena_is_auto(arena)) {
+		return;
+	}
+	extent_list_remove(&bin->slabs_full, slab);
+}
+
+static void
+arena_bin_reset(tsd_t *tsd, arena_t *arena, bin_t *bin) {
+	extent_t *slab;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	if (bin->slabcur != NULL) {
+		slab = bin->slabcur;
+		bin->slabcur = NULL;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_slab_dalloc(tsd_tsdn(tsd), arena, slab);
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	}
+	while ((slab = extent_heap_remove_first(&bin->slabs_nonfull)) != NULL) {
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_slab_dalloc(tsd_tsdn(tsd), arena, slab);
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	}
+	for (slab = extent_list_first(&bin->slabs_full); slab != NULL;
+	     slab = extent_list_first(&bin->slabs_full)) {
+		arena_bin_slabs_full_remove(arena, bin, slab);
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_slab_dalloc(tsd_tsdn(tsd), arena, slab);
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	}
+	if (config_stats) {
+		bin->stats.curregs = 0;
+		bin->stats.curslabs = 0;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+}
+
+void
+arena_reset(tsd_t *tsd, arena_t *arena) {
+	/*
+	 * Locking in this function is unintuitive.  The caller guarantees that
+	 * no concurrent operations are happening in this arena, but there are
+	 * still reasons that some locking is necessary:
+	 *
+	 * - Some of the functions in the transitive closure of calls assume
+	 *   appropriate locks are held, and in some cases these locks are
+	 *   temporarily dropped to avoid lock order reversal or deadlock due to
+	 *   reentry.
+	 * - mallctl("epoch", ...) may concurrently refresh stats.  While
+	 *   strictly speaking this is a "concurrent operation", disallowing
+	 *   stats refreshes would impose an inconvenient burden.
+	 */
+
+	/* Large allocations. */
+	malloc_mutex_lock(tsd_tsdn(tsd), &arena->large_mtx);
+
+	for (extent_t *extent = extent_list_first(&arena->large); extent !=
+	    NULL; extent = extent_list_first(&arena->large)) {
+		void *ptr = extent_base_get(extent);
+		size_t usize;
+
+		malloc_mutex_unlock(tsd_tsdn(tsd), &arena->large_mtx);
+		alloc_ctx_t alloc_ctx;
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+		assert(alloc_ctx.szind != SC_NSIZES);
+
+		if (config_stats || (config_prof && opt_prof)) {
+			usize = sz_index2size(alloc_ctx.szind);
+			assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+		}
+		/* Remove large allocation from prof sample set. */
+		if (config_prof && opt_prof) {
+			prof_free(tsd, ptr, usize, &alloc_ctx);
+		}
+		large_dalloc(tsd_tsdn(tsd), extent);
+		malloc_mutex_lock(tsd_tsdn(tsd), &arena->large_mtx);
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &arena->large_mtx);
+
+	/* Bins. */
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			arena_bin_reset(tsd, arena,
+			    &arena->bins[i].bin_shards[j]);
+		}
+	}
+
+	atomic_store_zu(&arena->nactive, 0, ATOMIC_RELAXED);
+}
+
+static void
+arena_destroy_retained(tsdn_t *tsdn, arena_t *arena) {
+	/*
+	 * Iterate over the retained extents and destroy them.  This gives the
+	 * extent allocator underlying the extent hooks an opportunity to unmap
+	 * all retained memory without having to keep its own metadata
+	 * structures.  In practice, virtual memory for dss-allocated extents is
+	 * leaked here, so best practice is to avoid dss for arenas to be
+	 * destroyed, or provide custom extent hooks that track retained
+	 * dss-based extents for later reuse.
+	 */
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+	extent_t *extent;
+	while ((extent = extents_evict(tsdn, arena, &extent_hooks,
+	    &arena->extents_retained, 0)) != NULL) {
+		extent_destroy_wrapper(tsdn, arena, &extent_hooks, extent);
+	}
+}
+
+void
+arena_destroy(tsd_t *tsd, arena_t *arena) {
+	assert(base_ind_get(arena->base) >= narenas_auto);
+	assert(arena_nthreads_get(arena, false) == 0);
+	assert(arena_nthreads_get(arena, true) == 0);
+
+	/*
+	 * No allocations have occurred since arena_reset() was called.
+	 * Furthermore, the caller (arena_i_destroy_ctl()) purged all cached
+	 * extents, so only retained extents may remain.
+	 */
+	assert(extents_npages_get(&arena->extents_dirty) == 0);
+	assert(extents_npages_get(&arena->extents_muzzy) == 0);
+
+	/* Deallocate retained memory. */
+	arena_destroy_retained(tsd_tsdn(tsd), arena);
+
+	/*
+	 * Remove the arena pointer from the arenas array.  We rely on the fact
+	 * that there is no way for the application to get a dirty read from the
+	 * arenas array unless there is an inherent race in the application
+	 * involving access of an arena being concurrently destroyed.  The
+	 * application must synchronize knowledge of the arena's validity, so as
+	 * long as we use an atomic write to update the arenas array, the
+	 * application will get a clean read any time after it synchronizes
+	 * knowledge that the arena is no longer valid.
+	 */
+	arena_set(base_ind_get(arena->base), NULL);
+
+	/*
+	 * Destroy the base allocator, which manages all metadata ever mapped by
+	 * this arena.
+	 */
+	base_delete(tsd_tsdn(tsd), arena->base);
+}
+
+static extent_t *
+arena_slab_alloc_hard(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, const bin_info_t *bin_info,
+    szind_t szind) {
+	extent_t *slab;
+	bool zero, commit;
+
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	zero = false;
+	commit = true;
+	slab = extent_alloc_wrapper(tsdn, arena, r_extent_hooks, NULL,
+	    bin_info->slab_size, 0, PAGE, true, szind, &zero, &commit);
+
+	if (config_stats && slab != NULL) {
+		arena_stats_mapped_add(tsdn, &arena->stats,
+		    bin_info->slab_size);
+	}
+
+	return slab;
+}
+
+static extent_t *
+arena_slab_alloc(tsdn_t *tsdn, arena_t *arena, szind_t binind, unsigned binshard,
+    const bin_info_t *bin_info) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+	szind_t szind = sz_size2index(bin_info->reg_size);
+	bool zero = false;
+	bool commit = true;
+	extent_t *slab = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_dirty, NULL, bin_info->slab_size, 0, PAGE, true,
+	    binind, &zero, &commit);
+	if (slab == NULL && arena_may_have_muzzy(arena)) {
+		slab = extents_alloc(tsdn, arena, &extent_hooks,
+		    &arena->extents_muzzy, NULL, bin_info->slab_size, 0, PAGE,
+		    true, binind, &zero, &commit);
+	}
+	if (slab == NULL) {
+		slab = arena_slab_alloc_hard(tsdn, arena, &extent_hooks,
+		    bin_info, szind);
+		if (slab == NULL) {
+			return NULL;
+		}
+	}
+	assert(extent_slab_get(slab));
+
+	/* Initialize slab internals. */
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+	extent_nfree_binshard_set(slab, bin_info->nregs, binshard);
+	bitmap_init(slab_data->bitmap, &bin_info->bitmap_info, false);
+
+	arena_nactive_add(arena, extent_size_get(slab) >> LG_PAGE);
+
+	return slab;
+}
+
+static extent_t *
+arena_bin_nonfull_slab_get(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, unsigned binshard) {
+	extent_t *slab;
+	const bin_info_t *bin_info;
+
+	/* Look for a usable slab. */
+	slab = arena_bin_slabs_nonfull_tryget(bin);
+	if (slab != NULL) {
+		return slab;
+	}
+	/* No existing slabs have any space available. */
+
+	bin_info = &bin_infos[binind];
+
+	/* Allocate a new slab. */
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	/******************************/
+	slab = arena_slab_alloc(tsdn, arena, binind, binshard, bin_info);
+	/********************************/
+	malloc_mutex_lock(tsdn, &bin->lock);
+	if (slab != NULL) {
+		if (config_stats) {
+			bin->stats.nslabs++;
+			bin->stats.curslabs++;
+		}
+		return slab;
+	}
+
+	/*
+	 * arena_slab_alloc() failed, but another thread may have made
+	 * sufficient memory available while this one dropped bin->lock above,
+	 * so search one more time.
+	 */
+	slab = arena_bin_slabs_nonfull_tryget(bin);
+	if (slab != NULL) {
+		return slab;
+	}
+
+	return NULL;
+}
+
+/* Re-fill bin->slabcur, then call arena_slab_reg_alloc(). */
+static void *
+arena_bin_malloc_hard(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, unsigned binshard) {
+	const bin_info_t *bin_info;
+	extent_t *slab;
+
+	bin_info = &bin_infos[binind];
+	if (!arena_is_auto(arena) && bin->slabcur != NULL) {
+		arena_bin_slabs_full_insert(arena, bin, bin->slabcur);
+		bin->slabcur = NULL;
+	}
+	slab = arena_bin_nonfull_slab_get(tsdn, arena, bin, binind, binshard);
+	if (bin->slabcur != NULL) {
+		/*
+		 * Another thread updated slabcur while this one ran without the
+		 * bin lock in arena_bin_nonfull_slab_get().
+		 */
+		if (extent_nfree_get(bin->slabcur) > 0) {
+			void *ret = arena_slab_reg_alloc(bin->slabcur,
+			    bin_info);
+			if (slab != NULL) {
+				/*
+				 * arena_slab_alloc() may have allocated slab,
+				 * or it may have been pulled from
+				 * slabs_nonfull.  Therefore it is unsafe to
+				 * make any assumptions about how slab has
+				 * previously been used, and
+				 * arena_bin_lower_slab() must be called, as if
+				 * a region were just deallocated from the slab.
+				 */
+				if (extent_nfree_get(slab) == bin_info->nregs) {
+					arena_dalloc_bin_slab(tsdn, arena, slab,
+					    bin);
+				} else {
+					arena_bin_lower_slab(tsdn, arena, slab,
+					    bin);
+				}
+			}
+			return ret;
+		}
+
+		arena_bin_slabs_full_insert(arena, bin, bin->slabcur);
+		bin->slabcur = NULL;
+	}
+
+	if (slab == NULL) {
+		return NULL;
+	}
+	bin->slabcur = slab;
+
+	assert(extent_nfree_get(bin->slabcur) > 0);
+
+	return arena_slab_reg_alloc(slab, bin_info);
+}
+
+/* Choose a bin shard and return the locked bin. */
+bin_t *
+arena_bin_choose_lock(tsdn_t *tsdn, arena_t *arena, szind_t binind,
+    unsigned *binshard) {
+	bin_t *bin;
+	if (tsdn_null(tsdn) || tsd_arena_get(tsdn_tsd(tsdn)) == NULL) {
+		*binshard = 0;
+	} else {
+		*binshard = tsd_binshardsp_get(tsdn_tsd(tsdn))->binshard[binind];
+	}
+	assert(*binshard < bin_infos[binind].n_shards);
+	bin = &arena->bins[binind].bin_shards[*binshard];
+	malloc_mutex_lock(tsdn, &bin->lock);
+
+	return bin;
+}
+
+void
+arena_tcache_fill_small(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, uint64_t prof_accumbytes) {
+	unsigned i, nfill, cnt;
+
+	assert(tbin->ncached == 0);
+
+	if (config_prof && arena_prof_accum(tsdn, arena, prof_accumbytes)) {
+		prof_idump(tsdn);
+	}
+
+	unsigned binshard;
+	bin_t *bin = arena_bin_choose_lock(tsdn, arena, binind, &binshard);
+
+	for (i = 0, nfill = (tcache_bin_info[binind].ncached_max >>
+	    tcache->lg_fill_div[binind]); i < nfill; i += cnt) {
+		extent_t *slab;
+		if ((slab = bin->slabcur) != NULL && extent_nfree_get(slab) >
+		    0) {
+			unsigned tofill = nfill - i;
+			cnt = tofill < extent_nfree_get(slab) ?
+				tofill : extent_nfree_get(slab);
+			arena_slab_reg_alloc_batch(
+			   slab, &bin_infos[binind], cnt,
+			   tbin->avail - nfill + i);
+		} else {
+			cnt = 1;
+			void *ptr = arena_bin_malloc_hard(tsdn, arena, bin,
+			    binind, binshard);
+			/*
+			 * OOM.  tbin->avail isn't yet filled down to its first
+			 * element, so the successful allocations (if any) must
+			 * be moved just before tbin->avail before bailing out.
+			 */
+			if (ptr == NULL) {
+				if (i > 0) {
+					memmove(tbin->avail - i,
+						tbin->avail - nfill,
+						i * sizeof(void *));
+				}
+				break;
+			}
+			/* Insert such that low regions get used first. */
+			*(tbin->avail - nfill + i) = ptr;
+		}
+		if (config_fill && unlikely(opt_junk_alloc)) {
+			for (unsigned j = 0; j < cnt; j++) {
+				void* ptr = *(tbin->avail - nfill + i + j);
+				arena_alloc_junk_small(ptr, &bin_infos[binind],
+							true);
+			}
+		}
+	}
+	if (config_stats) {
+		bin->stats.nmalloc += i;
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		bin->stats.curregs += i;
+		bin->stats.nfills++;
+		tbin->tstats.nrequests = 0;
+	}
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	tbin->ncached = i;
+	arena_decay_tick(tsdn, arena);
+}
+
+void
+arena_alloc_junk_small(void *ptr, const bin_info_t *bin_info, bool zero) {
+	if (!zero) {
+		memset(ptr, JEMALLOC_ALLOC_JUNK, bin_info->reg_size);
+	}
+}
+
+static void
+arena_dalloc_junk_small_impl(void *ptr, const bin_info_t *bin_info) {
+	memset(ptr, JEMALLOC_FREE_JUNK, bin_info->reg_size);
+}
+arena_dalloc_junk_small_t *JET_MUTABLE arena_dalloc_junk_small =
+    arena_dalloc_junk_small_impl;
+
+static void *
+arena_malloc_small(tsdn_t *tsdn, arena_t *arena, szind_t binind, bool zero) {
+	void *ret;
+	bin_t *bin;
+	size_t usize;
+	extent_t *slab;
+
+	assert(binind < SC_NBINS);
+	usize = sz_index2size(binind);
+	unsigned binshard;
+	bin = arena_bin_choose_lock(tsdn, arena, binind, &binshard);
+
+	if ((slab = bin->slabcur) != NULL && extent_nfree_get(slab) > 0) {
+		ret = arena_slab_reg_alloc(slab, &bin_infos[binind]);
+	} else {
+		ret = arena_bin_malloc_hard(tsdn, arena, bin, binind, binshard);
+	}
+
+	if (ret == NULL) {
+		malloc_mutex_unlock(tsdn, &bin->lock);
+		return NULL;
+	}
+
+	if (config_stats) {
+		bin->stats.nmalloc++;
+		bin->stats.nrequests++;
+		bin->stats.curregs++;
+	}
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	if (config_prof && arena_prof_accum(tsdn, arena, usize)) {
+		prof_idump(tsdn);
+	}
+
+	if (!zero) {
+		if (config_fill) {
+			if (unlikely(opt_junk_alloc)) {
+				arena_alloc_junk_small(ret,
+				    &bin_infos[binind], false);
+			} else if (unlikely(opt_zero)) {
+				memset(ret, 0, usize);
+			}
+		}
+	} else {
+		if (config_fill && unlikely(opt_junk_alloc)) {
+			arena_alloc_junk_small(ret, &bin_infos[binind],
+			    true);
+		}
+		memset(ret, 0, usize);
+	}
+
+	arena_decay_tick(tsdn, arena);
+	return ret;
+}
+
+void *
+arena_malloc_hard(tsdn_t *tsdn, arena_t *arena, size_t size, szind_t ind,
+    bool zero) {
+	assert(!tsdn_null(tsdn) || arena != NULL);
+
+	if (likely(!tsdn_null(tsdn))) {
+		arena = arena_choose_maybe_huge(tsdn_tsd(tsdn), arena, size);
+	}
+	if (unlikely(arena == NULL)) {
+		return NULL;
+	}
+
+	if (likely(size <= SC_SMALL_MAXCLASS)) {
+		return arena_malloc_small(tsdn, arena, ind, zero);
+	}
+	return large_malloc(tsdn, arena, sz_index2size(ind), zero);
+}
+
+void *
+arena_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
+    bool zero, tcache_t *tcache) {
+	void *ret;
+
+	if (usize <= SC_SMALL_MAXCLASS
+	    && (alignment < PAGE
+	    || (alignment == PAGE && (usize & PAGE_MASK) == 0))) {
+		/* Small; alignment doesn't require special slab placement. */
+		ret = arena_malloc(tsdn, arena, usize, sz_size2index(usize),
+		    zero, tcache, true);
+	} else {
+		if (likely(alignment <= CACHELINE)) {
+			ret = large_malloc(tsdn, arena, usize, zero);
+		} else {
+			ret = large_palloc(tsdn, arena, usize, alignment, zero);
+		}
+	}
+	return ret;
+}
+
+void
+arena_prof_promote(tsdn_t *tsdn, void *ptr, size_t usize) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(isalloc(tsdn, ptr) == SC_LARGE_MINCLASS);
+	assert(usize <= SC_SMALL_MAXCLASS);
+
+	if (config_opt_safety_checks) {
+		safety_check_set_redzone(ptr, usize, SC_LARGE_MINCLASS);
+	}
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent = rtree_extent_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+	arena_t *arena = extent_arena_get(extent);
+
+	szind_t szind = sz_size2index(usize);
+	extent_szind_set(extent, szind);
+	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
+	    szind, false);
+
+	prof_accum_cancel(tsdn, &arena->prof_accum, usize);
+
+	assert(isalloc(tsdn, ptr) == usize);
+}
+
+static size_t
+arena_prof_demote(tsdn_t *tsdn, extent_t *extent, const void *ptr) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_szind_set(extent, SC_NBINS);
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
+	    SC_NBINS, false);
+
+	assert(isalloc(tsdn, ptr) == SC_LARGE_MINCLASS);
+
+	return SC_LARGE_MINCLASS;
+}
+
+void
+arena_dalloc_promoted(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    bool slow_path) {
+	cassert(config_prof);
+	assert(opt_prof);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	size_t usize = extent_usize_get(extent);
+	size_t bumped_usize = arena_prof_demote(tsdn, extent, ptr);
+	if (config_opt_safety_checks && usize < SC_LARGE_MINCLASS) {
+		/*
+		 * Currently, we only do redzoning for small sampled
+		 * allocations.
+		 */
+		assert(bumped_usize == SC_LARGE_MINCLASS);
+		safety_check_verify_redzone(ptr, usize, bumped_usize);
+	}
+	if (bumped_usize <= tcache_maxclass && tcache != NULL) {
+		tcache_dalloc_large(tsdn_tsd(tsdn), tcache, ptr,
+		    sz_size2index(bumped_usize), slow_path);
+	} else {
+		large_dalloc(tsdn, extent);
+	}
+}
+
+static void
+arena_dissociate_bin_slab(arena_t *arena, extent_t *slab, bin_t *bin) {
+	/* Dissociate slab from bin. */
+	if (slab == bin->slabcur) {
+		bin->slabcur = NULL;
+	} else {
+		szind_t binind = extent_szind_get(slab);
+		const bin_info_t *bin_info = &bin_infos[binind];
+
+		/*
+		 * The following block's conditional is necessary because if the
+		 * slab only contains one region, then it never gets inserted
+		 * into the non-full slabs heap.
+		 */
+		if (bin_info->nregs == 1) {
+			arena_bin_slabs_full_remove(arena, bin, slab);
+		} else {
+			arena_bin_slabs_nonfull_remove(bin, slab);
+		}
+	}
+}
+
+static void
+arena_dalloc_bin_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin) {
+	assert(slab != bin->slabcur);
+
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	/******************************/
+	arena_slab_dalloc(tsdn, arena, slab);
+	/****************************/
+	malloc_mutex_lock(tsdn, &bin->lock);
+	if (config_stats) {
+		bin->stats.curslabs--;
+	}
+}
+
+static void
+arena_bin_lower_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin) {
+	assert(extent_nfree_get(slab) > 0);
+
+	/*
+	 * Make sure that if bin->slabcur is non-NULL, it refers to the
+	 * oldest/lowest non-full slab.  It is okay to NULL slabcur out rather
+	 * than proactively keeping it pointing at the oldest/lowest non-full
+	 * slab.
+	 */
+	if (bin->slabcur != NULL && extent_snad_comp(bin->slabcur, slab) > 0) {
+		/* Switch slabcur. */
+		if (extent_nfree_get(bin->slabcur) > 0) {
+			arena_bin_slabs_nonfull_insert(bin, bin->slabcur);
+		} else {
+			arena_bin_slabs_full_insert(arena, bin, bin->slabcur);
+		}
+		bin->slabcur = slab;
+		if (config_stats) {
+			bin->stats.reslabs++;
+		}
+	} else {
+		arena_bin_slabs_nonfull_insert(bin, slab);
+	}
+}
+
+static void
+arena_dalloc_bin_locked_impl(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, extent_t *slab, void *ptr, bool junked) {
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+	const bin_info_t *bin_info = &bin_infos[binind];
+
+	if (!junked && config_fill && unlikely(opt_junk_free)) {
+		arena_dalloc_junk_small(ptr, bin_info);
+	}
+
+	arena_slab_reg_dalloc(slab, slab_data, ptr);
+	unsigned nfree = extent_nfree_get(slab);
+	if (nfree == bin_info->nregs) {
+		arena_dissociate_bin_slab(arena, slab, bin);
+		arena_dalloc_bin_slab(tsdn, arena, slab, bin);
+	} else if (nfree == 1 && slab != bin->slabcur) {
+		arena_bin_slabs_full_remove(arena, bin, slab);
+		arena_bin_lower_slab(tsdn, arena, slab, bin);
+	}
+
+	if (config_stats) {
+		bin->stats.ndalloc++;
+		bin->stats.curregs--;
+	}
+}
+
+void
+arena_dalloc_bin_junked_locked(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, extent_t *extent, void *ptr) {
+	arena_dalloc_bin_locked_impl(tsdn, arena, bin, binind, extent, ptr,
+	    true);
+}
+
+static void
+arena_dalloc_bin(tsdn_t *tsdn, arena_t *arena, extent_t *extent, void *ptr) {
+	szind_t binind = extent_szind_get(extent);
+	unsigned binshard = extent_binshard_get(extent);
+	bin_t *bin = &arena->bins[binind].bin_shards[binshard];
+
+	malloc_mutex_lock(tsdn, &bin->lock);
+	arena_dalloc_bin_locked_impl(tsdn, arena, bin, binind, extent, ptr,
+	    false);
+	malloc_mutex_unlock(tsdn, &bin->lock);
+}
+
+void
+arena_dalloc_small(tsdn_t *tsdn, void *ptr) {
+	extent_t *extent = iealloc(tsdn, ptr);
+	arena_t *arena = extent_arena_get(extent);
+
+	arena_dalloc_bin(tsdn, arena, extent, ptr);
+	arena_decay_tick(tsdn, arena);
+}
+
+bool
+arena_ralloc_no_move(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero, size_t *newsize) {
+	bool ret;
+	/* Calls with non-zero extra had to clamp extra. */
+	assert(extra == 0 || size + extra <= SC_LARGE_MAXCLASS);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		ret = true;
+		goto done;
+	}
+
+	size_t usize_min = sz_s2u(size);
+	size_t usize_max = sz_s2u(size + extra);
+	if (likely(oldsize <= SC_SMALL_MAXCLASS && usize_min
+	    <= SC_SMALL_MAXCLASS)) {
+		/*
+		 * Avoid moving the allocation if the size class can be left the
+		 * same.
+		 */
+		assert(bin_infos[sz_size2index(oldsize)].reg_size ==
+		    oldsize);
+		if ((usize_max > SC_SMALL_MAXCLASS
+		    || sz_size2index(usize_max) != sz_size2index(oldsize))
+		    && (size > oldsize || usize_max < oldsize)) {
+			ret = true;
+			goto done;
+		}
+
+		arena_decay_tick(tsdn, extent_arena_get(extent));
+		ret = false;
+	} else if (oldsize >= SC_LARGE_MINCLASS
+	    && usize_max >= SC_LARGE_MINCLASS) {
+		ret = large_ralloc_no_move(tsdn, extent, usize_min, usize_max,
+		    zero);
+	} else {
+		ret = true;
+	}
+done:
+	assert(extent == iealloc(tsdn, ptr));
+	*newsize = extent_usize_get(extent);
+
+	return ret;
+}
+
+static void *
+arena_ralloc_move_helper(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache) {
+	if (alignment == 0) {
+		return arena_malloc(tsdn, arena, usize, sz_size2index(usize),
+		    zero, tcache, true);
+	}
+	usize = sz_sa2u(usize, alignment);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+	return ipalloct(tsdn, usize, alignment, zero, tcache, arena);
+}
+
+void *
+arena_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t oldsize,
+    size_t size, size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args) {
+	size_t usize = sz_s2u(size);
+	if (unlikely(usize == 0 || size > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+
+	if (likely(usize <= SC_SMALL_MAXCLASS)) {
+		/* Try to avoid moving the allocation. */
+		UNUSED size_t newsize;
+		if (!arena_ralloc_no_move(tsdn, ptr, oldsize, usize, 0, zero,
+		    &newsize)) {
+			hook_invoke_expand(hook_args->is_realloc
+			    ? hook_expand_realloc : hook_expand_rallocx,
+			    ptr, oldsize, usize, (uintptr_t)ptr,
+			    hook_args->args);
+			return ptr;
+		}
+	}
+
+	if (oldsize >= SC_LARGE_MINCLASS
+	    && usize >= SC_LARGE_MINCLASS) {
+		return large_ralloc(tsdn, arena, ptr, usize,
+		    alignment, zero, tcache, hook_args);
+	}
+
+	/*
+	 * size and oldsize are different enough that we need to move the
+	 * object.  In that case, fall back to allocating new space and copying.
+	 */
+	void *ret = arena_ralloc_move_helper(tsdn, arena, usize, alignment,
+	    zero, tcache);
+	if (ret == NULL) {
+		return NULL;
+	}
+
+	hook_invoke_alloc(hook_args->is_realloc
+	    ? hook_alloc_realloc : hook_alloc_rallocx, ret, (uintptr_t)ret,
+	    hook_args->args);
+	hook_invoke_dalloc(hook_args->is_realloc
+	    ? hook_dalloc_realloc : hook_dalloc_rallocx, ptr, hook_args->args);
+
+	/*
+	 * Junk/zero-filling were already done by
+	 * ipalloc()/arena_malloc().
+	 */
+	size_t copysize = (usize < oldsize) ? usize : oldsize;
+	memcpy(ret, ptr, copysize);
+	isdalloct(tsdn, ptr, oldsize, tcache, NULL, true);
+	return ret;
+}
+
+dss_prec_t
+arena_dss_prec_get(arena_t *arena) {
+	return (dss_prec_t)atomic_load_u(&arena->dss_prec, ATOMIC_ACQUIRE);
+}
+
+bool
+arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec) {
+	if (!have_dss) {
+		return (dss_prec != dss_prec_disabled);
+	}
+	atomic_store_u(&arena->dss_prec, (unsigned)dss_prec, ATOMIC_RELEASE);
+	return false;
+}
+
+ssize_t
+arena_dirty_decay_ms_default_get(void) {
+	return atomic_load_zd(&dirty_decay_ms_default, ATOMIC_RELAXED);
+}
+
+bool
+arena_dirty_decay_ms_default_set(ssize_t decay_ms) {
+	if (!arena_decay_ms_valid(decay_ms)) {
+		return true;
+	}
+	atomic_store_zd(&dirty_decay_ms_default, decay_ms, ATOMIC_RELAXED);
+	return false;
+}
+
+ssize_t
+arena_muzzy_decay_ms_default_get(void) {
+	return atomic_load_zd(&muzzy_decay_ms_default, ATOMIC_RELAXED);
+}
+
+bool
+arena_muzzy_decay_ms_default_set(ssize_t decay_ms) {
+	if (!arena_decay_ms_valid(decay_ms)) {
+		return true;
+	}
+	atomic_store_zd(&muzzy_decay_ms_default, decay_ms, ATOMIC_RELAXED);
+	return false;
+}
+
+bool
+arena_retain_grow_limit_get_set(tsd_t *tsd, arena_t *arena, size_t *old_limit,
+    size_t *new_limit) {
+	assert(opt_retain);
+
+	pszind_t new_ind JEMALLOC_CC_SILENCE_INIT(0);
+	if (new_limit != NULL) {
+		size_t limit = *new_limit;
+		/* Grow no more than the new limit. */
+		if ((new_ind = sz_psz2ind(limit + 1) - 1) >= SC_NPSIZES) {
+			return true;
+		}
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &arena->extent_grow_mtx);
+	if (old_limit != NULL) {
+		*old_limit = sz_pind2sz(arena->retain_grow_limit);
+	}
+	if (new_limit != NULL) {
+		arena->retain_grow_limit = new_ind;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &arena->extent_grow_mtx);
+
+	return false;
+}
+
+unsigned
+arena_nthreads_get(arena_t *arena, bool internal) {
+	return atomic_load_u(&arena->nthreads[internal], ATOMIC_RELAXED);
+}
+
+void
+arena_nthreads_inc(arena_t *arena, bool internal) {
+	atomic_fetch_add_u(&arena->nthreads[internal], 1, ATOMIC_RELAXED);
+}
+
+void
+arena_nthreads_dec(arena_t *arena, bool internal) {
+	atomic_fetch_sub_u(&arena->nthreads[internal], 1, ATOMIC_RELAXED);
+}
+
+size_t
+arena_extent_sn_next(arena_t *arena) {
+	return atomic_fetch_add_zu(&arena->extent_sn_next, 1, ATOMIC_RELAXED);
+}
+
+arena_t *
+arena_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	arena_t *arena;
+	base_t *base;
+	unsigned i;
+
+	if (ind == 0) {
+		base = b0get();
+	} else {
+		base = base_new(tsdn, ind, extent_hooks);
+		if (base == NULL) {
+			return NULL;
+		}
+	}
+
+	unsigned nbins_total = 0;
+	for (i = 0; i < SC_NBINS; i++) {
+		nbins_total += bin_infos[i].n_shards;
+	}
+	size_t arena_size = sizeof(arena_t) + sizeof(bin_t) * nbins_total;
+	arena = (arena_t *)base_alloc(tsdn, base, arena_size, CACHELINE);
+	if (arena == NULL) {
+		goto label_error;
+	}
+
+	atomic_store_u(&arena->nthreads[0], 0, ATOMIC_RELAXED);
+	atomic_store_u(&arena->nthreads[1], 0, ATOMIC_RELAXED);
+	arena->last_thd = NULL;
+
+	if (config_stats) {
+		if (arena_stats_init(tsdn, &arena->stats)) {
+			goto label_error;
+		}
+
+		ql_new(&arena->tcache_ql);
+		ql_new(&arena->cache_bin_array_descriptor_ql);
+		if (malloc_mutex_init(&arena->tcache_ql_mtx, "tcache_ql",
+		    WITNESS_RANK_TCACHE_QL, malloc_mutex_rank_exclusive)) {
+			goto label_error;
+		}
+	}
+
+	if (config_prof) {
+		if (prof_accum_init(tsdn, &arena->prof_accum)) {
+			goto label_error;
+		}
+	}
+
+	if (config_cache_oblivious) {
+		/*
+		 * A nondeterministic seed based on the address of arena reduces
+		 * the likelihood of lockstep non-uniform cache index
+		 * utilization among identical concurrent processes, but at the
+		 * cost of test repeatability.  For debug builds, instead use a
+		 * deterministic seed.
+		 */
+		atomic_store_zu(&arena->offset_state, config_debug ? ind :
+		    (size_t)(uintptr_t)arena, ATOMIC_RELAXED);
+	}
+
+	atomic_store_zu(&arena->extent_sn_next, 0, ATOMIC_RELAXED);
+
+	atomic_store_u(&arena->dss_prec, (unsigned)extent_dss_prec_get(),
+	    ATOMIC_RELAXED);
+
+	atomic_store_zu(&arena->nactive, 0, ATOMIC_RELAXED);
+
+	extent_list_init(&arena->large);
+	if (malloc_mutex_init(&arena->large_mtx, "arena_large",
+	    WITNESS_RANK_ARENA_LARGE, malloc_mutex_rank_exclusive)) {
+		goto label_error;
+	}
+
+	/*
+	 * Delay coalescing for dirty extents despite the disruptive effect on
+	 * memory layout for best-fit extent allocation, since cached extents
+	 * are likely to be reused soon after deallocation, and the cost of
+	 * merging/splitting extents is non-trivial.
+	 */
+	if (extents_init(tsdn, &arena->extents_dirty, extent_state_dirty,
+	    true)) {
+		goto label_error;
+	}
+	/*
+	 * Coalesce muzzy extents immediately, because operations on them are in
+	 * the critical path much less often than for dirty extents.
+	 */
+	if (extents_init(tsdn, &arena->extents_muzzy, extent_state_muzzy,
+	    false)) {
+		goto label_error;
+	}
+	/*
+	 * Coalesce retained extents immediately, in part because they will
+	 * never be evicted (and therefore there's no opportunity for delayed
+	 * coalescing), but also because operations on retained extents are not
+	 * in the critical path.
+	 */
+	if (extents_init(tsdn, &arena->extents_retained, extent_state_retained,
+	    false)) {
+		goto label_error;
+	}
+
+	if (arena_decay_init(&arena->decay_dirty,
+	    arena_dirty_decay_ms_default_get(), &arena->stats.decay_dirty)) {
+		goto label_error;
+	}
+	if (arena_decay_init(&arena->decay_muzzy,
+	    arena_muzzy_decay_ms_default_get(), &arena->stats.decay_muzzy)) {
+		goto label_error;
+	}
+
+	arena->extent_grow_next = sz_psz2ind(HUGEPAGE);
+	arena->retain_grow_limit = sz_psz2ind(SC_LARGE_MAXCLASS);
+	if (malloc_mutex_init(&arena->extent_grow_mtx, "extent_grow",
+	    WITNESS_RANK_EXTENT_GROW, malloc_mutex_rank_exclusive)) {
+		goto label_error;
+	}
+
+	extent_avail_new(&arena->extent_avail);
+	if (malloc_mutex_init(&arena->extent_avail_mtx, "extent_avail",
+	    WITNESS_RANK_EXTENT_AVAIL, malloc_mutex_rank_exclusive)) {
+		goto label_error;
+	}
+
+	/* Initialize bins. */
+	uintptr_t bin_addr = (uintptr_t)arena + sizeof(arena_t);
+	atomic_store_u(&arena->binshard_next, 0, ATOMIC_RELEASE);
+	for (i = 0; i < SC_NBINS; i++) {
+		unsigned nshards = bin_infos[i].n_shards;
+		arena->bins[i].bin_shards = (bin_t *)bin_addr;
+		bin_addr += nshards * sizeof(bin_t);
+		for (unsigned j = 0; j < nshards; j++) {
+			bool err = bin_init(&arena->bins[i].bin_shards[j]);
+			if (err) {
+				goto label_error;
+			}
+		}
+	}
+	assert(bin_addr == (uintptr_t)arena + arena_size);
+
+	arena->base = base;
+	/* Set arena before creating background threads. */
+	arena_set(ind, arena);
+
+	nstime_init(&arena->create_time, 0);
+	nstime_update(&arena->create_time);
+
+	/* We don't support reentrancy for arena 0 bootstrapping. */
+	if (ind != 0) {
+		/*
+		 * If we're here, then arena 0 already exists, so bootstrapping
+		 * is done enough that we should have tsd.
+		 */
+		assert(!tsdn_null(tsdn));
+		pre_reentrancy(tsdn_tsd(tsdn), arena);
+		if (test_hooks_arena_new_hook) {
+			test_hooks_arena_new_hook();
+		}
+		post_reentrancy(tsdn_tsd(tsdn));
+	}
+
+	return arena;
+label_error:
+	if (ind != 0) {
+		base_delete(tsdn, base);
+	}
+	return NULL;
+}
+
+arena_t *
+arena_choose_huge(tsd_t *tsd) {
+	/* huge_arena_ind can be 0 during init (will use a0). */
+	if (huge_arena_ind == 0) {
+		assert(!malloc_initialized());
+	}
+
+	arena_t *huge_arena = arena_get(tsd_tsdn(tsd), huge_arena_ind, false);
+	if (huge_arena == NULL) {
+		/* Create the huge arena on demand. */
+		assert(huge_arena_ind != 0);
+		huge_arena = arena_get(tsd_tsdn(tsd), huge_arena_ind, true);
+		if (huge_arena == NULL) {
+			return NULL;
+		}
+		/*
+		 * Purge eagerly for huge allocations, because: 1) number of
+		 * huge allocations is usually small, which means ticker based
+		 * decay is not reliable; and 2) less immediate reuse is
+		 * expected for huge allocations.
+		 */
+		if (arena_dirty_decay_ms_default_get() > 0) {
+			arena_dirty_decay_ms_set(tsd_tsdn(tsd), huge_arena, 0);
+		}
+		if (arena_muzzy_decay_ms_default_get() > 0) {
+			arena_muzzy_decay_ms_set(tsd_tsdn(tsd), huge_arena, 0);
+		}
+	}
+
+	return huge_arena;
+}
+
+bool
+arena_init_huge(void) {
+	bool huge_enabled;
+
+	/* The threshold should be large size class. */
+	if (opt_oversize_threshold > SC_LARGE_MAXCLASS ||
+	    opt_oversize_threshold < SC_LARGE_MINCLASS) {
+		opt_oversize_threshold = 0;
+		oversize_threshold = SC_LARGE_MAXCLASS + PAGE;
+		huge_enabled = false;
+	} else {
+		/* Reserve the index for the huge arena. */
+		huge_arena_ind = narenas_total_get();
+		oversize_threshold = opt_oversize_threshold;
+		huge_enabled = true;
+	}
+
+	return huge_enabled;
+}
+
+bool
+arena_is_huge(unsigned arena_ind) {
+	if (huge_arena_ind == 0) {
+		return false;
+	}
+	return (arena_ind == huge_arena_ind);
+}
+
+void
+arena_boot(sc_data_t *sc_data) {
+	arena_dirty_decay_ms_default_set(opt_dirty_decay_ms);
+	arena_muzzy_decay_ms_default_set(opt_muzzy_decay_ms);
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		sc_t *sc = &sc_data->sc[i];
+		div_init(&arena_binind_div_info[i],
+		    (1U << sc->lg_base) + (sc->ndelta << sc->lg_delta));
+	}
+}
+
+void
+arena_prefork0(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_prefork(tsdn, &arena->decay_muzzy.mtx);
+}
+
+void
+arena_prefork1(tsdn_t *tsdn, arena_t *arena) {
+	if (config_stats) {
+		malloc_mutex_prefork(tsdn, &arena->tcache_ql_mtx);
+	}
+}
+
+void
+arena_prefork2(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->extent_grow_mtx);
+}
+
+void
+arena_prefork3(tsdn_t *tsdn, arena_t *arena) {
+	extents_prefork(tsdn, &arena->extents_dirty);
+	extents_prefork(tsdn, &arena->extents_muzzy);
+	extents_prefork(tsdn, &arena->extents_retained);
+}
+
+void
+arena_prefork4(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->extent_avail_mtx);
+}
+
+void
+arena_prefork5(tsdn_t *tsdn, arena_t *arena) {
+	base_prefork(tsdn, arena->base);
+}
+
+void
+arena_prefork6(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->large_mtx);
+}
+
+void
+arena_prefork7(tsdn_t *tsdn, arena_t *arena) {
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_prefork(tsdn, &arena->bins[i].bin_shards[j]);
+		}
+	}
+}
+
+void
+arena_postfork_parent(tsdn_t *tsdn, arena_t *arena) {
+	unsigned i;
+
+	for (i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_postfork_parent(tsdn,
+			    &arena->bins[i].bin_shards[j]);
+		}
+	}
+	malloc_mutex_postfork_parent(tsdn, &arena->large_mtx);
+	base_postfork_parent(tsdn, arena->base);
+	malloc_mutex_postfork_parent(tsdn, &arena->extent_avail_mtx);
+	extents_postfork_parent(tsdn, &arena->extents_dirty);
+	extents_postfork_parent(tsdn, &arena->extents_muzzy);
+	extents_postfork_parent(tsdn, &arena->extents_retained);
+	malloc_mutex_postfork_parent(tsdn, &arena->extent_grow_mtx);
+	malloc_mutex_postfork_parent(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_postfork_parent(tsdn, &arena->decay_muzzy.mtx);
+	if (config_stats) {
+		malloc_mutex_postfork_parent(tsdn, &arena->tcache_ql_mtx);
+	}
+}
+
+void
+arena_postfork_child(tsdn_t *tsdn, arena_t *arena) {
+	unsigned i;
+
+	atomic_store_u(&arena->nthreads[0], 0, ATOMIC_RELAXED);
+	atomic_store_u(&arena->nthreads[1], 0, ATOMIC_RELAXED);
+	if (tsd_arena_get(tsdn_tsd(tsdn)) == arena) {
+		arena_nthreads_inc(arena, false);
+	}
+	if (tsd_iarena_get(tsdn_tsd(tsdn)) == arena) {
+		arena_nthreads_inc(arena, true);
+	}
+	if (config_stats) {
+		ql_new(&arena->tcache_ql);
+		ql_new(&arena->cache_bin_array_descriptor_ql);
+		tcache_t *tcache = tcache_get(tsdn_tsd(tsdn));
+		if (tcache != NULL && tcache->arena == arena) {
+			ql_elm_new(tcache, link);
+			ql_tail_insert(&arena->tcache_ql, tcache, link);
+			cache_bin_array_descriptor_init(
+			    &tcache->cache_bin_array_descriptor,
+			    tcache->bins_small, tcache->bins_large);
+			ql_tail_insert(&arena->cache_bin_array_descriptor_ql,
+			    &tcache->cache_bin_array_descriptor, link);
+		}
+	}
+
+	for (i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_postfork_child(tsdn, &arena->bins[i].bin_shards[j]);
+		}
+	}
+	malloc_mutex_postfork_child(tsdn, &arena->large_mtx);
+	base_postfork_child(tsdn, arena->base);
+	malloc_mutex_postfork_child(tsdn, &arena->extent_avail_mtx);
+	extents_postfork_child(tsdn, &arena->extents_dirty);
+	extents_postfork_child(tsdn, &arena->extents_muzzy);
+	extents_postfork_child(tsdn, &arena->extents_retained);
+	malloc_mutex_postfork_child(tsdn, &arena->extent_grow_mtx);
+	malloc_mutex_postfork_child(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_postfork_child(tsdn, &arena->decay_muzzy.mtx);
+	if (config_stats) {
+		malloc_mutex_postfork_child(tsdn, &arena->tcache_ql_mtx);
+	}
+}
diff --git a/jemalloc/src/background_thread.c b/jemalloc/src/background_thread.c
new file mode 100644
index 0000000..57b9b25
--- /dev/null
+++ b/jemalloc/src/background_thread.c
@@ -0,0 +1,939 @@
+#define JEMALLOC_BACKGROUND_THREAD_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+/******************************************************************************/
+/* Data. */
+
+/* This option should be opt-in only. */
+#define BACKGROUND_THREAD_DEFAULT false
+/* Read-only after initialization. */
+bool opt_background_thread = BACKGROUND_THREAD_DEFAULT;
+size_t opt_max_background_threads = MAX_BACKGROUND_THREAD_LIMIT + 1;
+
+/* Used for thread creation, termination and stats. */
+malloc_mutex_t background_thread_lock;
+/* Indicates global state.  Atomic because decay reads this w/o locking. */
+atomic_b_t background_thread_enabled_state;
+size_t n_background_threads;
+size_t max_background_threads;
+/* Thread info per-index. */
+background_thread_info_t *background_thread_info;
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+
+static int (*pthread_create_fptr)(pthread_t *__restrict, const pthread_attr_t *,
+    void *(*)(void *), void *__restrict);
+
+static void
+pthread_create_wrapper_init(void) {
+#ifdef JEMALLOC_LAZY_LOCK
+	if (!isthreaded) {
+		isthreaded = true;
+	}
+#endif
+}
+
+int
+pthread_create_wrapper(pthread_t *__restrict thread, const pthread_attr_t *attr,
+    void *(*start_routine)(void *), void *__restrict arg) {
+	pthread_create_wrapper_init();
+
+	return pthread_create_fptr(thread, attr, start_routine, arg);
+}
+#endif /* JEMALLOC_PTHREAD_CREATE_WRAPPER */
+
+#ifndef JEMALLOC_BACKGROUND_THREAD
+#define NOT_REACHED { not_reached(); }
+bool background_thread_create(tsd_t *tsd, unsigned arena_ind) NOT_REACHED
+bool background_threads_enable(tsd_t *tsd) NOT_REACHED
+bool background_threads_disable(tsd_t *tsd) NOT_REACHED
+void background_thread_interval_check(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, size_t npages_new) NOT_REACHED
+void background_thread_prefork0(tsdn_t *tsdn) NOT_REACHED
+void background_thread_prefork1(tsdn_t *tsdn) NOT_REACHED
+void background_thread_postfork_parent(tsdn_t *tsdn) NOT_REACHED
+void background_thread_postfork_child(tsdn_t *tsdn) NOT_REACHED
+bool background_thread_stats_read(tsdn_t *tsdn,
+    background_thread_stats_t *stats) NOT_REACHED
+void background_thread_ctl_init(tsdn_t *tsdn) NOT_REACHED
+#undef NOT_REACHED
+#else
+
+static bool background_thread_enabled_at_fork;
+
+static void
+background_thread_info_init(tsdn_t *tsdn, background_thread_info_t *info) {
+	background_thread_wakeup_time_set(tsdn, info, 0);
+	info->npages_to_purge_new = 0;
+	if (config_stats) {
+		info->tot_n_runs = 0;
+		nstime_init(&info->tot_sleep_time, 0);
+	}
+}
+
+static inline bool
+set_current_thread_affinity(int cpu) {
+#if defined(JEMALLOC_HAVE_SCHED_SETAFFINITY)
+	cpu_set_t cpuset;
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+	int ret = sched_setaffinity(0, sizeof(cpu_set_t), &cpuset);
+
+	return (ret != 0);
+#else
+	return false;
+#endif
+}
+
+/* Threshold for determining when to wake up the background thread. */
+#define BACKGROUND_THREAD_NPAGES_THRESHOLD UINT64_C(1024)
+#define BILLION UINT64_C(1000000000)
+/* Minimal sleep interval 100 ms. */
+#define BACKGROUND_THREAD_MIN_INTERVAL_NS (BILLION / 10)
+
+static inline size_t
+decay_npurge_after_interval(arena_decay_t *decay, size_t interval) {
+	size_t i;
+	uint64_t sum = 0;
+	for (i = 0; i < interval; i++) {
+		sum += decay->backlog[i] * h_steps[i];
+	}
+	for (; i < SMOOTHSTEP_NSTEPS; i++) {
+		sum += decay->backlog[i] * (h_steps[i] - h_steps[i - interval]);
+	}
+
+	return (size_t)(sum >> SMOOTHSTEP_BFP);
+}
+
+static uint64_t
+arena_decay_compute_purge_interval_impl(tsdn_t *tsdn, arena_decay_t *decay,
+    extents_t *extents) {
+	if (malloc_mutex_trylock(tsdn, &decay->mtx)) {
+		/* Use minimal interval if decay is contended. */
+		return BACKGROUND_THREAD_MIN_INTERVAL_NS;
+	}
+
+	uint64_t interval;
+	ssize_t decay_time = atomic_load_zd(&decay->time_ms, ATOMIC_RELAXED);
+	if (decay_time <= 0) {
+		/* Purging is eagerly done or disabled currently. */
+		interval = BACKGROUND_THREAD_INDEFINITE_SLEEP;
+		goto label_done;
+	}
+
+	uint64_t decay_interval_ns = nstime_ns(&decay->interval);
+	assert(decay_interval_ns > 0);
+	size_t npages = extents_npages_get(extents);
+	if (npages == 0) {
+		unsigned i;
+		for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+			if (decay->backlog[i] > 0) {
+				break;
+			}
+		}
+		if (i == SMOOTHSTEP_NSTEPS) {
+			/* No dirty pages recorded.  Sleep indefinitely. */
+			interval = BACKGROUND_THREAD_INDEFINITE_SLEEP;
+			goto label_done;
+		}
+	}
+	if (npages <= BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		/* Use max interval. */
+		interval = decay_interval_ns * SMOOTHSTEP_NSTEPS;
+		goto label_done;
+	}
+
+	size_t lb = BACKGROUND_THREAD_MIN_INTERVAL_NS / decay_interval_ns;
+	size_t ub = SMOOTHSTEP_NSTEPS;
+	/* Minimal 2 intervals to ensure reaching next epoch deadline. */
+	lb = (lb < 2) ? 2 : lb;
+	if ((decay_interval_ns * ub <= BACKGROUND_THREAD_MIN_INTERVAL_NS) ||
+	    (lb + 2 > ub)) {
+		interval = BACKGROUND_THREAD_MIN_INTERVAL_NS;
+		goto label_done;
+	}
+
+	assert(lb + 2 <= ub);
+	size_t npurge_lb, npurge_ub;
+	npurge_lb = decay_npurge_after_interval(decay, lb);
+	if (npurge_lb > BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		interval = decay_interval_ns * lb;
+		goto label_done;
+	}
+	npurge_ub = decay_npurge_after_interval(decay, ub);
+	if (npurge_ub < BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		interval = decay_interval_ns * ub;
+		goto label_done;
+	}
+
+	unsigned n_search = 0;
+	size_t target, npurge;
+	while ((npurge_lb + BACKGROUND_THREAD_NPAGES_THRESHOLD < npurge_ub)
+	    && (lb + 2 < ub)) {
+		target = (lb + ub) / 2;
+		npurge = decay_npurge_after_interval(decay, target);
+		if (npurge > BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+			ub = target;
+			npurge_ub = npurge;
+		} else {
+			lb = target;
+			npurge_lb = npurge;
+		}
+		assert(n_search++ < lg_floor(SMOOTHSTEP_NSTEPS) + 1);
+	}
+	interval = decay_interval_ns * (ub + lb) / 2;
+label_done:
+	interval = (interval < BACKGROUND_THREAD_MIN_INTERVAL_NS) ?
+	    BACKGROUND_THREAD_MIN_INTERVAL_NS : interval;
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	return interval;
+}
+
+/* Compute purge interval for background threads. */
+static uint64_t
+arena_decay_compute_purge_interval(tsdn_t *tsdn, arena_t *arena) {
+	uint64_t i1, i2;
+	i1 = arena_decay_compute_purge_interval_impl(tsdn, &arena->decay_dirty,
+	    &arena->extents_dirty);
+	if (i1 == BACKGROUND_THREAD_MIN_INTERVAL_NS) {
+		return i1;
+	}
+	i2 = arena_decay_compute_purge_interval_impl(tsdn, &arena->decay_muzzy,
+	    &arena->extents_muzzy);
+
+	return i1 < i2 ? i1 : i2;
+}
+
+static void
+background_thread_sleep(tsdn_t *tsdn, background_thread_info_t *info,
+    uint64_t interval) {
+	if (config_stats) {
+		info->tot_n_runs++;
+	}
+	info->npages_to_purge_new = 0;
+
+	struct timeval tv;
+	/* Specific clock required by timedwait. */
+	gettimeofday(&tv, NULL);
+	nstime_t before_sleep;
+	nstime_init2(&before_sleep, tv.tv_sec, tv.tv_usec * 1000);
+
+	int ret;
+	if (interval == BACKGROUND_THREAD_INDEFINITE_SLEEP) {
+		assert(background_thread_indefinite_sleep(info));
+		ret = pthread_cond_wait(&info->cond, &info->mtx.lock);
+		assert(ret == 0);
+	} else {
+		assert(interval >= BACKGROUND_THREAD_MIN_INTERVAL_NS &&
+		    interval <= BACKGROUND_THREAD_INDEFINITE_SLEEP);
+		/* We need malloc clock (can be different from tv). */
+		nstime_t next_wakeup;
+		nstime_init(&next_wakeup, 0);
+		nstime_update(&next_wakeup);
+		nstime_iadd(&next_wakeup, interval);
+		assert(nstime_ns(&next_wakeup) <
+		    BACKGROUND_THREAD_INDEFINITE_SLEEP);
+		background_thread_wakeup_time_set(tsdn, info,
+		    nstime_ns(&next_wakeup));
+
+		nstime_t ts_wakeup;
+		nstime_copy(&ts_wakeup, &before_sleep);
+		nstime_iadd(&ts_wakeup, interval);
+		struct timespec ts;
+		ts.tv_sec = (size_t)nstime_sec(&ts_wakeup);
+		ts.tv_nsec = (size_t)nstime_nsec(&ts_wakeup);
+
+		assert(!background_thread_indefinite_sleep(info));
+		ret = pthread_cond_timedwait(&info->cond, &info->mtx.lock, &ts);
+		assert(ret == ETIMEDOUT || ret == 0);
+		background_thread_wakeup_time_set(tsdn, info,
+		    BACKGROUND_THREAD_INDEFINITE_SLEEP);
+	}
+	if (config_stats) {
+		gettimeofday(&tv, NULL);
+		nstime_t after_sleep;
+		nstime_init2(&after_sleep, tv.tv_sec, tv.tv_usec * 1000);
+		if (nstime_compare(&after_sleep, &before_sleep) > 0) {
+			nstime_subtract(&after_sleep, &before_sleep);
+			nstime_add(&info->tot_sleep_time, &after_sleep);
+		}
+	}
+}
+
+static bool
+background_thread_pause_check(tsdn_t *tsdn, background_thread_info_t *info) {
+	if (unlikely(info->state == background_thread_paused)) {
+		malloc_mutex_unlock(tsdn, &info->mtx);
+		/* Wait on global lock to update status. */
+		malloc_mutex_lock(tsdn, &background_thread_lock);
+		malloc_mutex_unlock(tsdn, &background_thread_lock);
+		malloc_mutex_lock(tsdn, &info->mtx);
+		return true;
+	}
+
+	return false;
+}
+
+static inline void
+background_work_sleep_once(tsdn_t *tsdn, background_thread_info_t *info, unsigned ind) {
+	uint64_t min_interval = BACKGROUND_THREAD_INDEFINITE_SLEEP;
+	unsigned narenas = narenas_total_get();
+
+	for (unsigned i = ind; i < narenas; i += max_background_threads) {
+		arena_t *arena = arena_get(tsdn, i, false);
+		if (!arena) {
+			continue;
+		}
+		arena_decay(tsdn, arena, true, false);
+		if (min_interval == BACKGROUND_THREAD_MIN_INTERVAL_NS) {
+			/* Min interval will be used. */
+			continue;
+		}
+		uint64_t interval = arena_decay_compute_purge_interval(tsdn,
+		    arena);
+		assert(interval >= BACKGROUND_THREAD_MIN_INTERVAL_NS);
+		if (min_interval > interval) {
+			min_interval = interval;
+		}
+	}
+	background_thread_sleep(tsdn, info, min_interval);
+}
+
+static bool
+background_threads_disable_single(tsd_t *tsd, background_thread_info_t *info) {
+	if (info == &background_thread_info[0]) {
+		malloc_mutex_assert_owner(tsd_tsdn(tsd),
+		    &background_thread_lock);
+	} else {
+		malloc_mutex_assert_not_owner(tsd_tsdn(tsd),
+		    &background_thread_lock);
+	}
+
+	pre_reentrancy(tsd, NULL);
+	malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	bool has_thread;
+	assert(info->state != background_thread_paused);
+	if (info->state == background_thread_started) {
+		has_thread = true;
+		info->state = background_thread_stopped;
+		pthread_cond_signal(&info->cond);
+	} else {
+		has_thread = false;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+
+	if (!has_thread) {
+		post_reentrancy(tsd);
+		return false;
+	}
+	void *ret;
+	if (pthread_join(info->thread, &ret)) {
+		post_reentrancy(tsd);
+		return true;
+	}
+	assert(ret == NULL);
+	n_background_threads--;
+	post_reentrancy(tsd);
+
+	return false;
+}
+
+static void *background_thread_entry(void *ind_arg);
+
+static int
+background_thread_create_signals_masked(pthread_t *thread,
+    const pthread_attr_t *attr, void *(*start_routine)(void *), void *arg) {
+	/*
+	 * Mask signals during thread creation so that the thread inherits
+	 * an empty signal set.
+	 */
+	sigset_t set;
+	sigfillset(&set);
+	sigset_t oldset;
+	int mask_err = pthread_sigmask(SIG_SETMASK, &set, &oldset);
+	if (mask_err != 0) {
+		return mask_err;
+	}
+	int create_err = pthread_create_wrapper(thread, attr, start_routine,
+	    arg);
+	/*
+	 * Restore the signal mask.  Failure to restore the signal mask here
+	 * changes program behavior.
+	 */
+	int restore_err = pthread_sigmask(SIG_SETMASK, &oldset, NULL);
+	if (restore_err != 0) {
+		malloc_printf("<jemalloc>: background thread creation "
+		    "failed (%d), and signal mask restoration failed "
+		    "(%d)\n", create_err, restore_err);
+		if (opt_abort) {
+			abort();
+		}
+	}
+	return create_err;
+}
+
+static bool
+check_background_thread_creation(tsd_t *tsd, unsigned *n_created,
+    bool *created_threads) {
+	bool ret = false;
+	if (likely(*n_created == n_background_threads)) {
+		return ret;
+	}
+
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+	malloc_mutex_unlock(tsdn, &background_thread_info[0].mtx);
+	for (unsigned i = 1; i < max_background_threads; i++) {
+		if (created_threads[i]) {
+			continue;
+		}
+		background_thread_info_t *info = &background_thread_info[i];
+		malloc_mutex_lock(tsdn, &info->mtx);
+		/*
+		 * In case of the background_thread_paused state because of
+		 * arena reset, delay the creation.
+		 */
+		bool create = (info->state == background_thread_started);
+		malloc_mutex_unlock(tsdn, &info->mtx);
+		if (!create) {
+			continue;
+		}
+
+		pre_reentrancy(tsd, NULL);
+		int err = background_thread_create_signals_masked(&info->thread,
+		    NULL, background_thread_entry, (void *)(uintptr_t)i);
+		post_reentrancy(tsd);
+
+		if (err == 0) {
+			(*n_created)++;
+			created_threads[i] = true;
+		} else {
+			malloc_printf("<jemalloc>: background thread "
+			    "creation failed (%d)\n", err);
+			if (opt_abort) {
+				abort();
+			}
+		}
+		/* Return to restart the loop since we unlocked. */
+		ret = true;
+		break;
+	}
+	malloc_mutex_lock(tsdn, &background_thread_info[0].mtx);
+
+	return ret;
+}
+
+static void
+background_thread0_work(tsd_t *tsd) {
+	/* Thread0 is also responsible for launching / terminating threads. */
+	VARIABLE_ARRAY(bool, created_threads, max_background_threads);
+	unsigned i;
+	for (i = 1; i < max_background_threads; i++) {
+		created_threads[i] = false;
+	}
+	/* Start working, and create more threads when asked. */
+	unsigned n_created = 1;
+	while (background_thread_info[0].state != background_thread_stopped) {
+		if (background_thread_pause_check(tsd_tsdn(tsd),
+		    &background_thread_info[0])) {
+			continue;
+		}
+		if (check_background_thread_creation(tsd, &n_created,
+		    (bool *)&created_threads)) {
+			continue;
+		}
+		background_work_sleep_once(tsd_tsdn(tsd),
+		    &background_thread_info[0], 0);
+	}
+
+	/*
+	 * Shut down other threads at exit.  Note that the ctl thread is holding
+	 * the global background_thread mutex (and is waiting) for us.
+	 */
+	assert(!background_thread_enabled());
+	for (i = 1; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		assert(info->state != background_thread_paused);
+		if (created_threads[i]) {
+			background_threads_disable_single(tsd, info);
+		} else {
+			malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+			if (info->state != background_thread_stopped) {
+				/* The thread was not created. */
+				assert(info->state ==
+				    background_thread_started);
+				n_background_threads--;
+				info->state = background_thread_stopped;
+			}
+			malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		}
+	}
+	background_thread_info[0].state = background_thread_stopped;
+	assert(n_background_threads == 1);
+}
+
+static void
+background_work(tsd_t *tsd, unsigned ind) {
+	background_thread_info_t *info = &background_thread_info[ind];
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	background_thread_wakeup_time_set(tsd_tsdn(tsd), info,
+	    BACKGROUND_THREAD_INDEFINITE_SLEEP);
+	if (ind == 0) {
+		background_thread0_work(tsd);
+	} else {
+		while (info->state != background_thread_stopped) {
+			if (background_thread_pause_check(tsd_tsdn(tsd),
+			    info)) {
+				continue;
+			}
+			background_work_sleep_once(tsd_tsdn(tsd), info, ind);
+		}
+	}
+	assert(info->state == background_thread_stopped);
+	background_thread_wakeup_time_set(tsd_tsdn(tsd), info, 0);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+}
+
+static void *
+background_thread_entry(void *ind_arg) {
+	unsigned thread_ind = (unsigned)(uintptr_t)ind_arg;
+	assert(thread_ind < max_background_threads);
+#ifdef JEMALLOC_HAVE_PTHREAD_SETNAME_NP
+	pthread_setname_np(pthread_self(), "jemalloc_bg_thd");
+#elif defined(__FreeBSD__)
+	pthread_set_name_np(pthread_self(), "jemalloc_bg_thd");
+#endif
+	if (opt_percpu_arena != percpu_arena_disabled) {
+		set_current_thread_affinity((int)thread_ind);
+	}
+	/*
+	 * Start periodic background work.  We use internal tsd which avoids
+	 * side effects, for example triggering new arena creation (which in
+	 * turn triggers another background thread creation).
+	 */
+	background_work(tsd_internal_fetch(), thread_ind);
+	assert(pthread_equal(pthread_self(),
+	    background_thread_info[thread_ind].thread));
+
+	return NULL;
+}
+
+static void
+background_thread_init(tsd_t *tsd, background_thread_info_t *info) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+	info->state = background_thread_started;
+	background_thread_info_init(tsd_tsdn(tsd), info);
+	n_background_threads++;
+}
+
+static bool
+background_thread_create_locked(tsd_t *tsd, unsigned arena_ind) {
+	assert(have_background_thread);
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+
+	/* We create at most NCPUs threads. */
+	size_t thread_ind = arena_ind % max_background_threads;
+	background_thread_info_t *info = &background_thread_info[thread_ind];
+
+	bool need_new_thread;
+	malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	need_new_thread = background_thread_enabled() &&
+	    (info->state == background_thread_stopped);
+	if (need_new_thread) {
+		background_thread_init(tsd, info);
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+	if (!need_new_thread) {
+		return false;
+	}
+	if (arena_ind != 0) {
+		/* Threads are created asynchronously by Thread 0. */
+		background_thread_info_t *t0 = &background_thread_info[0];
+		malloc_mutex_lock(tsd_tsdn(tsd), &t0->mtx);
+		assert(t0->state == background_thread_started);
+		pthread_cond_signal(&t0->cond);
+		malloc_mutex_unlock(tsd_tsdn(tsd), &t0->mtx);
+
+		return false;
+	}
+
+	pre_reentrancy(tsd, NULL);
+	/*
+	 * To avoid complications (besides reentrancy), create internal
+	 * background threads with the underlying pthread_create.
+	 */
+	int err = background_thread_create_signals_masked(&info->thread, NULL,
+	    background_thread_entry, (void *)thread_ind);
+	post_reentrancy(tsd);
+
+	if (err != 0) {
+		malloc_printf("<jemalloc>: arena 0 background thread creation "
+		    "failed (%d)\n", err);
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+		info->state = background_thread_stopped;
+		n_background_threads--;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+
+		return true;
+	}
+
+	return false;
+}
+
+/* Create a new background thread if needed. */
+bool
+background_thread_create(tsd_t *tsd, unsigned arena_ind) {
+	assert(have_background_thread);
+
+	bool ret;
+	malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+	ret = background_thread_create_locked(tsd, arena_ind);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+
+	return ret;
+}
+
+bool
+background_threads_enable(tsd_t *tsd) {
+	assert(n_background_threads == 0);
+	assert(background_thread_enabled());
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+
+	VARIABLE_ARRAY(bool, marked, max_background_threads);
+	unsigned i, nmarked;
+	for (i = 0; i < max_background_threads; i++) {
+		marked[i] = false;
+	}
+	nmarked = 0;
+	/* Thread 0 is required and created at the end. */
+	marked[0] = true;
+	/* Mark the threads we need to create for thread 0. */
+	unsigned n = narenas_total_get();
+	for (i = 1; i < n; i++) {
+		if (marked[i % max_background_threads] ||
+		    arena_get(tsd_tsdn(tsd), i, false) == NULL) {
+			continue;
+		}
+		background_thread_info_t *info = &background_thread_info[
+		    i % max_background_threads];
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+		assert(info->state == background_thread_stopped);
+		background_thread_init(tsd, info);
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		marked[i % max_background_threads] = true;
+		if (++nmarked == max_background_threads) {
+			break;
+		}
+	}
+
+	return background_thread_create_locked(tsd, 0);
+}
+
+bool
+background_threads_disable(tsd_t *tsd) {
+	assert(!background_thread_enabled());
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+
+	/* Thread 0 will be responsible for terminating other threads. */
+	if (background_threads_disable_single(tsd,
+	    &background_thread_info[0])) {
+		return true;
+	}
+	assert(n_background_threads == 0);
+
+	return false;
+}
+
+/* Check if we need to signal the background thread early. */
+void
+background_thread_interval_check(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, size_t npages_new) {
+	background_thread_info_t *info = arena_background_thread_info_get(
+	    arena);
+	if (malloc_mutex_trylock(tsdn, &info->mtx)) {
+		/*
+		 * Background thread may hold the mutex for a long period of
+		 * time.  We'd like to avoid the variance on application
+		 * threads.  So keep this non-blocking, and leave the work to a
+		 * future epoch.
+		 */
+		return;
+	}
+
+	if (info->state != background_thread_started) {
+		goto label_done;
+	}
+	if (malloc_mutex_trylock(tsdn, &decay->mtx)) {
+		goto label_done;
+	}
+
+	ssize_t decay_time = atomic_load_zd(&decay->time_ms, ATOMIC_RELAXED);
+	if (decay_time <= 0) {
+		/* Purging is eagerly done or disabled currently. */
+		goto label_done_unlock2;
+	}
+	uint64_t decay_interval_ns = nstime_ns(&decay->interval);
+	assert(decay_interval_ns > 0);
+
+	nstime_t diff;
+	nstime_init(&diff, background_thread_wakeup_time_get(info));
+	if (nstime_compare(&diff, &decay->epoch) <= 0) {
+		goto label_done_unlock2;
+	}
+	nstime_subtract(&diff, &decay->epoch);
+	if (nstime_ns(&diff) < BACKGROUND_THREAD_MIN_INTERVAL_NS) {
+		goto label_done_unlock2;
+	}
+
+	if (npages_new > 0) {
+		size_t n_epoch = (size_t)(nstime_ns(&diff) / decay_interval_ns);
+		/*
+		 * Compute how many new pages we would need to purge by the next
+		 * wakeup, which is used to determine if we should signal the
+		 * background thread.
+		 */
+		uint64_t npurge_new;
+		if (n_epoch >= SMOOTHSTEP_NSTEPS) {
+			npurge_new = npages_new;
+		} else {
+			uint64_t h_steps_max = h_steps[SMOOTHSTEP_NSTEPS - 1];
+			assert(h_steps_max >=
+			    h_steps[SMOOTHSTEP_NSTEPS - 1 - n_epoch]);
+			npurge_new = npages_new * (h_steps_max -
+			    h_steps[SMOOTHSTEP_NSTEPS - 1 - n_epoch]);
+			npurge_new >>= SMOOTHSTEP_BFP;
+		}
+		info->npages_to_purge_new += npurge_new;
+	}
+
+	bool should_signal;
+	if (info->npages_to_purge_new > BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		should_signal = true;
+	} else if (unlikely(background_thread_indefinite_sleep(info)) &&
+	    (extents_npages_get(&arena->extents_dirty) > 0 ||
+	    extents_npages_get(&arena->extents_muzzy) > 0 ||
+	    info->npages_to_purge_new > 0)) {
+		should_signal = true;
+	} else {
+		should_signal = false;
+	}
+
+	if (should_signal) {
+		info->npages_to_purge_new = 0;
+		pthread_cond_signal(&info->cond);
+	}
+label_done_unlock2:
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+label_done:
+	malloc_mutex_unlock(tsdn, &info->mtx);
+}
+
+void
+background_thread_prefork0(tsdn_t *tsdn) {
+	malloc_mutex_prefork(tsdn, &background_thread_lock);
+	background_thread_enabled_at_fork = background_thread_enabled();
+}
+
+void
+background_thread_prefork1(tsdn_t *tsdn) {
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		malloc_mutex_prefork(tsdn, &background_thread_info[i].mtx);
+	}
+}
+
+void
+background_thread_postfork_parent(tsdn_t *tsdn) {
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		malloc_mutex_postfork_parent(tsdn,
+		    &background_thread_info[i].mtx);
+	}
+	malloc_mutex_postfork_parent(tsdn, &background_thread_lock);
+}
+
+void
+background_thread_postfork_child(tsdn_t *tsdn) {
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		malloc_mutex_postfork_child(tsdn,
+		    &background_thread_info[i].mtx);
+	}
+	malloc_mutex_postfork_child(tsdn, &background_thread_lock);
+	if (!background_thread_enabled_at_fork) {
+		return;
+	}
+
+	/* Clear background_thread state (reset to disabled for child). */
+	malloc_mutex_lock(tsdn, &background_thread_lock);
+	n_background_threads = 0;
+	background_thread_enabled_set(tsdn, false);
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		malloc_mutex_lock(tsdn, &info->mtx);
+		info->state = background_thread_stopped;
+		int ret = pthread_cond_init(&info->cond, NULL);
+		assert(ret == 0);
+		background_thread_info_init(tsdn, info);
+		malloc_mutex_unlock(tsdn, &info->mtx);
+	}
+	malloc_mutex_unlock(tsdn, &background_thread_lock);
+}
+
+bool
+background_thread_stats_read(tsdn_t *tsdn, background_thread_stats_t *stats) {
+	assert(config_stats);
+	malloc_mutex_lock(tsdn, &background_thread_lock);
+	if (!background_thread_enabled()) {
+		malloc_mutex_unlock(tsdn, &background_thread_lock);
+		return true;
+	}
+
+	stats->num_threads = n_background_threads;
+	uint64_t num_runs = 0;
+	nstime_init(&stats->run_interval, 0);
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		if (malloc_mutex_trylock(tsdn, &info->mtx)) {
+			/*
+			 * Each background thread run may take a long time;
+			 * avoid waiting on the stats if the thread is active.
+			 */
+			continue;
+		}
+		if (info->state != background_thread_stopped) {
+			num_runs += info->tot_n_runs;
+			nstime_add(&stats->run_interval, &info->tot_sleep_time);
+		}
+		malloc_mutex_unlock(tsdn, &info->mtx);
+	}
+	stats->num_runs = num_runs;
+	if (num_runs > 0) {
+		nstime_idivide(&stats->run_interval, num_runs);
+	}
+	malloc_mutex_unlock(tsdn, &background_thread_lock);
+
+	return false;
+}
+
+#undef BACKGROUND_THREAD_NPAGES_THRESHOLD
+#undef BILLION
+#undef BACKGROUND_THREAD_MIN_INTERVAL_NS
+
+#ifdef JEMALLOC_HAVE_DLSYM
+#include <dlfcn.h>
+#endif
+
+static bool
+pthread_create_fptr_init(void) {
+	if (pthread_create_fptr != NULL) {
+		return false;
+	}
+	/*
+	 * Try the next symbol first, because 1) when use lazy_lock we have a
+	 * wrapper for pthread_create; and 2) application may define its own
+	 * wrapper as well (and can call malloc within the wrapper).
+	 */
+#ifdef JEMALLOC_HAVE_DLSYM
+	pthread_create_fptr = dlsym(RTLD_NEXT, "pthread_create");
+#else
+	pthread_create_fptr = NULL;
+#endif
+	if (pthread_create_fptr == NULL) {
+		if (config_lazy_lock) {
+			malloc_write("<jemalloc>: Error in dlsym(RTLD_NEXT, "
+			    "\"pthread_create\")\n");
+			abort();
+		} else {
+			/* Fall back to the default symbol. */
+			pthread_create_fptr = pthread_create;
+		}
+	}
+
+	return false;
+}
+
+/*
+ * When lazy lock is enabled, we need to make sure setting isthreaded before
+ * taking any background_thread locks.  This is called early in ctl (instead of
+ * wait for the pthread_create calls to trigger) because the mutex is required
+ * before creating background threads.
+ */
+void
+background_thread_ctl_init(tsdn_t *tsdn) {
+	malloc_mutex_assert_not_owner(tsdn, &background_thread_lock);
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+	pthread_create_fptr_init();
+	pthread_create_wrapper_init();
+#endif
+}
+
+#endif /* defined(JEMALLOC_BACKGROUND_THREAD) */
+
+bool
+background_thread_boot0(void) {
+	if (!have_background_thread && opt_background_thread) {
+		malloc_printf("<jemalloc>: option background_thread currently "
+		    "supports pthread only\n");
+		return true;
+	}
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+	if ((config_lazy_lock || opt_background_thread) &&
+	    pthread_create_fptr_init()) {
+		return true;
+	}
+#endif
+	return false;
+}
+
+bool
+background_thread_boot1(tsdn_t *tsdn) {
+#ifdef JEMALLOC_BACKGROUND_THREAD
+	assert(have_background_thread);
+	assert(narenas_total_get() > 0);
+
+	if (opt_max_background_threads > MAX_BACKGROUND_THREAD_LIMIT) {
+		opt_max_background_threads = DEFAULT_NUM_BACKGROUND_THREAD;
+	}
+	max_background_threads = opt_max_background_threads;
+
+	background_thread_enabled_set(tsdn, opt_background_thread);
+	if (malloc_mutex_init(&background_thread_lock,
+	    "background_thread_global",
+	    WITNESS_RANK_BACKGROUND_THREAD_GLOBAL,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	background_thread_info = (background_thread_info_t *)base_alloc(tsdn,
+	    b0get(), opt_max_background_threads *
+	    sizeof(background_thread_info_t), CACHELINE);
+	if (background_thread_info == NULL) {
+		return true;
+	}
+
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		/* Thread mutex is rank_inclusive because of thread0. */
+		if (malloc_mutex_init(&info->mtx, "background_thread",
+		    WITNESS_RANK_BACKGROUND_THREAD,
+		    malloc_mutex_address_ordered)) {
+			return true;
+		}
+		if (pthread_cond_init(&info->cond, NULL)) {
+			return true;
+		}
+		malloc_mutex_lock(tsdn, &info->mtx);
+		info->state = background_thread_stopped;
+		background_thread_info_init(tsdn, info);
+		malloc_mutex_unlock(tsdn, &info->mtx);
+	}
+#endif
+
+	return false;
+}
diff --git a/jemalloc/src/base.c b/jemalloc/src/base.c
new file mode 100644
index 0000000..f3c6166
--- /dev/null
+++ b/jemalloc/src/base.c
@@ -0,0 +1,514 @@
+#define JEMALLOC_BASE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/sz.h"
+
+/******************************************************************************/
+/* Data. */
+
+static base_t *b0;
+
+metadata_thp_mode_t opt_metadata_thp = METADATA_THP_DEFAULT;
+
+const char *metadata_thp_mode_names[] = {
+	"disabled",
+	"auto",
+	"always"
+};
+
+/******************************************************************************/
+
+static inline bool
+metadata_thp_madvise(void) {
+	return (metadata_thp_enabled() &&
+	    (init_system_thp_mode == thp_mode_default));
+}
+
+static void *
+base_map(tsdn_t *tsdn, extent_hooks_t *extent_hooks, unsigned ind, size_t size) {
+	void *addr;
+	bool zero = true;
+	bool commit = true;
+
+	/* Use huge page sizes and alignment regardless of opt_metadata_thp. */
+	assert(size == HUGEPAGE_CEILING(size));
+	size_t alignment = HUGEPAGE;
+	if (extent_hooks == &extent_hooks_default) {
+		addr = extent_alloc_mmap(NULL, size, alignment, &zero, &commit);
+	} else {
+		/* No arena context as we are creating new arenas. */
+		tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+		pre_reentrancy(tsd, NULL);
+		addr = extent_hooks->alloc(extent_hooks, NULL, size, alignment,
+		    &zero, &commit, ind);
+		post_reentrancy(tsd);
+	}
+
+	return addr;
+}
+
+static void
+base_unmap(tsdn_t *tsdn, extent_hooks_t *extent_hooks, unsigned ind, void *addr,
+    size_t size) {
+	/*
+	 * Cascade through dalloc, decommit, purge_forced, and purge_lazy,
+	 * stopping at first success.  This cascade is performed for consistency
+	 * with the cascade in extent_dalloc_wrapper() because an application's
+	 * custom hooks may not support e.g. dalloc.  This function is only ever
+	 * called as a side effect of arena destruction, so although it might
+	 * seem pointless to do anything besides dalloc here, the application
+	 * may in fact want the end state of all associated virtual memory to be
+	 * in some consistent-but-allocated state.
+	 */
+	if (extent_hooks == &extent_hooks_default) {
+		if (!extent_dalloc_mmap(addr, size)) {
+			goto label_done;
+		}
+		if (!pages_decommit(addr, size)) {
+			goto label_done;
+		}
+		if (!pages_purge_forced(addr, size)) {
+			goto label_done;
+		}
+		if (!pages_purge_lazy(addr, size)) {
+			goto label_done;
+		}
+		/* Nothing worked.  This should never happen. */
+		not_reached();
+	} else {
+		tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+		pre_reentrancy(tsd, NULL);
+		if (extent_hooks->dalloc != NULL &&
+		    !extent_hooks->dalloc(extent_hooks, addr, size, true,
+		    ind)) {
+			goto label_post_reentrancy;
+		}
+		if (extent_hooks->decommit != NULL &&
+		    !extent_hooks->decommit(extent_hooks, addr, size, 0, size,
+		    ind)) {
+			goto label_post_reentrancy;
+		}
+		if (extent_hooks->purge_forced != NULL &&
+		    !extent_hooks->purge_forced(extent_hooks, addr, size, 0,
+		    size, ind)) {
+			goto label_post_reentrancy;
+		}
+		if (extent_hooks->purge_lazy != NULL &&
+		    !extent_hooks->purge_lazy(extent_hooks, addr, size, 0, size,
+		    ind)) {
+			goto label_post_reentrancy;
+		}
+		/* Nothing worked.  That's the application's problem. */
+	label_post_reentrancy:
+		post_reentrancy(tsd);
+	}
+label_done:
+	if (metadata_thp_madvise()) {
+		/* Set NOHUGEPAGE after unmap to avoid kernel defrag. */
+		assert(((uintptr_t)addr & HUGEPAGE_MASK) == 0 &&
+		    (size & HUGEPAGE_MASK) == 0);
+		pages_nohuge(addr, size);
+	}
+}
+
+static void
+base_extent_init(size_t *extent_sn_next, extent_t *extent, void *addr,
+    size_t size) {
+	size_t sn;
+
+	sn = *extent_sn_next;
+	(*extent_sn_next)++;
+
+	extent_binit(extent, addr, size, sn);
+}
+
+static size_t
+base_get_num_blocks(base_t *base, bool with_new_block) {
+	base_block_t *b = base->blocks;
+	assert(b != NULL);
+
+	size_t n_blocks = with_new_block ? 2 : 1;
+	while (b->next != NULL) {
+		n_blocks++;
+		b = b->next;
+	}
+
+	return n_blocks;
+}
+
+static void
+base_auto_thp_switch(tsdn_t *tsdn, base_t *base) {
+	assert(opt_metadata_thp == metadata_thp_auto);
+	malloc_mutex_assert_owner(tsdn, &base->mtx);
+	if (base->auto_thp_switched) {
+		return;
+	}
+	/* Called when adding a new block. */
+	bool should_switch;
+	if (base_ind_get(base) != 0) {
+		should_switch = (base_get_num_blocks(base, true) ==
+		    BASE_AUTO_THP_THRESHOLD);
+	} else {
+		should_switch = (base_get_num_blocks(base, true) ==
+		    BASE_AUTO_THP_THRESHOLD_A0);
+	}
+	if (!should_switch) {
+		return;
+	}
+
+	base->auto_thp_switched = true;
+	assert(!config_stats || base->n_thp == 0);
+	/* Make the initial blocks THP lazily. */
+	base_block_t *block = base->blocks;
+	while (block != NULL) {
+		assert((block->size & HUGEPAGE_MASK) == 0);
+		pages_huge(block, block->size);
+		if (config_stats) {
+			base->n_thp += HUGEPAGE_CEILING(block->size -
+			    extent_bsize_get(&block->extent)) >> LG_HUGEPAGE;
+		}
+		block = block->next;
+		assert(block == NULL || (base_ind_get(base) == 0));
+	}
+}
+
+static void *
+base_extent_bump_alloc_helper(extent_t *extent, size_t *gap_size, size_t size,
+    size_t alignment) {
+	void *ret;
+
+	assert(alignment == ALIGNMENT_CEILING(alignment, QUANTUM));
+	assert(size == ALIGNMENT_CEILING(size, alignment));
+
+	*gap_size = ALIGNMENT_CEILING((uintptr_t)extent_addr_get(extent),
+	    alignment) - (uintptr_t)extent_addr_get(extent);
+	ret = (void *)((uintptr_t)extent_addr_get(extent) + *gap_size);
+	assert(extent_bsize_get(extent) >= *gap_size + size);
+	extent_binit(extent, (void *)((uintptr_t)extent_addr_get(extent) +
+	    *gap_size + size), extent_bsize_get(extent) - *gap_size - size,
+	    extent_sn_get(extent));
+	return ret;
+}
+
+static void
+base_extent_bump_alloc_post(base_t *base, extent_t *extent, size_t gap_size,
+    void *addr, size_t size) {
+	if (extent_bsize_get(extent) > 0) {
+		/*
+		 * Compute the index for the largest size class that does not
+		 * exceed extent's size.
+		 */
+		szind_t index_floor =
+		    sz_size2index(extent_bsize_get(extent) + 1) - 1;
+		extent_heap_insert(&base->avail[index_floor], extent);
+	}
+
+	if (config_stats) {
+		base->allocated += size;
+		/*
+		 * Add one PAGE to base_resident for every page boundary that is
+		 * crossed by the new allocation. Adjust n_thp similarly when
+		 * metadata_thp is enabled.
+		 */
+		base->resident += PAGE_CEILING((uintptr_t)addr + size) -
+		    PAGE_CEILING((uintptr_t)addr - gap_size);
+		assert(base->allocated <= base->resident);
+		assert(base->resident <= base->mapped);
+		if (metadata_thp_madvise() && (opt_metadata_thp ==
+		    metadata_thp_always || base->auto_thp_switched)) {
+			base->n_thp += (HUGEPAGE_CEILING((uintptr_t)addr + size)
+			    - HUGEPAGE_CEILING((uintptr_t)addr - gap_size)) >>
+			    LG_HUGEPAGE;
+			assert(base->mapped >= base->n_thp << LG_HUGEPAGE);
+		}
+	}
+}
+
+static void *
+base_extent_bump_alloc(base_t *base, extent_t *extent, size_t size,
+    size_t alignment) {
+	void *ret;
+	size_t gap_size;
+
+	ret = base_extent_bump_alloc_helper(extent, &gap_size, size, alignment);
+	base_extent_bump_alloc_post(base, extent, gap_size, ret, size);
+	return ret;
+}
+
+/*
+ * Allocate a block of virtual memory that is large enough to start with a
+ * base_block_t header, followed by an object of specified size and alignment.
+ * On success a pointer to the initialized base_block_t header is returned.
+ */
+static base_block_t *
+base_block_alloc(tsdn_t *tsdn, base_t *base, extent_hooks_t *extent_hooks,
+    unsigned ind, pszind_t *pind_last, size_t *extent_sn_next, size_t size,
+    size_t alignment) {
+	alignment = ALIGNMENT_CEILING(alignment, QUANTUM);
+	size_t usize = ALIGNMENT_CEILING(size, alignment);
+	size_t header_size = sizeof(base_block_t);
+	size_t gap_size = ALIGNMENT_CEILING(header_size, alignment) -
+	    header_size;
+	/*
+	 * Create increasingly larger blocks in order to limit the total number
+	 * of disjoint virtual memory ranges.  Choose the next size in the page
+	 * size class series (skipping size classes that are not a multiple of
+	 * HUGEPAGE), or a size large enough to satisfy the requested size and
+	 * alignment, whichever is larger.
+	 */
+	size_t min_block_size = HUGEPAGE_CEILING(sz_psz2u(header_size + gap_size
+	    + usize));
+	pszind_t pind_next = (*pind_last + 1 < sz_psz2ind(SC_LARGE_MAXCLASS)) ?
+	    *pind_last + 1 : *pind_last;
+	size_t next_block_size = HUGEPAGE_CEILING(sz_pind2sz(pind_next));
+	size_t block_size = (min_block_size > next_block_size) ? min_block_size
+	    : next_block_size;
+	base_block_t *block = (base_block_t *)base_map(tsdn, extent_hooks, ind,
+	    block_size);
+	if (block == NULL) {
+		return NULL;
+	}
+
+	if (metadata_thp_madvise()) {
+		void *addr = (void *)block;
+		assert(((uintptr_t)addr & HUGEPAGE_MASK) == 0 &&
+		    (block_size & HUGEPAGE_MASK) == 0);
+		if (opt_metadata_thp == metadata_thp_always) {
+			pages_huge(addr, block_size);
+		} else if (opt_metadata_thp == metadata_thp_auto &&
+		    base != NULL) {
+			/* base != NULL indicates this is not a new base. */
+			malloc_mutex_lock(tsdn, &base->mtx);
+			base_auto_thp_switch(tsdn, base);
+			if (base->auto_thp_switched) {
+				pages_huge(addr, block_size);
+			}
+			malloc_mutex_unlock(tsdn, &base->mtx);
+		}
+	}
+
+	*pind_last = sz_psz2ind(block_size);
+	block->size = block_size;
+	block->next = NULL;
+	assert(block_size >= header_size);
+	base_extent_init(extent_sn_next, &block->extent,
+	    (void *)((uintptr_t)block + header_size), block_size - header_size);
+	return block;
+}
+
+/*
+ * Allocate an extent that is at least as large as specified size, with
+ * specified alignment.
+ */
+static extent_t *
+base_extent_alloc(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment) {
+	malloc_mutex_assert_owner(tsdn, &base->mtx);
+
+	extent_hooks_t *extent_hooks = base_extent_hooks_get(base);
+	/*
+	 * Drop mutex during base_block_alloc(), because an extent hook will be
+	 * called.
+	 */
+	malloc_mutex_unlock(tsdn, &base->mtx);
+	base_block_t *block = base_block_alloc(tsdn, base, extent_hooks,
+	    base_ind_get(base), &base->pind_last, &base->extent_sn_next, size,
+	    alignment);
+	malloc_mutex_lock(tsdn, &base->mtx);
+	if (block == NULL) {
+		return NULL;
+	}
+	block->next = base->blocks;
+	base->blocks = block;
+	if (config_stats) {
+		base->allocated += sizeof(base_block_t);
+		base->resident += PAGE_CEILING(sizeof(base_block_t));
+		base->mapped += block->size;
+		if (metadata_thp_madvise() &&
+		    !(opt_metadata_thp == metadata_thp_auto
+		      && !base->auto_thp_switched)) {
+			assert(base->n_thp > 0);
+			base->n_thp += HUGEPAGE_CEILING(sizeof(base_block_t)) >>
+			    LG_HUGEPAGE;
+		}
+		assert(base->allocated <= base->resident);
+		assert(base->resident <= base->mapped);
+		assert(base->n_thp << LG_HUGEPAGE <= base->mapped);
+	}
+	return &block->extent;
+}
+
+base_t *
+b0get(void) {
+	return b0;
+}
+
+base_t *
+base_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	pszind_t pind_last = 0;
+	size_t extent_sn_next = 0;
+	base_block_t *block = base_block_alloc(tsdn, NULL, extent_hooks, ind,
+	    &pind_last, &extent_sn_next, sizeof(base_t), QUANTUM);
+	if (block == NULL) {
+		return NULL;
+	}
+
+	size_t gap_size;
+	size_t base_alignment = CACHELINE;
+	size_t base_size = ALIGNMENT_CEILING(sizeof(base_t), base_alignment);
+	base_t *base = (base_t *)base_extent_bump_alloc_helper(&block->extent,
+	    &gap_size, base_size, base_alignment);
+	base->ind = ind;
+	atomic_store_p(&base->extent_hooks, extent_hooks, ATOMIC_RELAXED);
+	if (malloc_mutex_init(&base->mtx, "base", WITNESS_RANK_BASE,
+	    malloc_mutex_rank_exclusive)) {
+		base_unmap(tsdn, extent_hooks, ind, block, block->size);
+		return NULL;
+	}
+	base->pind_last = pind_last;
+	base->extent_sn_next = extent_sn_next;
+	base->blocks = block;
+	base->auto_thp_switched = false;
+	for (szind_t i = 0; i < SC_NSIZES; i++) {
+		extent_heap_new(&base->avail[i]);
+	}
+	if (config_stats) {
+		base->allocated = sizeof(base_block_t);
+		base->resident = PAGE_CEILING(sizeof(base_block_t));
+		base->mapped = block->size;
+		base->n_thp = (opt_metadata_thp == metadata_thp_always) &&
+		    metadata_thp_madvise() ? HUGEPAGE_CEILING(sizeof(base_block_t))
+		    >> LG_HUGEPAGE : 0;
+		assert(base->allocated <= base->resident);
+		assert(base->resident <= base->mapped);
+		assert(base->n_thp << LG_HUGEPAGE <= base->mapped);
+	}
+	base_extent_bump_alloc_post(base, &block->extent, gap_size, base,
+	    base_size);
+
+	return base;
+}
+
+void
+base_delete(tsdn_t *tsdn, base_t *base) {
+	extent_hooks_t *extent_hooks = base_extent_hooks_get(base);
+	base_block_t *next = base->blocks;
+	do {
+		base_block_t *block = next;
+		next = block->next;
+		base_unmap(tsdn, extent_hooks, base_ind_get(base), block,
+		    block->size);
+	} while (next != NULL);
+}
+
+extent_hooks_t *
+base_extent_hooks_get(base_t *base) {
+	return (extent_hooks_t *)atomic_load_p(&base->extent_hooks,
+	    ATOMIC_ACQUIRE);
+}
+
+extent_hooks_t *
+base_extent_hooks_set(base_t *base, extent_hooks_t *extent_hooks) {
+	extent_hooks_t *old_extent_hooks = base_extent_hooks_get(base);
+	atomic_store_p(&base->extent_hooks, extent_hooks, ATOMIC_RELEASE);
+	return old_extent_hooks;
+}
+
+static void *
+base_alloc_impl(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment,
+    size_t *esn) {
+	alignment = QUANTUM_CEILING(alignment);
+	size_t usize = ALIGNMENT_CEILING(size, alignment);
+	size_t asize = usize + alignment - QUANTUM;
+
+	extent_t *extent = NULL;
+	malloc_mutex_lock(tsdn, &base->mtx);
+	for (szind_t i = sz_size2index(asize); i < SC_NSIZES; i++) {
+		extent = extent_heap_remove_first(&base->avail[i]);
+		if (extent != NULL) {
+			/* Use existing space. */
+			break;
+		}
+	}
+	if (extent == NULL) {
+		/* Try to allocate more space. */
+		extent = base_extent_alloc(tsdn, base, usize, alignment);
+	}
+	void *ret;
+	if (extent == NULL) {
+		ret = NULL;
+		goto label_return;
+	}
+
+	ret = base_extent_bump_alloc(base, extent, usize, alignment);
+	if (esn != NULL) {
+		*esn = extent_sn_get(extent);
+	}
+label_return:
+	malloc_mutex_unlock(tsdn, &base->mtx);
+	return ret;
+}
+
+/*
+ * base_alloc() returns zeroed memory, which is always demand-zeroed for the
+ * auto arenas, in order to make multi-page sparse data structures such as radix
+ * tree nodes efficient with respect to physical memory usage.  Upon success a
+ * pointer to at least size bytes with specified alignment is returned.  Note
+ * that size is rounded up to the nearest multiple of alignment to avoid false
+ * sharing.
+ */
+void *
+base_alloc(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment) {
+	return base_alloc_impl(tsdn, base, size, alignment, NULL);
+}
+
+extent_t *
+base_alloc_extent(tsdn_t *tsdn, base_t *base) {
+	size_t esn;
+	extent_t *extent = base_alloc_impl(tsdn, base, sizeof(extent_t),
+	    CACHELINE, &esn);
+	if (extent == NULL) {
+		return NULL;
+	}
+	extent_esn_set(extent, esn);
+	return extent;
+}
+
+void
+base_stats_get(tsdn_t *tsdn, base_t *base, size_t *allocated, size_t *resident,
+    size_t *mapped, size_t *n_thp) {
+	cassert(config_stats);
+
+	malloc_mutex_lock(tsdn, &base->mtx);
+	assert(base->allocated <= base->resident);
+	assert(base->resident <= base->mapped);
+	*allocated = base->allocated;
+	*resident = base->resident;
+	*mapped = base->mapped;
+	*n_thp = base->n_thp;
+	malloc_mutex_unlock(tsdn, &base->mtx);
+}
+
+void
+base_prefork(tsdn_t *tsdn, base_t *base) {
+	malloc_mutex_prefork(tsdn, &base->mtx);
+}
+
+void
+base_postfork_parent(tsdn_t *tsdn, base_t *base) {
+	malloc_mutex_postfork_parent(tsdn, &base->mtx);
+}
+
+void
+base_postfork_child(tsdn_t *tsdn, base_t *base) {
+	malloc_mutex_postfork_child(tsdn, &base->mtx);
+}
+
+bool
+base_boot(tsdn_t *tsdn) {
+	b0 = base_new(tsdn, 0, (extent_hooks_t *)&extent_hooks_default);
+	return (b0 == NULL);
+}
diff --git a/jemalloc/src/bin.c b/jemalloc/src/bin.c
new file mode 100644
index 0000000..bca6b12
--- /dev/null
+++ b/jemalloc/src/bin.c
@@ -0,0 +1,95 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/witness.h"
+
+bin_info_t bin_infos[SC_NBINS];
+
+static void
+bin_infos_init(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS],
+    bin_info_t bin_infos[SC_NBINS]) {
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		bin_info_t *bin_info = &bin_infos[i];
+		sc_t *sc = &sc_data->sc[i];
+		bin_info->reg_size = ((size_t)1U << sc->lg_base)
+		    + ((size_t)sc->ndelta << sc->lg_delta);
+		bin_info->slab_size = (sc->pgs << LG_PAGE);
+		bin_info->nregs =
+		    (uint32_t)(bin_info->slab_size / bin_info->reg_size);
+		bin_info->n_shards = bin_shard_sizes[i];
+		bitmap_info_t bitmap_info = BITMAP_INFO_INITIALIZER(
+		    bin_info->nregs);
+		bin_info->bitmap_info = bitmap_info;
+	}
+}
+
+bool
+bin_update_shard_size(unsigned bin_shard_sizes[SC_NBINS], size_t start_size,
+    size_t end_size, size_t nshards) {
+	if (nshards > BIN_SHARDS_MAX || nshards == 0) {
+		return true;
+	}
+
+	if (start_size > SC_SMALL_MAXCLASS) {
+		return false;
+	}
+	if (end_size > SC_SMALL_MAXCLASS) {
+		end_size = SC_SMALL_MAXCLASS;
+	}
+
+	/* Compute the index since this may happen before sz init. */
+	szind_t ind1 = sz_size2index_compute(start_size);
+	szind_t ind2 = sz_size2index_compute(end_size);
+	for (unsigned i = ind1; i <= ind2; i++) {
+		bin_shard_sizes[i] = (unsigned)nshards;
+	}
+
+	return false;
+}
+
+void
+bin_shard_sizes_boot(unsigned bin_shard_sizes[SC_NBINS]) {
+	/* Load the default number of shards. */
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		bin_shard_sizes[i] = N_BIN_SHARDS_DEFAULT;
+	}
+}
+
+void
+bin_boot(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS]) {
+	assert(sc_data->initialized);
+	bin_infos_init(sc_data, bin_shard_sizes, bin_infos);
+}
+
+bool
+bin_init(bin_t *bin) {
+	if (malloc_mutex_init(&bin->lock, "bin", WITNESS_RANK_BIN,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	bin->slabcur = NULL;
+	extent_heap_new(&bin->slabs_nonfull);
+	extent_list_init(&bin->slabs_full);
+	if (config_stats) {
+		memset(&bin->stats, 0, sizeof(bin_stats_t));
+	}
+	return false;
+}
+
+void
+bin_prefork(tsdn_t *tsdn, bin_t *bin) {
+	malloc_mutex_prefork(tsdn, &bin->lock);
+}
+
+void
+bin_postfork_parent(tsdn_t *tsdn, bin_t *bin) {
+	malloc_mutex_postfork_parent(tsdn, &bin->lock);
+}
+
+void
+bin_postfork_child(tsdn_t *tsdn, bin_t *bin) {
+	malloc_mutex_postfork_child(tsdn, &bin->lock);
+}
diff --git a/jemalloc/src/bitmap.c b/jemalloc/src/bitmap.c
new file mode 100644
index 0000000..468b317
--- /dev/null
+++ b/jemalloc/src/bitmap.c
@@ -0,0 +1,121 @@
+#define JEMALLOC_BITMAP_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+
+/******************************************************************************/
+
+#ifdef BITMAP_USE_TREE
+
+void
+bitmap_info_init(bitmap_info_t *binfo, size_t nbits) {
+	unsigned i;
+	size_t group_count;
+
+	assert(nbits > 0);
+	assert(nbits <= (ZU(1) << LG_BITMAP_MAXBITS));
+
+	/*
+	 * Compute the number of groups necessary to store nbits bits, and
+	 * progressively work upward through the levels until reaching a level
+	 * that requires only one group.
+	 */
+	binfo->levels[0].group_offset = 0;
+	group_count = BITMAP_BITS2GROUPS(nbits);
+	for (i = 1; group_count > 1; i++) {
+		assert(i < BITMAP_MAX_LEVELS);
+		binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+		    + group_count;
+		group_count = BITMAP_BITS2GROUPS(group_count);
+	}
+	binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+	    + group_count;
+	assert(binfo->levels[i].group_offset <= BITMAP_GROUPS_MAX);
+	binfo->nlevels = i;
+	binfo->nbits = nbits;
+}
+
+static size_t
+bitmap_info_ngroups(const bitmap_info_t *binfo) {
+	return binfo->levels[binfo->nlevels].group_offset;
+}
+
+void
+bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo, bool fill) {
+	size_t extra;
+	unsigned i;
+
+	/*
+	 * Bits are actually inverted with regard to the external bitmap
+	 * interface.
+	 */
+
+	if (fill) {
+		/* The "filled" bitmap starts out with all 0 bits. */
+		memset(bitmap, 0, bitmap_size(binfo));
+		return;
+	}
+
+	/*
+	 * The "empty" bitmap starts out with all 1 bits, except for trailing
+	 * unused bits (if any).  Note that each group uses bit 0 to correspond
+	 * to the first logical bit in the group, so extra bits are the most
+	 * significant bits of the last group.
+	 */
+	memset(bitmap, 0xffU, bitmap_size(binfo));
+	extra = (BITMAP_GROUP_NBITS - (binfo->nbits & BITMAP_GROUP_NBITS_MASK))
+	    & BITMAP_GROUP_NBITS_MASK;
+	if (extra != 0) {
+		bitmap[binfo->levels[1].group_offset - 1] >>= extra;
+	}
+	for (i = 1; i < binfo->nlevels; i++) {
+		size_t group_count = binfo->levels[i].group_offset -
+		    binfo->levels[i-1].group_offset;
+		extra = (BITMAP_GROUP_NBITS - (group_count &
+		    BITMAP_GROUP_NBITS_MASK)) & BITMAP_GROUP_NBITS_MASK;
+		if (extra != 0) {
+			bitmap[binfo->levels[i+1].group_offset - 1] >>= extra;
+		}
+	}
+}
+
+#else /* BITMAP_USE_TREE */
+
+void
+bitmap_info_init(bitmap_info_t *binfo, size_t nbits) {
+	assert(nbits > 0);
+	assert(nbits <= (ZU(1) << LG_BITMAP_MAXBITS));
+
+	binfo->ngroups = BITMAP_BITS2GROUPS(nbits);
+	binfo->nbits = nbits;
+}
+
+static size_t
+bitmap_info_ngroups(const bitmap_info_t *binfo) {
+	return binfo->ngroups;
+}
+
+void
+bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo, bool fill) {
+	size_t extra;
+
+	if (fill) {
+		memset(bitmap, 0, bitmap_size(binfo));
+		return;
+	}
+
+	memset(bitmap, 0xffU, bitmap_size(binfo));
+	extra = (BITMAP_GROUP_NBITS - (binfo->nbits & BITMAP_GROUP_NBITS_MASK))
+	    & BITMAP_GROUP_NBITS_MASK;
+	if (extra != 0) {
+		bitmap[binfo->ngroups - 1] >>= extra;
+	}
+}
+
+#endif /* BITMAP_USE_TREE */
+
+size_t
+bitmap_size(const bitmap_info_t *binfo) {
+	return (bitmap_info_ngroups(binfo) << LG_SIZEOF_BITMAP);
+}
diff --git a/jemalloc/src/ckh.c b/jemalloc/src/ckh.c
new file mode 100644
index 0000000..1bf6df5
--- /dev/null
+++ b/jemalloc/src/ckh.c
@@ -0,0 +1,570 @@
+/*
+ *******************************************************************************
+ * Implementation of (2^1+,2) cuckoo hashing, where 2^1+ indicates that each
+ * hash bucket contains 2^n cells, for n >= 1, and 2 indicates that two hash
+ * functions are employed.  The original cuckoo hashing algorithm was described
+ * in:
+ *
+ *   Pagh, R., F.F. Rodler (2004) Cuckoo Hashing.  Journal of Algorithms
+ *     51(2):122-144.
+ *
+ * Generalization of cuckoo hashing was discussed in:
+ *
+ *   Erlingsson, U., M. Manasse, F. McSherry (2006) A cool and practical
+ *     alternative to traditional hash tables.  In Proceedings of the 7th
+ *     Workshop on Distributed Data and Structures (WDAS'06), Santa Clara, CA,
+ *     January 2006.
+ *
+ * This implementation uses precisely two hash functions because that is the
+ * fewest that can work, and supporting multiple hashes is an implementation
+ * burden.  Here is a reproduction of Figure 1 from Erlingsson et al. (2006)
+ * that shows approximate expected maximum load factors for various
+ * configurations:
+ *
+ *           |         #cells/bucket         |
+ *   #hashes |   1   |   2   |   4   |   8   |
+ *   --------+-------+-------+-------+-------+
+ *         1 | 0.006 | 0.006 | 0.03  | 0.12  |
+ *         2 | 0.49  | 0.86  |>0.93< |>0.96< |
+ *         3 | 0.91  | 0.97  | 0.98  | 0.999 |
+ *         4 | 0.97  | 0.99  | 0.999 |       |
+ *
+ * The number of cells per bucket is chosen such that a bucket fits in one cache
+ * line.  So, on 32- and 64-bit systems, we use (8,2) and (4,2) cuckoo hashing,
+ * respectively.
+ *
+ ******************************************************************************/
+#define JEMALLOC_CKH_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/ckh.h"
+
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static bool	ckh_grow(tsd_t *tsd, ckh_t *ckh);
+static void	ckh_shrink(tsd_t *tsd, ckh_t *ckh);
+
+/******************************************************************************/
+
+/*
+ * Search bucket for key and return the cell number if found; SIZE_T_MAX
+ * otherwise.
+ */
+static size_t
+ckh_bucket_search(ckh_t *ckh, size_t bucket, const void *key) {
+	ckhc_t *cell;
+	unsigned i;
+
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		if (cell->key != NULL && ckh->keycomp(key, cell->key)) {
+			return (bucket << LG_CKH_BUCKET_CELLS) + i;
+		}
+	}
+
+	return SIZE_T_MAX;
+}
+
+/*
+ * Search table for key and return cell number if found; SIZE_T_MAX otherwise.
+ */
+static size_t
+ckh_isearch(ckh_t *ckh, const void *key) {
+	size_t hashes[2], bucket, cell;
+
+	assert(ckh != NULL);
+
+	ckh->hash(key, hashes);
+
+	/* Search primary bucket. */
+	bucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	if (cell != SIZE_T_MAX) {
+		return cell;
+	}
+
+	/* Search secondary bucket. */
+	bucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	return cell;
+}
+
+static bool
+ckh_try_bucket_insert(ckh_t *ckh, size_t bucket, const void *key,
+    const void *data) {
+	ckhc_t *cell;
+	unsigned offset, i;
+
+	/*
+	 * Cycle through the cells in the bucket, starting at a random position.
+	 * The randomness avoids worst-case search overhead as buckets fill up.
+	 */
+	offset = (unsigned)prng_lg_range_u64(&ckh->prng_state,
+	    LG_CKH_BUCKET_CELLS);
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) +
+		    ((i + offset) & ((ZU(1) << LG_CKH_BUCKET_CELLS) - 1))];
+		if (cell->key == NULL) {
+			cell->key = key;
+			cell->data = data;
+			ckh->count++;
+			return false;
+		}
+	}
+
+	return true;
+}
+
+/*
+ * No space is available in bucket.  Randomly evict an item, then try to find an
+ * alternate location for that item.  Iteratively repeat this
+ * eviction/relocation procedure until either success or detection of an
+ * eviction/relocation bucket cycle.
+ */
+static bool
+ckh_evict_reloc_insert(ckh_t *ckh, size_t argbucket, void const **argkey,
+    void const **argdata) {
+	const void *key, *data, *tkey, *tdata;
+	ckhc_t *cell;
+	size_t hashes[2], bucket, tbucket;
+	unsigned i;
+
+	bucket = argbucket;
+	key = *argkey;
+	data = *argdata;
+	while (true) {
+		/*
+		 * Choose a random item within the bucket to evict.  This is
+		 * critical to correct function, because without (eventually)
+		 * evicting all items within a bucket during iteration, it
+		 * would be possible to get stuck in an infinite loop if there
+		 * were an item for which both hashes indicated the same
+		 * bucket.
+		 */
+		i = (unsigned)prng_lg_range_u64(&ckh->prng_state,
+		    LG_CKH_BUCKET_CELLS);
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		assert(cell->key != NULL);
+
+		/* Swap cell->{key,data} and {key,data} (evict). */
+		tkey = cell->key; tdata = cell->data;
+		cell->key = key; cell->data = data;
+		key = tkey; data = tdata;
+
+#ifdef CKH_COUNT
+		ckh->nrelocs++;
+#endif
+
+		/* Find the alternate bucket for the evicted item. */
+		ckh->hash(key, hashes);
+		tbucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+		if (tbucket == bucket) {
+			tbucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets)
+			    - 1);
+			/*
+			 * It may be that (tbucket == bucket) still, if the
+			 * item's hashes both indicate this bucket.  However,
+			 * we are guaranteed to eventually escape this bucket
+			 * during iteration, assuming pseudo-random item
+			 * selection (true randomness would make infinite
+			 * looping a remote possibility).  The reason we can
+			 * never get trapped forever is that there are two
+			 * cases:
+			 *
+			 * 1) This bucket == argbucket, so we will quickly
+			 *    detect an eviction cycle and terminate.
+			 * 2) An item was evicted to this bucket from another,
+			 *    which means that at least one item in this bucket
+			 *    has hashes that indicate distinct buckets.
+			 */
+		}
+		/* Check for a cycle. */
+		if (tbucket == argbucket) {
+			*argkey = key;
+			*argdata = data;
+			return true;
+		}
+
+		bucket = tbucket;
+		if (!ckh_try_bucket_insert(ckh, bucket, key, data)) {
+			return false;
+		}
+	}
+}
+
+static bool
+ckh_try_insert(ckh_t *ckh, void const**argkey, void const**argdata) {
+	size_t hashes[2], bucket;
+	const void *key = *argkey;
+	const void *data = *argdata;
+
+	ckh->hash(key, hashes);
+
+	/* Try to insert in primary bucket. */
+	bucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (!ckh_try_bucket_insert(ckh, bucket, key, data)) {
+		return false;
+	}
+
+	/* Try to insert in secondary bucket. */
+	bucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (!ckh_try_bucket_insert(ckh, bucket, key, data)) {
+		return false;
+	}
+
+	/*
+	 * Try to find a place for this item via iterative eviction/relocation.
+	 */
+	return ckh_evict_reloc_insert(ckh, bucket, argkey, argdata);
+}
+
+/*
+ * Try to rebuild the hash table from scratch by inserting all items from the
+ * old table into the new.
+ */
+static bool
+ckh_rebuild(ckh_t *ckh, ckhc_t *aTab) {
+	size_t count, i, nins;
+	const void *key, *data;
+
+	count = ckh->count;
+	ckh->count = 0;
+	for (i = nins = 0; nins < count; i++) {
+		if (aTab[i].key != NULL) {
+			key = aTab[i].key;
+			data = aTab[i].data;
+			if (ckh_try_insert(ckh, &key, &data)) {
+				ckh->count = count;
+				return true;
+			}
+			nins++;
+		}
+	}
+
+	return false;
+}
+
+static bool
+ckh_grow(tsd_t *tsd, ckh_t *ckh) {
+	bool ret;
+	ckhc_t *tab, *ttab;
+	unsigned lg_prevbuckets, lg_curcells;
+
+#ifdef CKH_COUNT
+	ckh->ngrows++;
+#endif
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table will have to be doubled more than once in order to create a
+	 * usable table.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS;
+	while (true) {
+		size_t usize;
+
+		lg_curcells++;
+		usize = sz_sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+		if (unlikely(usize == 0
+		    || usize > SC_LARGE_MAXCLASS)) {
+			ret = true;
+			goto label_return;
+		}
+		tab = (ckhc_t *)ipallocztm(tsd_tsdn(tsd), usize, CACHELINE,
+		    true, NULL, true, arena_ichoose(tsd, NULL));
+		if (tab == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		/* Swap in new table. */
+		ttab = ckh->tab;
+		ckh->tab = tab;
+		tab = ttab;
+		ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+		if (!ckh_rebuild(ckh, tab)) {
+			idalloctm(tsd_tsdn(tsd), tab, NULL, NULL, true, true);
+			break;
+		}
+
+		/* Rebuilding failed, so back out partially rebuilt table. */
+		idalloctm(tsd_tsdn(tsd), ckh->tab, NULL, NULL, true, true);
+		ckh->tab = tab;
+		ckh->lg_curbuckets = lg_prevbuckets;
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+static void
+ckh_shrink(tsd_t *tsd, ckh_t *ckh) {
+	ckhc_t *tab, *ttab;
+	size_t usize;
+	unsigned lg_prevbuckets, lg_curcells;
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table rebuild will fail.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS - 1;
+	usize = sz_sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		return;
+	}
+	tab = (ckhc_t *)ipallocztm(tsd_tsdn(tsd), usize, CACHELINE, true, NULL,
+	    true, arena_ichoose(tsd, NULL));
+	if (tab == NULL) {
+		/*
+		 * An OOM error isn't worth propagating, since it doesn't
+		 * prevent this or future operations from proceeding.
+		 */
+		return;
+	}
+	/* Swap in new table. */
+	ttab = ckh->tab;
+	ckh->tab = tab;
+	tab = ttab;
+	ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+	if (!ckh_rebuild(ckh, tab)) {
+		idalloctm(tsd_tsdn(tsd), tab, NULL, NULL, true, true);
+#ifdef CKH_COUNT
+		ckh->nshrinks++;
+#endif
+		return;
+	}
+
+	/* Rebuilding failed, so back out partially rebuilt table. */
+	idalloctm(tsd_tsdn(tsd), ckh->tab, NULL, NULL, true, true);
+	ckh->tab = tab;
+	ckh->lg_curbuckets = lg_prevbuckets;
+#ifdef CKH_COUNT
+	ckh->nshrinkfails++;
+#endif
+}
+
+bool
+ckh_new(tsd_t *tsd, ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
+    ckh_keycomp_t *keycomp) {
+	bool ret;
+	size_t mincells, usize;
+	unsigned lg_mincells;
+
+	assert(minitems > 0);
+	assert(hash != NULL);
+	assert(keycomp != NULL);
+
+#ifdef CKH_COUNT
+	ckh->ngrows = 0;
+	ckh->nshrinks = 0;
+	ckh->nshrinkfails = 0;
+	ckh->ninserts = 0;
+	ckh->nrelocs = 0;
+#endif
+	ckh->prng_state = 42; /* Value doesn't really matter. */
+	ckh->count = 0;
+
+	/*
+	 * Find the minimum power of 2 that is large enough to fit minitems
+	 * entries.  We are using (2+,2) cuckoo hashing, which has an expected
+	 * maximum load factor of at least ~0.86, so 0.75 is a conservative load
+	 * factor that will typically allow mincells items to fit without ever
+	 * growing the table.
+	 */
+	assert(LG_CKH_BUCKET_CELLS > 0);
+	mincells = ((minitems + (3 - (minitems % 3))) / 3) << 2;
+	for (lg_mincells = LG_CKH_BUCKET_CELLS;
+	    (ZU(1) << lg_mincells) < mincells;
+	    lg_mincells++) {
+		/* Do nothing. */
+	}
+	ckh->lg_minbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->lg_curbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->hash = hash;
+	ckh->keycomp = keycomp;
+
+	usize = sz_sa2u(sizeof(ckhc_t) << lg_mincells, CACHELINE);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		ret = true;
+		goto label_return;
+	}
+	ckh->tab = (ckhc_t *)ipallocztm(tsd_tsdn(tsd), usize, CACHELINE, true,
+	    NULL, true, arena_ichoose(tsd, NULL));
+	if (ckh->tab == NULL) {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+void
+ckh_delete(tsd_t *tsd, ckh_t *ckh) {
+	assert(ckh != NULL);
+
+#ifdef CKH_VERBOSE
+	malloc_printf(
+	    "%s(%p): ngrows: %"FMTu64", nshrinks: %"FMTu64","
+	    " nshrinkfails: %"FMTu64", ninserts: %"FMTu64","
+	    " nrelocs: %"FMTu64"\n", __func__, ckh,
+	    (unsigned long long)ckh->ngrows,
+	    (unsigned long long)ckh->nshrinks,
+	    (unsigned long long)ckh->nshrinkfails,
+	    (unsigned long long)ckh->ninserts,
+	    (unsigned long long)ckh->nrelocs);
+#endif
+
+	idalloctm(tsd_tsdn(tsd), ckh->tab, NULL, NULL, true, true);
+	if (config_debug) {
+		memset(ckh, JEMALLOC_FREE_JUNK, sizeof(ckh_t));
+	}
+}
+
+size_t
+ckh_count(ckh_t *ckh) {
+	assert(ckh != NULL);
+
+	return ckh->count;
+}
+
+bool
+ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data) {
+	size_t i, ncells;
+
+	for (i = *tabind, ncells = (ZU(1) << (ckh->lg_curbuckets +
+	    LG_CKH_BUCKET_CELLS)); i < ncells; i++) {
+		if (ckh->tab[i].key != NULL) {
+			if (key != NULL) {
+				*key = (void *)ckh->tab[i].key;
+			}
+			if (data != NULL) {
+				*data = (void *)ckh->tab[i].data;
+			}
+			*tabind = i + 1;
+			return false;
+		}
+	}
+
+	return true;
+}
+
+bool
+ckh_insert(tsd_t *tsd, ckh_t *ckh, const void *key, const void *data) {
+	bool ret;
+
+	assert(ckh != NULL);
+	assert(ckh_search(ckh, key, NULL, NULL));
+
+#ifdef CKH_COUNT
+	ckh->ninserts++;
+#endif
+
+	while (ckh_try_insert(ckh, &key, &data)) {
+		if (ckh_grow(tsd, ckh)) {
+			ret = true;
+			goto label_return;
+		}
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+bool
+ckh_remove(tsd_t *tsd, ckh_t *ckh, const void *searchkey, void **key,
+    void **data) {
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL) {
+			*key = (void *)ckh->tab[cell].key;
+		}
+		if (data != NULL) {
+			*data = (void *)ckh->tab[cell].data;
+		}
+		ckh->tab[cell].key = NULL;
+		ckh->tab[cell].data = NULL; /* Not necessary. */
+
+		ckh->count--;
+		/* Try to halve the table if it is less than 1/4 full. */
+		if (ckh->count < (ZU(1) << (ckh->lg_curbuckets
+		    + LG_CKH_BUCKET_CELLS - 2)) && ckh->lg_curbuckets
+		    > ckh->lg_minbuckets) {
+			/* Ignore error due to OOM. */
+			ckh_shrink(tsd, ckh);
+		}
+
+		return false;
+	}
+
+	return true;
+}
+
+bool
+ckh_search(ckh_t *ckh, const void *searchkey, void **key, void **data) {
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL) {
+			*key = (void *)ckh->tab[cell].key;
+		}
+		if (data != NULL) {
+			*data = (void *)ckh->tab[cell].data;
+		}
+		return false;
+	}
+
+	return true;
+}
+
+void
+ckh_string_hash(const void *key, size_t r_hash[2]) {
+	hash(key, strlen((const char *)key), 0x94122f33U, r_hash);
+}
+
+bool
+ckh_string_keycomp(const void *k1, const void *k2) {
+	assert(k1 != NULL);
+	assert(k2 != NULL);
+
+	return !strcmp((char *)k1, (char *)k2);
+}
+
+void
+ckh_pointer_hash(const void *key, size_t r_hash[2]) {
+	union {
+		const void	*v;
+		size_t		i;
+	} u;
+
+	assert(sizeof(u.v) == sizeof(u.i));
+	u.v = key;
+	hash(&u.i, sizeof(u.i), 0xd983396eU, r_hash);
+}
+
+bool
+ckh_pointer_keycomp(const void *k1, const void *k2) {
+	return (k1 == k2);
+}
diff --git a/jemalloc/src/ctl.c b/jemalloc/src/ctl.c
new file mode 100644
index 0000000..37d4610
--- /dev/null
+++ b/jemalloc/src/ctl.c
@@ -0,0 +1,3436 @@
+#define JEMALLOC_CTL_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+/* Data. */
+
+/*
+ * ctl_mtx protects the following:
+ * - ctl_stats->*
+ */
+static malloc_mutex_t	ctl_mtx;
+static bool		ctl_initialized;
+static ctl_stats_t	*ctl_stats;
+static ctl_arenas_t	*ctl_arenas;
+
+/******************************************************************************/
+/* Helpers for named and indexed nodes. */
+
+static const ctl_named_node_t *
+ctl_named_node(const ctl_node_t *node) {
+	return ((node->named) ? (const ctl_named_node_t *)node : NULL);
+}
+
+static const ctl_named_node_t *
+ctl_named_children(const ctl_named_node_t *node, size_t index) {
+	const ctl_named_node_t *children = ctl_named_node(node->children);
+
+	return (children ? &children[index] : NULL);
+}
+
+static const ctl_indexed_node_t *
+ctl_indexed_node(const ctl_node_t *node) {
+	return (!node->named ? (const ctl_indexed_node_t *)node : NULL);
+}
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+#define CTL_PROTO(n)							\
+static int	n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,	\
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen);
+
+#define INDEX_PROTO(n)							\
+static const ctl_named_node_t	*n##_index(tsdn_t *tsdn,		\
+    const size_t *mib, size_t miblen, size_t i);
+
+CTL_PROTO(version)
+CTL_PROTO(epoch)
+CTL_PROTO(background_thread)
+CTL_PROTO(max_background_threads)
+CTL_PROTO(thread_tcache_enabled)
+CTL_PROTO(thread_tcache_flush)
+CTL_PROTO(thread_prof_name)
+CTL_PROTO(thread_prof_active)
+CTL_PROTO(thread_arena)
+CTL_PROTO(thread_allocated)
+CTL_PROTO(thread_allocatedp)
+CTL_PROTO(thread_deallocated)
+CTL_PROTO(thread_deallocatedp)
+CTL_PROTO(config_cache_oblivious)
+CTL_PROTO(config_debug)
+CTL_PROTO(config_fill)
+CTL_PROTO(config_lazy_lock)
+CTL_PROTO(config_malloc_conf)
+CTL_PROTO(config_opt_safety_checks)
+CTL_PROTO(config_prof)
+CTL_PROTO(config_prof_libgcc)
+CTL_PROTO(config_prof_libunwind)
+CTL_PROTO(config_stats)
+CTL_PROTO(config_utrace)
+CTL_PROTO(config_xmalloc)
+CTL_PROTO(opt_abort)
+CTL_PROTO(opt_abort_conf)
+CTL_PROTO(opt_confirm_conf)
+CTL_PROTO(opt_metadata_thp)
+CTL_PROTO(opt_retain)
+CTL_PROTO(opt_dss)
+CTL_PROTO(opt_narenas)
+CTL_PROTO(opt_percpu_arena)
+CTL_PROTO(opt_oversize_threshold)
+CTL_PROTO(opt_background_thread)
+CTL_PROTO(opt_max_background_threads)
+CTL_PROTO(opt_dirty_decay_ms)
+CTL_PROTO(opt_muzzy_decay_ms)
+CTL_PROTO(opt_stats_print)
+CTL_PROTO(opt_stats_print_opts)
+CTL_PROTO(opt_junk)
+CTL_PROTO(opt_zero)
+CTL_PROTO(opt_utrace)
+CTL_PROTO(opt_xmalloc)
+CTL_PROTO(opt_tcache)
+CTL_PROTO(opt_thp)
+CTL_PROTO(opt_lg_extent_max_active_fit)
+CTL_PROTO(opt_lg_tcache_max)
+CTL_PROTO(opt_prof)
+CTL_PROTO(opt_prof_prefix)
+CTL_PROTO(opt_prof_active)
+CTL_PROTO(opt_prof_thread_active_init)
+CTL_PROTO(opt_lg_prof_sample)
+CTL_PROTO(opt_lg_prof_interval)
+CTL_PROTO(opt_prof_gdump)
+CTL_PROTO(opt_prof_final)
+CTL_PROTO(opt_prof_leak)
+CTL_PROTO(opt_prof_accum)
+CTL_PROTO(tcache_create)
+CTL_PROTO(tcache_flush)
+CTL_PROTO(tcache_destroy)
+CTL_PROTO(arena_i_initialized)
+CTL_PROTO(arena_i_decay)
+CTL_PROTO(arena_i_purge)
+CTL_PROTO(arena_i_reset)
+CTL_PROTO(arena_i_destroy)
+CTL_PROTO(arena_i_dss)
+CTL_PROTO(arena_i_dirty_decay_ms)
+CTL_PROTO(arena_i_muzzy_decay_ms)
+CTL_PROTO(arena_i_extent_hooks)
+CTL_PROTO(arena_i_retain_grow_limit)
+INDEX_PROTO(arena_i)
+CTL_PROTO(arenas_bin_i_size)
+CTL_PROTO(arenas_bin_i_nregs)
+CTL_PROTO(arenas_bin_i_slab_size)
+CTL_PROTO(arenas_bin_i_nshards)
+INDEX_PROTO(arenas_bin_i)
+CTL_PROTO(arenas_lextent_i_size)
+INDEX_PROTO(arenas_lextent_i)
+CTL_PROTO(arenas_narenas)
+CTL_PROTO(arenas_dirty_decay_ms)
+CTL_PROTO(arenas_muzzy_decay_ms)
+CTL_PROTO(arenas_quantum)
+CTL_PROTO(arenas_page)
+CTL_PROTO(arenas_tcache_max)
+CTL_PROTO(arenas_nbins)
+CTL_PROTO(arenas_nhbins)
+CTL_PROTO(arenas_nlextents)
+CTL_PROTO(arenas_create)
+CTL_PROTO(arenas_lookup)
+CTL_PROTO(prof_thread_active_init)
+CTL_PROTO(prof_active)
+CTL_PROTO(prof_dump)
+CTL_PROTO(prof_gdump)
+CTL_PROTO(prof_reset)
+CTL_PROTO(prof_interval)
+CTL_PROTO(lg_prof_sample)
+CTL_PROTO(prof_log_start)
+CTL_PROTO(prof_log_stop)
+CTL_PROTO(stats_arenas_i_small_allocated)
+CTL_PROTO(stats_arenas_i_small_nmalloc)
+CTL_PROTO(stats_arenas_i_small_ndalloc)
+CTL_PROTO(stats_arenas_i_small_nrequests)
+CTL_PROTO(stats_arenas_i_small_nfills)
+CTL_PROTO(stats_arenas_i_small_nflushes)
+CTL_PROTO(stats_arenas_i_large_allocated)
+CTL_PROTO(stats_arenas_i_large_nmalloc)
+CTL_PROTO(stats_arenas_i_large_ndalloc)
+CTL_PROTO(stats_arenas_i_large_nrequests)
+CTL_PROTO(stats_arenas_i_large_nfills)
+CTL_PROTO(stats_arenas_i_large_nflushes)
+CTL_PROTO(stats_arenas_i_bins_j_nmalloc)
+CTL_PROTO(stats_arenas_i_bins_j_ndalloc)
+CTL_PROTO(stats_arenas_i_bins_j_nrequests)
+CTL_PROTO(stats_arenas_i_bins_j_curregs)
+CTL_PROTO(stats_arenas_i_bins_j_nfills)
+CTL_PROTO(stats_arenas_i_bins_j_nflushes)
+CTL_PROTO(stats_arenas_i_bins_j_nslabs)
+CTL_PROTO(stats_arenas_i_bins_j_nreslabs)
+CTL_PROTO(stats_arenas_i_bins_j_curslabs)
+CTL_PROTO(stats_arenas_i_bins_j_nonfull_slabs)
+INDEX_PROTO(stats_arenas_i_bins_j)
+CTL_PROTO(stats_arenas_i_lextents_j_nmalloc)
+CTL_PROTO(stats_arenas_i_lextents_j_ndalloc)
+CTL_PROTO(stats_arenas_i_lextents_j_nrequests)
+CTL_PROTO(stats_arenas_i_lextents_j_curlextents)
+INDEX_PROTO(stats_arenas_i_lextents_j)
+CTL_PROTO(stats_arenas_i_extents_j_ndirty)
+CTL_PROTO(stats_arenas_i_extents_j_nmuzzy)
+CTL_PROTO(stats_arenas_i_extents_j_nretained)
+CTL_PROTO(stats_arenas_i_extents_j_dirty_bytes)
+CTL_PROTO(stats_arenas_i_extents_j_muzzy_bytes)
+CTL_PROTO(stats_arenas_i_extents_j_retained_bytes)
+INDEX_PROTO(stats_arenas_i_extents_j)
+CTL_PROTO(stats_arenas_i_nthreads)
+CTL_PROTO(stats_arenas_i_uptime)
+CTL_PROTO(stats_arenas_i_dss)
+CTL_PROTO(stats_arenas_i_dirty_decay_ms)
+CTL_PROTO(stats_arenas_i_muzzy_decay_ms)
+CTL_PROTO(stats_arenas_i_pactive)
+CTL_PROTO(stats_arenas_i_pdirty)
+CTL_PROTO(stats_arenas_i_pmuzzy)
+CTL_PROTO(stats_arenas_i_mapped)
+CTL_PROTO(stats_arenas_i_retained)
+CTL_PROTO(stats_arenas_i_extent_avail)
+CTL_PROTO(stats_arenas_i_dirty_npurge)
+CTL_PROTO(stats_arenas_i_dirty_nmadvise)
+CTL_PROTO(stats_arenas_i_dirty_purged)
+CTL_PROTO(stats_arenas_i_muzzy_npurge)
+CTL_PROTO(stats_arenas_i_muzzy_nmadvise)
+CTL_PROTO(stats_arenas_i_muzzy_purged)
+CTL_PROTO(stats_arenas_i_base)
+CTL_PROTO(stats_arenas_i_internal)
+CTL_PROTO(stats_arenas_i_metadata_thp)
+CTL_PROTO(stats_arenas_i_tcache_bytes)
+CTL_PROTO(stats_arenas_i_resident)
+CTL_PROTO(stats_arenas_i_abandoned_vm)
+INDEX_PROTO(stats_arenas_i)
+CTL_PROTO(stats_allocated)
+CTL_PROTO(stats_active)
+CTL_PROTO(stats_background_thread_num_threads)
+CTL_PROTO(stats_background_thread_num_runs)
+CTL_PROTO(stats_background_thread_run_interval)
+CTL_PROTO(stats_metadata)
+CTL_PROTO(stats_metadata_thp)
+CTL_PROTO(stats_resident)
+CTL_PROTO(stats_mapped)
+CTL_PROTO(stats_retained)
+CTL_PROTO(experimental_hooks_install)
+CTL_PROTO(experimental_hooks_remove)
+CTL_PROTO(experimental_utilization_query)
+CTL_PROTO(experimental_utilization_batch_query)
+CTL_PROTO(experimental_arenas_i_pactivep)
+INDEX_PROTO(experimental_arenas_i)
+
+#define MUTEX_STATS_CTL_PROTO_GEN(n)					\
+CTL_PROTO(stats_##n##_num_ops)						\
+CTL_PROTO(stats_##n##_num_wait)						\
+CTL_PROTO(stats_##n##_num_spin_acq)					\
+CTL_PROTO(stats_##n##_num_owner_switch)					\
+CTL_PROTO(stats_##n##_total_wait_time)					\
+CTL_PROTO(stats_##n##_max_wait_time)					\
+CTL_PROTO(stats_##n##_max_num_thds)
+
+/* Global mutexes. */
+#define OP(mtx) MUTEX_STATS_CTL_PROTO_GEN(mutexes_##mtx)
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+
+/* Per arena mutexes. */
+#define OP(mtx) MUTEX_STATS_CTL_PROTO_GEN(arenas_i_mutexes_##mtx)
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+
+/* Arena bin mutexes. */
+MUTEX_STATS_CTL_PROTO_GEN(arenas_i_bins_j_mutex)
+#undef MUTEX_STATS_CTL_PROTO_GEN
+
+CTL_PROTO(stats_mutexes_reset)
+
+/******************************************************************************/
+/* mallctl tree. */
+
+#define NAME(n)	{true},	n
+#define CHILD(t, c)							\
+	sizeof(c##_node) / sizeof(ctl_##t##_node_t),			\
+	(ctl_node_t *)c##_node,						\
+	NULL
+#define CTL(c)	0, NULL, c##_ctl
+
+/*
+ * Only handles internal indexed nodes, since there are currently no external
+ * ones.
+ */
+#define INDEX(i)	{false},	i##_index
+
+static const ctl_named_node_t	thread_tcache_node[] = {
+	{NAME("enabled"),	CTL(thread_tcache_enabled)},
+	{NAME("flush"),		CTL(thread_tcache_flush)}
+};
+
+static const ctl_named_node_t	thread_prof_node[] = {
+	{NAME("name"),		CTL(thread_prof_name)},
+	{NAME("active"),	CTL(thread_prof_active)}
+};
+
+static const ctl_named_node_t	thread_node[] = {
+	{NAME("arena"),		CTL(thread_arena)},
+	{NAME("allocated"),	CTL(thread_allocated)},
+	{NAME("allocatedp"),	CTL(thread_allocatedp)},
+	{NAME("deallocated"),	CTL(thread_deallocated)},
+	{NAME("deallocatedp"),	CTL(thread_deallocatedp)},
+	{NAME("tcache"),	CHILD(named, thread_tcache)},
+	{NAME("prof"),		CHILD(named, thread_prof)}
+};
+
+static const ctl_named_node_t	config_node[] = {
+	{NAME("cache_oblivious"), CTL(config_cache_oblivious)},
+	{NAME("debug"),		CTL(config_debug)},
+	{NAME("fill"),		CTL(config_fill)},
+	{NAME("lazy_lock"),	CTL(config_lazy_lock)},
+	{NAME("malloc_conf"),	CTL(config_malloc_conf)},
+	{NAME("opt_safety_checks"),	CTL(config_opt_safety_checks)},
+	{NAME("prof"),		CTL(config_prof)},
+	{NAME("prof_libgcc"),	CTL(config_prof_libgcc)},
+	{NAME("prof_libunwind"), CTL(config_prof_libunwind)},
+	{NAME("stats"),		CTL(config_stats)},
+	{NAME("utrace"),	CTL(config_utrace)},
+	{NAME("xmalloc"),	CTL(config_xmalloc)}
+};
+
+static const ctl_named_node_t opt_node[] = {
+	{NAME("abort"),		CTL(opt_abort)},
+	{NAME("abort_conf"),	CTL(opt_abort_conf)},
+	{NAME("confirm_conf"),	CTL(opt_confirm_conf)},
+	{NAME("metadata_thp"),	CTL(opt_metadata_thp)},
+	{NAME("retain"),	CTL(opt_retain)},
+	{NAME("dss"),		CTL(opt_dss)},
+	{NAME("narenas"),	CTL(opt_narenas)},
+	{NAME("percpu_arena"),	CTL(opt_percpu_arena)},
+	{NAME("oversize_threshold"),	CTL(opt_oversize_threshold)},
+	{NAME("background_thread"),	CTL(opt_background_thread)},
+	{NAME("max_background_threads"),	CTL(opt_max_background_threads)},
+	{NAME("dirty_decay_ms"), CTL(opt_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(opt_muzzy_decay_ms)},
+	{NAME("stats_print"),	CTL(opt_stats_print)},
+	{NAME("stats_print_opts"),	CTL(opt_stats_print_opts)},
+	{NAME("junk"),		CTL(opt_junk)},
+	{NAME("zero"),		CTL(opt_zero)},
+	{NAME("utrace"),	CTL(opt_utrace)},
+	{NAME("xmalloc"),	CTL(opt_xmalloc)},
+	{NAME("tcache"),	CTL(opt_tcache)},
+	{NAME("thp"),		CTL(opt_thp)},
+	{NAME("lg_extent_max_active_fit"), CTL(opt_lg_extent_max_active_fit)},
+	{NAME("lg_tcache_max"),	CTL(opt_lg_tcache_max)},
+	{NAME("prof"),		CTL(opt_prof)},
+	{NAME("prof_prefix"),	CTL(opt_prof_prefix)},
+	{NAME("prof_active"),	CTL(opt_prof_active)},
+	{NAME("prof_thread_active_init"), CTL(opt_prof_thread_active_init)},
+	{NAME("lg_prof_sample"), CTL(opt_lg_prof_sample)},
+	{NAME("lg_prof_interval"), CTL(opt_lg_prof_interval)},
+	{NAME("prof_gdump"),	CTL(opt_prof_gdump)},
+	{NAME("prof_final"),	CTL(opt_prof_final)},
+	{NAME("prof_leak"),	CTL(opt_prof_leak)},
+	{NAME("prof_accum"),	CTL(opt_prof_accum)}
+};
+
+static const ctl_named_node_t	tcache_node[] = {
+	{NAME("create"),	CTL(tcache_create)},
+	{NAME("flush"),		CTL(tcache_flush)},
+	{NAME("destroy"),	CTL(tcache_destroy)}
+};
+
+static const ctl_named_node_t arena_i_node[] = {
+	{NAME("initialized"),	CTL(arena_i_initialized)},
+	{NAME("decay"),		CTL(arena_i_decay)},
+	{NAME("purge"),		CTL(arena_i_purge)},
+	{NAME("reset"),		CTL(arena_i_reset)},
+	{NAME("destroy"),	CTL(arena_i_destroy)},
+	{NAME("dss"),		CTL(arena_i_dss)},
+	{NAME("dirty_decay_ms"), CTL(arena_i_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(arena_i_muzzy_decay_ms)},
+	{NAME("extent_hooks"),	CTL(arena_i_extent_hooks)},
+	{NAME("retain_grow_limit"),	CTL(arena_i_retain_grow_limit)}
+};
+static const ctl_named_node_t super_arena_i_node[] = {
+	{NAME(""),		CHILD(named, arena_i)}
+};
+
+static const ctl_indexed_node_t arena_node[] = {
+	{INDEX(arena_i)}
+};
+
+static const ctl_named_node_t arenas_bin_i_node[] = {
+	{NAME("size"),		CTL(arenas_bin_i_size)},
+	{NAME("nregs"),		CTL(arenas_bin_i_nregs)},
+	{NAME("slab_size"),	CTL(arenas_bin_i_slab_size)},
+	{NAME("nshards"),	CTL(arenas_bin_i_nshards)}
+};
+static const ctl_named_node_t super_arenas_bin_i_node[] = {
+	{NAME(""),		CHILD(named, arenas_bin_i)}
+};
+
+static const ctl_indexed_node_t arenas_bin_node[] = {
+	{INDEX(arenas_bin_i)}
+};
+
+static const ctl_named_node_t arenas_lextent_i_node[] = {
+	{NAME("size"),		CTL(arenas_lextent_i_size)}
+};
+static const ctl_named_node_t super_arenas_lextent_i_node[] = {
+	{NAME(""),		CHILD(named, arenas_lextent_i)}
+};
+
+static const ctl_indexed_node_t arenas_lextent_node[] = {
+	{INDEX(arenas_lextent_i)}
+};
+
+static const ctl_named_node_t arenas_node[] = {
+	{NAME("narenas"),	CTL(arenas_narenas)},
+	{NAME("dirty_decay_ms"), CTL(arenas_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(arenas_muzzy_decay_ms)},
+	{NAME("quantum"),	CTL(arenas_quantum)},
+	{NAME("page"),		CTL(arenas_page)},
+	{NAME("tcache_max"),	CTL(arenas_tcache_max)},
+	{NAME("nbins"),		CTL(arenas_nbins)},
+	{NAME("nhbins"),	CTL(arenas_nhbins)},
+	{NAME("bin"),		CHILD(indexed, arenas_bin)},
+	{NAME("nlextents"),	CTL(arenas_nlextents)},
+	{NAME("lextent"),	CHILD(indexed, arenas_lextent)},
+	{NAME("create"),	CTL(arenas_create)},
+	{NAME("lookup"),	CTL(arenas_lookup)}
+};
+
+static const ctl_named_node_t	prof_node[] = {
+	{NAME("thread_active_init"), CTL(prof_thread_active_init)},
+	{NAME("active"),	CTL(prof_active)},
+	{NAME("dump"),		CTL(prof_dump)},
+	{NAME("gdump"),		CTL(prof_gdump)},
+	{NAME("reset"),		CTL(prof_reset)},
+	{NAME("interval"),	CTL(prof_interval)},
+	{NAME("lg_sample"),	CTL(lg_prof_sample)},
+	{NAME("log_start"),	CTL(prof_log_start)},
+	{NAME("log_stop"),	CTL(prof_log_stop)}
+};
+static const ctl_named_node_t stats_arenas_i_small_node[] = {
+	{NAME("allocated"),	CTL(stats_arenas_i_small_allocated)},
+	{NAME("nmalloc"),	CTL(stats_arenas_i_small_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_small_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_small_nrequests)},
+	{NAME("nfills"),	CTL(stats_arenas_i_small_nfills)},
+	{NAME("nflushes"),	CTL(stats_arenas_i_small_nflushes)}
+};
+
+static const ctl_named_node_t stats_arenas_i_large_node[] = {
+	{NAME("allocated"),	CTL(stats_arenas_i_large_allocated)},
+	{NAME("nmalloc"),	CTL(stats_arenas_i_large_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_large_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_large_nrequests)},
+	{NAME("nfills"),	CTL(stats_arenas_i_large_nfills)},
+	{NAME("nflushes"),	CTL(stats_arenas_i_large_nflushes)}
+};
+
+#define MUTEX_PROF_DATA_NODE(prefix)					\
+static const ctl_named_node_t stats_##prefix##_node[] = {		\
+	{NAME("num_ops"),						\
+	 CTL(stats_##prefix##_num_ops)},				\
+	{NAME("num_wait"),						\
+	 CTL(stats_##prefix##_num_wait)},				\
+	{NAME("num_spin_acq"),						\
+	 CTL(stats_##prefix##_num_spin_acq)},				\
+	{NAME("num_owner_switch"),					\
+	 CTL(stats_##prefix##_num_owner_switch)},			\
+	{NAME("total_wait_time"),					\
+	 CTL(stats_##prefix##_total_wait_time)},			\
+	{NAME("max_wait_time"),						\
+	 CTL(stats_##prefix##_max_wait_time)},				\
+	{NAME("max_num_thds"),						\
+	 CTL(stats_##prefix##_max_num_thds)}				\
+	/* Note that # of current waiting thread not provided. */	\
+};
+
+MUTEX_PROF_DATA_NODE(arenas_i_bins_j_mutex)
+
+static const ctl_named_node_t stats_arenas_i_bins_j_node[] = {
+	{NAME("nmalloc"),	CTL(stats_arenas_i_bins_j_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_bins_j_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_bins_j_nrequests)},
+	{NAME("curregs"),	CTL(stats_arenas_i_bins_j_curregs)},
+	{NAME("nfills"),	CTL(stats_arenas_i_bins_j_nfills)},
+	{NAME("nflushes"),	CTL(stats_arenas_i_bins_j_nflushes)},
+	{NAME("nslabs"),	CTL(stats_arenas_i_bins_j_nslabs)},
+	{NAME("nreslabs"),	CTL(stats_arenas_i_bins_j_nreslabs)},
+	{NAME("curslabs"),	CTL(stats_arenas_i_bins_j_curslabs)},
+	{NAME("nonfull_slabs"),	CTL(stats_arenas_i_bins_j_nonfull_slabs)},
+	{NAME("mutex"),		CHILD(named, stats_arenas_i_bins_j_mutex)}
+};
+
+static const ctl_named_node_t super_stats_arenas_i_bins_j_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i_bins_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_bins_node[] = {
+	{INDEX(stats_arenas_i_bins_j)}
+};
+
+static const ctl_named_node_t stats_arenas_i_lextents_j_node[] = {
+	{NAME("nmalloc"),	CTL(stats_arenas_i_lextents_j_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_lextents_j_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_lextents_j_nrequests)},
+	{NAME("curlextents"),	CTL(stats_arenas_i_lextents_j_curlextents)}
+};
+static const ctl_named_node_t super_stats_arenas_i_lextents_j_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i_lextents_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_lextents_node[] = {
+	{INDEX(stats_arenas_i_lextents_j)}
+};
+
+static const ctl_named_node_t stats_arenas_i_extents_j_node[] = {
+	{NAME("ndirty"),	CTL(stats_arenas_i_extents_j_ndirty)},
+	{NAME("nmuzzy"),	CTL(stats_arenas_i_extents_j_nmuzzy)},
+	{NAME("nretained"),	CTL(stats_arenas_i_extents_j_nretained)},
+	{NAME("dirty_bytes"),	CTL(stats_arenas_i_extents_j_dirty_bytes)},
+	{NAME("muzzy_bytes"),	CTL(stats_arenas_i_extents_j_muzzy_bytes)},
+	{NAME("retained_bytes"), CTL(stats_arenas_i_extents_j_retained_bytes)}
+};
+
+static const ctl_named_node_t super_stats_arenas_i_extents_j_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i_extents_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_extents_node[] = {
+	{INDEX(stats_arenas_i_extents_j)}
+};
+
+#define OP(mtx)  MUTEX_PROF_DATA_NODE(arenas_i_mutexes_##mtx)
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+
+static const ctl_named_node_t stats_arenas_i_mutexes_node[] = {
+#define OP(mtx) {NAME(#mtx), CHILD(named, stats_arenas_i_mutexes_##mtx)},
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+};
+
+static const ctl_named_node_t stats_arenas_i_node[] = {
+	{NAME("nthreads"),	CTL(stats_arenas_i_nthreads)},
+	{NAME("uptime"),	CTL(stats_arenas_i_uptime)},
+	{NAME("dss"),		CTL(stats_arenas_i_dss)},
+	{NAME("dirty_decay_ms"), CTL(stats_arenas_i_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(stats_arenas_i_muzzy_decay_ms)},
+	{NAME("pactive"),	CTL(stats_arenas_i_pactive)},
+	{NAME("pdirty"),	CTL(stats_arenas_i_pdirty)},
+	{NAME("pmuzzy"),	CTL(stats_arenas_i_pmuzzy)},
+	{NAME("mapped"),	CTL(stats_arenas_i_mapped)},
+	{NAME("retained"),	CTL(stats_arenas_i_retained)},
+	{NAME("extent_avail"),	CTL(stats_arenas_i_extent_avail)},
+	{NAME("dirty_npurge"),	CTL(stats_arenas_i_dirty_npurge)},
+	{NAME("dirty_nmadvise"), CTL(stats_arenas_i_dirty_nmadvise)},
+	{NAME("dirty_purged"),	CTL(stats_arenas_i_dirty_purged)},
+	{NAME("muzzy_npurge"),	CTL(stats_arenas_i_muzzy_npurge)},
+	{NAME("muzzy_nmadvise"), CTL(stats_arenas_i_muzzy_nmadvise)},
+	{NAME("muzzy_purged"),	CTL(stats_arenas_i_muzzy_purged)},
+	{NAME("base"),		CTL(stats_arenas_i_base)},
+	{NAME("internal"),	CTL(stats_arenas_i_internal)},
+	{NAME("metadata_thp"),	CTL(stats_arenas_i_metadata_thp)},
+	{NAME("tcache_bytes"),	CTL(stats_arenas_i_tcache_bytes)},
+	{NAME("resident"),	CTL(stats_arenas_i_resident)},
+	{NAME("abandoned_vm"),	CTL(stats_arenas_i_abandoned_vm)},
+	{NAME("small"),		CHILD(named, stats_arenas_i_small)},
+	{NAME("large"),		CHILD(named, stats_arenas_i_large)},
+	{NAME("bins"),		CHILD(indexed, stats_arenas_i_bins)},
+	{NAME("lextents"),	CHILD(indexed, stats_arenas_i_lextents)},
+	{NAME("extents"),	CHILD(indexed, stats_arenas_i_extents)},
+	{NAME("mutexes"),	CHILD(named, stats_arenas_i_mutexes)}
+};
+static const ctl_named_node_t super_stats_arenas_i_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i)}
+};
+
+static const ctl_indexed_node_t stats_arenas_node[] = {
+	{INDEX(stats_arenas_i)}
+};
+
+static const ctl_named_node_t stats_background_thread_node[] = {
+	{NAME("num_threads"),	CTL(stats_background_thread_num_threads)},
+	{NAME("num_runs"),	CTL(stats_background_thread_num_runs)},
+	{NAME("run_interval"),	CTL(stats_background_thread_run_interval)}
+};
+
+#define OP(mtx) MUTEX_PROF_DATA_NODE(mutexes_##mtx)
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+
+static const ctl_named_node_t stats_mutexes_node[] = {
+#define OP(mtx) {NAME(#mtx), CHILD(named, stats_mutexes_##mtx)},
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+	{NAME("reset"),		CTL(stats_mutexes_reset)}
+};
+#undef MUTEX_PROF_DATA_NODE
+
+static const ctl_named_node_t stats_node[] = {
+	{NAME("allocated"),	CTL(stats_allocated)},
+	{NAME("active"),	CTL(stats_active)},
+	{NAME("metadata"),	CTL(stats_metadata)},
+	{NAME("metadata_thp"),	CTL(stats_metadata_thp)},
+	{NAME("resident"),	CTL(stats_resident)},
+	{NAME("mapped"),	CTL(stats_mapped)},
+	{NAME("retained"),	CTL(stats_retained)},
+	{NAME("background_thread"),
+	 CHILD(named, stats_background_thread)},
+	{NAME("mutexes"),	CHILD(named, stats_mutexes)},
+	{NAME("arenas"),	CHILD(indexed, stats_arenas)}
+};
+
+static const ctl_named_node_t experimental_hooks_node[] = {
+	{NAME("install"),	CTL(experimental_hooks_install)},
+	{NAME("remove"),	CTL(experimental_hooks_remove)}
+};
+
+static const ctl_named_node_t experimental_utilization_node[] = {
+	{NAME("query"),		CTL(experimental_utilization_query)},
+	{NAME("batch_query"),	CTL(experimental_utilization_batch_query)}
+};
+
+static const ctl_named_node_t experimental_arenas_i_node[] = {
+	{NAME("pactivep"),	CTL(experimental_arenas_i_pactivep)}
+};
+static const ctl_named_node_t super_experimental_arenas_i_node[] = {
+	{NAME(""),		CHILD(named, experimental_arenas_i)}
+};
+
+static const ctl_indexed_node_t experimental_arenas_node[] = {
+	{INDEX(experimental_arenas_i)}
+};
+
+static const ctl_named_node_t experimental_node[] = {
+	{NAME("hooks"),		CHILD(named, experimental_hooks)},
+	{NAME("utilization"),	CHILD(named, experimental_utilization)},
+	{NAME("arenas"),	CHILD(indexed, experimental_arenas)}
+};
+
+static const ctl_named_node_t	root_node[] = {
+	{NAME("version"),	CTL(version)},
+	{NAME("epoch"),		CTL(epoch)},
+	{NAME("background_thread"),	CTL(background_thread)},
+	{NAME("max_background_threads"),	CTL(max_background_threads)},
+	{NAME("thread"),	CHILD(named, thread)},
+	{NAME("config"),	CHILD(named, config)},
+	{NAME("opt"),		CHILD(named, opt)},
+	{NAME("tcache"),	CHILD(named, tcache)},
+	{NAME("arena"),		CHILD(indexed, arena)},
+	{NAME("arenas"),	CHILD(named, arenas)},
+	{NAME("prof"),		CHILD(named, prof)},
+	{NAME("stats"),		CHILD(named, stats)},
+	{NAME("experimental"),	CHILD(named, experimental)}
+};
+static const ctl_named_node_t super_root_node[] = {
+	{NAME(""),		CHILD(named, root)}
+};
+
+#undef NAME
+#undef CHILD
+#undef CTL
+#undef INDEX
+
+/******************************************************************************/
+
+/*
+ * Sets *dst + *src non-atomically.  This is safe, since everything is
+ * synchronized by the ctl mutex.
+ */
+static void
+ctl_accum_arena_stats_u64(arena_stats_u64_t *dst, arena_stats_u64_t *src) {
+#ifdef JEMALLOC_ATOMIC_U64
+	uint64_t cur_dst = atomic_load_u64(dst, ATOMIC_RELAXED);
+	uint64_t cur_src = atomic_load_u64(src, ATOMIC_RELAXED);
+	atomic_store_u64(dst, cur_dst + cur_src, ATOMIC_RELAXED);
+#else
+	*dst += *src;
+#endif
+}
+
+/* Likewise: with ctl mutex synchronization, reading is simple. */
+static uint64_t
+ctl_arena_stats_read_u64(arena_stats_u64_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_u64(p, ATOMIC_RELAXED);
+#else
+	return *p;
+#endif
+}
+
+static void
+accum_atomic_zu(atomic_zu_t *dst, atomic_zu_t *src) {
+	size_t cur_dst = atomic_load_zu(dst, ATOMIC_RELAXED);
+	size_t cur_src = atomic_load_zu(src, ATOMIC_RELAXED);
+	atomic_store_zu(dst, cur_dst + cur_src, ATOMIC_RELAXED);
+}
+
+/******************************************************************************/
+
+static unsigned
+arenas_i2a_impl(size_t i, bool compat, bool validate) {
+	unsigned a;
+
+	switch (i) {
+	case MALLCTL_ARENAS_ALL:
+		a = 0;
+		break;
+	case MALLCTL_ARENAS_DESTROYED:
+		a = 1;
+		break;
+	default:
+		if (compat && i == ctl_arenas->narenas) {
+			/*
+			 * Provide deprecated backward compatibility for
+			 * accessing the merged stats at index narenas rather
+			 * than via MALLCTL_ARENAS_ALL.  This is scheduled for
+			 * removal in 6.0.0.
+			 */
+			a = 0;
+		} else if (validate && i >= ctl_arenas->narenas) {
+			a = UINT_MAX;
+		} else {
+			/*
+			 * This function should never be called for an index
+			 * more than one past the range of indices that have
+			 * initialized ctl data.
+			 */
+			assert(i < ctl_arenas->narenas || (!validate && i ==
+			    ctl_arenas->narenas));
+			a = (unsigned)i + 2;
+		}
+		break;
+	}
+
+	return a;
+}
+
+static unsigned
+arenas_i2a(size_t i) {
+	return arenas_i2a_impl(i, true, false);
+}
+
+static ctl_arena_t *
+arenas_i_impl(tsd_t *tsd, size_t i, bool compat, bool init) {
+	ctl_arena_t *ret;
+
+	assert(!compat || !init);
+
+	ret = ctl_arenas->arenas[arenas_i2a_impl(i, compat, false)];
+	if (init && ret == NULL) {
+		if (config_stats) {
+			struct container_s {
+				ctl_arena_t		ctl_arena;
+				ctl_arena_stats_t	astats;
+			};
+			struct container_s *cont =
+			    (struct container_s *)base_alloc(tsd_tsdn(tsd),
+			    b0get(), sizeof(struct container_s), QUANTUM);
+			if (cont == NULL) {
+				return NULL;
+			}
+			ret = &cont->ctl_arena;
+			ret->astats = &cont->astats;
+		} else {
+			ret = (ctl_arena_t *)base_alloc(tsd_tsdn(tsd), b0get(),
+			    sizeof(ctl_arena_t), QUANTUM);
+			if (ret == NULL) {
+				return NULL;
+			}
+		}
+		ret->arena_ind = (unsigned)i;
+		ctl_arenas->arenas[arenas_i2a_impl(i, compat, false)] = ret;
+	}
+
+	assert(ret == NULL || arenas_i2a(ret->arena_ind) == arenas_i2a(i));
+	return ret;
+}
+
+static ctl_arena_t *
+arenas_i(size_t i) {
+	ctl_arena_t *ret = arenas_i_impl(tsd_fetch(), i, true, false);
+	assert(ret != NULL);
+	return ret;
+}
+
+static void
+ctl_arena_clear(ctl_arena_t *ctl_arena) {
+	ctl_arena->nthreads = 0;
+	ctl_arena->dss = dss_prec_names[dss_prec_limit];
+	ctl_arena->dirty_decay_ms = -1;
+	ctl_arena->muzzy_decay_ms = -1;
+	ctl_arena->pactive = 0;
+	ctl_arena->pdirty = 0;
+	ctl_arena->pmuzzy = 0;
+	if (config_stats) {
+		memset(&ctl_arena->astats->astats, 0, sizeof(arena_stats_t));
+		ctl_arena->astats->allocated_small = 0;
+		ctl_arena->astats->nmalloc_small = 0;
+		ctl_arena->astats->ndalloc_small = 0;
+		ctl_arena->astats->nrequests_small = 0;
+		ctl_arena->astats->nfills_small = 0;
+		ctl_arena->astats->nflushes_small = 0;
+		memset(ctl_arena->astats->bstats, 0, SC_NBINS *
+		    sizeof(bin_stats_t));
+		memset(ctl_arena->astats->lstats, 0, (SC_NSIZES - SC_NBINS) *
+		    sizeof(arena_stats_large_t));
+		memset(ctl_arena->astats->estats, 0, SC_NPSIZES *
+		    sizeof(arena_stats_extents_t));
+	}
+}
+
+static void
+ctl_arena_stats_amerge(tsdn_t *tsdn, ctl_arena_t *ctl_arena, arena_t *arena) {
+	unsigned i;
+
+	if (config_stats) {
+		arena_stats_merge(tsdn, arena, &ctl_arena->nthreads,
+		    &ctl_arena->dss, &ctl_arena->dirty_decay_ms,
+		    &ctl_arena->muzzy_decay_ms, &ctl_arena->pactive,
+		    &ctl_arena->pdirty, &ctl_arena->pmuzzy,
+		    &ctl_arena->astats->astats, ctl_arena->astats->bstats,
+		    ctl_arena->astats->lstats, ctl_arena->astats->estats);
+
+		for (i = 0; i < SC_NBINS; i++) {
+			ctl_arena->astats->allocated_small +=
+			    ctl_arena->astats->bstats[i].curregs *
+			    sz_index2size(i);
+			ctl_arena->astats->nmalloc_small +=
+			    ctl_arena->astats->bstats[i].nmalloc;
+			ctl_arena->astats->ndalloc_small +=
+			    ctl_arena->astats->bstats[i].ndalloc;
+			ctl_arena->astats->nrequests_small +=
+			    ctl_arena->astats->bstats[i].nrequests;
+			ctl_arena->astats->nfills_small +=
+			    ctl_arena->astats->bstats[i].nfills;
+			ctl_arena->astats->nflushes_small +=
+			    ctl_arena->astats->bstats[i].nflushes;
+		}
+	} else {
+		arena_basic_stats_merge(tsdn, arena, &ctl_arena->nthreads,
+		    &ctl_arena->dss, &ctl_arena->dirty_decay_ms,
+		    &ctl_arena->muzzy_decay_ms, &ctl_arena->pactive,
+		    &ctl_arena->pdirty, &ctl_arena->pmuzzy);
+	}
+}
+
+static void
+ctl_arena_stats_sdmerge(ctl_arena_t *ctl_sdarena, ctl_arena_t *ctl_arena,
+    bool destroyed) {
+	unsigned i;
+
+	if (!destroyed) {
+		ctl_sdarena->nthreads += ctl_arena->nthreads;
+		ctl_sdarena->pactive += ctl_arena->pactive;
+		ctl_sdarena->pdirty += ctl_arena->pdirty;
+		ctl_sdarena->pmuzzy += ctl_arena->pmuzzy;
+	} else {
+		assert(ctl_arena->nthreads == 0);
+		assert(ctl_arena->pactive == 0);
+		assert(ctl_arena->pdirty == 0);
+		assert(ctl_arena->pmuzzy == 0);
+	}
+
+	if (config_stats) {
+		ctl_arena_stats_t *sdstats = ctl_sdarena->astats;
+		ctl_arena_stats_t *astats = ctl_arena->astats;
+
+		if (!destroyed) {
+			accum_atomic_zu(&sdstats->astats.mapped,
+			    &astats->astats.mapped);
+			accum_atomic_zu(&sdstats->astats.retained,
+			    &astats->astats.retained);
+			accum_atomic_zu(&sdstats->astats.extent_avail,
+			    &astats->astats.extent_avail);
+		}
+
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_dirty.npurge,
+		    &astats->astats.decay_dirty.npurge);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_dirty.nmadvise,
+		    &astats->astats.decay_dirty.nmadvise);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_dirty.purged,
+		    &astats->astats.decay_dirty.purged);
+
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_muzzy.npurge,
+		    &astats->astats.decay_muzzy.npurge);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_muzzy.nmadvise,
+		    &astats->astats.decay_muzzy.nmadvise);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_muzzy.purged,
+		    &astats->astats.decay_muzzy.purged);
+
+#define OP(mtx) malloc_mutex_prof_merge(				\
+		    &(sdstats->astats.mutex_prof_data[			\
+		        arena_prof_mutex_##mtx]),			\
+		    &(astats->astats.mutex_prof_data[			\
+		        arena_prof_mutex_##mtx]));
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+		if (!destroyed) {
+			accum_atomic_zu(&sdstats->astats.base,
+			    &astats->astats.base);
+			accum_atomic_zu(&sdstats->astats.internal,
+			    &astats->astats.internal);
+			accum_atomic_zu(&sdstats->astats.resident,
+			    &astats->astats.resident);
+			accum_atomic_zu(&sdstats->astats.metadata_thp,
+			    &astats->astats.metadata_thp);
+		} else {
+			assert(atomic_load_zu(
+			    &astats->astats.internal, ATOMIC_RELAXED) == 0);
+		}
+
+		if (!destroyed) {
+			sdstats->allocated_small += astats->allocated_small;
+		} else {
+			assert(astats->allocated_small == 0);
+		}
+		sdstats->nmalloc_small += astats->nmalloc_small;
+		sdstats->ndalloc_small += astats->ndalloc_small;
+		sdstats->nrequests_small += astats->nrequests_small;
+		sdstats->nfills_small += astats->nfills_small;
+		sdstats->nflushes_small += astats->nflushes_small;
+
+		if (!destroyed) {
+			accum_atomic_zu(&sdstats->astats.allocated_large,
+			    &astats->astats.allocated_large);
+		} else {
+			assert(atomic_load_zu(&astats->astats.allocated_large,
+			    ATOMIC_RELAXED) == 0);
+		}
+		ctl_accum_arena_stats_u64(&sdstats->astats.nmalloc_large,
+		    &astats->astats.nmalloc_large);
+		ctl_accum_arena_stats_u64(&sdstats->astats.ndalloc_large,
+		    &astats->astats.ndalloc_large);
+		ctl_accum_arena_stats_u64(&sdstats->astats.nrequests_large,
+		    &astats->astats.nrequests_large);
+		accum_atomic_zu(&sdstats->astats.abandoned_vm,
+		    &astats->astats.abandoned_vm);
+
+		accum_atomic_zu(&sdstats->astats.tcache_bytes,
+		    &astats->astats.tcache_bytes);
+
+		if (ctl_arena->arena_ind == 0) {
+			sdstats->astats.uptime = astats->astats.uptime;
+		}
+
+		/* Merge bin stats. */
+		for (i = 0; i < SC_NBINS; i++) {
+			sdstats->bstats[i].nmalloc += astats->bstats[i].nmalloc;
+			sdstats->bstats[i].ndalloc += astats->bstats[i].ndalloc;
+			sdstats->bstats[i].nrequests +=
+			    astats->bstats[i].nrequests;
+			if (!destroyed) {
+				sdstats->bstats[i].curregs +=
+				    astats->bstats[i].curregs;
+			} else {
+				assert(astats->bstats[i].curregs == 0);
+			}
+			sdstats->bstats[i].nfills += astats->bstats[i].nfills;
+			sdstats->bstats[i].nflushes +=
+			    astats->bstats[i].nflushes;
+			sdstats->bstats[i].nslabs += astats->bstats[i].nslabs;
+			sdstats->bstats[i].reslabs += astats->bstats[i].reslabs;
+			if (!destroyed) {
+				sdstats->bstats[i].curslabs +=
+				    astats->bstats[i].curslabs;
+				sdstats->bstats[i].nonfull_slabs +=
+				    astats->bstats[i].nonfull_slabs;
+			} else {
+				assert(astats->bstats[i].curslabs == 0);
+				assert(astats->bstats[i].nonfull_slabs == 0);
+			}
+			malloc_mutex_prof_merge(&sdstats->bstats[i].mutex_data,
+			    &astats->bstats[i].mutex_data);
+		}
+
+		/* Merge stats for large allocations. */
+		for (i = 0; i < SC_NSIZES - SC_NBINS; i++) {
+			ctl_accum_arena_stats_u64(&sdstats->lstats[i].nmalloc,
+			    &astats->lstats[i].nmalloc);
+			ctl_accum_arena_stats_u64(&sdstats->lstats[i].ndalloc,
+			    &astats->lstats[i].ndalloc);
+			ctl_accum_arena_stats_u64(&sdstats->lstats[i].nrequests,
+			    &astats->lstats[i].nrequests);
+			if (!destroyed) {
+				sdstats->lstats[i].curlextents +=
+				    astats->lstats[i].curlextents;
+			} else {
+				assert(astats->lstats[i].curlextents == 0);
+			}
+		}
+
+		/* Merge extents stats. */
+		for (i = 0; i < SC_NPSIZES; i++) {
+			accum_atomic_zu(&sdstats->estats[i].ndirty,
+			    &astats->estats[i].ndirty);
+			accum_atomic_zu(&sdstats->estats[i].nmuzzy,
+			    &astats->estats[i].nmuzzy);
+			accum_atomic_zu(&sdstats->estats[i].nretained,
+			    &astats->estats[i].nretained);
+			accum_atomic_zu(&sdstats->estats[i].dirty_bytes,
+			    &astats->estats[i].dirty_bytes);
+			accum_atomic_zu(&sdstats->estats[i].muzzy_bytes,
+			    &astats->estats[i].muzzy_bytes);
+			accum_atomic_zu(&sdstats->estats[i].retained_bytes,
+			    &astats->estats[i].retained_bytes);
+		}
+	}
+}
+
+static void
+ctl_arena_refresh(tsdn_t *tsdn, arena_t *arena, ctl_arena_t *ctl_sdarena,
+    unsigned i, bool destroyed) {
+	ctl_arena_t *ctl_arena = arenas_i(i);
+
+	ctl_arena_clear(ctl_arena);
+	ctl_arena_stats_amerge(tsdn, ctl_arena, arena);
+	/* Merge into sum stats as well. */
+	ctl_arena_stats_sdmerge(ctl_sdarena, ctl_arena, destroyed);
+}
+
+static unsigned
+ctl_arena_init(tsd_t *tsd, extent_hooks_t *extent_hooks) {
+	unsigned arena_ind;
+	ctl_arena_t *ctl_arena;
+
+	if ((ctl_arena = ql_last(&ctl_arenas->destroyed, destroyed_link)) !=
+	    NULL) {
+		ql_remove(&ctl_arenas->destroyed, ctl_arena, destroyed_link);
+		arena_ind = ctl_arena->arena_ind;
+	} else {
+		arena_ind = ctl_arenas->narenas;
+	}
+
+	/* Trigger stats allocation. */
+	if (arenas_i_impl(tsd, arena_ind, false, true) == NULL) {
+		return UINT_MAX;
+	}
+
+	/* Initialize new arena. */
+	if (arena_init(tsd_tsdn(tsd), arena_ind, extent_hooks) == NULL) {
+		return UINT_MAX;
+	}
+
+	if (arena_ind == ctl_arenas->narenas) {
+		ctl_arenas->narenas++;
+	}
+
+	return arena_ind;
+}
+
+static void
+ctl_background_thread_stats_read(tsdn_t *tsdn) {
+	background_thread_stats_t *stats = &ctl_stats->background_thread;
+	if (!have_background_thread ||
+	    background_thread_stats_read(tsdn, stats)) {
+		memset(stats, 0, sizeof(background_thread_stats_t));
+		nstime_init(&stats->run_interval, 0);
+	}
+}
+
+static void
+ctl_refresh(tsdn_t *tsdn) {
+	unsigned i;
+	ctl_arena_t *ctl_sarena = arenas_i(MALLCTL_ARENAS_ALL);
+	VARIABLE_ARRAY(arena_t *, tarenas, ctl_arenas->narenas);
+
+	/*
+	 * Clear sum stats, since they will be merged into by
+	 * ctl_arena_refresh().
+	 */
+	ctl_arena_clear(ctl_sarena);
+
+	for (i = 0; i < ctl_arenas->narenas; i++) {
+		tarenas[i] = arena_get(tsdn, i, false);
+	}
+
+	for (i = 0; i < ctl_arenas->narenas; i++) {
+		ctl_arena_t *ctl_arena = arenas_i(i);
+		bool initialized = (tarenas[i] != NULL);
+
+		ctl_arena->initialized = initialized;
+		if (initialized) {
+			ctl_arena_refresh(tsdn, tarenas[i], ctl_sarena, i,
+			    false);
+		}
+	}
+
+	if (config_stats) {
+		ctl_stats->allocated = ctl_sarena->astats->allocated_small +
+		    atomic_load_zu(&ctl_sarena->astats->astats.allocated_large,
+			ATOMIC_RELAXED);
+		ctl_stats->active = (ctl_sarena->pactive << LG_PAGE);
+		ctl_stats->metadata = atomic_load_zu(
+		    &ctl_sarena->astats->astats.base, ATOMIC_RELAXED) +
+		    atomic_load_zu(&ctl_sarena->astats->astats.internal,
+			ATOMIC_RELAXED);
+		ctl_stats->metadata_thp = atomic_load_zu(
+		    &ctl_sarena->astats->astats.metadata_thp, ATOMIC_RELAXED);
+		ctl_stats->resident = atomic_load_zu(
+		    &ctl_sarena->astats->astats.resident, ATOMIC_RELAXED);
+		ctl_stats->mapped = atomic_load_zu(
+		    &ctl_sarena->astats->astats.mapped, ATOMIC_RELAXED);
+		ctl_stats->retained = atomic_load_zu(
+		    &ctl_sarena->astats->astats.retained, ATOMIC_RELAXED);
+
+		ctl_background_thread_stats_read(tsdn);
+
+#define READ_GLOBAL_MUTEX_PROF_DATA(i, mtx)				\
+    malloc_mutex_lock(tsdn, &mtx);					\
+    malloc_mutex_prof_read(tsdn, &ctl_stats->mutex_prof_data[i], &mtx);	\
+    malloc_mutex_unlock(tsdn, &mtx);
+
+		if (config_prof && opt_prof) {
+			READ_GLOBAL_MUTEX_PROF_DATA(global_prof_mutex_prof,
+			    bt2gctx_mtx);
+		}
+		if (have_background_thread) {
+			READ_GLOBAL_MUTEX_PROF_DATA(
+			    global_prof_mutex_background_thread,
+			    background_thread_lock);
+		} else {
+			memset(&ctl_stats->mutex_prof_data[
+			    global_prof_mutex_background_thread], 0,
+			    sizeof(mutex_prof_data_t));
+		}
+		/* We own ctl mutex already. */
+		malloc_mutex_prof_read(tsdn,
+		    &ctl_stats->mutex_prof_data[global_prof_mutex_ctl],
+		    &ctl_mtx);
+#undef READ_GLOBAL_MUTEX_PROF_DATA
+	}
+	ctl_arenas->epoch++;
+}
+
+static bool
+ctl_init(tsd_t *tsd) {
+	bool ret;
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	if (!ctl_initialized) {
+		ctl_arena_t *ctl_sarena, *ctl_darena;
+		unsigned i;
+
+		/*
+		 * Allocate demand-zeroed space for pointers to the full
+		 * range of supported arena indices.
+		 */
+		if (ctl_arenas == NULL) {
+			ctl_arenas = (ctl_arenas_t *)base_alloc(tsdn,
+			    b0get(), sizeof(ctl_arenas_t), QUANTUM);
+			if (ctl_arenas == NULL) {
+				ret = true;
+				goto label_return;
+			}
+		}
+
+		if (config_stats && ctl_stats == NULL) {
+			ctl_stats = (ctl_stats_t *)base_alloc(tsdn, b0get(),
+			    sizeof(ctl_stats_t), QUANTUM);
+			if (ctl_stats == NULL) {
+				ret = true;
+				goto label_return;
+			}
+		}
+
+		/*
+		 * Allocate space for the current full range of arenas
+		 * here rather than doing it lazily elsewhere, in order
+		 * to limit when OOM-caused errors can occur.
+		 */
+		if ((ctl_sarena = arenas_i_impl(tsd, MALLCTL_ARENAS_ALL, false,
+		    true)) == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		ctl_sarena->initialized = true;
+
+		if ((ctl_darena = arenas_i_impl(tsd, MALLCTL_ARENAS_DESTROYED,
+		    false, true)) == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		ctl_arena_clear(ctl_darena);
+		/*
+		 * Don't toggle ctl_darena to initialized until an arena is
+		 * actually destroyed, so that arena.<i>.initialized can be used
+		 * to query whether the stats are relevant.
+		 */
+
+		ctl_arenas->narenas = narenas_total_get();
+		for (i = 0; i < ctl_arenas->narenas; i++) {
+			if (arenas_i_impl(tsd, i, false, true) == NULL) {
+				ret = true;
+				goto label_return;
+			}
+		}
+
+		ql_new(&ctl_arenas->destroyed);
+		ctl_refresh(tsdn);
+
+		ctl_initialized = true;
+	}
+
+	ret = false;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+static int
+ctl_lookup(tsdn_t *tsdn, const char *name, ctl_node_t const **nodesp,
+    size_t *mibp, size_t *depthp) {
+	int ret;
+	const char *elm, *tdot, *dot;
+	size_t elen, i, j;
+	const ctl_named_node_t *node;
+
+	elm = name;
+	/* Equivalent to strchrnul(). */
+	dot = ((tdot = strchr(elm, '.')) != NULL) ? tdot : strchr(elm, '\0');
+	elen = (size_t)((uintptr_t)dot - (uintptr_t)elm);
+	if (elen == 0) {
+		ret = ENOENT;
+		goto label_return;
+	}
+	node = super_root_node;
+	for (i = 0; i < *depthp; i++) {
+		assert(node);
+		assert(node->nchildren > 0);
+		if (ctl_named_node(node->children) != NULL) {
+			const ctl_named_node_t *pnode = node;
+
+			/* Children are named. */
+			for (j = 0; j < node->nchildren; j++) {
+				const ctl_named_node_t *child =
+				    ctl_named_children(node, j);
+				if (strlen(child->name) == elen &&
+				    strncmp(elm, child->name, elen) == 0) {
+					node = child;
+					if (nodesp != NULL) {
+						nodesp[i] =
+						    (const ctl_node_t *)node;
+					}
+					mibp[i] = j;
+					break;
+				}
+			}
+			if (node == pnode) {
+				ret = ENOENT;
+				goto label_return;
+			}
+		} else {
+			uintmax_t index;
+			const ctl_indexed_node_t *inode;
+
+			/* Children are indexed. */
+			index = malloc_strtoumax(elm, NULL, 10);
+			if (index == UINTMAX_MAX || index > SIZE_T_MAX) {
+				ret = ENOENT;
+				goto label_return;
+			}
+
+			inode = ctl_indexed_node(node->children);
+			node = inode->index(tsdn, mibp, *depthp, (size_t)index);
+			if (node == NULL) {
+				ret = ENOENT;
+				goto label_return;
+			}
+
+			if (nodesp != NULL) {
+				nodesp[i] = (const ctl_node_t *)node;
+			}
+			mibp[i] = (size_t)index;
+		}
+
+		if (node->ctl != NULL) {
+			/* Terminal node. */
+			if (*dot != '\0') {
+				/*
+				 * The name contains more elements than are
+				 * in this path through the tree.
+				 */
+				ret = ENOENT;
+				goto label_return;
+			}
+			/* Complete lookup successful. */
+			*depthp = i + 1;
+			break;
+		}
+
+		/* Update elm. */
+		if (*dot == '\0') {
+			/* No more elements. */
+			ret = ENOENT;
+			goto label_return;
+		}
+		elm = &dot[1];
+		dot = ((tdot = strchr(elm, '.')) != NULL) ? tdot :
+		    strchr(elm, '\0');
+		elen = (size_t)((uintptr_t)dot - (uintptr_t)elm);
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+int
+ctl_byname(tsd_t *tsd, const char *name, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen) {
+	int ret;
+	size_t depth;
+	ctl_node_t const *nodes[CTL_MAX_DEPTH];
+	size_t mib[CTL_MAX_DEPTH];
+	const ctl_named_node_t *node;
+
+	if (!ctl_initialized && ctl_init(tsd)) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	depth = CTL_MAX_DEPTH;
+	ret = ctl_lookup(tsd_tsdn(tsd), name, nodes, mib, &depth);
+	if (ret != 0) {
+		goto label_return;
+	}
+
+	node = ctl_named_node(nodes[depth-1]);
+	if (node != NULL && node->ctl) {
+		ret = node->ctl(tsd, mib, depth, oldp, oldlenp, newp, newlen);
+	} else {
+		/* The name refers to a partial path through the ctl tree. */
+		ret = ENOENT;
+	}
+
+label_return:
+	return(ret);
+}
+
+int
+ctl_nametomib(tsd_t *tsd, const char *name, size_t *mibp, size_t *miblenp) {
+	int ret;
+
+	if (!ctl_initialized && ctl_init(tsd)) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	ret = ctl_lookup(tsd_tsdn(tsd), name, NULL, mibp, miblenp);
+label_return:
+	return(ret);
+}
+
+int
+ctl_bymib(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	const ctl_named_node_t *node;
+	size_t i;
+
+	if (!ctl_initialized && ctl_init(tsd)) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	/* Iterate down the tree. */
+	node = super_root_node;
+	for (i = 0; i < miblen; i++) {
+		assert(node);
+		assert(node->nchildren > 0);
+		if (ctl_named_node(node->children) != NULL) {
+			/* Children are named. */
+			if (node->nchildren <= mib[i]) {
+				ret = ENOENT;
+				goto label_return;
+			}
+			node = ctl_named_children(node, mib[i]);
+		} else {
+			const ctl_indexed_node_t *inode;
+
+			/* Indexed element. */
+			inode = ctl_indexed_node(node->children);
+			node = inode->index(tsd_tsdn(tsd), mib, miblen, mib[i]);
+			if (node == NULL) {
+				ret = ENOENT;
+				goto label_return;
+			}
+		}
+	}
+
+	/* Call the ctl function. */
+	if (node && node->ctl) {
+		ret = node->ctl(tsd, mib, miblen, oldp, oldlenp, newp, newlen);
+	} else {
+		/* Partial MIB. */
+		ret = ENOENT;
+	}
+
+label_return:
+	return(ret);
+}
+
+bool
+ctl_boot(void) {
+	if (malloc_mutex_init(&ctl_mtx, "ctl", WITNESS_RANK_CTL,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	ctl_initialized = false;
+
+	return false;
+}
+
+void
+ctl_prefork(tsdn_t *tsdn) {
+	malloc_mutex_prefork(tsdn, &ctl_mtx);
+}
+
+void
+ctl_postfork_parent(tsdn_t *tsdn) {
+	malloc_mutex_postfork_parent(tsdn, &ctl_mtx);
+}
+
+void
+ctl_postfork_child(tsdn_t *tsdn) {
+	malloc_mutex_postfork_child(tsdn, &ctl_mtx);
+}
+
+/******************************************************************************/
+/* *_ctl() functions. */
+
+#define READONLY()	do {						\
+	if (newp != NULL || newlen != 0) {				\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define WRITEONLY()	do {						\
+	if (oldp != NULL || oldlenp != NULL) {				\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define READ_XOR_WRITE()	do {					\
+	if ((oldp != NULL && oldlenp != NULL) && (newp != NULL ||	\
+	    newlen != 0)) {						\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define READ(v, t)	do {						\
+	if (oldp != NULL && oldlenp != NULL) {				\
+		if (*oldlenp != sizeof(t)) {				\
+			size_t	copylen = (sizeof(t) <= *oldlenp)	\
+			    ? sizeof(t) : *oldlenp;			\
+			memcpy(oldp, (void *)&(v), copylen);		\
+			ret = EINVAL;					\
+			goto label_return;				\
+		}							\
+		*(t *)oldp = (v);					\
+	}								\
+} while (0)
+
+#define WRITE(v, t)	do {						\
+	if (newp != NULL) {						\
+		if (newlen != sizeof(t)) {				\
+			ret = EINVAL;					\
+			goto label_return;				\
+		}							\
+		(v) = *(t *)newp;					\
+	}								\
+} while (0)
+
+#define MIB_UNSIGNED(v, i) do {						\
+	if (mib[i] > UINT_MAX) {					\
+		ret = EFAULT;						\
+		goto label_return;					\
+	}								\
+	v = (unsigned)mib[i];						\
+} while (0)
+
+/*
+ * There's a lot of code duplication in the following macros due to limitations
+ * in how nested cpp macros are expanded.
+ */
+#define CTL_RO_CLGEN(c, l, n, v, t)					\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,	\
+    size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	if (l) {							\
+		malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);		\
+	}								\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	if (l) {							\
+		malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);		\
+	}								\
+	return ret;							\
+}
+
+#define CTL_RO_CGEN(c, n, v, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);			\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);			\
+	return ret;							\
+}
+
+#define CTL_RO_GEN(n, v, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,	\
+    size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);			\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);			\
+	return ret;							\
+}
+
+/*
+ * ctl_mtx is not acquired, under the assumption that no pertinent data will
+ * mutate during the call.
+ */
+#define CTL_RO_NL_CGEN(c, n, v, t)					\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+#define CTL_RO_NL_GEN(n, v, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+#define CTL_TSD_RO_NL_CGEN(c, n, m, t)					\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,	\
+    size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	READONLY();							\
+	oldval = (m(tsd));						\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+#define CTL_RO_CONFIG_GEN(n, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	READONLY();							\
+	oldval = n;							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+/******************************************************************************/
+
+CTL_RO_NL_GEN(version, JEMALLOC_VERSION, const char *)
+
+static int
+epoch_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	UNUSED uint64_t newval;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	WRITE(newval, uint64_t);
+	if (newp != NULL) {
+		ctl_refresh(tsd_tsdn(tsd));
+	}
+	READ(ctl_arenas->epoch, uint64_t);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+background_thread_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!have_background_thread) {
+		return ENOENT;
+	}
+	background_thread_ctl_init(tsd_tsdn(tsd));
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+	if (newp == NULL) {
+		oldval = background_thread_enabled();
+		READ(oldval, bool);
+	} else {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = background_thread_enabled();
+		READ(oldval, bool);
+
+		bool newval = *(bool *)newp;
+		if (newval == oldval) {
+			ret = 0;
+			goto label_return;
+		}
+
+		background_thread_enabled_set(tsd_tsdn(tsd), newval);
+		if (newval) {
+			if (background_threads_enable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		} else {
+			if (background_threads_disable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		}
+	}
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+
+	return ret;
+}
+
+static int
+max_background_threads_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	size_t oldval;
+
+	if (!have_background_thread) {
+		return ENOENT;
+	}
+	background_thread_ctl_init(tsd_tsdn(tsd));
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+	if (newp == NULL) {
+		oldval = max_background_threads;
+		READ(oldval, size_t);
+	} else {
+		if (newlen != sizeof(size_t)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = max_background_threads;
+		READ(oldval, size_t);
+
+		size_t newval = *(size_t *)newp;
+		if (newval == oldval) {
+			ret = 0;
+			goto label_return;
+		}
+		if (newval > opt_max_background_threads) {
+			ret = EINVAL;
+			goto label_return;
+		}
+
+		if (background_thread_enabled()) {
+			background_thread_enabled_set(tsd_tsdn(tsd), false);
+			if (background_threads_disable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+			max_background_threads = newval;
+			background_thread_enabled_set(tsd_tsdn(tsd), true);
+			if (background_threads_enable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		} else {
+			max_background_threads = newval;
+		}
+	}
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+
+	return ret;
+}
+
+/******************************************************************************/
+
+CTL_RO_CONFIG_GEN(config_cache_oblivious, bool)
+CTL_RO_CONFIG_GEN(config_debug, bool)
+CTL_RO_CONFIG_GEN(config_fill, bool)
+CTL_RO_CONFIG_GEN(config_lazy_lock, bool)
+CTL_RO_CONFIG_GEN(config_malloc_conf, const char *)
+CTL_RO_CONFIG_GEN(config_opt_safety_checks, bool)
+CTL_RO_CONFIG_GEN(config_prof, bool)
+CTL_RO_CONFIG_GEN(config_prof_libgcc, bool)
+CTL_RO_CONFIG_GEN(config_prof_libunwind, bool)
+CTL_RO_CONFIG_GEN(config_stats, bool)
+CTL_RO_CONFIG_GEN(config_utrace, bool)
+CTL_RO_CONFIG_GEN(config_xmalloc, bool)
+
+/******************************************************************************/
+
+CTL_RO_NL_GEN(opt_abort, opt_abort, bool)
+CTL_RO_NL_GEN(opt_abort_conf, opt_abort_conf, bool)
+CTL_RO_NL_GEN(opt_confirm_conf, opt_confirm_conf, bool)
+CTL_RO_NL_GEN(opt_metadata_thp, metadata_thp_mode_names[opt_metadata_thp],
+    const char *)
+CTL_RO_NL_GEN(opt_retain, opt_retain, bool)
+CTL_RO_NL_GEN(opt_dss, opt_dss, const char *)
+CTL_RO_NL_GEN(opt_narenas, opt_narenas, unsigned)
+CTL_RO_NL_GEN(opt_percpu_arena, percpu_arena_mode_names[opt_percpu_arena],
+    const char *)
+CTL_RO_NL_GEN(opt_oversize_threshold, opt_oversize_threshold, size_t)
+CTL_RO_NL_GEN(opt_background_thread, opt_background_thread, bool)
+CTL_RO_NL_GEN(opt_max_background_threads, opt_max_background_threads, size_t)
+CTL_RO_NL_GEN(opt_dirty_decay_ms, opt_dirty_decay_ms, ssize_t)
+CTL_RO_NL_GEN(opt_muzzy_decay_ms, opt_muzzy_decay_ms, ssize_t)
+CTL_RO_NL_GEN(opt_stats_print, opt_stats_print, bool)
+CTL_RO_NL_GEN(opt_stats_print_opts, opt_stats_print_opts, const char *)
+CTL_RO_NL_CGEN(config_fill, opt_junk, opt_junk, const char *)
+CTL_RO_NL_CGEN(config_fill, opt_zero, opt_zero, bool)
+CTL_RO_NL_CGEN(config_utrace, opt_utrace, opt_utrace, bool)
+CTL_RO_NL_CGEN(config_xmalloc, opt_xmalloc, opt_xmalloc, bool)
+CTL_RO_NL_GEN(opt_tcache, opt_tcache, bool)
+CTL_RO_NL_GEN(opt_thp, thp_mode_names[opt_thp], const char *)
+CTL_RO_NL_GEN(opt_lg_extent_max_active_fit, opt_lg_extent_max_active_fit,
+    size_t)
+CTL_RO_NL_GEN(opt_lg_tcache_max, opt_lg_tcache_max, ssize_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof, opt_prof, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_prefix, opt_prof_prefix, const char *)
+CTL_RO_NL_CGEN(config_prof, opt_prof_active, opt_prof_active, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_thread_active_init,
+    opt_prof_thread_active_init, bool)
+CTL_RO_NL_CGEN(config_prof, opt_lg_prof_sample, opt_lg_prof_sample, size_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof_accum, opt_prof_accum, bool)
+CTL_RO_NL_CGEN(config_prof, opt_lg_prof_interval, opt_lg_prof_interval, ssize_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof_gdump, opt_prof_gdump, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_final, opt_prof_final, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_leak, opt_prof_leak, bool)
+
+/******************************************************************************/
+
+static int
+thread_arena_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	arena_t *oldarena;
+	unsigned newind, oldind;
+
+	oldarena = arena_choose(tsd, NULL);
+	if (oldarena == NULL) {
+		return EAGAIN;
+	}
+	newind = oldind = arena_ind_get(oldarena);
+	WRITE(newind, unsigned);
+	READ(oldind, unsigned);
+
+	if (newind != oldind) {
+		arena_t *newarena;
+
+		if (newind >= narenas_total_get()) {
+			/* New arena index is out of range. */
+			ret = EFAULT;
+			goto label_return;
+		}
+
+		if (have_percpu_arena &&
+		    PERCPU_ARENA_ENABLED(opt_percpu_arena)) {
+			if (newind < percpu_arena_ind_limit(opt_percpu_arena)) {
+				/*
+				 * If perCPU arena is enabled, thread_arena
+				 * control is not allowed for the auto arena
+				 * range.
+				 */
+				ret = EPERM;
+				goto label_return;
+			}
+		}
+
+		/* Initialize arena if necessary. */
+		newarena = arena_get(tsd_tsdn(tsd), newind, true);
+		if (newarena == NULL) {
+			ret = EAGAIN;
+			goto label_return;
+		}
+		/* Set new arena/tcache associations. */
+		arena_migrate(tsd, oldind, newind);
+		if (tcache_available(tsd)) {
+			tcache_arena_reassociate(tsd_tsdn(tsd),
+			    tsd_tcachep_get(tsd), newarena);
+		}
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+CTL_TSD_RO_NL_CGEN(config_stats, thread_allocated, tsd_thread_allocated_get,
+    uint64_t)
+CTL_TSD_RO_NL_CGEN(config_stats, thread_allocatedp, tsd_thread_allocatedp_get,
+    uint64_t *)
+CTL_TSD_RO_NL_CGEN(config_stats, thread_deallocated, tsd_thread_deallocated_get,
+    uint64_t)
+CTL_TSD_RO_NL_CGEN(config_stats, thread_deallocatedp,
+    tsd_thread_deallocatedp_get, uint64_t *)
+
+static int
+thread_tcache_enabled_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	bool oldval;
+
+	oldval = tcache_enabled_get(tsd);
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		tcache_enabled_set(tsd, *(bool *)newp);
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+thread_tcache_flush_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+
+	if (!tcache_available(tsd)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	READONLY();
+	WRITEONLY();
+
+	tcache_flush(tsd);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+thread_prof_name_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	READ_XOR_WRITE();
+
+	if (newp != NULL) {
+		if (newlen != sizeof(const char *)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+
+		if ((ret = prof_thread_name_set(tsd, *(const char **)newp)) !=
+		    0) {
+			goto label_return;
+		}
+	} else {
+		const char *oldname = prof_thread_name_get(tsd);
+		READ(oldname, const char *);
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+thread_prof_active_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	oldval = prof_thread_active_get(tsd);
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		if (prof_thread_active_set(tsd, *(bool *)newp)) {
+			ret = EAGAIN;
+			goto label_return;
+		}
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+tcache_create_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned tcache_ind;
+
+	READONLY();
+	if (tcaches_create(tsd, &tcache_ind)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+	READ(tcache_ind, unsigned);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+tcache_flush_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned tcache_ind;
+
+	WRITEONLY();
+	tcache_ind = UINT_MAX;
+	WRITE(tcache_ind, unsigned);
+	if (tcache_ind == UINT_MAX) {
+		ret = EFAULT;
+		goto label_return;
+	}
+	tcaches_flush(tsd, tcache_ind);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+tcache_destroy_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned tcache_ind;
+
+	WRITEONLY();
+	tcache_ind = UINT_MAX;
+	WRITE(tcache_ind, unsigned);
+	if (tcache_ind == UINT_MAX) {
+		ret = EFAULT;
+		goto label_return;
+	}
+	tcaches_destroy(tsd, tcache_ind);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+arena_i_initialized_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+	unsigned arena_ind;
+	bool initialized;
+
+	READONLY();
+	MIB_UNSIGNED(arena_ind, 1);
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	initialized = arenas_i(arena_ind)->initialized;
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+
+	READ(initialized, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static void
+arena_i_decay(tsdn_t *tsdn, unsigned arena_ind, bool all) {
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	{
+		unsigned narenas = ctl_arenas->narenas;
+
+		/*
+		 * Access via index narenas is deprecated, and scheduled for
+		 * removal in 6.0.0.
+		 */
+		if (arena_ind == MALLCTL_ARENAS_ALL || arena_ind == narenas) {
+			unsigned i;
+			VARIABLE_ARRAY(arena_t *, tarenas, narenas);
+
+			for (i = 0; i < narenas; i++) {
+				tarenas[i] = arena_get(tsdn, i, false);
+			}
+
+			/*
+			 * No further need to hold ctl_mtx, since narenas and
+			 * tarenas contain everything needed below.
+			 */
+			malloc_mutex_unlock(tsdn, &ctl_mtx);
+
+			for (i = 0; i < narenas; i++) {
+				if (tarenas[i] != NULL) {
+					arena_decay(tsdn, tarenas[i], false,
+					    all);
+				}
+			}
+		} else {
+			arena_t *tarena;
+
+			assert(arena_ind < narenas);
+
+			tarena = arena_get(tsdn, arena_ind, false);
+
+			/* No further need to hold ctl_mtx. */
+			malloc_mutex_unlock(tsdn, &ctl_mtx);
+
+			if (tarena != NULL) {
+				arena_decay(tsdn, tarena, false, all);
+			}
+		}
+	}
+}
+
+static int
+arena_i_decay_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+
+	READONLY();
+	WRITEONLY();
+	MIB_UNSIGNED(arena_ind, 1);
+	arena_i_decay(tsd_tsdn(tsd), arena_ind, false);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arena_i_purge_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+
+	READONLY();
+	WRITEONLY();
+	MIB_UNSIGNED(arena_ind, 1);
+	arena_i_decay(tsd_tsdn(tsd), arena_ind, true);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arena_i_reset_destroy_helper(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen, unsigned *arena_ind,
+    arena_t **arena) {
+	int ret;
+
+	READONLY();
+	WRITEONLY();
+	MIB_UNSIGNED(*arena_ind, 1);
+
+	*arena = arena_get(tsd_tsdn(tsd), *arena_ind, false);
+	if (*arena == NULL || arena_is_auto(*arena)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static void
+arena_reset_prepare_background_thread(tsd_t *tsd, unsigned arena_ind) {
+	/* Temporarily disable the background thread during arena reset. */
+	if (have_background_thread) {
+		malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+		if (background_thread_enabled()) {
+			background_thread_info_t *info =
+			    background_thread_info_get(arena_ind);
+			assert(info->state == background_thread_started);
+			malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+			info->state = background_thread_paused;
+			malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		}
+	}
+}
+
+static void
+arena_reset_finish_background_thread(tsd_t *tsd, unsigned arena_ind) {
+	if (have_background_thread) {
+		if (background_thread_enabled()) {
+			background_thread_info_t *info =
+			    background_thread_info_get(arena_ind);
+			assert(info->state == background_thread_paused);
+			malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+			info->state = background_thread_started;
+			malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+	}
+}
+
+static int
+arena_i_reset_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	ret = arena_i_reset_destroy_helper(tsd, mib, miblen, oldp, oldlenp,
+	    newp, newlen, &arena_ind, &arena);
+	if (ret != 0) {
+		return ret;
+	}
+
+	arena_reset_prepare_background_thread(tsd, arena_ind);
+	arena_reset(tsd, arena);
+	arena_reset_finish_background_thread(tsd, arena_ind);
+
+	return ret;
+}
+
+static int
+arena_i_destroy_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+	ctl_arena_t *ctl_darena, *ctl_arena;
+
+	ret = arena_i_reset_destroy_helper(tsd, mib, miblen, oldp, oldlenp,
+	    newp, newlen, &arena_ind, &arena);
+	if (ret != 0) {
+		goto label_return;
+	}
+
+	if (arena_nthreads_get(arena, false) != 0 || arena_nthreads_get(arena,
+	    true) != 0) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	arena_reset_prepare_background_thread(tsd, arena_ind);
+	/* Merge stats after resetting and purging arena. */
+	arena_reset(tsd, arena);
+	arena_decay(tsd_tsdn(tsd), arena, false, true);
+	ctl_darena = arenas_i(MALLCTL_ARENAS_DESTROYED);
+	ctl_darena->initialized = true;
+	ctl_arena_refresh(tsd_tsdn(tsd), arena, ctl_darena, arena_ind, true);
+	/* Destroy arena. */
+	arena_destroy(tsd, arena);
+	ctl_arena = arenas_i(arena_ind);
+	ctl_arena->initialized = false;
+	/* Record arena index for later recycling via arenas.create. */
+	ql_elm_new(ctl_arena, destroyed_link);
+	ql_tail_insert(&ctl_arenas->destroyed, ctl_arena, destroyed_link);
+	arena_reset_finish_background_thread(tsd, arena_ind);
+
+	assert(ret == 0);
+label_return:
+	return ret;
+}
+
+static int
+arena_i_dss_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	const char *dss = NULL;
+	unsigned arena_ind;
+	dss_prec_t dss_prec_old = dss_prec_limit;
+	dss_prec_t dss_prec = dss_prec_limit;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	WRITE(dss, const char *);
+	MIB_UNSIGNED(arena_ind, 1);
+	if (dss != NULL) {
+		int i;
+		bool match = false;
+
+		for (i = 0; i < dss_prec_limit; i++) {
+			if (strcmp(dss_prec_names[i], dss) == 0) {
+				dss_prec = i;
+				match = true;
+				break;
+			}
+		}
+
+		if (!match) {
+			ret = EINVAL;
+			goto label_return;
+		}
+	}
+
+	/*
+	 * Access via index narenas is deprecated, and scheduled for removal in
+	 * 6.0.0.
+	 */
+	if (arena_ind == MALLCTL_ARENAS_ALL || arena_ind ==
+	    ctl_arenas->narenas) {
+		if (dss_prec != dss_prec_limit &&
+		    extent_dss_prec_set(dss_prec)) {
+			ret = EFAULT;
+			goto label_return;
+		}
+		dss_prec_old = extent_dss_prec_get();
+	} else {
+		arena_t *arena = arena_get(tsd_tsdn(tsd), arena_ind, false);
+		if (arena == NULL || (dss_prec != dss_prec_limit &&
+		    arena_dss_prec_set(arena, dss_prec))) {
+			ret = EFAULT;
+			goto label_return;
+		}
+		dss_prec_old = arena_dss_prec_get(arena);
+	}
+
+	dss = dss_prec_names[dss_prec_old];
+	READ(dss, const char *);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arena_i_decay_ms_ctl_impl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen, bool dirty) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	MIB_UNSIGNED(arena_ind, 1);
+	arena = arena_get(tsd_tsdn(tsd), arena_ind, false);
+	if (arena == NULL) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	if (oldp != NULL && oldlenp != NULL) {
+		size_t oldval = dirty ? arena_dirty_decay_ms_get(arena) :
+		    arena_muzzy_decay_ms_get(arena);
+		READ(oldval, ssize_t);
+	}
+	if (newp != NULL) {
+		if (newlen != sizeof(ssize_t)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		if (arena_is_huge(arena_ind) && *(ssize_t *)newp > 0) {
+			/*
+			 * By default the huge arena purges eagerly.  If it is
+			 * set to non-zero decay time afterwards, background
+			 * thread might be needed.
+			 */
+			if (background_thread_create(tsd, arena_ind)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		}
+		if (dirty ? arena_dirty_decay_ms_set(tsd_tsdn(tsd), arena,
+		    *(ssize_t *)newp) : arena_muzzy_decay_ms_set(tsd_tsdn(tsd),
+		    arena, *(ssize_t *)newp)) {
+			ret = EFAULT;
+			goto label_return;
+		}
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arena_i_dirty_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arena_i_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, true);
+}
+
+static int
+arena_i_muzzy_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arena_i_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, false);
+}
+
+static int
+arena_i_extent_hooks_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	MIB_UNSIGNED(arena_ind, 1);
+	if (arena_ind < narenas_total_get()) {
+		extent_hooks_t *old_extent_hooks;
+		arena = arena_get(tsd_tsdn(tsd), arena_ind, false);
+		if (arena == NULL) {
+			if (arena_ind >= narenas_auto) {
+				ret = EFAULT;
+				goto label_return;
+			}
+			old_extent_hooks =
+			    (extent_hooks_t *)&extent_hooks_default;
+			READ(old_extent_hooks, extent_hooks_t *);
+			if (newp != NULL) {
+				/* Initialize a new arena as a side effect. */
+				extent_hooks_t *new_extent_hooks
+				    JEMALLOC_CC_SILENCE_INIT(NULL);
+				WRITE(new_extent_hooks, extent_hooks_t *);
+				arena = arena_init(tsd_tsdn(tsd), arena_ind,
+				    new_extent_hooks);
+				if (arena == NULL) {
+					ret = EFAULT;
+					goto label_return;
+				}
+			}
+		} else {
+			if (newp != NULL) {
+				extent_hooks_t *new_extent_hooks
+				    JEMALLOC_CC_SILENCE_INIT(NULL);
+				WRITE(new_extent_hooks, extent_hooks_t *);
+				old_extent_hooks = extent_hooks_set(tsd, arena,
+				    new_extent_hooks);
+				READ(old_extent_hooks, extent_hooks_t *);
+			} else {
+				old_extent_hooks = extent_hooks_get(arena);
+				READ(old_extent_hooks, extent_hooks_t *);
+			}
+		}
+	} else {
+		ret = EFAULT;
+		goto label_return;
+	}
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arena_i_retain_grow_limit_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	if (!opt_retain) {
+		/* Only relevant when retain is enabled. */
+		return ENOENT;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	MIB_UNSIGNED(arena_ind, 1);
+	if (arena_ind < narenas_total_get() && (arena =
+	    arena_get(tsd_tsdn(tsd), arena_ind, false)) != NULL) {
+		size_t old_limit, new_limit;
+		if (newp != NULL) {
+			WRITE(new_limit, size_t);
+		}
+		bool err = arena_retain_grow_limit_get_set(tsd, arena,
+		    &old_limit, newp != NULL ? &new_limit : NULL);
+		if (!err) {
+			READ(old_limit, size_t);
+			ret = 0;
+		} else {
+			ret = EFAULT;
+		}
+	} else {
+		ret = EFAULT;
+	}
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static const ctl_named_node_t *
+arena_i_index(tsdn_t *tsdn, const size_t *mib, size_t miblen,
+    size_t i) {
+	const ctl_named_node_t *ret;
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	switch (i) {
+	case MALLCTL_ARENAS_ALL:
+	case MALLCTL_ARENAS_DESTROYED:
+		break;
+	default:
+		if (i > ctl_arenas->narenas) {
+			ret = NULL;
+			goto label_return;
+		}
+		break;
+	}
+
+	ret = super_arena_i_node;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+arenas_narenas_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned narenas;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	READONLY();
+	if (*oldlenp != sizeof(unsigned)) {
+		ret = EINVAL;
+		goto label_return;
+	}
+	narenas = ctl_arenas->narenas;
+	READ(narenas, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arenas_decay_ms_ctl_impl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen, bool dirty) {
+	int ret;
+
+	if (oldp != NULL && oldlenp != NULL) {
+		size_t oldval = (dirty ? arena_dirty_decay_ms_default_get() :
+		    arena_muzzy_decay_ms_default_get());
+		READ(oldval, ssize_t);
+	}
+	if (newp != NULL) {
+		if (newlen != sizeof(ssize_t)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		if (dirty ? arena_dirty_decay_ms_default_set(*(ssize_t *)newp)
+		    : arena_muzzy_decay_ms_default_set(*(ssize_t *)newp)) {
+			ret = EFAULT;
+			goto label_return;
+		}
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arenas_dirty_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arenas_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, true);
+}
+
+static int
+arenas_muzzy_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arenas_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, false);
+}
+
+CTL_RO_NL_GEN(arenas_quantum, QUANTUM, size_t)
+CTL_RO_NL_GEN(arenas_page, PAGE, size_t)
+CTL_RO_NL_GEN(arenas_tcache_max, tcache_maxclass, size_t)
+CTL_RO_NL_GEN(arenas_nbins, SC_NBINS, unsigned)
+CTL_RO_NL_GEN(arenas_nhbins, nhbins, unsigned)
+CTL_RO_NL_GEN(arenas_bin_i_size, bin_infos[mib[2]].reg_size, size_t)
+CTL_RO_NL_GEN(arenas_bin_i_nregs, bin_infos[mib[2]].nregs, uint32_t)
+CTL_RO_NL_GEN(arenas_bin_i_slab_size, bin_infos[mib[2]].slab_size, size_t)
+CTL_RO_NL_GEN(arenas_bin_i_nshards, bin_infos[mib[2]].n_shards, uint32_t)
+static const ctl_named_node_t *
+arenas_bin_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	if (i > SC_NBINS) {
+		return NULL;
+	}
+	return super_arenas_bin_i_node;
+}
+
+CTL_RO_NL_GEN(arenas_nlextents, SC_NSIZES - SC_NBINS, unsigned)
+CTL_RO_NL_GEN(arenas_lextent_i_size, sz_index2size(SC_NBINS+(szind_t)mib[2]),
+    size_t)
+static const ctl_named_node_t *
+arenas_lextent_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	if (i > SC_NSIZES - SC_NBINS) {
+		return NULL;
+	}
+	return super_arenas_lextent_i_node;
+}
+
+static int
+arenas_create_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	extent_hooks_t *extent_hooks;
+	unsigned arena_ind;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+
+	extent_hooks = (extent_hooks_t *)&extent_hooks_default;
+	WRITE(extent_hooks, extent_hooks_t *);
+	if ((arena_ind = ctl_arena_init(tsd, extent_hooks)) == UINT_MAX) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+	READ(arena_ind, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arenas_lookup_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	void *ptr;
+	extent_t *extent;
+	arena_t *arena;
+
+	ptr = NULL;
+	ret = EINVAL;
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	WRITE(ptr, void *);
+	extent = iealloc(tsd_tsdn(tsd), ptr);
+	if (extent == NULL)
+		goto label_return;
+
+	arena = extent_arena_get(extent);
+	if (arena == NULL)
+		goto label_return;
+
+	arena_ind = arena_ind_get(arena);
+	READ(arena_ind, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+prof_thread_active_init_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = prof_thread_active_init_set(tsd_tsdn(tsd),
+		    *(bool *)newp);
+	} else {
+		oldval = prof_thread_active_init_get(tsd_tsdn(tsd));
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_active_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = prof_active_set(tsd_tsdn(tsd), *(bool *)newp);
+	} else {
+		oldval = prof_active_get(tsd_tsdn(tsd));
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_dump_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	const char *filename = NULL;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	WRITEONLY();
+	WRITE(filename, const char *);
+
+	if (prof_mdump(tsd, filename)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_gdump_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = prof_gdump_set(tsd_tsdn(tsd), *(bool *)newp);
+	} else {
+		oldval = prof_gdump_get(tsd_tsdn(tsd));
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_reset_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	size_t lg_sample = lg_prof_sample;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	WRITEONLY();
+	WRITE(lg_sample, size_t);
+	if (lg_sample >= (sizeof(uint64_t) << 3)) {
+		lg_sample = (sizeof(uint64_t) << 3) - 1;
+	}
+
+	prof_reset(tsd, lg_sample);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+CTL_RO_NL_CGEN(config_prof, prof_interval, prof_interval, uint64_t)
+CTL_RO_NL_CGEN(config_prof, lg_prof_sample, lg_prof_sample, size_t)
+
+static int
+prof_log_start_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+
+	const char *filename = NULL;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	WRITEONLY();
+	WRITE(filename, const char *);
+
+	if (prof_log_start(tsd_tsdn(tsd), filename)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_log_stop_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (prof_log_stop(tsd_tsdn(tsd))) {
+		return EFAULT;
+	}
+
+	return 0;
+}
+
+/******************************************************************************/
+
+CTL_RO_CGEN(config_stats, stats_allocated, ctl_stats->allocated, size_t)
+CTL_RO_CGEN(config_stats, stats_active, ctl_stats->active, size_t)
+CTL_RO_CGEN(config_stats, stats_metadata, ctl_stats->metadata, size_t)
+CTL_RO_CGEN(config_stats, stats_metadata_thp, ctl_stats->metadata_thp, size_t)
+CTL_RO_CGEN(config_stats, stats_resident, ctl_stats->resident, size_t)
+CTL_RO_CGEN(config_stats, stats_mapped, ctl_stats->mapped, size_t)
+CTL_RO_CGEN(config_stats, stats_retained, ctl_stats->retained, size_t)
+
+CTL_RO_CGEN(config_stats, stats_background_thread_num_threads,
+    ctl_stats->background_thread.num_threads, size_t)
+CTL_RO_CGEN(config_stats, stats_background_thread_num_runs,
+    ctl_stats->background_thread.num_runs, uint64_t)
+CTL_RO_CGEN(config_stats, stats_background_thread_run_interval,
+    nstime_ns(&ctl_stats->background_thread.run_interval), uint64_t)
+
+CTL_RO_GEN(stats_arenas_i_dss, arenas_i(mib[2])->dss, const char *)
+CTL_RO_GEN(stats_arenas_i_dirty_decay_ms, arenas_i(mib[2])->dirty_decay_ms,
+    ssize_t)
+CTL_RO_GEN(stats_arenas_i_muzzy_decay_ms, arenas_i(mib[2])->muzzy_decay_ms,
+    ssize_t)
+CTL_RO_GEN(stats_arenas_i_nthreads, arenas_i(mib[2])->nthreads, unsigned)
+CTL_RO_GEN(stats_arenas_i_uptime,
+    nstime_ns(&arenas_i(mib[2])->astats->astats.uptime), uint64_t)
+CTL_RO_GEN(stats_arenas_i_pactive, arenas_i(mib[2])->pactive, size_t)
+CTL_RO_GEN(stats_arenas_i_pdirty, arenas_i(mib[2])->pdirty, size_t)
+CTL_RO_GEN(stats_arenas_i_pmuzzy, arenas_i(mib[2])->pmuzzy, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_mapped,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.mapped, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_retained,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.retained, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_extent_avail,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.extent_avail,
+        ATOMIC_RELAXED),
+    size_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_dirty_npurge,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_dirty.npurge), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_dirty_nmadvise,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_dirty.nmadvise), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_dirty_purged,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_dirty.purged), uint64_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_muzzy_npurge,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_muzzy.npurge), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_muzzy_nmadvise,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_muzzy.nmadvise), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_muzzy_purged,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_muzzy.purged), uint64_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_base,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.base, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_internal,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.internal, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_metadata_thp,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.metadata_thp,
+    ATOMIC_RELAXED), size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_tcache_bytes,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.tcache_bytes,
+    ATOMIC_RELAXED), size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_resident,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.resident, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_abandoned_vm,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.abandoned_vm,
+    ATOMIC_RELAXED), size_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_allocated,
+    arenas_i(mib[2])->astats->allocated_small, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nmalloc,
+    arenas_i(mib[2])->astats->nmalloc_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_ndalloc,
+    arenas_i(mib[2])->astats->ndalloc_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nrequests,
+    arenas_i(mib[2])->astats->nrequests_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nfills,
+    arenas_i(mib[2])->astats->nfills_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nflushes,
+    arenas_i(mib[2])->astats->nflushes_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_allocated,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.allocated_large,
+    ATOMIC_RELAXED), size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nmalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nmalloc_large), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_ndalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.ndalloc_large), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nrequests,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nrequests_large), uint64_t)
+/*
+ * Note: "nmalloc_large" here instead of "nfills" in the read.  This is
+ * intentional (large has no batch fill).
+ */
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nfills,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nmalloc_large), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nflushes,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nflushes_large), uint64_t)
+
+/* Lock profiling related APIs below. */
+#define RO_MUTEX_CTL_GEN(n, l)						\
+CTL_RO_CGEN(config_stats, stats_##n##_num_ops,				\
+    l.n_lock_ops, uint64_t)						\
+CTL_RO_CGEN(config_stats, stats_##n##_num_wait,				\
+    l.n_wait_times, uint64_t)						\
+CTL_RO_CGEN(config_stats, stats_##n##_num_spin_acq,			\
+    l.n_spin_acquired, uint64_t)					\
+CTL_RO_CGEN(config_stats, stats_##n##_num_owner_switch,			\
+    l.n_owner_switches, uint64_t) 					\
+CTL_RO_CGEN(config_stats, stats_##n##_total_wait_time,			\
+    nstime_ns(&l.tot_wait_time), uint64_t)				\
+CTL_RO_CGEN(config_stats, stats_##n##_max_wait_time,			\
+    nstime_ns(&l.max_wait_time), uint64_t)				\
+CTL_RO_CGEN(config_stats, stats_##n##_max_num_thds,			\
+    l.max_n_thds, uint32_t)
+
+/* Global mutexes. */
+#define OP(mtx)								\
+    RO_MUTEX_CTL_GEN(mutexes_##mtx,					\
+        ctl_stats->mutex_prof_data[global_prof_mutex_##mtx])
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+
+/* Per arena mutexes */
+#define OP(mtx) RO_MUTEX_CTL_GEN(arenas_i_mutexes_##mtx,		\
+    arenas_i(mib[2])->astats->astats.mutex_prof_data[arena_prof_mutex_##mtx])
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+
+/* tcache bin mutex */
+RO_MUTEX_CTL_GEN(arenas_i_bins_j_mutex,
+    arenas_i(mib[2])->astats->bstats[mib[4]].mutex_data)
+#undef RO_MUTEX_CTL_GEN
+
+/* Resets all mutex stats, including global, arena and bin mutexes. */
+static int
+stats_mutexes_reset_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen) {
+	if (!config_stats) {
+		return ENOENT;
+	}
+
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+
+#define MUTEX_PROF_RESET(mtx)						\
+    malloc_mutex_lock(tsdn, &mtx);					\
+    malloc_mutex_prof_data_reset(tsdn, &mtx);				\
+    malloc_mutex_unlock(tsdn, &mtx);
+
+	/* Global mutexes: ctl and prof. */
+	MUTEX_PROF_RESET(ctl_mtx);
+	if (have_background_thread) {
+		MUTEX_PROF_RESET(background_thread_lock);
+	}
+	if (config_prof && opt_prof) {
+		MUTEX_PROF_RESET(bt2gctx_mtx);
+	}
+
+
+	/* Per arena mutexes. */
+	unsigned n = narenas_total_get();
+
+	for (unsigned i = 0; i < n; i++) {
+		arena_t *arena = arena_get(tsdn, i, false);
+		if (!arena) {
+			continue;
+		}
+		MUTEX_PROF_RESET(arena->large_mtx);
+		MUTEX_PROF_RESET(arena->extent_avail_mtx);
+		MUTEX_PROF_RESET(arena->extents_dirty.mtx);
+		MUTEX_PROF_RESET(arena->extents_muzzy.mtx);
+		MUTEX_PROF_RESET(arena->extents_retained.mtx);
+		MUTEX_PROF_RESET(arena->decay_dirty.mtx);
+		MUTEX_PROF_RESET(arena->decay_muzzy.mtx);
+		MUTEX_PROF_RESET(arena->tcache_ql_mtx);
+		MUTEX_PROF_RESET(arena->base->mtx);
+
+		for (szind_t i = 0; i < SC_NBINS; i++) {
+			for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+				bin_t *bin = &arena->bins[i].bin_shards[j];
+				MUTEX_PROF_RESET(bin->lock);
+			}
+		}
+	}
+#undef MUTEX_PROF_RESET
+	return 0;
+}
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nmalloc,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nmalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_ndalloc,
+    arenas_i(mib[2])->astats->bstats[mib[4]].ndalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nrequests,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nrequests, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_curregs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].curregs, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nfills,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nfills, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nflushes,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nflushes, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nslabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nslabs, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nreslabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].reslabs, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_curslabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].curslabs, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nonfull_slabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nonfull_slabs, size_t)
+
+static const ctl_named_node_t *
+stats_arenas_i_bins_j_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t j) {
+	if (j > SC_NBINS) {
+		return NULL;
+	}
+	return super_stats_arenas_i_bins_j_node;
+}
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_nmalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->lstats[mib[4]].nmalloc), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_ndalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->lstats[mib[4]].ndalloc), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_nrequests,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->lstats[mib[4]].nrequests), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_curlextents,
+    arenas_i(mib[2])->astats->lstats[mib[4]].curlextents, size_t)
+
+static const ctl_named_node_t *
+stats_arenas_i_lextents_j_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t j) {
+	if (j > SC_NSIZES - SC_NBINS) {
+		return NULL;
+	}
+	return super_stats_arenas_i_lextents_j_node;
+}
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_ndirty,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].ndirty,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_nmuzzy,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].nmuzzy,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_nretained,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].nretained,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_dirty_bytes,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].dirty_bytes,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_muzzy_bytes,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].muzzy_bytes,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_retained_bytes,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].retained_bytes,
+	ATOMIC_RELAXED), size_t);
+
+static const ctl_named_node_t *
+stats_arenas_i_extents_j_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t j) {
+	if (j >= SC_NPSIZES) {
+		return NULL;
+	}
+	return super_stats_arenas_i_extents_j_node;
+}
+
+static bool
+ctl_arenas_i_verify(size_t i) {
+	size_t a = arenas_i2a_impl(i, true, true);
+	if (a == UINT_MAX || !ctl_arenas->arenas[a]->initialized) {
+		return true;
+	}
+
+	return false;
+}
+
+static const ctl_named_node_t *
+stats_arenas_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	const ctl_named_node_t *ret;
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	if (ctl_arenas_i_verify(i)) {
+		ret = NULL;
+		goto label_return;
+	}
+
+	ret = super_stats_arenas_i_node;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+static int
+experimental_hooks_install_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	if (oldp == NULL || oldlenp == NULL|| newp == NULL) {
+		ret = EINVAL;
+		goto label_return;
+	}
+	/*
+	 * Note: this is a *private* struct.  This is an experimental interface;
+	 * forcing the user to know the jemalloc internals well enough to
+	 * extract the ABI hopefully ensures nobody gets too comfortable with
+	 * this API, which can change at a moment's notice.
+	 */
+	hooks_t hooks;
+	WRITE(hooks, hooks_t);
+	void *handle = hook_install(tsd_tsdn(tsd), &hooks);
+	if (handle == NULL) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+	READ(handle, void *);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+experimental_hooks_remove_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	WRITEONLY();
+	void *handle = NULL;
+	WRITE(handle, void *);
+	if (handle == NULL) {
+		ret = EINVAL;
+		goto label_return;
+	}
+	hook_remove(tsd_tsdn(tsd), handle);
+	ret = 0;
+label_return:
+	return ret;
+}
+
+/*
+ * Output six memory utilization entries for an input pointer, the first one of
+ * type (void *) and the remaining five of type size_t, describing the following
+ * (in the same order):
+ *
+ * (a) memory address of the extent a potential reallocation would go into,
+ * == the five fields below describe about the extent the pointer resides in ==
+ * (b) number of free regions in the extent,
+ * (c) number of regions in the extent,
+ * (d) size of the extent in terms of bytes,
+ * (e) total number of free regions in the bin the extent belongs to, and
+ * (f) total number of regions in the bin the extent belongs to.
+ *
+ * Note that "(e)" and "(f)" are only available when stats are enabled;
+ * otherwise their values are undefined.
+ *
+ * This API is mainly intended for small class allocations, where extents are
+ * used as slab.  Note that if the bin the extent belongs to is completely
+ * full, "(a)" will be NULL.
+ *
+ * In case of large class allocations, "(a)" will be NULL, and "(e)" and "(f)"
+ * will be zero (if stats are enabled; otherwise undefined).  The other three
+ * fields will be properly set though the values are trivial: "(b)" will be 0,
+ * "(c)" will be 1, and "(d)" will be the usable size.
+ *
+ * The input pointer and size are respectively passed in by newp and newlen,
+ * and the output fields and size are respectively oldp and *oldlenp.
+ *
+ * It can be beneficial to define the following macros to make it easier to
+ * access the output:
+ *
+ * #define SLABCUR_READ(out) (*(void **)out)
+ * #define COUNTS(out) ((size_t *)((void **)out + 1))
+ * #define NFREE_READ(out) COUNTS(out)[0]
+ * #define NREGS_READ(out) COUNTS(out)[1]
+ * #define SIZE_READ(out) COUNTS(out)[2]
+ * #define BIN_NFREE_READ(out) COUNTS(out)[3]
+ * #define BIN_NREGS_READ(out) COUNTS(out)[4]
+ *
+ * and then write e.g. NFREE_READ(oldp) to fetch the output.  See the unit test
+ * test_query in test/unit/extent_util.c for an example.
+ *
+ * For a typical defragmentation workflow making use of this API for
+ * understanding the fragmentation level, please refer to the comment for
+ * experimental_utilization_batch_query_ctl.
+ *
+ * It's up to the application how to determine the significance of
+ * fragmentation relying on the outputs returned.  Possible choices are:
+ *
+ * (a) if extent utilization ratio is below certain threshold,
+ * (b) if extent memory consumption is above certain threshold,
+ * (c) if extent utilization ratio is significantly below bin utilization ratio,
+ * (d) if input pointer deviates a lot from potential reallocation address, or
+ * (e) some selection/combination of the above.
+ *
+ * The caller needs to make sure that the input/output arguments are valid,
+ * in particular, that the size of the output is correct, i.e.:
+ *
+ *     *oldlenp = sizeof(void *) + sizeof(size_t) * 5
+ *
+ * Otherwise, the function immediately returns EINVAL without touching anything.
+ *
+ * In the rare case where there's no associated extent found for the input
+ * pointer, the function zeros out all output fields and return.  Please refer
+ * to the comment for experimental_utilization_batch_query_ctl to understand the
+ * motivation from C++.
+ */
+static int
+experimental_utilization_query_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+
+	assert(sizeof(extent_util_stats_verbose_t)
+	    == sizeof(void *) + sizeof(size_t) * 5);
+
+	if (oldp == NULL || oldlenp == NULL
+	    || *oldlenp != sizeof(extent_util_stats_verbose_t)
+	    || newp == NULL) {
+		ret = EINVAL;
+		goto label_return;
+	}
+
+	void *ptr = NULL;
+	WRITE(ptr, void *);
+	extent_util_stats_verbose_t *util_stats
+	    = (extent_util_stats_verbose_t *)oldp;
+	extent_util_stats_verbose_get(tsd_tsdn(tsd), ptr,
+	    &util_stats->nfree, &util_stats->nregs, &util_stats->size,
+	    &util_stats->bin_nfree, &util_stats->bin_nregs,
+	    &util_stats->slabcur_addr);
+	ret = 0;
+
+label_return:
+	return ret;
+}
+
+/*
+ * Given an input array of pointers, output three memory utilization entries of
+ * type size_t for each input pointer about the extent it resides in:
+ *
+ * (a) number of free regions in the extent,
+ * (b) number of regions in the extent, and
+ * (c) size of the extent in terms of bytes.
+ *
+ * This API is mainly intended for small class allocations, where extents are
+ * used as slab.  In case of large class allocations, the outputs are trivial:
+ * "(a)" will be 0, "(b)" will be 1, and "(c)" will be the usable size.
+ *
+ * Note that multiple input pointers may reside on a same extent so the output
+ * fields may contain duplicates.
+ *
+ * The format of the input/output looks like:
+ *
+ * input[0]:  1st_pointer_to_query	|  output[0]: 1st_extent_n_free_regions
+ *					|  output[1]: 1st_extent_n_regions
+ *					|  output[2]: 1st_extent_size
+ * input[1]:  2nd_pointer_to_query	|  output[3]: 2nd_extent_n_free_regions
+ *					|  output[4]: 2nd_extent_n_regions
+ *					|  output[5]: 2nd_extent_size
+ * ...					|  ...
+ *
+ * The input array and size are respectively passed in by newp and newlen, and
+ * the output array and size are respectively oldp and *oldlenp.
+ *
+ * It can be beneficial to define the following macros to make it easier to
+ * access the output:
+ *
+ * #define NFREE_READ(out, i) out[(i) * 3]
+ * #define NREGS_READ(out, i) out[(i) * 3 + 1]
+ * #define SIZE_READ(out, i) out[(i) * 3 + 2]
+ *
+ * and then write e.g. NFREE_READ(oldp, i) to fetch the output.  See the unit
+ * test test_batch in test/unit/extent_util.c for a concrete example.
+ *
+ * A typical workflow would be composed of the following steps:
+ *
+ * (1) flush tcache: mallctl("thread.tcache.flush", ...)
+ * (2) initialize input array of pointers to query fragmentation
+ * (3) allocate output array to hold utilization statistics
+ * (4) query utilization: mallctl("experimental.utilization.batch_query", ...)
+ * (5) (optional) decide if it's worthwhile to defragment; otherwise stop here
+ * (6) disable tcache: mallctl("thread.tcache.enabled", ...)
+ * (7) defragment allocations with significant fragmentation, e.g.:
+ *         for each allocation {
+ *             if it's fragmented {
+ *                 malloc(...);
+ *                 memcpy(...);
+ *                 free(...);
+ *             }
+ *         }
+ * (8) enable tcache: mallctl("thread.tcache.enabled", ...)
+ *
+ * The application can determine the significance of fragmentation themselves
+ * relying on the statistics returned, both at the overall level i.e. step "(5)"
+ * and at individual allocation level i.e. within step "(7)".  Possible choices
+ * are:
+ *
+ * (a) whether memory utilization ratio is below certain threshold,
+ * (b) whether memory consumption is above certain threshold, or
+ * (c) some combination of the two.
+ *
+ * The caller needs to make sure that the input/output arrays are valid and
+ * their sizes are proper as well as matched, meaning:
+ *
+ * (a) newlen = n_pointers * sizeof(const void *)
+ * (b) *oldlenp = n_pointers * sizeof(size_t) * 3
+ * (c) n_pointers > 0
+ *
+ * Otherwise, the function immediately returns EINVAL without touching anything.
+ *
+ * In the rare case where there's no associated extent found for some pointers,
+ * rather than immediately terminating the computation and raising an error,
+ * the function simply zeros out the corresponding output fields and continues
+ * the computation until all input pointers are handled.  The motivations of
+ * such a design are as follows:
+ *
+ * (a) The function always either processes nothing or processes everything, and
+ * never leaves the output half touched and half untouched.
+ *
+ * (b) It facilitates usage needs especially common in C++.  A vast variety of
+ * C++ objects are instantiated with multiple dynamic memory allocations.  For
+ * example, std::string and std::vector typically use at least two allocations,
+ * one for the metadata and one for the actual content.  Other types may use
+ * even more allocations.  When inquiring about utilization statistics, the
+ * caller often wants to examine into all such allocations, especially internal
+ * one(s), rather than just the topmost one.  The issue comes when some
+ * implementations do certain optimizations to reduce/aggregate some internal
+ * allocations, e.g. putting short strings directly into the metadata, and such
+ * decisions are not known to the caller.  Therefore, we permit pointers to
+ * memory usages that may not be returned by previous malloc calls, and we
+ * provide the caller a convenient way to identify such cases.
+ */
+static int
+experimental_utilization_batch_query_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+
+	assert(sizeof(extent_util_stats_t) == sizeof(size_t) * 3);
+
+	const size_t len = newlen / sizeof(const void *);
+	if (oldp == NULL || oldlenp == NULL || newp == NULL || newlen == 0
+	    || newlen != len * sizeof(const void *)
+	    || *oldlenp != len * sizeof(extent_util_stats_t)) {
+		ret = EINVAL;
+		goto label_return;
+	}
+
+	void **ptrs = (void **)newp;
+	extent_util_stats_t *util_stats = (extent_util_stats_t *)oldp;
+	size_t i;
+	for (i = 0; i < len; ++i) {
+		extent_util_stats_get(tsd_tsdn(tsd), ptrs[i],
+		    &util_stats[i].nfree, &util_stats[i].nregs,
+		    &util_stats[i].size);
+	}
+	ret = 0;
+
+label_return:
+	return ret;
+}
+
+static const ctl_named_node_t *
+experimental_arenas_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	const ctl_named_node_t *ret;
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	if (ctl_arenas_i_verify(i)) {
+		ret = NULL;
+		goto label_return;
+	}
+	ret = super_experimental_arenas_i_node;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+static int
+experimental_arenas_i_pactivep_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	if (!config_stats) {
+		return ENOENT;
+	}
+	if (oldp == NULL || oldlenp == NULL || *oldlenp != sizeof(size_t *)) {
+		return EINVAL;
+	}
+
+	unsigned arena_ind;
+	arena_t *arena;
+	int ret;
+	size_t *pactivep;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	READONLY();
+	MIB_UNSIGNED(arena_ind, 2);
+	if (arena_ind < narenas_total_get() && (arena =
+	    arena_get(tsd_tsdn(tsd), arena_ind, false)) != NULL) {
+#if defined(JEMALLOC_GCC_ATOMIC_ATOMICS) ||				\
+    defined(JEMALLOC_GCC_SYNC_ATOMICS) || defined(_MSC_VER)
+		/* Expose the underlying counter for fast read. */
+		pactivep = (size_t *)&(arena->nactive.repr);
+		READ(pactivep, size_t *);
+		ret = 0;
+#else
+		ret = EFAULT;
+#endif
+	} else {
+		ret = EFAULT;
+	}
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
diff --git a/jemalloc/src/div.c b/jemalloc/src/div.c
new file mode 100644
index 0000000..808892a
--- /dev/null
+++ b/jemalloc/src/div.c
@@ -0,0 +1,55 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/div.h"
+
+#include "jemalloc/internal/assert.h"
+
+/*
+ * Suppose we have n = q * d, all integers. We know n and d, and want q = n / d.
+ *
+ * For any k, we have (here, all division is exact; not C-style rounding):
+ * floor(ceil(2^k / d) * n / 2^k) = floor((2^k + r) / d * n / 2^k), where
+ * r = (-2^k) mod d.
+ *
+ * Expanding this out:
+ * ... = floor(2^k / d * n / 2^k + r / d * n / 2^k)
+ *     = floor(n / d + (r / d) * (n / 2^k)).
+ *
+ * The fractional part of n / d is 0 (because of the assumption that d divides n
+ * exactly), so we have:
+ * ... = n / d + floor((r / d) * (n / 2^k))
+ *
+ * So that our initial expression is equal to the quantity we seek, so long as
+ * (r / d) * (n / 2^k) < 1.
+ *
+ * r is a remainder mod d, so r < d and r / d < 1 always. We can make
+ * n / 2 ^ k < 1 by setting k = 32. This gets us a value of magic that works.
+ */
+
+void
+div_init(div_info_t *div_info, size_t d) {
+	/* Nonsensical. */
+	assert(d != 0);
+	/*
+	 * This would make the value of magic too high to fit into a uint32_t
+	 * (we would want magic = 2^32 exactly). This would mess with code gen
+	 * on 32-bit machines.
+	 */
+	assert(d != 1);
+
+	uint64_t two_to_k = ((uint64_t)1 << 32);
+	uint32_t magic = (uint32_t)(two_to_k / d);
+
+	/*
+	 * We want magic = ceil(2^k / d), but C gives us floor. We have to
+	 * increment it unless the result was exact (i.e. unless d is a power of
+	 * two).
+	 */
+	if (two_to_k % d != 0) {
+		magic++;
+	}
+	div_info->magic = magic;
+#ifdef JEMALLOC_DEBUG
+	div_info->d = d;
+#endif
+}
diff --git a/jemalloc/src/extent.c b/jemalloc/src/extent.c
new file mode 100644
index 0000000..efdcdc0
--- /dev/null
+++ b/jemalloc/src/extent.c
@@ -0,0 +1,2408 @@
+#define JEMALLOC_EXTENT_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+
+/******************************************************************************/
+/* Data. */
+
+rtree_t		extents_rtree;
+/* Keyed by the address of the extent_t being protected. */
+mutex_pool_t	extent_mutex_pool;
+
+size_t opt_lg_extent_max_active_fit = LG_EXTENT_MAX_ACTIVE_FIT_DEFAULT;
+
+static const bitmap_info_t extents_bitmap_info =
+    BITMAP_INFO_INITIALIZER(SC_NPSIZES+1);
+
+static void *extent_alloc_default(extent_hooks_t *extent_hooks, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit,
+    unsigned arena_ind);
+static bool extent_dalloc_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static void extent_destroy_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static bool extent_commit_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool extent_commit_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained);
+static bool extent_decommit_default(extent_hooks_t *extent_hooks,
+    void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);
+#ifdef PAGES_CAN_PURGE_LAZY
+static bool extent_purge_lazy_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+#endif
+static bool extent_purge_lazy_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained);
+#ifdef PAGES_CAN_PURGE_FORCED
+static bool extent_purge_forced_default(extent_hooks_t *extent_hooks,
+    void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);
+#endif
+static bool extent_purge_forced_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained);
+static bool extent_split_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t size_a, size_t size_b, bool committed,
+    unsigned arena_ind);
+static extent_t *extent_split_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b,
+    bool growing_retained);
+static bool extent_merge_default(extent_hooks_t *extent_hooks, void *addr_a,
+    size_t size_a, void *addr_b, size_t size_b, bool committed,
+    unsigned arena_ind);
+static bool extent_merge_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b,
+    bool growing_retained);
+
+const extent_hooks_t	extent_hooks_default = {
+	extent_alloc_default,
+	extent_dalloc_default,
+	extent_destroy_default,
+	extent_commit_default,
+	extent_decommit_default
+#ifdef PAGES_CAN_PURGE_LAZY
+	,
+	extent_purge_lazy_default
+#else
+	,
+	NULL
+#endif
+#ifdef PAGES_CAN_PURGE_FORCED
+	,
+	extent_purge_forced_default
+#else
+	,
+	NULL
+#endif
+	,
+	extent_split_default,
+	extent_merge_default
+};
+
+/* Used exclusively for gdump triggering. */
+static atomic_zu_t curpages;
+static atomic_zu_t highpages;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static void extent_deregister(tsdn_t *tsdn, extent_t *extent);
+static extent_t *extent_recycle(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, void *new_addr,
+    size_t usize, size_t pad, size_t alignment, bool slab, szind_t szind,
+    bool *zero, bool *commit, bool growing_retained);
+static extent_t *extent_try_coalesce(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained);
+static void extent_record(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, extent_t *extent,
+    bool growing_retained);
+
+/******************************************************************************/
+
+#define ATTR_NONE /* does nothing */
+
+ph_gen(ATTR_NONE, extent_avail_, extent_tree_t, extent_t, ph_link,
+    extent_esnead_comp)
+
+#undef ATTR_NONE
+
+typedef enum {
+	lock_result_success,
+	lock_result_failure,
+	lock_result_no_extent
+} lock_result_t;
+
+static lock_result_t
+extent_rtree_leaf_elm_try_lock(tsdn_t *tsdn, rtree_leaf_elm_t *elm,
+    extent_t **result, bool inactive_only) {
+	extent_t *extent1 = rtree_leaf_elm_extent_read(tsdn, &extents_rtree,
+	    elm, true);
+
+	/* Slab implies active extents and should be skipped. */
+	if (extent1 == NULL || (inactive_only && rtree_leaf_elm_slab_read(tsdn,
+	    &extents_rtree, elm, true))) {
+		return lock_result_no_extent;
+	}
+
+	/*
+	 * It's possible that the extent changed out from under us, and with it
+	 * the leaf->extent mapping.  We have to recheck while holding the lock.
+	 */
+	extent_lock(tsdn, extent1);
+	extent_t *extent2 = rtree_leaf_elm_extent_read(tsdn,
+	    &extents_rtree, elm, true);
+
+	if (extent1 == extent2) {
+		*result = extent1;
+		return lock_result_success;
+	} else {
+		extent_unlock(tsdn, extent1);
+		return lock_result_failure;
+	}
+}
+
+/*
+ * Returns a pool-locked extent_t * if there's one associated with the given
+ * address, and NULL otherwise.
+ */
+static extent_t *
+extent_lock_from_addr(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx, void *addr,
+    bool inactive_only) {
+	extent_t *ret = NULL;
+	rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, &extents_rtree,
+	    rtree_ctx, (uintptr_t)addr, false, false);
+	if (elm == NULL) {
+		return NULL;
+	}
+	lock_result_t lock_result;
+	do {
+		lock_result = extent_rtree_leaf_elm_try_lock(tsdn, elm, &ret,
+		    inactive_only);
+	} while (lock_result == lock_result_failure);
+	return ret;
+}
+
+extent_t *
+extent_alloc(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_lock(tsdn, &arena->extent_avail_mtx);
+	extent_t *extent = extent_avail_first(&arena->extent_avail);
+	if (extent == NULL) {
+		malloc_mutex_unlock(tsdn, &arena->extent_avail_mtx);
+		return base_alloc_extent(tsdn, arena->base);
+	}
+	extent_avail_remove(&arena->extent_avail, extent);
+	atomic_fetch_sub_zu(&arena->extent_avail_cnt, 1, ATOMIC_RELAXED);
+	malloc_mutex_unlock(tsdn, &arena->extent_avail_mtx);
+	return extent;
+}
+
+void
+extent_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	malloc_mutex_lock(tsdn, &arena->extent_avail_mtx);
+	extent_avail_insert(&arena->extent_avail, extent);
+	atomic_fetch_add_zu(&arena->extent_avail_cnt, 1, ATOMIC_RELAXED);
+	malloc_mutex_unlock(tsdn, &arena->extent_avail_mtx);
+}
+
+extent_hooks_t *
+extent_hooks_get(arena_t *arena) {
+	return base_extent_hooks_get(arena->base);
+}
+
+extent_hooks_t *
+extent_hooks_set(tsd_t *tsd, arena_t *arena, extent_hooks_t *extent_hooks) {
+	background_thread_info_t *info;
+	if (have_background_thread) {
+		info = arena_background_thread_info_get(arena);
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	}
+	extent_hooks_t *ret = base_extent_hooks_set(arena->base, extent_hooks);
+	if (have_background_thread) {
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+	}
+
+	return ret;
+}
+
+static void
+extent_hooks_assure_initialized(arena_t *arena,
+    extent_hooks_t **r_extent_hooks) {
+	if (*r_extent_hooks == EXTENT_HOOKS_INITIALIZER) {
+		*r_extent_hooks = extent_hooks_get(arena);
+	}
+}
+
+#ifndef JEMALLOC_JET
+static
+#endif
+size_t
+extent_size_quantize_floor(size_t size) {
+	size_t ret;
+	pszind_t pind;
+
+	assert(size > 0);
+	assert((size & PAGE_MASK) == 0);
+
+	pind = sz_psz2ind(size - sz_large_pad + 1);
+	if (pind == 0) {
+		/*
+		 * Avoid underflow.  This short-circuit would also do the right
+		 * thing for all sizes in the range for which there are
+		 * PAGE-spaced size classes, but it's simplest to just handle
+		 * the one case that would cause erroneous results.
+		 */
+		return size;
+	}
+	ret = sz_pind2sz(pind - 1) + sz_large_pad;
+	assert(ret <= size);
+	return ret;
+}
+
+#ifndef JEMALLOC_JET
+static
+#endif
+size_t
+extent_size_quantize_ceil(size_t size) {
+	size_t ret;
+
+	assert(size > 0);
+	assert(size - sz_large_pad <= SC_LARGE_MAXCLASS);
+	assert((size & PAGE_MASK) == 0);
+
+	ret = extent_size_quantize_floor(size);
+	if (ret < size) {
+		/*
+		 * Skip a quantization that may have an adequately large extent,
+		 * because under-sized extents may be mixed in.  This only
+		 * happens when an unusual size is requested, i.e. for aligned
+		 * allocation, and is just one of several places where linear
+		 * search would potentially find sufficiently aligned available
+		 * memory somewhere lower.
+		 */
+		ret = sz_pind2sz(sz_psz2ind(ret - sz_large_pad + 1)) +
+		    sz_large_pad;
+	}
+	return ret;
+}
+
+/* Generate pairing heap functions. */
+ph_gen(, extent_heap_, extent_heap_t, extent_t, ph_link, extent_snad_comp)
+
+bool
+extents_init(tsdn_t *tsdn, extents_t *extents, extent_state_t state,
+    bool delay_coalesce) {
+	if (malloc_mutex_init(&extents->mtx, "extents", WITNESS_RANK_EXTENTS,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	for (unsigned i = 0; i < SC_NPSIZES + 1; i++) {
+		extent_heap_new(&extents->heaps[i]);
+	}
+	bitmap_init(extents->bitmap, &extents_bitmap_info, true);
+	extent_list_init(&extents->lru);
+	atomic_store_zu(&extents->npages, 0, ATOMIC_RELAXED);
+	extents->state = state;
+	extents->delay_coalesce = delay_coalesce;
+	return false;
+}
+
+extent_state_t
+extents_state_get(const extents_t *extents) {
+	return extents->state;
+}
+
+size_t
+extents_npages_get(extents_t *extents) {
+	return atomic_load_zu(&extents->npages, ATOMIC_RELAXED);
+}
+
+size_t
+extents_nextents_get(extents_t *extents, pszind_t pind) {
+	return atomic_load_zu(&extents->nextents[pind], ATOMIC_RELAXED);
+}
+
+size_t
+extents_nbytes_get(extents_t *extents, pszind_t pind) {
+	return atomic_load_zu(&extents->nbytes[pind], ATOMIC_RELAXED);
+}
+
+static void
+extents_stats_add(extents_t *extent, pszind_t pind, size_t sz) {
+	size_t cur = atomic_load_zu(&extent->nextents[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nextents[pind], cur + 1, ATOMIC_RELAXED);
+	cur = atomic_load_zu(&extent->nbytes[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nbytes[pind], cur + sz, ATOMIC_RELAXED);
+}
+
+static void
+extents_stats_sub(extents_t *extent, pszind_t pind, size_t sz) {
+	size_t cur = atomic_load_zu(&extent->nextents[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nextents[pind], cur - 1, ATOMIC_RELAXED);
+	cur = atomic_load_zu(&extent->nbytes[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nbytes[pind], cur - sz, ATOMIC_RELAXED);
+}
+
+static void
+extents_insert_locked(tsdn_t *tsdn, extents_t *extents, extent_t *extent) {
+	malloc_mutex_assert_owner(tsdn, &extents->mtx);
+	assert(extent_state_get(extent) == extents->state);
+
+	size_t size = extent_size_get(extent);
+	size_t psz = extent_size_quantize_floor(size);
+	pszind_t pind = sz_psz2ind(psz);
+	if (extent_heap_empty(&extents->heaps[pind])) {
+		bitmap_unset(extents->bitmap, &extents_bitmap_info,
+		    (size_t)pind);
+	}
+	extent_heap_insert(&extents->heaps[pind], extent);
+
+	if (config_stats) {
+		extents_stats_add(extents, pind, size);
+	}
+
+	extent_list_append(&extents->lru, extent);
+	size_t npages = size >> LG_PAGE;
+	/*
+	 * All modifications to npages hold the mutex (as asserted above), so we
+	 * don't need an atomic fetch-add; we can get by with a load followed by
+	 * a store.
+	 */
+	size_t cur_extents_npages =
+	    atomic_load_zu(&extents->npages, ATOMIC_RELAXED);
+	atomic_store_zu(&extents->npages, cur_extents_npages + npages,
+	    ATOMIC_RELAXED);
+}
+
+static void
+extents_remove_locked(tsdn_t *tsdn, extents_t *extents, extent_t *extent) {
+	malloc_mutex_assert_owner(tsdn, &extents->mtx);
+	assert(extent_state_get(extent) == extents->state);
+
+	size_t size = extent_size_get(extent);
+	size_t psz = extent_size_quantize_floor(size);
+	pszind_t pind = sz_psz2ind(psz);
+	extent_heap_remove(&extents->heaps[pind], extent);
+
+	if (config_stats) {
+		extents_stats_sub(extents, pind, size);
+	}
+
+	if (extent_heap_empty(&extents->heaps[pind])) {
+		bitmap_set(extents->bitmap, &extents_bitmap_info,
+		    (size_t)pind);
+	}
+	extent_list_remove(&extents->lru, extent);
+	size_t npages = size >> LG_PAGE;
+	/*
+	 * As in extents_insert_locked, we hold extents->mtx and so don't need
+	 * atomic operations for updating extents->npages.
+	 */
+	size_t cur_extents_npages =
+	    atomic_load_zu(&extents->npages, ATOMIC_RELAXED);
+	assert(cur_extents_npages >= npages);
+	atomic_store_zu(&extents->npages,
+	    cur_extents_npages - (size >> LG_PAGE), ATOMIC_RELAXED);
+}
+
+/*
+ * Find an extent with size [min_size, max_size) to satisfy the alignment
+ * requirement.  For each size, try only the first extent in the heap.
+ */
+static extent_t *
+extents_fit_alignment(extents_t *extents, size_t min_size, size_t max_size,
+    size_t alignment) {
+        pszind_t pind = sz_psz2ind(extent_size_quantize_ceil(min_size));
+        pszind_t pind_max = sz_psz2ind(extent_size_quantize_ceil(max_size));
+
+	for (pszind_t i = (pszind_t)bitmap_ffu(extents->bitmap,
+	    &extents_bitmap_info, (size_t)pind); i < pind_max; i =
+	    (pszind_t)bitmap_ffu(extents->bitmap, &extents_bitmap_info,
+	    (size_t)i+1)) {
+		assert(i < SC_NPSIZES);
+		assert(!extent_heap_empty(&extents->heaps[i]));
+		extent_t *extent = extent_heap_first(&extents->heaps[i]);
+		uintptr_t base = (uintptr_t)extent_base_get(extent);
+		size_t candidate_size = extent_size_get(extent);
+		assert(candidate_size >= min_size);
+
+		uintptr_t next_align = ALIGNMENT_CEILING((uintptr_t)base,
+		    PAGE_CEILING(alignment));
+		if (base > next_align || base + candidate_size <= next_align) {
+			/* Overflow or not crossing the next alignment. */
+			continue;
+		}
+
+		size_t leadsize = next_align - base;
+		if (candidate_size - leadsize >= min_size) {
+			return extent;
+		}
+	}
+
+	return NULL;
+}
+
+/*
+ * Do first-fit extent selection, i.e. select the oldest/lowest extent that is
+ * large enough.
+ */
+static extent_t *
+extents_first_fit_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    size_t size) {
+	extent_t *ret = NULL;
+
+	pszind_t pind = sz_psz2ind(extent_size_quantize_ceil(size));
+
+	if (!maps_coalesce && !opt_retain) {
+		/*
+		 * No split / merge allowed (Windows w/o retain). Try exact fit
+		 * only.
+		 */
+		return extent_heap_empty(&extents->heaps[pind]) ? NULL :
+		    extent_heap_first(&extents->heaps[pind]);
+	}
+
+	for (pszind_t i = (pszind_t)bitmap_ffu(extents->bitmap,
+	    &extents_bitmap_info, (size_t)pind);
+	    i < SC_NPSIZES + 1;
+	    i = (pszind_t)bitmap_ffu(extents->bitmap, &extents_bitmap_info,
+	    (size_t)i+1)) {
+		assert(!extent_heap_empty(&extents->heaps[i]));
+		extent_t *extent = extent_heap_first(&extents->heaps[i]);
+		assert(extent_size_get(extent) >= size);
+		/*
+		 * In order to reduce fragmentation, avoid reusing and splitting
+		 * large extents for much smaller sizes.
+		 *
+		 * Only do check for dirty extents (delay_coalesce).
+		 */
+		if (extents->delay_coalesce &&
+		    (sz_pind2sz(i) >> opt_lg_extent_max_active_fit) > size) {
+			break;
+		}
+		if (ret == NULL || extent_snad_comp(extent, ret) < 0) {
+			ret = extent;
+		}
+		if (i == SC_NPSIZES) {
+			break;
+		}
+		assert(i < SC_NPSIZES);
+	}
+
+	return ret;
+}
+
+/*
+ * Do first-fit extent selection, where the selection policy choice is
+ * based on extents->delay_coalesce.
+ */
+static extent_t *
+extents_fit_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    size_t esize, size_t alignment) {
+	malloc_mutex_assert_owner(tsdn, &extents->mtx);
+
+	size_t max_size = esize + PAGE_CEILING(alignment) - PAGE;
+	/* Beware size_t wrap-around. */
+	if (max_size < esize) {
+		return NULL;
+	}
+
+	extent_t *extent =
+	    extents_first_fit_locked(tsdn, arena, extents, max_size);
+
+	if (alignment > PAGE && extent == NULL) {
+		/*
+		 * max_size guarantees the alignment requirement but is rather
+		 * pessimistic.  Next we try to satisfy the aligned allocation
+		 * with sizes in [esize, max_size).
+		 */
+		extent = extents_fit_alignment(extents, esize, max_size,
+		    alignment);
+	}
+
+	return extent;
+}
+
+static bool
+extent_try_delayed_coalesce(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent) {
+	extent_state_set(extent, extent_state_active);
+	bool coalesced;
+	extent = extent_try_coalesce(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, extent, &coalesced, false);
+	extent_state_set(extent, extents_state_get(extents));
+
+	if (!coalesced) {
+		return true;
+	}
+	extents_insert_locked(tsdn, extents, extent);
+	return false;
+}
+
+extent_t *
+extents_alloc(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	assert(size + pad != 0);
+	assert(alignment != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_t *extent = extent_recycle(tsdn, arena, r_extent_hooks, extents,
+	    new_addr, size, pad, alignment, slab, szind, zero, commit, false);
+	assert(extent == NULL || extent_dumpable_get(extent));
+	return extent;
+}
+
+void
+extents_dalloc(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *extent) {
+	assert(extent_base_get(extent) != NULL);
+	assert(extent_size_get(extent) != 0);
+	assert(extent_dumpable_get(extent));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_addr_set(extent, extent_base_get(extent));
+	extent_zeroed_set(extent, false);
+
+	extent_record(tsdn, arena, r_extent_hooks, extents, extent, false);
+}
+
+extent_t *
+extents_evict(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, size_t npages_min) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	malloc_mutex_lock(tsdn, &extents->mtx);
+
+	/*
+	 * Get the LRU coalesced extent, if any.  If coalescing was delayed,
+	 * the loop will iterate until the LRU extent is fully coalesced.
+	 */
+	extent_t *extent;
+	while (true) {
+		/* Get the LRU extent, if any. */
+		extent = extent_list_first(&extents->lru);
+		if (extent == NULL) {
+			goto label_return;
+		}
+		/* Check the eviction limit. */
+		size_t extents_npages = atomic_load_zu(&extents->npages,
+		    ATOMIC_RELAXED);
+		if (extents_npages <= npages_min) {
+			extent = NULL;
+			goto label_return;
+		}
+		extents_remove_locked(tsdn, extents, extent);
+		if (!extents->delay_coalesce) {
+			break;
+		}
+		/* Try to coalesce. */
+		if (extent_try_delayed_coalesce(tsdn, arena, r_extent_hooks,
+		    rtree_ctx, extents, extent)) {
+			break;
+		}
+		/*
+		 * The LRU extent was just coalesced and the result placed in
+		 * the LRU at its neighbor's position.  Start over.
+		 */
+	}
+
+	/*
+	 * Either mark the extent active or deregister it to protect against
+	 * concurrent operations.
+	 */
+	switch (extents_state_get(extents)) {
+	case extent_state_active:
+		not_reached();
+	case extent_state_dirty:
+	case extent_state_muzzy:
+		extent_state_set(extent, extent_state_active);
+		break;
+	case extent_state_retained:
+		extent_deregister(tsdn, extent);
+		break;
+	default:
+		not_reached();
+	}
+
+label_return:
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+	return extent;
+}
+
+/*
+ * This can only happen when we fail to allocate a new extent struct (which
+ * indicates OOM), e.g. when trying to split an existing extent.
+ */
+static void
+extents_abandon_vm(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *extent, bool growing_retained) {
+	size_t sz = extent_size_get(extent);
+	if (config_stats) {
+		arena_stats_accum_zu(&arena->stats.abandoned_vm, sz);
+	}
+	/*
+	 * Leak extent after making sure its pages have already been purged, so
+	 * that this is only a virtual memory leak.
+	 */
+	if (extents_state_get(extents) == extent_state_dirty) {
+		if (extent_purge_lazy_impl(tsdn, arena, r_extent_hooks,
+		    extent, 0, sz, growing_retained)) {
+			extent_purge_forced_impl(tsdn, arena, r_extent_hooks,
+			    extent, 0, extent_size_get(extent),
+			    growing_retained);
+		}
+	}
+	extent_dalloc(tsdn, arena, extent);
+}
+
+void
+extents_prefork(tsdn_t *tsdn, extents_t *extents) {
+	malloc_mutex_prefork(tsdn, &extents->mtx);
+}
+
+void
+extents_postfork_parent(tsdn_t *tsdn, extents_t *extents) {
+	malloc_mutex_postfork_parent(tsdn, &extents->mtx);
+}
+
+void
+extents_postfork_child(tsdn_t *tsdn, extents_t *extents) {
+	malloc_mutex_postfork_child(tsdn, &extents->mtx);
+}
+
+static void
+extent_deactivate_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    extent_t *extent) {
+	assert(extent_arena_get(extent) == arena);
+	assert(extent_state_get(extent) == extent_state_active);
+
+	extent_state_set(extent, extents_state_get(extents));
+	extents_insert_locked(tsdn, extents, extent);
+}
+
+static void
+extent_deactivate(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    extent_t *extent) {
+	malloc_mutex_lock(tsdn, &extents->mtx);
+	extent_deactivate_locked(tsdn, arena, extents, extent);
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+}
+
+static void
+extent_activate_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    extent_t *extent) {
+	assert(extent_arena_get(extent) == arena);
+	assert(extent_state_get(extent) == extents_state_get(extents));
+
+	extents_remove_locked(tsdn, extents, extent);
+	extent_state_set(extent, extent_state_active);
+}
+
+static bool
+extent_rtree_leaf_elms_lookup(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx,
+    const extent_t *extent, bool dependent, bool init_missing,
+    rtree_leaf_elm_t **r_elm_a, rtree_leaf_elm_t **r_elm_b) {
+	*r_elm_a = rtree_leaf_elm_lookup(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_base_get(extent), dependent, init_missing);
+	if (!dependent && *r_elm_a == NULL) {
+		return true;
+	}
+	assert(*r_elm_a != NULL);
+
+	*r_elm_b = rtree_leaf_elm_lookup(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_last_get(extent), dependent, init_missing);
+	if (!dependent && *r_elm_b == NULL) {
+		return true;
+	}
+	assert(*r_elm_b != NULL);
+
+	return false;
+}
+
+static void
+extent_rtree_write_acquired(tsdn_t *tsdn, rtree_leaf_elm_t *elm_a,
+    rtree_leaf_elm_t *elm_b, extent_t *extent, szind_t szind, bool slab) {
+	rtree_leaf_elm_write(tsdn, &extents_rtree, elm_a, extent, szind, slab);
+	if (elm_b != NULL) {
+		rtree_leaf_elm_write(tsdn, &extents_rtree, elm_b, extent, szind,
+		    slab);
+	}
+}
+
+static void
+extent_interior_register(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx, extent_t *extent,
+    szind_t szind) {
+	assert(extent_slab_get(extent));
+
+	/* Register interior. */
+	for (size_t i = 1; i < (extent_size_get(extent) >> LG_PAGE) - 1; i++) {
+		rtree_write(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)extent_base_get(extent) + (uintptr_t)(i <<
+		    LG_PAGE), extent, szind, true);
+	}
+}
+
+static void
+extent_gdump_add(tsdn_t *tsdn, const extent_t *extent) {
+	cassert(config_prof);
+	/* prof_gdump() requirement. */
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (opt_prof && extent_state_get(extent) == extent_state_active) {
+		size_t nadd = extent_size_get(extent) >> LG_PAGE;
+		size_t cur = atomic_fetch_add_zu(&curpages, nadd,
+		    ATOMIC_RELAXED) + nadd;
+		size_t high = atomic_load_zu(&highpages, ATOMIC_RELAXED);
+		while (cur > high && !atomic_compare_exchange_weak_zu(
+		    &highpages, &high, cur, ATOMIC_RELAXED, ATOMIC_RELAXED)) {
+			/*
+			 * Don't refresh cur, because it may have decreased
+			 * since this thread lost the highpages update race.
+			 * Note that high is updated in case of CAS failure.
+			 */
+		}
+		if (cur > high && prof_gdump_get_unlocked()) {
+			prof_gdump(tsdn);
+		}
+	}
+}
+
+static void
+extent_gdump_sub(tsdn_t *tsdn, const extent_t *extent) {
+	cassert(config_prof);
+
+	if (opt_prof && extent_state_get(extent) == extent_state_active) {
+		size_t nsub = extent_size_get(extent) >> LG_PAGE;
+		assert(atomic_load_zu(&curpages, ATOMIC_RELAXED) >= nsub);
+		atomic_fetch_sub_zu(&curpages, nsub, ATOMIC_RELAXED);
+	}
+}
+
+static bool
+extent_register_impl(tsdn_t *tsdn, extent_t *extent, bool gdump_add) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *elm_a, *elm_b;
+
+	/*
+	 * We need to hold the lock to protect against a concurrent coalesce
+	 * operation that sees us in a partial state.
+	 */
+	extent_lock(tsdn, extent);
+
+	if (extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, extent, false, true,
+	    &elm_a, &elm_b)) {
+		extent_unlock(tsdn, extent);
+		return true;
+	}
+
+	szind_t szind = extent_szind_get_maybe_invalid(extent);
+	bool slab = extent_slab_get(extent);
+	extent_rtree_write_acquired(tsdn, elm_a, elm_b, extent, szind, slab);
+	if (slab) {
+		extent_interior_register(tsdn, rtree_ctx, extent, szind);
+	}
+
+	extent_unlock(tsdn, extent);
+
+	if (config_prof && gdump_add) {
+		extent_gdump_add(tsdn, extent);
+	}
+
+	return false;
+}
+
+static bool
+extent_register(tsdn_t *tsdn, extent_t *extent) {
+	return extent_register_impl(tsdn, extent, true);
+}
+
+static bool
+extent_register_no_gdump_add(tsdn_t *tsdn, extent_t *extent) {
+	return extent_register_impl(tsdn, extent, false);
+}
+
+static void
+extent_reregister(tsdn_t *tsdn, extent_t *extent) {
+	bool err = extent_register(tsdn, extent);
+	assert(!err);
+}
+
+/*
+ * Removes all pointers to the given extent from the global rtree indices for
+ * its interior.  This is relevant for slab extents, for which we need to do
+ * metadata lookups at places other than the head of the extent.  We deregister
+ * on the interior, then, when an extent moves from being an active slab to an
+ * inactive state.
+ */
+static void
+extent_interior_deregister(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx,
+    extent_t *extent) {
+	size_t i;
+
+	assert(extent_slab_get(extent));
+
+	for (i = 1; i < (extent_size_get(extent) >> LG_PAGE) - 1; i++) {
+		rtree_clear(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)extent_base_get(extent) + (uintptr_t)(i <<
+		    LG_PAGE));
+	}
+}
+
+/*
+ * Removes all pointers to the given extent from the global rtree.
+ */
+static void
+extent_deregister_impl(tsdn_t *tsdn, extent_t *extent, bool gdump) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *elm_a, *elm_b;
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, extent, true, false,
+	    &elm_a, &elm_b);
+
+	extent_lock(tsdn, extent);
+
+	extent_rtree_write_acquired(tsdn, elm_a, elm_b, NULL, SC_NSIZES, false);
+	if (extent_slab_get(extent)) {
+		extent_interior_deregister(tsdn, rtree_ctx, extent);
+		extent_slab_set(extent, false);
+	}
+
+	extent_unlock(tsdn, extent);
+
+	if (config_prof && gdump) {
+		extent_gdump_sub(tsdn, extent);
+	}
+}
+
+static void
+extent_deregister(tsdn_t *tsdn, extent_t *extent) {
+	extent_deregister_impl(tsdn, extent, true);
+}
+
+static void
+extent_deregister_no_gdump_sub(tsdn_t *tsdn, extent_t *extent) {
+	extent_deregister_impl(tsdn, extent, false);
+}
+
+/*
+ * Tries to find and remove an extent from extents that can be used for the
+ * given allocation request.
+ */
+static extent_t *
+extent_recycle_extract(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    void *new_addr, size_t size, size_t pad, size_t alignment, bool slab,
+    bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+	assert(alignment > 0);
+	if (config_debug && new_addr != NULL) {
+		/*
+		 * Non-NULL new_addr has two use cases:
+		 *
+		 *   1) Recycle a known-extant extent, e.g. during purging.
+		 *   2) Perform in-place expanding reallocation.
+		 *
+		 * Regardless of use case, new_addr must either refer to a
+		 * non-existing extent, or to the base of an extant extent,
+		 * since only active slabs support interior lookups (which of
+		 * course cannot be recycled).
+		 */
+		assert(PAGE_ADDR2BASE(new_addr) == new_addr);
+		assert(pad == 0);
+		assert(alignment <= PAGE);
+	}
+
+	size_t esize = size + pad;
+	malloc_mutex_lock(tsdn, &extents->mtx);
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	extent_t *extent;
+	if (new_addr != NULL) {
+		extent = extent_lock_from_addr(tsdn, rtree_ctx, new_addr,
+		    false);
+		if (extent != NULL) {
+			/*
+			 * We might null-out extent to report an error, but we
+			 * still need to unlock the associated mutex after.
+			 */
+			extent_t *unlock_extent = extent;
+			assert(extent_base_get(extent) == new_addr);
+			if (extent_arena_get(extent) != arena ||
+			    extent_size_get(extent) < esize ||
+			    extent_state_get(extent) !=
+			    extents_state_get(extents)) {
+				extent = NULL;
+			}
+			extent_unlock(tsdn, unlock_extent);
+		}
+	} else {
+		extent = extents_fit_locked(tsdn, arena, extents, esize,
+		    alignment);
+	}
+	if (extent == NULL) {
+		malloc_mutex_unlock(tsdn, &extents->mtx);
+		return NULL;
+	}
+
+	extent_activate_locked(tsdn, arena, extents, extent);
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+
+	return extent;
+}
+
+/*
+ * Given an allocation request and an extent guaranteed to be able to satisfy
+ * it, this splits off lead and trail extents, leaving extent pointing to an
+ * extent satisfying the allocation.
+ * This function doesn't put lead or trail into any extents_t; it's the caller's
+ * job to ensure that they can be reused.
+ */
+typedef enum {
+	/*
+	 * Split successfully.  lead, extent, and trail, are modified to extents
+	 * describing the ranges before, in, and after the given allocation.
+	 */
+	extent_split_interior_ok,
+	/*
+	 * The extent can't satisfy the given allocation request.  None of the
+	 * input extent_t *s are touched.
+	 */
+	extent_split_interior_cant_alloc,
+	/*
+	 * In a potentially invalid state.  Must leak (if *to_leak is non-NULL),
+	 * and salvage what's still salvageable (if *to_salvage is non-NULL).
+	 * None of lead, extent, or trail are valid.
+	 */
+	extent_split_interior_error
+} extent_split_interior_result_t;
+
+static extent_split_interior_result_t
+extent_split_interior(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx,
+    /* The result of splitting, in case of success. */
+    extent_t **extent, extent_t **lead, extent_t **trail,
+    /* The mess to clean up, in case of error. */
+    extent_t **to_leak, extent_t **to_salvage,
+    void *new_addr, size_t size, size_t pad, size_t alignment, bool slab,
+    szind_t szind, bool growing_retained) {
+	size_t esize = size + pad;
+	size_t leadsize = ALIGNMENT_CEILING((uintptr_t)extent_base_get(*extent),
+	    PAGE_CEILING(alignment)) - (uintptr_t)extent_base_get(*extent);
+	assert(new_addr == NULL || leadsize == 0);
+	if (extent_size_get(*extent) < leadsize + esize) {
+		return extent_split_interior_cant_alloc;
+	}
+	size_t trailsize = extent_size_get(*extent) - leadsize - esize;
+
+	*lead = NULL;
+	*trail = NULL;
+	*to_leak = NULL;
+	*to_salvage = NULL;
+
+	/* Split the lead. */
+	if (leadsize != 0) {
+		*lead = *extent;
+		*extent = extent_split_impl(tsdn, arena, r_extent_hooks,
+		    *lead, leadsize, SC_NSIZES, false, esize + trailsize, szind,
+		    slab, growing_retained);
+		if (*extent == NULL) {
+			*to_leak = *lead;
+			*lead = NULL;
+			return extent_split_interior_error;
+		}
+	}
+
+	/* Split the trail. */
+	if (trailsize != 0) {
+		*trail = extent_split_impl(tsdn, arena, r_extent_hooks, *extent,
+		    esize, szind, slab, trailsize, SC_NSIZES, false,
+		    growing_retained);
+		if (*trail == NULL) {
+			*to_leak = *extent;
+			*to_salvage = *lead;
+			*lead = NULL;
+			*extent = NULL;
+			return extent_split_interior_error;
+		}
+	}
+
+	if (leadsize == 0 && trailsize == 0) {
+		/*
+		 * Splitting causes szind to be set as a side effect, but no
+		 * splitting occurred.
+		 */
+		extent_szind_set(*extent, szind);
+		if (szind != SC_NSIZES) {
+			rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx,
+			    (uintptr_t)extent_addr_get(*extent), szind, slab);
+			if (slab && extent_size_get(*extent) > PAGE) {
+				rtree_szind_slab_update(tsdn, &extents_rtree,
+				    rtree_ctx,
+				    (uintptr_t)extent_past_get(*extent) -
+				    (uintptr_t)PAGE, szind, slab);
+			}
+		}
+	}
+
+	return extent_split_interior_ok;
+}
+
+/*
+ * This fulfills the indicated allocation request out of the given extent (which
+ * the caller should have ensured was big enough).  If there's any unused space
+ * before or after the resulting allocation, that space is given its own extent
+ * and put back into extents.
+ */
+static extent_t *
+extent_recycle_split(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    void *new_addr, size_t size, size_t pad, size_t alignment, bool slab,
+    szind_t szind, extent_t *extent, bool growing_retained) {
+	extent_t *lead;
+	extent_t *trail;
+	extent_t *to_leak;
+	extent_t *to_salvage;
+
+	extent_split_interior_result_t result = extent_split_interior(
+	    tsdn, arena, r_extent_hooks, rtree_ctx, &extent, &lead, &trail,
+	    &to_leak, &to_salvage, new_addr, size, pad, alignment, slab, szind,
+	    growing_retained);
+
+	if (!maps_coalesce && result != extent_split_interior_ok
+	    && !opt_retain) {
+		/*
+		 * Split isn't supported (implies Windows w/o retain).  Avoid
+		 * leaking the extents.
+		 */
+		assert(to_leak != NULL && lead == NULL && trail == NULL);
+		extent_deactivate(tsdn, arena, extents, to_leak);
+		return NULL;
+	}
+
+	if (result == extent_split_interior_ok) {
+		if (lead != NULL) {
+			extent_deactivate(tsdn, arena, extents, lead);
+		}
+		if (trail != NULL) {
+			extent_deactivate(tsdn, arena, extents, trail);
+		}
+		return extent;
+	} else {
+		/*
+		 * We should have picked an extent that was large enough to
+		 * fulfill our allocation request.
+		 */
+		assert(result == extent_split_interior_error);
+		if (to_salvage != NULL) {
+			extent_deregister(tsdn, to_salvage);
+		}
+		if (to_leak != NULL) {
+			void *leak = extent_base_get(to_leak);
+			extent_deregister_no_gdump_sub(tsdn, to_leak);
+			extents_abandon_vm(tsdn, arena, r_extent_hooks, extents,
+			    to_leak, growing_retained);
+			assert(extent_lock_from_addr(tsdn, rtree_ctx, leak,
+			    false) == NULL);
+		}
+		return NULL;
+	}
+	unreachable();
+}
+
+static bool
+extent_need_manual_zero(arena_t *arena) {
+	/*
+	 * Need to manually zero the extent on repopulating if either; 1) non
+	 * default extent hooks installed (in which case the purge semantics may
+	 * change); or 2) transparent huge pages enabled.
+	 */
+	return (!arena_has_default_hooks(arena) ||
+		(opt_thp == thp_mode_always));
+}
+
+/*
+ * Tries to satisfy the given allocation request by reusing one of the extents
+ * in the given extents_t.
+ */
+static extent_t *
+extent_recycle(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit,
+    bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+	assert(new_addr == NULL || !slab);
+	assert(pad == 0 || !slab);
+	assert(!*zero || !slab);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent = extent_recycle_extract(tsdn, arena, r_extent_hooks,
+	    rtree_ctx, extents, new_addr, size, pad, alignment, slab,
+	    growing_retained);
+	if (extent == NULL) {
+		return NULL;
+	}
+
+	extent = extent_recycle_split(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, new_addr, size, pad, alignment, slab, szind, extent,
+	    growing_retained);
+	if (extent == NULL) {
+		return NULL;
+	}
+
+	if (*commit && !extent_committed_get(extent)) {
+		if (extent_commit_impl(tsdn, arena, r_extent_hooks, extent,
+		    0, extent_size_get(extent), growing_retained)) {
+			extent_record(tsdn, arena, r_extent_hooks, extents,
+			    extent, growing_retained);
+			return NULL;
+		}
+		if (!extent_need_manual_zero(arena)) {
+			extent_zeroed_set(extent, true);
+		}
+	}
+
+	if (extent_committed_get(extent)) {
+		*commit = true;
+	}
+	if (extent_zeroed_get(extent)) {
+		*zero = true;
+	}
+
+	if (pad != 0) {
+		extent_addr_randomize(tsdn, extent, alignment);
+	}
+	assert(extent_state_get(extent) == extent_state_active);
+	if (slab) {
+		extent_slab_set(extent, slab);
+		extent_interior_register(tsdn, rtree_ctx, extent, szind);
+	}
+
+	if (*zero) {
+		void *addr = extent_base_get(extent);
+		if (!extent_zeroed_get(extent)) {
+			size_t size = extent_size_get(extent);
+			if (extent_need_manual_zero(arena) ||
+			    pages_purge_forced(addr, size)) {
+				memset(addr, 0, size);
+			}
+		} else if (config_debug) {
+			size_t *p = (size_t *)(uintptr_t)addr;
+			/* Check the first page only. */
+			for (size_t i = 0; i < PAGE / sizeof(size_t); i++) {
+				assert(p[i] == 0);
+			}
+		}
+	}
+	return extent;
+}
+
+/*
+ * If the caller specifies (!*zero), it is still possible to receive zeroed
+ * memory, in which case *zero is toggled to true.  arena_extent_alloc() takes
+ * advantage of this to avoid demanding zeroed extents, but taking advantage of
+ * them if they are returned.
+ */
+static void *
+extent_alloc_core(tsdn_t *tsdn, arena_t *arena, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit, dss_prec_t dss_prec) {
+	void *ret;
+
+	assert(size != 0);
+	assert(alignment != 0);
+
+	/* "primary" dss. */
+	if (have_dss && dss_prec == dss_prec_primary && (ret =
+	    extent_alloc_dss(tsdn, arena, new_addr, size, alignment, zero,
+	    commit)) != NULL) {
+		return ret;
+	}
+	/* mmap. */
+	if ((ret = extent_alloc_mmap(new_addr, size, alignment, zero, commit))
+	    != NULL) {
+		return ret;
+	}
+	/* "secondary" dss. */
+	if (have_dss && dss_prec == dss_prec_secondary && (ret =
+	    extent_alloc_dss(tsdn, arena, new_addr, size, alignment, zero,
+	    commit)) != NULL) {
+		return ret;
+	}
+
+	/* All strategies for allocation failed. */
+	return NULL;
+}
+
+static void *
+extent_alloc_default_impl(tsdn_t *tsdn, arena_t *arena, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit) {
+	void *ret = extent_alloc_core(tsdn, arena, new_addr, size, alignment, zero,
+	    commit, (dss_prec_t)atomic_load_u(&arena->dss_prec,
+	    ATOMIC_RELAXED));
+	if (have_madvise_huge && ret) {
+		pages_set_thp_state(ret, size);
+	}
+	return ret;
+}
+
+static void *
+extent_alloc_default(extent_hooks_t *extent_hooks, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit, unsigned arena_ind) {
+	tsdn_t *tsdn;
+	arena_t *arena;
+
+	tsdn = tsdn_fetch();
+	arena = arena_get(tsdn, arena_ind, false);
+	/*
+	 * The arena we're allocating on behalf of must have been initialized
+	 * already.
+	 */
+	assert(arena != NULL);
+
+	return extent_alloc_default_impl(tsdn, arena, new_addr, size,
+	    ALIGNMENT_CEILING(alignment, PAGE), zero, commit);
+}
+
+static void
+extent_hook_pre_reentrancy(tsdn_t *tsdn, arena_t *arena) {
+	tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+	if (arena == arena_get(tsd_tsdn(tsd), 0, false)) {
+		/*
+		 * The only legitimate case of customized extent hooks for a0 is
+		 * hooks with no allocation activities.  One such example is to
+		 * place metadata on pre-allocated resources such as huge pages.
+		 * In that case, rely on reentrancy_level checks to catch
+		 * infinite recursions.
+		 */
+		pre_reentrancy(tsd, NULL);
+	} else {
+		pre_reentrancy(tsd, arena);
+	}
+}
+
+static void
+extent_hook_post_reentrancy(tsdn_t *tsdn) {
+	tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+	post_reentrancy(tsd);
+}
+
+/*
+ * If virtual memory is retained, create increasingly larger extents from which
+ * to split requested extents in order to limit the total number of disjoint
+ * virtual memory ranges retained by each arena.
+ */
+static extent_t *
+extent_grow_retained(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, size_t size, size_t pad, size_t alignment,
+    bool slab, szind_t szind, bool *zero, bool *commit) {
+	malloc_mutex_assert_owner(tsdn, &arena->extent_grow_mtx);
+	assert(pad == 0 || !slab);
+	assert(!*zero || !slab);
+
+	size_t esize = size + pad;
+	size_t alloc_size_min = esize + PAGE_CEILING(alignment) - PAGE;
+	/* Beware size_t wrap-around. */
+	if (alloc_size_min < esize) {
+		goto label_err;
+	}
+	/*
+	 * Find the next extent size in the series that would be large enough to
+	 * satisfy this request.
+	 */
+	pszind_t egn_skip = 0;
+	size_t alloc_size = sz_pind2sz(arena->extent_grow_next + egn_skip);
+	while (alloc_size < alloc_size_min) {
+		egn_skip++;
+		if (arena->extent_grow_next + egn_skip >=
+		    sz_psz2ind(SC_LARGE_MAXCLASS)) {
+			/* Outside legal range. */
+			goto label_err;
+		}
+		alloc_size = sz_pind2sz(arena->extent_grow_next + egn_skip);
+	}
+
+	extent_t *extent = extent_alloc(tsdn, arena);
+	if (extent == NULL) {
+		goto label_err;
+	}
+	bool zeroed = false;
+	bool committed = false;
+
+	void *ptr;
+	if (*r_extent_hooks == &extent_hooks_default) {
+		ptr = extent_alloc_default_impl(tsdn, arena, NULL,
+		    alloc_size, PAGE, &zeroed, &committed);
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		ptr = (*r_extent_hooks)->alloc(*r_extent_hooks, NULL,
+		    alloc_size, PAGE, &zeroed, &committed,
+		    arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	extent_init(extent, arena, ptr, alloc_size, false, SC_NSIZES,
+	    arena_extent_sn_next(arena), extent_state_active, zeroed,
+	    committed, true, EXTENT_IS_HEAD);
+	if (ptr == NULL) {
+		extent_dalloc(tsdn, arena, extent);
+		goto label_err;
+	}
+
+	if (extent_register_no_gdump_add(tsdn, extent)) {
+		extent_dalloc(tsdn, arena, extent);
+		goto label_err;
+	}
+
+	if (extent_zeroed_get(extent) && extent_committed_get(extent)) {
+		*zero = true;
+	}
+	if (extent_committed_get(extent)) {
+		*commit = true;
+	}
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *lead;
+	extent_t *trail;
+	extent_t *to_leak;
+	extent_t *to_salvage;
+	extent_split_interior_result_t result = extent_split_interior(
+	    tsdn, arena, r_extent_hooks, rtree_ctx, &extent, &lead, &trail,
+	    &to_leak, &to_salvage, NULL, size, pad, alignment, slab, szind,
+	    true);
+
+	if (result == extent_split_interior_ok) {
+		if (lead != NULL) {
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, lead, true);
+		}
+		if (trail != NULL) {
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, trail, true);
+		}
+	} else {
+		/*
+		 * We should have allocated a sufficiently large extent; the
+		 * cant_alloc case should not occur.
+		 */
+		assert(result == extent_split_interior_error);
+		if (to_salvage != NULL) {
+			if (config_prof) {
+				extent_gdump_add(tsdn, to_salvage);
+			}
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, to_salvage, true);
+		}
+		if (to_leak != NULL) {
+			extent_deregister_no_gdump_sub(tsdn, to_leak);
+			extents_abandon_vm(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, to_leak, true);
+		}
+		goto label_err;
+	}
+
+	if (*commit && !extent_committed_get(extent)) {
+		if (extent_commit_impl(tsdn, arena, r_extent_hooks, extent, 0,
+		    extent_size_get(extent), true)) {
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, extent, true);
+			goto label_err;
+		}
+		if (!extent_need_manual_zero(arena)) {
+			extent_zeroed_set(extent, true);
+		}
+	}
+
+	/*
+	 * Increment extent_grow_next if doing so wouldn't exceed the allowed
+	 * range.
+	 */
+	if (arena->extent_grow_next + egn_skip + 1 <=
+	    arena->retain_grow_limit) {
+		arena->extent_grow_next += egn_skip + 1;
+	} else {
+		arena->extent_grow_next = arena->retain_grow_limit;
+	}
+	/* All opportunities for failure are past. */
+	malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+
+	if (config_prof) {
+		/* Adjust gdump stats now that extent is final size. */
+		extent_gdump_add(tsdn, extent);
+	}
+	if (pad != 0) {
+		extent_addr_randomize(tsdn, extent, alignment);
+	}
+	if (slab) {
+		rtree_ctx_t rtree_ctx_fallback;
+		rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+		    &rtree_ctx_fallback);
+
+		extent_slab_set(extent, true);
+		extent_interior_register(tsdn, rtree_ctx, extent, szind);
+	}
+	if (*zero && !extent_zeroed_get(extent)) {
+		void *addr = extent_base_get(extent);
+		size_t size = extent_size_get(extent);
+		if (extent_need_manual_zero(arena) ||
+		    pages_purge_forced(addr, size)) {
+			memset(addr, 0, size);
+		}
+	}
+
+	return extent;
+label_err:
+	malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+	return NULL;
+}
+
+static extent_t *
+extent_alloc_retained(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	assert(size != 0);
+	assert(alignment != 0);
+
+	malloc_mutex_lock(tsdn, &arena->extent_grow_mtx);
+
+	extent_t *extent = extent_recycle(tsdn, arena, r_extent_hooks,
+	    &arena->extents_retained, new_addr, size, pad, alignment, slab,
+	    szind, zero, commit, true);
+	if (extent != NULL) {
+		malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+		if (config_prof) {
+			extent_gdump_add(tsdn, extent);
+		}
+	} else if (opt_retain && new_addr == NULL) {
+		extent = extent_grow_retained(tsdn, arena, r_extent_hooks, size,
+		    pad, alignment, slab, szind, zero, commit);
+		/* extent_grow_retained() always releases extent_grow_mtx. */
+	} else {
+		malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+	}
+	malloc_mutex_assert_not_owner(tsdn, &arena->extent_grow_mtx);
+
+	return extent;
+}
+
+static extent_t *
+extent_alloc_wrapper_hard(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	size_t esize = size + pad;
+	extent_t *extent = extent_alloc(tsdn, arena);
+	if (extent == NULL) {
+		return NULL;
+	}
+	void *addr;
+	size_t palignment = ALIGNMENT_CEILING(alignment, PAGE);
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		addr = extent_alloc_default_impl(tsdn, arena, new_addr, esize,
+		    palignment, zero, commit);
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		addr = (*r_extent_hooks)->alloc(*r_extent_hooks, new_addr,
+		    esize, palignment, zero, commit, arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+	if (addr == NULL) {
+		extent_dalloc(tsdn, arena, extent);
+		return NULL;
+	}
+	extent_init(extent, arena, addr, esize, slab, szind,
+	    arena_extent_sn_next(arena), extent_state_active, *zero, *commit,
+	    true, EXTENT_NOT_HEAD);
+	if (pad != 0) {
+		extent_addr_randomize(tsdn, extent, alignment);
+	}
+	if (extent_register(tsdn, extent)) {
+		extent_dalloc(tsdn, arena, extent);
+		return NULL;
+	}
+
+	return extent;
+}
+
+extent_t *
+extent_alloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	extent_t *extent = extent_alloc_retained(tsdn, arena, r_extent_hooks,
+	    new_addr, size, pad, alignment, slab, szind, zero, commit);
+	if (extent == NULL) {
+		if (opt_retain && new_addr != NULL) {
+			/*
+			 * When retain is enabled and new_addr is set, we do not
+			 * attempt extent_alloc_wrapper_hard which does mmap
+			 * that is very unlikely to succeed (unless it happens
+			 * to be at the end).
+			 */
+			return NULL;
+		}
+		extent = extent_alloc_wrapper_hard(tsdn, arena, r_extent_hooks,
+		    new_addr, size, pad, alignment, slab, szind, zero, commit);
+	}
+
+	assert(extent == NULL || extent_dumpable_get(extent));
+	return extent;
+}
+
+static bool
+extent_can_coalesce(arena_t *arena, extents_t *extents, const extent_t *inner,
+    const extent_t *outer) {
+	assert(extent_arena_get(inner) == arena);
+	if (extent_arena_get(outer) != arena) {
+		return false;
+	}
+
+	assert(extent_state_get(inner) == extent_state_active);
+	if (extent_state_get(outer) != extents->state) {
+		return false;
+	}
+
+	if (extent_committed_get(inner) != extent_committed_get(outer)) {
+		return false;
+	}
+
+	return true;
+}
+
+static bool
+extent_coalesce(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *inner, extent_t *outer, bool forward,
+    bool growing_retained) {
+	assert(extent_can_coalesce(arena, extents, inner, outer));
+
+	extent_activate_locked(tsdn, arena, extents, outer);
+
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+	bool err = extent_merge_impl(tsdn, arena, r_extent_hooks,
+	    forward ? inner : outer, forward ? outer : inner, growing_retained);
+	malloc_mutex_lock(tsdn, &extents->mtx);
+
+	if (err) {
+		extent_deactivate_locked(tsdn, arena, extents, outer);
+	}
+
+	return err;
+}
+
+static extent_t *
+extent_try_coalesce_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained,
+    bool inactive_only) {
+	/*
+	 * We avoid checking / locking inactive neighbors for large size
+	 * classes, since they are eagerly coalesced on deallocation which can
+	 * cause lock contention.
+	 */
+	/*
+	 * Continue attempting to coalesce until failure, to protect against
+	 * races with other threads that are thwarted by this one.
+	 */
+	bool again;
+	do {
+		again = false;
+
+		/* Try to coalesce forward. */
+		extent_t *next = extent_lock_from_addr(tsdn, rtree_ctx,
+		    extent_past_get(extent), inactive_only);
+		if (next != NULL) {
+			/*
+			 * extents->mtx only protects against races for
+			 * like-state extents, so call extent_can_coalesce()
+			 * before releasing next's pool lock.
+			 */
+			bool can_coalesce = extent_can_coalesce(arena, extents,
+			    extent, next);
+
+			extent_unlock(tsdn, next);
+
+			if (can_coalesce && !extent_coalesce(tsdn, arena,
+			    r_extent_hooks, extents, extent, next, true,
+			    growing_retained)) {
+				if (extents->delay_coalesce) {
+					/* Do minimal coalescing. */
+					*coalesced = true;
+					return extent;
+				}
+				again = true;
+			}
+		}
+
+		/* Try to coalesce backward. */
+		extent_t *prev = extent_lock_from_addr(tsdn, rtree_ctx,
+		    extent_before_get(extent), inactive_only);
+		if (prev != NULL) {
+			bool can_coalesce = extent_can_coalesce(arena, extents,
+			    extent, prev);
+			extent_unlock(tsdn, prev);
+
+			if (can_coalesce && !extent_coalesce(tsdn, arena,
+			    r_extent_hooks, extents, extent, prev, false,
+			    growing_retained)) {
+				extent = prev;
+				if (extents->delay_coalesce) {
+					/* Do minimal coalescing. */
+					*coalesced = true;
+					return extent;
+				}
+				again = true;
+			}
+		}
+	} while (again);
+
+	if (extents->delay_coalesce) {
+		*coalesced = false;
+	}
+	return extent;
+}
+
+static extent_t *
+extent_try_coalesce(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained) {
+	return extent_try_coalesce_impl(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, extent, coalesced, growing_retained, false);
+}
+
+static extent_t *
+extent_try_coalesce_large(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained) {
+	return extent_try_coalesce_impl(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, extent, coalesced, growing_retained, true);
+}
+
+/*
+ * Does the metadata management portions of putting an unused extent into the
+ * given extents_t (coalesces, deregisters slab interiors, the heap operations).
+ */
+static void
+extent_record(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *extent, bool growing_retained) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	assert((extents_state_get(extents) != extent_state_dirty &&
+	    extents_state_get(extents) != extent_state_muzzy) ||
+	    !extent_zeroed_get(extent));
+
+	malloc_mutex_lock(tsdn, &extents->mtx);
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	extent_szind_set(extent, SC_NSIZES);
+	if (extent_slab_get(extent)) {
+		extent_interior_deregister(tsdn, rtree_ctx, extent);
+		extent_slab_set(extent, false);
+	}
+
+	assert(rtree_extent_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_base_get(extent), true) == extent);
+
+	if (!extents->delay_coalesce) {
+		extent = extent_try_coalesce(tsdn, arena, r_extent_hooks,
+		    rtree_ctx, extents, extent, NULL, growing_retained);
+	} else if (extent_size_get(extent) >= SC_LARGE_MINCLASS) {
+		assert(extents == &arena->extents_dirty);
+		/* Always coalesce large extents eagerly. */
+		bool coalesced;
+		do {
+			assert(extent_state_get(extent) == extent_state_active);
+			extent = extent_try_coalesce_large(tsdn, arena,
+			    r_extent_hooks, rtree_ctx, extents, extent,
+			    &coalesced, growing_retained);
+		} while (coalesced);
+		if (extent_size_get(extent) >= oversize_threshold) {
+			/* Shortcut to purge the oversize extent eagerly. */
+			malloc_mutex_unlock(tsdn, &extents->mtx);
+			arena_decay_extent(tsdn, arena, r_extent_hooks, extent);
+			return;
+		}
+	}
+	extent_deactivate_locked(tsdn, arena, extents, extent);
+
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+}
+
+void
+extent_dalloc_gap(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (extent_register(tsdn, extent)) {
+		extent_dalloc(tsdn, arena, extent);
+		return;
+	}
+	extent_dalloc_wrapper(tsdn, arena, &extent_hooks, extent);
+}
+
+static bool
+extent_may_dalloc(void) {
+	/* With retain enabled, the default dalloc always fails. */
+	return !opt_retain;
+}
+
+static bool
+extent_dalloc_default_impl(void *addr, size_t size) {
+	if (!have_dss || !extent_in_dss(addr)) {
+		return extent_dalloc_mmap(addr, size);
+	}
+	return true;
+}
+
+static bool
+extent_dalloc_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	return extent_dalloc_default_impl(addr, size);
+}
+
+static bool
+extent_dalloc_wrapper_try(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	bool err;
+
+	assert(extent_base_get(extent) != NULL);
+	assert(extent_size_get(extent) != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_addr_set(extent, extent_base_get(extent));
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	/* Try to deallocate. */
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		err = extent_dalloc_default_impl(extent_base_get(extent),
+		    extent_size_get(extent));
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		err = ((*r_extent_hooks)->dalloc == NULL ||
+		    (*r_extent_hooks)->dalloc(*r_extent_hooks,
+		    extent_base_get(extent), extent_size_get(extent),
+		    extent_committed_get(extent), arena_ind_get(arena)));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	if (!err) {
+		extent_dalloc(tsdn, arena, extent);
+	}
+
+	return err;
+}
+
+void
+extent_dalloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	assert(extent_dumpable_get(extent));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	/* Avoid calling the default extent_dalloc unless have to. */
+	if (*r_extent_hooks != &extent_hooks_default || extent_may_dalloc()) {
+		/*
+		 * Deregister first to avoid a race with other allocating
+		 * threads, and reregister if deallocation fails.
+		 */
+		extent_deregister(tsdn, extent);
+		if (!extent_dalloc_wrapper_try(tsdn, arena, r_extent_hooks,
+		    extent)) {
+			return;
+		}
+		extent_reregister(tsdn, extent);
+	}
+
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	/* Try to decommit; purge if that fails. */
+	bool zeroed;
+	if (!extent_committed_get(extent)) {
+		zeroed = true;
+	} else if (!extent_decommit_wrapper(tsdn, arena, r_extent_hooks, extent,
+	    0, extent_size_get(extent))) {
+		zeroed = true;
+	} else if ((*r_extent_hooks)->purge_forced != NULL &&
+	    !(*r_extent_hooks)->purge_forced(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), 0,
+	    extent_size_get(extent), arena_ind_get(arena))) {
+		zeroed = true;
+	} else if (extent_state_get(extent) == extent_state_muzzy ||
+	    ((*r_extent_hooks)->purge_lazy != NULL &&
+	    !(*r_extent_hooks)->purge_lazy(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), 0,
+	    extent_size_get(extent), arena_ind_get(arena)))) {
+		zeroed = false;
+	} else {
+		zeroed = false;
+	}
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	extent_zeroed_set(extent, zeroed);
+
+	if (config_prof) {
+		extent_gdump_sub(tsdn, extent);
+	}
+
+	extent_record(tsdn, arena, r_extent_hooks, &arena->extents_retained,
+	    extent, false);
+}
+
+static void
+extent_destroy_default_impl(void *addr, size_t size) {
+	if (!have_dss || !extent_in_dss(addr)) {
+		pages_unmap(addr, size);
+	}
+}
+
+static void
+extent_destroy_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	extent_destroy_default_impl(addr, size);
+}
+
+void
+extent_destroy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	assert(extent_base_get(extent) != NULL);
+	assert(extent_size_get(extent) != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	/* Deregister first to avoid a race with other allocating threads. */
+	extent_deregister(tsdn, extent);
+
+	extent_addr_set(extent, extent_base_get(extent));
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	/* Try to destroy; silently fail otherwise. */
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		extent_destroy_default_impl(extent_base_get(extent),
+		    extent_size_get(extent));
+	} else if ((*r_extent_hooks)->destroy != NULL) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		(*r_extent_hooks)->destroy(*r_extent_hooks,
+		    extent_base_get(extent), extent_size_get(extent),
+		    extent_committed_get(extent), arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	extent_dalloc(tsdn, arena, extent);
+}
+
+static bool
+extent_commit_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	return pages_commit((void *)((uintptr_t)addr + (uintptr_t)offset),
+	    length);
+}
+
+static bool
+extent_commit_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = ((*r_extent_hooks)->commit == NULL ||
+	    (*r_extent_hooks)->commit(*r_extent_hooks, extent_base_get(extent),
+	    extent_size_get(extent), offset, length, arena_ind_get(arena)));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	extent_committed_set(extent, extent_committed_get(extent) || !err);
+	return err;
+}
+
+bool
+extent_commit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	return extent_commit_impl(tsdn, arena, r_extent_hooks, extent, offset,
+	    length, false);
+}
+
+static bool
+extent_decommit_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	return pages_decommit((void *)((uintptr_t)addr + (uintptr_t)offset),
+	    length);
+}
+
+bool
+extent_decommit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = ((*r_extent_hooks)->decommit == NULL ||
+	    (*r_extent_hooks)->decommit(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), offset, length,
+	    arena_ind_get(arena)));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	extent_committed_set(extent, extent_committed_get(extent) && err);
+	return err;
+}
+
+#ifdef PAGES_CAN_PURGE_LAZY
+static bool
+extent_purge_lazy_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	assert(addr != NULL);
+	assert((offset & PAGE_MASK) == 0);
+	assert(length != 0);
+	assert((length & PAGE_MASK) == 0);
+
+	return pages_purge_lazy((void *)((uintptr_t)addr + (uintptr_t)offset),
+	    length);
+}
+#endif
+
+static bool
+extent_purge_lazy_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->purge_lazy == NULL) {
+		return true;
+	}
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = (*r_extent_hooks)->purge_lazy(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), offset, length,
+	    arena_ind_get(arena));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	return err;
+}
+
+bool
+extent_purge_lazy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	return extent_purge_lazy_impl(tsdn, arena, r_extent_hooks, extent,
+	    offset, length, false);
+}
+
+#ifdef PAGES_CAN_PURGE_FORCED
+static bool
+extent_purge_forced_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind) {
+	assert(addr != NULL);
+	assert((offset & PAGE_MASK) == 0);
+	assert(length != 0);
+	assert((length & PAGE_MASK) == 0);
+
+	return pages_purge_forced((void *)((uintptr_t)addr +
+	    (uintptr_t)offset), length);
+}
+#endif
+
+static bool
+extent_purge_forced_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->purge_forced == NULL) {
+		return true;
+	}
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = (*r_extent_hooks)->purge_forced(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), offset, length,
+	    arena_ind_get(arena));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	return err;
+}
+
+bool
+extent_purge_forced_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	return extent_purge_forced_impl(tsdn, arena, r_extent_hooks, extent,
+	    offset, length, false);
+}
+
+static bool
+extent_split_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t size_a, size_t size_b, bool committed, unsigned arena_ind) {
+	if (!maps_coalesce) {
+		/*
+		 * Without retain, only whole regions can be purged (required by
+		 * MEM_RELEASE on Windows) -- therefore disallow splitting.  See
+		 * comments in extent_head_no_merge().
+		 */
+		return !opt_retain;
+	}
+
+	return false;
+}
+
+/*
+ * Accepts the extent to split, and the characteristics of each side of the
+ * split.  The 'a' parameters go with the 'lead' of the resulting pair of
+ * extents (the lower addressed portion of the split), and the 'b' parameters go
+ * with the trail (the higher addressed portion).  This makes 'extent' the lead,
+ * and returns the trail (except in case of error).
+ */
+static extent_t *
+extent_split_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b,
+    bool growing_retained) {
+	assert(extent_size_get(extent) == size_a + size_b);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->split == NULL) {
+		return NULL;
+	}
+
+	extent_t *trail = extent_alloc(tsdn, arena);
+	if (trail == NULL) {
+		goto label_error_a;
+	}
+
+	extent_init(trail, arena, (void *)((uintptr_t)extent_base_get(extent) +
+	    size_a), size_b, slab_b, szind_b, extent_sn_get(extent),
+	    extent_state_get(extent), extent_zeroed_get(extent),
+	    extent_committed_get(extent), extent_dumpable_get(extent),
+	    EXTENT_NOT_HEAD);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *lead_elm_a, *lead_elm_b;
+	{
+		extent_t lead;
+
+		extent_init(&lead, arena, extent_addr_get(extent), size_a,
+		    slab_a, szind_a, extent_sn_get(extent),
+		    extent_state_get(extent), extent_zeroed_get(extent),
+		    extent_committed_get(extent), extent_dumpable_get(extent),
+		    EXTENT_NOT_HEAD);
+
+		extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, &lead, false,
+		    true, &lead_elm_a, &lead_elm_b);
+	}
+	rtree_leaf_elm_t *trail_elm_a, *trail_elm_b;
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, trail, false, true,
+	    &trail_elm_a, &trail_elm_b);
+
+	if (lead_elm_a == NULL || lead_elm_b == NULL || trail_elm_a == NULL
+	    || trail_elm_b == NULL) {
+		goto label_error_b;
+	}
+
+	extent_lock2(tsdn, extent, trail);
+
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = (*r_extent_hooks)->split(*r_extent_hooks, extent_base_get(extent),
+	    size_a + size_b, size_a, size_b, extent_committed_get(extent),
+	    arena_ind_get(arena));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	if (err) {
+		goto label_error_c;
+	}
+
+	extent_size_set(extent, size_a);
+	extent_szind_set(extent, szind_a);
+
+	extent_rtree_write_acquired(tsdn, lead_elm_a, lead_elm_b, extent,
+	    szind_a, slab_a);
+	extent_rtree_write_acquired(tsdn, trail_elm_a, trail_elm_b, trail,
+	    szind_b, slab_b);
+
+	extent_unlock2(tsdn, extent, trail);
+
+	return trail;
+label_error_c:
+	extent_unlock2(tsdn, extent, trail);
+label_error_b:
+	extent_dalloc(tsdn, arena, trail);
+label_error_a:
+	return NULL;
+}
+
+extent_t *
+extent_split_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b) {
+	return extent_split_impl(tsdn, arena, r_extent_hooks, extent, size_a,
+	    szind_a, slab_a, size_b, szind_b, slab_b, false);
+}
+
+static bool
+extent_merge_default_impl(void *addr_a, void *addr_b) {
+	if (!maps_coalesce && !opt_retain) {
+		return true;
+	}
+	if (have_dss && !extent_dss_mergeable(addr_a, addr_b)) {
+		return true;
+	}
+
+	return false;
+}
+
+/*
+ * Returns true if the given extents can't be merged because of their head bit
+ * settings.  Assumes the second extent has the higher address.
+ */
+static bool
+extent_head_no_merge(extent_t *a, extent_t *b) {
+	assert(extent_base_get(a) < extent_base_get(b));
+	/*
+	 * When coalesce is not always allowed (Windows), only merge extents
+	 * from the same VirtualAlloc region under opt.retain (in which case
+	 * MEM_DECOMMIT is utilized for purging).
+	 */
+	if (maps_coalesce) {
+		return false;
+	}
+	if (!opt_retain) {
+		return true;
+	}
+	/* If b is a head extent, disallow the cross-region merge. */
+	if (extent_is_head_get(b)) {
+		/*
+		 * Additionally, sn should not overflow with retain; sanity
+		 * check that different regions have unique sn.
+		 */
+		assert(extent_sn_comp(a, b) != 0);
+		return true;
+	}
+	assert(extent_sn_comp(a, b) == 0);
+
+	return false;
+}
+
+static bool
+extent_merge_default(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a,
+    void *addr_b, size_t size_b, bool committed, unsigned arena_ind) {
+	if (!maps_coalesce) {
+		tsdn_t *tsdn = tsdn_fetch();
+		extent_t *a = iealloc(tsdn, addr_a);
+		extent_t *b = iealloc(tsdn, addr_b);
+		if (extent_head_no_merge(a, b)) {
+			return true;
+		}
+	}
+	return extent_merge_default_impl(addr_a, addr_b);
+}
+
+static bool
+extent_merge_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b,
+    bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+	assert(extent_base_get(a) < extent_base_get(b));
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->merge == NULL || extent_head_no_merge(a, b)) {
+		return true;
+	}
+
+	bool err;
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		err = extent_merge_default_impl(extent_base_get(a),
+		    extent_base_get(b));
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		err = (*r_extent_hooks)->merge(*r_extent_hooks,
+		    extent_base_get(a), extent_size_get(a), extent_base_get(b),
+		    extent_size_get(b), extent_committed_get(a),
+		    arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	if (err) {
+		return true;
+	}
+
+	/*
+	 * The rtree writes must happen while all the relevant elements are
+	 * owned, so the following code uses decomposed helper functions rather
+	 * than extent_{,de}register() to do things in the right order.
+	 */
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *a_elm_a, *a_elm_b, *b_elm_a, *b_elm_b;
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, a, true, false, &a_elm_a,
+	    &a_elm_b);
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, b, true, false, &b_elm_a,
+	    &b_elm_b);
+
+	extent_lock2(tsdn, a, b);
+
+	if (a_elm_b != NULL) {
+		rtree_leaf_elm_write(tsdn, &extents_rtree, a_elm_b, NULL,
+		    SC_NSIZES, false);
+	}
+	if (b_elm_b != NULL) {
+		rtree_leaf_elm_write(tsdn, &extents_rtree, b_elm_a, NULL,
+		    SC_NSIZES, false);
+	} else {
+		b_elm_b = b_elm_a;
+	}
+
+	extent_size_set(a, extent_size_get(a) + extent_size_get(b));
+	extent_szind_set(a, SC_NSIZES);
+	extent_sn_set(a, (extent_sn_get(a) < extent_sn_get(b)) ?
+	    extent_sn_get(a) : extent_sn_get(b));
+	extent_zeroed_set(a, extent_zeroed_get(a) && extent_zeroed_get(b));
+
+	extent_rtree_write_acquired(tsdn, a_elm_a, b_elm_b, a, SC_NSIZES,
+	    false);
+
+	extent_unlock2(tsdn, a, b);
+
+	extent_dalloc(tsdn, extent_arena_get(b), b);
+
+	return false;
+}
+
+bool
+extent_merge_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b) {
+	return extent_merge_impl(tsdn, arena, r_extent_hooks, a, b, false);
+}
+
+bool
+extent_boot(void) {
+	if (rtree_new(&extents_rtree, true)) {
+		return true;
+	}
+
+	if (mutex_pool_init(&extent_mutex_pool, "extent_mutex_pool",
+	    WITNESS_RANK_EXTENT_POOL)) {
+		return true;
+	}
+
+	if (have_dss) {
+		extent_dss_boot();
+	}
+
+	return false;
+}
+
+void
+extent_util_stats_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size) {
+	assert(ptr != NULL && nfree != NULL && nregs != NULL && size != NULL);
+
+	const extent_t *extent = iealloc(tsdn, ptr);
+	if (unlikely(extent == NULL)) {
+		*nfree = *nregs = *size = 0;
+		return;
+	}
+
+	*size = extent_size_get(extent);
+	if (!extent_slab_get(extent)) {
+		*nfree = 0;
+		*nregs = 1;
+	} else {
+		*nfree = extent_nfree_get(extent);
+		*nregs = bin_infos[extent_szind_get(extent)].nregs;
+		assert(*nfree <= *nregs);
+		assert(*nfree * extent_usize_get(extent) <= *size);
+	}
+}
+
+void
+extent_util_stats_verbose_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size,
+    size_t *bin_nfree, size_t *bin_nregs, void **slabcur_addr) {
+	assert(ptr != NULL && nfree != NULL && nregs != NULL && size != NULL
+	    && bin_nfree != NULL && bin_nregs != NULL && slabcur_addr != NULL);
+
+	const extent_t *extent = iealloc(tsdn, ptr);
+	if (unlikely(extent == NULL)) {
+		*nfree = *nregs = *size = *bin_nfree = *bin_nregs = 0;
+		*slabcur_addr = NULL;
+		return;
+	}
+
+	*size = extent_size_get(extent);
+	if (!extent_slab_get(extent)) {
+		*nfree = *bin_nfree = *bin_nregs = 0;
+		*nregs = 1;
+		*slabcur_addr = NULL;
+		return;
+	}
+
+	*nfree = extent_nfree_get(extent);
+	const szind_t szind = extent_szind_get(extent);
+	*nregs = bin_infos[szind].nregs;
+	assert(*nfree <= *nregs);
+	assert(*nfree * extent_usize_get(extent) <= *size);
+
+	const arena_t *arena = extent_arena_get(extent);
+	assert(arena != NULL);
+	const unsigned binshard = extent_binshard_get(extent);
+	bin_t *bin = &arena->bins[szind].bin_shards[binshard];
+
+	malloc_mutex_lock(tsdn, &bin->lock);
+	if (config_stats) {
+		*bin_nregs = *nregs * bin->stats.curslabs;
+		assert(*bin_nregs >= bin->stats.curregs);
+		*bin_nfree = *bin_nregs - bin->stats.curregs;
+	} else {
+		*bin_nfree = *bin_nregs = 0;
+	}
+	extent_t *slab;
+	if (bin->slabcur != NULL) {
+		slab = bin->slabcur;
+	} else {
+		slab = extent_heap_first(&bin->slabs_nonfull);
+	}
+	*slabcur_addr = slab != NULL ? extent_addr_get(slab) : NULL;
+	malloc_mutex_unlock(tsdn, &bin->lock);
+}
diff --git a/jemalloc/src/extent_dss.c b/jemalloc/src/extent_dss.c
new file mode 100644
index 0000000..8581789
--- /dev/null
+++ b/jemalloc/src/extent_dss.c
@@ -0,0 +1,271 @@
+#define JEMALLOC_EXTENT_DSS_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/spin.h"
+
+/******************************************************************************/
+/* Data. */
+
+const char	*opt_dss = DSS_DEFAULT;
+
+const char	*dss_prec_names[] = {
+	"disabled",
+	"primary",
+	"secondary",
+	"N/A"
+};
+
+/*
+ * Current dss precedence default, used when creating new arenas.  NB: This is
+ * stored as unsigned rather than dss_prec_t because in principle there's no
+ * guarantee that sizeof(dss_prec_t) is the same as sizeof(unsigned), and we use
+ * atomic operations to synchronize the setting.
+ */
+static atomic_u_t	dss_prec_default = ATOMIC_INIT(
+    (unsigned)DSS_PREC_DEFAULT);
+
+/* Base address of the DSS. */
+static void		*dss_base;
+/* Atomic boolean indicating whether a thread is currently extending DSS. */
+static atomic_b_t	dss_extending;
+/* Atomic boolean indicating whether the DSS is exhausted. */
+static atomic_b_t	dss_exhausted;
+/* Atomic current upper limit on DSS addresses. */
+static atomic_p_t	dss_max;
+
+/******************************************************************************/
+
+static void *
+extent_dss_sbrk(intptr_t increment) {
+#ifdef JEMALLOC_DSS
+	return sbrk(increment);
+#else
+	not_implemented();
+	return NULL;
+#endif
+}
+
+dss_prec_t
+extent_dss_prec_get(void) {
+	dss_prec_t ret;
+
+	if (!have_dss) {
+		return dss_prec_disabled;
+	}
+	ret = (dss_prec_t)atomic_load_u(&dss_prec_default, ATOMIC_ACQUIRE);
+	return ret;
+}
+
+bool
+extent_dss_prec_set(dss_prec_t dss_prec) {
+	if (!have_dss) {
+		return (dss_prec != dss_prec_disabled);
+	}
+	atomic_store_u(&dss_prec_default, (unsigned)dss_prec, ATOMIC_RELEASE);
+	return false;
+}
+
+static void
+extent_dss_extending_start(void) {
+	spin_t spinner = SPIN_INITIALIZER;
+	while (true) {
+		bool expected = false;
+		if (atomic_compare_exchange_weak_b(&dss_extending, &expected,
+		    true, ATOMIC_ACQ_REL, ATOMIC_RELAXED)) {
+			break;
+		}
+		spin_adaptive(&spinner);
+	}
+}
+
+static void
+extent_dss_extending_finish(void) {
+	assert(atomic_load_b(&dss_extending, ATOMIC_RELAXED));
+
+	atomic_store_b(&dss_extending, false, ATOMIC_RELEASE);
+}
+
+static void *
+extent_dss_max_update(void *new_addr) {
+	/*
+	 * Get the current end of the DSS as max_cur and assure that dss_max is
+	 * up to date.
+	 */
+	void *max_cur = extent_dss_sbrk(0);
+	if (max_cur == (void *)-1) {
+		return NULL;
+	}
+	atomic_store_p(&dss_max, max_cur, ATOMIC_RELEASE);
+	/* Fixed new_addr can only be supported if it is at the edge of DSS. */
+	if (new_addr != NULL && max_cur != new_addr) {
+		return NULL;
+	}
+	return max_cur;
+}
+
+void *
+extent_alloc_dss(tsdn_t *tsdn, arena_t *arena, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit) {
+	extent_t *gap;
+
+	cassert(have_dss);
+	assert(size > 0);
+	assert(alignment == ALIGNMENT_CEILING(alignment, PAGE));
+
+	/*
+	 * sbrk() uses a signed increment argument, so take care not to
+	 * interpret a large allocation request as a negative increment.
+	 */
+	if ((intptr_t)size < 0) {
+		return NULL;
+	}
+
+	gap = extent_alloc(tsdn, arena);
+	if (gap == NULL) {
+		return NULL;
+	}
+
+	extent_dss_extending_start();
+	if (!atomic_load_b(&dss_exhausted, ATOMIC_ACQUIRE)) {
+		/*
+		 * The loop is necessary to recover from races with other
+		 * threads that are using the DSS for something other than
+		 * malloc.
+		 */
+		while (true) {
+			void *max_cur = extent_dss_max_update(new_addr);
+			if (max_cur == NULL) {
+				goto label_oom;
+			}
+
+			/*
+			 * Compute how much page-aligned gap space (if any) is
+			 * necessary to satisfy alignment.  This space can be
+			 * recycled for later use.
+			 */
+			void *gap_addr_page = (void *)(PAGE_CEILING(
+			    (uintptr_t)max_cur));
+			void *ret = (void *)ALIGNMENT_CEILING(
+			    (uintptr_t)gap_addr_page, alignment);
+			size_t gap_size_page = (uintptr_t)ret -
+			    (uintptr_t)gap_addr_page;
+			if (gap_size_page != 0) {
+				extent_init(gap, arena, gap_addr_page,
+				    gap_size_page, false, SC_NSIZES,
+				    arena_extent_sn_next(arena),
+				    extent_state_active, false, true, true,
+				    EXTENT_NOT_HEAD);
+			}
+			/*
+			 * Compute the address just past the end of the desired
+			 * allocation space.
+			 */
+			void *dss_next = (void *)((uintptr_t)ret + size);
+			if ((uintptr_t)ret < (uintptr_t)max_cur ||
+			    (uintptr_t)dss_next < (uintptr_t)max_cur) {
+				goto label_oom; /* Wrap-around. */
+			}
+			/* Compute the increment, including subpage bytes. */
+			void *gap_addr_subpage = max_cur;
+			size_t gap_size_subpage = (uintptr_t)ret -
+			    (uintptr_t)gap_addr_subpage;
+			intptr_t incr = gap_size_subpage + size;
+
+			assert((uintptr_t)max_cur + incr == (uintptr_t)ret +
+			    size);
+
+			/* Try to allocate. */
+			void *dss_prev = extent_dss_sbrk(incr);
+			if (dss_prev == max_cur) {
+				/* Success. */
+				atomic_store_p(&dss_max, dss_next,
+				    ATOMIC_RELEASE);
+				extent_dss_extending_finish();
+
+				if (gap_size_page != 0) {
+					extent_dalloc_gap(tsdn, arena, gap);
+				} else {
+					extent_dalloc(tsdn, arena, gap);
+				}
+				if (!*commit) {
+					*commit = pages_decommit(ret, size);
+				}
+				if (*zero && *commit) {
+					extent_hooks_t *extent_hooks =
+					    EXTENT_HOOKS_INITIALIZER;
+					extent_t extent;
+
+					extent_init(&extent, arena, ret, size,
+					    size, false, SC_NSIZES,
+					    extent_state_active, false, true,
+					    true, EXTENT_NOT_HEAD);
+					if (extent_purge_forced_wrapper(tsdn,
+					    arena, &extent_hooks, &extent, 0,
+					    size)) {
+						memset(ret, 0, size);
+					}
+				}
+				return ret;
+			}
+			/*
+			 * Failure, whether due to OOM or a race with a raw
+			 * sbrk() call from outside the allocator.
+			 */
+			if (dss_prev == (void *)-1) {
+				/* OOM. */
+				atomic_store_b(&dss_exhausted, true,
+				    ATOMIC_RELEASE);
+				goto label_oom;
+			}
+		}
+	}
+label_oom:
+	extent_dss_extending_finish();
+	extent_dalloc(tsdn, arena, gap);
+	return NULL;
+}
+
+static bool
+extent_in_dss_helper(void *addr, void *max) {
+	return ((uintptr_t)addr >= (uintptr_t)dss_base && (uintptr_t)addr <
+	    (uintptr_t)max);
+}
+
+bool
+extent_in_dss(void *addr) {
+	cassert(have_dss);
+
+	return extent_in_dss_helper(addr, atomic_load_p(&dss_max,
+	    ATOMIC_ACQUIRE));
+}
+
+bool
+extent_dss_mergeable(void *addr_a, void *addr_b) {
+	void *max;
+
+	cassert(have_dss);
+
+	if ((uintptr_t)addr_a < (uintptr_t)dss_base && (uintptr_t)addr_b <
+	    (uintptr_t)dss_base) {
+		return true;
+	}
+
+	max = atomic_load_p(&dss_max, ATOMIC_ACQUIRE);
+	return (extent_in_dss_helper(addr_a, max) ==
+	    extent_in_dss_helper(addr_b, max));
+}
+
+void
+extent_dss_boot(void) {
+	cassert(have_dss);
+
+	dss_base = extent_dss_sbrk(0);
+	atomic_store_b(&dss_extending, false, ATOMIC_RELAXED);
+	atomic_store_b(&dss_exhausted, dss_base == (void *)-1, ATOMIC_RELAXED);
+	atomic_store_p(&dss_max, dss_base, ATOMIC_RELAXED);
+}
+
+/******************************************************************************/
diff --git a/jemalloc/src/extent_mmap.c b/jemalloc/src/extent_mmap.c
new file mode 100644
index 0000000..17fd1c8
--- /dev/null
+++ b/jemalloc/src/extent_mmap.c
@@ -0,0 +1,42 @@
+#define JEMALLOC_EXTENT_MMAP_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_mmap.h"
+
+/******************************************************************************/
+/* Data. */
+
+bool	opt_retain =
+#ifdef JEMALLOC_RETAIN
+    true
+#else
+    false
+#endif
+    ;
+
+/******************************************************************************/
+
+void *
+extent_alloc_mmap(void *new_addr, size_t size, size_t alignment, bool *zero,
+    bool *commit) {
+	assert(alignment == ALIGNMENT_CEILING(alignment, PAGE));
+	void *ret = pages_map(new_addr, size, alignment, commit);
+	if (ret == NULL) {
+		return NULL;
+	}
+	assert(ret != NULL);
+	if (*commit) {
+		*zero = true;
+	}
+	return ret;
+}
+
+bool
+extent_dalloc_mmap(void *addr, size_t size) {
+	if (!opt_retain) {
+		pages_unmap(addr, size);
+	}
+	return opt_retain;
+}
diff --git a/jemalloc/src/hash.c b/jemalloc/src/hash.c
new file mode 100644
index 0000000..7b2bdc2
--- /dev/null
+++ b/jemalloc/src/hash.c
@@ -0,0 +1,3 @@
+#define JEMALLOC_HASH_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
diff --git a/jemalloc/src/hook.c b/jemalloc/src/hook.c
new file mode 100644
index 0000000..9ac703c
--- /dev/null
+++ b/jemalloc/src/hook.c
@@ -0,0 +1,195 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/hook.h"
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/seq.h"
+
+typedef struct hooks_internal_s hooks_internal_t;
+struct hooks_internal_s {
+	hooks_t hooks;
+	bool in_use;
+};
+
+seq_define(hooks_internal_t, hooks)
+
+static atomic_u_t nhooks = ATOMIC_INIT(0);
+static seq_hooks_t hooks[HOOK_MAX];
+static malloc_mutex_t hooks_mu;
+
+bool
+hook_boot() {
+	return malloc_mutex_init(&hooks_mu, "hooks", WITNESS_RANK_HOOK,
+	    malloc_mutex_rank_exclusive);
+}
+
+static void *
+hook_install_locked(hooks_t *to_install) {
+	hooks_internal_t hooks_internal;
+	for (int i = 0; i < HOOK_MAX; i++) {
+		bool success = seq_try_load_hooks(&hooks_internal, &hooks[i]);
+		/* We hold mu; no concurrent access. */
+		assert(success);
+		if (!hooks_internal.in_use) {
+			hooks_internal.hooks = *to_install;
+			hooks_internal.in_use = true;
+			seq_store_hooks(&hooks[i], &hooks_internal);
+			atomic_store_u(&nhooks,
+			    atomic_load_u(&nhooks, ATOMIC_RELAXED) + 1,
+			    ATOMIC_RELAXED);
+			return &hooks[i];
+		}
+	}
+	return NULL;
+}
+
+void *
+hook_install(tsdn_t *tsdn, hooks_t *to_install) {
+	malloc_mutex_lock(tsdn, &hooks_mu);
+	void *ret = hook_install_locked(to_install);
+	if (ret != NULL) {
+		tsd_global_slow_inc(tsdn);
+	}
+	malloc_mutex_unlock(tsdn, &hooks_mu);
+	return ret;
+}
+
+static void
+hook_remove_locked(seq_hooks_t *to_remove) {
+	hooks_internal_t hooks_internal;
+	bool success = seq_try_load_hooks(&hooks_internal, to_remove);
+	/* We hold mu; no concurrent access. */
+	assert(success);
+	/* Should only remove hooks that were added. */
+	assert(hooks_internal.in_use);
+	hooks_internal.in_use = false;
+	seq_store_hooks(to_remove, &hooks_internal);
+	atomic_store_u(&nhooks, atomic_load_u(&nhooks, ATOMIC_RELAXED) - 1,
+	    ATOMIC_RELAXED);
+}
+
+void
+hook_remove(tsdn_t *tsdn, void *opaque) {
+	if (config_debug) {
+		char *hooks_begin = (char *)&hooks[0];
+		char *hooks_end = (char *)&hooks[HOOK_MAX];
+		char *hook = (char *)opaque;
+		assert(hooks_begin <= hook && hook < hooks_end
+		    && (hook - hooks_begin) % sizeof(seq_hooks_t) == 0);
+	}
+	malloc_mutex_lock(tsdn, &hooks_mu);
+	hook_remove_locked((seq_hooks_t *)opaque);
+	tsd_global_slow_dec(tsdn);
+	malloc_mutex_unlock(tsdn, &hooks_mu);
+}
+
+#define FOR_EACH_HOOK_BEGIN(hooks_internal_ptr)				\
+for (int for_each_hook_counter = 0;					\
+    for_each_hook_counter < HOOK_MAX;					\
+    for_each_hook_counter++) {						\
+	bool for_each_hook_success = seq_try_load_hooks(		\
+	    (hooks_internal_ptr), &hooks[for_each_hook_counter]);	\
+	if (!for_each_hook_success) {					\
+		continue;						\
+	}								\
+	if (!(hooks_internal_ptr)->in_use) {				\
+		continue;						\
+	}
+#define FOR_EACH_HOOK_END						\
+}
+
+static bool *
+hook_reentrantp() {
+	/*
+	 * We prevent user reentrancy within hooks.  This is basically just a
+	 * thread-local bool that triggers an early-exit.
+	 *
+	 * We don't fold in_hook into reentrancy.  There are two reasons for
+	 * this:
+	 * - Right now, we turn on reentrancy during things like extent hook
+	 *   execution.  Allocating during extent hooks is not officially
+	 *   supported, but we don't want to break it for the time being.  These
+	 *   sorts of allocations should probably still be hooked, though.
+	 * - If a hook allocates, we may want it to be relatively fast (after
+	 *   all, it executes on every allocator operation).  Turning on
+	 *   reentrancy is a fairly heavyweight mode (disabling tcache,
+	 *   redirecting to arena 0, etc.).  It's possible we may one day want
+	 *   to turn on reentrant mode here, if it proves too difficult to keep
+	 *   this working.  But that's fairly easy for us to see; OTOH, people
+	 *   not using hooks because they're too slow is easy for us to miss.
+	 *
+	 * The tricky part is
+	 * that this code might get invoked even if we don't have access to tsd.
+	 * This function mimics getting a pointer to thread-local data, except
+	 * that it might secretly return a pointer to some global data if we
+	 * know that the caller will take the early-exit path.
+	 * If we return a bool that indicates that we are reentrant, then the
+	 * caller will go down the early exit path, leaving the global
+	 * untouched.
+	 */
+	static bool in_hook_global = true;
+	tsdn_t *tsdn = tsdn_fetch();
+	tcache_t *tcache = tsdn_tcachep_get(tsdn);
+	if (tcache != NULL) {
+		return &tcache->in_hook;
+	}
+	return &in_hook_global;
+}
+
+#define HOOK_PROLOGUE							\
+	if (likely(atomic_load_u(&nhooks, ATOMIC_RELAXED) == 0)) {	\
+		return;							\
+	}								\
+	bool *in_hook = hook_reentrantp();				\
+	if (*in_hook) {							\
+		return;							\
+	}								\
+	*in_hook = true;
+
+#define HOOK_EPILOGUE							\
+	*in_hook = false;
+
+void
+hook_invoke_alloc(hook_alloc_t type, void *result, uintptr_t result_raw,
+    uintptr_t args_raw[3]) {
+	HOOK_PROLOGUE
+
+	hooks_internal_t hook;
+	FOR_EACH_HOOK_BEGIN(&hook)
+		hook_alloc h = hook.hooks.alloc_hook;
+		if (h != NULL) {
+			h(hook.hooks.extra, type, result, result_raw, args_raw);
+		}
+	FOR_EACH_HOOK_END
+
+	HOOK_EPILOGUE
+}
+
+void
+hook_invoke_dalloc(hook_dalloc_t type, void *address, uintptr_t args_raw[3]) {
+	HOOK_PROLOGUE
+	hooks_internal_t hook;
+	FOR_EACH_HOOK_BEGIN(&hook)
+		hook_dalloc h = hook.hooks.dalloc_hook;
+		if (h != NULL) {
+			h(hook.hooks.extra, type, address, args_raw);
+		}
+	FOR_EACH_HOOK_END
+	HOOK_EPILOGUE
+}
+
+void
+hook_invoke_expand(hook_expand_t type, void *address, size_t old_usize,
+    size_t new_usize, uintptr_t result_raw, uintptr_t args_raw[4]) {
+	HOOK_PROLOGUE
+	hooks_internal_t hook;
+	FOR_EACH_HOOK_BEGIN(&hook)
+		hook_expand h = hook.hooks.expand_hook;
+		if (h != NULL) {
+			h(hook.hooks.extra, type, address, old_usize, new_usize,
+			    result_raw, args_raw);
+		}
+	FOR_EACH_HOOK_END
+	HOOK_EPILOGUE
+}
diff --git a/jemalloc/src/jemalloc.c b/jemalloc/src/jemalloc.c
new file mode 100644
index 0000000..ed13718
--- /dev/null
+++ b/jemalloc/src/jemalloc.c
@@ -0,0 +1,3922 @@
+#define JEMALLOC_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/log.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/spin.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/ticker.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+/* Data. */
+
+/* Runtime configuration options. */
+const char	*je_malloc_conf
+#ifndef _WIN32
+    JEMALLOC_ATTR(weak)
+#endif
+    ;
+bool	opt_abort =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+bool	opt_abort_conf =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+/* Intentionally default off, even with debug builds. */
+bool	opt_confirm_conf = false;
+const char	*opt_junk =
+#if (defined(JEMALLOC_DEBUG) && defined(JEMALLOC_FILL))
+    "true"
+#else
+    "false"
+#endif
+    ;
+bool	opt_junk_alloc =
+#if (defined(JEMALLOC_DEBUG) && defined(JEMALLOC_FILL))
+    true
+#else
+    false
+#endif
+    ;
+bool	opt_junk_free =
+#if (defined(JEMALLOC_DEBUG) && defined(JEMALLOC_FILL))
+    true
+#else
+    false
+#endif
+    ;
+
+bool	opt_utrace = false;
+bool	opt_xmalloc = false;
+bool	opt_zero = false;
+unsigned	opt_narenas = 0;
+
+unsigned	ncpus;
+
+/* Protects arenas initialization. */
+malloc_mutex_t arenas_lock;
+/*
+ * Arenas that are used to service external requests.  Not all elements of the
+ * arenas array are necessarily used; arenas are created lazily as needed.
+ *
+ * arenas[0..narenas_auto) are used for automatic multiplexing of threads and
+ * arenas.  arenas[narenas_auto..narenas_total) are only used if the application
+ * takes some action to create them and allocate from them.
+ *
+ * Points to an arena_t.
+ */
+JEMALLOC_ALIGNED(CACHELINE)
+atomic_p_t		arenas[MALLOCX_ARENA_LIMIT];
+static atomic_u_t	narenas_total; /* Use narenas_total_*(). */
+/* Below three are read-only after initialization. */
+static arena_t		*a0; /* arenas[0]. */
+unsigned		narenas_auto;
+unsigned		manual_arena_base;
+
+typedef enum {
+	malloc_init_uninitialized	= 3,
+	malloc_init_a0_initialized	= 2,
+	malloc_init_recursible		= 1,
+	malloc_init_initialized		= 0 /* Common case --> jnz. */
+} malloc_init_t;
+static malloc_init_t	malloc_init_state = malloc_init_uninitialized;
+
+/* False should be the common case.  Set to true to trigger initialization. */
+bool			malloc_slow = true;
+
+/* When malloc_slow is true, set the corresponding bits for sanity check. */
+enum {
+	flag_opt_junk_alloc	= (1U),
+	flag_opt_junk_free	= (1U << 1),
+	flag_opt_zero		= (1U << 2),
+	flag_opt_utrace		= (1U << 3),
+	flag_opt_xmalloc	= (1U << 4)
+};
+static uint8_t	malloc_slow_flags;
+
+#ifdef JEMALLOC_THREADED_INIT
+/* Used to let the initializing thread recursively allocate. */
+#  define NO_INITIALIZER	((unsigned long)0)
+#  define INITIALIZER		pthread_self()
+#  define IS_INITIALIZER	(malloc_initializer == pthread_self())
+static pthread_t		malloc_initializer = NO_INITIALIZER;
+#else
+#  define NO_INITIALIZER	false
+#  define INITIALIZER		true
+#  define IS_INITIALIZER	malloc_initializer
+static bool			malloc_initializer = NO_INITIALIZER;
+#endif
+
+/* Used to avoid initialization races. */
+#ifdef _WIN32
+#if _WIN32_WINNT >= 0x0600
+static malloc_mutex_t	init_lock = SRWLOCK_INIT;
+#else
+static malloc_mutex_t	init_lock;
+static bool init_lock_initialized = false;
+
+JEMALLOC_ATTR(constructor)
+static void WINAPI
+_init_init_lock(void) {
+	/*
+	 * If another constructor in the same binary is using mallctl to e.g.
+	 * set up extent hooks, it may end up running before this one, and
+	 * malloc_init_hard will crash trying to lock the uninitialized lock. So
+	 * we force an initialization of the lock in malloc_init_hard as well.
+	 * We don't try to care about atomicity of the accessed to the
+	 * init_lock_initialized boolean, since it really only matters early in
+	 * the process creation, before any separate thread normally starts
+	 * doing anything.
+	 */
+	if (!init_lock_initialized) {
+		malloc_mutex_init(&init_lock, "init", WITNESS_RANK_INIT,
+		    malloc_mutex_rank_exclusive);
+	}
+	init_lock_initialized = true;
+}
+
+#ifdef _MSC_VER
+#  pragma section(".CRT$XCU", read)
+JEMALLOC_SECTION(".CRT$XCU") JEMALLOC_ATTR(used)
+static const void (WINAPI *init_init_lock)(void) = _init_init_lock;
+#endif
+#endif
+#else
+static malloc_mutex_t	init_lock = MALLOC_MUTEX_INITIALIZER;
+#endif
+
+typedef struct {
+	void	*p;	/* Input pointer (as in realloc(p, s)). */
+	size_t	s;	/* Request size. */
+	void	*r;	/* Result pointer. */
+} malloc_utrace_t;
+
+#ifdef JEMALLOC_UTRACE
+#  define UTRACE(a, b, c) do {						\
+	if (unlikely(opt_utrace)) {					\
+		int utrace_serrno = errno;				\
+		malloc_utrace_t ut;					\
+		ut.p = (a);						\
+		ut.s = (b);						\
+		ut.r = (c);						\
+		utrace(&ut, sizeof(ut));				\
+		errno = utrace_serrno;					\
+	}								\
+} while (0)
+#else
+#  define UTRACE(a, b, c)
+#endif
+
+/* Whether encountered any invalid config options. */
+static bool had_conf_error = false;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static bool	malloc_init_hard_a0(void);
+static bool	malloc_init_hard(void);
+
+/******************************************************************************/
+/*
+ * Begin miscellaneous support functions.
+ */
+
+bool
+malloc_initialized(void) {
+	return (malloc_init_state == malloc_init_initialized);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+malloc_init_a0(void) {
+	if (unlikely(malloc_init_state == malloc_init_uninitialized)) {
+		return malloc_init_hard_a0();
+	}
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+malloc_init(void) {
+	if (unlikely(!malloc_initialized()) && malloc_init_hard()) {
+		return true;
+	}
+	return false;
+}
+
+/*
+ * The a0*() functions are used instead of i{d,}alloc() in situations that
+ * cannot tolerate TLS variable access.
+ */
+
+static void *
+a0ialloc(size_t size, bool zero, bool is_internal) {
+	if (unlikely(malloc_init_a0())) {
+		return NULL;
+	}
+
+	return iallocztm(TSDN_NULL, size, sz_size2index(size), zero, NULL,
+	    is_internal, arena_get(TSDN_NULL, 0, true), true);
+}
+
+static void
+a0idalloc(void *ptr, bool is_internal) {
+	idalloctm(TSDN_NULL, ptr, NULL, NULL, is_internal, true);
+}
+
+void *
+a0malloc(size_t size) {
+	return a0ialloc(size, false, true);
+}
+
+void
+a0dalloc(void *ptr) {
+	a0idalloc(ptr, true);
+}
+
+/*
+ * FreeBSD's libc uses the bootstrap_*() functions in bootstrap-senstive
+ * situations that cannot tolerate TLS variable access (TLS allocation and very
+ * early internal data structure initialization).
+ */
+
+void *
+bootstrap_malloc(size_t size) {
+	if (unlikely(size == 0)) {
+		size = 1;
+	}
+
+	return a0ialloc(size, false, false);
+}
+
+void *
+bootstrap_calloc(size_t num, size_t size) {
+	size_t num_size;
+
+	num_size = num * size;
+	if (unlikely(num_size == 0)) {
+		assert(num == 0 || size == 0);
+		num_size = 1;
+	}
+
+	return a0ialloc(num_size, true, false);
+}
+
+void
+bootstrap_free(void *ptr) {
+	if (unlikely(ptr == NULL)) {
+		return;
+	}
+
+	a0idalloc(ptr, false);
+}
+
+void
+arena_set(unsigned ind, arena_t *arena) {
+	atomic_store_p(&arenas[ind], arena, ATOMIC_RELEASE);
+}
+
+static void
+narenas_total_set(unsigned narenas) {
+	atomic_store_u(&narenas_total, narenas, ATOMIC_RELEASE);
+}
+
+static void
+narenas_total_inc(void) {
+	atomic_fetch_add_u(&narenas_total, 1, ATOMIC_RELEASE);
+}
+
+unsigned
+narenas_total_get(void) {
+	return atomic_load_u(&narenas_total, ATOMIC_ACQUIRE);
+}
+
+/* Create a new arena and insert it into the arenas array at index ind. */
+static arena_t *
+arena_init_locked(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	arena_t *arena;
+
+	assert(ind <= narenas_total_get());
+	if (ind >= MALLOCX_ARENA_LIMIT) {
+		return NULL;
+	}
+	if (ind == narenas_total_get()) {
+		narenas_total_inc();
+	}
+
+	/*
+	 * Another thread may have already initialized arenas[ind] if it's an
+	 * auto arena.
+	 */
+	arena = arena_get(tsdn, ind, false);
+	if (arena != NULL) {
+		assert(arena_is_auto(arena));
+		return arena;
+	}
+
+	/* Actually initialize the arena. */
+	arena = arena_new(tsdn, ind, extent_hooks);
+
+	return arena;
+}
+
+static void
+arena_new_create_background_thread(tsdn_t *tsdn, unsigned ind) {
+	if (ind == 0) {
+		return;
+	}
+	/*
+	 * Avoid creating a new background thread just for the huge arena, which
+	 * purges eagerly by default.
+	 */
+	if (have_background_thread && !arena_is_huge(ind)) {
+		if (background_thread_create(tsdn_tsd(tsdn), ind)) {
+			malloc_printf("<jemalloc>: error in background thread "
+				      "creation for arena %u. Abort.\n", ind);
+			abort();
+		}
+	}
+}
+
+arena_t *
+arena_init(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	arena_t *arena;
+
+	malloc_mutex_lock(tsdn, &arenas_lock);
+	arena = arena_init_locked(tsdn, ind, extent_hooks);
+	malloc_mutex_unlock(tsdn, &arenas_lock);
+
+	arena_new_create_background_thread(tsdn, ind);
+
+	return arena;
+}
+
+static void
+arena_bind(tsd_t *tsd, unsigned ind, bool internal) {
+	arena_t *arena = arena_get(tsd_tsdn(tsd), ind, false);
+	arena_nthreads_inc(arena, internal);
+
+	if (internal) {
+		tsd_iarena_set(tsd, arena);
+	} else {
+		tsd_arena_set(tsd, arena);
+		unsigned shard = atomic_fetch_add_u(&arena->binshard_next, 1,
+		    ATOMIC_RELAXED);
+		tsd_binshards_t *bins = tsd_binshardsp_get(tsd);
+		for (unsigned i = 0; i < SC_NBINS; i++) {
+			assert(bin_infos[i].n_shards > 0 &&
+			    bin_infos[i].n_shards <= BIN_SHARDS_MAX);
+			bins->binshard[i] = shard % bin_infos[i].n_shards;
+		}
+	}
+}
+
+void
+arena_migrate(tsd_t *tsd, unsigned oldind, unsigned newind) {
+	arena_t *oldarena, *newarena;
+
+	oldarena = arena_get(tsd_tsdn(tsd), oldind, false);
+	newarena = arena_get(tsd_tsdn(tsd), newind, false);
+	arena_nthreads_dec(oldarena, false);
+	arena_nthreads_inc(newarena, false);
+	tsd_arena_set(tsd, newarena);
+}
+
+static void
+arena_unbind(tsd_t *tsd, unsigned ind, bool internal) {
+	arena_t *arena;
+
+	arena = arena_get(tsd_tsdn(tsd), ind, false);
+	arena_nthreads_dec(arena, internal);
+
+	if (internal) {
+		tsd_iarena_set(tsd, NULL);
+	} else {
+		tsd_arena_set(tsd, NULL);
+	}
+}
+
+arena_tdata_t *
+arena_tdata_get_hard(tsd_t *tsd, unsigned ind) {
+	arena_tdata_t *tdata, *arenas_tdata_old;
+	arena_tdata_t *arenas_tdata = tsd_arenas_tdata_get(tsd);
+	unsigned narenas_tdata_old, i;
+	unsigned narenas_tdata = tsd_narenas_tdata_get(tsd);
+	unsigned narenas_actual = narenas_total_get();
+
+	/*
+	 * Dissociate old tdata array (and set up for deallocation upon return)
+	 * if it's too small.
+	 */
+	if (arenas_tdata != NULL && narenas_tdata < narenas_actual) {
+		arenas_tdata_old = arenas_tdata;
+		narenas_tdata_old = narenas_tdata;
+		arenas_tdata = NULL;
+		narenas_tdata = 0;
+		tsd_arenas_tdata_set(tsd, arenas_tdata);
+		tsd_narenas_tdata_set(tsd, narenas_tdata);
+	} else {
+		arenas_tdata_old = NULL;
+		narenas_tdata_old = 0;
+	}
+
+	/* Allocate tdata array if it's missing. */
+	if (arenas_tdata == NULL) {
+		bool *arenas_tdata_bypassp = tsd_arenas_tdata_bypassp_get(tsd);
+		narenas_tdata = (ind < narenas_actual) ? narenas_actual : ind+1;
+
+		if (tsd_nominal(tsd) && !*arenas_tdata_bypassp) {
+			*arenas_tdata_bypassp = true;
+			arenas_tdata = (arena_tdata_t *)a0malloc(
+			    sizeof(arena_tdata_t) * narenas_tdata);
+			*arenas_tdata_bypassp = false;
+		}
+		if (arenas_tdata == NULL) {
+			tdata = NULL;
+			goto label_return;
+		}
+		assert(tsd_nominal(tsd) && !*arenas_tdata_bypassp);
+		tsd_arenas_tdata_set(tsd, arenas_tdata);
+		tsd_narenas_tdata_set(tsd, narenas_tdata);
+	}
+
+	/*
+	 * Copy to tdata array.  It's possible that the actual number of arenas
+	 * has increased since narenas_total_get() was called above, but that
+	 * causes no correctness issues unless two threads concurrently execute
+	 * the arenas.create mallctl, which we trust mallctl synchronization to
+	 * prevent.
+	 */
+
+	/* Copy/initialize tickers. */
+	for (i = 0; i < narenas_actual; i++) {
+		if (i < narenas_tdata_old) {
+			ticker_copy(&arenas_tdata[i].decay_ticker,
+			    &arenas_tdata_old[i].decay_ticker);
+		} else {
+			ticker_init(&arenas_tdata[i].decay_ticker,
+			    DECAY_NTICKS_PER_UPDATE);
+		}
+	}
+	if (narenas_tdata > narenas_actual) {
+		memset(&arenas_tdata[narenas_actual], 0, sizeof(arena_tdata_t)
+		    * (narenas_tdata - narenas_actual));
+	}
+
+	/* Read the refreshed tdata array. */
+	tdata = &arenas_tdata[ind];
+label_return:
+	if (arenas_tdata_old != NULL) {
+		a0dalloc(arenas_tdata_old);
+	}
+	return tdata;
+}
+
+/* Slow path, called only by arena_choose(). */
+arena_t *
+arena_choose_hard(tsd_t *tsd, bool internal) {
+	arena_t *ret JEMALLOC_CC_SILENCE_INIT(NULL);
+
+	if (have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena)) {
+		unsigned choose = percpu_arena_choose();
+		ret = arena_get(tsd_tsdn(tsd), choose, true);
+		assert(ret != NULL);
+		arena_bind(tsd, arena_ind_get(ret), false);
+		arena_bind(tsd, arena_ind_get(ret), true);
+
+		return ret;
+	}
+
+	if (narenas_auto > 1) {
+		unsigned i, j, choose[2], first_null;
+		bool is_new_arena[2];
+
+		/*
+		 * Determine binding for both non-internal and internal
+		 * allocation.
+		 *
+		 *   choose[0]: For application allocation.
+		 *   choose[1]: For internal metadata allocation.
+		 */
+
+		for (j = 0; j < 2; j++) {
+			choose[j] = 0;
+			is_new_arena[j] = false;
+		}
+
+		first_null = narenas_auto;
+		malloc_mutex_lock(tsd_tsdn(tsd), &arenas_lock);
+		assert(arena_get(tsd_tsdn(tsd), 0, false) != NULL);
+		for (i = 1; i < narenas_auto; i++) {
+			if (arena_get(tsd_tsdn(tsd), i, false) != NULL) {
+				/*
+				 * Choose the first arena that has the lowest
+				 * number of threads assigned to it.
+				 */
+				for (j = 0; j < 2; j++) {
+					if (arena_nthreads_get(arena_get(
+					    tsd_tsdn(tsd), i, false), !!j) <
+					    arena_nthreads_get(arena_get(
+					    tsd_tsdn(tsd), choose[j], false),
+					    !!j)) {
+						choose[j] = i;
+					}
+				}
+			} else if (first_null == narenas_auto) {
+				/*
+				 * Record the index of the first uninitialized
+				 * arena, in case all extant arenas are in use.
+				 *
+				 * NB: It is possible for there to be
+				 * discontinuities in terms of initialized
+				 * versus uninitialized arenas, due to the
+				 * "thread.arena" mallctl.
+				 */
+				first_null = i;
+			}
+		}
+
+		for (j = 0; j < 2; j++) {
+			if (arena_nthreads_get(arena_get(tsd_tsdn(tsd),
+			    choose[j], false), !!j) == 0 || first_null ==
+			    narenas_auto) {
+				/*
+				 * Use an unloaded arena, or the least loaded
+				 * arena if all arenas are already initialized.
+				 */
+				if (!!j == internal) {
+					ret = arena_get(tsd_tsdn(tsd),
+					    choose[j], false);
+				}
+			} else {
+				arena_t *arena;
+
+				/* Initialize a new arena. */
+				choose[j] = first_null;
+				arena = arena_init_locked(tsd_tsdn(tsd),
+				    choose[j],
+				    (extent_hooks_t *)&extent_hooks_default);
+				if (arena == NULL) {
+					malloc_mutex_unlock(tsd_tsdn(tsd),
+					    &arenas_lock);
+					return NULL;
+				}
+				is_new_arena[j] = true;
+				if (!!j == internal) {
+					ret = arena;
+				}
+			}
+			arena_bind(tsd, choose[j], !!j);
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &arenas_lock);
+
+		for (j = 0; j < 2; j++) {
+			if (is_new_arena[j]) {
+				assert(choose[j] > 0);
+				arena_new_create_background_thread(
+				    tsd_tsdn(tsd), choose[j]);
+			}
+		}
+
+	} else {
+		ret = arena_get(tsd_tsdn(tsd), 0, false);
+		arena_bind(tsd, 0, false);
+		arena_bind(tsd, 0, true);
+	}
+
+	return ret;
+}
+
+void
+iarena_cleanup(tsd_t *tsd) {
+	arena_t *iarena;
+
+	iarena = tsd_iarena_get(tsd);
+	if (iarena != NULL) {
+		arena_unbind(tsd, arena_ind_get(iarena), true);
+	}
+}
+
+void
+arena_cleanup(tsd_t *tsd) {
+	arena_t *arena;
+
+	arena = tsd_arena_get(tsd);
+	if (arena != NULL) {
+		arena_unbind(tsd, arena_ind_get(arena), false);
+	}
+}
+
+void
+arenas_tdata_cleanup(tsd_t *tsd) {
+	arena_tdata_t *arenas_tdata;
+
+	/* Prevent tsd->arenas_tdata from being (re)created. */
+	*tsd_arenas_tdata_bypassp_get(tsd) = true;
+
+	arenas_tdata = tsd_arenas_tdata_get(tsd);
+	if (arenas_tdata != NULL) {
+		tsd_arenas_tdata_set(tsd, NULL);
+		a0dalloc(arenas_tdata);
+	}
+}
+
+static void
+stats_print_atexit(void) {
+	if (config_stats) {
+		tsdn_t *tsdn;
+		unsigned narenas, i;
+
+		tsdn = tsdn_fetch();
+
+		/*
+		 * Merge stats from extant threads.  This is racy, since
+		 * individual threads do not lock when recording tcache stats
+		 * events.  As a consequence, the final stats may be slightly
+		 * out of date by the time they are reported, if other threads
+		 * continue to allocate.
+		 */
+		for (i = 0, narenas = narenas_total_get(); i < narenas; i++) {
+			arena_t *arena = arena_get(tsdn, i, false);
+			if (arena != NULL) {
+				tcache_t *tcache;
+
+				malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+				ql_foreach(tcache, &arena->tcache_ql, link) {
+					tcache_stats_merge(tsdn, tcache, arena);
+				}
+				malloc_mutex_unlock(tsdn,
+				    &arena->tcache_ql_mtx);
+			}
+		}
+	}
+	je_malloc_stats_print(NULL, NULL, opt_stats_print_opts);
+}
+
+/*
+ * Ensure that we don't hold any locks upon entry to or exit from allocator
+ * code (in a "broad" sense that doesn't count a reentrant allocation as an
+ * entrance or exit).
+ */
+JEMALLOC_ALWAYS_INLINE void
+check_entry_exit_locking(tsdn_t *tsdn) {
+	if (!config_debug) {
+		return;
+	}
+	if (tsdn_null(tsdn)) {
+		return;
+	}
+	tsd_t *tsd = tsdn_tsd(tsdn);
+	/*
+	 * It's possible we hold locks at entry/exit if we're in a nested
+	 * allocation.
+	 */
+	int8_t reentrancy_level = tsd_reentrancy_level_get(tsd);
+	if (reentrancy_level != 0) {
+		return;
+	}
+	witness_assert_lockless(tsdn_witness_tsdp_get(tsdn));
+}
+
+/*
+ * End miscellaneous support functions.
+ */
+/******************************************************************************/
+/*
+ * Begin initialization functions.
+ */
+
+static char *
+jemalloc_secure_getenv(const char *name) {
+#ifdef JEMALLOC_HAVE_SECURE_GETENV
+	return secure_getenv(name);
+#else
+#  ifdef JEMALLOC_HAVE_ISSETUGID
+	if (issetugid() != 0) {
+		return NULL;
+	}
+#  endif
+	return getenv(name);
+#endif
+}
+
+static unsigned
+malloc_ncpus(void) {
+	long result;
+
+#ifdef _WIN32
+	SYSTEM_INFO si;
+	GetSystemInfo(&si);
+	result = si.dwNumberOfProcessors;
+#elif defined(JEMALLOC_GLIBC_MALLOC_HOOK) && defined(CPU_COUNT)
+	/*
+	 * glibc >= 2.6 has the CPU_COUNT macro.
+	 *
+	 * glibc's sysconf() uses isspace().  glibc allocates for the first time
+	 * *before* setting up the isspace tables.  Therefore we need a
+	 * different method to get the number of CPUs.
+	 */
+	{
+		cpu_set_t set;
+
+		pthread_getaffinity_np(pthread_self(), sizeof(set), &set);
+		result = CPU_COUNT(&set);
+	}
+#else
+	result = sysconf(_SC_NPROCESSORS_ONLN);
+#endif
+	return ((result == -1) ? 1 : (unsigned)result);
+}
+
+static void
+init_opt_stats_print_opts(const char *v, size_t vlen) {
+	size_t opts_len = strlen(opt_stats_print_opts);
+	assert(opts_len <= stats_print_tot_num_options);
+
+	for (size_t i = 0; i < vlen; i++) {
+		switch (v[i]) {
+#define OPTION(o, v, d, s) case o: break;
+			STATS_PRINT_OPTIONS
+#undef OPTION
+		default: continue;
+		}
+
+		if (strchr(opt_stats_print_opts, v[i]) != NULL) {
+			/* Ignore repeated. */
+			continue;
+		}
+
+		opt_stats_print_opts[opts_len++] = v[i];
+		opt_stats_print_opts[opts_len] = '\0';
+		assert(opts_len <= stats_print_tot_num_options);
+	}
+	assert(opts_len == strlen(opt_stats_print_opts));
+}
+
+/* Reads the next size pair in a multi-sized option. */
+static bool
+malloc_conf_multi_sizes_next(const char **slab_size_segment_cur,
+    size_t *vlen_left, size_t *slab_start, size_t *slab_end, size_t *new_size) {
+	const char *cur = *slab_size_segment_cur;
+	char *end;
+	uintmax_t um;
+
+	set_errno(0);
+
+	/* First number, then '-' */
+	um = malloc_strtoumax(cur, &end, 0);
+	if (get_errno() != 0 || *end != '-') {
+		return true;
+	}
+	*slab_start = (size_t)um;
+	cur = end + 1;
+
+	/* Second number, then ':' */
+	um = malloc_strtoumax(cur, &end, 0);
+	if (get_errno() != 0 || *end != ':') {
+		return true;
+	}
+	*slab_end = (size_t)um;
+	cur = end + 1;
+
+	/* Last number */
+	um = malloc_strtoumax(cur, &end, 0);
+	if (get_errno() != 0) {
+		return true;
+	}
+	*new_size = (size_t)um;
+
+	/* Consume the separator if there is one. */
+	if (*end == '|') {
+		end++;
+	}
+
+	*vlen_left -= end - *slab_size_segment_cur;
+	*slab_size_segment_cur = end;
+
+	return false;
+}
+
+static bool
+malloc_conf_next(char const **opts_p, char const **k_p, size_t *klen_p,
+    char const **v_p, size_t *vlen_p) {
+	bool accept;
+	const char *opts = *opts_p;
+
+	*k_p = opts;
+
+	for (accept = false; !accept;) {
+		switch (*opts) {
+		case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
+		case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
+		case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
+		case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
+		case 'Y': case 'Z':
+		case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
+		case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
+		case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
+		case 's': case 't': case 'u': case 'v': case 'w': case 'x':
+		case 'y': case 'z':
+		case '0': case '1': case '2': case '3': case '4': case '5':
+		case '6': case '7': case '8': case '9':
+		case '_':
+			opts++;
+			break;
+		case ':':
+			opts++;
+			*klen_p = (uintptr_t)opts - 1 - (uintptr_t)*k_p;
+			*v_p = opts;
+			accept = true;
+			break;
+		case '\0':
+			if (opts != *opts_p) {
+				malloc_write("<jemalloc>: Conf string ends "
+				    "with key\n");
+			}
+			return true;
+		default:
+			malloc_write("<jemalloc>: Malformed conf string\n");
+			return true;
+		}
+	}
+
+	for (accept = false; !accept;) {
+		switch (*opts) {
+		case ',':
+			opts++;
+			/*
+			 * Look ahead one character here, because the next time
+			 * this function is called, it will assume that end of
+			 * input has been cleanly reached if no input remains,
+			 * but we have optimistically already consumed the
+			 * comma if one exists.
+			 */
+			if (*opts == '\0') {
+				malloc_write("<jemalloc>: Conf string ends "
+				    "with comma\n");
+			}
+			*vlen_p = (uintptr_t)opts - 1 - (uintptr_t)*v_p;
+			accept = true;
+			break;
+		case '\0':
+			*vlen_p = (uintptr_t)opts - (uintptr_t)*v_p;
+			accept = true;
+			break;
+		default:
+			opts++;
+			break;
+		}
+	}
+
+	*opts_p = opts;
+	return false;
+}
+
+static void
+malloc_abort_invalid_conf(void) {
+	assert(opt_abort_conf);
+	malloc_printf("<jemalloc>: Abort (abort_conf:true) on invalid conf "
+	    "value (see above).\n");
+	abort();
+}
+
+static void
+malloc_conf_error(const char *msg, const char *k, size_t klen, const char *v,
+    size_t vlen) {
+	malloc_printf("<jemalloc>: %s: %.*s:%.*s\n", msg, (int)klen, k,
+	    (int)vlen, v);
+	/* If abort_conf is set, error out after processing all options. */
+	const char *experimental = "experimental_";
+	if (strncmp(k, experimental, strlen(experimental)) == 0) {
+		/* However, tolerate experimental features. */
+		return;
+	}
+	had_conf_error = true;
+}
+
+static void
+malloc_slow_flag_init(void) {
+	/*
+	 * Combine the runtime options into malloc_slow for fast path.  Called
+	 * after processing all the options.
+	 */
+	malloc_slow_flags |= (opt_junk_alloc ? flag_opt_junk_alloc : 0)
+	    | (opt_junk_free ? flag_opt_junk_free : 0)
+	    | (opt_zero ? flag_opt_zero : 0)
+	    | (opt_utrace ? flag_opt_utrace : 0)
+	    | (opt_xmalloc ? flag_opt_xmalloc : 0);
+
+	malloc_slow = (malloc_slow_flags != 0);
+}
+
+/* Number of sources for initializing malloc_conf */
+#define MALLOC_CONF_NSOURCES 4
+
+static const char *
+obtain_malloc_conf(unsigned which_source, char buf[PATH_MAX + 1]) {
+	if (config_debug) {
+		static unsigned read_source = 0;
+		/*
+		 * Each source should only be read once, to minimize # of
+		 * syscalls on init.
+		 */
+		assert(read_source++ == which_source);
+	}
+	assert(which_source < MALLOC_CONF_NSOURCES);
+
+	const char *ret;
+	switch (which_source) {
+	case 0:
+		ret = config_malloc_conf;
+		break;
+	case 1:
+		if (je_malloc_conf != NULL) {
+			/* Use options that were compiled into the program. */
+			ret = je_malloc_conf;
+		} else {
+			/* No configuration specified. */
+			ret = NULL;
+		}
+		break;
+	case 2: {
+		ssize_t linklen = 0;
+#ifndef _WIN32
+		int saved_errno = errno;
+		const char *linkname =
+#  ifdef JEMALLOC_PREFIX
+		    "/etc/"JEMALLOC_PREFIX"malloc.conf"
+#  else
+		    "/etc/malloc.conf"
+#  endif
+		    ;
+
+		/*
+		 * Try to use the contents of the "/etc/malloc.conf" symbolic
+		 * link's name.
+		 */
+#ifndef JEMALLOC_READLINKAT
+		linklen = readlink(linkname, buf, PATH_MAX);
+#else
+		linklen = readlinkat(AT_FDCWD, linkname, buf, PATH_MAX);
+#endif
+		if (linklen == -1) {
+			/* No configuration specified. */
+			linklen = 0;
+			/* Restore errno. */
+			set_errno(saved_errno);
+		}
+#endif
+		buf[linklen] = '\0';
+		ret = buf;
+		break;
+	} case 3: {
+		const char *envname =
+#ifdef JEMALLOC_PREFIX
+		    JEMALLOC_CPREFIX"MALLOC_CONF"
+#else
+		    "MALLOC_CONF"
+#endif
+		    ;
+
+		if ((ret = jemalloc_secure_getenv(envname)) != NULL) {
+			/*
+			 * Do nothing; opts is already initialized to the value
+			 * of the MALLOC_CONF environment variable.
+			 */
+		} else {
+			/* No configuration specified. */
+			ret = NULL;
+		}
+		break;
+	} default:
+		not_reached();
+		ret = NULL;
+	}
+	return ret;
+}
+
+static void
+malloc_conf_init_helper(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS],
+    bool initial_call, const char *opts_cache[MALLOC_CONF_NSOURCES],
+    char buf[PATH_MAX + 1]) {
+	static const char *opts_explain[MALLOC_CONF_NSOURCES] = {
+		"string specified via --with-malloc-conf",
+		"string pointed to by the global variable malloc_conf",
+		"\"name\" of the file referenced by the symbolic link named "
+		    "/etc/malloc.conf",
+		"value of the environment variable MALLOC_CONF"
+	};
+	unsigned i;
+	const char *opts, *k, *v;
+	size_t klen, vlen;
+
+	for (i = 0; i < MALLOC_CONF_NSOURCES; i++) {
+		/* Get runtime configuration. */
+		if (initial_call) {
+			opts_cache[i] = obtain_malloc_conf(i, buf);
+		}
+		opts = opts_cache[i];
+		if (!initial_call && opt_confirm_conf) {
+			malloc_printf(
+			    "<jemalloc>: malloc_conf #%u (%s): \"%s\"\n",
+			    i + 1, opts_explain[i], opts != NULL ? opts : "");
+		}
+		if (opts == NULL) {
+			continue;
+		}
+
+		while (*opts != '\0' && !malloc_conf_next(&opts, &k, &klen, &v,
+		    &vlen)) {
+
+#define CONF_ERROR(msg, k, klen, v, vlen)				\
+			if (!initial_call) {				\
+				malloc_conf_error(			\
+				    msg, k, klen, v, vlen);		\
+				cur_opt_valid = false;			\
+			}
+#define CONF_CONTINUE	{						\
+				if (!initial_call && opt_confirm_conf	\
+				    && cur_opt_valid) {			\
+					malloc_printf("<jemalloc>: -- "	\
+					    "Set conf value: %.*s:%.*s"	\
+					    "\n", (int)klen, k,		\
+					    (int)vlen, v);		\
+				}					\
+				continue;				\
+			}
+#define CONF_MATCH(n)							\
+	(sizeof(n)-1 == klen && strncmp(n, k, klen) == 0)
+#define CONF_MATCH_VALUE(n)						\
+	(sizeof(n)-1 == vlen && strncmp(n, v, vlen) == 0)
+#define CONF_HANDLE_BOOL(o, n)						\
+			if (CONF_MATCH(n)) {				\
+				if (CONF_MATCH_VALUE("true")) {		\
+					o = true;			\
+				} else if (CONF_MATCH_VALUE("false")) {	\
+					o = false;			\
+				} else {				\
+					CONF_ERROR("Invalid conf value",\
+					    k, klen, v, vlen);		\
+				}					\
+				CONF_CONTINUE;				\
+			}
+      /*
+       * One of the CONF_MIN macros below expands, in one of the use points,
+       * to "unsigned integer < 0", which is always false, triggering the
+       * GCC -Wtype-limits warning, which we disable here and re-enable below.
+       */
+      JEMALLOC_DIAGNOSTIC_PUSH
+      JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS
+
+#define CONF_DONT_CHECK_MIN(um, min)	false
+#define CONF_CHECK_MIN(um, min)	((um) < (min))
+#define CONF_DONT_CHECK_MAX(um, max)	false
+#define CONF_CHECK_MAX(um, max)	((um) > (max))
+#define CONF_HANDLE_T_U(t, o, n, min, max, check_min, check_max, clip)	\
+			if (CONF_MATCH(n)) {				\
+				uintmax_t um;				\
+				char *end;				\
+									\
+				set_errno(0);				\
+				um = malloc_strtoumax(v, &end, 0);	\
+				if (get_errno() != 0 || (uintptr_t)end -\
+				    (uintptr_t)v != vlen) {		\
+					CONF_ERROR("Invalid conf value",\
+					    k, klen, v, vlen);		\
+				} else if (clip) {			\
+					if (check_min(um, (t)(min))) {	\
+						o = (t)(min);		\
+					} else if (			\
+					    check_max(um, (t)(max))) {	\
+						o = (t)(max);		\
+					} else {			\
+						o = (t)um;		\
+					}				\
+				} else {				\
+					if (check_min(um, (t)(min)) ||	\
+					    check_max(um, (t)(max))) {	\
+						CONF_ERROR(		\
+						    "Out-of-range "	\
+						    "conf value",	\
+						    k, klen, v, vlen);	\
+					} else {			\
+						o = (t)um;		\
+					}				\
+				}					\
+				CONF_CONTINUE;				\
+			}
+#define CONF_HANDLE_UNSIGNED(o, n, min, max, check_min, check_max,	\
+    clip)								\
+			CONF_HANDLE_T_U(unsigned, o, n, min, max,	\
+			    check_min, check_max, clip)
+#define CONF_HANDLE_SIZE_T(o, n, min, max, check_min, check_max, clip)	\
+			CONF_HANDLE_T_U(size_t, o, n, min, max,		\
+			    check_min, check_max, clip)
+#define CONF_HANDLE_SSIZE_T(o, n, min, max)				\
+			if (CONF_MATCH(n)) {				\
+				long l;					\
+				char *end;				\
+									\
+				set_errno(0);				\
+				l = strtol(v, &end, 0);			\
+				if (get_errno() != 0 || (uintptr_t)end -\
+				    (uintptr_t)v != vlen) {		\
+					CONF_ERROR("Invalid conf value",\
+					    k, klen, v, vlen);		\
+				} else if (l < (ssize_t)(min) || l >	\
+				    (ssize_t)(max)) {			\
+					CONF_ERROR(			\
+					    "Out-of-range conf value",	\
+					    k, klen, v, vlen);		\
+				} else {				\
+					o = l;				\
+				}					\
+				CONF_CONTINUE;				\
+			}
+#define CONF_HANDLE_CHAR_P(o, n, d)					\
+			if (CONF_MATCH(n)) {				\
+				size_t cpylen = (vlen <=		\
+				    sizeof(o)-1) ? vlen :		\
+				    sizeof(o)-1;			\
+				strncpy(o, v, cpylen);			\
+				o[cpylen] = '\0';			\
+				CONF_CONTINUE;				\
+			}
+
+			bool cur_opt_valid = true;
+
+			CONF_HANDLE_BOOL(opt_confirm_conf, "confirm_conf")
+			if (initial_call) {
+				continue;
+			}
+
+			CONF_HANDLE_BOOL(opt_abort, "abort")
+			CONF_HANDLE_BOOL(opt_abort_conf, "abort_conf")
+			if (strncmp("metadata_thp", k, klen) == 0) {
+				int i;
+				bool match = false;
+				for (i = 0; i < metadata_thp_mode_limit; i++) {
+					if (strncmp(metadata_thp_mode_names[i],
+					    v, vlen) == 0) {
+						opt_metadata_thp = i;
+						match = true;
+						break;
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_BOOL(opt_retain, "retain")
+			if (strncmp("dss", k, klen) == 0) {
+				int i;
+				bool match = false;
+				for (i = 0; i < dss_prec_limit; i++) {
+					if (strncmp(dss_prec_names[i], v, vlen)
+					    == 0) {
+						if (extent_dss_prec_set(i)) {
+							CONF_ERROR(
+							    "Error setting dss",
+							    k, klen, v, vlen);
+						} else {
+							opt_dss =
+							    dss_prec_names[i];
+							match = true;
+							break;
+						}
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_UNSIGNED(opt_narenas, "narenas", 1,
+			    UINT_MAX, CONF_CHECK_MIN, CONF_DONT_CHECK_MAX,
+			    false)
+			if (CONF_MATCH("bin_shards")) {
+				const char *bin_shards_segment_cur = v;
+				size_t vlen_left = vlen;
+				do {
+					size_t size_start;
+					size_t size_end;
+					size_t nshards;
+					bool err = malloc_conf_multi_sizes_next(
+					    &bin_shards_segment_cur, &vlen_left,
+					    &size_start, &size_end, &nshards);
+					if (err || bin_update_shard_size(
+					    bin_shard_sizes, size_start,
+					    size_end, nshards)) {
+						CONF_ERROR(
+						    "Invalid settings for "
+						    "bin_shards", k, klen, v,
+						    vlen);
+						break;
+					}
+				} while (vlen_left > 0);
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_SSIZE_T(opt_dirty_decay_ms,
+			    "dirty_decay_ms", -1, NSTIME_SEC_MAX * KQU(1000) <
+			    QU(SSIZE_MAX) ? NSTIME_SEC_MAX * KQU(1000) :
+			    SSIZE_MAX);
+			CONF_HANDLE_SSIZE_T(opt_muzzy_decay_ms,
+			    "muzzy_decay_ms", -1, NSTIME_SEC_MAX * KQU(1000) <
+			    QU(SSIZE_MAX) ? NSTIME_SEC_MAX * KQU(1000) :
+			    SSIZE_MAX);
+			CONF_HANDLE_BOOL(opt_stats_print, "stats_print")
+			if (CONF_MATCH("stats_print_opts")) {
+				init_opt_stats_print_opts(v, vlen);
+				CONF_CONTINUE;
+			}
+			if (config_fill) {
+				if (CONF_MATCH("junk")) {
+					if (CONF_MATCH_VALUE("true")) {
+						opt_junk = "true";
+						opt_junk_alloc = opt_junk_free =
+						    true;
+					} else if (CONF_MATCH_VALUE("false")) {
+						opt_junk = "false";
+						opt_junk_alloc = opt_junk_free =
+						    false;
+					} else if (CONF_MATCH_VALUE("alloc")) {
+						opt_junk = "alloc";
+						opt_junk_alloc = true;
+						opt_junk_free = false;
+					} else if (CONF_MATCH_VALUE("free")) {
+						opt_junk = "free";
+						opt_junk_alloc = false;
+						opt_junk_free = true;
+					} else {
+						CONF_ERROR(
+						    "Invalid conf value",
+						    k, klen, v, vlen);
+					}
+					CONF_CONTINUE;
+				}
+				CONF_HANDLE_BOOL(opt_zero, "zero")
+			}
+			if (config_utrace) {
+				CONF_HANDLE_BOOL(opt_utrace, "utrace")
+			}
+			if (config_xmalloc) {
+				CONF_HANDLE_BOOL(opt_xmalloc, "xmalloc")
+			}
+			CONF_HANDLE_BOOL(opt_tcache, "tcache")
+			CONF_HANDLE_SSIZE_T(opt_lg_tcache_max, "lg_tcache_max",
+			    -1, (sizeof(size_t) << 3) - 1)
+
+			/*
+			 * The runtime option of oversize_threshold remains
+			 * undocumented.  It may be tweaked in the next major
+			 * release (6.0).  The default value 8M is rather
+			 * conservative / safe.  Tuning it further down may
+			 * improve fragmentation a bit more, but may also cause
+			 * contention on the huge arena.
+			 */
+			CONF_HANDLE_SIZE_T(opt_oversize_threshold,
+			    "oversize_threshold", 0, SC_LARGE_MAXCLASS,
+			    CONF_DONT_CHECK_MIN, CONF_CHECK_MAX, false)
+			CONF_HANDLE_SIZE_T(opt_lg_extent_max_active_fit,
+			    "lg_extent_max_active_fit", 0,
+			    (sizeof(size_t) << 3), CONF_DONT_CHECK_MIN,
+			    CONF_CHECK_MAX, false)
+
+			if (strncmp("percpu_arena", k, klen) == 0) {
+				bool match = false;
+				for (int i = percpu_arena_mode_names_base; i <
+				    percpu_arena_mode_names_limit; i++) {
+					if (strncmp(percpu_arena_mode_names[i],
+					    v, vlen) == 0) {
+						if (!have_percpu_arena) {
+							CONF_ERROR(
+							    "No getcpu support",
+							    k, klen, v, vlen);
+						}
+						opt_percpu_arena = i;
+						match = true;
+						break;
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_BOOL(opt_background_thread,
+			    "background_thread");
+			CONF_HANDLE_SIZE_T(opt_max_background_threads,
+					   "max_background_threads", 1,
+					   opt_max_background_threads,
+					   CONF_CHECK_MIN, CONF_CHECK_MAX,
+					   true);
+			if (CONF_MATCH("slab_sizes")) {
+				bool err;
+				const char *slab_size_segment_cur = v;
+				size_t vlen_left = vlen;
+				do {
+					size_t slab_start;
+					size_t slab_end;
+					size_t pgs;
+					err = malloc_conf_multi_sizes_next(
+					    &slab_size_segment_cur,
+					    &vlen_left, &slab_start, &slab_end,
+					    &pgs);
+					if (!err) {
+						sc_data_update_slab_size(
+						    sc_data, slab_start,
+						    slab_end, (int)pgs);
+					} else {
+						CONF_ERROR("Invalid settings "
+						    "for slab_sizes",
+						    k, klen, v, vlen);
+					}
+				} while (!err && vlen_left > 0);
+				CONF_CONTINUE;
+			}
+			if (config_prof) {
+				CONF_HANDLE_BOOL(opt_prof, "prof")
+				CONF_HANDLE_CHAR_P(opt_prof_prefix,
+				    "prof_prefix", "jeprof")
+				CONF_HANDLE_BOOL(opt_prof_active, "prof_active")
+				CONF_HANDLE_BOOL(opt_prof_thread_active_init,
+				    "prof_thread_active_init")
+				CONF_HANDLE_SIZE_T(opt_lg_prof_sample,
+				    "lg_prof_sample", 0, (sizeof(uint64_t) << 3)
+				    - 1, CONF_DONT_CHECK_MIN, CONF_CHECK_MAX,
+				    true)
+				CONF_HANDLE_BOOL(opt_prof_accum, "prof_accum")
+				CONF_HANDLE_SSIZE_T(opt_lg_prof_interval,
+				    "lg_prof_interval", -1,
+				    (sizeof(uint64_t) << 3) - 1)
+				CONF_HANDLE_BOOL(opt_prof_gdump, "prof_gdump")
+				CONF_HANDLE_BOOL(opt_prof_final, "prof_final")
+				CONF_HANDLE_BOOL(opt_prof_leak, "prof_leak")
+				CONF_HANDLE_BOOL(opt_prof_log, "prof_log")
+			}
+			if (config_log) {
+				if (CONF_MATCH("log")) {
+					size_t cpylen = (
+					    vlen <= sizeof(log_var_names) ?
+					    vlen : sizeof(log_var_names) - 1);
+					strncpy(log_var_names, v, cpylen);
+					log_var_names[cpylen] = '\0';
+					CONF_CONTINUE;
+				}
+			}
+			if (CONF_MATCH("thp")) {
+				bool match = false;
+				for (int i = 0; i < thp_mode_names_limit; i++) {
+					if (strncmp(thp_mode_names[i],v, vlen)
+					    == 0) {
+						if (!have_madvise_huge) {
+							CONF_ERROR(
+							    "No THP support",
+							    k, klen, v, vlen);
+						}
+						opt_thp = i;
+						match = true;
+						break;
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_ERROR("Invalid conf pair", k, klen, v, vlen);
+#undef CONF_ERROR
+#undef CONF_CONTINUE
+#undef CONF_MATCH
+#undef CONF_MATCH_VALUE
+#undef CONF_HANDLE_BOOL
+#undef CONF_DONT_CHECK_MIN
+#undef CONF_CHECK_MIN
+#undef CONF_DONT_CHECK_MAX
+#undef CONF_CHECK_MAX
+#undef CONF_HANDLE_T_U
+#undef CONF_HANDLE_UNSIGNED
+#undef CONF_HANDLE_SIZE_T
+#undef CONF_HANDLE_SSIZE_T
+#undef CONF_HANDLE_CHAR_P
+    /* Re-enable diagnostic "-Wtype-limits" */
+    JEMALLOC_DIAGNOSTIC_POP
+		}
+		if (opt_abort_conf && had_conf_error) {
+			malloc_abort_invalid_conf();
+		}
+	}
+	atomic_store_b(&log_init_done, true, ATOMIC_RELEASE);
+}
+
+static void
+malloc_conf_init(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS]) {
+	const char *opts_cache[MALLOC_CONF_NSOURCES] = {NULL, NULL, NULL, NULL};
+	char buf[PATH_MAX + 1];
+
+	/* The first call only set the confirm_conf option and opts_cache */
+	malloc_conf_init_helper(NULL, NULL, true, opts_cache, buf);
+	malloc_conf_init_helper(sc_data, bin_shard_sizes, false, opts_cache,
+	    NULL);
+}
+
+#undef MALLOC_CONF_NSOURCES
+
+static bool
+malloc_init_hard_needed(void) {
+	if (malloc_initialized() || (IS_INITIALIZER && malloc_init_state ==
+	    malloc_init_recursible)) {
+		/*
+		 * Another thread initialized the allocator before this one
+		 * acquired init_lock, or this thread is the initializing
+		 * thread, and it is recursively allocating.
+		 */
+		return false;
+	}
+#ifdef JEMALLOC_THREADED_INIT
+	if (malloc_initializer != NO_INITIALIZER && !IS_INITIALIZER) {
+		/* Busy-wait until the initializing thread completes. */
+		spin_t spinner = SPIN_INITIALIZER;
+		do {
+			malloc_mutex_unlock(TSDN_NULL, &init_lock);
+			spin_adaptive(&spinner);
+			malloc_mutex_lock(TSDN_NULL, &init_lock);
+		} while (!malloc_initialized());
+		return false;
+	}
+#endif
+	return true;
+}
+
+static bool
+malloc_init_hard_a0_locked() {
+	malloc_initializer = INITIALIZER;
+
+	JEMALLOC_DIAGNOSTIC_PUSH
+	JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+	sc_data_t sc_data = {0};
+	JEMALLOC_DIAGNOSTIC_POP
+
+	/*
+	 * Ordering here is somewhat tricky; we need sc_boot() first, since that
+	 * determines what the size classes will be, and then
+	 * malloc_conf_init(), since any slab size tweaking will need to be done
+	 * before sz_boot and bin_boot, which assume that the values they read
+	 * out of sc_data_global are final.
+	 */
+	sc_boot(&sc_data);
+	unsigned bin_shard_sizes[SC_NBINS];
+	bin_shard_sizes_boot(bin_shard_sizes);
+	/*
+	 * prof_boot0 only initializes opt_prof_prefix.  We need to do it before
+	 * we parse malloc_conf options, in case malloc_conf parsing overwrites
+	 * it.
+	 */
+	if (config_prof) {
+		prof_boot0();
+	}
+	malloc_conf_init(&sc_data, bin_shard_sizes);
+	sz_boot(&sc_data);
+	bin_boot(&sc_data, bin_shard_sizes);
+
+	if (opt_stats_print) {
+		/* Print statistics at exit. */
+		if (atexit(stats_print_atexit) != 0) {
+			malloc_write("<jemalloc>: Error in atexit()\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+	}
+	if (pages_boot()) {
+		return true;
+	}
+	if (base_boot(TSDN_NULL)) {
+		return true;
+	}
+	if (extent_boot()) {
+		return true;
+	}
+	if (ctl_boot()) {
+		return true;
+	}
+	if (config_prof) {
+		prof_boot1();
+	}
+	arena_boot(&sc_data);
+	if (tcache_boot(TSDN_NULL)) {
+		return true;
+	}
+	if (malloc_mutex_init(&arenas_lock, "arenas", WITNESS_RANK_ARENAS,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	hook_boot();
+	/*
+	 * Create enough scaffolding to allow recursive allocation in
+	 * malloc_ncpus().
+	 */
+	narenas_auto = 1;
+	manual_arena_base = narenas_auto + 1;
+	memset(arenas, 0, sizeof(arena_t *) * narenas_auto);
+	/*
+	 * Initialize one arena here.  The rest are lazily created in
+	 * arena_choose_hard().
+	 */
+	if (arena_init(TSDN_NULL, 0, (extent_hooks_t *)&extent_hooks_default)
+	    == NULL) {
+		return true;
+	}
+	a0 = arena_get(TSDN_NULL, 0, false);
+	malloc_init_state = malloc_init_a0_initialized;
+
+	return false;
+}
+
+static bool
+malloc_init_hard_a0(void) {
+	bool ret;
+
+	malloc_mutex_lock(TSDN_NULL, &init_lock);
+	ret = malloc_init_hard_a0_locked();
+	malloc_mutex_unlock(TSDN_NULL, &init_lock);
+	return ret;
+}
+
+/* Initialize data structures which may trigger recursive allocation. */
+static bool
+malloc_init_hard_recursible(void) {
+	malloc_init_state = malloc_init_recursible;
+
+	ncpus = malloc_ncpus();
+
+#if (defined(JEMALLOC_HAVE_PTHREAD_ATFORK) && !defined(JEMALLOC_MUTEX_INIT_CB) \
+    && !defined(JEMALLOC_ZONE) && !defined(_WIN32) && \
+    !defined(__native_client__))
+	/* LinuxThreads' pthread_atfork() allocates. */
+	if (pthread_atfork(jemalloc_prefork, jemalloc_postfork_parent,
+	    jemalloc_postfork_child) != 0) {
+		malloc_write("<jemalloc>: Error in pthread_atfork()\n");
+		if (opt_abort) {
+			abort();
+		}
+		return true;
+	}
+#endif
+
+	if (background_thread_boot0()) {
+		return true;
+	}
+
+	return false;
+}
+
+static unsigned
+malloc_narenas_default(void) {
+	assert(ncpus > 0);
+	/*
+	 * For SMP systems, create more than one arena per CPU by
+	 * default.
+	 */
+	if (ncpus > 1) {
+		return ncpus << 2;
+	} else {
+		return 1;
+	}
+}
+
+static percpu_arena_mode_t
+percpu_arena_as_initialized(percpu_arena_mode_t mode) {
+	assert(!malloc_initialized());
+	assert(mode <= percpu_arena_disabled);
+
+	if (mode != percpu_arena_disabled) {
+		mode += percpu_arena_mode_enabled_base;
+	}
+
+	return mode;
+}
+
+static bool
+malloc_init_narenas(void) {
+	assert(ncpus > 0);
+
+	if (opt_percpu_arena != percpu_arena_disabled) {
+		if (!have_percpu_arena || malloc_getcpu() < 0) {
+			opt_percpu_arena = percpu_arena_disabled;
+			malloc_printf("<jemalloc>: perCPU arena getcpu() not "
+			    "available. Setting narenas to %u.\n", opt_narenas ?
+			    opt_narenas : malloc_narenas_default());
+			if (opt_abort) {
+				abort();
+			}
+		} else {
+			if (ncpus >= MALLOCX_ARENA_LIMIT) {
+				malloc_printf("<jemalloc>: narenas w/ percpu"
+				    "arena beyond limit (%d)\n", ncpus);
+				if (opt_abort) {
+					abort();
+				}
+				return true;
+			}
+			/* NB: opt_percpu_arena isn't fully initialized yet. */
+			if (percpu_arena_as_initialized(opt_percpu_arena) ==
+			    per_phycpu_arena && ncpus % 2 != 0) {
+				malloc_printf("<jemalloc>: invalid "
+				    "configuration -- per physical CPU arena "
+				    "with odd number (%u) of CPUs (no hyper "
+				    "threading?).\n", ncpus);
+				if (opt_abort)
+					abort();
+			}
+			unsigned n = percpu_arena_ind_limit(
+			    percpu_arena_as_initialized(opt_percpu_arena));
+			if (opt_narenas < n) {
+				/*
+				 * If narenas is specified with percpu_arena
+				 * enabled, actual narenas is set as the greater
+				 * of the two. percpu_arena_choose will be free
+				 * to use any of the arenas based on CPU
+				 * id. This is conservative (at a small cost)
+				 * but ensures correctness.
+				 *
+				 * If for some reason the ncpus determined at
+				 * boot is not the actual number (e.g. because
+				 * of affinity setting from numactl), reserving
+				 * narenas this way provides a workaround for
+				 * percpu_arena.
+				 */
+				opt_narenas = n;
+			}
+		}
+	}
+	if (opt_narenas == 0) {
+		opt_narenas = malloc_narenas_default();
+	}
+	assert(opt_narenas > 0);
+
+	narenas_auto = opt_narenas;
+	/*
+	 * Limit the number of arenas to the indexing range of MALLOCX_ARENA().
+	 */
+	if (narenas_auto >= MALLOCX_ARENA_LIMIT) {
+		narenas_auto = MALLOCX_ARENA_LIMIT - 1;
+		malloc_printf("<jemalloc>: Reducing narenas to limit (%d)\n",
+		    narenas_auto);
+	}
+	narenas_total_set(narenas_auto);
+	if (arena_init_huge()) {
+		narenas_total_inc();
+	}
+	manual_arena_base = narenas_total_get();
+
+	return false;
+}
+
+static void
+malloc_init_percpu(void) {
+	opt_percpu_arena = percpu_arena_as_initialized(opt_percpu_arena);
+}
+
+static bool
+malloc_init_hard_finish(void) {
+	if (malloc_mutex_boot()) {
+		return true;
+	}
+
+	malloc_init_state = malloc_init_initialized;
+	malloc_slow_flag_init();
+
+	return false;
+}
+
+static void
+malloc_init_hard_cleanup(tsdn_t *tsdn, bool reentrancy_set) {
+	malloc_mutex_assert_owner(tsdn, &init_lock);
+	malloc_mutex_unlock(tsdn, &init_lock);
+	if (reentrancy_set) {
+		assert(!tsdn_null(tsdn));
+		tsd_t *tsd = tsdn_tsd(tsdn);
+		assert(tsd_reentrancy_level_get(tsd) > 0);
+		post_reentrancy(tsd);
+	}
+}
+
+static bool
+malloc_init_hard(void) {
+	tsd_t *tsd;
+
+#if defined(_WIN32) && _WIN32_WINNT < 0x0600
+	_init_init_lock();
+#endif
+	malloc_mutex_lock(TSDN_NULL, &init_lock);
+
+#define UNLOCK_RETURN(tsdn, ret, reentrancy)		\
+	malloc_init_hard_cleanup(tsdn, reentrancy);	\
+	return ret;
+
+	if (!malloc_init_hard_needed()) {
+		UNLOCK_RETURN(TSDN_NULL, false, false)
+	}
+
+	if (malloc_init_state != malloc_init_a0_initialized &&
+	    malloc_init_hard_a0_locked()) {
+		UNLOCK_RETURN(TSDN_NULL, true, false)
+	}
+
+	malloc_mutex_unlock(TSDN_NULL, &init_lock);
+	/* Recursive allocation relies on functional tsd. */
+	tsd = malloc_tsd_boot0();
+	if (tsd == NULL) {
+		return true;
+	}
+	if (malloc_init_hard_recursible()) {
+		return true;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &init_lock);
+	/* Set reentrancy level to 1 during init. */
+	pre_reentrancy(tsd, NULL);
+	/* Initialize narenas before prof_boot2 (for allocation). */
+	if (malloc_init_narenas() || background_thread_boot1(tsd_tsdn(tsd))) {
+		UNLOCK_RETURN(tsd_tsdn(tsd), true, true)
+	}
+	if (config_prof && prof_boot2(tsd)) {
+		UNLOCK_RETURN(tsd_tsdn(tsd), true, true)
+	}
+
+	malloc_init_percpu();
+
+	if (malloc_init_hard_finish()) {
+		UNLOCK_RETURN(tsd_tsdn(tsd), true, true)
+	}
+	post_reentrancy(tsd);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &init_lock);
+
+	witness_assert_lockless(witness_tsd_tsdn(
+	    tsd_witness_tsdp_get_unsafe(tsd)));
+	malloc_tsd_boot1();
+	/* Update TSD after tsd_boot1. */
+	tsd = tsd_fetch();
+	if (opt_background_thread) {
+		assert(have_background_thread);
+		/*
+		 * Need to finish init & unlock first before creating background
+		 * threads (pthread_create depends on malloc).  ctl_init (which
+		 * sets isthreaded) needs to be called without holding any lock.
+		 */
+		background_thread_ctl_init(tsd_tsdn(tsd));
+		if (background_thread_create(tsd, 0)) {
+			return true;
+		}
+	}
+#undef UNLOCK_RETURN
+	return false;
+}
+
+/*
+ * End initialization functions.
+ */
+/******************************************************************************/
+/*
+ * Begin allocation-path internal functions and data structures.
+ */
+
+/*
+ * Settings determined by the documented behavior of the allocation functions.
+ */
+typedef struct static_opts_s static_opts_t;
+struct static_opts_s {
+	/* Whether or not allocation size may overflow. */
+	bool may_overflow;
+
+	/*
+	 * Whether or not allocations (with alignment) of size 0 should be
+	 * treated as size 1.
+	 */
+	bool bump_empty_aligned_alloc;
+	/*
+	 * Whether to assert that allocations are not of size 0 (after any
+	 * bumping).
+	 */
+	bool assert_nonempty_alloc;
+
+	/*
+	 * Whether or not to modify the 'result' argument to malloc in case of
+	 * error.
+	 */
+	bool null_out_result_on_error;
+	/* Whether to set errno when we encounter an error condition. */
+	bool set_errno_on_error;
+
+	/*
+	 * The minimum valid alignment for functions requesting aligned storage.
+	 */
+	size_t min_alignment;
+
+	/* The error string to use if we oom. */
+	const char *oom_string;
+	/* The error string to use if the passed-in alignment is invalid. */
+	const char *invalid_alignment_string;
+
+	/*
+	 * False if we're configured to skip some time-consuming operations.
+	 *
+	 * This isn't really a malloc "behavior", but it acts as a useful
+	 * summary of several other static (or at least, static after program
+	 * initialization) options.
+	 */
+	bool slow;
+	/*
+	 * Return size.
+	 */
+	bool usize;
+};
+
+JEMALLOC_ALWAYS_INLINE void
+static_opts_init(static_opts_t *static_opts) {
+	static_opts->may_overflow = false;
+	static_opts->bump_empty_aligned_alloc = false;
+	static_opts->assert_nonempty_alloc = false;
+	static_opts->null_out_result_on_error = false;
+	static_opts->set_errno_on_error = false;
+	static_opts->min_alignment = 0;
+	static_opts->oom_string = "";
+	static_opts->invalid_alignment_string = "";
+	static_opts->slow = false;
+	static_opts->usize = false;
+}
+
+/*
+ * These correspond to the macros in jemalloc/jemalloc_macros.h.  Broadly, we
+ * should have one constant here per magic value there.  Note however that the
+ * representations need not be related.
+ */
+#define TCACHE_IND_NONE ((unsigned)-1)
+#define TCACHE_IND_AUTOMATIC ((unsigned)-2)
+#define ARENA_IND_AUTOMATIC ((unsigned)-1)
+
+typedef struct dynamic_opts_s dynamic_opts_t;
+struct dynamic_opts_s {
+	void **result;
+	size_t usize;
+	size_t num_items;
+	size_t item_size;
+	size_t alignment;
+	bool zero;
+	unsigned tcache_ind;
+	unsigned arena_ind;
+};
+
+JEMALLOC_ALWAYS_INLINE void
+dynamic_opts_init(dynamic_opts_t *dynamic_opts) {
+	dynamic_opts->result = NULL;
+	dynamic_opts->usize = 0;
+	dynamic_opts->num_items = 0;
+	dynamic_opts->item_size = 0;
+	dynamic_opts->alignment = 0;
+	dynamic_opts->zero = false;
+	dynamic_opts->tcache_ind = TCACHE_IND_AUTOMATIC;
+	dynamic_opts->arena_ind = ARENA_IND_AUTOMATIC;
+}
+
+/* ind is ignored if dopts->alignment > 0. */
+JEMALLOC_ALWAYS_INLINE void *
+imalloc_no_sample(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd,
+    size_t size, size_t usize, szind_t ind) {
+	tcache_t *tcache;
+	arena_t *arena;
+
+	/* Fill in the tcache. */
+	if (dopts->tcache_ind == TCACHE_IND_AUTOMATIC) {
+		if (likely(!sopts->slow)) {
+			/* Getting tcache ptr unconditionally. */
+			tcache = tsd_tcachep_get(tsd);
+			assert(tcache == tcache_get(tsd));
+		} else {
+			tcache = tcache_get(tsd);
+		}
+	} else if (dopts->tcache_ind == TCACHE_IND_NONE) {
+		tcache = NULL;
+	} else {
+		tcache = tcaches_get(tsd, dopts->tcache_ind);
+	}
+
+	/* Fill in the arena. */
+	if (dopts->arena_ind == ARENA_IND_AUTOMATIC) {
+		/*
+		 * In case of automatic arena management, we defer arena
+		 * computation until as late as we can, hoping to fill the
+		 * allocation out of the tcache.
+		 */
+		arena = NULL;
+	} else {
+		arena = arena_get(tsd_tsdn(tsd), dopts->arena_ind, true);
+	}
+
+	if (unlikely(dopts->alignment != 0)) {
+		return ipalloct(tsd_tsdn(tsd), usize, dopts->alignment,
+		    dopts->zero, tcache, arena);
+	}
+
+	return iallocztm(tsd_tsdn(tsd), size, ind, dopts->zero, tcache, false,
+	    arena, sopts->slow);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+imalloc_sample(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd,
+    size_t usize, szind_t ind) {
+	void *ret;
+
+	/*
+	 * For small allocations, sampling bumps the usize.  If so, we allocate
+	 * from the ind_large bucket.
+	 */
+	szind_t ind_large;
+	size_t bumped_usize = usize;
+
+	if (usize <= SC_SMALL_MAXCLASS) {
+		assert(((dopts->alignment == 0) ?
+		    sz_s2u(SC_LARGE_MINCLASS) :
+		    sz_sa2u(SC_LARGE_MINCLASS, dopts->alignment))
+			== SC_LARGE_MINCLASS);
+		ind_large = sz_size2index(SC_LARGE_MINCLASS);
+		bumped_usize = sz_s2u(SC_LARGE_MINCLASS);
+		ret = imalloc_no_sample(sopts, dopts, tsd, bumped_usize,
+		    bumped_usize, ind_large);
+		if (unlikely(ret == NULL)) {
+			return NULL;
+		}
+		arena_prof_promote(tsd_tsdn(tsd), ret, usize);
+	} else {
+		ret = imalloc_no_sample(sopts, dopts, tsd, usize, usize, ind);
+	}
+
+	return ret;
+}
+
+/*
+ * Returns true if the allocation will overflow, and false otherwise.  Sets
+ * *size to the product either way.
+ */
+JEMALLOC_ALWAYS_INLINE bool
+compute_size_with_overflow(bool may_overflow, dynamic_opts_t *dopts,
+    size_t *size) {
+	/*
+	 * This function is just num_items * item_size, except that we may have
+	 * to check for overflow.
+	 */
+
+	if (!may_overflow) {
+		assert(dopts->num_items == 1);
+		*size = dopts->item_size;
+		return false;
+	}
+
+	/* A size_t with its high-half bits all set to 1. */
+	static const size_t high_bits = SIZE_T_MAX << (sizeof(size_t) * 8 / 2);
+
+	*size = dopts->item_size * dopts->num_items;
+
+	if (unlikely(*size == 0)) {
+		return (dopts->num_items != 0 && dopts->item_size != 0);
+	}
+
+	/*
+	 * We got a non-zero size, but we don't know if we overflowed to get
+	 * there.  To avoid having to do a divide, we'll be clever and note that
+	 * if both A and B can be represented in N/2 bits, then their product
+	 * can be represented in N bits (without the possibility of overflow).
+	 */
+	if (likely((high_bits & (dopts->num_items | dopts->item_size)) == 0)) {
+		return false;
+	}
+	if (likely(*size / dopts->item_size == dopts->num_items)) {
+		return false;
+	}
+	return true;
+}
+
+JEMALLOC_ALWAYS_INLINE int
+imalloc_body(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd) {
+	/* Where the actual allocated memory will live. */
+	void *allocation = NULL;
+	/* Filled in by compute_size_with_overflow below. */
+	size_t size = 0;
+	/*
+	 * For unaligned allocations, we need only ind.  For aligned
+	 * allocations, or in case of stats or profiling we need usize.
+	 *
+	 * These are actually dead stores, in that their values are reset before
+	 * any branch on their value is taken.  Sometimes though, it's
+	 * convenient to pass them as arguments before this point.  To avoid
+	 * undefined behavior then, we initialize them with dummy stores.
+	 */
+	szind_t ind = 0;
+	size_t usize = 0;
+
+	/* Reentrancy is only checked on slow path. */
+	int8_t reentrancy_level;
+
+	/* Compute the amount of memory the user wants. */
+	if (unlikely(compute_size_with_overflow(sopts->may_overflow, dopts,
+	    &size))) {
+		goto label_oom;
+	}
+
+	if (unlikely(dopts->alignment < sopts->min_alignment
+	    || (dopts->alignment & (dopts->alignment - 1)) != 0)) {
+		goto label_invalid_alignment;
+	}
+
+	/* This is the beginning of the "core" algorithm. */
+
+	if (dopts->alignment == 0) {
+		ind = sz_size2index(size);
+		if (unlikely(ind >= SC_NSIZES)) {
+			goto label_oom;
+		}
+		if (config_stats || (config_prof && opt_prof) || sopts->usize) {
+			usize = sz_index2size(ind);
+			dopts->usize = usize;
+			assert(usize > 0 && usize
+			    <= SC_LARGE_MAXCLASS);
+		}
+	} else {
+		if (sopts->bump_empty_aligned_alloc) {
+			if (unlikely(size == 0)) {
+				size = 1;
+			}
+		}
+		usize = sz_sa2u(size, dopts->alignment);
+		dopts->usize = usize;
+		if (unlikely(usize == 0
+		    || usize > SC_LARGE_MAXCLASS)) {
+			goto label_oom;
+		}
+	}
+	/* Validate the user input. */
+	if (sopts->assert_nonempty_alloc) {
+		assert (size != 0);
+	}
+
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	/*
+	 * If we need to handle reentrancy, we can do it out of a
+	 * known-initialized arena (i.e. arena 0).
+	 */
+	reentrancy_level = tsd_reentrancy_level_get(tsd);
+	if (sopts->slow && unlikely(reentrancy_level > 0)) {
+		/*
+		 * We should never specify particular arenas or tcaches from
+		 * within our internal allocations.
+		 */
+		assert(dopts->tcache_ind == TCACHE_IND_AUTOMATIC ||
+		    dopts->tcache_ind == TCACHE_IND_NONE);
+		assert(dopts->arena_ind == ARENA_IND_AUTOMATIC);
+		dopts->tcache_ind = TCACHE_IND_NONE;
+		/* We know that arena 0 has already been initialized. */
+		dopts->arena_ind = 0;
+	}
+
+	/* If profiling is on, get our profiling context. */
+	if (config_prof && opt_prof) {
+		/*
+		 * Note that if we're going down this path, usize must have been
+		 * initialized in the previous if statement.
+		 */
+		prof_tctx_t *tctx = prof_alloc_prep(
+		    tsd, usize, prof_active_get_unlocked(), true);
+
+		alloc_ctx_t alloc_ctx;
+		if (likely((uintptr_t)tctx == (uintptr_t)1U)) {
+			alloc_ctx.slab = (usize
+			    <= SC_SMALL_MAXCLASS);
+			allocation = imalloc_no_sample(
+			    sopts, dopts, tsd, usize, usize, ind);
+		} else if ((uintptr_t)tctx > (uintptr_t)1U) {
+			/*
+			 * Note that ind might still be 0 here.  This is fine;
+			 * imalloc_sample ignores ind if dopts->alignment > 0.
+			 */
+			allocation = imalloc_sample(
+			    sopts, dopts, tsd, usize, ind);
+			alloc_ctx.slab = false;
+		} else {
+			allocation = NULL;
+		}
+
+		if (unlikely(allocation == NULL)) {
+			prof_alloc_rollback(tsd, tctx, true);
+			goto label_oom;
+		}
+		prof_malloc(tsd_tsdn(tsd), allocation, usize, &alloc_ctx, tctx);
+	} else {
+		/*
+		 * If dopts->alignment > 0, then ind is still 0, but usize was
+		 * computed in the previous if statement.  Down the positive
+		 * alignment path, imalloc_no_sample ignores ind and size
+		 * (relying only on usize).
+		 */
+		allocation = imalloc_no_sample(sopts, dopts, tsd, size, usize,
+		    ind);
+		if (unlikely(allocation == NULL)) {
+			goto label_oom;
+		}
+	}
+
+	/*
+	 * Allocation has been done at this point.  We still have some
+	 * post-allocation work to do though.
+	 */
+	assert(dopts->alignment == 0
+	    || ((uintptr_t)allocation & (dopts->alignment - 1)) == ZU(0));
+
+	if (config_stats) {
+		assert(usize == isalloc(tsd_tsdn(tsd), allocation));
+		*tsd_thread_allocatedp_get(tsd) += usize;
+	}
+
+	if (sopts->slow) {
+		UTRACE(0, size, allocation);
+	}
+
+	/* Success! */
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	*dopts->result = allocation;
+	return 0;
+
+label_oom:
+	if (unlikely(sopts->slow) && config_xmalloc && unlikely(opt_xmalloc)) {
+		malloc_write(sopts->oom_string);
+		abort();
+	}
+
+	if (sopts->slow) {
+		UTRACE(NULL, size, NULL);
+	}
+
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	if (sopts->set_errno_on_error) {
+		set_errno(ENOMEM);
+	}
+
+	if (sopts->null_out_result_on_error) {
+		*dopts->result = NULL;
+	}
+
+	return ENOMEM;
+
+	/*
+	 * This label is only jumped to by one goto; we move it out of line
+	 * anyways to avoid obscuring the non-error paths, and for symmetry with
+	 * the oom case.
+	 */
+label_invalid_alignment:
+	if (config_xmalloc && unlikely(opt_xmalloc)) {
+		malloc_write(sopts->invalid_alignment_string);
+		abort();
+	}
+
+	if (sopts->set_errno_on_error) {
+		set_errno(EINVAL);
+	}
+
+	if (sopts->slow) {
+		UTRACE(NULL, size, NULL);
+	}
+
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	if (sopts->null_out_result_on_error) {
+		*dopts->result = NULL;
+	}
+
+	return EINVAL;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+imalloc_init_check(static_opts_t *sopts, dynamic_opts_t *dopts) {
+	if (unlikely(!malloc_initialized()) && unlikely(malloc_init())) {
+		if (config_xmalloc && unlikely(opt_xmalloc)) {
+			malloc_write(sopts->oom_string);
+			abort();
+		}
+		UTRACE(NULL, dopts->num_items * dopts->item_size, NULL);
+		set_errno(ENOMEM);
+		*dopts->result = NULL;
+
+		return false;
+	}
+
+	return true;
+}
+
+/* Returns the errno-style error code of the allocation. */
+JEMALLOC_ALWAYS_INLINE int
+imalloc(static_opts_t *sopts, dynamic_opts_t *dopts) {
+	if (tsd_get_allocates() && !imalloc_init_check(sopts, dopts)) {
+		return ENOMEM;
+	}
+
+	/* We always need the tsd.  Let's grab it right away. */
+	tsd_t *tsd = tsd_fetch();
+	assert(tsd);
+	if (likely(tsd_fast(tsd))) {
+		/* Fast and common path. */
+		tsd_assert_fast(tsd);
+		sopts->slow = false;
+		return imalloc_body(sopts, dopts, tsd);
+	} else {
+		if (!tsd_get_allocates() && !imalloc_init_check(sopts, dopts)) {
+			return ENOMEM;
+		}
+
+		sopts->slow = true;
+		return imalloc_body(sopts, dopts, tsd);
+	}
+}
+
+JEMALLOC_NOINLINE
+void *
+malloc_default(size_t size) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.malloc.entry", "size: %zu", size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.null_out_result_on_error = true;
+	sopts.set_errno_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in malloc(): out of memory\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+
+	imalloc(&sopts, &dopts);
+	/*
+	 * Note that this branch gets optimized away -- it immediately follows
+	 * the check on tsd_fast that sets sopts.slow.
+	 */
+	if (sopts.slow) {
+		uintptr_t args[3] = {size};
+		hook_invoke_alloc(hook_alloc_malloc, ret, (uintptr_t)ret, args);
+	}
+
+	LOG("core.malloc.exit", "result: %p", ret);
+
+	return ret;
+}
+
+/******************************************************************************/
+/*
+ * Begin malloc(3)-compatible functions.
+ */
+
+/*
+ * malloc() fastpath.
+ *
+ * Fastpath assumes size <= SC_LOOKUP_MAXCLASS, and that we hit
+ * tcache.  If either of these is false, we tail-call to the slowpath,
+ * malloc_default().  Tail-calling is used to avoid any caller-saved
+ * registers.
+ *
+ * fastpath supports ticker and profiling, both of which will also
+ * tail-call to the slowpath if they fire.
+ */
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1)
+je_malloc(size_t size) {
+	LOG("core.malloc.entry", "size: %zu", size);
+
+	if (tsd_get_allocates() && unlikely(!malloc_initialized())) {
+		return malloc_default(size);
+	}
+
+	tsd_t *tsd = tsd_get(false);
+	if (unlikely(!tsd || !tsd_fast(tsd) || (size > SC_LOOKUP_MAXCLASS))) {
+		return malloc_default(size);
+	}
+
+	tcache_t *tcache = tsd_tcachep_get(tsd);
+
+	if (unlikely(ticker_trytick(&tcache->gc_ticker))) {
+		return malloc_default(size);
+	}
+
+	szind_t ind = sz_size2index_lookup(size);
+	size_t usize;
+	if (config_stats || config_prof) {
+		usize = sz_index2size(ind);
+	}
+	/* Fast path relies on size being a bin. I.e. SC_LOOKUP_MAXCLASS < SC_SMALL_MAXCLASS */
+	assert(ind < SC_NBINS);
+	assert(size <= SC_SMALL_MAXCLASS);
+
+	if (config_prof) {
+		int64_t bytes_until_sample = tsd_bytes_until_sample_get(tsd);
+		bytes_until_sample -= usize;
+		tsd_bytes_until_sample_set(tsd, bytes_until_sample);
+
+		if (unlikely(bytes_until_sample < 0)) {
+			/*
+			 * Avoid a prof_active check on the fastpath.
+			 * If prof_active is false, set bytes_until_sample to
+			 * a large value.  If prof_active is set to true,
+			 * bytes_until_sample will be reset.
+			 */
+			if (!prof_active) {
+				tsd_bytes_until_sample_set(tsd, SSIZE_MAX);
+			}
+			return malloc_default(size);
+		}
+	}
+
+	cache_bin_t *bin = tcache_small_bin_get(tcache, ind);
+	bool tcache_success;
+	void* ret = cache_bin_alloc_easy(bin, &tcache_success);
+
+	if (tcache_success) {
+		if (config_stats) {
+			*tsd_thread_allocatedp_get(tsd) += usize;
+			bin->tstats.nrequests++;
+		}
+		if (config_prof) {
+			tcache->prof_accumbytes += usize;
+		}
+
+		LOG("core.malloc.exit", "result: %p", ret);
+
+		/* Fastpath success */
+		return ret;
+	}
+
+	return malloc_default(size);
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+JEMALLOC_ATTR(nonnull(1))
+je_posix_memalign(void **memptr, size_t alignment, size_t size) {
+	int ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.posix_memalign.entry", "mem ptr: %p, alignment: %zu, "
+	    "size: %zu", memptr, alignment, size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.bump_empty_aligned_alloc = true;
+	sopts.min_alignment = sizeof(void *);
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+
+	dopts.result = memptr;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = alignment;
+
+	ret = imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {(uintptr_t)memptr, (uintptr_t)alignment,
+			(uintptr_t)size};
+		hook_invoke_alloc(hook_alloc_posix_memalign, *memptr,
+		    (uintptr_t)ret, args);
+	}
+
+	LOG("core.posix_memalign.exit", "result: %d, alloc ptr: %p", ret,
+	    *memptr);
+
+	return ret;
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(2)
+je_aligned_alloc(size_t alignment, size_t size) {
+	void *ret;
+
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.aligned_alloc.entry", "alignment: %zu, size: %zu\n",
+	    alignment, size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.bump_empty_aligned_alloc = true;
+	sopts.null_out_result_on_error = true;
+	sopts.set_errno_on_error = true;
+	sopts.min_alignment = 1;
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = alignment;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {(uintptr_t)alignment, (uintptr_t)size};
+		hook_invoke_alloc(hook_alloc_aligned_alloc, ret,
+		    (uintptr_t)ret, args);
+	}
+
+	LOG("core.aligned_alloc.exit", "result: %p", ret);
+
+	return ret;
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE2(1, 2)
+je_calloc(size_t num, size_t size) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.calloc.entry", "num: %zu, size: %zu\n", num, size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.may_overflow = true;
+	sopts.null_out_result_on_error = true;
+	sopts.set_errno_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in calloc(): out of memory\n";
+
+	dopts.result = &ret;
+	dopts.num_items = num;
+	dopts.item_size = size;
+	dopts.zero = true;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {(uintptr_t)num, (uintptr_t)size};
+		hook_invoke_alloc(hook_alloc_calloc, ret, (uintptr_t)ret, args);
+	}
+
+	LOG("core.calloc.exit", "result: %p", ret);
+
+	return ret;
+}
+
+static void *
+irealloc_prof_sample(tsd_t *tsd, void *old_ptr, size_t old_usize, size_t usize,
+    prof_tctx_t *tctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+
+	if (tctx == NULL) {
+		return NULL;
+	}
+	if (usize <= SC_SMALL_MAXCLASS) {
+		p = iralloc(tsd, old_ptr, old_usize,
+		    SC_LARGE_MINCLASS, 0, false, hook_args);
+		if (p == NULL) {
+			return NULL;
+		}
+		arena_prof_promote(tsd_tsdn(tsd), p, usize);
+	} else {
+		p = iralloc(tsd, old_ptr, old_usize, usize, 0, false,
+		    hook_args);
+	}
+
+	return p;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+irealloc_prof(tsd_t *tsd, void *old_ptr, size_t old_usize, size_t usize,
+   alloc_ctx_t *alloc_ctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+	bool prof_active;
+	prof_tctx_t *old_tctx, *tctx;
+
+	prof_active = prof_active_get_unlocked();
+	old_tctx = prof_tctx_get(tsd_tsdn(tsd), old_ptr, alloc_ctx);
+	tctx = prof_alloc_prep(tsd, usize, prof_active, true);
+	if (unlikely((uintptr_t)tctx != (uintptr_t)1U)) {
+		p = irealloc_prof_sample(tsd, old_ptr, old_usize, usize, tctx,
+		    hook_args);
+	} else {
+		p = iralloc(tsd, old_ptr, old_usize, usize, 0, false,
+		    hook_args);
+	}
+	if (unlikely(p == NULL)) {
+		prof_alloc_rollback(tsd, tctx, true);
+		return NULL;
+	}
+	prof_realloc(tsd, p, usize, tctx, prof_active, true, old_ptr, old_usize,
+	    old_tctx);
+
+	return p;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+ifree(tsd_t *tsd, void *ptr, tcache_t *tcache, bool slow_path) {
+	if (!slow_path) {
+		tsd_assert_fast(tsd);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	if (tsd_reentrancy_level_get(tsd) != 0) {
+		assert(slow_path);
+	}
+
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	alloc_ctx_t alloc_ctx;
+	rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+	assert(alloc_ctx.szind != SC_NSIZES);
+
+	size_t usize;
+	if (config_prof && opt_prof) {
+		usize = sz_index2size(alloc_ctx.szind);
+		prof_free(tsd, ptr, usize, &alloc_ctx);
+	} else if (config_stats) {
+		usize = sz_index2size(alloc_ctx.szind);
+	}
+	if (config_stats) {
+		*tsd_thread_deallocatedp_get(tsd) += usize;
+	}
+
+	if (likely(!slow_path)) {
+		idalloctm(tsd_tsdn(tsd), ptr, tcache, &alloc_ctx, false,
+		    false);
+	} else {
+		idalloctm(tsd_tsdn(tsd), ptr, tcache, &alloc_ctx, false,
+		    true);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+isfree(tsd_t *tsd, void *ptr, size_t usize, tcache_t *tcache, bool slow_path) {
+	if (!slow_path) {
+		tsd_assert_fast(tsd);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	if (tsd_reentrancy_level_get(tsd) != 0) {
+		assert(slow_path);
+	}
+
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	alloc_ctx_t alloc_ctx, *ctx;
+	if (!config_cache_oblivious && ((uintptr_t)ptr & PAGE_MASK) != 0) {
+		/*
+		 * When cache_oblivious is disabled and ptr is not page aligned,
+		 * the allocation was not sampled -- usize can be used to
+		 * determine szind directly.
+		 */
+		alloc_ctx.szind = sz_size2index(usize);
+		alloc_ctx.slab = true;
+		ctx = &alloc_ctx;
+		if (config_debug) {
+			alloc_ctx_t dbg_ctx;
+			rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+			rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree,
+			    rtree_ctx, (uintptr_t)ptr, true, &dbg_ctx.szind,
+			    &dbg_ctx.slab);
+			assert(dbg_ctx.szind == alloc_ctx.szind);
+			assert(dbg_ctx.slab == alloc_ctx.slab);
+		}
+	} else if (config_prof && opt_prof) {
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+		assert(alloc_ctx.szind == sz_size2index(usize));
+		ctx = &alloc_ctx;
+	} else {
+		ctx = NULL;
+	}
+
+	if (config_prof && opt_prof) {
+		prof_free(tsd, ptr, usize, ctx);
+	}
+	if (config_stats) {
+		*tsd_thread_deallocatedp_get(tsd) += usize;
+	}
+
+	if (likely(!slow_path)) {
+		isdalloct(tsd_tsdn(tsd), ptr, usize, tcache, ctx, false);
+	} else {
+		isdalloct(tsd_tsdn(tsd), ptr, usize, tcache, ctx, true);
+	}
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ALLOC_SIZE(2)
+je_realloc(void *ptr, size_t arg_size) {
+	void *ret;
+	tsdn_t *tsdn JEMALLOC_CC_SILENCE_INIT(NULL);
+	size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+	size_t old_usize = 0;
+	size_t size = arg_size;
+
+	LOG("core.realloc.entry", "ptr: %p, size: %zu\n", ptr, size);
+
+	if (unlikely(size == 0)) {
+		if (ptr != NULL) {
+			/* realloc(ptr, 0) is equivalent to free(ptr). */
+			UTRACE(ptr, 0, 0);
+			tcache_t *tcache;
+			tsd_t *tsd = tsd_fetch();
+			if (tsd_reentrancy_level_get(tsd) == 0) {
+				tcache = tcache_get(tsd);
+			} else {
+				tcache = NULL;
+			}
+
+			uintptr_t args[3] = {(uintptr_t)ptr, size};
+			hook_invoke_dalloc(hook_dalloc_realloc, ptr, args);
+
+			ifree(tsd, ptr, tcache, true);
+
+			LOG("core.realloc.exit", "result: %p", NULL);
+			return NULL;
+		}
+		size = 1;
+	}
+
+	if (likely(ptr != NULL)) {
+		assert(malloc_initialized() || IS_INITIALIZER);
+		tsd_t *tsd = tsd_fetch();
+
+		check_entry_exit_locking(tsd_tsdn(tsd));
+
+
+		hook_ralloc_args_t hook_args = {true, {(uintptr_t)ptr,
+			(uintptr_t)arg_size, 0, 0}};
+
+		alloc_ctx_t alloc_ctx;
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+		assert(alloc_ctx.szind != SC_NSIZES);
+		old_usize = sz_index2size(alloc_ctx.szind);
+		assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
+		if (config_prof && opt_prof) {
+			usize = sz_s2u(size);
+			if (unlikely(usize == 0
+			    || usize > SC_LARGE_MAXCLASS)) {
+				ret = NULL;
+			} else {
+				ret = irealloc_prof(tsd, ptr, old_usize, usize,
+				    &alloc_ctx, &hook_args);
+			}
+		} else {
+			if (config_stats) {
+				usize = sz_s2u(size);
+			}
+			ret = iralloc(tsd, ptr, old_usize, size, 0, false,
+			    &hook_args);
+		}
+		tsdn = tsd_tsdn(tsd);
+	} else {
+		/* realloc(NULL, size) is equivalent to malloc(size). */
+		static_opts_t sopts;
+		dynamic_opts_t dopts;
+
+		static_opts_init(&sopts);
+		dynamic_opts_init(&dopts);
+
+		sopts.null_out_result_on_error = true;
+		sopts.set_errno_on_error = true;
+		sopts.oom_string =
+		    "<jemalloc>: Error in realloc(): out of memory\n";
+
+		dopts.result = &ret;
+		dopts.num_items = 1;
+		dopts.item_size = size;
+
+		imalloc(&sopts, &dopts);
+		if (sopts.slow) {
+			uintptr_t args[3] = {(uintptr_t)ptr, arg_size};
+			hook_invoke_alloc(hook_alloc_realloc, ret,
+			    (uintptr_t)ret, args);
+		}
+
+		return ret;
+	}
+
+	if (unlikely(ret == NULL)) {
+		if (config_xmalloc && unlikely(opt_xmalloc)) {
+			malloc_write("<jemalloc>: Error in realloc(): "
+			    "out of memory\n");
+			abort();
+		}
+		set_errno(ENOMEM);
+	}
+	if (config_stats && likely(ret != NULL)) {
+		tsd_t *tsd;
+
+		assert(usize == isalloc(tsdn, ret));
+		tsd = tsdn_tsd(tsdn);
+		*tsd_thread_allocatedp_get(tsd) += usize;
+		*tsd_thread_deallocatedp_get(tsd) += old_usize;
+	}
+	UTRACE(ptr, size, ret);
+	check_entry_exit_locking(tsdn);
+
+	LOG("core.realloc.exit", "result: %p", ret);
+	return ret;
+}
+
+JEMALLOC_NOINLINE
+void
+free_default(void *ptr) {
+	UTRACE(ptr, 0, 0);
+	if (likely(ptr != NULL)) {
+		/*
+		 * We avoid setting up tsd fully (e.g. tcache, arena binding)
+		 * based on only free() calls -- other activities trigger the
+		 * minimal to full transition.  This is because free() may
+		 * happen during thread shutdown after tls deallocation: if a
+		 * thread never had any malloc activities until then, a
+		 * fully-setup tsd won't be destructed properly.
+		 */
+		tsd_t *tsd = tsd_fetch_min();
+		check_entry_exit_locking(tsd_tsdn(tsd));
+
+		tcache_t *tcache;
+		if (likely(tsd_fast(tsd))) {
+			tsd_assert_fast(tsd);
+			/* Unconditionally get tcache ptr on fast path. */
+			tcache = tsd_tcachep_get(tsd);
+			ifree(tsd, ptr, tcache, false);
+		} else {
+			if (likely(tsd_reentrancy_level_get(tsd) == 0)) {
+				tcache = tcache_get(tsd);
+			} else {
+				tcache = NULL;
+			}
+			uintptr_t args_raw[3] = {(uintptr_t)ptr};
+			hook_invoke_dalloc(hook_dalloc_free, ptr, args_raw);
+			ifree(tsd, ptr, tcache, true);
+		}
+		check_entry_exit_locking(tsd_tsdn(tsd));
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE
+bool free_fastpath(void *ptr, size_t size, bool size_hint) {
+	tsd_t *tsd = tsd_get(false);
+	if (unlikely(!tsd || !tsd_fast(tsd))) {
+		return false;
+	}
+
+	tcache_t *tcache = tsd_tcachep_get(tsd);
+
+	alloc_ctx_t alloc_ctx;
+	/*
+	 * If !config_cache_oblivious, we can check PAGE alignment to
+	 * detect sampled objects.  Otherwise addresses are
+	 * randomized, and we have to look it up in the rtree anyway.
+	 * See also isfree().
+	 */
+	if (!size_hint || config_cache_oblivious) {
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		bool res = rtree_szind_slab_read_fast(tsd_tsdn(tsd), &extents_rtree,
+						      rtree_ctx, (uintptr_t)ptr,
+						      &alloc_ctx.szind, &alloc_ctx.slab);
+
+		/* Note: profiled objects will have alloc_ctx.slab set */
+		if (!res || !alloc_ctx.slab) {
+			return false;
+		}
+		assert(alloc_ctx.szind != SC_NSIZES);
+	} else {
+		/*
+		 * Check for both sizes that are too large, and for sampled objects.
+		 * Sampled objects are always page-aligned.  The sampled object check
+		 * will also check for null ptr.
+		 */
+		if (size > SC_LOOKUP_MAXCLASS || (((uintptr_t)ptr & PAGE_MASK) == 0)) {
+			return false;
+		}
+		alloc_ctx.szind = sz_size2index_lookup(size);
+	}
+
+	if (unlikely(ticker_trytick(&tcache->gc_ticker))) {
+		return false;
+	}
+
+	cache_bin_t *bin = tcache_small_bin_get(tcache, alloc_ctx.szind);
+	cache_bin_info_t *bin_info = &tcache_bin_info[alloc_ctx.szind];
+	if (!cache_bin_dalloc_easy(bin, bin_info, ptr)) {
+		return false;
+	}
+
+	if (config_stats) {
+		size_t usize = sz_index2size(alloc_ctx.szind);
+		*tsd_thread_deallocatedp_get(tsd) += usize;
+	}
+
+	return true;
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_free(void *ptr) {
+	LOG("core.free.entry", "ptr: %p", ptr);
+
+	if (!free_fastpath(ptr, 0, false)) {
+		free_default(ptr);
+	}
+
+	LOG("core.free.exit", "");
+}
+
+/*
+ * End malloc(3)-compatible functions.
+ */
+/******************************************************************************/
+/*
+ * Begin non-standard override functions.
+ */
+
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc)
+je_memalign(size_t alignment, size_t size) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.memalign.entry", "alignment: %zu, size: %zu\n", alignment,
+	    size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.min_alignment = 1;
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+	sopts.null_out_result_on_error = true;
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = alignment;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {alignment, size};
+		hook_invoke_alloc(hook_alloc_memalign, ret, (uintptr_t)ret,
+		    args);
+	}
+
+	LOG("core.memalign.exit", "result: %p", ret);
+	return ret;
+}
+#endif
+
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc)
+je_valloc(size_t size) {
+	void *ret;
+
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.valloc.entry", "size: %zu\n", size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.null_out_result_on_error = true;
+	sopts.min_alignment = PAGE;
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = PAGE;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {size};
+		hook_invoke_alloc(hook_alloc_valloc, ret, (uintptr_t)ret, args);
+	}
+
+	LOG("core.valloc.exit", "result: %p\n", ret);
+	return ret;
+}
+#endif
+
+#if defined(JEMALLOC_IS_MALLOC) && defined(JEMALLOC_GLIBC_MALLOC_HOOK)
+/*
+ * glibc provides the RTLD_DEEPBIND flag for dlopen which can make it possible
+ * to inconsistently reference libc's malloc(3)-compatible functions
+ * (https://bugzilla.mozilla.org/show_bug.cgi?id=493541).
+ *
+ * These definitions interpose hooks in glibc.  The functions are actually
+ * passed an extra argument for the caller return address, which will be
+ * ignored.
+ */
+JEMALLOC_EXPORT void (*__free_hook)(void *ptr) = je_free;
+JEMALLOC_EXPORT void *(*__malloc_hook)(size_t size) = je_malloc;
+JEMALLOC_EXPORT void *(*__realloc_hook)(void *ptr, size_t size) = je_realloc;
+#  ifdef JEMALLOC_GLIBC_MEMALIGN_HOOK
+JEMALLOC_EXPORT void *(*__memalign_hook)(size_t alignment, size_t size) =
+    je_memalign;
+#  endif
+
+#  ifdef CPU_COUNT
+/*
+ * To enable static linking with glibc, the libc specific malloc interface must
+ * be implemented also, so none of glibc's malloc.o functions are added to the
+ * link.
+ */
+#    define ALIAS(je_fn)	__attribute__((alias (#je_fn), used))
+/* To force macro expansion of je_ prefix before stringification. */
+#    define PREALIAS(je_fn)	ALIAS(je_fn)
+#    ifdef JEMALLOC_OVERRIDE___LIBC_CALLOC
+void *__libc_calloc(size_t n, size_t size) PREALIAS(je_calloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_FREE
+void __libc_free(void* ptr) PREALIAS(je_free);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_MALLOC
+void *__libc_malloc(size_t size) PREALIAS(je_malloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_MEMALIGN
+void *__libc_memalign(size_t align, size_t s) PREALIAS(je_memalign);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_REALLOC
+void *__libc_realloc(void* ptr, size_t size) PREALIAS(je_realloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_VALLOC
+void *__libc_valloc(size_t size) PREALIAS(je_valloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___POSIX_MEMALIGN
+int __posix_memalign(void** r, size_t a, size_t s) PREALIAS(je_posix_memalign);
+#    endif
+#    undef PREALIAS
+#    undef ALIAS
+#  endif
+#endif
+
+/*
+ * End non-standard override functions.
+ */
+/******************************************************************************/
+/*
+ * Begin non-standard functions.
+ */
+
+#ifdef JEMALLOC_EXPERIMENTAL_SMALLOCX_API
+
+#define JEMALLOC_SMALLOCX_CONCAT_HELPER(x, y) x ## y
+#define JEMALLOC_SMALLOCX_CONCAT_HELPER2(x, y)  \
+  JEMALLOC_SMALLOCX_CONCAT_HELPER(x, y)
+
+typedef struct {
+	void *ptr;
+	size_t size;
+} smallocx_return_t;
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+smallocx_return_t JEMALLOC_NOTHROW
+/*
+ * The attribute JEMALLOC_ATTR(malloc) cannot be used due to:
+ *  - https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86488
+ */
+JEMALLOC_SMALLOCX_CONCAT_HELPER2(je_smallocx_, JEMALLOC_VERSION_GID_IDENT)
+  (size_t size, int flags) {
+	/*
+	 * Note: the attribute JEMALLOC_ALLOC_SIZE(1) cannot be
+	 * used here because it makes writing beyond the `size`
+	 * of the `ptr` undefined behavior, but the objective
+	 * of this function is to allow writing beyond `size`
+	 * up to `smallocx_return_t::size`.
+	 */
+	smallocx_return_t ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.smallocx.entry", "size: %zu, flags: %d", size, flags);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.assert_nonempty_alloc = true;
+	sopts.null_out_result_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in mallocx(): out of memory\n";
+	sopts.usize = true;
+
+	dopts.result = &ret.ptr;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	if (unlikely(flags != 0)) {
+		if ((flags & MALLOCX_LG_ALIGN_MASK) != 0) {
+			dopts.alignment = MALLOCX_ALIGN_GET_SPECIFIED(flags);
+		}
+
+		dopts.zero = MALLOCX_ZERO_GET(flags);
+
+		if ((flags & MALLOCX_TCACHE_MASK) != 0) {
+			if ((flags & MALLOCX_TCACHE_MASK)
+			    == MALLOCX_TCACHE_NONE) {
+				dopts.tcache_ind = TCACHE_IND_NONE;
+			} else {
+				dopts.tcache_ind = MALLOCX_TCACHE_GET(flags);
+			}
+		} else {
+			dopts.tcache_ind = TCACHE_IND_AUTOMATIC;
+		}
+
+		if ((flags & MALLOCX_ARENA_MASK) != 0)
+			dopts.arena_ind = MALLOCX_ARENA_GET(flags);
+	}
+
+	imalloc(&sopts, &dopts);
+	assert(dopts.usize == je_nallocx(size, flags));
+	ret.size = dopts.usize;
+
+	LOG("core.smallocx.exit", "result: %p, size: %zu", ret.ptr, ret.size);
+	return ret;
+}
+#undef JEMALLOC_SMALLOCX_CONCAT_HELPER
+#undef JEMALLOC_SMALLOCX_CONCAT_HELPER2
+#endif
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1)
+je_mallocx(size_t size, int flags) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.mallocx.entry", "size: %zu, flags: %d", size, flags);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.assert_nonempty_alloc = true;
+	sopts.null_out_result_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in mallocx(): out of memory\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	if (unlikely(flags != 0)) {
+		if ((flags & MALLOCX_LG_ALIGN_MASK) != 0) {
+			dopts.alignment = MALLOCX_ALIGN_GET_SPECIFIED(flags);
+		}
+
+		dopts.zero = MALLOCX_ZERO_GET(flags);
+
+		if ((flags & MALLOCX_TCACHE_MASK) != 0) {
+			if ((flags & MALLOCX_TCACHE_MASK)
+			    == MALLOCX_TCACHE_NONE) {
+				dopts.tcache_ind = TCACHE_IND_NONE;
+			} else {
+				dopts.tcache_ind = MALLOCX_TCACHE_GET(flags);
+			}
+		} else {
+			dopts.tcache_ind = TCACHE_IND_AUTOMATIC;
+		}
+
+		if ((flags & MALLOCX_ARENA_MASK) != 0)
+			dopts.arena_ind = MALLOCX_ARENA_GET(flags);
+	}
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {size, flags};
+		hook_invoke_alloc(hook_alloc_mallocx, ret, (uintptr_t)ret,
+		    args);
+	}
+
+	LOG("core.mallocx.exit", "result: %p", ret);
+	return ret;
+}
+
+static void *
+irallocx_prof_sample(tsdn_t *tsdn, void *old_ptr, size_t old_usize,
+    size_t usize, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena,
+    prof_tctx_t *tctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+
+	if (tctx == NULL) {
+		return NULL;
+	}
+	if (usize <= SC_SMALL_MAXCLASS) {
+		p = iralloct(tsdn, old_ptr, old_usize,
+		    SC_LARGE_MINCLASS, alignment, zero, tcache,
+		    arena, hook_args);
+		if (p == NULL) {
+			return NULL;
+		}
+		arena_prof_promote(tsdn, p, usize);
+	} else {
+		p = iralloct(tsdn, old_ptr, old_usize, usize, alignment, zero,
+		    tcache, arena, hook_args);
+	}
+
+	return p;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+irallocx_prof(tsd_t *tsd, void *old_ptr, size_t old_usize, size_t size,
+    size_t alignment, size_t *usize, bool zero, tcache_t *tcache,
+    arena_t *arena, alloc_ctx_t *alloc_ctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+	bool prof_active;
+	prof_tctx_t *old_tctx, *tctx;
+
+	prof_active = prof_active_get_unlocked();
+	old_tctx = prof_tctx_get(tsd_tsdn(tsd), old_ptr, alloc_ctx);
+	tctx = prof_alloc_prep(tsd, *usize, prof_active, false);
+	if (unlikely((uintptr_t)tctx != (uintptr_t)1U)) {
+		p = irallocx_prof_sample(tsd_tsdn(tsd), old_ptr, old_usize,
+		    *usize, alignment, zero, tcache, arena, tctx, hook_args);
+	} else {
+		p = iralloct(tsd_tsdn(tsd), old_ptr, old_usize, size, alignment,
+		    zero, tcache, arena, hook_args);
+	}
+	if (unlikely(p == NULL)) {
+		prof_alloc_rollback(tsd, tctx, false);
+		return NULL;
+	}
+
+	if (p == old_ptr && alignment != 0) {
+		/*
+		 * The allocation did not move, so it is possible that the size
+		 * class is smaller than would guarantee the requested
+		 * alignment, and that the alignment constraint was
+		 * serendipitously satisfied.  Additionally, old_usize may not
+		 * be the same as the current usize because of in-place large
+		 * reallocation.  Therefore, query the actual value of usize.
+		 */
+		*usize = isalloc(tsd_tsdn(tsd), p);
+	}
+	prof_realloc(tsd, p, *usize, tctx, prof_active, false, old_ptr,
+	    old_usize, old_tctx);
+
+	return p;
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ALLOC_SIZE(2)
+je_rallocx(void *ptr, size_t size, int flags) {
+	void *p;
+	tsd_t *tsd;
+	size_t usize;
+	size_t old_usize;
+	size_t alignment = MALLOCX_ALIGN_GET(flags);
+	bool zero = flags & MALLOCX_ZERO;
+	arena_t *arena;
+	tcache_t *tcache;
+
+	LOG("core.rallocx.entry", "ptr: %p, size: %zu, flags: %d", ptr,
+	    size, flags);
+
+
+	assert(ptr != NULL);
+	assert(size != 0);
+	assert(malloc_initialized() || IS_INITIALIZER);
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	if (unlikely((flags & MALLOCX_ARENA_MASK) != 0)) {
+		unsigned arena_ind = MALLOCX_ARENA_GET(flags);
+		arena = arena_get(tsd_tsdn(tsd), arena_ind, true);
+		if (unlikely(arena == NULL)) {
+			goto label_oom;
+		}
+	} else {
+		arena = NULL;
+	}
+
+	if (unlikely((flags & MALLOCX_TCACHE_MASK) != 0)) {
+		if ((flags & MALLOCX_TCACHE_MASK) == MALLOCX_TCACHE_NONE) {
+			tcache = NULL;
+		} else {
+			tcache = tcaches_get(tsd, MALLOCX_TCACHE_GET(flags));
+		}
+	} else {
+		tcache = tcache_get(tsd);
+	}
+
+	alloc_ctx_t alloc_ctx;
+	rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+	assert(alloc_ctx.szind != SC_NSIZES);
+	old_usize = sz_index2size(alloc_ctx.szind);
+	assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
+
+	hook_ralloc_args_t hook_args = {false, {(uintptr_t)ptr, size, flags,
+		0}};
+	if (config_prof && opt_prof) {
+		usize = (alignment == 0) ?
+		    sz_s2u(size) : sz_sa2u(size, alignment);
+		if (unlikely(usize == 0
+		    || usize > SC_LARGE_MAXCLASS)) {
+			goto label_oom;
+		}
+		p = irallocx_prof(tsd, ptr, old_usize, size, alignment, &usize,
+		    zero, tcache, arena, &alloc_ctx, &hook_args);
+		if (unlikely(p == NULL)) {
+			goto label_oom;
+		}
+	} else {
+		p = iralloct(tsd_tsdn(tsd), ptr, old_usize, size, alignment,
+		    zero, tcache, arena, &hook_args);
+		if (unlikely(p == NULL)) {
+			goto label_oom;
+		}
+		if (config_stats) {
+			usize = isalloc(tsd_tsdn(tsd), p);
+		}
+	}
+	assert(alignment == 0 || ((uintptr_t)p & (alignment - 1)) == ZU(0));
+
+	if (config_stats) {
+		*tsd_thread_allocatedp_get(tsd) += usize;
+		*tsd_thread_deallocatedp_get(tsd) += old_usize;
+	}
+	UTRACE(ptr, size, p);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.rallocx.exit", "result: %p", p);
+	return p;
+label_oom:
+	if (config_xmalloc && unlikely(opt_xmalloc)) {
+		malloc_write("<jemalloc>: Error in rallocx(): out of memory\n");
+		abort();
+	}
+	UTRACE(ptr, size, 0);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.rallocx.exit", "result: %p", NULL);
+	return NULL;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ixallocx_helper(tsdn_t *tsdn, void *ptr, size_t old_usize, size_t size,
+    size_t extra, size_t alignment, bool zero) {
+	size_t newsize;
+
+	if (ixalloc(tsdn, ptr, old_usize, size, extra, alignment, zero,
+	    &newsize)) {
+		return old_usize;
+	}
+
+	return newsize;
+}
+
+static size_t
+ixallocx_prof_sample(tsdn_t *tsdn, void *ptr, size_t old_usize, size_t size,
+    size_t extra, size_t alignment, bool zero, prof_tctx_t *tctx) {
+	size_t usize;
+
+	if (tctx == NULL) {
+		return old_usize;
+	}
+	usize = ixallocx_helper(tsdn, ptr, old_usize, size, extra, alignment,
+	    zero);
+
+	return usize;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ixallocx_prof(tsd_t *tsd, void *ptr, size_t old_usize, size_t size,
+    size_t extra, size_t alignment, bool zero, alloc_ctx_t *alloc_ctx) {
+	size_t usize_max, usize;
+	bool prof_active;
+	prof_tctx_t *old_tctx, *tctx;
+
+	prof_active = prof_active_get_unlocked();
+	old_tctx = prof_tctx_get(tsd_tsdn(tsd), ptr, alloc_ctx);
+	/*
+	 * usize isn't knowable before ixalloc() returns when extra is non-zero.
+	 * Therefore, compute its maximum possible value and use that in
+	 * prof_alloc_prep() to decide whether to capture a backtrace.
+	 * prof_realloc() will use the actual usize to decide whether to sample.
+	 */
+	if (alignment == 0) {
+		usize_max = sz_s2u(size+extra);
+		assert(usize_max > 0
+		    && usize_max <= SC_LARGE_MAXCLASS);
+	} else {
+		usize_max = sz_sa2u(size+extra, alignment);
+		if (unlikely(usize_max == 0
+		    || usize_max > SC_LARGE_MAXCLASS)) {
+			/*
+			 * usize_max is out of range, and chances are that
+			 * allocation will fail, but use the maximum possible
+			 * value and carry on with prof_alloc_prep(), just in
+			 * case allocation succeeds.
+			 */
+			usize_max = SC_LARGE_MAXCLASS;
+		}
+	}
+	tctx = prof_alloc_prep(tsd, usize_max, prof_active, false);
+
+	if (unlikely((uintptr_t)tctx != (uintptr_t)1U)) {
+		usize = ixallocx_prof_sample(tsd_tsdn(tsd), ptr, old_usize,
+		    size, extra, alignment, zero, tctx);
+	} else {
+		usize = ixallocx_helper(tsd_tsdn(tsd), ptr, old_usize, size,
+		    extra, alignment, zero);
+	}
+	if (usize == old_usize) {
+		prof_alloc_rollback(tsd, tctx, false);
+		return usize;
+	}
+	prof_realloc(tsd, ptr, usize, tctx, prof_active, false, ptr, old_usize,
+	    old_tctx);
+
+	return usize;
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+je_xallocx(void *ptr, size_t size, size_t extra, int flags) {
+	tsd_t *tsd;
+	size_t usize, old_usize;
+	size_t alignment = MALLOCX_ALIGN_GET(flags);
+	bool zero = flags & MALLOCX_ZERO;
+
+	LOG("core.xallocx.entry", "ptr: %p, size: %zu, extra: %zu, "
+	    "flags: %d", ptr, size, extra, flags);
+
+	assert(ptr != NULL);
+	assert(size != 0);
+	assert(SIZE_T_MAX - size >= extra);
+	assert(malloc_initialized() || IS_INITIALIZER);
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	alloc_ctx_t alloc_ctx;
+	rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+	assert(alloc_ctx.szind != SC_NSIZES);
+	old_usize = sz_index2size(alloc_ctx.szind);
+	assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
+	/*
+	 * The API explicitly absolves itself of protecting against (size +
+	 * extra) numerical overflow, but we may need to clamp extra to avoid
+	 * exceeding SC_LARGE_MAXCLASS.
+	 *
+	 * Ordinarily, size limit checking is handled deeper down, but here we
+	 * have to check as part of (size + extra) clamping, since we need the
+	 * clamped value in the above helper functions.
+	 */
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		usize = old_usize;
+		goto label_not_resized;
+	}
+	if (unlikely(SC_LARGE_MAXCLASS - size < extra)) {
+		extra = SC_LARGE_MAXCLASS - size;
+	}
+
+	if (config_prof && opt_prof) {
+		usize = ixallocx_prof(tsd, ptr, old_usize, size, extra,
+		    alignment, zero, &alloc_ctx);
+	} else {
+		usize = ixallocx_helper(tsd_tsdn(tsd), ptr, old_usize, size,
+		    extra, alignment, zero);
+	}
+	if (unlikely(usize == old_usize)) {
+		goto label_not_resized;
+	}
+
+	if (config_stats) {
+		*tsd_thread_allocatedp_get(tsd) += usize;
+		*tsd_thread_deallocatedp_get(tsd) += old_usize;
+	}
+label_not_resized:
+	if (unlikely(!tsd_fast(tsd))) {
+		uintptr_t args[4] = {(uintptr_t)ptr, size, extra, flags};
+		hook_invoke_expand(hook_expand_xallocx, ptr, old_usize,
+		    usize, (uintptr_t)usize, args);
+	}
+
+	UTRACE(ptr, size, ptr);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.xallocx.exit", "result: %zu", usize);
+	return usize;
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+JEMALLOC_ATTR(pure)
+je_sallocx(const void *ptr, int flags) {
+	size_t usize;
+	tsdn_t *tsdn;
+
+	LOG("core.sallocx.entry", "ptr: %p, flags: %d", ptr, flags);
+
+	assert(malloc_initialized() || IS_INITIALIZER);
+	assert(ptr != NULL);
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+
+	if (config_debug || force_ivsalloc) {
+		usize = ivsalloc(tsdn, ptr);
+		assert(force_ivsalloc || usize != 0);
+	} else {
+		usize = isalloc(tsdn, ptr);
+	}
+
+	check_entry_exit_locking(tsdn);
+
+	LOG("core.sallocx.exit", "result: %zu", usize);
+	return usize;
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_dallocx(void *ptr, int flags) {
+	LOG("core.dallocx.entry", "ptr: %p, flags: %d", ptr, flags);
+
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	tsd_t *tsd = tsd_fetch();
+	bool fast = tsd_fast(tsd);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	tcache_t *tcache;
+	if (unlikely((flags & MALLOCX_TCACHE_MASK) != 0)) {
+		/* Not allowed to be reentrant and specify a custom tcache. */
+		assert(tsd_reentrancy_level_get(tsd) == 0);
+		if ((flags & MALLOCX_TCACHE_MASK) == MALLOCX_TCACHE_NONE) {
+			tcache = NULL;
+		} else {
+			tcache = tcaches_get(tsd, MALLOCX_TCACHE_GET(flags));
+		}
+	} else {
+		if (likely(fast)) {
+			tcache = tsd_tcachep_get(tsd);
+			assert(tcache == tcache_get(tsd));
+		} else {
+			if (likely(tsd_reentrancy_level_get(tsd) == 0)) {
+				tcache = tcache_get(tsd);
+			}  else {
+				tcache = NULL;
+			}
+		}
+	}
+
+	UTRACE(ptr, 0, 0);
+	if (likely(fast)) {
+		tsd_assert_fast(tsd);
+		ifree(tsd, ptr, tcache, false);
+	} else {
+		uintptr_t args_raw[3] = {(uintptr_t)ptr, flags};
+		hook_invoke_dalloc(hook_dalloc_dallocx, ptr, args_raw);
+		ifree(tsd, ptr, tcache, true);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.dallocx.exit", "");
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+inallocx(tsdn_t *tsdn, size_t size, int flags) {
+	check_entry_exit_locking(tsdn);
+
+	size_t usize;
+	if (likely((flags & MALLOCX_LG_ALIGN_MASK) == 0)) {
+		usize = sz_s2u(size);
+	} else {
+		usize = sz_sa2u(size, MALLOCX_ALIGN_GET_SPECIFIED(flags));
+	}
+	check_entry_exit_locking(tsdn);
+	return usize;
+}
+
+JEMALLOC_NOINLINE void
+sdallocx_default(void *ptr, size_t size, int flags) {
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	tsd_t *tsd = tsd_fetch();
+	bool fast = tsd_fast(tsd);
+	size_t usize = inallocx(tsd_tsdn(tsd), size, flags);
+	assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	tcache_t *tcache;
+	if (unlikely((flags & MALLOCX_TCACHE_MASK) != 0)) {
+		/* Not allowed to be reentrant and specify a custom tcache. */
+		assert(tsd_reentrancy_level_get(tsd) == 0);
+		if ((flags & MALLOCX_TCACHE_MASK) == MALLOCX_TCACHE_NONE) {
+			tcache = NULL;
+		} else {
+			tcache = tcaches_get(tsd, MALLOCX_TCACHE_GET(flags));
+		}
+	} else {
+		if (likely(fast)) {
+			tcache = tsd_tcachep_get(tsd);
+			assert(tcache == tcache_get(tsd));
+		} else {
+			if (likely(tsd_reentrancy_level_get(tsd) == 0)) {
+				tcache = tcache_get(tsd);
+			} else {
+				tcache = NULL;
+			}
+		}
+	}
+
+	UTRACE(ptr, 0, 0);
+	if (likely(fast)) {
+		tsd_assert_fast(tsd);
+		isfree(tsd, ptr, usize, tcache, false);
+	} else {
+		uintptr_t args_raw[3] = {(uintptr_t)ptr, size, flags};
+		hook_invoke_dalloc(hook_dalloc_sdallocx, ptr, args_raw);
+		isfree(tsd, ptr, usize, tcache, true);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_sdallocx(void *ptr, size_t size, int flags) {
+	LOG("core.sdallocx.entry", "ptr: %p, size: %zu, flags: %d", ptr,
+		size, flags);
+
+	if (flags !=0 || !free_fastpath(ptr, size, true)) {
+		sdallocx_default(ptr, size, flags);
+	}
+
+	LOG("core.sdallocx.exit", "");
+}
+
+void JEMALLOC_NOTHROW
+je_sdallocx_noflags(void *ptr, size_t size) {
+	LOG("core.sdallocx.entry", "ptr: %p, size: %zu, flags: 0", ptr,
+		size);
+
+	if (!free_fastpath(ptr, size, true)) {
+		sdallocx_default(ptr, size, 0);
+	}
+
+	LOG("core.sdallocx.exit", "");
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+JEMALLOC_ATTR(pure)
+je_nallocx(size_t size, int flags) {
+	size_t usize;
+	tsdn_t *tsdn;
+
+	assert(size != 0);
+
+	if (unlikely(malloc_init())) {
+		LOG("core.nallocx.exit", "result: %zu", ZU(0));
+		return 0;
+	}
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+
+	usize = inallocx(tsdn, size, flags);
+	if (unlikely(usize > SC_LARGE_MAXCLASS)) {
+		LOG("core.nallocx.exit", "result: %zu", ZU(0));
+		return 0;
+	}
+
+	check_entry_exit_locking(tsdn);
+	LOG("core.nallocx.exit", "result: %zu", usize);
+	return usize;
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+je_mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	tsd_t *tsd;
+
+	LOG("core.mallctl.entry", "name: %s", name);
+
+	if (unlikely(malloc_init())) {
+		LOG("core.mallctl.exit", "result: %d", EAGAIN);
+		return EAGAIN;
+	}
+
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	ret = ctl_byname(tsd, name, oldp, oldlenp, newp, newlen);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.mallctl.exit", "result: %d", ret);
+	return ret;
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+je_mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp) {
+	int ret;
+
+	LOG("core.mallctlnametomib.entry", "name: %s", name);
+
+	if (unlikely(malloc_init())) {
+		LOG("core.mallctlnametomib.exit", "result: %d", EAGAIN);
+		return EAGAIN;
+	}
+
+	tsd_t *tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	ret = ctl_nametomib(tsd, name, mibp, miblenp);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.mallctlnametomib.exit", "result: %d", ret);
+	return ret;
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+je_mallctlbymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+  void *newp, size_t newlen) {
+	int ret;
+	tsd_t *tsd;
+
+	LOG("core.mallctlbymib.entry", "");
+
+	if (unlikely(malloc_init())) {
+		LOG("core.mallctlbymib.exit", "result: %d", EAGAIN);
+		return EAGAIN;
+	}
+
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	ret = ctl_bymib(tsd, mib, miblen, oldp, oldlenp, newp, newlen);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	LOG("core.mallctlbymib.exit", "result: %d", ret);
+	return ret;
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_malloc_stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts) {
+	tsdn_t *tsdn;
+
+	LOG("core.malloc_stats_print.entry", "");
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+	stats_print(write_cb, cbopaque, opts);
+	check_entry_exit_locking(tsdn);
+	LOG("core.malloc_stats_print.exit", "");
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+je_malloc_usable_size(JEMALLOC_USABLE_SIZE_CONST void *ptr) {
+	size_t ret;
+	tsdn_t *tsdn;
+
+	LOG("core.malloc_usable_size.entry", "ptr: %p", ptr);
+
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+
+	if (unlikely(ptr == NULL)) {
+		ret = 0;
+	} else {
+		if (config_debug || force_ivsalloc) {
+			ret = ivsalloc(tsdn, ptr);
+			assert(force_ivsalloc || ret != 0);
+		} else {
+			ret = isalloc(tsdn, ptr);
+		}
+	}
+
+	check_entry_exit_locking(tsdn);
+	LOG("core.malloc_usable_size.exit", "result: %zu", ret);
+	return ret;
+}
+
+/*
+ * End non-standard functions.
+ */
+/******************************************************************************/
+/*
+ * The following functions are used by threading libraries for protection of
+ * malloc during fork().
+ */
+
+/*
+ * If an application creates a thread before doing any allocation in the main
+ * thread, then calls fork(2) in the main thread followed by memory allocation
+ * in the child process, a race can occur that results in deadlock within the
+ * child: the main thread may have forked while the created thread had
+ * partially initialized the allocator.  Ordinarily jemalloc prevents
+ * fork/malloc races via the following functions it registers during
+ * initialization using pthread_atfork(), but of course that does no good if
+ * the allocator isn't fully initialized at fork time.  The following library
+ * constructor is a partial solution to this problem.  It may still be possible
+ * to trigger the deadlock described above, but doing so would involve forking
+ * via a library constructor that runs before jemalloc's runs.
+ */
+#ifndef JEMALLOC_JET
+JEMALLOC_ATTR(constructor)
+static void
+jemalloc_constructor(void) {
+	malloc_init();
+}
+#endif
+
+#ifndef JEMALLOC_MUTEX_INIT_CB
+void
+jemalloc_prefork(void)
+#else
+JEMALLOC_EXPORT void
+_malloc_prefork(void)
+#endif
+{
+	tsd_t *tsd;
+	unsigned i, j, narenas;
+	arena_t *arena;
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	if (!malloc_initialized()) {
+		return;
+	}
+#endif
+	assert(malloc_initialized());
+
+	tsd = tsd_fetch();
+
+	narenas = narenas_total_get();
+
+	witness_prefork(tsd_witness_tsdp_get(tsd));
+	/* Acquire all mutexes in a safe order. */
+	ctl_prefork(tsd_tsdn(tsd));
+	tcache_prefork(tsd_tsdn(tsd));
+	malloc_mutex_prefork(tsd_tsdn(tsd), &arenas_lock);
+	if (have_background_thread) {
+		background_thread_prefork0(tsd_tsdn(tsd));
+	}
+	prof_prefork0(tsd_tsdn(tsd));
+	if (have_background_thread) {
+		background_thread_prefork1(tsd_tsdn(tsd));
+	}
+	/* Break arena prefork into stages to preserve lock order. */
+	for (i = 0; i < 8; i++) {
+		for (j = 0; j < narenas; j++) {
+			if ((arena = arena_get(tsd_tsdn(tsd), j, false)) !=
+			    NULL) {
+				switch (i) {
+				case 0:
+					arena_prefork0(tsd_tsdn(tsd), arena);
+					break;
+				case 1:
+					arena_prefork1(tsd_tsdn(tsd), arena);
+					break;
+				case 2:
+					arena_prefork2(tsd_tsdn(tsd), arena);
+					break;
+				case 3:
+					arena_prefork3(tsd_tsdn(tsd), arena);
+					break;
+				case 4:
+					arena_prefork4(tsd_tsdn(tsd), arena);
+					break;
+				case 5:
+					arena_prefork5(tsd_tsdn(tsd), arena);
+					break;
+				case 6:
+					arena_prefork6(tsd_tsdn(tsd), arena);
+					break;
+				case 7:
+					arena_prefork7(tsd_tsdn(tsd), arena);
+					break;
+				default: not_reached();
+				}
+			}
+		}
+	}
+	prof_prefork1(tsd_tsdn(tsd));
+	tsd_prefork(tsd);
+}
+
+#ifndef JEMALLOC_MUTEX_INIT_CB
+void
+jemalloc_postfork_parent(void)
+#else
+JEMALLOC_EXPORT void
+_malloc_postfork(void)
+#endif
+{
+	tsd_t *tsd;
+	unsigned i, narenas;
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	if (!malloc_initialized()) {
+		return;
+	}
+#endif
+	assert(malloc_initialized());
+
+	tsd = tsd_fetch();
+
+	tsd_postfork_parent(tsd);
+
+	witness_postfork_parent(tsd_witness_tsdp_get(tsd));
+	/* Release all mutexes, now that fork() has completed. */
+	for (i = 0, narenas = narenas_total_get(); i < narenas; i++) {
+		arena_t *arena;
+
+		if ((arena = arena_get(tsd_tsdn(tsd), i, false)) != NULL) {
+			arena_postfork_parent(tsd_tsdn(tsd), arena);
+		}
+	}
+	prof_postfork_parent(tsd_tsdn(tsd));
+	if (have_background_thread) {
+		background_thread_postfork_parent(tsd_tsdn(tsd));
+	}
+	malloc_mutex_postfork_parent(tsd_tsdn(tsd), &arenas_lock);
+	tcache_postfork_parent(tsd_tsdn(tsd));
+	ctl_postfork_parent(tsd_tsdn(tsd));
+}
+
+void
+jemalloc_postfork_child(void) {
+	tsd_t *tsd;
+	unsigned i, narenas;
+
+	assert(malloc_initialized());
+
+	tsd = tsd_fetch();
+
+	tsd_postfork_child(tsd);
+
+	witness_postfork_child(tsd_witness_tsdp_get(tsd));
+	/* Release all mutexes, now that fork() has completed. */
+	for (i = 0, narenas = narenas_total_get(); i < narenas; i++) {
+		arena_t *arena;
+
+		if ((arena = arena_get(tsd_tsdn(tsd), i, false)) != NULL) {
+			arena_postfork_child(tsd_tsdn(tsd), arena);
+		}
+	}
+	prof_postfork_child(tsd_tsdn(tsd));
+	if (have_background_thread) {
+		background_thread_postfork_child(tsd_tsdn(tsd));
+	}
+	malloc_mutex_postfork_child(tsd_tsdn(tsd), &arenas_lock);
+	tcache_postfork_child(tsd_tsdn(tsd));
+	ctl_postfork_child(tsd_tsdn(tsd));
+}
+
+/******************************************************************************/
diff --git a/jemalloc/src/jemalloc_cpp.cpp b/jemalloc/src/jemalloc_cpp.cpp
new file mode 100644
index 0000000..da0441a
--- /dev/null
+++ b/jemalloc/src/jemalloc_cpp.cpp
@@ -0,0 +1,141 @@
+#include <mutex>
+#include <new>
+
+#define JEMALLOC_CPP_CPP_
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#ifdef __cplusplus
+}
+#endif
+
+// All operators in this file are exported.
+
+// Possibly alias hidden versions of malloc and sdallocx to avoid an extra plt
+// thunk?
+//
+// extern __typeof (sdallocx) sdallocx_int
+//  __attribute ((alias ("sdallocx"),
+//		visibility ("hidden")));
+//
+// ... but it needs to work with jemalloc namespaces.
+
+void	*operator new(std::size_t size);
+void	*operator new[](std::size_t size);
+void	*operator new(std::size_t size, const std::nothrow_t &) noexcept;
+void	*operator new[](std::size_t size, const std::nothrow_t &) noexcept;
+void	operator delete(void *ptr) noexcept;
+void	operator delete[](void *ptr) noexcept;
+void	operator delete(void *ptr, const std::nothrow_t &) noexcept;
+void	operator delete[](void *ptr, const std::nothrow_t &) noexcept;
+
+#if __cpp_sized_deallocation >= 201309
+/* C++14's sized-delete operators. */
+void	operator delete(void *ptr, std::size_t size) noexcept;
+void	operator delete[](void *ptr, std::size_t size) noexcept;
+#endif
+
+JEMALLOC_NOINLINE
+static void *
+handleOOM(std::size_t size, bool nothrow) {
+	void *ptr = nullptr;
+
+	while (ptr == nullptr) {
+		std::new_handler handler;
+		// GCC-4.8 and clang 4.0 do not have std::get_new_handler.
+		{
+			static std::mutex mtx;
+			std::lock_guard<std::mutex> lock(mtx);
+
+			handler = std::set_new_handler(nullptr);
+			std::set_new_handler(handler);
+		}
+		if (handler == nullptr)
+			break;
+
+		try {
+			handler();
+		} catch (const std::bad_alloc &) {
+			break;
+		}
+
+		ptr = je_malloc(size);
+	}
+
+	if (ptr == nullptr && !nothrow)
+		std::__throw_bad_alloc();
+	return ptr;
+}
+
+template <bool IsNoExcept>
+JEMALLOC_ALWAYS_INLINE
+void *
+newImpl(std::size_t size) noexcept(IsNoExcept) {
+	void *ptr = je_malloc(size);
+	if (likely(ptr != nullptr))
+		return ptr;
+
+	return handleOOM(size, IsNoExcept);
+}
+
+void *
+operator new(std::size_t size) {
+	return newImpl<false>(size);
+}
+
+void *
+operator new[](std::size_t size) {
+	return newImpl<false>(size);
+}
+
+void *
+operator new(std::size_t size, const std::nothrow_t &) noexcept {
+	return newImpl<true>(size);
+}
+
+void *
+operator new[](std::size_t size, const std::nothrow_t &) noexcept {
+	return newImpl<true>(size);
+}
+
+void
+operator delete(void *ptr) noexcept {
+	je_free(ptr);
+}
+
+void
+operator delete[](void *ptr) noexcept {
+	je_free(ptr);
+}
+
+void
+operator delete(void *ptr, const std::nothrow_t &) noexcept {
+	je_free(ptr);
+}
+
+void operator delete[](void *ptr, const std::nothrow_t &) noexcept {
+	je_free(ptr);
+}
+
+#if __cpp_sized_deallocation >= 201309
+
+void
+operator delete(void *ptr, std::size_t size) noexcept {
+	if (unlikely(ptr == nullptr)) {
+		return;
+	}
+	je_sdallocx_noflags(ptr, size);
+}
+
+void operator delete[](void *ptr, std::size_t size) noexcept {
+	if (unlikely(ptr == nullptr)) {
+		return;
+	}
+	je_sdallocx_noflags(ptr, size);
+}
+
+#endif  // __cpp_sized_deallocation
diff --git a/jemalloc/src/large.c b/jemalloc/src/large.c
new file mode 100644
index 0000000..8e7a781
--- /dev/null
+++ b/jemalloc/src/large.c
@@ -0,0 +1,395 @@
+#define JEMALLOC_LARGE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+
+void *
+large_malloc(tsdn_t *tsdn, arena_t *arena, size_t usize, bool zero) {
+	assert(usize == sz_s2u(usize));
+
+	return large_palloc(tsdn, arena, usize, CACHELINE, zero);
+}
+
+void *
+large_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
+    bool zero) {
+	size_t ausize;
+	extent_t *extent;
+	bool is_zeroed;
+	UNUSED bool idump JEMALLOC_CC_SILENCE_INIT(false);
+
+	assert(!tsdn_null(tsdn) || arena != NULL);
+
+	ausize = sz_sa2u(usize, alignment);
+	if (unlikely(ausize == 0 || ausize > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+
+	if (config_fill && unlikely(opt_zero)) {
+		zero = true;
+	}
+	/*
+	 * Copy zero into is_zeroed and pass the copy when allocating the
+	 * extent, so that it is possible to make correct junk/zero fill
+	 * decisions below, even if is_zeroed ends up true when zero is false.
+	 */
+	is_zeroed = zero;
+	if (likely(!tsdn_null(tsdn))) {
+		arena = arena_choose_maybe_huge(tsdn_tsd(tsdn), arena, usize);
+	}
+	if (unlikely(arena == NULL) || (extent = arena_extent_alloc_large(tsdn,
+	    arena, usize, alignment, &is_zeroed)) == NULL) {
+		return NULL;
+	}
+
+	/* See comments in arena_bin_slabs_full_insert(). */
+	if (!arena_is_auto(arena)) {
+		/* Insert extent into large. */
+		malloc_mutex_lock(tsdn, &arena->large_mtx);
+		extent_list_append(&arena->large, extent);
+		malloc_mutex_unlock(tsdn, &arena->large_mtx);
+	}
+	if (config_prof && arena_prof_accum(tsdn, arena, usize)) {
+		prof_idump(tsdn);
+	}
+
+	if (zero) {
+		assert(is_zeroed);
+	} else if (config_fill && unlikely(opt_junk_alloc)) {
+		memset(extent_addr_get(extent), JEMALLOC_ALLOC_JUNK,
+		    extent_usize_get(extent));
+	}
+
+	arena_decay_tick(tsdn, arena);
+	return extent_addr_get(extent);
+}
+
+static void
+large_dalloc_junk_impl(void *ptr, size_t size) {
+	memset(ptr, JEMALLOC_FREE_JUNK, size);
+}
+large_dalloc_junk_t *JET_MUTABLE large_dalloc_junk = large_dalloc_junk_impl;
+
+static void
+large_dalloc_maybe_junk_impl(void *ptr, size_t size) {
+	if (config_fill && have_dss && unlikely(opt_junk_free)) {
+		/*
+		 * Only bother junk filling if the extent isn't about to be
+		 * unmapped.
+		 */
+		if (opt_retain || (have_dss && extent_in_dss(ptr))) {
+			large_dalloc_junk(ptr, size);
+		}
+	}
+}
+large_dalloc_maybe_junk_t *JET_MUTABLE large_dalloc_maybe_junk =
+    large_dalloc_maybe_junk_impl;
+
+static bool
+large_ralloc_no_move_shrink(tsdn_t *tsdn, extent_t *extent, size_t usize) {
+	arena_t *arena = extent_arena_get(extent);
+	size_t oldusize = extent_usize_get(extent);
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+	size_t diff = extent_size_get(extent) - (usize + sz_large_pad);
+
+	assert(oldusize > usize);
+
+	if (extent_hooks->split == NULL) {
+		return true;
+	}
+
+	/* Split excess pages. */
+	if (diff != 0) {
+		extent_t *trail = extent_split_wrapper(tsdn, arena,
+		    &extent_hooks, extent, usize + sz_large_pad,
+		    sz_size2index(usize), false, diff, SC_NSIZES, false);
+		if (trail == NULL) {
+			return true;
+		}
+
+		if (config_fill && unlikely(opt_junk_free)) {
+			large_dalloc_maybe_junk(extent_addr_get(trail),
+			    extent_size_get(trail));
+		}
+
+		arena_extents_dirty_dalloc(tsdn, arena, &extent_hooks, trail);
+	}
+
+	arena_extent_ralloc_large_shrink(tsdn, arena, extent, oldusize);
+
+	return false;
+}
+
+static bool
+large_ralloc_no_move_expand(tsdn_t *tsdn, extent_t *extent, size_t usize,
+    bool zero) {
+	arena_t *arena = extent_arena_get(extent);
+	size_t oldusize = extent_usize_get(extent);
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+	size_t trailsize = usize - oldusize;
+
+	if (extent_hooks->merge == NULL) {
+		return true;
+	}
+
+	if (config_fill && unlikely(opt_zero)) {
+		zero = true;
+	}
+	/*
+	 * Copy zero into is_zeroed_trail and pass the copy when allocating the
+	 * extent, so that it is possible to make correct junk/zero fill
+	 * decisions below, even if is_zeroed_trail ends up true when zero is
+	 * false.
+	 */
+	bool is_zeroed_trail = zero;
+	bool commit = true;
+	extent_t *trail;
+	bool new_mapping;
+	if ((trail = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_dirty, extent_past_get(extent), trailsize, 0,
+	    CACHELINE, false, SC_NSIZES, &is_zeroed_trail, &commit)) != NULL
+	    || (trail = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_muzzy, extent_past_get(extent), trailsize, 0,
+	    CACHELINE, false, SC_NSIZES, &is_zeroed_trail, &commit)) != NULL) {
+		if (config_stats) {
+			new_mapping = false;
+		}
+	} else {
+		if ((trail = extent_alloc_wrapper(tsdn, arena, &extent_hooks,
+		    extent_past_get(extent), trailsize, 0, CACHELINE, false,
+		    SC_NSIZES, &is_zeroed_trail, &commit)) == NULL) {
+			return true;
+		}
+		if (config_stats) {
+			new_mapping = true;
+		}
+	}
+
+	if (extent_merge_wrapper(tsdn, arena, &extent_hooks, extent, trail)) {
+		extent_dalloc_wrapper(tsdn, arena, &extent_hooks, trail);
+		return true;
+	}
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	szind_t szind = sz_size2index(usize);
+	extent_szind_set(extent, szind);
+	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_addr_get(extent), szind, false);
+
+	if (config_stats && new_mapping) {
+		arena_stats_mapped_add(tsdn, &arena->stats, trailsize);
+	}
+
+	if (zero) {
+		if (config_cache_oblivious) {
+			/*
+			 * Zero the trailing bytes of the original allocation's
+			 * last page, since they are in an indeterminate state.
+			 * There will always be trailing bytes, because ptr's
+			 * offset from the beginning of the extent is a multiple
+			 * of CACHELINE in [0 .. PAGE).
+			 */
+			void *zbase = (void *)
+			    ((uintptr_t)extent_addr_get(extent) + oldusize);
+			void *zpast = PAGE_ADDR2BASE((void *)((uintptr_t)zbase +
+			    PAGE));
+			size_t nzero = (uintptr_t)zpast - (uintptr_t)zbase;
+			assert(nzero > 0);
+			memset(zbase, 0, nzero);
+		}
+		assert(is_zeroed_trail);
+	} else if (config_fill && unlikely(opt_junk_alloc)) {
+		memset((void *)((uintptr_t)extent_addr_get(extent) + oldusize),
+		    JEMALLOC_ALLOC_JUNK, usize - oldusize);
+	}
+
+	arena_extent_ralloc_large_expand(tsdn, arena, extent, oldusize);
+
+	return false;
+}
+
+bool
+large_ralloc_no_move(tsdn_t *tsdn, extent_t *extent, size_t usize_min,
+    size_t usize_max, bool zero) {
+	size_t oldusize = extent_usize_get(extent);
+
+	/* The following should have been caught by callers. */
+	assert(usize_min > 0 && usize_max <= SC_LARGE_MAXCLASS);
+	/* Both allocation sizes must be large to avoid a move. */
+	assert(oldusize >= SC_LARGE_MINCLASS
+	    && usize_max >= SC_LARGE_MINCLASS);
+
+	if (usize_max > oldusize) {
+		/* Attempt to expand the allocation in-place. */
+		if (!large_ralloc_no_move_expand(tsdn, extent, usize_max,
+		    zero)) {
+			arena_decay_tick(tsdn, extent_arena_get(extent));
+			return false;
+		}
+		/* Try again, this time with usize_min. */
+		if (usize_min < usize_max && usize_min > oldusize &&
+		    large_ralloc_no_move_expand(tsdn, extent, usize_min,
+		    zero)) {
+			arena_decay_tick(tsdn, extent_arena_get(extent));
+			return false;
+		}
+	}
+
+	/*
+	 * Avoid moving the allocation if the existing extent size accommodates
+	 * the new size.
+	 */
+	if (oldusize >= usize_min && oldusize <= usize_max) {
+		arena_decay_tick(tsdn, extent_arena_get(extent));
+		return false;
+	}
+
+	/* Attempt to shrink the allocation in-place. */
+	if (oldusize > usize_max) {
+		if (!large_ralloc_no_move_shrink(tsdn, extent, usize_max)) {
+			arena_decay_tick(tsdn, extent_arena_get(extent));
+			return false;
+		}
+	}
+	return true;
+}
+
+static void *
+large_ralloc_move_helper(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool zero) {
+	if (alignment <= CACHELINE) {
+		return large_malloc(tsdn, arena, usize, zero);
+	}
+	return large_palloc(tsdn, arena, usize, alignment, zero);
+}
+
+void *
+large_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args) {
+	extent_t *extent = iealloc(tsdn, ptr);
+
+	size_t oldusize = extent_usize_get(extent);
+	/* The following should have been caught by callers. */
+	assert(usize > 0 && usize <= SC_LARGE_MAXCLASS);
+	/* Both allocation sizes must be large to avoid a move. */
+	assert(oldusize >= SC_LARGE_MINCLASS
+	    && usize >= SC_LARGE_MINCLASS);
+
+	/* Try to avoid moving the allocation. */
+	if (!large_ralloc_no_move(tsdn, extent, usize, usize, zero)) {
+		hook_invoke_expand(hook_args->is_realloc
+		    ? hook_expand_realloc : hook_expand_rallocx, ptr, oldusize,
+		    usize, (uintptr_t)ptr, hook_args->args);
+		return extent_addr_get(extent);
+	}
+
+	/*
+	 * usize and old size are different enough that we need to use a
+	 * different size class.  In that case, fall back to allocating new
+	 * space and copying.
+	 */
+	void *ret = large_ralloc_move_helper(tsdn, arena, usize, alignment,
+	    zero);
+	if (ret == NULL) {
+		return NULL;
+	}
+
+	hook_invoke_alloc(hook_args->is_realloc
+	    ? hook_alloc_realloc : hook_alloc_rallocx, ret, (uintptr_t)ret,
+	    hook_args->args);
+	hook_invoke_dalloc(hook_args->is_realloc
+	    ? hook_dalloc_realloc : hook_dalloc_rallocx, ptr, hook_args->args);
+
+	size_t copysize = (usize < oldusize) ? usize : oldusize;
+	memcpy(ret, extent_addr_get(extent), copysize);
+	isdalloct(tsdn, extent_addr_get(extent), oldusize, tcache, NULL, true);
+	return ret;
+}
+
+/*
+ * junked_locked indicates whether the extent's data have been junk-filled, and
+ * whether the arena's large_mtx is currently held.
+ */
+static void
+large_dalloc_prep_impl(tsdn_t *tsdn, arena_t *arena, extent_t *extent,
+    bool junked_locked) {
+	if (!junked_locked) {
+		/* See comments in arena_bin_slabs_full_insert(). */
+		if (!arena_is_auto(arena)) {
+			malloc_mutex_lock(tsdn, &arena->large_mtx);
+			extent_list_remove(&arena->large, extent);
+			malloc_mutex_unlock(tsdn, &arena->large_mtx);
+		}
+		large_dalloc_maybe_junk(extent_addr_get(extent),
+		    extent_usize_get(extent));
+	} else {
+		/* Only hold the large_mtx if necessary. */
+		if (!arena_is_auto(arena)) {
+			malloc_mutex_assert_owner(tsdn, &arena->large_mtx);
+			extent_list_remove(&arena->large, extent);
+		}
+	}
+	arena_extent_dalloc_large_prep(tsdn, arena, extent);
+}
+
+static void
+large_dalloc_finish_impl(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+	arena_extents_dirty_dalloc(tsdn, arena, &extent_hooks, extent);
+}
+
+void
+large_dalloc_prep_junked_locked(tsdn_t *tsdn, extent_t *extent) {
+	large_dalloc_prep_impl(tsdn, extent_arena_get(extent), extent, true);
+}
+
+void
+large_dalloc_finish(tsdn_t *tsdn, extent_t *extent) {
+	large_dalloc_finish_impl(tsdn, extent_arena_get(extent), extent);
+}
+
+void
+large_dalloc(tsdn_t *tsdn, extent_t *extent) {
+	arena_t *arena = extent_arena_get(extent);
+	large_dalloc_prep_impl(tsdn, arena, extent, false);
+	large_dalloc_finish_impl(tsdn, arena, extent);
+	arena_decay_tick(tsdn, arena);
+}
+
+size_t
+large_salloc(tsdn_t *tsdn, const extent_t *extent) {
+	return extent_usize_get(extent);
+}
+
+prof_tctx_t *
+large_prof_tctx_get(tsdn_t *tsdn, const extent_t *extent) {
+	return extent_prof_tctx_get(extent);
+}
+
+void
+large_prof_tctx_set(tsdn_t *tsdn, extent_t *extent, prof_tctx_t *tctx) {
+	extent_prof_tctx_set(extent, tctx);
+}
+
+void
+large_prof_tctx_reset(tsdn_t *tsdn, extent_t *extent) {
+	large_prof_tctx_set(tsdn, extent, (prof_tctx_t *)(uintptr_t)1U);
+}
+
+nstime_t
+large_prof_alloc_time_get(const extent_t *extent) {
+	return extent_prof_alloc_time_get(extent);
+}
+
+void
+large_prof_alloc_time_set(extent_t *extent, nstime_t t) {
+	extent_prof_alloc_time_set(extent, t);
+}
diff --git a/jemalloc/src/log.c b/jemalloc/src/log.c
new file mode 100644
index 0000000..778902f
--- /dev/null
+++ b/jemalloc/src/log.c
@@ -0,0 +1,78 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/log.h"
+
+char log_var_names[JEMALLOC_LOG_VAR_BUFSIZE];
+atomic_b_t log_init_done = ATOMIC_INIT(false);
+
+/*
+ * Returns true if we were able to pick out a segment.  Fills in r_segment_end
+ * with a pointer to the first character after the end of the string.
+ */
+static const char *
+log_var_extract_segment(const char* segment_begin) {
+	const char *end;
+	for (end = segment_begin; *end != '\0' && *end != '|'; end++) {
+	}
+	return end;
+}
+
+static bool
+log_var_matches_segment(const char *segment_begin, const char *segment_end,
+    const char *log_var_begin, const char *log_var_end) {
+	assert(segment_begin <= segment_end);
+	assert(log_var_begin < log_var_end);
+
+	ptrdiff_t segment_len = segment_end - segment_begin;
+	ptrdiff_t log_var_len = log_var_end - log_var_begin;
+	/* The special '.' segment matches everything. */
+	if (segment_len == 1 && *segment_begin == '.') {
+		return true;
+	}
+        if (segment_len == log_var_len) {
+		return strncmp(segment_begin, log_var_begin, segment_len) == 0;
+	} else if (segment_len < log_var_len) {
+		return strncmp(segment_begin, log_var_begin, segment_len) == 0
+		    && log_var_begin[segment_len] == '.';
+        } else {
+		return false;
+	}
+}
+
+unsigned
+log_var_update_state(log_var_t *log_var) {
+	const char *log_var_begin = log_var->name;
+	const char *log_var_end = log_var->name + strlen(log_var->name);
+
+	/* Pointer to one before the beginning of the current segment. */
+	const char *segment_begin = log_var_names;
+
+	/*
+	 * If log_init done is false, we haven't parsed the malloc conf yet.  To
+	 * avoid log-spew, we default to not displaying anything.
+	 */
+	if (!atomic_load_b(&log_init_done, ATOMIC_ACQUIRE)) {
+		return LOG_INITIALIZED_NOT_ENABLED;
+	}
+
+	while (true) {
+		const char *segment_end = log_var_extract_segment(
+		    segment_begin);
+		assert(segment_end < log_var_names + JEMALLOC_LOG_VAR_BUFSIZE);
+		if (log_var_matches_segment(segment_begin, segment_end,
+		    log_var_begin, log_var_end)) {
+			atomic_store_u(&log_var->state, LOG_ENABLED,
+			    ATOMIC_RELAXED);
+			return LOG_ENABLED;
+		}
+		if (*segment_end == '\0') {
+			/* Hit the end of the segment string with no match. */
+			atomic_store_u(&log_var->state,
+			    LOG_INITIALIZED_NOT_ENABLED, ATOMIC_RELAXED);
+			return LOG_INITIALIZED_NOT_ENABLED;
+		}
+		/* Otherwise, skip the delimiter and continue. */
+		segment_begin = segment_end + 1;
+	}
+}
diff --git a/jemalloc/src/malloc_io.c b/jemalloc/src/malloc_io.c
new file mode 100644
index 0000000..d7cb0f5
--- /dev/null
+++ b/jemalloc/src/malloc_io.c
@@ -0,0 +1,675 @@
+#define JEMALLOC_MALLOC_IO_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/util.h"
+
+#ifdef assert
+#  undef assert
+#endif
+#ifdef not_reached
+#  undef not_reached
+#endif
+#ifdef not_implemented
+#  undef not_implemented
+#endif
+#ifdef assert_not_implemented
+#  undef assert_not_implemented
+#endif
+
+/*
+ * Define simple versions of assertion macros that won't recurse in case
+ * of assertion failures in malloc_*printf().
+ */
+#define assert(e) do {							\
+	if (config_debug && !(e)) {					\
+		malloc_write("<jemalloc>: Failed assertion\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define not_reached() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Unreachable code reached\n");	\
+		abort();						\
+	}								\
+	unreachable();							\
+} while (0)
+
+#define not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Not implemented\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define assert_not_implemented(e) do {					\
+	if (unlikely(config_debug && !(e))) {				\
+		not_implemented();					\
+	}								\
+} while (0)
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void wrtmessage(void *cbopaque, const char *s);
+#define U2S_BUFSIZE ((1U << (LG_SIZEOF_INTMAX_T + 3)) + 1)
+static char *u2s(uintmax_t x, unsigned base, bool uppercase, char *s,
+    size_t *slen_p);
+#define D2S_BUFSIZE (1 + U2S_BUFSIZE)
+static char *d2s(intmax_t x, char sign, char *s, size_t *slen_p);
+#define O2S_BUFSIZE (1 + U2S_BUFSIZE)
+static char *o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p);
+#define X2S_BUFSIZE (2 + U2S_BUFSIZE)
+static char *x2s(uintmax_t x, bool alt_form, bool uppercase, char *s,
+    size_t *slen_p);
+
+/******************************************************************************/
+
+/* malloc_message() setup. */
+static void
+wrtmessage(void *cbopaque, const char *s) {
+	malloc_write_fd(STDERR_FILENO, s, strlen(s));
+}
+
+JEMALLOC_EXPORT void	(*je_malloc_message)(void *, const char *s);
+
+/*
+ * Wrapper around malloc_message() that avoids the need for
+ * je_malloc_message(...) throughout the code.
+ */
+void
+malloc_write(const char *s) {
+	if (je_malloc_message != NULL) {
+		je_malloc_message(NULL, s);
+	} else {
+		wrtmessage(NULL, s);
+	}
+}
+
+/*
+ * glibc provides a non-standard strerror_r() when _GNU_SOURCE is defined, so
+ * provide a wrapper.
+ */
+int
+buferror(int err, char *buf, size_t buflen) {
+#ifdef _WIN32
+	FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM, NULL, err, 0,
+	    (LPSTR)buf, (DWORD)buflen, NULL);
+	return 0;
+#elif defined(JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE) && defined(_GNU_SOURCE)
+	char *b = strerror_r(err, buf, buflen);
+	if (b != buf) {
+		strncpy(buf, b, buflen);
+		buf[buflen-1] = '\0';
+	}
+	return 0;
+#else
+	return strerror_r(err, buf, buflen);
+#endif
+}
+
+uintmax_t
+malloc_strtoumax(const char *restrict nptr, char **restrict endptr, int base) {
+	uintmax_t ret, digit;
+	unsigned b;
+	bool neg;
+	const char *p, *ns;
+
+	p = nptr;
+	if (base < 0 || base == 1 || base > 36) {
+		ns = p;
+		set_errno(EINVAL);
+		ret = UINTMAX_MAX;
+		goto label_return;
+	}
+	b = base;
+
+	/* Swallow leading whitespace and get sign, if any. */
+	neg = false;
+	while (true) {
+		switch (*p) {
+		case '\t': case '\n': case '\v': case '\f': case '\r': case ' ':
+			p++;
+			break;
+		case '-':
+			neg = true;
+			/* Fall through. */
+		case '+':
+			p++;
+			/* Fall through. */
+		default:
+			goto label_prefix;
+		}
+	}
+
+	/* Get prefix, if any. */
+	label_prefix:
+	/*
+	 * Note where the first non-whitespace/sign character is so that it is
+	 * possible to tell whether any digits are consumed (e.g., "  0" vs.
+	 * "  -x").
+	 */
+	ns = p;
+	if (*p == '0') {
+		switch (p[1]) {
+		case '0': case '1': case '2': case '3': case '4': case '5':
+		case '6': case '7':
+			if (b == 0) {
+				b = 8;
+			}
+			if (b == 8) {
+				p++;
+			}
+			break;
+		case 'X': case 'x':
+			switch (p[2]) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+				if (b == 0) {
+					b = 16;
+				}
+				if (b == 16) {
+					p += 2;
+				}
+				break;
+			default:
+				break;
+			}
+			break;
+		default:
+			p++;
+			ret = 0;
+			goto label_return;
+		}
+	}
+	if (b == 0) {
+		b = 10;
+	}
+
+	/* Convert. */
+	ret = 0;
+	while ((*p >= '0' && *p <= '9' && (digit = *p - '0') < b)
+	    || (*p >= 'A' && *p <= 'Z' && (digit = 10 + *p - 'A') < b)
+	    || (*p >= 'a' && *p <= 'z' && (digit = 10 + *p - 'a') < b)) {
+		uintmax_t pret = ret;
+		ret *= b;
+		ret += digit;
+		if (ret < pret) {
+			/* Overflow. */
+			set_errno(ERANGE);
+			ret = UINTMAX_MAX;
+			goto label_return;
+		}
+		p++;
+	}
+	if (neg) {
+		ret = (uintmax_t)(-((intmax_t)ret));
+	}
+
+	if (p == ns) {
+		/* No conversion performed. */
+		set_errno(EINVAL);
+		ret = UINTMAX_MAX;
+		goto label_return;
+	}
+
+label_return:
+	if (endptr != NULL) {
+		if (p == ns) {
+			/* No characters were converted. */
+			*endptr = (char *)nptr;
+		} else {
+			*endptr = (char *)p;
+		}
+	}
+	return ret;
+}
+
+static char *
+u2s(uintmax_t x, unsigned base, bool uppercase, char *s, size_t *slen_p) {
+	unsigned i;
+
+	i = U2S_BUFSIZE - 1;
+	s[i] = '\0';
+	switch (base) {
+	case 10:
+		do {
+			i--;
+			s[i] = "0123456789"[x % (uint64_t)10];
+			x /= (uint64_t)10;
+		} while (x > 0);
+		break;
+	case 16: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEF"
+		    : "0123456789abcdef";
+
+		do {
+			i--;
+			s[i] = digits[x & 0xf];
+			x >>= 4;
+		} while (x > 0);
+		break;
+	} default: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+		    : "0123456789abcdefghijklmnopqrstuvwxyz";
+
+		assert(base >= 2 && base <= 36);
+		do {
+			i--;
+			s[i] = digits[x % (uint64_t)base];
+			x /= (uint64_t)base;
+		} while (x > 0);
+	}}
+
+	*slen_p = U2S_BUFSIZE - 1 - i;
+	return &s[i];
+}
+
+static char *
+d2s(intmax_t x, char sign, char *s, size_t *slen_p) {
+	bool neg;
+
+	if ((neg = (x < 0))) {
+		x = -x;
+	}
+	s = u2s(x, 10, false, s, slen_p);
+	if (neg) {
+		sign = '-';
+	}
+	switch (sign) {
+	case '-':
+		if (!neg) {
+			break;
+		}
+		/* Fall through. */
+	case ' ':
+	case '+':
+		s--;
+		(*slen_p)++;
+		*s = sign;
+		break;
+	default: not_reached();
+	}
+	return s;
+}
+
+static char *
+o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p) {
+	s = u2s(x, 8, false, s, slen_p);
+	if (alt_form && *s != '0') {
+		s--;
+		(*slen_p)++;
+		*s = '0';
+	}
+	return s;
+}
+
+static char *
+x2s(uintmax_t x, bool alt_form, bool uppercase, char *s, size_t *slen_p) {
+	s = u2s(x, 16, uppercase, s, slen_p);
+	if (alt_form) {
+		s -= 2;
+		(*slen_p) += 2;
+		memcpy(s, uppercase ? "0X" : "0x", 2);
+	}
+	return s;
+}
+
+size_t
+malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) {
+	size_t i;
+	const char *f;
+
+#define APPEND_C(c) do {						\
+	if (i < size) {							\
+		str[i] = (c);						\
+	}								\
+	i++;								\
+} while (0)
+#define APPEND_S(s, slen) do {						\
+	if (i < size) {							\
+		size_t cpylen = (slen <= size - i) ? slen : size - i;	\
+		memcpy(&str[i], s, cpylen);				\
+	}								\
+	i += slen;							\
+} while (0)
+#define APPEND_PADDED_S(s, slen, width, left_justify) do {		\
+	/* Left padding. */						\
+	size_t pad_len = (width == -1) ? 0 : ((slen < (size_t)width) ?	\
+	    (size_t)width - slen : 0);					\
+	if (!left_justify && pad_len != 0) {				\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++) {				\
+			APPEND_C(' ');					\
+		}							\
+	}								\
+	/* Value. */							\
+	APPEND_S(s, slen);						\
+	/* Right padding. */						\
+	if (left_justify && pad_len != 0) {				\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++) {				\
+			APPEND_C(' ');					\
+		}							\
+	}								\
+} while (0)
+#define GET_ARG_NUMERIC(val, len) do {					\
+	switch ((unsigned char)len) {					\
+	case '?':							\
+		val = va_arg(ap, int);					\
+		break;							\
+	case '?' | 0x80:						\
+		val = va_arg(ap, unsigned int);				\
+		break;							\
+	case 'l':							\
+		val = va_arg(ap, long);					\
+		break;							\
+	case 'l' | 0x80:						\
+		val = va_arg(ap, unsigned long);			\
+		break;							\
+	case 'q':							\
+		val = va_arg(ap, long long);				\
+		break;							\
+	case 'q' | 0x80:						\
+		val = va_arg(ap, unsigned long long);			\
+		break;							\
+	case 'j':							\
+		val = va_arg(ap, intmax_t);				\
+		break;							\
+	case 'j' | 0x80:						\
+		val = va_arg(ap, uintmax_t);				\
+		break;							\
+	case 't':							\
+		val = va_arg(ap, ptrdiff_t);				\
+		break;							\
+	case 'z':							\
+		val = va_arg(ap, ssize_t);				\
+		break;							\
+	case 'z' | 0x80:						\
+		val = va_arg(ap, size_t);				\
+		break;							\
+	case 'p': /* Synthetic; used for %p. */				\
+		val = va_arg(ap, uintptr_t);				\
+		break;							\
+	default:							\
+		not_reached();						\
+		val = 0;						\
+	}								\
+} while (0)
+
+	i = 0;
+	f = format;
+	while (true) {
+		switch (*f) {
+		case '\0': goto label_out;
+		case '%': {
+			bool alt_form = false;
+			bool left_justify = false;
+			bool plus_space = false;
+			bool plus_plus = false;
+			int prec = -1;
+			int width = -1;
+			unsigned char len = '?';
+			char *s;
+			size_t slen;
+
+			f++;
+			/* Flags. */
+			while (true) {
+				switch (*f) {
+				case '#':
+					assert(!alt_form);
+					alt_form = true;
+					break;
+				case '-':
+					assert(!left_justify);
+					left_justify = true;
+					break;
+				case ' ':
+					assert(!plus_space);
+					plus_space = true;
+					break;
+				case '+':
+					assert(!plus_plus);
+					plus_plus = true;
+					break;
+				default: goto label_width;
+				}
+				f++;
+			}
+			/* Width. */
+			label_width:
+			switch (*f) {
+			case '*':
+				width = va_arg(ap, int);
+				f++;
+				if (width < 0) {
+					left_justify = true;
+					width = -width;
+				}
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uwidth;
+				set_errno(0);
+				uwidth = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uwidth != UINTMAX_MAX || get_errno() !=
+				    ERANGE);
+				width = (int)uwidth;
+				break;
+			} default:
+				break;
+			}
+			/* Width/precision separator. */
+			if (*f == '.') {
+				f++;
+			} else {
+				goto label_length;
+			}
+			/* Precision. */
+			switch (*f) {
+			case '*':
+				prec = va_arg(ap, int);
+				f++;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uprec;
+				set_errno(0);
+				uprec = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uprec != UINTMAX_MAX || get_errno() !=
+				    ERANGE);
+				prec = (int)uprec;
+				break;
+			}
+			default: break;
+			}
+			/* Length. */
+			label_length:
+			switch (*f) {
+			case 'l':
+				f++;
+				if (*f == 'l') {
+					len = 'q';
+					f++;
+				} else {
+					len = 'l';
+				}
+				break;
+			case 'q': case 'j': case 't': case 'z':
+				len = *f;
+				f++;
+				break;
+			default: break;
+			}
+			/* Conversion specifier. */
+			switch (*f) {
+			case '%':
+				/* %% */
+				APPEND_C(*f);
+				f++;
+				break;
+			case 'd': case 'i': {
+				intmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[D2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len);
+				s = d2s(val, (plus_plus ? '+' : (plus_space ?
+				    ' ' : '-')), buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'o': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[O2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = o2s(val, alt_form, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'u': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[U2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = u2s(val, 10, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'x': case 'X': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = x2s(val, alt_form, *f == 'X', buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'c': {
+				unsigned char val;
+				char buf[2];
+
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				val = va_arg(ap, int);
+				buf[0] = val;
+				buf[1] = '\0';
+				APPEND_PADDED_S(buf, 1, width, left_justify);
+				f++;
+				break;
+			} case 's':
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				s = va_arg(ap, char *);
+				slen = (prec < 0) ? strlen(s) : (size_t)prec;
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			case 'p': {
+				uintmax_t val;
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, 'p');
+				s = x2s(val, true, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} default: not_reached();
+			}
+			break;
+		} default: {
+			APPEND_C(*f);
+			f++;
+			break;
+		}}
+	}
+	label_out:
+	if (i < size) {
+		str[i] = '\0';
+	} else {
+		str[size - 1] = '\0';
+	}
+
+#undef APPEND_C
+#undef APPEND_S
+#undef APPEND_PADDED_S
+#undef GET_ARG_NUMERIC
+	return i;
+}
+
+JEMALLOC_FORMAT_PRINTF(3, 4)
+size_t
+malloc_snprintf(char *str, size_t size, const char *format, ...) {
+	size_t ret;
+	va_list ap;
+
+	va_start(ap, format);
+	ret = malloc_vsnprintf(str, size, format, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+void
+malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap) {
+	char buf[MALLOC_PRINTF_BUFSIZE];
+
+	if (write_cb == NULL) {
+		/*
+		 * The caller did not provide an alternate write_cb callback
+		 * function, so use the default one.  malloc_write() is an
+		 * inline function, so use malloc_message() directly here.
+		 */
+		write_cb = (je_malloc_message != NULL) ? je_malloc_message :
+		    wrtmessage;
+	}
+
+	malloc_vsnprintf(buf, sizeof(buf), format, ap);
+	write_cb(cbopaque, buf);
+}
+
+/*
+ * Print to a callback function in such a way as to (hopefully) avoid memory
+ * allocation.
+ */
+JEMALLOC_FORMAT_PRINTF(3, 4)
+void
+malloc_cprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(write_cb, cbopaque, format, ap);
+	va_end(ap);
+}
+
+/* Print to stderr in such a way as to avoid memory allocation. */
+JEMALLOC_FORMAT_PRINTF(1, 2)
+void
+malloc_printf(const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+}
+
+/*
+ * Restore normal assertion macros, in order to make it possible to compile all
+ * C files as a single concatenation.
+ */
+#undef assert
+#undef not_reached
+#undef not_implemented
+#undef assert_not_implemented
+#include "jemalloc/internal/assert.h"
diff --git a/jemalloc/src/mutex.c b/jemalloc/src/mutex.c
new file mode 100644
index 0000000..3f920f5
--- /dev/null
+++ b/jemalloc/src/mutex.c
@@ -0,0 +1,223 @@
+#define JEMALLOC_MUTEX_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/spin.h"
+
+#ifndef _CRT_SPINCOUNT
+#define _CRT_SPINCOUNT 4000
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+#ifdef JEMALLOC_LAZY_LOCK
+bool isthreaded = false;
+#endif
+#ifdef JEMALLOC_MUTEX_INIT_CB
+static bool		postpone_init = true;
+static malloc_mutex_t	*postponed_mutexes = NULL;
+#endif
+
+/******************************************************************************/
+/*
+ * We intercept pthread_create() calls in order to toggle isthreaded if the
+ * process goes multi-threaded.
+ */
+
+#if defined(JEMALLOC_LAZY_LOCK) && !defined(_WIN32)
+JEMALLOC_EXPORT int
+pthread_create(pthread_t *__restrict thread,
+    const pthread_attr_t *__restrict attr, void *(*start_routine)(void *),
+    void *__restrict arg) {
+	return pthread_create_wrapper(thread, attr, start_routine, arg);
+}
+#endif
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+JEMALLOC_EXPORT int	_pthread_mutex_init_calloc_cb(pthread_mutex_t *mutex,
+    void *(calloc_cb)(size_t, size_t));
+#endif
+
+void
+malloc_mutex_lock_slow(malloc_mutex_t *mutex) {
+	mutex_prof_data_t *data = &mutex->prof_data;
+	nstime_t before = NSTIME_ZERO_INITIALIZER;
+
+	if (ncpus == 1) {
+		goto label_spin_done;
+	}
+
+	int cnt = 0, max_cnt = MALLOC_MUTEX_MAX_SPIN;
+	do {
+		spin_cpu_spinwait();
+		if (!atomic_load_b(&mutex->locked, ATOMIC_RELAXED)
+                    && !malloc_mutex_trylock_final(mutex)) {
+			data->n_spin_acquired++;
+			return;
+		}
+	} while (cnt++ < max_cnt);
+
+	if (!config_stats) {
+		/* Only spin is useful when stats is off. */
+		malloc_mutex_lock_final(mutex);
+		return;
+	}
+label_spin_done:
+	nstime_update(&before);
+	/* Copy before to after to avoid clock skews. */
+	nstime_t after;
+	nstime_copy(&after, &before);
+	uint32_t n_thds = atomic_fetch_add_u32(&data->n_waiting_thds, 1,
+	    ATOMIC_RELAXED) + 1;
+	/* One last try as above two calls may take quite some cycles. */
+	if (!malloc_mutex_trylock_final(mutex)) {
+		atomic_fetch_sub_u32(&data->n_waiting_thds, 1, ATOMIC_RELAXED);
+		data->n_spin_acquired++;
+		return;
+	}
+
+	/* True slow path. */
+	malloc_mutex_lock_final(mutex);
+	/* Update more slow-path only counters. */
+	atomic_fetch_sub_u32(&data->n_waiting_thds, 1, ATOMIC_RELAXED);
+	nstime_update(&after);
+
+	nstime_t delta;
+	nstime_copy(&delta, &after);
+	nstime_subtract(&delta, &before);
+
+	data->n_wait_times++;
+	nstime_add(&data->tot_wait_time, &delta);
+	if (nstime_compare(&data->max_wait_time, &delta) < 0) {
+		nstime_copy(&data->max_wait_time, &delta);
+	}
+	if (n_thds > data->max_n_thds) {
+		data->max_n_thds = n_thds;
+	}
+}
+
+static void
+mutex_prof_data_init(mutex_prof_data_t *data) {
+	memset(data, 0, sizeof(mutex_prof_data_t));
+	nstime_init(&data->max_wait_time, 0);
+	nstime_init(&data->tot_wait_time, 0);
+	data->prev_owner = NULL;
+}
+
+void
+malloc_mutex_prof_data_reset(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	malloc_mutex_assert_owner(tsdn, mutex);
+	mutex_prof_data_init(&mutex->prof_data);
+}
+
+static int
+mutex_addr_comp(const witness_t *witness1, void *mutex1,
+    const witness_t *witness2, void *mutex2) {
+	assert(mutex1 != NULL);
+	assert(mutex2 != NULL);
+	uintptr_t mu1int = (uintptr_t)mutex1;
+	uintptr_t mu2int = (uintptr_t)mutex2;
+	if (mu1int < mu2int) {
+		return -1;
+	} else if (mu1int == mu2int) {
+		return 0;
+	} else {
+		return 1;
+	}
+}
+
+bool
+malloc_mutex_init(malloc_mutex_t *mutex, const char *name,
+    witness_rank_t rank, malloc_mutex_lock_order_t lock_order) {
+	mutex_prof_data_init(&mutex->prof_data);
+#ifdef _WIN32
+#  if _WIN32_WINNT >= 0x0600
+	InitializeSRWLock(&mutex->lock);
+#  else
+	if (!InitializeCriticalSectionAndSpinCount(&mutex->lock,
+	    _CRT_SPINCOUNT)) {
+		return true;
+	}
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+       mutex->lock = OS_UNFAIR_LOCK_INIT;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+	if (postpone_init) {
+		mutex->postponed_next = postponed_mutexes;
+		postponed_mutexes = mutex;
+	} else {
+		if (_pthread_mutex_init_calloc_cb(&mutex->lock,
+		    bootstrap_calloc) != 0) {
+			return true;
+		}
+	}
+#else
+	pthread_mutexattr_t attr;
+
+	if (pthread_mutexattr_init(&attr) != 0) {
+		return true;
+	}
+	pthread_mutexattr_settype(&attr, MALLOC_MUTEX_TYPE);
+	if (pthread_mutex_init(&mutex->lock, &attr) != 0) {
+		pthread_mutexattr_destroy(&attr);
+		return true;
+	}
+	pthread_mutexattr_destroy(&attr);
+#endif
+	if (config_debug) {
+		mutex->lock_order = lock_order;
+		if (lock_order == malloc_mutex_address_ordered) {
+			witness_init(&mutex->witness, name, rank,
+			    mutex_addr_comp, mutex);
+		} else {
+			witness_init(&mutex->witness, name, rank, NULL, NULL);
+		}
+	}
+	return false;
+}
+
+void
+malloc_mutex_prefork(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	malloc_mutex_lock(tsdn, mutex);
+}
+
+void
+malloc_mutex_postfork_parent(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	malloc_mutex_unlock(tsdn, mutex);
+}
+
+void
+malloc_mutex_postfork_child(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	malloc_mutex_unlock(tsdn, mutex);
+#else
+	if (malloc_mutex_init(mutex, mutex->witness.name,
+	    mutex->witness.rank, mutex->lock_order)) {
+		malloc_printf("<jemalloc>: Error re-initializing mutex in "
+		    "child\n");
+		if (opt_abort) {
+			abort();
+		}
+	}
+#endif
+}
+
+bool
+malloc_mutex_boot(void) {
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	postpone_init = false;
+	while (postponed_mutexes != NULL) {
+		if (_pthread_mutex_init_calloc_cb(&postponed_mutexes->lock,
+		    bootstrap_calloc) != 0) {
+			return true;
+		}
+		postponed_mutexes = postponed_mutexes->postponed_next;
+	}
+#endif
+	return false;
+}
diff --git a/jemalloc/src/mutex_pool.c b/jemalloc/src/mutex_pool.c
new file mode 100644
index 0000000..f24d10e
--- /dev/null
+++ b/jemalloc/src/mutex_pool.c
@@ -0,0 +1,18 @@
+#define JEMALLOC_MUTEX_POOL_C_
+
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+
+bool
+mutex_pool_init(mutex_pool_t *pool, const char *name, witness_rank_t rank) {
+	for (int i = 0; i < MUTEX_POOL_SIZE; ++i) {
+		if (malloc_mutex_init(&pool->mutexes[i], name, rank,
+		    malloc_mutex_address_ordered)) {
+			return true;
+		}
+	}
+	return false;
+}
diff --git a/jemalloc/src/nstime.c b/jemalloc/src/nstime.c
new file mode 100644
index 0000000..71db353
--- /dev/null
+++ b/jemalloc/src/nstime.c
@@ -0,0 +1,170 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/nstime.h"
+
+#include "jemalloc/internal/assert.h"
+
+#define BILLION	UINT64_C(1000000000)
+#define MILLION	UINT64_C(1000000)
+
+void
+nstime_init(nstime_t *time, uint64_t ns) {
+	time->ns = ns;
+}
+
+void
+nstime_init2(nstime_t *time, uint64_t sec, uint64_t nsec) {
+	time->ns = sec * BILLION + nsec;
+}
+
+uint64_t
+nstime_ns(const nstime_t *time) {
+	return time->ns;
+}
+
+uint64_t
+nstime_msec(const nstime_t *time) {
+	return time->ns / MILLION;
+}
+
+uint64_t
+nstime_sec(const nstime_t *time) {
+	return time->ns / BILLION;
+}
+
+uint64_t
+nstime_nsec(const nstime_t *time) {
+	return time->ns % BILLION;
+}
+
+void
+nstime_copy(nstime_t *time, const nstime_t *source) {
+	*time = *source;
+}
+
+int
+nstime_compare(const nstime_t *a, const nstime_t *b) {
+	return (a->ns > b->ns) - (a->ns < b->ns);
+}
+
+void
+nstime_add(nstime_t *time, const nstime_t *addend) {
+	assert(UINT64_MAX - time->ns >= addend->ns);
+
+	time->ns += addend->ns;
+}
+
+void
+nstime_iadd(nstime_t *time, uint64_t addend) {
+	assert(UINT64_MAX - time->ns >= addend);
+
+	time->ns += addend;
+}
+
+void
+nstime_subtract(nstime_t *time, const nstime_t *subtrahend) {
+	assert(nstime_compare(time, subtrahend) >= 0);
+
+	time->ns -= subtrahend->ns;
+}
+
+void
+nstime_isubtract(nstime_t *time, uint64_t subtrahend) {
+	assert(time->ns >= subtrahend);
+
+	time->ns -= subtrahend;
+}
+
+void
+nstime_imultiply(nstime_t *time, uint64_t multiplier) {
+	assert((((time->ns | multiplier) & (UINT64_MAX << (sizeof(uint64_t) <<
+	    2))) == 0) || ((time->ns * multiplier) / multiplier == time->ns));
+
+	time->ns *= multiplier;
+}
+
+void
+nstime_idivide(nstime_t *time, uint64_t divisor) {
+	assert(divisor != 0);
+
+	time->ns /= divisor;
+}
+
+uint64_t
+nstime_divide(const nstime_t *time, const nstime_t *divisor) {
+	assert(divisor->ns != 0);
+
+	return time->ns / divisor->ns;
+}
+
+#ifdef _WIN32
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	FILETIME ft;
+	uint64_t ticks_100ns;
+
+	GetSystemTimeAsFileTime(&ft);
+	ticks_100ns = (((uint64_t)ft.dwHighDateTime) << 32) | ft.dwLowDateTime;
+
+	nstime_init(time, ticks_100ns * 100);
+}
+#elif defined(JEMALLOC_HAVE_CLOCK_MONOTONIC_COARSE)
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC_COARSE, &ts);
+	nstime_init2(time, ts.tv_sec, ts.tv_nsec);
+}
+#elif defined(JEMALLOC_HAVE_CLOCK_MONOTONIC)
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts);
+	nstime_init2(time, ts.tv_sec, ts.tv_nsec);
+}
+#elif defined(JEMALLOC_HAVE_MACH_ABSOLUTE_TIME)
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	nstime_init(time, mach_absolute_time());
+}
+#else
+#  define NSTIME_MONOTONIC false
+static void
+nstime_get(nstime_t *time) {
+	struct timeval tv;
+
+	gettimeofday(&tv, NULL);
+	nstime_init2(time, tv.tv_sec, tv.tv_usec * 1000);
+}
+#endif
+
+static bool
+nstime_monotonic_impl(void) {
+	return NSTIME_MONOTONIC;
+#undef NSTIME_MONOTONIC
+}
+nstime_monotonic_t *JET_MUTABLE nstime_monotonic = nstime_monotonic_impl;
+
+static bool
+nstime_update_impl(nstime_t *time) {
+	nstime_t old_time;
+
+	nstime_copy(&old_time, time);
+	nstime_get(time);
+
+	/* Handle non-monotonic clocks. */
+	if (unlikely(nstime_compare(&old_time, time) > 0)) {
+		nstime_copy(time, &old_time);
+		return true;
+	}
+
+	return false;
+}
+nstime_update_t *JET_MUTABLE nstime_update = nstime_update_impl;
diff --git a/jemalloc/src/pages.c b/jemalloc/src/pages.c
new file mode 100644
index 0000000..b601578
--- /dev/null
+++ b/jemalloc/src/pages.c
@@ -0,0 +1,649 @@
+#define JEMALLOC_PAGES_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/pages.h"
+
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/malloc_io.h"
+
+#ifdef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+#include <sys/sysctl.h>
+#ifdef __FreeBSD__
+#include <vm/vm_param.h>
+#endif
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+/* Actual operating system page size, detected during bootstrap, <= PAGE. */
+static size_t	os_page;
+
+#ifndef _WIN32
+#  define PAGES_PROT_COMMIT (PROT_READ | PROT_WRITE)
+#  define PAGES_PROT_DECOMMIT (PROT_NONE)
+static int	mmap_flags;
+#endif
+static bool	os_overcommits;
+
+const char *thp_mode_names[] = {
+	"default",
+	"always",
+	"never",
+	"not supported"
+};
+thp_mode_t opt_thp = THP_MODE_DEFAULT;
+thp_mode_t init_system_thp_mode;
+
+/* Runtime support for lazy purge. Irrelevant when !pages_can_purge_lazy. */
+static bool pages_can_purge_lazy_runtime = true;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static void os_pages_unmap(void *addr, size_t size);
+
+/******************************************************************************/
+
+static void *
+os_pages_map(void *addr, size_t size, size_t alignment, bool *commit) {
+	assert(ALIGNMENT_ADDR2BASE(addr, os_page) == addr);
+	assert(ALIGNMENT_CEILING(size, os_page) == size);
+	assert(size != 0);
+
+	if (os_overcommits) {
+		*commit = true;
+	}
+
+	void *ret;
+#ifdef _WIN32
+	/*
+	 * If VirtualAlloc can't allocate at the given address when one is
+	 * given, it fails and returns NULL.
+	 */
+	ret = VirtualAlloc(addr, size, MEM_RESERVE | (*commit ? MEM_COMMIT : 0),
+	    PAGE_READWRITE);
+#else
+	/*
+	 * We don't use MAP_FIXED here, because it can cause the *replacement*
+	 * of existing mappings, and we only want to create new mappings.
+	 */
+	{
+		int prot = *commit ? PAGES_PROT_COMMIT : PAGES_PROT_DECOMMIT;
+
+		ret = mmap(addr, size, prot, mmap_flags, -1, 0);
+	}
+	assert(ret != NULL);
+
+	if (ret == MAP_FAILED) {
+		ret = NULL;
+	} else if (addr != NULL && ret != addr) {
+		/*
+		 * We succeeded in mapping memory, but not in the right place.
+		 */
+		os_pages_unmap(ret, size);
+		ret = NULL;
+	}
+#endif
+	assert(ret == NULL || (addr == NULL && ret != addr) || (addr != NULL &&
+	    ret == addr));
+	return ret;
+}
+
+static void *
+os_pages_trim(void *addr, size_t alloc_size, size_t leadsize, size_t size,
+    bool *commit) {
+	void *ret = (void *)((uintptr_t)addr + leadsize);
+
+	assert(alloc_size >= leadsize + size);
+#ifdef _WIN32
+	os_pages_unmap(addr, alloc_size);
+	void *new_addr = os_pages_map(ret, size, PAGE, commit);
+	if (new_addr == ret) {
+		return ret;
+	}
+	if (new_addr != NULL) {
+		os_pages_unmap(new_addr, size);
+	}
+	return NULL;
+#else
+	size_t trailsize = alloc_size - leadsize - size;
+
+	if (leadsize != 0) {
+		os_pages_unmap(addr, leadsize);
+	}
+	if (trailsize != 0) {
+		os_pages_unmap((void *)((uintptr_t)ret + size), trailsize);
+	}
+	return ret;
+#endif
+}
+
+static void
+os_pages_unmap(void *addr, size_t size) {
+	assert(ALIGNMENT_ADDR2BASE(addr, os_page) == addr);
+	assert(ALIGNMENT_CEILING(size, os_page) == size);
+
+#ifdef _WIN32
+	if (VirtualFree(addr, 0, MEM_RELEASE) == 0)
+#else
+	if (munmap(addr, size) == -1)
+#endif
+	{
+		char buf[BUFERROR_BUF];
+
+		buferror(get_errno(), buf, sizeof(buf));
+		malloc_printf("<jemalloc>: Error in "
+#ifdef _WIN32
+		    "VirtualFree"
+#else
+		    "munmap"
+#endif
+		    "(): %s\n", buf);
+		if (opt_abort) {
+			abort();
+		}
+	}
+}
+
+static void *
+pages_map_slow(size_t size, size_t alignment, bool *commit) {
+	size_t alloc_size = size + alignment - os_page;
+	/* Beware size_t wrap-around. */
+	if (alloc_size < size) {
+		return NULL;
+	}
+
+	void *ret;
+	do {
+		void *pages = os_pages_map(NULL, alloc_size, alignment, commit);
+		if (pages == NULL) {
+			return NULL;
+		}
+		size_t leadsize = ALIGNMENT_CEILING((uintptr_t)pages, alignment)
+		    - (uintptr_t)pages;
+		ret = os_pages_trim(pages, alloc_size, leadsize, size, commit);
+	} while (ret == NULL);
+
+	assert(ret != NULL);
+	assert(PAGE_ADDR2BASE(ret) == ret);
+	return ret;
+}
+
+void *
+pages_map(void *addr, size_t size, size_t alignment, bool *commit) {
+	assert(alignment >= PAGE);
+	assert(ALIGNMENT_ADDR2BASE(addr, alignment) == addr);
+
+#if defined(__FreeBSD__) && defined(MAP_EXCL)
+	/*
+	 * FreeBSD has mechanisms both to mmap at specific address without
+	 * touching existing mappings, and to mmap with specific alignment.
+	 */
+	{
+		if (os_overcommits) {
+			*commit = true;
+		}
+
+		int prot = *commit ? PAGES_PROT_COMMIT : PAGES_PROT_DECOMMIT;
+		int flags = mmap_flags;
+
+		if (addr != NULL) {
+			flags |= MAP_FIXED | MAP_EXCL;
+		} else {
+			unsigned alignment_bits = ffs_zu(alignment);
+			assert(alignment_bits > 1);
+			flags |= MAP_ALIGNED(alignment_bits - 1);
+		}
+
+		void *ret = mmap(addr, size, prot, flags, -1, 0);
+		if (ret == MAP_FAILED) {
+			ret = NULL;
+		}
+
+		return ret;
+	}
+#endif
+	/*
+	 * Ideally, there would be a way to specify alignment to mmap() (like
+	 * NetBSD has), but in the absence of such a feature, we have to work
+	 * hard to efficiently create aligned mappings.  The reliable, but
+	 * slow method is to create a mapping that is over-sized, then trim the
+	 * excess.  However, that always results in one or two calls to
+	 * os_pages_unmap(), and it can leave holes in the process's virtual
+	 * memory map if memory grows downward.
+	 *
+	 * Optimistically try mapping precisely the right amount before falling
+	 * back to the slow method, with the expectation that the optimistic
+	 * approach works most of the time.
+	 */
+
+	void *ret = os_pages_map(addr, size, os_page, commit);
+	if (ret == NULL || ret == addr) {
+		return ret;
+	}
+	assert(addr == NULL);
+	if (ALIGNMENT_ADDR2OFFSET(ret, alignment) != 0) {
+		os_pages_unmap(ret, size);
+		return pages_map_slow(size, alignment, commit);
+	}
+
+	assert(PAGE_ADDR2BASE(ret) == ret);
+	return ret;
+}
+
+void
+pages_unmap(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	os_pages_unmap(addr, size);
+}
+
+static bool
+pages_commit_impl(void *addr, size_t size, bool commit) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	if (os_overcommits) {
+		return true;
+	}
+
+#ifdef _WIN32
+	return (commit ? (addr != VirtualAlloc(addr, size, MEM_COMMIT,
+	    PAGE_READWRITE)) : (!VirtualFree(addr, size, MEM_DECOMMIT)));
+#else
+	{
+		int prot = commit ? PAGES_PROT_COMMIT : PAGES_PROT_DECOMMIT;
+		void *result = mmap(addr, size, prot, mmap_flags | MAP_FIXED,
+		    -1, 0);
+		if (result == MAP_FAILED) {
+			return true;
+		}
+		if (result != addr) {
+			/*
+			 * We succeeded in mapping memory, but not in the right
+			 * place.
+			 */
+			os_pages_unmap(result, size);
+			return true;
+		}
+		return false;
+	}
+#endif
+}
+
+bool
+pages_commit(void *addr, size_t size) {
+	return pages_commit_impl(addr, size, true);
+}
+
+bool
+pages_decommit(void *addr, size_t size) {
+	return pages_commit_impl(addr, size, false);
+}
+
+bool
+pages_purge_lazy(void *addr, size_t size) {
+	assert(ALIGNMENT_ADDR2BASE(addr, os_page) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	if (!pages_can_purge_lazy) {
+		return true;
+	}
+	if (!pages_can_purge_lazy_runtime) {
+		/*
+		 * Built with lazy purge enabled, but detected it was not
+		 * supported on the current system.
+		 */
+		return true;
+	}
+
+#ifdef _WIN32
+	VirtualAlloc(addr, size, MEM_RESET, PAGE_READWRITE);
+	return false;
+#elif defined(JEMALLOC_PURGE_MADVISE_FREE)
+	return (madvise(addr, size,
+#  ifdef MADV_FREE
+	    MADV_FREE
+#  else
+	    JEMALLOC_MADV_FREE
+#  endif
+	    ) != 0);
+#elif defined(JEMALLOC_PURGE_MADVISE_DONTNEED) && \
+    !defined(JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS)
+	return (madvise(addr, size, MADV_DONTNEED) != 0);
+#else
+	not_reached();
+#endif
+}
+
+bool
+pages_purge_forced(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	if (!pages_can_purge_forced) {
+		return true;
+	}
+
+#if defined(JEMALLOC_PURGE_MADVISE_DONTNEED) && \
+    defined(JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS)
+	return (madvise(addr, size, MADV_DONTNEED) != 0);
+#elif defined(JEMALLOC_MAPS_COALESCE)
+	/* Try to overlay a new demand-zeroed mapping. */
+	return pages_commit(addr, size);
+#else
+	not_reached();
+#endif
+}
+
+static bool
+pages_huge_impl(void *addr, size_t size, bool aligned) {
+	if (aligned) {
+		assert(HUGEPAGE_ADDR2BASE(addr) == addr);
+		assert(HUGEPAGE_CEILING(size) == size);
+	}
+#ifdef JEMALLOC_HAVE_MADVISE_HUGE
+	return (madvise(addr, size, MADV_HUGEPAGE) != 0);
+#else
+	return true;
+#endif
+}
+
+bool
+pages_huge(void *addr, size_t size) {
+	return pages_huge_impl(addr, size, true);
+}
+
+static bool
+pages_huge_unaligned(void *addr, size_t size) {
+	return pages_huge_impl(addr, size, false);
+}
+
+static bool
+pages_nohuge_impl(void *addr, size_t size, bool aligned) {
+	if (aligned) {
+		assert(HUGEPAGE_ADDR2BASE(addr) == addr);
+		assert(HUGEPAGE_CEILING(size) == size);
+	}
+
+#ifdef JEMALLOC_HAVE_MADVISE_HUGE
+	return (madvise(addr, size, MADV_NOHUGEPAGE) != 0);
+#else
+	return false;
+#endif
+}
+
+bool
+pages_nohuge(void *addr, size_t size) {
+	return pages_nohuge_impl(addr, size, true);
+}
+
+static bool
+pages_nohuge_unaligned(void *addr, size_t size) {
+	return pages_nohuge_impl(addr, size, false);
+}
+
+bool
+pages_dontdump(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+#ifdef JEMALLOC_MADVISE_DONTDUMP
+	return madvise(addr, size, MADV_DONTDUMP) != 0;
+#else
+	return false;
+#endif
+}
+
+bool
+pages_dodump(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+#ifdef JEMALLOC_MADVISE_DONTDUMP
+	return madvise(addr, size, MADV_DODUMP) != 0;
+#else
+	return false;
+#endif
+}
+
+
+static size_t
+os_page_detect(void) {
+#ifdef _WIN32
+	SYSTEM_INFO si;
+	GetSystemInfo(&si);
+	return si.dwPageSize;
+#elif defined(__FreeBSD__)
+	/*
+	 * This returns the value obtained from
+	 * the auxv vector, avoiding a syscall.
+	 */
+	return getpagesize();
+#else
+	long result = sysconf(_SC_PAGESIZE);
+	if (result == -1) {
+		return LG_PAGE;
+	}
+	return (size_t)result;
+#endif
+}
+
+#ifdef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+static bool
+os_overcommits_sysctl(void) {
+	int vm_overcommit;
+	size_t sz;
+
+	sz = sizeof(vm_overcommit);
+#if defined(__FreeBSD__) && defined(VM_OVERCOMMIT)
+	int mib[2];
+
+	mib[0] = CTL_VM;
+	mib[1] = VM_OVERCOMMIT;
+	if (sysctl(mib, 2, &vm_overcommit, &sz, NULL, 0) != 0) {
+		return false; /* Error. */
+	}
+#else
+	if (sysctlbyname("vm.overcommit", &vm_overcommit, &sz, NULL, 0) != 0) {
+		return false; /* Error. */
+	}
+#endif
+
+	return ((vm_overcommit & 0x3) == 0);
+}
+#endif
+
+#ifdef JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY
+/*
+ * Use syscall(2) rather than {open,read,close}(2) when possible to avoid
+ * reentry during bootstrapping if another library has interposed system call
+ * wrappers.
+ */
+static bool
+os_overcommits_proc(void) {
+	int fd;
+	char buf[1];
+
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_open)
+	#if defined(O_CLOEXEC)
+		fd = (int)syscall(SYS_open, "/proc/sys/vm/overcommit_memory", O_RDONLY |
+			O_CLOEXEC);
+	#else
+		fd = (int)syscall(SYS_open, "/proc/sys/vm/overcommit_memory", O_RDONLY);
+		if (fd != -1) {
+			fcntl(fd, F_SETFD, fcntl(fd, F_GETFD) | FD_CLOEXEC);
+		}
+	#endif
+#elif defined(JEMALLOC_USE_SYSCALL) && defined(SYS_openat)
+	#if defined(O_CLOEXEC)
+		fd = (int)syscall(SYS_openat,
+			AT_FDCWD, "/proc/sys/vm/overcommit_memory", O_RDONLY | O_CLOEXEC);
+	#else
+		fd = (int)syscall(SYS_openat,
+			AT_FDCWD, "/proc/sys/vm/overcommit_memory", O_RDONLY);
+		if (fd != -1) {
+			fcntl(fd, F_SETFD, fcntl(fd, F_GETFD) | FD_CLOEXEC);
+		}
+	#endif
+#else
+	#if defined(O_CLOEXEC)
+		fd = open("/proc/sys/vm/overcommit_memory", O_RDONLY | O_CLOEXEC);
+	#else
+		fd = open("/proc/sys/vm/overcommit_memory", O_RDONLY);
+		if (fd != -1) {
+			fcntl(fd, F_SETFD, fcntl(fd, F_GETFD) | FD_CLOEXEC);
+		}
+	#endif
+#endif
+
+	if (fd == -1) {
+		return false; /* Error. */
+	}
+
+	ssize_t nread = malloc_read_fd(fd, &buf, sizeof(buf));
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_close)
+	syscall(SYS_close, fd);
+#else
+	close(fd);
+#endif
+
+	if (nread < 1) {
+		return false; /* Error. */
+	}
+	/*
+	 * /proc/sys/vm/overcommit_memory meanings:
+	 * 0: Heuristic overcommit.
+	 * 1: Always overcommit.
+	 * 2: Never overcommit.
+	 */
+	return (buf[0] == '0' || buf[0] == '1');
+}
+#endif
+
+void
+pages_set_thp_state (void *ptr, size_t size) {
+	if (opt_thp == thp_mode_default || opt_thp == init_system_thp_mode) {
+		return;
+	}
+	assert(opt_thp != thp_mode_not_supported &&
+	    init_system_thp_mode != thp_mode_not_supported);
+
+	if (opt_thp == thp_mode_always
+	    && init_system_thp_mode != thp_mode_never) {
+		assert(init_system_thp_mode == thp_mode_default);
+		pages_huge_unaligned(ptr, size);
+	} else if (opt_thp == thp_mode_never) {
+		assert(init_system_thp_mode == thp_mode_default ||
+		    init_system_thp_mode == thp_mode_always);
+		pages_nohuge_unaligned(ptr, size);
+	}
+}
+
+static void
+init_thp_state(void) {
+	if (!have_madvise_huge) {
+		if (metadata_thp_enabled() && opt_abort) {
+			malloc_write("<jemalloc>: no MADV_HUGEPAGE support\n");
+			abort();
+		}
+		goto label_error;
+	}
+
+	static const char sys_state_madvise[] = "always [madvise] never\n";
+	static const char sys_state_always[] = "[always] madvise never\n";
+	static const char sys_state_never[] = "always madvise [never]\n";
+	char buf[sizeof(sys_state_madvise)];
+
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_open)
+	int fd = (int)syscall(SYS_open,
+	    "/sys/kernel/mm/transparent_hugepage/enabled", O_RDONLY);
+#else
+	int fd = open("/sys/kernel/mm/transparent_hugepage/enabled", O_RDONLY);
+#endif
+	if (fd == -1) {
+		goto label_error;
+	}
+
+	ssize_t nread = malloc_read_fd(fd, &buf, sizeof(buf));
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_close)
+	syscall(SYS_close, fd);
+#else
+	close(fd);
+#endif
+
+        if (nread < 0) {
+		goto label_error;
+        }
+
+	if (strncmp(buf, sys_state_madvise, (size_t)nread) == 0) {
+		init_system_thp_mode = thp_mode_default;
+	} else if (strncmp(buf, sys_state_always, (size_t)nread) == 0) {
+		init_system_thp_mode = thp_mode_always;
+	} else if (strncmp(buf, sys_state_never, (size_t)nread) == 0) {
+		init_system_thp_mode = thp_mode_never;
+	} else {
+		goto label_error;
+	}
+	return;
+label_error:
+	opt_thp = init_system_thp_mode = thp_mode_not_supported;
+}
+
+bool
+pages_boot(void) {
+	os_page = os_page_detect();
+	if (os_page > PAGE) {
+		malloc_write("<jemalloc>: Unsupported system page size\n");
+		if (opt_abort) {
+			abort();
+		}
+		return true;
+	}
+
+#ifndef _WIN32
+	mmap_flags = MAP_PRIVATE | MAP_ANON;
+#endif
+
+#ifdef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+	os_overcommits = os_overcommits_sysctl();
+#elif defined(JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY)
+	os_overcommits = os_overcommits_proc();
+#  ifdef MAP_NORESERVE
+	if (os_overcommits) {
+		mmap_flags |= MAP_NORESERVE;
+	}
+#  endif
+#else
+	os_overcommits = false;
+#endif
+
+	init_thp_state();
+
+#ifdef __FreeBSD__
+	/*
+	 * FreeBSD doesn't need the check; madvise(2) is known to work.
+	 */
+#else
+	/* Detect lazy purge runtime support. */
+	if (pages_can_purge_lazy) {
+		bool committed = false;
+		void *madv_free_page = os_pages_map(NULL, PAGE, PAGE, &committed);
+		if (madv_free_page == NULL) {
+			return true;
+		}
+		assert(pages_can_purge_lazy_runtime);
+		if (pages_purge_lazy(madv_free_page, PAGE)) {
+			pages_can_purge_lazy_runtime = false;
+		}
+		os_pages_unmap(madv_free_page, PAGE);
+	}
+#endif
+
+	return false;
+}
diff --git a/jemalloc/src/prng.c b/jemalloc/src/prng.c
new file mode 100644
index 0000000..83c04bf
--- /dev/null
+++ b/jemalloc/src/prng.c
@@ -0,0 +1,3 @@
+#define JEMALLOC_PRNG_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
diff --git a/jemalloc/src/prof.c b/jemalloc/src/prof.c
new file mode 100644
index 0000000..13334cb
--- /dev/null
+++ b/jemalloc/src/prof.c
@@ -0,0 +1,3160 @@
+#define JEMALLOC_PROF_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/emitter.h"
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_PROF_LIBUNWIND
+#define UNW_LOCAL_ONLY
+#include <libunwind.h>
+#endif
+
+#ifdef JEMALLOC_PROF_LIBGCC
+/*
+ * We have a circular dependency -- jemalloc_internal.h tells us if we should
+ * use libgcc's unwinding functionality, but after we've included that, we've
+ * already hooked _Unwind_Backtrace.  We'll temporarily disable hooking.
+ */
+#undef _Unwind_Backtrace
+#include <unwind.h>
+#define _Unwind_Backtrace JEMALLOC_HOOK(_Unwind_Backtrace, test_hooks_libc_hook)
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+bool		opt_prof = false;
+bool		opt_prof_active = true;
+bool		opt_prof_thread_active_init = true;
+size_t		opt_lg_prof_sample = LG_PROF_SAMPLE_DEFAULT;
+ssize_t		opt_lg_prof_interval = LG_PROF_INTERVAL_DEFAULT;
+bool		opt_prof_gdump = false;
+bool		opt_prof_final = false;
+bool		opt_prof_leak = false;
+bool		opt_prof_accum = false;
+bool		opt_prof_log = false;
+char		opt_prof_prefix[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PATH_MAX +
+#endif
+    1];
+
+/*
+ * Initialized as opt_prof_active, and accessed via
+ * prof_active_[gs]et{_unlocked,}().
+ */
+bool			prof_active;
+static malloc_mutex_t	prof_active_mtx;
+
+/*
+ * Initialized as opt_prof_thread_active_init, and accessed via
+ * prof_thread_active_init_[gs]et().
+ */
+static bool		prof_thread_active_init;
+static malloc_mutex_t	prof_thread_active_init_mtx;
+
+/*
+ * Initialized as opt_prof_gdump, and accessed via
+ * prof_gdump_[gs]et{_unlocked,}().
+ */
+bool			prof_gdump_val;
+static malloc_mutex_t	prof_gdump_mtx;
+
+uint64_t	prof_interval = 0;
+
+size_t		lg_prof_sample;
+
+typedef enum prof_logging_state_e prof_logging_state_t;
+enum prof_logging_state_e {
+	prof_logging_state_stopped,
+	prof_logging_state_started,
+	prof_logging_state_dumping
+};
+
+/*
+ * - stopped: log_start never called, or previous log_stop has completed.
+ * - started: log_start called, log_stop not called yet. Allocations are logged.
+ * - dumping: log_stop called but not finished; samples are not logged anymore.
+ */
+prof_logging_state_t prof_logging_state = prof_logging_state_stopped;
+
+#ifdef JEMALLOC_JET
+static bool prof_log_dummy = false;
+#endif
+
+/* Incremented for every log file that is output. */
+static uint64_t log_seq = 0;
+static char log_filename[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PATH_MAX +
+#endif
+    1];
+
+/* Timestamp for most recent call to log_start(). */
+static nstime_t log_start_timestamp = NSTIME_ZERO_INITIALIZER;
+
+/* Increment these when adding to the log_bt and log_thr linked lists. */
+static size_t log_bt_index = 0;
+static size_t log_thr_index = 0;
+
+/* Linked list node definitions. These are only used in prof.c. */
+typedef struct prof_bt_node_s prof_bt_node_t;
+
+struct prof_bt_node_s {
+	prof_bt_node_t *next;
+	size_t index;
+	prof_bt_t bt;
+	/* Variable size backtrace vector pointed to by bt. */
+	void *vec[1];
+};
+
+typedef struct prof_thr_node_s prof_thr_node_t;
+
+struct prof_thr_node_s {
+	prof_thr_node_t *next;
+	size_t index;
+	uint64_t thr_uid;
+	/* Variable size based on thr_name_sz. */
+	char name[1];
+};
+
+typedef struct prof_alloc_node_s prof_alloc_node_t;
+
+/* This is output when logging sampled allocations. */
+struct prof_alloc_node_s {
+	prof_alloc_node_t *next;
+	/* Indices into an array of thread data. */
+	size_t alloc_thr_ind;
+	size_t free_thr_ind;
+
+	/* Indices into an array of backtraces. */
+	size_t alloc_bt_ind;
+	size_t free_bt_ind;
+
+	uint64_t alloc_time_ns;
+	uint64_t free_time_ns;
+
+	size_t usize;
+};
+
+/*
+ * Created on the first call to prof_log_start and deleted on prof_log_stop.
+ * These are the backtraces and threads that have already been logged by an
+ * allocation.
+ */
+static bool log_tables_initialized = false;
+static ckh_t log_bt_node_set;
+static ckh_t log_thr_node_set;
+
+/* Store linked lists for logged data. */
+static prof_bt_node_t *log_bt_first = NULL;
+static prof_bt_node_t *log_bt_last = NULL;
+static prof_thr_node_t *log_thr_first = NULL;
+static prof_thr_node_t *log_thr_last = NULL;
+static prof_alloc_node_t *log_alloc_first = NULL;
+static prof_alloc_node_t *log_alloc_last = NULL;
+
+/* Protects the prof_logging_state and any log_{...} variable. */
+static malloc_mutex_t log_mtx;
+
+/*
+ * Table of mutexes that are shared among gctx's.  These are leaf locks, so
+ * there is no problem with using them for more than one gctx at the same time.
+ * The primary motivation for this sharing though is that gctx's are ephemeral,
+ * and destroying mutexes causes complications for systems that allocate when
+ * creating/destroying mutexes.
+ */
+static malloc_mutex_t	*gctx_locks;
+static atomic_u_t	cum_gctxs; /* Atomic counter. */
+
+/*
+ * Table of mutexes that are shared among tdata's.  No operations require
+ * holding multiple tdata locks, so there is no problem with using them for more
+ * than one tdata at the same time, even though a gctx lock may be acquired
+ * while holding a tdata lock.
+ */
+static malloc_mutex_t	*tdata_locks;
+
+/*
+ * Global hash of (prof_bt_t *)-->(prof_gctx_t *).  This is the master data
+ * structure that knows about all backtraces currently captured.
+ */
+static ckh_t		bt2gctx;
+/* Non static to enable profiling. */
+malloc_mutex_t		bt2gctx_mtx;
+
+/*
+ * Tree of all extant prof_tdata_t structures, regardless of state,
+ * {attached,detached,expired}.
+ */
+static prof_tdata_tree_t	tdatas;
+static malloc_mutex_t	tdatas_mtx;
+
+static uint64_t		next_thr_uid;
+static malloc_mutex_t	next_thr_uid_mtx;
+
+static malloc_mutex_t	prof_dump_seq_mtx;
+static uint64_t		prof_dump_seq;
+static uint64_t		prof_dump_iseq;
+static uint64_t		prof_dump_mseq;
+static uint64_t		prof_dump_useq;
+
+/*
+ * This buffer is rather large for stack allocation, so use a single buffer for
+ * all profile dumps.
+ */
+static malloc_mutex_t	prof_dump_mtx;
+static char		prof_dump_buf[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PROF_DUMP_BUFSIZE
+#else
+    1
+#endif
+];
+static size_t		prof_dump_buf_end;
+static int		prof_dump_fd;
+
+/* Do not dump any profiles until bootstrapping is complete. */
+static bool		prof_booted = false;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static bool	prof_tctx_should_destroy(tsdn_t *tsdn, prof_tctx_t *tctx);
+static void	prof_tctx_destroy(tsd_t *tsd, prof_tctx_t *tctx);
+static bool	prof_tdata_should_destroy(tsdn_t *tsdn, prof_tdata_t *tdata,
+    bool even_if_attached);
+static void	prof_tdata_destroy(tsd_t *tsd, prof_tdata_t *tdata,
+    bool even_if_attached);
+static char	*prof_thread_name_alloc(tsdn_t *tsdn, const char *thread_name);
+
+/* Hashtable functions for log_bt_node_set and log_thr_node_set. */
+static void prof_thr_node_hash(const void *key, size_t r_hash[2]);
+static bool prof_thr_node_keycomp(const void *k1, const void *k2);
+static void prof_bt_node_hash(const void *key, size_t r_hash[2]);
+static bool prof_bt_node_keycomp(const void *k1, const void *k2);
+
+/******************************************************************************/
+/* Red-black trees. */
+
+static int
+prof_tctx_comp(const prof_tctx_t *a, const prof_tctx_t *b) {
+	uint64_t a_thr_uid = a->thr_uid;
+	uint64_t b_thr_uid = b->thr_uid;
+	int ret = (a_thr_uid > b_thr_uid) - (a_thr_uid < b_thr_uid);
+	if (ret == 0) {
+		uint64_t a_thr_discrim = a->thr_discrim;
+		uint64_t b_thr_discrim = b->thr_discrim;
+		ret = (a_thr_discrim > b_thr_discrim) - (a_thr_discrim <
+		    b_thr_discrim);
+		if (ret == 0) {
+			uint64_t a_tctx_uid = a->tctx_uid;
+			uint64_t b_tctx_uid = b->tctx_uid;
+			ret = (a_tctx_uid > b_tctx_uid) - (a_tctx_uid <
+			    b_tctx_uid);
+		}
+	}
+	return ret;
+}
+
+rb_gen(static UNUSED, tctx_tree_, prof_tctx_tree_t, prof_tctx_t,
+    tctx_link, prof_tctx_comp)
+
+static int
+prof_gctx_comp(const prof_gctx_t *a, const prof_gctx_t *b) {
+	unsigned a_len = a->bt.len;
+	unsigned b_len = b->bt.len;
+	unsigned comp_len = (a_len < b_len) ? a_len : b_len;
+	int ret = memcmp(a->bt.vec, b->bt.vec, comp_len * sizeof(void *));
+	if (ret == 0) {
+		ret = (a_len > b_len) - (a_len < b_len);
+	}
+	return ret;
+}
+
+rb_gen(static UNUSED, gctx_tree_, prof_gctx_tree_t, prof_gctx_t, dump_link,
+    prof_gctx_comp)
+
+static int
+prof_tdata_comp(const prof_tdata_t *a, const prof_tdata_t *b) {
+	int ret;
+	uint64_t a_uid = a->thr_uid;
+	uint64_t b_uid = b->thr_uid;
+
+	ret = ((a_uid > b_uid) - (a_uid < b_uid));
+	if (ret == 0) {
+		uint64_t a_discrim = a->thr_discrim;
+		uint64_t b_discrim = b->thr_discrim;
+
+		ret = ((a_discrim > b_discrim) - (a_discrim < b_discrim));
+	}
+	return ret;
+}
+
+rb_gen(static UNUSED, tdata_tree_, prof_tdata_tree_t, prof_tdata_t, tdata_link,
+    prof_tdata_comp)
+
+/******************************************************************************/
+
+void
+prof_alloc_rollback(tsd_t *tsd, prof_tctx_t *tctx, bool updated) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	if (updated) {
+		/*
+		 * Compute a new sample threshold.  This isn't very important in
+		 * practice, because this function is rarely executed, so the
+		 * potential for sample bias is minimal except in contrived
+		 * programs.
+		 */
+		tdata = prof_tdata_get(tsd, true);
+		if (tdata != NULL) {
+			prof_sample_threshold_update(tdata);
+		}
+	}
+
+	if ((uintptr_t)tctx > (uintptr_t)1U) {
+		malloc_mutex_lock(tsd_tsdn(tsd), tctx->tdata->lock);
+		tctx->prepared = false;
+		if (prof_tctx_should_destroy(tsd_tsdn(tsd), tctx)) {
+			prof_tctx_destroy(tsd, tctx);
+		} else {
+			malloc_mutex_unlock(tsd_tsdn(tsd), tctx->tdata->lock);
+		}
+	}
+}
+
+void
+prof_malloc_sample_object(tsdn_t *tsdn, const void *ptr, size_t usize,
+    prof_tctx_t *tctx) {
+	prof_tctx_set(tsdn, ptr, usize, NULL, tctx);
+
+	/* Get the current time and set this in the extent_t. We'll read this
+	 * when free() is called. */
+	nstime_t t = NSTIME_ZERO_INITIALIZER;
+	nstime_update(&t);
+	prof_alloc_time_set(tsdn, ptr, NULL, t);
+
+	malloc_mutex_lock(tsdn, tctx->tdata->lock);
+	tctx->cnts.curobjs++;
+	tctx->cnts.curbytes += usize;
+	if (opt_prof_accum) {
+		tctx->cnts.accumobjs++;
+		tctx->cnts.accumbytes += usize;
+	}
+	tctx->prepared = false;
+	malloc_mutex_unlock(tsdn, tctx->tdata->lock);
+}
+
+static size_t
+prof_log_bt_index(tsd_t *tsd, prof_bt_t *bt) {
+	assert(prof_logging_state == prof_logging_state_started);
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &log_mtx);
+
+	prof_bt_node_t dummy_node;
+	dummy_node.bt = *bt;
+	prof_bt_node_t *node;
+
+	/* See if this backtrace is already cached in the table. */
+	if (ckh_search(&log_bt_node_set, (void *)(&dummy_node),
+	    (void **)(&node), NULL)) {
+		size_t sz = offsetof(prof_bt_node_t, vec) +
+			        (bt->len * sizeof(void *));
+		prof_bt_node_t *new_node = (prof_bt_node_t *)
+		    iallocztm(tsd_tsdn(tsd), sz, sz_size2index(sz), false, NULL,
+		    true, arena_get(TSDN_NULL, 0, true), true);
+		if (log_bt_first == NULL) {
+			log_bt_first = new_node;
+			log_bt_last = new_node;
+		} else {
+			log_bt_last->next = new_node;
+			log_bt_last = new_node;
+		}
+
+		new_node->next = NULL;
+		new_node->index = log_bt_index;
+		/*
+		 * Copy the backtrace: bt is inside a tdata or gctx, which
+		 * might die before prof_log_stop is called.
+		 */
+		new_node->bt.len = bt->len;
+		memcpy(new_node->vec, bt->vec, bt->len * sizeof(void *));
+		new_node->bt.vec = new_node->vec;
+
+		log_bt_index++;
+		ckh_insert(tsd, &log_bt_node_set, (void *)new_node, NULL);
+		return new_node->index;
+	} else {
+		return node->index;
+	}
+}
+static size_t
+prof_log_thr_index(tsd_t *tsd, uint64_t thr_uid, const char *name) {
+	assert(prof_logging_state == prof_logging_state_started);
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &log_mtx);
+
+	prof_thr_node_t dummy_node;
+	dummy_node.thr_uid = thr_uid;
+	prof_thr_node_t *node;
+
+	/* See if this thread is already cached in the table. */
+	if (ckh_search(&log_thr_node_set, (void *)(&dummy_node),
+	    (void **)(&node), NULL)) {
+		size_t sz = offsetof(prof_thr_node_t, name) + strlen(name) + 1;
+		prof_thr_node_t *new_node = (prof_thr_node_t *)
+		    iallocztm(tsd_tsdn(tsd), sz, sz_size2index(sz), false, NULL,
+		    true, arena_get(TSDN_NULL, 0, true), true);
+		if (log_thr_first == NULL) {
+			log_thr_first = new_node;
+			log_thr_last = new_node;
+		} else {
+			log_thr_last->next = new_node;
+			log_thr_last = new_node;
+		}
+
+		new_node->next = NULL;
+		new_node->index = log_thr_index;
+		new_node->thr_uid = thr_uid;
+		strcpy(new_node->name, name);
+
+		log_thr_index++;
+		ckh_insert(tsd, &log_thr_node_set, (void *)new_node, NULL);
+		return new_node->index;
+	} else {
+		return node->index;
+	}
+}
+
+static void
+prof_try_log(tsd_t *tsd, const void *ptr, size_t usize, prof_tctx_t *tctx) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	prof_tdata_t *cons_tdata = prof_tdata_get(tsd, false);
+	if (cons_tdata == NULL) {
+		/*
+		 * We decide not to log these allocations. cons_tdata will be
+		 * NULL only when the current thread is in a weird state (e.g.
+		 * it's being destroyed).
+		 */
+		return;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &log_mtx);
+
+	if (prof_logging_state != prof_logging_state_started) {
+		goto label_done;
+	}
+
+	if (!log_tables_initialized) {
+		bool err1 = ckh_new(tsd, &log_bt_node_set, PROF_CKH_MINITEMS,
+				prof_bt_node_hash, prof_bt_node_keycomp);
+		bool err2 = ckh_new(tsd, &log_thr_node_set, PROF_CKH_MINITEMS,
+				prof_thr_node_hash, prof_thr_node_keycomp);
+		if (err1 || err2) {
+			goto label_done;
+		}
+		log_tables_initialized = true;
+	}
+
+	nstime_t alloc_time = prof_alloc_time_get(tsd_tsdn(tsd), ptr,
+			          (alloc_ctx_t *)NULL);
+	nstime_t free_time = NSTIME_ZERO_INITIALIZER;
+	nstime_update(&free_time);
+
+	size_t sz = sizeof(prof_alloc_node_t);
+	prof_alloc_node_t *new_node = (prof_alloc_node_t *)
+	    iallocztm(tsd_tsdn(tsd), sz, sz_size2index(sz), false, NULL, true,
+	    arena_get(TSDN_NULL, 0, true), true);
+
+	const char *prod_thr_name = (tctx->tdata->thread_name == NULL)?
+				        "" : tctx->tdata->thread_name;
+	const char *cons_thr_name = prof_thread_name_get(tsd);
+
+	prof_bt_t bt;
+	/* Initialize the backtrace, using the buffer in tdata to store it. */
+	bt_init(&bt, cons_tdata->vec);
+	prof_backtrace(&bt);
+	prof_bt_t *cons_bt = &bt;
+
+	/* We haven't destroyed tctx yet, so gctx should be good to read. */
+	prof_bt_t *prod_bt = &tctx->gctx->bt;
+
+	new_node->next = NULL;
+	new_node->alloc_thr_ind = prof_log_thr_index(tsd, tctx->tdata->thr_uid,
+				      prod_thr_name);
+	new_node->free_thr_ind = prof_log_thr_index(tsd, cons_tdata->thr_uid,
+				     cons_thr_name);
+	new_node->alloc_bt_ind = prof_log_bt_index(tsd, prod_bt);
+	new_node->free_bt_ind = prof_log_bt_index(tsd, cons_bt);
+	new_node->alloc_time_ns = nstime_ns(&alloc_time);
+	new_node->free_time_ns = nstime_ns(&free_time);
+	new_node->usize = usize;
+
+	if (log_alloc_first == NULL) {
+		log_alloc_first = new_node;
+		log_alloc_last = new_node;
+	} else {
+		log_alloc_last->next = new_node;
+		log_alloc_last = new_node;
+	}
+
+label_done:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &log_mtx);
+}
+
+void
+prof_free_sampled_object(tsd_t *tsd, const void *ptr, size_t usize,
+    prof_tctx_t *tctx) {
+	malloc_mutex_lock(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	assert(tctx->cnts.curobjs > 0);
+	assert(tctx->cnts.curbytes >= usize);
+	tctx->cnts.curobjs--;
+	tctx->cnts.curbytes -= usize;
+
+	prof_try_log(tsd, ptr, usize, tctx);
+
+	if (prof_tctx_should_destroy(tsd_tsdn(tsd), tctx)) {
+		prof_tctx_destroy(tsd, tctx);
+	} else {
+		malloc_mutex_unlock(tsd_tsdn(tsd), tctx->tdata->lock);
+	}
+}
+
+void
+bt_init(prof_bt_t *bt, void **vec) {
+	cassert(config_prof);
+
+	bt->vec = vec;
+	bt->len = 0;
+}
+
+static void
+prof_enter(tsd_t *tsd, prof_tdata_t *tdata) {
+	cassert(config_prof);
+	assert(tdata == prof_tdata_get(tsd, false));
+
+	if (tdata != NULL) {
+		assert(!tdata->enq);
+		tdata->enq = true;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &bt2gctx_mtx);
+}
+
+static void
+prof_leave(tsd_t *tsd, prof_tdata_t *tdata) {
+	cassert(config_prof);
+	assert(tdata == prof_tdata_get(tsd, false));
+
+	malloc_mutex_unlock(tsd_tsdn(tsd), &bt2gctx_mtx);
+
+	if (tdata != NULL) {
+		bool idump, gdump;
+
+		assert(tdata->enq);
+		tdata->enq = false;
+		idump = tdata->enq_idump;
+		tdata->enq_idump = false;
+		gdump = tdata->enq_gdump;
+		tdata->enq_gdump = false;
+
+		if (idump) {
+			prof_idump(tsd_tsdn(tsd));
+		}
+		if (gdump) {
+			prof_gdump(tsd_tsdn(tsd));
+		}
+	}
+}
+
+#ifdef JEMALLOC_PROF_LIBUNWIND
+void
+prof_backtrace(prof_bt_t *bt) {
+	int nframes;
+
+	cassert(config_prof);
+	assert(bt->len == 0);
+	assert(bt->vec != NULL);
+
+	nframes = unw_backtrace(bt->vec, PROF_BT_MAX);
+	if (nframes <= 0) {
+		return;
+	}
+	bt->len = nframes;
+}
+#elif (defined(JEMALLOC_PROF_LIBGCC))
+static _Unwind_Reason_Code
+prof_unwind_init_callback(struct _Unwind_Context *context, void *arg) {
+	cassert(config_prof);
+
+	return _URC_NO_REASON;
+}
+
+static _Unwind_Reason_Code
+prof_unwind_callback(struct _Unwind_Context *context, void *arg) {
+	prof_unwind_data_t *data = (prof_unwind_data_t *)arg;
+	void *ip;
+
+	cassert(config_prof);
+
+	ip = (void *)_Unwind_GetIP(context);
+	if (ip == NULL) {
+		return _URC_END_OF_STACK;
+	}
+	data->bt->vec[data->bt->len] = ip;
+	data->bt->len++;
+	if (data->bt->len == data->max) {
+		return _URC_END_OF_STACK;
+	}
+
+	return _URC_NO_REASON;
+}
+
+void
+prof_backtrace(prof_bt_t *bt) {
+	prof_unwind_data_t data = {bt, PROF_BT_MAX};
+
+	cassert(config_prof);
+
+	_Unwind_Backtrace(prof_unwind_callback, &data);
+}
+#elif (defined(JEMALLOC_PROF_GCC))
+void
+prof_backtrace(prof_bt_t *bt) {
+#define BT_FRAME(i)							\
+	if ((i) < PROF_BT_MAX) {					\
+		void *p;						\
+		if (__builtin_frame_address(i) == 0) {			\
+			return;						\
+		}							\
+		p = __builtin_return_address(i);			\
+		if (p == NULL) {					\
+			return;						\
+		}							\
+		bt->vec[(i)] = p;					\
+		bt->len = (i) + 1;					\
+	} else {							\
+		return;							\
+	}
+
+	cassert(config_prof);
+
+	BT_FRAME(0)
+	BT_FRAME(1)
+	BT_FRAME(2)
+	BT_FRAME(3)
+	BT_FRAME(4)
+	BT_FRAME(5)
+	BT_FRAME(6)
+	BT_FRAME(7)
+	BT_FRAME(8)
+	BT_FRAME(9)
+
+	BT_FRAME(10)
+	BT_FRAME(11)
+	BT_FRAME(12)
+	BT_FRAME(13)
+	BT_FRAME(14)
+	BT_FRAME(15)
+	BT_FRAME(16)
+	BT_FRAME(17)
+	BT_FRAME(18)
+	BT_FRAME(19)
+
+	BT_FRAME(20)
+	BT_FRAME(21)
+	BT_FRAME(22)
+	BT_FRAME(23)
+	BT_FRAME(24)
+	BT_FRAME(25)
+	BT_FRAME(26)
+	BT_FRAME(27)
+	BT_FRAME(28)
+	BT_FRAME(29)
+
+	BT_FRAME(30)
+	BT_FRAME(31)
+	BT_FRAME(32)
+	BT_FRAME(33)
+	BT_FRAME(34)
+	BT_FRAME(35)
+	BT_FRAME(36)
+	BT_FRAME(37)
+	BT_FRAME(38)
+	BT_FRAME(39)
+
+	BT_FRAME(40)
+	BT_FRAME(41)
+	BT_FRAME(42)
+	BT_FRAME(43)
+	BT_FRAME(44)
+	BT_FRAME(45)
+	BT_FRAME(46)
+	BT_FRAME(47)
+	BT_FRAME(48)
+	BT_FRAME(49)
+
+	BT_FRAME(50)
+	BT_FRAME(51)
+	BT_FRAME(52)
+	BT_FRAME(53)
+	BT_FRAME(54)
+	BT_FRAME(55)
+	BT_FRAME(56)
+	BT_FRAME(57)
+	BT_FRAME(58)
+	BT_FRAME(59)
+
+	BT_FRAME(60)
+	BT_FRAME(61)
+	BT_FRAME(62)
+	BT_FRAME(63)
+	BT_FRAME(64)
+	BT_FRAME(65)
+	BT_FRAME(66)
+	BT_FRAME(67)
+	BT_FRAME(68)
+	BT_FRAME(69)
+
+	BT_FRAME(70)
+	BT_FRAME(71)
+	BT_FRAME(72)
+	BT_FRAME(73)
+	BT_FRAME(74)
+	BT_FRAME(75)
+	BT_FRAME(76)
+	BT_FRAME(77)
+	BT_FRAME(78)
+	BT_FRAME(79)
+
+	BT_FRAME(80)
+	BT_FRAME(81)
+	BT_FRAME(82)
+	BT_FRAME(83)
+	BT_FRAME(84)
+	BT_FRAME(85)
+	BT_FRAME(86)
+	BT_FRAME(87)
+	BT_FRAME(88)
+	BT_FRAME(89)
+
+	BT_FRAME(90)
+	BT_FRAME(91)
+	BT_FRAME(92)
+	BT_FRAME(93)
+	BT_FRAME(94)
+	BT_FRAME(95)
+	BT_FRAME(96)
+	BT_FRAME(97)
+	BT_FRAME(98)
+	BT_FRAME(99)
+
+	BT_FRAME(100)
+	BT_FRAME(101)
+	BT_FRAME(102)
+	BT_FRAME(103)
+	BT_FRAME(104)
+	BT_FRAME(105)
+	BT_FRAME(106)
+	BT_FRAME(107)
+	BT_FRAME(108)
+	BT_FRAME(109)
+
+	BT_FRAME(110)
+	BT_FRAME(111)
+	BT_FRAME(112)
+	BT_FRAME(113)
+	BT_FRAME(114)
+	BT_FRAME(115)
+	BT_FRAME(116)
+	BT_FRAME(117)
+	BT_FRAME(118)
+	BT_FRAME(119)
+
+	BT_FRAME(120)
+	BT_FRAME(121)
+	BT_FRAME(122)
+	BT_FRAME(123)
+	BT_FRAME(124)
+	BT_FRAME(125)
+	BT_FRAME(126)
+	BT_FRAME(127)
+#undef BT_FRAME
+}
+#else
+void
+prof_backtrace(prof_bt_t *bt) {
+	cassert(config_prof);
+	not_reached();
+}
+#endif
+
+static malloc_mutex_t *
+prof_gctx_mutex_choose(void) {
+	unsigned ngctxs = atomic_fetch_add_u(&cum_gctxs, 1, ATOMIC_RELAXED);
+
+	return &gctx_locks[(ngctxs - 1) % PROF_NCTX_LOCKS];
+}
+
+static malloc_mutex_t *
+prof_tdata_mutex_choose(uint64_t thr_uid) {
+	return &tdata_locks[thr_uid % PROF_NTDATA_LOCKS];
+}
+
+static prof_gctx_t *
+prof_gctx_create(tsdn_t *tsdn, prof_bt_t *bt) {
+	/*
+	 * Create a single allocation that has space for vec of length bt->len.
+	 */
+	size_t size = offsetof(prof_gctx_t, vec) + (bt->len * sizeof(void *));
+	prof_gctx_t *gctx = (prof_gctx_t *)iallocztm(tsdn, size,
+	    sz_size2index(size), false, NULL, true, arena_get(TSDN_NULL, 0, true),
+	    true);
+	if (gctx == NULL) {
+		return NULL;
+	}
+	gctx->lock = prof_gctx_mutex_choose();
+	/*
+	 * Set nlimbo to 1, in order to avoid a race condition with
+	 * prof_tctx_destroy()/prof_gctx_try_destroy().
+	 */
+	gctx->nlimbo = 1;
+	tctx_tree_new(&gctx->tctxs);
+	/* Duplicate bt. */
+	memcpy(gctx->vec, bt->vec, bt->len * sizeof(void *));
+	gctx->bt.vec = gctx->vec;
+	gctx->bt.len = bt->len;
+	return gctx;
+}
+
+static void
+prof_gctx_try_destroy(tsd_t *tsd, prof_tdata_t *tdata_self, prof_gctx_t *gctx,
+    prof_tdata_t *tdata) {
+	cassert(config_prof);
+
+	/*
+	 * Check that gctx is still unused by any thread cache before destroying
+	 * it.  prof_lookup() increments gctx->nlimbo in order to avoid a race
+	 * condition with this function, as does prof_tctx_destroy() in order to
+	 * avoid a race between the main body of prof_tctx_destroy() and entry
+	 * into this function.
+	 */
+	prof_enter(tsd, tdata_self);
+	malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+	assert(gctx->nlimbo != 0);
+	if (tctx_tree_empty(&gctx->tctxs) && gctx->nlimbo == 1) {
+		/* Remove gctx from bt2gctx. */
+		if (ckh_remove(tsd, &bt2gctx, &gctx->bt, NULL, NULL)) {
+			not_reached();
+		}
+		prof_leave(tsd, tdata_self);
+		/* Destroy gctx. */
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+		idalloctm(tsd_tsdn(tsd), gctx, NULL, NULL, true, true);
+	} else {
+		/*
+		 * Compensate for increment in prof_tctx_destroy() or
+		 * prof_lookup().
+		 */
+		gctx->nlimbo--;
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+		prof_leave(tsd, tdata_self);
+	}
+}
+
+static bool
+prof_tctx_should_destroy(tsdn_t *tsdn, prof_tctx_t *tctx) {
+	malloc_mutex_assert_owner(tsdn, tctx->tdata->lock);
+
+	if (opt_prof_accum) {
+		return false;
+	}
+	if (tctx->cnts.curobjs != 0) {
+		return false;
+	}
+	if (tctx->prepared) {
+		return false;
+	}
+	return true;
+}
+
+static bool
+prof_gctx_should_destroy(prof_gctx_t *gctx) {
+	if (opt_prof_accum) {
+		return false;
+	}
+	if (!tctx_tree_empty(&gctx->tctxs)) {
+		return false;
+	}
+	if (gctx->nlimbo != 0) {
+		return false;
+	}
+	return true;
+}
+
+static void
+prof_tctx_destroy(tsd_t *tsd, prof_tctx_t *tctx) {
+	prof_tdata_t *tdata = tctx->tdata;
+	prof_gctx_t *gctx = tctx->gctx;
+	bool destroy_tdata, destroy_tctx, destroy_gctx;
+
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	assert(tctx->cnts.curobjs == 0);
+	assert(tctx->cnts.curbytes == 0);
+	assert(!opt_prof_accum);
+	assert(tctx->cnts.accumobjs == 0);
+	assert(tctx->cnts.accumbytes == 0);
+
+	ckh_remove(tsd, &tdata->bt2tctx, &gctx->bt, NULL, NULL);
+	destroy_tdata = prof_tdata_should_destroy(tsd_tsdn(tsd), tdata, false);
+	malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+
+	malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+	switch (tctx->state) {
+	case prof_tctx_state_nominal:
+		tctx_tree_remove(&gctx->tctxs, tctx);
+		destroy_tctx = true;
+		if (prof_gctx_should_destroy(gctx)) {
+			/*
+			 * Increment gctx->nlimbo in order to keep another
+			 * thread from winning the race to destroy gctx while
+			 * this one has gctx->lock dropped.  Without this, it
+			 * would be possible for another thread to:
+			 *
+			 * 1) Sample an allocation associated with gctx.
+			 * 2) Deallocate the sampled object.
+			 * 3) Successfully prof_gctx_try_destroy(gctx).
+			 *
+			 * The result would be that gctx no longer exists by the
+			 * time this thread accesses it in
+			 * prof_gctx_try_destroy().
+			 */
+			gctx->nlimbo++;
+			destroy_gctx = true;
+		} else {
+			destroy_gctx = false;
+		}
+		break;
+	case prof_tctx_state_dumping:
+		/*
+		 * A dumping thread needs tctx to remain valid until dumping
+		 * has finished.  Change state such that the dumping thread will
+		 * complete destruction during a late dump iteration phase.
+		 */
+		tctx->state = prof_tctx_state_purgatory;
+		destroy_tctx = false;
+		destroy_gctx = false;
+		break;
+	default:
+		not_reached();
+		destroy_tctx = false;
+		destroy_gctx = false;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+	if (destroy_gctx) {
+		prof_gctx_try_destroy(tsd, prof_tdata_get(tsd, false), gctx,
+		    tdata);
+	}
+
+	malloc_mutex_assert_not_owner(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	if (destroy_tdata) {
+		prof_tdata_destroy(tsd, tdata, false);
+	}
+
+	if (destroy_tctx) {
+		idalloctm(tsd_tsdn(tsd), tctx, NULL, NULL, true, true);
+	}
+}
+
+static bool
+prof_lookup_global(tsd_t *tsd, prof_bt_t *bt, prof_tdata_t *tdata,
+    void **p_btkey, prof_gctx_t **p_gctx, bool *p_new_gctx) {
+	union {
+		prof_gctx_t	*p;
+		void		*v;
+	} gctx, tgctx;
+	union {
+		prof_bt_t	*p;
+		void		*v;
+	} btkey;
+	bool new_gctx;
+
+	prof_enter(tsd, tdata);
+	if (ckh_search(&bt2gctx, bt, &btkey.v, &gctx.v)) {
+		/* bt has never been seen before.  Insert it. */
+		prof_leave(tsd, tdata);
+		tgctx.p = prof_gctx_create(tsd_tsdn(tsd), bt);
+		if (tgctx.v == NULL) {
+			return true;
+		}
+		prof_enter(tsd, tdata);
+		if (ckh_search(&bt2gctx, bt, &btkey.v, &gctx.v)) {
+			gctx.p = tgctx.p;
+			btkey.p = &gctx.p->bt;
+			if (ckh_insert(tsd, &bt2gctx, btkey.v, gctx.v)) {
+				/* OOM. */
+				prof_leave(tsd, tdata);
+				idalloctm(tsd_tsdn(tsd), gctx.v, NULL, NULL,
+				    true, true);
+				return true;
+			}
+			new_gctx = true;
+		} else {
+			new_gctx = false;
+		}
+	} else {
+		tgctx.v = NULL;
+		new_gctx = false;
+	}
+
+	if (!new_gctx) {
+		/*
+		 * Increment nlimbo, in order to avoid a race condition with
+		 * prof_tctx_destroy()/prof_gctx_try_destroy().
+		 */
+		malloc_mutex_lock(tsd_tsdn(tsd), gctx.p->lock);
+		gctx.p->nlimbo++;
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx.p->lock);
+		new_gctx = false;
+
+		if (tgctx.v != NULL) {
+			/* Lost race to insert. */
+			idalloctm(tsd_tsdn(tsd), tgctx.v, NULL, NULL, true,
+			    true);
+		}
+	}
+	prof_leave(tsd, tdata);
+
+	*p_btkey = btkey.v;
+	*p_gctx = gctx.p;
+	*p_new_gctx = new_gctx;
+	return false;
+}
+
+prof_tctx_t *
+prof_lookup(tsd_t *tsd, prof_bt_t *bt) {
+	union {
+		prof_tctx_t	*p;
+		void		*v;
+	} ret;
+	prof_tdata_t *tdata;
+	bool not_found;
+
+	cassert(config_prof);
+
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return NULL;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), tdata->lock);
+	not_found = ckh_search(&tdata->bt2tctx, bt, NULL, &ret.v);
+	if (!not_found) { /* Note double negative! */
+		ret.p->prepared = true;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+	if (not_found) {
+		void *btkey;
+		prof_gctx_t *gctx;
+		bool new_gctx, error;
+
+		/*
+		 * This thread's cache lacks bt.  Look for it in the global
+		 * cache.
+		 */
+		if (prof_lookup_global(tsd, bt, tdata, &btkey, &gctx,
+		    &new_gctx)) {
+			return NULL;
+		}
+
+		/* Link a prof_tctx_t into gctx for this thread. */
+		ret.v = iallocztm(tsd_tsdn(tsd), sizeof(prof_tctx_t),
+		    sz_size2index(sizeof(prof_tctx_t)), false, NULL, true,
+		    arena_ichoose(tsd, NULL), true);
+		if (ret.p == NULL) {
+			if (new_gctx) {
+				prof_gctx_try_destroy(tsd, tdata, gctx, tdata);
+			}
+			return NULL;
+		}
+		ret.p->tdata = tdata;
+		ret.p->thr_uid = tdata->thr_uid;
+		ret.p->thr_discrim = tdata->thr_discrim;
+		memset(&ret.p->cnts, 0, sizeof(prof_cnt_t));
+		ret.p->gctx = gctx;
+		ret.p->tctx_uid = tdata->tctx_uid_next++;
+		ret.p->prepared = true;
+		ret.p->state = prof_tctx_state_initializing;
+		malloc_mutex_lock(tsd_tsdn(tsd), tdata->lock);
+		error = ckh_insert(tsd, &tdata->bt2tctx, btkey, ret.v);
+		malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+		if (error) {
+			if (new_gctx) {
+				prof_gctx_try_destroy(tsd, tdata, gctx, tdata);
+			}
+			idalloctm(tsd_tsdn(tsd), ret.v, NULL, NULL, true, true);
+			return NULL;
+		}
+		malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+		ret.p->state = prof_tctx_state_nominal;
+		tctx_tree_insert(&gctx->tctxs, ret.p);
+		gctx->nlimbo--;
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+	}
+
+	return ret.p;
+}
+
+/*
+ * The bodies of this function and prof_leakcheck() are compiled out unless heap
+ * profiling is enabled, so that it is possible to compile jemalloc with
+ * floating point support completely disabled.  Avoiding floating point code is
+ * important on memory-constrained systems, but it also enables a workaround for
+ * versions of glibc that don't properly save/restore floating point registers
+ * during dynamic lazy symbol loading (which internally calls into whatever
+ * malloc implementation happens to be integrated into the application).  Note
+ * that some compilers (e.g.  gcc 4.8) may use floating point registers for fast
+ * memory moves, so jemalloc must be compiled with such optimizations disabled
+ * (e.g.
+ * -mno-sse) in order for the workaround to be complete.
+ */
+void
+prof_sample_threshold_update(prof_tdata_t *tdata) {
+#ifdef JEMALLOC_PROF
+	if (!config_prof) {
+		return;
+	}
+
+	if (lg_prof_sample == 0) {
+		tsd_bytes_until_sample_set(tsd_fetch(), 0);
+		return;
+	}
+
+	/*
+	 * Compute sample interval as a geometrically distributed random
+	 * variable with mean (2^lg_prof_sample).
+	 *
+	 *                             __        __
+	 *                             |  log(u)  |                     1
+	 * tdata->bytes_until_sample = | -------- |, where p = ---------------
+	 *                             | log(1-p) |             lg_prof_sample
+	 *                                                     2
+	 *
+	 * For more information on the math, see:
+	 *
+	 *   Non-Uniform Random Variate Generation
+	 *   Luc Devroye
+	 *   Springer-Verlag, New York, 1986
+	 *   pp 500
+	 *   (http://luc.devroye.org/rnbookindex.html)
+	 */
+	uint64_t r = prng_lg_range_u64(&tdata->prng_state, 53);
+	double u = (double)r * (1.0/9007199254740992.0L);
+	uint64_t bytes_until_sample = (uint64_t)(log(u) /
+	    log(1.0 - (1.0 / (double)((uint64_t)1U << lg_prof_sample))))
+	    + (uint64_t)1U;
+	if (bytes_until_sample > SSIZE_MAX) {
+		bytes_until_sample = SSIZE_MAX;
+	}
+	tsd_bytes_until_sample_set(tsd_fetch(), bytes_until_sample);
+
+#endif
+}
+
+#ifdef JEMALLOC_JET
+static prof_tdata_t *
+prof_tdata_count_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *arg) {
+	size_t *tdata_count = (size_t *)arg;
+
+	(*tdata_count)++;
+
+	return NULL;
+}
+
+size_t
+prof_tdata_count(void) {
+	size_t tdata_count = 0;
+	tsdn_t *tsdn;
+
+	tsdn = tsdn_fetch();
+	malloc_mutex_lock(tsdn, &tdatas_mtx);
+	tdata_tree_iter(&tdatas, NULL, prof_tdata_count_iter,
+	    (void *)&tdata_count);
+	malloc_mutex_unlock(tsdn, &tdatas_mtx);
+
+	return tdata_count;
+}
+
+size_t
+prof_bt_count(void) {
+	size_t bt_count;
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+
+	tsd = tsd_fetch();
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return 0;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &bt2gctx_mtx);
+	bt_count = ckh_count(&bt2gctx);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &bt2gctx_mtx);
+
+	return bt_count;
+}
+#endif
+
+static int
+prof_dump_open_impl(bool propagate_err, const char *filename) {
+	int fd;
+
+	fd = creat(filename, 0644);
+	if (fd == -1 && !propagate_err) {
+		malloc_printf("<jemalloc>: creat(\"%s\"), 0644) failed\n",
+		    filename);
+		if (opt_abort) {
+			abort();
+		}
+	}
+
+	return fd;
+}
+prof_dump_open_t *JET_MUTABLE prof_dump_open = prof_dump_open_impl;
+
+static bool
+prof_dump_flush(bool propagate_err) {
+	bool ret = false;
+	ssize_t err;
+
+	cassert(config_prof);
+
+	err = malloc_write_fd(prof_dump_fd, prof_dump_buf, prof_dump_buf_end);
+	if (err == -1) {
+		if (!propagate_err) {
+			malloc_write("<jemalloc>: write() failed during heap "
+			    "profile flush\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+		ret = true;
+	}
+	prof_dump_buf_end = 0;
+
+	return ret;
+}
+
+static bool
+prof_dump_close(bool propagate_err) {
+	bool ret;
+
+	assert(prof_dump_fd != -1);
+	ret = prof_dump_flush(propagate_err);
+	close(prof_dump_fd);
+	prof_dump_fd = -1;
+
+	return ret;
+}
+
+static bool
+prof_dump_write(bool propagate_err, const char *s) {
+	size_t i, slen, n;
+
+	cassert(config_prof);
+
+	i = 0;
+	slen = strlen(s);
+	while (i < slen) {
+		/* Flush the buffer if it is full. */
+		if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) {
+			if (prof_dump_flush(propagate_err) && propagate_err) {
+				return true;
+			}
+		}
+
+		if (prof_dump_buf_end + slen - i <= PROF_DUMP_BUFSIZE) {
+			/* Finish writing. */
+			n = slen - i;
+		} else {
+			/* Write as much of s as will fit. */
+			n = PROF_DUMP_BUFSIZE - prof_dump_buf_end;
+		}
+		memcpy(&prof_dump_buf[prof_dump_buf_end], &s[i], n);
+		prof_dump_buf_end += n;
+		i += n;
+	}
+	assert(i == slen);
+
+	return false;
+}
+
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static bool
+prof_dump_printf(bool propagate_err, const char *format, ...) {
+	bool ret;
+	va_list ap;
+	char buf[PROF_PRINTF_BUFSIZE];
+
+	va_start(ap, format);
+	malloc_vsnprintf(buf, sizeof(buf), format, ap);
+	va_end(ap);
+	ret = prof_dump_write(propagate_err, buf);
+
+	return ret;
+}
+
+static void
+prof_tctx_merge_tdata(tsdn_t *tsdn, prof_tctx_t *tctx, prof_tdata_t *tdata) {
+	malloc_mutex_assert_owner(tsdn, tctx->tdata->lock);
+
+	malloc_mutex_lock(tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_initializing:
+		malloc_mutex_unlock(tsdn, tctx->gctx->lock);
+		return;
+	case prof_tctx_state_nominal:
+		tctx->state = prof_tctx_state_dumping;
+		malloc_mutex_unlock(tsdn, tctx->gctx->lock);
+
+		memcpy(&tctx->dump_cnts, &tctx->cnts, sizeof(prof_cnt_t));
+
+		tdata->cnt_summed.curobjs += tctx->dump_cnts.curobjs;
+		tdata->cnt_summed.curbytes += tctx->dump_cnts.curbytes;
+		if (opt_prof_accum) {
+			tdata->cnt_summed.accumobjs +=
+			    tctx->dump_cnts.accumobjs;
+			tdata->cnt_summed.accumbytes +=
+			    tctx->dump_cnts.accumbytes;
+		}
+		break;
+	case prof_tctx_state_dumping:
+	case prof_tctx_state_purgatory:
+		not_reached();
+	}
+}
+
+static void
+prof_tctx_merge_gctx(tsdn_t *tsdn, prof_tctx_t *tctx, prof_gctx_t *gctx) {
+	malloc_mutex_assert_owner(tsdn, gctx->lock);
+
+	gctx->cnt_summed.curobjs += tctx->dump_cnts.curobjs;
+	gctx->cnt_summed.curbytes += tctx->dump_cnts.curbytes;
+	if (opt_prof_accum) {
+		gctx->cnt_summed.accumobjs += tctx->dump_cnts.accumobjs;
+		gctx->cnt_summed.accumbytes += tctx->dump_cnts.accumbytes;
+	}
+}
+
+static prof_tctx_t *
+prof_tctx_merge_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *arg) {
+	tsdn_t *tsdn = (tsdn_t *)arg;
+
+	malloc_mutex_assert_owner(tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_nominal:
+		/* New since dumping started; ignore. */
+		break;
+	case prof_tctx_state_dumping:
+	case prof_tctx_state_purgatory:
+		prof_tctx_merge_gctx(tsdn, tctx, tctx->gctx);
+		break;
+	default:
+		not_reached();
+	}
+
+	return NULL;
+}
+
+struct prof_tctx_dump_iter_arg_s {
+	tsdn_t	*tsdn;
+	bool	propagate_err;
+};
+
+static prof_tctx_t *
+prof_tctx_dump_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *opaque) {
+	struct prof_tctx_dump_iter_arg_s *arg =
+	    (struct prof_tctx_dump_iter_arg_s *)opaque;
+
+	malloc_mutex_assert_owner(arg->tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_initializing:
+	case prof_tctx_state_nominal:
+		/* Not captured by this dump. */
+		break;
+	case prof_tctx_state_dumping:
+	case prof_tctx_state_purgatory:
+		if (prof_dump_printf(arg->propagate_err,
+		    "  t%"FMTu64": %"FMTu64": %"FMTu64" [%"FMTu64": "
+		    "%"FMTu64"]\n", tctx->thr_uid, tctx->dump_cnts.curobjs,
+		    tctx->dump_cnts.curbytes, tctx->dump_cnts.accumobjs,
+		    tctx->dump_cnts.accumbytes)) {
+			return tctx;
+		}
+		break;
+	default:
+		not_reached();
+	}
+	return NULL;
+}
+
+static prof_tctx_t *
+prof_tctx_finish_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *arg) {
+	tsdn_t *tsdn = (tsdn_t *)arg;
+	prof_tctx_t *ret;
+
+	malloc_mutex_assert_owner(tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_nominal:
+		/* New since dumping started; ignore. */
+		break;
+	case prof_tctx_state_dumping:
+		tctx->state = prof_tctx_state_nominal;
+		break;
+	case prof_tctx_state_purgatory:
+		ret = tctx;
+		goto label_return;
+	default:
+		not_reached();
+	}
+
+	ret = NULL;
+label_return:
+	return ret;
+}
+
+static void
+prof_dump_gctx_prep(tsdn_t *tsdn, prof_gctx_t *gctx, prof_gctx_tree_t *gctxs) {
+	cassert(config_prof);
+
+	malloc_mutex_lock(tsdn, gctx->lock);
+
+	/*
+	 * Increment nlimbo so that gctx won't go away before dump.
+	 * Additionally, link gctx into the dump list so that it is included in
+	 * prof_dump()'s second pass.
+	 */
+	gctx->nlimbo++;
+	gctx_tree_insert(gctxs, gctx);
+
+	memset(&gctx->cnt_summed, 0, sizeof(prof_cnt_t));
+
+	malloc_mutex_unlock(tsdn, gctx->lock);
+}
+
+struct prof_gctx_merge_iter_arg_s {
+	tsdn_t	*tsdn;
+	size_t	leak_ngctx;
+};
+
+static prof_gctx_t *
+prof_gctx_merge_iter(prof_gctx_tree_t *gctxs, prof_gctx_t *gctx, void *opaque) {
+	struct prof_gctx_merge_iter_arg_s *arg =
+	    (struct prof_gctx_merge_iter_arg_s *)opaque;
+
+	malloc_mutex_lock(arg->tsdn, gctx->lock);
+	tctx_tree_iter(&gctx->tctxs, NULL, prof_tctx_merge_iter,
+	    (void *)arg->tsdn);
+	if (gctx->cnt_summed.curobjs != 0) {
+		arg->leak_ngctx++;
+	}
+	malloc_mutex_unlock(arg->tsdn, gctx->lock);
+
+	return NULL;
+}
+
+static void
+prof_gctx_finish(tsd_t *tsd, prof_gctx_tree_t *gctxs) {
+	prof_tdata_t *tdata = prof_tdata_get(tsd, false);
+	prof_gctx_t *gctx;
+
+	/*
+	 * Standard tree iteration won't work here, because as soon as we
+	 * decrement gctx->nlimbo and unlock gctx, another thread can
+	 * concurrently destroy it, which will corrupt the tree.  Therefore,
+	 * tear down the tree one node at a time during iteration.
+	 */
+	while ((gctx = gctx_tree_first(gctxs)) != NULL) {
+		gctx_tree_remove(gctxs, gctx);
+		malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+		{
+			prof_tctx_t *next;
+
+			next = NULL;
+			do {
+				prof_tctx_t *to_destroy =
+				    tctx_tree_iter(&gctx->tctxs, next,
+				    prof_tctx_finish_iter,
+				    (void *)tsd_tsdn(tsd));
+				if (to_destroy != NULL) {
+					next = tctx_tree_next(&gctx->tctxs,
+					    to_destroy);
+					tctx_tree_remove(&gctx->tctxs,
+					    to_destroy);
+					idalloctm(tsd_tsdn(tsd), to_destroy,
+					    NULL, NULL, true, true);
+				} else {
+					next = NULL;
+				}
+			} while (next != NULL);
+		}
+		gctx->nlimbo--;
+		if (prof_gctx_should_destroy(gctx)) {
+			gctx->nlimbo++;
+			malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+			prof_gctx_try_destroy(tsd, tdata, gctx, tdata);
+		} else {
+			malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+		}
+	}
+}
+
+struct prof_tdata_merge_iter_arg_s {
+	tsdn_t		*tsdn;
+	prof_cnt_t	cnt_all;
+};
+
+static prof_tdata_t *
+prof_tdata_merge_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *opaque) {
+	struct prof_tdata_merge_iter_arg_s *arg =
+	    (struct prof_tdata_merge_iter_arg_s *)opaque;
+
+	malloc_mutex_lock(arg->tsdn, tdata->lock);
+	if (!tdata->expired) {
+		size_t tabind;
+		union {
+			prof_tctx_t	*p;
+			void		*v;
+		} tctx;
+
+		tdata->dumping = true;
+		memset(&tdata->cnt_summed, 0, sizeof(prof_cnt_t));
+		for (tabind = 0; !ckh_iter(&tdata->bt2tctx, &tabind, NULL,
+		    &tctx.v);) {
+			prof_tctx_merge_tdata(arg->tsdn, tctx.p, tdata);
+		}
+
+		arg->cnt_all.curobjs += tdata->cnt_summed.curobjs;
+		arg->cnt_all.curbytes += tdata->cnt_summed.curbytes;
+		if (opt_prof_accum) {
+			arg->cnt_all.accumobjs += tdata->cnt_summed.accumobjs;
+			arg->cnt_all.accumbytes += tdata->cnt_summed.accumbytes;
+		}
+	} else {
+		tdata->dumping = false;
+	}
+	malloc_mutex_unlock(arg->tsdn, tdata->lock);
+
+	return NULL;
+}
+
+static prof_tdata_t *
+prof_tdata_dump_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *arg) {
+	bool propagate_err = *(bool *)arg;
+
+	if (!tdata->dumping) {
+		return NULL;
+	}
+
+	if (prof_dump_printf(propagate_err,
+	    "  t%"FMTu64": %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]%s%s\n",
+	    tdata->thr_uid, tdata->cnt_summed.curobjs,
+	    tdata->cnt_summed.curbytes, tdata->cnt_summed.accumobjs,
+	    tdata->cnt_summed.accumbytes,
+	    (tdata->thread_name != NULL) ? " " : "",
+	    (tdata->thread_name != NULL) ? tdata->thread_name : "")) {
+		return tdata;
+	}
+	return NULL;
+}
+
+static bool
+prof_dump_header_impl(tsdn_t *tsdn, bool propagate_err,
+    const prof_cnt_t *cnt_all) {
+	bool ret;
+
+	if (prof_dump_printf(propagate_err,
+	    "heap_v2/%"FMTu64"\n"
+	    "  t*: %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]\n",
+	    ((uint64_t)1U << lg_prof_sample), cnt_all->curobjs,
+	    cnt_all->curbytes, cnt_all->accumobjs, cnt_all->accumbytes)) {
+		return true;
+	}
+
+	malloc_mutex_lock(tsdn, &tdatas_mtx);
+	ret = (tdata_tree_iter(&tdatas, NULL, prof_tdata_dump_iter,
+	    (void *)&propagate_err) != NULL);
+	malloc_mutex_unlock(tsdn, &tdatas_mtx);
+	return ret;
+}
+prof_dump_header_t *JET_MUTABLE prof_dump_header = prof_dump_header_impl;
+
+static bool
+prof_dump_gctx(tsdn_t *tsdn, bool propagate_err, prof_gctx_t *gctx,
+    const prof_bt_t *bt, prof_gctx_tree_t *gctxs) {
+	bool ret;
+	unsigned i;
+	struct prof_tctx_dump_iter_arg_s prof_tctx_dump_iter_arg;
+
+	cassert(config_prof);
+	malloc_mutex_assert_owner(tsdn, gctx->lock);
+
+	/* Avoid dumping such gctx's that have no useful data. */
+	if ((!opt_prof_accum && gctx->cnt_summed.curobjs == 0) ||
+	    (opt_prof_accum && gctx->cnt_summed.accumobjs == 0)) {
+		assert(gctx->cnt_summed.curobjs == 0);
+		assert(gctx->cnt_summed.curbytes == 0);
+		assert(gctx->cnt_summed.accumobjs == 0);
+		assert(gctx->cnt_summed.accumbytes == 0);
+		ret = false;
+		goto label_return;
+	}
+
+	if (prof_dump_printf(propagate_err, "@")) {
+		ret = true;
+		goto label_return;
+	}
+	for (i = 0; i < bt->len; i++) {
+		if (prof_dump_printf(propagate_err, " %#"FMTxPTR,
+		    (uintptr_t)bt->vec[i])) {
+			ret = true;
+			goto label_return;
+		}
+	}
+
+	if (prof_dump_printf(propagate_err,
+	    "\n"
+	    "  t*: %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]\n",
+	    gctx->cnt_summed.curobjs, gctx->cnt_summed.curbytes,
+	    gctx->cnt_summed.accumobjs, gctx->cnt_summed.accumbytes)) {
+		ret = true;
+		goto label_return;
+	}
+
+	prof_tctx_dump_iter_arg.tsdn = tsdn;
+	prof_tctx_dump_iter_arg.propagate_err = propagate_err;
+	if (tctx_tree_iter(&gctx->tctxs, NULL, prof_tctx_dump_iter,
+	    (void *)&prof_tctx_dump_iter_arg) != NULL) {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+#ifndef _WIN32
+JEMALLOC_FORMAT_PRINTF(1, 2)
+static int
+prof_open_maps(const char *format, ...) {
+	int mfd;
+	va_list ap;
+	char filename[PATH_MAX + 1];
+
+	va_start(ap, format);
+	malloc_vsnprintf(filename, sizeof(filename), format, ap);
+	va_end(ap);
+
+#if defined(O_CLOEXEC)
+	mfd = open(filename, O_RDONLY | O_CLOEXEC);
+#else
+	mfd = open(filename, O_RDONLY);
+	if (mfd != -1) {
+		fcntl(mfd, F_SETFD, fcntl(mfd, F_GETFD) | FD_CLOEXEC);
+	}
+#endif
+
+	return mfd;
+}
+#endif
+
+static int
+prof_getpid(void) {
+#ifdef _WIN32
+	return GetCurrentProcessId();
+#else
+	return getpid();
+#endif
+}
+
+static bool
+prof_dump_maps(bool propagate_err) {
+	bool ret;
+	int mfd;
+
+	cassert(config_prof);
+#ifdef __FreeBSD__
+	mfd = prof_open_maps("/proc/curproc/map");
+#elif defined(_WIN32)
+	mfd = -1; // Not implemented
+#else
+	{
+		int pid = prof_getpid();
+
+		mfd = prof_open_maps("/proc/%d/task/%d/maps", pid, pid);
+		if (mfd == -1) {
+			mfd = prof_open_maps("/proc/%d/maps", pid);
+		}
+	}
+#endif
+	if (mfd != -1) {
+		ssize_t nread;
+
+		if (prof_dump_write(propagate_err, "\nMAPPED_LIBRARIES:\n") &&
+		    propagate_err) {
+			ret = true;
+			goto label_return;
+		}
+		nread = 0;
+		do {
+			prof_dump_buf_end += nread;
+			if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) {
+				/* Make space in prof_dump_buf before read(). */
+				if (prof_dump_flush(propagate_err) &&
+				    propagate_err) {
+					ret = true;
+					goto label_return;
+				}
+			}
+			nread = malloc_read_fd(mfd,
+			    &prof_dump_buf[prof_dump_buf_end], PROF_DUMP_BUFSIZE
+			    - prof_dump_buf_end);
+		} while (nread > 0);
+	} else {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	if (mfd != -1) {
+		close(mfd);
+	}
+	return ret;
+}
+
+/*
+ * See prof_sample_threshold_update() comment for why the body of this function
+ * is conditionally compiled.
+ */
+static void
+prof_leakcheck(const prof_cnt_t *cnt_all, size_t leak_ngctx,
+    const char *filename) {
+#ifdef JEMALLOC_PROF
+	/*
+	 * Scaling is equivalent AdjustSamples() in jeprof, but the result may
+	 * differ slightly from what jeprof reports, because here we scale the
+	 * summary values, whereas jeprof scales each context individually and
+	 * reports the sums of the scaled values.
+	 */
+	if (cnt_all->curbytes != 0) {
+		double sample_period = (double)((uint64_t)1 << lg_prof_sample);
+		double ratio = (((double)cnt_all->curbytes) /
+		    (double)cnt_all->curobjs) / sample_period;
+		double scale_factor = 1.0 / (1.0 - exp(-ratio));
+		uint64_t curbytes = (uint64_t)round(((double)cnt_all->curbytes)
+		    * scale_factor);
+		uint64_t curobjs = (uint64_t)round(((double)cnt_all->curobjs) *
+		    scale_factor);
+
+		malloc_printf("<jemalloc>: Leak approximation summary: ~%"FMTu64
+		    " byte%s, ~%"FMTu64" object%s, >= %zu context%s\n",
+		    curbytes, (curbytes != 1) ? "s" : "", curobjs, (curobjs !=
+		    1) ? "s" : "", leak_ngctx, (leak_ngctx != 1) ? "s" : "");
+		malloc_printf(
+		    "<jemalloc>: Run jeprof on \"%s\" for leak detail\n",
+		    filename);
+	}
+#endif
+}
+
+struct prof_gctx_dump_iter_arg_s {
+	tsdn_t	*tsdn;
+	bool	propagate_err;
+};
+
+static prof_gctx_t *
+prof_gctx_dump_iter(prof_gctx_tree_t *gctxs, prof_gctx_t *gctx, void *opaque) {
+	prof_gctx_t *ret;
+	struct prof_gctx_dump_iter_arg_s *arg =
+	    (struct prof_gctx_dump_iter_arg_s *)opaque;
+
+	malloc_mutex_lock(arg->tsdn, gctx->lock);
+
+	if (prof_dump_gctx(arg->tsdn, arg->propagate_err, gctx, &gctx->bt,
+	    gctxs)) {
+		ret = gctx;
+		goto label_return;
+	}
+
+	ret = NULL;
+label_return:
+	malloc_mutex_unlock(arg->tsdn, gctx->lock);
+	return ret;
+}
+
+static void
+prof_dump_prep(tsd_t *tsd, prof_tdata_t *tdata,
+    struct prof_tdata_merge_iter_arg_s *prof_tdata_merge_iter_arg,
+    struct prof_gctx_merge_iter_arg_s *prof_gctx_merge_iter_arg,
+    prof_gctx_tree_t *gctxs) {
+	size_t tabind;
+	union {
+		prof_gctx_t	*p;
+		void		*v;
+	} gctx;
+
+	prof_enter(tsd, tdata);
+
+	/*
+	 * Put gctx's in limbo and clear their counters in preparation for
+	 * summing.
+	 */
+	gctx_tree_new(gctxs);
+	for (tabind = 0; !ckh_iter(&bt2gctx, &tabind, NULL, &gctx.v);) {
+		prof_dump_gctx_prep(tsd_tsdn(tsd), gctx.p, gctxs);
+	}
+
+	/*
+	 * Iterate over tdatas, and for the non-expired ones snapshot their tctx
+	 * stats and merge them into the associated gctx's.
+	 */
+	prof_tdata_merge_iter_arg->tsdn = tsd_tsdn(tsd);
+	memset(&prof_tdata_merge_iter_arg->cnt_all, 0, sizeof(prof_cnt_t));
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+	tdata_tree_iter(&tdatas, NULL, prof_tdata_merge_iter,
+	    (void *)prof_tdata_merge_iter_arg);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+
+	/* Merge tctx stats into gctx's. */
+	prof_gctx_merge_iter_arg->tsdn = tsd_tsdn(tsd);
+	prof_gctx_merge_iter_arg->leak_ngctx = 0;
+	gctx_tree_iter(gctxs, NULL, prof_gctx_merge_iter,
+	    (void *)prof_gctx_merge_iter_arg);
+
+	prof_leave(tsd, tdata);
+}
+
+static bool
+prof_dump_file(tsd_t *tsd, bool propagate_err, const char *filename,
+    bool leakcheck, prof_tdata_t *tdata,
+    struct prof_tdata_merge_iter_arg_s *prof_tdata_merge_iter_arg,
+    struct prof_gctx_merge_iter_arg_s *prof_gctx_merge_iter_arg,
+    struct prof_gctx_dump_iter_arg_s *prof_gctx_dump_iter_arg,
+    prof_gctx_tree_t *gctxs) {
+	/* Create dump file. */
+	if ((prof_dump_fd = prof_dump_open(propagate_err, filename)) == -1) {
+		return true;
+	}
+
+	/* Dump profile header. */
+	if (prof_dump_header(tsd_tsdn(tsd), propagate_err,
+	    &prof_tdata_merge_iter_arg->cnt_all)) {
+		goto label_write_error;
+	}
+
+	/* Dump per gctx profile stats. */
+	prof_gctx_dump_iter_arg->tsdn = tsd_tsdn(tsd);
+	prof_gctx_dump_iter_arg->propagate_err = propagate_err;
+	if (gctx_tree_iter(gctxs, NULL, prof_gctx_dump_iter,
+	    (void *)prof_gctx_dump_iter_arg) != NULL) {
+		goto label_write_error;
+	}
+
+	/* Dump /proc/<pid>/maps if possible. */
+	if (prof_dump_maps(propagate_err)) {
+		goto label_write_error;
+	}
+
+	if (prof_dump_close(propagate_err)) {
+		return true;
+	}
+
+	return false;
+label_write_error:
+	prof_dump_close(propagate_err);
+	return true;
+}
+
+static bool
+prof_dump(tsd_t *tsd, bool propagate_err, const char *filename,
+    bool leakcheck) {
+	cassert(config_prof);
+	assert(tsd_reentrancy_level_get(tsd) == 0);
+
+	prof_tdata_t * tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return true;
+	}
+
+	pre_reentrancy(tsd, NULL);
+	malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_mtx);
+
+	prof_gctx_tree_t gctxs;
+	struct prof_tdata_merge_iter_arg_s prof_tdata_merge_iter_arg;
+	struct prof_gctx_merge_iter_arg_s prof_gctx_merge_iter_arg;
+	struct prof_gctx_dump_iter_arg_s prof_gctx_dump_iter_arg;
+	prof_dump_prep(tsd, tdata, &prof_tdata_merge_iter_arg,
+	    &prof_gctx_merge_iter_arg, &gctxs);
+	bool err = prof_dump_file(tsd, propagate_err, filename, leakcheck, tdata,
+	    &prof_tdata_merge_iter_arg, &prof_gctx_merge_iter_arg,
+	    &prof_gctx_dump_iter_arg, &gctxs);
+	prof_gctx_finish(tsd, &gctxs);
+
+	malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_mtx);
+	post_reentrancy(tsd);
+
+	if (err) {
+		return true;
+	}
+
+	if (leakcheck) {
+		prof_leakcheck(&prof_tdata_merge_iter_arg.cnt_all,
+		    prof_gctx_merge_iter_arg.leak_ngctx, filename);
+	}
+	return false;
+}
+
+#ifdef JEMALLOC_JET
+void
+prof_cnt_all(uint64_t *curobjs, uint64_t *curbytes, uint64_t *accumobjs,
+    uint64_t *accumbytes) {
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+	struct prof_tdata_merge_iter_arg_s prof_tdata_merge_iter_arg;
+	struct prof_gctx_merge_iter_arg_s prof_gctx_merge_iter_arg;
+	prof_gctx_tree_t gctxs;
+
+	tsd = tsd_fetch();
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		if (curobjs != NULL) {
+			*curobjs = 0;
+		}
+		if (curbytes != NULL) {
+			*curbytes = 0;
+		}
+		if (accumobjs != NULL) {
+			*accumobjs = 0;
+		}
+		if (accumbytes != NULL) {
+			*accumbytes = 0;
+		}
+		return;
+	}
+
+	prof_dump_prep(tsd, tdata, &prof_tdata_merge_iter_arg,
+	    &prof_gctx_merge_iter_arg, &gctxs);
+	prof_gctx_finish(tsd, &gctxs);
+
+	if (curobjs != NULL) {
+		*curobjs = prof_tdata_merge_iter_arg.cnt_all.curobjs;
+	}
+	if (curbytes != NULL) {
+		*curbytes = prof_tdata_merge_iter_arg.cnt_all.curbytes;
+	}
+	if (accumobjs != NULL) {
+		*accumobjs = prof_tdata_merge_iter_arg.cnt_all.accumobjs;
+	}
+	if (accumbytes != NULL) {
+		*accumbytes = prof_tdata_merge_iter_arg.cnt_all.accumbytes;
+	}
+}
+#endif
+
+#define DUMP_FILENAME_BUFSIZE	(PATH_MAX + 1)
+#define VSEQ_INVALID		UINT64_C(0xffffffffffffffff)
+static void
+prof_dump_filename(char *filename, char v, uint64_t vseq) {
+	cassert(config_prof);
+
+	if (vseq != VSEQ_INVALID) {
+	        /* "<prefix>.<pid>.<seq>.v<vseq>.heap" */
+		malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE,
+		    "%s.%d.%"FMTu64".%c%"FMTu64".heap",
+		    opt_prof_prefix, prof_getpid(), prof_dump_seq, v, vseq);
+	} else {
+	        /* "<prefix>.<pid>.<seq>.<v>.heap" */
+		malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE,
+		    "%s.%d.%"FMTu64".%c.heap",
+		    opt_prof_prefix, prof_getpid(), prof_dump_seq, v);
+	}
+	prof_dump_seq++;
+}
+
+static void
+prof_fdump(void) {
+	tsd_t *tsd;
+	char filename[DUMP_FILENAME_BUFSIZE];
+
+	cassert(config_prof);
+	assert(opt_prof_final);
+	assert(opt_prof_prefix[0] != '\0');
+
+	if (!prof_booted) {
+		return;
+	}
+	tsd = tsd_fetch();
+	assert(tsd_reentrancy_level_get(tsd) == 0);
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+	prof_dump_filename(filename, 'f', VSEQ_INVALID);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+	prof_dump(tsd, false, filename, opt_prof_leak);
+}
+
+bool
+prof_accum_init(tsdn_t *tsdn, prof_accum_t *prof_accum) {
+	cassert(config_prof);
+
+#ifndef JEMALLOC_ATOMIC_U64
+	if (malloc_mutex_init(&prof_accum->mtx, "prof_accum",
+	    WITNESS_RANK_PROF_ACCUM, malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	prof_accum->accumbytes = 0;
+#else
+	atomic_store_u64(&prof_accum->accumbytes, 0, ATOMIC_RELAXED);
+#endif
+	return false;
+}
+
+void
+prof_idump(tsdn_t *tsdn) {
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	if (!prof_booted || tsdn_null(tsdn) || !prof_active_get_unlocked()) {
+		return;
+	}
+	tsd = tsdn_tsd(tsdn);
+	if (tsd_reentrancy_level_get(tsd) > 0) {
+		return;
+	}
+
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return;
+	}
+	if (tdata->enq) {
+		tdata->enq_idump = true;
+		return;
+	}
+
+	if (opt_prof_prefix[0] != '\0') {
+		char filename[PATH_MAX + 1];
+		malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		prof_dump_filename(filename, 'i', prof_dump_iseq);
+		prof_dump_iseq++;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		prof_dump(tsd, false, filename, false);
+	}
+}
+
+bool
+prof_mdump(tsd_t *tsd, const char *filename) {
+	cassert(config_prof);
+	assert(tsd_reentrancy_level_get(tsd) == 0);
+
+	if (!opt_prof || !prof_booted) {
+		return true;
+	}
+	char filename_buf[DUMP_FILENAME_BUFSIZE];
+	if (filename == NULL) {
+		/* No filename specified, so automatically generate one. */
+		if (opt_prof_prefix[0] == '\0') {
+			return true;
+		}
+		malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		prof_dump_filename(filename_buf, 'm', prof_dump_mseq);
+		prof_dump_mseq++;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		filename = filename_buf;
+	}
+	return prof_dump(tsd, true, filename, false);
+}
+
+void
+prof_gdump(tsdn_t *tsdn) {
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	if (!prof_booted || tsdn_null(tsdn) || !prof_active_get_unlocked()) {
+		return;
+	}
+	tsd = tsdn_tsd(tsdn);
+	if (tsd_reentrancy_level_get(tsd) > 0) {
+		return;
+	}
+
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return;
+	}
+	if (tdata->enq) {
+		tdata->enq_gdump = true;
+		return;
+	}
+
+	if (opt_prof_prefix[0] != '\0') {
+		char filename[DUMP_FILENAME_BUFSIZE];
+		malloc_mutex_lock(tsdn, &prof_dump_seq_mtx);
+		prof_dump_filename(filename, 'u', prof_dump_useq);
+		prof_dump_useq++;
+		malloc_mutex_unlock(tsdn, &prof_dump_seq_mtx);
+		prof_dump(tsd, false, filename, false);
+	}
+}
+
+static void
+prof_bt_hash(const void *key, size_t r_hash[2]) {
+	prof_bt_t *bt = (prof_bt_t *)key;
+
+	cassert(config_prof);
+
+	hash(bt->vec, bt->len * sizeof(void *), 0x94122f33U, r_hash);
+}
+
+static bool
+prof_bt_keycomp(const void *k1, const void *k2) {
+	const prof_bt_t *bt1 = (prof_bt_t *)k1;
+	const prof_bt_t *bt2 = (prof_bt_t *)k2;
+
+	cassert(config_prof);
+
+	if (bt1->len != bt2->len) {
+		return false;
+	}
+	return (memcmp(bt1->vec, bt2->vec, bt1->len * sizeof(void *)) == 0);
+}
+
+static void
+prof_bt_node_hash(const void *key, size_t r_hash[2]) {
+	const prof_bt_node_t *bt_node = (prof_bt_node_t *)key;
+	prof_bt_hash((void *)(&bt_node->bt), r_hash);
+}
+
+static bool
+prof_bt_node_keycomp(const void *k1, const void *k2) {
+	const prof_bt_node_t *bt_node1 = (prof_bt_node_t *)k1;
+	const prof_bt_node_t *bt_node2 = (prof_bt_node_t *)k2;
+	return prof_bt_keycomp((void *)(&bt_node1->bt),
+	    (void *)(&bt_node2->bt));
+}
+
+static void
+prof_thr_node_hash(const void *key, size_t r_hash[2]) {
+	const prof_thr_node_t *thr_node = (prof_thr_node_t *)key;
+	hash(&thr_node->thr_uid, sizeof(uint64_t), 0x94122f35U, r_hash);
+}
+
+static bool
+prof_thr_node_keycomp(const void *k1, const void *k2) {
+	const prof_thr_node_t *thr_node1 = (prof_thr_node_t *)k1;
+	const prof_thr_node_t *thr_node2 = (prof_thr_node_t *)k2;
+	return thr_node1->thr_uid == thr_node2->thr_uid;
+}
+
+static uint64_t
+prof_thr_uid_alloc(tsdn_t *tsdn) {
+	uint64_t thr_uid;
+
+	malloc_mutex_lock(tsdn, &next_thr_uid_mtx);
+	thr_uid = next_thr_uid;
+	next_thr_uid++;
+	malloc_mutex_unlock(tsdn, &next_thr_uid_mtx);
+
+	return thr_uid;
+}
+
+static prof_tdata_t *
+prof_tdata_init_impl(tsd_t *tsd, uint64_t thr_uid, uint64_t thr_discrim,
+    char *thread_name, bool active) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	/* Initialize an empty cache for this thread. */
+	tdata = (prof_tdata_t *)iallocztm(tsd_tsdn(tsd), sizeof(prof_tdata_t),
+	    sz_size2index(sizeof(prof_tdata_t)), false, NULL, true,
+	    arena_get(TSDN_NULL, 0, true), true);
+	if (tdata == NULL) {
+		return NULL;
+	}
+
+	tdata->lock = prof_tdata_mutex_choose(thr_uid);
+	tdata->thr_uid = thr_uid;
+	tdata->thr_discrim = thr_discrim;
+	tdata->thread_name = thread_name;
+	tdata->attached = true;
+	tdata->expired = false;
+	tdata->tctx_uid_next = 0;
+
+	if (ckh_new(tsd, &tdata->bt2tctx, PROF_CKH_MINITEMS, prof_bt_hash,
+	    prof_bt_keycomp)) {
+		idalloctm(tsd_tsdn(tsd), tdata, NULL, NULL, true, true);
+		return NULL;
+	}
+
+	tdata->prng_state = (uint64_t)(uintptr_t)tdata;
+	prof_sample_threshold_update(tdata);
+
+	tdata->enq = false;
+	tdata->enq_idump = false;
+	tdata->enq_gdump = false;
+
+	tdata->dumping = false;
+	tdata->active = active;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+	tdata_tree_insert(&tdatas, tdata);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+
+	return tdata;
+}
+
+prof_tdata_t *
+prof_tdata_init(tsd_t *tsd) {
+	return prof_tdata_init_impl(tsd, prof_thr_uid_alloc(tsd_tsdn(tsd)), 0,
+	    NULL, prof_thread_active_init_get(tsd_tsdn(tsd)));
+}
+
+static bool
+prof_tdata_should_destroy_unlocked(prof_tdata_t *tdata, bool even_if_attached) {
+	if (tdata->attached && !even_if_attached) {
+		return false;
+	}
+	if (ckh_count(&tdata->bt2tctx) != 0) {
+		return false;
+	}
+	return true;
+}
+
+static bool
+prof_tdata_should_destroy(tsdn_t *tsdn, prof_tdata_t *tdata,
+    bool even_if_attached) {
+	malloc_mutex_assert_owner(tsdn, tdata->lock);
+
+	return prof_tdata_should_destroy_unlocked(tdata, even_if_attached);
+}
+
+static void
+prof_tdata_destroy_locked(tsd_t *tsd, prof_tdata_t *tdata,
+    bool even_if_attached) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &tdatas_mtx);
+
+	tdata_tree_remove(&tdatas, tdata);
+
+	assert(prof_tdata_should_destroy_unlocked(tdata, even_if_attached));
+
+	if (tdata->thread_name != NULL) {
+		idalloctm(tsd_tsdn(tsd), tdata->thread_name, NULL, NULL, true,
+		    true);
+	}
+	ckh_delete(tsd, &tdata->bt2tctx);
+	idalloctm(tsd_tsdn(tsd), tdata, NULL, NULL, true, true);
+}
+
+static void
+prof_tdata_destroy(tsd_t *tsd, prof_tdata_t *tdata, bool even_if_attached) {
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+	prof_tdata_destroy_locked(tsd, tdata, even_if_attached);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+}
+
+static void
+prof_tdata_detach(tsd_t *tsd, prof_tdata_t *tdata) {
+	bool destroy_tdata;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), tdata->lock);
+	if (tdata->attached) {
+		destroy_tdata = prof_tdata_should_destroy(tsd_tsdn(tsd), tdata,
+		    true);
+		/*
+		 * Only detach if !destroy_tdata, because detaching would allow
+		 * another thread to win the race to destroy tdata.
+		 */
+		if (!destroy_tdata) {
+			tdata->attached = false;
+		}
+		tsd_prof_tdata_set(tsd, NULL);
+	} else {
+		destroy_tdata = false;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+	if (destroy_tdata) {
+		prof_tdata_destroy(tsd, tdata, true);
+	}
+}
+
+prof_tdata_t *
+prof_tdata_reinit(tsd_t *tsd, prof_tdata_t *tdata) {
+	uint64_t thr_uid = tdata->thr_uid;
+	uint64_t thr_discrim = tdata->thr_discrim + 1;
+	char *thread_name = (tdata->thread_name != NULL) ?
+	    prof_thread_name_alloc(tsd_tsdn(tsd), tdata->thread_name) : NULL;
+	bool active = tdata->active;
+
+	prof_tdata_detach(tsd, tdata);
+	return prof_tdata_init_impl(tsd, thr_uid, thr_discrim, thread_name,
+	    active);
+}
+
+static bool
+prof_tdata_expire(tsdn_t *tsdn, prof_tdata_t *tdata) {
+	bool destroy_tdata;
+
+	malloc_mutex_lock(tsdn, tdata->lock);
+	if (!tdata->expired) {
+		tdata->expired = true;
+		destroy_tdata = tdata->attached ? false :
+		    prof_tdata_should_destroy(tsdn, tdata, false);
+	} else {
+		destroy_tdata = false;
+	}
+	malloc_mutex_unlock(tsdn, tdata->lock);
+
+	return destroy_tdata;
+}
+
+static prof_tdata_t *
+prof_tdata_reset_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *arg) {
+	tsdn_t *tsdn = (tsdn_t *)arg;
+
+	return (prof_tdata_expire(tsdn, tdata) ? tdata : NULL);
+}
+
+void
+prof_reset(tsd_t *tsd, size_t lg_sample) {
+	prof_tdata_t *next;
+
+	assert(lg_sample < (sizeof(uint64_t) << 3));
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_mtx);
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+
+	lg_prof_sample = lg_sample;
+
+	next = NULL;
+	do {
+		prof_tdata_t *to_destroy = tdata_tree_iter(&tdatas, next,
+		    prof_tdata_reset_iter, (void *)tsd);
+		if (to_destroy != NULL) {
+			next = tdata_tree_next(&tdatas, to_destroy);
+			prof_tdata_destroy_locked(tsd, to_destroy, false);
+		} else {
+			next = NULL;
+		}
+	} while (next != NULL);
+
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_mtx);
+}
+
+void
+prof_tdata_cleanup(tsd_t *tsd) {
+	prof_tdata_t *tdata;
+
+	if (!config_prof) {
+		return;
+	}
+
+	tdata = tsd_prof_tdata_get(tsd);
+	if (tdata != NULL) {
+		prof_tdata_detach(tsd, tdata);
+	}
+}
+
+bool
+prof_active_get(tsdn_t *tsdn) {
+	bool prof_active_current;
+
+	malloc_mutex_lock(tsdn, &prof_active_mtx);
+	prof_active_current = prof_active;
+	malloc_mutex_unlock(tsdn, &prof_active_mtx);
+	return prof_active_current;
+}
+
+bool
+prof_active_set(tsdn_t *tsdn, bool active) {
+	bool prof_active_old;
+
+	malloc_mutex_lock(tsdn, &prof_active_mtx);
+	prof_active_old = prof_active;
+	prof_active = active;
+	malloc_mutex_unlock(tsdn, &prof_active_mtx);
+	return prof_active_old;
+}
+
+#ifdef JEMALLOC_JET
+size_t
+prof_log_bt_count(void) {
+	size_t cnt = 0;
+	prof_bt_node_t *node = log_bt_first;
+	while (node != NULL) {
+		cnt++;
+		node = node->next;
+	}
+	return cnt;
+}
+
+size_t
+prof_log_alloc_count(void) {
+	size_t cnt = 0;
+	prof_alloc_node_t *node = log_alloc_first;
+	while (node != NULL) {
+		cnt++;
+		node = node->next;
+	}
+	return cnt;
+}
+
+size_t
+prof_log_thr_count(void) {
+	size_t cnt = 0;
+	prof_thr_node_t *node = log_thr_first;
+	while (node != NULL) {
+		cnt++;
+		node = node->next;
+	}
+	return cnt;
+}
+
+bool
+prof_log_is_logging(void) {
+	return prof_logging_state == prof_logging_state_started;
+}
+
+bool
+prof_log_rep_check(void) {
+	if (prof_logging_state == prof_logging_state_stopped
+	    && log_tables_initialized) {
+		return true;
+	}
+
+	if (log_bt_last != NULL && log_bt_last->next != NULL) {
+		return true;
+	}
+	if (log_thr_last != NULL && log_thr_last->next != NULL) {
+		return true;
+	}
+	if (log_alloc_last != NULL && log_alloc_last->next != NULL) {
+		return true;
+	}
+
+	size_t bt_count = prof_log_bt_count();
+	size_t thr_count = prof_log_thr_count();
+	size_t alloc_count = prof_log_alloc_count();
+
+
+	if (prof_logging_state == prof_logging_state_stopped) {
+		if (bt_count != 0 || thr_count != 0 || alloc_count || 0) {
+			return true;
+		}
+	}
+
+	prof_alloc_node_t *node = log_alloc_first;
+	while (node != NULL) {
+		if (node->alloc_bt_ind >= bt_count) {
+			return true;
+		}
+		if (node->free_bt_ind >= bt_count) {
+			return true;
+		}
+		if (node->alloc_thr_ind >= thr_count) {
+			return true;
+		}
+		if (node->free_thr_ind >= thr_count) {
+			return true;
+		}
+		if (node->alloc_time_ns > node->free_time_ns) {
+			return true;
+		}
+		node = node->next;
+	}
+
+	return false;
+}
+
+void
+prof_log_dummy_set(bool new_value) {
+	prof_log_dummy = new_value;
+}
+#endif
+
+bool
+prof_log_start(tsdn_t *tsdn, const char *filename) {
+	if (!opt_prof || !prof_booted) {
+		return true;
+	}
+
+	bool ret = false;
+	size_t buf_size = PATH_MAX + 1;
+
+	malloc_mutex_lock(tsdn, &log_mtx);
+
+	if (prof_logging_state != prof_logging_state_stopped) {
+		ret = true;
+	} else if (filename == NULL) {
+		/* Make default name. */
+		malloc_snprintf(log_filename, buf_size, "%s.%d.%"FMTu64".json",
+		    opt_prof_prefix, prof_getpid(), log_seq);
+		log_seq++;
+		prof_logging_state = prof_logging_state_started;
+	} else if (strlen(filename) >= buf_size) {
+		ret = true;
+	} else {
+		strcpy(log_filename, filename);
+		prof_logging_state = prof_logging_state_started;
+	}
+
+	if (!ret) {
+		nstime_update(&log_start_timestamp);
+	}
+
+	malloc_mutex_unlock(tsdn, &log_mtx);
+
+	return ret;
+}
+
+/* Used as an atexit function to stop logging on exit. */
+static void
+prof_log_stop_final(void) {
+	tsd_t *tsd = tsd_fetch();
+	prof_log_stop(tsd_tsdn(tsd));
+}
+
+struct prof_emitter_cb_arg_s {
+	int fd;
+	ssize_t ret;
+};
+
+static void
+prof_emitter_write_cb(void *opaque, const char *to_write) {
+	struct prof_emitter_cb_arg_s *arg =
+	    (struct prof_emitter_cb_arg_s *)opaque;
+	size_t bytes = strlen(to_write);
+#ifdef JEMALLOC_JET
+	if (prof_log_dummy) {
+		return;
+	}
+#endif
+	arg->ret = write(arg->fd, (void *)to_write, bytes);
+}
+
+/*
+ * prof_log_emit_{...} goes through the appropriate linked list, emitting each
+ * node to the json and deallocating it.
+ */
+static void
+prof_log_emit_threads(tsd_t *tsd, emitter_t *emitter) {
+	emitter_json_array_kv_begin(emitter, "threads");
+	prof_thr_node_t *thr_node = log_thr_first;
+	prof_thr_node_t *thr_old_node;
+	while (thr_node != NULL) {
+		emitter_json_object_begin(emitter);
+
+		emitter_json_kv(emitter, "thr_uid", emitter_type_uint64,
+		    &thr_node->thr_uid);
+
+		char *thr_name = thr_node->name;
+
+		emitter_json_kv(emitter, "thr_name", emitter_type_string,
+		    &thr_name);
+
+		emitter_json_object_end(emitter);
+		thr_old_node = thr_node;
+		thr_node = thr_node->next;
+		idalloc(tsd, thr_old_node);
+	}
+	emitter_json_array_end(emitter);
+}
+
+static void
+prof_log_emit_traces(tsd_t *tsd, emitter_t *emitter) {
+	emitter_json_array_kv_begin(emitter, "stack_traces");
+	prof_bt_node_t *bt_node = log_bt_first;
+	prof_bt_node_t *bt_old_node;
+	/*
+	 * Calculate how many hex digits we need: twice number of bytes, two for
+	 * "0x", and then one more for terminating '\0'.
+	 */
+	char buf[2 * sizeof(intptr_t) + 3];
+	size_t buf_sz = sizeof(buf);
+	while (bt_node != NULL) {
+		emitter_json_array_begin(emitter);
+		size_t i;
+		for (i = 0; i < bt_node->bt.len; i++) {
+			malloc_snprintf(buf, buf_sz, "%p", bt_node->bt.vec[i]);
+			char *trace_str = buf;
+			emitter_json_value(emitter, emitter_type_string,
+			    &trace_str);
+		}
+		emitter_json_array_end(emitter);
+
+		bt_old_node = bt_node;
+		bt_node = bt_node->next;
+		idalloc(tsd, bt_old_node);
+	}
+	emitter_json_array_end(emitter);
+}
+
+static void
+prof_log_emit_allocs(tsd_t *tsd, emitter_t *emitter) {
+	emitter_json_array_kv_begin(emitter, "allocations");
+	prof_alloc_node_t *alloc_node = log_alloc_first;
+	prof_alloc_node_t *alloc_old_node;
+	while (alloc_node != NULL) {
+		emitter_json_object_begin(emitter);
+
+		emitter_json_kv(emitter, "alloc_thread", emitter_type_size,
+		    &alloc_node->alloc_thr_ind);
+
+		emitter_json_kv(emitter, "free_thread", emitter_type_size,
+		    &alloc_node->free_thr_ind);
+
+		emitter_json_kv(emitter, "alloc_trace", emitter_type_size,
+		    &alloc_node->alloc_bt_ind);
+
+		emitter_json_kv(emitter, "free_trace", emitter_type_size,
+		    &alloc_node->free_bt_ind);
+
+		emitter_json_kv(emitter, "alloc_timestamp",
+		    emitter_type_uint64, &alloc_node->alloc_time_ns);
+
+		emitter_json_kv(emitter, "free_timestamp", emitter_type_uint64,
+		    &alloc_node->free_time_ns);
+
+		emitter_json_kv(emitter, "usize", emitter_type_uint64,
+		    &alloc_node->usize);
+
+		emitter_json_object_end(emitter);
+
+		alloc_old_node = alloc_node;
+		alloc_node = alloc_node->next;
+		idalloc(tsd, alloc_old_node);
+	}
+	emitter_json_array_end(emitter);
+}
+
+static void
+prof_log_emit_metadata(emitter_t *emitter) {
+	emitter_json_object_kv_begin(emitter, "info");
+
+	nstime_t now = NSTIME_ZERO_INITIALIZER;
+
+	nstime_update(&now);
+	uint64_t ns = nstime_ns(&now) - nstime_ns(&log_start_timestamp);
+	emitter_json_kv(emitter, "duration", emitter_type_uint64, &ns);
+
+	char *vers = JEMALLOC_VERSION;
+	emitter_json_kv(emitter, "version",
+	    emitter_type_string, &vers);
+
+	emitter_json_kv(emitter, "lg_sample_rate",
+	    emitter_type_int, &lg_prof_sample);
+
+	int pid = prof_getpid();
+	emitter_json_kv(emitter, "pid", emitter_type_int, &pid);
+
+	emitter_json_object_end(emitter);
+}
+
+
+bool
+prof_log_stop(tsdn_t *tsdn) {
+	if (!opt_prof || !prof_booted) {
+		return true;
+	}
+
+	tsd_t *tsd = tsdn_tsd(tsdn);
+	malloc_mutex_lock(tsdn, &log_mtx);
+
+	if (prof_logging_state != prof_logging_state_started) {
+		malloc_mutex_unlock(tsdn, &log_mtx);
+		return true;
+	}
+
+	/*
+	 * Set the state to dumping. We'll set it to stopped when we're done.
+	 * Since other threads won't be able to start/stop/log when the state is
+	 * dumping, we don't have to hold the lock during the whole method.
+	 */
+	prof_logging_state = prof_logging_state_dumping;
+	malloc_mutex_unlock(tsdn, &log_mtx);
+
+
+	emitter_t emitter;
+
+	/* Create a file. */
+
+	int fd;
+#ifdef JEMALLOC_JET
+	if (prof_log_dummy) {
+		fd = 0;
+	} else {
+		fd = creat(log_filename, 0644);
+	}
+#else
+	fd = creat(log_filename, 0644);
+#endif
+
+	if (fd == -1) {
+		malloc_printf("<jemalloc>: creat() for log file \"%s\" "
+			      " failed with %d\n", log_filename, errno);
+		if (opt_abort) {
+			abort();
+		}
+		return true;
+	}
+
+	/* Emit to json. */
+	struct prof_emitter_cb_arg_s arg;
+	arg.fd = fd;
+	emitter_init(&emitter, emitter_output_json, &prof_emitter_write_cb,
+	    (void *)(&arg));
+
+	emitter_begin(&emitter);
+	prof_log_emit_metadata(&emitter);
+	prof_log_emit_threads(tsd, &emitter);
+	prof_log_emit_traces(tsd, &emitter);
+	prof_log_emit_allocs(tsd, &emitter);
+	emitter_end(&emitter);
+
+	/* Reset global state. */
+	if (log_tables_initialized) {
+		ckh_delete(tsd, &log_bt_node_set);
+		ckh_delete(tsd, &log_thr_node_set);
+	}
+	log_tables_initialized = false;
+	log_bt_index = 0;
+	log_thr_index = 0;
+	log_bt_first = NULL;
+	log_bt_last = NULL;
+	log_thr_first = NULL;
+	log_thr_last = NULL;
+	log_alloc_first = NULL;
+	log_alloc_last = NULL;
+
+	malloc_mutex_lock(tsdn, &log_mtx);
+	prof_logging_state = prof_logging_state_stopped;
+	malloc_mutex_unlock(tsdn, &log_mtx);
+
+#ifdef JEMALLOC_JET
+	if (prof_log_dummy) {
+		return false;
+	}
+#endif
+	return close(fd);
+}
+
+const char *
+prof_thread_name_get(tsd_t *tsd) {
+	prof_tdata_t *tdata;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return "";
+	}
+	return (tdata->thread_name != NULL ? tdata->thread_name : "");
+}
+
+static char *
+prof_thread_name_alloc(tsdn_t *tsdn, const char *thread_name) {
+	char *ret;
+	size_t size;
+
+	if (thread_name == NULL) {
+		return NULL;
+	}
+
+	size = strlen(thread_name) + 1;
+	if (size == 1) {
+		return "";
+	}
+
+	ret = iallocztm(tsdn, size, sz_size2index(size), false, NULL, true,
+	    arena_get(TSDN_NULL, 0, true), true);
+	if (ret == NULL) {
+		return NULL;
+	}
+	memcpy(ret, thread_name, size);
+	return ret;
+}
+
+int
+prof_thread_name_set(tsd_t *tsd, const char *thread_name) {
+	prof_tdata_t *tdata;
+	unsigned i;
+	char *s;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return EAGAIN;
+	}
+
+	/* Validate input. */
+	if (thread_name == NULL) {
+		return EFAULT;
+	}
+	for (i = 0; thread_name[i] != '\0'; i++) {
+		char c = thread_name[i];
+		if (!isgraph(c) && !isblank(c)) {
+			return EFAULT;
+		}
+	}
+
+	s = prof_thread_name_alloc(tsd_tsdn(tsd), thread_name);
+	if (s == NULL) {
+		return EAGAIN;
+	}
+
+	if (tdata->thread_name != NULL) {
+		idalloctm(tsd_tsdn(tsd), tdata->thread_name, NULL, NULL, true,
+		    true);
+		tdata->thread_name = NULL;
+	}
+	if (strlen(s) > 0) {
+		tdata->thread_name = s;
+	}
+	return 0;
+}
+
+bool
+prof_thread_active_get(tsd_t *tsd) {
+	prof_tdata_t *tdata;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return false;
+	}
+	return tdata->active;
+}
+
+bool
+prof_thread_active_set(tsd_t *tsd, bool active) {
+	prof_tdata_t *tdata;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return true;
+	}
+	tdata->active = active;
+	return false;
+}
+
+bool
+prof_thread_active_init_get(tsdn_t *tsdn) {
+	bool active_init;
+
+	malloc_mutex_lock(tsdn, &prof_thread_active_init_mtx);
+	active_init = prof_thread_active_init;
+	malloc_mutex_unlock(tsdn, &prof_thread_active_init_mtx);
+	return active_init;
+}
+
+bool
+prof_thread_active_init_set(tsdn_t *tsdn, bool active_init) {
+	bool active_init_old;
+
+	malloc_mutex_lock(tsdn, &prof_thread_active_init_mtx);
+	active_init_old = prof_thread_active_init;
+	prof_thread_active_init = active_init;
+	malloc_mutex_unlock(tsdn, &prof_thread_active_init_mtx);
+	return active_init_old;
+}
+
+bool
+prof_gdump_get(tsdn_t *tsdn) {
+	bool prof_gdump_current;
+
+	malloc_mutex_lock(tsdn, &prof_gdump_mtx);
+	prof_gdump_current = prof_gdump_val;
+	malloc_mutex_unlock(tsdn, &prof_gdump_mtx);
+	return prof_gdump_current;
+}
+
+bool
+prof_gdump_set(tsdn_t *tsdn, bool gdump) {
+	bool prof_gdump_old;
+
+	malloc_mutex_lock(tsdn, &prof_gdump_mtx);
+	prof_gdump_old = prof_gdump_val;
+	prof_gdump_val = gdump;
+	malloc_mutex_unlock(tsdn, &prof_gdump_mtx);
+	return prof_gdump_old;
+}
+
+void
+prof_boot0(void) {
+	cassert(config_prof);
+
+	memcpy(opt_prof_prefix, PROF_PREFIX_DEFAULT,
+	    sizeof(PROF_PREFIX_DEFAULT));
+}
+
+void
+prof_boot1(void) {
+	cassert(config_prof);
+
+	/*
+	 * opt_prof must be in its final state before any arenas are
+	 * initialized, so this function must be executed early.
+	 */
+
+	if (opt_prof_leak && !opt_prof) {
+		/*
+		 * Enable opt_prof, but in such a way that profiles are never
+		 * automatically dumped.
+		 */
+		opt_prof = true;
+		opt_prof_gdump = false;
+	} else if (opt_prof) {
+		if (opt_lg_prof_interval >= 0) {
+			prof_interval = (((uint64_t)1U) <<
+			    opt_lg_prof_interval);
+		}
+	}
+}
+
+bool
+prof_boot2(tsd_t *tsd) {
+	cassert(config_prof);
+
+	if (opt_prof) {
+		unsigned i;
+
+		lg_prof_sample = opt_lg_prof_sample;
+
+		prof_active = opt_prof_active;
+		if (malloc_mutex_init(&prof_active_mtx, "prof_active",
+		    WITNESS_RANK_PROF_ACTIVE, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		prof_gdump_val = opt_prof_gdump;
+		if (malloc_mutex_init(&prof_gdump_mtx, "prof_gdump",
+		    WITNESS_RANK_PROF_GDUMP, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		prof_thread_active_init = opt_prof_thread_active_init;
+		if (malloc_mutex_init(&prof_thread_active_init_mtx,
+		    "prof_thread_active_init",
+		    WITNESS_RANK_PROF_THREAD_ACTIVE_INIT,
+		    malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (ckh_new(tsd, &bt2gctx, PROF_CKH_MINITEMS, prof_bt_hash,
+		    prof_bt_keycomp)) {
+			return true;
+		}
+		if (malloc_mutex_init(&bt2gctx_mtx, "prof_bt2gctx",
+		    WITNESS_RANK_PROF_BT2GCTX, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		tdata_tree_new(&tdatas);
+		if (malloc_mutex_init(&tdatas_mtx, "prof_tdatas",
+		    WITNESS_RANK_PROF_TDATAS, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		next_thr_uid = 0;
+		if (malloc_mutex_init(&next_thr_uid_mtx, "prof_next_thr_uid",
+		    WITNESS_RANK_PROF_NEXT_THR_UID, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (malloc_mutex_init(&prof_dump_seq_mtx, "prof_dump_seq",
+		    WITNESS_RANK_PROF_DUMP_SEQ, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+		if (malloc_mutex_init(&prof_dump_mtx, "prof_dump",
+		    WITNESS_RANK_PROF_DUMP, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (opt_prof_final && opt_prof_prefix[0] != '\0' &&
+		    atexit(prof_fdump) != 0) {
+			malloc_write("<jemalloc>: Error in atexit()\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+
+		if (opt_prof_log) {
+			prof_log_start(tsd_tsdn(tsd), NULL);
+		}
+
+		if (atexit(prof_log_stop_final) != 0) {
+			malloc_write("<jemalloc>: Error in atexit() "
+				     "for logging\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+
+		if (malloc_mutex_init(&log_mtx, "prof_log",
+		    WITNESS_RANK_PROF_LOG, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (ckh_new(tsd, &log_bt_node_set, PROF_CKH_MINITEMS,
+		    prof_bt_node_hash, prof_bt_node_keycomp)) {
+			return true;
+		}
+
+		if (ckh_new(tsd, &log_thr_node_set, PROF_CKH_MINITEMS,
+		    prof_thr_node_hash, prof_thr_node_keycomp)) {
+			return true;
+		}
+
+		log_tables_initialized = true;
+
+		gctx_locks = (malloc_mutex_t *)base_alloc(tsd_tsdn(tsd),
+		    b0get(), PROF_NCTX_LOCKS * sizeof(malloc_mutex_t),
+		    CACHELINE);
+		if (gctx_locks == NULL) {
+			return true;
+		}
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			if (malloc_mutex_init(&gctx_locks[i], "prof_gctx",
+			    WITNESS_RANK_PROF_GCTX,
+			    malloc_mutex_rank_exclusive)) {
+				return true;
+			}
+		}
+
+		tdata_locks = (malloc_mutex_t *)base_alloc(tsd_tsdn(tsd),
+		    b0get(), PROF_NTDATA_LOCKS * sizeof(malloc_mutex_t),
+		    CACHELINE);
+		if (tdata_locks == NULL) {
+			return true;
+		}
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			if (malloc_mutex_init(&tdata_locks[i], "prof_tdata",
+			    WITNESS_RANK_PROF_TDATA,
+			    malloc_mutex_rank_exclusive)) {
+				return true;
+			}
+		}
+#ifdef JEMALLOC_PROF_LIBGCC
+		/*
+		 * Cause the backtracing machinery to allocate its internal
+		 * state before enabling profiling.
+		 */
+		_Unwind_Backtrace(prof_unwind_init_callback, NULL);
+#endif
+	}
+	prof_booted = true;
+
+	return false;
+}
+
+void
+prof_prefork0(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		unsigned i;
+
+		malloc_mutex_prefork(tsdn, &prof_dump_mtx);
+		malloc_mutex_prefork(tsdn, &bt2gctx_mtx);
+		malloc_mutex_prefork(tsdn, &tdatas_mtx);
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			malloc_mutex_prefork(tsdn, &tdata_locks[i]);
+		}
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			malloc_mutex_prefork(tsdn, &gctx_locks[i]);
+		}
+	}
+}
+
+void
+prof_prefork1(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		malloc_mutex_prefork(tsdn, &prof_active_mtx);
+		malloc_mutex_prefork(tsdn, &prof_dump_seq_mtx);
+		malloc_mutex_prefork(tsdn, &prof_gdump_mtx);
+		malloc_mutex_prefork(tsdn, &next_thr_uid_mtx);
+		malloc_mutex_prefork(tsdn, &prof_thread_active_init_mtx);
+	}
+}
+
+void
+prof_postfork_parent(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		unsigned i;
+
+		malloc_mutex_postfork_parent(tsdn,
+		    &prof_thread_active_init_mtx);
+		malloc_mutex_postfork_parent(tsdn, &next_thr_uid_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_gdump_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_dump_seq_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_active_mtx);
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			malloc_mutex_postfork_parent(tsdn, &gctx_locks[i]);
+		}
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			malloc_mutex_postfork_parent(tsdn, &tdata_locks[i]);
+		}
+		malloc_mutex_postfork_parent(tsdn, &tdatas_mtx);
+		malloc_mutex_postfork_parent(tsdn, &bt2gctx_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_dump_mtx);
+	}
+}
+
+void
+prof_postfork_child(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		unsigned i;
+
+		malloc_mutex_postfork_child(tsdn, &prof_thread_active_init_mtx);
+		malloc_mutex_postfork_child(tsdn, &next_thr_uid_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_gdump_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_dump_seq_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_active_mtx);
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			malloc_mutex_postfork_child(tsdn, &gctx_locks[i]);
+		}
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			malloc_mutex_postfork_child(tsdn, &tdata_locks[i]);
+		}
+		malloc_mutex_postfork_child(tsdn, &tdatas_mtx);
+		malloc_mutex_postfork_child(tsdn, &bt2gctx_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_dump_mtx);
+	}
+}
+
+/******************************************************************************/
diff --git a/jemalloc/src/rtree.c b/jemalloc/src/rtree.c
new file mode 100644
index 0000000..4ae41fe
--- /dev/null
+++ b/jemalloc/src/rtree.c
@@ -0,0 +1,320 @@
+#define JEMALLOC_RTREE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/mutex.h"
+
+/*
+ * Only the most significant bits of keys passed to rtree_{read,write}() are
+ * used.
+ */
+bool
+rtree_new(rtree_t *rtree, bool zeroed) {
+#ifdef JEMALLOC_JET
+	if (!zeroed) {
+		memset(rtree, 0, sizeof(rtree_t)); /* Clear root. */
+	}
+#else
+	assert(zeroed);
+#endif
+
+	if (malloc_mutex_init(&rtree->init_lock, "rtree", WITNESS_RANK_RTREE,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	return false;
+}
+
+static rtree_node_elm_t *
+rtree_node_alloc_impl(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	return (rtree_node_elm_t *)base_alloc(tsdn, b0get(), nelms *
+	    sizeof(rtree_node_elm_t), CACHELINE);
+}
+rtree_node_alloc_t *JET_MUTABLE rtree_node_alloc = rtree_node_alloc_impl;
+
+static void
+rtree_node_dalloc_impl(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *node) {
+	/* Nodes are never deleted during normal operation. */
+	not_reached();
+}
+rtree_node_dalloc_t *JET_MUTABLE rtree_node_dalloc =
+    rtree_node_dalloc_impl;
+
+static rtree_leaf_elm_t *
+rtree_leaf_alloc_impl(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	return (rtree_leaf_elm_t *)base_alloc(tsdn, b0get(), nelms *
+	    sizeof(rtree_leaf_elm_t), CACHELINE);
+}
+rtree_leaf_alloc_t *JET_MUTABLE rtree_leaf_alloc = rtree_leaf_alloc_impl;
+
+static void
+rtree_leaf_dalloc_impl(tsdn_t *tsdn, rtree_t *rtree, rtree_leaf_elm_t *leaf) {
+	/* Leaves are never deleted during normal operation. */
+	not_reached();
+}
+rtree_leaf_dalloc_t *JET_MUTABLE rtree_leaf_dalloc =
+    rtree_leaf_dalloc_impl;
+
+#ifdef JEMALLOC_JET
+#  if RTREE_HEIGHT > 1
+static void
+rtree_delete_subtree(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *subtree,
+    unsigned level) {
+	size_t nchildren = ZU(1) << rtree_levels[level].bits;
+	if (level + 2 < RTREE_HEIGHT) {
+		for (size_t i = 0; i < nchildren; i++) {
+			rtree_node_elm_t *node =
+			    (rtree_node_elm_t *)atomic_load_p(&subtree[i].child,
+			    ATOMIC_RELAXED);
+			if (node != NULL) {
+				rtree_delete_subtree(tsdn, rtree, node, level +
+				    1);
+			}
+		}
+	} else {
+		for (size_t i = 0; i < nchildren; i++) {
+			rtree_leaf_elm_t *leaf =
+			    (rtree_leaf_elm_t *)atomic_load_p(&subtree[i].child,
+			    ATOMIC_RELAXED);
+			if (leaf != NULL) {
+				rtree_leaf_dalloc(tsdn, rtree, leaf);
+			}
+		}
+	}
+
+	if (subtree != rtree->root) {
+		rtree_node_dalloc(tsdn, rtree, subtree);
+	}
+}
+#  endif
+
+void
+rtree_delete(tsdn_t *tsdn, rtree_t *rtree) {
+#  if RTREE_HEIGHT > 1
+	rtree_delete_subtree(tsdn, rtree, rtree->root, 0);
+#  endif
+}
+#endif
+
+static rtree_node_elm_t *
+rtree_node_init(tsdn_t *tsdn, rtree_t *rtree, unsigned level,
+    atomic_p_t *elmp) {
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+	/*
+	 * If *elmp is non-null, then it was initialized with the init lock
+	 * held, so we can get by with 'relaxed' here.
+	 */
+	rtree_node_elm_t *node = atomic_load_p(elmp, ATOMIC_RELAXED);
+	if (node == NULL) {
+		node = rtree_node_alloc(tsdn, rtree, ZU(1) <<
+		    rtree_levels[level].bits);
+		if (node == NULL) {
+			malloc_mutex_unlock(tsdn, &rtree->init_lock);
+			return NULL;
+		}
+		/*
+		 * Even though we hold the lock, a later reader might not; we
+		 * need release semantics.
+		 */
+		atomic_store_p(elmp, node, ATOMIC_RELEASE);
+	}
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+
+	return node;
+}
+
+static rtree_leaf_elm_t *
+rtree_leaf_init(tsdn_t *tsdn, rtree_t *rtree, atomic_p_t *elmp) {
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+	/*
+	 * If *elmp is non-null, then it was initialized with the init lock
+	 * held, so we can get by with 'relaxed' here.
+	 */
+	rtree_leaf_elm_t *leaf = atomic_load_p(elmp, ATOMIC_RELAXED);
+	if (leaf == NULL) {
+		leaf = rtree_leaf_alloc(tsdn, rtree, ZU(1) <<
+		    rtree_levels[RTREE_HEIGHT-1].bits);
+		if (leaf == NULL) {
+			malloc_mutex_unlock(tsdn, &rtree->init_lock);
+			return NULL;
+		}
+		/*
+		 * Even though we hold the lock, a later reader might not; we
+		 * need release semantics.
+		 */
+		atomic_store_p(elmp, leaf, ATOMIC_RELEASE);
+	}
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+
+	return leaf;
+}
+
+static bool
+rtree_node_valid(rtree_node_elm_t *node) {
+	return ((uintptr_t)node != (uintptr_t)0);
+}
+
+static bool
+rtree_leaf_valid(rtree_leaf_elm_t *leaf) {
+	return ((uintptr_t)leaf != (uintptr_t)0);
+}
+
+static rtree_node_elm_t *
+rtree_child_node_tryread(rtree_node_elm_t *elm, bool dependent) {
+	rtree_node_elm_t *node;
+
+	if (dependent) {
+		node = (rtree_node_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_RELAXED);
+	} else {
+		node = (rtree_node_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_ACQUIRE);
+	}
+
+	assert(!dependent || node != NULL);
+	return node;
+}
+
+static rtree_node_elm_t *
+rtree_child_node_read(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *elm,
+    unsigned level, bool dependent) {
+	rtree_node_elm_t *node;
+
+	node = rtree_child_node_tryread(elm, dependent);
+	if (!dependent && unlikely(!rtree_node_valid(node))) {
+		node = rtree_node_init(tsdn, rtree, level + 1, &elm->child);
+	}
+	assert(!dependent || node != NULL);
+	return node;
+}
+
+static rtree_leaf_elm_t *
+rtree_child_leaf_tryread(rtree_node_elm_t *elm, bool dependent) {
+	rtree_leaf_elm_t *leaf;
+
+	if (dependent) {
+		leaf = (rtree_leaf_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_RELAXED);
+	} else {
+		leaf = (rtree_leaf_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_ACQUIRE);
+	}
+
+	assert(!dependent || leaf != NULL);
+	return leaf;
+}
+
+static rtree_leaf_elm_t *
+rtree_child_leaf_read(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *elm,
+    unsigned level, bool dependent) {
+	rtree_leaf_elm_t *leaf;
+
+	leaf = rtree_child_leaf_tryread(elm, dependent);
+	if (!dependent && unlikely(!rtree_leaf_valid(leaf))) {
+		leaf = rtree_leaf_init(tsdn, rtree, &elm->child);
+	}
+	assert(!dependent || leaf != NULL);
+	return leaf;
+}
+
+rtree_leaf_elm_t *
+rtree_leaf_elm_lookup_hard(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, bool init_missing) {
+	rtree_node_elm_t *node;
+	rtree_leaf_elm_t *leaf;
+#if RTREE_HEIGHT > 1
+	node = rtree->root;
+#else
+	leaf = rtree->root;
+#endif
+
+	if (config_debug) {
+		uintptr_t leafkey = rtree_leafkey(key);
+		for (unsigned i = 0; i < RTREE_CTX_NCACHE; i++) {
+			assert(rtree_ctx->cache[i].leafkey != leafkey);
+		}
+		for (unsigned i = 0; i < RTREE_CTX_NCACHE_L2; i++) {
+			assert(rtree_ctx->l2_cache[i].leafkey != leafkey);
+		}
+	}
+
+#define RTREE_GET_CHILD(level) {					\
+		assert(level < RTREE_HEIGHT-1);				\
+		if (level != 0 && !dependent &&				\
+		    unlikely(!rtree_node_valid(node))) {		\
+			return NULL;					\
+		}							\
+		uintptr_t subkey = rtree_subkey(key, level);		\
+		if (level + 2 < RTREE_HEIGHT) {				\
+			node = init_missing ?				\
+			    rtree_child_node_read(tsdn, rtree,		\
+			    &node[subkey], level, dependent) :		\
+			    rtree_child_node_tryread(&node[subkey],	\
+			    dependent);					\
+		} else {						\
+			leaf = init_missing ?				\
+			    rtree_child_leaf_read(tsdn, rtree,		\
+			    &node[subkey], level, dependent) :		\
+			    rtree_child_leaf_tryread(&node[subkey],	\
+			    dependent);					\
+		}							\
+	}
+	/*
+	 * Cache replacement upon hard lookup (i.e. L1 & L2 rtree cache miss):
+	 * (1) evict last entry in L2 cache; (2) move the collision slot from L1
+	 * cache down to L2; and 3) fill L1.
+	 */
+#define RTREE_GET_LEAF(level) {						\
+		assert(level == RTREE_HEIGHT-1);			\
+		if (!dependent && unlikely(!rtree_leaf_valid(leaf))) {	\
+			return NULL;					\
+		}							\
+		if (RTREE_CTX_NCACHE_L2 > 1) {				\
+			memmove(&rtree_ctx->l2_cache[1],		\
+			    &rtree_ctx->l2_cache[0],			\
+			    sizeof(rtree_ctx_cache_elm_t) *		\
+			    (RTREE_CTX_NCACHE_L2 - 1));			\
+		}							\
+		size_t slot = rtree_cache_direct_map(key);		\
+		rtree_ctx->l2_cache[0].leafkey =			\
+		    rtree_ctx->cache[slot].leafkey;			\
+		rtree_ctx->l2_cache[0].leaf =				\
+		    rtree_ctx->cache[slot].leaf;			\
+		uintptr_t leafkey = rtree_leafkey(key);			\
+		rtree_ctx->cache[slot].leafkey = leafkey;		\
+		rtree_ctx->cache[slot].leaf = leaf;			\
+		uintptr_t subkey = rtree_subkey(key, level);		\
+		return &leaf[subkey];					\
+	}
+	if (RTREE_HEIGHT > 1) {
+		RTREE_GET_CHILD(0)
+	}
+	if (RTREE_HEIGHT > 2) {
+		RTREE_GET_CHILD(1)
+	}
+	if (RTREE_HEIGHT > 3) {
+		for (unsigned i = 2; i < RTREE_HEIGHT-1; i++) {
+			RTREE_GET_CHILD(i)
+		}
+	}
+	RTREE_GET_LEAF(RTREE_HEIGHT-1)
+#undef RTREE_GET_CHILD
+#undef RTREE_GET_LEAF
+	not_reached();
+}
+
+void
+rtree_ctx_data_init(rtree_ctx_t *ctx) {
+	for (unsigned i = 0; i < RTREE_CTX_NCACHE; i++) {
+		rtree_ctx_cache_elm_t *cache = &ctx->cache[i];
+		cache->leafkey = RTREE_LEAFKEY_INVALID;
+		cache->leaf = NULL;
+	}
+	for (unsigned i = 0; i < RTREE_CTX_NCACHE_L2; i++) {
+		rtree_ctx_cache_elm_t *cache = &ctx->l2_cache[i];
+		cache->leafkey = RTREE_LEAFKEY_INVALID;
+		cache->leaf = NULL;
+	}
+}
diff --git a/jemalloc/src/safety_check.c b/jemalloc/src/safety_check.c
new file mode 100644
index 0000000..804155d
--- /dev/null
+++ b/jemalloc/src/safety_check.c
@@ -0,0 +1,24 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+static void (*safety_check_abort)(const char *message);
+
+void safety_check_set_abort(void (*abort_fn)(const char *)) {
+	safety_check_abort = abort_fn;
+}
+
+void safety_check_fail(const char *format, ...) {
+	char buf[MALLOC_PRINTF_BUFSIZE];
+
+	va_list ap;
+	va_start(ap, format);
+	malloc_vsnprintf(buf, MALLOC_PRINTF_BUFSIZE, format, ap);
+	va_end(ap);
+
+	if (safety_check_abort == NULL) {
+		malloc_write(buf);
+		abort();
+	} else {
+		safety_check_abort(buf);
+	}
+}
diff --git a/jemalloc/src/sc.c b/jemalloc/src/sc.c
new file mode 100644
index 0000000..6cfcc1c
--- /dev/null
+++ b/jemalloc/src/sc.c
@@ -0,0 +1,313 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/sc.h"
+
+/*
+ * This module computes the size classes used to satisfy allocations.  The logic
+ * here was ported more or less line-by-line from a shell script, and because of
+ * that is not the most idiomatic C.  Eventually we should fix this, but for now
+ * at least the damage is compartmentalized to this file.
+ */
+
+sc_data_t sc_data_global;
+
+static size_t
+reg_size_compute(int lg_base, int lg_delta, int ndelta) {
+	return (ZU(1) << lg_base) + (ZU(ndelta) << lg_delta);
+}
+
+/* Returns the number of pages in the slab. */
+static int
+slab_size(int lg_page, int lg_base, int lg_delta, int ndelta) {
+	size_t page = (ZU(1) << lg_page);
+	size_t reg_size = reg_size_compute(lg_base, lg_delta, ndelta);
+
+	size_t try_slab_size = page;
+	size_t try_nregs = try_slab_size / reg_size;
+	size_t perfect_slab_size = 0;
+	bool perfect = false;
+	/*
+	 * This loop continues until we find the least common multiple of the
+	 * page size and size class size.  Size classes are all of the form
+	 * base + ndelta * delta == (ndelta + base/ndelta) * delta, which is
+	 * (ndelta + ngroup) * delta.  The way we choose slabbing strategies
+	 * means that delta is at most the page size and ndelta < ngroup.  So
+	 * the loop executes for at most 2 * ngroup - 1 iterations, which is
+	 * also the bound on the number of pages in a slab chosen by default.
+	 * With the current default settings, this is at most 7.
+	 */
+	while (!perfect) {
+		perfect_slab_size = try_slab_size;
+		size_t perfect_nregs = try_nregs;
+		try_slab_size += page;
+		try_nregs = try_slab_size / reg_size;
+		if (perfect_slab_size == perfect_nregs * reg_size) {
+			perfect = true;
+		}
+	}
+	return (int)(perfect_slab_size / page);
+}
+
+static void
+size_class(
+    /* Output. */
+    sc_t *sc,
+    /* Configuration decisions. */
+    int lg_max_lookup, int lg_page, int lg_ngroup,
+    /* Inputs specific to the size class. */
+    int index, int lg_base, int lg_delta, int ndelta) {
+	sc->index = index;
+	sc->lg_base = lg_base;
+	sc->lg_delta = lg_delta;
+	sc->ndelta = ndelta;
+	sc->psz = (reg_size_compute(lg_base, lg_delta, ndelta)
+	    % (ZU(1) << lg_page) == 0);
+	size_t size = (ZU(1) << lg_base) + (ZU(ndelta) << lg_delta);
+	if (index == 0) {
+		assert(!sc->psz);
+	}
+	if (size < (ZU(1) << (lg_page + lg_ngroup))) {
+		sc->bin = true;
+		sc->pgs = slab_size(lg_page, lg_base, lg_delta, ndelta);
+	} else {
+		sc->bin = false;
+		sc->pgs = 0;
+	}
+	if (size <= (ZU(1) << lg_max_lookup)) {
+		sc->lg_delta_lookup = lg_delta;
+	} else {
+		sc->lg_delta_lookup = 0;
+	}
+}
+
+static void
+size_classes(
+    /* Output. */
+    sc_data_t *sc_data,
+    /* Determined by the system. */
+    size_t lg_ptr_size, int lg_quantum,
+    /* Configuration decisions. */
+    int lg_tiny_min, int lg_max_lookup, int lg_page, int lg_ngroup) {
+	int ptr_bits = (1 << lg_ptr_size) * 8;
+	int ngroup = (1 << lg_ngroup);
+	int ntiny = 0;
+	int nlbins = 0;
+	int lg_tiny_maxclass = (unsigned)-1;
+	int nbins = 0;
+	int npsizes = 0;
+
+	int index = 0;
+
+	int ndelta = 0;
+	int lg_base = lg_tiny_min;
+	int lg_delta = lg_base;
+
+	/* Outputs that we update as we go. */
+	size_t lookup_maxclass = 0;
+	size_t small_maxclass = 0;
+	int lg_large_minclass = 0;
+	size_t large_maxclass = 0;
+
+	/* Tiny size classes. */
+	while (lg_base < lg_quantum) {
+		sc_t *sc = &sc_data->sc[index];
+		size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+		    lg_base, lg_delta, ndelta);
+		if (sc->lg_delta_lookup != 0) {
+			nlbins = index + 1;
+		}
+		if (sc->psz) {
+			npsizes++;
+		}
+		if (sc->bin) {
+			nbins++;
+		}
+		ntiny++;
+		/* Final written value is correct. */
+		lg_tiny_maxclass = lg_base;
+		index++;
+		lg_delta = lg_base;
+		lg_base++;
+	}
+
+	/* First non-tiny (pseudo) group. */
+	if (ntiny != 0) {
+		sc_t *sc = &sc_data->sc[index];
+		/*
+		 * See the note in sc.h; the first non-tiny size class has an
+		 * unusual encoding.
+		 */
+		lg_base--;
+		ndelta = 1;
+		size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+		    lg_base, lg_delta, ndelta);
+		index++;
+		lg_base++;
+		lg_delta++;
+		if (sc->psz) {
+			npsizes++;
+		}
+		if (sc->bin) {
+			nbins++;
+		}
+	}
+	while (ndelta < ngroup) {
+		sc_t *sc = &sc_data->sc[index];
+		size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+		    lg_base, lg_delta, ndelta);
+		index++;
+		ndelta++;
+		if (sc->psz) {
+			npsizes++;
+		}
+		if (sc->bin) {
+			nbins++;
+		}
+	}
+
+	/* All remaining groups. */
+	lg_base = lg_base + lg_ngroup;
+	while (lg_base < ptr_bits - 1) {
+		ndelta = 1;
+		int ndelta_limit;
+		if (lg_base == ptr_bits - 2) {
+			ndelta_limit = ngroup - 1;
+		} else {
+			ndelta_limit = ngroup;
+		}
+		while (ndelta <= ndelta_limit) {
+			sc_t *sc = &sc_data->sc[index];
+			size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+			    lg_base, lg_delta, ndelta);
+			if (sc->lg_delta_lookup != 0) {
+				nlbins = index + 1;
+				/* Final written value is correct. */
+				lookup_maxclass = (ZU(1) << lg_base)
+				    + (ZU(ndelta) << lg_delta);
+			}
+			if (sc->psz) {
+				npsizes++;
+			}
+			if (sc->bin) {
+				nbins++;
+				/* Final written value is correct. */
+				small_maxclass = (ZU(1) << lg_base)
+				    + (ZU(ndelta) << lg_delta);
+				if (lg_ngroup > 0) {
+					lg_large_minclass = lg_base + 1;
+				} else {
+					lg_large_minclass = lg_base + 2;
+				}
+			}
+			large_maxclass = (ZU(1) << lg_base)
+			    + (ZU(ndelta) << lg_delta);
+			index++;
+			ndelta++;
+		}
+		lg_base++;
+		lg_delta++;
+	}
+	/* Additional outputs. */
+	int nsizes = index;
+	unsigned lg_ceil_nsizes = lg_ceil(nsizes);
+
+	/* Fill in the output data. */
+	sc_data->ntiny = ntiny;
+	sc_data->nlbins = nlbins;
+	sc_data->nbins = nbins;
+	sc_data->nsizes = nsizes;
+	sc_data->lg_ceil_nsizes = lg_ceil_nsizes;
+	sc_data->npsizes = npsizes;
+	sc_data->lg_tiny_maxclass = lg_tiny_maxclass;
+	sc_data->lookup_maxclass = lookup_maxclass;
+	sc_data->small_maxclass = small_maxclass;
+	sc_data->lg_large_minclass = lg_large_minclass;
+	sc_data->large_minclass = (ZU(1) << lg_large_minclass);
+	sc_data->large_maxclass = large_maxclass;
+
+	/*
+	 * We compute these values in two ways:
+	 *   - Incrementally, as above.
+	 *   - In macros, in sc.h.
+	 * The computation is easier when done incrementally, but putting it in
+	 * a constant makes it available to the fast paths without having to
+	 * touch the extra global cacheline.  We assert, however, that the two
+	 * computations are equivalent.
+	 */
+	assert(sc_data->npsizes == SC_NPSIZES);
+	assert(sc_data->lg_tiny_maxclass == SC_LG_TINY_MAXCLASS);
+	assert(sc_data->small_maxclass == SC_SMALL_MAXCLASS);
+	assert(sc_data->large_minclass == SC_LARGE_MINCLASS);
+	assert(sc_data->lg_large_minclass == SC_LG_LARGE_MINCLASS);
+	assert(sc_data->large_maxclass == SC_LARGE_MAXCLASS);
+
+	/*
+	 * In the allocation fastpath, we want to assume that we can
+	 * unconditionally subtract the requested allocation size from
+	 * a ssize_t, and detect passing through 0 correctly.  This
+	 * results in optimal generated code.  For this to work, the
+	 * maximum allocation size must be less than SSIZE_MAX.
+	 */
+	assert(SC_LARGE_MAXCLASS < SSIZE_MAX);
+}
+
+void
+sc_data_init(sc_data_t *sc_data) {
+	assert(!sc_data->initialized);
+
+	int lg_max_lookup = 12;
+
+	size_classes(sc_data, LG_SIZEOF_PTR, LG_QUANTUM, SC_LG_TINY_MIN,
+	    lg_max_lookup, LG_PAGE, 2);
+
+	sc_data->initialized = true;
+}
+
+static void
+sc_data_update_sc_slab_size(sc_t *sc, size_t reg_size, size_t pgs_guess) {
+	size_t min_pgs = reg_size / PAGE;
+	if (reg_size % PAGE != 0) {
+		min_pgs++;
+	}
+	/*
+	 * BITMAP_MAXBITS is actually determined by putting the smallest
+	 * possible size-class on one page, so this can never be 0.
+	 */
+	size_t max_pgs = BITMAP_MAXBITS * reg_size / PAGE;
+
+	assert(min_pgs <= max_pgs);
+	assert(min_pgs > 0);
+	assert(max_pgs >= 1);
+	if (pgs_guess < min_pgs) {
+		sc->pgs = (int)min_pgs;
+	} else if (pgs_guess > max_pgs) {
+		sc->pgs = (int)max_pgs;
+	} else {
+		sc->pgs = (int)pgs_guess;
+	}
+}
+
+void
+sc_data_update_slab_size(sc_data_t *data, size_t begin, size_t end, int pgs) {
+	assert(data->initialized);
+	for (int i = 0; i < data->nsizes; i++) {
+		sc_t *sc = &data->sc[i];
+		if (!sc->bin) {
+			break;
+		}
+		size_t reg_size = reg_size_compute(sc->lg_base, sc->lg_delta,
+		    sc->ndelta);
+		if (begin <= reg_size && reg_size <= end) {
+			sc_data_update_sc_slab_size(sc, reg_size, pgs);
+		}
+	}
+}
+
+void
+sc_boot(sc_data_t *data) {
+	sc_data_init(data);
+}
diff --git a/jemalloc/src/stats.c b/jemalloc/src/stats.c
new file mode 100644
index 0000000..118e05d
--- /dev/null
+++ b/jemalloc/src/stats.c
@@ -0,0 +1,1457 @@
+#define JEMALLOC_STATS_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/emitter.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_prof.h"
+
+const char *global_mutex_names[mutex_prof_num_global_mutexes] = {
+#define OP(mtx) #mtx,
+	MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+};
+
+const char *arena_mutex_names[mutex_prof_num_arena_mutexes] = {
+#define OP(mtx) #mtx,
+	MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+};
+
+#define CTL_GET(n, v, t) do {						\
+	size_t sz = sizeof(t);						\
+	xmallctl(n, (void *)v, &sz, NULL, 0);				\
+} while (0)
+
+#define CTL_M2_GET(n, i, v, t) do {					\
+	size_t mib[CTL_MAX_DEPTH];					\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = (i);							\
+	xmallctlbymib(mib, miblen, (void *)v, &sz, NULL, 0);		\
+} while (0)
+
+#define CTL_M2_M4_GET(n, i, j, v, t) do {				\
+	size_t mib[CTL_MAX_DEPTH];					\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = (i);							\
+	mib[4] = (j);							\
+	xmallctlbymib(mib, miblen, (void *)v, &sz, NULL, 0);		\
+} while (0)
+
+/******************************************************************************/
+/* Data. */
+
+bool opt_stats_print = false;
+char opt_stats_print_opts[stats_print_tot_num_options+1] = "";
+
+/******************************************************************************/
+
+static uint64_t
+rate_per_second(uint64_t value, uint64_t uptime_ns) {
+	uint64_t billion = 1000000000;
+	if (uptime_ns == 0 || value == 0) {
+		return 0;
+	}
+	if (uptime_ns < billion) {
+		return value;
+	} else {
+		uint64_t uptime_s = uptime_ns / billion;
+		return value / uptime_s;
+	}
+}
+
+/* Calculate x.yyy and output a string (takes a fixed sized char array). */
+static bool
+get_rate_str(uint64_t dividend, uint64_t divisor, char str[6]) {
+	if (divisor == 0 || dividend > divisor) {
+		/* The rate is not supposed to be greater than 1. */
+		return true;
+	}
+	if (dividend > 0) {
+		assert(UINT64_MAX / dividend >= 1000);
+	}
+
+	unsigned n = (unsigned)((dividend * 1000) / divisor);
+	if (n < 10) {
+		malloc_snprintf(str, 6, "0.00%u", n);
+	} else if (n < 100) {
+		malloc_snprintf(str, 6, "0.0%u", n);
+	} else if (n < 1000) {
+		malloc_snprintf(str, 6, "0.%u", n);
+	} else {
+		malloc_snprintf(str, 6, "1");
+	}
+
+	return false;
+}
+
+#define MUTEX_CTL_STR_MAX_LENGTH 128
+static void
+gen_mutex_ctl_str(char *str, size_t buf_len, const char *prefix,
+    const char *mutex, const char *counter) {
+	malloc_snprintf(str, buf_len, "stats.%s.%s.%s", prefix, mutex, counter);
+}
+
+static void
+mutex_stats_init_cols(emitter_row_t *row, const char *table_name,
+    emitter_col_t *name,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters]) {
+	mutex_prof_uint64_t_counter_ind_t k_uint64_t = 0;
+	mutex_prof_uint32_t_counter_ind_t k_uint32_t = 0;
+
+	emitter_col_t *col;
+
+	if (name != NULL) {
+		emitter_col_init(name, row);
+		name->justify = emitter_justify_left;
+		name->width = 21;
+		name->type = emitter_type_title;
+		name->str_val = table_name;
+	}
+
+#define WIDTH_uint32_t 12
+#define WIDTH_uint64_t 16
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	col = &col_##counter_type[k_##counter_type];			\
+	++k_##counter_type;						\
+	emitter_col_init(col, row);					\
+	col->justify = emitter_justify_right;				\
+	col->width = derived ? 8 : WIDTH_##counter_type;		\
+	col->type = emitter_type_title;					\
+	col->str_val = human;
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef WIDTH_uint32_t
+#undef WIDTH_uint64_t
+	col_uint64_t[mutex_counter_total_wait_time_ps].width = 10;
+}
+
+static void
+mutex_stats_read_global(const char *name, emitter_col_t *col_name,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters],
+    uint64_t uptime) {
+	char cmd[MUTEX_CTL_STR_MAX_LENGTH];
+
+	col_name->str_val = name;
+
+	emitter_col_t *dst;
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	dst = &col_##counter_type[mutex_counter_##counter];		\
+	dst->type = EMITTER_TYPE_##counter_type;			\
+	if (!derived) {							\
+		gen_mutex_ctl_str(cmd, MUTEX_CTL_STR_MAX_LENGTH,	\
+		    "mutexes", name, #counter);				\
+		CTL_GET(cmd, (counter_type *)&dst->bool_val, counter_type);	\
+	} else { \
+	    emitter_col_t *base = &col_##counter_type[mutex_counter_##base_counter];	\
+	    dst->counter_type##_val = rate_per_second(base->counter_type##_val, uptime); \
+	}
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+static void
+mutex_stats_read_arena(unsigned arena_ind, mutex_prof_arena_ind_t mutex_ind,
+    const char *name, emitter_col_t *col_name,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters],
+    uint64_t uptime) {
+	char cmd[MUTEX_CTL_STR_MAX_LENGTH];
+
+	col_name->str_val = name;
+
+	emitter_col_t *dst;
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	dst = &col_##counter_type[mutex_counter_##counter];		\
+	dst->type = EMITTER_TYPE_##counter_type;			\
+	if (!derived) {                                   \
+		gen_mutex_ctl_str(cmd, MUTEX_CTL_STR_MAX_LENGTH,        \
+		    "arenas.0.mutexes", arena_mutex_names[mutex_ind], #counter);\
+		CTL_M2_GET(cmd, arena_ind, (counter_type *)&dst->bool_val, counter_type); \
+	} else {                      \
+		emitter_col_t *base = &col_##counter_type[mutex_counter_##base_counter];	\
+		dst->counter_type##_val = rate_per_second(base->counter_type##_val, uptime); \
+	}
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+static void
+mutex_stats_read_arena_bin(unsigned arena_ind, unsigned bin_ind,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters],
+    uint64_t uptime) {
+	char cmd[MUTEX_CTL_STR_MAX_LENGTH];
+	emitter_col_t *dst;
+
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	dst = &col_##counter_type[mutex_counter_##counter];		\
+	dst->type = EMITTER_TYPE_##counter_type;			\
+	if (!derived) {                                   \
+		gen_mutex_ctl_str(cmd, MUTEX_CTL_STR_MAX_LENGTH,        \
+		    "arenas.0.bins.0","mutex", #counter);            \
+		CTL_M2_M4_GET(cmd, arena_ind, bin_ind,                \
+		    (counter_type *)&dst->bool_val, counter_type);  \
+	} else {                      \
+		emitter_col_t *base = &col_##counter_type[mutex_counter_##base_counter]; \
+		dst->counter_type##_val = rate_per_second(base->counter_type##_val, uptime); \
+	}
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+/* "row" can be NULL to avoid emitting in table mode. */
+static void
+mutex_stats_emit(emitter_t *emitter, emitter_row_t *row,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters]) {
+	if (row != NULL) {
+		emitter_table_row(emitter, row);
+	}
+
+	mutex_prof_uint64_t_counter_ind_t k_uint64_t = 0;
+	mutex_prof_uint32_t_counter_ind_t k_uint32_t = 0;
+
+	emitter_col_t *col;
+
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, type, human, derived, base_counter)		\
+	if (!derived) {                    \
+		col = &col_##type[k_##type];                        \
+		++k_##type;                            \
+		emitter_json_kv(emitter, #counter, EMITTER_TYPE_##type,        \
+		    (const void *)&col->bool_val); \
+	}
+	MUTEX_PROF_COUNTERS;
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+#define COL(row_name, column_name, left_or_right, col_width, etype)      \
+	emitter_col_t col_##column_name;                                     \
+	emitter_col_init(&col_##column_name, &row_name);                     \
+	col_##column_name.justify = emitter_justify_##left_or_right;         \
+	col_##column_name.width = col_width;                                 \
+	col_##column_name.type = emitter_type_##etype;
+
+#define COL_HDR(row_name, column_name, human, left_or_right, col_width, etype)  \
+	COL(row_name, column_name, left_or_right, col_width, etype)	         \
+	emitter_col_t header_##column_name;                                  \
+	emitter_col_init(&header_##column_name, &header_##row_name);         \
+	header_##column_name.justify = emitter_justify_##left_or_right;      \
+	header_##column_name.width = col_width;                              \
+	header_##column_name.type = emitter_type_title;                      \
+	header_##column_name.str_val = human ? human : #column_name;
+
+
+static void
+stats_arena_bins_print(emitter_t *emitter, bool mutex, unsigned i, uint64_t uptime) {
+	size_t page;
+	bool in_gap, in_gap_prev;
+	unsigned nbins, j;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+
+	emitter_row_t header_row;
+	emitter_row_init(&header_row);
+
+	emitter_row_t row;
+	emitter_row_init(&row);
+
+	COL_HDR(row, size, NULL, right, 20, size)
+	COL_HDR(row, ind, NULL, right, 4, unsigned)
+	COL_HDR(row, allocated, NULL, right, 13, uint64)
+	COL_HDR(row, nmalloc, NULL, right, 13, uint64)
+	COL_HDR(row, nmalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, ndalloc, NULL, right, 13, uint64)
+	COL_HDR(row, ndalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nrequests, NULL, right, 13, uint64)
+	COL_HDR(row, nrequests_ps, "(#/sec)", right, 10, uint64)
+	COL_HDR(row, nshards, NULL, right, 9, unsigned)
+	COL_HDR(row, curregs, NULL, right, 13, size)
+	COL_HDR(row, curslabs, NULL, right, 13, size)
+	COL_HDR(row, nonfull_slabs, NULL, right, 15, size)
+	COL_HDR(row, regs, NULL, right, 5, unsigned)
+	COL_HDR(row, pgs, NULL, right, 4, size)
+	/* To buffer a right- and left-justified column. */
+	COL_HDR(row, justify_spacer, NULL, right, 1, title)
+	COL_HDR(row, util, NULL, right, 6, title)
+	COL_HDR(row, nfills, NULL, right, 13, uint64)
+	COL_HDR(row, nfills_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nflushes, NULL, right, 13, uint64)
+	COL_HDR(row, nflushes_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nslabs, NULL, right, 13, uint64)
+	COL_HDR(row, nreslabs, NULL, right, 13, uint64)
+	COL_HDR(row, nreslabs_ps, "(#/sec)", right, 8, uint64)
+
+	/* Don't want to actually print the name. */
+	header_justify_spacer.str_val = " ";
+	col_justify_spacer.str_val = " ";
+
+	emitter_col_t col_mutex64[mutex_prof_num_uint64_t_counters];
+	emitter_col_t col_mutex32[mutex_prof_num_uint32_t_counters];
+
+	emitter_col_t header_mutex64[mutex_prof_num_uint64_t_counters];
+	emitter_col_t header_mutex32[mutex_prof_num_uint32_t_counters];
+
+	if (mutex) {
+		mutex_stats_init_cols(&row, NULL, NULL, col_mutex64,
+		    col_mutex32);
+		mutex_stats_init_cols(&header_row, NULL, NULL, header_mutex64,
+		    header_mutex32);
+	}
+
+	/*
+	 * We print a "bins:" header as part of the table row; we need to adjust
+	 * the header size column to compensate.
+	 */
+	header_size.width -=5;
+	emitter_table_printf(emitter, "bins:");
+	emitter_table_row(emitter, &header_row);
+	emitter_json_array_kv_begin(emitter, "bins");
+
+	for (j = 0, in_gap = false; j < nbins; j++) {
+		uint64_t nslabs;
+		size_t reg_size, slab_size, curregs;
+		size_t curslabs;
+		size_t nonfull_slabs;
+		uint32_t nregs, nshards;
+		uint64_t nmalloc, ndalloc, nrequests, nfills, nflushes;
+		uint64_t nreslabs;
+
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nslabs", i, j, &nslabs,
+		    uint64_t);
+		in_gap_prev = in_gap;
+		in_gap = (nslabs == 0);
+
+		if (in_gap_prev && !in_gap) {
+			emitter_table_printf(emitter,
+			    "                     ---\n");
+		}
+
+		CTL_M2_GET("arenas.bin.0.size", j, &reg_size, size_t);
+		CTL_M2_GET("arenas.bin.0.nregs", j, &nregs, uint32_t);
+		CTL_M2_GET("arenas.bin.0.slab_size", j, &slab_size, size_t);
+		CTL_M2_GET("arenas.bin.0.nshards", j, &nshards, uint32_t);
+
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nmalloc", i, j, &nmalloc,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.ndalloc", i, j, &ndalloc,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.curregs", i, j, &curregs,
+		    size_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nrequests", i, j,
+		    &nrequests, uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nfills", i, j, &nfills,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nflushes", i, j, &nflushes,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nreslabs", i, j, &nreslabs,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.curslabs", i, j, &curslabs,
+		    size_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nonfull_slabs", i, j, &nonfull_slabs,
+		    size_t);
+
+		if (mutex) {
+			mutex_stats_read_arena_bin(i, j, col_mutex64,
+			    col_mutex32, uptime);
+		}
+
+		emitter_json_object_begin(emitter);
+		emitter_json_kv(emitter, "nmalloc", emitter_type_uint64,
+		    &nmalloc);
+		emitter_json_kv(emitter, "ndalloc", emitter_type_uint64,
+		    &ndalloc);
+		emitter_json_kv(emitter, "curregs", emitter_type_size,
+		    &curregs);
+		emitter_json_kv(emitter, "nrequests", emitter_type_uint64,
+		    &nrequests);
+		emitter_json_kv(emitter, "nfills", emitter_type_uint64,
+		    &nfills);
+		emitter_json_kv(emitter, "nflushes", emitter_type_uint64,
+		    &nflushes);
+		emitter_json_kv(emitter, "nreslabs", emitter_type_uint64,
+		    &nreslabs);
+		emitter_json_kv(emitter, "curslabs", emitter_type_size,
+		    &curslabs);
+		emitter_json_kv(emitter, "nonfull_slabs", emitter_type_size,
+		    &nonfull_slabs);
+		if (mutex) {
+			emitter_json_object_kv_begin(emitter, "mutex");
+			mutex_stats_emit(emitter, NULL, col_mutex64,
+			    col_mutex32);
+			emitter_json_object_end(emitter);
+		}
+		emitter_json_object_end(emitter);
+
+		size_t availregs = nregs * curslabs;
+		char util[6];
+		if (get_rate_str((uint64_t)curregs, (uint64_t)availregs, util))
+		{
+			if (availregs == 0) {
+				malloc_snprintf(util, sizeof(util), "1");
+			} else if (curregs > availregs) {
+				/*
+				 * Race detected: the counters were read in
+				 * separate mallctl calls and concurrent
+				 * operations happened in between.  In this case
+				 * no meaningful utilization can be computed.
+				 */
+				malloc_snprintf(util, sizeof(util), " race");
+			} else {
+				not_reached();
+			}
+		}
+
+		col_size.size_val = reg_size;
+		col_ind.unsigned_val = j;
+		col_allocated.size_val = curregs * reg_size;
+		col_nmalloc.uint64_val = nmalloc;
+		col_nmalloc_ps.uint64_val = rate_per_second(nmalloc, uptime);
+		col_ndalloc.uint64_val = ndalloc;
+		col_ndalloc_ps.uint64_val = rate_per_second(ndalloc, uptime);
+		col_nrequests.uint64_val = nrequests;
+		col_nrequests_ps.uint64_val = rate_per_second(nrequests, uptime);
+		col_nshards.unsigned_val = nshards;
+		col_curregs.size_val = curregs;
+		col_curslabs.size_val = curslabs;
+		col_nonfull_slabs.size_val = nonfull_slabs;
+		col_regs.unsigned_val = nregs;
+		col_pgs.size_val = slab_size / page;
+		col_util.str_val = util;
+		col_nfills.uint64_val = nfills;
+		col_nfills_ps.uint64_val = rate_per_second(nfills, uptime);
+		col_nflushes.uint64_val = nflushes;
+		col_nflushes_ps.uint64_val = rate_per_second(nflushes, uptime);
+		col_nslabs.uint64_val = nslabs;
+		col_nreslabs.uint64_val = nreslabs;
+		col_nreslabs_ps.uint64_val = rate_per_second(nreslabs, uptime);
+
+		/*
+		 * Note that mutex columns were initialized above, if mutex ==
+		 * true.
+		 */
+
+		emitter_table_row(emitter, &row);
+	}
+	emitter_json_array_end(emitter); /* Close "bins". */
+
+	if (in_gap) {
+		emitter_table_printf(emitter, "                     ---\n");
+	}
+}
+
+static void
+stats_arena_lextents_print(emitter_t *emitter, unsigned i, uint64_t uptime) {
+	unsigned nbins, nlextents, j;
+	bool in_gap, in_gap_prev;
+
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+	CTL_GET("arenas.nlextents", &nlextents, unsigned);
+
+	emitter_row_t header_row;
+	emitter_row_init(&header_row);
+	emitter_row_t row;
+	emitter_row_init(&row);
+
+	COL_HDR(row, size, NULL, right, 20, size)
+	COL_HDR(row, ind, NULL, right, 4, unsigned)
+	COL_HDR(row, allocated, NULL, right, 13, size)
+	COL_HDR(row, nmalloc, NULL, right, 13, uint64)
+	COL_HDR(row, nmalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, ndalloc, NULL, right, 13, uint64)
+	COL_HDR(row, ndalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nrequests, NULL, right, 13, uint64)
+	COL_HDR(row, nrequests_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, curlextents, NULL, right, 13, size)
+
+	/* As with bins, we label the large extents table. */
+	header_size.width -= 6;
+	emitter_table_printf(emitter, "large:");
+	emitter_table_row(emitter, &header_row);
+	emitter_json_array_kv_begin(emitter, "lextents");
+
+	for (j = 0, in_gap = false; j < nlextents; j++) {
+		uint64_t nmalloc, ndalloc, nrequests;
+		size_t lextent_size, curlextents;
+
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.nmalloc", i, j,
+		    &nmalloc, uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.ndalloc", i, j,
+		    &ndalloc, uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.nrequests", i, j,
+		    &nrequests, uint64_t);
+		in_gap_prev = in_gap;
+		in_gap = (nrequests == 0);
+
+		if (in_gap_prev && !in_gap) {
+			emitter_table_printf(emitter,
+			    "                     ---\n");
+		}
+
+		CTL_M2_GET("arenas.lextent.0.size", j, &lextent_size, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.curlextents", i, j,
+		    &curlextents, size_t);
+
+		emitter_json_object_begin(emitter);
+		emitter_json_kv(emitter, "curlextents", emitter_type_size,
+		    &curlextents);
+		emitter_json_object_end(emitter);
+
+		col_size.size_val = lextent_size;
+		col_ind.unsigned_val = nbins + j;
+		col_allocated.size_val = curlextents * lextent_size;
+		col_nmalloc.uint64_val = nmalloc;
+		col_nmalloc_ps.uint64_val = rate_per_second(nmalloc, uptime);
+		col_ndalloc.uint64_val = ndalloc;
+		col_ndalloc_ps.uint64_val = rate_per_second(ndalloc, uptime);
+		col_nrequests.uint64_val = nrequests;
+		col_nrequests_ps.uint64_val = rate_per_second(nrequests, uptime);
+		col_curlextents.size_val = curlextents;
+
+		if (!in_gap) {
+			emitter_table_row(emitter, &row);
+		}
+	}
+	emitter_json_array_end(emitter); /* Close "lextents". */
+	if (in_gap) {
+		emitter_table_printf(emitter, "                     ---\n");
+	}
+}
+
+static void
+stats_arena_extents_print(emitter_t *emitter, unsigned i) {
+	unsigned j;
+	bool in_gap, in_gap_prev;
+	emitter_row_t header_row;
+	emitter_row_init(&header_row);
+	emitter_row_t row;
+	emitter_row_init(&row);
+
+	COL_HDR(row, size, NULL, right, 20, size)
+	COL_HDR(row, ind, NULL, right, 4, unsigned)
+	COL_HDR(row, ndirty, NULL, right, 13, size)
+	COL_HDR(row, dirty, NULL, right, 13, size)
+	COL_HDR(row, nmuzzy, NULL, right, 13, size)
+	COL_HDR(row, muzzy, NULL, right, 13, size)
+	COL_HDR(row, nretained, NULL, right, 13, size)
+	COL_HDR(row, retained, NULL, right, 13, size)
+	COL_HDR(row, ntotal, NULL, right, 13, size)
+	COL_HDR(row, total, NULL, right, 13, size)
+
+	/* Label this section. */
+	header_size.width -= 8;
+	emitter_table_printf(emitter, "extents:");
+	emitter_table_row(emitter, &header_row);
+	emitter_json_array_kv_begin(emitter, "extents");
+
+	in_gap = false;
+	for (j = 0; j < SC_NPSIZES; j++) {
+		size_t ndirty, nmuzzy, nretained, total, dirty_bytes,
+		    muzzy_bytes, retained_bytes, total_bytes;
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.ndirty", i, j,
+		    &ndirty, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.nmuzzy", i, j,
+		    &nmuzzy, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.nretained", i, j,
+		    &nretained, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.dirty_bytes", i, j,
+		    &dirty_bytes, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.muzzy_bytes", i, j,
+		    &muzzy_bytes, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.retained_bytes", i, j,
+		    &retained_bytes, size_t);
+		total = ndirty + nmuzzy + nretained;
+		total_bytes = dirty_bytes + muzzy_bytes + retained_bytes;
+
+		in_gap_prev = in_gap;
+		in_gap = (total == 0);
+
+		if (in_gap_prev && !in_gap) {
+			emitter_table_printf(emitter,
+			    "                     ---\n");
+		}
+
+		emitter_json_object_begin(emitter);
+		emitter_json_kv(emitter, "ndirty", emitter_type_size, &ndirty);
+		emitter_json_kv(emitter, "nmuzzy", emitter_type_size, &nmuzzy);
+		emitter_json_kv(emitter, "nretained", emitter_type_size,
+		    &nretained);
+
+		emitter_json_kv(emitter, "dirty_bytes", emitter_type_size,
+		    &dirty_bytes);
+		emitter_json_kv(emitter, "muzzy_bytes", emitter_type_size,
+		    &muzzy_bytes);
+		emitter_json_kv(emitter, "retained_bytes", emitter_type_size,
+		    &retained_bytes);
+		emitter_json_object_end(emitter);
+
+		col_size.size_val = sz_pind2sz(j);
+		col_ind.size_val = j;
+		col_ndirty.size_val = ndirty;
+		col_dirty.size_val = dirty_bytes;
+		col_nmuzzy.size_val = nmuzzy;
+		col_muzzy.size_val = muzzy_bytes;
+		col_nretained.size_val = nretained;
+		col_retained.size_val = retained_bytes;
+		col_ntotal.size_val = total;
+		col_total.size_val = total_bytes;
+
+		if (!in_gap) {
+			emitter_table_row(emitter, &row);
+		}
+	}
+	emitter_json_array_end(emitter); /* Close "extents". */
+	if (in_gap) {
+		emitter_table_printf(emitter, "                     ---\n");
+	}
+}
+
+static void
+stats_arena_mutexes_print(emitter_t *emitter, unsigned arena_ind, uint64_t uptime) {
+	emitter_row_t row;
+	emitter_col_t col_name;
+	emitter_col_t col64[mutex_prof_num_uint64_t_counters];
+	emitter_col_t col32[mutex_prof_num_uint32_t_counters];
+
+	emitter_row_init(&row);
+	mutex_stats_init_cols(&row, "", &col_name, col64, col32);
+
+	emitter_json_object_kv_begin(emitter, "mutexes");
+	emitter_table_row(emitter, &row);
+
+	for (mutex_prof_arena_ind_t i = 0; i < mutex_prof_num_arena_mutexes;
+	    i++) {
+		const char *name = arena_mutex_names[i];
+		emitter_json_object_kv_begin(emitter, name);
+		mutex_stats_read_arena(arena_ind, i, name, &col_name, col64,
+		    col32, uptime);
+		mutex_stats_emit(emitter, &row, col64, col32);
+		emitter_json_object_end(emitter); /* Close the mutex dict. */
+	}
+	emitter_json_object_end(emitter); /* End "mutexes". */
+}
+
+static void
+stats_arena_print(emitter_t *emitter, unsigned i, bool bins, bool large,
+    bool mutex, bool extents) {
+	unsigned nthreads;
+	const char *dss;
+	ssize_t dirty_decay_ms, muzzy_decay_ms;
+	size_t page, pactive, pdirty, pmuzzy, mapped, retained;
+	size_t base, internal, resident, metadata_thp, extent_avail;
+	uint64_t dirty_npurge, dirty_nmadvise, dirty_purged;
+	uint64_t muzzy_npurge, muzzy_nmadvise, muzzy_purged;
+	size_t small_allocated;
+	uint64_t small_nmalloc, small_ndalloc, small_nrequests, small_nfills,
+	    small_nflushes;
+	size_t large_allocated;
+	uint64_t large_nmalloc, large_ndalloc, large_nrequests, large_nfills,
+	    large_nflushes;
+	size_t tcache_bytes, abandoned_vm;
+	uint64_t uptime;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_M2_GET("stats.arenas.0.nthreads", i, &nthreads, unsigned);
+	emitter_kv(emitter, "nthreads", "assigned threads",
+	    emitter_type_unsigned, &nthreads);
+
+	CTL_M2_GET("stats.arenas.0.uptime", i, &uptime, uint64_t);
+	emitter_kv(emitter, "uptime_ns", "uptime", emitter_type_uint64,
+	    &uptime);
+
+	CTL_M2_GET("stats.arenas.0.dss", i, &dss, const char *);
+	emitter_kv(emitter, "dss", "dss allocation precedence",
+	    emitter_type_string, &dss);
+
+	CTL_M2_GET("stats.arenas.0.dirty_decay_ms", i, &dirty_decay_ms,
+	    ssize_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_decay_ms", i, &muzzy_decay_ms,
+	    ssize_t);
+	CTL_M2_GET("stats.arenas.0.pactive", i, &pactive, size_t);
+	CTL_M2_GET("stats.arenas.0.pdirty", i, &pdirty, size_t);
+	CTL_M2_GET("stats.arenas.0.pmuzzy", i, &pmuzzy, size_t);
+	CTL_M2_GET("stats.arenas.0.dirty_npurge", i, &dirty_npurge, uint64_t);
+	CTL_M2_GET("stats.arenas.0.dirty_nmadvise", i, &dirty_nmadvise,
+	    uint64_t);
+	CTL_M2_GET("stats.arenas.0.dirty_purged", i, &dirty_purged, uint64_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_npurge", i, &muzzy_npurge, uint64_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_nmadvise", i, &muzzy_nmadvise,
+	    uint64_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_purged", i, &muzzy_purged, uint64_t);
+
+	emitter_row_t decay_row;
+	emitter_row_init(&decay_row);
+
+	/* JSON-style emission. */
+	emitter_json_kv(emitter, "dirty_decay_ms", emitter_type_ssize,
+	    &dirty_decay_ms);
+	emitter_json_kv(emitter, "muzzy_decay_ms", emitter_type_ssize,
+	    &muzzy_decay_ms);
+
+	emitter_json_kv(emitter, "pactive", emitter_type_size, &pactive);
+	emitter_json_kv(emitter, "pdirty", emitter_type_size, &pdirty);
+	emitter_json_kv(emitter, "pmuzzy", emitter_type_size, &pmuzzy);
+
+	emitter_json_kv(emitter, "dirty_npurge", emitter_type_uint64,
+	    &dirty_npurge);
+	emitter_json_kv(emitter, "dirty_nmadvise", emitter_type_uint64,
+	    &dirty_nmadvise);
+	emitter_json_kv(emitter, "dirty_purged", emitter_type_uint64,
+	    &dirty_purged);
+
+	emitter_json_kv(emitter, "muzzy_npurge", emitter_type_uint64,
+	    &muzzy_npurge);
+	emitter_json_kv(emitter, "muzzy_nmadvise", emitter_type_uint64,
+	    &muzzy_nmadvise);
+	emitter_json_kv(emitter, "muzzy_purged", emitter_type_uint64,
+	    &muzzy_purged);
+
+	/* Table-style emission. */
+	COL(decay_row, decay_type, right, 9, title);
+	col_decay_type.str_val = "decaying:";
+
+	COL(decay_row, decay_time, right, 6, title);
+	col_decay_time.str_val = "time";
+
+	COL(decay_row, decay_npages, right, 13, title);
+	col_decay_npages.str_val = "npages";
+
+	COL(decay_row, decay_sweeps, right, 13, title);
+	col_decay_sweeps.str_val = "sweeps";
+
+	COL(decay_row, decay_madvises, right, 13, title);
+	col_decay_madvises.str_val = "madvises";
+
+	COL(decay_row, decay_purged, right, 13, title);
+	col_decay_purged.str_val = "purged";
+
+	/* Title row. */
+	emitter_table_row(emitter, &decay_row);
+
+	/* Dirty row. */
+	col_decay_type.str_val = "dirty:";
+
+	if (dirty_decay_ms >= 0) {
+		col_decay_time.type = emitter_type_ssize;
+		col_decay_time.ssize_val = dirty_decay_ms;
+	} else {
+		col_decay_time.type = emitter_type_title;
+		col_decay_time.str_val = "N/A";
+	}
+
+	col_decay_npages.type = emitter_type_size;
+	col_decay_npages.size_val = pdirty;
+
+	col_decay_sweeps.type = emitter_type_uint64;
+	col_decay_sweeps.uint64_val = dirty_npurge;
+
+	col_decay_madvises.type = emitter_type_uint64;
+	col_decay_madvises.uint64_val = dirty_nmadvise;
+
+	col_decay_purged.type = emitter_type_uint64;
+	col_decay_purged.uint64_val = dirty_purged;
+
+	emitter_table_row(emitter, &decay_row);
+
+	/* Muzzy row. */
+	col_decay_type.str_val = "muzzy:";
+
+	if (muzzy_decay_ms >= 0) {
+		col_decay_time.type = emitter_type_ssize;
+		col_decay_time.ssize_val = muzzy_decay_ms;
+	} else {
+		col_decay_time.type = emitter_type_title;
+		col_decay_time.str_val = "N/A";
+	}
+
+	col_decay_npages.type = emitter_type_size;
+	col_decay_npages.size_val = pmuzzy;
+
+	col_decay_sweeps.type = emitter_type_uint64;
+	col_decay_sweeps.uint64_val = muzzy_npurge;
+
+	col_decay_madvises.type = emitter_type_uint64;
+	col_decay_madvises.uint64_val = muzzy_nmadvise;
+
+	col_decay_purged.type = emitter_type_uint64;
+	col_decay_purged.uint64_val = muzzy_purged;
+
+	emitter_table_row(emitter, &decay_row);
+
+	/* Small / large / total allocation counts. */
+	emitter_row_t alloc_count_row;
+	emitter_row_init(&alloc_count_row);
+
+	COL(alloc_count_row, count_title, left, 21, title);
+	col_count_title.str_val = "";
+
+	COL(alloc_count_row, count_allocated, right, 16, title);
+	col_count_allocated.str_val = "allocated";
+
+	COL(alloc_count_row, count_nmalloc, right, 16, title);
+	col_count_nmalloc.str_val = "nmalloc";
+	COL(alloc_count_row, count_nmalloc_ps, right, 8, title);
+	col_count_nmalloc_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_ndalloc, right, 16, title);
+	col_count_ndalloc.str_val = "ndalloc";
+	COL(alloc_count_row, count_ndalloc_ps, right, 8, title);
+	col_count_ndalloc_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_nrequests, right, 16, title);
+	col_count_nrequests.str_val = "nrequests";
+	COL(alloc_count_row, count_nrequests_ps, right, 10, title);
+	col_count_nrequests_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_nfills, right, 16, title);
+	col_count_nfills.str_val = "nfill";
+	COL(alloc_count_row, count_nfills_ps, right, 10, title);
+	col_count_nfills_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_nflushes, right, 16, title);
+	col_count_nflushes.str_val = "nflush";
+	COL(alloc_count_row, count_nflushes_ps, right, 10, title);
+	col_count_nflushes_ps.str_val = "(#/sec)";
+
+	emitter_table_row(emitter, &alloc_count_row);
+
+	col_count_nmalloc_ps.type = emitter_type_uint64;
+	col_count_ndalloc_ps.type = emitter_type_uint64;
+	col_count_nrequests_ps.type = emitter_type_uint64;
+	col_count_nfills_ps.type = emitter_type_uint64;
+	col_count_nflushes_ps.type = emitter_type_uint64;
+
+#define GET_AND_EMIT_ALLOC_STAT(small_or_large, name, valtype)		\
+	CTL_M2_GET("stats.arenas.0." #small_or_large "." #name, i,	\
+	    &small_or_large##_##name, valtype##_t);			\
+	emitter_json_kv(emitter, #name, emitter_type_##valtype,		\
+	    &small_or_large##_##name);					\
+	col_count_##name.type = emitter_type_##valtype;		\
+	col_count_##name.valtype##_val = small_or_large##_##name;
+
+	emitter_json_object_kv_begin(emitter, "small");
+	col_count_title.str_val = "small:";
+
+	GET_AND_EMIT_ALLOC_STAT(small, allocated, size)
+	GET_AND_EMIT_ALLOC_STAT(small, nmalloc, uint64)
+	col_count_nmalloc_ps.uint64_val =
+	    rate_per_second(col_count_nmalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, ndalloc, uint64)
+	col_count_ndalloc_ps.uint64_val =
+	    rate_per_second(col_count_ndalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, nrequests, uint64)
+	col_count_nrequests_ps.uint64_val =
+	    rate_per_second(col_count_nrequests.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, nfills, uint64)
+	col_count_nfills_ps.uint64_val =
+	    rate_per_second(col_count_nfills.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, nflushes, uint64)
+	col_count_nflushes_ps.uint64_val =
+	    rate_per_second(col_count_nflushes.uint64_val, uptime);
+
+	emitter_table_row(emitter, &alloc_count_row);
+	emitter_json_object_end(emitter); /* Close "small". */
+
+	emitter_json_object_kv_begin(emitter, "large");
+	col_count_title.str_val = "large:";
+
+	GET_AND_EMIT_ALLOC_STAT(large, allocated, size)
+	GET_AND_EMIT_ALLOC_STAT(large, nmalloc, uint64)
+	col_count_nmalloc_ps.uint64_val =
+	    rate_per_second(col_count_nmalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, ndalloc, uint64)
+	col_count_ndalloc_ps.uint64_val =
+	    rate_per_second(col_count_ndalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, nrequests, uint64)
+	col_count_nrequests_ps.uint64_val =
+	    rate_per_second(col_count_nrequests.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, nfills, uint64)
+	col_count_nfills_ps.uint64_val =
+	    rate_per_second(col_count_nfills.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, nflushes, uint64)
+	col_count_nflushes_ps.uint64_val =
+	    rate_per_second(col_count_nflushes.uint64_val, uptime);
+
+	emitter_table_row(emitter, &alloc_count_row);
+	emitter_json_object_end(emitter); /* Close "large". */
+
+#undef GET_AND_EMIT_ALLOC_STAT
+
+	/* Aggregated small + large stats are emitter only in table mode. */
+	col_count_title.str_val = "total:";
+	col_count_allocated.size_val = small_allocated + large_allocated;
+	col_count_nmalloc.uint64_val = small_nmalloc + large_nmalloc;
+	col_count_ndalloc.uint64_val = small_ndalloc + large_ndalloc;
+	col_count_nrequests.uint64_val = small_nrequests + large_nrequests;
+	col_count_nfills.uint64_val = small_nfills + large_nfills;
+	col_count_nflushes.uint64_val = small_nflushes + large_nflushes;
+	col_count_nmalloc_ps.uint64_val =
+	    rate_per_second(col_count_nmalloc.uint64_val, uptime);
+	col_count_ndalloc_ps.uint64_val =
+	    rate_per_second(col_count_ndalloc.uint64_val, uptime);
+	col_count_nrequests_ps.uint64_val =
+	    rate_per_second(col_count_nrequests.uint64_val, uptime);
+	col_count_nfills_ps.uint64_val =
+	    rate_per_second(col_count_nfills.uint64_val, uptime);
+	col_count_nflushes_ps.uint64_val =
+	    rate_per_second(col_count_nflushes.uint64_val, uptime);
+	emitter_table_row(emitter, &alloc_count_row);
+
+	emitter_row_t mem_count_row;
+	emitter_row_init(&mem_count_row);
+
+	emitter_col_t mem_count_title;
+	emitter_col_init(&mem_count_title, &mem_count_row);
+	mem_count_title.justify = emitter_justify_left;
+	mem_count_title.width = 21;
+	mem_count_title.type = emitter_type_title;
+	mem_count_title.str_val = "";
+
+	emitter_col_t mem_count_val;
+	emitter_col_init(&mem_count_val, &mem_count_row);
+	mem_count_val.justify = emitter_justify_right;
+	mem_count_val.width = 16;
+	mem_count_val.type = emitter_type_title;
+	mem_count_val.str_val = "";
+
+	emitter_table_row(emitter, &mem_count_row);
+	mem_count_val.type = emitter_type_size;
+
+	/* Active count in bytes is emitted only in table mode. */
+	mem_count_title.str_val = "active:";
+	mem_count_val.size_val = pactive * page;
+	emitter_table_row(emitter, &mem_count_row);
+
+#define GET_AND_EMIT_MEM_STAT(stat)					\
+	CTL_M2_GET("stats.arenas.0."#stat, i, &stat, size_t);		\
+	emitter_json_kv(emitter, #stat, emitter_type_size, &stat);	\
+	mem_count_title.str_val = #stat":";				\
+	mem_count_val.size_val = stat;					\
+	emitter_table_row(emitter, &mem_count_row);
+
+	GET_AND_EMIT_MEM_STAT(mapped)
+	GET_AND_EMIT_MEM_STAT(retained)
+	GET_AND_EMIT_MEM_STAT(base)
+	GET_AND_EMIT_MEM_STAT(internal)
+	GET_AND_EMIT_MEM_STAT(metadata_thp)
+	GET_AND_EMIT_MEM_STAT(tcache_bytes)
+	GET_AND_EMIT_MEM_STAT(resident)
+	GET_AND_EMIT_MEM_STAT(abandoned_vm)
+	GET_AND_EMIT_MEM_STAT(extent_avail)
+#undef GET_AND_EMIT_MEM_STAT
+
+	if (mutex) {
+		stats_arena_mutexes_print(emitter, i, uptime);
+	}
+	if (bins) {
+		stats_arena_bins_print(emitter, mutex, i, uptime);
+	}
+	if (large) {
+		stats_arena_lextents_print(emitter, i, uptime);
+	}
+	if (extents) {
+		stats_arena_extents_print(emitter, i);
+	}
+}
+
+static void
+stats_general_print(emitter_t *emitter) {
+	const char *cpv;
+	bool bv, bv2;
+	unsigned uv;
+	uint32_t u32v;
+	uint64_t u64v;
+	ssize_t ssv, ssv2;
+	size_t sv, bsz, usz, ssz, sssz, cpsz;
+
+	bsz = sizeof(bool);
+	usz = sizeof(unsigned);
+	ssz = sizeof(size_t);
+	sssz = sizeof(ssize_t);
+	cpsz = sizeof(const char *);
+
+	CTL_GET("version", &cpv, const char *);
+	emitter_kv(emitter, "version", "Version", emitter_type_string, &cpv);
+
+	/* config. */
+	emitter_dict_begin(emitter, "config", "Build-time option settings");
+#define CONFIG_WRITE_BOOL(name)						\
+	do {								\
+		CTL_GET("config."#name, &bv, bool);			\
+		emitter_kv(emitter, #name, "config."#name,		\
+		    emitter_type_bool, &bv);				\
+	} while (0)
+
+	CONFIG_WRITE_BOOL(cache_oblivious);
+	CONFIG_WRITE_BOOL(debug);
+	CONFIG_WRITE_BOOL(fill);
+	CONFIG_WRITE_BOOL(lazy_lock);
+	emitter_kv(emitter, "malloc_conf", "config.malloc_conf",
+	    emitter_type_string, &config_malloc_conf);
+
+	CONFIG_WRITE_BOOL(opt_safety_checks);
+	CONFIG_WRITE_BOOL(prof);
+	CONFIG_WRITE_BOOL(prof_libgcc);
+	CONFIG_WRITE_BOOL(prof_libunwind);
+	CONFIG_WRITE_BOOL(stats);
+	CONFIG_WRITE_BOOL(utrace);
+	CONFIG_WRITE_BOOL(xmalloc);
+#undef CONFIG_WRITE_BOOL
+	emitter_dict_end(emitter); /* Close "config" dict. */
+
+	/* opt. */
+#define OPT_WRITE(name, var, size, emitter_type)			\
+	if (je_mallctl("opt."name, (void *)&var, &size, NULL, 0) ==	\
+	    0) {							\
+		emitter_kv(emitter, name, "opt."name, emitter_type,	\
+		    &var);						\
+	}
+
+#define OPT_WRITE_MUTABLE(name, var1, var2, size, emitter_type,		\
+    altname)								\
+	if (je_mallctl("opt."name, (void *)&var1, &size, NULL, 0) ==	\
+	    0 && je_mallctl(altname, (void *)&var2, &size, NULL, 0)	\
+	    == 0) {							\
+		emitter_kv_note(emitter, name, "opt."name,		\
+		    emitter_type, &var1, altname, emitter_type,		\
+		    &var2);						\
+	}
+
+#define OPT_WRITE_BOOL(name) OPT_WRITE(name, bv, bsz, emitter_type_bool)
+#define OPT_WRITE_BOOL_MUTABLE(name, altname)				\
+	OPT_WRITE_MUTABLE(name, bv, bv2, bsz, emitter_type_bool, altname)
+
+#define OPT_WRITE_UNSIGNED(name)					\
+	OPT_WRITE(name, uv, usz, emitter_type_unsigned)
+
+#define OPT_WRITE_SIZE_T(name)						\
+	OPT_WRITE(name, sv, ssz, emitter_type_size)
+#define OPT_WRITE_SSIZE_T(name)						\
+	OPT_WRITE(name, ssv, sssz, emitter_type_ssize)
+#define OPT_WRITE_SSIZE_T_MUTABLE(name, altname)			\
+	OPT_WRITE_MUTABLE(name, ssv, ssv2, sssz, emitter_type_ssize,	\
+	    altname)
+
+#define OPT_WRITE_CHAR_P(name)						\
+	OPT_WRITE(name, cpv, cpsz, emitter_type_string)
+
+	emitter_dict_begin(emitter, "opt", "Run-time option settings");
+
+	OPT_WRITE_BOOL("abort")
+	OPT_WRITE_BOOL("abort_conf")
+	OPT_WRITE_BOOL("confirm_conf")
+	OPT_WRITE_BOOL("retain")
+	OPT_WRITE_CHAR_P("dss")
+	OPT_WRITE_UNSIGNED("narenas")
+	OPT_WRITE_CHAR_P("percpu_arena")
+	OPT_WRITE_SIZE_T("oversize_threshold")
+	OPT_WRITE_CHAR_P("metadata_thp")
+	OPT_WRITE_BOOL_MUTABLE("background_thread", "background_thread")
+	OPT_WRITE_SSIZE_T_MUTABLE("dirty_decay_ms", "arenas.dirty_decay_ms")
+	OPT_WRITE_SSIZE_T_MUTABLE("muzzy_decay_ms", "arenas.muzzy_decay_ms")
+	OPT_WRITE_SIZE_T("lg_extent_max_active_fit")
+	OPT_WRITE_CHAR_P("junk")
+	OPT_WRITE_BOOL("zero")
+	OPT_WRITE_BOOL("utrace")
+	OPT_WRITE_BOOL("xmalloc")
+	OPT_WRITE_BOOL("tcache")
+	OPT_WRITE_SSIZE_T("lg_tcache_max")
+	OPT_WRITE_CHAR_P("thp")
+	OPT_WRITE_BOOL("prof")
+	OPT_WRITE_CHAR_P("prof_prefix")
+	OPT_WRITE_BOOL_MUTABLE("prof_active", "prof.active")
+	OPT_WRITE_BOOL_MUTABLE("prof_thread_active_init",
+	    "prof.thread_active_init")
+	OPT_WRITE_SSIZE_T_MUTABLE("lg_prof_sample", "prof.lg_sample")
+	OPT_WRITE_BOOL("prof_accum")
+	OPT_WRITE_SSIZE_T("lg_prof_interval")
+	OPT_WRITE_BOOL("prof_gdump")
+	OPT_WRITE_BOOL("prof_final")
+	OPT_WRITE_BOOL("prof_leak")
+	OPT_WRITE_BOOL("stats_print")
+	OPT_WRITE_CHAR_P("stats_print_opts")
+
+	emitter_dict_end(emitter);
+
+#undef OPT_WRITE
+#undef OPT_WRITE_MUTABLE
+#undef OPT_WRITE_BOOL
+#undef OPT_WRITE_BOOL_MUTABLE
+#undef OPT_WRITE_UNSIGNED
+#undef OPT_WRITE_SSIZE_T
+#undef OPT_WRITE_SSIZE_T_MUTABLE
+#undef OPT_WRITE_CHAR_P
+
+	/* prof. */
+	if (config_prof) {
+		emitter_dict_begin(emitter, "prof", "Profiling settings");
+
+		CTL_GET("prof.thread_active_init", &bv, bool);
+		emitter_kv(emitter, "thread_active_init",
+		    "prof.thread_active_init", emitter_type_bool, &bv);
+
+		CTL_GET("prof.active", &bv, bool);
+		emitter_kv(emitter, "active", "prof.active", emitter_type_bool,
+		    &bv);
+
+		CTL_GET("prof.gdump", &bv, bool);
+		emitter_kv(emitter, "gdump", "prof.gdump", emitter_type_bool,
+		    &bv);
+
+		CTL_GET("prof.interval", &u64v, uint64_t);
+		emitter_kv(emitter, "interval", "prof.interval",
+		    emitter_type_uint64, &u64v);
+
+		CTL_GET("prof.lg_sample", &ssv, ssize_t);
+		emitter_kv(emitter, "lg_sample", "prof.lg_sample",
+		    emitter_type_ssize, &ssv);
+
+		emitter_dict_end(emitter); /* Close "prof". */
+	}
+
+	/* arenas. */
+	/*
+	 * The json output sticks arena info into an "arenas" dict; the table
+	 * output puts them at the top-level.
+	 */
+	emitter_json_object_kv_begin(emitter, "arenas");
+
+	CTL_GET("arenas.narenas", &uv, unsigned);
+	emitter_kv(emitter, "narenas", "Arenas", emitter_type_unsigned, &uv);
+
+	/*
+	 * Decay settings are emitted only in json mode; in table mode, they're
+	 * emitted as notes with the opt output, above.
+	 */
+	CTL_GET("arenas.dirty_decay_ms", &ssv, ssize_t);
+	emitter_json_kv(emitter, "dirty_decay_ms", emitter_type_ssize, &ssv);
+
+	CTL_GET("arenas.muzzy_decay_ms", &ssv, ssize_t);
+	emitter_json_kv(emitter, "muzzy_decay_ms", emitter_type_ssize, &ssv);
+
+	CTL_GET("arenas.quantum", &sv, size_t);
+	emitter_kv(emitter, "quantum", "Quantum size", emitter_type_size, &sv);
+
+	CTL_GET("arenas.page", &sv, size_t);
+	emitter_kv(emitter, "page", "Page size", emitter_type_size, &sv);
+
+	if (je_mallctl("arenas.tcache_max", (void *)&sv, &ssz, NULL, 0) == 0) {
+		emitter_kv(emitter, "tcache_max",
+		    "Maximum thread-cached size class", emitter_type_size, &sv);
+	}
+
+	unsigned nbins;
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+	emitter_kv(emitter, "nbins", "Number of bin size classes",
+	    emitter_type_unsigned, &nbins);
+
+	unsigned nhbins;
+	CTL_GET("arenas.nhbins", &nhbins, unsigned);
+	emitter_kv(emitter, "nhbins", "Number of thread-cache bin size classes",
+	    emitter_type_unsigned, &nhbins);
+
+	/*
+	 * We do enough mallctls in a loop that we actually want to omit them
+	 * (not just omit the printing).
+	 */
+	if (emitter->output == emitter_output_json) {
+		emitter_json_array_kv_begin(emitter, "bin");
+		for (unsigned i = 0; i < nbins; i++) {
+			emitter_json_object_begin(emitter);
+
+			CTL_M2_GET("arenas.bin.0.size", i, &sv, size_t);
+			emitter_json_kv(emitter, "size", emitter_type_size,
+			    &sv);
+
+			CTL_M2_GET("arenas.bin.0.nregs", i, &u32v, uint32_t);
+			emitter_json_kv(emitter, "nregs", emitter_type_uint32,
+			    &u32v);
+
+			CTL_M2_GET("arenas.bin.0.slab_size", i, &sv, size_t);
+			emitter_json_kv(emitter, "slab_size", emitter_type_size,
+			    &sv);
+
+			CTL_M2_GET("arenas.bin.0.nshards", i, &u32v, uint32_t);
+			emitter_json_kv(emitter, "nshards", emitter_type_uint32,
+			    &u32v);
+
+			emitter_json_object_end(emitter);
+		}
+		emitter_json_array_end(emitter); /* Close "bin". */
+	}
+
+	unsigned nlextents;
+	CTL_GET("arenas.nlextents", &nlextents, unsigned);
+	emitter_kv(emitter, "nlextents", "Number of large size classes",
+	    emitter_type_unsigned, &nlextents);
+
+	if (emitter->output == emitter_output_json) {
+		emitter_json_array_kv_begin(emitter, "lextent");
+		for (unsigned i = 0; i < nlextents; i++) {
+			emitter_json_object_begin(emitter);
+
+			CTL_M2_GET("arenas.lextent.0.size", i, &sv, size_t);
+			emitter_json_kv(emitter, "size", emitter_type_size,
+			    &sv);
+
+			emitter_json_object_end(emitter);
+		}
+		emitter_json_array_end(emitter); /* Close "lextent". */
+	}
+
+	emitter_json_object_end(emitter); /* Close "arenas" */
+}
+
+static void
+stats_print_helper(emitter_t *emitter, bool merged, bool destroyed,
+    bool unmerged, bool bins, bool large, bool mutex, bool extents) {
+	/*
+	 * These should be deleted.  We keep them around for a while, to aid in
+	 * the transition to the emitter code.
+	 */
+	size_t allocated, active, metadata, metadata_thp, resident, mapped,
+	    retained;
+	size_t num_background_threads;
+	uint64_t background_thread_num_runs, background_thread_run_interval;
+
+	CTL_GET("stats.allocated", &allocated, size_t);
+	CTL_GET("stats.active", &active, size_t);
+	CTL_GET("stats.metadata", &metadata, size_t);
+	CTL_GET("stats.metadata_thp", &metadata_thp, size_t);
+	CTL_GET("stats.resident", &resident, size_t);
+	CTL_GET("stats.mapped", &mapped, size_t);
+	CTL_GET("stats.retained", &retained, size_t);
+
+	if (have_background_thread) {
+		CTL_GET("stats.background_thread.num_threads",
+		    &num_background_threads, size_t);
+		CTL_GET("stats.background_thread.num_runs",
+		    &background_thread_num_runs, uint64_t);
+		CTL_GET("stats.background_thread.run_interval",
+		    &background_thread_run_interval, uint64_t);
+	} else {
+		num_background_threads = 0;
+		background_thread_num_runs = 0;
+		background_thread_run_interval = 0;
+	}
+
+	/* Generic global stats. */
+	emitter_json_object_kv_begin(emitter, "stats");
+	emitter_json_kv(emitter, "allocated", emitter_type_size, &allocated);
+	emitter_json_kv(emitter, "active", emitter_type_size, &active);
+	emitter_json_kv(emitter, "metadata", emitter_type_size, &metadata);
+	emitter_json_kv(emitter, "metadata_thp", emitter_type_size,
+	    &metadata_thp);
+	emitter_json_kv(emitter, "resident", emitter_type_size, &resident);
+	emitter_json_kv(emitter, "mapped", emitter_type_size, &mapped);
+	emitter_json_kv(emitter, "retained", emitter_type_size, &retained);
+
+	emitter_table_printf(emitter, "Allocated: %zu, active: %zu, "
+	    "metadata: %zu (n_thp %zu), resident: %zu, mapped: %zu, "
+	    "retained: %zu\n", allocated, active, metadata, metadata_thp,
+	    resident, mapped, retained);
+
+	/* Background thread stats. */
+	emitter_json_object_kv_begin(emitter, "background_thread");
+	emitter_json_kv(emitter, "num_threads", emitter_type_size,
+	    &num_background_threads);
+	emitter_json_kv(emitter, "num_runs", emitter_type_uint64,
+	    &background_thread_num_runs);
+	emitter_json_kv(emitter, "run_interval", emitter_type_uint64,
+	    &background_thread_run_interval);
+	emitter_json_object_end(emitter); /* Close "background_thread". */
+
+	emitter_table_printf(emitter, "Background threads: %zu, "
+	    "num_runs: %"FMTu64", run_interval: %"FMTu64" ns\n",
+	    num_background_threads, background_thread_num_runs,
+	    background_thread_run_interval);
+
+	if (mutex) {
+		emitter_row_t row;
+		emitter_col_t name;
+		emitter_col_t col64[mutex_prof_num_uint64_t_counters];
+		emitter_col_t col32[mutex_prof_num_uint32_t_counters];
+		uint64_t uptime;
+
+		emitter_row_init(&row);
+		mutex_stats_init_cols(&row, "", &name, col64, col32);
+
+		emitter_table_row(emitter, &row);
+		emitter_json_object_kv_begin(emitter, "mutexes");
+
+		CTL_M2_GET("stats.arenas.0.uptime", 0, &uptime, uint64_t);
+
+		for (int i = 0; i < mutex_prof_num_global_mutexes; i++) {
+			mutex_stats_read_global(global_mutex_names[i], &name,
+			    col64, col32, uptime);
+			emitter_json_object_kv_begin(emitter, global_mutex_names[i]);
+			mutex_stats_emit(emitter, &row, col64, col32);
+			emitter_json_object_end(emitter);
+		}
+
+		emitter_json_object_end(emitter); /* Close "mutexes". */
+	}
+
+	emitter_json_object_end(emitter); /* Close "stats". */
+
+	if (merged || destroyed || unmerged) {
+		unsigned narenas;
+
+		emitter_json_object_kv_begin(emitter, "stats.arenas");
+
+		CTL_GET("arenas.narenas", &narenas, unsigned);
+		size_t mib[3];
+		size_t miblen = sizeof(mib) / sizeof(size_t);
+		size_t sz;
+		VARIABLE_ARRAY(bool, initialized, narenas);
+		bool destroyed_initialized;
+		unsigned i, j, ninitialized;
+
+		xmallctlnametomib("arena.0.initialized", mib, &miblen);
+		for (i = ninitialized = 0; i < narenas; i++) {
+			mib[1] = i;
+			sz = sizeof(bool);
+			xmallctlbymib(mib, miblen, &initialized[i], &sz,
+			    NULL, 0);
+			if (initialized[i]) {
+				ninitialized++;
+			}
+		}
+		mib[1] = MALLCTL_ARENAS_DESTROYED;
+		sz = sizeof(bool);
+		xmallctlbymib(mib, miblen, &destroyed_initialized, &sz,
+		    NULL, 0);
+
+		/* Merged stats. */
+		if (merged && (ninitialized > 1 || !unmerged)) {
+			/* Print merged arena stats. */
+			emitter_table_printf(emitter, "Merged arenas stats:\n");
+			emitter_json_object_kv_begin(emitter, "merged");
+			stats_arena_print(emitter, MALLCTL_ARENAS_ALL, bins,
+			    large, mutex, extents);
+			emitter_json_object_end(emitter); /* Close "merged". */
+		}
+
+		/* Destroyed stats. */
+		if (destroyed_initialized && destroyed) {
+			/* Print destroyed arena stats. */
+			emitter_table_printf(emitter,
+			    "Destroyed arenas stats:\n");
+			emitter_json_object_kv_begin(emitter, "destroyed");
+			stats_arena_print(emitter, MALLCTL_ARENAS_DESTROYED,
+			    bins, large, mutex, extents);
+			emitter_json_object_end(emitter); /* Close "destroyed". */
+		}
+
+		/* Unmerged stats. */
+		if (unmerged) {
+			for (i = j = 0; i < narenas; i++) {
+				if (initialized[i]) {
+					char arena_ind_str[20];
+					malloc_snprintf(arena_ind_str,
+					    sizeof(arena_ind_str), "%u", i);
+					emitter_json_object_kv_begin(emitter,
+					    arena_ind_str);
+					emitter_table_printf(emitter,
+					    "arenas[%s]:\n", arena_ind_str);
+					stats_arena_print(emitter, i, bins,
+					    large, mutex, extents);
+					/* Close "<arena-ind>". */
+					emitter_json_object_end(emitter);
+				}
+			}
+		}
+		emitter_json_object_end(emitter); /* Close "stats.arenas". */
+	}
+}
+
+void
+stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts) {
+	int err;
+	uint64_t epoch;
+	size_t u64sz;
+#define OPTION(o, v, d, s) bool v = d;
+	STATS_PRINT_OPTIONS
+#undef OPTION
+
+	/*
+	 * Refresh stats, in case mallctl() was called by the application.
+	 *
+	 * Check for OOM here, since refreshing the ctl cache can trigger
+	 * allocation.  In practice, none of the subsequent mallctl()-related
+	 * calls in this function will cause OOM if this one succeeds.
+	 * */
+	epoch = 1;
+	u64sz = sizeof(uint64_t);
+	err = je_mallctl("epoch", (void *)&epoch, &u64sz, (void *)&epoch,
+	    sizeof(uint64_t));
+	if (err != 0) {
+		if (err == EAGAIN) {
+			malloc_write("<jemalloc>: Memory allocation failure in "
+			    "mallctl(\"epoch\", ...)\n");
+			return;
+		}
+		malloc_write("<jemalloc>: Failure in mallctl(\"epoch\", "
+		    "...)\n");
+		abort();
+	}
+
+	if (opts != NULL) {
+		for (unsigned i = 0; opts[i] != '\0'; i++) {
+			switch (opts[i]) {
+#define OPTION(o, v, d, s) case o: v = s; break;
+				STATS_PRINT_OPTIONS
+#undef OPTION
+			default:;
+			}
+		}
+	}
+
+	emitter_t emitter;
+	emitter_init(&emitter,
+	    json ? emitter_output_json : emitter_output_table, write_cb,
+	    cbopaque);
+	emitter_begin(&emitter);
+	emitter_table_printf(&emitter, "___ Begin jemalloc statistics ___\n");
+	emitter_json_object_kv_begin(&emitter, "jemalloc");
+
+	if (general) {
+		stats_general_print(&emitter);
+	}
+	if (config_stats) {
+		stats_print_helper(&emitter, merged, destroyed, unmerged,
+		    bins, large, mutex, extents);
+	}
+
+	emitter_json_object_end(&emitter); /* Closes the "jemalloc" dict. */
+	emitter_table_printf(&emitter, "--- End jemalloc statistics ---\n");
+	emitter_end(&emitter);
+}
diff --git a/jemalloc/src/sz.c b/jemalloc/src/sz.c
new file mode 100644
index 0000000..8633fb0
--- /dev/null
+++ b/jemalloc/src/sz.c
@@ -0,0 +1,64 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/sz.h"
+
+JEMALLOC_ALIGNED(CACHELINE)
+size_t sz_pind2sz_tab[SC_NPSIZES+1];
+
+static void
+sz_boot_pind2sz_tab(const sc_data_t *sc_data) {
+	int pind = 0;
+	for (unsigned i = 0; i < SC_NSIZES; i++) {
+		const sc_t *sc = &sc_data->sc[i];
+		if (sc->psz) {
+			sz_pind2sz_tab[pind] = (ZU(1) << sc->lg_base)
+			    + (ZU(sc->ndelta) << sc->lg_delta);
+			pind++;
+		}
+	}
+	for (int i = pind; i <= (int)SC_NPSIZES; i++) {
+		sz_pind2sz_tab[pind] = sc_data->large_maxclass + PAGE;
+	}
+}
+
+JEMALLOC_ALIGNED(CACHELINE)
+size_t sz_index2size_tab[SC_NSIZES];
+
+static void
+sz_boot_index2size_tab(const sc_data_t *sc_data) {
+	for (unsigned i = 0; i < SC_NSIZES; i++) {
+		const sc_t *sc = &sc_data->sc[i];
+		sz_index2size_tab[i] = (ZU(1) << sc->lg_base)
+		    + (ZU(sc->ndelta) << (sc->lg_delta));
+	}
+}
+
+/*
+ * To keep this table small, we divide sizes by the tiny min size, which gives
+ * the smallest interval for which the result can change.
+ */
+JEMALLOC_ALIGNED(CACHELINE)
+uint8_t sz_size2index_tab[(SC_LOOKUP_MAXCLASS >> SC_LG_TINY_MIN) + 1];
+
+static void
+sz_boot_size2index_tab(const sc_data_t *sc_data) {
+	size_t dst_max = (SC_LOOKUP_MAXCLASS >> SC_LG_TINY_MIN) + 1;
+	size_t dst_ind = 0;
+	for (unsigned sc_ind = 0; sc_ind < SC_NSIZES && dst_ind < dst_max;
+	    sc_ind++) {
+		const sc_t *sc = &sc_data->sc[sc_ind];
+		size_t sz = (ZU(1) << sc->lg_base)
+		    + (ZU(sc->ndelta) << sc->lg_delta);
+		size_t max_ind = ((sz + (ZU(1) << SC_LG_TINY_MIN) - 1)
+				   >> SC_LG_TINY_MIN);
+		for (; dst_ind <= max_ind && dst_ind < dst_max; dst_ind++) {
+			sz_size2index_tab[dst_ind] = sc_ind;
+		}
+	}
+}
+
+void
+sz_boot(const sc_data_t *sc_data) {
+	sz_boot_pind2sz_tab(sc_data);
+	sz_boot_index2size_tab(sc_data);
+	sz_boot_size2index_tab(sc_data);
+}
diff --git a/jemalloc/src/tcache.c b/jemalloc/src/tcache.c
new file mode 100644
index 0000000..50099a9
--- /dev/null
+++ b/jemalloc/src/tcache.c
@@ -0,0 +1,798 @@
+#define JEMALLOC_TCACHE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/sc.h"
+
+/******************************************************************************/
+/* Data. */
+
+bool	opt_tcache = true;
+ssize_t	opt_lg_tcache_max = LG_TCACHE_MAXCLASS_DEFAULT;
+
+cache_bin_info_t	*tcache_bin_info;
+static unsigned		stack_nelms; /* Total stack elms per tcache. */
+
+unsigned		nhbins;
+size_t			tcache_maxclass;
+
+tcaches_t		*tcaches;
+
+/* Index of first element within tcaches that has never been used. */
+static unsigned		tcaches_past;
+
+/* Head of singly linked list tracking available tcaches elements. */
+static tcaches_t	*tcaches_avail;
+
+/* Protects tcaches{,_past,_avail}. */
+static malloc_mutex_t	tcaches_mtx;
+
+/******************************************************************************/
+
+size_t
+tcache_salloc(tsdn_t *tsdn, const void *ptr) {
+	return arena_salloc(tsdn, ptr);
+}
+
+void
+tcache_event_hard(tsd_t *tsd, tcache_t *tcache) {
+	szind_t binind = tcache->next_gc_bin;
+
+	cache_bin_t *tbin;
+	if (binind < SC_NBINS) {
+		tbin = tcache_small_bin_get(tcache, binind);
+	} else {
+		tbin = tcache_large_bin_get(tcache, binind);
+	}
+	if (tbin->low_water > 0) {
+		/*
+		 * Flush (ceiling) 3/4 of the objects below the low water mark.
+		 */
+		if (binind < SC_NBINS) {
+			tcache_bin_flush_small(tsd, tcache, tbin, binind,
+			    tbin->ncached - tbin->low_water + (tbin->low_water
+			    >> 2));
+			/*
+			 * Reduce fill count by 2X.  Limit lg_fill_div such that
+			 * the fill count is always at least 1.
+			 */
+			cache_bin_info_t *tbin_info = &tcache_bin_info[binind];
+			if ((tbin_info->ncached_max >>
+			     (tcache->lg_fill_div[binind] + 1)) >= 1) {
+				tcache->lg_fill_div[binind]++;
+			}
+		} else {
+			tcache_bin_flush_large(tsd, tbin, binind, tbin->ncached
+			    - tbin->low_water + (tbin->low_water >> 2), tcache);
+		}
+	} else if (tbin->low_water < 0) {
+		/*
+		 * Increase fill count by 2X for small bins.  Make sure
+		 * lg_fill_div stays greater than 0.
+		 */
+		if (binind < SC_NBINS && tcache->lg_fill_div[binind] > 1) {
+			tcache->lg_fill_div[binind]--;
+		}
+	}
+	tbin->low_water = tbin->ncached;
+
+	tcache->next_gc_bin++;
+	if (tcache->next_gc_bin == nhbins) {
+		tcache->next_gc_bin = 0;
+	}
+}
+
+void *
+tcache_alloc_small_hard(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, bool *tcache_success) {
+	void *ret;
+
+	assert(tcache->arena != NULL);
+	arena_tcache_fill_small(tsdn, arena, tcache, tbin, binind,
+	    config_prof ? tcache->prof_accumbytes : 0);
+	if (config_prof) {
+		tcache->prof_accumbytes = 0;
+	}
+	ret = cache_bin_alloc_easy(tbin, tcache_success);
+
+	return ret;
+}
+
+/* Enabled with --enable-extra-size-check. */
+static void
+tbin_extents_lookup_size_check(tsdn_t *tsdn, cache_bin_t *tbin, szind_t binind,
+    size_t nflush, extent_t **extents){
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	/*
+	 * Verify that the items in the tcache all have the correct size; this
+	 * is useful for catching sized deallocation bugs, also to fail early
+	 * instead of corrupting metadata.  Since this can be turned on for opt
+	 * builds, avoid the branch in the loop.
+	 */
+	szind_t szind;
+	size_t sz_sum = binind * nflush;
+	for (unsigned i = 0 ; i < nflush; i++) {
+		rtree_extent_szind_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)*(tbin->avail - 1 - i), true,
+		    &extents[i], &szind);
+		sz_sum -= szind;
+	}
+	if (sz_sum != 0) {
+		safety_check_fail("<jemalloc>: size mismatch in thread cache "
+		    "detected, likely caused by sized deallocation bugs by "
+		    "application. Abort.\n");
+		abort();
+	}
+}
+
+void
+tcache_bin_flush_small(tsd_t *tsd, tcache_t *tcache, cache_bin_t *tbin,
+    szind_t binind, unsigned rem) {
+	bool merged_stats = false;
+
+	assert(binind < SC_NBINS);
+	assert((cache_bin_sz_t)rem <= tbin->ncached);
+
+	arena_t *arena = tcache->arena;
+	assert(arena != NULL);
+	unsigned nflush = tbin->ncached - rem;
+	VARIABLE_ARRAY(extent_t *, item_extent, nflush);
+
+	/* Look up extent once per item. */
+	if (config_opt_safety_checks) {
+		tbin_extents_lookup_size_check(tsd_tsdn(tsd), tbin, binind,
+		    nflush, item_extent);
+	} else {
+		for (unsigned i = 0 ; i < nflush; i++) {
+			item_extent[i] = iealloc(tsd_tsdn(tsd),
+			    *(tbin->avail - 1 - i));
+		}
+	}
+	while (nflush > 0) {
+		/* Lock the arena bin associated with the first object. */
+		extent_t *extent = item_extent[0];
+		unsigned bin_arena_ind = extent_arena_ind_get(extent);
+		arena_t *bin_arena = arena_get(tsd_tsdn(tsd), bin_arena_ind,
+		    false);
+		unsigned binshard = extent_binshard_get(extent);
+		assert(binshard < bin_infos[binind].n_shards);
+		bin_t *bin = &bin_arena->bins[binind].bin_shards[binshard];
+
+		if (config_prof && bin_arena == arena) {
+			if (arena_prof_accum(tsd_tsdn(tsd), arena,
+			    tcache->prof_accumbytes)) {
+				prof_idump(tsd_tsdn(tsd));
+			}
+			tcache->prof_accumbytes = 0;
+		}
+
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+		if (config_stats && bin_arena == arena && !merged_stats) {
+			merged_stats = true;
+			bin->stats.nflushes++;
+			bin->stats.nrequests += tbin->tstats.nrequests;
+			tbin->tstats.nrequests = 0;
+		}
+		unsigned ndeferred = 0;
+		for (unsigned i = 0; i < nflush; i++) {
+			void *ptr = *(tbin->avail - 1 - i);
+			extent = item_extent[i];
+			assert(ptr != NULL && extent != NULL);
+
+			if (extent_arena_ind_get(extent) == bin_arena_ind
+			    && extent_binshard_get(extent) == binshard) {
+				arena_dalloc_bin_junked_locked(tsd_tsdn(tsd),
+				    bin_arena, bin, binind, extent, ptr);
+			} else {
+				/*
+				 * This object was allocated via a different
+				 * arena bin than the one that is currently
+				 * locked.  Stash the object, so that it can be
+				 * handled in a future pass.
+				 */
+				*(tbin->avail - 1 - ndeferred) = ptr;
+				item_extent[ndeferred] = extent;
+				ndeferred++;
+			}
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_decay_ticks(tsd_tsdn(tsd), bin_arena, nflush - ndeferred);
+		nflush = ndeferred;
+	}
+	if (config_stats && !merged_stats) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		unsigned binshard;
+		bin_t *bin = arena_bin_choose_lock(tsd_tsdn(tsd), arena, binind,
+		    &binshard);
+		bin->stats.nflushes++;
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+	}
+
+	memmove(tbin->avail - rem, tbin->avail - tbin->ncached, rem *
+	    sizeof(void *));
+	tbin->ncached = rem;
+	if (tbin->ncached < tbin->low_water) {
+		tbin->low_water = tbin->ncached;
+	}
+}
+
+void
+tcache_bin_flush_large(tsd_t *tsd, cache_bin_t *tbin, szind_t binind,
+    unsigned rem, tcache_t *tcache) {
+	bool merged_stats = false;
+
+	assert(binind < nhbins);
+	assert((cache_bin_sz_t)rem <= tbin->ncached);
+
+	arena_t *tcache_arena = tcache->arena;
+	assert(tcache_arena != NULL);
+	unsigned nflush = tbin->ncached - rem;
+	VARIABLE_ARRAY(extent_t *, item_extent, nflush);
+
+#ifndef JEMALLOC_EXTRA_SIZE_CHECK
+	/* Look up extent once per item. */
+	for (unsigned i = 0 ; i < nflush; i++) {
+		item_extent[i] = iealloc(tsd_tsdn(tsd), *(tbin->avail - 1 - i));
+	}
+#else
+	tbin_extents_lookup_size_check(tsd_tsdn(tsd), tbin, binind, nflush,
+	    item_extent);
+#endif
+	while (nflush > 0) {
+		/* Lock the arena associated with the first object. */
+		extent_t *extent = item_extent[0];
+		unsigned locked_arena_ind = extent_arena_ind_get(extent);
+		arena_t *locked_arena = arena_get(tsd_tsdn(tsd),
+		    locked_arena_ind, false);
+		bool idump;
+
+		if (config_prof) {
+			idump = false;
+		}
+
+		bool lock_large = !arena_is_auto(locked_arena);
+		if (lock_large) {
+			malloc_mutex_lock(tsd_tsdn(tsd), &locked_arena->large_mtx);
+		}
+		for (unsigned i = 0; i < nflush; i++) {
+			void *ptr = *(tbin->avail - 1 - i);
+			assert(ptr != NULL);
+			extent = item_extent[i];
+			if (extent_arena_ind_get(extent) == locked_arena_ind) {
+				large_dalloc_prep_junked_locked(tsd_tsdn(tsd),
+				    extent);
+			}
+		}
+		if ((config_prof || config_stats) &&
+		    (locked_arena == tcache_arena)) {
+			if (config_prof) {
+				idump = arena_prof_accum(tsd_tsdn(tsd),
+				    tcache_arena, tcache->prof_accumbytes);
+				tcache->prof_accumbytes = 0;
+			}
+			if (config_stats) {
+				merged_stats = true;
+				arena_stats_large_flush_nrequests_add(
+				    tsd_tsdn(tsd), &tcache_arena->stats, binind,
+				    tbin->tstats.nrequests);
+				tbin->tstats.nrequests = 0;
+			}
+		}
+		if (lock_large) {
+			malloc_mutex_unlock(tsd_tsdn(tsd), &locked_arena->large_mtx);
+		}
+
+		unsigned ndeferred = 0;
+		for (unsigned i = 0; i < nflush; i++) {
+			void *ptr = *(tbin->avail - 1 - i);
+			extent = item_extent[i];
+			assert(ptr != NULL && extent != NULL);
+
+			if (extent_arena_ind_get(extent) == locked_arena_ind) {
+				large_dalloc_finish(tsd_tsdn(tsd), extent);
+			} else {
+				/*
+				 * This object was allocated via a different
+				 * arena than the one that is currently locked.
+				 * Stash the object, so that it can be handled
+				 * in a future pass.
+				 */
+				*(tbin->avail - 1 - ndeferred) = ptr;
+				item_extent[ndeferred] = extent;
+				ndeferred++;
+			}
+		}
+		if (config_prof && idump) {
+			prof_idump(tsd_tsdn(tsd));
+		}
+		arena_decay_ticks(tsd_tsdn(tsd), locked_arena, nflush -
+		    ndeferred);
+		nflush = ndeferred;
+	}
+	if (config_stats && !merged_stats) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		arena_stats_large_flush_nrequests_add(tsd_tsdn(tsd),
+		    &tcache_arena->stats, binind, tbin->tstats.nrequests);
+		tbin->tstats.nrequests = 0;
+	}
+
+	memmove(tbin->avail - rem, tbin->avail - tbin->ncached, rem *
+	    sizeof(void *));
+	tbin->ncached = rem;
+	if (tbin->ncached < tbin->low_water) {
+		tbin->low_water = tbin->ncached;
+	}
+}
+
+void
+tcache_arena_associate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) {
+	assert(tcache->arena == NULL);
+	tcache->arena = arena;
+
+	if (config_stats) {
+		/* Link into list of extant tcaches. */
+		malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+
+		ql_elm_new(tcache, link);
+		ql_tail_insert(&arena->tcache_ql, tcache, link);
+		cache_bin_array_descriptor_init(
+		    &tcache->cache_bin_array_descriptor, tcache->bins_small,
+		    tcache->bins_large);
+		ql_tail_insert(&arena->cache_bin_array_descriptor_ql,
+		    &tcache->cache_bin_array_descriptor, link);
+
+		malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx);
+	}
+}
+
+static void
+tcache_arena_dissociate(tsdn_t *tsdn, tcache_t *tcache) {
+	arena_t *arena = tcache->arena;
+	assert(arena != NULL);
+	if (config_stats) {
+		/* Unlink from list of extant tcaches. */
+		malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+		if (config_debug) {
+			bool in_ql = false;
+			tcache_t *iter;
+			ql_foreach(iter, &arena->tcache_ql, link) {
+				if (iter == tcache) {
+					in_ql = true;
+					break;
+				}
+			}
+			assert(in_ql);
+		}
+		ql_remove(&arena->tcache_ql, tcache, link);
+		ql_remove(&arena->cache_bin_array_descriptor_ql,
+		    &tcache->cache_bin_array_descriptor, link);
+		tcache_stats_merge(tsdn, tcache, arena);
+		malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx);
+	}
+	tcache->arena = NULL;
+}
+
+void
+tcache_arena_reassociate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) {
+	tcache_arena_dissociate(tsdn, tcache);
+	tcache_arena_associate(tsdn, tcache, arena);
+}
+
+bool
+tsd_tcache_enabled_data_init(tsd_t *tsd) {
+	/* Called upon tsd initialization. */
+	tsd_tcache_enabled_set(tsd, opt_tcache);
+	tsd_slow_update(tsd);
+
+	if (opt_tcache) {
+		/* Trigger tcache init. */
+		tsd_tcache_data_init(tsd);
+	}
+
+	return false;
+}
+
+/* Initialize auto tcache (embedded in TSD). */
+static void
+tcache_init(tsd_t *tsd, tcache_t *tcache, void *avail_stack) {
+	memset(&tcache->link, 0, sizeof(ql_elm(tcache_t)));
+	tcache->prof_accumbytes = 0;
+	tcache->next_gc_bin = 0;
+	tcache->arena = NULL;
+
+	ticker_init(&tcache->gc_ticker, TCACHE_GC_INCR);
+
+	size_t stack_offset = 0;
+	assert((TCACHE_NSLOTS_SMALL_MAX & 1U) == 0);
+	memset(tcache->bins_small, 0, sizeof(cache_bin_t) * SC_NBINS);
+	memset(tcache->bins_large, 0, sizeof(cache_bin_t) * (nhbins - SC_NBINS));
+	unsigned i = 0;
+	for (; i < SC_NBINS; i++) {
+		tcache->lg_fill_div[i] = 1;
+		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *);
+		/*
+		 * avail points past the available space.  Allocations will
+		 * access the slots toward higher addresses (for the benefit of
+		 * prefetch).
+		 */
+		tcache_small_bin_get(tcache, i)->avail =
+		    (void **)((uintptr_t)avail_stack + (uintptr_t)stack_offset);
+	}
+	for (; i < nhbins; i++) {
+		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *);
+		tcache_large_bin_get(tcache, i)->avail =
+		    (void **)((uintptr_t)avail_stack + (uintptr_t)stack_offset);
+	}
+	assert(stack_offset == stack_nelms * sizeof(void *));
+}
+
+/* Initialize auto tcache (embedded in TSD). */
+bool
+tsd_tcache_data_init(tsd_t *tsd) {
+	tcache_t *tcache = tsd_tcachep_get_unsafe(tsd);
+	assert(tcache_small_bin_get(tcache, 0)->avail == NULL);
+	size_t size = stack_nelms * sizeof(void *);
+	/* Avoid false cacheline sharing. */
+	size = sz_sa2u(size, CACHELINE);
+
+	void *avail_array = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true,
+	    NULL, true, arena_get(TSDN_NULL, 0, true));
+	if (avail_array == NULL) {
+		return true;
+	}
+
+	tcache_init(tsd, tcache, avail_array);
+	/*
+	 * Initialization is a bit tricky here.  After malloc init is done, all
+	 * threads can rely on arena_choose and associate tcache accordingly.
+	 * However, the thread that does actual malloc bootstrapping relies on
+	 * functional tsd, and it can only rely on a0.  In that case, we
+	 * associate its tcache to a0 temporarily, and later on
+	 * arena_choose_hard() will re-associate properly.
+	 */
+	tcache->arena = NULL;
+	arena_t *arena;
+	if (!malloc_initialized()) {
+		/* If in initialization, assign to a0. */
+		arena = arena_get(tsd_tsdn(tsd), 0, false);
+		tcache_arena_associate(tsd_tsdn(tsd), tcache, arena);
+	} else {
+		arena = arena_choose(tsd, NULL);
+		/* This may happen if thread.tcache.enabled is used. */
+		if (tcache->arena == NULL) {
+			tcache_arena_associate(tsd_tsdn(tsd), tcache, arena);
+		}
+	}
+	assert(arena == tcache->arena);
+
+	return false;
+}
+
+/* Created manual tcache for tcache.create mallctl. */
+tcache_t *
+tcache_create_explicit(tsd_t *tsd) {
+	tcache_t *tcache;
+	size_t size, stack_offset;
+
+	size = sizeof(tcache_t);
+	/* Naturally align the pointer stacks. */
+	size = PTR_CEILING(size);
+	stack_offset = size;
+	size += stack_nelms * sizeof(void *);
+	/* Avoid false cacheline sharing. */
+	size = sz_sa2u(size, CACHELINE);
+
+	tcache = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true, NULL, true,
+	    arena_get(TSDN_NULL, 0, true));
+	if (tcache == NULL) {
+		return NULL;
+	}
+
+	tcache_init(tsd, tcache,
+	    (void *)((uintptr_t)tcache + (uintptr_t)stack_offset));
+	tcache_arena_associate(tsd_tsdn(tsd), tcache, arena_ichoose(tsd, NULL));
+
+	return tcache;
+}
+
+static void
+tcache_flush_cache(tsd_t *tsd, tcache_t *tcache) {
+	assert(tcache->arena != NULL);
+
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		cache_bin_t *tbin = tcache_small_bin_get(tcache, i);
+		tcache_bin_flush_small(tsd, tcache, tbin, i, 0);
+
+		if (config_stats) {
+			assert(tbin->tstats.nrequests == 0);
+		}
+	}
+	for (unsigned i = SC_NBINS; i < nhbins; i++) {
+		cache_bin_t *tbin = tcache_large_bin_get(tcache, i);
+		tcache_bin_flush_large(tsd, tbin, i, 0, tcache);
+
+		if (config_stats) {
+			assert(tbin->tstats.nrequests == 0);
+		}
+	}
+
+	if (config_prof && tcache->prof_accumbytes > 0 &&
+	    arena_prof_accum(tsd_tsdn(tsd), tcache->arena,
+	    tcache->prof_accumbytes)) {
+		prof_idump(tsd_tsdn(tsd));
+	}
+}
+
+void
+tcache_flush(tsd_t *tsd) {
+	assert(tcache_available(tsd));
+	tcache_flush_cache(tsd, tsd_tcachep_get(tsd));
+}
+
+static void
+tcache_destroy(tsd_t *tsd, tcache_t *tcache, bool tsd_tcache) {
+	tcache_flush_cache(tsd, tcache);
+	arena_t *arena = tcache->arena;
+	tcache_arena_dissociate(tsd_tsdn(tsd), tcache);
+
+	if (tsd_tcache) {
+		/* Release the avail array for the TSD embedded auto tcache. */
+		void *avail_array =
+		    (void *)((uintptr_t)tcache_small_bin_get(tcache, 0)->avail -
+		    (uintptr_t)tcache_bin_info[0].ncached_max * sizeof(void *));
+		idalloctm(tsd_tsdn(tsd), avail_array, NULL, NULL, true, true);
+	} else {
+		/* Release both the tcache struct and avail array. */
+		idalloctm(tsd_tsdn(tsd), tcache, NULL, NULL, true, true);
+	}
+
+	/*
+	 * The deallocation and tcache flush above may not trigger decay since
+	 * we are on the tcache shutdown path (potentially with non-nominal
+	 * tsd).  Manually trigger decay to avoid pathological cases.  Also
+	 * include arena 0 because the tcache array is allocated from it.
+	 */
+	arena_decay(tsd_tsdn(tsd), arena_get(tsd_tsdn(tsd), 0, false),
+	    false, false);
+
+	if (arena_nthreads_get(arena, false) == 0 &&
+	    !background_thread_enabled()) {
+		/* Force purging when no threads assigned to the arena anymore. */
+		arena_decay(tsd_tsdn(tsd), arena, false, true);
+	} else {
+		arena_decay(tsd_tsdn(tsd), arena, false, false);
+	}
+}
+
+/* For auto tcache (embedded in TSD) only. */
+void
+tcache_cleanup(tsd_t *tsd) {
+	tcache_t *tcache = tsd_tcachep_get(tsd);
+	if (!tcache_available(tsd)) {
+		assert(tsd_tcache_enabled_get(tsd) == false);
+		if (config_debug) {
+			assert(tcache_small_bin_get(tcache, 0)->avail == NULL);
+		}
+		return;
+	}
+	assert(tsd_tcache_enabled_get(tsd));
+	assert(tcache_small_bin_get(tcache, 0)->avail != NULL);
+
+	tcache_destroy(tsd, tcache, true);
+	if (config_debug) {
+		tcache_small_bin_get(tcache, 0)->avail = NULL;
+	}
+}
+
+void
+tcache_stats_merge(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) {
+	unsigned i;
+
+	cassert(config_stats);
+
+	/* Merge and reset tcache stats. */
+	for (i = 0; i < SC_NBINS; i++) {
+		cache_bin_t *tbin = tcache_small_bin_get(tcache, i);
+		unsigned binshard;
+		bin_t *bin = arena_bin_choose_lock(tsdn, arena, i, &binshard);
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		malloc_mutex_unlock(tsdn, &bin->lock);
+		tbin->tstats.nrequests = 0;
+	}
+
+	for (; i < nhbins; i++) {
+		cache_bin_t *tbin = tcache_large_bin_get(tcache, i);
+		arena_stats_large_flush_nrequests_add(tsdn, &arena->stats, i,
+		    tbin->tstats.nrequests);
+		tbin->tstats.nrequests = 0;
+	}
+}
+
+static bool
+tcaches_create_prep(tsd_t *tsd) {
+	bool err;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+
+	if (tcaches == NULL) {
+		tcaches = base_alloc(tsd_tsdn(tsd), b0get(), sizeof(tcache_t *)
+		    * (MALLOCX_TCACHE_MAX+1), CACHELINE);
+		if (tcaches == NULL) {
+			err = true;
+			goto label_return;
+		}
+	}
+
+	if (tcaches_avail == NULL && tcaches_past > MALLOCX_TCACHE_MAX) {
+		err = true;
+		goto label_return;
+	}
+
+	err = false;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+	return err;
+}
+
+bool
+tcaches_create(tsd_t *tsd, unsigned *r_ind) {
+	witness_assert_depth(tsdn_witness_tsdp_get(tsd_tsdn(tsd)), 0);
+
+	bool err;
+
+	if (tcaches_create_prep(tsd)) {
+		err = true;
+		goto label_return;
+	}
+
+	tcache_t *tcache = tcache_create_explicit(tsd);
+	if (tcache == NULL) {
+		err = true;
+		goto label_return;
+	}
+
+	tcaches_t *elm;
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+	if (tcaches_avail != NULL) {
+		elm = tcaches_avail;
+		tcaches_avail = tcaches_avail->next;
+		elm->tcache = tcache;
+		*r_ind = (unsigned)(elm - tcaches);
+	} else {
+		elm = &tcaches[tcaches_past];
+		elm->tcache = tcache;
+		*r_ind = tcaches_past;
+		tcaches_past++;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+
+	err = false;
+label_return:
+	witness_assert_depth(tsdn_witness_tsdp_get(tsd_tsdn(tsd)), 0);
+	return err;
+}
+
+static tcache_t *
+tcaches_elm_remove(tsd_t *tsd, tcaches_t *elm, bool allow_reinit) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &tcaches_mtx);
+
+	if (elm->tcache == NULL) {
+		return NULL;
+	}
+	tcache_t *tcache = elm->tcache;
+	if (allow_reinit) {
+		elm->tcache = TCACHES_ELM_NEED_REINIT;
+	} else {
+		elm->tcache = NULL;
+	}
+
+	if (tcache == TCACHES_ELM_NEED_REINIT) {
+		return NULL;
+	}
+	return tcache;
+}
+
+void
+tcaches_flush(tsd_t *tsd, unsigned ind) {
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+	tcache_t *tcache = tcaches_elm_remove(tsd, &tcaches[ind], true);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+	if (tcache != NULL) {
+		/* Destroy the tcache; recreate in tcaches_get() if needed. */
+		tcache_destroy(tsd, tcache, false);
+	}
+}
+
+void
+tcaches_destroy(tsd_t *tsd, unsigned ind) {
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+	tcaches_t *elm = &tcaches[ind];
+	tcache_t *tcache = tcaches_elm_remove(tsd, elm, false);
+	elm->next = tcaches_avail;
+	tcaches_avail = elm;
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+	if (tcache != NULL) {
+		tcache_destroy(tsd, tcache, false);
+	}
+}
+
+bool
+tcache_boot(tsdn_t *tsdn) {
+	/* If necessary, clamp opt_lg_tcache_max. */
+	if (opt_lg_tcache_max < 0 || (ZU(1) << opt_lg_tcache_max) <
+	    SC_SMALL_MAXCLASS) {
+		tcache_maxclass = SC_SMALL_MAXCLASS;
+	} else {
+		tcache_maxclass = (ZU(1) << opt_lg_tcache_max);
+	}
+
+	if (malloc_mutex_init(&tcaches_mtx, "tcaches", WITNESS_RANK_TCACHES,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	nhbins = sz_size2index(tcache_maxclass) + 1;
+
+	/* Initialize tcache_bin_info. */
+	tcache_bin_info = (cache_bin_info_t *)base_alloc(tsdn, b0get(), nhbins
+	    * sizeof(cache_bin_info_t), CACHELINE);
+	if (tcache_bin_info == NULL) {
+		return true;
+	}
+	stack_nelms = 0;
+	unsigned i;
+	for (i = 0; i < SC_NBINS; i++) {
+		if ((bin_infos[i].nregs << 1) <= TCACHE_NSLOTS_SMALL_MIN) {
+			tcache_bin_info[i].ncached_max =
+			    TCACHE_NSLOTS_SMALL_MIN;
+		} else if ((bin_infos[i].nregs << 1) <=
+		    TCACHE_NSLOTS_SMALL_MAX) {
+			tcache_bin_info[i].ncached_max =
+			    (bin_infos[i].nregs << 1);
+		} else {
+			tcache_bin_info[i].ncached_max =
+			    TCACHE_NSLOTS_SMALL_MAX;
+		}
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+	for (; i < nhbins; i++) {
+		tcache_bin_info[i].ncached_max = TCACHE_NSLOTS_LARGE;
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+
+	return false;
+}
+
+void
+tcache_prefork(tsdn_t *tsdn) {
+	if (!config_prof && opt_tcache) {
+		malloc_mutex_prefork(tsdn, &tcaches_mtx);
+	}
+}
+
+void
+tcache_postfork_parent(tsdn_t *tsdn) {
+	if (!config_prof && opt_tcache) {
+		malloc_mutex_postfork_parent(tsdn, &tcaches_mtx);
+	}
+}
+
+void
+tcache_postfork_child(tsdn_t *tsdn) {
+	if (!config_prof && opt_tcache) {
+		malloc_mutex_postfork_child(tsdn, &tcaches_mtx);
+	}
+}
diff --git a/jemalloc/src/test_hooks.c b/jemalloc/src/test_hooks.c
new file mode 100644
index 0000000..ace00d9
--- /dev/null
+++ b/jemalloc/src/test_hooks.c
@@ -0,0 +1,12 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+/*
+ * The hooks are a little bit screwy -- they're not genuinely exported in the
+ * sense that we want them available to end-users, but we do want them visible
+ * from outside the generated library, so that we can use them in test code.
+ */
+JEMALLOC_EXPORT
+void (*test_hooks_arena_new_hook)() = NULL;
+
+JEMALLOC_EXPORT
+void (*test_hooks_libc_hook)() = NULL;
diff --git a/jemalloc/src/ticker.c b/jemalloc/src/ticker.c
new file mode 100644
index 0000000..d7b8cd2
--- /dev/null
+++ b/jemalloc/src/ticker.c
@@ -0,0 +1,3 @@
+#define JEMALLOC_TICKER_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
diff --git a/jemalloc/src/tsd.c b/jemalloc/src/tsd.c
new file mode 100644
index 0000000..a31f6b9
--- /dev/null
+++ b/jemalloc/src/tsd.c
@@ -0,0 +1,534 @@
+#define JEMALLOC_TSD_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+
+/******************************************************************************/
+/* Data. */
+
+static unsigned ncleanups;
+static malloc_tsd_cleanup_t cleanups[MALLOC_TSD_CLEANUPS_MAX];
+
+/* TSD_INITIALIZER triggers "-Wmissing-field-initializer" */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls = TSD_INITIALIZER;
+JEMALLOC_TSD_TYPE_ATTR(bool) JEMALLOC_TLS_MODEL tsd_initialized = false;
+bool tsd_booted = false;
+#elif (defined(JEMALLOC_TLS))
+JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls = TSD_INITIALIZER;
+pthread_key_t tsd_tsd;
+bool tsd_booted = false;
+#elif (defined(_WIN32))
+DWORD tsd_tsd;
+tsd_wrapper_t tsd_boot_wrapper = {false, TSD_INITIALIZER};
+bool tsd_booted = false;
+#else
+
+/*
+ * This contains a mutex, but it's pretty convenient to allow the mutex code to
+ * have a dependency on tsd.  So we define the struct here, and only refer to it
+ * by pointer in the header.
+ */
+struct tsd_init_head_s {
+	ql_head(tsd_init_block_t) blocks;
+	malloc_mutex_t lock;
+};
+
+pthread_key_t tsd_tsd;
+tsd_init_head_t	tsd_init_head = {
+	ql_head_initializer(blocks),
+	MALLOC_MUTEX_INITIALIZER
+};
+
+tsd_wrapper_t tsd_boot_wrapper = {
+	false,
+	TSD_INITIALIZER
+};
+bool tsd_booted = false;
+#endif
+
+JEMALLOC_DIAGNOSTIC_POP
+
+/******************************************************************************/
+
+/* A list of all the tsds in the nominal state. */
+typedef ql_head(tsd_t) tsd_list_t;
+static tsd_list_t tsd_nominal_tsds = ql_head_initializer(tsd_nominal_tsds);
+static malloc_mutex_t tsd_nominal_tsds_lock;
+
+/* How many slow-path-enabling features are turned on. */
+static atomic_u32_t tsd_global_slow_count = ATOMIC_INIT(0);
+
+static bool
+tsd_in_nominal_list(tsd_t *tsd) {
+	tsd_t *tsd_list;
+	bool found = false;
+	/*
+	 * We don't know that tsd is nominal; it might not be safe to get data
+	 * out of it here.
+	 */
+	malloc_mutex_lock(TSDN_NULL, &tsd_nominal_tsds_lock);
+	ql_foreach(tsd_list, &tsd_nominal_tsds, TSD_MANGLE(tcache).tsd_link) {
+		if (tsd == tsd_list) {
+			found = true;
+			break;
+		}
+	}
+	malloc_mutex_unlock(TSDN_NULL, &tsd_nominal_tsds_lock);
+	return found;
+}
+
+static void
+tsd_add_nominal(tsd_t *tsd) {
+	assert(!tsd_in_nominal_list(tsd));
+	assert(tsd_state_get(tsd) <= tsd_state_nominal_max);
+	ql_elm_new(tsd, TSD_MANGLE(tcache).tsd_link);
+	malloc_mutex_lock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+	ql_tail_insert(&tsd_nominal_tsds, tsd, TSD_MANGLE(tcache).tsd_link);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+static void
+tsd_remove_nominal(tsd_t *tsd) {
+	assert(tsd_in_nominal_list(tsd));
+	assert(tsd_state_get(tsd) <= tsd_state_nominal_max);
+	malloc_mutex_lock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+	ql_remove(&tsd_nominal_tsds, tsd, TSD_MANGLE(tcache).tsd_link);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+static void
+tsd_force_recompute(tsdn_t *tsdn) {
+	/*
+	 * The stores to tsd->state here need to synchronize with the exchange
+	 * in tsd_slow_update.
+	 */
+	atomic_fence(ATOMIC_RELEASE);
+	malloc_mutex_lock(tsdn, &tsd_nominal_tsds_lock);
+	tsd_t *remote_tsd;
+	ql_foreach(remote_tsd, &tsd_nominal_tsds, TSD_MANGLE(tcache).tsd_link) {
+		assert(tsd_atomic_load(&remote_tsd->state, ATOMIC_RELAXED)
+		    <= tsd_state_nominal_max);
+		tsd_atomic_store(&remote_tsd->state, tsd_state_nominal_recompute,
+		    ATOMIC_RELAXED);
+	}
+	malloc_mutex_unlock(tsdn, &tsd_nominal_tsds_lock);
+}
+
+void
+tsd_global_slow_inc(tsdn_t *tsdn) {
+	atomic_fetch_add_u32(&tsd_global_slow_count, 1, ATOMIC_RELAXED);
+	/*
+	 * We unconditionally force a recompute, even if the global slow count
+	 * was already positive.  If we didn't, then it would be possible for us
+	 * to return to the user, have the user synchronize externally with some
+	 * other thread, and then have that other thread not have picked up the
+	 * update yet (since the original incrementing thread might still be
+	 * making its way through the tsd list).
+	 */
+	tsd_force_recompute(tsdn);
+}
+
+void tsd_global_slow_dec(tsdn_t *tsdn) {
+	atomic_fetch_sub_u32(&tsd_global_slow_count, 1, ATOMIC_RELAXED);
+	/* See the note in ..._inc(). */
+	tsd_force_recompute(tsdn);
+}
+
+static bool
+tsd_local_slow(tsd_t *tsd) {
+	return !tsd_tcache_enabled_get(tsd)
+	    || tsd_reentrancy_level_get(tsd) > 0;
+}
+
+bool
+tsd_global_slow() {
+	return atomic_load_u32(&tsd_global_slow_count, ATOMIC_RELAXED) > 0;
+}
+
+/******************************************************************************/
+
+static uint8_t
+tsd_state_compute(tsd_t *tsd) {
+	if (!tsd_nominal(tsd)) {
+		return tsd_state_get(tsd);
+	}
+	/* We're in *a* nominal state; but which one? */
+	if (malloc_slow || tsd_local_slow(tsd) || tsd_global_slow()) {
+		return tsd_state_nominal_slow;
+	} else {
+		return tsd_state_nominal;
+	}
+}
+
+void
+tsd_slow_update(tsd_t *tsd) {
+	uint8_t old_state;
+	do {
+		uint8_t new_state = tsd_state_compute(tsd);
+		old_state = tsd_atomic_exchange(&tsd->state, new_state,
+		    ATOMIC_ACQUIRE);
+	} while (old_state == tsd_state_nominal_recompute);
+}
+
+void
+tsd_state_set(tsd_t *tsd, uint8_t new_state) {
+	/* Only the tsd module can change the state *to* recompute. */
+	assert(new_state != tsd_state_nominal_recompute);
+	uint8_t old_state = tsd_atomic_load(&tsd->state, ATOMIC_RELAXED);
+	if (old_state > tsd_state_nominal_max) {
+		/*
+		 * Not currently in the nominal list, but it might need to be
+		 * inserted there.
+		 */
+		assert(!tsd_in_nominal_list(tsd));
+		tsd_atomic_store(&tsd->state, new_state, ATOMIC_RELAXED);
+		if (new_state <= tsd_state_nominal_max) {
+			tsd_add_nominal(tsd);
+		}
+	} else {
+		/*
+		 * We're currently nominal.  If the new state is non-nominal,
+		 * great; we take ourselves off the list and just enter the new
+		 * state.
+		 */
+		assert(tsd_in_nominal_list(tsd));
+		if (new_state > tsd_state_nominal_max) {
+			tsd_remove_nominal(tsd);
+			tsd_atomic_store(&tsd->state, new_state,
+			    ATOMIC_RELAXED);
+		} else {
+			/*
+			 * This is the tricky case.  We're transitioning from
+			 * one nominal state to another.  The caller can't know
+			 * about any races that are occuring at the same time,
+			 * so we always have to recompute no matter what.
+			 */
+			tsd_slow_update(tsd);
+		}
+	}
+}
+
+static bool
+tsd_data_init(tsd_t *tsd) {
+	/*
+	 * We initialize the rtree context first (before the tcache), since the
+	 * tcache initialization depends on it.
+	 */
+	rtree_ctx_data_init(tsd_rtree_ctxp_get_unsafe(tsd));
+
+	/*
+	 * A nondeterministic seed based on the address of tsd reduces
+	 * the likelihood of lockstep non-uniform cache index
+	 * utilization among identical concurrent processes, but at the
+	 * cost of test repeatability.  For debug builds, instead use a
+	 * deterministic seed.
+	 */
+	*tsd_offset_statep_get(tsd) = config_debug ? 0 :
+	    (uint64_t)(uintptr_t)tsd;
+
+	return tsd_tcache_enabled_data_init(tsd);
+}
+
+static void
+assert_tsd_data_cleanup_done(tsd_t *tsd) {
+	assert(!tsd_nominal(tsd));
+	assert(!tsd_in_nominal_list(tsd));
+	assert(*tsd_arenap_get_unsafe(tsd) == NULL);
+	assert(*tsd_iarenap_get_unsafe(tsd) == NULL);
+	assert(*tsd_arenas_tdata_bypassp_get_unsafe(tsd) == true);
+	assert(*tsd_arenas_tdatap_get_unsafe(tsd) == NULL);
+	assert(*tsd_tcache_enabledp_get_unsafe(tsd) == false);
+	assert(*tsd_prof_tdatap_get_unsafe(tsd) == NULL);
+}
+
+static bool
+tsd_data_init_nocleanup(tsd_t *tsd) {
+	assert(tsd_state_get(tsd) == tsd_state_reincarnated ||
+	    tsd_state_get(tsd) == tsd_state_minimal_initialized);
+	/*
+	 * During reincarnation, there is no guarantee that the cleanup function
+	 * will be called (deallocation may happen after all tsd destructors).
+	 * We set up tsd in a way that no cleanup is needed.
+	 */
+	rtree_ctx_data_init(tsd_rtree_ctxp_get_unsafe(tsd));
+	*tsd_arenas_tdata_bypassp_get(tsd) = true;
+	*tsd_tcache_enabledp_get_unsafe(tsd) = false;
+	*tsd_reentrancy_levelp_get(tsd) = 1;
+	assert_tsd_data_cleanup_done(tsd);
+
+	return false;
+}
+
+tsd_t *
+tsd_fetch_slow(tsd_t *tsd, bool minimal) {
+	assert(!tsd_fast(tsd));
+
+	if (tsd_state_get(tsd) == tsd_state_nominal_slow) {
+		/*
+		 * On slow path but no work needed.  Note that we can't
+		 * necessarily *assert* that we're slow, because we might be
+		 * slow because of an asynchronous modification to global state,
+		 * which might be asynchronously modified *back*.
+		 */
+	} else if (tsd_state_get(tsd) == tsd_state_nominal_recompute) {
+		tsd_slow_update(tsd);
+	} else if (tsd_state_get(tsd) == tsd_state_uninitialized) {
+		if (!minimal) {
+			if (tsd_booted) {
+				tsd_state_set(tsd, tsd_state_nominal);
+				tsd_slow_update(tsd);
+				/* Trigger cleanup handler registration. */
+				tsd_set(tsd);
+				tsd_data_init(tsd);
+			}
+		} else {
+			tsd_state_set(tsd, tsd_state_minimal_initialized);
+			tsd_set(tsd);
+			tsd_data_init_nocleanup(tsd);
+		}
+	} else if (tsd_state_get(tsd) == tsd_state_minimal_initialized) {
+		if (!minimal) {
+			/* Switch to fully initialized. */
+			tsd_state_set(tsd, tsd_state_nominal);
+			assert(*tsd_reentrancy_levelp_get(tsd) >= 1);
+			(*tsd_reentrancy_levelp_get(tsd))--;
+			tsd_slow_update(tsd);
+			tsd_data_init(tsd);
+		} else {
+			assert_tsd_data_cleanup_done(tsd);
+		}
+	} else if (tsd_state_get(tsd) == tsd_state_purgatory) {
+		tsd_state_set(tsd, tsd_state_reincarnated);
+		tsd_set(tsd);
+		tsd_data_init_nocleanup(tsd);
+	} else {
+		assert(tsd_state_get(tsd) == tsd_state_reincarnated);
+	}
+
+	return tsd;
+}
+
+void *
+malloc_tsd_malloc(size_t size) {
+	return a0malloc(CACHELINE_CEILING(size));
+}
+
+void
+malloc_tsd_dalloc(void *wrapper) {
+	a0dalloc(wrapper);
+}
+
+#if defined(JEMALLOC_MALLOC_THREAD_CLEANUP) || defined(_WIN32)
+#ifndef _WIN32
+JEMALLOC_EXPORT
+#endif
+void
+_malloc_thread_cleanup(void) {
+	bool pending[MALLOC_TSD_CLEANUPS_MAX], again;
+	unsigned i;
+
+	for (i = 0; i < ncleanups; i++) {
+		pending[i] = true;
+	}
+
+	do {
+		again = false;
+		for (i = 0; i < ncleanups; i++) {
+			if (pending[i]) {
+				pending[i] = cleanups[i]();
+				if (pending[i]) {
+					again = true;
+				}
+			}
+		}
+	} while (again);
+}
+#endif
+
+void
+malloc_tsd_cleanup_register(bool (*f)(void)) {
+	assert(ncleanups < MALLOC_TSD_CLEANUPS_MAX);
+	cleanups[ncleanups] = f;
+	ncleanups++;
+}
+
+static void
+tsd_do_data_cleanup(tsd_t *tsd) {
+	prof_tdata_cleanup(tsd);
+	iarena_cleanup(tsd);
+	arena_cleanup(tsd);
+	arenas_tdata_cleanup(tsd);
+	tcache_cleanup(tsd);
+	witnesses_cleanup(tsd_witness_tsdp_get_unsafe(tsd));
+}
+
+void
+tsd_cleanup(void *arg) {
+	tsd_t *tsd = (tsd_t *)arg;
+
+	switch (tsd_state_get(tsd)) {
+	case tsd_state_uninitialized:
+		/* Do nothing. */
+		break;
+	case tsd_state_minimal_initialized:
+		/* This implies the thread only did free() in its life time. */
+		/* Fall through. */
+	case tsd_state_reincarnated:
+		/*
+		 * Reincarnated means another destructor deallocated memory
+		 * after the destructor was called.  Cleanup isn't required but
+		 * is still called for testing and completeness.
+		 */
+		assert_tsd_data_cleanup_done(tsd);
+		/* Fall through. */
+	case tsd_state_nominal:
+	case tsd_state_nominal_slow:
+		tsd_do_data_cleanup(tsd);
+		tsd_state_set(tsd, tsd_state_purgatory);
+		tsd_set(tsd);
+		break;
+	case tsd_state_purgatory:
+		/*
+		 * The previous time this destructor was called, we set the
+		 * state to tsd_state_purgatory so that other destructors
+		 * wouldn't cause re-creation of the tsd.  This time, do
+		 * nothing, and do not request another callback.
+		 */
+		break;
+	default:
+		not_reached();
+	}
+#ifdef JEMALLOC_JET
+	test_callback_t test_callback = *tsd_test_callbackp_get_unsafe(tsd);
+	int *data = tsd_test_datap_get_unsafe(tsd);
+	if (test_callback != NULL) {
+		test_callback(data);
+	}
+#endif
+}
+
+tsd_t *
+malloc_tsd_boot0(void) {
+	tsd_t *tsd;
+
+	ncleanups = 0;
+	if (malloc_mutex_init(&tsd_nominal_tsds_lock, "tsd_nominal_tsds_lock",
+	    WITNESS_RANK_OMIT, malloc_mutex_rank_exclusive)) {
+		return NULL;
+	}
+	if (tsd_boot0()) {
+		return NULL;
+	}
+	tsd = tsd_fetch();
+	*tsd_arenas_tdata_bypassp_get(tsd) = true;
+	return tsd;
+}
+
+void
+malloc_tsd_boot1(void) {
+	tsd_boot1();
+	tsd_t *tsd = tsd_fetch();
+	/* malloc_slow has been set properly.  Update tsd_slow. */
+	tsd_slow_update(tsd);
+	*tsd_arenas_tdata_bypassp_get(tsd) = false;
+}
+
+#ifdef _WIN32
+static BOOL WINAPI
+_tls_callback(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved) {
+	switch (fdwReason) {
+#ifdef JEMALLOC_LAZY_LOCK
+	case DLL_THREAD_ATTACH:
+		isthreaded = true;
+		break;
+#endif
+	case DLL_THREAD_DETACH:
+		_malloc_thread_cleanup();
+		break;
+	default:
+		break;
+	}
+	return true;
+}
+
+/*
+ * We need to be able to say "read" here (in the "pragma section"), but have
+ * hooked "read". We won't read for the rest of the file, so we can get away
+ * with unhooking.
+ */
+#ifdef read
+#  undef read
+#endif
+
+#ifdef _MSC_VER
+#  ifdef _M_IX86
+#    pragma comment(linker, "/INCLUDE:__tls_used")
+#    pragma comment(linker, "/INCLUDE:_tls_callback")
+#  else
+#    pragma comment(linker, "/INCLUDE:_tls_used")
+#    pragma comment(linker, "/INCLUDE:" STRINGIFY(tls_callback) )
+#  endif
+#  pragma section(".CRT$XLY",long,read)
+#endif
+JEMALLOC_SECTION(".CRT$XLY") JEMALLOC_ATTR(used)
+BOOL	(WINAPI *const tls_callback)(HINSTANCE hinstDLL,
+    DWORD fdwReason, LPVOID lpvReserved) = _tls_callback;
+#endif
+
+#if (!defined(JEMALLOC_MALLOC_THREAD_CLEANUP) && !defined(JEMALLOC_TLS) && \
+    !defined(_WIN32))
+void *
+tsd_init_check_recursion(tsd_init_head_t *head, tsd_init_block_t *block) {
+	pthread_t self = pthread_self();
+	tsd_init_block_t *iter;
+
+	/* Check whether this thread has already inserted into the list. */
+	malloc_mutex_lock(TSDN_NULL, &head->lock);
+	ql_foreach(iter, &head->blocks, link) {
+		if (iter->thread == self) {
+			malloc_mutex_unlock(TSDN_NULL, &head->lock);
+			return iter->data;
+		}
+	}
+	/* Insert block into list. */
+	ql_elm_new(block, link);
+	block->thread = self;
+	ql_tail_insert(&head->blocks, block, link);
+	malloc_mutex_unlock(TSDN_NULL, &head->lock);
+	return NULL;
+}
+
+void
+tsd_init_finish(tsd_init_head_t *head, tsd_init_block_t *block) {
+	malloc_mutex_lock(TSDN_NULL, &head->lock);
+	ql_remove(&head->blocks, block, link);
+	malloc_mutex_unlock(TSDN_NULL, &head->lock);
+}
+#endif
+
+void
+tsd_prefork(tsd_t *tsd) {
+	malloc_mutex_prefork(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+void
+tsd_postfork_parent(tsd_t *tsd) {
+	malloc_mutex_postfork_parent(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+void
+tsd_postfork_child(tsd_t *tsd) {
+	malloc_mutex_postfork_child(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+	ql_new(&tsd_nominal_tsds);
+
+	if (tsd_state_get(tsd) <= tsd_state_nominal_max) {
+		tsd_add_nominal(tsd);
+	}
+}
diff --git a/jemalloc/src/witness.c b/jemalloc/src/witness.c
new file mode 100644
index 0000000..f42b72a
--- /dev/null
+++ b/jemalloc/src/witness.c
@@ -0,0 +1,100 @@
+#define JEMALLOC_WITNESS_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/malloc_io.h"
+
+void
+witness_init(witness_t *witness, const char *name, witness_rank_t rank,
+    witness_comp_t *comp, void *opaque) {
+	witness->name = name;
+	witness->rank = rank;
+	witness->comp = comp;
+	witness->opaque = opaque;
+}
+
+static void
+witness_lock_error_impl(const witness_list_t *witnesses,
+    const witness_t *witness) {
+	witness_t *w;
+
+	malloc_printf("<jemalloc>: Lock rank order reversal:");
+	ql_foreach(w, witnesses, link) {
+		malloc_printf(" %s(%u)", w->name, w->rank);
+	}
+	malloc_printf(" %s(%u)\n", witness->name, witness->rank);
+	abort();
+}
+witness_lock_error_t *JET_MUTABLE witness_lock_error = witness_lock_error_impl;
+
+static void
+witness_owner_error_impl(const witness_t *witness) {
+	malloc_printf("<jemalloc>: Should own %s(%u)\n", witness->name,
+	    witness->rank);
+	abort();
+}
+witness_owner_error_t *JET_MUTABLE witness_owner_error =
+    witness_owner_error_impl;
+
+static void
+witness_not_owner_error_impl(const witness_t *witness) {
+	malloc_printf("<jemalloc>: Should not own %s(%u)\n", witness->name,
+	    witness->rank);
+	abort();
+}
+witness_not_owner_error_t *JET_MUTABLE witness_not_owner_error =
+    witness_not_owner_error_impl;
+
+static void
+witness_depth_error_impl(const witness_list_t *witnesses,
+    witness_rank_t rank_inclusive, unsigned depth) {
+	witness_t *w;
+
+	malloc_printf("<jemalloc>: Should own %u lock%s of rank >= %u:", depth,
+	    (depth != 1) ?  "s" : "", rank_inclusive);
+	ql_foreach(w, witnesses, link) {
+		malloc_printf(" %s(%u)", w->name, w->rank);
+	}
+	malloc_printf("\n");
+	abort();
+}
+witness_depth_error_t *JET_MUTABLE witness_depth_error =
+    witness_depth_error_impl;
+
+void
+witnesses_cleanup(witness_tsd_t *witness_tsd) {
+	witness_assert_lockless(witness_tsd_tsdn(witness_tsd));
+
+	/* Do nothing. */
+}
+
+void
+witness_prefork(witness_tsd_t *witness_tsd) {
+	if (!config_debug) {
+		return;
+	}
+	witness_tsd->forking = true;
+}
+
+void
+witness_postfork_parent(witness_tsd_t *witness_tsd) {
+	if (!config_debug) {
+		return;
+	}
+	witness_tsd->forking = false;
+}
+
+void
+witness_postfork_child(witness_tsd_t *witness_tsd) {
+	if (!config_debug) {
+		return;
+	}
+#ifndef JEMALLOC_MUTEX_INIT_CB
+	witness_list_t *witnesses;
+
+	witnesses = &witness_tsd->witnesses;
+	ql_new(witnesses);
+#endif
+	witness_tsd->forking = false;
+}
diff --git a/jemalloc/src/zone.c b/jemalloc/src/zone.c
new file mode 100644
index 0000000..23dfdd0
--- /dev/null
+++ b/jemalloc/src/zone.c
@@ -0,0 +1,469 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+
+#ifndef JEMALLOC_ZONE
+#  error "This source file is for zones on Darwin (OS X)."
+#endif
+
+/* Definitions of the following structs in malloc/malloc.h might be too old
+ * for the built binary to run on newer versions of OSX. So use the newest
+ * possible version of those structs.
+ */
+typedef struct _malloc_zone_t {
+	void *reserved1;
+	void *reserved2;
+	size_t (*size)(struct _malloc_zone_t *, const void *);
+	void *(*malloc)(struct _malloc_zone_t *, size_t);
+	void *(*calloc)(struct _malloc_zone_t *, size_t, size_t);
+	void *(*valloc)(struct _malloc_zone_t *, size_t);
+	void (*free)(struct _malloc_zone_t *, void *);
+	void *(*realloc)(struct _malloc_zone_t *, void *, size_t);
+	void (*destroy)(struct _malloc_zone_t *);
+	const char *zone_name;
+	unsigned (*batch_malloc)(struct _malloc_zone_t *, size_t, void **, unsigned);
+	void (*batch_free)(struct _malloc_zone_t *, void **, unsigned);
+	struct malloc_introspection_t *introspect;
+	unsigned version;
+	void *(*memalign)(struct _malloc_zone_t *, size_t, size_t);
+	void (*free_definite_size)(struct _malloc_zone_t *, void *, size_t);
+	size_t (*pressure_relief)(struct _malloc_zone_t *, size_t);
+} malloc_zone_t;
+
+typedef struct {
+	vm_address_t address;
+	vm_size_t size;
+} vm_range_t;
+
+typedef struct malloc_statistics_t {
+	unsigned blocks_in_use;
+	size_t size_in_use;
+	size_t max_size_in_use;
+	size_t size_allocated;
+} malloc_statistics_t;
+
+typedef kern_return_t memory_reader_t(task_t, vm_address_t, vm_size_t, void **);
+
+typedef void vm_range_recorder_t(task_t, void *, unsigned type, vm_range_t *, unsigned);
+
+typedef struct malloc_introspection_t {
+	kern_return_t (*enumerator)(task_t, void *, unsigned, vm_address_t, memory_reader_t, vm_range_recorder_t);
+	size_t (*good_size)(malloc_zone_t *, size_t);
+	boolean_t (*check)(malloc_zone_t *);
+	void (*print)(malloc_zone_t *, boolean_t);
+	void (*log)(malloc_zone_t *, void *);
+	void (*force_lock)(malloc_zone_t *);
+	void (*force_unlock)(malloc_zone_t *);
+	void (*statistics)(malloc_zone_t *, malloc_statistics_t *);
+	boolean_t (*zone_locked)(malloc_zone_t *);
+	boolean_t (*enable_discharge_checking)(malloc_zone_t *);
+	boolean_t (*disable_discharge_checking)(malloc_zone_t *);
+	void (*discharge)(malloc_zone_t *, void *);
+#ifdef __BLOCKS__
+	void (*enumerate_discharged_pointers)(malloc_zone_t *, void (^)(void *, void *));
+#else
+	void *enumerate_unavailable_without_blocks;
+#endif
+	void (*reinit_lock)(malloc_zone_t *);
+} malloc_introspection_t;
+
+extern kern_return_t malloc_get_all_zones(task_t, memory_reader_t, vm_address_t **, unsigned *);
+
+extern malloc_zone_t *malloc_default_zone(void);
+
+extern void malloc_zone_register(malloc_zone_t *zone);
+
+extern void malloc_zone_unregister(malloc_zone_t *zone);
+
+/*
+ * The malloc_default_purgeable_zone() function is only available on >= 10.6.
+ * We need to check whether it is present at runtime, thus the weak_import.
+ */
+extern malloc_zone_t *malloc_default_purgeable_zone(void)
+JEMALLOC_ATTR(weak_import);
+
+/******************************************************************************/
+/* Data. */
+
+static malloc_zone_t *default_zone, *purgeable_zone;
+static malloc_zone_t jemalloc_zone;
+static struct malloc_introspection_t jemalloc_zone_introspect;
+static pid_t zone_force_lock_pid = -1;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static size_t	zone_size(malloc_zone_t *zone, const void *ptr);
+static void	*zone_malloc(malloc_zone_t *zone, size_t size);
+static void	*zone_calloc(malloc_zone_t *zone, size_t num, size_t size);
+static void	*zone_valloc(malloc_zone_t *zone, size_t size);
+static void	zone_free(malloc_zone_t *zone, void *ptr);
+static void	*zone_realloc(malloc_zone_t *zone, void *ptr, size_t size);
+static void	*zone_memalign(malloc_zone_t *zone, size_t alignment,
+    size_t size);
+static void	zone_free_definite_size(malloc_zone_t *zone, void *ptr,
+    size_t size);
+static void	zone_destroy(malloc_zone_t *zone);
+static unsigned	zone_batch_malloc(struct _malloc_zone_t *zone, size_t size,
+    void **results, unsigned num_requested);
+static void	zone_batch_free(struct _malloc_zone_t *zone,
+    void **to_be_freed, unsigned num_to_be_freed);
+static size_t	zone_pressure_relief(struct _malloc_zone_t *zone, size_t goal);
+static size_t	zone_good_size(malloc_zone_t *zone, size_t size);
+static kern_return_t	zone_enumerator(task_t task, void *data, unsigned type_mask,
+    vm_address_t zone_address, memory_reader_t reader,
+    vm_range_recorder_t recorder);
+static boolean_t	zone_check(malloc_zone_t *zone);
+static void	zone_print(malloc_zone_t *zone, boolean_t verbose);
+static void	zone_log(malloc_zone_t *zone, void *address);
+static void	zone_force_lock(malloc_zone_t *zone);
+static void	zone_force_unlock(malloc_zone_t *zone);
+static void	zone_statistics(malloc_zone_t *zone,
+    malloc_statistics_t *stats);
+static boolean_t	zone_locked(malloc_zone_t *zone);
+static void	zone_reinit_lock(malloc_zone_t *zone);
+
+/******************************************************************************/
+/*
+ * Functions.
+ */
+
+static size_t
+zone_size(malloc_zone_t *zone, const void *ptr) {
+	/*
+	 * There appear to be places within Darwin (such as setenv(3)) that
+	 * cause calls to this function with pointers that *no* zone owns.  If
+	 * we knew that all pointers were owned by *some* zone, we could split
+	 * our zone into two parts, and use one as the default allocator and
+	 * the other as the default deallocator/reallocator.  Since that will
+	 * not work in practice, we must check all pointers to assure that they
+	 * reside within a mapped extent before determining size.
+	 */
+	return ivsalloc(tsdn_fetch(), ptr);
+}
+
+static void *
+zone_malloc(malloc_zone_t *zone, size_t size) {
+	return je_malloc(size);
+}
+
+static void *
+zone_calloc(malloc_zone_t *zone, size_t num, size_t size) {
+	return je_calloc(num, size);
+}
+
+static void *
+zone_valloc(malloc_zone_t *zone, size_t size) {
+	void *ret = NULL; /* Assignment avoids useless compiler warning. */
+
+	je_posix_memalign(&ret, PAGE, size);
+
+	return ret;
+}
+
+static void
+zone_free(malloc_zone_t *zone, void *ptr) {
+	if (ivsalloc(tsdn_fetch(), ptr) != 0) {
+		je_free(ptr);
+		return;
+	}
+
+	free(ptr);
+}
+
+static void *
+zone_realloc(malloc_zone_t *zone, void *ptr, size_t size) {
+	if (ivsalloc(tsdn_fetch(), ptr) != 0) {
+		return je_realloc(ptr, size);
+	}
+
+	return realloc(ptr, size);
+}
+
+static void *
+zone_memalign(malloc_zone_t *zone, size_t alignment, size_t size) {
+	void *ret = NULL; /* Assignment avoids useless compiler warning. */
+
+	je_posix_memalign(&ret, alignment, size);
+
+	return ret;
+}
+
+static void
+zone_free_definite_size(malloc_zone_t *zone, void *ptr, size_t size) {
+	size_t alloc_size;
+
+	alloc_size = ivsalloc(tsdn_fetch(), ptr);
+	if (alloc_size != 0) {
+		assert(alloc_size == size);
+		je_free(ptr);
+		return;
+	}
+
+	free(ptr);
+}
+
+static void
+zone_destroy(malloc_zone_t *zone) {
+	/* This function should never be called. */
+	not_reached();
+}
+
+static unsigned
+zone_batch_malloc(struct _malloc_zone_t *zone, size_t size, void **results,
+    unsigned num_requested) {
+	unsigned i;
+
+	for (i = 0; i < num_requested; i++) {
+		results[i] = je_malloc(size);
+		if (!results[i])
+			break;
+	}
+
+	return i;
+}
+
+static void
+zone_batch_free(struct _malloc_zone_t *zone, void **to_be_freed,
+    unsigned num_to_be_freed) {
+	unsigned i;
+
+	for (i = 0; i < num_to_be_freed; i++) {
+		zone_free(zone, to_be_freed[i]);
+		to_be_freed[i] = NULL;
+	}
+}
+
+static size_t
+zone_pressure_relief(struct _malloc_zone_t *zone, size_t goal) {
+	return 0;
+}
+
+static size_t
+zone_good_size(malloc_zone_t *zone, size_t size) {
+	if (size == 0) {
+		size = 1;
+	}
+	return sz_s2u(size);
+}
+
+static kern_return_t
+zone_enumerator(task_t task, void *data, unsigned type_mask,
+    vm_address_t zone_address, memory_reader_t reader,
+    vm_range_recorder_t recorder) {
+	return KERN_SUCCESS;
+}
+
+static boolean_t
+zone_check(malloc_zone_t *zone) {
+	return true;
+}
+
+static void
+zone_print(malloc_zone_t *zone, boolean_t verbose) {
+}
+
+static void
+zone_log(malloc_zone_t *zone, void *address) {
+}
+
+static void
+zone_force_lock(malloc_zone_t *zone) {
+	if (isthreaded) {
+		/*
+		 * See the note in zone_force_unlock, below, to see why we need
+		 * this.
+		 */
+		assert(zone_force_lock_pid == -1);
+		zone_force_lock_pid = getpid();
+		jemalloc_prefork();
+	}
+}
+
+static void
+zone_force_unlock(malloc_zone_t *zone) {
+	/*
+	 * zone_force_lock and zone_force_unlock are the entry points to the
+	 * forking machinery on OS X.  The tricky thing is, the child is not
+	 * allowed to unlock mutexes locked in the parent, even if owned by the
+	 * forking thread (and the mutex type we use in OS X will fail an assert
+	 * if we try).  In the child, we can get away with reinitializing all
+	 * the mutexes, which has the effect of unlocking them.  In the parent,
+	 * doing this would mean we wouldn't wake any waiters blocked on the
+	 * mutexes we unlock.  So, we record the pid of the current thread in
+	 * zone_force_lock, and use that to detect if we're in the parent or
+	 * child here, to decide which unlock logic we need.
+	 */
+	if (isthreaded) {
+		assert(zone_force_lock_pid != -1);
+		if (getpid() == zone_force_lock_pid) {
+			jemalloc_postfork_parent();
+		} else {
+			jemalloc_postfork_child();
+		}
+		zone_force_lock_pid = -1;
+	}
+}
+
+static void
+zone_statistics(malloc_zone_t *zone, malloc_statistics_t *stats) {
+	/* We make no effort to actually fill the values */
+	stats->blocks_in_use = 0;
+	stats->size_in_use = 0;
+	stats->max_size_in_use = 0;
+	stats->size_allocated = 0;
+}
+
+static boolean_t
+zone_locked(malloc_zone_t *zone) {
+	/* Pretend no lock is being held */
+	return false;
+}
+
+static void
+zone_reinit_lock(malloc_zone_t *zone) {
+	/* As of OSX 10.12, this function is only used when force_unlock would
+	 * be used if the zone version were < 9. So just use force_unlock. */
+	zone_force_unlock(zone);
+}
+
+static void
+zone_init(void) {
+	jemalloc_zone.size = zone_size;
+	jemalloc_zone.malloc = zone_malloc;
+	jemalloc_zone.calloc = zone_calloc;
+	jemalloc_zone.valloc = zone_valloc;
+	jemalloc_zone.free = zone_free;
+	jemalloc_zone.realloc = zone_realloc;
+	jemalloc_zone.destroy = zone_destroy;
+	jemalloc_zone.zone_name = "jemalloc_zone";
+	jemalloc_zone.batch_malloc = zone_batch_malloc;
+	jemalloc_zone.batch_free = zone_batch_free;
+	jemalloc_zone.introspect = &jemalloc_zone_introspect;
+	jemalloc_zone.version = 9;
+	jemalloc_zone.memalign = zone_memalign;
+	jemalloc_zone.free_definite_size = zone_free_definite_size;
+	jemalloc_zone.pressure_relief = zone_pressure_relief;
+
+	jemalloc_zone_introspect.enumerator = zone_enumerator;
+	jemalloc_zone_introspect.good_size = zone_good_size;
+	jemalloc_zone_introspect.check = zone_check;
+	jemalloc_zone_introspect.print = zone_print;
+	jemalloc_zone_introspect.log = zone_log;
+	jemalloc_zone_introspect.force_lock = zone_force_lock;
+	jemalloc_zone_introspect.force_unlock = zone_force_unlock;
+	jemalloc_zone_introspect.statistics = zone_statistics;
+	jemalloc_zone_introspect.zone_locked = zone_locked;
+	jemalloc_zone_introspect.enable_discharge_checking = NULL;
+	jemalloc_zone_introspect.disable_discharge_checking = NULL;
+	jemalloc_zone_introspect.discharge = NULL;
+#ifdef __BLOCKS__
+	jemalloc_zone_introspect.enumerate_discharged_pointers = NULL;
+#else
+	jemalloc_zone_introspect.enumerate_unavailable_without_blocks = NULL;
+#endif
+	jemalloc_zone_introspect.reinit_lock = zone_reinit_lock;
+}
+
+static malloc_zone_t *
+zone_default_get(void) {
+	malloc_zone_t **zones = NULL;
+	unsigned int num_zones = 0;
+
+	/*
+	 * On OSX 10.12, malloc_default_zone returns a special zone that is not
+	 * present in the list of registered zones. That zone uses a "lite zone"
+	 * if one is present (apparently enabled when malloc stack logging is
+	 * enabled), or the first registered zone otherwise. In practice this
+	 * means unless malloc stack logging is enabled, the first registered
+	 * zone is the default.  So get the list of zones to get the first one,
+	 * instead of relying on malloc_default_zone.
+	 */
+	if (KERN_SUCCESS != malloc_get_all_zones(0, NULL,
+	    (vm_address_t**)&zones, &num_zones)) {
+		/*
+		 * Reset the value in case the failure happened after it was
+		 * set.
+		 */
+		num_zones = 0;
+	}
+
+	if (num_zones) {
+		return zones[0];
+	}
+
+	return malloc_default_zone();
+}
+
+/* As written, this function can only promote jemalloc_zone. */
+static void
+zone_promote(void) {
+	malloc_zone_t *zone;
+
+	do {
+		/*
+		 * Unregister and reregister the default zone.  On OSX >= 10.6,
+		 * unregistering takes the last registered zone and places it
+		 * at the location of the specified zone.  Unregistering the
+		 * default zone thus makes the last registered one the default.
+		 * On OSX < 10.6, unregistering shifts all registered zones.
+		 * The first registered zone then becomes the default.
+		 */
+		malloc_zone_unregister(default_zone);
+		malloc_zone_register(default_zone);
+
+		/*
+		 * On OSX 10.6, having the default purgeable zone appear before
+		 * the default zone makes some things crash because it thinks it
+		 * owns the default zone allocated pointers.  We thus
+		 * unregister/re-register it in order to ensure it's always
+		 * after the default zone.  On OSX < 10.6, there is no purgeable
+		 * zone, so this does nothing.  On OSX >= 10.6, unregistering
+		 * replaces the purgeable zone with the last registered zone
+		 * above, i.e. the default zone.  Registering it again then puts
+		 * it at the end, obviously after the default zone.
+		 */
+		if (purgeable_zone != NULL) {
+			malloc_zone_unregister(purgeable_zone);
+			malloc_zone_register(purgeable_zone);
+		}
+
+		zone = zone_default_get();
+	} while (zone != &jemalloc_zone);
+}
+
+JEMALLOC_ATTR(constructor)
+void
+zone_register(void) {
+	/*
+	 * If something else replaced the system default zone allocator, don't
+	 * register jemalloc's.
+	 */
+	default_zone = zone_default_get();
+	if (!default_zone->zone_name || strcmp(default_zone->zone_name,
+	    "DefaultMallocZone") != 0) {
+		return;
+	}
+
+	/*
+	 * The default purgeable zone is created lazily by OSX's libc.  It uses
+	 * the default zone when it is created for "small" allocations
+	 * (< 15 KiB), but assumes the default zone is a scalable_zone.  This
+	 * obviously fails when the default zone is the jemalloc zone, so
+	 * malloc_default_purgeable_zone() is called beforehand so that the
+	 * default purgeable zone is created when the default zone is still
+	 * a scalable_zone.  As purgeable zones only exist on >= 10.6, we need
+	 * to check for the existence of malloc_default_purgeable_zone() at
+	 * run time.
+	 */
+	purgeable_zone = (malloc_default_purgeable_zone == NULL) ? NULL :
+	    malloc_default_purgeable_zone();
+
+	/* Register the custom zone.  At this point it won't be the default. */
+	zone_init();
+	malloc_zone_register(&jemalloc_zone);
+
+	/* Promote the custom zone to be default. */
+	zone_promote();
+}
diff --git a/jemalloc/test/include/test/SFMT-alti.h b/jemalloc/test/include/test/SFMT-alti.h
new file mode 100644
index 0000000..a1885db
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-alti.h
@@ -0,0 +1,186 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/**
+ * @file SFMT-alti.h
+ *
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT)
+ * pseudorandom number generator
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * Copyright (C) 2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software.
+ * see LICENSE.txt
+ */
+
+#ifndef SFMT_ALTI_H
+#define SFMT_ALTI_H
+
+/**
+ * This function represents the recursion formula in AltiVec and BIG ENDIAN.
+ * @param a a 128-bit part of the interal state array
+ * @param b a 128-bit part of the interal state array
+ * @param c a 128-bit part of the interal state array
+ * @param d a 128-bit part of the interal state array
+ * @return output
+ */
+JEMALLOC_ALWAYS_INLINE
+vector unsigned int vec_recursion(vector unsigned int a,
+						vector unsigned int b,
+						vector unsigned int c,
+						vector unsigned int d) {
+
+    const vector unsigned int sl1 = ALTI_SL1;
+    const vector unsigned int sr1 = ALTI_SR1;
+#ifdef ONLY64
+    const vector unsigned int mask = ALTI_MSK64;
+    const vector unsigned char perm_sl = ALTI_SL2_PERM64;
+    const vector unsigned char perm_sr = ALTI_SR2_PERM64;
+#else
+    const vector unsigned int mask = ALTI_MSK;
+    const vector unsigned char perm_sl = ALTI_SL2_PERM;
+    const vector unsigned char perm_sr = ALTI_SR2_PERM;
+#endif
+    vector unsigned int v, w, x, y, z;
+    x = vec_perm(a, (vector unsigned int)perm_sl, perm_sl);
+    v = a;
+    y = vec_sr(b, sr1);
+    z = vec_perm(c, (vector unsigned int)perm_sr, perm_sr);
+    w = vec_sl(d, sl1);
+    z = vec_xor(z, w);
+    y = vec_and(y, mask);
+    v = vec_xor(v, x);
+    z = vec_xor(z, y);
+    z = vec_xor(z, v);
+    return z;
+}
+
+/**
+ * This function fills the internal state array with pseudorandom
+ * integers.
+ */
+static inline void gen_rand_all(sfmt_t *ctx) {
+    int i;
+    vector unsigned int r, r1, r2;
+
+    r1 = ctx->sfmt[N - 2].s;
+    r2 = ctx->sfmt[N - 1].s;
+    for (i = 0; i < N - POS1; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, ctx->sfmt[i + POS1].s, r1, r2);
+	ctx->sfmt[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, ctx->sfmt[i + POS1 - N].s, r1, r2);
+	ctx->sfmt[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+/**
+ * This function fills the user-specified array with pseudorandom
+ * integers.
+ *
+ * @param array an 128-bit array to be filled by pseudorandom numbers.
+ * @param size number of 128-bit pesudorandom numbers to be generated.
+ */
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size) {
+    int i, j;
+    vector unsigned int r, r1, r2;
+
+    r1 = ctx->sfmt[N - 2].s;
+    r2 = ctx->sfmt[N - 1].s;
+    for (i = 0; i < N - POS1; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, ctx->sfmt[i + POS1].s, r1, r2);
+	array[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, array[i + POS1 - N].s, r1, r2);
+	array[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    /* main loop */
+    for (; i < size - N; i++) {
+	r = vec_recursion(array[i - N].s, array[i + POS1 - N].s, r1, r2);
+	array[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    for (j = 0; j < 2 * N - size; j++) {
+	ctx->sfmt[j].s = array[j + size - N].s;
+    }
+    for (; i < size; i++) {
+	r = vec_recursion(array[i - N].s, array[i + POS1 - N].s, r1, r2);
+	array[i].s = r;
+	ctx->sfmt[j++].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+#ifndef ONLY64
+#if defined(__APPLE__)
+#define ALTI_SWAP (vector unsigned char) \
+	(4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11)
+#else
+#define ALTI_SWAP {4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11}
+#endif
+/**
+ * This function swaps high and low 32-bit of 64-bit integers in user
+ * specified array.
+ *
+ * @param array an 128-bit array to be swaped.
+ * @param size size of 128-bit array.
+ */
+static inline void swap(w128_t *array, int size) {
+    int i;
+    const vector unsigned char perm = ALTI_SWAP;
+
+    for (i = 0; i < size; i++) {
+	array[i].s = vec_perm(array[i].s, (vector unsigned int)perm, perm);
+    }
+}
+#endif
+
+#endif
diff --git a/jemalloc/test/include/test/SFMT-params.h b/jemalloc/test/include/test/SFMT-params.h
new file mode 100644
index 0000000..ade6622
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params.h
@@ -0,0 +1,132 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS_H
+#define SFMT_PARAMS_H
+
+#if !defined(MEXP)
+#ifdef __GNUC__
+  #warning "MEXP is not defined. I assume MEXP is 19937."
+#endif
+  #define MEXP 19937
+#endif
+/*-----------------
+  BASIC DEFINITIONS
+  -----------------*/
+/** Mersenne Exponent. The period of the sequence 
+ *  is a multiple of 2^MEXP-1.
+ * #define MEXP 19937 */
+/** SFMT generator has an internal state array of 128-bit integers,
+ * and N is its size. */
+#define N (MEXP / 128 + 1)
+/** N32 is the size of internal state array when regarded as an array
+ * of 32-bit integers.*/
+#define N32 (N * 4)
+/** N64 is the size of internal state array when regarded as an array
+ * of 64-bit integers.*/
+#define N64 (N * 2)
+
+/*----------------------
+  the parameters of SFMT
+  following definitions are in paramsXXXX.h file.
+  ----------------------*/
+/** the pick up position of the array.
+#define POS1 122 
+*/
+
+/** the parameter of shift left as four 32-bit registers.
+#define SL1 18
+ */
+
+/** the parameter of shift left as one 128-bit register. 
+ * The 128-bit integer is shifted by (SL2 * 8) bits. 
+#define SL2 1 
+*/
+
+/** the parameter of shift right as four 32-bit registers.
+#define SR1 11
+*/
+
+/** the parameter of shift right as one 128-bit register. 
+ * The 128-bit integer is shifted by (SL2 * 8) bits. 
+#define SR2 1 
+*/
+
+/** A bitmask, used in the recursion.  These parameters are introduced
+ * to break symmetry of SIMD.
+#define MSK1 0xdfffffefU
+#define MSK2 0xddfecb7fU
+#define MSK3 0xbffaffffU
+#define MSK4 0xbffffff6U 
+*/
+
+/** These definitions are part of a 128-bit period certification vector.
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0xc98e126aU
+*/
+
+#if MEXP == 607
+  #include "test/SFMT-params607.h"
+#elif MEXP == 1279
+  #include "test/SFMT-params1279.h"
+#elif MEXP == 2281
+  #include "test/SFMT-params2281.h"
+#elif MEXP == 4253
+  #include "test/SFMT-params4253.h"
+#elif MEXP == 11213
+  #include "test/SFMT-params11213.h"
+#elif MEXP == 19937
+  #include "test/SFMT-params19937.h"
+#elif MEXP == 44497
+  #include "test/SFMT-params44497.h"
+#elif MEXP == 86243
+  #include "test/SFMT-params86243.h"
+#elif MEXP == 132049
+  #include "test/SFMT-params132049.h"
+#elif MEXP == 216091
+  #include "test/SFMT-params216091.h"
+#else
+#ifdef __GNUC__
+  #error "MEXP is not valid."
+  #undef MEXP
+#else
+  #undef MEXP
+#endif
+
+#endif
+
+#endif /* SFMT_PARAMS_H */
diff --git a/jemalloc/test/include/test/SFMT-params11213.h b/jemalloc/test/include/test/SFMT-params11213.h
new file mode 100644
index 0000000..2994bd2
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params11213.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS11213_H
+#define SFMT_PARAMS11213_H
+
+#define POS1	68
+#define SL1	14
+#define SL2	3
+#define SR1	7
+#define SR2	3
+#define MSK1	0xeffff7fbU
+#define MSK2	0xffffffefU
+#define MSK3	0xdfdfbfffU
+#define MSK4	0x7fffdbfdU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0xe8148000U
+#define PARITY4	0xd0c7afa3U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12}
+    #define ALTI_SR2_PERM64	{13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-11213:68-14-3-7-3:effff7fb-ffffffef-dfdfbfff-7fffdbfd"
+
+#endif /* SFMT_PARAMS11213_H */
diff --git a/jemalloc/test/include/test/SFMT-params1279.h b/jemalloc/test/include/test/SFMT-params1279.h
new file mode 100644
index 0000000..d7959f9
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params1279.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS1279_H
+#define SFMT_PARAMS1279_H
+
+#define POS1	7
+#define SL1	14
+#define SL2	3
+#define SR1	5
+#define SR2	1
+#define MSK1	0xf7fefffdU
+#define MSK2	0x7fefcfffU
+#define MSK3	0xaff3ef3fU
+#define MSK4	0xb5ffff7fU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x20000000U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-1279:7-14-3-5-1:f7fefffd-7fefcfff-aff3ef3f-b5ffff7f"
+
+#endif /* SFMT_PARAMS1279_H */
diff --git a/jemalloc/test/include/test/SFMT-params132049.h b/jemalloc/test/include/test/SFMT-params132049.h
new file mode 100644
index 0000000..a1dcec3
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params132049.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS132049_H
+#define SFMT_PARAMS132049_H
+
+#define POS1	110
+#define SL1	19
+#define SL2	1
+#define SR1	21
+#define SR2	1
+#define MSK1	0xffffbb5fU
+#define MSK2	0xfb6ebf95U
+#define MSK3	0xfffefffaU
+#define MSK4	0xcff77fffU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0xcb520000U
+#define PARITY4	0xc7e91c7dU
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-132049:110-19-1-21-1:ffffbb5f-fb6ebf95-fffefffa-cff77fff"
+
+#endif /* SFMT_PARAMS132049_H */
diff --git a/jemalloc/test/include/test/SFMT-params19937.h b/jemalloc/test/include/test/SFMT-params19937.h
new file mode 100644
index 0000000..fb92b4c
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params19937.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS19937_H
+#define SFMT_PARAMS19937_H
+
+#define POS1	122
+#define SL1	18
+#define SL2	1
+#define SR1	11
+#define SR2	1
+#define MSK1	0xdfffffefU
+#define MSK2	0xddfecb7fU
+#define MSK3	0xbffaffffU
+#define MSK4	0xbffffff6U
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x13c9e684U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-19937:122-18-1-11-1:dfffffef-ddfecb7f-bffaffff-bffffff6"
+
+#endif /* SFMT_PARAMS19937_H */
diff --git a/jemalloc/test/include/test/SFMT-params216091.h b/jemalloc/test/include/test/SFMT-params216091.h
new file mode 100644
index 0000000..125ce28
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params216091.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS216091_H
+#define SFMT_PARAMS216091_H
+
+#define POS1	627
+#define SL1	11
+#define SL2	3
+#define SR1	10
+#define SR2	1
+#define MSK1	0xbff7bff7U
+#define MSK2	0xbfffffffU
+#define MSK3	0xbffffa7fU
+#define MSK4	0xffddfbfbU
+#define PARITY1	0xf8000001U
+#define PARITY2	0x89e80709U
+#define PARITY3	0x3bd2b64bU
+#define PARITY4	0x0c64b1e4U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-216091:627-11-3-10-1:bff7bff7-bfffffff-bffffa7f-ffddfbfb"
+
+#endif /* SFMT_PARAMS216091_H */
diff --git a/jemalloc/test/include/test/SFMT-params2281.h b/jemalloc/test/include/test/SFMT-params2281.h
new file mode 100644
index 0000000..0ef85c4
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params2281.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS2281_H
+#define SFMT_PARAMS2281_H
+
+#define POS1	12
+#define SL1	19
+#define SL2	1
+#define SR1	5
+#define SR2	1
+#define MSK1	0xbff7ffbfU
+#define MSK2	0xfdfffffeU
+#define MSK3	0xf7ffef7fU
+#define MSK4	0xf2f7cbbfU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x41dfa600U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-2281:12-19-1-5-1:bff7ffbf-fdfffffe-f7ffef7f-f2f7cbbf"
+
+#endif /* SFMT_PARAMS2281_H */
diff --git a/jemalloc/test/include/test/SFMT-params4253.h b/jemalloc/test/include/test/SFMT-params4253.h
new file mode 100644
index 0000000..9f07bc6
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params4253.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS4253_H
+#define SFMT_PARAMS4253_H
+
+#define POS1	17
+#define SL1	20
+#define SL2	1
+#define SR1	7
+#define SR2	1
+#define MSK1	0x9f7bffffU
+#define MSK2	0x9fffff5fU
+#define MSK3	0x3efffffbU
+#define MSK4	0xfffff7bbU
+#define PARITY1	0xa8000001U
+#define PARITY2	0xaf5390a3U
+#define PARITY3	0xb740b3f8U
+#define PARITY4	0x6c11486dU
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-4253:17-20-1-7-1:9f7bffff-9fffff5f-3efffffb-fffff7bb"
+
+#endif /* SFMT_PARAMS4253_H */
diff --git a/jemalloc/test/include/test/SFMT-params44497.h b/jemalloc/test/include/test/SFMT-params44497.h
new file mode 100644
index 0000000..85598fe
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params44497.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS44497_H
+#define SFMT_PARAMS44497_H
+
+#define POS1	330
+#define SL1	5
+#define SL2	3
+#define SR1	9
+#define SR2	3
+#define MSK1	0xeffffffbU
+#define MSK2	0xdfbebfffU
+#define MSK3	0xbfbf7befU
+#define MSK4	0x9ffd7bffU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0xa3ac4000U
+#define PARITY4	0xecc1327aU
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12}
+    #define ALTI_SR2_PERM64	{13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-44497:330-5-3-9-3:effffffb-dfbebfff-bfbf7bef-9ffd7bff"
+
+#endif /* SFMT_PARAMS44497_H */
diff --git a/jemalloc/test/include/test/SFMT-params607.h b/jemalloc/test/include/test/SFMT-params607.h
new file mode 100644
index 0000000..bc76485
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params607.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS607_H
+#define SFMT_PARAMS607_H
+
+#define POS1	2
+#define SL1	15
+#define SL2	3
+#define SR1	13
+#define SR2	3
+#define MSK1	0xfdff37ffU
+#define MSK2	0xef7f3f7dU
+#define MSK3	0xff777b7dU
+#define MSK4	0x7ff7fb2fU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x5986f054U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12}
+    #define ALTI_SR2_PERM64	{13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-607:2-15-3-13-3:fdff37ff-ef7f3f7d-ff777b7d-7ff7fb2f"
+
+#endif /* SFMT_PARAMS607_H */
diff --git a/jemalloc/test/include/test/SFMT-params86243.h b/jemalloc/test/include/test/SFMT-params86243.h
new file mode 100644
index 0000000..5e4d783
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-params86243.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS86243_H
+#define SFMT_PARAMS86243_H
+
+#define POS1	366
+#define SL1	6
+#define SL2	7
+#define SR1	19
+#define SR2	1
+#define MSK1	0xfdbffbffU
+#define MSK2	0xbff7ff3fU
+#define MSK3	0xfd77efffU
+#define MSK4	0xbf9ff3ffU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0xe9528d85U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(25,25,25,25,3,25,25,25,7,0,1,2,11,4,5,6)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(7,25,25,25,25,25,25,25,15,0,1,2,3,4,5,6)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{25,25,25,25,3,25,25,25,7,0,1,2,11,4,5,6}
+    #define ALTI_SL2_PERM64	{7,25,25,25,25,25,25,25,15,0,1,2,3,4,5,6}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-86243:366-6-7-19-1:fdbffbff-bff7ff3f-fd77efff-bf9ff3ff"
+
+#endif /* SFMT_PARAMS86243_H */
diff --git a/jemalloc/test/include/test/SFMT-sse2.h b/jemalloc/test/include/test/SFMT-sse2.h
new file mode 100644
index 0000000..169ad55
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT-sse2.h
@@ -0,0 +1,157 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/**
+ * @file  SFMT-sse2.h
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT) for Intel SSE2
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * @note We assume LITTLE ENDIAN in this file
+ *
+ * Copyright (C) 2006, 2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software, see LICENSE.txt
+ */
+
+#ifndef SFMT_SSE2_H
+#define SFMT_SSE2_H
+
+/**
+ * This function represents the recursion formula.
+ * @param a a 128-bit part of the interal state array
+ * @param b a 128-bit part of the interal state array
+ * @param c a 128-bit part of the interal state array
+ * @param d a 128-bit part of the interal state array
+ * @param mask 128-bit mask
+ * @return output
+ */
+JEMALLOC_ALWAYS_INLINE __m128i mm_recursion(__m128i *a, __m128i *b,
+				   __m128i c, __m128i d, __m128i mask) {
+    __m128i v, x, y, z;
+
+    x = _mm_load_si128(a);
+    y = _mm_srli_epi32(*b, SR1);
+    z = _mm_srli_si128(c, SR2);
+    v = _mm_slli_epi32(d, SL1);
+    z = _mm_xor_si128(z, x);
+    z = _mm_xor_si128(z, v);
+    x = _mm_slli_si128(x, SL2);
+    y = _mm_and_si128(y, mask);
+    z = _mm_xor_si128(z, x);
+    z = _mm_xor_si128(z, y);
+    return z;
+}
+
+/**
+ * This function fills the internal state array with pseudorandom
+ * integers.
+ */
+static inline void gen_rand_all(sfmt_t *ctx) {
+    int i;
+    __m128i r, r1, r2, mask;
+    mask = _mm_set_epi32(MSK4, MSK3, MSK2, MSK1);
+
+    r1 = _mm_load_si128(&ctx->sfmt[N - 2].si);
+    r2 = _mm_load_si128(&ctx->sfmt[N - 1].si);
+    for (i = 0; i < N - POS1; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &ctx->sfmt[i + POS1].si, r1, r2,
+	  mask);
+	_mm_store_si128(&ctx->sfmt[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &ctx->sfmt[i + POS1 - N].si, r1, r2,
+	  mask);
+	_mm_store_si128(&ctx->sfmt[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+/**
+ * This function fills the user-specified array with pseudorandom
+ * integers.
+ *
+ * @param array an 128-bit array to be filled by pseudorandom numbers.
+ * @param size number of 128-bit pesudorandom numbers to be generated.
+ */
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size) {
+    int i, j;
+    __m128i r, r1, r2, mask;
+    mask = _mm_set_epi32(MSK4, MSK3, MSK2, MSK1);
+
+    r1 = _mm_load_si128(&ctx->sfmt[N - 2].si);
+    r2 = _mm_load_si128(&ctx->sfmt[N - 1].si);
+    for (i = 0; i < N - POS1; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &ctx->sfmt[i + POS1].si, r1, r2,
+	  mask);
+	_mm_store_si128(&array[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &array[i + POS1 - N].si, r1, r2,
+	  mask);
+	_mm_store_si128(&array[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    /* main loop */
+    for (; i < size - N; i++) {
+	r = mm_recursion(&array[i - N].si, &array[i + POS1 - N].si, r1, r2,
+			 mask);
+	_mm_store_si128(&array[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    for (j = 0; j < 2 * N - size; j++) {
+	r = _mm_load_si128(&array[j + size - N].si);
+	_mm_store_si128(&ctx->sfmt[j].si, r);
+    }
+    for (; i < size; i++) {
+	r = mm_recursion(&array[i - N].si, &array[i + POS1 - N].si, r1, r2,
+			 mask);
+	_mm_store_si128(&array[i].si, r);
+	_mm_store_si128(&ctx->sfmt[j++].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+#endif
diff --git a/jemalloc/test/include/test/SFMT.h b/jemalloc/test/include/test/SFMT.h
new file mode 100644
index 0000000..863fc55
--- /dev/null
+++ b/jemalloc/test/include/test/SFMT.h
@@ -0,0 +1,146 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/** 
+ * @file SFMT.h 
+ *
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT) pseudorandom
+ * number generator
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * Copyright (C) 2006, 2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software.
+ * see LICENSE.txt
+ *
+ * @note We assume that your system has inttypes.h.  If your system
+ * doesn't have inttypes.h, you have to typedef uint32_t and uint64_t,
+ * and you have to define PRIu64 and PRIx64 in this file as follows:
+ * @verbatim
+ typedef unsigned int uint32_t
+ typedef unsigned long long uint64_t  
+ #define PRIu64 "llu"
+ #define PRIx64 "llx"
+@endverbatim
+ * uint32_t must be exactly 32-bit unsigned integer type (no more, no
+ * less), and uint64_t must be exactly 64-bit unsigned integer type.
+ * PRIu64 and PRIx64 are used for printf function to print 64-bit
+ * unsigned int and 64-bit unsigned int in hexadecimal format.
+ */
+
+#ifndef SFMT_H
+#define SFMT_H
+
+typedef struct sfmt_s sfmt_t;
+
+uint32_t gen_rand32(sfmt_t *ctx);
+uint32_t gen_rand32_range(sfmt_t *ctx, uint32_t limit);
+uint64_t gen_rand64(sfmt_t *ctx);
+uint64_t gen_rand64_range(sfmt_t *ctx, uint64_t limit);
+void fill_array32(sfmt_t *ctx, uint32_t *array, int size);
+void fill_array64(sfmt_t *ctx, uint64_t *array, int size);
+sfmt_t *init_gen_rand(uint32_t seed);
+sfmt_t *init_by_array(uint32_t *init_key, int key_length);
+void fini_gen_rand(sfmt_t *ctx);
+const char *get_idstring(void);
+int get_min_array_size32(void);
+int get_min_array_size64(void);
+
+/* These real versions are due to Isaku Wada */
+/** generates a random number on [0,1]-real-interval */
+static inline double to_real1(uint32_t v) {
+    return v * (1.0/4294967295.0); 
+    /* divided by 2^32-1 */ 
+}
+
+/** generates a random number on [0,1]-real-interval */
+static inline double genrand_real1(sfmt_t *ctx) {
+    return to_real1(gen_rand32(ctx));
+}
+
+/** generates a random number on [0,1)-real-interval */
+static inline double to_real2(uint32_t v) {
+    return v * (1.0/4294967296.0); 
+    /* divided by 2^32 */
+}
+
+/** generates a random number on [0,1)-real-interval */
+static inline double genrand_real2(sfmt_t *ctx) {
+    return to_real2(gen_rand32(ctx));
+}
+
+/** generates a random number on (0,1)-real-interval */
+static inline double to_real3(uint32_t v) {
+    return (((double)v) + 0.5)*(1.0/4294967296.0); 
+    /* divided by 2^32 */
+}
+
+/** generates a random number on (0,1)-real-interval */
+static inline double genrand_real3(sfmt_t *ctx) {
+    return to_real3(gen_rand32(ctx));
+}
+/** These real versions are due to Isaku Wada */
+
+/** generates a random number on [0,1) with 53-bit resolution*/
+static inline double to_res53(uint64_t v) {
+    return v * (1.0/18446744073709551616.0L);
+}
+
+/** generates a random number on [0,1) with 53-bit resolution from two
+ * 32 bit integers */
+static inline double to_res53_mix(uint32_t x, uint32_t y) {
+    return to_res53(x | ((uint64_t)y << 32));
+}
+
+/** generates a random number on [0,1) with 53-bit resolution
+ */
+static inline double genrand_res53(sfmt_t *ctx) {
+    return to_res53(gen_rand64(ctx));
+}
+
+/** generates a random number on [0,1) with 53-bit resolution
+    using 32bit integer.
+ */
+static inline double genrand_res53_mix(sfmt_t *ctx) {
+    uint32_t x, y;
+
+    x = gen_rand32(ctx);
+    y = gen_rand32(ctx);
+    return to_res53_mix(x, y);
+}
+#endif
diff --git a/jemalloc/test/include/test/btalloc.h b/jemalloc/test/include/test/btalloc.h
new file mode 100644
index 0000000..5877ea7
--- /dev/null
+++ b/jemalloc/test/include/test/btalloc.h
@@ -0,0 +1,30 @@
+/* btalloc() provides a mechanism for allocating via permuted backtraces. */
+void	*btalloc(size_t size, unsigned bits);
+
+#define btalloc_n_proto(n)						\
+void	*btalloc_##n(size_t size, unsigned bits);
+btalloc_n_proto(0)
+btalloc_n_proto(1)
+
+#define btalloc_n_gen(n)						\
+void *									\
+btalloc_##n(size_t size, unsigned bits) {				\
+	void *p;							\
+									\
+	if (bits == 0) {						\
+		p = mallocx(size, 0);					\
+	} else {							\
+		switch (bits & 0x1U) {					\
+		case 0:							\
+			p = (btalloc_0(size, bits >> 1));		\
+			break;						\
+		case 1:							\
+			p = (btalloc_1(size, bits >> 1));		\
+			break;						\
+		default: not_reached();					\
+		}							\
+	}								\
+	/* Intentionally sabotage tail call optimization. */		\
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");		\
+	return p;							\
+}
diff --git a/jemalloc/test/include/test/extent_hooks.h b/jemalloc/test/include/test/extent_hooks.h
new file mode 100644
index 0000000..1f06201
--- /dev/null
+++ b/jemalloc/test/include/test/extent_hooks.h
@@ -0,0 +1,289 @@
+/*
+ * Boilerplate code used for testing extent hooks via interception and
+ * passthrough.
+ */
+
+static void	*extent_alloc_hook(extent_hooks_t *extent_hooks, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit,
+    unsigned arena_ind);
+static bool	extent_dalloc_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static void	extent_destroy_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static bool	extent_commit_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_decommit_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_purge_lazy_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_purge_forced_hook(extent_hooks_t *extent_hooks,
+    void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_split_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t size_a, size_t size_b, bool committed,
+    unsigned arena_ind);
+static bool	extent_merge_hook(extent_hooks_t *extent_hooks, void *addr_a,
+    size_t size_a, void *addr_b, size_t size_b, bool committed,
+    unsigned arena_ind);
+
+static extent_hooks_t *default_hooks;
+static extent_hooks_t hooks = {
+	extent_alloc_hook,
+	extent_dalloc_hook,
+	extent_destroy_hook,
+	extent_commit_hook,
+	extent_decommit_hook,
+	extent_purge_lazy_hook,
+	extent_purge_forced_hook,
+	extent_split_hook,
+	extent_merge_hook
+};
+
+/* Control whether hook functions pass calls through to default hooks. */
+static bool try_alloc = true;
+static bool try_dalloc = true;
+static bool try_destroy = true;
+static bool try_commit = true;
+static bool try_decommit = true;
+static bool try_purge_lazy = true;
+static bool try_purge_forced = true;
+static bool try_split = true;
+static bool try_merge = true;
+
+/* Set to false prior to operations, then introspect after operations. */
+static bool called_alloc;
+static bool called_dalloc;
+static bool called_destroy;
+static bool called_commit;
+static bool called_decommit;
+static bool called_purge_lazy;
+static bool called_purge_forced;
+static bool called_split;
+static bool called_merge;
+
+/* Set to false prior to operations, then introspect after operations. */
+static bool did_alloc;
+static bool did_dalloc;
+static bool did_destroy;
+static bool did_commit;
+static bool did_decommit;
+static bool did_purge_lazy;
+static bool did_purge_forced;
+static bool did_split;
+static bool did_merge;
+
+#if 0
+#  define TRACE_HOOK(fmt, ...) malloc_printf(fmt, __VA_ARGS__)
+#else
+#  define TRACE_HOOK(fmt, ...)
+#endif
+
+static void *
+extent_alloc_hook(extent_hooks_t *extent_hooks, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit, unsigned arena_ind) {
+	void *ret;
+
+	TRACE_HOOK("%s(extent_hooks=%p, new_addr=%p, size=%zu, alignment=%zu, "
+	    "*zero=%s, *commit=%s, arena_ind=%u)\n", __func__, extent_hooks,
+	    new_addr, size, alignment, *zero ?  "true" : "false", *commit ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->alloc, extent_alloc_hook,
+	    "Wrong hook function");
+	called_alloc = true;
+	if (!try_alloc) {
+		return NULL;
+	}
+	ret = default_hooks->alloc(default_hooks, new_addr, size, alignment,
+	    zero, commit, 0);
+	did_alloc = (ret != NULL);
+	return ret;
+}
+
+static bool
+extent_dalloc_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, committed=%s, "
+	    "arena_ind=%u)\n", __func__, extent_hooks, addr, size, committed ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->dalloc, extent_dalloc_hook,
+	    "Wrong hook function");
+	called_dalloc = true;
+	if (!try_dalloc) {
+		return true;
+	}
+	err = default_hooks->dalloc(default_hooks, addr, size, committed, 0);
+	did_dalloc = !err;
+	return err;
+}
+
+static void
+extent_destroy_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, committed=%s, "
+	    "arena_ind=%u)\n", __func__, extent_hooks, addr, size, committed ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->destroy, extent_destroy_hook,
+	    "Wrong hook function");
+	called_destroy = true;
+	if (!try_destroy) {
+		return;
+	}
+	default_hooks->destroy(default_hooks, addr, size, committed, 0);
+	did_destroy = true;
+}
+
+static bool
+extent_commit_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu, arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->commit, extent_commit_hook,
+	    "Wrong hook function");
+	called_commit = true;
+	if (!try_commit) {
+		return true;
+	}
+	err = default_hooks->commit(default_hooks, addr, size, offset, length,
+	    0);
+	did_commit = !err;
+	return err;
+}
+
+static bool
+extent_decommit_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu, arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->decommit, extent_decommit_hook,
+	    "Wrong hook function");
+	called_decommit = true;
+	if (!try_decommit) {
+		return true;
+	}
+	err = default_hooks->decommit(default_hooks, addr, size, offset, length,
+	    0);
+	did_decommit = !err;
+	return err;
+}
+
+static bool
+extent_purge_lazy_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->purge_lazy, extent_purge_lazy_hook,
+	    "Wrong hook function");
+	called_purge_lazy = true;
+	if (!try_purge_lazy) {
+		return true;
+	}
+	err = default_hooks->purge_lazy == NULL ||
+	    default_hooks->purge_lazy(default_hooks, addr, size, offset, length,
+	    0);
+	did_purge_lazy = !err;
+	return err;
+}
+
+static bool
+extent_purge_forced_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->purge_forced, extent_purge_forced_hook,
+	    "Wrong hook function");
+	called_purge_forced = true;
+	if (!try_purge_forced) {
+		return true;
+	}
+	err = default_hooks->purge_forced == NULL ||
+	    default_hooks->purge_forced(default_hooks, addr, size, offset,
+	    length, 0);
+	did_purge_forced = !err;
+	return err;
+}
+
+static bool
+extent_split_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t size_a, size_t size_b, bool committed, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, size_a=%zu, "
+	    "size_b=%zu, committed=%s, arena_ind=%u)\n", __func__, extent_hooks,
+	    addr, size, size_a, size_b, committed ? "true" : "false",
+	    arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->split, extent_split_hook,
+	    "Wrong hook function");
+	called_split = true;
+	if (!try_split) {
+		return true;
+	}
+	err = (default_hooks->split == NULL ||
+	    default_hooks->split(default_hooks, addr, size, size_a, size_b,
+	    committed, 0));
+	did_split = !err;
+	return err;
+}
+
+static bool
+extent_merge_hook(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a,
+    void *addr_b, size_t size_b, bool committed, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr_a=%p, size_a=%zu, addr_b=%p "
+	    "size_b=%zu, committed=%s, arena_ind=%u)\n", __func__, extent_hooks,
+	    addr_a, size_a, addr_b, size_b, committed ? "true" : "false",
+	    arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->merge, extent_merge_hook,
+	    "Wrong hook function");
+	assert_ptr_eq((void *)((uintptr_t)addr_a + size_a), addr_b,
+	    "Extents not mergeable");
+	called_merge = true;
+	if (!try_merge) {
+		return true;
+	}
+	err = (default_hooks->merge == NULL ||
+	    default_hooks->merge(default_hooks, addr_a, size_a, addr_b, size_b,
+	    committed, 0));
+	did_merge = !err;
+	return err;
+}
+
+static void
+extent_hooks_prep(void) {
+	size_t sz;
+
+	sz = sizeof(default_hooks);
+	assert_d_eq(mallctl("arena.0.extent_hooks", (void *)&default_hooks, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() error");
+}
diff --git a/jemalloc/test/include/test/jemalloc_test.h.in b/jemalloc/test/include/test/jemalloc_test.h.in
new file mode 100644
index 0000000..c46af5d
--- /dev/null
+++ b/jemalloc/test/include/test/jemalloc_test.h.in
@@ -0,0 +1,173 @@
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <limits.h>
+#ifndef SIZE_T_MAX
+#  define SIZE_T_MAX	SIZE_MAX
+#endif
+#include <stdlib.h>
+#include <stdarg.h>
+#include <stdbool.h>
+#include <errno.h>
+#include <math.h>
+#include <string.h>
+#ifdef _WIN32
+#  include "msvc_compat/strings.h"
+#endif
+
+#ifdef _WIN32
+#  include <windows.h>
+#  include "msvc_compat/windows_extra.h"
+#else
+#  include <pthread.h>
+#endif
+
+#include "test/jemalloc_test_defs.h"
+
+#if defined(JEMALLOC_OSATOMIC)
+#  include <libkern/OSAtomic.h>
+#endif
+
+#if defined(HAVE_ALTIVEC) && !defined(__APPLE__)
+#  include <altivec.h>
+#endif
+#ifdef HAVE_SSE2
+#  include <emmintrin.h>
+#endif
+
+/******************************************************************************/
+/*
+ * For unit tests, expose all public and private interfaces.
+ */
+#ifdef JEMALLOC_UNIT_TEST
+#  define JEMALLOC_JET
+#  define JEMALLOC_MANGLE
+#  include "jemalloc/internal/jemalloc_preamble.h"
+#  include "jemalloc/internal/jemalloc_internal_includes.h"
+
+/******************************************************************************/
+/*
+ * For integration tests, expose the public jemalloc interfaces, but only
+ * expose the minimum necessary internal utility code (to avoid re-implementing
+ * essentially identical code within the test infrastructure).
+ */
+#elif defined(JEMALLOC_INTEGRATION_TEST) || \
+    defined(JEMALLOC_INTEGRATION_CPP_TEST)
+#  define JEMALLOC_MANGLE
+#  include "jemalloc/jemalloc@install_suffix@.h"
+#  include "jemalloc/internal/jemalloc_internal_defs.h"
+#  include "jemalloc/internal/jemalloc_internal_macros.h"
+
+static const bool config_debug =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+
+#  define JEMALLOC_N(n) @private_namespace@##n
+#  include "jemalloc/internal/private_namespace.h"
+#  include "jemalloc/internal/test_hooks.h"
+
+/* Hermetic headers. */
+#  include "jemalloc/internal/assert.h"
+#  include "jemalloc/internal/malloc_io.h"
+#  include "jemalloc/internal/nstime.h"
+#  include "jemalloc/internal/util.h"
+
+/* Non-hermetic headers. */
+#  include "jemalloc/internal/qr.h"
+#  include "jemalloc/internal/ql.h"
+
+/******************************************************************************/
+/*
+ * For stress tests, expose the public jemalloc interfaces with name mangling
+ * so that they can be tested as e.g. malloc() and free().  Also expose the
+ * public jemalloc interfaces with jet_ prefixes, so that stress tests can use
+ * a separate allocator for their internal data structures.
+ */
+#elif defined(JEMALLOC_STRESS_TEST)
+#  include "jemalloc/jemalloc@install_suffix@.h"
+
+#  include "jemalloc/jemalloc_protos_jet.h"
+
+#  define JEMALLOC_JET
+#  include "jemalloc/internal/jemalloc_preamble.h"
+#  include "jemalloc/internal/jemalloc_internal_includes.h"
+#  include "jemalloc/internal/public_unnamespace.h"
+#  undef JEMALLOC_JET
+
+#  include "jemalloc/jemalloc_rename.h"
+#  define JEMALLOC_MANGLE
+#  ifdef JEMALLOC_STRESS_TESTLIB
+#    include "jemalloc/jemalloc_mangle_jet.h"
+#  else
+#    include "jemalloc/jemalloc_mangle.h"
+#  endif
+
+/******************************************************************************/
+/*
+ * This header does dangerous things, the effects of which only test code
+ * should be subject to.
+ */
+#else
+#  error "This header cannot be included outside a testing context"
+#endif
+
+/******************************************************************************/
+/*
+ * Common test utilities.
+ */
+#include "test/btalloc.h"
+#include "test/math.h"
+#include "test/mtx.h"
+#include "test/mq.h"
+#include "test/test.h"
+#include "test/timer.h"
+#include "test/thd.h"
+#define MEXP 19937
+#include "test/SFMT.h"
+
+/******************************************************************************/
+/*
+ * Define always-enabled assertion macros, so that test assertions execute even
+ * if assertions are disabled in the library code.
+ */
+#undef assert
+#undef not_reached
+#undef not_implemented
+#undef assert_not_implemented
+
+#define assert(e) do {							\
+	if (!(e)) {							\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Failed assertion: \"%s\"\n",	\
+		    __FILE__, __LINE__, #e);				\
+		abort();						\
+	}								\
+} while (0)
+
+#define not_reached() do {						\
+	malloc_printf(							\
+	    "<jemalloc>: %s:%d: Unreachable code reached\n",		\
+	    __FILE__, __LINE__);					\
+	abort();							\
+} while (0)
+
+#define not_implemented() do {						\
+	malloc_printf("<jemalloc>: %s:%d: Not implemented\n",		\
+	    __FILE__, __LINE__);					\
+	abort();							\
+} while (0)
+
+#define assert_not_implemented(e) do {					\
+	if (!(e)) {							\
+		not_implemented();					\
+	}								\
+} while (0)
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/jemalloc/test/include/test/jemalloc_test_defs.h.in b/jemalloc/test/include/test/jemalloc_test_defs.h.in
new file mode 100644
index 0000000..5cc8532
--- /dev/null
+++ b/jemalloc/test/include/test/jemalloc_test_defs.h.in
@@ -0,0 +1,9 @@
+#include "jemalloc/internal/jemalloc_internal_defs.h"
+#include "jemalloc/internal/jemalloc_internal_decls.h"
+
+/*
+ * For use by SFMT.  configure.ac doesn't actually define HAVE_SSE2 because its
+ * dependencies are notoriously unportable in practice.
+ */
+#undef HAVE_SSE2
+#undef HAVE_ALTIVEC
diff --git a/jemalloc/test/include/test/math.h b/jemalloc/test/include/test/math.h
new file mode 100644
index 0000000..efba086
--- /dev/null
+++ b/jemalloc/test/include/test/math.h
@@ -0,0 +1,306 @@
+/*
+ * Compute the natural log of Gamma(x), accurate to 10 decimal places.
+ *
+ * This implementation is based on:
+ *
+ *   Pike, M.C., I.D. Hill (1966) Algorithm 291: Logarithm of Gamma function
+ *   [S14].  Communications of the ACM 9(9):684.
+ */
+static inline double
+ln_gamma(double x) {
+	double f, z;
+
+	assert(x > 0.0);
+
+	if (x < 7.0) {
+		f = 1.0;
+		z = x;
+		while (z < 7.0) {
+			f *= z;
+			z += 1.0;
+		}
+		x = z;
+		f = -log(f);
+	} else {
+		f = 0.0;
+	}
+
+	z = 1.0 / (x * x);
+
+	return f + (x-0.5) * log(x) - x + 0.918938533204673 +
+	    (((-0.000595238095238 * z + 0.000793650793651) * z -
+	    0.002777777777778) * z + 0.083333333333333) / x;
+}
+
+/*
+ * Compute the incomplete Gamma ratio for [0..x], where p is the shape
+ * parameter, and ln_gamma_p is ln_gamma(p).
+ *
+ * This implementation is based on:
+ *
+ *   Bhattacharjee, G.P. (1970) Algorithm AS 32: The incomplete Gamma integral.
+ *   Applied Statistics 19:285-287.
+ */
+static inline double
+i_gamma(double x, double p, double ln_gamma_p) {
+	double acu, factor, oflo, gin, term, rn, a, b, an, dif;
+	double pn[6];
+	unsigned i;
+
+	assert(p > 0.0);
+	assert(x >= 0.0);
+
+	if (x == 0.0) {
+		return 0.0;
+	}
+
+	acu = 1.0e-10;
+	oflo = 1.0e30;
+	gin = 0.0;
+	factor = exp(p * log(x) - x - ln_gamma_p);
+
+	if (x <= 1.0 || x < p) {
+		/* Calculation by series expansion. */
+		gin = 1.0;
+		term = 1.0;
+		rn = p;
+
+		while (true) {
+			rn += 1.0;
+			term *= x / rn;
+			gin += term;
+			if (term <= acu) {
+				gin *= factor / p;
+				return gin;
+			}
+		}
+	} else {
+		/* Calculation by continued fraction. */
+		a = 1.0 - p;
+		b = a + x + 1.0;
+		term = 0.0;
+		pn[0] = 1.0;
+		pn[1] = x;
+		pn[2] = x + 1.0;
+		pn[3] = x * b;
+		gin = pn[2] / pn[3];
+
+		while (true) {
+			a += 1.0;
+			b += 2.0;
+			term += 1.0;
+			an = a * term;
+			for (i = 0; i < 2; i++) {
+				pn[i+4] = b * pn[i+2] - an * pn[i];
+			}
+			if (pn[5] != 0.0) {
+				rn = pn[4] / pn[5];
+				dif = fabs(gin - rn);
+				if (dif <= acu && dif <= acu * rn) {
+					gin = 1.0 - factor * gin;
+					return gin;
+				}
+				gin = rn;
+			}
+			for (i = 0; i < 4; i++) {
+				pn[i] = pn[i+2];
+			}
+
+			if (fabs(pn[4]) >= oflo) {
+				for (i = 0; i < 4; i++) {
+					pn[i] /= oflo;
+				}
+			}
+		}
+	}
+}
+
+/*
+ * Given a value p in [0..1] of the lower tail area of the normal distribution,
+ * compute the limit on the definite integral from [-inf..z] that satisfies p,
+ * accurate to 16 decimal places.
+ *
+ * This implementation is based on:
+ *
+ *   Wichura, M.J. (1988) Algorithm AS 241: The percentage points of the normal
+ *   distribution.  Applied Statistics 37(3):477-484.
+ */
+static inline double
+pt_norm(double p) {
+	double q, r, ret;
+
+	assert(p > 0.0 && p < 1.0);
+
+	q = p - 0.5;
+	if (fabs(q) <= 0.425) {
+		/* p close to 1/2. */
+		r = 0.180625 - q * q;
+		return q * (((((((2.5090809287301226727e3 * r +
+		    3.3430575583588128105e4) * r + 6.7265770927008700853e4) * r
+		    + 4.5921953931549871457e4) * r + 1.3731693765509461125e4) *
+		    r + 1.9715909503065514427e3) * r + 1.3314166789178437745e2)
+		    * r + 3.3871328727963666080e0) /
+		    (((((((5.2264952788528545610e3 * r +
+		    2.8729085735721942674e4) * r + 3.9307895800092710610e4) * r
+		    + 2.1213794301586595867e4) * r + 5.3941960214247511077e3) *
+		    r + 6.8718700749205790830e2) * r + 4.2313330701600911252e1)
+		    * r + 1.0);
+	} else {
+		if (q < 0.0) {
+			r = p;
+		} else {
+			r = 1.0 - p;
+		}
+		assert(r > 0.0);
+
+		r = sqrt(-log(r));
+		if (r <= 5.0) {
+			/* p neither close to 1/2 nor 0 or 1. */
+			r -= 1.6;
+			ret = ((((((((7.74545014278341407640e-4 * r +
+			    2.27238449892691845833e-2) * r +
+			    2.41780725177450611770e-1) * r +
+			    1.27045825245236838258e0) * r +
+			    3.64784832476320460504e0) * r +
+			    5.76949722146069140550e0) * r +
+			    4.63033784615654529590e0) * r +
+			    1.42343711074968357734e0) /
+			    (((((((1.05075007164441684324e-9 * r +
+			    5.47593808499534494600e-4) * r +
+			    1.51986665636164571966e-2)
+			    * r + 1.48103976427480074590e-1) * r +
+			    6.89767334985100004550e-1) * r +
+			    1.67638483018380384940e0) * r +
+			    2.05319162663775882187e0) * r + 1.0));
+		} else {
+			/* p near 0 or 1. */
+			r -= 5.0;
+			ret = ((((((((2.01033439929228813265e-7 * r +
+			    2.71155556874348757815e-5) * r +
+			    1.24266094738807843860e-3) * r +
+			    2.65321895265761230930e-2) * r +
+			    2.96560571828504891230e-1) * r +
+			    1.78482653991729133580e0) * r +
+			    5.46378491116411436990e0) * r +
+			    6.65790464350110377720e0) /
+			    (((((((2.04426310338993978564e-15 * r +
+			    1.42151175831644588870e-7) * r +
+			    1.84631831751005468180e-5) * r +
+			    7.86869131145613259100e-4) * r +
+			    1.48753612908506148525e-2) * r +
+			    1.36929880922735805310e-1) * r +
+			    5.99832206555887937690e-1)
+			    * r + 1.0));
+		}
+		if (q < 0.0) {
+			ret = -ret;
+		}
+		return ret;
+	}
+}
+
+/*
+ * Given a value p in [0..1] of the lower tail area of the Chi^2 distribution
+ * with df degrees of freedom, where ln_gamma_df_2 is ln_gamma(df/2.0), compute
+ * the upper limit on the definite integral from [0..z] that satisfies p,
+ * accurate to 12 decimal places.
+ *
+ * This implementation is based on:
+ *
+ *   Best, D.J., D.E. Roberts (1975) Algorithm AS 91: The percentage points of
+ *   the Chi^2 distribution.  Applied Statistics 24(3):385-388.
+ *
+ *   Shea, B.L. (1991) Algorithm AS R85: A remark on AS 91: The percentage
+ *   points of the Chi^2 distribution.  Applied Statistics 40(1):233-235.
+ */
+static inline double
+pt_chi2(double p, double df, double ln_gamma_df_2) {
+	double e, aa, xx, c, ch, a, q, p1, p2, t, x, b, s1, s2, s3, s4, s5, s6;
+	unsigned i;
+
+	assert(p >= 0.0 && p < 1.0);
+	assert(df > 0.0);
+
+	e = 5.0e-7;
+	aa = 0.6931471805;
+
+	xx = 0.5 * df;
+	c = xx - 1.0;
+
+	if (df < -1.24 * log(p)) {
+		/* Starting approximation for small Chi^2. */
+		ch = pow(p * xx * exp(ln_gamma_df_2 + xx * aa), 1.0 / xx);
+		if (ch - e < 0.0) {
+			return ch;
+		}
+	} else {
+		if (df > 0.32) {
+			x = pt_norm(p);
+			/*
+			 * Starting approximation using Wilson and Hilferty
+			 * estimate.
+			 */
+			p1 = 0.222222 / df;
+			ch = df * pow(x * sqrt(p1) + 1.0 - p1, 3.0);
+			/* Starting approximation for p tending to 1. */
+			if (ch > 2.2 * df + 6.0) {
+				ch = -2.0 * (log(1.0 - p) - c * log(0.5 * ch) +
+				    ln_gamma_df_2);
+			}
+		} else {
+			ch = 0.4;
+			a = log(1.0 - p);
+			while (true) {
+				q = ch;
+				p1 = 1.0 + ch * (4.67 + ch);
+				p2 = ch * (6.73 + ch * (6.66 + ch));
+				t = -0.5 + (4.67 + 2.0 * ch) / p1 - (6.73 + ch
+				    * (13.32 + 3.0 * ch)) / p2;
+				ch -= (1.0 - exp(a + ln_gamma_df_2 + 0.5 * ch +
+				    c * aa) * p2 / p1) / t;
+				if (fabs(q / ch - 1.0) - 0.01 <= 0.0) {
+					break;
+				}
+			}
+		}
+	}
+
+	for (i = 0; i < 20; i++) {
+		/* Calculation of seven-term Taylor series. */
+		q = ch;
+		p1 = 0.5 * ch;
+		if (p1 < 0.0) {
+			return -1.0;
+		}
+		p2 = p - i_gamma(p1, xx, ln_gamma_df_2);
+		t = p2 * exp(xx * aa + ln_gamma_df_2 + p1 - c * log(ch));
+		b = t / ch;
+		a = 0.5 * t - b * c;
+		s1 = (210.0 + a * (140.0 + a * (105.0 + a * (84.0 + a * (70.0 +
+		    60.0 * a))))) / 420.0;
+		s2 = (420.0 + a * (735.0 + a * (966.0 + a * (1141.0 + 1278.0 *
+		    a)))) / 2520.0;
+		s3 = (210.0 + a * (462.0 + a * (707.0 + 932.0 * a))) / 2520.0;
+		s4 = (252.0 + a * (672.0 + 1182.0 * a) + c * (294.0 + a *
+		    (889.0 + 1740.0 * a))) / 5040.0;
+		s5 = (84.0 + 264.0 * a + c * (175.0 + 606.0 * a)) / 2520.0;
+		s6 = (120.0 + c * (346.0 + 127.0 * c)) / 5040.0;
+		ch += t * (1.0 + 0.5 * t * s1 - b * c * (s1 - b * (s2 - b * (s3
+		    - b * (s4 - b * (s5 - b * s6))))));
+		if (fabs(q / ch - 1.0) <= e) {
+			break;
+		}
+	}
+
+	return ch;
+}
+
+/*
+ * Given a value p in [0..1] and Gamma distribution shape and scale parameters,
+ * compute the upper limit on the definite integral from [0..z] that satisfies
+ * p.
+ */
+static inline double
+pt_gamma(double p, double shape, double scale, double ln_gamma_shape) {
+	return pt_chi2(p, shape * 2.0, ln_gamma_shape) * 0.5 * scale;
+}
diff --git a/jemalloc/test/include/test/mq.h b/jemalloc/test/include/test/mq.h
new file mode 100644
index 0000000..af2c078
--- /dev/null
+++ b/jemalloc/test/include/test/mq.h
@@ -0,0 +1,107 @@
+void	mq_nanosleep(unsigned ns);
+
+/*
+ * Simple templated message queue implementation that relies on only mutexes for
+ * synchronization (which reduces portability issues).  Given the following
+ * setup:
+ *
+ *   typedef struct mq_msg_s mq_msg_t;
+ *   struct mq_msg_s {
+ *           mq_msg(mq_msg_t) link;
+ *           [message data]
+ *   };
+ *   mq_gen(, mq_, mq_t, mq_msg_t, link)
+ *
+ * The API is as follows:
+ *
+ *   bool mq_init(mq_t *mq);
+ *   void mq_fini(mq_t *mq);
+ *   unsigned mq_count(mq_t *mq);
+ *   mq_msg_t *mq_tryget(mq_t *mq);
+ *   mq_msg_t *mq_get(mq_t *mq);
+ *   void mq_put(mq_t *mq, mq_msg_t *msg);
+ *
+ * The message queue linkage embedded in each message is to be treated as
+ * externally opaque (no need to initialize or clean up externally).  mq_fini()
+ * does not perform any cleanup of messages, since it knows nothing of their
+ * payloads.
+ */
+#define mq_msg(a_mq_msg_type)	ql_elm(a_mq_msg_type)
+
+#define mq_gen(a_attr, a_prefix, a_mq_type, a_mq_msg_type, a_field)	\
+typedef struct {							\
+	mtx_t			lock;					\
+	ql_head(a_mq_msg_type)	msgs;					\
+	unsigned		count;					\
+} a_mq_type;								\
+a_attr bool								\
+a_prefix##init(a_mq_type *mq) {						\
+									\
+	if (mtx_init(&mq->lock)) {					\
+		return true;						\
+	}								\
+	ql_new(&mq->msgs);						\
+	mq->count = 0;							\
+	return false;							\
+}									\
+a_attr void								\
+a_prefix##fini(a_mq_type *mq) {						\
+	mtx_fini(&mq->lock);						\
+}									\
+a_attr unsigned								\
+a_prefix##count(a_mq_type *mq) {					\
+	unsigned count;							\
+									\
+	mtx_lock(&mq->lock);						\
+	count = mq->count;						\
+	mtx_unlock(&mq->lock);						\
+	return count;							\
+}									\
+a_attr a_mq_msg_type *							\
+a_prefix##tryget(a_mq_type *mq) {					\
+	a_mq_msg_type *msg;						\
+									\
+	mtx_lock(&mq->lock);						\
+	msg = ql_first(&mq->msgs);					\
+	if (msg != NULL) {						\
+		ql_head_remove(&mq->msgs, a_mq_msg_type, a_field);	\
+		mq->count--;						\
+	}								\
+	mtx_unlock(&mq->lock);						\
+	return msg;							\
+}									\
+a_attr a_mq_msg_type *							\
+a_prefix##get(a_mq_type *mq) {						\
+	a_mq_msg_type *msg;						\
+	unsigned ns;							\
+									\
+	msg = a_prefix##tryget(mq);					\
+	if (msg != NULL) {						\
+		return msg;						\
+	}								\
+									\
+	ns = 1;								\
+	while (true) {							\
+		mq_nanosleep(ns);					\
+		msg = a_prefix##tryget(mq);				\
+		if (msg != NULL) {					\
+			return msg;					\
+		}							\
+		if (ns < 1000*1000*1000) {				\
+			/* Double sleep time, up to max 1 second. */	\
+			ns <<= 1;					\
+			if (ns > 1000*1000*1000) {			\
+				ns = 1000*1000*1000;			\
+			}						\
+		}							\
+	}								\
+}									\
+a_attr void								\
+a_prefix##put(a_mq_type *mq, a_mq_msg_type *msg) {			\
+									\
+	mtx_lock(&mq->lock);						\
+	ql_elm_new(msg, a_field);					\
+	ql_tail_insert(&mq->msgs, msg, a_field);			\
+	mq->count++;							\
+	mtx_unlock(&mq->lock);						\
+}
diff --git a/jemalloc/test/include/test/mtx.h b/jemalloc/test/include/test/mtx.h
new file mode 100644
index 0000000..066a213
--- /dev/null
+++ b/jemalloc/test/include/test/mtx.h
@@ -0,0 +1,21 @@
+/*
+ * mtx is a slightly simplified version of malloc_mutex.  This code duplication
+ * is unfortunate, but there are allocator bootstrapping considerations that
+ * would leak into the test infrastructure if malloc_mutex were used directly
+ * in tests.
+ */
+
+typedef struct {
+#ifdef _WIN32
+	CRITICAL_SECTION	lock;
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	os_unfair_lock		lock;
+#else
+	pthread_mutex_t		lock;
+#endif
+} mtx_t;
+
+bool	mtx_init(mtx_t *mtx);
+void	mtx_fini(mtx_t *mtx);
+void	mtx_lock(mtx_t *mtx);
+void	mtx_unlock(mtx_t *mtx);
diff --git a/jemalloc/test/include/test/test.h b/jemalloc/test/include/test/test.h
new file mode 100644
index 0000000..fd0e526
--- /dev/null
+++ b/jemalloc/test/include/test/test.h
@@ -0,0 +1,338 @@
+#define ASSERT_BUFSIZE	256
+
+#define assert_cmp(t, a, b, cmp, neg_cmp, pri, ...) do {		\
+	t a_ = (a);							\
+	t b_ = (b);							\
+	if (!(a_ cmp b_)) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) " #cmp " (%s) --> "				\
+		    "%" pri " " #neg_cmp " %" pri ": ",			\
+		    __func__, __FILE__, __LINE__,			\
+		    #a, #b, a_, b_);					\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+
+#define assert_ptr_eq(a, b, ...)	assert_cmp(void *, a, b, ==,	\
+    !=, "p", __VA_ARGS__)
+#define assert_ptr_ne(a, b, ...)	assert_cmp(void *, a, b, !=,	\
+    ==, "p", __VA_ARGS__)
+#define assert_ptr_null(a, ...)		assert_cmp(void *, a, NULL, ==,	\
+    !=, "p", __VA_ARGS__)
+#define assert_ptr_not_null(a, ...)	assert_cmp(void *, a, NULL, !=,	\
+    ==, "p", __VA_ARGS__)
+
+#define assert_c_eq(a, b, ...)	assert_cmp(char, a, b, ==, !=, "c", __VA_ARGS__)
+#define assert_c_ne(a, b, ...)	assert_cmp(char, a, b, !=, ==, "c", __VA_ARGS__)
+#define assert_c_lt(a, b, ...)	assert_cmp(char, a, b, <, >=, "c", __VA_ARGS__)
+#define assert_c_le(a, b, ...)	assert_cmp(char, a, b, <=, >, "c", __VA_ARGS__)
+#define assert_c_ge(a, b, ...)	assert_cmp(char, a, b, >=, <, "c", __VA_ARGS__)
+#define assert_c_gt(a, b, ...)	assert_cmp(char, a, b, >, <=, "c", __VA_ARGS__)
+
+#define assert_x_eq(a, b, ...)	assert_cmp(int, a, b, ==, !=, "#x", __VA_ARGS__)
+#define assert_x_ne(a, b, ...)	assert_cmp(int, a, b, !=, ==, "#x", __VA_ARGS__)
+#define assert_x_lt(a, b, ...)	assert_cmp(int, a, b, <, >=, "#x", __VA_ARGS__)
+#define assert_x_le(a, b, ...)	assert_cmp(int, a, b, <=, >, "#x", __VA_ARGS__)
+#define assert_x_ge(a, b, ...)	assert_cmp(int, a, b, >=, <, "#x", __VA_ARGS__)
+#define assert_x_gt(a, b, ...)	assert_cmp(int, a, b, >, <=, "#x", __VA_ARGS__)
+
+#define assert_d_eq(a, b, ...)	assert_cmp(int, a, b, ==, !=, "d", __VA_ARGS__)
+#define assert_d_ne(a, b, ...)	assert_cmp(int, a, b, !=, ==, "d", __VA_ARGS__)
+#define assert_d_lt(a, b, ...)	assert_cmp(int, a, b, <, >=, "d", __VA_ARGS__)
+#define assert_d_le(a, b, ...)	assert_cmp(int, a, b, <=, >, "d", __VA_ARGS__)
+#define assert_d_ge(a, b, ...)	assert_cmp(int, a, b, >=, <, "d", __VA_ARGS__)
+#define assert_d_gt(a, b, ...)	assert_cmp(int, a, b, >, <=, "d", __VA_ARGS__)
+
+#define assert_u_eq(a, b, ...)	assert_cmp(int, a, b, ==, !=, "u", __VA_ARGS__)
+#define assert_u_ne(a, b, ...)	assert_cmp(int, a, b, !=, ==, "u", __VA_ARGS__)
+#define assert_u_lt(a, b, ...)	assert_cmp(int, a, b, <, >=, "u", __VA_ARGS__)
+#define assert_u_le(a, b, ...)	assert_cmp(int, a, b, <=, >, "u", __VA_ARGS__)
+#define assert_u_ge(a, b, ...)	assert_cmp(int, a, b, >=, <, "u", __VA_ARGS__)
+#define assert_u_gt(a, b, ...)	assert_cmp(int, a, b, >, <=, "u", __VA_ARGS__)
+
+#define assert_ld_eq(a, b, ...)	assert_cmp(long, a, b, ==,	\
+    !=, "ld", __VA_ARGS__)
+#define assert_ld_ne(a, b, ...)	assert_cmp(long, a, b, !=,	\
+    ==, "ld", __VA_ARGS__)
+#define assert_ld_lt(a, b, ...)	assert_cmp(long, a, b, <,	\
+    >=, "ld", __VA_ARGS__)
+#define assert_ld_le(a, b, ...)	assert_cmp(long, a, b, <=,	\
+    >, "ld", __VA_ARGS__)
+#define assert_ld_ge(a, b, ...)	assert_cmp(long, a, b, >=,	\
+    <, "ld", __VA_ARGS__)
+#define assert_ld_gt(a, b, ...)	assert_cmp(long, a, b, >,	\
+    <=, "ld", __VA_ARGS__)
+
+#define assert_lu_eq(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, ==, !=, "lu", __VA_ARGS__)
+#define assert_lu_ne(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, !=, ==, "lu", __VA_ARGS__)
+#define assert_lu_lt(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, <, >=, "lu", __VA_ARGS__)
+#define assert_lu_le(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, <=, >, "lu", __VA_ARGS__)
+#define assert_lu_ge(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, >=, <, "lu", __VA_ARGS__)
+#define assert_lu_gt(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, >, <=, "lu", __VA_ARGS__)
+
+#define assert_qd_eq(a, b, ...)	assert_cmp(long long, a, b, ==,	\
+    !=, "qd", __VA_ARGS__)
+#define assert_qd_ne(a, b, ...)	assert_cmp(long long, a, b, !=,	\
+    ==, "qd", __VA_ARGS__)
+#define assert_qd_lt(a, b, ...)	assert_cmp(long long, a, b, <,	\
+    >=, "qd", __VA_ARGS__)
+#define assert_qd_le(a, b, ...)	assert_cmp(long long, a, b, <=,	\
+    >, "qd", __VA_ARGS__)
+#define assert_qd_ge(a, b, ...)	assert_cmp(long long, a, b, >=,	\
+    <, "qd", __VA_ARGS__)
+#define assert_qd_gt(a, b, ...)	assert_cmp(long long, a, b, >,	\
+    <=, "qd", __VA_ARGS__)
+
+#define assert_qu_eq(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, ==, !=, "qu", __VA_ARGS__)
+#define assert_qu_ne(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, !=, ==, "qu", __VA_ARGS__)
+#define assert_qu_lt(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, <, >=, "qu", __VA_ARGS__)
+#define assert_qu_le(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, <=, >, "qu", __VA_ARGS__)
+#define assert_qu_ge(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, >=, <, "qu", __VA_ARGS__)
+#define assert_qu_gt(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, >, <=, "qu", __VA_ARGS__)
+
+#define assert_jd_eq(a, b, ...)	assert_cmp(intmax_t, a, b, ==,	\
+    !=, "jd", __VA_ARGS__)
+#define assert_jd_ne(a, b, ...)	assert_cmp(intmax_t, a, b, !=,	\
+    ==, "jd", __VA_ARGS__)
+#define assert_jd_lt(a, b, ...)	assert_cmp(intmax_t, a, b, <,	\
+    >=, "jd", __VA_ARGS__)
+#define assert_jd_le(a, b, ...)	assert_cmp(intmax_t, a, b, <=,	\
+    >, "jd", __VA_ARGS__)
+#define assert_jd_ge(a, b, ...)	assert_cmp(intmax_t, a, b, >=,	\
+    <, "jd", __VA_ARGS__)
+#define assert_jd_gt(a, b, ...)	assert_cmp(intmax_t, a, b, >,	\
+    <=, "jd", __VA_ARGS__)
+
+#define assert_ju_eq(a, b, ...)	assert_cmp(uintmax_t, a, b, ==,	\
+    !=, "ju", __VA_ARGS__)
+#define assert_ju_ne(a, b, ...)	assert_cmp(uintmax_t, a, b, !=,	\
+    ==, "ju", __VA_ARGS__)
+#define assert_ju_lt(a, b, ...)	assert_cmp(uintmax_t, a, b, <,	\
+    >=, "ju", __VA_ARGS__)
+#define assert_ju_le(a, b, ...)	assert_cmp(uintmax_t, a, b, <=,	\
+    >, "ju", __VA_ARGS__)
+#define assert_ju_ge(a, b, ...)	assert_cmp(uintmax_t, a, b, >=,	\
+    <, "ju", __VA_ARGS__)
+#define assert_ju_gt(a, b, ...)	assert_cmp(uintmax_t, a, b, >,	\
+    <=, "ju", __VA_ARGS__)
+
+#define assert_zd_eq(a, b, ...)	assert_cmp(ssize_t, a, b, ==,	\
+    !=, "zd", __VA_ARGS__)
+#define assert_zd_ne(a, b, ...)	assert_cmp(ssize_t, a, b, !=,	\
+    ==, "zd", __VA_ARGS__)
+#define assert_zd_lt(a, b, ...)	assert_cmp(ssize_t, a, b, <,	\
+    >=, "zd", __VA_ARGS__)
+#define assert_zd_le(a, b, ...)	assert_cmp(ssize_t, a, b, <=,	\
+    >, "zd", __VA_ARGS__)
+#define assert_zd_ge(a, b, ...)	assert_cmp(ssize_t, a, b, >=,	\
+    <, "zd", __VA_ARGS__)
+#define assert_zd_gt(a, b, ...)	assert_cmp(ssize_t, a, b, >,	\
+    <=, "zd", __VA_ARGS__)
+
+#define assert_zu_eq(a, b, ...)	assert_cmp(size_t, a, b, ==,	\
+    !=, "zu", __VA_ARGS__)
+#define assert_zu_ne(a, b, ...)	assert_cmp(size_t, a, b, !=,	\
+    ==, "zu", __VA_ARGS__)
+#define assert_zu_lt(a, b, ...)	assert_cmp(size_t, a, b, <,	\
+    >=, "zu", __VA_ARGS__)
+#define assert_zu_le(a, b, ...)	assert_cmp(size_t, a, b, <=,	\
+    >, "zu", __VA_ARGS__)
+#define assert_zu_ge(a, b, ...)	assert_cmp(size_t, a, b, >=,	\
+    <, "zu", __VA_ARGS__)
+#define assert_zu_gt(a, b, ...)	assert_cmp(size_t, a, b, >,	\
+    <=, "zu", __VA_ARGS__)
+
+#define assert_d32_eq(a, b, ...)	assert_cmp(int32_t, a, b, ==,	\
+    !=, FMTd32, __VA_ARGS__)
+#define assert_d32_ne(a, b, ...)	assert_cmp(int32_t, a, b, !=,	\
+    ==, FMTd32, __VA_ARGS__)
+#define assert_d32_lt(a, b, ...)	assert_cmp(int32_t, a, b, <,	\
+    >=, FMTd32, __VA_ARGS__)
+#define assert_d32_le(a, b, ...)	assert_cmp(int32_t, a, b, <=,	\
+    >, FMTd32, __VA_ARGS__)
+#define assert_d32_ge(a, b, ...)	assert_cmp(int32_t, a, b, >=,	\
+    <, FMTd32, __VA_ARGS__)
+#define assert_d32_gt(a, b, ...)	assert_cmp(int32_t, a, b, >,	\
+    <=, FMTd32, __VA_ARGS__)
+
+#define assert_u32_eq(a, b, ...)	assert_cmp(uint32_t, a, b, ==,	\
+    !=, FMTu32, __VA_ARGS__)
+#define assert_u32_ne(a, b, ...)	assert_cmp(uint32_t, a, b, !=,	\
+    ==, FMTu32, __VA_ARGS__)
+#define assert_u32_lt(a, b, ...)	assert_cmp(uint32_t, a, b, <,	\
+    >=, FMTu32, __VA_ARGS__)
+#define assert_u32_le(a, b, ...)	assert_cmp(uint32_t, a, b, <=,	\
+    >, FMTu32, __VA_ARGS__)
+#define assert_u32_ge(a, b, ...)	assert_cmp(uint32_t, a, b, >=,	\
+    <, FMTu32, __VA_ARGS__)
+#define assert_u32_gt(a, b, ...)	assert_cmp(uint32_t, a, b, >,	\
+    <=, FMTu32, __VA_ARGS__)
+
+#define assert_d64_eq(a, b, ...)	assert_cmp(int64_t, a, b, ==,	\
+    !=, FMTd64, __VA_ARGS__)
+#define assert_d64_ne(a, b, ...)	assert_cmp(int64_t, a, b, !=,	\
+    ==, FMTd64, __VA_ARGS__)
+#define assert_d64_lt(a, b, ...)	assert_cmp(int64_t, a, b, <,	\
+    >=, FMTd64, __VA_ARGS__)
+#define assert_d64_le(a, b, ...)	assert_cmp(int64_t, a, b, <=,	\
+    >, FMTd64, __VA_ARGS__)
+#define assert_d64_ge(a, b, ...)	assert_cmp(int64_t, a, b, >=,	\
+    <, FMTd64, __VA_ARGS__)
+#define assert_d64_gt(a, b, ...)	assert_cmp(int64_t, a, b, >,	\
+    <=, FMTd64, __VA_ARGS__)
+
+#define assert_u64_eq(a, b, ...)	assert_cmp(uint64_t, a, b, ==,	\
+    !=, FMTu64, __VA_ARGS__)
+#define assert_u64_ne(a, b, ...)	assert_cmp(uint64_t, a, b, !=,	\
+    ==, FMTu64, __VA_ARGS__)
+#define assert_u64_lt(a, b, ...)	assert_cmp(uint64_t, a, b, <,	\
+    >=, FMTu64, __VA_ARGS__)
+#define assert_u64_le(a, b, ...)	assert_cmp(uint64_t, a, b, <=,	\
+    >, FMTu64, __VA_ARGS__)
+#define assert_u64_ge(a, b, ...)	assert_cmp(uint64_t, a, b, >=,	\
+    <, FMTu64, __VA_ARGS__)
+#define assert_u64_gt(a, b, ...)	assert_cmp(uint64_t, a, b, >,	\
+    <=, FMTu64, __VA_ARGS__)
+
+#define assert_b_eq(a, b, ...) do {					\
+	bool a_ = (a);							\
+	bool b_ = (b);							\
+	if (!(a_ == b_)) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) == (%s) --> %s != %s: ",			\
+		    __func__, __FILE__, __LINE__,			\
+		    #a, #b, a_ ? "true" : "false",			\
+		    b_ ? "true" : "false");				\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+#define assert_b_ne(a, b, ...) do {					\
+	bool a_ = (a);							\
+	bool b_ = (b);							\
+	if (!(a_ != b_)) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) != (%s) --> %s == %s: ",			\
+		    __func__, __FILE__, __LINE__,			\
+		    #a, #b, a_ ? "true" : "false",			\
+		    b_ ? "true" : "false");				\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+#define assert_true(a, ...)	assert_b_eq(a, true, __VA_ARGS__)
+#define assert_false(a, ...)	assert_b_eq(a, false, __VA_ARGS__)
+
+#define assert_str_eq(a, b, ...) do {				\
+	if (strcmp((a), (b))) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) same as (%s) --> "				\
+		    "\"%s\" differs from \"%s\": ",			\
+		    __func__, __FILE__, __LINE__, #a, #b, a, b);	\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+#define assert_str_ne(a, b, ...) do {				\
+	if (!strcmp((a), (b))) {					\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) differs from (%s) --> "			\
+		    "\"%s\" same as \"%s\": ",				\
+		    __func__, __FILE__, __LINE__, #a, #b, a, b);	\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+
+#define assert_not_reached(...) do {					\
+	char prefix[ASSERT_BUFSIZE];					\
+	char message[ASSERT_BUFSIZE];					\
+	malloc_snprintf(prefix, sizeof(prefix),				\
+	    "%s:%s:%d: Unreachable code reached: ",			\
+	    __func__, __FILE__, __LINE__);				\
+	malloc_snprintf(message, sizeof(message), __VA_ARGS__);		\
+	p_test_fail(prefix, message);					\
+} while (0)
+
+/*
+ * If this enum changes, corresponding changes in test/test.sh.in are also
+ * necessary.
+ */
+typedef enum {
+	test_status_pass = 0,
+	test_status_skip = 1,
+	test_status_fail = 2,
+
+	test_status_count = 3
+} test_status_t;
+
+typedef void (test_t)(void);
+
+#define TEST_BEGIN(f)							\
+static void								\
+f(void) {								\
+	p_test_init(#f);
+
+#define TEST_END							\
+	goto label_test_end;						\
+label_test_end:								\
+	p_test_fini();							\
+}
+
+#define test(...)							\
+	p_test(__VA_ARGS__, NULL)
+
+#define test_no_reentrancy(...)							\
+	p_test_no_reentrancy(__VA_ARGS__, NULL)
+
+#define test_no_malloc_init(...)					\
+	p_test_no_malloc_init(__VA_ARGS__, NULL)
+
+#define test_skip_if(e) do {						\
+	if (e) {							\
+		test_skip("%s:%s:%d: Test skipped: (%s)",		\
+		    __func__, __FILE__, __LINE__, #e);			\
+		goto label_test_end;					\
+	}								\
+} while (0)
+
+bool test_is_reentrant();
+
+void	test_skip(const char *format, ...) JEMALLOC_FORMAT_PRINTF(1, 2);
+void	test_fail(const char *format, ...) JEMALLOC_FORMAT_PRINTF(1, 2);
+
+/* For private use by macros. */
+test_status_t	p_test(test_t *t, ...);
+test_status_t	p_test_no_reentrancy(test_t *t, ...);
+test_status_t	p_test_no_malloc_init(test_t *t, ...);
+void	p_test_init(const char *name);
+void	p_test_fini(void);
+void	p_test_fail(const char *prefix, const char *message);
diff --git a/jemalloc/test/include/test/thd.h b/jemalloc/test/include/test/thd.h
new file mode 100644
index 0000000..47a5126
--- /dev/null
+++ b/jemalloc/test/include/test/thd.h
@@ -0,0 +1,9 @@
+/* Abstraction layer for threading in tests. */
+#ifdef _WIN32
+typedef HANDLE thd_t;
+#else
+typedef pthread_t thd_t;
+#endif
+
+void	thd_create(thd_t *thd, void *(*proc)(void *), void *arg);
+void	thd_join(thd_t thd, void **ret);
diff --git a/jemalloc/test/include/test/timer.h b/jemalloc/test/include/test/timer.h
new file mode 100644
index 0000000..ace6191
--- /dev/null
+++ b/jemalloc/test/include/test/timer.h
@@ -0,0 +1,11 @@
+/* Simple timer, for use in benchmark reporting. */
+
+typedef struct {
+	nstime_t t0;
+	nstime_t t1;
+} timedelta_t;
+
+void	timer_start(timedelta_t *timer);
+void	timer_stop(timedelta_t *timer);
+uint64_t	timer_usec(const timedelta_t *timer);
+void	timer_ratio(timedelta_t *a, timedelta_t *b, char *buf, size_t buflen);
diff --git a/jemalloc/test/integration/MALLOCX_ARENA.c b/jemalloc/test/integration/MALLOCX_ARENA.c
new file mode 100644
index 0000000..222164d
--- /dev/null
+++ b/jemalloc/test/integration/MALLOCX_ARENA.c
@@ -0,0 +1,66 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS 10
+
+static bool have_dss =
+#ifdef JEMALLOC_DSS
+    true
+#else
+    false
+#endif
+    ;
+
+void *
+thd_start(void *arg) {
+	unsigned thread_ind = (unsigned)(uintptr_t)arg;
+	unsigned arena_ind;
+	void *p;
+	size_t sz;
+
+	sz = sizeof(arena_ind);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Error in arenas.create");
+
+	if (thread_ind % 4 != 3) {
+		size_t mib[3];
+		size_t miblen = sizeof(mib) / sizeof(size_t);
+		const char *dss_precs[] = {"disabled", "primary", "secondary"};
+		unsigned prec_ind = thread_ind %
+		    (sizeof(dss_precs)/sizeof(char*));
+		const char *dss = dss_precs[prec_ind];
+		int expected_err = (have_dss || prec_ind == 0) ? 0 : EFAULT;
+		assert_d_eq(mallctlnametomib("arena.0.dss", mib, &miblen), 0,
+		    "Error in mallctlnametomib()");
+		mib[1] = arena_ind;
+		assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, (void *)&dss,
+		    sizeof(const char *)), expected_err,
+		    "Error in mallctlbymib()");
+	}
+
+	p = mallocx(1, MALLOCX_ARENA(arena_ind));
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	dallocx(p, 0);
+
+	return NULL;
+}
+
+TEST_BEGIN(test_MALLOCX_ARENA) {
+	thd_t thds[NTHREADS];
+	unsigned i;
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start,
+		    (void *)(uintptr_t)i);
+	}
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_MALLOCX_ARENA);
+}
diff --git a/jemalloc/test/integration/aligned_alloc.c b/jemalloc/test/integration/aligned_alloc.c
new file mode 100644
index 0000000..4375b17
--- /dev/null
+++ b/jemalloc/test/integration/aligned_alloc.c
@@ -0,0 +1,157 @@
+#include "test/jemalloc_test.h"
+
+#define MAXALIGN (((size_t)1) << 23)
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+TEST_BEGIN(test_alignment_errors) {
+	size_t alignment;
+	void *p;
+
+	alignment = 0;
+	set_errno(0);
+	p = aligned_alloc(alignment, 1);
+	assert_false(p != NULL || get_errno() != EINVAL,
+	    "Expected error for invalid alignment %zu", alignment);
+
+	for (alignment = sizeof(size_t); alignment < MAXALIGN;
+	    alignment <<= 1) {
+		set_errno(0);
+		p = aligned_alloc(alignment + 1, 1);
+		assert_false(p != NULL || get_errno() != EINVAL,
+		    "Expected error for invalid alignment %zu",
+		    alignment + 1);
+	}
+}
+TEST_END
+
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_oom_errors) {
+	size_t alignment, size;
+	void *p;
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x8000000000000000);
+	size      = UINT64_C(0x8000000000000000);
+#else
+	alignment = 0x80000000LU;
+	size      = 0x80000000LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	assert_false(p != NULL || get_errno() != ENOMEM,
+	    "Expected error for aligned_alloc(%zu, %zu)",
+	    alignment, size);
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x4000000000000000);
+	size      = UINT64_C(0xc000000000000001);
+#else
+	alignment = 0x40000000LU;
+	size      = 0xc0000001LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	assert_false(p != NULL || get_errno() != ENOMEM,
+	    "Expected error for aligned_alloc(%zu, %zu)",
+	    alignment, size);
+
+	alignment = 0x10LU;
+#if LG_SIZEOF_PTR == 3
+	size = UINT64_C(0xfffffffffffffff0);
+#else
+	size = 0xfffffff0LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	assert_false(p != NULL || get_errno() != ENOMEM,
+	    "Expected error for aligned_alloc(&p, %zu, %zu)",
+	    alignment, size);
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+TEST_BEGIN(test_alignment_and_size) {
+#define NITER 4
+	size_t alignment, size, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (size = 1;
+		    size < 3 * alignment && size < (1U << 31);
+		    size += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				ps[i] = aligned_alloc(alignment, size);
+				if (ps[i] == NULL) {
+					char buf[BUFERROR_BUF];
+
+					buferror(get_errno(), buf, sizeof(buf));
+					test_fail(
+					    "Error for alignment=%zu, "
+					    "size=%zu (%#zx): %s",
+					    alignment, size, size, buf);
+				}
+				total += malloc_usable_size(ps[i]);
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					free(ps[i]);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef NITER
+}
+TEST_END
+
+TEST_BEGIN(test_zero_alloc) {
+	void *res = aligned_alloc(8, 0);
+	assert(res);
+	size_t usable = malloc_usable_size(res);
+	assert(usable > 0);
+	free(res);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_alignment_errors,
+	    test_oom_errors,
+	    test_alignment_and_size,
+	    test_zero_alloc);
+}
diff --git a/jemalloc/test/integration/allocated.c b/jemalloc/test/integration/allocated.c
new file mode 100644
index 0000000..1425fd0
--- /dev/null
+++ b/jemalloc/test/integration/allocated.c
@@ -0,0 +1,124 @@
+#include "test/jemalloc_test.h"
+
+static const bool config_stats =
+#ifdef JEMALLOC_STATS
+    true
+#else
+    false
+#endif
+    ;
+
+void *
+thd_start(void *arg) {
+	int err;
+	void *p;
+	uint64_t a0, a1, d0, d1;
+	uint64_t *ap0, *ap1, *dp0, *dp1;
+	size_t sz, usize;
+
+	sz = sizeof(a0);
+	if ((err = mallctl("thread.allocated", (void *)&a0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	sz = sizeof(ap0);
+	if ((err = mallctl("thread.allocatedp", (void *)&ap0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	assert_u64_eq(*ap0, a0,
+	    "\"thread.allocatedp\" should provide a pointer to internal "
+	    "storage");
+
+	sz = sizeof(d0);
+	if ((err = mallctl("thread.deallocated", (void *)&d0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	sz = sizeof(dp0);
+	if ((err = mallctl("thread.deallocatedp", (void *)&dp0, &sz, NULL,
+	    0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	assert_u64_eq(*dp0, d0,
+	    "\"thread.deallocatedp\" should provide a pointer to internal "
+	    "storage");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() error");
+
+	sz = sizeof(a1);
+	mallctl("thread.allocated", (void *)&a1, &sz, NULL, 0);
+	sz = sizeof(ap1);
+	mallctl("thread.allocatedp", (void *)&ap1, &sz, NULL, 0);
+	assert_u64_eq(*ap1, a1,
+	    "Dereferenced \"thread.allocatedp\" value should equal "
+	    "\"thread.allocated\" value");
+	assert_ptr_eq(ap0, ap1,
+	    "Pointer returned by \"thread.allocatedp\" should not change");
+
+	usize = malloc_usable_size(p);
+	assert_u64_le(a0 + usize, a1,
+	    "Allocated memory counter should increase by at least the amount "
+	    "explicitly allocated");
+
+	free(p);
+
+	sz = sizeof(d1);
+	mallctl("thread.deallocated", (void *)&d1, &sz, NULL, 0);
+	sz = sizeof(dp1);
+	mallctl("thread.deallocatedp", (void *)&dp1, &sz, NULL, 0);
+	assert_u64_eq(*dp1, d1,
+	    "Dereferenced \"thread.deallocatedp\" value should equal "
+	    "\"thread.deallocated\" value");
+	assert_ptr_eq(dp0, dp1,
+	    "Pointer returned by \"thread.deallocatedp\" should not change");
+
+	assert_u64_le(d0 + usize, d1,
+	    "Deallocated memory counter should increase by at least the amount "
+	    "explicitly deallocated");
+
+	return NULL;
+label_ENOENT:
+	assert_false(config_stats,
+	    "ENOENT should only be returned if stats are disabled");
+	test_skip("\"thread.allocated\" mallctl not available");
+	return NULL;
+}
+
+TEST_BEGIN(test_main_thread) {
+	thd_start(NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_subthread) {
+	thd_t thd;
+
+	thd_create(&thd, thd_start, NULL);
+	thd_join(thd, NULL);
+}
+TEST_END
+
+int
+main(void) {
+	/* Run tests multiple times to check for bad interactions. */
+	return test(
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread);
+}
diff --git a/jemalloc/test/integration/extent.c b/jemalloc/test/integration/extent.c
new file mode 100644
index 0000000..b5db087
--- /dev/null
+++ b/jemalloc/test/integration/extent.c
@@ -0,0 +1,248 @@
+#include "test/jemalloc_test.h"
+
+#include "test/extent_hooks.h"
+
+static bool
+check_background_thread_enabled(void) {
+	bool enabled;
+	size_t sz = sizeof(bool);
+	int ret = mallctl("background_thread", (void *)&enabled, &sz, NULL,0);
+	if (ret == ENOENT) {
+		return false;
+	}
+	assert_d_eq(ret, 0, "Unexpected mallctl error");
+	return enabled;
+}
+
+static void
+test_extent_body(unsigned arena_ind) {
+	void *p;
+	size_t large0, large1, large2, sz;
+	size_t purge_mib[3];
+	size_t purge_miblen;
+	int flags;
+	bool xallocx_success_a, xallocx_success_b, xallocx_success_c;
+
+	flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+
+	/* Get large size classes. */
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large0, &sz, NULL,
+	    0), 0, "Unexpected arenas.lextent.0.size failure");
+	assert_d_eq(mallctl("arenas.lextent.1.size", (void *)&large1, &sz, NULL,
+	    0), 0, "Unexpected arenas.lextent.1.size failure");
+	assert_d_eq(mallctl("arenas.lextent.2.size", (void *)&large2, &sz, NULL,
+	    0), 0, "Unexpected arenas.lextent.2.size failure");
+
+	/* Test dalloc/decommit/purge cascade. */
+	purge_miblen = sizeof(purge_mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.purge", purge_mib, &purge_miblen),
+	    0, "Unexpected mallctlnametomib() failure");
+	purge_mib[1] = (size_t)arena_ind;
+	called_alloc = false;
+	try_alloc = true;
+	try_dalloc = false;
+	try_decommit = false;
+	p = mallocx(large0 * 2, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	assert_true(called_alloc, "Expected alloc call");
+	called_dalloc = false;
+	called_decommit = false;
+	did_purge_lazy = false;
+	did_purge_forced = false;
+	called_split = false;
+	xallocx_success_a = (xallocx(p, large0, 0, flags) == large0);
+	assert_d_eq(mallctlbymib(purge_mib, purge_miblen, NULL, NULL, NULL, 0),
+	    0, "Unexpected arena.%u.purge error", arena_ind);
+	if (xallocx_success_a) {
+		assert_true(called_dalloc, "Expected dalloc call");
+		assert_true(called_decommit, "Expected decommit call");
+		assert_true(did_purge_lazy || did_purge_forced,
+		    "Expected purge");
+	}
+	assert_true(called_split, "Expected split call");
+	dallocx(p, flags);
+	try_dalloc = true;
+
+	/* Test decommit/commit and observe split/merge. */
+	try_dalloc = false;
+	try_decommit = true;
+	p = mallocx(large0 * 2, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	did_decommit = false;
+	did_commit = false;
+	called_split = false;
+	did_split = false;
+	did_merge = false;
+	xallocx_success_b = (xallocx(p, large0, 0, flags) == large0);
+	assert_d_eq(mallctlbymib(purge_mib, purge_miblen, NULL, NULL, NULL, 0),
+	    0, "Unexpected arena.%u.purge error", arena_ind);
+	if (xallocx_success_b) {
+		assert_true(did_split, "Expected split");
+	}
+	xallocx_success_c = (xallocx(p, large0 * 2, 0, flags) == large0 * 2);
+	if (did_split) {
+		assert_b_eq(did_decommit, did_commit,
+		    "Expected decommit/commit match");
+	}
+	if (xallocx_success_b && xallocx_success_c) {
+		assert_true(did_merge, "Expected merge");
+	}
+	dallocx(p, flags);
+	try_dalloc = true;
+	try_decommit = false;
+
+	/* Make sure non-large allocation succeeds. */
+	p = mallocx(42, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	dallocx(p, flags);
+}
+
+static void
+test_manual_hook_auto_arena(void) {
+	unsigned narenas;
+	size_t old_size, new_size, sz;
+	size_t hooks_mib[3];
+	size_t hooks_miblen;
+	extent_hooks_t *new_hooks, *old_hooks;
+
+	extent_hooks_prep();
+
+	sz = sizeof(unsigned);
+	/* Get number of auto arenas. */
+	assert_d_eq(mallctl("opt.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	if (narenas == 1) {
+		return;
+	}
+
+	/* Install custom extent hooks on arena 1 (might not be initialized). */
+	hooks_miblen = sizeof(hooks_mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.extent_hooks", hooks_mib,
+	    &hooks_miblen), 0, "Unexpected mallctlnametomib() failure");
+	hooks_mib[1] = 1;
+	old_size = sizeof(extent_hooks_t *);
+	new_hooks = &hooks;
+	new_size = sizeof(extent_hooks_t *);
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, (void *)&old_hooks,
+	    &old_size, (void *)&new_hooks, new_size), 0,
+	    "Unexpected extent_hooks error");
+	static bool auto_arena_created = false;
+	if (old_hooks != &hooks) {
+		assert_b_eq(auto_arena_created, false,
+		    "Expected auto arena 1 created only once.");
+		auto_arena_created = true;
+	}
+}
+
+static void
+test_manual_hook_body(void) {
+	unsigned arena_ind;
+	size_t old_size, new_size, sz;
+	size_t hooks_mib[3];
+	size_t hooks_miblen;
+	extent_hooks_t *new_hooks, *old_hooks;
+
+	extent_hooks_prep();
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+
+	/* Install custom extent hooks. */
+	hooks_miblen = sizeof(hooks_mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.extent_hooks", hooks_mib,
+	    &hooks_miblen), 0, "Unexpected mallctlnametomib() failure");
+	hooks_mib[1] = (size_t)arena_ind;
+	old_size = sizeof(extent_hooks_t *);
+	new_hooks = &hooks;
+	new_size = sizeof(extent_hooks_t *);
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, (void *)&old_hooks,
+	    &old_size, (void *)&new_hooks, new_size), 0,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->alloc, extent_alloc_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->dalloc, extent_dalloc_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->commit, extent_commit_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->decommit, extent_decommit_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->purge_lazy, extent_purge_lazy_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->purge_forced, extent_purge_forced_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->split, extent_split_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->merge, extent_merge_hook,
+	    "Unexpected extent_hooks error");
+
+	if (!check_background_thread_enabled()) {
+		test_extent_body(arena_ind);
+	}
+
+	/* Restore extent hooks. */
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, NULL, NULL,
+	    (void *)&old_hooks, new_size), 0, "Unexpected extent_hooks error");
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, (void *)&old_hooks,
+	    &old_size, NULL, 0), 0, "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks, default_hooks, "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->alloc, default_hooks->alloc,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->dalloc, default_hooks->dalloc,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->commit, default_hooks->commit,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->decommit, default_hooks->decommit,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->purge_lazy, default_hooks->purge_lazy,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->purge_forced, default_hooks->purge_forced,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->split, default_hooks->split,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->merge, default_hooks->merge,
+	    "Unexpected extent_hooks error");
+}
+
+TEST_BEGIN(test_extent_manual_hook) {
+	test_manual_hook_auto_arena();
+	test_manual_hook_body();
+
+	/* Test failure paths. */
+	try_split = false;
+	test_manual_hook_body();
+	try_merge = false;
+	test_manual_hook_body();
+	try_purge_lazy = false;
+	try_purge_forced = false;
+	test_manual_hook_body();
+
+	try_split = try_merge = try_purge_lazy = try_purge_forced = true;
+}
+TEST_END
+
+TEST_BEGIN(test_extent_auto_hook) {
+	unsigned arena_ind;
+	size_t new_size, sz;
+	extent_hooks_t *new_hooks;
+
+	extent_hooks_prep();
+
+	sz = sizeof(unsigned);
+	new_hooks = &hooks;
+	new_size = sizeof(extent_hooks_t *);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz,
+	    (void *)&new_hooks, new_size), 0, "Unexpected mallctl() failure");
+
+	test_skip_if(check_background_thread_enabled());
+	test_extent_body(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_extent_manual_hook,
+	    test_extent_auto_hook);
+}
diff --git a/jemalloc/test/integration/extent.sh b/jemalloc/test/integration/extent.sh
new file mode 100644
index 0000000..0cc2187
--- /dev/null
+++ b/jemalloc/test/integration/extent.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="junk:false"
+fi
diff --git a/jemalloc/test/integration/malloc.c b/jemalloc/test/integration/malloc.c
new file mode 100644
index 0000000..8b33bc8
--- /dev/null
+++ b/jemalloc/test/integration/malloc.c
@@ -0,0 +1,16 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_zero_alloc) {
+	void *res = malloc(0);
+	assert(res);
+	size_t usable = malloc_usable_size(res);
+	assert(usable > 0);
+	free(res);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_zero_alloc);
+}
diff --git a/jemalloc/test/integration/mallocx.c b/jemalloc/test/integration/mallocx.c
new file mode 100644
index 0000000..645d4db
--- /dev/null
+++ b/jemalloc/test/integration/mallocx.c
@@ -0,0 +1,274 @@
+#include "test/jemalloc_test.h"
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	size_t largemax;
+
+	largemax = get_large_size(get_nlarge()-1);
+
+	assert_ptr_null(mallocx(largemax+1, 0),
+	    "Expected OOM for mallocx(size=%#zx, 0)", largemax+1);
+
+	assert_ptr_null(mallocx(ZU(PTRDIFF_MAX)+1, 0),
+	    "Expected OOM for mallocx(size=%#zx, 0)", ZU(PTRDIFF_MAX)+1);
+
+	assert_ptr_null(mallocx(SIZE_T_MAX, 0),
+	    "Expected OOM for mallocx(size=%#zx, 0)", SIZE_T_MAX);
+
+	assert_ptr_null(mallocx(1, MALLOCX_ALIGN(ZU(PTRDIFF_MAX)+1)),
+	    "Expected OOM for mallocx(size=1, MALLOCX_ALIGN(%#zx))",
+	    ZU(PTRDIFF_MAX)+1);
+}
+TEST_END
+
+static void *
+remote_alloc(void *arg) {
+	unsigned arena;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	size_t large_sz;
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large_sz, &sz,
+	    NULL, 0), 0, "Unexpected mallctl failure");
+
+	void *ptr = mallocx(large_sz, MALLOCX_ARENA(arena)
+	    | MALLOCX_TCACHE_NONE);
+	void **ret = (void **)arg;
+	*ret = ptr;
+
+	return NULL;
+}
+
+TEST_BEGIN(test_remote_free) {
+	thd_t thd;
+	void *ret;
+	thd_create(&thd, remote_alloc, (void *)&ret);
+	thd_join(thd, NULL);
+	assert_ptr_not_null(ret, "Unexpected mallocx failure");
+
+	/* Avoid TCACHE_NONE to explicitly test tcache_flush(). */
+	dallocx(ret, 0);
+	mallctl("thread.tcache.flush", NULL, NULL, NULL, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_oom) {
+	size_t largemax;
+	bool oom;
+	void *ptrs[3];
+	unsigned i;
+
+	/*
+	 * It should be impossible to allocate three objects that each consume
+	 * nearly half the virtual address space.
+	 */
+	largemax = get_large_size(get_nlarge()-1);
+	oom = false;
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		ptrs[i] = mallocx(largemax, MALLOCX_ARENA(0));
+		if (ptrs[i] == NULL) {
+			oom = true;
+		}
+	}
+	assert_true(oom,
+	    "Expected OOM during series of calls to mallocx(size=%zu, 0)",
+	    largemax);
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		if (ptrs[i] != NULL) {
+			dallocx(ptrs[i], 0);
+		}
+	}
+	purge();
+
+#if LG_SIZEOF_PTR == 3
+	assert_ptr_null(mallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x8000000000000000ULL)),
+	    "Expected OOM for mallocx()");
+	assert_ptr_null(mallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x80000000)),
+	    "Expected OOM for mallocx()");
+#else
+	assert_ptr_null(mallocx(0x80000000UL, MALLOCX_ALIGN(0x80000000UL)),
+	    "Expected OOM for mallocx()");
+#endif
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+TEST_BEGIN(test_basic) {
+#define MAXSZ (((size_t)1) << 23)
+	size_t sz;
+
+	for (sz = 1; sz < MAXSZ; sz = nallocx(sz, 0) + 1) {
+		size_t nsz, rsz;
+		void *p;
+		nsz = nallocx(sz, 0);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		p = mallocx(sz, 0);
+		assert_ptr_not_null(p,
+		    "Unexpected mallocx(size=%zx, flags=0) error", sz);
+		rsz = sallocx(p, 0);
+		assert_zu_ge(rsz, sz, "Real size smaller than expected");
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() size mismatch");
+		dallocx(p, 0);
+
+		p = mallocx(sz, 0);
+		assert_ptr_not_null(p,
+		    "Unexpected mallocx(size=%zx, flags=0) error", sz);
+		dallocx(p, 0);
+
+		nsz = nallocx(sz, MALLOCX_ZERO);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		p = mallocx(sz, MALLOCX_ZERO);
+		assert_ptr_not_null(p,
+		    "Unexpected mallocx(size=%zx, flags=MALLOCX_ZERO) error",
+		    nsz);
+		rsz = sallocx(p, 0);
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() rsize mismatch");
+		dallocx(p, 0);
+		purge();
+	}
+#undef MAXSZ
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+	const char *percpu_arena;
+	size_t sz = sizeof(percpu_arena);
+
+	if(mallctl("opt.percpu_arena", (void *)&percpu_arena, &sz, NULL, 0) ||
+	    strcmp(percpu_arena, "disabled") != 0) {
+		test_skip("test_alignment_and_size skipped: "
+		    "not working with percpu arena.");
+	};
+#define MAXALIGN (((size_t)1) << 23)
+#define NITER 4
+	size_t nsz, rsz, alignment, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (sz = 1;
+		    sz < 3 * alignment && sz < (1U << 31);
+		    sz += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				nsz = nallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO | MALLOCX_ARENA(0));
+				assert_zu_ne(nsz, 0,
+				    "nallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				ps[i] = mallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO | MALLOCX_ARENA(0));
+				assert_ptr_not_null(ps[i],
+				    "mallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				rsz = sallocx(ps[i], 0);
+				assert_zu_ge(rsz, sz,
+				    "Real size smaller than expected for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_zu_eq(nsz, rsz,
+				    "nallocx()/sallocx() size mismatch for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_ptr_null(
+				    (void *)((uintptr_t)ps[i] & (alignment-1)),
+				    "%p inadequately aligned for"
+				    " alignment=%zu, size=%zu", ps[i],
+				    alignment, sz);
+				total += rsz;
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					dallocx(ps[i], 0);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef MAXALIGN
+#undef NITER
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_overflow,
+	    test_oom,
+	    test_remote_free,
+	    test_basic,
+	    test_alignment_and_size);
+}
diff --git a/jemalloc/test/integration/mallocx.sh b/jemalloc/test/integration/mallocx.sh
new file mode 100644
index 0000000..0cc2187
--- /dev/null
+++ b/jemalloc/test/integration/mallocx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="junk:false"
+fi
diff --git a/jemalloc/test/integration/overflow.c b/jemalloc/test/integration/overflow.c
new file mode 100644
index 0000000..748ebb6
--- /dev/null
+++ b/jemalloc/test/integration/overflow.c
@@ -0,0 +1,59 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	unsigned nlextents;
+	size_t mib[4];
+	size_t sz, miblen, max_size_class;
+	void *p;
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nlextents", (void *)&nlextents, &sz, NULL,
+	    0), 0, "Unexpected mallctl() error");
+
+	miblen = sizeof(mib) / sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arenas.lextent.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() error");
+	mib[2] = nlextents - 1;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&max_size_class, &sz,
+	    NULL, 0), 0, "Unexpected mallctlbymib() error");
+
+	assert_ptr_null(malloc(max_size_class + 1),
+	    "Expected OOM due to over-sized allocation request");
+	assert_ptr_null(malloc(SIZE_T_MAX),
+	    "Expected OOM due to over-sized allocation request");
+
+	assert_ptr_null(calloc(1, max_size_class + 1),
+	    "Expected OOM due to over-sized allocation request");
+	assert_ptr_null(calloc(1, SIZE_T_MAX),
+	    "Expected OOM due to over-sized allocation request");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() OOM");
+	assert_ptr_null(realloc(p, max_size_class + 1),
+	    "Expected OOM due to over-sized allocation request");
+	assert_ptr_null(realloc(p, SIZE_T_MAX),
+	    "Expected OOM due to over-sized allocation request");
+	free(p);
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+int
+main(void) {
+	return test(
+	    test_overflow);
+}
diff --git a/jemalloc/test/integration/posix_memalign.c b/jemalloc/test/integration/posix_memalign.c
new file mode 100644
index 0000000..d992260
--- /dev/null
+++ b/jemalloc/test/integration/posix_memalign.c
@@ -0,0 +1,128 @@
+#include "test/jemalloc_test.h"
+
+#define MAXALIGN (((size_t)1) << 23)
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+TEST_BEGIN(test_alignment_errors) {
+	size_t alignment;
+	void *p;
+
+	for (alignment = 0; alignment < sizeof(void *); alignment++) {
+		assert_d_eq(posix_memalign(&p, alignment, 1), EINVAL,
+		    "Expected error for invalid alignment %zu",
+		    alignment);
+	}
+
+	for (alignment = sizeof(size_t); alignment < MAXALIGN;
+	    alignment <<= 1) {
+		assert_d_ne(posix_memalign(&p, alignment + 1, 1), 0,
+		    "Expected error for invalid alignment %zu",
+		    alignment + 1);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_oom_errors) {
+	size_t alignment, size;
+	void *p;
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x8000000000000000);
+	size      = UINT64_C(0x8000000000000000);
+#else
+	alignment = 0x80000000LU;
+	size      = 0x80000000LU;
+#endif
+	assert_d_ne(posix_memalign(&p, alignment, size), 0,
+	    "Expected error for posix_memalign(&p, %zu, %zu)",
+	    alignment, size);
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x4000000000000000);
+	size      = UINT64_C(0xc000000000000001);
+#else
+	alignment = 0x40000000LU;
+	size      = 0xc0000001LU;
+#endif
+	assert_d_ne(posix_memalign(&p, alignment, size), 0,
+	    "Expected error for posix_memalign(&p, %zu, %zu)",
+	    alignment, size);
+
+	alignment = 0x10LU;
+#if LG_SIZEOF_PTR == 3
+	size = UINT64_C(0xfffffffffffffff0);
+#else
+	size = 0xfffffff0LU;
+#endif
+	assert_d_ne(posix_memalign(&p, alignment, size), 0,
+	    "Expected error for posix_memalign(&p, %zu, %zu)",
+	    alignment, size);
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+#define NITER 4
+	size_t alignment, size, total;
+	unsigned i;
+	int err;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (size = 0;
+		    size < 3 * alignment && size < (1U << 31);
+		    size += ((size == 0) ? 1 :
+		    (alignment >> (LG_SIZEOF_PTR-1)) - 1)) {
+			for (i = 0; i < NITER; i++) {
+				err = posix_memalign(&ps[i],
+				    alignment, size);
+				if (err) {
+					char buf[BUFERROR_BUF];
+
+					buferror(get_errno(), buf, sizeof(buf));
+					test_fail(
+					    "Error for alignment=%zu, "
+					    "size=%zu (%#zx): %s",
+					    alignment, size, size, buf);
+				}
+				total += malloc_usable_size(ps[i]);
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					free(ps[i]);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef NITER
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_alignment_errors,
+	    test_oom_errors,
+	    test_alignment_and_size);
+}
diff --git a/jemalloc/test/integration/rallocx.c b/jemalloc/test/integration/rallocx.c
new file mode 100644
index 0000000..08ed08d
--- /dev/null
+++ b/jemalloc/test/integration/rallocx.c
@@ -0,0 +1,258 @@
+#include "test/jemalloc_test.h"
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+TEST_BEGIN(test_grow_and_shrink) {
+	void *p, *q;
+	size_t tsz;
+#define NCYCLES 3
+	unsigned i, j;
+#define NSZS 1024
+	size_t szs[NSZS];
+#define MAXSZ ZU(12 * 1024 * 1024)
+
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	szs[0] = sallocx(p, 0);
+
+	for (i = 0; i < NCYCLES; i++) {
+		for (j = 1; j < NSZS && szs[j-1] < MAXSZ; j++) {
+			q = rallocx(p, szs[j-1]+1, 0);
+			assert_ptr_not_null(q,
+			    "Unexpected rallocx() error for size=%zu-->%zu",
+			    szs[j-1], szs[j-1]+1);
+			szs[j] = sallocx(q, 0);
+			assert_zu_ne(szs[j], szs[j-1]+1,
+			    "Expected size to be at least: %zu", szs[j-1]+1);
+			p = q;
+		}
+
+		for (j--; j > 0; j--) {
+			q = rallocx(p, szs[j-1], 0);
+			assert_ptr_not_null(q,
+			    "Unexpected rallocx() error for size=%zu-->%zu",
+			    szs[j], szs[j-1]);
+			tsz = sallocx(q, 0);
+			assert_zu_eq(tsz, szs[j-1],
+			    "Expected size=%zu, got size=%zu", szs[j-1], tsz);
+			p = q;
+		}
+	}
+
+	dallocx(p, 0);
+#undef MAXSZ
+#undef NSZS
+#undef NCYCLES
+}
+TEST_END
+
+static bool
+validate_fill(const void *p, uint8_t c, size_t offset, size_t len) {
+	bool ret = false;
+	const uint8_t *buf = (const uint8_t *)p;
+	size_t i;
+
+	for (i = 0; i < len; i++) {
+		uint8_t b = buf[offset+i];
+		if (b != c) {
+			test_fail("Allocation at %p (len=%zu) contains %#x "
+			    "rather than %#x at offset %zu", p, len, b, c,
+			    offset+i);
+			ret = true;
+		}
+	}
+
+	return ret;
+}
+
+TEST_BEGIN(test_zero) {
+	void *p, *q;
+	size_t psz, qsz, i, j;
+	size_t start_sizes[] = {1, 3*1024, 63*1024, 4095*1024};
+#define FILL_BYTE 0xaaU
+#define RANGE 2048
+
+	for (i = 0; i < sizeof(start_sizes)/sizeof(size_t); i++) {
+		size_t start_size = start_sizes[i];
+		p = mallocx(start_size, MALLOCX_ZERO);
+		assert_ptr_not_null(p, "Unexpected mallocx() error");
+		psz = sallocx(p, 0);
+
+		assert_false(validate_fill(p, 0, 0, psz),
+		    "Expected zeroed memory");
+		memset(p, FILL_BYTE, psz);
+		assert_false(validate_fill(p, FILL_BYTE, 0, psz),
+		    "Expected filled memory");
+
+		for (j = 1; j < RANGE; j++) {
+			q = rallocx(p, start_size+j, MALLOCX_ZERO);
+			assert_ptr_not_null(q, "Unexpected rallocx() error");
+			qsz = sallocx(q, 0);
+			if (q != p || qsz != psz) {
+				assert_false(validate_fill(q, FILL_BYTE, 0,
+				    psz), "Expected filled memory");
+				assert_false(validate_fill(q, 0, psz, qsz-psz),
+				    "Expected zeroed memory");
+			}
+			if (psz != qsz) {
+				memset((void *)((uintptr_t)q+psz), FILL_BYTE,
+				    qsz-psz);
+				psz = qsz;
+			}
+			p = q;
+		}
+		assert_false(validate_fill(p, FILL_BYTE, 0, psz),
+		    "Expected filled memory");
+		dallocx(p, 0);
+	}
+#undef FILL_BYTE
+}
+TEST_END
+
+TEST_BEGIN(test_align) {
+	void *p, *q;
+	size_t align;
+#define MAX_ALIGN (ZU(1) << 25)
+
+	align = ZU(1);
+	p = mallocx(1, MALLOCX_ALIGN(align));
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	for (align <<= 1; align <= MAX_ALIGN; align <<= 1) {
+		q = rallocx(p, 1, MALLOCX_ALIGN(align));
+		assert_ptr_not_null(q,
+		    "Unexpected rallocx() error for align=%zu", align);
+		assert_ptr_null(
+		    (void *)((uintptr_t)q & (align-1)),
+		    "%p inadequately aligned for align=%zu",
+		    q, align);
+		p = q;
+	}
+	dallocx(p, 0);
+#undef MAX_ALIGN
+}
+TEST_END
+
+TEST_BEGIN(test_lg_align_and_zero) {
+	void *p, *q;
+	unsigned lg_align;
+	size_t sz;
+#define MAX_LG_ALIGN 25
+#define MAX_VALIDATE (ZU(1) << 22)
+
+	lg_align = 0;
+	p = mallocx(1, MALLOCX_LG_ALIGN(lg_align)|MALLOCX_ZERO);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	for (lg_align++; lg_align <= MAX_LG_ALIGN; lg_align++) {
+		q = rallocx(p, 1, MALLOCX_LG_ALIGN(lg_align)|MALLOCX_ZERO);
+		assert_ptr_not_null(q,
+		    "Unexpected rallocx() error for lg_align=%u", lg_align);
+		assert_ptr_null(
+		    (void *)((uintptr_t)q & ((ZU(1) << lg_align)-1)),
+		    "%p inadequately aligned for lg_align=%u", q, lg_align);
+		sz = sallocx(q, 0);
+		if ((sz << 1) <= MAX_VALIDATE) {
+			assert_false(validate_fill(q, 0, 0, sz),
+			    "Expected zeroed memory");
+		} else {
+			assert_false(validate_fill(q, 0, 0, MAX_VALIDATE),
+			    "Expected zeroed memory");
+			assert_false(validate_fill(
+			    (void *)((uintptr_t)q+sz-MAX_VALIDATE),
+			    0, 0, MAX_VALIDATE), "Expected zeroed memory");
+		}
+		p = q;
+	}
+	dallocx(p, 0);
+#undef MAX_VALIDATE
+#undef MAX_LG_ALIGN
+}
+TEST_END
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	size_t largemax;
+	void *p;
+
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_ptr_null(rallocx(p, largemax+1, 0),
+	    "Expected OOM for rallocx(p, size=%#zx, 0)", largemax+1);
+
+	assert_ptr_null(rallocx(p, ZU(PTRDIFF_MAX)+1, 0),
+	    "Expected OOM for rallocx(p, size=%#zx, 0)", ZU(PTRDIFF_MAX)+1);
+
+	assert_ptr_null(rallocx(p, SIZE_T_MAX, 0),
+	    "Expected OOM for rallocx(p, size=%#zx, 0)", SIZE_T_MAX);
+
+	assert_ptr_null(rallocx(p, 1, MALLOCX_ALIGN(ZU(PTRDIFF_MAX)+1)),
+	    "Expected OOM for rallocx(p, size=1, MALLOCX_ALIGN(%#zx))",
+	    ZU(PTRDIFF_MAX)+1);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+int
+main(void) {
+	return test(
+	    test_grow_and_shrink,
+	    test_zero,
+	    test_align,
+	    test_lg_align_and_zero,
+	    test_overflow);
+}
diff --git a/jemalloc/test/integration/sdallocx.c b/jemalloc/test/integration/sdallocx.c
new file mode 100644
index 0000000..ca01448
--- /dev/null
+++ b/jemalloc/test/integration/sdallocx.c
@@ -0,0 +1,55 @@
+#include "test/jemalloc_test.h"
+
+#define MAXALIGN (((size_t)1) << 22)
+#define NITER 3
+
+TEST_BEGIN(test_basic) {
+	void *ptr = mallocx(64, 0);
+	sdallocx(ptr, 64, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+	size_t nsz, sz, alignment, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (sz = 1;
+		    sz < 3 * alignment && sz < (1U << 31);
+		    sz += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				nsz = nallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO);
+				ps[i] = mallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO);
+				total += nsz;
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					sdallocx(ps[i], sz,
+					    MALLOCX_ALIGN(alignment));
+					ps[i] = NULL;
+				}
+			}
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_basic,
+	    test_alignment_and_size);
+}
diff --git a/jemalloc/test/integration/slab_sizes.c b/jemalloc/test/integration/slab_sizes.c
new file mode 100644
index 0000000..af250c3
--- /dev/null
+++ b/jemalloc/test/integration/slab_sizes.c
@@ -0,0 +1,80 @@
+#include "test/jemalloc_test.h"
+
+/* Note that this test relies on the unusual slab sizes set in slab_sizes.sh. */
+
+TEST_BEGIN(test_slab_sizes) {
+	unsigned nbins;
+	size_t page;
+	size_t sizemib[4];
+	size_t slabmib[4];
+	size_t len;
+
+	len = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", &nbins, &len, NULL, 0), 0,
+	    "nbins mallctl failure");
+
+	len = sizeof(page);
+	assert_d_eq(mallctl("arenas.page", &page, &len, NULL, 0), 0,
+	    "page mallctl failure");
+
+	len = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.size", sizemib, &len), 0,
+	    "bin size mallctlnametomib failure");
+
+	len = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.slab_size", slabmib, &len),
+	    0, "slab size mallctlnametomib failure");
+
+	size_t biggest_slab_seen = 0;
+
+	for (unsigned i = 0; i < nbins; i++) {
+		size_t bin_size;
+		size_t slab_size;
+		len = sizeof(size_t);
+		sizemib[2] = i;
+		slabmib[2] = i;
+		assert_d_eq(mallctlbymib(sizemib, 4, (void *)&bin_size, &len,
+		    NULL, 0), 0, "bin size mallctlbymib failure");
+
+		len = sizeof(size_t);
+		assert_d_eq(mallctlbymib(slabmib, 4, (void *)&slab_size, &len,
+		    NULL, 0), 0, "slab size mallctlbymib failure");
+
+		if (bin_size < 100) {
+			/*
+			 * Then we should be as close to 17 as possible.  Since
+			 * not all page sizes are valid (because of bitmap
+			 * limitations on the number of items in a slab), we
+			 * should at least make sure that the number of pages
+			 * goes up.
+			 */
+			assert_zu_ge(slab_size, biggest_slab_seen,
+			    "Slab sizes should go up");
+			biggest_slab_seen = slab_size;
+		} else if (
+		    (100 <= bin_size && bin_size < 128)
+		    || (128 < bin_size && bin_size <= 200)) {
+			assert_zu_eq(slab_size, page,
+			    "Forced-small slabs should be small");
+		} else if (bin_size == 128) {
+			assert_zu_eq(slab_size, 2 * page,
+			    "Forced-2-page slab should be 2 pages");
+		} else if (200 < bin_size && bin_size <= 4096) {
+			assert_zu_ge(slab_size, biggest_slab_seen,
+			    "Slab sizes should go up");
+			biggest_slab_seen = slab_size;
+		}
+	}
+	/*
+	 * For any reasonable configuration, 17 pages should be a valid slab
+	 * size for 4096-byte items.
+	 */
+	assert_zu_eq(biggest_slab_seen, 17 * page, "Didn't hit page target");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_slab_sizes);
+}
diff --git a/jemalloc/test/integration/slab_sizes.sh b/jemalloc/test/integration/slab_sizes.sh
new file mode 100644
index 0000000..07e3db8
--- /dev/null
+++ b/jemalloc/test/integration/slab_sizes.sh
@@ -0,0 +1,4 @@
+#!/bin/sh
+
+# Some screwy-looking slab sizes.
+export MALLOC_CONF="slab_sizes:1-4096:17|100-200:1|128-128:2"
diff --git a/jemalloc/test/integration/smallocx.c b/jemalloc/test/integration/smallocx.c
new file mode 100644
index 0000000..2486752
--- /dev/null
+++ b/jemalloc/test/integration/smallocx.c
@@ -0,0 +1,312 @@
+#include "test/jemalloc_test.h"
+#include "jemalloc/jemalloc_macros.h"
+
+#define STR_HELPER(x) #x
+#define STR(x) STR_HELPER(x)
+
+#ifndef JEMALLOC_VERSION_GID_IDENT
+  #error "JEMALLOC_VERSION_GID_IDENT not defined"
+#endif
+
+#define JOIN(x, y) x ## y
+#define JOIN2(x, y) JOIN(x, y)
+#define smallocx JOIN2(smallocx_, JEMALLOC_VERSION_GID_IDENT)
+
+typedef struct {
+	void *ptr;
+	size_t size;
+} smallocx_return_t;
+
+extern smallocx_return_t
+smallocx(size_t size, int flags);
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	size_t largemax;
+
+	largemax = get_large_size(get_nlarge()-1);
+
+	assert_ptr_null(smallocx(largemax+1, 0).ptr,
+	    "Expected OOM for smallocx(size=%#zx, 0)", largemax+1);
+
+	assert_ptr_null(smallocx(ZU(PTRDIFF_MAX)+1, 0).ptr,
+	    "Expected OOM for smallocx(size=%#zx, 0)", ZU(PTRDIFF_MAX)+1);
+
+	assert_ptr_null(smallocx(SIZE_T_MAX, 0).ptr,
+	    "Expected OOM for smallocx(size=%#zx, 0)", SIZE_T_MAX);
+
+	assert_ptr_null(smallocx(1, MALLOCX_ALIGN(ZU(PTRDIFF_MAX)+1)).ptr,
+	    "Expected OOM for smallocx(size=1, MALLOCX_ALIGN(%#zx))",
+	    ZU(PTRDIFF_MAX)+1);
+}
+TEST_END
+
+static void *
+remote_alloc(void *arg) {
+	unsigned arena;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	size_t large_sz;
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large_sz, &sz,
+	    NULL, 0), 0, "Unexpected mallctl failure");
+
+	smallocx_return_t r
+	    = smallocx(large_sz, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE);
+	void *ptr = r.ptr;
+	assert_zu_eq(r.size,
+	    nallocx(large_sz, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE),
+	    "Expected smalloc(size,flags).size == nallocx(size,flags)");
+	void **ret = (void **)arg;
+	*ret = ptr;
+
+	return NULL;
+}
+
+TEST_BEGIN(test_remote_free) {
+	thd_t thd;
+	void *ret;
+	thd_create(&thd, remote_alloc, (void *)&ret);
+	thd_join(thd, NULL);
+	assert_ptr_not_null(ret, "Unexpected smallocx failure");
+
+	/* Avoid TCACHE_NONE to explicitly test tcache_flush(). */
+	dallocx(ret, 0);
+	mallctl("thread.tcache.flush", NULL, NULL, NULL, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_oom) {
+	size_t largemax;
+	bool oom;
+	void *ptrs[3];
+	unsigned i;
+
+	/*
+	 * It should be impossible to allocate three objects that each consume
+	 * nearly half the virtual address space.
+	 */
+	largemax = get_large_size(get_nlarge()-1);
+	oom = false;
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		ptrs[i] = smallocx(largemax, 0).ptr;
+		if (ptrs[i] == NULL) {
+			oom = true;
+		}
+	}
+	assert_true(oom,
+	    "Expected OOM during series of calls to smallocx(size=%zu, 0)",
+	    largemax);
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		if (ptrs[i] != NULL) {
+			dallocx(ptrs[i], 0);
+		}
+	}
+	purge();
+
+#if LG_SIZEOF_PTR == 3
+	assert_ptr_null(smallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x8000000000000000ULL)).ptr,
+	    "Expected OOM for smallocx()");
+	assert_ptr_null(smallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x80000000)).ptr,
+	    "Expected OOM for smallocx()");
+#else
+	assert_ptr_null(smallocx(0x80000000UL, MALLOCX_ALIGN(0x80000000UL)).ptr,
+	    "Expected OOM for smallocx()");
+#endif
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+TEST_BEGIN(test_basic) {
+#define MAXSZ (((size_t)1) << 23)
+	size_t sz;
+
+	for (sz = 1; sz < MAXSZ; sz = nallocx(sz, 0) + 1) {
+		smallocx_return_t ret;
+		size_t nsz, rsz, smz;
+		void *p;
+		nsz = nallocx(sz, 0);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		ret = smallocx(sz, 0);
+		p = ret.ptr;
+		smz = ret.size;
+		assert_ptr_not_null(p,
+		    "Unexpected smallocx(size=%zx, flags=0) error", sz);
+		rsz = sallocx(p, 0);
+		assert_zu_ge(rsz, sz, "Real size smaller than expected");
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() size mismatch");
+		assert_zu_eq(nsz, smz, "nallocx()/smallocx() size mismatch");
+		dallocx(p, 0);
+
+		ret = smallocx(sz, 0);
+		p = ret.ptr;
+		smz = ret.size;
+		assert_ptr_not_null(p,
+		    "Unexpected smallocx(size=%zx, flags=0) error", sz);
+		dallocx(p, 0);
+
+		nsz = nallocx(sz, MALLOCX_ZERO);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		assert_zu_ne(smz, 0, "Unexpected smallocx() error");
+		ret = smallocx(sz, MALLOCX_ZERO);
+		p = ret.ptr;
+		assert_ptr_not_null(p,
+		    "Unexpected smallocx(size=%zx, flags=MALLOCX_ZERO) error",
+		    nsz);
+		rsz = sallocx(p, 0);
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() rsize mismatch");
+		assert_zu_eq(nsz, smz, "nallocx()/smallocx() size mismatch");
+		dallocx(p, 0);
+		purge();
+	}
+#undef MAXSZ
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+	const char *percpu_arena;
+	size_t sz = sizeof(percpu_arena);
+
+	if(mallctl("opt.percpu_arena", (void *)&percpu_arena, &sz, NULL, 0) ||
+	    strcmp(percpu_arena, "disabled") != 0) {
+		test_skip("test_alignment_and_size skipped: "
+		    "not working with percpu arena.");
+	};
+#define MAXALIGN (((size_t)1) << 23)
+#define NITER 4
+	size_t nsz, rsz, smz, alignment, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (sz = 1;
+		    sz < 3 * alignment && sz < (1U << 31);
+		    sz += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				nsz = nallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO);
+				assert_zu_ne(nsz, 0,
+				    "nallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				smallocx_return_t ret
+				    = smallocx(sz, MALLOCX_ALIGN(alignment) | MALLOCX_ZERO);
+				ps[i] = ret.ptr;
+				assert_ptr_not_null(ps[i],
+				    "smallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				rsz = sallocx(ps[i], 0);
+				smz = ret.size;
+				assert_zu_ge(rsz, sz,
+				    "Real size smaller than expected for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_zu_eq(nsz, rsz,
+				    "nallocx()/sallocx() size mismatch for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_zu_eq(nsz, smz,
+				    "nallocx()/smallocx() size mismatch for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_ptr_null(
+				    (void *)((uintptr_t)ps[i] & (alignment-1)),
+				    "%p inadequately aligned for"
+				    " alignment=%zu, size=%zu", ps[i],
+				    alignment, sz);
+				total += rsz;
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					dallocx(ps[i], 0);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef MAXALIGN
+#undef NITER
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_overflow,
+	    test_oom,
+	    test_remote_free,
+	    test_basic,
+	    test_alignment_and_size);
+}
diff --git a/jemalloc/test/integration/smallocx.sh b/jemalloc/test/integration/smallocx.sh
new file mode 100644
index 0000000..d07f10f
--- /dev/null
+++ b/jemalloc/test/integration/smallocx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+    export MALLOC_CONF="junk:false"
+fi
diff --git a/jemalloc/test/integration/thread_arena.c b/jemalloc/test/integration/thread_arena.c
new file mode 100644
index 0000000..1e5ec05
--- /dev/null
+++ b/jemalloc/test/integration/thread_arena.c
@@ -0,0 +1,86 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS 10
+
+void *
+thd_start(void *arg) {
+	unsigned main_arena_ind = *(unsigned *)arg;
+	void *p;
+	unsigned arena_ind;
+	size_t size;
+	int err;
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Error in malloc()");
+	free(p);
+
+	size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", (void *)&arena_ind, &size,
+	    (void *)&main_arena_ind, sizeof(main_arena_ind)))) {
+		char buf[BUFERROR_BUF];
+
+		buferror(err, buf, sizeof(buf));
+		test_fail("Error in mallctl(): %s", buf);
+	}
+
+	size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", (void *)&arena_ind, &size, NULL,
+	    0))) {
+		char buf[BUFERROR_BUF];
+
+		buferror(err, buf, sizeof(buf));
+		test_fail("Error in mallctl(): %s", buf);
+	}
+	assert_u_eq(arena_ind, main_arena_ind,
+	    "Arena index should be same as for main thread");
+
+	return NULL;
+}
+
+static void
+mallctl_failure(int err) {
+	char buf[BUFERROR_BUF];
+
+	buferror(err, buf, sizeof(buf));
+	test_fail("Error in mallctl(): %s", buf);
+}
+
+TEST_BEGIN(test_thread_arena) {
+	void *p;
+	int err;
+	thd_t thds[NTHREADS];
+	unsigned i;
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Error in malloc()");
+
+	unsigned arena_ind, old_arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Arena creation failure");
+
+	size_t size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", (void *)&old_arena_ind, &size,
+	    (void *)&arena_ind, sizeof(arena_ind))) != 0) {
+		mallctl_failure(err);
+	}
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start,
+		    (void *)&arena_ind);
+	}
+
+	for (i = 0; i < NTHREADS; i++) {
+		intptr_t join_ret;
+		thd_join(thds[i], (void *)&join_ret);
+		assert_zd_eq(join_ret, 0, "Unexpected thread join error");
+	}
+	free(p);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_thread_arena);
+}
diff --git a/jemalloc/test/integration/thread_tcache_enabled.c b/jemalloc/test/integration/thread_tcache_enabled.c
new file mode 100644
index 0000000..95c9acc
--- /dev/null
+++ b/jemalloc/test/integration/thread_tcache_enabled.c
@@ -0,0 +1,87 @@
+#include "test/jemalloc_test.h"
+
+void *
+thd_start(void *arg) {
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	if (e0) {
+		e1 = false;
+		assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+		    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+		assert_true(e0, "tcache should be enabled");
+	}
+
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	free(malloc(1));
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	free(malloc(1));
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	free(malloc(1));
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	free(malloc(1));
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	free(malloc(1));
+	return NULL;
+}
+
+TEST_BEGIN(test_main_thread) {
+	thd_start(NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_subthread) {
+	thd_t thd;
+
+	thd_create(&thd, thd_start, NULL);
+	thd_join(thd, NULL);
+}
+TEST_END
+
+int
+main(void) {
+	/* Run tests multiple times to check for bad interactions. */
+	return test(
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread);
+}
diff --git a/jemalloc/test/integration/xallocx.c b/jemalloc/test/integration/xallocx.c
new file mode 100644
index 0000000..cd0ca04
--- /dev/null
+++ b/jemalloc/test/integration/xallocx.c
@@ -0,0 +1,384 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * Use a separate arena for xallocx() extension/contraction tests so that
+ * internal allocation e.g. by heap profiling can't interpose allocations where
+ * xallocx() would ordinarily be able to extend.
+ */
+static unsigned
+arena_ind(void) {
+	static unsigned ind = 0;
+
+	if (ind == 0) {
+		size_t sz = sizeof(ind);
+		assert_d_eq(mallctl("arenas.create", (void *)&ind, &sz, NULL,
+		    0), 0, "Unexpected mallctl failure creating arena");
+	}
+
+	return ind;
+}
+
+TEST_BEGIN(test_same_size) {
+	void *p;
+	size_t sz, tsz;
+
+	p = mallocx(42, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	sz = sallocx(p, 0);
+
+	tsz = xallocx(p, sz, 0, 0);
+	assert_zu_eq(tsz, sz, "Unexpected size change: %zu --> %zu", sz, tsz);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_extra_no_move) {
+	void *p;
+	size_t sz, tsz;
+
+	p = mallocx(42, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	sz = sallocx(p, 0);
+
+	tsz = xallocx(p, sz, sz-42, 0);
+	assert_zu_eq(tsz, sz, "Unexpected size change: %zu --> %zu", sz, tsz);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_no_move_fail) {
+	void *p;
+	size_t sz, tsz;
+
+	p = mallocx(42, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	sz = sallocx(p, 0);
+
+	tsz = xallocx(p, sz + 5, 0, 0);
+	assert_zu_eq(tsz, sz, "Unexpected size change: %zu --> %zu", sz, tsz);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nsmall(void) {
+	return get_nsizes_impl("arenas.nbins");
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_small_size(size_t ind) {
+	return get_size_impl("arenas.bin.0.size", ind);
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+TEST_BEGIN(test_size) {
+	size_t small0, largemax;
+	void *p;
+
+	/* Get size classes. */
+	small0 = get_small_size(0);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(small0, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	/* Test smallest supported size. */
+	assert_zu_eq(xallocx(p, 1, 0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	/* Test largest supported size. */
+	assert_zu_le(xallocx(p, largemax, 0, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	/* Test size overflow. */
+	assert_zu_le(xallocx(p, largemax+1, 0, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, SIZE_T_MAX, 0, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_size_extra_overflow) {
+	size_t small0, largemax;
+	void *p;
+
+	/* Get size classes. */
+	small0 = get_small_size(0);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(small0, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	/* Test overflows that can be resolved by clamping extra. */
+	assert_zu_le(xallocx(p, largemax-1, 2, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, largemax, 1, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	/* Test overflow such that largemax-size underflows. */
+	assert_zu_le(xallocx(p, largemax+1, 2, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, largemax+2, 3, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, SIZE_T_MAX-2, 2, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, SIZE_T_MAX-1, 1, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_extra_small) {
+	size_t small0, small1, largemax;
+	void *p;
+
+	/* Get size classes. */
+	small0 = get_small_size(0);
+	small1 = get_small_size(1);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(small0, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	assert_zu_eq(xallocx(p, small1, 0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_eq(xallocx(p, small1, 0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_eq(xallocx(p, small0, small1 - small0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	/* Test size+extra overflow. */
+	assert_zu_eq(xallocx(p, small0, largemax - small0 + 1, 0), small0,
+	    "Unexpected xallocx() behavior");
+	assert_zu_eq(xallocx(p, small0, SIZE_T_MAX - small0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_extra_large) {
+	int flags = MALLOCX_ARENA(arena_ind());
+	size_t smallmax, large1, large2, large3, largemax;
+	void *p;
+
+	/* Get size classes. */
+	smallmax = get_small_size(get_nsmall()-1);
+	large1 = get_large_size(1);
+	large2 = get_large_size(2);
+	large3 = get_large_size(3);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(large3, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	assert_zu_eq(xallocx(p, large3, 0, flags), large3,
+	    "Unexpected xallocx() behavior");
+	/* Test size decrease with zero extra. */
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	assert_zu_ge(xallocx(p, smallmax, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+
+	if (xallocx(p, large3, 0, flags) != large3) {
+		p = rallocx(p, large3, flags);
+		assert_ptr_not_null(p, "Unexpected rallocx() failure");
+	}
+	/* Test size decrease with non-zero extra. */
+	assert_zu_eq(xallocx(p, large1, large3 - large1, flags), large3,
+	    "Unexpected xallocx() behavior");
+	assert_zu_eq(xallocx(p, large2, large3 - large2, flags), large3,
+	    "Unexpected xallocx() behavior");
+	assert_zu_ge(xallocx(p, large1, large2 - large1, flags), large2,
+	    "Unexpected xallocx() behavior");
+	assert_zu_ge(xallocx(p, smallmax, large1 - smallmax, flags), large1,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	/* Test size increase with zero extra. */
+	assert_zu_le(xallocx(p, large3, 0, flags), large3,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, largemax+1, 0, flags), large3,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	/* Test size increase with non-zero extra. */
+	assert_zu_le(xallocx(p, large1, SIZE_T_MAX - large1, flags), largemax,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	/* Test size increase with non-zero extra. */
+	assert_zu_le(xallocx(p, large1, large3 - large1, flags), large3,
+	    "Unexpected xallocx() behavior");
+
+	if (xallocx(p, large3, 0, flags) != large3) {
+		p = rallocx(p, large3, flags);
+		assert_ptr_not_null(p, "Unexpected rallocx() failure");
+	}
+	/* Test size+extra overflow. */
+	assert_zu_le(xallocx(p, large3, largemax - large3 + 1, flags), largemax,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, flags);
+}
+TEST_END
+
+static void
+print_filled_extents(const void *p, uint8_t c, size_t len) {
+	const uint8_t *pc = (const uint8_t *)p;
+	size_t i, range0;
+	uint8_t c0;
+
+	malloc_printf("  p=%p, c=%#x, len=%zu:", p, c, len);
+	range0 = 0;
+	c0 = pc[0];
+	for (i = 0; i < len; i++) {
+		if (pc[i] != c0) {
+			malloc_printf(" %#x[%zu..%zu)", c0, range0, i);
+			range0 = i;
+			c0 = pc[i];
+		}
+	}
+	malloc_printf(" %#x[%zu..%zu)\n", c0, range0, i);
+}
+
+static bool
+validate_fill(const void *p, uint8_t c, size_t offset, size_t len) {
+	const uint8_t *pc = (const uint8_t *)p;
+	bool err;
+	size_t i;
+
+	for (i = offset, err = false; i < offset+len; i++) {
+		if (pc[i] != c) {
+			err = true;
+		}
+	}
+
+	if (err) {
+		print_filled_extents(p, c, offset + len);
+	}
+
+	return err;
+}
+
+static void
+test_zero(size_t szmin, size_t szmax) {
+	int flags = MALLOCX_ARENA(arena_ind()) | MALLOCX_ZERO;
+	size_t sz, nsz;
+	void *p;
+#define FILL_BYTE 0x7aU
+
+	sz = szmax;
+	p = mallocx(sz, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	assert_false(validate_fill(p, 0x00, 0, sz), "Memory not filled: sz=%zu",
+	    sz);
+
+	/*
+	 * Fill with non-zero so that non-debug builds are more likely to detect
+	 * errors.
+	 */
+	memset(p, FILL_BYTE, sz);
+	assert_false(validate_fill(p, FILL_BYTE, 0, sz),
+	    "Memory not filled: sz=%zu", sz);
+
+	/* Shrink in place so that we can expect growing in place to succeed. */
+	sz = szmin;
+	if (xallocx(p, sz, 0, flags) != sz) {
+		p = rallocx(p, sz, flags);
+		assert_ptr_not_null(p, "Unexpected rallocx() failure");
+	}
+	assert_false(validate_fill(p, FILL_BYTE, 0, sz),
+	    "Memory not filled: sz=%zu", sz);
+
+	for (sz = szmin; sz < szmax; sz = nsz) {
+		nsz = nallocx(sz+1, flags);
+		if (xallocx(p, sz+1, 0, flags) != nsz) {
+			p = rallocx(p, sz+1, flags);
+			assert_ptr_not_null(p, "Unexpected rallocx() failure");
+		}
+		assert_false(validate_fill(p, FILL_BYTE, 0, sz),
+		    "Memory not filled: sz=%zu", sz);
+		assert_false(validate_fill(p, 0x00, sz, nsz-sz),
+		    "Memory not filled: sz=%zu, nsz-sz=%zu", sz, nsz-sz);
+		memset((void *)((uintptr_t)p + sz), FILL_BYTE, nsz-sz);
+		assert_false(validate_fill(p, FILL_BYTE, 0, nsz),
+		    "Memory not filled: nsz=%zu", nsz);
+	}
+
+	dallocx(p, flags);
+}
+
+TEST_BEGIN(test_zero_large) {
+	size_t large0, large1;
+
+	/* Get size classes. */
+	large0 = get_large_size(0);
+	large1 = get_large_size(1);
+
+	test_zero(large1, large0 * 2);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_same_size,
+	    test_extra_no_move,
+	    test_no_move_fail,
+	    test_size,
+	    test_size_extra_overflow,
+	    test_extra_small,
+	    test_extra_large,
+	    test_zero_large);
+}
diff --git a/jemalloc/test/integration/xallocx.sh b/jemalloc/test/integration/xallocx.sh
new file mode 100644
index 0000000..0cc2187
--- /dev/null
+++ b/jemalloc/test/integration/xallocx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="junk:false"
+fi
diff --git a/jemalloc/test/src/SFMT.c b/jemalloc/test/src/SFMT.c
new file mode 100644
index 0000000..c05e218
--- /dev/null
+++ b/jemalloc/test/src/SFMT.c
@@ -0,0 +1,719 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/**
+ * @file  SFMT.c
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT)
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * Copyright (C) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software, see LICENSE.txt
+ */
+#define SFMT_C_
+#include "test/jemalloc_test.h"
+#include "test/SFMT-params.h"
+
+#if defined(JEMALLOC_BIG_ENDIAN) && !defined(BIG_ENDIAN64)
+#define BIG_ENDIAN64 1
+#endif
+#if defined(__BIG_ENDIAN__) && !defined(__amd64) && !defined(BIG_ENDIAN64)
+#define BIG_ENDIAN64 1
+#endif
+#if defined(HAVE_ALTIVEC) && !defined(BIG_ENDIAN64)
+#define BIG_ENDIAN64 1
+#endif
+#if defined(ONLY64) && !defined(BIG_ENDIAN64)
+  #if defined(__GNUC__)
+    #error "-DONLY64 must be specified with -DBIG_ENDIAN64"
+  #endif
+#undef ONLY64
+#endif
+/*------------------------------------------------------
+  128-bit SIMD data type for Altivec, SSE2 or standard C
+  ------------------------------------------------------*/
+#if defined(HAVE_ALTIVEC)
+/** 128-bit data structure */
+union W128_T {
+    vector unsigned int s;
+    uint32_t u[4];
+};
+/** 128-bit data type */
+typedef union W128_T w128_t;
+
+#elif defined(HAVE_SSE2)
+/** 128-bit data structure */
+union W128_T {
+    __m128i si;
+    uint32_t u[4];
+};
+/** 128-bit data type */
+typedef union W128_T w128_t;
+
+#else
+
+/** 128-bit data structure */
+struct W128_T {
+    uint32_t u[4];
+};
+/** 128-bit data type */
+typedef struct W128_T w128_t;
+
+#endif
+
+struct sfmt_s {
+    /** the 128-bit internal state array */
+    w128_t sfmt[N];
+    /** index counter to the 32-bit internal state array */
+    int idx;
+    /** a flag: it is 0 if and only if the internal state is not yet
+     * initialized. */
+    int initialized;
+};
+
+/*--------------------------------------
+  FILE GLOBAL VARIABLES
+  internal state, index counter and flag
+  --------------------------------------*/
+
+/** a parity check vector which certificate the period of 2^{MEXP} */
+static uint32_t parity[4] = {PARITY1, PARITY2, PARITY3, PARITY4};
+
+/*----------------
+  STATIC FUNCTIONS
+  ----------------*/
+static inline int idxof(int i);
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+static inline void rshift128(w128_t *out,  w128_t const *in, int shift);
+static inline void lshift128(w128_t *out,  w128_t const *in, int shift);
+#endif
+static inline void gen_rand_all(sfmt_t *ctx);
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size);
+static inline uint32_t func1(uint32_t x);
+static inline uint32_t func2(uint32_t x);
+static void period_certification(sfmt_t *ctx);
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+static inline void swap(w128_t *array, int size);
+#endif
+
+#if defined(HAVE_ALTIVEC)
+  #include "test/SFMT-alti.h"
+#elif defined(HAVE_SSE2)
+  #include "test/SFMT-sse2.h"
+#endif
+
+/**
+ * This function simulate a 64-bit index of LITTLE ENDIAN
+ * in BIG ENDIAN machine.
+ */
+#ifdef ONLY64
+static inline int idxof(int i) {
+    return i ^ 1;
+}
+#else
+static inline int idxof(int i) {
+    return i;
+}
+#endif
+/**
+ * This function simulates SIMD 128-bit right shift by the standard C.
+ * The 128-bit integer given in in is shifted by (shift * 8) bits.
+ * This function simulates the LITTLE ENDIAN SIMD.
+ * @param out the output of this function
+ * @param in the 128-bit data to be shifted
+ * @param shift the shift value
+ */
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+#ifdef ONLY64
+static inline void rshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[2] << 32) | ((uint64_t)in->u[3]);
+    tl = ((uint64_t)in->u[0] << 32) | ((uint64_t)in->u[1]);
+
+    oh = th >> (shift * 8);
+    ol = tl >> (shift * 8);
+    ol |= th << (64 - shift * 8);
+    out->u[0] = (uint32_t)(ol >> 32);
+    out->u[1] = (uint32_t)ol;
+    out->u[2] = (uint32_t)(oh >> 32);
+    out->u[3] = (uint32_t)oh;
+}
+#else
+static inline void rshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[3] << 32) | ((uint64_t)in->u[2]);
+    tl = ((uint64_t)in->u[1] << 32) | ((uint64_t)in->u[0]);
+
+    oh = th >> (shift * 8);
+    ol = tl >> (shift * 8);
+    ol |= th << (64 - shift * 8);
+    out->u[1] = (uint32_t)(ol >> 32);
+    out->u[0] = (uint32_t)ol;
+    out->u[3] = (uint32_t)(oh >> 32);
+    out->u[2] = (uint32_t)oh;
+}
+#endif
+/**
+ * This function simulates SIMD 128-bit left shift by the standard C.
+ * The 128-bit integer given in in is shifted by (shift * 8) bits.
+ * This function simulates the LITTLE ENDIAN SIMD.
+ * @param out the output of this function
+ * @param in the 128-bit data to be shifted
+ * @param shift the shift value
+ */
+#ifdef ONLY64
+static inline void lshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[2] << 32) | ((uint64_t)in->u[3]);
+    tl = ((uint64_t)in->u[0] << 32) | ((uint64_t)in->u[1]);
+
+    oh = th << (shift * 8);
+    ol = tl << (shift * 8);
+    oh |= tl >> (64 - shift * 8);
+    out->u[0] = (uint32_t)(ol >> 32);
+    out->u[1] = (uint32_t)ol;
+    out->u[2] = (uint32_t)(oh >> 32);
+    out->u[3] = (uint32_t)oh;
+}
+#else
+static inline void lshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[3] << 32) | ((uint64_t)in->u[2]);
+    tl = ((uint64_t)in->u[1] << 32) | ((uint64_t)in->u[0]);
+
+    oh = th << (shift * 8);
+    ol = tl << (shift * 8);
+    oh |= tl >> (64 - shift * 8);
+    out->u[1] = (uint32_t)(ol >> 32);
+    out->u[0] = (uint32_t)ol;
+    out->u[3] = (uint32_t)(oh >> 32);
+    out->u[2] = (uint32_t)oh;
+}
+#endif
+#endif
+
+/**
+ * This function represents the recursion formula.
+ * @param r output
+ * @param a a 128-bit part of the internal state array
+ * @param b a 128-bit part of the internal state array
+ * @param c a 128-bit part of the internal state array
+ * @param d a 128-bit part of the internal state array
+ */
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+#ifdef ONLY64
+static inline void do_recursion(w128_t *r, w128_t *a, w128_t *b, w128_t *c,
+				w128_t *d) {
+    w128_t x;
+    w128_t y;
+
+    lshift128(&x, a, SL2);
+    rshift128(&y, c, SR2);
+    r->u[0] = a->u[0] ^ x.u[0] ^ ((b->u[0] >> SR1) & MSK2) ^ y.u[0]
+	^ (d->u[0] << SL1);
+    r->u[1] = a->u[1] ^ x.u[1] ^ ((b->u[1] >> SR1) & MSK1) ^ y.u[1]
+	^ (d->u[1] << SL1);
+    r->u[2] = a->u[2] ^ x.u[2] ^ ((b->u[2] >> SR1) & MSK4) ^ y.u[2]
+	^ (d->u[2] << SL1);
+    r->u[3] = a->u[3] ^ x.u[3] ^ ((b->u[3] >> SR1) & MSK3) ^ y.u[3]
+	^ (d->u[3] << SL1);
+}
+#else
+static inline void do_recursion(w128_t *r, w128_t *a, w128_t *b, w128_t *c,
+				w128_t *d) {
+    w128_t x;
+    w128_t y;
+
+    lshift128(&x, a, SL2);
+    rshift128(&y, c, SR2);
+    r->u[0] = a->u[0] ^ x.u[0] ^ ((b->u[0] >> SR1) & MSK1) ^ y.u[0]
+	^ (d->u[0] << SL1);
+    r->u[1] = a->u[1] ^ x.u[1] ^ ((b->u[1] >> SR1) & MSK2) ^ y.u[1]
+	^ (d->u[1] << SL1);
+    r->u[2] = a->u[2] ^ x.u[2] ^ ((b->u[2] >> SR1) & MSK3) ^ y.u[2]
+	^ (d->u[2] << SL1);
+    r->u[3] = a->u[3] ^ x.u[3] ^ ((b->u[3] >> SR1) & MSK4) ^ y.u[3]
+	^ (d->u[3] << SL1);
+}
+#endif
+#endif
+
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+/**
+ * This function fills the internal state array with pseudorandom
+ * integers.
+ */
+static inline void gen_rand_all(sfmt_t *ctx) {
+    int i;
+    w128_t *r1, *r2;
+
+    r1 = &ctx->sfmt[N - 2];
+    r2 = &ctx->sfmt[N - 1];
+    for (i = 0; i < N - POS1; i++) {
+	do_recursion(&ctx->sfmt[i], &ctx->sfmt[i], &ctx->sfmt[i + POS1], r1,
+	  r2);
+	r1 = r2;
+	r2 = &ctx->sfmt[i];
+    }
+    for (; i < N; i++) {
+	do_recursion(&ctx->sfmt[i], &ctx->sfmt[i], &ctx->sfmt[i + POS1 - N], r1,
+	  r2);
+	r1 = r2;
+	r2 = &ctx->sfmt[i];
+    }
+}
+
+/**
+ * This function fills the user-specified array with pseudorandom
+ * integers.
+ *
+ * @param array an 128-bit array to be filled by pseudorandom numbers.
+ * @param size number of 128-bit pseudorandom numbers to be generated.
+ */
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size) {
+    int i, j;
+    w128_t *r1, *r2;
+
+    r1 = &ctx->sfmt[N - 2];
+    r2 = &ctx->sfmt[N - 1];
+    for (i = 0; i < N - POS1; i++) {
+	do_recursion(&array[i], &ctx->sfmt[i], &ctx->sfmt[i + POS1], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+    }
+    for (; i < N; i++) {
+	do_recursion(&array[i], &ctx->sfmt[i], &array[i + POS1 - N], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+    }
+    for (; i < size - N; i++) {
+	do_recursion(&array[i], &array[i - N], &array[i + POS1 - N], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+    }
+    for (j = 0; j < 2 * N - size; j++) {
+	ctx->sfmt[j] = array[j + size - N];
+    }
+    for (; i < size; i++, j++) {
+	do_recursion(&array[i], &array[i - N], &array[i + POS1 - N], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+	ctx->sfmt[j] = array[i];
+    }
+}
+#endif
+
+#if defined(BIG_ENDIAN64) && !defined(ONLY64) && !defined(HAVE_ALTIVEC)
+static inline void swap(w128_t *array, int size) {
+    int i;
+    uint32_t x, y;
+
+    for (i = 0; i < size; i++) {
+	x = array[i].u[0];
+	y = array[i].u[2];
+	array[i].u[0] = array[i].u[1];
+	array[i].u[2] = array[i].u[3];
+	array[i].u[1] = x;
+	array[i].u[3] = y;
+    }
+}
+#endif
+/**
+ * This function represents a function used in the initialization
+ * by init_by_array
+ * @param x 32-bit integer
+ * @return 32-bit integer
+ */
+static uint32_t func1(uint32_t x) {
+    return (x ^ (x >> 27)) * (uint32_t)1664525UL;
+}
+
+/**
+ * This function represents a function used in the initialization
+ * by init_by_array
+ * @param x 32-bit integer
+ * @return 32-bit integer
+ */
+static uint32_t func2(uint32_t x) {
+    return (x ^ (x >> 27)) * (uint32_t)1566083941UL;
+}
+
+/**
+ * This function certificate the period of 2^{MEXP}
+ */
+static void period_certification(sfmt_t *ctx) {
+    int inner = 0;
+    int i, j;
+    uint32_t work;
+    uint32_t *psfmt32 = &ctx->sfmt[0].u[0];
+
+    for (i = 0; i < 4; i++)
+	inner ^= psfmt32[idxof(i)] & parity[i];
+    for (i = 16; i > 0; i >>= 1)
+	inner ^= inner >> i;
+    inner &= 1;
+    /* check OK */
+    if (inner == 1) {
+	return;
+    }
+    /* check NG, and modification */
+    for (i = 0; i < 4; i++) {
+	work = 1;
+	for (j = 0; j < 32; j++) {
+	    if ((work & parity[i]) != 0) {
+		psfmt32[idxof(i)] ^= work;
+		return;
+	    }
+	    work = work << 1;
+	}
+    }
+}
+
+/*----------------
+  PUBLIC FUNCTIONS
+  ----------------*/
+/**
+ * This function returns the identification string.
+ * The string shows the word size, the Mersenne exponent,
+ * and all parameters of this generator.
+ */
+const char *get_idstring(void) {
+    return IDSTR;
+}
+
+/**
+ * This function returns the minimum size of array used for \b
+ * fill_array32() function.
+ * @return minimum size of array used for fill_array32() function.
+ */
+int get_min_array_size32(void) {
+    return N32;
+}
+
+/**
+ * This function returns the minimum size of array used for \b
+ * fill_array64() function.
+ * @return minimum size of array used for fill_array64() function.
+ */
+int get_min_array_size64(void) {
+    return N64;
+}
+
+#ifndef ONLY64
+/**
+ * This function generates and returns 32-bit pseudorandom number.
+ * init_gen_rand or init_by_array must be called before this function.
+ * @return 32-bit pseudorandom number
+ */
+uint32_t gen_rand32(sfmt_t *ctx) {
+    uint32_t r;
+    uint32_t *psfmt32 = &ctx->sfmt[0].u[0];
+
+    assert(ctx->initialized);
+    if (ctx->idx >= N32) {
+	gen_rand_all(ctx);
+	ctx->idx = 0;
+    }
+    r = psfmt32[ctx->idx++];
+    return r;
+}
+
+/* Generate a random integer in [0..limit). */
+uint32_t gen_rand32_range(sfmt_t *ctx, uint32_t limit) {
+    uint32_t ret, above;
+
+    above = 0xffffffffU - (0xffffffffU % limit);
+    while (1) {
+	ret = gen_rand32(ctx);
+	if (ret < above) {
+	    ret %= limit;
+	    break;
+	}
+    }
+    return ret;
+}
+#endif
+/**
+ * This function generates and returns 64-bit pseudorandom number.
+ * init_gen_rand or init_by_array must be called before this function.
+ * The function gen_rand64 should not be called after gen_rand32,
+ * unless an initialization is again executed.
+ * @return 64-bit pseudorandom number
+ */
+uint64_t gen_rand64(sfmt_t *ctx) {
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+    uint32_t r1, r2;
+    uint32_t *psfmt32 = &ctx->sfmt[0].u[0];
+#else
+    uint64_t r;
+    uint64_t *psfmt64 = (uint64_t *)&ctx->sfmt[0].u[0];
+#endif
+
+    assert(ctx->initialized);
+    assert(ctx->idx % 2 == 0);
+
+    if (ctx->idx >= N32) {
+	gen_rand_all(ctx);
+	ctx->idx = 0;
+    }
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+    r1 = psfmt32[ctx->idx];
+    r2 = psfmt32[ctx->idx + 1];
+    ctx->idx += 2;
+    return ((uint64_t)r2 << 32) | r1;
+#else
+    r = psfmt64[ctx->idx / 2];
+    ctx->idx += 2;
+    return r;
+#endif
+}
+
+/* Generate a random integer in [0..limit). */
+uint64_t gen_rand64_range(sfmt_t *ctx, uint64_t limit) {
+    uint64_t ret, above;
+
+    above = KQU(0xffffffffffffffff) - (KQU(0xffffffffffffffff) % limit);
+    while (1) {
+	ret = gen_rand64(ctx);
+	if (ret < above) {
+	    ret %= limit;
+	    break;
+	}
+    }
+    return ret;
+}
+
+#ifndef ONLY64
+/**
+ * This function generates pseudorandom 32-bit integers in the
+ * specified array[] by one call. The number of pseudorandom integers
+ * is specified by the argument size, which must be at least 624 and a
+ * multiple of four.  The generation by this function is much faster
+ * than the following gen_rand function.
+ *
+ * For initialization, init_gen_rand or init_by_array must be called
+ * before the first call of this function. This function can not be
+ * used after calling gen_rand function, without initialization.
+ *
+ * @param array an array where pseudorandom 32-bit integers are filled
+ * by this function.  The pointer to the array must be \b "aligned"
+ * (namely, must be a multiple of 16) in the SIMD version, since it
+ * refers to the address of a 128-bit integer.  In the standard C
+ * version, the pointer is arbitrary.
+ *
+ * @param size the number of 32-bit pseudorandom integers to be
+ * generated.  size must be a multiple of 4, and greater than or equal
+ * to (MEXP / 128 + 1) * 4.
+ *
+ * @note \b memalign or \b posix_memalign is available to get aligned
+ * memory. Mac OSX doesn't have these functions, but \b malloc of OSX
+ * returns the pointer to the aligned memory block.
+ */
+void fill_array32(sfmt_t *ctx, uint32_t *array, int size) {
+    assert(ctx->initialized);
+    assert(ctx->idx == N32);
+    assert(size % 4 == 0);
+    assert(size >= N32);
+
+    gen_rand_array(ctx, (w128_t *)array, size / 4);
+    ctx->idx = N32;
+}
+#endif
+
+/**
+ * This function generates pseudorandom 64-bit integers in the
+ * specified array[] by one call. The number of pseudorandom integers
+ * is specified by the argument size, which must be at least 312 and a
+ * multiple of two.  The generation by this function is much faster
+ * than the following gen_rand function.
+ *
+ * For initialization, init_gen_rand or init_by_array must be called
+ * before the first call of this function. This function can not be
+ * used after calling gen_rand function, without initialization.
+ *
+ * @param array an array where pseudorandom 64-bit integers are filled
+ * by this function.  The pointer to the array must be "aligned"
+ * (namely, must be a multiple of 16) in the SIMD version, since it
+ * refers to the address of a 128-bit integer.  In the standard C
+ * version, the pointer is arbitrary.
+ *
+ * @param size the number of 64-bit pseudorandom integers to be
+ * generated.  size must be a multiple of 2, and greater than or equal
+ * to (MEXP / 128 + 1) * 2
+ *
+ * @note \b memalign or \b posix_memalign is available to get aligned
+ * memory. Mac OSX doesn't have these functions, but \b malloc of OSX
+ * returns the pointer to the aligned memory block.
+ */
+void fill_array64(sfmt_t *ctx, uint64_t *array, int size) {
+    assert(ctx->initialized);
+    assert(ctx->idx == N32);
+    assert(size % 2 == 0);
+    assert(size >= N64);
+
+    gen_rand_array(ctx, (w128_t *)array, size / 2);
+    ctx->idx = N32;
+
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+    swap((w128_t *)array, size /2);
+#endif
+}
+
+/**
+ * This function initializes the internal state array with a 32-bit
+ * integer seed.
+ *
+ * @param seed a 32-bit integer used as the seed.
+ */
+sfmt_t *init_gen_rand(uint32_t seed) {
+    void *p;
+    sfmt_t *ctx;
+    int i;
+    uint32_t *psfmt32;
+
+    if (posix_memalign(&p, sizeof(w128_t), sizeof(sfmt_t)) != 0) {
+	return NULL;
+    }
+    ctx = (sfmt_t *)p;
+    psfmt32 = &ctx->sfmt[0].u[0];
+
+    psfmt32[idxof(0)] = seed;
+    for (i = 1; i < N32; i++) {
+	psfmt32[idxof(i)] = 1812433253UL * (psfmt32[idxof(i - 1)]
+					    ^ (psfmt32[idxof(i - 1)] >> 30))
+	    + i;
+    }
+    ctx->idx = N32;
+    period_certification(ctx);
+    ctx->initialized = 1;
+
+    return ctx;
+}
+
+/**
+ * This function initializes the internal state array,
+ * with an array of 32-bit integers used as the seeds
+ * @param init_key the array of 32-bit integers, used as a seed.
+ * @param key_length the length of init_key.
+ */
+sfmt_t *init_by_array(uint32_t *init_key, int key_length) {
+    void *p;
+    sfmt_t *ctx;
+    int i, j, count;
+    uint32_t r;
+    int lag;
+    int mid;
+    int size = N * 4;
+    uint32_t *psfmt32;
+
+    if (posix_memalign(&p, sizeof(w128_t), sizeof(sfmt_t)) != 0) {
+	return NULL;
+    }
+    ctx = (sfmt_t *)p;
+    psfmt32 = &ctx->sfmt[0].u[0];
+
+    if (size >= 623) {
+	lag = 11;
+    } else if (size >= 68) {
+	lag = 7;
+    } else if (size >= 39) {
+	lag = 5;
+    } else {
+	lag = 3;
+    }
+    mid = (size - lag) / 2;
+
+    memset(ctx->sfmt, 0x8b, sizeof(ctx->sfmt));
+    if (key_length + 1 > N32) {
+	count = key_length + 1;
+    } else {
+	count = N32;
+    }
+    r = func1(psfmt32[idxof(0)] ^ psfmt32[idxof(mid)]
+	      ^ psfmt32[idxof(N32 - 1)]);
+    psfmt32[idxof(mid)] += r;
+    r += key_length;
+    psfmt32[idxof(mid + lag)] += r;
+    psfmt32[idxof(0)] = r;
+
+    count--;
+    for (i = 1, j = 0; (j < count) && (j < key_length); j++) {
+	r = func1(psfmt32[idxof(i)] ^ psfmt32[idxof((i + mid) % N32)]
+		  ^ psfmt32[idxof((i + N32 - 1) % N32)]);
+	psfmt32[idxof((i + mid) % N32)] += r;
+	r += init_key[j] + i;
+	psfmt32[idxof((i + mid + lag) % N32)] += r;
+	psfmt32[idxof(i)] = r;
+	i = (i + 1) % N32;
+    }
+    for (; j < count; j++) {
+	r = func1(psfmt32[idxof(i)] ^ psfmt32[idxof((i + mid) % N32)]
+		  ^ psfmt32[idxof((i + N32 - 1) % N32)]);
+	psfmt32[idxof((i + mid) % N32)] += r;
+	r += i;
+	psfmt32[idxof((i + mid + lag) % N32)] += r;
+	psfmt32[idxof(i)] = r;
+	i = (i + 1) % N32;
+    }
+    for (j = 0; j < N32; j++) {
+	r = func2(psfmt32[idxof(i)] + psfmt32[idxof((i + mid) % N32)]
+		  + psfmt32[idxof((i + N32 - 1) % N32)]);
+	psfmt32[idxof((i + mid) % N32)] ^= r;
+	r -= i;
+	psfmt32[idxof((i + mid + lag) % N32)] ^= r;
+	psfmt32[idxof(i)] = r;
+	i = (i + 1) % N32;
+    }
+
+    ctx->idx = N32;
+    period_certification(ctx);
+    ctx->initialized = 1;
+
+    return ctx;
+}
+
+void fini_gen_rand(sfmt_t *ctx) {
+    assert(ctx != NULL);
+
+    ctx->initialized = 0;
+    free(ctx);
+}
diff --git a/jemalloc/test/src/btalloc.c b/jemalloc/test/src/btalloc.c
new file mode 100644
index 0000000..d570952
--- /dev/null
+++ b/jemalloc/test/src/btalloc.c
@@ -0,0 +1,6 @@
+#include "test/jemalloc_test.h"
+
+void *
+btalloc(size_t size, unsigned bits) {
+	return btalloc_0(size, bits);
+}
diff --git a/jemalloc/test/src/btalloc_0.c b/jemalloc/test/src/btalloc_0.c
new file mode 100644
index 0000000..77d8904
--- /dev/null
+++ b/jemalloc/test/src/btalloc_0.c
@@ -0,0 +1,3 @@
+#include "test/jemalloc_test.h"
+
+btalloc_n_gen(0)
diff --git a/jemalloc/test/src/btalloc_1.c b/jemalloc/test/src/btalloc_1.c
new file mode 100644
index 0000000..4c126c3
--- /dev/null
+++ b/jemalloc/test/src/btalloc_1.c
@@ -0,0 +1,3 @@
+#include "test/jemalloc_test.h"
+
+btalloc_n_gen(1)
diff --git a/jemalloc/test/src/math.c b/jemalloc/test/src/math.c
new file mode 100644
index 0000000..1758c67
--- /dev/null
+++ b/jemalloc/test/src/math.c
@@ -0,0 +1,2 @@
+#define MATH_C_
+#include "test/jemalloc_test.h"
diff --git a/jemalloc/test/src/mq.c b/jemalloc/test/src/mq.c
new file mode 100644
index 0000000..9b5f672
--- /dev/null
+++ b/jemalloc/test/src/mq.c
@@ -0,0 +1,27 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * Sleep for approximately ns nanoseconds.  No lower *nor* upper bound on sleep
+ * time is guaranteed.
+ */
+void
+mq_nanosleep(unsigned ns) {
+	assert(ns <= 1000*1000*1000);
+
+#ifdef _WIN32
+	Sleep(ns / 1000);
+#else
+	{
+		struct timespec timeout;
+
+		if (ns < 1000*1000*1000) {
+			timeout.tv_sec = 0;
+			timeout.tv_nsec = ns;
+		} else {
+			timeout.tv_sec = 1;
+			timeout.tv_nsec = 0;
+		}
+		nanosleep(&timeout, NULL);
+	}
+#endif
+}
diff --git a/jemalloc/test/src/mtx.c b/jemalloc/test/src/mtx.c
new file mode 100644
index 0000000..d9ce375
--- /dev/null
+++ b/jemalloc/test/src/mtx.c
@@ -0,0 +1,61 @@
+#include "test/jemalloc_test.h"
+
+#ifndef _CRT_SPINCOUNT
+#define _CRT_SPINCOUNT 4000
+#endif
+
+bool
+mtx_init(mtx_t *mtx) {
+#ifdef _WIN32
+	if (!InitializeCriticalSectionAndSpinCount(&mtx->lock,
+	    _CRT_SPINCOUNT)) {
+		return true;
+	}
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	mtx->lock = OS_UNFAIR_LOCK_INIT;
+#else
+	pthread_mutexattr_t attr;
+
+	if (pthread_mutexattr_init(&attr) != 0) {
+		return true;
+	}
+	pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_DEFAULT);
+	if (pthread_mutex_init(&mtx->lock, &attr) != 0) {
+		pthread_mutexattr_destroy(&attr);
+		return true;
+	}
+	pthread_mutexattr_destroy(&attr);
+#endif
+	return false;
+}
+
+void
+mtx_fini(mtx_t *mtx) {
+#ifdef _WIN32
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#else
+	pthread_mutex_destroy(&mtx->lock);
+#endif
+}
+
+void
+mtx_lock(mtx_t *mtx) {
+#ifdef _WIN32
+	EnterCriticalSection(&mtx->lock);
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	os_unfair_lock_lock(&mtx->lock);
+#else
+	pthread_mutex_lock(&mtx->lock);
+#endif
+}
+
+void
+mtx_unlock(mtx_t *mtx) {
+#ifdef _WIN32
+	LeaveCriticalSection(&mtx->lock);
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	os_unfair_lock_unlock(&mtx->lock);
+#else
+	pthread_mutex_unlock(&mtx->lock);
+#endif
+}
diff --git a/jemalloc/test/src/test.c b/jemalloc/test/src/test.c
new file mode 100644
index 0000000..f97ce4d
--- /dev/null
+++ b/jemalloc/test/src/test.c
@@ -0,0 +1,234 @@
+#include "test/jemalloc_test.h"
+
+/* Test status state. */
+
+static unsigned		test_count = 0;
+static test_status_t	test_counts[test_status_count] = {0, 0, 0};
+static test_status_t	test_status = test_status_pass;
+static const char *	test_name = "";
+
+/* Reentrancy testing helpers. */
+
+#define NUM_REENTRANT_ALLOCS 20
+typedef enum {
+	non_reentrant = 0,
+	libc_reentrant = 1,
+	arena_new_reentrant = 2
+} reentrancy_t;
+static reentrancy_t reentrancy;
+
+static bool libc_hook_ran = false;
+static bool arena_new_hook_ran = false;
+
+static const char *
+reentrancy_t_str(reentrancy_t r) {
+	switch (r) {
+	case non_reentrant:
+		return "non-reentrant";
+	case libc_reentrant:
+		return "libc-reentrant";
+	case arena_new_reentrant:
+		return "arena_new-reentrant";
+	default:
+		unreachable();
+	}
+}
+
+static void
+do_hook(bool *hook_ran, void (**hook)()) {
+	*hook_ran = true;
+	*hook = NULL;
+
+	size_t alloc_size = 1;
+	for (int i = 0; i < NUM_REENTRANT_ALLOCS; i++) {
+		free(malloc(alloc_size));
+		alloc_size *= 2;
+	}
+}
+
+static void
+libc_reentrancy_hook() {
+	do_hook(&libc_hook_ran, &test_hooks_libc_hook);
+}
+
+static void
+arena_new_reentrancy_hook() {
+	do_hook(&arena_new_hook_ran, &test_hooks_arena_new_hook);
+}
+
+/* Actual test infrastructure. */
+bool
+test_is_reentrant() {
+	return reentrancy != non_reentrant;
+}
+
+JEMALLOC_FORMAT_PRINTF(1, 2)
+void
+test_skip(const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+	malloc_printf("\n");
+	test_status = test_status_skip;
+}
+
+JEMALLOC_FORMAT_PRINTF(1, 2)
+void
+test_fail(const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+	malloc_printf("\n");
+	test_status = test_status_fail;
+}
+
+static const char *
+test_status_string(test_status_t test_status) {
+	switch (test_status) {
+	case test_status_pass: return "pass";
+	case test_status_skip: return "skip";
+	case test_status_fail: return "fail";
+	default: not_reached();
+	}
+}
+
+void
+p_test_init(const char *name) {
+	test_count++;
+	test_status = test_status_pass;
+	test_name = name;
+}
+
+void
+p_test_fini(void) {
+	test_counts[test_status]++;
+	malloc_printf("%s (%s): %s\n", test_name, reentrancy_t_str(reentrancy),
+	    test_status_string(test_status));
+}
+
+static void
+check_global_slow(test_status_t *status) {
+#ifdef JEMALLOC_UNIT_TEST
+	/*
+	 * This check needs to peek into tsd internals, which is why it's only
+	 * exposed in unit tests.
+	 */
+	if (tsd_global_slow()) {
+		malloc_printf("Testing increased global slow count\n");
+		*status = test_status_fail;
+	}
+#endif
+}
+
+static test_status_t
+p_test_impl(bool do_malloc_init, bool do_reentrant, test_t *t, va_list ap) {
+	test_status_t ret;
+
+	if (do_malloc_init) {
+		/*
+		 * Make sure initialization occurs prior to running tests.
+		 * Tests are special because they may use internal facilities
+		 * prior to triggering initialization as a side effect of
+		 * calling into the public API.
+		 */
+		if (nallocx(1, 0) == 0) {
+			malloc_printf("Initialization error");
+			return test_status_fail;
+		}
+	}
+
+	ret = test_status_pass;
+	for (; t != NULL; t = va_arg(ap, test_t *)) {
+		/* Non-reentrant run. */
+		reentrancy = non_reentrant;
+		test_hooks_arena_new_hook = test_hooks_libc_hook = NULL;
+		t();
+		if (test_status > ret) {
+			ret = test_status;
+		}
+		check_global_slow(&ret);
+		/* Reentrant run. */
+		if (do_reentrant) {
+			reentrancy = libc_reentrant;
+			test_hooks_arena_new_hook = NULL;
+			test_hooks_libc_hook = &libc_reentrancy_hook;
+			t();
+			if (test_status > ret) {
+				ret = test_status;
+			}
+			check_global_slow(&ret);
+
+			reentrancy = arena_new_reentrant;
+			test_hooks_libc_hook = NULL;
+			test_hooks_arena_new_hook = &arena_new_reentrancy_hook;
+			t();
+			if (test_status > ret) {
+				ret = test_status;
+			}
+			check_global_slow(&ret);
+		}
+	}
+
+	malloc_printf("--- %s: %u/%u, %s: %u/%u, %s: %u/%u ---\n",
+	    test_status_string(test_status_pass),
+	    test_counts[test_status_pass], test_count,
+	    test_status_string(test_status_skip),
+	    test_counts[test_status_skip], test_count,
+	    test_status_string(test_status_fail),
+	    test_counts[test_status_fail], test_count);
+
+	return ret;
+}
+
+test_status_t
+p_test(test_t *t, ...) {
+	test_status_t ret;
+	va_list ap;
+
+	ret = test_status_pass;
+	va_start(ap, t);
+	ret = p_test_impl(true, true, t, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+test_status_t
+p_test_no_reentrancy(test_t *t, ...) {
+	test_status_t ret;
+	va_list ap;
+
+	ret = test_status_pass;
+	va_start(ap, t);
+	ret = p_test_impl(true, false, t, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+test_status_t
+p_test_no_malloc_init(test_t *t, ...) {
+	test_status_t ret;
+	va_list ap;
+
+	ret = test_status_pass;
+	va_start(ap, t);
+	/*
+	 * We also omit reentrancy from bootstrapping tests, since we don't
+	 * (yet) care about general reentrancy during bootstrapping.
+	 */
+	ret = p_test_impl(false, false, t, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+void
+p_test_fail(const char *prefix, const char *message) {
+	malloc_cprintf(NULL, NULL, "%s%s\n", prefix, message);
+	test_status = test_status_fail;
+}
diff --git a/jemalloc/test/src/thd.c b/jemalloc/test/src/thd.c
new file mode 100644
index 0000000..9a15eab
--- /dev/null
+++ b/jemalloc/test/src/thd.c
@@ -0,0 +1,34 @@
+#include "test/jemalloc_test.h"
+
+#ifdef _WIN32
+void
+thd_create(thd_t *thd, void *(*proc)(void *), void *arg) {
+	LPTHREAD_START_ROUTINE routine = (LPTHREAD_START_ROUTINE)proc;
+	*thd = CreateThread(NULL, 0, routine, arg, 0, NULL);
+	if (*thd == NULL) {
+		test_fail("Error in CreateThread()\n");
+	}
+}
+
+void
+thd_join(thd_t thd, void **ret) {
+	if (WaitForSingleObject(thd, INFINITE) == WAIT_OBJECT_0 && ret) {
+		DWORD exit_code;
+		GetExitCodeThread(thd, (LPDWORD) &exit_code);
+		*ret = (void *)(uintptr_t)exit_code;
+	}
+}
+
+#else
+void
+thd_create(thd_t *thd, void *(*proc)(void *), void *arg) {
+	if (pthread_create(thd, NULL, proc, arg) != 0) {
+		test_fail("Error in pthread_create()\n");
+	}
+}
+
+void
+thd_join(thd_t thd, void **ret) {
+	pthread_join(thd, ret);
+}
+#endif
diff --git a/jemalloc/test/src/timer.c b/jemalloc/test/src/timer.c
new file mode 100644
index 0000000..c451c63
--- /dev/null
+++ b/jemalloc/test/src/timer.c
@@ -0,0 +1,56 @@
+#include "test/jemalloc_test.h"
+
+void
+timer_start(timedelta_t *timer) {
+	nstime_init(&timer->t0, 0);
+	nstime_update(&timer->t0);
+}
+
+void
+timer_stop(timedelta_t *timer) {
+	nstime_copy(&timer->t1, &timer->t0);
+	nstime_update(&timer->t1);
+}
+
+uint64_t
+timer_usec(const timedelta_t *timer) {
+	nstime_t delta;
+
+	nstime_copy(&delta, &timer->t1);
+	nstime_subtract(&delta, &timer->t0);
+	return nstime_ns(&delta) / 1000;
+}
+
+void
+timer_ratio(timedelta_t *a, timedelta_t *b, char *buf, size_t buflen) {
+	uint64_t t0 = timer_usec(a);
+	uint64_t t1 = timer_usec(b);
+	uint64_t mult;
+	size_t i = 0;
+	size_t j, n;
+
+	/* Whole. */
+	n = malloc_snprintf(&buf[i], buflen-i, "%"FMTu64, t0 / t1);
+	i += n;
+	if (i >= buflen) {
+		return;
+	}
+	mult = 1;
+	for (j = 0; j < n; j++) {
+		mult *= 10;
+	}
+
+	/* Decimal. */
+	n = malloc_snprintf(&buf[i], buflen-i, ".");
+	i += n;
+
+	/* Fraction. */
+	while (i < buflen-1) {
+		uint64_t round = (i+1 == buflen-1 && ((t0 * mult * 10 / t1) % 10
+		    >= 5)) ? 1 : 0;
+		n = malloc_snprintf(&buf[i], buflen-i,
+		    "%"FMTu64, (t0 * mult / t1) % 10 + round);
+		i += n;
+		mult *= 10;
+	}
+}
diff --git a/jemalloc/test/stress/hookbench.c b/jemalloc/test/stress/hookbench.c
new file mode 100644
index 0000000..97e90b0
--- /dev/null
+++ b/jemalloc/test/stress/hookbench.c
@@ -0,0 +1,73 @@
+#include "test/jemalloc_test.h"
+
+static void
+noop_alloc_hook(void *extra, hook_alloc_t type, void *result,
+    uintptr_t result_raw, uintptr_t args_raw[3]) {
+}
+
+static void
+noop_dalloc_hook(void *extra, hook_dalloc_t type, void *address,
+    uintptr_t args_raw[3]) {
+}
+
+static void
+noop_expand_hook(void *extra, hook_expand_t type, void *address,
+    size_t old_usize, size_t new_usize, uintptr_t result_raw,
+    uintptr_t args_raw[4]) {
+}
+
+static void
+malloc_free_loop(int iters) {
+	for (int i = 0; i < iters; i++) {
+		void *p = mallocx(1, 0);
+		free(p);
+	}
+}
+
+static void
+test_hooked(int iters) {
+	hooks_t hooks = {&noop_alloc_hook, &noop_dalloc_hook, &noop_expand_hook,
+		NULL};
+
+	int err;
+	void *handles[HOOK_MAX];
+	size_t sz = sizeof(handles[0]);
+
+	for (int i = 0; i < HOOK_MAX; i++) {
+		err = mallctl("experimental.hooks.install", &handles[i],
+		    &sz, &hooks, sizeof(hooks));
+		assert(err == 0);
+
+		timedelta_t timer;
+		timer_start(&timer);
+		malloc_free_loop(iters);
+		timer_stop(&timer);
+		malloc_printf("With %d hook%s: %"FMTu64"us\n", i + 1,
+		    i + 1 == 1 ? "" : "s", timer_usec(&timer));
+	}
+	for (int i = 0; i < HOOK_MAX; i++) {
+		err = mallctl("experimental.hooks.remove", NULL, NULL,
+		    &handles[i], sizeof(handles[i]));
+		assert(err == 0);
+	}
+}
+
+static void
+test_unhooked(int iters) {
+	timedelta_t timer;
+	timer_start(&timer);
+	malloc_free_loop(iters);
+	timer_stop(&timer);
+
+	malloc_printf("Without hooks: %"FMTu64"us\n", timer_usec(&timer));
+}
+
+int
+main(void) {
+	/* Initialize */
+	free(mallocx(1, 0));
+	int iters = 10 * 1000 * 1000;
+	malloc_printf("Benchmarking hooks with %d iterations:\n", iters);
+	test_hooked(iters);
+	test_unhooked(iters);
+}
diff --git a/jemalloc/test/stress/microbench.c b/jemalloc/test/stress/microbench.c
new file mode 100644
index 0000000..988b793
--- /dev/null
+++ b/jemalloc/test/stress/microbench.c
@@ -0,0 +1,165 @@
+#include "test/jemalloc_test.h"
+
+static inline void
+time_func(timedelta_t *timer, uint64_t nwarmup, uint64_t niter,
+    void (*func)(void)) {
+	uint64_t i;
+
+	for (i = 0; i < nwarmup; i++) {
+		func();
+	}
+	timer_start(timer);
+	for (i = 0; i < niter; i++) {
+		func();
+	}
+	timer_stop(timer);
+}
+
+void
+compare_funcs(uint64_t nwarmup, uint64_t niter, const char *name_a,
+    void (*func_a), const char *name_b, void (*func_b)) {
+	timedelta_t timer_a, timer_b;
+	char ratio_buf[6];
+	void *p;
+
+	p = mallocx(1, 0);
+	if (p == NULL) {
+		test_fail("Unexpected mallocx() failure");
+		return;
+	}
+
+	time_func(&timer_a, nwarmup, niter, func_a);
+	time_func(&timer_b, nwarmup, niter, func_b);
+
+	timer_ratio(&timer_a, &timer_b, ratio_buf, sizeof(ratio_buf));
+	malloc_printf("%"FMTu64" iterations, %s=%"FMTu64"us, "
+	    "%s=%"FMTu64"us, ratio=1:%s\n",
+	    niter, name_a, timer_usec(&timer_a), name_b, timer_usec(&timer_b),
+	    ratio_buf);
+
+	dallocx(p, 0);
+}
+
+static void
+malloc_free(void) {
+	/* The compiler can optimize away free(malloc(1))! */
+	void *p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	free(p);
+}
+
+static void
+mallocx_free(void) {
+	void *p = mallocx(1, 0);
+	if (p == NULL) {
+		test_fail("Unexpected mallocx() failure");
+		return;
+	}
+	free(p);
+}
+
+TEST_BEGIN(test_malloc_vs_mallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "malloc",
+	    malloc_free, "mallocx", mallocx_free);
+}
+TEST_END
+
+static void
+malloc_dallocx(void) {
+	void *p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	dallocx(p, 0);
+}
+
+static void
+malloc_sdallocx(void) {
+	void *p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	sdallocx(p, 1, 0);
+}
+
+TEST_BEGIN(test_free_vs_dallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "free", malloc_free,
+	    "dallocx", malloc_dallocx);
+}
+TEST_END
+
+TEST_BEGIN(test_dallocx_vs_sdallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "dallocx", malloc_dallocx,
+	    "sdallocx", malloc_sdallocx);
+}
+TEST_END
+
+static void
+malloc_mus_free(void) {
+	void *p;
+
+	p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	malloc_usable_size(p);
+	free(p);
+}
+
+static void
+malloc_sallocx_free(void) {
+	void *p;
+
+	p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	if (sallocx(p, 0) < 1) {
+		test_fail("Unexpected sallocx() failure");
+	}
+	free(p);
+}
+
+TEST_BEGIN(test_mus_vs_sallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "malloc_usable_size",
+	    malloc_mus_free, "sallocx", malloc_sallocx_free);
+}
+TEST_END
+
+static void
+malloc_nallocx_free(void) {
+	void *p;
+
+	p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	if (nallocx(1, 0) < 1) {
+		test_fail("Unexpected nallocx() failure");
+	}
+	free(p);
+}
+
+TEST_BEGIN(test_sallocx_vs_nallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "sallocx",
+	    malloc_sallocx_free, "nallocx", malloc_nallocx_free);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_malloc_vs_mallocx,
+	    test_free_vs_dallocx,
+	    test_dallocx_vs_sdallocx,
+	    test_mus_vs_sallocx,
+	    test_sallocx_vs_nallocx);
+}
diff --git a/jemalloc/test/test.sh.in b/jemalloc/test/test.sh.in
new file mode 100644
index 0000000..39302ff
--- /dev/null
+++ b/jemalloc/test/test.sh.in
@@ -0,0 +1,80 @@
+#!/bin/sh
+
+case @abi@ in
+  macho)
+    export DYLD_FALLBACK_LIBRARY_PATH="@objroot@lib"
+    ;;
+  pecoff)
+    export PATH="${PATH}:@objroot@lib"
+    ;;
+  *)
+    ;;
+esac
+
+# Make a copy of the @JEMALLOC_CPREFIX@MALLOC_CONF passed in to this script, so
+# it can be repeatedly concatenated with per test settings.
+export MALLOC_CONF_ALL=${@JEMALLOC_CPREFIX@MALLOC_CONF}
+# Concatenate the individual test's MALLOC_CONF and MALLOC_CONF_ALL.
+export_malloc_conf() {
+  if [ "x${MALLOC_CONF}" != "x" -a "x${MALLOC_CONF_ALL}" != "x" ] ; then
+    export @JEMALLOC_CPREFIX@MALLOC_CONF="${MALLOC_CONF},${MALLOC_CONF_ALL}"
+  else
+    export @JEMALLOC_CPREFIX@MALLOC_CONF="${MALLOC_CONF}${MALLOC_CONF_ALL}"
+  fi
+}
+
+# Corresponds to test_status_t.
+pass_code=0
+skip_code=1
+fail_code=2
+
+pass_count=0
+skip_count=0
+fail_count=0
+for t in $@; do
+  if [ $pass_count -ne 0 -o $skip_count -ne 0 -o $fail_count != 0 ] ; then
+    echo
+  fi
+  echo "=== ${t} ==="
+  if [ -e "@srcroot@${t}.sh" ] ; then
+    # Source the shell script corresponding to the test in a subshell and
+    # execute the test.  This allows the shell script to set MALLOC_CONF, which
+    # is then used to set @JEMALLOC_CPREFIX@MALLOC_CONF (thus allowing the
+    # per test shell script to ignore the @JEMALLOC_CPREFIX@ detail).
+    enable_fill=@enable_fill@ \
+    enable_prof=@enable_prof@ \
+    . @srcroot@${t}.sh && \
+    export_malloc_conf && \
+    $JEMALLOC_TEST_PREFIX ${t}@exe@ @abs_srcroot@ @abs_objroot@
+  else
+    export MALLOC_CONF= && \
+    export_malloc_conf && \
+    $JEMALLOC_TEST_PREFIX ${t}@exe@ @abs_srcroot@ @abs_objroot@
+  fi
+  result_code=$?
+  case ${result_code} in
+    ${pass_code})
+      pass_count=$((pass_count+1))
+      ;;
+    ${skip_code})
+      skip_count=$((skip_count+1))
+      ;;
+    ${fail_code})
+      fail_count=$((fail_count+1))
+      ;;
+    *)
+      echo "Test harness error: ${t} w/ MALLOC_CONF=\"${MALLOC_CONF}\"" 1>&2
+      echo "Use prefix to debug, e.g. JEMALLOC_TEST_PREFIX=\"gdb --args\" sh test/test.sh ${t}" 1>&2
+      exit 1
+  esac
+done
+
+total_count=`expr ${pass_count} + ${skip_count} + ${fail_count}`
+echo
+echo "Test suite summary: pass: ${pass_count}/${total_count}, skip: ${skip_count}/${total_count}, fail: ${fail_count}/${total_count}"
+
+if [ ${fail_count} -eq 0 ] ; then
+  exit 0
+else
+  exit 1
+fi
diff --git a/jemalloc/test/unit/SFMT.c b/jemalloc/test/unit/SFMT.c
new file mode 100644
index 0000000..1fc8cf1
--- /dev/null
+++ b/jemalloc/test/unit/SFMT.c
@@ -0,0 +1,1599 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "test/jemalloc_test.h"
+
+#define BLOCK_SIZE 10000
+#define BLOCK_SIZE64 (BLOCK_SIZE / 2)
+#define COUNT_1 1000
+#define COUNT_2 700
+
+static const uint32_t init_gen_rand_32_expected[] = {
+	3440181298U, 1564997079U, 1510669302U, 2930277156U, 1452439940U,
+	3796268453U,  423124208U, 2143818589U, 3827219408U, 2987036003U,
+	2674978610U, 1536842514U, 2027035537U, 2534897563U, 1686527725U,
+	 545368292U, 1489013321U, 1370534252U, 4231012796U, 3994803019U,
+	1764869045U,  824597505U,  862581900U, 2469764249U,  812862514U,
+	 359318673U,  116957936U, 3367389672U, 2327178354U, 1898245200U,
+	3206507879U, 2378925033U, 1040214787U, 2524778605U, 3088428700U,
+	1417665896U,  964324147U, 2282797708U, 2456269299U,  313400376U,
+	2245093271U, 1015729427U, 2694465011U, 3246975184U, 1992793635U,
+	 463679346U, 3721104591U, 3475064196U,  856141236U, 1499559719U,
+	3522818941U, 3721533109U, 1954826617U, 1282044024U, 1543279136U,
+	1301863085U, 2669145051U, 4221477354U, 3896016841U, 3392740262U,
+	 462466863U, 1037679449U, 1228140306U,  922298197U, 1205109853U,
+	1872938061U, 3102547608U, 2742766808U, 1888626088U, 4028039414U,
+	 157593879U, 1136901695U, 4038377686U, 3572517236U, 4231706728U,
+	2997311961U, 1189931652U, 3981543765U, 2826166703U,   87159245U,
+	1721379072U, 3897926942U, 1790395498U, 2569178939U, 1047368729U,
+	2340259131U, 3144212906U, 2301169789U, 2442885464U, 3034046771U,
+	3667880593U, 3935928400U, 2372805237U, 1666397115U, 2460584504U,
+	 513866770U, 3810869743U, 2147400037U, 2792078025U, 2941761810U,
+	3212265810U,  984692259U,  346590253U, 1804179199U, 3298543443U,
+	 750108141U, 2880257022U,  243310542U, 1869036465U, 1588062513U,
+	2983949551U, 1931450364U, 4034505847U, 2735030199U, 1628461061U,
+	2539522841U,  127965585U, 3992448871U,  913388237U,  559130076U,
+	1202933193U, 4087643167U, 2590021067U, 2256240196U, 1746697293U,
+	1013913783U, 1155864921U, 2715773730U,  915061862U, 1948766573U,
+	2322882854U, 3761119102U, 1343405684U, 3078711943U, 3067431651U,
+	3245156316U, 3588354584U, 3484623306U, 3899621563U, 4156689741U,
+	3237090058U, 3880063844U,  862416318U, 4039923869U, 2303788317U,
+	3073590536U,  701653667U, 2131530884U, 3169309950U, 2028486980U,
+	 747196777U, 3620218225U,  432016035U, 1449580595U, 2772266392U,
+	 444224948U, 1662832057U, 3184055582U, 3028331792U, 1861686254U,
+	1104864179U,  342430307U, 1350510923U, 3024656237U, 1028417492U,
+	2870772950U,  290847558U, 3675663500U,  508431529U, 4264340390U,
+	2263569913U, 1669302976U,  519511383U, 2706411211U, 3764615828U,
+	3883162495U, 4051445305U, 2412729798U, 3299405164U, 3991911166U,
+	2348767304U, 2664054906U, 3763609282U,  593943581U, 3757090046U,
+	2075338894U, 2020550814U, 4287452920U, 4290140003U, 1422957317U,
+	2512716667U, 2003485045U, 2307520103U, 2288472169U, 3940751663U,
+	4204638664U, 2892583423U, 1710068300U, 3904755993U, 2363243951U,
+	3038334120U,  547099465U,  771105860U, 3199983734U, 4282046461U,
+	2298388363U,  934810218U, 2837827901U, 3952500708U, 2095130248U,
+	3083335297U,   26885281U, 3932155283U, 1531751116U, 1425227133U,
+	 495654159U, 3279634176U, 3855562207U, 3957195338U, 4159985527U,
+	 893375062U, 1875515536U, 1327247422U, 3754140693U, 1028923197U,
+	1729880440U,  805571298U,  448971099U, 2726757106U, 2749436461U,
+	2485987104U,  175337042U, 3235477922U, 3882114302U, 2020970972U,
+	 943926109U, 2762587195U, 1904195558U, 3452650564U,  108432281U,
+	3893463573U, 3977583081U, 2636504348U, 1110673525U, 3548479841U,
+	4258854744U,  980047703U, 4057175418U, 3890008292U,  145653646U,
+	3141868989U, 3293216228U, 1194331837U, 1254570642U, 3049934521U,
+	2868313360U, 2886032750U, 1110873820U,  279553524U, 3007258565U,
+	1104807822U, 3186961098U,  315764646U, 2163680838U, 3574508994U,
+	3099755655U,  191957684U, 3642656737U, 3317946149U, 3522087636U,
+	 444526410U,  779157624U, 1088229627U, 1092460223U, 1856013765U,
+	3659877367U,  368270451U,  503570716U, 3000984671U, 2742789647U,
+	 928097709U, 2914109539U,  308843566U, 2816161253U, 3667192079U,
+	2762679057U, 3395240989U, 2928925038U, 1491465914U, 3458702834U,
+	3787782576U, 2894104823U, 1296880455U, 1253636503U,  989959407U,
+	2291560361U, 2776790436U, 1913178042U, 1584677829U,  689637520U,
+	1898406878U,  688391508U, 3385234998U,  845493284U, 1943591856U,
+	2720472050U,  222695101U, 1653320868U, 2904632120U, 4084936008U,
+	1080720688U, 3938032556U,  387896427U, 2650839632U,   99042991U,
+	1720913794U, 1047186003U, 1877048040U, 2090457659U,  517087501U,
+	4172014665U, 2129713163U, 2413533132U, 2760285054U, 4129272496U,
+	1317737175U, 2309566414U, 2228873332U, 3889671280U, 1110864630U,
+	3576797776U, 2074552772U,  832002644U, 3097122623U, 2464859298U,
+	2679603822U, 1667489885U, 3237652716U, 1478413938U, 1719340335U,
+	2306631119U,  639727358U, 3369698270U,  226902796U, 2099920751U,
+	1892289957U, 2201594097U, 3508197013U, 3495811856U, 3900381493U,
+	 841660320U, 3974501451U, 3360949056U, 1676829340U,  728899254U,
+	2047809627U, 2390948962U,  670165943U, 3412951831U, 4189320049U,
+	1911595255U, 2055363086U,  507170575U,  418219594U, 4141495280U,
+	2692088692U, 4203630654U, 3540093932U,  791986533U, 2237921051U,
+	2526864324U, 2956616642U, 1394958700U, 1983768223U, 1893373266U,
+	 591653646U,  228432437U, 1611046598U, 3007736357U, 1040040725U,
+	2726180733U, 2789804360U, 4263568405U,  829098158U, 3847722805U,
+	1123578029U, 1804276347U,  997971319U, 4203797076U, 4185199713U,
+	2811733626U, 2343642194U, 2985262313U, 1417930827U, 3759587724U,
+	1967077982U, 1585223204U, 1097475516U, 1903944948U,  740382444U,
+	1114142065U, 1541796065U, 1718384172U, 1544076191U, 1134682254U,
+	3519754455U, 2866243923U,  341865437U,  645498576U, 2690735853U,
+	1046963033U, 2493178460U, 1187604696U, 1619577821U,  488503634U,
+	3255768161U, 2306666149U, 1630514044U, 2377698367U, 2751503746U,
+	3794467088U, 1796415981U, 3657173746U,  409136296U, 1387122342U,
+	1297726519U,  219544855U, 4270285558U,  437578827U, 1444698679U,
+	2258519491U,  963109892U, 3982244073U, 3351535275U,  385328496U,
+	1804784013U,  698059346U, 3920535147U,  708331212U,  784338163U,
+	 785678147U, 1238376158U, 1557298846U, 2037809321U,  271576218U,
+	4145155269U, 1913481602U, 2763691931U,  588981080U, 1201098051U,
+	3717640232U, 1509206239U,  662536967U, 3180523616U, 1133105435U,
+	2963500837U, 2253971215U, 3153642623U, 1066925709U, 2582781958U,
+	3034720222U, 1090798544U, 2942170004U, 4036187520U,  686972531U,
+	2610990302U, 2641437026U, 1837562420U,  722096247U, 1315333033U,
+	2102231203U, 3402389208U, 3403698140U, 1312402831U, 2898426558U,
+	 814384596U,  385649582U, 1916643285U, 1924625106U, 2512905582U,
+	2501170304U, 4275223366U, 2841225246U, 1467663688U, 3563567847U,
+	2969208552U,  884750901U,  102992576U,  227844301U, 3681442994U,
+	3502881894U, 4034693299U, 1166727018U, 1697460687U, 1737778332U,
+	1787161139U, 1053003655U, 1215024478U, 2791616766U, 2525841204U,
+	1629323443U,    3233815U, 2003823032U, 3083834263U, 2379264872U,
+	3752392312U, 1287475550U, 3770904171U, 3004244617U, 1502117784U,
+	 918698423U, 2419857538U, 3864502062U, 1751322107U, 2188775056U,
+	4018728324U,  983712955U,  440071928U, 3710838677U, 2001027698U,
+	3994702151U,   22493119U, 3584400918U, 3446253670U, 4254789085U,
+	1405447860U, 1240245579U, 1800644159U, 1661363424U, 3278326132U,
+	3403623451U,   67092802U, 2609352193U, 3914150340U, 1814842761U,
+	3610830847U,  591531412U, 3880232807U, 1673505890U, 2585326991U,
+	1678544474U, 3148435887U, 3457217359U, 1193226330U, 2816576908U,
+	 154025329U,  121678860U, 1164915738U,  973873761U,  269116100U,
+	  52087970U,  744015362U,  498556057U,   94298882U, 1563271621U,
+	2383059628U, 4197367290U, 3958472990U, 2592083636U, 2906408439U,
+	1097742433U, 3924840517U,  264557272U, 2292287003U, 3203307984U,
+	4047038857U, 3820609705U, 2333416067U, 1839206046U, 3600944252U,
+	3412254904U,  583538222U, 2390557166U, 4140459427U, 2810357445U,
+	 226777499U, 2496151295U, 2207301712U, 3283683112U,  611630281U,
+	1933218215U, 3315610954U, 3889441987U, 3719454256U, 3957190521U,
+	1313998161U, 2365383016U, 3146941060U, 1801206260U,  796124080U,
+	2076248581U, 1747472464U, 3254365145U,  595543130U, 3573909503U,
+	3758250204U, 2020768540U, 2439254210U,   93368951U, 3155792250U,
+	2600232980U, 3709198295U, 3894900440U, 2971850836U, 1578909644U,
+	1443493395U, 2581621665U, 3086506297U, 2443465861U,  558107211U,
+	1519367835U,  249149686U,  908102264U, 2588765675U, 1232743965U,
+	1001330373U, 3561331654U, 2259301289U, 1564977624U, 3835077093U,
+	 727244906U, 4255738067U, 1214133513U, 2570786021U, 3899704621U,
+	1633861986U, 1636979509U, 1438500431U,   58463278U, 2823485629U,
+	2297430187U, 2926781924U, 3371352948U, 1864009023U, 2722267973U,
+	1444292075U,  437703973U, 1060414512U,  189705863U,  910018135U,
+	4077357964U,  884213423U, 2644986052U, 3973488374U, 1187906116U,
+	2331207875U,  780463700U, 3713351662U, 3854611290U,  412805574U,
+	2978462572U, 2176222820U,  829424696U, 2790788332U, 2750819108U,
+	1594611657U, 3899878394U, 3032870364U, 1702887682U, 1948167778U,
+	  14130042U,  192292500U,  947227076U,   90719497U, 3854230320U,
+	 784028434U, 2142399787U, 1563449646U, 2844400217U,  819143172U,
+	2883302356U, 2328055304U, 1328532246U, 2603885363U, 3375188924U,
+	 933941291U, 3627039714U, 2129697284U, 2167253953U, 2506905438U,
+	1412424497U, 2981395985U, 1418359660U, 2925902456U,   52752784U,
+	3713667988U, 3924669405U,  648975707U, 1145520213U, 4018650664U,
+	3805915440U, 2380542088U, 2013260958U, 3262572197U, 2465078101U,
+	1114540067U, 3728768081U, 2396958768U,  590672271U,  904818725U,
+	4263660715U,  700754408U, 1042601829U, 4094111823U, 4274838909U,
+	2512692617U, 2774300207U, 2057306915U, 3470942453U,   99333088U,
+	1142661026U, 2889931380U,   14316674U, 2201179167U,  415289459U,
+	 448265759U, 3515142743U, 3254903683U,  246633281U, 1184307224U,
+	2418347830U, 2092967314U, 2682072314U, 2558750234U, 2000352263U,
+	1544150531U,  399010405U, 1513946097U,  499682937U,  461167460U,
+	3045570638U, 1633669705U,  851492362U, 4052801922U, 2055266765U,
+	 635556996U,  368266356U, 2385737383U, 3218202352U, 2603772408U,
+	 349178792U,  226482567U, 3102426060U, 3575998268U, 2103001871U,
+	3243137071U,  225500688U, 1634718593U, 4283311431U, 4292122923U,
+	3842802787U,  811735523U,  105712518U,  663434053U, 1855889273U,
+	2847972595U, 1196355421U, 2552150115U, 4254510614U, 3752181265U,
+	3430721819U, 3828705396U, 3436287905U, 3441964937U, 4123670631U,
+	 353001539U,  459496439U, 3799690868U, 1293777660U, 2761079737U,
+	 498096339U, 3398433374U, 4080378380U, 2304691596U, 2995729055U,
+	4134660419U, 3903444024U, 3576494993U,  203682175U, 3321164857U,
+	2747963611U,   79749085U, 2992890370U, 1240278549U, 1772175713U,
+	2111331972U, 2655023449U, 1683896345U, 2836027212U, 3482868021U,
+	2489884874U,  756853961U, 2298874501U, 4013448667U, 4143996022U,
+	2948306858U, 4132920035U, 1283299272U,  995592228U, 3450508595U,
+	1027845759U, 1766942720U, 3861411826U, 1446861231U,   95974993U,
+	3502263554U, 1487532194U,  601502472U, 4129619129U,  250131773U,
+	2050079547U, 3198903947U, 3105589778U, 4066481316U, 3026383978U,
+	2276901713U,  365637751U, 2260718426U, 1394775634U, 1791172338U,
+	2690503163U, 2952737846U, 1568710462U,  732623190U, 2980358000U,
+	1053631832U, 1432426951U, 3229149635U, 1854113985U, 3719733532U,
+	3204031934U,  735775531U,  107468620U, 3734611984U,  631009402U,
+	3083622457U, 4109580626U,  159373458U, 1301970201U, 4132389302U,
+	1293255004U,  847182752U, 4170022737U,   96712900U, 2641406755U,
+	1381727755U,  405608287U, 4287919625U, 1703554290U, 3589580244U,
+	2911403488U,    2166565U, 2647306451U, 2330535117U, 1200815358U,
+	1165916754U,  245060911U, 4040679071U, 3684908771U, 2452834126U,
+	2486872773U, 2318678365U, 2940627908U, 1837837240U, 3447897409U,
+	4270484676U, 1495388728U, 3754288477U, 4204167884U, 1386977705U,
+	2692224733U, 3076249689U, 4109568048U, 4170955115U, 4167531356U,
+	4020189950U, 4261855038U, 3036907575U, 3410399885U, 3076395737U,
+	1046178638U,  144496770U,  230725846U, 3349637149U,   17065717U,
+	2809932048U, 2054581785U, 3608424964U, 3259628808U,  134897388U,
+	3743067463U,  257685904U, 3795656590U, 1562468719U, 3589103904U,
+	3120404710U,  254684547U, 2653661580U, 3663904795U, 2631942758U,
+	1063234347U, 2609732900U, 2332080715U, 3521125233U, 1180599599U,
+	1935868586U, 4110970440U,  296706371U, 2128666368U, 1319875791U,
+	1570900197U, 3096025483U, 1799882517U, 1928302007U, 1163707758U,
+	1244491489U, 3533770203U,  567496053U, 2757924305U, 2781639343U,
+	2818420107U,  560404889U, 2619609724U, 4176035430U, 2511289753U,
+	2521842019U, 3910553502U, 2926149387U, 3302078172U, 4237118867U,
+	 330725126U,  367400677U,  888239854U,  545570454U, 4259590525U,
+	 134343617U, 1102169784U, 1647463719U, 3260979784U, 1518840883U,
+	3631537963U, 3342671457U, 1301549147U, 2083739356U,  146593792U,
+	3217959080U,  652755743U, 2032187193U, 3898758414U, 1021358093U,
+	4037409230U, 2176407931U, 3427391950U, 2883553603U,  985613827U,
+	3105265092U, 3423168427U, 3387507672U,  467170288U, 2141266163U,
+	3723870208U,  916410914U, 1293987799U, 2652584950U,  769160137U,
+	3205292896U, 1561287359U, 1684510084U, 3136055621U, 3765171391U,
+	 639683232U, 2639569327U, 1218546948U, 4263586685U, 3058215773U,
+	2352279820U,  401870217U, 2625822463U, 1529125296U, 2981801895U,
+	1191285226U, 4027725437U, 3432700217U, 4098835661U,  971182783U,
+	2443861173U, 3881457123U, 3874386651U,  457276199U, 2638294160U,
+	4002809368U,  421169044U, 1112642589U, 3076213779U, 3387033971U,
+	2499610950U, 3057240914U, 1662679783U,  461224431U, 1168395933U
+};
+static const uint32_t init_by_array_32_expected[] = {
+	2920711183U, 3885745737U, 3501893680U,  856470934U, 1421864068U,
+	 277361036U, 1518638004U, 2328404353U, 3355513634U,   64329189U,
+	1624587673U, 3508467182U, 2481792141U, 3706480799U, 1925859037U,
+	2913275699U,  882658412U,  384641219U,  422202002U, 1873384891U,
+	2006084383U, 3924929912U, 1636718106U, 3108838742U, 1245465724U,
+	4195470535U,  779207191U, 1577721373U, 1390469554U, 2928648150U,
+	 121399709U, 3170839019U, 4044347501U,  953953814U, 3821710850U,
+	3085591323U, 3666535579U, 3577837737U, 2012008410U, 3565417471U,
+	4044408017U,  433600965U, 1637785608U, 1798509764U,  860770589U,
+	3081466273U, 3982393409U, 2451928325U, 3437124742U, 4093828739U,
+	3357389386U, 2154596123U,  496568176U, 2650035164U, 2472361850U,
+	   3438299U, 2150366101U, 1577256676U, 3802546413U, 1787774626U,
+	4078331588U, 3706103141U,  170391138U, 3806085154U, 1680970100U,
+	1961637521U, 3316029766U,  890610272U, 1453751581U, 1430283664U,
+	3051057411U, 3597003186U,  542563954U, 3796490244U, 1690016688U,
+	3448752238U,  440702173U,  347290497U, 1121336647U, 2540588620U,
+	 280881896U, 2495136428U,  213707396U,   15104824U, 2946180358U,
+	 659000016U,  566379385U, 2614030979U, 2855760170U,  334526548U,
+	2315569495U, 2729518615U,  564745877U, 1263517638U, 3157185798U,
+	1604852056U, 1011639885U, 2950579535U, 2524219188U,  312951012U,
+	1528896652U, 1327861054U, 2846910138U, 3966855905U, 2536721582U,
+	 855353911U, 1685434729U, 3303978929U, 1624872055U, 4020329649U,
+	3164802143U, 1642802700U, 1957727869U, 1792352426U, 3334618929U,
+	2631577923U, 3027156164U,  842334259U, 3353446843U, 1226432104U,
+	1742801369U, 3552852535U, 3471698828U, 1653910186U, 3380330939U,
+	2313782701U, 3351007196U, 2129839995U, 1800682418U, 4085884420U,
+	1625156629U, 3669701987U,  615211810U, 3294791649U, 4131143784U,
+	2590843588U, 3207422808U, 3275066464U,  561592872U, 3957205738U,
+	3396578098U,   48410678U, 3505556445U, 1005764855U, 3920606528U,
+	2936980473U, 2378918600U, 2404449845U, 1649515163U,  701203563U,
+	3705256349U,   83714199U, 3586854132U,  922978446U, 2863406304U,
+	3523398907U, 2606864832U, 2385399361U, 3171757816U, 4262841009U,
+	3645837721U, 1169579486U, 3666433897U, 3174689479U, 1457866976U,
+	3803895110U, 3346639145U, 1907224409U, 1978473712U, 1036712794U,
+	 980754888U, 1302782359U, 1765252468U,  459245755U, 3728923860U,
+	1512894209U, 2046491914U,  207860527U,  514188684U, 2288713615U,
+	1597354672U, 3349636117U, 2357291114U, 3995796221U,  945364213U,
+	1893326518U, 3770814016U, 1691552714U, 2397527410U,  967486361U,
+	 776416472U, 4197661421U,  951150819U, 1852770983U, 4044624181U,
+	1399439738U, 4194455275U, 2284037669U, 1550734958U, 3321078108U,
+	1865235926U, 2912129961U, 2664980877U, 1357572033U, 2600196436U,
+	2486728200U, 2372668724U, 1567316966U, 2374111491U, 1839843570U,
+	  20815612U, 3727008608U, 3871996229U,  824061249U, 1932503978U,
+	3404541726U,  758428924U, 2609331364U, 1223966026U, 1299179808U,
+	 648499352U, 2180134401U,  880821170U, 3781130950U,  113491270U,
+	1032413764U, 4185884695U, 2490396037U, 1201932817U, 4060951446U,
+	4165586898U, 1629813212U, 2887821158U,  415045333U,  628926856U,
+	2193466079U, 3391843445U, 2227540681U, 1907099846U, 2848448395U,
+	1717828221U, 1372704537U, 1707549841U, 2294058813U, 2101214437U,
+	2052479531U, 1695809164U, 3176587306U, 2632770465U,   81634404U,
+	1603220563U,  644238487U,  302857763U,  897352968U, 2613146653U,
+	1391730149U, 4245717312U, 4191828749U, 1948492526U, 2618174230U,
+	3992984522U, 2178852787U, 3596044509U, 3445573503U, 2026614616U,
+	 915763564U, 3415689334U, 2532153403U, 3879661562U, 2215027417U,
+	3111154986U, 2929478371U,  668346391U, 1152241381U, 2632029711U,
+	3004150659U, 2135025926U,  948690501U, 2799119116U, 4228829406U,
+	1981197489U, 4209064138U,  684318751U, 3459397845U,  201790843U,
+	4022541136U, 3043635877U,  492509624U, 3263466772U, 1509148086U,
+	 921459029U, 3198857146U,  705479721U, 3835966910U, 3603356465U,
+	 576159741U, 1742849431U,  594214882U, 2055294343U, 3634861861U,
+	 449571793U, 3246390646U, 3868232151U, 1479156585U, 2900125656U,
+	2464815318U, 3960178104U, 1784261920U,   18311476U, 3627135050U,
+	 644609697U,  424968996U,  919890700U, 2986824110U,  816423214U,
+	4003562844U, 1392714305U, 1757384428U, 2569030598U,  995949559U,
+	3875659880U, 2933807823U, 2752536860U, 2993858466U, 4030558899U,
+	2770783427U, 2775406005U, 2777781742U, 1931292655U,  472147933U,
+	3865853827U, 2726470545U, 2668412860U, 2887008249U,  408979190U,
+	3578063323U, 3242082049U, 1778193530U,   27981909U, 2362826515U,
+	 389875677U, 1043878156U,  581653903U, 3830568952U,  389535942U,
+	3713523185U, 2768373359U, 2526101582U, 1998618197U, 1160859704U,
+	3951172488U, 1098005003U,  906275699U, 3446228002U, 2220677963U,
+	2059306445U,  132199571U,  476838790U, 1868039399U, 3097344807U,
+	 857300945U,  396345050U, 2835919916U, 1782168828U, 1419519470U,
+	4288137521U,  819087232U,  596301494U,  872823172U, 1526888217U,
+	 805161465U, 1116186205U, 2829002754U, 2352620120U,  620121516U,
+	 354159268U, 3601949785U,  209568138U, 1352371732U, 2145977349U,
+	4236871834U, 1539414078U, 3558126206U, 3224857093U, 4164166682U,
+	3817553440U, 3301780278U, 2682696837U, 3734994768U, 1370950260U,
+	1477421202U, 2521315749U, 1330148125U, 1261554731U, 2769143688U,
+	3554756293U, 4235882678U, 3254686059U, 3530579953U, 1215452615U,
+	3574970923U, 4057131421U,  589224178U, 1000098193U,  171190718U,
+	2521852045U, 2351447494U, 2284441580U, 2646685513U, 3486933563U,
+	3789864960U, 1190528160U, 1702536782U, 1534105589U, 4262946827U,
+	2726686826U, 3584544841U, 2348270128U, 2145092281U, 2502718509U,
+	1027832411U, 3571171153U, 1287361161U, 4011474411U, 3241215351U,
+	2419700818U,  971242709U, 1361975763U, 1096842482U, 3271045537U,
+	  81165449U,  612438025U, 3912966678U, 1356929810U,  733545735U,
+	 537003843U, 1282953084U,  884458241U,  588930090U, 3930269801U,
+	2961472450U, 1219535534U, 3632251943U,  268183903U, 1441240533U,
+	3653903360U, 3854473319U, 2259087390U, 2548293048U, 2022641195U,
+	2105543911U, 1764085217U, 3246183186U,  482438805U,  888317895U,
+	2628314765U, 2466219854U,  717546004U, 2322237039U,  416725234U,
+	1544049923U, 1797944973U, 3398652364U, 3111909456U,  485742908U,
+	2277491072U, 1056355088U, 3181001278U,  129695079U, 2693624550U,
+	1764438564U, 3797785470U,  195503713U, 3266519725U, 2053389444U,
+	1961527818U, 3400226523U, 3777903038U, 2597274307U, 4235851091U,
+	4094406648U, 2171410785U, 1781151386U, 1378577117U,  654643266U,
+	3424024173U, 3385813322U,  679385799U,  479380913U,  681715441U,
+	3096225905U,  276813409U, 3854398070U, 2721105350U,  831263315U,
+	3276280337U, 2628301522U, 3984868494U, 1466099834U, 2104922114U,
+	1412672743U,  820330404U, 3491501010U,  942735832U,  710652807U,
+	3972652090U,  679881088U,   40577009U, 3705286397U, 2815423480U,
+	3566262429U,  663396513U, 3777887429U, 4016670678U,  404539370U,
+	1142712925U, 1140173408U, 2913248352U, 2872321286U,  263751841U,
+	3175196073U, 3162557581U, 2878996619U,   75498548U, 3836833140U,
+	3284664959U, 1157523805U,  112847376U,  207855609U, 1337979698U,
+	1222578451U,  157107174U,  901174378U, 3883717063U, 1618632639U,
+	1767889440U, 4264698824U, 1582999313U,  884471997U, 2508825098U,
+	3756370771U, 2457213553U, 3565776881U, 3709583214U,  915609601U,
+	 460833524U, 1091049576U,   85522880U,    2553251U,  132102809U,
+	2429882442U, 2562084610U, 1386507633U, 4112471229U,   21965213U,
+	1981516006U, 2418435617U, 3054872091U, 4251511224U, 2025783543U,
+	1916911512U, 2454491136U, 3938440891U, 3825869115U, 1121698605U,
+	3463052265U,  802340101U, 1912886800U, 4031997367U, 3550640406U,
+	1596096923U,  610150600U,  431464457U, 2541325046U,  486478003U,
+	 739704936U, 2862696430U, 3037903166U, 1129749694U, 2611481261U,
+	1228993498U,  510075548U, 3424962587U, 2458689681U,  818934833U,
+	4233309125U, 1608196251U, 3419476016U, 1858543939U, 2682166524U,
+	3317854285U,  631986188U, 3008214764U,  613826412U, 3567358221U,
+	3512343882U, 1552467474U, 3316162670U, 1275841024U, 4142173454U,
+	 565267881U,  768644821U,  198310105U, 2396688616U, 1837659011U,
+	 203429334U,  854539004U, 4235811518U, 3338304926U, 3730418692U,
+	3852254981U, 3032046452U, 2329811860U, 2303590566U, 2696092212U,
+	3894665932U,  145835667U,  249563655U, 1932210840U, 2431696407U,
+	3312636759U,  214962629U, 2092026914U, 3020145527U, 4073039873U,
+	2739105705U, 1308336752U,  855104522U, 2391715321U,   67448785U,
+	 547989482U,  854411802U, 3608633740U,  431731530U,  537375589U,
+	3888005760U,  696099141U,  397343236U, 1864511780U,   44029739U,
+	1729526891U, 1993398655U, 2010173426U, 2591546756U,  275223291U,
+	1503900299U, 4217765081U, 2185635252U, 1122436015U, 3550155364U,
+	 681707194U, 3260479338U,  933579397U, 2983029282U, 2505504587U,
+	2667410393U, 2962684490U, 4139721708U, 2658172284U, 2452602383U,
+	2607631612U, 1344296217U, 3075398709U, 2949785295U, 1049956168U,
+	3917185129U, 2155660174U, 3280524475U, 1503827867U,  674380765U,
+	1918468193U, 3843983676U,  634358221U, 2538335643U, 1873351298U,
+	3368723763U, 2129144130U, 3203528633U, 3087174986U, 2691698871U,
+	2516284287U,   24437745U, 1118381474U, 2816314867U, 2448576035U,
+	4281989654U,  217287825U,  165872888U, 2628995722U, 3533525116U,
+	2721669106U,  872340568U, 3429930655U, 3309047304U, 3916704967U,
+	3270160355U, 1348884255U, 1634797670U,  881214967U, 4259633554U,
+	 174613027U, 1103974314U, 1625224232U, 2678368291U, 1133866707U,
+	3853082619U, 4073196549U, 1189620777U,  637238656U,  930241537U,
+	4042750792U, 3842136042U, 2417007212U, 2524907510U, 1243036827U,
+	1282059441U, 3764588774U, 1394459615U, 2323620015U, 1166152231U,
+	3307479609U, 3849322257U, 3507445699U, 4247696636U,  758393720U,
+	 967665141U, 1095244571U, 1319812152U,  407678762U, 2640605208U,
+	2170766134U, 3663594275U, 4039329364U, 2512175520U,  725523154U,
+	2249807004U, 3312617979U, 2414634172U, 1278482215U,  349206484U,
+	1573063308U, 1196429124U, 3873264116U, 2400067801U,  268795167U,
+	 226175489U, 2961367263U, 1968719665U,   42656370U, 1010790699U,
+	 561600615U, 2422453992U, 3082197735U, 1636700484U, 3977715296U,
+	3125350482U, 3478021514U, 2227819446U, 1540868045U, 3061908980U,
+	1087362407U, 3625200291U,  361937537U,  580441897U, 1520043666U,
+	2270875402U, 1009161260U, 2502355842U, 4278769785U,  473902412U,
+	1057239083U, 1905829039U, 1483781177U, 2080011417U, 1207494246U,
+	1806991954U, 2194674403U, 3455972205U,  807207678U, 3655655687U,
+	 674112918U,  195425752U, 3917890095U, 1874364234U, 1837892715U,
+	3663478166U, 1548892014U, 2570748714U, 2049929836U, 2167029704U,
+	 697543767U, 3499545023U, 3342496315U, 1725251190U, 3561387469U,
+	2905606616U, 1580182447U, 3934525927U, 4103172792U, 1365672522U,
+	1534795737U, 3308667416U, 2841911405U, 3943182730U, 4072020313U,
+	3494770452U, 3332626671U,   55327267U,  478030603U,  411080625U,
+	3419529010U, 1604767823U, 3513468014U,  570668510U,  913790824U,
+	2283967995U,  695159462U, 3825542932U, 4150698144U, 1829758699U,
+	 202895590U, 1609122645U, 1267651008U, 2910315509U, 2511475445U,
+	2477423819U, 3932081579U,  900879979U, 2145588390U, 2670007504U,
+	 580819444U, 1864996828U, 2526325979U, 1019124258U,  815508628U,
+	2765933989U, 1277301341U, 3006021786U,  855540956U,  288025710U,
+	1919594237U, 2331223864U,  177452412U, 2475870369U, 2689291749U,
+	 865194284U,  253432152U, 2628531804U, 2861208555U, 2361597573U,
+	1653952120U, 1039661024U, 2159959078U, 3709040440U, 3564718533U,
+	2596878672U, 2041442161U,   31164696U, 2662962485U, 3665637339U,
+	1678115244U, 2699839832U, 3651968520U, 3521595541U,  458433303U,
+	2423096824U,   21831741U,  380011703U, 2498168716U,  861806087U,
+	1673574843U, 4188794405U, 2520563651U, 2632279153U, 2170465525U,
+	4171949898U, 3886039621U, 1661344005U, 3424285243U,  992588372U,
+	2500984144U, 2993248497U, 3590193895U, 1535327365U,  515645636U,
+	 131633450U, 3729760261U, 1613045101U, 3254194278U,   15889678U,
+	1493590689U,  244148718U, 2991472662U, 1401629333U,  777349878U,
+	2501401703U, 4285518317U, 3794656178U,  955526526U, 3442142820U,
+	3970298374U,  736025417U, 2737370764U, 1271509744U,  440570731U,
+	 136141826U, 1596189518U,  923399175U,  257541519U, 3505774281U,
+	2194358432U, 2518162991U, 1379893637U, 2667767062U, 3748146247U,
+	1821712620U, 3923161384U, 1947811444U, 2392527197U, 4127419685U,
+	1423694998U, 4156576871U, 1382885582U, 3420127279U, 3617499534U,
+	2994377493U, 4038063986U, 1918458672U, 2983166794U, 4200449033U,
+	 353294540U, 1609232588U,  243926648U, 2332803291U,  507996832U,
+	2392838793U, 4075145196U, 2060984340U, 4287475136U,   88232602U,
+	2491531140U, 4159725633U, 2272075455U,  759298618U,  201384554U,
+	 838356250U, 1416268324U,  674476934U,   90795364U,  141672229U,
+	3660399588U, 4196417251U, 3249270244U, 3774530247U,   59587265U,
+	3683164208U,   19392575U, 1463123697U, 1882205379U,  293780489U,
+	2553160622U, 2933904694U,  675638239U, 2851336944U, 1435238743U,
+	2448730183U,  804436302U, 2119845972U,  322560608U, 4097732704U,
+	2987802540U,  641492617U, 2575442710U, 4217822703U, 3271835300U,
+	2836418300U, 3739921620U, 2138378768U, 2879771855U, 4294903423U,
+	3121097946U, 2603440486U, 2560820391U, 1012930944U, 2313499967U,
+	 584489368U, 3431165766U,  897384869U, 2062537737U, 2847889234U,
+	3742362450U, 2951174585U, 4204621084U, 1109373893U, 3668075775U,
+	2750138839U, 3518055702U,  733072558U, 4169325400U,  788493625U
+};
+static const uint64_t init_gen_rand_64_expected[] = {
+	KQU(16924766246869039260), KQU( 8201438687333352714),
+	KQU( 2265290287015001750), KQU(18397264611805473832),
+	KQU( 3375255223302384358), KQU( 6345559975416828796),
+	KQU(18229739242790328073), KQU( 7596792742098800905),
+	KQU(  255338647169685981), KQU( 2052747240048610300),
+	KQU(18328151576097299343), KQU(12472905421133796567),
+	KQU(11315245349717600863), KQU(16594110197775871209),
+	KQU(15708751964632456450), KQU(10452031272054632535),
+	KQU(11097646720811454386), KQU( 4556090668445745441),
+	KQU(17116187693090663106), KQU(14931526836144510645),
+	KQU( 9190752218020552591), KQU( 9625800285771901401),
+	KQU(13995141077659972832), KQU( 5194209094927829625),
+	KQU( 4156788379151063303), KQU( 8523452593770139494),
+	KQU(14082382103049296727), KQU( 2462601863986088483),
+	KQU( 3030583461592840678), KQU( 5221622077872827681),
+	KQU( 3084210671228981236), KQU(13956758381389953823),
+	KQU(13503889856213423831), KQU(15696904024189836170),
+	KQU( 4612584152877036206), KQU( 6231135538447867881),
+	KQU(10172457294158869468), KQU( 6452258628466708150),
+	KQU(14044432824917330221), KQU(  370168364480044279),
+	KQU(10102144686427193359), KQU(  667870489994776076),
+	KQU( 2732271956925885858), KQU(18027788905977284151),
+	KQU(15009842788582923859), KQU( 7136357960180199542),
+	KQU(15901736243475578127), KQU(16951293785352615701),
+	KQU(10551492125243691632), KQU(17668869969146434804),
+	KQU(13646002971174390445), KQU( 9804471050759613248),
+	KQU( 5511670439655935493), KQU(18103342091070400926),
+	KQU(17224512747665137533), KQU(15534627482992618168),
+	KQU( 1423813266186582647), KQU(15821176807932930024),
+	KQU(   30323369733607156), KQU(11599382494723479403),
+	KQU(  653856076586810062), KQU( 3176437395144899659),
+	KQU(14028076268147963917), KQU(16156398271809666195),
+	KQU( 3166955484848201676), KQU( 5746805620136919390),
+	KQU(17297845208891256593), KQU(11691653183226428483),
+	KQU(17900026146506981577), KQU(15387382115755971042),
+	KQU(16923567681040845943), KQU( 8039057517199388606),
+	KQU(11748409241468629263), KQU(  794358245539076095),
+	KQU(13438501964693401242), KQU(14036803236515618962),
+	KQU( 5252311215205424721), KQU(17806589612915509081),
+	KQU( 6802767092397596006), KQU(14212120431184557140),
+	KQU( 1072951366761385712), KQU(13098491780722836296),
+	KQU( 9466676828710797353), KQU(12673056849042830081),
+	KQU(12763726623645357580), KQU(16468961652999309493),
+	KQU(15305979875636438926), KQU(17444713151223449734),
+	KQU( 5692214267627883674), KQU(13049589139196151505),
+	KQU(  880115207831670745), KQU( 1776529075789695498),
+	KQU(16695225897801466485), KQU(10666901778795346845),
+	KQU( 6164389346722833869), KQU( 2863817793264300475),
+	KQU( 9464049921886304754), KQU( 3993566636740015468),
+	KQU( 9983749692528514136), KQU(16375286075057755211),
+	KQU(16042643417005440820), KQU(11445419662923489877),
+	KQU( 7999038846885158836), KQU( 6721913661721511535),
+	KQU( 5363052654139357320), KQU( 1817788761173584205),
+	KQU(13290974386445856444), KQU( 4650350818937984680),
+	KQU( 8219183528102484836), KQU( 1569862923500819899),
+	KQU( 4189359732136641860), KQU(14202822961683148583),
+	KQU( 4457498315309429058), KQU(13089067387019074834),
+	KQU(11075517153328927293), KQU(10277016248336668389),
+	KQU( 7070509725324401122), KQU(17808892017780289380),
+	KQU(13143367339909287349), KQU( 1377743745360085151),
+	KQU( 5749341807421286485), KQU(14832814616770931325),
+	KQU( 7688820635324359492), KQU(10960474011539770045),
+	KQU(   81970066653179790), KQU(12619476072607878022),
+	KQU( 4419566616271201744), KQU(15147917311750568503),
+	KQU( 5549739182852706345), KQU( 7308198397975204770),
+	KQU(13580425496671289278), KQU(17070764785210130301),
+	KQU( 8202832846285604405), KQU( 6873046287640887249),
+	KQU( 6927424434308206114), KQU( 6139014645937224874),
+	KQU(10290373645978487639), KQU(15904261291701523804),
+	KQU( 9628743442057826883), KQU(18383429096255546714),
+	KQU( 4977413265753686967), KQU( 7714317492425012869),
+	KQU( 9025232586309926193), KQU(14627338359776709107),
+	KQU(14759849896467790763), KQU(10931129435864423252),
+	KQU( 4588456988775014359), KQU(10699388531797056724),
+	KQU(  468652268869238792), KQU( 5755943035328078086),
+	KQU( 2102437379988580216), KQU( 9986312786506674028),
+	KQU( 2654207180040945604), KQU( 8726634790559960062),
+	KQU(  100497234871808137), KQU( 2800137176951425819),
+	KQU( 6076627612918553487), KQU( 5780186919186152796),
+	KQU( 8179183595769929098), KQU( 6009426283716221169),
+	KQU( 2796662551397449358), KQU( 1756961367041986764),
+	KQU( 6972897917355606205), KQU(14524774345368968243),
+	KQU( 2773529684745706940), KQU( 4853632376213075959),
+	KQU( 4198177923731358102), KQU( 8271224913084139776),
+	KQU( 2741753121611092226), KQU(16782366145996731181),
+	KQU(15426125238972640790), KQU(13595497100671260342),
+	KQU( 3173531022836259898), KQU( 6573264560319511662),
+	KQU(18041111951511157441), KQU( 2351433581833135952),
+	KQU( 3113255578908173487), KQU( 1739371330877858784),
+	KQU(16046126562789165480), KQU( 8072101652214192925),
+	KQU(15267091584090664910), KQU( 9309579200403648940),
+	KQU( 5218892439752408722), KQU(14492477246004337115),
+	KQU(17431037586679770619), KQU( 7385248135963250480),
+	KQU( 9580144956565560660), KQU( 4919546228040008720),
+	KQU(15261542469145035584), KQU(18233297270822253102),
+	KQU( 5453248417992302857), KQU( 9309519155931460285),
+	KQU(10342813012345291756), KQU(15676085186784762381),
+	KQU(15912092950691300645), KQU( 9371053121499003195),
+	KQU( 9897186478226866746), KQU(14061858287188196327),
+	KQU(  122575971620788119), KQU(12146750969116317754),
+	KQU( 4438317272813245201), KQU( 8332576791009527119),
+	KQU(13907785691786542057), KQU(10374194887283287467),
+	KQU( 2098798755649059566), KQU( 3416235197748288894),
+	KQU( 8688269957320773484), KQU( 7503964602397371571),
+	KQU(16724977015147478236), KQU( 9461512855439858184),
+	KQU(13259049744534534727), KQU( 3583094952542899294),
+	KQU( 8764245731305528292), KQU(13240823595462088985),
+	KQU(13716141617617910448), KQU(18114969519935960955),
+	KQU( 2297553615798302206), KQU( 4585521442944663362),
+	KQU(17776858680630198686), KQU( 4685873229192163363),
+	KQU(  152558080671135627), KQU(15424900540842670088),
+	KQU(13229630297130024108), KQU(17530268788245718717),
+	KQU(16675633913065714144), KQU( 3158912717897568068),
+	KQU(15399132185380087288), KQU( 7401418744515677872),
+	KQU(13135412922344398535), KQU( 6385314346100509511),
+	KQU(13962867001134161139), KQU(10272780155442671999),
+	KQU(12894856086597769142), KQU(13340877795287554994),
+	KQU(12913630602094607396), KQU(12543167911119793857),
+	KQU(17343570372251873096), KQU(10959487764494150545),
+	KQU( 6966737953093821128), KQU(13780699135496988601),
+	KQU( 4405070719380142046), KQU(14923788365607284982),
+	KQU( 2869487678905148380), KQU( 6416272754197188403),
+	KQU(15017380475943612591), KQU( 1995636220918429487),
+	KQU( 3402016804620122716), KQU(15800188663407057080),
+	KQU(11362369990390932882), KQU(15262183501637986147),
+	KQU(10239175385387371494), KQU( 9352042420365748334),
+	KQU( 1682457034285119875), KQU( 1724710651376289644),
+	KQU( 2038157098893817966), KQU( 9897825558324608773),
+	KQU( 1477666236519164736), KQU(16835397314511233640),
+	KQU(10370866327005346508), KQU(10157504370660621982),
+	KQU(12113904045335882069), KQU(13326444439742783008),
+	KQU(11302769043000765804), KQU(13594979923955228484),
+	KQU(11779351762613475968), KQU( 3786101619539298383),
+	KQU( 8021122969180846063), KQU(15745904401162500495),
+	KQU(10762168465993897267), KQU(13552058957896319026),
+	KQU(11200228655252462013), KQU( 5035370357337441226),
+	KQU( 7593918984545500013), KQU( 5418554918361528700),
+	KQU( 4858270799405446371), KQU( 9974659566876282544),
+	KQU(18227595922273957859), KQU( 2772778443635656220),
+	KQU(14285143053182085385), KQU( 9939700992429600469),
+	KQU(12756185904545598068), KQU( 2020783375367345262),
+	KQU(   57026775058331227), KQU(  950827867930065454),
+	KQU( 6602279670145371217), KQU( 2291171535443566929),
+	KQU( 5832380724425010313), KQU( 1220343904715982285),
+	KQU(17045542598598037633), KQU(15460481779702820971),
+	KQU(13948388779949365130), KQU(13975040175430829518),
+	KQU(17477538238425541763), KQU(11104663041851745725),
+	KQU(15860992957141157587), KQU(14529434633012950138),
+	KQU( 2504838019075394203), KQU( 7512113882611121886),
+	KQU( 4859973559980886617), KQU( 1258601555703250219),
+	KQU(15594548157514316394), KQU( 4516730171963773048),
+	KQU(11380103193905031983), KQU( 6809282239982353344),
+	KQU(18045256930420065002), KQU( 2453702683108791859),
+	KQU(  977214582986981460), KQU( 2006410402232713466),
+	KQU( 6192236267216378358), KQU( 3429468402195675253),
+	KQU(18146933153017348921), KQU(17369978576367231139),
+	KQU( 1246940717230386603), KQU(11335758870083327110),
+	KQU(14166488801730353682), KQU( 9008573127269635732),
+	KQU(10776025389820643815), KQU(15087605441903942962),
+	KQU( 1359542462712147922), KQU(13898874411226454206),
+	KQU(17911176066536804411), KQU( 9435590428600085274),
+	KQU(  294488509967864007), KQU( 8890111397567922046),
+	KQU( 7987823476034328778), KQU(13263827582440967651),
+	KQU( 7503774813106751573), KQU(14974747296185646837),
+	KQU( 8504765037032103375), KQU(17340303357444536213),
+	KQU( 7704610912964485743), KQU( 8107533670327205061),
+	KQU( 9062969835083315985), KQU(16968963142126734184),
+	KQU(12958041214190810180), KQU( 2720170147759570200),
+	KQU( 2986358963942189566), KQU(14884226322219356580),
+	KQU(  286224325144368520), KQU(11313800433154279797),
+	KQU(18366849528439673248), KQU(17899725929482368789),
+	KQU( 3730004284609106799), KQU( 1654474302052767205),
+	KQU( 5006698007047077032), KQU( 8196893913601182838),
+	KQU(15214541774425211640), KQU(17391346045606626073),
+	KQU( 8369003584076969089), KQU( 3939046733368550293),
+	KQU(10178639720308707785), KQU( 2180248669304388697),
+	KQU(   62894391300126322), KQU( 9205708961736223191),
+	KQU( 6837431058165360438), KQU( 3150743890848308214),
+	KQU(17849330658111464583), KQU(12214815643135450865),
+	KQU(13410713840519603402), KQU( 3200778126692046802),
+	KQU(13354780043041779313), KQU(  800850022756886036),
+	KQU(15660052933953067433), KQU( 6572823544154375676),
+	KQU(11030281857015819266), KQU(12682241941471433835),
+	KQU(11654136407300274693), KQU( 4517795492388641109),
+	KQU( 9757017371504524244), KQU(17833043400781889277),
+	KQU(12685085201747792227), KQU(10408057728835019573),
+	KQU(   98370418513455221), KQU( 6732663555696848598),
+	KQU(13248530959948529780), KQU( 3530441401230622826),
+	KQU(18188251992895660615), KQU( 1847918354186383756),
+	KQU( 1127392190402660921), KQU(11293734643143819463),
+	KQU( 3015506344578682982), KQU(13852645444071153329),
+	KQU( 2121359659091349142), KQU( 1294604376116677694),
+	KQU( 5616576231286352318), KQU( 7112502442954235625),
+	KQU(11676228199551561689), KQU(12925182803007305359),
+	KQU( 7852375518160493082), KQU( 1136513130539296154),
+	KQU( 5636923900916593195), KQU( 3221077517612607747),
+	KQU(17784790465798152513), KQU( 3554210049056995938),
+	KQU(17476839685878225874), KQU( 3206836372585575732),
+	KQU( 2765333945644823430), KQU(10080070903718799528),
+	KQU( 5412370818878286353), KQU( 9689685887726257728),
+	KQU( 8236117509123533998), KQU( 1951139137165040214),
+	KQU( 4492205209227980349), KQU(16541291230861602967),
+	KQU( 1424371548301437940), KQU( 9117562079669206794),
+	KQU(14374681563251691625), KQU(13873164030199921303),
+	KQU( 6680317946770936731), KQU(15586334026918276214),
+	KQU(10896213950976109802), KQU( 9506261949596413689),
+	KQU( 9903949574308040616), KQU( 6038397344557204470),
+	KQU(  174601465422373648), KQU(15946141191338238030),
+	KQU(17142225620992044937), KQU( 7552030283784477064),
+	KQU( 2947372384532947997), KQU(  510797021688197711),
+	KQU( 4962499439249363461), KQU(   23770320158385357),
+	KQU(  959774499105138124), KQU( 1468396011518788276),
+	KQU( 2015698006852312308), KQU( 4149400718489980136),
+	KQU( 5992916099522371188), KQU(10819182935265531076),
+	KQU(16189787999192351131), KQU(  342833961790261950),
+	KQU(12470830319550495336), KQU(18128495041912812501),
+	KQU( 1193600899723524337), KQU( 9056793666590079770),
+	KQU( 2154021227041669041), KQU( 4963570213951235735),
+	KQU( 4865075960209211409), KQU( 2097724599039942963),
+	KQU( 2024080278583179845), KQU(11527054549196576736),
+	KQU(10650256084182390252), KQU( 4808408648695766755),
+	KQU( 1642839215013788844), KQU(10607187948250398390),
+	KQU( 7076868166085913508), KQU(  730522571106887032),
+	KQU(12500579240208524895), KQU( 4484390097311355324),
+	KQU(15145801330700623870), KQU( 8055827661392944028),
+	KQU( 5865092976832712268), KQU(15159212508053625143),
+	KQU( 3560964582876483341), KQU( 4070052741344438280),
+	KQU( 6032585709886855634), KQU(15643262320904604873),
+	KQU( 2565119772293371111), KQU(  318314293065348260),
+	KQU(15047458749141511872), KQU( 7772788389811528730),
+	KQU( 7081187494343801976), KQU( 6465136009467253947),
+	KQU(10425940692543362069), KQU(  554608190318339115),
+	KQU(14796699860302125214), KQU( 1638153134431111443),
+	KQU(10336967447052276248), KQU( 8412308070396592958),
+	KQU( 4004557277152051226), KQU( 8143598997278774834),
+	KQU(16413323996508783221), KQU(13139418758033994949),
+	KQU( 9772709138335006667), KQU( 2818167159287157659),
+	KQU(17091740573832523669), KQU(14629199013130751608),
+	KQU(18268322711500338185), KQU( 8290963415675493063),
+	KQU( 8830864907452542588), KQU( 1614839084637494849),
+	KQU(14855358500870422231), KQU( 3472996748392519937),
+	KQU(15317151166268877716), KQU( 5825895018698400362),
+	KQU(16730208429367544129), KQU(10481156578141202800),
+	KQU( 4746166512382823750), KQU(12720876014472464998),
+	KQU( 8825177124486735972), KQU(13733447296837467838),
+	KQU( 6412293741681359625), KQU( 8313213138756135033),
+	KQU(11421481194803712517), KQU( 7997007691544174032),
+	KQU( 6812963847917605930), KQU( 9683091901227558641),
+	KQU(14703594165860324713), KQU( 1775476144519618309),
+	KQU( 2724283288516469519), KQU(  717642555185856868),
+	KQU( 8736402192215092346), KQU(11878800336431381021),
+	KQU( 4348816066017061293), KQU( 6115112756583631307),
+	KQU( 9176597239667142976), KQU(12615622714894259204),
+	KQU(10283406711301385987), KQU( 5111762509485379420),
+	KQU( 3118290051198688449), KQU( 7345123071632232145),
+	KQU( 9176423451688682359), KQU( 4843865456157868971),
+	KQU(12008036363752566088), KQU(12058837181919397720),
+	KQU( 2145073958457347366), KQU( 1526504881672818067),
+	KQU( 3488830105567134848), KQU(13208362960674805143),
+	KQU( 4077549672899572192), KQU( 7770995684693818365),
+	KQU( 1398532341546313593), KQU(12711859908703927840),
+	KQU( 1417561172594446813), KQU(17045191024194170604),
+	KQU( 4101933177604931713), KQU(14708428834203480320),
+	KQU(17447509264469407724), KQU(14314821973983434255),
+	KQU(17990472271061617265), KQU( 5087756685841673942),
+	KQU(12797820586893859939), KQU( 1778128952671092879),
+	KQU( 3535918530508665898), KQU( 9035729701042481301),
+	KQU(14808661568277079962), KQU(14587345077537747914),
+	KQU(11920080002323122708), KQU( 6426515805197278753),
+	KQU( 3295612216725984831), KQU(11040722532100876120),
+	KQU(12305952936387598754), KQU(16097391899742004253),
+	KQU( 4908537335606182208), KQU(12446674552196795504),
+	KQU(16010497855816895177), KQU( 9194378874788615551),
+	KQU( 3382957529567613384), KQU( 5154647600754974077),
+	KQU( 9801822865328396141), KQU( 9023662173919288143),
+	KQU(17623115353825147868), KQU( 8238115767443015816),
+	KQU(15811444159859002560), KQU( 9085612528904059661),
+	KQU( 6888601089398614254), KQU(  258252992894160189),
+	KQU( 6704363880792428622), KQU( 6114966032147235763),
+	KQU(11075393882690261875), KQU( 8797664238933620407),
+	KQU( 5901892006476726920), KQU( 5309780159285518958),
+	KQU(14940808387240817367), KQU(14642032021449656698),
+	KQU( 9808256672068504139), KQU( 3670135111380607658),
+	KQU(11211211097845960152), KQU( 1474304506716695808),
+	KQU(15843166204506876239), KQU( 7661051252471780561),
+	KQU(10170905502249418476), KQU( 7801416045582028589),
+	KQU( 2763981484737053050), KQU( 9491377905499253054),
+	KQU(16201395896336915095), KQU( 9256513756442782198),
+	KQU( 5411283157972456034), KQU( 5059433122288321676),
+	KQU( 4327408006721123357), KQU( 9278544078834433377),
+	KQU( 7601527110882281612), KQU(11848295896975505251),
+	KQU(12096998801094735560), KQU(14773480339823506413),
+	KQU(15586227433895802149), KQU(12786541257830242872),
+	KQU( 6904692985140503067), KQU( 5309011515263103959),
+	KQU(12105257191179371066), KQU(14654380212442225037),
+	KQU( 2556774974190695009), KQU( 4461297399927600261),
+	KQU(14888225660915118646), KQU(14915459341148291824),
+	KQU( 2738802166252327631), KQU( 6047155789239131512),
+	KQU(12920545353217010338), KQU(10697617257007840205),
+	KQU( 2751585253158203504), KQU(13252729159780047496),
+	KQU(14700326134672815469), KQU(14082527904374600529),
+	KQU(16852962273496542070), KQU(17446675504235853907),
+	KQU(15019600398527572311), KQU(12312781346344081551),
+	KQU(14524667935039810450), KQU( 5634005663377195738),
+	KQU(11375574739525000569), KQU( 2423665396433260040),
+	KQU( 5222836914796015410), KQU( 4397666386492647387),
+	KQU( 4619294441691707638), KQU(  665088602354770716),
+	KQU(13246495665281593610), KQU( 6564144270549729409),
+	KQU(10223216188145661688), KQU( 3961556907299230585),
+	KQU(11543262515492439914), KQU(16118031437285993790),
+	KQU( 7143417964520166465), KQU(13295053515909486772),
+	KQU(   40434666004899675), KQU(17127804194038347164),
+	KQU( 8599165966560586269), KQU( 8214016749011284903),
+	KQU(13725130352140465239), KQU( 5467254474431726291),
+	KQU( 7748584297438219877), KQU(16933551114829772472),
+	KQU( 2169618439506799400), KQU( 2169787627665113463),
+	KQU(17314493571267943764), KQU(18053575102911354912),
+	KQU(11928303275378476973), KQU(11593850925061715550),
+	KQU(17782269923473589362), KQU( 3280235307704747039),
+	KQU( 6145343578598685149), KQU(17080117031114086090),
+	KQU(18066839902983594755), KQU( 6517508430331020706),
+	KQU( 8092908893950411541), KQU(12558378233386153732),
+	KQU( 4476532167973132976), KQU(16081642430367025016),
+	KQU( 4233154094369139361), KQU( 8693630486693161027),
+	KQU(11244959343027742285), KQU(12273503967768513508),
+	KQU(14108978636385284876), KQU( 7242414665378826984),
+	KQU( 6561316938846562432), KQU( 8601038474994665795),
+	KQU(17532942353612365904), KQU(17940076637020912186),
+	KQU( 7340260368823171304), KQU( 7061807613916067905),
+	KQU(10561734935039519326), KQU(17990796503724650862),
+	KQU( 6208732943911827159), KQU(  359077562804090617),
+	KQU(14177751537784403113), KQU(10659599444915362902),
+	KQU(15081727220615085833), KQU(13417573895659757486),
+	KQU(15513842342017811524), KQU(11814141516204288231),
+	KQU( 1827312513875101814), KQU( 2804611699894603103),
+	KQU(17116500469975602763), KQU(12270191815211952087),
+	KQU(12256358467786024988), KQU(18435021722453971267),
+	KQU(  671330264390865618), KQU(  476504300460286050),
+	KQU(16465470901027093441), KQU( 4047724406247136402),
+	KQU( 1322305451411883346), KQU( 1388308688834322280),
+	KQU( 7303989085269758176), KQU( 9323792664765233642),
+	KQU( 4542762575316368936), KQU(17342696132794337618),
+	KQU( 4588025054768498379), KQU(13415475057390330804),
+	KQU(17880279491733405570), KQU(10610553400618620353),
+	KQU( 3180842072658960139), KQU(13002966655454270120),
+	KQU( 1665301181064982826), KQU( 7083673946791258979),
+	KQU(  190522247122496820), KQU(17388280237250677740),
+	KQU( 8430770379923642945), KQU(12987180971921668584),
+	KQU( 2311086108365390642), KQU( 2870984383579822345),
+	KQU(14014682609164653318), KQU(14467187293062251484),
+	KQU(  192186361147413298), KQU(15171951713531796524),
+	KQU( 9900305495015948728), KQU(17958004775615466344),
+	KQU(14346380954498606514), KQU(18040047357617407096),
+	KQU( 5035237584833424532), KQU(15089555460613972287),
+	KQU( 4131411873749729831), KQU( 1329013581168250330),
+	KQU(10095353333051193949), KQU(10749518561022462716),
+	KQU( 9050611429810755847), KQU(15022028840236655649),
+	KQU( 8775554279239748298), KQU(13105754025489230502),
+	KQU(15471300118574167585), KQU(   89864764002355628),
+	KQU( 8776416323420466637), KQU( 5280258630612040891),
+	KQU( 2719174488591862912), KQU( 7599309137399661994),
+	KQU(15012887256778039979), KQU(14062981725630928925),
+	KQU(12038536286991689603), KQU( 7089756544681775245),
+	KQU(10376661532744718039), KQU( 1265198725901533130),
+	KQU(13807996727081142408), KQU( 2935019626765036403),
+	KQU( 7651672460680700141), KQU( 3644093016200370795),
+	KQU( 2840982578090080674), KQU(17956262740157449201),
+	KQU(18267979450492880548), KQU(11799503659796848070),
+	KQU( 9942537025669672388), KQU(11886606816406990297),
+	KQU( 5488594946437447576), KQU( 7226714353282744302),
+	KQU( 3784851653123877043), KQU(  878018453244803041),
+	KQU(12110022586268616085), KQU(  734072179404675123),
+	KQU(11869573627998248542), KQU(  469150421297783998),
+	KQU(  260151124912803804), KQU(11639179410120968649),
+	KQU( 9318165193840846253), KQU(12795671722734758075),
+	KQU(15318410297267253933), KQU(  691524703570062620),
+	KQU( 5837129010576994601), KQU(15045963859726941052),
+	KQU( 5850056944932238169), KQU(12017434144750943807),
+	KQU( 7447139064928956574), KQU( 3101711812658245019),
+	KQU(16052940704474982954), KQU(18195745945986994042),
+	KQU( 8932252132785575659), KQU(13390817488106794834),
+	KQU(11582771836502517453), KQU( 4964411326683611686),
+	KQU( 2195093981702694011), KQU(14145229538389675669),
+	KQU(16459605532062271798), KQU(  866316924816482864),
+	KQU( 4593041209937286377), KQU( 8415491391910972138),
+	KQU( 4171236715600528969), KQU(16637569303336782889),
+	KQU( 2002011073439212680), KQU(17695124661097601411),
+	KQU( 4627687053598611702), KQU( 7895831936020190403),
+	KQU( 8455951300917267802), KQU( 2923861649108534854),
+	KQU( 8344557563927786255), KQU( 6408671940373352556),
+	KQU(12210227354536675772), KQU(14294804157294222295),
+	KQU(10103022425071085127), KQU(10092959489504123771),
+	KQU( 6554774405376736268), KQU(12629917718410641774),
+	KQU( 6260933257596067126), KQU( 2460827021439369673),
+	KQU( 2541962996717103668), KQU(  597377203127351475),
+	KQU( 5316984203117315309), KQU( 4811211393563241961),
+	KQU(13119698597255811641), KQU( 8048691512862388981),
+	KQU(10216818971194073842), KQU( 4612229970165291764),
+	KQU(10000980798419974770), KQU( 6877640812402540687),
+	KQU( 1488727563290436992), KQU( 2227774069895697318),
+	KQU(11237754507523316593), KQU(13478948605382290972),
+	KQU( 1963583846976858124), KQU( 5512309205269276457),
+	KQU( 3972770164717652347), KQU( 3841751276198975037),
+	KQU(10283343042181903117), KQU( 8564001259792872199),
+	KQU(16472187244722489221), KQU( 8953493499268945921),
+	KQU( 3518747340357279580), KQU( 4003157546223963073),
+	KQU( 3270305958289814590), KQU( 3966704458129482496),
+	KQU( 8122141865926661939), KQU(14627734748099506653),
+	KQU(13064426990862560568), KQU( 2414079187889870829),
+	KQU( 5378461209354225306), KQU(10841985740128255566),
+	KQU(  538582442885401738), KQU( 7535089183482905946),
+	KQU(16117559957598879095), KQU( 8477890721414539741),
+	KQU( 1459127491209533386), KQU(17035126360733620462),
+	KQU( 8517668552872379126), KQU(10292151468337355014),
+	KQU(17081267732745344157), KQU(13751455337946087178),
+	KQU(14026945459523832966), KQU( 6653278775061723516),
+	KQU(10619085543856390441), KQU( 2196343631481122885),
+	KQU(10045966074702826136), KQU(10082317330452718282),
+	KQU( 5920859259504831242), KQU( 9951879073426540617),
+	KQU( 7074696649151414158), KQU(15808193543879464318),
+	KQU( 7385247772746953374), KQU( 3192003544283864292),
+	KQU(18153684490917593847), KQU(12423498260668568905),
+	KQU(10957758099756378169), KQU(11488762179911016040),
+	KQU( 2099931186465333782), KQU(11180979581250294432),
+	KQU( 8098916250668367933), KQU( 3529200436790763465),
+	KQU(12988418908674681745), KQU( 6147567275954808580),
+	KQU( 3207503344604030989), KQU(10761592604898615360),
+	KQU(  229854861031893504), KQU( 8809853962667144291),
+	KQU(13957364469005693860), KQU( 7634287665224495886),
+	KQU(12353487366976556874), KQU( 1134423796317152034),
+	KQU( 2088992471334107068), KQU( 7393372127190799698),
+	KQU( 1845367839871058391), KQU(  207922563987322884),
+	KQU(11960870813159944976), KQU(12182120053317317363),
+	KQU(17307358132571709283), KQU(13871081155552824936),
+	KQU(18304446751741566262), KQU( 7178705220184302849),
+	KQU(10929605677758824425), KQU(16446976977835806844),
+	KQU(13723874412159769044), KQU( 6942854352100915216),
+	KQU( 1726308474365729390), KQU( 2150078766445323155),
+	KQU(15345558947919656626), KQU(12145453828874527201),
+	KQU( 2054448620739726849), KQU( 2740102003352628137),
+	KQU(11294462163577610655), KQU(  756164283387413743),
+	KQU(17841144758438810880), KQU(10802406021185415861),
+	KQU( 8716455530476737846), KQU( 6321788834517649606),
+	KQU(14681322910577468426), KQU(17330043563884336387),
+	KQU(12701802180050071614), KQU(14695105111079727151),
+	KQU( 5112098511654172830), KQU( 4957505496794139973),
+	KQU( 8270979451952045982), KQU(12307685939199120969),
+	KQU(12425799408953443032), KQU( 8376410143634796588),
+	KQU(16621778679680060464), KQU( 3580497854566660073),
+	KQU( 1122515747803382416), KQU(  857664980960597599),
+	KQU( 6343640119895925918), KQU(12878473260854462891),
+	KQU(10036813920765722626), KQU(14451335468363173812),
+	KQU( 5476809692401102807), KQU(16442255173514366342),
+	KQU(13060203194757167104), KQU(14354124071243177715),
+	KQU(15961249405696125227), KQU(13703893649690872584),
+	KQU(  363907326340340064), KQU( 6247455540491754842),
+	KQU(12242249332757832361), KQU(  156065475679796717),
+	KQU( 9351116235749732355), KQU( 4590350628677701405),
+	KQU( 1671195940982350389), KQU(13501398458898451905),
+	KQU( 6526341991225002255), KQU( 1689782913778157592),
+	KQU( 7439222350869010334), KQU(13975150263226478308),
+	KQU(11411961169932682710), KQU(17204271834833847277),
+	KQU(  541534742544435367), KQU( 6591191931218949684),
+	KQU( 2645454775478232486), KQU( 4322857481256485321),
+	KQU( 8477416487553065110), KQU(12902505428548435048),
+	KQU(  971445777981341415), KQU(14995104682744976712),
+	KQU( 4243341648807158063), KQU( 8695061252721927661),
+	KQU( 5028202003270177222), KQU( 2289257340915567840),
+	KQU(13870416345121866007), KQU(13994481698072092233),
+	KQU( 6912785400753196481), KQU( 2278309315841980139),
+	KQU( 4329765449648304839), KQU( 5963108095785485298),
+	KQU( 4880024847478722478), KQU(16015608779890240947),
+	KQU( 1866679034261393544), KQU(  914821179919731519),
+	KQU( 9643404035648760131), KQU( 2418114953615593915),
+	KQU(  944756836073702374), KQU(15186388048737296834),
+	KQU( 7723355336128442206), KQU( 7500747479679599691),
+	KQU(18013961306453293634), KQU( 2315274808095756456),
+	KQU(13655308255424029566), KQU(17203800273561677098),
+	KQU( 1382158694422087756), KQU( 5090390250309588976),
+	KQU(  517170818384213989), KQU( 1612709252627729621),
+	KQU( 1330118955572449606), KQU(  300922478056709885),
+	KQU(18115693291289091987), KQU(13491407109725238321),
+	KQU(15293714633593827320), KQU( 5151539373053314504),
+	KQU( 5951523243743139207), KQU(14459112015249527975),
+	KQU( 5456113959000700739), KQU( 3877918438464873016),
+	KQU(12534071654260163555), KQU(15871678376893555041),
+	KQU(11005484805712025549), KQU(16353066973143374252),
+	KQU( 4358331472063256685), KQU( 8268349332210859288),
+	KQU(12485161590939658075), KQU(13955993592854471343),
+	KQU( 5911446886848367039), KQU(14925834086813706974),
+	KQU( 6590362597857994805), KQU( 1280544923533661875),
+	KQU( 1637756018947988164), KQU( 4734090064512686329),
+	KQU(16693705263131485912), KQU( 6834882340494360958),
+	KQU( 8120732176159658505), KQU( 2244371958905329346),
+	KQU(10447499707729734021), KQU( 7318742361446942194),
+	KQU( 8032857516355555296), KQU(14023605983059313116),
+	KQU( 1032336061815461376), KQU( 9840995337876562612),
+	KQU( 9869256223029203587), KQU(12227975697177267636),
+	KQU(12728115115844186033), KQU( 7752058479783205470),
+	KQU(  729733219713393087), KQU(12954017801239007622)
+};
+static const uint64_t init_by_array_64_expected[] = {
+	KQU( 2100341266307895239), KQU( 8344256300489757943),
+	KQU(15687933285484243894), KQU( 8268620370277076319),
+	KQU(12371852309826545459), KQU( 8800491541730110238),
+	KQU(18113268950100835773), KQU( 2886823658884438119),
+	KQU( 3293667307248180724), KQU( 9307928143300172731),
+	KQU( 7688082017574293629), KQU(  900986224735166665),
+	KQU( 9977972710722265039), KQU( 6008205004994830552),
+	KQU(  546909104521689292), KQU( 7428471521869107594),
+	KQU(14777563419314721179), KQU(16116143076567350053),
+	KQU( 5322685342003142329), KQU( 4200427048445863473),
+	KQU( 4693092150132559146), KQU(13671425863759338582),
+	KQU( 6747117460737639916), KQU( 4732666080236551150),
+	KQU( 5912839950611941263), KQU( 3903717554504704909),
+	KQU( 2615667650256786818), KQU(10844129913887006352),
+	KQU(13786467861810997820), KQU(14267853002994021570),
+	KQU(13767807302847237439), KQU(16407963253707224617),
+	KQU( 4802498363698583497), KQU( 2523802839317209764),
+	KQU( 3822579397797475589), KQU( 8950320572212130610),
+	KQU( 3745623504978342534), KQU(16092609066068482806),
+	KQU( 9817016950274642398), KQU(10591660660323829098),
+	KQU(11751606650792815920), KQU( 5122873818577122211),
+	KQU(17209553764913936624), KQU( 6249057709284380343),
+	KQU(15088791264695071830), KQU(15344673071709851930),
+	KQU( 4345751415293646084), KQU( 2542865750703067928),
+	KQU(13520525127852368784), KQU(18294188662880997241),
+	KQU( 3871781938044881523), KQU( 2873487268122812184),
+	KQU(15099676759482679005), KQU(15442599127239350490),
+	KQU( 6311893274367710888), KQU( 3286118760484672933),
+	KQU( 4146067961333542189), KQU(13303942567897208770),
+	KQU( 8196013722255630418), KQU( 4437815439340979989),
+	KQU(15433791533450605135), KQU( 4254828956815687049),
+	KQU( 1310903207708286015), KQU(10529182764462398549),
+	KQU(14900231311660638810), KQU( 9727017277104609793),
+	KQU( 1821308310948199033), KQU(11628861435066772084),
+	KQU( 9469019138491546924), KQU( 3145812670532604988),
+	KQU( 9938468915045491919), KQU( 1562447430672662142),
+	KQU(13963995266697989134), KQU( 3356884357625028695),
+	KQU( 4499850304584309747), KQU( 8456825817023658122),
+	KQU(10859039922814285279), KQU( 8099512337972526555),
+	KQU(  348006375109672149), KQU(11919893998241688603),
+	KQU( 1104199577402948826), KQU(16689191854356060289),
+	KQU(10992552041730168078), KQU( 7243733172705465836),
+	KQU( 5668075606180319560), KQU(18182847037333286970),
+	KQU( 4290215357664631322), KQU( 4061414220791828613),
+	KQU(13006291061652989604), KQU( 7140491178917128798),
+	KQU(12703446217663283481), KQU( 5500220597564558267),
+	KQU(10330551509971296358), KQU(15958554768648714492),
+	KQU( 5174555954515360045), KQU( 1731318837687577735),
+	KQU( 3557700801048354857), KQU(13764012341928616198),
+	KQU(13115166194379119043), KQU( 7989321021560255519),
+	KQU( 2103584280905877040), KQU( 9230788662155228488),
+	KQU(16396629323325547654), KQU(  657926409811318051),
+	KQU(15046700264391400727), KQU( 5120132858771880830),
+	KQU( 7934160097989028561), KQU( 6963121488531976245),
+	KQU(17412329602621742089), KQU(15144843053931774092),
+	KQU(17204176651763054532), KQU(13166595387554065870),
+	KQU( 8590377810513960213), KQU( 5834365135373991938),
+	KQU( 7640913007182226243), KQU( 3479394703859418425),
+	KQU(16402784452644521040), KQU( 4993979809687083980),
+	KQU(13254522168097688865), KQU(15643659095244365219),
+	KQU( 5881437660538424982), KQU(11174892200618987379),
+	KQU(  254409966159711077), KQU(17158413043140549909),
+	KQU( 3638048789290376272), KQU( 1376816930299489190),
+	KQU( 4622462095217761923), KQU(15086407973010263515),
+	KQU(13253971772784692238), KQU( 5270549043541649236),
+	KQU(11182714186805411604), KQU(12283846437495577140),
+	KQU( 5297647149908953219), KQU(10047451738316836654),
+	KQU( 4938228100367874746), KQU(12328523025304077923),
+	KQU( 3601049438595312361), KQU( 9313624118352733770),
+	KQU(13322966086117661798), KQU(16660005705644029394),
+	KQU(11337677526988872373), KQU(13869299102574417795),
+	KQU(15642043183045645437), KQU( 3021755569085880019),
+	KQU( 4979741767761188161), KQU(13679979092079279587),
+	KQU( 3344685842861071743), KQU(13947960059899588104),
+	KQU(  305806934293368007), KQU( 5749173929201650029),
+	KQU(11123724852118844098), KQU(15128987688788879802),
+	KQU(15251651211024665009), KQU( 7689925933816577776),
+	KQU(16732804392695859449), KQU(17087345401014078468),
+	KQU(14315108589159048871), KQU( 4820700266619778917),
+	KQU(16709637539357958441), KQU( 4936227875177351374),
+	KQU( 2137907697912987247), KQU(11628565601408395420),
+	KQU( 2333250549241556786), KQU( 5711200379577778637),
+	KQU( 5170680131529031729), KQU(12620392043061335164),
+	KQU(   95363390101096078), KQU( 5487981914081709462),
+	KQU( 1763109823981838620), KQU( 3395861271473224396),
+	KQU( 1300496844282213595), KQU( 6894316212820232902),
+	KQU(10673859651135576674), KQU( 5911839658857903252),
+	KQU(17407110743387299102), KQU( 8257427154623140385),
+	KQU(11389003026741800267), KQU( 4070043211095013717),
+	KQU(11663806997145259025), KQU(15265598950648798210),
+	KQU(  630585789434030934), KQU( 3524446529213587334),
+	KQU( 7186424168495184211), KQU(10806585451386379021),
+	KQU(11120017753500499273), KQU( 1586837651387701301),
+	KQU(17530454400954415544), KQU( 9991670045077880430),
+	KQU( 7550997268990730180), KQU( 8640249196597379304),
+	KQU( 3522203892786893823), KQU(10401116549878854788),
+	KQU(13690285544733124852), KQU( 8295785675455774586),
+	KQU(15535716172155117603), KQU( 3112108583723722511),
+	KQU(17633179955339271113), KQU(18154208056063759375),
+	KQU( 1866409236285815666), KQU(13326075895396412882),
+	KQU( 8756261842948020025), KQU( 6281852999868439131),
+	KQU(15087653361275292858), KQU(10333923911152949397),
+	KQU( 5265567645757408500), KQU(12728041843210352184),
+	KQU( 6347959327507828759), KQU(  154112802625564758),
+	KQU(18235228308679780218), KQU( 3253805274673352418),
+	KQU( 4849171610689031197), KQU(17948529398340432518),
+	KQU(13803510475637409167), KQU(13506570190409883095),
+	KQU(15870801273282960805), KQU( 8451286481299170773),
+	KQU( 9562190620034457541), KQU( 8518905387449138364),
+	KQU(12681306401363385655), KQU( 3788073690559762558),
+	KQU( 5256820289573487769), KQU( 2752021372314875467),
+	KQU( 6354035166862520716), KQU( 4328956378309739069),
+	KQU(  449087441228269600), KQU( 5533508742653090868),
+	KQU( 1260389420404746988), KQU(18175394473289055097),
+	KQU( 1535467109660399420), KQU( 8818894282874061442),
+	KQU(12140873243824811213), KQU(15031386653823014946),
+	KQU( 1286028221456149232), KQU( 6329608889367858784),
+	KQU( 9419654354945132725), KQU( 6094576547061672379),
+	KQU(17706217251847450255), KQU( 1733495073065878126),
+	KQU(16918923754607552663), KQU( 8881949849954945044),
+	KQU(12938977706896313891), KQU(14043628638299793407),
+	KQU(18393874581723718233), KQU( 6886318534846892044),
+	KQU(14577870878038334081), KQU(13541558383439414119),
+	KQU(13570472158807588273), KQU(18300760537910283361),
+	KQU(  818368572800609205), KQU( 1417000585112573219),
+	KQU(12337533143867683655), KQU(12433180994702314480),
+	KQU(  778190005829189083), KQU(13667356216206524711),
+	KQU( 9866149895295225230), KQU(11043240490417111999),
+	KQU( 1123933826541378598), KQU( 6469631933605123610),
+	KQU(14508554074431980040), KQU(13918931242962026714),
+	KQU( 2870785929342348285), KQU(14786362626740736974),
+	KQU(13176680060902695786), KQU( 9591778613541679456),
+	KQU( 9097662885117436706), KQU(  749262234240924947),
+	KQU( 1944844067793307093), KQU( 4339214904577487742),
+	KQU( 8009584152961946551), KQU(16073159501225501777),
+	KQU( 3335870590499306217), KQU(17088312653151202847),
+	KQU( 3108893142681931848), KQU(16636841767202792021),
+	KQU(10423316431118400637), KQU( 8008357368674443506),
+	KQU(11340015231914677875), KQU(17687896501594936090),
+	KQU(15173627921763199958), KQU(  542569482243721959),
+	KQU(15071714982769812975), KQU( 4466624872151386956),
+	KQU( 1901780715602332461), KQU( 9822227742154351098),
+	KQU( 1479332892928648780), KQU( 6981611948382474400),
+	KQU( 7620824924456077376), KQU(14095973329429406782),
+	KQU( 7902744005696185404), KQU(15830577219375036920),
+	KQU(10287076667317764416), KQU(12334872764071724025),
+	KQU( 4419302088133544331), KQU(14455842851266090520),
+	KQU(12488077416504654222), KQU( 7953892017701886766),
+	KQU( 6331484925529519007), KQU( 4902145853785030022),
+	KQU(17010159216096443073), KQU(11945354668653886087),
+	KQU(15112022728645230829), KQU(17363484484522986742),
+	KQU( 4423497825896692887), KQU( 8155489510809067471),
+	KQU(  258966605622576285), KQU( 5462958075742020534),
+	KQU( 6763710214913276228), KQU( 2368935183451109054),
+	KQU(14209506165246453811), KQU( 2646257040978514881),
+	KQU( 3776001911922207672), KQU( 1419304601390147631),
+	KQU(14987366598022458284), KQU( 3977770701065815721),
+	KQU(  730820417451838898), KQU( 3982991703612885327),
+	KQU( 2803544519671388477), KQU(17067667221114424649),
+	KQU( 2922555119737867166), KQU( 1989477584121460932),
+	KQU(15020387605892337354), KQU( 9293277796427533547),
+	KQU(10722181424063557247), KQU(16704542332047511651),
+	KQU( 5008286236142089514), KQU(16174732308747382540),
+	KQU(17597019485798338402), KQU(13081745199110622093),
+	KQU( 8850305883842258115), KQU(12723629125624589005),
+	KQU( 8140566453402805978), KQU(15356684607680935061),
+	KQU(14222190387342648650), KQU(11134610460665975178),
+	KQU( 1259799058620984266), KQU(13281656268025610041),
+	KQU(  298262561068153992), KQU(12277871700239212922),
+	KQU(13911297774719779438), KQU(16556727962761474934),
+	KQU(17903010316654728010), KQU( 9682617699648434744),
+	KQU(14757681836838592850), KQU( 1327242446558524473),
+	KQU(11126645098780572792), KQU( 1883602329313221774),
+	KQU( 2543897783922776873), KQU(15029168513767772842),
+	KQU(12710270651039129878), KQU(16118202956069604504),
+	KQU(15010759372168680524), KQU( 2296827082251923948),
+	KQU(10793729742623518101), KQU(13829764151845413046),
+	KQU(17769301223184451213), KQU( 3118268169210783372),
+	KQU(17626204544105123127), KQU( 7416718488974352644),
+	KQU(10450751996212925994), KQU( 9352529519128770586),
+	KQU(  259347569641110140), KQU( 8048588892269692697),
+	KQU( 1774414152306494058), KQU(10669548347214355622),
+	KQU(13061992253816795081), KQU(18432677803063861659),
+	KQU( 8879191055593984333), KQU(12433753195199268041),
+	KQU(14919392415439730602), KQU( 6612848378595332963),
+	KQU( 6320986812036143628), KQU(10465592420226092859),
+	KQU( 4196009278962570808), KQU( 3747816564473572224),
+	KQU(17941203486133732898), KQU( 2350310037040505198),
+	KQU( 5811779859134370113), KQU(10492109599506195126),
+	KQU( 7699650690179541274), KQU( 1954338494306022961),
+	KQU(14095816969027231152), KQU( 5841346919964852061),
+	KQU(14945969510148214735), KQU( 3680200305887550992),
+	KQU( 6218047466131695792), KQU( 8242165745175775096),
+	KQU(11021371934053307357), KQU( 1265099502753169797),
+	KQU( 4644347436111321718), KQU( 3609296916782832859),
+	KQU( 8109807992218521571), KQU(18387884215648662020),
+	KQU(14656324896296392902), KQU(17386819091238216751),
+	KQU(17788300878582317152), KQU( 7919446259742399591),
+	KQU( 4466613134576358004), KQU(12928181023667938509),
+	KQU(13147446154454932030), KQU(16552129038252734620),
+	KQU( 8395299403738822450), KQU(11313817655275361164),
+	KQU(  434258809499511718), KQU( 2074882104954788676),
+	KQU( 7929892178759395518), KQU( 9006461629105745388),
+	KQU( 5176475650000323086), KQU(11128357033468341069),
+	KQU(12026158851559118955), KQU(14699716249471156500),
+	KQU(  448982497120206757), KQU( 4156475356685519900),
+	KQU( 6063816103417215727), KQU(10073289387954971479),
+	KQU( 8174466846138590962), KQU( 2675777452363449006),
+	KQU( 9090685420572474281), KQU( 6659652652765562060),
+	KQU(12923120304018106621), KQU(11117480560334526775),
+	KQU(  937910473424587511), KQU( 1838692113502346645),
+	KQU(11133914074648726180), KQU( 7922600945143884053),
+	KQU(13435287702700959550), KQU( 5287964921251123332),
+	KQU(11354875374575318947), KQU(17955724760748238133),
+	KQU(13728617396297106512), KQU( 4107449660118101255),
+	KQU( 1210269794886589623), KQU(11408687205733456282),
+	KQU( 4538354710392677887), KQU(13566803319341319267),
+	KQU(17870798107734050771), KQU( 3354318982568089135),
+	KQU( 9034450839405133651), KQU(13087431795753424314),
+	KQU(  950333102820688239), KQU( 1968360654535604116),
+	KQU(16840551645563314995), KQU( 8867501803892924995),
+	KQU(11395388644490626845), KQU( 1529815836300732204),
+	KQU(13330848522996608842), KQU( 1813432878817504265),
+	KQU( 2336867432693429560), KQU(15192805445973385902),
+	KQU( 2528593071076407877), KQU(  128459777936689248),
+	KQU( 9976345382867214866), KQU( 6208885766767996043),
+	KQU(14982349522273141706), KQU( 3099654362410737822),
+	KQU(13776700761947297661), KQU( 8806185470684925550),
+	KQU( 8151717890410585321), KQU(  640860591588072925),
+	KQU(14592096303937307465), KQU( 9056472419613564846),
+	KQU(14861544647742266352), KQU(12703771500398470216),
+	KQU( 3142372800384138465), KQU( 6201105606917248196),
+	KQU(18337516409359270184), KQU(15042268695665115339),
+	KQU(15188246541383283846), KQU(12800028693090114519),
+	KQU( 5992859621101493472), KQU(18278043971816803521),
+	KQU( 9002773075219424560), KQU( 7325707116943598353),
+	KQU( 7930571931248040822), KQU( 5645275869617023448),
+	KQU( 7266107455295958487), KQU( 4363664528273524411),
+	KQU(14313875763787479809), KQU(17059695613553486802),
+	KQU( 9247761425889940932), KQU(13704726459237593128),
+	KQU( 2701312427328909832), KQU(17235532008287243115),
+	KQU(14093147761491729538), KQU( 6247352273768386516),
+	KQU( 8268710048153268415), KQU( 7985295214477182083),
+	KQU(15624495190888896807), KQU( 3772753430045262788),
+	KQU( 9133991620474991698), KQU( 5665791943316256028),
+	KQU( 7551996832462193473), KQU(13163729206798953877),
+	KQU( 9263532074153846374), KQU( 1015460703698618353),
+	KQU(17929874696989519390), KQU(18257884721466153847),
+	KQU(16271867543011222991), KQU( 3905971519021791941),
+	KQU(16814488397137052085), KQU( 1321197685504621613),
+	KQU( 2870359191894002181), KQU(14317282970323395450),
+	KQU(13663920845511074366), KQU( 2052463995796539594),
+	KQU(14126345686431444337), KQU( 1727572121947022534),
+	KQU(17793552254485594241), KQU( 6738857418849205750),
+	KQU( 1282987123157442952), KQU(16655480021581159251),
+	KQU( 6784587032080183866), KQU(14726758805359965162),
+	KQU( 7577995933961987349), KQU(12539609320311114036),
+	KQU(10789773033385439494), KQU( 8517001497411158227),
+	KQU(10075543932136339710), KQU(14838152340938811081),
+	KQU( 9560840631794044194), KQU(17445736541454117475),
+	KQU(10633026464336393186), KQU(15705729708242246293),
+	KQU( 1117517596891411098), KQU( 4305657943415886942),
+	KQU( 4948856840533979263), KQU(16071681989041789593),
+	KQU(13723031429272486527), KQU( 7639567622306509462),
+	KQU(12670424537483090390), KQU( 9715223453097197134),
+	KQU( 5457173389992686394), KQU(  289857129276135145),
+	KQU(17048610270521972512), KQU(  692768013309835485),
+	KQU(14823232360546632057), KQU(18218002361317895936),
+	KQU( 3281724260212650204), KQU(16453957266549513795),
+	KQU( 8592711109774511881), KQU(  929825123473369579),
+	KQU(15966784769764367791), KQU( 9627344291450607588),
+	KQU(10849555504977813287), KQU( 9234566913936339275),
+	KQU( 6413807690366911210), KQU(10862389016184219267),
+	KQU(13842504799335374048), KQU( 1531994113376881174),
+	KQU( 2081314867544364459), KQU(16430628791616959932),
+	KQU( 8314714038654394368), KQU( 9155473892098431813),
+	KQU(12577843786670475704), KQU( 4399161106452401017),
+	KQU( 1668083091682623186), KQU( 1741383777203714216),
+	KQU( 2162597285417794374), KQU(15841980159165218736),
+	KQU( 1971354603551467079), KQU( 1206714764913205968),
+	KQU( 4790860439591272330), KQU(14699375615594055799),
+	KQU( 8374423871657449988), KQU(10950685736472937738),
+	KQU(  697344331343267176), KQU(10084998763118059810),
+	KQU(12897369539795983124), KQU(12351260292144383605),
+	KQU( 1268810970176811234), KQU( 7406287800414582768),
+	KQU(  516169557043807831), KQU( 5077568278710520380),
+	KQU( 3828791738309039304), KQU( 7721974069946943610),
+	KQU( 3534670260981096460), KQU( 4865792189600584891),
+	KQU(16892578493734337298), KQU( 9161499464278042590),
+	KQU(11976149624067055931), KQU(13219479887277343990),
+	KQU(14161556738111500680), KQU(14670715255011223056),
+	KQU( 4671205678403576558), KQU(12633022931454259781),
+	KQU(14821376219869187646), KQU(  751181776484317028),
+	KQU( 2192211308839047070), KQU(11787306362361245189),
+	KQU(10672375120744095707), KQU( 4601972328345244467),
+	KQU(15457217788831125879), KQU( 8464345256775460809),
+	KQU(10191938789487159478), KQU( 6184348739615197613),
+	KQU(11425436778806882100), KQU( 2739227089124319793),
+	KQU(  461464518456000551), KQU( 4689850170029177442),
+	KQU( 6120307814374078625), KQU(11153579230681708671),
+	KQU( 7891721473905347926), KQU(10281646937824872400),
+	KQU( 3026099648191332248), KQU( 8666750296953273818),
+	KQU(14978499698844363232), KQU(13303395102890132065),
+	KQU( 8182358205292864080), KQU(10560547713972971291),
+	KQU(11981635489418959093), KQU( 3134621354935288409),
+	KQU(11580681977404383968), KQU(14205530317404088650),
+	KQU( 5997789011854923157), KQU(13659151593432238041),
+	KQU(11664332114338865086), KQU( 7490351383220929386),
+	KQU( 7189290499881530378), KQU(15039262734271020220),
+	KQU( 2057217285976980055), KQU(  555570804905355739),
+	KQU(11235311968348555110), KQU(13824557146269603217),
+	KQU(16906788840653099693), KQU( 7222878245455661677),
+	KQU( 5245139444332423756), KQU( 4723748462805674292),
+	KQU(12216509815698568612), KQU(17402362976648951187),
+	KQU(17389614836810366768), KQU( 4880936484146667711),
+	KQU( 9085007839292639880), KQU(13837353458498535449),
+	KQU(11914419854360366677), KQU(16595890135313864103),
+	KQU( 6313969847197627222), KQU(18296909792163910431),
+	KQU(10041780113382084042), KQU( 2499478551172884794),
+	KQU(11057894246241189489), KQU( 9742243032389068555),
+	KQU(12838934582673196228), KQU(13437023235248490367),
+	KQU(13372420669446163240), KQU( 6752564244716909224),
+	KQU( 7157333073400313737), KQU(12230281516370654308),
+	KQU( 1182884552219419117), KQU( 2955125381312499218),
+	KQU(10308827097079443249), KQU( 1337648572986534958),
+	KQU(16378788590020343939), KQU(  108619126514420935),
+	KQU( 3990981009621629188), KQU( 5460953070230946410),
+	KQU( 9703328329366531883), KQU(13166631489188077236),
+	KQU( 1104768831213675170), KQU( 3447930458553877908),
+	KQU( 8067172487769945676), KQU( 5445802098190775347),
+	KQU( 3244840981648973873), KQU(17314668322981950060),
+	KQU( 5006812527827763807), KQU(18158695070225526260),
+	KQU( 2824536478852417853), KQU(13974775809127519886),
+	KQU( 9814362769074067392), KQU(17276205156374862128),
+	KQU(11361680725379306967), KQU( 3422581970382012542),
+	KQU(11003189603753241266), KQU(11194292945277862261),
+	KQU( 6839623313908521348), KQU(11935326462707324634),
+	KQU( 1611456788685878444), KQU(13112620989475558907),
+	KQU(  517659108904450427), KQU(13558114318574407624),
+	KQU(15699089742731633077), KQU( 4988979278862685458),
+	KQU( 8111373583056521297), KQU( 3891258746615399627),
+	KQU( 8137298251469718086), KQU(12748663295624701649),
+	KQU( 4389835683495292062), KQU( 5775217872128831729),
+	KQU( 9462091896405534927), KQU( 8498124108820263989),
+	KQU( 8059131278842839525), KQU(10503167994254090892),
+	KQU(11613153541070396656), KQU(18069248738504647790),
+	KQU(  570657419109768508), KQU( 3950574167771159665),
+	KQU( 5514655599604313077), KQU( 2908460854428484165),
+	KQU(10777722615935663114), KQU(12007363304839279486),
+	KQU( 9800646187569484767), KQU( 8795423564889864287),
+	KQU(14257396680131028419), KQU( 6405465117315096498),
+	KQU( 7939411072208774878), KQU(17577572378528990006),
+	KQU(14785873806715994850), KQU(16770572680854747390),
+	KQU(18127549474419396481), KQU(11637013449455757750),
+	KQU(14371851933996761086), KQU( 3601181063650110280),
+	KQU( 4126442845019316144), KQU(10198287239244320669),
+	KQU(18000169628555379659), KQU(18392482400739978269),
+	KQU( 6219919037686919957), KQU( 3610085377719446052),
+	KQU( 2513925039981776336), KQU(16679413537926716955),
+	KQU(12903302131714909434), KQU( 5581145789762985009),
+	KQU(12325955044293303233), KQU(17216111180742141204),
+	KQU( 6321919595276545740), KQU( 3507521147216174501),
+	KQU( 9659194593319481840), KQU(11473976005975358326),
+	KQU(14742730101435987026), KQU(  492845897709954780),
+	KQU(16976371186162599676), KQU(17712703422837648655),
+	KQU( 9881254778587061697), KQU( 8413223156302299551),
+	KQU( 1563841828254089168), KQU( 9996032758786671975),
+	KQU(  138877700583772667), KQU(13003043368574995989),
+	KQU( 4390573668650456587), KQU( 8610287390568126755),
+	KQU(15126904974266642199), KQU( 6703637238986057662),
+	KQU( 2873075592956810157), KQU( 6035080933946049418),
+	KQU(13382846581202353014), KQU( 7303971031814642463),
+	KQU(18418024405307444267), KQU( 5847096731675404647),
+	KQU( 4035880699639842500), KQU(11525348625112218478),
+	KQU( 3041162365459574102), KQU( 2604734487727986558),
+	KQU(15526341771636983145), KQU(14556052310697370254),
+	KQU(12997787077930808155), KQU( 9601806501755554499),
+	KQU(11349677952521423389), KQU(14956777807644899350),
+	KQU(16559736957742852721), KQU(12360828274778140726),
+	KQU( 6685373272009662513), KQU(16932258748055324130),
+	KQU(15918051131954158508), KQU( 1692312913140790144),
+	KQU(  546653826801637367), KQU( 5341587076045986652),
+	KQU(14975057236342585662), KQU(12374976357340622412),
+	KQU(10328833995181940552), KQU(12831807101710443149),
+	KQU(10548514914382545716), KQU( 2217806727199715993),
+	KQU(12627067369242845138), KQU( 4598965364035438158),
+	KQU(  150923352751318171), KQU(14274109544442257283),
+	KQU( 4696661475093863031), KQU( 1505764114384654516),
+	KQU(10699185831891495147), KQU( 2392353847713620519),
+	KQU( 3652870166711788383), KQU( 8640653276221911108),
+	KQU( 3894077592275889704), KQU( 4918592872135964845),
+	KQU(16379121273281400789), KQU(12058465483591683656),
+	KQU(11250106829302924945), KQU( 1147537556296983005),
+	KQU( 6376342756004613268), KQU(14967128191709280506),
+	KQU(18007449949790627628), KQU( 9497178279316537841),
+	KQU( 7920174844809394893), KQU(10037752595255719907),
+	KQU(15875342784985217697), KQU(15311615921712850696),
+	KQU( 9552902652110992950), KQU(14054979450099721140),
+	KQU( 5998709773566417349), KQU(18027910339276320187),
+	KQU( 8223099053868585554), KQU( 7842270354824999767),
+	KQU( 4896315688770080292), KQU(12969320296569787895),
+	KQU( 2674321489185759961), KQU( 4053615936864718439),
+	KQU(11349775270588617578), KQU( 4743019256284553975),
+	KQU( 5602100217469723769), KQU(14398995691411527813),
+	KQU( 7412170493796825470), KQU(  836262406131744846),
+	KQU( 8231086633845153022), KQU( 5161377920438552287),
+	KQU( 8828731196169924949), KQU(16211142246465502680),
+	KQU( 3307990879253687818), KQU( 5193405406899782022),
+	KQU( 8510842117467566693), KQU( 6070955181022405365),
+	KQU(14482950231361409799), KQU(12585159371331138077),
+	KQU( 3511537678933588148), KQU( 2041849474531116417),
+	KQU(10944936685095345792), KQU(18303116923079107729),
+	KQU( 2720566371239725320), KQU( 4958672473562397622),
+	KQU( 3032326668253243412), KQU(13689418691726908338),
+	KQU( 1895205511728843996), KQU( 8146303515271990527),
+	KQU(16507343500056113480), KQU(  473996939105902919),
+	KQU( 9897686885246881481), KQU(14606433762712790575),
+	KQU( 6732796251605566368), KQU( 1399778120855368916),
+	KQU(  935023885182833777), KQU(16066282816186753477),
+	KQU( 7291270991820612055), KQU(17530230393129853844),
+	KQU(10223493623477451366), KQU(15841725630495676683),
+	KQU(17379567246435515824), KQU( 8588251429375561971),
+	KQU(18339511210887206423), KQU(17349587430725976100),
+	KQU(12244876521394838088), KQU( 6382187714147161259),
+	KQU(12335807181848950831), KQU(16948885622305460665),
+	KQU(13755097796371520506), KQU(14806740373324947801),
+	KQU( 4828699633859287703), KQU( 8209879281452301604),
+	KQU(12435716669553736437), KQU(13970976859588452131),
+	KQU( 6233960842566773148), KQU(12507096267900505759),
+	KQU( 1198713114381279421), KQU(14989862731124149015),
+	KQU(15932189508707978949), KQU( 2526406641432708722),
+	KQU(   29187427817271982), KQU( 1499802773054556353),
+	KQU(10816638187021897173), KQU( 5436139270839738132),
+	KQU( 6659882287036010082), KQU( 2154048955317173697),
+	KQU(10887317019333757642), KQU(16281091802634424955),
+	KQU(10754549879915384901), KQU(10760611745769249815),
+	KQU( 2161505946972504002), KQU( 5243132808986265107),
+	KQU(10129852179873415416), KQU(  710339480008649081),
+	KQU( 7802129453068808528), KQU(17967213567178907213),
+	KQU(15730859124668605599), KQU(13058356168962376502),
+	KQU( 3701224985413645909), KQU(14464065869149109264),
+	KQU( 9959272418844311646), KQU(10157426099515958752),
+	KQU(14013736814538268528), KQU(17797456992065653951),
+	KQU(17418878140257344806), KQU(15457429073540561521),
+	KQU( 2184426881360949378), KQU( 2062193041154712416),
+	KQU( 8553463347406931661), KQU( 4913057625202871854),
+	KQU( 2668943682126618425), KQU(17064444737891172288),
+	KQU( 4997115903913298637), KQU(12019402608892327416),
+	KQU(17603584559765897352), KQU(11367529582073647975),
+	KQU( 8211476043518436050), KQU( 8676849804070323674),
+	KQU(18431829230394475730), KQU(10490177861361247904),
+	KQU( 9508720602025651349), KQU( 7409627448555722700),
+	KQU( 5804047018862729008), KQU(11943858176893142594),
+	KQU(11908095418933847092), KQU( 5415449345715887652),
+	KQU( 1554022699166156407), KQU( 9073322106406017161),
+	KQU( 7080630967969047082), KQU(18049736940860732943),
+	KQU(12748714242594196794), KQU( 1226992415735156741),
+	KQU(17900981019609531193), KQU(11720739744008710999),
+	KQU( 3006400683394775434), KQU(11347974011751996028),
+	KQU( 3316999628257954608), KQU( 8384484563557639101),
+	KQU(18117794685961729767), KQU( 1900145025596618194),
+	KQU(17459527840632892676), KQU( 5634784101865710994),
+	KQU( 7918619300292897158), KQU( 3146577625026301350),
+	KQU( 9955212856499068767), KQU( 1873995843681746975),
+	KQU( 1561487759967972194), KQU( 8322718804375878474),
+	KQU(11300284215327028366), KQU( 4667391032508998982),
+	KQU( 9820104494306625580), KQU(17922397968599970610),
+	KQU( 1784690461886786712), KQU(14940365084341346821),
+	KQU( 5348719575594186181), KQU(10720419084507855261),
+	KQU(14210394354145143274), KQU( 2426468692164000131),
+	KQU(16271062114607059202), KQU(14851904092357070247),
+	KQU( 6524493015693121897), KQU( 9825473835127138531),
+	KQU(14222500616268569578), KQU(15521484052007487468),
+	KQU(14462579404124614699), KQU(11012375590820665520),
+	KQU(11625327350536084927), KQU(14452017765243785417),
+	KQU( 9989342263518766305), KQU( 3640105471101803790),
+	KQU( 4749866455897513242), KQU(13963064946736312044),
+	KQU(10007416591973223791), KQU(18314132234717431115),
+	KQU( 3286596588617483450), KQU( 7726163455370818765),
+	KQU( 7575454721115379328), KQU( 5308331576437663422),
+	KQU(18288821894903530934), KQU( 8028405805410554106),
+	KQU(15744019832103296628), KQU(  149765559630932100),
+	KQU( 6137705557200071977), KQU(14513416315434803615),
+	KQU(11665702820128984473), KQU(  218926670505601386),
+	KQU( 6868675028717769519), KQU(15282016569441512302),
+	KQU( 5707000497782960236), KQU( 6671120586555079567),
+	KQU( 2194098052618985448), KQU(16849577895477330978),
+	KQU(12957148471017466283), KQU( 1997805535404859393),
+	KQU( 1180721060263860490), KQU(13206391310193756958),
+	KQU(12980208674461861797), KQU( 3825967775058875366),
+	KQU(17543433670782042631), KQU( 1518339070120322730),
+	KQU(16344584340890991669), KQU( 2611327165318529819),
+	KQU(11265022723283422529), KQU( 4001552800373196817),
+	KQU(14509595890079346161), KQU( 3528717165416234562),
+	KQU(18153222571501914072), KQU( 9387182977209744425),
+	KQU(10064342315985580021), KQU(11373678413215253977),
+	KQU( 2308457853228798099), KQU( 9729042942839545302),
+	KQU( 7833785471140127746), KQU( 6351049900319844436),
+	KQU(14454610627133496067), KQU(12533175683634819111),
+	KQU(15570163926716513029), KQU(13356980519185762498)
+};
+
+TEST_BEGIN(test_gen_rand_32) {
+	uint32_t array32[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	uint32_t array32_2[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint32_t r32;
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size32(), BLOCK_SIZE,
+	    "Array size too small");
+	ctx = init_gen_rand(1234);
+	fill_array32(ctx, array32, BLOCK_SIZE);
+	fill_array32(ctx, array32_2, BLOCK_SIZE);
+	fini_gen_rand(ctx);
+
+	ctx = init_gen_rand(1234);
+	for (i = 0; i < BLOCK_SIZE; i++) {
+		if (i < COUNT_1) {
+			assert_u32_eq(array32[i], init_gen_rand_32_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32[i],
+		    "Mismatch at array32[%d]=%x, gen=%x", i, array32[i], r32);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32_2[i],
+		    "Mismatch at array32_2[%d]=%x, gen=%x", i, array32_2[i],
+		    r32);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+TEST_BEGIN(test_by_array_32) {
+	uint32_t array32[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	uint32_t array32_2[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint32_t ini[4] = {0x1234, 0x5678, 0x9abc, 0xdef0};
+	uint32_t r32;
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size32(), BLOCK_SIZE,
+	    "Array size too small");
+	ctx = init_by_array(ini, 4);
+	fill_array32(ctx, array32, BLOCK_SIZE);
+	fill_array32(ctx, array32_2, BLOCK_SIZE);
+	fini_gen_rand(ctx);
+
+	ctx = init_by_array(ini, 4);
+	for (i = 0; i < BLOCK_SIZE; i++) {
+		if (i < COUNT_1) {
+			assert_u32_eq(array32[i], init_by_array_32_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32[i],
+		    "Mismatch at array32[%d]=%x, gen=%x", i, array32[i], r32);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32_2[i],
+		    "Mismatch at array32_2[%d]=%x, gen=%x", i, array32_2[i],
+		    r32);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+TEST_BEGIN(test_gen_rand_64) {
+	uint64_t array64[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	uint64_t array64_2[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint64_t r;
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size64(), BLOCK_SIZE64,
+	    "Array size too small");
+	ctx = init_gen_rand(4321);
+	fill_array64(ctx, array64, BLOCK_SIZE64);
+	fill_array64(ctx, array64_2, BLOCK_SIZE64);
+	fini_gen_rand(ctx);
+
+	ctx = init_gen_rand(4321);
+	for (i = 0; i < BLOCK_SIZE64; i++) {
+		if (i < COUNT_1) {
+			assert_u64_eq(array64[i], init_gen_rand_64_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64[i],
+		    "Mismatch at array64[%d]=%"FMTx64", gen=%"FMTx64, i,
+		    array64[i], r);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64_2[i],
+		    "Mismatch at array64_2[%d]=%"FMTx64" gen=%"FMTx64"", i,
+		    array64_2[i], r);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+TEST_BEGIN(test_by_array_64) {
+	uint64_t array64[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	uint64_t array64_2[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint64_t r;
+	uint32_t ini[] = {5, 4, 3, 2, 1};
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size64(), BLOCK_SIZE64,
+	    "Array size too small");
+	ctx = init_by_array(ini, 5);
+	fill_array64(ctx, array64, BLOCK_SIZE64);
+	fill_array64(ctx, array64_2, BLOCK_SIZE64);
+	fini_gen_rand(ctx);
+
+	ctx = init_by_array(ini, 5);
+	for (i = 0; i < BLOCK_SIZE64; i++) {
+		if (i < COUNT_1) {
+			assert_u64_eq(array64[i], init_by_array_64_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64[i],
+		    "Mismatch at array64[%d]=%"FMTx64" gen=%"FMTx64, i,
+		    array64[i], r);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64_2[i],
+		    "Mismatch at array64_2[%d]=%"FMTx64" gen=%"FMTx64, i,
+		    array64_2[i], r);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_gen_rand_32,
+	    test_by_array_32,
+	    test_gen_rand_64,
+	    test_by_array_64);
+}
diff --git a/jemalloc/test/unit/a0.c b/jemalloc/test/unit/a0.c
new file mode 100644
index 0000000..a27ab3f
--- /dev/null
+++ b/jemalloc/test/unit/a0.c
@@ -0,0 +1,16 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_a0) {
+	void *p;
+
+	p = a0malloc(1);
+	assert_ptr_not_null(p, "Unexpected a0malloc() error");
+	a0dalloc(p);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_malloc_init(
+	    test_a0);
+}
diff --git a/jemalloc/test/unit/arena_reset.c b/jemalloc/test/unit/arena_reset.c
new file mode 100644
index 0000000..b182f31
--- /dev/null
+++ b/jemalloc/test/unit/arena_reset.c
@@ -0,0 +1,349 @@
+#ifndef ARENA_RESET_PROF_C_
+#include "test/jemalloc_test.h"
+#endif
+
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/rtree.h"
+
+#include "test/extent_hooks.h"
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nsmall(void) {
+	return get_nsizes_impl("arenas.nbins");
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_small_size(size_t ind) {
+	return get_size_impl("arenas.bin.0.size", ind);
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+/* Like ivsalloc(), but safe to call on discarded allocations. */
+static size_t
+vsalloc(tsdn_t *tsdn, const void *ptr) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent;
+	szind_t szind;
+	if (rtree_extent_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, false, &extent, &szind)) {
+		return 0;
+	}
+
+	if (extent == NULL) {
+		return 0;
+	}
+	if (extent_state_get(extent) != extent_state_active) {
+		return 0;
+	}
+
+	if (szind == SC_NSIZES) {
+		return 0;
+	}
+
+	return sz_index2size(szind);
+}
+
+static unsigned
+do_arena_create(extent_hooks_t *h) {
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz,
+	    (void *)(h != NULL ? &h : NULL), (h != NULL ? sizeof(h) : 0)), 0,
+	    "Unexpected mallctl() failure");
+	return arena_ind;
+}
+
+static void
+do_arena_reset_pre(unsigned arena_ind, void ***ptrs, unsigned *nptrs) {
+#define NLARGE	32
+	unsigned nsmall, nlarge, i;
+	size_t sz;
+	int flags;
+	tsdn_t *tsdn;
+
+	flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+
+	nsmall = get_nsmall();
+	nlarge = get_nlarge() > NLARGE ? NLARGE : get_nlarge();
+	*nptrs = nsmall + nlarge;
+	*ptrs = (void **)malloc(*nptrs * sizeof(void *));
+	assert_ptr_not_null(*ptrs, "Unexpected malloc() failure");
+
+	/* Allocate objects with a wide range of sizes. */
+	for (i = 0; i < nsmall; i++) {
+		sz = get_small_size(i);
+		(*ptrs)[i] = mallocx(sz, flags);
+		assert_ptr_not_null((*ptrs)[i],
+		    "Unexpected mallocx(%zu, %#x) failure", sz, flags);
+	}
+	for (i = 0; i < nlarge; i++) {
+		sz = get_large_size(i);
+		(*ptrs)[nsmall + i] = mallocx(sz, flags);
+		assert_ptr_not_null((*ptrs)[i],
+		    "Unexpected mallocx(%zu, %#x) failure", sz, flags);
+	}
+
+	tsdn = tsdn_fetch();
+
+	/* Verify allocations. */
+	for (i = 0; i < *nptrs; i++) {
+		assert_zu_gt(ivsalloc(tsdn, (*ptrs)[i]), 0,
+		    "Allocation should have queryable size");
+	}
+}
+
+static void
+do_arena_reset_post(void **ptrs, unsigned nptrs, unsigned arena_ind) {
+	tsdn_t *tsdn;
+	unsigned i;
+
+	tsdn = tsdn_fetch();
+
+	if (have_background_thread) {
+		malloc_mutex_lock(tsdn,
+		    &background_thread_info_get(arena_ind)->mtx);
+	}
+	/* Verify allocations no longer exist. */
+	for (i = 0; i < nptrs; i++) {
+		assert_zu_eq(vsalloc(tsdn, ptrs[i]), 0,
+		    "Allocation should no longer exist");
+	}
+	if (have_background_thread) {
+		malloc_mutex_unlock(tsdn,
+		    &background_thread_info_get(arena_ind)->mtx);
+	}
+
+	free(ptrs);
+}
+
+static void
+do_arena_reset_destroy(const char *name, unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib(name, mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+static void
+do_arena_reset(unsigned arena_ind) {
+	do_arena_reset_destroy("arena.0.reset", arena_ind);
+}
+
+static void
+do_arena_destroy(unsigned arena_ind) {
+	do_arena_reset_destroy("arena.0.destroy", arena_ind);
+}
+
+TEST_BEGIN(test_arena_reset) {
+	unsigned arena_ind;
+	void **ptrs;
+	unsigned nptrs;
+
+	arena_ind = do_arena_create(NULL);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+	do_arena_reset(arena_ind);
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+}
+TEST_END
+
+static bool
+arena_i_initialized(unsigned arena_ind, bool refresh) {
+	bool initialized;
+	size_t mib[3];
+	size_t miblen, sz;
+
+	if (refresh) {
+		uint64_t epoch = 1;
+		assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+		    sizeof(epoch)), 0, "Unexpected mallctl() failure");
+	}
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.initialized", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	sz = sizeof(initialized);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&initialized, &sz, NULL,
+	    0), 0, "Unexpected mallctlbymib() failure");
+
+	return initialized;
+}
+
+TEST_BEGIN(test_arena_destroy_initial) {
+	assert_false(arena_i_initialized(MALLCTL_ARENAS_DESTROYED, false),
+	    "Destroyed arena stats should not be initialized");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_destroy_hooks_default) {
+	unsigned arena_ind, arena_ind_another, arena_ind_prev;
+	void **ptrs;
+	unsigned nptrs;
+
+	arena_ind = do_arena_create(NULL);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+
+	assert_false(arena_i_initialized(arena_ind, false),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(arena_ind, true),
+	    "Arena stats should be initialized");
+
+	/*
+	 * Create another arena before destroying one, to better verify arena
+	 * index reuse.
+	 */
+	arena_ind_another = do_arena_create(NULL);
+
+	do_arena_destroy(arena_ind);
+
+	assert_false(arena_i_initialized(arena_ind, true),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(MALLCTL_ARENAS_DESTROYED, false),
+	    "Destroyed arena stats should be initialized");
+
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+
+	arena_ind_prev = arena_ind;
+	arena_ind = do_arena_create(NULL);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+	assert_u_eq(arena_ind, arena_ind_prev,
+	    "Arena index should have been recycled");
+	do_arena_destroy(arena_ind);
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+
+	do_arena_destroy(arena_ind_another);
+}
+TEST_END
+
+/*
+ * Actually unmap extents, regardless of opt_retain, so that attempts to access
+ * a destroyed arena's memory will segfault.
+ */
+static bool
+extent_dalloc_unmap(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, committed=%s, "
+	    "arena_ind=%u)\n", __func__, extent_hooks, addr, size, committed ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->dalloc, extent_dalloc_unmap,
+	    "Wrong hook function");
+	called_dalloc = true;
+	if (!try_dalloc) {
+		return true;
+	}
+	did_dalloc = true;
+	if (!maps_coalesce && opt_retain) {
+		return true;
+	}
+	pages_unmap(addr, size);
+	return false;
+}
+
+static extent_hooks_t hooks_orig;
+
+static extent_hooks_t hooks_unmap = {
+	extent_alloc_hook,
+	extent_dalloc_unmap, /* dalloc */
+	extent_destroy_hook,
+	extent_commit_hook,
+	extent_decommit_hook,
+	extent_purge_lazy_hook,
+	extent_purge_forced_hook,
+	extent_split_hook,
+	extent_merge_hook
+};
+
+TEST_BEGIN(test_arena_destroy_hooks_unmap) {
+	unsigned arena_ind;
+	void **ptrs;
+	unsigned nptrs;
+
+	extent_hooks_prep();
+	if (maps_coalesce) {
+		try_decommit = false;
+	}
+	memcpy(&hooks_orig, &hooks, sizeof(extent_hooks_t));
+	memcpy(&hooks, &hooks_unmap, sizeof(extent_hooks_t));
+
+	did_alloc = false;
+	arena_ind = do_arena_create(&hooks);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+
+	assert_true(did_alloc, "Expected alloc");
+
+	assert_false(arena_i_initialized(arena_ind, false),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(arena_ind, true),
+	    "Arena stats should be initialized");
+
+	did_dalloc = false;
+	do_arena_destroy(arena_ind);
+	assert_true(did_dalloc, "Expected dalloc");
+
+	assert_false(arena_i_initialized(arena_ind, true),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(MALLCTL_ARENAS_DESTROYED, false),
+	    "Destroyed arena stats should be initialized");
+
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+
+	memcpy(&hooks, &hooks_orig, sizeof(extent_hooks_t));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_arena_reset,
+	    test_arena_destroy_initial,
+	    test_arena_destroy_hooks_default,
+	    test_arena_destroy_hooks_unmap);
+}
diff --git a/jemalloc/test/unit/arena_reset_prof.c b/jemalloc/test/unit/arena_reset_prof.c
new file mode 100644
index 0000000..38d8012
--- /dev/null
+++ b/jemalloc/test/unit/arena_reset_prof.c
@@ -0,0 +1,4 @@
+#include "test/jemalloc_test.h"
+#define ARENA_RESET_PROF_C_
+
+#include "arena_reset.c"
diff --git a/jemalloc/test/unit/arena_reset_prof.sh b/jemalloc/test/unit/arena_reset_prof.sh
new file mode 100644
index 0000000..041dc1c
--- /dev/null
+++ b/jemalloc/test/unit/arena_reset_prof.sh
@@ -0,0 +1,3 @@
+#!/bin/sh
+
+export MALLOC_CONF="prof:true,lg_prof_sample:0"
diff --git a/jemalloc/test/unit/atomic.c b/jemalloc/test/unit/atomic.c
new file mode 100644
index 0000000..572d8d2
--- /dev/null
+++ b/jemalloc/test/unit/atomic.c
@@ -0,0 +1,229 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * We *almost* have consistent short names (e.g. "u32" for uint32_t, "b" for
+ * bool, etc.  The one exception is that the short name for void * is "p" in
+ * some places and "ptr" in others.  In the long run it would be nice to unify
+ * these, but in the short run we'll use this shim.
+ */
+#define assert_p_eq assert_ptr_eq
+
+/*
+ * t: the non-atomic type, like "uint32_t".
+ * ta: the short name for the type, like "u32".
+ * val[1,2,3]: Values of the given type.  The CAS tests use val2 for expected,
+ * and val3 for desired.
+ */
+
+#define DO_TESTS(t, ta, val1, val2, val3) do {				\
+	t val;								\
+	t expected;							\
+	bool success;							\
+	/* This (along with the load below) also tests ATOMIC_LOAD. */	\
+	atomic_##ta##_t atom = ATOMIC_INIT(val1);			\
+									\
+	/* ATOMIC_INIT and load. */					\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1, val, "Load or init failed");		\
+									\
+	/* Store. */							\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	atomic_store_##ta(&atom, val2, ATOMIC_RELAXED);			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val2, val, "Store failed");			\
+									\
+	/* Exchange. */							\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_exchange_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val, "Exchange returned invalid value");	\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val2, val, "Exchange store invalid value");	\
+									\
+	/* 								\
+	 * Weak CAS.  Spurious failures are allowed, so we loop a few	\
+	 * times.							\
+	 */								\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	success = false;						\
+	for (int i = 0; i < 10 && !success; i++) {			\
+		expected = val2;					\
+		success = atomic_compare_exchange_weak_##ta(&atom,	\
+		    &expected, val3, ATOMIC_RELAXED, ATOMIC_RELAXED);	\
+		assert_##ta##_eq(val1, expected, 			\
+		    "CAS should update expected");			\
+	}								\
+	assert_b_eq(val1 == val2, success,				\
+	    "Weak CAS did the wrong state update");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	if (success) {							\
+		assert_##ta##_eq(val3, val,				\
+		    "Successful CAS should update atomic");		\
+	} else {							\
+		assert_##ta##_eq(val1, val,				\
+		    "Unsuccessful CAS should not update atomic");	\
+	}								\
+									\
+	/* Strong CAS. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	expected = val2;						\
+	success = atomic_compare_exchange_strong_##ta(&atom, &expected,	\
+	    val3, ATOMIC_RELAXED, ATOMIC_RELAXED);			\
+	assert_b_eq(val1 == val2, success,				\
+	    "Strong CAS did the wrong state update");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	if (success) {							\
+		assert_##ta##_eq(val3, val,				\
+		    "Successful CAS should update atomic");		\
+	} else {							\
+		assert_##ta##_eq(val1, val,				\
+		    "Unsuccessful CAS should not update atomic");	\
+	}								\
+									\
+									\
+} while (0)
+
+#define DO_INTEGER_TESTS(t, ta, val1, val2) do {			\
+	atomic_##ta##_t atom;						\
+	t val;								\
+									\
+	/* Fetch-add. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_add_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-add should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 + val2, val,				\
+	    "Fetch-add should update atomic");				\
+									\
+	/* Fetch-sub. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_sub_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-sub should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 - val2, val,				\
+	    "Fetch-sub should update atomic");				\
+									\
+	/* Fetch-and. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_and_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-and should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 & val2, val,				\
+	    "Fetch-and should update atomic");				\
+									\
+	/* Fetch-or. */							\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_or_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-or should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 | val2, val,				\
+	    "Fetch-or should update atomic");				\
+									\
+	/* Fetch-xor. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_xor_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-xor should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 ^ val2, val,				\
+	    "Fetch-xor should update atomic");				\
+} while (0)
+
+#define TEST_STRUCT(t, ta)						\
+typedef struct {							\
+	t val1;								\
+	t val2;								\
+	t val3;								\
+} ta##_test_t;
+
+#define TEST_CASES(t) {							\
+	{(t)-1, (t)-1, (t)-2},						\
+	{(t)-1, (t) 0, (t)-2},						\
+	{(t)-1, (t) 1, (t)-2},						\
+									\
+	{(t) 0, (t)-1, (t)-2},						\
+	{(t) 0, (t) 0, (t)-2},						\
+	{(t) 0, (t) 1, (t)-2},						\
+									\
+	{(t) 1, (t)-1, (t)-2},						\
+	{(t) 1, (t) 0, (t)-2},						\
+	{(t) 1, (t) 1, (t)-2},						\
+									\
+	{(t)0, (t)-(1 << 22), (t)-2},					\
+	{(t)0, (t)(1 << 22), (t)-2},					\
+	{(t)(1 << 22), (t)-(1 << 22), (t)-2},				\
+	{(t)(1 << 22), (t)(1 << 22), (t)-2}				\
+}
+
+#define TEST_BODY(t, ta) do {						\
+	const ta##_test_t tests[] = TEST_CASES(t);			\
+	for (unsigned i = 0; i < sizeof(tests)/sizeof(tests[0]); i++) {	\
+		ta##_test_t test = tests[i];				\
+		DO_TESTS(t, ta, test.val1, test.val2, test.val3);	\
+	}								\
+} while (0)
+
+#define INTEGER_TEST_BODY(t, ta) do {					\
+	const ta##_test_t tests[] = TEST_CASES(t);			\
+	for (unsigned i = 0; i < sizeof(tests)/sizeof(tests[0]); i++) {	\
+		ta##_test_t test = tests[i];				\
+		DO_TESTS(t, ta, test.val1, test.val2, test.val3);	\
+		DO_INTEGER_TESTS(t, ta, test.val1, test.val2);		\
+	}								\
+} while (0)
+
+TEST_STRUCT(uint64_t, u64);
+TEST_BEGIN(test_atomic_u64) {
+#if !(LG_SIZEOF_PTR == 3 || LG_SIZEOF_INT == 3)
+	test_skip("64-bit atomic operations not supported");
+#else
+	INTEGER_TEST_BODY(uint64_t, u64);
+#endif
+}
+TEST_END
+
+
+TEST_STRUCT(uint32_t, u32);
+TEST_BEGIN(test_atomic_u32) {
+	INTEGER_TEST_BODY(uint32_t, u32);
+}
+TEST_END
+
+TEST_STRUCT(void *, p);
+TEST_BEGIN(test_atomic_p) {
+	TEST_BODY(void *, p);
+}
+TEST_END
+
+TEST_STRUCT(size_t, zu);
+TEST_BEGIN(test_atomic_zu) {
+	INTEGER_TEST_BODY(size_t, zu);
+}
+TEST_END
+
+TEST_STRUCT(ssize_t, zd);
+TEST_BEGIN(test_atomic_zd) {
+	INTEGER_TEST_BODY(ssize_t, zd);
+}
+TEST_END
+
+
+TEST_STRUCT(unsigned, u);
+TEST_BEGIN(test_atomic_u) {
+	INTEGER_TEST_BODY(unsigned, u);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_atomic_u64,
+	    test_atomic_u32,
+	    test_atomic_p,
+	    test_atomic_zu,
+	    test_atomic_zd,
+	    test_atomic_u);
+}
diff --git a/jemalloc/test/unit/background_thread.c b/jemalloc/test/unit/background_thread.c
new file mode 100644
index 0000000..f7bd37c
--- /dev/null
+++ b/jemalloc/test/unit/background_thread.c
@@ -0,0 +1,119 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/util.h"
+
+static void
+test_switch_background_thread_ctl(bool new_val) {
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+
+	e1 = new_val;
+	assert_d_eq(mallctl("background_thread", (void *)&e0, &sz,
+	    &e1, sz), 0, "Unexpected mallctl() failure");
+	assert_b_eq(e0, !e1,
+	    "background_thread should be %d before.\n", !e1);
+	if (e1) {
+		assert_zu_gt(n_background_threads, 0,
+		    "Number of background threads should be non zero.\n");
+	} else {
+		assert_zu_eq(n_background_threads, 0,
+		    "Number of background threads should be zero.\n");
+	}
+}
+
+static void
+test_repeat_background_thread_ctl(bool before) {
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+
+	e1 = before;
+	assert_d_eq(mallctl("background_thread", (void *)&e0, &sz,
+	    &e1, sz), 0, "Unexpected mallctl() failure");
+	assert_b_eq(e0, before,
+	    "background_thread should be %d.\n", before);
+	if (e1) {
+		assert_zu_gt(n_background_threads, 0,
+		    "Number of background threads should be non zero.\n");
+	} else {
+		assert_zu_eq(n_background_threads, 0,
+		    "Number of background threads should be zero.\n");
+	}
+}
+
+TEST_BEGIN(test_background_thread_ctl) {
+	test_skip_if(!have_background_thread);
+
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+
+	assert_d_eq(mallctl("opt.background_thread", (void *)&e0, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctl("background_thread", (void *)&e1, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() failure");
+	assert_b_eq(e0, e1,
+	    "Default and opt.background_thread does not match.\n");
+	if (e0) {
+		test_switch_background_thread_ctl(false);
+	}
+	assert_zu_eq(n_background_threads, 0,
+	    "Number of background threads should be 0.\n");
+
+	for (unsigned i = 0; i < 4; i++) {
+		test_switch_background_thread_ctl(true);
+		test_repeat_background_thread_ctl(true);
+		test_repeat_background_thread_ctl(true);
+
+		test_switch_background_thread_ctl(false);
+		test_repeat_background_thread_ctl(false);
+		test_repeat_background_thread_ctl(false);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_background_thread_running) {
+	test_skip_if(!have_background_thread);
+	test_skip_if(!config_stats);
+
+#if defined(JEMALLOC_BACKGROUND_THREAD)
+	tsd_t *tsd = tsd_fetch();
+	background_thread_info_t *info = &background_thread_info[0];
+
+	test_repeat_background_thread_ctl(false);
+	test_switch_background_thread_ctl(true);
+	assert_b_eq(info->state, background_thread_started,
+	    "Background_thread did not start.\n");
+
+	nstime_t start, now;
+	nstime_init(&start, 0);
+	nstime_update(&start);
+
+	bool ran = false;
+	while (true) {
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+		if (info->tot_n_runs > 0) {
+			ran = true;
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		if (ran) {
+			break;
+		}
+
+		nstime_init(&now, 0);
+		nstime_update(&now);
+		nstime_subtract(&now, &start);
+		assert_u64_lt(nstime_sec(&now), 1000,
+		    "Background threads did not run for 1000 seconds.");
+		sleep(1);
+	}
+	test_switch_background_thread_ctl(false);
+#endif
+}
+TEST_END
+
+int
+main(void) {
+	/* Background_thread creation tests reentrancy naturally. */
+	return test_no_reentrancy(
+	    test_background_thread_ctl,
+	    test_background_thread_running);
+}
diff --git a/jemalloc/test/unit/background_thread_enable.c b/jemalloc/test/unit/background_thread_enable.c
new file mode 100644
index 0000000..d894e93
--- /dev/null
+++ b/jemalloc/test/unit/background_thread_enable.c
@@ -0,0 +1,85 @@
+#include "test/jemalloc_test.h"
+
+const char *malloc_conf = "background_thread:false,narenas:1,max_background_threads:20";
+
+TEST_BEGIN(test_deferred) {
+	test_skip_if(!have_background_thread);
+
+	unsigned id;
+	size_t sz_u = sizeof(unsigned);
+
+	/*
+	 * 10 here is somewhat arbitrary, except insofar as we want to ensure
+	 * that the number of background threads is smaller than the number of
+	 * arenas.  I'll ragequit long before we have to spin up 10 threads per
+	 * cpu to handle background purging, so this is a conservative
+	 * approximation.
+	 */
+	for (unsigned i = 0; i < 10 * ncpus; i++) {
+		assert_d_eq(mallctl("arenas.create", &id, &sz_u, NULL, 0), 0,
+		    "Failed to create arena");
+	}
+
+	bool enable = true;
+	size_t sz_b = sizeof(bool);
+	assert_d_eq(mallctl("background_thread", NULL, NULL, &enable, sz_b), 0,
+	    "Failed to enable background threads");
+	enable = false;
+	assert_d_eq(mallctl("background_thread", NULL, NULL, &enable, sz_b), 0,
+	    "Failed to disable background threads");
+}
+TEST_END
+
+TEST_BEGIN(test_max_background_threads) {
+	test_skip_if(!have_background_thread);
+
+	size_t max_n_thds;
+	size_t opt_max_n_thds;
+	size_t sz_m = sizeof(max_n_thds);
+	assert_d_eq(mallctl("opt.max_background_threads",
+	    &opt_max_n_thds, &sz_m, NULL, 0), 0,
+	    "Failed to get opt.max_background_threads");
+	assert_d_eq(mallctl("max_background_threads", &max_n_thds, &sz_m, NULL,
+	    0), 0, "Failed to get max background threads");
+	assert_zu_eq(opt_max_n_thds, max_n_thds,
+	    "max_background_threads and "
+	    "opt.max_background_threads should match");
+	assert_d_eq(mallctl("max_background_threads", NULL, NULL, &max_n_thds,
+	    sz_m), 0, "Failed to set max background threads");
+
+	unsigned id;
+	size_t sz_u = sizeof(unsigned);
+
+	for (unsigned i = 0; i < 10 * ncpus; i++) {
+		assert_d_eq(mallctl("arenas.create", &id, &sz_u, NULL, 0), 0,
+		    "Failed to create arena");
+	}
+
+	bool enable = true;
+	size_t sz_b = sizeof(bool);
+	assert_d_eq(mallctl("background_thread", NULL, NULL, &enable, sz_b), 0,
+	    "Failed to enable background threads");
+	assert_zu_eq(n_background_threads, max_n_thds,
+	    "Number of background threads should not change.\n");
+	size_t new_max_thds = max_n_thds - 1;
+	if (new_max_thds > 0) {
+		assert_d_eq(mallctl("max_background_threads", NULL, NULL,
+		    &new_max_thds, sz_m), 0,
+		    "Failed to set max background threads");
+		assert_zu_eq(n_background_threads, new_max_thds,
+		    "Number of background threads should decrease by 1.\n");
+	}
+	new_max_thds = 1;
+	assert_d_eq(mallctl("max_background_threads", NULL, NULL, &new_max_thds,
+	    sz_m), 0, "Failed to set max background threads");
+	assert_zu_eq(n_background_threads, new_max_thds,
+	    "Number of background threads should be 1.\n");
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+		test_deferred,
+		test_max_background_threads);
+}
diff --git a/jemalloc/test/unit/base.c b/jemalloc/test/unit/base.c
new file mode 100644
index 0000000..6b792cf
--- /dev/null
+++ b/jemalloc/test/unit/base.c
@@ -0,0 +1,234 @@
+#include "test/jemalloc_test.h"
+
+#include "test/extent_hooks.h"
+
+static extent_hooks_t hooks_null = {
+	extent_alloc_hook,
+	NULL, /* dalloc */
+	NULL, /* destroy */
+	NULL, /* commit */
+	NULL, /* decommit */
+	NULL, /* purge_lazy */
+	NULL, /* purge_forced */
+	NULL, /* split */
+	NULL /* merge */
+};
+
+static extent_hooks_t hooks_not_null = {
+	extent_alloc_hook,
+	extent_dalloc_hook,
+	extent_destroy_hook,
+	NULL, /* commit */
+	extent_decommit_hook,
+	extent_purge_lazy_hook,
+	extent_purge_forced_hook,
+	NULL, /* split */
+	NULL /* merge */
+};
+
+TEST_BEGIN(test_base_hooks_default) {
+	base_t *base;
+	size_t allocated0, allocated1, resident, mapped, n_thp;
+
+	tsdn_t *tsdn = tsd_tsdn(tsd_fetch());
+	base = base_new(tsdn, 0, (extent_hooks_t *)&extent_hooks_default);
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated0, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated0, sizeof(base_t),
+		    "Base header should count as allocated");
+		if (opt_metadata_thp == metadata_thp_always) {
+			assert_zu_gt(n_thp, 0,
+			    "Base should have 1 THP at least.");
+		}
+	}
+
+	assert_ptr_not_null(base_alloc(tsdn, base, 42, 1),
+	    "Unexpected base_alloc() failure");
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated1, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated1 - allocated0, 42,
+		    "At least 42 bytes were allocated by base_alloc()");
+	}
+
+	base_delete(tsdn, base);
+}
+TEST_END
+
+TEST_BEGIN(test_base_hooks_null) {
+	extent_hooks_t hooks_orig;
+	base_t *base;
+	size_t allocated0, allocated1, resident, mapped, n_thp;
+
+	extent_hooks_prep();
+	try_dalloc = false;
+	try_destroy = true;
+	try_decommit = false;
+	try_purge_lazy = false;
+	try_purge_forced = false;
+	memcpy(&hooks_orig, &hooks, sizeof(extent_hooks_t));
+	memcpy(&hooks, &hooks_null, sizeof(extent_hooks_t));
+
+	tsdn_t *tsdn = tsd_tsdn(tsd_fetch());
+	base = base_new(tsdn, 0, &hooks);
+	assert_ptr_not_null(base, "Unexpected base_new() failure");
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated0, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated0, sizeof(base_t),
+		    "Base header should count as allocated");
+		if (opt_metadata_thp == metadata_thp_always) {
+			assert_zu_gt(n_thp, 0,
+			    "Base should have 1 THP at least.");
+		}
+	}
+
+	assert_ptr_not_null(base_alloc(tsdn, base, 42, 1),
+	    "Unexpected base_alloc() failure");
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated1, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated1 - allocated0, 42,
+		    "At least 42 bytes were allocated by base_alloc()");
+	}
+
+	base_delete(tsdn, base);
+
+	memcpy(&hooks, &hooks_orig, sizeof(extent_hooks_t));
+}
+TEST_END
+
+TEST_BEGIN(test_base_hooks_not_null) {
+	extent_hooks_t hooks_orig;
+	base_t *base;
+	void *p, *q, *r, *r_exp;
+
+	extent_hooks_prep();
+	try_dalloc = false;
+	try_destroy = true;
+	try_decommit = false;
+	try_purge_lazy = false;
+	try_purge_forced = false;
+	memcpy(&hooks_orig, &hooks, sizeof(extent_hooks_t));
+	memcpy(&hooks, &hooks_not_null, sizeof(extent_hooks_t));
+
+	tsdn_t *tsdn = tsd_tsdn(tsd_fetch());
+	did_alloc = false;
+	base = base_new(tsdn, 0, &hooks);
+	assert_ptr_not_null(base, "Unexpected base_new() failure");
+	assert_true(did_alloc, "Expected alloc");
+
+	/*
+	 * Check for tight packing at specified alignment under simple
+	 * conditions.
+	 */
+	{
+		const size_t alignments[] = {
+			1,
+			QUANTUM,
+			QUANTUM << 1,
+			CACHELINE,
+			CACHELINE << 1,
+		};
+		unsigned i;
+
+		for (i = 0; i < sizeof(alignments) / sizeof(size_t); i++) {
+			size_t alignment = alignments[i];
+			size_t align_ceil = ALIGNMENT_CEILING(alignment,
+			    QUANTUM);
+			p = base_alloc(tsdn, base, 1, alignment);
+			assert_ptr_not_null(p,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq(p,
+			    (void *)(ALIGNMENT_CEILING((uintptr_t)p,
+			    alignment)), "Expected quantum alignment");
+			q = base_alloc(tsdn, base, alignment, alignment);
+			assert_ptr_not_null(q,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq((void *)((uintptr_t)p + align_ceil), q,
+			    "Minimal allocation should take up %zu bytes",
+			    align_ceil);
+			r = base_alloc(tsdn, base, 1, alignment);
+			assert_ptr_not_null(r,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq((void *)((uintptr_t)q + align_ceil), r,
+			    "Minimal allocation should take up %zu bytes",
+			    align_ceil);
+		}
+	}
+
+	/*
+	 * Allocate an object that cannot fit in the first block, then verify
+	 * that the first block's remaining space is considered for subsequent
+	 * allocation.
+	 */
+	assert_zu_ge(extent_bsize_get(&base->blocks->extent), QUANTUM,
+	    "Remainder insufficient for test");
+	/* Use up all but one quantum of block. */
+	while (extent_bsize_get(&base->blocks->extent) > QUANTUM) {
+		p = base_alloc(tsdn, base, QUANTUM, QUANTUM);
+		assert_ptr_not_null(p, "Unexpected base_alloc() failure");
+	}
+	r_exp = extent_addr_get(&base->blocks->extent);
+	assert_zu_eq(base->extent_sn_next, 1, "One extant block expected");
+	q = base_alloc(tsdn, base, QUANTUM + 1, QUANTUM);
+	assert_ptr_not_null(q, "Unexpected base_alloc() failure");
+	assert_ptr_ne(q, r_exp, "Expected allocation from new block");
+	assert_zu_eq(base->extent_sn_next, 2, "Two extant blocks expected");
+	r = base_alloc(tsdn, base, QUANTUM, QUANTUM);
+	assert_ptr_not_null(r, "Unexpected base_alloc() failure");
+	assert_ptr_eq(r, r_exp, "Expected allocation from first block");
+	assert_zu_eq(base->extent_sn_next, 2, "Two extant blocks expected");
+
+	/*
+	 * Check for proper alignment support when normal blocks are too small.
+	 */
+	{
+		const size_t alignments[] = {
+			HUGEPAGE,
+			HUGEPAGE << 1
+		};
+		unsigned i;
+
+		for (i = 0; i < sizeof(alignments) / sizeof(size_t); i++) {
+			size_t alignment = alignments[i];
+			p = base_alloc(tsdn, base, QUANTUM, alignment);
+			assert_ptr_not_null(p,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq(p,
+			    (void *)(ALIGNMENT_CEILING((uintptr_t)p,
+			    alignment)), "Expected %zu-byte alignment",
+			    alignment);
+		}
+	}
+
+	called_dalloc = called_destroy = called_decommit = called_purge_lazy =
+	    called_purge_forced = false;
+	base_delete(tsdn, base);
+	assert_true(called_dalloc, "Expected dalloc call");
+	assert_true(!called_destroy, "Unexpected destroy call");
+	assert_true(called_decommit, "Expected decommit call");
+	assert_true(called_purge_lazy, "Expected purge_lazy call");
+	assert_true(called_purge_forced, "Expected purge_forced call");
+
+	try_dalloc = true;
+	try_destroy = true;
+	try_decommit = true;
+	try_purge_lazy = true;
+	try_purge_forced = true;
+	memcpy(&hooks, &hooks_orig, sizeof(extent_hooks_t));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_base_hooks_default,
+	    test_base_hooks_null,
+	    test_base_hooks_not_null);
+}
diff --git a/jemalloc/test/unit/binshard.c b/jemalloc/test/unit/binshard.c
new file mode 100644
index 0000000..d7a8df8
--- /dev/null
+++ b/jemalloc/test/unit/binshard.c
@@ -0,0 +1,154 @@
+#include "test/jemalloc_test.h"
+
+/* Config -- "narenas:1,bin_shards:1-160:16|129-512:4|256-256:8" */
+
+#define NTHREADS 16
+#define REMOTE_NALLOC 256
+
+static void *
+thd_producer(void *varg) {
+	void **mem = varg;
+	unsigned arena, i;
+	size_t sz;
+
+	sz = sizeof(arena);
+	/* Remote arena. */
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	for (i = 0; i < REMOTE_NALLOC / 2; i++) {
+		mem[i] = mallocx(1, MALLOCX_TCACHE_NONE | MALLOCX_ARENA(arena));
+	}
+
+	/* Remote bin. */
+	for (; i < REMOTE_NALLOC; i++) {
+		mem[i] = mallocx(1, MALLOCX_TCACHE_NONE | MALLOCX_ARENA(0));
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_producer_consumer) {
+	thd_t thds[NTHREADS];
+	void *mem[NTHREADS][REMOTE_NALLOC];
+	unsigned i;
+
+	/* Create producer threads to allocate. */
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_producer, mem[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+	/* Remote deallocation by the current thread. */
+	for (i = 0; i < NTHREADS; i++) {
+		for (unsigned j = 0; j < REMOTE_NALLOC; j++) {
+			assert_ptr_not_null(mem[i][j],
+			    "Unexpected remote allocation failure");
+			dallocx(mem[i][j], 0);
+		}
+	}
+}
+TEST_END
+
+static void *
+thd_start(void *varg) {
+	void *ptr, *ptr2;
+	extent_t *extent;
+	unsigned shard1, shard2;
+
+	tsdn_t *tsdn = tsdn_fetch();
+	/* Try triggering allocations from sharded bins. */
+	for (unsigned i = 0; i < 1024; i++) {
+		ptr = mallocx(1, MALLOCX_TCACHE_NONE);
+		ptr2 = mallocx(129, MALLOCX_TCACHE_NONE);
+
+		extent = iealloc(tsdn, ptr);
+		shard1 = extent_binshard_get(extent);
+		dallocx(ptr, 0);
+		assert_u_lt(shard1, 16, "Unexpected bin shard used");
+
+		extent = iealloc(tsdn, ptr2);
+		shard2 = extent_binshard_get(extent);
+		dallocx(ptr2, 0);
+		assert_u_lt(shard2, 4, "Unexpected bin shard used");
+
+		if (shard1 > 0 || shard2 > 0) {
+			/* Triggered sharded bin usage. */
+			return (void *)(uintptr_t)shard1;
+		}
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_bin_shard_mt) {
+	test_skip_if(have_percpu_arena &&
+	    PERCPU_ARENA_ENABLED(opt_percpu_arena));
+
+	thd_t thds[NTHREADS];
+	unsigned i;
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start, NULL);
+	}
+	bool sharded = false;
+	for (i = 0; i < NTHREADS; i++) {
+		void *ret;
+		thd_join(thds[i], &ret);
+		if (ret != NULL) {
+			sharded = true;
+		}
+	}
+	assert_b_eq(sharded, true, "Did not find sharded bins");
+}
+TEST_END
+
+TEST_BEGIN(test_bin_shard) {
+	unsigned nbins, i;
+	size_t mib[4], mib2[4];
+	size_t miblen, miblen2, len;
+
+	len = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &len, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	miblen = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.nshards", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	miblen2 = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.size", mib2, &miblen2), 0,
+	    "Unexpected mallctlnametomib() failure");
+
+	for (i = 0; i < nbins; i++) {
+		uint32_t nshards;
+		size_t size, sz1, sz2;
+
+		mib[2] = i;
+		sz1 = sizeof(nshards);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&nshards, &sz1,
+		    NULL, 0), 0, "Unexpected mallctlbymib() failure");
+
+		mib2[2] = i;
+		sz2 = sizeof(size);
+		assert_d_eq(mallctlbymib(mib2, miblen2, (void *)&size, &sz2,
+		    NULL, 0), 0, "Unexpected mallctlbymib() failure");
+
+		if (size >= 1 && size <= 128) {
+			assert_u_eq(nshards, 16, "Unexpected nshards");
+		} else if (size == 256) {
+			assert_u_eq(nshards, 8, "Unexpected nshards");
+		} else if (size > 128 && size <= 512) {
+			assert_u_eq(nshards, 4, "Unexpected nshards");
+		} else {
+			assert_u_eq(nshards, 1, "Unexpected nshards");
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_bin_shard,
+	    test_bin_shard_mt,
+	    test_producer_consumer);
+}
diff --git a/jemalloc/test/unit/binshard.sh b/jemalloc/test/unit/binshard.sh
new file mode 100644
index 0000000..c1d58c8
--- /dev/null
+++ b/jemalloc/test/unit/binshard.sh
@@ -0,0 +1,3 @@
+#!/bin/sh
+
+export MALLOC_CONF="narenas:1,bin_shards:1-160:16|129-512:4|256-256:8"
diff --git a/jemalloc/test/unit/bit_util.c b/jemalloc/test/unit/bit_util.c
new file mode 100644
index 0000000..b747deb
--- /dev/null
+++ b/jemalloc/test/unit/bit_util.c
@@ -0,0 +1,111 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/bit_util.h"
+
+#define TEST_POW2_CEIL(t, suf, pri) do {				\
+	unsigned i, pow2;						\
+	t x;								\
+									\
+	assert_##suf##_eq(pow2_ceil_##suf(0), 0, "Unexpected result");	\
+									\
+	for (i = 0; i < sizeof(t) * 8; i++) {				\
+		assert_##suf##_eq(pow2_ceil_##suf(((t)1) << i), ((t)1)	\
+		    << i, "Unexpected result");				\
+	}								\
+									\
+	for (i = 2; i < sizeof(t) * 8; i++) {				\
+		assert_##suf##_eq(pow2_ceil_##suf((((t)1) << i) - 1),	\
+		    ((t)1) << i, "Unexpected result");			\
+	}								\
+									\
+	for (i = 0; i < sizeof(t) * 8 - 1; i++) {			\
+		assert_##suf##_eq(pow2_ceil_##suf((((t)1) << i) + 1),	\
+		    ((t)1) << (i+1), "Unexpected result");		\
+	}								\
+									\
+	for (pow2 = 1; pow2 < 25; pow2++) {				\
+		for (x = (((t)1) << (pow2-1)) + 1; x <= ((t)1) << pow2;	\
+		    x++) {						\
+			assert_##suf##_eq(pow2_ceil_##suf(x),		\
+			    ((t)1) << pow2,				\
+			    "Unexpected result, x=%"pri, x);		\
+		}							\
+	}								\
+} while (0)
+
+TEST_BEGIN(test_pow2_ceil_u64) {
+	TEST_POW2_CEIL(uint64_t, u64, FMTu64);
+}
+TEST_END
+
+TEST_BEGIN(test_pow2_ceil_u32) {
+	TEST_POW2_CEIL(uint32_t, u32, FMTu32);
+}
+TEST_END
+
+TEST_BEGIN(test_pow2_ceil_zu) {
+	TEST_POW2_CEIL(size_t, zu, "zu");
+}
+TEST_END
+
+void
+assert_lg_ceil_range(size_t input, unsigned answer) {
+	if (input == 1) {
+		assert_u_eq(0, answer, "Got %u as lg_ceil of 1", answer);
+		return;
+	}
+	assert_zu_le(input, (ZU(1) << answer),
+	    "Got %u as lg_ceil of %zu", answer, input);
+	assert_zu_gt(input, (ZU(1) << (answer - 1)),
+	    "Got %u as lg_ceil of %zu", answer, input);
+}
+
+void
+assert_lg_floor_range(size_t input, unsigned answer) {
+	if (input == 1) {
+		assert_u_eq(0, answer, "Got %u as lg_floor of 1", answer);
+		return;
+	}
+	assert_zu_ge(input, (ZU(1) << answer),
+	    "Got %u as lg_floor of %zu", answer, input);
+	assert_zu_lt(input, (ZU(1) << (answer + 1)),
+	    "Got %u as lg_floor of %zu", answer, input);
+}
+
+TEST_BEGIN(test_lg_ceil_floor) {
+	for (size_t i = 1; i < 10 * 1000 * 1000; i++) {
+		assert_lg_ceil_range(i, lg_ceil(i));
+		assert_lg_ceil_range(i, LG_CEIL(i));
+		assert_lg_floor_range(i, lg_floor(i));
+		assert_lg_floor_range(i, LG_FLOOR(i));
+	}
+	for (int i = 10; i < 8 * (1 << LG_SIZEOF_PTR) - 5; i++) {
+		for (size_t j = 0; j < (1 << 4); j++) {
+			size_t num1 = ((size_t)1 << i)
+			    - j * ((size_t)1 << (i - 4));
+			size_t num2 = ((size_t)1 << i)
+			    + j * ((size_t)1 << (i - 4));
+			assert_zu_ne(num1, 0, "Invalid lg argument");
+			assert_zu_ne(num2, 0, "Invalid lg argument");
+			assert_lg_ceil_range(num1, lg_ceil(num1));
+			assert_lg_ceil_range(num1, LG_CEIL(num1));
+			assert_lg_ceil_range(num2, lg_ceil(num2));
+			assert_lg_ceil_range(num2, LG_CEIL(num2));
+
+			assert_lg_floor_range(num1, lg_floor(num1));
+			assert_lg_floor_range(num1, LG_FLOOR(num1));
+			assert_lg_floor_range(num2, lg_floor(num2));
+			assert_lg_floor_range(num2, LG_FLOOR(num2));
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_pow2_ceil_u64,
+	    test_pow2_ceil_u32,
+	    test_pow2_ceil_zu,
+	    test_lg_ceil_floor);
+}
diff --git a/jemalloc/test/unit/bitmap.c b/jemalloc/test/unit/bitmap.c
new file mode 100644
index 0000000..cafb203
--- /dev/null
+++ b/jemalloc/test/unit/bitmap.c
@@ -0,0 +1,431 @@
+#include "test/jemalloc_test.h"
+
+#define NBITS_TAB \
+    NB( 1) \
+    NB( 2) \
+    NB( 3) \
+    NB( 4) \
+    NB( 5) \
+    NB( 6) \
+    NB( 7) \
+    NB( 8) \
+    NB( 9) \
+    NB(10) \
+    NB(11) \
+    NB(12) \
+    NB(13) \
+    NB(14) \
+    NB(15) \
+    NB(16) \
+    NB(17) \
+    NB(18) \
+    NB(19) \
+    NB(20) \
+    NB(21) \
+    NB(22) \
+    NB(23) \
+    NB(24) \
+    NB(25) \
+    NB(26) \
+    NB(27) \
+    NB(28) \
+    NB(29) \
+    NB(30) \
+    NB(31) \
+    NB(32) \
+    \
+    NB(33) \
+    NB(34) \
+    NB(35) \
+    NB(36) \
+    NB(37) \
+    NB(38) \
+    NB(39) \
+    NB(40) \
+    NB(41) \
+    NB(42) \
+    NB(43) \
+    NB(44) \
+    NB(45) \
+    NB(46) \
+    NB(47) \
+    NB(48) \
+    NB(49) \
+    NB(50) \
+    NB(51) \
+    NB(52) \
+    NB(53) \
+    NB(54) \
+    NB(55) \
+    NB(56) \
+    NB(57) \
+    NB(58) \
+    NB(59) \
+    NB(60) \
+    NB(61) \
+    NB(62) \
+    NB(63) \
+    NB(64) \
+    NB(65) \
+    \
+    NB(126) \
+    NB(127) \
+    NB(128) \
+    NB(129) \
+    NB(130) \
+    \
+    NB(254) \
+    NB(255) \
+    NB(256) \
+    NB(257) \
+    NB(258) \
+    \
+    NB(510) \
+    NB(511) \
+    NB(512) \
+    NB(513) \
+    NB(514) \
+    \
+    NB(1024) \
+    NB(2048) \
+    NB(4096) \
+    NB(8192) \
+    NB(16384) \
+
+static void
+test_bitmap_initializer_body(const bitmap_info_t *binfo, size_t nbits) {
+	bitmap_info_t binfo_dyn;
+	bitmap_info_init(&binfo_dyn, nbits);
+
+	assert_zu_eq(bitmap_size(binfo), bitmap_size(&binfo_dyn),
+	    "Unexpected difference between static and dynamic initialization, "
+	    "nbits=%zu", nbits);
+	assert_zu_eq(binfo->nbits, binfo_dyn.nbits,
+	    "Unexpected difference between static and dynamic initialization, "
+	    "nbits=%zu", nbits);
+#ifdef BITMAP_USE_TREE
+	assert_u_eq(binfo->nlevels, binfo_dyn.nlevels,
+	    "Unexpected difference between static and dynamic initialization, "
+	    "nbits=%zu", nbits);
+	{
+		unsigned i;
+
+		for (i = 0; i < binfo->nlevels; i++) {
+			assert_zu_eq(binfo->levels[i].group_offset,
+			    binfo_dyn.levels[i].group_offset,
+			    "Unexpected difference between static and dynamic "
+			    "initialization, nbits=%zu, level=%u", nbits, i);
+		}
+	}
+#else
+	assert_zu_eq(binfo->ngroups, binfo_dyn.ngroups,
+	    "Unexpected difference between static and dynamic initialization");
+#endif
+}
+
+TEST_BEGIN(test_bitmap_initializer) {
+#define NB(nbits) {							\
+		if (nbits <= BITMAP_MAXBITS) {				\
+			bitmap_info_t binfo =				\
+			    BITMAP_INFO_INITIALIZER(nbits);		\
+			test_bitmap_initializer_body(&binfo, nbits);	\
+		}							\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static size_t
+test_bitmap_size_body(const bitmap_info_t *binfo, size_t nbits,
+    size_t prev_size) {
+	size_t size = bitmap_size(binfo);
+	assert_zu_ge(size, (nbits >> 3),
+	    "Bitmap size is smaller than expected");
+	assert_zu_ge(size, prev_size, "Bitmap size is smaller than expected");
+	return size;
+}
+
+TEST_BEGIN(test_bitmap_size) {
+	size_t nbits, prev_size;
+
+	prev_size = 0;
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		prev_size = test_bitmap_size_body(&binfo, nbits, prev_size);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		prev_size = test_bitmap_size_body(&binfo, nbits,	\
+		    prev_size);						\
+	}
+	prev_size = 0;
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_init_body(const bitmap_info_t *binfo, size_t nbits) {
+	size_t i;
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+
+	bitmap_init(bitmap, binfo, false);
+	for (i = 0; i < nbits; i++) {
+		assert_false(bitmap_get(bitmap, binfo, i),
+		    "Bit should be unset");
+	}
+
+	bitmap_init(bitmap, binfo, true);
+	for (i = 0; i < nbits; i++) {
+		assert_true(bitmap_get(bitmap, binfo, i), "Bit should be set");
+	}
+
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_init) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_init_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_init_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_set_body(const bitmap_info_t *binfo, size_t nbits) {
+	size_t i;
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+	bitmap_init(bitmap, binfo, false);
+
+	for (i = 0; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i);
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_set) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_set_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_set_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_unset_body(const bitmap_info_t *binfo, size_t nbits) {
+	size_t i;
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+	bitmap_init(bitmap, binfo, false);
+
+	for (i = 0; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i);
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+	for (i = 0; i < nbits; i++) {
+		bitmap_unset(bitmap, binfo, i);
+	}
+	for (i = 0; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i);
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_unset) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_unset_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_unset_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_xfu_body(const bitmap_info_t *binfo, size_t nbits) {
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+	bitmap_init(bitmap, binfo, false);
+
+	/* Iteratively set bits starting at the beginning. */
+	for (size_t i = 0; i < nbits; i++) {
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, (i > 0) ? i-1 : i), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+
+	/*
+	 * Iteratively unset bits starting at the end, and verify that
+	 * bitmap_sfu() reaches the unset bits.
+	 */
+	for (size_t i = nbits - 1; i < nbits; i--) { /* (nbits..0] */
+		bitmap_unset(bitmap, binfo, i);
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), i,
+		    "First unset bit should the bit previously unset");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, (i > 0) ? i-1 : i), i,
+		    "First unset bit should the bit previously unset");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+		    "First unset bit should the bit previously unset");
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+		    "First unset bit should the bit previously unset");
+		bitmap_unset(bitmap, binfo, i);
+	}
+	assert_false(bitmap_get(bitmap, binfo, 0), "Bit should be unset");
+
+	/*
+	 * Iteratively set bits starting at the beginning, and verify that
+	 * bitmap_sfu() looks past them.
+	 */
+	for (size_t i = 1; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i - 1);
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, (i > 0) ? i-1 : i), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		bitmap_unset(bitmap, binfo, i);
+	}
+	assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), nbits - 1,
+	    "First unset bit should be the last bit");
+	assert_zu_eq(bitmap_ffu(bitmap, binfo, (nbits > 1) ? nbits-2 : nbits-1),
+	    nbits - 1, "First unset bit should be the last bit");
+	assert_zu_eq(bitmap_ffu(bitmap, binfo, nbits - 1), nbits - 1,
+	    "First unset bit should be the last bit");
+	assert_zu_eq(bitmap_sfu(bitmap, binfo), nbits - 1,
+	    "First unset bit should be the last bit");
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+
+	/*
+	 * Bubble a "usu" pattern through the bitmap and verify that
+	 * bitmap_ffu() finds the correct bit for all five min_bit cases.
+	 */
+	if (nbits >= 3) {
+		for (size_t i = 0; i < nbits-2; i++) {
+			bitmap_unset(bitmap, binfo, i);
+			bitmap_unset(bitmap, binfo, i+2);
+			if (i > 0) {
+				assert_zu_eq(bitmap_ffu(bitmap, binfo, i-1), i,
+				    "Unexpected first unset bit");
+			}
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i+1), i+2,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i+2), i+2,
+			    "Unexpected first unset bit");
+			if (i + 3 < nbits) {
+				assert_zu_eq(bitmap_ffu(bitmap, binfo, i+3),
+				    nbits, "Unexpected first unset bit");
+			}
+			assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_sfu(bitmap, binfo), i+2,
+			    "Unexpected first unset bit");
+		}
+	}
+
+	/*
+	 * Unset the last bit, bubble another unset bit through the bitmap, and
+	 * verify that bitmap_ffu() finds the correct bit for all four min_bit
+	 * cases.
+	 */
+	if (nbits >= 3) {
+		bitmap_unset(bitmap, binfo, nbits-1);
+		for (size_t i = 0; i < nbits-1; i++) {
+			bitmap_unset(bitmap, binfo, i);
+			if (i > 0) {
+				assert_zu_eq(bitmap_ffu(bitmap, binfo, i-1), i,
+				    "Unexpected first unset bit");
+			}
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i+1), nbits-1,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, nbits-1),
+			    nbits-1, "Unexpected first unset bit");
+
+			assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+			    "Unexpected first unset bit");
+		}
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), nbits-1,
+		    "Unexpected first unset bit");
+	}
+
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_xfu) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_xfu_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_xfu_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_bitmap_initializer,
+	    test_bitmap_size,
+	    test_bitmap_init,
+	    test_bitmap_set,
+	    test_bitmap_unset,
+	    test_bitmap_xfu);
+}
diff --git a/jemalloc/test/unit/ckh.c b/jemalloc/test/unit/ckh.c
new file mode 100644
index 0000000..707ea5f
--- /dev/null
+++ b/jemalloc/test/unit/ckh.c
@@ -0,0 +1,211 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_new_delete) {
+	tsd_t *tsd;
+	ckh_t ckh;
+
+	tsd = tsd_fetch();
+
+	assert_false(ckh_new(tsd, &ckh, 2, ckh_string_hash,
+	    ckh_string_keycomp), "Unexpected ckh_new() error");
+	ckh_delete(tsd, &ckh);
+
+	assert_false(ckh_new(tsd, &ckh, 3, ckh_pointer_hash,
+	    ckh_pointer_keycomp), "Unexpected ckh_new() error");
+	ckh_delete(tsd, &ckh);
+}
+TEST_END
+
+TEST_BEGIN(test_count_insert_search_remove) {
+	tsd_t *tsd;
+	ckh_t ckh;
+	const char *strs[] = {
+	    "a string",
+	    "A string",
+	    "a string.",
+	    "A string."
+	};
+	const char *missing = "A string not in the hash table.";
+	size_t i;
+
+	tsd = tsd_fetch();
+
+	assert_false(ckh_new(tsd, &ckh, 2, ckh_string_hash,
+	    ckh_string_keycomp), "Unexpected ckh_new() error");
+	assert_zu_eq(ckh_count(&ckh), 0,
+	    "ckh_count() should return %zu, but it returned %zu", ZU(0),
+	    ckh_count(&ckh));
+
+	/* Insert. */
+	for (i = 0; i < sizeof(strs)/sizeof(const char *); i++) {
+		ckh_insert(tsd, &ckh, strs[i], strs[i]);
+		assert_zu_eq(ckh_count(&ckh), i+1,
+		    "ckh_count() should return %zu, but it returned %zu", i+1,
+		    ckh_count(&ckh));
+	}
+
+	/* Search. */
+	for (i = 0; i < sizeof(strs)/sizeof(const char *); i++) {
+		union {
+			void *p;
+			const char *s;
+		} k, v;
+		void **kp, **vp;
+		const char *ks, *vs;
+
+		kp = (i & 1) ? &k.p : NULL;
+		vp = (i & 2) ? &v.p : NULL;
+		k.p = NULL;
+		v.p = NULL;
+		assert_false(ckh_search(&ckh, strs[i], kp, vp),
+		    "Unexpected ckh_search() error");
+
+		ks = (i & 1) ? strs[i] : (const char *)NULL;
+		vs = (i & 2) ? strs[i] : (const char *)NULL;
+		assert_ptr_eq((void *)ks, (void *)k.s, "Key mismatch, i=%zu",
+		    i);
+		assert_ptr_eq((void *)vs, (void *)v.s, "Value mismatch, i=%zu",
+		    i);
+	}
+	assert_true(ckh_search(&ckh, missing, NULL, NULL),
+	    "Unexpected ckh_search() success");
+
+	/* Remove. */
+	for (i = 0; i < sizeof(strs)/sizeof(const char *); i++) {
+		union {
+			void *p;
+			const char *s;
+		} k, v;
+		void **kp, **vp;
+		const char *ks, *vs;
+
+		kp = (i & 1) ? &k.p : NULL;
+		vp = (i & 2) ? &v.p : NULL;
+		k.p = NULL;
+		v.p = NULL;
+		assert_false(ckh_remove(tsd, &ckh, strs[i], kp, vp),
+		    "Unexpected ckh_remove() error");
+
+		ks = (i & 1) ? strs[i] : (const char *)NULL;
+		vs = (i & 2) ? strs[i] : (const char *)NULL;
+		assert_ptr_eq((void *)ks, (void *)k.s, "Key mismatch, i=%zu",
+		    i);
+		assert_ptr_eq((void *)vs, (void *)v.s, "Value mismatch, i=%zu",
+		    i);
+		assert_zu_eq(ckh_count(&ckh),
+		    sizeof(strs)/sizeof(const char *) - i - 1,
+		    "ckh_count() should return %zu, but it returned %zu",
+		        sizeof(strs)/sizeof(const char *) - i - 1,
+		    ckh_count(&ckh));
+	}
+
+	ckh_delete(tsd, &ckh);
+}
+TEST_END
+
+TEST_BEGIN(test_insert_iter_remove) {
+#define NITEMS ZU(1000)
+	tsd_t *tsd;
+	ckh_t ckh;
+	void **p[NITEMS];
+	void *q, *r;
+	size_t i;
+
+	tsd = tsd_fetch();
+
+	assert_false(ckh_new(tsd, &ckh, 2, ckh_pointer_hash,
+	    ckh_pointer_keycomp), "Unexpected ckh_new() error");
+
+	for (i = 0; i < NITEMS; i++) {
+		p[i] = mallocx(i+1, 0);
+		assert_ptr_not_null(p[i], "Unexpected mallocx() failure");
+	}
+
+	for (i = 0; i < NITEMS; i++) {
+		size_t j;
+
+		for (j = i; j < NITEMS; j++) {
+			assert_false(ckh_insert(tsd, &ckh, p[j], p[j]),
+			    "Unexpected ckh_insert() failure");
+			assert_false(ckh_search(&ckh, p[j], &q, &r),
+			    "Unexpected ckh_search() failure");
+			assert_ptr_eq(p[j], q, "Key pointer mismatch");
+			assert_ptr_eq(p[j], r, "Value pointer mismatch");
+		}
+
+		assert_zu_eq(ckh_count(&ckh), NITEMS,
+		    "ckh_count() should return %zu, but it returned %zu",
+		    NITEMS, ckh_count(&ckh));
+
+		for (j = i + 1; j < NITEMS; j++) {
+			assert_false(ckh_search(&ckh, p[j], NULL, NULL),
+			    "Unexpected ckh_search() failure");
+			assert_false(ckh_remove(tsd, &ckh, p[j], &q, &r),
+			    "Unexpected ckh_remove() failure");
+			assert_ptr_eq(p[j], q, "Key pointer mismatch");
+			assert_ptr_eq(p[j], r, "Value pointer mismatch");
+			assert_true(ckh_search(&ckh, p[j], NULL, NULL),
+			    "Unexpected ckh_search() success");
+			assert_true(ckh_remove(tsd, &ckh, p[j], &q, &r),
+			    "Unexpected ckh_remove() success");
+		}
+
+		{
+			bool seen[NITEMS];
+			size_t tabind;
+
+			memset(seen, 0, sizeof(seen));
+
+			for (tabind = 0; !ckh_iter(&ckh, &tabind, &q, &r);) {
+				size_t k;
+
+				assert_ptr_eq(q, r, "Key and val not equal");
+
+				for (k = 0; k < NITEMS; k++) {
+					if (p[k] == q) {
+						assert_false(seen[k],
+						    "Item %zu already seen", k);
+						seen[k] = true;
+						break;
+					}
+				}
+			}
+
+			for (j = 0; j < i + 1; j++) {
+				assert_true(seen[j], "Item %zu not seen", j);
+			}
+			for (; j < NITEMS; j++) {
+				assert_false(seen[j], "Item %zu seen", j);
+			}
+		}
+	}
+
+	for (i = 0; i < NITEMS; i++) {
+		assert_false(ckh_search(&ckh, p[i], NULL, NULL),
+		    "Unexpected ckh_search() failure");
+		assert_false(ckh_remove(tsd, &ckh, p[i], &q, &r),
+		    "Unexpected ckh_remove() failure");
+		assert_ptr_eq(p[i], q, "Key pointer mismatch");
+		assert_ptr_eq(p[i], r, "Value pointer mismatch");
+		assert_true(ckh_search(&ckh, p[i], NULL, NULL),
+		    "Unexpected ckh_search() success");
+		assert_true(ckh_remove(tsd, &ckh, p[i], &q, &r),
+		    "Unexpected ckh_remove() success");
+		dallocx(p[i], 0);
+	}
+
+	assert_zu_eq(ckh_count(&ckh), 0,
+	    "ckh_count() should return %zu, but it returned %zu",
+	    ZU(0), ckh_count(&ckh));
+	ckh_delete(tsd, &ckh);
+#undef NITEMS
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_new_delete,
+	    test_count_insert_search_remove,
+	    test_insert_iter_remove);
+}
diff --git a/jemalloc/test/unit/decay.c b/jemalloc/test/unit/decay.c
new file mode 100644
index 0000000..cf3c079
--- /dev/null
+++ b/jemalloc/test/unit/decay.c
@@ -0,0 +1,605 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ticker.h"
+
+static nstime_monotonic_t *nstime_monotonic_orig;
+static nstime_update_t *nstime_update_orig;
+
+static unsigned nupdates_mock;
+static nstime_t time_mock;
+static bool monotonic_mock;
+
+static bool
+check_background_thread_enabled(void) {
+	bool enabled;
+	size_t sz = sizeof(bool);
+	int ret = mallctl("background_thread", (void *)&enabled, &sz, NULL,0);
+	if (ret == ENOENT) {
+		return false;
+	}
+	assert_d_eq(ret, 0, "Unexpected mallctl error");
+	return enabled;
+}
+
+static bool
+nstime_monotonic_mock(void) {
+	return monotonic_mock;
+}
+
+static bool
+nstime_update_mock(nstime_t *time) {
+	nupdates_mock++;
+	if (monotonic_mock) {
+		nstime_copy(time, &time_mock);
+	}
+	return !monotonic_mock;
+}
+
+static unsigned
+do_arena_create(ssize_t dirty_decay_ms, ssize_t muzzy_decay_ms) {
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib("arena.0.dirty_decay_ms", mib, &miblen),
+	    0, "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(dirty_decay_ms)), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	assert_d_eq(mallctlnametomib("arena.0.muzzy_decay_ms", mib, &miblen),
+	    0, "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(muzzy_decay_ms)), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	return arena_ind;
+}
+
+static void
+do_arena_destroy(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.destroy", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+void
+do_epoch(void) {
+	uint64_t epoch = 1;
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+}
+
+void
+do_purge(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.purge", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+void
+do_decay(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.decay", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+static uint64_t
+get_arena_npurge_impl(const char *mibname, unsigned arena_ind) {
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib(mibname, mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[2] = (size_t)arena_ind;
+	uint64_t npurge = 0;
+	size_t sz = sizeof(npurge);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&npurge, &sz, NULL, 0),
+	    config_stats ? 0 : ENOENT, "Unexpected mallctlbymib() failure");
+	return npurge;
+}
+
+static uint64_t
+get_arena_dirty_npurge(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.dirty_npurge", arena_ind);
+}
+
+static uint64_t
+get_arena_dirty_purged(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.dirty_purged", arena_ind);
+}
+
+static uint64_t
+get_arena_muzzy_npurge(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.muzzy_npurge", arena_ind);
+}
+
+static uint64_t
+get_arena_npurge(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.dirty_npurge", arena_ind) +
+	    get_arena_npurge_impl("stats.arenas.0.muzzy_npurge", arena_ind);
+}
+
+static size_t
+get_arena_pdirty(unsigned arena_ind) {
+	do_epoch();
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("stats.arenas.0.pdirty", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[2] = (size_t)arena_ind;
+	size_t pdirty;
+	size_t sz = sizeof(pdirty);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&pdirty, &sz, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+	return pdirty;
+}
+
+static size_t
+get_arena_pmuzzy(unsigned arena_ind) {
+	do_epoch();
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("stats.arenas.0.pmuzzy", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[2] = (size_t)arena_ind;
+	size_t pmuzzy;
+	size_t sz = sizeof(pmuzzy);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&pmuzzy, &sz, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+	return pmuzzy;
+}
+
+static void *
+do_mallocx(size_t size, int flags) {
+	void *p = mallocx(size, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	return p;
+}
+
+static void
+generate_dirty(unsigned arena_ind, size_t size) {
+	int flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+	void *p = do_mallocx(size, flags);
+	dallocx(p, flags);
+}
+
+TEST_BEGIN(test_decay_ticks) {
+	test_skip_if(check_background_thread_enabled());
+
+	ticker_t *decay_ticker;
+	unsigned tick0, tick1, arena_ind;
+	size_t sz, large0;
+	void *p;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large0, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	/* Set up a manually managed arena for test. */
+	arena_ind = do_arena_create(0, 0);
+
+	/* Migrate to the new arena, and get the ticker. */
+	unsigned old_arena_ind;
+	size_t sz_arena_ind = sizeof(old_arena_ind);
+	assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind,
+	    &sz_arena_ind, (void *)&arena_ind, sizeof(arena_ind)), 0,
+	    "Unexpected mallctl() failure");
+	decay_ticker = decay_ticker_get(tsd_fetch(), arena_ind);
+	assert_ptr_not_null(decay_ticker,
+	    "Unexpected failure getting decay ticker");
+
+	/*
+	 * Test the standard APIs using a large size class, since we can't
+	 * control tcache interactions for small size classes (except by
+	 * completely disabling tcache for the entire test program).
+	 */
+
+	/* malloc(). */
+	tick0 = ticker_read(decay_ticker);
+	p = malloc(large0);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during malloc()");
+	/* free(). */
+	tick0 = ticker_read(decay_ticker);
+	free(p);
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during free()");
+
+	/* calloc(). */
+	tick0 = ticker_read(decay_ticker);
+	p = calloc(1, large0);
+	assert_ptr_not_null(p, "Unexpected calloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during calloc()");
+	free(p);
+
+	/* posix_memalign(). */
+	tick0 = ticker_read(decay_ticker);
+	assert_d_eq(posix_memalign(&p, sizeof(size_t), large0), 0,
+	    "Unexpected posix_memalign() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0,
+	    "Expected ticker to tick during posix_memalign()");
+	free(p);
+
+	/* aligned_alloc(). */
+	tick0 = ticker_read(decay_ticker);
+	p = aligned_alloc(sizeof(size_t), large0);
+	assert_ptr_not_null(p, "Unexpected aligned_alloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0,
+	    "Expected ticker to tick during aligned_alloc()");
+	free(p);
+
+	/* realloc(). */
+	/* Allocate. */
+	tick0 = ticker_read(decay_ticker);
+	p = realloc(NULL, large0);
+	assert_ptr_not_null(p, "Unexpected realloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during realloc()");
+	/* Reallocate. */
+	tick0 = ticker_read(decay_ticker);
+	p = realloc(p, large0);
+	assert_ptr_not_null(p, "Unexpected realloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during realloc()");
+	/* Deallocate. */
+	tick0 = ticker_read(decay_ticker);
+	realloc(p, 0);
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during realloc()");
+
+	/*
+	 * Test the *allocx() APIs using large and small size classes, with
+	 * tcache explicitly disabled.
+	 */
+	{
+		unsigned i;
+		size_t allocx_sizes[2];
+		allocx_sizes[0] = large0;
+		allocx_sizes[1] = 1;
+
+		for (i = 0; i < sizeof(allocx_sizes) / sizeof(size_t); i++) {
+			sz = allocx_sizes[i];
+
+			/* mallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			p = mallocx(sz, MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p, "Unexpected mallocx() failure");
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during mallocx() (sz=%zu)",
+			    sz);
+			/* rallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			p = rallocx(p, sz, MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p, "Unexpected rallocx() failure");
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during rallocx() (sz=%zu)",
+			    sz);
+			/* xallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			xallocx(p, sz, 0, MALLOCX_TCACHE_NONE);
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during xallocx() (sz=%zu)",
+			    sz);
+			/* dallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			dallocx(p, MALLOCX_TCACHE_NONE);
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during dallocx() (sz=%zu)",
+			    sz);
+			/* sdallocx(). */
+			p = mallocx(sz, MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p, "Unexpected mallocx() failure");
+			tick0 = ticker_read(decay_ticker);
+			sdallocx(p, sz, MALLOCX_TCACHE_NONE);
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during sdallocx() "
+			    "(sz=%zu)", sz);
+		}
+	}
+
+	/*
+	 * Test tcache fill/flush interactions for large and small size classes,
+	 * using an explicit tcache.
+	 */
+	unsigned tcache_ind, i;
+	size_t tcache_sizes[2];
+	tcache_sizes[0] = large0;
+	tcache_sizes[1] = 1;
+
+	size_t tcache_max, sz_tcache_max;
+	sz_tcache_max = sizeof(tcache_max);
+	assert_d_eq(mallctl("arenas.tcache_max", (void *)&tcache_max,
+	    &sz_tcache_max, NULL, 0), 0, "Unexpected mallctl() failure");
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("tcache.create", (void *)&tcache_ind, &sz,
+	    NULL, 0), 0, "Unexpected mallctl failure");
+
+	for (i = 0; i < sizeof(tcache_sizes) / sizeof(size_t); i++) {
+		sz = tcache_sizes[i];
+
+		/* tcache fill. */
+		tick0 = ticker_read(decay_ticker);
+		p = mallocx(sz, MALLOCX_TCACHE(tcache_ind));
+		assert_ptr_not_null(p, "Unexpected mallocx() failure");
+		tick1 = ticker_read(decay_ticker);
+		assert_u32_ne(tick1, tick0,
+		    "Expected ticker to tick during tcache fill "
+		    "(sz=%zu)", sz);
+		/* tcache flush. */
+		dallocx(p, MALLOCX_TCACHE(tcache_ind));
+		tick0 = ticker_read(decay_ticker);
+		assert_d_eq(mallctl("tcache.flush", NULL, NULL,
+		    (void *)&tcache_ind, sizeof(unsigned)), 0,
+		    "Unexpected mallctl failure");
+		tick1 = ticker_read(decay_ticker);
+
+		/* Will only tick if it's in tcache. */
+		if (sz <= tcache_max) {
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during tcache "
+			    "flush (sz=%zu)", sz);
+		} else {
+			assert_u32_eq(tick1, tick0,
+			    "Unexpected ticker tick during tcache "
+			    "flush (sz=%zu)", sz);
+		}
+	}
+}
+TEST_END
+
+static void
+decay_ticker_helper(unsigned arena_ind, int flags, bool dirty, ssize_t dt,
+    uint64_t dirty_npurge0, uint64_t muzzy_npurge0, bool terminate_asap) {
+#define NINTERVALS 101
+	nstime_t time, update_interval, decay_ms, deadline;
+
+	nstime_init(&time, 0);
+	nstime_update(&time);
+
+	nstime_init2(&decay_ms, dt, 0);
+	nstime_copy(&deadline, &time);
+	nstime_add(&deadline, &decay_ms);
+
+	nstime_init2(&update_interval, dt, 0);
+	nstime_idivide(&update_interval, NINTERVALS);
+
+	/*
+	 * Keep q's slab from being deallocated during the looping below.  If a
+	 * cached slab were to repeatedly come and go during looping, it could
+	 * prevent the decay backlog ever becoming empty.
+	 */
+	void *p = do_mallocx(1, flags);
+	uint64_t dirty_npurge1, muzzy_npurge1;
+	do {
+		for (unsigned i = 0; i < DECAY_NTICKS_PER_UPDATE / 2;
+		    i++) {
+			void *q = do_mallocx(1, flags);
+			dallocx(q, flags);
+		}
+		dirty_npurge1 = get_arena_dirty_npurge(arena_ind);
+		muzzy_npurge1 = get_arena_muzzy_npurge(arena_ind);
+
+		nstime_add(&time_mock, &update_interval);
+		nstime_update(&time);
+	} while (nstime_compare(&time, &deadline) <= 0 && ((dirty_npurge1 ==
+	    dirty_npurge0 && muzzy_npurge1 == muzzy_npurge0) ||
+	    !terminate_asap));
+	dallocx(p, flags);
+
+	if (config_stats) {
+		assert_u64_gt(dirty_npurge1 + muzzy_npurge1, dirty_npurge0 +
+		    muzzy_npurge0, "Expected purging to occur");
+	}
+#undef NINTERVALS
+}
+
+TEST_BEGIN(test_decay_ticker) {
+	test_skip_if(check_background_thread_enabled());
+#define NPS 2048
+	ssize_t ddt = opt_dirty_decay_ms;
+	ssize_t mdt = opt_muzzy_decay_ms;
+	unsigned arena_ind = do_arena_create(ddt, mdt);
+	int flags = (MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE);
+	void *ps[NPS];
+	size_t large;
+
+	/*
+	 * Allocate a bunch of large objects, pause the clock, deallocate every
+	 * other object (to fragment virtual memory), restore the clock, then
+	 * [md]allocx() in a tight loop while advancing time rapidly to verify
+	 * the ticker triggers purging.
+	 */
+
+	size_t tcache_max;
+	size_t sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.tcache_max", (void *)&tcache_max, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+	large = nallocx(tcache_max + 1, flags);
+
+	do_purge(arena_ind);
+	uint64_t dirty_npurge0 = get_arena_dirty_npurge(arena_ind);
+	uint64_t muzzy_npurge0 = get_arena_muzzy_npurge(arena_ind);
+
+	for (unsigned i = 0; i < NPS; i++) {
+		ps[i] = do_mallocx(large, flags);
+	}
+
+	nupdates_mock = 0;
+	nstime_init(&time_mock, 0);
+	nstime_update(&time_mock);
+	monotonic_mock = true;
+
+	nstime_monotonic_orig = nstime_monotonic;
+	nstime_update_orig = nstime_update;
+	nstime_monotonic = nstime_monotonic_mock;
+	nstime_update = nstime_update_mock;
+
+	for (unsigned i = 0; i < NPS; i += 2) {
+		dallocx(ps[i], flags);
+		unsigned nupdates0 = nupdates_mock;
+		do_decay(arena_ind);
+		assert_u_gt(nupdates_mock, nupdates0,
+		    "Expected nstime_update() to be called");
+	}
+
+	decay_ticker_helper(arena_ind, flags, true, ddt, dirty_npurge0,
+	    muzzy_npurge0, true);
+	decay_ticker_helper(arena_ind, flags, false, ddt+mdt, dirty_npurge0,
+	    muzzy_npurge0, false);
+
+	do_arena_destroy(arena_ind);
+
+	nstime_monotonic = nstime_monotonic_orig;
+	nstime_update = nstime_update_orig;
+#undef NPS
+}
+TEST_END
+
+TEST_BEGIN(test_decay_nonmonotonic) {
+	test_skip_if(check_background_thread_enabled());
+#define NPS (SMOOTHSTEP_NSTEPS + 1)
+	int flags = (MALLOCX_ARENA(0) | MALLOCX_TCACHE_NONE);
+	void *ps[NPS];
+	uint64_t npurge0 = 0;
+	uint64_t npurge1 = 0;
+	size_t sz, large0;
+	unsigned i, nupdates0;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large0, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure");
+	do_epoch();
+	sz = sizeof(uint64_t);
+	npurge0 = get_arena_npurge(0);
+
+	nupdates_mock = 0;
+	nstime_init(&time_mock, 0);
+	nstime_update(&time_mock);
+	monotonic_mock = false;
+
+	nstime_monotonic_orig = nstime_monotonic;
+	nstime_update_orig = nstime_update;
+	nstime_monotonic = nstime_monotonic_mock;
+	nstime_update = nstime_update_mock;
+
+	for (i = 0; i < NPS; i++) {
+		ps[i] = mallocx(large0, flags);
+		assert_ptr_not_null(ps[i], "Unexpected mallocx() failure");
+	}
+
+	for (i = 0; i < NPS; i++) {
+		dallocx(ps[i], flags);
+		nupdates0 = nupdates_mock;
+		assert_d_eq(mallctl("arena.0.decay", NULL, NULL, NULL, 0), 0,
+		    "Unexpected arena.0.decay failure");
+		assert_u_gt(nupdates_mock, nupdates0,
+		    "Expected nstime_update() to be called");
+	}
+
+	do_epoch();
+	sz = sizeof(uint64_t);
+	npurge1 = get_arena_npurge(0);
+
+	if (config_stats) {
+		assert_u64_eq(npurge0, npurge1, "Unexpected purging occurred");
+	}
+
+	nstime_monotonic = nstime_monotonic_orig;
+	nstime_update = nstime_update_orig;
+#undef NPS
+}
+TEST_END
+
+TEST_BEGIN(test_decay_now) {
+	test_skip_if(check_background_thread_enabled());
+
+	unsigned arena_ind = do_arena_create(0, 0);
+	assert_zu_eq(get_arena_pdirty(arena_ind), 0, "Unexpected dirty pages");
+	assert_zu_eq(get_arena_pmuzzy(arena_ind), 0, "Unexpected muzzy pages");
+	size_t sizes[] = {16, PAGE<<2, HUGEPAGE<<2};
+	/* Verify that dirty/muzzy pages never linger after deallocation. */
+	for (unsigned i = 0; i < sizeof(sizes)/sizeof(size_t); i++) {
+		size_t size = sizes[i];
+		generate_dirty(arena_ind, size);
+		assert_zu_eq(get_arena_pdirty(arena_ind), 0,
+		    "Unexpected dirty pages");
+		assert_zu_eq(get_arena_pmuzzy(arena_ind), 0,
+		    "Unexpected muzzy pages");
+	}
+	do_arena_destroy(arena_ind);
+}
+TEST_END
+
+TEST_BEGIN(test_decay_never) {
+	test_skip_if(check_background_thread_enabled() || !config_stats);
+
+	unsigned arena_ind = do_arena_create(-1, -1);
+	int flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+	assert_zu_eq(get_arena_pdirty(arena_ind), 0, "Unexpected dirty pages");
+	assert_zu_eq(get_arena_pmuzzy(arena_ind), 0, "Unexpected muzzy pages");
+	size_t sizes[] = {16, PAGE<<2, HUGEPAGE<<2};
+	void *ptrs[sizeof(sizes)/sizeof(size_t)];
+	for (unsigned i = 0; i < sizeof(sizes)/sizeof(size_t); i++) {
+		ptrs[i] = do_mallocx(sizes[i], flags);
+	}
+	/* Verify that each deallocation generates additional dirty pages. */
+	size_t pdirty_prev = get_arena_pdirty(arena_ind);
+	size_t pmuzzy_prev = get_arena_pmuzzy(arena_ind);
+	assert_zu_eq(pdirty_prev, 0, "Unexpected dirty pages");
+	assert_zu_eq(pmuzzy_prev, 0, "Unexpected muzzy pages");
+	for (unsigned i = 0; i < sizeof(sizes)/sizeof(size_t); i++) {
+		dallocx(ptrs[i], flags);
+		size_t pdirty = get_arena_pdirty(arena_ind);
+		size_t pmuzzy = get_arena_pmuzzy(arena_ind);
+		assert_zu_gt(pdirty + (size_t)get_arena_dirty_purged(arena_ind),
+		    pdirty_prev, "Expected dirty pages to increase.");
+		assert_zu_eq(pmuzzy, 0, "Unexpected muzzy pages");
+		pdirty_prev = pdirty;
+	}
+	do_arena_destroy(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_decay_ticks,
+	    test_decay_ticker,
+	    test_decay_nonmonotonic,
+	    test_decay_now,
+	    test_decay_never);
+}
diff --git a/jemalloc/test/unit/decay.sh b/jemalloc/test/unit/decay.sh
new file mode 100644
index 0000000..45aeccf
--- /dev/null
+++ b/jemalloc/test/unit/decay.sh
@@ -0,0 +1,3 @@
+#!/bin/sh
+
+export MALLOC_CONF="dirty_decay_ms:1000,muzzy_decay_ms:1000,lg_tcache_max:0"
diff --git a/jemalloc/test/unit/div.c b/jemalloc/test/unit/div.c
new file mode 100644
index 0000000..b47f10b
--- /dev/null
+++ b/jemalloc/test/unit/div.c
@@ -0,0 +1,29 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/div.h"
+
+TEST_BEGIN(test_div_exhaustive) {
+	for (size_t divisor = 2; divisor < 1000 * 1000; ++divisor) {
+		div_info_t div_info;
+		div_init(&div_info, divisor);
+		size_t max = 1000 * divisor;
+		if (max < 1000 * 1000) {
+			max = 1000 * 1000;
+		}
+		for (size_t dividend = 0; dividend < 1000 * divisor;
+		    dividend += divisor) {
+			size_t quotient = div_compute(
+			    &div_info, dividend);
+			assert_zu_eq(dividend, quotient * divisor,
+			    "With divisor = %zu, dividend = %zu, "
+			    "got quotient %zu", divisor, dividend, quotient);
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_div_exhaustive);
+}
diff --git a/jemalloc/test/unit/emitter.c b/jemalloc/test/unit/emitter.c
new file mode 100644
index 0000000..b4a693f
--- /dev/null
+++ b/jemalloc/test/unit/emitter.c
@@ -0,0 +1,469 @@
+#include "test/jemalloc_test.h"
+#include "jemalloc/internal/emitter.h"
+
+/*
+ * This is so useful for debugging and feature work, we'll leave printing
+ * functionality committed but disabled by default.
+ */
+/* Print the text as it will appear. */
+static bool print_raw = false;
+/* Print the text escaped, so it can be copied back into the test case. */
+static bool print_escaped = false;
+
+typedef struct buf_descriptor_s buf_descriptor_t;
+struct buf_descriptor_s {
+	char *buf;
+	size_t len;
+	bool mid_quote;
+};
+
+/*
+ * Forwards all writes to the passed-in buf_v (which should be cast from a
+ * buf_descriptor_t *).
+ */
+static void
+forwarding_cb(void *buf_descriptor_v, const char *str) {
+	buf_descriptor_t *buf_descriptor = (buf_descriptor_t *)buf_descriptor_v;
+
+	if (print_raw) {
+		malloc_printf("%s", str);
+	}
+	if (print_escaped) {
+		const char *it = str;
+		while (*it != '\0') {
+			if (!buf_descriptor->mid_quote) {
+				malloc_printf("\"");
+				buf_descriptor->mid_quote = true;
+			}
+			switch (*it) {
+			case '\\':
+				malloc_printf("\\");
+				break;
+			case '\"':
+				malloc_printf("\\\"");
+				break;
+			case '\t':
+				malloc_printf("\\t");
+				break;
+			case '\n':
+				malloc_printf("\\n\"\n");
+				buf_descriptor->mid_quote = false;
+				break;
+			default:
+				malloc_printf("%c", *it);
+			}
+			it++;
+		}
+	}
+
+	size_t written = malloc_snprintf(buf_descriptor->buf,
+	    buf_descriptor->len, "%s", str);
+	assert_zu_eq(written, strlen(str), "Buffer overflow!");
+	buf_descriptor->buf += written;
+	buf_descriptor->len -= written;
+	assert_zu_gt(buf_descriptor->len, 0, "Buffer out of space!");
+}
+
+static void
+assert_emit_output(void (*emit_fn)(emitter_t *),
+    const char *expected_json_output, const char *expected_table_output) {
+	emitter_t emitter;
+	char buf[MALLOC_PRINTF_BUFSIZE];
+	buf_descriptor_t buf_descriptor;
+
+	buf_descriptor.buf = buf;
+	buf_descriptor.len = MALLOC_PRINTF_BUFSIZE;
+	buf_descriptor.mid_quote = false;
+
+	emitter_init(&emitter, emitter_output_json, &forwarding_cb,
+	    &buf_descriptor);
+	(*emit_fn)(&emitter);
+	assert_str_eq(expected_json_output, buf, "json output failure");
+
+	buf_descriptor.buf = buf;
+	buf_descriptor.len = MALLOC_PRINTF_BUFSIZE;
+	buf_descriptor.mid_quote = false;
+
+	emitter_init(&emitter, emitter_output_table, &forwarding_cb,
+	    &buf_descriptor);
+	(*emit_fn)(&emitter);
+	assert_str_eq(expected_table_output, buf, "table output failure");
+}
+
+static void
+emit_dict(emitter_t *emitter) {
+	bool b_false = false;
+	bool b_true = true;
+	int i_123 = 123;
+	const char *str = "a string";
+
+	emitter_begin(emitter);
+	emitter_dict_begin(emitter, "foo", "This is the foo table:");
+	emitter_kv(emitter, "abc", "ABC", emitter_type_bool, &b_false);
+	emitter_kv(emitter, "def", "DEF", emitter_type_bool, &b_true);
+	emitter_kv_note(emitter, "ghi", "GHI", emitter_type_int, &i_123,
+	    "note_key1", emitter_type_string, &str);
+	emitter_kv_note(emitter, "jkl", "JKL", emitter_type_string, &str,
+	    "note_key2", emitter_type_bool, &b_false);
+	emitter_dict_end(emitter);
+	emitter_end(emitter);
+}
+static const char *dict_json =
+"{\n"
+"\t\"foo\": {\n"
+"\t\t\"abc\": false,\n"
+"\t\t\"def\": true,\n"
+"\t\t\"ghi\": 123,\n"
+"\t\t\"jkl\": \"a string\"\n"
+"\t}\n"
+"}\n";
+static const char *dict_table =
+"This is the foo table:\n"
+"  ABC: false\n"
+"  DEF: true\n"
+"  GHI: 123 (note_key1: \"a string\")\n"
+"  JKL: \"a string\" (note_key2: false)\n";
+
+TEST_BEGIN(test_dict) {
+	assert_emit_output(&emit_dict, dict_json, dict_table);
+}
+TEST_END
+
+static void
+emit_table_printf(emitter_t *emitter) {
+	emitter_begin(emitter);
+	emitter_table_printf(emitter, "Table note 1\n");
+	emitter_table_printf(emitter, "Table note 2 %s\n",
+	    "with format string");
+	emitter_end(emitter);
+}
+
+static const char *table_printf_json =
+"{\n"
+"}\n";
+
+static const char *table_printf_table =
+"Table note 1\n"
+"Table note 2 with format string\n";
+
+TEST_BEGIN(test_table_printf) {
+	assert_emit_output(&emit_table_printf, table_printf_json,
+	    table_printf_table);
+}
+TEST_END
+
+static void emit_nested_dict(emitter_t *emitter) {
+	int val = 123;
+	emitter_begin(emitter);
+	emitter_dict_begin(emitter, "json1", "Dict 1");
+	emitter_dict_begin(emitter, "json2", "Dict 2");
+	emitter_kv(emitter, "primitive", "A primitive", emitter_type_int, &val);
+	emitter_dict_end(emitter); /* Close 2 */
+	emitter_dict_begin(emitter, "json3", "Dict 3");
+	emitter_dict_end(emitter); /* Close 3 */
+	emitter_dict_end(emitter); /* Close 1 */
+	emitter_dict_begin(emitter, "json4", "Dict 4");
+	emitter_kv(emitter, "primitive", "Another primitive",
+	    emitter_type_int, &val);
+	emitter_dict_end(emitter); /* Close 4 */
+	emitter_end(emitter);
+}
+
+static const char *nested_object_json =
+"{\n"
+"\t\"json1\": {\n"
+"\t\t\"json2\": {\n"
+"\t\t\t\"primitive\": 123\n"
+"\t\t},\n"
+"\t\t\"json3\": {\n"
+"\t\t}\n"
+"\t},\n"
+"\t\"json4\": {\n"
+"\t\t\"primitive\": 123\n"
+"\t}\n"
+"}\n";
+
+static const char *nested_object_table =
+"Dict 1\n"
+"  Dict 2\n"
+"    A primitive: 123\n"
+"  Dict 3\n"
+"Dict 4\n"
+"  Another primitive: 123\n";
+
+TEST_BEGIN(test_nested_dict) {
+	assert_emit_output(&emit_nested_dict, nested_object_json,
+	    nested_object_table);
+}
+TEST_END
+
+static void
+emit_types(emitter_t *emitter) {
+	bool b = false;
+	int i = -123;
+	unsigned u = 123;
+	ssize_t zd = -456;
+	size_t zu = 456;
+	const char *str = "string";
+	uint32_t u32 = 789;
+	uint64_t u64 = 10000000000ULL;
+
+	emitter_begin(emitter);
+	emitter_kv(emitter, "k1", "K1", emitter_type_bool, &b);
+	emitter_kv(emitter, "k2", "K2", emitter_type_int, &i);
+	emitter_kv(emitter, "k3", "K3", emitter_type_unsigned, &u);
+	emitter_kv(emitter, "k4", "K4", emitter_type_ssize, &zd);
+	emitter_kv(emitter, "k5", "K5", emitter_type_size, &zu);
+	emitter_kv(emitter, "k6", "K6", emitter_type_string, &str);
+	emitter_kv(emitter, "k7", "K7", emitter_type_uint32, &u32);
+	emitter_kv(emitter, "k8", "K8", emitter_type_uint64, &u64);
+	/*
+	 * We don't test the title type, since it's only used for tables.  It's
+	 * tested in the emitter_table_row tests.
+	 */
+	emitter_end(emitter);
+}
+
+static const char *types_json =
+"{\n"
+"\t\"k1\": false,\n"
+"\t\"k2\": -123,\n"
+"\t\"k3\": 123,\n"
+"\t\"k4\": -456,\n"
+"\t\"k5\": 456,\n"
+"\t\"k6\": \"string\",\n"
+"\t\"k7\": 789,\n"
+"\t\"k8\": 10000000000\n"
+"}\n";
+
+static const char *types_table =
+"K1: false\n"
+"K2: -123\n"
+"K3: 123\n"
+"K4: -456\n"
+"K5: 456\n"
+"K6: \"string\"\n"
+"K7: 789\n"
+"K8: 10000000000\n";
+
+TEST_BEGIN(test_types) {
+	assert_emit_output(&emit_types, types_json, types_table);
+}
+TEST_END
+
+static void
+emit_modal(emitter_t *emitter) {
+	int val = 123;
+	emitter_begin(emitter);
+	emitter_dict_begin(emitter, "j0", "T0");
+	emitter_json_key(emitter, "j1");
+	emitter_json_object_begin(emitter);
+	emitter_kv(emitter, "i1", "I1", emitter_type_int, &val);
+	emitter_json_kv(emitter, "i2", emitter_type_int, &val);
+	emitter_table_kv(emitter, "I3", emitter_type_int, &val);
+	emitter_table_dict_begin(emitter, "T1");
+	emitter_kv(emitter, "i4", "I4", emitter_type_int, &val);
+	emitter_json_object_end(emitter); /* Close j1 */
+	emitter_kv(emitter, "i5", "I5", emitter_type_int, &val);
+	emitter_table_dict_end(emitter); /* Close T1 */
+	emitter_kv(emitter, "i6", "I6", emitter_type_int, &val);
+	emitter_dict_end(emitter); /* Close j0 / T0 */
+	emitter_end(emitter);
+}
+
+const char *modal_json =
+"{\n"
+"\t\"j0\": {\n"
+"\t\t\"j1\": {\n"
+"\t\t\t\"i1\": 123,\n"
+"\t\t\t\"i2\": 123,\n"
+"\t\t\t\"i4\": 123\n"
+"\t\t},\n"
+"\t\t\"i5\": 123,\n"
+"\t\t\"i6\": 123\n"
+"\t}\n"
+"}\n";
+
+const char *modal_table =
+"T0\n"
+"  I1: 123\n"
+"  I3: 123\n"
+"  T1\n"
+"    I4: 123\n"
+"    I5: 123\n"
+"  I6: 123\n";
+
+TEST_BEGIN(test_modal) {
+	assert_emit_output(&emit_modal, modal_json, modal_table);
+}
+TEST_END
+
+static void
+emit_json_arr(emitter_t *emitter) {
+	int ival = 123;
+
+	emitter_begin(emitter);
+	emitter_json_key(emitter, "dict");
+	emitter_json_object_begin(emitter);
+	emitter_json_key(emitter, "arr");
+	emitter_json_array_begin(emitter);
+	emitter_json_object_begin(emitter);
+	emitter_json_kv(emitter, "foo", emitter_type_int, &ival);
+	emitter_json_object_end(emitter); /* Close arr[0] */
+	/* arr[1] and arr[2] are primitives. */
+	emitter_json_value(emitter, emitter_type_int, &ival);
+	emitter_json_value(emitter, emitter_type_int, &ival);
+	emitter_json_object_begin(emitter);
+	emitter_json_kv(emitter, "bar", emitter_type_int, &ival);
+	emitter_json_kv(emitter, "baz", emitter_type_int, &ival);
+	emitter_json_object_end(emitter); /* Close arr[3]. */
+	emitter_json_array_end(emitter); /* Close arr. */
+	emitter_json_object_end(emitter); /* Close dict. */
+	emitter_end(emitter);
+}
+
+static const char *json_array_json =
+"{\n"
+"\t\"dict\": {\n"
+"\t\t\"arr\": [\n"
+"\t\t\t{\n"
+"\t\t\t\t\"foo\": 123\n"
+"\t\t\t},\n"
+"\t\t\t123,\n"
+"\t\t\t123,\n"
+"\t\t\t{\n"
+"\t\t\t\t\"bar\": 123,\n"
+"\t\t\t\t\"baz\": 123\n"
+"\t\t\t}\n"
+"\t\t]\n"
+"\t}\n"
+"}\n";
+
+static const char *json_array_table = "";
+
+TEST_BEGIN(test_json_arr) {
+	assert_emit_output(&emit_json_arr, json_array_json, json_array_table);
+}
+TEST_END
+
+static void
+emit_json_nested_array(emitter_t *emitter) {
+	int ival = 123;
+	char *sval = "foo";
+	emitter_begin(emitter);
+	emitter_json_array_begin(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_value(emitter, emitter_type_string, &sval);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_value(emitter, emitter_type_string, &sval);
+		emitter_json_array_end(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_array_end(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_value(emitter, emitter_type_string, &sval);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_array_end(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_array_end(emitter);
+	emitter_json_array_end(emitter);
+	emitter_end(emitter);
+}
+
+static const char *json_nested_array_json =
+"{\n"
+"\t[\n"
+"\t\t[\n"
+"\t\t\t123,\n"
+"\t\t\t\"foo\",\n"
+"\t\t\t123,\n"
+"\t\t\t\"foo\"\n"
+"\t\t],\n"
+"\t\t[\n"
+"\t\t\t123\n"
+"\t\t],\n"
+"\t\t[\n"
+"\t\t\t\"foo\",\n"
+"\t\t\t123\n"
+"\t\t],\n"
+"\t\t[\n"
+"\t\t]\n"
+"\t]\n"
+"}\n";
+
+TEST_BEGIN(test_json_nested_arr) {
+	assert_emit_output(&emit_json_nested_array, json_nested_array_json,
+	    json_array_table);
+}
+TEST_END
+
+static void
+emit_table_row(emitter_t *emitter) {
+	emitter_begin(emitter);
+	emitter_row_t row;
+	emitter_col_t abc = {emitter_justify_left, 10, emitter_type_title, {0}, {0, 0}};
+	abc.str_val = "ABC title";
+	emitter_col_t def = {emitter_justify_right, 15, emitter_type_title, {0}, {0, 0}};
+	def.str_val = "DEF title";
+	emitter_col_t ghi = {emitter_justify_right, 5, emitter_type_title, {0}, {0, 0}};
+	ghi.str_val = "GHI";
+
+	emitter_row_init(&row);
+	emitter_col_init(&abc, &row);
+	emitter_col_init(&def, &row);
+	emitter_col_init(&ghi, &row);
+
+	emitter_table_row(emitter, &row);
+
+	abc.type = emitter_type_int;
+	def.type = emitter_type_bool;
+	ghi.type = emitter_type_int;
+
+	abc.int_val = 123;
+	def.bool_val = true;
+	ghi.int_val = 456;
+	emitter_table_row(emitter, &row);
+
+	abc.int_val = 789;
+	def.bool_val = false;
+	ghi.int_val = 1011;
+	emitter_table_row(emitter, &row);
+
+	abc.type = emitter_type_string;
+	abc.str_val = "a string";
+	def.bool_val = false;
+	ghi.type = emitter_type_title;
+	ghi.str_val = "ghi";
+	emitter_table_row(emitter, &row);
+
+	emitter_end(emitter);
+}
+
+static const char *table_row_json =
+"{\n"
+"}\n";
+
+static const char *table_row_table =
+"ABC title       DEF title  GHI\n"
+"123                  true  456\n"
+"789                 false 1011\n"
+"\"a string\"          false  ghi\n";
+
+TEST_BEGIN(test_table_row) {
+	assert_emit_output(&emit_table_row, table_row_json, table_row_table);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_dict,
+	    test_table_printf,
+	    test_nested_dict,
+	    test_types,
+	    test_modal,
+	    test_json_arr,
+	    test_json_nested_arr,
+	    test_table_row);
+}
diff --git a/jemalloc/test/unit/extent_quantize.c b/jemalloc/test/unit/extent_quantize.c
new file mode 100644
index 0000000..0ca7a75
--- /dev/null
+++ b/jemalloc/test/unit/extent_quantize.c
@@ -0,0 +1,141 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_small_extent_size) {
+	unsigned nbins, i;
+	size_t sz, extent_size;
+	size_t mib[4];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+
+	/*
+	 * Iterate over all small size classes, get their extent sizes, and
+	 * verify that the quantized size is the same as the extent size.
+	 */
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &sz, NULL, 0), 0,
+	    "Unexpected mallctl failure");
+
+	assert_d_eq(mallctlnametomib("arenas.bin.0.slab_size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib failure");
+	for (i = 0; i < nbins; i++) {
+		mib[2] = i;
+		sz = sizeof(size_t);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&extent_size, &sz,
+		    NULL, 0), 0, "Unexpected mallctlbymib failure");
+		assert_zu_eq(extent_size,
+		    extent_size_quantize_floor(extent_size),
+		    "Small extent quantization should be a no-op "
+		    "(extent_size=%zu)", extent_size);
+		assert_zu_eq(extent_size,
+		    extent_size_quantize_ceil(extent_size),
+		    "Small extent quantization should be a no-op "
+		    "(extent_size=%zu)", extent_size);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_large_extent_size) {
+	bool cache_oblivious;
+	unsigned nlextents, i;
+	size_t sz, extent_size_prev, ceil_prev;
+	size_t mib[4];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+
+	/*
+	 * Iterate over all large size classes, get their extent sizes, and
+	 * verify that the quantized size is the same as the extent size.
+	 */
+
+	sz = sizeof(bool);
+	assert_d_eq(mallctl("config.cache_oblivious", (void *)&cache_oblivious,
+	    &sz, NULL, 0), 0, "Unexpected mallctl failure");
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nlextents", (void *)&nlextents, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	assert_d_eq(mallctlnametomib("arenas.lextent.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib failure");
+	for (i = 0; i < nlextents; i++) {
+		size_t lextent_size, extent_size, floor, ceil;
+
+		mib[2] = i;
+		sz = sizeof(size_t);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&lextent_size,
+		    &sz, NULL, 0), 0, "Unexpected mallctlbymib failure");
+		extent_size = cache_oblivious ? lextent_size + PAGE :
+		    lextent_size;
+		floor = extent_size_quantize_floor(extent_size);
+		ceil = extent_size_quantize_ceil(extent_size);
+
+		assert_zu_eq(extent_size, floor,
+		    "Extent quantization should be a no-op for precise size "
+		    "(lextent_size=%zu, extent_size=%zu)", lextent_size,
+		    extent_size);
+		assert_zu_eq(extent_size, ceil,
+		    "Extent quantization should be a no-op for precise size "
+		    "(lextent_size=%zu, extent_size=%zu)", lextent_size,
+		    extent_size);
+
+		if (i > 0) {
+			assert_zu_eq(extent_size_prev,
+			    extent_size_quantize_floor(extent_size - PAGE),
+			    "Floor should be a precise size");
+			if (extent_size_prev < ceil_prev) {
+				assert_zu_eq(ceil_prev, extent_size,
+				    "Ceiling should be a precise size "
+				    "(extent_size_prev=%zu, ceil_prev=%zu, "
+				    "extent_size=%zu)", extent_size_prev,
+				    ceil_prev, extent_size);
+			}
+		}
+		if (i + 1 < nlextents) {
+			extent_size_prev = floor;
+			ceil_prev = extent_size_quantize_ceil(extent_size +
+			    PAGE);
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_monotonic) {
+#define SZ_MAX	ZU(4 * 1024 * 1024)
+	unsigned i;
+	size_t floor_prev, ceil_prev;
+
+	floor_prev = 0;
+	ceil_prev = 0;
+	for (i = 1; i <= SZ_MAX >> LG_PAGE; i++) {
+		size_t extent_size, floor, ceil;
+
+		extent_size = i << LG_PAGE;
+		floor = extent_size_quantize_floor(extent_size);
+		ceil = extent_size_quantize_ceil(extent_size);
+
+		assert_zu_le(floor, extent_size,
+		    "Floor should be <= (floor=%zu, extent_size=%zu, ceil=%zu)",
+		    floor, extent_size, ceil);
+		assert_zu_ge(ceil, extent_size,
+		    "Ceiling should be >= (floor=%zu, extent_size=%zu, "
+		    "ceil=%zu)", floor, extent_size, ceil);
+
+		assert_zu_le(floor_prev, floor, "Floor should be monotonic "
+		    "(floor_prev=%zu, floor=%zu, extent_size=%zu, ceil=%zu)",
+		    floor_prev, floor, extent_size, ceil);
+		assert_zu_le(ceil_prev, ceil, "Ceiling should be monotonic "
+		    "(floor=%zu, extent_size=%zu, ceil_prev=%zu, ceil=%zu)",
+		    floor, extent_size, ceil_prev, ceil);
+
+		floor_prev = floor;
+		ceil_prev = ceil;
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_small_extent_size,
+	    test_large_extent_size,
+	    test_monotonic);
+}
diff --git a/jemalloc/test/unit/extent_util.c b/jemalloc/test/unit/extent_util.c
new file mode 100644
index 0000000..4de0b04
--- /dev/null
+++ b/jemalloc/test/unit/extent_util.c
@@ -0,0 +1,267 @@
+#include "test/jemalloc_test.h"
+
+#define TEST_UTIL_EINVAL(node, a, b, c, d, why_inval) do {		\
+	assert_d_eq(mallctl("experimental.utilization." node,		\
+	    a, b, c, d), EINVAL, "Should fail when " why_inval);	\
+	assert_zu_eq(out_sz, out_sz_ref,				\
+	    "Output size touched when given invalid arguments");	\
+	assert_d_eq(memcmp(out, out_ref, out_sz_ref), 0,		\
+	    "Output content touched when given invalid arguments");	\
+} while (0)
+
+#define TEST_UTIL_QUERY_EINVAL(a, b, c, d, why_inval)			\
+	TEST_UTIL_EINVAL("query", a, b, c, d, why_inval)
+#define TEST_UTIL_BATCH_EINVAL(a, b, c, d, why_inval)			\
+	TEST_UTIL_EINVAL("batch_query", a, b, c, d, why_inval)
+
+#define TEST_UTIL_VALID(node) do {					\
+        assert_d_eq(mallctl("experimental.utilization." node,		\
+	    out, &out_sz, in, in_sz), 0,				\
+	    "Should return 0 on correct arguments");			\
+        assert_zu_eq(out_sz, out_sz_ref, "incorrect output size");	\
+	assert_d_ne(memcmp(out, out_ref, out_sz_ref), 0,		\
+	    "Output content should be changed");			\
+} while (0)
+
+#define TEST_UTIL_BATCH_VALID TEST_UTIL_VALID("batch_query")
+
+#define TEST_MAX_SIZE (1 << 20)
+
+TEST_BEGIN(test_query) {
+	size_t sz;
+	/*
+	 * Select some sizes that can span both small and large sizes, and are
+	 * numerically unrelated to any size boundaries.
+	 */
+	for (sz = 7; sz <= TEST_MAX_SIZE && sz <= SC_LARGE_MAXCLASS;
+	    sz += (sz <= SC_SMALL_MAXCLASS ? 1009 : 99989)) {
+		void *p = mallocx(sz, 0);
+		void **in = &p;
+		size_t in_sz = sizeof(const void *);
+		size_t out_sz = sizeof(void *) + sizeof(size_t) * 5;
+		void *out = mallocx(out_sz, 0);
+		void *out_ref = mallocx(out_sz, 0);
+		size_t out_sz_ref = out_sz;
+
+		assert_ptr_not_null(p,
+		    "test pointer allocation failed");
+		assert_ptr_not_null(out,
+		    "test output allocation failed");
+		assert_ptr_not_null(out_ref,
+		    "test reference output allocation failed");
+
+#define SLABCUR_READ(out) (*(void **)out)
+#define COUNTS(out) ((size_t *)((void **)out + 1))
+#define NFREE_READ(out) COUNTS(out)[0]
+#define NREGS_READ(out) COUNTS(out)[1]
+#define SIZE_READ(out) COUNTS(out)[2]
+#define BIN_NFREE_READ(out) COUNTS(out)[3]
+#define BIN_NREGS_READ(out) COUNTS(out)[4]
+
+		SLABCUR_READ(out) = NULL;
+		NFREE_READ(out) = NREGS_READ(out) = SIZE_READ(out) = -1;
+		BIN_NFREE_READ(out) = BIN_NREGS_READ(out) = -1;
+		memcpy(out_ref, out, out_sz);
+
+		/* Test invalid argument(s) errors */
+		TEST_UTIL_QUERY_EINVAL(NULL, &out_sz, in, in_sz,
+		    "old is NULL");
+		TEST_UTIL_QUERY_EINVAL(out, NULL, in, in_sz,
+		    "oldlenp is NULL");
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, NULL, in_sz,
+		    "newp is NULL");
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, in, 0,
+		    "newlen is zero");
+		in_sz -= 1;
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, in, in_sz,
+		    "invalid newlen");
+		in_sz += 1;
+		out_sz_ref = out_sz -= 2 * sizeof(size_t);
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, in, in_sz,
+		    "invalid *oldlenp");
+		out_sz_ref = out_sz += 2 * sizeof(size_t);
+
+		/* Examine output for valid call */
+		TEST_UTIL_VALID("query");
+		assert_zu_le(sz, SIZE_READ(out),
+		    "Extent size should be at least allocation size");
+		assert_zu_eq(SIZE_READ(out) & (PAGE - 1), 0,
+		    "Extent size should be a multiple of page size");
+		if (sz <= SC_SMALL_MAXCLASS) {
+			assert_zu_le(NFREE_READ(out), NREGS_READ(out),
+			    "Extent free count exceeded region count");
+			assert_zu_le(NREGS_READ(out), SIZE_READ(out),
+			    "Extent region count exceeded size");
+			assert_zu_ne(NREGS_READ(out), 0,
+			    "Extent region count must be positive");
+			assert_true(NFREE_READ(out) == 0 || (SLABCUR_READ(out)
+			    != NULL && SLABCUR_READ(out) <= p),
+			    "Allocation should follow first fit principle");
+			if (config_stats) {
+				assert_zu_le(BIN_NFREE_READ(out),
+				    BIN_NREGS_READ(out),
+				    "Bin free count exceeded region count");
+				assert_zu_ne(BIN_NREGS_READ(out), 0,
+				    "Bin region count must be positive");
+				assert_zu_le(NFREE_READ(out),
+				    BIN_NFREE_READ(out),
+				    "Extent free count exceeded bin free count");
+				assert_zu_le(NREGS_READ(out),
+				    BIN_NREGS_READ(out),
+				    "Extent region count exceeded "
+				    "bin region count");
+				assert_zu_eq(BIN_NREGS_READ(out)
+				    % NREGS_READ(out), 0,
+				    "Bin region count isn't a multiple of "
+				    "extent region count");
+				assert_zu_le(
+				    BIN_NFREE_READ(out) - NFREE_READ(out),
+				    BIN_NREGS_READ(out) - NREGS_READ(out),
+				    "Free count in other extents in the bin "
+				    "exceeded region count in other extents "
+				    "in the bin");
+				assert_zu_le(NREGS_READ(out) - NFREE_READ(out),
+				    BIN_NREGS_READ(out) - BIN_NFREE_READ(out),
+				    "Extent utilized count exceeded "
+				    "bin utilized count");
+			}
+		} else {
+			assert_zu_eq(NFREE_READ(out), 0,
+			    "Extent free count should be zero");
+			assert_zu_eq(NREGS_READ(out), 1,
+			    "Extent region count should be one");
+			assert_ptr_null(SLABCUR_READ(out),
+			    "Current slab must be null for large size classes");
+			if (config_stats) {
+				assert_zu_eq(BIN_NFREE_READ(out), 0,
+				    "Bin free count must be zero for "
+				    "large sizes");
+				assert_zu_eq(BIN_NREGS_READ(out), 0,
+				    "Bin region count must be zero for "
+				    "large sizes");
+			}
+		}
+
+#undef BIN_NREGS_READ
+#undef BIN_NFREE_READ
+#undef SIZE_READ
+#undef NREGS_READ
+#undef NFREE_READ
+#undef COUNTS
+#undef SLABCUR_READ
+
+		free(out_ref);
+		free(out);
+		free(p);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_batch) {
+	size_t sz;
+	/*
+	 * Select some sizes that can span both small and large sizes, and are
+	 * numerically unrelated to any size boundaries.
+	 */
+	for (sz = 17; sz <= TEST_MAX_SIZE && sz <= SC_LARGE_MAXCLASS;
+	    sz += (sz <= SC_SMALL_MAXCLASS ? 1019 : 99991)) {
+		void *p = mallocx(sz, 0);
+		void *q = mallocx(sz, 0);
+		void *in[] = {p, q};
+		size_t in_sz = sizeof(const void *) * 2;
+		size_t out[] = {-1, -1, -1, -1, -1, -1};
+		size_t out_sz = sizeof(size_t) * 6;
+		size_t out_ref[] = {-1, -1, -1, -1, -1, -1};
+		size_t out_sz_ref = out_sz;
+
+		assert_ptr_not_null(p, "test pointer allocation failed");
+		assert_ptr_not_null(q, "test pointer allocation failed");
+
+		/* Test invalid argument(s) errors */
+		TEST_UTIL_BATCH_EINVAL(NULL, &out_sz, in, in_sz,
+		    "old is NULL");
+		TEST_UTIL_BATCH_EINVAL(out, NULL, in, in_sz,
+		    "oldlenp is NULL");
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, NULL, in_sz,
+		    "newp is NULL");
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, 0,
+		    "newlen is zero");
+		in_sz -= 1;
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, in_sz,
+		    "newlen is not an exact multiple");
+		in_sz += 1;
+		out_sz_ref = out_sz -= 2 * sizeof(size_t);
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, in_sz,
+		    "*oldlenp is not an exact multiple");
+		out_sz_ref = out_sz += 2 * sizeof(size_t);
+		in_sz -= sizeof(const void *);
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, in_sz,
+		    "*oldlenp and newlen do not match");
+		in_sz += sizeof(const void *);
+
+	/* Examine output for valid calls */
+#define TEST_EQUAL_REF(i, message) \
+	assert_d_eq(memcmp(out + (i) * 3, out_ref + (i) * 3, 3), 0, message)
+
+#define NFREE_READ(out, i) out[(i) * 3]
+#define NREGS_READ(out, i) out[(i) * 3 + 1]
+#define SIZE_READ(out, i) out[(i) * 3 + 2]
+
+		out_sz_ref = out_sz /= 2;
+		in_sz /= 2;
+		TEST_UTIL_BATCH_VALID;
+		assert_zu_le(sz, SIZE_READ(out, 0),
+		    "Extent size should be at least allocation size");
+		assert_zu_eq(SIZE_READ(out, 0) & (PAGE - 1), 0,
+		    "Extent size should be a multiple of page size");
+		if (sz <= SC_SMALL_MAXCLASS) {
+			assert_zu_le(NFREE_READ(out, 0), NREGS_READ(out, 0),
+			    "Extent free count exceeded region count");
+			assert_zu_le(NREGS_READ(out, 0), SIZE_READ(out, 0),
+			    "Extent region count exceeded size");
+			assert_zu_ne(NREGS_READ(out, 0), 0,
+			    "Extent region count must be positive");
+		} else {
+			assert_zu_eq(NFREE_READ(out, 0), 0,
+			    "Extent free count should be zero");
+			assert_zu_eq(NREGS_READ(out, 0), 1,
+			    "Extent region count should be one");
+		}
+		TEST_EQUAL_REF(1,
+		    "Should not overwrite content beyond what's needed");
+		in_sz *= 2;
+		out_sz_ref = out_sz *= 2;
+
+		memcpy(out_ref, out, 3 * sizeof(size_t));
+		TEST_UTIL_BATCH_VALID;
+		TEST_EQUAL_REF(0, "Statistics should be stable across calls");
+		if (sz <= SC_SMALL_MAXCLASS) {
+			assert_zu_le(NFREE_READ(out, 1), NREGS_READ(out, 1),
+			    "Extent free count exceeded region count");
+		} else {
+			assert_zu_eq(NFREE_READ(out, 0), 0,
+			    "Extent free count should be zero");
+		}
+		assert_zu_eq(NREGS_READ(out, 0), NREGS_READ(out, 1),
+		    "Extent region count should be same for same region size");
+		assert_zu_eq(SIZE_READ(out, 0), SIZE_READ(out, 1),
+		    "Extent size should be same for same region size");
+
+#undef SIZE_READ
+#undef NREGS_READ
+#undef NFREE_READ
+
+#undef TEST_EQUAL_REF
+
+		free(q);
+		free(p);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	assert_zu_lt(SC_SMALL_MAXCLASS, TEST_MAX_SIZE,
+	    "Test case cannot cover large classes");
+	return test(test_query, test_batch);
+}
diff --git a/jemalloc/test/unit/fork.c b/jemalloc/test/unit/fork.c
new file mode 100644
index 0000000..b169075
--- /dev/null
+++ b/jemalloc/test/unit/fork.c
@@ -0,0 +1,141 @@
+#include "test/jemalloc_test.h"
+
+#ifndef _WIN32
+#include <sys/wait.h>
+#endif
+
+#ifndef _WIN32
+static void
+wait_for_child_exit(int pid) {
+	int status;
+	while (true) {
+		if (waitpid(pid, &status, 0) == -1) {
+			test_fail("Unexpected waitpid() failure.");
+		}
+		if (WIFSIGNALED(status)) {
+			test_fail("Unexpected child termination due to "
+			    "signal %d", WTERMSIG(status));
+			break;
+		}
+		if (WIFEXITED(status)) {
+			if (WEXITSTATUS(status) != 0) {
+				test_fail("Unexpected child exit value %d",
+				    WEXITSTATUS(status));
+			}
+			break;
+		}
+	}
+}
+#endif
+
+TEST_BEGIN(test_fork) {
+#ifndef _WIN32
+	void *p;
+	pid_t pid;
+
+	/* Set up a manually managed arena for test. */
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+
+	/* Migrate to the new arena. */
+	unsigned old_arena_ind;
+	sz = sizeof(old_arena_ind);
+	assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+	    (void *)&arena_ind, sizeof(arena_ind)), 0,
+	    "Unexpected mallctl() failure");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	pid = fork();
+
+	free(p);
+
+	p = malloc(64);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+	free(p);
+
+	if (pid == -1) {
+		/* Error. */
+		test_fail("Unexpected fork() failure");
+	} else if (pid == 0) {
+		/* Child. */
+		_exit(0);
+	} else {
+		wait_for_child_exit(pid);
+	}
+#else
+	test_skip("fork(2) is irrelevant to Windows");
+#endif
+}
+TEST_END
+
+#ifndef _WIN32
+static void *
+do_fork_thd(void *arg) {
+	malloc(1);
+	int pid = fork();
+	if (pid == -1) {
+		/* Error. */
+		test_fail("Unexpected fork() failure");
+	} else if (pid == 0) {
+		/* Child. */
+		char *args[] = {"true", NULL};
+		execvp(args[0], args);
+		test_fail("Exec failed");
+	} else {
+		/* Parent */
+		wait_for_child_exit(pid);
+	}
+	return NULL;
+}
+#endif
+
+#ifndef _WIN32
+static void
+do_test_fork_multithreaded() {
+	thd_t child;
+	thd_create(&child, do_fork_thd, NULL);
+	do_fork_thd(NULL);
+	thd_join(child, NULL);
+}
+#endif
+
+TEST_BEGIN(test_fork_multithreaded) {
+#ifndef _WIN32
+	/*
+	 * We've seen bugs involving hanging on arenas_lock (though the same
+	 * class of bugs can happen on any mutex).  The bugs are intermittent
+	 * though, so we want to run the test multiple times.  Since we hold the
+	 * arenas lock only early in the process lifetime, we can't just run
+	 * this test in a loop (since, after all the arenas are initialized, we
+	 * won't acquire arenas_lock any further).  We therefore repeat the test
+	 * with multiple processes.
+	 */
+	for (int i = 0; i < 100; i++) {
+		int pid = fork();
+		if (pid == -1) {
+			/* Error. */
+			test_fail("Unexpected fork() failure,");
+		} else if (pid == 0) {
+			/* Child. */
+			do_test_fork_multithreaded();
+			_exit(0);
+		} else {
+			wait_for_child_exit(pid);
+		}
+	}
+#else
+	test_skip("fork(2) is irrelevant to Windows");
+#endif
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_fork,
+	    test_fork_multithreaded);
+}
diff --git a/jemalloc/test/unit/hash.c b/jemalloc/test/unit/hash.c
new file mode 100644
index 0000000..7cc034f
--- /dev/null
+++ b/jemalloc/test/unit/hash.c
@@ -0,0 +1,173 @@
+/*
+ * This file is based on code that is part of SMHasher
+ * (https://code.google.com/p/smhasher/), and is subject to the MIT license
+ * (http://www.opensource.org/licenses/mit-license.php).  Both email addresses
+ * associated with the source code's revision history belong to Austin Appleby,
+ * and the revision history ranges from 2010 to 2012.  Therefore the copyright
+ * and license are here taken to be:
+ *
+ * Copyright (c) 2010-2012 Austin Appleby
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "test/jemalloc_test.h"
+#include "jemalloc/internal/hash.h"
+
+typedef enum {
+	hash_variant_x86_32,
+	hash_variant_x86_128,
+	hash_variant_x64_128
+} hash_variant_t;
+
+static int
+hash_variant_bits(hash_variant_t variant) {
+	switch (variant) {
+	case hash_variant_x86_32: return 32;
+	case hash_variant_x86_128: return 128;
+	case hash_variant_x64_128: return 128;
+	default: not_reached();
+	}
+}
+
+static const char *
+hash_variant_string(hash_variant_t variant) {
+	switch (variant) {
+	case hash_variant_x86_32: return "hash_x86_32";
+	case hash_variant_x86_128: return "hash_x86_128";
+	case hash_variant_x64_128: return "hash_x64_128";
+	default: not_reached();
+	}
+}
+
+#define KEY_SIZE	256
+static void
+hash_variant_verify_key(hash_variant_t variant, uint8_t *key) {
+	const int hashbytes = hash_variant_bits(variant) / 8;
+	const int hashes_size = hashbytes * 256;
+	VARIABLE_ARRAY(uint8_t, hashes, hashes_size);
+	VARIABLE_ARRAY(uint8_t, final, hashbytes);
+	unsigned i;
+	uint32_t computed, expected;
+
+	memset(key, 0, KEY_SIZE);
+	memset(hashes, 0, hashes_size);
+	memset(final, 0, hashbytes);
+
+	/*
+	 * Hash keys of the form {0}, {0,1}, {0,1,2}, ..., {0,1,...,255} as the
+	 * seed.
+	 */
+	for (i = 0; i < 256; i++) {
+		key[i] = (uint8_t)i;
+		switch (variant) {
+		case hash_variant_x86_32: {
+			uint32_t out;
+			out = hash_x86_32(key, i, 256-i);
+			memcpy(&hashes[i*hashbytes], &out, hashbytes);
+			break;
+		} case hash_variant_x86_128: {
+			uint64_t out[2];
+			hash_x86_128(key, i, 256-i, out);
+			memcpy(&hashes[i*hashbytes], out, hashbytes);
+			break;
+		} case hash_variant_x64_128: {
+			uint64_t out[2];
+			hash_x64_128(key, i, 256-i, out);
+			memcpy(&hashes[i*hashbytes], out, hashbytes);
+			break;
+		} default: not_reached();
+		}
+	}
+
+	/* Hash the result array. */
+	switch (variant) {
+	case hash_variant_x86_32: {
+		uint32_t out = hash_x86_32(hashes, hashes_size, 0);
+		memcpy(final, &out, sizeof(out));
+		break;
+	} case hash_variant_x86_128: {
+		uint64_t out[2];
+		hash_x86_128(hashes, hashes_size, 0, out);
+		memcpy(final, out, sizeof(out));
+		break;
+	} case hash_variant_x64_128: {
+		uint64_t out[2];
+		hash_x64_128(hashes, hashes_size, 0, out);
+		memcpy(final, out, sizeof(out));
+		break;
+	} default: not_reached();
+	}
+
+	computed = (final[0] << 0) | (final[1] << 8) | (final[2] << 16) |
+	    (final[3] << 24);
+
+	switch (variant) {
+#ifdef JEMALLOC_BIG_ENDIAN
+	case hash_variant_x86_32: expected = 0x6213303eU; break;
+	case hash_variant_x86_128: expected = 0x266820caU; break;
+	case hash_variant_x64_128: expected = 0xcc622b6fU; break;
+#else
+	case hash_variant_x86_32: expected = 0xb0f57ee3U; break;
+	case hash_variant_x86_128: expected = 0xb3ece62aU; break;
+	case hash_variant_x64_128: expected = 0x6384ba69U; break;
+#endif
+	default: not_reached();
+	}
+
+	assert_u32_eq(computed, expected,
+	    "Hash mismatch for %s(): expected %#x but got %#x",
+	    hash_variant_string(variant), expected, computed);
+}
+
+static void
+hash_variant_verify(hash_variant_t variant) {
+#define MAX_ALIGN	16
+	uint8_t key[KEY_SIZE + (MAX_ALIGN - 1)];
+	unsigned i;
+
+	for (i = 0; i < MAX_ALIGN; i++) {
+		hash_variant_verify_key(variant, &key[i]);
+	}
+#undef MAX_ALIGN
+}
+#undef KEY_SIZE
+
+TEST_BEGIN(test_hash_x86_32) {
+	hash_variant_verify(hash_variant_x86_32);
+}
+TEST_END
+
+TEST_BEGIN(test_hash_x86_128) {
+	hash_variant_verify(hash_variant_x86_128);
+}
+TEST_END
+
+TEST_BEGIN(test_hash_x64_128) {
+	hash_variant_verify(hash_variant_x64_128);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_hash_x86_32,
+	    test_hash_x86_128,
+	    test_hash_x64_128);
+}
diff --git a/jemalloc/test/unit/hook.c b/jemalloc/test/unit/hook.c
new file mode 100644
index 0000000..72fcc43
--- /dev/null
+++ b/jemalloc/test/unit/hook.c
@@ -0,0 +1,580 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/hook.h"
+
+static void *arg_extra;
+static int arg_type;
+static void *arg_result;
+static void *arg_address;
+static size_t arg_old_usize;
+static size_t arg_new_usize;
+static uintptr_t arg_result_raw;
+static uintptr_t arg_args_raw[4];
+
+static int call_count = 0;
+
+static void
+reset_args() {
+	arg_extra = NULL;
+	arg_type = 12345;
+	arg_result = NULL;
+	arg_address = NULL;
+	arg_old_usize = 0;
+	arg_new_usize = 0;
+	arg_result_raw = 0;
+	memset(arg_args_raw, 77, sizeof(arg_args_raw));
+}
+
+static void
+alloc_free_size(size_t sz) {
+	void *ptr = mallocx(1, 0);
+	free(ptr);
+	ptr = mallocx(1, 0);
+	free(ptr);
+	ptr = mallocx(1, MALLOCX_TCACHE_NONE);
+	dallocx(ptr, MALLOCX_TCACHE_NONE);
+}
+
+/*
+ * We want to support a degree of user reentrancy.  This tests a variety of
+ * allocation scenarios.
+ */
+static void
+be_reentrant() {
+	/* Let's make sure the tcache is non-empty if enabled. */
+	alloc_free_size(1);
+	alloc_free_size(1024);
+	alloc_free_size(64 * 1024);
+	alloc_free_size(256 * 1024);
+	alloc_free_size(1024 * 1024);
+
+	/* Some reallocation. */
+	void *ptr = mallocx(129, 0);
+	ptr = rallocx(ptr, 130, 0);
+	free(ptr);
+
+	ptr = mallocx(2 * 1024 * 1024, 0);
+	free(ptr);
+	ptr = mallocx(1 * 1024 * 1024, 0);
+	ptr = rallocx(ptr, 2 * 1024 * 1024, 0);
+	free(ptr);
+
+	ptr = mallocx(1, 0);
+	ptr = rallocx(ptr, 1000, 0);
+	free(ptr);
+}
+
+static void
+set_args_raw(uintptr_t *args_raw, int nargs) {
+	memcpy(arg_args_raw, args_raw, sizeof(uintptr_t) * nargs);
+}
+
+static void
+assert_args_raw(uintptr_t *args_raw_expected, int nargs) {
+	int cmp = memcmp(args_raw_expected, arg_args_raw,
+	    sizeof(uintptr_t) * nargs);
+	assert_d_eq(cmp, 0, "Raw args mismatch");
+}
+
+static void
+reset() {
+	call_count = 0;
+	reset_args();
+}
+
+static void
+test_alloc_hook(void *extra, hook_alloc_t type, void *result,
+    uintptr_t result_raw, uintptr_t args_raw[3]) {
+	call_count++;
+	arg_extra = extra;
+	arg_type = (int)type;
+	arg_result = result;
+	arg_result_raw = result_raw;
+	set_args_raw(args_raw, 3);
+	be_reentrant();
+}
+
+static void
+test_dalloc_hook(void *extra, hook_dalloc_t type, void *address,
+    uintptr_t args_raw[3]) {
+	call_count++;
+	arg_extra = extra;
+	arg_type = (int)type;
+	arg_address = address;
+	set_args_raw(args_raw, 3);
+	be_reentrant();
+}
+
+static void
+test_expand_hook(void *extra, hook_expand_t type, void *address,
+    size_t old_usize, size_t new_usize, uintptr_t result_raw,
+    uintptr_t args_raw[4]) {
+	call_count++;
+	arg_extra = extra;
+	arg_type = (int)type;
+	arg_address = address;
+	arg_old_usize = old_usize;
+	arg_new_usize = new_usize;
+	arg_result_raw = result_raw;
+	set_args_raw(args_raw, 4);
+	be_reentrant();
+}
+
+TEST_BEGIN(test_hooks_basic) {
+	/* Just verify that the record their arguments correctly. */
+	hooks_t hooks = {
+		&test_alloc_hook, &test_dalloc_hook, &test_expand_hook,
+		(void *)111};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	uintptr_t args_raw[4] = {10, 20, 30, 40};
+
+	/* Alloc */
+	reset_args();
+	hook_invoke_alloc(hook_alloc_posix_memalign, (void *)222, 333,
+	    args_raw);
+	assert_ptr_eq(arg_extra, (void *)111, "Passed wrong user pointer");
+	assert_d_eq((int)hook_alloc_posix_memalign, arg_type,
+	    "Passed wrong alloc type");
+	assert_ptr_eq((void *)222, arg_result, "Passed wrong result address");
+	assert_u64_eq(333, arg_result_raw, "Passed wrong result");
+	assert_args_raw(args_raw, 3);
+
+	/* Dalloc */
+	reset_args();
+	hook_invoke_dalloc(hook_dalloc_sdallocx, (void *)222, args_raw);
+	assert_d_eq((int)hook_dalloc_sdallocx, arg_type,
+	    "Passed wrong dalloc type");
+	assert_ptr_eq((void *)111, arg_extra, "Passed wrong user pointer");
+	assert_ptr_eq((void *)222, arg_address, "Passed wrong address");
+	assert_args_raw(args_raw, 3);
+
+	/* Expand */
+	reset_args();
+	hook_invoke_expand(hook_expand_xallocx, (void *)222, 333, 444, 555,
+	    args_raw);
+	assert_d_eq((int)hook_expand_xallocx, arg_type,
+	    "Passed wrong expand type");
+	assert_ptr_eq((void *)111, arg_extra, "Passed wrong user pointer");
+	assert_ptr_eq((void *)222, arg_address, "Passed wrong address");
+	assert_zu_eq(333, arg_old_usize, "Passed wrong old usize");
+	assert_zu_eq(444, arg_new_usize, "Passed wrong new usize");
+	assert_zu_eq(555, arg_result_raw, "Passed wrong result");
+	assert_args_raw(args_raw, 4);
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_null) {
+	/* Null hooks should be ignored, not crash. */
+	hooks_t hooks1 = {NULL, NULL, NULL, NULL};
+	hooks_t hooks2 = {&test_alloc_hook, NULL, NULL, NULL};
+	hooks_t hooks3 = {NULL, &test_dalloc_hook, NULL, NULL};
+	hooks_t hooks4 = {NULL, NULL, &test_expand_hook, NULL};
+
+	void *handle1 = hook_install(TSDN_NULL, &hooks1);
+	void *handle2 = hook_install(TSDN_NULL, &hooks2);
+	void *handle3 = hook_install(TSDN_NULL, &hooks3);
+	void *handle4 = hook_install(TSDN_NULL, &hooks4);
+
+	assert_ptr_ne(handle1, NULL, "Hook installation failed");
+	assert_ptr_ne(handle2, NULL, "Hook installation failed");
+	assert_ptr_ne(handle3, NULL, "Hook installation failed");
+	assert_ptr_ne(handle4, NULL, "Hook installation failed");
+
+	uintptr_t args_raw[4] = {10, 20, 30, 40};
+
+	call_count = 0;
+	hook_invoke_alloc(hook_alloc_malloc, NULL, 0, args_raw);
+	assert_d_eq(call_count, 1, "Called wrong number of times");
+
+	call_count = 0;
+	hook_invoke_dalloc(hook_dalloc_free, NULL, args_raw);
+	assert_d_eq(call_count, 1, "Called wrong number of times");
+
+	call_count = 0;
+	hook_invoke_expand(hook_expand_realloc, NULL, 0, 0, 0, args_raw);
+	assert_d_eq(call_count, 1, "Called wrong number of times");
+
+	hook_remove(TSDN_NULL, handle1);
+	hook_remove(TSDN_NULL, handle2);
+	hook_remove(TSDN_NULL, handle3);
+	hook_remove(TSDN_NULL, handle4);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_remove) {
+	hooks_t hooks = {&test_alloc_hook, NULL, NULL, NULL};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+	call_count = 0;
+	uintptr_t args_raw[4] = {10, 20, 30, 40};
+	hook_invoke_alloc(hook_alloc_malloc, NULL, 0, args_raw);
+	assert_d_eq(call_count, 1, "Hook not invoked");
+
+	call_count = 0;
+	hook_remove(TSDN_NULL, handle);
+	hook_invoke_alloc(hook_alloc_malloc, NULL, 0, NULL);
+	assert_d_eq(call_count, 0, "Hook invoked after removal");
+
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_alloc_simple) {
+	/* "Simple" in the sense that we're not in a realloc variant. */
+	hooks_t hooks = {&test_alloc_hook, NULL, NULL, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	/* Stop malloc from being optimized away. */
+	volatile int err;
+	void *volatile ptr;
+
+	/* malloc */
+	reset();
+	ptr = malloc(1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_malloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[0], "Wrong argument");
+	free(ptr);
+
+	/* posix_memalign */
+	reset();
+	err = posix_memalign((void **)&ptr, 1024, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_posix_memalign,
+	    "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)err, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)&ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1024, arg_args_raw[1], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[2], "Wrong argument");
+	free(ptr);
+
+	/* aligned_alloc */
+	reset();
+	ptr = aligned_alloc(1024, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_aligned_alloc,
+	    "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1024, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/* calloc */
+	reset();
+	ptr = calloc(11, 13);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_calloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)11, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)13, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/* memalign */
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+	reset();
+	ptr = memalign(1024, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_memalign, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1024, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+#endif /* JEMALLOC_OVERRIDE_MEMALIGN */
+
+	/* valloc */
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+	reset();
+	ptr = valloc(1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_valloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[0], "Wrong argument");
+	free(ptr);
+#endif /* JEMALLOC_OVERRIDE_VALLOC */
+
+	/* mallocx */
+	reset();
+	ptr = mallocx(1, MALLOCX_LG_ALIGN(10));
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_mallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)MALLOCX_LG_ALIGN(10), arg_args_raw[1],
+	    "Wrong flags");
+	free(ptr);
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_dalloc_simple) {
+	/* "Simple" in the sense that we're not in a realloc variant. */
+	hooks_t hooks = {NULL, &test_dalloc_hook, NULL, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+
+	/* free() */
+	reset();
+	ptr = malloc(1);
+	free(ptr);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_free, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+
+	/* dallocx() */
+	reset();
+	ptr = malloc(1);
+	dallocx(ptr, MALLOCX_TCACHE_NONE);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_dallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)MALLOCX_TCACHE_NONE, arg_args_raw[1],
+	    "Wrong raw arg");
+
+	/* sdallocx() */
+	reset();
+	ptr = malloc(1);
+	sdallocx(ptr, 1, MALLOCX_TCACHE_NONE);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_sdallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)MALLOCX_TCACHE_NONE, arg_args_raw[2],
+	    "Wrong raw arg");
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_expand_simple) {
+	/* "Simple" in the sense that we're not in a realloc variant. */
+	hooks_t hooks = {NULL, NULL, &test_expand_hook, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+
+	/* xallocx() */
+	reset();
+	ptr = malloc(1);
+	size_t new_usize = xallocx(ptr, 100, 200, MALLOCX_TCACHE_NONE);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_expand_xallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer expanded");
+	assert_u64_eq(arg_old_usize, nallocx(1, 0), "Wrong old usize");
+	assert_u64_eq(arg_new_usize, sallocx(ptr, 0), "Wrong new usize");
+	assert_u64_eq(new_usize, arg_result_raw, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong arg");
+	assert_u64_eq(100, arg_args_raw[1], "Wrong arg");
+	assert_u64_eq(200, arg_args_raw[2], "Wrong arg");
+	assert_u64_eq(MALLOCX_TCACHE_NONE, arg_args_raw[3], "Wrong arg");
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_realloc_as_malloc_or_free) {
+	hooks_t hooks = {&test_alloc_hook, &test_dalloc_hook,
+		&test_expand_hook, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+
+	/* realloc(NULL, size) as malloc */
+	reset();
+	ptr = realloc(NULL, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_realloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)NULL, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/* realloc(ptr, 0) as free */
+	ptr = malloc(1);
+	reset();
+	realloc(ptr, 0);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_realloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)0, arg_args_raw[1], "Wrong raw arg");
+
+	/* realloc(NULL, 0) as malloc(0) */
+	reset();
+	ptr = realloc(NULL, 0);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_realloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)NULL, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)0, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+static void
+do_realloc_test(void *(*ralloc)(void *, size_t, int), int flags,
+    int expand_type, int dalloc_type) {
+	hooks_t hooks = {&test_alloc_hook, &test_dalloc_hook,
+		&test_expand_hook, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+	void *volatile ptr2;
+
+	/* Realloc in-place, small. */
+	ptr = malloc(129);
+	reset();
+	ptr2 = ralloc(ptr, 130, flags);
+	assert_ptr_eq(ptr, ptr2, "Small realloc moved");
+
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, expand_type, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)130, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/*
+	 * Realloc in-place, large.  Since we can't guarantee the large case
+	 * across all platforms, we stay resilient to moving results.
+	 */
+	ptr = malloc(2 * 1024 * 1024);
+	free(ptr);
+	ptr2 = malloc(1 * 1024 * 1024);
+	reset();
+	ptr = ralloc(ptr2, 2 * 1024 * 1024, flags);
+	/* ptr is the new address, ptr2 is the old address. */
+	if (ptr == ptr2) {
+		assert_d_eq(call_count, 1, "Hook not called");
+		assert_d_eq(arg_type, expand_type, "Wrong hook type");
+	} else {
+		assert_d_eq(call_count, 2, "Wrong hooks called");
+		assert_ptr_eq(ptr, arg_result, "Wrong address");
+		assert_d_eq(arg_type, dalloc_type, "Wrong hook type");
+	}
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_ptr_eq(ptr2, arg_address, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr2, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)2 * 1024 * 1024, arg_args_raw[1],
+	    "Wrong argument");
+	free(ptr);
+
+	/* Realloc with move, small. */
+	ptr = malloc(8);
+	reset();
+	ptr2 = ralloc(ptr, 128, flags);
+	assert_ptr_ne(ptr, ptr2, "Small realloc didn't move");
+
+	assert_d_eq(call_count, 2, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, dalloc_type, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong address");
+	assert_ptr_eq(ptr2, arg_result, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr2, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)128, arg_args_raw[1], "Wrong argument");
+	free(ptr2);
+
+	/* Realloc with move, large. */
+	ptr = malloc(1);
+	reset();
+	ptr2 = ralloc(ptr, 2 * 1024 * 1024, flags);
+	assert_ptr_ne(ptr, ptr2, "Large realloc didn't move");
+
+	assert_d_eq(call_count, 2, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, dalloc_type, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong address");
+	assert_ptr_eq(ptr2, arg_result, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr2, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)2 * 1024 * 1024, arg_args_raw[1],
+	    "Wrong argument");
+	free(ptr2);
+
+	hook_remove(TSDN_NULL, handle);
+}
+
+static void *
+realloc_wrapper(void *ptr, size_t size, UNUSED int flags) {
+	return realloc(ptr, size);
+}
+
+TEST_BEGIN(test_hooks_realloc) {
+	do_realloc_test(&realloc_wrapper, 0, hook_expand_realloc,
+	    hook_dalloc_realloc);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_rallocx) {
+	do_realloc_test(&rallocx, MALLOCX_TCACHE_NONE, hook_expand_rallocx,
+	    hook_dalloc_rallocx);
+}
+TEST_END
+
+int
+main(void) {
+	/* We assert on call counts. */
+	return test_no_reentrancy(
+	    test_hooks_basic,
+	    test_hooks_null,
+	    test_hooks_remove,
+	    test_hooks_alloc_simple,
+	    test_hooks_dalloc_simple,
+	    test_hooks_expand_simple,
+	    test_hooks_realloc_as_malloc_or_free,
+	    test_hooks_realloc,
+	    test_hooks_rallocx);
+}
diff --git a/jemalloc/test/unit/huge.c b/jemalloc/test/unit/huge.c
new file mode 100644
index 0000000..ab72cf0
--- /dev/null
+++ b/jemalloc/test/unit/huge.c
@@ -0,0 +1,108 @@
+#include "test/jemalloc_test.h"
+
+/* Threshold: 2 << 20 = 2097152. */
+const char *malloc_conf = "oversize_threshold:2097152";
+
+#define HUGE_SZ (2 << 20)
+#define SMALL_SZ (8)
+
+TEST_BEGIN(huge_bind_thread) {
+	unsigned arena1, arena2;
+	size_t sz = sizeof(unsigned);
+
+	/* Bind to a manual arena. */
+	assert_d_eq(mallctl("arenas.create", &arena1, &sz, NULL, 0), 0,
+	    "Failed to create arena");
+	assert_d_eq(mallctl("thread.arena", NULL, NULL, &arena1,
+	    sizeof(arena1)), 0, "Fail to bind thread");
+
+	void *ptr = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_eq(arena1, arena2, "Wrong arena used after binding");
+	dallocx(ptr, 0);
+
+	/* Switch back to arena 0. */
+	test_skip_if(have_percpu_arena &&
+	    PERCPU_ARENA_ENABLED(opt_percpu_arena));
+	arena2 = 0;
+	assert_d_eq(mallctl("thread.arena", NULL, NULL, &arena2,
+	    sizeof(arena2)), 0, "Fail to bind thread");
+	ptr = mallocx(SMALL_SZ, MALLOCX_TCACHE_NONE);
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_eq(arena2, 0, "Wrong arena used after binding");
+	dallocx(ptr, MALLOCX_TCACHE_NONE);
+
+	/* Then huge allocation should use the huge arena. */
+	ptr = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena2, 0, "Wrong arena used after binding");
+	assert_u_ne(arena1, arena2, "Wrong arena used after binding");
+	dallocx(ptr, 0);
+}
+TEST_END
+
+TEST_BEGIN(huge_mallocx) {
+	unsigned arena1, arena2;
+	size_t sz = sizeof(unsigned);
+
+	assert_d_eq(mallctl("arenas.create", &arena1, &sz, NULL, 0), 0,
+	    "Failed to create arena");
+	void *huge = mallocx(HUGE_SZ, MALLOCX_ARENA(arena1));
+	assert_ptr_not_null(huge, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &huge,
+	    sizeof(huge)), 0, "Unexpected mallctl() failure");
+	assert_u_eq(arena1, arena2, "Wrong arena used for mallocx");
+	dallocx(huge, MALLOCX_ARENA(arena1));
+
+	void *huge2 = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(huge, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &huge2,
+	    sizeof(huge2)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena1, arena2,
+	    "Huge allocation should not come from the manual arena.");
+	assert_u_ne(arena2, 0,
+	    "Huge allocation should not come from the arena 0.");
+	dallocx(huge2, 0);
+}
+TEST_END
+
+TEST_BEGIN(huge_allocation) {
+	unsigned arena1, arena2;
+
+	void *ptr = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate huge size");
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.lookup", &arena1, &sz, &ptr, sizeof(ptr)),
+	    0, "Unexpected mallctl() failure");
+	assert_u_gt(arena1, 0, "Huge allocation should not come from arena 0");
+	dallocx(ptr, 0);
+
+	ptr = mallocx(HUGE_SZ >> 1, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate half huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena1, arena2, "Wrong arena used for half huge");
+	dallocx(ptr, 0);
+
+	ptr = mallocx(SMALL_SZ, MALLOCX_TCACHE_NONE);
+	assert_ptr_not_null(ptr, "Fail to allocate small size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena1, arena2,
+	    "Huge and small should be from different arenas");
+	dallocx(ptr, 0);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    huge_allocation,
+	    huge_mallocx,
+	    huge_bind_thread);
+}
diff --git a/jemalloc/test/unit/junk.c b/jemalloc/test/unit/junk.c
new file mode 100644
index 0000000..57e3ad4
--- /dev/null
+++ b/jemalloc/test/unit/junk.c
@@ -0,0 +1,141 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/util.h"
+
+static arena_dalloc_junk_small_t *arena_dalloc_junk_small_orig;
+static large_dalloc_junk_t *large_dalloc_junk_orig;
+static large_dalloc_maybe_junk_t *large_dalloc_maybe_junk_orig;
+static void *watch_for_junking;
+static bool saw_junking;
+
+static void
+watch_junking(void *p) {
+	watch_for_junking = p;
+	saw_junking = false;
+}
+
+static void
+arena_dalloc_junk_small_intercept(void *ptr, const bin_info_t *bin_info) {
+	size_t i;
+
+	arena_dalloc_junk_small_orig(ptr, bin_info);
+	for (i = 0; i < bin_info->reg_size; i++) {
+		assert_u_eq(((uint8_t *)ptr)[i], JEMALLOC_FREE_JUNK,
+		    "Missing junk fill for byte %zu/%zu of deallocated region",
+		    i, bin_info->reg_size);
+	}
+	if (ptr == watch_for_junking) {
+		saw_junking = true;
+	}
+}
+
+static void
+large_dalloc_junk_intercept(void *ptr, size_t usize) {
+	size_t i;
+
+	large_dalloc_junk_orig(ptr, usize);
+	for (i = 0; i < usize; i++) {
+		assert_u_eq(((uint8_t *)ptr)[i], JEMALLOC_FREE_JUNK,
+		    "Missing junk fill for byte %zu/%zu of deallocated region",
+		    i, usize);
+	}
+	if (ptr == watch_for_junking) {
+		saw_junking = true;
+	}
+}
+
+static void
+large_dalloc_maybe_junk_intercept(void *ptr, size_t usize) {
+	large_dalloc_maybe_junk_orig(ptr, usize);
+	if (ptr == watch_for_junking) {
+		saw_junking = true;
+	}
+}
+
+static void
+test_junk(size_t sz_min, size_t sz_max) {
+	uint8_t *s;
+	size_t sz_prev, sz, i;
+
+	if (opt_junk_free) {
+		arena_dalloc_junk_small_orig = arena_dalloc_junk_small;
+		arena_dalloc_junk_small = arena_dalloc_junk_small_intercept;
+		large_dalloc_junk_orig = large_dalloc_junk;
+		large_dalloc_junk = large_dalloc_junk_intercept;
+		large_dalloc_maybe_junk_orig = large_dalloc_maybe_junk;
+		large_dalloc_maybe_junk = large_dalloc_maybe_junk_intercept;
+	}
+
+	sz_prev = 0;
+	s = (uint8_t *)mallocx(sz_min, 0);
+	assert_ptr_not_null((void *)s, "Unexpected mallocx() failure");
+
+	for (sz = sallocx(s, 0); sz <= sz_max;
+	    sz_prev = sz, sz = sallocx(s, 0)) {
+		if (sz_prev > 0) {
+			assert_u_eq(s[0], 'a',
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    ZU(0), sz_prev);
+			assert_u_eq(s[sz_prev-1], 'a',
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    sz_prev-1, sz_prev);
+		}
+
+		for (i = sz_prev; i < sz; i++) {
+			if (opt_junk_alloc) {
+				assert_u_eq(s[i], JEMALLOC_ALLOC_JUNK,
+				    "Newly allocated byte %zu/%zu isn't "
+				    "junk-filled", i, sz);
+			}
+			s[i] = 'a';
+		}
+
+		if (xallocx(s, sz+1, 0, 0) == sz) {
+			uint8_t *t;
+			watch_junking(s);
+			t = (uint8_t *)rallocx(s, sz+1, 0);
+			assert_ptr_not_null((void *)t,
+			    "Unexpected rallocx() failure");
+			assert_zu_ge(sallocx(t, 0), sz+1,
+			    "Unexpectedly small rallocx() result");
+			if (!background_thread_enabled()) {
+				assert_ptr_ne(s, t,
+				    "Unexpected in-place rallocx()");
+				assert_true(!opt_junk_free || saw_junking,
+				    "Expected region of size %zu to be "
+				    "junk-filled", sz);
+			}
+			s = t;
+		}
+	}
+
+	watch_junking(s);
+	dallocx(s, 0);
+	assert_true(!opt_junk_free || saw_junking,
+	    "Expected region of size %zu to be junk-filled", sz);
+
+	if (opt_junk_free) {
+		arena_dalloc_junk_small = arena_dalloc_junk_small_orig;
+		large_dalloc_junk = large_dalloc_junk_orig;
+		large_dalloc_maybe_junk = large_dalloc_maybe_junk_orig;
+	}
+}
+
+TEST_BEGIN(test_junk_small) {
+	test_skip_if(!config_fill);
+	test_junk(1, SC_SMALL_MAXCLASS - 1);
+}
+TEST_END
+
+TEST_BEGIN(test_junk_large) {
+	test_skip_if(!config_fill);
+	test_junk(SC_SMALL_MAXCLASS + 1, (1U << (SC_LG_LARGE_MINCLASS + 1)));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_junk_small,
+	    test_junk_large);
+}
diff --git a/jemalloc/test/unit/junk.sh b/jemalloc/test/unit/junk.sh
new file mode 100644
index 0000000..97cd8ca
--- /dev/null
+++ b/jemalloc/test/unit/junk.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,zero:false,junk:true"
+fi
diff --git a/jemalloc/test/unit/junk_alloc.c b/jemalloc/test/unit/junk_alloc.c
new file mode 100644
index 0000000..a442a0c
--- /dev/null
+++ b/jemalloc/test/unit/junk_alloc.c
@@ -0,0 +1 @@
+#include "junk.c"
diff --git a/jemalloc/test/unit/junk_alloc.sh b/jemalloc/test/unit/junk_alloc.sh
new file mode 100644
index 0000000..e1008c2
--- /dev/null
+++ b/jemalloc/test/unit/junk_alloc.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,zero:false,junk:alloc"
+fi
diff --git a/jemalloc/test/unit/junk_free.c b/jemalloc/test/unit/junk_free.c
new file mode 100644
index 0000000..a442a0c
--- /dev/null
+++ b/jemalloc/test/unit/junk_free.c
@@ -0,0 +1 @@
+#include "junk.c"
diff --git a/jemalloc/test/unit/junk_free.sh b/jemalloc/test/unit/junk_free.sh
new file mode 100644
index 0000000..402196c
--- /dev/null
+++ b/jemalloc/test/unit/junk_free.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,zero:false,junk:free"
+fi
diff --git a/jemalloc/test/unit/log.c b/jemalloc/test/unit/log.c
new file mode 100644
index 0000000..a52bd73
--- /dev/null
+++ b/jemalloc/test/unit/log.c
@@ -0,0 +1,193 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/log.h"
+
+static void
+expect_no_logging(const char *names) {
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	log_var_t log_l2 = LOG_VAR_INIT("l2");
+	log_var_t log_l2_a = LOG_VAR_INIT("l2.a");
+
+	strcpy(log_var_names, names);
+
+	int count = 0;
+
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+			count++;
+		log_do_end(log_l1)
+
+		log_do_begin(log_l2)
+			count++;
+		log_do_end(log_l2)
+
+		log_do_begin(log_l2_a)
+			count++;
+		log_do_end(log_l2_a)
+	}
+	assert_d_eq(count, 0, "Disabled logging not ignored!");
+}
+
+TEST_BEGIN(test_log_disabled) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	expect_no_logging("");
+	expect_no_logging("abc");
+	expect_no_logging("a.b.c");
+	expect_no_logging("l12");
+	expect_no_logging("l123|a456|b789");
+	expect_no_logging("|||");
+}
+TEST_END
+
+TEST_BEGIN(test_log_enabled_direct) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	log_var_t log_l1_a = LOG_VAR_INIT("l1.a");
+	log_var_t log_l2 = LOG_VAR_INIT("l2");
+
+	int count;
+
+	count = 0;
+	strcpy(log_var_names, "l1");
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+			count++;
+		log_do_end(log_l1)
+	}
+	assert_d_eq(count, 10, "Mis-logged!");
+
+	count = 0;
+	strcpy(log_var_names, "l1.a");
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1_a)
+			count++;
+		log_do_end(log_l1_a)
+	}
+	assert_d_eq(count, 10, "Mis-logged!");
+
+	count = 0;
+	strcpy(log_var_names, "l1.a|abc|l2|def");
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1_a)
+			count++;
+		log_do_end(log_l1_a)
+
+		log_do_begin(log_l2)
+			count++;
+		log_do_end(log_l2)
+	}
+	assert_d_eq(count, 20, "Mis-logged!");
+}
+TEST_END
+
+TEST_BEGIN(test_log_enabled_indirect) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	strcpy(log_var_names, "l0|l1|abc|l2.b|def");
+
+	/* On. */
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	/* Off. */
+	log_var_t log_l1a = LOG_VAR_INIT("l1a");
+	/* On. */
+	log_var_t log_l1_a = LOG_VAR_INIT("l1.a");
+	/* Off. */
+	log_var_t log_l2_a = LOG_VAR_INIT("l2.a");
+	/* On. */
+	log_var_t log_l2_b_a = LOG_VAR_INIT("l2.b.a");
+	/* On. */
+	log_var_t log_l2_b_b = LOG_VAR_INIT("l2.b.b");
+
+	/* 4 are on total, so should sum to 40. */
+	int count = 0;
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+			count++;
+		log_do_end(log_l1)
+
+		log_do_begin(log_l1a)
+			count++;
+		log_do_end(log_l1a)
+
+		log_do_begin(log_l1_a)
+			count++;
+		log_do_end(log_l1_a)
+
+		log_do_begin(log_l2_a)
+			count++;
+		log_do_end(log_l2_a)
+
+		log_do_begin(log_l2_b_a)
+			count++;
+		log_do_end(log_l2_b_a)
+
+		log_do_begin(log_l2_b_b)
+			count++;
+		log_do_end(log_l2_b_b)
+	}
+
+	assert_d_eq(count, 40, "Mis-logged!");
+}
+TEST_END
+
+TEST_BEGIN(test_log_enabled_global) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	strcpy(log_var_names, "abc|.|def");
+
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	log_var_t log_l2_a_a = LOG_VAR_INIT("l2.a.a");
+
+	int count = 0;
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+		    count++;
+		log_do_end(log_l1)
+
+		log_do_begin(log_l2_a_a)
+		    count++;
+		log_do_end(log_l2_a_a)
+	}
+	assert_d_eq(count, 20, "Mis-logged!");
+}
+TEST_END
+
+TEST_BEGIN(test_logs_if_no_init) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, false, ATOMIC_RELAXED);
+
+	log_var_t l = LOG_VAR_INIT("definitely.not.enabled");
+
+	int count = 0;
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(l)
+			count++;
+		log_do_end(l)
+	}
+	assert_d_eq(count, 0, "Logging shouldn't happen if not initialized.");
+}
+TEST_END
+
+/*
+ * This really just checks to make sure that this usage compiles; we don't have
+ * any test code to run.
+ */
+TEST_BEGIN(test_log_only_format_string) {
+	if (false) {
+		LOG("log_str", "No arguments follow this format string.");
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_log_disabled,
+	    test_log_enabled_direct,
+	    test_log_enabled_indirect,
+	    test_log_enabled_global,
+	    test_logs_if_no_init,
+	    test_log_only_format_string);
+}
diff --git a/jemalloc/test/unit/mallctl.c b/jemalloc/test/unit/mallctl.c
new file mode 100644
index 0000000..3a75ac0
--- /dev/null
+++ b/jemalloc/test/unit/mallctl.c
@@ -0,0 +1,888 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/util.h"
+
+TEST_BEGIN(test_mallctl_errors) {
+	uint64_t epoch;
+	size_t sz;
+
+	assert_d_eq(mallctl("no_such_name", NULL, NULL, NULL, 0), ENOENT,
+	    "mallctl() should return ENOENT for non-existent names");
+
+	assert_d_eq(mallctl("version", NULL, NULL, "0.0.0", strlen("0.0.0")),
+	    EPERM, "mallctl() should return EPERM on attempt to write "
+	    "read-only value");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)-1), EINVAL,
+	    "mallctl() should return EINVAL for input size mismatch");
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)+1), EINVAL,
+	    "mallctl() should return EINVAL for input size mismatch");
+
+	sz = sizeof(epoch)-1;
+	assert_d_eq(mallctl("epoch", (void *)&epoch, &sz, NULL, 0), EINVAL,
+	    "mallctl() should return EINVAL for output size mismatch");
+	sz = sizeof(epoch)+1;
+	assert_d_eq(mallctl("epoch", (void *)&epoch, &sz, NULL, 0), EINVAL,
+	    "mallctl() should return EINVAL for output size mismatch");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctlnametomib_errors) {
+	size_t mib[1];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("no_such_name", mib, &miblen), ENOENT,
+	    "mallctlnametomib() should return ENOENT for non-existent names");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctlbymib_errors) {
+	uint64_t epoch;
+	size_t sz;
+	size_t mib[1];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("version", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, "0.0.0",
+	    strlen("0.0.0")), EPERM, "mallctl() should return EPERM on "
+	    "attempt to write read-only value");
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("epoch", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)-1), EINVAL,
+	    "mallctlbymib() should return EINVAL for input size mismatch");
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)+1), EINVAL,
+	    "mallctlbymib() should return EINVAL for input size mismatch");
+
+	sz = sizeof(epoch)-1;
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&epoch, &sz, NULL, 0),
+	    EINVAL,
+	    "mallctlbymib() should return EINVAL for output size mismatch");
+	sz = sizeof(epoch)+1;
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&epoch, &sz, NULL, 0),
+	    EINVAL,
+	    "mallctlbymib() should return EINVAL for output size mismatch");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctl_read_write) {
+	uint64_t old_epoch, new_epoch;
+	size_t sz = sizeof(old_epoch);
+
+	/* Blind. */
+	assert_d_eq(mallctl("epoch", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+
+	/* Read. */
+	assert_d_eq(mallctl("epoch", (void *)&old_epoch, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+
+	/* Write. */
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&new_epoch,
+	    sizeof(new_epoch)), 0, "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+
+	/* Read+write. */
+	assert_d_eq(mallctl("epoch", (void *)&old_epoch, &sz,
+	    (void *)&new_epoch, sizeof(new_epoch)), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctlnametomib_short_mib) {
+	size_t mib[4];
+	size_t miblen;
+
+	miblen = 3;
+	mib[3] = 42;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.nregs", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	assert_zu_eq(miblen, 3, "Unexpected mib output length");
+	assert_zu_eq(mib[3], 42,
+	    "mallctlnametomib() wrote past the end of the input mib");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctl_config) {
+#define TEST_MALLCTL_CONFIG(config, t) do {				\
+	t oldval;							\
+	size_t sz = sizeof(oldval);					\
+	assert_d_eq(mallctl("config."#config, (void *)&oldval, &sz,	\
+	    NULL, 0), 0, "Unexpected mallctl() failure");		\
+	assert_b_eq(oldval, config_##config, "Incorrect config value");	\
+	assert_zu_eq(sz, sizeof(oldval), "Unexpected output size");	\
+} while (0)
+
+	TEST_MALLCTL_CONFIG(cache_oblivious, bool);
+	TEST_MALLCTL_CONFIG(debug, bool);
+	TEST_MALLCTL_CONFIG(fill, bool);
+	TEST_MALLCTL_CONFIG(lazy_lock, bool);
+	TEST_MALLCTL_CONFIG(malloc_conf, const char *);
+	TEST_MALLCTL_CONFIG(prof, bool);
+	TEST_MALLCTL_CONFIG(prof_libgcc, bool);
+	TEST_MALLCTL_CONFIG(prof_libunwind, bool);
+	TEST_MALLCTL_CONFIG(stats, bool);
+	TEST_MALLCTL_CONFIG(utrace, bool);
+	TEST_MALLCTL_CONFIG(xmalloc, bool);
+
+#undef TEST_MALLCTL_CONFIG
+}
+TEST_END
+
+TEST_BEGIN(test_mallctl_opt) {
+	bool config_always = true;
+
+#define TEST_MALLCTL_OPT(t, opt, config) do {				\
+	t oldval;							\
+	size_t sz = sizeof(oldval);					\
+	int expected = config_##config ? 0 : ENOENT;			\
+	int result = mallctl("opt."#opt, (void *)&oldval, &sz, NULL,	\
+	    0);								\
+	assert_d_eq(result, expected,					\
+	    "Unexpected mallctl() result for opt."#opt);		\
+	assert_zu_eq(sz, sizeof(oldval), "Unexpected output size");	\
+} while (0)
+
+	TEST_MALLCTL_OPT(bool, abort, always);
+	TEST_MALLCTL_OPT(bool, abort_conf, always);
+	TEST_MALLCTL_OPT(bool, confirm_conf, always);
+	TEST_MALLCTL_OPT(const char *, metadata_thp, always);
+	TEST_MALLCTL_OPT(bool, retain, always);
+	TEST_MALLCTL_OPT(const char *, dss, always);
+	TEST_MALLCTL_OPT(unsigned, narenas, always);
+	TEST_MALLCTL_OPT(const char *, percpu_arena, always);
+	TEST_MALLCTL_OPT(size_t, oversize_threshold, always);
+	TEST_MALLCTL_OPT(bool, background_thread, always);
+	TEST_MALLCTL_OPT(ssize_t, dirty_decay_ms, always);
+	TEST_MALLCTL_OPT(ssize_t, muzzy_decay_ms, always);
+	TEST_MALLCTL_OPT(bool, stats_print, always);
+	TEST_MALLCTL_OPT(const char *, junk, fill);
+	TEST_MALLCTL_OPT(bool, zero, fill);
+	TEST_MALLCTL_OPT(bool, utrace, utrace);
+	TEST_MALLCTL_OPT(bool, xmalloc, xmalloc);
+	TEST_MALLCTL_OPT(bool, tcache, always);
+	TEST_MALLCTL_OPT(size_t, lg_extent_max_active_fit, always);
+	TEST_MALLCTL_OPT(size_t, lg_tcache_max, always);
+	TEST_MALLCTL_OPT(const char *, thp, always);
+	TEST_MALLCTL_OPT(bool, prof, prof);
+	TEST_MALLCTL_OPT(const char *, prof_prefix, prof);
+	TEST_MALLCTL_OPT(bool, prof_active, prof);
+	TEST_MALLCTL_OPT(ssize_t, lg_prof_sample, prof);
+	TEST_MALLCTL_OPT(bool, prof_accum, prof);
+	TEST_MALLCTL_OPT(ssize_t, lg_prof_interval, prof);
+	TEST_MALLCTL_OPT(bool, prof_gdump, prof);
+	TEST_MALLCTL_OPT(bool, prof_final, prof);
+	TEST_MALLCTL_OPT(bool, prof_leak, prof);
+
+#undef TEST_MALLCTL_OPT
+}
+TEST_END
+
+TEST_BEGIN(test_manpage_example) {
+	unsigned nbins, i;
+	size_t mib[4];
+	size_t len, miblen;
+
+	len = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &len, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	miblen = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	for (i = 0; i < nbins; i++) {
+		size_t bin_size;
+
+		mib[2] = i;
+		len = sizeof(bin_size);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&bin_size, &len,
+		    NULL, 0), 0, "Unexpected mallctlbymib() failure");
+		/* Do something with bin_size... */
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_tcache_none) {
+	test_skip_if(!opt_tcache);
+
+	/* Allocate p and q. */
+	void *p0 = mallocx(42, 0);
+	assert_ptr_not_null(p0, "Unexpected mallocx() failure");
+	void *q = mallocx(42, 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+
+	/* Deallocate p and q, but bypass the tcache for q. */
+	dallocx(p0, 0);
+	dallocx(q, MALLOCX_TCACHE_NONE);
+
+	/* Make sure that tcache-based allocation returns p, not q. */
+	void *p1 = mallocx(42, 0);
+	assert_ptr_not_null(p1, "Unexpected mallocx() failure");
+	assert_ptr_eq(p0, p1, "Expected tcache to allocate cached region");
+
+	/* Clean up. */
+	dallocx(p1, MALLOCX_TCACHE_NONE);
+}
+TEST_END
+
+TEST_BEGIN(test_tcache) {
+#define NTCACHES	10
+	unsigned tis[NTCACHES];
+	void *ps[NTCACHES];
+	void *qs[NTCACHES];
+	unsigned i;
+	size_t sz, psz, qsz;
+
+	psz = 42;
+	qsz = nallocx(psz, 0) + 1;
+
+	/* Create tcaches. */
+	for (i = 0; i < NTCACHES; i++) {
+		sz = sizeof(unsigned);
+		assert_d_eq(mallctl("tcache.create", (void *)&tis[i], &sz, NULL,
+		    0), 0, "Unexpected mallctl() failure, i=%u", i);
+	}
+
+	/* Exercise tcache ID recycling. */
+	for (i = 0; i < NTCACHES; i++) {
+		assert_d_eq(mallctl("tcache.destroy", NULL, NULL,
+		    (void *)&tis[i], sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure, i=%u", i);
+	}
+	for (i = 0; i < NTCACHES; i++) {
+		sz = sizeof(unsigned);
+		assert_d_eq(mallctl("tcache.create", (void *)&tis[i], &sz, NULL,
+		    0), 0, "Unexpected mallctl() failure, i=%u", i);
+	}
+
+	/* Flush empty tcaches. */
+	for (i = 0; i < NTCACHES; i++) {
+		assert_d_eq(mallctl("tcache.flush", NULL, NULL, (void *)&tis[i],
+		    sizeof(unsigned)), 0, "Unexpected mallctl() failure, i=%u",
+		    i);
+	}
+
+	/* Cache some allocations. */
+	for (i = 0; i < NTCACHES; i++) {
+		ps[i] = mallocx(psz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(ps[i], "Unexpected mallocx() failure, i=%u",
+		    i);
+		dallocx(ps[i], MALLOCX_TCACHE(tis[i]));
+
+		qs[i] = mallocx(qsz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(qs[i], "Unexpected mallocx() failure, i=%u",
+		    i);
+		dallocx(qs[i], MALLOCX_TCACHE(tis[i]));
+	}
+
+	/* Verify that tcaches allocate cached regions. */
+	for (i = 0; i < NTCACHES; i++) {
+		void *p0 = ps[i];
+		ps[i] = mallocx(psz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(ps[i], "Unexpected mallocx() failure, i=%u",
+		    i);
+		assert_ptr_eq(ps[i], p0,
+		    "Expected mallocx() to allocate cached region, i=%u", i);
+	}
+
+	/* Verify that reallocation uses cached regions. */
+	for (i = 0; i < NTCACHES; i++) {
+		void *q0 = qs[i];
+		qs[i] = rallocx(ps[i], qsz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(qs[i], "Unexpected rallocx() failure, i=%u",
+		    i);
+		assert_ptr_eq(qs[i], q0,
+		    "Expected rallocx() to allocate cached region, i=%u", i);
+		/* Avoid undefined behavior in case of test failure. */
+		if (qs[i] == NULL) {
+			qs[i] = ps[i];
+		}
+	}
+	for (i = 0; i < NTCACHES; i++) {
+		dallocx(qs[i], MALLOCX_TCACHE(tis[i]));
+	}
+
+	/* Flush some non-empty tcaches. */
+	for (i = 0; i < NTCACHES/2; i++) {
+		assert_d_eq(mallctl("tcache.flush", NULL, NULL, (void *)&tis[i],
+		    sizeof(unsigned)), 0, "Unexpected mallctl() failure, i=%u",
+		    i);
+	}
+
+	/* Destroy tcaches. */
+	for (i = 0; i < NTCACHES; i++) {
+		assert_d_eq(mallctl("tcache.destroy", NULL, NULL,
+		    (void *)&tis[i], sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure, i=%u", i);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_thread_arena) {
+	unsigned old_arena_ind, new_arena_ind, narenas;
+
+	const char *opa;
+	size_t sz = sizeof(opa);
+	assert_d_eq(mallctl("opt.percpu_arena", (void *)&opa, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	if (opt_oversize_threshold != 0) {
+		narenas--;
+	}
+	assert_u_eq(narenas, opt_narenas, "Number of arenas incorrect");
+
+	if (strcmp(opa, "disabled") == 0) {
+		new_arena_ind = narenas - 1;
+		assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+		    (void *)&new_arena_ind, sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure");
+		new_arena_ind = 0;
+		assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+		    (void *)&new_arena_ind, sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure");
+	} else {
+		assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+		    NULL, 0), 0, "Unexpected mallctl() failure");
+		new_arena_ind = percpu_arena_ind_limit(opt_percpu_arena) - 1;
+		if (old_arena_ind != new_arena_ind) {
+			assert_d_eq(mallctl("thread.arena",
+			    (void *)&old_arena_ind, &sz, (void *)&new_arena_ind,
+			    sizeof(unsigned)), EPERM, "thread.arena ctl "
+			    "should not be allowed with percpu arena");
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_initialized) {
+	unsigned narenas, i;
+	size_t sz;
+	size_t mib[3];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+	bool initialized;
+
+	sz = sizeof(narenas);
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctlnametomib("arena.0.initialized", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	for (i = 0; i < narenas; i++) {
+		mib[1] = i;
+		sz = sizeof(initialized);
+		assert_d_eq(mallctlbymib(mib, miblen, &initialized, &sz, NULL,
+		    0), 0, "Unexpected mallctl() failure");
+	}
+
+	mib[1] = MALLCTL_ARENAS_ALL;
+	sz = sizeof(initialized);
+	assert_d_eq(mallctlbymib(mib, miblen, &initialized, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_true(initialized,
+	    "Merged arena statistics should always be initialized");
+
+	/* Equivalent to the above but using mallctl() directly. */
+	sz = sizeof(initialized);
+	assert_d_eq(mallctl(
+	    "arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".initialized",
+	    (void *)&initialized, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_true(initialized,
+	    "Merged arena statistics should always be initialized");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_dirty_decay_ms) {
+	ssize_t dirty_decay_ms, orig_dirty_decay_ms, prev_dirty_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arena.0.dirty_decay_ms",
+	    (void *)&orig_dirty_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	dirty_decay_ms = -2;
+	assert_d_eq(mallctl("arena.0.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	dirty_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arena.0.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), 0,
+	    "Unexpected mallctl() failure");
+
+	for (prev_dirty_decay_ms = dirty_decay_ms, dirty_decay_ms = -1;
+	    dirty_decay_ms < 20; prev_dirty_decay_ms = dirty_decay_ms,
+	    dirty_decay_ms++) {
+		ssize_t old_dirty_decay_ms;
+
+		assert_d_eq(mallctl("arena.0.dirty_decay_ms",
+		    (void *)&old_dirty_decay_ms, &sz, (void *)&dirty_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_dirty_decay_ms, prev_dirty_decay_ms,
+		    "Unexpected old arena.0.dirty_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_muzzy_decay_ms) {
+	ssize_t muzzy_decay_ms, orig_muzzy_decay_ms, prev_muzzy_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arena.0.muzzy_decay_ms",
+	    (void *)&orig_muzzy_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	muzzy_decay_ms = -2;
+	assert_d_eq(mallctl("arena.0.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	muzzy_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arena.0.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), 0,
+	    "Unexpected mallctl() failure");
+
+	for (prev_muzzy_decay_ms = muzzy_decay_ms, muzzy_decay_ms = -1;
+	    muzzy_decay_ms < 20; prev_muzzy_decay_ms = muzzy_decay_ms,
+	    muzzy_decay_ms++) {
+		ssize_t old_muzzy_decay_ms;
+
+		assert_d_eq(mallctl("arena.0.muzzy_decay_ms",
+		    (void *)&old_muzzy_decay_ms, &sz, (void *)&muzzy_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_muzzy_decay_ms, prev_muzzy_decay_ms,
+		    "Unexpected old arena.0.muzzy_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_purge) {
+	unsigned narenas;
+	size_t sz = sizeof(unsigned);
+	size_t mib[3];
+	size_t miblen = 3;
+
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlnametomib("arena.0.purge", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = narenas;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	mib[1] = MALLCTL_ARENAS_ALL;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_decay) {
+	unsigned narenas;
+	size_t sz = sizeof(unsigned);
+	size_t mib[3];
+	size_t miblen = 3;
+
+	assert_d_eq(mallctl("arena.0.decay", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlnametomib("arena.0.decay", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = narenas;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	mib[1] = MALLCTL_ARENAS_ALL;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_dss) {
+	const char *dss_prec_old, *dss_prec_new;
+	size_t sz = sizeof(dss_prec_old);
+	size_t mib[3];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.dss", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() error");
+
+	dss_prec_new = "disabled";
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz,
+	    (void *)&dss_prec_new, sizeof(dss_prec_new)), 0,
+	    "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected default for dss precedence");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_new, &sz,
+	    (void *)&dss_prec_old, sizeof(dss_prec_old)), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected value for dss precedence");
+
+	mib[1] = narenas_total_get();
+	dss_prec_new = "disabled";
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz,
+	    (void *)&dss_prec_new, sizeof(dss_prec_new)), 0,
+	    "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected default for dss precedence");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_new, &sz,
+	    (void *)&dss_prec_old, sizeof(dss_prec_new)), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected value for dss precedence");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_retain_grow_limit) {
+	size_t old_limit, new_limit, default_limit;
+	size_t mib[3];
+	size_t miblen;
+
+	bool retain_enabled;
+	size_t sz = sizeof(retain_enabled);
+	assert_d_eq(mallctl("opt.retain", &retain_enabled, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	test_skip_if(!retain_enabled);
+
+	sz = sizeof(default_limit);
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.retain_grow_limit", mib, &miblen),
+	    0, "Unexpected mallctlnametomib() error");
+
+	assert_d_eq(mallctlbymib(mib, miblen, &default_limit, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(default_limit, SC_LARGE_MAXCLASS,
+	    "Unexpected default for retain_grow_limit");
+
+	new_limit = PAGE - 1;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &new_limit,
+	    sizeof(new_limit)), EFAULT, "Unexpected mallctl() success");
+
+	new_limit = PAGE + 1;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &new_limit,
+	    sizeof(new_limit)), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlbymib(mib, miblen, &old_limit, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(old_limit, PAGE,
+	    "Unexpected value for retain_grow_limit");
+
+	/* Expect grow less than psize class 10. */
+	new_limit = sz_pind2sz(10) - 1;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &new_limit,
+	    sizeof(new_limit)), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlbymib(mib, miblen, &old_limit, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(old_limit, sz_pind2sz(9),
+	    "Unexpected value for retain_grow_limit");
+
+	/* Restore to default. */
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &default_limit,
+	    sizeof(default_limit)), 0, "Unexpected mallctl() failure");
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_dirty_decay_ms) {
+	ssize_t dirty_decay_ms, orig_dirty_decay_ms, prev_dirty_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arenas.dirty_decay_ms",
+	    (void *)&orig_dirty_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	dirty_decay_ms = -2;
+	assert_d_eq(mallctl("arenas.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	dirty_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arenas.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), 0,
+	    "Expected mallctl() failure");
+
+	for (prev_dirty_decay_ms = dirty_decay_ms, dirty_decay_ms = -1;
+	    dirty_decay_ms < 20; prev_dirty_decay_ms = dirty_decay_ms,
+	    dirty_decay_ms++) {
+		ssize_t old_dirty_decay_ms;
+
+		assert_d_eq(mallctl("arenas.dirty_decay_ms",
+		    (void *)&old_dirty_decay_ms, &sz, (void *)&dirty_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_dirty_decay_ms, prev_dirty_decay_ms,
+		    "Unexpected old arenas.dirty_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_muzzy_decay_ms) {
+	ssize_t muzzy_decay_ms, orig_muzzy_decay_ms, prev_muzzy_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arenas.muzzy_decay_ms",
+	    (void *)&orig_muzzy_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	muzzy_decay_ms = -2;
+	assert_d_eq(mallctl("arenas.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	muzzy_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arenas.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), 0,
+	    "Expected mallctl() failure");
+
+	for (prev_muzzy_decay_ms = muzzy_decay_ms, muzzy_decay_ms = -1;
+	    muzzy_decay_ms < 20; prev_muzzy_decay_ms = muzzy_decay_ms,
+	    muzzy_decay_ms++) {
+		ssize_t old_muzzy_decay_ms;
+
+		assert_d_eq(mallctl("arenas.muzzy_decay_ms",
+		    (void *)&old_muzzy_decay_ms, &sz, (void *)&muzzy_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_muzzy_decay_ms, prev_muzzy_decay_ms,
+		    "Unexpected old arenas.muzzy_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_constants) {
+#define TEST_ARENAS_CONSTANT(t, name, expected) do {			\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("arenas."#name, (void *)&name, &sz, NULL,	\
+	    0), 0, "Unexpected mallctl() failure");			\
+	assert_zu_eq(name, expected, "Incorrect "#name" size");		\
+} while (0)
+
+	TEST_ARENAS_CONSTANT(size_t, quantum, QUANTUM);
+	TEST_ARENAS_CONSTANT(size_t, page, PAGE);
+	TEST_ARENAS_CONSTANT(unsigned, nbins, SC_NBINS);
+	TEST_ARENAS_CONSTANT(unsigned, nlextents, SC_NSIZES - SC_NBINS);
+
+#undef TEST_ARENAS_CONSTANT
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_bin_constants) {
+#define TEST_ARENAS_BIN_CONSTANT(t, name, expected) do {		\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("arenas.bin.0."#name, (void *)&name, &sz,	\
+	    NULL, 0), 0, "Unexpected mallctl() failure");		\
+	assert_zu_eq(name, expected, "Incorrect "#name" size");		\
+} while (0)
+
+	TEST_ARENAS_BIN_CONSTANT(size_t, size, bin_infos[0].reg_size);
+	TEST_ARENAS_BIN_CONSTANT(uint32_t, nregs, bin_infos[0].nregs);
+	TEST_ARENAS_BIN_CONSTANT(size_t, slab_size,
+	    bin_infos[0].slab_size);
+	TEST_ARENAS_BIN_CONSTANT(uint32_t, nshards, bin_infos[0].n_shards);
+
+#undef TEST_ARENAS_BIN_CONSTANT
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_lextent_constants) {
+#define TEST_ARENAS_LEXTENT_CONSTANT(t, name, expected) do {		\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("arenas.lextent.0."#name, (void *)&name,	\
+	    &sz, NULL, 0), 0, "Unexpected mallctl() failure");		\
+	assert_zu_eq(name, expected, "Incorrect "#name" size");		\
+} while (0)
+
+	TEST_ARENAS_LEXTENT_CONSTANT(size_t, size,
+	    SC_LARGE_MINCLASS);
+
+#undef TEST_ARENAS_LEXTENT_CONSTANT
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_create) {
+	unsigned narenas_before, arena, narenas_after;
+	size_t sz = sizeof(unsigned);
+
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas_before, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas_after, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+
+	assert_u_eq(narenas_before+1, narenas_after,
+	    "Unexpected number of arenas before versus after extension");
+	assert_u_eq(arena, narenas_after-1, "Unexpected arena index");
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_lookup) {
+	unsigned arena, arena1;
+	void *ptr;
+	size_t sz = sizeof(unsigned);
+
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	ptr = mallocx(42, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE);
+	assert_ptr_not_null(ptr, "Unexpected mallocx() failure");
+	assert_d_eq(mallctl("arenas.lookup", &arena1, &sz, &ptr, sizeof(ptr)),
+	    0, "Unexpected mallctl() failure");
+	assert_u_eq(arena, arena1, "Unexpected arena index");
+	dallocx(ptr, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas) {
+#define TEST_STATS_ARENAS(t, name) do {					\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("stats.arenas.0."#name, (void *)&name, &sz,	\
+	    NULL, 0), 0, "Unexpected mallctl() failure");		\
+} while (0)
+
+	TEST_STATS_ARENAS(unsigned, nthreads);
+	TEST_STATS_ARENAS(const char *, dss);
+	TEST_STATS_ARENAS(ssize_t, dirty_decay_ms);
+	TEST_STATS_ARENAS(ssize_t, muzzy_decay_ms);
+	TEST_STATS_ARENAS(size_t, pactive);
+	TEST_STATS_ARENAS(size_t, pdirty);
+
+#undef TEST_STATS_ARENAS
+}
+TEST_END
+
+static void
+alloc_hook(void *extra, UNUSED hook_alloc_t type, UNUSED void *result,
+    UNUSED uintptr_t result_raw, UNUSED uintptr_t args_raw[3]) {
+	*(bool *)extra = true;
+}
+
+static void
+dalloc_hook(void *extra, UNUSED hook_dalloc_t type,
+    UNUSED void *address, UNUSED uintptr_t args_raw[3]) {
+	*(bool *)extra = true;
+}
+
+TEST_BEGIN(test_hooks) {
+	bool hook_called = false;
+	hooks_t hooks = {&alloc_hook, &dalloc_hook, NULL, &hook_called};
+	void *handle = NULL;
+	size_t sz = sizeof(handle);
+	int err = mallctl("experimental.hooks.install", &handle, &sz, &hooks,
+	    sizeof(hooks));
+	assert_d_eq(err, 0, "Hook installation failed");
+	assert_ptr_ne(handle, NULL, "Hook installation gave null handle");
+	void *ptr = mallocx(1, 0);
+	assert_true(hook_called, "Alloc hook not called");
+	hook_called = false;
+	free(ptr);
+	assert_true(hook_called, "Free hook not called");
+
+	err = mallctl("experimental.hooks.remove", NULL, NULL, &handle,
+	    sizeof(handle));
+	assert_d_eq(err, 0, "Hook removal failed");
+	hook_called = false;
+	ptr = mallocx(1, 0);
+	free(ptr);
+	assert_false(hook_called, "Hook called after removal");
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_exhaustion) {
+	bool hook_called = false;
+	hooks_t hooks = {&alloc_hook, &dalloc_hook, NULL, &hook_called};
+
+	void *handle;
+	void *handles[HOOK_MAX];
+	size_t sz = sizeof(handle);
+	int err;
+	for (int i = 0; i < HOOK_MAX; i++) {
+		handle = NULL;
+		err = mallctl("experimental.hooks.install", &handle, &sz,
+		    &hooks, sizeof(hooks));
+		assert_d_eq(err, 0, "Error installation hooks");
+		assert_ptr_ne(handle, NULL, "Got NULL handle");
+		handles[i] = handle;
+	}
+	err = mallctl("experimental.hooks.install", &handle, &sz, &hooks,
+	    sizeof(hooks));
+	assert_d_eq(err, EAGAIN, "Should have failed hook installation");
+	for (int i = 0; i < HOOK_MAX; i++) {
+		err = mallctl("experimental.hooks.remove", NULL, NULL,
+		    &handles[i], sizeof(handles[i]));
+		assert_d_eq(err, 0, "Hook removal failed");
+	}
+	/* Insertion failed, but then we removed some; it should work now. */
+	handle = NULL;
+	err = mallctl("experimental.hooks.install", &handle, &sz, &hooks,
+	    sizeof(hooks));
+	assert_d_eq(err, 0, "Hook insertion failed");
+	assert_ptr_ne(handle, NULL, "Got NULL handle");
+	err = mallctl("experimental.hooks.remove", NULL, NULL, &handle,
+	    sizeof(handle));
+	assert_d_eq(err, 0, "Hook removal failed");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_mallctl_errors,
+	    test_mallctlnametomib_errors,
+	    test_mallctlbymib_errors,
+	    test_mallctl_read_write,
+	    test_mallctlnametomib_short_mib,
+	    test_mallctl_config,
+	    test_mallctl_opt,
+	    test_manpage_example,
+	    test_tcache_none,
+	    test_tcache,
+	    test_thread_arena,
+	    test_arena_i_initialized,
+	    test_arena_i_dirty_decay_ms,
+	    test_arena_i_muzzy_decay_ms,
+	    test_arena_i_purge,
+	    test_arena_i_decay,
+	    test_arena_i_dss,
+	    test_arena_i_retain_grow_limit,
+	    test_arenas_dirty_decay_ms,
+	    test_arenas_muzzy_decay_ms,
+	    test_arenas_constants,
+	    test_arenas_bin_constants,
+	    test_arenas_lextent_constants,
+	    test_arenas_create,
+	    test_arenas_lookup,
+	    test_stats_arenas,
+	    test_hooks,
+	    test_hooks_exhaustion);
+}
diff --git a/jemalloc/test/unit/malloc_io.c b/jemalloc/test/unit/malloc_io.c
new file mode 100644
index 0000000..79ba7fc
--- /dev/null
+++ b/jemalloc/test/unit/malloc_io.c
@@ -0,0 +1,258 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_malloc_strtoumax_no_endptr) {
+	int err;
+
+	set_errno(0);
+	assert_ju_eq(malloc_strtoumax("0", NULL, 0), 0, "Unexpected result");
+	err = get_errno();
+	assert_d_eq(err, 0, "Unexpected failure");
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_strtoumax) {
+	struct test_s {
+		const char *input;
+		const char *expected_remainder;
+		int base;
+		int expected_errno;
+		const char *expected_errno_name;
+		uintmax_t expected_x;
+	};
+#define ERR(e)		e, #e
+#define KUMAX(x)	((uintmax_t)x##ULL)
+#define KSMAX(x)	((uintmax_t)(intmax_t)x##LL)
+	struct test_s tests[] = {
+		{"0",		"0",	-1,	ERR(EINVAL),	UINTMAX_MAX},
+		{"0",		"0",	1,	ERR(EINVAL),	UINTMAX_MAX},
+		{"0",		"0",	37,	ERR(EINVAL),	UINTMAX_MAX},
+
+		{"",		"",	0,	ERR(EINVAL),	UINTMAX_MAX},
+		{"+",		"+",	0,	ERR(EINVAL),	UINTMAX_MAX},
+		{"++3",		"++3",	0,	ERR(EINVAL),	UINTMAX_MAX},
+		{"-",		"-",	0,	ERR(EINVAL),	UINTMAX_MAX},
+
+		{"42",		"",	0,	ERR(0),		KUMAX(42)},
+		{"+42",		"",	0,	ERR(0),		KUMAX(42)},
+		{"-42",		"",	0,	ERR(0),		KSMAX(-42)},
+		{"042",		"",	0,	ERR(0),		KUMAX(042)},
+		{"+042",	"",	0,	ERR(0),		KUMAX(042)},
+		{"-042",	"",	0,	ERR(0),		KSMAX(-042)},
+		{"0x42",	"",	0,	ERR(0),		KUMAX(0x42)},
+		{"+0x42",	"",	0,	ERR(0),		KUMAX(0x42)},
+		{"-0x42",	"",	0,	ERR(0),		KSMAX(-0x42)},
+
+		{"0",		"",	0,	ERR(0),		KUMAX(0)},
+		{"1",		"",	0,	ERR(0),		KUMAX(1)},
+
+		{"42",		"",	0,	ERR(0),		KUMAX(42)},
+		{" 42",		"",	0,	ERR(0),		KUMAX(42)},
+		{"42 ",		" ",	0,	ERR(0),		KUMAX(42)},
+		{"0x",		"x",	0,	ERR(0),		KUMAX(0)},
+		{"42x",		"x",	0,	ERR(0),		KUMAX(42)},
+
+		{"07",		"",	0,	ERR(0),		KUMAX(7)},
+		{"010",		"",	0,	ERR(0),		KUMAX(8)},
+		{"08",		"8",	0,	ERR(0),		KUMAX(0)},
+		{"0_",		"_",	0,	ERR(0),		KUMAX(0)},
+
+		{"0x",		"x",	0,	ERR(0),		KUMAX(0)},
+		{"0X",		"X",	0,	ERR(0),		KUMAX(0)},
+		{"0xg",		"xg",	0,	ERR(0),		KUMAX(0)},
+		{"0XA",		"",	0,	ERR(0),		KUMAX(10)},
+
+		{"010",		"",	10,	ERR(0),		KUMAX(10)},
+		{"0x3",		"x3",	10,	ERR(0),		KUMAX(0)},
+
+		{"12",		"2",	2,	ERR(0),		KUMAX(1)},
+		{"78",		"8",	8,	ERR(0),		KUMAX(7)},
+		{"9a",		"a",	10,	ERR(0),		KUMAX(9)},
+		{"9A",		"A",	10,	ERR(0),		KUMAX(9)},
+		{"fg",		"g",	16,	ERR(0),		KUMAX(15)},
+		{"FG",		"G",	16,	ERR(0),		KUMAX(15)},
+		{"0xfg",	"g",	16,	ERR(0),		KUMAX(15)},
+		{"0XFG",	"G",	16,	ERR(0),		KUMAX(15)},
+		{"z_",		"_",	36,	ERR(0),		KUMAX(35)},
+		{"Z_",		"_",	36,	ERR(0),		KUMAX(35)}
+	};
+#undef ERR
+#undef KUMAX
+#undef KSMAX
+	unsigned i;
+
+	for (i = 0; i < sizeof(tests)/sizeof(struct test_s); i++) {
+		struct test_s *test = &tests[i];
+		int err;
+		uintmax_t result;
+		char *remainder;
+
+		set_errno(0);
+		result = malloc_strtoumax(test->input, &remainder, test->base);
+		err = get_errno();
+		assert_d_eq(err, test->expected_errno,
+		    "Expected errno %s for \"%s\", base %d",
+		    test->expected_errno_name, test->input, test->base);
+		assert_str_eq(remainder, test->expected_remainder,
+		    "Unexpected remainder for \"%s\", base %d",
+		    test->input, test->base);
+		if (err == 0) {
+			assert_ju_eq(result, test->expected_x,
+			    "Unexpected result for \"%s\", base %d",
+			    test->input, test->base);
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_snprintf_truncated) {
+#define BUFLEN	15
+	char buf[BUFLEN];
+	size_t result;
+	size_t len;
+#define TEST(expected_str_untruncated, ...) do {			\
+	result = malloc_snprintf(buf, len, __VA_ARGS__);		\
+	assert_d_eq(strncmp(buf, expected_str_untruncated, len-1), 0,	\
+	    "Unexpected string inequality (\"%s\" vs \"%s\")",		\
+	    buf, expected_str_untruncated);				\
+	assert_zu_eq(result, strlen(expected_str_untruncated),		\
+	    "Unexpected result");					\
+} while (0)
+
+	for (len = 1; len < BUFLEN; len++) {
+		TEST("012346789",	"012346789");
+		TEST("a0123b",		"a%sb", "0123");
+		TEST("a01234567",	"a%s%s", "0123", "4567");
+		TEST("a0123  ",		"a%-6s", "0123");
+		TEST("a  0123",		"a%6s", "0123");
+		TEST("a   012",		"a%6.3s", "0123");
+		TEST("a   012",		"a%*.*s", 6, 3, "0123");
+		TEST("a 123b",		"a% db", 123);
+		TEST("a123b",		"a%-db", 123);
+		TEST("a-123b",		"a%-db", -123);
+		TEST("a+123b",		"a%+db", 123);
+	}
+#undef BUFLEN
+#undef TEST
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_snprintf) {
+#define BUFLEN	128
+	char buf[BUFLEN];
+	size_t result;
+#define TEST(expected_str, ...) do {					\
+	result = malloc_snprintf(buf, sizeof(buf), __VA_ARGS__);	\
+	assert_str_eq(buf, expected_str, "Unexpected output");		\
+	assert_zu_eq(result, strlen(expected_str), "Unexpected result");\
+} while (0)
+
+	TEST("hello", "hello");
+
+	TEST("50%, 100%", "50%%, %d%%", 100);
+
+	TEST("a0123b", "a%sb", "0123");
+
+	TEST("a 0123b", "a%5sb", "0123");
+	TEST("a 0123b", "a%*sb", 5, "0123");
+
+	TEST("a0123 b", "a%-5sb", "0123");
+	TEST("a0123b", "a%*sb", -1, "0123");
+	TEST("a0123 b", "a%*sb", -5, "0123");
+	TEST("a0123 b", "a%-*sb", -5, "0123");
+
+	TEST("a012b", "a%.3sb", "0123");
+	TEST("a012b", "a%.*sb", 3, "0123");
+	TEST("a0123b", "a%.*sb", -3, "0123");
+
+	TEST("a  012b", "a%5.3sb", "0123");
+	TEST("a  012b", "a%5.*sb", 3, "0123");
+	TEST("a  012b", "a%*.3sb", 5, "0123");
+	TEST("a  012b", "a%*.*sb", 5, 3, "0123");
+	TEST("a 0123b", "a%*.*sb", 5, -3, "0123");
+
+	TEST("_abcd_", "_%x_", 0xabcd);
+	TEST("_0xabcd_", "_%#x_", 0xabcd);
+	TEST("_1234_", "_%o_", 01234);
+	TEST("_01234_", "_%#o_", 01234);
+	TEST("_1234_", "_%u_", 1234);
+
+	TEST("_1234_", "_%d_", 1234);
+	TEST("_ 1234_", "_% d_", 1234);
+	TEST("_+1234_", "_%+d_", 1234);
+	TEST("_-1234_", "_%d_", -1234);
+	TEST("_-1234_", "_% d_", -1234);
+	TEST("_-1234_", "_%+d_", -1234);
+
+	TEST("_-1234_", "_%d_", -1234);
+	TEST("_1234_", "_%d_", 1234);
+	TEST("_-1234_", "_%i_", -1234);
+	TEST("_1234_", "_%i_", 1234);
+	TEST("_01234_", "_%#o_", 01234);
+	TEST("_1234_", "_%u_", 1234);
+	TEST("_0x1234abc_", "_%#x_", 0x1234abc);
+	TEST("_0X1234ABC_", "_%#X_", 0x1234abc);
+	TEST("_c_", "_%c_", 'c');
+	TEST("_string_", "_%s_", "string");
+	TEST("_0x42_", "_%p_", ((void *)0x42));
+
+	TEST("_-1234_", "_%ld_", ((long)-1234));
+	TEST("_1234_", "_%ld_", ((long)1234));
+	TEST("_-1234_", "_%li_", ((long)-1234));
+	TEST("_1234_", "_%li_", ((long)1234));
+	TEST("_01234_", "_%#lo_", ((long)01234));
+	TEST("_1234_", "_%lu_", ((long)1234));
+	TEST("_0x1234abc_", "_%#lx_", ((long)0x1234abc));
+	TEST("_0X1234ABC_", "_%#lX_", ((long)0x1234ABC));
+
+	TEST("_-1234_", "_%lld_", ((long long)-1234));
+	TEST("_1234_", "_%lld_", ((long long)1234));
+	TEST("_-1234_", "_%lli_", ((long long)-1234));
+	TEST("_1234_", "_%lli_", ((long long)1234));
+	TEST("_01234_", "_%#llo_", ((long long)01234));
+	TEST("_1234_", "_%llu_", ((long long)1234));
+	TEST("_0x1234abc_", "_%#llx_", ((long long)0x1234abc));
+	TEST("_0X1234ABC_", "_%#llX_", ((long long)0x1234ABC));
+
+	TEST("_-1234_", "_%qd_", ((long long)-1234));
+	TEST("_1234_", "_%qd_", ((long long)1234));
+	TEST("_-1234_", "_%qi_", ((long long)-1234));
+	TEST("_1234_", "_%qi_", ((long long)1234));
+	TEST("_01234_", "_%#qo_", ((long long)01234));
+	TEST("_1234_", "_%qu_", ((long long)1234));
+	TEST("_0x1234abc_", "_%#qx_", ((long long)0x1234abc));
+	TEST("_0X1234ABC_", "_%#qX_", ((long long)0x1234ABC));
+
+	TEST("_-1234_", "_%jd_", ((intmax_t)-1234));
+	TEST("_1234_", "_%jd_", ((intmax_t)1234));
+	TEST("_-1234_", "_%ji_", ((intmax_t)-1234));
+	TEST("_1234_", "_%ji_", ((intmax_t)1234));
+	TEST("_01234_", "_%#jo_", ((intmax_t)01234));
+	TEST("_1234_", "_%ju_", ((intmax_t)1234));
+	TEST("_0x1234abc_", "_%#jx_", ((intmax_t)0x1234abc));
+	TEST("_0X1234ABC_", "_%#jX_", ((intmax_t)0x1234ABC));
+
+	TEST("_1234_", "_%td_", ((ptrdiff_t)1234));
+	TEST("_-1234_", "_%td_", ((ptrdiff_t)-1234));
+	TEST("_1234_", "_%ti_", ((ptrdiff_t)1234));
+	TEST("_-1234_", "_%ti_", ((ptrdiff_t)-1234));
+
+	TEST("_-1234_", "_%zd_", ((ssize_t)-1234));
+	TEST("_1234_", "_%zd_", ((ssize_t)1234));
+	TEST("_-1234_", "_%zi_", ((ssize_t)-1234));
+	TEST("_1234_", "_%zi_", ((ssize_t)1234));
+	TEST("_01234_", "_%#zo_", ((ssize_t)01234));
+	TEST("_1234_", "_%zu_", ((ssize_t)1234));
+	TEST("_0x1234abc_", "_%#zx_", ((ssize_t)0x1234abc));
+	TEST("_0X1234ABC_", "_%#zX_", ((ssize_t)0x1234ABC));
+#undef BUFLEN
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_malloc_strtoumax_no_endptr,
+	    test_malloc_strtoumax,
+	    test_malloc_snprintf_truncated,
+	    test_malloc_snprintf);
+}
diff --git a/jemalloc/test/unit/math.c b/jemalloc/test/unit/math.c
new file mode 100644
index 0000000..09ef20c
--- /dev/null
+++ b/jemalloc/test/unit/math.c
@@ -0,0 +1,390 @@
+#include "test/jemalloc_test.h"
+
+#define MAX_REL_ERR 1.0e-9
+#define MAX_ABS_ERR 1.0e-9
+
+#include <float.h>
+
+#ifdef __PGI
+#undef INFINITY
+#endif
+
+#ifndef INFINITY
+#define INFINITY (DBL_MAX + DBL_MAX)
+#endif
+
+static bool
+double_eq_rel(double a, double b, double max_rel_err, double max_abs_err) {
+	double rel_err;
+
+	if (fabs(a - b) < max_abs_err) {
+		return true;
+	}
+	rel_err = (fabs(b) > fabs(a)) ? fabs((a-b)/b) : fabs((a-b)/a);
+	return (rel_err < max_rel_err);
+}
+
+static uint64_t
+factorial(unsigned x) {
+	uint64_t ret = 1;
+	unsigned i;
+
+	for (i = 2; i <= x; i++) {
+		ret *= (uint64_t)i;
+	}
+
+	return ret;
+}
+
+TEST_BEGIN(test_ln_gamma_factorial) {
+	unsigned x;
+
+	/* exp(ln_gamma(x)) == (x-1)! for integer x. */
+	for (x = 1; x <= 21; x++) {
+		assert_true(double_eq_rel(exp(ln_gamma(x)),
+		    (double)factorial(x-1), MAX_REL_ERR, MAX_ABS_ERR),
+		    "Incorrect factorial result for x=%u", x);
+	}
+}
+TEST_END
+
+/* Expected ln_gamma([0.0..100.0] increment=0.25). */
+static const double ln_gamma_misc_expected[] = {
+	INFINITY,
+	1.28802252469807743, 0.57236494292470008, 0.20328095143129538,
+	0.00000000000000000, -0.09827183642181320, -0.12078223763524518,
+	-0.08440112102048555, 0.00000000000000000, 0.12487171489239651,
+	0.28468287047291918, 0.47521466691493719, 0.69314718055994529,
+	0.93580193110872523, 1.20097360234707429, 1.48681557859341718,
+	1.79175946922805496, 2.11445692745037128, 2.45373657084244234,
+	2.80857141857573644, 3.17805383034794575, 3.56137591038669710,
+	3.95781396761871651, 4.36671603662228680, 4.78749174278204581,
+	5.21960398699022932, 5.66256205985714178, 6.11591589143154568,
+	6.57925121201010121, 7.05218545073853953, 7.53436423675873268,
+	8.02545839631598312, 8.52516136106541467, 9.03318691960512332,
+	9.54926725730099690, 10.07315123968123949, 10.60460290274525086,
+	11.14340011995171231, 11.68933342079726856, 12.24220494005076176,
+	12.80182748008146909, 13.36802367147604720, 13.94062521940376342,
+	14.51947222506051816, 15.10441257307551943, 15.69530137706046524,
+	16.29200047656724237, 16.89437797963419285, 17.50230784587389010,
+	18.11566950571089407, 18.73434751193644843, 19.35823122022435427,
+	19.98721449566188468, 20.62119544270163018, 21.26007615624470048,
+	21.90376249182879320, 22.55216385312342098, 23.20519299513386002,
+	23.86276584168908954, 24.52480131594137802, 25.19122118273868338,
+	25.86194990184851861, 26.53691449111561340, 27.21604439872720604,
+	27.89927138384089389, 28.58652940490193828, 29.27775451504081516,
+	29.97288476399884871, 30.67186010608067548, 31.37462231367769050,
+	32.08111489594735843, 32.79128302226991565, 33.50507345013689076,
+	34.22243445715505317, 34.94331577687681545, 35.66766853819134298,
+	36.39544520803305261, 37.12659953718355865, 37.86108650896109395,
+	38.59886229060776230, 39.33988418719949465, 40.08411059791735198,
+	40.83150097453079752, 41.58201578195490100, 42.33561646075348506,
+	43.09226539146988699, 43.85192586067515208, 44.61456202863158893,
+	45.38013889847690052, 46.14862228684032885, 46.91997879580877395,
+	47.69417578616628361, 48.47118135183522014, 49.25096429545256882,
+	50.03349410501914463, 50.81874093156324790, 51.60667556776436982,
+	52.39726942748592364, 53.19049452616926743, 53.98632346204390586,
+	54.78472939811231157, 55.58568604486942633, 56.38916764371992940,
+	57.19514895105859864, 58.00360522298051080, 58.81451220059079787,
+	59.62784609588432261, 60.44358357816834371, 61.26170176100199427,
+	62.08217818962842927, 62.90499082887649962, 63.73011805151035958,
+	64.55753862700632340, 65.38723171073768015, 66.21917683354901385,
+	67.05335389170279825, 67.88974313718154008, 68.72832516833013017,
+	69.56908092082363737, 70.41199165894616385, 71.25703896716800045,
+	72.10420474200799390, 72.95347118416940191, 73.80482079093779646,
+	74.65823634883015814, 75.51370092648485866, 76.37119786778275454,
+	77.23071078519033961, 78.09222355331530707, 78.95572030266725960,
+	79.82118541361435859, 80.68860351052903468, 81.55795945611502873,
+	82.42923834590904164, 83.30242550295004378, 84.17750647261028973,
+	85.05446701758152983, 85.93329311301090456, 86.81397094178107920,
+	87.69648688992882057, 88.58082754219766741, 89.46697967771913795,
+	90.35493026581838194, 91.24466646193963015, 92.13617560368709292,
+	93.02944520697742803, 93.92446296229978486, 94.82121673107967297,
+	95.71969454214321615, 96.61988458827809723, 97.52177522288820910,
+	98.42535495673848800, 99.33061245478741341, 100.23753653310367895,
+	101.14611615586458981, 102.05634043243354370, 102.96819861451382394,
+	103.88168009337621811, 104.79677439715833032, 105.71347118823287303,
+	106.63176026064346047, 107.55163153760463501, 108.47307506906540198,
+	109.39608102933323153, 110.32063971475740516, 111.24674154146920557,
+	112.17437704317786995, 113.10353686902013237, 114.03421178146170689,
+	114.96639265424990128, 115.90007047041454769, 116.83523632031698014,
+	117.77188139974506953, 118.70999700805310795, 119.64957454634490830,
+	120.59060551569974962, 121.53308151543865279, 122.47699424143097247,
+	123.42233548443955726, 124.36909712850338394, 125.31727114935689826,
+	126.26684961288492559, 127.21782467361175861, 128.17018857322420899,
+	129.12393363912724453, 130.07905228303084755, 131.03553699956862033,
+	131.99338036494577864, 132.95257503561629164, 133.91311374698926784,
+	134.87498931216194364, 135.83819462068046846, 136.80272263732638294,
+	137.76856640092901785, 138.73571902320256299, 139.70417368760718091,
+	140.67392364823425055, 141.64496222871400732, 142.61728282114600574,
+	143.59087888505104047, 144.56574394634486680, 145.54187159633210058,
+	146.51925549072063859, 147.49788934865566148, 148.47776695177302031,
+	149.45888214327129617, 150.44122882700193600, 151.42480096657754984,
+	152.40959258449737490, 153.39559776128982094, 154.38281063467164245,
+	155.37122539872302696, 156.36083630307879844, 157.35163765213474107,
+	158.34362380426921391, 159.33678917107920370, 160.33112821663092973,
+	161.32663545672428995, 162.32330545817117695, 163.32113283808695314,
+	164.32011226319519892, 165.32023844914485267, 166.32150615984036790,
+	167.32391020678358018, 168.32744544842768164, 169.33210678954270634,
+	170.33788918059275375, 171.34478761712384198, 172.35279713916281707,
+	173.36191283062726143, 174.37212981874515094, 175.38344327348534080,
+	176.39584840699734514, 177.40934047306160437, 178.42391476654847793,
+	179.43956662288721304, 180.45629141754378111, 181.47408456550741107,
+	182.49294152078630304, 183.51285777591152737, 184.53382886144947861,
+	185.55585034552262869, 186.57891783333786861, 187.60302696672312095,
+	188.62817342367162610, 189.65435291789341932, 190.68156119837468054,
+	191.70979404894376330, 192.73904728784492590, 193.76931676731820176,
+	194.80059837318714244, 195.83288802445184729, 196.86618167288995096,
+	197.90047530266301123, 198.93576492992946214, 199.97204660246373464,
+	201.00931639928148797, 202.04757043027063901, 203.08680483582807597,
+	204.12701578650228385, 205.16819948264117102, 206.21035215404597807,
+	207.25347005962987623, 208.29754948708190909, 209.34258675253678916,
+	210.38857820024875878, 211.43552020227099320, 212.48340915813977858,
+	213.53224149456323744, 214.58201366511514152, 215.63272214993284592,
+	216.68436345542014010, 217.73693411395422004, 218.79043068359703739,
+	219.84484974781133815, 220.90018791517996988, 221.95644181913033322,
+	223.01360811766215875, 224.07168349307951871, 225.13066465172661879,
+	226.19054832372759734, 227.25133126272962159, 228.31301024565024704,
+	229.37558207242807384, 230.43904356577689896, 231.50339157094342113,
+	232.56862295546847008, 233.63473460895144740, 234.70172344281823484,
+	235.76958639009222907, 236.83832040516844586, 237.90792246359117712,
+	238.97838956183431947, 240.04971871708477238, 241.12190696702904802,
+	242.19495136964280846, 243.26884900298270509, 244.34359696498191283,
+	245.41919237324782443, 246.49563236486270057, 247.57291409618682110,
+	248.65103474266476269, 249.72999149863338175, 250.80978157713354904,
+	251.89040220972316320, 252.97185064629374551, 254.05412415488834199,
+	255.13722002152300661, 256.22113555000953511, 257.30586806178126835,
+	258.39141489572085675, 259.47777340799029844, 260.56494097186322279,
+	261.65291497755913497, 262.74169283208021852, 263.83127195904967266,
+	264.92164979855277807, 266.01282380697938379, 267.10479145686849733,
+	268.19755023675537586, 269.29109765101975427, 270.38543121973674488,
+	271.48054847852881721, 272.57644697842033565, 273.67312428569374561,
+	274.77057798174683967, 275.86880566295326389, 276.96780494052313770,
+	278.06757344036617496, 279.16810880295668085, 280.26940868320008349,
+	281.37147075030043197, 282.47429268763045229, 283.57787219260217171,
+	284.68220697654078322, 285.78729476455760050, 286.89313329542699194,
+	287.99972032146268930, 289.10705360839756395, 290.21513093526289140,
+	291.32395009427028754, 292.43350889069523646, 293.54380514276073200,
+	294.65483668152336350, 295.76660135076059532, 296.87909700685889902,
+	297.99232151870342022, 299.10627276756946458, 300.22094864701409733,
+	301.33634706277030091, 302.45246593264130297, 303.56930318639643929,
+	304.68685676566872189, 305.80512462385280514, 306.92410472600477078,
+	308.04379504874236773, 309.16419358014690033, 310.28529831966631036,
+	311.40710727801865687, 312.52961847709792664, 313.65282994987899201,
+	314.77673974032603610, 315.90134590329950015, 317.02664650446632777,
+	318.15263962020929966, 319.27932333753892635, 320.40669575400545455,
+	321.53475497761127144, 322.66349912672620803, 323.79292633000159185,
+	324.92303472628691452, 326.05382246454587403, 327.18528770377525916,
+	328.31742861292224234, 329.45024337080525356, 330.58373016603343331,
+	331.71788719692847280, 332.85271267144611329, 333.98820480709991898,
+	335.12436183088397001, 336.26118197919845443, 337.39866349777429377,
+	338.53680464159958774, 339.67560367484657036, 340.81505887079896411,
+	341.95516851178109619, 343.09593088908627578, 344.23734430290727460,
+	345.37940706226686416, 346.52211748494903532, 347.66547389743118401,
+	348.80947463481720661, 349.95411804077025408, 351.09940246744753267,
+	352.24532627543504759, 353.39188783368263103, 354.53908551944078908,
+	355.68691771819692349, 356.83538282361303118, 357.98447923746385868,
+	359.13420536957539753
+};
+
+TEST_BEGIN(test_ln_gamma_misc) {
+	unsigned i;
+
+	for (i = 1; i < sizeof(ln_gamma_misc_expected)/sizeof(double); i++) {
+		double x = (double)i * 0.25;
+		assert_true(double_eq_rel(ln_gamma(x),
+		    ln_gamma_misc_expected[i], MAX_REL_ERR, MAX_ABS_ERR),
+		    "Incorrect ln_gamma result for i=%u", i);
+	}
+}
+TEST_END
+
+/* Expected pt_norm([0.01..0.99] increment=0.01). */
+static const double pt_norm_expected[] = {
+	-INFINITY,
+	-2.32634787404084076, -2.05374891063182252, -1.88079360815125085,
+	-1.75068607125216946, -1.64485362695147264, -1.55477359459685305,
+	-1.47579102817917063, -1.40507156030963221, -1.34075503369021654,
+	-1.28155156554460081, -1.22652812003661049, -1.17498679206608991,
+	-1.12639112903880045, -1.08031934081495606, -1.03643338949378938,
+	-0.99445788320975281, -0.95416525314619416, -0.91536508784281390,
+	-0.87789629505122846, -0.84162123357291418, -0.80642124701824025,
+	-0.77219321418868492, -0.73884684918521371, -0.70630256284008752,
+	-0.67448975019608171, -0.64334540539291685, -0.61281299101662701,
+	-0.58284150727121620, -0.55338471955567281, -0.52440051270804067,
+	-0.49585034734745320, -0.46769879911450812, -0.43991316567323380,
+	-0.41246312944140462, -0.38532046640756751, -0.35845879325119373,
+	-0.33185334643681652, -0.30548078809939738, -0.27931903444745404,
+	-0.25334710313579978, -0.22754497664114931, -0.20189347914185077,
+	-0.17637416478086135, -0.15096921549677725, -0.12566134685507399,
+	-0.10043372051146975, -0.07526986209982976, -0.05015358346473352,
+	-0.02506890825871106, 0.00000000000000000, 0.02506890825871106,
+	0.05015358346473366, 0.07526986209982990, 0.10043372051146990,
+	0.12566134685507413, 0.15096921549677739, 0.17637416478086146,
+	0.20189347914185105, 0.22754497664114931, 0.25334710313579978,
+	0.27931903444745404, 0.30548078809939738, 0.33185334643681652,
+	0.35845879325119373, 0.38532046640756762, 0.41246312944140484,
+	0.43991316567323391, 0.46769879911450835, 0.49585034734745348,
+	0.52440051270804111, 0.55338471955567303, 0.58284150727121620,
+	0.61281299101662701, 0.64334540539291685, 0.67448975019608171,
+	0.70630256284008752, 0.73884684918521371, 0.77219321418868492,
+	0.80642124701824036, 0.84162123357291441, 0.87789629505122879,
+	0.91536508784281423, 0.95416525314619460, 0.99445788320975348,
+	1.03643338949378938, 1.08031934081495606, 1.12639112903880045,
+	1.17498679206608991, 1.22652812003661049, 1.28155156554460081,
+	1.34075503369021654, 1.40507156030963265, 1.47579102817917085,
+	1.55477359459685394, 1.64485362695147308, 1.75068607125217102,
+	1.88079360815125041, 2.05374891063182208, 2.32634787404084076
+};
+
+TEST_BEGIN(test_pt_norm) {
+	unsigned i;
+
+	for (i = 1; i < sizeof(pt_norm_expected)/sizeof(double); i++) {
+		double p = (double)i * 0.01;
+		assert_true(double_eq_rel(pt_norm(p), pt_norm_expected[i],
+		    MAX_REL_ERR, MAX_ABS_ERR),
+		    "Incorrect pt_norm result for i=%u", i);
+	}
+}
+TEST_END
+
+/*
+ * Expected pt_chi2(p=[0.01..0.99] increment=0.07,
+ *                  df={0.1, 1.1, 10.1, 100.1, 1000.1}).
+ */
+static const double pt_chi2_df[] = {0.1, 1.1, 10.1, 100.1, 1000.1};
+static const double pt_chi2_expected[] = {
+	1.168926411457320e-40, 1.347680397072034e-22, 3.886980416666260e-17,
+	8.245951724356564e-14, 2.068936347497604e-11, 1.562561743309233e-09,
+	5.459543043426564e-08, 1.114775688149252e-06, 1.532101202364371e-05,
+	1.553884683726585e-04, 1.239396954915939e-03, 8.153872320255721e-03,
+	4.631183739647523e-02, 2.473187311701327e-01, 2.175254800183617e+00,
+
+	0.0003729887888876379, 0.0164409238228929513, 0.0521523015190650113,
+	0.1064701372271216612, 0.1800913735793082115, 0.2748704281195626931,
+	0.3939246282787986497, 0.5420727552260817816, 0.7267265822221973259,
+	0.9596554296000253670, 1.2607440376386165326, 1.6671185084541604304,
+	2.2604828984738705167, 3.2868613342148607082, 6.9298574921692139839,
+
+	2.606673548632508, 4.602913725294877, 5.646152813924212,
+	6.488971315540869, 7.249823275816285, 7.977314231410841,
+	8.700354939944047, 9.441728024225892, 10.224338321374127,
+	11.076435368801061, 12.039320937038386, 13.183878752697167,
+	14.657791935084575, 16.885728216339373, 23.361991680031817,
+
+	70.14844087392152, 80.92379498849355, 85.53325420085891,
+	88.94433120715347, 91.83732712857017, 94.46719943606301,
+	96.96896479994635, 99.43412843510363, 101.94074719829733,
+	104.57228644307247, 107.43900093448734, 110.71844673417287,
+	114.76616819871325, 120.57422505959563, 135.92318818757556,
+
+	899.0072447849649, 937.9271278858220, 953.8117189560207,
+	965.3079371501154, 974.8974061207954, 983.4936235182347,
+	991.5691170518946, 999.4334123954690, 1007.3391826856553,
+	1015.5445154999951, 1024.3777075619569, 1034.3538789836223,
+	1046.4872561869577, 1063.5717461999654, 1107.0741966053859
+};
+
+TEST_BEGIN(test_pt_chi2) {
+	unsigned i, j;
+	unsigned e = 0;
+
+	for (i = 0; i < sizeof(pt_chi2_df)/sizeof(double); i++) {
+		double df = pt_chi2_df[i];
+		double ln_gamma_df = ln_gamma(df * 0.5);
+		for (j = 1; j < 100; j += 7) {
+			double p = (double)j * 0.01;
+			assert_true(double_eq_rel(pt_chi2(p, df, ln_gamma_df),
+			    pt_chi2_expected[e], MAX_REL_ERR, MAX_ABS_ERR),
+			    "Incorrect pt_chi2 result for i=%u, j=%u", i, j);
+			e++;
+		}
+	}
+}
+TEST_END
+
+/*
+ * Expected pt_gamma(p=[0.1..0.99] increment=0.07,
+ *                   shape=[0.5..3.0] increment=0.5).
+ */
+static const double pt_gamma_shape[] = {0.5, 1.0, 1.5, 2.0, 2.5, 3.0};
+static const double pt_gamma_expected[] = {
+	7.854392895485103e-05, 5.043466107888016e-03, 1.788288957794883e-02,
+	3.900956150232906e-02, 6.913847560638034e-02, 1.093710833465766e-01,
+	1.613412523825817e-01, 2.274682115597864e-01, 3.114117323127083e-01,
+	4.189466220207417e-01, 5.598106789059246e-01, 7.521856146202706e-01,
+	1.036125427911119e+00, 1.532450860038180e+00, 3.317448300510606e+00,
+
+	0.01005033585350144, 0.08338160893905107, 0.16251892949777497,
+	0.24846135929849966, 0.34249030894677596, 0.44628710262841947,
+	0.56211891815354142, 0.69314718055994529, 0.84397007029452920,
+	1.02165124753198167, 1.23787435600161766, 1.51412773262977574,
+	1.89711998488588196, 2.52572864430825783, 4.60517018598809091,
+
+	0.05741590094955853, 0.24747378084860744, 0.39888572212236084,
+	0.54394139997444901, 0.69048812513915159, 0.84311389861296104,
+	1.00580622221479898, 1.18298694218766931, 1.38038096305861213,
+	1.60627736383027453, 1.87396970522337947, 2.20749220408081070,
+	2.65852391865854942, 3.37934630984842244, 5.67243336507218476,
+
+	0.1485547402532659, 0.4657458011640391, 0.6832386130709406,
+	0.8794297834672100, 1.0700752852474524, 1.2629614217350744,
+	1.4638400448580779, 1.6783469900166610, 1.9132338090606940,
+	2.1778589228618777, 2.4868823970010991, 2.8664695666264195,
+	3.3724415436062114, 4.1682658512758071, 6.6383520679938108,
+
+	0.2771490383641385, 0.7195001279643727, 0.9969081732265243,
+	1.2383497880608061, 1.4675206597269927, 1.6953064251816552,
+	1.9291243435606809, 2.1757300955477641, 2.4428032131216391,
+	2.7406534569230616, 3.0851445039665513, 3.5043101122033367,
+	4.0575997065264637, 4.9182956424675286, 7.5431362346944937,
+
+	0.4360451650782932, 0.9983600902486267, 1.3306365880734528,
+	1.6129750834753802, 1.8767241606994294, 2.1357032436097660,
+	2.3988853336865565, 2.6740603137235603, 2.9697561737517959,
+	3.2971457713883265, 3.6731795898504660, 4.1275751617770631,
+	4.7230515633946677, 5.6417477865306020, 8.4059469148854635
+};
+
+TEST_BEGIN(test_pt_gamma_shape) {
+	unsigned i, j;
+	unsigned e = 0;
+
+	for (i = 0; i < sizeof(pt_gamma_shape)/sizeof(double); i++) {
+		double shape = pt_gamma_shape[i];
+		double ln_gamma_shape = ln_gamma(shape);
+		for (j = 1; j < 100; j += 7) {
+			double p = (double)j * 0.01;
+			assert_true(double_eq_rel(pt_gamma(p, shape, 1.0,
+			    ln_gamma_shape), pt_gamma_expected[e], MAX_REL_ERR,
+			    MAX_ABS_ERR),
+			    "Incorrect pt_gamma result for i=%u, j=%u", i, j);
+			e++;
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_pt_gamma_scale) {
+	double shape = 1.0;
+	double ln_gamma_shape = ln_gamma(shape);
+
+	assert_true(double_eq_rel(
+	    pt_gamma(0.5, shape, 1.0, ln_gamma_shape) * 10.0,
+	    pt_gamma(0.5, shape, 10.0, ln_gamma_shape), MAX_REL_ERR,
+	    MAX_ABS_ERR),
+	    "Scale should be trivially equivalent to external multiplication");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ln_gamma_factorial,
+	    test_ln_gamma_misc,
+	    test_pt_norm,
+	    test_pt_chi2,
+	    test_pt_gamma_shape,
+	    test_pt_gamma_scale);
+}
diff --git a/jemalloc/test/unit/mq.c b/jemalloc/test/unit/mq.c
new file mode 100644
index 0000000..57a4d54
--- /dev/null
+++ b/jemalloc/test/unit/mq.c
@@ -0,0 +1,89 @@
+#include "test/jemalloc_test.h"
+
+#define NSENDERS	3
+#define NMSGS		100000
+
+typedef struct mq_msg_s mq_msg_t;
+struct mq_msg_s {
+	mq_msg(mq_msg_t)	link;
+};
+mq_gen(static, mq_, mq_t, mq_msg_t, link)
+
+TEST_BEGIN(test_mq_basic) {
+	mq_t mq;
+	mq_msg_t msg;
+
+	assert_false(mq_init(&mq), "Unexpected mq_init() failure");
+	assert_u_eq(mq_count(&mq), 0, "mq should be empty");
+	assert_ptr_null(mq_tryget(&mq),
+	    "mq_tryget() should fail when the queue is empty");
+
+	mq_put(&mq, &msg);
+	assert_u_eq(mq_count(&mq), 1, "mq should contain one message");
+	assert_ptr_eq(mq_tryget(&mq), &msg, "mq_tryget() should return msg");
+
+	mq_put(&mq, &msg);
+	assert_ptr_eq(mq_get(&mq), &msg, "mq_get() should return msg");
+
+	mq_fini(&mq);
+}
+TEST_END
+
+static void *
+thd_receiver_start(void *arg) {
+	mq_t *mq = (mq_t *)arg;
+	unsigned i;
+
+	for (i = 0; i < (NSENDERS * NMSGS); i++) {
+		mq_msg_t *msg = mq_get(mq);
+		assert_ptr_not_null(msg, "mq_get() should never return NULL");
+		dallocx(msg, 0);
+	}
+	return NULL;
+}
+
+static void *
+thd_sender_start(void *arg) {
+	mq_t *mq = (mq_t *)arg;
+	unsigned i;
+
+	for (i = 0; i < NMSGS; i++) {
+		mq_msg_t *msg;
+		void *p;
+		p = mallocx(sizeof(mq_msg_t), 0);
+		assert_ptr_not_null(p, "Unexpected mallocx() failure");
+		msg = (mq_msg_t *)p;
+		mq_put(mq, msg);
+	}
+	return NULL;
+}
+
+TEST_BEGIN(test_mq_threaded) {
+	mq_t mq;
+	thd_t receiver;
+	thd_t senders[NSENDERS];
+	unsigned i;
+
+	assert_false(mq_init(&mq), "Unexpected mq_init() failure");
+
+	thd_create(&receiver, thd_receiver_start, (void *)&mq);
+	for (i = 0; i < NSENDERS; i++) {
+		thd_create(&senders[i], thd_sender_start, (void *)&mq);
+	}
+
+	thd_join(receiver, NULL);
+	for (i = 0; i < NSENDERS; i++) {
+		thd_join(senders[i], NULL);
+	}
+
+	mq_fini(&mq);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_mq_basic,
+	    test_mq_threaded);
+}
+
diff --git a/jemalloc/test/unit/mtx.c b/jemalloc/test/unit/mtx.c
new file mode 100644
index 0000000..424587b
--- /dev/null
+++ b/jemalloc/test/unit/mtx.c
@@ -0,0 +1,57 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS	2
+#define NINCRS		2000000
+
+TEST_BEGIN(test_mtx_basic) {
+	mtx_t mtx;
+
+	assert_false(mtx_init(&mtx), "Unexpected mtx_init() failure");
+	mtx_lock(&mtx);
+	mtx_unlock(&mtx);
+	mtx_fini(&mtx);
+}
+TEST_END
+
+typedef struct {
+	mtx_t		mtx;
+	unsigned	x;
+} thd_start_arg_t;
+
+static void *
+thd_start(void *varg) {
+	thd_start_arg_t *arg = (thd_start_arg_t *)varg;
+	unsigned i;
+
+	for (i = 0; i < NINCRS; i++) {
+		mtx_lock(&arg->mtx);
+		arg->x++;
+		mtx_unlock(&arg->mtx);
+	}
+	return NULL;
+}
+
+TEST_BEGIN(test_mtx_race) {
+	thd_start_arg_t arg;
+	thd_t thds[NTHREADS];
+	unsigned i;
+
+	assert_false(mtx_init(&arg.mtx), "Unexpected mtx_init() failure");
+	arg.x = 0;
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start, (void *)&arg);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+	assert_u_eq(arg.x, NTHREADS * NINCRS,
+	    "Race-related counter corruption");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_mtx_basic,
+	    test_mtx_race);
+}
diff --git a/jemalloc/test/unit/nstime.c b/jemalloc/test/unit/nstime.c
new file mode 100644
index 0000000..f313780
--- /dev/null
+++ b/jemalloc/test/unit/nstime.c
@@ -0,0 +1,249 @@
+#include "test/jemalloc_test.h"
+
+#define BILLION	UINT64_C(1000000000)
+
+TEST_BEGIN(test_nstime_init) {
+	nstime_t nst;
+
+	nstime_init(&nst, 42000000043);
+	assert_u64_eq(nstime_ns(&nst), 42000000043, "ns incorrectly read");
+	assert_u64_eq(nstime_sec(&nst), 42, "sec incorrectly read");
+	assert_u64_eq(nstime_nsec(&nst), 43, "nsec incorrectly read");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_init2) {
+	nstime_t nst;
+
+	nstime_init2(&nst, 42, 43);
+	assert_u64_eq(nstime_sec(&nst), 42, "sec incorrectly read");
+	assert_u64_eq(nstime_nsec(&nst), 43, "nsec incorrectly read");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_copy) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_init(&nstb, 0);
+	nstime_copy(&nstb, &nsta);
+	assert_u64_eq(nstime_sec(&nstb), 42, "sec incorrectly copied");
+	assert_u64_eq(nstime_nsec(&nstb), 43, "nsec incorrectly copied");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_compare) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0, "Times should be equal");
+	assert_d_eq(nstime_compare(&nstb, &nsta), 0, "Times should be equal");
+
+	nstime_init2(&nstb, 42, 42);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 1,
+	    "nsta should be greater than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), -1,
+	    "nstb should be less than nsta");
+
+	nstime_init2(&nstb, 42, 44);
+	assert_d_eq(nstime_compare(&nsta, &nstb), -1,
+	    "nsta should be less than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), 1,
+	    "nstb should be greater than nsta");
+
+	nstime_init2(&nstb, 41, BILLION - 1);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 1,
+	    "nsta should be greater than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), -1,
+	    "nstb should be less than nsta");
+
+	nstime_init2(&nstb, 43, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), -1,
+	    "nsta should be less than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), 1,
+	    "nstb should be greater than nsta");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_add) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_add(&nsta, &nstb);
+	nstime_init2(&nstb, 84, 86);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+
+	nstime_init2(&nsta, 42, BILLION - 1);
+	nstime_copy(&nstb, &nsta);
+	nstime_add(&nsta, &nstb);
+	nstime_init2(&nstb, 85, BILLION - 2);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_iadd) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, BILLION - 1);
+	nstime_iadd(&nsta, 1);
+	nstime_init2(&nstb, 43, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+
+	nstime_init2(&nsta, 42, 1);
+	nstime_iadd(&nsta, BILLION + 1);
+	nstime_init2(&nstb, 43, 2);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_subtract) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_subtract(&nsta, &nstb);
+	nstime_init(&nstb, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_init2(&nstb, 41, 44);
+	nstime_subtract(&nsta, &nstb);
+	nstime_init2(&nstb, 0, BILLION - 1);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_isubtract) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_isubtract(&nsta, 42*BILLION + 43);
+	nstime_init(&nstb, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_isubtract(&nsta, 41*BILLION + 44);
+	nstime_init2(&nstb, 0, BILLION - 1);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_imultiply) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_imultiply(&nsta, 10);
+	nstime_init2(&nstb, 420, 430);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect multiplication result");
+
+	nstime_init2(&nsta, 42, 666666666);
+	nstime_imultiply(&nsta, 3);
+	nstime_init2(&nstb, 127, 999999998);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect multiplication result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_idivide) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	nstime_idivide(&nsta, 10);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect division result");
+
+	nstime_init2(&nsta, 42, 666666666);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 3);
+	nstime_idivide(&nsta, 3);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect division result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_divide) {
+	nstime_t nsta, nstb, nstc;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	assert_u64_eq(nstime_divide(&nsta, &nstb), 10,
+	    "Incorrect division result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	nstime_init(&nstc, 1);
+	nstime_add(&nsta, &nstc);
+	assert_u64_eq(nstime_divide(&nsta, &nstb), 10,
+	    "Incorrect division result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	nstime_init(&nstc, 1);
+	nstime_subtract(&nsta, &nstc);
+	assert_u64_eq(nstime_divide(&nsta, &nstb), 9,
+	    "Incorrect division result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_monotonic) {
+	nstime_monotonic();
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_update) {
+	nstime_t nst;
+
+	nstime_init(&nst, 0);
+
+	assert_false(nstime_update(&nst), "Basic time update failed.");
+
+	/* Only Rip Van Winkle sleeps this long. */
+	{
+		nstime_t addend;
+		nstime_init2(&addend, 631152000, 0);
+		nstime_add(&nst, &addend);
+	}
+	{
+		nstime_t nst0;
+		nstime_copy(&nst0, &nst);
+		assert_true(nstime_update(&nst),
+		    "Update should detect time roll-back.");
+		assert_d_eq(nstime_compare(&nst, &nst0), 0,
+		    "Time should not have been modified");
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_nstime_init,
+	    test_nstime_init2,
+	    test_nstime_copy,
+	    test_nstime_compare,
+	    test_nstime_add,
+	    test_nstime_iadd,
+	    test_nstime_subtract,
+	    test_nstime_isubtract,
+	    test_nstime_imultiply,
+	    test_nstime_idivide,
+	    test_nstime_divide,
+	    test_nstime_monotonic,
+	    test_nstime_update);
+}
diff --git a/jemalloc/test/unit/pack.c b/jemalloc/test/unit/pack.c
new file mode 100644
index 0000000..fc188b0
--- /dev/null
+++ b/jemalloc/test/unit/pack.c
@@ -0,0 +1,166 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * Size class that is a divisor of the page size, ideally 4+ regions per run.
+ */
+#if LG_PAGE <= 14
+#define SZ	(ZU(1) << (LG_PAGE - 2))
+#else
+#define SZ	ZU(4096)
+#endif
+
+/*
+ * Number of slabs to consume at high water mark.  Should be at least 2 so that
+ * if mmap()ed memory grows downward, downward growth of mmap()ed memory is
+ * tested.
+ */
+#define NSLABS	8
+
+static unsigned
+binind_compute(void) {
+	size_t sz;
+	unsigned nbins, i;
+
+	sz = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &sz, NULL, 0), 0,
+	    "Unexpected mallctl failure");
+
+	for (i = 0; i < nbins; i++) {
+		size_t mib[4];
+		size_t miblen = sizeof(mib)/sizeof(size_t);
+		size_t size;
+
+		assert_d_eq(mallctlnametomib("arenas.bin.0.size", mib,
+		    &miblen), 0, "Unexpected mallctlnametomb failure");
+		mib[2] = (size_t)i;
+
+		sz = sizeof(size);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&size, &sz, NULL,
+		    0), 0, "Unexpected mallctlbymib failure");
+		if (size == SZ) {
+			return i;
+		}
+	}
+
+	test_fail("Unable to compute nregs_per_run");
+	return 0;
+}
+
+static size_t
+nregs_per_run_compute(void) {
+	uint32_t nregs;
+	size_t sz;
+	unsigned binind = binind_compute();
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib("arenas.bin.0.nregs", mib, &miblen), 0,
+	    "Unexpected mallctlnametomb failure");
+	mib[2] = (size_t)binind;
+	sz = sizeof(nregs);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&nregs, &sz, NULL,
+	    0), 0, "Unexpected mallctlbymib failure");
+	return nregs;
+}
+
+static unsigned
+arenas_create_mallctl(void) {
+	unsigned arena_ind;
+	size_t sz;
+
+	sz = sizeof(arena_ind);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Error in arenas.create");
+
+	return arena_ind;
+}
+
+static void
+arena_reset_mallctl(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib("arena.0.reset", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+TEST_BEGIN(test_pack) {
+	bool prof_enabled;
+	size_t sz = sizeof(prof_enabled);
+	if (mallctl("opt.prof", (void *)&prof_enabled, &sz, NULL, 0) == 0) {
+		test_skip_if(prof_enabled);
+	}
+
+	unsigned arena_ind = arenas_create_mallctl();
+	size_t nregs_per_run = nregs_per_run_compute();
+	size_t nregs = nregs_per_run * NSLABS;
+	VARIABLE_ARRAY(void *, ptrs, nregs);
+	size_t i, j, offset;
+
+	/* Fill matrix. */
+	for (i = offset = 0; i < NSLABS; i++) {
+		for (j = 0; j < nregs_per_run; j++) {
+			void *p = mallocx(SZ, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p,
+			    "Unexpected mallocx(%zu, MALLOCX_ARENA(%u) |"
+			    " MALLOCX_TCACHE_NONE) failure, run=%zu, reg=%zu",
+			    SZ, arena_ind, i, j);
+			ptrs[(i * nregs_per_run) + j] = p;
+		}
+	}
+
+	/*
+	 * Free all but one region of each run, but rotate which region is
+	 * preserved, so that subsequent allocations exercise the within-run
+	 * layout policy.
+	 */
+	offset = 0;
+	for (i = offset = 0;
+	    i < NSLABS;
+	    i++, offset = (offset + 1) % nregs_per_run) {
+		for (j = 0; j < nregs_per_run; j++) {
+			void *p = ptrs[(i * nregs_per_run) + j];
+			if (offset == j) {
+				continue;
+			}
+			dallocx(p, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE);
+		}
+	}
+
+	/*
+	 * Logically refill matrix, skipping preserved regions and verifying
+	 * that the matrix is unmodified.
+	 */
+	offset = 0;
+	for (i = offset = 0;
+	    i < NSLABS;
+	    i++, offset = (offset + 1) % nregs_per_run) {
+		for (j = 0; j < nregs_per_run; j++) {
+			void *p;
+
+			if (offset == j) {
+				continue;
+			}
+			p = mallocx(SZ, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE);
+			assert_ptr_eq(p, ptrs[(i * nregs_per_run) + j],
+			    "Unexpected refill discrepancy, run=%zu, reg=%zu\n",
+			    i, j);
+		}
+	}
+
+	/* Clean up. */
+	arena_reset_mallctl(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_pack);
+}
diff --git a/jemalloc/test/unit/pack.sh b/jemalloc/test/unit/pack.sh
new file mode 100644
index 0000000..6f45148
--- /dev/null
+++ b/jemalloc/test/unit/pack.sh
@@ -0,0 +1,4 @@
+#!/bin/sh
+
+# Immediately purge to minimize fragmentation.
+export MALLOC_CONF="dirty_decay_ms:0,muzzy_decay_ms:0"
diff --git a/jemalloc/test/unit/pages.c b/jemalloc/test/unit/pages.c
new file mode 100644
index 0000000..ee729ee
--- /dev/null
+++ b/jemalloc/test/unit/pages.c
@@ -0,0 +1,29 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_pages_huge) {
+	size_t alloc_size;
+	bool commit;
+	void *pages, *hugepage;
+
+	alloc_size = HUGEPAGE * 2 - PAGE;
+	commit = true;
+	pages = pages_map(NULL, alloc_size, PAGE, &commit);
+	assert_ptr_not_null(pages, "Unexpected pages_map() error");
+
+	if (init_system_thp_mode == thp_mode_default) {
+	    hugepage = (void *)(ALIGNMENT_CEILING((uintptr_t)pages, HUGEPAGE));
+	    assert_b_ne(pages_huge(hugepage, HUGEPAGE), have_madvise_huge,
+	        "Unexpected pages_huge() result");
+	    assert_false(pages_nohuge(hugepage, HUGEPAGE),
+	        "Unexpected pages_nohuge() result");
+	}
+
+	pages_unmap(pages, alloc_size);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_pages_huge);
+}
diff --git a/jemalloc/test/unit/ph.c b/jemalloc/test/unit/ph.c
new file mode 100644
index 0000000..88bf56f
--- /dev/null
+++ b/jemalloc/test/unit/ph.c
@@ -0,0 +1,318 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ph.h"
+
+typedef struct node_s node_t;
+
+struct node_s {
+#define NODE_MAGIC 0x9823af7e
+	uint32_t magic;
+	phn(node_t) link;
+	uint64_t key;
+};
+
+static int
+node_cmp(const node_t *a, const node_t *b) {
+	int ret;
+
+	ret = (a->key > b->key) - (a->key < b->key);
+	if (ret == 0) {
+		/*
+		 * Duplicates are not allowed in the heap, so force an
+		 * arbitrary ordering for non-identical items with equal keys.
+		 */
+		ret = (((uintptr_t)a) > ((uintptr_t)b))
+		    - (((uintptr_t)a) < ((uintptr_t)b));
+	}
+	return ret;
+}
+
+static int
+node_cmp_magic(const node_t *a, const node_t *b) {
+
+	assert_u32_eq(a->magic, NODE_MAGIC, "Bad magic");
+	assert_u32_eq(b->magic, NODE_MAGIC, "Bad magic");
+
+	return node_cmp(a, b);
+}
+
+typedef ph(node_t) heap_t;
+ph_gen(static, heap_, heap_t, node_t, link, node_cmp_magic);
+
+static void
+node_print(const node_t *node, unsigned depth) {
+	unsigned i;
+	node_t *leftmost_child, *sibling;
+
+	for (i = 0; i < depth; i++) {
+		malloc_printf("\t");
+	}
+	malloc_printf("%2"FMTu64"\n", node->key);
+
+	leftmost_child = phn_lchild_get(node_t, link, node);
+	if (leftmost_child == NULL) {
+		return;
+	}
+	node_print(leftmost_child, depth + 1);
+
+	for (sibling = phn_next_get(node_t, link, leftmost_child); sibling !=
+	    NULL; sibling = phn_next_get(node_t, link, sibling)) {
+		node_print(sibling, depth + 1);
+	}
+}
+
+static void
+heap_print(const heap_t *heap) {
+	node_t *auxelm;
+
+	malloc_printf("vvv heap %p vvv\n", heap);
+	if (heap->ph_root == NULL) {
+		goto label_return;
+	}
+
+	node_print(heap->ph_root, 0);
+
+	for (auxelm = phn_next_get(node_t, link, heap->ph_root); auxelm != NULL;
+	    auxelm = phn_next_get(node_t, link, auxelm)) {
+		assert_ptr_eq(phn_next_get(node_t, link, phn_prev_get(node_t,
+		    link, auxelm)), auxelm,
+		    "auxelm's prev doesn't link to auxelm");
+		node_print(auxelm, 0);
+	}
+
+label_return:
+	malloc_printf("^^^ heap %p ^^^\n", heap);
+}
+
+static unsigned
+node_validate(const node_t *node, const node_t *parent) {
+	unsigned nnodes = 1;
+	node_t *leftmost_child, *sibling;
+
+	if (parent != NULL) {
+		assert_d_ge(node_cmp_magic(node, parent), 0,
+		    "Child is less than parent");
+	}
+
+	leftmost_child = phn_lchild_get(node_t, link, node);
+	if (leftmost_child == NULL) {
+		return nnodes;
+	}
+	assert_ptr_eq((void *)phn_prev_get(node_t, link, leftmost_child),
+	    (void *)node, "Leftmost child does not link to node");
+	nnodes += node_validate(leftmost_child, node);
+
+	for (sibling = phn_next_get(node_t, link, leftmost_child); sibling !=
+	    NULL; sibling = phn_next_get(node_t, link, sibling)) {
+		assert_ptr_eq(phn_next_get(node_t, link, phn_prev_get(node_t,
+		    link, sibling)), sibling,
+		    "sibling's prev doesn't link to sibling");
+		nnodes += node_validate(sibling, node);
+	}
+	return nnodes;
+}
+
+static unsigned
+heap_validate(const heap_t *heap) {
+	unsigned nnodes = 0;
+	node_t *auxelm;
+
+	if (heap->ph_root == NULL) {
+		goto label_return;
+	}
+
+	nnodes += node_validate(heap->ph_root, NULL);
+
+	for (auxelm = phn_next_get(node_t, link, heap->ph_root); auxelm != NULL;
+	    auxelm = phn_next_get(node_t, link, auxelm)) {
+		assert_ptr_eq(phn_next_get(node_t, link, phn_prev_get(node_t,
+		    link, auxelm)), auxelm,
+		    "auxelm's prev doesn't link to auxelm");
+		nnodes += node_validate(auxelm, NULL);
+	}
+
+label_return:
+	if (false) {
+		heap_print(heap);
+	}
+	return nnodes;
+}
+
+TEST_BEGIN(test_ph_empty) {
+	heap_t heap;
+
+	heap_new(&heap);
+	assert_true(heap_empty(&heap), "Heap should be empty");
+	assert_ptr_null(heap_first(&heap), "Unexpected node");
+	assert_ptr_null(heap_any(&heap), "Unexpected node");
+}
+TEST_END
+
+static void
+node_remove(heap_t *heap, node_t *node) {
+	heap_remove(heap, node);
+
+	node->magic = 0;
+}
+
+static node_t *
+node_remove_first(heap_t *heap) {
+	node_t *node = heap_remove_first(heap);
+	node->magic = 0;
+	return node;
+}
+
+static node_t *
+node_remove_any(heap_t *heap) {
+	node_t *node = heap_remove_any(heap);
+	node->magic = 0;
+	return node;
+}
+
+TEST_BEGIN(test_ph_random) {
+#define NNODES 25
+#define NBAGS 250
+#define SEED 42
+	sfmt_t *sfmt;
+	uint64_t bag[NNODES];
+	heap_t heap;
+	node_t nodes[NNODES];
+	unsigned i, j, k;
+
+	sfmt = init_gen_rand(SEED);
+	for (i = 0; i < NBAGS; i++) {
+		switch (i) {
+		case 0:
+			/* Insert in order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = j;
+			}
+			break;
+		case 1:
+			/* Insert in reverse order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = NNODES - j - 1;
+			}
+			break;
+		default:
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = gen_rand64_range(sfmt, NNODES);
+			}
+		}
+
+		for (j = 1; j <= NNODES; j++) {
+			/* Initialize heap and nodes. */
+			heap_new(&heap);
+			assert_u_eq(heap_validate(&heap), 0,
+			    "Incorrect node count");
+			for (k = 0; k < j; k++) {
+				nodes[k].magic = NODE_MAGIC;
+				nodes[k].key = bag[k];
+			}
+
+			/* Insert nodes. */
+			for (k = 0; k < j; k++) {
+				heap_insert(&heap, &nodes[k]);
+				if (i % 13 == 12) {
+					assert_ptr_not_null(heap_any(&heap),
+					    "Heap should not be empty");
+					/* Trigger merging. */
+					assert_ptr_not_null(heap_first(&heap),
+					    "Heap should not be empty");
+				}
+				assert_u_eq(heap_validate(&heap), k + 1,
+				    "Incorrect node count");
+			}
+
+			assert_false(heap_empty(&heap),
+			    "Heap should not be empty");
+
+			/* Remove nodes. */
+			switch (i % 6) {
+			case 0:
+				for (k = 0; k < j; k++) {
+					assert_u_eq(heap_validate(&heap), j - k,
+					    "Incorrect node count");
+					node_remove(&heap, &nodes[k]);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+				}
+				break;
+			case 1:
+				for (k = j; k > 0; k--) {
+					node_remove(&heap, &nodes[k-1]);
+					assert_u_eq(heap_validate(&heap), k - 1,
+					    "Incorrect node count");
+				}
+				break;
+			case 2: {
+				node_t *prev = NULL;
+				for (k = 0; k < j; k++) {
+					node_t *node = node_remove_first(&heap);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+					if (prev != NULL) {
+						assert_d_ge(node_cmp(node,
+						    prev), 0,
+						    "Bad removal order");
+					}
+					prev = node;
+				}
+				break;
+			} case 3: {
+				node_t *prev = NULL;
+				for (k = 0; k < j; k++) {
+					node_t *node = heap_first(&heap);
+					assert_u_eq(heap_validate(&heap), j - k,
+					    "Incorrect node count");
+					if (prev != NULL) {
+						assert_d_ge(node_cmp(node,
+						    prev), 0,
+						    "Bad removal order");
+					}
+					node_remove(&heap, node);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+					prev = node;
+				}
+				break;
+			} case 4: {
+				for (k = 0; k < j; k++) {
+					node_remove_any(&heap);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+				}
+				break;
+			} case 5: {
+				for (k = 0; k < j; k++) {
+					node_t *node = heap_any(&heap);
+					assert_u_eq(heap_validate(&heap), j - k,
+					    "Incorrect node count");
+					node_remove(&heap, node);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+				}
+				break;
+			} default:
+				not_reached();
+			}
+
+			assert_ptr_null(heap_first(&heap),
+			    "Heap should be empty");
+			assert_ptr_null(heap_any(&heap),
+			    "Heap should be empty");
+			assert_true(heap_empty(&heap), "Heap should be empty");
+		}
+	}
+	fini_gen_rand(sfmt);
+#undef NNODES
+#undef SEED
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ph_empty,
+	    test_ph_random);
+}
diff --git a/jemalloc/test/unit/prng.c b/jemalloc/test/unit/prng.c
new file mode 100644
index 0000000..b5795c2
--- /dev/null
+++ b/jemalloc/test/unit/prng.c
@@ -0,0 +1,237 @@
+#include "test/jemalloc_test.h"
+
+static void
+test_prng_lg_range_u32(bool atomic) {
+	atomic_u32_t sa, sb;
+	uint32_t ra, rb;
+	unsigned lg_range;
+
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_u32(&sa, 32, atomic);
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_u32(&sa, 32, atomic);
+	assert_u32_eq(ra, rb,
+	    "Repeated generation should produce repeated results");
+
+	atomic_store_u32(&sb, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_u32(&sb, 32, atomic);
+	assert_u32_eq(ra, rb,
+	    "Equivalent generation should produce equivalent results");
+
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_u32(&sa, 32, atomic);
+	rb = prng_lg_range_u32(&sa, 32, atomic);
+	assert_u32_ne(ra, rb,
+	    "Full-width results must not immediately repeat");
+
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_u32(&sa, 32, atomic);
+	for (lg_range = 31; lg_range > 0; lg_range--) {
+		atomic_store_u32(&sb, 42, ATOMIC_RELAXED);
+		rb = prng_lg_range_u32(&sb, lg_range, atomic);
+		assert_u32_eq((rb & (UINT32_C(0xffffffff) << lg_range)),
+		    0, "High order bits should be 0, lg_range=%u", lg_range);
+		assert_u32_eq(rb, (ra >> (32 - lg_range)),
+		    "Expected high order bits of full-width result, "
+		    "lg_range=%u", lg_range);
+	}
+}
+
+static void
+test_prng_lg_range_u64(void) {
+	uint64_t sa, sb, ra, rb;
+	unsigned lg_range;
+
+	sa = 42;
+	ra = prng_lg_range_u64(&sa, 64);
+	sa = 42;
+	rb = prng_lg_range_u64(&sa, 64);
+	assert_u64_eq(ra, rb,
+	    "Repeated generation should produce repeated results");
+
+	sb = 42;
+	rb = prng_lg_range_u64(&sb, 64);
+	assert_u64_eq(ra, rb,
+	    "Equivalent generation should produce equivalent results");
+
+	sa = 42;
+	ra = prng_lg_range_u64(&sa, 64);
+	rb = prng_lg_range_u64(&sa, 64);
+	assert_u64_ne(ra, rb,
+	    "Full-width results must not immediately repeat");
+
+	sa = 42;
+	ra = prng_lg_range_u64(&sa, 64);
+	for (lg_range = 63; lg_range > 0; lg_range--) {
+		sb = 42;
+		rb = prng_lg_range_u64(&sb, lg_range);
+		assert_u64_eq((rb & (UINT64_C(0xffffffffffffffff) << lg_range)),
+		    0, "High order bits should be 0, lg_range=%u", lg_range);
+		assert_u64_eq(rb, (ra >> (64 - lg_range)),
+		    "Expected high order bits of full-width result, "
+		    "lg_range=%u", lg_range);
+	}
+}
+
+static void
+test_prng_lg_range_zu(bool atomic) {
+	atomic_zu_t sa, sb;
+	size_t ra, rb;
+	unsigned lg_range;
+
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	assert_zu_eq(ra, rb,
+	    "Repeated generation should produce repeated results");
+
+	atomic_store_zu(&sb, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_zu(&sb, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	assert_zu_eq(ra, rb,
+	    "Equivalent generation should produce equivalent results");
+
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	rb = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	assert_zu_ne(ra, rb,
+	    "Full-width results must not immediately repeat");
+
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	for (lg_range = (ZU(1) << (3 + LG_SIZEOF_PTR)) - 1; lg_range > 0;
+	    lg_range--) {
+		atomic_store_zu(&sb, 42, ATOMIC_RELAXED);
+		rb = prng_lg_range_zu(&sb, lg_range, atomic);
+		assert_zu_eq((rb & (SIZE_T_MAX << lg_range)),
+		    0, "High order bits should be 0, lg_range=%u", lg_range);
+		assert_zu_eq(rb, (ra >> ((ZU(1) << (3 + LG_SIZEOF_PTR)) -
+		    lg_range)), "Expected high order bits of full-width "
+		    "result, lg_range=%u", lg_range);
+	}
+}
+
+TEST_BEGIN(test_prng_lg_range_u32_nonatomic) {
+	test_prng_lg_range_u32(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_u32_atomic) {
+	test_prng_lg_range_u32(true);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_u64_nonatomic) {
+	test_prng_lg_range_u64();
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_zu_nonatomic) {
+	test_prng_lg_range_zu(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_zu_atomic) {
+	test_prng_lg_range_zu(true);
+}
+TEST_END
+
+static void
+test_prng_range_u32(bool atomic) {
+	uint32_t range;
+#define MAX_RANGE	10000000
+#define RANGE_STEP	97
+#define NREPS		10
+
+	for (range = 2; range < MAX_RANGE; range += RANGE_STEP) {
+		atomic_u32_t s;
+		unsigned rep;
+
+		atomic_store_u32(&s, range, ATOMIC_RELAXED);
+		for (rep = 0; rep < NREPS; rep++) {
+			uint32_t r = prng_range_u32(&s, range, atomic);
+
+			assert_u32_lt(r, range, "Out of range");
+		}
+	}
+}
+
+static void
+test_prng_range_u64(void) {
+	uint64_t range;
+#define MAX_RANGE	10000000
+#define RANGE_STEP	97
+#define NREPS		10
+
+	for (range = 2; range < MAX_RANGE; range += RANGE_STEP) {
+		uint64_t s;
+		unsigned rep;
+
+		s = range;
+		for (rep = 0; rep < NREPS; rep++) {
+			uint64_t r = prng_range_u64(&s, range);
+
+			assert_u64_lt(r, range, "Out of range");
+		}
+	}
+}
+
+static void
+test_prng_range_zu(bool atomic) {
+	size_t range;
+#define MAX_RANGE	10000000
+#define RANGE_STEP	97
+#define NREPS		10
+
+	for (range = 2; range < MAX_RANGE; range += RANGE_STEP) {
+		atomic_zu_t s;
+		unsigned rep;
+
+		atomic_store_zu(&s, range, ATOMIC_RELAXED);
+		for (rep = 0; rep < NREPS; rep++) {
+			size_t r = prng_range_zu(&s, range, atomic);
+
+			assert_zu_lt(r, range, "Out of range");
+		}
+	}
+}
+
+TEST_BEGIN(test_prng_range_u32_nonatomic) {
+	test_prng_range_u32(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_u32_atomic) {
+	test_prng_range_u32(true);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_u64_nonatomic) {
+	test_prng_range_u64();
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_zu_nonatomic) {
+	test_prng_range_zu(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_zu_atomic) {
+	test_prng_range_zu(true);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_prng_lg_range_u32_nonatomic,
+	    test_prng_lg_range_u32_atomic,
+	    test_prng_lg_range_u64_nonatomic,
+	    test_prng_lg_range_zu_nonatomic,
+	    test_prng_lg_range_zu_atomic,
+	    test_prng_range_u32_nonatomic,
+	    test_prng_range_u32_atomic,
+	    test_prng_range_u64_nonatomic,
+	    test_prng_range_zu_nonatomic,
+	    test_prng_range_zu_atomic);
+}
diff --git a/jemalloc/test/unit/prof_accum.c b/jemalloc/test/unit/prof_accum.c
new file mode 100644
index 0000000..2522006
--- /dev/null
+++ b/jemalloc/test/unit/prof_accum.c
@@ -0,0 +1,81 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS		4
+#define NALLOCS_PER_THREAD	50
+#define DUMP_INTERVAL		1
+#define BT_COUNT_CHECK_INTERVAL	5
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+static void *
+alloc_from_permuted_backtrace(unsigned thd_ind, unsigned iteration) {
+	return btalloc(1, thd_ind*NALLOCS_PER_THREAD + iteration);
+}
+
+static void *
+thd_start(void *varg) {
+	unsigned thd_ind = *(unsigned *)varg;
+	size_t bt_count_prev, bt_count;
+	unsigned i_prev, i;
+
+	i_prev = 0;
+	bt_count_prev = 0;
+	for (i = 0; i < NALLOCS_PER_THREAD; i++) {
+		void *p = alloc_from_permuted_backtrace(thd_ind, i);
+		dallocx(p, 0);
+		if (i % DUMP_INTERVAL == 0) {
+			assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+			    0, "Unexpected error while dumping heap profile");
+		}
+
+		if (i % BT_COUNT_CHECK_INTERVAL == 0 ||
+		    i+1 == NALLOCS_PER_THREAD) {
+			bt_count = prof_bt_count();
+			assert_zu_le(bt_count_prev+(i-i_prev), bt_count,
+			    "Expected larger backtrace count increase");
+			i_prev = i;
+			bt_count_prev = bt_count;
+		}
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_idump) {
+	bool active;
+	thd_t thds[NTHREADS];
+	unsigned thd_args[NTHREADS];
+	unsigned i;
+
+	test_skip_if(!config_prof);
+
+	active = true;
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0,
+	    "Unexpected mallctl failure while activating profiling");
+
+	prof_dump_open = prof_dump_open_intercept;
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_args[i] = i;
+		thd_create(&thds[i], thd_start, (void *)&thd_args[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_idump);
+}
diff --git a/jemalloc/test/unit/prof_accum.sh b/jemalloc/test/unit/prof_accum.sh
new file mode 100644
index 0000000..b3e13fc
--- /dev/null
+++ b/jemalloc/test/unit/prof_accum.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_accum:true,prof_active:false,lg_prof_sample:0"
+fi
diff --git a/jemalloc/test/unit/prof_active.c b/jemalloc/test/unit/prof_active.c
new file mode 100644
index 0000000..850a24a
--- /dev/null
+++ b/jemalloc/test/unit/prof_active.c
@@ -0,0 +1,117 @@
+#include "test/jemalloc_test.h"
+
+static void
+mallctl_bool_get(const char *name, bool expected, const char *func, int line) {
+	bool old;
+	size_t sz;
+
+	sz = sizeof(old);
+	assert_d_eq(mallctl(name, (void *)&old, &sz, NULL, 0), 0,
+	    "%s():%d: Unexpected mallctl failure reading %s", func, line, name);
+	assert_b_eq(old, expected, "%s():%d: Unexpected %s value", func, line,
+	    name);
+}
+
+static void
+mallctl_bool_set(const char *name, bool old_expected, bool val_new,
+    const char *func, int line) {
+	bool old;
+	size_t sz;
+
+	sz = sizeof(old);
+	assert_d_eq(mallctl(name, (void *)&old, &sz, (void *)&val_new,
+	    sizeof(val_new)), 0,
+	    "%s():%d: Unexpected mallctl failure reading/writing %s", func,
+	    line, name);
+	assert_b_eq(old, old_expected, "%s():%d: Unexpected %s value", func,
+	    line, name);
+}
+
+static void
+mallctl_prof_active_get_impl(bool prof_active_old_expected, const char *func,
+    int line) {
+	mallctl_bool_get("prof.active", prof_active_old_expected, func, line);
+}
+#define mallctl_prof_active_get(a)					\
+	mallctl_prof_active_get_impl(a, __func__, __LINE__)
+
+static void
+mallctl_prof_active_set_impl(bool prof_active_old_expected,
+    bool prof_active_new, const char *func, int line) {
+	mallctl_bool_set("prof.active", prof_active_old_expected,
+	    prof_active_new, func, line);
+}
+#define mallctl_prof_active_set(a, b)					\
+	mallctl_prof_active_set_impl(a, b, __func__, __LINE__)
+
+static void
+mallctl_thread_prof_active_get_impl(bool thread_prof_active_old_expected,
+    const char *func, int line) {
+	mallctl_bool_get("thread.prof.active", thread_prof_active_old_expected,
+	    func, line);
+}
+#define mallctl_thread_prof_active_get(a)				\
+	mallctl_thread_prof_active_get_impl(a, __func__, __LINE__)
+
+static void
+mallctl_thread_prof_active_set_impl(bool thread_prof_active_old_expected,
+    bool thread_prof_active_new, const char *func, int line) {
+	mallctl_bool_set("thread.prof.active", thread_prof_active_old_expected,
+	    thread_prof_active_new, func, line);
+}
+#define mallctl_thread_prof_active_set(a, b)				\
+	mallctl_thread_prof_active_set_impl(a, b, __func__, __LINE__)
+
+static void
+prof_sampling_probe_impl(bool expect_sample, const char *func, int line) {
+	void *p;
+	size_t expected_backtraces = expect_sample ? 1 : 0;
+
+	assert_zu_eq(prof_bt_count(), 0, "%s():%d: Expected 0 backtraces", func,
+	    line);
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	assert_zu_eq(prof_bt_count(), expected_backtraces,
+	    "%s():%d: Unexpected backtrace count", func, line);
+	dallocx(p, 0);
+}
+#define prof_sampling_probe(a)						\
+	prof_sampling_probe_impl(a, __func__, __LINE__)
+
+TEST_BEGIN(test_prof_active) {
+	test_skip_if(!config_prof);
+
+	mallctl_prof_active_get(true);
+	mallctl_thread_prof_active_get(false);
+
+	mallctl_prof_active_set(true, true);
+	mallctl_thread_prof_active_set(false, false);
+	/* prof.active, !thread.prof.active. */
+	prof_sampling_probe(false);
+
+	mallctl_prof_active_set(true, false);
+	mallctl_thread_prof_active_set(false, false);
+	/* !prof.active, !thread.prof.active. */
+	prof_sampling_probe(false);
+
+	mallctl_prof_active_set(false, false);
+	mallctl_thread_prof_active_set(false, true);
+	/* !prof.active, thread.prof.active. */
+	prof_sampling_probe(false);
+
+	mallctl_prof_active_set(false, true);
+	mallctl_thread_prof_active_set(true, true);
+	/* prof.active, thread.prof.active. */
+	prof_sampling_probe(true);
+
+	/* Restore settings. */
+	mallctl_prof_active_set(true, true);
+	mallctl_thread_prof_active_set(true, false);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_prof_active);
+}
diff --git a/jemalloc/test/unit/prof_active.sh b/jemalloc/test/unit/prof_active.sh
new file mode 100644
index 0000000..0167cb1
--- /dev/null
+++ b/jemalloc/test/unit/prof_active.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_thread_active_init:false,lg_prof_sample:0"
+fi
diff --git a/jemalloc/test/unit/prof_gdump.c b/jemalloc/test/unit/prof_gdump.c
new file mode 100644
index 0000000..f7e0aac
--- /dev/null
+++ b/jemalloc/test/unit/prof_gdump.c
@@ -0,0 +1,74 @@
+#include "test/jemalloc_test.h"
+
+static bool did_prof_dump_open;
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	did_prof_dump_open = true;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+TEST_BEGIN(test_gdump) {
+	bool active, gdump, gdump_old;
+	void *p, *q, *r, *s;
+	size_t sz;
+
+	test_skip_if(!config_prof);
+
+	active = true;
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0,
+	    "Unexpected mallctl failure while activating profiling");
+
+	prof_dump_open = prof_dump_open_intercept;
+
+	did_prof_dump_open = false;
+	p = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+
+	did_prof_dump_open = false;
+	q = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+
+	gdump = false;
+	sz = sizeof(gdump_old);
+	assert_d_eq(mallctl("prof.gdump", (void *)&gdump_old, &sz,
+	    (void *)&gdump, sizeof(gdump)), 0,
+	    "Unexpected mallctl failure while disabling prof.gdump");
+	assert(gdump_old);
+	did_prof_dump_open = false;
+	r = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+	assert_false(did_prof_dump_open, "Unexpected profile dump");
+
+	gdump = true;
+	sz = sizeof(gdump_old);
+	assert_d_eq(mallctl("prof.gdump", (void *)&gdump_old, &sz,
+	    (void *)&gdump, sizeof(gdump)), 0,
+	    "Unexpected mallctl failure while enabling prof.gdump");
+	assert(!gdump_old);
+	did_prof_dump_open = false;
+	s = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+
+	dallocx(p, 0);
+	dallocx(q, 0);
+	dallocx(r, 0);
+	dallocx(s, 0);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_gdump);
+}
diff --git a/jemalloc/test/unit/prof_gdump.sh b/jemalloc/test/unit/prof_gdump.sh
new file mode 100644
index 0000000..3f600d2
--- /dev/null
+++ b/jemalloc/test/unit/prof_gdump.sh
@@ -0,0 +1,6 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_active:false,prof_gdump:true"
+fi
+
diff --git a/jemalloc/test/unit/prof_idump.c b/jemalloc/test/unit/prof_idump.c
new file mode 100644
index 0000000..1cc6c98
--- /dev/null
+++ b/jemalloc/test/unit/prof_idump.c
@@ -0,0 +1,42 @@
+#include "test/jemalloc_test.h"
+
+static bool did_prof_dump_open;
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	did_prof_dump_open = true;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+TEST_BEGIN(test_idump) {
+	bool active;
+	void *p;
+
+	test_skip_if(!config_prof);
+
+	active = true;
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0,
+	    "Unexpected mallctl failure while activating profiling");
+
+	prof_dump_open = prof_dump_open_intercept;
+
+	did_prof_dump_open = false;
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	dallocx(p, 0);
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_idump);
+}
diff --git a/jemalloc/test/unit/prof_idump.sh b/jemalloc/test/unit/prof_idump.sh
new file mode 100644
index 0000000..4dc599a
--- /dev/null
+++ b/jemalloc/test/unit/prof_idump.sh
@@ -0,0 +1,8 @@
+#!/bin/sh
+
+export MALLOC_CONF="tcache:false"
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="${MALLOC_CONF},prof:true,prof_accum:true,prof_active:false,lg_prof_sample:0,lg_prof_interval:0"
+fi
+
+
diff --git a/jemalloc/test/unit/prof_log.c b/jemalloc/test/unit/prof_log.c
new file mode 100644
index 0000000..92fbd7c
--- /dev/null
+++ b/jemalloc/test/unit/prof_log.c
@@ -0,0 +1,148 @@
+#include "test/jemalloc_test.h"
+
+#define N_PARAM 100
+#define N_THREADS 10
+
+static void assert_rep() {
+	assert_b_eq(prof_log_rep_check(), false, "Rep check failed");
+}
+
+static void assert_log_empty() {
+	assert_zu_eq(prof_log_bt_count(), 0,
+	    "The log has backtraces; it isn't empty");
+	assert_zu_eq(prof_log_thr_count(), 0,
+	    "The log has threads; it isn't empty");
+	assert_zu_eq(prof_log_alloc_count(), 0,
+	    "The log has allocations; it isn't empty");
+}
+
+void *buf[N_PARAM];
+
+static void f() {
+	int i;
+	for (i = 0; i < N_PARAM; i++) {
+		buf[i] = malloc(100);
+	}
+	for (i = 0; i < N_PARAM; i++) {
+		free(buf[i]);
+	}
+}
+
+TEST_BEGIN(test_prof_log_many_logs) {
+	int i;
+
+	test_skip_if(!config_prof);
+
+	for (i = 0; i < N_PARAM; i++) {
+		assert_b_eq(prof_log_is_logging(), false,
+		    "Logging shouldn't have started yet");
+		assert_d_eq(mallctl("prof.log_start", NULL, NULL, NULL, 0), 0,
+		    "Unexpected mallctl failure when starting logging");
+		assert_b_eq(prof_log_is_logging(), true,
+		    "Logging should be started by now");
+		assert_log_empty();
+		assert_rep();
+		f();
+		assert_zu_eq(prof_log_thr_count(), 1, "Wrong thread count");
+		assert_rep();
+		assert_b_eq(prof_log_is_logging(), true,
+		    "Logging should still be on");
+		assert_d_eq(mallctl("prof.log_stop", NULL, NULL, NULL, 0), 0,
+		    "Unexpected mallctl failure when stopping logging");
+		assert_b_eq(prof_log_is_logging(), false,
+		    "Logging should have turned off");
+	}
+}
+TEST_END
+
+thd_t thr_buf[N_THREADS];
+
+static void *f_thread(void *unused) {
+	int i;
+	for (i = 0; i < N_PARAM; i++) {
+		void *p = malloc(100);
+		memset(p, 100, sizeof(char));
+		free(p);
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_prof_log_many_threads) {
+
+	test_skip_if(!config_prof);
+
+	int i;
+	assert_d_eq(mallctl("prof.log_start", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when starting logging");
+	for (i = 0; i < N_THREADS; i++) {
+		thd_create(&thr_buf[i], &f_thread, NULL);
+	}
+
+	for (i = 0; i < N_THREADS; i++) {
+		thd_join(thr_buf[i], NULL);
+	}
+	assert_zu_eq(prof_log_thr_count(), N_THREADS,
+	    "Wrong number of thread entries");
+	assert_rep();
+	assert_d_eq(mallctl("prof.log_stop", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when stopping logging");
+}
+TEST_END
+
+static void f3() {
+	void *p = malloc(100);
+	free(p);
+}
+
+static void f1() {
+	void *p = malloc(100);
+	f3();
+	free(p);
+}
+
+static void f2() {
+	void *p = malloc(100);
+	free(p);
+}
+
+TEST_BEGIN(test_prof_log_many_traces) {
+
+	test_skip_if(!config_prof);
+
+	assert_d_eq(mallctl("prof.log_start", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when starting logging");
+	int i;
+	assert_rep();
+	assert_log_empty();
+	for (i = 0; i < N_PARAM; i++) {
+		assert_rep();
+		f1();
+		assert_rep();
+		f2();
+		assert_rep();
+		f3();
+		assert_rep();
+	}
+	/*
+	 * There should be 8 total backtraces: two for malloc/free in f1(), two
+	 * for malloc/free in f2(), two for malloc/free in f3(), and then two
+	 * for malloc/free in f1()'s call to f3().  However compiler
+	 * optimizations such as loop unrolling might generate more call sites.
+	 * So >= 8 traces are expected.
+	 */
+	assert_zu_ge(prof_log_bt_count(), 8,
+	    "Expect at least 8 backtraces given sample workload");
+	assert_d_eq(mallctl("prof.log_stop", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when stopping logging");
+}
+TEST_END
+
+int
+main(void) {
+	prof_log_dummy_set(true);
+	return test_no_reentrancy(
+	    test_prof_log_many_logs,
+	    test_prof_log_many_traces,
+	    test_prof_log_many_threads);
+}
diff --git a/jemalloc/test/unit/prof_log.sh b/jemalloc/test/unit/prof_log.sh
new file mode 100644
index 0000000..8fcc7d8
--- /dev/null
+++ b/jemalloc/test/unit/prof_log.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,lg_prof_sample:0"
+fi
diff --git a/jemalloc/test/unit/prof_reset.c b/jemalloc/test/unit/prof_reset.c
new file mode 100644
index 0000000..7cce42d
--- /dev/null
+++ b/jemalloc/test/unit/prof_reset.c
@@ -0,0 +1,286 @@
+#include "test/jemalloc_test.h"
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+static void
+set_prof_active(bool active) {
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0, "Unexpected mallctl failure");
+}
+
+static size_t
+get_lg_prof_sample(void) {
+	size_t lg_prof_sample;
+	size_t sz = sizeof(size_t);
+
+	assert_d_eq(mallctl("prof.lg_sample", (void *)&lg_prof_sample, &sz,
+	    NULL, 0), 0,
+	    "Unexpected mallctl failure while reading profiling sample rate");
+	return lg_prof_sample;
+}
+
+static void
+do_prof_reset(size_t lg_prof_sample) {
+	assert_d_eq(mallctl("prof.reset", NULL, NULL,
+	    (void *)&lg_prof_sample, sizeof(size_t)), 0,
+	    "Unexpected mallctl failure while resetting profile data");
+	assert_zu_eq(lg_prof_sample, get_lg_prof_sample(),
+	    "Expected profile sample rate change");
+}
+
+TEST_BEGIN(test_prof_reset_basic) {
+	size_t lg_prof_sample_orig, lg_prof_sample, lg_prof_sample_next;
+	size_t sz;
+	unsigned i;
+
+	test_skip_if(!config_prof);
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("opt.lg_prof_sample", (void *)&lg_prof_sample_orig,
+	    &sz, NULL, 0), 0,
+	    "Unexpected mallctl failure while reading profiling sample rate");
+	assert_zu_eq(lg_prof_sample_orig, 0,
+	    "Unexpected profiling sample rate");
+	lg_prof_sample = get_lg_prof_sample();
+	assert_zu_eq(lg_prof_sample_orig, lg_prof_sample,
+	    "Unexpected disagreement between \"opt.lg_prof_sample\" and "
+	    "\"prof.lg_sample\"");
+
+	/* Test simple resets. */
+	for (i = 0; i < 2; i++) {
+		assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0), 0,
+		    "Unexpected mallctl failure while resetting profile data");
+		lg_prof_sample = get_lg_prof_sample();
+		assert_zu_eq(lg_prof_sample_orig, lg_prof_sample,
+		    "Unexpected profile sample rate change");
+	}
+
+	/* Test resets with prof.lg_sample changes. */
+	lg_prof_sample_next = 1;
+	for (i = 0; i < 2; i++) {
+		do_prof_reset(lg_prof_sample_next);
+		lg_prof_sample = get_lg_prof_sample();
+		assert_zu_eq(lg_prof_sample, lg_prof_sample_next,
+		    "Expected profile sample rate change");
+		lg_prof_sample_next = lg_prof_sample_orig;
+	}
+
+	/* Make sure the test code restored prof.lg_sample. */
+	lg_prof_sample = get_lg_prof_sample();
+	assert_zu_eq(lg_prof_sample_orig, lg_prof_sample,
+	    "Unexpected disagreement between \"opt.lg_prof_sample\" and "
+	    "\"prof.lg_sample\"");
+}
+TEST_END
+
+bool prof_dump_header_intercepted = false;
+prof_cnt_t cnt_all_copy = {0, 0, 0, 0};
+static bool
+prof_dump_header_intercept(tsdn_t *tsdn, bool propagate_err,
+    const prof_cnt_t *cnt_all) {
+	prof_dump_header_intercepted = true;
+	memcpy(&cnt_all_copy, cnt_all, sizeof(prof_cnt_t));
+
+	return false;
+}
+
+TEST_BEGIN(test_prof_reset_cleanup) {
+	void *p;
+	prof_dump_header_t *prof_dump_header_orig;
+
+	test_skip_if(!config_prof);
+
+	set_prof_active(true);
+
+	assert_zu_eq(prof_bt_count(), 0, "Expected 0 backtraces");
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	assert_zu_eq(prof_bt_count(), 1, "Expected 1 backtrace");
+
+	prof_dump_header_orig = prof_dump_header;
+	prof_dump_header = prof_dump_header_intercept;
+	assert_false(prof_dump_header_intercepted, "Unexpected intercept");
+
+	assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+	    0, "Unexpected error while dumping heap profile");
+	assert_true(prof_dump_header_intercepted, "Expected intercept");
+	assert_u64_eq(cnt_all_copy.curobjs, 1, "Expected 1 allocation");
+
+	assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0), 0,
+	    "Unexpected error while resetting heap profile data");
+	assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+	    0, "Unexpected error while dumping heap profile");
+	assert_u64_eq(cnt_all_copy.curobjs, 0, "Expected 0 allocations");
+	assert_zu_eq(prof_bt_count(), 1, "Expected 1 backtrace");
+
+	prof_dump_header = prof_dump_header_orig;
+
+	dallocx(p, 0);
+	assert_zu_eq(prof_bt_count(), 0, "Expected 0 backtraces");
+
+	set_prof_active(false);
+}
+TEST_END
+
+#define NTHREADS		4
+#define NALLOCS_PER_THREAD	(1U << 13)
+#define OBJ_RING_BUF_COUNT	1531
+#define RESET_INTERVAL		(1U << 10)
+#define DUMP_INTERVAL		3677
+static void *
+thd_start(void *varg) {
+	unsigned thd_ind = *(unsigned *)varg;
+	unsigned i;
+	void *objs[OBJ_RING_BUF_COUNT];
+
+	memset(objs, 0, sizeof(objs));
+
+	for (i = 0; i < NALLOCS_PER_THREAD; i++) {
+		if (i % RESET_INTERVAL == 0) {
+			assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0),
+			    0, "Unexpected error while resetting heap profile "
+			    "data");
+		}
+
+		if (i % DUMP_INTERVAL == 0) {
+			assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+			    0, "Unexpected error while dumping heap profile");
+		}
+
+		{
+			void **pp = &objs[i % OBJ_RING_BUF_COUNT];
+			if (*pp != NULL) {
+				dallocx(*pp, 0);
+				*pp = NULL;
+			}
+			*pp = btalloc(1, thd_ind*NALLOCS_PER_THREAD + i);
+			assert_ptr_not_null(*pp,
+			    "Unexpected btalloc() failure");
+		}
+	}
+
+	/* Clean up any remaining objects. */
+	for (i = 0; i < OBJ_RING_BUF_COUNT; i++) {
+		void **pp = &objs[i % OBJ_RING_BUF_COUNT];
+		if (*pp != NULL) {
+			dallocx(*pp, 0);
+			*pp = NULL;
+		}
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_prof_reset) {
+	size_t lg_prof_sample_orig;
+	thd_t thds[NTHREADS];
+	unsigned thd_args[NTHREADS];
+	unsigned i;
+	size_t bt_count, tdata_count;
+
+	test_skip_if(!config_prof);
+
+	bt_count = prof_bt_count();
+	assert_zu_eq(bt_count, 0,
+	    "Unexpected pre-existing tdata structures");
+	tdata_count = prof_tdata_count();
+
+	lg_prof_sample_orig = get_lg_prof_sample();
+	do_prof_reset(5);
+
+	set_prof_active(true);
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_args[i] = i;
+		thd_create(&thds[i], thd_start, (void *)&thd_args[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+
+	assert_zu_eq(prof_bt_count(), bt_count,
+	    "Unexpected bactrace count change");
+	assert_zu_eq(prof_tdata_count(), tdata_count,
+	    "Unexpected remaining tdata structures");
+
+	set_prof_active(false);
+
+	do_prof_reset(lg_prof_sample_orig);
+}
+TEST_END
+#undef NTHREADS
+#undef NALLOCS_PER_THREAD
+#undef OBJ_RING_BUF_COUNT
+#undef RESET_INTERVAL
+#undef DUMP_INTERVAL
+
+/* Test sampling at the same allocation site across resets. */
+#define NITER 10
+TEST_BEGIN(test_xallocx) {
+	size_t lg_prof_sample_orig;
+	unsigned i;
+	void *ptrs[NITER];
+
+	test_skip_if(!config_prof);
+
+	lg_prof_sample_orig = get_lg_prof_sample();
+	set_prof_active(true);
+
+	/* Reset profiling. */
+	do_prof_reset(0);
+
+	for (i = 0; i < NITER; i++) {
+		void *p;
+		size_t sz, nsz;
+
+		/* Reset profiling. */
+		do_prof_reset(0);
+
+		/* Allocate small object (which will be promoted). */
+		p = ptrs[i] = mallocx(1, 0);
+		assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+		/* Reset profiling. */
+		do_prof_reset(0);
+
+		/* Perform successful xallocx(). */
+		sz = sallocx(p, 0);
+		assert_zu_eq(xallocx(p, sz, 0, 0), sz,
+		    "Unexpected xallocx() failure");
+
+		/* Perform unsuccessful xallocx(). */
+		nsz = nallocx(sz+1, 0);
+		assert_zu_eq(xallocx(p, nsz, 0, 0), sz,
+		    "Unexpected xallocx() success");
+	}
+
+	for (i = 0; i < NITER; i++) {
+		/* dallocx. */
+		dallocx(ptrs[i], 0);
+	}
+
+	set_prof_active(false);
+	do_prof_reset(lg_prof_sample_orig);
+}
+TEST_END
+#undef NITER
+
+int
+main(void) {
+	/* Intercept dumping prior to running any tests. */
+	prof_dump_open = prof_dump_open_intercept;
+
+	return test_no_reentrancy(
+	    test_prof_reset_basic,
+	    test_prof_reset_cleanup,
+	    test_prof_reset,
+	    test_xallocx);
+}
diff --git a/jemalloc/test/unit/prof_reset.sh b/jemalloc/test/unit/prof_reset.sh
new file mode 100644
index 0000000..43c516a
--- /dev/null
+++ b/jemalloc/test/unit/prof_reset.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_active:false,lg_prof_sample:0"
+fi
diff --git a/jemalloc/test/unit/prof_tctx.c b/jemalloc/test/unit/prof_tctx.c
new file mode 100644
index 0000000..ff3b2b0
--- /dev/null
+++ b/jemalloc/test/unit/prof_tctx.c
@@ -0,0 +1,46 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_prof_realloc) {
+	tsdn_t *tsdn;
+	int flags;
+	void *p, *q;
+	prof_tctx_t *tctx_p, *tctx_q;
+	uint64_t curobjs_0, curobjs_1, curobjs_2, curobjs_3;
+
+	test_skip_if(!config_prof);
+
+	tsdn = tsdn_fetch();
+	flags = MALLOCX_TCACHE_NONE;
+
+	prof_cnt_all(&curobjs_0, NULL, NULL, NULL);
+	p = mallocx(1024, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	tctx_p = prof_tctx_get(tsdn, p, NULL);
+	assert_ptr_ne(tctx_p, (prof_tctx_t *)(uintptr_t)1U,
+	    "Expected valid tctx");
+	prof_cnt_all(&curobjs_1, NULL, NULL, NULL);
+	assert_u64_eq(curobjs_0 + 1, curobjs_1,
+	    "Allocation should have increased sample size");
+
+	q = rallocx(p, 2048, flags);
+	assert_ptr_ne(p, q, "Expected move");
+	assert_ptr_not_null(p, "Unexpected rmallocx() failure");
+	tctx_q = prof_tctx_get(tsdn, q, NULL);
+	assert_ptr_ne(tctx_q, (prof_tctx_t *)(uintptr_t)1U,
+	    "Expected valid tctx");
+	prof_cnt_all(&curobjs_2, NULL, NULL, NULL);
+	assert_u64_eq(curobjs_1, curobjs_2,
+	    "Reallocation should not have changed sample size");
+
+	dallocx(q, flags);
+	prof_cnt_all(&curobjs_3, NULL, NULL, NULL);
+	assert_u64_eq(curobjs_0, curobjs_3,
+	    "Sample size should have returned to base level");
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_prof_realloc);
+}
diff --git a/jemalloc/test/unit/prof_tctx.sh b/jemalloc/test/unit/prof_tctx.sh
new file mode 100644
index 0000000..8fcc7d8
--- /dev/null
+++ b/jemalloc/test/unit/prof_tctx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,lg_prof_sample:0"
+fi
diff --git a/jemalloc/test/unit/prof_thread_name.c b/jemalloc/test/unit/prof_thread_name.c
new file mode 100644
index 0000000..c9c2a2b
--- /dev/null
+++ b/jemalloc/test/unit/prof_thread_name.c
@@ -0,0 +1,120 @@
+#include "test/jemalloc_test.h"
+
+static void
+mallctl_thread_name_get_impl(const char *thread_name_expected, const char *func,
+    int line) {
+	const char *thread_name_old;
+	size_t sz;
+
+	sz = sizeof(thread_name_old);
+	assert_d_eq(mallctl("thread.prof.name", (void *)&thread_name_old, &sz,
+	    NULL, 0), 0,
+	    "%s():%d: Unexpected mallctl failure reading thread.prof.name",
+	    func, line);
+	assert_str_eq(thread_name_old, thread_name_expected,
+	    "%s():%d: Unexpected thread.prof.name value", func, line);
+}
+#define mallctl_thread_name_get(a)					\
+	mallctl_thread_name_get_impl(a, __func__, __LINE__)
+
+static void
+mallctl_thread_name_set_impl(const char *thread_name, const char *func,
+    int line) {
+	assert_d_eq(mallctl("thread.prof.name", NULL, NULL,
+	    (void *)&thread_name, sizeof(thread_name)), 0,
+	    "%s():%d: Unexpected mallctl failure reading thread.prof.name",
+	    func, line);
+	mallctl_thread_name_get_impl(thread_name, func, line);
+}
+#define mallctl_thread_name_set(a)					\
+	mallctl_thread_name_set_impl(a, __func__, __LINE__)
+
+TEST_BEGIN(test_prof_thread_name_validation) {
+	const char *thread_name;
+
+	test_skip_if(!config_prof);
+
+	mallctl_thread_name_get("");
+	mallctl_thread_name_set("hi there");
+
+	/* NULL input shouldn't be allowed. */
+	thread_name = NULL;
+	assert_d_eq(mallctl("thread.prof.name", NULL, NULL,
+	    (void *)&thread_name, sizeof(thread_name)), EFAULT,
+	    "Unexpected mallctl result writing \"%s\" to thread.prof.name",
+	    thread_name);
+
+	/* '\n' shouldn't be allowed. */
+	thread_name = "hi\nthere";
+	assert_d_eq(mallctl("thread.prof.name", NULL, NULL,
+	    (void *)&thread_name, sizeof(thread_name)), EFAULT,
+	    "Unexpected mallctl result writing \"%s\" to thread.prof.name",
+	    thread_name);
+
+	/* Simultaneous read/write shouldn't be allowed. */
+	{
+		const char *thread_name_old;
+		size_t sz;
+
+		sz = sizeof(thread_name_old);
+		assert_d_eq(mallctl("thread.prof.name",
+		    (void *)&thread_name_old, &sz, (void *)&thread_name,
+		    sizeof(thread_name)), EPERM,
+		    "Unexpected mallctl result writing \"%s\" to "
+		    "thread.prof.name", thread_name);
+	}
+
+	mallctl_thread_name_set("");
+}
+TEST_END
+
+#define NTHREADS	4
+#define NRESET		25
+static void *
+thd_start(void *varg) {
+	unsigned thd_ind = *(unsigned *)varg;
+	char thread_name[16] = "";
+	unsigned i;
+
+	malloc_snprintf(thread_name, sizeof(thread_name), "thread %u", thd_ind);
+
+	mallctl_thread_name_get("");
+	mallctl_thread_name_set(thread_name);
+
+	for (i = 0; i < NRESET; i++) {
+		assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0), 0,
+		    "Unexpected error while resetting heap profile data");
+		mallctl_thread_name_get(thread_name);
+	}
+
+	mallctl_thread_name_set(thread_name);
+	mallctl_thread_name_set("");
+
+	return NULL;
+}
+
+TEST_BEGIN(test_prof_thread_name_threaded) {
+	thd_t thds[NTHREADS];
+	unsigned thd_args[NTHREADS];
+	unsigned i;
+
+	test_skip_if(!config_prof);
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_args[i] = i;
+		thd_create(&thds[i], thd_start, (void *)&thd_args[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+}
+TEST_END
+#undef NTHREADS
+#undef NRESET
+
+int
+main(void) {
+	return test(
+	    test_prof_thread_name_validation,
+	    test_prof_thread_name_threaded);
+}
diff --git a/jemalloc/test/unit/prof_thread_name.sh b/jemalloc/test/unit/prof_thread_name.sh
new file mode 100644
index 0000000..298c105
--- /dev/null
+++ b/jemalloc/test/unit/prof_thread_name.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_active:false"
+fi
diff --git a/jemalloc/test/unit/ql.c b/jemalloc/test/unit/ql.c
new file mode 100644
index 0000000..b76c24c
--- /dev/null
+++ b/jemalloc/test/unit/ql.c
@@ -0,0 +1,204 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ql.h"
+
+/* Number of ring entries, in [2..26]. */
+#define NENTRIES 9
+
+typedef struct list_s list_t;
+typedef ql_head(list_t) list_head_t;
+
+struct list_s {
+	ql_elm(list_t) link;
+	char id;
+};
+
+static void
+test_empty_list(list_head_t *head) {
+	list_t *t;
+	unsigned i;
+
+	assert_ptr_null(ql_first(head), "Unexpected element for empty list");
+	assert_ptr_null(ql_last(head, link),
+	    "Unexpected element for empty list");
+
+	i = 0;
+	ql_foreach(t, head, link) {
+		i++;
+	}
+	assert_u_eq(i, 0, "Unexpected element for empty list");
+
+	i = 0;
+	ql_reverse_foreach(t, head, link) {
+		i++;
+	}
+	assert_u_eq(i, 0, "Unexpected element for empty list");
+}
+
+TEST_BEGIN(test_ql_empty) {
+	list_head_t head;
+
+	ql_new(&head);
+	test_empty_list(&head);
+}
+TEST_END
+
+static void
+init_entries(list_t *entries, unsigned nentries) {
+	unsigned i;
+
+	for (i = 0; i < nentries; i++) {
+		entries[i].id = 'a' + i;
+		ql_elm_new(&entries[i], link);
+	}
+}
+
+static void
+test_entries_list(list_head_t *head, list_t *entries, unsigned nentries) {
+	list_t *t;
+	unsigned i;
+
+	assert_c_eq(ql_first(head)->id, entries[0].id, "Element id mismatch");
+	assert_c_eq(ql_last(head, link)->id, entries[nentries-1].id,
+	    "Element id mismatch");
+
+	i = 0;
+	ql_foreach(t, head, link) {
+		assert_c_eq(t->id, entries[i].id, "Element id mismatch");
+		i++;
+	}
+
+	i = 0;
+	ql_reverse_foreach(t, head, link) {
+		assert_c_eq(t->id, entries[nentries-i-1].id,
+		    "Element id mismatch");
+		i++;
+	}
+
+	for (i = 0; i < nentries-1; i++) {
+		t = ql_next(head, &entries[i], link);
+		assert_c_eq(t->id, entries[i+1].id, "Element id mismatch");
+	}
+	assert_ptr_null(ql_next(head, &entries[nentries-1], link),
+	    "Unexpected element");
+
+	assert_ptr_null(ql_prev(head, &entries[0], link), "Unexpected element");
+	for (i = 1; i < nentries; i++) {
+		t = ql_prev(head, &entries[i], link);
+		assert_c_eq(t->id, entries[i-1].id, "Element id mismatch");
+	}
+}
+
+TEST_BEGIN(test_ql_tail_insert) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_tail_insert(&head, &entries[i], link);
+	}
+
+	test_entries_list(&head, entries, NENTRIES);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_tail_remove) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_tail_insert(&head, &entries[i], link);
+	}
+
+	for (i = 0; i < NENTRIES; i++) {
+		test_entries_list(&head, entries, NENTRIES-i);
+		ql_tail_remove(&head, list_t, link);
+	}
+	test_empty_list(&head);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_head_insert) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_head_insert(&head, &entries[NENTRIES-i-1], link);
+	}
+
+	test_entries_list(&head, entries, NENTRIES);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_head_remove) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_head_insert(&head, &entries[NENTRIES-i-1], link);
+	}
+
+	for (i = 0; i < NENTRIES; i++) {
+		test_entries_list(&head, &entries[i], NENTRIES-i);
+		ql_head_remove(&head, list_t, link);
+	}
+	test_empty_list(&head);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_insert) {
+	list_head_t head;
+	list_t entries[8];
+	list_t *a, *b, *c, *d, *e, *f, *g, *h;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	a = &entries[0];
+	b = &entries[1];
+	c = &entries[2];
+	d = &entries[3];
+	e = &entries[4];
+	f = &entries[5];
+	g = &entries[6];
+	h = &entries[7];
+
+	/*
+	 * ql_remove(), ql_before_insert(), and ql_after_insert() are used
+	 * internally by other macros that are already tested, so there's no
+	 * need to test them completely.  However, insertion/deletion from the
+	 * middle of lists is not otherwise tested; do so here.
+	 */
+	ql_tail_insert(&head, f, link);
+	ql_before_insert(&head, f, b, link);
+	ql_before_insert(&head, f, c, link);
+	ql_after_insert(f, h, link);
+	ql_after_insert(f, g, link);
+	ql_before_insert(&head, b, a, link);
+	ql_after_insert(c, d, link);
+	ql_before_insert(&head, f, e, link);
+
+	test_entries_list(&head, entries, sizeof(entries)/sizeof(list_t));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ql_empty,
+	    test_ql_tail_insert,
+	    test_ql_tail_remove,
+	    test_ql_head_insert,
+	    test_ql_head_remove,
+	    test_ql_insert);
+}
diff --git a/jemalloc/test/unit/qr.c b/jemalloc/test/unit/qr.c
new file mode 100644
index 0000000..271a109
--- /dev/null
+++ b/jemalloc/test/unit/qr.c
@@ -0,0 +1,243 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/qr.h"
+
+/* Number of ring entries, in [2..26]. */
+#define NENTRIES 9
+/* Split index, in [1..NENTRIES). */
+#define SPLIT_INDEX 5
+
+typedef struct ring_s ring_t;
+
+struct ring_s {
+	qr(ring_t) link;
+	char id;
+};
+
+static void
+init_entries(ring_t *entries) {
+	unsigned i;
+
+	for (i = 0; i < NENTRIES; i++) {
+		qr_new(&entries[i], link);
+		entries[i].id = 'a' + i;
+	}
+}
+
+static void
+test_independent_entries(ring_t *entries) {
+	ring_t *t;
+	unsigned i, j;
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			j++;
+		}
+		assert_u_eq(j, 1,
+		    "Iteration over single-element ring should visit precisely "
+		    "one element");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			j++;
+		}
+		assert_u_eq(j, 1,
+		    "Iteration over single-element ring should visit precisely "
+		    "one element");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_next(&entries[i], link);
+		assert_ptr_eq(t, &entries[i],
+		    "Next element in single-element ring should be same as "
+		    "current element");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_prev(&entries[i], link);
+		assert_ptr_eq(t, &entries[i],
+		    "Previous element in single-element ring should be same as "
+		    "current element");
+	}
+}
+
+TEST_BEGIN(test_qr_one) {
+	ring_t entries[NENTRIES];
+
+	init_entries(entries);
+	test_independent_entries(entries);
+}
+TEST_END
+
+static void
+test_entries_ring(ring_t *entries) {
+	ring_t *t;
+	unsigned i, j;
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(i+j) % NENTRIES].id,
+			    "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(NENTRIES+i-j-1) %
+			    NENTRIES].id, "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_next(&entries[i], link);
+		assert_c_eq(t->id, entries[(i+1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_prev(&entries[i], link);
+		assert_c_eq(t->id, entries[(NENTRIES+i-1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+}
+
+TEST_BEGIN(test_qr_after_insert) {
+	ring_t entries[NENTRIES];
+	unsigned i;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_after_insert(&entries[i - 1], &entries[i], link);
+	}
+	test_entries_ring(entries);
+}
+TEST_END
+
+TEST_BEGIN(test_qr_remove) {
+	ring_t entries[NENTRIES];
+	ring_t *t;
+	unsigned i, j;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_after_insert(&entries[i - 1], &entries[i], link);
+	}
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[i+j].id,
+			    "Element id mismatch");
+			j++;
+		}
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[NENTRIES - 1 - j].id,
+			"Element id mismatch");
+			j++;
+		}
+		qr_remove(&entries[i], link);
+	}
+	test_independent_entries(entries);
+}
+TEST_END
+
+TEST_BEGIN(test_qr_before_insert) {
+	ring_t entries[NENTRIES];
+	ring_t *t;
+	unsigned i, j;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_before_insert(&entries[i - 1], &entries[i], link);
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(NENTRIES+i-j) %
+			    NENTRIES].id, "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(i+j+1) % NENTRIES].id,
+			    "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_next(&entries[i], link);
+		assert_c_eq(t->id, entries[(NENTRIES+i-1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_prev(&entries[i], link);
+		assert_c_eq(t->id, entries[(i+1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+}
+TEST_END
+
+static void
+test_split_entries(ring_t *entries) {
+	ring_t *t;
+	unsigned i, j;
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			if (i < SPLIT_INDEX) {
+				assert_c_eq(t->id,
+				    entries[(i+j) % SPLIT_INDEX].id,
+				    "Element id mismatch");
+			} else {
+				assert_c_eq(t->id, entries[(i+j-SPLIT_INDEX) %
+				    (NENTRIES-SPLIT_INDEX) + SPLIT_INDEX].id,
+				    "Element id mismatch");
+			}
+			j++;
+		}
+	}
+}
+
+TEST_BEGIN(test_qr_meld_split) {
+	ring_t entries[NENTRIES];
+	unsigned i;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_after_insert(&entries[i - 1], &entries[i], link);
+	}
+
+	qr_split(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_split_entries(entries);
+
+	qr_meld(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_entries_ring(entries);
+
+	qr_meld(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_split_entries(entries);
+
+	qr_split(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_entries_ring(entries);
+
+	qr_split(&entries[0], &entries[0], ring_t, link);
+	test_entries_ring(entries);
+
+	qr_meld(&entries[0], &entries[0], ring_t, link);
+	test_entries_ring(entries);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_qr_one,
+	    test_qr_after_insert,
+	    test_qr_remove,
+	    test_qr_before_insert,
+	    test_qr_meld_split);
+}
diff --git a/jemalloc/test/unit/rb.c b/jemalloc/test/unit/rb.c
new file mode 100644
index 0000000..65c0492
--- /dev/null
+++ b/jemalloc/test/unit/rb.c
@@ -0,0 +1,355 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/rb.h"
+
+#define rbtn_black_height(a_type, a_field, a_rbt, r_height) do {	\
+	a_type *rbp_bh_t;						\
+	for (rbp_bh_t = (a_rbt)->rbt_root, (r_height) = 0; rbp_bh_t !=	\
+	    NULL; rbp_bh_t = rbtn_left_get(a_type, a_field,		\
+	    rbp_bh_t)) {						\
+		if (!rbtn_red_get(a_type, a_field, rbp_bh_t)) {		\
+		(r_height)++;						\
+		}							\
+	}								\
+} while (0)
+
+typedef struct node_s node_t;
+
+struct node_s {
+#define NODE_MAGIC 0x9823af7e
+	uint32_t magic;
+	rb_node(node_t) link;
+	uint64_t key;
+};
+
+static int
+node_cmp(const node_t *a, const node_t *b) {
+	int ret;
+
+	assert_u32_eq(a->magic, NODE_MAGIC, "Bad magic");
+	assert_u32_eq(b->magic, NODE_MAGIC, "Bad magic");
+
+	ret = (a->key > b->key) - (a->key < b->key);
+	if (ret == 0) {
+		/*
+		 * Duplicates are not allowed in the tree, so force an
+		 * arbitrary ordering for non-identical items with equal keys.
+		 */
+		ret = (((uintptr_t)a) > ((uintptr_t)b))
+		    - (((uintptr_t)a) < ((uintptr_t)b));
+	}
+	return ret;
+}
+
+typedef rb_tree(node_t) tree_t;
+rb_gen(static, tree_, tree_t, node_t, link, node_cmp);
+
+TEST_BEGIN(test_rb_empty) {
+	tree_t tree;
+	node_t key;
+
+	tree_new(&tree);
+
+	assert_true(tree_empty(&tree), "Tree should be empty");
+	assert_ptr_null(tree_first(&tree), "Unexpected node");
+	assert_ptr_null(tree_last(&tree), "Unexpected node");
+
+	key.key = 0;
+	key.magic = NODE_MAGIC;
+	assert_ptr_null(tree_search(&tree, &key), "Unexpected node");
+
+	key.key = 0;
+	key.magic = NODE_MAGIC;
+	assert_ptr_null(tree_nsearch(&tree, &key), "Unexpected node");
+
+	key.key = 0;
+	key.magic = NODE_MAGIC;
+	assert_ptr_null(tree_psearch(&tree, &key), "Unexpected node");
+}
+TEST_END
+
+static unsigned
+tree_recurse(node_t *node, unsigned black_height, unsigned black_depth) {
+	unsigned ret = 0;
+	node_t *left_node;
+	node_t *right_node;
+
+	if (node == NULL) {
+		return ret;
+	}
+
+	left_node = rbtn_left_get(node_t, link, node);
+	right_node = rbtn_right_get(node_t, link, node);
+
+	if (!rbtn_red_get(node_t, link, node)) {
+		black_depth++;
+	}
+
+	/* Red nodes must be interleaved with black nodes. */
+	if (rbtn_red_get(node_t, link, node)) {
+		if (left_node != NULL) {
+			assert_false(rbtn_red_get(node_t, link, left_node),
+				"Node should be black");
+		}
+		if (right_node != NULL) {
+			assert_false(rbtn_red_get(node_t, link, right_node),
+			    "Node should be black");
+		}
+	}
+
+	/* Self. */
+	assert_u32_eq(node->magic, NODE_MAGIC, "Bad magic");
+
+	/* Left subtree. */
+	if (left_node != NULL) {
+		ret += tree_recurse(left_node, black_height, black_depth);
+	} else {
+		ret += (black_depth != black_height);
+	}
+
+	/* Right subtree. */
+	if (right_node != NULL) {
+		ret += tree_recurse(right_node, black_height, black_depth);
+	} else {
+		ret += (black_depth != black_height);
+	}
+
+	return ret;
+}
+
+static node_t *
+tree_iterate_cb(tree_t *tree, node_t *node, void *data) {
+	unsigned *i = (unsigned *)data;
+	node_t *search_node;
+
+	assert_u32_eq(node->magic, NODE_MAGIC, "Bad magic");
+
+	/* Test rb_search(). */
+	search_node = tree_search(tree, node);
+	assert_ptr_eq(search_node, node,
+	    "tree_search() returned unexpected node");
+
+	/* Test rb_nsearch(). */
+	search_node = tree_nsearch(tree, node);
+	assert_ptr_eq(search_node, node,
+	    "tree_nsearch() returned unexpected node");
+
+	/* Test rb_psearch(). */
+	search_node = tree_psearch(tree, node);
+	assert_ptr_eq(search_node, node,
+	    "tree_psearch() returned unexpected node");
+
+	(*i)++;
+
+	return NULL;
+}
+
+static unsigned
+tree_iterate(tree_t *tree) {
+	unsigned i;
+
+	i = 0;
+	tree_iter(tree, NULL, tree_iterate_cb, (void *)&i);
+
+	return i;
+}
+
+static unsigned
+tree_iterate_reverse(tree_t *tree) {
+	unsigned i;
+
+	i = 0;
+	tree_reverse_iter(tree, NULL, tree_iterate_cb, (void *)&i);
+
+	return i;
+}
+
+static void
+node_remove(tree_t *tree, node_t *node, unsigned nnodes) {
+	node_t *search_node;
+	unsigned black_height, imbalances;
+
+	tree_remove(tree, node);
+
+	/* Test rb_nsearch(). */
+	search_node = tree_nsearch(tree, node);
+	if (search_node != NULL) {
+		assert_u64_ge(search_node->key, node->key,
+		    "Key ordering error");
+	}
+
+	/* Test rb_psearch(). */
+	search_node = tree_psearch(tree, node);
+	if (search_node != NULL) {
+		assert_u64_le(search_node->key, node->key,
+		    "Key ordering error");
+	}
+
+	node->magic = 0;
+
+	rbtn_black_height(node_t, link, tree, black_height);
+	imbalances = tree_recurse(tree->rbt_root, black_height, 0);
+	assert_u_eq(imbalances, 0, "Tree is unbalanced");
+	assert_u_eq(tree_iterate(tree), nnodes-1,
+	    "Unexpected node iteration count");
+	assert_u_eq(tree_iterate_reverse(tree), nnodes-1,
+	    "Unexpected node iteration count");
+}
+
+static node_t *
+remove_iterate_cb(tree_t *tree, node_t *node, void *data) {
+	unsigned *nnodes = (unsigned *)data;
+	node_t *ret = tree_next(tree, node);
+
+	node_remove(tree, node, *nnodes);
+
+	return ret;
+}
+
+static node_t *
+remove_reverse_iterate_cb(tree_t *tree, node_t *node, void *data) {
+	unsigned *nnodes = (unsigned *)data;
+	node_t *ret = tree_prev(tree, node);
+
+	node_remove(tree, node, *nnodes);
+
+	return ret;
+}
+
+static void
+destroy_cb(node_t *node, void *data) {
+	unsigned *nnodes = (unsigned *)data;
+
+	assert_u_gt(*nnodes, 0, "Destruction removed too many nodes");
+	(*nnodes)--;
+}
+
+TEST_BEGIN(test_rb_random) {
+#define NNODES 25
+#define NBAGS 250
+#define SEED 42
+	sfmt_t *sfmt;
+	uint64_t bag[NNODES];
+	tree_t tree;
+	node_t nodes[NNODES];
+	unsigned i, j, k, black_height, imbalances;
+
+	sfmt = init_gen_rand(SEED);
+	for (i = 0; i < NBAGS; i++) {
+		switch (i) {
+		case 0:
+			/* Insert in order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = j;
+			}
+			break;
+		case 1:
+			/* Insert in reverse order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = NNODES - j - 1;
+			}
+			break;
+		default:
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = gen_rand64_range(sfmt, NNODES);
+			}
+		}
+
+		for (j = 1; j <= NNODES; j++) {
+			/* Initialize tree and nodes. */
+			tree_new(&tree);
+			for (k = 0; k < j; k++) {
+				nodes[k].magic = NODE_MAGIC;
+				nodes[k].key = bag[k];
+			}
+
+			/* Insert nodes. */
+			for (k = 0; k < j; k++) {
+				tree_insert(&tree, &nodes[k]);
+
+				rbtn_black_height(node_t, link, &tree,
+				    black_height);
+				imbalances = tree_recurse(tree.rbt_root,
+				    black_height, 0);
+				assert_u_eq(imbalances, 0,
+				    "Tree is unbalanced");
+
+				assert_u_eq(tree_iterate(&tree), k+1,
+				    "Unexpected node iteration count");
+				assert_u_eq(tree_iterate_reverse(&tree), k+1,
+				    "Unexpected node iteration count");
+
+				assert_false(tree_empty(&tree),
+				    "Tree should not be empty");
+				assert_ptr_not_null(tree_first(&tree),
+				    "Tree should not be empty");
+				assert_ptr_not_null(tree_last(&tree),
+				    "Tree should not be empty");
+
+				tree_next(&tree, &nodes[k]);
+				tree_prev(&tree, &nodes[k]);
+			}
+
+			/* Remove nodes. */
+			switch (i % 5) {
+			case 0:
+				for (k = 0; k < j; k++) {
+					node_remove(&tree, &nodes[k], j - k);
+				}
+				break;
+			case 1:
+				for (k = j; k > 0; k--) {
+					node_remove(&tree, &nodes[k-1], k);
+				}
+				break;
+			case 2: {
+				node_t *start;
+				unsigned nnodes = j;
+
+				start = NULL;
+				do {
+					start = tree_iter(&tree, start,
+					    remove_iterate_cb, (void *)&nnodes);
+					nnodes--;
+				} while (start != NULL);
+				assert_u_eq(nnodes, 0,
+				    "Removal terminated early");
+				break;
+			} case 3: {
+				node_t *start;
+				unsigned nnodes = j;
+
+				start = NULL;
+				do {
+					start = tree_reverse_iter(&tree, start,
+					    remove_reverse_iterate_cb,
+					    (void *)&nnodes);
+					nnodes--;
+				} while (start != NULL);
+				assert_u_eq(nnodes, 0,
+				    "Removal terminated early");
+				break;
+			} case 4: {
+				unsigned nnodes = j;
+				tree_destroy(&tree, destroy_cb, &nnodes);
+				assert_u_eq(nnodes, 0,
+				    "Destruction terminated early");
+				break;
+			} default:
+				not_reached();
+			}
+		}
+	}
+	fini_gen_rand(sfmt);
+#undef NNODES
+#undef NBAGS
+#undef SEED
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_rb_empty,
+	    test_rb_random);
+}
diff --git a/jemalloc/test/unit/retained.c b/jemalloc/test/unit/retained.c
new file mode 100644
index 0000000..7993fd3
--- /dev/null
+++ b/jemalloc/test/unit/retained.c
@@ -0,0 +1,184 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/spin.h"
+
+static unsigned		arena_ind;
+static size_t		sz;
+static size_t		esz;
+#define NEPOCHS		8
+#define PER_THD_NALLOCS	1
+static atomic_u_t	epoch;
+static atomic_u_t	nfinished;
+
+static unsigned
+do_arena_create(extent_hooks_t *h) {
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz,
+	    (void *)(h != NULL ? &h : NULL), (h != NULL ? sizeof(h) : 0)), 0,
+	    "Unexpected mallctl() failure");
+	return arena_ind;
+}
+
+static void
+do_arena_destroy(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.destroy", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+static void
+do_refresh(void) {
+	uint64_t epoch = 1;
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)), 0, "Unexpected mallctl() failure");
+}
+
+static size_t
+do_get_size_impl(const char *cmd, unsigned arena_ind) {
+	size_t mib[4];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+	size_t z = sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = arena_ind;
+	size_t size;
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&size, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\"], ...) failure", cmd);
+
+	return size;
+}
+
+static size_t
+do_get_active(unsigned arena_ind) {
+	return do_get_size_impl("stats.arenas.0.pactive", arena_ind) * PAGE;
+}
+
+static size_t
+do_get_mapped(unsigned arena_ind) {
+	return do_get_size_impl("stats.arenas.0.mapped", arena_ind);
+}
+
+static void *
+thd_start(void *arg) {
+	for (unsigned next_epoch = 1; next_epoch < NEPOCHS; next_epoch++) {
+		/* Busy-wait for next epoch. */
+		unsigned cur_epoch;
+		spin_t spinner = SPIN_INITIALIZER;
+		while ((cur_epoch = atomic_load_u(&epoch, ATOMIC_ACQUIRE)) !=
+		    next_epoch) {
+			spin_adaptive(&spinner);
+		}
+		assert_u_eq(cur_epoch, next_epoch, "Unexpected epoch");
+
+		/*
+		 * Allocate.  The main thread will reset the arena, so there's
+		 * no need to deallocate.
+		 */
+		for (unsigned i = 0; i < PER_THD_NALLOCS; i++) {
+			void *p = mallocx(sz, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE
+			    );
+			assert_ptr_not_null(p,
+			    "Unexpected mallocx() failure\n");
+		}
+
+		/* Let the main thread know we've finished this iteration. */
+		atomic_fetch_add_u(&nfinished, 1, ATOMIC_RELEASE);
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_retained) {
+	test_skip_if(!config_stats);
+
+	arena_ind = do_arena_create(NULL);
+	sz = nallocx(HUGEPAGE, 0);
+	esz = sz + sz_large_pad;
+
+	atomic_store_u(&epoch, 0, ATOMIC_RELAXED);
+
+	unsigned nthreads = ncpus * 2;
+	if (LG_SIZEOF_PTR < 3 && nthreads > 16) {
+		nthreads = 16; /* 32-bit platform could run out of vaddr. */
+	}
+	VARIABLE_ARRAY(thd_t, threads, nthreads);
+	for (unsigned i = 0; i < nthreads; i++) {
+		thd_create(&threads[i], thd_start, NULL);
+	}
+
+	for (unsigned e = 1; e < NEPOCHS; e++) {
+		atomic_store_u(&nfinished, 0, ATOMIC_RELEASE);
+		atomic_store_u(&epoch, e, ATOMIC_RELEASE);
+
+		/* Wait for threads to finish allocating. */
+		spin_t spinner = SPIN_INITIALIZER;
+		while (atomic_load_u(&nfinished, ATOMIC_ACQUIRE) < nthreads) {
+			spin_adaptive(&spinner);
+		}
+
+		/*
+		 * Assert that retained is no more than the sum of size classes
+		 * that should have been used to satisfy the worker threads'
+		 * requests, discounting per growth fragmentation.
+		 */
+		do_refresh();
+
+		size_t allocated = esz * nthreads * PER_THD_NALLOCS;
+		size_t active = do_get_active(arena_ind);
+		assert_zu_le(allocated, active, "Unexpected active memory");
+		size_t mapped = do_get_mapped(arena_ind);
+		assert_zu_le(active, mapped, "Unexpected mapped memory");
+
+		arena_t *arena = arena_get(tsdn_fetch(), arena_ind, false);
+		size_t usable = 0;
+		size_t fragmented = 0;
+		for (pszind_t pind = sz_psz2ind(HUGEPAGE); pind <
+		    arena->extent_grow_next; pind++) {
+			size_t psz = sz_pind2sz(pind);
+			size_t psz_fragmented = psz % esz;
+			size_t psz_usable = psz - psz_fragmented;
+			/*
+			 * Only consider size classes that wouldn't be skipped.
+			 */
+			if (psz_usable > 0) {
+				assert_zu_lt(usable, allocated,
+				    "Excessive retained memory "
+				    "(%#zx[+%#zx] > %#zx)", usable, psz_usable,
+				    allocated);
+				fragmented += psz_fragmented;
+				usable += psz_usable;
+			}
+		}
+
+		/*
+		 * Clean up arena.  Destroying and recreating the arena
+		 * is simpler that specifying extent hooks that deallocate
+		 * (rather than retaining) during reset.
+		 */
+		do_arena_destroy(arena_ind);
+		assert_u_eq(do_arena_create(NULL), arena_ind,
+		    "Unexpected arena index");
+	}
+
+	for (unsigned i = 0; i < nthreads; i++) {
+		thd_join(threads[i], NULL);
+	}
+
+	do_arena_destroy(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_retained);
+}
diff --git a/jemalloc/test/unit/rtree.c b/jemalloc/test/unit/rtree.c
new file mode 100644
index 0000000..90adca1
--- /dev/null
+++ b/jemalloc/test/unit/rtree.c
@@ -0,0 +1,228 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/rtree.h"
+
+rtree_node_alloc_t *rtree_node_alloc_orig;
+rtree_node_dalloc_t *rtree_node_dalloc_orig;
+rtree_leaf_alloc_t *rtree_leaf_alloc_orig;
+rtree_leaf_dalloc_t *rtree_leaf_dalloc_orig;
+
+/* Potentially too large to safely place on the stack. */
+rtree_t test_rtree;
+
+static rtree_node_elm_t *
+rtree_node_alloc_intercept(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	rtree_node_elm_t *node;
+
+	if (rtree != &test_rtree) {
+		return rtree_node_alloc_orig(tsdn, rtree, nelms);
+	}
+
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+	node = (rtree_node_elm_t *)calloc(nelms, sizeof(rtree_node_elm_t));
+	assert_ptr_not_null(node, "Unexpected calloc() failure");
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+
+	return node;
+}
+
+static void
+rtree_node_dalloc_intercept(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_node_elm_t *node) {
+	if (rtree != &test_rtree) {
+		rtree_node_dalloc_orig(tsdn, rtree, node);
+		return;
+	}
+
+	free(node);
+}
+
+static rtree_leaf_elm_t *
+rtree_leaf_alloc_intercept(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	rtree_leaf_elm_t *leaf;
+
+	if (rtree != &test_rtree) {
+		return rtree_leaf_alloc_orig(tsdn, rtree, nelms);
+	}
+
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+	leaf = (rtree_leaf_elm_t *)calloc(nelms, sizeof(rtree_leaf_elm_t));
+	assert_ptr_not_null(leaf, "Unexpected calloc() failure");
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+
+	return leaf;
+}
+
+static void
+rtree_leaf_dalloc_intercept(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *leaf) {
+	if (rtree != &test_rtree) {
+		rtree_leaf_dalloc_orig(tsdn, rtree, leaf);
+		return;
+	}
+
+	free(leaf);
+}
+
+TEST_BEGIN(test_rtree_read_empty) {
+	tsdn_t *tsdn;
+
+	tsdn = tsdn_fetch();
+
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+	assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx, PAGE,
+	    false), "rtree_extent_read() should return NULL for empty tree");
+	rtree_delete(tsdn, rtree);
+}
+TEST_END
+
+#undef NTHREADS
+#undef NITERS
+#undef SEED
+
+TEST_BEGIN(test_rtree_extrema) {
+	extent_t extent_a, extent_b;
+	extent_init(&extent_a, NULL, NULL, SC_LARGE_MINCLASS, false,
+	    sz_size2index(SC_LARGE_MINCLASS), 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+	extent_init(&extent_b, NULL, NULL, 0, false, SC_NSIZES, 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+
+	tsdn_t *tsdn = tsdn_fetch();
+
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+
+	assert_false(rtree_write(tsdn, rtree, &rtree_ctx, PAGE, &extent_a,
+	    extent_szind_get(&extent_a), extent_slab_get(&extent_a)),
+	    "Unexpected rtree_write() failure");
+	rtree_szind_slab_update(tsdn, rtree, &rtree_ctx, PAGE,
+	    extent_szind_get(&extent_a), extent_slab_get(&extent_a));
+	assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx, PAGE, true),
+	    &extent_a,
+	    "rtree_extent_read() should return previously set value");
+
+	assert_false(rtree_write(tsdn, rtree, &rtree_ctx, ~((uintptr_t)0),
+	    &extent_b, extent_szind_get_maybe_invalid(&extent_b),
+	    extent_slab_get(&extent_b)), "Unexpected rtree_write() failure");
+	assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+	    ~((uintptr_t)0), true), &extent_b,
+	    "rtree_extent_read() should return previously set value");
+
+	rtree_delete(tsdn, rtree);
+}
+TEST_END
+
+TEST_BEGIN(test_rtree_bits) {
+	tsdn_t *tsdn = tsdn_fetch();
+
+	uintptr_t keys[] = {PAGE, PAGE + 1,
+	    PAGE + (((uintptr_t)1) << LG_PAGE) - 1};
+
+	extent_t extent;
+	extent_init(&extent, NULL, NULL, 0, false, SC_NSIZES, 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+
+	for (unsigned i = 0; i < sizeof(keys)/sizeof(uintptr_t); i++) {
+		assert_false(rtree_write(tsdn, rtree, &rtree_ctx, keys[i],
+		    &extent, SC_NSIZES, false),
+		    "Unexpected rtree_write() failure");
+		for (unsigned j = 0; j < sizeof(keys)/sizeof(uintptr_t); j++) {
+			assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+			    keys[j], true), &extent,
+			    "rtree_extent_read() should return previously set "
+			    "value and ignore insignificant key bits; i=%u, "
+			    "j=%u, set key=%#"FMTxPTR", get key=%#"FMTxPTR, i,
+			    j, keys[i], keys[j]);
+		}
+		assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    (((uintptr_t)2) << LG_PAGE), false),
+		    "Only leftmost rtree leaf should be set; i=%u", i);
+		rtree_clear(tsdn, rtree, &rtree_ctx, keys[i]);
+	}
+
+	rtree_delete(tsdn, rtree);
+}
+TEST_END
+
+TEST_BEGIN(test_rtree_random) {
+#define NSET 16
+#define SEED 42
+	sfmt_t *sfmt = init_gen_rand(SEED);
+	tsdn_t *tsdn = tsdn_fetch();
+	uintptr_t keys[NSET];
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+
+	extent_t extent;
+	extent_init(&extent, NULL, NULL, 0, false, SC_NSIZES, 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+
+	for (unsigned i = 0; i < NSET; i++) {
+		keys[i] = (uintptr_t)gen_rand64(sfmt);
+		rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, rtree,
+		    &rtree_ctx, keys[i], false, true);
+		assert_ptr_not_null(elm,
+		    "Unexpected rtree_leaf_elm_lookup() failure");
+		rtree_leaf_elm_write(tsdn, rtree, elm, &extent, SC_NSIZES,
+		    false);
+		assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true), &extent,
+		    "rtree_extent_read() should return previously set value");
+	}
+	for (unsigned i = 0; i < NSET; i++) {
+		assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true), &extent,
+		    "rtree_extent_read() should return previously set value, "
+		    "i=%u", i);
+	}
+
+	for (unsigned i = 0; i < NSET; i++) {
+		rtree_clear(tsdn, rtree, &rtree_ctx, keys[i]);
+		assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true),
+		   "rtree_extent_read() should return previously set value");
+	}
+	for (unsigned i = 0; i < NSET; i++) {
+		assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true),
+		    "rtree_extent_read() should return previously set value");
+	}
+
+	rtree_delete(tsdn, rtree);
+	fini_gen_rand(sfmt);
+#undef NSET
+#undef SEED
+}
+TEST_END
+
+int
+main(void) {
+	rtree_node_alloc_orig = rtree_node_alloc;
+	rtree_node_alloc = rtree_node_alloc_intercept;
+	rtree_node_dalloc_orig = rtree_node_dalloc;
+	rtree_node_dalloc = rtree_node_dalloc_intercept;
+	rtree_leaf_alloc_orig = rtree_leaf_alloc;
+	rtree_leaf_alloc = rtree_leaf_alloc_intercept;
+	rtree_leaf_dalloc_orig = rtree_leaf_dalloc;
+	rtree_leaf_dalloc = rtree_leaf_dalloc_intercept;
+
+	return test(
+	    test_rtree_read_empty,
+	    test_rtree_extrema,
+	    test_rtree_bits,
+	    test_rtree_random);
+}
diff --git a/jemalloc/test/unit/safety_check.c b/jemalloc/test/unit/safety_check.c
new file mode 100644
index 0000000..bf4bd86
--- /dev/null
+++ b/jemalloc/test/unit/safety_check.c
@@ -0,0 +1,156 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/safety_check.h"
+
+/*
+ * Note that we get called through safety_check.sh, which turns on sampling for
+ * everything.
+ */
+
+bool fake_abort_called;
+void fake_abort(const char *message) {
+	(void)message;
+	fake_abort_called = true;
+}
+
+TEST_BEGIN(test_malloc_free_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	free(ptr);
+	safety_check_set_abort(NULL);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_mallocx_dallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = mallocx(128, 0);
+	ptr[128] = 0;
+	dallocx(ptr, 0);
+	safety_check_set_abort(NULL);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_sdallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	sdallocx(ptr, 128, 0);
+	safety_check_set_abort(NULL);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_realloc_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	ptr = realloc(ptr, 129);
+	safety_check_set_abort(NULL);
+	free(ptr);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_rallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	ptr = rallocx(ptr, 129, 0);
+	safety_check_set_abort(NULL);
+	free(ptr);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_xallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	size_t result = xallocx(ptr, 129, 0, 0);
+	assert_zu_eq(result, 128, "");
+	free(ptr);
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+	safety_check_set_abort(NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_realloc_no_overflow) {
+	char* ptr = malloc(128);
+	ptr = realloc(ptr, 256);
+	ptr[128] = 0;
+	ptr[255] = 0;
+	free(ptr);
+
+	ptr = malloc(128);
+	ptr = realloc(ptr, 64);
+	ptr[63] = 0;
+	ptr[0] = 0;
+	free(ptr);
+}
+TEST_END
+
+TEST_BEGIN(test_rallocx_no_overflow) {
+	char* ptr = malloc(128);
+	ptr = rallocx(ptr, 256, 0);
+	ptr[128] = 0;
+	ptr[255] = 0;
+	free(ptr);
+
+	ptr = malloc(128);
+	ptr = rallocx(ptr, 64, 0);
+	ptr[63] = 0;
+	ptr[0] = 0;
+	free(ptr);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_malloc_free_overflow,
+	    test_mallocx_dallocx_overflow,
+	    test_malloc_sdallocx_overflow,
+	    test_realloc_overflow,
+	    test_rallocx_overflow,
+	    test_xallocx_overflow,
+	    test_realloc_no_overflow,
+	    test_rallocx_no_overflow);
+}
diff --git a/jemalloc/test/unit/safety_check.sh b/jemalloc/test/unit/safety_check.sh
new file mode 100644
index 0000000..8fcc7d8
--- /dev/null
+++ b/jemalloc/test/unit/safety_check.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,lg_prof_sample:0"
+fi
diff --git a/jemalloc/test/unit/sc.c b/jemalloc/test/unit/sc.c
new file mode 100644
index 0000000..bf51d8e
--- /dev/null
+++ b/jemalloc/test/unit/sc.c
@@ -0,0 +1,33 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_update_slab_size) {
+	sc_data_t data;
+	memset(&data, 0, sizeof(data));
+	sc_data_init(&data);
+	sc_t *tiny = &data.sc[0];
+	size_t tiny_size = (ZU(1) << tiny->lg_base)
+	    + (ZU(tiny->ndelta) << tiny->lg_delta);
+	size_t pgs_too_big = (tiny_size * BITMAP_MAXBITS + PAGE - 1) / PAGE + 1;
+	sc_data_update_slab_size(&data, tiny_size, tiny_size, (int)pgs_too_big);
+	assert_zu_lt((size_t)tiny->pgs, pgs_too_big, "Allowed excessive pages");
+
+	sc_data_update_slab_size(&data, 1, 10 * PAGE, 1);
+	for (int i = 0; i < data.nbins; i++) {
+		sc_t *sc = &data.sc[i];
+		size_t reg_size = (ZU(1) << sc->lg_base)
+		    + (ZU(sc->ndelta) << sc->lg_delta);
+		if (reg_size <= PAGE) {
+			assert_d_eq(sc->pgs, 1, "Ignored valid page size hint");
+		} else {
+			assert_d_gt(sc->pgs, 1,
+			    "Allowed invalid page size hint");
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_update_slab_size);
+}
diff --git a/jemalloc/test/unit/seq.c b/jemalloc/test/unit/seq.c
new file mode 100644
index 0000000..19613b0
--- /dev/null
+++ b/jemalloc/test/unit/seq.c
@@ -0,0 +1,95 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/seq.h"
+
+typedef struct data_s data_t;
+struct data_s {
+	int arr[10];
+};
+
+static void
+set_data(data_t *data, int num) {
+	for (int i = 0; i < 10; i++) {
+		data->arr[i] = num;
+	}
+}
+
+static void
+assert_data(data_t *data) {
+	int num = data->arr[0];
+	for (int i = 0; i < 10; i++) {
+		assert_d_eq(num, data->arr[i], "Data consistency error");
+	}
+}
+
+seq_define(data_t, data)
+
+typedef struct thd_data_s thd_data_t;
+struct thd_data_s {
+	seq_data_t data;
+};
+
+static void *
+seq_reader_thd(void *arg) {
+	thd_data_t *thd_data = (thd_data_t *)arg;
+	int iter = 0;
+	data_t local_data;
+	while (iter < 1000 * 1000 - 1) {
+		bool success = seq_try_load_data(&local_data, &thd_data->data);
+		if (success) {
+			assert_data(&local_data);
+			assert_d_le(iter, local_data.arr[0],
+			    "Seq read went back in time.");
+			iter = local_data.arr[0];
+		}
+	}
+	return NULL;
+}
+
+static void *
+seq_writer_thd(void *arg) {
+	thd_data_t *thd_data = (thd_data_t *)arg;
+	data_t local_data;
+	memset(&local_data, 0, sizeof(local_data));
+	for (int i = 0; i < 1000 * 1000; i++) {
+		set_data(&local_data, i);
+		seq_store_data(&thd_data->data, &local_data);
+	}
+	return NULL;
+}
+
+TEST_BEGIN(test_seq_threaded) {
+	thd_data_t thd_data;
+	memset(&thd_data, 0, sizeof(thd_data));
+
+	thd_t reader;
+	thd_t writer;
+
+	thd_create(&reader, seq_reader_thd, &thd_data);
+	thd_create(&writer, seq_writer_thd, &thd_data);
+
+	thd_join(reader, NULL);
+	thd_join(writer, NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_seq_simple) {
+	data_t data;
+	seq_data_t seq;
+	memset(&seq, 0, sizeof(seq));
+	for (int i = 0; i < 1000 * 1000; i++) {
+		set_data(&data, i);
+		seq_store_data(&seq, &data);
+		set_data(&data, 0);
+		bool success = seq_try_load_data(&data, &seq);
+		assert_b_eq(success, true, "Failed non-racing read");
+		assert_data(&data);
+	}
+}
+TEST_END
+
+int main(void) {
+	return test_no_reentrancy(
+	    test_seq_simple,
+	    test_seq_threaded);
+}
diff --git a/jemalloc/test/unit/size_classes.c b/jemalloc/test/unit/size_classes.c
new file mode 100644
index 0000000..6947336
--- /dev/null
+++ b/jemalloc/test/unit/size_classes.c
@@ -0,0 +1,188 @@
+#include "test/jemalloc_test.h"
+
+static size_t
+get_max_size_class(void) {
+	unsigned nlextents;
+	size_t mib[4];
+	size_t sz, miblen, max_size_class;
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nlextents", (void *)&nlextents, &sz, NULL,
+	    0), 0, "Unexpected mallctl() error");
+
+	miblen = sizeof(mib) / sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arenas.lextent.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() error");
+	mib[2] = nlextents - 1;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&max_size_class, &sz,
+	    NULL, 0), 0, "Unexpected mallctlbymib() error");
+
+	return max_size_class;
+}
+
+TEST_BEGIN(test_size_classes) {
+	size_t size_class, max_size_class;
+	szind_t index, max_index;
+
+	max_size_class = get_max_size_class();
+	max_index = sz_size2index(max_size_class);
+
+	for (index = 0, size_class = sz_index2size(index); index < max_index ||
+	    size_class < max_size_class; index++, size_class =
+	    sz_index2size(index)) {
+		assert_true(index < max_index,
+		    "Loop conditionals should be equivalent; index=%u, "
+		    "size_class=%zu (%#zx)", index, size_class, size_class);
+		assert_true(size_class < max_size_class,
+		    "Loop conditionals should be equivalent; index=%u, "
+		    "size_class=%zu (%#zx)", index, size_class, size_class);
+
+		assert_u_eq(index, sz_size2index(size_class),
+		    "sz_size2index() does not reverse sz_index2size(): index=%u"
+		    " --> size_class=%zu --> index=%u --> size_class=%zu",
+		    index, size_class, sz_size2index(size_class),
+		    sz_index2size(sz_size2index(size_class)));
+		assert_zu_eq(size_class,
+		    sz_index2size(sz_size2index(size_class)),
+		    "sz_index2size() does not reverse sz_size2index(): index=%u"
+		    " --> size_class=%zu --> index=%u --> size_class=%zu",
+		    index, size_class, sz_size2index(size_class),
+		    sz_index2size(sz_size2index(size_class)));
+
+		assert_u_eq(index+1, sz_size2index(size_class+1),
+		    "Next size_class does not round up properly");
+
+		assert_zu_eq(size_class, (index > 0) ?
+		    sz_s2u(sz_index2size(index-1)+1) : sz_s2u(1),
+		    "sz_s2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_s2u(size_class-1),
+		    "sz_s2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_s2u(size_class),
+		    "sz_s2u() does not compute same size class");
+		assert_zu_eq(sz_s2u(size_class+1), sz_index2size(index+1),
+		    "sz_s2u() does not round up to next size class");
+	}
+
+	assert_u_eq(index, sz_size2index(sz_index2size(index)),
+	    "sz_size2index() does not reverse sz_index2size()");
+	assert_zu_eq(max_size_class, sz_index2size(
+	    sz_size2index(max_size_class)),
+	    "sz_index2size() does not reverse sz_size2index()");
+
+	assert_zu_eq(size_class, sz_s2u(sz_index2size(index-1)+1),
+	    "sz_s2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_s2u(size_class-1),
+	    "sz_s2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_s2u(size_class),
+	    "sz_s2u() does not compute same size class");
+}
+TEST_END
+
+TEST_BEGIN(test_psize_classes) {
+	size_t size_class, max_psz;
+	pszind_t pind, max_pind;
+
+	max_psz = get_max_size_class() + PAGE;
+	max_pind = sz_psz2ind(max_psz);
+
+	for (pind = 0, size_class = sz_pind2sz(pind);
+	    pind < max_pind || size_class < max_psz;
+	    pind++, size_class = sz_pind2sz(pind)) {
+		assert_true(pind < max_pind,
+		    "Loop conditionals should be equivalent; pind=%u, "
+		    "size_class=%zu (%#zx)", pind, size_class, size_class);
+		assert_true(size_class < max_psz,
+		    "Loop conditionals should be equivalent; pind=%u, "
+		    "size_class=%zu (%#zx)", pind, size_class, size_class);
+
+		assert_u_eq(pind, sz_psz2ind(size_class),
+		    "sz_psz2ind() does not reverse sz_pind2sz(): pind=%u -->"
+		    " size_class=%zu --> pind=%u --> size_class=%zu", pind,
+		    size_class, sz_psz2ind(size_class),
+		    sz_pind2sz(sz_psz2ind(size_class)));
+		assert_zu_eq(size_class, sz_pind2sz(sz_psz2ind(size_class)),
+		    "sz_pind2sz() does not reverse sz_psz2ind(): pind=%u -->"
+		    " size_class=%zu --> pind=%u --> size_class=%zu", pind,
+		    size_class, sz_psz2ind(size_class),
+		    sz_pind2sz(sz_psz2ind(size_class)));
+
+		if (size_class == SC_LARGE_MAXCLASS) {
+			assert_u_eq(SC_NPSIZES, sz_psz2ind(size_class + 1),
+			    "Next size_class does not round up properly");
+		} else {
+			assert_u_eq(pind + 1, sz_psz2ind(size_class + 1),
+			    "Next size_class does not round up properly");
+		}
+
+		assert_zu_eq(size_class, (pind > 0) ?
+		    sz_psz2u(sz_pind2sz(pind-1)+1) : sz_psz2u(1),
+		    "sz_psz2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_psz2u(size_class-1),
+		    "sz_psz2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_psz2u(size_class),
+		    "sz_psz2u() does not compute same size class");
+		assert_zu_eq(sz_psz2u(size_class+1), sz_pind2sz(pind+1),
+		    "sz_psz2u() does not round up to next size class");
+	}
+
+	assert_u_eq(pind, sz_psz2ind(sz_pind2sz(pind)),
+	    "sz_psz2ind() does not reverse sz_pind2sz()");
+	assert_zu_eq(max_psz, sz_pind2sz(sz_psz2ind(max_psz)),
+	    "sz_pind2sz() does not reverse sz_psz2ind()");
+
+	assert_zu_eq(size_class, sz_psz2u(sz_pind2sz(pind-1)+1),
+	    "sz_psz2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_psz2u(size_class-1),
+	    "sz_psz2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_psz2u(size_class),
+	    "sz_psz2u() does not compute same size class");
+}
+TEST_END
+
+TEST_BEGIN(test_overflow) {
+	size_t max_size_class, max_psz;
+
+	max_size_class = get_max_size_class();
+	max_psz = max_size_class + PAGE;
+
+	assert_u_eq(sz_size2index(max_size_class+1), SC_NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+	assert_u_eq(sz_size2index(ZU(PTRDIFF_MAX)+1), SC_NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+	assert_u_eq(sz_size2index(SIZE_T_MAX), SC_NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+
+	assert_zu_eq(sz_s2u(max_size_class+1), 0,
+	    "sz_s2u() should return 0 for unsupported size");
+	assert_zu_eq(sz_s2u(ZU(PTRDIFF_MAX)+1), 0,
+	    "sz_s2u() should return 0 for unsupported size");
+	assert_zu_eq(sz_s2u(SIZE_T_MAX), 0,
+	    "sz_s2u() should return 0 on overflow");
+
+	assert_u_eq(sz_psz2ind(max_size_class+1), SC_NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+	assert_u_eq(sz_psz2ind(ZU(PTRDIFF_MAX)+1), SC_NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+	assert_u_eq(sz_psz2ind(SIZE_T_MAX), SC_NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+
+	assert_zu_eq(sz_psz2u(max_size_class+1), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported"
+	    " size");
+	assert_zu_eq(sz_psz2u(ZU(PTRDIFF_MAX)+1), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported "
+	    "size");
+	assert_zu_eq(sz_psz2u(SIZE_T_MAX), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) on overflow");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_size_classes,
+	    test_psize_classes,
+	    test_overflow);
+}
diff --git a/jemalloc/test/unit/slab.c b/jemalloc/test/unit/slab.c
new file mode 100644
index 0000000..c56af25
--- /dev/null
+++ b/jemalloc/test/unit/slab.c
@@ -0,0 +1,33 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_arena_slab_regind) {
+	szind_t binind;
+
+	for (binind = 0; binind < SC_NBINS; binind++) {
+		size_t regind;
+		extent_t slab;
+		const bin_info_t *bin_info = &bin_infos[binind];
+		extent_init(&slab, NULL, mallocx(bin_info->slab_size,
+		    MALLOCX_LG_ALIGN(LG_PAGE)), bin_info->slab_size, true,
+		    binind, 0, extent_state_active, false, true, true,
+		    EXTENT_NOT_HEAD);
+		assert_ptr_not_null(extent_addr_get(&slab),
+		    "Unexpected malloc() failure");
+		for (regind = 0; regind < bin_info->nregs; regind++) {
+			void *reg = (void *)((uintptr_t)extent_addr_get(&slab) +
+			    (bin_info->reg_size * regind));
+			assert_zu_eq(arena_slab_regind(&slab, binind, reg),
+			    regind,
+			    "Incorrect region index computed for size %zu",
+			    bin_info->reg_size);
+		}
+		free(extent_addr_get(&slab));
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_arena_slab_regind);
+}
diff --git a/jemalloc/test/unit/smoothstep.c b/jemalloc/test/unit/smoothstep.c
new file mode 100644
index 0000000..7c5dbb7
--- /dev/null
+++ b/jemalloc/test/unit/smoothstep.c
@@ -0,0 +1,102 @@
+#include "test/jemalloc_test.h"
+
+static const uint64_t smoothstep_tab[] = {
+#define STEP(step, h, x, y)			\
+	h,
+	SMOOTHSTEP
+#undef STEP
+};
+
+TEST_BEGIN(test_smoothstep_integral) {
+	uint64_t sum, min, max;
+	unsigned i;
+
+	/*
+	 * The integral of smoothstep in the [0..1] range equals 1/2.  Verify
+	 * that the fixed point representation's integral is no more than
+	 * rounding error distant from 1/2.  Regarding rounding, each table
+	 * element is rounded down to the nearest fixed point value, so the
+	 * integral may be off by as much as SMOOTHSTEP_NSTEPS ulps.
+	 */
+	sum = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+		sum += smoothstep_tab[i];
+	}
+
+	max = (KQU(1) << (SMOOTHSTEP_BFP-1)) * (SMOOTHSTEP_NSTEPS+1);
+	min = max - SMOOTHSTEP_NSTEPS;
+
+	assert_u64_ge(sum, min,
+	    "Integral too small, even accounting for truncation");
+	assert_u64_le(sum, max, "Integral exceeds 1/2");
+	if (false) {
+		malloc_printf("%"FMTu64" ulps under 1/2 (limit %d)\n",
+		    max - sum, SMOOTHSTEP_NSTEPS);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_smoothstep_monotonic) {
+	uint64_t prev_h;
+	unsigned i;
+
+	/*
+	 * The smoothstep function is monotonic in [0..1], i.e. its slope is
+	 * non-negative.  In practice we want to parametrize table generation
+	 * such that piecewise slope is greater than zero, but do not require
+	 * that here.
+	 */
+	prev_h = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+		uint64_t h = smoothstep_tab[i];
+		assert_u64_ge(h, prev_h, "Piecewise non-monotonic, i=%u", i);
+		prev_h = h;
+	}
+	assert_u64_eq(smoothstep_tab[SMOOTHSTEP_NSTEPS-1],
+	    (KQU(1) << SMOOTHSTEP_BFP), "Last step must equal 1");
+}
+TEST_END
+
+TEST_BEGIN(test_smoothstep_slope) {
+	uint64_t prev_h, prev_delta;
+	unsigned i;
+
+	/*
+	 * The smoothstep slope strictly increases until x=0.5, and then
+	 * strictly decreases until x=1.0.  Verify the slightly weaker
+	 * requirement of monotonicity, so that inadequate table precision does
+	 * not cause false test failures.
+	 */
+	prev_h = 0;
+	prev_delta = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS / 2 + SMOOTHSTEP_NSTEPS % 2; i++) {
+		uint64_t h = smoothstep_tab[i];
+		uint64_t delta = h - prev_h;
+		assert_u64_ge(delta, prev_delta,
+		    "Slope must monotonically increase in 0.0 <= x <= 0.5, "
+		    "i=%u", i);
+		prev_h = h;
+		prev_delta = delta;
+	}
+
+	prev_h = KQU(1) << SMOOTHSTEP_BFP;
+	prev_delta = 0;
+	for (i = SMOOTHSTEP_NSTEPS-1; i >= SMOOTHSTEP_NSTEPS / 2; i--) {
+		uint64_t h = smoothstep_tab[i];
+		uint64_t delta = prev_h - h;
+		assert_u64_ge(delta, prev_delta,
+		    "Slope must monotonically decrease in 0.5 <= x <= 1.0, "
+		    "i=%u", i);
+		prev_h = h;
+		prev_delta = delta;
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_smoothstep_integral,
+	    test_smoothstep_monotonic,
+	    test_smoothstep_slope);
+}
diff --git a/jemalloc/test/unit/spin.c b/jemalloc/test/unit/spin.c
new file mode 100644
index 0000000..b965f74
--- /dev/null
+++ b/jemalloc/test/unit/spin.c
@@ -0,0 +1,18 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/spin.h"
+
+TEST_BEGIN(test_spin) {
+	spin_t spinner = SPIN_INITIALIZER;
+
+	for (unsigned i = 0; i < 100; i++) {
+		spin_adaptive(&spinner);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_spin);
+}
diff --git a/jemalloc/test/unit/stats.c b/jemalloc/test/unit/stats.c
new file mode 100644
index 0000000..646768e
--- /dev/null
+++ b/jemalloc/test/unit/stats.c
@@ -0,0 +1,374 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_stats_summary) {
+	size_t sz, allocated, active, resident, mapped;
+	int expected = config_stats ? 0 : ENOENT;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.allocated", (void *)&allocated, &sz, NULL,
+	    0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.active", (void *)&active, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.resident", (void *)&resident, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.mapped", (void *)&mapped, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_le(allocated, active,
+		    "allocated should be no larger than active");
+		assert_zu_lt(active, resident,
+		    "active should be less than resident");
+		assert_zu_lt(active, mapped,
+		    "active should be less than mapped");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_stats_large) {
+	void *p;
+	uint64_t epoch;
+	size_t allocated;
+	uint64_t nmalloc, ndalloc, nrequests;
+	size_t sz;
+	int expected = config_stats ? 0 : ENOENT;
+
+	p = mallocx(SC_SMALL_MAXCLASS + 1, MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.allocated",
+	    (void *)&allocated, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.nmalloc", (void *)&nmalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.large.ndalloc", (void *)&ndalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.large.nrequests",
+	    (void *)&nrequests, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_gt(allocated, 0,
+		    "allocated should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_le(nmalloc, nrequests,
+		    "nmalloc should no larger than nrequests");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas_summary) {
+	void *little, *large;
+	uint64_t epoch;
+	size_t sz;
+	int expected = config_stats ? 0 : ENOENT;
+	size_t mapped;
+	uint64_t dirty_npurge, dirty_nmadvise, dirty_purged;
+	uint64_t muzzy_npurge, muzzy_nmadvise, muzzy_purged;
+
+	little = mallocx(SC_SMALL_MAXCLASS, MALLOCX_ARENA(0));
+	assert_ptr_not_null(little, "Unexpected mallocx() failure");
+	large = mallocx((1U << SC_LG_LARGE_MINCLASS),
+	    MALLOCX_ARENA(0));
+	assert_ptr_not_null(large, "Unexpected mallocx() failure");
+
+	dallocx(little, 0);
+	dallocx(large, 0);
+
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.mapped", (void *)&mapped, &sz, NULL,
+	    0), expected, "Unexepected mallctl() result");
+
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.dirty_npurge",
+	    (void *)&dirty_npurge, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.dirty_nmadvise",
+	    (void *)&dirty_nmadvise, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.dirty_purged",
+	    (void *)&dirty_purged, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.muzzy_npurge",
+	    (void *)&muzzy_npurge, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.muzzy_nmadvise",
+	    (void *)&muzzy_nmadvise, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.muzzy_purged",
+	    (void *)&muzzy_purged, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+
+	if (config_stats) {
+		if (!background_thread_enabled()) {
+			assert_u64_gt(dirty_npurge + muzzy_npurge, 0,
+			    "At least one purge should have occurred");
+		}
+		assert_u64_le(dirty_nmadvise, dirty_purged,
+		    "dirty_nmadvise should be no greater than dirty_purged");
+		assert_u64_le(muzzy_nmadvise, muzzy_purged,
+		    "muzzy_nmadvise should be no greater than muzzy_purged");
+	}
+}
+TEST_END
+
+void *
+thd_start(void *arg) {
+	return NULL;
+}
+
+static void
+no_lazy_lock(void) {
+	thd_t thd;
+
+	thd_create(&thd, thd_start, NULL);
+	thd_join(thd, NULL);
+}
+
+TEST_BEGIN(test_stats_arenas_small) {
+	void *p;
+	size_t sz, allocated;
+	uint64_t epoch, nmalloc, ndalloc, nrequests;
+	int expected = config_stats ? 0 : ENOENT;
+
+	no_lazy_lock(); /* Lazy locking would dodge tcache testing. */
+
+	p = mallocx(SC_SMALL_MAXCLASS, MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.small.allocated",
+	    (void *)&allocated, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.small.nmalloc", (void *)&nmalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.small.ndalloc", (void *)&ndalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.small.nrequests",
+	    (void *)&nrequests, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_gt(allocated, 0,
+		    "allocated should be greater than zero");
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be no greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_gt(nrequests, 0,
+		    "nrequests should be greater than zero");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas_large) {
+	void *p;
+	size_t sz, allocated;
+	uint64_t epoch, nmalloc, ndalloc;
+	int expected = config_stats ? 0 : ENOENT;
+
+	p = mallocx((1U << SC_LG_LARGE_MINCLASS), MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.allocated",
+	    (void *)&allocated, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.nmalloc", (void *)&nmalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.large.ndalloc", (void *)&ndalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_gt(allocated, 0,
+		    "allocated should be greater than zero");
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+static void
+gen_mallctl_str(char *cmd, char *name, unsigned arena_ind) {
+	sprintf(cmd, "stats.arenas.%u.bins.0.%s", arena_ind, name);
+}
+
+TEST_BEGIN(test_stats_arenas_bins) {
+	void *p;
+	size_t sz, curslabs, curregs, nonfull_slabs;
+	uint64_t epoch, nmalloc, ndalloc, nrequests, nfills, nflushes;
+	uint64_t nslabs, nreslabs;
+	int expected = config_stats ? 0 : ENOENT;
+
+	/* Make sure allocation below isn't satisfied by tcache. */
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+
+	unsigned arena_ind, old_arena_ind;
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Arena creation failure");
+	sz = sizeof(arena_ind);
+	assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+	    (void *)&arena_ind, sizeof(arena_ind)), 0,
+	    "Unexpected mallctl() failure");
+
+	p = malloc(bin_infos[0].reg_size);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	char cmd[128];
+	sz = sizeof(uint64_t);
+	gen_mallctl_str(cmd, "nmalloc", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nmalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "ndalloc", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&ndalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nrequests", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nrequests, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(size_t);
+	gen_mallctl_str(cmd, "curregs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&curregs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	sz = sizeof(uint64_t);
+	gen_mallctl_str(cmd, "nfills", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nfills, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nflushes", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nflushes, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	gen_mallctl_str(cmd, "nslabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nslabs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nreslabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nreslabs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(size_t);
+	gen_mallctl_str(cmd, "curslabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&curslabs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nonfull_slabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nonfull_slabs, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_gt(nrequests, 0,
+		    "nrequests should be greater than zero");
+		assert_zu_gt(curregs, 0,
+		    "allocated should be greater than zero");
+		if (opt_tcache) {
+			assert_u64_gt(nfills, 0,
+			    "At least one fill should have occurred");
+			assert_u64_gt(nflushes, 0,
+			    "At least one flush should have occurred");
+		}
+		assert_u64_gt(nslabs, 0,
+		    "At least one slab should have been allocated");
+		assert_zu_gt(curslabs, 0,
+		    "At least one slab should be currently allocated");
+		assert_zu_eq(nonfull_slabs, 0,
+		    "slabs_nonfull should be empty");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas_lextents) {
+	void *p;
+	uint64_t epoch, nmalloc, ndalloc;
+	size_t curlextents, sz, hsize;
+	int expected = config_stats ? 0 : ENOENT;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&hsize, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+
+	p = mallocx(hsize, MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.lextents.0.nmalloc",
+	    (void *)&nmalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.lextents.0.ndalloc",
+	    (void *)&ndalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.lextents.0.curlextents",
+	    (void *)&curlextents, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_gt(curlextents, 0,
+		    "At least one extent should be currently allocated");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_stats_summary,
+	    test_stats_large,
+	    test_stats_arenas_summary,
+	    test_stats_arenas_small,
+	    test_stats_arenas_large,
+	    test_stats_arenas_bins,
+	    test_stats_arenas_lextents);
+}
diff --git a/jemalloc/test/unit/stats_print.c b/jemalloc/test/unit/stats_print.c
new file mode 100644
index 0000000..014d002
--- /dev/null
+++ b/jemalloc/test/unit/stats_print.c
@@ -0,0 +1,999 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/util.h"
+
+typedef enum {
+	TOKEN_TYPE_NONE,
+	TOKEN_TYPE_ERROR,
+	TOKEN_TYPE_EOI,
+	TOKEN_TYPE_NULL,
+	TOKEN_TYPE_FALSE,
+	TOKEN_TYPE_TRUE,
+	TOKEN_TYPE_LBRACKET,
+	TOKEN_TYPE_RBRACKET,
+	TOKEN_TYPE_LBRACE,
+	TOKEN_TYPE_RBRACE,
+	TOKEN_TYPE_COLON,
+	TOKEN_TYPE_COMMA,
+	TOKEN_TYPE_STRING,
+	TOKEN_TYPE_NUMBER
+} token_type_t;
+
+typedef struct parser_s parser_t;
+typedef struct {
+	parser_t	*parser;
+	token_type_t	token_type;
+	size_t		pos;
+	size_t		len;
+	size_t		line;
+	size_t		col;
+} token_t;
+
+struct parser_s {
+	bool verbose;
+	char	*buf; /* '\0'-terminated. */
+	size_t	len; /* Number of characters preceding '\0' in buf. */
+	size_t	pos;
+	size_t	line;
+	size_t	col;
+	token_t	token;
+};
+
+static void
+token_init(token_t *token, parser_t *parser, token_type_t token_type,
+    size_t pos, size_t len, size_t line, size_t col) {
+	token->parser = parser;
+	token->token_type = token_type;
+	token->pos = pos;
+	token->len = len;
+	token->line = line;
+	token->col = col;
+}
+
+static void
+token_error(token_t *token) {
+	if (!token->parser->verbose) {
+		return;
+	}
+	switch (token->token_type) {
+	case TOKEN_TYPE_NONE:
+		not_reached();
+	case TOKEN_TYPE_ERROR:
+		malloc_printf("%zu:%zu: Unexpected character in token: ",
+		    token->line, token->col);
+		break;
+	default:
+		malloc_printf("%zu:%zu: Unexpected token: ", token->line,
+		    token->col);
+		break;
+	}
+	UNUSED ssize_t err = malloc_write_fd(STDERR_FILENO,
+	    &token->parser->buf[token->pos], token->len);
+	malloc_printf("\n");
+}
+
+static void
+parser_init(parser_t *parser, bool verbose) {
+	parser->verbose = verbose;
+	parser->buf = NULL;
+	parser->len = 0;
+	parser->pos = 0;
+	parser->line = 1;
+	parser->col = 0;
+}
+
+static void
+parser_fini(parser_t *parser) {
+	if (parser->buf != NULL) {
+		dallocx(parser->buf, MALLOCX_TCACHE_NONE);
+	}
+}
+
+static bool
+parser_append(parser_t *parser, const char *str) {
+	size_t len = strlen(str);
+	char *buf = (parser->buf == NULL) ? mallocx(len + 1,
+	    MALLOCX_TCACHE_NONE) : rallocx(parser->buf, parser->len + len + 1,
+	    MALLOCX_TCACHE_NONE);
+	if (buf == NULL) {
+		return true;
+	}
+	memcpy(&buf[parser->len], str, len + 1);
+	parser->buf = buf;
+	parser->len += len;
+	return false;
+}
+
+static bool
+parser_tokenize(parser_t *parser) {
+	enum {
+		STATE_START,
+		STATE_EOI,
+		STATE_N, STATE_NU, STATE_NUL, STATE_NULL,
+		STATE_F, STATE_FA, STATE_FAL, STATE_FALS, STATE_FALSE,
+		STATE_T, STATE_TR, STATE_TRU, STATE_TRUE,
+		STATE_LBRACKET,
+		STATE_RBRACKET,
+		STATE_LBRACE,
+		STATE_RBRACE,
+		STATE_COLON,
+		STATE_COMMA,
+		STATE_CHARS,
+		STATE_CHAR_ESCAPE,
+		STATE_CHAR_U, STATE_CHAR_UD, STATE_CHAR_UDD, STATE_CHAR_UDDD,
+		STATE_STRING,
+		STATE_MINUS,
+		STATE_LEADING_ZERO,
+		STATE_DIGITS,
+		STATE_DECIMAL,
+		STATE_FRAC_DIGITS,
+		STATE_EXP,
+		STATE_EXP_SIGN,
+		STATE_EXP_DIGITS,
+		STATE_ACCEPT
+	} state = STATE_START;
+	size_t token_pos JEMALLOC_CC_SILENCE_INIT(0);
+	size_t token_line JEMALLOC_CC_SILENCE_INIT(1);
+	size_t token_col JEMALLOC_CC_SILENCE_INIT(0);
+
+	assert_zu_le(parser->pos, parser->len,
+	    "Position is past end of buffer");
+
+	while (state != STATE_ACCEPT) {
+		char c = parser->buf[parser->pos];
+
+		switch (state) {
+		case STATE_START:
+			token_pos = parser->pos;
+			token_line = parser->line;
+			token_col = parser->col;
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+				break;
+			case '\0':
+				state = STATE_EOI;
+				break;
+			case 'n':
+				state = STATE_N;
+				break;
+			case 'f':
+				state = STATE_F;
+				break;
+			case 't':
+				state = STATE_T;
+				break;
+			case '[':
+				state = STATE_LBRACKET;
+				break;
+			case ']':
+				state = STATE_RBRACKET;
+				break;
+			case '{':
+				state = STATE_LBRACE;
+				break;
+			case '}':
+				state = STATE_RBRACE;
+				break;
+			case ':':
+				state = STATE_COLON;
+				break;
+			case ',':
+				state = STATE_COMMA;
+				break;
+			case '"':
+				state = STATE_CHARS;
+				break;
+			case '-':
+				state = STATE_MINUS;
+				break;
+			case '0':
+				state = STATE_LEADING_ZERO;
+				break;
+			case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_EOI:
+			token_init(&parser->token, parser,
+			    TOKEN_TYPE_EOI, token_pos, parser->pos -
+			    token_pos, token_line, token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_N:
+			switch (c) {
+			case 'u':
+				state = STATE_NU;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_NU:
+			switch (c) {
+			case 'l':
+				state = STATE_NUL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_NUL:
+			switch (c) {
+			case 'l':
+				state = STATE_NULL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_NULL:
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+			case '\0':
+			case '[': case ']': case '{': case '}': case ':':
+			case ',':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			token_init(&parser->token, parser, TOKEN_TYPE_NULL,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_F:
+			switch (c) {
+			case 'a':
+				state = STATE_FA;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FA:
+			switch (c) {
+			case 'l':
+				state = STATE_FAL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FAL:
+			switch (c) {
+			case 's':
+				state = STATE_FALS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FALS:
+			switch (c) {
+			case 'e':
+				state = STATE_FALSE;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FALSE:
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+			case '\0':
+			case '[': case ']': case '{': case '}': case ':':
+			case ',':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			token_init(&parser->token, parser,
+			    TOKEN_TYPE_FALSE, token_pos, parser->pos -
+			    token_pos, token_line, token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_T:
+			switch (c) {
+			case 'r':
+				state = STATE_TR;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_TR:
+			switch (c) {
+			case 'u':
+				state = STATE_TRU;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_TRU:
+			switch (c) {
+			case 'e':
+				state = STATE_TRUE;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_TRUE:
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+			case '\0':
+			case '[': case ']': case '{': case '}': case ':':
+			case ',':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			token_init(&parser->token, parser, TOKEN_TYPE_TRUE,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_LBRACKET:
+			token_init(&parser->token, parser, TOKEN_TYPE_LBRACKET,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_RBRACKET:
+			token_init(&parser->token, parser, TOKEN_TYPE_RBRACKET,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_LBRACE:
+			token_init(&parser->token, parser, TOKEN_TYPE_LBRACE,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_RBRACE:
+			token_init(&parser->token, parser, TOKEN_TYPE_RBRACE,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_COLON:
+			token_init(&parser->token, parser, TOKEN_TYPE_COLON,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_COMMA:
+			token_init(&parser->token, parser, TOKEN_TYPE_COMMA,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_CHARS:
+			switch (c) {
+			case '\\':
+				state = STATE_CHAR_ESCAPE;
+				break;
+			case '"':
+				state = STATE_STRING;
+				break;
+			case 0x00: case 0x01: case 0x02: case 0x03: case 0x04:
+			case 0x05: case 0x06: case 0x07: case 0x08: case 0x09:
+			case 0x0a: case 0x0b: case 0x0c: case 0x0d: case 0x0e:
+			case 0x0f: case 0x10: case 0x11: case 0x12: case 0x13:
+			case 0x14: case 0x15: case 0x16: case 0x17: case 0x18:
+			case 0x19: case 0x1a: case 0x1b: case 0x1c: case 0x1d:
+			case 0x1e: case 0x1f:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			default:
+				break;
+			}
+			break;
+		case STATE_CHAR_ESCAPE:
+			switch (c) {
+			case '"': case '\\': case '/': case 'b': case 'n':
+			case 'r': case 't':
+				state = STATE_CHARS;
+				break;
+			case 'u':
+				state = STATE_CHAR_U;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_U:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHAR_UD;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_UD:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHAR_UDD;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_UDD:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHAR_UDDD;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_UDDD:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHARS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_STRING:
+			token_init(&parser->token, parser, TOKEN_TYPE_STRING,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_MINUS:
+			switch (c) {
+			case '0':
+				state = STATE_LEADING_ZERO;
+				break;
+			case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_LEADING_ZERO:
+			switch (c) {
+			case '.':
+				state = STATE_DECIMAL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		case STATE_DIGITS:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				break;
+			case '.':
+				state = STATE_DECIMAL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		case STATE_DECIMAL:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_FRAC_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FRAC_DIGITS:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				break;
+			case 'e': case 'E':
+				state = STATE_EXP;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		case STATE_EXP:
+			switch (c) {
+			case '-': case '+':
+				state = STATE_EXP_SIGN;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_EXP_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_EXP_SIGN:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_EXP_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_EXP_DIGITS:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		default:
+			not_reached();
+		}
+
+		if (state != STATE_ACCEPT) {
+			if (c == '\n') {
+				parser->line++;
+				parser->col = 0;
+			} else {
+				parser->col++;
+			}
+			parser->pos++;
+		}
+	}
+	return false;
+}
+
+static bool	parser_parse_array(parser_t *parser);
+static bool	parser_parse_object(parser_t *parser);
+
+static bool
+parser_parse_value(parser_t *parser) {
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_NULL:
+	case TOKEN_TYPE_FALSE:
+	case TOKEN_TYPE_TRUE:
+	case TOKEN_TYPE_STRING:
+	case TOKEN_TYPE_NUMBER:
+		return false;
+	case TOKEN_TYPE_LBRACE:
+		return parser_parse_object(parser);
+	case TOKEN_TYPE_LBRACKET:
+		return parser_parse_array(parser);
+	default:
+		return true;
+	}
+	not_reached();
+}
+
+static bool
+parser_parse_pair(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_STRING,
+	    "Pair should start with string");
+	if (parser_tokenize(parser)) {
+		return true;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_COLON:
+		if (parser_tokenize(parser)) {
+			return true;
+		}
+		return parser_parse_value(parser);
+	default:
+		return true;
+	}
+}
+
+static bool
+parser_parse_values(parser_t *parser) {
+	if (parser_parse_value(parser)) {
+		return true;
+	}
+
+	while (true) {
+		if (parser_tokenize(parser)) {
+			return true;
+		}
+		switch (parser->token.token_type) {
+		case TOKEN_TYPE_COMMA:
+			if (parser_tokenize(parser)) {
+				return true;
+			}
+			if (parser_parse_value(parser)) {
+				return true;
+			}
+			break;
+		case TOKEN_TYPE_RBRACKET:
+			return false;
+		default:
+			return true;
+		}
+	}
+}
+
+static bool
+parser_parse_array(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_LBRACKET,
+	    "Array should start with [");
+	if (parser_tokenize(parser)) {
+		return true;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_RBRACKET:
+		return false;
+	default:
+		return parser_parse_values(parser);
+	}
+	not_reached();
+}
+
+static bool
+parser_parse_pairs(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_STRING,
+	    "Object should start with string");
+	if (parser_parse_pair(parser)) {
+		return true;
+	}
+
+	while (true) {
+		if (parser_tokenize(parser)) {
+			return true;
+		}
+		switch (parser->token.token_type) {
+		case TOKEN_TYPE_COMMA:
+			if (parser_tokenize(parser)) {
+				return true;
+			}
+			switch (parser->token.token_type) {
+			case TOKEN_TYPE_STRING:
+				if (parser_parse_pair(parser)) {
+					return true;
+				}
+				break;
+			default:
+				return true;
+			}
+			break;
+		case TOKEN_TYPE_RBRACE:
+			return false;
+		default:
+			return true;
+		}
+	}
+}
+
+static bool
+parser_parse_object(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_LBRACE,
+	    "Object should start with {");
+	if (parser_tokenize(parser)) {
+		return true;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_STRING:
+		return parser_parse_pairs(parser);
+	case TOKEN_TYPE_RBRACE:
+		return false;
+	default:
+		return true;
+	}
+	not_reached();
+}
+
+static bool
+parser_parse(parser_t *parser) {
+	if (parser_tokenize(parser)) {
+		goto label_error;
+	}
+	if (parser_parse_value(parser)) {
+		goto label_error;
+	}
+
+	if (parser_tokenize(parser)) {
+		goto label_error;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_EOI:
+		return false;
+	default:
+		goto label_error;
+	}
+	not_reached();
+
+label_error:
+	token_error(&parser->token);
+	return true;
+}
+
+TEST_BEGIN(test_json_parser) {
+	size_t i;
+	const char *invalid_inputs[] = {
+		/* Tokenizer error case tests. */
+		"{ \"string\": X }",
+		"{ \"string\": nXll }",
+		"{ \"string\": nuXl }",
+		"{ \"string\": nulX }",
+		"{ \"string\": nullX }",
+		"{ \"string\": fXlse }",
+		"{ \"string\": faXse }",
+		"{ \"string\": falXe }",
+		"{ \"string\": falsX }",
+		"{ \"string\": falseX }",
+		"{ \"string\": tXue }",
+		"{ \"string\": trXe }",
+		"{ \"string\": truX }",
+		"{ \"string\": trueX }",
+		"{ \"string\": \"\n\" }",
+		"{ \"string\": \"\\z\" }",
+		"{ \"string\": \"\\uX000\" }",
+		"{ \"string\": \"\\u0X00\" }",
+		"{ \"string\": \"\\u00X0\" }",
+		"{ \"string\": \"\\u000X\" }",
+		"{ \"string\": -X }",
+		"{ \"string\": 0.X }",
+		"{ \"string\": 0.0eX }",
+		"{ \"string\": 0.0e+X }",
+
+		/* Parser error test cases. */
+		"{\"string\": }",
+		"{\"string\" }",
+		"{\"string\": [ 0 }",
+		"{\"string\": {\"a\":0, 1 } }",
+		"{\"string\": {\"a\":0: } }",
+		"{",
+		"{}{",
+	};
+	const char *valid_inputs[] = {
+		/* Token tests. */
+		"null",
+		"false",
+		"true",
+		"{}",
+		"{\"a\": 0}",
+		"[]",
+		"[0, 1]",
+		"0",
+		"1",
+		"10",
+		"-10",
+		"10.23",
+		"10.23e4",
+		"10.23e-4",
+		"10.23e+4",
+		"10.23E4",
+		"10.23E-4",
+		"10.23E+4",
+		"-10.23",
+		"-10.23e4",
+		"-10.23e-4",
+		"-10.23e+4",
+		"-10.23E4",
+		"-10.23E-4",
+		"-10.23E+4",
+		"\"value\"",
+		"\" \\\" \\/ \\b \\n \\r \\t \\u0abc \\u1DEF \"",
+
+		/* Parser test with various nesting. */
+		"{\"a\":null, \"b\":[1,[{\"c\":2},3]], \"d\":{\"e\":true}}",
+	};
+
+	for (i = 0; i < sizeof(invalid_inputs)/sizeof(const char *); i++) {
+		const char *input = invalid_inputs[i];
+		parser_t parser;
+		parser_init(&parser, false);
+		assert_false(parser_append(&parser, input),
+		    "Unexpected input appending failure");
+		assert_true(parser_parse(&parser),
+		    "Unexpected parse success for input: %s", input);
+		parser_fini(&parser);
+	}
+
+	for (i = 0; i < sizeof(valid_inputs)/sizeof(const char *); i++) {
+		const char *input = valid_inputs[i];
+		parser_t parser;
+		parser_init(&parser, true);
+		assert_false(parser_append(&parser, input),
+		    "Unexpected input appending failure");
+		assert_false(parser_parse(&parser),
+		    "Unexpected parse error for input: %s", input);
+		parser_fini(&parser);
+	}
+}
+TEST_END
+
+void
+write_cb(void *opaque, const char *str) {
+	parser_t *parser = (parser_t *)opaque;
+	if (parser_append(parser, str)) {
+		test_fail("Unexpected input appending failure");
+	}
+}
+
+TEST_BEGIN(test_stats_print_json) {
+	const char *opts[] = {
+		"J",
+		"Jg",
+		"Jm",
+		"Jd",
+		"Jmd",
+		"Jgd",
+		"Jgm",
+		"Jgmd",
+		"Ja",
+		"Jb",
+		"Jl",
+		"Jx",
+		"Jbl",
+		"Jal",
+		"Jab",
+		"Jabl",
+		"Jax",
+		"Jbx",
+		"Jlx",
+		"Jablx",
+		"Jgmdablx",
+	};
+	unsigned arena_ind, i;
+
+	for (i = 0; i < 3; i++) {
+		unsigned j;
+
+		switch (i) {
+		case 0:
+			break;
+		case 1: {
+			size_t sz = sizeof(arena_ind);
+			assert_d_eq(mallctl("arenas.create", (void *)&arena_ind,
+			    &sz, NULL, 0), 0, "Unexpected mallctl failure");
+			break;
+		} case 2: {
+			size_t mib[3];
+			size_t miblen = sizeof(mib)/sizeof(size_t);
+			assert_d_eq(mallctlnametomib("arena.0.destroy",
+			    mib, &miblen), 0,
+			    "Unexpected mallctlnametomib failure");
+			mib[1] = arena_ind;
+			assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL,
+			    0), 0, "Unexpected mallctlbymib failure");
+			break;
+		} default:
+			not_reached();
+		}
+
+		for (j = 0; j < sizeof(opts)/sizeof(const char *); j++) {
+			parser_t parser;
+
+			parser_init(&parser, true);
+			malloc_stats_print(write_cb, (void *)&parser, opts[j]);
+			assert_false(parser_parse(&parser),
+			    "Unexpected parse error, opts=\"%s\"", opts[j]);
+			parser_fini(&parser);
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_json_parser,
+	    test_stats_print_json);
+}
diff --git a/jemalloc/test/unit/test_hooks.c b/jemalloc/test/unit/test_hooks.c
new file mode 100644
index 0000000..ded8698
--- /dev/null
+++ b/jemalloc/test/unit/test_hooks.c
@@ -0,0 +1,38 @@
+#include "test/jemalloc_test.h"
+
+static bool hook_called = false;
+
+static void
+hook() {
+	hook_called = true;
+}
+
+static int
+func_to_hook(int arg1, int arg2) {
+	return arg1 + arg2;
+}
+
+#define func_to_hook JEMALLOC_HOOK(func_to_hook, test_hooks_libc_hook)
+
+TEST_BEGIN(unhooked_call) {
+	test_hooks_libc_hook = NULL;
+	hook_called = false;
+	assert_d_eq(3, func_to_hook(1, 2), "Hooking changed return value.");
+	assert_false(hook_called, "Nulling out hook didn't take.");
+}
+TEST_END
+
+TEST_BEGIN(hooked_call) {
+	test_hooks_libc_hook = &hook;
+	hook_called = false;
+	assert_d_eq(3, func_to_hook(1, 2), "Hooking changed return value.");
+	assert_true(hook_called, "Hook should have executed.");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    unhooked_call,
+	    hooked_call);
+}
diff --git a/jemalloc/test/unit/ticker.c b/jemalloc/test/unit/ticker.c
new file mode 100644
index 0000000..e5790a3
--- /dev/null
+++ b/jemalloc/test/unit/ticker.c
@@ -0,0 +1,73 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ticker.h"
+
+TEST_BEGIN(test_ticker_tick) {
+#define NREPS 2
+#define NTICKS 3
+	ticker_t ticker;
+	int32_t i, j;
+
+	ticker_init(&ticker, NTICKS);
+	for (i = 0; i < NREPS; i++) {
+		for (j = 0; j < NTICKS; j++) {
+			assert_u_eq(ticker_read(&ticker), NTICKS - j,
+			    "Unexpected ticker value (i=%d, j=%d)", i, j);
+			assert_false(ticker_tick(&ticker),
+			    "Unexpected ticker fire (i=%d, j=%d)", i, j);
+		}
+		assert_u32_eq(ticker_read(&ticker), 0,
+		    "Expected ticker depletion");
+		assert_true(ticker_tick(&ticker),
+		    "Expected ticker fire (i=%d)", i);
+		assert_u32_eq(ticker_read(&ticker), NTICKS,
+		    "Expected ticker reset");
+	}
+#undef NTICKS
+}
+TEST_END
+
+TEST_BEGIN(test_ticker_ticks) {
+#define NTICKS 3
+	ticker_t ticker;
+
+	ticker_init(&ticker, NTICKS);
+
+	assert_u_eq(ticker_read(&ticker), NTICKS, "Unexpected ticker value");
+	assert_false(ticker_ticks(&ticker, NTICKS), "Unexpected ticker fire");
+	assert_u_eq(ticker_read(&ticker), 0, "Unexpected ticker value");
+	assert_true(ticker_ticks(&ticker, NTICKS), "Expected ticker fire");
+	assert_u_eq(ticker_read(&ticker), NTICKS, "Unexpected ticker value");
+
+	assert_true(ticker_ticks(&ticker, NTICKS + 1), "Expected ticker fire");
+	assert_u_eq(ticker_read(&ticker), NTICKS, "Unexpected ticker value");
+#undef NTICKS
+}
+TEST_END
+
+TEST_BEGIN(test_ticker_copy) {
+#define NTICKS 3
+	ticker_t ta, tb;
+
+	ticker_init(&ta, NTICKS);
+	ticker_copy(&tb, &ta);
+	assert_u_eq(ticker_read(&tb), NTICKS, "Unexpected ticker value");
+	assert_true(ticker_ticks(&tb, NTICKS + 1), "Expected ticker fire");
+	assert_u_eq(ticker_read(&tb), NTICKS, "Unexpected ticker value");
+
+	ticker_tick(&ta);
+	ticker_copy(&tb, &ta);
+	assert_u_eq(ticker_read(&tb), NTICKS - 1, "Unexpected ticker value");
+	assert_true(ticker_ticks(&tb, NTICKS), "Expected ticker fire");
+	assert_u_eq(ticker_read(&tb), NTICKS, "Unexpected ticker value");
+#undef NTICKS
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ticker_tick,
+	    test_ticker_ticks,
+	    test_ticker_copy);
+}
diff --git a/jemalloc/test/unit/tsd.c b/jemalloc/test/unit/tsd.c
new file mode 100644
index 0000000..917884d
--- /dev/null
+++ b/jemalloc/test/unit/tsd.c
@@ -0,0 +1,267 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * If we're e.g. in debug mode, we *never* enter the fast path, and so shouldn't
+ * be asserting that we're on one.
+ */
+static bool originally_fast;
+static int data_cleanup_count;
+
+void
+data_cleanup(int *data) {
+	if (data_cleanup_count == 0) {
+		assert_x_eq(*data, MALLOC_TSD_TEST_DATA_INIT,
+		    "Argument passed into cleanup function should match tsd "
+		    "value");
+	}
+	++data_cleanup_count;
+
+	/*
+	 * Allocate during cleanup for two rounds, in order to assure that
+	 * jemalloc's internal tsd reinitialization happens.
+	 */
+	bool reincarnate = false;
+	switch (*data) {
+	case MALLOC_TSD_TEST_DATA_INIT:
+		*data = 1;
+		reincarnate = true;
+		break;
+	case 1:
+		*data = 2;
+		reincarnate = true;
+		break;
+	case 2:
+		return;
+	default:
+		not_reached();
+	}
+
+	if (reincarnate) {
+		void *p = mallocx(1, 0);
+		assert_ptr_not_null(p, "Unexpeced mallocx() failure");
+		dallocx(p, 0);
+	}
+}
+
+static void *
+thd_start(void *arg) {
+	int d = (int)(uintptr_t)arg;
+	void *p;
+
+	tsd_t *tsd = tsd_fetch();
+	assert_x_eq(tsd_test_data_get(tsd), MALLOC_TSD_TEST_DATA_INIT,
+	    "Initial tsd get should return initialization value");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	tsd_test_data_set(tsd, d);
+	assert_x_eq(tsd_test_data_get(tsd), d,
+	    "After tsd set, tsd get should return value that was set");
+
+	d = 0;
+	assert_x_eq(tsd_test_data_get(tsd), (int)(uintptr_t)arg,
+	    "Resetting local data should have no effect on tsd");
+
+	tsd_test_callback_set(tsd, &data_cleanup);
+
+	free(p);
+	return NULL;
+}
+
+TEST_BEGIN(test_tsd_main_thread) {
+	thd_start((void *)(uintptr_t)0xa5f3e329);
+}
+TEST_END
+
+TEST_BEGIN(test_tsd_sub_thread) {
+	thd_t thd;
+
+	data_cleanup_count = 0;
+	thd_create(&thd, thd_start, (void *)MALLOC_TSD_TEST_DATA_INIT);
+	thd_join(thd, NULL);
+	/*
+	 * We reincarnate twice in the data cleanup, so it should execute at
+	 * least 3 times.
+	 */
+	assert_x_ge(data_cleanup_count, 3,
+	    "Cleanup function should have executed multiple times.");
+}
+TEST_END
+
+static void *
+thd_start_reincarnated(void *arg) {
+	tsd_t *tsd = tsd_fetch();
+	assert(tsd);
+
+	void *p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	/* Manually trigger reincarnation. */
+	assert_ptr_not_null(tsd_arena_get(tsd),
+	    "Should have tsd arena set.");
+	tsd_cleanup((void *)tsd);
+	assert_ptr_null(*tsd_arenap_get_unsafe(tsd),
+	    "TSD arena should have been cleared.");
+	assert_u_eq(tsd_state_get(tsd), tsd_state_purgatory,
+	    "TSD state should be purgatory\n");
+
+	free(p);
+	assert_u_eq(tsd_state_get(tsd), tsd_state_reincarnated,
+	    "TSD state should be reincarnated\n");
+	p = mallocx(1, MALLOCX_TCACHE_NONE);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+	assert_ptr_null(*tsd_arenap_get_unsafe(tsd),
+	    "Should not have tsd arena set after reincarnation.");
+
+	free(p);
+	tsd_cleanup((void *)tsd);
+	assert_ptr_null(*tsd_arenap_get_unsafe(tsd),
+	    "TSD arena should have been cleared after 2nd cleanup.");
+
+	return NULL;
+}
+
+TEST_BEGIN(test_tsd_reincarnation) {
+	thd_t thd;
+	thd_create(&thd, thd_start_reincarnated, NULL);
+	thd_join(thd, NULL);
+}
+TEST_END
+
+typedef struct {
+	atomic_u32_t phase;
+	atomic_b_t error;
+} global_slow_data_t;
+
+static void *
+thd_start_global_slow(void *arg) {
+	/* PHASE 0 */
+	global_slow_data_t *data = (global_slow_data_t *)arg;
+	free(mallocx(1, 0));
+
+	tsd_t *tsd = tsd_fetch();
+	/*
+	 * No global slowness has happened yet; there was an error if we were
+	 * originally fast but aren't now.
+	 */
+	atomic_store_b(&data->error, originally_fast && !tsd_fast(tsd),
+	    ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 1, ATOMIC_SEQ_CST);
+
+	/* PHASE 2 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 2) {
+	}
+	free(mallocx(1, 0));
+	atomic_store_b(&data->error, tsd_fast(tsd), ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 3, ATOMIC_SEQ_CST);
+
+	/* PHASE 4 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 4) {
+	}
+	free(mallocx(1, 0));
+	atomic_store_b(&data->error, tsd_fast(tsd), ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 5, ATOMIC_SEQ_CST);
+
+	/* PHASE 6 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 6) {
+	}
+	free(mallocx(1, 0));
+	/* Only one decrement so far. */
+	atomic_store_b(&data->error, tsd_fast(tsd), ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 7, ATOMIC_SEQ_CST);
+
+	/* PHASE 8 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 8) {
+	}
+	free(mallocx(1, 0));
+	/*
+	 * Both decrements happened; we should be fast again (if we ever
+	 * were)
+	 */
+	atomic_store_b(&data->error, originally_fast && !tsd_fast(tsd),
+	    ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 9, ATOMIC_SEQ_CST);
+
+	return NULL;
+}
+
+TEST_BEGIN(test_tsd_global_slow) {
+	global_slow_data_t data = {ATOMIC_INIT(0), ATOMIC_INIT(false)};
+	/*
+	 * Note that the "mallocx" here (vs. malloc) is important, since the
+	 * compiler is allowed to optimize away free(malloc(1)) but not
+	 * free(mallocx(1)).
+	 */
+	free(mallocx(1, 0));
+	tsd_t *tsd = tsd_fetch();
+	originally_fast = tsd_fast(tsd);
+
+	thd_t thd;
+	thd_create(&thd, thd_start_global_slow, (void *)&data.phase);
+	/* PHASE 1 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 1) {
+		/*
+		 * We don't have a portable condvar/semaphore mechanism.
+		 * Spin-wait.
+		 */
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	tsd_global_slow_inc(tsd_tsdn(tsd));
+	free(mallocx(1, 0));
+	assert_false(tsd_fast(tsd), "");
+	atomic_store_u32(&data.phase, 2, ATOMIC_SEQ_CST);
+
+	/* PHASE 3 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 3) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	/* Increase again, so that we can test multiple fast/slow changes. */
+	tsd_global_slow_inc(tsd_tsdn(tsd));
+	atomic_store_u32(&data.phase, 4, ATOMIC_SEQ_CST);
+	free(mallocx(1, 0));
+	assert_false(tsd_fast(tsd), "");
+
+	/* PHASE 5 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 5) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	tsd_global_slow_dec(tsd_tsdn(tsd));
+	atomic_store_u32(&data.phase, 6, ATOMIC_SEQ_CST);
+	/* We only decreased once; things should still be slow. */
+	free(mallocx(1, 0));
+	assert_false(tsd_fast(tsd), "");
+
+	/* PHASE 7 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 7) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	tsd_global_slow_dec(tsd_tsdn(tsd));
+	atomic_store_u32(&data.phase, 8, ATOMIC_SEQ_CST);
+	/* We incremented and then decremented twice; we should be fast now. */
+	free(mallocx(1, 0));
+	assert_true(!originally_fast || tsd_fast(tsd), "");
+
+	/* PHASE 9 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 9) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+
+	thd_join(thd, NULL);
+}
+TEST_END
+
+int
+main(void) {
+	/* Ensure tsd bootstrapped. */
+	if (nallocx(1, 0) == 0) {
+		malloc_printf("Initialization error");
+		return test_status_fail;
+	}
+
+	return test_no_reentrancy(
+	    test_tsd_main_thread,
+	    test_tsd_sub_thread,
+	    test_tsd_reincarnation,
+	    test_tsd_global_slow);
+}
diff --git a/jemalloc/test/unit/witness.c b/jemalloc/test/unit/witness.c
new file mode 100644
index 0000000..5986da4
--- /dev/null
+++ b/jemalloc/test/unit/witness.c
@@ -0,0 +1,280 @@
+#include "test/jemalloc_test.h"
+
+static witness_lock_error_t *witness_lock_error_orig;
+static witness_owner_error_t *witness_owner_error_orig;
+static witness_not_owner_error_t *witness_not_owner_error_orig;
+static witness_depth_error_t *witness_depth_error_orig;
+
+static bool saw_lock_error;
+static bool saw_owner_error;
+static bool saw_not_owner_error;
+static bool saw_depth_error;
+
+static void
+witness_lock_error_intercept(const witness_list_t *witnesses,
+    const witness_t *witness) {
+	saw_lock_error = true;
+}
+
+static void
+witness_owner_error_intercept(const witness_t *witness) {
+	saw_owner_error = true;
+}
+
+static void
+witness_not_owner_error_intercept(const witness_t *witness) {
+	saw_not_owner_error = true;
+}
+
+static void
+witness_depth_error_intercept(const witness_list_t *witnesses,
+    witness_rank_t rank_inclusive, unsigned depth) {
+	saw_depth_error = true;
+}
+
+static int
+witness_comp(const witness_t *a, void *oa, const witness_t *b, void *ob) {
+	assert_u_eq(a->rank, b->rank, "Witnesses should have equal rank");
+
+	assert(oa == (void *)a);
+	assert(ob == (void *)b);
+
+	return strcmp(a->name, b->name);
+}
+
+static int
+witness_comp_reverse(const witness_t *a, void *oa, const witness_t *b,
+    void *ob) {
+	assert_u_eq(a->rank, b->rank, "Witnesses should have equal rank");
+
+	assert(oa == (void *)a);
+	assert(ob == (void *)b);
+
+	return -strcmp(a->name, b->name);
+}
+
+TEST_BEGIN(test_witness) {
+	witness_t a, b;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 0);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+	witness_assert_not_owner(&witness_tsdn, &a);
+	witness_lock(&witness_tsdn, &a);
+	witness_assert_owner(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)2U, 0);
+
+	witness_init(&b, "b", 2, NULL, NULL);
+	witness_assert_not_owner(&witness_tsdn, &b);
+	witness_lock(&witness_tsdn, &b);
+	witness_assert_owner(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 2);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 2);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)2U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)3U, 0);
+
+	witness_unlock(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)2U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)3U, 0);
+	witness_unlock(&witness_tsdn, &b);
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_witness_comp) {
+	witness_t a, b, c, d;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, witness_comp, &a);
+	witness_assert_not_owner(&witness_tsdn, &a);
+	witness_lock(&witness_tsdn, &a);
+	witness_assert_owner(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	witness_init(&b, "b", 1, witness_comp, &b);
+	witness_assert_not_owner(&witness_tsdn, &b);
+	witness_lock(&witness_tsdn, &b);
+	witness_assert_owner(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 2);
+	witness_unlock(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	witness_lock_error_orig = witness_lock_error;
+	witness_lock_error = witness_lock_error_intercept;
+	saw_lock_error = false;
+
+	witness_init(&c, "c", 1, witness_comp_reverse, &c);
+	witness_assert_not_owner(&witness_tsdn, &c);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	witness_lock(&witness_tsdn, &c);
+	assert_true(saw_lock_error, "Expected witness lock error");
+	witness_unlock(&witness_tsdn, &c);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	saw_lock_error = false;
+
+	witness_init(&d, "d", 1, NULL, NULL);
+	witness_assert_not_owner(&witness_tsdn, &d);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	witness_lock(&witness_tsdn, &d);
+	assert_true(saw_lock_error, "Expected witness lock error");
+	witness_unlock(&witness_tsdn, &d);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	witness_unlock(&witness_tsdn, &a);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_lock_error = witness_lock_error_orig;
+}
+TEST_END
+
+TEST_BEGIN(test_witness_reversal) {
+	witness_t a, b;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_lock_error_orig = witness_lock_error;
+	witness_lock_error = witness_lock_error_intercept;
+	saw_lock_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+	witness_init(&b, "b", 2, NULL, NULL);
+
+	witness_lock(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 1);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	witness_lock(&witness_tsdn, &a);
+	assert_true(saw_lock_error, "Expected witness lock error");
+
+	witness_unlock(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+	witness_unlock(&witness_tsdn, &b);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_lock_error = witness_lock_error_orig;
+}
+TEST_END
+
+TEST_BEGIN(test_witness_recursive) {
+	witness_t a;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_not_owner_error_orig = witness_not_owner_error;
+	witness_not_owner_error = witness_not_owner_error_intercept;
+	saw_not_owner_error = false;
+
+	witness_lock_error_orig = witness_lock_error;
+	witness_lock_error = witness_lock_error_intercept;
+	saw_lock_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+
+	witness_lock(&witness_tsdn, &a);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	assert_false(saw_not_owner_error, "Unexpected witness not owner error");
+	witness_lock(&witness_tsdn, &a);
+	assert_true(saw_lock_error, "Expected witness lock error");
+	assert_true(saw_not_owner_error, "Expected witness not owner error");
+
+	witness_unlock(&witness_tsdn, &a);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_owner_error = witness_owner_error_orig;
+	witness_lock_error = witness_lock_error_orig;
+
+}
+TEST_END
+
+TEST_BEGIN(test_witness_unlock_not_owned) {
+	witness_t a;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_owner_error_orig = witness_owner_error;
+	witness_owner_error = witness_owner_error_intercept;
+	saw_owner_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+
+	assert_false(saw_owner_error, "Unexpected owner error");
+	witness_unlock(&witness_tsdn, &a);
+	assert_true(saw_owner_error, "Expected owner error");
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_owner_error = witness_owner_error_orig;
+}
+TEST_END
+
+TEST_BEGIN(test_witness_depth) {
+	witness_t a;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_depth_error_orig = witness_depth_error;
+	witness_depth_error = witness_depth_error_intercept;
+	saw_depth_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+
+	assert_false(saw_depth_error, "Unexpected depth error");
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+
+	witness_lock(&witness_tsdn, &a);
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+	assert_true(saw_depth_error, "Expected depth error");
+
+	witness_unlock(&witness_tsdn, &a);
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+
+	witness_depth_error = witness_depth_error_orig;
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_witness,
+	    test_witness_comp,
+	    test_witness_reversal,
+	    test_witness_recursive,
+	    test_witness_unlock_not_owned,
+	    test_witness_depth);
+}
diff --git a/jemalloc/test/unit/zero.c b/jemalloc/test/unit/zero.c
new file mode 100644
index 0000000..271fd5c
--- /dev/null
+++ b/jemalloc/test/unit/zero.c
@@ -0,0 +1,59 @@
+#include "test/jemalloc_test.h"
+
+static void
+test_zero(size_t sz_min, size_t sz_max) {
+	uint8_t *s;
+	size_t sz_prev, sz, i;
+#define MAGIC	((uint8_t)0x61)
+
+	sz_prev = 0;
+	s = (uint8_t *)mallocx(sz_min, 0);
+	assert_ptr_not_null((void *)s, "Unexpected mallocx() failure");
+
+	for (sz = sallocx(s, 0); sz <= sz_max;
+	    sz_prev = sz, sz = sallocx(s, 0)) {
+		if (sz_prev > 0) {
+			assert_u_eq(s[0], MAGIC,
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    ZU(0), sz_prev);
+			assert_u_eq(s[sz_prev-1], MAGIC,
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    sz_prev-1, sz_prev);
+		}
+
+		for (i = sz_prev; i < sz; i++) {
+			assert_u_eq(s[i], 0x0,
+			    "Newly allocated byte %zu/%zu isn't zero-filled",
+			    i, sz);
+			s[i] = MAGIC;
+		}
+
+		if (xallocx(s, sz+1, 0, 0) == sz) {
+			s = (uint8_t *)rallocx(s, sz+1, 0);
+			assert_ptr_not_null((void *)s,
+			    "Unexpected rallocx() failure");
+		}
+	}
+
+	dallocx(s, 0);
+#undef MAGIC
+}
+
+TEST_BEGIN(test_zero_small) {
+	test_skip_if(!config_fill);
+	test_zero(1, SC_SMALL_MAXCLASS - 1);
+}
+TEST_END
+
+TEST_BEGIN(test_zero_large) {
+	test_skip_if(!config_fill);
+	test_zero(SC_SMALL_MAXCLASS + 1, 1U << (SC_LG_LARGE_MINCLASS + 1));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_zero_small,
+	    test_zero_large);
+}
diff --git a/jemalloc/test/unit/zero.sh b/jemalloc/test/unit/zero.sh
new file mode 100644
index 0000000..b4540b2
--- /dev/null
+++ b/jemalloc/test/unit/zero.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,junk:false,zero:true"
+fi
diff --git a/make_rpm.mk b/make_rpm.mk
new file mode 100644
index 0000000..cfafd8e
--- /dev/null
+++ b/make_rpm.mk
@@ -0,0 +1,106 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+package_prefix ?=
+name = $(package_prefix)memkind
+arch = $(shell uname -p)
+version = 0.0.0
+release = 1
+
+src = $(shell cat MANIFEST)
+
+topdir = $(HOME)/rpmbuild
+rpm = $(topdir)/RPMS/$(arch)/$(name)-devel-$(version)-$(release).$(arch).rpm
+trpm = $(topdir)/TRPMS/$(arch)/$(name)-tests-$(version)-$(release).$(arch).rpm
+srpm = $(topdir)/SRPMS/$(name)-$(version)-$(release).src.rpm
+specfile = $(topdir)/SPECS/$(name)-$(version).spec
+source_tar = $(topdir)/SOURCES/memkind-$(version).tar.gz
+source_tmp_dir := $(topdir)/SOURCES/$(name)-tmp-$(shell date +%s)
+rpmbuild_flags = -E '%define _topdir $(topdir)'
+rpmclean_flags = $(rpmbuild_flags) --clean --rmsource --rmspec
+memkind_test_dir = $(MPSS_TEST_BASEDIR)/memkind-dt
+exclude_source_files = test/memkind-afts.ts \
+test/memkind-afts-ext.ts \
+test/memkind-slts.ts \
+test/memkind-perf.ts \
+test/memkind-perf-ext.ts \
+test/memkind-pytests.ts \
+test/python_framework/cmd_helper.py \
+test/python_framework/huge_page_organizer.py \
+test/python_framework/__init__.py \
+test/hbw_detection_test.py \
+test/dax_kmem_env_var_test.py \
+test/autohbw_test.py \
+test/trace_mechanism_test.py \
+test/draw_plots.py \
+VERSION
+
+all: $(rpm)
+
+$(rpm): $(specfile) $(source_tar)
+	rpmbuild $(rpmbuild_flags) $(specfile) -ba
+	mkdir -p $(topdir)/TRPMS/$(arch)
+	mv $(topdir)/RPMS/$(arch)/$(name)-tests-$(version)-$(release).$(arch).rpm $(trpm)
+
+$(source_tar): $(topdir)/.setup $(src) MANIFEST
+	mkdir -p $(source_tmp_dir)
+	set -e ; \
+	for f in $(exclude_source_files); do \
+		echo $$f >> $(source_tmp_dir)/EXCLUDE ; \
+	done
+	tar cf $(source_tmp_dir)/tmp.tar -T MANIFEST --transform="s|^|$(name)-$(version)/|" -X $(source_tmp_dir)/EXCLUDE
+	cd $(source_tmp_dir) && tar xf tmp.tar
+	set -e ; \
+	for f in $(exclude_source_files); do \
+		mkdir -p `dirname $(source_tmp_dir)/$(name)-$(version)/$$f` ; \
+		cp $$f $(source_tmp_dir)/$(name)-$(version)/$$f ; \
+	done
+	cd $(source_tmp_dir)/$(name)-$(version) && ./autogen.sh && ./configure && make dist; \
+	# tar.gz produced by "make dist" from above produces memkind-$(version).tar.gz
+	# If $(package_prefix) is not empty, then need to repackage that tar.gz to $(name)-$(version)
+	# thus below command. Otherwise, rpmbuild will fail.
+	if [ -n "$(package_prefix)" ]; then \
+	    cd $(source_tmp_dir)/$(name)-$(version) && \
+	    tar xf memkind-$(version).tar.gz && \
+	    rm -rf memkind-$(version).tar.gz && \
+	    mv memkind-$(version) $(name)-$(version) && \
+	    tar cfz memkind-$(version).tar.gz $(name)-$(version); \
+	fi
+	mv $(source_tmp_dir)/$(name)-$(version)/memkind-$(version).tar.gz $@
+	rm -rf $(source_tmp_dir)
+
+$(specfile): $(topdir)/.setup memkind.spec.mk
+	@echo "$$memkind_spec" > $@
+
+$(topdir)/.setup:
+	mkdir -p $(topdir)/{SOURCES,SPECS}
+	touch $@
+
+clean:
+	-rpmbuild $(rpmclean_flags) $(specfile)
+
+.PHONY: all clean
+
+include memkind.spec.mk
+
diff --git a/man/autohbw.7 b/man/autohbw.7
new file mode 100644
index 0000000..a1a4232
--- /dev/null
+++ b/man/autohbw.7
@@ -0,0 +1,195 @@
+.\"
+.\" Copyright (C) 2014-2016 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "AUTOHBW" 7 "2016-07-28" "Intel Corporation" "AUTOHBW" \" -*- nroff -*-
+.SH "NAME"
+libautohbw.so \- An interposer library for redirecting heap allocations
+.SH "SYNOPSIS"
+.BR LD_PRELOAD=libautohbw.so
+command {arguments ...}
+.SH "DESCRIPTION"
+.B AutoHBW
+library
+.BR (libautohbw.so)
+is an interposer library for redirecting heap allocations
+.B (malloc, calloc, realloc, valloc, posix_memalign, memlign)
+to high-bandwidth
+.B (HBW)
+memory. Consequently, AutoHBW library can be used to
+automatically allocate high-bandwidth memory without any modification to
+source code of an application.
+
+.br
+For instance, the following command-line runs existing binary /bin/ls with
+AutoHBW library, automatically redirecting heap allocations (larger than a given
+threshold) to high-bandwidth memory.
+.IP
+.B LD_PRELOAD=libautohbw.so
+/bin/ls
+
+.SH "ENVIRONMENT"
+
+The behavior of AutoHBW library is controlled by the following environment
+variables.
+
+.PP
+.B AUTO_HBW_SIZE=x:[y]
+.br
+Indicates that any allocation larger than
+.I x
+and smaller than
+.I y
+should be
+allocated in HBW memory.
+.I x
+and
+.I y
+can be followed by a K, M, or G to indicate
+the size in Kilo/Memga/Giga bytes (e.g., 4K, 3M, 2G).
+.br
+
+Examples:
+.IP AUTO_HBW_SIZE=4K
+# allocations larger than 4K allocated in HBW
+.IP AUTO_HBW_SIZE=1M:5M
+# allocations between 1M and 5M allocated in HBW
+
+.PP
+.B AUTO_HBW_LOG=level
+.br
+Sets the value of logging (printing)
+.I level.
+If
+.I level
+is:
+.br
+.IP -1
+no messages are printed
+.br
+.IP 0
+no allocations messages are printed but INFO messages are printed
+.br
+.IP 1
+a log message is printed for each allocation (Default)
+.br
+.IP 2
+a log message is printed for each allocation with a backtrace.
+Redirect this output and use
+.B autohbw_get_src_lines.pl
+to find source lines for each allocation. Your application must
+be compiled with
+.B -g
+to see source lines.
+.PP
+Notes:
+.IP
+1. Logging adds extra overhead. Therefore, for performance
+critical runs, logging level should be 0
+.IP
+2. The amount of free memory printed with log messages is only
+approximate -- e.g. pages that are not touched yet are excluded
+.PP
+Examples:
+.IP AUTO_HBW_LOG=1
+
+.PP
+.B AUTO_HBW_MEM_TYPE=memory_type
+.br
+Sets the type of memory type that should be automatically allocated. By
+default, this type is MEMKIND_HBW_PREFERRED, if MCDRAM is found in your
+system; otherwise, the default is MEMKIND_DEFAULT. The names of memory
+types are defined in
+.B memkind(3)
+man page.
+.B memory_type
+has to be one of
+.B MEMKIND_DEFAULT, MEMKIND_HUGETLB, MEMKIND_INTERLEAVE, MEMKIND_HBW,
+.B MEMKIND_HBW_PREFERRED, MEMKIND_HBW_HUGETLB, MEMKIND_HBW_PREFERRED_HUGETLB,
+.B MEMKIND_HBW_GBTLB (DEPRECATED), MEMKIND_HBW_PREFERRED_GBTLB (DEPRECATED), MEMKIND_GBTLB (DEPRECATED),
+.B MEMKIND_HBW_INTERLEAVE
+
+If you are requesting any huge
+TLB pages, please make sure that the requested type is currently enabled
+in your OS.
+
+Examples:
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_HBW_PREFERRED
+# (default, if MCDRAM present)
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_DEFAULT
+# (default, if MCDRAM absent)
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_HBW_HUGETLB
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_HUGETLB
+
+.PP
+.B AUTO_HBW_DEBUG=0|1|2
+.br
+Set the debug message printing level. Default is 0. This is mainly for
+development.
+
+.SH "NOTES"
+It is possible to temporarily disable/enable automatic HBW allocations by
+calling disableAutoHBW() and enableAutoHBW() in source code. To call
+these routines, please include
+.B autohbw_api.h
+header file and link with -lautohbw.
+
+
+.br
+If high-bandwidth memory is not physically present in your system,
+the environment variable
+.B MEMKIND_HBW_NODES
+must be set to indicate the high-bandwidth node as indicated in
+.B memkind(3).
+
+
+.SH "EXAMPLES"
+.br
+The following will run /bin/ls with AutoHBW library. Make sure that paths to
+both libautohbw.so and libmemkind.so are included in
+.B LD_LIBRARY_PATH.
+.IP
+.B LD_PRELOAD=libautohbw.so
+/bin/ls -l
+.PP
+To run with MPI, a shell script must be created, with the correct LD_PRELOAD
+command for each rank. For example, if we put
+.B LD_PRELOAD=libautohbw.so /bin/ls
+in a shell script named autohbw_test.sh, it can be executed with 2 MPI ranks as:
+.br
+.IP
+.B mpirun -n 2 ./autohbw_test.sh
+
+.SH "COPYRIGHT"
+Copyright (C) 2014, 2015, 2016 Intel Corporation. All rights reserved.
+
+.SH "SEE ALSO"
+.BR memkind(3)
+.BR malloc (3),
+.BR numactl (8),
+
+
+
+
+
+
+
diff --git a/man/hbwallocator.3 b/man/hbwallocator.3
new file mode 100644
index 0000000..031c577
--- /dev/null
+++ b/man/hbwallocator.3
@@ -0,0 +1,125 @@
+.\"
+.\" Copyright (C) 2015 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "HBWALLOCATOR" 3 "2015-11-02" "Intel Corporation" "HBWALLOCATOR" \" -*- nroff -*-
+.SH "NAME"
+hbw::allocator<T> \- The C++ allocator compatible with the C++ standard library allocator concepts
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.nf
+.B #include <hbw_allocator.h>
+.sp
+.B Link with -lmemkind
+.sp
+.BI "hbw::allocator();"
+.br
+.BI "template <class U>hbw::allocator<T>::allocator(const hbw::allocator<U>&);
+.br
+.BI "hbw::allocator<T>::~allocator()'"
+.br
+.BI "hbw::allocator<T>::pointer hbw::allocator<T>::address(hbw::allocator<T>::reference " "x" );
+.br
+.BI "hbw::allocator<T>::const_pointer hbw::allocator<T>::address(hbw::allocator<T>::const_reference " "x" );
+.br
+.BI "hbw::allocator<T>::pointer hbw::allocator<T>::allocate(hbw::allocator<T>::size_type " "n" ", const void * = 0);
+.br
+.BI "void hbw::allocator<T>::deallocate(hbw::allocator<T>::pointer " "p" ", hbw::allocator<T>::size_type " "n" );
+.br
+.BI "hbw::allocator<T>::size_type  hbw::allocator<T>::max_size();
+.br
+.BI "void hbw::allocator<T>::construct(hbw::allocator<T>::pointer " "p" ", const hbw::allocator<T>::value_type& " "val" );
+.br
+.BI "void hbw::allocator<T>::destroy(hbw::allocator<T>::pointer " "p" );
+.fi
+.SH "DESCRIPTION"
+The
+.BR hbw::allocator<T>
+is intended to be used with STL containers to allocate high bandwidth memory. Memory management is based on hbwmalloc (memkind library), enabling users to gain performance in multithreaded applications. Refer
+.BR hbwmalloc (3)
+and
+.BR memkind (3)
+man page for more details.
+.PP
+All public member types and functions corresponds to standard library allocator concepts and definitions. The current implementation supports C++03 standard.
+.PP
+Template arguments:
+.br
+.I T
+is an object type aliased by value_type.
+.br
+.I U
+is an object type.
+.PP
+.B Note:
+.br
+.BI "hbw::allocator<T>::pointer hbw::allocator<T>::allocate(hbw::allocator<T>::size_type " "n" ", const void * = 0)"
+allocates high bandwidth memory using
+.BR hbw_malloc ().
+Throw
+.I std::bad_alloc
+when:
+.br
+.RS
+.I n
+= 0,
+.RE
+.br
+.RS
+.I n
+>
+.BR max_size ()
+.RE
+.br
+.RS
+or there is not enough memory to satisfy the request.
+.RE
+.PP
+.BI "hbw::allocator<T>::deallocate(hbw::allocator<T>::pointer " "p" ", hbw::allocator<T>::size_type " "n")
+deallocates memory associated with pointer returned by
+.BR allocate ()
+using
+.BR hbw_free ().
+
+.SH ERRORS
+The same as described in
+.B ERRORS
+section of
+.BR hbwmalloc (3)
+man page.
+.SH "NOTES"
+The
+.BR hbw::allocator<T>
+behavior depends on hbwmalloc heap management policy. To get and set the policy please use
+.BR hbw_get_policy ()
+and
+.BR hbw_set_policy ()
+respectively.
+.SH "COPYRIGHT"
+Copyright (C) 2015 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR hbwmalloc (3),
+.BR numa (3),
+.BR numactl (8),
+.BR memkind (3)
diff --git a/man/hbwmalloc.3 b/man/hbwmalloc.3
new file mode 100644
index 0000000..4c7329a
--- /dev/null
+++ b/man/hbwmalloc.3
@@ -0,0 +1,515 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "HBWMALLOC" 3 "2015-03-31" "Intel Corporation" "HBWMALLOC" \" -*- nroff -*-
+.SH "NAME"
+hbwmalloc \- The high bandwidth memory interface
+.br
+.BR Note:
+.I hbwmalloc.h
+functionality is considered as stable API (STANDARD API).
+.SH "SYNOPSIS"
+.nf
+.B #include <hbwmalloc.h>
+.sp
+.B Link with -lmemkind
+.sp
+.B int hbw_check_available(void);
+.br
+.BI "void* hbw_malloc(size_t " "size" );
+.br
+.BI "void* hbw_calloc(size_t " "nmemb" ", size_t " "size" );
+.br
+.BI "void* hbw_realloc (void " "*ptr" ", size_t " "size" );
+.br
+.BI "void hbw_free(void " "*ptr" );
+.br
+.BI "size_t hbw_malloc_usable_size(void " "*ptr" );
+.br
+.BI "int hbw_posix_memalign(void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.br
+.BI "int hbw_posix_memalign_psize(void " "**memptr" ", size_t " "alignment" ", size_t " "size" ", hbw_pagesize_t " "pagesize" );
+.br
+.B hbw_policy_t hbw_get_policy(void);
+.br
+.BI "int hbw_set_policy(hbw_policy_t " "mode" );
+.br
+.BI "int hbw_verify_memory_region(void " "*addr" ", size_t " "size" ", int " "flags" );
+.fi
+.SH "DESCRIPTION"
+.BR hbw_check_available ()
+returns zero if high bandwidth memory is available or an error code
+described in the
+.B ERRORS
+section if not.
+.PP
+.BR hbw_malloc ()
+allocates
+.I size
+bytes of uninitialized high bandwidth memory. The allocated space is
+suitably aligned (after possible pointer coercion) for storage of any
+type of object. If
+.I size
+is zero then
+.BR hbw_malloc ()
+returns
+.IR "NULL" .
+.PP
+.BR hbw_calloc ()
+allocates space for
+.I nmemb
+objects in high bandwidth memory, each
+.I size
+bytes in length. The result is identical to calling
+.BR hbw_malloc ()
+with an argument of
+.IR "nmemb"
+*
+.IR "size" ,
+with the exception that the allocated memory is explicitly
+initialized to zero bytes.  If
+.I nmemb
+or
+.I size
+is 0, then
+.BR hbw_calloc ()
+returns
+.IR "NULL" .
+.PP
+.BR hbw_realloc ()
+changes the size of the previously allocated high bandwidth memory
+referenced by
+.I ptr
+to
+.I size
+bytes. The contents of the memory remain unchanged up to the lesser of
+the new and old sizes. If the new size is larger, the contents of the
+newly allocated portion of the memory are undefined. Upon success, the
+memory referenced by
+.I ptr
+is freed and a pointer to the newly allocated high bandwidth memory is
+returned.
+
+.B Note:
+.BR hbw_realloc ()
+may move the memory allocation, resulting in a different return value
+than
+.IR "ptr" .
+
+If
+.I ptr
+is
+.IR "NULL" ,
+the
+.BR hbw_realloc ()
+function behaves identically to
+.BR hbw_malloc ()
+for the specified size.
+If
+.I size
+is equal to zero, and
+.I ptr
+is not
+.IR "NULL" ,
+then the call is equivalent to
+.I hbw_free(ptr)
+and
+.I NULL
+is returned. The address
+.IR "ptr" ,
+if not
+.IR "NULL" ,
+was returned by a previous call to
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc ()
+or
+.BR hbw_posix_memalign ().
+Otherwise, or if
+.I hbw_free(ptr)
+was called before, undefined behavior occurs.
+
+.B Note:
+.BR hbw_realloc ()
+cannot be used with a pointer returned by
+.BR hbw_posix_memalign_psize ().
+
+.PP
+.BR hbw_free ()
+causes the allocated memory referenced by
+.I ptr
+to be made available for future allocations. If
+.I ptr
+is
+.IR "NULL" ,
+no action occurs.
+The address
+.IR "ptr" ,
+if not
+.IR "NULL" ,
+must have been returned by a previous call to
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc (),
+.BR hbw_posix_memalign ()
+or
+.BR hbw_posix_memalign_psize ().
+Otherwise, if
+.I hbw_free(ptr)
+was called before, undefined behavior occurs.
+.PP
+.BR hbw_malloc_usable_size ()
+returns the number of usable bytes in the block pointed to by
+.IR "ptr" ,
+a pointer to a block of memory allocated by
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc (),
+.BR hbw_posix_memalign (),
+or
+.BR hbw_posix_memalign_psize ().
+.PP
+.BR hbw_posix_memalign ()
+allocates
+.I size
+bytes of high bandwidth memory such that the allocation's base address
+is an even multiple of
+.IR "alignment" ,
+and returns the allocation in the value pointed to by
+.IR "memptr" .
+The requested
+.I alignment
+must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+If
+.I size
+is 0, then
+.BR hbw_posix_memalign ()
+returns 0, with a
+.I NULL
+returned in
+.IR "memptr" .
+.PP
+.BR hbw_posix_memalign_psize ()
+allocates
+.I size
+bytes of high bandwidth memory such that the allocation's base address
+is an even multiple of
+.IR "alignment" ,
+and returns the allocation in the value pointed to by
+.IR "memptr" .
+The requested
+.I alignment
+must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+The memory will be allocated using pages determined by the
+.IR "pagesize"
+variable which may be one of the following enumerated values:
+.TP
+.B HBW_PAGESIZE_4KB
+The four kilobyte page size option. Note that with transparent huge
+pages enabled these allocations may be promoted by the operating
+system to two megabyte pages.
+.TP
+.B HBW_PAGESIZE_2MB
+The two megabyte page size option.
+.BR Note:
+This page size requires huge pages configuration described in
+.BR "SYSTEM CONFIGURATION"
+section.
+.TP
+.B HBW_PAGESIZE_1GB (DEPRECATED)
+This option allows the user to specify arbitrary sizes backed by
+1GB chunks of huge pages. Huge pages are allocated even if the
+size is not a modulo of 1GB.
+.BR Note:
+This page size requires huge pages configuration described in
+.BR "SYSTEM CONFIGURATION"
+section.
+.TP
+.B HBW_PAGESIZE_1GB_STRICT (DEPRECATED)
+The total size of the allocation must be a multiple of 1GB with
+this option, otherwise the allocation will fail.
+.BR Note:
+This page size requires huge pages configuration described in
+.BR "SYSTEM CONFIGURATION"
+section.
+.PP
+.BR Note:
+.BR HBW_PAGESIZE_2MB ,
+.BR HBW_PAGESIZE_1GB
+and
+.BR HBW_PAGESIZE_1GB_STRICT
+options are not supported with
+.BR HBW_POLICY_INTERLEAVE
+policy.
+.PP
+.BR hbw_get_policy ()
+returns the current fallback policy when insufficient high bandwidth
+memory is available.
+.PP
+.BR hbw_set_policy ()
+sets the current fallback policy. The policy can be modified only once in the lifetime of an application and before calling
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc (),
+.BR hbw_posix_memalign (),
+or
+.BR hbw_posix_memalign_psize ()
+function.
+.br
+.BR Note:
+If the policy is not set, than
+.B HBW_POLICY_PREFERRED
+will be used by default.
+.TP
+.B HBW_POLICY_BIND
+If insufficient high bandwidth memory from the nearest NUMA node is
+available to satisfy a request, the allocated pointer is set to
+.I NULL
+and
+.I errno
+is set to
+.BR ENOMEM .
+If insufficient high bandwidth memory pages are
+available at fault time the Out Of Memory (OOM) Killer is triggered.
+Note that pages are faulted exclusively from the high bandwidth NUMA
+node nearest at time of allocation, not at time of fault.
+.TP
+.B HBW_POLICY_BIND_ALL
+If insufficient high bandwidth memory is available to satisfy a request,
+the allocated pointer is set to
+.I NULL
+and
+.I errno
+is set to
+.BR ENOMEM .
+If insufficient high bandwidth memory pages are
+available at fault time the Out Of Memory (OOM) Killer is triggered.
+Note that pages are faulted from the high bandwidth NUMA nodes.
+Nearest NUMA node is selected at time of page fault.
+.TP
+.B HBW_POLICY_PREFERRED
+If insufficient memory is available from the high bandwidth NUMA node
+closest at allocation time, fall back to standard memory (default)
+with the smallest NUMA distance.
+.TP
+.B HBW_POLICY_INTERLEAVE
+Interleave faulted pages from across all high bandwidth NUMA nodes
+using standard size pages (the Transparent Huge Page feature is
+disabled).
+.PP
+.BR hbw_verify_memory_region ()
+verifies if memory region fully falls into high bandwidth memory. Returns
+0 if memory address range from
+.IR "addr"
+to
+.IR "addr"
++
+.IR "size"
+is allocated in high bandwidth memory,
+-1 if any fragment of memory was not backed by high bandwidth memory (e.g. when memory is not initialized)
+or one of error codes described in
+.B ERRORS
+section.
+
+Using this function in production code may result in serious performance penalty.
+
+The
+.IR Flags
+argument may include optional flags that modify function behavior:
+.TP
+.B HBW_TOUCH_PAGES
+Before checking pages, function will touch first byte of all pages in address range starting from
+.IR "addr"
+to
+.IR "addr"
++
+.IR "size"
+by read and write (so the content will be overwritten by the same data as it was read).
+Using this option may trigger Out Of Memory Killer.
+.SH "RETURN VALUE"
+.BR hbw_get_policy ()
+returns
+.BR HBW_POLICY_BIND,
+.BR HBW_POLICY_BIND_ALL,
+.BR HBW_POLICY_PREFERRED
+or
+.BR HBW_POLICY_INTERLEAVE
+which represents the current high bandwidth policy.
+.BR hbw_free ()
+do not have return value.
+.BR hbw_malloc ()
+.BR hbw_calloc ()
+and
+.BR hbw_realloc ()
+return the pointer to the allocated memory, or
+.I NULL
+if the request fails.
+.BR hbw_posix_memalign (),
+.BR hbw_posix_memalign_psize ()
+and
+.BR hbw_set_policy ()
+return zero on success and return an error code
+as described in the
+.B ERRORS
+section below on failure.
+.SH ERRORS
+.TP
+Error codes described here are the POSIX standard error codes as defined in
+.I <errno.h>
+.TP
+.BR hbw_check_available ()
+returns
+.BR ENODEV
+if high-bandwidth memory is unavailable.
+.TP
+.BR "hbw_posix_memalign" "() and " "hbw_posix_memalign_psize" "()"
+If the
+.I alignment
+parameter is not a power of two, or was not a multiple of
+.IR "sizeof(void*)" ,
+then
+.B EINVAL
+is returned.
+If the policy and
+.I pagesize
+combination is unsupported then
+.B EINVAL
+is returned.
+If there was insufficient memory to satisfy the request then
+.B ENOMEM
+is returned.
+.TP
+.BR hbw_set_policy ()
+returns
+.B EPERM
+if
+.BR hbw_set_policy ()
+was called more than once, or
+.B EINVAL
+if
+.I mode
+argument was neither
+.BR HBW_POLICY_PREFERRED ,
+.BR HBW_POLICY_BIND ,
+.BR HBW_POLICY_BIND_ALL
+nor
+.BR HBW_POLICY_INTERLEAVE .
+.TP
+.BR hbw_verify_memory_region ()
+returns
+.B EINVAL
+if
+.IR "addr"
+is
+.IR "NULL" ,
+.IR "size"
+equals 0 or
+.IR "flags"
+contained unsupported bit set. If memory pointed by
+.IR "addr"
+could not be verified then
+.B EFAULT
+is returned.
+.SH "NOTES"
+The
+.I <hbwmalloc.h>
+file defines the external functions and enumerations for the hbwmalloc
+library. These interfaces define a heap manager that targets high
+bandwidth memory numa nodes.
+.SH "FILES"
+.TP
+.I /usr/bin/memkind-hbw-nodes
+Prints a comma-separated list of high bandwidth nodes.
+.SH "ENVIRONMENT"
+.TP
+.B MEMKIND_HBW_NODES
+This environment variable is a comma-separated list of NUMA nodes that
+are treated as high bandwidth. Uses the
+.I libnuma
+routine
+.BR numa_parse_nodestring ()
+for parsing, so the syntax described in the
+.BR numa (3)
+man page for this routine applies for example: 1-3,5 is a valid setting.
+.TP
+.B MEMKIND_ARENA_NUM_PER_KIND
+This environment variable allows leveraging internal mechanism of
+the library for setting number of arenas per kind. Value should be
+a positive integer (not greater than
+.B INT_MAX
+defined in
+.IR <limits.h> ).
+The user should set the value based on the characteristics
+of application that is using the library. Higher value can
+provide better performance in extremely multithreaded applications at
+the cost of memory overhead. See section
+.BR "IMPLEMENTATION NOTES"
+of
+.BR jemalloc (3)
+for more details about arenas.
+.TP
+.B MEMKIND_HEAP_MANAGER
+Controls heap management behavior in memkind library by switching to one of the available heap managers.
+.br
+Values:
+.br
+    JEMALLOC - sets the jemalloc heap manager
+.br
+    TBB - sets the Intel Threading Building Blocks heap manager. This option requires installed
+    Intel Threading Building Blocks library.
+.PP
+.BR Note:
+If the
+.B MEMKIND_HEAP_MANAGER
+is not set then the jemalloc heap manager will be used by default.
+.SH "SYSTEM CONFIGURATION"
+Interfaces for obtaining 2MB (HUGETLB) memory need allocated
+huge pages in the kernel's huge page pool.
+.TP
+.B HUGETLB (huge pages)
+Current number of "persistent" huge pages can be read from
+.I /proc/sys/vm/nr_hugepages
+file.
+Proposed way of setting hugepages is:
+.BR "sudo sysctl vm.nr_hugepages=<number_of_hugepages>" .
+More information can be found here:
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.SH "KNOWN ISSUES"
+.TP
+.B HUGETLB (huge pages)
+There might be some overhead in huge pages consumption caused by heap management.
+If your allocation fails because of OOM, please try to allocate extra huge pages (e.g. 8 huge pages).
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR malloc (3),
+.BR numa (3),
+.BR numactl (8),
+.BR mbind (2),
+.BR mmap (2),
+.BR move_pages (2),
+.BR jemalloc (3),
+.BR memkind (3)
diff --git a/man/memkind-auto-dax-kmem-nodes.1 b/man/memkind-auto-dax-kmem-nodes.1
new file mode 100644
index 0000000..4ec49ae
--- /dev/null
+++ b/man/memkind-auto-dax-kmem-nodes.1
@@ -0,0 +1,56 @@
+.\"
+.\" Copyright (C) 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND-AUTO-DAX-KMEM-NODES" 1 "2019-08-29" "Intel Corporation" "MEMKIND-AUTO-DAX-KMEM-NODES" \" -*- nroff -*-
+.SH "NAME"
+memkind-auto-dax-kmem-nodes - prints comma-separated list of automatically detected persistent memory NUMA nodes.
+.br
+Note: This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes), or removed.
+.SH "SYNOPSIS"
+memkind-auto-dax-kmem-nodes [-h|--help]
+.SH "DESCRIPTION"
+.BR memkind-auto-dax-kmem-nodes
+prints a comma-separated list of automatically detected persistent memory NUMA nodes that can be used with the numactl --membind option.
+.SH "OPTIONS"
+.BR " -h, --help"
+              Displays help text and exit.
+.SH "EXIT STATUS"
+.BR 0
+on success
+.br
+.BR 1
+if automatic detection is disabled (memkind wasn't build with suitable libdaxctl-devel version)
+.br
+.BR 2
+if argument is invalid
+.br
+.BR 3
+on other failure
+.SH "COPYRIGHT"
+Copyright (C) 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR daxctl (3),
+.BR memkind (3),
+.BR numactl (8)
+.SH "AUTHOR"
+Michal Biesek <michal.biesek@intel.com>
diff --git a/man/memkind-hbw-nodes.1 b/man/memkind-hbw-nodes.1
new file mode 100644
index 0000000..1d4fd34
--- /dev/null
+++ b/man/memkind-hbw-nodes.1
@@ -0,0 +1,47 @@
+.\"
+.\" Copyright (C) 2018 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND-HBW_NODES" 1 "2018-03-12" "Intel Corporation" "MEMKIND-HBW_NODES" \" -*- nroff -*-
+.SH "NAME"
+memkind-hbw-nodes - print comma-separated list of high bandwidth nodes.
+.br
+Note: This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes), or removed.
+.SH "SYNOPSIS"
+memkind-hbw-nodes [-h|--help]
+.SH "DESCRIPTION"
+.BR memkind-hbw-nodes
+prints a comma-separated list of high bandwidth NUMA nodes that can be used with the numactl --membind option.
+.SH "OPTIONS"
+.BR " -h, --help"
+              Displays help text and exit.
+.SH "EXIT STATUS"
+.BR memkind-hbw-nodes
+exits with status 0 on success, 1 on failure, 2 on invalid argument.
+.SH "COPYRIGHT"
+Copyright (C) 2016 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR hbwmalloc (3),
+.BR memkind (3),
+.BR numactl (8)
+.SH "AUTHOR"
+Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>
diff --git a/man/memkind.3 b/man/memkind.3
new file mode 100644
index 0000000..2658de3
--- /dev/null
+++ b/man/memkind.3
@@ -0,0 +1,1092 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND" 3 "2015-03-31" "Intel Corporation" "MEMKIND" \" -*- nroff -*-
+.SH "NAME"
+memkind \- Heap manager that enables allocations to memory with different properties.
+.br
+This header expose STANDARD and EXPERIMENTAL API. API Standards are described below in this man page.
+.SH "SYNOPSIS"
+.nf
+.B #include <memkind.h>
+.sp
+.B Link with -lmemkind
+.br
+.SS "EXPERIMENTAL API:"
+.sp
+.B "HEAP MANAGEMENT:"
+.br
+.BI "int memkind_posix_memalign(memkind_t " "kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.sp
+.B "KIND MANAGEMENT:"
+.br
+.BI "int memkind_create_kind(memkind_memtype_t " "memtype_flags" ", memkind_policy_t " "policy" ", memkind_bits_t " "flags" ", memkind_t " "*kind" );
+.sp
+.SS "STANDARD API:"
+.sp
+.B "ERROR HANDLING:"
+.br
+.BI "void memkind_error_message(int " "err" ", char " "*msg" ", size_t " "size" );
+.sp
+.B "LIBRARY VERSION:"
+.br
+.BI "int memkind_get_version();"
+.sp
+.B "HEAP MANAGEMENT:"
+.br
+.BI "void *memkind_malloc(memkind_t " "kind" ", size_t " "size" );
+.br
+.BI "void *memkind_calloc(memkind_t " "kind" ", size_t " "num" ", size_t " "size" );
+.br
+.BI "void *memkind_realloc(memkind_t " "kind" ", void " "*ptr" ", size_t " "size" );
+.br
+.BI "void memkind_free(memkind_t " "kind" ", void " "*ptr" );
+.br
+.BI "size_t memkind_malloc_usable_size(memkind_t " "kind" ", void " "*ptr" );
+.br
+.BI "void *memkind_defrag_reallocate(memkind_t " "kind" ", void " "*ptr" );
+.br
+.BI "memkind_t memkind_detect_kind(void " "*ptr" );
+.sp
+.B "KIND CONFIGURATION MANAGEMENT:"
+.br
+.BI "struct memkind_config *memkind_config_new();"
+.br
+.BI "void memkind_config_delete(struct memkind_config " "*cfg" );
+.br
+.BI "void memkind_config_set_path(struct memkind_config " "*cfg" ", const char " "*pmem_dir" );
+.br
+.BI "void memkind_config_set_size(struct memkind_config " "*cfg" ", size_t " "pmem_size" );
+.br
+.BI "void memkind_config_set_memory_usage_policy(struct memkind_config " "*cfg" ", memkind_mem_usage_policy " "policy" );
+.sp
+.B "KIND MANAGEMENT:"
+.br
+.BI "int memkind_create_pmem(const char " "*dir" ", size_t " "max_size" ", memkind_t " "*kind" );
+.br
+.BI "int memkind_create_pmem_with_config(struct memkind_config " "*cfg" ", memkind_t " "*kind" );
+.br
+.BI "int memkind_destroy_kind(memkind_t " "kind" );
+.br
+.BI "int memkind_check_available(memkind_t " "kind" );
+.sp
+.B "STATISTICS:"
+.br
+.BI "int memkind_update_cached_stats(void);"
+.br
+.BI "int memkind_get_stat(memkind_t " "kind" ", memkind_stat " "stat" ", size_t " "*value" );
+.sp
+.B "DECORATORS:"
+.br
+.BI "void memkind_malloc_pre(memkind_t " "*kind" ", size_t " "*size" );
+.br
+.BI "void memkind_malloc_post(memkind_t " "kind" ", size_t " "size" ", void " "**result" );
+.br
+.BI "void memkind_calloc_pre(memkind_t " "*kind" ", size_t " "*nmemb" ", size_t " "*size" );
+.br
+.BI "void memkind_calloc_post(memkind_t " "kind" ", size_t " "nmemb" ", size_t " "size" ", void " "**result" );
+.br
+.BI "void memkind_posix_memalign_pre(memkind_t " "*kind" ", void " "**memptr" ", size_t " "*alignment" ", size_t " "*size" );
+.br
+.BI "void memkind_posix_memalign_post(memkind_t " "kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" ", int " "*err" );
+.br
+.BI "void memkind_realloc_pre(memkind_t " "*kind" ", void " "**ptr" ", size_t " "*size" );
+.br
+.BI "void memkind_realloc_post(memkind_t " "*kind" ", void " "*ptr" ", size_t " "size" ", void " "**result" );
+.br
+.BI "void memkind_free_pre(memkind_t " "*kind" ", void " "**ptr" );
+.br
+.BI "void memkind_free_post(memkind_t " "kind" ", void " "*ptr" );
+.sp
+.sp
+.br
+.SH "DESCRIPTION"
+.PP
+.BR memkind_error_message ()
+converts an error number
+.I err
+returned by a member of the memkind
+interface to an error message
+.I msg
+where the maximum size of the message is passed by the
+.I size
+parameter.
+
+.B "HEAP MANAGEMENT:"
+.br
+The functions described in this section define a heap manager with an
+interface modeled on the ISO C standard API's, except that the user
+must specify the
+.I kind
+of memory with the first argument to each function. See the
+.B KINDS
+section below for a full description of the implemented kinds.
+For file-backed kind of memory see
+.BR memkind_create_pmem ()
+or
+.BR memkind_create_pmem_with_config ().
+.PP
+.BR memkind_malloc ()
+allocates
+.I size
+bytes of uninitialized memory of the specified
+.IR "kind" .
+The allocated space is suitably aligned (after possible pointer
+coercion) for storage of any type of object. If
+.I size
+is 0, then
+.BR memkind_malloc ()
+returns
+.IR "NULL" .
+.PP
+.BR memkind_calloc ()
+allocates space for
+.I num
+objects each
+.I size
+bytes in length in memory of the specified
+.IR "kind" .
+The result is identical to calling
+.BR memkind_malloc ()
+with an argument of
+.IR num
+*
+.IR "size" ,
+with the exception that the allocated memory is explicitly
+initialized to zero bytes.
+If
+.I num
+or
+.I size
+is 0, then
+.BR memkind_calloc ()
+returns
+.IR "NULL" .
+.PP
+.BR memkind_realloc ()
+changes the size of the previously allocated memory referenced by
+.I ptr
+to
+.I size
+bytes of the specified
+.IR "kind" .
+The contents of the memory remain unchanged up to the lesser of
+the new and old sizes. If the new size is larger, the contents of the
+newly allocated portion of the memory are undefined. Upon success, the
+memory referenced by
+.I ptr
+is freed and a pointer to the newly allocated memory is returned.
+
+.BR Note:
+.BR memkind_realloc ()
+may move the memory allocation, resulting in a different return value
+than
+.IR "ptr" .
+
+If
+.I ptr
+is
+.IR "NULL" ,
+the
+.BR memkind_realloc ()
+function behaves identically to
+.BR memkind_malloc ()
+for the specified size.
+If
+.I size
+is equal to zero, and
+.I ptr
+is not
+.IR "NULL" ,
+then the call is equivalent to
+.IR "memkind_free(kind, ptr)"
+and
+.I NULL
+is returned. The address
+.IR "ptr" ,
+if not
+.IR "NULL" ,
+must have been returned by a previous call to
+.BR memkind_malloc (),
+.BR memkind_calloc (),
+.BR memkind_realloc (),
+.BR memkind_defrag_reallocate ()
+or
+.BR memkind_posix_memalign ()
+with the same
+.I kind
+as specified to the call to
+.BR memkind_realloc ().
+Otherwise, if
+.I memkind_free(kind, ptr)
+was called before, undefined behavior occurs.
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_realloc ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_realloc (),
+but this will require an internal look up for correct kind.
+.BR Note:
+The look up for
+.I kind
+could result in serious performance penalty,
+which can be avoided by specifying a correct
+.IR kind .
+If
+.I kind
+is
+.I NULL
+and
+.I ptr
+is
+.IR "NULL" ,
+then
+.BR memkind_realloc ()
+returns
+.I NULL
+and sets
+.I errno
+to
+.BR EINVAL .
+.PP
+.BR memkind_posix_memalign ()
+allocates
+.I size
+bytes of memory of a specified
+.I kind
+such that the allocation's base address
+is an even multiple of
+.IR "alignment" ,
+and returns the allocation in the value pointed to by
+.IR "memptr" .
+The requested
+.I alignment
+must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+If
+.I size
+is 0, then
+.BR memkind_posix_memalign ()
+returns 0, with a
+.I NULL
+returned in
+.IR "memptr" .
+.PP
+.BR memkind_malloc_usable_size ()
+function provides the same semantics as
+.BR malloc_usable_size (3),
+but operates on specified
+.IR "kind" .
+.br
+.BR Note:
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_malloc_usable_size ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_malloc_usable_size (),
+but this could require a internal look up for correct kind.
+.BR memkind_malloc_usable_size ()
+is supported by TBB heap manager described in
+.B ENVIRONMENT
+section since Intel TBB 2019 Update 4.
+.PP
+.BR memkind_defrag_reallocate ()
+reallocates the object conditionally inside specific
+.IR "kind" .
+Function determines if it's worthwhile to move allocation to reduce degree of external fragmentation of the heap.
+In case of failure function returns
+.IR "NULL" ,
+otherwise function returns a pointer to reallocated memory and memory referenced by
+.I ptr
+was released and should not be accessed.
+If
+.I ptr
+is
+.IR "NULL" ,
+then
+.BR memkind_defrag_reallocate ()
+returns
+.IR "NULL" .
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_defrag_reallocate ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_defrag_reallocate (),
+but this will require an internal look up for correct kind.
+.BR Note:
+The look up for
+.I kind
+could result in serious performance penalty,
+which can be avoided by specifying a correct
+.IR kind .
+.PP
+.BR memkind_detect_kind ()
+returns the kind associated with allocated memory referenced by
+.IR ptr .
+This pointer
+must have been returned by a previous call to
+.BR memkind_malloc (),
+.BR memkind_calloc (),
+.BR memkind_realloc (),
+.BR memkind_defrag_reallocate ()
+or
+.BR memkind_posix_memalign ().
+If
+.I ptr
+is
+.IR "NULL" ,
+then
+.BR memkind_detect_kind ()
+returns
+.IR "NULL" .
+.BR Note:
+This function has non-trivial performance overhead.
+.PP
+.BR memkind_free ()
+causes the allocated memory referenced by
+.I ptr
+to be made available for future allocations. This pointer
+must have been returned by a previous call to
+.BR memkind_malloc (),
+.BR memkind_calloc (),
+.BR memkind_realloc (),
+.BR memkind_defrag_reallocate ()
+or
+.BR memkind_posix_memalign ().
+Otherwise, if
+.I memkind_free(kind, ptr)
+was already called before, undefined behavior occurs.
+If
+.I ptr
+is
+.IR "NULL" ,
+no operation is performed.
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_free ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_free (),
+but this will require an internal look up for correct kind.
+.BR Note:
+The look up for
+.I kind
+could result in serious performance penalty,
+which can be avoided by specifying a correct
+.IR kind .
+.sp
+.B "KIND CONFIGURATION MANAGEMENT:"
+.br
+The functions described in this section define a way to create, delete and update kind specific configuration.
+Except of
+.BR memkind_config_new (),
+user must specify the memkind configuration with the first argument to each function.
+API described here is most useful with file-backed kind of memory, e.g.
+.BR memkind_create_pmem_with_config ()
+method.
+.PP
+.BR memkind_config_new ()
+creates the memkind configuration.
+.PP
+.BR memkind_config_delete ()
+deletes previously created memkind configuration, which must have been returned by a previous call to
+.BR memkind_config_new ().
+.PP
+.BR memkind_config_set_path ()
+updates the memkind
+.IR pmem_dir
+configuration parameter, which specifies directory path, where file-backed kind of memory will be created.
+.BR Note:
+This function does not validate that
+.I pmem_dir
+specifies a valid path.
+.PP
+.BR memkind_config_set_size ()
+updates the memkind
+.IR pmem_size
+configuration parameter, which allows to limit the file-backed kind memory partition.
+.BR Note:
+This function does not validate that
+.I pmem_size
+is in valid range.
+.PP
+.BR memkind_config_set_memory_usage_policy ()
+updates the memkind
+.IR policy
+configuration parameter, which allows to tune up memory utilization.
+The user should set the value based on the characteristics of application that is using
+the library (e.g. prioritize memory usage, CPU utilization), for more details about
+.IR policy
+see the
+.BR "MEMORY USAGE POLICY"
+section below.
+.BR Note:
+This function does not validate that
+.I policy
+is in valid range.
+.sp
+.B "KIND MANAGEMENT:"
+.br
+There are built-in kinds that are always available and these are enumerated in the
+.B KINDS
+section. The user can also create their own kinds of memory. This
+section describes the API's that enable the tracking of the different
+kinds of memory and determining their properties.
+.PP
+.BR memkind_create_pmem ()
+is a convenient function used to create a file-backed kind of memory.
+It allocates a temporary file in the given directory
+.IR dir .
+The file is created in a fashion similar to
+.BR tmpfile (3),
+so that the file name does not appear when the directory is listed and
+the space is automatically freed when the program terminates.
+The file is truncated to a size of
+.I max_size
+bytes and the resulting space is memory-mapped.
+.br
+Note that the actual file system space is not allocated immediately, but only
+on a call to
+.BR memkind_pmem_mmap ()
+(see
+.BR memkind_pmem (3)).
+This allows to create a pmem memkind of a pretty large size without the
+need to reserve in advance the corresponding file system space for the entire
+heap. If the value of
+.I max_size
+equals 0, pmem memkind is only limited by the capacity of the file system mounted under
+.I dir
+argument.
+The minimum
+.I max_size
+value which allows to limit the size of kind by the library is defined as
+.BR MEMKIND_PMEM_MIN_SIZE .
+Calling
+.BR memkind_create_pmem ()
+with a size smaller than that and different than 0 will return an error.
+The maximum allowed size is not limited by
+.BR memkind ,
+but by the file system specified by the
+.I dir
+argument.
+The
+.I max_size
+passed in is the raw size of the memory pool and
+.B jemalloc
+will use some of that space for its own metadata.
+Returns zero if the pmem memkind is created successfully or an error code from the
+.B ERRORS
+section if not.
+.PP
+.BR memkind_create_pmem_with_config ()
+is a second function used to create a file-backed kind of memory.
+Function behaves simillar to
+.BR memkind_create_pmem ()
+but instead of passing
+.I dir
+and
+.I max_size
+arguments,
+it uses
+.I config
+param to specify characteristics of created file-backed kind of memory (see
+.B KIND CONFIGURATION MANAGEMENT
+section).
+.PP
+.BR memkind_create_kind ()
+creates kind that allocates memory with specific memory type, memory binding policy and flags (see
+.B MEMORY FLAGS
+section).
+The
+.IR memtype_flags
+(see
+.B MEMORY TYPES
+section) determine memory types to allocate,
+.IR policy
+argument is policy for specifying page binding to memory types selected by
+.IR memtype_flags .
+Returns zero if the specified kind is created successfully or an error code from the
+.B ERRORS
+section if not.
+.PP
+.BR memkind_destroy_kind ()
+destroys previously created kind object, which must have been returned by a previous call to
+.BR memkind_create_pmem (),
+.BR memkind_create_pmem_with_config ()
+or
+.BR memkind_create_kind ().
+Otherwise, or if
+.I memkind_destroy_kind(kind)
+was already called before, undefined behavior occurs.
+Note that, when the kind was returned by
+.BR memkind_create_kind ()
+all allocated memory must be freed before kind is destroyed,
+otherwise this will cause memory leak. When the kind was returned by
+.BR memkind_create_pmem ()
+or
+.BR memkind_create_pmem_with_config ()
+all allocated memory will be freed after kind will be destroyed.
+.PP
+.BR memkind_check_available ()
+returns zero if the specified
+.I kind
+is available or an error code from the
+.B ERRORS
+section if it is not.
+.PP
+.BR MEMKIND_PMEM_MIN_SIZE
+The minimum size which allows to limit the file-backed memory partition.
+.sp
+.B "STATISTICS:"
+.br
+The functions described in this section define a way to get specific memory allocation statistics.
+.PP
+.BR memkind_update_cached_stats ()
+is used to force an update of cached dynamic allocator statistics.
+Statistics are not updated real-time by memkind library and this method allows to force its update.
+.PP
+.BR memkind_get_stat ()
+retrieves statistic of the specified type
+and returns it in
+.IR "value".
+For more details about
+.I stat
+see the
+.B "MEMORY STATISTICS TYPE"
+section below.
+Measured statistic applies
+to specific
+.IR "kind",
+when
+.I NULL
+is given as
+.I kind
+then statistic applies to memory used by the whole memkind library.
+.BR Note:
+You need to call
+.BR memkind_update_cached_stats ()
+before calling
+.BR memkind_get_stat ()
+because statistics are cached by memkind library.
+.sp
+.B "DECORATORS:"
+.br
+The memkind library enables the user to define decorator functions that
+can be called before and after each memkind heap management API. The
+decorators that are called at the beginning of the function end are named
+after that function with
+.I _pre
+appended to the name and those that are called at the end of the
+function are named after that function with
+.I _post
+appended to the name. These are weak symbols and if they are not
+present at link time they are not called. The memkind library does
+not define these symbols which are reserved for user definition.
+These decorators can be used to track calls to the heap management
+interface or to modify parameters. The decorators that are called at
+the beginning of the allocator pass all inputs by reference and the
+decorators that are called at the end of the allocator pass the output
+by reference. This enables the modification of the input and output
+of each heap management function by the decorators.
+.sp
+.B "LIBRARY VERSION:"
+.br
+The memkind library version scheme consist major, minor and patch numbers separated by dot. Combining those numbers, we got the following representation:
+.br
+major.minor.patch, where:
+.br
+	-major number is incremented whenever API is changed (loss of backward compatibility),
+.br
+	-minor number is incremented whenever additional extensions are introduced or behavior has been changed,
+.br
+	-patch number is incremented whenever small bug fixes are added.
+.sp
+memkind library provide numeric representation of the version by exposing the following API:
+.PP
+.BR memkind_get_version ()
+returns version number represented by a single integer number, obtained from the formula:
+.br
+major * 1000000 + minor * 1000 + patch
+.sp
+.BR Note:
+major < 1 means unstable API.
+.sp
+API standards:
+.br
+-STANDARD API, API is considered as stable
+.br
+-NON-STANDARD API, API is considered as stable, however this is not a standard way to use memkind
+.br
+-EXPERIMENTAL API, API is considered as unstable and the subject to change
+.br
+.sp
+.SH "RETURN VALUE"
+.BR memkind_calloc (),
+.BR memkind_malloc (),
+.BR memkind_realloc ()
+and
+.BR memkind_defrag_reallocate ()
+returns the pointer to the allocated memory or
+.I NULL
+if the request fails.
+.BR memkind_malloc_usable_size ()
+returns the number of usable bytes in the block of allocated memory pointed to by
+.IR "ptr" ,
+a pointer to a block of memory allocated by
+.BR memkind_malloc ()
+or a related function. If
+.I ptr
+is
+.IR "NULL" ,
+0 is returned.
+.BR memkind_free ()
+and
+.BR memkind_error_message ()
+do not have return values.
+All other memkind API's return 0 upon
+success and an error code defined in the
+.B ERRORS
+section upon failure.
+The memkind library avoids setting
+.I errno
+directly, but calls to underlying libraries and system calls may set
+.IR errno
+(e.g.
+.BR memkind_create_pmem ()).
+.SH "KINDS"
+The available kinds of memory:
+.TP
+.B MEMKIND_DEFAULT
+Default allocation using standard memory and default page size.
+.TP
+.B MEMKIND_HUGETLB
+Allocate from standard memory using huge pages.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_GBTLB (DEPRECATED)
+Allocate from standard memory using 1GB chunks backed by huge pages.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_INTERLEAVE
+Allocate pages interleaved across all NUMA nodes with transparent huge
+pages disabled.
+.TP
+.B MEMKIND_HBW
+Allocate from the closest high bandwidth memory NUMA node at the time
+of allocation. If there is not enough high bandwidth memory to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_HBW_ALL
+Same as
+.B MEMKIND_HBW
+except decision regarding closest NUMA node is postponed until the time of first write.
+.TP
+.B MEMKIND_HBW_HUGETLB
+Same as
+.B MEMKIND_HBW
+except the allocation is backed by huge pages.
+.BR Note:
+This kind requires
+huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_ALL_HUGETLB
+Combination of
+.B MEMKIND_HBW_ALL
+and
+.B MEMKIND_HBW_HUGETLB
+properties.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_PREFERRED
+Same as
+.B MEMKIND_HBW
+except that if there is not enough high bandwidth memory to satisfy
+the request, the allocation will fall back on standard memory.
+.TP
+.B MEMKIND_HBW_PREFERRED_HUGETLB
+Same as
+.B MEMKIND_HBW_PREFERRED
+except the allocation is backed by huge pages.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_GBTLB (DEPRECATED)
+Same as
+.B MEMKIND_HBW
+except the allocation is backed by 1GB chunks of huge pages. Note that
+.I size
+can take on any value, but full gigabyte pages will allocated for each
+request, so remainder of the last page will be wasted.
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_PREFERRED_GBTLB (DEPRECATED)
+Same as
+.B MEMKIND_HBW_GBTLB
+except that if there is not enough high bandwidth memory to satisfy
+the request, the allocation will fall back on standard memory.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_INTERLEAVE
+Same as
+.B MEMKIND_HBW
+except that the pages that support the allocation are interleaved
+across all high bandwidth nodes and transparent huge pages are
+disabled.
+.TP
+.B MEMKIND_DAX_KMEM
+Allocate from the closest persistent memory NUMA node at the time
+of allocation. If there is not enough memory in the closest persistent memory NUMA node to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_DAX_KMEM_ALL
+Allocate from the closest persistent memory NUMA node available at the time
+of allocation. If there is not enough memory on any of persistent memory NUMA nodes to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_DAX_KMEM_PREFERRED
+Same as
+.B MEMKIND_DAX_KMEM
+except that if there is not enough memory in the closest persistent memory NUMA node to satisfy
+the request, the allocation will fall back on other memory NUMA nodes.
+.BR Note:
+For this kind, the allocation will not succeed if two or more
+persistent memory NUMA nodes are in the same shortest distance to the same CPU on which process is eligible to run.
+Check on that eligibility is done upon starting the application.
+.TP
+.B MEMKIND_REGULAR
+Allocate from regular memory using the default page size. Regular means general purpose memory
+from the NUMA nodes containing CPUs.
+.SH "MEMORY TYPES"
+The available types of memory:
+.TP
+.B MEMKIND_MEMTYPE_DEFAULT
+Standard memory, the same as process uses.
+.TP
+.B MEMKIND_MEMTYPE_HIGH_BANDWIDTH
+High bandwidth memory (HBM). There must be at least two memory types with different bandwidth to determine which is the HBM.
+.SH "MEMORY BINDING POLICY"
+The available types of memory binding policy:
+.TP
+.B MEMKIND_POLICY_BIND_LOCAL
+Allocate local memory. If there is not enough memory to satisfy the request
+.I errno
+is set to
+.BR ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_POLICY_BIND_ALL
+Memory locality is ignored. If there is not enough memory to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_POLICY_PREFERRED_LOCAL
+Allocate preferred memory that is local.
+If there is not enough preferred memory to satisfy the request or
+preferred memory is not available, the allocation will fall back on any other memory.
+.TP
+.B MEMKIND_POLICY_INTERLEAVE_LOCAL
+Interleave allocation across local memory.
+For n memory types the allocation will be interleaved across all of them.
+.TP
+.B MEMKIND_POLICY_INTERLEAVE_ALL
+Interleave allocation. Locality is ignored.
+For n memory types the allocation will be interleaved across all of them.
+.TP
+.B MEMKIND_POLICY_MAX_VALUE
+Max policy value.
+.SH "MEMORY FLAGS"
+The available types of memory flags:
+.TP
+.B MEMKIND_MASK_PAGE_SIZE_2MB
+Allocation backed by 2MB page size.
+.SH "MEMORY USAGE POLICY"
+The available types of memory usage policy:
+.TP
+.B MEMKIND_MEM_USAGE_POLICY_DEFAULT
+Default memory usage policy.
+.TP
+.B MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE
+Conservative memory usage policy - prioritize memory usage at cost of performance.
+.BR Note:
+Memory usage policies have no effect for TBB heap manager described in
+.B ENVIRONMENT
+section.
+.SH "MEMORY STATISTICS TYPE"
+The available types of memory statistics:
+.TP
+.B MEMKIND_STAT_TYPE_RESIDENT
+Maximum number of bytes in physically resident data pages mapped.
+.TP
+.B MEMKIND_STAT_TYPE_ACTIVE
+Total number of bytes in active pages.
+.TP
+.B MEMKIND_STAT_TYPE_ALLOCATED
+Total number of allocated bytes.
+.SH "ERRORS"
+.TP
+.BR memkind_posix_memalign ()
+returns the one of the POSIX standard error codes
+.B EINVAL
+or
+.B ENOMEM
+as defined in
+.I <errno.h>
+if an error occurs (these have positive values).
+If the
+.I alignment
+parameter is not a power of two or is not a multiple of
+.IR "sizeof(void*)" ,
+then
+.B EINVAL
+is returned. If there is insufficient memory to satisfy the request then
+.B ENOMEM
+is returned.
+.PP
+All functions other than
+.BR memkind_posix_memalign ()
+which have an integer return type return one of the negative error
+codes as defined in
+.I <memkind.h>
+and described below.
+.TP
+.B MEMKIND_ERROR_UNAVAILABLE
+Requested memory kind is not available
+.TP
+.B MEMKIND_ERROR_MBIND
+Call to
+.BR mbind (2)
+failed
+.TP
+.B MEMKIND_ERROR_MMAP
+Call to
+.BR mmap (2)
+failed
+.TP
+.B MEMKIND_ERROR_MALLOC
+Call to jemalloc's
+.BR malloc ()
+failed
+.TP
+.B MEMKIND_ERROR_ENVIRON
+Error parsing environment variable
+.I MEMKIND_*
+.TP
+.B MEMKIND_ERROR_INVALID
+Invalid input arguments to memkind routine
+.TP
+.B MEMKIND_ERROR_TOOMANY
+Error trying to initialize more than maximum
+.B MEMKIND_MAX_KIND
+number of kinds
+.TP
+.B MEMKIND_ERROR_BADOPS
+Error memkind operation structure is missing or invalid
+.TP
+.B MEMKIND_ERROR_HUGETLB
+Unable to allocate huge pages
+.TP
+.B MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE
+Error requested memory type is not available
+.TP
+.B MEMKIND_ERROR_OPERATION_FAILED
+Error memkind operation failed
+.TP
+.B MEMKIND_ERROR_ARENAS_CREATE
+Call to jemalloc's
+.BR arenas.create ()
+failed
+.TP
+.B MEMKIND_ERROR_RUNTIME
+Unspecified run-time error
+.SH "FILES"
+.TP
+.I /usr/bin/memkind-hbw-nodes
+Prints a comma-separated list of high bandwidth nodes.
+.TP
+.I /usr/bin/memkind-auto-dax-kmem-nodes
+Prints a comma-separated list of persistent memory NUMA nodes, which are automatically detected.
+.SH "ENVIRONMENT"
+.TP
+.B MEMKIND_HBW_NODES
+This environment variable is a comma-separated list of NUMA nodes that
+are treated as high bandwidth. Uses the
+.I libnuma
+routine
+.BR numa_parse_nodestring ()
+for parsing, so the syntax described in the
+.BR numa (3)
+man page for this routine applies: e.g. 1-3,5 is a valid setting.
+.TP
+.B MEMKIND_DAX_KMEM_NODES
+This environment variable is a comma-separated list of NUMA nodes that
+are treated as PMEM memory. Uses the
+.I libnuma
+routine
+.BR numa_parse_nodestring ()
+for parsing, so the syntax described in the
+.BR numa (3)
+man page for this routine applies: e.g. 1-3,5 is a valid setting.
+.TP
+.B MEMKIND_ARENA_NUM_PER_KIND
+This environment variable allows leveraging internal mechanism of
+the library for setting number of arenas per kind. Value should be
+a positive integer (not greater than
+.B INT_MAX
+defined in
+.IR <limits.h> ).
+The user should set the value based on the characteristics
+of application that is using the library. Higher value can
+provide better performance in extremely multithreaded applications at
+the cost of memory overhead. See section
+.BR "IMPLEMENTATION NOTES"
+of
+.BR jemalloc (3)
+for more details about arenas.
+.TP
+.B MEMKIND_HOG_MEMORY
+Controls behavior of memkind with regards to returning memory to underlying OS. Setting
+.B MEMKIND_HOG_MEMORY
+to 1 causes memkind to not release memory to OS in anticipation of memory reuse soon. This will
+improve latency of 'free' operations but increase memory usage.
+.TP
+.B MEMKIND_DEBUG
+Controls logging mechanism in memkind. Setting
+.B MEMKIND_DEBUG
+to 1 enables printing messages like errors and general information about environment to
+.IR stderr .
+.TP
+.B MEMKIND_BACKGROUND_THREAD_LIMIT
+Enable background worker threads.
+Value should be from range 0 to maximum number of cpus.
+Setting
+.B MEMKIND_BACKGROUND_THREAD_LIMIT
+to specific value will limit maximum number of background worker threads to this value.
+0 means maximum number of background worker threads will be limited to maximum number of cpus.
+.TP
+.B MEMKIND_HEAP_MANAGER
+Controls heap management behavior in memkind library by switching to one of the available heap managers.
+.br
+Values:
+.br
+    JEMALLOC - sets the jemalloc heap manager
+.br
+    TBB - sets the Intel Threading Building Blocks heap manager. This option requires installed
+    Intel Threading Building Blocks library.
+.PP
+If the
+.B MEMKIND_HEAP_MANAGER
+is not set then the jemalloc heap manager will be used by default.
+.SH "SYSTEM CONFIGURATION"
+Interfaces for obtaining 2MB (HUGETLB) memory need allocated
+huge pages in the kernel's huge page pool.
+.TP
+.B HUGETLB (huge pages)
+Current number of "persistent" huge pages can be read from
+.I /proc/sys/vm/nr_hugepages
+file.
+Proposed way of setting hugepages is:
+.BR "sudo sysctl vm.nr_hugepages=<number_of_hugepages>" .
+More information can be found here:
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.SH "STATIC LINKING"
+When linking statically against memkind,
+.I libmemkind.a
+should be used together with its dependencies
+.I libnuma
+and pthread. Pthread can be linked by adding
+.I /usr/lib64/libpthread.a
+as a dependency (exact path may vary). Typically
+.I libnuma
+will need to be compiled from sources to use it as a static dependency.
+.I libnuma
+can be reached on GitHub:
+.UR https://github.com/numactl/numactl
+.UE
+.SH "KNOWN ISSUES"
+.TP
+.B HUGETLB (huge pages)
+There might be some overhead in huge pages consumption caused by heap management.
+If your allocation fails because of OOM, please try to allocate extra huge pages (e.g. 8 huge pages).
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR malloc (3),
+.BR malloc_usable_size (3),
+.BR numa (3),
+.BR numactl (8),
+.BR mbind (2),
+.BR mmap (2),
+.BR move_pages (2),
+.BR jemalloc (3),
+.BR memkind_dax_kmem (3),
+.BR memkind_default (3),
+.BR memkind_arena (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3)
diff --git a/man/memkind_arena.3 b/man/memkind_arena.3
new file mode 100644
index 0000000..45a4d1a
--- /dev/null
+++ b/man/memkind_arena.3
@@ -0,0 +1,222 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_ARENA" 3 "2015-04-21" "Intel Corporation" "MEMKIND_ARENA" \" -*- nroff -*-
+.SH "NAME"
+memkind_arena \- jemalloc arena allocation memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_arena_create(struct memkind " "*kind" ", struct memkind_ops " "*ops" ", const char " "*name" );
+.BI "int memkind_arena_create_map(struct memkind " "*kind" ", extent_hooks_t " "*hooks" );
+.BI "int memkind_arena_destroy(struct memkind " "*kind" );
+.BI "void *memkind_arena_malloc(struct memkind " "*kind" ", size_t " "size" );
+.BI "void *memkind_arena_calloc(struct memkind " "*kind" ", size_t " "num" ", size_t " "size" );
+.BI "int memkind_arena_posix_memalign(struct memkind " "*kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.BI "void *memkind_arena_realloc(struct memkind " "*kind" ", void " "*ptr" ", size_t " "size" );
+.BI "void *memkind_arena_realloc_with_kind_detect(void " "*ptr" ", size_t " "size" );
+.BI "int memkind_thread_get_arena(struct memkind " "*kind" ", unsigned int " "*arena" ", size_t " "size" );
+.BI "int memkind_bijective_get_arena(struct memkind " "*kind" ", unsigned int " "*arena" ", size_t " "size" );
+.BI "struct memkind *get_kind_by_arena(unsigned " "arena_ind" );
+.BI "struct memkind *memkind_arena_detect_kind(void " "*ptr" );
+.BI "int memkind_arena_finalize(struct memkind " "*kind" );
+.BI "void memkind_arena_init(struct memkind " "*kind" );
+.BI "void memkind_arena_free(struct memkind " "*kind" ", void " "*ptr" );
+.BI "void memkind_arena_free_with_kind_detect(void " "*ptr" );
+.BI "size_t memkind_arena_malloc_usable_size(void " "*ptr" );
+.BI "int memkind_arena_update_memory_usage_policy(struct memkind " "*kind" ", memkind_mem_usage_policy " "policy" );
+.BI "int memkind_arena_enable_background_threads(size_t threads_limit);"
+.br
+.SH DESCRIPTION
+This header file is a collection of functions can be used to populate
+the memkind operations structure for memory kinds that use jemalloc.
+.PP
+.BR memkind_arena_create ()
+is an implementation of the memkind "create" operation for memory
+kinds that use jemalloc.  This calls
+.BR memkind_default_create ()
+(see
+.BR memkind_default (3))
+followed by
+.BR memkind_arena_create_map ()
+described below.
+.PP
+.BR memkind_arena_create_map ()
+creates the
+.I arena_map
+array for the memkind structure pointed to by
+.IR kind
+which can be indexed by the
+.BR ops.get_arena ()
+function from the kind's operations.  If get_arena points
+.BR memkind_thread_get_arena ()
+then there will be four arenas created for each processor,
+and if get_arena points to
+.BR memkind_bijective_get_arena ()
+then just one arena is created.
+.PP
+.BR memkind_arena_destroy ()
+is an implementation of the memkind "destroy" operation for memory
+kinds that use jemalloc.  This releases all of the resources
+allocated by
+.BR memkind_arena_create ().
+.PP
+.BR memkind_arena_malloc ()
+is an implementation of the memkind "malloc" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.
+.PP
+.BR memkind_arena_calloc ()
+is an implementation of the memkind "calloc" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.
+.PP
+.BR memkind_arena_posix_memalign ()
+is an implementation of the memkind "posix_memalign" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.  The POSIX standard requires that
+.BR posix_memalign (3)
+may not set
+.I errno
+however the jemalloc's
+.BR mallocx ()
+routine may.  In an attempt to abide by the standard
+.I errno
+is recorded before calling jemalloc's
+.BR mallocx ()
+and then reset after the call.
+.PP
+.BR memkind_arena_realloc ()
+is an implementation of the memkind "realloc" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.
+.PP
+.BR memkind_arena_realloc_with_kind_detect ()
+function will look up for kind associated to the allocated memory referenced by
+.I ptr
+and call (depending on kind value)
+.BR memkind_arena_realloc ()
+or
+.BR memkind_default_realloc ().
+.PP
+.BR memkind_thread_get_arena ()
+retrieves the
+.I arena
+index that is bound to to the calling thread based on a hash of its
+thread ID.  The
+.I arena
+index can be used with the MALLOCX_ARENA macro to set flags for jemalloc's
+.BR mallocx ().
+.PP
+.BR memkind_bijective_arena_get_arena ()
+retrieves the
+.I arena
+index to be used with the MALLOCX_ARENA macro to set flags for jemalloc's
+.BR mallocx ().
+Use of this operation implies that only one arena is used for the
+.IR kind .
+.PP
+.BR memkind_arena_free ()
+is an implementation of the memkind "free" operation for memory
+kinds that use jemalloc.  It causes the allocated memory referenced by
+.IR ptr ,
+which must have been returned by a previous call to
+.BR memkind_arena_malloc (),
+.BR memkind_arena_calloc ()
+or
+.BR memkind_arena_realloc ()
+to be made available for future allocations.
+It uses the memkind "get_arena" operation to select the arena.
+.PP
+.BR memkind_arena_free_with_kind_detect ()
+function will look up for kind associated to the allocated memory referenced by
+.I ptr
+and call
+.BR memkind_arena_free ().
+.PP
+.BR memkind_arena_detect_kind ()
+returns pointer to memory kind structure associated with given allocated memory referenced by
+.IR ptr .
+.PP
+.BR get_kind_by_arena ()
+returns pointer to memory kind structure associated with given arena.
+.PP
+.BR memkind_arena_malloc_usable_size ()
+is an implementation of the memkind "usable_size" operation for memory kinds that use jemalloc.
+.PP
+.BR memkind_arena_finalize ()
+is an implementation of the memkind "finalize" operation for memory kinds that
+use jemalloc. This function releases all resources allocated by
+.BR memkind_arena_create ()
+and it's called when
+.BR main ()
+finishes or after calling
+.BR exit ()
+function.
+.PP
+.BR memkind_arena_init ()
+creates arena map with proper hooks per specified kind.
+.PP
+.BR memkind_arena_update_memory_usage_policy ()
+function changes time, which determine how fast jemalloc returns unused pages back to
+the operating system, in other words how fast it deallocates file space.
+.PP
+.BR memkind_arena_enable_background_threads ()
+enables internal background worker threads in jemalloc.
+The
+.I threads_limit
+specify limit of background threads which will be enabled ( 0 means no limit).
+.PP
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/man/memkind_dax_kmem.3 b/man/memkind_dax_kmem.3
new file mode 100644
index 0000000..99300a9
--- /dev/null
+++ b/man/memkind_dax_kmem.3
@@ -0,0 +1,60 @@
+.\"
+.\" Copyright (C) 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_DAX_KMEM" 3 "2019-07-19" "Intel Corporation" "MEMKIND_DAX_KMEM" \" -*- nroff -*-
+.SH "NAME"
+memkind_dax_kmem \- DAX KMEM memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_dax_kmem_all_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.SH DESCRIPTION
+.PP
+DAX KMEM memory memkind operations.
+.PP
+.BR memkind_dax_kmem_all_get_mbind_nodemask ()
+sets the
+.I nodemask
+bits to one that correspond to the all persistent memory NUMA nodes in
+the system. All other bits up to
+.I maxnode
+are set to zero.
+The
+.I nodemask
+can be used in conjunction with the
+.BR mbind (2)
+system call.
+.SH "COPYRIGHT"
+Copyright (C) 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/man/memkind_default.3 b/man/memkind_default.3
new file mode 100644
index 0000000..3880f18
--- /dev/null
+++ b/man/memkind_default.3
@@ -0,0 +1,226 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_DEFAULT" 3 "2015-04-21" "Intel Corporation" "MEMKIND_DEFAULT" \" -*- nroff -*-
+.SH "NAME"
+memkind_default \- default implementations for memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes), or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_default_create(struct memkind " "*kind" ", struct memkind_ops " "*ops" ", const char " "*name" );
+.br
+.BI "int memkind_default_destroy(struct memkind " "*kind" );
+.br
+.BI "void *memkind_default_malloc(struct memkind " "*kind" ", size_t " "size" );
+.br
+.BI "void *memkind_default_calloc(struct memkind " "*kind" ", size_t " "num" ", size_t " "size" );
+.br
+.BI "int memkind_default_posix_memalign(struct memkind " "*kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.br
+.BI "void *memkind_default_realloc(struct memkind " "*kind" ", void " "*ptr" ", size_t " "size" );
+.br
+.BI "void memkind_default_free(struct memkind " "*kind" ", void " "*ptr" );
+.br
+.BI "void *memkind_default_mmap(struct memkind " "*kind" ", void " "*addr" ", size_t " "size" );
+.br
+.BI "int memkind_default_mbind(struct memkind " "*kind" ", void " "*ptr" ", size_t " "len" );
+.br
+.BI "int memkind_default_get_mmap_flags(struct memkind " "*kind" ", int " "*flags" );
+.br
+.BI "int memkind_default_get_mbind_mode(struct memkind " "*kind" ", int " "*mode" );
+.br
+.BI "size_t memkind_default_malloc_usable_size(struct memkind " "*kind" ", void " "*ptr" );
+.br
+.BI "int memkind_preferred_get_mbind_mode(struct memkind " "*kind" ", int " "*mode" );
+.br
+.BI "int memkind_interleave_get_mbind_mode(struct memkind " "*kind" ", int " "*mode" );
+.br
+.BI "int memkind_nohugepage_madvise(struct memkind " "*kind" ", void " "*addr" ", size_t " "size" );
+.br
+.BI "int memkind_posix_check_alignment(struct memkind " "*kind" ", size_t " "alignment" );
+.br
+.BI "int memkind_default_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.BI "void memkind_default_init_once(void);"
+.br
+.BI "bool size_out_of_bounds(size_t " "size" );
+.br
+.SH DESCRIPTION
+.PP
+Default implementations for memkind operations which include a several
+useful methods that are not part of the
+.B MEMKIND_DEFAULT
+kind which is a fall through to the jemalloc implementation.
+.PP
+.BR memkind_default_create ()
+implements the required start up for every kind. If a kind does not
+point to this function directly for its
+.BR ops.create ()
+operation, then the function that it points to must call
+.BR memkind_default_create ()
+at its start.
+.PP
+.BR memkind_default_destroy ()
+implements the required shutdown for every kind. If a kind does not
+point to this function directly for its
+.BR ops.destroy ()
+operation, then the function that it points to must call
+.BR memkind_default_destroy ()
+at its end.
+.PP
+.BR memkind_default_malloc ()
+is a direct call through the jemalloc's
+.BR malloc ().
+.PP
+.BR memkind_default_calloc ()
+is a direct call through the jemalloc's
+.BR calloc ().
+.PP
+.BR memkind_default_posix_memalign ()
+is a direct call through the jemalloc's
+.BR posix_memalign ().
+.PP
+.BR memkind_default_realloc ()
+is a direct call through the jemalloc's
+.BR realloc ().
+.PP
+.BR memkind_default_free ()
+is a direct call through the jemalloc's
+.BR free ().
+Note that this method can be called on any pointer returned by a
+jemalloc allocation, and in particular, all of the arena
+allocations described in
+.BR memkind_arena (3)
+can use this function for freeing.
+.PP
+.BR memkind_default_mmap ()
+This calls the ops->get_mmap_flags()
+operations for the kind, or falls back on the default implementations
+if the function pointers are
+.IR "NULL" .
+The results of these calls are
+passed to the
+.BR mmap (2)
+call to allocate pages from the operating system.  The
+.I addr
+is the hint passed through to
+.BR mmap (2)
+and
+.I size
+is the size of the buffer to be allocated.  The return value is the
+allocated buffer or
+.B MAP_FAILED
+in the case of an error.
+.PP
+.BR memkind_default_mbind ()
+makes calls the kind's
+.BR ops.get_mbind_nodemask ()
+and
+.BR ops.get_mbind_mode ()
+operations to gather inputs and then calls the
+.BR mbind (2)
+system call using the results along with and user input
+.I ptr
+and
+.IR len .
+.PP
+.BR memkind_default_get_mmap_flags ()
+sets
+.I flags
+to
+.BR "MAP_PRIVATE | MAP_ANONYMOUS" .
+See
+.BR mmap (2)
+for more information about these flags.
+.PP
+.BR memkind_default_get_mbind_mode ()
+sets
+.I mode
+to
+.BR MPOL_BIND .
+See
+.BR mbind (2)
+for more information about this flag.
+.PP
+.BR memkind_default_malloc_usable_size ()
+is a direct call through the jemalloc's
+.BR malloc_usable_size ().
+.PP
+.BR memkind_preferred_get_mbind_mode ()
+sets
+.I mode
+to
+.BR MPOL_PREFERRED .
+See
+.BR mbind (2)
+for more information about this flag.
+.PP
+.BR memkind_interleave_get_mbind_mode ()
+sets
+.I mode
+to
+.BR MPOL_INTERLEAVE .
+See
+.BR mbind (2)
+for more information about this flag.
+.PP
+.BR memkind_nohugepage_madvise ()
+calls
+.BR madvise (2)
+with the
+.B MADV_NOHUGEPAGE
+advice.
+See
+.BR madvise (2)
+for more information about this option.
+.PP
+.BR memkind_posix_check_alignment ()
+can be used to check the alignment value for
+.BR memkind_posix_memalign ()
+to ensure that is abides by the POSIX requirements:
+alignment must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+.PP
+.BR memkind_default_get_mbind_nodemask ()
+wraps jemalloc's
+.BR copy_bitmask_to_bitmask .
+This function copies body of the bitmask structure into passed pointer.
+.PP
+.BR memkind_default_init_once ()
+initializes heap manager.
+.PP
+.BR size_out_of_bounds ()
+returns true if given size is out of bounds, otherwise will return false.
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/man/memkind_hbw.3 b/man/memkind_hbw.3
new file mode 100644
index 0000000..afd9c72
--- /dev/null
+++ b/man/memkind_hbw.3
@@ -0,0 +1,160 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_HBW" 3 "2016-06-13" "Intel Corporation" "MEMKIND_HBW" \" -*- nroff -*-
+.SH "NAME"
+memkind_hbw \- high bandwidth memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.nf
+.sp
+.BI "int memkind_hbw_check_available(struct memkind " "*kind" );
+.br
+.BI "int memkind_hbw_hugetlb_check_available(struct memkind " "*kind" );
+.br
+.BI "int memkind_hbw_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.BI "int memkind_hbw_all_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.BI "void memkind_hbw_init_once(void);"
+.br
+.BI "void memkind_hbw_hugetlb_init_once(void);"
+.br
+.BI "void memkind_hbw_all_hugetlb_init_once(void);"
+.br
+.BI "void memkind_hbw_preferred_init_once(void);"
+.br
+.BI "void memkind_hbw_preferred_hugetlb_init_once(void);"
+.br
+.BI "void memkind_hbw_interleave_init_once(void);"
+.br
+.SH DESCRIPTION
+.PP
+High bandwidth memory memkind operations.
+.PP
+.BR memkind_hbw_check_available ()
+returns zero if library was able to detect heterogeneous NUMA node
+bandwidths.   Returns
+.B MEMKIND_ERROR_UNAVAILABLE
+if the detection mechanism failed.
+Detection mechanism can be also overridden by the
+environment variable
+.B MEMKIND_HBW_NODES
+as described in the
+.BR memkind (3)
+man page.
+.PP
+.BR memkind_hbw_hugetlb_check_available ()
+In addition to checking for high bandwidth memory as is done by
+.BR memkind_hbw_check_available (),
+this also checks for 2MB huge pages as
+is done by
+.BR memkind_hugetlb_check_available_2mb ().
+.PP
+.BR memkind_hbw_get_mbind_nodemask ()
+sets the
+.I nodemask
+bit to one that corresponds to the high bandwidth NUMA node that has
+the closest NUMA distance to the CPU of the calling process.
+All other bits up to
+.I maxnode
+are set to zero.
+The
+.I nodemask
+can be used in conjunction with the
+.BR mbind (2)
+system call.
+.PP
+.BR memkind_hbw_all_get_mbind_nodemask ()
+sets the
+.I nodemask
+bits to one that correspond to the all high bandwidth NUMA nodes in
+the system. All other bits up to
+.I maxnode
+are set to zero.
+The
+.I nodemask
+can be used in conjunction with the
+.BR mbind (2)
+system call.
+.PP
+.BR memkind_hbw_init_once ()
+This function initializes
+.B MEMKIND_HBW
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_hugetlb_init_once ()
+This function initializes
+.B MEMKIND_HBW_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_hugetlb_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_preferred_init_once ()
+This function initializes
+.B MEMKIND_HBW_PREFERRED
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_preferred_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_preferred_hugetlb_init_once ()
+This function initializes
+.B MEMKIND_HBW_PREFERRED_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_preferred_hugetlb_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_all_hugetlb_init_once ()
+This function initializes
+.B MEMKIND_HBW_ALL_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_all_hugetlb_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_interleave_init_once ()
+This function initializes
+.B MEMKIND_HBW_INTERLEAVE
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_interleave_init_once ()
+may reserve some extra memory.
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/man/memkind_hugetlb.3 b/man/memkind_hugetlb.3
new file mode 100644
index 0000000..52c42b1
--- /dev/null
+++ b/man/memkind_hugetlb.3
@@ -0,0 +1,77 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_HUGETLB" 3 "2015-04-21" "Intel Corporation" "MEMKIND_HUGETLB" \" -*- nroff -*-
+.SH "NAME"
+memkind_hugetlb \- hugetlb memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_hugetlb_check_available_2mb(struct memkind " "*kind" );
+.br
+.BI "int memkind_hugetlb_get_mmap_flags(struct memkind " "*kind" ", int " "*flags" );
+.br
+.BI "void memkind_hugetlb_init_once(void);"
+.br
+
+.SH DESCRIPTION
+.PP
+The hugetlb memory memkind operations enable memory kinds which use
+the Linux hugetlbfs file system.  For more information about the
+hugetlbfs see link below.
+.br
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.PP
+.BR memkind_hugetlb_check_available_2mb ()
+check if there are 2MB pages reserved in the default hugetlbfs.  If
+the kind implements
+.BR ops.get_mbind_nodemask (),
+then only the NUMA nodes
+set by the nodemask are checked, otherwise every NUMA node is checked.
+.PP
+.BR memkind_hugetlb_get_mmap_flags ()
+sets the flags for the
+.BR mmap ()
+system call such that the hugetlbfs is utilized for allocations.
+.PP
+.BR memkind_hugetlb_init_once ()
+this function initializes
+.B MEMKIND_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hugetlb_init_once ()
+may reserve some extra memory.
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/man/memkind_interleave.3 b/man/memkind_interleave.3
new file mode 100644
index 0000000..9e110f2
--- /dev/null
+++ b/man/memkind_interleave.3
@@ -0,0 +1,51 @@
+.\"
+.\" Copyright (C) 2016 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_INTERLEAVE" 3 "2016-02-19" "Intel Corporation" "MEMKIND_INTERLEAVE" \" -*- nroff -*-
+.SH "NAME"
+memkind_interleave \- interleave memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "void memkind_interleave_init_once(void);"
+.br
+.SH DESCRIPTION
+.PP
+.BR memkind_interleave_init_once ()
+initializes MEMKIND_INTERLEAVE kind and it should not be called more than once.
+.BR Note:
+.BR memkind_interleave_init_once ()
+may reserve some extra memory.
+.SH "COPYRIGHT"
+Copyright (C) 2016 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/man/memkind_pmem.3 b/man/memkind_pmem.3
new file mode 100644
index 0000000..93d02a1
--- /dev/null
+++ b/man/memkind_pmem.3
@@ -0,0 +1,114 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_PMEM" 3 "2015-04-21" "Intel Corporation" "MEMKIND_PMEM" \" -*- nroff -*-
+.SH "NAME"
+memkind_pmem \- file-backed memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_pmem_create(struct memkind " "*kind" ", struct memkind_ops " "*ops" ", const char " "*name" );
+.br
+.BI "int memkind_pmem_destroy(struct memkind " "*kind" );
+.br
+.BI "void *memkind_pmem_mmap(struct memkind " "*kind" ", void " "*addr" ", size_t " "size" );
+.br
+.BI "int memkind_pmem_get_mmap_flags(struct memkind " "*kind" ", int " "*flags" );
+.br
+.SH DESCRIPTION
+.PP
+The pmem memory memkind operations enable memory kinds built on memory-mapped
+files.  These support traditional
+.B volatile
+memory allocation in a fashion similar to
+.BR libvmem (3)
+library.  It uses the
+.BR mmap (2)
+system call to create a pool of volatile memory.  Such memory may have different
+attributes, depending on the file system containing the memory-mapped files.
+(See also
+.IR http://pmem.io/pmdk/libvmem ).
+.PP
+The pmem memkinds are most useful when used with
+.I Direct Access
+storage (DAX), which is memory-addressable persistent storage
+that supports load/store access without being paged via the system page cache.
+A Persistent Memory-aware file system is typically used to provide this
+type of access.
+.PP
+The most convenient way to create pmem memkinds is to use
+.BR memkind_create_pmem ()
+or
+.BR memkind_create_pmem_with_config ()
+(see
+.BR memkind (3)).
+.PP
+.BR memkind_pmem_create ()
+is an implementation of the memkind "create" operation for file-backed memory
+kinds.  This allocates a space for some pmem-specific metadata, then calls
+.BR memkind_arena_create ()
+(see
+.BR memkind_arena (3))
+.PP
+.BR memkind_pmem_destroy ()
+is an implementation of the memkind "destroy" operation for file-backed memory
+kinds.  This releases all of the resources
+allocated by
+.BR memkind_pmem_create ()
+and allows the file system space to be reclaimed.
+.PP
+.BR memkind_pmem_mmap ()
+allocates the file system space for a block of
+.I size
+bytes in the memory-mapped file associated with given kind.
+The
+.I addr
+hint is ignored.  The return value is the address of mapped memory region or
+.B MAP_FAILED
+in the case of an error.
+.PP
+.BR memkind_pmem_get_mmap_flags ()
+sets
+.I flags
+to
+.BR "MAP_SHARED" .
+See
+.BR mmap (2)
+for more information about these flags.
+.TP
+.B MEMKIND_PMEM_CHUNK_SIZE
+The size of the PMEM chunk size.
+.SH "COPYRIGHT"
+Copyright (C) 2015 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR libvmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/man/memkindallocator.3 b/man/memkindallocator.3
new file mode 100644
index 0000000..9533519
--- /dev/null
+++ b/man/memkindallocator.3
@@ -0,0 +1,194 @@
+.\"
+.\" Copyright (C) 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKINDALLOCATOR" 3 "2019-09-24" "Intel Corporation" "MEMKINDALLOCATOR" \" -*- nroff -*-
+.SH "NAME"
+libmemkind::static_kind::allocator<T> \- The C++ allocator compatible with the C++ standard library allocator concepts
+.br
+.BR Note:
+.I memkind_allocator.h
+functionality is considered as stable API (STANDARD API).
+.SH "SYNOPSIS"
+.nf
+.B #include <memkind_allocator.h>
+.sp
+.B Link with -lmemkind
+.sp
+.BI "libmemkind::static_kind::allocator(libmemkind::kinds " "kind" );
+.br
+.BI "template <typename U> libmemkind::static_kind::allocator<T>::allocator(const libmemkind::static_kind::allocator<U>&)" " "noexcept;
+.br
+.BI "template <typename U> libmemkind::static_kind::allocator(const allocator<U>&& " "other" ) " "noexcept;
+.br
+.BI "libmemkind::static_kind::allocator<T>::~allocator();
+.br
+.BI "T *libmemkind::static_kind::allocator<T>::allocate(std::size_t " "n" ) " "const;
+.br
+.BI "void libmemkind::static_kind::allocator<T>::deallocate(T " "*p" ", std::size_t " "n" ) " "const;
+.br
+.BI "template <class U, class... Args> void libmemkind::static_kind::allocator<T>::construct(U " "*p" ", Args... " "args" ) " "const;
+.br
+.BI "void libmemkind::static_kind::allocator<T>::destroy(T " "*p" ) " "const;
+.fi
+.SH "DESCRIPTION"
+The
+.BR libmemkind::static_kind::allocator<T>
+is intended to be used with STL containers to allocate from static kinds memory. Memory management is based on memkind library. Refer
+.BR memkind (3)
+man page for more details.
+.PP
+The
+.BR libmemkind::kinds
+specifies allocator static kinds of memory, representing type of memory which offers different characteristics. The available types of allocator kinds of memory:
+.PP
+.B libmemkind::kinds::DEFAULT
+Default allocation using standard memory and default page size.
+.PP
+.B libmemkind::kinds::HUGETLB
+Allocate from standard memory using huge pages. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::INTERLEAVE
+Allocate pages interleaved across all NUMA nodes with transparent huge pages disabled.
+.PP
+.B libmemkind::kinds::HBW
+Allocate from the closest high bandwidth memory NUMA node at the time of allocation. If there is not enough high bandwidth memory to satisfy the request, errno is set to ENOMEM and the allocated pointer is set to
+.IR "NULL" .
+.PP
+.B libmemkind::kinds::HBW_ALL
+Same as
+.B libmemkind::kinds::HBW
+except decision regarding closest NUMA node is postponed until the time of first write.
+.PP
+.B libmemkind::kinds::HBW_HUGETLB
+Same as
+.B libmemkind::kinds::HBW
+except the allocation is backed by huge pages. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::HBW_ALL_HUGETLB
+Combination of
+.B libmemkind::kinds::HBW_ALL
+and
+.B libmemkind::kinds::HBW_HUGETLB
+properties. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::HBW_PREFERRED
+Same as
+.B libmemkind::kinds::HBW
+except that if there is not enough high bandwidth memory to satisfy the request, the allocation will fall back on standard memory.
+.PP
+.B libmemkind::kinds::HBW_PREFERRED_HUGETLB
+Same as
+.B libmemkind::kinds::HBW_PREFERRED
+except the allocation is backed by huge pages. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::HBW_INTERLEAVE
+Same as
+.B libmemkind::kinds::HBW
+except that the pages that support the allocation are interleaved across all high bandwidth nodes and transparent huge pages are disabled.
+.PP
+.B libmemkind::kinds::REGULAR
+Allocate from regular memory using the default page size. Regular means general purpose memory from the NUMA nodes containing CPUs.
+.PP
+.B libmemkind::kinds::DAX_KMEM
+Allocate from the closest persistent memory NUMA node at the time of allocation. If there is not enough memory in the closest persistent memory NUMA node to satisfy the request,
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.PP
+.B libmemkind::kinds::DAX_KMEM_ALL
+Allocate from the closest persistent memory NUMA node available at the time of allocation. If there is not enough memory on any of persistent memory NUMA nodes to satisfy the request,
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.PP
+.B libmemkind::kinds::DAX_KMEM_PREFERRED
+Same as
+.B libmemkind::kinds::DAX_KMEM
+except that if there is not enough memory in the closest persistent memory NUMA node to satisfy the request, the allocation will fall back on other memory NUMA nodes.
+.BR Note:
+For this kind, the allocation will not succeed if two or more
+persistent memory NUMA nodes are in the same shortest distance to the same CPU on which process is eligible to run.
+Check on that eligibility is done upon starting the application.
+.PP
+All public member types and functions correspond to standard library allocator concepts and definitions. The current implementation supports C++11 standard.
+.PP
+Template arguments:
+.br
+.I T
+is an object type aliased by value_type.
+.br
+.I U
+is an object type.
+.PP
+.BR Note:
+.br
+.BI "T *libmemkind::static_kind::allocator<T>::allocate(std::size_t " "n")
+allocates memory using
+.BR memkind_malloc ().
+Throw
+.I std::bad_alloc
+when:
+.RS
+.I n
+= 0
+.RE
+.br
+.RS
+.br
+or there is not enough memory to satisfy the request.
+.RE
+.PP
+.BI "libmemkind::static_kind::allocator<T>::deallocate(T " "*p" ", std::size_t " "n")
+deallocates memory associated with pointer returned by
+.BR allocate ()
+using
+.BR memkind_free ().
+.SH "SYSTEM CONFIGURATION"
+Interfaces for obtaining 2MB (HUGETLB) memory need allocated
+huge pages in the kernel's huge page pool.
+.TP
+.B HUGETLB (huge pages)
+Current number of "persistent" huge pages can be read from
+.I /proc/sys/vm/nr_hugepages
+file.
+Proposed way of setting hugepages is:
+.BR "sudo sysctl vm.nr_hugepages=<number_of_hugepages>" .
+More information can be found here:
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.SH "COPYRIGHT"
+Copyright (C) 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3)
diff --git a/man/pmemallocator.3 b/man/pmemallocator.3
new file mode 100644
index 0000000..3c07a4c
--- /dev/null
+++ b/man/pmemallocator.3
@@ -0,0 +1,116 @@
+.\"
+.\" Copyright (C) 2018 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "PMEMALLOCATOR" 3 "2018-09-13" "Intel Corporation" "PMEMALLOCATOR" \" -*- nroff -*-
+.SH "NAME"
+libmemkind::pmem::allocator<T> \- The C++ allocator compatible with the C++ standard library allocator concepts
+.br
+.BR Note:
+.I pmem_allocator.h
+functionality is considered as stable API (STANDARD API).
+.SH "SYNOPSIS"
+.nf
+.B #include <pmem_allocator.h>
+.sp
+.B Link with -lmemkind
+.sp
+.BI "libmemkind::pmem::allocator(const char " "*dir" ", size_t " "max_size" );
+.br
+.BI "libmemkind::pmem::llocator(const char " "&dir" ", size_t " "max_size" ", libmemkind::allocation_policy " "alloc_policy" );
+.br
+.BI "libmemkind::pmem::allocator(const std::string " "&dir" ", size_t " "max_size" );
+.br
+.BI "libmemkind::pmem::allocator(const std::string " "&dir" ", size_t " "max_size" ", libmemkind::allocation_policy " "alloc_policy" );
+.br
+.BI "template <typename U> libmemkind::pmem::allocator<T>::allocator(const libmemkind::pmem::allocator<U>&)" " "noexcept;
+.br
+.BI "template <typename U> libmemkind::pmem::allocator(const allocator<U>&& " "other" ) " "noexcept;
+.br
+.BI "libmemkind::pmem::allocator<T>::~allocator();
+.br
+.BI "T *libmemkind::pmem::allocator<T>::allocate(std::size_t " "n" ) " "const;
+.br
+.BI "void libmemkind::pmem::allocator<T>::deallocate(T " "*p" ", std::size_t " "n" ) " "const;
+.br
+.BI "template <class U, class... Args> void libmemkind::pmem::allocator<T>::construct(U " "*p" ", Args... " "args" ) " "const;
+.br
+.BI "void libmemkind::pmem::allocator<T>::destroy(T " "*p" ) " "const;
+.fi
+.SH "DESCRIPTION"
+The
+.BR libmemkind::pmem::allocator<T>
+is intended to be used with STL containers to allocate persistent memory. Memory management is based on memkind_pmem (memkind library). Refer
+.BR memkind_pmem (3)
+and
+.BR memkind (3)
+man page for more details.
+.PP
+The
+.BR libmemkind::allocation_policy
+specifies allocator memory usage policy, which allows to tune up memory utilization. The available types of allocator usage policy:
+.PP
+.B libmemkind::allocation_policy::DEFAULT
+Default allocator memory usage policy.
+.PP
+.B libmemkind::allocation_policy::CONSERVATIVE
+Conservative allocator memory usage policy - prioritize memory usage at cost of performance.
+.PP
+All public member types and functions corresponds to standard library allocator concepts and definitions. The current implementation supports C++11 standard.
+.PP
+Template arguments:
+.br
+.I T
+is an object type aliased by value_type.
+.br
+.I U
+is an object type.
+.PP
+.BR Note:
+.br
+.BI "T *libmemkind::pmem::allocator<T>::allocate(std::size_t " "n")
+allocates persistent memory using
+.BR memkind_malloc ().
+Throw
+.I std::bad_alloc
+when:
+.br
+.RS
+.I n
+= 0
+.RE
+.br
+.RS
+or there is not enough memory to satisfy the request.
+.RE
+.PP
+.BI "libmemkind::pmem::allocator<T>::deallocate(T " "*p" ", std::size_t " "n")
+deallocates memory associated with pointer returned by
+.BR allocate ()
+using
+.BR memkind_free ().
+
+.SH "COPYRIGHT"
+Copyright (C) 2018 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind_pmem (3),
+.BR memkind (3)
diff --git a/memkind-1.10.0/.codecov.yml b/memkind-1.10.0/.codecov.yml
new file mode 100644
index 0000000..d42f217
--- /dev/null
+++ b/memkind-1.10.0/.codecov.yml
@@ -0,0 +1,10 @@
+ignore:
+  - "autohbw/.*"
+  - "examples/autohbw*"
+  - "examples/*example*"
+  - "examples/memkind*"
+  - "test/.*"
+comment:
+  layout: "diff"
+  behaviour: default
+  require_changes: true
diff --git a/memkind-1.10.0/.gitignore b/memkind-1.10.0/.gitignore
new file mode 100644
index 0000000..b581c84
--- /dev/null
+++ b/memkind-1.10.0/.gitignore
@@ -0,0 +1,126 @@
+*.a
+*.la
+*.lo
+*.o
+*.so
+.cproject
+.deps/
+.dirstamp
+.libs/
+.project
+/Makefile
+/Makefile.in
+/VERSION
+/aclocal.m4
+/ar-lib
+/compile
+/config.guess
+/config.h
+/config.h.in
+/config.log
+/config.status
+/config.sub
+/config_tls.h
+/configure
+/depcomp
+/examples/*.la
+/examples/*.lo
+/examples/*.o
+/examples/.deps/
+/examples/.dirstamp
+/examples/.libs/
+/examples/autohbw_candidates
+/examples/filter_memkind
+/examples/hello_hbw
+/examples/hello_memkind
+/examples/hello_memkind_debug
+/examples/memkind_allocated
+/examples/memkind_cpp_allocator
+/examples/memkind_get_stat
+/examples/pmem_alignment
+/examples/pmem_and_dax_kmem_kind
+/examples/pmem_and_default_kind
+/examples/pmem_config
+/examples/pmem_cpp_allocator
+/examples/pmem_detect_kind
+/examples/pmem_free_with_unknown_kind
+/examples/pmem_kinds
+/examples/pmem_malloc
+/examples/pmem_malloc_unlimited
+/examples/pmem_multithreads
+/examples/pmem_multithreads_onekind
+/examples/pmem_usable_size
+/install-sh
+/jemalloc/configure
+/jemalloc/obj/
+/libtool
+/ltmain.sh
+/m4/
+/memkind-*.spec
+/memkind-*.tar.gz
+/memkind-auto-dax-kmem-nodes
+/memkind-hbw-nodes
+/missing
+/src/memkind_defines.h.in
+/stamp-h1
+/stamp-h2
+/test-driver
+/test-suite.log
+/test/*.o
+/test/*.pyc
+/test/*.so
+/test/.deps/
+/test/.dirstamp
+/test/.libs/
+/test/all_tests
+/test/alloc_benchmark_glibc
+/test/alloc_benchmark_hbw
+/test/alloc_benchmark_pmem
+/test/alloc_benchmark_tbb
+/test/allocator_perf_tool_tests
+/test/autohbw_candidates
+/test/autohbw_test_helper
+/test/check.sh.log
+/test/check.sh.trs
+/test/dax_kmem_test
+/test/decorator_test
+/test/environ_err_dax_kmem_malloc_positive_test
+/test/environ_err_dax_kmem_malloc_test
+/test/environ_err_hbw_malloc_test
+/test/environerr_hbw_malloc_test
+/test/environerr_test
+/test/filter_memkind
+/test/fragmentation_benchmark_pmem
+/test/freeing_memory_segfault_test
+/test/gb_page_tests_bind_policy
+/test/hello_hbw
+/test/hello_memkind
+/test/hello_memkind_debug
+/test/locality_test
+/test/mallocerr_test
+/test/memkind_allocated
+/test/memkind_cpp_allocator
+/test/memkind_get_stat
+/test/memkind_stat_test
+/test/perf_tool
+/test/performance_test
+/test/performance/.deps/
+/test/pmem_alignment
+/test/pmem_and_dax_kmem_kind
+/test/pmem_and_default_kind
+/test/pmem_config
+/test/pmem_cpp_allocator
+/test/pmem_detect_kind
+/test/pmem_free_with_unknown_kind
+/test/pmem_kinds
+/test/pmem_malloc
+/test/pmem_malloc_unlimited
+/test/pmem_multithreads
+/test/pmem_multithreads_onekind
+/test/pmem_test
+/test/pmem_usable_size
+/test/trace_mechanism_test_helper
+/test/defrag_reallocate
+/vs
+autom4te.cache/
+memkind.pc
diff --git a/memkind-1.10.0/.travis.yml b/memkind-1.10.0/.travis.yml
new file mode 100644
index 0000000..3899848
--- /dev/null
+++ b/memkind-1.10.0/.travis.yml
@@ -0,0 +1,65 @@
+dist: bionic
+sudo: required
+language: cpp
+cache: ccache
+git:
+  depth: false
+matrix:
+  fast_finish: true
+  include:
+
+    - name: "Astyle"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+          packages:
+            - astyle
+      script:
+        - ./astyle.sh
+
+    - name: "GCC 4.8"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+          packages:
+            - g++-4.8
+            - libnuma-dev
+      env:
+         - MATRIX_EVAL="CC=gcc-4.8 && CXX=g++-4.8"
+      install:
+      script:
+        - ./build.sh
+
+    - name: "GCC 8.0"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+          packages:
+            - g++-8
+            - libnuma-dev
+      env:
+         - MATRIX_EVAL="CC=gcc-8 && CXX=g++-8"
+      install:
+      script:
+        - ./build.sh
+
+    - name: "Clang 8"
+      addons:
+        apt:
+          sources:
+            - ubuntu-toolchain-r-test
+            - llvm-toolchain-bionic-8
+          packages:
+            - clang-8
+            - libnuma-dev
+      env:
+        - MATRIX_EVAL="CC=clang-8 && CXX=clang++-8"
+      install:
+      script:
+        - ./build.sh
+
+before_install:
+  - eval "${MATRIX_EVAL}"
diff --git a/memkind-1.10.0/AUTHORS b/memkind-1.10.0/AUTHORS
new file mode 100644
index 0000000..ea4ca98
--- /dev/null
+++ b/memkind-1.10.0/AUTHORS
@@ -0,0 +1,26 @@
+Christopher Cantalupo <christopher.m.cantalupo@intel.com>
+Vishwanath Venkatesan <vishwanath.venkatesan@intel.com>
+Steven Hampson <steven.t.hampson@intel.com>
+Krzysztof Czurylo <krzysztof.czurylo@intel.com>
+Leobardo Rountree <leobardo.r.rountree@intel.com>
+Hector Barajas Villalobos <hector.a.barajas.villalobos@intel.com>
+Ruchira Sasanka <ruchira.sasanka@intel.com>
+Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>
+Rafael Aquini <aquini@redhat.com>
+Alexandr Konovalov <alexandr.konovalov@intel.com>
+Grzegorz Ozanski <grzegorz.ozanski@intel.com>
+Artur Koziej <artur.koziej@intel.com>
+Jakub Dlugolecki <jakub.dlugolecki@intel.com>
+Lukasz Anaczkowski <lukasz.anaczkowski@intel.com>
+Agata Wozniak <agata.wozniak@intel.com>
+Grzegorz Andrejczuk <grzegorz.andrejczuk@intel.com>
+Kamil Patelczyk <kamil.patelczyk@intel.com>
+Jeff Hammond <jeff.r.hammond@intel.com>
+Aleksandra Hernik <aleksandra.hernik@intel.com>
+Pawel Kochanek <pawel.b.kochanek@intel.com>
+Katarzyna Wasiuta <katarzyna.wasiuta@intel.com>
+Krzysztof Filipek <krzysztof.filipek@intel.com>
+Michal Biesek <michal.biesek@intel.com>
+Adam Borowski <adam.borowski@intel.com>
+Yuan Zhou <yuan.zhou@intel.com>
+Harald Servat <harald.servat@intel.com>
diff --git a/memkind-1.10.0/CONTRIBUTING b/memkind-1.10.0/CONTRIBUTING
new file mode 100644
index 0000000..ca0039b
--- /dev/null
+++ b/memkind-1.10.0/CONTRIBUTING
@@ -0,0 +1,134 @@
+CONTRIBUTING
+============
+
+This file is intended to help those interested in contributing to the
+memkind library.
+
+
+COMMUNICATION
+=============
+
+Please participate in the memkind mailing list:
+
+    https://lists.01.org/mailman/listinfo/memkind
+
+There is also the option of opening an issue through github:
+
+    https://github.com/memkind/memkind/issues
+
+The mailing list is intended for discussion and the issues are useful
+for tracking progress on tasks.  The TODO file lists out a number of
+topics, and in order to prioritize one of them please open a github
+issue with a related subject.
+
+
+TESTING
+=======
+
+The tests require a Linux kernel newer than 3.11 (the details are
+documented in the memkind README), and the reservation of 3000 huge
+pages.  The huge pages can be reserved with the following command:
+
+    $ sudo echo 3000 > /proc/sys/vm/nr_hugepages
+
+Only in the case where gigabyte pages have been reserved will the
+tests associated with gigabyte pages be executed.  Reserving gigabyte
+pages may require a modification to the kernel command line unless the
+kernel is quite recent.
+
+To test memkind simply execute the "make check" target after building.
+This target calls memkind/test/test.sh with parameters
+depending on the environment.
+
+Most of the tests are written within the gtest framework, however, as
+part of testing the example programs are also executed and the return
+code of the executable determines pass or fail.  The autotools test
+infrastructure is used as a high level executor, but it does not track
+individual tests.  The consequence of this is that the autotools
+output records only one high level test which passes in the case where
+every underlying test was successful and fails if any underlying test
+fails.  The individual test results are recorded in the directory
+called "gtest_output." Here you will find the log of the tests in
+gtest_output/test.out and a set of junit style xml results: one for
+each test.  Note that a side effect of having only one autotools test
+is that autotools parallel testing is disabled.  We have
+multi-threaded tests that use the OpenMP run-time which enables more
+purposeful and deterministic testing of threading issues.  Note that
+the OpenMP run-time is only required for testing, it is not used by
+the memkind library internally.
+
+
+CODING STYLE
+============
+
+Before submitting a patch for inclusion, please run the modified
+source code files through astyle with the following options
+
+    $ astyle --style=linux --indent=spaces=4 -S --max-continuation-indent=80 \
+             --max-code-length=80 --break-after-logical --indent-namespaces -z2 \
+             --align-pointer=name
+
+More information about astyle can be found here:
+
+    http://astyle.sourceforge.net/
+
+In C, source code constants are in all caps and everything else is in
+all lower case.  Underscores are used to separate words within a
+symbol name. No camel case shall be used in C code.  The test code is
+largely written in C++.  Here camel-case should be used for class
+names and should always have a capitalized first letter.  Other
+symbols in C++ should generally follow the C style.
+
+Most symbols with global scope shall be prefixed with "memkind_" or
+"hbw_" depending on which interface they are a part of.
+
+Any global variable shall have _g appended to the variable name and in
+most cases shall be statically scoped within a single compilation
+unit.  The exception to that rule are static memory kinds that are
+declared as extern within the associated interface header and are
+constant after initialization.  Global variables should be used in a
+very narrow set of circumstances and in most cases modifications
+should be guarded with pthread_once(3).
+
+Functions not used outside of the compilation unit shall be declared
+as static.  All functions which are not declared as static shall be
+documented in a man page and have an associated interface header file.
+
+Preprocessor mark-up is discouraged when other options are possible.
+Please use enum in place of #define when value control at configure or
+build time is not required.
+
+
+TESTS
+=====
+
+The current state of the tests is not nearly as well organized as it
+could be.  That being said, it is quite simple to add a new test.
+Most C++ files in the test directory are associated with a single
+gtest testing::Test class.  These classes usually have several
+associated test fixtures in the same file.  If a new test can be added
+as a fixture to an existing class, simply add the fixture to the file
+and the test will be incorporated into the test infrastructure.
+If a new class is required, create a new file and add it to the list
+of "test_all_tests_SOURCES" in memkind/test/Makfile.mk and it will
+be incorporated into the test infrastructure.
+
+There are a few files which define classes which are not google test
+classes.  These are check.cpp, trial_generator.cpp and main.cpp.  The
+check.cpp file defines a class Check that can be used to validate
+fundamental memkind features like NUMA node location, and page size.
+The trial_generator.cpp file is used to abstract a sequence of
+allocation and deallocation calls while performing checks on the
+results of each call; this can be used to apply similar tests to all
+of the different allocation APIs.  The main.cpp file is a simple
+wrapper around testing::InitGoogleTest and RUN_ALL_TESTS().
+
+
+SUBMITTING A PATCH
+==================
+
+Please be sure that all tests pass before submission and that the
+style conforms to the specifications given here.  If a new feature is
+implemented in the patch, please also include unit tests and an
+example which exercises this feature.  Once these requirements have
+been met, please submit a pull request through github.
diff --git a/memkind-1.10.0/COPYING b/memkind-1.10.0/COPYING
new file mode 100644
index 0000000..cc1c9cc
--- /dev/null
+++ b/memkind-1.10.0/COPYING
@@ -0,0 +1,25 @@
+Unless otherwise specified, files in the memkind source distribution are
+subject to the following license:
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+    * Redistributions of source code must retain the above copyright notice,
+      this list of conditions and the following disclaimer.
+    * Redistributions in binary form must reproduce the above copyright
+      notice, this list of conditions and the following disclaimer in the
+      documentation and/or other materials provided with the distribution.
+    * Neither the name of Intel Corporation nor the names of its contributors
+      may be used to endorse or promote products derived from this software
+      without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
diff --git a/memkind-1.10.0/ChangeLog b/memkind-1.10.0/ChangeLog
new file mode 100644
index 0000000..d887e18
--- /dev/null
+++ b/memkind-1.10.0/ChangeLog
@@ -0,0 +1,242 @@
+* Mon Dec 16 2019 Michal Biesek <michal.biesek@intel.com> v1.10.0
+- Provided another way to use Persistent Memory in memkind (MEMKIND_DAX_KMEM_* kinds)
+- Added C++ allocator for static kinds (including MEMKIND_DAX_KMEM_* kinds)
+- Provided an interface to get memkind allocation statistics
+- Provided a method to reduce fragmentation - memkind_defrag_reallocate()
+- Added support for background thread
+- Updated internally used jemalloc to upstream version 5.2.1
+- Extended hbw interface with hbw_malloc_usable_size()
+- Simplified building process
+
+* Wed Apr 3 2019 Michal Biesek <michal.biesek@intel.com> v1.9.0
+- Provided a second function to create PMEM kind memkind_create_pmem_with_config()
+- Provided the memkind_detect_kind() function to recognize a kind from memory allocation
+- Provided memkind_config_set_memory_usage_policy() function
+  that modifies the default memory usage policy behavior
+- Provided the ability to pass a NULL kind to memkind_malloc_usable_size() and memkind_realloc()
+- Unified the memkind_realloc() behavior for all kinds
+- Added the support for the pool_msize() function to TBB
+- Extended build arch to ppc64, ppc64le, s390x
+- Removed disable-stats from used jemalloc
+- Fixed compilation under the Clang and rpm build configuration
+- Provided the support for Dockerfile
+
+* Mon Dec 10 2018 Michal Biesek <michal.biesek@intel.com> v1.8.0
+- Fixed error with realloc/free method with passing thread-specific cache flag
+- Fixed error with memkind_create_pmem(), if other PMEM kind was destroyed before
+- Fixed error with zeroing large allocations in PMEM kind
+- Added support to create kind without maximum size limit of PMEM kind (max_size=0)
+- Extended memkind API with memkind_malloc_usable_size()
+- Removed EXPERIMENTAL from most methods in memkind API
+- Added MEMKIND_ERROR_ARENAS_CREATE code
+- Added C++ allocator for PMEM kind
+- Extended PMEM examples
+- Fixed integration with Travis CI
+- Extended Travis CI configuration with Astyle and Coverage
+- Added PMEM kind tests
+
+* Wed Jan 24 2018 Pawel Kochanek <pawel.b.kochanek@intel.com> v1.7.0
+- Updated internally used jemalloc to upstream version 5.0.
+- Fixed error that has been occurring while memkind was dynamically loaded.
+- Fixed MEMKIND_HBW_NODES behavior for single NUMA node system.
+- Removed licenses other than BSD 3-clause from COPYING.
+- Changed build instructions.
+- Added configurable jemalloc prefix in build scripts.
+- Upgraded gtest to version 1.8.0.
+- Added memory footprint tests.
+- Added locality test for MEMKIND_PREFERRED.
+- Applied test parametrization in BATests.
+- Fixed problems with pytest testing in Travis.
+- Added huge page configuration in several tests.
+- Removed several symbols that are no longer exposed in public API from man pages.
+- Fixed HBW_POLICY_BIND_ALL documentation.
+
+* Tue Jun 13 2017 Artur Koziej <artur.koziej@intel.com> v1.6.0
+- Introduced new policy HBW_POLICY_BIND_ALL.
+- Introduced MEMKIND_HBW_ALL, MEMKIND_HBW_ALL_HUGETLB and MEMKIND_REGULAR kinds.
+- Fixed hbw_posix_memalign_psize() return value in case of unsupported policy and
+  page size combination.
+- Documented supported policy and page size combination for
+  hbw_posix_memalign_psize().
+- Merged libmemkind.a and libjemalloc_pic.a into a single static library.
+- Fixed OOM during tcache acquisition.
+- Tuned huge pages configuration in memkind tests.
+- Added stress tests for allocations with large random sizes.
+- Added stress tests for allocations with random kinds.
+- Extended policy and NUMA node binding tests.
+- Fixed several Travis build problems.
+- Removed gb_page_tests_preferred_policy tests.
+
+* Tue Feb 21 2017 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.5.0
+- Intel(R) TBB scalable_allocator can be used as heap management alternative
+  to jemalloc (requires installed Intel (R) TBB package and environment variable
+  MEMKIND_HEAP_MANAGER=TBB)
+- Updated internally used jemalloc to upstream version 4.3.1 (previously modified 3.5.1)
+- Introduced new environment option MEMKIND_HOG_MEMORY. Setting it to "1" will prevent
+  memkind from releasing memory to OS.
+- Introduced support for KNM processor from Intel(R) Xeon Phi(TM) family
+- General cleanup of build scripts (e.g. fixed issues found on Ubuntu,
+  introduced simple build.sh script)
+- Deprecated support for gigabyte-pages
+- Removed number of previously deprecated symbols
+
+* Thu Nov 17 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.4.0
+- Introduced hbw_verify_memory_region() for verifying if provided memory
+  region fully falls into high-bandwidth memory (for details please take a look
+  at man/hbwmalloc.3).
+- New API for creating kinds using set of predefined attributes:
+  memkind_create_kind(), memkind_destroy_kind() (for details please take a look
+  at man/memkind.3).
+- Fixed issue in autohbw where it was reporting limit error regardless of
+  value provided.
+- Added logging on error paths.
+- Test base fixes and improvements.
+
+* Tue Sep 27 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.3.0
+- Introduced logging mechanism (for details please take a look at
+  man/memkind.3).
+- Deprecated symbols: memkind_finalize(), memkind_get_num_kind(),
+  memkind_get_kind_by_partition(), memkind_get_kind_by_name(),
+  memkind_partition_mmap(), memkind_get_size(), MEMKIND_ERROR_MEMALIGN,
+  MEMKIND_ERROR_MALLCTL,  MEMKIND_ERROR_GETCPU, MEMKIND_ERROR_PMTT,
+  MEMKIND_ERROR_TIEDISTANCE, MEMKIND_ERROR_ALIGNMENT, MEMKIND_ERROR_MALLOCX,
+  MEMKIND_ERROR_REPNAME, MEMKIND_ERROR_PTHREAD, MEMKIND_ERROR_BADPOLICY,
+  MEMKIND_ERROR_REPPOLICY.
+- Added integration with hwloc (turned on with --with-hwloc).
+- Cleanup of symbols exposed by libmemkind.so (e.g. no longer exposing libnuma
+  and jemalloc symbols).
+- AutoHBW files have been moved to "memkind/autohbw" directory, code has
+  been refactored and tests have been added to appropriate scenarios.
+- Library is now built with flags improving security (can be turned off
+  with --disable-secure configure time option).
+- Changed configuration of jemalloc to turn off unused features.
+
+* Thu Aug 18 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.2.0
+- memkind_create() and memkind_ops have been deprecated (moved to
+  memkind_deprecated.h).
+- Deprecated the headers from memkind/internal and added compile-time
+  warnings to them (plan is to remove those from rpm packages).
+- autohbw files have been moved to memkind/autohbw directory.
+- API decorators (memkind_malloc_pre, memkind_malloc_post etc.) now
+  need to be enabled with configure-time option --enable-decorators.
+- Allocation time optimizations with up to 20% improvement.
+
+* Wed Jul 13 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.1.1
+- Performance improvement for memkind_free() in scenario where NULL
+  was passed as kind (reduced by 60%).
+- Introduced integration with Travis CI.
+- Fixed issue where memory returned from calloc was not filled with zeroes
+  when using memkind_pmem kinds.
+- Fixed issue where interleave kinds was failing on systems without
+  Transparent Huge Pages module configured.
+- Resolved several issues that was causing compilation errors on some systems.
+- test/test.sh is now able to detect system configuration, and run only those
+  tests which requirements are meet.
+- Added gtest to repo to avoid downloading it during build proecess.
+
+* Mon May 23 2016 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v1.1.0
+- Implemented algorithm for detecting high-bandwidth memory on Intel Xeon Phi X200
+  without parsing PMTT.
+- PMTT-parsing service has been removed.
+- memkind struct definition from memkind.h has been replaced by forward
+  declaration. Definition has been moved to /internal/memkind_private.h and
+  shouldn't be considered as part of library interface anymore.
+- Performance improvement for hugetlb kinds (check_available() is no longer
+  parsing sysfs every time).
+- Performance fix by removing --enable-safe from jemmaloc build flags.
+- New tests for validating high-bandwidth memory detection has been added.
+- AFTS has been split into two groups: memkind-afts.ts and memkind-afts-ext.ts,
+  based on amount of required memory (for details view test/README).
+
+* Thu Mar 03 2016 Artur Koziej <artur.koziej@intel.com> v1.0.0
+- hbwmalloc.h released as stable API.
+- Introduce API standards: STANDARD API, NON-STANDARD API, EXPERIMENTAL API.
+- Introduce memkind versioning API.
+- Fix type names in hbwmalloc API.
+- Change error codes to POSIX standard ones in hbwmalloc API.
+- Move API headers to include directory.
+- Move NON-STANDARD API, or EXPERIMENTAL API to include/memkind/internal directory.
+- New man page (memkind_interleave.3) describing interleave kind.
+- Significant documentation improvement.
+- Performance fix for jemalloc - significantly decrease of library initialization time.
+- Enforce 2MB alignment in jemalloc due to the Linux kernel bug (munmap() fails for
+  huge pages, when the size is not aligned).
+- Move to systemd defined service (memkind.service), drop init.d script.
+- New rpm layout (memkind and memkind-devel).
+- Remove false dependency on OpenMP.
+- Extend test base for stress, longevity and initialization performance tests.
+- Fix memory leak in ctl_growk().
+- Rename namespace: hbwmalloc -> hbw.
+- Rename class: hbwmalloc_allocator -> allocator.
+
+* Thu Oct 15 2015 Krzysztof Kulakowski <krzysztof.kulakowski@intel.com> v0.3.0
+- Added two new kinds: MEMKIND_INTERLEAVE and MEMKIND_HBW_INTERLEAVE
+  (for details please take a look at man/memkind.3)
+- Added support for file-backed memory heaps
+  (for details please take a look at man/memkind_pmem.3)
+- Added autohbw library, that intercepts the standard heap allocations,
+  which let use high-bandwidth memory without modifying existing codebase
+  (for details please take a look at examples/autohbw_README)
+- jemalloc is now, statically linked, internal component of memkind RPM
+  instead of rpm dependency.
+- Added memkind_allocated example which demonstrates usage of memkind
+  with alignas() alignment specifier introduced in C++11
+  (for details please take a look at examples/README)
+- Added support for using thread local storage to improve performance in
+  multithreaded applications (enabled by configure time option: --enable-tls)
+- Added possibility to set number of jemalloc arenas per kind by
+  configure time parameter or environmental variable
+  (for details please take a look at MEMKIND_ARENA_NUM_PER_KIND section
+  of man/memkind.3).
+- Added decorators to the memkind allocation APIs.
+  These are weak symbols (pre and post for each API) which can modify
+  the input and output of each of the calls.
+- Significantly extended test base (new groups of tests: performance,
+  multithreaded, PMTT, nagtive, stress).
+
+* Fri Jan 9 2015 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.2.0
+- Bumped memkind ABI version to 0:1:0.
+- Removed memkind_get_kind_for_free() from externally facing API's.  Instead if
+  memkind_free() is called with zero passed for the kind then
+  memkind_get_kind_for_free() is called internally.
+- Moved to single callback memkind_partition_mmap() to simplify
+  jemalloc modifications.
+- Added hooks for setting file descriptor and offset for mmap enabling
+  file-backed memory kinds.
+- Added a void pointer called "priv" to memkind structure for storing data for
+  user-defined kinds.
+- Removed call to sched_getcpu(), now thread id is hashed to determine the arena
+  index.
+- Added weak symbol hooks for decorating the heap management functions.
+- Fixed several issues with init.d/memkind script and spec file scriptlets that
+  are exposed by SLES-12.
+- Introduced an example library called numakind.
+  The numakind library will allocate from the closest NUMA node to a thread as
+  measured when that thread makes its first allocation call.
+- Fixed error handling in memkind_gbtlb_mmap() that could cause a segfault when
+  gigabyte pages are not available.
+- Added tests for PMTT parser.
+- Removed binary mock PMTT table from source code, replaced it with a hexdump.
+- Fixed a number of issues in test scripts which were suppressing errors.
+- Removed unnecessary includes from header files.
+- Better error checking in example code.
+- Documentation update and clean up.
+
+* Thu Nov 13 2014 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.1.0
+- Increased test code coverage significantly.
+- Fixed bug in memkind_error_message() for MEMKIND_ERROR_TOOMANY.
+- Removed memkind_arena_free() API since it was redundant with memkind_default_free().
+- Static memkind structs are now declared as extern in the headers and defined in the
+  source files files rather than being statically defined in the headers.
+
+* Thu Oct 30 2014 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.0.9
+- Now building with autotools.
+- Updated documentation.
+- Fixed typo in copyright.
+- Fixed test scripts to properly handle return code of each test.
+- Added C++03 standard allocator that uses hbw_malloc and hbw_free.
+* Tue Sep 30 2014 Christopher Cantalupo <christopher.m.cantalupo@intel.com> v0.0.8
+- Added GBTLB functionality, code clean up, documentation updates,
+  examples directory.  Examples includes stream modified to use
+  memkind interface.  Code coverage still lacking, and documentation
+  incomplete.
diff --git a/memkind-1.10.0/MANIFEST b/memkind-1.10.0/MANIFEST
new file mode 100644
index 0000000..f5aa3e4
--- /dev/null
+++ b/memkind-1.10.0/MANIFEST
@@ -0,0 +1,552 @@
+.codecov.yml
+.travis.yml
+AUTHORS
+CONTRIBUTING
+COPYING
+ChangeLog
+Makefile.am
+NEWS
+PULL_REQUEST_TEMPLATE.md
+README
+astyle.sh
+autogen.sh
+autohbw/Makefile.mk
+autohbw/autohbw.c
+autohbw/autohbw_README
+autohbw/autohbw_get_src_lines.pl
+autohbw/autohbw_test.sh
+build.sh
+configure.ac
+examples/Makefile.mk
+examples/README
+examples/autohbw_candidates.c
+examples/filter_example.c
+examples/hello_hbw_example.c
+examples/hello_memkind_example.c
+examples/memkind_allocated.hpp
+examples/memkind_allocated_example.cpp
+examples/memkind_cpp_allocator.cpp
+examples/memkind_decorator_debug.c
+examples/memkind_get_stat.c
+examples/pmem_alignment.c
+examples/pmem_and_default_kind.c
+examples/pmem_and_dax_kmem.c
+examples/pmem_config.c
+examples/pmem_cpp_allocator.cpp
+examples/pmem_detect_kind.c
+examples/pmem_free_with_unknown_kind.c
+examples/pmem_kinds.c
+examples/pmem_malloc.c
+examples/pmem_malloc_unlimited.c
+examples/pmem_multithreads.c
+examples/pmem_multithreads_onekind.c
+examples/pmem_usable_size.c
+include/hbw_allocator.h
+include/hbwmalloc.h
+include/memkind.h
+include/memkind/internal/heap_manager.h
+include/memkind/internal/memkind_arena.h
+include/memkind/internal/memkind_bandwidth.h
+include/memkind/internal/memkind_dax_kmem.h
+include/memkind/internal/memkind_default.h
+include/memkind/internal/memkind_gbtlb.h
+include/memkind/internal/memkind_hbw.h
+include/memkind/internal/memkind_hugetlb.h
+include/memkind/internal/memkind_interleave.h
+include/memkind/internal/memkind_log.h
+include/memkind/internal/memkind_pmem.h
+include/memkind/internal/memkind_private.h
+include/memkind/internal/memkind_regular.h
+include/memkind/internal/tbb_mem_pool_policy.h
+include/memkind/internal/tbb_wrapper.h
+include/memkind/internal/vec.h
+include/memkind_allocator.h
+include/memkind_deprecated.h
+include/pmem_allocator.h
+install_astyle.sh
+jemalloc/.appveyor.yml
+jemalloc/.autom4te.cfg
+jemalloc/.gitattributes
+jemalloc/.gitignore
+jemalloc/.travis.yml
+jemalloc/autogen.sh
+jemalloc/bin/jemalloc-config.in
+jemalloc/bin/jemalloc.sh.in
+jemalloc/bin/jeprof.in
+jemalloc/build-aux/config.guess
+jemalloc/build-aux/config.sub
+jemalloc/build-aux/install-sh
+jemalloc/ChangeLog
+jemalloc/config.stamp.in
+jemalloc/configure.ac
+jemalloc/COPYING
+jemalloc/doc/html.xsl.in
+jemalloc/doc/jemalloc.xml.in
+jemalloc/doc/manpages.xsl.in
+jemalloc/doc/stylesheet.xsl
+jemalloc/include/jemalloc/internal/arena_externs.h
+jemalloc/include/jemalloc/internal/arena_inlines_a.h
+jemalloc/include/jemalloc/internal/arena_inlines_b.h
+jemalloc/include/jemalloc/internal/arena_stats.h
+jemalloc/include/jemalloc/internal/arena_structs_a.h
+jemalloc/include/jemalloc/internal/arena_structs_b.h
+jemalloc/include/jemalloc/internal/arena_types.h
+jemalloc/include/jemalloc/internal/assert.h
+jemalloc/include/jemalloc/internal/atomic.h
+jemalloc/include/jemalloc/internal/atomic_c11.h
+jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
+jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
+jemalloc/include/jemalloc/internal/atomic_msvc.h
+jemalloc/include/jemalloc/internal/background_thread_externs.h
+jemalloc/include/jemalloc/internal/background_thread_inlines.h
+jemalloc/include/jemalloc/internal/background_thread_structs.h
+jemalloc/include/jemalloc/internal/base_externs.h
+jemalloc/include/jemalloc/internal/base_inlines.h
+jemalloc/include/jemalloc/internal/base_structs.h
+jemalloc/include/jemalloc/internal/base_types.h
+jemalloc/include/jemalloc/internal/bin.h
+jemalloc/include/jemalloc/internal/bin_stats.h
+jemalloc/include/jemalloc/internal/bin_types.h
+jemalloc/include/jemalloc/internal/bitmap.h
+jemalloc/include/jemalloc/internal/bit_util.h
+jemalloc/include/jemalloc/internal/cache_bin.h
+jemalloc/include/jemalloc/internal/ckh.h
+jemalloc/include/jemalloc/internal/ctl.h
+jemalloc/include/jemalloc/internal/div.h
+jemalloc/include/jemalloc/internal/emitter.h
+jemalloc/include/jemalloc/internal/extent_dss.h
+jemalloc/include/jemalloc/internal/extent_externs.h
+jemalloc/include/jemalloc/internal/extent_inlines.h
+jemalloc/include/jemalloc/internal/extent_mmap.h
+jemalloc/include/jemalloc/internal/extent_structs.h
+jemalloc/include/jemalloc/internal/extent_types.h
+jemalloc/include/jemalloc/internal/hash.h
+jemalloc/include/jemalloc/internal/hook.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
+jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
+jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
+jemalloc/include/jemalloc/internal/large_externs.h
+jemalloc/include/jemalloc/internal/log.h
+jemalloc/include/jemalloc/internal/malloc_io.h
+jemalloc/include/jemalloc/internal/mutex.h
+jemalloc/include/jemalloc/internal/mutex_pool.h
+jemalloc/include/jemalloc/internal/mutex_prof.h
+jemalloc/include/jemalloc/internal/nstime.h
+jemalloc/include/jemalloc/internal/pages.h
+jemalloc/include/jemalloc/internal/ph.h
+jemalloc/include/jemalloc/internal/private_namespace.sh
+jemalloc/include/jemalloc/internal/private_symbols.sh
+jemalloc/include/jemalloc/internal/prng.h
+jemalloc/include/jemalloc/internal/prof_externs.h
+jemalloc/include/jemalloc/internal/prof_inlines_a.h
+jemalloc/include/jemalloc/internal/prof_inlines_b.h
+jemalloc/include/jemalloc/internal/prof_structs.h
+jemalloc/include/jemalloc/internal/prof_types.h
+jemalloc/include/jemalloc/internal/public_namespace.sh
+jemalloc/include/jemalloc/internal/public_unnamespace.sh
+jemalloc/include/jemalloc/internal/ql.h
+jemalloc/include/jemalloc/internal/qr.h
+jemalloc/include/jemalloc/internal/quantum.h
+jemalloc/include/jemalloc/internal/rb.h
+jemalloc/include/jemalloc/internal/rtree.h
+jemalloc/include/jemalloc/internal/rtree_tsd.h
+jemalloc/include/jemalloc/internal/safety_check.h
+jemalloc/include/jemalloc/internal/sc.h
+jemalloc/include/jemalloc/internal/seq.h
+jemalloc/include/jemalloc/internal/smoothstep.h
+jemalloc/include/jemalloc/internal/smoothstep.sh
+jemalloc/include/jemalloc/internal/spin.h
+jemalloc/include/jemalloc/internal/stats.h
+jemalloc/include/jemalloc/internal/sz.h
+jemalloc/include/jemalloc/internal/tcache_externs.h
+jemalloc/include/jemalloc/internal/tcache_inlines.h
+jemalloc/include/jemalloc/internal/tcache_structs.h
+jemalloc/include/jemalloc/internal/tcache_types.h
+jemalloc/include/jemalloc/internal/test_hooks.h
+jemalloc/include/jemalloc/internal/ticker.h
+jemalloc/include/jemalloc/internal/tsd.h
+jemalloc/include/jemalloc/internal/tsd_generic.h
+jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
+jemalloc/include/jemalloc/internal/tsd_tls.h
+jemalloc/include/jemalloc/internal/tsd_types.h
+jemalloc/include/jemalloc/internal/tsd_win.h
+jemalloc/include/jemalloc/internal/util.h
+jemalloc/include/jemalloc/internal/witness.h
+jemalloc/include/jemalloc/jemalloc.sh
+jemalloc/include/jemalloc/jemalloc_defs.h.in
+jemalloc/include/jemalloc/jemalloc_macros.h.in
+jemalloc/include/jemalloc/jemalloc_mangle.sh
+jemalloc/include/jemalloc/jemalloc_protos.h.in
+jemalloc/include/jemalloc/jemalloc_rename.sh
+jemalloc/include/jemalloc/jemalloc_typedefs.h.in
+jemalloc/include/msvc_compat/C99/stdbool.h
+jemalloc/include/msvc_compat/C99/stdint.h
+jemalloc/include/msvc_compat/strings.h
+jemalloc/include/msvc_compat/windows_extra.h
+jemalloc/INSTALL.md
+jemalloc/jemalloc.pc.in
+jemalloc/m4/ax_cxx_compile_stdcxx.m4
+jemalloc/Makefile.in
+jemalloc/msvc/jemalloc_vc2015.sln
+jemalloc/msvc/jemalloc_vc2017.sln
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/ReadMe.txt
+jemalloc/msvc/test_threads/test_threads.cpp
+jemalloc/msvc/test_threads/test_threads.h
+jemalloc/msvc/test_threads/test_threads_main.cpp
+jemalloc/README
+jemalloc/run_tests.sh
+jemalloc/scripts/gen_run_tests.py
+jemalloc/scripts/gen_travis.py
+jemalloc/src/arena.c
+jemalloc/src/background_thread.c
+jemalloc/src/base.c
+jemalloc/src/bin.c
+jemalloc/src/bitmap.c
+jemalloc/src/ckh.c
+jemalloc/src/ctl.c
+jemalloc/src/div.c
+jemalloc/src/extent.c
+jemalloc/src/extent_dss.c
+jemalloc/src/extent_mmap.c
+jemalloc/src/hash.c
+jemalloc/src/hook.c
+jemalloc/src/jemalloc.c
+jemalloc/src/jemalloc_cpp.cpp
+jemalloc/src/large.c
+jemalloc/src/log.c
+jemalloc/src/malloc_io.c
+jemalloc/src/mutex.c
+jemalloc/src/mutex_pool.c
+jemalloc/src/nstime.c
+jemalloc/src/pages.c
+jemalloc/src/prng.c
+jemalloc/src/prof.c
+jemalloc/src/rtree.c
+jemalloc/src/safety_check.c
+jemalloc/src/sc.c
+jemalloc/src/stats.c
+jemalloc/src/sz.c
+jemalloc/src/tcache.c
+jemalloc/src/test_hooks.c
+jemalloc/src/ticker.c
+jemalloc/src/tsd.c
+jemalloc/src/witness.c
+jemalloc/src/zone.c
+jemalloc/test/include/test/btalloc.h
+jemalloc/test/include/test/extent_hooks.h
+jemalloc/test/include/test/jemalloc_test.h.in
+jemalloc/test/include/test/jemalloc_test_defs.h.in
+jemalloc/test/include/test/math.h
+jemalloc/test/include/test/mq.h
+jemalloc/test/include/test/mtx.h
+jemalloc/test/include/test/SFMT-alti.h
+jemalloc/test/include/test/SFMT-params.h
+jemalloc/test/include/test/SFMT-params607.h
+jemalloc/test/include/test/SFMT-params1279.h
+jemalloc/test/include/test/SFMT-params2281.h
+jemalloc/test/include/test/SFMT-params4253.h
+jemalloc/test/include/test/SFMT-params11213.h
+jemalloc/test/include/test/SFMT-params19937.h
+jemalloc/test/include/test/SFMT-params44497.h
+jemalloc/test/include/test/SFMT-params86243.h
+jemalloc/test/include/test/SFMT-params132049.h
+jemalloc/test/include/test/SFMT-params216091.h
+jemalloc/test/include/test/SFMT-sse2.h
+jemalloc/test/include/test/SFMT.h
+jemalloc/test/include/test/test.h
+jemalloc/test/include/test/thd.h
+jemalloc/test/include/test/timer.h
+jemalloc/test/integration/aligned_alloc.c
+jemalloc/test/integration/allocated.c
+jemalloc/test/integration/extent.c
+jemalloc/test/integration/extent.sh
+jemalloc/test/integration/malloc.c
+jemalloc/test/integration/mallocx.c
+jemalloc/test/integration/mallocx.sh
+jemalloc/test/integration/MALLOCX_ARENA.c
+jemalloc/test/integration/overflow.c
+jemalloc/test/integration/posix_memalign.c
+jemalloc/test/integration/rallocx.c
+jemalloc/test/integration/sdallocx.c
+jemalloc/test/integration/slab_sizes.c
+jemalloc/test/integration/slab_sizes.sh
+jemalloc/test/integration/smallocx.c
+jemalloc/test/integration/smallocx.sh
+jemalloc/test/integration/thread_arena.c
+jemalloc/test/integration/thread_tcache_enabled.c
+jemalloc/test/integration/xallocx.c
+jemalloc/test/integration/xallocx.sh
+jemalloc/test/src/btalloc.c
+jemalloc/test/src/btalloc_0.c
+jemalloc/test/src/btalloc_1.c
+jemalloc/test/src/math.c
+jemalloc/test/src/mq.c
+jemalloc/test/src/mtx.c
+jemalloc/test/src/SFMT.c
+jemalloc/test/src/test.c
+jemalloc/test/src/thd.c
+jemalloc/test/src/timer.c
+jemalloc/test/stress/hookbench.c
+jemalloc/test/stress/microbench.c
+jemalloc/test/test.sh.in
+jemalloc/test/unit/a0.c
+jemalloc/test/unit/arena_reset.c
+jemalloc/test/unit/arena_reset_prof.c
+jemalloc/test/unit/arena_reset_prof.sh
+jemalloc/test/unit/atomic.c
+jemalloc/test/unit/background_thread.c
+jemalloc/test/unit/background_thread_enable.c
+jemalloc/test/unit/base.c
+jemalloc/test/unit/binshard.c
+jemalloc/test/unit/binshard.sh
+jemalloc/test/unit/bitmap.c
+jemalloc/test/unit/bit_util.c
+jemalloc/test/unit/ckh.c
+jemalloc/test/unit/decay.c
+jemalloc/test/unit/decay.sh
+jemalloc/test/unit/div.c
+jemalloc/test/unit/emitter.c
+jemalloc/test/unit/extent_quantize.c
+jemalloc/test/unit/extent_util.c
+jemalloc/test/unit/fork.c
+jemalloc/test/unit/hash.c
+jemalloc/test/unit/hook.c
+jemalloc/test/unit/huge.c
+jemalloc/test/unit/junk.c
+jemalloc/test/unit/junk.sh
+jemalloc/test/unit/junk_alloc.c
+jemalloc/test/unit/junk_alloc.sh
+jemalloc/test/unit/junk_free.c
+jemalloc/test/unit/junk_free.sh
+jemalloc/test/unit/log.c
+jemalloc/test/unit/mallctl.c
+jemalloc/test/unit/malloc_io.c
+jemalloc/test/unit/math.c
+jemalloc/test/unit/mq.c
+jemalloc/test/unit/mtx.c
+jemalloc/test/unit/nstime.c
+jemalloc/test/unit/pack.c
+jemalloc/test/unit/pack.sh
+jemalloc/test/unit/pages.c
+jemalloc/test/unit/ph.c
+jemalloc/test/unit/prng.c
+jemalloc/test/unit/prof_accum.c
+jemalloc/test/unit/prof_accum.sh
+jemalloc/test/unit/prof_active.c
+jemalloc/test/unit/prof_active.sh
+jemalloc/test/unit/prof_gdump.c
+jemalloc/test/unit/prof_gdump.sh
+jemalloc/test/unit/prof_idump.c
+jemalloc/test/unit/prof_idump.sh
+jemalloc/test/unit/prof_log.c
+jemalloc/test/unit/prof_log.sh
+jemalloc/test/unit/prof_reset.c
+jemalloc/test/unit/prof_reset.sh
+jemalloc/test/unit/prof_tctx.c
+jemalloc/test/unit/prof_tctx.sh
+jemalloc/test/unit/prof_thread_name.c
+jemalloc/test/unit/prof_thread_name.sh
+jemalloc/test/unit/ql.c
+jemalloc/test/unit/qr.c
+jemalloc/test/unit/rb.c
+jemalloc/test/unit/retained.c
+jemalloc/test/unit/rtree.c
+jemalloc/test/unit/safety_check.c
+jemalloc/test/unit/safety_check.sh
+jemalloc/test/unit/sc.c
+jemalloc/test/unit/seq.c
+jemalloc/test/unit/SFMT.c
+jemalloc/test/unit/size_classes.c
+jemalloc/test/unit/slab.c
+jemalloc/test/unit/smoothstep.c
+jemalloc/test/unit/spin.c
+jemalloc/test/unit/stats.c
+jemalloc/test/unit/stats_print.c
+jemalloc/test/unit/test_hooks.c
+jemalloc/test/unit/ticker.c
+jemalloc/test/unit/tsd.c
+jemalloc/test/unit/witness.c
+jemalloc/test/unit/zero.c
+jemalloc/test/unit/zero.sh
+jemalloc/TUNING.md
+m4/ax_cxx_compile_stdcxx.m4
+m4/ax_cxx_compile_stdcxx_11.m4
+m4/ax_pthread.m4
+man/autohbw.7
+man/hbwallocator.3
+man/hbwmalloc.3
+man/memkind-auto-dax-kmem-nodes.1
+man/memkind-hbw-nodes.1
+man/memkind.3
+man/memkind_arena.3
+man/memkind_dax_kmem.3
+man/memkind_default.3
+man/memkind_hbw.3
+man/memkind_hugetlb.3
+man/memkind_pmem.3
+man/memkindallocator.3
+man/pmemallocator.3
+memkind.pc.in
+memkind.spec.mk
+src/Makefile.mk
+src/hbwmalloc.c
+src/heap_manager.c
+man/memkind-auto-dax-kmem-nodes.c
+src/memkind-hbw-nodes.c
+src/memkind.c
+src/memkind_arena.c
+src/memkind_bandwidth.c
+src/memkind_dax_kmem.c
+src/memkind_default.c
+src/memkind_gbtlb.c
+src/memkind_hbw.c
+src/memkind_hugetlb.c
+src/memkind_interleave.c
+src/memkind_log.c
+src/memkind_pmem.c
+src/memkind_regular.c
+src/tbb_wrapper.c
+test/Allocator.hpp
+test/Makefile.mk
+test/TestPolicy.hpp
+test/alloc_benchmark.c
+test/alloc_performance_tests.cpp
+test/allocate_to_max_stress_test.cpp
+test/allocator_perf_tool/AllocationSizes.hpp
+test/allocator_perf_tool/Allocation_info.cpp
+test/allocator_perf_tool/Allocation_info.hpp
+test/allocator_perf_tool/Allocator.hpp
+test/allocator_perf_tool/AllocatorFactory.hpp
+test/allocator_perf_tool/CSVLogger.hpp
+test/allocator_perf_tool/CommandLine.hpp
+test/allocator_perf_tool/Configuration.hpp
+test/allocator_perf_tool/ConsoleLog.hpp
+test/allocator_perf_tool/FunctionCalls.hpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.h
+test/allocator_perf_tool/GTestAdapter.hpp
+test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/HugePageOrganizer.hpp
+test/allocator_perf_tool/HugePageUnmap.hpp
+test/allocator_perf_tool/Iterator.hpp
+test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/Makefile
+test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
+test/allocator_perf_tool/Numastat.hpp
+test/allocator_perf_tool/PmemMockup.cpp
+test/allocator_perf_tool/PmemMockup.hpp
+test/allocator_perf_tool/Runnable.hpp
+test/allocator_perf_tool/ScenarioWorkload.cpp
+test/allocator_perf_tool/ScenarioWorkload.h
+test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
+test/allocator_perf_tool/Stats.hpp
+test/allocator_perf_tool/StressIncreaseToMax.cpp
+test/allocator_perf_tool/StressIncreaseToMax.h
+test/allocator_perf_tool/Task.hpp
+test/allocator_perf_tool/TaskFactory.hpp
+test/allocator_perf_tool/Tests.hpp
+test/allocator_perf_tool/Thread.hpp
+test/allocator_perf_tool/TimerSysTime.hpp
+test/allocator_perf_tool/VectorIterator.hpp
+test/allocator_perf_tool/Workload.hpp
+test/allocator_perf_tool/WrappersMacros.hpp
+test/allocator_perf_tool/main.cpp
+test/autohbw_test.py
+test/autohbw_test_helper.c
+test/bat_tests.cpp
+test/check.cpp
+test/check.h
+test/common.h
+test/dax_kmem_nodes.cpp
+test/dax_kmem_nodes.h
+test/dax_kmem_env_var_test.py
+test/decorator_test.cpp
+test/decorator_test.h
+test/dlopen_test.cpp
+test/draw_plots.py
+test/environ_err_dax_kmem_malloc_positive_test.cpp
+test/environ_err_dax_kmem_malloc_test.cpp
+test/environ_err_hbw_malloc_test.cpp
+test/error_message_tests.cpp
+test/fragmentation_benchmark_pmem.cpp
+test/freeing_memory_segfault_test.cpp
+test/gb_page_tests_bind_policy.cpp
+test/get_arena_test.cpp
+test/gtest_fused/gtest/gtest-all.cc
+test/gtest_fused/gtest/gtest.h
+test/hbw_allocator_performance_tests.cpp
+test/hbw_allocator_tests.cpp
+test/hbw_detection_test.py
+test/hbw_verify_function_test.cpp
+test/heap_manager_init_perf_test.cpp
+test/huge_page_test.cpp
+test/load_tbbmalloc_symbols.c
+test/locality_test.cpp
+test/main.cpp
+test/memkind_allocator_tests.cpp
+test/memkind-afts-ext.ts
+test/memkind-afts.ts
+test/memkind-perf-ext.ts
+test/memkind-perf.ts
+test/memkind-pytests.ts
+test/memkind-slts.ts
+test/memkind_dax_kmem_test.cpp
+test/memkind_detect_kind_tests.cpp
+test/memkind_null_kind_test.cpp
+test/memkind_pmem_config_tests.cpp
+test/memkind_pmem_long_time_tests.cpp
+test/memkind_pmem_tests.cpp
+test/memkind_stat_test.cpp
+test/memkind_defrag_reallocate.cpp
+test/memkind_versioning_tests.cpp
+test/memory_footprint_test.cpp
+test/memory_manager.h
+test/multithreaded_tests.cpp
+test/negative_tests.cpp
+test/performance/framework.cpp
+test/performance/framework.hpp
+test/performance/operations.hpp
+test/performance/perf_tests.cpp
+test/performance/perf_tests.hpp
+test/pmem_alloc_performance_tests.cpp
+test/pmem_allocator_tests.cpp
+test/proc_stat.h
+test/python_framework/__init__.py
+test/python_framework/cmd_helper.py
+test/python_framework/huge_page_organizer.py
+test/random_sizes_allocator.h
+test/run_alloc_benchmark.sh
+test/static_kinds_list.h
+test/static_kinds_tests.cpp
+test/tbbmalloc.h
+test/test.sh
+test/test_dax_kmem.sh
+test/trace_mechanism_test.py
+test/trace_mechanism_test_helper.c
+test/trial_generator.cpp
+test/trial_generator.h
+utils/docker/Dockerfile.fedora-31
+utils/docker/Dockerfile.ubuntu-18.04
+utils/docker/README.md
+utils/docker/docker_install_ndctl.sh
+utils/docker/docker_install_tbb.sh
+utils/docker/docker_run_build.sh
+utils/docker/docker_run_coverage.sh
+utils/docker/docker_run_test.sh
+utils/docker/run_local.sh
+utils/docker/set_host_configuration.sh
diff --git a/memkind-1.10.0/Makefile.am b/memkind-1.10.0/Makefile.am
new file mode 100644
index 0000000..d3b2039
--- /dev/null
+++ b/memkind-1.10.0/Makefile.am
@@ -0,0 +1,526 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+ACLOCAL_AMFLAGS = -I m4
+AM_CPPFLAGS = -I$(top_srcdir)/include -I$(top_srcdir)/jemalloc/include
+MAN2HTML = groff -mandoc -Thtml
+
+lib_LTLIBRARIES = libmemkind.la
+
+libmemkind_la_SOURCES = src/hbwmalloc.c \
+                        src/heap_manager.c \
+                        src/memkind.c \
+                        src/memkind_arena.c \
+                        src/memkind_bandwidth.c \
+                        src/memkind_dax_kmem.c \
+                        src/memkind_default.c \
+                        src/memkind_gbtlb.c \
+                        src/memkind_hbw.c \
+                        src/memkind_hugetlb.c \
+                        src/memkind_interleave.c \
+                        src/memkind_log.c \
+                        src/memkind_pmem.c \
+                        src/memkind_regular.c \
+                        src/tbb_wrapper.c \
+                        # end
+
+
+libmemkind_la_LIBADD = jemalloc/lib/libjemalloc_pic.a
+libmemkind_la_LDFLAGS = -version-info 0:1:0 -ldl -lrt
+include_HEADERS = include/hbw_allocator.h \
+                  include/hbwmalloc.h \
+                  include/memkind.h \
+                  include/memkind_allocator.h \
+                  include/memkind_deprecated.h \
+                  include/pmem_allocator.h \
+                  # end
+
+noinst_HEADERS =  include/memkind/internal/heap_manager.h \
+                  include/memkind/internal/memkind_arena.h \
+                  include/memkind/internal/memkind_bandwidth.h \
+                  include/memkind/internal/memkind_dax_kmem.h \
+                  include/memkind/internal/memkind_default.h \
+                  include/memkind/internal/memkind_gbtlb.h \
+                  include/memkind/internal/memkind_hbw.h \
+                  include/memkind/internal/memkind_hugetlb.h \
+                  include/memkind/internal/memkind_interleave.h \
+                  include/memkind/internal/memkind_log.h \
+                  include/memkind/internal/memkind_pmem.h \
+                  include/memkind/internal/memkind_private.h \
+                  include/memkind/internal/memkind_regular.h \
+                  include/memkind/internal/tbb_mem_pool_policy.h \
+                  include/memkind/internal/tbb_wrapper.h \
+                  include/memkind/internal/vec.h \
+                  # end
+
+EXTRA_DIST = CONTRIBUTING \
+             VERSION \
+             astyle.sh \
+             autogen.sh \
+             build.sh \
+             examples/README \
+             jemalloc/.appveyor.yml \
+             jemalloc/.autom4te.cfg \
+             jemalloc/.gitattributes\
+             jemalloc/.gitignore \
+             jemalloc/.travis.yml \
+             jemalloc/autogen.sh \
+             jemalloc/bin/jemalloc-config.in \
+             jemalloc/bin/jemalloc.sh.in \
+             jemalloc/bin/jeprof.in \
+             jemalloc/build-aux/config.guess \
+             jemalloc/build-aux/config.sub \
+             jemalloc/build-aux/install-sh \
+             jemalloc/ChangeLog \
+             jemalloc/config.stamp.in \
+             jemalloc/configure.ac \
+             jemalloc/COPYING \
+             jemalloc/doc/html.xsl.in \
+             jemalloc/doc/jemalloc.xml.in \
+             jemalloc/doc/manpages.xsl.in \
+             jemalloc/doc/stylesheet.xsl \
+             jemalloc/include/jemalloc/internal/arena_externs.h \
+             jemalloc/include/jemalloc/internal/arena_inlines_a.h \
+             jemalloc/include/jemalloc/internal/arena_inlines_b.h \
+             jemalloc/include/jemalloc/internal/arena_stats.h \
+             jemalloc/include/jemalloc/internal/arena_structs_a.h \
+             jemalloc/include/jemalloc/internal/arena_structs_b.h \
+             jemalloc/include/jemalloc/internal/arena_types.h \
+             jemalloc/include/jemalloc/internal/assert.h \
+             jemalloc/include/jemalloc/internal/atomic.h \
+             jemalloc/include/jemalloc/internal/atomic_c11.h \
+             jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h \
+             jemalloc/include/jemalloc/internal/atomic_gcc_sync.h \
+             jemalloc/include/jemalloc/internal/atomic_msvc.h \
+             jemalloc/include/jemalloc/internal/background_thread_externs.h \
+             jemalloc/include/jemalloc/internal/background_thread_inlines.h \
+             jemalloc/include/jemalloc/internal/background_thread_structs.h \
+             jemalloc/include/jemalloc/internal/base_externs.h \
+             jemalloc/include/jemalloc/internal/base_inlines.h \
+             jemalloc/include/jemalloc/internal/base_structs.h \
+             jemalloc/include/jemalloc/internal/base_types.h \
+             jemalloc/include/jemalloc/internal/bin.h \
+             jemalloc/include/jemalloc/internal/bin_stats.h \
+             jemalloc/include/jemalloc/internal/bin_types.h \
+             jemalloc/include/jemalloc/internal/bitmap.h \
+             jemalloc/include/jemalloc/internal/bit_util.h \
+             jemalloc/include/jemalloc/internal/cache_bin.h \
+             jemalloc/include/jemalloc/internal/ckh.h \
+             jemalloc/include/jemalloc/internal/ctl.h \
+             jemalloc/include/jemalloc/internal/div.h \
+             jemalloc/include/jemalloc/internal/emitter.h \
+             jemalloc/include/jemalloc/internal/extent_dss.h \
+             jemalloc/include/jemalloc/internal/extent_externs.h \
+             jemalloc/include/jemalloc/internal/extent_inlines.h \
+             jemalloc/include/jemalloc/internal/extent_mmap.h \
+             jemalloc/include/jemalloc/internal/extent_structs.h \
+             jemalloc/include/jemalloc/internal/extent_types.h \
+             jemalloc/include/jemalloc/internal/hash.h \
+             jemalloc/include/jemalloc/internal/hook.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h \
+             jemalloc/include/jemalloc/internal/jemalloc_internal_types.h \
+             jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in \
+             jemalloc/include/jemalloc/internal/large_externs.h \
+             jemalloc/include/jemalloc/internal/log.h \
+             jemalloc/include/jemalloc/internal/malloc_io.h \
+             jemalloc/include/jemalloc/internal/mutex.h \
+             jemalloc/include/jemalloc/internal/mutex_pool.h \
+             jemalloc/include/jemalloc/internal/mutex_prof.h \
+             jemalloc/include/jemalloc/internal/nstime.h \
+             jemalloc/include/jemalloc/internal/pages.h \
+             jemalloc/include/jemalloc/internal/ph.h \
+             jemalloc/include/jemalloc/internal/private_namespace.sh \
+             jemalloc/include/jemalloc/internal/private_symbols.sh \
+             jemalloc/include/jemalloc/internal/prng.h \
+             jemalloc/include/jemalloc/internal/prof_externs.h \
+             jemalloc/include/jemalloc/internal/prof_inlines_a.h \
+             jemalloc/include/jemalloc/internal/prof_inlines_b.h \
+             jemalloc/include/jemalloc/internal/prof_structs.h \
+             jemalloc/include/jemalloc/internal/prof_types.h \
+             jemalloc/include/jemalloc/internal/public_namespace.sh \
+             jemalloc/include/jemalloc/internal/public_unnamespace.sh \
+             jemalloc/include/jemalloc/internal/ql.h \
+             jemalloc/include/jemalloc/internal/qr.h \
+             jemalloc/include/jemalloc/internal/quantum.h \
+             jemalloc/include/jemalloc/internal/rb.h \
+             jemalloc/include/jemalloc/internal/rtree.h \
+             jemalloc/include/jemalloc/internal/rtree_tsd.h \
+             jemalloc/include/jemalloc/internal/safety_check.h \
+             jemalloc/include/jemalloc/internal/sc.h \
+             jemalloc/include/jemalloc/internal/seq.h \
+             jemalloc/include/jemalloc/internal/smoothstep.h \
+             jemalloc/include/jemalloc/internal/smoothstep.sh \
+             jemalloc/include/jemalloc/internal/spin.h \
+             jemalloc/include/jemalloc/internal/stats.h \
+             jemalloc/include/jemalloc/internal/sz.h \
+             jemalloc/include/jemalloc/internal/tcache_externs.h \
+             jemalloc/include/jemalloc/internal/tcache_inlines.h \
+             jemalloc/include/jemalloc/internal/tcache_structs.h \
+             jemalloc/include/jemalloc/internal/tcache_types.h \
+             jemalloc/include/jemalloc/internal/test_hooks.h \
+             jemalloc/include/jemalloc/internal/ticker.h \
+             jemalloc/include/jemalloc/internal/tsd.h \
+             jemalloc/include/jemalloc/internal/tsd_generic.h \
+             jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h \
+             jemalloc/include/jemalloc/internal/tsd_tls.h \
+             jemalloc/include/jemalloc/internal/tsd_types.h \
+             jemalloc/include/jemalloc/internal/tsd_win.h \
+             jemalloc/include/jemalloc/internal/util.h \
+             jemalloc/include/jemalloc/internal/witness.h \
+             jemalloc/include/jemalloc/jemalloc.sh \
+             jemalloc/include/jemalloc/jemalloc_defs.h.in \
+             jemalloc/include/jemalloc/jemalloc_macros.h.in \
+             jemalloc/include/jemalloc/jemalloc_mangle.sh \
+             jemalloc/include/jemalloc/jemalloc_protos.h.in \
+             jemalloc/include/jemalloc/jemalloc_rename.sh \
+             jemalloc/include/jemalloc/jemalloc_typedefs.h.in \
+             jemalloc/include/msvc_compat/C99/stdbool.h \
+             jemalloc/include/msvc_compat/C99/stdint.h \
+             jemalloc/include/msvc_compat/strings.h \
+             jemalloc/include/msvc_compat/windows_extra.h \
+             jemalloc/INSTALL.md \
+             jemalloc/jemalloc.pc.in \
+             jemalloc/m4/ax_cxx_compile_stdcxx.m4 \
+             jemalloc/Makefile.in \
+             jemalloc/msvc/jemalloc_vc2015.sln \
+             jemalloc/msvc/jemalloc_vc2017.sln \
+             jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj \
+             jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters \
+             jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj \
+             jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters \
+             jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj \
+             jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters \
+             jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj \
+             jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters \
+             jemalloc/msvc/ReadMe.txt \
+             jemalloc/msvc/test_threads/test_threads.cpp \
+             jemalloc/msvc/test_threads/test_threads.h \
+             jemalloc/msvc/test_threads/test_threads_main.cpp \
+             jemalloc/README \
+             jemalloc/run_tests.sh \
+             jemalloc/scripts/gen_run_tests.py \
+             jemalloc/scripts/gen_travis.py \
+             jemalloc/src/arena.c \
+             jemalloc/src/background_thread.c \
+             jemalloc/src/base.c \
+             jemalloc/src/bin.c \
+             jemalloc/src/bitmap.c \
+             jemalloc/src/ckh.c \
+             jemalloc/src/ctl.c \
+             jemalloc/src/div.c \
+             jemalloc/src/extent.c \
+             jemalloc/src/extent_dss.c \
+             jemalloc/src/extent_mmap.c \
+             jemalloc/src/hash.c \
+             jemalloc/src/hook.c \
+             jemalloc/src/jemalloc.c \
+             jemalloc/src/jemalloc_cpp.cpp \
+             jemalloc/src/large.c \
+             jemalloc/src/log.c \
+             jemalloc/src/malloc_io.c \
+             jemalloc/src/mutex.c \
+             jemalloc/src/mutex_pool.c \
+             jemalloc/src/nstime.c \
+             jemalloc/src/pages.c \
+             jemalloc/src/prng.c \
+             jemalloc/src/prof.c \
+             jemalloc/src/rtree.c \
+             jemalloc/src/safety_check.c \
+             jemalloc/src/sc.c \
+             jemalloc/src/stats.c \
+             jemalloc/src/sz.c \
+             jemalloc/src/tcache.c \
+             jemalloc/src/test_hooks.c \
+             jemalloc/src/ticker.c \
+             jemalloc/src/tsd.c \
+             jemalloc/src/witness.c \
+             jemalloc/src/zone.c \
+             jemalloc/test/include/test/btalloc.h \
+             jemalloc/test/include/test/extent_hooks.h \
+             jemalloc/test/include/test/jemalloc_test.h.in \
+             jemalloc/test/include/test/jemalloc_test_defs.h.in \
+             jemalloc/test/include/test/math.h \
+             jemalloc/test/include/test/mq.h \
+             jemalloc/test/include/test/mtx.h \
+             jemalloc/test/include/test/SFMT-alti.h \
+             jemalloc/test/include/test/SFMT-params.h \
+             jemalloc/test/include/test/SFMT-params607.h \
+             jemalloc/test/include/test/SFMT-params1279.h \
+             jemalloc/test/include/test/SFMT-params2281.h \
+             jemalloc/test/include/test/SFMT-params4253.h \
+             jemalloc/test/include/test/SFMT-params11213.h \
+             jemalloc/test/include/test/SFMT-params19937.h \
+             jemalloc/test/include/test/SFMT-params44497.h \
+             jemalloc/test/include/test/SFMT-params86243.h \
+             jemalloc/test/include/test/SFMT-params132049.h \
+             jemalloc/test/include/test/SFMT-params216091.h \
+             jemalloc/test/include/test/SFMT-sse2.h \
+             jemalloc/test/include/test/SFMT.h \
+             jemalloc/test/include/test/test.h \
+             jemalloc/test/include/test/thd.h \
+             jemalloc/test/include/test/timer.h \
+             jemalloc/test/integration/aligned_alloc.c \
+             jemalloc/test/integration/allocated.c \
+             jemalloc/test/integration/extent.c \
+             jemalloc/test/integration/extent.sh \
+             jemalloc/test/integration/malloc.c \
+             jemalloc/test/integration/mallocx.c \
+             jemalloc/test/integration/mallocx.sh \
+             jemalloc/test/integration/MALLOCX_ARENA.c \
+             jemalloc/test/integration/overflow.c \
+             jemalloc/test/integration/posix_memalign.c \
+             jemalloc/test/integration/rallocx.c \
+             jemalloc/test/integration/sdallocx.c \
+             jemalloc/test/integration/slab_sizes.c \
+             jemalloc/test/integration/slab_sizes.sh \
+             jemalloc/test/integration/smallocx.c \
+             jemalloc/test/integration/smallocx.sh \
+             jemalloc/test/integration/thread_arena.c \
+             jemalloc/test/integration/thread_tcache_enabled.c \
+             jemalloc/test/integration/xallocx.c \
+             jemalloc/test/integration/xallocx.sh \
+             jemalloc/test/src/btalloc.c \
+             jemalloc/test/src/btalloc_0.c \
+             jemalloc/test/src/btalloc_1.c \
+             jemalloc/test/src/math.c \
+             jemalloc/test/src/mq.c \
+             jemalloc/test/src/mtx.c \
+             jemalloc/test/src/SFMT.c \
+             jemalloc/test/src/test.c \
+             jemalloc/test/src/thd.c  \
+             jemalloc/test/src/timer.c \
+             jemalloc/test/stress/hookbench.c \
+             jemalloc/test/stress/microbench.c \
+             jemalloc/test/test.sh.in \
+             jemalloc/test/unit/a0.c \
+             jemalloc/test/unit/arena_reset.c \
+             jemalloc/test/unit/arena_reset_prof.c \
+             jemalloc/test/unit/arena_reset_prof.sh \
+             jemalloc/test/unit/atomic.c \
+             jemalloc/test/unit/background_thread.c \
+             jemalloc/test/unit/background_thread_enable.c \
+             jemalloc/test/unit/base.c \
+             jemalloc/test/unit/binshard.c \
+             jemalloc/test/unit/binshard.sh \
+             jemalloc/test/unit/bitmap.c \
+             jemalloc/test/unit/bit_util.c \
+             jemalloc/test/unit/ckh.c \
+             jemalloc/test/unit/decay.c \
+             jemalloc/test/unit/decay.sh \
+             jemalloc/test/unit/div.c \
+             jemalloc/test/unit/emitter.c \
+             jemalloc/test/unit/extent_quantize.c\
+             jemalloc/test/unit/extent_util.c \
+             jemalloc/test/unit/fork.c \
+             jemalloc/test/unit/hash.c \
+             jemalloc/test/unit/hook.c \
+             jemalloc/test/unit/huge.c \
+             jemalloc/test/unit/junk.c \
+             jemalloc/test/unit/junk.sh \
+             jemalloc/test/unit/junk_alloc.c \
+             jemalloc/test/unit/junk_alloc.sh \
+             jemalloc/test/unit/junk_free.c \
+             jemalloc/test/unit/junk_free.sh \
+             jemalloc/test/unit/log.c \
+             jemalloc/test/unit/mallctl.c \
+             jemalloc/test/unit/malloc_io.c \
+             jemalloc/test/unit/math.c \
+             jemalloc/test/unit/mq.c \
+             jemalloc/test/unit/mtx.c \
+             jemalloc/test/unit/nstime.c \
+             jemalloc/test/unit/pack.c \
+             jemalloc/test/unit/pack.sh \
+             jemalloc/test/unit/pages.c \
+             jemalloc/test/unit/ph.c \
+             jemalloc/test/unit/prng.c \
+             jemalloc/test/unit/prof_accum.c \
+             jemalloc/test/unit/prof_accum.sh \
+             jemalloc/test/unit/prof_active.c \
+             jemalloc/test/unit/prof_active.sh \
+             jemalloc/test/unit/prof_gdump.c \
+             jemalloc/test/unit/prof_gdump.sh \
+             jemalloc/test/unit/prof_idump.c \
+             jemalloc/test/unit/prof_idump.sh \
+             jemalloc/test/unit/prof_log.c \
+             jemalloc/test/unit/prof_log.sh \
+             jemalloc/test/unit/prof_reset.c \
+             jemalloc/test/unit/prof_reset.sh \
+             jemalloc/test/unit/prof_tctx.c \
+             jemalloc/test/unit/prof_tctx.sh \
+             jemalloc/test/unit/prof_thread_name.c \
+             jemalloc/test/unit/prof_thread_name.sh\
+             jemalloc/test/unit/ql.c \
+             jemalloc/test/unit/qr.c \
+             jemalloc/test/unit/rb.c \
+             jemalloc/test/unit/retained.c \
+             jemalloc/test/unit/rtree.c \
+             jemalloc/test/unit/safety_check.c \
+             jemalloc/test/unit/safety_check.sh \
+             jemalloc/test/unit/sc.c \
+             jemalloc/test/unit/seq.c \
+             jemalloc/test/unit/SFMT.c \
+             jemalloc/test/unit/size_classes.c \
+             jemalloc/test/unit/slab.c \
+             jemalloc/test/unit/smoothstep.c \
+             jemalloc/test/unit/spin.c \
+             jemalloc/test/unit/stats.c \
+             jemalloc/test/unit/stats_print.c \
+             jemalloc/test/unit/test_hooks.c \
+             jemalloc/test/unit/ticker.c \
+             jemalloc/test/unit/tsd.c \
+             jemalloc/test/unit/witness.c \
+             jemalloc/test/unit/zero.c \
+             jemalloc/test/unit/zero.sh \
+             jemalloc/TUNING.md \
+             memkind-$(VERSION).spec \
+             memkind.pc.in \
+             memkind.spec.mk \
+             test/test.sh \
+             test/test_dax_kmem.sh \
+             utils/docker/Dockerfile.fedora-31 \
+             utils/docker/Dockerfile.ubuntu-18.04 \
+             utils/docker/README.md \
+             utils/docker/docker_install_ndctl.sh \
+             utils/docker/docker_install_tbb.sh \
+             utils/docker/docker_run_build.sh \
+             utils/docker/docker_run_coverage.sh \
+             utils/docker/docker_run_test.sh \
+             utils/docker/run_local.sh \
+             utils/docker/set_host_configuration.sh \
+             # end
+
+dist_doc_DATA = COPYING README VERSION
+dist_man_MANS =  man/autohbw.7 \
+                 man/hbwallocator.3 \
+                 man/hbwmalloc.3 \
+                 man/memkind-auto-dax-kmem-nodes.1 \
+                 man/memkind-hbw-nodes.1 \
+                 man/memkind.3 \
+                 man/memkind_arena.3 \
+                 man/memkind_dax_kmem.3 \
+                 man/memkind_default.3 \
+                 man/memkind_hbw.3 \
+                 man/memkind_hugetlb.3 \
+                 man/memkind_pmem.3 \
+                 man/memkindallocator.3 \
+                 man/pmemallocator.3 \
+                 # end
+
+pkgconfigdir = $(libdir)/pkgconfig
+pkgconfig_DATA = memkind.pc
+
+CLEANFILES = memkind-$(VERSION).spec $(libmemkind_la_LIBADD)
+DISTCLEANFILES = VERSION configure pkgconfig_DATA
+
+bin_PROGRAMS = memkind-auto-dax-kmem-nodes memkind-hbw-nodes
+
+memkind_auto_dax_kmem_nodes_SOURCES = src/memkind-auto-dax-kmem-nodes.c
+memkind_auto_dax_kmem_nodes_LDADD = libmemkind.la
+memkind_auto_dax_kmem_nodes_LDFLAGS = $(DAXCTL_LIBS)
+
+memkind_hbw_nodes_SOURCES = src/memkind-hbw-nodes.c
+memkind_hbw_nodes_LDADD = libmemkind.la
+
+bin_SCRIPTS =
+check_PROGRAMS =
+noinst_PROGRAMS =
+noinst_LTLIBRARIES =
+TESTS =
+
+req_flags = -fvisibility=hidden -Wall -Werror -D_GNU_SOURCE -DJE_PREFIX=@memkind_prefix@
+
+AM_CFLAGS = $(req_flags)
+AM_CXXFLAGS = $(req_flags)
+
+clean-local: clean-jemalloc_deps
+
+html-local:
+	$(MAN2HTML) man/hbwmalloc.3 > man/hbwmalloc.html
+	$(MAN2HTML) man/memkind.3 > man/memkind.html
+	$(MAN2HTML) man/memkindallocator.3 > man/memkindallocator.html
+	$(MAN2HTML) man/pmemallocator.3 > man/pmemallocator.html
+
+clean-jemalloc_deps:
+	[ -e jemalloc/configure ] && (cd jemalloc && $(MAKE) clean) || exit 0
+	[ -e jemalloc/configure ] && (cd jemalloc && rm configure) || exit 0
+
+.PHONY: jemalloc_deps checkprogs clean-jemalloc_deps html-local
+.NOTPARALLEL: static_lib
+
+# build check programs without running tests
+checkprogs: libmemkind.la $(check_PROGRAMS)
+
+memkind-$(VERSION).spec:
+	$(MAKE) version="$(VERSION)" -f memkind.spec.mk $@
+
+rpm: dist
+	$(MAKE) version="$(VERSION)" -f memkind.spec.mk $@
+
+all: jemalloc_deps static_lib
+
+# the script merge memkind and jemalloc libraries into one static library.
+define ar_prog
+create libmemkind.a\n\
+addlib .libs/libmemkind.a\n\
+addlib $(libmemkind_la_LIBADD)\n\
+save\n\
+end
+endef
+
+static_lib: libmemkind.la
+	bash -c "ar -M < <(echo -e '$(ar_prog)')"
+	cp libmemkind.a .libs/
+	rm libmemkind.a
+
+JEMALLOC_CONFIG = --enable-autogen --without-export --with-version=5.2.1-0-gea6b3e973b477b8061e0076bb257dbd7f3faa756 \
+			--disable-fill --disable-initial-exec-tls --with-jemalloc-prefix=@memkind_prefix@ \
+			--with-malloc-conf="narenas:@auto_arenas@,lg_tcache_max:@tcache_max_size_class@" \
+			#end
+
+jemalloc_deps:
+	[ ! -e jemalloc/configure ] && (cd jemalloc && autoconf) || exit 0
+	[ ! -e $(libmemkind_la_LIBADD) ] && (cd jemalloc && ./configure $(JEMALLOC_CONFIG) && $(MAKE)) || exit 0
+
+unit_tests: checkprogs
+	test/all_tests --gtest_filter=*free_with_NULL_kind_4096_bytes:NegativeTest.*:ErrorMessage.*:GetArenaTest.*:MemkindVersioningTests.*
+	test/decorator_test
+
+PMEM_PATH=/tmp/
+unit_tests_pmem: checkprogs
+	test/pmem_test -d $(PMEM_PATH)
+
+unit_tests_dax_kmem: checkprogs
+	test/test_dax_kmem.sh
+
+include autohbw/Makefile.mk
+include test/Makefile.mk
+include examples/Makefile.mk
+include src/Makefile.mk
diff --git a/memkind-1.10.0/NEWS b/memkind-1.10.0/NEWS
new file mode 100644
index 0000000..c72d191
--- /dev/null
+++ b/memkind-1.10.0/NEWS
@@ -0,0 +1,42 @@
+OPEN DEVELOPMENT
+================
+
+We are moving to a more open development model with memkind.
+
+- The memkind github organization:
+  https://github.com/memkind
+
+- Rather than posting bulk patches at time of tag, we are posting more
+  incremental patches with finer grain commit details.
+
+- There is now a "dev" branch in the memkind repository and a
+  "memkind-dev" branch in the memkind fork of the jemalloc repository.
+  These will be used for open development work.
+
+- There is a memkind web page here:
+  http://memkind.github.io/memkind
+
+- We have posted an architecture document here:
+  http://memkind.github.io/memkind/memkind_arch_20150318.pdf
+
+- We invite you to join the memkind mailing list:
+  https://lists.01.org/mailman/listinfo/memkind
+
+- Feel free to open a github issue, especially if there is a TODO item
+  that you would like to prioritize.
+
+- We have posted source and binary RPMs for several distributions here:
+  http://download.opensuse.org/repositories/home:/cmcantalupo/
+
+- memkind is also available in Fedora package repository (devel, EPEL7):
+  https://admin.fedoraproject.org/pkgdb/package/memkind/
+
+- Pull requests are welcome.
+
+- White space formatting generally conforms to
+  $ astyle --style=linux --indent=spaces=4 -S --max-continuation-indent=80 \
+           --max-code-length=80 --break-after-logical --indent-namespaces -z2 \
+           --align-pointer=name
+
+- For more information about astyle:
+  http://astyle.sourceforge.net/
diff --git a/memkind-1.10.0/PULL_REQUEST_TEMPLATE.md b/memkind-1.10.0/PULL_REQUEST_TEMPLATE.md
new file mode 100644
index 0000000..41b7945
--- /dev/null
+++ b/memkind-1.10.0/PULL_REQUEST_TEMPLATE.md
@@ -0,0 +1,29 @@
+<!--- Provide a general summary of your changes in the Title above -->
+
+### Description
+<!--- Describe your changes in detail -->
+<!--- If it fixes an open issue, please link to the issue here. -->
+
+### Types of changes
+<!--- Put an `x` in the box(es) that apply -->
+
+- [ ] Bugfix (non-breaking change which fixes issue linked in Description above)
+- [ ] New feature (non-breaking change which adds functionality)
+- [ ] Breaking change (fix or feature that would cause existing functionality to not work as expected)
+- [ ] Documentation (correction or otherwise)
+- [ ] Cosmetics (whitespace, appearance)
+- [ ] Other
+
+### Checklist
+<!--- Put an `x` in the box(es) that apply -->
+
+- [ ] Code compiles without errors
+- [ ] All tests pass locally with my changes (see TESTING section in CONTRIBUTING file)
+- [ ] Created tests which will fail without the change (if possible)
+- [ ] Extended the README/documentation (if necessary)
+- [ ] All newly added files have proprietary license (if necessary)
+- [ ] All newly added files are referenced in MANIFEST files (if necessary)
+
+## Further comments
+<!--- If this is a relatively large or complex change, kick off the discussion by explaining why you -->
+<!--- choose the solution you did and what alternatives you considered, etc... -->
diff --git a/memkind-1.10.0/README b/memkind-1.10.0/README
new file mode 100644
index 0000000..f76195a
--- /dev/null
+++ b/memkind-1.10.0/README
@@ -0,0 +1,390 @@
+# **MEMKIND**
+
+[![Build Status](https://travis-ci.org/memkind/memkind.svg)](https://travis-ci.org/memkind/memkind)
+[![MEMKIND version](https://img.shields.io/github/tag/memkind/memkind.svg)](https://github.com/memkind/memkind/releases/latest)
+[![Coverage Status](http://codecov.io/github/memkind/memkind/coverage.svg?branch=master)](http://codecov.io/gh/memkind/memkind?branch=master)
+[![Packaging status](https://repology.org/badge/tiny-repos/memkind.svg)](https://repology.org/project/memkind/versions)
+
+The memkind library is a user extensible heap manager built on top of
+jemalloc which enables control of memory characteristics and a
+partitioning of the heap between kinds of memory.  The kinds of memory
+are defined by operating system memory policies that have been applied
+to virtual address ranges.  Memory characteristics supported by
+memkind without user extension include control of NUMA and page size
+features.  The jemalloc non-standard interface has been extended to
+enable specialized arenas to make requests for virtual memory from the
+operating system through the memkind partition interface.  Through the
+other memkind interfaces the user can control and extend memory
+partition features and allocate memory while selecting enabled
+features.  Memkind interface allows to create and control file-backed memory
+(PMEM kind) on specified device.
+
+## Contents
+
+1. [Interfaces](#interfaces)
+2. [Dependencies](#dependencies)
+3. [Building and Installing](#building-and-installing)
+	* [jemalloc](#jemalloc)
+4. [Run Requirements](#run-requirements)
+5. [Kind Requirements](#kind-requirements)
+6. [Kernel](#kernel)
+7. [NVDIMM volatile usage](#nvdimm-volatile-usage)
+	* [DAX device](#dax-device)
+	* [DAX filesystem](#dax-filesystem)
+8. [The Detection Mechanism of the Kind](#the-detection-mechanism-of-the-kind)
+9. [Setting Logging Mechanism](#setting-logging-mechanism)
+10. [Setting Heap Manager](#setting-heap-manager)
+11. [Testing](#testing)
+12. [Simulating High Bandwidth Memory](#simulating-high-bandwidth-memory)
+13. [Notes](#notes)
+14. [Status](#status)
+15. [Disclaimer](#disclaimer)
+
+## Interfaces
+The memkind library delivers four interfaces:
+ * hbwmalloc.h - recommended for high-bandwidth memory use cases (stable)
+ * memkind.h - generic interface for more complex use cases (stable)
+ * pmem_allocator.h - the C++ allocator that satisfies the C++ Allocator
+   [requirements](https://en.cppreference.com/w/cpp/named_req/Allocator)
+   used for PMEM memory use cases (stable)
+ * memkind_allocator.h - the C++ allocator that satisfies the C++ Allocator
+   [requirements](https://en.cppreference.com/w/cpp/named_req/Allocator)
+   used for static kinds (stable)
+
+For more detailed information about those interfaces see
+corresponding manpages (located in man/ subdir):
+
+    man memkind
+
+    man hbwmalloc
+
+    man pmemallocator
+
+    man memkindallocator
+
+## Dependencies
+
+You will need to install required packages on the build system:
+
+* **autoconf**
+* **automake**
+* **gcc-c++**
+* **libnuma-devel**
+* **libtool**
+* **numactl-devel**
+* **unzip**
+
+For using automatic recognition of PMEM NUMA in MEMKIND_DAX_KMEM:
+
+* **libdaxctl-devel** (v66 or later)
+
+## Building and Installing
+
+The memkind library has a dependency on a related fork of jemalloc.
+The configure scripts and gtest source code are distributed with the
+source tarball included in the source RPM, and this tarball is created
+with the memkind "make dist" target. In contrast to the distributed source
+tarball, the git repository does not include any generated files.
+For this reason some additional steps are required when building
+from a checkout of the git repo. Those steps include running
+the bash script called "autogen.sh" prior to configure.  This script
+will populate a VERSION file based on "git describe", and use
+autoreconf to generate a configure script.
+
+Building and installing memkind in standard system location can be as simple as
+typing the following while in the root directory of the source tree:
+
+    ./autogen.sh
+    ./configure
+    make
+    make install
+
+To install this library into **other locations**, you can use the prefix variable, e.g.:
+
+    ./autogen.sh
+    ./configure --prefix=/usr
+    make
+    make install
+
+This will install files to /usr/lib, /usr/include, /usr/share/doc/, usr/share/man.
+
+See the output of:
+
+    ./configure --help
+
+for more information about either the memkind or the jemalloc configuration options.
+
+## jemalloc
+The jemalloc source was forked from jemalloc version 5.2.1.  This source tree
+is located within the jemalloc subdirectory of the memkind source.  The jemalloc
+source code has been kept close to the original form, and in particular
+the build system has been lightly modified.
+
+## Run Requirements
+
+You will need to install required packages for applications,
+which are using the memkind library for dynamic linking at run time:
+
+* **libnuma**
+* **numactl**
+* **pthread**
+
+## Kind Requirements
+
+| Memory kind                   | NUMA  |  HBW Memory | Hugepages | Device DAX | Filesystem supporting hole punching |
+| ----------------------------- |:-----:|:-----------:|:---------:|:----------:|:-----------------------------------:|
+| MEMKIND_DEFAULT               |       |             |           |            |                                     |
+| MEMKIND_HUGETLB               | X     | X           | X         |            |                                     |
+| MEMKIND_HBW                   | X     | X           |           |            |                                     |
+| MEMKIND_HBW_ALL               | X     | X           |           |            |                                     |
+| MEMKIND_HBW_HUGETLB           | X     | X           | X         |            |                                     |
+| MEMKIND_HBW_ALL_HUGETLB       | X     | X           | X         |            |                                     |
+| MEMKIND_HBW_PREFERRED         | X     | X           |           |            |                                     |
+| MEMKIND_HBW_PREFERRED_HUGETLB | X     | X           | X         |            |                                     |
+| MEMKIND_HBW_INTERLEAVE        | X     | X           |           |            |                                     |
+| MEMKIND_REGULAR               | X     |             |           |            |                                     |
+| MEMKIND_DAX_KMEM              | X     |             |           | X          |                                     |
+| MEMKIND_DAX_KMEM_ALL          | X     |             |           | X          |                                     |
+| MEMKIND_DAX_KMEM_PREFERRED    | X     |             |           | X          |                                     |
+| PMEM kind                     |       |             |           |            | X                                   |
+
+## Kernel
+To correctly control of NUMA, huge pages and file-backed memory following
+requirements regarding Linux kernel must be satisfied:
+
+* **NUMA**
+
+Requires kernel
+[patch](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=3964acd0dbec123aa0a621973a2a0580034b4788)
+introduced in Linux v3.11 that impacts
+functionality of the NUMA system calls.  Red Hat has back-ported this patch to the
+v3.10 kernel in the RHEL 7.0 GA release, so RHEL 7.0 onward supports
+memkind even though this kernel version predates v3.11.
+
+* **Hugepages**
+
+Functionality related to hugepages allocation require patches
+[patch1](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=e0ec90ee7e6f6cbaa6d59ffb48d2a7af5e80e61d) and
+[patch2](https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=099730d67417dfee273e9b10ac2560ca7fac7eb9)
+Without them physical memory may end up being located on incorrect NUMA node.
+
+* **2MB Pages**
+
+To use the interfaces for obtaining 2MB pages please be sure to follow the instructions
+[here](https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt) and pay particular
+attention to the use of the procfs files:
+
+    /proc/sys/vm/nr_hugepages
+    /proc/sys/vm/nr_overcommit_hugepages
+
+for enabling the kernel's huge page pool.
+
+* **Filesystem supporting hole punching**
+
+To use the PMEM kind, please be sure that filesystem which is used for PMEM creation supports
+[FALLOC_FL_PUNCH_HOLE](http://man7.org/linux/man-pages/man2/fallocate.2.html) flag.
+
+* **Device DAX**
+
+To use MEMKIND_DAX_KMEM_* kinds you need at least Linux Kernel 5.1 (with enabled DEV_DAX_KMEM Kernel option) and created DAX device.
+If you have loaded dax_pmem_compat module instead of dax_pmem, please read
+[this](https://pmem.io/ndctl/daxctl-migrate-device-model.html) article and migrate device model
+to use alternative device dax drivers, e.g. kmem.
+
+If you want to migrate device model type:
+
+    daxctl migrate-device-model
+
+To create a new DAX device you can type:
+
+    ndctl create-namespace --mode=devdax --map=mem
+
+To display a list of created devices:
+
+    ndctl list
+
+If you have already created device in another mode, you can reconfigure the device using:
+
+    ndctl create-namespace --mode=devdax --map=mem --force -e namespaceX.Y
+
+Where namespaceX.Y means namespace you want to reconfigure.
+
+For more details about creating new namespace read [this](https://pmem.io/ndctl/ndctl-create-namespace.html).
+
+Conversion from device dax to NUMA node can be performed using following commands:
+
+    daxctl reconfigure-device daxX.Y --mode=system-ram
+
+Where daxX.Y is DAX device you want to reconfigure.
+
+This will migrate device from device_dax to kmem driver. After this step daxctl should be able to
+see devices in system-ram mode:
+
+    daxctl list
+
+Example output after reconfigure dax2.0 as system-ram:
+
+    [
+      {
+        "chardev":"dax1.0",
+        "size":263182090240,
+        "target_node":3,
+        "mode":"devdax"
+      },
+      {
+        "chardev":"dax2.0",
+        "size":263182090240,
+        "target_node":4,
+        "mode":"system-ram"
+      }
+    ]
+
+Also you should be able to check new NUMA node configuration by:
+
+    numactl -H
+
+Example output, where NUMA node 4 is NUMA node created from persistent memory:
+
+    available: 3 nodes (0-1,4)
+    node 0 cpus: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83
+    node 0 size: 192112 MB
+    node 0 free: 185575 MB
+    node 1 cpus: 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
+    node 1 size: 193522 MB
+    node 1 free: 193107 MB
+    node 4 cpus:
+    node 4 size: 250880 MB
+    node 4 free: 250879 MB
+    node distances:
+    node   0   1   4
+      0:  10  21  17
+      1:  21  10  28
+      4:  17  28  10
+
+## NVDIMM volatile usage
+
+Memkind supports using persistent memory as an extension of DRAM.
+This volatile memory solution is provided by the library with two separate ways described below.
+
+## DAX device
+
+NVDIMM memory as DAX device is supported by MEMKIND_DAX_KMEM_* kinds. With this solution
+persistent memory will be seen in OS as separate NUMA nodes.
+
+Memkind allows two ways to use this kind:
+
+- first implicitly, by allowing memkind library for automatic recognition of NUMA nodes
+created from persistent memory using [libdaxctl-devel](#dependencies)
+- secondary explicitly, by using MEMKIND_DAX_KMEM_NODES environment variable set to
+comma separated list of NUMA nodes which will be treated as NUMA nodes created from persistent memory,
+this solution overrides the first one
+
+## DAX filesystem
+PMEM kind supports the traditional malloc/free interfaces on a memory mapped file.
+This allows the use of [persistent memory](https://pmem.io/)
+as volatile memory, for cases where the region of persistent memory
+is useful to an application, but when the application doesn't need it to be persistent.
+PMEM kind is most useful when used with Direct Access storage
+[DAX](https://www.kernel.org/doc/Documentation/filesystems/dax.txt), which is
+memory-addressable persistent storage that supports load/store access without being paged via the system page cache.
+
+Application using memkind library supports managing a data placement:
+
+| Data placement | Memory kind               |
+|----------------|---------------------------|
+| PMEM (fsdax)   | PMEM kind                 |
+| PMEM (devdax)  | MEMKIND_DAX_KMEM kind     |
+| DRAM           | e.g. MEMKIND_DEFAULT kind |
+
+Currently, the PMEM kind is supported only by the jemalloc heap manager.
+
+## The Detection Mechanism of the Kind
+One of the notable features of the memkind is to detect the correct kind of previously allocated memory.
+
+| Operations                                  | Memkind API function                   |
+|---------------------------------------------|----------------------------------------|
+| Freeing memory                              | memkind_free(kind, ptr)                |
+| Reallocating memory                         | memkind_realloc(kind, ptr, size)       |
+| Obtaining the size of allocated memory      | memkind_malloc_usable_size(kind, ptr)  |
+| Reallocating memory to reduce fragmentation | memkind_defrag_reallocate(kind, ptr)   |
+
+Operations above could be unified for all used memory kinds
+by passing a NULL value as a kind to the functions mentioned above.
+
+For more details, please see the following
+[example](https://github.com/memkind/memkind/blob/master/examples/pmem_free_with_unknown_kind.c).
+
+**Important Notes:**
+The look up for correct kind could result in serious performance penalty,
+which can be avoided by specifying a correct kind explicitly.
+
+## Setting Logging Mechanism
+In memkind library logging mechanism could be enabled by setting MEMKIND_DEBUG
+environment variable. Setting MEMKIND_DEBUG to "1" enables printing messages
+like errors and general information about environment to stderr.
+
+## Setting Heap Manager
+In memkind library heap management can be adjusted with MEMKIND_HEAP_MANAGER
+environment variable, which allows for switching to one of the available
+heap managers.
+Values:
+- JEMALLOC - sets the jemalloc heap manager
+- TBB - sets Intel Threading Building Blocks heap manager. This option requires installed
+Intel Threading Building Blocks library.
+
+If the MEMKIND_HEAP_MANAGER is not set then the jemalloc heap manager will be used by default.
+
+## Testing
+All existing tests pass. For more information on how to execute tests
+see the [CONTRIBUTING](CONTRIBUTING) file.
+
+When tests are run on a NUMA platform without high bandwidth memory
+the MEMKIND_HBW_NODES environment variable is used in conjunction with
+"numactl --membind" to force standard allocations to one NUMA node and
+high bandwidth allocations through a different NUMA node.  See next
+section for more details.
+
+
+## Simulating High Bandwidth Memory
+A method for testing for the benefit of high bandwidth memory on a
+dual socket Intel(R) Xeon(TM) system is to use the QPI bus to simulate
+slow memory.  This is not an accurate model of the bandwidth and
+latency characteristics of the Intel's 2nd generation Intel(R) Xeon Phi(TM)
+Product Family on package memory, but is a reasonable way to determine
+which data structures rely critically on bandwidth.
+
+If the application a.out has been modified to use high bandwidth
+memory with the memkind library then this can be done with numactl as
+follows with the bash shell:
+
+    export MEMKIND_HBW_NODES=0
+    numactl --membind=1 --cpunodebind=0 a.out
+
+or with csh:
+
+    setenv MEMKIND_HBW_NODES 0
+    numactl --membind=1 --cpunodebind=0 a.out
+
+The MEMKIND_HBW_NODES environment variable set to zero will bind high
+bandwidth allocations to NUMA node 0.  The --membind=1 flag to numactl
+will bind standard allocations, static and stack variables to NUMA
+node 1.  The --cpunodebind=0 option to numactl will bind the process
+threads to CPUs associated with NUMA node 0.  With this configuration
+standard allocations will be fetched across the QPI bus, and high
+bandwidth allocations will be local to the process CPU.
+
+## Notes
+* Using memkind with Transparent Huge Pages enabled may result in
+ undesirably high memory footprint. To avoid that disable THP using following
+ [instruction](https://www.kernel.org/doc/Documentation/vm/transhuge.txt)
+
+## Status
+Different interfaces can represent different maturity level
+(as described in corresponding man pages).
+Feedback on design and implementation is greatly appreciated.
+
+## Disclaimer
+SEE [COPYING](COPYING) FILE FOR LICENSE INFORMATION.
+
+THIS SOFTWARE IS PROVIDED AS A DEVELOPMENT SNAPSHOT TO AID
+COLLABORATION AND WAS NOT ISSUED AS A RELEASED PRODUCT BY INTEL.
diff --git a/memkind-1.10.0/README.md b/memkind-1.10.0/README.md
new file mode 120000
index 0000000..100b938
--- /dev/null
+++ b/memkind-1.10.0/README.md
@@ -0,0 +1 @@
+README
\ No newline at end of file
diff --git a/memkind-1.10.0/astyle.sh b/memkind-1.10.0/astyle.sh
new file mode 100755
index 0000000..96b801b
--- /dev/null
+++ b/memkind-1.10.0/astyle.sh
@@ -0,0 +1,51 @@
+#!/bin/bash
+#
+#  Copyright (C) 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+ASTYLE_MIN_VER="3.1"
+ASTYLE_OPT="--style=linux --indent=spaces=4 -S -r --max-continuation-indent=80 "
+ASTYLE_OPT+="--max-code-length=80 --break-after-logical --indent-namespaces -z2 "
+ASTYLE_OPT+="--align-pointer=name"
+if ! ASTYLE=$(which astyle)
+then
+    echo "Package astyle was not found. Unable to check source files format policy." >&2
+    exit 1
+fi
+ASTYLE_VER=$(astyle --version 2>&1 | awk '{print $NF}')
+if (( $(echo "$ASTYLE_VER < $ASTYLE_MIN_VER" | bc -l) ));
+then
+    echo "Minimal required version of astyle is $ASTYLE_MIN_VER. Detected version is $ASTYLE_VER" >&2
+    echo "Unable to check source files format policy." >&2
+    exit 1
+fi
+$ASTYLE $ASTYLE_OPT ./*.c,*.cpp,*.h,*.hpp --exclude=jemalloc > astyle.out
+if TEST=$(cat astyle.out | grep -c Formatted)
+then
+    cat astyle.out
+    git --no-pager diff
+    echo "Please fix style issues as shown above"
+    exit 1
+else
+    echo "Source code is compliant with format policy. No style issues were found."
+    exit 0
+fi
diff --git a/memkind-1.10.0/autogen.sh b/memkind-1.10.0/autogen.sh
new file mode 100755
index 0000000..8949097
--- /dev/null
+++ b/memkind-1.10.0/autogen.sh
@@ -0,0 +1,54 @@
+#!/bin/bash
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+set -e
+
+# If the VERSION file does not exist, then create it based on git
+# describe or if not in a git repo just set VERSION to 0.0.0.
+# Determine VERSION scheme, based on git:
+# -if HEAD is tag annotated - it is official release e.g. setting VERSION to "1.9.0"
+# -if HEAD is not tag annotated - it is snapshot e.g. setting VERSION to "1.8.0+dev19+g1c93b2d"
+
+if [ ! -f VERSION ]; then
+    if [ -f .git/config ]; then
+        sha=$(git describe --long | awk -F- '{print $(NF)}')
+        release=$(git describe --long | awk -F- '{print $(NF-1)}')
+        version=$(git describe --long | sed -e "s|\(.*\)-$release-$sha|\1|" -e "s|-|+|g" -e "s|^v||")
+        if [ ${release} != "0" ]; then
+            echo "WARNING: No annotated tag referring to this commit was found, setting version to development build " 2>&1
+            version=${version}+dev${release}+${sha}
+        else
+            echo "Annotated tag referring to this commit was found, setting version as an official release" 2>&1
+            version=${version}
+        fi
+    else
+        echo "WARNING: VERSION file does not exist and working directory is not a git repository, setting version to 0.0.0" 2>&1
+        version=0.0.0
+    fi
+    echo $version > VERSION
+fi
+
+autoreconf --install
+
diff --git a/memkind-1.10.0/autohbw/Makefile.mk b/memkind-1.10.0/autohbw/Makefile.mk
new file mode 100644
index 0000000..9d2fd12
--- /dev/null
+++ b/memkind-1.10.0/autohbw/Makefile.mk
@@ -0,0 +1,37 @@
+#
+#  Copyright (C) 2014 - 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+lib_LTLIBRARIES += autohbw/libautohbw.la \
+                   # end
+
+EXTRA_DIST += autohbw/autohbw_get_src_lines.pl
+
+autohbw_libautohbw_la_LIBADD = libmemkind.la
+
+autohbw_libautohbw_la_SOURCES = autohbw/autohbw.c
+
+clean-local: autohbw-clean
+
+autohbw-clean:
+	rm -f autohbw/*.gcno
diff --git a/memkind-1.10.0/autohbw/autohbw.c b/memkind-1.10.0/autohbw/autohbw.c
new file mode 100644
index 0000000..4bff27b
--- /dev/null
+++ b/memkind-1.10.0/autohbw/autohbw.c
@@ -0,0 +1,457 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////
+// File   : autohbw.c
+// Purpose: Library to automatically allocate HBW (MCDRAM)
+// Author : Ruchira Sasanka (ruchira DOT sasanka AT intel DOT com)
+// Date   : Jan 30, 2015
+///////////////////////////////////////////////////////////////////////////
+
+#include <memkind.h>
+
+#include <assert.h>
+#include <ctype.h>
+#include <errno.h>
+#include <stdbool.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#define AUTOHBW_EXPORT __attribute__((visibility("default")))
+#define AUTOHBW_INIT __attribute__((constructor))
+
+//-2 = nothing is printed
+//-1 = critical messages are printed
+// 0 = no log messages for allocations are printed but INFO messages are printed
+// 1 = a log message is printed for each allocation (Default)
+// 2 = a log message is printed for each allocation with a backtrace
+enum {
+    ALWAYS = -1,
+    INFO,
+    ALLOC,
+    VERBOSE
+};
+
+// Default is to print allocations
+static int LogLevel = ALLOC;
+
+// Allocations of size greater than low limit promoted to HBW memory.
+// If there is a high limit specified, allocations larger than this limit
+// will not be allocated in HBW.
+static size_t HBWLowLimit = 1 * 1024 * 1024;
+static size_t HBWHighLimit = -1ull;
+
+// Whether we have initialized HBW arena of memkind library -- by making
+// a dummy call to it. HBW arena (and hence any memkind_* call with kind
+// HBW) must NOT be used until this flag is set true.
+static bool MemkindInitDone = false;
+
+// Following is the type of HBW memory that is allocated using memkind.
+// By changing this type, this library can be used to allocate other
+// types of memory types (e.g., MEMKIND_HUGETLB, MEMKIND_GBTLB,
+// MEMKIND_HBW_HUGETLB etc.)
+static memkind_t hbw_kind;
+
+// API control for HBW allocations.
+static bool isAutoHBWEnabled = true;
+
+#define LOG(level, ...)                                                                            \
+    do {                                                                                           \
+        if (LogLevel >= level) {                                                                   \
+            fprintf(stderr, __VA_ARGS__);                                                          \
+        }                                                                                          \
+    } while (0)
+
+static bool isAllocInHBW(size_t size)
+{
+    if (!MemkindInitDone)
+        return false;
+
+    if (!isAutoHBWEnabled)
+        return false;
+
+    if (size < HBWLowLimit)
+        return false;
+
+    if (size > HBWHighLimit)
+        return false;
+
+    return true;
+}
+
+// Returns the limit in bytes using a limit value and a multiplier
+// character like K, M, G
+static size_t getLimit(size_t limit, char lchar)
+{
+    // Now read the trailing character (e.g., K, M, G)
+    // Based on the character, determine the multiplier
+    if ((limit > 0) && isalpha(lchar)) {
+        long mult = 1;
+
+        switch (toupper(lchar)) {
+            case 'G':
+                mult *= 1024;
+            case 'M':
+                mult *= 1024;
+            case 'K':
+                mult *= 1024;
+        }
+
+        // check for overflow, saturate at max
+        if (limit >= -1ull / mult)
+            return -1ull;
+
+        return limit * mult;
+    }
+
+    return limit;
+}
+
+// Once HBWLowLimit (and HBWHighLimit) are set, call this method to
+// inform the user about the size range of arrays that will be allocated
+// in HBW
+static void printLimits()
+{
+    // Inform according to the limits set
+    if ((HBWLowLimit > 0) && (HBWHighLimit < -1ull)) {
+        // if both high and low limits are specified, we use a range
+        LOG(INFO, "INFO: Allocations between %ldK - %ldK will be allocated in "
+            "HBW. Set AUTO_HBW_SIZE=X:Y to change this limit.\n",
+            HBWLowLimit / 1024, HBWHighLimit / 1024);
+    } else if (HBWLowLimit > 0) {
+        // if only a low limit is provided, use that
+        LOG(INFO, "INFO: Allocations greater than %ldK will be allocated in HBW."
+            " Set AUTO_HBW_SIZE=X:Y to change this limit.\n",
+            HBWLowLimit / 1024);
+    } else if (HBWHighLimit < -1ull) {
+        // if only a high limit is provided, use that
+        LOG(INFO, "INFO: Allocations smaller than %ldK will be allocated in HBW. "
+            "Set AUTO_HBW_SIZE=X:Y to change this limit.\n",
+            HBWHighLimit / 1024);
+    } else {
+        // none of limits is set to non-edge value, everything goes to HBW
+        LOG(INFO, "INFO: All allocation will be done in HBW.");
+    }
+}
+
+struct kind_name_t {
+    memkind_t *kind;
+    const char *name;
+};
+
+static struct kind_name_t named_kinds[] = {
+    { &MEMKIND_DEFAULT, "memkind_default" },
+    { &MEMKIND_HUGETLB, "memkind_hugetlb" },
+    { &MEMKIND_INTERLEAVE, "memkind_interleave" },
+    { &MEMKIND_HBW, "memkind_hbw" },
+    { &MEMKIND_HBW_PREFERRED, "memkind_hbw_preferred" },
+    { &MEMKIND_HBW_HUGETLB, "memkind_hbw_hugetlb" },
+    { &MEMKIND_HBW_PREFERRED_HUGETLB, "memkind_hbw_preferred_hugetlb" },
+    { &MEMKIND_HBW_GBTLB, "memkind_hbw_gbtlb" },
+    { &MEMKIND_HBW_PREFERRED_GBTLB, "memkind_hbw_preferred_gbtlb" },
+    { &MEMKIND_GBTLB, "memkind_gbtlb" },
+    { &MEMKIND_HBW_INTERLEAVE, "memkind_hbw_interleave" },
+};
+
+static memkind_t get_kind_by_name(const char *name)
+{
+    int i;
+    for (i = 0; i < sizeof(named_kinds) / sizeof(named_kinds[0]); ++i)
+        if (strcasecmp(named_kinds[i].name, name) == 0)
+            return *named_kinds[i].kind;
+
+    return 0;
+}
+
+// Read from the environment and sets global variables
+// Env variables are:
+//   AUTO_HBW_SIZE = gives the size for auto HBW allocation
+//   AUTO_HBW_LOG = gives logging level
+static void setEnvValues()
+{
+    // STEP: Read the log level from the env variable. Do this early because
+    //       printing depends on this
+    char *log_str = secure_getenv("AUTO_HBW_LOG");
+    if (log_str && strlen(log_str)) {
+        int level = atoi(log_str);
+        LogLevel = level;
+        LOG(ALWAYS, "INFO: Setting log level to %d\n", LogLevel);
+    }
+
+    if (LogLevel == INFO) {
+        LOG(INFO, "INFO: HBW allocation stats will not be printed. "
+            "Set AUTO_HBW_LOG to enable.\n");
+    } else if (LogLevel == ALLOC) {
+        LOG(INFO, "INFO: Only HBW allocations will be printed. "
+            "Set AUTO_HBW_LOG to disable/enable.\n");
+    } else if (LogLevel == VERBOSE) {
+        LOG(INFO, "INFO: HBW allocation with backtrace info will be printed. "
+            "Set AUTO_HBW_LOG to disable.\n");
+    }
+
+    // Set the memory type allocated by this library. By default, it is
+    // MEMKIND_HBW, but we can use this library to allocate other memory
+    // types
+    const char *memtype_str = secure_getenv("AUTO_HBW_MEM_TYPE");
+    if (memtype_str && strlen(memtype_str)) {
+        // Find the memkind_t using the name the user has provided in the env variable
+        memkind_t mty = get_kind_by_name(memtype_str);
+        if (mty != 0) {
+            hbw_kind = mty;
+            LOG(INFO, "INFO: Setting HBW memory type to %s\n", memtype_str);
+        } else {
+            LOG(ALWAYS, "WARN: Memory type %s not recognized. Using default type\n",
+                memtype_str);
+        }
+    }
+
+    // STEP: Set the size limits (thresholds) for HBW allocation
+    //
+    // Reads the environment variable
+    const char *size_str = secure_getenv("AUTO_HBW_SIZE");
+    if (size_str) {
+        size_t lowlim = HBWLowLimit / 1024;
+        size_t highlim = HBWHighLimit / 1024;
+        char lowC = 'K', highC = 'K';
+
+        if (size_str) {
+            char *ptr = (char *)size_str;
+            lowlim = strtoll(ptr, &ptr, 10);
+            if (*ptr != 0 && *ptr != ':')
+                lowC = *ptr++;
+            else
+                lowC = ' ';
+
+            if (*ptr++ == ':') {
+                highlim = strtoll(ptr, &ptr, 10);
+                if (*ptr)
+                    highC = *ptr;
+                else
+                    highC = ' ';
+            }
+
+            LOG(INFO, "INFO: lowlim=%zu(%c), highlim=%zu(%c)\n", lowlim, lowC, highlim,
+                highC);
+        }
+
+        HBWLowLimit = getLimit(lowlim, lowC);
+        HBWHighLimit = getLimit(highlim, highC);
+
+        if (HBWLowLimit >= HBWHighLimit) {
+            LOG(ALWAYS, "WARN: In AUTO_HBW_SIZE=X:Y, X cannot be greater or equal to Y. "
+                "None of allocations will use HBW memory.\n");
+        }
+    } else {
+        // if the user did not specify any limits, inform that we are using
+        // default limits
+        LOG(INFO, "INFO: Using default values for array size thresholds. "
+            "Set AUTO_HBW_SIZE=X:Y to change.\n");
+    }
+
+    // inform the user about limits
+    printLimits();
+}
+
+// This function is executed at library load time.
+// Initialize HBW arena by making a dummy allocation/free at library load
+// time. Until HBW initialization is complete, we must not call any
+// allocation routines with HBW as kind.
+static void AUTOHBW_INIT autohbw_load(void)
+{
+    // First set the default memory type this library allocates. This can
+    // be overridden by env variable
+    // Note: 'memkind_hbw_preferred' will allow falling back to DDR but
+    //       'memkind_hbw will not'
+    // Note: If HBM is not installed on a system, memkind_hbw_preferred call
+    //       would fail. Therefore, we need to check for availability first.
+    if (memkind_check_available(MEMKIND_HBW) != 0) {
+        LOG(ALWAYS, "WARN: *** No HBM found in system. Will use default (DDR) "
+            "OR user specified type ***\n");
+        hbw_kind = MEMKIND_DEFAULT;
+    } else {
+        hbw_kind = MEMKIND_HBW_PREFERRED;
+    }
+
+    // Read any env variables. This has to be done first because DbgLevel
+    // is set using env variables and debug printing is used below
+    setEnvValues(); // read any env variables
+
+    LOG(INFO, "INFO: autohbw.so loaded!\n");
+
+    // dummy HBW call to initialize HBW arena
+    void *pp = memkind_malloc(hbw_kind, 16);
+    if (pp == 0) {
+        LOG(ALWAYS, "\t-HBW init call FAILED. "
+            "Is required memory type present on your system?\n");
+        abort();
+    }
+
+    LOG(ALWAYS, "\t-HBW int call succeeded\n");
+    memkind_free(hbw_kind, pp);
+
+    MemkindInitDone = true; // enable HBW allocation
+}
+
+static void *MemkindMalloc(size_t size)
+{
+    LOG(VERBOSE, "In my memkind malloc sz:%ld ... ", size);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    void *ptr = memkind_malloc(kind, size);
+
+    LOG(VERBOSE, "\tptr:%p\n", ptr);
+    return ptr;
+}
+
+static void *MemkindCalloc(size_t nmemb, size_t size)
+{
+    LOG(VERBOSE, "In my memkind calloc sz:%ld ..", size * nmemb);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    void *ptr = memkind_calloc(kind, nmemb, size);
+
+    LOG(VERBOSE, "\tptr:%p\n", ptr);
+    return ptr;
+}
+
+static void *MemkindRealloc(void *ptr, size_t size)
+{
+    LOG(VERBOSE, "In my memkind realloc sz:%ld, p1:%p ..", size, ptr);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    void *nptr = memkind_realloc(kind, ptr, size);
+
+    LOG(VERBOSE, "\tptr=%p\n", nptr);
+    return nptr;
+}
+
+static int MemkindAlign(void **memptr, size_t alignment, size_t size)
+{
+    LOG(VERBOSE, "In my memkind align sz:%ld .. ", size);
+
+    bool useHbw = isAllocInHBW(size);
+    memkind_t kind = useHbw ? hbw_kind : MEMKIND_DEFAULT;
+
+    if (useHbw)
+        LOG(VERBOSE, "\tHBW");
+
+    int ret = memkind_posix_memalign(kind, memptr, alignment, size);
+
+    LOG(VERBOSE, "\tptr:%p\n", *memptr);
+    return ret;
+}
+
+// memkind_free does not need the exact kind, if kind is 0. Then
+// the library can figure out the proper kind itself.
+static void MemkindFree(void *ptr)
+{
+    // avoid to many useless logs
+    if (ptr)
+        LOG(VERBOSE, "In my memkind free, ptr:%p\n", ptr);
+    memkind_free(0, ptr);
+}
+
+//--------------------------------------------------------------------------
+// ------------------ Public API of autohbw           ----------------------
+//--------------------------------------------------------------------------
+
+AUTOHBW_EXPORT void enableAutoHBW()
+{
+    isAutoHBWEnabled = true;
+    LOG(INFO, "INFO: HBW allocations enabled by application (for this rank)\n");
+}
+
+AUTOHBW_EXPORT void disableAutoHBW()
+{
+    isAutoHBWEnabled = false;
+    LOG(INFO, "INFO: HBW allocations disabled by application (for this rank)\n");
+}
+
+AUTOHBW_EXPORT void *malloc(size_t size)
+{
+    return MemkindMalloc(size);
+}
+
+AUTOHBW_EXPORT void *calloc(size_t nmemb, size_t size)
+{
+    return MemkindCalloc(nmemb, size);
+}
+
+AUTOHBW_EXPORT void *realloc(void *ptr, size_t size)
+{
+    return MemkindRealloc(ptr, size);
+}
+
+AUTOHBW_EXPORT int posix_memalign(void **memptr, size_t alignment, size_t size)
+{
+    return MemkindAlign(memptr, alignment, size);
+}
+
+// Warn about deprecated function usage.
+AUTOHBW_EXPORT void *valloc(size_t size)
+{
+    LOG(ALWAYS, "use of deprecated valloc. Use posix_memalign instead\n");
+    void *memptr = 0;
+    size_t boundary = sysconf(_SC_PAGESIZE);
+    int status = MemkindAlign(&memptr, boundary, size);
+    if (status == 0 && memptr != 0)
+        return memptr;
+
+    return 0;
+}
+
+// Warn about deprecated function usage.
+AUTOHBW_EXPORT void *memalign(size_t boundary, size_t size)
+{
+    LOG(ALWAYS, "use of deprecated memalign. Use posix_memalign instead\n");
+    void *memptr = 0;
+    int status = MemkindAlign(&memptr, boundary, size);
+    if (status == 0 && memptr != 0)
+        return memptr;
+
+    return 0;
+}
+
+AUTOHBW_EXPORT void free(void *ptr)
+{
+    return MemkindFree(ptr);
+}
diff --git a/memkind-1.10.0/autohbw/autohbw_README b/memkind-1.10.0/autohbw/autohbw_README
new file mode 100644
index 0000000..cf23a0f
--- /dev/null
+++ b/memkind-1.10.0/autohbw/autohbw_README
@@ -0,0 +1,92 @@
+README for auothbw library
+Ruchira Sasanka (ruchira DOT sasanka AT intel DOT com)
+2015 April 8
+
+PURPOSE
+-------
+The autohbw library is used to automatically allocate high-bandwidth (HBW)
+memory without any modifications to source code of your application.
+This library intercepts the standard heap allocations (e.g., malloc, calloc)
+in your program so that it can serve those allocations out of HBW memory.
+
+There are two libraries installed at the same location as memkind library:
+  1) libautohbw.so -- dynamic library
+  2) libautohbw.a  -- static library (provided for static linking)
+
+USAGE
+-----
+To use the dynamic library, insert LD_PRELOAD=libautohbw.so and libmekind.so
+before your commandline. For instance,
+
+   LD_PRELOAD=libautohbw.so /bin/ls
+
+executes Unix ls utility with autohbw library. You can execute the
+above command on the prompt or put it in a shell script. An example of
+the latter approach is given in ./autohbw_test.sh. Make sure that
+LD_LIBRARY_PATH environment variable contains the path to libautohbw.so
+and libmekind.so before you exeute the above command.
+
+To use the statically linked library, insert -lautohbw -lmemkind
+on your link line. You will have to provide the library path with -L to
+point to the above libraries. If you are linking your program statically
+(e.g., with -static or -fast flags in ifort/icc), you must use the static
+version of autohbw library (libautohbw.a).
+
+It is possible to temporarily disable/enable automatic HBM allocations by
+calling disableAutoHBW() and enableAutoHBW() in your source code. To call
+these routines, please include autohbw_api.h header file and link with
+-lautohbw.
+
+
+Usage with MPI programs
+-----------------------
+To use with MPI programs, you need to create a shell script to do the
+LD_PRELOAD. E.g.,
+   mpirun -n 2 ./autohbw_test.sh
+
+ENVIRONMENT VARIABLES
+---------------------
+The following environment variables control the behavior of the autohbw
+library:
+
+  AUTO_HBW_SIZE=x[:y]
+  Indicates that any allocation larger than x and smaller than y should be
+  allocated in HBW memory. x and y can be followed by a K, M, or G to
+  indicate the size in Kilo/Memga/Giga bytes (e.g., 4K, 3M, 2G).
+  Examples:
+    AUTO_HBW_SIZE=4K     # all allocation larger than 4K allocated in HBW
+    AUTO_HBW_SIZE=1M:5M  # allocations between 1M and 5M allocated in HBW
+
+
+  AUTO_HBW_LOG=level
+  Sets the value of logging (printing) level. If level is:
+   -1 = no messages are printed
+    0 = no log messages for allocations are printed but INFO messages
+        are printed
+    1 = a log message is printed for each allocation (Default)
+    2 = a log message is printed for each allocation with a backtrace
+        Redirect this output and use autohbw_get_src_lines.pl to find
+        source lines for each allocation. Your application must be compiled
+        with -g to see source lines.
+  Notes:
+    1. Logging adds extra overhead. Therefore, performance critical runs,
+       logging level should be 0
+    2. The amount of free memory printed with log messages is only
+       approximate -- e.g. pages that are not touched yet are excluded
+  Examples:
+    AUTO_HBW_LOG=1
+
+
+  AUTO_HBW_MEM_TYPE=memory_type
+  Sets the type of memory type that should be automatically allocated. By
+  default, this type is MEMKIND_HBW_PREFERRED, if MCDRAM is found in your
+  system; otherwise, the default is MEMKIND_DEFAULT. The names of memory
+  types are definedin memkind(3) man page. If you are requesting any huge
+  TLB pages, pleasemake sure that the requested type is currently enabled
+  in your OS.
+  Examples:
+    AUTO_HBW_MEM_TYPE=MEMKIND_HBW_PREFERRED   (default, if MCDRAM present)
+    AUTO_HBW_MEM_TYPE=MEMKIND_DEFAULT         (default, if MCDRAM absent)
+    AUTO_HBW_MEM_TYPE=MEMKIND_HBW_HUGETLB
+    AUTO_HBW_MEM_TYPE=MEMKIND_HUGETLB
+
diff --git a/memkind-1.10.0/autohbw/autohbw_api.h b/memkind-1.10.0/autohbw/autohbw_api.h
new file mode 100644
index 0000000..45f666a
--- /dev/null
+++ b/memkind-1.10.0/autohbw/autohbw_api.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (C) 2015 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////
+// File   : autohbw_api.h
+// Purpose: Header file to include, to use API calls to AutoHBW
+// Author : Ruchira Sasanka (ruchira.sasanka AT intel.com)
+// Date   : Jan 30, 2015
+///////////////////////////////////////////////////////////////////////////
+
+#ifndef AUTOHBW_API_H
+#define AUTOHBW_API_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+// Temporarily enable HBM allocations
+void enableAutoHBW();
+
+// Temporarily disable HBM allocations
+void disableAutoHBW();
+
+#ifdef __cplusplus
+}
+#endif
+
+#endif
diff --git a/memkind-1.10.0/autohbw/autohbw_get_src_lines.pl b/memkind-1.10.0/autohbw/autohbw_get_src_lines.pl
new file mode 100644
index 0000000..81ccb67
--- /dev/null
+++ b/memkind-1.10.0/autohbw/autohbw_get_src_lines.pl
@@ -0,0 +1,106 @@
+#!/usr/bin/perl
+#
+#  Copyright (C) 2015 - 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+use strict;
+
+my $usage = "Usage: get_autohbw_srclines.pl output_log_of_AutoHBW executable";
+
+# Check for 2 arguments
+#
+if (@ARGV != 2) {
+    print $usage, "\n";
+    exit;
+}
+
+# Read the command line arguments
+#
+my $LogF = shift @ARGV;
+my $BinaryF = shift @ARGV;
+
+&main();
+
+sub main {
+
+
+    print("Info: Reading AutoHBW log from: $LogF\n");
+    print("Info: Binary file: $BinaryF\n");
+
+    # open the log file produced by AutoHBM and look at lines starting
+    # with Log
+    open LOGF, "grep Log $LogF |" or die "Can't open log file $LogF";
+
+    my $line;
+
+    # Read each log line
+    #
+    while ($line = <LOGF>) {
+
+        # if the line contain 3 backtrace addresses, try to find the source
+        # lines for them
+        #
+        if ($line =~ /^(Log:.*)Backtrace:.*0x([0-9a-f]+).*0x([0-9a-f]+).*0x([0-9a-f]+)/ ) {
+
+            #  Read the pointers
+            #
+            my @ptrs;
+
+            my $pre = $1;
+
+            $ptrs[0] = $2;
+            $ptrs[1] = $3;
+            $ptrs[2] = $4;
+
+            # prints the first portion of the line
+            #
+            print $pre, "\n";
+
+            # for each of the pointers, lookup its source line using
+            # addr2line and print the src line(s) if found
+            #
+            my $i=0;
+            for ($i=0; $i < @ptrs; $i++) {
+
+                my $addr = $ptrs[$i];
+
+                open SRCL, "addr2line -e $BinaryF 0x$addr |"
+                    or die "addr2line fail";
+
+
+                my $srcl = <SRCL>;
+
+                if ($srcl =~ /^\?/) {
+
+                } else {
+
+                    print "\t- Src: $srcl";
+                }
+
+            }
+
+        }
+
+    }
+
+}
diff --git a/memkind-1.10.0/autohbw/autohbw_test.sh b/memkind-1.10.0/autohbw/autohbw_test.sh
new file mode 100755
index 0000000..52cf313
--- /dev/null
+++ b/memkind-1.10.0/autohbw/autohbw_test.sh
@@ -0,0 +1,37 @@
+#!/bin/bash
+
+#  Copyright (C) 2015 - 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# The following is needed if your system does not really have HBW nodes.
+# Please see 'man memkind' for a description of this variable
+#
+# export MEMKIND_HBW_NODES=0
+
+# Set any autohbw env variables
+#
+export AUTO_HBW_SIZE=4K
+
+# Make sure you have set LD_LIBRARY_PATH to lib directory with libautohbw.so
+# and libmemkind.so
+#
+LD_PRELOAD=libautohbw.so:libmemkind.so /bin/ls
diff --git a/memkind-1.10.0/build.sh b/memkind-1.10.0/build.sh
new file mode 100755
index 0000000..6dfd578
--- /dev/null
+++ b/memkind-1.10.0/build.sh
@@ -0,0 +1,35 @@
+#!/bin/bash
+#
+#  Copyright (C) 2016 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+set -e
+
+cd $(dirname $0)
+EXTRA_CONF=$@
+
+./autogen.sh
+./configure $EXTRA_CONF
+
+#use V=1 for full cmdlines of build
+make all -j`nproc`
+make checkprogs -j`nproc`
diff --git a/memkind-1.10.0/configure.ac b/memkind-1.10.0/configure.ac
new file mode 100644
index 0000000..95116a1
--- /dev/null
+++ b/memkind-1.10.0/configure.ac
@@ -0,0 +1,234 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+#                                               -*- Autoconf -*-
+# Process this file with autoconf to produce a configure script.
+
+AC_PREREQ([2.63])
+AC_INIT([memkind],m4_esyscmd([tr -d '\n' < VERSION]))
+
+AC_CONFIG_MACRO_DIR([m4])
+AC_CONFIG_HEADERS([config.h])
+AC_CONFIG_SRCDIR([memkind.spec.mk])
+
+AM_INIT_AUTOMAKE([-Wall -Werror foreign 1.11 silent-rules subdir-objects parallel-tests tar-pax])
+AM_SILENT_RULES([yes])
+
+# Checks for programs and libraries.
+m4_ifdef([AM_PROG_AR], [AM_PROG_AR])
+AC_PROG_CXX
+AC_PROG_CC
+AC_OPENMP
+AC_CHECK_LIB(numa, numa_available, [], [AC_MSG_ERROR([libnuma is required dependency])])
+AX_PTHREAD([LIBS="$PTHREAD_LIBS $LIBS" CFLAGS="$CFLAGS $PTHREAD_CFLAGS" CC="$PTHREAD_CC"],
+           [AC_MSG_ERROR([pthreads are required dependency])])
+
+AM_PROG_CC_C_O
+
+#============================tls===============================================
+# Check for thread local storage support
+AC_ARG_ENABLE([tls],
+  [AS_HELP_STRING([--enable-tls], [Enable thread-local storage (__thread keyword)])],
+[if test "x$enable_tls" = "xyes" ; then
+  enable_tls="1"
+else
+  enable_tls="0"
+fi
+],
+[enable_tls="0"]
+)
+if test "x${enable_tls}" = "x1" ; then
+AC_MSG_CHECKING([for TLS support])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+    __thread int x;
+]], [[
+    x = 1234;
+]])],
+    AC_MSG_RESULT([yes]),
+    AC_MSG_RESULT([no])
+    enable_tls="0")
+fi
+if test "x${enable_tls}" = "x1" ; then
+  AC_DEFINE([MEMKIND_TLS], [ ], [Enables TLS usage for mapping arenas to threads])
+fi
+AC_SUBST([enable_tls])
+
+#============================decorators========================================
+AC_ARG_ENABLE([decorators],
+  [AS_HELP_STRING([--enable-decorators], [Enable decorators])],
+[if test "x$enable_decorators" = "xyes" ; then
+  enable_decorators="1"
+else
+  enable_decorators="0"
+fi
+],
+[enable_decorators="0"]
+)
+if test "x${enable_decorators}" = "x1" ; then
+  AC_DEFINE([MEMKIND_DECORATION_ENABLED], [ ], [Enables decorators])
+fi
+AC_SUBST([enable_decorators])
+
+#============================debug=============================================
+AC_ARG_ENABLE([debug],
+  [AS_HELP_STRING([--enable-debug], [Build debugging code and compile with -O0 -g])],
+[if test "x$enable_debug" = "xno" ; then
+  enable_debug="0"
+else
+  enable_debug="1"
+fi
+],
+[enable_debug="0"]
+)
+if test "x$enable_debug" = "x1" ; then
+  AC_DEFINE([MEMKIND_DEBUG], [ ], [Enables code for debugging])
+  CFLAGS="$CFLAGS -O0 -g"
+  CXXFLAGS="$CXXFLAGS -O0 -g"
+fi
+AC_SUBST([enable_debug])
+
+#============================gcov==============================================
+AC_ARG_ENABLE([gcov],
+  [AS_HELP_STRING([--enable-gcov], [Build code with gcov instructions])],
+[if test "x$enable_gcov" = "xno" ; then
+  enable_gcov="0"
+else
+  enable_gcov="1"
+fi
+],
+[enable_gcov="0"]
+)
+if test "x$enable_gcov" = "x1" ; then
+  CFLAGS="$CFLAGS -O0 -fprofile-arcs -ftest-coverage"
+  CXXFLAGS="$CXXFLAGS -O0 -fprofile-arcs -ftest-coverage"
+fi
+AC_SUBST([enable_gcov])
+
+#============================secure_flags======================================
+AC_ARG_ENABLE([secure],
+  [AS_HELP_STRING([--enable-secure], [Build library with security enchantments])],
+[if test "x$enable_secure" = "xno" ; then
+  enable_secure="0"
+else
+  enable_secure="1"
+fi
+],
+[enable_secure="1"]
+)
+if test "x$enable_secure" = "x1" ; then
+  CFLAGS="$CFLAGS -fstack-protector"
+  LDFLAGS="$LDFLAGS -Wl,-z,relro,-z,now"
+
+  if test "$CFLAGS" != "${CFLAGS%-O0*}" ; then # if CFLAGS contains -O0
+      echo "WARNING: Could not apply FORTIFY_SOURCE=2 due to lack of optimization (-O0)"
+  else
+      CFLAGS="$CFLAGS -D_FORTIFY_SOURCE=2" #FORTITFY_SOURCE does not work with -O0 (ex. if enable_debug=1 or enable_gcov=1)
+  fi
+fi
+
+AC_SUBST([enable_secure])
+
+#============================tcache_max_size_class======================
+
+tcache_max_size_class=12;
+AC_DEFINE_UNQUOTED([JEMALLOC_TCACHE_CLASS], $tcache_max_size_class,
+  [Maximum size class (log base 2) to cache in jemalloc tcache])
+
+AC_SUBST(tcache_max_size_class)
+
+#============================auto_arenas================================
+auto_arenas=256;
+AC_DEFINE_UNQUOTED([ARENA_LIMIT_DEFAULT_KIND], $auto_arenas,
+  [Maximum number of automatic managed arenas by jemalloc (used by MEMKIND_DEFAULT)])
+
+AC_SUBST(auto_arenas)
+
+#============================arena_limit=======================================
+arena_limit=256;
+AC_ARG_VAR(ARENA_LIMIT,
+  [Upper bound for number of arenas per kind, if set to 0 then no limit]
+)
+if test "$ARENA_LIMIT" != "" ; then
+  arena_limit=$ARENA_LIMIT;
+fi
+
+AC_DEFINE_UNQUOTED([ARENA_LIMIT_PER_KIND], $arena_limit, [Upper bound for number of arenas per kind])
+
+#============================memkind_prefix=======================================
+memkind_prefix=jemk_;
+AC_ARG_VAR(MEMKIND_PREFIX,
+  [Prefix used for all jemalloc public API under memkind, default value jemk_]
+)
+if test "$MEMKIND_PREFIX" != "" ; then
+  memkind_prefix=$MEMKIND_PREFIX;
+fi
+
+AC_SUBST(memkind_prefix)
+
+#===============================daxctl=========================================
+AC_ARG_ENABLE([daxctl],
+  [AS_HELP_STRING([--enable-daxctl], [Build library with daxctl support])],
+[if test "x$enable_daxctl" = "xno" ; then
+  enable_daxctl="0"
+else
+  enable_daxctl="1"
+fi
+],
+[enable_daxctl="1"]
+)
+if test "x$enable_daxctl" = "x1" ; then
+    AC_CHECK_LIB(daxctl, daxctl_dev_get_memory, [daxctl_kmem=yes], [daxctl_kmem=no])
+fi
+if test $daxctl_kmem = "yes" ; then
+    AC_DEFINE([MEMKIND_DAXCTL_KMEM], [1], [Automatic recognition of PMEM as system memory in MEMKIND_DAX_KMEM])
+    AC_SUBST(DAXCTL_LIBS, [-ldaxctl])
+    LDFLAGS="$LDFLAGS -ldaxctl"
+fi
+AC_SUBST([enable_daxctl])
+
+#============================cxx11=============================================
+
+AX_CXX_COMPILE_STDCXX_11([noext], [optional])
+AM_CONDITIONAL([HAVE_CXX11], [test "x$HAVE_CXX11" = x1])
+
+LT_PREREQ([2.2])
+LT_INIT
+
+AC_CONFIG_FILES([Makefile memkind.pc])
+
+AC_OUTPUT
+AC_MSG_RESULT([=================================================================================])
+AC_MSG_RESULT([
+               Memkind version $VERSION])
+if test $daxctl_kmem = "yes" ; then
+AC_MSG_RESULT([
+               Automatic recognition of PMEM NUMA in MEMKIND_DAX_KMEM is enabled])
+else
+AC_MSG_RESULT([
+               Automatic recognition of PMEM NUMA in MEMKIND_DAX_KMEM is disabled
+               To enable this feature please install libdaxctl-devel v66 or later
+               and run configure again])
+fi
+AC_MSG_RESULT([=================================================================================])
diff --git a/memkind-1.10.0/copying_headers/MANIFEST.EXEMPT b/memkind-1.10.0/copying_headers/MANIFEST.EXEMPT
new file mode 100644
index 0000000..0e6326d
--- /dev/null
+++ b/memkind-1.10.0/copying_headers/MANIFEST.EXEMPT
@@ -0,0 +1,347 @@
+.codecov.yml
+.gitignore
+.travis.yml
+AUTHORS
+CONTRIBUTING
+COPYING
+ChangeLog
+NEWS
+PULL_REQUEST_TEMPLATE.md
+README
+VERSION
+autohbw/autohbw_README
+examples/README
+include/memkind/internal/tbb_mem_pool_policy.h
+jemalloc/.appveyor.yml
+jemalloc/.autom4te.cfg
+jemalloc/.gitattributes
+jemalloc/.gitignore
+jemalloc/.travis.yml
+jemalloc/autogen.sh
+jemalloc/bin/jemalloc-config.in
+jemalloc/bin/jemalloc.sh.in
+jemalloc/bin/jeprof.in
+jemalloc/build-aux/config.guess
+jemalloc/build-aux/config.sub
+jemalloc/build-aux/install-sh
+jemalloc/ChangeLog
+jemalloc/config.stamp.in
+jemalloc/configure.ac
+jemalloc/COPYING
+jemalloc/doc/html.xsl.in
+jemalloc/doc/jemalloc.xml.in
+jemalloc/doc/manpages.xsl.in
+jemalloc/doc/stylesheet.xsl
+jemalloc/include/jemalloc/internal/arena_externs.h
+jemalloc/include/jemalloc/internal/arena_inlines_a.h
+jemalloc/include/jemalloc/internal/arena_inlines_b.h
+jemalloc/include/jemalloc/internal/arena_stats.h
+jemalloc/include/jemalloc/internal/arena_structs_a.h
+jemalloc/include/jemalloc/internal/arena_structs_b.h
+jemalloc/include/jemalloc/internal/arena_types.h
+jemalloc/include/jemalloc/internal/assert.h
+jemalloc/include/jemalloc/internal/atomic.h
+jemalloc/include/jemalloc/internal/atomic_c11.h
+jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
+jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
+jemalloc/include/jemalloc/internal/atomic_msvc.h
+jemalloc/include/jemalloc/internal/background_thread_externs.h
+jemalloc/include/jemalloc/internal/background_thread_inlines.h
+jemalloc/include/jemalloc/internal/background_thread_structs.h
+jemalloc/include/jemalloc/internal/base_externs.h
+jemalloc/include/jemalloc/internal/base_inlines.h
+jemalloc/include/jemalloc/internal/base_structs.h
+jemalloc/include/jemalloc/internal/base_types.h
+jemalloc/include/jemalloc/internal/bin.h
+jemalloc/include/jemalloc/internal/bin_stats.h
+jemalloc/include/jemalloc/internal/bin_types.h
+jemalloc/include/jemalloc/internal/bitmap.h
+jemalloc/include/jemalloc/internal/bit_util.h
+jemalloc/include/jemalloc/internal/cache_bin.h
+jemalloc/include/jemalloc/internal/ckh.h
+jemalloc/include/jemalloc/internal/ctl.h
+jemalloc/include/jemalloc/internal/div.h
+jemalloc/include/jemalloc/internal/emitter.h
+jemalloc/include/jemalloc/internal/extent_dss.h
+jemalloc/include/jemalloc/internal/extent_externs.h
+jemalloc/include/jemalloc/internal/extent_inlines.h
+jemalloc/include/jemalloc/internal/extent_mmap.h
+jemalloc/include/jemalloc/internal/extent_structs.h
+jemalloc/include/jemalloc/internal/extent_types.h
+jemalloc/include/jemalloc/internal/hash.h
+jemalloc/include/jemalloc/internal/hook.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
+jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
+jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
+jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
+jemalloc/include/jemalloc/internal/large_externs.h
+jemalloc/include/jemalloc/internal/log.h
+jemalloc/include/jemalloc/internal/malloc_io.h
+jemalloc/include/jemalloc/internal/mutex.h
+jemalloc/include/jemalloc/internal/mutex_pool.h
+jemalloc/include/jemalloc/internal/mutex_prof.h
+jemalloc/include/jemalloc/internal/nstime.h
+jemalloc/include/jemalloc/internal/pages.h
+jemalloc/include/jemalloc/internal/ph.h
+jemalloc/include/jemalloc/internal/private_namespace.sh
+jemalloc/include/jemalloc/internal/private_symbols.sh
+jemalloc/include/jemalloc/internal/prng.h
+jemalloc/include/jemalloc/internal/prof_externs.h
+jemalloc/include/jemalloc/internal/prof_inlines_a.h
+jemalloc/include/jemalloc/internal/prof_inlines_b.h
+jemalloc/include/jemalloc/internal/prof_structs.h
+jemalloc/include/jemalloc/internal/prof_types.h
+jemalloc/include/jemalloc/internal/public_namespace.sh
+jemalloc/include/jemalloc/internal/public_unnamespace.sh
+jemalloc/include/jemalloc/internal/ql.h
+jemalloc/include/jemalloc/internal/qr.h
+jemalloc/include/jemalloc/internal/quantum.h
+jemalloc/include/jemalloc/internal/rb.h
+jemalloc/include/jemalloc/internal/rtree.h
+jemalloc/include/jemalloc/internal/rtree_tsd.h
+jemalloc/include/jemalloc/internal/safety_check.h
+jemalloc/include/jemalloc/internal/sc.h
+jemalloc/include/jemalloc/internal/seq.h
+jemalloc/include/jemalloc/internal/smoothstep.h
+jemalloc/include/jemalloc/internal/smoothstep.sh
+jemalloc/include/jemalloc/internal/spin.h
+jemalloc/include/jemalloc/internal/stats.h
+jemalloc/include/jemalloc/internal/sz.h
+jemalloc/include/jemalloc/internal/tcache_externs.h
+jemalloc/include/jemalloc/internal/tcache_inlines.h
+jemalloc/include/jemalloc/internal/tcache_structs.h
+jemalloc/include/jemalloc/internal/tcache_types.h
+jemalloc/include/jemalloc/internal/test_hooks.h
+jemalloc/include/jemalloc/internal/ticker.h
+jemalloc/include/jemalloc/internal/tsd.h
+jemalloc/include/jemalloc/internal/tsd_generic.h
+jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
+jemalloc/include/jemalloc/internal/tsd_tls.h
+jemalloc/include/jemalloc/internal/tsd_types.h
+jemalloc/include/jemalloc/internal/tsd_win.h
+jemalloc/include/jemalloc/internal/util.h
+jemalloc/include/jemalloc/internal/witness.h
+jemalloc/include/jemalloc/jemalloc.sh
+jemalloc/include/jemalloc/jemalloc_defs.h.in
+jemalloc/include/jemalloc/jemalloc_macros.h.in
+jemalloc/include/jemalloc/jemalloc_mangle.sh
+jemalloc/include/jemalloc/jemalloc_protos.h.in
+jemalloc/include/jemalloc/jemalloc_rename.sh
+jemalloc/include/jemalloc/jemalloc_typedefs.h.in
+jemalloc/include/msvc_compat/C99/stdbool.h
+jemalloc/include/msvc_compat/C99/stdint.h
+jemalloc/include/msvc_compat/strings.h
+jemalloc/include/msvc_compat/windows_extra.h
+jemalloc/INSTALL.md
+jemalloc/jemalloc.pc.in
+jemalloc/m4/ax_cxx_compile_stdcxx.m4
+jemalloc/Makefile.in
+jemalloc/msvc/jemalloc_vc2015.sln
+jemalloc/msvc/jemalloc_vc2017.sln
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
+jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
+jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
+jemalloc/msvc/ReadMe.txt
+jemalloc/msvc/test_threads/test_threads.cpp
+jemalloc/msvc/test_threads/test_threads.h
+jemalloc/msvc/test_threads/test_threads_main.cpp
+jemalloc/README
+jemalloc/run_tests.sh
+jemalloc/scripts/gen_run_tests.py
+jemalloc/scripts/gen_travis.py
+jemalloc/src/arena.c
+jemalloc/src/background_thread.c
+jemalloc/src/base.c
+jemalloc/src/bin.c
+jemalloc/src/bitmap.c
+jemalloc/src/ckh.c
+jemalloc/src/ctl.c
+jemalloc/src/div.c
+jemalloc/src/extent.c
+jemalloc/src/extent_dss.c
+jemalloc/src/extent_mmap.c
+jemalloc/src/hash.c
+jemalloc/src/hook.c
+jemalloc/src/jemalloc.c
+jemalloc/src/jemalloc_cpp.cpp
+jemalloc/src/large.c
+jemalloc/src/log.c
+jemalloc/src/malloc_io.c
+jemalloc/src/mutex.c
+jemalloc/src/mutex_pool.c
+jemalloc/src/nstime.c
+jemalloc/src/pages.c
+jemalloc/src/prng.c
+jemalloc/src/prof.c
+jemalloc/src/rtree.c
+jemalloc/src/safety_check.c
+jemalloc/src/sc.c
+jemalloc/src/stats.c
+jemalloc/src/sz.c
+jemalloc/src/tcache.c
+jemalloc/src/test_hooks.c
+jemalloc/src/ticker.c
+jemalloc/src/tsd.c
+jemalloc/src/witness.c
+jemalloc/src/zone.c
+jemalloc/test/include/test/btalloc.h
+jemalloc/test/include/test/extent_hooks.h
+jemalloc/test/include/test/jemalloc_test.h.in
+jemalloc/test/include/test/jemalloc_test_defs.h.in
+jemalloc/test/include/test/math.h
+jemalloc/test/include/test/mq.h
+jemalloc/test/include/test/mtx.h
+jemalloc/test/include/test/SFMT-alti.h
+jemalloc/test/include/test/SFMT-params.h
+jemalloc/test/include/test/SFMT-params607.h
+jemalloc/test/include/test/SFMT-params1279.h
+jemalloc/test/include/test/SFMT-params2281.h
+jemalloc/test/include/test/SFMT-params4253.h
+jemalloc/test/include/test/SFMT-params11213.h
+jemalloc/test/include/test/SFMT-params19937.h
+jemalloc/test/include/test/SFMT-params44497.h
+jemalloc/test/include/test/SFMT-params86243.h
+jemalloc/test/include/test/SFMT-params132049.h
+jemalloc/test/include/test/SFMT-params216091.h
+jemalloc/test/include/test/SFMT-sse2.h
+jemalloc/test/include/test/SFMT.h
+jemalloc/test/include/test/test.h
+jemalloc/test/include/test/thd.h
+jemalloc/test/include/test/timer.h
+jemalloc/test/integration/aligned_alloc.c
+jemalloc/test/integration/allocated.c
+jemalloc/test/integration/extent.c
+jemalloc/test/integration/extent.sh
+jemalloc/test/integration/malloc.c
+jemalloc/test/integration/mallocx.c
+jemalloc/test/integration/mallocx.sh
+jemalloc/test/integration/MALLOCX_ARENA.c
+jemalloc/test/integration/overflow.c
+jemalloc/test/integration/posix_memalign.c
+jemalloc/test/integration/rallocx.c
+jemalloc/test/integration/sdallocx.c
+jemalloc/test/integration/slab_sizes.c
+jemalloc/test/integration/slab_sizes.sh
+jemalloc/test/integration/smallocx.c
+jemalloc/test/integration/smallocx.sh
+jemalloc/test/integration/thread_arena.c
+jemalloc/test/integration/thread_tcache_enabled.c
+jemalloc/test/integration/xallocx.c
+jemalloc/test/integration/xallocx.sh
+jemalloc/test/src/btalloc.c
+jemalloc/test/src/btalloc_0.c
+jemalloc/test/src/btalloc_1.c
+jemalloc/test/src/math.c
+jemalloc/test/src/mq.c
+jemalloc/test/src/mtx.c
+jemalloc/test/src/SFMT.c
+jemalloc/test/src/test.c
+jemalloc/test/src/thd.c
+jemalloc/test/src/timer.c
+jemalloc/test/stress/hookbench.c
+jemalloc/test/stress/microbench.c
+jemalloc/test/test.sh.in
+jemalloc/test/unit/a0.c
+jemalloc/test/unit/arena_reset.c
+jemalloc/test/unit/arena_reset_prof.c
+jemalloc/test/unit/arena_reset_prof.sh
+jemalloc/test/unit/atomic.c
+jemalloc/test/unit/background_thread.c
+jemalloc/test/unit/background_thread_enable.c
+jemalloc/test/unit/base.c
+jemalloc/test/unit/binshard.c
+jemalloc/test/unit/binshard.sh
+jemalloc/test/unit/bitmap.c
+jemalloc/test/unit/bit_util.c
+jemalloc/test/unit/ckh.c
+jemalloc/test/unit/decay.c
+jemalloc/test/unit/decay.sh
+jemalloc/test/unit/div.c
+jemalloc/test/unit/emitter.c
+jemalloc/test/unit/extent_quantize.c
+jemalloc/test/unit/extent_util.c
+jemalloc/test/unit/fork.c
+jemalloc/test/unit/hash.c
+jemalloc/test/unit/hook.c
+jemalloc/test/unit/huge.c
+jemalloc/test/unit/junk.c
+jemalloc/test/unit/junk.sh
+jemalloc/test/unit/junk_alloc.c
+jemalloc/test/unit/junk_alloc.sh
+jemalloc/test/unit/junk_free.c
+jemalloc/test/unit/junk_free.sh
+jemalloc/test/unit/log.c
+jemalloc/test/unit/mallctl.c
+jemalloc/test/unit/malloc_io.c
+jemalloc/test/unit/math.c
+jemalloc/test/unit/mq.c
+jemalloc/test/unit/mtx.c
+jemalloc/test/unit/nstime.c
+jemalloc/test/unit/pack.c
+jemalloc/test/unit/pack.sh
+jemalloc/test/unit/pages.c
+jemalloc/test/unit/ph.c
+jemalloc/test/unit/prng.c
+jemalloc/test/unit/prof_accum.c
+jemalloc/test/unit/prof_accum.sh
+jemalloc/test/unit/prof_active.c
+jemalloc/test/unit/prof_active.sh
+jemalloc/test/unit/prof_gdump.c
+jemalloc/test/unit/prof_gdump.sh
+jemalloc/test/unit/prof_idump.c
+jemalloc/test/unit/prof_idump.sh
+jemalloc/test/unit/prof_log.c
+jemalloc/test/unit/prof_log.sh
+jemalloc/test/unit/prof_reset.c
+jemalloc/test/unit/prof_reset.sh
+jemalloc/test/unit/prof_tctx.c
+jemalloc/test/unit/prof_tctx.sh
+jemalloc/test/unit/prof_thread_name.c
+jemalloc/test/unit/prof_thread_name.sh
+jemalloc/test/unit/ql.c
+jemalloc/test/unit/qr.c
+jemalloc/test/unit/rb.c
+jemalloc/test/unit/retained.c
+jemalloc/test/unit/rtree.c
+jemalloc/test/unit/safety_check.c
+jemalloc/test/unit/safety_check.sh
+jemalloc/test/unit/sc.c
+jemalloc/test/unit/seq.c
+jemalloc/test/unit/SFMT.c
+jemalloc/test/unit/size_classes.c
+jemalloc/test/unit/slab.c
+jemalloc/test/unit/smoothstep.c
+jemalloc/test/unit/spin.c
+jemalloc/test/unit/stats.c
+jemalloc/test/unit/stats_print.c
+jemalloc/test/unit/test_hooks.c
+jemalloc/test/unit/ticker.c
+jemalloc/test/unit/tsd.c
+jemalloc/test/unit/witness.c
+jemalloc/test/unit/zero.c
+jemalloc/test/unit/zero.sh
+jemalloc/TUNING.md
+m4/ax_cxx_compile_stdcxx.m4
+m4/ax_cxx_compile_stdcxx_11.m4
+m4/ax_pthread.m4
+memkind.pc.in
+test/gtest_fused/gtest/gtest-all.cc
+test/gtest_fused/gtest/gtest.h
+test/memkind-afts-ext.ts
+test/memkind-afts.ts
+test/memkind-perf-ext.ts
+test/memkind-perf.ts
+test/memkind-pytests.ts
+test/memkind-slts.ts
+utils/docker/README.md
diff --git a/memkind-1.10.0/copying_headers/MANIFEST.freeBSD b/memkind-1.10.0/copying_headers/MANIFEST.freeBSD
new file mode 100644
index 0000000..56798ed
--- /dev/null
+++ b/memkind-1.10.0/copying_headers/MANIFEST.freeBSD
@@ -0,0 +1,193 @@
+Makefile.am
+astyle.sh
+autogen.sh
+autohbw/Makefile.mk
+autohbw/autohbw.c
+autohbw/autohbw_get_src_lines.pl
+autohbw/autohbw_test.sh
+build.sh
+configure.ac
+examples/Makefile.mk
+examples/autohbw_candidates.c
+examples/filter_example.c
+examples/hello_hbw_example.c
+examples/hello_memkind_example.c
+examples/memkind_allocated.hpp
+examples/memkind_allocated_example.cpp
+examples/memkind_decorator_debug.c
+examples/memkind_get_stat.c
+include/hbw_allocator.h
+include/hbwmalloc.h
+include/memkind.h
+include/memkind/internal/heap_manager.h
+include/memkind/internal/memkind_arena.h
+include/memkind/internal/memkind_bandwidth.h
+include/memkind/internal/memkind_dax_kmem.h
+include/memkind/internal/memkind_default.h
+include/memkind/internal/memkind_gbtlb.h
+include/memkind/internal/memkind_hbw.h
+include/memkind/internal/memkind_hugetlb.h
+include/memkind/internal/memkind_interleave.h
+include/memkind/internal/memkind_log.h
+include/memkind/internal/memkind_pmem.h
+include/memkind/internal/memkind_private.h
+include/memkind/internal/memkind_regular.h
+include/memkind/internal/tbb_wrapper.h
+include/memkind/internal/vec.h
+include/memkind_allocator.h
+include/memkind_deprecated.h
+include/pmem_allocator.h
+install_astyle.sh
+man/autohbw.7
+man/hbwallocator.3
+man/hbwmalloc.3
+mna/memkind-auto-dax-kmem-nodes.1
+man/memkind-hbw-nodes.1
+man/memkind.3
+man/memkind_arena.3
+man/memkind_dax_kmem.3
+man/memkind_default.3
+man/memkind_hbw.3
+man/memkind_hugetlb.3
+man/memkind_pmem.3
+man/memkindallocator.3
+man/pmemallocator.3
+memkind.spec.mk
+src/Makefile.mk
+src/hbwmalloc.c
+src/heap_manager.c
+src/memkind-auto-dax-kmem-nodes.c
+src/memkind-hbw-nodes.c
+src/memkind.c
+src/memkind_arena.c
+src/memkind_bandwidth.c
+src/memkind_dax_kmem.c
+src/memkind_default.c
+src/memkind_gbtlb.c
+src/memkind_hbw.c
+src/memkind_hugetlb.c
+src/memkind_interleave.c
+src/memkind_log.c
+src/memkind_pmem.c
+src/memkind_regular.c
+src/tbb_wrapper.c
+test/Allocator.hpp
+test/Makefile.mk
+test/TestPolicy.hpp
+test/alloc_benchmark.c
+test/alloc_performance_tests.cpp
+test/allocate_to_max_stress_test.cpp
+test/allocator_perf_tool/AllocationSizes.hpp
+test/allocator_perf_tool/Allocation_info.cpp
+test/allocator_perf_tool/Allocation_info.hpp
+test/allocator_perf_tool/Allocator.hpp
+test/allocator_perf_tool/AllocatorFactory.hpp
+test/allocator_perf_tool/CSVLogger.hpp
+test/allocator_perf_tool/CommandLine.hpp
+test/allocator_perf_tool/Configuration.hpp
+test/allocator_perf_tool/ConsoleLog.hpp
+test/allocator_perf_tool/FunctionCalls.hpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
+test/allocator_perf_tool/FunctionCallsPerformanceTask.h
+test/allocator_perf_tool/GTestAdapter.hpp
+test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/HugePageOrganizer.hpp
+test/allocator_perf_tool/HugePageUnmap.hpp
+test/allocator_perf_tool/Iterator.hpp
+test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
+test/allocator_perf_tool/Makefile
+test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
+test/allocator_perf_tool/Numastat.hpp
+test/allocator_perf_tool/PmemMockup.cpp
+test/allocator_perf_tool/PmemMockup.hpp
+test/allocator_perf_tool/Runnable.hpp
+test/allocator_perf_tool/ScenarioWorkload.cpp
+test/allocator_perf_tool/ScenarioWorkload.h
+test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
+test/allocator_perf_tool/Stats.hpp
+test/allocator_perf_tool/StressIncreaseToMax.cpp
+test/allocator_perf_tool/StressIncreaseToMax.h
+test/allocator_perf_tool/Task.hpp
+test/allocator_perf_tool/TaskFactory.hpp
+test/allocator_perf_tool/Tests.hpp
+test/allocator_perf_tool/Thread.hpp
+test/allocator_perf_tool/TimerSysTime.hpp
+test/allocator_perf_tool/VectorIterator.hpp
+test/allocator_perf_tool/Workload.hpp
+test/allocator_perf_tool/WrappersMacros.hpp
+test/allocator_perf_tool/main.cpp
+test/autohbw_test.py
+test/autohbw_test_helper.c
+test/bat_tests.cpp
+test/check.cpp
+test/check.h
+test/common.h
+test/dax_kmem_nodes.cpp
+test/dax_kmem_nodes.h
+test/dax_kmem_env_var_test.py
+test/decorator_test.cpp
+test/decorator_test.h
+test/dlopen_test.cpp
+test/draw_plots.py
+test/environ_err_dax_kmem_malloc_positive_test.cpp
+test/environ_err_dax_kmem_malloc_test.cpp
+test/environ_err_hbw_malloc_test.cpp
+test/error_message_tests.cpp
+test/fragmentation_benchmark_pmem.cpp
+test/freeing_memory_segfault_test.cpp
+test/gb_page_tests_bind_policy.cpp
+test/get_arena_test.cpp
+test/hbw_allocator_performance_tests.cpp
+test/hbw_allocator_tests.cpp
+test/hbw_detection_test.py
+test/hbw_verify_function_test.cpp
+test/heap_manager_init_perf_test.cpp
+test/huge_page_test.cpp
+test/load_tbbmalloc_symbols.c
+test/locality_test.cpp
+test/main.cpp
+test/memkind_allocator_tests.cpp
+test/memkind_dax_kmem_test.cpp
+test/memkind_detect_kind_tests.cpp
+test/memkind_null_kind_test.cpp
+test/memkind_pmem_config_tests.cpp
+test/memkind_pmem_long_time_tests.cpp
+test/memkind_pmem_tests.cpp
+test/memkind_stat_test.cpp
+test/memkind_defrag_reallocate.cpp
+test/memkind_versioning_tests.cpp
+test/memory_footprint_test.cpp
+test/memory_manager.h
+test/multithreaded_tests.cpp
+test/negative_tests.cpp
+test/performance/framework.cpp
+test/performance/framework.hpp
+test/performance/operations.hpp
+test/performance/perf_tests.cpp
+test/performance/perf_tests.hpp
+test/pmem_alloc_performance_tests.cpp
+test/pmem_allocator_tests.cpp
+test/proc_stat.h
+test/python_framework/__init__.py
+test/python_framework/cmd_helper.py
+test/python_framework/huge_page_organizer.py
+test/random_sizes_allocator.h
+test/run_alloc_benchmark.sh
+test/static_kinds_list.h
+test/static_kinds_tests.cpp
+test/tbbmalloc.h
+test/test.sh
+test/test_dax_kmem.sh
+test/trace_mechanism_test.py
+test/trace_mechanism_test_helper.c
+test/trial_generator.cpp
+test/trial_generator.h
+utils/docker/Dockerfile.fedora-31
+utils/docker/Dockerfile.ubuntu-18.04
+utils/docker/docker_install_ndctl.sh
+utils/docker/docker_install_tbb.sh
+utils/docker/docker_run_build.sh
+utils/docker/docker_run_coverage.sh
+utils/docker/docker_run_test.sh
+utils/docker/run_local.sh
+utils/docker/set_host_configuration.sh
diff --git a/memkind-1.10.0/copying_headers/MANIFEST.freeBSD3 b/memkind-1.10.0/copying_headers/MANIFEST.freeBSD3
new file mode 100644
index 0000000..553a45e
--- /dev/null
+++ b/memkind-1.10.0/copying_headers/MANIFEST.freeBSD3
@@ -0,0 +1,14 @@
+examples/memkind_cpp_allocator.cpp
+examples/pmem_alignment.c
+examples/pmem_and_dax_kmem_kind.c
+examples/pmem_and_default_kind.c
+examples/pmem_config.c
+examples/pmem_cpp_allocator.cpp
+examples/pmem_detect_kind.c
+examples/pmem_free_with_unknown_kind.c
+examples/pmem_kinds.c
+examples/pmem_malloc.c
+examples/pmem_malloc_unlimited.c
+examples/pmem_multithreads.c
+examples/pmem_multithreads_onekind.c
+examples/pmem_usable_size.c
diff --git a/memkind-1.10.0/copying_headers/header.freeBSD b/memkind-1.10.0/copying_headers/header.freeBSD
new file mode 100644
index 0000000..db5a089
--- /dev/null
+++ b/memkind-1.10.0/copying_headers/header.freeBSD
@@ -0,0 +1,23 @@
+/*
+ * Copyright (C) 2014 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
\ No newline at end of file
diff --git a/memkind-1.10.0/copying_headers/header.freeBSD3 b/memkind-1.10.0/copying_headers/header.freeBSD3
new file mode 100644
index 0000000..9119412
--- /dev/null
+++ b/memkind-1.10.0/copying_headers/header.freeBSD3
@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2015, Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
diff --git a/memkind-1.10.0/copying_headers/header.intel-acpi b/memkind-1.10.0/copying_headers/header.intel-acpi
new file mode 100644
index 0000000..c4603f8
--- /dev/null
+++ b/memkind-1.10.0/copying_headers/header.intel-acpi
@@ -0,0 +1,36 @@
+/*
+ * Copyright (C) 2000 - 2014, Intel Corp.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions, and the following disclaimer,
+ *    without modification.
+ * 2. Redistributions in binary form must reproduce at minimum a disclaimer
+ *    substantially similar to the "NO WARRANTY" disclaimer below
+ *    ("Disclaimer") and any redistribution must be conditioned upon
+ *    including a substantially similar Disclaimer requirement for further
+ *    binary redistribution.
+ * 3. Neither the names of the above-listed copyright holders nor the names
+ *    of any contributors may be used to endorse or promote products derived
+ *    from this software without specific prior written permission.
+ *
+ * Alternatively, this software may be distributed under the terms of the
+ * GNU General Public License ("GPL") version 2 as published by the Free
+ * Software Foundation.
+ *
+ * NO WARRANTY
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTIBILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * HOLDERS OR CONTRIBUTORS BE LIABLE FOR SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
+ * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGES.
+ */
\ No newline at end of file
diff --git a/memkind-1.10.0/debian/changelog b/memkind-1.10.0/debian/changelog
new file mode 100644
index 0000000..8d118a4
--- /dev/null
+++ b/memkind-1.10.0/debian/changelog
@@ -0,0 +1,5 @@
+memkind (0.0.0) UNRELEASED; urgency=low
+
+  * See ChangeLog for more details
+
+ -- Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>  Mon, 12 Dec 2016 15:53:31 +0000
diff --git a/memkind-1.10.0/debian/compat b/memkind-1.10.0/debian/compat
new file mode 100644
index 0000000..ec63514
--- /dev/null
+++ b/memkind-1.10.0/debian/compat
@@ -0,0 +1 @@
+9
diff --git a/memkind-1.10.0/debian/control b/memkind-1.10.0/debian/control
new file mode 100644
index 0000000..5a13b16
--- /dev/null
+++ b/memkind-1.10.0/debian/control
@@ -0,0 +1,25 @@
+Source: memkind
+Section: utils
+Priority: optional
+Maintainer: Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>
+Build-Depends: debhelper (>= 8.0.0)
+Standards-Version: 3.9.4
+Homepage: https://github.com/memkind/memkind
+
+Package: memkind
+Architecture: any
+Depends: ${shlibs:Depends}, ${misc:Depends}
+Description:
+ The memkind library is an user extensible heap manager built on top of
+ jemalloc which enables control of memory characteristics and a
+ partitioning of the heap between kinds of memory. The kinds of memory
+ are defined by operating system memory policies that have been applied
+ to virtual address ranges. Memory characteristics supported by
+ memkind without user extension include control of NUMA and page size
+ features. The jemalloc non-standard interface has been extended to
+ enable specialized arenas to make requests for virtual memory from the
+ operating system through the memkind partition interface. Through the
+ other memkind interfaces the user can control and extend memory
+ partition features and allocate memory while selecting enabled
+ features. This software is being made available for early evaluation.
+ Feedback on design or implementation is greatly appreciated.
diff --git a/memkind-1.10.0/debian/rules b/memkind-1.10.0/debian/rules
new file mode 100755
index 0000000..5e7f167
--- /dev/null
+++ b/memkind-1.10.0/debian/rules
@@ -0,0 +1,21 @@
+#!/usr/bin/make -f
+# -*- makefile -*-
+
+# Uncomment this to turn on verbose mode.
+#export DH_VERBOSE=1
+
+# ignore tests during build
+override_dh_auto_test:
+
+override_dh_auto_configure:
+	./build_jemalloc.sh
+	./autogen.sh
+	dh_auto_configure -- --prefix=/usr --libdir=/usr/lib \
+           --includedir=/usr/include --sbindir=/usr/sbin --enable-cxx11 \
+           --mandir=/usr/share/man --docdir=/usr/share/doc/memkind
+
+override_dh_auto_build:
+	dh_auto_build -- checkprogs
+
+%:
+	dh $@
diff --git a/memkind-1.10.0/examples/Makefile.mk b/memkind-1.10.0/examples/Makefile.mk
new file mode 100644
index 0000000..bcc8ba2
--- /dev/null
+++ b/memkind-1.10.0/examples/Makefile.mk
@@ -0,0 +1,99 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+noinst_PROGRAMS += examples/autohbw_candidates \
+                   examples/filter_memkind \
+                   examples/hello_hbw \
+                   examples/hello_memkind \
+                   examples/hello_memkind_debug \
+                   examples/memkind_get_stat \
+                   examples/pmem_alignment \
+                   examples/pmem_and_dax_kmem_kind \
+                   examples/pmem_and_default_kind \
+                   examples/pmem_config \
+                   examples/pmem_detect_kind \
+                   examples/pmem_free_with_unknown_kind \
+                   examples/pmem_kinds \
+                   examples/pmem_malloc \
+                   examples/pmem_malloc_unlimited \
+                   examples/pmem_multithreads \
+                   examples/pmem_multithreads_onekind \
+                   examples/pmem_usable_size \
+                   # end
+if HAVE_CXX11
+noinst_PROGRAMS += examples/memkind_allocated
+noinst_PROGRAMS += examples/memkind_cpp_allocator
+noinst_PROGRAMS += examples/pmem_cpp_allocator
+endif
+
+examples_autohbw_candidates_LDADD = libmemkind.la
+examples_filter_memkind_LDADD = libmemkind.la
+examples_hello_hbw_LDADD = libmemkind.la
+examples_hello_memkind_LDADD = libmemkind.la
+examples_hello_memkind_debug_LDADD = libmemkind.la
+examples_memkind_get_stat_LDADD = libmemkind.la
+examples_pmem_alignment_LDADD = libmemkind.la
+examples_pmem_and_dax_kmem_kind_LDADD = libmemkind.la
+examples_pmem_and_default_kind_LDADD = libmemkind.la
+examples_pmem_config_LDADD = libmemkind.la
+examples_pmem_detect_kind_LDADD = libmemkind.la
+examples_pmem_free_with_unknown_kind_LDADD = libmemkind.la
+examples_pmem_kinds_LDADD = libmemkind.la
+examples_pmem_malloc_LDADD = libmemkind.la
+examples_pmem_malloc_unlimited_LDADD = libmemkind.la
+examples_pmem_multithreads_LDADD = libmemkind.la
+examples_pmem_multithreads_onekind_LDADD = libmemkind.la
+examples_pmem_usable_size_LDADD = libmemkind.la
+
+if HAVE_CXX11
+examples_memkind_allocated_LDADD = libmemkind.la
+examples_memkind_cpp_allocator_LDADD  = libmemkind.la
+examples_pmem_cpp_allocator_LDADD = libmemkind.la
+endif
+
+examples_autohbw_candidates_SOURCES = examples/autohbw_candidates.c
+examples_filter_memkind_SOURCES = examples/filter_example.c
+examples_hello_hbw_SOURCES = examples/hello_hbw_example.c
+examples_hello_memkind_SOURCES = examples/hello_memkind_example.c
+examples_hello_memkind_debug_SOURCES = examples/hello_memkind_example.c examples/memkind_decorator_debug.c
+examples_pmem_alignment_SOURCES = examples/pmem_alignment.c
+examples_pmem_and_dax_kmem_kind_SOURCES = examples/pmem_and_dax_kmem_kind.c
+examples_pmem_and_default_kind_SOURCES = examples/pmem_and_default_kind.c
+examples_pmem_config_SOURCES = examples/pmem_config.c
+examples_pmem_detect_kind_SOURCES = examples/pmem_detect_kind.c
+examples_pmem_free_with_unknown_kind_SOURCES = examples/pmem_free_with_unknown_kind.c
+examples_pmem_kinds_SOURCES = examples/pmem_kinds.c
+examples_pmem_malloc_SOURCES = examples/pmem_malloc.c
+examples_pmem_malloc_unlimited_SOURCES = examples/pmem_malloc_unlimited.c
+examples_pmem_multithreads_SOURCES = examples/pmem_multithreads.c
+examples_pmem_multithreads_onekind_SOURCES = examples/pmem_multithreads_onekind.c
+examples_pmem_usable_size_SOURCES = examples/pmem_usable_size.c
+if HAVE_CXX11
+examples_memkind_allocated_SOURCES = examples/memkind_allocated_example.cpp examples/memkind_allocated.hpp
+examples_memkind_cpp_allocator_SOURCES = examples/memkind_cpp_allocator.cpp
+examples_pmem_cpp_allocator_SOURCES = examples/pmem_cpp_allocator.cpp
+endif
+
+clean-local:
+	rm -f examples/*.gcno examples/*.gcda
diff --git a/memkind-1.10.0/examples/README b/memkind-1.10.0/examples/README
new file mode 100644
index 0000000..79e4408
--- /dev/null
+++ b/memkind-1.10.0/examples/README
@@ -0,0 +1,104 @@
+# Memkind examples
+
+The example directory contains example codes that use the memkind
+interface.
+
+## PMEM
+
+The pmem_*.c(pp) demonstrates how to create and use a file-backed memory kind.
+The default pmem path is "/tmp/".
+Custom directory is pass as first argument to all of PMEM example programs,
+e.g. to execute pmem_malloc example in /mnt/pmem location, call:
+
+    ./pmem_malloc /mnt/pmem/
+
+### pmem_kinds.c
+
+This example shows how to create and destroy pmem kind with defined or unlimited size.
+
+### pmem_malloc.c
+
+This example shows how to allocate memory and possibility to exceed pmem kind size.
+
+### pmem_malloc_unlimited.c
+
+This example shows how to allocate memory with unlimited kind size.
+
+### pmem_usable_size.c
+
+This example shows difference between the expected and the actual allocation size.
+
+### pmem_alignment.c
+
+This example shows how to use memkind alignment and how it affects allocations.
+
+### pmem_multithreads.c
+
+This example shows how to use multithreading with independent pmem kinds.
+
+### pmem_multithreads_onekind.c
+
+This example shows how to use multithreading with one main pmem kind.
+
+### pmem_and_dax_kmem_kind.c
+
+This example shows how to allocate to PMEM memory using file-backed memory (pmem kind)
+and persistent memory NUMA node (MEMKIND_DAX_KMEM).
+
+### pmem_and_default_kind.c
+
+This example shows how to allocate in standard memory and file-backed memory (pmem kind).
+
+### pmem_detect_kind.c
+
+This example shows how to distinguish allocation from different kinds using detect kind function.
+
+### pmem_config.c
+
+This example shows how to use custom configuration to create pmem kind.
+
+### pmem_free_with_unknown_kind.c
+
+This example shows how to allocate in standard memory and file-backed memory (pmem kind)
+and free memory without a need to remember which kind it belongs to.
+
+### pmem_cpp_allocator.cpp
+
+This example shows usage of C++ allocator mechanism designed for file-backed memory
+kind with different data structures like: vector, list and map.
+
+## Other memkind examples
+
+### memkind_get_stat.c
+
+This example shows how to get information about allocator and kind statistics.
+
+### other
+
+The simplest example is the hello_example.c which is a hello world
+variant.  The filter_example.c shows how you would use high bandwidth
+memory to store a reduction of a larger data set stored in DDR. There is
+also an example of how to create user defined kinds.  This example
+creates kinds which isolate allocations to a single NUMA node each
+backed by a single arena.
+
+### memkind_cpp_allocator.cpp
+
+This example shows usage of C++ allocator mechanism designed for static
+kinds with different data structures like: forward list, deque and multiset.
+
+The memkind_allocated example is simple usage of memkind in C++11 which
+shows how memkind can be used to allocate objects, and consists of two files:
+memkind_allocated.hpp - which is definition of template class that should be
+inherited and parametrized by derived type (curiously recurring template
+pattern), to let deriving class allocate their objects using specified kind.
+memkind_allocated_example.cpp - which is usage example of this approach.
+Logic of memkind_allocated is based on overriding operator new() in template,
+and allocating memory on kind specified in new() parameter, or by overridable
+static method getClassKind(). This implementation also supports alignment
+specifier's (alignas() - new feature in C++11).
+The downside of this approach is that it will work properly only if
+memkind_allocated template is inherited once in inheritance chain (which
+probably makes that not very useful for most scenarios). Other thing is that it
+overriding class new() operator which can cause various problems if used
+unwisely.
diff --git a/memkind-1.10.0/examples/autohbw_candidates.c b/memkind-1.10.0/examples/autohbw_candidates.c
new file mode 100644
index 0000000..f4844ce
--- /dev/null
+++ b/memkind-1.10.0/examples/autohbw_candidates.c
@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////
+// File   : autohbw_candidates.c
+// Purpose: Shows which functions are interposed by AutoHBW library.
+//        : These functions can be used for testing purposes
+// Author : Ruchira Sasanka (ruchira.sasanka AT intel.com)
+// Date   : Sept 10, 2015
+///////////////////////////////////////////////////////////////////////////
+
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+
+
+///////////////////////////////////////////////////////////////////////////
+// This function contains an example case for each heap allocation function
+// intercepted by the AutoHBW library
+///////////////////////////////////////////////////////////////////////////
+
+//volatile is needed to prevent optimizing out below hooks
+volatile int memkind_called_g;
+
+void memkind_malloc_post(struct memkind *kind, size_t size, void **result)
+{
+    memkind_called_g = 1;
+}
+void memkind_calloc_post(struct memkind *kind, size_t nmemb, size_t size,
+                         void **result)
+{
+    memkind_called_g = 1;
+}
+void memkind_posix_memalign_post(struct memkind *kind, void **memptr,
+                                 size_t alignment, size_t size, int *err)
+{
+    memkind_called_g = 1;
+}
+void memkind_realloc_post(struct memkind *kind, void *ptr, size_t size,
+                          void **result)
+{
+    memkind_called_g = 1;
+}
+void memkind_free_pre(struct memkind **kind, void **ptr)
+{
+    memkind_called_g = 1;
+}
+
+void finish_testcase(int fail_condition, const char *fail_message, int *err)
+{
+
+    if(memkind_called_g != 1 || fail_condition) {
+        printf("%s\n", fail_message);
+        *err= -1;
+    }
+    memkind_called_g = 0;
+}
+
+int main()
+{
+    int err = 0;
+    const size_t size = 1024 * 1024;   // 1M of data
+
+    void *buf = NULL;
+    memkind_called_g = 0;
+
+    // Test 1: Test malloc and free
+    buf = malloc(size);
+    finish_testcase(buf==NULL, "Malloc failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after malloc failed!", &err);
+
+    // Test 2: Test calloc and free
+    buf = calloc(size, 1);
+    finish_testcase(buf==NULL, "Calloc failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after calloc failed!", &err);
+
+    // Test 3: Test realloc and free
+    buf = malloc(size);
+    finish_testcase(buf==NULL, "Malloc before realloc failed!", &err);
+
+    buf = realloc(buf,  size * 2);
+    finish_testcase(buf==NULL, "Realloc failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after realloc failed!", &err);
+    buf = NULL;
+
+    // Test 4: Test posix_memalign and free
+    int ret = posix_memalign(&buf,  64, size);
+    finish_testcase(ret, "Posix_memalign failed!", &err);
+
+    free(buf);
+    finish_testcase(0, "Free after posix_memalign failed!", &err);
+
+    return err;
+}
diff --git a/memkind-1.10.0/examples/filter_example.c b/memkind-1.10.0/examples/filter_example.c
new file mode 100644
index 0000000..f128d71
--- /dev/null
+++ b/memkind-1.10.0/examples/filter_example.c
@@ -0,0 +1,87 @@
+/*
+ * Copyright (C) 2014 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <errno.h>
+
+int main(int argc, char **argv)
+{
+    const size_t stream_len = 1024 * 1024;
+    const size_t filter_len = 1024;
+    const size_t num_filter = stream_len / filter_len;
+    size_t i, j;
+    double *stream = NULL;
+    double *filter = NULL;
+    double *result = NULL;
+
+    srandom(0);
+
+    stream = (double *)memkind_malloc(MEMKIND_DEFAULT, stream_len * sizeof(double));
+    if (stream == NULL) {
+        perror("<memkind>");
+        fprintf(stderr, "Unable to allocate stream\n");
+        return errno ? -errno : 1;
+    }
+
+    filter = (double *)memkind_malloc(MEMKIND_HBW, filter_len * sizeof(double));
+    if (filter == NULL) {
+        perror("<memkind>");
+        fprintf(stderr, "Unable to allocate filter\n");
+        return errno ? -errno : 1;
+    }
+
+    result = (double *)memkind_calloc(MEMKIND_HBW, filter_len, sizeof(double));
+    if (result == NULL) {
+        perror("<memkind>");
+        fprintf(stderr, "Unable to allocate result\n");
+        return errno ? -errno : 1;
+    }
+
+    for (i = 0; i < stream_len; i++) {
+        stream[i] = (double)(random())/(double)(RAND_MAX);
+    }
+
+    for (i = 0; i < filter_len; i++) {
+        filter[i] = (double)(i)/(double)(filter_len);
+    }
+
+    for (i = 0; i < num_filter; i++) {
+        for (j = 0; j < filter_len; j++) {
+            result[j] += stream[i * filter_len + j] * filter[j];
+        }
+    }
+
+    for (i = 0; i < filter_len; i++) {
+        fprintf(stdout, "%.6e\n", result[i]);
+    }
+
+    memkind_free(MEMKIND_HBW, result);
+    memkind_free(MEMKIND_HBW, filter);
+    memkind_free(MEMKIND_DEFAULT, stream);
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/hello_hbw_example.c b/memkind-1.10.0/examples/hello_hbw_example.c
new file mode 100644
index 0000000..7b044fa
--- /dev/null
+++ b/memkind-1.10.0/examples/hello_hbw_example.c
@@ -0,0 +1,81 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbwmalloc.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <errno.h>
+
+int main(int argc, char **argv)
+{
+    const size_t size = 512;
+    char *default_str = NULL;
+    char *hbw_str = NULL;
+    char *hbw_hugetlb_str = NULL;
+    int err = 0;
+
+    default_str = (char *)malloc(size);
+    if (default_str == NULL) {
+        perror("malloc()");
+        fprintf(stderr, "Unable to allocate default string\n");
+        err = errno ? -errno : 1;
+        goto exit;
+    }
+    hbw_str = (char *)hbw_malloc(size);
+    if (hbw_str == NULL) {
+        perror("hbw_malloc()");
+        fprintf(stderr, "Unable to allocate hbw string\n");
+        err = errno ? -errno : 1;
+        goto exit;
+    }
+    err = hbw_posix_memalign_psize((void **)&hbw_hugetlb_str, 2097152, size,
+                                   HBW_PAGESIZE_2MB);
+    if (err) {
+        perror("hbw_posix_memalign()");
+        fprintf(stderr, "Unable to allocate hbw hugetlb string\n");
+        err = errno ? -errno : 1;
+        goto exit;
+    }
+
+    sprintf(default_str, "Hello world from standard memory\n");
+    sprintf(hbw_str, "Hello world from high bandwidth memory\n");
+    sprintf(hbw_hugetlb_str, "Hello world from high bandwidth 2 MB paged memory\n");
+
+    fprintf(stdout, "%s", default_str);
+    fprintf(stdout, "%s", hbw_str);
+    fprintf(stdout, "%s", hbw_hugetlb_str);
+
+exit:
+    if (hbw_hugetlb_str) {
+        hbw_free(hbw_hugetlb_str);
+    }
+    if (hbw_str) {
+        hbw_free(hbw_str);
+    }
+    if (default_str) {
+        free(default_str);
+    }
+    return err;
+}
diff --git a/memkind-1.10.0/examples/hello_memkind_example.c b/memkind-1.10.0/examples/hello_memkind_example.c
new file mode 100644
index 0000000..e85e30c
--- /dev/null
+++ b/memkind-1.10.0/examples/hello_memkind_example.c
@@ -0,0 +1,115 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <errno.h>
+
+int main(int argc, char **argv)
+{
+    const size_t size = 512;
+    char *default_str = NULL;
+    char *hugetlb_str = NULL;
+    char *hbw_str = NULL;
+    char *hbw_hugetlb_str = NULL;
+    char *hbw_preferred_str = NULL;
+    char *hbw_preferred_hugetlb_str = NULL;
+    char *hbw_interleave_str = NULL;
+
+    default_str = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    if (default_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate default string\n");
+        return errno ? -errno : 1;
+    }
+    hugetlb_str = (char *)memkind_malloc(MEMKIND_HUGETLB, size);
+    if (hugetlb_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hugetlb string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_str = (char *)memkind_malloc(MEMKIND_HBW, size);
+    if (hbw_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_hugetlb_str = (char *)memkind_malloc(MEMKIND_HBW_HUGETLB, size);
+    if (hbw_hugetlb_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw_hugetlb string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_preferred_str = (char *)memkind_malloc(MEMKIND_HBW_PREFERRED, size);
+    if (hbw_preferred_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw_preferred string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_preferred_hugetlb_str = (char *)memkind_malloc(
+                                    MEMKIND_HBW_PREFERRED_HUGETLB, size);
+    if (hbw_preferred_hugetlb_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr, "Unable to allocate hbw_preferred_hugetlb string\n");
+        return errno ? -errno : 1;
+    }
+    hbw_interleave_str = (char *)memkind_malloc(MEMKIND_HBW_INTERLEAVE, size);
+    if (hbw_interleave_str == NULL) {
+        perror("memkind_malloc()");
+        fprintf(stderr,"Unable to allocate hbw_interleave string\n");
+        return errno ? -errno : 1;
+    }
+
+    sprintf(default_str, "Hello world from standard memory\n");
+    sprintf(hugetlb_str, "Hello world from standard memory with 2 MB pages\n");
+    sprintf(hbw_str, "Hello world from high bandwidth memory\n");
+    sprintf(hbw_hugetlb_str, "Hello world from high bandwidth 2 MB paged memory\n");
+    sprintf(hbw_preferred_str,
+            "Hello world from high bandwidth memory if sufficient resources exist\n");
+    sprintf(hbw_preferred_hugetlb_str,
+            "Hello world from high bandwidth 2 MB paged memory if sufficient resources exist\n");
+
+    sprintf(hbw_interleave_str,
+            "Hello world from high bandwidth interleaved memory\n");
+
+    fprintf(stdout, "%s", default_str);
+    fprintf(stdout, "%s", hugetlb_str);
+    fprintf(stdout, "%s", hbw_str);
+    fprintf(stdout, "%s", hbw_hugetlb_str);
+    fprintf(stdout, "%s", hbw_preferred_str);
+    fprintf(stdout, "%s", hbw_preferred_hugetlb_str);
+    fprintf(stdout, "%s", hbw_interleave_str);
+
+    memkind_free(MEMKIND_HBW_INTERLEAVE, hbw_interleave_str);
+    memkind_free(MEMKIND_HBW_PREFERRED_HUGETLB, hbw_preferred_hugetlb_str);
+    memkind_free(MEMKIND_HBW_PREFERRED, hbw_preferred_str);
+    memkind_free(MEMKIND_HBW_HUGETLB, hbw_hugetlb_str);
+    memkind_free(MEMKIND_HBW, hbw_str);
+    memkind_free(MEMKIND_HUGETLB, hugetlb_str);
+    memkind_free(MEMKIND_DEFAULT, default_str);
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/memkind_allocated.hpp b/memkind-1.10.0/examples/memkind_allocated.hpp
new file mode 100644
index 0000000..dd75bef
--- /dev/null
+++ b/memkind-1.10.0/examples/memkind_allocated.hpp
@@ -0,0 +1,107 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+#include <memkind.h>
+
+#include <cstdlib>
+#include <new>
+
+template <class deriving_class>
+class memkind_allocated
+{
+public:
+    static memkind_t getClassKind()
+    {
+        return MEMKIND_DEFAULT;
+    }
+
+    void *operator new(std::size_t size)
+    {
+        return deriving_class::operator new(size, deriving_class::getClassKind());
+    }
+
+    void *operator new[](std::size_t size)
+    {
+        return deriving_class::operator new(size, deriving_class::getClassKind());
+    }
+
+    void *operator new(std::size_t size, memkind_t memory_kind)
+    {
+        void *result_ptr = NULL;
+        int allocation_result = 0;
+
+        //This check if deriving_class has specified alignment, which is suitable
+        //to be used with posix_memalign()
+        if(alignof(deriving_class) <  sizeof(void *)) {
+            result_ptr = memkind_malloc(memory_kind, size);
+            allocation_result = result_ptr ? 1 : 0;
+        } else {
+            allocation_result = memkind_posix_memalign(memory_kind, &result_ptr,
+                                                       alignof(deriving_class), size);
+        }
+
+        if(allocation_result) {
+            throw std::bad_alloc();
+        }
+
+        return result_ptr;
+    }
+
+    void *operator new[](std::size_t size, memkind_t memory_kind)
+    {
+        return deriving_class::operator new(size, memory_kind);
+    }
+
+    void operator delete(void *ptr, memkind_t memory_kind)
+    {
+        memkind_free(memory_kind, ptr);
+    }
+
+    void operator delete(void *ptr)
+    {
+        memkind_free(0, ptr);
+    }
+
+    void operator delete[](void *ptr)
+    {
+        deriving_class::operator delete(ptr);
+    }
+
+    void operator delete[](void *ptr, memkind_t memory_kind)
+    {
+        deriving_class::operator delete(ptr, memory_kind);
+    }
+
+protected:
+    memkind_allocated()
+    {
+    }
+
+    ~memkind_allocated()
+    {
+    }
+
+};
diff --git a/memkind-1.10.0/examples/memkind_allocated_example.cpp b/memkind-1.10.0/examples/memkind_allocated_example.cpp
new file mode 100644
index 0000000..52f916b
--- /dev/null
+++ b/memkind-1.10.0/examples/memkind_allocated_example.cpp
@@ -0,0 +1,105 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+///////////////////////////////////////////////////////////////////////////////////////////////////////////
+
+#include <memkind.h>
+#include "memkind_allocated.hpp"
+
+// This code is example usage of C++11 features with custom allocator,
+// support for C++11 is required.
+#include <iostream>
+
+#if __cplusplus > 199711L
+
+#include <cstdlib>
+#include <new>
+#include <string>
+
+//example class definition, which derive from  memkind_allocated template to
+//have objects allocated with memkind, and have alignment specified by alignas()
+class alignas(128) memkind_allocated_example : public
+    memkind_allocated<memkind_allocated_example>
+{
+    std::string message;
+
+public:
+    //Override method for returning class default kind to make objects be by default allocated on High-Bandwidth Memory
+    static memkind_t getClassKind()
+    {
+        return MEMKIND_HBW;
+    }
+
+    memkind_allocated_example(std::string my_message)
+    {
+        this->message = my_message;
+    }
+
+    memkind_allocated_example()
+    {
+    }
+
+    void print_message()
+    {
+        std::cout << message << std::endl;
+        std::cout << "Memory address of this object is: " <<  (void *)this << std::endl
+                  << std::endl;
+    }
+};
+
+int main()
+{
+    memkind_t specified_kind = MEMKIND_HBW_HUGETLB;
+
+    memkind_allocated_example *default_kind_example = new memkind_allocated_example(
+        std::string("This object has been allocated using class default kind, which is: MEMKIND_DEFAULT") );
+    default_kind_example->print_message();
+    delete default_kind_example;
+
+    memkind_allocated_example *specified_kind_example = new(
+        specified_kind) memkind_allocated_example(
+        std::string("This object has been allocated using specified kind, which is: MEMKIND_HBW_HUGETLB") );
+    specified_kind_example->print_message();
+    delete specified_kind_example;
+
+    //examples for using same approach for allocating arrays of objects, note that objects created that way can be initialized only with default (unparameterized) constructor
+    memkind_allocated_example *default_kind_array_example = new
+    memkind_allocated_example[5]();
+    delete[] default_kind_array_example;
+
+    memkind_allocated_example *specified_kind_array_example = new(
+        specified_kind) memkind_allocated_example[5]();
+    delete[] specified_kind_array_example;
+
+    return 0;
+}
+#else //If C++11 is not available - do nothing.
+int main()
+{
+    std::cout << "WARNING: because your compiler does not support C++11 standard,"
+              << std::endl;
+    std::cout << "this example is only as a dummy placeholder." << std::endl;
+    return 0;
+}
+#endif
diff --git a/memkind-1.10.0/examples/memkind_cpp_allocator.cpp b/memkind-1.10.0/examples/memkind_cpp_allocator.cpp
new file mode 100644
index 0000000..98b165f
--- /dev/null
+++ b/memkind-1.10.0/examples/memkind_cpp_allocator.cpp
@@ -0,0 +1,126 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#include "memkind_allocator.h"
+
+#include <cassert>
+#include <forward_list>
+#include <iostream>
+#include <scoped_allocator>
+#include <set>
+#include <string>
+#include <deque>
+
+#define STL_FWLIST_TEST
+#define STL_DEQUE_TEST
+#if _GLIBCXX_USE_CXX11_ABI
+#define STL_MULTISET_TEST
+#endif
+
+int main(int argc, char *argv[])
+{
+    std::cout << "TEST SCOPE: HELLO" << std::endl;
+
+#ifdef STL_FWLIST_TEST
+    {
+        std::cout << "FORWARD LIST OPEN" << std::endl;
+        libmemkind::static_kind::allocator<int> alc{ libmemkind::kinds::REGULAR };
+        std::forward_list<int, libmemkind::static_kind::allocator<int>> fwlist {alc};
+        int i;
+
+        for (i = 0; i <= 20; ++i) {
+            fwlist.push_front(i);
+        }
+
+        fwlist.sort(std::greater<int>());
+
+        for(auto &el: fwlist) {
+            i--;
+            assert(el == i);
+        }
+
+        fwlist.clear();
+
+        std::cout << "FORWARD LIST CLOSE" << std::endl;
+    }
+#endif
+
+#ifdef STL_DEQUE_TEST
+    std::cout << "DEQUE OPEN" << std::endl;
+    libmemkind::static_kind::allocator<int> int_alc{ libmemkind::kinds::REGULAR };
+    std::deque<int, libmemkind::static_kind::allocator<int>> deque {int_alc};
+
+    for (int i=0; i<10; i++) {
+        deque.push_back(i);
+    }
+
+    assert(deque.size() == 10);
+
+    while(!deque.empty()) {
+        deque.pop_front();
+    }
+
+    assert(deque.size() == 0);
+
+    std::cout << "DEQUE CLOSE" << std::endl;
+#endif
+
+#ifdef STL_MULTISET_TEST
+    {
+        std::cout << "MULTISET OPEN" << std::endl;
+        typedef libmemkind::static_kind::allocator<std::string> multiset_alloc_t;
+        multiset_alloc_t multiset_alloc { libmemkind::kinds::DEFAULT };
+        std::multiset<std::string, std::less<std::string>, std::scoped_allocator_adaptor<multiset_alloc_t> >
+        multiset{ std::scoped_allocator_adaptor<multiset_alloc_t>(multiset_alloc) };
+
+        multiset.insert("S");
+        multiset.insert("A");
+        multiset.insert("C");
+        multiset.insert("S");
+        multiset.insert("C");
+        multiset.insert("E");
+
+        std::string test;
+        for (auto &n : multiset) {
+            test.append(n);
+        }
+        assert(test == "ACCESS");
+
+        std::cout << "MULTISET CLOSE" << std::endl;
+    }
+#endif
+
+    std::cout << "TEST SCOPE: GOODBYE" << std::endl;
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/memkind_decorator_debug.c b/memkind-1.10.0/examples/memkind_decorator_debug.c
new file mode 100644
index 0000000..51183d3
--- /dev/null
+++ b/memkind-1.10.0/examples/memkind_decorator_debug.c
@@ -0,0 +1,64 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+
+/* This is an example that enables debug printing on every alloction call */
+
+static void memkind_debug(const char *func, memkind_t kind, size_t size,
+                          void *ptr)
+{
+    fprintf(stderr, "[ DEBUG ] func=%s kind=%p size=%zu ptr=0x%lx\n", func, kind,
+            size, (size_t)ptr);
+}
+
+void memkind_malloc_post(memkind_t kind, size_t size, void **result)
+{
+    memkind_debug("memkind_malloc", kind, size, *result);
+}
+
+void memkind_calloc_post(memkind_t kind, size_t nmemb, size_t size,
+                         void **result)
+{
+    memkind_debug("memkind_calloc", kind, nmemb * size, *result);
+}
+
+void memkind_posix_memalign_post(memkind_t kind, void **memptr,
+                                 size_t alignment, size_t size, int *err)
+{
+    memkind_debug("memkind_posix_memalign", kind, size, *memptr);
+}
+
+void memkind_realloc_post(memkind_t kind, void *ptr, size_t size, void **result)
+{
+    memkind_debug("memkind_realloc", kind, size, *result);
+}
+
+void memkind_free_pre(memkind_t kind, void **ptr)
+{
+    memkind_debug("memkind_free", kind, 0, *ptr);
+}
diff --git a/memkind-1.10.0/examples/memkind_get_stat.c b/memkind-1.10.0/examples/memkind_get_stat.c
new file mode 100644
index 0000000..7e8d0d5
--- /dev/null
+++ b/memkind-1.10.0/examples/memkind_get_stat.c
@@ -0,0 +1,69 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include <stdio.h>
+
+int main()
+{
+    const size_t size = 1024;
+
+    size_t stats_active;
+    size_t stats_resident;
+    size_t stats_allocated;
+    size_t stats_kind_regular_active;
+    size_t stats_kind_regular_resident;
+    size_t stats_kind_regular_allocated;
+
+    char *ptr_regular = (char *)memkind_malloc(MEMKIND_REGULAR, size);
+
+    fprintf(stdout,
+            "This example shows how to use memkind API to retrieve information about allocation stats.\n");
+
+    snprintf(ptr_regular, size,
+             "Hello world from regular kind memory - ptr_regular.\n");
+
+    memkind_update_cached_stats();
+    memkind_get_stat(NULL, MEMKIND_STAT_TYPE_RESIDENT, &stats_resident);
+    memkind_get_stat(NULL, MEMKIND_STAT_TYPE_ACTIVE, &stats_active);
+    memkind_get_stat(NULL, MEMKIND_STAT_TYPE_ALLOCATED, &stats_allocated);
+    memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_RESIDENT,
+                     &stats_kind_regular_resident);
+    memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_ACTIVE,
+                     &stats_kind_regular_active);
+    memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_ALLOCATED,
+                     &stats_kind_regular_allocated);
+
+    fprintf(stdout,
+            "\n Global stats \nresident %zu \nactive %zu, \nallocated %zu \n",
+            stats_resident, stats_active, stats_allocated);
+    fprintf(stdout,
+            "\n MEMKIND_REGULAR stats \nresident %zu \nactive %zu, \nallocated %zu \n",
+            stats_kind_regular_resident, stats_kind_regular_active,
+            stats_kind_regular_allocated);
+
+    memkind_free(NULL, ptr_regular);
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_alignment.c b/memkind-1.10.0/examples/pmem_alignment.c
new file mode 100644
index 0000000..3c314fc
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_alignment.c
@@ -0,0 +1,132 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+
+static char path[PATH_MAX] = "/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use memkind alignment and how it affects allocations.\nPMEM kind directory: %s\n",
+            path);
+
+    err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str10 = NULL;
+    char *pmem_str11 = NULL;
+
+    // Perform two allocations - 32 bytes
+    pmem_str10 = (char *)memkind_malloc(pmem_kind, 32);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    pmem_str11 = (char *)memkind_malloc(pmem_kind, 32);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11)\n");
+        return 1;
+    }
+
+    // The allocations should be very close to each other in memory
+    if (pmem_str11 - pmem_str10 != 32) {
+        fprintf(stderr, "Something went wrong with allocations.\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind, pmem_str10);
+    memkind_free(pmem_kind, pmem_str11);
+
+    // Perform two allocations - 32 bytes with alignment 64
+    err = memkind_posix_memalign(pmem_kind, (void **)&pmem_str10, 64, 32);
+    if (err) {
+        fprintf(stderr,
+                "Unable to allocate pmem string (pmem_str10) with alignment.\n");
+        return 1;
+    }
+
+    err = memkind_posix_memalign(pmem_kind, (void **)&pmem_str11, 64, 32);
+    if (err) {
+        fprintf(stderr,
+                "Unable to allocate pmem string (pmem_str11) with alignment.\n");
+        return 1;
+    }
+
+    // The addresses of allocations are not close to each other in memory, they are aligned to 64
+    if (pmem_str11 - pmem_str10 != 64) {
+        fprintf(stderr, "Something went wrong with alignment allocations.\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind, pmem_str10);
+    memkind_free(pmem_kind, pmem_str11);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "The memory has been successfully allocated using memkind alignment.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_and_dax_kmem_kind.c b/memkind-1.10.0/examples/pmem_and_dax_kmem_kind.c
new file mode 100644
index 0000000..45a544f
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_and_dax_kmem_kind.c
@@ -0,0 +1,109 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    const size_t size = 512;
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2) {
+        if (realpath(argv[1], path) == NULL) {
+            fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+            return 1;
+        }
+    }
+
+    char *ptr_dax_kmem = NULL;
+    char *ptr_pmem = NULL;
+
+    fprintf(stdout,
+            "This example shows how to allocate to PMEM memory using file-backed memory (pmem kind) "
+            "and persistent memory NUMA node (MEMKIND_DAX_KMEM).\nPMEM kind directory: %s\n",
+            path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    ptr_dax_kmem = (char *)memkind_malloc(MEMKIND_DAX_KMEM, size);
+    if (!ptr_dax_kmem) {
+        fprintf(stderr, "Unable allocate 512 bytes in persistent memory NUMA node.\n");
+        return 1;
+    }
+
+    ptr_pmem = (char *)memkind_malloc(pmem_kind, size);
+    if (!ptr_pmem) {
+        fprintf(stderr, "Unable allocate 512 bytes in file-backed memory.\n");
+        return 1;
+    }
+
+    snprintf(ptr_dax_kmem, size,
+             "Hello world from persistent memory NUMA node - ptr_dax_kmem.\n");
+    snprintf(ptr_pmem, size, "Hello world from file-backed memory - ptr_pmem.\n");
+
+    fprintf(stdout, "%s", ptr_dax_kmem);
+    fprintf(stdout, "%s", ptr_pmem);
+
+    memkind_free(MEMKIND_DAX_KMEM, ptr_dax_kmem);
+    memkind_free(pmem_kind, ptr_pmem);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_and_default_kind.c b/memkind-1.10.0/examples/pmem_and_default_kind.c
new file mode 100644
index 0000000..4b0ec38
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_and_default_kind.c
@@ -0,0 +1,143 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <errno.h>
+#include <sys/resource.h>
+
+#define MB (1024 * 1024)
+#define HEAP_LIMIT_SIMULATE (1024 * MB)
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    const size_t size = 512;
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+    errno = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2) {
+        if (realpath(argv[1], path) == NULL) {
+            fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+            return 1;
+        }
+    }
+
+    // Operation below limit the current size of heap
+    // to show different place of allocation
+    const struct rlimit heap_limit = { HEAP_LIMIT_SIMULATE, HEAP_LIMIT_SIMULATE };
+    err = setrlimit(RLIMIT_DATA, &heap_limit);
+    if (err) {
+        fprintf(stderr, "Unable to set heap limit.\n");
+        return 1;
+    }
+
+    char *ptr_default = NULL;
+    char *ptr_default_not_possible = NULL;
+    char *ptr_pmem = NULL;
+
+    fprintf(stdout,
+            "This example shows how to allocate memory using standard memory (MEMKIND_DEFAULT) "
+            "and file-backed kind of memory (PMEM).\nPMEM kind directory: %s\n", path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    ptr_default = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    if (!ptr_default) {
+        fprintf(stderr, "Unable allocate 512 bytes in standard memory.\n");
+        return 1;
+    }
+
+    errno = 0;
+    ptr_default_not_possible = (char *)memkind_malloc(MEMKIND_DEFAULT,
+                                                      HEAP_LIMIT_SIMULATE);
+    if (ptr_default_not_possible) {
+        fprintf(stderr,
+                "Failure, this allocation should not be possible "
+                "(expected result was NULL), because of setlimit function.\n");
+        return 1;
+    }
+    if (errno != ENOMEM) {
+        fprintf(stderr,
+                "Failure, this allocation should set errno to ENOMEM value, because of setlimit function.\n");
+        return 1;
+    }
+
+    errno = 0;
+    ptr_pmem = (char *)memkind_malloc(pmem_kind, HEAP_LIMIT_SIMULATE);
+    if (!ptr_pmem) {
+        fprintf(stderr, "Unable allocate HEAP_LIMIT_SIMULATE in file-backed memory.\n");
+        return 1;
+    }
+    if (errno != 0) {
+        fprintf(stderr, "Failure, this allocation should not set errno value.\n");
+        return 1;
+    }
+
+    sprintf(ptr_default, "Hello world from standard memory - ptr_default.\n");
+    sprintf(ptr_pmem, "Hello world from file-backed memory - ptr_pmem.\n");
+
+    fprintf(stdout, "%s", ptr_default);
+    fprintf(stdout, "%s", ptr_pmem);
+
+    memkind_free(MEMKIND_DEFAULT, ptr_default);
+    memkind_free(pmem_kind, ptr_pmem);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_config.c b/memkind-1.10.0/examples/pmem_config.c
new file mode 100644
index 0000000..cd7cc00
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_config.c
@@ -0,0 +1,98 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+
+static char path[PATH_MAX] = "/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use custom configuration to create pmem kind."
+            "\nPMEM kind directory: %s\n", path);
+
+    struct memkind_config *test_cfg = memkind_config_new();
+    if (!test_cfg) {
+        fprintf(stderr, "Unable to create memkind cfg.\n");
+        return 1;
+    }
+
+    memkind_config_set_path(test_cfg, path);
+    memkind_config_set_size(test_cfg, PMEM_MAX_SIZE);
+    memkind_config_set_memory_usage_policy(test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+
+
+    // Create PMEM partition with specific configuration
+    err =  memkind_create_pmem_with_config(test_cfg, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    memkind_config_delete(test_cfg);
+
+    fprintf(stdout,
+            "PMEM kind and configuration was successfully created and destroyed.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_cpp_allocator.cpp b/memkind-1.10.0/examples/pmem_cpp_allocator.cpp
new file mode 100644
index 0000000..e3decd1
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_cpp_allocator.cpp
@@ -0,0 +1,167 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#include "pmem_allocator.h"
+
+#include <sys/stat.h>
+#include <iostream>
+#include <vector>
+#include <list>
+#include <map>
+#include <string>
+#include <scoped_allocator>
+#include <cassert>
+
+#define STL_VECTOR_TEST
+#define STL_LIST_TEST
+#if _GLIBCXX_USE_CXX11_ABI
+#define STL_VEC_STRING_TEST
+#define STL_MAP_INT_STRING_TEST
+#endif
+
+void cpp_allocator_test(const char *pmem_directory)
+{
+    std::cout << "TEST SCOPE: HELLO" << std::endl;
+
+    size_t pmem_max_size = 1024*1024*1024;
+
+#ifdef STL_VECTOR_TEST
+    {
+        std::cout << "VECTOR OPEN" << std::endl;
+        libmemkind::pmem::allocator<int> alc{ pmem_directory, pmem_max_size };
+        std::vector<int, libmemkind::pmem::allocator<int>> vector{ alc };
+
+        for (int i = 0; i < 20; ++i) {
+            vector.push_back(0xDEAD + i);
+            assert(vector.back() == 0xDEAD + i);
+        }
+
+        std::cout << "VECTOR CLOSE" << std::endl;
+    }
+#endif
+
+#ifdef STL_LIST_TEST
+    {
+        std::cout << "LIST OPEN" << std::endl;
+        libmemkind::pmem::allocator<int> alc{ pmem_directory, pmem_max_size };
+        std::list<int, libmemkind::pmem::allocator<int>> list{ alc };
+
+        const int nx2 = 4;
+        for (int i = 0; i < nx2; ++i) {
+            list.emplace_back(0xBEAC011 + i);
+            assert(list.back() == 0xBEAC011 + i);
+        }
+
+        for (int i = 0; i < nx2; ++i) {
+            list.pop_back();
+        }
+
+        std::cout << "LIST CLOSE" << std::endl;
+    }
+#endif
+
+#ifdef STL_VEC_STRING_TEST
+    {
+        std::cout << "STRINGED VECTOR OPEN" << std::endl;
+        typedef libmemkind::pmem::allocator<char> str_alloc_t;
+        typedef std::basic_string<char, std::char_traits<char>, str_alloc_t>
+        pmem_string;
+        typedef libmemkind::pmem::allocator<pmem_string> vec_alloc_t;
+
+        vec_alloc_t vec_alloc{ pmem_directory, pmem_max_size };
+        str_alloc_t str_alloc{ pmem_directory, pmem_max_size };
+
+        std::vector<pmem_string, std::scoped_allocator_adaptor<vec_alloc_t> >
+        vec{ std::scoped_allocator_adaptor<vec_alloc_t>(vec_alloc) };
+
+        pmem_string arg{ "Very very loooong striiiing", str_alloc };
+
+        vec.push_back(arg);
+        assert(vec.back() == arg);
+
+        std::cout << "STRINGED VECTOR CLOSE" << std::endl;
+    }
+
+#endif
+
+#ifdef STL_MAP_INT_STRING_TEST
+    {
+        std::cout << "INT_STRING MAP OPEN" << std::endl;
+        typedef std::basic_string<char, std::char_traits<char>, libmemkind::pmem::allocator<char>>
+                pmem_string;
+        typedef int key_t;
+        typedef pmem_string value_t;
+        typedef libmemkind::pmem::allocator<std::pair<const key_t, value_t>>
+                                                                          allocator_t;
+        typedef std::map<key_t, value_t, std::less<key_t>, std::scoped_allocator_adaptor<allocator_t>>
+                map_t;
+
+        allocator_t allocator( pmem_directory, pmem_max_size );
+
+        value_t source_str1("Lorem ipsum dolor ", allocator);
+        value_t source_str2("sit amet consectetuer adipiscing elit", allocator );
+
+        map_t target_map{ std::scoped_allocator_adaptor<allocator_t>(allocator) };
+
+        target_map[key_t(165)] = source_str1;
+        assert(target_map[key_t(165)] == source_str1);
+        target_map[key_t(165)] = source_str2;
+        assert(target_map[key_t(165)] == source_str2);
+
+        std::cout << "INT_STRING MAP CLOSE" << std::endl;
+    }
+#endif
+    std::cout << "TEST SCOPE: GOODBYE" << std::endl;
+}
+
+int main(int argc, char *argv[])
+{
+    const char *pmem_directory = "/tmp/";
+
+    if (argc > 2) {
+        std::cerr << "Usage: pmem_cpp_allocator [directory path]\n"
+                  << "\t[directory path] - directory where temporary file is created (default = \"/tmp/\")"
+                  << std::endl;
+        return 0;
+    } else if (argc == 2) {
+        struct stat st;
+        if (stat(argv[1], &st) != 0 || !S_ISDIR(st.st_mode)) {
+            fprintf(stderr,"%s : Invalid path to pmem kind directory\n", argv[1]);
+            return 1;
+        }
+        pmem_directory = argv[1];
+    }
+
+    cpp_allocator_test(pmem_directory);
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_detect_kind.c b/memkind-1.10.0/examples/pmem_detect_kind.c
new file mode 100644
index 0000000..ba4c609
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_detect_kind.c
@@ -0,0 +1,165 @@
+/*
+ * Copyright (c) 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+
+#define MALLOC_SIZE 512U
+#define REALLOC_SIZE 2048U
+#define ALLOC_LIMIT  1000U
+
+static void *alloc_buffer[ALLOC_LIMIT];
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+static int allocate_pmem_and_default_kind(struct memkind *pmem_kind)
+{
+    unsigned i;
+    for(i = 0; i < ALLOC_LIMIT; i++) {
+        if (i%2)
+            alloc_buffer[i] = memkind_malloc(pmem_kind, MALLOC_SIZE);
+        else
+            alloc_buffer[i] = memkind_malloc(MEMKIND_DEFAULT, MALLOC_SIZE);
+
+        if (!alloc_buffer[i]) {
+            return 1;
+        }
+    }
+
+    return 0;
+}
+
+static int realloc_using_get_kind_only_on_pmem()
+{
+    unsigned i;
+    for(i = 0; i < ALLOC_LIMIT; i++) {
+        if (memkind_detect_kind(alloc_buffer[i]) != MEMKIND_DEFAULT) {
+            void *temp = memkind_realloc(NULL, alloc_buffer[i], REALLOC_SIZE);
+            if (!temp) {
+                return 1;
+            }
+            alloc_buffer[i] = temp;
+        }
+    }
+
+    return 0;
+}
+
+
+static int verify_allocation_size(struct memkind *pmem_kind, size_t pmem_size)
+{
+    unsigned i;
+    for(i = 0; i < ALLOC_LIMIT; i++) {
+        void *val = alloc_buffer[i];
+        if (i%2) {
+            if (memkind_malloc_usable_size(pmem_kind, val) != pmem_size ) {
+                return 1;
+            }
+        } else {
+            if (memkind_malloc_usable_size(MEMKIND_DEFAULT, val) != MALLOC_SIZE) {
+                return 1;
+            }
+        }
+    }
+
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to distinguish allocation from different kinds using detect kind function"
+            "\nPMEM kind directory: %s\n", path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Allocate to PMEM and DEFAULT kind.\n");
+
+    if (allocate_pmem_and_default_kind(pmem_kind)) {
+        fprintf(stderr, "allocate_pmem_and_default_kind().\n");
+        return 1;
+    }
+
+    if (verify_allocation_size(pmem_kind, MALLOC_SIZE)) {
+        fprintf(stderr, "verify_allocation_size() before resize.\n");
+        return 1;
+    }
+
+    fprintf(stdout,
+            "Reallocate memory only on PMEM kind using memkind_detect_kind().\n");
+
+    if (realloc_using_get_kind_only_on_pmem()) {
+        fprintf(stderr, "realloc_using_get_kind_only_on_pmem().\n");
+        return 1;
+    }
+
+    if (verify_allocation_size(pmem_kind, REALLOC_SIZE)) {
+        fprintf(stderr, "verify_allocation_size() after resize.\n");
+        return 1;
+    }
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory from PMEM kind was successfully reallocated.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_free_with_unknown_kind.c b/memkind-1.10.0/examples/pmem_free_with_unknown_kind.c
new file mode 100644
index 0000000..cedd18a
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_free_with_unknown_kind.c
@@ -0,0 +1,115 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+static const size_t PMEM_PART_SIZE = MEMKIND_PMEM_MIN_SIZE + 4 * 1024;
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char **argv)
+{
+    const size_t size = 512;
+    struct memkind *pmem_kind = NULL;
+    const int arraySize = 100;
+    char *ptr[100] = { NULL };
+    int i = 0;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use memkind_free with unknown kind as a parameter.\n");
+
+    err = memkind_create_pmem(path, PMEM_PART_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    for (i = 0; i < arraySize; ++i) {
+        if (i < 50) {
+            ptr[i] = memkind_malloc(MEMKIND_DEFAULT, size);
+            if (ptr[i] == NULL) {
+                fprintf(stderr, "Unable to allocate memkind default.\n");
+                return 1;
+            }
+        } else {
+            ptr[i] = memkind_malloc(pmem_kind, size);
+            if (ptr[i] == NULL) {
+                fprintf(stderr, "Unable to allocate pmem.\n");
+                return 1;
+            }
+        }
+    }
+    fprintf(stdout,
+            "Memory was successfully allocated in default kind and pmem kind.\n");
+
+    sprintf(ptr[10], "Hello world from standard memory - ptr[10].\n");
+    sprintf(ptr[40], "Hello world from standard memory - ptr[40].\n");
+    sprintf(ptr[80], "Hello world from persistent memory - ptr[80].\n");
+
+    fprintf(stdout, "%s", ptr[10]);
+    fprintf(stdout, "%s", ptr[40]);
+    fprintf(stdout, "%s", ptr[80]);
+
+    fprintf(stdout, "Free memory without specifying kind.\n");
+    for (i = 0; i < arraySize; ++i) {
+        memkind_free(NULL, ptr[i]);
+    }
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully released.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_kinds.c b/memkind-1.10.0/examples/pmem_kinds.c
new file mode 100644
index 0000000..0571428
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_kinds.c
@@ -0,0 +1,119 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+#define NUM_KINDS 10
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kinds[NUM_KINDS] = {NULL};
+    struct memkind *pmem_kind = NULL;
+    struct memkind *pmem_kind_unlimited = NULL;
+
+    int err = 0, i = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to create and destroy pmem kind with defined or unlimited size."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create first PMEM partition with specific size
+    err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Create second PMEM partition with unlimited size
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Destroy both PMEM partitions
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    err = memkind_destroy_kind(pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Create many PMEM kinds with same specific size
+    for (i = 0; i < NUM_KINDS; i++) {
+        err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kinds[i]);
+        if (err) {
+            print_err_message(err);
+            return 1;
+        }
+    }
+
+    // Destroy all of them
+    for (i = 0; i < NUM_KINDS; i++) {
+        err = memkind_destroy_kind(pmem_kinds[i]);
+        if (err) {
+            print_err_message(err);
+            return 1;
+        }
+    }
+
+    fprintf(stdout, "PMEM kinds have been successfully created and destroyed.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_malloc.c b/memkind-1.10.0/examples/pmem_malloc.c
new file mode 100644
index 0000000..61e201d
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_malloc.c
@@ -0,0 +1,129 @@
+/*
+ * Copyright (c) 2015 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to allocate memory and possibility to exceed pmem kind size."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create PMEM partition with specific size
+    err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str1 = NULL;
+    char *pmem_str2 = NULL;
+    char *pmem_str3 = NULL;
+    char *pmem_str4 = NULL;
+
+    // Allocate 512 Bytes of 32 MB available
+    pmem_str1 = (char *)memkind_malloc(pmem_kind, 512);
+    if (pmem_str1 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str1).\n");
+        return 1;
+    }
+
+    // Allocate 8 MB of 31.9 MB available
+    pmem_str2 = (char *)memkind_malloc(pmem_kind, 8 * 1024 * 1024);
+    if (pmem_str2 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str2).\n");
+        return 1;
+    }
+
+    // Allocate 16 MB of 23.9 MB available
+    pmem_str3 = (char *)memkind_malloc(pmem_kind, 16 * 1024 * 1024);
+    if (pmem_str3 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str3).\n");
+        return 1;
+    }
+
+    // Allocate 16 MB of 7.9 MB available - Out Of Memory expected
+    pmem_str4 = (char *)memkind_malloc(pmem_kind, 16 * 1024 * 1024);
+    if (pmem_str4 != NULL) {
+        fprintf(stderr,
+                "Failure, this allocation should not be possible (expected result was NULL).\n");
+        return 1;
+    }
+
+    sprintf(pmem_str1, "Hello world from pmem - pmem_str1.\n");
+    sprintf(pmem_str2, "Hello world from pmem - pmem_str2.\n");
+    sprintf(pmem_str3, "Hello world from persistent memory - pmem_str3.\n");
+
+    fprintf(stdout, "%s", pmem_str1);
+    fprintf(stdout, "%s", pmem_str2);
+    fprintf(stdout, "%s", pmem_str3);
+
+    memkind_free(pmem_kind, pmem_str1);
+    memkind_free(pmem_kind, pmem_str2);
+    memkind_free(pmem_kind, pmem_str3);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_malloc_unlimited.c b/memkind-1.10.0/examples/pmem_malloc_unlimited.c
new file mode 100644
index 0000000..a267069
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_malloc_unlimited.c
@@ -0,0 +1,101 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind_unlimited = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to allocate memory with unlimited kind size."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create PMEM partition with unlimited size
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str10 = NULL;
+    char *pmem_str11 = NULL;
+
+    // Huge allocation
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_unlimited, 32 * 1024 * 1024);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    // Another huge allocation, kind size is only limited by OS resources
+    pmem_str11 = (char *)memkind_malloc(pmem_kind_unlimited, 32 * 1024 * 1024);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11).\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind_unlimited, pmem_str10);
+    memkind_free(pmem_kind_unlimited, pmem_str11);
+
+    err = memkind_destroy_kind(pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout, "Memory was successfully allocated and released.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/examples/pmem_multithreads.c b/memkind-1.10.0/examples/pmem_multithreads.c
new file mode 100644
index 0000000..fe72a43
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_multithreads.c
@@ -0,0 +1,138 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <pthread.h>
+#include <unistd.h>
+
+#define PMEM_MAX_SIZE (1024 * 1024 * 32)
+#define NUM_THREADS 10
+
+static char path[PATH_MAX]="/tmp/";
+
+void *thread_ind(void *arg);
+
+static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
+static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use multithreading with independent pmem kinds."
+            "\nPMEM kind directory: %s\n", path);
+
+    pthread_t pmem_threads[NUM_THREADS];
+    int t;
+
+    // Create many independent threads
+    for (t = 0; t < NUM_THREADS; t++) {
+        if (pthread_create(&pmem_threads[t], NULL, thread_ind, NULL) != 0) {
+            fprintf(stderr, "Unable to create a thread.\n");
+            return 1;
+        }
+    }
+
+    sleep(1);
+    if (pthread_cond_broadcast(&cond) != 0) {
+        fprintf(stderr, "Unable to broadcast a condition.\n");
+        return 1;
+    }
+
+    for (t = 0; t < NUM_THREADS; t++) {
+        if (pthread_join(pmem_threads[t], NULL) != 0) {
+            fprintf(stderr, "Thread join failed.\n");
+            return 1;
+        }
+    }
+
+    fprintf(stdout, "Threads successfully allocated memory in the PMEM kinds.\n");
+
+    return 0;
+}
+
+void *thread_ind(void *arg)
+{
+    struct memkind *pmem_kind;
+
+    if (pthread_mutex_lock(&mutex) != 0) {
+        fprintf(stderr, "Failed to acquire mutex.\n");
+        return NULL;
+    }
+    if (pthread_cond_wait(&cond, &mutex) != 0) {
+        fprintf(stderr, "Failed to block mutex on condition.\n");
+        return NULL;
+    }
+    if (pthread_mutex_unlock(&mutex) != 0) {
+        fprintf(stderr, "Failed to release mutex.\n");
+        return NULL;
+    }
+
+    int err = memkind_create_pmem(path, PMEM_MAX_SIZE, &pmem_kind);
+    if (err) {
+        print_err_message(err);
+        return NULL;
+    }
+
+    void *test = memkind_malloc(pmem_kind, 32);
+    if (test == NULL) {
+        fprintf(stderr, "Unable to allocate pmem (test) in thread.\n");
+        return NULL;
+    }
+
+    memkind_free(pmem_kind, test);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        print_err_message(err);
+    }
+
+    return NULL;
+}
diff --git a/memkind-1.10.0/examples/pmem_multithreads_onekind.c b/memkind-1.10.0/examples/pmem_multithreads_onekind.c
new file mode 100644
index 0000000..ed7bc07
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_multithreads_onekind.c
@@ -0,0 +1,167 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <pthread.h>
+#include <unistd.h>
+
+#define NUM_THREADS 10
+#define NUM_ALLOCS 100
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+struct arg_struct {
+    int id;
+    struct memkind *kind;
+    int **ptr;
+};
+
+void *thread_onekind(void *arg);
+
+static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
+static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind_unlimited = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows how to use multithreading with one main pmem kind."
+            "\nPMEM kind directory: %s\n", path);
+
+    // Create PMEM partition with unlimited size
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    // Create a few threads which will access to our main pmem_kind
+    pthread_t pmem_threads[NUM_THREADS];
+    int *pmem_tint[NUM_THREADS][NUM_ALLOCS];
+    int t = 0, i = 0;
+
+    struct arg_struct *args[NUM_THREADS];
+
+    for (t = 0; t<NUM_THREADS; t++) {
+        args[t] = malloc(sizeof(struct arg_struct));
+        args[t]->id = t;
+        args[t]->ptr = &pmem_tint[t][0];
+        args[t]->kind = pmem_kind_unlimited;
+
+        if (pthread_create(&pmem_threads[t], NULL, thread_onekind,
+                           (void *)args[t])!= 0) {
+            fprintf(stderr, "Unable to create a thread.\n");
+            return 1;
+        }
+    }
+
+    sleep(1);
+    if (pthread_cond_broadcast(&cond) != 0) {
+        fprintf(stderr, "Unable to broadcast a condition.\n");
+        return 1;
+    }
+
+    for (t = 0; t < NUM_THREADS; t++) {
+        if (pthread_join(pmem_threads[t], NULL) != 0) {
+            fprintf(stderr, "Thread join failed.\n");
+            return 1;
+        }
+    }
+
+    // Check if we can read the values outside of threads and free resources
+    for (t = 0; t < NUM_THREADS; t++) {
+        for (i = 0; i < NUM_ALLOCS; i++) {
+            if(*pmem_tint[t][i] != t) {
+                fprintf(stderr,
+                        "pmem_tint value has not been saved correctly in the thread.\n");
+                return 1;
+            }
+            memkind_free(args[t]->kind, *(args[t]->ptr+i));
+        }
+        free(args[t]);
+    }
+
+    fprintf(stdout, "Threads successfully allocated memory in the PMEM kind.\n");
+
+    return 0;
+}
+
+void *thread_onekind(void *arg)
+{
+    struct arg_struct *args = (struct arg_struct *)arg;
+    int i;
+
+    if (pthread_mutex_lock(&mutex) != 0) {
+        fprintf(stderr, "Failed to acquire mutex.\n");
+        return NULL;
+    }
+    if (pthread_cond_wait(&cond, &mutex) != 0) {
+        fprintf(stderr, "Failed to block mutex on condition.\n");
+        return NULL;
+    }
+    if (pthread_mutex_unlock(&mutex) != 0) {
+        fprintf(stderr, "Failed to release mutex.\n");
+        return NULL;
+    }
+
+    // Lets alloc int and put there thread ID
+    for (i = 0; i < NUM_ALLOCS; i++) {
+        *(args->ptr+i) = (int *)memkind_malloc(args->kind, sizeof(int));
+        if (*(args->ptr+i) == NULL) {
+            fprintf(stderr, "Unable to allocate pmem int.\n");
+            return NULL;
+        }
+        **(args->ptr+i) = args->id;
+    }
+
+    return NULL;
+}
diff --git a/memkind-1.10.0/examples/pmem_usable_size.c b/memkind-1.10.0/examples/pmem_usable_size.c
new file mode 100644
index 0000000..92dd078
--- /dev/null
+++ b/memkind-1.10.0/examples/pmem_usable_size.c
@@ -0,0 +1,159 @@
+/*
+ * Copyright (c) 2018 - 2019 Intel Corporation
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *     * Redistributions of source code must retain the above copyright
+ *       notice, this list of conditions and the following disclaimer.
+ *
+ *     * Redistributions in binary form must reproduce the above copyright
+ *       notice, this list of conditions and the following disclaimer in
+ *       the documentation and/or other materials provided with the
+ *       distribution.
+ *
+ *     * Neither the name of Intel Corporation nor the names of its
+ *       contributors may be used to endorse or promote products derived
+ *       from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY LOG OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#define MB (1024 * 1024)
+
+static char path[PATH_MAX]="/tmp/";
+
+static void print_err_message(int err)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(err, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    fprintf(stderr, "%s\n", error_message);
+}
+
+int main(int argc, char *argv[])
+{
+    struct memkind *pmem_kind_unlimited = NULL;
+    int err = 0;
+
+    if (argc > 2) {
+        fprintf(stderr, "Usage: %s [pmem_kind_dir_path]\n", argv[0]);
+        return 1;
+    } else if (argc == 2 && (realpath(argv[1], path) == NULL)) {
+        fprintf(stderr, "Incorrect pmem_kind_dir_path %s\n", argv[1]);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "This example shows difference between the expected and the actual allocation size."
+            "\nPMEM kind directory: %s\n", path);
+
+    err = memkind_create_pmem(path, 0, &pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    char *pmem_str10 = NULL;
+    char *pmem_str11 = NULL;
+    char *pmem_str12 = NULL;
+
+    // 32 bytes allocation
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_unlimited, 32);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str10) != 32) {
+        fprintf(stderr, "Wrong usable size for small allocation (pmem_str10).\n");
+        return 1;
+    }
+
+    // 31 bytes allocation
+    pmem_str11 = (char *)memkind_malloc(pmem_kind_unlimited, 31);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation, its 32 again
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str11) != 32) {
+        fprintf(stderr, "Wrong usable size for small allocation (pmem_str11).\n");
+        return 1;
+    }
+
+    // 33 bytes allocation
+    pmem_str12 = (char *)memkind_malloc(pmem_kind_unlimited, 33);
+    if (pmem_str12 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str12).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation, its 48 now
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str12) != 48) {
+        fprintf(stderr, "Wrong usable size for small allocation (pmem_str12).\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind_unlimited, pmem_str10);
+    memkind_free(pmem_kind_unlimited, pmem_str11);
+    memkind_free(pmem_kind_unlimited, pmem_str12);
+
+    // 5MB allocation
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_unlimited, 5 * MB);
+    if (pmem_str10 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str10).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str10) != 5 * MB) {
+        fprintf(stderr, "Wrong usable size for large allocation (pmem_str10).\n");
+        return 1;
+    }
+
+    // 5MB + 1B allocation
+    pmem_str11 = (char *)memkind_malloc(pmem_kind_unlimited, 5 * MB + 1);
+    if (pmem_str11 == NULL) {
+        fprintf(stderr, "Unable to allocate pmem string (pmem_str11).\n");
+        return 1;
+    }
+
+    // Check real usable size for this allocation, its 6MB now
+    if (memkind_malloc_usable_size(pmem_kind_unlimited, pmem_str11) != 6 * MB) {
+        fprintf(stderr, "Wrong usable size for large allocation (pmem_str11).\n");
+        return 1;
+    }
+
+    memkind_free(pmem_kind_unlimited, pmem_str10);
+    memkind_free(pmem_kind_unlimited, pmem_str11);
+
+    err = memkind_destroy_kind(pmem_kind_unlimited);
+    if (err) {
+        print_err_message(err);
+        return 1;
+    }
+
+    fprintf(stdout,
+            "The real size of the allocation has been successfully read.\n");
+
+    return 0;
+}
diff --git a/memkind-1.10.0/include/hbw_allocator.h b/memkind-1.10.0/include/hbw_allocator.h
new file mode 100644
index 0000000..ec9c1c0
--- /dev/null
+++ b/memkind-1.10.0/include/hbw_allocator.h
@@ -0,0 +1,134 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <hbwmalloc.h>
+
+#include <stddef.h>
+#include <new>
+/*
+ * Header file for the C++ allocator compatible with the C++ standard library allocator concepts.
+ * More details in hbwallocator(3) man page.
+ * Note: memory heap management is based on hbwmalloc, refer to the hbwmalloc man page for more information.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+namespace hbw
+{
+
+    template <class T>
+    class allocator
+    {
+    public:
+        /*
+         *  Public member types required and defined by the standard library allocator concepts.
+         */
+        typedef size_t size_type;
+        typedef ptrdiff_t difference_type;
+        typedef T *pointer;
+        typedef const T *const_pointer;
+        typedef T &reference;
+        typedef const T &const_reference;
+        typedef T value_type;
+
+        template <class U>
+        struct rebind {
+            typedef hbw::allocator<U> other;
+        };
+
+        /*
+         *  Public member functions required and defined by the standard library allocator concepts.
+         */
+        allocator() throw() { }
+
+        template <class U>
+        allocator(const allocator<U> &) throw() { }
+
+        ~allocator() throw() { }
+
+        pointer address(reference x) const
+        {
+            return &x;
+        }
+
+        const_pointer address(const_reference x) const
+        {
+            return &x;
+        }
+
+        /*
+         *  Allocates n*sizeof(T) bytes of high bandwidth memory using hbw_malloc().
+         *  Throws std::bad_alloc when cannot allocate memory.
+         */
+        pointer allocate(size_type n, const void * = 0)
+        {
+            if (n > this->max_size()) {
+                throw std::bad_alloc();
+            }
+            pointer result = static_cast<pointer>(hbw_malloc(n * sizeof(T)));
+            if (!result) {
+                throw std::bad_alloc();
+            }
+            return result;
+        }
+
+        /*
+         *  Deallocates memory associated with pointer returned by allocate() using hbw_free().
+         */
+        void deallocate(pointer p, size_type n)
+        {
+            hbw_free(static_cast<void *>(p));
+        }
+
+        size_type max_size() const throw()
+        {
+            return size_t(-1) / sizeof(T);
+        }
+
+        void construct(pointer p, const_reference val)
+        {
+            ::new(p) value_type(val);
+        }
+
+        void destroy(pointer p)
+        {
+            p->~T();
+        }
+    };
+
+    template <class T, class U>
+    bool operator==(const allocator<T> &, const allocator<U> &)
+    {
+        return true;
+    }
+
+    template <class T, class U>
+    bool operator!=(const allocator<T> &, const allocator<U> &)
+    {
+        return false;
+    }
+
+}
diff --git a/memkind-1.10.0/include/hbwmalloc.h b/memkind-1.10.0/include/hbwmalloc.h
new file mode 100644
index 0000000..3e8eac8
--- /dev/null
+++ b/memkind-1.10.0/include/hbwmalloc.h
@@ -0,0 +1,245 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stdlib.h>
+/*
+ *  Header file for the high bandwidth memory interface.
+ *
+ *  This file defines the external API's and enumerations for the
+ *  hbwmalloc library.  These interfaces define a heap manager that
+ *  targets the high bandwidth memory numa nodes.
+ *
+ *  hbwmalloc.h functionality is considered as stable API (STANDARD API).
+ *
+ *  Please read hbwmalloc(3) man page for or more details.
+ */
+
+/*
+ *  Fallback policy.
+ *
+ *  Policy that determines behavior when there is not enough free high
+ *  bandwidth memory to satisfy a user request.  This enum is used with
+ *  hbw_get_policy() and hbw_set_policy().
+ */
+typedef enum {
+    /*
+     *  If insufficient high bandwidth memory pages on nearest NUMA node are
+     *  available then OOM killer will be triggered.
+     */
+    HBW_POLICY_BIND = 1,
+    /*
+     *  If insufficient high bandwidth memory pages are available fall
+     *  back on standard memory pages.
+     */
+    HBW_POLICY_PREFERRED = 2,
+    /*
+     *  Interleave pages across high bandwidth nodes. If insufficient memory
+     *  pages are available then OOM killer will be triggered.
+     */
+    HBW_POLICY_INTERLEAVE = 3,
+    /*
+     *  If insufficient high bandwidth memory pages are available then
+     *  OOM killer will be triggered.
+     */
+    HBW_POLICY_BIND_ALL = 4,
+} hbw_policy_t;
+
+/*
+ *  Page size selection.
+ *
+ *  The hbw_posix_memalign_psize() API gives the user the option to
+ *  select the page size from this enumerated list.
+ */
+typedef enum {
+
+    /*
+     * The four kilobyte page size option. Note that with transparent huge
+     * pages enabled these allocations may be promoted by the operating system
+     * to two megabyte pages.
+     */
+    HBW_PAGESIZE_4KB           = 1,
+
+    /*
+     * The two megabyte page size option.
+     */
+    HBW_PAGESIZE_2MB           = 2,
+    /*
+     * This option is deprecated.
+     * Allocate high bandwidth memory using 1GB chunks backed by huge pages.
+     */
+    HBW_PAGESIZE_1GB_STRICT    = 3,
+
+    /*
+     * This option is deprecated.
+     * Allocate high bandwidth memory using 1GB chunks backed by huge pages.
+     */
+    HBW_PAGESIZE_1GB           = 4,
+
+    /*
+     * Helper representing value of the last enum element incremented by 1.
+     * Shall not be treated as a valid value for functions taking hbw_pagesize_t
+     * as parameter.
+     */
+    HBW_PAGESIZE_MAX_VALUE
+} hbw_pagesize_t;
+
+/*
+ * Flags for hbw_verify_ptr function
+ */
+enum {
+
+    /*
+     * This option touches first byte of all pages in address range starting from "addr" to "addr" + "size"
+     * by read and write (so the content will be overwritten by the same data as it was read).
+     */
+    HBW_TOUCH_PAGES      = (1 << 0)
+};
+
+/*
+ * Returns the current fallback policy when insufficient high bandwidth memory
+ * is available.
+ */
+hbw_policy_t hbw_get_policy(void);
+
+/*
+ * Sets the current fallback policy. The policy can be modified only once in
+ * the lifetime of an application and before calling hbw_*alloc() or
+ * hbw_posix_memalign*() function.
+ * Note: If the policy is not set, than HBW_POLICY_PREFERRED will be used by
+ * default.
+ *
+ * Returns:
+ *   0: on success
+ *   EPERM: if hbw_set_policy () was called more than once
+ *   EINVAL: if mode argument was neither HBW_POLICY_PREFERRED, HBW_POLICY_BIND, HBW_POLICY_BIND_ALL nor HBW_POLICY_INTERLEAVE
+ */
+int hbw_set_policy(hbw_policy_t mode);
+
+/*
+ * Verifies high bandwidth memory availability.
+ * Returns:
+ *   0: if high bandwidth memory is available
+ *   ENODEV: if high-bandwidth memory is unavailable.
+ */
+int hbw_check_available(void);
+
+/*
+ * Verifies if allocated memory fully fall into high bandwidth memory.
+ * Returns:
+ *   0: if memory in address range from "addr" to "addr" + "size" is allocated in high bandwidth memory
+ *   -1: if any region of memory was not allocated in high bandwidth memory
+ *   EINVAL: if addr is NULL, size equals 0 or flags contained unsupported bit set
+ *   EFAULT: could not verify memory
+ */
+int hbw_verify_memory_region(void *addr, size_t size, int flags);
+
+/*
+ * Allocates size bytes of uninitialized high bandwidth memory.
+ * The allocated space is  suitably  aligned (after  possible  pointer
+ * coercion) for storage of any type of object. If size is zero then
+ * hbw_malloc() returns NULL.
+ */
+void *hbw_malloc(size_t size);
+
+/*
+ * Allocates space for num objects in high bandwidth memory, each size bytes
+ * in length.
+ * The result is identical to calling hbw_malloc() with an argument of
+ * num*size, with the exception that the allocated memory is explicitly
+ * initialized to zero bytes.
+ * If num or size is 0, then hbw_calloc() returns NULL.
+ */
+void *hbw_calloc(size_t num, size_t size);
+
+/*
+ * Allocates size bytes of high bandwidth memory such that the allocation's
+ * base address is an even multiple of alignment, and returns the allocation
+ * in the value pointed to by memptr.  The requested alignment must be a power
+ * of 2 at least as large as sizeof(void *).
+ * Returns:
+ *   0: on success
+ *   ENOMEM: if there was insufficient memory to satisfy the request
+ *   EINVAL: if the alignment parameter was not a power of two, or was less than sizeof(void *)
+ */
+int hbw_posix_memalign(void **memptr, size_t alignment, size_t size);
+
+/*
+ * Allocates size bytes of high bandwidth memory such that the allocation's
+ * base address is an even multiple of alignment, and returns the allocation
+ * in the value pointed to by memptr. The requested alignment must be a power
+ * of 2 at least as large as sizeof(void  *). The memory will be allocated
+ * using pages determined by the pagesize variable.
+ * Returns:
+ *   0: on success
+ *   ENOMEM: if there was insufficient memory to satisfy the request
+ *   EINVAL: if the alignment parameter was not a power of two, or was less than sizeof(void *)
+ */
+int hbw_posix_memalign_psize(void **memptr, size_t alignment, size_t size,
+                             hbw_pagesize_t pagesize);
+
+/*
+ * Changes the size of the previously allocated memory referenced by ptr to
+ * size bytes of the specified kind. The contents of the memory are unchanged
+ * up to the lesser of the new and old size.
+ * If the new size is larger, the contents of the newly allocated portion
+ * of the memory are undefined.
+ * Upon success, the memory referenced by ptr is freed and a pointer to the
+ * newly allocated high bandwidth memory is returned.
+ * Note: memkind_realloc() may move the memory allocation, resulting in a
+ * different return value than ptr.
+ * If ptr is NULL, the hbw_realloc() function behaves identically to
+ * hbw_malloc() for the specified size. The address ptr, if not NULL,
+ * was returned by a previous call to hbw_malloc(), hbw_calloc(),
+ * hbw_realloc(), or hbw_posix_memalign(). Otherwise, or if hbw_free(ptr)
+ * was called before, undefined behavior occurs.
+ * Note: hbw_realloc() cannot be used with a pointer returned by
+ * hbw_posix_memalign_psize().
+ */
+void *hbw_realloc(void *ptr, size_t size);
+
+/*
+ * Causes the allocated memory referenced by ptr to be made
+ * available for future allocations. If ptr is NULL, no action occurs.
+ * The address ptr, if not NULL, must have been returned by a previous call
+ * to hbw_malloc(), hbw_calloc(), hbw_realloc(), hbw_posix_memalign(), or
+ * hbw_posix_memalign_psize(). Otherwise, if hbw_free(ptr) was called before,
+ * undefined behavior occurs.
+ */
+void hbw_free(void *ptr);
+
+/*
+ * Returns the number of usable bytes in the block pointed to by ptr, a pointer
+ * to a block of memory allocated by hbw_malloc(), hbw_calloc(), hbw_realloc(),
+ * hbw_posix_memalign(), or hbw_posix_memalign_psize().
+ */
+size_t hbw_malloc_usable_size(void *ptr);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind.h b/memkind-1.10.0/include/memkind.h
new file mode 100644
index 0000000..0be640e
--- /dev/null
+++ b/memkind-1.10.0/include/memkind.h
@@ -0,0 +1,433 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <sys/types.h>
+
+/**
+ * Header file for the memkind heap manager.
+ * More details in memkind(3) man page.
+ *
+ * API standards are described in memkind(3) man page.
+ */
+
+/// \warning EXPERIMENTAL API
+#define _MEMKIND_BIT(N) (1ull << N)
+
+/// \brief Memkind memory types
+/// \warning EXPERIMENTAL API
+typedef enum memkind_memtype_t {
+
+    /**
+     * Select standard memory, the same as process use.
+     */
+    MEMKIND_MEMTYPE_DEFAULT = _MEMKIND_BIT(0),
+
+    /**
+     * Select high bandwidth memory (HBM).
+     * There must be at least two memories with different bandwidth to
+     * determine the HBM.
+     */
+    MEMKIND_MEMTYPE_HIGH_BANDWIDTH = _MEMKIND_BIT(1)
+
+} memkind_memtype_t;
+
+#undef _MEMKIND_BIT
+
+/// \brief Memkind policy
+/// \warning EXPERIMENTAL API
+typedef enum memkind_policy_t {
+
+    /**
+     * Allocate local memory.
+     * If there is not enough memory to satisfy the request errno is set to
+     * ENOMEM and the allocated pointer is set to NULL.
+     */
+    MEMKIND_POLICY_BIND_LOCAL = 0,
+
+    /**
+     * Memory locality is ignored.
+     * If there is not enough memory to satisfy the request errno is set to
+     * ENOMEM and the allocated pointer is set to NULL.
+     */
+    MEMKIND_POLICY_BIND_ALL,
+
+    /**
+     * Allocate preferred memory that is local.
+     * If there is not enough preferred memory to satisfy the request or
+     * preferred memory is not available, the allocation will fall back on any
+     * other memory.
+     */
+    MEMKIND_POLICY_PREFERRED_LOCAL,
+
+    /**
+     * Interleave allocation across local memory.
+     * For n memory types the allocation will be interleaved across all of
+     * them.
+     */
+    MEMKIND_POLICY_INTERLEAVE_LOCAL,
+
+    /**
+     * Interleave allocation. Locality is ignored.
+     * For n memory types the allocation will be interleaved across all of
+     * them.
+     */
+    MEMKIND_POLICY_INTERLEAVE_ALL,
+
+    /**
+     * Max policy value.
+     */
+    MEMKIND_POLICY_MAX_VALUE
+
+} memkind_policy_t;
+
+/// \brief Memkind bits definition
+/// \warning EXPERIMENTAL API
+/// \note The bits specify flags and masks. Bits <0,1,2,...,7> are reserved for page size, where page sizes are encoded
+///       by base-2 logarithm. If the page size bits are set to zero value, than default page size will be used.
+typedef enum memkind_bits_t {
+    MEMKIND_MASK_PAGE_SIZE_2MB = 21ull,  /**<  Allocations backed by 2 MB page size (2^21 = 2MB) */
+} memkind_bits_t;
+
+/// \brief Memkind type definition
+/// \warning EXPERIMENTAL API
+typedef struct memkind *memkind_t;
+
+
+/// \brief Memkind constant values
+/// \warning EXPERIMENTAL API
+enum memkind_const {
+    MEMKIND_MAX_KIND = 512,                     /**<  Maximum number of kinds */
+    MEMKIND_ERROR_MESSAGE_SIZE = 128,           /**<  Error message size */
+    MEMKIND_PMEM_MIN_SIZE = (1024 * 1024 * 16)  /**<  The minimum size which allows to limit the file-backed memory partition */
+};
+
+/// \brief Memkind operation statuses
+/// \warning EXPERIMENTAL API
+enum {
+    MEMKIND_SUCCESS = 0,                        /**<  Operation success */
+    MEMKIND_ERROR_UNAVAILABLE = -1,             /**<  Error: Memory kind is not available */
+    MEMKIND_ERROR_MBIND = -2,                   /**<  Error: Call to mbind() failed */
+    MEMKIND_ERROR_MMAP  = -3,                   /**<  Error: Call to mmap() failed */
+    MEMKIND_ERROR_MALLOC = -6,                  /**<  Error: Call to malloc() failed */
+    MEMKIND_ERROR_ENVIRON = -12,                /**<  Error: Unable to parse environment variable */
+    MEMKIND_ERROR_INVALID = -13,                /**<  Error: Invalid argument */
+    MEMKIND_ERROR_TOOMANY = -15,                /**<  Error: Attempt to initialize more than MEMKIND_MAX_KIND number of kinds */
+    MEMKIND_ERROR_BADOPS = -17,                 /**<  Error: Invalid memkind_ops structure */
+    MEMKIND_ERROR_HUGETLB = -18,                /**<  Error: Unable to allocate huge pages */
+    MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE = -20,  /**<  Error: Requested memory type is not available */
+    MEMKIND_ERROR_OPERATION_FAILED = -21,       /**<  Error: Operation failed */
+    MEMKIND_ERROR_ARENAS_CREATE = -22,          /**<  Error: Call to jemalloc's arenas.create failed */
+    MEMKIND_ERROR_RUNTIME = -255                /**<  Error: Unspecified run-time error */
+};
+
+/* KIND CONFIGURATION MANAGEMENT INTERFACE */
+
+/// \brief Memkind memory usage policy
+typedef enum memkind_mem_usage_policy {
+    MEMKIND_MEM_USAGE_POLICY_DEFAULT      = 0,        /**<  Default memory usage */
+    MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE = 1,        /**<  Minimize memory usage at all costs, */
+    MEMKIND_MEM_USAGE_POLICY_MAX_VALUE
+} memkind_mem_usage_policy;
+
+/// \brief Memkind memory statistics type
+typedef enum memkind_stat_type {
+    MEMKIND_STAT_TYPE_RESIDENT      = 0,        /**<  Maximum number of bytes in physically resident data pages mapped */
+    MEMKIND_STAT_TYPE_ACTIVE        = 1,        /**<  Total number of bytes in active pages */
+    MEMKIND_STAT_TYPE_ALLOCATED     = 2,        /**<  Total number of allocated bytes */
+    MEMKIND_STAT_TYPE_MAX_VALUE
+} memkind_stat_type;
+
+/// \brief Forward declaration of memkind configuration
+struct memkind_config;
+
+///
+/// \brief Create a memkind configuration
+/// \note STANDARD API
+/// \return Memkind configuration, NULL on failure
+///
+struct memkind_config *memkind_config_new(void);
+
+///
+/// \brief Delete memkind configuration
+/// \note STANDARD API
+/// \param cfg memkind configuration
+///
+void memkind_config_delete(struct memkind_config *cfg);
+
+///
+/// \brief Update memkind configuration with path to specified directory parameter
+/// \note STANDARD API
+/// \param cfg memkind configuration
+/// \param pmem_dir path to specified directory for PMEM kind
+///
+void memkind_config_set_path(struct memkind_config *cfg, const char *pmem_dir);
+
+///
+/// \brief Update memkind configuration with PMEM kind size
+/// \note STANDARD API
+/// \param cfg memkind configuration
+/// \param pmem_size size limit for PMEM kind
+///
+void memkind_config_set_size(struct memkind_config *cfg, size_t pmem_size);
+
+///
+/// \brief Update memkind configuration with memory usage policy parameter
+/// \note STANDARD API
+/// \param cfg memkind configuration
+/// \param policy memkind memory usage policy
+///
+void memkind_config_set_memory_usage_policy(struct memkind_config *cfg,
+                                            memkind_mem_usage_policy policy);
+
+///
+/// \brief Create kind that allocates memory with specific memory type, memory binding policy and flags.
+/// \warning EXPERIMENTAL API
+/// \note Currently implemented memory type and policy configurations:
+///.      {MEMKIND_MEMTYPE_DEFAULT, MEMKIND_POLICY_PREFERRED_LOCAL},
+///.      {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_BIND_LOCAL},
+///       {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_PREFERRED_LOCAL},
+///       {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_INTERLEAVE_ALL},
+///       {MEMKIND_MEMTYPE_DEFAULT | MEMKIND_MEMTYPE_HIGH_BANDWIDTH, MEMKIND_POLICY_INTERLEAVE_ALL}.
+/// \param memtype_flags determine the memory types to allocate from by combination of memkind_memtype_t values.
+///        This field cannot have zero value.
+/// \param policy specify policy for page binding to memory types selected by  memtype_flags.
+///        This field must be set to memkind_policy_t value. If policy is set to MEMKIND_POLICY_PREFERRED_LOCAL then only one memory
+///        type must be selected. Note: the value cannot be set to MEMKIND_POLICY_MAX_VALUE.
+/// \param flags the field must be set to a combination of memkind_bits_t values.
+/// \param kind pointer to kind which will be created
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE or MEMKIND_ERROR_INVALID on failure
+///
+int memkind_create_kind(memkind_memtype_t memtype_flags,
+                        memkind_policy_t policy,
+                        memkind_bits_t flags,
+                        memkind_t *kind);
+
+///
+/// \brief Destroy previously created kind object, which must have been returned
+///        by a call to memkind_create_kind() or memkind_create_pmem().
+///        The function has undefined behavior when the handle is invalid or
+///        memkind_destroy_kind(kind) was already called before
+/// \warning EXPERIMENTAL API
+/// \note if the kind was returned by memkind_create_kind() all allocated memory must be freed
+///       before kind is destroyed, otherwise this will cause memory leak.
+/// \param kind specified memory kind
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, MEMKIND_ERROR_OPERATION_FAILED on failure
+///
+int memkind_destroy_kind(memkind_t kind);
+
+///
+/// \brief Get kind associated with allocated memory referenced by ptr
+/// \note STANDARD API
+/// \note This function has non-trivial performance overhead
+/// \param ptr pointer to the allocated memory
+/// \return Kind associated with allocated memory, NULL on failure
+///
+memkind_t memkind_detect_kind(void *ptr);
+
+#include "memkind_deprecated.h"
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_REGULAR;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_DEFAULT;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_ALL;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_PREFERRED;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_ALL_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_PREFERRED_HUGETLB;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_HBW_INTERLEAVE;
+
+/// \warning EXPERIMENTAL API
+extern memkind_t MEMKIND_INTERLEAVE;
+
+/// \note STANDARD API
+extern memkind_t MEMKIND_DAX_KMEM;
+
+/// \note STANDARD API
+extern memkind_t MEMKIND_DAX_KMEM_ALL;
+
+/// \note STANDARD API
+extern memkind_t MEMKIND_DAX_KMEM_PREFERRED;
+
+///
+/// \brief Get Memkind API version
+/// \note STANDARD API
+/// \return Version number represented by a single integer number(major * 1000000 + minor * 1000 + patch)
+///
+int memkind_get_version();
+
+///
+/// \brief Convert error number into an error message
+/// \note STANDARD API
+/// \param err error number
+/// \param msg error message
+/// \param size size of message
+///
+void memkind_error_message(int err, char *msg, size_t size);
+
+///
+/// \brief Create a new PMEM (file-backed) kind of given size on top of a temporary file
+///        in the given directory dir
+/// \note STANDARD API
+/// \param dir path to specified directory to temporary file
+/// \param max_size size limit for kind
+/// \param kind pointer to kind which will be created
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_create_pmem(const char *dir, size_t max_size, memkind_t *kind);
+
+///
+/// \brief Create a new PMEM kind with given memkind configuration
+/// \note STANDARD API
+/// \param cfg memkind configuration for specifying PMEM parameters
+/// \param kind pointer to kind which will be created
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_create_pmem_with_config(struct memkind_config *cfg,
+                                    memkind_t *kind);
+
+///
+/// \brief Check if kind is available
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_check_available(memkind_t kind);
+
+///
+/// \brief Update memkind cached statistics
+/// \note STANDARD API
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_update_cached_stats(void);
+
+///
+/// \brief Get memkind statistic
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param stat specified type of memory statistic
+/// \param value reference to value of memory statistic
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, other values on failure
+///
+int memkind_get_stat(memkind_t kind, memkind_stat_type stat, size_t *value);
+
+/* HEAP MANAGEMENT INTERFACE */
+
+///
+/// \brief Allocates size bytes of uninitialized storage of the specified kind
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param size number of bytes to allocate
+/// \return Pointer to the allocated memory
+///
+void *memkind_malloc(memkind_t kind, size_t size);
+
+///
+/// \brief Obtain size of block of memory allocated with the memkind API
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the allocated memory
+/// \return Number of usable bytes
+///
+size_t memkind_malloc_usable_size(memkind_t kind, void *ptr);
+
+///
+/// \brief Allocates memory of the specified kind for an array of num elements
+///        of size bytes each and initializes all bytes in the allocated storage to zero
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param num number of objects
+/// \param size specified size of each element
+/// \return Pointer to the allocated memory
+///
+void *memkind_calloc(memkind_t kind, size_t num, size_t size);
+
+///
+/// \brief Allocates size bytes of the specified kind and places the address of the allocated memory
+///        in *memptr. The address of the allocated memory will be a multiple of alignment,
+///        which must be a power of two and a multiple of sizeof(void *)
+/// \note EXPERIMENTAL API
+/// \param kind specified memory kind
+/// \param memptr address of the allocated memory
+/// \param alignment specified alignment of bytes
+/// \param size specified size of bytes
+/// \return Memkind operation status, MEMKIND_SUCCESS on success, EINVAL or ENOMEM on failure
+///
+int memkind_posix_memalign(memkind_t kind, void **memptr, size_t alignment,
+                           size_t size);
+
+///
+/// \brief Reallocates memory of the specified kind
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the memory block to be reallocated
+/// \param size new size for the memory block in bytes
+/// \return Pointer to the allocated memory
+///
+void *memkind_realloc(memkind_t kind, void *ptr, size_t size);
+
+///
+/// \brief Free the memory space of the specified kind pointed by ptr
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the allocated memory
+///
+void memkind_free(memkind_t kind, void *ptr);
+
+///
+/// \brief Try to reallocate allocation to reduce fragmentation
+/// \note STANDARD API
+/// \param kind specified memory kind
+/// \param ptr pointer to the allocated memory
+/// \return Pointer to newly transferred allocated memory
+///
+void *memkind_defrag_reallocate(memkind_t kind, void *ptr);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/heap_manager.h b/memkind-1.10.0/include/memkind/internal/heap_manager.h
new file mode 100644
index 0000000..2686f1f
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/heap_manager.h
@@ -0,0 +1,36 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <memkind.h>
+
+void heap_manager_init(struct memkind *kind);
+void heap_manager_free(void *ptr);
+size_t heap_manager_malloc_usable_size(void *ptr);
+void *heap_manager_realloc(void *ptr, size_t size);
+struct memkind *heap_manager_detect_kind(void *ptr);
+int heap_manager_update_cached_stats(void);
+int heap_manager_get_stat(memkind_stat_type stat, size_t *value);
+void *heap_manager_defrag_reallocate(void *ptr);
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_arena.h b/memkind-1.10.0/include/memkind/internal/memkind_arena.h
new file mode 100644
index 0000000..210c3b2
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_arena.h
@@ -0,0 +1,77 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+#include <jemalloc/jemalloc.h>
+#include <memkind/internal/memkind_private.h>
+
+/*
+ * Header file for the jemalloc arena allocation memkind operations.
+ * More details in memkind_arena(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+struct memkind *get_kind_by_arena(unsigned arena_ind);
+struct memkind *memkind_arena_detect_kind(void *ptr);
+int memkind_arena_create(struct memkind *kind, struct memkind_ops *ops,
+                         const char *name);
+int memkind_arena_create_map(struct memkind *kind, extent_hooks_t *hooks);
+int memkind_arena_destroy(struct memkind *kind);
+void *memkind_arena_malloc(struct memkind *kind, size_t size);
+void *memkind_arena_calloc(struct memkind *kind, size_t num, size_t size);
+int memkind_arena_posix_memalign(struct memkind *kind, void **memptr,
+                                 size_t alignment, size_t size);
+void *memkind_arena_realloc(struct memkind *kind, void *ptr, size_t size);
+void *memkind_arena_realloc_with_kind_detect(void *ptr, size_t size);
+int memkind_bijective_get_arena(struct memkind *kind, unsigned int *arena,
+                                size_t size);
+int memkind_thread_get_arena(struct memkind *kind, unsigned int *arena,
+                             size_t size);
+int memkind_arena_finalize(struct memkind *kind);
+void memkind_arena_init(struct memkind *kind);
+void memkind_arena_free(struct memkind *kind, void *ptr);
+void memkind_arena_free_with_kind_detect(void *ptr);
+size_t memkind_arena_malloc_usable_size(void *ptr);
+int memkind_arena_update_memory_usage_policy(struct memkind *kind,
+                                             memkind_mem_usage_policy policy);
+int memkind_arena_enable_background_threads(size_t threads_limit);
+int memkind_arena_update_cached_stats(void);
+int memkind_arena_get_kind_stat(struct memkind *kind,
+                                memkind_stat_type stat_type,
+                                size_t *stat);
+int memkind_arena_get_stat_with_check_init(struct memkind *kind,
+                                           memkind_stat_type stat, bool check_init, size_t *value);
+int memkind_arena_get_global_stat(memkind_stat_type stat_type, size_t *stat);
+void *memkind_arena_defrag_reallocate(struct memkind *kind, void *ptr);
+void *memkind_arena_defrag_reallocate_with_kind_detect(void *ptr);
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_bandwidth.h b/memkind-1.10.0/include/memkind/internal/memkind_bandwidth.h
new file mode 100644
index 0000000..e741a0b
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_bandwidth.h
@@ -0,0 +1,51 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind/internal/vec.h>
+#include <numa.h>
+#include <stdbool.h>
+
+struct vec_cpu_node;
+
+typedef int (*get_node_bitmask)(struct bitmask *);
+typedef int (*fill_bandwidth_values)(int *);
+
+int bandwidth_fill(int *bandwidth, get_node_bitmask get_bitmask);
+int set_closest_numanode(fill_bandwidth_values fill, const char *env,
+                         struct vec_cpu_node **closest_numanode, int num_cpu, bool is_single_node);
+void set_bitmask_for_all_closest_numanodes(unsigned long *nodemask,
+                                           unsigned long maxnode, const struct vec_cpu_node *closest_numanode,
+                                           int num_cpu);
+int set_bitmask_for_current_closest_numanode(unsigned long *nodemask,
+                                             unsigned long maxnode, const struct vec_cpu_node *closest_numanode,
+                                             int num_cpu);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_dax_kmem.h b/memkind-1.10.0/include/memkind/internal/memkind_dax_kmem.h
new file mode 100644
index 0000000..2bbc193
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_dax_kmem.h
@@ -0,0 +1,49 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the DAX KMEM memory memkind operations.
+ * More details in memkind_dax_kmem(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_dax_kmem_all_get_mbind_nodemask(struct memkind *kind,
+                                            unsigned long *nodemask, unsigned long maxnode);
+
+extern struct memkind_ops MEMKIND_DAX_KMEM_OPS;
+extern struct memkind_ops MEMKIND_DAX_KMEM_ALL_OPS;
+extern struct memkind_ops MEMKIND_DAX_KMEM_PREFERRED_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_default.h b/memkind-1.10.0/include/memkind/internal/memkind_default.h
new file mode 100644
index 0000000..51e24d5
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_default.h
@@ -0,0 +1,71 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+#include <memkind/internal/memkind_private.h>
+
+/*
+ * Header file for the default implementations for memkind operations.
+ * More details in memkind_default(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_default_create(struct memkind *kind, struct memkind_ops *ops,
+                           const char *name);
+int memkind_default_destroy(struct memkind *kind);
+void *memkind_default_malloc(struct memkind *kind, size_t size);
+void *memkind_default_calloc(struct memkind *kind, size_t num, size_t size);
+int memkind_default_posix_memalign(struct memkind *kind, void **memptr,
+                                   size_t alignment, size_t size);
+void *memkind_default_realloc(struct memkind *kind, void *ptr, size_t size);
+void memkind_default_free(struct memkind *kind, void *ptr);
+void *memkind_default_mmap(struct memkind *kind, void *addr, size_t size);
+int memkind_default_mbind(struct memkind *kind, void *ptr, size_t size);
+int memkind_default_get_mmap_flags(struct memkind *kind, int *flags);
+int memkind_default_get_mbind_mode(struct memkind *kind, int *mode);
+int memkind_default_get_mbind_nodemask(struct memkind *kind,
+                                       unsigned long *nodemask, unsigned long maxnode);
+int memkind_preferred_get_mbind_mode(struct memkind *kind, int *mode);
+int memkind_interleave_get_mbind_mode(struct memkind *kind, int *mode);
+int memkind_nohugepage_madvise(struct memkind *kind, void *addr, size_t size);
+int memkind_posix_check_alignment(struct memkind *kind, size_t alignment);
+void memkind_default_init_once(void);
+size_t memkind_default_malloc_usable_size(struct memkind *kind, void *ptr);
+static inline bool size_out_of_bounds(size_t size)
+{
+    return !size;
+}
+
+extern struct memkind_ops MEMKIND_DEFAULT_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_gbtlb.h b/memkind-1.10.0/include/memkind/internal/memkind_gbtlb.h
new file mode 100755
index 0000000..97f0796
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_gbtlb.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the gigabyte TLB memkind operations.
+ *
+ * All function declarations has been moved to memkind_deprecated.h
+ * because of end of GBTLB support.
+ *
+ * API standards are described in memkind(3) man page.
+ */
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_hbw.h b/memkind-1.10.0/include/memkind/internal/memkind_hbw.h
new file mode 100755
index 0000000..7eb039d
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_hbw.h
@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the high bandwidth memory memkind operations.
+ * More details in memkind_hbw(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_hbw_check_available(struct memkind *kind);
+int memkind_hbw_hugetlb_check_available(struct memkind *kind);
+int memkind_hbw_get_mbind_nodemask(struct memkind *kind,
+                                   unsigned long *nodemask,
+                                   unsigned long maxnode);
+int memkind_hbw_all_get_mbind_nodemask(struct memkind *kind,
+                                       unsigned long *nodemask,
+                                       unsigned long maxnode);
+void memkind_hbw_init_once(void);
+void memkind_hbw_all_init_once(void);
+void memkind_hbw_hugetlb_init_once(void);
+void memkind_hbw_all_hugetlb_init_once(void);
+void memkind_hbw_preferred_init_once(void);
+void memkind_hbw_preferred_hugetlb_init_once(void);
+void memkind_hbw_interleave_init_once(void);
+
+extern struct memkind_ops MEMKIND_HBW_OPS;
+extern struct memkind_ops MEMKIND_HBW_ALL_OPS;
+extern struct memkind_ops MEMKIND_HBW_HUGETLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_ALL_HUGETLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_PREFERRED_OPS;
+extern struct memkind_ops MEMKIND_HBW_PREFERRED_HUGETLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_INTERLEAVE_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_hugetlb.h b/memkind-1.10.0/include/memkind/internal/memkind_hugetlb.h
new file mode 100755
index 0000000..d76915f
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_hugetlb.h
@@ -0,0 +1,48 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the hugetlb memory memkind operations.
+ * More details in memkind_hugetlb(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+int memkind_hugetlb_get_mmap_flags(struct memkind *kind, int *flags);
+void memkind_hugetlb_init_once(void);
+int memkind_hugetlb_check_available_2mb(struct memkind *kind);
+
+extern struct memkind_ops MEMKIND_HUGETLB_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_interleave.h b/memkind-1.10.0/include/memkind/internal/memkind_interleave.h
new file mode 100755
index 0000000..1913d6a
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_interleave.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/*
+ * Header file for the interleave memory memkind operations.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+void memkind_interleave_init_once(void);
+
+extern struct memkind_ops MEMKIND_INTERLEAVE_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_log.h b/memkind-1.10.0/include/memkind/internal/memkind_log.h
new file mode 100644
index 0000000..7e9b7e0
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_log.h
@@ -0,0 +1,52 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define PRINTF_FORMAT __attribute__ ((format (printf, 1, 2)))
+
+/*
+ * For printing informational messages
+ * Requires environment variable MEMKIND_DEBUG to be set to appropriate value
+ */
+void log_info(const char *format, ...) PRINTF_FORMAT;
+
+/*
+ * For printing messages regarding errors and failures
+ * Requires environment variable MEMKIND_DEBUG to be set to appropriate value
+ */
+void log_err(const char *format, ...) PRINTF_FORMAT;
+
+/*
+ * For printing messages regarding fatal errors before calling abort()
+ * Works *no matter* of MEMKIND_DEBUG state
+ */
+void log_fatal(const char *format, ...)PRINTF_FORMAT;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_pmem.h b/memkind-1.10.0/include/memkind/internal/memkind_pmem.h
new file mode 100644
index 0000000..b6098ba
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_pmem.h
@@ -0,0 +1,64 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+#include "memkind_default.h"
+#include "memkind_arena.h"
+
+#include <pthread.h>
+
+/*
+ * Header file for the file-backed memory memkind operations.
+ * More details in memkind_pmem(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+#define MEMKIND_PMEM_CHUNK_SIZE (1ull << 21ull) // 2MB
+
+int memkind_pmem_create(struct memkind *kind, struct memkind_ops *ops,
+                        const char *name);
+int memkind_pmem_destroy(struct memkind *kind);
+void *memkind_pmem_mmap(struct memkind *kind, void *addr, size_t size);
+int memkind_pmem_get_mmap_flags(struct memkind *kind, int *flags);
+
+struct memkind_pmem {
+    int fd;
+    off_t offset;
+    size_t max_size;
+    pthread_mutex_t pmem_lock;
+    size_t current_size;
+};
+
+extern struct memkind_ops MEMKIND_PMEM_OPS;
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_private.h b/memkind-1.10.0/include/memkind/internal/memkind_private.h
new file mode 100644
index 0000000..af01ba5
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_private.h
@@ -0,0 +1,125 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "memkind.h"
+
+#include <stdbool.h>
+#include <pthread.h>
+
+#ifdef __GNUC__
+#   define MEMKIND_LIKELY(x)       __builtin_expect(!!(x), 1)
+#   define MEMKIND_UNLIKELY(x)     __builtin_expect(!!(x), 0)
+#else
+#   define MEMKIND_LIKELY(x)       (x)
+#   define MEMKIND_UNLIKELY(x)     (x)
+#endif
+
+#ifndef MEMKIND_EXPORT
+#   define MEMKIND_EXPORT __attribute__((visibility("default")))
+#endif
+
+// This ladder call is required due to meanders of C's preprocessor logic.
+// Without it, JE_PREFIX would be used directly (i.e. 'JE_PREFIX') and not
+// substituted with defined value.
+#define JE_SYMBOL2(a, b) a ## b
+#define JE_SYMBOL1(a, b) JE_SYMBOL2(a, b)
+#define JE_SYMBOL(b)     JE_SYMBOL1(JE_PREFIX, b)
+
+// Redefine symbols
+#define jemk_malloc                 JE_SYMBOL(malloc)
+#define jemk_mallocx                JE_SYMBOL(mallocx)
+#define jemk_calloc                 JE_SYMBOL(calloc)
+#define jemk_rallocx                JE_SYMBOL(rallocx)
+#define jemk_realloc                JE_SYMBOL(realloc)
+#define jemk_mallctl                JE_SYMBOL(mallctl)
+#define jemk_memalign               JE_SYMBOL(memalign)
+#define jemk_posix_memalign         JE_SYMBOL(posix_memalign)
+#define jemk_free                   JE_SYMBOL(free)
+#define jemk_dallocx                JE_SYMBOL(dallocx)
+#define jemk_malloc_usable_size     JE_SYMBOL(malloc_usable_size)
+
+enum memkind_const_private {
+    MEMKIND_NAME_LENGTH_PRIV = 64
+};
+
+struct memkind_ops {
+    int (* create)(struct memkind *kind, struct memkind_ops *ops, const char *name);
+    int (* destroy)(struct memkind *kind);
+    void *(* malloc)(struct memkind *kind, size_t size);
+    void *(* calloc)(struct memkind *kind, size_t num, size_t size);
+    int (* posix_memalign)(struct memkind *kind, void **memptr, size_t alignment,
+                           size_t size);
+    void *(* realloc)(struct memkind *kind, void *ptr, size_t size);
+    void (* free)(struct memkind *kind, void *ptr);
+    void *(* mmap)(struct memkind *kind, void *addr, size_t size);
+    int (* mbind)(struct memkind *kind, void *ptr, size_t size);
+    int (* madvise)(struct memkind *kind, void *addr, size_t size);
+    int (* get_mmap_flags)(struct memkind *kind, int *flags);
+    int (* get_mbind_mode)(struct memkind *kind, int *mode);
+    int (* get_mbind_nodemask)(struct memkind *kind, unsigned long *nodemask,
+                               unsigned long maxnode);
+    int (* get_arena)(struct memkind *kind, unsigned int *arena, size_t size);
+    int (* check_available)(struct memkind *kind);
+    void (* init_once)(void);
+    int (* finalize)(struct memkind *kind);
+    size_t (* malloc_usable_size)(struct memkind *kind, void *ptr);
+    int (* update_memory_usage_policy)(struct memkind *kind,
+                                       memkind_mem_usage_policy policy);
+    int (* get_stat)(memkind_t kind, memkind_stat_type stat, size_t *value);
+    void *(* defrag_reallocate)(struct memkind *kind, void *ptr);
+};
+
+struct memkind {
+    struct memkind_ops *ops;
+    unsigned int partition;
+    char name[MEMKIND_NAME_LENGTH_PRIV];
+    pthread_once_t init_once;
+    unsigned int arena_map_len; // is power of 2
+    unsigned int *arena_map; // To be deleted beyond 1.2.0+
+    pthread_key_t arena_key;
+    void *priv;
+    unsigned int
+    arena_map_mask; // arena_map_len - 1 to optimize modulo operation on arena_map_len
+    unsigned int arena_zero; // index first jemalloc arena of this kind
+};
+
+struct memkind_config {
+    const char *pmem_dir;            //PMEM kind path
+    size_t pmem_size;                //PMEM kind size
+    memkind_mem_usage_policy policy; //kind memory usage policy
+};
+
+void memkind_init(memkind_t kind, bool check_numa);
+
+void *kind_mmap(struct memkind *kind, void *addr, size_t size);
+
+#ifdef __cplusplus
+}
+#endif
+
diff --git a/memkind-1.10.0/include/memkind/internal/memkind_regular.h b/memkind-1.10.0/include/memkind/internal/memkind_regular.h
new file mode 100755
index 0000000..067ea04
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/memkind_regular.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <memkind.h>
+
+/*
+ * Header file for the regular memory memkind operations.
+ * More details in memkind_regular(3) man page.
+ *
+ * Functionality defined in this header is considered as EXPERIMENTAL API.
+ * API standards are described in memkind(3) man page.
+ */
+
+extern struct memkind_ops MEMKIND_REGULAR_OPS;
+int memkind_regular_all_get_mbind_nodemask(struct memkind *kind,
+                                           unsigned long *nodemask,
+                                           unsigned long maxnode);
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/tbb_mem_pool_policy.h b/memkind-1.10.0/include/memkind/internal/tbb_mem_pool_policy.h
new file mode 100644
index 0000000..085cb29
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/tbb_mem_pool_policy.h
@@ -0,0 +1,34 @@
+/*
+    Copyright (c) 2005-2018 Intel Corporation
+
+    Licensed under the Apache License, Version 2.0 (the "License");
+    you may not use this file except in compliance with the License.
+    You may obtain a copy of the License at
+
+        http://www.apache.org/licenses/LICENSE-2.0
+
+    Unless required by applicable law or agreed to in writing, software
+    distributed under the License is distributed on an "AS IS" BASIS,
+    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+    See the License for the specific language governing permissions and
+    limitations under the License.
+
+
+
+
+*/
+
+#include <stdint.h>
+
+typedef void *(*rawAllocType)(intptr_t pool_id, size_t *bytes);
+typedef int   (*rawFreeType)(intptr_t pool_id, void *raw_ptr, size_t raw_bytes);
+
+struct MemPoolPolicy {
+    rawAllocType pAlloc;
+    rawFreeType  pFree;
+    size_t       granularity;
+    int          version;
+    unsigned     fixedPool : 1,
+                 keepAllMemory : 1,
+                 reserved : 30;
+};
diff --git a/memkind-1.10.0/include/memkind/internal/tbb_wrapper.h b/memkind-1.10.0/include/memkind/internal/tbb_wrapper.h
new file mode 100644
index 0000000..11615d8
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/tbb_wrapper.h
@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include <memkind.h>
+#include <memkind_deprecated.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/* dynamically load TBB symbols */
+void load_tbb_symbols(void);
+
+/* ops callbacks are replaced by TBB callbacks. */
+void tbb_initialize(struct memkind *kind);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+void tbb_pool_free_with_kind_detect(void *ptr);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+void *tbb_pool_realloc_with_kind_detect(void *ptr, size_t size);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+size_t tbb_pool_malloc_usable_size_with_kind_detect(void *ptr);
+
+/* ptr pointer must come from the valid TBB pool allocation */
+struct memkind *tbb_detect_kind(void *ptr);
+
+/* update cached stats for TBB (unsupported) */
+int tbb_update_cached_stats(void);
+
+/* get allocator stat for TBB (unsupported) */
+int tbb_get_global_stat(memkind_stat_type stat, size_t *value);
+
+/* defrag reallocate for TBB (unsupported) */
+void *tbb_pool_defrag_reallocate_with_kind_detect(void *ptr);
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind/internal/vec.h b/memkind-1.10.0/include/memkind/internal/vec.h
new file mode 100644
index 0000000..982618f
--- /dev/null
+++ b/memkind-1.10.0/include/memkind/internal/vec.h
@@ -0,0 +1,102 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <stddef.h>
+#include <stdlib.h>
+
+#define VEC_INIT_SIZE (8)
+
+#define VEC(name, type)\
+struct name {\
+    type *buffer;\
+    size_t size;\
+    size_t capacity;\
+}
+
+static inline int
+vec_reserve(void *vec, size_t ncapacity, size_t s)
+{
+    size_t ncap = ncapacity == 0 ? VEC_INIT_SIZE : ncapacity;
+    VEC(vvec, void) *vecp = (struct vvec *)vec;
+    void *tbuf = realloc(vecp->buffer, s * ncap);
+    if (tbuf == NULL) {
+        return -1;
+    }
+    vecp->buffer = tbuf;
+    vecp->capacity = ncap;
+    return 0;
+}
+
+#define VEC_RESERVE(vec, ncapacity)\
+(((vec)->size == 0 || (ncapacity) > (vec)->size) ?\
+    vec_reserve((void *)vec, ncapacity, sizeof(*(vec)->buffer)) :\
+    0)
+
+#define VEC_INSERT(vec, element)\
+((vec)->buffer[(vec)->size - 1] = (element), 0)
+
+#define VEC_INC_SIZE(vec)\
+(((vec)->size++), 0)
+
+#define VEC_INC_BACK(vec)\
+((vec)->capacity == (vec)->size ?\
+    (VEC_RESERVE((vec), ((vec)->capacity * 2)) == 0 ?\
+        VEC_INC_SIZE(vec) : -1) :\
+    VEC_INC_SIZE(vec))
+
+#define VEC_PUSH_BACK(vec, element)\
+(VEC_INC_BACK(vec) == 0? VEC_INSERT(vec, element) : -1)
+
+#define VEC_FOREACH(el, vec)\
+size_t _vec_i; \
+for (_vec_i = 0;\
+    _vec_i < (vec)->size && (((el) = (vec)->buffer[_vec_i]), 1);\
+    ++_vec_i)
+
+#define VEC_SIZE(vec)\
+((vec)->size)
+
+#define VEC_GET(vec, id)\
+(&(vec)->buffer[id])
+
+#define VEC_CLEAR(vec) do {\
+    (vec)->size = 0;\
+} while (0)
+
+#define VEC_DELETE(vec) do {\
+    free((vec)->buffer);\
+    (vec)->buffer = NULL;\
+    (vec)->size = 0;\
+    (vec)->capacity = 0;\
+} while (0)
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/memkind_allocator.h b/memkind-1.10.0/include/memkind_allocator.h
new file mode 100644
index 0000000..45bde75
--- /dev/null
+++ b/memkind-1.10.0/include/memkind_allocator.h
@@ -0,0 +1,231 @@
+/*
+ * Copyright (C) 2019 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <cmath>
+#include <map>
+#include <memory>
+#include <string>
+#include <exception>
+#include <type_traits>
+#include <cstddef>
+#include <stdexcept>
+
+#include "memkind.h"
+
+/*
+ * Header file for the C++ allocator compatible with the C++ standard library allocator concepts.
+ * More details in memkind(3) man page.
+ * Note: memory heap management is based on memkind_malloc, refer to the memkind(3) man page for more
+ * information.
+ *
+ * Functionality defined in this header is considered as stable API (STANDARD API).
+ * API standards are described in memkind(3) man page.
+ */
+namespace libmemkind
+{
+    enum class kinds {
+        DEFAULT = 0,
+        HUGETLB = 1,
+        INTERLEAVE = 2,
+        HBW = 3,
+        HBW_ALL = 4,
+        HBW_HUGETLB = 5,
+        HBW_ALL_HUGETLB = 6,
+        HBW_PREFERRED = 7,
+        HBW_PREFERRED_HUGETLB = 8,
+        HBW_INTERLEAVE = 9,
+        REGULAR = 10,
+        DAX_KMEM = 11,
+        DAX_KMEM_ALL = 12,
+        DAX_KMEM_PREFERRED = 13,
+    };
+
+    namespace static_kind
+    {
+        template<typename T>
+        class allocator
+        {
+        public:
+            using value_type = T;
+            using pointer = value_type*;
+            using const_pointer = const value_type*;
+            using reference = value_type&;
+            using const_reference = const value_type&;
+            using size_type = size_t;
+            using difference_type = ptrdiff_t;
+
+            template<class U>
+            struct rebind {
+                using other = allocator<U>;
+            };
+
+            template<typename U>
+            friend class allocator;
+
+#if !_GLIBCXX_USE_CXX11_ABI
+            /* This is a workaround for compilers (e.g GCC 4.8) that uses C++11 standard,
+             * but use old - non C++11 ABI */
+            template<typename V = void>
+            explicit allocator()
+            {
+                static_assert(std::is_same<V, void>::value,
+                              "libmemkind::static_kind::allocator cannot be compiled without CXX11 ABI");
+            }
+#endif
+
+            explicit allocator(libmemkind::kinds kind)
+            {
+                switch (kind) {
+                    case libmemkind::kinds::DEFAULT:
+                        _kind = MEMKIND_DEFAULT;
+                        break;
+                    case libmemkind::kinds::HUGETLB:
+                        _kind = MEMKIND_HUGETLB;
+                        break;
+                    case libmemkind::kinds::INTERLEAVE:
+                        _kind = MEMKIND_INTERLEAVE;
+                        break;
+                    case libmemkind::kinds::HBW:
+                        _kind = MEMKIND_HBW;
+                        break;
+                    case libmemkind::kinds::HBW_ALL:
+                        _kind = MEMKIND_HBW_ALL;
+                        break;
+                    case libmemkind::kinds::HBW_HUGETLB:
+                        _kind = MEMKIND_HBW_HUGETLB;
+                        break;
+                    case libmemkind::kinds::HBW_ALL_HUGETLB:
+                        _kind = MEMKIND_HBW_ALL_HUGETLB;
+                        break;
+                    case libmemkind::kinds::HBW_PREFERRED:
+                        _kind = MEMKIND_HBW_PREFERRED;
+                        break;
+                    case libmemkind::kinds::HBW_PREFERRED_HUGETLB:
+                        _kind = MEMKIND_HBW_PREFERRED_HUGETLB;
+                        break;
+                    case libmemkind::kinds::HBW_INTERLEAVE:
+                        _kind = MEMKIND_HBW_INTERLEAVE;
+                        break;
+                    case libmemkind::kinds::REGULAR:
+                        _kind = MEMKIND_REGULAR;
+                        break;
+                    case libmemkind::kinds::DAX_KMEM:
+                        _kind = MEMKIND_DAX_KMEM;
+                        break;
+                    case libmemkind::kinds::DAX_KMEM_ALL:
+                        _kind = MEMKIND_DAX_KMEM_ALL;
+                        break;
+                    case libmemkind::kinds::DAX_KMEM_PREFERRED:
+                        _kind = MEMKIND_DAX_KMEM_PREFERRED;
+                        break;
+                    default:
+                        throw std::runtime_error("Unknown libmemkind::kinds");
+                        break;
+                }
+            }
+
+            allocator(const allocator &other) = default;
+
+            template <typename U>
+            allocator(const allocator<U> &other) noexcept
+            {
+                _kind = other._kind;
+            }
+
+            allocator(allocator &&other) = default;
+
+            template <typename U>
+            allocator(const allocator<U> &&other) noexcept
+            {
+                _kind = std::move(other._kind);
+            }
+
+            allocator<T> &operator = (const allocator &other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (const allocator<U> &other) noexcept
+            {
+                _kind = other._kind;
+                return *this;
+            }
+
+            allocator<T> &operator = (allocator &&other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (allocator<U> &&other) noexcept
+            {
+                _kind = std::move(other._kind);
+                return *this;
+            }
+
+            pointer allocate(size_type n) const
+            {
+                pointer result = static_cast<pointer>(memkind_malloc(_kind, n*sizeof(T)));
+                if (!result) {
+                    throw std::bad_alloc();
+                }
+                return result;
+            }
+
+            void deallocate(pointer p, size_type n) const
+            {
+                memkind_free(_kind, static_cast<void *>(p));
+            }
+
+            template <class U, class... Args>
+            void construct(U *p, Args &&... args) const
+            {
+                ::new((void *)p) U(std::forward<Args>(args)...);
+            }
+
+            void destroy(pointer p) const
+            {
+                p->~value_type();
+            }
+
+            template <typename U, typename V>
+            friend bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs);
+
+            template <typename U, typename V>
+            friend bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs);
+
+        private:
+            memkind_t _kind;
+        };
+
+        template <typename U, typename V>
+        bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return lhs._kind == rhs._kind;
+        }
+
+        template <typename U, typename V>
+        bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return !(lhs._kind == rhs._kind);
+        }
+    }  // namespace static_kind
+}  // namespace libmemkind
diff --git a/memkind-1.10.0/include/memkind_deprecated.h b/memkind-1.10.0/include/memkind_deprecated.h
new file mode 100644
index 0000000..3e2c3c5
--- /dev/null
+++ b/memkind-1.10.0/include/memkind_deprecated.h
@@ -0,0 +1,88 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+/*
+ * !!!!!!!!!!!!!!!!!!!
+ * !!!   WARNING   !!!
+ * !!! PLEASE READ !!!
+ * !!!!!!!!!!!!!!!!!!!
+ *
+ * This header file contains all memkind deprecated symbols.
+ *
+ * Please avoid usage of this API in newly developed code, as
+ * eventually this code is subject for removal.
+ */
+
+#pragma once
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#ifndef MEMKIND_DEPRECATED
+
+#ifdef __GNUC__
+#define MEMKIND_DEPRECATED(func) func __attribute__ ((deprecated))
+#elif defined(_MSC_VER)
+#define MEMKIND_DEPRECATED(func) __declspec(deprecated) func
+#else
+#pragma message("WARNING: You need to implement MEMKIND_DEPRECATED for this compiler")
+#define MEMKIND_DEPRECATED(func) func
+#endif
+
+#endif
+
+/*
+ * Symbols related to GBTLB that are no longer supported
+ */
+extern memkind_t MEMKIND_HBW_GBTLB;
+extern memkind_t MEMKIND_HBW_PREFERRED_GBTLB;
+extern memkind_t MEMKIND_GBTLB;
+
+int MEMKIND_DEPRECATED(memkind_get_kind_by_partition(int partition,
+                                                     memkind_t *kind));
+
+enum memkind_base_partition {
+    MEMKIND_PARTITION_DEFAULT = 0,
+    MEMKIND_PARTITION_HBW = 1,
+    MEMKIND_PARTITION_HBW_HUGETLB = 2,
+    MEMKIND_PARTITION_HBW_PREFERRED = 3,
+    MEMKIND_PARTITION_HBW_PREFERRED_HUGETLB = 4,
+    MEMKIND_PARTITION_HUGETLB = 5,
+    MEMKIND_PARTITION_HBW_GBTLB = 6,
+    MEMKIND_PARTITION_HBW_PREFERRED_GBTLB = 7,
+    MEMKIND_PARTITION_GBTLB = 8,
+    MEMKIND_PARTITION_HBW_INTERLEAVE = 9,
+    MEMKIND_PARTITION_INTERLEAVE = 10,
+    MEMKIND_PARTITION_REGULAR = 11,
+    MEMKIND_PARTITION_HBW_ALL = 12,
+    MEMKIND_PARTITION_HBW_ALL_HUGETLB = 13,
+    MEMKIND_PARTITION_DAX_KMEM = 14,
+    MEMKIND_PARTITION_DAX_KMEM_ALL = 15,
+    MEMKIND_PARTITION_DAX_KMEM_PREFERRED = 16,
+    MEMKIND_NUM_BASE_KIND
+};
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/include/pmem_allocator.h b/memkind-1.10.0/include/pmem_allocator.h
new file mode 100644
index 0000000..d373e3c
--- /dev/null
+++ b/memkind-1.10.0/include/pmem_allocator.h
@@ -0,0 +1,236 @@
+/*
+ * Copyright (C) 2018 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <memory>
+#include <string>
+#include <exception>
+#include <type_traits>
+#include <atomic>
+#include <cstddef>
+#include <stdexcept>
+
+#include "memkind.h"
+
+/*
+ * Header file for the C++ allocator compatible with the C++ standard library allocator concepts.
+ * More details in pmemallocator(3) man page.
+ * Note: memory heap management is based on memkind_malloc, refer to the memkind(3) man page for more
+ * information.
+ *
+ * Functionality defined in this header is considered as stable API (STANDARD API).
+ * API standards are described in memkind(3) man page.
+ */
+namespace libmemkind
+{
+    enum class allocation_policy {
+        DEFAULT = MEMKIND_MEM_USAGE_POLICY_DEFAULT,
+        CONSERVATIVE = MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE,
+        MAX
+    };
+
+    namespace pmem
+    {
+        namespace internal
+        {
+            class kind_wrapper_t
+            {
+            public:
+                kind_wrapper_t(const char *dir, std::size_t max_size,
+                               libmemkind::allocation_policy alloc_policy=
+                                   libmemkind::allocation_policy::DEFAULT)
+                {
+                    cfg = memkind_config_new();
+                    if (!cfg) {
+                        throw std::runtime_error(
+                            std::string("An error occurred while creating pmem config"));
+                    }
+                    memkind_config_set_path(cfg, dir);
+                    memkind_config_set_size(cfg, max_size);
+                    memkind_config_set_memory_usage_policy(cfg,
+                                                           static_cast<memkind_mem_usage_policy>(alloc_policy));
+                    int err_c  = memkind_create_pmem_with_config(cfg, &kind);
+                    memkind_config_delete(cfg);
+                    if (err_c) {
+                        throw std::runtime_error(
+                            std::string("An error occurred while creating pmem kind; error code: ") +
+                            std::to_string(err_c));
+                    }
+                }
+
+                kind_wrapper_t(const kind_wrapper_t &) = delete;
+                void operator=(const kind_wrapper_t &) = delete;
+
+                ~kind_wrapper_t()
+                {
+                    memkind_destroy_kind(kind);
+                }
+
+                memkind_t get() const
+                {
+                    return kind;
+                }
+
+            private:
+                memkind_t kind;
+                struct memkind_config *cfg;
+            };
+        }
+
+        template<typename T>
+        class allocator
+        {
+            using kind_wrapper_t = internal::kind_wrapper_t;
+            std::shared_ptr<kind_wrapper_t> kind_wrapper_ptr;
+        public:
+            using value_type = T;
+            using pointer = value_type*;
+            using const_pointer = const value_type*;
+            using reference = value_type&;
+            using const_reference = const value_type&;
+            using size_type = size_t;
+            using difference_type = ptrdiff_t;
+
+            template<class U>
+            struct rebind {
+                using other = allocator<U>;
+            };
+
+            template<typename U>
+            friend class allocator;
+
+#if !_GLIBCXX_USE_CXX11_ABI
+            /* This is a workaround for compilers (e.g GCC 4.8) that uses C++11 standard,
+             * but use old - non C++11 ABI */
+            template<typename V = void>
+            explicit allocator()
+            {
+                static_assert(std::is_same<V, void>::value,
+                              "libmemkind::pmem::allocator cannot be compiled without CXX11 ABI");
+            }
+#endif
+
+            explicit allocator(const char *dir, size_t max_size) :
+                kind_wrapper_ptr(std::make_shared<kind_wrapper_t>(dir, max_size))
+            {
+            }
+
+            explicit allocator(const std::string &dir, size_t max_size) :
+                allocator(dir.c_str(), max_size)
+            {
+            }
+
+            explicit allocator(const char *dir, size_t max_size,
+                               libmemkind::allocation_policy alloc_policy) :
+                kind_wrapper_ptr(std::make_shared<kind_wrapper_t>(dir, max_size, alloc_policy))
+            {
+            }
+
+            explicit allocator(const std::string &dir, size_t max_size,
+                               libmemkind::allocation_policy alloc_policy) :
+                allocator(dir.c_str(), max_size, alloc_policy)
+            {
+            }
+
+            allocator(const allocator &other) = default;
+
+            template <typename U>
+            allocator(const allocator<U> &other) noexcept : kind_wrapper_ptr(
+                    other.kind_wrapper_ptr)
+            {
+            }
+
+            allocator(allocator &&other) = default;
+
+            template <typename U>
+            allocator(allocator<U> &&other) noexcept :
+                kind_wrapper_ptr(std::move(other.kind_wrapper_ptr))
+            {
+            }
+
+            allocator<T> &operator = (const allocator &other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (const allocator<U> &other) noexcept
+            {
+                kind_wrapper_ptr = other.kind_wrapper_ptr;
+                return *this;
+            }
+
+            allocator<T> &operator = (allocator &&other) = default;
+
+            template <typename U>
+            allocator<T> &operator = (allocator<U> &&other) noexcept
+            {
+                kind_wrapper_ptr = std::move(other.kind_wrapper_ptr);
+                return *this;
+            }
+
+            pointer allocate(size_type n) const
+            {
+                pointer result = static_cast<pointer>(memkind_malloc(kind_wrapper_ptr->get(),
+                                                                     n*sizeof(T)));
+                if (!result) {
+                    throw std::bad_alloc();
+                }
+                return result;
+            }
+
+            void deallocate(pointer p, size_type n) const
+            {
+                memkind_free(kind_wrapper_ptr->get(), static_cast<void *>(p));
+            }
+
+            template <class U, class... Args>
+            void construct(U *p, Args &&... args) const
+            {
+                ::new((void *)p) U(std::forward<Args>(args)...);
+            }
+
+            void destroy(pointer p) const
+            {
+                p->~value_type();
+            }
+
+            template <typename U, typename V>
+            friend bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs);
+
+            template <typename U, typename V>
+            friend bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs);
+        };
+
+        template <typename U, typename V>
+        bool operator ==(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return lhs.kind_wrapper_ptr->get() == rhs.kind_wrapper_ptr->get();
+        }
+
+        template <typename U, typename V>
+        bool operator !=(const allocator<U> &lhs, const allocator<V> &rhs)
+        {
+            return !(lhs == rhs);
+        }
+    }  // namespace pmem
+}  // namespace libmemkind
diff --git a/memkind-1.10.0/install_astyle.sh b/memkind-1.10.0/install_astyle.sh
new file mode 100755
index 0000000..58615fc
--- /dev/null
+++ b/memkind-1.10.0/install_astyle.sh
@@ -0,0 +1,31 @@
+#!/bin/bash
+#
+#  Copyright (C) 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+set -ex
+curl -SL https://sourceforge.net/projects/astyle/files/astyle/astyle%203.1/astyle_3.1_linux.tar.gz -o /tmp/astyle.tar.gz
+tar -xzvf /tmp/astyle.tar.gz -C /tmp/
+echo "Install astyle"
+cd /tmp/astyle/build/gcc
+make
+sudo make install
diff --git a/memkind-1.10.0/jemalloc/.appveyor.yml b/memkind-1.10.0/jemalloc/.appveyor.yml
new file mode 100644
index 0000000..90b0368
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/.appveyor.yml
@@ -0,0 +1,42 @@
+version: '{build}'
+
+environment:
+  matrix:
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    MSVC: amd64
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW32
+    CPU: i686
+    MSVC: x86
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW32
+    CPU: i686
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    MSVC: amd64
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+  - MSYSTEM: MINGW32
+    CPU: i686
+    MSVC: x86
+  - MSYSTEM: MINGW32
+    CPU: i686
+
+install:
+  - set PATH=c:\msys64\%MSYSTEM%\bin;c:\msys64\usr\bin;%PATH%
+  - if defined MSVC call "c:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\vcvarsall.bat" %MSVC%
+  - if defined MSVC pacman --noconfirm -Rsc mingw-w64-%CPU%-gcc gcc
+  - pacman --noconfirm -Suy mingw-w64-%CPU%-make
+
+build_script:
+  - bash -c "autoconf"
+  - bash -c "./configure $CONFIG_FLAGS"
+  - mingw32-make
+  - file lib/jemalloc.dll
+  - mingw32-make tests
+  - mingw32-make -k check
diff --git a/memkind-1.10.0/jemalloc/.autom4te.cfg b/memkind-1.10.0/jemalloc/.autom4te.cfg
new file mode 100644
index 0000000..fe2424d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/.autom4te.cfg
@@ -0,0 +1,3 @@
+begin-language: "Autoconf-without-aclocal-m4"
+args: --no-cache
+end-language: "Autoconf-without-aclocal-m4"
diff --git a/memkind-1.10.0/jemalloc/.gitattributes b/memkind-1.10.0/jemalloc/.gitattributes
new file mode 100644
index 0000000..6313b56
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/.gitattributes
@@ -0,0 +1 @@
+* text=auto eol=lf
diff --git a/memkind-1.10.0/jemalloc/.gitignore b/memkind-1.10.0/jemalloc/.gitignore
new file mode 100644
index 0000000..5ca0ad1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/.gitignore
@@ -0,0 +1,92 @@
+/bin/jemalloc-config
+/bin/jemalloc.sh
+/bin/jeprof
+
+/config.stamp
+/config.log
+/config.status
+/configure
+
+/doc/html.xsl
+/doc/manpages.xsl
+/doc/jemalloc.xml
+/doc/jemalloc.html
+/doc/jemalloc.3
+
+/jemalloc.pc
+
+/lib/
+
+/Makefile
+
+/include/jemalloc/internal/jemalloc_preamble.h
+/include/jemalloc/internal/jemalloc_internal_defs.h
+/include/jemalloc/internal/private_namespace.gen.h
+/include/jemalloc/internal/private_namespace.h
+/include/jemalloc/internal/private_namespace_jet.gen.h
+/include/jemalloc/internal/private_namespace_jet.h
+/include/jemalloc/internal/private_symbols.awk
+/include/jemalloc/internal/private_symbols_jet.awk
+/include/jemalloc/internal/public_namespace.h
+/include/jemalloc/internal/public_symbols.txt
+/include/jemalloc/internal/public_unnamespace.h
+/include/jemalloc/jemalloc.h
+/include/jemalloc/jemalloc_defs.h
+/include/jemalloc/jemalloc_macros.h
+/include/jemalloc/jemalloc_mangle.h
+/include/jemalloc/jemalloc_mangle_jet.h
+/include/jemalloc/jemalloc_protos.h
+/include/jemalloc/jemalloc_protos_jet.h
+/include/jemalloc/jemalloc_rename.h
+/include/jemalloc/jemalloc_typedefs.h
+
+/src/*.[od]
+/src/*.sym
+
+/run_tests.out/
+
+/test/test.sh
+test/include/test/jemalloc_test.h
+test/include/test/jemalloc_test_defs.h
+
+/test/integration/[A-Za-z]*
+!/test/integration/[A-Za-z]*.*
+/test/integration/*.[od]
+/test/integration/*.out
+
+/test/integration/cpp/[A-Za-z]*
+!/test/integration/cpp/[A-Za-z]*.*
+/test/integration/cpp/*.[od]
+/test/integration/cpp/*.out
+
+/test/src/*.[od]
+
+/test/stress/[A-Za-z]*
+!/test/stress/[A-Za-z]*.*
+/test/stress/*.[od]
+/test/stress/*.out
+
+/test/unit/[A-Za-z]*
+!/test/unit/[A-Za-z]*.*
+/test/unit/*.[od]
+/test/unit/*.out
+
+/VERSION
+
+*.pdb
+*.sdf
+*.opendb
+*.VC.db
+*.opensdf
+*.cachefile
+*.suo
+*.user
+*.sln.docstates
+*.tmp
+.vs/
+/msvc/Win32/
+/msvc/x64/
+/msvc/projects/*/*/Debug*/
+/msvc/projects/*/*/Release*/
+/msvc/projects/*/*/Win32/
+/msvc/projects/*/*/x64/
diff --git a/memkind-1.10.0/jemalloc/.travis.yml b/memkind-1.10.0/jemalloc/.travis.yml
new file mode 100644
index 0000000..2da5da8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/.travis.yml
@@ -0,0 +1,195 @@
+language: generic
+dist: precise
+
+matrix:
+  include:
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: &gcc_multilib
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons: *gcc_multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --disable-libdl" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --enable-opt-safety-checks" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-libdl --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-opt-safety-checks --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary,percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    # Development build
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-cache-oblivious --enable-stats --enable-log --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    # --enable-expermental-smallocx:
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-experimental-smallocx --enable-stats --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+
+    # Valgrind
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds" JEMALLOC_TEST_PREFIX="valgrind"
+      addons:
+        apt:
+          packages:
+            - valgrind
+
+
+before_script:
+  - autoconf
+  - scripts/gen_travis.py > travis_script && diff .travis.yml travis_script
+  - ./configure ${COMPILER_FLAGS:+       CC="$CC $COMPILER_FLAGS"       CXX="$CXX $COMPILER_FLAGS" }       $CONFIGURE_FLAGS
+  - make -j3
+  - make -j3 tests
+
+script:
+  - make check
+
diff --git a/memkind-1.10.0/jemalloc/COPYING b/memkind-1.10.0/jemalloc/COPYING
new file mode 100644
index 0000000..3b7fd35
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/COPYING
@@ -0,0 +1,27 @@
+Unless otherwise specified, files in the jemalloc source distribution are
+subject to the following license:
+--------------------------------------------------------------------------------
+Copyright (C) 2002-present Jason Evans <jasone@canonware.com>.
+All rights reserved.
+Copyright (C) 2007-2012 Mozilla Foundation.  All rights reserved.
+Copyright (C) 2009-present Facebook, Inc.  All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+1. Redistributions of source code must retain the above copyright notice(s),
+   this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright notice(s),
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+--------------------------------------------------------------------------------
diff --git a/memkind-1.10.0/jemalloc/ChangeLog b/memkind-1.10.0/jemalloc/ChangeLog
new file mode 100644
index 0000000..e55813b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/ChangeLog
@@ -0,0 +1,1518 @@
+Following are change highlights associated with official releases.  Important
+bug fixes are all mentioned, but some internal enhancements are omitted here for
+brevity.  Much more detail can be found in the git revision history:
+
+    https://github.com/jemalloc/jemalloc
+
+* 5.2.1 (August 5, 2019)
+
+  This release is primarily about Windows.  A critical virtual memory leak is
+  resolved on all Windows platforms.  The regression was present in all releases
+  since 5.0.0.
+
+  Bug fixes:
+  - Fix a severe virtual memory leak on Windows.  This regression was first
+    released in 5.0.0.  (@Ignition, @j0t, @frederik-h, @davidtgoldblatt,
+    @interwq)
+  - Fix size 0 handling in posix_memalign().  This regression was first released
+    in 5.2.0.  (@interwq)
+  - Fix the prof_log unit test which may observe unexpected backtraces from
+    compiler optimizations.  The test was first added in 5.2.0.  (@marxin,
+    @gnzlbg, @interwq)
+  - Fix the declaration of the extent_avail tree.  This regression was first
+    released in 5.1.0.  (@zoulasc)
+  - Fix an incorrect reference in jeprof.  This functionality was first released
+    in 3.0.0.  (@prehistoric-penguin)
+  - Fix an assertion on the deallocation fast-path.  This regression was first
+    released in 5.2.0.  (@yinan1048576)
+  - Fix the TLS_MODEL attribute in headers.  This regression was first released
+    in 5.0.0.  (@zoulasc, @interwq)
+
+  Optimizations and refactors:
+  - Implement opt.retain on Windows and enable by default on 64-bit.  (@interwq,
+    @davidtgoldblatt)
+  - Optimize away a branch on the operator delete[] path.  (@mgrice)
+  - Add format annotation to the format generator function.  (@zoulasc)
+  - Refactor and improve the size class header generation.  (@yinan1048576)
+  - Remove best fit.  (@djwatson)
+  - Avoid blocking on background thread locks for stats.  (@oranagra, @interwq)
+
+* 5.2.0 (April 2, 2019)
+
+  This release includes a few notable improvements, which are summarized below:
+  1) improved fast-path performance from the optimizations by @djwatson; 2)
+  reduced virtual memory fragmentation and metadata usage; and 3) bug fixes on
+  setting the number of background threads.  In addition, peak / spike memory
+  usage is improved with certain allocation patterns.  As usual, the release and
+  prior dev versions have gone through large-scale production testing.
+
+  New features:
+  - Implement oversize_threshold, which uses a dedicated arena for allocations
+    crossing the specified threshold to reduce fragmentation.  (@interwq)
+  - Add extents usage information to stats.  (@tyleretzel)
+  - Log time information for sampled allocations.  (@tyleretzel)
+  - Support 0 size in sdallocx.  (@djwatson)
+  - Output rate for certain counters in malloc_stats.  (@zinoale)
+  - Add configure option --enable-readlinkat, which allows the use of readlinkat
+    over readlink.  (@davidtgoldblatt)
+  - Add configure options --{enable,disable}-{static,shared} to allow not
+    building unwanted libraries.  (@Ericson2314)
+  - Add configure option --disable-libdl to enable fully static builds.
+    (@interwq)
+  - Add mallctl interfaces:
+	+ opt.oversize_threshold (@interwq)
+	+ stats.arenas.<i>.extent_avail (@tyleretzel)
+	+ stats.arenas.<i>.extents.<j>.n{dirty,muzzy,retained} (@tyleretzel)
+	+ stats.arenas.<i>.extents.<j>.{dirty,muzzy,retained}_bytes
+	  (@tyleretzel)
+
+  Portability improvements:
+  - Update MSVC builds.  (@maksqwe, @rustyx)
+  - Workaround a compiler optimizer bug on s390x.  (@rkmisra)
+  - Make use of pthread_set_name_np(3) on FreeBSD.  (@trasz)
+  - Implement malloc_getcpu() to enable percpu_arena for windows.  (@santagada)
+  - Link against -pthread instead of -lpthread.  (@paravoid)
+  - Make background_thread not dependent on libdl.  (@interwq)
+  - Add stringify to fix a linker directive issue on MSVC.  (@daverigby)
+  - Detect and fall back when 8-bit atomics are unavailable.  (@interwq)
+  - Fall back to the default pthread_create if dlsym(3) fails.  (@interwq)
+
+  Optimizations and refactors:
+  - Refactor the TSD module.  (@davidtgoldblatt)
+  - Avoid taking extents_muzzy mutex when muzzy is disabled.  (@interwq)
+  - Avoid taking large_mtx for auto arenas on the tcache flush path.  (@interwq)
+  - Optimize ixalloc by avoiding a size lookup.  (@interwq)
+  - Implement opt.oversize_threshold which uses a dedicated arena for requests
+    crossing the threshold, also eagerly purges the oversize extents.  Default
+    the threshold to 8 MiB.  (@interwq)
+  - Clean compilation with -Wextra.  (@gnzlbg, @jasone)
+  - Refactor the size class module.  (@davidtgoldblatt)
+  - Refactor the stats emitter.  (@tyleretzel)
+  - Optimize pow2_ceil.  (@rkmisra)
+  - Avoid runtime detection of lazy purging on FreeBSD.  (@trasz)
+  - Optimize mmap(2) alignment handling on FreeBSD.  (@trasz)
+  - Improve error handling for THP state initialization.  (@jsteemann)
+  - Rework the malloc() fast path.  (@djwatson)
+  - Rework the free() fast path.  (@djwatson)
+  - Refactor and optimize the tcache fill / flush paths.  (@djwatson)
+  - Optimize sync / lwsync on PowerPC.  (@chmeeedalf)
+  - Bypass extent_dalloc() when retain is enabled.  (@interwq)
+  - Optimize the locking on large deallocation.  (@interwq)
+  - Reduce the number of pages committed from sanity checking in debug build.
+    (@trasz, @interwq)
+  - Deprecate OSSpinLock.  (@interwq)
+  - Lower the default number of background threads to 4 (when the feature
+    is enabled).  (@interwq)
+  - Optimize the trylock spin wait.  (@djwatson)
+  - Use arena index for arena-matching checks.  (@interwq)
+  - Avoid forced decay on thread termination when using background threads.
+    (@interwq)
+  - Disable muzzy decay by default.  (@djwatson, @interwq)
+  - Only initialize libgcc unwinder when profiling is enabled.  (@paravoid,
+    @interwq)
+
+  Bug fixes (all only relevant to jemalloc 5.x):
+  - Fix background thread index issues with max_background_threads.  (@djwatson,
+    @interwq)
+  - Fix stats output for opt.lg_extent_max_active_fit.  (@interwq)
+  - Fix opt.prof_prefix initialization.  (@davidtgoldblatt)
+  - Properly trigger decay on tcache destroy.  (@interwq, @amosbird)
+  - Fix tcache.flush.  (@interwq)
+  - Detect whether explicit extent zero out is necessary with huge pages or
+    custom extent hooks, which may change the purge semantics.  (@interwq)
+  - Fix a side effect caused by extent_max_active_fit combined with decay-based
+    purging, where freed extents can accumulate and not be reused for an
+    extended period of time.  (@interwq, @mpghf)
+  - Fix a missing unlock on extent register error handling.  (@zoulasc)
+
+  Testing:
+  - Simplify the Travis script output.  (@gnzlbg)
+  - Update the test scripts for FreeBSD.  (@devnexen)
+  - Add unit tests for the producer-consumer pattern.  (@interwq)
+  - Add Cirrus-CI config for FreeBSD builds.  (@jasone)
+  - Add size-matching sanity checks on tcache flush.  (@davidtgoldblatt,
+    @interwq)
+
+  Incompatible changes:
+  - Remove --with-lg-page-sizes.  (@davidtgoldblatt)
+
+  Documentation:
+  - Attempt to build docs by default, however skip doc building when xsltproc
+    is missing. (@interwq, @cmuellner)
+
+* 5.1.0 (May 4, 2018)
+
+  This release is primarily about fine-tuning, ranging from several new features
+  to numerous notable performance and portability enhancements.  The release and
+  prior dev versions have been running in multiple large scale applications for
+  months, and the cumulative improvements are substantial in many cases.
+
+  Given the long and successful production runs, this release is likely a good
+  candidate for applications to upgrade, from both jemalloc 5.0 and before.  For
+  performance-critical applications, the newly added TUNING.md provides
+  guidelines on jemalloc tuning.
+
+  New features:
+  - Implement transparent huge page support for internal metadata.  (@interwq)
+  - Add opt.thp to allow enabling / disabling transparent huge pages for all
+    mappings.  (@interwq)
+  - Add maximum background thread count option.  (@djwatson)
+  - Allow prof_active to control opt.lg_prof_interval and prof.gdump.
+    (@interwq)
+  - Allow arena index lookup based on allocation addresses via mallctl.
+    (@lionkov)
+  - Allow disabling initial-exec TLS model.  (@davidtgoldblatt, @KenMacD)
+  - Add opt.lg_extent_max_active_fit to set the max ratio between the size of
+    the active extent selected (to split off from) and the size of the requested
+    allocation.  (@interwq, @davidtgoldblatt)
+  - Add retain_grow_limit to set the max size when growing virtual address
+    space.  (@interwq)
+  - Add mallctl interfaces:
+    + arena.<i>.retain_grow_limit  (@interwq)
+    + arenas.lookup  (@lionkov)
+    + max_background_threads  (@djwatson)
+    + opt.lg_extent_max_active_fit  (@interwq)
+    + opt.max_background_threads  (@djwatson)
+    + opt.metadata_thp  (@interwq)
+    + opt.thp  (@interwq)
+    + stats.metadata_thp  (@interwq)
+
+  Portability improvements:
+  - Support GNU/kFreeBSD configuration.  (@paravoid)
+  - Support m68k, nios2 and SH3 architectures.  (@paravoid)
+  - Fall back to FD_CLOEXEC when O_CLOEXEC is unavailable.  (@zonyitoo)
+  - Fix symbol listing for cross-compiling.  (@tamird)
+  - Fix high bits computation on ARM.  (@davidtgoldblatt, @paravoid)
+  - Disable the CPU_SPINWAIT macro for Power.  (@davidtgoldblatt, @marxin)
+  - Fix MSVC 2015 & 2017 builds.  (@rustyx)
+  - Improve RISC-V support.  (@EdSchouten)
+  - Set name mangling script in strict mode.  (@nicolov)
+  - Avoid MADV_HUGEPAGE on ARM.  (@marxin)
+  - Modify configure to determine return value of strerror_r.
+    (@davidtgoldblatt, @cferris1000)
+  - Make sure CXXFLAGS is tested with CPP compiler.  (@nehaljwani)
+  - Fix 32-bit build on MSVC.  (@rustyx)
+  - Fix external symbol on MSVC.  (@maksqwe)
+  - Avoid a printf format specifier warning.  (@jasone)
+  - Add configure option --disable-initial-exec-tls which can allow jemalloc to
+    be dynamically loaded after program startup.  (@davidtgoldblatt, @KenMacD)
+  - AArch64: Add ILP32 support.  (@cmuellner)
+  - Add --with-lg-vaddr configure option to support cross compiling.
+    (@cmuellner, @davidtgoldblatt)
+
+  Optimizations and refactors:
+  - Improve active extent fit with extent_max_active_fit.  This considerably
+    reduces fragmentation over time and improves virtual memory and metadata
+    usage.  (@davidtgoldblatt, @interwq)
+  - Eagerly coalesce large extents to reduce fragmentation.  (@interwq)
+  - sdallocx: only read size info when page aligned (i.e. possibly sampled),
+    which speeds up the sized deallocation path significantly.  (@interwq)
+  - Avoid attempting new mappings for in place expansion with retain, since
+    it rarely succeeds in practice and causes high overhead.  (@interwq)
+  - Refactor OOM handling in newImpl.  (@wqfish)
+  - Add internal fine-grained logging functionality for debugging use.
+    (@davidtgoldblatt)
+  - Refactor arena / tcache interactions.  (@davidtgoldblatt)
+  - Refactor extent management with dumpable flag.  (@davidtgoldblatt)
+  - Add runtime detection of lazy purging.  (@interwq)
+  - Use pairing heap instead of red-black tree for extents_avail.  (@djwatson)
+  - Use sysctl on startup in FreeBSD.  (@trasz)
+  - Use thread local prng state instead of atomic.  (@djwatson)
+  - Make decay to always purge one more extent than before, because in
+    practice large extents are usually the ones that cross the decay threshold.
+    Purging the additional extent helps save memory as well as reduce VM
+    fragmentation.  (@interwq)
+  - Fast division by dynamic values.  (@davidtgoldblatt)
+  - Improve the fit for aligned allocation.  (@interwq, @edwinsmith)
+  - Refactor extent_t bitpacking.  (@rkmisra)
+  - Optimize the generated assembly for ticker operations.  (@davidtgoldblatt)
+  - Convert stats printing to use a structured text emitter.  (@davidtgoldblatt)
+  - Remove preserve_lru feature for extents management.  (@djwatson)
+  - Consolidate two memory loads into one on the fast deallocation path.
+    (@davidtgoldblatt, @interwq)
+
+  Bug fixes (most of the issues are only relevant to jemalloc 5.0):
+  - Fix deadlock with multithreaded fork in OS X.  (@davidtgoldblatt)
+  - Validate returned file descriptor before use.  (@zonyitoo)
+  - Fix a few background thread initialization and shutdown issues.  (@interwq)
+  - Fix an extent coalesce + decay race by taking both coalescing extents off
+    the LRU list.  (@interwq)
+  - Fix potentially unbound increase during decay, caused by one thread keep
+    stashing memory to purge while other threads generating new pages.  The
+    number of pages to purge is checked to prevent this.  (@interwq)
+  - Fix a FreeBSD bootstrap assertion.  (@strejda, @interwq)
+  - Handle 32 bit mutex counters.  (@rkmisra)
+  - Fix a indexing bug when creating background threads.  (@davidtgoldblatt,
+    @binliu19)
+  - Fix arguments passed to extent_init.  (@yuleniwo, @interwq)
+  - Fix addresses used for ordering mutexes.  (@rkmisra)
+  - Fix abort_conf processing during bootstrap.  (@interwq)
+  - Fix include path order for out-of-tree builds.  (@cmuellner)
+
+  Incompatible changes:
+  - Remove --disable-thp.  (@interwq)
+  - Remove mallctl interfaces:
+    + config.thp  (@interwq)
+
+  Documentation:
+  - Add TUNING.md.  (@interwq, @davidtgoldblatt, @djwatson)
+
+* 5.0.1 (July 1, 2017)
+
+  This bugfix release fixes several issues, most of which are obscure enough
+  that typical applications are not impacted.
+
+  Bug fixes:
+  - Update decay->nunpurged before purging, in order to avoid potential update
+    races and subsequent incorrect purging volume.  (@interwq)
+  - Only abort on dlsym(3) error if the failure impacts an enabled feature (lazy
+    locking and/or background threads).  This mitigates an initialization
+    failure bug for which we still do not have a clear reproduction test case.
+    (@interwq)
+  - Modify tsd management so that it neither crashes nor leaks if a thread's
+    only allocation activity is to call free() after TLS destructors have been
+    executed.  This behavior was observed when operating with GNU libc, and is
+    unlikely to be an issue with other libc implementations.  (@interwq)
+  - Mask signals during background thread creation.  This prevents signals from
+    being inadvertently delivered to background threads.  (@jasone,
+    @davidtgoldblatt, @interwq)
+  - Avoid inactivity checks within background threads, in order to prevent
+    recursive mutex acquisition.  (@interwq)
+  - Fix extent_grow_retained() to use the specified hooks when the
+    arena.<i>.extent_hooks mallctl is used to override the default hooks.
+    (@interwq)
+  - Add missing reentrancy support for custom extent hooks which allocate.
+    (@interwq)
+  - Post-fork(2), re-initialize the list of tcaches associated with each arena
+    to contain no tcaches except the forking thread's.  (@interwq)
+  - Add missing post-fork(2) mutex reinitialization for extent_grow_mtx.  This
+    fixes potential deadlocks after fork(2).  (@interwq)
+  - Enforce minimum autoconf version (currently 2.68), since 2.63 is known to
+    generate corrupt configure scripts.  (@jasone)
+  - Ensure that the configured page size (--with-lg-page) is no larger than the
+    configured huge page size (--with-lg-hugepage).  (@jasone)
+
+* 5.0.0 (June 13, 2017)
+
+  Unlike all previous jemalloc releases, this release does not use naturally
+  aligned "chunks" for virtual memory management, and instead uses page-aligned
+  "extents".  This change has few externally visible effects, but the internal
+  impacts are... extensive.  Many other internal changes combine to make this
+  the most cohesively designed version of jemalloc so far, with ample
+  opportunity for further enhancements.
+
+  Continuous integration is now an integral aspect of development thanks to the
+  efforts of @davidtgoldblatt, and the dev branch tends to remain reasonably
+  stable on the tested platforms (Linux, FreeBSD, macOS, and Windows).  As a
+  side effect the official release frequency may decrease over time.
+
+  New features:
+  - Implement optional per-CPU arena support; threads choose which arena to use
+    based on current CPU rather than on fixed thread-->arena associations.
+    (@interwq)
+  - Implement two-phase decay of unused dirty pages.  Pages transition from
+    dirty-->muzzy-->clean, where the first phase transition relies on
+    madvise(... MADV_FREE) semantics, and the second phase transition discards
+    pages such that they are replaced with demand-zeroed pages on next access.
+    (@jasone)
+  - Increase decay time resolution from seconds to milliseconds.  (@jasone)
+  - Implement opt-in per CPU background threads, and use them for asynchronous
+    decay-driven unused dirty page purging.  (@interwq)
+  - Add mutex profiling, which collects a variety of statistics useful for
+    diagnosing overhead/contention issues.  (@interwq)
+  - Add C++ new/delete operator bindings.  (@djwatson)
+  - Support manually created arena destruction, such that all data and metadata
+    are discarded.  Add MALLCTL_ARENAS_DESTROYED for accessing merged stats
+    associated with destroyed arenas.  (@jasone)
+  - Add MALLCTL_ARENAS_ALL as a fixed index for use in accessing
+    merged/destroyed arena statistics via mallctl.  (@jasone)
+  - Add opt.abort_conf to optionally abort if invalid configuration options are
+    detected during initialization.  (@interwq)
+  - Add opt.stats_print_opts, so that e.g. JSON output can be selected for the
+    stats dumped during exit if opt.stats_print is true.  (@jasone)
+  - Add --with-version=VERSION for use when embedding jemalloc into another
+    project's git repository.  (@jasone)
+  - Add --disable-thp to support cross compiling.  (@jasone)
+  - Add --with-lg-hugepage to support cross compiling.  (@jasone)
+  - Add mallctl interfaces (various authors):
+    + background_thread
+    + opt.abort_conf
+    + opt.retain
+    + opt.percpu_arena
+    + opt.background_thread
+    + opt.{dirty,muzzy}_decay_ms
+    + opt.stats_print_opts
+    + arena.<i>.initialized
+    + arena.<i>.destroy
+    + arena.<i>.{dirty,muzzy}_decay_ms
+    + arena.<i>.extent_hooks
+    + arenas.{dirty,muzzy}_decay_ms
+    + arenas.bin.<i>.slab_size
+    + arenas.nlextents
+    + arenas.lextent.<i>.size
+    + arenas.create
+    + stats.background_thread.{num_threads,num_runs,run_interval}
+    + stats.mutexes.{ctl,background_thread,prof,reset}.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+    + stats.arenas.<i>.{dirty,muzzy}_decay_ms
+    + stats.arenas.<i>.uptime
+    + stats.arenas.<i>.{pmuzzy,base,internal,resident}
+    + stats.arenas.<i>.{dirty,muzzy}_{npurge,nmadvise,purged}
+    + stats.arenas.<i>.bins.<j>.{nslabs,reslabs,curslabs}
+    + stats.arenas.<i>.bins.<j>.mutex.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+    + stats.arenas.<i>.lextents.<j>.{nmalloc,ndalloc,nrequests,curlextents}
+    + stats.arenas.i.mutexes.{large,extent_avail,extents_dirty,extents_muzzy,
+      extents_retained,decay_dirty,decay_muzzy,base,tcache_list}.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+
+  Portability improvements:
+  - Improve reentrant allocation support, such that deadlock is less likely if
+    e.g. a system library call in turn allocates memory.  (@davidtgoldblatt,
+    @interwq)
+  - Support static linking of jemalloc with glibc.  (@djwatson)
+
+  Optimizations and refactors:
+  - Organize virtual memory as "extents" of virtual memory pages, rather than as
+    naturally aligned "chunks", and store all metadata in arbitrarily distant
+    locations.  This reduces virtual memory external fragmentation, and will
+    interact better with huge pages (not yet explicitly supported).  (@jasone)
+  - Fold large and huge size classes together; only small and large size classes
+    remain.  (@jasone)
+  - Unify the allocation paths, and merge most fast-path branching decisions.
+    (@davidtgoldblatt, @interwq)
+  - Embed per thread automatic tcache into thread-specific data, which reduces
+    conditional branches and dereferences.  Also reorganize tcache to increase
+    fast-path data locality.  (@interwq)
+  - Rewrite atomics to closely model the C11 API, convert various
+    synchronization from mutex-based to atomic, and use the explicit memory
+    ordering control to resolve various hypothetical races without increasing
+    synchronization overhead.  (@davidtgoldblatt)
+  - Extensively optimize rtree via various methods:
+    + Add multiple layers of rtree lookup caching, since rtree lookups are now
+      part of fast-path deallocation.  (@interwq)
+    + Determine rtree layout at compile time.  (@jasone)
+    + Make the tree shallower for common configurations.  (@jasone)
+    + Embed the root node in the top-level rtree data structure, thus avoiding
+      one level of indirection.  (@jasone)
+    + Further specialize leaf elements as compared to internal node elements,
+      and directly embed extent metadata needed for fast-path deallocation.
+      (@jasone)
+    + Ignore leading always-zero address bits (architecture-specific).
+      (@jasone)
+  - Reorganize headers (ongoing work) to make them hermetic, and disentangle
+    various module dependencies.  (@davidtgoldblatt)
+  - Convert various internal data structures such as size class metadata from
+    boot-time-initialized to compile-time-initialized.  Propagate resulting data
+    structure simplifications, such as making arena metadata fixed-size.
+    (@jasone)
+  - Simplify size class lookups when constrained to size classes that are
+    multiples of the page size.  This speeds lookups, but the primary benefit is
+    complexity reduction in code that was the source of numerous regressions.
+    (@jasone)
+  - Lock individual extents when possible for localized extent operations,
+    rather than relying on a top-level arena lock.  (@davidtgoldblatt, @jasone)
+  - Use first fit layout policy instead of best fit, in order to improve
+    packing.  (@jasone)
+  - If munmap(2) is not in use, use an exponential series to grow each arena's
+    virtual memory, so that the number of disjoint virtual memory mappings
+    remains low.  (@jasone)
+  - Implement per arena base allocators, so that arenas never share any virtual
+    memory pages.  (@jasone)
+  - Automatically generate private symbol name mangling macros.  (@jasone)
+
+  Incompatible changes:
+  - Replace chunk hooks with an expanded/normalized set of extent hooks.
+    (@jasone)
+  - Remove ratio-based purging.  (@jasone)
+  - Remove --disable-tcache.  (@jasone)
+  - Remove --disable-tls.  (@jasone)
+  - Remove --enable-ivsalloc.  (@jasone)
+  - Remove --with-lg-size-class-group.  (@jasone)
+  - Remove --with-lg-tiny-min.  (@jasone)
+  - Remove --disable-cc-silence.  (@jasone)
+  - Remove --enable-code-coverage.  (@jasone)
+  - Remove --disable-munmap (replaced by opt.retain).  (@jasone)
+  - Remove Valgrind support.  (@jasone)
+  - Remove quarantine support.  (@jasone)
+  - Remove redzone support.  (@jasone)
+  - Remove mallctl interfaces (various authors):
+    + config.munmap
+    + config.tcache
+    + config.tls
+    + config.valgrind
+    + opt.lg_chunk
+    + opt.purge
+    + opt.lg_dirty_mult
+    + opt.decay_time
+    + opt.quarantine
+    + opt.redzone
+    + opt.thp
+    + arena.<i>.lg_dirty_mult
+    + arena.<i>.decay_time
+    + arena.<i>.chunk_hooks
+    + arenas.initialized
+    + arenas.lg_dirty_mult
+    + arenas.decay_time
+    + arenas.bin.<i>.run_size
+    + arenas.nlruns
+    + arenas.lrun.<i>.size
+    + arenas.nhchunks
+    + arenas.hchunk.<i>.size
+    + arenas.extend
+    + stats.cactive
+    + stats.arenas.<i>.lg_dirty_mult
+    + stats.arenas.<i>.decay_time
+    + stats.arenas.<i>.metadata.{mapped,allocated}
+    + stats.arenas.<i>.{npurge,nmadvise,purged}
+    + stats.arenas.<i>.huge.{allocated,nmalloc,ndalloc,nrequests}
+    + stats.arenas.<i>.bins.<j>.{nruns,reruns,curruns}
+    + stats.arenas.<i>.lruns.<j>.{nmalloc,ndalloc,nrequests,curruns}
+    + stats.arenas.<i>.hchunks.<j>.{nmalloc,ndalloc,nrequests,curhchunks}
+
+  Bug fixes:
+  - Improve interval-based profile dump triggering to dump only one profile when
+    a single allocation's size exceeds the interval.  (@jasone)
+  - Use prefixed function names (as controlled by --with-jemalloc-prefix) when
+    pruning backtrace frames in jeprof.  (@jasone)
+
+* 4.5.0 (February 28, 2017)
+
+  This is the first release to benefit from much broader continuous integration
+  testing, thanks to @davidtgoldblatt.  Had we had this testing infrastructure
+  in place for prior releases, it would have caught all of the most serious
+  regressions fixed by this release.
+
+  New features:
+  - Add --disable-thp and the opt.thp mallctl to provide opt-out mechanisms for
+    transparent huge page integration.  (@jasone)
+  - Update zone allocator integration to work with macOS 10.12.  (@glandium)
+  - Restructure *CFLAGS configuration, so that CFLAGS behaves typically, and
+    EXTRA_CFLAGS provides a way to specify e.g. -Werror during building, but not
+    during configuration.  (@jasone, @ronawho)
+
+  Bug fixes:
+  - Fix DSS (sbrk(2)-based) allocation.  This regression was first released in
+    4.3.0.  (@jasone)
+  - Handle race in per size class utilization computation.  This functionality
+    was first released in 4.0.0.  (@interwq)
+  - Fix lock order reversal during gdump.  (@jasone)
+  - Fix/refactor tcache synchronization.  This regression was first released in
+    4.0.0.  (@jasone)
+  - Fix various JSON-formatted malloc_stats_print() bugs.  This functionality
+    was first released in 4.3.0.  (@jasone)
+  - Fix huge-aligned allocation.  This regression was first released in 4.4.0.
+    (@jasone)
+  - When transparent huge page integration is enabled, detect what state pages
+    start in according to the kernel's current operating mode, and only convert
+    arena chunks to non-huge during purging if that is not their initial state.
+    This functionality was first released in 4.4.0.  (@jasone)
+  - Fix lg_chunk clamping for the --enable-cache-oblivious --disable-fill case.
+    This regression was first released in 4.0.0.  (@jasone, @428desmo)
+  - Properly detect sparc64 when building for Linux.  (@glaubitz)
+
+* 4.4.0 (December 3, 2016)
+
+  New features:
+  - Add configure support for *-*-linux-android.  (@cferris1000, @jasone)
+  - Add the --disable-syscall configure option, for use on systems that place
+    security-motivated limitations on syscall(2).  (@jasone)
+  - Add support for Debian GNU/kFreeBSD.  (@thesam)
+
+  Optimizations:
+  - Add extent serial numbers and use them where appropriate as a sort key that
+    is higher priority than address, so that the allocation policy prefers older
+    extents.  This tends to improve locality (decrease fragmentation) when
+    memory grows downward.  (@jasone)
+  - Refactor madvise(2) configuration so that MADV_FREE is detected and utilized
+    on Linux 4.5 and newer.  (@jasone)
+  - Mark partially purged arena chunks as non-huge-page.  This improves
+    interaction with Linux's transparent huge page functionality.  (@jasone)
+
+  Bug fixes:
+  - Fix size class computations for edge conditions involving extremely large
+    allocations.  This regression was first released in 4.0.0.  (@jasone,
+    @ingvarha)
+  - Remove overly restrictive assertions related to the cactive statistic.  This
+    regression was first released in 4.1.0.  (@jasone)
+  - Implement a more reliable detection scheme for os_unfair_lock on macOS.
+    (@jszakmeister)
+
+* 4.3.1 (November 7, 2016)
+
+  Bug fixes:
+  - Fix a severe virtual memory leak.  This regression was first released in
+    4.3.0.  (@interwq, @jasone)
+  - Refactor atomic and prng APIs to restore support for 32-bit platforms that
+    use pre-C11 toolchains, e.g. FreeBSD's mips.  (@jasone)
+
+* 4.3.0 (November 4, 2016)
+
+  This is the first release that passes the test suite for multiple Windows
+  configurations, thanks in large part to @glandium setting up continuous
+  integration via AppVeyor (and Travis CI for Linux and OS X).
+
+  New features:
+  - Add "J" (JSON) support to malloc_stats_print().  (@jasone)
+  - Add Cray compiler support.  (@ronawho)
+
+  Optimizations:
+  - Add/use adaptive spinning for bootstrapping and radix tree node
+    initialization.  (@jasone)
+
+  Bug fixes:
+  - Fix large allocation to search starting in the optimal size class heap,
+    which can substantially reduce virtual memory churn and fragmentation.  This
+    regression was first released in 4.0.0.  (@mjp41, @jasone)
+  - Fix stats.arenas.<i>.nthreads accounting.  (@interwq)
+  - Fix and simplify decay-based purging.  (@jasone)
+  - Make DSS (sbrk(2)-related) operations lockless, which resolves potential
+    deadlocks during thread exit.  (@jasone)
+  - Fix over-sized allocation of radix tree leaf nodes.  (@mjp41, @ogaun,
+    @jasone)
+  - Fix over-sized allocation of arena_t (plus associated stats) data
+    structures.  (@jasone, @interwq)
+  - Fix EXTRA_CFLAGS to not affect configuration.  (@jasone)
+  - Fix a Valgrind integration bug.  (@ronawho)
+  - Disallow 0x5a junk filling when running in Valgrind.  (@jasone)
+  - Fix a file descriptor leak on Linux.  This regression was first released in
+    4.2.0.  (@vsarunas, @jasone)
+  - Fix static linking of jemalloc with glibc.  (@djwatson)
+  - Use syscall(2) rather than {open,read,close}(2) during boot on Linux.  This
+    works around other libraries' system call wrappers performing reentrant
+    allocation.  (@kspinka, @Whissi, @jasone)
+  - Fix OS X default zone replacement to work with OS X 10.12.  (@glandium,
+    @jasone)
+  - Fix cached memory management to avoid needless commit/decommit operations
+    during purging, which resolves permanent virtual memory map fragmentation
+    issues on Windows.  (@mjp41, @jasone)
+  - Fix TSD fetches to avoid (recursive) allocation.  This is relevant to
+    non-TLS and Windows configurations.  (@jasone)
+  - Fix malloc_conf overriding to work on Windows.  (@jasone)
+  - Forcibly disable lazy-lock on Windows (was forcibly *enabled*).  (@jasone)
+
+* 4.2.1 (June 8, 2016)
+
+  Bug fixes:
+  - Fix bootstrapping issues for configurations that require allocation during
+    tsd initialization (e.g. --disable-tls).  (@cferris1000, @jasone)
+  - Fix gettimeofday() version of nstime_update().  (@ronawho)
+  - Fix Valgrind regressions in calloc() and chunk_alloc_wrapper().  (@ronawho)
+  - Fix potential VM map fragmentation regression.  (@jasone)
+  - Fix opt_zero-triggered in-place huge reallocation zeroing.  (@jasone)
+  - Fix heap profiling context leaks in reallocation edge cases.  (@jasone)
+
+* 4.2.0 (May 12, 2016)
+
+  New features:
+  - Add the arena.<i>.reset mallctl, which makes it possible to discard all of
+    an arena's allocations in a single operation.  (@jasone)
+  - Add the stats.retained and stats.arenas.<i>.retained statistics.  (@jasone)
+  - Add the --with-version configure option.  (@jasone)
+  - Support --with-lg-page values larger than actual page size.  (@jasone)
+
+  Optimizations:
+  - Use pairing heaps rather than red-black trees for various hot data
+    structures.  (@djwatson, @jasone)
+  - Streamline fast paths of rtree operations.  (@jasone)
+  - Optimize the fast paths of calloc() and [m,d,sd]allocx().  (@jasone)
+  - Decommit unused virtual memory if the OS does not overcommit.  (@jasone)
+  - Specify MAP_NORESERVE on Linux if [heuristic] overcommit is active, in order
+    to avoid unfortunate interactions during fork(2).  (@jasone)
+
+  Bug fixes:
+  - Fix chunk accounting related to triggering gdump profiles.  (@jasone)
+  - Link against librt for clock_gettime(2) if glibc < 2.17.  (@jasone)
+  - Scale leak report summary according to sampling probability.  (@jasone)
+
+* 4.1.1 (May 3, 2016)
+
+  This bugfix release resolves a variety of mostly minor issues, though the
+  bitmap fix is critical for 64-bit Windows.
+
+  Bug fixes:
+  - Fix the linear scan version of bitmap_sfu() to shift by the proper amount
+    even when sizeof(long) is not the same as sizeof(void *), as on 64-bit
+    Windows.  (@jasone)
+  - Fix hashing functions to avoid unaligned memory accesses (and resulting
+    crashes).  This is relevant at least to some ARM-based platforms.
+    (@rkmisra)
+  - Fix fork()-related lock rank ordering reversals.  These reversals were
+    unlikely to cause deadlocks in practice except when heap profiling was
+    enabled and active.  (@jasone)
+  - Fix various chunk leaks in OOM code paths.  (@jasone)
+  - Fix malloc_stats_print() to print opt.narenas correctly.  (@jasone)
+  - Fix MSVC-specific build/test issues.  (@rustyx, @yuslepukhin)
+  - Fix a variety of test failures that were due to test fragility rather than
+    core bugs.  (@jasone)
+
+* 4.1.0 (February 28, 2016)
+
+  This release is primarily about optimizations, but it also incorporates a lot
+  of portability-motivated refactoring and enhancements.  Many people worked on
+  this release, to an extent that even with the omission here of minor changes
+  (see git revision history), and of the people who reported and diagnosed
+  issues, so much of the work was contributed that starting with this release,
+  changes are annotated with author credits to help reflect the collaborative
+  effort involved.
+
+  New features:
+  - Implement decay-based unused dirty page purging, a major optimization with
+    mallctl API impact.  This is an alternative to the existing ratio-based
+    unused dirty page purging, and is intended to eventually become the sole
+    purging mechanism.  New mallctls:
+    + opt.purge
+    + opt.decay_time
+    + arena.<i>.decay
+    + arena.<i>.decay_time
+    + arenas.decay_time
+    + stats.arenas.<i>.decay_time
+    (@jasone, @cevans87)
+  - Add --with-malloc-conf, which makes it possible to embed a default
+    options string during configuration.  This was motivated by the desire to
+    specify --with-malloc-conf=purge:decay , since the default must remain
+    purge:ratio until the 5.0.0 release.  (@jasone)
+  - Add MS Visual Studio 2015 support.  (@rustyx, @yuslepukhin)
+  - Make *allocx() size class overflow behavior defined.  The maximum
+    size class is now less than PTRDIFF_MAX to protect applications against
+    numerical overflow, and all allocation functions are guaranteed to indicate
+    errors rather than potentially crashing if the request size exceeds the
+    maximum size class.  (@jasone)
+  - jeprof:
+    + Add raw heap profile support.  (@jasone)
+    + Add --retain and --exclude for backtrace symbol filtering.  (@jasone)
+
+  Optimizations:
+  - Optimize the fast path to combine various bootstrapping and configuration
+    checks and execute more streamlined code in the common case.  (@interwq)
+  - Use linear scan for small bitmaps (used for small object tracking).  In
+    addition to speeding up bitmap operations on 64-bit systems, this reduces
+    allocator metadata overhead by approximately 0.2%.  (@djwatson)
+  - Separate arena_avail trees, which substantially speeds up run tree
+    operations.  (@djwatson)
+  - Use memoization (boot-time-computed table) for run quantization.  Separate
+    arena_avail trees reduced the importance of this optimization.  (@jasone)
+  - Attempt mmap-based in-place huge reallocation.  This can dramatically speed
+    up incremental huge reallocation.  (@jasone)
+
+  Incompatible changes:
+  - Make opt.narenas unsigned rather than size_t.  (@jasone)
+
+  Bug fixes:
+  - Fix stats.cactive accounting regression.  (@rustyx, @jasone)
+  - Handle unaligned keys in hash().  This caused problems for some ARM systems.
+    (@jasone, @cferris1000)
+  - Refactor arenas array.  In addition to fixing a fork-related deadlock, this
+    makes arena lookups faster and simpler.  (@jasone)
+  - Move retained memory allocation out of the default chunk allocation
+    function, to a location that gets executed even if the application installs
+    a custom chunk allocation function.  This resolves a virtual memory leak.
+    (@buchgr)
+  - Fix a potential tsd cleanup leak.  (@cferris1000, @jasone)
+  - Fix run quantization.  In practice this bug had no impact unless
+    applications requested memory with alignment exceeding one page.
+    (@jasone, @djwatson)
+  - Fix LinuxThreads-specific bootstrapping deadlock.  (Cosmin Paraschiv)
+  - jeprof:
+    + Don't discard curl options if timeout is not defined.  (@djwatson)
+    + Detect failed profile fetches.  (@djwatson)
+  - Fix stats.arenas.<i>.{dss,lg_dirty_mult,decay_time,pactive,pdirty} for
+    --disable-stats case.  (@jasone)
+
+* 4.0.4 (October 24, 2015)
+
+  This bugfix release fixes another xallocx() regression.  No other regressions
+  have come to light in over a month, so this is likely a good starting point
+  for people who prefer to wait for "dot one" releases with all the major issues
+  shaken out.
+
+  Bug fixes:
+  - Fix xallocx(..., MALLOCX_ZERO to zero the last full trailing page of large
+    allocations that have been randomly assigned an offset of 0 when
+    --enable-cache-oblivious configure option is enabled.
+
+* 4.0.3 (September 24, 2015)
+
+  This bugfix release continues the trend of xallocx() and heap profiling fixes.
+
+  Bug fixes:
+  - Fix xallocx(..., MALLOCX_ZERO) to zero all trailing bytes of large
+    allocations when --enable-cache-oblivious configure option is enabled.
+  - Fix xallocx(..., MALLOCX_ZERO) to zero trailing bytes of huge allocations
+    when resizing from/to a size class that is not a multiple of the chunk size.
+  - Fix prof_tctx_dump_iter() to filter out nodes that were created after heap
+    profile dumping started.
+  - Work around a potentially bad thread-specific data initialization
+    interaction with NPTL (glibc's pthreads implementation).
+
+* 4.0.2 (September 21, 2015)
+
+  This bugfix release addresses a few bugs specific to heap profiling.
+
+  Bug fixes:
+  - Fix ixallocx_prof_sample() to never modify nor create sampled small
+    allocations.  xallocx() is in general incapable of moving small allocations,
+    so this fix removes buggy code without loss of generality.
+  - Fix irallocx_prof_sample() to always allocate large regions, even when
+    alignment is non-zero.
+  - Fix prof_alloc_rollback() to read tdata from thread-specific data rather
+    than dereferencing a potentially invalid tctx.
+
+* 4.0.1 (September 15, 2015)
+
+  This is a bugfix release that is somewhat high risk due to the amount of
+  refactoring required to address deep xallocx() problems.  As a side effect of
+  these fixes, xallocx() now tries harder to partially fulfill requests for
+  optional extra space.  Note that a couple of minor heap profiling
+  optimizations are included, but these are better thought of as performance
+  fixes that were integral to discovering most of the other bugs.
+
+  Optimizations:
+  - Avoid a chunk metadata read in arena_prof_tctx_set(), since it is in the
+    fast path when heap profiling is enabled.  Additionally, split a special
+    case out into arena_prof_tctx_reset(), which also avoids chunk metadata
+    reads.
+  - Optimize irallocx_prof() to optimistically update the sampler state.  The
+    prior implementation appears to have been a holdover from when
+    rallocx()/xallocx() functionality was combined as rallocm().
+
+  Bug fixes:
+  - Fix TLS configuration such that it is enabled by default for platforms on
+    which it works correctly.
+  - Fix arenas_cache_cleanup() and arena_get_hard() to handle
+    allocation/deallocation within the application's thread-specific data
+    cleanup functions even after arenas_cache is torn down.
+  - Fix xallocx() bugs related to size+extra exceeding HUGE_MAXCLASS.
+  - Fix chunk purge hook calls for in-place huge shrinking reallocation to
+    specify the old chunk size rather than the new chunk size.  This bug caused
+    no correctness issues for the default chunk purge function, but was
+    visible to custom functions set via the "arena.<i>.chunk_hooks" mallctl.
+  - Fix heap profiling bugs:
+    + Fix heap profiling to distinguish among otherwise identical sample sites
+      with interposed resets (triggered via the "prof.reset" mallctl).  This bug
+      could cause data structure corruption that would most likely result in a
+      segfault.
+    + Fix irealloc_prof() to prof_alloc_rollback() on OOM.
+    + Make one call to prof_active_get_unlocked() per allocation event, and use
+      the result throughout the relevant functions that handle an allocation
+      event.  Also add a missing check in prof_realloc().  These fixes protect
+      allocation events against concurrent prof_active changes.
+    + Fix ixallocx_prof() to pass usize_max and zero to ixallocx_prof_sample()
+      in the correct order.
+    + Fix prof_realloc() to call prof_free_sampled_object() after calling
+      prof_malloc_sample_object().  Prior to this fix, if tctx and old_tctx were
+      the same, the tctx could have been prematurely destroyed.
+  - Fix portability bugs:
+    + Don't bitshift by negative amounts when encoding/decoding run sizes in
+      chunk header maps.  This affected systems with page sizes greater than 8
+      KiB.
+    + Rename index_t to szind_t to avoid an existing type on Solaris.
+    + Add JEMALLOC_CXX_THROW to the memalign() function prototype, in order to
+      match glibc and avoid compilation errors when including both
+      jemalloc/jemalloc.h and malloc.h in C++ code.
+    + Don't assume that /bin/sh is appropriate when running size_classes.sh
+      during configuration.
+    + Consider __sparcv9 a synonym for __sparc64__ when defining LG_QUANTUM.
+    + Link tests to librt if it contains clock_gettime(2).
+
+* 4.0.0 (August 17, 2015)
+
+  This version contains many speed and space optimizations, both minor and
+  major.  The major themes are generalization, unification, and simplification.
+  Although many of these optimizations cause no visible behavior change, their
+  cumulative effect is substantial.
+
+  New features:
+  - Normalize size class spacing to be consistent across the complete size
+    range.  By default there are four size classes per size doubling, but this
+    is now configurable via the --with-lg-size-class-group option.  Also add the
+    --with-lg-page, --with-lg-page-sizes, --with-lg-quantum, and
+    --with-lg-tiny-min options, which can be used to tweak page and size class
+    settings.  Impacts:
+    + Worst case performance for incrementally growing/shrinking reallocation
+      is improved because there are far fewer size classes, and therefore
+      copying happens less often.
+    + Internal fragmentation is limited to 20% for all but the smallest size
+      classes (those less than four times the quantum).  (1B + 4 KiB)
+      and (1B + 4 MiB) previously suffered nearly 50% internal fragmentation.
+    + Chunk fragmentation tends to be lower because there are fewer distinct run
+      sizes to pack.
+  - Add support for explicit tcaches.  The "tcache.create", "tcache.flush", and
+    "tcache.destroy" mallctls control tcache lifetime and flushing, and the
+    MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to the *allocx() API
+    control which tcache is used for each operation.
+  - Implement per thread heap profiling, as well as the ability to
+    enable/disable heap profiling on a per thread basis.  Add the "prof.reset",
+    "prof.lg_sample", "thread.prof.name", "thread.prof.active",
+    "opt.prof_thread_active_init", "prof.thread_active_init", and
+    "thread.prof.active" mallctls.
+  - Add support for per arena application-specified chunk allocators, configured
+    via the "arena.<i>.chunk_hooks" mallctl.
+  - Refactor huge allocation to be managed by arenas, so that arenas now
+    function as general purpose independent allocators.  This is important in
+    the context of user-specified chunk allocators, aside from the scalability
+    benefits.  Related new statistics:
+    + The "stats.arenas.<i>.huge.allocated", "stats.arenas.<i>.huge.nmalloc",
+      "stats.arenas.<i>.huge.ndalloc", and "stats.arenas.<i>.huge.nrequests"
+      mallctls provide high level per arena huge allocation statistics.
+    + The "arenas.nhchunks", "arenas.hchunk.<i>.size",
+      "stats.arenas.<i>.hchunks.<j>.nmalloc",
+      "stats.arenas.<i>.hchunks.<j>.ndalloc",
+      "stats.arenas.<i>.hchunks.<j>.nrequests", and
+      "stats.arenas.<i>.hchunks.<j>.curhchunks" mallctls provide per size class
+      statistics.
+  - Add the 'util' column to malloc_stats_print() output, which reports the
+    proportion of available regions that are currently in use for each small
+    size class.
+  - Add "alloc" and "free" modes for for junk filling (see the "opt.junk"
+    mallctl), so that it is possible to separately enable junk filling for
+    allocation versus deallocation.
+  - Add the jemalloc-config script, which provides information about how
+    jemalloc was configured, and how to integrate it into application builds.
+  - Add metadata statistics, which are accessible via the "stats.metadata",
+    "stats.arenas.<i>.metadata.mapped", and
+    "stats.arenas.<i>.metadata.allocated" mallctls.
+  - Add the "stats.resident" mallctl, which reports the upper limit of
+    physically resident memory mapped by the allocator.
+  - Add per arena control over unused dirty page purging, via the
+    "arenas.lg_dirty_mult", "arena.<i>.lg_dirty_mult", and
+    "stats.arenas.<i>.lg_dirty_mult" mallctls.
+  - Add the "prof.gdump" mallctl, which makes it possible to toggle the gdump
+    feature on/off during program execution.
+  - Add sdallocx(), which implements sized deallocation.  The primary
+    optimization over dallocx() is the removal of a metadata read, which often
+    suffers an L1 cache miss.
+  - Add missing header includes in jemalloc/jemalloc.h, so that applications
+    only have to #include <jemalloc/jemalloc.h>.
+  - Add support for additional platforms:
+    + Bitrig
+    + Cygwin
+    + DragonFlyBSD
+    + iOS
+    + OpenBSD
+    + OpenRISC/or1k
+
+  Optimizations:
+  - Maintain dirty runs in per arena LRUs rather than in per arena trees of
+    dirty-run-containing chunks.  In practice this change significantly reduces
+    dirty page purging volume.
+  - Integrate whole chunks into the unused dirty page purging machinery.  This
+    reduces the cost of repeated huge allocation/deallocation, because it
+    effectively introduces a cache of chunks.
+  - Split the arena chunk map into two separate arrays, in order to increase
+    cache locality for the frequently accessed bits.
+  - Move small run metadata out of runs, into arena chunk headers.  This reduces
+    run fragmentation, smaller runs reduce external fragmentation for small size
+    classes, and packed (less uniformly aligned) metadata layout improves CPU
+    cache set distribution.
+  - Randomly distribute large allocation base pointer alignment relative to page
+    boundaries in order to more uniformly utilize CPU cache sets.  This can be
+    disabled via the --disable-cache-oblivious configure option, and queried via
+    the "config.cache_oblivious" mallctl.
+  - Micro-optimize the fast paths for the public API functions.
+  - Refactor thread-specific data to reside in a single structure.  This assures
+    that only a single TLS read is necessary per call into the public API.
+  - Implement in-place huge allocation growing and shrinking.
+  - Refactor rtree (radix tree for chunk lookups) to be lock-free, and make
+    additional optimizations that reduce maximum lookup depth to one or two
+    levels.  This resolves what was a concurrency bottleneck for per arena huge
+    allocation, because a global data structure is critical for determining
+    which arenas own which huge allocations.
+
+  Incompatible changes:
+  - Replace --enable-cc-silence with --disable-cc-silence to suppress spurious
+    warnings by default.
+  - Assure that the constness of malloc_usable_size()'s return type matches that
+    of the system implementation.
+  - Change the heap profile dump format to support per thread heap profiling,
+    rename pprof to jeprof, and enhance it with the --thread=<n> option.  As a
+    result, the bundled jeprof must now be used rather than the upstream
+    (gperftools) pprof.
+  - Disable "opt.prof_final" by default, in order to avoid atexit(3), which can
+    internally deadlock on some platforms.
+  - Change the "arenas.nlruns" mallctl type from size_t to unsigned.
+  - Replace the "stats.arenas.<i>.bins.<j>.allocated" mallctl with
+    "stats.arenas.<i>.bins.<j>.curregs".
+  - Ignore MALLOC_CONF in set{uid,gid,cap} binaries.
+  - Ignore MALLOCX_ARENA(a) in dallocx(), in favor of using the
+    MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to control tcache usage.
+
+  Removed features:
+  - Remove the *allocm() API, which is superseded by the *allocx() API.
+  - Remove the --enable-dss options, and make dss non-optional on all platforms
+    which support sbrk(2).
+  - Remove the "arenas.purge" mallctl, which was obsoleted by the
+    "arena.<i>.purge" mallctl in 3.1.0.
+  - Remove the unnecessary "opt.valgrind" mallctl; jemalloc automatically
+    detects whether it is running inside Valgrind.
+  - Remove the "stats.huge.allocated", "stats.huge.nmalloc", and
+    "stats.huge.ndalloc" mallctls.
+  - Remove the --enable-mremap option.
+  - Remove the "stats.chunks.current", "stats.chunks.total", and
+    "stats.chunks.high" mallctls.
+
+  Bug fixes:
+  - Fix the cactive statistic to decrease (rather than increase) when active
+    memory decreases.  This regression was first released in 3.5.0.
+  - Fix OOM handling in memalign() and valloc().  A variant of this bug existed
+    in all releases since 2.0.0, which introduced these functions.
+  - Fix an OOM-related regression in arena_tcache_fill_small(), which could
+    cause cache corruption on OOM.  This regression was present in all releases
+    from 2.2.0 through 3.6.0.
+  - Fix size class overflow handling for malloc(), posix_memalign(), memalign(),
+    calloc(), and realloc() when profiling is enabled.
+  - Fix the "arena.<i>.dss" mallctl to return an error if "primary" or
+    "secondary" precedence is specified, but sbrk(2) is not supported.
+  - Fix fallback lg_floor() implementations to handle extremely large inputs.
+  - Ensure the default purgeable zone is after the default zone on OS X.
+  - Fix latent bugs in atomic_*().
+  - Fix the "arena.<i>.dss" mallctl to handle read-only calls.
+  - Fix tls_model configuration to enable the initial-exec model when possible.
+  - Mark malloc_conf as a weak symbol so that the application can override it.
+  - Correctly detect glibc's adaptive pthread mutexes.
+  - Fix the --without-export configure option.
+
+* 3.6.0 (March 31, 2014)
+
+  This version contains a critical bug fix for a regression present in 3.5.0 and
+  3.5.1.
+
+  Bug fixes:
+  - Fix a regression in arena_chunk_alloc() that caused crashes during
+    small/large allocation if chunk allocation failed.  In the absence of this
+    bug, chunk allocation failure would result in allocation failure, e.g.  NULL
+    return from malloc().  This regression was introduced in 3.5.0.
+  - Fix backtracing for gcc intrinsics-based backtracing by specifying
+    -fno-omit-frame-pointer to gcc.  Note that the application (and all the
+    libraries it links to) must also be compiled with this option for
+    backtracing to be reliable.
+  - Use dss allocation precedence for huge allocations as well as small/large
+    allocations.
+  - Fix test assertion failure message formatting.  This bug did not manifest on
+    x86_64 systems because of implementation subtleties in va_list.
+  - Fix inconsequential test failures for hash and SFMT code.
+
+  New features:
+  - Support heap profiling on FreeBSD.  This feature depends on the proc
+    filesystem being mounted during heap profile dumping.
+
+* 3.5.1 (February 25, 2014)
+
+  This version primarily addresses minor bugs in test code.
+
+  Bug fixes:
+  - Configure Solaris/Illumos to use MADV_FREE.
+  - Fix junk filling for mremap(2)-based huge reallocation.  This is only
+    relevant if configuring with the --enable-mremap option specified.
+  - Avoid compilation failure if 'restrict' C99 keyword is not supported by the
+    compiler.
+  - Add a configure test for SSE2 rather than assuming it is usable on i686
+    systems.  This fixes test compilation errors, especially on 32-bit Linux
+    systems.
+  - Fix mallctl argument size mismatches (size_t vs. uint64_t) in the stats unit
+    test.
+  - Fix/remove flawed alignment-related overflow tests.
+  - Prevent compiler optimizations that could change backtraces in the
+    prof_accum unit test.
+
+* 3.5.0 (January 22, 2014)
+
+  This version focuses on refactoring and automated testing, though it also
+  includes some non-trivial heap profiling optimizations not mentioned below.
+
+  New features:
+  - Add the *allocx() API, which is a successor to the experimental *allocm()
+    API.  The *allocx() functions are slightly simpler to use because they have
+    fewer parameters, they directly return the results of primary interest, and
+    mallocx()/rallocx() avoid the strict aliasing pitfall that
+    allocm()/rallocm() share with posix_memalign().  Note that *allocm() is
+    slated for removal in the next non-bugfix release.
+  - Add support for LinuxThreads.
+
+  Bug fixes:
+  - Unless heap profiling is enabled, disable floating point code and don't link
+    with libm.  This, in combination with e.g. EXTRA_CFLAGS=-mno-sse on x64
+    systems, makes it possible to completely disable floating point register
+    use.  Some versions of glibc neglect to save/restore caller-saved floating
+    point registers during dynamic lazy symbol loading, and the symbol loading
+    code uses whatever malloc the application happens to have linked/loaded
+    with, the result being potential floating point register corruption.
+  - Report ENOMEM rather than EINVAL if an OOM occurs during heap profiling
+    backtrace creation in imemalign().  This bug impacted posix_memalign() and
+    aligned_alloc().
+  - Fix a file descriptor leak in a prof_dump_maps() error path.
+  - Fix prof_dump() to close the dump file descriptor for all relevant error
+    paths.
+  - Fix rallocm() to use the arena specified by the ALLOCM_ARENA(s) flag for
+    allocation, not just deallocation.
+  - Fix a data race for large allocation stats counters.
+  - Fix a potential infinite loop during thread exit.  This bug occurred on
+    Solaris, and could affect other platforms with similar pthreads TSD
+    implementations.
+  - Don't junk-fill reallocations unless usable size changes.  This fixes a
+    violation of the *allocx()/*allocm() semantics.
+  - Fix growing large reallocation to junk fill new space.
+  - Fix huge deallocation to junk fill when munmap is disabled.
+  - Change the default private namespace prefix from empty to je_, and change
+    --with-private-namespace-prefix so that it prepends an additional prefix
+    rather than replacing je_.  This reduces the likelihood of applications
+    which statically link jemalloc experiencing symbol name collisions.
+  - Add missing private namespace mangling (relevant when
+    --with-private-namespace is specified).
+  - Add and use JEMALLOC_INLINE_C so that static inline functions are marked as
+    static even for debug builds.
+  - Add a missing mutex unlock in a malloc_init_hard() error path.  In practice
+    this error path is never executed.
+  - Fix numerous bugs in malloc_strotumax() error handling/reporting.  These
+    bugs had no impact except for malformed inputs.
+  - Fix numerous bugs in malloc_snprintf().  These bugs were not exercised by
+    existing calls, so they had no impact.
+
+* 3.4.1 (October 20, 2013)
+
+  Bug fixes:
+  - Fix a race in the "arenas.extend" mallctl that could cause memory corruption
+    of internal data structures and subsequent crashes.
+  - Fix Valgrind integration flaws that caused Valgrind warnings about reads of
+    uninitialized memory in:
+    + arena chunk headers
+    + internal zero-initialized data structures (relevant to tcache and prof
+      code)
+  - Preserve errno during the first allocation.  A readlink(2) call during
+    initialization fails unless /etc/malloc.conf exists, so errno was typically
+    set during the first allocation prior to this fix.
+  - Fix compilation warnings reported by gcc 4.8.1.
+
+* 3.4.0 (June 2, 2013)
+
+  This version is essentially a small bugfix release, but the addition of
+  aarch64 support requires that the minor version be incremented.
+
+  Bug fixes:
+  - Fix race-triggered deadlocks in chunk_record().  These deadlocks were
+    typically triggered by multiple threads concurrently deallocating huge
+    objects.
+
+  New features:
+  - Add support for the aarch64 architecture.
+
+* 3.3.1 (March 6, 2013)
+
+  This version fixes bugs that are typically encountered only when utilizing
+  custom run-time options.
+
+  Bug fixes:
+  - Fix a locking order bug that could cause deadlock during fork if heap
+    profiling were enabled.
+  - Fix a chunk recycling bug that could cause the allocator to lose track of
+    whether a chunk was zeroed.  On FreeBSD, NetBSD, and OS X, it could cause
+    corruption if allocating via sbrk(2) (unlikely unless running with the
+    "dss:primary" option specified).  This was completely harmless on Linux
+    unless using mlockall(2) (and unlikely even then, unless the
+    --disable-munmap configure option or the "dss:primary" option was
+    specified).  This regression was introduced in 3.1.0 by the
+    mlockall(2)/madvise(2) interaction fix.
+  - Fix TLS-related memory corruption that could occur during thread exit if the
+    thread never allocated memory.  Only the quarantine and prof facilities were
+    susceptible.
+  - Fix two quarantine bugs:
+    + Internal reallocation of the quarantined object array leaked the old
+      array.
+    + Reallocation failure for internal reallocation of the quarantined object
+      array (very unlikely) resulted in memory corruption.
+  - Fix Valgrind integration to annotate all internally allocated memory in a
+    way that keeps Valgrind happy about internal data structure access.
+  - Fix building for s390 systems.
+
+* 3.3.0 (January 23, 2013)
+
+  This version includes a few minor performance improvements in addition to the
+  listed new features and bug fixes.
+
+  New features:
+  - Add clipping support to lg_chunk option processing.
+  - Add the --enable-ivsalloc option.
+  - Add the --without-export option.
+  - Add the --disable-zone-allocator option.
+
+  Bug fixes:
+  - Fix "arenas.extend" mallctl to output the number of arenas.
+  - Fix chunk_recycle() to unconditionally inform Valgrind that returned memory
+    is undefined.
+  - Fix build break on FreeBSD related to alloca.h.
+
+* 3.2.0 (November 9, 2012)
+
+  In addition to a couple of bug fixes, this version modifies page run
+  allocation and dirty page purging algorithms in order to better control
+  page-level virtual memory fragmentation.
+
+  Incompatible changes:
+  - Change the "opt.lg_dirty_mult" default from 5 to 3 (32:1 to 8:1).
+
+  Bug fixes:
+  - Fix dss/mmap allocation precedence code to use recyclable mmap memory only
+    after primary dss allocation fails.
+  - Fix deadlock in the "arenas.purge" mallctl.  This regression was introduced
+    in 3.1.0 by the addition of the "arena.<i>.purge" mallctl.
+
+* 3.1.0 (October 16, 2012)
+
+  New features:
+  - Auto-detect whether running inside Valgrind, thus removing the need to
+    manually specify MALLOC_CONF=valgrind:true.
+  - Add the "arenas.extend" mallctl, which allows applications to create
+    manually managed arenas.
+  - Add the ALLOCM_ARENA() flag for {,r,d}allocm().
+  - Add the "opt.dss", "arena.<i>.dss", and "stats.arenas.<i>.dss" mallctls,
+    which provide control over dss/mmap precedence.
+  - Add the "arena.<i>.purge" mallctl, which obsoletes "arenas.purge".
+  - Define LG_QUANTUM for hppa.
+
+  Incompatible changes:
+  - Disable tcache by default if running inside Valgrind, in order to avoid
+    making unallocated objects appear reachable to Valgrind.
+  - Drop const from malloc_usable_size() argument on Linux.
+
+  Bug fixes:
+  - Fix heap profiling crash if sampled object is freed via realloc(p, 0).
+  - Remove const from __*_hook variable declarations, so that glibc can modify
+    them during process forking.
+  - Fix mlockall(2)/madvise(2) interaction.
+  - Fix fork(2)-related deadlocks.
+  - Fix error return value for "thread.tcache.enabled" mallctl.
+
+* 3.0.0 (May 11, 2012)
+
+  Although this version adds some major new features, the primary focus is on
+  internal code cleanup that facilitates maintainability and portability, most
+  of which is not reflected in the ChangeLog.  This is the first release to
+  incorporate substantial contributions from numerous other developers, and the
+  result is a more broadly useful allocator (see the git revision history for
+  contribution details).  Note that the license has been unified, thanks to
+  Facebook granting a license under the same terms as the other copyright
+  holders (see COPYING).
+
+  New features:
+  - Implement Valgrind support, redzones, and quarantine.
+  - Add support for additional platforms:
+    + FreeBSD
+    + Mac OS X Lion
+    + MinGW
+    + Windows (no support yet for replacing the system malloc)
+  - Add support for additional architectures:
+    + MIPS
+    + SH4
+    + Tilera
+  - Add support for cross compiling.
+  - Add nallocm(), which rounds a request size up to the nearest size class
+    without actually allocating.
+  - Implement aligned_alloc() (blame C11).
+  - Add the "thread.tcache.enabled" mallctl.
+  - Add the "opt.prof_final" mallctl.
+  - Update pprof (from gperftools 2.0).
+  - Add the --with-mangling option.
+  - Add the --disable-experimental option.
+  - Add the --disable-munmap option, and make it the default on Linux.
+  - Add the --enable-mremap option, which disables use of mremap(2) by default.
+
+  Incompatible changes:
+  - Enable stats by default.
+  - Enable fill by default.
+  - Disable lazy locking by default.
+  - Rename the "tcache.flush" mallctl to "thread.tcache.flush".
+  - Rename the "arenas.pagesize" mallctl to "arenas.page".
+  - Change the "opt.lg_prof_sample" default from 0 to 19 (1 B to 512 KiB).
+  - Change the "opt.prof_accum" default from true to false.
+
+  Removed features:
+  - Remove the swap feature, including the "config.swap", "swap.avail",
+    "swap.prezeroed", "swap.nfds", and "swap.fds" mallctls.
+  - Remove highruns statistics, including the
+    "stats.arenas.<i>.bins.<j>.highruns" and
+    "stats.arenas.<i>.lruns.<j>.highruns" mallctls.
+  - As part of small size class refactoring, remove the "opt.lg_[qc]space_max",
+    "arenas.cacheline", "arenas.subpage", "arenas.[tqcs]space_{min,max}", and
+    "arenas.[tqcs]bins" mallctls.
+  - Remove the "arenas.chunksize" mallctl.
+  - Remove the "opt.lg_prof_tcmax" option.
+  - Remove the "opt.lg_prof_bt_max" option.
+  - Remove the "opt.lg_tcache_gc_sweep" option.
+  - Remove the --disable-tiny option, including the "config.tiny" mallctl.
+  - Remove the --enable-dynamic-page-shift configure option.
+  - Remove the --enable-sysv configure option.
+
+  Bug fixes:
+  - Fix a statistics-related bug in the "thread.arena" mallctl that could cause
+    invalid statistics and crashes.
+  - Work around TLS deallocation via free() on Linux.  This bug could cause
+    write-after-free memory corruption.
+  - Fix a potential deadlock that could occur during interval- and
+    growth-triggered heap profile dumps.
+  - Fix large calloc() zeroing bugs due to dropping chunk map unzeroed flags.
+  - Fix chunk_alloc_dss() to stop claiming memory is zeroed.  This bug could
+    cause memory corruption and crashes with --enable-dss specified.
+  - Fix fork-related bugs that could cause deadlock in children between fork
+    and exec.
+  - Fix malloc_stats_print() to honor 'b' and 'l' in the opts parameter.
+  - Fix realloc(p, 0) to act like free(p).
+  - Do not enforce minimum alignment in memalign().
+  - Check for NULL pointer in malloc_usable_size().
+  - Fix an off-by-one heap profile statistics bug that could be observed in
+    interval- and growth-triggered heap profiles.
+  - Fix the "epoch" mallctl to update cached stats even if the passed in epoch
+    is 0.
+  - Fix bin->runcur management to fix a layout policy bug.  This bug did not
+    affect correctness.
+  - Fix a bug in choose_arena_hard() that potentially caused more arenas to be
+    initialized than necessary.
+  - Add missing "opt.lg_tcache_max" mallctl implementation.
+  - Use glibc allocator hooks to make mixed allocator usage less likely.
+  - Fix build issues for --disable-tcache.
+  - Don't mangle pthread_create() when --with-private-namespace is specified.
+
+* 2.2.5 (November 14, 2011)
+
+  Bug fixes:
+  - Fix huge_ralloc() race when using mremap(2).  This is a serious bug that
+    could cause memory corruption and/or crashes.
+  - Fix huge_ralloc() to maintain chunk statistics.
+  - Fix malloc_stats_print(..., "a") output.
+
+* 2.2.4 (November 5, 2011)
+
+  Bug fixes:
+  - Initialize arenas_tsd before using it.  This bug existed for 2.2.[0-3], as
+    well as for --disable-tls builds in earlier releases.
+  - Do not assume a 4 KiB page size in test/rallocm.c.
+
+* 2.2.3 (August 31, 2011)
+
+  This version fixes numerous bugs related to heap profiling.
+
+  Bug fixes:
+  - Fix a prof-related race condition.  This bug could cause memory corruption,
+    but only occurred in non-default configurations (prof_accum:false).
+  - Fix off-by-one backtracing issues (make sure that prof_alloc_prep() is
+    excluded from backtraces).
+  - Fix a prof-related bug in realloc() (only triggered by OOM errors).
+  - Fix prof-related bugs in allocm() and rallocm().
+  - Fix prof_tdata_cleanup() for --disable-tls builds.
+  - Fix a relative include path, to fix objdir builds.
+
+* 2.2.2 (July 30, 2011)
+
+  Bug fixes:
+  - Fix a build error for --disable-tcache.
+  - Fix assertions in arena_purge() (for real this time).
+  - Add the --with-private-namespace option.  This is a workaround for symbol
+    conflicts that can inadvertently arise when using static libraries.
+
+* 2.2.1 (March 30, 2011)
+
+  Bug fixes:
+  - Implement atomic operations for x86/x64.  This fixes compilation failures
+    for versions of gcc that are still in wide use.
+  - Fix an assertion in arena_purge().
+
+* 2.2.0 (March 22, 2011)
+
+  This version incorporates several improvements to algorithms and data
+  structures that tend to reduce fragmentation and increase speed.
+
+  New features:
+  - Add the "stats.cactive" mallctl.
+  - Update pprof (from google-perftools 1.7).
+  - Improve backtracing-related configuration logic, and add the
+    --disable-prof-libgcc option.
+
+  Bug fixes:
+  - Change default symbol visibility from "internal", to "hidden", which
+    decreases the overhead of library-internal function calls.
+  - Fix symbol visibility so that it is also set on OS X.
+  - Fix a build dependency regression caused by the introduction of the .pic.o
+    suffix for PIC object files.
+  - Add missing checks for mutex initialization failures.
+  - Don't use libgcc-based backtracing except on x64, where it is known to work.
+  - Fix deadlocks on OS X that were due to memory allocation in
+    pthread_mutex_lock().
+  - Heap profiling-specific fixes:
+    + Fix memory corruption due to integer overflow in small region index
+      computation, when using a small enough sample interval that profiling
+      context pointers are stored in small run headers.
+    + Fix a bootstrap ordering bug that only occurred with TLS disabled.
+    + Fix a rallocm() rsize bug.
+    + Fix error detection bugs for aligned memory allocation.
+
+* 2.1.3 (March 14, 2011)
+
+  Bug fixes:
+  - Fix a cpp logic regression (due to the "thread.{de,}allocatedp" mallctl fix
+    for OS X in 2.1.2).
+  - Fix a "thread.arena" mallctl bug.
+  - Fix a thread cache stats merging bug.
+
+* 2.1.2 (March 2, 2011)
+
+  Bug fixes:
+  - Fix "thread.{de,}allocatedp" mallctl for OS X.
+  - Add missing jemalloc.a to build system.
+
+* 2.1.1 (January 31, 2011)
+
+  Bug fixes:
+  - Fix aligned huge reallocation (affected allocm()).
+  - Fix the ALLOCM_LG_ALIGN macro definition.
+  - Fix a heap dumping deadlock.
+  - Fix a "thread.arena" mallctl bug.
+
+* 2.1.0 (December 3, 2010)
+
+  This version incorporates some optimizations that can't quite be considered
+  bug fixes.
+
+  New features:
+  - Use Linux's mremap(2) for huge object reallocation when possible.
+  - Avoid locking in mallctl*() when possible.
+  - Add the "thread.[de]allocatedp" mallctl's.
+  - Convert the manual page source from roff to DocBook, and generate both roff
+    and HTML manuals.
+
+  Bug fixes:
+  - Fix a crash due to incorrect bootstrap ordering.  This only impacted
+    --enable-debug --enable-dss configurations.
+  - Fix a minor statistics bug for mallctl("swap.avail", ...).
+
+* 2.0.1 (October 29, 2010)
+
+  Bug fixes:
+  - Fix a race condition in heap profiling that could cause undefined behavior
+    if "opt.prof_accum" were disabled.
+  - Add missing mutex unlocks for some OOM error paths in the heap profiling
+    code.
+  - Fix a compilation error for non-C99 builds.
+
+* 2.0.0 (October 24, 2010)
+
+  This version focuses on the experimental *allocm() API, and on improved
+  run-time configuration/introspection.  Nonetheless, numerous performance
+  improvements are also included.
+
+  New features:
+  - Implement the experimental {,r,s,d}allocm() API, which provides a superset
+    of the functionality available via malloc(), calloc(), posix_memalign(),
+    realloc(), malloc_usable_size(), and free().  These functions can be used to
+    allocate/reallocate aligned zeroed memory, ask for optional extra memory
+    during reallocation, prevent object movement during reallocation, etc.
+  - Replace JEMALLOC_OPTIONS/JEMALLOC_PROF_PREFIX with MALLOC_CONF, which is
+    more human-readable, and more flexible.  For example:
+      JEMALLOC_OPTIONS=AJP
+    is now:
+      MALLOC_CONF=abort:true,fill:true,stats_print:true
+  - Port to Apple OS X.  Sponsored by Mozilla.
+  - Make it possible for the application to control thread-->arena mappings via
+    the "thread.arena" mallctl.
+  - Add compile-time support for all TLS-related functionality via pthreads TSD.
+    This is mainly of interest for OS X, which does not support TLS, but has a
+    TSD implementation with similar performance.
+  - Override memalign() and valloc() if they are provided by the system.
+  - Add the "arenas.purge" mallctl, which can be used to synchronously purge all
+    dirty unused pages.
+  - Make cumulative heap profiling data optional, so that it is possible to
+    limit the amount of memory consumed by heap profiling data structures.
+  - Add per thread allocation counters that can be accessed via the
+    "thread.allocated" and "thread.deallocated" mallctls.
+
+  Incompatible changes:
+  - Remove JEMALLOC_OPTIONS and malloc_options (see MALLOC_CONF above).
+  - Increase default backtrace depth from 4 to 128 for heap profiling.
+  - Disable interval-based profile dumps by default.
+
+  Bug fixes:
+  - Remove bad assertions in fork handler functions.  These assertions could
+    cause aborts for some combinations of configure settings.
+  - Fix strerror_r() usage to deal with non-standard semantics in GNU libc.
+  - Fix leak context reporting.  This bug tended to cause the number of contexts
+    to be underreported (though the reported number of objects and bytes were
+    correct).
+  - Fix a realloc() bug for large in-place growing reallocation.  This bug could
+    cause memory corruption, but it was hard to trigger.
+  - Fix an allocation bug for small allocations that could be triggered if
+    multiple threads raced to create a new run of backing pages.
+  - Enhance the heap profiler to trigger samples based on usable size, rather
+    than request size.
+  - Fix a heap profiling bug due to sometimes losing track of requested object
+    size for sampled objects.
+
+* 1.0.3 (August 12, 2010)
+
+  Bug fixes:
+  - Fix the libunwind-based implementation of stack backtracing (used for heap
+    profiling).  This bug could cause zero-length backtraces to be reported.
+  - Add a missing mutex unlock in library initialization code.  If multiple
+    threads raced to initialize malloc, some of them could end up permanently
+    blocked.
+
+* 1.0.2 (May 11, 2010)
+
+  Bug fixes:
+  - Fix junk filling of large objects, which could cause memory corruption.
+  - Add MAP_NORESERVE support for chunk mapping, because otherwise virtual
+    memory limits could cause swap file configuration to fail.  Contributed by
+    Jordan DeLong.
+
+* 1.0.1 (April 14, 2010)
+
+  Bug fixes:
+  - Fix compilation when --enable-fill is specified.
+  - Fix threads-related profiling bugs that affected accuracy and caused memory
+    to be leaked during thread exit.
+  - Fix dirty page purging race conditions that could cause crashes.
+  - Fix crash in tcache flushing code during thread destruction.
+
+* 1.0.0 (April 11, 2010)
+
+  This release focuses on speed and run-time introspection.  Numerous
+  algorithmic improvements make this release substantially faster than its
+  predecessors.
+
+  New features:
+  - Implement autoconf-based configuration system.
+  - Add mallctl*(), for the purposes of introspection and run-time
+    configuration.
+  - Make it possible for the application to manually flush a thread's cache, via
+    the "tcache.flush" mallctl.
+  - Base maximum dirty page count on proportion of active memory.
+  - Compute various additional run-time statistics, including per size class
+    statistics for large objects.
+  - Expose malloc_stats_print(), which can be called repeatedly by the
+    application.
+  - Simplify the malloc_message() signature to only take one string argument,
+    and incorporate an opaque data pointer argument for use by the application
+    in combination with malloc_stats_print().
+  - Add support for allocation backed by one or more swap files, and allow the
+    application to disable over-commit if swap files are in use.
+  - Implement allocation profiling and leak checking.
+
+  Removed features:
+  - Remove the dynamic arena rebalancing code, since thread-specific caching
+    reduces its utility.
+
+  Bug fixes:
+  - Modify chunk allocation to work when address space layout randomization
+    (ASLR) is in use.
+  - Fix thread cleanup bugs related to TLS destruction.
+  - Handle 0-size allocation requests in posix_memalign().
+  - Fix a chunk leak.  The leaked chunks were never touched, so this impacted
+    virtual memory usage, but not physical memory usage.
+
+* linux_2008082[78]a (August 27/28, 2008)
+
+  These snapshot releases are the simple result of incorporating Linux-specific
+  support into the FreeBSD malloc sources.
+
+--------------------------------------------------------------------------------
+vim:filetype=text:textwidth=80
diff --git a/memkind-1.10.0/jemalloc/INSTALL.md b/memkind-1.10.0/jemalloc/INSTALL.md
new file mode 100644
index 0000000..b8f729b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/INSTALL.md
@@ -0,0 +1,421 @@
+Building and installing a packaged release of jemalloc can be as simple as
+typing the following while in the root directory of the source tree:
+
+    ./configure
+    make
+    make install
+
+If building from unpackaged developer sources, the simplest command sequence
+that might work is:
+
+    ./autogen.sh
+    make dist
+    make
+    make install
+
+Note that documentation is not built by the default target because doing so
+would create a dependency on xsltproc in packaged releases, hence the
+requirement to either run 'make dist' or avoid installing docs via the various
+install_* targets documented below.
+
+
+## Advanced configuration
+
+The 'configure' script supports numerous options that allow control of which
+functionality is enabled, where jemalloc is installed, etc.  Optionally, pass
+any of the following arguments (not a definitive list) to 'configure':
+
+* `--help`
+
+    Print a definitive list of options.
+
+* `--prefix=<install-root-dir>`
+
+    Set the base directory in which to install.  For example:
+
+        ./configure --prefix=/usr/local
+
+    will cause files to be installed into /usr/local/include, /usr/local/lib,
+    and /usr/local/man.
+
+* `--with-version=(<major>.<minor>.<bugfix>-<nrev>-g<gid>|VERSION)`
+
+    The VERSION file is mandatory for successful configuration, and the
+    following steps are taken to assure its presence:
+    1) If --with-version=<major>.<minor>.<bugfix>-<nrev>-g<gid> is specified,
+       generate VERSION using the specified value.
+    2) If --with-version is not specified in either form and the source
+       directory is inside a git repository, try to generate VERSION via 'git
+       describe' invocations that pattern-match release tags.
+    3) If VERSION is missing, generate it with a bogus version:
+       0.0.0-0-g0000000000000000000000000000000000000000
+
+    Note that --with-version=VERSION bypasses (1) and (2), which simplifies
+    VERSION configuration when embedding a jemalloc release into another
+    project's git repository.
+
+* `--with-rpath=<colon-separated-rpath>`
+
+    Embed one or more library paths, so that libjemalloc can find the libraries
+    it is linked to.  This works only on ELF-based systems.
+
+* `--with-mangling=<map>`
+
+    Mangle public symbols specified in <map> which is a comma-separated list of
+    name:mangled pairs.
+
+    For example, to use ld's --wrap option as an alternative method for
+    overriding libc's malloc implementation, specify something like:
+
+      --with-mangling=malloc:__wrap_malloc,free:__wrap_free[...]
+
+    Note that mangling happens prior to application of the prefix specified by
+    --with-jemalloc-prefix, and mangled symbols are then ignored when applying
+    the prefix.
+
+* `--with-jemalloc-prefix=<prefix>`
+
+    Prefix all public APIs with <prefix>.  For example, if <prefix> is
+    "prefix_", API changes like the following occur:
+
+      malloc()         --> prefix_malloc()
+      malloc_conf      --> prefix_malloc_conf
+      /etc/malloc.conf --> /etc/prefix_malloc.conf
+      MALLOC_CONF      --> PREFIX_MALLOC_CONF
+
+    This makes it possible to use jemalloc at the same time as the system
+    allocator, or even to use multiple copies of jemalloc simultaneously.
+
+    By default, the prefix is "", except on OS X, where it is "je_".  On OS X,
+    jemalloc overlays the default malloc zone, but makes no attempt to actually
+    replace the "malloc", "calloc", etc. symbols.
+
+* `--without-export`
+
+    Don't export public APIs.  This can be useful when building jemalloc as a
+    static library, or to avoid exporting public APIs when using the zone
+    allocator on OSX.
+
+* `--with-private-namespace=<prefix>`
+
+    Prefix all library-private APIs with <prefix>je_.  For shared libraries,
+    symbol visibility mechanisms prevent these symbols from being exported, but
+    for static libraries, naming collisions are a real possibility.  By
+    default, <prefix> is empty, which results in a symbol prefix of je_ .
+
+* `--with-install-suffix=<suffix>`
+
+    Append <suffix> to the base name of all installed files, such that multiple
+    versions of jemalloc can coexist in the same installation directory.  For
+    example, libjemalloc.so.0 becomes libjemalloc<suffix>.so.0.
+
+* `--with-malloc-conf=<malloc_conf>`
+
+    Embed `<malloc_conf>` as a run-time options string that is processed prior to
+    the malloc_conf global variable, the /etc/malloc.conf symlink, and the
+    MALLOC_CONF environment variable.  For example, to change the default decay
+    time to 30 seconds:
+
+      --with-malloc-conf=decay_ms:30000
+
+* `--enable-debug`
+
+    Enable assertions and validation code.  This incurs a substantial
+    performance hit, but is very useful during application development.
+
+* `--disable-stats`
+
+    Disable statistics gathering functionality.  See the "opt.stats_print"
+    option documentation for usage details.
+
+* `--enable-prof`
+
+    Enable heap profiling and leak detection functionality.  See the "opt.prof"
+    option documentation for usage details.  When enabled, there are several
+    approaches to backtracing, and the configure script chooses the first one
+    in the following list that appears to function correctly:
+
+    + libunwind      (requires --enable-prof-libunwind)
+    + libgcc         (unless --disable-prof-libgcc)
+    + gcc intrinsics (unless --disable-prof-gcc)
+
+* `--enable-prof-libunwind`
+
+    Use the libunwind library (http://www.nongnu.org/libunwind/) for stack
+    backtracing.
+
+* `--disable-prof-libgcc`
+
+    Disable the use of libgcc's backtracing functionality.
+
+* `--disable-prof-gcc`
+
+    Disable the use of gcc intrinsics for backtracing.
+
+* `--with-static-libunwind=<libunwind.a>`
+
+    Statically link against the specified libunwind.a rather than dynamically
+    linking with -lunwind.
+
+* `--disable-fill`
+
+    Disable support for junk/zero filling of memory.  See the "opt.junk" and
+    "opt.zero" option documentation for usage details.
+
+* `--disable-zone-allocator`
+
+    Disable zone allocator for Darwin.  This means jemalloc won't be hooked as
+    the default allocator on OSX/iOS.
+
+* `--enable-utrace`
+
+    Enable utrace(2)-based allocation tracing.  This feature is not broadly
+    portable (FreeBSD has it, but Linux and OS X do not).
+
+* `--enable-xmalloc`
+
+    Enable support for optional immediate termination due to out-of-memory
+    errors, as is commonly implemented by "xmalloc" wrapper function for malloc.
+    See the "opt.xmalloc" option documentation for usage details.
+
+* `--enable-lazy-lock`
+
+    Enable code that wraps pthread_create() to detect when an application
+    switches from single-threaded to multi-threaded mode, so that it can avoid
+    mutex locking/unlocking operations while in single-threaded mode.  In
+    practice, this feature usually has little impact on performance unless
+    thread-specific caching is disabled.
+
+* `--disable-cache-oblivious`
+
+    Disable cache-oblivious large allocation alignment for large allocation
+    requests with no alignment constraints.  If this feature is disabled, all
+    large allocations are page-aligned as an implementation artifact, which can
+    severely harm CPU cache utilization.  However, the cache-oblivious layout
+    comes at the cost of one extra page per large allocation, which in the
+    most extreme case increases physical memory usage for the 16 KiB size class
+    to 20 KiB.
+
+* `--disable-syscall`
+
+    Disable use of syscall(2) rather than {open,read,write,close}(2).  This is
+    intended as a workaround for systems that place security limitations on
+    syscall(2).
+
+* `--disable-cxx`
+
+    Disable C++ integration.  This will cause new and delete operator
+    implementations to be omitted.
+
+* `--with-xslroot=<path>`
+
+    Specify where to find DocBook XSL stylesheets when building the
+    documentation.
+
+* `--with-lg-page=<lg-page>`
+
+    Specify the base 2 log of the allocator page size, which must in turn be at
+    least as large as the system page size.  By default the configure script
+    determines the host's page size and sets the allocator page size equal to
+    the system page size, so this option need not be specified unless the
+    system page size may change between configuration and execution, e.g. when
+    cross compiling.
+
+* `--with-lg-hugepage=<lg-hugepage>`
+
+    Specify the base 2 log of the system huge page size.  This option is useful
+    when cross compiling, or when overriding the default for systems that do
+    not explicitly support huge pages.
+
+* `--with-lg-quantum=<lg-quantum>`
+
+    Specify the base 2 log of the minimum allocation alignment.  jemalloc needs
+    to know the minimum alignment that meets the following C standard
+    requirement (quoted from the April 12, 2011 draft of the C11 standard):
+
+    >  The pointer returned if the allocation succeeds is suitably aligned so
+      that it may be assigned to a pointer to any type of object with a
+      fundamental alignment requirement and then used to access such an object
+      or an array of such objects in the space allocated [...]
+
+    This setting is architecture-specific, and although jemalloc includes known
+    safe values for the most commonly used modern architectures, there is a
+    wrinkle related to GNU libc (glibc) that may impact your choice of
+    <lg-quantum>.  On most modern architectures, this mandates 16-byte
+    alignment (<lg-quantum>=4), but the glibc developers chose not to meet this
+    requirement for performance reasons.  An old discussion can be found at
+    <https://sourceware.org/bugzilla/show_bug.cgi?id=206> .  Unlike glibc,
+    jemalloc does follow the C standard by default (caveat: jemalloc
+    technically cheats for size classes smaller than the quantum), but the fact
+    that Linux systems already work around this allocator noncompliance means
+    that it is generally safe in practice to let jemalloc's minimum alignment
+    follow glibc's lead.  If you specify `--with-lg-quantum=3` during
+    configuration, jemalloc will provide additional size classes that are not
+    16-byte-aligned (24, 40, and 56).
+
+* `--with-lg-vaddr=<lg-vaddr>`
+
+    Specify the number of significant virtual address bits.  By default, the
+    configure script attempts to detect virtual address size on those platforms
+    where it knows how, and picks a default otherwise.  This option may be
+    useful when cross-compiling.
+
+* `--disable-initial-exec-tls`
+
+    Disable the initial-exec TLS model for jemalloc's internal thread-local
+    storage (on those platforms that support explicit settings).  This can allow
+    jemalloc to be dynamically loaded after program startup (e.g. using dlopen).
+    Note that in this case, there will be two malloc implementations operating
+    in the same process, which will almost certainly result in confusing runtime
+    crashes if pointers leak from one implementation to the other.
+
+* `--disable-libdl`
+
+    Disable the usage of libdl, namely dlsym(3) which is required by the lazy
+    lock option.  This can allow building static binaries.
+
+The following environment variables (not a definitive list) impact configure's
+behavior:
+
+* `CFLAGS="?"`
+* `CXXFLAGS="?"`
+
+    Pass these flags to the C/C++ compiler.  Any flags set by the configure
+    script are prepended, which means explicitly set flags generally take
+    precedence.  Take care when specifying flags such as -Werror, because
+    configure tests may be affected in undesirable ways.
+
+* `EXTRA_CFLAGS="?"`
+* `EXTRA_CXXFLAGS="?"`
+
+    Append these flags to CFLAGS/CXXFLAGS, without passing them to the
+    compiler(s) during configuration.  This makes it possible to add flags such
+    as -Werror, while allowing the configure script to determine what other
+    flags are appropriate for the specified configuration.
+
+* `CPPFLAGS="?"`
+
+    Pass these flags to the C preprocessor.  Note that CFLAGS is not passed to
+    'cpp' when 'configure' is looking for include files, so you must use
+    CPPFLAGS instead if you need to help 'configure' find header files.
+
+* `LD_LIBRARY_PATH="?"`
+
+    'ld' uses this colon-separated list to find libraries.
+
+* `LDFLAGS="?"`
+
+    Pass these flags when linking.
+
+* `PATH="?"`
+
+    'configure' uses this to find programs.
+
+In some cases it may be necessary to work around configuration results that do
+not match reality.  For example, Linux 4.5 added support for the MADV_FREE flag
+to madvise(2), which can cause problems if building on a host with MADV_FREE
+support and deploying to a target without.  To work around this, use a cache
+file to override the relevant configuration variable defined in configure.ac,
+e.g.:
+
+    echo "je_cv_madv_free=no" > config.cache && ./configure -C
+
+
+## Advanced compilation
+
+To build only parts of jemalloc, use the following targets:
+
+    build_lib_shared
+    build_lib_static
+    build_lib
+    build_doc_html
+    build_doc_man
+    build_doc
+
+To install only parts of jemalloc, use the following targets:
+
+    install_bin
+    install_include
+    install_lib_shared
+    install_lib_static
+    install_lib_pc
+    install_lib
+    install_doc_html
+    install_doc_man
+    install_doc
+
+To clean up build results to varying degrees, use the following make targets:
+
+    clean
+    distclean
+    relclean
+
+
+## Advanced installation
+
+Optionally, define make variables when invoking make, including (not
+exclusively):
+
+* `INCLUDEDIR="?"`
+
+    Use this as the installation prefix for header files.
+
+* `LIBDIR="?"`
+
+    Use this as the installation prefix for libraries.
+
+* `MANDIR="?"`
+
+    Use this as the installation prefix for man pages.
+
+* `DESTDIR="?"`
+
+    Prepend DESTDIR to INCLUDEDIR, LIBDIR, DATADIR, and MANDIR.  This is useful
+    when installing to a different path than was specified via --prefix.
+
+* `CC="?"`
+
+    Use this to invoke the C compiler.
+
+* `CFLAGS="?"`
+
+    Pass these flags to the compiler.
+
+* `CPPFLAGS="?"`
+
+    Pass these flags to the C preprocessor.
+
+* `LDFLAGS="?"`
+
+    Pass these flags when linking.
+
+* `PATH="?"`
+
+    Use this to search for programs used during configuration and building.
+
+
+## Development
+
+If you intend to make non-trivial changes to jemalloc, use the 'autogen.sh'
+script rather than 'configure'.  This re-generates 'configure', enables
+configuration dependency rules, and enables re-generation of automatically
+generated source files.
+
+The build system supports using an object directory separate from the source
+tree.  For example, you can create an 'obj' directory, and from within that
+directory, issue configuration and build commands:
+
+    autoconf
+    mkdir obj
+    cd obj
+    ../configure --enable-autogen
+    make
+
+
+## Documentation
+
+The manual page is generated in both html and roff formats.  Any web browser
+can be used to view the html manual.  The roff manual page can be formatted
+prior to installation via the following command:
+
+    nroff -man -t doc/jemalloc.3
diff --git a/memkind-1.10.0/jemalloc/Makefile.in b/memkind-1.10.0/jemalloc/Makefile.in
new file mode 100644
index 0000000..7128b00
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/Makefile.in
@@ -0,0 +1,623 @@
+# Clear out all vpaths, then set just one (default vpath) for the main build
+# directory.
+vpath
+vpath % .
+
+# Clear the default suffixes, so that built-in rules are not used.
+.SUFFIXES :
+
+SHELL := /bin/sh
+
+CC := @CC@
+CXX := @CXX@
+
+# Configuration parameters.
+DESTDIR =
+BINDIR := $(DESTDIR)@BINDIR@
+INCLUDEDIR := $(DESTDIR)@INCLUDEDIR@
+LIBDIR := $(DESTDIR)@LIBDIR@
+DATADIR := $(DESTDIR)@DATADIR@
+MANDIR := $(DESTDIR)@MANDIR@
+srcroot := @srcroot@
+objroot := @objroot@
+abs_srcroot := @abs_srcroot@
+abs_objroot := @abs_objroot@
+
+# Build parameters.
+CPPFLAGS := @CPPFLAGS@ -I$(objroot)include -I$(srcroot)include
+CONFIGURE_CFLAGS := @CONFIGURE_CFLAGS@
+SPECIFIED_CFLAGS := @SPECIFIED_CFLAGS@
+EXTRA_CFLAGS := @EXTRA_CFLAGS@
+CFLAGS := $(strip $(CONFIGURE_CFLAGS) $(SPECIFIED_CFLAGS) $(EXTRA_CFLAGS))
+CONFIGURE_CXXFLAGS := @CONFIGURE_CXXFLAGS@
+SPECIFIED_CXXFLAGS := @SPECIFIED_CXXFLAGS@
+EXTRA_CXXFLAGS := @EXTRA_CXXFLAGS@
+CXXFLAGS := $(strip $(CONFIGURE_CXXFLAGS) $(SPECIFIED_CXXFLAGS) $(EXTRA_CXXFLAGS))
+LDFLAGS := @LDFLAGS@
+EXTRA_LDFLAGS := @EXTRA_LDFLAGS@
+LIBS := @LIBS@
+RPATH_EXTRA := @RPATH_EXTRA@
+SO := @so@
+IMPORTLIB := @importlib@
+O := @o@
+A := @a@
+EXE := @exe@
+LIBPREFIX := @libprefix@
+REV := @rev@
+install_suffix := @install_suffix@
+ABI := @abi@
+XSLTPROC := @XSLTPROC@
+XSLROOT := @XSLROOT@
+AUTOCONF := @AUTOCONF@
+_RPATH = @RPATH@
+RPATH = $(if $(1),$(call _RPATH,$(1)))
+cfghdrs_in := $(addprefix $(srcroot),@cfghdrs_in@)
+cfghdrs_out := @cfghdrs_out@
+cfgoutputs_in := $(addprefix $(srcroot),@cfgoutputs_in@)
+cfgoutputs_out := @cfgoutputs_out@
+enable_autogen := @enable_autogen@
+enable_doc := @enable_doc@
+enable_shared := @enable_shared@
+enable_static := @enable_static@
+enable_prof := @enable_prof@
+enable_zone_allocator := @enable_zone_allocator@
+enable_experimental_smallocx := @enable_experimental_smallocx@
+MALLOC_CONF := @JEMALLOC_CPREFIX@MALLOC_CONF
+link_whole_archive := @link_whole_archive@
+DSO_LDFLAGS = @DSO_LDFLAGS@
+SOREV = @SOREV@
+PIC_CFLAGS = @PIC_CFLAGS@
+CTARGET = @CTARGET@
+LDTARGET = @LDTARGET@
+TEST_LD_MODE = @TEST_LD_MODE@
+MKLIB = @MKLIB@
+AR = @AR@
+ARFLAGS = @ARFLAGS@
+DUMP_SYMS = @DUMP_SYMS@
+AWK := @AWK@
+CC_MM = @CC_MM@
+LM := @LM@
+INSTALL = @INSTALL@
+
+ifeq (macho, $(ABI))
+TEST_LIBRARY_PATH := DYLD_FALLBACK_LIBRARY_PATH="$(objroot)lib"
+else
+ifeq (pecoff, $(ABI))
+TEST_LIBRARY_PATH := PATH="$(PATH):$(objroot)lib"
+else
+TEST_LIBRARY_PATH :=
+endif
+endif
+
+LIBJEMALLOC := $(LIBPREFIX)jemalloc$(install_suffix)
+
+# Lists of files.
+BINS := $(objroot)bin/jemalloc-config $(objroot)bin/jemalloc.sh $(objroot)bin/jeprof
+C_HDRS := $(objroot)include/jemalloc/jemalloc$(install_suffix).h
+C_SRCS := $(srcroot)src/jemalloc.c \
+	$(srcroot)src/arena.c \
+	$(srcroot)src/background_thread.c \
+	$(srcroot)src/base.c \
+	$(srcroot)src/bin.c \
+	$(srcroot)src/bitmap.c \
+	$(srcroot)src/ckh.c \
+	$(srcroot)src/ctl.c \
+	$(srcroot)src/div.c \
+	$(srcroot)src/extent.c \
+	$(srcroot)src/extent_dss.c \
+	$(srcroot)src/extent_mmap.c \
+	$(srcroot)src/hash.c \
+	$(srcroot)src/hook.c \
+	$(srcroot)src/large.c \
+	$(srcroot)src/log.c \
+	$(srcroot)src/malloc_io.c \
+	$(srcroot)src/mutex.c \
+	$(srcroot)src/mutex_pool.c \
+	$(srcroot)src/nstime.c \
+	$(srcroot)src/pages.c \
+	$(srcroot)src/prng.c \
+	$(srcroot)src/prof.c \
+	$(srcroot)src/rtree.c \
+	$(srcroot)src/safety_check.c \
+	$(srcroot)src/stats.c \
+	$(srcroot)src/sc.c \
+	$(srcroot)src/sz.c \
+	$(srcroot)src/tcache.c \
+	$(srcroot)src/test_hooks.c \
+	$(srcroot)src/ticker.c \
+	$(srcroot)src/tsd.c \
+	$(srcroot)src/witness.c
+ifeq ($(enable_zone_allocator), 1)
+C_SRCS += $(srcroot)src/zone.c
+endif
+ifeq ($(IMPORTLIB),$(SO))
+STATIC_LIBS := $(objroot)lib/$(LIBJEMALLOC).$(A)
+endif
+ifdef PIC_CFLAGS
+STATIC_LIBS += $(objroot)lib/$(LIBJEMALLOC)_pic.$(A)
+else
+STATIC_LIBS += $(objroot)lib/$(LIBJEMALLOC)_s.$(A)
+endif
+DSOS := $(objroot)lib/$(LIBJEMALLOC).$(SOREV)
+ifneq ($(SOREV),$(SO))
+DSOS += $(objroot)lib/$(LIBJEMALLOC).$(SO)
+endif
+ifeq (1, $(link_whole_archive))
+LJEMALLOC := -Wl,--whole-archive -L$(objroot)lib -l$(LIBJEMALLOC) -Wl,--no-whole-archive
+else
+LJEMALLOC := $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+endif
+PC := $(objroot)jemalloc.pc
+MAN3 := $(objroot)doc/jemalloc$(install_suffix).3
+DOCS_XML := $(objroot)doc/jemalloc$(install_suffix).xml
+DOCS_HTML := $(DOCS_XML:$(objroot)%.xml=$(objroot)%.html)
+DOCS_MAN3 := $(DOCS_XML:$(objroot)%.xml=$(objroot)%.3)
+DOCS := $(DOCS_HTML) $(DOCS_MAN3)
+C_TESTLIB_SRCS := $(srcroot)test/src/btalloc.c $(srcroot)test/src/btalloc_0.c \
+	$(srcroot)test/src/btalloc_1.c $(srcroot)test/src/math.c \
+	$(srcroot)test/src/mtx.c $(srcroot)test/src/mq.c \
+	$(srcroot)test/src/SFMT.c $(srcroot)test/src/test.c \
+	$(srcroot)test/src/thd.c $(srcroot)test/src/timer.c
+ifeq (1, $(link_whole_archive))
+C_UTIL_INTEGRATION_SRCS :=
+C_UTIL_CPP_SRCS :=
+else
+C_UTIL_INTEGRATION_SRCS := $(srcroot)src/nstime.c $(srcroot)src/malloc_io.c
+C_UTIL_CPP_SRCS := $(srcroot)src/nstime.c $(srcroot)src/malloc_io.c
+endif
+TESTS_UNIT := \
+	$(srcroot)test/unit/a0.c \
+	$(srcroot)test/unit/arena_reset.c \
+	$(srcroot)test/unit/atomic.c \
+	$(srcroot)test/unit/background_thread.c \
+	$(srcroot)test/unit/background_thread_enable.c \
+	$(srcroot)test/unit/base.c \
+	$(srcroot)test/unit/bitmap.c \
+	$(srcroot)test/unit/bit_util.c \
+	$(srcroot)test/unit/binshard.c \
+	$(srcroot)test/unit/ckh.c \
+	$(srcroot)test/unit/decay.c \
+	$(srcroot)test/unit/div.c \
+	$(srcroot)test/unit/emitter.c \
+	$(srcroot)test/unit/extent_quantize.c \
+	$(srcroot)test/unit/extent_util.c \
+	$(srcroot)test/unit/fork.c \
+	$(srcroot)test/unit/hash.c \
+	$(srcroot)test/unit/hook.c \
+	$(srcroot)test/unit/huge.c \
+	$(srcroot)test/unit/junk.c \
+	$(srcroot)test/unit/junk_alloc.c \
+	$(srcroot)test/unit/junk_free.c \
+	$(srcroot)test/unit/log.c \
+	$(srcroot)test/unit/mallctl.c \
+	$(srcroot)test/unit/malloc_io.c \
+	$(srcroot)test/unit/math.c \
+	$(srcroot)test/unit/mq.c \
+	$(srcroot)test/unit/mtx.c \
+	$(srcroot)test/unit/pack.c \
+	$(srcroot)test/unit/pages.c \
+	$(srcroot)test/unit/ph.c \
+	$(srcroot)test/unit/prng.c \
+	$(srcroot)test/unit/prof_accum.c \
+	$(srcroot)test/unit/prof_active.c \
+	$(srcroot)test/unit/prof_gdump.c \
+	$(srcroot)test/unit/prof_idump.c \
+	$(srcroot)test/unit/prof_log.c \
+	$(srcroot)test/unit/prof_reset.c \
+	$(srcroot)test/unit/prof_tctx.c \
+	$(srcroot)test/unit/prof_thread_name.c \
+	$(srcroot)test/unit/ql.c \
+	$(srcroot)test/unit/qr.c \
+	$(srcroot)test/unit/rb.c \
+	$(srcroot)test/unit/retained.c \
+	$(srcroot)test/unit/rtree.c \
+	$(srcroot)test/unit/safety_check.c \
+	$(srcroot)test/unit/seq.c \
+	$(srcroot)test/unit/SFMT.c \
+	$(srcroot)test/unit/sc.c \
+	$(srcroot)test/unit/size_classes.c \
+	$(srcroot)test/unit/slab.c \
+	$(srcroot)test/unit/smoothstep.c \
+	$(srcroot)test/unit/spin.c \
+	$(srcroot)test/unit/stats.c \
+	$(srcroot)test/unit/stats_print.c \
+	$(srcroot)test/unit/test_hooks.c \
+	$(srcroot)test/unit/ticker.c \
+	$(srcroot)test/unit/nstime.c \
+	$(srcroot)test/unit/tsd.c \
+	$(srcroot)test/unit/witness.c \
+	$(srcroot)test/unit/zero.c
+ifeq (@enable_prof@, 1)
+TESTS_UNIT += \
+	$(srcroot)test/unit/arena_reset_prof.c
+endif
+TESTS_INTEGRATION := $(srcroot)test/integration/aligned_alloc.c \
+	$(srcroot)test/integration/allocated.c \
+	$(srcroot)test/integration/extent.c \
+	$(srcroot)test/integration/malloc.c \
+	$(srcroot)test/integration/mallocx.c \
+	$(srcroot)test/integration/MALLOCX_ARENA.c \
+	$(srcroot)test/integration/overflow.c \
+	$(srcroot)test/integration/posix_memalign.c \
+	$(srcroot)test/integration/rallocx.c \
+	$(srcroot)test/integration/sdallocx.c \
+	$(srcroot)test/integration/slab_sizes.c \
+	$(srcroot)test/integration/thread_arena.c \
+	$(srcroot)test/integration/thread_tcache_enabled.c \
+	$(srcroot)test/integration/xallocx.c
+ifeq (@enable_experimental_smallocx@, 1)
+TESTS_INTEGRATION += \
+  $(srcroot)test/integration/smallocx.c
+endif
+ifeq (@enable_cxx@, 1)
+CPP_SRCS := $(srcroot)src/jemalloc_cpp.cpp
+TESTS_INTEGRATION_CPP := $(srcroot)test/integration/cpp/basic.cpp
+else
+CPP_SRCS :=
+TESTS_INTEGRATION_CPP :=
+endif
+TESTS_STRESS := $(srcroot)test/stress/microbench.c \
+	$(srcroot)test/stress/hookbench.c
+
+
+TESTS := $(TESTS_UNIT) $(TESTS_INTEGRATION) $(TESTS_INTEGRATION_CPP) $(TESTS_STRESS)
+
+PRIVATE_NAMESPACE_HDRS := $(objroot)include/jemalloc/internal/private_namespace.h $(objroot)include/jemalloc/internal/private_namespace_jet.h
+PRIVATE_NAMESPACE_GEN_HDRS := $(PRIVATE_NAMESPACE_HDRS:%.h=%.gen.h)
+C_SYM_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.sym.$(O))
+C_SYMS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.sym)
+C_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.$(O))
+CPP_OBJS := $(CPP_SRCS:$(srcroot)%.cpp=$(objroot)%.$(O))
+C_PIC_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.pic.$(O))
+CPP_PIC_OBJS := $(CPP_SRCS:$(srcroot)%.cpp=$(objroot)%.pic.$(O))
+C_JET_SYM_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.sym.$(O))
+C_JET_SYMS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.sym)
+C_JET_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.$(O))
+C_TESTLIB_UNIT_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.unit.$(O))
+C_TESTLIB_INTEGRATION_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.integration.$(O))
+C_UTIL_INTEGRATION_OBJS := $(C_UTIL_INTEGRATION_SRCS:$(srcroot)%.c=$(objroot)%.integration.$(O))
+C_TESTLIB_STRESS_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.stress.$(O))
+C_TESTLIB_OBJS := $(C_TESTLIB_UNIT_OBJS) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(C_TESTLIB_STRESS_OBJS)
+
+TESTS_UNIT_OBJS := $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_INTEGRATION_OBJS := $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_INTEGRATION_CPP_OBJS := $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%.$(O))
+TESTS_STRESS_OBJS := $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_OBJS := $(TESTS_UNIT_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_STRESS_OBJS)
+TESTS_CPP_OBJS := $(TESTS_INTEGRATION_CPP_OBJS)
+
+.PHONY: all dist build_doc_html build_doc_man build_doc
+.PHONY: install_bin install_include install_lib
+.PHONY: install_doc_html install_doc_man install_doc install
+.PHONY: tests check clean distclean relclean
+
+.SECONDARY : $(PRIVATE_NAMESPACE_GEN_HDRS) $(TESTS_OBJS) $(TESTS_CPP_OBJS)
+
+# Default target.
+all: build_lib
+
+dist: build_doc
+
+$(objroot)doc/%.html : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/html.xsl
+ifneq ($(XSLROOT),)
+	$(XSLTPROC) -o $@ $(objroot)doc/html.xsl $<
+else
+ifeq ($(wildcard $(DOCS_HTML)),)
+	@echo "<p>Missing xsltproc.  Doc not built.</p>" > $@
+endif
+	@echo "Missing xsltproc.  "$@" not (re)built."
+endif
+
+$(objroot)doc/%.3 : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/manpages.xsl
+ifneq ($(XSLROOT),)
+	$(XSLTPROC) -o $@ $(objroot)doc/manpages.xsl $<
+else
+ifeq ($(wildcard $(DOCS_MAN3)),)
+	@echo "Missing xsltproc.  Doc not built." > $@
+endif
+	@echo "Missing xsltproc.  "$@" not (re)built."
+endif
+
+build_doc_html: $(DOCS_HTML)
+build_doc_man: $(DOCS_MAN3)
+build_doc: $(DOCS)
+
+#
+# Include generated dependency files.
+#
+ifdef CC_MM
+-include $(C_SYM_OBJS:%.$(O)=%.d)
+-include $(C_OBJS:%.$(O)=%.d)
+-include $(CPP_OBJS:%.$(O)=%.d)
+-include $(C_PIC_OBJS:%.$(O)=%.d)
+-include $(CPP_PIC_OBJS:%.$(O)=%.d)
+-include $(C_JET_SYM_OBJS:%.$(O)=%.d)
+-include $(C_JET_OBJS:%.$(O)=%.d)
+-include $(C_TESTLIB_OBJS:%.$(O)=%.d)
+-include $(TESTS_OBJS:%.$(O)=%.d)
+-include $(TESTS_CPP_OBJS:%.$(O)=%.d)
+endif
+
+$(C_SYM_OBJS): $(objroot)src/%.sym.$(O): $(srcroot)src/%.c
+$(C_SYM_OBJS): CPPFLAGS += -DJEMALLOC_NO_PRIVATE_NAMESPACE
+$(C_SYMS): $(objroot)src/%.sym: $(objroot)src/%.sym.$(O)
+$(C_OBJS): $(objroot)src/%.$(O): $(srcroot)src/%.c
+$(CPP_OBJS): $(objroot)src/%.$(O): $(srcroot)src/%.cpp
+$(C_PIC_OBJS): $(objroot)src/%.pic.$(O): $(srcroot)src/%.c
+$(C_PIC_OBJS): CFLAGS += $(PIC_CFLAGS)
+$(CPP_PIC_OBJS): $(objroot)src/%.pic.$(O): $(srcroot)src/%.cpp
+$(CPP_PIC_OBJS): CXXFLAGS += $(PIC_CFLAGS)
+$(C_JET_SYM_OBJS): $(objroot)src/%.jet.sym.$(O): $(srcroot)src/%.c
+$(C_JET_SYM_OBJS): CPPFLAGS += -DJEMALLOC_JET -DJEMALLOC_NO_PRIVATE_NAMESPACE
+$(C_JET_SYMS): $(objroot)src/%.jet.sym: $(objroot)src/%.jet.sym.$(O)
+$(C_JET_OBJS): $(objroot)src/%.jet.$(O): $(srcroot)src/%.c
+$(C_JET_OBJS): CPPFLAGS += -DJEMALLOC_JET
+$(C_TESTLIB_UNIT_OBJS): $(objroot)test/src/%.unit.$(O): $(srcroot)test/src/%.c
+$(C_TESTLIB_UNIT_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
+$(C_TESTLIB_INTEGRATION_OBJS): $(objroot)test/src/%.integration.$(O): $(srcroot)test/src/%.c
+$(C_TESTLIB_INTEGRATION_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_TEST
+$(C_UTIL_INTEGRATION_OBJS): $(objroot)src/%.integration.$(O): $(srcroot)src/%.c
+$(C_TESTLIB_STRESS_OBJS): $(objroot)test/src/%.stress.$(O): $(srcroot)test/src/%.c
+$(C_TESTLIB_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST -DJEMALLOC_STRESS_TESTLIB
+$(C_TESTLIB_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
+$(TESTS_UNIT_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
+$(TESTS_INTEGRATION_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_TEST
+$(TESTS_INTEGRATION_CPP_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_CPP_TEST
+$(TESTS_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST
+$(TESTS_OBJS): $(objroot)test/%.$(O): $(srcroot)test/%.c
+$(TESTS_CPP_OBJS): $(objroot)test/%.$(O): $(srcroot)test/%.cpp
+$(TESTS_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
+$(TESTS_CPP_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
+ifneq ($(IMPORTLIB),$(SO))
+$(CPP_OBJS) $(C_SYM_OBJS) $(C_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS): CPPFLAGS += -DDLLEXPORT
+endif
+
+# Dependencies.
+ifndef CC_MM
+HEADER_DIRS = $(srcroot)include/jemalloc/internal \
+	$(objroot)include/jemalloc $(objroot)include/jemalloc/internal
+HEADERS = $(filter-out $(PRIVATE_NAMESPACE_HDRS),$(wildcard $(foreach dir,$(HEADER_DIRS),$(dir)/*.h)))
+$(C_SYM_OBJS) $(C_OBJS) $(CPP_OBJS) $(C_PIC_OBJS) $(CPP_PIC_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS) $(TESTS_CPP_OBJS): $(HEADERS)
+$(TESTS_OBJS) $(TESTS_CPP_OBJS): $(objroot)test/include/test/jemalloc_test.h
+endif
+
+$(C_OBJS) $(CPP_OBJS) $(C_PIC_OBJS) $(CPP_PIC_OBJS) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_INTEGRATION_CPP_OBJS): $(objroot)include/jemalloc/internal/private_namespace.h
+$(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS) $(C_TESTLIB_STRESS_OBJS) $(TESTS_UNIT_OBJS) $(TESTS_STRESS_OBJS): $(objroot)include/jemalloc/internal/private_namespace_jet.h
+
+$(C_SYM_OBJS) $(C_OBJS) $(C_PIC_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS): %.$(O):
+	@mkdir -p $(@D)
+	$(CC) $(CFLAGS) -c $(CPPFLAGS) $(CTARGET) $<
+ifdef CC_MM
+	@$(CC) -MM $(CPPFLAGS) -MT $@ -o $(@:%.$(O)=%.d) $<
+endif
+
+$(C_SYMS): %.sym:
+	@mkdir -p $(@D)
+	$(DUMP_SYMS) $< | $(AWK) -f $(objroot)include/jemalloc/internal/private_symbols.awk > $@
+
+$(C_JET_SYMS): %.sym:
+	@mkdir -p $(@D)
+	$(DUMP_SYMS) $< | $(AWK) -f $(objroot)include/jemalloc/internal/private_symbols_jet.awk > $@
+
+$(objroot)include/jemalloc/internal/private_namespace.gen.h: $(C_SYMS)
+	$(SHELL) $(srcroot)include/jemalloc/internal/private_namespace.sh $^ > $@
+
+$(objroot)include/jemalloc/internal/private_namespace_jet.gen.h: $(C_JET_SYMS)
+	$(SHELL) $(srcroot)include/jemalloc/internal/private_namespace.sh $^ > $@
+
+%.h: %.gen.h
+	@if ! `cmp -s $< $@` ; then echo "cp $< $<"; cp $< $@ ; fi
+
+$(CPP_OBJS) $(CPP_PIC_OBJS) $(TESTS_CPP_OBJS): %.$(O):
+	@mkdir -p $(@D)
+	$(CXX) $(CXXFLAGS) -c $(CPPFLAGS) $(CTARGET) $<
+ifdef CC_MM
+	@$(CXX) -MM $(CPPFLAGS) -MT $@ -o $(@:%.$(O)=%.d) $<
+endif
+
+ifneq ($(SOREV),$(SO))
+%.$(SO) : %.$(SOREV)
+	@mkdir -p $(@D)
+	ln -sf $(<F) $@
+endif
+
+$(objroot)lib/$(LIBJEMALLOC).$(SOREV) : $(if $(PIC_CFLAGS),$(C_PIC_OBJS),$(C_OBJS)) $(if $(PIC_CFLAGS),$(CPP_PIC_OBJS),$(CPP_OBJS))
+	@mkdir -p $(@D)
+	$(CC) $(DSO_LDFLAGS) $(call RPATH,$(RPATH_EXTRA)) $(LDTARGET) $+ $(LDFLAGS) $(LIBS) $(EXTRA_LDFLAGS)
+
+$(objroot)lib/$(LIBJEMALLOC)_pic.$(A) : $(C_PIC_OBJS) $(CPP_PIC_OBJS)
+$(objroot)lib/$(LIBJEMALLOC).$(A) : $(C_OBJS) $(CPP_OBJS)
+$(objroot)lib/$(LIBJEMALLOC)_s.$(A) : $(C_OBJS) $(CPP_OBJS)
+
+$(STATIC_LIBS):
+	@mkdir -p $(@D)
+	$(AR) $(ARFLAGS)@AROUT@ $+
+
+$(objroot)test/unit/%$(EXE): $(objroot)test/unit/%.$(O) $(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS)
+	@mkdir -p $(@D)
+	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LDFLAGS) $(filter-out -lm,$(LIBS)) $(LM) $(EXTRA_LDFLAGS)
+
+$(objroot)test/integration/%$(EXE): $(objroot)test/integration/%.$(O) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+	@mkdir -p $(@D)
+	$(CC) $(TEST_LD_MODE) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LJEMALLOC) $(LDFLAGS) $(filter-out -lm,$(filter -lrt -pthread -lstdc++,$(LIBS))) $(LM) $(EXTRA_LDFLAGS)
+
+$(objroot)test/integration/cpp/%$(EXE): $(objroot)test/integration/cpp/%.$(O) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+	@mkdir -p $(@D)
+	$(CXX) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(LIBS)) -lm $(EXTRA_LDFLAGS)
+
+$(objroot)test/stress/%$(EXE): $(objroot)test/stress/%.$(O) $(C_JET_OBJS) $(C_TESTLIB_STRESS_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+	@mkdir -p $(@D)
+	$(CC) $(TEST_LD_MODE) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(LIBS)) $(LM) $(EXTRA_LDFLAGS)
+
+build_lib_shared: $(DSOS)
+build_lib_static: $(STATIC_LIBS)
+ifeq ($(enable_shared), 1)
+build_lib: build_lib_shared
+endif
+ifeq ($(enable_static), 1)
+build_lib: build_lib_static
+endif
+
+install_bin:
+	$(INSTALL) -d $(BINDIR)
+	@for b in $(BINS); do \
+	echo "$(INSTALL) -m 755 $$b $(BINDIR)"; \
+	$(INSTALL) -m 755 $$b $(BINDIR); \
+done
+
+install_include:
+	$(INSTALL) -d $(INCLUDEDIR)/jemalloc
+	@for h in $(C_HDRS); do \
+	echo "$(INSTALL) -m 644 $$h $(INCLUDEDIR)/jemalloc"; \
+	$(INSTALL) -m 644 $$h $(INCLUDEDIR)/jemalloc; \
+done
+
+install_lib_shared: $(DSOS)
+	$(INSTALL) -d $(LIBDIR)
+	$(INSTALL) -m 755 $(objroot)lib/$(LIBJEMALLOC).$(SOREV) $(LIBDIR)
+ifneq ($(SOREV),$(SO))
+	ln -sf $(LIBJEMALLOC).$(SOREV) $(LIBDIR)/$(LIBJEMALLOC).$(SO)
+endif
+
+install_lib_static: $(STATIC_LIBS)
+	$(INSTALL) -d $(LIBDIR)
+	@for l in $(STATIC_LIBS); do \
+	echo "$(INSTALL) -m 755 $$l $(LIBDIR)"; \
+	$(INSTALL) -m 755 $$l $(LIBDIR); \
+done
+
+install_lib_pc: $(PC)
+	$(INSTALL) -d $(LIBDIR)/pkgconfig
+	@for l in $(PC); do \
+	echo "$(INSTALL) -m 644 $$l $(LIBDIR)/pkgconfig"; \
+	$(INSTALL) -m 644 $$l $(LIBDIR)/pkgconfig; \
+done
+
+ifeq ($(enable_shared), 1)
+install_lib: install_lib_shared
+endif
+ifeq ($(enable_static), 1)
+install_lib: install_lib_static
+endif
+install_lib: install_lib_pc
+
+install_doc_html:
+	$(INSTALL) -d $(DATADIR)/doc/jemalloc$(install_suffix)
+	@for d in $(DOCS_HTML); do \
+	echo "$(INSTALL) -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix)"; \
+	$(INSTALL) -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix); \
+done
+
+install_doc_man:
+	$(INSTALL) -d $(MANDIR)/man3
+	@for d in $(DOCS_MAN3); do \
+	echo "$(INSTALL) -m 644 $$d $(MANDIR)/man3"; \
+	$(INSTALL) -m 644 $$d $(MANDIR)/man3; \
+done
+
+install_doc: build_doc install_doc_html install_doc_man
+
+install: install_bin install_include install_lib
+
+ifeq ($(enable_doc), 1)
+install: install_doc
+endif
+
+tests_unit: $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%$(EXE))
+tests_integration: $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%$(EXE)) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%$(EXE))
+tests_stress: $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%$(EXE))
+tests: tests_unit tests_integration tests_stress
+
+check_unit_dir:
+	@mkdir -p $(objroot)test/unit
+check_integration_dir:
+	@mkdir -p $(objroot)test/integration
+stress_dir:
+	@mkdir -p $(objroot)test/stress
+check_dir: check_unit_dir check_integration_dir
+
+check_unit: tests_unit check_unit_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%)
+check_integration_prof: tests_integration check_integration_dir
+ifeq ($(enable_prof), 1)
+	$(MALLOC_CONF)="prof:true" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+	$(MALLOC_CONF)="prof:true,prof_active:false" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+endif
+check_integration_decay: tests_integration check_integration_dir
+	$(MALLOC_CONF)="dirty_decay_ms:-1,muzzy_decay_ms:-1" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+	$(MALLOC_CONF)="dirty_decay_ms:0,muzzy_decay_ms:0" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+check_integration: tests_integration check_integration_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+stress: tests_stress stress_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%)
+check: check_unit check_integration check_integration_decay check_integration_prof
+
+clean:
+	rm -f $(PRIVATE_NAMESPACE_HDRS)
+	rm -f $(PRIVATE_NAMESPACE_GEN_HDRS)
+	rm -f $(C_SYM_OBJS)
+	rm -f $(C_SYMS)
+	rm -f $(C_OBJS)
+	rm -f $(CPP_OBJS)
+	rm -f $(C_PIC_OBJS)
+	rm -f $(CPP_PIC_OBJS)
+	rm -f $(C_JET_SYM_OBJS)
+	rm -f $(C_JET_SYMS)
+	rm -f $(C_JET_OBJS)
+	rm -f $(C_TESTLIB_OBJS)
+	rm -f $(C_SYM_OBJS:%.$(O)=%.d)
+	rm -f $(C_OBJS:%.$(O)=%.d)
+	rm -f $(CPP_OBJS:%.$(O)=%.d)
+	rm -f $(C_PIC_OBJS:%.$(O)=%.d)
+	rm -f $(CPP_PIC_OBJS:%.$(O)=%.d)
+	rm -f $(C_JET_SYM_OBJS:%.$(O)=%.d)
+	rm -f $(C_JET_OBJS:%.$(O)=%.d)
+	rm -f $(C_TESTLIB_OBJS:%.$(O)=%.d)
+	rm -f $(TESTS_OBJS:%.$(O)=%$(EXE))
+	rm -f $(TESTS_OBJS)
+	rm -f $(TESTS_OBJS:%.$(O)=%.d)
+	rm -f $(TESTS_OBJS:%.$(O)=%.out)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%$(EXE))
+	rm -f $(TESTS_CPP_OBJS)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%.d)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%.out)
+	rm -f $(DSOS) $(STATIC_LIBS)
+
+distclean: clean
+	rm -f $(objroot)bin/jemalloc-config
+	rm -f $(objroot)bin/jemalloc.sh
+	rm -f $(objroot)bin/jeprof
+	rm -f $(objroot)config.log
+	rm -f $(objroot)config.status
+	rm -f $(objroot)config.stamp
+	rm -f $(cfghdrs_out)
+	rm -f $(cfgoutputs_out)
+
+relclean: distclean
+	rm -f $(objroot)configure
+	rm -f $(objroot)VERSION
+	rm -f $(DOCS_HTML)
+	rm -f $(DOCS_MAN3)
+
+#===============================================================================
+# Re-configuration rules.
+
+ifeq ($(enable_autogen), 1)
+$(srcroot)configure : $(srcroot)configure.ac
+	cd ./$(srcroot) && $(AUTOCONF)
+
+$(objroot)config.status : $(srcroot)configure
+	./$(objroot)config.status --recheck
+
+$(srcroot)config.stamp.in : $(srcroot)configure.ac
+	echo stamp > $(srcroot)config.stamp.in
+
+$(objroot)config.stamp : $(cfgoutputs_in) $(cfghdrs_in) $(srcroot)configure
+	./$(objroot)config.status
+	@touch $@
+
+# There must be some action in order for make to re-read Makefile when it is
+# out of date.
+$(cfgoutputs_out) $(cfghdrs_out) : $(objroot)config.stamp
+	@true
+endif
diff --git a/memkind-1.10.0/jemalloc/README b/memkind-1.10.0/jemalloc/README
new file mode 100644
index 0000000..3a6e0d2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/README
@@ -0,0 +1,20 @@
+jemalloc is a general purpose malloc(3) implementation that emphasizes
+fragmentation avoidance and scalable concurrency support.  jemalloc first came
+into use as the FreeBSD libc allocator in 2005, and since then it has found its
+way into numerous applications that rely on its predictable behavior.  In 2010
+jemalloc development efforts broadened to include developer support features
+such as heap profiling and extensive monitoring/tuning hooks.  Modern jemalloc
+releases continue to be integrated back into FreeBSD, and therefore versatility
+remains critical.  Ongoing development efforts trend toward making jemalloc
+among the best allocators for a broad range of demanding applications, and
+eliminating/mitigating weaknesses that have practical repercussions for real
+world applications.
+
+The COPYING file contains copyright and licensing information.
+
+The INSTALL file contains information on how to configure, build, and install
+jemalloc.
+
+The ChangeLog file contains a brief summary of changes for each release.
+
+URL: http://jemalloc.net/
diff --git a/memkind-1.10.0/jemalloc/TUNING.md b/memkind-1.10.0/jemalloc/TUNING.md
new file mode 100644
index 0000000..7d094f3
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/TUNING.md
@@ -0,0 +1,129 @@
+This document summarizes the common approaches for performance fine tuning with
+jemalloc (as of 5.1.0).  The default configuration of jemalloc tends to work
+reasonably well in practice, and most applications should not have to tune any
+options. However, in order to cover a wide range of applications and avoid
+pathological cases, the default setting is sometimes kept conservative and
+suboptimal, even for many common workloads.  When jemalloc is properly tuned for
+a specific application / workload, it is common to improve system level metrics
+by a few percent, or make favorable trade-offs.
+
+
+## Notable runtime options for performance tuning
+
+Runtime options can be set via
+[malloc_conf](http://jemalloc.net/jemalloc.3.html#tuning).
+
+* [background_thread](http://jemalloc.net/jemalloc.3.html#background_thread)
+
+    Enabling jemalloc background threads generally improves the tail latency for
+    application threads, since unused memory purging is shifted to the dedicated
+    background threads.  In addition, unintended purging delay caused by
+    application inactivity is avoided with background threads.
+
+    Suggested: `background_thread:true` when jemalloc managed threads can be
+    allowed.
+
+* [metadata_thp](http://jemalloc.net/jemalloc.3.html#opt.metadata_thp)
+
+    Allowing jemalloc to utilize transparent huge pages for its internal
+    metadata usually reduces TLB misses significantly, especially for programs
+    with large memory footprint and frequent allocation / deallocation
+    activities.  Metadata memory usage may increase due to the use of huge
+    pages.
+
+    Suggested for allocation intensive programs: `metadata_thp:auto` or
+    `metadata_thp:always`, which is expected to improve CPU utilization at a
+    small memory cost.
+
+* [dirty_decay_ms](http://jemalloc.net/jemalloc.3.html#opt.dirty_decay_ms) and
+  [muzzy_decay_ms](http://jemalloc.net/jemalloc.3.html#opt.muzzy_decay_ms)
+
+    Decay time determines how fast jemalloc returns unused pages back to the
+    operating system, and therefore provides a fairly straightforward trade-off
+    between CPU and memory usage.  Shorter decay time purges unused pages faster
+    to reduces memory usage (usually at the cost of more CPU cycles spent on
+    purging), and vice versa.
+
+    Suggested: tune the values based on the desired trade-offs.
+
+* [narenas](http://jemalloc.net/jemalloc.3.html#opt.narenas)
+
+    By default jemalloc uses multiple arenas to reduce internal lock contention.
+    However high arena count may also increase overall memory fragmentation,
+    since arenas manage memory independently.  When high degree of parallelism
+    is not expected at the allocator level, lower number of arenas often
+    improves memory usage.
+
+    Suggested: if low parallelism is expected, try lower arena count while
+    monitoring CPU and memory usage.
+
+* [percpu_arena](http://jemalloc.net/jemalloc.3.html#opt.percpu_arena)
+
+    Enable dynamic thread to arena association based on running CPU.  This has
+    the potential to improve locality, e.g. when thread to CPU affinity is
+    present.
+
+    Suggested: try `percpu_arena:percpu` or `percpu_arena:phycpu` if
+    thread migration between processors is expected to be infrequent.
+
+Examples:
+
+* High resource consumption application, prioritizing CPU utilization:
+
+    `background_thread:true,metadata_thp:auto` combined with relaxed decay time
+    (increased `dirty_decay_ms` and / or `muzzy_decay_ms`,
+    e.g. `dirty_decay_ms:30000,muzzy_decay_ms:30000`).
+
+* High resource consumption application, prioritizing memory usage:
+
+    `background_thread:true` combined with shorter decay time (decreased
+    `dirty_decay_ms` and / or `muzzy_decay_ms`,
+    e.g. `dirty_decay_ms:5000,muzzy_decay_ms:5000`), and lower arena count
+    (e.g. number of CPUs).
+
+* Low resource consumption application:
+
+    `narenas:1,lg_tcache_max:13` combined with shorter decay time (decreased
+    `dirty_decay_ms` and / or `muzzy_decay_ms`,e.g.
+    `dirty_decay_ms:1000,muzzy_decay_ms:0`).
+
+* Extremely conservative -- minimize memory usage at all costs, only suitable when
+allocation activity is very rare:
+
+    `narenas:1,tcache:false,dirty_decay_ms:0,muzzy_decay_ms:0`
+
+Note that it is recommended to combine the options with `abort_conf:true` which
+aborts immediately on illegal options.
+
+## Beyond runtime options
+
+In addition to the runtime options, there are a number of programmatic ways to
+improve application performance with jemalloc.
+
+* [Explicit arenas](http://jemalloc.net/jemalloc.3.html#arenas.create)
+
+    Manually created arenas can help performance in various ways, e.g. by
+    managing locality and contention for specific usages.  For example,
+    applications can explicitly allocate frequently accessed objects from a
+    dedicated arena with
+    [mallocx()](http://jemalloc.net/jemalloc.3.html#MALLOCX_ARENA) to improve
+    locality.  In addition, explicit arenas often benefit from individually
+    tuned options, e.g. relaxed [decay
+    time](http://jemalloc.net/jemalloc.3.html#arena.i.dirty_decay_ms) if
+    frequent reuse is expected.
+
+* [Extent hooks](http://jemalloc.net/jemalloc.3.html#arena.i.extent_hooks)
+
+    Extent hooks allow customization for managing underlying memory.  One use
+    case for performance purpose is to utilize huge pages -- for example,
+    [HHVM](https://github.com/facebook/hhvm/blob/master/hphp/util/alloc.cpp)
+    uses explicit arenas with customized extent hooks to manage 1GB huge pages
+    for frequently accessed data, which reduces TLB misses significantly.
+
+* [Explicit thread-to-arena
+  binding](http://jemalloc.net/jemalloc.3.html#thread.arena)
+
+    It is common for some threads in an application to have different memory
+    access / allocation patterns.  Threads with heavy workloads often benefit
+    from explicit binding, e.g. binding very active threads to dedicated arenas
+    may reduce contention at the allocator level.
diff --git a/memkind-1.10.0/jemalloc/autogen.sh b/memkind-1.10.0/jemalloc/autogen.sh
new file mode 100755
index 0000000..75f32da
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/autogen.sh
@@ -0,0 +1,17 @@
+#!/bin/sh
+
+for i in autoconf; do
+    echo "$i"
+    $i
+    if [ $? -ne 0 ]; then
+	echo "Error $? in $i"
+	exit 1
+    fi
+done
+
+echo "./configure --enable-autogen $@"
+./configure --enable-autogen $@
+if [ $? -ne 0 ]; then
+    echo "Error $? in ./configure"
+    exit 1
+fi
diff --git a/memkind-1.10.0/jemalloc/bin/jemalloc-config.in b/memkind-1.10.0/jemalloc/bin/jemalloc-config.in
new file mode 100644
index 0000000..80eca2e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/bin/jemalloc-config.in
@@ -0,0 +1,83 @@
+#!/bin/sh
+
+usage() {
+	cat <<EOF
+Usage:
+  @BINDIR@/jemalloc-config <option>
+Options:
+  --help | -h  : Print usage.
+  --version    : Print jemalloc version.
+  --revision   : Print shared library revision number.
+  --config     : Print configure options used to build jemalloc.
+  --prefix     : Print installation directory prefix.
+  --bindir     : Print binary installation directory.
+  --datadir    : Print data installation directory.
+  --includedir : Print include installation directory.
+  --libdir     : Print library installation directory.
+  --mandir     : Print manual page installation directory.
+  --cc         : Print compiler used to build jemalloc.
+  --cflags     : Print compiler flags used to build jemalloc.
+  --cppflags   : Print preprocessor flags used to build jemalloc.
+  --cxxflags   : Print C++ compiler flags used to build jemalloc.
+  --ldflags    : Print library flags used to build jemalloc.
+  --libs       : Print libraries jemalloc was linked against.
+EOF
+}
+
+prefix="@prefix@"
+exec_prefix="@exec_prefix@"
+
+case "$1" in
+--help | -h)
+	usage
+	exit 0
+	;;
+--version)
+	echo "@jemalloc_version@"
+	;;
+--revision)
+	echo "@rev@"
+	;;
+--config)
+	echo "@CONFIG@"
+	;;
+--prefix)
+	echo "@PREFIX@"
+	;;
+--bindir)
+	echo "@BINDIR@"
+	;;
+--datadir)
+	echo "@DATADIR@"
+	;;
+--includedir)
+	echo "@INCLUDEDIR@"
+	;;
+--libdir)
+	echo "@LIBDIR@"
+	;;
+--mandir)
+	echo "@MANDIR@"
+	;;
+--cc)
+	echo "@CC@"
+	;;
+--cflags)
+	echo "@CFLAGS@"
+	;;
+--cppflags)
+	echo "@CPPFLAGS@"
+	;;
+--cxxflags)
+	echo "@CXXFLAGS@"
+	;;
+--ldflags)
+	echo "@LDFLAGS@ @EXTRA_LDFLAGS@"
+	;;
+--libs)
+	echo "@LIBS@"
+	;;
+*)
+	usage
+	exit 1
+esac
diff --git a/memkind-1.10.0/jemalloc/bin/jemalloc.sh.in b/memkind-1.10.0/jemalloc/bin/jemalloc.sh.in
new file mode 100644
index 0000000..cdf3673
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/bin/jemalloc.sh.in
@@ -0,0 +1,9 @@
+#!/bin/sh
+
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+
+@LD_PRELOAD_VAR@=${libdir}/libjemalloc.@SOREV@
+export @LD_PRELOAD_VAR@
+exec "$@"
diff --git a/memkind-1.10.0/jemalloc/bin/jeprof.in b/memkind-1.10.0/jemalloc/bin/jeprof.in
new file mode 100644
index 0000000..3ed408c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/bin/jeprof.in
@@ -0,0 +1,5625 @@
+#! /usr/bin/env perl
+
+# Copyright (c) 1998-2007, Google Inc.
+# All rights reserved.
+#
+# Redistribution and use in source and binary forms, with or without
+# modification, are permitted provided that the following conditions are
+# met:
+#
+#     * Redistributions of source code must retain the above copyright
+# notice, this list of conditions and the following disclaimer.
+#     * Redistributions in binary form must reproduce the above
+# copyright notice, this list of conditions and the following disclaimer
+# in the documentation and/or other materials provided with the
+# distribution.
+#     * Neither the name of Google Inc. nor the names of its
+# contributors may be used to endorse or promote products derived from
+# this software without specific prior written permission.
+#
+# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# ---
+# Program for printing the profile generated by common/profiler.cc,
+# or by the heap profiler (common/debugallocation.cc)
+#
+# The profile contains a sequence of entries of the form:
+#       <count> <stack trace>
+# This program parses the profile, and generates user-readable
+# output.
+#
+# Examples:
+#
+# % tools/jeprof "program" "profile"
+#   Enters "interactive" mode
+#
+# % tools/jeprof --text "program" "profile"
+#   Generates one line per procedure
+#
+# % tools/jeprof --gv "program" "profile"
+#   Generates annotated call-graph and displays via "gv"
+#
+# % tools/jeprof --gv --focus=Mutex "program" "profile"
+#   Restrict to code paths that involve an entry that matches "Mutex"
+#
+# % tools/jeprof --gv --focus=Mutex --ignore=string "program" "profile"
+#   Restrict to code paths that involve an entry that matches "Mutex"
+#   and does not match "string"
+#
+# % tools/jeprof --list=IBF_CheckDocid "program" "profile"
+#   Generates disassembly listing of all routines with at least one
+#   sample that match the --list=<regexp> pattern.  The listing is
+#   annotated with the flat and cumulative sample counts at each line.
+#
+# % tools/jeprof --disasm=IBF_CheckDocid "program" "profile"
+#   Generates disassembly listing of all routines with at least one
+#   sample that match the --disasm=<regexp> pattern.  The listing is
+#   annotated with the flat and cumulative sample counts at each PC value.
+#
+# TODO: Use color to indicate files?
+
+use strict;
+use warnings;
+use Getopt::Long;
+use Cwd;
+
+my $JEPROF_VERSION = "@jemalloc_version@";
+my $PPROF_VERSION = "2.0";
+
+# These are the object tools we use which can come from a
+# user-specified location using --tools, from the JEPROF_TOOLS
+# environment variable, or from the environment.
+my %obj_tool_map = (
+  "objdump" => "objdump",
+  "nm" => "nm",
+  "addr2line" => "addr2line",
+  "c++filt" => "c++filt",
+  ## ConfigureObjTools may add architecture-specific entries:
+  #"nm_pdb" => "nm-pdb",       # for reading windows (PDB-format) executables
+  #"addr2line_pdb" => "addr2line-pdb",                                # ditto
+  #"otool" => "otool",         # equivalent of objdump on OS X
+);
+# NOTE: these are lists, so you can put in commandline flags if you want.
+my @DOT = ("dot");          # leave non-absolute, since it may be in /usr/local
+my @GV = ("gv");
+my @EVINCE = ("evince");    # could also be xpdf or perhaps acroread
+my @KCACHEGRIND = ("kcachegrind");
+my @PS2PDF = ("ps2pdf");
+# These are used for dynamic profiles
+my @URL_FETCHER = ("curl", "-s", "--fail");
+
+# These are the web pages that servers need to support for dynamic profiles
+my $HEAP_PAGE = "/pprof/heap";
+my $PROFILE_PAGE = "/pprof/profile";   # must support cgi-param "?seconds=#"
+my $PMUPROFILE_PAGE = "/pprof/pmuprofile(?:\\?.*)?"; # must support cgi-param
+                                                # ?seconds=#&event=x&period=n
+my $GROWTH_PAGE = "/pprof/growth";
+my $CONTENTION_PAGE = "/pprof/contention";
+my $WALL_PAGE = "/pprof/wall(?:\\?.*)?";  # accepts options like namefilter
+my $FILTEREDPROFILE_PAGE = "/pprof/filteredprofile(?:\\?.*)?";
+my $CENSUSPROFILE_PAGE = "/pprof/censusprofile(?:\\?.*)?"; # must support cgi-param
+                                                       # "?seconds=#",
+                                                       # "?tags_regexp=#" and
+                                                       # "?type=#".
+my $SYMBOL_PAGE = "/pprof/symbol";     # must support symbol lookup via POST
+my $PROGRAM_NAME_PAGE = "/pprof/cmdline";
+
+# These are the web pages that can be named on the command line.
+# All the alternatives must begin with /.
+my $PROFILES = "($HEAP_PAGE|$PROFILE_PAGE|$PMUPROFILE_PAGE|" .
+               "$GROWTH_PAGE|$CONTENTION_PAGE|$WALL_PAGE|" .
+               "$FILTEREDPROFILE_PAGE|$CENSUSPROFILE_PAGE)";
+
+# default binary name
+my $UNKNOWN_BINARY = "(unknown)";
+
+# There is a pervasive dependency on the length (in hex characters,
+# i.e., nibbles) of an address, distinguishing between 32-bit and
+# 64-bit profiles.  To err on the safe size, default to 64-bit here:
+my $address_length = 16;
+
+my $dev_null = "/dev/null";
+if (! -e $dev_null && $^O =~ /MSWin/) {    # $^O is the OS perl was built for
+  $dev_null = "nul";
+}
+
+# A list of paths to search for shared object files
+my @prefix_list = ();
+
+# Special routine name that should not have any symbols.
+# Used as separator to parse "addr2line -i" output.
+my $sep_symbol = '_fini';
+my $sep_address = undef;
+
+##### Argument parsing #####
+
+sub usage_string {
+  return <<EOF;
+Usage:
+jeprof [options] <program> <profiles>
+   <profiles> is a space separated list of profile names.
+jeprof [options] <symbolized-profiles>
+   <symbolized-profiles> is a list of profile files where each file contains
+   the necessary symbol mappings  as well as profile data (likely generated
+   with --raw).
+jeprof [options] <profile>
+   <profile> is a remote form.  Symbols are obtained from host:port$SYMBOL_PAGE
+
+   Each name can be:
+   /path/to/profile        - a path to a profile file
+   host:port[/<service>]   - a location of a service to get profile from
+
+   The /<service> can be $HEAP_PAGE, $PROFILE_PAGE, /pprof/pmuprofile,
+                         $GROWTH_PAGE, $CONTENTION_PAGE, /pprof/wall,
+                         $CENSUSPROFILE_PAGE, or /pprof/filteredprofile.
+   For instance:
+     jeprof http://myserver.com:80$HEAP_PAGE
+   If /<service> is omitted, the service defaults to $PROFILE_PAGE (cpu profiling).
+jeprof --symbols <program>
+   Maps addresses to symbol names.  In this mode, stdin should be a
+   list of library mappings, in the same format as is found in the heap-
+   and cpu-profile files (this loosely matches that of /proc/self/maps
+   on linux), followed by a list of hex addresses to map, one per line.
+
+   For more help with querying remote servers, including how to add the
+   necessary server-side support code, see this filename (or one like it):
+
+   /usr/doc/gperftools-$PPROF_VERSION/pprof_remote_servers.html
+
+Options:
+   --cum               Sort by cumulative data
+   --base=<base>       Subtract <base> from <profile> before display
+   --interactive       Run in interactive mode (interactive "help" gives help) [default]
+   --seconds=<n>       Length of time for dynamic profiles [default=30 secs]
+   --add_lib=<file>    Read additional symbols and line info from the given library
+   --lib_prefix=<dir>  Comma separated list of library path prefixes
+
+Reporting Granularity:
+   --addresses         Report at address level
+   --lines             Report at source line level
+   --functions         Report at function level [default]
+   --files             Report at source file level
+
+Output type:
+   --text              Generate text report
+   --callgrind         Generate callgrind format to stdout
+   --gv                Generate Postscript and display
+   --evince            Generate PDF and display
+   --web               Generate SVG and display
+   --list=<regexp>     Generate source listing of matching routines
+   --disasm=<regexp>   Generate disassembly of matching routines
+   --symbols           Print demangled symbol names found at given addresses
+   --dot               Generate DOT file to stdout
+   --ps                Generate Postcript to stdout
+   --pdf               Generate PDF to stdout
+   --svg               Generate SVG to stdout
+   --gif               Generate GIF to stdout
+   --raw               Generate symbolized jeprof data (useful with remote fetch)
+
+Heap-Profile Options:
+   --inuse_space       Display in-use (mega)bytes [default]
+   --inuse_objects     Display in-use objects
+   --alloc_space       Display allocated (mega)bytes
+   --alloc_objects     Display allocated objects
+   --show_bytes        Display space in bytes
+   --drop_negative     Ignore negative differences
+
+Contention-profile options:
+   --total_delay       Display total delay at each region [default]
+   --contentions       Display number of delays at each region
+   --mean_delay        Display mean delay at each region
+
+Call-graph Options:
+   --nodecount=<n>     Show at most so many nodes [default=80]
+   --nodefraction=<f>  Hide nodes below <f>*total [default=.005]
+   --edgefraction=<f>  Hide edges below <f>*total [default=.001]
+   --maxdegree=<n>     Max incoming/outgoing edges per node [default=8]
+   --focus=<regexp>    Focus on backtraces with nodes matching <regexp>
+   --thread=<n>        Show profile for thread <n>
+   --ignore=<regexp>   Ignore backtraces with nodes matching <regexp>
+   --scale=<n>         Set GV scaling [default=0]
+   --heapcheck         Make nodes with non-0 object counts
+                       (i.e. direct leak generators) more visible
+   --retain=<regexp>   Retain only nodes that match <regexp>
+   --exclude=<regexp>  Exclude all nodes that match <regexp>
+
+Miscellaneous:
+   --tools=<prefix or binary:fullpath>[,...]   \$PATH for object tool pathnames
+   --test              Run unit tests
+   --help              This message
+   --version           Version information
+
+Environment Variables:
+   JEPROF_TMPDIR        Profiles directory. Defaults to \$HOME/jeprof
+   JEPROF_TOOLS         Prefix for object tools pathnames
+
+Examples:
+
+jeprof /bin/ls ls.prof
+                       Enters "interactive" mode
+jeprof --text /bin/ls ls.prof
+                       Outputs one line per procedure
+jeprof --web /bin/ls ls.prof
+                       Displays annotated call-graph in web browser
+jeprof --gv /bin/ls ls.prof
+                       Displays annotated call-graph via 'gv'
+jeprof --gv --focus=Mutex /bin/ls ls.prof
+                       Restricts to code paths including a .*Mutex.* entry
+jeprof --gv --focus=Mutex --ignore=string /bin/ls ls.prof
+                       Code paths including Mutex but not string
+jeprof --list=getdir /bin/ls ls.prof
+                       (Per-line) annotated source listing for getdir()
+jeprof --disasm=getdir /bin/ls ls.prof
+                       (Per-PC) annotated disassembly for getdir()
+
+jeprof http://localhost:1234/
+                       Enters "interactive" mode
+jeprof --text localhost:1234
+                       Outputs one line per procedure for localhost:1234
+jeprof --raw localhost:1234 > ./local.raw
+jeprof --text ./local.raw
+                       Fetches a remote profile for later analysis and then
+                       analyzes it in text mode.
+EOF
+}
+
+sub version_string {
+  return <<EOF
+jeprof (part of jemalloc $JEPROF_VERSION)
+based on pprof (part of gperftools $PPROF_VERSION)
+
+Copyright 1998-2007 Google Inc.
+
+This is BSD licensed software; see the source for copying conditions
+and license information.
+There is NO warranty; not even for MERCHANTABILITY or FITNESS FOR A
+PARTICULAR PURPOSE.
+EOF
+}
+
+sub usage {
+  my $msg = shift;
+  print STDERR "$msg\n\n";
+  print STDERR usage_string();
+  print STDERR "\nFATAL ERROR: $msg\n";    # just as a reminder
+  exit(1);
+}
+
+sub Init() {
+  # Setup tmp-file name and handler to clean it up.
+  # We do this in the very beginning so that we can use
+  # error() and cleanup() function anytime here after.
+  $main::tmpfile_sym = "/tmp/jeprof$$.sym";
+  $main::tmpfile_ps = "/tmp/jeprof$$";
+  $main::next_tmpfile = 0;
+  $SIG{'INT'} = \&sighandler;
+
+  # Cache from filename/linenumber to source code
+  $main::source_cache = ();
+
+  $main::opt_help = 0;
+  $main::opt_version = 0;
+
+  $main::opt_cum = 0;
+  $main::opt_base = '';
+  $main::opt_addresses = 0;
+  $main::opt_lines = 0;
+  $main::opt_functions = 0;
+  $main::opt_files = 0;
+  $main::opt_lib_prefix = "";
+
+  $main::opt_text = 0;
+  $main::opt_callgrind = 0;
+  $main::opt_list = "";
+  $main::opt_disasm = "";
+  $main::opt_symbols = 0;
+  $main::opt_gv = 0;
+  $main::opt_evince = 0;
+  $main::opt_web = 0;
+  $main::opt_dot = 0;
+  $main::opt_ps = 0;
+  $main::opt_pdf = 0;
+  $main::opt_gif = 0;
+  $main::opt_svg = 0;
+  $main::opt_raw = 0;
+
+  $main::opt_nodecount = 80;
+  $main::opt_nodefraction = 0.005;
+  $main::opt_edgefraction = 0.001;
+  $main::opt_maxdegree = 8;
+  $main::opt_focus = '';
+  $main::opt_thread = undef;
+  $main::opt_ignore = '';
+  $main::opt_scale = 0;
+  $main::opt_heapcheck = 0;
+  $main::opt_retain = '';
+  $main::opt_exclude = '';
+  $main::opt_seconds = 30;
+  $main::opt_lib = "";
+
+  $main::opt_inuse_space   = 0;
+  $main::opt_inuse_objects = 0;
+  $main::opt_alloc_space   = 0;
+  $main::opt_alloc_objects = 0;
+  $main::opt_show_bytes    = 0;
+  $main::opt_drop_negative = 0;
+  $main::opt_interactive   = 0;
+
+  $main::opt_total_delay = 0;
+  $main::opt_contentions = 0;
+  $main::opt_mean_delay = 0;
+
+  $main::opt_tools   = "";
+  $main::opt_debug   = 0;
+  $main::opt_test    = 0;
+
+  # These are undocumented flags used only by unittests.
+  $main::opt_test_stride = 0;
+
+  # Are we using $SYMBOL_PAGE?
+  $main::use_symbol_page = 0;
+
+  # Files returned by TempName.
+  %main::tempnames = ();
+
+  # Type of profile we are dealing with
+  # Supported types:
+  #     cpu
+  #     heap
+  #     growth
+  #     contention
+  $main::profile_type = '';     # Empty type means "unknown"
+
+  GetOptions("help!"          => \$main::opt_help,
+             "version!"       => \$main::opt_version,
+             "cum!"           => \$main::opt_cum,
+             "base=s"         => \$main::opt_base,
+             "seconds=i"      => \$main::opt_seconds,
+             "add_lib=s"      => \$main::opt_lib,
+             "lib_prefix=s"   => \$main::opt_lib_prefix,
+             "functions!"     => \$main::opt_functions,
+             "lines!"         => \$main::opt_lines,
+             "addresses!"     => \$main::opt_addresses,
+             "files!"         => \$main::opt_files,
+             "text!"          => \$main::opt_text,
+             "callgrind!"     => \$main::opt_callgrind,
+             "list=s"         => \$main::opt_list,
+             "disasm=s"       => \$main::opt_disasm,
+             "symbols!"       => \$main::opt_symbols,
+             "gv!"            => \$main::opt_gv,
+             "evince!"        => \$main::opt_evince,
+             "web!"           => \$main::opt_web,
+             "dot!"           => \$main::opt_dot,
+             "ps!"            => \$main::opt_ps,
+             "pdf!"           => \$main::opt_pdf,
+             "svg!"           => \$main::opt_svg,
+             "gif!"           => \$main::opt_gif,
+             "raw!"           => \$main::opt_raw,
+             "interactive!"   => \$main::opt_interactive,
+             "nodecount=i"    => \$main::opt_nodecount,
+             "nodefraction=f" => \$main::opt_nodefraction,
+             "edgefraction=f" => \$main::opt_edgefraction,
+             "maxdegree=i"    => \$main::opt_maxdegree,
+             "focus=s"        => \$main::opt_focus,
+             "thread=s"       => \$main::opt_thread,
+             "ignore=s"       => \$main::opt_ignore,
+             "scale=i"        => \$main::opt_scale,
+             "heapcheck"      => \$main::opt_heapcheck,
+             "retain=s"       => \$main::opt_retain,
+             "exclude=s"      => \$main::opt_exclude,
+             "inuse_space!"   => \$main::opt_inuse_space,
+             "inuse_objects!" => \$main::opt_inuse_objects,
+             "alloc_space!"   => \$main::opt_alloc_space,
+             "alloc_objects!" => \$main::opt_alloc_objects,
+             "show_bytes!"    => \$main::opt_show_bytes,
+             "drop_negative!" => \$main::opt_drop_negative,
+             "total_delay!"   => \$main::opt_total_delay,
+             "contentions!"   => \$main::opt_contentions,
+             "mean_delay!"    => \$main::opt_mean_delay,
+             "tools=s"        => \$main::opt_tools,
+             "test!"          => \$main::opt_test,
+             "debug!"         => \$main::opt_debug,
+             # Undocumented flags used only by unittests:
+             "test_stride=i"  => \$main::opt_test_stride,
+      ) || usage("Invalid option(s)");
+
+  # Deal with the standard --help and --version
+  if ($main::opt_help) {
+    print usage_string();
+    exit(0);
+  }
+
+  if ($main::opt_version) {
+    print version_string();
+    exit(0);
+  }
+
+  # Disassembly/listing/symbols mode requires address-level info
+  if ($main::opt_disasm || $main::opt_list || $main::opt_symbols) {
+    $main::opt_functions = 0;
+    $main::opt_lines = 0;
+    $main::opt_addresses = 1;
+    $main::opt_files = 0;
+  }
+
+  # Check heap-profiling flags
+  if ($main::opt_inuse_space +
+      $main::opt_inuse_objects +
+      $main::opt_alloc_space +
+      $main::opt_alloc_objects > 1) {
+    usage("Specify at most on of --inuse/--alloc options");
+  }
+
+  # Check output granularities
+  my $grains =
+      $main::opt_functions +
+      $main::opt_lines +
+      $main::opt_addresses +
+      $main::opt_files +
+      0;
+  if ($grains > 1) {
+    usage("Only specify one output granularity option");
+  }
+  if ($grains == 0) {
+    $main::opt_functions = 1;
+  }
+
+  # Check output modes
+  my $modes =
+      $main::opt_text +
+      $main::opt_callgrind +
+      ($main::opt_list eq '' ? 0 : 1) +
+      ($main::opt_disasm eq '' ? 0 : 1) +
+      ($main::opt_symbols == 0 ? 0 : 1) +
+      $main::opt_gv +
+      $main::opt_evince +
+      $main::opt_web +
+      $main::opt_dot +
+      $main::opt_ps +
+      $main::opt_pdf +
+      $main::opt_svg +
+      $main::opt_gif +
+      $main::opt_raw +
+      $main::opt_interactive +
+      0;
+  if ($modes > 1) {
+    usage("Only specify one output mode");
+  }
+  if ($modes == 0) {
+    if (-t STDOUT) {  # If STDOUT is a tty, activate interactive mode
+      $main::opt_interactive = 1;
+    } else {
+      $main::opt_text = 1;
+    }
+  }
+
+  if ($main::opt_test) {
+    RunUnitTests();
+    # Should not return
+    exit(1);
+  }
+
+  # Binary name and profile arguments list
+  $main::prog = "";
+  @main::pfile_args = ();
+
+  # Remote profiling without a binary (using $SYMBOL_PAGE instead)
+  if (@ARGV > 0) {
+    if (IsProfileURL($ARGV[0])) {
+      $main::use_symbol_page = 1;
+    } elsif (IsSymbolizedProfileFile($ARGV[0])) {
+      $main::use_symbolized_profile = 1;
+      $main::prog = $UNKNOWN_BINARY;  # will be set later from the profile file
+    }
+  }
+
+  if ($main::use_symbol_page || $main::use_symbolized_profile) {
+    # We don't need a binary!
+    my %disabled = ('--lines' => $main::opt_lines,
+                    '--disasm' => $main::opt_disasm);
+    for my $option (keys %disabled) {
+      usage("$option cannot be used without a binary") if $disabled{$option};
+    }
+    # Set $main::prog later...
+    scalar(@ARGV) || usage("Did not specify profile file");
+  } elsif ($main::opt_symbols) {
+    # --symbols needs a binary-name (to run nm on, etc) but not profiles
+    $main::prog = shift(@ARGV) || usage("Did not specify program");
+  } else {
+    $main::prog = shift(@ARGV) || usage("Did not specify program");
+    scalar(@ARGV) || usage("Did not specify profile file");
+  }
+
+  # Parse profile file/location arguments
+  foreach my $farg (@ARGV) {
+    if ($farg =~ m/(.*)\@([0-9]+)(|\/.*)$/ ) {
+      my $machine = $1;
+      my $num_machines = $2;
+      my $path = $3;
+      for (my $i = 0; $i < $num_machines; $i++) {
+        unshift(@main::pfile_args, "$i.$machine$path");
+      }
+    } else {
+      unshift(@main::pfile_args, $farg);
+    }
+  }
+
+  if ($main::use_symbol_page) {
+    unless (IsProfileURL($main::pfile_args[0])) {
+      error("The first profile should be a remote form to use $SYMBOL_PAGE\n");
+    }
+    CheckSymbolPage();
+    $main::prog = FetchProgramName();
+  } elsif (!$main::use_symbolized_profile) {  # may not need objtools!
+    ConfigureObjTools($main::prog)
+  }
+
+  # Break the opt_lib_prefix into the prefix_list array
+  @prefix_list = split (',', $main::opt_lib_prefix);
+
+  # Remove trailing / from the prefixes, in the list to prevent
+  # searching things like /my/path//lib/mylib.so
+  foreach (@prefix_list) {
+    s|/+$||;
+  }
+}
+
+sub FilterAndPrint {
+  my ($profile, $symbols, $libs, $thread) = @_;
+
+  # Get total data in profile
+  my $total = TotalProfile($profile);
+
+  # Remove uniniteresting stack items
+  $profile = RemoveUninterestingFrames($symbols, $profile);
+
+  # Focus?
+  if ($main::opt_focus ne '') {
+    $profile = FocusProfile($symbols, $profile, $main::opt_focus);
+  }
+
+  # Ignore?
+  if ($main::opt_ignore ne '') {
+    $profile = IgnoreProfile($symbols, $profile, $main::opt_ignore);
+  }
+
+  my $calls = ExtractCalls($symbols, $profile);
+
+  # Reduce profiles to required output granularity, and also clean
+  # each stack trace so a given entry exists at most once.
+  my $reduced = ReduceProfile($symbols, $profile);
+
+  # Get derived profiles
+  my $flat = FlatProfile($reduced);
+  my $cumulative = CumulativeProfile($reduced);
+
+  # Print
+  if (!$main::opt_interactive) {
+    if ($main::opt_disasm) {
+      PrintDisassembly($libs, $flat, $cumulative, $main::opt_disasm);
+    } elsif ($main::opt_list) {
+      PrintListing($total, $libs, $flat, $cumulative, $main::opt_list, 0);
+    } elsif ($main::opt_text) {
+      # Make sure the output is empty when have nothing to report
+      # (only matters when --heapcheck is given but we must be
+      # compatible with old branches that did not pass --heapcheck always):
+      if ($total != 0) {
+        printf("Total%s: %s %s\n",
+               (defined($thread) ? " (t$thread)" : ""),
+               Unparse($total), Units());
+      }
+      PrintText($symbols, $flat, $cumulative, -1);
+    } elsif ($main::opt_raw) {
+      PrintSymbolizedProfile($symbols, $profile, $main::prog);
+    } elsif ($main::opt_callgrind) {
+      PrintCallgrind($calls);
+    } else {
+      if (PrintDot($main::prog, $symbols, $profile, $flat, $cumulative, $total)) {
+        if ($main::opt_gv) {
+          RunGV(TempName($main::next_tmpfile, "ps"), "");
+        } elsif ($main::opt_evince) {
+          RunEvince(TempName($main::next_tmpfile, "pdf"), "");
+        } elsif ($main::opt_web) {
+          my $tmp = TempName($main::next_tmpfile, "svg");
+          RunWeb($tmp);
+          # The command we run might hand the file name off
+          # to an already running browser instance and then exit.
+          # Normally, we'd remove $tmp on exit (right now),
+          # but fork a child to remove $tmp a little later, so that the
+          # browser has time to load it first.
+          delete $main::tempnames{$tmp};
+          if (fork() == 0) {
+            sleep 5;
+            unlink($tmp);
+            exit(0);
+          }
+        }
+      } else {
+        cleanup();
+        exit(1);
+      }
+    }
+  } else {
+    InteractiveMode($profile, $symbols, $libs, $total);
+  }
+}
+
+sub Main() {
+  Init();
+  $main::collected_profile = undef;
+  @main::profile_files = ();
+  $main::op_time = time();
+
+  # Printing symbols is special and requires a lot less info that most.
+  if ($main::opt_symbols) {
+    PrintSymbols(*STDIN);   # Get /proc/maps and symbols output from stdin
+    return;
+  }
+
+  # Fetch all profile data
+  FetchDynamicProfiles();
+
+  # this will hold symbols that we read from the profile files
+  my $symbol_map = {};
+
+  # Read one profile, pick the last item on the list
+  my $data = ReadProfile($main::prog, pop(@main::profile_files));
+  my $profile = $data->{profile};
+  my $pcs = $data->{pcs};
+  my $libs = $data->{libs};   # Info about main program and shared libraries
+  $symbol_map = MergeSymbols($symbol_map, $data->{symbols});
+
+  # Add additional profiles, if available.
+  if (scalar(@main::profile_files) > 0) {
+    foreach my $pname (@main::profile_files) {
+      my $data2 = ReadProfile($main::prog, $pname);
+      $profile = AddProfile($profile, $data2->{profile});
+      $pcs = AddPcs($pcs, $data2->{pcs});
+      $symbol_map = MergeSymbols($symbol_map, $data2->{symbols});
+    }
+  }
+
+  # Subtract base from profile, if specified
+  if ($main::opt_base ne '') {
+    my $base = ReadProfile($main::prog, $main::opt_base);
+    $profile = SubtractProfile($profile, $base->{profile});
+    $pcs = AddPcs($pcs, $base->{pcs});
+    $symbol_map = MergeSymbols($symbol_map, $base->{symbols});
+  }
+
+  # Collect symbols
+  my $symbols;
+  if ($main::use_symbolized_profile) {
+    $symbols = FetchSymbols($pcs, $symbol_map);
+  } elsif ($main::use_symbol_page) {
+    $symbols = FetchSymbols($pcs);
+  } else {
+    # TODO(csilvers): $libs uses the /proc/self/maps data from profile1,
+    # which may differ from the data from subsequent profiles, especially
+    # if they were run on different machines.  Use appropriate libs for
+    # each pc somehow.
+    $symbols = ExtractSymbols($libs, $pcs);
+  }
+
+  if (!defined($main::opt_thread)) {
+    FilterAndPrint($profile, $symbols, $libs);
+  }
+  if (defined($data->{threads})) {
+    foreach my $thread (sort { $a <=> $b } keys(%{$data->{threads}})) {
+      if (defined($main::opt_thread) &&
+          ($main::opt_thread eq '*' || $main::opt_thread == $thread)) {
+        my $thread_profile = $data->{threads}{$thread};
+        FilterAndPrint($thread_profile, $symbols, $libs, $thread);
+      }
+    }
+  }
+
+  cleanup();
+  exit(0);
+}
+
+##### Entry Point #####
+
+Main();
+
+# Temporary code to detect if we're running on a Goobuntu system.
+# These systems don't have the right stuff installed for the special
+# Readline libraries to work, so as a temporary workaround, we default
+# to using the normal stdio code, rather than the fancier readline-based
+# code
+sub ReadlineMightFail {
+  if (-e '/lib/libtermcap.so.2') {
+    return 0;  # libtermcap exists, so readline should be okay
+  } else {
+    return 1;
+  }
+}
+
+sub RunGV {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  if (!system(ShellEscape(@GV, "--version") . " >$dev_null 2>&1")) {
+    # Options using double dash are supported by this gv version.
+    # Also, turn on noantialias to better handle bug in gv for
+    # postscript files with large dimensions.
+    # TODO: Maybe we should not pass the --noantialias flag
+    # if the gv version is known to work properly without the flag.
+    system(ShellEscape(@GV, "--scale=$main::opt_scale", "--noantialias", $fname)
+           . $bg);
+  } else {
+    # Old gv version - only supports options that use single dash.
+    print STDERR ShellEscape(@GV, "-scale", $main::opt_scale) . "\n";
+    system(ShellEscape(@GV, "-scale", "$main::opt_scale", $fname) . $bg);
+  }
+}
+
+sub RunEvince {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  system(ShellEscape(@EVINCE, $fname) . $bg);
+}
+
+sub RunWeb {
+  my $fname = shift;
+  print STDERR "Loading web page file:///$fname\n";
+
+  if (`uname` =~ /Darwin/) {
+    # OS X: open will use standard preference for SVG files.
+    system("/usr/bin/open", $fname);
+    return;
+  }
+
+  # Some kind of Unix; try generic symlinks, then specific browsers.
+  # (Stop once we find one.)
+  # Works best if the browser is already running.
+  my @alt = (
+    "/etc/alternatives/gnome-www-browser",
+    "/etc/alternatives/x-www-browser",
+    "google-chrome",
+    "firefox",
+  );
+  foreach my $b (@alt) {
+    if (system($b, $fname) == 0) {
+      return;
+    }
+  }
+
+  print STDERR "Could not load web browser.\n";
+}
+
+sub RunKcachegrind {
+  my $fname = shift;
+  my $bg = shift;       # "" or " &" if we should run in background
+  print STDERR "Starting '@KCACHEGRIND " . $fname . $bg . "'\n";
+  system(ShellEscape(@KCACHEGRIND, $fname) . $bg);
+}
+
+
+##### Interactive helper routines #####
+
+sub InteractiveMode {
+  $| = 1;  # Make output unbuffered for interactive mode
+  my ($orig_profile, $symbols, $libs, $total) = @_;
+
+  print STDERR "Welcome to jeprof!  For help, type 'help'.\n";
+
+  # Use ReadLine if it's installed and input comes from a console.
+  if ( -t STDIN &&
+       !ReadlineMightFail() &&
+       defined(eval {require Term::ReadLine}) ) {
+    my $term = new Term::ReadLine 'jeprof';
+    while ( defined ($_ = $term->readline('(jeprof) '))) {
+      $term->addhistory($_) if /\S/;
+      if (!InteractiveCommand($orig_profile, $symbols, $libs, $total, $_)) {
+        last;    # exit when we get an interactive command to quit
+      }
+    }
+  } else {       # don't have readline
+    while (1) {
+      print STDERR "(jeprof) ";
+      $_ = <STDIN>;
+      last if ! defined $_ ;
+      s/\r//g;         # turn windows-looking lines into unix-looking lines
+
+      # Save some flags that might be reset by InteractiveCommand()
+      my $save_opt_lines = $main::opt_lines;
+
+      if (!InteractiveCommand($orig_profile, $symbols, $libs, $total, $_)) {
+        last;    # exit when we get an interactive command to quit
+      }
+
+      # Restore flags
+      $main::opt_lines = $save_opt_lines;
+    }
+  }
+}
+
+# Takes two args: orig profile, and command to run.
+# Returns 1 if we should keep going, or 0 if we were asked to quit
+sub InteractiveCommand {
+  my($orig_profile, $symbols, $libs, $total, $command) = @_;
+  $_ = $command;                # just to make future m//'s easier
+  if (!defined($_)) {
+    print STDERR "\n";
+    return 0;
+  }
+  if (m/^\s*quit/) {
+    return 0;
+  }
+  if (m/^\s*help/) {
+    InteractiveHelpMessage();
+    return 1;
+  }
+  # Clear all the mode options -- mode is controlled by "$command"
+  $main::opt_text = 0;
+  $main::opt_callgrind = 0;
+  $main::opt_disasm = 0;
+  $main::opt_list = 0;
+  $main::opt_gv = 0;
+  $main::opt_evince = 0;
+  $main::opt_cum = 0;
+
+  if (m/^\s*(text|top)(\d*)\s*(.*)/) {
+    $main::opt_text = 1;
+
+    my $line_limit = ($2 ne "") ? int($2) : 10;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($3);
+
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintText($symbols, $flat, $cumulative, $line_limit);
+    return 1;
+  }
+  if (m/^\s*callgrind\s*([^ \n]*)/) {
+    $main::opt_callgrind = 1;
+
+    # Get derived profiles
+    my $calls = ExtractCalls($symbols, $orig_profile);
+    my $filename = $1;
+    if ( $1 eq '' ) {
+      $filename = TempName($main::next_tmpfile, "callgrind");
+    }
+    PrintCallgrind($calls, $filename);
+    if ( $1 eq '' ) {
+      RunKcachegrind($filename, " & ");
+      $main::next_tmpfile++;
+    }
+
+    return 1;
+  }
+  if (m/^\s*(web)?list\s*(.+)/) {
+    my $html = (defined($1) && ($1 eq "web"));
+    $main::opt_list = 1;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($2);
+
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintListing($total, $libs, $flat, $cumulative, $routine, $html);
+    return 1;
+  }
+  if (m/^\s*disasm\s*(.+)/) {
+    $main::opt_disasm = 1;
+
+    my $routine;
+    my $ignore;
+    ($routine, $ignore) = ParseInteractiveArgs($1);
+
+    # Process current profile to account for various settings
+    my $profile = ProcessProfile($total, $orig_profile, $symbols, "", $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    PrintDisassembly($libs, $flat, $cumulative, $routine);
+    return 1;
+  }
+  if (m/^\s*(gv|web|evince)\s*(.*)/) {
+    $main::opt_gv = 0;
+    $main::opt_evince = 0;
+    $main::opt_web = 0;
+    if ($1 eq "gv") {
+      $main::opt_gv = 1;
+    } elsif ($1 eq "evince") {
+      $main::opt_evince = 1;
+    } elsif ($1 eq "web") {
+      $main::opt_web = 1;
+    }
+
+    my $focus;
+    my $ignore;
+    ($focus, $ignore) = ParseInteractiveArgs($2);
+
+    # Process current profile to account for various settings
+    my $profile = ProcessProfile($total, $orig_profile, $symbols,
+                                 $focus, $ignore);
+    my $reduced = ReduceProfile($symbols, $profile);
+
+    # Get derived profiles
+    my $flat = FlatProfile($reduced);
+    my $cumulative = CumulativeProfile($reduced);
+
+    if (PrintDot($main::prog, $symbols, $profile, $flat, $cumulative, $total)) {
+      if ($main::opt_gv) {
+        RunGV(TempName($main::next_tmpfile, "ps"), " &");
+      } elsif ($main::opt_evince) {
+        RunEvince(TempName($main::next_tmpfile, "pdf"), " &");
+      } elsif ($main::opt_web) {
+        RunWeb(TempName($main::next_tmpfile, "svg"));
+      }
+      $main::next_tmpfile++;
+    }
+    return 1;
+  }
+  if (m/^\s*$/) {
+    return 1;
+  }
+  print STDERR "Unknown command: try 'help'.\n";
+  return 1;
+}
+
+
+sub ProcessProfile {
+  my $total_count = shift;
+  my $orig_profile = shift;
+  my $symbols = shift;
+  my $focus = shift;
+  my $ignore = shift;
+
+  # Process current profile to account for various settings
+  my $profile = $orig_profile;
+  printf("Total: %s %s\n", Unparse($total_count), Units());
+  if ($focus ne '') {
+    $profile = FocusProfile($symbols, $profile, $focus);
+    my $focus_count = TotalProfile($profile);
+    printf("After focusing on '%s': %s %s of %s (%0.1f%%)\n",
+           $focus,
+           Unparse($focus_count), Units(),
+           Unparse($total_count), ($focus_count*100.0) / $total_count);
+  }
+  if ($ignore ne '') {
+    $profile = IgnoreProfile($symbols, $profile, $ignore);
+    my $ignore_count = TotalProfile($profile);
+    printf("After ignoring '%s': %s %s of %s (%0.1f%%)\n",
+           $ignore,
+           Unparse($ignore_count), Units(),
+           Unparse($total_count),
+           ($ignore_count*100.0) / $total_count);
+  }
+
+  return $profile;
+}
+
+sub InteractiveHelpMessage {
+  print STDERR <<ENDOFHELP;
+Interactive jeprof mode
+
+Commands:
+  gv
+  gv [focus] [-ignore1] [-ignore2]
+      Show graphical hierarchical display of current profile.  Without
+      any arguments, shows all samples in the profile.  With the optional
+      "focus" argument, restricts the samples shown to just those where
+      the "focus" regular expression matches a routine name on the stack
+      trace.
+
+  web
+  web [focus] [-ignore1] [-ignore2]
+      Like GV, but displays profile in your web browser instead of using
+      Ghostview. Works best if your web browser is already running.
+      To change the browser that gets used:
+      On Linux, set the /etc/alternatives/gnome-www-browser symlink.
+      On OS X, change the Finder association for SVG files.
+
+  list [routine_regexp] [-ignore1] [-ignore2]
+      Show source listing of routines whose names match "routine_regexp"
+
+  weblist [routine_regexp] [-ignore1] [-ignore2]
+     Displays a source listing of routines whose names match "routine_regexp"
+     in a web browser.  You can click on source lines to view the
+     corresponding disassembly.
+
+  top [--cum] [-ignore1] [-ignore2]
+  top20 [--cum] [-ignore1] [-ignore2]
+  top37 [--cum] [-ignore1] [-ignore2]
+      Show top lines ordered by flat profile count, or cumulative count
+      if --cum is specified.  If a number is present after 'top', the
+      top K routines will be shown (defaults to showing the top 10)
+
+  disasm [routine_regexp] [-ignore1] [-ignore2]
+      Show disassembly of routines whose names match "routine_regexp",
+      annotated with sample counts.
+
+  callgrind
+  callgrind [filename]
+      Generates callgrind file. If no filename is given, kcachegrind is called.
+
+  help - This listing
+  quit or ^D - End jeprof
+
+For commands that accept optional -ignore tags, samples where any routine in
+the stack trace matches the regular expression in any of the -ignore
+parameters will be ignored.
+
+Further pprof details are available at this location (or one similar):
+
+ /usr/doc/gperftools-$PPROF_VERSION/cpu_profiler.html
+ /usr/doc/gperftools-$PPROF_VERSION/heap_profiler.html
+
+ENDOFHELP
+}
+sub ParseInteractiveArgs {
+  my $args = shift;
+  my $focus = "";
+  my $ignore = "";
+  my @x = split(/ +/, $args);
+  foreach $a (@x) {
+    if ($a =~ m/^(--|-)lines$/) {
+      $main::opt_lines = 1;
+    } elsif ($a =~ m/^(--|-)cum$/) {
+      $main::opt_cum = 1;
+    } elsif ($a =~ m/^-(.*)/) {
+      $ignore .= (($ignore ne "") ? "|" : "" ) . $1;
+    } else {
+      $focus .= (($focus ne "") ? "|" : "" ) . $a;
+    }
+  }
+  if ($ignore ne "") {
+    print STDERR "Ignoring samples in call stacks that match '$ignore'\n";
+  }
+  return ($focus, $ignore);
+}
+
+##### Output code #####
+
+sub TempName {
+  my $fnum = shift;
+  my $ext = shift;
+  my $file = "$main::tmpfile_ps.$fnum.$ext";
+  $main::tempnames{$file} = 1;
+  return $file;
+}
+
+# Print profile data in packed binary format (64-bit) to standard out
+sub PrintProfileData {
+  my $profile = shift;
+
+  # print header (64-bit style)
+  # (zero) (header-size) (version) (sample-period) (zero)
+  print pack('L*', 0, 0, 3, 0, 0, 0, 1, 0, 0, 0);
+
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    if ($#addrs >= 0) {
+      my $depth = $#addrs + 1;
+      # int(foo / 2**32) is the only reliable way to get rid of bottom
+      # 32 bits on both 32- and 64-bit systems.
+      print pack('L*', $count & 0xFFFFFFFF, int($count / 2**32));
+      print pack('L*', $depth & 0xFFFFFFFF, int($depth / 2**32));
+
+      foreach my $full_addr (@addrs) {
+        my $addr = $full_addr;
+        $addr =~ s/0x0*//;  # strip off leading 0x, zeroes
+        if (length($addr) > 16) {
+          print STDERR "Invalid address in profile: $full_addr\n";
+          next;
+        }
+        my $low_addr = substr($addr, -8);       # get last 8 hex chars
+        my $high_addr = substr($addr, -16, 8);  # get up to 8 more hex chars
+        print pack('L*', hex('0x' . $low_addr), hex('0x' . $high_addr));
+      }
+    }
+  }
+}
+
+# Print symbols and profile data
+sub PrintSymbolizedProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $prog = shift;
+
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+
+  print '--- ', $symbol_marker, "\n";
+  if (defined($prog)) {
+    print 'binary=', $prog, "\n";
+  }
+  while (my ($pc, $name) = each(%{$symbols})) {
+    my $sep = ' ';
+    print '0x', $pc;
+    # We have a list of function names, which include the inlined
+    # calls.  They are separated (and terminated) by --, which is
+    # illegal in function names.
+    for (my $j = 2; $j <= $#{$name}; $j += 3) {
+      print $sep, $name->[$j];
+      $sep = '--';
+    }
+    print "\n";
+  }
+  print '---', "\n";
+
+  my $profile_marker;
+  if ($main::profile_type eq 'heap') {
+    $HEAP_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  } elsif ($main::profile_type eq 'growth') {
+    $GROWTH_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  } elsif ($main::profile_type eq 'contention') {
+    $CONTENTION_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  } else { # elsif ($main::profile_type eq 'cpu')
+    $PROFILE_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+    $profile_marker = $&;
+  }
+
+  print '--- ', $profile_marker, "\n";
+  if (defined($main::collected_profile)) {
+    # if used with remote fetch, simply dump the collected profile to output.
+    open(SRC, "<$main::collected_profile");
+    while (<SRC>) {
+      print $_;
+    }
+    close(SRC);
+  } else {
+    # --raw/http: For everything to work correctly for non-remote profiles, we
+    # would need to extend PrintProfileData() to handle all possible profile
+    # types, re-enable the code that is currently disabled in ReadCPUProfile()
+    # and FixCallerAddresses(), and remove the remote profile dumping code in
+    # the block above.
+    die "--raw/http: jeprof can only dump remote profiles for --raw\n";
+    # dump a cpu-format profile to standard out
+    PrintProfileData($profile);
+  }
+}
+
+# Print text output
+sub PrintText {
+  my $symbols = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $line_limit = shift;
+
+  my $total = TotalProfile($flat);
+
+  # Which profile to sort by?
+  my $s = $main::opt_cum ? $cumulative : $flat;
+
+  my $running_sum = 0;
+  my $lines = 0;
+  foreach my $k (sort { GetEntry($s, $b) <=> GetEntry($s, $a) || $a cmp $b }
+                 keys(%{$cumulative})) {
+    my $f = GetEntry($flat, $k);
+    my $c = GetEntry($cumulative, $k);
+    $running_sum += $f;
+
+    my $sym = $k;
+    if (exists($symbols->{$k})) {
+      $sym = $symbols->{$k}->[0] . " " . $symbols->{$k}->[1];
+      if ($main::opt_addresses) {
+        $sym = $k . " " . $sym;
+      }
+    }
+
+    if ($f != 0 || $c != 0) {
+      printf("%8s %6s %6s %8s %6s %s\n",
+             Unparse($f),
+             Percent($f, $total),
+             Percent($running_sum, $total),
+             Unparse($c),
+             Percent($c, $total),
+             $sym);
+    }
+    $lines++;
+    last if ($line_limit >= 0 && $lines >= $line_limit);
+  }
+}
+
+# Callgrind format has a compression for repeated function and file
+# names.  You show the name the first time, and just use its number
+# subsequently.  This can cut down the file to about a third or a
+# quarter of its uncompressed size.  $key and $val are the key/value
+# pair that would normally be printed by callgrind; $map is a map from
+# value to number.
+sub CompressedCGName {
+  my($key, $val, $map) = @_;
+  my $idx = $map->{$val};
+  # For very short keys, providing an index hurts rather than helps.
+  if (length($val) <= 3) {
+    return "$key=$val\n";
+  } elsif (defined($idx)) {
+    return "$key=($idx)\n";
+  } else {
+    # scalar(keys $map) gives the number of items in the map.
+    $idx = scalar(keys(%{$map})) + 1;
+    $map->{$val} = $idx;
+    return "$key=($idx) $val\n";
+  }
+}
+
+# Print the call graph in a way that's suiteable for callgrind.
+sub PrintCallgrind {
+  my $calls = shift;
+  my $filename;
+  my %filename_to_index_map;
+  my %fnname_to_index_map;
+
+  if ($main::opt_interactive) {
+    $filename = shift;
+    print STDERR "Writing callgrind file to '$filename'.\n"
+  } else {
+    $filename = "&STDOUT";
+  }
+  open(CG, ">$filename");
+  printf CG ("events: Hits\n\n");
+  foreach my $call ( map { $_->[0] }
+                     sort { $a->[1] cmp $b ->[1] ||
+                            $a->[2] <=> $b->[2] }
+                     map { /([^:]+):(\d+):([^ ]+)( -> ([^:]+):(\d+):(.+))?/;
+                           [$_, $1, $2] }
+                     keys %$calls ) {
+    my $count = int($calls->{$call});
+    $call =~ /([^:]+):(\d+):([^ ]+)( -> ([^:]+):(\d+):(.+))?/;
+    my ( $caller_file, $caller_line, $caller_function,
+         $callee_file, $callee_line, $callee_function ) =
+       ( $1, $2, $3, $5, $6, $7 );
+
+    # TODO(csilvers): for better compression, collect all the
+    # caller/callee_files and functions first, before printing
+    # anything, and only compress those referenced more than once.
+    printf CG CompressedCGName("fl", $caller_file, \%filename_to_index_map);
+    printf CG CompressedCGName("fn", $caller_function, \%fnname_to_index_map);
+    if (defined $6) {
+      printf CG CompressedCGName("cfl", $callee_file, \%filename_to_index_map);
+      printf CG CompressedCGName("cfn", $callee_function, \%fnname_to_index_map);
+      printf CG ("calls=$count $callee_line\n");
+    }
+    printf CG ("$caller_line $count\n\n");
+  }
+}
+
+# Print disassembly for all all routines that match $main::opt_disasm
+sub PrintDisassembly {
+  my $libs = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $disasm_opts = shift;
+
+  my $total = TotalProfile($flat);
+
+  foreach my $lib (@{$libs}) {
+    my $symbol_table = GetProcedureBoundaries($lib->[0], $disasm_opts);
+    my $offset = AddressSub($lib->[1], $lib->[3]);
+    foreach my $routine (sort ByName keys(%{$symbol_table})) {
+      my $start_addr = $symbol_table->{$routine}->[0];
+      my $end_addr = $symbol_table->{$routine}->[1];
+      # See if there are any samples in this routine
+      my $length = hex(AddressSub($end_addr, $start_addr));
+      my $addr = AddressAdd($start_addr, $offset);
+      for (my $i = 0; $i < $length; $i++) {
+        if (defined($cumulative->{$addr})) {
+          PrintDisassembledFunction($lib->[0], $offset,
+                                    $routine, $flat, $cumulative,
+                                    $start_addr, $end_addr, $total);
+          last;
+        }
+        $addr = AddressInc($addr);
+      }
+    }
+  }
+}
+
+# Return reference to array of tuples of the form:
+#       [start_address, filename, linenumber, instruction, limit_address]
+# E.g.,
+#       ["0x806c43d", "/foo/bar.cc", 131, "ret", "0x806c440"]
+sub Disassemble {
+  my $prog = shift;
+  my $offset = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+
+  my $objdump = $obj_tool_map{"objdump"};
+  my $cmd = ShellEscape($objdump, "-C", "-d", "-l", "--no-show-raw-insn",
+                        "--start-address=0x$start_addr",
+                        "--stop-address=0x$end_addr", $prog);
+  open(OBJDUMP, "$cmd |") || error("$cmd: $!\n");
+  my @result = ();
+  my $filename = "";
+  my $linenumber = -1;
+  my $last = ["", "", "", ""];
+  while (<OBJDUMP>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    chop;
+    if (m|\s*([^:\s]+):(\d+)\s*$|) {
+      # Location line of the form:
+      #   <filename>:<linenumber>
+      $filename = $1;
+      $linenumber = $2;
+    } elsif (m/^ +([0-9a-f]+):\s*(.*)/) {
+      # Disassembly line -- zero-extend address to full length
+      my $addr = HexExtend($1);
+      my $k = AddressAdd($addr, $offset);
+      $last->[4] = $k;   # Store ending address for previous instruction
+      $last = [$k, $filename, $linenumber, $2, $end_addr];
+      push(@result, $last);
+    }
+  }
+  close(OBJDUMP);
+  return @result;
+}
+
+# The input file should contain lines of the form /proc/maps-like
+# output (same format as expected from the profiles) or that looks
+# like hex addresses (like "0xDEADBEEF").  We will parse all
+# /proc/maps output, and for all the hex addresses, we will output
+# "short" symbol names, one per line, in the same order as the input.
+sub PrintSymbols {
+  my $maps_and_symbols_file = shift;
+
+  # ParseLibraries expects pcs to be in a set.  Fine by us...
+  my @pclist = ();   # pcs in sorted order
+  my $pcs = {};
+  my $map = "";
+  foreach my $line (<$maps_and_symbols_file>) {
+    $line =~ s/\r//g;    # turn windows-looking lines into unix-looking lines
+    if ($line =~ /\b(0x[0-9a-f]+)\b/i) {
+      push(@pclist, HexExtend($1));
+      $pcs->{$pclist[-1]} = 1;
+    } else {
+      $map .= $line;
+    }
+  }
+
+  my $libs = ParseLibraries($main::prog, $map, $pcs);
+  my $symbols = ExtractSymbols($libs, $pcs);
+
+  foreach my $pc (@pclist) {
+    # ->[0] is the shortname, ->[2] is the full name
+    print(($symbols->{$pc}->[0] || "??") . "\n");
+  }
+}
+
+
+# For sorting functions by name
+sub ByName {
+  return ShortFunctionName($a) cmp ShortFunctionName($b);
+}
+
+# Print source-listing for all all routines that match $list_opts
+sub PrintListing {
+  my $total = shift;
+  my $libs = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $list_opts = shift;
+  my $html = shift;
+
+  my $output = \*STDOUT;
+  my $fname = "";
+
+  if ($html) {
+    # Arrange to write the output to a temporary file
+    $fname = TempName($main::next_tmpfile, "html");
+    $main::next_tmpfile++;
+    if (!open(TEMP, ">$fname")) {
+      print STDERR "$fname: $!\n";
+      return;
+    }
+    $output = \*TEMP;
+    print $output HtmlListingHeader();
+    printf $output ("<div class=\"legend\">%s<br>Total: %s %s</div>\n",
+                    $main::prog, Unparse($total), Units());
+  }
+
+  my $listed = 0;
+  foreach my $lib (@{$libs}) {
+    my $symbol_table = GetProcedureBoundaries($lib->[0], $list_opts);
+    my $offset = AddressSub($lib->[1], $lib->[3]);
+    foreach my $routine (sort ByName keys(%{$symbol_table})) {
+      # Print if there are any samples in this routine
+      my $start_addr = $symbol_table->{$routine}->[0];
+      my $end_addr = $symbol_table->{$routine}->[1];
+      my $length = hex(AddressSub($end_addr, $start_addr));
+      my $addr = AddressAdd($start_addr, $offset);
+      for (my $i = 0; $i < $length; $i++) {
+        if (defined($cumulative->{$addr})) {
+          $listed += PrintSource(
+            $lib->[0], $offset,
+            $routine, $flat, $cumulative,
+            $start_addr, $end_addr,
+            $html,
+            $output);
+          last;
+        }
+        $addr = AddressInc($addr);
+      }
+    }
+  }
+
+  if ($html) {
+    if ($listed > 0) {
+      print $output HtmlListingFooter();
+      close($output);
+      RunWeb($fname);
+    } else {
+      close($output);
+      unlink($fname);
+    }
+  }
+}
+
+sub HtmlListingHeader {
+  return <<'EOF';
+<DOCTYPE html>
+<html>
+<head>
+<title>Pprof listing</title>
+<style type="text/css">
+body {
+  font-family: sans-serif;
+}
+h1 {
+  font-size: 1.5em;
+  margin-bottom: 4px;
+}
+.legend {
+  font-size: 1.25em;
+}
+.line {
+  color: #aaaaaa;
+}
+.nop {
+  color: #aaaaaa;
+}
+.unimportant {
+  color: #cccccc;
+}
+.disasmloc {
+  color: #000000;
+}
+.deadsrc {
+  cursor: pointer;
+}
+.deadsrc:hover {
+  background-color: #eeeeee;
+}
+.livesrc {
+  color: #0000ff;
+  cursor: pointer;
+}
+.livesrc:hover {
+  background-color: #eeeeee;
+}
+.asm {
+  color: #008800;
+  display: none;
+}
+</style>
+<script type="text/javascript">
+function jeprof_toggle_asm(e) {
+  var target;
+  if (!e) e = window.event;
+  if (e.target) target = e.target;
+  else if (e.srcElement) target = e.srcElement;
+
+  if (target) {
+    var asm = target.nextSibling;
+    if (asm && asm.className == "asm") {
+      asm.style.display = (asm.style.display == "block" ? "" : "block");
+      e.preventDefault();
+      return false;
+    }
+  }
+}
+</script>
+</head>
+<body>
+EOF
+}
+
+sub HtmlListingFooter {
+  return <<'EOF';
+</body>
+</html>
+EOF
+}
+
+sub HtmlEscape {
+  my $text = shift;
+  $text =~ s/&/&amp;/g;
+  $text =~ s/</&lt;/g;
+  $text =~ s/>/&gt;/g;
+  return $text;
+}
+
+# Returns the indentation of the line, if it has any non-whitespace
+# characters.  Otherwise, returns -1.
+sub Indentation {
+  my $line = shift;
+  if (m/^(\s*)\S/) {
+    return length($1);
+  } else {
+    return -1;
+  }
+}
+
+# If the symbol table contains inlining info, Disassemble() may tag an
+# instruction with a location inside an inlined function.  But for
+# source listings, we prefer to use the location in the function we
+# are listing.  So use MapToSymbols() to fetch full location
+# information for each instruction and then pick out the first
+# location from a location list (location list contains callers before
+# callees in case of inlining).
+#
+# After this routine has run, each entry in $instructions contains:
+#   [0] start address
+#   [1] filename for function we are listing
+#   [2] line number for function we are listing
+#   [3] disassembly
+#   [4] limit address
+#   [5] most specific filename (may be different from [1] due to inlining)
+#   [6] most specific line number (may be different from [2] due to inlining)
+sub GetTopLevelLineNumbers {
+  my ($lib, $offset, $instructions) = @_;
+  my $pcs = [];
+  for (my $i = 0; $i <= $#{$instructions}; $i++) {
+    push(@{$pcs}, $instructions->[$i]->[0]);
+  }
+  my $symbols = {};
+  MapToSymbols($lib, $offset, $pcs, $symbols);
+  for (my $i = 0; $i <= $#{$instructions}; $i++) {
+    my $e = $instructions->[$i];
+    push(@{$e}, $e->[1]);
+    push(@{$e}, $e->[2]);
+    my $addr = $e->[0];
+    my $sym = $symbols->{$addr};
+    if (defined($sym)) {
+      if ($#{$sym} >= 2 && $sym->[1] =~ m/^(.*):(\d+)$/) {
+        $e->[1] = $1;  # File name
+        $e->[2] = $2;  # Line number
+      }
+    }
+  }
+}
+
+# Print source-listing for one routine
+sub PrintSource {
+  my $prog = shift;
+  my $offset = shift;
+  my $routine = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+  my $html = shift;
+  my $output = shift;
+
+  # Disassemble all instructions (just to get line numbers)
+  my @instructions = Disassemble($prog, $offset, $start_addr, $end_addr);
+  GetTopLevelLineNumbers($prog, $offset, \@instructions);
+
+  # Hack 1: assume that the first source file encountered in the
+  # disassembly contains the routine
+  my $filename = undef;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    if ($instructions[$i]->[2] >= 0) {
+      $filename = $instructions[$i]->[1];
+      last;
+    }
+  }
+  if (!defined($filename)) {
+    print STDERR "no filename found in $routine\n";
+    return 0;
+  }
+
+  # Hack 2: assume that the largest line number from $filename is the
+  # end of the procedure.  This is typically safe since if P1 contains
+  # an inlined call to P2, then P2 usually occurs earlier in the
+  # source file.  If this does not work, we might have to compute a
+  # density profile or just print all regions we find.
+  my $lastline = 0;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    my $f = $instructions[$i]->[1];
+    my $l = $instructions[$i]->[2];
+    if (($f eq $filename) && ($l > $lastline)) {
+      $lastline = $l;
+    }
+  }
+
+  # Hack 3: assume the first source location from "filename" is the start of
+  # the source code.
+  my $firstline = 1;
+  for (my $i = 0; $i <= $#instructions; $i++) {
+    if ($instructions[$i]->[1] eq $filename) {
+      $firstline = $instructions[$i]->[2];
+      last;
+    }
+  }
+
+  # Hack 4: Extend last line forward until its indentation is less than
+  # the indentation we saw on $firstline
+  my $oldlastline = $lastline;
+  {
+    if (!open(FILE, "<$filename")) {
+      print STDERR "$filename: $!\n";
+      return 0;
+    }
+    my $l = 0;
+    my $first_indentation = -1;
+    while (<FILE>) {
+      s/\r//g;         # turn windows-looking lines into unix-looking lines
+      $l++;
+      my $indent = Indentation($_);
+      if ($l >= $firstline) {
+        if ($first_indentation < 0 && $indent >= 0) {
+          $first_indentation = $indent;
+          last if ($first_indentation == 0);
+        }
+      }
+      if ($l >= $lastline && $indent >= 0) {
+        if ($indent >= $first_indentation) {
+          $lastline = $l+1;
+        } else {
+          last;
+        }
+      }
+    }
+    close(FILE);
+  }
+
+  # Assign all samples to the range $firstline,$lastline,
+  # Hack 4: If an instruction does not occur in the range, its samples
+  # are moved to the next instruction that occurs in the range.
+  my $samples1 = {};        # Map from line number to flat count
+  my $samples2 = {};        # Map from line number to cumulative count
+  my $running1 = 0;         # Unassigned flat counts
+  my $running2 = 0;         # Unassigned cumulative counts
+  my $total1 = 0;           # Total flat counts
+  my $total2 = 0;           # Total cumulative counts
+  my %disasm = ();          # Map from line number to disassembly
+  my $running_disasm = "";  # Unassigned disassembly
+  my $skip_marker = "---\n";
+  if ($html) {
+    $skip_marker = "";
+    for (my $l = $firstline; $l <= $lastline; $l++) {
+      $disasm{$l} = "";
+    }
+  }
+  my $last_dis_filename = '';
+  my $last_dis_linenum = -1;
+  my $last_touched_line = -1;  # To detect gaps in disassembly for a line
+  foreach my $e (@instructions) {
+    # Add up counts for all address that fall inside this instruction
+    my $c1 = 0;
+    my $c2 = 0;
+    for (my $a = $e->[0]; $a lt $e->[4]; $a = AddressInc($a)) {
+      $c1 += GetEntry($flat, $a);
+      $c2 += GetEntry($cumulative, $a);
+    }
+
+    if ($html) {
+      my $dis = sprintf("      %6s %6s \t\t%8s: %s ",
+                        HtmlPrintNumber($c1),
+                        HtmlPrintNumber($c2),
+                        UnparseAddress($offset, $e->[0]),
+                        CleanDisassembly($e->[3]));
+
+      # Append the most specific source line associated with this instruction
+      if (length($dis) < 80) { $dis .= (' ' x (80 - length($dis))) };
+      $dis = HtmlEscape($dis);
+      my $f = $e->[5];
+      my $l = $e->[6];
+      if ($f ne $last_dis_filename) {
+        $dis .= sprintf("<span class=disasmloc>%s:%d</span>",
+                        HtmlEscape(CleanFileName($f)), $l);
+      } elsif ($l ne $last_dis_linenum) {
+        # De-emphasize the unchanged file name portion
+        $dis .= sprintf("<span class=unimportant>%s</span>" .
+                        "<span class=disasmloc>:%d</span>",
+                        HtmlEscape(CleanFileName($f)), $l);
+      } else {
+        # De-emphasize the entire location
+        $dis .= sprintf("<span class=unimportant>%s:%d</span>",
+                        HtmlEscape(CleanFileName($f)), $l);
+      }
+      $last_dis_filename = $f;
+      $last_dis_linenum = $l;
+      $running_disasm .= $dis;
+      $running_disasm .= "\n";
+    }
+
+    $running1 += $c1;
+    $running2 += $c2;
+    $total1 += $c1;
+    $total2 += $c2;
+    my $file = $e->[1];
+    my $line = $e->[2];
+    if (($file eq $filename) &&
+        ($line >= $firstline) &&
+        ($line <= $lastline)) {
+      # Assign all accumulated samples to this line
+      AddEntry($samples1, $line, $running1);
+      AddEntry($samples2, $line, $running2);
+      $running1 = 0;
+      $running2 = 0;
+      if ($html) {
+        if ($line != $last_touched_line && $disasm{$line} ne '') {
+          $disasm{$line} .= "\n";
+        }
+        $disasm{$line} .= $running_disasm;
+        $running_disasm = '';
+        $last_touched_line = $line;
+      }
+    }
+  }
+
+  # Assign any leftover samples to $lastline
+  AddEntry($samples1, $lastline, $running1);
+  AddEntry($samples2, $lastline, $running2);
+  if ($html) {
+    if ($lastline != $last_touched_line && $disasm{$lastline} ne '') {
+      $disasm{$lastline} .= "\n";
+    }
+    $disasm{$lastline} .= $running_disasm;
+  }
+
+  if ($html) {
+    printf $output (
+      "<h1>%s</h1>%s\n<pre onClick=\"jeprof_toggle_asm()\">\n" .
+      "Total:%6s %6s (flat / cumulative %s)\n",
+      HtmlEscape(ShortFunctionName($routine)),
+      HtmlEscape(CleanFileName($filename)),
+      Unparse($total1),
+      Unparse($total2),
+      Units());
+  } else {
+    printf $output (
+      "ROUTINE ====================== %s in %s\n" .
+      "%6s %6s Total %s (flat / cumulative)\n",
+      ShortFunctionName($routine),
+      CleanFileName($filename),
+      Unparse($total1),
+      Unparse($total2),
+      Units());
+  }
+  if (!open(FILE, "<$filename")) {
+    print STDERR "$filename: $!\n";
+    return 0;
+  }
+  my $l = 0;
+  while (<FILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    $l++;
+    if ($l >= $firstline - 5 &&
+        (($l <= $oldlastline + 5) || ($l <= $lastline))) {
+      chop;
+      my $text = $_;
+      if ($l == $firstline) { print $output $skip_marker; }
+      my $n1 = GetEntry($samples1, $l);
+      my $n2 = GetEntry($samples2, $l);
+      if ($html) {
+        # Emit a span that has one of the following classes:
+        #    livesrc -- has samples
+        #    deadsrc -- has disassembly, but with no samples
+        #    nop     -- has no matching disasembly
+        # Also emit an optional span containing disassembly.
+        my $dis = $disasm{$l};
+        my $asm = "";
+        if (defined($dis) && $dis ne '') {
+          $asm = "<span class=\"asm\">" . $dis . "</span>";
+        }
+        my $source_class = (($n1 + $n2 > 0)
+                            ? "livesrc"
+                            : (($asm ne "") ? "deadsrc" : "nop"));
+        printf $output (
+          "<span class=\"line\">%5d</span> " .
+          "<span class=\"%s\">%6s %6s %s</span>%s\n",
+          $l, $source_class,
+          HtmlPrintNumber($n1),
+          HtmlPrintNumber($n2),
+          HtmlEscape($text),
+          $asm);
+      } else {
+        printf $output(
+          "%6s %6s %4d: %s\n",
+          UnparseAlt($n1),
+          UnparseAlt($n2),
+          $l,
+          $text);
+      }
+      if ($l == $lastline)  { print $output $skip_marker; }
+    };
+  }
+  close(FILE);
+  if ($html) {
+    print $output "</pre>\n";
+  }
+  return 1;
+}
+
+# Return the source line for the specified file/linenumber.
+# Returns undef if not found.
+sub SourceLine {
+  my $file = shift;
+  my $line = shift;
+
+  # Look in cache
+  if (!defined($main::source_cache{$file})) {
+    if (100 < scalar keys(%main::source_cache)) {
+      # Clear the cache when it gets too big
+      $main::source_cache = ();
+    }
+
+    # Read all lines from the file
+    if (!open(FILE, "<$file")) {
+      print STDERR "$file: $!\n";
+      $main::source_cache{$file} = [];  # Cache the negative result
+      return undef;
+    }
+    my $lines = [];
+    push(@{$lines}, "");        # So we can use 1-based line numbers as indices
+    while (<FILE>) {
+      push(@{$lines}, $_);
+    }
+    close(FILE);
+
+    # Save the lines in the cache
+    $main::source_cache{$file} = $lines;
+  }
+
+  my $lines = $main::source_cache{$file};
+  if (($line < 0) || ($line > $#{$lines})) {
+    return undef;
+  } else {
+    return $lines->[$line];
+  }
+}
+
+# Print disassembly for one routine with interspersed source if available
+sub PrintDisassembledFunction {
+  my $prog = shift;
+  my $offset = shift;
+  my $routine = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $start_addr = shift;
+  my $end_addr = shift;
+  my $total = shift;
+
+  # Disassemble all instructions
+  my @instructions = Disassemble($prog, $offset, $start_addr, $end_addr);
+
+  # Make array of counts per instruction
+  my @flat_count = ();
+  my @cum_count = ();
+  my $flat_total = 0;
+  my $cum_total = 0;
+  foreach my $e (@instructions) {
+    # Add up counts for all address that fall inside this instruction
+    my $c1 = 0;
+    my $c2 = 0;
+    for (my $a = $e->[0]; $a lt $e->[4]; $a = AddressInc($a)) {
+      $c1 += GetEntry($flat, $a);
+      $c2 += GetEntry($cumulative, $a);
+    }
+    push(@flat_count, $c1);
+    push(@cum_count, $c2);
+    $flat_total += $c1;
+    $cum_total += $c2;
+  }
+
+  # Print header with total counts
+  printf("ROUTINE ====================== %s\n" .
+         "%6s %6s %s (flat, cumulative) %.1f%% of total\n",
+         ShortFunctionName($routine),
+         Unparse($flat_total),
+         Unparse($cum_total),
+         Units(),
+         ($cum_total * 100.0) / $total);
+
+  # Process instructions in order
+  my $current_file = "";
+  for (my $i = 0; $i <= $#instructions; ) {
+    my $e = $instructions[$i];
+
+    # Print the new file name whenever we switch files
+    if ($e->[1] ne $current_file) {
+      $current_file = $e->[1];
+      my $fname = $current_file;
+      $fname =~ s|^\./||;   # Trim leading "./"
+
+      # Shorten long file names
+      if (length($fname) >= 58) {
+        $fname = "..." . substr($fname, -55);
+      }
+      printf("-------------------- %s\n", $fname);
+    }
+
+    # TODO: Compute range of lines to print together to deal with
+    # small reorderings.
+    my $first_line = $e->[2];
+    my $last_line = $first_line;
+    my %flat_sum = ();
+    my %cum_sum = ();
+    for (my $l = $first_line; $l <= $last_line; $l++) {
+      $flat_sum{$l} = 0;
+      $cum_sum{$l} = 0;
+    }
+
+    # Find run of instructions for this range of source lines
+    my $first_inst = $i;
+    while (($i <= $#instructions) &&
+           ($instructions[$i]->[2] >= $first_line) &&
+           ($instructions[$i]->[2] <= $last_line)) {
+      $e = $instructions[$i];
+      $flat_sum{$e->[2]} += $flat_count[$i];
+      $cum_sum{$e->[2]} += $cum_count[$i];
+      $i++;
+    }
+    my $last_inst = $i - 1;
+
+    # Print source lines
+    for (my $l = $first_line; $l <= $last_line; $l++) {
+      my $line = SourceLine($current_file, $l);
+      if (!defined($line)) {
+        $line = "?\n";
+        next;
+      } else {
+        $line =~ s/^\s+//;
+      }
+      printf("%6s %6s %5d: %s",
+             UnparseAlt($flat_sum{$l}),
+             UnparseAlt($cum_sum{$l}),
+             $l,
+             $line);
+    }
+
+    # Print disassembly
+    for (my $x = $first_inst; $x <= $last_inst; $x++) {
+      my $e = $instructions[$x];
+      printf("%6s %6s    %8s: %6s\n",
+             UnparseAlt($flat_count[$x]),
+             UnparseAlt($cum_count[$x]),
+             UnparseAddress($offset, $e->[0]),
+             CleanDisassembly($e->[3]));
+    }
+  }
+}
+
+# Print DOT graph
+sub PrintDot {
+  my $prog = shift;
+  my $symbols = shift;
+  my $raw = shift;
+  my $flat = shift;
+  my $cumulative = shift;
+  my $overall_total = shift;
+
+  # Get total
+  my $local_total = TotalProfile($flat);
+  my $nodelimit = int($main::opt_nodefraction * $local_total);
+  my $edgelimit = int($main::opt_edgefraction * $local_total);
+  my $nodecount = $main::opt_nodecount;
+
+  # Find nodes to include
+  my @list = (sort { abs(GetEntry($cumulative, $b)) <=>
+                     abs(GetEntry($cumulative, $a))
+                     || $a cmp $b }
+              keys(%{$cumulative}));
+  my $last = $nodecount - 1;
+  if ($last > $#list) {
+    $last = $#list;
+  }
+  while (($last >= 0) &&
+         (abs(GetEntry($cumulative, $list[$last])) <= $nodelimit)) {
+    $last--;
+  }
+  if ($last < 0) {
+    print STDERR "No nodes to print\n";
+    return 0;
+  }
+
+  if ($nodelimit > 0 || $edgelimit > 0) {
+    printf STDERR ("Dropping nodes with <= %s %s; edges with <= %s abs(%s)\n",
+                   Unparse($nodelimit), Units(),
+                   Unparse($edgelimit), Units());
+  }
+
+  # Open DOT output file
+  my $output;
+  my $escaped_dot = ShellEscape(@DOT);
+  my $escaped_ps2pdf = ShellEscape(@PS2PDF);
+  if ($main::opt_gv) {
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "ps"));
+    $output = "| $escaped_dot -Tps2 >$escaped_outfile";
+  } elsif ($main::opt_evince) {
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "pdf"));
+    $output = "| $escaped_dot -Tps2 | $escaped_ps2pdf - $escaped_outfile";
+  } elsif ($main::opt_ps) {
+    $output = "| $escaped_dot -Tps2";
+  } elsif ($main::opt_pdf) {
+    $output = "| $escaped_dot -Tps2 | $escaped_ps2pdf - -";
+  } elsif ($main::opt_web || $main::opt_svg) {
+    # We need to post-process the SVG, so write to a temporary file always.
+    my $escaped_outfile = ShellEscape(TempName($main::next_tmpfile, "svg"));
+    $output = "| $escaped_dot -Tsvg >$escaped_outfile";
+  } elsif ($main::opt_gif) {
+    $output = "| $escaped_dot -Tgif";
+  } else {
+    $output = ">&STDOUT";
+  }
+  open(DOT, $output) || error("$output: $!\n");
+
+  # Title
+  printf DOT ("digraph \"%s; %s %s\" {\n",
+              $prog,
+              Unparse($overall_total),
+              Units());
+  if ($main::opt_pdf) {
+    # The output is more printable if we set the page size for dot.
+    printf DOT ("size=\"8,11\"\n");
+  }
+  printf DOT ("node [width=0.375,height=0.25];\n");
+
+  # Print legend
+  printf DOT ("Legend [shape=box,fontsize=24,shape=plaintext," .
+              "label=\"%s\\l%s\\l%s\\l%s\\l%s\\l\"];\n",
+              $prog,
+              sprintf("Total %s: %s", Units(), Unparse($overall_total)),
+              sprintf("Focusing on: %s", Unparse($local_total)),
+              sprintf("Dropped nodes with <= %s abs(%s)",
+                      Unparse($nodelimit), Units()),
+              sprintf("Dropped edges with <= %s %s",
+                      Unparse($edgelimit), Units())
+              );
+
+  # Print nodes
+  my %node = ();
+  my $nextnode = 1;
+  foreach my $a (@list[0..$last]) {
+    # Pick font size
+    my $f = GetEntry($flat, $a);
+    my $c = GetEntry($cumulative, $a);
+
+    my $fs = 8;
+    if ($local_total > 0) {
+      $fs = 8 + (50.0 * sqrt(abs($f * 1.0 / $local_total)));
+    }
+
+    $node{$a} = $nextnode++;
+    my $sym = $a;
+    $sym =~ s/\s+/\\n/g;
+    $sym =~ s/::/\\n/g;
+
+    # Extra cumulative info to print for non-leaves
+    my $extra = "";
+    if ($f != $c) {
+      $extra = sprintf("\\rof %s (%s)",
+                       Unparse($c),
+                       Percent($c, $local_total));
+    }
+    my $style = "";
+    if ($main::opt_heapcheck) {
+      if ($f > 0) {
+        # make leak-causing nodes more visible (add a background)
+        $style = ",style=filled,fillcolor=gray"
+      } elsif ($f < 0) {
+        # make anti-leak-causing nodes (which almost never occur)
+        # stand out as well (triple border)
+        $style = ",peripheries=3"
+      }
+    }
+
+    printf DOT ("N%d [label=\"%s\\n%s (%s)%s\\r" .
+                "\",shape=box,fontsize=%.1f%s];\n",
+                $node{$a},
+                $sym,
+                Unparse($f),
+                Percent($f, $local_total),
+                $extra,
+                $fs,
+                $style,
+               );
+  }
+
+  # Get edges and counts per edge
+  my %edge = ();
+  my $n;
+  my $fullname_to_shortname_map = {};
+  FillFullnameToShortnameMap($symbols, $fullname_to_shortname_map);
+  foreach my $k (keys(%{$raw})) {
+    # TODO: omit low %age edges
+    $n = $raw->{$k};
+    my @translated = TranslateStack($symbols, $fullname_to_shortname_map, $k);
+    for (my $i = 1; $i <= $#translated; $i++) {
+      my $src = $translated[$i];
+      my $dst = $translated[$i-1];
+      #next if ($src eq $dst);  # Avoid self-edges?
+      if (exists($node{$src}) && exists($node{$dst})) {
+        my $edge_label = "$src\001$dst";
+        if (!exists($edge{$edge_label})) {
+          $edge{$edge_label} = 0;
+        }
+        $edge{$edge_label} += $n;
+      }
+    }
+  }
+
+  # Print edges (process in order of decreasing counts)
+  my %indegree = ();   # Number of incoming edges added per node so far
+  my %outdegree = ();  # Number of outgoing edges added per node so far
+  foreach my $e (sort { $edge{$b} <=> $edge{$a} } keys(%edge)) {
+    my @x = split(/\001/, $e);
+    $n = $edge{$e};
+
+    # Initialize degree of kept incoming and outgoing edges if necessary
+    my $src = $x[0];
+    my $dst = $x[1];
+    if (!exists($outdegree{$src})) { $outdegree{$src} = 0; }
+    if (!exists($indegree{$dst})) { $indegree{$dst} = 0; }
+
+    my $keep;
+    if ($indegree{$dst} == 0) {
+      # Keep edge if needed for reachability
+      $keep = 1;
+    } elsif (abs($n) <= $edgelimit) {
+      # Drop if we are below --edgefraction
+      $keep = 0;
+    } elsif ($outdegree{$src} >= $main::opt_maxdegree ||
+             $indegree{$dst} >= $main::opt_maxdegree) {
+      # Keep limited number of in/out edges per node
+      $keep = 0;
+    } else {
+      $keep = 1;
+    }
+
+    if ($keep) {
+      $outdegree{$src}++;
+      $indegree{$dst}++;
+
+      # Compute line width based on edge count
+      my $fraction = abs($local_total ? (3 * ($n / $local_total)) : 0);
+      if ($fraction > 1) { $fraction = 1; }
+      my $w = $fraction * 2;
+      if ($w < 1 && ($main::opt_web || $main::opt_svg)) {
+        # SVG output treats line widths < 1 poorly.
+        $w = 1;
+      }
+
+      # Dot sometimes segfaults if given edge weights that are too large, so
+      # we cap the weights at a large value
+      my $edgeweight = abs($n) ** 0.7;
+      if ($edgeweight > 100000) { $edgeweight = 100000; }
+      $edgeweight = int($edgeweight);
+
+      my $style = sprintf("setlinewidth(%f)", $w);
+      if ($x[1] =~ m/\(inline\)/) {
+        $style .= ",dashed";
+      }
+
+      # Use a slightly squashed function of the edge count as the weight
+      printf DOT ("N%s -> N%s [label=%s, weight=%d, style=\"%s\"];\n",
+                  $node{$x[0]},
+                  $node{$x[1]},
+                  Unparse($n),
+                  $edgeweight,
+                  $style);
+    }
+  }
+
+  print DOT ("}\n");
+  close(DOT);
+
+  if ($main::opt_web || $main::opt_svg) {
+    # Rewrite SVG to be more usable inside web browser.
+    RewriteSvg(TempName($main::next_tmpfile, "svg"));
+  }
+
+  return 1;
+}
+
+sub RewriteSvg {
+  my $svgfile = shift;
+
+  open(SVG, $svgfile) || die "open temp svg: $!";
+  my @svg = <SVG>;
+  close(SVG);
+  unlink $svgfile;
+  my $svg = join('', @svg);
+
+  # Dot's SVG output is
+  #
+  #    <svg width="___" height="___"
+  #     viewBox="___" xmlns=...>
+  #    <g id="graph0" transform="...">
+  #    ...
+  #    </g>
+  #    </svg>
+  #
+  # Change it to
+  #
+  #    <svg width="100%" height="100%"
+  #     xmlns=...>
+  #    $svg_javascript
+  #    <g id="viewport" transform="translate(0,0)">
+  #    <g id="graph0" transform="...">
+  #    ...
+  #    </g>
+  #    </g>
+  #    </svg>
+
+  # Fix width, height; drop viewBox.
+  $svg =~ s/(?s)<svg width="[^"]+" height="[^"]+"(.*?)viewBox="[^"]+"/<svg width="100%" height="100%"$1/;
+
+  # Insert script, viewport <g> above first <g>
+  my $svg_javascript = SvgJavascript();
+  my $viewport = "<g id=\"viewport\" transform=\"translate(0,0)\">\n";
+  $svg =~ s/<g id="graph\d"/$svg_javascript$viewport$&/;
+
+  # Insert final </g> above </svg>.
+  $svg =~ s/(.*)(<\/svg>)/$1<\/g>$2/;
+  $svg =~ s/<g id="graph\d"(.*?)/<g id="viewport"$1/;
+
+  if ($main::opt_svg) {
+    # --svg: write to standard output.
+    print $svg;
+  } else {
+    # Write back to temporary file.
+    open(SVG, ">$svgfile") || die "open $svgfile: $!";
+    print SVG $svg;
+    close(SVG);
+  }
+}
+
+sub SvgJavascript {
+  return <<'EOF';
+<script type="text/ecmascript"><![CDATA[
+// SVGPan
+// http://www.cyberz.org/blog/2009/12/08/svgpan-a-javascript-svg-panzoomdrag-library/
+// Local modification: if(true || ...) below to force panning, never moving.
+
+/**
+ *  SVGPan library 1.2
+ * ====================
+ *
+ * Given an unique existing element with id "viewport", including the
+ * the library into any SVG adds the following capabilities:
+ *
+ *  - Mouse panning
+ *  - Mouse zooming (using the wheel)
+ *  - Object dargging
+ *
+ * Known issues:
+ *
+ *  - Zooming (while panning) on Safari has still some issues
+ *
+ * Releases:
+ *
+ * 1.2, Sat Mar 20 08:42:50 GMT 2010, Zeng Xiaohui
+ *	Fixed a bug with browser mouse handler interaction
+ *
+ * 1.1, Wed Feb  3 17:39:33 GMT 2010, Zeng Xiaohui
+ *	Updated the zoom code to support the mouse wheel on Safari/Chrome
+ *
+ * 1.0, Andrea Leofreddi
+ *	First release
+ *
+ * This code is licensed under the following BSD license:
+ *
+ * Copyright 2009-2010 Andrea Leofreddi <a.leofreddi@itcharm.com>. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without modification, are
+ * permitted provided that the following conditions are met:
+ *
+ *    1. Redistributions of source code must retain the above copyright notice, this list of
+ *       conditions and the following disclaimer.
+ *
+ *    2. Redistributions in binary form must reproduce the above copyright notice, this list
+ *       of conditions and the following disclaimer in the documentation and/or other materials
+ *       provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY Andrea Leofreddi ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
+ * FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL Andrea Leofreddi OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * The views and conclusions contained in the software and documentation are those of the
+ * authors and should not be interpreted as representing official policies, either expressed
+ * or implied, of Andrea Leofreddi.
+ */
+
+var root = document.documentElement;
+
+var state = 'none', stateTarget, stateOrigin, stateTf;
+
+setupHandlers(root);
+
+/**
+ * Register handlers
+ */
+function setupHandlers(root){
+	setAttributes(root, {
+		"onmouseup" : "add(evt)",
+		"onmousedown" : "handleMouseDown(evt)",
+		"onmousemove" : "handleMouseMove(evt)",
+		"onmouseup" : "handleMouseUp(evt)",
+		//"onmouseout" : "handleMouseUp(evt)", // Decomment this to stop the pan functionality when dragging out of the SVG element
+	});
+
+	if(navigator.userAgent.toLowerCase().indexOf('webkit') >= 0)
+		window.addEventListener('mousewheel', handleMouseWheel, false); // Chrome/Safari
+	else
+		window.addEventListener('DOMMouseScroll', handleMouseWheel, false); // Others
+
+	var g = svgDoc.getElementById("svg");
+	g.width = "100%";
+	g.height = "100%";
+}
+
+/**
+ * Instance an SVGPoint object with given event coordinates.
+ */
+function getEventPoint(evt) {
+	var p = root.createSVGPoint();
+
+	p.x = evt.clientX;
+	p.y = evt.clientY;
+
+	return p;
+}
+
+/**
+ * Sets the current transform matrix of an element.
+ */
+function setCTM(element, matrix) {
+	var s = "matrix(" + matrix.a + "," + matrix.b + "," + matrix.c + "," + matrix.d + "," + matrix.e + "," + matrix.f + ")";
+
+	element.setAttribute("transform", s);
+}
+
+/**
+ * Dumps a matrix to a string (useful for debug).
+ */
+function dumpMatrix(matrix) {
+	var s = "[ " + matrix.a + ", " + matrix.c + ", " + matrix.e + "\n  " + matrix.b + ", " + matrix.d + ", " + matrix.f + "\n  0, 0, 1 ]";
+
+	return s;
+}
+
+/**
+ * Sets attributes of an element.
+ */
+function setAttributes(element, attributes){
+	for (i in attributes)
+		element.setAttributeNS(null, i, attributes[i]);
+}
+
+/**
+ * Handle mouse move event.
+ */
+function handleMouseWheel(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var delta;
+
+	if(evt.wheelDelta)
+		delta = evt.wheelDelta / 3600; // Chrome/Safari
+	else
+		delta = evt.detail / -90; // Mozilla
+
+	var z = 1 + delta; // Zoom factor: 0.9/1.1
+
+	var g = svgDoc.getElementById("viewport");
+
+	var p = getEventPoint(evt);
+
+	p = p.matrixTransform(g.getCTM().inverse());
+
+	// Compute new scale matrix in current mouse position
+	var k = root.createSVGMatrix().translate(p.x, p.y).scale(z).translate(-p.x, -p.y);
+
+        setCTM(g, g.getCTM().multiply(k));
+
+	stateTf = stateTf.multiply(k.inverse());
+}
+
+/**
+ * Handle mouse move event.
+ */
+function handleMouseMove(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var g = svgDoc.getElementById("viewport");
+
+	if(state == 'pan') {
+		// Pan mode
+		var p = getEventPoint(evt).matrixTransform(stateTf);
+
+		setCTM(g, stateTf.inverse().translate(p.x - stateOrigin.x, p.y - stateOrigin.y));
+	} else if(state == 'move') {
+		// Move mode
+		var p = getEventPoint(evt).matrixTransform(g.getCTM().inverse());
+
+		setCTM(stateTarget, root.createSVGMatrix().translate(p.x - stateOrigin.x, p.y - stateOrigin.y).multiply(g.getCTM().inverse()).multiply(stateTarget.getCTM()));
+
+		stateOrigin = p;
+	}
+}
+
+/**
+ * Handle click event.
+ */
+function handleMouseDown(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	var g = svgDoc.getElementById("viewport");
+
+	if(true || evt.target.tagName == "svg") {
+		// Pan mode
+		state = 'pan';
+
+		stateTf = g.getCTM().inverse();
+
+		stateOrigin = getEventPoint(evt).matrixTransform(stateTf);
+	} else {
+		// Move mode
+		state = 'move';
+
+		stateTarget = evt.target;
+
+		stateTf = g.getCTM().inverse();
+
+		stateOrigin = getEventPoint(evt).matrixTransform(stateTf);
+	}
+}
+
+/**
+ * Handle mouse button release event.
+ */
+function handleMouseUp(evt) {
+	if(evt.preventDefault)
+		evt.preventDefault();
+
+	evt.returnValue = false;
+
+	var svgDoc = evt.target.ownerDocument;
+
+	if(state == 'pan' || state == 'move') {
+		// Quit pan mode
+		state = '';
+	}
+}
+
+]]></script>
+EOF
+}
+
+# Provides a map from fullname to shortname for cases where the
+# shortname is ambiguous.  The symlist has both the fullname and
+# shortname for all symbols, which is usually fine, but sometimes --
+# such as overloaded functions -- two different fullnames can map to
+# the same shortname.  In that case, we use the address of the
+# function to disambiguate the two.  This function fills in a map that
+# maps fullnames to modified shortnames in such cases.  If a fullname
+# is not present in the map, the 'normal' shortname provided by the
+# symlist is the appropriate one to use.
+sub FillFullnameToShortnameMap {
+  my $symbols = shift;
+  my $fullname_to_shortname_map = shift;
+  my $shortnames_seen_once = {};
+  my $shortnames_seen_more_than_once = {};
+
+  foreach my $symlist (values(%{$symbols})) {
+    # TODO(csilvers): deal with inlined symbols too.
+    my $shortname = $symlist->[0];
+    my $fullname = $symlist->[2];
+    if ($fullname !~ /<[0-9a-fA-F]+>$/) {  # fullname doesn't end in an address
+      next;       # the only collisions we care about are when addresses differ
+    }
+    if (defined($shortnames_seen_once->{$shortname}) &&
+        $shortnames_seen_once->{$shortname} ne $fullname) {
+      $shortnames_seen_more_than_once->{$shortname} = 1;
+    } else {
+      $shortnames_seen_once->{$shortname} = $fullname;
+    }
+  }
+
+  foreach my $symlist (values(%{$symbols})) {
+    my $shortname = $symlist->[0];
+    my $fullname = $symlist->[2];
+    # TODO(csilvers): take in a list of addresses we care about, and only
+    # store in the map if $symlist->[1] is in that list.  Saves space.
+    next if defined($fullname_to_shortname_map->{$fullname});
+    if (defined($shortnames_seen_more_than_once->{$shortname})) {
+      if ($fullname =~ /<0*([^>]*)>$/) {   # fullname has address at end of it
+        $fullname_to_shortname_map->{$fullname} = "$shortname\@$1";
+      }
+    }
+  }
+}
+
+# Return a small number that identifies the argument.
+# Multiple calls with the same argument will return the same number.
+# Calls with different arguments will return different numbers.
+sub ShortIdFor {
+  my $key = shift;
+  my $id = $main::uniqueid{$key};
+  if (!defined($id)) {
+    $id = keys(%main::uniqueid) + 1;
+    $main::uniqueid{$key} = $id;
+  }
+  return $id;
+}
+
+# Translate a stack of addresses into a stack of symbols
+sub TranslateStack {
+  my $symbols = shift;
+  my $fullname_to_shortname_map = shift;
+  my $k = shift;
+
+  my @addrs = split(/\n/, $k);
+  my @result = ();
+  for (my $i = 0; $i <= $#addrs; $i++) {
+    my $a = $addrs[$i];
+
+    # Skip large addresses since they sometimes show up as fake entries on RH9
+    if (length($a) > 8 && $a gt "7fffffffffffffff") {
+      next;
+    }
+
+    if ($main::opt_disasm || $main::opt_list) {
+      # We want just the address for the key
+      push(@result, $a);
+      next;
+    }
+
+    my $symlist = $symbols->{$a};
+    if (!defined($symlist)) {
+      $symlist = [$a, "", $a];
+    }
+
+    # We can have a sequence of symbols for a particular entry
+    # (more than one symbol in the case of inlining).  Callers
+    # come before callees in symlist, so walk backwards since
+    # the translated stack should contain callees before callers.
+    for (my $j = $#{$symlist}; $j >= 2; $j -= 3) {
+      my $func = $symlist->[$j-2];
+      my $fileline = $symlist->[$j-1];
+      my $fullfunc = $symlist->[$j];
+      if (defined($fullname_to_shortname_map->{$fullfunc})) {
+        $func = $fullname_to_shortname_map->{$fullfunc};
+      }
+      if ($j > 2) {
+        $func = "$func (inline)";
+      }
+
+      # Do not merge nodes corresponding to Callback::Run since that
+      # causes confusing cycles in dot display.  Instead, we synthesize
+      # a unique name for this frame per caller.
+      if ($func =~ m/Callback.*::Run$/) {
+        my $caller = ($i > 0) ? $addrs[$i-1] : 0;
+        $func = "Run#" . ShortIdFor($caller);
+      }
+
+      if ($main::opt_addresses) {
+        push(@result, "$a $func $fileline");
+      } elsif ($main::opt_lines) {
+        if ($func eq '??' && $fileline eq '??:0') {
+          push(@result, "$a");
+        } else {
+          push(@result, "$func $fileline");
+        }
+      } elsif ($main::opt_functions) {
+        if ($func eq '??') {
+          push(@result, "$a");
+        } else {
+          push(@result, $func);
+        }
+      } elsif ($main::opt_files) {
+        if ($fileline eq '??:0' || $fileline eq '') {
+          push(@result, "$a");
+        } else {
+          my $f = $fileline;
+          $f =~ s/:\d+$//;
+          push(@result, $f);
+        }
+      } else {
+        push(@result, $a);
+        last;  # Do not print inlined info
+      }
+    }
+  }
+
+  # print join(",", @addrs), " => ", join(",", @result), "\n";
+  return @result;
+}
+
+# Generate percent string for a number and a total
+sub Percent {
+  my $num = shift;
+  my $tot = shift;
+  if ($tot != 0) {
+    return sprintf("%.1f%%", $num * 100.0 / $tot);
+  } else {
+    return ($num == 0) ? "nan" : (($num > 0) ? "+inf" : "-inf");
+  }
+}
+
+# Generate pretty-printed form of number
+sub Unparse {
+  my $num = shift;
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    if ($main::opt_inuse_objects || $main::opt_alloc_objects) {
+      return sprintf("%d", $num);
+    } else {
+      if ($main::opt_show_bytes) {
+        return sprintf("%d", $num);
+      } else {
+        return sprintf("%.1f", $num / 1048576.0);
+      }
+    }
+  } elsif ($main::profile_type eq 'contention' && !$main::opt_contentions) {
+    return sprintf("%.3f", $num / 1e9); # Convert nanoseconds to seconds
+  } else {
+    return sprintf("%d", $num);
+  }
+}
+
+# Alternate pretty-printed form: 0 maps to "."
+sub UnparseAlt {
+  my $num = shift;
+  if ($num == 0) {
+    return ".";
+  } else {
+    return Unparse($num);
+  }
+}
+
+# Alternate pretty-printed form: 0 maps to ""
+sub HtmlPrintNumber {
+  my $num = shift;
+  if ($num == 0) {
+    return "";
+  } else {
+    return Unparse($num);
+  }
+}
+
+# Return output units
+sub Units {
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    if ($main::opt_inuse_objects || $main::opt_alloc_objects) {
+      return "objects";
+    } else {
+      if ($main::opt_show_bytes) {
+        return "B";
+      } else {
+        return "MB";
+      }
+    }
+  } elsif ($main::profile_type eq 'contention' && !$main::opt_contentions) {
+    return "seconds";
+  } else {
+    return "samples";
+  }
+}
+
+##### Profile manipulation code #####
+
+# Generate flattened profile:
+# If count is charged to stack [a,b,c,d], in generated profile,
+# it will be charged to [a]
+sub FlatProfile {
+  my $profile = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    if ($#addrs >= 0) {
+      AddEntry($result, $addrs[0], $count);
+    }
+  }
+  return $result;
+}
+
+# Generate cumulative profile:
+# If count is charged to stack [a,b,c,d], in generated profile,
+# it will be charged to [a], [b], [c], [d]
+sub CumulativeProfile {
+  my $profile = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    foreach my $a (@addrs) {
+      AddEntry($result, $a, $count);
+    }
+  }
+  return $result;
+}
+
+# If the second-youngest PC on the stack is always the same, returns
+# that pc.  Otherwise, returns undef.
+sub IsSecondPcAlwaysTheSame {
+  my $profile = shift;
+
+  my $second_pc = undef;
+  foreach my $k (keys(%{$profile})) {
+    my @addrs = split(/\n/, $k);
+    if ($#addrs < 1) {
+      return undef;
+    }
+    if (not defined $second_pc) {
+      $second_pc = $addrs[1];
+    } else {
+      if ($second_pc ne $addrs[1]) {
+        return undef;
+      }
+    }
+  }
+  return $second_pc;
+}
+
+sub ExtractSymbolLocation {
+  my $symbols = shift;
+  my $address = shift;
+  # 'addr2line' outputs "??:0" for unknown locations; we do the
+  # same to be consistent.
+  my $location = "??:0:unknown";
+  if (exists $symbols->{$address}) {
+    my $file = $symbols->{$address}->[1];
+    if ($file eq "?") {
+      $file = "??:0"
+    }
+    $location = $file . ":" . $symbols->{$address}->[0];
+  }
+  return $location;
+}
+
+# Extracts a graph of calls.
+sub ExtractCalls {
+  my $symbols = shift;
+  my $profile = shift;
+
+  my $calls = {};
+  while( my ($stack_trace, $count) = each %$profile ) {
+    my @address = split(/\n/, $stack_trace);
+    my $destination = ExtractSymbolLocation($symbols, $address[0]);
+    AddEntry($calls, $destination, $count);
+    for (my $i = 1; $i <= $#address; $i++) {
+      my $source = ExtractSymbolLocation($symbols, $address[$i]);
+      my $call = "$source -> $destination";
+      AddEntry($calls, $call, $count);
+      $destination = $source;
+    }
+  }
+
+  return $calls;
+}
+
+sub FilterFrames {
+  my $symbols = shift;
+  my $profile = shift;
+
+  if ($main::opt_retain eq '' && $main::opt_exclude eq '') {
+    return $profile;
+  }
+
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my @path = ();
+    foreach my $a (@addrs) {
+      my $sym;
+      if (exists($symbols->{$a})) {
+        $sym = $symbols->{$a}->[0];
+      } else {
+        $sym = $a;
+      }
+      if ($main::opt_retain ne '' && $sym !~ m/$main::opt_retain/) {
+        next;
+      }
+      if ($main::opt_exclude ne '' && $sym =~ m/$main::opt_exclude/) {
+        next;
+      }
+      push(@path, $a);
+    }
+    if (scalar(@path) > 0) {
+      my $reduced_path = join("\n", @path);
+      AddEntry($result, $reduced_path, $count);
+    }
+  }
+
+  return $result;
+}
+
+sub RemoveUninterestingFrames {
+  my $symbols = shift;
+  my $profile = shift;
+
+  # List of function names to skip
+  my %skip = ();
+  my $skip_regexp = 'NOMATCH';
+  if ($main::profile_type eq 'heap' || $main::profile_type eq 'growth') {
+    foreach my $name ('@JEMALLOC_PREFIX@calloc',
+                      'cfree',
+                      '@JEMALLOC_PREFIX@malloc',
+                      'newImpl',
+                      'void* newImpl',
+                      '@JEMALLOC_PREFIX@free',
+                      '@JEMALLOC_PREFIX@memalign',
+                      '@JEMALLOC_PREFIX@posix_memalign',
+                      '@JEMALLOC_PREFIX@aligned_alloc',
+                      'pvalloc',
+                      '@JEMALLOC_PREFIX@valloc',
+                      '@JEMALLOC_PREFIX@realloc',
+                      '@JEMALLOC_PREFIX@mallocx',
+                      '@JEMALLOC_PREFIX@rallocx',
+                      '@JEMALLOC_PREFIX@xallocx',
+                      '@JEMALLOC_PREFIX@dallocx',
+                      '@JEMALLOC_PREFIX@sdallocx',
+                      '@JEMALLOC_PREFIX@sdallocx_noflags',
+                      'tc_calloc',
+                      'tc_cfree',
+                      'tc_malloc',
+                      'tc_free',
+                      'tc_memalign',
+                      'tc_posix_memalign',
+                      'tc_pvalloc',
+                      'tc_valloc',
+                      'tc_realloc',
+                      'tc_new',
+                      'tc_delete',
+                      'tc_newarray',
+                      'tc_deletearray',
+                      'tc_new_nothrow',
+                      'tc_newarray_nothrow',
+                      'do_malloc',
+                      '::do_malloc',   # new name -- got moved to an unnamed ns
+                      '::do_malloc_or_cpp_alloc',
+                      'DoSampledAllocation',
+                      'simple_alloc::allocate',
+                      '__malloc_alloc_template::allocate',
+                      '__builtin_delete',
+                      '__builtin_new',
+                      '__builtin_vec_delete',
+                      '__builtin_vec_new',
+                      'operator new',
+                      'operator new[]',
+                      # The entry to our memory-allocation routines on OS X
+                      'malloc_zone_malloc',
+                      'malloc_zone_calloc',
+                      'malloc_zone_valloc',
+                      'malloc_zone_realloc',
+                      'malloc_zone_memalign',
+                      'malloc_zone_free',
+                      # These mark the beginning/end of our custom sections
+                      '__start_google_malloc',
+                      '__stop_google_malloc',
+                      '__start_malloc_hook',
+                      '__stop_malloc_hook') {
+      $skip{$name} = 1;
+      $skip{"_" . $name} = 1;   # Mach (OS X) adds a _ prefix to everything
+    }
+    # TODO: Remove TCMalloc once everything has been
+    # moved into the tcmalloc:: namespace and we have flushed
+    # old code out of the system.
+    $skip_regexp = "TCMalloc|^tcmalloc::";
+  } elsif ($main::profile_type eq 'contention') {
+    foreach my $vname ('base::RecordLockProfileData',
+                       'base::SubmitMutexProfileData',
+                       'base::SubmitSpinLockProfileData',
+                       'Mutex::Unlock',
+                       'Mutex::UnlockSlow',
+                       'Mutex::ReaderUnlock',
+                       'MutexLock::~MutexLock',
+                       'SpinLock::Unlock',
+                       'SpinLock::SlowUnlock',
+                       'SpinLockHolder::~SpinLockHolder') {
+      $skip{$vname} = 1;
+    }
+  } elsif ($main::profile_type eq 'cpu') {
+    # Drop signal handlers used for CPU profile collection
+    # TODO(dpeng): this should not be necessary; it's taken
+    # care of by the general 2nd-pc mechanism below.
+    foreach my $name ('ProfileData::Add',           # historical
+                      'ProfileData::prof_handler',  # historical
+                      'CpuProfiler::prof_handler',
+                      '__FRAME_END__',
+                      '__pthread_sighandler',
+                      '__restore') {
+      $skip{$name} = 1;
+    }
+  } else {
+    # Nothing skipped for unknown types
+  }
+
+  if ($main::profile_type eq 'cpu') {
+    # If all the second-youngest program counters are the same,
+    # this STRONGLY suggests that it is an artifact of measurement,
+    # i.e., stack frames pushed by the CPU profiler signal handler.
+    # Hence, we delete them.
+    # (The topmost PC is read from the signal structure, not from
+    # the stack, so it does not get involved.)
+    while (my $second_pc = IsSecondPcAlwaysTheSame($profile)) {
+      my $result = {};
+      my $func = '';
+      if (exists($symbols->{$second_pc})) {
+        $second_pc = $symbols->{$second_pc}->[0];
+      }
+      print STDERR "Removing $second_pc from all stack traces.\n";
+      foreach my $k (keys(%{$profile})) {
+        my $count = $profile->{$k};
+        my @addrs = split(/\n/, $k);
+        splice @addrs, 1, 1;
+        my $reduced_path = join("\n", @addrs);
+        AddEntry($result, $reduced_path, $count);
+      }
+      $profile = $result;
+    }
+  }
+
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my @path = ();
+    foreach my $a (@addrs) {
+      if (exists($symbols->{$a})) {
+        my $func = $symbols->{$a}->[0];
+        if ($skip{$func} || ($func =~ m/$skip_regexp/)) {
+          # Throw away the portion of the backtrace seen so far, under the
+          # assumption that previous frames were for functions internal to the
+          # allocator.
+          @path = ();
+          next;
+        }
+      }
+      push(@path, $a);
+    }
+    my $reduced_path = join("\n", @path);
+    AddEntry($result, $reduced_path, $count);
+  }
+
+  $result = FilterFrames($symbols, $result);
+
+  return $result;
+}
+
+# Reduce profile to granularity given by user
+sub ReduceProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $result = {};
+  my $fullname_to_shortname_map = {};
+  FillFullnameToShortnameMap($symbols, $fullname_to_shortname_map);
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @translated = TranslateStack($symbols, $fullname_to_shortname_map, $k);
+    my @path = ();
+    my %seen = ();
+    $seen{''} = 1;      # So that empty keys are skipped
+    foreach my $e (@translated) {
+      # To avoid double-counting due to recursion, skip a stack-trace
+      # entry if it has already been seen
+      if (!$seen{$e}) {
+        $seen{$e} = 1;
+        push(@path, $e);
+      }
+    }
+    my $reduced_path = join("\n", @path);
+    AddEntry($result, $reduced_path, $count);
+  }
+  return $result;
+}
+
+# Does the specified symbol array match the regexp?
+sub SymbolMatches {
+  my $sym = shift;
+  my $re = shift;
+  if (defined($sym)) {
+    for (my $i = 0; $i < $#{$sym}; $i += 3) {
+      if ($sym->[$i] =~ m/$re/ || $sym->[$i+1] =~ m/$re/) {
+        return 1;
+      }
+    }
+  }
+  return 0;
+}
+
+# Focus only on paths involving specified regexps
+sub FocusProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $focus = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    foreach my $a (@addrs) {
+      # Reply if it matches either the address/shortname/fileline
+      if (($a =~ m/$focus/) || SymbolMatches($symbols->{$a}, $focus)) {
+        AddEntry($result, $k, $count);
+        last;
+      }
+    }
+  }
+  return $result;
+}
+
+# Focus only on paths not involving specified regexps
+sub IgnoreProfile {
+  my $symbols = shift;
+  my $profile = shift;
+  my $ignore = shift;
+  my $result = {};
+  foreach my $k (keys(%{$profile})) {
+    my $count = $profile->{$k};
+    my @addrs = split(/\n/, $k);
+    my $matched = 0;
+    foreach my $a (@addrs) {
+      # Reply if it matches either the address/shortname/fileline
+      if (($a =~ m/$ignore/) || SymbolMatches($symbols->{$a}, $ignore)) {
+        $matched = 1;
+        last;
+      }
+    }
+    if (!$matched) {
+      AddEntry($result, $k, $count);
+    }
+  }
+  return $result;
+}
+
+# Get total count in profile
+sub TotalProfile {
+  my $profile = shift;
+  my $result = 0;
+  foreach my $k (keys(%{$profile})) {
+    $result += $profile->{$k};
+  }
+  return $result;
+}
+
+# Add A to B
+sub AddProfile {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  # add all keys in A
+  foreach my $k (keys(%{$A})) {
+    my $v = $A->{$k};
+    AddEntry($R, $k, $v);
+  }
+  # add all keys in B
+  foreach my $k (keys(%{$B})) {
+    my $v = $B->{$k};
+    AddEntry($R, $k, $v);
+  }
+  return $R;
+}
+
+# Merges symbol maps
+sub MergeSymbols {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  foreach my $k (keys(%{$A})) {
+    $R->{$k} = $A->{$k};
+  }
+  if (defined($B)) {
+    foreach my $k (keys(%{$B})) {
+      $R->{$k} = $B->{$k};
+    }
+  }
+  return $R;
+}
+
+
+# Add A to B
+sub AddPcs {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  # add all keys in A
+  foreach my $k (keys(%{$A})) {
+    $R->{$k} = 1
+  }
+  # add all keys in B
+  foreach my $k (keys(%{$B})) {
+    $R->{$k} = 1
+  }
+  return $R;
+}
+
+# Subtract B from A
+sub SubtractProfile {
+  my $A = shift;
+  my $B = shift;
+
+  my $R = {};
+  foreach my $k (keys(%{$A})) {
+    my $v = $A->{$k} - GetEntry($B, $k);
+    if ($v < 0 && $main::opt_drop_negative) {
+      $v = 0;
+    }
+    AddEntry($R, $k, $v);
+  }
+  if (!$main::opt_drop_negative) {
+    # Take care of when subtracted profile has more entries
+    foreach my $k (keys(%{$B})) {
+      if (!exists($A->{$k})) {
+        AddEntry($R, $k, 0 - $B->{$k});
+      }
+    }
+  }
+  return $R;
+}
+
+# Get entry from profile; zero if not present
+sub GetEntry {
+  my $profile = shift;
+  my $k = shift;
+  if (exists($profile->{$k})) {
+    return $profile->{$k};
+  } else {
+    return 0;
+  }
+}
+
+# Add entry to specified profile
+sub AddEntry {
+  my $profile = shift;
+  my $k = shift;
+  my $n = shift;
+  if (!exists($profile->{$k})) {
+    $profile->{$k} = 0;
+  }
+  $profile->{$k} += $n;
+}
+
+# Add a stack of entries to specified profile, and add them to the $pcs
+# list.
+sub AddEntries {
+  my $profile = shift;
+  my $pcs = shift;
+  my $stack = shift;
+  my $count = shift;
+  my @k = ();
+
+  foreach my $e (split(/\s+/, $stack)) {
+    my $pc = HexExtend($e);
+    $pcs->{$pc} = 1;
+    push @k, $pc;
+  }
+  AddEntry($profile, (join "\n", @k), $count);
+}
+
+##### Code to profile a server dynamically #####
+
+sub CheckSymbolPage {
+  my $url = SymbolPageURL();
+  my $command = ShellEscape(@URL_FETCHER, $url);
+  open(SYMBOL, "$command |") or error($command);
+  my $line = <SYMBOL>;
+  $line =~ s/\r//g;         # turn windows-looking lines into unix-looking lines
+  close(SYMBOL);
+  unless (defined($line)) {
+    error("$url doesn't exist\n");
+  }
+
+  if ($line =~ /^num_symbols:\s+(\d+)$/) {
+    if ($1 == 0) {
+      error("Stripped binary. No symbols available.\n");
+    }
+  } else {
+    error("Failed to get the number of symbols from $url\n");
+  }
+}
+
+sub IsProfileURL {
+  my $profile_name = shift;
+  if (-f $profile_name) {
+    printf STDERR "Using local file $profile_name.\n";
+    return 0;
+  }
+  return 1;
+}
+
+sub ParseProfileURL {
+  my $profile_name = shift;
+
+  if (!defined($profile_name) || $profile_name eq "") {
+    return ();
+  }
+
+  # Split profile URL - matches all non-empty strings, so no test.
+  $profile_name =~ m,^(https?://)?([^/]+)(.*?)(/|$PROFILES)?$,;
+
+  my $proto = $1 || "http://";
+  my $hostport = $2;
+  my $prefix = $3;
+  my $profile = $4 || "/";
+
+  my $host = $hostport;
+  $host =~ s/:.*//;
+
+  my $baseurl = "$proto$hostport$prefix";
+  return ($host, $baseurl, $profile);
+}
+
+# We fetch symbols from the first profile argument.
+sub SymbolPageURL {
+  my ($host, $baseURL, $path) = ParseProfileURL($main::pfile_args[0]);
+  return "$baseURL$SYMBOL_PAGE";
+}
+
+sub FetchProgramName() {
+  my ($host, $baseURL, $path) = ParseProfileURL($main::pfile_args[0]);
+  my $url = "$baseURL$PROGRAM_NAME_PAGE";
+  my $command_line = ShellEscape(@URL_FETCHER, $url);
+  open(CMDLINE, "$command_line |") or error($command_line);
+  my $cmdline = <CMDLINE>;
+  $cmdline =~ s/\r//g;   # turn windows-looking lines into unix-looking lines
+  close(CMDLINE);
+  error("Failed to get program name from $url\n") unless defined($cmdline);
+  $cmdline =~ s/\x00.+//;  # Remove argv[1] and latters.
+  $cmdline =~ s!\n!!g;  # Remove LFs.
+  return $cmdline;
+}
+
+# Gee, curl's -L (--location) option isn't reliable at least
+# with its 7.12.3 version.  Curl will forget to post data if
+# there is a redirection.  This function is a workaround for
+# curl.  Redirection happens on borg hosts.
+sub ResolveRedirectionForCurl {
+  my $url = shift;
+  my $command_line = ShellEscape(@URL_FETCHER, "--head", $url);
+  open(CMDLINE, "$command_line |") or error($command_line);
+  while (<CMDLINE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (/^Location: (.*)/) {
+      $url = $1;
+    }
+  }
+  close(CMDLINE);
+  return $url;
+}
+
+# Add a timeout flat to URL_FETCHER.  Returns a new list.
+sub AddFetchTimeout {
+  my $timeout = shift;
+  my @fetcher = @_;
+  if (defined($timeout)) {
+    if (join(" ", @fetcher) =~ m/\bcurl -s/) {
+      push(@fetcher, "--max-time", sprintf("%d", $timeout));
+    } elsif (join(" ", @fetcher) =~ m/\brpcget\b/) {
+      push(@fetcher, sprintf("--deadline=%d", $timeout));
+    }
+  }
+  return @fetcher;
+}
+
+# Reads a symbol map from the file handle name given as $1, returning
+# the resulting symbol map.  Also processes variables relating to symbols.
+# Currently, the only variable processed is 'binary=<value>' which updates
+# $main::prog to have the correct program name.
+sub ReadSymbols {
+  my $in = shift;
+  my $map = {};
+  while (<$in>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Removes all the leading zeroes from the symbols, see comment below.
+    if (m/^0x0*([0-9a-f]+)\s+(.+)/) {
+      $map->{$1} = $2;
+    } elsif (m/^---/) {
+      last;
+    } elsif (m/^([a-z][^=]*)=(.*)$/ ) {
+      my ($variable, $value) = ($1, $2);
+      for ($variable, $value) {
+        s/^\s+//;
+        s/\s+$//;
+      }
+      if ($variable eq "binary") {
+        if ($main::prog ne $UNKNOWN_BINARY && $main::prog ne $value) {
+          printf STDERR ("Warning: Mismatched binary name '%s', using '%s'.\n",
+                         $main::prog, $value);
+        }
+        $main::prog = $value;
+      } else {
+        printf STDERR ("Ignoring unknown variable in symbols list: " .
+            "'%s' = '%s'\n", $variable, $value);
+      }
+    }
+  }
+  return $map;
+}
+
+sub URLEncode {
+  my $str = shift;
+  $str =~ s/([^A-Za-z0-9\-_.!~*'()])/ sprintf "%%%02x", ord $1 /eg;
+  return $str;
+}
+
+sub AppendSymbolFilterParams {
+  my $url = shift;
+  my @params = ();
+  if ($main::opt_retain ne '') {
+    push(@params, sprintf("retain=%s", URLEncode($main::opt_retain)));
+  }
+  if ($main::opt_exclude ne '') {
+    push(@params, sprintf("exclude=%s", URLEncode($main::opt_exclude)));
+  }
+  if (scalar @params > 0) {
+    $url = sprintf("%s?%s", $url, join("&", @params));
+  }
+  return $url;
+}
+
+# Fetches and processes symbols to prepare them for use in the profile output
+# code.  If the optional 'symbol_map' arg is not given, fetches symbols from
+# $SYMBOL_PAGE for all PC values found in profile.  Otherwise, the raw symbols
+# are assumed to have already been fetched into 'symbol_map' and are simply
+# extracted and processed.
+sub FetchSymbols {
+  my $pcset = shift;
+  my $symbol_map = shift;
+
+  my %seen = ();
+  my @pcs = grep { !$seen{$_}++ } keys(%$pcset);  # uniq
+
+  if (!defined($symbol_map)) {
+    my $post_data = join("+", sort((map {"0x" . "$_"} @pcs)));
+
+    open(POSTFILE, ">$main::tmpfile_sym");
+    print POSTFILE $post_data;
+    close(POSTFILE);
+
+    my $url = SymbolPageURL();
+
+    my $command_line;
+    if (join(" ", @URL_FETCHER) =~ m/\bcurl -s/) {
+      $url = ResolveRedirectionForCurl($url);
+      $url = AppendSymbolFilterParams($url);
+      $command_line = ShellEscape(@URL_FETCHER, "-d", "\@$main::tmpfile_sym",
+                                  $url);
+    } else {
+      $url = AppendSymbolFilterParams($url);
+      $command_line = (ShellEscape(@URL_FETCHER, "--post", $url)
+                       . " < " . ShellEscape($main::tmpfile_sym));
+    }
+    # We use c++filt in case $SYMBOL_PAGE gives us mangled symbols.
+    my $escaped_cppfilt = ShellEscape($obj_tool_map{"c++filt"});
+    open(SYMBOL, "$command_line | $escaped_cppfilt |") or error($command_line);
+    $symbol_map = ReadSymbols(*SYMBOL{IO});
+    close(SYMBOL);
+  }
+
+  my $symbols = {};
+  foreach my $pc (@pcs) {
+    my $fullname;
+    # For 64 bits binaries, symbols are extracted with 8 leading zeroes.
+    # Then /symbol reads the long symbols in as uint64, and outputs
+    # the result with a "0x%08llx" format which get rid of the zeroes.
+    # By removing all the leading zeroes in both $pc and the symbols from
+    # /symbol, the symbols match and are retrievable from the map.
+    my $shortpc = $pc;
+    $shortpc =~ s/^0*//;
+    # Each line may have a list of names, which includes the function
+    # and also other functions it has inlined.  They are separated (in
+    # PrintSymbolizedProfile), by --, which is illegal in function names.
+    my $fullnames;
+    if (defined($symbol_map->{$shortpc})) {
+      $fullnames = $symbol_map->{$shortpc};
+    } else {
+      $fullnames = "0x" . $pc;  # Just use addresses
+    }
+    my $sym = [];
+    $symbols->{$pc} = $sym;
+    foreach my $fullname (split("--", $fullnames)) {
+      my $name = ShortFunctionName($fullname);
+      push(@{$sym}, $name, "?", $fullname);
+    }
+  }
+  return $symbols;
+}
+
+sub BaseName {
+  my $file_name = shift;
+  $file_name =~ s!^.*/!!;  # Remove directory name
+  return $file_name;
+}
+
+sub MakeProfileBaseName {
+  my ($binary_name, $profile_name) = @_;
+  my ($host, $baseURL, $path) = ParseProfileURL($profile_name);
+  my $binary_shortname = BaseName($binary_name);
+  return sprintf("%s.%s.%s",
+                 $binary_shortname, $main::op_time, $host);
+}
+
+sub FetchDynamicProfile {
+  my $binary_name = shift;
+  my $profile_name = shift;
+  my $fetch_name_only = shift;
+  my $encourage_patience = shift;
+
+  if (!IsProfileURL($profile_name)) {
+    return $profile_name;
+  } else {
+    my ($host, $baseURL, $path) = ParseProfileURL($profile_name);
+    if ($path eq "" || $path eq "/") {
+      # Missing type specifier defaults to cpu-profile
+      $path = $PROFILE_PAGE;
+    }
+
+    my $profile_file = MakeProfileBaseName($binary_name, $profile_name);
+
+    my $url = "$baseURL$path";
+    my $fetch_timeout = undef;
+    if ($path =~ m/$PROFILE_PAGE|$PMUPROFILE_PAGE/) {
+      if ($path =~ m/[?]/) {
+        $url .= "&";
+      } else {
+        $url .= "?";
+      }
+      $url .= sprintf("seconds=%d", $main::opt_seconds);
+      $fetch_timeout = $main::opt_seconds * 1.01 + 60;
+      # Set $profile_type for consumption by PrintSymbolizedProfile.
+      $main::profile_type = 'cpu';
+    } else {
+      # For non-CPU profiles, we add a type-extension to
+      # the target profile file name.
+      my $suffix = $path;
+      $suffix =~ s,/,.,g;
+      $profile_file .= $suffix;
+      # Set $profile_type for consumption by PrintSymbolizedProfile.
+      if ($path =~ m/$HEAP_PAGE/) {
+        $main::profile_type = 'heap';
+      } elsif ($path =~ m/$GROWTH_PAGE/) {
+        $main::profile_type = 'growth';
+      } elsif ($path =~ m/$CONTENTION_PAGE/) {
+        $main::profile_type = 'contention';
+      }
+    }
+
+    my $profile_dir = $ENV{"JEPROF_TMPDIR"} || ($ENV{HOME} . "/jeprof");
+    if (! -d $profile_dir) {
+      mkdir($profile_dir)
+          || die("Unable to create profile directory $profile_dir: $!\n");
+    }
+    my $tmp_profile = "$profile_dir/.tmp.$profile_file";
+    my $real_profile = "$profile_dir/$profile_file";
+
+    if ($fetch_name_only > 0) {
+      return $real_profile;
+    }
+
+    my @fetcher = AddFetchTimeout($fetch_timeout, @URL_FETCHER);
+    my $cmd = ShellEscape(@fetcher, $url) . " > " . ShellEscape($tmp_profile);
+    if ($path =~ m/$PROFILE_PAGE|$PMUPROFILE_PAGE|$CENSUSPROFILE_PAGE/){
+      print STDERR "Gathering CPU profile from $url for $main::opt_seconds seconds to\n  ${real_profile}\n";
+      if ($encourage_patience) {
+        print STDERR "Be patient...\n";
+      }
+    } else {
+      print STDERR "Fetching $path profile from $url to\n  ${real_profile}\n";
+    }
+
+    (system($cmd) == 0) || error("Failed to get profile: $cmd: $!\n");
+    (system("mv", $tmp_profile, $real_profile) == 0) || error("Unable to rename profile\n");
+    print STDERR "Wrote profile to $real_profile\n";
+    $main::collected_profile = $real_profile;
+    return $main::collected_profile;
+  }
+}
+
+# Collect profiles in parallel
+sub FetchDynamicProfiles {
+  my $items = scalar(@main::pfile_args);
+  my $levels = log($items) / log(2);
+
+  if ($items == 1) {
+    $main::profile_files[0] = FetchDynamicProfile($main::prog, $main::pfile_args[0], 0, 1);
+  } else {
+    # math rounding issues
+    if ((2 ** $levels) < $items) {
+     $levels++;
+    }
+    my $count = scalar(@main::pfile_args);
+    for (my $i = 0; $i < $count; $i++) {
+      $main::profile_files[$i] = FetchDynamicProfile($main::prog, $main::pfile_args[$i], 1, 0);
+    }
+    print STDERR "Fetching $count profiles, Be patient...\n";
+    FetchDynamicProfilesRecurse($levels, 0, 0);
+    $main::collected_profile = join(" \\\n    ", @main::profile_files);
+  }
+}
+
+# Recursively fork a process to get enough processes
+# collecting profiles
+sub FetchDynamicProfilesRecurse {
+  my $maxlevel = shift;
+  my $level = shift;
+  my $position = shift;
+
+  if (my $pid = fork()) {
+    $position = 0 | ($position << 1);
+    TryCollectProfile($maxlevel, $level, $position);
+    wait;
+  } else {
+    $position = 1 | ($position << 1);
+    TryCollectProfile($maxlevel, $level, $position);
+    cleanup();
+    exit(0);
+  }
+}
+
+# Collect a single profile
+sub TryCollectProfile {
+  my $maxlevel = shift;
+  my $level = shift;
+  my $position = shift;
+
+  if ($level >= ($maxlevel - 1)) {
+    if ($position < scalar(@main::pfile_args)) {
+      FetchDynamicProfile($main::prog, $main::pfile_args[$position], 0, 0);
+    }
+  } else {
+    FetchDynamicProfilesRecurse($maxlevel, $level+1, $position);
+  }
+}
+
+##### Parsing code #####
+
+# Provide a small streaming-read module to handle very large
+# cpu-profile files.  Stream in chunks along a sliding window.
+# Provides an interface to get one 'slot', correctly handling
+# endian-ness differences.  A slot is one 32-bit or 64-bit word
+# (depending on the input profile).  We tell endianness and bit-size
+# for the profile by looking at the first 8 bytes: in cpu profiles,
+# the second slot is always 3 (we'll accept anything that's not 0).
+BEGIN {
+  package CpuProfileStream;
+
+  sub new {
+    my ($class, $file, $fname) = @_;
+    my $self = { file        => $file,
+                 base        => 0,
+                 stride      => 512 * 1024,   # must be a multiple of bitsize/8
+                 slots       => [],
+                 unpack_code => "",           # N for big-endian, V for little
+                 perl_is_64bit => 1,          # matters if profile is 64-bit
+    };
+    bless $self, $class;
+    # Let unittests adjust the stride
+    if ($main::opt_test_stride > 0) {
+      $self->{stride} = $main::opt_test_stride;
+    }
+    # Read the first two slots to figure out bitsize and endianness.
+    my $slots = $self->{slots};
+    my $str;
+    read($self->{file}, $str, 8);
+    # Set the global $address_length based on what we see here.
+    # 8 is 32-bit (8 hexadecimal chars); 16 is 64-bit (16 hexadecimal chars).
+    $address_length = ($str eq (chr(0)x8)) ? 16 : 8;
+    if ($address_length == 8) {
+      if (substr($str, 6, 2) eq chr(0)x2) {
+        $self->{unpack_code} = 'V';  # Little-endian.
+      } elsif (substr($str, 4, 2) eq chr(0)x2) {
+        $self->{unpack_code} = 'N';  # Big-endian
+      } else {
+        ::error("$fname: header size >= 2**16\n");
+      }
+      @$slots = unpack($self->{unpack_code} . "*", $str);
+    } else {
+      # If we're a 64-bit profile, check if we're a 64-bit-capable
+      # perl.  Otherwise, each slot will be represented as a float
+      # instead of an int64, losing precision and making all the
+      # 64-bit addresses wrong.  We won't complain yet, but will
+      # later if we ever see a value that doesn't fit in 32 bits.
+      my $has_q = 0;
+      eval { $has_q = pack("Q", "1") ? 1 : 1; };
+      if (!$has_q) {
+        $self->{perl_is_64bit} = 0;
+      }
+      read($self->{file}, $str, 8);
+      if (substr($str, 4, 4) eq chr(0)x4) {
+        # We'd love to use 'Q', but it's a) not universal, b) not endian-proof.
+        $self->{unpack_code} = 'V';  # Little-endian.
+      } elsif (substr($str, 0, 4) eq chr(0)x4) {
+        $self->{unpack_code} = 'N';  # Big-endian
+      } else {
+        ::error("$fname: header size >= 2**32\n");
+      }
+      my @pair = unpack($self->{unpack_code} . "*", $str);
+      # Since we know one of the pair is 0, it's fine to just add them.
+      @$slots = (0, $pair[0] + $pair[1]);
+    }
+    return $self;
+  }
+
+  # Load more data when we access slots->get(X) which is not yet in memory.
+  sub overflow {
+    my ($self) = @_;
+    my $slots = $self->{slots};
+    $self->{base} += $#$slots + 1;   # skip over data we're replacing
+    my $str;
+    read($self->{file}, $str, $self->{stride});
+    if ($address_length == 8) {      # the 32-bit case
+      # This is the easy case: unpack provides 32-bit unpacking primitives.
+      @$slots = unpack($self->{unpack_code} . "*", $str);
+    } else {
+      # We need to unpack 32 bits at a time and combine.
+      my @b32_values = unpack($self->{unpack_code} . "*", $str);
+      my @b64_values = ();
+      for (my $i = 0; $i < $#b32_values; $i += 2) {
+        # TODO(csilvers): if this is a 32-bit perl, the math below
+        #    could end up in a too-large int, which perl will promote
+        #    to a double, losing necessary precision.  Deal with that.
+        #    Right now, we just die.
+        my ($lo, $hi) = ($b32_values[$i], $b32_values[$i+1]);
+        if ($self->{unpack_code} eq 'N') {    # big-endian
+          ($lo, $hi) = ($hi, $lo);
+        }
+        my $value = $lo + $hi * (2**32);
+        if (!$self->{perl_is_64bit} &&   # check value is exactly represented
+            (($value % (2**32)) != $lo || int($value / (2**32)) != $hi)) {
+          ::error("Need a 64-bit perl to process this 64-bit profile.\n");
+        }
+        push(@b64_values, $value);
+      }
+      @$slots = @b64_values;
+    }
+  }
+
+  # Access the i-th long in the file (logically), or -1 at EOF.
+  sub get {
+    my ($self, $idx) = @_;
+    my $slots = $self->{slots};
+    while ($#$slots >= 0) {
+      if ($idx < $self->{base}) {
+        # The only time we expect a reference to $slots[$i - something]
+        # after referencing $slots[$i] is reading the very first header.
+        # Since $stride > |header|, that shouldn't cause any lookback
+        # errors.  And everything after the header is sequential.
+        print STDERR "Unexpected look-back reading CPU profile";
+        return -1;   # shrug, don't know what better to return
+      } elsif ($idx > $self->{base} + $#$slots) {
+        $self->overflow();
+      } else {
+        return $slots->[$idx - $self->{base}];
+      }
+    }
+    # If we get here, $slots is [], which means we've reached EOF
+    return -1;  # unique since slots is supposed to hold unsigned numbers
+  }
+}
+
+# Reads the top, 'header' section of a profile, and returns the last
+# line of the header, commonly called a 'header line'.  The header
+# section of a profile consists of zero or more 'command' lines that
+# are instructions to jeprof, which jeprof executes when reading the
+# header.  All 'command' lines start with a %.  After the command
+# lines is the 'header line', which is a profile-specific line that
+# indicates what type of profile it is, and perhaps other global
+# information about the profile.  For instance, here's a header line
+# for a heap profile:
+#   heap profile:     53:    38236 [  5525:  1284029] @ heapprofile
+# For historical reasons, the CPU profile does not contain a text-
+# readable header line.  If the profile looks like a CPU profile,
+# this function returns "".  If no header line could be found, this
+# function returns undef.
+#
+# The following commands are recognized:
+#   %warn -- emit the rest of this line to stderr, prefixed by 'WARNING:'
+#
+# The input file should be in binmode.
+sub ReadProfileHeader {
+  local *PROFILE = shift;
+  my $firstchar = "";
+  my $line = "";
+  read(PROFILE, $firstchar, 1);
+  seek(PROFILE, -1, 1);                    # unread the firstchar
+  if ($firstchar !~ /[[:print:]]/) {       # is not a text character
+    return "";
+  }
+  while (defined($line = <PROFILE>)) {
+    $line =~ s/\r//g;   # turn windows-looking lines into unix-looking lines
+    if ($line =~ /^%warn\s+(.*)/) {        # 'warn' command
+      # Note this matches both '%warn blah\n' and '%warn\n'.
+      print STDERR "WARNING: $1\n";        # print the rest of the line
+    } elsif ($line =~ /^%/) {
+      print STDERR "Ignoring unknown command from profile header: $line";
+    } else {
+      # End of commands, must be the header line.
+      return $line;
+    }
+  }
+  return undef;     # got to EOF without seeing a header line
+}
+
+sub IsSymbolizedProfileFile {
+  my $file_name = shift;
+  if (!(-e $file_name) || !(-r $file_name)) {
+    return 0;
+  }
+  # Check if the file contains a symbol-section marker.
+  open(TFILE, "<$file_name");
+  binmode TFILE;
+  my $firstline = ReadProfileHeader(*TFILE);
+  close(TFILE);
+  if (!$firstline) {
+    return 0;
+  }
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+  return $firstline =~ /^--- *$symbol_marker/;
+}
+
+# Parse profile generated by common/profiler.cc and return a reference
+# to a map:
+#      $result->{version}     Version number of profile file
+#      $result->{period}      Sampling period (in microseconds)
+#      $result->{profile}     Profile object
+#      $result->{threads}     Map of thread IDs to profile objects
+#      $result->{map}         Memory map info from profile
+#      $result->{pcs}         Hash of all PC values seen, key is hex address
+sub ReadProfile {
+  my $prog = shift;
+  my $fname = shift;
+  my $result;            # return value
+
+  $CONTENTION_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $contention_marker = $&;
+  $GROWTH_PAGE  =~ m,[^/]+$,;    # matches everything after the last slash
+  my $growth_marker = $&;
+  $SYMBOL_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $symbol_marker = $&;
+  $PROFILE_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $profile_marker = $&;
+  $HEAP_PAGE =~ m,[^/]+$,;    # matches everything after the last slash
+  my $heap_marker = $&;
+
+  # Look at first line to see if it is a heap or a CPU profile.
+  # CPU profile may start with no header at all, and just binary data
+  # (starting with \0\0\0\0) -- in that case, don't try to read the
+  # whole firstline, since it may be gigabytes(!) of data.
+  open(PROFILE, "<$fname") || error("$fname: $!\n");
+  binmode PROFILE;      # New perls do UTF-8 processing
+  my $header = ReadProfileHeader(*PROFILE);
+  if (!defined($header)) {   # means "at EOF"
+    error("Profile is empty.\n");
+  }
+
+  my $symbols;
+  if ($header =~ m/^--- *$symbol_marker/o) {
+    # Verify that the user asked for a symbolized profile
+    if (!$main::use_symbolized_profile) {
+      # we have both a binary and symbolized profiles, abort
+      error("FATAL ERROR: Symbolized profile\n   $fname\ncannot be used with " .
+            "a binary arg. Try again without passing\n   $prog\n");
+    }
+    # Read the symbol section of the symbolized profile file.
+    $symbols = ReadSymbols(*PROFILE{IO});
+    # Read the next line to get the header for the remaining profile.
+    $header = ReadProfileHeader(*PROFILE) || "";
+  }
+
+  if ($header =~ m/^--- *($heap_marker|$growth_marker)/o) {
+    # Skip "--- ..." line for profile types that have their own headers.
+    $header = ReadProfileHeader(*PROFILE) || "";
+  }
+
+  $main::profile_type = '';
+
+  if ($header =~ m/^heap profile:.*$growth_marker/o) {
+    $main::profile_type = 'growth';
+    $result =  ReadHeapProfile($prog, *PROFILE, $header);
+  } elsif ($header =~ m/^heap profile:/) {
+    $main::profile_type = 'heap';
+    $result =  ReadHeapProfile($prog, *PROFILE, $header);
+  } elsif ($header =~ m/^heap/) {
+    $main::profile_type = 'heap';
+    $result = ReadThreadedHeapProfile($prog, $fname, $header);
+  } elsif ($header =~ m/^--- *$contention_marker/o) {
+    $main::profile_type = 'contention';
+    $result = ReadSynchProfile($prog, *PROFILE);
+  } elsif ($header =~ m/^--- *Stacks:/) {
+    print STDERR
+      "Old format contention profile: mistakenly reports " .
+      "condition variable signals as lock contentions.\n";
+    $main::profile_type = 'contention';
+    $result = ReadSynchProfile($prog, *PROFILE);
+  } elsif ($header =~ m/^--- *$profile_marker/) {
+    # the binary cpu profile data starts immediately after this line
+    $main::profile_type = 'cpu';
+    $result = ReadCPUProfile($prog, $fname, *PROFILE);
+  } else {
+    if (defined($symbols)) {
+      # a symbolized profile contains a format we don't recognize, bail out
+      error("$fname: Cannot recognize profile section after symbols.\n");
+    }
+    # no ascii header present -- must be a CPU profile
+    $main::profile_type = 'cpu';
+    $result = ReadCPUProfile($prog, $fname, *PROFILE);
+  }
+
+  close(PROFILE);
+
+  # if we got symbols along with the profile, return those as well
+  if (defined($symbols)) {
+    $result->{symbols} = $symbols;
+  }
+
+  return $result;
+}
+
+# Subtract one from caller pc so we map back to call instr.
+# However, don't do this if we're reading a symbolized profile
+# file, in which case the subtract-one was done when the file
+# was written.
+#
+# We apply the same logic to all readers, though ReadCPUProfile uses an
+# independent implementation.
+sub FixCallerAddresses {
+  my $stack = shift;
+  # --raw/http: Always subtract one from pc's, because PrintSymbolizedProfile()
+  # dumps unadjusted profiles.
+  {
+    $stack =~ /(\s)/;
+    my $delimiter = $1;
+    my @addrs = split(' ', $stack);
+    my @fixedaddrs;
+    $#fixedaddrs = $#addrs;
+    if ($#addrs >= 0) {
+      $fixedaddrs[0] = $addrs[0];
+    }
+    for (my $i = 1; $i <= $#addrs; $i++) {
+      $fixedaddrs[$i] = AddressSub($addrs[$i], "0x1");
+    }
+    return join $delimiter, @fixedaddrs;
+  }
+}
+
+# CPU profile reader
+sub ReadCPUProfile {
+  my $prog = shift;
+  my $fname = shift;       # just used for logging
+  local *PROFILE = shift;
+  my $version;
+  my $period;
+  my $i;
+  my $profile = {};
+  my $pcs = {};
+
+  # Parse string into array of slots.
+  my $slots = CpuProfileStream->new(*PROFILE, $fname);
+
+  # Read header.  The current header version is a 5-element structure
+  # containing:
+  #   0: header count (always 0)
+  #   1: header "words" (after this one: 3)
+  #   2: format version (0)
+  #   3: sampling period (usec)
+  #   4: unused padding (always 0)
+  if ($slots->get(0) != 0 ) {
+    error("$fname: not a profile file, or old format profile file\n");
+  }
+  $i = 2 + $slots->get(1);
+  $version = $slots->get(2);
+  $period = $slots->get(3);
+  # Do some sanity checking on these header values.
+  if ($version > (2**32) || $period > (2**32) || $i > (2**32) || $i < 5) {
+    error("$fname: not a profile file, or corrupted profile file\n");
+  }
+
+  # Parse profile
+  while ($slots->get($i) != -1) {
+    my $n = $slots->get($i++);
+    my $d = $slots->get($i++);
+    if ($d > (2**16)) {  # TODO(csilvers): what's a reasonable max-stack-depth?
+      my $addr = sprintf("0%o", $i * ($address_length == 8 ? 4 : 8));
+      print STDERR "At index $i (address $addr):\n";
+      error("$fname: stack trace depth >= 2**32\n");
+    }
+    if ($slots->get($i) == 0) {
+      # End of profile data marker
+      $i += $d;
+      last;
+    }
+
+    # Make key out of the stack entries
+    my @k = ();
+    for (my $j = 0; $j < $d; $j++) {
+      my $pc = $slots->get($i+$j);
+      # Subtract one from caller pc so we map back to call instr.
+      $pc--;
+      $pc = sprintf("%0*x", $address_length, $pc);
+      $pcs->{$pc} = 1;
+      push @k, $pc;
+    }
+
+    AddEntry($profile, (join "\n", @k), $n);
+    $i += $d;
+  }
+
+  # Parse map
+  my $map = '';
+  seek(PROFILE, $i * 4, 0);
+  read(PROFILE, $map, (stat PROFILE)[7]);
+
+  my $r = {};
+  $r->{version} = $version;
+  $r->{period} = $period;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+
+  return $r;
+}
+
+sub HeapProfileIndex {
+  my $index = 1;
+  if ($main::opt_inuse_space) {
+    $index = 1;
+  } elsif ($main::opt_inuse_objects) {
+    $index = 0;
+  } elsif ($main::opt_alloc_space) {
+    $index = 3;
+  } elsif ($main::opt_alloc_objects) {
+    $index = 2;
+  }
+  return $index;
+}
+
+sub ReadMappedLibraries {
+  my $fh = shift;
+  my $map = "";
+  # Read the /proc/self/maps data
+  while (<$fh>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    $map .= $_;
+  }
+  return $map;
+}
+
+sub ReadMemoryMap {
+  my $fh = shift;
+  my $map = "";
+  # Read /proc/self/maps data as formatted by DumpAddressMap()
+  my $buildvar = "";
+  while (<PROFILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Parse "build=<dir>" specification if supplied
+    if (m/^\s*build=(.*)\n/) {
+      $buildvar = $1;
+    }
+
+    # Expand "$build" variable if available
+    $_ =~ s/\$build\b/$buildvar/g;
+
+    $map .= $_;
+  }
+  return $map;
+}
+
+sub AdjustSamples {
+  my ($sample_adjustment, $sampling_algorithm, $n1, $s1, $n2, $s2) = @_;
+  if ($sample_adjustment) {
+    if ($sampling_algorithm == 2) {
+      # Remote-heap version 2
+      # The sampling frequency is the rate of a Poisson process.
+      # This means that the probability of sampling an allocation of
+      # size X with sampling rate Y is 1 - exp(-X/Y)
+      if ($n1 != 0) {
+        my $ratio = (($s1*1.0)/$n1)/($sample_adjustment);
+        my $scale_factor = 1/(1 - exp(-$ratio));
+        $n1 *= $scale_factor;
+        $s1 *= $scale_factor;
+      }
+      if ($n2 != 0) {
+        my $ratio = (($s2*1.0)/$n2)/($sample_adjustment);
+        my $scale_factor = 1/(1 - exp(-$ratio));
+        $n2 *= $scale_factor;
+        $s2 *= $scale_factor;
+      }
+    } else {
+      # Remote-heap version 1
+      my $ratio;
+      $ratio = (($s1*1.0)/$n1)/($sample_adjustment);
+      if ($ratio < 1) {
+        $n1 /= $ratio;
+        $s1 /= $ratio;
+      }
+      $ratio = (($s2*1.0)/$n2)/($sample_adjustment);
+      if ($ratio < 1) {
+        $n2 /= $ratio;
+        $s2 /= $ratio;
+      }
+    }
+  }
+  return ($n1, $s1, $n2, $s2);
+}
+
+sub ReadHeapProfile {
+  my $prog = shift;
+  local *PROFILE = shift;
+  my $header = shift;
+
+  my $index = HeapProfileIndex();
+
+  # Find the type of this profile.  The header line looks like:
+  #    heap profile:   1246:  8800744 [  1246:  8800744] @ <heap-url>/266053
+  # There are two pairs <count: size>, the first inuse objects/space, and the
+  # second allocated objects/space.  This is followed optionally by a profile
+  # type, and if that is present, optionally by a sampling frequency.
+  # For remote heap profiles (v1):
+  # The interpretation of the sampling frequency is that the profiler, for
+  # each sample, calculates a uniformly distributed random integer less than
+  # the given value, and records the next sample after that many bytes have
+  # been allocated.  Therefore, the expected sample interval is half of the
+  # given frequency.  By default, if not specified, the expected sample
+  # interval is 128KB.  Only remote-heap-page profiles are adjusted for
+  # sample size.
+  # For remote heap profiles (v2):
+  # The sampling frequency is the rate of a Poisson process. This means that
+  # the probability of sampling an allocation of size X with sampling rate Y
+  # is 1 - exp(-X/Y)
+  # For version 2, a typical header line might look like this:
+  # heap profile:   1922: 127792360 [  1922: 127792360] @ <heap-url>_v2/524288
+  # the trailing number (524288) is the sampling rate. (Version 1 showed
+  # double the 'rate' here)
+  my $sampling_algorithm = 0;
+  my $sample_adjustment = 0;
+  chomp($header);
+  my $type = "unknown";
+  if ($header =~ m"^heap profile:\s*(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\](\s*@\s*([^/]*)(/(\d+))?)?") {
+    if (defined($6) && ($6 ne '')) {
+      $type = $6;
+      my $sample_period = $8;
+      # $type is "heapprofile" for profiles generated by the
+      # heap-profiler, and either "heap" or "heap_v2" for profiles
+      # generated by sampling directly within tcmalloc.  It can also
+      # be "growth" for heap-growth profiles.  The first is typically
+      # found for profiles generated locally, and the others for
+      # remote profiles.
+      if (($type eq "heapprofile") || ($type !~ /heap/) ) {
+        # No need to adjust for the sampling rate with heap-profiler-derived data
+        $sampling_algorithm = 0;
+      } elsif ($type =~ /_v2/) {
+        $sampling_algorithm = 2;     # version 2 sampling
+        if (defined($sample_period) && ($sample_period ne '')) {
+          $sample_adjustment = int($sample_period);
+        }
+      } else {
+        $sampling_algorithm = 1;     # version 1 sampling
+        if (defined($sample_period) && ($sample_period ne '')) {
+          $sample_adjustment = int($sample_period)/2;
+        }
+      }
+    } else {
+      # We detect whether or not this is a remote-heap profile by checking
+      # that the total-allocated stats ($n2,$s2) are exactly the
+      # same as the in-use stats ($n1,$s1).  It is remotely conceivable
+      # that a non-remote-heap profile may pass this check, but it is hard
+      # to imagine how that could happen.
+      # In this case it's so old it's guaranteed to be remote-heap version 1.
+      my ($n1, $s1, $n2, $s2) = ($1, $2, $3, $4);
+      if (($n1 == $n2) && ($s1 == $s2)) {
+        # This is likely to be a remote-heap based sample profile
+        $sampling_algorithm = 1;
+      }
+    }
+  }
+
+  if ($sampling_algorithm > 0) {
+    # For remote-heap generated profiles, adjust the counts and sizes to
+    # account for the sample rate (we sample once every 128KB by default).
+    if ($sample_adjustment == 0) {
+      # Turn on profile adjustment.
+      $sample_adjustment = 128*1024;
+      print STDERR "Adjusting heap profiles for 1-in-128KB sampling rate\n";
+    } else {
+      printf STDERR ("Adjusting heap profiles for 1-in-%d sampling rate\n",
+                     $sample_adjustment);
+    }
+    if ($sampling_algorithm > 1) {
+      # We don't bother printing anything for the original version (version 1)
+      printf STDERR "Heap version $sampling_algorithm\n";
+    }
+  }
+
+  my $profile = {};
+  my $pcs = {};
+  my $map = "";
+
+  while (<PROFILE>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (/^MAPPED_LIBRARIES:/) {
+      $map .= ReadMappedLibraries(*PROFILE);
+      last;
+    }
+
+    if (/^--- Memory map:/) {
+      $map .= ReadMemoryMap(*PROFILE);
+      last;
+    }
+
+    # Read entry of the form:
+    #  <count1>: <bytes1> [<count2>: <bytes2>] @ a1 a2 a3 ... an
+    s/^\s*//;
+    s/\s*$//;
+    if (m/^\s*(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\]\s+@\s+(.*)$/) {
+      my $stack = $5;
+      my ($n1, $s1, $n2, $s2) = ($1, $2, $3, $4);
+      my @counts = AdjustSamples($sample_adjustment, $sampling_algorithm,
+                                 $n1, $s1, $n2, $s2);
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $counts[$index]);
+    }
+  }
+
+  my $r = {};
+  $r->{version} = "heap";
+  $r->{period} = 1;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+sub ReadThreadedHeapProfile {
+  my ($prog, $fname, $header) = @_;
+
+  my $index = HeapProfileIndex();
+  my $sampling_algorithm = 0;
+  my $sample_adjustment = 0;
+  chomp($header);
+  my $type = "unknown";
+  # Assuming a very specific type of header for now.
+  if ($header =~ m"^heap_v2/(\d+)") {
+    $type = "_v2";
+    $sampling_algorithm = 2;
+    $sample_adjustment = int($1);
+  }
+  if ($type ne "_v2" || !defined($sample_adjustment)) {
+    die "Threaded heap profiles require v2 sampling with a sample rate\n";
+  }
+
+  my $profile = {};
+  my $thread_profiles = {};
+  my $pcs = {};
+  my $map = "";
+  my $stack = "";
+
+  while (<PROFILE>) {
+    s/\r//g;
+    if (/^MAPPED_LIBRARIES:/) {
+      $map .= ReadMappedLibraries(*PROFILE);
+      last;
+    }
+
+    if (/^--- Memory map:/) {
+      $map .= ReadMemoryMap(*PROFILE);
+      last;
+    }
+
+    # Read entry of the form:
+    # @ a1 a2 ... an
+    #   t*: <count1>: <bytes1> [<count2>: <bytes2>]
+    #   t1: <count1>: <bytes1> [<count2>: <bytes2>]
+    #     ...
+    #   tn: <count1>: <bytes1> [<count2>: <bytes2>]
+    s/^\s*//;
+    s/\s*$//;
+    if (m/^@\s+(.*)$/) {
+      $stack = $1;
+    } elsif (m/^\s*(t(\*|\d+)):\s+(\d+):\s+(\d+)\s+\[\s*(\d+):\s+(\d+)\]$/) {
+      if ($stack eq "") {
+        # Still in the header, so this is just a per-thread summary.
+        next;
+      }
+      my $thread = $2;
+      my ($n1, $s1, $n2, $s2) = ($3, $4, $5, $6);
+      my @counts = AdjustSamples($sample_adjustment, $sampling_algorithm,
+                                 $n1, $s1, $n2, $s2);
+      if ($thread eq "*") {
+        AddEntries($profile, $pcs, FixCallerAddresses($stack), $counts[$index]);
+      } else {
+        if (!exists($thread_profiles->{$thread})) {
+          $thread_profiles->{$thread} = {};
+        }
+        AddEntries($thread_profiles->{$thread}, $pcs,
+                   FixCallerAddresses($stack), $counts[$index]);
+      }
+    }
+  }
+
+  my $r = {};
+  $r->{version} = "heap";
+  $r->{period} = 1;
+  $r->{profile} = $profile;
+  $r->{threads} = $thread_profiles;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+sub ReadSynchProfile {
+  my $prog = shift;
+  local *PROFILE = shift;
+  my $header = shift;
+
+  my $map = '';
+  my $profile = {};
+  my $pcs = {};
+  my $sampling_period = 1;
+  my $cyclespernanosec = 2.8;   # Default assumption for old binaries
+  my $seen_clockrate = 0;
+  my $line;
+
+  my $index = 0;
+  if ($main::opt_total_delay) {
+    $index = 0;
+  } elsif ($main::opt_contentions) {
+    $index = 1;
+  } elsif ($main::opt_mean_delay) {
+    $index = 2;
+  }
+
+  while ( $line = <PROFILE> ) {
+    $line =~ s/\r//g;      # turn windows-looking lines into unix-looking lines
+    if ( $line =~ /^\s*(\d+)\s+(\d+) \@\s*(.*?)\s*$/ ) {
+      my ($cycles, $count, $stack) = ($1, $2, $3);
+
+      # Convert cycles to nanoseconds
+      $cycles /= $cyclespernanosec;
+
+      # Adjust for sampling done by application
+      $cycles *= $sampling_period;
+      $count *= $sampling_period;
+
+      my @values = ($cycles, $count, $cycles / $count);
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $values[$index]);
+
+    } elsif ( $line =~ /^(slow release).*thread \d+  \@\s*(.*?)\s*$/ ||
+              $line =~ /^\s*(\d+) \@\s*(.*?)\s*$/ ) {
+      my ($cycles, $stack) = ($1, $2);
+      if ($cycles !~ /^\d+$/) {
+        next;
+      }
+
+      # Convert cycles to nanoseconds
+      $cycles /= $cyclespernanosec;
+
+      # Adjust for sampling done by application
+      $cycles *= $sampling_period;
+
+      AddEntries($profile, $pcs, FixCallerAddresses($stack), $cycles);
+
+    } elsif ( $line =~ m/^([a-z][^=]*)=(.*)$/ ) {
+      my ($variable, $value) = ($1,$2);
+      for ($variable, $value) {
+        s/^\s+//;
+        s/\s+$//;
+      }
+      if ($variable eq "cycles/second") {
+        $cyclespernanosec = $value / 1e9;
+        $seen_clockrate = 1;
+      } elsif ($variable eq "sampling period") {
+        $sampling_period = $value;
+      } elsif ($variable eq "ms since reset") {
+        # Currently nothing is done with this value in jeprof
+        # So we just silently ignore it for now
+      } elsif ($variable eq "discarded samples") {
+        # Currently nothing is done with this value in jeprof
+        # So we just silently ignore it for now
+      } else {
+        printf STDERR ("Ignoring unnknown variable in /contention output: " .
+                       "'%s' = '%s'\n",$variable,$value);
+      }
+    } else {
+      # Memory map entry
+      $map .= $line;
+    }
+  }
+
+  if (!$seen_clockrate) {
+    printf STDERR ("No cycles/second entry in profile; Guessing %.1f GHz\n",
+                   $cyclespernanosec);
+  }
+
+  my $r = {};
+  $r->{version} = 0;
+  $r->{period} = $sampling_period;
+  $r->{profile} = $profile;
+  $r->{libs} = ParseLibraries($prog, $map, $pcs);
+  $r->{pcs} = $pcs;
+  return $r;
+}
+
+# Given a hex value in the form "0x1abcd" or "1abcd", return either
+# "0001abcd" or "000000000001abcd", depending on the current (global)
+# address length.
+sub HexExtend {
+  my $addr = shift;
+
+  $addr =~ s/^(0x)?0*//;
+  my $zeros_needed = $address_length - length($addr);
+  if ($zeros_needed < 0) {
+    printf STDERR "Warning: address $addr is longer than address length $address_length\n";
+    return $addr;
+  }
+  return ("0" x $zeros_needed) . $addr;
+}
+
+##### Symbol extraction #####
+
+# Aggressively search the lib_prefix values for the given library
+# If all else fails, just return the name of the library unmodified.
+# If the lib_prefix is "/my/path,/other/path" and $file is "/lib/dir/mylib.so"
+# it will search the following locations in this order, until it finds a file:
+#   /my/path/lib/dir/mylib.so
+#   /other/path/lib/dir/mylib.so
+#   /my/path/dir/mylib.so
+#   /other/path/dir/mylib.so
+#   /my/path/mylib.so
+#   /other/path/mylib.so
+#   /lib/dir/mylib.so              (returned as last resort)
+sub FindLibrary {
+  my $file = shift;
+  my $suffix = $file;
+
+  # Search for the library as described above
+  do {
+    foreach my $prefix (@prefix_list) {
+      my $fullpath = $prefix . $suffix;
+      if (-e $fullpath) {
+        return $fullpath;
+      }
+    }
+  } while ($suffix =~ s|^/[^/]+/|/|);
+  return $file;
+}
+
+# Return path to library with debugging symbols.
+# For libc libraries, the copy in /usr/lib/debug contains debugging symbols
+sub DebuggingLibrary {
+  my $file = shift;
+  if ($file =~ m|^/|) {
+      if (-f "/usr/lib/debug$file") {
+        return "/usr/lib/debug$file";
+      } elsif (-f "/usr/lib/debug$file.debug") {
+        return "/usr/lib/debug$file.debug";
+      }
+  }
+  return undef;
+}
+
+# Parse text section header of a library using objdump
+sub ParseTextSectionHeaderFromObjdump {
+  my $lib = shift;
+
+  my $size = undef;
+  my $vma;
+  my $file_offset;
+  # Get objdump output from the library file to figure out how to
+  # map between mapped addresses and addresses in the library.
+  my $cmd = ShellEscape($obj_tool_map{"objdump"}, "-h", $lib);
+  open(OBJDUMP, "$cmd |") || error("$cmd: $!\n");
+  while (<OBJDUMP>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    # Idx Name          Size      VMA       LMA       File off  Algn
+    #  10 .text         00104b2c  420156f0  420156f0  000156f0  2**4
+    # For 64-bit objects, VMA and LMA will be 16 hex digits, size and file
+    # offset may still be 8.  But AddressSub below will still handle that.
+    my @x = split;
+    if (($#x >= 6) && ($x[1] eq '.text')) {
+      $size = $x[2];
+      $vma = $x[3];
+      $file_offset = $x[5];
+      last;
+    }
+  }
+  close(OBJDUMP);
+
+  if (!defined($size)) {
+    return undef;
+  }
+
+  my $r = {};
+  $r->{size} = $size;
+  $r->{vma} = $vma;
+  $r->{file_offset} = $file_offset;
+
+  return $r;
+}
+
+# Parse text section header of a library using otool (on OS X)
+sub ParseTextSectionHeaderFromOtool {
+  my $lib = shift;
+
+  my $size = undef;
+  my $vma = undef;
+  my $file_offset = undef;
+  # Get otool output from the library file to figure out how to
+  # map between mapped addresses and addresses in the library.
+  my $command = ShellEscape($obj_tool_map{"otool"}, "-l", $lib);
+  open(OTOOL, "$command |") || error("$command: $!\n");
+  my $cmd = "";
+  my $sectname = "";
+  my $segname = "";
+  foreach my $line (<OTOOL>) {
+    $line =~ s/\r//g;      # turn windows-looking lines into unix-looking lines
+    # Load command <#>
+    #       cmd LC_SEGMENT
+    # [...]
+    # Section
+    #   sectname __text
+    #    segname __TEXT
+    #       addr 0x000009f8
+    #       size 0x00018b9e
+    #     offset 2552
+    #      align 2^2 (4)
+    # We will need to strip off the leading 0x from the hex addresses,
+    # and convert the offset into hex.
+    if ($line =~ /Load command/) {
+      $cmd = "";
+      $sectname = "";
+      $segname = "";
+    } elsif ($line =~ /Section/) {
+      $sectname = "";
+      $segname = "";
+    } elsif ($line =~ /cmd (\w+)/) {
+      $cmd = $1;
+    } elsif ($line =~ /sectname (\w+)/) {
+      $sectname = $1;
+    } elsif ($line =~ /segname (\w+)/) {
+      $segname = $1;
+    } elsif (!(($cmd eq "LC_SEGMENT" || $cmd eq "LC_SEGMENT_64") &&
+               $sectname eq "__text" &&
+               $segname eq "__TEXT")) {
+      next;
+    } elsif ($line =~ /\baddr 0x([0-9a-fA-F]+)/) {
+      $vma = $1;
+    } elsif ($line =~ /\bsize 0x([0-9a-fA-F]+)/) {
+      $size = $1;
+    } elsif ($line =~ /\boffset ([0-9]+)/) {
+      $file_offset = sprintf("%016x", $1);
+    }
+    if (defined($vma) && defined($size) && defined($file_offset)) {
+      last;
+    }
+  }
+  close(OTOOL);
+
+  if (!defined($vma) || !defined($size) || !defined($file_offset)) {
+     return undef;
+  }
+
+  my $r = {};
+  $r->{size} = $size;
+  $r->{vma} = $vma;
+  $r->{file_offset} = $file_offset;
+
+  return $r;
+}
+
+sub ParseTextSectionHeader {
+  # obj_tool_map("otool") is only defined if we're in a Mach-O environment
+  if (defined($obj_tool_map{"otool"})) {
+    my $r = ParseTextSectionHeaderFromOtool(@_);
+    if (defined($r)){
+      return $r;
+    }
+  }
+  # If otool doesn't work, or we don't have it, fall back to objdump
+  return ParseTextSectionHeaderFromObjdump(@_);
+}
+
+# Split /proc/pid/maps dump into a list of libraries
+sub ParseLibraries {
+  return if $main::use_symbol_page;  # We don't need libraries info.
+  my $prog = Cwd::abs_path(shift);
+  my $map = shift;
+  my $pcs = shift;
+
+  my $result = [];
+  my $h = "[a-f0-9]+";
+  my $zero_offset = HexExtend("0");
+
+  my $buildvar = "";
+  foreach my $l (split("\n", $map)) {
+    if ($l =~ m/^\s*build=(.*)$/) {
+      $buildvar = $1;
+    }
+
+    my $start;
+    my $finish;
+    my $offset;
+    my $lib;
+    if ($l =~ /^($h)-($h)\s+..x.\s+($h)\s+\S+:\S+\s+\d+\s+(\S+\.(so|dll|dylib|bundle)((\.\d+)+\w*(\.\d+){0,3})?)$/i) {
+      # Full line from /proc/self/maps.  Example:
+      #   40000000-40015000 r-xp 00000000 03:01 12845071   /lib/ld-2.3.2.so
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = HexExtend($3);
+      $lib = $4;
+      $lib =~ s|\\|/|g;     # turn windows-style paths into unix-style paths
+    } elsif ($l =~ /^\s*($h)-($h):\s*(\S+\.so(\.\d+)*)/) {
+      # Cooked line from DumpAddressMap.  Example:
+      #   40000000-40015000: /lib/ld-2.3.2.so
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = $zero_offset;
+      $lib = $3;
+    } elsif (($l =~ /^($h)-($h)\s+..x.\s+($h)\s+\S+:\S+\s+\d+\s+(\S+)$/i) && ($4 eq $prog)) {
+      # PIEs and address space randomization do not play well with our
+      # default assumption that main executable is at lowest
+      # addresses. So we're detecting main executable in
+      # /proc/self/maps as well.
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = HexExtend($3);
+      $lib = $4;
+      $lib =~ s|\\|/|g;     # turn windows-style paths into unix-style paths
+    }
+    # FreeBSD 10.0 virtual memory map /proc/curproc/map as defined in
+    # function procfs_doprocmap (sys/fs/procfs/procfs_map.c)
+    #
+    # Example:
+    # 0x800600000 0x80061a000 26 0 0xfffff800035a0000 r-x 75 33 0x1004 COW NC vnode /libexec/ld-elf.s
+    # o.1 NCH -1
+    elsif ($l =~ /^(0x$h)\s(0x$h)\s\d+\s\d+\s0x$h\sr-x\s\d+\s\d+\s0x\d+\s(COW|NCO)\s(NC|NNC)\svnode\s(\S+\.so(\.\d+)*)/) {
+      $start = HexExtend($1);
+      $finish = HexExtend($2);
+      $offset = $zero_offset;
+      $lib = FindLibrary($5);
+
+    } else {
+      next;
+    }
+
+    # Expand "$build" variable if available
+    $lib =~ s/\$build\b/$buildvar/g;
+
+    $lib = FindLibrary($lib);
+
+    # Check for pre-relocated libraries, which use pre-relocated symbol tables
+    # and thus require adjusting the offset that we'll use to translate
+    # VM addresses into symbol table addresses.
+    # Only do this if we're not going to fetch the symbol table from a
+    # debugging copy of the library.
+    if (!DebuggingLibrary($lib)) {
+      my $text = ParseTextSectionHeader($lib);
+      if (defined($text)) {
+         my $vma_offset = AddressSub($text->{vma}, $text->{file_offset});
+         $offset = AddressAdd($offset, $vma_offset);
+      }
+    }
+
+    if($main::opt_debug) { printf STDERR "$start:$finish ($offset) $lib\n"; }
+    push(@{$result}, [$lib, $start, $finish, $offset]);
+  }
+
+  # Append special entry for additional library (not relocated)
+  if ($main::opt_lib ne "") {
+    my $text = ParseTextSectionHeader($main::opt_lib);
+    if (defined($text)) {
+       my $start = $text->{vma};
+       my $finish = AddressAdd($start, $text->{size});
+
+       push(@{$result}, [$main::opt_lib, $start, $finish, $start]);
+    }
+  }
+
+  # Append special entry for the main program.  This covers
+  # 0..max_pc_value_seen, so that we assume pc values not found in one
+  # of the library ranges will be treated as coming from the main
+  # program binary.
+  my $min_pc = HexExtend("0");
+  my $max_pc = $min_pc;          # find the maximal PC value in any sample
+  foreach my $pc (keys(%{$pcs})) {
+    if (HexExtend($pc) gt $max_pc) { $max_pc = HexExtend($pc); }
+  }
+  push(@{$result}, [$prog, $min_pc, $max_pc, $zero_offset]);
+
+  return $result;
+}
+
+# Add two hex addresses of length $address_length.
+# Run jeprof --test for unit test if this is changed.
+sub AddressAdd {
+  my $addr1 = shift;
+  my $addr2 = shift;
+  my $sum;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $sum = (hex($addr1)+hex($addr2)) % (0x10000000 * 16);
+    return sprintf("%08x", $sum);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize carry handling.
+
+    if ($main::opt_debug and $main::opt_test) {
+      print STDERR "AddressAdd $addr1 + $addr2 = ";
+    }
+
+    my $a1 = substr($addr1,-7);
+    $addr1 = substr($addr1,0,-7);
+    my $a2 = substr($addr2,-7);
+    $addr2 = substr($addr2,0,-7);
+    $sum = hex($a1) + hex($a2);
+    my $c = 0;
+    if ($sum > 0xfffffff) {
+      $c = 1;
+      $sum -= 0x10000000;
+    }
+    my $r = sprintf("%07x", $sum);
+
+    $a1 = substr($addr1,-7);
+    $addr1 = substr($addr1,0,-7);
+    $a2 = substr($addr2,-7);
+    $addr2 = substr($addr2,0,-7);
+    $sum = hex($a1) + hex($a2) + $c;
+    $c = 0;
+    if ($sum > 0xfffffff) {
+      $c = 1;
+      $sum -= 0x10000000;
+    }
+    $r = sprintf("%07x", $sum) . $r;
+
+    $sum = hex($addr1) + hex($addr2) + $c;
+    if ($sum > 0xff) { $sum -= 0x100; }
+    $r = sprintf("%02x", $sum) . $r;
+
+    if ($main::opt_debug and $main::opt_test) { print STDERR "$r\n"; }
+
+    return $r;
+  }
+}
+
+
+# Subtract two hex addresses of length $address_length.
+# Run jeprof --test for unit test if this is changed.
+sub AddressSub {
+  my $addr1 = shift;
+  my $addr2 = shift;
+  my $diff;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $diff = (hex($addr1)-hex($addr2)) % (0x10000000 * 16);
+    return sprintf("%08x", $diff);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize borrow handling.
+    # if ($main::opt_debug) { print STDERR "AddressSub $addr1 - $addr2 = "; }
+
+    my $a1 = hex(substr($addr1,-7));
+    $addr1 = substr($addr1,0,-7);
+    my $a2 = hex(substr($addr2,-7));
+    $addr2 = substr($addr2,0,-7);
+    my $b = 0;
+    if ($a2 > $a1) {
+      $b = 1;
+      $a1 += 0x10000000;
+    }
+    $diff = $a1 - $a2;
+    my $r = sprintf("%07x", $diff);
+
+    $a1 = hex(substr($addr1,-7));
+    $addr1 = substr($addr1,0,-7);
+    $a2 = hex(substr($addr2,-7)) + $b;
+    $addr2 = substr($addr2,0,-7);
+    $b = 0;
+    if ($a2 > $a1) {
+      $b = 1;
+      $a1 += 0x10000000;
+    }
+    $diff = $a1 - $a2;
+    $r = sprintf("%07x", $diff) . $r;
+
+    $a1 = hex($addr1);
+    $a2 = hex($addr2) + $b;
+    if ($a2 > $a1) { $a1 += 0x100; }
+    $diff = $a1 - $a2;
+    $r = sprintf("%02x", $diff) . $r;
+
+    # if ($main::opt_debug) { print STDERR "$r\n"; }
+
+    return $r;
+  }
+}
+
+# Increment a hex addresses of length $address_length.
+# Run jeprof --test for unit test if this is changed.
+sub AddressInc {
+  my $addr = shift;
+  my $sum;
+
+  if ($address_length == 8) {
+    # Perl doesn't cope with wraparound arithmetic, so do it explicitly:
+    $sum = (hex($addr)+1) % (0x10000000 * 16);
+    return sprintf("%08x", $sum);
+
+  } else {
+    # Do the addition in 7-nibble chunks to trivialize carry handling.
+    # We are always doing this to step through the addresses in a function,
+    # and will almost never overflow the first chunk, so we check for this
+    # case and exit early.
+
+    # if ($main::opt_debug) { print STDERR "AddressInc $addr1 = "; }
+
+    my $a1 = substr($addr,-7);
+    $addr = substr($addr,0,-7);
+    $sum = hex($a1) + 1;
+    my $r = sprintf("%07x", $sum);
+    if ($sum <= 0xfffffff) {
+      $r = $addr . $r;
+      # if ($main::opt_debug) { print STDERR "$r\n"; }
+      return HexExtend($r);
+    } else {
+      $r = "0000000";
+    }
+
+    $a1 = substr($addr,-7);
+    $addr = substr($addr,0,-7);
+    $sum = hex($a1) + 1;
+    $r = sprintf("%07x", $sum) . $r;
+    if ($sum <= 0xfffffff) {
+      $r = $addr . $r;
+      # if ($main::opt_debug) { print STDERR "$r\n"; }
+      return HexExtend($r);
+    } else {
+      $r = "00000000000000";
+    }
+
+    $sum = hex($addr) + 1;
+    if ($sum > 0xff) { $sum -= 0x100; }
+    $r = sprintf("%02x", $sum) . $r;
+
+    # if ($main::opt_debug) { print STDERR "$r\n"; }
+    return $r;
+  }
+}
+
+# Extract symbols for all PC values found in profile
+sub ExtractSymbols {
+  my $libs = shift;
+  my $pcset = shift;
+
+  my $symbols = {};
+
+  # Map each PC value to the containing library.  To make this faster,
+  # we sort libraries by their starting pc value (highest first), and
+  # advance through the libraries as we advance the pc.  Sometimes the
+  # addresses of libraries may overlap with the addresses of the main
+  # binary, so to make sure the libraries 'win', we iterate over the
+  # libraries in reverse order (which assumes the binary doesn't start
+  # in the middle of a library, which seems a fair assumption).
+  my @pcs = (sort { $a cmp $b } keys(%{$pcset}));  # pcset is 0-extended strings
+  foreach my $lib (sort {$b->[1] cmp $a->[1]} @{$libs}) {
+    my $libname = $lib->[0];
+    my $start = $lib->[1];
+    my $finish = $lib->[2];
+    my $offset = $lib->[3];
+
+    # Use debug library if it exists
+    my $debug_libname = DebuggingLibrary($libname);
+    if ($debug_libname) {
+        $libname = $debug_libname;
+    }
+
+    # Get list of pcs that belong in this library.
+    my $contained = [];
+    my ($start_pc_index, $finish_pc_index);
+    # Find smallest finish_pc_index such that $finish < $pc[$finish_pc_index].
+    for ($finish_pc_index = $#pcs + 1; $finish_pc_index > 0;
+         $finish_pc_index--) {
+      last if $pcs[$finish_pc_index - 1] le $finish;
+    }
+    # Find smallest start_pc_index such that $start <= $pc[$start_pc_index].
+    for ($start_pc_index = $finish_pc_index; $start_pc_index > 0;
+         $start_pc_index--) {
+      last if $pcs[$start_pc_index - 1] lt $start;
+    }
+    # This keeps PC values higher than $pc[$finish_pc_index] in @pcs,
+    # in case there are overlaps in libraries and the main binary.
+    @{$contained} = splice(@pcs, $start_pc_index,
+                           $finish_pc_index - $start_pc_index);
+    # Map to symbols
+    MapToSymbols($libname, AddressSub($start, $offset), $contained, $symbols);
+  }
+
+  return $symbols;
+}
+
+# Map list of PC values to symbols for a given image
+sub MapToSymbols {
+  my $image = shift;
+  my $offset = shift;
+  my $pclist = shift;
+  my $symbols = shift;
+
+  my $debug = 0;
+
+  # Ignore empty binaries
+  if ($#{$pclist} < 0) { return; }
+
+  # Figure out the addr2line command to use
+  my $addr2line = $obj_tool_map{"addr2line"};
+  my $cmd = ShellEscape($addr2line, "-f", "-C", "-e", $image);
+  if (exists $obj_tool_map{"addr2line_pdb"}) {
+    $addr2line = $obj_tool_map{"addr2line_pdb"};
+    $cmd = ShellEscape($addr2line, "--demangle", "-f", "-C", "-e", $image);
+  }
+
+  # If "addr2line" isn't installed on the system at all, just use
+  # nm to get what info we can (function names, but not line numbers).
+  if (system(ShellEscape($addr2line, "--help") . " >$dev_null 2>&1") != 0) {
+    MapSymbolsWithNM($image, $offset, $pclist, $symbols);
+    return;
+  }
+
+  # "addr2line -i" can produce a variable number of lines per input
+  # address, with no separator that allows us to tell when data for
+  # the next address starts.  So we find the address for a special
+  # symbol (_fini) and interleave this address between all real
+  # addresses passed to addr2line.  The name of this special symbol
+  # can then be used as a separator.
+  $sep_address = undef;  # May be filled in by MapSymbolsWithNM()
+  my $nm_symbols = {};
+  MapSymbolsWithNM($image, $offset, $pclist, $nm_symbols);
+  if (defined($sep_address)) {
+    # Only add " -i" to addr2line if the binary supports it.
+    # addr2line --help returns 0, but not if it sees an unknown flag first.
+    if (system("$cmd -i --help >$dev_null 2>&1") == 0) {
+      $cmd .= " -i";
+    } else {
+      $sep_address = undef;   # no need for sep_address if we don't support -i
+    }
+  }
+
+  # Make file with all PC values with intervening 'sep_address' so
+  # that we can reliably detect the end of inlined function list
+  open(ADDRESSES, ">$main::tmpfile_sym") || error("$main::tmpfile_sym: $!\n");
+  if ($debug) { print("---- $image ---\n"); }
+  for (my $i = 0; $i <= $#{$pclist}; $i++) {
+    # addr2line always reads hex addresses, and does not need '0x' prefix.
+    if ($debug) { printf STDERR ("%s\n", $pclist->[$i]); }
+    printf ADDRESSES ("%s\n", AddressSub($pclist->[$i], $offset));
+    if (defined($sep_address)) {
+      printf ADDRESSES ("%s\n", $sep_address);
+    }
+  }
+  close(ADDRESSES);
+  if ($debug) {
+    print("----\n");
+    system("cat", $main::tmpfile_sym);
+    print("----\n");
+    system("$cmd < " . ShellEscape($main::tmpfile_sym));
+    print("----\n");
+  }
+
+  open(SYMBOLS, "$cmd <" . ShellEscape($main::tmpfile_sym) . " |")
+      || error("$cmd: $!\n");
+  my $count = 0;   # Index in pclist
+  while (<SYMBOLS>) {
+    # Read fullfunction and filelineinfo from next pair of lines
+    s/\r?\n$//g;
+    my $fullfunction = $_;
+    $_ = <SYMBOLS>;
+    s/\r?\n$//g;
+    my $filelinenum = $_;
+
+    if (defined($sep_address) && $fullfunction eq $sep_symbol) {
+      # Terminating marker for data for this address
+      $count++;
+      next;
+    }
+
+    $filelinenum =~ s|\\|/|g; # turn windows-style paths into unix-style paths
+
+    my $pcstr = $pclist->[$count];
+    my $function = ShortFunctionName($fullfunction);
+    my $nms = $nm_symbols->{$pcstr};
+    if (defined($nms)) {
+      if ($fullfunction eq '??') {
+        # nm found a symbol for us.
+        $function = $nms->[0];
+        $fullfunction = $nms->[2];
+      } else {
+	# MapSymbolsWithNM tags each routine with its starting address,
+	# useful in case the image has multiple occurrences of this
+	# routine.  (It uses a syntax that resembles template paramters,
+	# that are automatically stripped out by ShortFunctionName().)
+	# addr2line does not provide the same information.  So we check
+	# if nm disambiguated our symbol, and if so take the annotated
+	# (nm) version of the routine-name.  TODO(csilvers): this won't
+	# catch overloaded, inlined symbols, which nm doesn't see.
+	# Better would be to do a check similar to nm's, in this fn.
+	if ($nms->[2] =~ m/^\Q$function\E/) {  # sanity check it's the right fn
+	  $function = $nms->[0];
+	  $fullfunction = $nms->[2];
+	}
+      }
+    }
+
+    # Prepend to accumulated symbols for pcstr
+    # (so that caller comes before callee)
+    my $sym = $symbols->{$pcstr};
+    if (!defined($sym)) {
+      $sym = [];
+      $symbols->{$pcstr} = $sym;
+    }
+    unshift(@{$sym}, $function, $filelinenum, $fullfunction);
+    if ($debug) { printf STDERR ("%s => [%s]\n", $pcstr, join(" ", @{$sym})); }
+    if (!defined($sep_address)) {
+      # Inlining is off, so this entry ends immediately
+      $count++;
+    }
+  }
+  close(SYMBOLS);
+}
+
+# Use nm to map the list of referenced PCs to symbols.  Return true iff we
+# are able to read procedure information via nm.
+sub MapSymbolsWithNM {
+  my $image = shift;
+  my $offset = shift;
+  my $pclist = shift;
+  my $symbols = shift;
+
+  # Get nm output sorted by increasing address
+  my $symbol_table = GetProcedureBoundaries($image, ".");
+  if (!%{$symbol_table}) {
+    return 0;
+  }
+  # Start addresses are already the right length (8 or 16 hex digits).
+  my @names = sort { $symbol_table->{$a}->[0] cmp $symbol_table->{$b}->[0] }
+    keys(%{$symbol_table});
+
+  if ($#names < 0) {
+    # No symbols: just use addresses
+    foreach my $pc (@{$pclist}) {
+      my $pcstr = "0x" . $pc;
+      $symbols->{$pc} = [$pcstr, "?", $pcstr];
+    }
+    return 0;
+  }
+
+  # Sort addresses so we can do a join against nm output
+  my $index = 0;
+  my $fullname = $names[0];
+  my $name = ShortFunctionName($fullname);
+  foreach my $pc (sort { $a cmp $b } @{$pclist}) {
+    # Adjust for mapped offset
+    my $mpc = AddressSub($pc, $offset);
+    while (($index < $#names) && ($mpc ge $symbol_table->{$fullname}->[1])){
+      $index++;
+      $fullname = $names[$index];
+      $name = ShortFunctionName($fullname);
+    }
+    if ($mpc lt $symbol_table->{$fullname}->[1]) {
+      $symbols->{$pc} = [$name, "?", $fullname];
+    } else {
+      my $pcstr = "0x" . $pc;
+      $symbols->{$pc} = [$pcstr, "?", $pcstr];
+    }
+  }
+  return 1;
+}
+
+sub ShortFunctionName {
+  my $function = shift;
+  while ($function =~ s/\([^()]*\)(\s*const)?//g) { }   # Argument types
+  while ($function =~ s/<[^<>]*>//g)  { }    # Remove template arguments
+  $function =~ s/^.*\s+(\w+::)/$1/;          # Remove leading type
+  return $function;
+}
+
+# Trim overly long symbols found in disassembler output
+sub CleanDisassembly {
+  my $d = shift;
+  while ($d =~ s/\([^()%]*\)(\s*const)?//g) { } # Argument types, not (%rax)
+  while ($d =~ s/(\w+)<[^<>]*>/$1/g)  { }       # Remove template arguments
+  return $d;
+}
+
+# Clean file name for display
+sub CleanFileName {
+  my ($f) = @_;
+  $f =~ s|^/proc/self/cwd/||;
+  $f =~ s|^\./||;
+  return $f;
+}
+
+# Make address relative to section and clean up for display
+sub UnparseAddress {
+  my ($offset, $address) = @_;
+  $address = AddressSub($address, $offset);
+  $address =~ s/^0x//;
+  $address =~ s/^0*//;
+  return $address;
+}
+
+##### Miscellaneous #####
+
+# Find the right versions of the above object tools to use.  The
+# argument is the program file being analyzed, and should be an ELF
+# 32-bit or ELF 64-bit executable file.  The location of the tools
+# is determined by considering the following options in this order:
+#   1) --tools option, if set
+#   2) JEPROF_TOOLS environment variable, if set
+#   3) the environment
+sub ConfigureObjTools {
+  my $prog_file = shift;
+
+  # Check for the existence of $prog_file because /usr/bin/file does not
+  # predictably return error status in prod.
+  (-e $prog_file)  || error("$prog_file does not exist.\n");
+
+  my $file_type = undef;
+  if (-e "/usr/bin/file") {
+    # Follow symlinks (at least for systems where "file" supports that).
+    my $escaped_prog_file = ShellEscape($prog_file);
+    $file_type = `/usr/bin/file -L $escaped_prog_file 2>$dev_null ||
+                  /usr/bin/file $escaped_prog_file`;
+  } elsif ($^O == "MSWin32") {
+    $file_type = "MS Windows";
+  } else {
+    print STDERR "WARNING: Can't determine the file type of $prog_file";
+  }
+
+  if ($file_type =~ /64-bit/) {
+    # Change $address_length to 16 if the program file is ELF 64-bit.
+    # We can't detect this from many (most?) heap or lock contention
+    # profiles, since the actual addresses referenced are generally in low
+    # memory even for 64-bit programs.
+    $address_length = 16;
+  }
+
+  if ($file_type =~ /MS Windows/) {
+    # For windows, we provide a version of nm and addr2line as part of
+    # the opensource release, which is capable of parsing
+    # Windows-style PDB executables.  It should live in the path, or
+    # in the same directory as jeprof.
+    $obj_tool_map{"nm_pdb"} = "nm-pdb";
+    $obj_tool_map{"addr2line_pdb"} = "addr2line-pdb";
+  }
+
+  if ($file_type =~ /Mach-O/) {
+    # OS X uses otool to examine Mach-O files, rather than objdump.
+    $obj_tool_map{"otool"} = "otool";
+    $obj_tool_map{"addr2line"} = "false";  # no addr2line
+    $obj_tool_map{"objdump"} = "false";  # no objdump
+  }
+
+  # Go fill in %obj_tool_map with the pathnames to use:
+  foreach my $tool (keys %obj_tool_map) {
+    $obj_tool_map{$tool} = ConfigureTool($obj_tool_map{$tool});
+  }
+}
+
+# Returns the path of a caller-specified object tool.  If --tools or
+# JEPROF_TOOLS are specified, then returns the full path to the tool
+# with that prefix.  Otherwise, returns the path unmodified (which
+# means we will look for it on PATH).
+sub ConfigureTool {
+  my $tool = shift;
+  my $path;
+
+  # --tools (or $JEPROF_TOOLS) is a comma separated list, where each
+  # item is either a) a pathname prefix, or b) a map of the form
+  # <tool>:<path>.  First we look for an entry of type (b) for our
+  # tool.  If one is found, we use it.  Otherwise, we consider all the
+  # pathname prefixes in turn, until one yields an existing file.  If
+  # none does, we use a default path.
+  my $tools = $main::opt_tools || $ENV{"JEPROF_TOOLS"} || "";
+  if ($tools =~ m/(,|^)\Q$tool\E:([^,]*)/) {
+    $path = $2;
+    # TODO(csilvers): sanity-check that $path exists?  Hard if it's relative.
+  } elsif ($tools ne '') {
+    foreach my $prefix (split(',', $tools)) {
+      next if ($prefix =~ /:/);    # ignore "tool:fullpath" entries in the list
+      if (-x $prefix . $tool) {
+        $path = $prefix . $tool;
+        last;
+      }
+    }
+    if (!$path) {
+      error("No '$tool' found with prefix specified by " .
+            "--tools (or \$JEPROF_TOOLS) '$tools'\n");
+    }
+  } else {
+    # ... otherwise use the version that exists in the same directory as
+    # jeprof.  If there's nothing there, use $PATH.
+    $0 =~ m,[^/]*$,;     # this is everything after the last slash
+    my $dirname = $`;    # this is everything up to and including the last slash
+    if (-x "$dirname$tool") {
+      $path = "$dirname$tool";
+    } else {
+      $path = $tool;
+    }
+  }
+  if ($main::opt_debug) { print STDERR "Using '$path' for '$tool'.\n"; }
+  return $path;
+}
+
+sub ShellEscape {
+  my @escaped_words = ();
+  foreach my $word (@_) {
+    my $escaped_word = $word;
+    if ($word =~ m![^a-zA-Z0-9/.,_=-]!) {  # check for anything not in whitelist
+      $escaped_word =~ s/'/'\\''/;
+      $escaped_word = "'$escaped_word'";
+    }
+    push(@escaped_words, $escaped_word);
+  }
+  return join(" ", @escaped_words);
+}
+
+sub cleanup {
+  unlink($main::tmpfile_sym);
+  unlink(keys %main::tempnames);
+
+  # We leave any collected profiles in $HOME/jeprof in case the user wants
+  # to look at them later.  We print a message informing them of this.
+  if ((scalar(@main::profile_files) > 0) &&
+      defined($main::collected_profile)) {
+    if (scalar(@main::profile_files) == 1) {
+      print STDERR "Dynamically gathered profile is in $main::collected_profile\n";
+    }
+    print STDERR "If you want to investigate this profile further, you can do:\n";
+    print STDERR "\n";
+    print STDERR "  jeprof \\\n";
+    print STDERR "    $main::prog \\\n";
+    print STDERR "    $main::collected_profile\n";
+    print STDERR "\n";
+  }
+}
+
+sub sighandler {
+  cleanup();
+  exit(1);
+}
+
+sub error {
+  my $msg = shift;
+  print STDERR $msg;
+  cleanup();
+  exit(1);
+}
+
+
+# Run $nm_command and get all the resulting procedure boundaries whose
+# names match "$regexp" and returns them in a hashtable mapping from
+# procedure name to a two-element vector of [start address, end address]
+sub GetProcedureBoundariesViaNm {
+  my $escaped_nm_command = shift;    # shell-escaped
+  my $regexp = shift;
+
+  my $symbol_table = {};
+  open(NM, "$escaped_nm_command |") || error("$escaped_nm_command: $!\n");
+  my $last_start = "0";
+  my $routine = "";
+  while (<NM>) {
+    s/\r//g;         # turn windows-looking lines into unix-looking lines
+    if (m/^\s*([0-9a-f]+) (.) (..*)/) {
+      my $start_val = $1;
+      my $type = $2;
+      my $this_routine = $3;
+
+      # It's possible for two symbols to share the same address, if
+      # one is a zero-length variable (like __start_google_malloc) or
+      # one symbol is a weak alias to another (like __libc_malloc).
+      # In such cases, we want to ignore all values except for the
+      # actual symbol, which in nm-speak has type "T".  The logic
+      # below does this, though it's a bit tricky: what happens when
+      # we have a series of lines with the same address, is the first
+      # one gets queued up to be processed.  However, it won't
+      # *actually* be processed until later, when we read a line with
+      # a different address.  That means that as long as we're reading
+      # lines with the same address, we have a chance to replace that
+      # item in the queue, which we do whenever we see a 'T' entry --
+      # that is, a line with type 'T'.  If we never see a 'T' entry,
+      # we'll just go ahead and process the first entry (which never
+      # got touched in the queue), and ignore the others.
+      if ($start_val eq $last_start && $type =~ /t/i) {
+        # We are the 'T' symbol at this address, replace previous symbol.
+        $routine = $this_routine;
+        next;
+      } elsif ($start_val eq $last_start) {
+        # We're not the 'T' symbol at this address, so ignore us.
+        next;
+      }
+
+      if ($this_routine eq $sep_symbol) {
+        $sep_address = HexExtend($start_val);
+      }
+
+      # Tag this routine with the starting address in case the image
+      # has multiple occurrences of this routine.  We use a syntax
+      # that resembles template parameters that are automatically
+      # stripped out by ShortFunctionName()
+      $this_routine .= "<$start_val>";
+
+      if (defined($routine) && $routine =~ m/$regexp/) {
+        $symbol_table->{$routine} = [HexExtend($last_start),
+                                     HexExtend($start_val)];
+      }
+      $last_start = $start_val;
+      $routine = $this_routine;
+    } elsif (m/^Loaded image name: (.+)/) {
+      # The win32 nm workalike emits information about the binary it is using.
+      if ($main::opt_debug) { print STDERR "Using Image $1\n"; }
+    } elsif (m/^PDB file name: (.+)/) {
+      # The win32 nm workalike emits information about the pdb it is using.
+      if ($main::opt_debug) { print STDERR "Using PDB $1\n"; }
+    }
+  }
+  close(NM);
+  # Handle the last line in the nm output.  Unfortunately, we don't know
+  # how big this last symbol is, because we don't know how big the file
+  # is.  For now, we just give it a size of 0.
+  # TODO(csilvers): do better here.
+  if (defined($routine) && $routine =~ m/$regexp/) {
+    $symbol_table->{$routine} = [HexExtend($last_start),
+                                 HexExtend($last_start)];
+  }
+  return $symbol_table;
+}
+
+# Gets the procedure boundaries for all routines in "$image" whose names
+# match "$regexp" and returns them in a hashtable mapping from procedure
+# name to a two-element vector of [start address, end address].
+# Will return an empty map if nm is not installed or not working properly.
+sub GetProcedureBoundaries {
+  my $image = shift;
+  my $regexp = shift;
+
+  # If $image doesn't start with /, then put ./ in front of it.  This works
+  # around an obnoxious bug in our probing of nm -f behavior.
+  # "nm -f $image" is supposed to fail on GNU nm, but if:
+  #
+  # a. $image starts with [BbSsPp] (for example, bin/foo/bar), AND
+  # b. you have a.out in your current directory (a not uncommon occurence)
+  #
+  # then "nm -f $image" succeeds because -f only looks at the first letter of
+  # the argument, which looks valid because it's [BbSsPp], and then since
+  # there's no image provided, it looks for a.out and finds it.
+  #
+  # This regex makes sure that $image starts with . or /, forcing the -f
+  # parsing to fail since . and / are not valid formats.
+  $image =~ s#^[^/]#./$&#;
+
+  # For libc libraries, the copy in /usr/lib/debug contains debugging symbols
+  my $debugging = DebuggingLibrary($image);
+  if ($debugging) {
+    $image = $debugging;
+  }
+
+  my $nm = $obj_tool_map{"nm"};
+  my $cppfilt = $obj_tool_map{"c++filt"};
+
+  # nm can fail for two reasons: 1) $image isn't a debug library; 2) nm
+  # binary doesn't support --demangle.  In addition, for OS X we need
+  # to use the -f flag to get 'flat' nm output (otherwise we don't sort
+  # properly and get incorrect results).  Unfortunately, GNU nm uses -f
+  # in an incompatible way.  So first we test whether our nm supports
+  # --demangle and -f.
+  my $demangle_flag = "";
+  my $cppfilt_flag = "";
+  my $to_devnull = ">$dev_null 2>&1";
+  if (system(ShellEscape($nm, "--demangle", $image) . $to_devnull) == 0) {
+    # In this mode, we do "nm --demangle <foo>"
+    $demangle_flag = "--demangle";
+    $cppfilt_flag = "";
+  } elsif (system(ShellEscape($cppfilt, $image) . $to_devnull) == 0) {
+    # In this mode, we do "nm <foo> | c++filt"
+    $cppfilt_flag = " | " . ShellEscape($cppfilt);
+  };
+  my $flatten_flag = "";
+  if (system(ShellEscape($nm, "-f", $image) . $to_devnull) == 0) {
+    $flatten_flag = "-f";
+  }
+
+  # Finally, in the case $imagie isn't a debug library, we try again with
+  # -D to at least get *exported* symbols.  If we can't use --demangle,
+  # we use c++filt instead, if it exists on this system.
+  my @nm_commands = (ShellEscape($nm, "-n", $flatten_flag, $demangle_flag,
+                                 $image) . " 2>$dev_null $cppfilt_flag",
+                     ShellEscape($nm, "-D", "-n", $flatten_flag, $demangle_flag,
+                                 $image) . " 2>$dev_null $cppfilt_flag",
+                     # 6nm is for Go binaries
+                     ShellEscape("6nm", "$image") . " 2>$dev_null | sort",
+                     );
+
+  # If the executable is an MS Windows PDB-format executable, we'll
+  # have set up obj_tool_map("nm_pdb").  In this case, we actually
+  # want to use both unix nm and windows-specific nm_pdb, since
+  # PDB-format executables can apparently include dwarf .o files.
+  if (exists $obj_tool_map{"nm_pdb"}) {
+    push(@nm_commands,
+         ShellEscape($obj_tool_map{"nm_pdb"}, "--demangle", $image)
+         . " 2>$dev_null");
+  }
+
+  foreach my $nm_command (@nm_commands) {
+    my $symbol_table = GetProcedureBoundariesViaNm($nm_command, $regexp);
+    return $symbol_table if (%{$symbol_table});
+  }
+  my $symbol_table = {};
+  return $symbol_table;
+}
+
+
+# The test vectors for AddressAdd/Sub/Inc are 8-16-nibble hex strings.
+# To make them more readable, we add underscores at interesting places.
+# This routine removes the underscores, producing the canonical representation
+# used by jeprof to represent addresses, particularly in the tested routines.
+sub CanonicalHex {
+  my $arg = shift;
+  return join '', (split '_',$arg);
+}
+
+
+# Unit test for AddressAdd:
+sub AddressAddUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressAddUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressAdd ($row->[0], $row->[1]);
+    if ($sum ne $row->[2]) {
+      printf STDERR "ERROR: %s != %s + %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[2];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressAdd 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressAdd (CanonicalHex($row->[0]), CanonicalHex($row->[1]));
+    my $expected = join '', (split '_',$row->[2]);
+    if ($sum ne CanonicalHex($row->[2])) {
+      printf STDERR "ERROR: %s != %s + %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[2];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressAdd 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Unit test for AddressSub:
+sub AddressSubUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressSubUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressSub ($row->[0], $row->[1]);
+    if ($sum ne $row->[3]) {
+      printf STDERR "ERROR: %s != %s - %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[3];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressSub 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressSub (CanonicalHex($row->[0]), CanonicalHex($row->[1]));
+    if ($sum ne CanonicalHex($row->[3])) {
+      printf STDERR "ERROR: %s != %s - %s = %s\n", $sum,
+             $row->[0], $row->[1], $row->[3];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressSub 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Unit test for AddressInc:
+sub AddressIncUnitTest {
+  my $test_data_8 = shift;
+  my $test_data_16 = shift;
+  my $error_count = 0;
+  my $fail_count = 0;
+  my $pass_count = 0;
+  # print STDERR "AddressIncUnitTest: ", 1+$#{$test_data_8}, " tests\n";
+
+  # First a few 8-nibble addresses.  Note that this implementation uses
+  # plain old arithmetic, so a quick sanity check along with verifying what
+  # happens to overflow (we want it to wrap):
+  $address_length = 8;
+  foreach my $row (@{$test_data_8}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressInc ($row->[0]);
+    if ($sum ne $row->[4]) {
+      printf STDERR "ERROR: %s != %s + 1 = %s\n", $sum,
+             $row->[0], $row->[4];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressInc 32-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count = $fail_count;
+  $fail_count = 0;
+  $pass_count = 0;
+
+  # Now 16-nibble addresses.
+  $address_length = 16;
+  foreach my $row (@{$test_data_16}) {
+    if ($main::opt_debug and $main::opt_test) { print STDERR "@{$row}\n"; }
+    my $sum = AddressInc (CanonicalHex($row->[0]));
+    if ($sum ne CanonicalHex($row->[4])) {
+      printf STDERR "ERROR: %s != %s + 1 = %s\n", $sum,
+             $row->[0], $row->[4];
+      ++$fail_count;
+    } else {
+      ++$pass_count;
+    }
+  }
+  printf STDERR "AddressInc 64-bit tests: %d passes, %d failures\n",
+         $pass_count, $fail_count;
+  $error_count += $fail_count;
+
+  return $error_count;
+}
+
+
+# Driver for unit tests.
+# Currently just the address add/subtract/increment routines for 64-bit.
+sub RunUnitTests {
+  my $error_count = 0;
+
+  # This is a list of tuples [a, b, a+b, a-b, a+1]
+  my $unit_test_data_8 = [
+    [qw(aaaaaaaa 50505050 fafafafa 5a5a5a5a aaaaaaab)],
+    [qw(50505050 aaaaaaaa fafafafa a5a5a5a6 50505051)],
+    [qw(ffffffff aaaaaaaa aaaaaaa9 55555555 00000000)],
+    [qw(00000001 ffffffff 00000000 00000002 00000002)],
+    [qw(00000001 fffffff0 fffffff1 00000011 00000002)],
+  ];
+  my $unit_test_data_16 = [
+    # The implementation handles data in 7-nibble chunks, so those are the
+    # interesting boundaries.
+    [qw(aaaaaaaa 50505050
+        00_000000f_afafafa 00_0000005_a5a5a5a 00_000000a_aaaaaab)],
+    [qw(50505050 aaaaaaaa
+        00_000000f_afafafa ff_ffffffa_5a5a5a6 00_0000005_0505051)],
+    [qw(ffffffff aaaaaaaa
+        00_000001a_aaaaaa9 00_0000005_5555555 00_0000010_0000000)],
+    [qw(00000001 ffffffff
+        00_0000010_0000000 ff_ffffff0_0000002 00_0000000_0000002)],
+    [qw(00000001 fffffff0
+        00_000000f_ffffff1 ff_ffffff0_0000011 00_0000000_0000002)],
+
+    [qw(00_a00000a_aaaaaaa 50505050
+        00_a00000f_afafafa 00_a000005_a5a5a5a 00_a00000a_aaaaaab)],
+    [qw(0f_fff0005_0505050 aaaaaaaa
+        0f_fff000f_afafafa 0f_ffefffa_5a5a5a6 0f_fff0005_0505051)],
+    [qw(00_000000f_fffffff 01_800000a_aaaaaaa
+        01_800001a_aaaaaa9 fe_8000005_5555555 00_0000010_0000000)],
+    [qw(00_0000000_0000001 ff_fffffff_fffffff
+        00_0000000_0000000 00_0000000_0000002 00_0000000_0000002)],
+    [qw(00_0000000_0000001 ff_fffffff_ffffff0
+        ff_fffffff_ffffff1 00_0000000_0000011 00_0000000_0000002)],
+  ];
+
+  $error_count += AddressAddUnitTest($unit_test_data_8, $unit_test_data_16);
+  $error_count += AddressSubUnitTest($unit_test_data_8, $unit_test_data_16);
+  $error_count += AddressIncUnitTest($unit_test_data_8, $unit_test_data_16);
+  if ($error_count > 0) {
+    print STDERR $error_count, " errors: FAILED\n";
+  } else {
+    print STDERR "PASS\n";
+  }
+  exit ($error_count);
+}
diff --git a/memkind-1.10.0/jemalloc/build-aux/config.guess b/memkind-1.10.0/jemalloc/build-aux/config.guess
new file mode 100755
index 0000000..2e9ad7f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/build-aux/config.guess
@@ -0,0 +1,1462 @@
+#! /bin/sh
+# Attempt to guess a canonical system name.
+#   Copyright 1992-2016 Free Software Foundation, Inc.
+
+timestamp='2016-10-02'
+
+# This file is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+#
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that
+# program.  This Exception is an additional permission under section 7
+# of the GNU General Public License, version 3 ("GPLv3").
+#
+# Originally written by Per Bothner; maintained since 2000 by Ben Elliston.
+#
+# You can get the latest version of this script from:
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess
+#
+# Please send patches to <config-patches@gnu.org>.
+
+
+me=`echo "$0" | sed -e 's,.*/,,'`
+
+usage="\
+Usage: $0 [OPTION]
+
+Output the configuration name of the system \`$me' is run on.
+
+Operation modes:
+  -h, --help         print this help, then exit
+  -t, --time-stamp   print date of last modification, then exit
+  -v, --version      print version number, then exit
+
+Report bugs and patches to <config-patches@gnu.org>."
+
+version="\
+GNU config.guess ($timestamp)
+
+Originally written by Per Bothner.
+Copyright 1992-2016 Free Software Foundation, Inc.
+
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
+
+help="
+Try \`$me --help' for more information."
+
+# Parse command line
+while test $# -gt 0 ; do
+  case $1 in
+    --time-stamp | --time* | -t )
+       echo "$timestamp" ; exit ;;
+    --version | -v )
+       echo "$version" ; exit ;;
+    --help | --h* | -h )
+       echo "$usage"; exit ;;
+    -- )     # Stop option processing
+       shift; break ;;
+    - )	# Use stdin as input.
+       break ;;
+    -* )
+       echo "$me: invalid option $1$help" >&2
+       exit 1 ;;
+    * )
+       break ;;
+  esac
+done
+
+if test $# != 0; then
+  echo "$me: too many arguments$help" >&2
+  exit 1
+fi
+
+trap 'exit 1' 1 2 15
+
+# CC_FOR_BUILD -- compiler used by this script. Note that the use of a
+# compiler to aid in system detection is discouraged as it requires
+# temporary files to be created and, as you can see below, it is a
+# headache to deal with in a portable fashion.
+
+# Historically, `CC_FOR_BUILD' used to be named `HOST_CC'. We still
+# use `HOST_CC' if defined, but it is deprecated.
+
+# Portable tmp directory creation inspired by the Autoconf team.
+
+set_cc_for_build='
+trap "exitcode=\$?; (rm -f \$tmpfiles 2>/dev/null; rmdir \$tmp 2>/dev/null) && exit \$exitcode" 0 ;
+trap "rm -f \$tmpfiles 2>/dev/null; rmdir \$tmp 2>/dev/null; exit 1" 1 2 13 15 ;
+: ${TMPDIR=/tmp} ;
+ { tmp=`(umask 077 && mktemp -d "$TMPDIR/cgXXXXXX") 2>/dev/null` && test -n "$tmp" && test -d "$tmp" ; } ||
+ { test -n "$RANDOM" && tmp=$TMPDIR/cg$$-$RANDOM && (umask 077 && mkdir $tmp) ; } ||
+ { tmp=$TMPDIR/cg-$$ && (umask 077 && mkdir $tmp) && echo "Warning: creating insecure temp directory" >&2 ; } ||
+ { echo "$me: cannot create a temporary directory in $TMPDIR" >&2 ; exit 1 ; } ;
+dummy=$tmp/dummy ;
+tmpfiles="$dummy.c $dummy.o $dummy.rel $dummy" ;
+case $CC_FOR_BUILD,$HOST_CC,$CC in
+ ,,)    echo "int x;" > $dummy.c ;
+	for c in cc gcc c89 c99 ; do
+	  if ($c -c -o $dummy.o $dummy.c) >/dev/null 2>&1 ; then
+	     CC_FOR_BUILD="$c"; break ;
+	  fi ;
+	done ;
+	if test x"$CC_FOR_BUILD" = x ; then
+	  CC_FOR_BUILD=no_compiler_found ;
+	fi
+	;;
+ ,,*)   CC_FOR_BUILD=$CC ;;
+ ,*,*)  CC_FOR_BUILD=$HOST_CC ;;
+esac ; set_cc_for_build= ;'
+
+# This is needed to find uname on a Pyramid OSx when run in the BSD universe.
+# (ghazi@noc.rutgers.edu 1994-08-24)
+if (test -f /.attbin/uname) >/dev/null 2>&1 ; then
+	PATH=$PATH:/.attbin ; export PATH
+fi
+
+UNAME_MACHINE=`(uname -m) 2>/dev/null` || UNAME_MACHINE=unknown
+UNAME_RELEASE=`(uname -r) 2>/dev/null` || UNAME_RELEASE=unknown
+UNAME_SYSTEM=`(uname -s) 2>/dev/null`  || UNAME_SYSTEM=unknown
+UNAME_VERSION=`(uname -v) 2>/dev/null` || UNAME_VERSION=unknown
+
+case "${UNAME_SYSTEM}" in
+Linux|GNU|GNU/*)
+	# If the system lacks a compiler, then just pick glibc.
+	# We could probably try harder.
+	LIBC=gnu
+
+	eval $set_cc_for_build
+	cat <<-EOF > $dummy.c
+	#include <features.h>
+	#if defined(__UCLIBC__)
+	LIBC=uclibc
+	#elif defined(__dietlibc__)
+	LIBC=dietlibc
+	#else
+	LIBC=gnu
+	#endif
+	EOF
+	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^LIBC' | sed 's, ,,g'`
+	;;
+esac
+
+# Note: order is significant - the case branches are not exclusive.
+
+case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
+    *:NetBSD:*:*)
+	# NetBSD (nbsd) targets should (where applicable) match one or
+	# more of the tuples: *-*-netbsdelf*, *-*-netbsdaout*,
+	# *-*-netbsdecoff* and *-*-netbsd*.  For targets that recently
+	# switched to ELF, *-*-netbsd* would select the old
+	# object file format.  This provides both forward
+	# compatibility and a consistent mechanism for selecting the
+	# object file format.
+	#
+	# Note: NetBSD doesn't particularly care about the vendor
+	# portion of the name.  We always set it to "unknown".
+	sysctl="sysctl -n hw.machine_arch"
+	UNAME_MACHINE_ARCH=`(uname -p 2>/dev/null || \
+	    /sbin/$sysctl 2>/dev/null || \
+	    /usr/sbin/$sysctl 2>/dev/null || \
+	    echo unknown)`
+	case "${UNAME_MACHINE_ARCH}" in
+	    armeb) machine=armeb-unknown ;;
+	    arm*) machine=arm-unknown ;;
+	    sh3el) machine=shl-unknown ;;
+	    sh3eb) machine=sh-unknown ;;
+	    sh5el) machine=sh5le-unknown ;;
+	    earmv*)
+		arch=`echo ${UNAME_MACHINE_ARCH} | sed -e 's,^e\(armv[0-9]\).*$,\1,'`
+		endian=`echo ${UNAME_MACHINE_ARCH} | sed -ne 's,^.*\(eb\)$,\1,p'`
+		machine=${arch}${endian}-unknown
+		;;
+	    *) machine=${UNAME_MACHINE_ARCH}-unknown ;;
+	esac
+	# The Operating System including object format, if it has switched
+	# to ELF recently (or will in the future) and ABI.
+	case "${UNAME_MACHINE_ARCH}" in
+	    earm*)
+		os=netbsdelf
+		;;
+	    arm*|i386|m68k|ns32k|sh3*|sparc|vax)
+		eval $set_cc_for_build
+		if echo __ELF__ | $CC_FOR_BUILD -E - 2>/dev/null \
+			| grep -q __ELF__
+		then
+		    # Once all utilities can be ECOFF (netbsdecoff) or a.out (netbsdaout).
+		    # Return netbsd for either.  FIX?
+		    os=netbsd
+		else
+		    os=netbsdelf
+		fi
+		;;
+	    *)
+		os=netbsd
+		;;
+	esac
+	# Determine ABI tags.
+	case "${UNAME_MACHINE_ARCH}" in
+	    earm*)
+		expr='s/^earmv[0-9]/-eabi/;s/eb$//'
+		abi=`echo ${UNAME_MACHINE_ARCH} | sed -e "$expr"`
+		;;
+	esac
+	# The OS release
+	# Debian GNU/NetBSD machines have a different userland, and
+	# thus, need a distinct triplet. However, they do not need
+	# kernel version information, so it can be replaced with a
+	# suitable tag, in the style of linux-gnu.
+	case "${UNAME_VERSION}" in
+	    Debian*)
+		release='-gnu'
+		;;
+	    *)
+		release=`echo ${UNAME_RELEASE} | sed -e 's/[-_].*//' | cut -d. -f1,2`
+		;;
+	esac
+	# Since CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM:
+	# contains redundant information, the shorter form:
+	# CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM is used.
+	echo "${machine}-${os}${release}${abi}"
+	exit ;;
+    *:Bitrig:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/Bitrig.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-bitrig${UNAME_RELEASE}
+	exit ;;
+    *:OpenBSD:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/OpenBSD.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-openbsd${UNAME_RELEASE}
+	exit ;;
+    *:LibertyBSD:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/^.*BSD\.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-libertybsd${UNAME_RELEASE}
+	exit ;;
+    *:ekkoBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-ekkobsd${UNAME_RELEASE}
+	exit ;;
+    *:SolidBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-solidbsd${UNAME_RELEASE}
+	exit ;;
+    macppc:MirBSD:*:*)
+	echo powerpc-unknown-mirbsd${UNAME_RELEASE}
+	exit ;;
+    *:MirBSD:*:*)
+	echo ${UNAME_MACHINE}-unknown-mirbsd${UNAME_RELEASE}
+	exit ;;
+    *:Sortix:*:*)
+	echo ${UNAME_MACHINE}-unknown-sortix
+	exit ;;
+    alpha:OSF1:*:*)
+	case $UNAME_RELEASE in
+	*4.0)
+		UNAME_RELEASE=`/usr/sbin/sizer -v | awk '{print $3}'`
+		;;
+	*5.*)
+		UNAME_RELEASE=`/usr/sbin/sizer -v | awk '{print $4}'`
+		;;
+	esac
+	# According to Compaq, /usr/sbin/psrinfo has been available on
+	# OSF/1 and Tru64 systems produced since 1995.  I hope that
+	# covers most systems running today.  This code pipes the CPU
+	# types through head -n 1, so we only detect the type of CPU 0.
+	ALPHA_CPU_TYPE=`/usr/sbin/psrinfo -v | sed -n -e 's/^  The alpha \(.*\) processor.*$/\1/p' | head -n 1`
+	case "$ALPHA_CPU_TYPE" in
+	    "EV4 (21064)")
+		UNAME_MACHINE=alpha ;;
+	    "EV4.5 (21064)")
+		UNAME_MACHINE=alpha ;;
+	    "LCA4 (21066/21068)")
+		UNAME_MACHINE=alpha ;;
+	    "EV5 (21164)")
+		UNAME_MACHINE=alphaev5 ;;
+	    "EV5.6 (21164A)")
+		UNAME_MACHINE=alphaev56 ;;
+	    "EV5.6 (21164PC)")
+		UNAME_MACHINE=alphapca56 ;;
+	    "EV5.7 (21164PC)")
+		UNAME_MACHINE=alphapca57 ;;
+	    "EV6 (21264)")
+		UNAME_MACHINE=alphaev6 ;;
+	    "EV6.7 (21264A)")
+		UNAME_MACHINE=alphaev67 ;;
+	    "EV6.8CB (21264C)")
+		UNAME_MACHINE=alphaev68 ;;
+	    "EV6.8AL (21264B)")
+		UNAME_MACHINE=alphaev68 ;;
+	    "EV6.8CX (21264D)")
+		UNAME_MACHINE=alphaev68 ;;
+	    "EV6.9A (21264/EV69A)")
+		UNAME_MACHINE=alphaev69 ;;
+	    "EV7 (21364)")
+		UNAME_MACHINE=alphaev7 ;;
+	    "EV7.9 (21364A)")
+		UNAME_MACHINE=alphaev79 ;;
+	esac
+	# A Pn.n version is a patched version.
+	# A Vn.n version is a released version.
+	# A Tn.n version is a released field test version.
+	# A Xn.n version is an unreleased experimental baselevel.
+	# 1.2 uses "1.2" for uname -r.
+	echo ${UNAME_MACHINE}-dec-osf`echo ${UNAME_RELEASE} | sed -e 's/^[PVTX]//' | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz`
+	# Reset EXIT trap before exiting to avoid spurious non-zero exit code.
+	exitcode=$?
+	trap '' 0
+	exit $exitcode ;;
+    Alpha\ *:Windows_NT*:*)
+	# How do we know it's Interix rather than the generic POSIX subsystem?
+	# Should we change UNAME_MACHINE based on the output of uname instead
+	# of the specific Alpha model?
+	echo alpha-pc-interix
+	exit ;;
+    21064:Windows_NT:50:3)
+	echo alpha-dec-winnt3.5
+	exit ;;
+    Amiga*:UNIX_System_V:4.0:*)
+	echo m68k-unknown-sysv4
+	exit ;;
+    *:[Aa]miga[Oo][Ss]:*:*)
+	echo ${UNAME_MACHINE}-unknown-amigaos
+	exit ;;
+    *:[Mm]orph[Oo][Ss]:*:*)
+	echo ${UNAME_MACHINE}-unknown-morphos
+	exit ;;
+    *:OS/390:*:*)
+	echo i370-ibm-openedition
+	exit ;;
+    *:z/VM:*:*)
+	echo s390-ibm-zvmoe
+	exit ;;
+    *:OS400:*:*)
+	echo powerpc-ibm-os400
+	exit ;;
+    arm:RISC*:1.[012]*:*|arm:riscix:1.[012]*:*)
+	echo arm-acorn-riscix${UNAME_RELEASE}
+	exit ;;
+    arm*:riscos:*:*|arm*:RISCOS:*:*)
+	echo arm-unknown-riscos
+	exit ;;
+    SR2?01:HI-UX/MPP:*:* | SR8000:HI-UX/MPP:*:*)
+	echo hppa1.1-hitachi-hiuxmpp
+	exit ;;
+    Pyramid*:OSx*:*:* | MIS*:OSx*:*:* | MIS*:SMP_DC-OSx*:*:*)
+	# akee@wpdis03.wpafb.af.mil (Earle F. Ake) contributed MIS and NILE.
+	if test "`(/bin/universe) 2>/dev/null`" = att ; then
+		echo pyramid-pyramid-sysv3
+	else
+		echo pyramid-pyramid-bsd
+	fi
+	exit ;;
+    NILE*:*:*:dcosx)
+	echo pyramid-pyramid-svr4
+	exit ;;
+    DRS?6000:unix:4.0:6*)
+	echo sparc-icl-nx6
+	exit ;;
+    DRS?6000:UNIX_SV:4.2*:7* | DRS?6000:isis:4.2*:7*)
+	case `/usr/bin/uname -p` in
+	    sparc) echo sparc-icl-nx7; exit ;;
+	esac ;;
+    s390x:SunOS:*:*)
+	echo ${UNAME_MACHINE}-ibm-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4H:SunOS:5.*:*)
+	echo sparc-hal-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:5.*:* | tadpole*:SunOS:5.*:*)
+	echo sparc-sun-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    i86pc:AuroraUX:5.*:* | i86xen:AuroraUX:5.*:*)
+	echo i386-pc-auroraux${UNAME_RELEASE}
+	exit ;;
+    i86pc:SunOS:5.*:* | i86xen:SunOS:5.*:*)
+	eval $set_cc_for_build
+	SUN_ARCH=i386
+	# If there is a compiler, see if it is configured for 64-bit objects.
+	# Note that the Sun cc does not turn __LP64__ into 1 like gcc does.
+	# This test works for both compilers.
+	if [ "$CC_FOR_BUILD" != no_compiler_found ]; then
+	    if (echo '#ifdef __amd64'; echo IS_64BIT_ARCH; echo '#endif') | \
+		(CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) | \
+		grep IS_64BIT_ARCH >/dev/null
+	    then
+		SUN_ARCH=x86_64
+	    fi
+	fi
+	echo ${SUN_ARCH}-pc-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:6*:*)
+	# According to config.sub, this is the proper way to canonicalize
+	# SunOS6.  Hard to guess exactly what SunOS6 will be like, but
+	# it's likely to be more like Solaris than SunOS4.
+	echo sparc-sun-solaris3`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    sun4*:SunOS:*:*)
+	case "`/usr/bin/arch -k`" in
+	    Series*|S4*)
+		UNAME_RELEASE=`uname -v`
+		;;
+	esac
+	# Japanese Language versions have a version number like `4.1.3-JL'.
+	echo sparc-sun-sunos`echo ${UNAME_RELEASE}|sed -e 's/-/_/'`
+	exit ;;
+    sun3*:SunOS:*:*)
+	echo m68k-sun-sunos${UNAME_RELEASE}
+	exit ;;
+    sun*:*:4.2BSD:*)
+	UNAME_RELEASE=`(sed 1q /etc/motd | awk '{print substr($5,1,3)}') 2>/dev/null`
+	test "x${UNAME_RELEASE}" = x && UNAME_RELEASE=3
+	case "`/bin/arch`" in
+	    sun3)
+		echo m68k-sun-sunos${UNAME_RELEASE}
+		;;
+	    sun4)
+		echo sparc-sun-sunos${UNAME_RELEASE}
+		;;
+	esac
+	exit ;;
+    aushp:SunOS:*:*)
+	echo sparc-auspex-sunos${UNAME_RELEASE}
+	exit ;;
+    # The situation for MiNT is a little confusing.  The machine name
+    # can be virtually everything (everything which is not
+    # "atarist" or "atariste" at least should have a processor
+    # > m68000).  The system name ranges from "MiNT" over "FreeMiNT"
+    # to the lowercase version "mint" (or "freemint").  Finally
+    # the system name "TOS" denotes a system which is actually not
+    # MiNT.  But MiNT is downward compatible to TOS, so this should
+    # be no problem.
+    atarist[e]:*MiNT:*:* | atarist[e]:*mint:*:* | atarist[e]:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    atari*:*MiNT:*:* | atari*:*mint:*:* | atarist[e]:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    *falcon*:*MiNT:*:* | *falcon*:*mint:*:* | *falcon*:*TOS:*:*)
+	echo m68k-atari-mint${UNAME_RELEASE}
+	exit ;;
+    milan*:*MiNT:*:* | milan*:*mint:*:* | *milan*:*TOS:*:*)
+	echo m68k-milan-mint${UNAME_RELEASE}
+	exit ;;
+    hades*:*MiNT:*:* | hades*:*mint:*:* | *hades*:*TOS:*:*)
+	echo m68k-hades-mint${UNAME_RELEASE}
+	exit ;;
+    *:*MiNT:*:* | *:*mint:*:* | *:*TOS:*:*)
+	echo m68k-unknown-mint${UNAME_RELEASE}
+	exit ;;
+    m68k:machten:*:*)
+	echo m68k-apple-machten${UNAME_RELEASE}
+	exit ;;
+    powerpc:machten:*:*)
+	echo powerpc-apple-machten${UNAME_RELEASE}
+	exit ;;
+    RISC*:Mach:*:*)
+	echo mips-dec-mach_bsd4.3
+	exit ;;
+    RISC*:ULTRIX:*:*)
+	echo mips-dec-ultrix${UNAME_RELEASE}
+	exit ;;
+    VAX*:ULTRIX*:*:*)
+	echo vax-dec-ultrix${UNAME_RELEASE}
+	exit ;;
+    2020:CLIX:*:* | 2430:CLIX:*:*)
+	echo clipper-intergraph-clix${UNAME_RELEASE}
+	exit ;;
+    mips:*:*:UMIPS | mips:*:*:RISCos)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+#ifdef __cplusplus
+#include <stdio.h>  /* for printf() prototype */
+	int main (int argc, char *argv[]) {
+#else
+	int main (argc, argv) int argc; char *argv[]; {
+#endif
+	#if defined (host_mips) && defined (MIPSEB)
+	#if defined (SYSTYPE_SYSV)
+	  printf ("mips-mips-riscos%ssysv\n", argv[1]); exit (0);
+	#endif
+	#if defined (SYSTYPE_SVR4)
+	  printf ("mips-mips-riscos%ssvr4\n", argv[1]); exit (0);
+	#endif
+	#if defined (SYSTYPE_BSD43) || defined(SYSTYPE_BSD)
+	  printf ("mips-mips-riscos%sbsd\n", argv[1]); exit (0);
+	#endif
+	#endif
+	  exit (-1);
+	}
+EOF
+	$CC_FOR_BUILD -o $dummy $dummy.c &&
+	  dummyarg=`echo "${UNAME_RELEASE}" | sed -n 's/\([0-9]*\).*/\1/p'` &&
+	  SYSTEM_NAME=`$dummy $dummyarg` &&
+	    { echo "$SYSTEM_NAME"; exit; }
+	echo mips-mips-riscos${UNAME_RELEASE}
+	exit ;;
+    Motorola:PowerMAX_OS:*:*)
+	echo powerpc-motorola-powermax
+	exit ;;
+    Motorola:*:4.3:PL8-*)
+	echo powerpc-harris-powermax
+	exit ;;
+    Night_Hawk:*:*:PowerMAX_OS | Synergy:PowerMAX_OS:*:*)
+	echo powerpc-harris-powermax
+	exit ;;
+    Night_Hawk:Power_UNIX:*:*)
+	echo powerpc-harris-powerunix
+	exit ;;
+    m88k:CX/UX:7*:*)
+	echo m88k-harris-cxux7
+	exit ;;
+    m88k:*:4*:R4*)
+	echo m88k-motorola-sysv4
+	exit ;;
+    m88k:*:3*:R3*)
+	echo m88k-motorola-sysv3
+	exit ;;
+    AViiON:dgux:*:*)
+	# DG/UX returns AViiON for all architectures
+	UNAME_PROCESSOR=`/usr/bin/uname -p`
+	if [ $UNAME_PROCESSOR = mc88100 ] || [ $UNAME_PROCESSOR = mc88110 ]
+	then
+	    if [ ${TARGET_BINARY_INTERFACE}x = m88kdguxelfx ] || \
+	       [ ${TARGET_BINARY_INTERFACE}x = x ]
+	    then
+		echo m88k-dg-dgux${UNAME_RELEASE}
+	    else
+		echo m88k-dg-dguxbcs${UNAME_RELEASE}
+	    fi
+	else
+	    echo i586-dg-dgux${UNAME_RELEASE}
+	fi
+	exit ;;
+    M88*:DolphinOS:*:*)	# DolphinOS (SVR3)
+	echo m88k-dolphin-sysv3
+	exit ;;
+    M88*:*:R3*:*)
+	# Delta 88k system running SVR3
+	echo m88k-motorola-sysv3
+	exit ;;
+    XD88*:*:*:*) # Tektronix XD88 system running UTekV (SVR3)
+	echo m88k-tektronix-sysv3
+	exit ;;
+    Tek43[0-9][0-9]:UTek:*:*) # Tektronix 4300 system running UTek (BSD)
+	echo m68k-tektronix-bsd
+	exit ;;
+    *:IRIX*:*:*)
+	echo mips-sgi-irix`echo ${UNAME_RELEASE}|sed -e 's/-/_/g'`
+	exit ;;
+    ????????:AIX?:[12].1:2)   # AIX 2.2.1 or AIX 2.1.1 is RT/PC AIX.
+	echo romp-ibm-aix     # uname -m gives an 8 hex-code CPU id
+	exit ;;               # Note that: echo "'`uname -s`'" gives 'AIX '
+    i*86:AIX:*:*)
+	echo i386-ibm-aix
+	exit ;;
+    ia64:AIX:*:*)
+	if [ -x /usr/bin/oslevel ] ; then
+		IBM_REV=`/usr/bin/oslevel`
+	else
+		IBM_REV=${UNAME_VERSION}.${UNAME_RELEASE}
+	fi
+	echo ${UNAME_MACHINE}-ibm-aix${IBM_REV}
+	exit ;;
+    *:AIX:2:3)
+	if grep bos325 /usr/include/stdio.h >/dev/null 2>&1; then
+		eval $set_cc_for_build
+		sed 's/^		//' << EOF >$dummy.c
+		#include <sys/systemcfg.h>
+
+		main()
+			{
+			if (!__power_pc())
+				exit(1);
+			puts("powerpc-ibm-aix3.2.5");
+			exit(0);
+			}
+EOF
+		if $CC_FOR_BUILD -o $dummy $dummy.c && SYSTEM_NAME=`$dummy`
+		then
+			echo "$SYSTEM_NAME"
+		else
+			echo rs6000-ibm-aix3.2.5
+		fi
+	elif grep bos324 /usr/include/stdio.h >/dev/null 2>&1; then
+		echo rs6000-ibm-aix3.2.4
+	else
+		echo rs6000-ibm-aix3.2
+	fi
+	exit ;;
+    *:AIX:*:[4567])
+	IBM_CPU_ID=`/usr/sbin/lsdev -C -c processor -S available | sed 1q | awk '{ print $1 }'`
+	if /usr/sbin/lsattr -El ${IBM_CPU_ID} | grep ' POWER' >/dev/null 2>&1; then
+		IBM_ARCH=rs6000
+	else
+		IBM_ARCH=powerpc
+	fi
+	if [ -x /usr/bin/lslpp ] ; then
+		IBM_REV=`/usr/bin/lslpp -Lqc bos.rte.libc |
+			   awk -F: '{ print $3 }' | sed s/[0-9]*$/0/`
+	else
+		IBM_REV=${UNAME_VERSION}.${UNAME_RELEASE}
+	fi
+	echo ${IBM_ARCH}-ibm-aix${IBM_REV}
+	exit ;;
+    *:AIX:*:*)
+	echo rs6000-ibm-aix
+	exit ;;
+    ibmrt:4.4BSD:*|romp-ibm:BSD:*)
+	echo romp-ibm-bsd4.4
+	exit ;;
+    ibmrt:*BSD:*|romp-ibm:BSD:*)            # covers RT/PC BSD and
+	echo romp-ibm-bsd${UNAME_RELEASE}   # 4.3 with uname added to
+	exit ;;                             # report: romp-ibm BSD 4.3
+    *:BOSX:*:*)
+	echo rs6000-bull-bosx
+	exit ;;
+    DPX/2?00:B.O.S.:*:*)
+	echo m68k-bull-sysv3
+	exit ;;
+    9000/[34]??:4.3bsd:1.*:*)
+	echo m68k-hp-bsd
+	exit ;;
+    hp300:4.4BSD:*:* | 9000/[34]??:4.3bsd:2.*:*)
+	echo m68k-hp-bsd4.4
+	exit ;;
+    9000/[34678]??:HP-UX:*:*)
+	HPUX_REV=`echo ${UNAME_RELEASE}|sed -e 's/[^.]*.[0B]*//'`
+	case "${UNAME_MACHINE}" in
+	    9000/31? )            HP_ARCH=m68000 ;;
+	    9000/[34]?? )         HP_ARCH=m68k ;;
+	    9000/[678][0-9][0-9])
+		if [ -x /usr/bin/getconf ]; then
+		    sc_cpu_version=`/usr/bin/getconf SC_CPU_VERSION 2>/dev/null`
+		    sc_kernel_bits=`/usr/bin/getconf SC_KERNEL_BITS 2>/dev/null`
+		    case "${sc_cpu_version}" in
+		      523) HP_ARCH=hppa1.0 ;; # CPU_PA_RISC1_0
+		      528) HP_ARCH=hppa1.1 ;; # CPU_PA_RISC1_1
+		      532)                      # CPU_PA_RISC2_0
+			case "${sc_kernel_bits}" in
+			  32) HP_ARCH=hppa2.0n ;;
+			  64) HP_ARCH=hppa2.0w ;;
+			  '') HP_ARCH=hppa2.0 ;;   # HP-UX 10.20
+			esac ;;
+		    esac
+		fi
+		if [ "${HP_ARCH}" = "" ]; then
+		    eval $set_cc_for_build
+		    sed 's/^		//' << EOF >$dummy.c
+
+		#define _HPUX_SOURCE
+		#include <stdlib.h>
+		#include <unistd.h>
+
+		int main ()
+		{
+		#if defined(_SC_KERNEL_BITS)
+		    long bits = sysconf(_SC_KERNEL_BITS);
+		#endif
+		    long cpu  = sysconf (_SC_CPU_VERSION);
+
+		    switch (cpu)
+			{
+			case CPU_PA_RISC1_0: puts ("hppa1.0"); break;
+			case CPU_PA_RISC1_1: puts ("hppa1.1"); break;
+			case CPU_PA_RISC2_0:
+		#if defined(_SC_KERNEL_BITS)
+			    switch (bits)
+				{
+				case 64: puts ("hppa2.0w"); break;
+				case 32: puts ("hppa2.0n"); break;
+				default: puts ("hppa2.0"); break;
+				} break;
+		#else  /* !defined(_SC_KERNEL_BITS) */
+			    puts ("hppa2.0"); break;
+		#endif
+			default: puts ("hppa1.0"); break;
+			}
+		    exit (0);
+		}
+EOF
+		    (CCOPTS="" $CC_FOR_BUILD -o $dummy $dummy.c 2>/dev/null) && HP_ARCH=`$dummy`
+		    test -z "$HP_ARCH" && HP_ARCH=hppa
+		fi ;;
+	esac
+	if [ ${HP_ARCH} = hppa2.0w ]
+	then
+	    eval $set_cc_for_build
+
+	    # hppa2.0w-hp-hpux* has a 64-bit kernel and a compiler generating
+	    # 32-bit code.  hppa64-hp-hpux* has the same kernel and a compiler
+	    # generating 64-bit code.  GNU and HP use different nomenclature:
+	    #
+	    # $ CC_FOR_BUILD=cc ./config.guess
+	    # => hppa2.0w-hp-hpux11.23
+	    # $ CC_FOR_BUILD="cc +DA2.0w" ./config.guess
+	    # => hppa64-hp-hpux11.23
+
+	    if echo __LP64__ | (CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) |
+		grep -q __LP64__
+	    then
+		HP_ARCH=hppa2.0w
+	    else
+		HP_ARCH=hppa64
+	    fi
+	fi
+	echo ${HP_ARCH}-hp-hpux${HPUX_REV}
+	exit ;;
+    ia64:HP-UX:*:*)
+	HPUX_REV=`echo ${UNAME_RELEASE}|sed -e 's/[^.]*.[0B]*//'`
+	echo ia64-hp-hpux${HPUX_REV}
+	exit ;;
+    3050*:HI-UX:*:*)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+	#include <unistd.h>
+	int
+	main ()
+	{
+	  long cpu = sysconf (_SC_CPU_VERSION);
+	  /* The order matters, because CPU_IS_HP_MC68K erroneously returns
+	     true for CPU_PA_RISC1_0.  CPU_IS_PA_RISC returns correct
+	     results, however.  */
+	  if (CPU_IS_PA_RISC (cpu))
+	    {
+	      switch (cpu)
+		{
+		  case CPU_PA_RISC1_0: puts ("hppa1.0-hitachi-hiuxwe2"); break;
+		  case CPU_PA_RISC1_1: puts ("hppa1.1-hitachi-hiuxwe2"); break;
+		  case CPU_PA_RISC2_0: puts ("hppa2.0-hitachi-hiuxwe2"); break;
+		  default: puts ("hppa-hitachi-hiuxwe2"); break;
+		}
+	    }
+	  else if (CPU_IS_HP_MC68K (cpu))
+	    puts ("m68k-hitachi-hiuxwe2");
+	  else puts ("unknown-hitachi-hiuxwe2");
+	  exit (0);
+	}
+EOF
+	$CC_FOR_BUILD -o $dummy $dummy.c && SYSTEM_NAME=`$dummy` &&
+		{ echo "$SYSTEM_NAME"; exit; }
+	echo unknown-hitachi-hiuxwe2
+	exit ;;
+    9000/7??:4.3bsd:*:* | 9000/8?[79]:4.3bsd:*:* )
+	echo hppa1.1-hp-bsd
+	exit ;;
+    9000/8??:4.3bsd:*:*)
+	echo hppa1.0-hp-bsd
+	exit ;;
+    *9??*:MPE/iX:*:* | *3000*:MPE/iX:*:*)
+	echo hppa1.0-hp-mpeix
+	exit ;;
+    hp7??:OSF1:*:* | hp8?[79]:OSF1:*:* )
+	echo hppa1.1-hp-osf
+	exit ;;
+    hp8??:OSF1:*:*)
+	echo hppa1.0-hp-osf
+	exit ;;
+    i*86:OSF1:*:*)
+	if [ -x /usr/sbin/sysversion ] ; then
+	    echo ${UNAME_MACHINE}-unknown-osf1mk
+	else
+	    echo ${UNAME_MACHINE}-unknown-osf1
+	fi
+	exit ;;
+    parisc*:Lites*:*:*)
+	echo hppa1.1-hp-lites
+	exit ;;
+    C1*:ConvexOS:*:* | convex:ConvexOS:C1*:*)
+	echo c1-convex-bsd
+	exit ;;
+    C2*:ConvexOS:*:* | convex:ConvexOS:C2*:*)
+	if getsysinfo -f scalar_acc
+	then echo c32-convex-bsd
+	else echo c2-convex-bsd
+	fi
+	exit ;;
+    C34*:ConvexOS:*:* | convex:ConvexOS:C34*:*)
+	echo c34-convex-bsd
+	exit ;;
+    C38*:ConvexOS:*:* | convex:ConvexOS:C38*:*)
+	echo c38-convex-bsd
+	exit ;;
+    C4*:ConvexOS:*:* | convex:ConvexOS:C4*:*)
+	echo c4-convex-bsd
+	exit ;;
+    CRAY*Y-MP:*:*:*)
+	echo ymp-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*[A-Z]90:*:*:*)
+	echo ${UNAME_MACHINE}-cray-unicos${UNAME_RELEASE} \
+	| sed -e 's/CRAY.*\([A-Z]90\)/\1/' \
+	      -e y/ABCDEFGHIJKLMNOPQRSTUVWXYZ/abcdefghijklmnopqrstuvwxyz/ \
+	      -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*TS:*:*:*)
+	echo t90-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*T3E:*:*:*)
+	echo alphaev5-cray-unicosmk${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    CRAY*SV1:*:*:*)
+	echo sv1-cray-unicos${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    *:UNICOS/mp:*:*)
+	echo craynv-cray-unicosmp${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
+	exit ;;
+    F30[01]:UNIX_System_V:*:* | F700:UNIX_System_V:*:*)
+	FUJITSU_PROC=`uname -m | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz`
+	FUJITSU_SYS=`uname -p | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/\///'`
+	FUJITSU_REL=`echo ${UNAME_RELEASE} | sed -e 's/ /_/'`
+	echo "${FUJITSU_PROC}-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
+	exit ;;
+    5000:UNIX_System_V:4.*:*)
+	FUJITSU_SYS=`uname -p | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/\///'`
+	FUJITSU_REL=`echo ${UNAME_RELEASE} | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/ /_/'`
+	echo "sparc-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
+	exit ;;
+    i*86:BSD/386:*:* | i*86:BSD/OS:*:* | *:Ascend\ Embedded/OS:*:*)
+	echo ${UNAME_MACHINE}-pc-bsdi${UNAME_RELEASE}
+	exit ;;
+    sparc*:BSD/OS:*:*)
+	echo sparc-unknown-bsdi${UNAME_RELEASE}
+	exit ;;
+    *:BSD/OS:*:*)
+	echo ${UNAME_MACHINE}-unknown-bsdi${UNAME_RELEASE}
+	exit ;;
+    *:FreeBSD:*:*)
+	UNAME_PROCESSOR=`/usr/bin/uname -p`
+	case ${UNAME_PROCESSOR} in
+	    amd64)
+		echo x86_64-unknown-freebsd`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'` ;;
+	    *)
+		echo ${UNAME_PROCESSOR}-unknown-freebsd`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'` ;;
+	esac
+	exit ;;
+    i*:CYGWIN*:*)
+	echo ${UNAME_MACHINE}-pc-cygwin
+	exit ;;
+    *:MINGW64*:*)
+	echo ${UNAME_MACHINE}-pc-mingw64
+	exit ;;
+    *:MINGW*:*)
+	echo ${UNAME_MACHINE}-pc-mingw32
+	exit ;;
+    *:MSYS*:*)
+	echo ${UNAME_MACHINE}-pc-msys
+	exit ;;
+    i*:windows32*:*)
+	# uname -m includes "-pc" on this system.
+	echo ${UNAME_MACHINE}-mingw32
+	exit ;;
+    i*:PW*:*)
+	echo ${UNAME_MACHINE}-pc-pw32
+	exit ;;
+    *:Interix*:*)
+	case ${UNAME_MACHINE} in
+	    x86)
+		echo i586-pc-interix${UNAME_RELEASE}
+		exit ;;
+	    authenticamd | genuineintel | EM64T)
+		echo x86_64-unknown-interix${UNAME_RELEASE}
+		exit ;;
+	    IA64)
+		echo ia64-unknown-interix${UNAME_RELEASE}
+		exit ;;
+	esac ;;
+    [345]86:Windows_95:* | [345]86:Windows_98:* | [345]86:Windows_NT:*)
+	echo i${UNAME_MACHINE}-pc-mks
+	exit ;;
+    8664:Windows_NT:*)
+	echo x86_64-pc-mks
+	exit ;;
+    i*:Windows_NT*:* | Pentium*:Windows_NT*:*)
+	# How do we know it's Interix rather than the generic POSIX subsystem?
+	# It also conflicts with pre-2.0 versions of AT&T UWIN. Should we
+	# UNAME_MACHINE based on the output of uname instead of i386?
+	echo i586-pc-interix
+	exit ;;
+    i*:UWIN*:*)
+	echo ${UNAME_MACHINE}-pc-uwin
+	exit ;;
+    amd64:CYGWIN*:*:* | x86_64:CYGWIN*:*:*)
+	echo x86_64-unknown-cygwin
+	exit ;;
+    p*:CYGWIN*:*)
+	echo powerpcle-unknown-cygwin
+	exit ;;
+    prep*:SunOS:5.*:*)
+	echo powerpcle-unknown-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
+	exit ;;
+    *:GNU:*:*)
+	# the GNU system
+	echo `echo ${UNAME_MACHINE}|sed -e 's,[-/].*$,,'`-unknown-${LIBC}`echo ${UNAME_RELEASE}|sed -e 's,/.*$,,'`
+	exit ;;
+    *:GNU/*:*:*)
+	# other systems with GNU libc and userland
+	echo ${UNAME_MACHINE}-unknown-`echo ${UNAME_SYSTEM} | sed 's,^[^/]*/,,' | tr "[:upper:]" "[:lower:]"``echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`-${LIBC}
+	exit ;;
+    i*86:Minix:*:*)
+	echo ${UNAME_MACHINE}-pc-minix
+	exit ;;
+    aarch64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    aarch64_be:Linux:*:*)
+	UNAME_MACHINE=aarch64_be
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    alpha:Linux:*:*)
+	case `sed -n '/^cpu model/s/^.*: \(.*\)/\1/p' < /proc/cpuinfo` in
+	  EV5)   UNAME_MACHINE=alphaev5 ;;
+	  EV56)  UNAME_MACHINE=alphaev56 ;;
+	  PCA56) UNAME_MACHINE=alphapca56 ;;
+	  PCA57) UNAME_MACHINE=alphapca56 ;;
+	  EV6)   UNAME_MACHINE=alphaev6 ;;
+	  EV67)  UNAME_MACHINE=alphaev67 ;;
+	  EV68*) UNAME_MACHINE=alphaev68 ;;
+	esac
+	objdump --private-headers /bin/sh | grep -q ld.so.1
+	if test "$?" = 0 ; then LIBC=gnulibc1 ; fi
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    arc:Linux:*:* | arceb:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    arm*:Linux:*:*)
+	eval $set_cc_for_build
+	if echo __ARM_EABI__ | $CC_FOR_BUILD -E - 2>/dev/null \
+	    | grep -q __ARM_EABI__
+	then
+	    echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	else
+	    if echo __ARM_PCS_VFP | $CC_FOR_BUILD -E - 2>/dev/null \
+		| grep -q __ARM_PCS_VFP
+	    then
+		echo ${UNAME_MACHINE}-unknown-linux-${LIBC}eabi
+	    else
+		echo ${UNAME_MACHINE}-unknown-linux-${LIBC}eabihf
+	    fi
+	fi
+	exit ;;
+    avr32*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    cris:Linux:*:*)
+	echo ${UNAME_MACHINE}-axis-linux-${LIBC}
+	exit ;;
+    crisv32:Linux:*:*)
+	echo ${UNAME_MACHINE}-axis-linux-${LIBC}
+	exit ;;
+    e2k:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    frv:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    hexagon:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    i*86:Linux:*:*)
+	echo ${UNAME_MACHINE}-pc-linux-${LIBC}
+	exit ;;
+    ia64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    k1om:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    m32r*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    m68*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    mips:Linux:*:* | mips64:Linux:*:*)
+	eval $set_cc_for_build
+	sed 's/^	//' << EOF >$dummy.c
+	#undef CPU
+	#undef ${UNAME_MACHINE}
+	#undef ${UNAME_MACHINE}el
+	#if defined(__MIPSEL__) || defined(__MIPSEL) || defined(_MIPSEL) || defined(MIPSEL)
+	CPU=${UNAME_MACHINE}el
+	#else
+	#if defined(__MIPSEB__) || defined(__MIPSEB) || defined(_MIPSEB) || defined(MIPSEB)
+	CPU=${UNAME_MACHINE}
+	#else
+	CPU=
+	#endif
+	#endif
+EOF
+	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^CPU'`
+	test x"${CPU}" != x && { echo "${CPU}-unknown-linux-${LIBC}"; exit; }
+	;;
+    mips64el:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    openrisc*:Linux:*:*)
+	echo or1k-unknown-linux-${LIBC}
+	exit ;;
+    or32:Linux:*:* | or1k*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    padre:Linux:*:*)
+	echo sparc-unknown-linux-${LIBC}
+	exit ;;
+    parisc64:Linux:*:* | hppa64:Linux:*:*)
+	echo hppa64-unknown-linux-${LIBC}
+	exit ;;
+    parisc:Linux:*:* | hppa:Linux:*:*)
+	# Look for CPU level
+	case `grep '^cpu[^a-z]*:' /proc/cpuinfo 2>/dev/null | cut -d' ' -f2` in
+	  PA7*) echo hppa1.1-unknown-linux-${LIBC} ;;
+	  PA8*) echo hppa2.0-unknown-linux-${LIBC} ;;
+	  *)    echo hppa-unknown-linux-${LIBC} ;;
+	esac
+	exit ;;
+    ppc64:Linux:*:*)
+	echo powerpc64-unknown-linux-${LIBC}
+	exit ;;
+    ppc:Linux:*:*)
+	echo powerpc-unknown-linux-${LIBC}
+	exit ;;
+    ppc64le:Linux:*:*)
+	echo powerpc64le-unknown-linux-${LIBC}
+	exit ;;
+    ppcle:Linux:*:*)
+	echo powerpcle-unknown-linux-${LIBC}
+	exit ;;
+    riscv32:Linux:*:* | riscv64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    s390:Linux:*:* | s390x:Linux:*:*)
+	echo ${UNAME_MACHINE}-ibm-linux-${LIBC}
+	exit ;;
+    sh64*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    sh*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    sparc:Linux:*:* | sparc64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    tile*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    vax:Linux:*:*)
+	echo ${UNAME_MACHINE}-dec-linux-${LIBC}
+	exit ;;
+    x86_64:Linux:*:*)
+	echo ${UNAME_MACHINE}-pc-linux-${LIBC}
+	exit ;;
+    xtensa*:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
+    i*86:DYNIX/ptx:4*:*)
+	# ptx 4.0 does uname -s correctly, with DYNIX/ptx in there.
+	# earlier versions are messed up and put the nodename in both
+	# sysname and nodename.
+	echo i386-sequent-sysv4
+	exit ;;
+    i*86:UNIX_SV:4.2MP:2.*)
+	# Unixware is an offshoot of SVR4, but it has its own version
+	# number series starting with 2...
+	# I am not positive that other SVR4 systems won't match this,
+	# I just have to hope.  -- rms.
+	# Use sysv4.2uw... so that sysv4* matches it.
+	echo ${UNAME_MACHINE}-pc-sysv4.2uw${UNAME_VERSION}
+	exit ;;
+    i*86:OS/2:*:*)
+	# If we were able to find `uname', then EMX Unix compatibility
+	# is probably installed.
+	echo ${UNAME_MACHINE}-pc-os2-emx
+	exit ;;
+    i*86:XTS-300:*:STOP)
+	echo ${UNAME_MACHINE}-unknown-stop
+	exit ;;
+    i*86:atheos:*:*)
+	echo ${UNAME_MACHINE}-unknown-atheos
+	exit ;;
+    i*86:syllable:*:*)
+	echo ${UNAME_MACHINE}-pc-syllable
+	exit ;;
+    i*86:LynxOS:2.*:* | i*86:LynxOS:3.[01]*:* | i*86:LynxOS:4.[02]*:*)
+	echo i386-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    i*86:*DOS:*:*)
+	echo ${UNAME_MACHINE}-pc-msdosdjgpp
+	exit ;;
+    i*86:*:4.*:* | i*86:SYSTEM_V:4.*:*)
+	UNAME_REL=`echo ${UNAME_RELEASE} | sed 's/\/MP$//'`
+	if grep Novell /usr/include/link.h >/dev/null 2>/dev/null; then
+		echo ${UNAME_MACHINE}-univel-sysv${UNAME_REL}
+	else
+		echo ${UNAME_MACHINE}-pc-sysv${UNAME_REL}
+	fi
+	exit ;;
+    i*86:*:5:[678]*)
+	# UnixWare 7.x, OpenUNIX and OpenServer 6.
+	case `/bin/uname -X | grep "^Machine"` in
+	    *486*)	     UNAME_MACHINE=i486 ;;
+	    *Pentium)	     UNAME_MACHINE=i586 ;;
+	    *Pent*|*Celeron) UNAME_MACHINE=i686 ;;
+	esac
+	echo ${UNAME_MACHINE}-unknown-sysv${UNAME_RELEASE}${UNAME_SYSTEM}${UNAME_VERSION}
+	exit ;;
+    i*86:*:3.2:*)
+	if test -f /usr/options/cb.name; then
+		UNAME_REL=`sed -n 's/.*Version //p' </usr/options/cb.name`
+		echo ${UNAME_MACHINE}-pc-isc$UNAME_REL
+	elif /bin/uname -X 2>/dev/null >/dev/null ; then
+		UNAME_REL=`(/bin/uname -X|grep Release|sed -e 's/.*= //')`
+		(/bin/uname -X|grep i80486 >/dev/null) && UNAME_MACHINE=i486
+		(/bin/uname -X|grep '^Machine.*Pentium' >/dev/null) \
+			&& UNAME_MACHINE=i586
+		(/bin/uname -X|grep '^Machine.*Pent *II' >/dev/null) \
+			&& UNAME_MACHINE=i686
+		(/bin/uname -X|grep '^Machine.*Pentium Pro' >/dev/null) \
+			&& UNAME_MACHINE=i686
+		echo ${UNAME_MACHINE}-pc-sco$UNAME_REL
+	else
+		echo ${UNAME_MACHINE}-pc-sysv32
+	fi
+	exit ;;
+    pc:*:*:*)
+	# Left here for compatibility:
+	# uname -m prints for DJGPP always 'pc', but it prints nothing about
+	# the processor, so we play safe by assuming i586.
+	# Note: whatever this is, it MUST be the same as what config.sub
+	# prints for the "djgpp" host, or else GDB configure will decide that
+	# this is a cross-build.
+	echo i586-pc-msdosdjgpp
+	exit ;;
+    Intel:Mach:3*:*)
+	echo i386-pc-mach3
+	exit ;;
+    paragon:*:*:*)
+	echo i860-intel-osf1
+	exit ;;
+    i860:*:4.*:*) # i860-SVR4
+	if grep Stardent /usr/include/sys/uadmin.h >/dev/null 2>&1 ; then
+	  echo i860-stardent-sysv${UNAME_RELEASE} # Stardent Vistra i860-SVR4
+	else # Add other i860-SVR4 vendors below as they are discovered.
+	  echo i860-unknown-sysv${UNAME_RELEASE}  # Unknown i860-SVR4
+	fi
+	exit ;;
+    mini*:CTIX:SYS*5:*)
+	# "miniframe"
+	echo m68010-convergent-sysv
+	exit ;;
+    mc68k:UNIX:SYSTEM5:3.51m)
+	echo m68k-convergent-sysv
+	exit ;;
+    M680?0:D-NIX:5.3:*)
+	echo m68k-diab-dnix
+	exit ;;
+    M68*:*:R3V[5678]*:*)
+	test -r /sysV68 && { echo 'm68k-motorola-sysv'; exit; } ;;
+    3[345]??:*:4.0:3.0 | 3[34]??A:*:4.0:3.0 | 3[34]??,*:*:4.0:3.0 | 3[34]??/*:*:4.0:3.0 | 4400:*:4.0:3.0 | 4850:*:4.0:3.0 | SKA40:*:4.0:3.0 | SDS2:*:4.0:3.0 | SHG2:*:4.0:3.0 | S7501*:*:4.0:3.0)
+	OS_REL=''
+	test -r /etc/.relid \
+	&& OS_REL=.`sed -n 's/[^ ]* [^ ]* \([0-9][0-9]\).*/\1/p' < /etc/.relid`
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	  && { echo i486-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep entium >/dev/null \
+	  && { echo i586-ncr-sysv4.3${OS_REL}; exit; } ;;
+    3[34]??:*:4.0:* | 3[34]??,*:*:4.0:*)
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	  && { echo i486-ncr-sysv4; exit; } ;;
+    NCR*:*:4.2:* | MPRAS*:*:4.2:*)
+	OS_REL='.3'
+	test -r /etc/.relid \
+	    && OS_REL=.`sed -n 's/[^ ]* [^ ]* \([0-9][0-9]\).*/\1/p' < /etc/.relid`
+	/bin/uname -p 2>/dev/null | grep 86 >/dev/null \
+	    && { echo i486-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep entium >/dev/null \
+	    && { echo i586-ncr-sysv4.3${OS_REL}; exit; }
+	/bin/uname -p 2>/dev/null | /bin/grep pteron >/dev/null \
+	    && { echo i586-ncr-sysv4.3${OS_REL}; exit; } ;;
+    m68*:LynxOS:2.*:* | m68*:LynxOS:3.0*:*)
+	echo m68k-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    mc68030:UNIX_System_V:4.*:*)
+	echo m68k-atari-sysv4
+	exit ;;
+    TSUNAMI:LynxOS:2.*:*)
+	echo sparc-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    rs6000:LynxOS:2.*:*)
+	echo rs6000-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    PowerPC:LynxOS:2.*:* | PowerPC:LynxOS:3.[01]*:* | PowerPC:LynxOS:4.[02]*:*)
+	echo powerpc-unknown-lynxos${UNAME_RELEASE}
+	exit ;;
+    SM[BE]S:UNIX_SV:*:*)
+	echo mips-dde-sysv${UNAME_RELEASE}
+	exit ;;
+    RM*:ReliantUNIX-*:*:*)
+	echo mips-sni-sysv4
+	exit ;;
+    RM*:SINIX-*:*:*)
+	echo mips-sni-sysv4
+	exit ;;
+    *:SINIX-*:*:*)
+	if uname -p 2>/dev/null >/dev/null ; then
+		UNAME_MACHINE=`(uname -p) 2>/dev/null`
+		echo ${UNAME_MACHINE}-sni-sysv4
+	else
+		echo ns32k-sni-sysv
+	fi
+	exit ;;
+    PENTIUM:*:4.0*:*)	# Unisys `ClearPath HMP IX 4000' SVR4/MP effort
+			# says <Richard.M.Bartel@ccMail.Census.GOV>
+	echo i586-unisys-sysv4
+	exit ;;
+    *:UNIX_System_V:4*:FTX*)
+	# From Gerald Hewes <hewes@openmarket.com>.
+	# How about differentiating between stratus architectures? -djm
+	echo hppa1.1-stratus-sysv4
+	exit ;;
+    *:*:*:FTX*)
+	# From seanf@swdc.stratus.com.
+	echo i860-stratus-sysv4
+	exit ;;
+    i*86:VOS:*:*)
+	# From Paul.Green@stratus.com.
+	echo ${UNAME_MACHINE}-stratus-vos
+	exit ;;
+    *:VOS:*:*)
+	# From Paul.Green@stratus.com.
+	echo hppa1.1-stratus-vos
+	exit ;;
+    mc68*:A/UX:*:*)
+	echo m68k-apple-aux${UNAME_RELEASE}
+	exit ;;
+    news*:NEWS-OS:6*:*)
+	echo mips-sony-newsos6
+	exit ;;
+    R[34]000:*System_V*:*:* | R4000:UNIX_SYSV:*:* | R*000:UNIX_SV:*:*)
+	if [ -d /usr/nec ]; then
+		echo mips-nec-sysv${UNAME_RELEASE}
+	else
+		echo mips-unknown-sysv${UNAME_RELEASE}
+	fi
+	exit ;;
+    BeBox:BeOS:*:*)	# BeOS running on hardware made by Be, PPC only.
+	echo powerpc-be-beos
+	exit ;;
+    BeMac:BeOS:*:*)	# BeOS running on Mac or Mac clone, PPC only.
+	echo powerpc-apple-beos
+	exit ;;
+    BePC:BeOS:*:*)	# BeOS running on Intel PC compatible.
+	echo i586-pc-beos
+	exit ;;
+    BePC:Haiku:*:*)	# Haiku running on Intel PC compatible.
+	echo i586-pc-haiku
+	exit ;;
+    x86_64:Haiku:*:*)
+	echo x86_64-unknown-haiku
+	exit ;;
+    SX-4:SUPER-UX:*:*)
+	echo sx4-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-5:SUPER-UX:*:*)
+	echo sx5-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-6:SUPER-UX:*:*)
+	echo sx6-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-7:SUPER-UX:*:*)
+	echo sx7-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-8:SUPER-UX:*:*)
+	echo sx8-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-8R:SUPER-UX:*:*)
+	echo sx8r-nec-superux${UNAME_RELEASE}
+	exit ;;
+    SX-ACE:SUPER-UX:*:*)
+	echo sxace-nec-superux${UNAME_RELEASE}
+	exit ;;
+    Power*:Rhapsody:*:*)
+	echo powerpc-apple-rhapsody${UNAME_RELEASE}
+	exit ;;
+    *:Rhapsody:*:*)
+	echo ${UNAME_MACHINE}-apple-rhapsody${UNAME_RELEASE}
+	exit ;;
+    *:Darwin:*:*)
+	UNAME_PROCESSOR=`uname -p` || UNAME_PROCESSOR=unknown
+	eval $set_cc_for_build
+	if test "$UNAME_PROCESSOR" = unknown ; then
+	    UNAME_PROCESSOR=powerpc
+	fi
+	if test `echo "$UNAME_RELEASE" | sed -e 's/\..*//'` -le 10 ; then
+	    if [ "$CC_FOR_BUILD" != no_compiler_found ]; then
+		if (echo '#ifdef __LP64__'; echo IS_64BIT_ARCH; echo '#endif') | \
+		    (CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) | \
+		    grep IS_64BIT_ARCH >/dev/null
+		then
+		    case $UNAME_PROCESSOR in
+			i386) UNAME_PROCESSOR=x86_64 ;;
+			powerpc) UNAME_PROCESSOR=powerpc64 ;;
+		    esac
+		fi
+	    fi
+	elif test "$UNAME_PROCESSOR" = i386 ; then
+	    # Avoid executing cc on OS X 10.9, as it ships with a stub
+	    # that puts up a graphical alert prompting to install
+	    # developer tools.  Any system running Mac OS X 10.7 or
+	    # later (Darwin 11 and later) is required to have a 64-bit
+	    # processor. This is not true of the ARM version of Darwin
+	    # that Apple uses in portable devices.
+	    UNAME_PROCESSOR=x86_64
+	fi
+	echo ${UNAME_PROCESSOR}-apple-darwin${UNAME_RELEASE}
+	exit ;;
+    *:procnto*:*:* | *:QNX:[0123456789]*:*)
+	UNAME_PROCESSOR=`uname -p`
+	if test "$UNAME_PROCESSOR" = x86; then
+		UNAME_PROCESSOR=i386
+		UNAME_MACHINE=pc
+	fi
+	echo ${UNAME_PROCESSOR}-${UNAME_MACHINE}-nto-qnx${UNAME_RELEASE}
+	exit ;;
+    *:QNX:*:4*)
+	echo i386-pc-qnx
+	exit ;;
+    NEO-?:NONSTOP_KERNEL:*:*)
+	echo neo-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    NSE-*:NONSTOP_KERNEL:*:*)
+	echo nse-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    NSR-?:NONSTOP_KERNEL:*:*)
+	echo nsr-tandem-nsk${UNAME_RELEASE}
+	exit ;;
+    *:NonStop-UX:*:*)
+	echo mips-compaq-nonstopux
+	exit ;;
+    BS2000:POSIX*:*:*)
+	echo bs2000-siemens-sysv
+	exit ;;
+    DS/*:UNIX_System_V:*:*)
+	echo ${UNAME_MACHINE}-${UNAME_SYSTEM}-${UNAME_RELEASE}
+	exit ;;
+    *:Plan9:*:*)
+	# "uname -m" is not consistent, so use $cputype instead. 386
+	# is converted to i386 for consistency with other x86
+	# operating systems.
+	if test "$cputype" = 386; then
+	    UNAME_MACHINE=i386
+	else
+	    UNAME_MACHINE="$cputype"
+	fi
+	echo ${UNAME_MACHINE}-unknown-plan9
+	exit ;;
+    *:TOPS-10:*:*)
+	echo pdp10-unknown-tops10
+	exit ;;
+    *:TENEX:*:*)
+	echo pdp10-unknown-tenex
+	exit ;;
+    KS10:TOPS-20:*:* | KL10:TOPS-20:*:* | TYPE4:TOPS-20:*:*)
+	echo pdp10-dec-tops20
+	exit ;;
+    XKL-1:TOPS-20:*:* | TYPE5:TOPS-20:*:*)
+	echo pdp10-xkl-tops20
+	exit ;;
+    *:TOPS-20:*:*)
+	echo pdp10-unknown-tops20
+	exit ;;
+    *:ITS:*:*)
+	echo pdp10-unknown-its
+	exit ;;
+    SEI:*:*:SEIUX)
+	echo mips-sei-seiux${UNAME_RELEASE}
+	exit ;;
+    *:DragonFly:*:*)
+	echo ${UNAME_MACHINE}-unknown-dragonfly`echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`
+	exit ;;
+    *:*VMS:*:*)
+	UNAME_MACHINE=`(uname -p) 2>/dev/null`
+	case "${UNAME_MACHINE}" in
+	    A*) echo alpha-dec-vms ; exit ;;
+	    I*) echo ia64-dec-vms ; exit ;;
+	    V*) echo vax-dec-vms ; exit ;;
+	esac ;;
+    *:XENIX:*:SysV)
+	echo i386-pc-xenix
+	exit ;;
+    i*86:skyos:*:*)
+	echo ${UNAME_MACHINE}-pc-skyos`echo ${UNAME_RELEASE} | sed -e 's/ .*$//'`
+	exit ;;
+    i*86:rdos:*:*)
+	echo ${UNAME_MACHINE}-pc-rdos
+	exit ;;
+    i*86:AROS:*:*)
+	echo ${UNAME_MACHINE}-pc-aros
+	exit ;;
+    x86_64:VMkernel:*:*)
+	echo ${UNAME_MACHINE}-unknown-esx
+	exit ;;
+    amd64:Isilon\ OneFS:*:*)
+	echo x86_64-unknown-onefs
+	exit ;;
+esac
+
+cat >&2 <<EOF
+$0: unable to guess system type
+
+This script (version $timestamp), has failed to recognize the
+operating system you are using. If your script is old, overwrite
+config.guess and config.sub with the latest versions from:
+
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess
+and
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub
+
+If $0 has already been updated, send the following data and any
+information you think might be pertinent to config-patches@gnu.org to
+provide the necessary information to handle your system.
+
+config.guess timestamp = $timestamp
+
+uname -m = `(uname -m) 2>/dev/null || echo unknown`
+uname -r = `(uname -r) 2>/dev/null || echo unknown`
+uname -s = `(uname -s) 2>/dev/null || echo unknown`
+uname -v = `(uname -v) 2>/dev/null || echo unknown`
+
+/usr/bin/uname -p = `(/usr/bin/uname -p) 2>/dev/null`
+/bin/uname -X     = `(/bin/uname -X) 2>/dev/null`
+
+hostinfo               = `(hostinfo) 2>/dev/null`
+/bin/universe          = `(/bin/universe) 2>/dev/null`
+/usr/bin/arch -k       = `(/usr/bin/arch -k) 2>/dev/null`
+/bin/arch              = `(/bin/arch) 2>/dev/null`
+/usr/bin/oslevel       = `(/usr/bin/oslevel) 2>/dev/null`
+/usr/convex/getsysinfo = `(/usr/convex/getsysinfo) 2>/dev/null`
+
+UNAME_MACHINE = ${UNAME_MACHINE}
+UNAME_RELEASE = ${UNAME_RELEASE}
+UNAME_SYSTEM  = ${UNAME_SYSTEM}
+UNAME_VERSION = ${UNAME_VERSION}
+EOF
+
+exit 1
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "timestamp='"
+# time-stamp-format: "%:y-%02m-%02d"
+# time-stamp-end: "'"
+# End:
diff --git a/memkind-1.10.0/jemalloc/build-aux/config.sub b/memkind-1.10.0/jemalloc/build-aux/config.sub
new file mode 100755
index 0000000..dd2ca93
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/build-aux/config.sub
@@ -0,0 +1,1825 @@
+#! /bin/sh
+# Configuration validation subroutine script.
+#   Copyright 1992-2016 Free Software Foundation, Inc.
+
+timestamp='2016-11-04'
+
+# This file is free software; you can redistribute it and/or modify it
+# under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful, but
+# WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+# General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, see <http://www.gnu.org/licenses/>.
+#
+# As a special exception to the GNU General Public License, if you
+# distribute this file as part of a program that contains a
+# configuration script generated by Autoconf, you may include it under
+# the same distribution terms that you use for the rest of that
+# program.  This Exception is an additional permission under section 7
+# of the GNU General Public License, version 3 ("GPLv3").
+
+
+# Please send patches to <config-patches@gnu.org>.
+#
+# Configuration subroutine to validate and canonicalize a configuration type.
+# Supply the specified configuration type as an argument.
+# If it is invalid, we print an error message on stderr and exit with code 1.
+# Otherwise, we print the canonical config type on stdout and succeed.
+
+# You can get the latest version of this script from:
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub
+
+# This file is supposed to be the same for all GNU packages
+# and recognize all the CPU types, system types and aliases
+# that are meaningful with *any* GNU software.
+# Each package is responsible for reporting which valid configurations
+# it does not support.  The user should be able to distinguish
+# a failure to support a valid configuration from a meaningless
+# configuration.
+
+# The goal of this file is to map all the various variations of a given
+# machine specification into a single specification in the form:
+#	CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM
+# or in some cases, the newer four-part form:
+#	CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM
+# It is wrong to echo any other type of specification.
+
+me=`echo "$0" | sed -e 's,.*/,,'`
+
+usage="\
+Usage: $0 [OPTION] CPU-MFR-OPSYS or ALIAS
+
+Canonicalize a configuration name.
+
+Operation modes:
+  -h, --help         print this help, then exit
+  -t, --time-stamp   print date of last modification, then exit
+  -v, --version      print version number, then exit
+
+Report bugs and patches to <config-patches@gnu.org>."
+
+version="\
+GNU config.sub ($timestamp)
+
+Copyright 1992-2016 Free Software Foundation, Inc.
+
+This is free software; see the source for copying conditions.  There is NO
+warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
+
+help="
+Try \`$me --help' for more information."
+
+# Parse command line
+while test $# -gt 0 ; do
+  case $1 in
+    --time-stamp | --time* | -t )
+       echo "$timestamp" ; exit ;;
+    --version | -v )
+       echo "$version" ; exit ;;
+    --help | --h* | -h )
+       echo "$usage"; exit ;;
+    -- )     # Stop option processing
+       shift; break ;;
+    - )	# Use stdin as input.
+       break ;;
+    -* )
+       echo "$me: invalid option $1$help"
+       exit 1 ;;
+
+    *local*)
+       # First pass through any local machine types.
+       echo $1
+       exit ;;
+
+    * )
+       break ;;
+  esac
+done
+
+case $# in
+ 0) echo "$me: missing argument$help" >&2
+    exit 1;;
+ 1) ;;
+ *) echo "$me: too many arguments$help" >&2
+    exit 1;;
+esac
+
+# Separate what the user gave into CPU-COMPANY and OS or KERNEL-OS (if any).
+# Here we must recognize all the valid KERNEL-OS combinations.
+maybe_os=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\2/'`
+case $maybe_os in
+  nto-qnx* | linux-gnu* | linux-android* | linux-dietlibc | linux-newlib* | \
+  linux-musl* | linux-uclibc* | uclinux-uclibc* | uclinux-gnu* | kfreebsd*-gnu* | \
+  knetbsd*-gnu* | netbsd*-gnu* | netbsd*-eabi* | \
+  kopensolaris*-gnu* | cloudabi*-eabi* | \
+  storm-chaos* | os2-emx* | rtmk-nova*)
+    os=-$maybe_os
+    basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`
+    ;;
+  android-linux)
+    os=-linux-android
+    basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`-unknown
+    ;;
+  *)
+    basic_machine=`echo $1 | sed 's/-[^-]*$//'`
+    if [ $basic_machine != $1 ]
+    then os=`echo $1 | sed 's/.*-/-/'`
+    else os=; fi
+    ;;
+esac
+
+### Let's recognize common machines as not being operating systems so
+### that things like config.sub decstation-3100 work.  We also
+### recognize some manufacturers as not being operating systems, so we
+### can provide default operating systems below.
+case $os in
+	-sun*os*)
+		# Prevent following clause from handling this invalid input.
+		;;
+	-dec* | -mips* | -sequent* | -encore* | -pc532* | -sgi* | -sony* | \
+	-att* | -7300* | -3300* | -delta* | -motorola* | -sun[234]* | \
+	-unicom* | -ibm* | -next | -hp | -isi* | -apollo | -altos* | \
+	-convergent* | -ncr* | -news | -32* | -3600* | -3100* | -hitachi* |\
+	-c[123]* | -convex* | -sun | -crds | -omron* | -dg | -ultra | -tti* | \
+	-harris | -dolphin | -highlevel | -gould | -cbm | -ns | -masscomp | \
+	-apple | -axis | -knuth | -cray | -microblaze*)
+		os=
+		basic_machine=$1
+		;;
+	-bluegene*)
+		os=-cnk
+		;;
+	-sim | -cisco | -oki | -wec | -winbond)
+		os=
+		basic_machine=$1
+		;;
+	-scout)
+		;;
+	-wrs)
+		os=-vxworks
+		basic_machine=$1
+		;;
+	-chorusos*)
+		os=-chorusos
+		basic_machine=$1
+		;;
+	-chorusrdb)
+		os=-chorusrdb
+		basic_machine=$1
+		;;
+	-hiux*)
+		os=-hiuxwe2
+		;;
+	-sco6)
+		os=-sco5v6
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco5)
+		os=-sco3.2v5
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco4)
+		os=-sco3.2v4
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco3.2.[4-9]*)
+		os=`echo $os | sed -e 's/sco3.2./sco3.2v/'`
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco3.2v[4-9]*)
+		# Don't forget version if it is 3.2v4 or newer.
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco5v6*)
+		# Don't forget version if it is 3.2v4 or newer.
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-sco*)
+		os=-sco3.2v2
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-udk*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-isc)
+		os=-isc2.2
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-clix*)
+		basic_machine=clipper-intergraph
+		;;
+	-isc*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-pc/'`
+		;;
+	-lynx*178)
+		os=-lynxos178
+		;;
+	-lynx*5)
+		os=-lynxos5
+		;;
+	-lynx*)
+		os=-lynxos
+		;;
+	-ptx*)
+		basic_machine=`echo $1 | sed -e 's/86-.*/86-sequent/'`
+		;;
+	-windowsnt*)
+		os=`echo $os | sed -e 's/windowsnt/winnt/'`
+		;;
+	-psos*)
+		os=-psos
+		;;
+	-mint | -mint[0-9]*)
+		basic_machine=m68k-atari
+		os=-mint
+		;;
+esac
+
+# Decode aliases for certain CPU-COMPANY combinations.
+case $basic_machine in
+	# Recognize the basic CPU types without company name.
+	# Some are omitted here because they have special meanings below.
+	1750a | 580 \
+	| a29k \
+	| aarch64 | aarch64_be \
+	| alpha | alphaev[4-8] | alphaev56 | alphaev6[78] | alphapca5[67] \
+	| alpha64 | alpha64ev[4-8] | alpha64ev56 | alpha64ev6[78] | alpha64pca5[67] \
+	| am33_2.0 \
+	| arc | arceb \
+	| arm | arm[bl]e | arme[lb] | armv[2-8] | armv[3-8][lb] | armv7[arm] \
+	| avr | avr32 \
+	| ba \
+	| be32 | be64 \
+	| bfin \
+	| c4x | c8051 | clipper \
+	| d10v | d30v | dlx | dsp16xx \
+	| e2k | epiphany \
+	| fido | fr30 | frv | ft32 \
+	| h8300 | h8500 | hppa | hppa1.[01] | hppa2.0 | hppa2.0[nw] | hppa64 \
+	| hexagon \
+	| i370 | i860 | i960 | ia64 \
+	| ip2k | iq2000 \
+	| k1om \
+	| le32 | le64 \
+	| lm32 \
+	| m32c | m32r | m32rle | m68000 | m68k | m88k \
+	| maxq | mb | microblaze | microblazeel | mcore | mep | metag \
+	| mips | mipsbe | mipseb | mipsel | mipsle \
+	| mips16 \
+	| mips64 | mips64el \
+	| mips64octeon | mips64octeonel \
+	| mips64orion | mips64orionel \
+	| mips64r5900 | mips64r5900el \
+	| mips64vr | mips64vrel \
+	| mips64vr4100 | mips64vr4100el \
+	| mips64vr4300 | mips64vr4300el \
+	| mips64vr5000 | mips64vr5000el \
+	| mips64vr5900 | mips64vr5900el \
+	| mipsisa32 | mipsisa32el \
+	| mipsisa32r2 | mipsisa32r2el \
+	| mipsisa32r6 | mipsisa32r6el \
+	| mipsisa64 | mipsisa64el \
+	| mipsisa64r2 | mipsisa64r2el \
+	| mipsisa64r6 | mipsisa64r6el \
+	| mipsisa64sb1 | mipsisa64sb1el \
+	| mipsisa64sr71k | mipsisa64sr71kel \
+	| mipsr5900 | mipsr5900el \
+	| mipstx39 | mipstx39el \
+	| mn10200 | mn10300 \
+	| moxie \
+	| mt \
+	| msp430 \
+	| nds32 | nds32le | nds32be \
+	| nios | nios2 | nios2eb | nios2el \
+	| ns16k | ns32k \
+	| open8 | or1k | or1knd | or32 \
+	| pdp10 | pdp11 | pj | pjl \
+	| powerpc | powerpc64 | powerpc64le | powerpcle \
+	| pru \
+	| pyramid \
+	| riscv32 | riscv64 \
+	| rl78 | rx \
+	| score \
+	| sh | sh[1234] | sh[24]a | sh[24]aeb | sh[23]e | sh[234]eb | sheb | shbe | shle | sh[1234]le | sh3ele \
+	| sh64 | sh64le \
+	| sparc | sparc64 | sparc64b | sparc64v | sparc86x | sparclet | sparclite \
+	| sparcv8 | sparcv9 | sparcv9b | sparcv9v \
+	| spu \
+	| tahoe | tic4x | tic54x | tic55x | tic6x | tic80 | tron \
+	| ubicom32 \
+	| v850 | v850e | v850e1 | v850e2 | v850es | v850e2v3 \
+	| visium \
+	| we32k \
+	| x86 | xc16x | xstormy16 | xtensa \
+	| z8k | z80)
+		basic_machine=$basic_machine-unknown
+		;;
+	c54x)
+		basic_machine=tic54x-unknown
+		;;
+	c55x)
+		basic_machine=tic55x-unknown
+		;;
+	c6x)
+		basic_machine=tic6x-unknown
+		;;
+	leon|leon[3-9])
+		basic_machine=sparc-$basic_machine
+		;;
+	m6811 | m68hc11 | m6812 | m68hc12 | m68hcs12x | nvptx | picochip)
+		basic_machine=$basic_machine-unknown
+		os=-none
+		;;
+	m88110 | m680[12346]0 | m683?2 | m68360 | m5200 | v70 | w65 | z8k)
+		;;
+	ms1)
+		basic_machine=mt-unknown
+		;;
+
+	strongarm | thumb | xscale)
+		basic_machine=arm-unknown
+		;;
+	xgate)
+		basic_machine=$basic_machine-unknown
+		os=-none
+		;;
+	xscaleeb)
+		basic_machine=armeb-unknown
+		;;
+
+	xscaleel)
+		basic_machine=armel-unknown
+		;;
+
+	# We use `pc' rather than `unknown'
+	# because (1) that's what they normally are, and
+	# (2) the word "unknown" tends to confuse beginning users.
+	i*86 | x86_64)
+	  basic_machine=$basic_machine-pc
+	  ;;
+	# Object if more than one company name word.
+	*-*-*)
+		echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2
+		exit 1
+		;;
+	# Recognize the basic CPU types with company name.
+	580-* \
+	| a29k-* \
+	| aarch64-* | aarch64_be-* \
+	| alpha-* | alphaev[4-8]-* | alphaev56-* | alphaev6[78]-* \
+	| alpha64-* | alpha64ev[4-8]-* | alpha64ev56-* | alpha64ev6[78]-* \
+	| alphapca5[67]-* | alpha64pca5[67]-* | arc-* | arceb-* \
+	| arm-*  | armbe-* | armle-* | armeb-* | armv*-* \
+	| avr-* | avr32-* \
+	| ba-* \
+	| be32-* | be64-* \
+	| bfin-* | bs2000-* \
+	| c[123]* | c30-* | [cjt]90-* | c4x-* \
+	| c8051-* | clipper-* | craynv-* | cydra-* \
+	| d10v-* | d30v-* | dlx-* \
+	| e2k-* | elxsi-* \
+	| f30[01]-* | f700-* | fido-* | fr30-* | frv-* | fx80-* \
+	| h8300-* | h8500-* \
+	| hppa-* | hppa1.[01]-* | hppa2.0-* | hppa2.0[nw]-* | hppa64-* \
+	| hexagon-* \
+	| i*86-* | i860-* | i960-* | ia64-* \
+	| ip2k-* | iq2000-* \
+	| k1om-* \
+	| le32-* | le64-* \
+	| lm32-* \
+	| m32c-* | m32r-* | m32rle-* \
+	| m68000-* | m680[012346]0-* | m68360-* | m683?2-* | m68k-* \
+	| m88110-* | m88k-* | maxq-* | mcore-* | metag-* \
+	| microblaze-* | microblazeel-* \
+	| mips-* | mipsbe-* | mipseb-* | mipsel-* | mipsle-* \
+	| mips16-* \
+	| mips64-* | mips64el-* \
+	| mips64octeon-* | mips64octeonel-* \
+	| mips64orion-* | mips64orionel-* \
+	| mips64r5900-* | mips64r5900el-* \
+	| mips64vr-* | mips64vrel-* \
+	| mips64vr4100-* | mips64vr4100el-* \
+	| mips64vr4300-* | mips64vr4300el-* \
+	| mips64vr5000-* | mips64vr5000el-* \
+	| mips64vr5900-* | mips64vr5900el-* \
+	| mipsisa32-* | mipsisa32el-* \
+	| mipsisa32r2-* | mipsisa32r2el-* \
+	| mipsisa32r6-* | mipsisa32r6el-* \
+	| mipsisa64-* | mipsisa64el-* \
+	| mipsisa64r2-* | mipsisa64r2el-* \
+	| mipsisa64r6-* | mipsisa64r6el-* \
+	| mipsisa64sb1-* | mipsisa64sb1el-* \
+	| mipsisa64sr71k-* | mipsisa64sr71kel-* \
+	| mipsr5900-* | mipsr5900el-* \
+	| mipstx39-* | mipstx39el-* \
+	| mmix-* \
+	| mt-* \
+	| msp430-* \
+	| nds32-* | nds32le-* | nds32be-* \
+	| nios-* | nios2-* | nios2eb-* | nios2el-* \
+	| none-* | np1-* | ns16k-* | ns32k-* \
+	| open8-* \
+	| or1k*-* \
+	| orion-* \
+	| pdp10-* | pdp11-* | pj-* | pjl-* | pn-* | power-* \
+	| powerpc-* | powerpc64-* | powerpc64le-* | powerpcle-* \
+	| pru-* \
+	| pyramid-* \
+	| riscv32-* | riscv64-* \
+	| rl78-* | romp-* | rs6000-* | rx-* \
+	| sh-* | sh[1234]-* | sh[24]a-* | sh[24]aeb-* | sh[23]e-* | sh[34]eb-* | sheb-* | shbe-* \
+	| shle-* | sh[1234]le-* | sh3ele-* | sh64-* | sh64le-* \
+	| sparc-* | sparc64-* | sparc64b-* | sparc64v-* | sparc86x-* | sparclet-* \
+	| sparclite-* \
+	| sparcv8-* | sparcv9-* | sparcv9b-* | sparcv9v-* | sv1-* | sx*-* \
+	| tahoe-* \
+	| tic30-* | tic4x-* | tic54x-* | tic55x-* | tic6x-* | tic80-* \
+	| tile*-* \
+	| tron-* \
+	| ubicom32-* \
+	| v850-* | v850e-* | v850e1-* | v850es-* | v850e2-* | v850e2v3-* \
+	| vax-* \
+	| visium-* \
+	| we32k-* \
+	| x86-* | x86_64-* | xc16x-* | xps100-* \
+	| xstormy16-* | xtensa*-* \
+	| ymp-* \
+	| z8k-* | z80-*)
+		;;
+	# Recognize the basic CPU types without company name, with glob match.
+	xtensa*)
+		basic_machine=$basic_machine-unknown
+		;;
+	# Recognize the various machine names and aliases which stand
+	# for a CPU type and a company and sometimes even an OS.
+	386bsd)
+		basic_machine=i386-unknown
+		os=-bsd
+		;;
+	3b1 | 7300 | 7300-att | att-7300 | pc7300 | safari | unixpc)
+		basic_machine=m68000-att
+		;;
+	3b*)
+		basic_machine=we32k-att
+		;;
+	a29khif)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	abacus)
+		basic_machine=abacus-unknown
+		;;
+	adobe68k)
+		basic_machine=m68010-adobe
+		os=-scout
+		;;
+	alliant | fx80)
+		basic_machine=fx80-alliant
+		;;
+	altos | altos3068)
+		basic_machine=m68k-altos
+		;;
+	am29k)
+		basic_machine=a29k-none
+		os=-bsd
+		;;
+	amd64)
+		basic_machine=x86_64-pc
+		;;
+	amd64-*)
+		basic_machine=x86_64-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	amdahl)
+		basic_machine=580-amdahl
+		os=-sysv
+		;;
+	amiga | amiga-*)
+		basic_machine=m68k-unknown
+		;;
+	amigaos | amigados)
+		basic_machine=m68k-unknown
+		os=-amigaos
+		;;
+	amigaunix | amix)
+		basic_machine=m68k-unknown
+		os=-sysv4
+		;;
+	apollo68)
+		basic_machine=m68k-apollo
+		os=-sysv
+		;;
+	apollo68bsd)
+		basic_machine=m68k-apollo
+		os=-bsd
+		;;
+	aros)
+		basic_machine=i386-pc
+		os=-aros
+		;;
+	asmjs)
+		basic_machine=asmjs-unknown
+		;;
+	aux)
+		basic_machine=m68k-apple
+		os=-aux
+		;;
+	balance)
+		basic_machine=ns32k-sequent
+		os=-dynix
+		;;
+	blackfin)
+		basic_machine=bfin-unknown
+		os=-linux
+		;;
+	blackfin-*)
+		basic_machine=bfin-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	bluegene*)
+		basic_machine=powerpc-ibm
+		os=-cnk
+		;;
+	c54x-*)
+		basic_machine=tic54x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c55x-*)
+		basic_machine=tic55x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c6x-*)
+		basic_machine=tic6x-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	c90)
+		basic_machine=c90-cray
+		os=-unicos
+		;;
+	cegcc)
+		basic_machine=arm-unknown
+		os=-cegcc
+		;;
+	convex-c1)
+		basic_machine=c1-convex
+		os=-bsd
+		;;
+	convex-c2)
+		basic_machine=c2-convex
+		os=-bsd
+		;;
+	convex-c32)
+		basic_machine=c32-convex
+		os=-bsd
+		;;
+	convex-c34)
+		basic_machine=c34-convex
+		os=-bsd
+		;;
+	convex-c38)
+		basic_machine=c38-convex
+		os=-bsd
+		;;
+	cray | j90)
+		basic_machine=j90-cray
+		os=-unicos
+		;;
+	craynv)
+		basic_machine=craynv-cray
+		os=-unicosmp
+		;;
+	cr16 | cr16-*)
+		basic_machine=cr16-unknown
+		os=-elf
+		;;
+	crds | unos)
+		basic_machine=m68k-crds
+		;;
+	crisv32 | crisv32-* | etraxfs*)
+		basic_machine=crisv32-axis
+		;;
+	cris | cris-* | etrax*)
+		basic_machine=cris-axis
+		;;
+	crx)
+		basic_machine=crx-unknown
+		os=-elf
+		;;
+	da30 | da30-*)
+		basic_machine=m68k-da30
+		;;
+	decstation | decstation-3100 | pmax | pmax-* | pmin | dec3100 | decstatn)
+		basic_machine=mips-dec
+		;;
+	decsystem10* | dec10*)
+		basic_machine=pdp10-dec
+		os=-tops10
+		;;
+	decsystem20* | dec20*)
+		basic_machine=pdp10-dec
+		os=-tops20
+		;;
+	delta | 3300 | motorola-3300 | motorola-delta \
+	      | 3300-motorola | delta-motorola)
+		basic_machine=m68k-motorola
+		;;
+	delta88)
+		basic_machine=m88k-motorola
+		os=-sysv3
+		;;
+	dicos)
+		basic_machine=i686-pc
+		os=-dicos
+		;;
+	djgpp)
+		basic_machine=i586-pc
+		os=-msdosdjgpp
+		;;
+	dpx20 | dpx20-*)
+		basic_machine=rs6000-bull
+		os=-bosx
+		;;
+	dpx2* | dpx2*-bull)
+		basic_machine=m68k-bull
+		os=-sysv3
+		;;
+	e500v[12])
+		basic_machine=powerpc-unknown
+		os=$os"spe"
+		;;
+	e500v[12]-*)
+		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=$os"spe"
+		;;
+	ebmon29k)
+		basic_machine=a29k-amd
+		os=-ebmon
+		;;
+	elxsi)
+		basic_machine=elxsi-elxsi
+		os=-bsd
+		;;
+	encore | umax | mmax)
+		basic_machine=ns32k-encore
+		;;
+	es1800 | OSE68k | ose68k | ose | OSE)
+		basic_machine=m68k-ericsson
+		os=-ose
+		;;
+	fx2800)
+		basic_machine=i860-alliant
+		;;
+	genix)
+		basic_machine=ns32k-ns
+		;;
+	gmicro)
+		basic_machine=tron-gmicro
+		os=-sysv
+		;;
+	go32)
+		basic_machine=i386-pc
+		os=-go32
+		;;
+	h3050r* | hiux*)
+		basic_machine=hppa1.1-hitachi
+		os=-hiuxwe2
+		;;
+	h8300hms)
+		basic_machine=h8300-hitachi
+		os=-hms
+		;;
+	h8300xray)
+		basic_machine=h8300-hitachi
+		os=-xray
+		;;
+	h8500hms)
+		basic_machine=h8500-hitachi
+		os=-hms
+		;;
+	harris)
+		basic_machine=m88k-harris
+		os=-sysv3
+		;;
+	hp300-*)
+		basic_machine=m68k-hp
+		;;
+	hp300bsd)
+		basic_machine=m68k-hp
+		os=-bsd
+		;;
+	hp300hpux)
+		basic_machine=m68k-hp
+		os=-hpux
+		;;
+	hp3k9[0-9][0-9] | hp9[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hp9k2[0-9][0-9] | hp9k31[0-9])
+		basic_machine=m68000-hp
+		;;
+	hp9k3[2-9][0-9])
+		basic_machine=m68k-hp
+		;;
+	hp9k6[0-9][0-9] | hp6[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hp9k7[0-79][0-9] | hp7[0-79][0-9])
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k78[0-9] | hp78[0-9])
+		# FIXME: really hppa2.0-hp
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[67]1 | hp8[67]1 | hp9k80[24] | hp80[24] | hp9k8[78]9 | hp8[78]9 | hp9k893 | hp893)
+		# FIXME: really hppa2.0-hp
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[0-9][13679] | hp8[0-9][13679])
+		basic_machine=hppa1.1-hp
+		;;
+	hp9k8[0-9][0-9] | hp8[0-9][0-9])
+		basic_machine=hppa1.0-hp
+		;;
+	hppa-next)
+		os=-nextstep3
+		;;
+	hppaosf)
+		basic_machine=hppa1.1-hp
+		os=-osf
+		;;
+	hppro)
+		basic_machine=hppa1.1-hp
+		os=-proelf
+		;;
+	i370-ibm* | ibm*)
+		basic_machine=i370-ibm
+		;;
+	i*86v32)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv32
+		;;
+	i*86v4*)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv4
+		;;
+	i*86v)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-sysv
+		;;
+	i*86sol2)
+		basic_machine=`echo $1 | sed -e 's/86.*/86-pc/'`
+		os=-solaris2
+		;;
+	i386mach)
+		basic_machine=i386-mach
+		os=-mach
+		;;
+	i386-vsta | vsta)
+		basic_machine=i386-unknown
+		os=-vsta
+		;;
+	iris | iris4d)
+		basic_machine=mips-sgi
+		case $os in
+		    -irix*)
+			;;
+		    *)
+			os=-irix4
+			;;
+		esac
+		;;
+	isi68 | isi)
+		basic_machine=m68k-isi
+		os=-sysv
+		;;
+	leon-*|leon[3-9]-*)
+		basic_machine=sparc-`echo $basic_machine | sed 's/-.*//'`
+		;;
+	m68knommu)
+		basic_machine=m68k-unknown
+		os=-linux
+		;;
+	m68knommu-*)
+		basic_machine=m68k-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	m88k-omron*)
+		basic_machine=m88k-omron
+		;;
+	magnum | m3230)
+		basic_machine=mips-mips
+		os=-sysv
+		;;
+	merlin)
+		basic_machine=ns32k-utek
+		os=-sysv
+		;;
+	microblaze*)
+		basic_machine=microblaze-xilinx
+		;;
+	mingw64)
+		basic_machine=x86_64-pc
+		os=-mingw64
+		;;
+	mingw32)
+		basic_machine=i686-pc
+		os=-mingw32
+		;;
+	mingw32ce)
+		basic_machine=arm-unknown
+		os=-mingw32ce
+		;;
+	miniframe)
+		basic_machine=m68000-convergent
+		;;
+	*mint | -mint[0-9]* | *MiNT | *MiNT[0-9]*)
+		basic_machine=m68k-atari
+		os=-mint
+		;;
+	mips3*-*)
+		basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'`
+		;;
+	mips3*)
+		basic_machine=`echo $basic_machine | sed -e 's/mips3/mips64/'`-unknown
+		;;
+	monitor)
+		basic_machine=m68k-rom68k
+		os=-coff
+		;;
+	morphos)
+		basic_machine=powerpc-unknown
+		os=-morphos
+		;;
+	moxiebox)
+		basic_machine=moxie-unknown
+		os=-moxiebox
+		;;
+	msdos)
+		basic_machine=i386-pc
+		os=-msdos
+		;;
+	ms1-*)
+		basic_machine=`echo $basic_machine | sed -e 's/ms1-/mt-/'`
+		;;
+	msys)
+		basic_machine=i686-pc
+		os=-msys
+		;;
+	mvs)
+		basic_machine=i370-ibm
+		os=-mvs
+		;;
+	nacl)
+		basic_machine=le32-unknown
+		os=-nacl
+		;;
+	ncr3000)
+		basic_machine=i486-ncr
+		os=-sysv4
+		;;
+	netbsd386)
+		basic_machine=i386-unknown
+		os=-netbsd
+		;;
+	netwinder)
+		basic_machine=armv4l-rebel
+		os=-linux
+		;;
+	news | news700 | news800 | news900)
+		basic_machine=m68k-sony
+		os=-newsos
+		;;
+	news1000)
+		basic_machine=m68030-sony
+		os=-newsos
+		;;
+	news-3600 | risc-news)
+		basic_machine=mips-sony
+		os=-newsos
+		;;
+	necv70)
+		basic_machine=v70-nec
+		os=-sysv
+		;;
+	next | m*-next )
+		basic_machine=m68k-next
+		case $os in
+		    -nextstep* )
+			;;
+		    -ns2*)
+		      os=-nextstep2
+			;;
+		    *)
+		      os=-nextstep3
+			;;
+		esac
+		;;
+	nh3000)
+		basic_machine=m68k-harris
+		os=-cxux
+		;;
+	nh[45]000)
+		basic_machine=m88k-harris
+		os=-cxux
+		;;
+	nindy960)
+		basic_machine=i960-intel
+		os=-nindy
+		;;
+	mon960)
+		basic_machine=i960-intel
+		os=-mon960
+		;;
+	nonstopux)
+		basic_machine=mips-compaq
+		os=-nonstopux
+		;;
+	np1)
+		basic_machine=np1-gould
+		;;
+	neo-tandem)
+		basic_machine=neo-tandem
+		;;
+	nse-tandem)
+		basic_machine=nse-tandem
+		;;
+	nsr-tandem)
+		basic_machine=nsr-tandem
+		;;
+	op50n-* | op60c-*)
+		basic_machine=hppa1.1-oki
+		os=-proelf
+		;;
+	openrisc | openrisc-*)
+		basic_machine=or32-unknown
+		;;
+	os400)
+		basic_machine=powerpc-ibm
+		os=-os400
+		;;
+	OSE68000 | ose68000)
+		basic_machine=m68000-ericsson
+		os=-ose
+		;;
+	os68k)
+		basic_machine=m68k-none
+		os=-os68k
+		;;
+	pa-hitachi)
+		basic_machine=hppa1.1-hitachi
+		os=-hiuxwe2
+		;;
+	paragon)
+		basic_machine=i860-intel
+		os=-osf
+		;;
+	parisc)
+		basic_machine=hppa-unknown
+		os=-linux
+		;;
+	parisc-*)
+		basic_machine=hppa-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=-linux
+		;;
+	pbd)
+		basic_machine=sparc-tti
+		;;
+	pbb)
+		basic_machine=m68k-tti
+		;;
+	pc532 | pc532-*)
+		basic_machine=ns32k-pc532
+		;;
+	pc98)
+		basic_machine=i386-pc
+		;;
+	pc98-*)
+		basic_machine=i386-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentium | p5 | k5 | k6 | nexgen | viac3)
+		basic_machine=i586-pc
+		;;
+	pentiumpro | p6 | 6x86 | athlon | athlon_*)
+		basic_machine=i686-pc
+		;;
+	pentiumii | pentium2 | pentiumiii | pentium3)
+		basic_machine=i686-pc
+		;;
+	pentium4)
+		basic_machine=i786-pc
+		;;
+	pentium-* | p5-* | k5-* | k6-* | nexgen-* | viac3-*)
+		basic_machine=i586-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentiumpro-* | p6-* | 6x86-* | athlon-*)
+		basic_machine=i686-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentiumii-* | pentium2-* | pentiumiii-* | pentium3-*)
+		basic_machine=i686-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pentium4-*)
+		basic_machine=i786-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	pn)
+		basic_machine=pn-gould
+		;;
+	power)	basic_machine=power-ibm
+		;;
+	ppc | ppcbe)	basic_machine=powerpc-unknown
+		;;
+	ppc-* | ppcbe-*)
+		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppcle | powerpclittle)
+		basic_machine=powerpcle-unknown
+		;;
+	ppcle-* | powerpclittle-*)
+		basic_machine=powerpcle-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppc64)	basic_machine=powerpc64-unknown
+		;;
+	ppc64-*) basic_machine=powerpc64-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ppc64le | powerpc64little)
+		basic_machine=powerpc64le-unknown
+		;;
+	ppc64le-* | powerpc64little-*)
+		basic_machine=powerpc64le-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	ps2)
+		basic_machine=i386-ibm
+		;;
+	pw32)
+		basic_machine=i586-unknown
+		os=-pw32
+		;;
+	rdos | rdos64)
+		basic_machine=x86_64-pc
+		os=-rdos
+		;;
+	rdos32)
+		basic_machine=i386-pc
+		os=-rdos
+		;;
+	rom68k)
+		basic_machine=m68k-rom68k
+		os=-coff
+		;;
+	rm[46]00)
+		basic_machine=mips-siemens
+		;;
+	rtpc | rtpc-*)
+		basic_machine=romp-ibm
+		;;
+	s390 | s390-*)
+		basic_machine=s390-ibm
+		;;
+	s390x | s390x-*)
+		basic_machine=s390x-ibm
+		;;
+	sa29200)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	sb1)
+		basic_machine=mipsisa64sb1-unknown
+		;;
+	sb1el)
+		basic_machine=mipsisa64sb1el-unknown
+		;;
+	sde)
+		basic_machine=mipsisa32-sde
+		os=-elf
+		;;
+	sei)
+		basic_machine=mips-sei
+		os=-seiux
+		;;
+	sequent)
+		basic_machine=i386-sequent
+		;;
+	sh)
+		basic_machine=sh-hitachi
+		os=-hms
+		;;
+	sh5el)
+		basic_machine=sh5le-unknown
+		;;
+	sh64)
+		basic_machine=sh64-unknown
+		;;
+	sparclite-wrs | simso-wrs)
+		basic_machine=sparclite-wrs
+		os=-vxworks
+		;;
+	sps7)
+		basic_machine=m68k-bull
+		os=-sysv2
+		;;
+	spur)
+		basic_machine=spur-unknown
+		;;
+	st2000)
+		basic_machine=m68k-tandem
+		;;
+	stratus)
+		basic_machine=i860-stratus
+		os=-sysv4
+		;;
+	strongarm-* | thumb-*)
+		basic_machine=arm-`echo $basic_machine | sed 's/^[^-]*-//'`
+		;;
+	sun2)
+		basic_machine=m68000-sun
+		;;
+	sun2os3)
+		basic_machine=m68000-sun
+		os=-sunos3
+		;;
+	sun2os4)
+		basic_machine=m68000-sun
+		os=-sunos4
+		;;
+	sun3os3)
+		basic_machine=m68k-sun
+		os=-sunos3
+		;;
+	sun3os4)
+		basic_machine=m68k-sun
+		os=-sunos4
+		;;
+	sun4os3)
+		basic_machine=sparc-sun
+		os=-sunos3
+		;;
+	sun4os4)
+		basic_machine=sparc-sun
+		os=-sunos4
+		;;
+	sun4sol2)
+		basic_machine=sparc-sun
+		os=-solaris2
+		;;
+	sun3 | sun3-*)
+		basic_machine=m68k-sun
+		;;
+	sun4)
+		basic_machine=sparc-sun
+		;;
+	sun386 | sun386i | roadrunner)
+		basic_machine=i386-sun
+		;;
+	sv1)
+		basic_machine=sv1-cray
+		os=-unicos
+		;;
+	symmetry)
+		basic_machine=i386-sequent
+		os=-dynix
+		;;
+	t3e)
+		basic_machine=alphaev5-cray
+		os=-unicos
+		;;
+	t90)
+		basic_machine=t90-cray
+		os=-unicos
+		;;
+	tile*)
+		basic_machine=$basic_machine-unknown
+		os=-linux-gnu
+		;;
+	tx39)
+		basic_machine=mipstx39-unknown
+		;;
+	tx39el)
+		basic_machine=mipstx39el-unknown
+		;;
+	toad1)
+		basic_machine=pdp10-xkl
+		os=-tops20
+		;;
+	tower | tower-32)
+		basic_machine=m68k-ncr
+		;;
+	tpf)
+		basic_machine=s390x-ibm
+		os=-tpf
+		;;
+	udi29k)
+		basic_machine=a29k-amd
+		os=-udi
+		;;
+	ultra3)
+		basic_machine=a29k-nyu
+		os=-sym1
+		;;
+	v810 | necv810)
+		basic_machine=v810-nec
+		os=-none
+		;;
+	vaxv)
+		basic_machine=vax-dec
+		os=-sysv
+		;;
+	vms)
+		basic_machine=vax-dec
+		os=-vms
+		;;
+	vpp*|vx|vx-*)
+		basic_machine=f301-fujitsu
+		;;
+	vxworks960)
+		basic_machine=i960-wrs
+		os=-vxworks
+		;;
+	vxworks68)
+		basic_machine=m68k-wrs
+		os=-vxworks
+		;;
+	vxworks29k)
+		basic_machine=a29k-wrs
+		os=-vxworks
+		;;
+	w65*)
+		basic_machine=w65-wdc
+		os=-none
+		;;
+	w89k-*)
+		basic_machine=hppa1.1-winbond
+		os=-proelf
+		;;
+	xbox)
+		basic_machine=i686-pc
+		os=-mingw32
+		;;
+	xps | xps100)
+		basic_machine=xps100-honeywell
+		;;
+	xscale-* | xscalee[bl]-*)
+		basic_machine=`echo $basic_machine | sed 's/^xscale/arm/'`
+		;;
+	ymp)
+		basic_machine=ymp-cray
+		os=-unicos
+		;;
+	z8k-*-coff)
+		basic_machine=z8k-unknown
+		os=-sim
+		;;
+	z80-*-coff)
+		basic_machine=z80-unknown
+		os=-sim
+		;;
+	none)
+		basic_machine=none-none
+		os=-none
+		;;
+
+# Here we handle the default manufacturer of certain CPU types.  It is in
+# some cases the only manufacturer, in others, it is the most popular.
+	w89k)
+		basic_machine=hppa1.1-winbond
+		;;
+	op50n)
+		basic_machine=hppa1.1-oki
+		;;
+	op60c)
+		basic_machine=hppa1.1-oki
+		;;
+	romp)
+		basic_machine=romp-ibm
+		;;
+	mmix)
+		basic_machine=mmix-knuth
+		;;
+	rs6000)
+		basic_machine=rs6000-ibm
+		;;
+	vax)
+		basic_machine=vax-dec
+		;;
+	pdp10)
+		# there are many clones, so DEC is not a safe bet
+		basic_machine=pdp10-unknown
+		;;
+	pdp11)
+		basic_machine=pdp11-dec
+		;;
+	we32k)
+		basic_machine=we32k-att
+		;;
+	sh[1234] | sh[24]a | sh[24]aeb | sh[34]eb | sh[1234]le | sh[23]ele)
+		basic_machine=sh-unknown
+		;;
+	sparc | sparcv8 | sparcv9 | sparcv9b | sparcv9v)
+		basic_machine=sparc-sun
+		;;
+	cydra)
+		basic_machine=cydra-cydrome
+		;;
+	orion)
+		basic_machine=orion-highlevel
+		;;
+	orion105)
+		basic_machine=clipper-highlevel
+		;;
+	mac | mpw | mac-mpw)
+		basic_machine=m68k-apple
+		;;
+	pmac | pmac-mpw)
+		basic_machine=powerpc-apple
+		;;
+	*-unknown)
+		# Make sure to match an already-canonicalized machine name.
+		;;
+	*)
+		echo Invalid configuration \`$1\': machine \`$basic_machine\' not recognized 1>&2
+		exit 1
+		;;
+esac
+
+# Here we canonicalize certain aliases for manufacturers.
+case $basic_machine in
+	*-digital*)
+		basic_machine=`echo $basic_machine | sed 's/digital.*/dec/'`
+		;;
+	*-commodore*)
+		basic_machine=`echo $basic_machine | sed 's/commodore.*/cbm/'`
+		;;
+	*)
+		;;
+esac
+
+# Decode manufacturer-specific aliases for certain operating systems.
+
+if [ x"$os" != x"" ]
+then
+case $os in
+	# First match some system type aliases
+	# that might get confused with valid system types.
+	# -solaris* is a basic system type, with this one exception.
+	-auroraux)
+		os=-auroraux
+		;;
+	-solaris1 | -solaris1.*)
+		os=`echo $os | sed -e 's|solaris1|sunos4|'`
+		;;
+	-solaris)
+		os=-solaris2
+		;;
+	-svr4*)
+		os=-sysv4
+		;;
+	-unixware*)
+		os=-sysv4.2uw
+		;;
+	-gnu/linux*)
+		os=`echo $os | sed -e 's|gnu/linux|linux-gnu|'`
+		;;
+	# First accept the basic system types.
+	# The portable systems comes first.
+	# Each alternative MUST END IN A *, to match a version number.
+	# -sysv* is not here because it comes later, after sysvr4.
+	-gnu* | -bsd* | -mach* | -minix* | -genix* | -ultrix* | -irix* \
+	      | -*vms* | -sco* | -esix* | -isc* | -aix* | -cnk* | -sunos | -sunos[34]*\
+	      | -hpux* | -unos* | -osf* | -luna* | -dgux* | -auroraux* | -solaris* \
+	      | -sym* | -kopensolaris* | -plan9* \
+	      | -amigaos* | -amigados* | -msdos* | -newsos* | -unicos* | -aof* \
+	      | -aos* | -aros* | -cloudabi* | -sortix* \
+	      | -nindy* | -vxsim* | -vxworks* | -ebmon* | -hms* | -mvs* \
+	      | -clix* | -riscos* | -uniplus* | -iris* | -rtu* | -xenix* \
+	      | -hiux* | -386bsd* | -knetbsd* | -mirbsd* | -netbsd* \
+	      | -bitrig* | -openbsd* | -solidbsd* | -libertybsd* \
+	      | -ekkobsd* | -kfreebsd* | -freebsd* | -riscix* | -lynxos* \
+	      | -bosx* | -nextstep* | -cxux* | -aout* | -elf* | -oabi* \
+	      | -ptx* | -coff* | -ecoff* | -winnt* | -domain* | -vsta* \
+	      | -udi* | -eabi* | -lites* | -ieee* | -go32* | -aux* \
+	      | -chorusos* | -chorusrdb* | -cegcc* \
+	      | -cygwin* | -msys* | -pe* | -psos* | -moss* | -proelf* | -rtems* \
+	      | -midipix* | -mingw32* | -mingw64* | -linux-gnu* | -linux-android* \
+	      | -linux-newlib* | -linux-musl* | -linux-uclibc* \
+	      | -uxpv* | -beos* | -mpeix* | -udk* | -moxiebox* \
+	      | -interix* | -uwin* | -mks* | -rhapsody* | -darwin* | -opened* \
+	      | -openstep* | -oskit* | -conix* | -pw32* | -nonstopux* \
+	      | -storm-chaos* | -tops10* | -tenex* | -tops20* | -its* \
+	      | -os2* | -vos* | -palmos* | -uclinux* | -nucleus* \
+	      | -morphos* | -superux* | -rtmk* | -rtmk-nova* | -windiss* \
+	      | -powermax* | -dnix* | -nx6 | -nx7 | -sei* | -dragonfly* \
+	      | -skyos* | -haiku* | -rdos* | -toppers* | -drops* | -es* \
+	      | -onefs* | -tirtos* | -phoenix* | -fuchsia*)
+	# Remember, each alternative MUST END IN *, to match a version number.
+		;;
+	-qnx*)
+		case $basic_machine in
+		    x86-* | i*86-*)
+			;;
+		    *)
+			os=-nto$os
+			;;
+		esac
+		;;
+	-nto-qnx*)
+		;;
+	-nto*)
+		os=`echo $os | sed -e 's|nto|nto-qnx|'`
+		;;
+	-sim | -es1800* | -hms* | -xray | -os68k* | -none* | -v88r* \
+	      | -windows* | -osx | -abug | -netware* | -os9* | -beos* | -haiku* \
+	      | -macos* | -mpw* | -magic* | -mmixware* | -mon960* | -lnews*)
+		;;
+	-mac*)
+		os=`echo $os | sed -e 's|mac|macos|'`
+		;;
+	-linux-dietlibc)
+		os=-linux-dietlibc
+		;;
+	-linux*)
+		os=`echo $os | sed -e 's|linux|linux-gnu|'`
+		;;
+	-sunos5*)
+		os=`echo $os | sed -e 's|sunos5|solaris2|'`
+		;;
+	-sunos6*)
+		os=`echo $os | sed -e 's|sunos6|solaris3|'`
+		;;
+	-opened*)
+		os=-openedition
+		;;
+	-os400*)
+		os=-os400
+		;;
+	-wince*)
+		os=-wince
+		;;
+	-osfrose*)
+		os=-osfrose
+		;;
+	-osf*)
+		os=-osf
+		;;
+	-utek*)
+		os=-bsd
+		;;
+	-dynix*)
+		os=-bsd
+		;;
+	-acis*)
+		os=-aos
+		;;
+	-atheos*)
+		os=-atheos
+		;;
+	-syllable*)
+		os=-syllable
+		;;
+	-386bsd)
+		os=-bsd
+		;;
+	-ctix* | -uts*)
+		os=-sysv
+		;;
+	-nova*)
+		os=-rtmk-nova
+		;;
+	-ns2 )
+		os=-nextstep2
+		;;
+	-nsk*)
+		os=-nsk
+		;;
+	# Preserve the version number of sinix5.
+	-sinix5.*)
+		os=`echo $os | sed -e 's|sinix|sysv|'`
+		;;
+	-sinix*)
+		os=-sysv4
+		;;
+	-tpf*)
+		os=-tpf
+		;;
+	-triton*)
+		os=-sysv3
+		;;
+	-oss*)
+		os=-sysv3
+		;;
+	-svr4)
+		os=-sysv4
+		;;
+	-svr3)
+		os=-sysv3
+		;;
+	-sysvr4)
+		os=-sysv4
+		;;
+	# This must come after -sysvr4.
+	-sysv*)
+		;;
+	-ose*)
+		os=-ose
+		;;
+	-es1800*)
+		os=-ose
+		;;
+	-xenix)
+		os=-xenix
+		;;
+	-*mint | -mint[0-9]* | -*MiNT | -MiNT[0-9]*)
+		os=-mint
+		;;
+	-aros*)
+		os=-aros
+		;;
+	-zvmoe)
+		os=-zvmoe
+		;;
+	-dicos*)
+		os=-dicos
+		;;
+	-nacl*)
+		;;
+	-ios)
+		;;
+	-none)
+		;;
+	*)
+		# Get rid of the `-' at the beginning of $os.
+		os=`echo $os | sed 's/[^-]*-//'`
+		echo Invalid configuration \`$1\': system \`$os\' not recognized 1>&2
+		exit 1
+		;;
+esac
+else
+
+# Here we handle the default operating systems that come with various machines.
+# The value should be what the vendor currently ships out the door with their
+# machine or put another way, the most popular os provided with the machine.
+
+# Note that if you're going to try to match "-MANUFACTURER" here (say,
+# "-sun"), then you have to tell the case statement up towards the top
+# that MANUFACTURER isn't an operating system.  Otherwise, code above
+# will signal an error saying that MANUFACTURER isn't an operating
+# system, and we'll never get to this point.
+
+case $basic_machine in
+	score-*)
+		os=-elf
+		;;
+	spu-*)
+		os=-elf
+		;;
+	*-acorn)
+		os=-riscix1.2
+		;;
+	arm*-rebel)
+		os=-linux
+		;;
+	arm*-semi)
+		os=-aout
+		;;
+	c4x-* | tic4x-*)
+		os=-coff
+		;;
+	c8051-*)
+		os=-elf
+		;;
+	hexagon-*)
+		os=-elf
+		;;
+	tic54x-*)
+		os=-coff
+		;;
+	tic55x-*)
+		os=-coff
+		;;
+	tic6x-*)
+		os=-coff
+		;;
+	# This must come before the *-dec entry.
+	pdp10-*)
+		os=-tops20
+		;;
+	pdp11-*)
+		os=-none
+		;;
+	*-dec | vax-*)
+		os=-ultrix4.2
+		;;
+	m68*-apollo)
+		os=-domain
+		;;
+	i386-sun)
+		os=-sunos4.0.2
+		;;
+	m68000-sun)
+		os=-sunos3
+		;;
+	m68*-cisco)
+		os=-aout
+		;;
+	mep-*)
+		os=-elf
+		;;
+	mips*-cisco)
+		os=-elf
+		;;
+	mips*-*)
+		os=-elf
+		;;
+	or32-*)
+		os=-coff
+		;;
+	*-tti)	# must be before sparc entry or we get the wrong os.
+		os=-sysv3
+		;;
+	sparc-* | *-sun)
+		os=-sunos4.1.1
+		;;
+	*-be)
+		os=-beos
+		;;
+	*-haiku)
+		os=-haiku
+		;;
+	*-ibm)
+		os=-aix
+		;;
+	*-knuth)
+		os=-mmixware
+		;;
+	*-wec)
+		os=-proelf
+		;;
+	*-winbond)
+		os=-proelf
+		;;
+	*-oki)
+		os=-proelf
+		;;
+	*-hp)
+		os=-hpux
+		;;
+	*-hitachi)
+		os=-hiux
+		;;
+	i860-* | *-att | *-ncr | *-altos | *-motorola | *-convergent)
+		os=-sysv
+		;;
+	*-cbm)
+		os=-amigaos
+		;;
+	*-dg)
+		os=-dgux
+		;;
+	*-dolphin)
+		os=-sysv3
+		;;
+	m68k-ccur)
+		os=-rtu
+		;;
+	m88k-omron*)
+		os=-luna
+		;;
+	*-next )
+		os=-nextstep
+		;;
+	*-sequent)
+		os=-ptx
+		;;
+	*-crds)
+		os=-unos
+		;;
+	*-ns)
+		os=-genix
+		;;
+	i370-*)
+		os=-mvs
+		;;
+	*-next)
+		os=-nextstep3
+		;;
+	*-gould)
+		os=-sysv
+		;;
+	*-highlevel)
+		os=-bsd
+		;;
+	*-encore)
+		os=-bsd
+		;;
+	*-sgi)
+		os=-irix
+		;;
+	*-siemens)
+		os=-sysv4
+		;;
+	*-masscomp)
+		os=-rtu
+		;;
+	f30[01]-fujitsu | f700-fujitsu)
+		os=-uxpv
+		;;
+	*-rom68k)
+		os=-coff
+		;;
+	*-*bug)
+		os=-coff
+		;;
+	*-apple)
+		os=-macos
+		;;
+	*-atari*)
+		os=-mint
+		;;
+	*)
+		os=-none
+		;;
+esac
+fi
+
+# Here we handle the case where we know the os, and the CPU type, but not the
+# manufacturer.  We pick the logical manufacturer.
+vendor=unknown
+case $basic_machine in
+	*-unknown)
+		case $os in
+			-riscix*)
+				vendor=acorn
+				;;
+			-sunos*)
+				vendor=sun
+				;;
+			-cnk*|-aix*)
+				vendor=ibm
+				;;
+			-beos*)
+				vendor=be
+				;;
+			-hpux*)
+				vendor=hp
+				;;
+			-mpeix*)
+				vendor=hp
+				;;
+			-hiux*)
+				vendor=hitachi
+				;;
+			-unos*)
+				vendor=crds
+				;;
+			-dgux*)
+				vendor=dg
+				;;
+			-luna*)
+				vendor=omron
+				;;
+			-genix*)
+				vendor=ns
+				;;
+			-mvs* | -opened*)
+				vendor=ibm
+				;;
+			-os400*)
+				vendor=ibm
+				;;
+			-ptx*)
+				vendor=sequent
+				;;
+			-tpf*)
+				vendor=ibm
+				;;
+			-vxsim* | -vxworks* | -windiss*)
+				vendor=wrs
+				;;
+			-aux*)
+				vendor=apple
+				;;
+			-hms*)
+				vendor=hitachi
+				;;
+			-mpw* | -macos*)
+				vendor=apple
+				;;
+			-*mint | -mint[0-9]* | -*MiNT | -MiNT[0-9]*)
+				vendor=atari
+				;;
+			-vos*)
+				vendor=stratus
+				;;
+		esac
+		basic_machine=`echo $basic_machine | sed "s/unknown/$vendor/"`
+		;;
+esac
+
+echo $basic_machine$os
+exit
+
+# Local variables:
+# eval: (add-hook 'write-file-hooks 'time-stamp)
+# time-stamp-start: "timestamp='"
+# time-stamp-format: "%:y-%02m-%02d"
+# time-stamp-end: "'"
+# End:
diff --git a/memkind-1.10.0/jemalloc/build-aux/install-sh b/memkind-1.10.0/jemalloc/build-aux/install-sh
new file mode 100755
index 0000000..ebc6691
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/build-aux/install-sh
@@ -0,0 +1,250 @@
+#! /bin/sh
+#
+# install - install a program, script, or datafile
+# This comes from X11R5 (mit/util/scripts/install.sh).
+#
+# Copyright 1991 by the Massachusetts Institute of Technology
+#
+# Permission to use, copy, modify, distribute, and sell this software and its
+# documentation for any purpose is hereby granted without fee, provided that
+# the above copyright notice appear in all copies and that both that
+# copyright notice and this permission notice appear in supporting
+# documentation, and that the name of M.I.T. not be used in advertising or
+# publicity pertaining to distribution of the software without specific,
+# written prior permission.  M.I.T. makes no representations about the
+# suitability of this software for any purpose.  It is provided "as is"
+# without express or implied warranty.
+#
+# Calling this script install-sh is preferred over install.sh, to prevent
+# `make' implicit rules from creating a file called install from it
+# when there is no Makefile.
+#
+# This script is compatible with the BSD install script, but was written
+# from scratch.  It can only install one file at a time, a restriction
+# shared with many OS's install programs.
+
+
+# set DOITPROG to echo to test this script
+
+# Don't use :- since 4.3BSD and earlier shells don't like it.
+doit="${DOITPROG-}"
+
+
+# put in absolute paths if you don't have them in your path; or use env. vars.
+
+mvprog="${MVPROG-mv}"
+cpprog="${CPPROG-cp}"
+chmodprog="${CHMODPROG-chmod}"
+chownprog="${CHOWNPROG-chown}"
+chgrpprog="${CHGRPPROG-chgrp}"
+stripprog="${STRIPPROG-strip}"
+rmprog="${RMPROG-rm}"
+mkdirprog="${MKDIRPROG-mkdir}"
+
+transformbasename=""
+transform_arg=""
+instcmd="$mvprog"
+chmodcmd="$chmodprog 0755"
+chowncmd=""
+chgrpcmd=""
+stripcmd=""
+rmcmd="$rmprog -f"
+mvcmd="$mvprog"
+src=""
+dst=""
+dir_arg=""
+
+while [ x"$1" != x ]; do
+    case $1 in
+	-c) instcmd="$cpprog"
+	    shift
+	    continue;;
+
+	-d) dir_arg=true
+	    shift
+	    continue;;
+
+	-m) chmodcmd="$chmodprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-o) chowncmd="$chownprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-g) chgrpcmd="$chgrpprog $2"
+	    shift
+	    shift
+	    continue;;
+
+	-s) stripcmd="$stripprog"
+	    shift
+	    continue;;
+
+	-t=*) transformarg=`echo $1 | sed 's/-t=//'`
+	    shift
+	    continue;;
+
+	-b=*) transformbasename=`echo $1 | sed 's/-b=//'`
+	    shift
+	    continue;;
+
+	*)  if [ x"$src" = x ]
+	    then
+		src=$1
+	    else
+		# this colon is to work around a 386BSD /bin/sh bug
+		:
+		dst=$1
+	    fi
+	    shift
+	    continue;;
+    esac
+done
+
+if [ x"$src" = x ]
+then
+	echo "install:	no input file specified"
+	exit 1
+else
+	true
+fi
+
+if [ x"$dir_arg" != x ]; then
+	dst=$src
+	src=""
+	
+	if [ -d $dst ]; then
+		instcmd=:
+	else
+		instcmd=mkdir
+	fi
+else
+
+# Waiting for this to be detected by the "$instcmd $src $dsttmp" command
+# might cause directories to be created, which would be especially bad 
+# if $src (and thus $dsttmp) contains '*'.
+
+	if [ -f $src -o -d $src ]
+	then
+		true
+	else
+		echo "install:  $src does not exist"
+		exit 1
+	fi
+	
+	if [ x"$dst" = x ]
+	then
+		echo "install:	no destination specified"
+		exit 1
+	else
+		true
+	fi
+
+# If destination is a directory, append the input filename; if your system
+# does not like double slashes in filenames, you may need to add some logic
+
+	if [ -d $dst ]
+	then
+		dst="$dst"/`basename $src`
+	else
+		true
+	fi
+fi
+
+## this sed command emulates the dirname command
+dstdir=`echo $dst | sed -e 's,[^/]*$,,;s,/$,,;s,^$,.,'`
+
+# Make sure that the destination directory exists.
+#  this part is taken from Noah Friedman's mkinstalldirs script
+
+# Skip lots of stat calls in the usual case.
+if [ ! -d "$dstdir" ]; then
+defaultIFS='	
+'
+IFS="${IFS-${defaultIFS}}"
+
+oIFS="${IFS}"
+# Some sh's can't handle IFS=/ for some reason.
+IFS='%'
+set - `echo ${dstdir} | sed -e 's@/@%@g' -e 's@^%@/@'`
+IFS="${oIFS}"
+
+pathcomp=''
+
+while [ $# -ne 0 ] ; do
+	pathcomp="${pathcomp}${1}"
+	shift
+
+	if [ ! -d "${pathcomp}" ] ;
+        then
+		$mkdirprog "${pathcomp}"
+	else
+		true
+	fi
+
+	pathcomp="${pathcomp}/"
+done
+fi
+
+if [ x"$dir_arg" != x ]
+then
+	$doit $instcmd $dst &&
+
+	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dst; else true ; fi &&
+	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dst; else true ; fi &&
+	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dst; else true ; fi &&
+	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dst; else true ; fi
+else
+
+# If we're going to rename the final executable, determine the name now.
+
+	if [ x"$transformarg" = x ] 
+	then
+		dstfile=`basename $dst`
+	else
+		dstfile=`basename $dst $transformbasename | 
+			sed $transformarg`$transformbasename
+	fi
+
+# don't allow the sed command to completely eliminate the filename
+
+	if [ x"$dstfile" = x ] 
+	then
+		dstfile=`basename $dst`
+	else
+		true
+	fi
+
+# Make a temp file name in the proper directory.
+
+	dsttmp=$dstdir/#inst.$$#
+
+# Move or copy the file name to the temp name
+
+	$doit $instcmd $src $dsttmp &&
+
+	trap "rm -f ${dsttmp}" 0 &&
+
+# and set any options; do chmod last to preserve setuid bits
+
+# If any of these fail, we abort the whole thing.  If we want to
+# ignore errors from any of these, just make sure not to ignore
+# errors from the above "$doit $instcmd $src $dsttmp" command.
+
+	if [ x"$chowncmd" != x ]; then $doit $chowncmd $dsttmp; else true;fi &&
+	if [ x"$chgrpcmd" != x ]; then $doit $chgrpcmd $dsttmp; else true;fi &&
+	if [ x"$stripcmd" != x ]; then $doit $stripcmd $dsttmp; else true;fi &&
+	if [ x"$chmodcmd" != x ]; then $doit $chmodcmd $dsttmp; else true;fi &&
+
+# Now rename the file to the real destination.
+
+	$doit $rmcmd -f $dstdir/$dstfile &&
+	$doit $mvcmd $dsttmp $dstdir/$dstfile 
+
+fi &&
+
+
+exit 0
diff --git a/memkind-1.10.0/jemalloc/config.stamp.in b/memkind-1.10.0/jemalloc/config.stamp.in
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/config.stamp.in
diff --git a/memkind-1.10.0/jemalloc/configure.ac b/memkind-1.10.0/jemalloc/configure.ac
new file mode 100644
index 0000000..ea6b756
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/configure.ac
@@ -0,0 +1,2410 @@
+dnl Process this file with autoconf to produce a configure script.
+AC_PREREQ(2.63)
+AC_INIT([Makefile.in])
+
+AC_CONFIG_AUX_DIR([build-aux])
+
+dnl ============================================================================
+dnl Custom macro definitions.
+
+dnl JE_CONCAT_VVV(r, a, b)
+dnl
+dnl Set $r to the concatenation of $a and $b, with a space separating them iff
+dnl both $a and $b are non-empty.
+AC_DEFUN([JE_CONCAT_VVV],
+if test "x[$]{$2}" = "x" -o "x[$]{$3}" = "x" ; then
+  $1="[$]{$2}[$]{$3}"
+else
+  $1="[$]{$2} [$]{$3}"
+fi
+)
+
+dnl JE_APPEND_VS(a, b)
+dnl
+dnl Set $a to the concatenation of $a and b, with a space separating them iff
+dnl both $a and b are non-empty.
+AC_DEFUN([JE_APPEND_VS],
+  T_APPEND_V=$2
+  JE_CONCAT_VVV($1, $1, T_APPEND_V)
+)
+
+CONFIGURE_CFLAGS=
+SPECIFIED_CFLAGS="${CFLAGS}"
+dnl JE_CFLAGS_ADD(cflag)
+dnl
+dnl CFLAGS is the concatenation of CONFIGURE_CFLAGS and SPECIFIED_CFLAGS
+dnl (ignoring EXTRA_CFLAGS, which does not impact configure tests.  This macro
+dnl appends to CONFIGURE_CFLAGS and regenerates CFLAGS.
+AC_DEFUN([JE_CFLAGS_ADD],
+[
+AC_MSG_CHECKING([whether compiler supports $1])
+T_CONFIGURE_CFLAGS="${CONFIGURE_CFLAGS}"
+JE_APPEND_VS(CONFIGURE_CFLAGS, $1)
+JE_CONCAT_VVV(CFLAGS, CONFIGURE_CFLAGS, SPECIFIED_CFLAGS)
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+]], [[
+    return 0;
+]])],
+              [je_cv_cflags_added=$1]
+              AC_MSG_RESULT([yes]),
+              [je_cv_cflags_added=]
+              AC_MSG_RESULT([no])
+              [CONFIGURE_CFLAGS="${T_CONFIGURE_CFLAGS}"]
+)
+JE_CONCAT_VVV(CFLAGS, CONFIGURE_CFLAGS, SPECIFIED_CFLAGS)
+])
+
+dnl JE_CFLAGS_SAVE()
+dnl JE_CFLAGS_RESTORE()
+dnl
+dnl Save/restore CFLAGS.  Nesting is not supported.
+AC_DEFUN([JE_CFLAGS_SAVE],
+SAVED_CONFIGURE_CFLAGS="${CONFIGURE_CFLAGS}"
+)
+AC_DEFUN([JE_CFLAGS_RESTORE],
+CONFIGURE_CFLAGS="${SAVED_CONFIGURE_CFLAGS}"
+JE_CONCAT_VVV(CFLAGS, CONFIGURE_CFLAGS, SPECIFIED_CFLAGS)
+)
+
+CONFIGURE_CXXFLAGS=
+SPECIFIED_CXXFLAGS="${CXXFLAGS}"
+dnl JE_CXXFLAGS_ADD(cxxflag)
+AC_DEFUN([JE_CXXFLAGS_ADD],
+[
+AC_MSG_CHECKING([whether compiler supports $1])
+T_CONFIGURE_CXXFLAGS="${CONFIGURE_CXXFLAGS}"
+JE_APPEND_VS(CONFIGURE_CXXFLAGS, $1)
+JE_CONCAT_VVV(CXXFLAGS, CONFIGURE_CXXFLAGS, SPECIFIED_CXXFLAGS)
+AC_LANG_PUSH([C++])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+]], [[
+    return 0;
+]])],
+              [je_cv_cxxflags_added=$1]
+              AC_MSG_RESULT([yes]),
+              [je_cv_cxxflags_added=]
+              AC_MSG_RESULT([no])
+              [CONFIGURE_CXXFLAGS="${T_CONFIGURE_CXXFLAGS}"]
+)
+AC_LANG_POP([C++])
+JE_CONCAT_VVV(CXXFLAGS, CONFIGURE_CXXFLAGS, SPECIFIED_CXXFLAGS)
+])
+
+dnl JE_COMPILABLE(label, hcode, mcode, rvar)
+dnl
+dnl Use AC_LINK_IFELSE() rather than AC_COMPILE_IFELSE() so that linker errors
+dnl cause failure.
+AC_DEFUN([JE_COMPILABLE],
+[
+AC_CACHE_CHECK([whether $1 is compilable],
+               [$4],
+               [AC_LINK_IFELSE([AC_LANG_PROGRAM([$2],
+                                                [$3])],
+                               [$4=yes],
+                               [$4=no])])
+])
+
+dnl ============================================================================
+
+CONFIG=`echo ${ac_configure_args} | sed -e 's#'"'"'\([^ ]*\)'"'"'#\1#g'`
+AC_SUBST([CONFIG])
+
+dnl Library revision.
+rev=2
+AC_SUBST([rev])
+
+srcroot=$srcdir
+if test "x${srcroot}" = "x." ; then
+  srcroot=""
+else
+  srcroot="${srcroot}/"
+fi
+AC_SUBST([srcroot])
+abs_srcroot="`cd \"${srcdir}\"; pwd`/"
+AC_SUBST([abs_srcroot])
+
+objroot=""
+AC_SUBST([objroot])
+abs_objroot="`pwd`/"
+AC_SUBST([abs_objroot])
+
+dnl Munge install path variables.
+if test "x$prefix" = "xNONE" ; then
+  prefix="/usr/local"
+fi
+if test "x$exec_prefix" = "xNONE" ; then
+  exec_prefix=$prefix
+fi
+PREFIX=$prefix
+AC_SUBST([PREFIX])
+BINDIR=`eval echo $bindir`
+BINDIR=`eval echo $BINDIR`
+AC_SUBST([BINDIR])
+INCLUDEDIR=`eval echo $includedir`
+INCLUDEDIR=`eval echo $INCLUDEDIR`
+AC_SUBST([INCLUDEDIR])
+LIBDIR=`eval echo $libdir`
+LIBDIR=`eval echo $LIBDIR`
+AC_SUBST([LIBDIR])
+DATADIR=`eval echo $datadir`
+DATADIR=`eval echo $DATADIR`
+AC_SUBST([DATADIR])
+MANDIR=`eval echo $mandir`
+MANDIR=`eval echo $MANDIR`
+AC_SUBST([MANDIR])
+
+dnl Support for building documentation.
+AC_PATH_PROG([XSLTPROC], [xsltproc], [false], [$PATH])
+if test -d "/usr/share/xml/docbook/stylesheet/docbook-xsl" ; then
+  DEFAULT_XSLROOT="/usr/share/xml/docbook/stylesheet/docbook-xsl"
+elif test -d "/usr/share/sgml/docbook/xsl-stylesheets" ; then
+  DEFAULT_XSLROOT="/usr/share/sgml/docbook/xsl-stylesheets"
+else
+  dnl Documentation building will fail if this default gets used.
+  DEFAULT_XSLROOT=""
+fi
+AC_ARG_WITH([xslroot],
+  [AS_HELP_STRING([--with-xslroot=<path>], [XSL stylesheet root path])], [
+if test "x$with_xslroot" = "xno" ; then
+  XSLROOT="${DEFAULT_XSLROOT}"
+else
+  XSLROOT="${with_xslroot}"
+fi
+],
+  XSLROOT="${DEFAULT_XSLROOT}"
+)
+if test "x$XSLTPROC" = "xfalse" ; then
+  XSLROOT=""
+fi
+AC_SUBST([XSLROOT])
+
+dnl If CFLAGS isn't defined, set CFLAGS to something reasonable.  Otherwise,
+dnl just prevent autoconf from molesting CFLAGS.
+CFLAGS=$CFLAGS
+AC_PROG_CC
+
+if test "x$GCC" != "xyes" ; then
+  AC_CACHE_CHECK([whether compiler is MSVC],
+                 [je_cv_msvc],
+                 [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([],
+                                                     [
+#ifndef _MSC_VER
+  int fail[-1];
+#endif
+])],
+                               [je_cv_msvc=yes],
+                               [je_cv_msvc=no])])
+fi
+
+dnl check if a cray prgenv wrapper compiler is being used
+je_cv_cray_prgenv_wrapper=""
+if test "x${PE_ENV}" != "x" ; then
+  case "${CC}" in
+    CC|cc)
+	je_cv_cray_prgenv_wrapper="yes"
+	;;
+    *)
+       ;;
+  esac
+fi
+
+AC_CACHE_CHECK([whether compiler is cray],
+              [je_cv_cray],
+              [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([],
+                                                  [
+#ifndef _CRAYC
+  int fail[-1];
+#endif
+])],
+                            [je_cv_cray=yes],
+                            [je_cv_cray=no])])
+
+if test "x${je_cv_cray}" = "xyes" ; then
+  AC_CACHE_CHECK([whether cray compiler version is 8.4],
+                [je_cv_cray_84],
+                [AC_COMPILE_IFELSE([AC_LANG_PROGRAM([],
+                                                      [
+#if !(_RELEASE_MAJOR == 8 && _RELEASE_MINOR == 4)
+  int fail[-1];
+#endif
+])],
+                              [je_cv_cray_84=yes],
+                              [je_cv_cray_84=no])])
+fi
+
+if test "x$GCC" = "xyes" ; then
+  JE_CFLAGS_ADD([-std=gnu11])
+  if test "x$je_cv_cflags_added" = "x-std=gnu11" ; then
+    AC_DEFINE_UNQUOTED([JEMALLOC_HAS_RESTRICT])
+  else
+    JE_CFLAGS_ADD([-std=gnu99])
+    if test "x$je_cv_cflags_added" = "x-std=gnu99" ; then
+      AC_DEFINE_UNQUOTED([JEMALLOC_HAS_RESTRICT])
+    fi
+  fi
+  JE_CFLAGS_ADD([-Wall])
+  JE_CFLAGS_ADD([-Wextra])
+  JE_CFLAGS_ADD([-Wshorten-64-to-32])
+  JE_CFLAGS_ADD([-Wsign-compare])
+  JE_CFLAGS_ADD([-Wundef])
+  JE_CFLAGS_ADD([-Wno-format-zero-length])
+  JE_CFLAGS_ADD([-pipe])
+  JE_CFLAGS_ADD([-g3])
+elif test "x$je_cv_msvc" = "xyes" ; then
+  CC="$CC -nologo"
+  JE_CFLAGS_ADD([-Zi])
+  JE_CFLAGS_ADD([-MT])
+  JE_CFLAGS_ADD([-W3])
+  JE_CFLAGS_ADD([-FS])
+  JE_APPEND_VS(CPPFLAGS, -I${srcdir}/include/msvc_compat)
+fi
+if test "x$je_cv_cray" = "xyes" ; then
+  dnl cray compiler 8.4 has an inlining bug
+  if test "x$je_cv_cray_84" = "xyes" ; then
+    JE_CFLAGS_ADD([-hipa2])
+    JE_CFLAGS_ADD([-hnognu])
+  fi
+  dnl ignore unreachable code warning
+  JE_CFLAGS_ADD([-hnomessage=128])
+  dnl ignore redefinition of "malloc", "free", etc warning
+  JE_CFLAGS_ADD([-hnomessage=1357])
+fi
+AC_SUBST([CONFIGURE_CFLAGS])
+AC_SUBST([SPECIFIED_CFLAGS])
+AC_SUBST([EXTRA_CFLAGS])
+AC_PROG_CPP
+
+AC_ARG_ENABLE([cxx],
+  [AS_HELP_STRING([--disable-cxx], [Disable C++ integration])],
+if test "x$enable_cxx" = "xno" ; then
+  enable_cxx="0"
+else
+  enable_cxx="1"
+fi
+,
+enable_cxx="1"
+)
+if test "x$enable_cxx" = "x1" ; then
+  dnl Require at least c++14, which is the first version to support sized
+  dnl deallocation.  C++ support is not compiled otherwise.
+  m4_include([m4/ax_cxx_compile_stdcxx.m4])
+  AX_CXX_COMPILE_STDCXX([14], [noext], [optional])
+  if test "x${HAVE_CXX14}" = "x1" ; then
+    JE_CXXFLAGS_ADD([-Wall])
+    JE_CXXFLAGS_ADD([-Wextra])
+    JE_CXXFLAGS_ADD([-g3])
+
+    SAVED_LIBS="${LIBS}"
+    JE_APPEND_VS(LIBS, -lstdc++)
+    JE_COMPILABLE([libstdc++ linkage], [
+#include <stdlib.h>
+], [[
+	int *arr = (int *)malloc(sizeof(int) * 42);
+	if (arr == NULL)
+		return 1;
+]], [je_cv_libstdcxx])
+    if test "x${je_cv_libstdcxx}" = "xno" ; then
+      LIBS="${SAVED_LIBS}"
+    fi
+  else
+    enable_cxx="0"
+  fi
+fi
+AC_SUBST([enable_cxx])
+AC_SUBST([CONFIGURE_CXXFLAGS])
+AC_SUBST([SPECIFIED_CXXFLAGS])
+AC_SUBST([EXTRA_CXXFLAGS])
+
+AC_C_BIGENDIAN([ac_cv_big_endian=1], [ac_cv_big_endian=0])
+if test "x${ac_cv_big_endian}" = "x1" ; then
+  AC_DEFINE_UNQUOTED([JEMALLOC_BIG_ENDIAN], [ ])
+fi
+
+if test "x${je_cv_msvc}" = "xyes" -a "x${ac_cv_header_inttypes_h}" = "xno"; then
+  JE_APPEND_VS(CPPFLAGS, -I${srcdir}/include/msvc_compat/C99)
+fi
+
+if test "x${je_cv_msvc}" = "xyes" ; then
+  LG_SIZEOF_PTR=LG_SIZEOF_PTR_WIN
+  AC_MSG_RESULT([Using a predefined value for sizeof(void *): 4 for 32-bit, 8 for 64-bit])
+else
+  AC_CHECK_SIZEOF([void *])
+  if test "x${ac_cv_sizeof_void_p}" = "x8" ; then
+    LG_SIZEOF_PTR=3
+  elif test "x${ac_cv_sizeof_void_p}" = "x4" ; then
+    LG_SIZEOF_PTR=2
+  else
+    AC_MSG_ERROR([Unsupported pointer size: ${ac_cv_sizeof_void_p}])
+  fi
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_PTR], [$LG_SIZEOF_PTR])
+
+AC_CHECK_SIZEOF([int])
+if test "x${ac_cv_sizeof_int}" = "x8" ; then
+  LG_SIZEOF_INT=3
+elif test "x${ac_cv_sizeof_int}" = "x4" ; then
+  LG_SIZEOF_INT=2
+else
+  AC_MSG_ERROR([Unsupported int size: ${ac_cv_sizeof_int}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_INT], [$LG_SIZEOF_INT])
+
+AC_CHECK_SIZEOF([long])
+if test "x${ac_cv_sizeof_long}" = "x8" ; then
+  LG_SIZEOF_LONG=3
+elif test "x${ac_cv_sizeof_long}" = "x4" ; then
+  LG_SIZEOF_LONG=2
+else
+  AC_MSG_ERROR([Unsupported long size: ${ac_cv_sizeof_long}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_LONG], [$LG_SIZEOF_LONG])
+
+AC_CHECK_SIZEOF([long long])
+if test "x${ac_cv_sizeof_long_long}" = "x8" ; then
+  LG_SIZEOF_LONG_LONG=3
+elif test "x${ac_cv_sizeof_long_long}" = "x4" ; then
+  LG_SIZEOF_LONG_LONG=2
+else
+  AC_MSG_ERROR([Unsupported long long size: ${ac_cv_sizeof_long_long}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_LONG_LONG], [$LG_SIZEOF_LONG_LONG])
+
+AC_CHECK_SIZEOF([intmax_t])
+if test "x${ac_cv_sizeof_intmax_t}" = "x16" ; then
+  LG_SIZEOF_INTMAX_T=4
+elif test "x${ac_cv_sizeof_intmax_t}" = "x8" ; then
+  LG_SIZEOF_INTMAX_T=3
+elif test "x${ac_cv_sizeof_intmax_t}" = "x4" ; then
+  LG_SIZEOF_INTMAX_T=2
+else
+  AC_MSG_ERROR([Unsupported intmax_t size: ${ac_cv_sizeof_intmax_t}])
+fi
+AC_DEFINE_UNQUOTED([LG_SIZEOF_INTMAX_T], [$LG_SIZEOF_INTMAX_T])
+
+AC_CANONICAL_HOST
+dnl CPU-specific settings.
+CPU_SPINWAIT=""
+case "${host_cpu}" in
+  i686|x86_64)
+	HAVE_CPU_SPINWAIT=1
+	if test "x${je_cv_msvc}" = "xyes" ; then
+	    AC_CACHE_VAL([je_cv_pause_msvc],
+	      [JE_COMPILABLE([pause instruction MSVC], [],
+					[[_mm_pause(); return 0;]],
+					[je_cv_pause_msvc])])
+	    if test "x${je_cv_pause_msvc}" = "xyes" ; then
+		CPU_SPINWAIT='_mm_pause()'
+	    fi
+	else
+	    AC_CACHE_VAL([je_cv_pause],
+	      [JE_COMPILABLE([pause instruction], [],
+					[[__asm__ volatile("pause"); return 0;]],
+					[je_cv_pause])])
+	    if test "x${je_cv_pause}" = "xyes" ; then
+		CPU_SPINWAIT='__asm__ volatile("pause")'
+	    fi
+	fi
+	;;
+  *)
+	HAVE_CPU_SPINWAIT=0
+	;;
+esac
+AC_DEFINE_UNQUOTED([HAVE_CPU_SPINWAIT], [$HAVE_CPU_SPINWAIT])
+AC_DEFINE_UNQUOTED([CPU_SPINWAIT], [$CPU_SPINWAIT])
+
+AC_ARG_WITH([lg_vaddr],
+  [AS_HELP_STRING([--with-lg-vaddr=<lg-vaddr>], [Number of significant virtual address bits])],
+  [LG_VADDR="$with_lg_vaddr"], [LG_VADDR="detect"])
+
+case "${host_cpu}" in
+  aarch64)
+    if test "x$LG_VADDR" = "xdetect"; then
+      AC_MSG_CHECKING([number of significant virtual address bits])
+      if test "x${LG_SIZEOF_PTR}" = "x2" ; then
+        #aarch64 ILP32
+        LG_VADDR=32
+      else
+        #aarch64 LP64
+        LG_VADDR=48
+      fi
+      AC_MSG_RESULT([$LG_VADDR])
+    fi
+    ;;
+  x86_64)
+    if test "x$LG_VADDR" = "xdetect"; then
+      AC_CACHE_CHECK([number of significant virtual address bits],
+                     [je_cv_lg_vaddr],
+                     AC_RUN_IFELSE([AC_LANG_PROGRAM(
+[[
+#include <stdio.h>
+#ifdef _WIN32
+#include <limits.h>
+#include <intrin.h>
+typedef unsigned __int32 uint32_t;
+#else
+#include <stdint.h>
+#endif
+]], [[
+	uint32_t r[[4]];
+	uint32_t eax_in = 0x80000008U;
+#ifdef _WIN32
+	__cpuid((int *)r, (int)eax_in);
+#else
+	asm volatile ("cpuid"
+	    : "=a" (r[[0]]), "=b" (r[[1]]), "=c" (r[[2]]), "=d" (r[[3]])
+	    : "a" (eax_in), "c" (0)
+	);
+#endif
+	uint32_t eax_out = r[[0]];
+	uint32_t vaddr = ((eax_out & 0x0000ff00U) >> 8);
+	FILE *f = fopen("conftest.out", "w");
+	if (f == NULL) {
+		return 1;
+	}
+	if (vaddr > (sizeof(void *) << 3)) {
+		vaddr = sizeof(void *) << 3;
+	}
+	fprintf(f, "%u", vaddr);
+	fclose(f);
+	return 0;
+]])],
+                   [je_cv_lg_vaddr=`cat conftest.out`],
+                   [je_cv_lg_vaddr=error],
+                   [je_cv_lg_vaddr=57]))
+      if test "x${je_cv_lg_vaddr}" != "x" ; then
+        LG_VADDR="${je_cv_lg_vaddr}"
+      fi
+      if test "x${LG_VADDR}" != "xerror" ; then
+        AC_DEFINE_UNQUOTED([LG_VADDR], [$LG_VADDR])
+      else
+        AC_MSG_ERROR([cannot determine number of significant virtual address bits])
+      fi
+    fi
+    ;;
+  *)
+    if test "x$LG_VADDR" = "xdetect"; then
+      AC_MSG_CHECKING([number of significant virtual address bits])
+      if test "x${LG_SIZEOF_PTR}" = "x3" ; then
+        LG_VADDR=64
+      elif test "x${LG_SIZEOF_PTR}" = "x2" ; then
+        LG_VADDR=32
+      elif test "x${LG_SIZEOF_PTR}" = "xLG_SIZEOF_PTR_WIN" ; then
+        LG_VADDR="(1U << (LG_SIZEOF_PTR_WIN+3))"
+      else
+        AC_MSG_ERROR([Unsupported lg(pointer size): ${LG_SIZEOF_PTR}])
+      fi
+      AC_MSG_RESULT([$LG_VADDR])
+    fi
+    ;;
+esac
+AC_DEFINE_UNQUOTED([LG_VADDR], [$LG_VADDR])
+
+LD_PRELOAD_VAR="LD_PRELOAD"
+so="so"
+importlib="${so}"
+o="$ac_objext"
+a="a"
+exe="$ac_exeext"
+libprefix="lib"
+link_whole_archive="0"
+DSO_LDFLAGS='-shared -Wl,-soname,$(@F)'
+RPATH='-Wl,-rpath,$(1)'
+SOREV="${so}.${rev}"
+PIC_CFLAGS='-fPIC -DPIC'
+CTARGET='-o $@'
+LDTARGET='-o $@'
+TEST_LD_MODE=
+EXTRA_LDFLAGS=
+ARFLAGS='crus'
+AROUT=' $@'
+CC_MM=1
+
+if test "x$je_cv_cray_prgenv_wrapper" = "xyes" ; then
+  TEST_LD_MODE='-dynamic'
+fi
+
+if test "x${je_cv_cray}" = "xyes" ; then
+  CC_MM=
+fi
+
+AN_MAKEVAR([AR], [AC_PROG_AR])
+AN_PROGRAM([ar], [AC_PROG_AR])
+AC_DEFUN([AC_PROG_AR], [AC_CHECK_TOOL(AR, ar, :)])
+AC_PROG_AR
+
+AN_MAKEVAR([NM], [AC_PROG_NM])
+AN_PROGRAM([nm], [AC_PROG_NM])
+AC_DEFUN([AC_PROG_NM], [AC_CHECK_TOOL(NM, nm, :)])
+AC_PROG_NM
+
+AC_PROG_AWK
+
+dnl ============================================================================
+dnl jemalloc version.
+dnl
+
+AC_ARG_WITH([version],
+  [AS_HELP_STRING([--with-version=<major>.<minor>.<bugfix>-<nrev>-g<gid>],
+   [Version string])],
+  [
+    echo "${with_version}" | grep ['^[0-9]\+\.[0-9]\+\.[0-9]\+-[0-9]\+-g[0-9a-f]\+$'] 2>&1 1>/dev/null
+    if test $? -eq 0 ; then
+      echo "$with_version" > "${objroot}VERSION"
+    else
+      echo "${with_version}" | grep ['^VERSION$'] 2>&1 1>/dev/null
+      if test $? -ne 0 ; then
+        AC_MSG_ERROR([${with_version} does not match <major>.<minor>.<bugfix>-<nrev>-g<gid> or VERSION])
+      fi
+    fi
+  ], [
+    dnl Set VERSION if source directory is inside a git repository.
+    if test "x`test ! \"${srcroot}\" && cd \"${srcroot}\"; git rev-parse --is-inside-work-tree 2>/dev/null`" = "xtrue" ; then
+      dnl Pattern globs aren't powerful enough to match both single- and
+      dnl double-digit version numbers, so iterate over patterns to support up
+      dnl to version 99.99.99 without any accidental matches.
+      for pattern in ['[0-9].[0-9].[0-9]' '[0-9].[0-9].[0-9][0-9]' \
+                     '[0-9].[0-9][0-9].[0-9]' '[0-9].[0-9][0-9].[0-9][0-9]' \
+                     '[0-9][0-9].[0-9].[0-9]' '[0-9][0-9].[0-9].[0-9][0-9]' \
+                     '[0-9][0-9].[0-9][0-9].[0-9]' \
+                     '[0-9][0-9].[0-9][0-9].[0-9][0-9]']; do
+        (test ! "${srcroot}" && cd "${srcroot}"; git describe --long --abbrev=40 --match="${pattern}") > "${objroot}VERSION.tmp" 2>/dev/null
+        if test $? -eq 0 ; then
+          mv "${objroot}VERSION.tmp" "${objroot}VERSION"
+          break
+        fi
+      done
+    fi
+    rm -f "${objroot}VERSION.tmp"
+  ])
+
+if test ! -e "${objroot}VERSION" ; then
+  if test ! -e "${srcroot}VERSION" ; then
+    AC_MSG_RESULT(
+      [Missing VERSION file, and unable to generate it; creating bogus VERSION])
+    echo "0.0.0-0-g0000000000000000000000000000000000000000" > "${objroot}VERSION"
+  else
+    cp ${srcroot}VERSION ${objroot}VERSION
+  fi
+fi
+jemalloc_version=`cat "${objroot}VERSION"`
+jemalloc_version_major=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]1}'`
+jemalloc_version_minor=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]2}'`
+jemalloc_version_bugfix=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]3}'`
+jemalloc_version_nrev=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]4}'`
+jemalloc_version_gid=`echo ${jemalloc_version} | tr ".g-" " " | awk '{print [$]5}'`
+AC_SUBST([jemalloc_version])
+AC_SUBST([jemalloc_version_major])
+AC_SUBST([jemalloc_version_minor])
+AC_SUBST([jemalloc_version_bugfix])
+AC_SUBST([jemalloc_version_nrev])
+AC_SUBST([jemalloc_version_gid])
+
+dnl Platform-specific settings.  abi and RPATH can probably be determined
+dnl programmatically, but doing so is error-prone, which makes it generally
+dnl not worth the trouble.
+dnl
+dnl Define cpp macros in CPPFLAGS, rather than doing AC_DEFINE(macro), since the
+dnl definitions need to be seen before any headers are included, which is a pain
+dnl to make happen otherwise.
+default_retain="0"
+maps_coalesce="1"
+DUMP_SYMS="${NM} -a"
+SYM_PREFIX=""
+case "${host}" in
+  *-*-darwin* | *-*-ios*)
+	abi="macho"
+	RPATH=""
+	LD_PRELOAD_VAR="DYLD_INSERT_LIBRARIES"
+	so="dylib"
+	importlib="${so}"
+	force_tls="0"
+	DSO_LDFLAGS='-shared -Wl,-install_name,$(LIBDIR)/$(@F)'
+	SOREV="${rev}.${so}"
+	sbrk_deprecated="1"
+	SYM_PREFIX="_"
+	;;
+  *-*-freebsd*)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_SYSCTL_VM_OVERCOMMIT], [ ])
+	force_lazy_lock="1"
+	;;
+  *-*-dragonfly*)
+	abi="elf"
+	;;
+  *-*-openbsd*)
+	abi="elf"
+	force_tls="0"
+	;;
+  *-*-bitrig*)
+	abi="elf"
+	;;
+  *-*-linux-android)
+	dnl syscall(2) and secure_getenv(3) are exposed by _GNU_SOURCE.
+	JE_APPEND_VS(CPPFLAGS, -D_GNU_SOURCE)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS], [ ])
+	AC_DEFINE([JEMALLOC_HAS_ALLOCA_H])
+	AC_DEFINE([JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY], [ ])
+	AC_DEFINE([JEMALLOC_THREADED_INIT], [ ])
+	AC_DEFINE([JEMALLOC_C11_ATOMICS])
+	force_tls="0"
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  default_retain="1"
+	fi
+	;;
+  *-*-linux*)
+	dnl syscall(2) and secure_getenv(3) are exposed by _GNU_SOURCE.
+	JE_APPEND_VS(CPPFLAGS, -D_GNU_SOURCE)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS], [ ])
+	AC_DEFINE([JEMALLOC_HAS_ALLOCA_H])
+	AC_DEFINE([JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY], [ ])
+	AC_DEFINE([JEMALLOC_THREADED_INIT], [ ])
+	AC_DEFINE([JEMALLOC_USE_CXX_THROW], [ ])
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  default_retain="1"
+	fi
+	;;
+  *-*-kfreebsd*)
+	dnl syscall(2) and secure_getenv(3) are exposed by _GNU_SOURCE.
+	JE_APPEND_VS(CPPFLAGS, -D_GNU_SOURCE)
+	abi="elf"
+	AC_DEFINE([JEMALLOC_HAS_ALLOCA_H])
+	AC_DEFINE([JEMALLOC_SYSCTL_VM_OVERCOMMIT], [ ])
+	AC_DEFINE([JEMALLOC_THREADED_INIT], [ ])
+	AC_DEFINE([JEMALLOC_USE_CXX_THROW], [ ])
+	;;
+  *-*-netbsd*)
+	AC_MSG_CHECKING([ABI])
+        AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[#ifdef __ELF__
+/* ELF */
+#else
+#error aout
+#endif
+]])],
+                          [abi="elf"],
+                          [abi="aout"])
+	AC_MSG_RESULT([$abi])
+	;;
+  *-*-solaris2*)
+	abi="elf"
+	RPATH='-Wl,-R,$(1)'
+	dnl Solaris needs this for sigwait().
+	JE_APPEND_VS(CPPFLAGS, -D_POSIX_PTHREAD_SEMANTICS)
+	JE_APPEND_VS(LIBS, -lposix4 -lsocket -lnsl)
+	;;
+  *-ibm-aix*)
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  dnl 64bit AIX
+	  LD_PRELOAD_VAR="LDR_PRELOAD64"
+	else
+	  dnl 32bit AIX
+	  LD_PRELOAD_VAR="LDR_PRELOAD"
+	fi
+	abi="xcoff"
+	;;
+  *-*-mingw* | *-*-cygwin*)
+	abi="pecoff"
+	force_tls="0"
+	maps_coalesce="0"
+	RPATH=""
+	so="dll"
+	if test "x$je_cv_msvc" = "xyes" ; then
+	  importlib="lib"
+	  DSO_LDFLAGS="-LD"
+	  EXTRA_LDFLAGS="-link -DEBUG"
+	  CTARGET='-Fo$@'
+	  LDTARGET='-Fe$@'
+	  AR='lib'
+	  ARFLAGS='-nologo -out:'
+	  AROUT='$@'
+	  CC_MM=
+        else
+	  importlib="${so}"
+	  DSO_LDFLAGS="-shared"
+	  link_whole_archive="1"
+	fi
+	case "${host}" in
+	  *-*-cygwin*)
+	    DUMP_SYMS="dumpbin /SYMBOLS"
+	    ;;
+	  *)
+	    ;;
+	esac
+	a="lib"
+	libprefix=""
+	SOREV="${so}"
+	PIC_CFLAGS=""
+	if test "${LG_SIZEOF_PTR}" = "3"; then
+	  default_retain="1"
+	fi
+	;;
+  *)
+	AC_MSG_RESULT([Unsupported operating system: ${host}])
+	abi="elf"
+	;;
+esac
+
+JEMALLOC_USABLE_SIZE_CONST=const
+AC_CHECK_HEADERS([malloc.h], [
+  AC_MSG_CHECKING([whether malloc_usable_size definition can use const argument])
+  AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+    [#include <malloc.h>
+     #include <stddef.h>
+    size_t malloc_usable_size(const void *ptr);
+    ],
+    [])],[
+                AC_MSG_RESULT([yes])
+         ],[
+                JEMALLOC_USABLE_SIZE_CONST=
+                AC_MSG_RESULT([no])
+         ])
+])
+AC_DEFINE_UNQUOTED([JEMALLOC_USABLE_SIZE_CONST], [$JEMALLOC_USABLE_SIZE_CONST])
+AC_SUBST([abi])
+AC_SUBST([RPATH])
+AC_SUBST([LD_PRELOAD_VAR])
+AC_SUBST([so])
+AC_SUBST([importlib])
+AC_SUBST([o])
+AC_SUBST([a])
+AC_SUBST([exe])
+AC_SUBST([libprefix])
+AC_SUBST([link_whole_archive])
+AC_SUBST([DSO_LDFLAGS])
+AC_SUBST([EXTRA_LDFLAGS])
+AC_SUBST([SOREV])
+AC_SUBST([PIC_CFLAGS])
+AC_SUBST([CTARGET])
+AC_SUBST([LDTARGET])
+AC_SUBST([TEST_LD_MODE])
+AC_SUBST([MKLIB])
+AC_SUBST([ARFLAGS])
+AC_SUBST([AROUT])
+AC_SUBST([DUMP_SYMS])
+AC_SUBST([CC_MM])
+
+dnl Determine whether libm must be linked to use e.g. log(3).
+AC_SEARCH_LIBS([log], [m], , [AC_MSG_ERROR([Missing math functions])])
+if test "x$ac_cv_search_log" != "xnone required" ; then
+  LM="$ac_cv_search_log"
+else
+  LM=
+fi
+AC_SUBST(LM)
+
+JE_COMPILABLE([__attribute__ syntax],
+              [static __attribute__((unused)) void foo(void){}],
+              [],
+              [je_cv_attribute])
+if test "x${je_cv_attribute}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR], [ ])
+  if test "x${GCC}" = "xyes" -a "x${abi}" = "xelf"; then
+    JE_CFLAGS_ADD([-fvisibility=hidden])
+    JE_CXXFLAGS_ADD([-fvisibility=hidden])
+  fi
+fi
+dnl Check for tls_model attribute support (clang 3.0 still lacks support).
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([tls_model attribute], [],
+              [static __thread int
+               __attribute__((tls_model("initial-exec"), unused)) foo;
+               foo = 0;],
+              [je_cv_tls_model])
+JE_CFLAGS_RESTORE()
+dnl (Setting of JEMALLOC_TLS_MODEL is done later, after we've checked for
+dnl --disable-initial-exec-tls)
+
+dnl Check for alloc_size attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([alloc_size attribute], [#include <stdlib.h>],
+              [void *foo(size_t size) __attribute__((alloc_size(1)));],
+              [je_cv_alloc_size])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_alloc_size}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_ALLOC_SIZE], [ ])
+fi
+dnl Check for format(gnu_printf, ...) attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([format(gnu_printf, ...) attribute], [#include <stdlib.h>],
+              [void *foo(const char *format, ...) __attribute__((format(gnu_printf, 1, 2)));],
+              [je_cv_format_gnu_printf])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_format_gnu_printf}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF], [ ])
+fi
+dnl Check for format(printf, ...) attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([format(printf, ...) attribute], [#include <stdlib.h>],
+              [void *foo(const char *format, ...) __attribute__((format(printf, 1, 2)));],
+              [je_cv_format_printf])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_format_printf}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_FORMAT_PRINTF], [ ])
+fi
+
+dnl Check for format_arg(...) attribute support.
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([format(printf, ...) attribute], [#include <stdlib.h>],
+              [const char * __attribute__((__format_arg__(1))) foo(const char *format);],
+              [je_cv_format_arg])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_format_arg}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ATTR_FORMAT_ARG], [ ])
+fi
+
+dnl Support optional additions to rpath.
+AC_ARG_WITH([rpath],
+  [AS_HELP_STRING([--with-rpath=<rpath>], [Colon-separated rpath (ELF systems only)])],
+if test "x$with_rpath" = "xno" ; then
+  RPATH_EXTRA=
+else
+  RPATH_EXTRA="`echo $with_rpath | tr \":\" \" \"`"
+fi,
+  RPATH_EXTRA=
+)
+AC_SUBST([RPATH_EXTRA])
+
+dnl Disable rules that do automatic regeneration of configure output by default.
+AC_ARG_ENABLE([autogen],
+  [AS_HELP_STRING([--enable-autogen], [Automatically regenerate configure output])],
+if test "x$enable_autogen" = "xno" ; then
+  enable_autogen="0"
+else
+  enable_autogen="1"
+fi
+,
+enable_autogen="0"
+)
+AC_SUBST([enable_autogen])
+
+AC_PROG_INSTALL
+AC_PROG_RANLIB
+AC_PATH_PROG([LD], [ld], [false], [$PATH])
+AC_PATH_PROG([AUTOCONF], [autoconf], [false], [$PATH])
+
+dnl Enable documentation
+AC_ARG_ENABLE([doc],
+	      [AS_HELP_STRING([--enable-documentation], [Build documentation])],
+if test "x$enable_doc" = "xno" ; then
+  enable_doc="0"
+else
+  enable_doc="1"
+fi
+,
+enable_doc="1"
+)
+AC_SUBST([enable_doc])
+
+dnl Enable shared libs
+AC_ARG_ENABLE([shared],
+  [AS_HELP_STRING([--enable-shared], [Build shared libaries])],
+if test "x$enable_shared" = "xno" ; then
+  enable_shared="0"
+else
+  enable_shared="1"
+fi
+,
+enable_shared="1"
+)
+AC_SUBST([enable_shared])
+
+dnl Enable static libs
+AC_ARG_ENABLE([static],
+  [AS_HELP_STRING([--enable-static], [Build static libaries])],
+if test "x$enable_static" = "xno" ; then
+  enable_static="0"
+else
+  enable_static="1"
+fi
+,
+enable_static="1"
+)
+AC_SUBST([enable_static])
+
+if test "$enable_shared$enable_static" = "00" ; then
+  AC_MSG_ERROR([Please enable one of shared or static builds])
+fi
+
+dnl Perform no name mangling by default.
+AC_ARG_WITH([mangling],
+  [AS_HELP_STRING([--with-mangling=<map>], [Mangle symbols in <map>])],
+  [mangling_map="$with_mangling"], [mangling_map=""])
+
+dnl Do not prefix public APIs by default.
+AC_ARG_WITH([jemalloc_prefix],
+  [AS_HELP_STRING([--with-jemalloc-prefix=<prefix>], [Prefix to prepend to all public APIs])],
+  [JEMALLOC_PREFIX="$with_jemalloc_prefix"],
+  [if test "x$abi" != "xmacho" -a "x$abi" != "xpecoff"; then
+  JEMALLOC_PREFIX=""
+else
+  JEMALLOC_PREFIX="je_"
+fi]
+)
+if test "x$JEMALLOC_PREFIX" = "x" ; then
+  AC_DEFINE([JEMALLOC_IS_MALLOC])
+else
+  JEMALLOC_CPREFIX=`echo ${JEMALLOC_PREFIX} | tr "a-z" "A-Z"`
+  AC_DEFINE_UNQUOTED([JEMALLOC_PREFIX], ["$JEMALLOC_PREFIX"])
+  AC_DEFINE_UNQUOTED([JEMALLOC_CPREFIX], ["$JEMALLOC_CPREFIX"])
+fi
+AC_SUBST([JEMALLOC_PREFIX])
+AC_SUBST([JEMALLOC_CPREFIX])
+
+AC_ARG_WITH([export],
+  [AS_HELP_STRING([--without-export], [disable exporting jemalloc public APIs])],
+  [if test "x$with_export" = "xno"; then
+  AC_DEFINE([JEMALLOC_EXPORT],[])
+fi]
+)
+
+public_syms="aligned_alloc calloc dallocx free mallctl mallctlbymib mallctlnametomib malloc malloc_conf malloc_message malloc_stats_print malloc_usable_size mallocx smallocx_${jemalloc_version_gid} nallocx posix_memalign rallocx realloc sallocx sdallocx xallocx"
+dnl Check for additional platform-specific public API functions.
+AC_CHECK_FUNC([memalign],
+	      [AC_DEFINE([JEMALLOC_OVERRIDE_MEMALIGN], [ ])
+	       public_syms="${public_syms} memalign"])
+AC_CHECK_FUNC([valloc],
+	      [AC_DEFINE([JEMALLOC_OVERRIDE_VALLOC], [ ])
+	       public_syms="${public_syms} valloc"])
+
+dnl Check for allocator-related functions that should be wrapped.
+wrap_syms=
+if test "x${JEMALLOC_PREFIX}" = "x" ; then
+  AC_CHECK_FUNC([__libc_calloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_CALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_calloc"])
+  AC_CHECK_FUNC([__libc_free],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_FREE], [ ])
+		 wrap_syms="${wrap_syms} __libc_free"])
+  AC_CHECK_FUNC([__libc_malloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_MALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_malloc"])
+  AC_CHECK_FUNC([__libc_memalign],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_MEMALIGN], [ ])
+		 wrap_syms="${wrap_syms} __libc_memalign"])
+  AC_CHECK_FUNC([__libc_realloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_REALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_realloc"])
+  AC_CHECK_FUNC([__libc_valloc],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___LIBC_VALLOC], [ ])
+		 wrap_syms="${wrap_syms} __libc_valloc"])
+  AC_CHECK_FUNC([__posix_memalign],
+		[AC_DEFINE([JEMALLOC_OVERRIDE___POSIX_MEMALIGN], [ ])
+		 wrap_syms="${wrap_syms} __posix_memalign"])
+fi
+
+case "${host}" in
+  *-*-mingw* | *-*-cygwin*)
+    wrap_syms="${wrap_syms} tls_callback"
+    ;;
+  *)
+    ;;
+esac
+
+dnl Mangle library-private APIs.
+AC_ARG_WITH([private_namespace],
+  [AS_HELP_STRING([--with-private-namespace=<prefix>], [Prefix to prepend to all library-private APIs])],
+  [JEMALLOC_PRIVATE_NAMESPACE="${with_private_namespace}je_"],
+  [JEMALLOC_PRIVATE_NAMESPACE="je_"]
+)
+AC_DEFINE_UNQUOTED([JEMALLOC_PRIVATE_NAMESPACE], [$JEMALLOC_PRIVATE_NAMESPACE])
+private_namespace="$JEMALLOC_PRIVATE_NAMESPACE"
+AC_SUBST([private_namespace])
+
+dnl Do not add suffix to installed files by default.
+AC_ARG_WITH([install_suffix],
+  [AS_HELP_STRING([--with-install-suffix=<suffix>], [Suffix to append to all installed files])],
+  [INSTALL_SUFFIX="$with_install_suffix"],
+  [INSTALL_SUFFIX=]
+)
+install_suffix="$INSTALL_SUFFIX"
+AC_SUBST([install_suffix])
+
+dnl Specify default malloc_conf.
+AC_ARG_WITH([malloc_conf],
+  [AS_HELP_STRING([--with-malloc-conf=<malloc_conf>], [config.malloc_conf options string])],
+  [JEMALLOC_CONFIG_MALLOC_CONF="$with_malloc_conf"],
+  [JEMALLOC_CONFIG_MALLOC_CONF=""]
+)
+config_malloc_conf="$JEMALLOC_CONFIG_MALLOC_CONF"
+AC_DEFINE_UNQUOTED([JEMALLOC_CONFIG_MALLOC_CONF], ["$config_malloc_conf"])
+
+dnl Substitute @je_@ in jemalloc_protos.h.in, primarily to make generation of
+dnl jemalloc_protos_jet.h easy.
+je_="je_"
+AC_SUBST([je_])
+
+cfgoutputs_in="Makefile.in"
+cfgoutputs_in="${cfgoutputs_in} jemalloc.pc.in"
+cfgoutputs_in="${cfgoutputs_in} doc/html.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} doc/manpages.xsl.in"
+cfgoutputs_in="${cfgoutputs_in} doc/jemalloc.xml.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/jemalloc_macros.h.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/jemalloc_protos.h.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/jemalloc_typedefs.h.in"
+cfgoutputs_in="${cfgoutputs_in} include/jemalloc/internal/jemalloc_preamble.h.in"
+cfgoutputs_in="${cfgoutputs_in} test/test.sh.in"
+cfgoutputs_in="${cfgoutputs_in} test/include/test/jemalloc_test.h.in"
+
+cfgoutputs_out="Makefile"
+cfgoutputs_out="${cfgoutputs_out} jemalloc.pc"
+cfgoutputs_out="${cfgoutputs_out} doc/html.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/manpages.xsl"
+cfgoutputs_out="${cfgoutputs_out} doc/jemalloc.xml"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc_macros.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc_protos.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/jemalloc_typedefs.h"
+cfgoutputs_out="${cfgoutputs_out} include/jemalloc/internal/jemalloc_preamble.h"
+cfgoutputs_out="${cfgoutputs_out} test/test.sh"
+cfgoutputs_out="${cfgoutputs_out} test/include/test/jemalloc_test.h"
+
+cfgoutputs_tup="Makefile"
+cfgoutputs_tup="${cfgoutputs_tup} jemalloc.pc:jemalloc.pc.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/html.xsl:doc/html.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/manpages.xsl:doc/manpages.xsl.in"
+cfgoutputs_tup="${cfgoutputs_tup} doc/jemalloc.xml:doc/jemalloc.xml.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc_macros.h:include/jemalloc/jemalloc_macros.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc_protos.h:include/jemalloc/jemalloc_protos.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/jemalloc_typedefs.h:include/jemalloc/jemalloc_typedefs.h.in"
+cfgoutputs_tup="${cfgoutputs_tup} include/jemalloc/internal/jemalloc_preamble.h"
+cfgoutputs_tup="${cfgoutputs_tup} test/test.sh:test/test.sh.in"
+cfgoutputs_tup="${cfgoutputs_tup} test/include/test/jemalloc_test.h:test/include/test/jemalloc_test.h.in"
+
+cfghdrs_in="include/jemalloc/jemalloc_defs.h.in"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/jemalloc_internal_defs.h.in"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/private_symbols.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/private_namespace.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/public_namespace.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/internal/public_unnamespace.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/jemalloc_rename.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/jemalloc_mangle.sh"
+cfghdrs_in="${cfghdrs_in} include/jemalloc/jemalloc.sh"
+cfghdrs_in="${cfghdrs_in} test/include/test/jemalloc_test_defs.h.in"
+
+cfghdrs_out="include/jemalloc/jemalloc_defs.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc${install_suffix}.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/private_symbols.awk"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/private_symbols_jet.awk"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/public_symbols.txt"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/public_namespace.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/public_unnamespace.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_protos_jet.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_rename.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_mangle.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/jemalloc_mangle_jet.h"
+cfghdrs_out="${cfghdrs_out} include/jemalloc/internal/jemalloc_internal_defs.h"
+cfghdrs_out="${cfghdrs_out} test/include/test/jemalloc_test_defs.h"
+
+cfghdrs_tup="include/jemalloc/jemalloc_defs.h:include/jemalloc/jemalloc_defs.h.in"
+cfghdrs_tup="${cfghdrs_tup} include/jemalloc/internal/jemalloc_internal_defs.h:include/jemalloc/internal/jemalloc_internal_defs.h.in"
+cfghdrs_tup="${cfghdrs_tup} test/include/test/jemalloc_test_defs.h:test/include/test/jemalloc_test_defs.h.in"
+
+dnl ============================================================================
+dnl jemalloc build options.
+dnl
+
+dnl Do not compile with debugging by default.
+AC_ARG_ENABLE([debug],
+  [AS_HELP_STRING([--enable-debug],
+                  [Build debugging code])],
+[if test "x$enable_debug" = "xno" ; then
+  enable_debug="0"
+else
+  enable_debug="1"
+fi
+],
+[enable_debug="0"]
+)
+if test "x$enable_debug" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DEBUG], [ ])
+fi
+if test "x$enable_debug" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DEBUG], [ ])
+fi
+AC_SUBST([enable_debug])
+
+dnl Only optimize if not debugging.
+if test "x$enable_debug" = "x0" ; then
+  if test "x$GCC" = "xyes" ; then
+    JE_CFLAGS_ADD([-O3])
+    JE_CXXFLAGS_ADD([-O3])
+    JE_CFLAGS_ADD([-funroll-loops])
+  elif test "x$je_cv_msvc" = "xyes" ; then
+    JE_CFLAGS_ADD([-O2])
+    JE_CXXFLAGS_ADD([-O2])
+  else
+    JE_CFLAGS_ADD([-O])
+    JE_CXXFLAGS_ADD([-O])
+  fi
+fi
+
+dnl Enable statistics calculation by default.
+AC_ARG_ENABLE([stats],
+  [AS_HELP_STRING([--disable-stats],
+                  [Disable statistics calculation/reporting])],
+[if test "x$enable_stats" = "xno" ; then
+  enable_stats="0"
+else
+  enable_stats="1"
+fi
+],
+[enable_stats="1"]
+)
+if test "x$enable_stats" = "x1" ; then
+  AC_DEFINE([JEMALLOC_STATS], [ ])
+fi
+AC_SUBST([enable_stats])
+
+dnl Do not enable smallocx by default.
+AC_ARG_ENABLE([experimental_smallocx],
+  [AS_HELP_STRING([--enable-experimental-smallocx], [Enable experimental smallocx API])],
+[if test "x$enable_experimental_smallocx" = "xno" ; then
+enable_experimental_smallocx="0"
+else
+enable_experimental_smallocx="1"
+fi
+],
+[enable_experimental_smallocx="0"]
+)
+if test "x$enable_experimental_smallocx" = "x1" ; then
+  AC_DEFINE([JEMALLOC_EXPERIMENTAL_SMALLOCX_API])
+fi
+AC_SUBST([enable_experimental_smallocx])
+
+dnl Do not enable profiling by default.
+AC_ARG_ENABLE([prof],
+  [AS_HELP_STRING([--enable-prof], [Enable allocation profiling])],
+[if test "x$enable_prof" = "xno" ; then
+  enable_prof="0"
+else
+  enable_prof="1"
+fi
+],
+[enable_prof="0"]
+)
+if test "x$enable_prof" = "x1" ; then
+  backtrace_method=""
+else
+  backtrace_method="N/A"
+fi
+
+AC_ARG_ENABLE([prof-libunwind],
+  [AS_HELP_STRING([--enable-prof-libunwind], [Use libunwind for backtracing])],
+[if test "x$enable_prof_libunwind" = "xno" ; then
+  enable_prof_libunwind="0"
+else
+  enable_prof_libunwind="1"
+fi
+],
+[enable_prof_libunwind="0"]
+)
+AC_ARG_WITH([static_libunwind],
+  [AS_HELP_STRING([--with-static-libunwind=<libunwind.a>],
+  [Path to static libunwind library; use rather than dynamically linking])],
+if test "x$with_static_libunwind" = "xno" ; then
+  LUNWIND="-lunwind"
+else
+  if test ! -f "$with_static_libunwind" ; then
+    AC_MSG_ERROR([Static libunwind not found: $with_static_libunwind])
+  fi
+  LUNWIND="$with_static_libunwind"
+fi,
+  LUNWIND="-lunwind"
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libunwind" = "x1" ; then
+  AC_CHECK_HEADERS([libunwind.h], , [enable_prof_libunwind="0"])
+  if test "x$LUNWIND" = "x-lunwind" ; then
+    AC_CHECK_LIB([unwind], [unw_backtrace], [JE_APPEND_VS(LIBS, $LUNWIND)],
+                 [enable_prof_libunwind="0"])
+  else
+    JE_APPEND_VS(LIBS, $LUNWIND)
+  fi
+  if test "x${enable_prof_libunwind}" = "x1" ; then
+    backtrace_method="libunwind"
+    AC_DEFINE([JEMALLOC_PROF_LIBUNWIND], [ ])
+  fi
+fi
+
+AC_ARG_ENABLE([prof-libgcc],
+  [AS_HELP_STRING([--disable-prof-libgcc],
+  [Do not use libgcc for backtracing])],
+[if test "x$enable_prof_libgcc" = "xno" ; then
+  enable_prof_libgcc="0"
+else
+  enable_prof_libgcc="1"
+fi
+],
+[enable_prof_libgcc="1"]
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_libgcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  AC_CHECK_HEADERS([unwind.h], , [enable_prof_libgcc="0"])
+  if test "x${enable_prof_libgcc}" = "x1" ; then
+    AC_CHECK_LIB([gcc], [_Unwind_Backtrace], [JE_APPEND_VS(LIBS, -lgcc)], [enable_prof_libgcc="0"])
+  fi
+  if test "x${enable_prof_libgcc}" = "x1" ; then
+    backtrace_method="libgcc"
+    AC_DEFINE([JEMALLOC_PROF_LIBGCC], [ ])
+  fi
+else
+  enable_prof_libgcc="0"
+fi
+
+AC_ARG_ENABLE([prof-gcc],
+  [AS_HELP_STRING([--disable-prof-gcc],
+  [Do not use gcc intrinsics for backtracing])],
+[if test "x$enable_prof_gcc" = "xno" ; then
+  enable_prof_gcc="0"
+else
+  enable_prof_gcc="1"
+fi
+],
+[enable_prof_gcc="1"]
+)
+if test "x$backtrace_method" = "x" -a "x$enable_prof_gcc" = "x1" \
+     -a "x$GCC" = "xyes" ; then
+  JE_CFLAGS_ADD([-fno-omit-frame-pointer])
+  backtrace_method="gcc intrinsics"
+  AC_DEFINE([JEMALLOC_PROF_GCC], [ ])
+else
+  enable_prof_gcc="0"
+fi
+
+if test "x$backtrace_method" = "x" ; then
+  backtrace_method="none (disabling profiling)"
+  enable_prof="0"
+fi
+AC_MSG_CHECKING([configured backtracing method])
+AC_MSG_RESULT([$backtrace_method])
+if test "x$enable_prof" = "x1" ; then
+  dnl Heap profiling uses the log(3) function.
+  JE_APPEND_VS(LIBS, $LM)
+
+  AC_DEFINE([JEMALLOC_PROF], [ ])
+fi
+AC_SUBST([enable_prof])
+
+dnl Indicate whether adjacent virtual memory mappings automatically coalesce
+dnl (and fragment on demand).
+if test "x${maps_coalesce}" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MAPS_COALESCE], [ ])
+fi
+
+dnl Indicate whether to retain memory (rather than using munmap()) by default.
+if test "x$default_retain" = "x1" ; then
+  AC_DEFINE([JEMALLOC_RETAIN], [ ])
+fi
+
+dnl Enable allocation from DSS if supported by the OS.
+have_dss="1"
+dnl Check whether the BSD/SUSv1 sbrk() exists.  If not, disable DSS support.
+AC_CHECK_FUNC([sbrk], [have_sbrk="1"], [have_sbrk="0"])
+if test "x$have_sbrk" = "x1" ; then
+  if test "x$sbrk_deprecated" = "x1" ; then
+    AC_MSG_RESULT([Disabling dss allocation because sbrk is deprecated])
+    have_dss="0"
+  fi
+else
+  have_dss="0"
+fi
+
+if test "x$have_dss" = "x1" ; then
+  AC_DEFINE([JEMALLOC_DSS], [ ])
+fi
+
+dnl Support the junk/zero filling option by default.
+AC_ARG_ENABLE([fill],
+  [AS_HELP_STRING([--disable-fill], [Disable support for junk/zero filling])],
+[if test "x$enable_fill" = "xno" ; then
+  enable_fill="0"
+else
+  enable_fill="1"
+fi
+],
+[enable_fill="1"]
+)
+if test "x$enable_fill" = "x1" ; then
+  AC_DEFINE([JEMALLOC_FILL], [ ])
+fi
+AC_SUBST([enable_fill])
+
+dnl Disable utrace(2)-based tracing by default.
+AC_ARG_ENABLE([utrace],
+  [AS_HELP_STRING([--enable-utrace], [Enable utrace(2)-based tracing])],
+[if test "x$enable_utrace" = "xno" ; then
+  enable_utrace="0"
+else
+  enable_utrace="1"
+fi
+],
+[enable_utrace="0"]
+)
+JE_COMPILABLE([utrace(2)], [
+#include <sys/types.h>
+#include <sys/param.h>
+#include <sys/time.h>
+#include <sys/uio.h>
+#include <sys/ktrace.h>
+], [
+	utrace((void *)0, 0);
+], [je_cv_utrace])
+if test "x${je_cv_utrace}" = "xno" ; then
+  enable_utrace="0"
+fi
+if test "x$enable_utrace" = "x1" ; then
+  AC_DEFINE([JEMALLOC_UTRACE], [ ])
+fi
+AC_SUBST([enable_utrace])
+
+dnl Do not support the xmalloc option by default.
+AC_ARG_ENABLE([xmalloc],
+  [AS_HELP_STRING([--enable-xmalloc], [Support xmalloc option])],
+[if test "x$enable_xmalloc" = "xno" ; then
+  enable_xmalloc="0"
+else
+  enable_xmalloc="1"
+fi
+],
+[enable_xmalloc="0"]
+)
+if test "x$enable_xmalloc" = "x1" ; then
+  AC_DEFINE([JEMALLOC_XMALLOC], [ ])
+fi
+AC_SUBST([enable_xmalloc])
+
+dnl Support cache-oblivious allocation alignment by default.
+AC_ARG_ENABLE([cache-oblivious],
+  [AS_HELP_STRING([--disable-cache-oblivious],
+                  [Disable support for cache-oblivious allocation alignment])],
+[if test "x$enable_cache_oblivious" = "xno" ; then
+  enable_cache_oblivious="0"
+else
+  enable_cache_oblivious="1"
+fi
+],
+[enable_cache_oblivious="1"]
+)
+if test "x$enable_cache_oblivious" = "x1" ; then
+  AC_DEFINE([JEMALLOC_CACHE_OBLIVIOUS], [ ])
+fi
+AC_SUBST([enable_cache_oblivious])
+
+dnl Do not log by default.
+AC_ARG_ENABLE([log],
+  [AS_HELP_STRING([--enable-log], [Support debug logging])],
+[if test "x$enable_log" = "xno" ; then
+  enable_log="0"
+else
+  enable_log="1"
+fi
+],
+[enable_log="0"]
+)
+if test "x$enable_log" = "x1" ; then
+  AC_DEFINE([JEMALLOC_LOG], [ ])
+fi
+AC_SUBST([enable_log])
+
+dnl Do not use readlinkat by default
+AC_ARG_ENABLE([readlinkat],
+  [AS_HELP_STRING([--enable-readlinkat], [Use readlinkat over readlink])],
+[if test "x$enable_readlinkat" = "xno" ; then
+  enable_readlinkat="0"
+else
+  enable_readlinkat="1"
+fi
+],
+[enable_readlinkat="0"]
+)
+if test "x$enable_readlinkat" = "x1" ; then
+  AC_DEFINE([JEMALLOC_READLINKAT], [ ])
+fi
+AC_SUBST([enable_readlinkat])
+
+dnl Avoid extra safety checks by default
+AC_ARG_ENABLE([opt-safety-checks],
+  [AS_HELP_STRING([--enable-opt-safety-checks],
+  [Perform certain low-overhead checks, even in opt mode])],
+[if test "x$enable_opt_safety_checks" = "xno" ; then
+  enable_opt_safety_checks="0"
+else
+  enable_opt_safety_checks="1"
+fi
+],
+[enable_opt_safety_checks="0"]
+)
+if test "x$enable_opt_safety_checks" = "x1" ; then
+  AC_DEFINE([JEMALLOC_OPT_SAFETY_CHECKS], [ ])
+fi
+AC_SUBST([enable_opt_safety_checks])
+
+JE_COMPILABLE([a program using __builtin_unreachable], [
+void foo (void) {
+  __builtin_unreachable();
+}
+], [
+	{
+		foo();
+	}
+], [je_cv_gcc_builtin_unreachable])
+if test "x${je_cv_gcc_builtin_unreachable}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_INTERNAL_UNREACHABLE], [__builtin_unreachable])
+else
+  AC_DEFINE([JEMALLOC_INTERNAL_UNREACHABLE], [abort])
+fi
+
+dnl ============================================================================
+dnl Check for  __builtin_ffsl(), then ffsl(3), and fail if neither are found.
+dnl One of those two functions should (theoretically) exist on all platforms
+dnl that jemalloc currently has a chance of functioning on without modification.
+dnl We additionally assume ffs[ll]() or __builtin_ffs[ll]() are defined if
+dnl ffsl() or __builtin_ffsl() are defined, respectively.
+JE_COMPILABLE([a program using __builtin_ffsl], [
+#include <stdio.h>
+#include <strings.h>
+#include <string.h>
+], [
+	{
+		int rv = __builtin_ffsl(0x08);
+		printf("%d\n", rv);
+	}
+], [je_cv_gcc_builtin_ffsl])
+if test "x${je_cv_gcc_builtin_ffsl}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_INTERNAL_FFSLL], [__builtin_ffsll])
+  AC_DEFINE([JEMALLOC_INTERNAL_FFSL], [__builtin_ffsl])
+  AC_DEFINE([JEMALLOC_INTERNAL_FFS], [__builtin_ffs])
+else
+  JE_COMPILABLE([a program using ffsl], [
+  #include <stdio.h>
+  #include <strings.h>
+  #include <string.h>
+  ], [
+	{
+		int rv = ffsl(0x08);
+		printf("%d\n", rv);
+	}
+  ], [je_cv_function_ffsl])
+  if test "x${je_cv_function_ffsl}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_INTERNAL_FFSLL], [ffsll])
+    AC_DEFINE([JEMALLOC_INTERNAL_FFSL], [ffsl])
+    AC_DEFINE([JEMALLOC_INTERNAL_FFS], [ffs])
+  else
+    AC_MSG_ERROR([Cannot build without ffsl(3) or __builtin_ffsl()])
+  fi
+fi
+
+JE_COMPILABLE([a program using __builtin_popcountl], [
+#include <stdio.h>
+#include <strings.h>
+#include <string.h>
+], [
+	{
+		int rv = __builtin_popcountl(0x08);
+		printf("%d\n", rv);
+	}
+], [je_cv_gcc_builtin_popcountl])
+if test "x${je_cv_gcc_builtin_popcountl}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_INTERNAL_POPCOUNT], [__builtin_popcount])
+  AC_DEFINE([JEMALLOC_INTERNAL_POPCOUNTL], [__builtin_popcountl])
+fi
+
+AC_ARG_WITH([lg_quantum],
+  [AS_HELP_STRING([--with-lg-quantum=<lg-quantum>],
+   [Base 2 log of minimum allocation alignment])],
+  [LG_QUANTA="$with_lg_quantum"],
+  [LG_QUANTA="3 4"])
+if test "x$with_lg_quantum" != "x" ; then
+  AC_DEFINE_UNQUOTED([LG_QUANTUM], [$with_lg_quantum])
+fi
+
+AC_ARG_WITH([lg_page],
+  [AS_HELP_STRING([--with-lg-page=<lg-page>], [Base 2 log of system page size])],
+  [LG_PAGE="$with_lg_page"], [LG_PAGE="detect"])
+if test "x$LG_PAGE" = "xdetect"; then
+  AC_CACHE_CHECK([LG_PAGE],
+               [je_cv_lg_page],
+               AC_RUN_IFELSE([AC_LANG_PROGRAM(
+[[
+#include <strings.h>
+#ifdef _WIN32
+#include <windows.h>
+#else
+#include <unistd.h>
+#endif
+#include <stdio.h>
+]],
+[[
+    int result;
+    FILE *f;
+
+#ifdef _WIN32
+    SYSTEM_INFO si;
+    GetSystemInfo(&si);
+    result = si.dwPageSize;
+#else
+    result = sysconf(_SC_PAGESIZE);
+#endif
+    if (result == -1) {
+	return 1;
+    }
+    result = JEMALLOC_INTERNAL_FFSL(result) - 1;
+
+    f = fopen("conftest.out", "w");
+    if (f == NULL) {
+	return 1;
+    }
+    fprintf(f, "%d", result);
+    fclose(f);
+
+    return 0;
+]])],
+                             [je_cv_lg_page=`cat conftest.out`],
+                             [je_cv_lg_page=undefined],
+                             [je_cv_lg_page=12]))
+fi
+if test "x${je_cv_lg_page}" != "x" ; then
+  LG_PAGE="${je_cv_lg_page}"
+fi
+if test "x${LG_PAGE}" != "xundefined" ; then
+   AC_DEFINE_UNQUOTED([LG_PAGE], [$LG_PAGE])
+else
+   AC_MSG_ERROR([cannot determine value for LG_PAGE])
+fi
+
+AC_ARG_WITH([lg_hugepage],
+  [AS_HELP_STRING([--with-lg-hugepage=<lg-hugepage>],
+   [Base 2 log of system huge page size])],
+  [je_cv_lg_hugepage="${with_lg_hugepage}"],
+  [je_cv_lg_hugepage=""])
+if test "x${je_cv_lg_hugepage}" = "x" ; then
+  dnl Look in /proc/meminfo (Linux-specific) for information on the default huge
+  dnl page size, if any.  The relevant line looks like:
+  dnl
+  dnl   Hugepagesize:       2048 kB
+  if test -e "/proc/meminfo" ; then
+    hpsk=[`cat /proc/meminfo 2>/dev/null | \
+          grep -e '^Hugepagesize:[[:space:]]\+[0-9]\+[[:space:]]kB$' | \
+          awk '{print $2}'`]
+    if test "x${hpsk}" != "x" ; then
+      je_cv_lg_hugepage=10
+      while test "${hpsk}" -gt 1 ; do
+        hpsk="$((hpsk / 2))"
+        je_cv_lg_hugepage="$((je_cv_lg_hugepage + 1))"
+      done
+    fi
+  fi
+
+  dnl Set default if unable to automatically configure.
+  if test "x${je_cv_lg_hugepage}" = "x" ; then
+    je_cv_lg_hugepage=21
+  fi
+fi
+if test "x${LG_PAGE}" != "xundefined" -a \
+        "${je_cv_lg_hugepage}" -lt "${LG_PAGE}" ; then
+  AC_MSG_ERROR([Huge page size (2^${je_cv_lg_hugepage}) must be at least page size (2^${LG_PAGE})])
+fi
+AC_DEFINE_UNQUOTED([LG_HUGEPAGE], [${je_cv_lg_hugepage}])
+
+dnl ============================================================================
+dnl Enable libdl by default.
+AC_ARG_ENABLE([libdl],
+  [AS_HELP_STRING([--disable-libdl],
+  [Do not use libdl])],
+[if test "x$enable_libdl" = "xno" ; then
+  enable_libdl="0"
+else
+  enable_libdl="1"
+fi
+],
+[enable_libdl="1"]
+)
+AC_SUBST([libdl])
+
+dnl ============================================================================
+dnl Configure pthreads.
+
+if test "x$abi" != "xpecoff" ; then
+  AC_DEFINE([JEMALLOC_HAVE_PTHREAD], [ ])
+  AC_CHECK_HEADERS([pthread.h], , [AC_MSG_ERROR([pthread.h is missing])])
+  dnl Some systems may embed pthreads functionality in libc; check for libpthread
+  dnl first, but try libc too before failing.
+  AC_CHECK_LIB([pthread], [pthread_create], [JE_APPEND_VS(LIBS, -pthread)],
+               [AC_SEARCH_LIBS([pthread_create], , ,
+                               AC_MSG_ERROR([libpthread is missing]))])
+  wrap_syms="${wrap_syms} pthread_create"
+  have_pthread="1"
+
+dnl Check if we have dlsym support.
+  if test "x$enable_libdl" = "x1" ; then
+    have_dlsym="1"
+    AC_CHECK_HEADERS([dlfcn.h], , [have_dlsym="0"])
+    check_dlsym_in_libdl="0"
+    if test "x$have_dlsym" = "x1" ; then
+      AC_CHECK_FUNC([dlsym], [], [check_dlsym_in_libdl="1"])
+    fi
+    if test "x$check_dlsym_in_libdl" = "x1" ; then
+      AC_CHECK_LIB([dl], [dlsym], [LIBS="$LIBS -ldl"], [have_dlsym="0"])
+    fi
+    if test "x$have_dlsym" = "x1" ; then
+      AC_DEFINE([JEMALLOC_HAVE_DLSYM], [ ])
+    fi
+  else
+    have_dlsym="0"
+  fi
+
+  JE_COMPILABLE([pthread_atfork(3)], [
+#include <pthread.h>
+], [
+  pthread_atfork((void *)0, (void *)0, (void *)0);
+], [je_cv_pthread_atfork])
+  if test "x${je_cv_pthread_atfork}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_HAVE_PTHREAD_ATFORK], [ ])
+  fi
+  dnl Check if pthread_setname_np is available with the expected API.
+  JE_COMPILABLE([pthread_setname_np(3)], [
+#include <pthread.h>
+], [
+  pthread_setname_np(pthread_self(), "setname_test");
+], [je_cv_pthread_setname_np])
+  if test "x${je_cv_pthread_setname_np}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_HAVE_PTHREAD_SETNAME_NP], [ ])
+  fi
+fi
+
+JE_APPEND_VS(CPPFLAGS, -D_REENTRANT)
+
+dnl Check whether clock_gettime(2) is in libc or librt.
+AC_SEARCH_LIBS([clock_gettime], [rt])
+
+dnl Cray wrapper compiler often adds `-lrt` when using `-static`. Check with
+dnl `-dynamic` as well in case a user tries to dynamically link in jemalloc
+if test "x$je_cv_cray_prgenv_wrapper" = "xyes" ; then
+  if test "$ac_cv_search_clock_gettime" != "-lrt"; then
+    JE_CFLAGS_SAVE()
+
+    unset ac_cv_search_clock_gettime
+    JE_CFLAGS_ADD([-dynamic])
+    AC_SEARCH_LIBS([clock_gettime], [rt])
+
+    JE_CFLAGS_RESTORE()
+  fi
+fi
+
+dnl check for CLOCK_MONOTONIC_COARSE (Linux-specific).
+JE_COMPILABLE([clock_gettime(CLOCK_MONOTONIC_COARSE, ...)], [
+#include <time.h>
+], [
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC_COARSE, &ts);
+], [je_cv_clock_monotonic_coarse])
+if test "x${je_cv_clock_monotonic_coarse}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_CLOCK_MONOTONIC_COARSE])
+fi
+
+dnl check for CLOCK_MONOTONIC.
+JE_COMPILABLE([clock_gettime(CLOCK_MONOTONIC, ...)], [
+#include <unistd.h>
+#include <time.h>
+], [
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts);
+#if !defined(_POSIX_MONOTONIC_CLOCK) || _POSIX_MONOTONIC_CLOCK < 0
+#  error _POSIX_MONOTONIC_CLOCK missing/invalid
+#endif
+], [je_cv_clock_monotonic])
+if test "x${je_cv_clock_monotonic}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_CLOCK_MONOTONIC])
+fi
+
+dnl Check for mach_absolute_time().
+JE_COMPILABLE([mach_absolute_time()], [
+#include <mach/mach_time.h>
+], [
+	mach_absolute_time();
+], [je_cv_mach_absolute_time])
+if test "x${je_cv_mach_absolute_time}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_MACH_ABSOLUTE_TIME])
+fi
+
+dnl Use syscall(2) (if available) by default.
+AC_ARG_ENABLE([syscall],
+  [AS_HELP_STRING([--disable-syscall], [Disable use of syscall(2)])],
+[if test "x$enable_syscall" = "xno" ; then
+  enable_syscall="0"
+else
+  enable_syscall="1"
+fi
+],
+[enable_syscall="1"]
+)
+if test "x$enable_syscall" = "x1" ; then
+  dnl Check if syscall(2) is usable.  Treat warnings as errors, so that e.g. OS
+  dnl X 10.12's deprecation warning prevents use.
+  JE_CFLAGS_SAVE()
+  JE_CFLAGS_ADD([-Werror])
+  JE_COMPILABLE([syscall(2)], [
+#include <sys/syscall.h>
+#include <unistd.h>
+], [
+	syscall(SYS_write, 2, "hello", 5);
+],
+                [je_cv_syscall])
+  JE_CFLAGS_RESTORE()
+  if test "x$je_cv_syscall" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_USE_SYSCALL], [ ])
+  fi
+fi
+
+dnl Check if the GNU-specific secure_getenv function exists.
+AC_CHECK_FUNC([secure_getenv],
+              [have_secure_getenv="1"],
+              [have_secure_getenv="0"]
+             )
+if test "x$have_secure_getenv" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_SECURE_GETENV], [ ])
+fi
+
+dnl Check if the GNU-specific sched_getcpu function exists.
+AC_CHECK_FUNC([sched_getcpu],
+              [have_sched_getcpu="1"],
+              [have_sched_getcpu="0"]
+             )
+if test "x$have_sched_getcpu" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_SCHED_GETCPU], [ ])
+fi
+
+dnl Check if the GNU-specific sched_setaffinity function exists.
+AC_CHECK_FUNC([sched_setaffinity],
+              [have_sched_setaffinity="1"],
+              [have_sched_setaffinity="0"]
+             )
+if test "x$have_sched_setaffinity" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_SCHED_SETAFFINITY], [ ])
+fi
+
+dnl Check if the Solaris/BSD issetugid function exists.
+AC_CHECK_FUNC([issetugid],
+              [have_issetugid="1"],
+              [have_issetugid="0"]
+             )
+if test "x$have_issetugid" = "x1" ; then
+  AC_DEFINE([JEMALLOC_HAVE_ISSETUGID], [ ])
+fi
+
+dnl Check whether the BSD-specific _malloc_thread_cleanup() exists.  If so, use
+dnl it rather than pthreads TSD cleanup functions to support cleanup during
+dnl thread exit, in order to avoid pthreads library recursion during
+dnl bootstrapping.
+AC_CHECK_FUNC([_malloc_thread_cleanup],
+              [have__malloc_thread_cleanup="1"],
+              [have__malloc_thread_cleanup="0"]
+             )
+if test "x$have__malloc_thread_cleanup" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MALLOC_THREAD_CLEANUP], [ ])
+  wrap_syms="${wrap_syms} _malloc_thread_cleanup"
+  force_tls="1"
+fi
+
+dnl Check whether the BSD-specific _pthread_mutex_init_calloc_cb() exists.  If
+dnl so, mutex initialization causes allocation, and we need to implement this
+dnl callback function in order to prevent recursive allocation.
+AC_CHECK_FUNC([_pthread_mutex_init_calloc_cb],
+              [have__pthread_mutex_init_calloc_cb="1"],
+              [have__pthread_mutex_init_calloc_cb="0"]
+             )
+if test "x$have__pthread_mutex_init_calloc_cb" = "x1" ; then
+  AC_DEFINE([JEMALLOC_MUTEX_INIT_CB])
+  wrap_syms="${wrap_syms} _malloc_prefork _malloc_postfork"
+fi
+
+dnl Disable lazy locking by default.
+AC_ARG_ENABLE([lazy_lock],
+  [AS_HELP_STRING([--enable-lazy-lock],
+  [Enable lazy locking (only lock when multi-threaded)])],
+[if test "x$enable_lazy_lock" = "xno" ; then
+  enable_lazy_lock="0"
+else
+  enable_lazy_lock="1"
+fi
+],
+[enable_lazy_lock=""]
+)
+if test "x${enable_lazy_lock}" = "x" ; then
+  if test "x${force_lazy_lock}" = "x1" ; then
+    AC_MSG_RESULT([Forcing lazy-lock to avoid allocator/threading bootstrap issues])
+    enable_lazy_lock="1"
+  else
+    enable_lazy_lock="0"
+  fi
+fi
+if test "x${enable_lazy_lock}" = "x1" -a "x${abi}" = "xpecoff" ; then
+  AC_MSG_RESULT([Forcing no lazy-lock because thread creation monitoring is unimplemented])
+  enable_lazy_lock="0"
+fi
+if test "x$enable_lazy_lock" = "x1" ; then
+  if test "x$have_dlsym" = "x1" ; then
+    AC_DEFINE([JEMALLOC_LAZY_LOCK], [ ])
+  else
+    AC_MSG_ERROR([Missing dlsym support: lazy-lock cannot be enabled.])
+  fi
+fi
+AC_SUBST([enable_lazy_lock])
+
+dnl Automatically configure TLS.
+if test "x${force_tls}" = "x1" ; then
+  enable_tls="1"
+elif test "x${force_tls}" = "x0" ; then
+  enable_tls="0"
+else
+  enable_tls="1"
+fi
+if test "x${enable_tls}" = "x1" ; then
+AC_MSG_CHECKING([for TLS])
+AC_COMPILE_IFELSE([AC_LANG_PROGRAM(
+[[
+    __thread int x;
+]], [[
+    x = 42;
+
+    return 0;
+]])],
+              AC_MSG_RESULT([yes]),
+              AC_MSG_RESULT([no])
+              enable_tls="0")
+else
+  enable_tls="0"
+fi
+AC_SUBST([enable_tls])
+if test "x${enable_tls}" = "x1" ; then
+  AC_DEFINE_UNQUOTED([JEMALLOC_TLS], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for C11 atomics.
+
+JE_COMPILABLE([C11 atomics], [
+#include <stdint.h>
+#if (__STDC_VERSION__ >= 201112L) && !defined(__STDC_NO_ATOMICS__)
+#include <stdatomic.h>
+#else
+#error Atomics not available
+#endif
+], [
+    uint64_t *p = (uint64_t *)0;
+    uint64_t x = 1;
+    volatile atomic_uint_least64_t *a = (volatile atomic_uint_least64_t *)p;
+    uint64_t r = atomic_fetch_add(a, x) + x;
+    return r == 0;
+], [je_cv_c11_atomics])
+if test "x${je_cv_c11_atomics}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_C11_ATOMICS])
+fi
+
+dnl ============================================================================
+dnl Check for GCC-style __atomic atomics.
+
+JE_COMPILABLE([GCC __atomic atomics], [
+], [
+    int x = 0;
+    int val = 1;
+    int y = __atomic_fetch_add(&x, val, __ATOMIC_RELAXED);
+    int after_add = x;
+    return after_add == 1;
+], [je_cv_gcc_atomic_atomics])
+if test "x${je_cv_gcc_atomic_atomics}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_GCC_ATOMIC_ATOMICS])
+
+  dnl check for 8-bit atomic support
+  JE_COMPILABLE([GCC 8-bit __atomic atomics], [
+  ], [
+      unsigned char x = 0;
+      int val = 1;
+      int y = __atomic_fetch_add(&x, val, __ATOMIC_RELAXED);
+      int after_add = (int)x;
+      return after_add == 1;
+  ], [je_cv_gcc_u8_atomic_atomics])
+  if test "x${je_cv_gcc_u8_atomic_atomics}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_GCC_U8_ATOMIC_ATOMICS])
+  fi
+fi
+
+dnl ============================================================================
+dnl Check for GCC-style __sync atomics.
+
+JE_COMPILABLE([GCC __sync atomics], [
+], [
+    int x = 0;
+    int before_add = __sync_fetch_and_add(&x, 1);
+    int after_add = x;
+    return (before_add == 0) && (after_add == 1);
+], [je_cv_gcc_sync_atomics])
+if test "x${je_cv_gcc_sync_atomics}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_GCC_SYNC_ATOMICS])
+
+  dnl check for 8-bit atomic support
+  JE_COMPILABLE([GCC 8-bit __sync atomics], [
+  ], [
+      unsigned char x = 0;
+      int before_add = __sync_fetch_and_add(&x, 1);
+      int after_add = (int)x;
+      return (before_add == 0) && (after_add == 1);
+  ], [je_cv_gcc_u8_sync_atomics])
+  if test "x${je_cv_gcc_u8_sync_atomics}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_GCC_U8_SYNC_ATOMICS])
+  fi
+fi
+
+dnl ============================================================================
+dnl Check for atomic(3) operations as provided on Darwin.
+dnl We need this not for the atomic operations (which are provided above), but
+dnl rather for the OS_unfair_lock type it exposes.
+
+JE_COMPILABLE([Darwin OSAtomic*()], [
+#include <libkern/OSAtomic.h>
+#include <inttypes.h>
+], [
+	{
+		int32_t x32 = 0;
+		volatile int32_t *x32p = &x32;
+		OSAtomicAdd32(1, x32p);
+	}
+	{
+		int64_t x64 = 0;
+		volatile int64_t *x64p = &x64;
+		OSAtomicAdd64(1, x64p);
+	}
+], [je_cv_osatomic])
+if test "x${je_cv_osatomic}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_OSATOMIC], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for madvise(2).
+
+JE_COMPILABLE([madvise(2)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, 0);
+], [je_cv_madvise])
+if test "x${je_cv_madvise}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_MADVISE], [ ])
+
+  dnl Check for madvise(..., MADV_FREE).
+  JE_COMPILABLE([madvise(..., MADV_FREE)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_FREE);
+], [je_cv_madv_free])
+  if test "x${je_cv_madv_free}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
+  elif test "x${je_cv_madvise}" = "xyes" ; then
+    case "${host_cpu}" in i686|x86_64)
+        case "${host}" in *-*-linux*)
+            AC_DEFINE([JEMALLOC_PURGE_MADVISE_FREE], [ ])
+            AC_DEFINE([JEMALLOC_DEFINE_MADVISE_FREE], [ ])
+	    ;;
+        esac
+        ;;
+    esac
+  fi
+
+  dnl Check for madvise(..., MADV_DONTNEED).
+  JE_COMPILABLE([madvise(..., MADV_DONTNEED)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_DONTNEED);
+], [je_cv_madv_dontneed])
+  if test "x${je_cv_madv_dontneed}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_PURGE_MADVISE_DONTNEED], [ ])
+  fi
+
+  dnl Check for madvise(..., MADV_DO[NT]DUMP).
+  JE_COMPILABLE([madvise(..., MADV_DO[[NT]]DUMP)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_DONTDUMP);
+	madvise((void *)0, 0, MADV_DODUMP);
+], [je_cv_madv_dontdump])
+  if test "x${je_cv_madv_dontdump}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_MADVISE_DONTDUMP], [ ])
+  fi
+
+  dnl Check for madvise(..., MADV_[NO]HUGEPAGE).
+  JE_COMPILABLE([madvise(..., MADV_[[NO]]HUGEPAGE)], [
+#include <sys/mman.h>
+], [
+	madvise((void *)0, 0, MADV_HUGEPAGE);
+	madvise((void *)0, 0, MADV_NOHUGEPAGE);
+], [je_cv_thp])
+case "${host_cpu}" in
+  arm*)
+    ;;
+  *)
+  if test "x${je_cv_thp}" = "xyes" ; then
+    AC_DEFINE([JEMALLOC_HAVE_MADVISE_HUGE], [ ])
+  fi
+  ;;
+esac
+fi
+
+dnl ============================================================================
+dnl Check for __builtin_clz() and __builtin_clzl().
+
+AC_CACHE_CHECK([for __builtin_clz],
+               [je_cv_builtin_clz],
+               [AC_LINK_IFELSE([AC_LANG_PROGRAM([],
+                                                [
+                                                {
+                                                        unsigned x = 0;
+                                                        int y = __builtin_clz(x);
+                                                }
+                                                {
+                                                        unsigned long x = 0;
+                                                        int y = __builtin_clzl(x);
+                                                }
+                                                ])],
+                               [je_cv_builtin_clz=yes],
+                               [je_cv_builtin_clz=no])])
+
+if test "x${je_cv_builtin_clz}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_BUILTIN_CLZ], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for os_unfair_lock operations as provided on Darwin.
+
+JE_COMPILABLE([Darwin os_unfair_lock_*()], [
+#include <os/lock.h>
+#include <AvailabilityMacros.h>
+], [
+	#if MAC_OS_X_VERSION_MIN_REQUIRED < 101200
+	#error "os_unfair_lock is not supported"
+	#else
+	os_unfair_lock lock = OS_UNFAIR_LOCK_INIT;
+	os_unfair_lock_lock(&lock);
+	os_unfair_lock_unlock(&lock);
+	#endif
+], [je_cv_os_unfair_lock])
+if test "x${je_cv_os_unfair_lock}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_OS_UNFAIR_LOCK], [ ])
+fi
+
+dnl ============================================================================
+dnl Darwin-related configuration.
+
+AC_ARG_ENABLE([zone-allocator],
+  [AS_HELP_STRING([--disable-zone-allocator],
+                  [Disable zone allocator for Darwin])],
+[if test "x$enable_zone_allocator" = "xno" ; then
+  enable_zone_allocator="0"
+else
+  enable_zone_allocator="1"
+fi
+],
+[if test "x${abi}" = "xmacho"; then
+  enable_zone_allocator="1"
+fi
+]
+)
+AC_SUBST([enable_zone_allocator])
+
+if test "x${enable_zone_allocator}" = "x1" ; then
+  if test "x${abi}" != "xmacho"; then
+    AC_MSG_ERROR([--enable-zone-allocator is only supported on Darwin])
+  fi
+  AC_DEFINE([JEMALLOC_ZONE], [ ])
+fi
+
+dnl ============================================================================
+dnl Use initial-exec TLS by default.
+AC_ARG_ENABLE([initial-exec-tls],
+  [AS_HELP_STRING([--disable-initial-exec-tls],
+                  [Disable the initial-exec tls model])],
+[if test "x$enable_initial_exec_tls" = "xno" ; then
+  enable_initial_exec_tls="0"
+else
+  enable_initial_exec_tls="1"
+fi
+],
+[enable_initial_exec_tls="1"]
+)
+AC_SUBST([enable_initial_exec_tls])
+
+if test "x${je_cv_tls_model}" = "xyes" -a \
+       "x${enable_initial_exec_tls}" = "x1" ; then
+  AC_DEFINE([JEMALLOC_TLS_MODEL],
+            [__attribute__((tls_model("initial-exec")))])
+else
+  AC_DEFINE([JEMALLOC_TLS_MODEL], [ ])
+fi
+
+dnl ============================================================================
+dnl Enable background threads if possible.
+
+if test "x${have_pthread}" = "x1" -a "x${je_cv_os_unfair_lock}" != "xyes" ; then
+  AC_DEFINE([JEMALLOC_BACKGROUND_THREAD])
+fi
+
+dnl ============================================================================
+dnl Check for glibc malloc hooks
+
+JE_COMPILABLE([glibc malloc hook], [
+#include <stddef.h>
+
+extern void (* __free_hook)(void *ptr);
+extern void *(* __malloc_hook)(size_t size);
+extern void *(* __realloc_hook)(void *ptr, size_t size);
+], [
+  void *ptr = 0L;
+  if (__malloc_hook) ptr = __malloc_hook(1);
+  if (__realloc_hook) ptr = __realloc_hook(ptr, 2);
+  if (__free_hook && ptr) __free_hook(ptr);
+], [je_cv_glibc_malloc_hook])
+if test "x${je_cv_glibc_malloc_hook}" = "xyes" ; then
+  if test "x${JEMALLOC_PREFIX}" = "x" ; then
+    AC_DEFINE([JEMALLOC_GLIBC_MALLOC_HOOK], [ ])
+    wrap_syms="${wrap_syms} __free_hook __malloc_hook __realloc_hook"
+  fi
+fi
+
+JE_COMPILABLE([glibc memalign hook], [
+#include <stddef.h>
+
+extern void *(* __memalign_hook)(size_t alignment, size_t size);
+], [
+  void *ptr = 0L;
+  if (__memalign_hook) ptr = __memalign_hook(16, 7);
+], [je_cv_glibc_memalign_hook])
+if test "x${je_cv_glibc_memalign_hook}" = "xyes" ; then
+  if test "x${JEMALLOC_PREFIX}" = "x" ; then
+    AC_DEFINE([JEMALLOC_GLIBC_MEMALIGN_HOOK], [ ])
+    wrap_syms="${wrap_syms} __memalign_hook"
+  fi
+fi
+
+JE_COMPILABLE([pthreads adaptive mutexes], [
+#include <pthread.h>
+], [
+  pthread_mutexattr_t attr;
+  pthread_mutexattr_init(&attr);
+  pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_ADAPTIVE_NP);
+  pthread_mutexattr_destroy(&attr);
+], [je_cv_pthread_mutex_adaptive_np])
+if test "x${je_cv_pthread_mutex_adaptive_np}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_HAVE_PTHREAD_MUTEX_ADAPTIVE_NP], [ ])
+fi
+
+JE_CFLAGS_SAVE()
+JE_CFLAGS_ADD([-D_GNU_SOURCE])
+JE_CFLAGS_ADD([-Werror])
+JE_CFLAGS_ADD([-herror_on_warning])
+JE_COMPILABLE([strerror_r returns char with gnu source], [
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+], [
+  char *buffer = (char *) malloc(100);
+  char *error = strerror_r(EINVAL, buffer, 100);
+  printf("%s\n", error);
+], [je_cv_strerror_r_returns_char_with_gnu_source])
+JE_CFLAGS_RESTORE()
+if test "x${je_cv_strerror_r_returns_char_with_gnu_source}" = "xyes" ; then
+  AC_DEFINE([JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE], [ ])
+fi
+
+dnl ============================================================================
+dnl Check for typedefs, structures, and compiler characteristics.
+AC_HEADER_STDBOOL
+
+dnl ============================================================================
+dnl Define commands that generate output files.
+
+AC_CONFIG_COMMANDS([include/jemalloc/internal/public_symbols.txt], [
+  f="${objroot}include/jemalloc/internal/public_symbols.txt"
+  mkdir -p "${objroot}include/jemalloc/internal"
+  cp /dev/null "${f}"
+  for nm in `echo ${mangling_map} |tr ',' ' '` ; do
+    n=`echo ${nm} |tr ':' ' ' |awk '{print $[]1}'`
+    m=`echo ${nm} |tr ':' ' ' |awk '{print $[]2}'`
+    echo "${n}:${m}" >> "${f}"
+    dnl Remove name from public_syms so that it isn't redefined later.
+    public_syms=`for sym in ${public_syms}; do echo "${sym}"; done |grep -v "^${n}\$" |tr '\n' ' '`
+  done
+  for sym in ${public_syms} ; do
+    n="${sym}"
+    m="${JEMALLOC_PREFIX}${sym}"
+    echo "${n}:${m}" >> "${f}"
+  done
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  mangling_map="${mangling_map}"
+  public_syms="${public_syms}"
+  JEMALLOC_PREFIX="${JEMALLOC_PREFIX}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/private_symbols.awk], [
+  f="${objroot}include/jemalloc/internal/private_symbols.awk"
+  mkdir -p "${objroot}include/jemalloc/internal"
+  export_syms=`for sym in ${public_syms}; do echo "${JEMALLOC_PREFIX}${sym}"; done; for sym in ${wrap_syms}; do echo "${sym}"; done;`
+  "${srcdir}/include/jemalloc/internal/private_symbols.sh" "${SYM_PREFIX}" ${export_syms} > "${objroot}include/jemalloc/internal/private_symbols.awk"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  public_syms="${public_syms}"
+  wrap_syms="${wrap_syms}"
+  SYM_PREFIX="${SYM_PREFIX}"
+  JEMALLOC_PREFIX="${JEMALLOC_PREFIX}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/private_symbols_jet.awk], [
+  f="${objroot}include/jemalloc/internal/private_symbols_jet.awk"
+  mkdir -p "${objroot}include/jemalloc/internal"
+  export_syms=`for sym in ${public_syms}; do echo "jet_${sym}"; done; for sym in ${wrap_syms}; do echo "${sym}"; done;`
+  "${srcdir}/include/jemalloc/internal/private_symbols.sh" "${SYM_PREFIX}" ${export_syms} > "${objroot}include/jemalloc/internal/private_symbols_jet.awk"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  public_syms="${public_syms}"
+  wrap_syms="${wrap_syms}"
+  SYM_PREFIX="${SYM_PREFIX}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/public_namespace.h], [
+  mkdir -p "${objroot}include/jemalloc/internal"
+  "${srcdir}/include/jemalloc/internal/public_namespace.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" > "${objroot}include/jemalloc/internal/public_namespace.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/internal/public_unnamespace.h], [
+  mkdir -p "${objroot}include/jemalloc/internal"
+  "${srcdir}/include/jemalloc/internal/public_unnamespace.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" > "${objroot}include/jemalloc/internal/public_unnamespace.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_protos_jet.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  cat "${srcdir}/include/jemalloc/jemalloc_protos.h.in" | sed -e 's/@je_@/jet_/g' > "${objroot}include/jemalloc/jemalloc_protos_jet.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_rename.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc_rename.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" > "${objroot}include/jemalloc/jemalloc_rename.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_mangle.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc_mangle.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" je_ > "${objroot}include/jemalloc/jemalloc_mangle.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc_mangle_jet.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc_mangle.sh" "${objroot}include/jemalloc/internal/public_symbols.txt" jet_ > "${objroot}include/jemalloc/jemalloc_mangle_jet.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+])
+AC_CONFIG_COMMANDS([include/jemalloc/jemalloc.h], [
+  mkdir -p "${objroot}include/jemalloc"
+  "${srcdir}/include/jemalloc/jemalloc.sh" "${objroot}" > "${objroot}include/jemalloc/jemalloc${install_suffix}.h"
+], [
+  srcdir="${srcdir}"
+  objroot="${objroot}"
+  install_suffix="${install_suffix}"
+])
+
+dnl Process .in files.
+AC_SUBST([cfghdrs_in])
+AC_SUBST([cfghdrs_out])
+AC_CONFIG_HEADERS([$cfghdrs_tup])
+
+dnl ============================================================================
+dnl Generate outputs.
+
+AC_CONFIG_FILES([$cfgoutputs_tup config.stamp bin/jemalloc-config bin/jemalloc.sh bin/jeprof])
+AC_SUBST([cfgoutputs_in])
+AC_SUBST([cfgoutputs_out])
+AC_OUTPUT
+
+dnl ============================================================================
+dnl Print out the results of configuration.
+AC_MSG_RESULT([===============================================================================])
+AC_MSG_RESULT([jemalloc version   : ${jemalloc_version}])
+AC_MSG_RESULT([library revision   : ${rev}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([CONFIG             : ${CONFIG}])
+AC_MSG_RESULT([CC                 : ${CC}])
+AC_MSG_RESULT([CONFIGURE_CFLAGS   : ${CONFIGURE_CFLAGS}])
+AC_MSG_RESULT([SPECIFIED_CFLAGS   : ${SPECIFIED_CFLAGS}])
+AC_MSG_RESULT([EXTRA_CFLAGS       : ${EXTRA_CFLAGS}])
+AC_MSG_RESULT([CPPFLAGS           : ${CPPFLAGS}])
+AC_MSG_RESULT([CXX                : ${CXX}])
+AC_MSG_RESULT([CONFIGURE_CXXFLAGS : ${CONFIGURE_CXXFLAGS}])
+AC_MSG_RESULT([SPECIFIED_CXXFLAGS : ${SPECIFIED_CXXFLAGS}])
+AC_MSG_RESULT([EXTRA_CXXFLAGS     : ${EXTRA_CXXFLAGS}])
+AC_MSG_RESULT([LDFLAGS            : ${LDFLAGS}])
+AC_MSG_RESULT([EXTRA_LDFLAGS      : ${EXTRA_LDFLAGS}])
+AC_MSG_RESULT([DSO_LDFLAGS        : ${DSO_LDFLAGS}])
+AC_MSG_RESULT([LIBS               : ${LIBS}])
+AC_MSG_RESULT([RPATH_EXTRA        : ${RPATH_EXTRA}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([XSLTPROC           : ${XSLTPROC}])
+AC_MSG_RESULT([XSLROOT            : ${XSLROOT}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([PREFIX             : ${PREFIX}])
+AC_MSG_RESULT([BINDIR             : ${BINDIR}])
+AC_MSG_RESULT([DATADIR            : ${DATADIR}])
+AC_MSG_RESULT([INCLUDEDIR         : ${INCLUDEDIR}])
+AC_MSG_RESULT([LIBDIR             : ${LIBDIR}])
+AC_MSG_RESULT([MANDIR             : ${MANDIR}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([srcroot            : ${srcroot}])
+AC_MSG_RESULT([abs_srcroot        : ${abs_srcroot}])
+AC_MSG_RESULT([objroot            : ${objroot}])
+AC_MSG_RESULT([abs_objroot        : ${abs_objroot}])
+AC_MSG_RESULT([])
+AC_MSG_RESULT([JEMALLOC_PREFIX    : ${JEMALLOC_PREFIX}])
+AC_MSG_RESULT([JEMALLOC_PRIVATE_NAMESPACE])
+AC_MSG_RESULT([                   : ${JEMALLOC_PRIVATE_NAMESPACE}])
+AC_MSG_RESULT([install_suffix     : ${install_suffix}])
+AC_MSG_RESULT([malloc_conf        : ${config_malloc_conf}])
+AC_MSG_RESULT([documentation      : ${enable_doc}])
+AC_MSG_RESULT([shared libs        : ${enable_shared}])
+AC_MSG_RESULT([static libs        : ${enable_static}])
+AC_MSG_RESULT([autogen            : ${enable_autogen}])
+AC_MSG_RESULT([debug              : ${enable_debug}])
+AC_MSG_RESULT([stats              : ${enable_stats}])
+AC_MSG_RESULT([experimetal_smallocx : ${enable_experimental_smallocx}])
+AC_MSG_RESULT([prof               : ${enable_prof}])
+AC_MSG_RESULT([prof-libunwind     : ${enable_prof_libunwind}])
+AC_MSG_RESULT([prof-libgcc        : ${enable_prof_libgcc}])
+AC_MSG_RESULT([prof-gcc           : ${enable_prof_gcc}])
+AC_MSG_RESULT([fill               : ${enable_fill}])
+AC_MSG_RESULT([utrace             : ${enable_utrace}])
+AC_MSG_RESULT([xmalloc            : ${enable_xmalloc}])
+AC_MSG_RESULT([log                : ${enable_log}])
+AC_MSG_RESULT([lazy_lock          : ${enable_lazy_lock}])
+AC_MSG_RESULT([cache-oblivious    : ${enable_cache_oblivious}])
+AC_MSG_RESULT([cxx                : ${enable_cxx}])
+AC_MSG_RESULT([===============================================================================])
diff --git a/memkind-1.10.0/jemalloc/doc/html.xsl.in b/memkind-1.10.0/jemalloc/doc/html.xsl.in
new file mode 100644
index 0000000..ec4fa65
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/doc/html.xsl.in
@@ -0,0 +1,5 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:import href="@XSLROOT@/html/docbook.xsl"/>
+  <xsl:import href="@abs_srcroot@doc/stylesheet.xsl"/>
+  <xsl:output method="xml" encoding="utf-8"/>
+</xsl:stylesheet>
diff --git a/memkind-1.10.0/jemalloc/doc/jemalloc.xml.in b/memkind-1.10.0/jemalloc/doc/jemalloc.xml.in
new file mode 100644
index 0000000..7fecda7
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/doc/jemalloc.xml.in
@@ -0,0 +1,3513 @@
+<?xml version='1.0' encoding='UTF-8'?>
+<?xml-stylesheet type="text/xsl"
+        href="http://docbook.sourceforge.net/release/xsl/current/manpages/docbook.xsl"?>
+<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.4//EN"
+        "http://www.oasis-open.org/docbook/xml/4.4/docbookx.dtd" [
+]>
+
+<refentry>
+  <refentryinfo>
+    <title>User Manual</title>
+    <productname>jemalloc</productname>
+    <releaseinfo role="version">@jemalloc_version@</releaseinfo>
+    <authorgroup>
+      <author>
+        <firstname>Jason</firstname>
+        <surname>Evans</surname>
+        <personblurb>Author</personblurb>
+      </author>
+    </authorgroup>
+  </refentryinfo>
+  <refmeta>
+    <refentrytitle>JEMALLOC</refentrytitle>
+    <manvolnum>3</manvolnum>
+  </refmeta>
+  <refnamediv>
+    <refdescriptor>jemalloc</refdescriptor>
+    <refname>jemalloc</refname>
+    <!-- Each refname causes a man page file to be created.  Only if this were
+         the system malloc(3) implementation would these files be appropriate.
+    <refname>malloc</refname>
+    <refname>calloc</refname>
+    <refname>posix_memalign</refname>
+    <refname>aligned_alloc</refname>
+    <refname>realloc</refname>
+    <refname>free</refname>
+    <refname>mallocx</refname>
+    <refname>rallocx</refname>
+    <refname>xallocx</refname>
+    <refname>sallocx</refname>
+    <refname>dallocx</refname>
+    <refname>sdallocx</refname>
+    <refname>nallocx</refname>
+    <refname>mallctl</refname>
+    <refname>mallctlnametomib</refname>
+    <refname>mallctlbymib</refname>
+    <refname>malloc_stats_print</refname>
+    <refname>malloc_usable_size</refname>
+    -->
+    <refpurpose>general purpose memory allocation functions</refpurpose>
+  </refnamediv>
+  <refsect1 id="library">
+    <title>LIBRARY</title>
+    <para>This manual describes jemalloc @jemalloc_version@.  More information
+    can be found at the <ulink
+    url="http://jemalloc.net/">jemalloc website</ulink>.</para>
+  </refsect1>
+  <refsynopsisdiv>
+    <title>SYNOPSIS</title>
+    <funcsynopsis>
+      <funcsynopsisinfo>#include &lt;<filename class="headerfile">jemalloc/jemalloc.h</filename>&gt;</funcsynopsisinfo>
+      <refsect2>
+        <title>Standard API</title>
+        <funcprototype>
+          <funcdef>void *<function>malloc</function></funcdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>calloc</function></funcdef>
+          <paramdef>size_t <parameter>number</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>posix_memalign</function></funcdef>
+          <paramdef>void **<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>aligned_alloc</function></funcdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>realloc</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>free</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+        </funcprototype>
+      </refsect2>
+      <refsect2>
+        <title>Non-standard API</title>
+        <funcprototype>
+          <funcdef>void *<function>mallocx</function></funcdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void *<function>rallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>xallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>extra</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>sallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>dallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>sdallocx</function></funcdef>
+          <paramdef>void *<parameter>ptr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>nallocx</function></funcdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>int <parameter>flags</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctl</function></funcdef>
+          <paramdef>const char *<parameter>name</parameter></paramdef>
+          <paramdef>void *<parameter>oldp</parameter></paramdef>
+          <paramdef>size_t *<parameter>oldlenp</parameter></paramdef>
+          <paramdef>void *<parameter>newp</parameter></paramdef>
+          <paramdef>size_t <parameter>newlen</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctlnametomib</function></funcdef>
+          <paramdef>const char *<parameter>name</parameter></paramdef>
+          <paramdef>size_t *<parameter>mibp</parameter></paramdef>
+          <paramdef>size_t *<parameter>miblenp</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>int <function>mallctlbymib</function></funcdef>
+          <paramdef>const size_t *<parameter>mib</parameter></paramdef>
+          <paramdef>size_t <parameter>miblen</parameter></paramdef>
+          <paramdef>void *<parameter>oldp</parameter></paramdef>
+          <paramdef>size_t *<parameter>oldlenp</parameter></paramdef>
+          <paramdef>void *<parameter>newp</parameter></paramdef>
+          <paramdef>size_t <parameter>newlen</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>malloc_stats_print</function></funcdef>
+          <paramdef>void <parameter>(*write_cb)</parameter>
+            <funcparams>void *, const char *</funcparams>
+          </paramdef>
+          <paramdef>void *<parameter>cbopaque</parameter></paramdef>
+          <paramdef>const char *<parameter>opts</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>size_t <function>malloc_usable_size</function></funcdef>
+          <paramdef>const void *<parameter>ptr</parameter></paramdef>
+        </funcprototype>
+        <funcprototype>
+          <funcdef>void <function>(*malloc_message)</function></funcdef>
+          <paramdef>void *<parameter>cbopaque</parameter></paramdef>
+          <paramdef>const char *<parameter>s</parameter></paramdef>
+        </funcprototype>
+        <para><type>const char *</type><varname>malloc_conf</varname>;</para>
+      </refsect2>
+    </funcsynopsis>
+  </refsynopsisdiv>
+  <refsect1 id="description">
+    <title>DESCRIPTION</title>
+    <refsect2>
+      <title>Standard API</title>
+
+      <para>The <function>malloc()</function> function allocates
+      <parameter>size</parameter> bytes of uninitialized memory.  The allocated
+      space is suitably aligned (after possible pointer coercion) for storage
+      of any type of object.</para>
+
+      <para>The <function>calloc()</function> function allocates
+      space for <parameter>number</parameter> objects, each
+      <parameter>size</parameter> bytes in length.  The result is identical to
+      calling <function>malloc()</function> with an argument of
+      <parameter>number</parameter> * <parameter>size</parameter>, with the
+      exception that the allocated memory is explicitly initialized to zero
+      bytes.</para>
+
+      <para>The <function>posix_memalign()</function> function
+      allocates <parameter>size</parameter> bytes of memory such that the
+      allocation's base address is a multiple of
+      <parameter>alignment</parameter>, and returns the allocation in the value
+      pointed to by <parameter>ptr</parameter>.  The requested
+      <parameter>alignment</parameter> must be a power of 2 at least as large as
+      <code language="C">sizeof(<type>void *</type>)</code>.</para>
+
+      <para>The <function>aligned_alloc()</function> function
+      allocates <parameter>size</parameter> bytes of memory such that the
+      allocation's base address is a multiple of
+      <parameter>alignment</parameter>.  The requested
+      <parameter>alignment</parameter> must be a power of 2.  Behavior is
+      undefined if <parameter>size</parameter> is not an integral multiple of
+      <parameter>alignment</parameter>.</para>
+
+      <para>The <function>realloc()</function> function changes the
+      size of the previously allocated memory referenced by
+      <parameter>ptr</parameter> to <parameter>size</parameter> bytes.  The
+      contents of the memory are unchanged up to the lesser of the new and old
+      sizes.  If the new size is larger, the contents of the newly allocated
+      portion of the memory are undefined.  Upon success, the memory referenced
+      by <parameter>ptr</parameter> is freed and a pointer to the newly
+      allocated memory is returned.  Note that
+      <function>realloc()</function> may move the memory allocation,
+      resulting in a different return value than <parameter>ptr</parameter>.
+      If <parameter>ptr</parameter> is <constant>NULL</constant>, the
+      <function>realloc()</function> function behaves identically to
+      <function>malloc()</function> for the specified size.</para>
+
+      <para>The <function>free()</function> function causes the
+      allocated memory referenced by <parameter>ptr</parameter> to be made
+      available for future allocations.  If <parameter>ptr</parameter> is
+      <constant>NULL</constant>, no action occurs.</para>
+    </refsect2>
+    <refsect2>
+      <title>Non-standard API</title>
+      <para>The <function>mallocx()</function>,
+      <function>rallocx()</function>,
+      <function>xallocx()</function>,
+      <function>sallocx()</function>,
+      <function>dallocx()</function>,
+      <function>sdallocx()</function>, and
+      <function>nallocx()</function> functions all have a
+      <parameter>flags</parameter> argument that can be used to specify
+      options.  The functions only check the options that are contextually
+      relevant.  Use bitwise or (<code language="C">|</code>) operations to
+      specify one or more of the following:
+        <variablelist>
+          <varlistentry id="MALLOCX_LG_ALIGN">
+            <term><constant>MALLOCX_LG_ALIGN(<parameter>la</parameter>)
+            </constant></term>
+
+            <listitem><para>Align the memory allocation to start at an address
+            that is a multiple of <code language="C">(1 &lt;&lt;
+            <parameter>la</parameter>)</code>.  This macro does not validate
+            that <parameter>la</parameter> is within the valid
+            range.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_ALIGN">
+            <term><constant>MALLOCX_ALIGN(<parameter>a</parameter>)
+            </constant></term>
+
+            <listitem><para>Align the memory allocation to start at an address
+            that is a multiple of <parameter>a</parameter>, where
+            <parameter>a</parameter> is a power of two.  This macro does not
+            validate that <parameter>a</parameter> is a power of 2.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_ZERO">
+            <term><constant>MALLOCX_ZERO</constant></term>
+
+            <listitem><para>Initialize newly allocated memory to contain zero
+            bytes.  In the growing reallocation case, the real size prior to
+            reallocation defines the boundary between untouched bytes and those
+            that are initialized to contain zero bytes.  If this macro is
+            absent, newly allocated memory is uninitialized.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_TCACHE">
+            <term><constant>MALLOCX_TCACHE(<parameter>tc</parameter>)
+            </constant></term>
+
+            <listitem><para>Use the thread-specific cache (tcache) specified by
+            the identifier <parameter>tc</parameter>, which must have been
+            acquired via the <link
+            linkend="tcache.create"><mallctl>tcache.create</mallctl></link>
+            mallctl.  This macro does not validate that
+            <parameter>tc</parameter> specifies a valid
+            identifier.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOC_TCACHE_NONE">
+            <term><constant>MALLOCX_TCACHE_NONE</constant></term>
+
+            <listitem><para>Do not use a thread-specific cache (tcache).  Unless
+            <constant>MALLOCX_TCACHE(<parameter>tc</parameter>)</constant> or
+            <constant>MALLOCX_TCACHE_NONE</constant> is specified, an
+            automatically managed tcache will be used under many circumstances.
+            This macro cannot be used in the same <parameter>flags</parameter>
+            argument as
+            <constant>MALLOCX_TCACHE(<parameter>tc</parameter>)</constant>.</para></listitem>
+          </varlistentry>
+          <varlistentry id="MALLOCX_ARENA">
+            <term><constant>MALLOCX_ARENA(<parameter>a</parameter>)
+            </constant></term>
+
+            <listitem><para>Use the arena specified by the index
+            <parameter>a</parameter>.  This macro has no effect for regions that
+            were allocated via an arena other than the one specified.  This
+            macro does not validate that <parameter>a</parameter> specifies an
+            arena index in the valid range.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>mallocx()</function> function allocates at
+      least <parameter>size</parameter> bytes of memory, and returns a pointer
+      to the base address of the allocation.  Behavior is undefined if
+      <parameter>size</parameter> is <constant>0</constant>.</para>
+
+      <para>The <function>rallocx()</function> function resizes the
+      allocation at <parameter>ptr</parameter> to be at least
+      <parameter>size</parameter> bytes, and returns a pointer to the base
+      address of the resulting allocation, which may or may not have moved from
+      its original location.  Behavior is undefined if
+      <parameter>size</parameter> is <constant>0</constant>.</para>
+
+      <para>The <function>xallocx()</function> function resizes the
+      allocation at <parameter>ptr</parameter> in place to be at least
+      <parameter>size</parameter> bytes, and returns the real size of the
+      allocation.  If <parameter>extra</parameter> is non-zero, an attempt is
+      made to resize the allocation to be at least <code
+      language="C">(<parameter>size</parameter> +
+      <parameter>extra</parameter>)</code> bytes, though inability to allocate
+      the extra byte(s) will not by itself result in failure to resize.
+      Behavior is undefined if <parameter>size</parameter> is
+      <constant>0</constant>, or if <code
+      language="C">(<parameter>size</parameter> + <parameter>extra</parameter>
+      &gt; <constant>SIZE_T_MAX</constant>)</code>.</para>
+
+      <para>The <function>sallocx()</function> function returns the
+      real size of the allocation at <parameter>ptr</parameter>.</para>
+
+      <para>The <function>dallocx()</function> function causes the
+      memory referenced by <parameter>ptr</parameter> to be made available for
+      future allocations.</para>
+
+      <para>The <function>sdallocx()</function> function is an
+      extension of <function>dallocx()</function> with a
+      <parameter>size</parameter> parameter to allow the caller to pass in the
+      allocation size as an optimization.  The minimum valid input size is the
+      original requested size of the allocation, and the maximum valid input
+      size is the corresponding value returned by
+      <function>nallocx()</function> or
+      <function>sallocx()</function>.</para>
+
+      <para>The <function>nallocx()</function> function allocates no
+      memory, but it performs the same size computation as the
+      <function>mallocx()</function> function, and returns the real
+      size of the allocation that would result from the equivalent
+      <function>mallocx()</function> function call, or
+      <constant>0</constant> if the inputs exceed the maximum supported size
+      class and/or alignment.  Behavior is undefined if
+      <parameter>size</parameter> is <constant>0</constant>.</para>
+
+      <para>The <function>mallctl()</function> function provides a
+      general interface for introspecting the memory allocator, as well as
+      setting modifiable parameters and triggering actions.  The
+      period-separated <parameter>name</parameter> argument specifies a
+      location in a tree-structured namespace; see the <xref
+      linkend="mallctl_namespace" xrefstyle="template:%t"/> section for
+      documentation on the tree contents.  To read a value, pass a pointer via
+      <parameter>oldp</parameter> to adequate space to contain the value, and a
+      pointer to its length via <parameter>oldlenp</parameter>; otherwise pass
+      <constant>NULL</constant> and <constant>NULL</constant>.  Similarly, to
+      write a value, pass a pointer to the value via
+      <parameter>newp</parameter>, and its length via
+      <parameter>newlen</parameter>; otherwise pass <constant>NULL</constant>
+      and <constant>0</constant>.</para>
+
+      <para>The <function>mallctlnametomib()</function> function
+      provides a way to avoid repeated name lookups for applications that
+      repeatedly query the same portion of the namespace, by translating a name
+      to a <quote>Management Information Base</quote> (MIB) that can be passed
+      repeatedly to <function>mallctlbymib()</function>.  Upon
+      successful return from <function>mallctlnametomib()</function>,
+      <parameter>mibp</parameter> contains an array of
+      <parameter>*miblenp</parameter> integers, where
+      <parameter>*miblenp</parameter> is the lesser of the number of components
+      in <parameter>name</parameter> and the input value of
+      <parameter>*miblenp</parameter>.  Thus it is possible to pass a
+      <parameter>*miblenp</parameter> that is smaller than the number of
+      period-separated name components, which results in a partial MIB that can
+      be used as the basis for constructing a complete MIB.  For name
+      components that are integers (e.g. the 2 in
+      <link
+      linkend="arenas.bin.i.size"><mallctl>arenas.bin.2.size</mallctl></link>),
+      the corresponding MIB component will always be that integer.  Therefore,
+      it is legitimate to construct code like the following: <programlisting
+      language="C"><![CDATA[
+unsigned nbins, i;
+size_t mib[4];
+size_t len, miblen;
+
+len = sizeof(nbins);
+mallctl("arenas.nbins", &nbins, &len, NULL, 0);
+
+miblen = 4;
+mallctlnametomib("arenas.bin.0.size", mib, &miblen);
+for (i = 0; i < nbins; i++) {
+	size_t bin_size;
+
+	mib[2] = i;
+	len = sizeof(bin_size);
+	mallctlbymib(mib, miblen, (void *)&bin_size, &len, NULL, 0);
+	/* Do something with bin_size... */
+}]]></programlisting></para>
+
+      <varlistentry id="malloc_stats_print_opts">
+      </varlistentry>
+      <para>The <function>malloc_stats_print()</function> function writes
+      summary statistics via the <parameter>write_cb</parameter> callback
+      function pointer and <parameter>cbopaque</parameter> data passed to
+      <parameter>write_cb</parameter>, or <function>malloc_message()</function>
+      if <parameter>write_cb</parameter> is <constant>NULL</constant>.  The
+      statistics are presented in human-readable form unless <quote>J</quote> is
+      specified as a character within the <parameter>opts</parameter> string, in
+      which case the statistics are presented in <ulink
+      url="http://www.json.org/">JSON format</ulink>.  This function can be
+      called repeatedly.  General information that never changes during
+      execution can be omitted by specifying <quote>g</quote> as a character
+      within the <parameter>opts</parameter> string.  Note that
+      <function>malloc_stats_print()</function> uses the
+      <function>mallctl*()</function> functions internally, so inconsistent
+      statistics can be reported if multiple threads use these functions
+      simultaneously.  If <option>--enable-stats</option> is specified during
+      configuration, <quote>m</quote>, <quote>d</quote>, and <quote>a</quote>
+      can be specified to omit merged arena, destroyed merged arena, and per
+      arena statistics, respectively; <quote>b</quote> and <quote>l</quote> can
+      be specified to omit per size class statistics for bins and large objects,
+      respectively; <quote>x</quote> can be specified to omit all mutex
+      statistics; <quote>e</quote> can be used to omit extent statistics.
+      Unrecognized characters are silently ignored.  Note that thread caching
+      may prevent some statistics from being completely up to date, since extra
+      locking would be required to merge counters that track thread cache
+      operations.</para>
+
+      <para>The <function>malloc_usable_size()</function> function
+      returns the usable size of the allocation pointed to by
+      <parameter>ptr</parameter>.  The return value may be larger than the size
+      that was requested during allocation.  The
+      <function>malloc_usable_size()</function> function is not a
+      mechanism for in-place <function>realloc()</function>; rather
+      it is provided solely as a tool for introspection purposes.  Any
+      discrepancy between the requested allocation size and the size reported
+      by <function>malloc_usable_size()</function> should not be
+      depended on, since such behavior is entirely implementation-dependent.
+      </para>
+    </refsect2>
+  </refsect1>
+  <refsect1 id="tuning">
+    <title>TUNING</title>
+    <para>Once, when the first call is made to one of the memory allocation
+    routines, the allocator initializes its internals based in part on various
+    options that can be specified at compile- or run-time.</para>
+
+    <para>The string specified via <option>--with-malloc-conf</option>, the
+    string pointed to by the global variable <varname>malloc_conf</varname>, the
+    <quote>name</quote> of the file referenced by the symbolic link named
+    <filename class="symlink">/etc/malloc.conf</filename>, and the value of the
+    environment variable <envar>MALLOC_CONF</envar>, will be interpreted, in
+    that order, from left to right as options.  Note that
+    <varname>malloc_conf</varname> may be read before
+    <function>main()</function> is entered, so the declaration of
+    <varname>malloc_conf</varname> should specify an initializer that contains
+    the final value to be read by jemalloc.  <option>--with-malloc-conf</option>
+    and <varname>malloc_conf</varname> are compile-time mechanisms, whereas
+    <filename class="symlink">/etc/malloc.conf</filename> and
+    <envar>MALLOC_CONF</envar> can be safely set any time prior to program
+    invocation.</para>
+
+    <para>An options string is a comma-separated list of option:value pairs.
+    There is one key corresponding to each <link
+    linkend="opt.abort"><mallctl>opt.*</mallctl></link> mallctl (see the <xref
+    linkend="mallctl_namespace" xrefstyle="template:%t"/> section for options
+    documentation).  For example, <literal>abort:true,narenas:1</literal> sets
+    the <link linkend="opt.abort"><mallctl>opt.abort</mallctl></link> and <link
+    linkend="opt.narenas"><mallctl>opt.narenas</mallctl></link> options.  Some
+    options have boolean values (true/false), others have integer values (base
+    8, 10, or 16, depending on prefix), and yet others have raw string
+    values.</para>
+  </refsect1>
+  <refsect1 id="implementation_notes">
+    <title>IMPLEMENTATION NOTES</title>
+    <para>Traditionally, allocators have used
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> to obtain memory, which is
+    suboptimal for several reasons, including race conditions, increased
+    fragmentation, and artificial limitations on maximum usable memory.  If
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> is supported by the operating
+    system, this allocator uses both
+    <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> and
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>, in that order of preference;
+    otherwise only <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry> is used.</para>
+
+    <para>This allocator uses multiple arenas in order to reduce lock
+    contention for threaded programs on multi-processor systems.  This works
+    well with regard to threading scalability, but incurs some costs.  There is
+    a small fixed per-arena overhead, and additionally, arenas manage memory
+    completely independently of each other, which means a small fixed increase
+    in overall memory fragmentation.  These overheads are not generally an
+    issue, given the number of arenas normally used.  Note that using
+    substantially more arenas than the default is not likely to improve
+    performance, mainly due to reduced cache performance.  However, it may make
+    sense to reduce the number of arenas if an application does not make much
+    use of the allocation functions.</para>
+
+    <para>In addition to multiple arenas, this allocator supports
+    thread-specific caching, in order to make it possible to completely avoid
+    synchronization for most allocation requests.  Such caching allows very fast
+    allocation in the common case, but it increases memory usage and
+    fragmentation, since a bounded number of objects can remain allocated in
+    each thread cache.</para>
+
+    <para>Memory is conceptually broken into extents.  Extents are always
+    aligned to multiples of the page size.  This alignment makes it possible to
+    find metadata for user objects quickly.  User objects are broken into two
+    categories according to size: small and large.  Contiguous small objects
+    comprise a slab, which resides within a single extent, whereas large objects
+    each have their own extents backing them.</para>
+
+    <para>Small objects are managed in groups by slabs.  Each slab maintains
+    a bitmap to track which regions are in use.  Allocation requests that are no
+    more than half the quantum (8 or 16, depending on architecture) are rounded
+    up to the nearest power of two that is at least <code
+    language="C">sizeof(<type>double</type>)</code>.  All other object size
+    classes are multiples of the quantum, spaced such that there are four size
+    classes for each doubling in size, which limits internal fragmentation to
+    approximately 20% for all but the smallest size classes.  Small size classes
+    are smaller than four times the page size, and large size classes extend
+    from four times the page size up to the largest size class that does not
+    exceed <constant>PTRDIFF_MAX</constant>.</para>
+
+    <para>Allocations are packed tightly together, which can be an issue for
+    multi-threaded applications.  If you need to assure that allocations do not
+    suffer from cacheline sharing, round your allocation requests up to the
+    nearest multiple of the cacheline size, or specify cacheline alignment when
+    allocating.</para>
+
+    <para>The <function>realloc()</function>,
+    <function>rallocx()</function>, and
+    <function>xallocx()</function> functions may resize allocations
+    without moving them under limited circumstances.  Unlike the
+    <function>*allocx()</function> API, the standard API does not
+    officially round up the usable size of an allocation to the nearest size
+    class, so technically it is necessary to call
+    <function>realloc()</function> to grow e.g. a 9-byte allocation to
+    16 bytes, or shrink a 16-byte allocation to 9 bytes.  Growth and shrinkage
+    trivially succeeds in place as long as the pre-size and post-size both round
+    up to the same size class.  No other API guarantees are made regarding
+    in-place resizing, but the current implementation also tries to resize large
+    allocations in place, as long as the pre-size and post-size are both large.
+    For shrinkage to succeed, the extent allocator must support splitting (see
+    <link
+    linkend="arena.i.extent_hooks"><mallctl>arena.&lt;i&gt;.extent_hooks</mallctl></link>).
+    Growth only succeeds if the trailing memory is currently available, and the
+    extent allocator supports merging.</para>
+
+    <para>Assuming 4 KiB pages and a 16-byte quantum on a 64-bit system, the
+    size classes in each category are as shown in <xref linkend="size_classes"
+    xrefstyle="template:Table %n"/>.</para>
+
+    <table xml:id="size_classes" frame="all">
+      <title>Size classes</title>
+      <tgroup cols="3" colsep="1" rowsep="1">
+      <colspec colname="c1" align="left"/>
+      <colspec colname="c2" align="right"/>
+      <colspec colname="c3" align="left"/>
+      <thead>
+        <row>
+          <entry>Category</entry>
+          <entry>Spacing</entry>
+          <entry>Size</entry>
+        </row>
+      </thead>
+      <tbody>
+        <row>
+          <entry morerows="8">Small</entry>
+          <entry>lg</entry>
+          <entry>[8]</entry>
+        </row>
+        <row>
+          <entry>16</entry>
+          <entry>[16, 32, 48, 64, 80, 96, 112, 128]</entry>
+        </row>
+        <row>
+          <entry>32</entry>
+          <entry>[160, 192, 224, 256]</entry>
+        </row>
+        <row>
+          <entry>64</entry>
+          <entry>[320, 384, 448, 512]</entry>
+        </row>
+        <row>
+          <entry>128</entry>
+          <entry>[640, 768, 896, 1024]</entry>
+        </row>
+        <row>
+          <entry>256</entry>
+          <entry>[1280, 1536, 1792, 2048]</entry>
+        </row>
+        <row>
+          <entry>512</entry>
+          <entry>[2560, 3072, 3584, 4096]</entry>
+        </row>
+        <row>
+          <entry>1 KiB</entry>
+          <entry>[5 KiB, 6 KiB, 7 KiB, 8 KiB]</entry>
+        </row>
+        <row>
+          <entry>2 KiB</entry>
+          <entry>[10 KiB, 12 KiB, 14 KiB]</entry>
+        </row>
+        <row>
+          <entry morerows="15">Large</entry>
+          <entry>2 KiB</entry>
+          <entry>[16 KiB]</entry>
+        </row>
+        <row>
+          <entry>4 KiB</entry>
+          <entry>[20 KiB, 24 KiB, 28 KiB, 32 KiB]</entry>
+        </row>
+        <row>
+          <entry>8 KiB</entry>
+          <entry>[40 KiB, 48 KiB, 54 KiB, 64 KiB]</entry>
+        </row>
+        <row>
+          <entry>16 KiB</entry>
+          <entry>[80 KiB, 96 KiB, 112 KiB, 128 KiB]</entry>
+        </row>
+        <row>
+          <entry>32 KiB</entry>
+          <entry>[160 KiB, 192 KiB, 224 KiB, 256 KiB]</entry>
+        </row>
+        <row>
+          <entry>64 KiB</entry>
+          <entry>[320 KiB, 384 KiB, 448 KiB, 512 KiB]</entry>
+        </row>
+        <row>
+          <entry>128 KiB</entry>
+          <entry>[640 KiB, 768 KiB, 896 KiB, 1 MiB]</entry>
+        </row>
+        <row>
+          <entry>256 KiB</entry>
+          <entry>[1280 KiB, 1536 KiB, 1792 KiB, 2 MiB]</entry>
+        </row>
+        <row>
+          <entry>512 KiB</entry>
+          <entry>[2560 KiB, 3 MiB, 3584 KiB, 4 MiB]</entry>
+        </row>
+        <row>
+          <entry>1 MiB</entry>
+          <entry>[5 MiB, 6 MiB, 7 MiB, 8 MiB]</entry>
+        </row>
+        <row>
+          <entry>2 MiB</entry>
+          <entry>[10 MiB, 12 MiB, 14 MiB, 16 MiB]</entry>
+        </row>
+        <row>
+          <entry>4 MiB</entry>
+          <entry>[20 MiB, 24 MiB, 28 MiB, 32 MiB]</entry>
+        </row>
+        <row>
+          <entry>8 MiB</entry>
+          <entry>[40 MiB, 48 MiB, 56 MiB, 64 MiB]</entry>
+        </row>
+        <row>
+          <entry>...</entry>
+          <entry>...</entry>
+        </row>
+        <row>
+          <entry>512 PiB</entry>
+          <entry>[2560 PiB, 3 EiB, 3584 PiB, 4 EiB]</entry>
+        </row>
+        <row>
+          <entry>1 EiB</entry>
+          <entry>[5 EiB, 6 EiB, 7 EiB]</entry>
+        </row>
+      </tbody>
+      </tgroup>
+    </table>
+  </refsect1>
+  <refsect1 id="mallctl_namespace">
+    <title>MALLCTL NAMESPACE</title>
+    <para>The following names are defined in the namespace accessible via the
+    <function>mallctl*()</function> functions.  Value types are specified in
+    parentheses, their readable/writable statuses are encoded as
+    <literal>rw</literal>, <literal>r-</literal>, <literal>-w</literal>, or
+    <literal>--</literal>, and required build configuration flags follow, if
+    any.  A name element encoded as <literal>&lt;i&gt;</literal> or
+    <literal>&lt;j&gt;</literal> indicates an integer component, where the
+    integer varies from 0 to some upper value that must be determined via
+    introspection.  In the case of <mallctl>stats.arenas.&lt;i&gt;.*</mallctl>
+    and <mallctl>arena.&lt;i&gt;.{initialized,purge,decay,dss}</mallctl>,
+    <literal>&lt;i&gt;</literal> equal to
+    <constant>MALLCTL_ARENAS_ALL</constant> can be used to operate on all arenas
+    or access the summation of statistics from all arenas; similarly
+    <literal>&lt;i&gt;</literal> equal to
+    <constant>MALLCTL_ARENAS_DESTROYED</constant> can be used to access the
+    summation of statistics from all destroyed arenas.  These constants can be
+    utilized either via <function>mallctlnametomib()</function> followed by
+    <function>mallctlbymib()</function>, or via code such as the following:
+    <programlisting language="C"><![CDATA[
+#define STRINGIFY_HELPER(x) #x
+#define STRINGIFY(x) STRINGIFY_HELPER(x)
+
+mallctl("arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".decay",
+    NULL, NULL, NULL, 0);]]></programlisting>
+    Take special note of the <link
+    linkend="epoch"><mallctl>epoch</mallctl></link> mallctl, which controls
+    refreshing of cached dynamic statistics.</para>
+
+    <variablelist>
+      <varlistentry id="version">
+        <term>
+          <mallctl>version</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Return the jemalloc version string.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="epoch">
+        <term>
+          <mallctl>epoch</mallctl>
+          (<type>uint64_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>If a value is passed in, refresh the data from which
+        the <function>mallctl*()</function> functions report values,
+        and increment the epoch.  Return the current epoch.  This is useful for
+        detecting whether another thread caused a refresh.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="background_thread">
+        <term>
+          <mallctl>background_thread</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Enable/disable internal background worker threads.  When
+        set to true, background threads are created on demand (the number of
+        background threads will be no more than the number of CPUs or active
+        arenas).  Threads run periodically, and handle <link
+        linkend="arena.i.decay">purging</link> asynchronously.  When switching
+        off, background threads are terminated synchronously.  Note that after
+        <citerefentry><refentrytitle>fork</refentrytitle><manvolnum>2</manvolnum></citerefentry>
+        function, the state in the child process will be disabled regardless
+        the state in parent process. See <link
+        linkend="stats.background_thread.num_threads"><mallctl>stats.background_thread</mallctl></link>
+        for related stats.  <link
+        linkend="opt.background_thread"><mallctl>opt.background_thread</mallctl></link>
+        can be used to set the default option.  This option is only available on
+        selected pthread-based platforms.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="max_background_threads">
+        <term>
+          <mallctl>max_background_threads</mallctl>
+          (<type>size_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Maximum number of background worker threads that will
+        be created.  This value is capped at <link
+        linkend="opt.max_background_threads"><mallctl>opt.max_background_threads</mallctl></link> at
+        startup.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.cache_oblivious">
+        <term>
+          <mallctl>config.cache_oblivious</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-cache-oblivious</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.debug">
+        <term>
+          <mallctl>config.debug</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-debug</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.fill">
+        <term>
+          <mallctl>config.fill</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-fill</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.lazy_lock">
+        <term>
+          <mallctl>config.lazy_lock</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-lazy-lock</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.malloc_conf">
+        <term>
+          <mallctl>config.malloc_conf</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Embedded configure-time-specified run-time options
+        string, empty unless <option>--with-malloc-conf</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.prof">
+        <term>
+          <mallctl>config.prof</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-prof</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.prof_libgcc">
+        <term>
+          <mallctl>config.prof_libgcc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--disable-prof-libgcc</option> was not
+        specified during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.prof_libunwind">
+        <term>
+          <mallctl>config.prof_libunwind</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-prof-libunwind</option> was specified
+        during build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.stats">
+        <term>
+          <mallctl>config.stats</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-stats</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+
+      <varlistentry id="config.utrace">
+        <term>
+          <mallctl>config.utrace</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-utrace</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="config.xmalloc">
+        <term>
+          <mallctl>config.xmalloc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para><option>--enable-xmalloc</option> was specified during
+        build configuration.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.abort">
+        <term>
+          <mallctl>opt.abort</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Abort-on-warning enabled/disabled.  If true, most
+        warnings are fatal.  Note that runtime option warnings are not included
+        (see <link
+        linkend="opt.abort_conf"><mallctl>opt.abort_conf</mallctl></link> for
+        that). The process will call
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> in these cases.  This option is
+        disabled by default unless <option>--enable-debug</option> is
+        specified during configuration, in which case it is enabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.confirm_conf">
+        <term>
+          <mallctl>opt.confirm_conf</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+	<listitem><para>Confirm-runtime-options-when-program-starts
+	enabled/disabled.  If true, the string specified via
+	<option>--with-malloc-conf</option>, the string pointed to by the
+	global variable <varname>malloc_conf</varname>, the <quote>name</quote>
+	of the file referenced by the symbolic link named
+	<filename class="symlink">/etc/malloc.conf</filename>, and the value of
+	the environment variable <envar>MALLOC_CONF</envar>, will be printed in
+	order.  Then, each option being set will be individually printed.  This
+	option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.abort_conf">
+        <term>
+          <mallctl>opt.abort_conf</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Abort-on-invalid-configuration enabled/disabled.  If
+        true, invalid runtime options are fatal.  The process will call
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> in these cases.  This option is
+        disabled by default unless <option>--enable-debug</option> is
+        specified during configuration, in which case it is enabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.metadata_thp">
+        <term>
+          <mallctl>opt.metadata_thp</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Controls whether to allow jemalloc to use transparent
+        huge page (THP) for internal metadata (see <link
+        linkend="stats.metadata">stats.metadata</link>).  <quote>always</quote>
+        allows such usage.  <quote>auto</quote> uses no THP initially, but may
+        begin to do so when metadata usage reaches certain level.  The default
+        is <quote>disabled</quote>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.retain">
+        <term>
+          <mallctl>opt.retain</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>If true, retain unused virtual memory for later reuse
+        rather than discarding it by calling
+        <citerefentry><refentrytitle>munmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> or equivalent (see <link
+        linkend="stats.retained">stats.retained</link> for related details).
+        It also makes jemalloc use <citerefentry>
+        <refentrytitle>mmap</refentrytitle><manvolnum>2</manvolnum>
+        </citerefentry> or equivalent in a more greedy way, mapping larger
+        chunks in one go.  This option is disabled by default unless discarding
+        virtual memory is known to trigger platform-specific performance
+        problems, namely 1) for [64-bit] Linux, which has a quirk in its virtual
+        memory allocation algorithm that causes semi-permanent VM map holes
+        under normal jemalloc operation; and 2) for [64-bit] Windows, which
+        disallows split / merged regions with
+        <parameter><constant>MEM_RELEASE</constant></parameter>.  Although the
+        same issues may present on 32-bit platforms as well, retaining virtual
+        memory for 32-bit Linux and Windows is disabled by default due to the
+        practical possibility of address space exhaustion.  </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.dss">
+        <term>
+          <mallctl>opt.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>dss (<citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry>) allocation precedence as
+        related to <citerefentry><refentrytitle>mmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> allocation.  The following
+        settings are supported if
+        <citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> is supported by the operating
+        system: <quote>disabled</quote>, <quote>primary</quote>, and
+        <quote>secondary</quote>; otherwise only <quote>disabled</quote> is
+        supported.  The default is <quote>secondary</quote> if
+        <citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> is supported by the operating
+        system; <quote>disabled</quote> otherwise.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.narenas">
+        <term>
+          <mallctl>opt.narenas</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum number of arenas to use for automatic
+        multiplexing of threads and arenas.  The default is four times the
+        number of CPUs, or one if there is a single CPU.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.oversize_threshold">
+        <term>
+          <mallctl>opt.oversize_threshold</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>The threshold in bytes of which requests are considered
+        oversize.  Allocation requests with greater sizes are fulfilled from a
+        dedicated arena (automatically managed, however not within
+        <literal>narenas</literal>), in order to reduce fragmentation by not
+        mixing huge allocations with small ones.  In addition, the decay API
+        guarantees on the extents greater than the specified threshold may be
+        overridden.  Note that requests with arena index specified via
+        <constant>MALLOCX_ARENA</constant>, or threads associated with explicit
+        arenas will not be considered.  The default threshold is 8MiB.  Values
+        not within large size classes disables this feature.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.percpu_arena">
+        <term>
+          <mallctl>opt.percpu_arena</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Per CPU arena mode.  Use the <quote>percpu</quote>
+        setting to enable this feature, which uses number of CPUs to determine
+        number of arenas, and bind threads to arenas dynamically based on the
+        CPU the thread runs on currently.  <quote>phycpu</quote> setting uses
+        one arena per physical CPU, which means the two hyper threads on the
+        same CPU share one arena.  Note that no runtime checking regarding the
+        availability of hyper threading is done at the moment.  When set to
+        <quote>disabled</quote>, narenas and thread to arena association will
+        not be impacted by this option.  The default is <quote>disabled</quote>.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.background_thread">
+        <term>
+          <mallctl>opt.background_thread</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Internal background worker threads enabled/disabled.
+        Because of potential circular dependencies, enabling background thread
+        using this option may cause crash or deadlock during initialization. For
+        a reliable way to use this feature, see <link
+        linkend="background_thread">background_thread</link> for dynamic control
+        options and details.  This option is disabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.max_background_threads">
+        <term>
+          <mallctl>opt.max_background_threads</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum number of background threads that will be created
+        if <link linkend="background_thread">background_thread</link> is set.
+        Defaults to number of cpus.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.dirty_decay_ms">
+        <term>
+          <mallctl>opt.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused dirty pages until an equivalent set of unused dirty pages
+        is purged (i.e. converted to muzzy via e.g.
+        <function>madvise(<parameter>...</parameter><parameter><constant>MADV_FREE</constant></parameter>)</function>
+        if supported by the operating system, or converted to clean otherwise)
+        and/or reused.  Dirty pages are defined as previously having been
+        potentially written to by the application, and therefore consuming
+        physical memory, yet having no current use.  The pages are incrementally
+        purged according to a sigmoidal decay curve that starts and ends with
+        zero purge rate.  A decay time of 0 causes all unused dirty pages to be
+        purged immediately upon creation.  A decay time of -1 disables purging.
+        The default decay time is 10 seconds.  See <link
+        linkend="arenas.dirty_decay_ms"><mallctl>arenas.dirty_decay_ms</mallctl></link>
+        and <link
+        linkend="arena.i.dirty_decay_ms"><mallctl>arena.&lt;i&gt;.dirty_decay_ms</mallctl></link>
+        for related dynamic control options.  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for a description of muzzy pages.for a description of muzzy pages.  Note
+        that when the <link
+        linkend="opt.oversize_threshold"><mallctl>oversize_threshold</mallctl></link>
+        feature is enabled, the arenas reserved for oversize requests may have
+        its own default decay settings.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.muzzy_decay_ms">
+        <term>
+          <mallctl>opt.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused muzzy pages until an equivalent set of unused muzzy pages
+        is purged (i.e. converted to clean) and/or reused.  Muzzy pages are
+        defined as previously having been unused dirty pages that were
+        subsequently purged in a manner that left them subject to the
+        reclamation whims of the operating system (e.g.
+        <function>madvise(<parameter>...</parameter><parameter><constant>MADV_FREE</constant></parameter>)</function>),
+        and therefore in an indeterminate state.  The pages are incrementally
+        purged according to a sigmoidal decay curve that starts and ends with
+        zero purge rate.  A decay time of 0 causes all unused muzzy pages to be
+        purged immediately upon creation.  A decay time of -1 disables purging.
+        The default decay time is 10 seconds.  See <link
+        linkend="arenas.muzzy_decay_ms"><mallctl>arenas.muzzy_decay_ms</mallctl></link>
+        and <link
+        linkend="arena.i.muzzy_decay_ms"><mallctl>arena.&lt;i&gt;.muzzy_decay_ms</mallctl></link>
+        for related dynamic control options.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_extent_max_active_fit">
+        <term>
+          <mallctl>opt.lg_extent_max_active_fit</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>When reusing dirty extents, this determines the (log
+        base 2 of the) maximum ratio between the size of the active extent
+        selected (to split off from) and the size of the requested allocation.
+        This prevents the splitting of large active extents for smaller
+        allocations, which can reduce fragmentation over the long run
+        (especially for non-active extents).  Lower value may reduce
+        fragmentation, at the cost of extra active extents.  The default value
+        is 6, which gives a maximum ratio of 64 (2^6).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.stats_print">
+        <term>
+          <mallctl>opt.stats_print</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Enable/disable statistics printing at exit.  If
+        enabled, the <function>malloc_stats_print()</function>
+        function is called at program exit via an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function.  <link
+        linkend="opt.stats_print_opts"><mallctl>opt.stats_print_opts</mallctl></link>
+        can be combined to specify output options. If
+        <option>--enable-stats</option> is specified during configuration, this
+        has the potential to cause deadlock for a multi-threaded process that
+        exits while one or more threads are executing in the memory allocation
+        functions.  Furthermore, <function>atexit()</function> may
+        allocate memory during application initialization and then deadlock
+        internally when jemalloc in turn calls
+        <function>atexit()</function>, so this option is not
+        universally usable (though the application can register its own
+        <function>atexit()</function> function with equivalent
+        functionality).  Therefore, this option should only be used with care;
+        it is primarily intended as a performance tuning aid during application
+        development.  This option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.stats_print_opts">
+        <term>
+          <mallctl>opt.stats_print_opts</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Options (the <parameter>opts</parameter> string) to pass
+        to the <function>malloc_stats_print()</function> at exit (enabled
+        through <link
+        linkend="opt.stats_print"><mallctl>opt.stats_print</mallctl></link>). See
+        available options in <link
+        linkend="malloc_stats_print_opts"><function>malloc_stats_print()</function></link>.
+        Has no effect unless <link
+        linkend="opt.stats_print"><mallctl>opt.stats_print</mallctl></link> is
+        enabled.  The default is <quote></quote>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.junk">
+        <term>
+          <mallctl>opt.junk</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Junk filling.  If set to <quote>alloc</quote>, each byte
+        of uninitialized allocated memory will be initialized to
+        <literal>0xa5</literal>.  If set to <quote>free</quote>, all deallocated
+        memory will be initialized to <literal>0x5a</literal>.  If set to
+        <quote>true</quote>, both allocated and deallocated memory will be
+        initialized, and if set to <quote>false</quote>, junk filling be
+        disabled entirely.  This is intended for debugging and will impact
+        performance negatively.  This option is <quote>false</quote> by default
+        unless <option>--enable-debug</option> is specified during
+        configuration, in which case it is <quote>true</quote> by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.zero">
+        <term>
+          <mallctl>opt.zero</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-fill</option>]
+        </term>
+        <listitem><para>Zero filling enabled/disabled.  If enabled, each byte
+        of uninitialized allocated memory will be initialized to 0.  Note that
+        this initialization only happens once for each byte, so
+        <function>realloc()</function> and
+        <function>rallocx()</function> calls do not zero memory that
+        was previously allocated.  This is intended for debugging and will
+        impact performance negatively.  This option is disabled by default.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.utrace">
+        <term>
+          <mallctl>opt.utrace</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-utrace</option>]
+        </term>
+        <listitem><para>Allocation tracing based on
+        <citerefentry><refentrytitle>utrace</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> enabled/disabled.  This option
+        is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.xmalloc">
+        <term>
+          <mallctl>opt.xmalloc</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-xmalloc</option>]
+        </term>
+        <listitem><para>Abort-on-out-of-memory enabled/disabled.  If enabled,
+        rather than returning failure for any allocation function, display a
+        diagnostic message on <constant>STDERR_FILENO</constant> and cause the
+        program to drop core (using
+        <citerefentry><refentrytitle>abort</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry>).  If an application is
+        designed to depend on this behavior, set the option at compile time by
+        including the following in the source code:
+        <programlisting language="C"><![CDATA[
+malloc_conf = "xmalloc:true";]]></programlisting>
+        This option is disabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.tcache">
+        <term>
+          <mallctl>opt.tcache</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Thread-specific caching (tcache) enabled/disabled.  When
+        there are multiple threads, each thread uses a tcache for objects up to
+        a certain size.  Thread-specific caching allows many allocations to be
+        satisfied without performing any thread synchronization, at the cost of
+        increased memory use.  See the <link
+        linkend="opt.lg_tcache_max"><mallctl>opt.lg_tcache_max</mallctl></link>
+        option for related tuning information.  This option is enabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_tcache_max">
+        <term>
+          <mallctl>opt.lg_tcache_max</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size class (log base 2) to cache in the
+        thread-specific cache (tcache).  At a minimum, all small size classes
+        are cached, and at a maximum all large size classes are cached.  The
+        default maximum is 32 KiB (2^15).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.thp">
+        <term>
+          <mallctl>opt.thp</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Transparent hugepage (THP) mode. Settings "always",
+        "never" and "default" are available if THP is supported by the operating
+        system.  The "always" setting enables transparent hugepage for all user
+        memory mappings with
+        <parameter><constant>MADV_HUGEPAGE</constant></parameter>; "never"
+        ensures no transparent hugepage with
+        <parameter><constant>MADV_NOHUGEPAGE</constant></parameter>; the default
+        setting "default" makes no changes.  Note that: this option does not
+        affect THP for jemalloc internal metadata (see <link
+        linkend="opt.metadata_thp"><mallctl>opt.metadata_thp</mallctl></link>);
+        in addition, for arenas with customized <link
+        linkend="arena.i.extent_hooks"><mallctl>extent_hooks</mallctl></link>,
+        this option is bypassed as it is implemented as part of the default
+        extent hooks.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof">
+        <term>
+          <mallctl>opt.prof</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Memory profiling enabled/disabled.  If enabled, profile
+        memory allocation activity.  See the <link
+        linkend="opt.prof_active"><mallctl>opt.prof_active</mallctl></link>
+        option for on-the-fly activation/deactivation.  See the <link
+        linkend="opt.lg_prof_sample"><mallctl>opt.lg_prof_sample</mallctl></link>
+        option for probabilistic sampling control.  See the <link
+        linkend="opt.prof_accum"><mallctl>opt.prof_accum</mallctl></link>
+        option for control of cumulative sample reporting.  See the <link
+        linkend="opt.lg_prof_interval"><mallctl>opt.lg_prof_interval</mallctl></link>
+        option for information on interval-triggered profile dumping, the <link
+        linkend="opt.prof_gdump"><mallctl>opt.prof_gdump</mallctl></link>
+        option for information on high-water-triggered profile dumping, and the
+        <link linkend="opt.prof_final"><mallctl>opt.prof_final</mallctl></link>
+        option for final profile dumping.  Profile output is compatible with
+        the <command>jeprof</command> command, which is based on the
+        <command>pprof</command> that is developed as part of the <ulink
+        url="http://code.google.com/p/gperftools/">gperftools
+        package</ulink>.  See <link linkend="heap_profile_format">HEAP PROFILE
+        FORMAT</link> for heap profile format documentation.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_prefix">
+        <term>
+          <mallctl>opt.prof_prefix</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Filename prefix for profile dumps.  If the prefix is
+        set to the empty string, no automatic dumps will occur; this is
+        primarily useful for disabling the automatic final heap dump (which
+        also disables leak reporting, if enabled).  The default prefix is
+        <filename>jeprof</filename>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_active">
+        <term>
+          <mallctl>opt.prof_active</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Profiling activated/deactivated.  This is a secondary
+        control mechanism that makes it possible to start the application with
+        profiling enabled (see the <link
+        linkend="opt.prof"><mallctl>opt.prof</mallctl></link> option) but
+        inactive, then toggle profiling at any time during program execution
+        with the <link
+        linkend="prof.active"><mallctl>prof.active</mallctl></link> mallctl.
+        This option is enabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_thread_active_init">
+        <term>
+          <mallctl>opt.prof_thread_active_init</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Initial setting for <link
+        linkend="thread.prof.active"><mallctl>thread.prof.active</mallctl></link>
+        in newly created threads.  The initial setting for newly created threads
+        can also be changed during execution via the <link
+        linkend="prof.thread_active_init"><mallctl>prof.thread_active_init</mallctl></link>
+        mallctl.  This option is enabled by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_prof_sample">
+        <term>
+          <mallctl>opt.lg_prof_sample</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average interval (log base 2) between allocation
+        samples, as measured in bytes of allocation activity.  Increasing the
+        sampling interval decreases profile fidelity, but also decreases the
+        computational overhead.  The default sample interval is 512 KiB (2^19
+        B).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_accum">
+        <term>
+          <mallctl>opt.prof_accum</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Reporting of cumulative object/byte counts in profile
+        dumps enabled/disabled.  If this option is enabled, every unique
+        backtrace must be stored for the duration of execution.  Depending on
+        the application, this can impose a large memory overhead, and the
+        cumulative counts are not always of interest.  This option is disabled
+        by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.lg_prof_interval">
+        <term>
+          <mallctl>opt.lg_prof_interval</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average interval (log base 2) between memory profile
+        dumps, as measured in bytes of allocation activity.  The actual
+        interval between dumps may be sporadic because decentralized allocation
+        counters are used to avoid synchronization bottlenecks.  Profiles are
+        dumped to files named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.i&lt;iseq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the
+        <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.  By default, interval-triggered profile dumping is disabled
+        (encoded as -1).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_gdump">
+        <term>
+          <mallctl>opt.prof_gdump</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Set the initial state of <link
+        linkend="prof.gdump"><mallctl>prof.gdump</mallctl></link>, which when
+        enabled triggers a memory profile dump every time the total virtual
+        memory exceeds the previous maximum.  This option is disabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_final">
+        <term>
+          <mallctl>opt.prof_final</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Use an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function to dump final memory
+        usage to a file named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.f.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.  Note that <function>atexit()</function> may allocate
+        memory during application initialization and then deadlock internally
+        when jemalloc in turn calls <function>atexit()</function>, so
+        this option is not universally usable (though the application can
+        register its own <function>atexit()</function> function with
+        equivalent functionality).  This option is disabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="opt.prof_leak">
+        <term>
+          <mallctl>opt.prof_leak</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Leak reporting enabled/disabled.  If enabled, use an
+        <citerefentry><refentrytitle>atexit</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> function to report memory leaks
+        detected by allocation sampling.  See the
+        <link linkend="opt.prof"><mallctl>opt.prof</mallctl></link> option for
+        information on analyzing heap profile output.  This option is disabled
+        by default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.arena">
+        <term>
+          <mallctl>thread.arena</mallctl>
+          (<type>unsigned</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Get or set the arena associated with the calling
+        thread.  If the specified arena was not initialized beforehand (see the
+        <link
+        linkend="arena.i.initialized"><mallctl>arena.i.initialized</mallctl></link>
+        mallctl), it will be automatically initialized as a side effect of
+        calling this interface.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.allocated">
+        <term>
+          <mallctl>thread.allocated</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get the total number of bytes ever allocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.allocatedp">
+        <term>
+          <mallctl>thread.allocatedp</mallctl>
+          (<type>uint64_t *</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get a pointer to the the value that is returned by the
+        <link
+        linkend="thread.allocated"><mallctl>thread.allocated</mallctl></link>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <function>mallctl*()</function> calls.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.deallocated">
+        <term>
+          <mallctl>thread.deallocated</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get the total number of bytes ever deallocated by the
+        calling thread.  This counter has the potential to wrap around; it is
+        up to the application to appropriately interpret the counter in such
+        cases.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.deallocatedp">
+        <term>
+          <mallctl>thread.deallocatedp</mallctl>
+          (<type>uint64_t *</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Get a pointer to the the value that is returned by the
+        <link
+        linkend="thread.deallocated"><mallctl>thread.deallocated</mallctl></link>
+        mallctl.  This is useful for avoiding the overhead of repeated
+        <function>mallctl*()</function> calls.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.tcache.enabled">
+        <term>
+          <mallctl>thread.tcache.enabled</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Enable/disable calling thread's tcache.  The tcache is
+        implicitly flushed as a side effect of becoming
+        disabled (see <link
+        linkend="thread.tcache.flush"><mallctl>thread.tcache.flush</mallctl></link>).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.tcache.flush">
+        <term>
+          <mallctl>thread.tcache.flush</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Flush calling thread's thread-specific cache (tcache).
+        This interface releases all cached objects and internal data structures
+        associated with the calling thread's tcache.  Ordinarily, this interface
+        need not be called, since automatic periodic incremental garbage
+        collection occurs, and the thread cache is automatically discarded when
+        a thread exits.  However, garbage collection is triggered by allocation
+        activity, so it is possible for a thread that stops
+        allocating/deallocating to retain its cache indefinitely, in which case
+        the developer may find manual flushing useful.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.prof.name">
+        <term>
+          <mallctl>thread.prof.name</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal> or
+          <literal>-w</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Get/set the descriptive name associated with the calling
+        thread in memory profile dumps.  An internal copy of the name string is
+        created, so the input string need not be maintained after this interface
+        completes execution.  The output string of this interface should be
+        copied for non-ephemeral uses, because multiple implementation details
+        can cause asynchronous string deallocation.  Furthermore, each
+        invocation of this interface can only read or write; simultaneous
+        read/write is not supported due to string lifetime limitations.  The
+        name string must be nil-terminated and comprised only of characters in
+        the sets recognized
+        by <citerefentry><refentrytitle>isgraph</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry> and
+        <citerefentry><refentrytitle>isblank</refentrytitle>
+        <manvolnum>3</manvolnum></citerefentry>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="thread.prof.active">
+        <term>
+          <mallctl>thread.prof.active</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Control whether sampling is currently active for the
+        calling thread.  This is an activation mechanism in addition to <link
+        linkend="prof.active"><mallctl>prof.active</mallctl></link>; both must
+        be active for the calling thread to sample.  This flag is enabled by
+        default.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="tcache.create">
+        <term>
+          <mallctl>tcache.create</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Create an explicit thread-specific cache (tcache) and
+        return an identifier that can be passed to the <link
+        linkend="MALLOCX_TCACHE"><constant>MALLOCX_TCACHE(<parameter>tc</parameter>)</constant></link>
+        macro to explicitly use the specified cache rather than the
+        automatically managed one that is used by default.  Each explicit cache
+        can be used by only one thread at a time; the application must assure
+        that this constraint holds.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="tcache.flush">
+        <term>
+          <mallctl>tcache.flush</mallctl>
+          (<type>unsigned</type>)
+          <literal>-w</literal>
+        </term>
+        <listitem><para>Flush the specified thread-specific cache (tcache).  The
+        same considerations apply to this interface as to <link
+        linkend="thread.tcache.flush"><mallctl>thread.tcache.flush</mallctl></link>,
+        except that the tcache will never be automatically discarded.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="tcache.destroy">
+        <term>
+          <mallctl>tcache.destroy</mallctl>
+          (<type>unsigned</type>)
+          <literal>-w</literal>
+        </term>
+        <listitem><para>Flush the specified thread-specific cache (tcache) and
+        make the identifier available for use during a future tcache creation.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.initialized">
+        <term>
+          <mallctl>arena.&lt;i&gt;.initialized</mallctl>
+          (<type>bool</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Get whether the specified arena's statistics are
+        initialized (i.e. the arena was initialized prior to the current epoch).
+        This interface can also be nominally used to query whether the merged
+        statistics corresponding to <constant>MALLCTL_ARENAS_ALL</constant> are
+        initialized (always true).</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.decay">
+        <term>
+          <mallctl>arena.&lt;i&gt;.decay</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Trigger decay-based purging of unused dirty/muzzy pages
+        for arena &lt;i&gt;, or for all arenas if &lt;i&gt; equals
+        <constant>MALLCTL_ARENAS_ALL</constant>.  The proportion of unused
+        dirty/muzzy pages to be purged depends on the current time; see <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        and <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzy_decay_ms</mallctl></link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.purge">
+        <term>
+          <mallctl>arena.&lt;i&gt;.purge</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Purge all unused dirty pages for arena &lt;i&gt;, or for
+        all arenas if &lt;i&gt; equals <constant>MALLCTL_ARENAS_ALL</constant>.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.reset">
+        <term>
+          <mallctl>arena.&lt;i&gt;.reset</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Discard all of the arena's extant allocations.  This
+        interface can only be used with arenas explicitly created via <link
+        linkend="arenas.create"><mallctl>arenas.create</mallctl></link>.  None
+        of the arena's discarded/cached allocations may accessed afterward.  As
+        part of this requirement, all thread caches which were used to
+        allocate/deallocate in conjunction with the arena must be flushed
+        beforehand.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.destroy">
+        <term>
+          <mallctl>arena.&lt;i&gt;.destroy</mallctl>
+          (<type>void</type>)
+          <literal>--</literal>
+        </term>
+        <listitem><para>Destroy the arena.  Discard all of the arena's extant
+        allocations using the same mechanism as for <link
+        linkend="arena.i.reset"><mallctl>arena.&lt;i&gt;.reset</mallctl></link>
+        (with all the same constraints and side effects), merge the arena stats
+        into those accessible at arena index
+        <constant>MALLCTL_ARENAS_DESTROYED</constant>, and then completely
+        discard all metadata associated with the arena.  Future calls to <link
+        linkend="arenas.create"><mallctl>arenas.create</mallctl></link> may
+        recycle the arena index.  Destruction will fail if any threads are
+        currently associated with the arena as a result of calls to <link
+        linkend="thread.arena"><mallctl>thread.arena</mallctl></link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.dss">
+        <term>
+          <mallctl>arena.&lt;i&gt;.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Set the precedence of dss allocation as related to mmap
+        allocation for arena &lt;i&gt;, or for all arenas if &lt;i&gt; equals
+        <constant>MALLCTL_ARENAS_ALL</constant>.  See <link
+        linkend="opt.dss"><mallctl>opt.dss</mallctl></link> for supported
+        settings.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.dirty_decay_ms">
+        <term>
+          <mallctl>arena.&lt;i&gt;.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current per-arena approximate time in milliseconds from
+        the creation of a set of unused dirty pages until an equivalent set of
+        unused dirty pages is purged and/or reused.  Each time this interface is
+        set, all currently unused dirty pages are considered to have fully
+        decayed, which causes immediate purging of all unused dirty pages unless
+        the decay time is set to -1 (i.e. purging disabled).  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.muzzy_decay_ms">
+        <term>
+          <mallctl>arena.&lt;i&gt;.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current per-arena approximate time in milliseconds from
+        the creation of a set of unused muzzy pages until an equivalent set of
+        unused muzzy pages is purged and/or reused.  Each time this interface is
+        set, all currently unused muzzy pages are considered to have fully
+        decayed, which causes immediate purging of all unused muzzy pages unless
+        the decay time is set to -1 (i.e. purging disabled).  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.retain_grow_limit">
+        <term>
+          <mallctl>arena.&lt;i&gt;.retain_grow_limit</mallctl>
+          (<type>size_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Maximum size to grow retained region (only relevant when
+        <link linkend="opt.retain"><mallctl>opt.retain</mallctl></link> is
+        enabled).  This controls the maximum increment to expand virtual memory,
+        or allocation through <link
+        linkend="arena.i.extent_hooks"><mallctl>arena.&lt;i&gt;extent_hooks</mallctl></link>.
+        In particular, if customized extent hooks reserve physical memory
+        (e.g. 1G huge pages), this is useful to control the allocation hook's
+        input size.  The default is no limit.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arena.i.extent_hooks">
+        <term>
+          <mallctl>arena.&lt;i&gt;.extent_hooks</mallctl>
+          (<type>extent_hooks_t *</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Get or set the extent management hook functions for
+        arena &lt;i&gt;.  The functions must be capable of operating on all
+        extant extents associated with arena &lt;i&gt;, usually by passing
+        unknown extents to the replaced functions.  In practice, it is feasible
+        to control allocation for arenas explicitly created via <link
+        linkend="arenas.create"><mallctl>arenas.create</mallctl></link> such
+        that all extents originate from an application-supplied extent allocator
+        (by specifying the custom extent hook functions during arena creation).
+        However, the API guarantees for the automatically created arenas may be
+        relaxed -- hooks set there may be called in a "best effort" fashion; in
+        addition there may be extents created prior to the application having an
+        opportunity to take over extent allocation.</para>
+
+        <programlisting language="C"><![CDATA[
+typedef extent_hooks_s extent_hooks_t;
+struct extent_hooks_s {
+	extent_alloc_t		*alloc;
+	extent_dalloc_t		*dalloc;
+	extent_destroy_t	*destroy;
+	extent_commit_t		*commit;
+	extent_decommit_t	*decommit;
+	extent_purge_t		*purge_lazy;
+	extent_purge_t		*purge_forced;
+	extent_split_t		*split;
+	extent_merge_t		*merge;
+};]]></programlisting>
+        <para>The <type>extent_hooks_t</type> structure comprises function
+        pointers which are described individually below.  jemalloc uses these
+        functions to manage extent lifetime, which starts off with allocation of
+        mapped committed memory, in the simplest case followed by deallocation.
+        However, there are performance and platform reasons to retain extents
+        for later reuse.  Cleanup attempts cascade from deallocation to decommit
+        to forced purging to lazy purging, which gives the extent management
+        functions opportunities to reject the most permanent cleanup operations
+        in favor of less permanent (and often less costly) operations.  All
+        operations except allocation can be universally opted out of by setting
+        the hook pointers to <constant>NULL</constant>, or selectively opted out
+        of by returning failure.  Note that once the extent hook is set, the
+        structure is accessed directly by the associated arenas, so it must
+        remain valid for the entire lifetime of the arenas.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef void *<function>(extent_alloc_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>new_addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>alignment</parameter></paramdef>
+          <paramdef>bool *<parameter>zero</parameter></paramdef>
+          <paramdef>bool *<parameter>commit</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent allocation function conforms to the
+        <type>extent_alloc_t</type> type and upon success returns a pointer to
+        <parameter>size</parameter> bytes of mapped memory on behalf of arena
+        <parameter>arena_ind</parameter> such that the extent's base address is
+        a multiple of <parameter>alignment</parameter>, as well as setting
+        <parameter>*zero</parameter> to indicate whether the extent is zeroed
+        and <parameter>*commit</parameter> to indicate whether the extent is
+        committed.  Upon error the function returns <constant>NULL</constant>
+        and leaves <parameter>*zero</parameter> and
+        <parameter>*commit</parameter> unmodified.  The
+        <parameter>size</parameter> parameter is always a multiple of the page
+        size.  The <parameter>alignment</parameter> parameter is always a power
+        of two at least as large as the page size.  Zeroing is mandatory if
+        <parameter>*zero</parameter> is true upon function entry.  Committing is
+        mandatory if <parameter>*commit</parameter> is true upon function entry.
+        If <parameter>new_addr</parameter> is not <constant>NULL</constant>, the
+        returned pointer must be <parameter>new_addr</parameter> on success or
+        <constant>NULL</constant> on error.  Committed memory may be committed
+        in absolute terms as on a system that does not overcommit, or in
+        implicit terms as on a system that overcommits and satisfies physical
+        memory needs on demand via soft page faults.  Note that replacing the
+        default extent allocation function makes the arena's <link
+        linkend="arena.i.dss"><mallctl>arena.&lt;i&gt;.dss</mallctl></link>
+        setting irrelevant.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_dalloc_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>
+        An extent deallocation function conforms to the
+        <type>extent_dalloc_t</type> type and deallocates an extent at given
+        <parameter>addr</parameter> and <parameter>size</parameter> with
+        <parameter>committed</parameter>/decommited memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>, returning false upon
+        success.  If the function returns true, this indicates opt-out from
+        deallocation; the virtual memory mapping associated with the extent
+        remains mapped, in the same commit state, and available for future use,
+        in which case it will be automatically retained for later reuse.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef void <function>(extent_destroy_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>
+        An extent destruction function conforms to the
+        <type>extent_destroy_t</type> type and unconditionally destroys an
+        extent at given <parameter>addr</parameter> and
+        <parameter>size</parameter> with
+        <parameter>committed</parameter>/decommited memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>.  This function may be
+        called to destroy retained extents during arena destruction (see <link
+        linkend="arena.i.destroy"><mallctl>arena.&lt;i&gt;.destroy</mallctl></link>).</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_commit_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>offset</parameter></paramdef>
+          <paramdef>size_t <parameter>length</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent commit function conforms to the
+        <type>extent_commit_t</type> type and commits zeroed physical memory to
+        back pages within an extent at given <parameter>addr</parameter> and
+        <parameter>size</parameter> at <parameter>offset</parameter> bytes,
+        extending for <parameter>length</parameter> on behalf of arena
+        <parameter>arena_ind</parameter>, returning false upon success.
+        Committed memory may be committed in absolute terms as on a system that
+        does not overcommit, or in implicit terms as on a system that
+        overcommits and satisfies physical memory needs on demand via soft page
+        faults. If the function returns true, this indicates insufficient
+        physical memory to satisfy the request.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_decommit_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>offset</parameter></paramdef>
+          <paramdef>size_t <parameter>length</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent decommit function conforms to the
+        <type>extent_decommit_t</type> type and decommits any physical memory
+        that is backing pages within an extent at given
+        <parameter>addr</parameter> and <parameter>size</parameter> at
+        <parameter>offset</parameter> bytes, extending for
+        <parameter>length</parameter> on behalf of arena
+        <parameter>arena_ind</parameter>, returning false upon success, in which
+        case the pages will be committed via the extent commit function before
+        being reused.  If the function returns true, this indicates opt-out from
+        decommit; the memory remains committed and available for future use, in
+        which case it will be automatically retained for later reuse.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_purge_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>offset</parameter></paramdef>
+          <paramdef>size_t <parameter>length</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent purge function conforms to the
+        <type>extent_purge_t</type> type and discards physical pages
+        within the virtual memory mapping associated with an extent at given
+        <parameter>addr</parameter> and <parameter>size</parameter> at
+        <parameter>offset</parameter> bytes, extending for
+        <parameter>length</parameter> on behalf of arena
+        <parameter>arena_ind</parameter>.  A lazy extent purge function (e.g.
+        implemented via
+        <function>madvise(<parameter>...</parameter><parameter><constant>MADV_FREE</constant></parameter>)</function>)
+        can delay purging indefinitely and leave the pages within the purged
+        virtual memory range in an indeterminite state, whereas a forced extent
+        purge function immediately purges, and the pages within the virtual
+        memory range will be zero-filled the next time they are accessed.  If
+        the function returns true, this indicates failure to purge.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_split_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr</parameter></paramdef>
+          <paramdef>size_t <parameter>size</parameter></paramdef>
+          <paramdef>size_t <parameter>size_a</parameter></paramdef>
+          <paramdef>size_t <parameter>size_b</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent split function conforms to the
+        <type>extent_split_t</type> type and optionally splits an extent at
+        given <parameter>addr</parameter> and <parameter>size</parameter> into
+        two adjacent extents, the first of <parameter>size_a</parameter> bytes,
+        and the second of <parameter>size_b</parameter> bytes, operating on
+        <parameter>committed</parameter>/decommitted memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>, returning false upon
+        success.  If the function returns true, this indicates that the extent
+        remains unsplit and therefore should continue to be operated on as a
+        whole.</para>
+
+        <funcsynopsis><funcprototype>
+          <funcdef>typedef bool <function>(extent_merge_t)</function></funcdef>
+          <paramdef>extent_hooks_t *<parameter>extent_hooks</parameter></paramdef>
+          <paramdef>void *<parameter>addr_a</parameter></paramdef>
+          <paramdef>size_t <parameter>size_a</parameter></paramdef>
+          <paramdef>void *<parameter>addr_b</parameter></paramdef>
+          <paramdef>size_t <parameter>size_b</parameter></paramdef>
+          <paramdef>bool <parameter>committed</parameter></paramdef>
+          <paramdef>unsigned <parameter>arena_ind</parameter></paramdef>
+        </funcprototype></funcsynopsis>
+        <literallayout></literallayout>
+        <para>An extent merge function conforms to the
+        <type>extent_merge_t</type> type and optionally merges adjacent extents,
+        at given <parameter>addr_a</parameter> and <parameter>size_a</parameter>
+        with given <parameter>addr_b</parameter> and
+        <parameter>size_b</parameter> into one contiguous extent, operating on
+        <parameter>committed</parameter>/decommitted memory as indicated, on
+        behalf of arena <parameter>arena_ind</parameter>, returning false upon
+        success.  If the function returns true, this indicates that the extents
+        remain distinct mappings and therefore should continue to be operated on
+        independently.</para>
+        </listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.narenas">
+        <term>
+          <mallctl>arenas.narenas</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Current limit on number of arenas.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.dirty_decay_ms">
+        <term>
+          <mallctl>arenas.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current default per-arena approximate time in
+        milliseconds from the creation of a set of unused dirty pages until an
+        equivalent set of unused dirty pages is purged and/or reused, used to
+        initialize <link
+        linkend="arena.i.dirty_decay_ms"><mallctl>arena.&lt;i&gt;.dirty_decay_ms</mallctl></link>
+        during arena creation.  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.muzzy_decay_ms">
+        <term>
+          <mallctl>arenas.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Current default per-arena approximate time in
+        milliseconds from the creation of a set of unused muzzy pages until an
+        equivalent set of unused muzzy pages is purged and/or reused, used to
+        initialize <link
+        linkend="arena.i.muzzy_decay_ms"><mallctl>arena.&lt;i&gt;.muzzy_decay_ms</mallctl></link>
+        during arena creation.  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.quantum">
+        <term>
+          <mallctl>arenas.quantum</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Quantum size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.page">
+        <term>
+          <mallctl>arenas.page</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Page size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.tcache_max">
+        <term>
+          <mallctl>arenas.tcache_max</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum thread-cached size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.nbins">
+        <term>
+          <mallctl>arenas.nbins</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of bin size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.nhbins">
+        <term>
+          <mallctl>arenas.nhbins</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Total number of thread cache bin size
+        classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.bin.i.size">
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size supported by size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.bin.i.nregs">
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.nregs</mallctl>
+          (<type>uint32_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of regions per slab.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.bin.i.slab_size">
+        <term>
+          <mallctl>arenas.bin.&lt;i&gt;.slab_size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of bytes per slab.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.nlextents">
+        <term>
+          <mallctl>arenas.nlextents</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Total number of large size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.lextent.i.size">
+        <term>
+          <mallctl>arenas.lextent.&lt;i&gt;.size</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Maximum size supported by this large size
+        class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.create">
+        <term>
+          <mallctl>arenas.create</mallctl>
+          (<type>unsigned</type>, <type>extent_hooks_t *</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Explicitly create a new arena outside the range of
+        automatically managed arenas, with optionally specified extent hooks,
+        and return the new arena index.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="arenas.lookup">
+        <term>
+          <mallctl>arenas.lookup</mallctl>
+          (<type>unsigned</type>, <type>void*</type>)
+          <literal>rw</literal>
+        </term>
+        <listitem><para>Index of the arena to which an allocation belongs to.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.thread_active_init">
+        <term>
+          <mallctl>prof.thread_active_init</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Control the initial setting for <link
+        linkend="thread.prof.active"><mallctl>thread.prof.active</mallctl></link>
+        in newly created threads.  See the <link
+        linkend="opt.prof_thread_active_init"><mallctl>opt.prof_thread_active_init</mallctl></link>
+        option for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.active">
+        <term>
+          <mallctl>prof.active</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Control whether sampling is currently active.  See the
+        <link
+        linkend="opt.prof_active"><mallctl>opt.prof_active</mallctl></link>
+        option for additional information, as well as the interrelated <link
+        linkend="thread.prof.active"><mallctl>thread.prof.active</mallctl></link>
+        mallctl.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.dump">
+        <term>
+          <mallctl>prof.dump</mallctl>
+          (<type>const char *</type>)
+          <literal>-w</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Dump a memory profile to the specified file, or if NULL
+        is specified, to a file according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.m&lt;mseq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the
+        <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.gdump">
+        <term>
+          <mallctl>prof.gdump</mallctl>
+          (<type>bool</type>)
+          <literal>rw</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>When enabled, trigger a memory profile dump every time
+        the total virtual memory exceeds the previous maximum.  Profiles are
+        dumped to files named according to the pattern
+        <filename>&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.u&lt;useq&gt;.heap</filename>,
+        where <literal>&lt;prefix&gt;</literal> is controlled by the <link
+        linkend="opt.prof_prefix"><mallctl>opt.prof_prefix</mallctl></link>
+        option.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.reset">
+        <term>
+          <mallctl>prof.reset</mallctl>
+          (<type>size_t</type>)
+          <literal>-w</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Reset all memory profile statistics, and optionally
+        update the sample rate (see <link
+        linkend="opt.lg_prof_sample"><mallctl>opt.lg_prof_sample</mallctl></link>
+        and <link
+        linkend="prof.lg_sample"><mallctl>prof.lg_sample</mallctl></link>).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.lg_sample">
+        <term>
+          <mallctl>prof.lg_sample</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Get the current sample rate (see <link
+        linkend="opt.lg_prof_sample"><mallctl>opt.lg_prof_sample</mallctl></link>).
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="prof.interval">
+        <term>
+          <mallctl>prof.interval</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-prof</option>]
+        </term>
+        <listitem><para>Average number of bytes allocated between
+        interval-based profile dumps.  See the
+        <link
+        linkend="opt.lg_prof_interval"><mallctl>opt.lg_prof_interval</mallctl></link>
+        option for additional information.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.allocated">
+        <term>
+          <mallctl>stats.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes allocated by the
+        application.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.active">
+        <term>
+          <mallctl>stats.active</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in active pages allocated by the
+        application.  This is a multiple of the page size, and greater than or
+        equal to <link
+        linkend="stats.allocated"><mallctl>stats.allocated</mallctl></link>.
+        This does not include <link linkend="stats.arenas.i.pdirty">
+        <mallctl>stats.arenas.&lt;i&gt;.pdirty</mallctl></link>,
+        <link linkend="stats.arenas.i.pmuzzy">
+        <mallctl>stats.arenas.&lt;i&gt;.pmuzzy</mallctl></link>, nor pages
+        entirely devoted to allocator metadata.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.metadata">
+        <term>
+          <mallctl>stats.metadata</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes dedicated to metadata, which
+        comprise base allocations used for bootstrap-sensitive allocator
+        metadata structures (see <link
+        linkend="stats.arenas.i.base"><mallctl>stats.arenas.&lt;i&gt;.base</mallctl></link>)
+        and internal allocations (see <link
+        linkend="stats.arenas.i.internal"><mallctl>stats.arenas.&lt;i&gt;.internal</mallctl></link>).
+        Transparent huge page (enabled with <link
+        linkend="opt.metadata_thp">opt.metadata_thp</link>) usage is not
+        considered.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.metadata_thp">
+        <term>
+          <mallctl>stats.metadata_thp</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of transparent huge pages (THP) used for
+        metadata.  See <link
+        linkend="stats.metadata"><mallctl>stats.metadata</mallctl></link> and
+        <link linkend="opt.metadata_thp">opt.metadata_thp</link>) for
+        details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.resident">
+        <term>
+          <mallctl>stats.resident</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Maximum number of bytes in physically resident data
+        pages mapped by the allocator, comprising all pages dedicated to
+        allocator metadata, pages backing active allocations, and unused dirty
+        pages.  This is a maximum rather than precise because pages may not
+        actually be physically resident if they correspond to demand-zeroed
+        virtual memory that has not yet been touched.  This is a multiple of the
+        page size, and is larger than <link
+        linkend="stats.active"><mallctl>stats.active</mallctl></link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mapped">
+        <term>
+          <mallctl>stats.mapped</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in active extents mapped by the
+        allocator.  This is larger than <link
+        linkend="stats.active"><mallctl>stats.active</mallctl></link>.  This
+        does not include inactive extents, even those that contain unused dirty
+        pages, which means that there is no strict ordering between this and
+        <link
+        linkend="stats.resident"><mallctl>stats.resident</mallctl></link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.retained">
+        <term>
+          <mallctl>stats.retained</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Total number of bytes in virtual memory mappings that
+        were retained rather than being returned to the operating system via
+        e.g. <citerefentry><refentrytitle>munmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> or similar.  Retained virtual
+        memory is typically untouched, decommitted, or purged, so it has no
+        strongly associated physical memory (see <link
+        linkend="arena.i.extent_hooks">extent hooks</link> for details).
+        Retained memory is excluded from mapped memory statistics, e.g. <link
+        linkend="stats.mapped"><mallctl>stats.mapped</mallctl></link>.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.background_thread.num_threads">
+        <term>
+          <mallctl>stats.background_thread.num_threads</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Number of <link linkend="background_thread">background
+        threads</link> running currently.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.background_thread.num_runs">
+        <term>
+          <mallctl>stats.background_thread.num_runs</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Total number of runs from all <link
+        linkend="background_thread">background threads</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.background_thread.run_interval">
+        <term>
+          <mallctl>stats.background_thread.run_interval</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Average run interval in nanoseconds of <link
+        linkend="background_thread">background threads</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.ctl">
+        <term>
+          <mallctl>stats.mutexes.ctl.{counter};</mallctl>
+          (<type>counter specific type</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>ctl</varname> mutex (global
+        scope; mallctl related).  <mallctl>{counter}</mallctl> is one of the
+        counters below:</para>
+        <varlistentry id="mutex_counters">
+          <listitem><para><varname>num_ops</varname> (<type>uint64_t</type>):
+          Total number of lock acquisition operations on this mutex.</para>
+
+	  <para><varname>num_spin_acq</varname> (<type>uint64_t</type>): Number
+	  of times the mutex was spin-acquired.  When the mutex is currently
+	  locked and cannot be acquired immediately, a short period of
+	  spin-retry within jemalloc will be performed.  Acquired through spin
+	  generally means the contention was lightweight and not causing context
+	  switches.</para>
+
+	  <para><varname>num_wait</varname> (<type>uint64_t</type>): Number of
+	  times the mutex was wait-acquired, which means the mutex contention
+	  was not solved by spin-retry, and blocking operation was likely
+	  involved in order to acquire the mutex.  This event generally implies
+	  higher cost / longer delay, and should be investigated if it happens
+	  often.</para>
+
+	  <para><varname>max_wait_time</varname> (<type>uint64_t</type>):
+	  Maximum length of time in nanoseconds spent on a single wait-acquired
+	  lock operation.  Note that to avoid profiling overhead on the common
+	  path, this does not consider spin-acquired cases.</para>
+
+	  <para><varname>total_wait_time</varname> (<type>uint64_t</type>):
+	  Cumulative time in nanoseconds spent on wait-acquired lock operations.
+	  Similarly, spin-acquired cases are not considered.</para>
+
+	  <para><varname>max_num_thds</varname> (<type>uint32_t</type>): Maximum
+	  number of threads waiting on this mutex simultaneously.  Similarly,
+	  spin-acquired cases are not considered.</para>
+
+	  <para><varname>num_owner_switch</varname> (<type>uint64_t</type>):
+	  Number of times the current mutex owner is different from the previous
+	  one.  This event does not generally imply an issue; rather it is an
+	  indicator of how often the protected data are accessed by different
+	  threads.
+	  </para>
+	  </listitem>
+	</varlistentry>
+	</listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.background_thread">
+        <term>
+          <mallctl>stats.mutexes.background_thread.{counter}</mallctl>
+	  (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>background_thread</varname> mutex
+        (global scope; <link
+        linkend="background_thread"><mallctl>background_thread</mallctl></link>
+        related).  <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.prof">
+        <term>
+          <mallctl>stats.mutexes.prof.{counter}</mallctl>
+	  (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>prof</varname> mutex (global
+        scope; profiling related).  <mallctl>{counter}</mallctl> is one of the
+        counters in <link linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.mutexes.reset">
+        <term>
+          <mallctl>stats.mutexes.reset</mallctl>
+	  (<type>void</type>) <literal>--</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Reset all mutex profile statistics, including global
+        mutexes, arena mutexes and bin mutexes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dss">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dss</mallctl>
+          (<type>const char *</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>dss (<citerefentry><refentrytitle>sbrk</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry>) allocation precedence as
+        related to <citerefentry><refentrytitle>mmap</refentrytitle>
+        <manvolnum>2</manvolnum></citerefentry> allocation.  See <link
+        linkend="opt.dss"><mallctl>opt.dss</mallctl></link> for details.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_decay_ms">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused dirty pages until an equivalent set of unused dirty pages
+        is purged and/or reused.  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_decay_ms">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_decay_ms</mallctl>
+          (<type>ssize_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Approximate time in milliseconds from the creation of a
+        set of unused muzzy pages until an equivalent set of unused muzzy pages
+        is purged and/or reused.  See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.nthreads">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.nthreads</mallctl>
+          (<type>unsigned</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of threads currently assigned to
+        arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.uptime">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.uptime</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Time elapsed (in nanoseconds) since the arena was
+        created.  If &lt;i&gt; equals <constant>0</constant> or
+        <constant>MALLCTL_ARENAS_ALL</constant>, this is the uptime since malloc
+        initialization.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.pactive">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pactive</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages in active extents.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.pdirty">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pdirty</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages within unused extents that are
+        potentially dirty, and for which <function>madvise()</function> or
+        similar has not been called.  See <link
+        linkend="opt.dirty_decay_ms"><mallctl>opt.dirty_decay_ms</mallctl></link>
+        for a description of dirty pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.pmuzzy">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.pmuzzy</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Number of pages within unused extents that are muzzy.
+        See <link
+        linkend="opt.muzzy_decay_ms"><mallctl>opt.muzzy_decay_ms</mallctl></link>
+        for a description of muzzy pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mapped">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mapped</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of mapped bytes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.retained">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.retained</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of retained bytes.  See <link
+        linkend="stats.retained"><mallctl>stats.retained</mallctl></link> for
+        details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.extent_avail">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.extent_avail</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of allocated (but unused) extent structs in this
+	arena.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.base">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.base</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>
+        Number of bytes dedicated to bootstrap-sensitive allocator metadata
+        structures.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.internal">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.internal</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes dedicated to internal allocations.
+        Internal allocations differ from application-originated allocations in
+        that they are for internal use, and that they are omitted from heap
+        profiles.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.metadata_thp">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.metadata_thp</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of transparent huge pages (THP) used for
+        metadata.  See <link linkend="opt.metadata_thp">opt.metadata_thp</link>
+        for details.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.resident">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.resident</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Maximum number of bytes in physically resident data
+        pages mapped by the arena, comprising all pages dedicated to allocator
+        metadata, pages backing active allocations, and unused dirty pages.
+        This is a maximum rather than precise because pages may not actually be
+        physically resident if they correspond to demand-zeroed virtual memory
+        that has not yet been touched.  This is a multiple of the page
+        size.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_npurge">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_npurge</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of dirty page purge sweeps performed.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_nmadvise">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_nmadvise</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of <function>madvise()</function> or similar
+        calls made to purge dirty pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.dirty_purged">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.dirty_purged</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of dirty pages purged.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_npurge">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_npurge</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of muzzy page purge sweeps performed.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_nmadvise">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_nmadvise</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of <function>madvise()</function> or similar
+        calls made to purge muzzy pages.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.muzzy_purged">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.muzzy_purged</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of muzzy pages purged.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.allocated">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes currently allocated by small objects.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a small allocation was
+        requested from the arena's bins, whether to fill the relevant tcache if
+        <link linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is
+        enabled, or to directly satisfy an allocation request
+        otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a small allocation was
+        returned to the arena's bins, whether to flush the relevant tcache if
+        <link linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is
+        enabled, or to directly deallocate an allocation
+        otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        all bin size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nfills">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nfills</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache fills by all small size
+	classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.small.nflushes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.small.nflushes</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache flushes by all small size
+        classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.allocated">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.allocated</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Number of bytes currently allocated by large objects.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent was allocated
+        from the arena, whether to fill the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly satisfy
+        an allocation request otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent was returned
+        to the arena, whether to flush the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly
+        deallocate an allocation otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        all large size classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nfills">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nfills</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache fills by all large size
+	classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.large.nflushes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.large.nflushes</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of tcache flushes by all large size
+        classes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a bin region of the
+        corresponding size class was allocated from the arena, whether to fill
+        the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled, or
+        to directly satisfy an allocation request otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a bin region of the
+        corresponding size class was returned to the arena, whether to flush the
+        relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled, or
+        to directly deallocate an allocation otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        bin regions of the corresponding size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.curregs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curregs</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of regions for this size
+        class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nfills">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nfills</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Cumulative number of tcache fills.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nflushes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nflushes</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+        </term>
+        <listitem><para>Cumulative number of tcache flushes.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nslabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nslabs</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of slabs created.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.nreslabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nreslabs</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times the current slab from which
+        to allocate changed.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.j.curslabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curslabs</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of slabs.</para></listitem>
+      </varlistentry>
+
+
+      <varlistentry id="stats.arenas.i.bins.j.nonfull_slabs">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nonfull_slabs</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of nonfull slabs.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.bins.mutex">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.mutex.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on
+        <varname>arena.&lt;i&gt;.bins.&lt;j&gt;</varname> mutex (arena bin
+        scope; bin operation related).  <mallctl>{counter}</mallctl> is one of
+        the counters in <link linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.extents.n">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.extents.&lt;j&gt;.n{extent_type}</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para> Number of extents of the given type in this arena in
+	the bucket corresponding to page size index &lt;j&gt;. The extent type
+	is one of dirty, muzzy, or retained.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.extents.bytes">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.extents.&lt;j&gt;.{extent_type}_bytes</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+	<listitem><para> Sum of the bytes managed by extents of the given type
+	in this arena in the bucket corresponding to page size index &lt;j&gt;.
+	The extent type is one of dirty, muzzy, or retained.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.nmalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.nmalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent of the
+        corresponding size class was allocated from the arena, whether to fill
+        the relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly satisfy
+        an allocation request otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.ndalloc">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.ndalloc</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of times a large extent of the
+        corresponding size class was returned to the arena, whether to flush the
+        relevant tcache if <link
+        linkend="opt.tcache"><mallctl>opt.tcache</mallctl></link> is enabled and
+        the size class is within the range being cached, or to directly
+        deallocate an allocation otherwise.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.nrequests">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.nrequests</mallctl>
+          (<type>uint64_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Cumulative number of allocation requests satisfied by
+        large extents of the corresponding size class.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.lextents.j.curlextents">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.lextents.&lt;j&gt;.curlextents</mallctl>
+          (<type>size_t</type>)
+          <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Current number of large allocations for this size class.
+        </para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.large">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.large.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.large</varname>
+        mutex (arena scope; large allocation related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extent_avail">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extent_avail.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extent_avail
+        </varname> mutex (arena scope; extent avail related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extents_dirty">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extents_dirty.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extents_dirty
+        </varname> mutex (arena scope; dirty extents related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extents_muzzy">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extents_muzzy.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extents_muzzy
+        </varname> mutex (arena scope; muzzy extents related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.extents_retained">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.extents_retained.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.extents_retained
+        </varname> mutex (arena scope; retained extents related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.decay_dirty">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.decay_dirty.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.decay_dirty
+        </varname> mutex (arena scope; decay for dirty pages related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.decay_muzzy">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.decay_muzzy.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.decay_muzzy
+        </varname> mutex (arena scope; decay for muzzy pages related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.base">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.base.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on <varname>arena.&lt;i&gt;.base</varname>
+        mutex (arena scope; base allocator related).
+        <mallctl>{counter}</mallctl> is one of the counters in <link
+        linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+      <varlistentry id="stats.arenas.i.mutexes.tcache_list">
+        <term>
+          <mallctl>stats.arenas.&lt;i&gt;.mutexes.tcache_list.{counter}</mallctl>
+          (<type>counter specific type</type>) <literal>r-</literal>
+          [<option>--enable-stats</option>]
+        </term>
+        <listitem><para>Statistics on
+        <varname>arena.&lt;i&gt;.tcache_list</varname> mutex (arena scope;
+        tcache to arena association related).  This mutex is expected to be
+        accessed less often.  <mallctl>{counter}</mallctl> is one of the
+        counters in <link linkend="mutex_counters">mutex profiling
+        counters</link>.</para></listitem>
+      </varlistentry>
+
+    </variablelist>
+  </refsect1>
+  <refsect1 id="heap_profile_format">
+    <title>HEAP PROFILE FORMAT</title>
+    <para>Although the heap profiling functionality was originally designed to
+    be compatible with the
+    <command>pprof</command> command that is developed as part of the <ulink
+    url="http://code.google.com/p/gperftools/">gperftools
+    package</ulink>, the addition of per thread heap profiling functionality
+    required a different heap profile format.  The <command>jeprof</command>
+    command is derived from <command>pprof</command>, with enhancements to
+    support the heap profile format described here.</para>
+
+    <para>In the following hypothetical heap profile, <constant>[...]</constant>
+    indicates elision for the sake of compactness.  <programlisting><![CDATA[
+heap_v2/524288
+  t*: 28106: 56637512 [0: 0]
+  [...]
+  t3: 352: 16777344 [0: 0]
+  [...]
+  t99: 17754: 29341640 [0: 0]
+  [...]
+@ 0x5f86da8 0x5f5a1dc [...] 0x29e4d4e 0xa200316 0xabb2988 [...]
+  t*: 13: 6688 [0: 0]
+  t3: 12: 6496 [0: ]
+  t99: 1: 192 [0: 0]
+[...]
+
+MAPPED_LIBRARIES:
+[...]]]></programlisting> The following matches the above heap profile, but most
+tokens are replaced with <constant>&lt;description&gt;</constant> to indicate
+descriptions of the corresponding fields.  <programlisting><![CDATA[
+<heap_profile_format_version>/<mean_sample_interval>
+  <aggregate>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+  [...]
+  <thread_3_aggregate>: <curobjs>: <curbytes>[<cumobjs>: <cumbytes>]
+  [...]
+  <thread_99_aggregate>: <curobjs>: <curbytes>[<cumobjs>: <cumbytes>]
+  [...]
+@ <top_frame> <frame> [...] <frame> <frame> <frame> [...]
+  <backtrace_aggregate>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+  <backtrace_thread_3>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+  <backtrace_thread_99>: <curobjs>: <curbytes> [<cumobjs>: <cumbytes>]
+[...]
+
+MAPPED_LIBRARIES:
+</proc/<pid>/maps>]]></programlisting></para>
+  </refsect1>
+
+  <refsect1 id="debugging_malloc_problems">
+    <title>DEBUGGING MALLOC PROBLEMS</title>
+    <para>When debugging, it is a good idea to configure/build jemalloc with
+    the <option>--enable-debug</option> and <option>--enable-fill</option>
+    options, and recompile the program with suitable options and symbols for
+    debugger support.  When so configured, jemalloc incorporates a wide variety
+    of run-time assertions that catch application errors such as double-free,
+    write-after-free, etc.</para>
+
+    <para>Programs often accidentally depend on <quote>uninitialized</quote>
+    memory actually being filled with zero bytes.  Junk filling
+    (see the <link linkend="opt.junk"><mallctl>opt.junk</mallctl></link>
+    option) tends to expose such bugs in the form of obviously incorrect
+    results and/or coredumps.  Conversely, zero
+    filling (see the <link
+    linkend="opt.zero"><mallctl>opt.zero</mallctl></link> option) eliminates
+    the symptoms of such bugs.  Between these two options, it is usually
+    possible to quickly detect, diagnose, and eliminate such bugs.</para>
+
+    <para>This implementation does not provide much detail about the problems
+    it detects, because the performance impact for storing such information
+    would be prohibitive.</para>
+  </refsect1>
+  <refsect1 id="diagnostic_messages">
+    <title>DIAGNOSTIC MESSAGES</title>
+    <para>If any of the memory allocation/deallocation functions detect an
+    error or warning condition, a message will be printed to file descriptor
+    <constant>STDERR_FILENO</constant>.  Errors will result in the process
+    dumping core.  If the <link
+    linkend="opt.abort"><mallctl>opt.abort</mallctl></link> option is set, most
+    warnings are treated as errors.</para>
+
+    <para>The <varname>malloc_message</varname> variable allows the programmer
+    to override the function which emits the text strings forming the errors
+    and warnings if for some reason the <constant>STDERR_FILENO</constant> file
+    descriptor is not suitable for this.
+    <function>malloc_message()</function> takes the
+    <parameter>cbopaque</parameter> pointer argument that is
+    <constant>NULL</constant> unless overridden by the arguments in a call to
+    <function>malloc_stats_print()</function>, followed by a string
+    pointer.  Please note that doing anything which tries to allocate memory in
+    this function is likely to result in a crash or deadlock.</para>
+
+    <para>All messages are prefixed by
+    <quote><computeroutput>&lt;jemalloc&gt;: </computeroutput></quote>.</para>
+  </refsect1>
+  <refsect1 id="return_values">
+    <title>RETURN VALUES</title>
+    <refsect2>
+      <title>Standard API</title>
+      <para>The <function>malloc()</function> and
+      <function>calloc()</function> functions return a pointer to the
+      allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned and <varname>errno</varname> is set to
+      <errorname>ENOMEM</errorname>.</para>
+
+      <para>The <function>posix_memalign()</function> function
+      returns the value 0 if successful; otherwise it returns an error value.
+      The <function>posix_memalign()</function> function will fail
+      if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para>The <parameter>alignment</parameter> parameter is
+            not a power of 2 at least as large as
+            <code language="C">sizeof(<type>void *</type>)</code>.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOMEM</errorname></term>
+
+            <listitem><para>Memory allocation error.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>aligned_alloc()</function> function returns
+      a pointer to the allocated memory if successful; otherwise a
+      <constant>NULL</constant> pointer is returned and
+      <varname>errno</varname> is set.  The
+      <function>aligned_alloc()</function> function will fail if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para>The <parameter>alignment</parameter> parameter is
+            not a power of 2.
+            </para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOMEM</errorname></term>
+
+            <listitem><para>Memory allocation error.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>realloc()</function> function returns a
+      pointer, possibly identical to <parameter>ptr</parameter>, to the
+      allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned, and <varname>errno</varname> is set to
+      <errorname>ENOMEM</errorname> if the error was the result of an
+      allocation failure.  The <function>realloc()</function>
+      function always leaves the original buffer intact when an error occurs.
+      </para>
+
+      <para>The <function>free()</function> function returns no
+      value.</para>
+    </refsect2>
+    <refsect2>
+      <title>Non-standard API</title>
+      <para>The <function>mallocx()</function> and
+      <function>rallocx()</function> functions return a pointer to
+      the allocated memory if successful; otherwise a <constant>NULL</constant>
+      pointer is returned to indicate insufficient contiguous memory was
+      available to service the allocation request.  </para>
+
+      <para>The <function>xallocx()</function> function returns the
+      real size of the resulting resized allocation pointed to by
+      <parameter>ptr</parameter>, which is a value less than
+      <parameter>size</parameter> if the allocation could not be adequately
+      grown in place.  </para>
+
+      <para>The <function>sallocx()</function> function returns the
+      real size of the allocation pointed to by <parameter>ptr</parameter>.
+      </para>
+
+      <para>The <function>nallocx()</function> returns the real size
+      that would result from a successful equivalent
+      <function>mallocx()</function> function call, or zero if
+      insufficient memory is available to perform the size computation.  </para>
+
+      <para>The <function>mallctl()</function>,
+      <function>mallctlnametomib()</function>, and
+      <function>mallctlbymib()</function> functions return 0 on
+      success; otherwise they return an error value.  The functions will fail
+      if:
+        <variablelist>
+          <varlistentry>
+            <term><errorname>EINVAL</errorname></term>
+
+            <listitem><para><parameter>newp</parameter> is not
+            <constant>NULL</constant>, and <parameter>newlen</parameter> is too
+            large or too small.  Alternatively, <parameter>*oldlenp</parameter>
+            is too large or too small; in this case as much data as possible
+            are read despite the error.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>ENOENT</errorname></term>
+
+            <listitem><para><parameter>name</parameter> or
+            <parameter>mib</parameter> specifies an unknown/invalid
+            value.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EPERM</errorname></term>
+
+            <listitem><para>Attempt to read or write void value, or attempt to
+            write read-only value.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EAGAIN</errorname></term>
+
+            <listitem><para>A memory allocation failure
+            occurred.</para></listitem>
+          </varlistentry>
+          <varlistentry>
+            <term><errorname>EFAULT</errorname></term>
+
+            <listitem><para>An interface with side effects failed in some way
+            not directly related to <function>mallctl*()</function>
+            read/write processing.</para></listitem>
+          </varlistentry>
+        </variablelist>
+      </para>
+
+      <para>The <function>malloc_usable_size()</function> function
+      returns the usable size of the allocation pointed to by
+      <parameter>ptr</parameter>.  </para>
+    </refsect2>
+  </refsect1>
+  <refsect1 id="environment">
+    <title>ENVIRONMENT</title>
+    <para>The following environment variable affects the execution of the
+    allocation functions:
+      <variablelist>
+        <varlistentry>
+          <term><envar>MALLOC_CONF</envar></term>
+
+          <listitem><para>If the environment variable
+          <envar>MALLOC_CONF</envar> is set, the characters it contains
+          will be interpreted as options.</para></listitem>
+        </varlistentry>
+      </variablelist>
+    </para>
+  </refsect1>
+  <refsect1 id="examples">
+    <title>EXAMPLES</title>
+    <para>To dump core whenever a problem occurs:
+      <screen>ln -s 'abort:true' /etc/malloc.conf</screen>
+    </para>
+    <para>To specify in the source that only one arena should be automatically
+    created:
+      <programlisting language="C"><![CDATA[
+malloc_conf = "narenas:1";]]></programlisting></para>
+  </refsect1>
+  <refsect1 id="see_also">
+    <title>SEE ALSO</title>
+    <para><citerefentry><refentrytitle>madvise</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>mmap</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>sbrk</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>utrace</refentrytitle>
+    <manvolnum>2</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>alloca</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>atexit</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry>,
+    <citerefentry><refentrytitle>getpagesize</refentrytitle>
+    <manvolnum>3</manvolnum></citerefentry></para>
+  </refsect1>
+  <refsect1 id="standards">
+    <title>STANDARDS</title>
+    <para>The <function>malloc()</function>,
+    <function>calloc()</function>,
+    <function>realloc()</function>, and
+    <function>free()</function> functions conform to ISO/IEC
+    9899:1990 (<quote>ISO C90</quote>).</para>
+
+    <para>The <function>posix_memalign()</function> function conforms
+    to IEEE Std 1003.1-2001 (<quote>POSIX.1</quote>).</para>
+  </refsect1>
+</refentry>
diff --git a/memkind-1.10.0/jemalloc/doc/manpages.xsl.in b/memkind-1.10.0/jemalloc/doc/manpages.xsl.in
new file mode 100644
index 0000000..88b2626
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/doc/manpages.xsl.in
@@ -0,0 +1,4 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:import href="@XSLROOT@/manpages/docbook.xsl"/>
+  <xsl:import href="@abs_srcroot@doc/stylesheet.xsl"/>
+</xsl:stylesheet>
diff --git a/memkind-1.10.0/jemalloc/doc/stylesheet.xsl b/memkind-1.10.0/jemalloc/doc/stylesheet.xsl
new file mode 100644
index 0000000..619365d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/doc/stylesheet.xsl
@@ -0,0 +1,10 @@
+<xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
+  <xsl:param name="funcsynopsis.style">ansi</xsl:param>
+  <xsl:param name="function.parens" select="0"/>
+  <xsl:template match="function">
+    <xsl:call-template name="inline.monoseq"/>
+  </xsl:template>
+  <xsl:template match="mallctl">
+    <quote><xsl:call-template name="inline.monoseq"/></quote>
+  </xsl:template>
+</xsl:stylesheet>
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_externs.h
new file mode 100644
index 0000000..a4523ae
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_externs.h
@@ -0,0 +1,104 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_EXTERNS_H
+#define JEMALLOC_INTERNAL_ARENA_EXTERNS_H
+
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/stats.h"
+
+extern ssize_t opt_dirty_decay_ms;
+extern ssize_t opt_muzzy_decay_ms;
+
+extern percpu_arena_mode_t opt_percpu_arena;
+extern const char *percpu_arena_mode_names[];
+
+extern const uint64_t h_steps[SMOOTHSTEP_NSTEPS];
+extern malloc_mutex_t arenas_lock;
+
+extern size_t opt_oversize_threshold;
+extern size_t oversize_threshold;
+
+void arena_basic_stats_merge(tsdn_t *tsdn, arena_t *arena,
+    unsigned *nthreads, const char **dss, ssize_t *dirty_decay_ms,
+    ssize_t *muzzy_decay_ms, size_t *nactive, size_t *ndirty, size_t *nmuzzy);
+void arena_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
+    const char **dss, ssize_t *dirty_decay_ms, ssize_t *muzzy_decay_ms,
+    size_t *nactive, size_t *ndirty, size_t *nmuzzy, arena_stats_t *astats,
+    bin_stats_t *bstats, arena_stats_large_t *lstats,
+    arena_stats_extents_t *estats);
+void arena_extents_dirty_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+#ifdef JEMALLOC_JET
+size_t arena_slab_regind(extent_t *slab, szind_t binind, const void *ptr);
+#endif
+extent_t *arena_extent_alloc_large(tsdn_t *tsdn, arena_t *arena,
+    size_t usize, size_t alignment, bool *zero);
+void arena_extent_dalloc_large_prep(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent);
+void arena_extent_ralloc_large_shrink(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent, size_t oldsize);
+void arena_extent_ralloc_large_expand(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent, size_t oldsize);
+ssize_t arena_dirty_decay_ms_get(arena_t *arena);
+bool arena_dirty_decay_ms_set(tsdn_t *tsdn, arena_t *arena, ssize_t decay_ms);
+ssize_t arena_muzzy_decay_ms_get(arena_t *arena);
+bool arena_muzzy_decay_ms_set(tsdn_t *tsdn, arena_t *arena, ssize_t decay_ms);
+void arena_decay(tsdn_t *tsdn, arena_t *arena, bool is_background_thread,
+    bool all);
+void arena_reset(tsd_t *tsd, arena_t *arena);
+void arena_destroy(tsd_t *tsd, arena_t *arena);
+void arena_tcache_fill_small(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, uint64_t prof_accumbytes);
+void arena_alloc_junk_small(void *ptr, const bin_info_t *bin_info,
+    bool zero);
+
+typedef void (arena_dalloc_junk_small_t)(void *, const bin_info_t *);
+extern arena_dalloc_junk_small_t *JET_MUTABLE arena_dalloc_junk_small;
+
+void *arena_malloc_hard(tsdn_t *tsdn, arena_t *arena, size_t size,
+    szind_t ind, bool zero);
+void *arena_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache);
+void arena_prof_promote(tsdn_t *tsdn, void *ptr, size_t usize);
+void arena_dalloc_promoted(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    bool slow_path);
+void arena_dalloc_bin_junked_locked(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, extent_t *extent, void *ptr);
+void arena_dalloc_small(tsdn_t *tsdn, void *ptr);
+bool arena_ralloc_no_move(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero, size_t *newsize);
+void *arena_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t oldsize,
+    size_t size, size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args);
+dss_prec_t arena_dss_prec_get(arena_t *arena);
+bool arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec);
+ssize_t arena_dirty_decay_ms_default_get(void);
+bool arena_dirty_decay_ms_default_set(ssize_t decay_ms);
+ssize_t arena_muzzy_decay_ms_default_get(void);
+bool arena_muzzy_decay_ms_default_set(ssize_t decay_ms);
+bool arena_retain_grow_limit_get_set(tsd_t *tsd, arena_t *arena,
+    size_t *old_limit, size_t *new_limit);
+unsigned arena_nthreads_get(arena_t *arena, bool internal);
+void arena_nthreads_inc(arena_t *arena, bool internal);
+void arena_nthreads_dec(arena_t *arena, bool internal);
+size_t arena_extent_sn_next(arena_t *arena);
+arena_t *arena_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+bool arena_init_huge(void);
+bool arena_is_huge(unsigned arena_ind);
+arena_t *arena_choose_huge(tsd_t *tsd);
+bin_t *arena_bin_choose_lock(tsdn_t *tsdn, arena_t *arena, szind_t binind,
+    unsigned *binshard);
+void arena_boot(sc_data_t *sc_data);
+void arena_prefork0(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork1(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork2(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork3(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork4(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork5(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork6(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork7(tsdn_t *tsdn, arena_t *arena);
+void arena_postfork_parent(tsdn_t *tsdn, arena_t *arena);
+void arena_postfork_child(tsdn_t *tsdn, arena_t *arena);
+
+#endif /* JEMALLOC_INTERNAL_ARENA_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_inlines_a.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_inlines_a.h
new file mode 100644
index 0000000..9abf7f6
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_inlines_a.h
@@ -0,0 +1,57 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_INLINES_A_H
+#define JEMALLOC_INTERNAL_ARENA_INLINES_A_H
+
+static inline unsigned
+arena_ind_get(const arena_t *arena) {
+	return base_ind_get(arena->base);
+}
+
+static inline void
+arena_internal_add(arena_t *arena, size_t size) {
+	atomic_fetch_add_zu(&arena->stats.internal, size, ATOMIC_RELAXED);
+}
+
+static inline void
+arena_internal_sub(arena_t *arena, size_t size) {
+	atomic_fetch_sub_zu(&arena->stats.internal, size, ATOMIC_RELAXED);
+}
+
+static inline size_t
+arena_internal_get(arena_t *arena) {
+	return atomic_load_zu(&arena->stats.internal, ATOMIC_RELAXED);
+}
+
+static inline bool
+arena_prof_accum(tsdn_t *tsdn, arena_t *arena, uint64_t accumbytes) {
+	cassert(config_prof);
+
+	if (likely(prof_interval == 0 || !prof_active_get_unlocked())) {
+		return false;
+	}
+
+	return prof_accum_add(tsdn, &arena->prof_accum, accumbytes);
+}
+
+static inline void
+percpu_arena_update(tsd_t *tsd, unsigned cpu) {
+	assert(have_percpu_arena);
+	arena_t *oldarena = tsd_arena_get(tsd);
+	assert(oldarena != NULL);
+	unsigned oldind = arena_ind_get(oldarena);
+
+	if (oldind != cpu) {
+		unsigned newind = cpu;
+		arena_t *newarena = arena_get(tsd_tsdn(tsd), newind, true);
+		assert(newarena != NULL);
+
+		/* Set new arena/tcache associations. */
+		arena_migrate(tsd, oldind, newind);
+		tcache_t *tcache = tcache_get(tsd);
+		if (tcache != NULL) {
+			tcache_arena_reassociate(tsd_tsdn(tsd), tcache,
+			    newarena);
+		}
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_INLINES_A_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_inlines_b.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_inlines_b.h
new file mode 100644
index 0000000..dd92657
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_inlines_b.h
@@ -0,0 +1,427 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_INLINES_B_H
+#define JEMALLOC_INTERNAL_ARENA_INLINES_B_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/ticker.h"
+
+JEMALLOC_ALWAYS_INLINE bool
+arena_has_default_hooks(arena_t *arena) {
+	return (extent_hooks_get(arena) == &extent_hooks_default);
+}
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+arena_choose_maybe_huge(tsd_t *tsd, arena_t *arena, size_t size) {
+	if (arena != NULL) {
+		return arena;
+	}
+
+	/*
+	 * For huge allocations, use the dedicated huge arena if both are true:
+	 * 1) is using auto arena selection (i.e. arena == NULL), and 2) the
+	 * thread is not assigned to a manual arena.
+	 */
+	if (unlikely(size >= oversize_threshold)) {
+		arena_t *tsd_arena = tsd_arena_get(tsd);
+		if (tsd_arena == NULL || arena_is_auto(tsd_arena)) {
+			return arena_choose_huge(tsd);
+		}
+	}
+
+	return arena_choose(tsd, NULL);
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+arena_prof_tctx_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	/* Static check. */
+	if (alloc_ctx == NULL) {
+		const extent_t *extent = iealloc(tsdn, ptr);
+		if (unlikely(!extent_slab_get(extent))) {
+			return large_prof_tctx_get(tsdn, extent);
+		}
+	} else {
+		if (unlikely(!alloc_ctx->slab)) {
+			return large_prof_tctx_get(tsdn, iealloc(tsdn, ptr));
+		}
+	}
+	return (prof_tctx_t *)(uintptr_t)1U;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_prof_tctx_set(tsdn_t *tsdn, const void *ptr, size_t usize,
+    alloc_ctx_t *alloc_ctx, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	/* Static check. */
+	if (alloc_ctx == NULL) {
+		extent_t *extent = iealloc(tsdn, ptr);
+		if (unlikely(!extent_slab_get(extent))) {
+			large_prof_tctx_set(tsdn, extent, tctx);
+		}
+	} else {
+		if (unlikely(!alloc_ctx->slab)) {
+			large_prof_tctx_set(tsdn, iealloc(tsdn, ptr), tctx);
+		}
+	}
+}
+
+static inline void
+arena_prof_tctx_reset(tsdn_t *tsdn, const void *ptr, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	assert(!extent_slab_get(extent));
+
+	large_prof_tctx_reset(tsdn, extent);
+}
+
+JEMALLOC_ALWAYS_INLINE nstime_t
+arena_prof_alloc_time_get(tsdn_t *tsdn, const void *ptr,
+    alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	/*
+	 * Unlike arena_prof_prof_tctx_{get, set}, we only call this once we're
+	 * sure we have a sampled allocation.
+	 */
+	assert(!extent_slab_get(extent));
+	return large_prof_alloc_time_get(extent);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_prof_alloc_time_set(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx,
+    nstime_t t) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	assert(!extent_slab_get(extent));
+	large_prof_alloc_time_set(extent, t);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_ticks(tsdn_t *tsdn, arena_t *arena, unsigned nticks) {
+	tsd_t *tsd;
+	ticker_t *decay_ticker;
+
+	if (unlikely(tsdn_null(tsdn))) {
+		return;
+	}
+	tsd = tsdn_tsd(tsdn);
+	decay_ticker = decay_ticker_get(tsd, arena_ind_get(arena));
+	if (unlikely(decay_ticker == NULL)) {
+		return;
+	}
+	if (unlikely(ticker_ticks(decay_ticker, nticks))) {
+		arena_decay(tsdn, arena, false, false);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_tick(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_assert_not_owner(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_assert_not_owner(tsdn, &arena->decay_muzzy.mtx);
+
+	arena_decay_ticks(tsdn, arena, 1);
+}
+
+/* Purge a single extent to retained / unmapped directly. */
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_extent(tsdn_t *tsdn,arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extent_t *extent) {
+	size_t extent_size = extent_size_get(extent);
+	extent_dalloc_wrapper(tsdn, arena,
+	    r_extent_hooks, extent);
+	if (config_stats) {
+		/* Update stats accordingly. */
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_stats_add_u64(tsdn, &arena->stats,
+		    &arena->decay_dirty.stats->nmadvise, 1);
+		arena_stats_add_u64(tsdn, &arena->stats,
+		    &arena->decay_dirty.stats->purged, extent_size >> LG_PAGE);
+		arena_stats_sub_zu(tsdn, &arena->stats, &arena->stats.mapped,
+		    extent_size);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+arena_malloc(tsdn_t *tsdn, arena_t *arena, size_t size, szind_t ind, bool zero,
+    tcache_t *tcache, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+
+	if (likely(tcache != NULL)) {
+		if (likely(size <= SC_SMALL_MAXCLASS)) {
+			return tcache_alloc_small(tsdn_tsd(tsdn), arena,
+			    tcache, size, ind, zero, slow_path);
+		}
+		if (likely(size <= tcache_maxclass)) {
+			return tcache_alloc_large(tsdn_tsd(tsdn), arena,
+			    tcache, size, ind, zero, slow_path);
+		}
+		/* (size > tcache_maxclass) case falls through. */
+		assert(size > tcache_maxclass);
+	}
+
+	return arena_malloc_hard(tsdn, arena, size, ind, zero);
+}
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+arena_aalloc(tsdn_t *tsdn, const void *ptr) {
+	return extent_arena_get(iealloc(tsdn, ptr));
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_salloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	szind_t szind = rtree_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+	assert(szind != SC_NSIZES);
+
+	return sz_index2size(szind);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_vsalloc(tsdn_t *tsdn, const void *ptr) {
+	/*
+	 * Return 0 if ptr is not within an extent managed by jemalloc.  This
+	 * function has two extra costs relative to isalloc():
+	 * - The rtree calls cannot claim to be dependent lookups, which induces
+	 *   rtree lookup load dependencies.
+	 * - The lookup may fail, so there is an extra branch to check for
+	 *   failure.
+	 */
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent;
+	szind_t szind;
+	if (rtree_extent_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, false, &extent, &szind)) {
+		return 0;
+	}
+
+	if (extent == NULL) {
+		return 0;
+	}
+	assert(extent_state_get(extent) == extent_state_active);
+	/* Only slab members should be looked up via interior pointers. */
+	assert(extent_addr_get(extent) == ptr || extent_slab_get(extent));
+
+	assert(szind != SC_NSIZES);
+
+	return sz_index2size(szind);
+}
+
+static inline void
+arena_dalloc_large_no_tcache(tsdn_t *tsdn, void *ptr, szind_t szind) {
+	if (config_prof && unlikely(szind < SC_NBINS)) {
+		arena_dalloc_promoted(tsdn, ptr, NULL, true);
+	} else {
+		extent_t *extent = iealloc(tsdn, ptr);
+		large_dalloc(tsdn, extent);
+	}
+}
+
+static inline void
+arena_dalloc_no_tcache(tsdn_t *tsdn, void *ptr) {
+	assert(ptr != NULL);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	szind_t szind;
+	bool slab;
+	rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
+	    true, &szind, &slab);
+
+	if (config_debug) {
+		extent_t *extent = rtree_extent_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(szind < SC_NSIZES);
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		arena_dalloc_small(tsdn, ptr);
+	} else {
+		arena_dalloc_large_no_tcache(tsdn, ptr, szind);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_dalloc_large(tsdn_t *tsdn, void *ptr, tcache_t *tcache, szind_t szind,
+    bool slow_path) {
+	if (szind < nhbins) {
+		if (config_prof && unlikely(szind < SC_NBINS)) {
+			arena_dalloc_promoted(tsdn, ptr, tcache, slow_path);
+		} else {
+			tcache_dalloc_large(tsdn_tsd(tsdn), tcache, ptr, szind,
+			    slow_path);
+		}
+	} else {
+		extent_t *extent = iealloc(tsdn, ptr);
+		large_dalloc(tsdn, extent);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_dalloc(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+	assert(ptr != NULL);
+
+	if (unlikely(tcache == NULL)) {
+		arena_dalloc_no_tcache(tsdn, ptr);
+		return;
+	}
+
+	szind_t szind;
+	bool slab;
+	rtree_ctx_t *rtree_ctx;
+	if (alloc_ctx != NULL) {
+		szind = alloc_ctx->szind;
+		slab = alloc_ctx->slab;
+		assert(szind != SC_NSIZES);
+	} else {
+		rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+	}
+
+	if (config_debug) {
+		rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		extent_t *extent = rtree_extent_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(szind < SC_NSIZES);
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		tcache_dalloc_small(tsdn_tsd(tsdn), tcache, ptr, szind,
+		    slow_path);
+	} else {
+		arena_dalloc_large(tsdn, ptr, tcache, szind, slow_path);
+	}
+}
+
+static inline void
+arena_sdalloc_no_tcache(tsdn_t *tsdn, void *ptr, size_t size) {
+	assert(ptr != NULL);
+	assert(size <= SC_LARGE_MAXCLASS);
+
+	szind_t szind;
+	bool slab;
+	if (!config_prof || !opt_prof) {
+		/*
+		 * There is no risk of being confused by a promoted sampled
+		 * object, so base szind and slab on the given size.
+		 */
+		szind = sz_size2index(size);
+		slab = (szind < SC_NBINS);
+	}
+
+	if ((config_prof && opt_prof) || config_debug) {
+		rtree_ctx_t rtree_ctx_fallback;
+		rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+		    &rtree_ctx_fallback);
+
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+
+		assert(szind == sz_size2index(size));
+		assert((config_prof && opt_prof) || slab == (szind < SC_NBINS));
+
+		if (config_debug) {
+			extent_t *extent = rtree_extent_read(tsdn,
+			    &extents_rtree, rtree_ctx, (uintptr_t)ptr, true);
+			assert(szind == extent_szind_get(extent));
+			assert(slab == extent_slab_get(extent));
+		}
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		arena_dalloc_small(tsdn, ptr);
+	} else {
+		arena_dalloc_large_no_tcache(tsdn, ptr, szind);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_sdalloc(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+	assert(ptr != NULL);
+	assert(size <= SC_LARGE_MAXCLASS);
+
+	if (unlikely(tcache == NULL)) {
+		arena_sdalloc_no_tcache(tsdn, ptr, size);
+		return;
+	}
+
+	szind_t szind;
+	bool slab;
+	alloc_ctx_t local_ctx;
+	if (config_prof && opt_prof) {
+		if (alloc_ctx == NULL) {
+			/* Uncommon case and should be a static check. */
+			rtree_ctx_t rtree_ctx_fallback;
+			rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+			    &rtree_ctx_fallback);
+			rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+			    (uintptr_t)ptr, true, &local_ctx.szind,
+			    &local_ctx.slab);
+			assert(local_ctx.szind == sz_size2index(size));
+			alloc_ctx = &local_ctx;
+		}
+		slab = alloc_ctx->slab;
+		szind = alloc_ctx->szind;
+	} else {
+		/*
+		 * There is no risk of being confused by a promoted sampled
+		 * object, so base szind and slab on the given size.
+		 */
+		szind = sz_size2index(size);
+		slab = (szind < SC_NBINS);
+	}
+
+	if (config_debug) {
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+		extent_t *extent = rtree_extent_read(tsdn,
+		    &extents_rtree, rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		tcache_dalloc_small(tsdn_tsd(tsdn), tcache, ptr, szind,
+		    slow_path);
+	} else {
+		arena_dalloc_large(tsdn, ptr, tcache, szind, slow_path);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_INLINES_B_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_stats.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_stats.h
new file mode 100644
index 0000000..23949ed
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_stats.h
@@ -0,0 +1,271 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STATS_H
+#define JEMALLOC_INTERNAL_ARENA_STATS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/sc.h"
+
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+/*
+ * In those architectures that support 64-bit atomics, we use atomic updates for
+ * our 64-bit values.  Otherwise, we use a plain uint64_t and synchronize
+ * externally.
+ */
+#ifdef JEMALLOC_ATOMIC_U64
+typedef atomic_u64_t arena_stats_u64_t;
+#else
+/* Must hold the arena stats mutex while reading atomically. */
+typedef uint64_t arena_stats_u64_t;
+#endif
+
+typedef struct arena_stats_large_s arena_stats_large_t;
+struct arena_stats_large_s {
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the arena.
+	 */
+	arena_stats_u64_t	nmalloc;
+	arena_stats_u64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to this size class.
+	 * This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	arena_stats_u64_t	nrequests; /* Partially derived. */
+	/*
+	 * Number of tcache fills / flushes for large (similarly, periodically
+	 * merged).  Note that there is no large tcache batch-fill currently
+	 * (i.e. only fill 1 at a time); however flush may be batched.
+	 */
+	arena_stats_u64_t	nfills; /* Partially derived. */
+	arena_stats_u64_t	nflushes; /* Partially derived. */
+
+	/* Current number of allocations of this size class. */
+	size_t		curlextents; /* Derived. */
+};
+
+typedef struct arena_stats_decay_s arena_stats_decay_t;
+struct arena_stats_decay_s {
+	/* Total number of purge sweeps. */
+	arena_stats_u64_t	npurge;
+	/* Total number of madvise calls made. */
+	arena_stats_u64_t	nmadvise;
+	/* Total number of pages purged. */
+	arena_stats_u64_t	purged;
+};
+
+typedef struct arena_stats_extents_s arena_stats_extents_t;
+struct arena_stats_extents_s {
+	/*
+	 * Stats for a given index in the range [0, SC_NPSIZES] in an extents_t.
+	 * We track both bytes and # of extents: two extents in the same bucket
+	 * may have different sizes if adjacent size classes differ by more than
+	 * a page, so bytes cannot always be derived from # of extents.
+	 */
+	atomic_zu_t ndirty;
+	atomic_zu_t dirty_bytes;
+	atomic_zu_t nmuzzy;
+	atomic_zu_t muzzy_bytes;
+	atomic_zu_t nretained;
+	atomic_zu_t retained_bytes;
+};
+
+/*
+ * Arena stats.  Note that fields marked "derived" are not directly maintained
+ * within the arena code; rather their values are derived during stats merge
+ * requests.
+ */
+typedef struct arena_stats_s arena_stats_t;
+struct arena_stats_s {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_t		mtx;
+#endif
+
+	/* Number of bytes currently mapped, excluding retained memory. */
+	atomic_zu_t		mapped; /* Partially derived. */
+
+	/*
+	 * Number of unused virtual memory bytes currently retained.  Retained
+	 * bytes are technically mapped (though always decommitted or purged),
+	 * but they are excluded from the mapped statistic (above).
+	 */
+	atomic_zu_t		retained; /* Derived. */
+
+	/* Number of extent_t structs allocated by base, but not being used. */
+	atomic_zu_t		extent_avail;
+
+	arena_stats_decay_t	decay_dirty;
+	arena_stats_decay_t	decay_muzzy;
+
+	atomic_zu_t		base; /* Derived. */
+	atomic_zu_t		internal;
+	atomic_zu_t		resident; /* Derived. */
+	atomic_zu_t		metadata_thp;
+
+	atomic_zu_t		allocated_large; /* Derived. */
+	arena_stats_u64_t	nmalloc_large; /* Derived. */
+	arena_stats_u64_t	ndalloc_large; /* Derived. */
+	arena_stats_u64_t	nfills_large; /* Derived. */
+	arena_stats_u64_t	nflushes_large; /* Derived. */
+	arena_stats_u64_t	nrequests_large; /* Derived. */
+
+	/* VM space had to be leaked (undocumented).  Normally 0. */
+	atomic_zu_t		abandoned_vm;
+
+	/* Number of bytes cached in tcache associated with this arena. */
+	atomic_zu_t		tcache_bytes; /* Derived. */
+
+	mutex_prof_data_t mutex_prof_data[mutex_prof_num_arena_mutexes];
+
+	/* One element for each large size class. */
+	arena_stats_large_t	lstats[SC_NSIZES - SC_NBINS];
+
+	/* Arena uptime. */
+	nstime_t		uptime;
+};
+
+static inline bool
+arena_stats_init(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+	if (config_debug) {
+		for (size_t i = 0; i < sizeof(arena_stats_t); i++) {
+			assert(((char *)arena_stats)[i] == 0);
+		}
+	}
+#ifndef JEMALLOC_ATOMIC_U64
+	if (malloc_mutex_init(&arena_stats->mtx, "arena_stats",
+	    WITNESS_RANK_ARENA_STATS, malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+#endif
+	/* Memory is zeroed, so there is no need to clear stats. */
+	return false;
+}
+
+static inline void
+arena_stats_lock(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_lock(tsdn, &arena_stats->mtx);
+#endif
+}
+
+static inline void
+arena_stats_unlock(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_unlock(tsdn, &arena_stats->mtx);
+#endif
+}
+
+static inline uint64_t
+arena_stats_read_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_u64(p, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	return *p;
+#endif
+}
+
+static inline void
+arena_stats_add_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p, uint64_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	atomic_fetch_add_u64(p, x, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	*p += x;
+#endif
+}
+
+static inline void
+arena_stats_sub_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p, uint64_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	uint64_t r = atomic_fetch_sub_u64(p, x, ATOMIC_RELAXED);
+	assert(r - x <= r);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	*p -= x;
+	assert(*p + x >= *p);
+#endif
+}
+
+/*
+ * Non-atomically sets *dst += src.  *dst needs external synchronization.
+ * This lets us avoid the cost of a fetch_add when its unnecessary (note that
+ * the types here are atomic).
+ */
+static inline void
+arena_stats_accum_u64(arena_stats_u64_t *dst, uint64_t src) {
+#ifdef JEMALLOC_ATOMIC_U64
+	uint64_t cur_dst = atomic_load_u64(dst, ATOMIC_RELAXED);
+	atomic_store_u64(dst, src + cur_dst, ATOMIC_RELAXED);
+#else
+	*dst += src;
+#endif
+}
+
+static inline size_t
+arena_stats_read_zu(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    atomic_zu_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_zu(p, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	return atomic_load_zu(p, ATOMIC_RELAXED);
+#endif
+}
+
+static inline void
+arena_stats_add_zu(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    atomic_zu_t *p, size_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	atomic_fetch_add_zu(p, x, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	size_t cur = atomic_load_zu(p, ATOMIC_RELAXED);
+	atomic_store_zu(p, cur + x, ATOMIC_RELAXED);
+#endif
+}
+
+static inline void
+arena_stats_sub_zu(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    atomic_zu_t *p, size_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	size_t r = atomic_fetch_sub_zu(p, x, ATOMIC_RELAXED);
+	assert(r - x <= r);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	size_t cur = atomic_load_zu(p, ATOMIC_RELAXED);
+	atomic_store_zu(p, cur - x, ATOMIC_RELAXED);
+#endif
+}
+
+/* Like the _u64 variant, needs an externally synchronized *dst. */
+static inline void
+arena_stats_accum_zu(atomic_zu_t *dst, size_t src) {
+	size_t cur_dst = atomic_load_zu(dst, ATOMIC_RELAXED);
+	atomic_store_zu(dst, src + cur_dst, ATOMIC_RELAXED);
+}
+
+static inline void
+arena_stats_large_flush_nrequests_add(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    szind_t szind, uint64_t nrequests) {
+	arena_stats_lock(tsdn, arena_stats);
+	arena_stats_large_t *lstats = &arena_stats->lstats[szind - SC_NBINS];
+	arena_stats_add_u64(tsdn, arena_stats, &lstats->nrequests, nrequests);
+	arena_stats_add_u64(tsdn, arena_stats, &lstats->nflushes, 1);
+	arena_stats_unlock(tsdn, arena_stats);
+}
+
+static inline void
+arena_stats_mapped_add(tsdn_t *tsdn, arena_stats_t *arena_stats, size_t size) {
+	arena_stats_lock(tsdn, arena_stats);
+	arena_stats_add_zu(tsdn, arena_stats, &arena_stats->mapped, size);
+	arena_stats_unlock(tsdn, arena_stats);
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STATS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_structs_a.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_structs_a.h
new file mode 100644
index 0000000..46aa77c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_structs_a.h
@@ -0,0 +1,11 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H
+#define JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H
+
+#include "jemalloc/internal/bitmap.h"
+
+struct arena_slab_data_s {
+	/* Per region allocated/deallocated bitmap. */
+	bitmap_t	bitmap[BITMAP_GROUPS_MAX];
+};
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_structs_b.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_structs_b.h
new file mode 100644
index 0000000..eeab57f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_structs_b.h
@@ -0,0 +1,232 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H
+#define JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H
+
+#include "jemalloc/internal/arena_stats.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/smoothstep.h"
+#include "jemalloc/internal/ticker.h"
+
+struct arena_decay_s {
+	/* Synchronizes all non-atomic fields. */
+	malloc_mutex_t		mtx;
+	/*
+	 * True if a thread is currently purging the extents associated with
+	 * this decay structure.
+	 */
+	bool			purging;
+	/*
+	 * Approximate time in milliseconds from the creation of a set of unused
+	 * dirty pages until an equivalent set of unused dirty pages is purged
+	 * and/or reused.
+	 */
+	atomic_zd_t		time_ms;
+	/* time / SMOOTHSTEP_NSTEPS. */
+	nstime_t		interval;
+	/*
+	 * Time at which the current decay interval logically started.  We do
+	 * not actually advance to a new epoch until sometime after it starts
+	 * because of scheduling and computation delays, and it is even possible
+	 * to completely skip epochs.  In all cases, during epoch advancement we
+	 * merge all relevant activity into the most recently recorded epoch.
+	 */
+	nstime_t		epoch;
+	/* Deadline randomness generator. */
+	uint64_t		jitter_state;
+	/*
+	 * Deadline for current epoch.  This is the sum of interval and per
+	 * epoch jitter which is a uniform random variable in [0..interval).
+	 * Epochs always advance by precise multiples of interval, but we
+	 * randomize the deadline to reduce the likelihood of arenas purging in
+	 * lockstep.
+	 */
+	nstime_t		deadline;
+	/*
+	 * Number of unpurged pages at beginning of current epoch.  During epoch
+	 * advancement we use the delta between arena->decay_*.nunpurged and
+	 * extents_npages_get(&arena->extents_*) to determine how many dirty
+	 * pages, if any, were generated.
+	 */
+	size_t			nunpurged;
+	/*
+	 * Trailing log of how many unused dirty pages were generated during
+	 * each of the past SMOOTHSTEP_NSTEPS decay epochs, where the last
+	 * element is the most recent epoch.  Corresponding epoch times are
+	 * relative to epoch.
+	 */
+	size_t			backlog[SMOOTHSTEP_NSTEPS];
+
+	/*
+	 * Pointer to associated stats.  These stats are embedded directly in
+	 * the arena's stats due to how stats structures are shared between the
+	 * arena and ctl code.
+	 *
+	 * Synchronization: Same as associated arena's stats field. */
+	arena_stats_decay_t	*stats;
+	/* Peak number of pages in associated extents.  Used for debug only. */
+	uint64_t		ceil_npages;
+};
+
+struct arena_s {
+	/*
+	 * Number of threads currently assigned to this arena.  Each thread has
+	 * two distinct assignments, one for application-serving allocation, and
+	 * the other for internal metadata allocation.  Internal metadata must
+	 * not be allocated from arenas explicitly created via the arenas.create
+	 * mallctl, because the arena.<i>.reset mallctl indiscriminately
+	 * discards all allocations for the affected arena.
+	 *
+	 *   0: Application allocation.
+	 *   1: Internal metadata allocation.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_u_t		nthreads[2];
+
+	/* Next bin shard for binding new threads. Synchronization: atomic. */
+	atomic_u_t		binshard_next;
+
+	/*
+	 * When percpu_arena is enabled, to amortize the cost of reading /
+	 * updating the current CPU id, track the most recent thread accessing
+	 * this arena, and only read CPU if there is a mismatch.
+	 */
+	tsdn_t		*last_thd;
+
+	/* Synchronization: internal. */
+	arena_stats_t		stats;
+
+	/*
+	 * Lists of tcaches and cache_bin_array_descriptors for extant threads
+	 * associated with this arena.  Stats from these are merged
+	 * incrementally, and at exit if opt_stats_print is enabled.
+	 *
+	 * Synchronization: tcache_ql_mtx.
+	 */
+	ql_head(tcache_t)			tcache_ql;
+	ql_head(cache_bin_array_descriptor_t)	cache_bin_array_descriptor_ql;
+	malloc_mutex_t				tcache_ql_mtx;
+
+	/* Synchronization: internal. */
+	prof_accum_t		prof_accum;
+
+	/*
+	 * PRNG state for cache index randomization of large allocation base
+	 * pointers.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		offset_state;
+
+	/*
+	 * Extent serial number generator state.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		extent_sn_next;
+
+	/*
+	 * Represents a dss_prec_t, but atomically.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_u_t		dss_prec;
+
+	/*
+	 * Number of pages in active extents.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		nactive;
+
+	/*
+	 * Extant large allocations.
+	 *
+	 * Synchronization: large_mtx.
+	 */
+	extent_list_t		large;
+	/* Synchronizes all large allocation/update/deallocation. */
+	malloc_mutex_t		large_mtx;
+
+	/*
+	 * Collections of extents that were previously allocated.  These are
+	 * used when allocating extents, in an attempt to re-use address space.
+	 *
+	 * Synchronization: internal.
+	 */
+	extents_t		extents_dirty;
+	extents_t		extents_muzzy;
+	extents_t		extents_retained;
+
+	/*
+	 * Decay-based purging state, responsible for scheduling extent state
+	 * transitions.
+	 *
+	 * Synchronization: internal.
+	 */
+	arena_decay_t		decay_dirty; /* dirty --> muzzy */
+	arena_decay_t		decay_muzzy; /* muzzy --> retained */
+
+	/*
+	 * Next extent size class in a growing series to use when satisfying a
+	 * request via the extent hooks (only if opt_retain).  This limits the
+	 * number of disjoint virtual memory ranges so that extent merging can
+	 * be effective even if multiple arenas' extent allocation requests are
+	 * highly interleaved.
+	 *
+	 * retain_grow_limit is the max allowed size ind to expand (unless the
+	 * required size is greater).  Default is no limit, and controlled
+	 * through mallctl only.
+	 *
+	 * Synchronization: extent_grow_mtx
+	 */
+	pszind_t		extent_grow_next;
+	pszind_t		retain_grow_limit;
+	malloc_mutex_t		extent_grow_mtx;
+
+	/*
+	 * Available extent structures that were allocated via
+	 * base_alloc_extent().
+	 *
+	 * Synchronization: extent_avail_mtx.
+	 */
+	extent_tree_t		extent_avail;
+	atomic_zu_t		extent_avail_cnt;
+	malloc_mutex_t		extent_avail_mtx;
+
+	/*
+	 * bins is used to store heaps of free regions.
+	 *
+	 * Synchronization: internal.
+	 */
+	bins_t			bins[SC_NBINS];
+
+	/*
+	 * Base allocator, from which arena metadata are allocated.
+	 *
+	 * Synchronization: internal.
+	 */
+	base_t			*base;
+	/* Used to determine uptime.  Read-only after initialization. */
+	nstime_t		create_time;
+};
+
+/* Used in conjunction with tsd for fast arena-related context lookup. */
+struct arena_tdata_s {
+	ticker_t		decay_ticker;
+};
+
+/* Used to pass rtree lookup context down the path. */
+struct alloc_ctx_s {
+	szind_t szind;
+	bool slab;
+};
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_types.h
new file mode 100644
index 0000000..624937e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/arena_types.h
@@ -0,0 +1,51 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_TYPES_H
+#define JEMALLOC_INTERNAL_ARENA_TYPES_H
+
+#include "jemalloc/internal/sc.h"
+
+/* Maximum number of regions in one slab. */
+#define LG_SLAB_MAXREGS		(LG_PAGE - SC_LG_TINY_MIN)
+#define SLAB_MAXREGS		(1U << LG_SLAB_MAXREGS)
+
+/* Default decay times in milliseconds. */
+#define DIRTY_DECAY_MS_DEFAULT	ZD(10 * 1000)
+#define MUZZY_DECAY_MS_DEFAULT	(0)
+/* Number of event ticks between time checks. */
+#define DECAY_NTICKS_PER_UPDATE	1000
+
+typedef struct arena_slab_data_s arena_slab_data_t;
+typedef struct arena_decay_s arena_decay_t;
+typedef struct arena_s arena_t;
+typedef struct arena_tdata_s arena_tdata_t;
+typedef struct alloc_ctx_s alloc_ctx_t;
+
+typedef enum {
+	percpu_arena_mode_names_base   = 0, /* Used for options processing. */
+
+	/*
+	 * *_uninit are used only during bootstrapping, and must correspond
+	 * to initialized variant plus percpu_arena_mode_enabled_base.
+	 */
+	percpu_arena_uninit            = 0,
+	per_phycpu_arena_uninit        = 1,
+
+	/* All non-disabled modes must come after percpu_arena_disabled. */
+	percpu_arena_disabled          = 2,
+
+	percpu_arena_mode_names_limit  = 3, /* Used for options processing. */
+	percpu_arena_mode_enabled_base = 3,
+
+	percpu_arena                   = 3,
+	per_phycpu_arena               = 4  /* Hyper threads share arena. */
+} percpu_arena_mode_t;
+
+#define PERCPU_ARENA_ENABLED(m)	((m) >= percpu_arena_mode_enabled_base)
+#define PERCPU_ARENA_DEFAULT	percpu_arena_disabled
+
+/*
+ * When allocation_size >= oversize_threshold, use the dedicated huge arena
+ * (unless have explicitly spicified arena index).  0 disables the feature.
+ */
+#define OVERSIZE_THRESHOLD_DEFAULT (8 << 20)
+
+#endif /* JEMALLOC_INTERNAL_ARENA_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/assert.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/assert.h
new file mode 100644
index 0000000..be4d45b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/assert.h
@@ -0,0 +1,56 @@
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/util.h"
+
+/*
+ * Define a custom assert() in order to reduce the chances of deadlock during
+ * assertion failure.
+ */
+#ifndef assert
+#define assert(e) do {							\
+	if (unlikely(config_debug && !(e))) {				\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Failed assertion: \"%s\"\n",	\
+		    __FILE__, __LINE__, #e);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#ifndef not_reached
+#define not_reached() do {						\
+	if (config_debug) {						\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Unreachable code reached\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+	unreachable();							\
+} while (0)
+#endif
+
+#ifndef not_implemented
+#define not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_printf("<jemalloc>: %s:%d: Not implemented\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#ifndef assert_not_implemented
+#define assert_not_implemented(e) do {					\
+	if (unlikely(config_debug && !(e))) {				\
+		not_implemented();					\
+	}								\
+} while (0)
+#endif
+
+/* Use to assert a particular configuration, e.g., cassert(config_debug). */
+#ifndef cassert
+#define cassert(c) do {							\
+	if (unlikely(!(c))) {						\
+		not_reached();						\
+	}								\
+} while (0)
+#endif
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic.h
new file mode 100644
index 0000000..a76f54c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic.h
@@ -0,0 +1,86 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_H
+#define JEMALLOC_INTERNAL_ATOMIC_H
+
+#define ATOMIC_INLINE JEMALLOC_ALWAYS_INLINE
+
+#define JEMALLOC_U8_ATOMICS
+#if defined(JEMALLOC_GCC_ATOMIC_ATOMICS)
+#  include "jemalloc/internal/atomic_gcc_atomic.h"
+#  if !defined(JEMALLOC_GCC_U8_ATOMIC_ATOMICS)
+#    undef JEMALLOC_U8_ATOMICS
+#  endif
+#elif defined(JEMALLOC_GCC_SYNC_ATOMICS)
+#  include "jemalloc/internal/atomic_gcc_sync.h"
+#  if !defined(JEMALLOC_GCC_U8_SYNC_ATOMICS)
+#    undef JEMALLOC_U8_ATOMICS
+#  endif
+#elif defined(_MSC_VER)
+#  include "jemalloc/internal/atomic_msvc.h"
+#elif defined(JEMALLOC_C11_ATOMICS)
+#  include "jemalloc/internal/atomic_c11.h"
+#else
+#  error "Don't have atomics implemented on this platform."
+#endif
+
+/*
+ * This header gives more or less a backport of C11 atomics. The user can write
+ * JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_sizeof_type); to generate
+ * counterparts of the C11 atomic functions for type, as so:
+ *   JEMALLOC_GENERATE_ATOMICS(int *, pi, 3);
+ * and then write things like:
+ *   int *some_ptr;
+ *   atomic_pi_t atomic_ptr_to_int;
+ *   atomic_store_pi(&atomic_ptr_to_int, some_ptr, ATOMIC_RELAXED);
+ *   int *prev_value = atomic_exchange_pi(&ptr_to_int, NULL, ATOMIC_ACQ_REL);
+ *   assert(some_ptr == prev_value);
+ * and expect things to work in the obvious way.
+ *
+ * Also included (with naming differences to avoid conflicts with the standard
+ * library):
+ *   atomic_fence(atomic_memory_order_t) (mimics C11's atomic_thread_fence).
+ *   ATOMIC_INIT (mimics C11's ATOMIC_VAR_INIT).
+ */
+
+/*
+ * Pure convenience, so that we don't have to type "atomic_memory_order_"
+ * quite so often.
+ */
+#define ATOMIC_RELAXED atomic_memory_order_relaxed
+#define ATOMIC_ACQUIRE atomic_memory_order_acquire
+#define ATOMIC_RELEASE atomic_memory_order_release
+#define ATOMIC_ACQ_REL atomic_memory_order_acq_rel
+#define ATOMIC_SEQ_CST atomic_memory_order_seq_cst
+
+/*
+ * Not all platforms have 64-bit atomics.  If we do, this #define exposes that
+ * fact.
+ */
+#if (LG_SIZEOF_PTR == 3 || LG_SIZEOF_INT == 3)
+#  define JEMALLOC_ATOMIC_U64
+#endif
+
+JEMALLOC_GENERATE_ATOMICS(void *, p, LG_SIZEOF_PTR)
+
+/*
+ * There's no actual guarantee that sizeof(bool) == 1, but it's true on the only
+ * platform that actually needs to know the size, MSVC.
+ */
+JEMALLOC_GENERATE_ATOMICS(bool, b, 0)
+
+JEMALLOC_GENERATE_INT_ATOMICS(unsigned, u, LG_SIZEOF_INT)
+
+JEMALLOC_GENERATE_INT_ATOMICS(size_t, zu, LG_SIZEOF_PTR)
+
+JEMALLOC_GENERATE_INT_ATOMICS(ssize_t, zd, LG_SIZEOF_PTR)
+
+JEMALLOC_GENERATE_INT_ATOMICS(uint8_t, u8, 0)
+
+JEMALLOC_GENERATE_INT_ATOMICS(uint32_t, u32, 2)
+
+#ifdef JEMALLOC_ATOMIC_U64
+JEMALLOC_GENERATE_INT_ATOMICS(uint64_t, u64, 3)
+#endif
+
+#undef ATOMIC_INLINE
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_c11.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_c11.h
new file mode 100644
index 0000000..a5f9313
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_c11.h
@@ -0,0 +1,97 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_C11_H
+#define JEMALLOC_INTERNAL_ATOMIC_C11_H
+
+#include <stdatomic.h>
+
+#define ATOMIC_INIT(...) ATOMIC_VAR_INIT(__VA_ARGS__)
+
+#define atomic_memory_order_t memory_order
+#define atomic_memory_order_relaxed memory_order_relaxed
+#define atomic_memory_order_acquire memory_order_acquire
+#define atomic_memory_order_release memory_order_release
+#define atomic_memory_order_acq_rel memory_order_acq_rel
+#define atomic_memory_order_seq_cst memory_order_seq_cst
+
+#define atomic_fence atomic_thread_fence
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef _Atomic(type) atomic_##short_type##_t;				\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	/*								\
+	 * A strict interpretation of the C standard prevents		\
+	 * atomic_load from taking a const argument, but it's		\
+	 * convenient for our purposes. This cast is a workaround.	\
+	 */								\
+	atomic_##short_type##_t* a_nonconst =				\
+	    (atomic_##short_type##_t*)a;				\
+	return atomic_load_explicit(a_nonconst, mo);			\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	atomic_store_explicit(a, val, mo);				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return atomic_exchange_explicit(a, val, mo);			\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return atomic_compare_exchange_weak_explicit(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return atomic_compare_exchange_strong_explicit(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}
+
+/*
+ * Integral types have some special operations available that non-integral ones
+ * lack.
+ */
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type, 		\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_add_explicit(a, val, mo);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_sub_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_and_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_or_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_xor_explicit(a, val, mo);			\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_C11_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
new file mode 100644
index 0000000..471515e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_gcc_atomic.h
@@ -0,0 +1,129 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H
+#define JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H
+
+#include "jemalloc/internal/assert.h"
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+ATOMIC_INLINE int
+atomic_enum_to_builtin(atomic_memory_order_t mo) {
+	switch (mo) {
+	case atomic_memory_order_relaxed:
+		return __ATOMIC_RELAXED;
+	case atomic_memory_order_acquire:
+		return __ATOMIC_ACQUIRE;
+	case atomic_memory_order_release:
+		return __ATOMIC_RELEASE;
+	case atomic_memory_order_acq_rel:
+		return __ATOMIC_ACQ_REL;
+	case atomic_memory_order_seq_cst:
+		return __ATOMIC_SEQ_CST;
+	}
+	/* Can't happen; the switch is exhaustive. */
+	not_reached();
+}
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	__atomic_thread_fence(atomic_enum_to_builtin(mo));
+}
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef struct {							\
+	type repr;							\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	type result;							\
+	__atomic_load(&a->repr, &result, atomic_enum_to_builtin(mo));	\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a, type val,		\
+    atomic_memory_order_t mo) {						\
+	__atomic_store(&a->repr, &val, atomic_enum_to_builtin(mo));	\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	type result;							\
+	__atomic_exchange(&a->repr, &val, &result,			\
+	    atomic_enum_to_builtin(mo));				\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    UNUSED type *expected, type desired,				\
+    atomic_memory_order_t success_mo,					\
+    atomic_memory_order_t failure_mo) {					\
+	return __atomic_compare_exchange(&a->repr, expected, &desired,	\
+	    true, atomic_enum_to_builtin(success_mo),			\
+	    atomic_enum_to_builtin(failure_mo));			\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    UNUSED type *expected, type desired,				\
+    atomic_memory_order_t success_mo,					\
+    atomic_memory_order_t failure_mo) {					\
+	return __atomic_compare_exchange(&a->repr, expected, &desired,	\
+	    false,							\
+	    atomic_enum_to_builtin(success_mo),				\
+	    atomic_enum_to_builtin(failure_mo));			\
+}
+
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_add(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_sub(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_and(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_or(&a->repr, val,				\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_xor(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_gcc_sync.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
new file mode 100644
index 0000000..e02b7cb
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_gcc_sync.h
@@ -0,0 +1,195 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H
+#define JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	/* Easy cases first: no barrier, and full barrier. */
+	if (mo == atomic_memory_order_relaxed) {
+		asm volatile("" ::: "memory");
+		return;
+	}
+	if (mo == atomic_memory_order_seq_cst) {
+		asm volatile("" ::: "memory");
+		__sync_synchronize();
+		asm volatile("" ::: "memory");
+		return;
+	}
+	asm volatile("" ::: "memory");
+#  if defined(__i386__) || defined(__x86_64__)
+	/* This is implicit on x86. */
+#  elif defined(__ppc64__)
+	asm volatile("lwsync");
+#  elif defined(__ppc__)
+	asm volatile("sync");
+#  elif defined(__sparc__) && defined(__arch64__)
+	if (mo == atomic_memory_order_acquire) {
+		asm volatile("membar #LoadLoad | #LoadStore");
+	} else if (mo == atomic_memory_order_release) {
+		asm volatile("membar #LoadStore | #StoreStore");
+	} else {
+		asm volatile("membar #LoadLoad | #LoadStore | #StoreStore");
+	}
+#  else
+	__sync_synchronize();
+#  endif
+	asm volatile("" ::: "memory");
+}
+
+/*
+ * A correct implementation of seq_cst loads and stores on weakly ordered
+ * architectures could do either of the following:
+ *   1. store() is weak-fence -> store -> strong fence, load() is load ->
+ *      strong-fence.
+ *   2. store() is strong-fence -> store, load() is strong-fence -> load ->
+ *      weak-fence.
+ * The tricky thing is, load() and store() above can be the load or store
+ * portions of a gcc __sync builtin, so we have to follow GCC's lead, which
+ * means going with strategy 2.
+ * On strongly ordered architectures, the natural strategy is to stick a strong
+ * fence after seq_cst stores, and have naked loads.  So we want the strong
+ * fences in different places on different architectures.
+ * atomic_pre_sc_load_fence and atomic_post_sc_store_fence allow us to
+ * accomplish this.
+ */
+
+ATOMIC_INLINE void
+atomic_pre_sc_load_fence() {
+#  if defined(__i386__) || defined(__x86_64__) ||			\
+    (defined(__sparc__) && defined(__arch64__))
+	atomic_fence(atomic_memory_order_relaxed);
+#  else
+	atomic_fence(atomic_memory_order_seq_cst);
+#  endif
+}
+
+ATOMIC_INLINE void
+atomic_post_sc_store_fence() {
+#  if defined(__i386__) || defined(__x86_64__) ||			\
+    (defined(__sparc__) && defined(__arch64__))
+	atomic_fence(atomic_memory_order_seq_cst);
+#  else
+	atomic_fence(atomic_memory_order_relaxed);
+#  endif
+
+}
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef struct {							\
+	type volatile repr;						\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_pre_sc_load_fence();				\
+	}								\
+	type result = a->repr;						\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_acquire);		\
+	}								\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_release);		\
+	}								\
+	a->repr = val;							\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_post_sc_store_fence();				\
+	}								\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val, \
+    atomic_memory_order_t mo) {                  					 \
+	/*								\
+	 * Because of FreeBSD, we care about gcc 4.2, which doesn't have\
+	 * an atomic exchange builtin.  We fake it with a CAS loop.	\
+	 */								\
+	while (true) {							\
+		type old = a->repr;					\
+		if (__sync_bool_compare_and_swap(&a->repr, old, val)) {	\
+			return old;					\
+		}							\
+	}								\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired,                                     \
+    atomic_memory_order_t success_mo,                          \
+    atomic_memory_order_t failure_mo) {				                \
+	type prev = __sync_val_compare_and_swap(&a->repr, *expected,	\
+	    desired);							\
+	if (prev == *expected) {					\
+		return true;						\
+	} else {							\
+		*expected = prev;					\
+		return false;						\
+	}								\
+}									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired,                                       \
+    atomic_memory_order_t success_mo,                            \
+    atomic_memory_order_t failure_mo) {                          \
+	type prev = __sync_val_compare_and_swap(&a->repr, *expected,	\
+	    desired);							\
+	if (prev == *expected) {					\
+		return true;						\
+	} else {							\
+		*expected = prev;					\
+		return false;						\
+	}								\
+}
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_add(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_sub(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_and(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_or(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_xor(&a->repr, val);			\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_msvc.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_msvc.h
new file mode 100644
index 0000000..67057ce
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/atomic_msvc.h
@@ -0,0 +1,158 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_MSVC_H
+#define JEMALLOC_INTERNAL_ATOMIC_MSVC_H
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+typedef char atomic_repr_0_t;
+typedef short atomic_repr_1_t;
+typedef long atomic_repr_2_t;
+typedef __int64 atomic_repr_3_t;
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	_ReadWriteBarrier();
+#  if defined(_M_ARM) || defined(_M_ARM64)
+	/* ARM needs a barrier for everything but relaxed. */
+	if (mo != atomic_memory_order_relaxed) {
+		MemoryBarrier();
+	}
+#  elif defined(_M_IX86) || defined (_M_X64)
+	/* x86 needs a barrier only for seq_cst. */
+	if (mo == atomic_memory_order_seq_cst) {
+		MemoryBarrier();
+	}
+#  else
+#  error "Don't know how to create atomics for this platform for MSVC."
+#  endif
+	_ReadWriteBarrier();
+}
+
+#define ATOMIC_INTERLOCKED_REPR(lg_size) atomic_repr_ ## lg_size ## _t
+
+#define ATOMIC_CONCAT(a, b) ATOMIC_RAW_CONCAT(a, b)
+#define ATOMIC_RAW_CONCAT(a, b) a ## b
+
+#define ATOMIC_INTERLOCKED_NAME(base_name, lg_size) ATOMIC_CONCAT(	\
+    base_name, ATOMIC_INTERLOCKED_SUFFIX(lg_size))
+
+#define ATOMIC_INTERLOCKED_SUFFIX(lg_size)				\
+    ATOMIC_CONCAT(ATOMIC_INTERLOCKED_SUFFIX_, lg_size)
+
+#define ATOMIC_INTERLOCKED_SUFFIX_0 8
+#define ATOMIC_INTERLOCKED_SUFFIX_1 16
+#define ATOMIC_INTERLOCKED_SUFFIX_2
+#define ATOMIC_INTERLOCKED_SUFFIX_3 64
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_size)		\
+typedef struct {							\
+	ATOMIC_INTERLOCKED_REPR(lg_size) repr;				\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	ATOMIC_INTERLOCKED_REPR(lg_size) ret = a->repr;			\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_acquire);		\
+	}								\
+	return (type) ret;						\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_release);		\
+	}								\
+	a->repr = (ATOMIC_INTERLOCKED_REPR(lg_size)) val;		\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_fence(atomic_memory_order_seq_cst);		\
+	}								\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedExchange,	\
+	    lg_size)(&a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);	\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	ATOMIC_INTERLOCKED_REPR(lg_size) e =				\
+	    (ATOMIC_INTERLOCKED_REPR(lg_size))*expected;		\
+	ATOMIC_INTERLOCKED_REPR(lg_size) d =				\
+	    (ATOMIC_INTERLOCKED_REPR(lg_size))desired;			\
+	ATOMIC_INTERLOCKED_REPR(lg_size) old =				\
+	    ATOMIC_INTERLOCKED_NAME(_InterlockedCompareExchange, 	\
+		lg_size)(&a->repr, d, e);				\
+	if (old == e) {							\
+		return true;						\
+	} else {							\
+		*expected = (type)old;					\
+		return false;						\
+	}								\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	/* We implement the weak version with strong semantics. */	\
+	return atomic_compare_exchange_weak_##short_type(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}
+
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type, lg_size)	\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_size)			\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedExchangeAdd,	\
+	    lg_size)(&a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);	\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	/*								\
+	 * MSVC warns on negation of unsigned operands, but for us it	\
+	 * gives exactly the right semantics (MAX_TYPE + 1 - operand).	\
+	 */								\
+	__pragma(warning(push))						\
+	__pragma(warning(disable: 4146))				\
+	return atomic_fetch_add_##short_type(a, -val, mo);		\
+	__pragma(warning(pop))						\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedAnd, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedOr, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedXor, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_MSVC_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_externs.h
new file mode 100644
index 0000000..0f997e1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_externs.h
@@ -0,0 +1,32 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H
+
+extern bool opt_background_thread;
+extern size_t opt_max_background_threads;
+extern malloc_mutex_t background_thread_lock;
+extern atomic_b_t background_thread_enabled_state;
+extern size_t n_background_threads;
+extern size_t max_background_threads;
+extern background_thread_info_t *background_thread_info;
+
+bool background_thread_create(tsd_t *tsd, unsigned arena_ind);
+bool background_threads_enable(tsd_t *tsd);
+bool background_threads_disable(tsd_t *tsd);
+void background_thread_interval_check(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, size_t npages_new);
+void background_thread_prefork0(tsdn_t *tsdn);
+void background_thread_prefork1(tsdn_t *tsdn);
+void background_thread_postfork_parent(tsdn_t *tsdn);
+void background_thread_postfork_child(tsdn_t *tsdn);
+bool background_thread_stats_read(tsdn_t *tsdn,
+    background_thread_stats_t *stats);
+void background_thread_ctl_init(tsdn_t *tsdn);
+
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+extern int pthread_create_wrapper(pthread_t *__restrict, const pthread_attr_t *,
+    void *(*)(void *), void *__restrict);
+#endif
+bool background_thread_boot0(void);
+bool background_thread_boot1(tsdn_t *tsdn);
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_inlines.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_inlines.h
new file mode 100644
index 0000000..f85e86f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_inlines.h
@@ -0,0 +1,62 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H
+
+JEMALLOC_ALWAYS_INLINE bool
+background_thread_enabled(void) {
+	return atomic_load_b(&background_thread_enabled_state, ATOMIC_RELAXED);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+background_thread_enabled_set(tsdn_t *tsdn, bool state) {
+	malloc_mutex_assert_owner(tsdn, &background_thread_lock);
+	atomic_store_b(&background_thread_enabled_state, state, ATOMIC_RELAXED);
+}
+
+JEMALLOC_ALWAYS_INLINE background_thread_info_t *
+arena_background_thread_info_get(arena_t *arena) {
+	unsigned arena_ind = arena_ind_get(arena);
+	return &background_thread_info[arena_ind % max_background_threads];
+}
+
+JEMALLOC_ALWAYS_INLINE background_thread_info_t *
+background_thread_info_get(size_t ind) {
+	return &background_thread_info[ind % max_background_threads];
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+background_thread_wakeup_time_get(background_thread_info_t *info) {
+	uint64_t next_wakeup = nstime_ns(&info->next_wakeup);
+	assert(atomic_load_b(&info->indefinite_sleep, ATOMIC_ACQUIRE) ==
+	    (next_wakeup == BACKGROUND_THREAD_INDEFINITE_SLEEP));
+	return next_wakeup;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+background_thread_wakeup_time_set(tsdn_t *tsdn, background_thread_info_t *info,
+    uint64_t wakeup_time) {
+	malloc_mutex_assert_owner(tsdn, &info->mtx);
+	atomic_store_b(&info->indefinite_sleep,
+	    wakeup_time == BACKGROUND_THREAD_INDEFINITE_SLEEP, ATOMIC_RELEASE);
+	nstime_init(&info->next_wakeup, wakeup_time);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+background_thread_indefinite_sleep(background_thread_info_t *info) {
+	return atomic_load_b(&info->indefinite_sleep, ATOMIC_ACQUIRE);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_background_thread_inactivity_check(tsdn_t *tsdn, arena_t *arena,
+    bool is_background_thread) {
+	if (!background_thread_enabled() || is_background_thread) {
+		return;
+	}
+	background_thread_info_t *info =
+	    arena_background_thread_info_get(arena);
+	if (background_thread_indefinite_sleep(info)) {
+		background_thread_interval_check(tsdn, arena,
+		    &arena->decay_dirty, 0);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_structs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_structs.h
new file mode 100644
index 0000000..c02aa43
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/background_thread_structs.h
@@ -0,0 +1,54 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H
+
+/* This file really combines "structs" and "types", but only transitionally. */
+
+#if defined(JEMALLOC_BACKGROUND_THREAD) || defined(JEMALLOC_LAZY_LOCK)
+#  define JEMALLOC_PTHREAD_CREATE_WRAPPER
+#endif
+
+#define BACKGROUND_THREAD_INDEFINITE_SLEEP UINT64_MAX
+#define MAX_BACKGROUND_THREAD_LIMIT MALLOCX_ARENA_LIMIT
+#define DEFAULT_NUM_BACKGROUND_THREAD 4
+
+typedef enum {
+	background_thread_stopped,
+	background_thread_started,
+	/* Thread waits on the global lock when paused (for arena_reset). */
+	background_thread_paused,
+} background_thread_state_t;
+
+struct background_thread_info_s {
+#ifdef JEMALLOC_BACKGROUND_THREAD
+	/* Background thread is pthread specific. */
+	pthread_t		thread;
+	pthread_cond_t		cond;
+#endif
+	malloc_mutex_t		mtx;
+	background_thread_state_t	state;
+	/* When true, it means no wakeup scheduled. */
+	atomic_b_t		indefinite_sleep;
+	/* Next scheduled wakeup time (absolute time in ns). */
+	nstime_t		next_wakeup;
+	/*
+	 *  Since the last background thread run, newly added number of pages
+	 *  that need to be purged by the next wakeup.  This is adjusted on
+	 *  epoch advance, and is used to determine whether we should signal the
+	 *  background thread to wake up earlier.
+	 */
+	size_t			npages_to_purge_new;
+	/* Stats: total number of runs since started. */
+	uint64_t		tot_n_runs;
+	/* Stats: total sleep time since started. */
+	nstime_t		tot_sleep_time;
+};
+typedef struct background_thread_info_s background_thread_info_t;
+
+struct background_thread_stats_s {
+	size_t num_threads;
+	uint64_t num_runs;
+	nstime_t run_interval;
+};
+typedef struct background_thread_stats_s background_thread_stats_t;
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_externs.h
new file mode 100644
index 0000000..7b705c9
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_externs.h
@@ -0,0 +1,22 @@
+#ifndef JEMALLOC_INTERNAL_BASE_EXTERNS_H
+#define JEMALLOC_INTERNAL_BASE_EXTERNS_H
+
+extern metadata_thp_mode_t opt_metadata_thp;
+extern const char *metadata_thp_mode_names[];
+
+base_t *b0get(void);
+base_t *base_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+void base_delete(tsdn_t *tsdn, base_t *base);
+extent_hooks_t *base_extent_hooks_get(base_t *base);
+extent_hooks_t *base_extent_hooks_set(base_t *base,
+    extent_hooks_t *extent_hooks);
+void *base_alloc(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment);
+extent_t *base_alloc_extent(tsdn_t *tsdn, base_t *base);
+void base_stats_get(tsdn_t *tsdn, base_t *base, size_t *allocated,
+    size_t *resident, size_t *mapped, size_t *n_thp);
+void base_prefork(tsdn_t *tsdn, base_t *base);
+void base_postfork_parent(tsdn_t *tsdn, base_t *base);
+void base_postfork_child(tsdn_t *tsdn, base_t *base);
+bool base_boot(tsdn_t *tsdn);
+
+#endif /* JEMALLOC_INTERNAL_BASE_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_inlines.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_inlines.h
new file mode 100644
index 0000000..aec0e2e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_inlines.h
@@ -0,0 +1,13 @@
+#ifndef JEMALLOC_INTERNAL_BASE_INLINES_H
+#define JEMALLOC_INTERNAL_BASE_INLINES_H
+
+static inline unsigned
+base_ind_get(const base_t *base) {
+	return base->ind;
+}
+
+static inline bool
+metadata_thp_enabled(void) {
+	return (opt_metadata_thp != metadata_thp_disabled);
+}
+#endif /* JEMALLOC_INTERNAL_BASE_INLINES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_structs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_structs.h
new file mode 100644
index 0000000..07f214e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_structs.h
@@ -0,0 +1,59 @@
+#ifndef JEMALLOC_INTERNAL_BASE_STRUCTS_H
+#define JEMALLOC_INTERNAL_BASE_STRUCTS_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/sc.h"
+
+/* Embedded at the beginning of every block of base-managed virtual memory. */
+struct base_block_s {
+	/* Total size of block's virtual memory mapping. */
+	size_t		size;
+
+	/* Next block in list of base's blocks. */
+	base_block_t	*next;
+
+	/* Tracks unused trailing space. */
+	extent_t	extent;
+};
+
+struct base_s {
+	/* Associated arena's index within the arenas array. */
+	unsigned	ind;
+
+	/*
+	 * User-configurable extent hook functions.  Points to an
+	 * extent_hooks_t.
+	 */
+	atomic_p_t	extent_hooks;
+
+	/* Protects base_alloc() and base_stats_get() operations. */
+	malloc_mutex_t	mtx;
+
+	/* Using THP when true (metadata_thp auto mode). */
+	bool		auto_thp_switched;
+	/*
+	 * Most recent size class in the series of increasingly large base
+	 * extents.  Logarithmic spacing between subsequent allocations ensures
+	 * that the total number of distinct mappings remains small.
+	 */
+	pszind_t	pind_last;
+
+	/* Serial number generation state. */
+	size_t		extent_sn_next;
+
+	/* Chain of all blocks associated with base. */
+	base_block_t	*blocks;
+
+	/* Heap of extents that track unused trailing space within blocks. */
+	extent_heap_t	avail[SC_NSIZES];
+
+	/* Stats, only maintained if config_stats. */
+	size_t		allocated;
+	size_t		resident;
+	size_t		mapped;
+	/* Number of THP regions touched. */
+	size_t		n_thp;
+};
+
+#endif /* JEMALLOC_INTERNAL_BASE_STRUCTS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_types.h
new file mode 100644
index 0000000..b6db77d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/base_types.h
@@ -0,0 +1,33 @@
+#ifndef JEMALLOC_INTERNAL_BASE_TYPES_H
+#define JEMALLOC_INTERNAL_BASE_TYPES_H
+
+typedef struct base_block_s base_block_t;
+typedef struct base_s base_t;
+
+#define METADATA_THP_DEFAULT metadata_thp_disabled
+
+/*
+ * In auto mode, arenas switch to huge pages for the base allocator on the
+ * second base block.  a0 switches to thp on the 5th block (after 20 megabytes
+ * of metadata), since more metadata (e.g. rtree nodes) come from a0's base.
+ */
+
+#define BASE_AUTO_THP_THRESHOLD    2
+#define BASE_AUTO_THP_THRESHOLD_A0 5
+
+typedef enum {
+	metadata_thp_disabled   = 0,
+	/*
+	 * Lazily enable hugepage for metadata. To avoid high RSS caused by THP
+	 * + low usage arena (i.e. THP becomes a significant percentage), the
+	 * "auto" option only starts using THP after a base allocator used up
+	 * the first THP region.  Starting from the second hugepage (in a single
+	 * arena), "auto" behaves the same as "always", i.e. madvise hugepage
+	 * right away.
+	 */
+	metadata_thp_auto       = 1,
+	metadata_thp_always     = 2,
+	metadata_thp_mode_limit = 3
+} metadata_thp_mode_t;
+
+#endif /* JEMALLOC_INTERNAL_BASE_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin.h
new file mode 100644
index 0000000..8547e89
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin.h
@@ -0,0 +1,123 @@
+#ifndef JEMALLOC_INTERNAL_BIN_H
+#define JEMALLOC_INTERNAL_BIN_H
+
+#include "jemalloc/internal/bin_stats.h"
+#include "jemalloc/internal/bin_types.h"
+#include "jemalloc/internal/extent_types.h"
+#include "jemalloc/internal/extent_structs.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/sc.h"
+
+/*
+ * A bin contains a set of extents that are currently being used for slab
+ * allocations.
+ */
+
+/*
+ * Read-only information associated with each element of arena_t's bins array
+ * is stored separately, partly to reduce memory usage (only one copy, rather
+ * than one per arena), but mainly to avoid false cacheline sharing.
+ *
+ * Each slab has the following layout:
+ *
+ *   /--------------------\
+ *   | region 0           |
+ *   |--------------------|
+ *   | region 1           |
+ *   |--------------------|
+ *   | ...                |
+ *   | ...                |
+ *   | ...                |
+ *   |--------------------|
+ *   | region nregs-1     |
+ *   \--------------------/
+ */
+typedef struct bin_info_s bin_info_t;
+struct bin_info_s {
+	/* Size of regions in a slab for this bin's size class. */
+	size_t			reg_size;
+
+	/* Total size of a slab for this bin's size class. */
+	size_t			slab_size;
+
+	/* Total number of regions in a slab for this bin's size class. */
+	uint32_t		nregs;
+
+	/* Number of sharded bins in each arena for this size class. */
+	uint32_t		n_shards;
+
+	/*
+	 * Metadata used to manipulate bitmaps for slabs associated with this
+	 * bin.
+	 */
+	bitmap_info_t		bitmap_info;
+};
+
+extern bin_info_t bin_infos[SC_NBINS];
+
+typedef struct bin_s bin_t;
+struct bin_s {
+	/* All operations on bin_t fields require lock ownership. */
+	malloc_mutex_t		lock;
+
+	/*
+	 * Current slab being used to service allocations of this bin's size
+	 * class.  slabcur is independent of slabs_{nonfull,full}; whenever
+	 * slabcur is reassigned, the previous slab must be deallocated or
+	 * inserted into slabs_{nonfull,full}.
+	 */
+	extent_t		*slabcur;
+
+	/*
+	 * Heap of non-full slabs.  This heap is used to assure that new
+	 * allocations come from the non-full slab that is oldest/lowest in
+	 * memory.
+	 */
+	extent_heap_t		slabs_nonfull;
+
+	/* List used to track full slabs. */
+	extent_list_t		slabs_full;
+
+	/* Bin statistics. */
+	bin_stats_t	stats;
+};
+
+/* A set of sharded bins of the same size class. */
+typedef struct bins_s bins_t;
+struct bins_s {
+	/* Sharded bins.  Dynamically sized. */
+	bin_t *bin_shards;
+};
+
+void bin_shard_sizes_boot(unsigned bin_shards[SC_NBINS]);
+bool bin_update_shard_size(unsigned bin_shards[SC_NBINS], size_t start_size,
+    size_t end_size, size_t nshards);
+void bin_boot(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS]);
+
+/* Initializes a bin to empty.  Returns true on error. */
+bool bin_init(bin_t *bin);
+
+/* Forking. */
+void bin_prefork(tsdn_t *tsdn, bin_t *bin);
+void bin_postfork_parent(tsdn_t *tsdn, bin_t *bin);
+void bin_postfork_child(tsdn_t *tsdn, bin_t *bin);
+
+/* Stats. */
+static inline void
+bin_stats_merge(tsdn_t *tsdn, bin_stats_t *dst_bin_stats, bin_t *bin) {
+	malloc_mutex_lock(tsdn, &bin->lock);
+	malloc_mutex_prof_accum(tsdn, &dst_bin_stats->mutex_data, &bin->lock);
+	dst_bin_stats->nmalloc += bin->stats.nmalloc;
+	dst_bin_stats->ndalloc += bin->stats.ndalloc;
+	dst_bin_stats->nrequests += bin->stats.nrequests;
+	dst_bin_stats->curregs += bin->stats.curregs;
+	dst_bin_stats->nfills += bin->stats.nfills;
+	dst_bin_stats->nflushes += bin->stats.nflushes;
+	dst_bin_stats->nslabs += bin->stats.nslabs;
+	dst_bin_stats->reslabs += bin->stats.reslabs;
+	dst_bin_stats->curslabs += bin->stats.curslabs;
+	dst_bin_stats->nonfull_slabs += bin->stats.nonfull_slabs;
+	malloc_mutex_unlock(tsdn, &bin->lock);
+}
+
+#endif /* JEMALLOC_INTERNAL_BIN_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin_stats.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin_stats.h
new file mode 100644
index 0000000..d04519c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin_stats.h
@@ -0,0 +1,54 @@
+#ifndef JEMALLOC_INTERNAL_BIN_STATS_H
+#define JEMALLOC_INTERNAL_BIN_STATS_H
+
+#include "jemalloc/internal/mutex_prof.h"
+
+typedef struct bin_stats_s bin_stats_t;
+struct bin_stats_s {
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the bin.  Note that tcache may allocate an object, then recycle it
+	 * many times, resulting many increments to nrequests, but only one
+	 * each to nmalloc and ndalloc.
+	 */
+	uint64_t	nmalloc;
+	uint64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to the size of this
+	 * bin.  This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	uint64_t	nrequests;
+
+	/*
+	 * Current number of regions of this size class, including regions
+	 * currently cached by tcache.
+	 */
+	size_t		curregs;
+
+	/* Number of tcache fills from this bin. */
+	uint64_t	nfills;
+
+	/* Number of tcache flushes to this bin. */
+	uint64_t	nflushes;
+
+	/* Total number of slabs created for this bin's size class. */
+	uint64_t	nslabs;
+
+	/*
+	 * Total number of slabs reused by extracting them from the slabs heap
+	 * for this bin's size class.
+	 */
+	uint64_t	reslabs;
+
+	/* Current number of slabs in this bin. */
+	size_t		curslabs;
+
+	/* Current size of nonfull slabs heap in this bin. */
+	size_t		nonfull_slabs;
+
+	mutex_prof_data_t mutex_data;
+};
+
+#endif /* JEMALLOC_INTERNAL_BIN_STATS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin_types.h
new file mode 100644
index 0000000..3533606
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bin_types.h
@@ -0,0 +1,17 @@
+#ifndef JEMALLOC_INTERNAL_BIN_TYPES_H
+#define JEMALLOC_INTERNAL_BIN_TYPES_H
+
+#include "jemalloc/internal/sc.h"
+
+#define BIN_SHARDS_MAX (1 << EXTENT_BITS_BINSHARD_WIDTH)
+#define N_BIN_SHARDS_DEFAULT 1
+
+/* Used in TSD static initializer only. Real init in arena_bind(). */
+#define TSD_BINSHARDS_ZERO_INITIALIZER {{UINT8_MAX}}
+
+typedef struct tsd_binshards_s tsd_binshards_t;
+struct tsd_binshards_s {
+	uint8_t binshard[SC_NBINS];
+};
+
+#endif /* JEMALLOC_INTERNAL_BIN_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/bit_util.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bit_util.h
new file mode 100644
index 0000000..c045eb8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bit_util.h
@@ -0,0 +1,239 @@
+#ifndef JEMALLOC_INTERNAL_BIT_UTIL_H
+#define JEMALLOC_INTERNAL_BIT_UTIL_H
+
+#include "jemalloc/internal/assert.h"
+
+#define BIT_UTIL_INLINE static inline
+
+/* Sanity check. */
+#if !defined(JEMALLOC_INTERNAL_FFSLL) || !defined(JEMALLOC_INTERNAL_FFSL) \
+    || !defined(JEMALLOC_INTERNAL_FFS)
+#  error JEMALLOC_INTERNAL_FFS{,L,LL} should have been defined by configure
+#endif
+
+
+BIT_UTIL_INLINE unsigned
+ffs_llu(unsigned long long bitmap) {
+	return JEMALLOC_INTERNAL_FFSLL(bitmap);
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_lu(unsigned long bitmap) {
+	return JEMALLOC_INTERNAL_FFSL(bitmap);
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u(unsigned bitmap) {
+	return JEMALLOC_INTERNAL_FFS(bitmap);
+}
+
+#ifdef JEMALLOC_INTERNAL_POPCOUNTL
+BIT_UTIL_INLINE unsigned
+popcount_lu(unsigned long bitmap) {
+  return JEMALLOC_INTERNAL_POPCOUNTL(bitmap);
+}
+#endif
+
+/*
+ * Clears first unset bit in bitmap, and returns
+ * place of bit.  bitmap *must not* be 0.
+ */
+
+BIT_UTIL_INLINE size_t
+cfs_lu(unsigned long* bitmap) {
+	size_t bit = ffs_lu(*bitmap) - 1;
+	*bitmap ^= ZU(1) << bit;
+	return bit;
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_zu(size_t bitmap) {
+#if LG_SIZEOF_PTR == LG_SIZEOF_INT
+	return ffs_u(bitmap);
+#elif LG_SIZEOF_PTR == LG_SIZEOF_LONG
+	return ffs_lu(bitmap);
+#elif LG_SIZEOF_PTR == LG_SIZEOF_LONG_LONG
+	return ffs_llu(bitmap);
+#else
+#error No implementation for size_t ffs()
+#endif
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u64(uint64_t bitmap) {
+#if LG_SIZEOF_LONG == 3
+	return ffs_lu(bitmap);
+#elif LG_SIZEOF_LONG_LONG == 3
+	return ffs_llu(bitmap);
+#else
+#error No implementation for 64-bit ffs()
+#endif
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u32(uint32_t bitmap) {
+#if LG_SIZEOF_INT == 2
+	return ffs_u(bitmap);
+#else
+#error No implementation for 32-bit ffs()
+#endif
+	return ffs_u(bitmap);
+}
+
+BIT_UTIL_INLINE uint64_t
+pow2_ceil_u64(uint64_t x) {
+#if (defined(__amd64__) || defined(__x86_64__) || defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+	if(unlikely(x <= 1)) {
+		return x;
+	}
+	size_t msb_on_index;
+#if (defined(__amd64__) || defined(__x86_64__))
+	asm ("bsrq %1, %0"
+			: "=r"(msb_on_index) // Outputs.
+			: "r"(x-1)           // Inputs.
+		);
+#elif (defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+	msb_on_index = (63 ^ __builtin_clzll(x - 1));
+#endif
+	assert(msb_on_index < 63);
+	return 1ULL << (msb_on_index + 1);
+#else
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x |= x >> 32;
+	x++;
+	return x;
+#endif
+}
+
+BIT_UTIL_INLINE uint32_t
+pow2_ceil_u32(uint32_t x) {
+#if ((defined(__i386__) || defined(JEMALLOC_HAVE_BUILTIN_CLZ)) && (!defined(__s390__)))
+	if(unlikely(x <= 1)) {
+		return x;
+	}
+	size_t msb_on_index;
+#if (defined(__i386__))
+	asm ("bsr %1, %0"
+			: "=r"(msb_on_index) // Outputs.
+			: "r"(x-1)           // Inputs.
+		);
+#elif (defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+	msb_on_index = (31 ^ __builtin_clz(x - 1));
+#endif
+	assert(msb_on_index < 31);
+	return 1U << (msb_on_index + 1);
+#else
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x++;
+	return x;
+#endif
+}
+
+/* Compute the smallest power of 2 that is >= x. */
+BIT_UTIL_INLINE size_t
+pow2_ceil_zu(size_t x) {
+#if (LG_SIZEOF_PTR == 3)
+	return pow2_ceil_u64(x);
+#else
+	return pow2_ceil_u32(x);
+#endif
+}
+
+#if (defined(__i386__) || defined(__amd64__) || defined(__x86_64__))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	size_t ret;
+	assert(x != 0);
+
+	asm ("bsr %1, %0"
+	    : "=r"(ret) // Outputs.
+	    : "r"(x)    // Inputs.
+	    );
+	assert(ret < UINT_MAX);
+	return (unsigned)ret;
+}
+#elif (defined(_MSC_VER))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	unsigned long ret;
+
+	assert(x != 0);
+
+#if (LG_SIZEOF_PTR == 3)
+	_BitScanReverse64(&ret, x);
+#elif (LG_SIZEOF_PTR == 2)
+	_BitScanReverse(&ret, x);
+#else
+#  error "Unsupported type size for lg_floor()"
+#endif
+	assert(ret < UINT_MAX);
+	return (unsigned)ret;
+}
+#elif (defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	assert(x != 0);
+
+#if (LG_SIZEOF_PTR == LG_SIZEOF_INT)
+	return ((8 << LG_SIZEOF_PTR) - 1) - __builtin_clz(x);
+#elif (LG_SIZEOF_PTR == LG_SIZEOF_LONG)
+	return ((8 << LG_SIZEOF_PTR) - 1) - __builtin_clzl(x);
+#else
+#  error "Unsupported type size for lg_floor()"
+#endif
+}
+#else
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	assert(x != 0);
+
+	x |= (x >> 1);
+	x |= (x >> 2);
+	x |= (x >> 4);
+	x |= (x >> 8);
+	x |= (x >> 16);
+#if (LG_SIZEOF_PTR == 3)
+	x |= (x >> 32);
+#endif
+	if (x == SIZE_T_MAX) {
+		return (8 << LG_SIZEOF_PTR) - 1;
+	}
+	x++;
+	return ffs_zu(x) - 2;
+}
+#endif
+
+BIT_UTIL_INLINE unsigned
+lg_ceil(size_t x) {
+	return lg_floor(x) + ((x & (x - 1)) == 0 ? 0 : 1);
+}
+
+#undef BIT_UTIL_INLINE
+
+/* A compile-time version of lg_floor and lg_ceil. */
+#define LG_FLOOR_1(x) 0
+#define LG_FLOOR_2(x) (x < (1ULL << 1) ? LG_FLOOR_1(x) : 1 + LG_FLOOR_1(x >> 1))
+#define LG_FLOOR_4(x) (x < (1ULL << 2) ? LG_FLOOR_2(x) : 2 + LG_FLOOR_2(x >> 2))
+#define LG_FLOOR_8(x) (x < (1ULL << 4) ? LG_FLOOR_4(x) : 4 + LG_FLOOR_4(x >> 4))
+#define LG_FLOOR_16(x) (x < (1ULL << 8) ? LG_FLOOR_8(x) : 8 + LG_FLOOR_8(x >> 8))
+#define LG_FLOOR_32(x) (x < (1ULL << 16) ? LG_FLOOR_16(x) : 16 + LG_FLOOR_16(x >> 16))
+#define LG_FLOOR_64(x) (x < (1ULL << 32) ? LG_FLOOR_32(x) : 32 + LG_FLOOR_32(x >> 32))
+#if LG_SIZEOF_PTR == 2
+#  define LG_FLOOR(x) LG_FLOOR_32((x))
+#else
+#  define LG_FLOOR(x) LG_FLOOR_64((x))
+#endif
+
+#define LG_CEIL(x) (LG_FLOOR(x) + (((x) & ((x) - 1)) == 0 ? 0 : 1))
+
+#endif /* JEMALLOC_INTERNAL_BIT_UTIL_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/bitmap.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bitmap.h
new file mode 100644
index 0000000..c3f9cb4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/bitmap.h
@@ -0,0 +1,369 @@
+#ifndef JEMALLOC_INTERNAL_BITMAP_H
+#define JEMALLOC_INTERNAL_BITMAP_H
+
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/sc.h"
+
+typedef unsigned long bitmap_t;
+#define LG_SIZEOF_BITMAP	LG_SIZEOF_LONG
+
+/* Maximum bitmap bit count is 2^LG_BITMAP_MAXBITS. */
+#if LG_SLAB_MAXREGS > LG_CEIL(SC_NSIZES)
+/* Maximum bitmap bit count is determined by maximum regions per slab. */
+#  define LG_BITMAP_MAXBITS	LG_SLAB_MAXREGS
+#else
+/* Maximum bitmap bit count is determined by number of extent size classes. */
+#  define LG_BITMAP_MAXBITS	LG_CEIL(SC_NSIZES)
+#endif
+#define BITMAP_MAXBITS		(ZU(1) << LG_BITMAP_MAXBITS)
+
+/* Number of bits per group. */
+#define LG_BITMAP_GROUP_NBITS		(LG_SIZEOF_BITMAP + 3)
+#define BITMAP_GROUP_NBITS		(1U << LG_BITMAP_GROUP_NBITS)
+#define BITMAP_GROUP_NBITS_MASK		(BITMAP_GROUP_NBITS-1)
+
+/*
+ * Do some analysis on how big the bitmap is before we use a tree.  For a brute
+ * force linear search, if we would have to call ffs_lu() more than 2^3 times,
+ * use a tree instead.
+ */
+#if LG_BITMAP_MAXBITS - LG_BITMAP_GROUP_NBITS > 3
+#  define BITMAP_USE_TREE
+#endif
+
+/* Number of groups required to store a given number of bits. */
+#define BITMAP_BITS2GROUPS(nbits)					\
+    (((nbits) + BITMAP_GROUP_NBITS_MASK) >> LG_BITMAP_GROUP_NBITS)
+
+/*
+ * Number of groups required at a particular level for a given number of bits.
+ */
+#define BITMAP_GROUPS_L0(nbits)						\
+    BITMAP_BITS2GROUPS(nbits)
+#define BITMAP_GROUPS_L1(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(nbits))
+#define BITMAP_GROUPS_L2(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS((nbits))))
+#define BITMAP_GROUPS_L3(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(		\
+	BITMAP_BITS2GROUPS((nbits)))))
+#define BITMAP_GROUPS_L4(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(		\
+	BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS((nbits))))))
+
+/*
+ * Assuming the number of levels, number of groups required for a given number
+ * of bits.
+ */
+#define BITMAP_GROUPS_1_LEVEL(nbits)					\
+    BITMAP_GROUPS_L0(nbits)
+#define BITMAP_GROUPS_2_LEVEL(nbits)					\
+    (BITMAP_GROUPS_1_LEVEL(nbits) + BITMAP_GROUPS_L1(nbits))
+#define BITMAP_GROUPS_3_LEVEL(nbits)					\
+    (BITMAP_GROUPS_2_LEVEL(nbits) + BITMAP_GROUPS_L2(nbits))
+#define BITMAP_GROUPS_4_LEVEL(nbits)					\
+    (BITMAP_GROUPS_3_LEVEL(nbits) + BITMAP_GROUPS_L3(nbits))
+#define BITMAP_GROUPS_5_LEVEL(nbits)					\
+    (BITMAP_GROUPS_4_LEVEL(nbits) + BITMAP_GROUPS_L4(nbits))
+
+/*
+ * Maximum number of groups required to support LG_BITMAP_MAXBITS.
+ */
+#ifdef BITMAP_USE_TREE
+
+#if LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_1_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_1_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 2
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_2_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_2_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 3
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_3_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_3_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 4
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_4_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_4_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 5
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_5_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_5_LEVEL(BITMAP_MAXBITS)
+#else
+#  error "Unsupported bitmap size"
+#endif
+
+/*
+ * Maximum number of levels possible.  This could be statically computed based
+ * on LG_BITMAP_MAXBITS:
+ *
+ * #define BITMAP_MAX_LEVELS \
+ *     (LG_BITMAP_MAXBITS / LG_SIZEOF_BITMAP) \
+ *     + !!(LG_BITMAP_MAXBITS % LG_SIZEOF_BITMAP)
+ *
+ * However, that would not allow the generic BITMAP_INFO_INITIALIZER() macro, so
+ * instead hardcode BITMAP_MAX_LEVELS to the largest number supported by the
+ * various cascading macros.  The only additional cost this incurs is some
+ * unused trailing entries in bitmap_info_t structures; the bitmaps themselves
+ * are not impacted.
+ */
+#define BITMAP_MAX_LEVELS	5
+
+#define BITMAP_INFO_INITIALIZER(nbits) {				\
+	/* nbits. */							\
+	nbits,								\
+	/* nlevels. */							\
+	(BITMAP_GROUPS_L0(nbits) > BITMAP_GROUPS_L1(nbits)) +		\
+	    (BITMAP_GROUPS_L1(nbits) > BITMAP_GROUPS_L2(nbits)) +	\
+	    (BITMAP_GROUPS_L2(nbits) > BITMAP_GROUPS_L3(nbits)) +	\
+	    (BITMAP_GROUPS_L3(nbits) > BITMAP_GROUPS_L4(nbits)) + 1,	\
+	/* levels. */							\
+	{								\
+		{0},							\
+		{BITMAP_GROUPS_L0(nbits)},				\
+		{BITMAP_GROUPS_L1(nbits) + BITMAP_GROUPS_L0(nbits)},	\
+		{BITMAP_GROUPS_L2(nbits) + BITMAP_GROUPS_L1(nbits) +	\
+		    BITMAP_GROUPS_L0(nbits)},				\
+		{BITMAP_GROUPS_L3(nbits) + BITMAP_GROUPS_L2(nbits) +	\
+		    BITMAP_GROUPS_L1(nbits) + BITMAP_GROUPS_L0(nbits)},	\
+		{BITMAP_GROUPS_L4(nbits) + BITMAP_GROUPS_L3(nbits) +	\
+		     BITMAP_GROUPS_L2(nbits) + BITMAP_GROUPS_L1(nbits)	\
+		     + BITMAP_GROUPS_L0(nbits)}				\
+	}								\
+}
+
+#else /* BITMAP_USE_TREE */
+
+#define BITMAP_GROUPS(nbits)	BITMAP_BITS2GROUPS(nbits)
+#define BITMAP_GROUPS_MAX	BITMAP_BITS2GROUPS(BITMAP_MAXBITS)
+
+#define BITMAP_INFO_INITIALIZER(nbits) {				\
+	/* nbits. */							\
+	nbits,								\
+	/* ngroups. */							\
+	BITMAP_BITS2GROUPS(nbits)					\
+}
+
+#endif /* BITMAP_USE_TREE */
+
+typedef struct bitmap_level_s {
+	/* Offset of this level's groups within the array of groups. */
+	size_t group_offset;
+} bitmap_level_t;
+
+typedef struct bitmap_info_s {
+	/* Logical number of bits in bitmap (stored at bottom level). */
+	size_t nbits;
+
+#ifdef BITMAP_USE_TREE
+	/* Number of levels necessary for nbits. */
+	unsigned nlevels;
+
+	/*
+	 * Only the first (nlevels+1) elements are used, and levels are ordered
+	 * bottom to top (e.g. the bottom level is stored in levels[0]).
+	 */
+	bitmap_level_t levels[BITMAP_MAX_LEVELS+1];
+#else /* BITMAP_USE_TREE */
+	/* Number of groups necessary for nbits. */
+	size_t ngroups;
+#endif /* BITMAP_USE_TREE */
+} bitmap_info_t;
+
+void bitmap_info_init(bitmap_info_t *binfo, size_t nbits);
+void bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo, bool fill);
+size_t bitmap_size(const bitmap_info_t *binfo);
+
+static inline bool
+bitmap_full(bitmap_t *bitmap, const bitmap_info_t *binfo) {
+#ifdef BITMAP_USE_TREE
+	size_t rgoff = binfo->levels[binfo->nlevels].group_offset - 1;
+	bitmap_t rg = bitmap[rgoff];
+	/* The bitmap is full iff the root group is 0. */
+	return (rg == 0);
+#else
+	size_t i;
+
+	for (i = 0; i < binfo->ngroups; i++) {
+		if (bitmap[i] != 0) {
+			return false;
+		}
+	}
+	return true;
+#endif
+}
+
+static inline bool
+bitmap_get(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit) {
+	size_t goff;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	g = bitmap[goff];
+	return !(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+}
+
+static inline void
+bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit) {
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	assert(!bitmap_get(bitmap, binfo, bit));
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	assert(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+	g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(bitmap_get(bitmap, binfo, bit));
+#ifdef BITMAP_USE_TREE
+	/* Propagate group state transitions up the tree. */
+	if (g == 0) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			assert(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+			g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (g != 0) {
+				break;
+			}
+		}
+	}
+#endif
+}
+
+/* ffu: find first unset >= bit. */
+static inline size_t
+bitmap_ffu(const bitmap_t *bitmap, const bitmap_info_t *binfo, size_t min_bit) {
+	assert(min_bit < binfo->nbits);
+
+#ifdef BITMAP_USE_TREE
+	size_t bit = 0;
+	for (unsigned level = binfo->nlevels; level--;) {
+		size_t lg_bits_per_group = (LG_BITMAP_GROUP_NBITS * (level +
+		    1));
+		bitmap_t group = bitmap[binfo->levels[level].group_offset + (bit
+		    >> lg_bits_per_group)];
+		unsigned group_nmask = (unsigned)(((min_bit > bit) ? (min_bit -
+		    bit) : 0) >> (lg_bits_per_group - LG_BITMAP_GROUP_NBITS));
+		assert(group_nmask <= BITMAP_GROUP_NBITS);
+		bitmap_t group_mask = ~((1LU << group_nmask) - 1);
+		bitmap_t group_masked = group & group_mask;
+		if (group_masked == 0LU) {
+			if (group == 0LU) {
+				return binfo->nbits;
+			}
+			/*
+			 * min_bit was preceded by one or more unset bits in
+			 * this group, but there are no other unset bits in this
+			 * group.  Try again starting at the first bit of the
+			 * next sibling.  This will recurse at most once per
+			 * non-root level.
+			 */
+			size_t sib_base = bit + (ZU(1) << lg_bits_per_group);
+			assert(sib_base > min_bit);
+			assert(sib_base > bit);
+			if (sib_base >= binfo->nbits) {
+				return binfo->nbits;
+			}
+			return bitmap_ffu(bitmap, binfo, sib_base);
+		}
+		bit += ((size_t)(ffs_lu(group_masked) - 1)) <<
+		    (lg_bits_per_group - LG_BITMAP_GROUP_NBITS);
+	}
+	assert(bit >= min_bit);
+	assert(bit < binfo->nbits);
+	return bit;
+#else
+	size_t i = min_bit >> LG_BITMAP_GROUP_NBITS;
+	bitmap_t g = bitmap[i] & ~((1LU << (min_bit & BITMAP_GROUP_NBITS_MASK))
+	    - 1);
+	size_t bit;
+	do {
+		bit = ffs_lu(g);
+		if (bit != 0) {
+			return (i << LG_BITMAP_GROUP_NBITS) + (bit - 1);
+		}
+		i++;
+		g = bitmap[i];
+	} while (i < binfo->ngroups);
+	return binfo->nbits;
+#endif
+}
+
+/* sfu: set first unset. */
+static inline size_t
+bitmap_sfu(bitmap_t *bitmap, const bitmap_info_t *binfo) {
+	size_t bit;
+	bitmap_t g;
+	unsigned i;
+
+	assert(!bitmap_full(bitmap, binfo));
+
+#ifdef BITMAP_USE_TREE
+	i = binfo->nlevels - 1;
+	g = bitmap[binfo->levels[i].group_offset];
+	bit = ffs_lu(g) - 1;
+	while (i > 0) {
+		i--;
+		g = bitmap[binfo->levels[i].group_offset + bit];
+		bit = (bit << LG_BITMAP_GROUP_NBITS) + (ffs_lu(g) - 1);
+	}
+#else
+	i = 0;
+	g = bitmap[0];
+	while ((bit = ffs_lu(g)) == 0) {
+		i++;
+		g = bitmap[i];
+	}
+	bit = (i << LG_BITMAP_GROUP_NBITS) + (bit - 1);
+#endif
+	bitmap_set(bitmap, binfo, bit);
+	return bit;
+}
+
+static inline void
+bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit) {
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+	UNUSED bool propagate;
+
+	assert(bit < binfo->nbits);
+	assert(bitmap_get(bitmap, binfo, bit));
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	propagate = (g == 0);
+	assert((g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK))) == 0);
+	g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(!bitmap_get(bitmap, binfo, bit));
+#ifdef BITMAP_USE_TREE
+	/* Propagate group state transitions up the tree. */
+	if (propagate) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			propagate = (g == 0);
+			assert((g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)))
+			    == 0);
+			g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (!propagate) {
+				break;
+			}
+		}
+	}
+#endif /* BITMAP_USE_TREE */
+}
+
+#endif /* JEMALLOC_INTERNAL_BITMAP_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/cache_bin.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/cache_bin.h
new file mode 100644
index 0000000..d14556a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/cache_bin.h
@@ -0,0 +1,131 @@
+#ifndef JEMALLOC_INTERNAL_CACHE_BIN_H
+#define JEMALLOC_INTERNAL_CACHE_BIN_H
+
+#include "jemalloc/internal/ql.h"
+
+/*
+ * The cache_bins are the mechanism that the tcache and the arena use to
+ * communicate.  The tcache fills from and flushes to the arena by passing a
+ * cache_bin_t to fill/flush.  When the arena needs to pull stats from the
+ * tcaches associated with it, it does so by iterating over its
+ * cache_bin_array_descriptor_t objects and reading out per-bin stats it
+ * contains.  This makes it so that the arena need not know about the existence
+ * of the tcache at all.
+ */
+
+
+/*
+ * The count of the number of cached allocations in a bin.  We make this signed
+ * so that negative numbers can encode "invalid" states (e.g. a low water mark
+ * of -1 for a cache that has been depleted).
+ */
+typedef int32_t cache_bin_sz_t;
+
+typedef struct cache_bin_stats_s cache_bin_stats_t;
+struct cache_bin_stats_s {
+	/*
+	 * Number of allocation requests that corresponded to the size of this
+	 * bin.
+	 */
+	uint64_t nrequests;
+};
+
+/*
+ * Read-only information associated with each element of tcache_t's tbins array
+ * is stored separately, mainly to reduce memory usage.
+ */
+typedef struct cache_bin_info_s cache_bin_info_t;
+struct cache_bin_info_s {
+	/* Upper limit on ncached. */
+	cache_bin_sz_t ncached_max;
+};
+
+typedef struct cache_bin_s cache_bin_t;
+struct cache_bin_s {
+	/* Min # cached since last GC. */
+	cache_bin_sz_t low_water;
+	/* # of cached objects. */
+	cache_bin_sz_t ncached;
+	/*
+	 * ncached and stats are both modified frequently.  Let's keep them
+	 * close so that they have a higher chance of being on the same
+	 * cacheline, thus less write-backs.
+	 */
+	cache_bin_stats_t tstats;
+	/*
+	 * Stack of available objects.
+	 *
+	 * To make use of adjacent cacheline prefetch, the items in the avail
+	 * stack goes to higher address for newer allocations.  avail points
+	 * just above the available space, which means that
+	 * avail[-ncached, ... -1] are available items and the lowest item will
+	 * be allocated first.
+	 */
+	void **avail;
+};
+
+typedef struct cache_bin_array_descriptor_s cache_bin_array_descriptor_t;
+struct cache_bin_array_descriptor_s {
+	/*
+	 * The arena keeps a list of the cache bins associated with it, for
+	 * stats collection.
+	 */
+	ql_elm(cache_bin_array_descriptor_t) link;
+	/* Pointers to the tcache bins. */
+	cache_bin_t *bins_small;
+	cache_bin_t *bins_large;
+};
+
+static inline void
+cache_bin_array_descriptor_init(cache_bin_array_descriptor_t *descriptor,
+    cache_bin_t *bins_small, cache_bin_t *bins_large) {
+	ql_elm_new(descriptor, link);
+	descriptor->bins_small = bins_small;
+	descriptor->bins_large = bins_large;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+cache_bin_alloc_easy(cache_bin_t *bin, bool *success) {
+	void *ret;
+
+	bin->ncached--;
+
+	/*
+	 * Check for both bin->ncached == 0 and ncached < low_water
+	 * in a single branch.
+	 */
+	if (unlikely(bin->ncached <= bin->low_water)) {
+		bin->low_water = bin->ncached;
+		if (bin->ncached == -1) {
+			bin->ncached = 0;
+			*success = false;
+			return NULL;
+		}
+	}
+
+	/*
+	 * success (instead of ret) should be checked upon the return of this
+	 * function.  We avoid checking (ret == NULL) because there is never a
+	 * null stored on the avail stack (which is unknown to the compiler),
+	 * and eagerly checking ret would cause pipeline stall (waiting for the
+	 * cacheline).
+	 */
+	*success = true;
+	ret = *(bin->avail - (bin->ncached + 1));
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+cache_bin_dalloc_easy(cache_bin_t *bin, cache_bin_info_t *bin_info, void *ptr) {
+	if (unlikely(bin->ncached == bin_info->ncached_max)) {
+		return false;
+	}
+	assert(bin->ncached < bin_info->ncached_max);
+	bin->ncached++;
+	*(bin->avail - bin->ncached) = ptr;
+
+	return true;
+}
+
+#endif /* JEMALLOC_INTERNAL_CACHE_BIN_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/ckh.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ckh.h
new file mode 100644
index 0000000..7b3850b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ckh.h
@@ -0,0 +1,101 @@
+#ifndef JEMALLOC_INTERNAL_CKH_H
+#define JEMALLOC_INTERNAL_CKH_H
+
+#include "jemalloc/internal/tsd.h"
+
+/* Cuckoo hashing implementation.  Skip to the end for the interface. */
+
+/******************************************************************************/
+/* INTERNAL DEFINITIONS -- IGNORE */
+/******************************************************************************/
+
+/* Maintain counters used to get an idea of performance. */
+/* #define CKH_COUNT */
+/* Print counter values in ckh_delete() (requires CKH_COUNT). */
+/* #define CKH_VERBOSE */
+
+/*
+ * There are 2^LG_CKH_BUCKET_CELLS cells in each hash table bucket.  Try to fit
+ * one bucket per L1 cache line.
+ */
+#define LG_CKH_BUCKET_CELLS (LG_CACHELINE - LG_SIZEOF_PTR - 1)
+
+/* Typedefs to allow easy function pointer passing. */
+typedef void ckh_hash_t (const void *, size_t[2]);
+typedef bool ckh_keycomp_t (const void *, const void *);
+
+/* Hash table cell. */
+typedef struct {
+	const void *key;
+	const void *data;
+} ckhc_t;
+
+/* The hash table itself. */
+typedef struct {
+#ifdef CKH_COUNT
+	/* Counters used to get an idea of performance. */
+	uint64_t ngrows;
+	uint64_t nshrinks;
+	uint64_t nshrinkfails;
+	uint64_t ninserts;
+	uint64_t nrelocs;
+#endif
+
+	/* Used for pseudo-random number generation. */
+	uint64_t prng_state;
+
+	/* Total number of items. */
+	size_t count;
+
+	/*
+	 * Minimum and current number of hash table buckets.  There are
+	 * 2^LG_CKH_BUCKET_CELLS cells per bucket.
+	 */
+	unsigned lg_minbuckets;
+	unsigned lg_curbuckets;
+
+	/* Hash and comparison functions. */
+	ckh_hash_t *hash;
+	ckh_keycomp_t *keycomp;
+
+	/* Hash table with 2^lg_curbuckets buckets. */
+	ckhc_t *tab;
+} ckh_t;
+
+/******************************************************************************/
+/* BEGIN PUBLIC API */
+/******************************************************************************/
+
+/* Lifetime management.  Minitems is the initial capacity. */
+bool ckh_new(tsd_t *tsd, ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
+    ckh_keycomp_t *keycomp);
+void ckh_delete(tsd_t *tsd, ckh_t *ckh);
+
+/* Get the number of elements in the set. */
+size_t ckh_count(ckh_t *ckh);
+
+/*
+ * To iterate over the elements in the table, initialize *tabind to 0 and call
+ * this function until it returns true.  Each call that returns false will
+ * update *key and *data to the next element in the table, assuming the pointers
+ * are non-NULL.
+ */
+bool ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data);
+
+/*
+ * Basic hash table operations -- insert, removal, lookup.  For ckh_remove and
+ * ckh_search, key or data can be NULL.  The hash-table only stores pointers to
+ * the key and value, and doesn't do any lifetime management.
+ */
+bool ckh_insert(tsd_t *tsd, ckh_t *ckh, const void *key, const void *data);
+bool ckh_remove(tsd_t *tsd, ckh_t *ckh, const void *searchkey, void **key,
+    void **data);
+bool ckh_search(ckh_t *ckh, const void *searchkey, void **key, void **data);
+
+/* Some useful hash and comparison functions for strings and pointers. */
+void ckh_string_hash(const void *key, size_t r_hash[2]);
+bool ckh_string_keycomp(const void *k1, const void *k2);
+void ckh_pointer_hash(const void *key, size_t r_hash[2]);
+bool ckh_pointer_keycomp(const void *k1, const void *k2);
+
+#endif /* JEMALLOC_INTERNAL_CKH_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/ctl.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ctl.h
new file mode 100644
index 0000000..1d1aacc
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ctl.h
@@ -0,0 +1,134 @@
+#ifndef JEMALLOC_INTERNAL_CTL_H
+#define JEMALLOC_INTERNAL_CTL_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/stats.h"
+
+/* Maximum ctl tree depth. */
+#define CTL_MAX_DEPTH	7
+
+typedef struct ctl_node_s {
+	bool named;
+} ctl_node_t;
+
+typedef struct ctl_named_node_s {
+	ctl_node_t node;
+	const char *name;
+	/* If (nchildren == 0), this is a terminal node. */
+	size_t nchildren;
+	const ctl_node_t *children;
+	int (*ctl)(tsd_t *, const size_t *, size_t, void *, size_t *, void *,
+	    size_t);
+} ctl_named_node_t;
+
+typedef struct ctl_indexed_node_s {
+	struct ctl_node_s node;
+	const ctl_named_node_t *(*index)(tsdn_t *, const size_t *, size_t,
+	    size_t);
+} ctl_indexed_node_t;
+
+typedef struct ctl_arena_stats_s {
+	arena_stats_t astats;
+
+	/* Aggregate stats for small size classes, based on bin stats. */
+	size_t allocated_small;
+	uint64_t nmalloc_small;
+	uint64_t ndalloc_small;
+	uint64_t nrequests_small;
+	uint64_t nfills_small;
+	uint64_t nflushes_small;
+
+	bin_stats_t bstats[SC_NBINS];
+	arena_stats_large_t lstats[SC_NSIZES - SC_NBINS];
+	arena_stats_extents_t estats[SC_NPSIZES];
+} ctl_arena_stats_t;
+
+typedef struct ctl_stats_s {
+	size_t allocated;
+	size_t active;
+	size_t metadata;
+	size_t metadata_thp;
+	size_t resident;
+	size_t mapped;
+	size_t retained;
+
+	background_thread_stats_t background_thread;
+	mutex_prof_data_t mutex_prof_data[mutex_prof_num_global_mutexes];
+} ctl_stats_t;
+
+typedef struct ctl_arena_s ctl_arena_t;
+struct ctl_arena_s {
+	unsigned arena_ind;
+	bool initialized;
+	ql_elm(ctl_arena_t) destroyed_link;
+
+	/* Basic stats, supported even if !config_stats. */
+	unsigned nthreads;
+	const char *dss;
+	ssize_t dirty_decay_ms;
+	ssize_t muzzy_decay_ms;
+	size_t pactive;
+	size_t pdirty;
+	size_t pmuzzy;
+
+	/* NULL if !config_stats. */
+	ctl_arena_stats_t *astats;
+};
+
+typedef struct ctl_arenas_s {
+	uint64_t epoch;
+	unsigned narenas;
+	ql_head(ctl_arena_t) destroyed;
+
+	/*
+	 * Element 0 corresponds to merged stats for extant arenas (accessed via
+	 * MALLCTL_ARENAS_ALL), element 1 corresponds to merged stats for
+	 * destroyed arenas (accessed via MALLCTL_ARENAS_DESTROYED), and the
+	 * remaining MALLOCX_ARENA_LIMIT elements correspond to arenas.
+	 */
+	ctl_arena_t *arenas[2 + MALLOCX_ARENA_LIMIT];
+} ctl_arenas_t;
+
+int ctl_byname(tsd_t *tsd, const char *name, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen);
+int ctl_nametomib(tsd_t *tsd, const char *name, size_t *mibp, size_t *miblenp);
+
+int ctl_bymib(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen);
+bool ctl_boot(void);
+void ctl_prefork(tsdn_t *tsdn);
+void ctl_postfork_parent(tsdn_t *tsdn);
+void ctl_postfork_child(tsdn_t *tsdn);
+
+#define xmallctl(name, oldp, oldlenp, newp, newlen) do {		\
+	if (je_mallctl(name, oldp, oldlenp, newp, newlen)		\
+	    != 0) {							\
+		malloc_printf(						\
+		    "<jemalloc>: Failure in xmallctl(\"%s\", ...)\n",	\
+		    name);						\
+		abort();						\
+	}								\
+} while (0)
+
+#define xmallctlnametomib(name, mibp, miblenp) do {			\
+	if (je_mallctlnametomib(name, mibp, miblenp) != 0) {		\
+		malloc_printf("<jemalloc>: Failure in "			\
+		    "xmallctlnametomib(\"%s\", ...)\n", name);		\
+		abort();						\
+	}								\
+} while (0)
+
+#define xmallctlbymib(mib, miblen, oldp, oldlenp, newp, newlen) do {	\
+	if (je_mallctlbymib(mib, miblen, oldp, oldlenp, newp,		\
+	    newlen) != 0) {						\
+		malloc_write(						\
+		    "<jemalloc>: Failure in xmallctlbymib()\n");	\
+		abort();						\
+	}								\
+} while (0)
+
+#endif /* JEMALLOC_INTERNAL_CTL_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/div.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/div.h
new file mode 100644
index 0000000..aebae93
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/div.h
@@ -0,0 +1,41 @@
+#ifndef JEMALLOC_INTERNAL_DIV_H
+#define JEMALLOC_INTERNAL_DIV_H
+
+#include "jemalloc/internal/assert.h"
+
+/*
+ * This module does the division that computes the index of a region in a slab,
+ * given its offset relative to the base.
+ * That is, given a divisor d, an n = i * d (all integers), we'll return i.
+ * We do some pre-computation to do this more quickly than a CPU division
+ * instruction.
+ * We bound n < 2^32, and don't support dividing by one.
+ */
+
+typedef struct div_info_s div_info_t;
+struct div_info_s {
+	uint32_t magic;
+#ifdef JEMALLOC_DEBUG
+	size_t d;
+#endif
+};
+
+void div_init(div_info_t *div_info, size_t divisor);
+
+static inline size_t
+div_compute(div_info_t *div_info, size_t n) {
+	assert(n <= (uint32_t)-1);
+	/*
+	 * This generates, e.g. mov; imul; shr on x86-64. On a 32-bit machine,
+	 * the compilers I tried were all smart enough to turn this into the
+	 * appropriate "get the high 32 bits of the result of a multiply" (e.g.
+	 * mul; mov edx eax; on x86, umull on arm, etc.).
+	 */
+	size_t i = ((uint64_t)n * (uint64_t)div_info->magic) >> 32;
+#ifdef JEMALLOC_DEBUG
+	assert(i * div_info->d == n);
+#endif
+	return i;
+}
+
+#endif /* JEMALLOC_INTERNAL_DIV_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/emitter.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/emitter.h
new file mode 100644
index 0000000..d140d38
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/emitter.h
@@ -0,0 +1,486 @@
+#ifndef JEMALLOC_INTERNAL_EMITTER_H
+#define JEMALLOC_INTERNAL_EMITTER_H
+
+#include "jemalloc/internal/ql.h"
+
+typedef enum emitter_output_e emitter_output_t;
+enum emitter_output_e {
+	emitter_output_json,
+	emitter_output_table
+};
+
+typedef enum emitter_justify_e emitter_justify_t;
+enum emitter_justify_e {
+	emitter_justify_left,
+	emitter_justify_right,
+	/* Not for users; just to pass to internal functions. */
+	emitter_justify_none
+};
+
+typedef enum emitter_type_e emitter_type_t;
+enum emitter_type_e {
+	emitter_type_bool,
+	emitter_type_int,
+	emitter_type_unsigned,
+	emitter_type_uint32,
+	emitter_type_uint64,
+	emitter_type_size,
+	emitter_type_ssize,
+	emitter_type_string,
+	/*
+	 * A title is a column title in a table; it's just a string, but it's
+	 * not quoted.
+	 */
+	emitter_type_title,
+};
+
+typedef struct emitter_col_s emitter_col_t;
+struct emitter_col_s {
+	/* Filled in by the user. */
+	emitter_justify_t justify;
+	int width;
+	emitter_type_t type;
+	union {
+		bool bool_val;
+		int int_val;
+		unsigned unsigned_val;
+		uint32_t uint32_val;
+		uint32_t uint32_t_val;
+		uint64_t uint64_val;
+		uint64_t uint64_t_val;
+		size_t size_val;
+		ssize_t ssize_val;
+		const char *str_val;
+	};
+
+	/* Filled in by initialization. */
+	ql_elm(emitter_col_t) link;
+};
+
+typedef struct emitter_row_s emitter_row_t;
+struct emitter_row_s {
+	ql_head(emitter_col_t) cols;
+};
+
+typedef struct emitter_s emitter_t;
+struct emitter_s {
+	emitter_output_t output;
+	/* The output information. */
+	void (*write_cb)(void *, const char *);
+	void *cbopaque;
+	int nesting_depth;
+	/* True if we've already emitted a value at the given depth. */
+	bool item_at_depth;
+	/* True if we emitted a key and will emit corresponding value next. */
+	bool emitted_key;
+};
+
+/* Internal convenience function.  Write to the emitter the given string. */
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static inline void
+emitter_printf(emitter_t *emitter, const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(emitter->write_cb, emitter->cbopaque, format, ap);
+	va_end(ap);
+}
+
+static inline const char * JEMALLOC_FORMAT_ARG(3)
+emitter_gen_fmt(char *out_fmt, size_t out_size, const char *fmt_specifier,
+    emitter_justify_t justify, int width) {
+	size_t written;
+	fmt_specifier++;
+	if (justify == emitter_justify_none) {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%%s", fmt_specifier);
+	} else if (justify == emitter_justify_left) {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%-%d%s", width, fmt_specifier);
+	} else {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%%d%s", width, fmt_specifier);
+	}
+	/* Only happens in case of bad format string, which *we* choose. */
+	assert(written <  out_size);
+	return out_fmt;
+}
+
+/*
+ * Internal.  Emit the given value type in the relevant encoding (so that the
+ * bool true gets mapped to json "true", but the string "true" gets mapped to
+ * json "\"true\"", for instance.
+ *
+ * Width is ignored if justify is emitter_justify_none.
+ */
+static inline void
+emitter_print_value(emitter_t *emitter, emitter_justify_t justify, int width,
+    emitter_type_t value_type, const void *value) {
+	size_t str_written;
+#define BUF_SIZE 256
+#define FMT_SIZE 10
+	/*
+	 * We dynamically generate a format string to emit, to let us use the
+	 * snprintf machinery.  This is kinda hacky, but gets the job done
+	 * quickly without having to think about the various snprintf edge
+	 * cases.
+	 */
+	char fmt[FMT_SIZE];
+	char buf[BUF_SIZE];
+
+#define EMIT_SIMPLE(type, format)					\
+	emitter_printf(emitter,						\
+	    emitter_gen_fmt(fmt, FMT_SIZE, format, justify, width),	\
+	    *(const type *)value);
+
+	switch (value_type) {
+	case emitter_type_bool:
+		emitter_printf(emitter,
+		    emitter_gen_fmt(fmt, FMT_SIZE, "%s", justify, width),
+		    *(const bool *)value ?  "true" : "false");
+		break;
+	case emitter_type_int:
+		EMIT_SIMPLE(int, "%d")
+		break;
+	case emitter_type_unsigned:
+		EMIT_SIMPLE(unsigned, "%u")
+		break;
+	case emitter_type_ssize:
+		EMIT_SIMPLE(ssize_t, "%zd")
+		break;
+	case emitter_type_size:
+		EMIT_SIMPLE(size_t, "%zu")
+		break;
+	case emitter_type_string:
+		str_written = malloc_snprintf(buf, BUF_SIZE, "\"%s\"",
+		    *(const char *const *)value);
+		/*
+		 * We control the strings we output; we shouldn't get anything
+		 * anywhere near the fmt size.
+		 */
+		assert(str_written < BUF_SIZE);
+		emitter_printf(emitter,
+		    emitter_gen_fmt(fmt, FMT_SIZE, "%s", justify, width), buf);
+		break;
+	case emitter_type_uint32:
+		EMIT_SIMPLE(uint32_t, "%" FMTu32)
+		break;
+	case emitter_type_uint64:
+		EMIT_SIMPLE(uint64_t, "%" FMTu64)
+		break;
+	case emitter_type_title:
+		EMIT_SIMPLE(char *const, "%s");
+		break;
+	default:
+		unreachable();
+	}
+#undef BUF_SIZE
+#undef FMT_SIZE
+}
+
+
+/* Internal functions.  In json mode, tracks nesting state. */
+static inline void
+emitter_nest_inc(emitter_t *emitter) {
+	emitter->nesting_depth++;
+	emitter->item_at_depth = false;
+}
+
+static inline void
+emitter_nest_dec(emitter_t *emitter) {
+	emitter->nesting_depth--;
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_indent(emitter_t *emitter) {
+	int amount = emitter->nesting_depth;
+	const char *indent_str;
+	if (emitter->output == emitter_output_json) {
+		indent_str = "\t";
+	} else {
+		amount *= 2;
+		indent_str = " ";
+	}
+	for (int i = 0; i < amount; i++) {
+		emitter_printf(emitter, "%s", indent_str);
+	}
+}
+
+static inline void
+emitter_json_key_prefix(emitter_t *emitter) {
+	if (emitter->emitted_key) {
+		emitter->emitted_key = false;
+		return;
+	}
+	emitter_printf(emitter, "%s\n", emitter->item_at_depth ? "," : "");
+	emitter_indent(emitter);
+}
+
+/******************************************************************************/
+/* Public functions for emitter_t. */
+
+static inline void
+emitter_init(emitter_t *emitter, emitter_output_t emitter_output,
+    void (*write_cb)(void *, const char *), void *cbopaque) {
+	emitter->output = emitter_output;
+	emitter->write_cb = write_cb;
+	emitter->cbopaque = cbopaque;
+	emitter->item_at_depth = false;
+	emitter->emitted_key = false;
+	emitter->nesting_depth = 0;
+}
+
+/******************************************************************************/
+/* JSON public API. */
+
+/*
+ * Emits a key (e.g. as appears in an object). The next json entity emitted will
+ * be the corresponding value.
+ */
+static inline void
+emitter_json_key(emitter_t *emitter, const char *json_key) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "\"%s\": ", json_key);
+		emitter->emitted_key = true;
+	}
+}
+
+static inline void
+emitter_json_value(emitter_t *emitter, emitter_type_t value_type,
+    const void *value) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_print_value(emitter, emitter_justify_none, -1,
+		    value_type, value);
+		emitter->item_at_depth = true;
+	}
+}
+
+/* Shorthand for calling emitter_json_key and then emitter_json_value. */
+static inline void
+emitter_json_kv(emitter_t *emitter, const char *json_key,
+    emitter_type_t value_type, const void *value) {
+	emitter_json_key(emitter, json_key);
+	emitter_json_value(emitter, value_type, value);
+}
+
+static inline void
+emitter_json_array_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "[");
+		emitter_nest_inc(emitter);
+	}
+}
+
+/* Shorthand for calling emitter_json_key and then emitter_json_array_begin. */
+static inline void
+emitter_json_array_kv_begin(emitter_t *emitter, const char *json_key) {
+	emitter_json_key(emitter, json_key);
+	emitter_json_array_begin(emitter);
+}
+
+static inline void
+emitter_json_array_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n");
+		emitter_indent(emitter);
+		emitter_printf(emitter, "]");
+	}
+}
+
+static inline void
+emitter_json_object_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "{");
+		emitter_nest_inc(emitter);
+	}
+}
+
+/* Shorthand for calling emitter_json_key and then emitter_json_object_begin. */
+static inline void
+emitter_json_object_kv_begin(emitter_t *emitter, const char *json_key) {
+	emitter_json_key(emitter, json_key);
+	emitter_json_object_begin(emitter);
+}
+
+static inline void
+emitter_json_object_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n");
+		emitter_indent(emitter);
+		emitter_printf(emitter, "}");
+	}
+}
+
+
+/******************************************************************************/
+/* Table public API. */
+
+static inline void
+emitter_table_dict_begin(emitter_t *emitter, const char *table_key) {
+	if (emitter->output == emitter_output_table) {
+		emitter_indent(emitter);
+		emitter_printf(emitter, "%s\n", table_key);
+		emitter_nest_inc(emitter);
+	}
+}
+
+static inline void
+emitter_table_dict_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_table) {
+		emitter_nest_dec(emitter);
+	}
+}
+
+static inline void
+emitter_table_kv_note(emitter_t *emitter, const char *table_key,
+    emitter_type_t value_type, const void *value,
+    const char *table_note_key, emitter_type_t table_note_value_type,
+    const void *table_note_value) {
+	if (emitter->output == emitter_output_table) {
+		emitter_indent(emitter);
+		emitter_printf(emitter, "%s: ", table_key);
+		emitter_print_value(emitter, emitter_justify_none, -1,
+		    value_type, value);
+		if (table_note_key != NULL) {
+			emitter_printf(emitter, " (%s: ", table_note_key);
+			emitter_print_value(emitter, emitter_justify_none, -1,
+			    table_note_value_type, table_note_value);
+			emitter_printf(emitter, ")");
+		}
+		emitter_printf(emitter, "\n");
+	}
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_table_kv(emitter_t *emitter, const char *table_key,
+    emitter_type_t value_type, const void *value) {
+	emitter_table_kv_note(emitter, table_key, value_type, value, NULL,
+	    emitter_type_bool, NULL);
+}
+
+
+/* Write to the emitter the given string, but only in table mode. */
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static inline void
+emitter_table_printf(emitter_t *emitter, const char *format, ...) {
+	if (emitter->output == emitter_output_table) {
+		va_list ap;
+		va_start(ap, format);
+		malloc_vcprintf(emitter->write_cb, emitter->cbopaque, format, ap);
+		va_end(ap);
+	}
+}
+
+static inline void
+emitter_table_row(emitter_t *emitter, emitter_row_t *row) {
+	if (emitter->output != emitter_output_table) {
+		return;
+	}
+	emitter_col_t *col;
+	ql_foreach(col, &row->cols, link) {
+		emitter_print_value(emitter, col->justify, col->width,
+		    col->type, (const void *)&col->bool_val);
+	}
+	emitter_table_printf(emitter, "\n");
+}
+
+static inline void
+emitter_row_init(emitter_row_t *row) {
+	ql_new(&row->cols);
+}
+
+static inline void
+emitter_col_init(emitter_col_t *col, emitter_row_t *row) {
+	ql_elm_new(col, link);
+	ql_tail_insert(&row->cols, col, link);
+}
+
+
+/******************************************************************************/
+/*
+ * Generalized public API. Emits using either JSON or table, according to
+ * settings in the emitter_t. */
+
+/*
+ * Note emits a different kv pair as well, but only in table mode.  Omits the
+ * note if table_note_key is NULL.
+ */
+static inline void
+emitter_kv_note(emitter_t *emitter, const char *json_key, const char *table_key,
+    emitter_type_t value_type, const void *value,
+    const char *table_note_key, emitter_type_t table_note_value_type,
+    const void *table_note_value) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key(emitter, json_key);
+		emitter_json_value(emitter, value_type, value);
+	} else {
+		emitter_table_kv_note(emitter, table_key, value_type, value,
+		    table_note_key, table_note_value_type, table_note_value);
+	}
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_kv(emitter_t *emitter, const char *json_key, const char *table_key,
+    emitter_type_t value_type, const void *value) {
+	emitter_kv_note(emitter, json_key, table_key, value_type, value, NULL,
+	    emitter_type_bool, NULL);
+}
+
+static inline void
+emitter_dict_begin(emitter_t *emitter, const char *json_key,
+    const char *table_header) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key(emitter, json_key);
+		emitter_json_object_begin(emitter);
+	} else {
+		emitter_table_dict_begin(emitter, table_header);
+	}
+}
+
+static inline void
+emitter_dict_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_object_end(emitter);
+	} else {
+		emitter_table_dict_end(emitter);
+	}
+}
+
+static inline void
+emitter_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth == 0);
+		emitter_printf(emitter, "{");
+		emitter_nest_inc(emitter);
+	} else {
+		/*
+		 * This guarantees that we always call write_cb at least once.
+		 * This is useful if some invariant is established by each call
+		 * to write_cb, but doesn't hold initially: e.g., some buffer
+		 * holds a null-terminated string.
+		 */
+		emitter_printf(emitter, "%s", "");
+	}
+}
+
+static inline void
+emitter_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth == 1);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n}\n");
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_EMITTER_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_dss.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_dss.h
new file mode 100644
index 0000000..e8f02ce
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_dss.h
@@ -0,0 +1,26 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_DSS_H
+#define JEMALLOC_INTERNAL_EXTENT_DSS_H
+
+typedef enum {
+	dss_prec_disabled  = 0,
+	dss_prec_primary   = 1,
+	dss_prec_secondary = 2,
+
+	dss_prec_limit     = 3
+} dss_prec_t;
+#define DSS_PREC_DEFAULT dss_prec_secondary
+#define DSS_DEFAULT "secondary"
+
+extern const char *dss_prec_names[];
+
+extern const char *opt_dss;
+
+dss_prec_t extent_dss_prec_get(void);
+bool extent_dss_prec_set(dss_prec_t dss_prec);
+void *extent_alloc_dss(tsdn_t *tsdn, arena_t *arena, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit);
+bool extent_in_dss(void *addr);
+bool extent_dss_mergeable(void *addr_a, void *addr_b);
+void extent_dss_boot(void);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_DSS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_externs.h
new file mode 100644
index 0000000..8aba576
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_externs.h
@@ -0,0 +1,83 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTENT_EXTERNS_H
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/rtree.h"
+
+extern size_t opt_lg_extent_max_active_fit;
+
+extern rtree_t extents_rtree;
+extern const extent_hooks_t extent_hooks_default;
+extern mutex_pool_t extent_mutex_pool;
+
+extent_t *extent_alloc(tsdn_t *tsdn, arena_t *arena);
+void extent_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *extent);
+
+extent_hooks_t *extent_hooks_get(arena_t *arena);
+extent_hooks_t *extent_hooks_set(tsd_t *tsd, arena_t *arena,
+    extent_hooks_t *extent_hooks);
+
+#ifdef JEMALLOC_JET
+size_t extent_size_quantize_floor(size_t size);
+size_t extent_size_quantize_ceil(size_t size);
+#endif
+
+ph_proto(, extent_avail_, extent_tree_t, extent_t)
+ph_proto(, extent_heap_, extent_heap_t, extent_t)
+
+bool extents_init(tsdn_t *tsdn, extents_t *extents, extent_state_t state,
+    bool delay_coalesce);
+extent_state_t extents_state_get(const extents_t *extents);
+size_t extents_npages_get(extents_t *extents);
+/* Get the number of extents in the given page size index. */
+size_t extents_nextents_get(extents_t *extents, pszind_t ind);
+/* Get the sum total bytes of the extents in the given page size index. */
+size_t extents_nbytes_get(extents_t *extents, pszind_t ind);
+extent_t *extents_alloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, void *new_addr,
+    size_t size, size_t pad, size_t alignment, bool slab, szind_t szind,
+    bool *zero, bool *commit);
+void extents_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, extent_t *extent);
+extent_t *extents_evict(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, size_t npages_min);
+void extents_prefork(tsdn_t *tsdn, extents_t *extents);
+void extents_postfork_parent(tsdn_t *tsdn, extents_t *extents);
+void extents_postfork_child(tsdn_t *tsdn, extents_t *extents);
+extent_t *extent_alloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit);
+void extent_dalloc_gap(tsdn_t *tsdn, arena_t *arena, extent_t *extent);
+void extent_dalloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+void extent_destroy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+bool extent_commit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_decommit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_purge_lazy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_purge_forced_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+extent_t *extent_split_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b);
+bool extent_merge_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b);
+
+bool extent_boot(void);
+
+void extent_util_stats_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size);
+void extent_util_stats_verbose_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size,
+    size_t *bin_nfree, size_t *bin_nregs, void **slabcur_addr);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_inlines.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_inlines.h
new file mode 100644
index 0000000..77fa4c4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_inlines.h
@@ -0,0 +1,501 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_INLINES_H
+#define JEMALLOC_INTERNAL_EXTENT_INLINES_H
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/sz.h"
+
+static inline void
+extent_lock(tsdn_t *tsdn, extent_t *extent) {
+	assert(extent != NULL);
+	mutex_pool_lock(tsdn, &extent_mutex_pool, (uintptr_t)extent);
+}
+
+static inline void
+extent_unlock(tsdn_t *tsdn, extent_t *extent) {
+	assert(extent != NULL);
+	mutex_pool_unlock(tsdn, &extent_mutex_pool, (uintptr_t)extent);
+}
+
+static inline void
+extent_lock2(tsdn_t *tsdn, extent_t *extent1, extent_t *extent2) {
+	assert(extent1 != NULL && extent2 != NULL);
+	mutex_pool_lock2(tsdn, &extent_mutex_pool, (uintptr_t)extent1,
+	    (uintptr_t)extent2);
+}
+
+static inline void
+extent_unlock2(tsdn_t *tsdn, extent_t *extent1, extent_t *extent2) {
+	assert(extent1 != NULL && extent2 != NULL);
+	mutex_pool_unlock2(tsdn, &extent_mutex_pool, (uintptr_t)extent1,
+	    (uintptr_t)extent2);
+}
+
+static inline unsigned
+extent_arena_ind_get(const extent_t *extent) {
+	unsigned arena_ind = (unsigned)((extent->e_bits &
+	    EXTENT_BITS_ARENA_MASK) >> EXTENT_BITS_ARENA_SHIFT);
+	assert(arena_ind < MALLOCX_ARENA_LIMIT);
+
+	return arena_ind;
+}
+
+static inline arena_t *
+extent_arena_get(const extent_t *extent) {
+	unsigned arena_ind = extent_arena_ind_get(extent);
+
+	return (arena_t *)atomic_load_p(&arenas[arena_ind], ATOMIC_ACQUIRE);
+}
+
+static inline szind_t
+extent_szind_get_maybe_invalid(const extent_t *extent) {
+	szind_t szind = (szind_t)((extent->e_bits & EXTENT_BITS_SZIND_MASK) >>
+	    EXTENT_BITS_SZIND_SHIFT);
+	assert(szind <= SC_NSIZES);
+	return szind;
+}
+
+static inline szind_t
+extent_szind_get(const extent_t *extent) {
+	szind_t szind = extent_szind_get_maybe_invalid(extent);
+	assert(szind < SC_NSIZES); /* Never call when "invalid". */
+	return szind;
+}
+
+static inline size_t
+extent_usize_get(const extent_t *extent) {
+	return sz_index2size(extent_szind_get(extent));
+}
+
+static inline unsigned
+extent_binshard_get(const extent_t *extent) {
+	unsigned binshard = (unsigned)((extent->e_bits &
+	    EXTENT_BITS_BINSHARD_MASK) >> EXTENT_BITS_BINSHARD_SHIFT);
+	assert(binshard < bin_infos[extent_szind_get(extent)].n_shards);
+	return binshard;
+}
+
+static inline size_t
+extent_sn_get(const extent_t *extent) {
+	return (size_t)((extent->e_bits & EXTENT_BITS_SN_MASK) >>
+	    EXTENT_BITS_SN_SHIFT);
+}
+
+static inline extent_state_t
+extent_state_get(const extent_t *extent) {
+	return (extent_state_t)((extent->e_bits & EXTENT_BITS_STATE_MASK) >>
+	    EXTENT_BITS_STATE_SHIFT);
+}
+
+static inline bool
+extent_zeroed_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_ZEROED_MASK) >>
+	    EXTENT_BITS_ZEROED_SHIFT);
+}
+
+static inline bool
+extent_committed_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_COMMITTED_MASK) >>
+	    EXTENT_BITS_COMMITTED_SHIFT);
+}
+
+static inline bool
+extent_dumpable_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_DUMPABLE_MASK) >>
+	    EXTENT_BITS_DUMPABLE_SHIFT);
+}
+
+static inline bool
+extent_slab_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_SLAB_MASK) >>
+	    EXTENT_BITS_SLAB_SHIFT);
+}
+
+static inline unsigned
+extent_nfree_get(const extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return (unsigned)((extent->e_bits & EXTENT_BITS_NFREE_MASK) >>
+	    EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void *
+extent_base_get(const extent_t *extent) {
+	assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
+	    !extent_slab_get(extent));
+	return PAGE_ADDR2BASE(extent->e_addr);
+}
+
+static inline void *
+extent_addr_get(const extent_t *extent) {
+	assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
+	    !extent_slab_get(extent));
+	return extent->e_addr;
+}
+
+static inline size_t
+extent_size_get(const extent_t *extent) {
+	return (extent->e_size_esn & EXTENT_SIZE_MASK);
+}
+
+static inline size_t
+extent_esn_get(const extent_t *extent) {
+	return (extent->e_size_esn & EXTENT_ESN_MASK);
+}
+
+static inline size_t
+extent_bsize_get(const extent_t *extent) {
+	return extent->e_bsize;
+}
+
+static inline void *
+extent_before_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) - PAGE);
+}
+
+static inline void *
+extent_last_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) +
+	    extent_size_get(extent) - PAGE);
+}
+
+static inline void *
+extent_past_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) +
+	    extent_size_get(extent));
+}
+
+static inline arena_slab_data_t *
+extent_slab_data_get(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return &extent->e_slab_data;
+}
+
+static inline const arena_slab_data_t *
+extent_slab_data_get_const(const extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return &extent->e_slab_data;
+}
+
+static inline prof_tctx_t *
+extent_prof_tctx_get(const extent_t *extent) {
+	return (prof_tctx_t *)atomic_load_p(&extent->e_prof_tctx,
+	    ATOMIC_ACQUIRE);
+}
+
+static inline nstime_t
+extent_prof_alloc_time_get(const extent_t *extent) {
+	return extent->e_alloc_time;
+}
+
+static inline void
+extent_arena_set(extent_t *extent, arena_t *arena) {
+	unsigned arena_ind = (arena != NULL) ? arena_ind_get(arena) : ((1U <<
+	    MALLOCX_ARENA_BITS) - 1);
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_ARENA_MASK) |
+	    ((uint64_t)arena_ind << EXTENT_BITS_ARENA_SHIFT);
+}
+
+static inline void
+extent_binshard_set(extent_t *extent, unsigned binshard) {
+	/* The assertion assumes szind is set already. */
+	assert(binshard < bin_infos[extent_szind_get(extent)].n_shards);
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_BINSHARD_MASK) |
+	    ((uint64_t)binshard << EXTENT_BITS_BINSHARD_SHIFT);
+}
+
+static inline void
+extent_addr_set(extent_t *extent, void *addr) {
+	extent->e_addr = addr;
+}
+
+static inline void
+extent_addr_randomize(tsdn_t *tsdn, extent_t *extent, size_t alignment) {
+	assert(extent_base_get(extent) == extent_addr_get(extent));
+
+	if (alignment < PAGE) {
+		unsigned lg_range = LG_PAGE -
+		    lg_floor(CACHELINE_CEILING(alignment));
+		size_t r;
+		if (!tsdn_null(tsdn)) {
+			tsd_t *tsd = tsdn_tsd(tsdn);
+			r = (size_t)prng_lg_range_u64(
+			    tsd_offset_statep_get(tsd), lg_range);
+		} else {
+			r = prng_lg_range_zu(
+			    &extent_arena_get(extent)->offset_state,
+			    lg_range, true);
+		}
+		uintptr_t random_offset = ((uintptr_t)r) << (LG_PAGE -
+		    lg_range);
+		extent->e_addr = (void *)((uintptr_t)extent->e_addr +
+		    random_offset);
+		assert(ALIGNMENT_ADDR2BASE(extent->e_addr, alignment) ==
+		    extent->e_addr);
+	}
+}
+
+static inline void
+extent_size_set(extent_t *extent, size_t size) {
+	assert((size & ~EXTENT_SIZE_MASK) == 0);
+	extent->e_size_esn = size | (extent->e_size_esn & ~EXTENT_SIZE_MASK);
+}
+
+static inline void
+extent_esn_set(extent_t *extent, size_t esn) {
+	extent->e_size_esn = (extent->e_size_esn & ~EXTENT_ESN_MASK) | (esn &
+	    EXTENT_ESN_MASK);
+}
+
+static inline void
+extent_bsize_set(extent_t *extent, size_t bsize) {
+	extent->e_bsize = bsize;
+}
+
+static inline void
+extent_szind_set(extent_t *extent, szind_t szind) {
+	assert(szind <= SC_NSIZES); /* SC_NSIZES means "invalid". */
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SZIND_MASK) |
+	    ((uint64_t)szind << EXTENT_BITS_SZIND_SHIFT);
+}
+
+static inline void
+extent_nfree_set(extent_t *extent, unsigned nfree) {
+	assert(extent_slab_get(extent));
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_NFREE_MASK) |
+	    ((uint64_t)nfree << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_binshard_set(extent_t *extent, unsigned nfree, unsigned binshard) {
+	/* The assertion assumes szind is set already. */
+	assert(binshard < bin_infos[extent_szind_get(extent)].n_shards);
+	extent->e_bits = (extent->e_bits &
+	    (~EXTENT_BITS_NFREE_MASK & ~EXTENT_BITS_BINSHARD_MASK)) |
+	    ((uint64_t)binshard << EXTENT_BITS_BINSHARD_SHIFT) |
+	    ((uint64_t)nfree << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_inc(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	extent->e_bits += ((uint64_t)1U << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_dec(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	extent->e_bits -= ((uint64_t)1U << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_sub(extent_t *extent, uint64_t n) {
+	assert(extent_slab_get(extent));
+	extent->e_bits -= (n << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_sn_set(extent_t *extent, size_t sn) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SN_MASK) |
+	    ((uint64_t)sn << EXTENT_BITS_SN_SHIFT);
+}
+
+static inline void
+extent_state_set(extent_t *extent, extent_state_t state) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_STATE_MASK) |
+	    ((uint64_t)state << EXTENT_BITS_STATE_SHIFT);
+}
+
+static inline void
+extent_zeroed_set(extent_t *extent, bool zeroed) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_ZEROED_MASK) |
+	    ((uint64_t)zeroed << EXTENT_BITS_ZEROED_SHIFT);
+}
+
+static inline void
+extent_committed_set(extent_t *extent, bool committed) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_COMMITTED_MASK) |
+	    ((uint64_t)committed << EXTENT_BITS_COMMITTED_SHIFT);
+}
+
+static inline void
+extent_dumpable_set(extent_t *extent, bool dumpable) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_DUMPABLE_MASK) |
+	    ((uint64_t)dumpable << EXTENT_BITS_DUMPABLE_SHIFT);
+}
+
+static inline void
+extent_slab_set(extent_t *extent, bool slab) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SLAB_MASK) |
+	    ((uint64_t)slab << EXTENT_BITS_SLAB_SHIFT);
+}
+
+static inline void
+extent_prof_tctx_set(extent_t *extent, prof_tctx_t *tctx) {
+	atomic_store_p(&extent->e_prof_tctx, tctx, ATOMIC_RELEASE);
+}
+
+static inline void
+extent_prof_alloc_time_set(extent_t *extent, nstime_t t) {
+	nstime_copy(&extent->e_alloc_time, &t);
+}
+
+static inline bool
+extent_is_head_get(extent_t *extent) {
+	if (maps_coalesce) {
+		not_reached();
+	}
+
+	return (bool)((extent->e_bits & EXTENT_BITS_IS_HEAD_MASK) >>
+	    EXTENT_BITS_IS_HEAD_SHIFT);
+}
+
+static inline void
+extent_is_head_set(extent_t *extent, bool is_head) {
+	if (maps_coalesce) {
+		not_reached();
+	}
+
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_IS_HEAD_MASK) |
+	    ((uint64_t)is_head << EXTENT_BITS_IS_HEAD_SHIFT);
+}
+
+static inline void
+extent_init(extent_t *extent, arena_t *arena, void *addr, size_t size,
+    bool slab, szind_t szind, size_t sn, extent_state_t state, bool zeroed,
+    bool committed, bool dumpable, extent_head_state_t is_head) {
+	assert(addr == PAGE_ADDR2BASE(addr) || !slab);
+
+	extent_arena_set(extent, arena);
+	extent_addr_set(extent, addr);
+	extent_size_set(extent, size);
+	extent_slab_set(extent, slab);
+	extent_szind_set(extent, szind);
+	extent_sn_set(extent, sn);
+	extent_state_set(extent, state);
+	extent_zeroed_set(extent, zeroed);
+	extent_committed_set(extent, committed);
+	extent_dumpable_set(extent, dumpable);
+	ql_elm_new(extent, ql_link);
+	if (!maps_coalesce) {
+		extent_is_head_set(extent, (is_head == EXTENT_IS_HEAD) ? true :
+		    false);
+	}
+	if (config_prof) {
+		extent_prof_tctx_set(extent, NULL);
+	}
+}
+
+static inline void
+extent_binit(extent_t *extent, void *addr, size_t bsize, size_t sn) {
+	extent_arena_set(extent, NULL);
+	extent_addr_set(extent, addr);
+	extent_bsize_set(extent, bsize);
+	extent_slab_set(extent, false);
+	extent_szind_set(extent, SC_NSIZES);
+	extent_sn_set(extent, sn);
+	extent_state_set(extent, extent_state_active);
+	extent_zeroed_set(extent, true);
+	extent_committed_set(extent, true);
+	extent_dumpable_set(extent, true);
+}
+
+static inline void
+extent_list_init(extent_list_t *list) {
+	ql_new(list);
+}
+
+static inline extent_t *
+extent_list_first(const extent_list_t *list) {
+	return ql_first(list);
+}
+
+static inline extent_t *
+extent_list_last(const extent_list_t *list) {
+	return ql_last(list, ql_link);
+}
+
+static inline void
+extent_list_append(extent_list_t *list, extent_t *extent) {
+	ql_tail_insert(list, extent, ql_link);
+}
+
+static inline void
+extent_list_prepend(extent_list_t *list, extent_t *extent) {
+	ql_head_insert(list, extent, ql_link);
+}
+
+static inline void
+extent_list_replace(extent_list_t *list, extent_t *to_remove,
+    extent_t *to_insert) {
+	ql_after_insert(to_remove, to_insert, ql_link);
+	ql_remove(list, to_remove, ql_link);
+}
+
+static inline void
+extent_list_remove(extent_list_t *list, extent_t *extent) {
+	ql_remove(list, extent, ql_link);
+}
+
+static inline int
+extent_sn_comp(const extent_t *a, const extent_t *b) {
+	size_t a_sn = extent_sn_get(a);
+	size_t b_sn = extent_sn_get(b);
+
+	return (a_sn > b_sn) - (a_sn < b_sn);
+}
+
+static inline int
+extent_esn_comp(const extent_t *a, const extent_t *b) {
+	size_t a_esn = extent_esn_get(a);
+	size_t b_esn = extent_esn_get(b);
+
+	return (a_esn > b_esn) - (a_esn < b_esn);
+}
+
+static inline int
+extent_ad_comp(const extent_t *a, const extent_t *b) {
+	uintptr_t a_addr = (uintptr_t)extent_addr_get(a);
+	uintptr_t b_addr = (uintptr_t)extent_addr_get(b);
+
+	return (a_addr > b_addr) - (a_addr < b_addr);
+}
+
+static inline int
+extent_ead_comp(const extent_t *a, const extent_t *b) {
+	uintptr_t a_eaddr = (uintptr_t)a;
+	uintptr_t b_eaddr = (uintptr_t)b;
+
+	return (a_eaddr > b_eaddr) - (a_eaddr < b_eaddr);
+}
+
+static inline int
+extent_snad_comp(const extent_t *a, const extent_t *b) {
+	int ret;
+
+	ret = extent_sn_comp(a, b);
+	if (ret != 0) {
+		return ret;
+	}
+
+	ret = extent_ad_comp(a, b);
+	return ret;
+}
+
+static inline int
+extent_esnead_comp(const extent_t *a, const extent_t *b) {
+	int ret;
+
+	ret = extent_esn_comp(a, b);
+	if (ret != 0) {
+		return ret;
+	}
+
+	ret = extent_ead_comp(a, b);
+	return ret;
+}
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_INLINES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_mmap.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_mmap.h
new file mode 100644
index 0000000..55f17ee
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_mmap.h
@@ -0,0 +1,10 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H
+
+extern bool opt_retain;
+
+void *extent_alloc_mmap(void *new_addr, size_t size, size_t alignment,
+    bool *zero, bool *commit);
+bool extent_dalloc_mmap(void *addr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_structs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_structs.h
new file mode 100644
index 0000000..767cd89
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_structs.h
@@ -0,0 +1,256 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_STRUCTS_H
+#define JEMALLOC_INTERNAL_EXTENT_STRUCTS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/sc.h"
+
+typedef enum {
+	extent_state_active   = 0,
+	extent_state_dirty    = 1,
+	extent_state_muzzy    = 2,
+	extent_state_retained = 3
+} extent_state_t;
+
+/* Extent (span of pages).  Use accessor functions for e_* fields. */
+struct extent_s {
+	/*
+	 * Bitfield containing several fields:
+	 *
+	 * a: arena_ind
+	 * b: slab
+	 * c: committed
+	 * d: dumpable
+	 * z: zeroed
+	 * t: state
+	 * i: szind
+	 * f: nfree
+	 * s: bin_shard
+	 * n: sn
+	 *
+	 * nnnnnnnn ... nnnnnnss ssssffff ffffffii iiiiiitt zdcbaaaa aaaaaaaa
+	 *
+	 * arena_ind: Arena from which this extent came, or all 1 bits if
+	 *            unassociated.
+	 *
+	 * slab: The slab flag indicates whether the extent is used for a slab
+	 *       of small regions.  This helps differentiate small size classes,
+	 *       and it indicates whether interior pointers can be looked up via
+	 *       iealloc().
+	 *
+	 * committed: The committed flag indicates whether physical memory is
+	 *            committed to the extent, whether explicitly or implicitly
+	 *            as on a system that overcommits and satisfies physical
+	 *            memory needs on demand via soft page faults.
+	 *
+	 * dumpable: The dumpable flag indicates whether or not we've set the
+	 *           memory in question to be dumpable.  Note that this
+	 *           interacts somewhat subtly with user-specified extent hooks,
+	 *           since we don't know if *they* are fiddling with
+	 *           dumpability (in which case, we don't want to undo whatever
+	 *           they're doing).  To deal with this scenario, we:
+	 *             - Make dumpable false only for memory allocated with the
+	 *               default hooks.
+	 *             - Only allow memory to go from non-dumpable to dumpable,
+	 *               and only once.
+	 *             - Never make the OS call to allow dumping when the
+	 *               dumpable bit is already set.
+	 *           These three constraints mean that we will never
+	 *           accidentally dump user memory that the user meant to set
+	 *           nondumpable with their extent hooks.
+	 *
+	 *
+	 * zeroed: The zeroed flag is used by extent recycling code to track
+	 *         whether memory is zero-filled.
+	 *
+	 * state: The state flag is an extent_state_t.
+	 *
+	 * szind: The szind flag indicates usable size class index for
+	 *        allocations residing in this extent, regardless of whether the
+	 *        extent is a slab.  Extent size and usable size often differ
+	 *        even for non-slabs, either due to sz_large_pad or promotion of
+	 *        sampled small regions.
+	 *
+	 * nfree: Number of free regions in slab.
+	 *
+	 * bin_shard: the shard of the bin from which this extent came.
+	 *
+	 * sn: Serial number (potentially non-unique).
+	 *
+	 *     Serial numbers may wrap around if !opt_retain, but as long as
+	 *     comparison functions fall back on address comparison for equal
+	 *     serial numbers, stable (if imperfect) ordering is maintained.
+	 *
+	 *     Serial numbers may not be unique even in the absence of
+	 *     wrap-around, e.g. when splitting an extent and assigning the same
+	 *     serial number to both resulting adjacent extents.
+	 */
+	uint64_t		e_bits;
+#define MASK(CURRENT_FIELD_WIDTH, CURRENT_FIELD_SHIFT) ((((((uint64_t)0x1U) << (CURRENT_FIELD_WIDTH)) - 1)) << (CURRENT_FIELD_SHIFT))
+
+#define EXTENT_BITS_ARENA_WIDTH  MALLOCX_ARENA_BITS
+#define EXTENT_BITS_ARENA_SHIFT  0
+#define EXTENT_BITS_ARENA_MASK  MASK(EXTENT_BITS_ARENA_WIDTH, EXTENT_BITS_ARENA_SHIFT)
+
+#define EXTENT_BITS_SLAB_WIDTH  1
+#define EXTENT_BITS_SLAB_SHIFT  (EXTENT_BITS_ARENA_WIDTH + EXTENT_BITS_ARENA_SHIFT)
+#define EXTENT_BITS_SLAB_MASK  MASK(EXTENT_BITS_SLAB_WIDTH, EXTENT_BITS_SLAB_SHIFT)
+
+#define EXTENT_BITS_COMMITTED_WIDTH  1
+#define EXTENT_BITS_COMMITTED_SHIFT  (EXTENT_BITS_SLAB_WIDTH + EXTENT_BITS_SLAB_SHIFT)
+#define EXTENT_BITS_COMMITTED_MASK  MASK(EXTENT_BITS_COMMITTED_WIDTH, EXTENT_BITS_COMMITTED_SHIFT)
+
+#define EXTENT_BITS_DUMPABLE_WIDTH  1
+#define EXTENT_BITS_DUMPABLE_SHIFT  (EXTENT_BITS_COMMITTED_WIDTH + EXTENT_BITS_COMMITTED_SHIFT)
+#define EXTENT_BITS_DUMPABLE_MASK  MASK(EXTENT_BITS_DUMPABLE_WIDTH, EXTENT_BITS_DUMPABLE_SHIFT)
+
+#define EXTENT_BITS_ZEROED_WIDTH  1
+#define EXTENT_BITS_ZEROED_SHIFT  (EXTENT_BITS_DUMPABLE_WIDTH + EXTENT_BITS_DUMPABLE_SHIFT)
+#define EXTENT_BITS_ZEROED_MASK  MASK(EXTENT_BITS_ZEROED_WIDTH, EXTENT_BITS_ZEROED_SHIFT)
+
+#define EXTENT_BITS_STATE_WIDTH  2
+#define EXTENT_BITS_STATE_SHIFT  (EXTENT_BITS_ZEROED_WIDTH + EXTENT_BITS_ZEROED_SHIFT)
+#define EXTENT_BITS_STATE_MASK  MASK(EXTENT_BITS_STATE_WIDTH, EXTENT_BITS_STATE_SHIFT)
+
+#define EXTENT_BITS_SZIND_WIDTH  LG_CEIL(SC_NSIZES)
+#define EXTENT_BITS_SZIND_SHIFT  (EXTENT_BITS_STATE_WIDTH + EXTENT_BITS_STATE_SHIFT)
+#define EXTENT_BITS_SZIND_MASK  MASK(EXTENT_BITS_SZIND_WIDTH, EXTENT_BITS_SZIND_SHIFT)
+
+#define EXTENT_BITS_NFREE_WIDTH  (LG_SLAB_MAXREGS + 1)
+#define EXTENT_BITS_NFREE_SHIFT  (EXTENT_BITS_SZIND_WIDTH + EXTENT_BITS_SZIND_SHIFT)
+#define EXTENT_BITS_NFREE_MASK  MASK(EXTENT_BITS_NFREE_WIDTH, EXTENT_BITS_NFREE_SHIFT)
+
+#define EXTENT_BITS_BINSHARD_WIDTH  6
+#define EXTENT_BITS_BINSHARD_SHIFT  (EXTENT_BITS_NFREE_WIDTH + EXTENT_BITS_NFREE_SHIFT)
+#define EXTENT_BITS_BINSHARD_MASK  MASK(EXTENT_BITS_BINSHARD_WIDTH, EXTENT_BITS_BINSHARD_SHIFT)
+
+#define EXTENT_BITS_IS_HEAD_WIDTH 1
+#define EXTENT_BITS_IS_HEAD_SHIFT  (EXTENT_BITS_BINSHARD_WIDTH + EXTENT_BITS_BINSHARD_SHIFT)
+#define EXTENT_BITS_IS_HEAD_MASK  MASK(EXTENT_BITS_IS_HEAD_WIDTH, EXTENT_BITS_IS_HEAD_SHIFT)
+
+#define EXTENT_BITS_SN_SHIFT   (EXTENT_BITS_IS_HEAD_WIDTH + EXTENT_BITS_IS_HEAD_SHIFT)
+#define EXTENT_BITS_SN_MASK  (UINT64_MAX << EXTENT_BITS_SN_SHIFT)
+
+	/* Pointer to the extent that this structure is responsible for. */
+	void			*e_addr;
+
+	union {
+		/*
+		 * Extent size and serial number associated with the extent
+		 * structure (different than the serial number for the extent at
+		 * e_addr).
+		 *
+		 * ssssssss [...] ssssssss ssssnnnn nnnnnnnn
+		 */
+		size_t			e_size_esn;
+	#define EXTENT_SIZE_MASK	((size_t)~(PAGE-1))
+	#define EXTENT_ESN_MASK		((size_t)PAGE-1)
+		/* Base extent size, which may not be a multiple of PAGE. */
+		size_t			e_bsize;
+	};
+
+	/*
+	 * List linkage, used by a variety of lists:
+	 * - bin_t's slabs_full
+	 * - extents_t's LRU
+	 * - stashed dirty extents
+	 * - arena's large allocations
+	 */
+	ql_elm(extent_t)	ql_link;
+
+	/*
+	 * Linkage for per size class sn/address-ordered heaps, and
+	 * for extent_avail
+	 */
+	phn(extent_t)		ph_link;
+
+	union {
+		/* Small region slab metadata. */
+		arena_slab_data_t	e_slab_data;
+
+		/* Profiling data, used for large objects. */
+		struct {
+			/* Time when this was allocated. */
+			nstime_t		e_alloc_time;
+			/* Points to a prof_tctx_t. */
+			atomic_p_t		e_prof_tctx;
+		};
+	};
+};
+typedef ql_head(extent_t) extent_list_t;
+typedef ph(extent_t) extent_tree_t;
+typedef ph(extent_t) extent_heap_t;
+
+/* Quantized collection of extents, with built-in LRU queue. */
+struct extents_s {
+	malloc_mutex_t		mtx;
+
+	/*
+	 * Quantized per size class heaps of extents.
+	 *
+	 * Synchronization: mtx.
+	 */
+	extent_heap_t		heaps[SC_NPSIZES + 1];
+	atomic_zu_t		nextents[SC_NPSIZES + 1];
+	atomic_zu_t		nbytes[SC_NPSIZES + 1];
+
+	/*
+	 * Bitmap for which set bits correspond to non-empty heaps.
+	 *
+	 * Synchronization: mtx.
+	 */
+	bitmap_t		bitmap[BITMAP_GROUPS(SC_NPSIZES + 1)];
+
+	/*
+	 * LRU of all extents in heaps.
+	 *
+	 * Synchronization: mtx.
+	 */
+	extent_list_t		lru;
+
+	/*
+	 * Page sum for all extents in heaps.
+	 *
+	 * The synchronization here is a little tricky.  Modifications to npages
+	 * must hold mtx, but reads need not (though, a reader who sees npages
+	 * without holding the mutex can't assume anything about the rest of the
+	 * state of the extents_t).
+	 */
+	atomic_zu_t		npages;
+
+	/* All stored extents must be in the same state. */
+	extent_state_t		state;
+
+	/*
+	 * If true, delay coalescing until eviction; otherwise coalesce during
+	 * deallocation.
+	 */
+	bool			delay_coalesce;
+};
+
+/*
+ * The following two structs are for experimental purposes. See
+ * experimental_utilization_query_ctl and
+ * experimental_utilization_batch_query_ctl in src/ctl.c.
+ */
+
+struct extent_util_stats_s {
+	size_t nfree;
+	size_t nregs;
+	size_t size;
+};
+
+struct extent_util_stats_verbose_s {
+	void *slabcur_addr;
+	size_t nfree;
+	size_t nregs;
+	size_t size;
+	size_t bin_nfree;
+	size_t bin_nregs;
+};
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_STRUCTS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_types.h
new file mode 100644
index 0000000..96925cf
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/extent_types.h
@@ -0,0 +1,23 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_TYPES_H
+#define JEMALLOC_INTERNAL_EXTENT_TYPES_H
+
+typedef struct extent_s extent_t;
+typedef struct extents_s extents_t;
+
+typedef struct extent_util_stats_s extent_util_stats_t;
+typedef struct extent_util_stats_verbose_s extent_util_stats_verbose_t;
+
+#define EXTENT_HOOKS_INITIALIZER	NULL
+
+/*
+ * When reuse (and split) an active extent, (1U << opt_lg_extent_max_active_fit)
+ * is the max ratio between the size of the active extent and the new extent.
+ */
+#define LG_EXTENT_MAX_ACTIVE_FIT_DEFAULT 6
+
+typedef enum {
+	EXTENT_NOT_HEAD,
+	EXTENT_IS_HEAD   /* Only relevant for Windows && opt.retain. */
+} extent_head_state_t;
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/hash.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/hash.h
new file mode 100644
index 0000000..0270034
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/hash.h
@@ -0,0 +1,319 @@
+#ifndef JEMALLOC_INTERNAL_HASH_H
+#define JEMALLOC_INTERNAL_HASH_H
+
+#include "jemalloc/internal/assert.h"
+
+/*
+ * The following hash function is based on MurmurHash3, placed into the public
+ * domain by Austin Appleby.  See https://github.com/aappleby/smhasher for
+ * details.
+ */
+
+/******************************************************************************/
+/* Internal implementation. */
+static inline uint32_t
+hash_rotl_32(uint32_t x, int8_t r) {
+	return ((x << r) | (x >> (32 - r)));
+}
+
+static inline uint64_t
+hash_rotl_64(uint64_t x, int8_t r) {
+	return ((x << r) | (x >> (64 - r)));
+}
+
+static inline uint32_t
+hash_get_block_32(const uint32_t *p, int i) {
+	/* Handle unaligned read. */
+	if (unlikely((uintptr_t)p & (sizeof(uint32_t)-1)) != 0) {
+		uint32_t ret;
+
+		memcpy(&ret, (uint8_t *)(p + i), sizeof(uint32_t));
+		return ret;
+	}
+
+	return p[i];
+}
+
+static inline uint64_t
+hash_get_block_64(const uint64_t *p, int i) {
+	/* Handle unaligned read. */
+	if (unlikely((uintptr_t)p & (sizeof(uint64_t)-1)) != 0) {
+		uint64_t ret;
+
+		memcpy(&ret, (uint8_t *)(p + i), sizeof(uint64_t));
+		return ret;
+	}
+
+	return p[i];
+}
+
+static inline uint32_t
+hash_fmix_32(uint32_t h) {
+	h ^= h >> 16;
+	h *= 0x85ebca6b;
+	h ^= h >> 13;
+	h *= 0xc2b2ae35;
+	h ^= h >> 16;
+
+	return h;
+}
+
+static inline uint64_t
+hash_fmix_64(uint64_t k) {
+	k ^= k >> 33;
+	k *= KQU(0xff51afd7ed558ccd);
+	k ^= k >> 33;
+	k *= KQU(0xc4ceb9fe1a85ec53);
+	k ^= k >> 33;
+
+	return k;
+}
+
+static inline uint32_t
+hash_x86_32(const void *key, int len, uint32_t seed) {
+	const uint8_t *data = (const uint8_t *) key;
+	const int nblocks = len / 4;
+
+	uint32_t h1 = seed;
+
+	const uint32_t c1 = 0xcc9e2d51;
+	const uint32_t c2 = 0x1b873593;
+
+	/* body */
+	{
+		const uint32_t *blocks = (const uint32_t *) (data + nblocks*4);
+		int i;
+
+		for (i = -nblocks; i; i++) {
+			uint32_t k1 = hash_get_block_32(blocks, i);
+
+			k1 *= c1;
+			k1 = hash_rotl_32(k1, 15);
+			k1 *= c2;
+
+			h1 ^= k1;
+			h1 = hash_rotl_32(h1, 13);
+			h1 = h1*5 + 0xe6546b64;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t *) (data + nblocks*4);
+
+		uint32_t k1 = 0;
+
+		switch (len & 3) {
+		case 3: k1 ^= tail[2] << 16; JEMALLOC_FALLTHROUGH
+		case 2: k1 ^= tail[1] << 8; JEMALLOC_FALLTHROUGH
+		case 1: k1 ^= tail[0]; k1 *= c1; k1 = hash_rotl_32(k1, 15);
+			k1 *= c2; h1 ^= k1;
+		}
+	}
+
+	/* finalization */
+	h1 ^= len;
+
+	h1 = hash_fmix_32(h1);
+
+	return h1;
+}
+
+static inline void
+hash_x86_128(const void *key, const int len, uint32_t seed,
+    uint64_t r_out[2]) {
+	const uint8_t * data = (const uint8_t *) key;
+	const int nblocks = len / 16;
+
+	uint32_t h1 = seed;
+	uint32_t h2 = seed;
+	uint32_t h3 = seed;
+	uint32_t h4 = seed;
+
+	const uint32_t c1 = 0x239b961b;
+	const uint32_t c2 = 0xab0e9789;
+	const uint32_t c3 = 0x38b34ae5;
+	const uint32_t c4 = 0xa1e38b93;
+
+	/* body */
+	{
+		const uint32_t *blocks = (const uint32_t *) (data + nblocks*16);
+		int i;
+
+		for (i = -nblocks; i; i++) {
+			uint32_t k1 = hash_get_block_32(blocks, i*4 + 0);
+			uint32_t k2 = hash_get_block_32(blocks, i*4 + 1);
+			uint32_t k3 = hash_get_block_32(blocks, i*4 + 2);
+			uint32_t k4 = hash_get_block_32(blocks, i*4 + 3);
+
+			k1 *= c1; k1 = hash_rotl_32(k1, 15); k1 *= c2; h1 ^= k1;
+
+			h1 = hash_rotl_32(h1, 19); h1 += h2;
+			h1 = h1*5 + 0x561ccd1b;
+
+			k2 *= c2; k2 = hash_rotl_32(k2, 16); k2 *= c3; h2 ^= k2;
+
+			h2 = hash_rotl_32(h2, 17); h2 += h3;
+			h2 = h2*5 + 0x0bcaa747;
+
+			k3 *= c3; k3 = hash_rotl_32(k3, 17); k3 *= c4; h3 ^= k3;
+
+			h3 = hash_rotl_32(h3, 15); h3 += h4;
+			h3 = h3*5 + 0x96cd1c35;
+
+			k4 *= c4; k4 = hash_rotl_32(k4, 18); k4 *= c1; h4 ^= k4;
+
+			h4 = hash_rotl_32(h4, 13); h4 += h1;
+			h4 = h4*5 + 0x32ac3b17;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t *) (data + nblocks*16);
+		uint32_t k1 = 0;
+		uint32_t k2 = 0;
+		uint32_t k3 = 0;
+		uint32_t k4 = 0;
+
+		switch (len & 15) {
+		case 15: k4 ^= tail[14] << 16; JEMALLOC_FALLTHROUGH
+		case 14: k4 ^= tail[13] << 8; JEMALLOC_FALLTHROUGH
+		case 13: k4 ^= tail[12] << 0;
+			k4 *= c4; k4 = hash_rotl_32(k4, 18); k4 *= c1; h4 ^= k4;
+      JEMALLOC_FALLTHROUGH
+		case 12: k3 ^= tail[11] << 24; JEMALLOC_FALLTHROUGH
+		case 11: k3 ^= tail[10] << 16; JEMALLOC_FALLTHROUGH
+		case 10: k3 ^= tail[ 9] << 8; JEMALLOC_FALLTHROUGH
+		case  9: k3 ^= tail[ 8] << 0;
+		     k3 *= c3; k3 = hash_rotl_32(k3, 17); k3 *= c4; h3 ^= k3;
+         JEMALLOC_FALLTHROUGH
+		case  8: k2 ^= tail[ 7] << 24; JEMALLOC_FALLTHROUGH
+		case  7: k2 ^= tail[ 6] << 16; JEMALLOC_FALLTHROUGH
+		case  6: k2 ^= tail[ 5] << 8; JEMALLOC_FALLTHROUGH
+		case  5: k2 ^= tail[ 4] << 0;
+			k2 *= c2; k2 = hash_rotl_32(k2, 16); k2 *= c3; h2 ^= k2;
+      JEMALLOC_FALLTHROUGH
+		case  4: k1 ^= tail[ 3] << 24; JEMALLOC_FALLTHROUGH
+		case  3: k1 ^= tail[ 2] << 16; JEMALLOC_FALLTHROUGH
+		case  2: k1 ^= tail[ 1] << 8; JEMALLOC_FALLTHROUGH
+		case  1: k1 ^= tail[ 0] << 0;
+			k1 *= c1; k1 = hash_rotl_32(k1, 15); k1 *= c2; h1 ^= k1;
+      JEMALLOC_FALLTHROUGH
+		}
+	}
+
+	/* finalization */
+	h1 ^= len; h2 ^= len; h3 ^= len; h4 ^= len;
+
+	h1 += h2; h1 += h3; h1 += h4;
+	h2 += h1; h3 += h1; h4 += h1;
+
+	h1 = hash_fmix_32(h1);
+	h2 = hash_fmix_32(h2);
+	h3 = hash_fmix_32(h3);
+	h4 = hash_fmix_32(h4);
+
+	h1 += h2; h1 += h3; h1 += h4;
+	h2 += h1; h3 += h1; h4 += h1;
+
+	r_out[0] = (((uint64_t) h2) << 32) | h1;
+	r_out[1] = (((uint64_t) h4) << 32) | h3;
+}
+
+static inline void
+hash_x64_128(const void *key, const int len, const uint32_t seed,
+    uint64_t r_out[2]) {
+	const uint8_t *data = (const uint8_t *) key;
+	const int nblocks = len / 16;
+
+	uint64_t h1 = seed;
+	uint64_t h2 = seed;
+
+	const uint64_t c1 = KQU(0x87c37b91114253d5);
+	const uint64_t c2 = KQU(0x4cf5ad432745937f);
+
+	/* body */
+	{
+		const uint64_t *blocks = (const uint64_t *) (data);
+		int i;
+
+		for (i = 0; i < nblocks; i++) {
+			uint64_t k1 = hash_get_block_64(blocks, i*2 + 0);
+			uint64_t k2 = hash_get_block_64(blocks, i*2 + 1);
+
+			k1 *= c1; k1 = hash_rotl_64(k1, 31); k1 *= c2; h1 ^= k1;
+
+			h1 = hash_rotl_64(h1, 27); h1 += h2;
+			h1 = h1*5 + 0x52dce729;
+
+			k2 *= c2; k2 = hash_rotl_64(k2, 33); k2 *= c1; h2 ^= k2;
+
+			h2 = hash_rotl_64(h2, 31); h2 += h1;
+			h2 = h2*5 + 0x38495ab5;
+		}
+	}
+
+	/* tail */
+	{
+		const uint8_t *tail = (const uint8_t*)(data + nblocks*16);
+		uint64_t k1 = 0;
+		uint64_t k2 = 0;
+
+		switch (len & 15) {
+		case 15: k2 ^= ((uint64_t)(tail[14])) << 48; JEMALLOC_FALLTHROUGH
+		case 14: k2 ^= ((uint64_t)(tail[13])) << 40; JEMALLOC_FALLTHROUGH
+		case 13: k2 ^= ((uint64_t)(tail[12])) << 32; JEMALLOC_FALLTHROUGH
+		case 12: k2 ^= ((uint64_t)(tail[11])) << 24; JEMALLOC_FALLTHROUGH
+		case 11: k2 ^= ((uint64_t)(tail[10])) << 16; JEMALLOC_FALLTHROUGH
+		case 10: k2 ^= ((uint64_t)(tail[ 9])) << 8;  JEMALLOC_FALLTHROUGH
+		case  9: k2 ^= ((uint64_t)(tail[ 8])) << 0;
+			k2 *= c2; k2 = hash_rotl_64(k2, 33); k2 *= c1; h2 ^= k2;
+			JEMALLOC_FALLTHROUGH
+		case  8: k1 ^= ((uint64_t)(tail[ 7])) << 56; JEMALLOC_FALLTHROUGH
+		case  7: k1 ^= ((uint64_t)(tail[ 6])) << 48; JEMALLOC_FALLTHROUGH
+		case  6: k1 ^= ((uint64_t)(tail[ 5])) << 40; JEMALLOC_FALLTHROUGH
+		case  5: k1 ^= ((uint64_t)(tail[ 4])) << 32; JEMALLOC_FALLTHROUGH
+		case  4: k1 ^= ((uint64_t)(tail[ 3])) << 24; JEMALLOC_FALLTHROUGH
+		case  3: k1 ^= ((uint64_t)(tail[ 2])) << 16; JEMALLOC_FALLTHROUGH
+		case  2: k1 ^= ((uint64_t)(tail[ 1])) << 8;  JEMALLOC_FALLTHROUGH
+		case  1: k1 ^= ((uint64_t)(tail[ 0])) << 0;
+			k1 *= c1; k1 = hash_rotl_64(k1, 31); k1 *= c2; h1 ^= k1;
+		}
+	}
+
+	/* finalization */
+	h1 ^= len; h2 ^= len;
+
+	h1 += h2;
+	h2 += h1;
+
+	h1 = hash_fmix_64(h1);
+	h2 = hash_fmix_64(h2);
+
+	h1 += h2;
+	h2 += h1;
+
+	r_out[0] = h1;
+	r_out[1] = h2;
+}
+
+/******************************************************************************/
+/* API. */
+static inline void
+hash(const void *key, size_t len, const uint32_t seed, size_t r_hash[2]) {
+	assert(len <= INT_MAX); /* Unfortunate implementation limitation. */
+
+#if (LG_SIZEOF_PTR == 3 && !defined(JEMALLOC_BIG_ENDIAN))
+	hash_x64_128(key, (int)len, seed, (uint64_t *)r_hash);
+#else
+	{
+		uint64_t hashes[2];
+		hash_x86_128(key, (int)len, seed, hashes);
+		r_hash[0] = (size_t)hashes[0];
+		r_hash[1] = (size_t)hashes[1];
+	}
+#endif
+}
+
+#endif /* JEMALLOC_INTERNAL_HASH_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/hook.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/hook.h
new file mode 100644
index 0000000..ee246b1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/hook.h
@@ -0,0 +1,163 @@
+#ifndef JEMALLOC_INTERNAL_HOOK_H
+#define JEMALLOC_INTERNAL_HOOK_H
+
+#include "jemalloc/internal/tsd.h"
+
+/*
+ * This API is *extremely* experimental, and may get ripped out, changed in API-
+ * and ABI-incompatible ways, be insufficiently or incorrectly documented, etc.
+ *
+ * It allows hooking the stateful parts of the API to see changes as they
+ * happen.
+ *
+ * Allocation hooks are called after the allocation is done, free hooks are
+ * called before the free is done, and expand hooks are called after the
+ * allocation is expanded.
+ *
+ * For realloc and rallocx, if the expansion happens in place, the expansion
+ * hook is called.  If it is moved, then the alloc hook is called on the new
+ * location, and then the free hook is called on the old location (i.e. both
+ * hooks are invoked in between the alloc and the dalloc).
+ *
+ * If we return NULL from OOM, then usize might not be trustworthy.  Calling
+ * realloc(NULL, size) only calls the alloc hook, and calling realloc(ptr, 0)
+ * only calls the free hook.  (Calling realloc(NULL, 0) is treated as malloc(0),
+ * and only calls the alloc hook).
+ *
+ * Reentrancy:
+ *   Reentrancy is guarded against from within the hook implementation.  If you
+ *   call allocator functions from within a hook, the hooks will not be invoked
+ *   again.
+ * Threading:
+ *   The installation of a hook synchronizes with all its uses.  If you can
+ *   prove the installation of a hook happens-before a jemalloc entry point,
+ *   then the hook will get invoked (unless there's a racing removal).
+ *
+ *   Hook insertion appears to be atomic at a per-thread level (i.e. if a thread
+ *   allocates and has the alloc hook invoked, then a subsequent free on the
+ *   same thread will also have the free hook invoked).
+ *
+ *   The *removal* of a hook does *not* block until all threads are done with
+ *   the hook.  Hook authors have to be resilient to this, and need some
+ *   out-of-band mechanism for cleaning up any dynamically allocated memory
+ *   associated with their hook.
+ * Ordering:
+ *   Order of hook execution is unspecified, and may be different than insertion
+ *   order.
+ */
+
+#define HOOK_MAX 4
+
+enum hook_alloc_e {
+	hook_alloc_malloc,
+	hook_alloc_posix_memalign,
+	hook_alloc_aligned_alloc,
+	hook_alloc_calloc,
+	hook_alloc_memalign,
+	hook_alloc_valloc,
+	hook_alloc_mallocx,
+
+	/* The reallocating functions have both alloc and dalloc variants */
+	hook_alloc_realloc,
+	hook_alloc_rallocx,
+};
+/*
+ * We put the enum typedef after the enum, since this file may get included by
+ * jemalloc_cpp.cpp, and C++ disallows enum forward declarations.
+ */
+typedef enum hook_alloc_e hook_alloc_t;
+
+enum hook_dalloc_e {
+	hook_dalloc_free,
+	hook_dalloc_dallocx,
+	hook_dalloc_sdallocx,
+
+	/*
+	 * The dalloc halves of reallocation (not called if in-place expansion
+	 * happens).
+	 */
+	hook_dalloc_realloc,
+	hook_dalloc_rallocx,
+};
+typedef enum hook_dalloc_e hook_dalloc_t;
+
+
+enum hook_expand_e {
+	hook_expand_realloc,
+	hook_expand_rallocx,
+	hook_expand_xallocx,
+};
+typedef enum hook_expand_e hook_expand_t;
+
+typedef void (*hook_alloc)(
+    void *extra, hook_alloc_t type, void *result, uintptr_t result_raw,
+    uintptr_t args_raw[3]);
+
+typedef void (*hook_dalloc)(
+    void *extra, hook_dalloc_t type, void *address, uintptr_t args_raw[3]);
+
+typedef void (*hook_expand)(
+    void *extra, hook_expand_t type, void *address, size_t old_usize,
+    size_t new_usize, uintptr_t result_raw, uintptr_t args_raw[4]);
+
+typedef struct hooks_s hooks_t;
+struct hooks_s {
+	hook_alloc alloc_hook;
+	hook_dalloc dalloc_hook;
+	hook_expand expand_hook;
+	void *extra;
+};
+
+/*
+ * Begin implementation details; everything above this point might one day live
+ * in a public API.  Everything below this point never will.
+ */
+
+/*
+ * The realloc pathways haven't gotten any refactoring love in a while, and it's
+ * fairly difficult to pass information from the entry point to the hooks.  We
+ * put the informaiton the hooks will need into a struct to encapsulate
+ * everything.
+ *
+ * Much of these pathways are force-inlined, so that the compiler can avoid
+ * materializing this struct until we hit an extern arena function.  For fairly
+ * goofy reasons, *many* of the realloc paths hit an extern arena function.
+ * These paths are cold enough that it doesn't matter; eventually, we should
+ * rewrite the realloc code to make the expand-in-place and the
+ * free-then-realloc paths more orthogonal, at which point we don't need to
+ * spread the hook logic all over the place.
+ */
+typedef struct hook_ralloc_args_s hook_ralloc_args_t;
+struct hook_ralloc_args_s {
+	/* I.e. as opposed to rallocx. */
+	bool is_realloc;
+	/*
+	 * The expand hook takes 4 arguments, even if only 3 are actually used;
+	 * we add an extra one in case the user decides to memcpy without
+	 * looking too closely at the hooked function.
+	 */
+	uintptr_t args[4];
+};
+
+/*
+ * Returns an opaque handle to be used when removing the hook.  NULL means that
+ * we couldn't install the hook.
+ */
+bool hook_boot();
+
+void *hook_install(tsdn_t *tsdn, hooks_t *hooks);
+/* Uninstalls the hook with the handle previously returned from hook_install. */
+void hook_remove(tsdn_t *tsdn, void *opaque);
+
+/* Hooks */
+
+void hook_invoke_alloc(hook_alloc_t type, void *result, uintptr_t result_raw,
+    uintptr_t args_raw[3]);
+
+void hook_invoke_dalloc(hook_dalloc_t type, void *address,
+    uintptr_t args_raw[3]);
+
+void hook_invoke_expand(hook_expand_t type, void *address, size_t old_usize,
+    size_t new_usize, uintptr_t result_raw, uintptr_t args_raw[4]);
+
+#endif /* JEMALLOC_INTERNAL_HOOK_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
new file mode 100644
index 0000000..7d6053e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_decls.h
@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_DECLS_H
+#define JEMALLOC_INTERNAL_DECLS_H
+
+#include <math.h>
+#ifdef _WIN32
+#  include <windows.h>
+#  include "msvc_compat/windows_extra.h"
+#  ifdef _WIN64
+#    if LG_VADDR <= 32
+#      error Generate the headers using x64 vcargs
+#    endif
+#  else
+#    if LG_VADDR > 32
+#      undef LG_VADDR
+#      define LG_VADDR 32
+#    endif
+#  endif
+#else
+#  include <sys/param.h>
+#  include <sys/mman.h>
+#  if !defined(__pnacl__) && !defined(__native_client__)
+#    include <sys/syscall.h>
+#    if !defined(SYS_write) && defined(__NR_write)
+#      define SYS_write __NR_write
+#    endif
+#    if defined(SYS_open) && defined(__aarch64__)
+       /* Android headers may define SYS_open to __NR_open even though
+        * __NR_open may not exist on AArch64 (superseded by __NR_openat). */
+#      undef SYS_open
+#    endif
+#    include <sys/uio.h>
+#  endif
+#  include <pthread.h>
+#  ifdef __FreeBSD__
+#  include <pthread_np.h>
+#  endif
+#  include <signal.h>
+#  ifdef JEMALLOC_OS_UNFAIR_LOCK
+#    include <os/lock.h>
+#  endif
+#  ifdef JEMALLOC_GLIBC_MALLOC_HOOK
+#    include <sched.h>
+#  endif
+#  include <errno.h>
+#  include <sys/time.h>
+#  include <time.h>
+#  ifdef JEMALLOC_HAVE_MACH_ABSOLUTE_TIME
+#    include <mach/mach_time.h>
+#  endif
+#endif
+#include <sys/types.h>
+
+#include <limits.h>
+#ifndef SIZE_T_MAX
+#  define SIZE_T_MAX	SIZE_MAX
+#endif
+#ifndef SSIZE_MAX
+#  define SSIZE_MAX	((ssize_t)(SIZE_T_MAX >> 1))
+#endif
+#include <stdarg.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <stddef.h>
+#ifndef offsetof
+#  define offsetof(type, member)	((size_t)&(((type *)NULL)->member))
+#endif
+#include <string.h>
+#include <strings.h>
+#include <ctype.h>
+#ifdef _MSC_VER
+#  include <io.h>
+typedef intptr_t ssize_t;
+#  define PATH_MAX 1024
+#  define STDERR_FILENO 2
+#  define __func__ __FUNCTION__
+#  ifdef JEMALLOC_HAS_RESTRICT
+#    define restrict __restrict
+#  endif
+/* Disable warnings about deprecated system functions. */
+#  pragma warning(disable: 4996)
+#if _MSC_VER < 1800
+static int
+isblank(int c) {
+	return (c == '\t' || c == ' ');
+}
+#endif
+#else
+#  include <unistd.h>
+#endif
+#include <fcntl.h>
+
+#endif /* JEMALLOC_INTERNAL_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
new file mode 100644
index 0000000..c442a21
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_defs.h.in
@@ -0,0 +1,366 @@
+#ifndef JEMALLOC_INTERNAL_DEFS_H_
+#define JEMALLOC_INTERNAL_DEFS_H_
+/*
+ * If JEMALLOC_PREFIX is defined via --with-jemalloc-prefix, it will cause all
+ * public APIs to be prefixed.  This makes it possible, with some care, to use
+ * multiple allocators simultaneously.
+ */
+#undef JEMALLOC_PREFIX
+#undef JEMALLOC_CPREFIX
+
+/*
+ * Define overrides for non-standard allocator-related functions if they are
+ * present on the system.
+ */
+#undef JEMALLOC_OVERRIDE___LIBC_CALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_FREE
+#undef JEMALLOC_OVERRIDE___LIBC_MALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_MEMALIGN
+#undef JEMALLOC_OVERRIDE___LIBC_REALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_VALLOC
+#undef JEMALLOC_OVERRIDE___POSIX_MEMALIGN
+
+/*
+ * JEMALLOC_PRIVATE_NAMESPACE is used as a prefix for all library-private APIs.
+ * For shared libraries, symbol visibility mechanisms prevent these symbols
+ * from being exported, but for static libraries, naming collisions are a real
+ * possibility.
+ */
+#undef JEMALLOC_PRIVATE_NAMESPACE
+
+/*
+ * Hyper-threaded CPUs may need a special instruction inside spin loops in
+ * order to yield to another virtual CPU.
+ */
+#undef CPU_SPINWAIT
+/* 1 if CPU_SPINWAIT is defined, 0 otherwise. */
+#undef HAVE_CPU_SPINWAIT
+
+/*
+ * Number of significant bits in virtual addresses.  This may be less than the
+ * total number of bits in a pointer, e.g. on x64, for which the uppermost 16
+ * bits are the same as bit 47.
+ */
+#undef LG_VADDR
+
+/* Defined if C11 atomics are available. */
+#undef JEMALLOC_C11_ATOMICS
+
+/* Defined if GCC __atomic atomics are available. */
+#undef JEMALLOC_GCC_ATOMIC_ATOMICS
+/* and the 8-bit variant support. */
+#undef JEMALLOC_GCC_U8_ATOMIC_ATOMICS
+
+/* Defined if GCC __sync atomics are available. */
+#undef JEMALLOC_GCC_SYNC_ATOMICS
+/* and the 8-bit variant support. */
+#undef JEMALLOC_GCC_U8_SYNC_ATOMICS
+
+/*
+ * Defined if __builtin_clz() and __builtin_clzl() are available.
+ */
+#undef JEMALLOC_HAVE_BUILTIN_CLZ
+
+/*
+ * Defined if os_unfair_lock_*() functions are available, as provided by Darwin.
+ */
+#undef JEMALLOC_OS_UNFAIR_LOCK
+
+/* Defined if syscall(2) is usable. */
+#undef JEMALLOC_USE_SYSCALL
+
+/*
+ * Defined if secure_getenv(3) is available.
+ */
+#undef JEMALLOC_HAVE_SECURE_GETENV
+
+/*
+ * Defined if issetugid(2) is available.
+ */
+#undef JEMALLOC_HAVE_ISSETUGID
+
+/* Defined if pthread_atfork(3) is available. */
+#undef JEMALLOC_HAVE_PTHREAD_ATFORK
+
+/* Defined if pthread_setname_np(3) is available. */
+#undef JEMALLOC_HAVE_PTHREAD_SETNAME_NP
+
+/*
+ * Defined if clock_gettime(CLOCK_MONOTONIC_COARSE, ...) is available.
+ */
+#undef JEMALLOC_HAVE_CLOCK_MONOTONIC_COARSE
+
+/*
+ * Defined if clock_gettime(CLOCK_MONOTONIC, ...) is available.
+ */
+#undef JEMALLOC_HAVE_CLOCK_MONOTONIC
+
+/*
+ * Defined if mach_absolute_time() is available.
+ */
+#undef JEMALLOC_HAVE_MACH_ABSOLUTE_TIME
+
+/*
+ * Defined if _malloc_thread_cleanup() exists.  At least in the case of
+ * FreeBSD, pthread_key_create() allocates, which if used during malloc
+ * bootstrapping will cause recursion into the pthreads library.  Therefore, if
+ * _malloc_thread_cleanup() exists, use it as the basis for thread cleanup in
+ * malloc_tsd.
+ */
+#undef JEMALLOC_MALLOC_THREAD_CLEANUP
+
+/*
+ * Defined if threaded initialization is known to be safe on this platform.
+ * Among other things, it must be possible to initialize a mutex without
+ * triggering allocation in order for threaded allocation to be safe.
+ */
+#undef JEMALLOC_THREADED_INIT
+
+/*
+ * Defined if the pthreads implementation defines
+ * _pthread_mutex_init_calloc_cb(), in which case the function is used in order
+ * to avoid recursive allocation during mutex initialization.
+ */
+#undef JEMALLOC_MUTEX_INIT_CB
+
+/* Non-empty if the tls_model attribute is supported. */
+#undef JEMALLOC_TLS_MODEL
+
+/*
+ * JEMALLOC_DEBUG enables assertions and other sanity checks, and disables
+ * inline functions.
+ */
+#undef JEMALLOC_DEBUG
+
+/* JEMALLOC_STATS enables statistics calculation. */
+#undef JEMALLOC_STATS
+
+/* JEMALLOC_EXPERIMENTAL_SMALLOCX_API enables experimental smallocx API. */
+#undef JEMALLOC_EXPERIMENTAL_SMALLOCX_API
+
+/* JEMALLOC_PROF enables allocation profiling. */
+#undef JEMALLOC_PROF
+
+/* Use libunwind for profile backtracing if defined. */
+#undef JEMALLOC_PROF_LIBUNWIND
+
+/* Use libgcc for profile backtracing if defined. */
+#undef JEMALLOC_PROF_LIBGCC
+
+/* Use gcc intrinsics for profile backtracing if defined. */
+#undef JEMALLOC_PROF_GCC
+
+/*
+ * JEMALLOC_DSS enables use of sbrk(2) to allocate extents from the data storage
+ * segment (DSS).
+ */
+#undef JEMALLOC_DSS
+
+/* Support memory filling (junk/zero). */
+#undef JEMALLOC_FILL
+
+/* Support utrace(2)-based tracing. */
+#undef JEMALLOC_UTRACE
+
+/* Support optional abort() on OOM. */
+#undef JEMALLOC_XMALLOC
+
+/* Support lazy locking (avoid locking unless a second thread is launched). */
+#undef JEMALLOC_LAZY_LOCK
+
+/*
+ * Minimum allocation alignment is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#undef LG_QUANTUM
+
+/* One page is 2^LG_PAGE bytes. */
+#undef LG_PAGE
+
+/*
+ * One huge page is 2^LG_HUGEPAGE bytes.  Note that this is defined even if the
+ * system does not explicitly support huge pages; system calls that require
+ * explicit huge page support are separately configured.
+ */
+#undef LG_HUGEPAGE
+
+/*
+ * If defined, adjacent virtual memory mappings with identical attributes
+ * automatically coalesce, and they fragment when changes are made to subranges.
+ * This is the normal order of things for mmap()/munmap(), but on Windows
+ * VirtualAlloc()/VirtualFree() operations must be precisely matched, i.e.
+ * mappings do *not* coalesce/fragment.
+ */
+#undef JEMALLOC_MAPS_COALESCE
+
+/*
+ * If defined, retain memory for later reuse by default rather than using e.g.
+ * munmap() to unmap freed extents.  This is enabled on 64-bit Linux because
+ * common sequences of mmap()/munmap() calls will cause virtual memory map
+ * holes.
+ */
+#undef JEMALLOC_RETAIN
+
+/* TLS is used to map arenas and magazine caches to threads. */
+#undef JEMALLOC_TLS
+
+/*
+ * Used to mark unreachable code to quiet "end of non-void" compiler warnings.
+ * Don't use this directly; instead use unreachable() from util.h
+ */
+#undef JEMALLOC_INTERNAL_UNREACHABLE
+
+/*
+ * ffs*() functions to use for bitmapping.  Don't use these directly; instead,
+ * use ffs_*() from util.h.
+ */
+#undef JEMALLOC_INTERNAL_FFSLL
+#undef JEMALLOC_INTERNAL_FFSL
+#undef JEMALLOC_INTERNAL_FFS
+
+/*
+ * popcount*() functions to use for bitmapping.
+ */
+#undef JEMALLOC_INTERNAL_POPCOUNTL
+#undef JEMALLOC_INTERNAL_POPCOUNT
+
+/*
+ * If defined, explicitly attempt to more uniformly distribute large allocation
+ * pointer alignments across all cache indices.
+ */
+#undef JEMALLOC_CACHE_OBLIVIOUS
+
+/*
+ * If defined, enable logging facilities.  We make this a configure option to
+ * avoid taking extra branches everywhere.
+ */
+#undef JEMALLOC_LOG
+
+/*
+ * If defined, use readlinkat() (instead of readlink()) to follow
+ * /etc/malloc_conf.
+ */
+#undef JEMALLOC_READLINKAT
+
+/*
+ * Darwin (OS X) uses zones to work around Mach-O symbol override shortcomings.
+ */
+#undef JEMALLOC_ZONE
+
+/*
+ * Methods for determining whether the OS overcommits.
+ * JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY: Linux's
+ *                                         /proc/sys/vm.overcommit_memory file.
+ * JEMALLOC_SYSCTL_VM_OVERCOMMIT: FreeBSD's vm.overcommit sysctl.
+ */
+#undef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+#undef JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY
+
+/* Defined if madvise(2) is available. */
+#undef JEMALLOC_HAVE_MADVISE
+
+/*
+ * Defined if transparent huge pages are supported via the MADV_[NO]HUGEPAGE
+ * arguments to madvise(2).
+ */
+#undef JEMALLOC_HAVE_MADVISE_HUGE
+
+/*
+ * Methods for purging unused pages differ between operating systems.
+ *
+ *   madvise(..., MADV_FREE) : This marks pages as being unused, such that they
+ *                             will be discarded rather than swapped out.
+ *   madvise(..., MADV_DONTNEED) : If JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS is
+ *                                 defined, this immediately discards pages,
+ *                                 such that new pages will be demand-zeroed if
+ *                                 the address region is later touched;
+ *                                 otherwise this behaves similarly to
+ *                                 MADV_FREE, though typically with higher
+ *                                 system overhead.
+ */
+#undef JEMALLOC_PURGE_MADVISE_FREE
+#undef JEMALLOC_PURGE_MADVISE_DONTNEED
+#undef JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS
+
+/* Defined if madvise(2) is available but MADV_FREE is not (x86 Linux only). */
+#undef JEMALLOC_DEFINE_MADVISE_FREE
+
+/*
+ * Defined if MADV_DO[NT]DUMP is supported as an argument to madvise.
+ */
+#undef JEMALLOC_MADVISE_DONTDUMP
+
+/*
+ * Defined if transparent huge pages (THPs) are supported via the
+ * MADV_[NO]HUGEPAGE arguments to madvise(2), and THP support is enabled.
+ */
+#undef JEMALLOC_THP
+
+/* Define if operating system has alloca.h header. */
+#undef JEMALLOC_HAS_ALLOCA_H
+
+/* C99 restrict keyword supported. */
+#undef JEMALLOC_HAS_RESTRICT
+
+/* For use by hash code. */
+#undef JEMALLOC_BIG_ENDIAN
+
+/* sizeof(int) == 2^LG_SIZEOF_INT. */
+#undef LG_SIZEOF_INT
+
+/* sizeof(long) == 2^LG_SIZEOF_LONG. */
+#undef LG_SIZEOF_LONG
+
+/* sizeof(long long) == 2^LG_SIZEOF_LONG_LONG. */
+#undef LG_SIZEOF_LONG_LONG
+
+/* sizeof(intmax_t) == 2^LG_SIZEOF_INTMAX_T. */
+#undef LG_SIZEOF_INTMAX_T
+
+/* glibc malloc hooks (__malloc_hook, __realloc_hook, __free_hook). */
+#undef JEMALLOC_GLIBC_MALLOC_HOOK
+
+/* glibc memalign hook. */
+#undef JEMALLOC_GLIBC_MEMALIGN_HOOK
+
+/* pthread support */
+#undef JEMALLOC_HAVE_PTHREAD
+
+/* dlsym() support */
+#undef JEMALLOC_HAVE_DLSYM
+
+/* Adaptive mutex support in pthreads. */
+#undef JEMALLOC_HAVE_PTHREAD_MUTEX_ADAPTIVE_NP
+
+/* GNU specific sched_getcpu support */
+#undef JEMALLOC_HAVE_SCHED_GETCPU
+
+/* GNU specific sched_setaffinity support */
+#undef JEMALLOC_HAVE_SCHED_SETAFFINITY
+
+/*
+ * If defined, all the features necessary for background threads are present.
+ */
+#undef JEMALLOC_BACKGROUND_THREAD
+
+/*
+ * If defined, jemalloc symbols are not exported (doesn't work when
+ * JEMALLOC_PREFIX is not defined).
+ */
+#undef JEMALLOC_EXPORT
+
+/* config.malloc_conf options string. */
+#undef JEMALLOC_CONFIG_MALLOC_CONF
+
+/* If defined, jemalloc takes the malloc/free/etc. symbol names. */
+#undef JEMALLOC_IS_MALLOC
+
+/*
+ * Defined if strerror_r returns char * if _GNU_SOURCE is defined.
+ */
+#undef JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE
+
+/* Performs additional safety checks when defined. */
+#undef JEMALLOC_OPT_SAFETY_CHECKS
+
+#endif /* JEMALLOC_INTERNAL_DEFS_H_ */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
new file mode 100644
index 0000000..d291170
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_externs.h
@@ -0,0 +1,57 @@
+#ifndef JEMALLOC_INTERNAL_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTERNS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/tsd_types.h"
+
+/* TSD checks this to set thread local slow state accordingly. */
+extern bool malloc_slow;
+
+/* Run-time options. */
+extern bool opt_abort;
+extern bool opt_abort_conf;
+extern bool opt_confirm_conf;
+extern const char *opt_junk;
+extern bool opt_junk_alloc;
+extern bool opt_junk_free;
+extern bool opt_utrace;
+extern bool opt_xmalloc;
+extern bool opt_zero;
+extern unsigned opt_narenas;
+
+/* Number of CPUs. */
+extern unsigned ncpus;
+
+/* Number of arenas used for automatic multiplexing of threads and arenas. */
+extern unsigned narenas_auto;
+
+/* Base index for manual arenas. */
+extern unsigned manual_arena_base;
+
+/*
+ * Arenas that are used to service external requests.  Not all elements of the
+ * arenas array are necessarily used; arenas are created lazily as needed.
+ */
+extern atomic_p_t arenas[];
+
+void *a0malloc(size_t size);
+void a0dalloc(void *ptr);
+void *bootstrap_malloc(size_t size);
+void *bootstrap_calloc(size_t num, size_t size);
+void bootstrap_free(void *ptr);
+void arena_set(unsigned ind, arena_t *arena);
+unsigned narenas_total_get(void);
+arena_t *arena_init(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+arena_tdata_t *arena_tdata_get_hard(tsd_t *tsd, unsigned ind);
+arena_t *arena_choose_hard(tsd_t *tsd, bool internal);
+void arena_migrate(tsd_t *tsd, unsigned oldind, unsigned newind);
+void iarena_cleanup(tsd_t *tsd);
+void arena_cleanup(tsd_t *tsd);
+void arenas_tdata_cleanup(tsd_t *tsd);
+void jemalloc_prefork(void);
+void jemalloc_postfork_parent(void);
+void jemalloc_postfork_child(void);
+bool malloc_initialized(void);
+void je_sdallocx_noflags(void *ptr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
new file mode 100644
index 0000000..437eaa4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_includes.h
@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_INCLUDES_H
+#define JEMALLOC_INTERNAL_INCLUDES_H
+
+/*
+ * jemalloc can conceptually be broken into components (arena, tcache, etc.),
+ * but there are circular dependencies that cannot be broken without
+ * substantial performance degradation.
+ *
+ * Historically, we dealt with this by each header into four sections (types,
+ * structs, externs, and inlines), and included each header file multiple times
+ * in this file, picking out the portion we want on each pass using the
+ * following #defines:
+ *   JEMALLOC_H_TYPES   : Preprocessor-defined constants and psuedo-opaque data
+ *                        types.
+ *   JEMALLOC_H_STRUCTS : Data structures.
+ *   JEMALLOC_H_EXTERNS : Extern data declarations and function prototypes.
+ *   JEMALLOC_H_INLINES : Inline functions.
+ *
+ * We're moving toward a world in which the dependencies are explicit; each file
+ * will #include the headers it depends on (rather than relying on them being
+ * implicitly available via this file including every header file in the
+ * project).
+ *
+ * We're now in an intermediate state: we've broken up the header files to avoid
+ * having to include each one multiple times, but have not yet moved the
+ * dependency information into the header files (i.e. we still rely on the
+ * ordering in this file to ensure all a header's dependencies are available in
+ * its translation unit).  Each component is now broken up into multiple header
+ * files, corresponding to the sections above (e.g. instead of "foo.h", we now
+ * have "foo_types.h", "foo_structs.h", "foo_externs.h", "foo_inlines.h").
+ *
+ * Those files which have been converted to explicitly include their
+ * inter-component dependencies are now in the initial HERMETIC HEADERS
+ * section.  All headers may still rely on jemalloc_preamble.h (which, by fiat,
+ * must be included first in every translation unit) for system headers and
+ * global jemalloc definitions, however.
+ */
+
+/******************************************************************************/
+/* TYPES */
+/******************************************************************************/
+
+#include "jemalloc/internal/extent_types.h"
+#include "jemalloc/internal/base_types.h"
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/tcache_types.h"
+#include "jemalloc/internal/prof_types.h"
+
+/******************************************************************************/
+/* STRUCTS */
+/******************************************************************************/
+
+#include "jemalloc/internal/arena_structs_a.h"
+#include "jemalloc/internal/extent_structs.h"
+#include "jemalloc/internal/base_structs.h"
+#include "jemalloc/internal/prof_structs.h"
+#include "jemalloc/internal/arena_structs_b.h"
+#include "jemalloc/internal/tcache_structs.h"
+#include "jemalloc/internal/background_thread_structs.h"
+
+/******************************************************************************/
+/* EXTERNS */
+/******************************************************************************/
+
+#include "jemalloc/internal/jemalloc_internal_externs.h"
+#include "jemalloc/internal/extent_externs.h"
+#include "jemalloc/internal/base_externs.h"
+#include "jemalloc/internal/arena_externs.h"
+#include "jemalloc/internal/large_externs.h"
+#include "jemalloc/internal/tcache_externs.h"
+#include "jemalloc/internal/prof_externs.h"
+#include "jemalloc/internal/background_thread_externs.h"
+
+/******************************************************************************/
+/* INLINES */
+/******************************************************************************/
+
+#include "jemalloc/internal/jemalloc_internal_inlines_a.h"
+#include "jemalloc/internal/base_inlines.h"
+/*
+ * Include portions of arena code interleaved with tcache code in order to
+ * resolve circular dependencies.
+ */
+#include "jemalloc/internal/prof_inlines_a.h"
+#include "jemalloc/internal/arena_inlines_a.h"
+#include "jemalloc/internal/extent_inlines.h"
+#include "jemalloc/internal/jemalloc_internal_inlines_b.h"
+#include "jemalloc/internal/tcache_inlines.h"
+#include "jemalloc/internal/arena_inlines_b.h"
+#include "jemalloc/internal/jemalloc_internal_inlines_c.h"
+#include "jemalloc/internal/prof_inlines_b.h"
+#include "jemalloc/internal/background_thread_inlines.h"
+
+#endif /* JEMALLOC_INTERNAL_INCLUDES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
new file mode 100644
index 0000000..ddde9b4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_a.h
@@ -0,0 +1,174 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_A_H
+#define JEMALLOC_INTERNAL_INLINES_A_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/ticker.h"
+
+JEMALLOC_ALWAYS_INLINE malloc_cpuid_t
+malloc_getcpu(void) {
+	assert(have_percpu_arena);
+#if defined(_WIN32)
+	return GetCurrentProcessorNumber();
+#elif defined(JEMALLOC_HAVE_SCHED_GETCPU)
+	return (malloc_cpuid_t)sched_getcpu();
+#else
+	not_reached();
+	return -1;
+#endif
+}
+
+/* Return the chosen arena index based on current cpu. */
+JEMALLOC_ALWAYS_INLINE unsigned
+percpu_arena_choose(void) {
+	assert(have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena));
+
+	malloc_cpuid_t cpuid = malloc_getcpu();
+	assert(cpuid >= 0);
+
+	unsigned arena_ind;
+	if ((opt_percpu_arena == percpu_arena) || ((unsigned)cpuid < ncpus /
+	    2)) {
+		arena_ind = cpuid;
+	} else {
+		assert(opt_percpu_arena == per_phycpu_arena);
+		/* Hyper threads on the same physical CPU share arena. */
+		arena_ind = cpuid - ncpus / 2;
+	}
+
+	return arena_ind;
+}
+
+/* Return the limit of percpu auto arena range, i.e. arenas[0...ind_limit). */
+JEMALLOC_ALWAYS_INLINE unsigned
+percpu_arena_ind_limit(percpu_arena_mode_t mode) {
+	assert(have_percpu_arena && PERCPU_ARENA_ENABLED(mode));
+	if (mode == per_phycpu_arena && ncpus > 1) {
+		if (ncpus % 2) {
+			/* This likely means a misconfig. */
+			return ncpus / 2 + 1;
+		}
+		return ncpus / 2;
+	} else {
+		return ncpus;
+	}
+}
+
+static inline arena_tdata_t *
+arena_tdata_get(tsd_t *tsd, unsigned ind, bool refresh_if_missing) {
+	arena_tdata_t *tdata;
+	arena_tdata_t *arenas_tdata = tsd_arenas_tdata_get(tsd);
+
+	if (unlikely(arenas_tdata == NULL)) {
+		/* arenas_tdata hasn't been initialized yet. */
+		return arena_tdata_get_hard(tsd, ind);
+	}
+	if (unlikely(ind >= tsd_narenas_tdata_get(tsd))) {
+		/*
+		 * ind is invalid, cache is old (too small), or tdata to be
+		 * initialized.
+		 */
+		return (refresh_if_missing ? arena_tdata_get_hard(tsd, ind) :
+		    NULL);
+	}
+
+	tdata = &arenas_tdata[ind];
+	if (likely(tdata != NULL) || !refresh_if_missing) {
+		return tdata;
+	}
+	return arena_tdata_get_hard(tsd, ind);
+}
+
+static inline arena_t *
+arena_get(tsdn_t *tsdn, unsigned ind, bool init_if_missing) {
+	arena_t *ret;
+
+	assert(ind < MALLOCX_ARENA_LIMIT);
+
+	ret = (arena_t *)atomic_load_p(&arenas[ind], ATOMIC_ACQUIRE);
+	if (unlikely(ret == NULL)) {
+		if (init_if_missing) {
+			ret = arena_init(tsdn, ind,
+			    (extent_hooks_t *)&extent_hooks_default);
+		}
+	}
+	return ret;
+}
+
+static inline ticker_t *
+decay_ticker_get(tsd_t *tsd, unsigned ind) {
+	arena_tdata_t *tdata;
+
+	tdata = arena_tdata_get(tsd, ind, true);
+	if (unlikely(tdata == NULL)) {
+		return NULL;
+	}
+	return &tdata->decay_ticker;
+}
+
+JEMALLOC_ALWAYS_INLINE cache_bin_t *
+tcache_small_bin_get(tcache_t *tcache, szind_t binind) {
+	assert(binind < SC_NBINS);
+	return &tcache->bins_small[binind];
+}
+
+JEMALLOC_ALWAYS_INLINE cache_bin_t *
+tcache_large_bin_get(tcache_t *tcache, szind_t binind) {
+	assert(binind >= SC_NBINS &&binind < nhbins);
+	return &tcache->bins_large[binind - SC_NBINS];
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tcache_available(tsd_t *tsd) {
+	/*
+	 * Thread specific auto tcache might be unavailable if: 1) during tcache
+	 * initialization, or 2) disabled through thread.tcache.enabled mallctl
+	 * or config options.  This check covers all cases.
+	 */
+	if (likely(tsd_tcache_enabled_get(tsd))) {
+		/* Associated arena == NULL implies tcache init in progress. */
+		assert(tsd_tcachep_get(tsd)->arena == NULL ||
+		    tcache_small_bin_get(tsd_tcachep_get(tsd), 0)->avail !=
+		    NULL);
+		return true;
+	}
+
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE tcache_t *
+tcache_get(tsd_t *tsd) {
+	if (!tcache_available(tsd)) {
+		return NULL;
+	}
+
+	return tsd_tcachep_get(tsd);
+}
+
+static inline void
+pre_reentrancy(tsd_t *tsd, arena_t *arena) {
+	/* arena is the current context.  Reentry from a0 is not allowed. */
+	assert(arena != arena_get(tsd_tsdn(tsd), 0, false));
+
+	bool fast = tsd_fast(tsd);
+	assert(tsd_reentrancy_level_get(tsd) < INT8_MAX);
+	++*tsd_reentrancy_levelp_get(tsd);
+	if (fast) {
+		/* Prepare slow path for reentrancy. */
+		tsd_slow_update(tsd);
+		assert(tsd_state_get(tsd) == tsd_state_nominal_slow);
+	}
+}
+
+static inline void
+post_reentrancy(tsd_t *tsd) {
+	int8_t *reentrancy_level = tsd_reentrancy_levelp_get(tsd);
+	assert(*reentrancy_level > 0);
+	if (--*reentrancy_level == 0) {
+		tsd_slow_update(tsd);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_A_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
new file mode 100644
index 0000000..70d6e57
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_b.h
@@ -0,0 +1,87 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_B_H
+#define JEMALLOC_INTERNAL_INLINES_B_H
+
+#include "jemalloc/internal/rtree.h"
+
+/* Choose an arena based on a per-thread value. */
+static inline arena_t *
+arena_choose_impl(tsd_t *tsd, arena_t *arena, bool internal) {
+	arena_t *ret;
+
+	if (arena != NULL) {
+		return arena;
+	}
+
+	/* During reentrancy, arena 0 is the safest bet. */
+	if (unlikely(tsd_reentrancy_level_get(tsd) > 0)) {
+		return arena_get(tsd_tsdn(tsd), 0, true);
+	}
+
+	ret = internal ? tsd_iarena_get(tsd) : tsd_arena_get(tsd);
+	if (unlikely(ret == NULL)) {
+		ret = arena_choose_hard(tsd, internal);
+		assert(ret);
+		if (tcache_available(tsd)) {
+			tcache_t *tcache = tcache_get(tsd);
+			if (tcache->arena != NULL) {
+				/* See comments in tcache_data_init().*/
+				assert(tcache->arena ==
+				    arena_get(tsd_tsdn(tsd), 0, false));
+				if (tcache->arena != ret) {
+					tcache_arena_reassociate(tsd_tsdn(tsd),
+					    tcache, ret);
+				}
+			} else {
+				tcache_arena_associate(tsd_tsdn(tsd), tcache,
+				    ret);
+			}
+		}
+	}
+
+	/*
+	 * Note that for percpu arena, if the current arena is outside of the
+	 * auto percpu arena range, (i.e. thread is assigned to a manually
+	 * managed arena), then percpu arena is skipped.
+	 */
+	if (have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena) &&
+	    !internal && (arena_ind_get(ret) <
+	    percpu_arena_ind_limit(opt_percpu_arena)) && (ret->last_thd !=
+	    tsd_tsdn(tsd))) {
+		unsigned ind = percpu_arena_choose();
+		if (arena_ind_get(ret) != ind) {
+			percpu_arena_update(tsd, ind);
+			ret = tsd_arena_get(tsd);
+		}
+		ret->last_thd = tsd_tsdn(tsd);
+	}
+
+	return ret;
+}
+
+static inline arena_t *
+arena_choose(tsd_t *tsd, arena_t *arena) {
+	return arena_choose_impl(tsd, arena, false);
+}
+
+static inline arena_t *
+arena_ichoose(tsd_t *tsd, arena_t *arena) {
+	return arena_choose_impl(tsd, arena, true);
+}
+
+static inline bool
+arena_is_auto(arena_t *arena) {
+	assert(narenas_auto > 0);
+
+	return (arena_ind_get(arena) < manual_arena_base);
+}
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+iealloc(tsdn_t *tsdn, const void *ptr) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	return rtree_extent_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_B_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
new file mode 100644
index 0000000..cdb10eb
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_inlines_c.h
@@ -0,0 +1,222 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_C_H
+#define JEMALLOC_INTERNAL_INLINES_C_H
+
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/witness.h"
+
+/*
+ * Translating the names of the 'i' functions:
+ *   Abbreviations used in the first part of the function name (before
+ *   alloc/dalloc) describe what that function accomplishes:
+ *     a: arena (query)
+ *     s: size (query, or sized deallocation)
+ *     e: extent (query)
+ *     p: aligned (allocates)
+ *     vs: size (query, without knowing that the pointer is into the heap)
+ *     r: rallocx implementation
+ *     x: xallocx implementation
+ *   Abbreviations used in the second part of the function name (after
+ *   alloc/dalloc) describe the arguments it takes
+ *     z: whether to return zeroed memory
+ *     t: accepts a tcache_t * parameter
+ *     m: accepts an arena_t * parameter
+ */
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+iaalloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	return arena_aalloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+isalloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	return arena_salloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iallocztm(tsdn_t *tsdn, size_t size, szind_t ind, bool zero, tcache_t *tcache,
+    bool is_internal, arena_t *arena, bool slow_path) {
+	void *ret;
+
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena == NULL || arena_is_auto(arena));
+	if (!tsdn_null(tsdn) && tsd_reentrancy_level_get(tsdn_tsd(tsdn)) == 0) {
+		witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+		    WITNESS_RANK_CORE, 0);
+	}
+
+	ret = arena_malloc(tsdn, arena, size, ind, zero, tcache, slow_path);
+	if (config_stats && is_internal && likely(ret != NULL)) {
+		arena_internal_add(iaalloc(tsdn, ret), isalloc(tsdn, ret));
+	}
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ialloc(tsd_t *tsd, size_t size, szind_t ind, bool zero, bool slow_path) {
+	return iallocztm(tsd_tsdn(tsd), size, ind, zero, tcache_get(tsd), false,
+	    NULL, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipallocztm(tsdn_t *tsdn, size_t usize, size_t alignment, bool zero,
+    tcache_t *tcache, bool is_internal, arena_t *arena) {
+	void *ret;
+
+	assert(usize != 0);
+	assert(usize == sz_sa2u(usize, alignment));
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena == NULL || arena_is_auto(arena));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	ret = arena_palloc(tsdn, arena, usize, alignment, zero, tcache);
+	assert(ALIGNMENT_ADDR2BASE(ret, alignment) == ret);
+	if (config_stats && is_internal && likely(ret != NULL)) {
+		arena_internal_add(iaalloc(tsdn, ret), isalloc(tsdn, ret));
+	}
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipalloct(tsdn_t *tsdn, size_t usize, size_t alignment, bool zero,
+    tcache_t *tcache, arena_t *arena) {
+	return ipallocztm(tsdn, usize, alignment, zero, tcache, false, arena);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipalloc(tsd_t *tsd, size_t usize, size_t alignment, bool zero) {
+	return ipallocztm(tsd_tsdn(tsd), usize, alignment, zero,
+	    tcache_get(tsd), false, NULL);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ivsalloc(tsdn_t *tsdn, const void *ptr) {
+	return arena_vsalloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idalloctm(tsdn_t *tsdn, void *ptr, tcache_t *tcache, alloc_ctx_t *alloc_ctx,
+    bool is_internal, bool slow_path) {
+	assert(ptr != NULL);
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena_is_auto(iaalloc(tsdn, ptr)));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	if (config_stats && is_internal) {
+		arena_internal_sub(iaalloc(tsdn, ptr), isalloc(tsdn, ptr));
+	}
+	if (!is_internal && !tsdn_null(tsdn) &&
+	    tsd_reentrancy_level_get(tsdn_tsd(tsdn)) != 0) {
+		assert(tcache == NULL);
+	}
+	arena_dalloc(tsdn, ptr, tcache, alloc_ctx, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idalloc(tsd_t *tsd, void *ptr) {
+	idalloctm(tsd_tsdn(tsd), ptr, tcache_get(tsd), NULL, false, true);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+isdalloct(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	arena_sdalloc(tsdn, ptr, size, tcache, alloc_ctx, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloct_realign(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t alignment, bool zero, tcache_t *tcache, arena_t *arena,
+    hook_ralloc_args_t *hook_args) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	void *p;
+	size_t usize, copysize;
+
+	usize = sz_sa2u(size, alignment);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+	p = ipalloct(tsdn, usize, alignment, zero, tcache, arena);
+	if (p == NULL) {
+		return NULL;
+	}
+	/*
+	 * Copy at most size bytes (not size+extra), since the caller has no
+	 * expectation that the extra bytes will be reliably preserved.
+	 */
+	copysize = (size < oldsize) ? size : oldsize;
+	memcpy(p, ptr, copysize);
+	hook_invoke_alloc(hook_args->is_realloc
+	    ? hook_alloc_realloc : hook_alloc_rallocx, p, (uintptr_t)p,
+	    hook_args->args);
+	hook_invoke_dalloc(hook_args->is_realloc
+	    ? hook_dalloc_realloc : hook_dalloc_rallocx, ptr, hook_args->args);
+	isdalloct(tsdn, ptr, oldsize, tcache, NULL, true);
+	return p;
+}
+
+/*
+ * is_realloc threads through the knowledge of whether or not this call comes
+ * from je_realloc (as opposed to je_rallocx); this ensures that we pass the
+ * correct entry point into any hooks.
+ * Note that these functions are all force-inlined, so no actual bool gets
+ * passed-around anywhere.
+ */
+JEMALLOC_ALWAYS_INLINE void *
+iralloct(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size, size_t alignment,
+    bool zero, tcache_t *tcache, arena_t *arena, hook_ralloc_args_t *hook_args)
+{
+	assert(ptr != NULL);
+	assert(size != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		/*
+		 * Existing object alignment is inadequate; allocate new space
+		 * and copy.
+		 */
+		return iralloct_realign(tsdn, ptr, oldsize, size, alignment,
+		    zero, tcache, arena, hook_args);
+	}
+
+	return arena_ralloc(tsdn, arena, ptr, oldsize, size, alignment, zero,
+	    tcache, hook_args);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloc(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t alignment,
+    bool zero, hook_ralloc_args_t *hook_args) {
+	return iralloct(tsd_tsdn(tsd), ptr, oldsize, size, alignment, zero,
+	    tcache_get(tsd), NULL, hook_args);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+ixalloc(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero, size_t *newsize) {
+	assert(ptr != NULL);
+	assert(size != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		/* Existing object alignment is inadequate. */
+		*newsize = oldsize;
+		return true;
+	}
+
+	return arena_ralloc_no_move(tsdn, ptr, oldsize, size, extra, zero,
+	    newsize);
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_C_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
new file mode 100644
index 0000000..d8ea06f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_macros.h
@@ -0,0 +1,114 @@
+#ifndef JEMALLOC_INTERNAL_MACROS_H
+#define JEMALLOC_INTERNAL_MACROS_H
+
+#ifdef JEMALLOC_DEBUG
+#  define JEMALLOC_ALWAYS_INLINE static inline
+#else
+#  define JEMALLOC_ALWAYS_INLINE JEMALLOC_ATTR(always_inline) static inline
+#endif
+#ifdef _MSC_VER
+#  define inline _inline
+#endif
+
+#define UNUSED JEMALLOC_ATTR(unused)
+
+#define ZU(z)	((size_t)z)
+#define ZD(z)	((ssize_t)z)
+#define QU(q)	((uint64_t)q)
+#define QD(q)	((int64_t)q)
+
+#define KZU(z)	ZU(z##ULL)
+#define KZD(z)	ZD(z##LL)
+#define KQU(q)	QU(q##ULL)
+#define KQD(q)	QI(q##LL)
+
+#ifndef __DECONST
+#  define	__DECONST(type, var)	((type)(uintptr_t)(const void *)(var))
+#endif
+
+#if !defined(JEMALLOC_HAS_RESTRICT) || defined(__cplusplus)
+#  define restrict
+#endif
+
+/* Various function pointers are static and immutable except during testing. */
+#ifdef JEMALLOC_JET
+#  define JET_MUTABLE
+#else
+#  define JET_MUTABLE const
+#endif
+
+#define JEMALLOC_VA_ARGS_HEAD(head, ...) head
+#define JEMALLOC_VA_ARGS_TAIL(head, ...) __VA_ARGS__
+
+#if (defined(__GNUC__) || defined(__GNUG__)) && !defined(__clang__) \
+  && defined(JEMALLOC_HAVE_ATTR) && (__GNUC__ >= 7)
+#define JEMALLOC_FALLTHROUGH JEMALLOC_ATTR(fallthrough);
+#else
+#define JEMALLOC_FALLTHROUGH /* falls through */
+#endif
+
+/* Diagnostic suppression macros */
+#if defined(_MSC_VER) && !defined(__clang__)
+#  define JEMALLOC_DIAGNOSTIC_PUSH __pragma(warning(push))
+#  define JEMALLOC_DIAGNOSTIC_POP __pragma(warning(pop))
+#  define JEMALLOC_DIAGNOSTIC_IGNORE(W) __pragma(warning(disable:W))
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+#  define JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+/* #pragma GCC diagnostic first appeared in gcc 4.6. */
+#elif (defined(__GNUC__) && ((__GNUC__ > 4) || ((__GNUC__ == 4) && \
+  (__GNUC_MINOR__ > 5)))) || defined(__clang__)
+/*
+ * The JEMALLOC_PRAGMA__ macro is an implementation detail of the GCC and Clang
+ * diagnostic suppression macros and should not be used anywhere else.
+ */
+#  define JEMALLOC_PRAGMA__(X) _Pragma(#X)
+#  define JEMALLOC_DIAGNOSTIC_PUSH JEMALLOC_PRAGMA__(GCC diagnostic push)
+#  define JEMALLOC_DIAGNOSTIC_POP JEMALLOC_PRAGMA__(GCC diagnostic pop)
+#  define JEMALLOC_DIAGNOSTIC_IGNORE(W) \
+     JEMALLOC_PRAGMA__(GCC diagnostic ignored W)
+
+/*
+ * The -Wmissing-field-initializers warning is buggy in GCC versions < 5.1 and
+ * all clang versions up to version 7 (currently trunk, unreleased).  This macro
+ * suppresses the warning for the affected compiler versions only.
+ */
+#  if ((defined(__GNUC__) && !defined(__clang__)) && (__GNUC__ < 5)) || \
+     defined(__clang__)
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS  \
+          JEMALLOC_DIAGNOSTIC_IGNORE("-Wmissing-field-initializers")
+#  else
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+#  endif
+
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS  \
+     JEMALLOC_DIAGNOSTIC_IGNORE("-Wtype-limits")
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_UNUSED_PARAMETER \
+     JEMALLOC_DIAGNOSTIC_IGNORE("-Wunused-parameter")
+#  if defined(__GNUC__) && !defined(__clang__) && (__GNUC__ >= 7)
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN \
+       JEMALLOC_DIAGNOSTIC_IGNORE("-Walloc-size-larger-than=")
+#  else
+#    define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+#  endif
+#  define JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS \
+  JEMALLOC_DIAGNOSTIC_PUSH \
+  JEMALLOC_DIAGNOSTIC_IGNORE_UNUSED_PARAMETER
+#else
+#  define JEMALLOC_DIAGNOSTIC_PUSH
+#  define JEMALLOC_DIAGNOSTIC_POP
+#  define JEMALLOC_DIAGNOSTIC_IGNORE(W)
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS
+#  define JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+#  define JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+#endif
+
+/*
+ * Disables spurious diagnostics for all headers.  Since these headers are not
+ * included by users directly, it does not affect their diagnostic settings.
+ */
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+#endif /* JEMALLOC_INTERNAL_MACROS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
new file mode 100644
index 0000000..e296c5a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_internal_types.h
@@ -0,0 +1,114 @@
+#ifndef JEMALLOC_INTERNAL_TYPES_H
+#define JEMALLOC_INTERNAL_TYPES_H
+
+#include "jemalloc/internal/quantum.h"
+
+/* Page size index type. */
+typedef unsigned pszind_t;
+
+/* Size class index type. */
+typedef unsigned szind_t;
+
+/* Processor / core id type. */
+typedef int malloc_cpuid_t;
+
+/*
+ * Flags bits:
+ *
+ * a: arena
+ * t: tcache
+ * 0: unused
+ * z: zero
+ * n: alignment
+ *
+ * aaaaaaaa aaaatttt tttttttt 0znnnnnn
+ */
+#define MALLOCX_ARENA_BITS	12
+#define MALLOCX_TCACHE_BITS	12
+#define MALLOCX_LG_ALIGN_BITS	6
+#define MALLOCX_ARENA_SHIFT	20
+#define MALLOCX_TCACHE_SHIFT	8
+#define MALLOCX_ARENA_MASK \
+    (((1 << MALLOCX_ARENA_BITS) - 1) << MALLOCX_ARENA_SHIFT)
+/* NB: Arena index bias decreases the maximum number of arenas by 1. */
+#define MALLOCX_ARENA_LIMIT	((1 << MALLOCX_ARENA_BITS) - 1)
+#define MALLOCX_TCACHE_MASK \
+    (((1 << MALLOCX_TCACHE_BITS) - 1) << MALLOCX_TCACHE_SHIFT)
+#define MALLOCX_TCACHE_MAX	((1 << MALLOCX_TCACHE_BITS) - 3)
+#define MALLOCX_LG_ALIGN_MASK	((1 << MALLOCX_LG_ALIGN_BITS) - 1)
+/* Use MALLOCX_ALIGN_GET() if alignment may not be specified in flags. */
+#define MALLOCX_ALIGN_GET_SPECIFIED(flags)				\
+    (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK))
+#define MALLOCX_ALIGN_GET(flags)					\
+    (MALLOCX_ALIGN_GET_SPECIFIED(flags) & (SIZE_T_MAX-1))
+#define MALLOCX_ZERO_GET(flags)						\
+    ((bool)(flags & MALLOCX_ZERO))
+
+#define MALLOCX_TCACHE_GET(flags)					\
+    (((unsigned)((flags & MALLOCX_TCACHE_MASK) >> MALLOCX_TCACHE_SHIFT)) - 2)
+#define MALLOCX_ARENA_GET(flags)					\
+    (((unsigned)(((unsigned)flags) >> MALLOCX_ARENA_SHIFT)) - 1)
+
+/* Smallest size class to support. */
+#define TINY_MIN		(1U << LG_TINY_MIN)
+
+#define LONG			((size_t)(1U << LG_SIZEOF_LONG))
+#define LONG_MASK		(LONG - 1)
+
+/* Return the smallest long multiple that is >= a. */
+#define LONG_CEILING(a)							\
+	(((a) + LONG_MASK) & ~LONG_MASK)
+
+#define SIZEOF_PTR		(1U << LG_SIZEOF_PTR)
+#define PTR_MASK		(SIZEOF_PTR - 1)
+
+/* Return the smallest (void *) multiple that is >= a. */
+#define PTR_CEILING(a)							\
+	(((a) + PTR_MASK) & ~PTR_MASK)
+
+/*
+ * Maximum size of L1 cache line.  This is used to avoid cache line aliasing.
+ * In addition, this controls the spacing of cacheline-spaced size classes.
+ *
+ * CACHELINE cannot be based on LG_CACHELINE because __declspec(align()) can
+ * only handle raw constants.
+ */
+#define LG_CACHELINE		6
+#define CACHELINE		64
+#define CACHELINE_MASK		(CACHELINE - 1)
+
+/* Return the smallest cacheline multiple that is >= s. */
+#define CACHELINE_CEILING(s)						\
+	(((s) + CACHELINE_MASK) & ~CACHELINE_MASK)
+
+/* Return the nearest aligned address at or below a. */
+#define ALIGNMENT_ADDR2BASE(a, alignment)				\
+	((void *)((uintptr_t)(a) & ((~(alignment)) + 1)))
+
+/* Return the offset between a and the nearest aligned address at or below a. */
+#define ALIGNMENT_ADDR2OFFSET(a, alignment)				\
+	((size_t)((uintptr_t)(a) & (alignment - 1)))
+
+/* Return the smallest alignment multiple that is >= s. */
+#define ALIGNMENT_CEILING(s, alignment)					\
+	(((s) + (alignment - 1)) & ((~(alignment)) + 1))
+
+/* Declare a variable-length array. */
+#if __STDC_VERSION__ < 199901L
+#  ifdef _MSC_VER
+#    include <malloc.h>
+#    define alloca _alloca
+#  else
+#    ifdef JEMALLOC_HAS_ALLOCA_H
+#      include <alloca.h>
+#    else
+#      include <stdlib.h>
+#    endif
+#  endif
+#  define VARIABLE_ARRAY(type, name, count) \
+	type *name = alloca(sizeof(type) * (count))
+#else
+#  define VARIABLE_ARRAY(type, name, count) type name[(count)]
+#endif
+
+#endif /* JEMALLOC_INTERNAL_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
new file mode 100644
index 0000000..3418cbf
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/jemalloc_preamble.h.in
@@ -0,0 +1,213 @@
+#ifndef JEMALLOC_PREAMBLE_H
+#define JEMALLOC_PREAMBLE_H
+
+#include "jemalloc_internal_defs.h"
+#include "jemalloc/internal/jemalloc_internal_decls.h"
+
+#ifdef JEMALLOC_UTRACE
+#include <sys/ktrace.h>
+#endif
+
+#define JEMALLOC_NO_DEMANGLE
+#ifdef JEMALLOC_JET
+#  undef JEMALLOC_IS_MALLOC
+#  define JEMALLOC_N(n) jet_##n
+#  include "jemalloc/internal/public_namespace.h"
+#  define JEMALLOC_NO_RENAME
+#  include "../jemalloc@install_suffix@.h"
+#  undef JEMALLOC_NO_RENAME
+#else
+#  define JEMALLOC_N(n) @private_namespace@##n
+#  include "../jemalloc@install_suffix@.h"
+#endif
+
+#if defined(JEMALLOC_OSATOMIC)
+#include <libkern/OSAtomic.h>
+#endif
+
+#ifdef JEMALLOC_ZONE
+#include <mach/mach_error.h>
+#include <mach/mach_init.h>
+#include <mach/vm_map.h>
+#endif
+
+#include "jemalloc/internal/jemalloc_internal_macros.h"
+
+/*
+ * Note that the ordering matters here; the hook itself is name-mangled.  We
+ * want the inclusion of hooks to happen early, so that we hook as much as
+ * possible.
+ */
+#ifndef JEMALLOC_NO_PRIVATE_NAMESPACE
+#  ifndef JEMALLOC_JET
+#    include "jemalloc/internal/private_namespace.h"
+#  else
+#    include "jemalloc/internal/private_namespace_jet.h"
+#  endif
+#endif
+#include "jemalloc/internal/test_hooks.h"
+
+#ifdef JEMALLOC_DEFINE_MADVISE_FREE
+#  define JEMALLOC_MADV_FREE 8
+#endif
+
+static const bool config_debug =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_dss =
+#ifdef JEMALLOC_DSS
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_madvise_huge =
+#ifdef JEMALLOC_HAVE_MADVISE_HUGE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_fill =
+#ifdef JEMALLOC_FILL
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_lazy_lock =
+#ifdef JEMALLOC_LAZY_LOCK
+    true
+#else
+    false
+#endif
+    ;
+static const char * const config_malloc_conf = JEMALLOC_CONFIG_MALLOC_CONF;
+static const bool config_prof =
+#ifdef JEMALLOC_PROF
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libgcc =
+#ifdef JEMALLOC_PROF_LIBGCC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libunwind =
+#ifdef JEMALLOC_PROF_LIBUNWIND
+    true
+#else
+    false
+#endif
+    ;
+static const bool maps_coalesce =
+#ifdef JEMALLOC_MAPS_COALESCE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_stats =
+#ifdef JEMALLOC_STATS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_tls =
+#ifdef JEMALLOC_TLS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_utrace =
+#ifdef JEMALLOC_UTRACE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_xmalloc =
+#ifdef JEMALLOC_XMALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_cache_oblivious =
+#ifdef JEMALLOC_CACHE_OBLIVIOUS
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Undocumented, for jemalloc development use only at the moment.  See the note
+ * in jemalloc/internal/log.h.
+ */
+static const bool config_log =
+#ifdef JEMALLOC_LOG
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Are extra safety checks enabled; things like checking the size of sized
+ * deallocations, double-frees, etc.
+ */
+static const bool config_opt_safety_checks =
+#ifdef JEMALLOC_OPT_SAFETY_CHECKS
+    true
+#elif defined(JEMALLOC_DEBUG)
+    /*
+     * This lets us only guard safety checks by one flag instead of two; fast
+     * checks can guard solely by config_opt_safety_checks and run in debug mode
+     * too.
+     */
+    true
+#else
+    false
+#endif
+    ;
+
+#if defined(_WIN32) || defined(JEMALLOC_HAVE_SCHED_GETCPU)
+/* Currently percpu_arena depends on sched_getcpu. */
+#define JEMALLOC_PERCPU_ARENA
+#endif
+static const bool have_percpu_arena =
+#ifdef JEMALLOC_PERCPU_ARENA
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Undocumented, and not recommended; the application should take full
+ * responsibility for tracking provenance.
+ */
+static const bool force_ivsalloc =
+#ifdef JEMALLOC_FORCE_IVSALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_background_thread =
+#ifdef JEMALLOC_BACKGROUND_THREAD
+    true
+#else
+    false
+#endif
+    ;
+
+#endif /* JEMALLOC_PREAMBLE_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/large_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/large_externs.h
new file mode 100644
index 0000000..a05019e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/large_externs.h
@@ -0,0 +1,32 @@
+#ifndef JEMALLOC_INTERNAL_LARGE_EXTERNS_H
+#define JEMALLOC_INTERNAL_LARGE_EXTERNS_H
+
+#include "jemalloc/internal/hook.h"
+
+void *large_malloc(tsdn_t *tsdn, arena_t *arena, size_t usize, bool zero);
+void *large_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
+    bool zero);
+bool large_ralloc_no_move(tsdn_t *tsdn, extent_t *extent, size_t usize_min,
+    size_t usize_max, bool zero);
+void *large_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args);
+
+typedef void (large_dalloc_junk_t)(void *, size_t);
+extern large_dalloc_junk_t *JET_MUTABLE large_dalloc_junk;
+
+typedef void (large_dalloc_maybe_junk_t)(void *, size_t);
+extern large_dalloc_maybe_junk_t *JET_MUTABLE large_dalloc_maybe_junk;
+
+void large_dalloc_prep_junked_locked(tsdn_t *tsdn, extent_t *extent);
+void large_dalloc_finish(tsdn_t *tsdn, extent_t *extent);
+void large_dalloc(tsdn_t *tsdn, extent_t *extent);
+size_t large_salloc(tsdn_t *tsdn, const extent_t *extent);
+prof_tctx_t *large_prof_tctx_get(tsdn_t *tsdn, const extent_t *extent);
+void large_prof_tctx_set(tsdn_t *tsdn, extent_t *extent, prof_tctx_t *tctx);
+void large_prof_tctx_reset(tsdn_t *tsdn, extent_t *extent);
+
+nstime_t large_prof_alloc_time_get(const extent_t *extent);
+void large_prof_alloc_time_set(extent_t *extent, nstime_t time);
+
+#endif /* JEMALLOC_INTERNAL_LARGE_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/log.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/log.h
new file mode 100644
index 0000000..6420858
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/log.h
@@ -0,0 +1,115 @@
+#ifndef JEMALLOC_INTERNAL_LOG_H
+#define JEMALLOC_INTERNAL_LOG_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex.h"
+
+#ifdef JEMALLOC_LOG
+#  define JEMALLOC_LOG_VAR_BUFSIZE 1000
+#else
+#  define JEMALLOC_LOG_VAR_BUFSIZE 1
+#endif
+
+#define JEMALLOC_LOG_BUFSIZE 4096
+
+/*
+ * The log malloc_conf option is a '|'-delimited list of log_var name segments
+ * which should be logged.  The names are themselves hierarchical, with '.' as
+ * the delimiter (a "segment" is just a prefix in the log namespace).  So, if
+ * you have:
+ *
+ * log("arena", "log msg for arena"); // 1
+ * log("arena.a", "log msg for arena.a"); // 2
+ * log("arena.b", "log msg for arena.b"); // 3
+ * log("arena.a.a", "log msg for arena.a.a"); // 4
+ * log("extent.a", "log msg for extent.a"); // 5
+ * log("extent.b", "log msg for extent.b"); // 6
+ *
+ * And your malloc_conf option is "log=arena.a|extent", then lines 2, 4, 5, and
+ * 6 will print at runtime.  You can enable logging from all log vars by
+ * writing "log=.".
+ *
+ * None of this should be regarded as a stable API for right now.  It's intended
+ * as a debugging interface, to let us keep around some of our printf-debugging
+ * statements.
+ */
+
+extern char log_var_names[JEMALLOC_LOG_VAR_BUFSIZE];
+extern atomic_b_t log_init_done;
+
+typedef struct log_var_s log_var_t;
+struct log_var_s {
+	/*
+	 * Lowest bit is "inited", second lowest is "enabled".  Putting them in
+	 * a single word lets us avoid any fences on weak architectures.
+	 */
+	atomic_u_t state;
+	const char *name;
+};
+
+#define LOG_NOT_INITIALIZED 0U
+#define LOG_INITIALIZED_NOT_ENABLED 1U
+#define LOG_ENABLED 2U
+
+#define LOG_VAR_INIT(name_str) {ATOMIC_INIT(LOG_NOT_INITIALIZED), name_str}
+
+/*
+ * Returns the value we should assume for state (which is not necessarily
+ * accurate; if logging is done before logging has finished initializing, then
+ * we default to doing the safe thing by logging everything).
+ */
+unsigned log_var_update_state(log_var_t *log_var);
+
+/* We factor out the metadata management to allow us to test more easily. */
+#define log_do_begin(log_var)						\
+if (config_log) {							\
+	unsigned log_state = atomic_load_u(&(log_var).state,		\
+	    ATOMIC_RELAXED);						\
+	if (unlikely(log_state == LOG_NOT_INITIALIZED)) {		\
+		log_state = log_var_update_state(&(log_var));		\
+		assert(log_state != LOG_NOT_INITIALIZED);		\
+	}								\
+	if (log_state == LOG_ENABLED) {					\
+		{
+			/* User code executes here. */
+#define log_do_end(log_var)						\
+		}							\
+	}								\
+}
+
+/*
+ * MSVC has some preprocessor bugs in its expansion of __VA_ARGS__ during
+ * preprocessing.  To work around this, we take all potential extra arguments in
+ * a var-args functions.  Since a varargs macro needs at least one argument in
+ * the "...", we accept the format string there, and require that the first
+ * argument in this "..." is a const char *.
+ */
+static inline void
+log_impl_varargs(const char *name, ...) {
+	char buf[JEMALLOC_LOG_BUFSIZE];
+	va_list ap;
+
+	va_start(ap, name);
+	const char *format = va_arg(ap, const char *);
+	size_t dst_offset = 0;
+	dst_offset += malloc_snprintf(buf, JEMALLOC_LOG_BUFSIZE, "%s: ", name);
+	dst_offset += malloc_vsnprintf(buf + dst_offset,
+	    JEMALLOC_LOG_BUFSIZE - dst_offset, format, ap);
+	dst_offset += malloc_snprintf(buf + dst_offset,
+	    JEMALLOC_LOG_BUFSIZE - dst_offset, "\n");
+	va_end(ap);
+
+	malloc_write(buf);
+}
+
+/* Call as log("log.var.str", "format_string %d", arg_for_format_string); */
+#define LOG(log_var_str, ...)						\
+do {									\
+	static log_var_t log_var = LOG_VAR_INIT(log_var_str);		\
+	log_do_begin(log_var)						\
+		log_impl_varargs((log_var).name, __VA_ARGS__);		\
+	log_do_end(log_var)						\
+} while (0)
+
+#endif /* JEMALLOC_INTERNAL_LOG_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/malloc_io.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/malloc_io.h
new file mode 100644
index 0000000..1d1a414
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/malloc_io.h
@@ -0,0 +1,102 @@
+#ifndef JEMALLOC_INTERNAL_MALLOC_IO_H
+#define JEMALLOC_INTERNAL_MALLOC_IO_H
+
+#ifdef _WIN32
+#  ifdef _WIN64
+#    define FMT64_PREFIX "ll"
+#    define FMTPTR_PREFIX "ll"
+#  else
+#    define FMT64_PREFIX "ll"
+#    define FMTPTR_PREFIX ""
+#  endif
+#  define FMTd32 "d"
+#  define FMTu32 "u"
+#  define FMTx32 "x"
+#  define FMTd64 FMT64_PREFIX "d"
+#  define FMTu64 FMT64_PREFIX "u"
+#  define FMTx64 FMT64_PREFIX "x"
+#  define FMTdPTR FMTPTR_PREFIX "d"
+#  define FMTuPTR FMTPTR_PREFIX "u"
+#  define FMTxPTR FMTPTR_PREFIX "x"
+#else
+#  include <inttypes.h>
+#  define FMTd32 PRId32
+#  define FMTu32 PRIu32
+#  define FMTx32 PRIx32
+#  define FMTd64 PRId64
+#  define FMTu64 PRIu64
+#  define FMTx64 PRIx64
+#  define FMTdPTR PRIdPTR
+#  define FMTuPTR PRIuPTR
+#  define FMTxPTR PRIxPTR
+#endif
+
+/* Size of stack-allocated buffer passed to buferror(). */
+#define BUFERROR_BUF		64
+
+/*
+ * Size of stack-allocated buffer used by malloc_{,v,vc}printf().  This must be
+ * large enough for all possible uses within jemalloc.
+ */
+#define MALLOC_PRINTF_BUFSIZE	4096
+
+int buferror(int err, char *buf, size_t buflen);
+uintmax_t malloc_strtoumax(const char *restrict nptr, char **restrict endptr,
+    int base);
+void malloc_write(const char *s);
+
+/*
+ * malloc_vsnprintf() supports a subset of snprintf(3) that avoids floating
+ * point math.
+ */
+size_t malloc_vsnprintf(char *str, size_t size, const char *format,
+    va_list ap);
+size_t malloc_snprintf(char *str, size_t size, const char *format, ...)
+    JEMALLOC_FORMAT_PRINTF(3, 4);
+/*
+ * The caller can set write_cb to null to choose to print with the
+ * je_malloc_message hook.
+ */
+void malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap);
+void malloc_cprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, ...) JEMALLOC_FORMAT_PRINTF(3, 4);
+void malloc_printf(const char *format, ...) JEMALLOC_FORMAT_PRINTF(1, 2);
+
+static inline ssize_t
+malloc_write_fd(int fd, const void *buf, size_t count) {
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_write)
+	/*
+	 * Use syscall(2) rather than write(2) when possible in order to avoid
+	 * the possibility of memory allocation within libc.  This is necessary
+	 * on FreeBSD; most operating systems do not have this problem though.
+	 *
+	 * syscall() returns long or int, depending on platform, so capture the
+	 * result in the widest plausible type to avoid compiler warnings.
+	 */
+	long result = syscall(SYS_write, fd, buf, count);
+#else
+	ssize_t result = (ssize_t)write(fd, buf,
+#ifdef _WIN32
+	    (unsigned int)
+#endif
+	    count);
+#endif
+	return (ssize_t)result;
+}
+
+static inline ssize_t
+malloc_read_fd(int fd, void *buf, size_t count) {
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_read)
+	long result = syscall(SYS_read, fd, buf, count);
+#else
+	ssize_t result = read(fd, buf,
+#ifdef _WIN32
+	    (unsigned int)
+#endif
+	    count);
+#endif
+	return (ssize_t)result;
+}
+
+#endif /* JEMALLOC_INTERNAL_MALLOC_IO_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex.h
new file mode 100644
index 0000000..18b170b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex.h
@@ -0,0 +1,288 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_H
+#define JEMALLOC_INTERNAL_MUTEX_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/witness.h"
+
+typedef enum {
+	/* Can only acquire one mutex of a given witness rank at a time. */
+	malloc_mutex_rank_exclusive,
+	/*
+	 * Can acquire multiple mutexes of the same witness rank, but in
+	 * address-ascending order only.
+	 */
+	malloc_mutex_address_ordered
+} malloc_mutex_lock_order_t;
+
+typedef struct malloc_mutex_s malloc_mutex_t;
+struct malloc_mutex_s {
+	union {
+		struct {
+			/*
+			 * prof_data is defined first to reduce cacheline
+			 * bouncing: the data is not touched by the mutex holder
+			 * during unlocking, while might be modified by
+			 * contenders.  Having it before the mutex itself could
+			 * avoid prefetching a modified cacheline (for the
+			 * unlocking thread).
+			 */
+			mutex_prof_data_t	prof_data;
+#ifdef _WIN32
+#  if _WIN32_WINNT >= 0x0600
+			SRWLOCK         	lock;
+#  else
+			CRITICAL_SECTION	lock;
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+			os_unfair_lock		lock;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+			pthread_mutex_t		lock;
+			malloc_mutex_t		*postponed_next;
+#else
+			pthread_mutex_t		lock;
+#endif
+			/*
+			 * Hint flag to avoid exclusive cache line contention
+			 * during spin waiting
+			 */
+			atomic_b_t		locked;
+		};
+		/*
+		 * We only touch witness when configured w/ debug.  However we
+		 * keep the field in a union when !debug so that we don't have
+		 * to pollute the code base with #ifdefs, while avoid paying the
+		 * memory cost.
+		 */
+#if !defined(JEMALLOC_DEBUG)
+		witness_t			witness;
+		malloc_mutex_lock_order_t	lock_order;
+#endif
+	};
+
+#if defined(JEMALLOC_DEBUG)
+	witness_t			witness;
+	malloc_mutex_lock_order_t	lock_order;
+#endif
+};
+
+/*
+ * Based on benchmark results, a fixed spin with this amount of retries works
+ * well for our critical sections.
+ */
+#define MALLOC_MUTEX_MAX_SPIN 250
+
+#ifdef _WIN32
+#  if _WIN32_WINNT >= 0x0600
+#    define MALLOC_MUTEX_LOCK(m)    AcquireSRWLockExclusive(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  ReleaseSRWLockExclusive(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!TryAcquireSRWLockExclusive(&(m)->lock))
+#  else
+#    define MALLOC_MUTEX_LOCK(m)    EnterCriticalSection(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  LeaveCriticalSection(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!TryEnterCriticalSection(&(m)->lock))
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#    define MALLOC_MUTEX_LOCK(m)    os_unfair_lock_lock(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  os_unfair_lock_unlock(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!os_unfair_lock_trylock(&(m)->lock))
+#else
+#    define MALLOC_MUTEX_LOCK(m)    pthread_mutex_lock(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  pthread_mutex_unlock(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (pthread_mutex_trylock(&(m)->lock) != 0)
+#endif
+
+#define LOCK_PROF_DATA_INITIALIZER					\
+    {NSTIME_ZERO_INITIALIZER, NSTIME_ZERO_INITIALIZER, 0, 0, 0,		\
+	    ATOMIC_INIT(0), 0, NULL, 0}
+
+#ifdef _WIN32
+#  define MALLOC_MUTEX_INITIALIZER
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#  if defined(JEMALLOC_DEBUG)
+#    define MALLOC_MUTEX_INITIALIZER					\
+  {{{LOCK_PROF_DATA_INITIALIZER, OS_UNFAIR_LOCK_INIT, ATOMIC_INIT(false)}}, \
+         WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT), 0}
+#  else
+#    define MALLOC_MUTEX_INITIALIZER                      \
+  {{{LOCK_PROF_DATA_INITIALIZER, OS_UNFAIR_LOCK_INIT, ATOMIC_INIT(false)}},  \
+      WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#  endif
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+#  if (defined(JEMALLOC_DEBUG))
+#     define MALLOC_MUTEX_INITIALIZER					\
+      {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, NULL, ATOMIC_INIT(false)}},	\
+           WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT), 0}
+#  else
+#     define MALLOC_MUTEX_INITIALIZER					\
+      {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, NULL, ATOMIC_INIT(false)}},	\
+           WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#  endif
+
+#else
+#    define MALLOC_MUTEX_TYPE PTHREAD_MUTEX_DEFAULT
+#  if defined(JEMALLOC_DEBUG)
+#    define MALLOC_MUTEX_INITIALIZER					\
+     {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, ATOMIC_INIT(false)}}, \
+           WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT), 0}
+#  else
+#    define MALLOC_MUTEX_INITIALIZER                          \
+     {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, ATOMIC_INIT(false)}},	\
+      WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#  endif
+#endif
+
+#ifdef JEMALLOC_LAZY_LOCK
+extern bool isthreaded;
+#else
+#  undef isthreaded /* Undo private_namespace.h definition. */
+#  define isthreaded true
+#endif
+
+bool malloc_mutex_init(malloc_mutex_t *mutex, const char *name,
+    witness_rank_t rank, malloc_mutex_lock_order_t lock_order);
+void malloc_mutex_prefork(tsdn_t *tsdn, malloc_mutex_t *mutex);
+void malloc_mutex_postfork_parent(tsdn_t *tsdn, malloc_mutex_t *mutex);
+void malloc_mutex_postfork_child(tsdn_t *tsdn, malloc_mutex_t *mutex);
+bool malloc_mutex_boot(void);
+void malloc_mutex_prof_data_reset(tsdn_t *tsdn, malloc_mutex_t *mutex);
+
+void malloc_mutex_lock_slow(malloc_mutex_t *mutex);
+
+static inline void
+malloc_mutex_lock_final(malloc_mutex_t *mutex) {
+	MALLOC_MUTEX_LOCK(mutex);
+	atomic_store_b(&mutex->locked, true, ATOMIC_RELAXED);
+}
+
+static inline bool
+malloc_mutex_trylock_final(malloc_mutex_t *mutex) {
+	return MALLOC_MUTEX_TRYLOCK(mutex);
+}
+
+static inline void
+mutex_owner_stats_update(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	if (config_stats) {
+		mutex_prof_data_t *data = &mutex->prof_data;
+		data->n_lock_ops++;
+		if (data->prev_owner != tsdn) {
+			data->prev_owner = tsdn;
+			data->n_owner_switches++;
+		}
+	}
+}
+
+/* Trylock: return false if the lock is successfully acquired. */
+static inline bool
+malloc_mutex_trylock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		if (malloc_mutex_trylock_final(mutex)) {
+			atomic_store_b(&mutex->locked, true, ATOMIC_RELAXED);
+			return true;
+		}
+		mutex_owner_stats_update(tsdn, mutex);
+	}
+	witness_lock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+
+	return false;
+}
+
+/* Aggregate lock prof data. */
+static inline void
+malloc_mutex_prof_merge(mutex_prof_data_t *sum, mutex_prof_data_t *data) {
+	nstime_add(&sum->tot_wait_time, &data->tot_wait_time);
+	if (nstime_compare(&sum->max_wait_time, &data->max_wait_time) < 0) {
+		nstime_copy(&sum->max_wait_time, &data->max_wait_time);
+	}
+
+	sum->n_wait_times += data->n_wait_times;
+	sum->n_spin_acquired += data->n_spin_acquired;
+
+	if (sum->max_n_thds < data->max_n_thds) {
+		sum->max_n_thds = data->max_n_thds;
+	}
+	uint32_t cur_n_waiting_thds = atomic_load_u32(&sum->n_waiting_thds,
+	    ATOMIC_RELAXED);
+	uint32_t new_n_waiting_thds = cur_n_waiting_thds + atomic_load_u32(
+	    &data->n_waiting_thds, ATOMIC_RELAXED);
+	atomic_store_u32(&sum->n_waiting_thds, new_n_waiting_thds,
+	    ATOMIC_RELAXED);
+	sum->n_owner_switches += data->n_owner_switches;
+	sum->n_lock_ops += data->n_lock_ops;
+}
+
+static inline void
+malloc_mutex_lock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		if (malloc_mutex_trylock_final(mutex)) {
+			malloc_mutex_lock_slow(mutex);
+			atomic_store_b(&mutex->locked, true, ATOMIC_RELAXED);
+		}
+		mutex_owner_stats_update(tsdn, mutex);
+	}
+	witness_lock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+static inline void
+malloc_mutex_unlock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	atomic_store_b(&mutex->locked, false, ATOMIC_RELAXED);
+	witness_unlock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		MALLOC_MUTEX_UNLOCK(mutex);
+	}
+}
+
+static inline void
+malloc_mutex_assert_owner(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+static inline void
+malloc_mutex_assert_not_owner(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+/* Copy the prof data from mutex for processing. */
+static inline void
+malloc_mutex_prof_read(tsdn_t *tsdn, mutex_prof_data_t *data,
+    malloc_mutex_t *mutex) {
+	mutex_prof_data_t *source = &mutex->prof_data;
+	/* Can only read holding the mutex. */
+	malloc_mutex_assert_owner(tsdn, mutex);
+
+	/*
+	 * Not *really* allowed (we shouldn't be doing non-atomic loads of
+	 * atomic data), but the mutex protection makes this safe, and writing
+	 * a member-for-member copy is tedious for this situation.
+	 */
+	*data = *source;
+	/* n_wait_thds is not reported (modified w/o locking). */
+	atomic_store_u32(&data->n_waiting_thds, 0, ATOMIC_RELAXED);
+}
+
+static inline void
+malloc_mutex_prof_accum(tsdn_t *tsdn, mutex_prof_data_t *data,
+    malloc_mutex_t *mutex) {
+	mutex_prof_data_t *source = &mutex->prof_data;
+	/* Can only read holding the mutex. */
+	malloc_mutex_assert_owner(tsdn, mutex);
+
+	nstime_add(&data->tot_wait_time, &source->tot_wait_time);
+	if (nstime_compare(&source->max_wait_time, &data->max_wait_time) > 0) {
+		nstime_copy(&data->max_wait_time, &source->max_wait_time);
+	}
+	data->n_wait_times += source->n_wait_times;
+	data->n_spin_acquired += source->n_spin_acquired;
+	if (data->max_n_thds < source->max_n_thds) {
+		data->max_n_thds = source->max_n_thds;
+	}
+	/* n_wait_thds is not reported. */
+	atomic_store_u32(&data->n_waiting_thds, 0, ATOMIC_RELAXED);
+	data->n_owner_switches += source->n_owner_switches;
+	data->n_lock_ops += source->n_lock_ops;
+}
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex_pool.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex_pool.h
new file mode 100644
index 0000000..726cece
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex_pool.h
@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_POOL_H
+#define JEMALLOC_INTERNAL_MUTEX_POOL_H
+
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/witness.h"
+
+/* We do mod reductions by this value, so it should be kept a power of 2. */
+#define MUTEX_POOL_SIZE 256
+
+typedef struct mutex_pool_s mutex_pool_t;
+struct mutex_pool_s {
+	malloc_mutex_t mutexes[MUTEX_POOL_SIZE];
+};
+
+bool mutex_pool_init(mutex_pool_t *pool, const char *name, witness_rank_t rank);
+
+/* Internal helper - not meant to be called outside this module. */
+static inline malloc_mutex_t *
+mutex_pool_mutex(mutex_pool_t *pool, uintptr_t key) {
+	size_t hash_result[2];
+	hash(&key, sizeof(key), 0xd50dcc1b, hash_result);
+	return &pool->mutexes[hash_result[0] % MUTEX_POOL_SIZE];
+}
+
+static inline void
+mutex_pool_assert_not_held(tsdn_t *tsdn, mutex_pool_t *pool) {
+	for (int i = 0; i < MUTEX_POOL_SIZE; i++) {
+		malloc_mutex_assert_not_owner(tsdn, &pool->mutexes[i]);
+	}
+}
+
+/*
+ * Note that a mutex pool doesn't work exactly the way an embdedded mutex would.
+ * You're not allowed to acquire mutexes in the pool one at a time.  You have to
+ * acquire all the mutexes you'll need in a single function call, and then
+ * release them all in a single function call.
+ */
+
+static inline void
+mutex_pool_lock(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	mutex_pool_assert_not_held(tsdn, pool);
+
+	malloc_mutex_t *mutex = mutex_pool_mutex(pool, key);
+	malloc_mutex_lock(tsdn, mutex);
+}
+
+static inline void
+mutex_pool_unlock(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	malloc_mutex_t *mutex = mutex_pool_mutex(pool, key);
+	malloc_mutex_unlock(tsdn, mutex);
+
+	mutex_pool_assert_not_held(tsdn, pool);
+}
+
+static inline void
+mutex_pool_lock2(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key1,
+    uintptr_t key2) {
+	mutex_pool_assert_not_held(tsdn, pool);
+
+	malloc_mutex_t *mutex1 = mutex_pool_mutex(pool, key1);
+	malloc_mutex_t *mutex2 = mutex_pool_mutex(pool, key2);
+	if ((uintptr_t)mutex1 < (uintptr_t)mutex2) {
+		malloc_mutex_lock(tsdn, mutex1);
+		malloc_mutex_lock(tsdn, mutex2);
+	} else if ((uintptr_t)mutex1 == (uintptr_t)mutex2) {
+		malloc_mutex_lock(tsdn, mutex1);
+	} else {
+		malloc_mutex_lock(tsdn, mutex2);
+		malloc_mutex_lock(tsdn, mutex1);
+	}
+}
+
+static inline void
+mutex_pool_unlock2(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key1,
+    uintptr_t key2) {
+	malloc_mutex_t *mutex1 = mutex_pool_mutex(pool, key1);
+	malloc_mutex_t *mutex2 = mutex_pool_mutex(pool, key2);
+	if (mutex1 == mutex2) {
+		malloc_mutex_unlock(tsdn, mutex1);
+	} else {
+		malloc_mutex_unlock(tsdn, mutex1);
+		malloc_mutex_unlock(tsdn, mutex2);
+	}
+
+	mutex_pool_assert_not_held(tsdn, pool);
+}
+
+static inline void
+mutex_pool_assert_owner(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	malloc_mutex_assert_owner(tsdn, mutex_pool_mutex(pool, key));
+}
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_POOL_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex_prof.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex_prof.h
new file mode 100644
index 0000000..2cb8fb0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/mutex_prof.h
@@ -0,0 +1,108 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_PROF_H
+#define JEMALLOC_INTERNAL_MUTEX_PROF_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/tsd_types.h"
+
+#define MUTEX_PROF_GLOBAL_MUTEXES					\
+    OP(background_thread)						\
+    OP(ctl)								\
+    OP(prof)
+
+typedef enum {
+#define OP(mtx) global_prof_mutex_##mtx,
+	MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+	mutex_prof_num_global_mutexes
+} mutex_prof_global_ind_t;
+
+#define MUTEX_PROF_ARENA_MUTEXES					\
+    OP(large)								\
+    OP(extent_avail)							\
+    OP(extents_dirty)							\
+    OP(extents_muzzy)							\
+    OP(extents_retained)						\
+    OP(decay_dirty)							\
+    OP(decay_muzzy)							\
+    OP(base)								\
+    OP(tcache_list)
+
+typedef enum {
+#define OP(mtx) arena_prof_mutex_##mtx,
+	MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+	mutex_prof_num_arena_mutexes
+} mutex_prof_arena_ind_t;
+
+/*
+ * The forth parameter is a boolean value that is true for derived rate counters
+ * and false for real ones.
+ */
+#define MUTEX_PROF_UINT64_COUNTERS					\
+    OP(num_ops, uint64_t, "n_lock_ops", false, num_ops)					\
+    OP(num_ops_ps, uint64_t, "(#/sec)", true, num_ops)				\
+    OP(num_wait, uint64_t, "n_waiting", false, num_wait)				\
+    OP(num_wait_ps, uint64_t, "(#/sec)", true, num_wait)				\
+    OP(num_spin_acq, uint64_t, "n_spin_acq", false, num_spin_acq)			\
+    OP(num_spin_acq_ps, uint64_t, "(#/sec)", true, num_spin_acq)			\
+    OP(num_owner_switch, uint64_t, "n_owner_switch", false, num_owner_switch)		\
+    OP(num_owner_switch_ps, uint64_t, "(#/sec)", true, num_owner_switch)	\
+    OP(total_wait_time, uint64_t, "total_wait_ns", false, total_wait_time)		\
+    OP(total_wait_time_ps, uint64_t, "(#/sec)", true, total_wait_time)		\
+    OP(max_wait_time, uint64_t, "max_wait_ns", false, max_wait_time)
+
+#define MUTEX_PROF_UINT32_COUNTERS					\
+    OP(max_num_thds, uint32_t, "max_n_thds", false, max_num_thds)
+
+#define MUTEX_PROF_COUNTERS						\
+		MUTEX_PROF_UINT64_COUNTERS				\
+		MUTEX_PROF_UINT32_COUNTERS
+
+#define OP(counter, type, human, derived, base_counter) mutex_counter_##counter,
+
+#define COUNTER_ENUM(counter_list, t)					\
+		typedef enum {						\
+			counter_list					\
+			mutex_prof_num_##t##_counters			\
+		} mutex_prof_##t##_counter_ind_t;
+
+COUNTER_ENUM(MUTEX_PROF_UINT64_COUNTERS, uint64_t)
+COUNTER_ENUM(MUTEX_PROF_UINT32_COUNTERS, uint32_t)
+
+#undef COUNTER_ENUM
+#undef OP
+
+typedef struct {
+	/*
+	 * Counters touched on the slow path, i.e. when there is lock
+	 * contention.  We update them once we have the lock.
+	 */
+	/* Total time (in nano seconds) spent waiting on this mutex. */
+	nstime_t		tot_wait_time;
+	/* Max time (in nano seconds) spent on a single lock operation. */
+	nstime_t		max_wait_time;
+	/* # of times have to wait for this mutex (after spinning). */
+	uint64_t		n_wait_times;
+	/* # of times acquired the mutex through local spinning. */
+	uint64_t		n_spin_acquired;
+	/* Max # of threads waiting for the mutex at the same time. */
+	uint32_t		max_n_thds;
+	/* Current # of threads waiting on the lock.  Atomic synced. */
+	atomic_u32_t		n_waiting_thds;
+
+	/*
+	 * Data touched on the fast path.  These are modified right after we
+	 * grab the lock, so it's placed closest to the end (i.e. right before
+	 * the lock) so that we have a higher chance of them being on the same
+	 * cacheline.
+	 */
+	/* # of times the mutex holder is different than the previous one. */
+	uint64_t		n_owner_switches;
+	/* Previous mutex holder, to facilitate n_owner_switches. */
+	tsdn_t			*prev_owner;
+	/* # of lock() operations in total. */
+	uint64_t		n_lock_ops;
+} mutex_prof_data_t;
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_PROF_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/nstime.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/nstime.h
new file mode 100644
index 0000000..17c177c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/nstime.h
@@ -0,0 +1,34 @@
+#ifndef JEMALLOC_INTERNAL_NSTIME_H
+#define JEMALLOC_INTERNAL_NSTIME_H
+
+/* Maximum supported number of seconds (~584 years). */
+#define NSTIME_SEC_MAX KQU(18446744072)
+#define NSTIME_ZERO_INITIALIZER {0}
+
+typedef struct {
+	uint64_t ns;
+} nstime_t;
+
+void nstime_init(nstime_t *time, uint64_t ns);
+void nstime_init2(nstime_t *time, uint64_t sec, uint64_t nsec);
+uint64_t nstime_ns(const nstime_t *time);
+uint64_t nstime_sec(const nstime_t *time);
+uint64_t nstime_msec(const nstime_t *time);
+uint64_t nstime_nsec(const nstime_t *time);
+void nstime_copy(nstime_t *time, const nstime_t *source);
+int nstime_compare(const nstime_t *a, const nstime_t *b);
+void nstime_add(nstime_t *time, const nstime_t *addend);
+void nstime_iadd(nstime_t *time, uint64_t addend);
+void nstime_subtract(nstime_t *time, const nstime_t *subtrahend);
+void nstime_isubtract(nstime_t *time, uint64_t subtrahend);
+void nstime_imultiply(nstime_t *time, uint64_t multiplier);
+void nstime_idivide(nstime_t *time, uint64_t divisor);
+uint64_t nstime_divide(const nstime_t *time, const nstime_t *divisor);
+
+typedef bool (nstime_monotonic_t)(void);
+extern nstime_monotonic_t *JET_MUTABLE nstime_monotonic;
+
+typedef bool (nstime_update_t)(nstime_t *);
+extern nstime_update_t *JET_MUTABLE nstime_update;
+
+#endif /* JEMALLOC_INTERNAL_NSTIME_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/pages.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/pages.h
new file mode 100644
index 0000000..7dae633
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/pages.h
@@ -0,0 +1,88 @@
+#ifndef JEMALLOC_INTERNAL_PAGES_EXTERNS_H
+#define JEMALLOC_INTERNAL_PAGES_EXTERNS_H
+
+/* Page size.  LG_PAGE is determined by the configure script. */
+#ifdef PAGE_MASK
+#  undef PAGE_MASK
+#endif
+#define PAGE		((size_t)(1U << LG_PAGE))
+#define PAGE_MASK	((size_t)(PAGE - 1))
+/* Return the page base address for the page containing address a. */
+#define PAGE_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~PAGE_MASK))
+/* Return the smallest pagesize multiple that is >= s. */
+#define PAGE_CEILING(s)							\
+	(((s) + PAGE_MASK) & ~PAGE_MASK)
+
+/* Huge page size.  LG_HUGEPAGE is determined by the configure script. */
+#define HUGEPAGE	((size_t)(1U << LG_HUGEPAGE))
+#define HUGEPAGE_MASK	((size_t)(HUGEPAGE - 1))
+/* Return the huge page base address for the huge page containing address a. */
+#define HUGEPAGE_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~HUGEPAGE_MASK))
+/* Return the smallest pagesize multiple that is >= s. */
+#define HUGEPAGE_CEILING(s)						\
+	(((s) + HUGEPAGE_MASK) & ~HUGEPAGE_MASK)
+
+/* PAGES_CAN_PURGE_LAZY is defined if lazy purging is supported. */
+#if defined(_WIN32) || defined(JEMALLOC_PURGE_MADVISE_FREE)
+#  define PAGES_CAN_PURGE_LAZY
+#endif
+/*
+ * PAGES_CAN_PURGE_FORCED is defined if forced purging is supported.
+ *
+ * The only supported way to hard-purge on Windows is to decommit and then
+ * re-commit, but doing so is racy, and if re-commit fails it's a pain to
+ * propagate the "poisoned" memory state.  Since we typically decommit as the
+ * next step after purging on Windows anyway, there's no point in adding such
+ * complexity.
+ */
+#if !defined(_WIN32) && ((defined(JEMALLOC_PURGE_MADVISE_DONTNEED) && \
+    defined(JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS)) || \
+    defined(JEMALLOC_MAPS_COALESCE))
+#  define PAGES_CAN_PURGE_FORCED
+#endif
+
+static const bool pages_can_purge_lazy =
+#ifdef PAGES_CAN_PURGE_LAZY
+    true
+#else
+    false
+#endif
+    ;
+static const bool pages_can_purge_forced =
+#ifdef PAGES_CAN_PURGE_FORCED
+    true
+#else
+    false
+#endif
+    ;
+
+typedef enum {
+	thp_mode_default       = 0, /* Do not change hugepage settings. */
+	thp_mode_always        = 1, /* Always set MADV_HUGEPAGE. */
+	thp_mode_never         = 2, /* Always set MADV_NOHUGEPAGE. */
+
+	thp_mode_names_limit   = 3, /* Used for option processing. */
+	thp_mode_not_supported = 3  /* No THP support detected. */
+} thp_mode_t;
+
+#define THP_MODE_DEFAULT thp_mode_default
+extern thp_mode_t opt_thp;
+extern thp_mode_t init_system_thp_mode; /* Initial system wide state. */
+extern const char *thp_mode_names[];
+
+void *pages_map(void *addr, size_t size, size_t alignment, bool *commit);
+void pages_unmap(void *addr, size_t size);
+bool pages_commit(void *addr, size_t size);
+bool pages_decommit(void *addr, size_t size);
+bool pages_purge_lazy(void *addr, size_t size);
+bool pages_purge_forced(void *addr, size_t size);
+bool pages_huge(void *addr, size_t size);
+bool pages_nohuge(void *addr, size_t size);
+bool pages_dontdump(void *addr, size_t size);
+bool pages_dodump(void *addr, size_t size);
+bool pages_boot(void);
+void pages_set_thp_state (void *ptr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_PAGES_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/ph.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ph.h
new file mode 100644
index 0000000..84d6778
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ph.h
@@ -0,0 +1,391 @@
+/*
+ * A Pairing Heap implementation.
+ *
+ * "The Pairing Heap: A New Form of Self-Adjusting Heap"
+ * https://www.cs.cmu.edu/~sleator/papers/pairing-heaps.pdf
+ *
+ * With auxiliary twopass list, described in a follow on paper.
+ *
+ * "Pairing Heaps: Experiments and Analysis"
+ * http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.106.2988&rep=rep1&type=pdf
+ *
+ *******************************************************************************
+ */
+
+#ifndef PH_H_
+#define PH_H_
+
+/* Node structure. */
+#define phn(a_type)							\
+struct {								\
+	a_type	*phn_prev;						\
+	a_type	*phn_next;						\
+	a_type	*phn_lchild;						\
+}
+
+/* Root structure. */
+#define ph(a_type)							\
+struct {								\
+	a_type	*ph_root;						\
+}
+
+/* Internal utility macros. */
+#define phn_lchild_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_lchild)
+#define phn_lchild_set(a_type, a_field, a_phn, a_lchild) do {		\
+	a_phn->a_field.phn_lchild = a_lchild;				\
+} while (0)
+
+#define phn_next_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_next)
+#define phn_prev_set(a_type, a_field, a_phn, a_prev) do {		\
+	a_phn->a_field.phn_prev = a_prev;				\
+} while (0)
+
+#define phn_prev_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_prev)
+#define phn_next_set(a_type, a_field, a_phn, a_next) do {		\
+	a_phn->a_field.phn_next = a_next;				\
+} while (0)
+
+#define phn_merge_ordered(a_type, a_field, a_phn0, a_phn1, a_cmp) do {	\
+	a_type *phn0child;						\
+									\
+	assert(a_phn0 != NULL);						\
+	assert(a_phn1 != NULL);						\
+	assert(a_cmp(a_phn0, a_phn1) <= 0);				\
+									\
+	phn_prev_set(a_type, a_field, a_phn1, a_phn0);			\
+	phn0child = phn_lchild_get(a_type, a_field, a_phn0);		\
+	phn_next_set(a_type, a_field, a_phn1, phn0child);		\
+	if (phn0child != NULL) {					\
+		phn_prev_set(a_type, a_field, phn0child, a_phn1);	\
+	}								\
+	phn_lchild_set(a_type, a_field, a_phn0, a_phn1);		\
+} while (0)
+
+#define phn_merge(a_type, a_field, a_phn0, a_phn1, a_cmp, r_phn) do {	\
+	if (a_phn0 == NULL) {						\
+		r_phn = a_phn1;						\
+	} else if (a_phn1 == NULL) {					\
+		r_phn = a_phn0;						\
+	} else if (a_cmp(a_phn0, a_phn1) < 0) {				\
+		phn_merge_ordered(a_type, a_field, a_phn0, a_phn1,	\
+		    a_cmp);						\
+		r_phn = a_phn0;						\
+	} else {							\
+		phn_merge_ordered(a_type, a_field, a_phn1, a_phn0,	\
+		    a_cmp);						\
+		r_phn = a_phn1;						\
+	}								\
+} while (0)
+
+#define ph_merge_siblings(a_type, a_field, a_phn, a_cmp, r_phn) do {	\
+	a_type *head = NULL;						\
+	a_type *tail = NULL;						\
+	a_type *phn0 = a_phn;						\
+	a_type *phn1 = phn_next_get(a_type, a_field, phn0);		\
+									\
+	/*								\
+	 * Multipass merge, wherein the first two elements of a FIFO	\
+	 * are repeatedly merged, and each result is appended to the	\
+	 * singly linked FIFO, until the FIFO contains only a single	\
+	 * element.  We start with a sibling list but no reference to	\
+	 * its tail, so we do a single pass over the sibling list to	\
+	 * populate the FIFO.						\
+	 */								\
+	if (phn1 != NULL) {						\
+		a_type *phnrest = phn_next_get(a_type, a_field, phn1);	\
+		if (phnrest != NULL) {					\
+			phn_prev_set(a_type, a_field, phnrest, NULL);	\
+		}							\
+		phn_prev_set(a_type, a_field, phn0, NULL);		\
+		phn_next_set(a_type, a_field, phn0, NULL);		\
+		phn_prev_set(a_type, a_field, phn1, NULL);		\
+		phn_next_set(a_type, a_field, phn1, NULL);		\
+		phn_merge(a_type, a_field, phn0, phn1, a_cmp, phn0);	\
+		head = tail = phn0;					\
+		phn0 = phnrest;						\
+		while (phn0 != NULL) {					\
+			phn1 = phn_next_get(a_type, a_field, phn0);	\
+			if (phn1 != NULL) {				\
+				phnrest = phn_next_get(a_type, a_field,	\
+				    phn1);				\
+				if (phnrest != NULL) {			\
+					phn_prev_set(a_type, a_field,	\
+					    phnrest, NULL);		\
+				}					\
+				phn_prev_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				phn_next_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				phn_prev_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_next_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_merge(a_type, a_field, phn0, phn1,	\
+				    a_cmp, phn0);			\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = phnrest;				\
+			} else {					\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = NULL;				\
+			}						\
+		}							\
+		phn0 = head;						\
+		phn1 = phn_next_get(a_type, a_field, phn0);		\
+		if (phn1 != NULL) {					\
+			while (true) {					\
+				head = phn_next_get(a_type, a_field,	\
+				    phn1);				\
+				assert(phn_prev_get(a_type, a_field,	\
+				    phn0) == NULL);			\
+				phn_next_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				assert(phn_prev_get(a_type, a_field,	\
+				    phn1) == NULL);			\
+				phn_next_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_merge(a_type, a_field, phn0, phn1,	\
+				    a_cmp, phn0);			\
+				if (head == NULL) {			\
+					break;				\
+				}					\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = head;				\
+				phn1 = phn_next_get(a_type, a_field,	\
+				    phn0);				\
+			}						\
+		}							\
+	}								\
+	r_phn = phn0;							\
+} while (0)
+
+#define ph_merge_aux(a_type, a_field, a_ph, a_cmp) do {			\
+	a_type *phn = phn_next_get(a_type, a_field, a_ph->ph_root);	\
+	if (phn != NULL) {						\
+		phn_prev_set(a_type, a_field, a_ph->ph_root, NULL);	\
+		phn_next_set(a_type, a_field, a_ph->ph_root, NULL);	\
+		phn_prev_set(a_type, a_field, phn, NULL);		\
+		ph_merge_siblings(a_type, a_field, phn, a_cmp, phn);	\
+		assert(phn_next_get(a_type, a_field, phn) == NULL);	\
+		phn_merge(a_type, a_field, a_ph->ph_root, phn, a_cmp,	\
+		    a_ph->ph_root);					\
+	}								\
+} while (0)
+
+#define ph_merge_children(a_type, a_field, a_phn, a_cmp, r_phn) do {	\
+	a_type *lchild = phn_lchild_get(a_type, a_field, a_phn);	\
+	if (lchild == NULL) {						\
+		r_phn = NULL;						\
+	} else {							\
+		ph_merge_siblings(a_type, a_field, lchild, a_cmp,	\
+		    r_phn);						\
+	}								\
+} while (0)
+
+/*
+ * The ph_proto() macro generates function prototypes that correspond to the
+ * functions generated by an equivalently parameterized call to ph_gen().
+ */
+#define ph_proto(a_attr, a_prefix, a_ph_type, a_type)			\
+a_attr void	a_prefix##new(a_ph_type *ph);				\
+a_attr bool	a_prefix##empty(a_ph_type *ph);				\
+a_attr a_type	*a_prefix##first(a_ph_type *ph);			\
+a_attr a_type	*a_prefix##any(a_ph_type *ph);				\
+a_attr void	a_prefix##insert(a_ph_type *ph, a_type *phn);		\
+a_attr a_type	*a_prefix##remove_first(a_ph_type *ph);			\
+a_attr a_type	*a_prefix##remove_any(a_ph_type *ph);			\
+a_attr void	a_prefix##remove(a_ph_type *ph, a_type *phn);
+
+/*
+ * The ph_gen() macro generates a type-specific pairing heap implementation,
+ * based on the above cpp macros.
+ */
+#define ph_gen(a_attr, a_prefix, a_ph_type, a_type, a_field, a_cmp)	\
+a_attr void								\
+a_prefix##new(a_ph_type *ph) {						\
+	memset(ph, 0, sizeof(ph(a_type)));				\
+}									\
+a_attr bool								\
+a_prefix##empty(a_ph_type *ph) {					\
+	return (ph->ph_root == NULL);					\
+}									\
+a_attr a_type *								\
+a_prefix##first(a_ph_type *ph) {					\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	ph_merge_aux(a_type, a_field, ph, a_cmp);			\
+	return ph->ph_root;						\
+}									\
+a_attr a_type *								\
+a_prefix##any(a_ph_type *ph) {						\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	a_type *aux = phn_next_get(a_type, a_field, ph->ph_root);	\
+	if (aux != NULL) {						\
+		return aux;						\
+	}								\
+	return ph->ph_root;						\
+}									\
+a_attr void								\
+a_prefix##insert(a_ph_type *ph, a_type *phn) {				\
+	memset(&phn->a_field, 0, sizeof(phn(a_type)));			\
+									\
+	/*								\
+	 * Treat the root as an aux list during insertion, and lazily	\
+	 * merge during a_prefix##remove_first().  For elements that	\
+	 * are inserted, then removed via a_prefix##remove() before the	\
+	 * aux list is ever processed, this makes insert/remove		\
+	 * constant-time, whereas eager merging would make insert	\
+	 * O(log n).							\
+	 */								\
+	if (ph->ph_root == NULL) {					\
+		ph->ph_root = phn;					\
+	} else {							\
+		phn_next_set(a_type, a_field, phn, phn_next_get(a_type,	\
+		    a_field, ph->ph_root));				\
+		if (phn_next_get(a_type, a_field, ph->ph_root) !=	\
+		    NULL) {						\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, ph->ph_root),	\
+			    phn);					\
+		}							\
+		phn_prev_set(a_type, a_field, phn, ph->ph_root);	\
+		phn_next_set(a_type, a_field, ph->ph_root, phn);	\
+	}								\
+}									\
+a_attr a_type *								\
+a_prefix##remove_first(a_ph_type *ph) {					\
+	a_type *ret;							\
+									\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	ph_merge_aux(a_type, a_field, ph, a_cmp);			\
+									\
+	ret = ph->ph_root;						\
+									\
+	ph_merge_children(a_type, a_field, ph->ph_root, a_cmp,		\
+	    ph->ph_root);						\
+									\
+	return ret;							\
+}									\
+a_attr a_type *								\
+a_prefix##remove_any(a_ph_type *ph) {					\
+	/*								\
+	 * Remove the most recently inserted aux list element, or the	\
+	 * root if the aux list is empty.  This has the effect of	\
+	 * behaving as a LIFO (and insertion/removal is therefore	\
+	 * constant-time) if a_prefix##[remove_]first() are never	\
+	 * called.							\
+	 */								\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	a_type *ret = phn_next_get(a_type, a_field, ph->ph_root);	\
+	if (ret != NULL) {						\
+		a_type *aux = phn_next_get(a_type, a_field, ret);	\
+		phn_next_set(a_type, a_field, ph->ph_root, aux);	\
+		if (aux != NULL) {					\
+			phn_prev_set(a_type, a_field, aux,		\
+			    ph->ph_root);				\
+		}							\
+		return ret;						\
+	}								\
+	ret = ph->ph_root;						\
+	ph_merge_children(a_type, a_field, ph->ph_root, a_cmp,		\
+	    ph->ph_root);						\
+	return ret;							\
+}									\
+a_attr void								\
+a_prefix##remove(a_ph_type *ph, a_type *phn) {				\
+	a_type *replace, *parent;					\
+									\
+	if (ph->ph_root == phn) {					\
+		/*							\
+		 * We can delete from aux list without merging it, but	\
+		 * we need to merge if we are dealing with the root	\
+		 * node and it has children.				\
+		 */							\
+		if (phn_lchild_get(a_type, a_field, phn) == NULL) {	\
+			ph->ph_root = phn_next_get(a_type, a_field,	\
+			    phn);					\
+			if (ph->ph_root != NULL) {			\
+				phn_prev_set(a_type, a_field,		\
+				    ph->ph_root, NULL);			\
+			}						\
+			return;						\
+		}							\
+		ph_merge_aux(a_type, a_field, ph, a_cmp);		\
+		if (ph->ph_root == phn) {				\
+			ph_merge_children(a_type, a_field, ph->ph_root,	\
+			    a_cmp, ph->ph_root);			\
+			return;						\
+		}							\
+	}								\
+									\
+	/* Get parent (if phn is leftmost child) before mutating. */	\
+	if ((parent = phn_prev_get(a_type, a_field, phn)) != NULL) {	\
+		if (phn_lchild_get(a_type, a_field, parent) != phn) {	\
+			parent = NULL;					\
+		}							\
+	}								\
+	/* Find a possible replacement node, and link to parent. */	\
+	ph_merge_children(a_type, a_field, phn, a_cmp, replace);	\
+	/* Set next/prev for sibling linked list. */			\
+	if (replace != NULL) {						\
+		if (parent != NULL) {					\
+			phn_prev_set(a_type, a_field, replace, parent);	\
+			phn_lchild_set(a_type, a_field, parent,		\
+			    replace);					\
+		} else {						\
+			phn_prev_set(a_type, a_field, replace,		\
+			    phn_prev_get(a_type, a_field, phn));	\
+			if (phn_prev_get(a_type, a_field, phn) !=	\
+			    NULL) {					\
+				phn_next_set(a_type, a_field,		\
+				    phn_prev_get(a_type, a_field, phn),	\
+				    replace);				\
+			}						\
+		}							\
+		phn_next_set(a_type, a_field, replace,			\
+		    phn_next_get(a_type, a_field, phn));		\
+		if (phn_next_get(a_type, a_field, phn) != NULL) {	\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, phn),		\
+			    replace);					\
+		}							\
+	} else {							\
+		if (parent != NULL) {					\
+			a_type *next = phn_next_get(a_type, a_field,	\
+			    phn);					\
+			phn_lchild_set(a_type, a_field, parent, next);	\
+			if (next != NULL) {				\
+				phn_prev_set(a_type, a_field, next,	\
+				    parent);				\
+			}						\
+		} else {						\
+			assert(phn_prev_get(a_type, a_field, phn) !=	\
+			    NULL);					\
+			phn_next_set(a_type, a_field,			\
+			    phn_prev_get(a_type, a_field, phn),		\
+			    phn_next_get(a_type, a_field, phn));	\
+		}							\
+		if (phn_next_get(a_type, a_field, phn) != NULL) {	\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, phn),		\
+			    phn_prev_get(a_type, a_field, phn));	\
+		}							\
+	}								\
+}
+
+#endif /* PH_H_ */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/private_namespace.sh b/memkind-1.10.0/jemalloc/include/jemalloc/internal/private_namespace.sh
new file mode 100755
index 0000000..6ef1346
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/private_namespace.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+for symbol in `cat "$@"` ; do
+  echo "#define ${symbol} JEMALLOC_N(${symbol})"
+done
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/private_symbols.sh b/memkind-1.10.0/jemalloc/include/jemalloc/internal/private_symbols.sh
new file mode 100755
index 0000000..442a259
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/private_symbols.sh
@@ -0,0 +1,51 @@
+#!/bin/sh
+#
+# Generate private_symbols[_jet].awk.
+#
+# Usage: private_symbols.sh <sym_prefix> <sym>*
+#
+# <sym_prefix> is typically "" or "_".
+
+sym_prefix=$1
+shift
+
+cat <<EOF
+#!/usr/bin/env awk -f
+
+BEGIN {
+  sym_prefix = "${sym_prefix}"
+  split("\\
+EOF
+
+for public_sym in "$@" ; do
+  cat <<EOF
+        ${sym_prefix}${public_sym} \\
+EOF
+done
+
+cat <<"EOF"
+        ", exported_symbol_names)
+  # Store exported symbol names as keys in exported_symbols.
+  for (i in exported_symbol_names) {
+    exported_symbols[exported_symbol_names[i]] = 1
+  }
+}
+
+# Process 'nm -a <c_source.o>' output.
+#
+# Handle lines like:
+#   0000000000000008 D opt_junk
+#   0000000000007574 T malloc_initialized
+(NF == 3 && $2 ~ /^[ABCDGRSTVW]$/ && !($3 in exported_symbols) && $3 ~ /^[A-Za-z0-9_]+$/) {
+  print substr($3, 1+length(sym_prefix), length($3)-length(sym_prefix))
+}
+
+# Process 'dumpbin /SYMBOLS <c_source.obj>' output.
+#
+# Handle lines like:
+#   353 00008098 SECT4  notype       External     | opt_junk
+#   3F1 00000000 SECT7  notype ()    External     | malloc_initialized
+($3 ~ /^SECT[0-9]+/ && $(NF-2) == "External" && !($NF in exported_symbols)) {
+  print $NF
+}
+EOF
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/prng.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prng.h
new file mode 100644
index 0000000..15cc2d1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prng.h
@@ -0,0 +1,185 @@
+#ifndef JEMALLOC_INTERNAL_PRNG_H
+#define JEMALLOC_INTERNAL_PRNG_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
+
+/*
+ * Simple linear congruential pseudo-random number generator:
+ *
+ *   prng(y) = (a*x + c) % m
+ *
+ * where the following constants ensure maximal period:
+ *
+ *   a == Odd number (relatively prime to 2^n), and (a-1) is a multiple of 4.
+ *   c == Odd number (relatively prime to 2^n).
+ *   m == 2^32
+ *
+ * See Knuth's TAOCP 3rd Ed., Vol. 2, pg. 17 for details on these constraints.
+ *
+ * This choice of m has the disadvantage that the quality of the bits is
+ * proportional to bit position.  For example, the lowest bit has a cycle of 2,
+ * the next has a cycle of 4, etc.  For this reason, we prefer to use the upper
+ * bits.
+ */
+
+/******************************************************************************/
+/* INTERNAL DEFINITIONS -- IGNORE */
+/******************************************************************************/
+#define PRNG_A_32	UINT32_C(1103515241)
+#define PRNG_C_32	UINT32_C(12347)
+
+#define PRNG_A_64	UINT64_C(6364136223846793005)
+#define PRNG_C_64	UINT64_C(1442695040888963407)
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_state_next_u32(uint32_t state) {
+	return (state * PRNG_A_32) + PRNG_C_32;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_state_next_u64(uint64_t state) {
+	return (state * PRNG_A_64) + PRNG_C_64;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_state_next_zu(size_t state) {
+#if LG_SIZEOF_PTR == 2
+	return (state * PRNG_A_32) + PRNG_C_32;
+#elif LG_SIZEOF_PTR == 3
+	return (state * PRNG_A_64) + PRNG_C_64;
+#else
+#error Unsupported pointer size
+#endif
+}
+
+/******************************************************************************/
+/* BEGIN PUBLIC API */
+/******************************************************************************/
+
+/*
+ * The prng_lg_range functions give a uniform int in the half-open range [0,
+ * 2**lg_range).  If atomic is true, they do so safely from multiple threads.
+ * Multithreaded 64-bit prngs aren't supported.
+ */
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_lg_range_u32(atomic_u32_t *state, unsigned lg_range, bool atomic) {
+	uint32_t ret, state0, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= 32);
+
+	state0 = atomic_load_u32(state, ATOMIC_RELAXED);
+
+	if (atomic) {
+		do {
+			state1 = prng_state_next_u32(state0);
+		} while (!atomic_compare_exchange_weak_u32(state, &state0,
+		    state1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+	} else {
+		state1 = prng_state_next_u32(state0);
+		atomic_store_u32(state, state1, ATOMIC_RELAXED);
+	}
+	ret = state1 >> (32 - lg_range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_lg_range_u64(uint64_t *state, unsigned lg_range) {
+	uint64_t ret, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= 64);
+
+	state1 = prng_state_next_u64(*state);
+	*state = state1;
+	ret = state1 >> (64 - lg_range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_lg_range_zu(atomic_zu_t *state, unsigned lg_range, bool atomic) {
+	size_t ret, state0, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= ZU(1) << (3 + LG_SIZEOF_PTR));
+
+	state0 = atomic_load_zu(state, ATOMIC_RELAXED);
+
+	if (atomic) {
+		do {
+			state1 = prng_state_next_zu(state0);
+		} while (atomic_compare_exchange_weak_zu(state, &state0,
+		    state1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+	} else {
+		state1 = prng_state_next_zu(state0);
+		atomic_store_zu(state, state1, ATOMIC_RELAXED);
+	}
+	ret = state1 >> ((ZU(1) << (3 + LG_SIZEOF_PTR)) - lg_range);
+
+	return ret;
+}
+
+/*
+ * The prng_range functions behave like the prng_lg_range, but return a result
+ * in [0, range) instead of [0, 2**lg_range).
+ */
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_range_u32(atomic_u32_t *state, uint32_t range, bool atomic) {
+	uint32_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u32(pow2_ceil_u32(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_u32(state, lg_range, atomic);
+	} while (ret >= range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_range_u64(uint64_t *state, uint64_t range) {
+	uint64_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u64(pow2_ceil_u64(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_u64(state, lg_range);
+	} while (ret >= range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_range_zu(atomic_zu_t *state, size_t range, bool atomic) {
+	size_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u64(pow2_ceil_u64(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_zu(state, lg_range, atomic);
+	} while (ret >= range);
+
+	return ret;
+}
+
+#endif /* JEMALLOC_INTERNAL_PRNG_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_externs.h
new file mode 100644
index 0000000..094f3e1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_externs.h
@@ -0,0 +1,105 @@
+#ifndef JEMALLOC_INTERNAL_PROF_EXTERNS_H
+#define JEMALLOC_INTERNAL_PROF_EXTERNS_H
+
+#include "jemalloc/internal/mutex.h"
+
+extern malloc_mutex_t	bt2gctx_mtx;
+
+extern bool	opt_prof;
+extern bool	opt_prof_active;
+extern bool	opt_prof_thread_active_init;
+extern size_t	opt_lg_prof_sample;   /* Mean bytes between samples. */
+extern ssize_t	opt_lg_prof_interval; /* lg(prof_interval). */
+extern bool	opt_prof_gdump;       /* High-water memory dumping. */
+extern bool	opt_prof_final;       /* Final profile dumping. */
+extern bool	opt_prof_leak;        /* Dump leak summary at exit. */
+extern bool	opt_prof_accum;       /* Report cumulative bytes. */
+extern bool	opt_prof_log;	      /* Turn logging on at boot. */
+extern char	opt_prof_prefix[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PATH_MAX +
+#endif
+    1];
+
+/* Accessed via prof_active_[gs]et{_unlocked,}(). */
+extern bool	prof_active;
+
+/* Accessed via prof_gdump_[gs]et{_unlocked,}(). */
+extern bool	prof_gdump_val;
+
+/*
+ * Profile dump interval, measured in bytes allocated.  Each arena triggers a
+ * profile dump when it reaches this threshold.  The effect is that the
+ * interval between profile dumps averages prof_interval, though the actual
+ * interval between dumps will tend to be sporadic, and the interval will be a
+ * maximum of approximately (prof_interval * narenas).
+ */
+extern uint64_t	prof_interval;
+
+/*
+ * Initialized as opt_lg_prof_sample, and potentially modified during profiling
+ * resets.
+ */
+extern size_t	lg_prof_sample;
+
+void prof_alloc_rollback(tsd_t *tsd, prof_tctx_t *tctx, bool updated);
+void prof_malloc_sample_object(tsdn_t *tsdn, const void *ptr, size_t usize,
+    prof_tctx_t *tctx);
+void prof_free_sampled_object(tsd_t *tsd, const void *ptr, size_t usize,
+    prof_tctx_t *tctx);
+void bt_init(prof_bt_t *bt, void **vec);
+void prof_backtrace(prof_bt_t *bt);
+prof_tctx_t *prof_lookup(tsd_t *tsd, prof_bt_t *bt);
+#ifdef JEMALLOC_JET
+size_t prof_tdata_count(void);
+size_t prof_bt_count(void);
+#endif
+typedef int (prof_dump_open_t)(bool, const char *);
+extern prof_dump_open_t *JET_MUTABLE prof_dump_open;
+
+typedef bool (prof_dump_header_t)(tsdn_t *, bool, const prof_cnt_t *);
+extern prof_dump_header_t *JET_MUTABLE prof_dump_header;
+#ifdef JEMALLOC_JET
+void prof_cnt_all(uint64_t *curobjs, uint64_t *curbytes, uint64_t *accumobjs,
+    uint64_t *accumbytes);
+#endif
+bool prof_accum_init(tsdn_t *tsdn, prof_accum_t *prof_accum);
+void prof_idump(tsdn_t *tsdn);
+bool prof_mdump(tsd_t *tsd, const char *filename);
+void prof_gdump(tsdn_t *tsdn);
+prof_tdata_t *prof_tdata_init(tsd_t *tsd);
+prof_tdata_t *prof_tdata_reinit(tsd_t *tsd, prof_tdata_t *tdata);
+void prof_reset(tsd_t *tsd, size_t lg_sample);
+void prof_tdata_cleanup(tsd_t *tsd);
+bool prof_active_get(tsdn_t *tsdn);
+bool prof_active_set(tsdn_t *tsdn, bool active);
+const char *prof_thread_name_get(tsd_t *tsd);
+int prof_thread_name_set(tsd_t *tsd, const char *thread_name);
+bool prof_thread_active_get(tsd_t *tsd);
+bool prof_thread_active_set(tsd_t *tsd, bool active);
+bool prof_thread_active_init_get(tsdn_t *tsdn);
+bool prof_thread_active_init_set(tsdn_t *tsdn, bool active_init);
+bool prof_gdump_get(tsdn_t *tsdn);
+bool prof_gdump_set(tsdn_t *tsdn, bool active);
+void prof_boot0(void);
+void prof_boot1(void);
+bool prof_boot2(tsd_t *tsd);
+void prof_prefork0(tsdn_t *tsdn);
+void prof_prefork1(tsdn_t *tsdn);
+void prof_postfork_parent(tsdn_t *tsdn);
+void prof_postfork_child(tsdn_t *tsdn);
+void prof_sample_threshold_update(prof_tdata_t *tdata);
+
+bool prof_log_start(tsdn_t *tsdn, const char *filename);
+bool prof_log_stop(tsdn_t *tsdn);
+#ifdef JEMALLOC_JET
+size_t prof_log_bt_count(void);
+size_t prof_log_alloc_count(void);
+size_t prof_log_thr_count(void);
+bool prof_log_is_logging(void);
+bool prof_log_rep_check(void);
+void prof_log_dummy_set(bool new_value);
+#endif
+
+#endif /* JEMALLOC_INTERNAL_PROF_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_inlines_a.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_inlines_a.h
new file mode 100644
index 0000000..471d985
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_inlines_a.h
@@ -0,0 +1,85 @@
+#ifndef JEMALLOC_INTERNAL_PROF_INLINES_A_H
+#define JEMALLOC_INTERNAL_PROF_INLINES_A_H
+
+#include "jemalloc/internal/mutex.h"
+
+static inline bool
+prof_accum_add(tsdn_t *tsdn, prof_accum_t *prof_accum,
+    uint64_t accumbytes) {
+	cassert(config_prof);
+
+	bool overflow;
+	uint64_t a0, a1;
+
+	/*
+	 * If the application allocates fast enough (and/or if idump is slow
+	 * enough), extreme overflow here (a1 >= prof_interval * 2) can cause
+	 * idump trigger coalescing.  This is an intentional mechanism that
+	 * avoids rate-limiting allocation.
+	 */
+#ifdef JEMALLOC_ATOMIC_U64
+	a0 = atomic_load_u64(&prof_accum->accumbytes, ATOMIC_RELAXED);
+	do {
+		a1 = a0 + accumbytes;
+		assert(a1 >= a0);
+		overflow = (a1 >= prof_interval);
+		if (overflow) {
+			a1 %= prof_interval;
+		}
+	} while (!atomic_compare_exchange_weak_u64(&prof_accum->accumbytes, &a0,
+	    a1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+#else
+	malloc_mutex_lock(tsdn, &prof_accum->mtx);
+	a0 = prof_accum->accumbytes;
+	a1 = a0 + accumbytes;
+	overflow = (a1 >= prof_interval);
+	if (overflow) {
+		a1 %= prof_interval;
+	}
+	prof_accum->accumbytes = a1;
+	malloc_mutex_unlock(tsdn, &prof_accum->mtx);
+#endif
+	return overflow;
+}
+
+static inline void
+prof_accum_cancel(tsdn_t *tsdn, prof_accum_t *prof_accum,
+    size_t usize) {
+	cassert(config_prof);
+
+	/*
+	 * Cancel out as much of the excessive prof_accumbytes increase as
+	 * possible without underflowing.  Interval-triggered dumps occur
+	 * slightly more often than intended as a result of incomplete
+	 * canceling.
+	 */
+	uint64_t a0, a1;
+#ifdef JEMALLOC_ATOMIC_U64
+	a0 = atomic_load_u64(&prof_accum->accumbytes, ATOMIC_RELAXED);
+	do {
+		a1 = (a0 >= SC_LARGE_MINCLASS - usize)
+		    ? a0 - (SC_LARGE_MINCLASS - usize) : 0;
+	} while (!atomic_compare_exchange_weak_u64(&prof_accum->accumbytes, &a0,
+	    a1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+#else
+	malloc_mutex_lock(tsdn, &prof_accum->mtx);
+	a0 = prof_accum->accumbytes;
+	a1 = (a0 >= SC_LARGE_MINCLASS - usize)
+	    ?  a0 - (SC_LARGE_MINCLASS - usize) : 0;
+	prof_accum->accumbytes = a1;
+	malloc_mutex_unlock(tsdn, &prof_accum->mtx);
+#endif
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_active_get_unlocked(void) {
+	/*
+	 * Even if opt_prof is true, sampling can be temporarily disabled by
+	 * setting prof_active to false.  No locking is used when reading
+	 * prof_active in the fast path, so there are no guarantees regarding
+	 * how long it will take for all threads to notice state changes.
+	 */
+	return prof_active;
+}
+
+#endif /* JEMALLOC_INTERNAL_PROF_INLINES_A_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_inlines_b.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_inlines_b.h
new file mode 100644
index 0000000..8ba8a1e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_inlines_b.h
@@ -0,0 +1,250 @@
+#ifndef JEMALLOC_INTERNAL_PROF_INLINES_B_H
+#define JEMALLOC_INTERNAL_PROF_INLINES_B_H
+
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/sz.h"
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_gdump_get_unlocked(void) {
+	/*
+	 * No locking is used when reading prof_gdump_val in the fast path, so
+	 * there are no guarantees regarding how long it will take for all
+	 * threads to notice state changes.
+	 */
+	return prof_gdump_val;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tdata_t *
+prof_tdata_get(tsd_t *tsd, bool create) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	tdata = tsd_prof_tdata_get(tsd);
+	if (create) {
+		if (unlikely(tdata == NULL)) {
+			if (tsd_nominal(tsd)) {
+				tdata = prof_tdata_init(tsd);
+				tsd_prof_tdata_set(tsd, tdata);
+			}
+		} else if (unlikely(tdata->expired)) {
+			tdata = prof_tdata_reinit(tsd, tdata);
+			tsd_prof_tdata_set(tsd, tdata);
+		}
+		assert(tdata == NULL || tdata->attached);
+	}
+
+	return tdata;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+prof_tctx_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	return arena_prof_tctx_get(tsdn, ptr, alloc_ctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_tctx_set(tsdn_t *tsdn, const void *ptr, size_t usize,
+    alloc_ctx_t *alloc_ctx, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_tctx_set(tsdn, ptr, usize, alloc_ctx, tctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_tctx_reset(tsdn_t *tsdn, const void *ptr, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_tctx_reset(tsdn, ptr, tctx);
+}
+
+JEMALLOC_ALWAYS_INLINE nstime_t
+prof_alloc_time_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	return arena_prof_alloc_time_get(tsdn, ptr, alloc_ctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_alloc_time_set(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx,
+    nstime_t t) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_alloc_time_set(tsdn, ptr, alloc_ctx, t);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_sample_check(tsd_t *tsd, size_t usize, bool update) {
+	ssize_t check = update ? 0 : usize;
+
+	int64_t bytes_until_sample = tsd_bytes_until_sample_get(tsd);
+	if (update) {
+		bytes_until_sample -= usize;
+		if (tsd_nominal(tsd)) {
+			tsd_bytes_until_sample_set(tsd, bytes_until_sample);
+		}
+	}
+	if (likely(bytes_until_sample >= check)) {
+		return true;
+	}
+
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_sample_accum_update(tsd_t *tsd, size_t usize, bool update,
+			 prof_tdata_t **tdata_out) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	/* Fastpath: no need to load tdata */
+	if (likely(prof_sample_check(tsd, usize, update))) {
+		return true;
+	}
+
+	bool booted = tsd_prof_tdata_get(tsd);
+	tdata = prof_tdata_get(tsd, true);
+	if (unlikely((uintptr_t)tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)) {
+		tdata = NULL;
+	}
+
+	if (tdata_out != NULL) {
+		*tdata_out = tdata;
+	}
+
+	if (unlikely(tdata == NULL)) {
+		return true;
+	}
+
+	/*
+	 * If this was the first creation of tdata, then
+	 * prof_tdata_get() reset bytes_until_sample, so decrement and
+	 * check it again
+	 */
+	if (!booted && prof_sample_check(tsd, usize, update)) {
+		return true;
+	}
+
+	if (tsd_reentrancy_level_get(tsd) > 0) {
+		return true;
+	}
+	/* Compute new sample threshold. */
+	if (update) {
+		prof_sample_threshold_update(tdata);
+	}
+	return !tdata->active;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+prof_alloc_prep(tsd_t *tsd, size_t usize, bool prof_active, bool update) {
+	prof_tctx_t *ret;
+	prof_tdata_t *tdata;
+	prof_bt_t bt;
+
+	assert(usize == sz_s2u(usize));
+
+	if (!prof_active || likely(prof_sample_accum_update(tsd, usize, update,
+	    &tdata))) {
+		ret = (prof_tctx_t *)(uintptr_t)1U;
+	} else {
+		bt_init(&bt, tdata->vec);
+		prof_backtrace(&bt);
+		ret = prof_lookup(tsd, &bt);
+	}
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_malloc(tsdn_t *tsdn, const void *ptr, size_t usize, alloc_ctx_t *alloc_ctx,
+    prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(usize == isalloc(tsdn, ptr));
+
+	if (unlikely((uintptr_t)tctx > (uintptr_t)1U)) {
+		prof_malloc_sample_object(tsdn, ptr, usize, tctx);
+	} else {
+		prof_tctx_set(tsdn, ptr, usize, alloc_ctx,
+		    (prof_tctx_t *)(uintptr_t)1U);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_realloc(tsd_t *tsd, const void *ptr, size_t usize, prof_tctx_t *tctx,
+    bool prof_active, bool updated, const void *old_ptr, size_t old_usize,
+    prof_tctx_t *old_tctx) {
+	bool sampled, old_sampled, moved;
+
+	cassert(config_prof);
+	assert(ptr != NULL || (uintptr_t)tctx <= (uintptr_t)1U);
+
+	if (prof_active && !updated && ptr != NULL) {
+		assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+		if (prof_sample_accum_update(tsd, usize, true, NULL)) {
+			/*
+			 * Don't sample.  The usize passed to prof_alloc_prep()
+			 * was larger than what actually got allocated, so a
+			 * backtrace was captured for this allocation, even
+			 * though its actual usize was insufficient to cross the
+			 * sample threshold.
+			 */
+			prof_alloc_rollback(tsd, tctx, true);
+			tctx = (prof_tctx_t *)(uintptr_t)1U;
+		}
+	}
+
+	sampled = ((uintptr_t)tctx > (uintptr_t)1U);
+	old_sampled = ((uintptr_t)old_tctx > (uintptr_t)1U);
+	moved = (ptr != old_ptr);
+
+	if (unlikely(sampled)) {
+		prof_malloc_sample_object(tsd_tsdn(tsd), ptr, usize, tctx);
+	} else if (moved) {
+		prof_tctx_set(tsd_tsdn(tsd), ptr, usize, NULL,
+		    (prof_tctx_t *)(uintptr_t)1U);
+	} else if (unlikely(old_sampled)) {
+		/*
+		 * prof_tctx_set() would work for the !moved case as well, but
+		 * prof_tctx_reset() is slightly cheaper, and the proper thing
+		 * to do here in the presence of explicit knowledge re: moved
+		 * state.
+		 */
+		prof_tctx_reset(tsd_tsdn(tsd), ptr, tctx);
+	} else {
+		assert((uintptr_t)prof_tctx_get(tsd_tsdn(tsd), ptr, NULL) ==
+		    (uintptr_t)1U);
+	}
+
+	/*
+	 * The prof_free_sampled_object() call must come after the
+	 * prof_malloc_sample_object() call, because tctx and old_tctx may be
+	 * the same, in which case reversing the call order could cause the tctx
+	 * to be prematurely destroyed as a side effect of momentarily zeroed
+	 * counters.
+	 */
+	if (unlikely(old_sampled)) {
+		prof_free_sampled_object(tsd, ptr, old_usize, old_tctx);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_free(tsd_t *tsd, const void *ptr, size_t usize, alloc_ctx_t *alloc_ctx) {
+	prof_tctx_t *tctx = prof_tctx_get(tsd_tsdn(tsd), ptr, alloc_ctx);
+
+	cassert(config_prof);
+	assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+
+	if (unlikely((uintptr_t)tctx > (uintptr_t)1U)) {
+		prof_free_sampled_object(tsd, ptr, usize, tctx);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_PROF_INLINES_B_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_structs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_structs.h
new file mode 100644
index 0000000..34ed482
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_structs.h
@@ -0,0 +1,200 @@
+#ifndef JEMALLOC_INTERNAL_PROF_STRUCTS_H
+#define JEMALLOC_INTERNAL_PROF_STRUCTS_H
+
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/rb.h"
+
+struct prof_bt_s {
+	/* Backtrace, stored as len program counters. */
+	void		**vec;
+	unsigned	len;
+};
+
+#ifdef JEMALLOC_PROF_LIBGCC
+/* Data structure passed to libgcc _Unwind_Backtrace() callback functions. */
+typedef struct {
+	prof_bt_t	*bt;
+	unsigned	max;
+} prof_unwind_data_t;
+#endif
+
+struct prof_accum_s {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_t	mtx;
+	uint64_t	accumbytes;
+#else
+	atomic_u64_t	accumbytes;
+#endif
+};
+
+struct prof_cnt_s {
+	/* Profiling counters. */
+	uint64_t	curobjs;
+	uint64_t	curbytes;
+	uint64_t	accumobjs;
+	uint64_t	accumbytes;
+};
+
+typedef enum {
+	prof_tctx_state_initializing,
+	prof_tctx_state_nominal,
+	prof_tctx_state_dumping,
+	prof_tctx_state_purgatory /* Dumper must finish destroying. */
+} prof_tctx_state_t;
+
+struct prof_tctx_s {
+	/* Thread data for thread that performed the allocation. */
+	prof_tdata_t		*tdata;
+
+	/*
+	 * Copy of tdata->thr_{uid,discrim}, necessary because tdata may be
+	 * defunct during teardown.
+	 */
+	uint64_t		thr_uid;
+	uint64_t		thr_discrim;
+
+	/* Profiling counters, protected by tdata->lock. */
+	prof_cnt_t		cnts;
+
+	/* Associated global context. */
+	prof_gctx_t		*gctx;
+
+	/*
+	 * UID that distinguishes multiple tctx's created by the same thread,
+	 * but coexisting in gctx->tctxs.  There are two ways that such
+	 * coexistence can occur:
+	 * - A dumper thread can cause a tctx to be retained in the purgatory
+	 *   state.
+	 * - Although a single "producer" thread must create all tctx's which
+	 *   share the same thr_uid, multiple "consumers" can each concurrently
+	 *   execute portions of prof_tctx_destroy().  prof_tctx_destroy() only
+	 *   gets called once each time cnts.cur{objs,bytes} drop to 0, but this
+	 *   threshold can be hit again before the first consumer finishes
+	 *   executing prof_tctx_destroy().
+	 */
+	uint64_t		tctx_uid;
+
+	/* Linkage into gctx's tctxs. */
+	rb_node(prof_tctx_t)	tctx_link;
+
+	/*
+	 * True during prof_alloc_prep()..prof_malloc_sample_object(), prevents
+	 * sample vs destroy race.
+	 */
+	bool			prepared;
+
+	/* Current dump-related state, protected by gctx->lock. */
+	prof_tctx_state_t	state;
+
+	/*
+	 * Copy of cnts snapshotted during early dump phase, protected by
+	 * dump_mtx.
+	 */
+	prof_cnt_t		dump_cnts;
+};
+typedef rb_tree(prof_tctx_t) prof_tctx_tree_t;
+
+struct prof_gctx_s {
+	/* Protects nlimbo, cnt_summed, and tctxs. */
+	malloc_mutex_t		*lock;
+
+	/*
+	 * Number of threads that currently cause this gctx to be in a state of
+	 * limbo due to one of:
+	 *   - Initializing this gctx.
+	 *   - Initializing per thread counters associated with this gctx.
+	 *   - Preparing to destroy this gctx.
+	 *   - Dumping a heap profile that includes this gctx.
+	 * nlimbo must be 1 (single destroyer) in order to safely destroy the
+	 * gctx.
+	 */
+	unsigned		nlimbo;
+
+	/*
+	 * Tree of profile counters, one for each thread that has allocated in
+	 * this context.
+	 */
+	prof_tctx_tree_t	tctxs;
+
+	/* Linkage for tree of contexts to be dumped. */
+	rb_node(prof_gctx_t)	dump_link;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* Associated backtrace. */
+	prof_bt_t		bt;
+
+	/* Backtrace vector, variable size, referred to by bt. */
+	void			*vec[1];
+};
+typedef rb_tree(prof_gctx_t) prof_gctx_tree_t;
+
+struct prof_tdata_s {
+	malloc_mutex_t		*lock;
+
+	/* Monotonically increasing unique thread identifier. */
+	uint64_t		thr_uid;
+
+	/*
+	 * Monotonically increasing discriminator among tdata structures
+	 * associated with the same thr_uid.
+	 */
+	uint64_t		thr_discrim;
+
+	/* Included in heap profile dumps if non-NULL. */
+	char			*thread_name;
+
+	bool			attached;
+	bool			expired;
+
+	rb_node(prof_tdata_t)	tdata_link;
+
+	/*
+	 * Counter used to initialize prof_tctx_t's tctx_uid.  No locking is
+	 * necessary when incrementing this field, because only one thread ever
+	 * does so.
+	 */
+	uint64_t		tctx_uid_next;
+
+	/*
+	 * Hash of (prof_bt_t *)-->(prof_tctx_t *).  Each thread tracks
+	 * backtraces for which it has non-zero allocation/deallocation counters
+	 * associated with thread-specific prof_tctx_t objects.  Other threads
+	 * may write to prof_tctx_t contents when freeing associated objects.
+	 */
+	ckh_t			bt2tctx;
+
+	/* Sampling state. */
+	uint64_t		prng_state;
+
+	/* State used to avoid dumping while operating on prof internals. */
+	bool			enq;
+	bool			enq_idump;
+	bool			enq_gdump;
+
+	/*
+	 * Set to true during an early dump phase for tdata's which are
+	 * currently being dumped.  New threads' tdata's have this initialized
+	 * to false so that they aren't accidentally included in later dump
+	 * phases.
+	 */
+	bool			dumping;
+
+	/*
+	 * True if profiling is active for this tdata's thread
+	 * (thread.prof.active mallctl).
+	 */
+	bool			active;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* Backtrace vector, used for calls to prof_backtrace(). */
+	void			*vec[PROF_BT_MAX];
+};
+typedef rb_tree(prof_tdata_t) prof_tdata_tree_t;
+
+#endif /* JEMALLOC_INTERNAL_PROF_STRUCTS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_types.h
new file mode 100644
index 0000000..1eff995
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/prof_types.h
@@ -0,0 +1,56 @@
+#ifndef JEMALLOC_INTERNAL_PROF_TYPES_H
+#define JEMALLOC_INTERNAL_PROF_TYPES_H
+
+typedef struct prof_bt_s prof_bt_t;
+typedef struct prof_accum_s prof_accum_t;
+typedef struct prof_cnt_s prof_cnt_t;
+typedef struct prof_tctx_s prof_tctx_t;
+typedef struct prof_gctx_s prof_gctx_t;
+typedef struct prof_tdata_s prof_tdata_t;
+
+/* Option defaults. */
+#ifdef JEMALLOC_PROF
+#  define PROF_PREFIX_DEFAULT		"jeprof"
+#else
+#  define PROF_PREFIX_DEFAULT		""
+#endif
+#define LG_PROF_SAMPLE_DEFAULT		19
+#define LG_PROF_INTERVAL_DEFAULT	-1
+
+/*
+ * Hard limit on stack backtrace depth.  The version of prof_backtrace() that
+ * is based on __builtin_return_address() necessarily has a hard-coded number
+ * of backtrace frame handlers, and should be kept in sync with this setting.
+ */
+#define PROF_BT_MAX			128
+
+/* Initial hash table size. */
+#define PROF_CKH_MINITEMS		64
+
+/* Size of memory buffer to use when writing dump files. */
+#define PROF_DUMP_BUFSIZE		65536
+
+/* Size of stack-allocated buffer used by prof_printf(). */
+#define PROF_PRINTF_BUFSIZE		128
+
+/*
+ * Number of mutexes shared among all gctx's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define PROF_NCTX_LOCKS			1024
+
+/*
+ * Number of mutexes shared among all tdata's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define PROF_NTDATA_LOCKS		256
+
+/*
+ * prof_tdata pointers close to NULL are used to encode state information that
+ * is used for cleaning up during thread shutdown.
+ */
+#define PROF_TDATA_STATE_REINCARNATED	((prof_tdata_t *)(uintptr_t)1)
+#define PROF_TDATA_STATE_PURGATORY	((prof_tdata_t *)(uintptr_t)2)
+#define PROF_TDATA_STATE_MAX		PROF_TDATA_STATE_PURGATORY
+
+#endif /* JEMALLOC_INTERNAL_PROF_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/public_namespace.sh b/memkind-1.10.0/jemalloc/include/jemalloc/internal/public_namespace.sh
new file mode 100755
index 0000000..4d415ba
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/public_namespace.sh
@@ -0,0 +1,6 @@
+#!/bin/sh
+
+for nm in `cat $1` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#define je_${n} JEMALLOC_N(${n})"
+done
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/public_unnamespace.sh b/memkind-1.10.0/jemalloc/include/jemalloc/internal/public_unnamespace.sh
new file mode 100755
index 0000000..4239d17
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/public_unnamespace.sh
@@ -0,0 +1,6 @@
+#!/bin/sh
+
+for nm in `cat $1` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#undef je_${n}"
+done
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/ql.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ql.h
new file mode 100644
index 0000000..8029040
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ql.h
@@ -0,0 +1,88 @@
+#ifndef JEMALLOC_INTERNAL_QL_H
+#define JEMALLOC_INTERNAL_QL_H
+
+#include "jemalloc/internal/qr.h"
+
+/* List definitions. */
+#define ql_head(a_type)							\
+struct {								\
+	a_type *qlh_first;						\
+}
+
+#define ql_head_initializer(a_head) {NULL}
+
+#define ql_elm(a_type)	qr(a_type)
+
+/* List functions. */
+#define ql_new(a_head) do {						\
+	(a_head)->qlh_first = NULL;					\
+} while (0)
+
+#define ql_elm_new(a_elm, a_field) qr_new((a_elm), a_field)
+
+#define ql_first(a_head) ((a_head)->qlh_first)
+
+#define ql_last(a_head, a_field)					\
+	((ql_first(a_head) != NULL)					\
+	    ? qr_prev(ql_first(a_head), a_field) : NULL)
+
+#define ql_next(a_head, a_elm, a_field)					\
+	((ql_last(a_head, a_field) != (a_elm))				\
+	    ? qr_next((a_elm), a_field)	: NULL)
+
+#define ql_prev(a_head, a_elm, a_field)					\
+	((ql_first(a_head) != (a_elm)) ? qr_prev((a_elm), a_field)	\
+				       : NULL)
+
+#define ql_before_insert(a_head, a_qlelm, a_elm, a_field) do {		\
+	qr_before_insert((a_qlelm), (a_elm), a_field);			\
+	if (ql_first(a_head) == (a_qlelm)) {				\
+		ql_first(a_head) = (a_elm);				\
+	}								\
+} while (0)
+
+#define ql_after_insert(a_qlelm, a_elm, a_field)			\
+	qr_after_insert((a_qlelm), (a_elm), a_field)
+
+#define ql_head_insert(a_head, a_elm, a_field) do {			\
+	if (ql_first(a_head) != NULL) {					\
+		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
+	}								\
+	ql_first(a_head) = (a_elm);					\
+} while (0)
+
+#define ql_tail_insert(a_head, a_elm, a_field) do {			\
+	if (ql_first(a_head) != NULL) {					\
+		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
+	}								\
+	ql_first(a_head) = qr_next((a_elm), a_field);			\
+} while (0)
+
+#define ql_remove(a_head, a_elm, a_field) do {				\
+	if (ql_first(a_head) == (a_elm)) {				\
+		ql_first(a_head) = qr_next(ql_first(a_head), a_field);	\
+	}								\
+	if (ql_first(a_head) != (a_elm)) {				\
+		qr_remove((a_elm), a_field);				\
+	} else {							\
+		ql_first(a_head) = NULL;				\
+	}								\
+} while (0)
+
+#define ql_head_remove(a_head, a_type, a_field) do {			\
+	a_type *t = ql_first(a_head);					\
+	ql_remove((a_head), t, a_field);				\
+} while (0)
+
+#define ql_tail_remove(a_head, a_type, a_field) do {			\
+	a_type *t = ql_last(a_head, a_field);				\
+	ql_remove((a_head), t, a_field);				\
+} while (0)
+
+#define ql_foreach(a_var, a_head, a_field)				\
+	qr_foreach((a_var), ql_first(a_head), a_field)
+
+#define ql_reverse_foreach(a_var, a_head, a_field)			\
+	qr_reverse_foreach((a_var), ql_first(a_head), a_field)
+
+#endif /* JEMALLOC_INTERNAL_QL_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/qr.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/qr.h
new file mode 100644
index 0000000..1e1056b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/qr.h
@@ -0,0 +1,72 @@
+#ifndef JEMALLOC_INTERNAL_QR_H
+#define JEMALLOC_INTERNAL_QR_H
+
+/* Ring definitions. */
+#define qr(a_type)							\
+struct {								\
+	a_type	*qre_next;						\
+	a_type	*qre_prev;						\
+}
+
+/* Ring functions. */
+#define qr_new(a_qr, a_field) do {					\
+	(a_qr)->a_field.qre_next = (a_qr);				\
+	(a_qr)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_next(a_qr, a_field) ((a_qr)->a_field.qre_next)
+
+#define qr_prev(a_qr, a_field) ((a_qr)->a_field.qre_prev)
+
+#define qr_before_insert(a_qrelm, a_qr, a_field) do {			\
+	(a_qr)->a_field.qre_prev = (a_qrelm)->a_field.qre_prev;		\
+	(a_qr)->a_field.qre_next = (a_qrelm);				\
+	(a_qr)->a_field.qre_prev->a_field.qre_next = (a_qr);		\
+	(a_qrelm)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_after_insert(a_qrelm, a_qr, a_field) do {			\
+	(a_qr)->a_field.qre_next = (a_qrelm)->a_field.qre_next;		\
+	(a_qr)->a_field.qre_prev = (a_qrelm);				\
+	(a_qr)->a_field.qre_next->a_field.qre_prev = (a_qr);		\
+	(a_qrelm)->a_field.qre_next = (a_qr);				\
+} while (0)
+
+#define qr_meld(a_qr_a, a_qr_b, a_type, a_field) do {			\
+	a_type *t;							\
+	(a_qr_a)->a_field.qre_prev->a_field.qre_next = (a_qr_b);	\
+	(a_qr_b)->a_field.qre_prev->a_field.qre_next = (a_qr_a);	\
+	t = (a_qr_a)->a_field.qre_prev;					\
+	(a_qr_a)->a_field.qre_prev = (a_qr_b)->a_field.qre_prev;	\
+	(a_qr_b)->a_field.qre_prev = t;					\
+} while (0)
+
+/*
+ * qr_meld() and qr_split() are functionally equivalent, so there's no need to
+ * have two copies of the code.
+ */
+#define qr_split(a_qr_a, a_qr_b, a_type, a_field)			\
+	qr_meld((a_qr_a), (a_qr_b), a_type, a_field)
+
+#define qr_remove(a_qr, a_field) do {					\
+	(a_qr)->a_field.qre_prev->a_field.qre_next			\
+	    = (a_qr)->a_field.qre_next;					\
+	(a_qr)->a_field.qre_next->a_field.qre_prev			\
+	    = (a_qr)->a_field.qre_prev;					\
+	(a_qr)->a_field.qre_next = (a_qr);				\
+	(a_qr)->a_field.qre_prev = (a_qr);				\
+} while (0)
+
+#define qr_foreach(var, a_qr, a_field)					\
+	for ((var) = (a_qr);						\
+	    (var) != NULL;						\
+	    (var) = (((var)->a_field.qre_next != (a_qr))		\
+	    ? (var)->a_field.qre_next : NULL))
+
+#define qr_reverse_foreach(var, a_qr, a_field)				\
+	for ((var) = ((a_qr) != NULL) ? qr_prev(a_qr, a_field) : NULL;	\
+	    (var) != NULL;						\
+	    (var) = (((var) != (a_qr))					\
+	    ? (var)->a_field.qre_prev : NULL))
+
+#endif /* JEMALLOC_INTERNAL_QR_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/quantum.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/quantum.h
new file mode 100644
index 0000000..821086e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/quantum.h
@@ -0,0 +1,77 @@
+#ifndef JEMALLOC_INTERNAL_QUANTUM_H
+#define JEMALLOC_INTERNAL_QUANTUM_H
+
+/*
+ * Minimum allocation alignment is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#ifndef LG_QUANTUM
+#  if (defined(__i386__) || defined(_M_IX86))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __ia64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __alpha__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__sparc64__) || defined(__sparcv9) || defined(__sparc_v9__))
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__amd64__) || defined(__x86_64__) || defined(_M_X64))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __arm__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __aarch64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __hppa__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __m68k__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __mips__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __nios2__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __or1k__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __powerpc__
+#    define LG_QUANTUM		4
+#  endif
+#  if defined(__riscv) || defined(__riscv__)
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __s390__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined (__SH3E__) || defined(__SH4_SINGLE__) || defined(__SH4__) || \
+	defined(__SH4_SINGLE_ONLY__))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __tile__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __le32__
+#    define LG_QUANTUM		4
+#  endif
+#  ifndef LG_QUANTUM
+#    error "Unknown minimum alignment for architecture; specify via "
+	 "--with-lg-quantum"
+#  endif
+#endif
+
+#define QUANTUM			((size_t)(1U << LG_QUANTUM))
+#define QUANTUM_MASK		(QUANTUM - 1)
+
+/* Return the smallest quantum multiple that is >= a. */
+#define QUANTUM_CEILING(a)						\
+	(((a) + QUANTUM_MASK) & ~QUANTUM_MASK)
+
+#endif /* JEMALLOC_INTERNAL_QUANTUM_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/rb.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/rb.h
new file mode 100644
index 0000000..47fa5ca
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/rb.h
@@ -0,0 +1,1006 @@
+/*-
+ *******************************************************************************
+ *
+ * cpp macro implementation of left-leaning 2-3 red-black trees.  Parent
+ * pointers are not used, and color bits are stored in the least significant
+ * bit of right-child pointers (if RB_COMPACT is defined), thus making node
+ * linkage as compact as is possible for red-black trees.
+ *
+ * Usage:
+ *
+ *   #include <stdint.h>
+ *   #include <stdbool.h>
+ *   #define NDEBUG // (Optional, see assert(3).)
+ *   #include <assert.h>
+ *   #define RB_COMPACT // (Optional, embed color bits in right-child pointers.)
+ *   #include <rb.h>
+ *   ...
+ *
+ *******************************************************************************
+ */
+
+#ifndef RB_H_
+#define RB_H_
+
+#ifndef __PGI
+#define RB_COMPACT
+#endif
+
+#ifdef RB_COMPACT
+/* Node structure. */
+#define rb_node(a_type)							\
+struct {								\
+    a_type *rbn_left;							\
+    a_type *rbn_right_red;						\
+}
+#else
+#define rb_node(a_type)							\
+struct {								\
+    a_type *rbn_left;							\
+    a_type *rbn_right;							\
+    bool rbn_red;							\
+}
+#endif
+
+/* Root structure. */
+#define rb_tree(a_type)							\
+struct {								\
+    a_type *rbt_root;							\
+}
+
+/* Left accessors. */
+#define rbtn_left_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_left)
+#define rbtn_left_set(a_type, a_field, a_node, a_left) do {		\
+    (a_node)->a_field.rbn_left = a_left;				\
+} while (0)
+
+#ifdef RB_COMPACT
+/* Right accessors. */
+#define rbtn_right_get(a_type, a_field, a_node)				\
+    ((a_type *) (((intptr_t) (a_node)->a_field.rbn_right_red)		\
+      & ((ssize_t)-2)))
+#define rbtn_right_set(a_type, a_field, a_node, a_right) do {		\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((uintptr_t) a_right)	\
+      | (((uintptr_t) (a_node)->a_field.rbn_right_red) & ((size_t)1)));	\
+} while (0)
+
+/* Color accessors. */
+#define rbtn_red_get(a_type, a_field, a_node)				\
+    ((bool) (((uintptr_t) (a_node)->a_field.rbn_right_red)		\
+      & ((size_t)1)))
+#define rbtn_color_set(a_type, a_field, a_node, a_red) do {		\
+    (a_node)->a_field.rbn_right_red = (a_type *) ((((intptr_t)		\
+      (a_node)->a_field.rbn_right_red) & ((ssize_t)-2))			\
+      | ((ssize_t)a_red));						\
+} while (0)
+#define rbtn_red_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((uintptr_t)		\
+      (a_node)->a_field.rbn_right_red) | ((size_t)1));			\
+} while (0)
+#define rbtn_black_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_right_red = (a_type *) (((intptr_t)		\
+      (a_node)->a_field.rbn_right_red) & ((ssize_t)-2));		\
+} while (0)
+
+/* Node initializer. */
+#define rbt_node_new(a_type, a_field, a_rbt, a_node) do {		\
+    /* Bookkeeping bit cannot be used by node pointer. */		\
+    assert(((uintptr_t)(a_node) & 0x1) == 0);				\
+    rbtn_left_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_right_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_red_set(a_type, a_field, (a_node));				\
+} while (0)
+#else
+/* Right accessors. */
+#define rbtn_right_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_right)
+#define rbtn_right_set(a_type, a_field, a_node, a_right) do {		\
+    (a_node)->a_field.rbn_right = a_right;				\
+} while (0)
+
+/* Color accessors. */
+#define rbtn_red_get(a_type, a_field, a_node)				\
+    ((a_node)->a_field.rbn_red)
+#define rbtn_color_set(a_type, a_field, a_node, a_red) do {		\
+    (a_node)->a_field.rbn_red = (a_red);				\
+} while (0)
+#define rbtn_red_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_red = true;					\
+} while (0)
+#define rbtn_black_set(a_type, a_field, a_node) do {			\
+    (a_node)->a_field.rbn_red = false;					\
+} while (0)
+
+/* Node initializer. */
+#define rbt_node_new(a_type, a_field, a_rbt, a_node) do {		\
+    rbtn_left_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_right_set(a_type, a_field, (a_node), NULL);	\
+    rbtn_red_set(a_type, a_field, (a_node));				\
+} while (0)
+#endif
+
+/* Tree initializer. */
+#define rb_new(a_type, a_field, a_rbt) do {				\
+    (a_rbt)->rbt_root = NULL;						\
+} while (0)
+
+/* Internal utility macros. */
+#define rbtn_first(a_type, a_field, a_rbt, a_root, r_node) do {		\
+    (r_node) = (a_root);						\
+    if ((r_node) != NULL) {						\
+	for (;								\
+	  rbtn_left_get(a_type, a_field, (r_node)) != NULL;		\
+	  (r_node) = rbtn_left_get(a_type, a_field, (r_node))) {	\
+	}								\
+    }									\
+} while (0)
+
+#define rbtn_last(a_type, a_field, a_rbt, a_root, r_node) do {		\
+    (r_node) = (a_root);						\
+    if ((r_node) != NULL) {						\
+	for (; rbtn_right_get(a_type, a_field, (r_node)) != NULL;	\
+	  (r_node) = rbtn_right_get(a_type, a_field, (r_node))) {	\
+	}								\
+    }									\
+} while (0)
+
+#define rbtn_rotate_left(a_type, a_field, a_node, r_node) do {		\
+    (r_node) = rbtn_right_get(a_type, a_field, (a_node));		\
+    rbtn_right_set(a_type, a_field, (a_node),				\
+      rbtn_left_get(a_type, a_field, (r_node)));			\
+    rbtn_left_set(a_type, a_field, (r_node), (a_node));			\
+} while (0)
+
+#define rbtn_rotate_right(a_type, a_field, a_node, r_node) do {		\
+    (r_node) = rbtn_left_get(a_type, a_field, (a_node));		\
+    rbtn_left_set(a_type, a_field, (a_node),				\
+      rbtn_right_get(a_type, a_field, (r_node)));			\
+    rbtn_right_set(a_type, a_field, (r_node), (a_node));		\
+} while (0)
+
+/*
+ * The rb_proto() macro generates function prototypes that correspond to the
+ * functions generated by an equivalently parameterized call to rb_gen().
+ */
+
+#define rb_proto(a_attr, a_prefix, a_rbt_type, a_type)			\
+a_attr void								\
+a_prefix##new(a_rbt_type *rbtree);					\
+a_attr bool								\
+a_prefix##empty(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##first(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##last(a_rbt_type *rbtree);					\
+a_attr a_type *								\
+a_prefix##next(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##prev(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##search(a_rbt_type *rbtree, const a_type *key);		\
+a_attr a_type *								\
+a_prefix##nsearch(a_rbt_type *rbtree, const a_type *key);		\
+a_attr a_type *								\
+a_prefix##psearch(a_rbt_type *rbtree, const a_type *key);		\
+a_attr void								\
+a_prefix##insert(a_rbt_type *rbtree, a_type *node);			\
+a_attr void								\
+a_prefix##remove(a_rbt_type *rbtree, a_type *node);			\
+a_attr a_type *								\
+a_prefix##iter(a_rbt_type *rbtree, a_type *start, a_type *(*cb)(	\
+  a_rbt_type *, a_type *, void *), void *arg);				\
+a_attr a_type *								\
+a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg);		\
+a_attr void								\
+a_prefix##destroy(a_rbt_type *rbtree, void (*cb)(a_type *, void *),	\
+  void *arg);
+
+/*
+ * The rb_gen() macro generates a type-specific red-black tree implementation,
+ * based on the above cpp macros.
+ *
+ * Arguments:
+ *
+ *   a_attr    : Function attribute for generated functions (ex: static).
+ *   a_prefix  : Prefix for generated functions (ex: ex_).
+ *   a_rb_type : Type for red-black tree data structure (ex: ex_t).
+ *   a_type    : Type for red-black tree node data structure (ex: ex_node_t).
+ *   a_field   : Name of red-black tree node linkage (ex: ex_link).
+ *   a_cmp     : Node comparison function name, with the following prototype:
+ *                 int (a_cmp *)(a_type *a_node, a_type *a_other);
+ *                                       ^^^^^^
+ *                                    or a_key
+ *               Interpretation of comparison function return values:
+ *                 -1 : a_node <  a_other
+ *                  0 : a_node == a_other
+ *                  1 : a_node >  a_other
+ *               In all cases, the a_node or a_key macro argument is the first
+ *               argument to the comparison function, which makes it possible
+ *               to write comparison functions that treat the first argument
+ *               specially.
+ *
+ * Assuming the following setup:
+ *
+ *   typedef struct ex_node_s ex_node_t;
+ *   struct ex_node_s {
+ *       rb_node(ex_node_t) ex_link;
+ *   };
+ *   typedef rb_tree(ex_node_t) ex_t;
+ *   rb_gen(static, ex_, ex_t, ex_node_t, ex_link, ex_cmp)
+ *
+ * The following API is generated:
+ *
+ *   static void
+ *   ex_new(ex_t *tree);
+ *       Description: Initialize a red-black tree structure.
+ *       Args:
+ *         tree: Pointer to an uninitialized red-black tree object.
+ *
+ *   static bool
+ *   ex_empty(ex_t *tree);
+ *       Description: Determine whether tree is empty.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *       Ret: True if tree is empty, false otherwise.
+ *
+ *   static ex_node_t *
+ *   ex_first(ex_t *tree);
+ *   static ex_node_t *
+ *   ex_last(ex_t *tree);
+ *       Description: Get the first/last node in tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *       Ret: First/last node in tree, or NULL if tree is empty.
+ *
+ *   static ex_node_t *
+ *   ex_next(ex_t *tree, ex_node_t *node);
+ *   static ex_node_t *
+ *   ex_prev(ex_t *tree, ex_node_t *node);
+ *       Description: Get node's successor/predecessor.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: A node in tree.
+ *       Ret: node's successor/predecessor in tree, or NULL if node is
+ *            last/first.
+ *
+ *   static ex_node_t *
+ *   ex_search(ex_t *tree, const ex_node_t *key);
+ *       Description: Search for node that matches key.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or NULL if no match.
+ *
+ *   static ex_node_t *
+ *   ex_nsearch(ex_t *tree, const ex_node_t *key);
+ *   static ex_node_t *
+ *   ex_psearch(ex_t *tree, const ex_node_t *key);
+ *       Description: Search for node that matches key.  If no match is found,
+ *                    return what would be key's successor/predecessor, were
+ *                    key in tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or if no match, hypothetical node's
+ *            successor/predecessor (NULL if no successor/predecessor).
+ *
+ *   static void
+ *   ex_insert(ex_t *tree, ex_node_t *node);
+ *       Description: Insert node into tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node to be inserted into tree.
+ *
+ *   static void
+ *   ex_remove(ex_t *tree, ex_node_t *node);
+ *       Description: Remove node from tree.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node in tree to be removed.
+ *
+ *   static ex_node_t *
+ *   ex_iter(ex_t *tree, ex_node_t *start, ex_node_t *(*cb)(ex_t *,
+ *     ex_node_t *, void *), void *arg);
+ *   static ex_node_t *
+ *   ex_reverse_iter(ex_t *tree, ex_node_t *start, ex_node *(*cb)(ex_t *,
+ *     ex_node_t *, void *), void *arg);
+ *       Description: Iterate forward/backward over tree, starting at node.  If
+ *                    tree is modified, iteration must be immediately
+ *                    terminated by the callback function that causes the
+ *                    modification.
+ *       Args:
+ *         tree : Pointer to an initialized red-black tree object.
+ *         start: Node at which to start iteration, or NULL to start at
+ *                first/last node.
+ *         cb   : Callback function, which is called for each node during
+ *                iteration.  Under normal circumstances the callback function
+ *                should return NULL, which causes iteration to continue.  If a
+ *                callback function returns non-NULL, iteration is immediately
+ *                terminated and the non-NULL return value is returned by the
+ *                iterator.  This is useful for re-starting iteration after
+ *                modifying tree.
+ *         arg  : Opaque pointer passed to cb().
+ *       Ret: NULL if iteration completed, or the non-NULL callback return value
+ *            that caused termination of the iteration.
+ *
+ *   static void
+ *   ex_destroy(ex_t *tree, void (*cb)(ex_node_t *, void *), void *arg);
+ *       Description: Iterate over the tree with post-order traversal, remove
+ *                    each node, and run the callback if non-null.  This is
+ *                    used for destroying a tree without paying the cost to
+ *                    rebalance it.  The tree must not be otherwise altered
+ *                    during traversal.
+ *       Args:
+ *         tree: Pointer to an initialized red-black tree object.
+ *         cb  : Callback function, which, if non-null, is called for each node
+ *               during iteration.  There is no way to stop iteration once it
+ *               has begun.
+ *         arg : Opaque pointer passed to cb().
+ */
+#define rb_gen(a_attr, a_prefix, a_rbt_type, a_type, a_field, a_cmp)	\
+a_attr void								\
+a_prefix##new(a_rbt_type *rbtree) {					\
+    rb_new(a_type, a_field, rbtree);					\
+}									\
+a_attr bool								\
+a_prefix##empty(a_rbt_type *rbtree) {					\
+    return (rbtree->rbt_root == NULL);					\
+}									\
+a_attr a_type *								\
+a_prefix##first(a_rbt_type *rbtree) {					\
+    a_type *ret;							\
+    rbtn_first(a_type, a_field, rbtree, rbtree->rbt_root, ret);		\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##last(a_rbt_type *rbtree) {					\
+    a_type *ret;							\
+    rbtn_last(a_type, a_field, rbtree, rbtree->rbt_root, ret);		\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##next(a_rbt_type *rbtree, a_type *node) {			\
+    a_type *ret;							\
+    if (rbtn_right_get(a_type, a_field, node) != NULL) {		\
+	rbtn_first(a_type, a_field, rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), ret);						\
+    } else {								\
+	a_type *tnode = rbtree->rbt_root;				\
+	assert(tnode != NULL);						\
+	ret = NULL;							\
+	while (true) {							\
+	    int cmp = (a_cmp)(node, tnode);				\
+	    if (cmp < 0) {						\
+		ret = tnode;						\
+		tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	    } else if (cmp > 0) {					\
+		tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	    } else {							\
+		break;							\
+	    }								\
+	    assert(tnode != NULL);					\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##prev(a_rbt_type *rbtree, a_type *node) {			\
+    a_type *ret;							\
+    if (rbtn_left_get(a_type, a_field, node) != NULL) {			\
+	rbtn_last(a_type, a_field, rbtree, rbtn_left_get(a_type,	\
+	  a_field, node), ret);						\
+    } else {								\
+	a_type *tnode = rbtree->rbt_root;				\
+	assert(tnode != NULL);						\
+	ret = NULL;							\
+	while (true) {							\
+	    int cmp = (a_cmp)(node, tnode);				\
+	    if (cmp < 0) {						\
+		tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	    } else if (cmp > 0) {					\
+		ret = tnode;						\
+		tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	    } else {							\
+		break;							\
+	    }								\
+	    assert(tnode != NULL);					\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##search(a_rbt_type *rbtree, const a_type *key) {		\
+    a_type *ret;							\
+    int cmp;								\
+    ret = rbtree->rbt_root;						\
+    while (ret != NULL							\
+      && (cmp = (a_cmp)(key, ret)) != 0) {				\
+	if (cmp < 0) {							\
+	    ret = rbtn_left_get(a_type, a_field, ret);			\
+	} else {							\
+	    ret = rbtn_right_get(a_type, a_field, ret);			\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##nsearch(a_rbt_type *rbtree, const a_type *key) {		\
+    a_type *ret;							\
+    a_type *tnode = rbtree->rbt_root;					\
+    ret = NULL;								\
+    while (tnode != NULL) {						\
+	int cmp = (a_cmp)(key, tnode);					\
+	if (cmp < 0) {							\
+	    ret = tnode;						\
+	    tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	} else if (cmp > 0) {						\
+	    tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	} else {							\
+	    ret = tnode;						\
+	    break;							\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##psearch(a_rbt_type *rbtree, const a_type *key) {		\
+    a_type *ret;							\
+    a_type *tnode = rbtree->rbt_root;					\
+    ret = NULL;								\
+    while (tnode != NULL) {						\
+	int cmp = (a_cmp)(key, tnode);					\
+	if (cmp < 0) {							\
+	    tnode = rbtn_left_get(a_type, a_field, tnode);		\
+	} else if (cmp > 0) {						\
+	    ret = tnode;						\
+	    tnode = rbtn_right_get(a_type, a_field, tnode);		\
+	} else {							\
+	    ret = tnode;						\
+	    break;							\
+	}								\
+    }									\
+    return ret;								\
+}									\
+a_attr void								\
+a_prefix##insert(a_rbt_type *rbtree, a_type *node) {			\
+    struct {								\
+	a_type *node;							\
+	int cmp;							\
+    } path[sizeof(void *) << 4], *pathp;				\
+    rbt_node_new(a_type, a_field, rbtree, node);			\
+    /* Wind. */								\
+    path->node = rbtree->rbt_root;					\
+    for (pathp = path; pathp->node != NULL; pathp++) {			\
+	int cmp = pathp->cmp = a_cmp(node, pathp->node);		\
+	assert(cmp != 0);						\
+	if (cmp < 0) {							\
+	    pathp[1].node = rbtn_left_get(a_type, a_field,		\
+	      pathp->node);						\
+	} else {							\
+	    pathp[1].node = rbtn_right_get(a_type, a_field,		\
+	      pathp->node);						\
+	}								\
+    }									\
+    pathp->node = node;							\
+    /* Unwind. */							\
+    for (pathp--; (uintptr_t)pathp >= (uintptr_t)path; pathp--) {	\
+	a_type *cnode = pathp->node;					\
+	if (pathp->cmp < 0) {						\
+	    a_type *left = pathp[1].node;				\
+	    rbtn_left_set(a_type, a_field, cnode, left);		\
+	    if (rbtn_red_get(a_type, a_field, left)) {			\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (leftleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  leftleft)) {						\
+		    /* Fix up 4-node. */				\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, cnode, tnode);	\
+		    cnode = tnode;					\
+		}							\
+	    } else {							\
+		return;							\
+	    }								\
+	} else {							\
+	    a_type *right = pathp[1].node;				\
+	    rbtn_right_set(a_type, a_field, cnode, right);		\
+	    if (rbtn_red_get(a_type, a_field, right)) {			\
+		a_type *left = rbtn_left_get(a_type, a_field, cnode);	\
+		if (left != NULL && rbtn_red_get(a_type, a_field,	\
+		  left)) {						\
+		    /* Split 4-node. */					\
+		    rbtn_black_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, right);		\
+		    rbtn_red_set(a_type, a_field, cnode);		\
+		} else {						\
+		    /* Lean left. */					\
+		    a_type *tnode;					\
+		    bool tred = rbtn_red_get(a_type, a_field, cnode);	\
+		    rbtn_rotate_left(a_type, a_field, cnode, tnode);	\
+		    rbtn_color_set(a_type, a_field, tnode, tred);	\
+		    rbtn_red_set(a_type, a_field, cnode);		\
+		    cnode = tnode;					\
+		}							\
+	    } else {							\
+		return;							\
+	    }								\
+	}								\
+	pathp->node = cnode;						\
+    }									\
+    /* Set root, and make it black. */					\
+    rbtree->rbt_root = path->node;					\
+    rbtn_black_set(a_type, a_field, rbtree->rbt_root);			\
+}									\
+a_attr void								\
+a_prefix##remove(a_rbt_type *rbtree, a_type *node) {			\
+    struct {								\
+	a_type *node;							\
+	int cmp;							\
+    } *pathp, *nodep, path[sizeof(void *) << 4];			\
+    /* Wind. */								\
+    nodep = NULL; /* Silence compiler warning. */			\
+    path->node = rbtree->rbt_root;					\
+    for (pathp = path; pathp->node != NULL; pathp++) {			\
+	int cmp = pathp->cmp = a_cmp(node, pathp->node);		\
+	if (cmp < 0) {							\
+	    pathp[1].node = rbtn_left_get(a_type, a_field,		\
+	      pathp->node);						\
+	} else {							\
+	    pathp[1].node = rbtn_right_get(a_type, a_field,		\
+	      pathp->node);						\
+	    if (cmp == 0) {						\
+	        /* Find node's successor, in preparation for swap. */	\
+		pathp->cmp = 1;						\
+		nodep = pathp;						\
+		for (pathp++; pathp->node != NULL; pathp++) {		\
+		    pathp->cmp = -1;					\
+		    pathp[1].node = rbtn_left_get(a_type, a_field,	\
+		      pathp->node);					\
+		}							\
+		break;							\
+	    }								\
+	}								\
+    }									\
+    assert(nodep->node == node);					\
+    pathp--;								\
+    if (pathp->node != node) {						\
+	/* Swap node with its successor. */				\
+	bool tred = rbtn_red_get(a_type, a_field, pathp->node);		\
+	rbtn_color_set(a_type, a_field, pathp->node,			\
+	  rbtn_red_get(a_type, a_field, node));				\
+	rbtn_left_set(a_type, a_field, pathp->node,			\
+	  rbtn_left_get(a_type, a_field, node));			\
+	/* If node's successor is its right child, the following code */\
+	/* will do the wrong thing for the right child pointer.       */\
+	/* However, it doesn't matter, because the pointer will be    */\
+	/* properly set when the successor is pruned.                 */\
+	rbtn_right_set(a_type, a_field, pathp->node,			\
+	  rbtn_right_get(a_type, a_field, node));			\
+	rbtn_color_set(a_type, a_field, node, tred);			\
+	/* The pruned leaf node's child pointers are never accessed   */\
+	/* again, so don't bother setting them to nil.                */\
+	nodep->node = pathp->node;					\
+	pathp->node = node;						\
+	if (nodep == path) {						\
+	    rbtree->rbt_root = nodep->node;				\
+	} else {							\
+	    if (nodep[-1].cmp < 0) {					\
+		rbtn_left_set(a_type, a_field, nodep[-1].node,		\
+		  nodep->node);						\
+	    } else {							\
+		rbtn_right_set(a_type, a_field, nodep[-1].node,		\
+		  nodep->node);						\
+	    }								\
+	}								\
+    } else {								\
+	a_type *left = rbtn_left_get(a_type, a_field, node);		\
+	if (left != NULL) {						\
+	    /* node has no successor, but it has a left child.        */\
+	    /* Splice node out, without losing the left child.        */\
+	    assert(!rbtn_red_get(a_type, a_field, node));		\
+	    assert(rbtn_red_get(a_type, a_field, left));		\
+	    rbtn_black_set(a_type, a_field, left);			\
+	    if (pathp == path) {					\
+		rbtree->rbt_root = left;				\
+	    } else {							\
+		if (pathp[-1].cmp < 0) {				\
+		    rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+		      left);						\
+		} else {						\
+		    rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+		      left);						\
+		}							\
+	    }								\
+	    return;							\
+	} else if (pathp == path) {					\
+	    /* The tree only contained one node. */			\
+	    rbtree->rbt_root = NULL;					\
+	    return;							\
+	}								\
+    }									\
+    if (rbtn_red_get(a_type, a_field, pathp->node)) {			\
+	/* Prune red node, which requires no fixup. */			\
+	assert(pathp[-1].cmp < 0);					\
+	rbtn_left_set(a_type, a_field, pathp[-1].node, NULL);		\
+	return;								\
+    }									\
+    /* The node to be pruned is black, so unwind until balance is     */\
+    /* restored.                                                      */\
+    pathp->node = NULL;							\
+    for (pathp--; (uintptr_t)pathp >= (uintptr_t)path; pathp--) {	\
+	assert(pathp->cmp != 0);					\
+	if (pathp->cmp < 0) {						\
+	    rbtn_left_set(a_type, a_field, pathp->node,			\
+	      pathp[1].node);						\
+	    if (rbtn_red_get(a_type, a_field, pathp->node)) {		\
+		a_type *right = rbtn_right_get(a_type, a_field,		\
+		  pathp->node);						\
+		a_type *rightleft = rbtn_left_get(a_type, a_field,	\
+		  right);						\
+		a_type *tnode;						\
+		if (rightleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  rightleft)) {						\
+		    /* In the following diagrams, ||, //, and \\      */\
+		    /* indicate the path to the removed node.         */\
+		    /*                                                */\
+		    /*      ||                                        */\
+		    /*    pathp(r)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (r)                                   */\
+		    /*                                                */\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    rbtn_rotate_right(a_type, a_field, right, tnode);	\
+		    rbtn_right_set(a_type, a_field, pathp->node, tnode);\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(r)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (b)                                   */\
+		    /*                                                */\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		}							\
+		/* Balance restored, but rotation modified subtree    */\
+		/* root.                                              */\
+		assert((uintptr_t)pathp > (uintptr_t)path);		\
+		if (pathp[-1].cmp < 0) {				\
+		    rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+		      tnode);						\
+		} else {						\
+		    rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+		      tnode);						\
+		}							\
+		return;							\
+	    } else {							\
+		a_type *right = rbtn_right_get(a_type, a_field,		\
+		  pathp->node);						\
+		a_type *rightleft = rbtn_left_get(a_type, a_field,	\
+		  right);						\
+		if (rightleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  rightleft)) {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (r)                                   */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, rightleft);		\
+		    rbtn_rotate_right(a_type, a_field, right, tnode);	\
+		    rbtn_right_set(a_type, a_field, pathp->node, tnode);\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root, which may actually be the tree   */\
+		    /* root.                                          */\
+		    if (pathp == path) {				\
+			/* Set root. */					\
+			rbtree->rbt_root = tnode;			\
+		    } else {						\
+			if (pathp[-1].cmp < 0) {			\
+			    rbtn_left_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			} else {					\
+			    rbtn_right_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			}						\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*  //        \                                   */\
+		    /* (b)        (b)                                 */\
+		    /*           /                                    */\
+		    /*          (b)                                   */\
+		    a_type *tnode;					\
+		    rbtn_red_set(a_type, a_field, pathp->node);		\
+		    rbtn_rotate_left(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    pathp->node = tnode;				\
+		}							\
+	    }								\
+	} else {							\
+	    a_type *left;						\
+	    rbtn_right_set(a_type, a_field, pathp->node,		\
+	      pathp[1].node);						\
+	    left = rbtn_left_get(a_type, a_field, pathp->node);		\
+	    if (rbtn_red_get(a_type, a_field, left)) {			\
+		a_type *tnode;						\
+		a_type *leftright = rbtn_right_get(a_type, a_field,	\
+		  left);						\
+		a_type *leftrightleft = rbtn_left_get(a_type, a_field,	\
+		  leftright);						\
+		if (leftrightleft != NULL && rbtn_red_get(a_type,	\
+		  a_field, leftrightleft)) {				\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*   /        \\                                  */\
+		    /* (r)        (b)                                 */\
+		    /*   \                                            */\
+		    /*   (b)                                          */\
+		    /*   /                                            */\
+		    /* (r)                                            */\
+		    a_type *unode;					\
+		    rbtn_black_set(a_type, a_field, leftrightleft);	\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      unode);						\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    rbtn_right_set(a_type, a_field, unode, tnode);	\
+		    rbtn_rotate_left(a_type, a_field, unode, tnode);	\
+		} else {						\
+		    /*      ||                                        */\
+		    /*    pathp(b)                                    */\
+		    /*   /        \\                                  */\
+		    /* (r)        (b)                                 */\
+		    /*   \                                            */\
+		    /*   (b)                                          */\
+		    /*   /                                            */\
+		    /* (b)                                            */\
+		    assert(leftright != NULL);				\
+		    rbtn_red_set(a_type, a_field, leftright);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    rbtn_black_set(a_type, a_field, tnode);		\
+		}							\
+		/* Balance restored, but rotation modified subtree    */\
+		/* root, which may actually be the tree root.         */\
+		if (pathp == path) {					\
+		    /* Set root. */					\
+		    rbtree->rbt_root = tnode;				\
+		} else {						\
+		    if (pathp[-1].cmp < 0) {				\
+			rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    } else {						\
+			rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    }							\
+		}							\
+		return;							\
+	    } else if (rbtn_red_get(a_type, a_field, pathp->node)) {	\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (leftleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  leftleft)) {						\
+		    /*        ||                                      */\
+		    /*      pathp(r)                                  */\
+		    /*     /        \\                                */\
+		    /*   (b)        (b)                               */\
+		    /*   /                                            */\
+		    /* (r)                                            */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    rbtn_red_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root.                                  */\
+		    assert((uintptr_t)pathp > (uintptr_t)path);		\
+		    if (pathp[-1].cmp < 0) {				\
+			rbtn_left_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    } else {						\
+			rbtn_right_set(a_type, a_field, pathp[-1].node,	\
+			  tnode);					\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*        ||                                      */\
+		    /*      pathp(r)                                  */\
+		    /*     /        \\                                */\
+		    /*   (b)        (b)                               */\
+		    /*   /                                            */\
+		    /* (b)                                            */\
+		    rbtn_red_set(a_type, a_field, left);		\
+		    rbtn_black_set(a_type, a_field, pathp->node);	\
+		    /* Balance restored. */				\
+		    return;						\
+		}							\
+	    } else {							\
+		a_type *leftleft = rbtn_left_get(a_type, a_field, left);\
+		if (leftleft != NULL && rbtn_red_get(a_type, a_field,	\
+		  leftleft)) {						\
+		    /*               ||                               */\
+		    /*             pathp(b)                           */\
+		    /*            /        \\                         */\
+		    /*          (b)        (b)                        */\
+		    /*          /                                     */\
+		    /*        (r)                                     */\
+		    a_type *tnode;					\
+		    rbtn_black_set(a_type, a_field, leftleft);		\
+		    rbtn_rotate_right(a_type, a_field, pathp->node,	\
+		      tnode);						\
+		    /* Balance restored, but rotation modified        */\
+		    /* subtree root, which may actually be the tree   */\
+		    /* root.                                          */\
+		    if (pathp == path) {				\
+			/* Set root. */					\
+			rbtree->rbt_root = tnode;			\
+		    } else {						\
+			if (pathp[-1].cmp < 0) {			\
+			    rbtn_left_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			} else {					\
+			    rbtn_right_set(a_type, a_field,		\
+			      pathp[-1].node, tnode);			\
+			}						\
+		    }							\
+		    return;						\
+		} else {						\
+		    /*               ||                               */\
+		    /*             pathp(b)                           */\
+		    /*            /        \\                         */\
+		    /*          (b)        (b)                        */\
+		    /*          /                                     */\
+		    /*        (b)                                     */\
+		    rbtn_red_set(a_type, a_field, left);		\
+		}							\
+	    }								\
+	}								\
+    }									\
+    /* Set root. */							\
+    rbtree->rbt_root = path->node;					\
+    assert(!rbtn_red_get(a_type, a_field, rbtree->rbt_root));		\
+}									\
+a_attr a_type *								\
+a_prefix##iter_recurse(a_rbt_type *rbtree, a_type *node,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    if (node == NULL) {							\
+	return NULL;							\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = a_prefix##iter_recurse(rbtree, rbtn_left_get(a_type,	\
+	  a_field, node), cb, arg)) != NULL || (ret = cb(rbtree, node,	\
+	  arg)) != NULL) {						\
+	    return ret;							\
+	}								\
+	return a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg);					\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##iter_start(a_rbt_type *rbtree, a_type *start, a_type *node,	\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    int cmp = a_cmp(start, node);					\
+    if (cmp < 0) {							\
+	a_type *ret;							\
+	if ((ret = a_prefix##iter_start(rbtree, start,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg)) != NULL ||	\
+	  (ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg);					\
+    } else if (cmp > 0) {						\
+	return a_prefix##iter_start(rbtree, start,			\
+	  rbtn_right_get(a_type, a_field, node), cb, arg);		\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##iter_recurse(rbtree, rbtn_right_get(a_type,	\
+	  a_field, node), cb, arg);					\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##iter(a_rbt_type *rbtree, a_type *start, a_type *(*cb)(	\
+  a_rbt_type *, a_type *, void *), void *arg) {				\
+    a_type *ret;							\
+    if (start != NULL) {						\
+	ret = a_prefix##iter_start(rbtree, start, rbtree->rbt_root,	\
+	  cb, arg);							\
+    } else {								\
+	ret = a_prefix##iter_recurse(rbtree, rbtree->rbt_root, cb, arg);\
+    }									\
+    return ret;								\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter_recurse(a_rbt_type *rbtree, a_type *node,	\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    if (node == NULL) {							\
+	return NULL;							\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = a_prefix##reverse_iter_recurse(rbtree,		\
+	  rbtn_right_get(a_type, a_field, node), cb, arg)) != NULL ||	\
+	  (ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter_start(a_rbt_type *rbtree, a_type *start,		\
+  a_type *node, a_type *(*cb)(a_rbt_type *, a_type *, void *),		\
+  void *arg) {								\
+    int cmp = a_cmp(start, node);					\
+    if (cmp > 0) {							\
+	a_type *ret;							\
+	if ((ret = a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtn_right_get(a_type, a_field, node), cb, arg)) != NULL ||	\
+	  (ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    } else if (cmp < 0) {						\
+	return a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    } else {								\
+	a_type *ret;							\
+	if ((ret = cb(rbtree, node, arg)) != NULL) {			\
+	    return ret;							\
+	}								\
+	return a_prefix##reverse_iter_recurse(rbtree,			\
+	  rbtn_left_get(a_type, a_field, node), cb, arg);		\
+    }									\
+}									\
+a_attr a_type *								\
+a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
+  a_type *(*cb)(a_rbt_type *, a_type *, void *), void *arg) {		\
+    a_type *ret;							\
+    if (start != NULL) {						\
+	ret = a_prefix##reverse_iter_start(rbtree, start,		\
+	  rbtree->rbt_root, cb, arg);					\
+    } else {								\
+	ret = a_prefix##reverse_iter_recurse(rbtree, rbtree->rbt_root,	\
+	  cb, arg);							\
+    }									\
+    return ret;								\
+}									\
+a_attr void								\
+a_prefix##destroy_recurse(a_rbt_type *rbtree, a_type *node, void (*cb)(	\
+  a_type *, void *), void *arg) {					\
+    if (node == NULL) {							\
+	return;								\
+    }									\
+    a_prefix##destroy_recurse(rbtree, rbtn_left_get(a_type, a_field,	\
+      node), cb, arg);							\
+    rbtn_left_set(a_type, a_field, (node), NULL);			\
+    a_prefix##destroy_recurse(rbtree, rbtn_right_get(a_type, a_field,	\
+      node), cb, arg);							\
+    rbtn_right_set(a_type, a_field, (node), NULL);			\
+    if (cb) {								\
+	cb(node, arg);							\
+    }									\
+}									\
+a_attr void								\
+a_prefix##destroy(a_rbt_type *rbtree, void (*cb)(a_type *, void *),	\
+  void *arg) {								\
+    a_prefix##destroy_recurse(rbtree, rbtree->rbt_root, cb, arg);	\
+    rbtree->rbt_root = NULL;						\
+}
+
+#endif /* RB_H_ */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/rtree.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/rtree.h
new file mode 100644
index 0000000..16ccbeb
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/rtree.h
@@ -0,0 +1,528 @@
+#ifndef JEMALLOC_INTERNAL_RTREE_H
+#define JEMALLOC_INTERNAL_RTREE_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree_tsd.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/tsd.h"
+
+/*
+ * This radix tree implementation is tailored to the singular purpose of
+ * associating metadata with extents that are currently owned by jemalloc.
+ *
+ *******************************************************************************
+ */
+
+/* Number of high insignificant bits. */
+#define RTREE_NHIB ((1U << (LG_SIZEOF_PTR+3)) - LG_VADDR)
+/* Number of low insigificant bits. */
+#define RTREE_NLIB LG_PAGE
+/* Number of significant bits. */
+#define RTREE_NSB (LG_VADDR - RTREE_NLIB)
+/* Number of levels in radix tree. */
+#if RTREE_NSB <= 10
+#  define RTREE_HEIGHT 1
+#elif RTREE_NSB <= 36
+#  define RTREE_HEIGHT 2
+#elif RTREE_NSB <= 52
+#  define RTREE_HEIGHT 3
+#else
+#  error Unsupported number of significant virtual address bits
+#endif
+/* Use compact leaf representation if virtual address encoding allows. */
+#if RTREE_NHIB >= LG_CEIL(SC_NSIZES)
+#  define RTREE_LEAF_COMPACT
+#endif
+
+/* Needed for initialization only. */
+#define RTREE_LEAFKEY_INVALID ((uintptr_t)1)
+
+typedef struct rtree_node_elm_s rtree_node_elm_t;
+struct rtree_node_elm_s {
+	atomic_p_t	child; /* (rtree_{node,leaf}_elm_t *) */
+};
+
+struct rtree_leaf_elm_s {
+#ifdef RTREE_LEAF_COMPACT
+	/*
+	 * Single pointer-width field containing all three leaf element fields.
+	 * For example, on a 64-bit x64 system with 48 significant virtual
+	 * memory address bits, the index, extent, and slab fields are packed as
+	 * such:
+	 *
+	 * x: index
+	 * e: extent
+	 * b: slab
+	 *
+	 *   00000000 xxxxxxxx eeeeeeee [...] eeeeeeee eeee000b
+	 */
+	atomic_p_t	le_bits;
+#else
+	atomic_p_t	le_extent; /* (extent_t *) */
+	atomic_u_t	le_szind; /* (szind_t) */
+	atomic_b_t	le_slab; /* (bool) */
+#endif
+};
+
+typedef struct rtree_level_s rtree_level_t;
+struct rtree_level_s {
+	/* Number of key bits distinguished by this level. */
+	unsigned		bits;
+	/*
+	 * Cumulative number of key bits distinguished by traversing to
+	 * corresponding tree level.
+	 */
+	unsigned		cumbits;
+};
+
+typedef struct rtree_s rtree_t;
+struct rtree_s {
+	malloc_mutex_t		init_lock;
+	/* Number of elements based on rtree_levels[0].bits. */
+#if RTREE_HEIGHT > 1
+	rtree_node_elm_t	root[1U << (RTREE_NSB/RTREE_HEIGHT)];
+#else
+	rtree_leaf_elm_t	root[1U << (RTREE_NSB/RTREE_HEIGHT)];
+#endif
+};
+
+/*
+ * Split the bits into one to three partitions depending on number of
+ * significant bits.  It the number of bits does not divide evenly into the
+ * number of levels, place one remainder bit per level starting at the leaf
+ * level.
+ */
+static const rtree_level_t rtree_levels[] = {
+#if RTREE_HEIGHT == 1
+	{RTREE_NSB, RTREE_NHIB + RTREE_NSB}
+#elif RTREE_HEIGHT == 2
+	{RTREE_NSB/2, RTREE_NHIB + RTREE_NSB/2},
+	{RTREE_NSB/2 + RTREE_NSB%2, RTREE_NHIB + RTREE_NSB}
+#elif RTREE_HEIGHT == 3
+	{RTREE_NSB/3, RTREE_NHIB + RTREE_NSB/3},
+	{RTREE_NSB/3 + RTREE_NSB%3/2,
+	    RTREE_NHIB + RTREE_NSB/3*2 + RTREE_NSB%3/2},
+	{RTREE_NSB/3 + RTREE_NSB%3 - RTREE_NSB%3/2, RTREE_NHIB + RTREE_NSB}
+#else
+#  error Unsupported rtree height
+#endif
+};
+
+bool rtree_new(rtree_t *rtree, bool zeroed);
+
+typedef rtree_node_elm_t *(rtree_node_alloc_t)(tsdn_t *, rtree_t *, size_t);
+extern rtree_node_alloc_t *JET_MUTABLE rtree_node_alloc;
+
+typedef rtree_leaf_elm_t *(rtree_leaf_alloc_t)(tsdn_t *, rtree_t *, size_t);
+extern rtree_leaf_alloc_t *JET_MUTABLE rtree_leaf_alloc;
+
+typedef void (rtree_node_dalloc_t)(tsdn_t *, rtree_t *, rtree_node_elm_t *);
+extern rtree_node_dalloc_t *JET_MUTABLE rtree_node_dalloc;
+
+typedef void (rtree_leaf_dalloc_t)(tsdn_t *, rtree_t *, rtree_leaf_elm_t *);
+extern rtree_leaf_dalloc_t *JET_MUTABLE rtree_leaf_dalloc;
+#ifdef JEMALLOC_JET
+void rtree_delete(tsdn_t *tsdn, rtree_t *rtree);
+#endif
+rtree_leaf_elm_t *rtree_leaf_elm_lookup_hard(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_ctx_t *rtree_ctx, uintptr_t key, bool dependent, bool init_missing);
+
+JEMALLOC_ALWAYS_INLINE uintptr_t
+rtree_leafkey(uintptr_t key) {
+	unsigned ptrbits = ZU(1) << (LG_SIZEOF_PTR+3);
+	unsigned cumbits = (rtree_levels[RTREE_HEIGHT-1].cumbits -
+	    rtree_levels[RTREE_HEIGHT-1].bits);
+	unsigned maskbits = ptrbits - cumbits;
+	uintptr_t mask = ~((ZU(1) << maskbits) - 1);
+	return (key & mask);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+rtree_cache_direct_map(uintptr_t key) {
+	unsigned ptrbits = ZU(1) << (LG_SIZEOF_PTR+3);
+	unsigned cumbits = (rtree_levels[RTREE_HEIGHT-1].cumbits -
+	    rtree_levels[RTREE_HEIGHT-1].bits);
+	unsigned maskbits = ptrbits - cumbits;
+	return (size_t)((key >> maskbits) & (RTREE_CTX_NCACHE - 1));
+}
+
+JEMALLOC_ALWAYS_INLINE uintptr_t
+rtree_subkey(uintptr_t key, unsigned level) {
+	unsigned ptrbits = ZU(1) << (LG_SIZEOF_PTR+3);
+	unsigned cumbits = rtree_levels[level].cumbits;
+	unsigned shiftbits = ptrbits - cumbits;
+	unsigned maskbits = rtree_levels[level].bits;
+	uintptr_t mask = (ZU(1) << maskbits) - 1;
+	return ((key >> shiftbits) & mask);
+}
+
+/*
+ * Atomic getters.
+ *
+ * dependent: Reading a value on behalf of a pointer to a valid allocation
+ *            is guaranteed to be a clean read even without synchronization,
+ *            because the rtree update became visible in memory before the
+ *            pointer came into existence.
+ * !dependent: An arbitrary read, e.g. on behalf of ivsalloc(), may not be
+ *             dependent on a previous rtree write, which means a stale read
+ *             could result if synchronization were omitted here.
+ */
+#  ifdef RTREE_LEAF_COMPACT
+JEMALLOC_ALWAYS_INLINE uintptr_t
+rtree_leaf_elm_bits_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+	return (uintptr_t)atomic_load_p(&elm->le_bits, dependent
+	    ? ATOMIC_RELAXED : ATOMIC_ACQUIRE);
+}
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+rtree_leaf_elm_bits_extent_get(uintptr_t bits) {
+#    ifdef __aarch64__
+	/*
+	 * aarch64 doesn't sign extend the highest virtual address bit to set
+	 * the higher ones.  Instead, the high bits gets zeroed.
+	 */
+	uintptr_t high_bit_mask = ((uintptr_t)1 << LG_VADDR) - 1;
+	/* Mask off the slab bit. */
+	uintptr_t low_bit_mask = ~(uintptr_t)1;
+	uintptr_t mask = high_bit_mask & low_bit_mask;
+	return (extent_t *)(bits & mask);
+#    else
+	/* Restore sign-extended high bits, mask slab bit. */
+	return (extent_t *)((uintptr_t)((intptr_t)(bits << RTREE_NHIB) >>
+	    RTREE_NHIB) & ~((uintptr_t)0x1));
+#    endif
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+rtree_leaf_elm_bits_szind_get(uintptr_t bits) {
+	return (szind_t)(bits >> LG_VADDR);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+rtree_leaf_elm_bits_slab_get(uintptr_t bits) {
+	return (bool)(bits & (uintptr_t)0x1);
+}
+
+#  endif
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+rtree_leaf_elm_extent_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	return rtree_leaf_elm_bits_extent_get(bits);
+#else
+	extent_t *extent = (extent_t *)atomic_load_p(&elm->le_extent, dependent
+	    ? ATOMIC_RELAXED : ATOMIC_ACQUIRE);
+	return extent;
+#endif
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+rtree_leaf_elm_szind_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	return rtree_leaf_elm_bits_szind_get(bits);
+#else
+	return (szind_t)atomic_load_u(&elm->le_szind, dependent ? ATOMIC_RELAXED
+	    : ATOMIC_ACQUIRE);
+#endif
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+rtree_leaf_elm_slab_read(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool dependent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	return rtree_leaf_elm_bits_slab_get(bits);
+#else
+	return atomic_load_b(&elm->le_slab, dependent ? ATOMIC_RELAXED :
+	    ATOMIC_ACQUIRE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_extent_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, extent_t *extent) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t old_bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, true);
+	uintptr_t bits = ((uintptr_t)rtree_leaf_elm_bits_szind_get(old_bits) <<
+	    LG_VADDR) | ((uintptr_t)extent & (((uintptr_t)0x1 << LG_VADDR) - 1))
+	    | ((uintptr_t)rtree_leaf_elm_bits_slab_get(old_bits));
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	atomic_store_p(&elm->le_extent, extent, ATOMIC_RELEASE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_szind_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, szind_t szind) {
+	assert(szind <= SC_NSIZES);
+
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t old_bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm,
+	    true);
+	uintptr_t bits = ((uintptr_t)szind << LG_VADDR) |
+	    ((uintptr_t)rtree_leaf_elm_bits_extent_get(old_bits) &
+	    (((uintptr_t)0x1 << LG_VADDR) - 1)) |
+	    ((uintptr_t)rtree_leaf_elm_bits_slab_get(old_bits));
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	atomic_store_u(&elm->le_szind, szind, ATOMIC_RELEASE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_slab_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, bool slab) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t old_bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm,
+	    true);
+	uintptr_t bits = ((uintptr_t)rtree_leaf_elm_bits_szind_get(old_bits) <<
+	    LG_VADDR) | ((uintptr_t)rtree_leaf_elm_bits_extent_get(old_bits) &
+	    (((uintptr_t)0x1 << LG_VADDR) - 1)) | ((uintptr_t)slab);
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	atomic_store_b(&elm->le_slab, slab, ATOMIC_RELEASE);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_write(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, extent_t *extent, szind_t szind, bool slab) {
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = ((uintptr_t)szind << LG_VADDR) |
+	    ((uintptr_t)extent & (((uintptr_t)0x1 << LG_VADDR) - 1)) |
+	    ((uintptr_t)slab);
+	atomic_store_p(&elm->le_bits, (void *)bits, ATOMIC_RELEASE);
+#else
+	rtree_leaf_elm_slab_write(tsdn, rtree, elm, slab);
+	rtree_leaf_elm_szind_write(tsdn, rtree, elm, szind);
+	/*
+	 * Write extent last, since the element is atomically considered valid
+	 * as soon as the extent field is non-NULL.
+	 */
+	rtree_leaf_elm_extent_write(tsdn, rtree, elm, extent);
+#endif
+}
+
+static inline void
+rtree_leaf_elm_szind_slab_update(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *elm, szind_t szind, bool slab) {
+	assert(!slab || szind < SC_NBINS);
+
+	/*
+	 * The caller implicitly assures that it is the only writer to the szind
+	 * and slab fields, and that the extent field cannot currently change.
+	 */
+	rtree_leaf_elm_slab_write(tsdn, rtree, elm, slab);
+	rtree_leaf_elm_szind_write(tsdn, rtree, elm, szind);
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_leaf_elm_t *
+rtree_leaf_elm_lookup(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, bool init_missing) {
+	assert(key != 0);
+	assert(!dependent || !init_missing);
+
+	size_t slot = rtree_cache_direct_map(key);
+	uintptr_t leafkey = rtree_leafkey(key);
+	assert(leafkey != RTREE_LEAFKEY_INVALID);
+
+	/* Fast path: L1 direct mapped cache. */
+	if (likely(rtree_ctx->cache[slot].leafkey == leafkey)) {
+		rtree_leaf_elm_t *leaf = rtree_ctx->cache[slot].leaf;
+		assert(leaf != NULL);
+		uintptr_t subkey = rtree_subkey(key, RTREE_HEIGHT-1);
+		return &leaf[subkey];
+	}
+	/*
+	 * Search the L2 LRU cache.  On hit, swap the matching element into the
+	 * slot in L1 cache, and move the position in L2 up by 1.
+	 */
+#define RTREE_CACHE_CHECK_L2(i) do {					\
+	if (likely(rtree_ctx->l2_cache[i].leafkey == leafkey)) {	\
+		rtree_leaf_elm_t *leaf = rtree_ctx->l2_cache[i].leaf;	\
+		assert(leaf != NULL);					\
+		if (i > 0) {						\
+			/* Bubble up by one. */				\
+			rtree_ctx->l2_cache[i].leafkey =		\
+				rtree_ctx->l2_cache[i - 1].leafkey;	\
+			rtree_ctx->l2_cache[i].leaf =			\
+				rtree_ctx->l2_cache[i - 1].leaf;	\
+			rtree_ctx->l2_cache[i - 1].leafkey =		\
+			    rtree_ctx->cache[slot].leafkey;		\
+			rtree_ctx->l2_cache[i - 1].leaf =		\
+			    rtree_ctx->cache[slot].leaf;		\
+		} else {						\
+			rtree_ctx->l2_cache[0].leafkey =		\
+			    rtree_ctx->cache[slot].leafkey;		\
+			rtree_ctx->l2_cache[0].leaf =			\
+			    rtree_ctx->cache[slot].leaf;		\
+		}							\
+		rtree_ctx->cache[slot].leafkey = leafkey;		\
+		rtree_ctx->cache[slot].leaf = leaf;			\
+		uintptr_t subkey = rtree_subkey(key, RTREE_HEIGHT-1);	\
+		return &leaf[subkey];					\
+	}								\
+} while (0)
+	/* Check the first cache entry. */
+	RTREE_CACHE_CHECK_L2(0);
+	/* Search the remaining cache elements. */
+	for (unsigned i = 1; i < RTREE_CTX_NCACHE_L2; i++) {
+		RTREE_CACHE_CHECK_L2(i);
+	}
+#undef RTREE_CACHE_CHECK_L2
+
+	return rtree_leaf_elm_lookup_hard(tsdn, rtree, rtree_ctx, key,
+	    dependent, init_missing);
+}
+
+static inline bool
+rtree_write(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx, uintptr_t key,
+    extent_t *extent, szind_t szind, bool slab) {
+	/* Use rtree_clear() to set the extent to NULL. */
+	assert(extent != NULL);
+
+	rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, rtree, rtree_ctx,
+	    key, false, true);
+	if (elm == NULL) {
+		return true;
+	}
+
+	assert(rtree_leaf_elm_extent_read(tsdn, rtree, elm, false) == NULL);
+	rtree_leaf_elm_write(tsdn, rtree, elm, extent, szind, slab);
+
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_leaf_elm_t *
+rtree_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx, uintptr_t key,
+    bool dependent) {
+	rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, rtree, rtree_ctx,
+	    key, dependent, false);
+	if (!dependent && elm == NULL) {
+		return NULL;
+	}
+	assert(elm != NULL);
+	return elm;
+}
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+rtree_extent_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return NULL;
+	}
+	return rtree_leaf_elm_extent_read(tsdn, rtree, elm, dependent);
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+rtree_szind_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return SC_NSIZES;
+	}
+	return rtree_leaf_elm_szind_read(tsdn, rtree, elm, dependent);
+}
+
+/*
+ * rtree_slab_read() is intentionally omitted because slab is always read in
+ * conjunction with szind, which makes rtree_szind_slab_read() a better choice.
+ */
+
+JEMALLOC_ALWAYS_INLINE bool
+rtree_extent_szind_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, extent_t **r_extent, szind_t *r_szind) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return true;
+	}
+	*r_extent = rtree_leaf_elm_extent_read(tsdn, rtree, elm, dependent);
+	*r_szind = rtree_leaf_elm_szind_read(tsdn, rtree, elm, dependent);
+	return false;
+}
+
+/*
+ * Try to read szind_slab from the L1 cache.  Returns true on a hit,
+ * and fills in r_szind and r_slab.  Otherwise returns false.
+ *
+ * Key is allowed to be NULL in order to save an extra branch on the
+ * fastpath.  returns false in this case.
+ */
+JEMALLOC_ALWAYS_INLINE bool
+rtree_szind_slab_read_fast(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+			    uintptr_t key, szind_t *r_szind, bool *r_slab) {
+	rtree_leaf_elm_t *elm;
+
+	size_t slot = rtree_cache_direct_map(key);
+	uintptr_t leafkey = rtree_leafkey(key);
+	assert(leafkey != RTREE_LEAFKEY_INVALID);
+
+	if (likely(rtree_ctx->cache[slot].leafkey == leafkey)) {
+		rtree_leaf_elm_t *leaf = rtree_ctx->cache[slot].leaf;
+		assert(leaf != NULL);
+		uintptr_t subkey = rtree_subkey(key, RTREE_HEIGHT-1);
+		elm = &leaf[subkey];
+
+#ifdef RTREE_LEAF_COMPACT
+		uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree,
+							  elm, true);
+		*r_szind = rtree_leaf_elm_bits_szind_get(bits);
+		*r_slab = rtree_leaf_elm_bits_slab_get(bits);
+#else
+		*r_szind = rtree_leaf_elm_szind_read(tsdn, rtree, elm, true);
+		*r_slab = rtree_leaf_elm_slab_read(tsdn, rtree, elm, true);
+#endif
+		return true;
+	} else {
+		return false;
+	}
+}
+JEMALLOC_ALWAYS_INLINE bool
+rtree_szind_slab_read(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, szind_t *r_szind, bool *r_slab) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key,
+	    dependent);
+	if (!dependent && elm == NULL) {
+		return true;
+	}
+#ifdef RTREE_LEAF_COMPACT
+	uintptr_t bits = rtree_leaf_elm_bits_read(tsdn, rtree, elm, dependent);
+	*r_szind = rtree_leaf_elm_bits_szind_get(bits);
+	*r_slab = rtree_leaf_elm_bits_slab_get(bits);
+#else
+	*r_szind = rtree_leaf_elm_szind_read(tsdn, rtree, elm, dependent);
+	*r_slab = rtree_leaf_elm_slab_read(tsdn, rtree, elm, dependent);
+#endif
+	return false;
+}
+
+static inline void
+rtree_szind_slab_update(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, szind_t szind, bool slab) {
+	assert(!slab || szind < SC_NBINS);
+
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key, true);
+	rtree_leaf_elm_szind_slab_update(tsdn, rtree, elm, szind, slab);
+}
+
+static inline void
+rtree_clear(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key) {
+	rtree_leaf_elm_t *elm = rtree_read(tsdn, rtree, rtree_ctx, key, true);
+	assert(rtree_leaf_elm_extent_read(tsdn, rtree, elm, false) !=
+	    NULL);
+	rtree_leaf_elm_write(tsdn, rtree, elm, NULL, SC_NSIZES, false);
+}
+
+#endif /* JEMALLOC_INTERNAL_RTREE_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/rtree_tsd.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/rtree_tsd.h
new file mode 100644
index 0000000..562e292
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/rtree_tsd.h
@@ -0,0 +1,50 @@
+#ifndef JEMALLOC_INTERNAL_RTREE_CTX_H
+#define JEMALLOC_INTERNAL_RTREE_CTX_H
+
+/*
+ * Number of leafkey/leaf pairs to cache in L1 and L2 level respectively.  Each
+ * entry supports an entire leaf, so the cache hit rate is typically high even
+ * with a small number of entries.  In rare cases extent activity will straddle
+ * the boundary between two leaf nodes.  Furthermore, an arena may use a
+ * combination of dss and mmap.  Note that as memory usage grows past the amount
+ * that this cache can directly cover, the cache will become less effective if
+ * locality of reference is low, but the consequence is merely cache misses
+ * while traversing the tree nodes.
+ *
+ * The L1 direct mapped cache offers consistent and low cost on cache hit.
+ * However collision could affect hit rate negatively.  This is resolved by
+ * combining with a L2 LRU cache, which requires linear search and re-ordering
+ * on access but suffers no collision.  Note that, the cache will itself suffer
+ * cache misses if made overly large, plus the cost of linear search in the LRU
+ * cache.
+ */
+#define RTREE_CTX_LG_NCACHE 4
+#define RTREE_CTX_NCACHE (1 << RTREE_CTX_LG_NCACHE)
+#define RTREE_CTX_NCACHE_L2 8
+
+/*
+ * Zero initializer required for tsd initialization only.  Proper initialization
+ * done via rtree_ctx_data_init().
+ */
+#define RTREE_CTX_ZERO_INITIALIZER {{{0, 0}}, {{0, 0}}}
+
+
+typedef struct rtree_leaf_elm_s rtree_leaf_elm_t;
+
+typedef struct rtree_ctx_cache_elm_s rtree_ctx_cache_elm_t;
+struct rtree_ctx_cache_elm_s {
+	uintptr_t		leafkey;
+	rtree_leaf_elm_t	*leaf;
+};
+
+typedef struct rtree_ctx_s rtree_ctx_t;
+struct rtree_ctx_s {
+	/* Direct mapped cache. */
+	rtree_ctx_cache_elm_t	cache[RTREE_CTX_NCACHE];
+	/* L2 LRU cache. */
+	rtree_ctx_cache_elm_t	l2_cache[RTREE_CTX_NCACHE_L2];
+};
+
+void rtree_ctx_data_init(rtree_ctx_t *ctx);
+
+#endif /* JEMALLOC_INTERNAL_RTREE_CTX_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/safety_check.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/safety_check.h
new file mode 100644
index 0000000..53339ac
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/safety_check.h
@@ -0,0 +1,26 @@
+#ifndef JEMALLOC_INTERNAL_SAFETY_CHECK_H
+#define JEMALLOC_INTERNAL_SAFETY_CHECK_H
+
+void safety_check_fail(const char *format, ...);
+/* Can set to NULL for a default. */
+void safety_check_set_abort(void (*abort_fn)());
+
+JEMALLOC_ALWAYS_INLINE void
+safety_check_set_redzone(void *ptr, size_t usize, size_t bumped_usize) {
+	assert(usize < bumped_usize);
+	for (size_t i = usize; i < bumped_usize && i < usize + 32; ++i) {
+		*((unsigned char *)ptr + i) = 0xBC;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+safety_check_verify_redzone(const void *ptr, size_t usize, size_t bumped_usize)
+{
+	for (size_t i = usize; i < bumped_usize && i < usize + 32; ++i) {
+		if (unlikely(*((unsigned char *)ptr + i) != 0xBC)) {
+			safety_check_fail("Use after free error\n");
+		}
+	}
+}
+
+#endif /*JEMALLOC_INTERNAL_SAFETY_CHECK_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/sc.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/sc.h
new file mode 100644
index 0000000..9a099d8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/sc.h
@@ -0,0 +1,333 @@
+#ifndef JEMALLOC_INTERNAL_SC_H
+#define JEMALLOC_INTERNAL_SC_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+
+/*
+ * Size class computations:
+ *
+ * These are a little tricky; we'll first start by describing how things
+ * generally work, and then describe some of the details.
+ *
+ * Ignore the first few size classes for a moment. We can then split all the
+ * remaining size classes into groups. The size classes in a group are spaced
+ * such that they cover allocation request sizes in a power-of-2 range. The
+ * power of two is called the base of the group, and the size classes in it
+ * satisfy allocations in the half-open range (base, base * 2]. There are
+ * SC_NGROUP size classes in each group, equally spaced in the range, so that
+ * each one covers allocations for base / SC_NGROUP possible allocation sizes.
+ * We call that value (base / SC_NGROUP) the delta of the group. Each size class
+ * is delta larger than the one before it (including the initial size class in a
+ * group, which is delta larger than base, the largest size class in the
+ * previous group).
+ * To make the math all work out nicely, we require that SC_NGROUP is a power of
+ * two, and define it in terms of SC_LG_NGROUP. We'll often talk in terms of
+ * lg_base and lg_delta. For each of these groups then, we have that
+ * lg_delta == lg_base - SC_LG_NGROUP.
+ * The size classes in a group with a given lg_base and lg_delta (which, recall,
+ * can be computed from lg_base for these groups) are therefore:
+ *   base + 1 * delta
+ *     which covers allocations in (base, base + 1 * delta]
+ *   base + 2 * delta
+ *     which covers allocations in (base + 1 * delta, base + 2 * delta].
+ *   base + 3 * delta
+ *     which covers allocations in (base + 2 * delta, base + 3 * delta].
+ *   ...
+ *   base + SC_NGROUP * delta ( == 2 * base)
+ *     which covers allocations in (base + (SC_NGROUP - 1) * delta, 2 * base].
+ * (Note that currently SC_NGROUP is always 4, so the "..." is empty in
+ * practice.)
+ * Note that the last size class in the group is the next power of two (after
+ * base), so that we've set up the induction correctly for the next group's
+ * selection of delta.
+ *
+ * Now, let's start considering the first few size classes. Two extra constants
+ * come into play here: LG_QUANTUM and SC_LG_TINY_MIN. LG_QUANTUM ensures
+ * correct platform alignment; all objects of size (1 << LG_QUANTUM) or larger
+ * are at least (1 << LG_QUANTUM) aligned; this can be used to ensure that we
+ * never return improperly aligned memory, by making (1 << LG_QUANTUM) equal the
+ * highest required alignment of a platform. For allocation sizes smaller than
+ * (1 << LG_QUANTUM) though, we can be more relaxed (since we don't support
+ * platforms with types with alignment larger than their size). To allow such
+ * allocations (without wasting space unnecessarily), we introduce tiny size
+ * classes; one per power of two, up until we hit the quantum size. There are
+ * therefore LG_QUANTUM - SC_LG_TINY_MIN such size classes.
+ *
+ * Next, we have a size class of size (1 << LG_QUANTUM).  This can't be the
+ * start of a group in the sense we described above (covering a power of two
+ * range) since, if we divided into it to pick a value of delta, we'd get a
+ * delta smaller than (1 << LG_QUANTUM) for sizes >= (1 << LG_QUANTUM), which
+ * is against the rules.
+ *
+ * The first base we can divide by SC_NGROUP while still being at least
+ * (1 << LG_QUANTUM) is SC_NGROUP * (1 << LG_QUANTUM). We can get there by
+ * having SC_NGROUP size classes, spaced (1 << LG_QUANTUM) apart. These size
+ * classes are:
+ *   1 * (1 << LG_QUANTUM)
+ *   2 * (1 << LG_QUANTUM)
+ *   3 * (1 << LG_QUANTUM)
+ *   ... (although, as above, this "..." is empty in practice)
+ *   SC_NGROUP * (1 << LG_QUANTUM).
+ *
+ * There are SC_NGROUP of these size classes, so we can regard it as a sort of
+ * pseudo-group, even though it spans multiple powers of 2, is divided
+ * differently, and both starts and ends on a power of 2 (as opposed to just
+ * ending). SC_NGROUP is itself a power of two, so the first group after the
+ * pseudo-group has the power-of-two base SC_NGROUP * (1 << LG_QUANTUM), for a
+ * lg_base of LG_QUANTUM + SC_LG_NGROUP. We can divide this base into SC_NGROUP
+ * sizes without violating our LG_QUANTUM requirements, so we can safely set
+ * lg_delta = lg_base - SC_LG_GROUP (== LG_QUANTUM).
+ *
+ * So, in order, the size classes are:
+ *
+ * Tiny size classes:
+ * - Count: LG_QUANTUM - SC_LG_TINY_MIN.
+ * - Sizes:
+ *     1 << SC_LG_TINY_MIN
+ *     1 << (SC_LG_TINY_MIN + 1)
+ *     1 << (SC_LG_TINY_MIN + 2)
+ *     ...
+ *     1 << (LG_QUANTUM - 1)
+ *
+ * Initial pseudo-group:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *     1 * (1 << LG_QUANTUM)
+ *     2 * (1 << LG_QUANTUM)
+ *     3 * (1 << LG_QUANTUM)
+ *     ...
+ *     SC_NGROUP * (1 << LG_QUANTUM)
+ *
+ * Regular group 0:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *   (relative to lg_base of LG_QUANTUM + SC_LG_NGROUP and lg_delta of
+ *   lg_base - SC_LG_NGROUP)
+ *     (1 << lg_base) + 1 * (1 << lg_delta)
+ *     (1 << lg_base) + 2 * (1 << lg_delta)
+ *     (1 << lg_base) + 3 * (1 << lg_delta)
+ *     ...
+ *     (1 << lg_base) + SC_NGROUP * (1 << lg_delta) [ == (1 << (lg_base + 1)) ]
+ *
+ * Regular group 1:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *   (relative to lg_base of LG_QUANTUM + SC_LG_NGROUP + 1 and lg_delta of
+ *   lg_base - SC_LG_NGROUP)
+ *     (1 << lg_base) + 1 * (1 << lg_delta)
+ *     (1 << lg_base) + 2 * (1 << lg_delta)
+ *     (1 << lg_base) + 3 * (1 << lg_delta)
+ *     ...
+ *     (1 << lg_base) + SC_NGROUP * (1 << lg_delta) [ == (1 << (lg_base + 1)) ]
+ *
+ * ...
+ *
+ * Regular group N:
+ * - Count: SC_NGROUP
+ * - Sizes:
+ *   (relative to lg_base of LG_QUANTUM + SC_LG_NGROUP + N and lg_delta of
+ *   lg_base - SC_LG_NGROUP)
+ *     (1 << lg_base) + 1 * (1 << lg_delta)
+ *     (1 << lg_base) + 2 * (1 << lg_delta)
+ *     (1 << lg_base) + 3 * (1 << lg_delta)
+ *     ...
+ *     (1 << lg_base) + SC_NGROUP * (1 << lg_delta) [ == (1 << (lg_base + 1)) ]
+ *
+ *
+ * Representation of metadata:
+ * To make the math easy, we'll mostly work in lg quantities. We record lg_base,
+ * lg_delta, and ndelta (i.e. number of deltas above the base) on a
+ * per-size-class basis, and maintain the invariant that, across all size
+ * classes, size == (1 << lg_base) + ndelta * (1 << lg_delta).
+ *
+ * For regular groups (i.e. those with lg_base >= LG_QUANTUM + SC_LG_NGROUP),
+ * lg_delta is lg_base - SC_LG_NGROUP, and ndelta goes from 1 to SC_NGROUP.
+ *
+ * For the initial tiny size classes (if any), lg_base is lg(size class size).
+ * lg_delta is lg_base for the first size class, and lg_base - 1 for all
+ * subsequent ones. ndelta is always 0.
+ *
+ * For the pseudo-group, if there are no tiny size classes, then we set
+ * lg_base == LG_QUANTUM, lg_delta == LG_QUANTUM, and have ndelta range from 0
+ * to SC_NGROUP - 1. (Note that delta == base, so base + (SC_NGROUP - 1) * delta
+ * is just SC_NGROUP * base, or (1 << (SC_LG_NGROUP + LG_QUANTUM)), so we do
+ * indeed get a power of two that way). If there *are* tiny size classes, then
+ * the first size class needs to have lg_delta relative to the largest tiny size
+ * class. We therefore set lg_base == LG_QUANTUM - 1,
+ * lg_delta == LG_QUANTUM - 1, and ndelta == 1, keeping the rest of the
+ * pseudo-group the same.
+ *
+ *
+ * Other terminology:
+ * "Small" size classes mean those that are allocated out of bins, which is the
+ * same as those that are slab allocated.
+ * "Large" size classes are those that are not small. The cutoff for counting as
+ * large is page size * group size.
+ */
+
+/*
+ * Size class N + (1 << SC_LG_NGROUP) twice the size of size class N.
+ */
+#define SC_LG_NGROUP 2
+#define SC_LG_TINY_MIN 3
+
+#if SC_LG_TINY_MIN == 0
+/* The div module doesn't support division by 1, which this would require. */
+#error "Unsupported LG_TINY_MIN"
+#endif
+
+/*
+ * The definitions below are all determined by the above settings and system
+ * characteristics.
+ */
+#define SC_NGROUP (1ULL << SC_LG_NGROUP)
+#define SC_PTR_BITS ((1ULL << LG_SIZEOF_PTR) * 8)
+#define SC_NTINY (LG_QUANTUM - SC_LG_TINY_MIN)
+#define SC_LG_TINY_MAXCLASS (LG_QUANTUM > SC_LG_TINY_MIN ? LG_QUANTUM - 1 : -1)
+#define SC_NPSEUDO SC_NGROUP
+#define SC_LG_FIRST_REGULAR_BASE (LG_QUANTUM + SC_LG_NGROUP)
+/*
+ * We cap allocations to be less than 2 ** (ptr_bits - 1), so the highest base
+ * we need is 2 ** (ptr_bits - 2). (This also means that the last group is 1
+ * size class shorter than the others).
+ * We could probably save some space in arenas by capping this at LG_VADDR size.
+ */
+#define SC_LG_BASE_MAX (SC_PTR_BITS - 2)
+#define SC_NREGULAR (SC_NGROUP * 					\
+    (SC_LG_BASE_MAX - SC_LG_FIRST_REGULAR_BASE + 1) - 1)
+#define SC_NSIZES (SC_NTINY + SC_NPSEUDO + SC_NREGULAR)
+
+/* The number of size classes that are a multiple of the page size. */
+#define SC_NPSIZES (							\
+    /* Start with all the size classes. */				\
+    SC_NSIZES								\
+    /* Subtract out those groups with too small a base. */		\
+    - (LG_PAGE - 1 - SC_LG_FIRST_REGULAR_BASE) * SC_NGROUP		\
+    /* And the pseudo-group. */						\
+    - SC_NPSEUDO							\
+    /* And the tiny group. */						\
+    - SC_NTINY								\
+    /* Sizes where ndelta*delta is not a multiple of the page size. */	\
+    - (SC_LG_NGROUP * SC_NGROUP))
+/*
+ * Note that the last line is computed as the sum of the second column in the
+ * following table:
+ *                      lg(base) | count of sizes to exclude
+ * ------------------------------|-----------------------------
+ *                   LG_PAGE - 1 | SC_NGROUP - 1
+ *                       LG_PAGE | SC_NGROUP - 1
+ *                   LG_PAGE + 1 | SC_NGROUP - 2
+ *                   LG_PAGE + 2 | SC_NGROUP - 4
+ *                           ... | ...
+ *  LG_PAGE + (SC_LG_NGROUP - 1) | SC_NGROUP - (SC_NGROUP / 2)
+ */
+
+/*
+ * We declare a size class is binnable if size < page size * group. Or, in other
+ * words, lg(size) < lg(page size) + lg(group size).
+ */
+#define SC_NBINS (							\
+    /* Sub-regular size classes. */					\
+    SC_NTINY + SC_NPSEUDO						\
+    /* Groups with lg_regular_min_base <= lg_base <= lg_base_max */	\
+    + SC_NGROUP * (LG_PAGE + SC_LG_NGROUP - SC_LG_FIRST_REGULAR_BASE)	\
+    /* Last SC of the last group hits the bound exactly; exclude it. */	\
+    - 1)
+
+/*
+ * The size2index_tab lookup table uses uint8_t to encode each bin index, so we
+ * cannot support more than 256 small size classes.
+ */
+#if (SC_NBINS > 256)
+#  error "Too many small size classes"
+#endif
+
+/* The largest size class in the lookup table. */
+#define SC_LOOKUP_MAXCLASS ((size_t)1 << 12)
+
+/* Internal, only used for the definition of SC_SMALL_MAXCLASS. */
+#define SC_SMALL_MAX_BASE ((size_t)1 << (LG_PAGE + SC_LG_NGROUP - 1))
+#define SC_SMALL_MAX_DELTA ((size_t)1 << (LG_PAGE - 1))
+
+/* The largest size class allocated out of a slab. */
+#define SC_SMALL_MAXCLASS (SC_SMALL_MAX_BASE				\
+    + (SC_NGROUP - 1) * SC_SMALL_MAX_DELTA)
+
+/* The smallest size class not allocated out of a slab. */
+#define SC_LARGE_MINCLASS ((size_t)1ULL << (LG_PAGE + SC_LG_NGROUP))
+#define SC_LG_LARGE_MINCLASS (LG_PAGE + SC_LG_NGROUP)
+
+/* Internal; only used for the definition of SC_LARGE_MAXCLASS. */
+#define SC_MAX_BASE ((size_t)1 << (SC_PTR_BITS - 2))
+#define SC_MAX_DELTA ((size_t)1 << (SC_PTR_BITS - 2 - SC_LG_NGROUP))
+
+/* The largest size class supported. */
+#define SC_LARGE_MAXCLASS (SC_MAX_BASE + (SC_NGROUP - 1) * SC_MAX_DELTA)
+
+typedef struct sc_s sc_t;
+struct sc_s {
+	/* Size class index, or -1 if not a valid size class. */
+	int index;
+	/* Lg group base size (no deltas added). */
+	int lg_base;
+	/* Lg delta to previous size class. */
+	int lg_delta;
+	/* Delta multiplier.  size == 1<<lg_base + ndelta<<lg_delta */
+	int ndelta;
+	/*
+	 * True if the size class is a multiple of the page size, false
+	 * otherwise.
+	 */
+	bool psz;
+	/*
+	 * True if the size class is a small, bin, size class. False otherwise.
+	 */
+	bool bin;
+	/* The slab page count if a small bin size class, 0 otherwise. */
+	int pgs;
+	/* Same as lg_delta if a lookup table size class, 0 otherwise. */
+	int lg_delta_lookup;
+};
+
+typedef struct sc_data_s sc_data_t;
+struct sc_data_s {
+	/* Number of tiny size classes. */
+	unsigned ntiny;
+	/* Number of bins supported by the lookup table. */
+	int nlbins;
+	/* Number of small size class bins. */
+	int nbins;
+	/* Number of size classes. */
+	int nsizes;
+	/* Number of bits required to store NSIZES. */
+	int lg_ceil_nsizes;
+	/* Number of size classes that are a multiple of (1U << LG_PAGE). */
+	unsigned npsizes;
+	/* Lg of maximum tiny size class (or -1, if none). */
+	int lg_tiny_maxclass;
+	/* Maximum size class included in lookup table. */
+	size_t lookup_maxclass;
+	/* Maximum small size class. */
+	size_t small_maxclass;
+	/* Lg of minimum large size class. */
+	int lg_large_minclass;
+	/* The minimum large size class. */
+	size_t large_minclass;
+	/* Maximum (large) size class. */
+	size_t large_maxclass;
+	/* True if the sc_data_t has been initialized (for debugging only). */
+	bool initialized;
+
+	sc_t sc[SC_NSIZES];
+};
+
+void sc_data_init(sc_data_t *data);
+/*
+ * Updates slab sizes in [begin, end] to be pgs pages in length, if possible.
+ * Otherwise, does its best to accomodate the request.
+ */
+void sc_data_update_slab_size(sc_data_t *data, size_t begin, size_t end,
+    int pgs);
+void sc_boot(sc_data_t *data);
+
+#endif /* JEMALLOC_INTERNAL_SC_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/seq.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/seq.h
new file mode 100644
index 0000000..ef2df4c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/seq.h
@@ -0,0 +1,55 @@
+#ifndef JEMALLOC_INTERNAL_SEQ_H
+#define JEMALLOC_INTERNAL_SEQ_H
+
+#include "jemalloc/internal/atomic.h"
+
+/*
+ * A simple seqlock implementation.
+ */
+
+#define seq_define(type, short_type)					\
+typedef struct {							\
+	atomic_zu_t seq;						\
+	atomic_zu_t data[						\
+	    (sizeof(type) + sizeof(size_t) - 1) / sizeof(size_t)];	\
+} seq_##short_type##_t;							\
+									\
+/*									\
+ * No internal synchronization -- the caller must ensure that there's	\
+ * only a single writer at a time.					\
+ */									\
+static inline void							\
+seq_store_##short_type(seq_##short_type##_t *dst, type *src) {		\
+	size_t buf[sizeof(dst->data) / sizeof(size_t)];			\
+	buf[sizeof(buf) / sizeof(size_t) - 1] = 0;			\
+	memcpy(buf, src, sizeof(type));					\
+	size_t old_seq = atomic_load_zu(&dst->seq, ATOMIC_RELAXED);	\
+	atomic_store_zu(&dst->seq, old_seq + 1, ATOMIC_RELAXED);	\
+	atomic_fence(ATOMIC_RELEASE);					\
+	for (size_t i = 0; i < sizeof(buf) / sizeof(size_t); i++) {	\
+		atomic_store_zu(&dst->data[i], buf[i], ATOMIC_RELAXED);	\
+	}								\
+	atomic_store_zu(&dst->seq, old_seq + 2, ATOMIC_RELEASE);	\
+}									\
+									\
+/* Returns whether or not the read was consistent. */			\
+static inline bool							\
+seq_try_load_##short_type(type *dst, seq_##short_type##_t *src) {	\
+	size_t buf[sizeof(src->data) / sizeof(size_t)];			\
+	size_t seq1 = atomic_load_zu(&src->seq, ATOMIC_ACQUIRE);	\
+	if (seq1 % 2 != 0) {						\
+		return false;						\
+	}								\
+	for (size_t i = 0; i < sizeof(buf) / sizeof(size_t); i++) {	\
+		buf[i] = atomic_load_zu(&src->data[i], ATOMIC_RELAXED);	\
+	}								\
+	atomic_fence(ATOMIC_ACQUIRE);					\
+	size_t seq2 = atomic_load_zu(&src->seq, ATOMIC_RELAXED);	\
+	if (seq1 != seq2) {						\
+		return false;						\
+	}								\
+	memcpy(dst, buf, sizeof(type));					\
+	return true;							\
+}
+
+#endif /* JEMALLOC_INTERNAL_SEQ_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/smoothstep.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/smoothstep.h
new file mode 100644
index 0000000..2e14430
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/smoothstep.h
@@ -0,0 +1,232 @@
+#ifndef JEMALLOC_INTERNAL_SMOOTHSTEP_H
+#define JEMALLOC_INTERNAL_SMOOTHSTEP_H
+
+/*
+ * This file was generated by the following command:
+ *   sh smoothstep.sh smoother 200 24 3 15
+ */
+/******************************************************************************/
+
+/*
+ * This header defines a precomputed table based on the smoothstep family of
+ * sigmoidal curves (https://en.wikipedia.org/wiki/Smoothstep) that grow from 0
+ * to 1 in 0 <= x <= 1.  The table is stored as integer fixed point values so
+ * that floating point math can be avoided.
+ *
+ *                      3     2
+ *   smoothstep(x) = -2x  + 3x
+ *
+ *                       5      4      3
+ *   smootherstep(x) = 6x  - 15x  + 10x
+ *
+ *                          7      6      5      4
+ *   smootheststep(x) = -20x  + 70x  - 84x  + 35x
+ */
+
+#define SMOOTHSTEP_VARIANT	"smoother"
+#define SMOOTHSTEP_NSTEPS	200
+#define SMOOTHSTEP_BFP		24
+#define SMOOTHSTEP \
+ /* STEP(step, h,                            x,     y) */ \
+    STEP(   1, UINT64_C(0x0000000000000014), 0.005, 0.000001240643750) \
+    STEP(   2, UINT64_C(0x00000000000000a5), 0.010, 0.000009850600000) \
+    STEP(   3, UINT64_C(0x0000000000000229), 0.015, 0.000032995181250) \
+    STEP(   4, UINT64_C(0x0000000000000516), 0.020, 0.000077619200000) \
+    STEP(   5, UINT64_C(0x00000000000009dc), 0.025, 0.000150449218750) \
+    STEP(   6, UINT64_C(0x00000000000010e8), 0.030, 0.000257995800000) \
+    STEP(   7, UINT64_C(0x0000000000001aa4), 0.035, 0.000406555756250) \
+    STEP(   8, UINT64_C(0x0000000000002777), 0.040, 0.000602214400000) \
+    STEP(   9, UINT64_C(0x00000000000037c2), 0.045, 0.000850847793750) \
+    STEP(  10, UINT64_C(0x0000000000004be6), 0.050, 0.001158125000000) \
+    STEP(  11, UINT64_C(0x000000000000643c), 0.055, 0.001529510331250) \
+    STEP(  12, UINT64_C(0x000000000000811f), 0.060, 0.001970265600000) \
+    STEP(  13, UINT64_C(0x000000000000a2e2), 0.065, 0.002485452368750) \
+    STEP(  14, UINT64_C(0x000000000000c9d8), 0.070, 0.003079934200000) \
+    STEP(  15, UINT64_C(0x000000000000f64f), 0.075, 0.003758378906250) \
+    STEP(  16, UINT64_C(0x0000000000012891), 0.080, 0.004525260800000) \
+    STEP(  17, UINT64_C(0x00000000000160e7), 0.085, 0.005384862943750) \
+    STEP(  18, UINT64_C(0x0000000000019f95), 0.090, 0.006341279400000) \
+    STEP(  19, UINT64_C(0x000000000001e4dc), 0.095, 0.007398417481250) \
+    STEP(  20, UINT64_C(0x00000000000230fc), 0.100, 0.008560000000000) \
+    STEP(  21, UINT64_C(0x0000000000028430), 0.105, 0.009829567518750) \
+    STEP(  22, UINT64_C(0x000000000002deb0), 0.110, 0.011210480600000) \
+    STEP(  23, UINT64_C(0x00000000000340b1), 0.115, 0.012705922056250) \
+    STEP(  24, UINT64_C(0x000000000003aa67), 0.120, 0.014318899200000) \
+    STEP(  25, UINT64_C(0x0000000000041c00), 0.125, 0.016052246093750) \
+    STEP(  26, UINT64_C(0x00000000000495a8), 0.130, 0.017908625800000) \
+    STEP(  27, UINT64_C(0x000000000005178b), 0.135, 0.019890532631250) \
+    STEP(  28, UINT64_C(0x000000000005a1cf), 0.140, 0.022000294400000) \
+    STEP(  29, UINT64_C(0x0000000000063498), 0.145, 0.024240074668750) \
+    STEP(  30, UINT64_C(0x000000000006d009), 0.150, 0.026611875000000) \
+    STEP(  31, UINT64_C(0x000000000007743f), 0.155, 0.029117537206250) \
+    STEP(  32, UINT64_C(0x0000000000082157), 0.160, 0.031758745600000) \
+    STEP(  33, UINT64_C(0x000000000008d76b), 0.165, 0.034537029243750) \
+    STEP(  34, UINT64_C(0x0000000000099691), 0.170, 0.037453764200000) \
+    STEP(  35, UINT64_C(0x00000000000a5edf), 0.175, 0.040510175781250) \
+    STEP(  36, UINT64_C(0x00000000000b3067), 0.180, 0.043707340800000) \
+    STEP(  37, UINT64_C(0x00000000000c0b38), 0.185, 0.047046189818750) \
+    STEP(  38, UINT64_C(0x00000000000cef5e), 0.190, 0.050527509400000) \
+    STEP(  39, UINT64_C(0x00000000000ddce6), 0.195, 0.054151944356250) \
+    STEP(  40, UINT64_C(0x00000000000ed3d8), 0.200, 0.057920000000000) \
+    STEP(  41, UINT64_C(0x00000000000fd439), 0.205, 0.061832044393750) \
+    STEP(  42, UINT64_C(0x000000000010de0e), 0.210, 0.065888310600000) \
+    STEP(  43, UINT64_C(0x000000000011f158), 0.215, 0.070088898931250) \
+    STEP(  44, UINT64_C(0x0000000000130e17), 0.220, 0.074433779200000) \
+    STEP(  45, UINT64_C(0x0000000000143448), 0.225, 0.078922792968750) \
+    STEP(  46, UINT64_C(0x00000000001563e7), 0.230, 0.083555655800000) \
+    STEP(  47, UINT64_C(0x0000000000169cec), 0.235, 0.088331959506250) \
+    STEP(  48, UINT64_C(0x000000000017df4f), 0.240, 0.093251174400000) \
+    STEP(  49, UINT64_C(0x0000000000192b04), 0.245, 0.098312651543750) \
+    STEP(  50, UINT64_C(0x00000000001a8000), 0.250, 0.103515625000000) \
+    STEP(  51, UINT64_C(0x00000000001bde32), 0.255, 0.108859214081250) \
+    STEP(  52, UINT64_C(0x00000000001d458b), 0.260, 0.114342425600000) \
+    STEP(  53, UINT64_C(0x00000000001eb5f8), 0.265, 0.119964156118750) \
+    STEP(  54, UINT64_C(0x0000000000202f65), 0.270, 0.125723194200000) \
+    STEP(  55, UINT64_C(0x000000000021b1bb), 0.275, 0.131618222656250) \
+    STEP(  56, UINT64_C(0x0000000000233ce3), 0.280, 0.137647820800000) \
+    STEP(  57, UINT64_C(0x000000000024d0c3), 0.285, 0.143810466693750) \
+    STEP(  58, UINT64_C(0x0000000000266d40), 0.290, 0.150104539400000) \
+    STEP(  59, UINT64_C(0x000000000028123d), 0.295, 0.156528321231250) \
+    STEP(  60, UINT64_C(0x000000000029bf9c), 0.300, 0.163080000000000) \
+    STEP(  61, UINT64_C(0x00000000002b753d), 0.305, 0.169757671268750) \
+    STEP(  62, UINT64_C(0x00000000002d32fe), 0.310, 0.176559340600000) \
+    STEP(  63, UINT64_C(0x00000000002ef8bc), 0.315, 0.183482925806250) \
+    STEP(  64, UINT64_C(0x000000000030c654), 0.320, 0.190526259200000) \
+    STEP(  65, UINT64_C(0x0000000000329b9f), 0.325, 0.197687089843750) \
+    STEP(  66, UINT64_C(0x0000000000347875), 0.330, 0.204963085800000) \
+    STEP(  67, UINT64_C(0x0000000000365cb0), 0.335, 0.212351836381250) \
+    STEP(  68, UINT64_C(0x0000000000384825), 0.340, 0.219850854400000) \
+    STEP(  69, UINT64_C(0x00000000003a3aa8), 0.345, 0.227457578418750) \
+    STEP(  70, UINT64_C(0x00000000003c340f), 0.350, 0.235169375000000) \
+    STEP(  71, UINT64_C(0x00000000003e342b), 0.355, 0.242983540956250) \
+    STEP(  72, UINT64_C(0x0000000000403ace), 0.360, 0.250897305600000) \
+    STEP(  73, UINT64_C(0x00000000004247c8), 0.365, 0.258907832993750) \
+    STEP(  74, UINT64_C(0x0000000000445ae9), 0.370, 0.267012224200000) \
+    STEP(  75, UINT64_C(0x0000000000467400), 0.375, 0.275207519531250) \
+    STEP(  76, UINT64_C(0x00000000004892d8), 0.380, 0.283490700800000) \
+    STEP(  77, UINT64_C(0x00000000004ab740), 0.385, 0.291858693568750) \
+    STEP(  78, UINT64_C(0x00000000004ce102), 0.390, 0.300308369400000) \
+    STEP(  79, UINT64_C(0x00000000004f0fe9), 0.395, 0.308836548106250) \
+    STEP(  80, UINT64_C(0x00000000005143bf), 0.400, 0.317440000000000) \
+    STEP(  81, UINT64_C(0x0000000000537c4d), 0.405, 0.326115448143750) \
+    STEP(  82, UINT64_C(0x000000000055b95b), 0.410, 0.334859570600000) \
+    STEP(  83, UINT64_C(0x000000000057fab1), 0.415, 0.343669002681250) \
+    STEP(  84, UINT64_C(0x00000000005a4015), 0.420, 0.352540339200000) \
+    STEP(  85, UINT64_C(0x00000000005c894e), 0.425, 0.361470136718750) \
+    STEP(  86, UINT64_C(0x00000000005ed622), 0.430, 0.370454915800000) \
+    STEP(  87, UINT64_C(0x0000000000612655), 0.435, 0.379491163256250) \
+    STEP(  88, UINT64_C(0x00000000006379ac), 0.440, 0.388575334400000) \
+    STEP(  89, UINT64_C(0x000000000065cfeb), 0.445, 0.397703855293750) \
+    STEP(  90, UINT64_C(0x00000000006828d6), 0.450, 0.406873125000000) \
+    STEP(  91, UINT64_C(0x00000000006a842f), 0.455, 0.416079517831250) \
+    STEP(  92, UINT64_C(0x00000000006ce1bb), 0.460, 0.425319385600000) \
+    STEP(  93, UINT64_C(0x00000000006f413a), 0.465, 0.434589059868750) \
+    STEP(  94, UINT64_C(0x000000000071a270), 0.470, 0.443884854200000) \
+    STEP(  95, UINT64_C(0x000000000074051d), 0.475, 0.453203066406250) \
+    STEP(  96, UINT64_C(0x0000000000766905), 0.480, 0.462539980800000) \
+    STEP(  97, UINT64_C(0x000000000078cde7), 0.485, 0.471891870443750) \
+    STEP(  98, UINT64_C(0x00000000007b3387), 0.490, 0.481254999400000) \
+    STEP(  99, UINT64_C(0x00000000007d99a4), 0.495, 0.490625624981250) \
+    STEP( 100, UINT64_C(0x0000000000800000), 0.500, 0.500000000000000) \
+    STEP( 101, UINT64_C(0x000000000082665b), 0.505, 0.509374375018750) \
+    STEP( 102, UINT64_C(0x000000000084cc78), 0.510, 0.518745000600000) \
+    STEP( 103, UINT64_C(0x0000000000873218), 0.515, 0.528108129556250) \
+    STEP( 104, UINT64_C(0x00000000008996fa), 0.520, 0.537460019200000) \
+    STEP( 105, UINT64_C(0x00000000008bfae2), 0.525, 0.546796933593750) \
+    STEP( 106, UINT64_C(0x00000000008e5d8f), 0.530, 0.556115145800000) \
+    STEP( 107, UINT64_C(0x000000000090bec5), 0.535, 0.565410940131250) \
+    STEP( 108, UINT64_C(0x0000000000931e44), 0.540, 0.574680614400000) \
+    STEP( 109, UINT64_C(0x0000000000957bd0), 0.545, 0.583920482168750) \
+    STEP( 110, UINT64_C(0x000000000097d729), 0.550, 0.593126875000000) \
+    STEP( 111, UINT64_C(0x00000000009a3014), 0.555, 0.602296144706250) \
+    STEP( 112, UINT64_C(0x00000000009c8653), 0.560, 0.611424665600000) \
+    STEP( 113, UINT64_C(0x00000000009ed9aa), 0.565, 0.620508836743750) \
+    STEP( 114, UINT64_C(0x0000000000a129dd), 0.570, 0.629545084200000) \
+    STEP( 115, UINT64_C(0x0000000000a376b1), 0.575, 0.638529863281250) \
+    STEP( 116, UINT64_C(0x0000000000a5bfea), 0.580, 0.647459660800000) \
+    STEP( 117, UINT64_C(0x0000000000a8054e), 0.585, 0.656330997318750) \
+    STEP( 118, UINT64_C(0x0000000000aa46a4), 0.590, 0.665140429400000) \
+    STEP( 119, UINT64_C(0x0000000000ac83b2), 0.595, 0.673884551856250) \
+    STEP( 120, UINT64_C(0x0000000000aebc40), 0.600, 0.682560000000000) \
+    STEP( 121, UINT64_C(0x0000000000b0f016), 0.605, 0.691163451893750) \
+    STEP( 122, UINT64_C(0x0000000000b31efd), 0.610, 0.699691630600000) \
+    STEP( 123, UINT64_C(0x0000000000b548bf), 0.615, 0.708141306431250) \
+    STEP( 124, UINT64_C(0x0000000000b76d27), 0.620, 0.716509299200000) \
+    STEP( 125, UINT64_C(0x0000000000b98c00), 0.625, 0.724792480468750) \
+    STEP( 126, UINT64_C(0x0000000000bba516), 0.630, 0.732987775800000) \
+    STEP( 127, UINT64_C(0x0000000000bdb837), 0.635, 0.741092167006250) \
+    STEP( 128, UINT64_C(0x0000000000bfc531), 0.640, 0.749102694400000) \
+    STEP( 129, UINT64_C(0x0000000000c1cbd4), 0.645, 0.757016459043750) \
+    STEP( 130, UINT64_C(0x0000000000c3cbf0), 0.650, 0.764830625000000) \
+    STEP( 131, UINT64_C(0x0000000000c5c557), 0.655, 0.772542421581250) \
+    STEP( 132, UINT64_C(0x0000000000c7b7da), 0.660, 0.780149145600000) \
+    STEP( 133, UINT64_C(0x0000000000c9a34f), 0.665, 0.787648163618750) \
+    STEP( 134, UINT64_C(0x0000000000cb878a), 0.670, 0.795036914200000) \
+    STEP( 135, UINT64_C(0x0000000000cd6460), 0.675, 0.802312910156250) \
+    STEP( 136, UINT64_C(0x0000000000cf39ab), 0.680, 0.809473740800000) \
+    STEP( 137, UINT64_C(0x0000000000d10743), 0.685, 0.816517074193750) \
+    STEP( 138, UINT64_C(0x0000000000d2cd01), 0.690, 0.823440659400000) \
+    STEP( 139, UINT64_C(0x0000000000d48ac2), 0.695, 0.830242328731250) \
+    STEP( 140, UINT64_C(0x0000000000d64063), 0.700, 0.836920000000000) \
+    STEP( 141, UINT64_C(0x0000000000d7edc2), 0.705, 0.843471678768750) \
+    STEP( 142, UINT64_C(0x0000000000d992bf), 0.710, 0.849895460600000) \
+    STEP( 143, UINT64_C(0x0000000000db2f3c), 0.715, 0.856189533306250) \
+    STEP( 144, UINT64_C(0x0000000000dcc31c), 0.720, 0.862352179200000) \
+    STEP( 145, UINT64_C(0x0000000000de4e44), 0.725, 0.868381777343750) \
+    STEP( 146, UINT64_C(0x0000000000dfd09a), 0.730, 0.874276805800000) \
+    STEP( 147, UINT64_C(0x0000000000e14a07), 0.735, 0.880035843881250) \
+    STEP( 148, UINT64_C(0x0000000000e2ba74), 0.740, 0.885657574400000) \
+    STEP( 149, UINT64_C(0x0000000000e421cd), 0.745, 0.891140785918750) \
+    STEP( 150, UINT64_C(0x0000000000e58000), 0.750, 0.896484375000000) \
+    STEP( 151, UINT64_C(0x0000000000e6d4fb), 0.755, 0.901687348456250) \
+    STEP( 152, UINT64_C(0x0000000000e820b0), 0.760, 0.906748825600000) \
+    STEP( 153, UINT64_C(0x0000000000e96313), 0.765, 0.911668040493750) \
+    STEP( 154, UINT64_C(0x0000000000ea9c18), 0.770, 0.916444344200000) \
+    STEP( 155, UINT64_C(0x0000000000ebcbb7), 0.775, 0.921077207031250) \
+    STEP( 156, UINT64_C(0x0000000000ecf1e8), 0.780, 0.925566220800000) \
+    STEP( 157, UINT64_C(0x0000000000ee0ea7), 0.785, 0.929911101068750) \
+    STEP( 158, UINT64_C(0x0000000000ef21f1), 0.790, 0.934111689400000) \
+    STEP( 159, UINT64_C(0x0000000000f02bc6), 0.795, 0.938167955606250) \
+    STEP( 160, UINT64_C(0x0000000000f12c27), 0.800, 0.942080000000000) \
+    STEP( 161, UINT64_C(0x0000000000f22319), 0.805, 0.945848055643750) \
+    STEP( 162, UINT64_C(0x0000000000f310a1), 0.810, 0.949472490600000) \
+    STEP( 163, UINT64_C(0x0000000000f3f4c7), 0.815, 0.952953810181250) \
+    STEP( 164, UINT64_C(0x0000000000f4cf98), 0.820, 0.956292659200000) \
+    STEP( 165, UINT64_C(0x0000000000f5a120), 0.825, 0.959489824218750) \
+    STEP( 166, UINT64_C(0x0000000000f6696e), 0.830, 0.962546235800000) \
+    STEP( 167, UINT64_C(0x0000000000f72894), 0.835, 0.965462970756250) \
+    STEP( 168, UINT64_C(0x0000000000f7dea8), 0.840, 0.968241254400000) \
+    STEP( 169, UINT64_C(0x0000000000f88bc0), 0.845, 0.970882462793750) \
+    STEP( 170, UINT64_C(0x0000000000f92ff6), 0.850, 0.973388125000000) \
+    STEP( 171, UINT64_C(0x0000000000f9cb67), 0.855, 0.975759925331250) \
+    STEP( 172, UINT64_C(0x0000000000fa5e30), 0.860, 0.977999705600000) \
+    STEP( 173, UINT64_C(0x0000000000fae874), 0.865, 0.980109467368750) \
+    STEP( 174, UINT64_C(0x0000000000fb6a57), 0.870, 0.982091374200000) \
+    STEP( 175, UINT64_C(0x0000000000fbe400), 0.875, 0.983947753906250) \
+    STEP( 176, UINT64_C(0x0000000000fc5598), 0.880, 0.985681100800000) \
+    STEP( 177, UINT64_C(0x0000000000fcbf4e), 0.885, 0.987294077943750) \
+    STEP( 178, UINT64_C(0x0000000000fd214f), 0.890, 0.988789519400000) \
+    STEP( 179, UINT64_C(0x0000000000fd7bcf), 0.895, 0.990170432481250) \
+    STEP( 180, UINT64_C(0x0000000000fdcf03), 0.900, 0.991440000000000) \
+    STEP( 181, UINT64_C(0x0000000000fe1b23), 0.905, 0.992601582518750) \
+    STEP( 182, UINT64_C(0x0000000000fe606a), 0.910, 0.993658720600000) \
+    STEP( 183, UINT64_C(0x0000000000fe9f18), 0.915, 0.994615137056250) \
+    STEP( 184, UINT64_C(0x0000000000fed76e), 0.920, 0.995474739200000) \
+    STEP( 185, UINT64_C(0x0000000000ff09b0), 0.925, 0.996241621093750) \
+    STEP( 186, UINT64_C(0x0000000000ff3627), 0.930, 0.996920065800000) \
+    STEP( 187, UINT64_C(0x0000000000ff5d1d), 0.935, 0.997514547631250) \
+    STEP( 188, UINT64_C(0x0000000000ff7ee0), 0.940, 0.998029734400000) \
+    STEP( 189, UINT64_C(0x0000000000ff9bc3), 0.945, 0.998470489668750) \
+    STEP( 190, UINT64_C(0x0000000000ffb419), 0.950, 0.998841875000000) \
+    STEP( 191, UINT64_C(0x0000000000ffc83d), 0.955, 0.999149152206250) \
+    STEP( 192, UINT64_C(0x0000000000ffd888), 0.960, 0.999397785600000) \
+    STEP( 193, UINT64_C(0x0000000000ffe55b), 0.965, 0.999593444243750) \
+    STEP( 194, UINT64_C(0x0000000000ffef17), 0.970, 0.999742004200000) \
+    STEP( 195, UINT64_C(0x0000000000fff623), 0.975, 0.999849550781250) \
+    STEP( 196, UINT64_C(0x0000000000fffae9), 0.980, 0.999922380800000) \
+    STEP( 197, UINT64_C(0x0000000000fffdd6), 0.985, 0.999967004818750) \
+    STEP( 198, UINT64_C(0x0000000000ffff5a), 0.990, 0.999990149400000) \
+    STEP( 199, UINT64_C(0x0000000000ffffeb), 0.995, 0.999998759356250) \
+    STEP( 200, UINT64_C(0x0000000001000000), 1.000, 1.000000000000000) \
+
+#endif /* JEMALLOC_INTERNAL_SMOOTHSTEP_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/smoothstep.sh b/memkind-1.10.0/jemalloc/include/jemalloc/internal/smoothstep.sh
new file mode 100755
index 0000000..65de97b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/smoothstep.sh
@@ -0,0 +1,101 @@
+#!/bin/sh
+#
+# Generate a discrete lookup table for a sigmoid function in the smoothstep
+# family (https://en.wikipedia.org/wiki/Smoothstep), where the lookup table
+# entries correspond to x in [1/nsteps, 2/nsteps, ..., nsteps/nsteps].  Encode
+# the entries using a binary fixed point representation.
+#
+# Usage: smoothstep.sh <variant> <nsteps> <bfp> <xprec> <yprec>
+#
+#        <variant> is in {smooth, smoother, smoothest}.
+#        <nsteps> must be greater than zero.
+#        <bfp> must be in [0..62]; reasonable values are roughly [10..30].
+#        <xprec> is x decimal precision.
+#        <yprec> is y decimal precision.
+
+#set -x
+
+cmd="sh smoothstep.sh $*"
+variant=$1
+nsteps=$2
+bfp=$3
+xprec=$4
+yprec=$5
+
+case "${variant}" in
+  smooth)
+    ;;
+  smoother)
+    ;;
+  smoothest)
+    ;;
+  *)
+    echo "Unsupported variant"
+    exit 1
+    ;;
+esac
+
+smooth() {
+  step=$1
+  y=`echo ${yprec} k ${step} ${nsteps} / sx _2 lx 3 ^ '*' 3 lx 2 ^ '*' + p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  h=`echo ${yprec} k 2 ${bfp} ^ ${y} '*' p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g' | tr '.' ' ' | awk '{print $1}' `
+}
+
+smoother() {
+  step=$1
+  y=`echo ${yprec} k ${step} ${nsteps} / sx 6 lx 5 ^ '*' _15 lx 4 ^ '*' + 10 lx 3 ^ '*' + p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  h=`echo ${yprec} k 2 ${bfp} ^ ${y} '*' p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g' | tr '.' ' ' | awk '{print $1}' `
+}
+
+smoothest() {
+  step=$1
+  y=`echo ${yprec} k ${step} ${nsteps} / sx _20 lx 7 ^ '*' 70 lx 6 ^ '*' + _84 lx 5 ^ '*' + 35 lx 4 ^ '*' + p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  h=`echo ${yprec} k 2 ${bfp} ^ ${y} '*' p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g' | tr '.' ' ' | awk '{print $1}' `
+}
+
+cat <<EOF
+#ifndef JEMALLOC_INTERNAL_SMOOTHSTEP_H
+#define JEMALLOC_INTERNAL_SMOOTHSTEP_H
+
+/*
+ * This file was generated by the following command:
+ *   $cmd
+ */
+/******************************************************************************/
+
+/*
+ * This header defines a precomputed table based on the smoothstep family of
+ * sigmoidal curves (https://en.wikipedia.org/wiki/Smoothstep) that grow from 0
+ * to 1 in 0 <= x <= 1.  The table is stored as integer fixed point values so
+ * that floating point math can be avoided.
+ *
+ *                      3     2
+ *   smoothstep(x) = -2x  + 3x
+ *
+ *                       5      4      3
+ *   smootherstep(x) = 6x  - 15x  + 10x
+ *
+ *                          7      6      5      4
+ *   smootheststep(x) = -20x  + 70x  - 84x  + 35x
+ */
+
+#define SMOOTHSTEP_VARIANT	"${variant}"
+#define SMOOTHSTEP_NSTEPS	${nsteps}
+#define SMOOTHSTEP_BFP		${bfp}
+#define SMOOTHSTEP \\
+ /* STEP(step, h,                            x,     y) */ \\
+EOF
+
+s=1
+while [ $s -le $nsteps ] ; do
+  $variant ${s}
+  x=`echo ${xprec} k ${s} ${nsteps} / p | dc | tr -d '\\\\\n' | sed -e 's#^\.#0.#g'`
+  printf '    STEP(%4d, UINT64_C(0x%016x), %s, %s) \\\n' ${s} ${h} ${x} ${y}
+
+  s=$((s+1))
+done
+echo
+
+cat <<EOF
+#endif /* JEMALLOC_INTERNAL_SMOOTHSTEP_H */
+EOF
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/spin.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/spin.h
new file mode 100644
index 0000000..22804c6
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/spin.h
@@ -0,0 +1,40 @@
+#ifndef JEMALLOC_INTERNAL_SPIN_H
+#define JEMALLOC_INTERNAL_SPIN_H
+
+#define SPIN_INITIALIZER {0U}
+
+typedef struct {
+	unsigned iteration;
+} spin_t;
+
+static inline void
+spin_cpu_spinwait() {
+#  if HAVE_CPU_SPINWAIT
+	CPU_SPINWAIT;
+#  else
+	volatile int x = 0;
+	x = x;
+#  endif
+}
+
+static inline void
+spin_adaptive(spin_t *spin) {
+	volatile uint32_t i;
+
+	if (spin->iteration < 5) {
+		for (i = 0; i < (1U << spin->iteration); i++) {
+			spin_cpu_spinwait();
+		}
+		spin->iteration++;
+	} else {
+#ifdef _WIN32
+		SwitchToThread();
+#else
+		sched_yield();
+#endif
+	}
+}
+
+#undef SPIN_INLINE
+
+#endif /* JEMALLOC_INTERNAL_SPIN_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/stats.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/stats.h
new file mode 100644
index 0000000..3b9e0ea
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/stats.h
@@ -0,0 +1,31 @@
+#ifndef JEMALLOC_INTERNAL_STATS_H
+#define JEMALLOC_INTERNAL_STATS_H
+
+/*  OPTION(opt,		var_name,	default,	set_value_to) */
+#define STATS_PRINT_OPTIONS						\
+    OPTION('J',		json,		false,		true)		\
+    OPTION('g',		general,	true,		false)		\
+    OPTION('m',		merged,		config_stats,	false)		\
+    OPTION('d',		destroyed,	config_stats,	false)		\
+    OPTION('a',		unmerged,	config_stats,	false)		\
+    OPTION('b',		bins,		true,		false)		\
+    OPTION('l',		large,		true,		false)		\
+    OPTION('x',		mutex,		true,		false)		\
+    OPTION('e',		extents,	true,		false)
+
+enum {
+#define OPTION(o, v, d, s) stats_print_option_num_##v,
+    STATS_PRINT_OPTIONS
+#undef OPTION
+    stats_print_tot_num_options
+};
+
+/* Options for stats_print. */
+extern bool opt_stats_print;
+extern char opt_stats_print_opts[stats_print_tot_num_options+1];
+
+/* Implements je_malloc_stats_print. */
+void stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts);
+
+#endif /* JEMALLOC_INTERNAL_STATS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/sz.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/sz.h
new file mode 100644
index 0000000..68e558a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/sz.h
@@ -0,0 +1,318 @@
+#ifndef JEMALLOC_INTERNAL_SIZE_H
+#define JEMALLOC_INTERNAL_SIZE_H
+
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/util.h"
+
+/*
+ * sz module: Size computations.
+ *
+ * Some abbreviations used here:
+ *   p: Page
+ *   ind: Index
+ *   s, sz: Size
+ *   u: Usable size
+ *   a: Aligned
+ *
+ * These are not always used completely consistently, but should be enough to
+ * interpret function names.  E.g. sz_psz2ind converts page size to page size
+ * index; sz_sa2u converts a (size, alignment) allocation request to the usable
+ * size that would result from such an allocation.
+ */
+
+/*
+ * sz_pind2sz_tab encodes the same information as could be computed by
+ * sz_pind2sz_compute().
+ */
+extern size_t sz_pind2sz_tab[SC_NPSIZES + 1];
+/*
+ * sz_index2size_tab encodes the same information as could be computed (at
+ * unacceptable cost in some code paths) by sz_index2size_compute().
+ */
+extern size_t sz_index2size_tab[SC_NSIZES];
+/*
+ * sz_size2index_tab is a compact lookup table that rounds request sizes up to
+ * size classes.  In order to reduce cache footprint, the table is compressed,
+ * and all accesses are via sz_size2index().
+ */
+extern uint8_t sz_size2index_tab[];
+
+static const size_t sz_large_pad =
+#ifdef JEMALLOC_CACHE_OBLIVIOUS
+    PAGE
+#else
+    0
+#endif
+    ;
+
+extern void sz_boot(const sc_data_t *sc_data);
+
+JEMALLOC_ALWAYS_INLINE pszind_t
+sz_psz2ind(size_t psz) {
+	if (unlikely(psz > SC_LARGE_MAXCLASS)) {
+		return SC_NPSIZES;
+	}
+	pszind_t x = lg_floor((psz<<1)-1);
+	pszind_t shift = (x < SC_LG_NGROUP + LG_PAGE) ?
+	    0 : x - (SC_LG_NGROUP + LG_PAGE);
+	pszind_t grp = shift << SC_LG_NGROUP;
+
+	pszind_t lg_delta = (x < SC_LG_NGROUP + LG_PAGE + 1) ?
+	    LG_PAGE : x - SC_LG_NGROUP - 1;
+
+	size_t delta_inverse_mask = ZU(-1) << lg_delta;
+	pszind_t mod = ((((psz-1) & delta_inverse_mask) >> lg_delta)) &
+	    ((ZU(1) << SC_LG_NGROUP) - 1);
+
+	pszind_t ind = grp + mod;
+	return ind;
+}
+
+static inline size_t
+sz_pind2sz_compute(pszind_t pind) {
+	if (unlikely(pind == SC_NPSIZES)) {
+		return SC_LARGE_MAXCLASS + PAGE;
+	}
+	size_t grp = pind >> SC_LG_NGROUP;
+	size_t mod = pind & ((ZU(1) << SC_LG_NGROUP) - 1);
+
+	size_t grp_size_mask = ~((!!grp)-1);
+	size_t grp_size = ((ZU(1) << (LG_PAGE + (SC_LG_NGROUP-1))) << grp)
+	    & grp_size_mask;
+
+	size_t shift = (grp == 0) ? 1 : grp;
+	size_t lg_delta = shift + (LG_PAGE-1);
+	size_t mod_size = (mod+1) << lg_delta;
+
+	size_t sz = grp_size + mod_size;
+	return sz;
+}
+
+static inline size_t
+sz_pind2sz_lookup(pszind_t pind) {
+	size_t ret = (size_t)sz_pind2sz_tab[pind];
+	assert(ret == sz_pind2sz_compute(pind));
+	return ret;
+}
+
+static inline size_t
+sz_pind2sz(pszind_t pind) {
+	assert(pind < SC_NPSIZES + 1);
+	return sz_pind2sz_lookup(pind);
+}
+
+static inline size_t
+sz_psz2u(size_t psz) {
+	if (unlikely(psz > SC_LARGE_MAXCLASS)) {
+		return SC_LARGE_MAXCLASS + PAGE;
+	}
+	size_t x = lg_floor((psz<<1)-1);
+	size_t lg_delta = (x < SC_LG_NGROUP + LG_PAGE + 1) ?
+	    LG_PAGE : x - SC_LG_NGROUP - 1;
+	size_t delta = ZU(1) << lg_delta;
+	size_t delta_mask = delta - 1;
+	size_t usize = (psz + delta_mask) & ~delta_mask;
+	return usize;
+}
+
+static inline szind_t
+sz_size2index_compute(size_t size) {
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		return SC_NSIZES;
+	}
+
+	if (size == 0) {
+		return 0;
+	}
+#if (SC_NTINY != 0)
+	if (size <= (ZU(1) << SC_LG_TINY_MAXCLASS)) {
+		szind_t lg_tmin = SC_LG_TINY_MAXCLASS - SC_NTINY + 1;
+		szind_t lg_ceil = lg_floor(pow2_ceil_zu(size));
+		return (lg_ceil < lg_tmin ? 0 : lg_ceil - lg_tmin);
+	}
+#endif
+	{
+		szind_t x = lg_floor((size<<1)-1);
+		szind_t shift = (x < SC_LG_NGROUP + LG_QUANTUM) ? 0 :
+		    x - (SC_LG_NGROUP + LG_QUANTUM);
+		szind_t grp = shift << SC_LG_NGROUP;
+
+		szind_t lg_delta = (x < SC_LG_NGROUP + LG_QUANTUM + 1)
+		    ? LG_QUANTUM : x - SC_LG_NGROUP - 1;
+
+		size_t delta_inverse_mask = ZU(-1) << lg_delta;
+		szind_t mod = ((((size-1) & delta_inverse_mask) >> lg_delta)) &
+		    ((ZU(1) << SC_LG_NGROUP) - 1);
+
+		szind_t index = SC_NTINY + grp + mod;
+		return index;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+sz_size2index_lookup(size_t size) {
+	assert(size <= SC_LOOKUP_MAXCLASS);
+	szind_t ret = (sz_size2index_tab[(size + (ZU(1) << SC_LG_TINY_MIN) - 1)
+					 >> SC_LG_TINY_MIN]);
+	assert(ret == sz_size2index_compute(size));
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE szind_t
+sz_size2index(size_t size) {
+	if (likely(size <= SC_LOOKUP_MAXCLASS)) {
+		return sz_size2index_lookup(size);
+	}
+	return sz_size2index_compute(size);
+}
+
+static inline size_t
+sz_index2size_compute(szind_t index) {
+#if (SC_NTINY > 0)
+	if (index < SC_NTINY) {
+		return (ZU(1) << (SC_LG_TINY_MAXCLASS - SC_NTINY + 1 + index));
+	}
+#endif
+	{
+		size_t reduced_index = index - SC_NTINY;
+		size_t grp = reduced_index >> SC_LG_NGROUP;
+		size_t mod = reduced_index & ((ZU(1) << SC_LG_NGROUP) -
+		    1);
+
+		size_t grp_size_mask = ~((!!grp)-1);
+		size_t grp_size = ((ZU(1) << (LG_QUANTUM +
+		    (SC_LG_NGROUP-1))) << grp) & grp_size_mask;
+
+		size_t shift = (grp == 0) ? 1 : grp;
+		size_t lg_delta = shift + (LG_QUANTUM-1);
+		size_t mod_size = (mod+1) << lg_delta;
+
+		size_t usize = grp_size + mod_size;
+		return usize;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_index2size_lookup(szind_t index) {
+	size_t ret = (size_t)sz_index2size_tab[index];
+	assert(ret == sz_index2size_compute(index));
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_index2size(szind_t index) {
+	assert(index < SC_NSIZES);
+	return sz_index2size_lookup(index);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u_compute(size_t size) {
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		return 0;
+	}
+
+	if (size == 0) {
+		size++;
+	}
+#if (SC_NTINY > 0)
+	if (size <= (ZU(1) << SC_LG_TINY_MAXCLASS)) {
+		size_t lg_tmin = SC_LG_TINY_MAXCLASS - SC_NTINY + 1;
+		size_t lg_ceil = lg_floor(pow2_ceil_zu(size));
+		return (lg_ceil < lg_tmin ? (ZU(1) << lg_tmin) :
+		    (ZU(1) << lg_ceil));
+	}
+#endif
+	{
+		size_t x = lg_floor((size<<1)-1);
+		size_t lg_delta = (x < SC_LG_NGROUP + LG_QUANTUM + 1)
+		    ?  LG_QUANTUM : x - SC_LG_NGROUP - 1;
+		size_t delta = ZU(1) << lg_delta;
+		size_t delta_mask = delta - 1;
+		size_t usize = (size + delta_mask) & ~delta_mask;
+		return usize;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u_lookup(size_t size) {
+	size_t ret = sz_index2size_lookup(sz_size2index_lookup(size));
+
+	assert(ret == sz_s2u_compute(size));
+	return ret;
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+sz_s2u(size_t size) {
+	if (likely(size <= SC_LOOKUP_MAXCLASS)) {
+		return sz_s2u_lookup(size);
+	}
+	return sz_s2u_compute(size);
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size and alignment.
+ */
+JEMALLOC_ALWAYS_INLINE size_t
+sz_sa2u(size_t size, size_t alignment) {
+	size_t usize;
+
+	assert(alignment != 0 && ((alignment - 1) & alignment) == 0);
+
+	/* Try for a small size class. */
+	if (size <= SC_SMALL_MAXCLASS && alignment < PAGE) {
+		/*
+		 * Round size up to the nearest multiple of alignment.
+		 *
+		 * This done, we can take advantage of the fact that for each
+		 * small size class, every object is aligned at the smallest
+		 * power of two that is non-zero in the base two representation
+		 * of the size.  For example:
+		 *
+		 *   Size |   Base 2 | Minimum alignment
+		 *   -----+----------+------------------
+		 *     96 |  1100000 |  32
+		 *    144 | 10100000 |  32
+		 *    192 | 11000000 |  64
+		 */
+		usize = sz_s2u(ALIGNMENT_CEILING(size, alignment));
+		if (usize < SC_LARGE_MINCLASS) {
+			return usize;
+		}
+	}
+
+	/* Large size class.  Beware of overflow. */
+
+	if (unlikely(alignment > SC_LARGE_MAXCLASS)) {
+		return 0;
+	}
+
+	/* Make sure result is a large size class. */
+	if (size <= SC_LARGE_MINCLASS) {
+		usize = SC_LARGE_MINCLASS;
+	} else {
+		usize = sz_s2u(size);
+		if (usize < size) {
+			/* size_t overflow. */
+			return 0;
+		}
+	}
+
+	/*
+	 * Calculate the multi-page mapping that large_palloc() would need in
+	 * order to guarantee the alignment.
+	 */
+	if (usize + sz_large_pad + PAGE_CEILING(alignment) - PAGE < usize) {
+		/* size_t overflow. */
+		return 0;
+	}
+	return usize;
+}
+
+#endif /* JEMALLOC_INTERNAL_SIZE_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_externs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_externs.h
new file mode 100644
index 0000000..d63eafd
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_externs.h
@@ -0,0 +1,53 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_EXTERNS_H
+#define JEMALLOC_INTERNAL_TCACHE_EXTERNS_H
+
+extern bool	opt_tcache;
+extern ssize_t	opt_lg_tcache_max;
+
+extern cache_bin_info_t	*tcache_bin_info;
+
+/*
+ * Number of tcache bins.  There are SC_NBINS small-object bins, plus 0 or more
+ * large-object bins.
+ */
+extern unsigned	nhbins;
+
+/* Maximum cached size class. */
+extern size_t	tcache_maxclass;
+
+/*
+ * Explicit tcaches, managed via the tcache.{create,flush,destroy} mallctls and
+ * usable via the MALLOCX_TCACHE() flag.  The automatic per thread tcaches are
+ * completely disjoint from this data structure.  tcaches starts off as a sparse
+ * array, so it has no physical memory footprint until individual pages are
+ * touched.  This allows the entire array to be allocated the first time an
+ * explicit tcache is created without a disproportionate impact on memory usage.
+ */
+extern tcaches_t	*tcaches;
+
+size_t	tcache_salloc(tsdn_t *tsdn, const void *ptr);
+void	tcache_event_hard(tsd_t *tsd, tcache_t *tcache);
+void	*tcache_alloc_small_hard(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, bool *tcache_success);
+void	tcache_bin_flush_small(tsd_t *tsd, tcache_t *tcache, cache_bin_t *tbin,
+    szind_t binind, unsigned rem);
+void	tcache_bin_flush_large(tsd_t *tsd, cache_bin_t *tbin, szind_t binind,
+    unsigned rem, tcache_t *tcache);
+void	tcache_arena_reassociate(tsdn_t *tsdn, tcache_t *tcache,
+    arena_t *arena);
+tcache_t *tcache_create_explicit(tsd_t *tsd);
+void	tcache_cleanup(tsd_t *tsd);
+void	tcache_stats_merge(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena);
+bool	tcaches_create(tsd_t *tsd, unsigned *r_ind);
+void	tcaches_flush(tsd_t *tsd, unsigned ind);
+void	tcaches_destroy(tsd_t *tsd, unsigned ind);
+bool	tcache_boot(tsdn_t *tsdn);
+void tcache_arena_associate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena);
+void tcache_prefork(tsdn_t *tsdn);
+void tcache_postfork_parent(tsdn_t *tsdn);
+void tcache_postfork_child(tsdn_t *tsdn);
+void tcache_flush(tsd_t *tsd);
+bool tsd_tcache_data_init(tsd_t *tsd);
+bool tsd_tcache_enabled_data_init(tsd_t *tsd);
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_EXTERNS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_inlines.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_inlines.h
new file mode 100644
index 0000000..5eca20e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_inlines.h
@@ -0,0 +1,227 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_INLINES_H
+#define JEMALLOC_INTERNAL_TCACHE_INLINES_H
+
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/ticker.h"
+#include "jemalloc/internal/util.h"
+
+static inline bool
+tcache_enabled_get(tsd_t *tsd) {
+	return tsd_tcache_enabled_get(tsd);
+}
+
+static inline void
+tcache_enabled_set(tsd_t *tsd, bool enabled) {
+	bool was_enabled = tsd_tcache_enabled_get(tsd);
+
+	if (!was_enabled && enabled) {
+		tsd_tcache_data_init(tsd);
+	} else if (was_enabled && !enabled) {
+		tcache_cleanup(tsd);
+	}
+	/* Commit the state last.  Above calls check current state. */
+	tsd_tcache_enabled_set(tsd, enabled);
+	tsd_slow_update(tsd);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_event(tsd_t *tsd, tcache_t *tcache) {
+	if (TCACHE_GC_INCR == 0) {
+		return;
+	}
+
+	if (unlikely(ticker_tick(&tcache->gc_ticker))) {
+		tcache_event_hard(tsd, tcache);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+tcache_alloc_small(tsd_t *tsd, arena_t *arena, tcache_t *tcache,
+    size_t size, szind_t binind, bool zero, bool slow_path) {
+	void *ret;
+	cache_bin_t *bin;
+	bool tcache_success;
+	size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+
+	assert(binind < SC_NBINS);
+	bin = tcache_small_bin_get(tcache, binind);
+	ret = cache_bin_alloc_easy(bin, &tcache_success);
+	assert(tcache_success == (ret != NULL));
+	if (unlikely(!tcache_success)) {
+		bool tcache_hard_success;
+		arena = arena_choose(tsd, arena);
+		if (unlikely(arena == NULL)) {
+			return NULL;
+		}
+
+		ret = tcache_alloc_small_hard(tsd_tsdn(tsd), arena, tcache,
+		    bin, binind, &tcache_hard_success);
+		if (tcache_hard_success == false) {
+			return NULL;
+		}
+	}
+
+	assert(ret);
+	/*
+	 * Only compute usize if required.  The checks in the following if
+	 * statement are all static.
+	 */
+	if (config_prof || (slow_path && config_fill) || unlikely(zero)) {
+		usize = sz_index2size(binind);
+		assert(tcache_salloc(tsd_tsdn(tsd), ret) == usize);
+	}
+
+	if (likely(!zero)) {
+		if (slow_path && config_fill) {
+			if (unlikely(opt_junk_alloc)) {
+				arena_alloc_junk_small(ret, &bin_infos[binind],
+				    false);
+			} else if (unlikely(opt_zero)) {
+				memset(ret, 0, usize);
+			}
+		}
+	} else {
+		if (slow_path && config_fill && unlikely(opt_junk_alloc)) {
+			arena_alloc_junk_small(ret, &bin_infos[binind], true);
+		}
+		memset(ret, 0, usize);
+	}
+
+	if (config_stats) {
+		bin->tstats.nrequests++;
+	}
+	if (config_prof) {
+		tcache->prof_accumbytes += usize;
+	}
+	tcache_event(tsd, tcache);
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+tcache_alloc_large(tsd_t *tsd, arena_t *arena, tcache_t *tcache, size_t size,
+    szind_t binind, bool zero, bool slow_path) {
+	void *ret;
+	cache_bin_t *bin;
+	bool tcache_success;
+
+	assert(binind >= SC_NBINS &&binind < nhbins);
+	bin = tcache_large_bin_get(tcache, binind);
+	ret = cache_bin_alloc_easy(bin, &tcache_success);
+	assert(tcache_success == (ret != NULL));
+	if (unlikely(!tcache_success)) {
+		/*
+		 * Only allocate one large object at a time, because it's quite
+		 * expensive to create one and not use it.
+		 */
+		arena = arena_choose(tsd, arena);
+		if (unlikely(arena == NULL)) {
+			return NULL;
+		}
+
+		ret = large_malloc(tsd_tsdn(tsd), arena, sz_s2u(size), zero);
+		if (ret == NULL) {
+			return NULL;
+		}
+	} else {
+		size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+
+		/* Only compute usize on demand */
+		if (config_prof || (slow_path && config_fill) ||
+		    unlikely(zero)) {
+			usize = sz_index2size(binind);
+			assert(usize <= tcache_maxclass);
+		}
+
+		if (likely(!zero)) {
+			if (slow_path && config_fill) {
+				if (unlikely(opt_junk_alloc)) {
+					memset(ret, JEMALLOC_ALLOC_JUNK,
+					    usize);
+				} else if (unlikely(opt_zero)) {
+					memset(ret, 0, usize);
+				}
+			}
+		} else {
+			memset(ret, 0, usize);
+		}
+
+		if (config_stats) {
+			bin->tstats.nrequests++;
+		}
+		if (config_prof) {
+			tcache->prof_accumbytes += usize;
+		}
+	}
+
+	tcache_event(tsd, tcache);
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_dalloc_small(tsd_t *tsd, tcache_t *tcache, void *ptr, szind_t binind,
+    bool slow_path) {
+	cache_bin_t *bin;
+	cache_bin_info_t *bin_info;
+
+	assert(tcache_salloc(tsd_tsdn(tsd), ptr)
+	    <= SC_SMALL_MAXCLASS);
+
+	if (slow_path && config_fill && unlikely(opt_junk_free)) {
+		arena_dalloc_junk_small(ptr, &bin_infos[binind]);
+	}
+
+	bin = tcache_small_bin_get(tcache, binind);
+	bin_info = &tcache_bin_info[binind];
+	if (unlikely(!cache_bin_dalloc_easy(bin, bin_info, ptr))) {
+		tcache_bin_flush_small(tsd, tcache, bin, binind,
+		    (bin_info->ncached_max >> 1));
+		bool ret = cache_bin_dalloc_easy(bin, bin_info, ptr);
+		assert(ret);
+	}
+
+	tcache_event(tsd, tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tcache_dalloc_large(tsd_t *tsd, tcache_t *tcache, void *ptr, szind_t binind,
+    bool slow_path) {
+	cache_bin_t *bin;
+	cache_bin_info_t *bin_info;
+
+	assert(tcache_salloc(tsd_tsdn(tsd), ptr)
+	    > SC_SMALL_MAXCLASS);
+	assert(tcache_salloc(tsd_tsdn(tsd), ptr) <= tcache_maxclass);
+
+	if (slow_path && config_fill && unlikely(opt_junk_free)) {
+		large_dalloc_junk(ptr, sz_index2size(binind));
+	}
+
+	bin = tcache_large_bin_get(tcache, binind);
+	bin_info = &tcache_bin_info[binind];
+	if (unlikely(bin->ncached == bin_info->ncached_max)) {
+		tcache_bin_flush_large(tsd, bin, binind,
+		    (bin_info->ncached_max >> 1), tcache);
+	}
+	assert(bin->ncached < bin_info->ncached_max);
+	bin->ncached++;
+	*(bin->avail - bin->ncached) = ptr;
+
+	tcache_event(tsd, tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE tcache_t *
+tcaches_get(tsd_t *tsd, unsigned ind) {
+	tcaches_t *elm = &tcaches[ind];
+	if (unlikely(elm->tcache == NULL)) {
+		malloc_printf("<jemalloc>: invalid tcache id (%u).\n", ind);
+		abort();
+	} else if (unlikely(elm->tcache == TCACHES_ELM_NEED_REINIT)) {
+		elm->tcache = tcache_create_explicit(tsd);
+	}
+	return elm->tcache;
+}
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_INLINES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_structs.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_structs.h
new file mode 100644
index 0000000..172ef90
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_structs.h
@@ -0,0 +1,70 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_STRUCTS_H
+#define JEMALLOC_INTERNAL_TCACHE_STRUCTS_H
+
+#include "jemalloc/internal/cache_bin.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/ticker.h"
+#include "jemalloc/internal/tsd_types.h"
+
+/* Various uses of this struct need it to be a named type. */
+typedef ql_elm(tsd_t) tsd_link_t;
+
+struct tcache_s {
+	/*
+	 * To minimize our cache-footprint, we put the frequently accessed data
+	 * together at the start of this struct.
+	 */
+
+	/* Cleared after arena_prof_accum(). */
+	uint64_t	prof_accumbytes;
+	/* Drives incremental GC. */
+	ticker_t	gc_ticker;
+	/*
+	 * The pointer stacks associated with bins follow as a contiguous array.
+	 * During tcache initialization, the avail pointer in each element of
+	 * tbins is initialized to point to the proper offset within this array.
+	 */
+	cache_bin_t	bins_small[SC_NBINS];
+
+	/*
+	 * This data is less hot; we can be a little less careful with our
+	 * footprint here.
+	 */
+	/* Lets us track all the tcaches in an arena. */
+	ql_elm(tcache_t) link;
+
+	/* Logically scoped to tsd, but put here for cache layout reasons. */
+	ql_elm(tsd_t) tsd_link;
+	bool in_hook;
+
+	/*
+	 * The descriptor lets the arena find our cache bins without seeing the
+	 * tcache definition.  This enables arenas to aggregate stats across
+	 * tcaches without having a tcache dependency.
+	 */
+	cache_bin_array_descriptor_t cache_bin_array_descriptor;
+
+	/* The arena this tcache is associated with. */
+	arena_t		*arena;
+	/* Next bin to GC. */
+	szind_t		next_gc_bin;
+	/* For small bins, fill (ncached_max >> lg_fill_div). */
+	uint8_t		lg_fill_div[SC_NBINS];
+	/*
+	 * We put the cache bins for large size classes at the end of the
+	 * struct, since some of them might not get used.  This might end up
+	 * letting us avoid touching an extra page if we don't have to.
+	 */
+	cache_bin_t	bins_large[SC_NSIZES-SC_NBINS];
+};
+
+/* Linkage for list of available (previously used) explicit tcache IDs. */
+struct tcaches_s {
+	union {
+		tcache_t	*tcache;
+		tcaches_t	*next;
+	};
+};
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_STRUCTS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_types.h
new file mode 100644
index 0000000..dce6938
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tcache_types.h
@@ -0,0 +1,59 @@
+#ifndef JEMALLOC_INTERNAL_TCACHE_TYPES_H
+#define JEMALLOC_INTERNAL_TCACHE_TYPES_H
+
+#include "jemalloc/internal/sc.h"
+
+typedef struct tcache_s tcache_t;
+typedef struct tcaches_s tcaches_t;
+
+/*
+ * tcache pointers close to NULL are used to encode state information that is
+ * used for two purposes: preventing thread caching on a per thread basis and
+ * cleaning up during thread shutdown.
+ */
+#define TCACHE_STATE_DISABLED		((tcache_t *)(uintptr_t)1)
+#define TCACHE_STATE_REINCARNATED	((tcache_t *)(uintptr_t)2)
+#define TCACHE_STATE_PURGATORY		((tcache_t *)(uintptr_t)3)
+#define TCACHE_STATE_MAX		TCACHE_STATE_PURGATORY
+
+/*
+ * Absolute minimum number of cache slots for each small bin.
+ */
+#define TCACHE_NSLOTS_SMALL_MIN		20
+
+/*
+ * Absolute maximum number of cache slots for each small bin in the thread
+ * cache.  This is an additional constraint beyond that imposed as: twice the
+ * number of regions per slab for this size class.
+ *
+ * This constant must be an even number.
+ */
+#define TCACHE_NSLOTS_SMALL_MAX		200
+
+/* Number of cache slots for large size classes. */
+#define TCACHE_NSLOTS_LARGE		20
+
+/* (1U << opt_lg_tcache_max) is used to compute tcache_maxclass. */
+#define LG_TCACHE_MAXCLASS_DEFAULT	15
+
+/*
+ * TCACHE_GC_SWEEP is the approximate number of allocation events between
+ * full GC sweeps.  Integer rounding may cause the actual number to be
+ * slightly higher, since GC is performed incrementally.
+ */
+#define TCACHE_GC_SWEEP			8192
+
+/* Number of tcache allocation/deallocation events between incremental GCs. */
+#define TCACHE_GC_INCR							\
+    ((TCACHE_GC_SWEEP / SC_NBINS) + ((TCACHE_GC_SWEEP / SC_NBINS == 0) ? 0 : 1))
+
+/* Used in TSD static initializer only. Real init in tcache_data_init(). */
+#define TCACHE_ZERO_INITIALIZER {0}
+
+/* Used in TSD static initializer only. Will be initialized to opt_tcache. */
+#define TCACHE_ENABLED_ZERO_INITIALIZER false
+
+/* Used for explicit tcache only. Means flushed but not destroyed. */
+#define TCACHES_ELM_NEED_REINIT ((tcache_t *)(uintptr_t)1)
+
+#endif /* JEMALLOC_INTERNAL_TCACHE_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/test_hooks.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/test_hooks.h
new file mode 100644
index 0000000..a6351e5
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/test_hooks.h
@@ -0,0 +1,19 @@
+#ifndef JEMALLOC_INTERNAL_TEST_HOOKS_H
+#define JEMALLOC_INTERNAL_TEST_HOOKS_H
+
+extern JEMALLOC_EXPORT void (*test_hooks_arena_new_hook)();
+extern JEMALLOC_EXPORT void (*test_hooks_libc_hook)();
+
+#define JEMALLOC_HOOK(fn, hook) ((void)(hook != NULL && (hook(), 0)), fn)
+
+#define open JEMALLOC_HOOK(open, test_hooks_libc_hook)
+#define read JEMALLOC_HOOK(read, test_hooks_libc_hook)
+#define write JEMALLOC_HOOK(write, test_hooks_libc_hook)
+#define readlink JEMALLOC_HOOK(readlink, test_hooks_libc_hook)
+#define close JEMALLOC_HOOK(close, test_hooks_libc_hook)
+#define creat JEMALLOC_HOOK(creat, test_hooks_libc_hook)
+#define secure_getenv JEMALLOC_HOOK(secure_getenv, test_hooks_libc_hook)
+/* Note that this is undef'd and re-define'd in src/prof.c. */
+#define _Unwind_Backtrace JEMALLOC_HOOK(_Unwind_Backtrace, test_hooks_libc_hook)
+
+#endif /* JEMALLOC_INTERNAL_TEST_HOOKS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/ticker.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ticker.h
new file mode 100644
index 0000000..0357800
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/ticker.h
@@ -0,0 +1,91 @@
+#ifndef JEMALLOC_INTERNAL_TICKER_H
+#define JEMALLOC_INTERNAL_TICKER_H
+
+#include "jemalloc/internal/util.h"
+
+/**
+ * A ticker makes it easy to count-down events until some limit.  You
+ * ticker_init the ticker to trigger every nticks events.  You then notify it
+ * that an event has occurred with calls to ticker_tick (or that nticks events
+ * have occurred with a call to ticker_ticks), which will return true (and reset
+ * the counter) if the countdown hit zero.
+ */
+
+typedef struct {
+	int32_t tick;
+	int32_t nticks;
+} ticker_t;
+
+static inline void
+ticker_init(ticker_t *ticker, int32_t nticks) {
+	ticker->tick = nticks;
+	ticker->nticks = nticks;
+}
+
+static inline void
+ticker_copy(ticker_t *ticker, const ticker_t *other) {
+	*ticker = *other;
+}
+
+static inline int32_t
+ticker_read(const ticker_t *ticker) {
+	return ticker->tick;
+}
+
+/*
+ * Not intended to be a public API.  Unfortunately, on x86, neither gcc nor
+ * clang seems smart enough to turn
+ *   ticker->tick -= nticks;
+ *   if (unlikely(ticker->tick < 0)) {
+ *     fixup ticker
+ *     return true;
+ *   }
+ *   return false;
+ * into
+ *   subq %nticks_reg, (%ticker_reg)
+ *   js fixup ticker
+ *
+ * unless we force "fixup ticker" out of line.  In that case, gcc gets it right,
+ * but clang now does worse than before.  So, on x86 with gcc, we force it out
+ * of line, but otherwise let the inlining occur.  Ordinarily this wouldn't be
+ * worth the hassle, but this is on the fast path of both malloc and free (via
+ * tcache_event).
+ */
+#if defined(__GNUC__) && !defined(__clang__)				\
+    && (defined(__x86_64__) || defined(__i386__))
+JEMALLOC_NOINLINE
+#endif
+static bool
+ticker_fixup(ticker_t *ticker) {
+	ticker->tick = ticker->nticks;
+	return true;
+}
+
+static inline bool
+ticker_ticks(ticker_t *ticker, int32_t nticks) {
+	ticker->tick -= nticks;
+	if (unlikely(ticker->tick < 0)) {
+		return ticker_fixup(ticker);
+	}
+	return false;
+}
+
+static inline bool
+ticker_tick(ticker_t *ticker) {
+	return ticker_ticks(ticker, 1);
+}
+
+/*
+ * Try to tick.  If ticker would fire, return true, but rely on
+ * slowpath to reset ticker.
+ */
+static inline bool
+ticker_trytick(ticker_t *ticker) {
+	--ticker->tick;
+	if (unlikely(ticker->tick < 0)) {
+		return true;
+	}
+	return false;
+}
+
+#endif /* JEMALLOC_INTERNAL_TICKER_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd.h
new file mode 100644
index 0000000..9ba2600
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd.h
@@ -0,0 +1,415 @@
+#ifndef JEMALLOC_INTERNAL_TSD_H
+#define JEMALLOC_INTERNAL_TSD_H
+
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/bin_types.h"
+#include "jemalloc/internal/jemalloc_internal_externs.h"
+#include "jemalloc/internal/prof_types.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/rtree_tsd.h"
+#include "jemalloc/internal/tcache_types.h"
+#include "jemalloc/internal/tcache_structs.h"
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/witness.h"
+
+/*
+ * Thread-Specific-Data layout
+ * --- data accessed on tcache fast path: state, rtree_ctx, stats, prof ---
+ * s: state
+ * e: tcache_enabled
+ * m: thread_allocated (config_stats)
+ * f: thread_deallocated (config_stats)
+ * p: prof_tdata (config_prof)
+ * c: rtree_ctx (rtree cache accessed on deallocation)
+ * t: tcache
+ * --- data not accessed on tcache fast path: arena-related fields ---
+ * d: arenas_tdata_bypass
+ * r: reentrancy_level
+ * x: narenas_tdata
+ * i: iarena
+ * a: arena
+ * o: arenas_tdata
+ * Loading TSD data is on the critical path of basically all malloc operations.
+ * In particular, tcache and rtree_ctx rely on hot CPU cache to be effective.
+ * Use a compact layout to reduce cache footprint.
+ * +--- 64-bit and 64B cacheline; 1B each letter; First byte on the left. ---+
+ * |----------------------------  1st cacheline  ----------------------------|
+ * | sedrxxxx mmmmmmmm ffffffff pppppppp [c * 32  ........ ........ .......] |
+ * |----------------------------  2nd cacheline  ----------------------------|
+ * | [c * 64  ........ ........ ........ ........ ........ ........ .......] |
+ * |----------------------------  3nd cacheline  ----------------------------|
+ * | [c * 32  ........ ........ .......] iiiiiiii aaaaaaaa oooooooo [t...... |
+ * +-------------------------------------------------------------------------+
+ * Note: the entire tcache is embedded into TSD and spans multiple cachelines.
+ *
+ * The last 3 members (i, a and o) before tcache isn't really needed on tcache
+ * fast path.  However we have a number of unused tcache bins and witnesses
+ * (never touched unless config_debug) at the end of tcache, so we place them
+ * there to avoid breaking the cachelines and possibly paging in an extra page.
+ */
+#ifdef JEMALLOC_JET
+typedef void (*test_callback_t)(int *);
+#  define MALLOC_TSD_TEST_DATA_INIT 0x72b65c10
+#  define MALLOC_TEST_TSD \
+    O(test_data,		int,			int)		\
+    O(test_callback,		test_callback_t,	int)
+#  define MALLOC_TEST_TSD_INITIALIZER , MALLOC_TSD_TEST_DATA_INIT, NULL
+#else
+#  define MALLOC_TEST_TSD
+#  define MALLOC_TEST_TSD_INITIALIZER
+#endif
+
+/*  O(name,			type,			nullable type */
+#define MALLOC_TSD							\
+    O(tcache_enabled,		bool,			bool)		\
+    O(arenas_tdata_bypass,	bool,			bool)		\
+    O(reentrancy_level,		int8_t,			int8_t)		\
+    O(narenas_tdata,		uint32_t,		uint32_t)	\
+    O(offset_state,		uint64_t,		uint64_t)	\
+    O(thread_allocated,		uint64_t,		uint64_t)	\
+    O(thread_deallocated,	uint64_t,		uint64_t)	\
+    O(bytes_until_sample,	int64_t,		int64_t)	\
+    O(prof_tdata,		prof_tdata_t *,		prof_tdata_t *)	\
+    O(rtree_ctx,		rtree_ctx_t,		rtree_ctx_t)	\
+    O(iarena,			arena_t *,		arena_t *)	\
+    O(arena,			arena_t *,		arena_t *)	\
+    O(arenas_tdata,		arena_tdata_t *,	arena_tdata_t *)\
+    O(binshards,		tsd_binshards_t,	tsd_binshards_t)\
+    O(tcache,			tcache_t,		tcache_t)	\
+    O(witness_tsd,              witness_tsd_t,		witness_tsdn_t)	\
+    MALLOC_TEST_TSD
+
+#define TSD_INITIALIZER {						\
+    ATOMIC_INIT(tsd_state_uninitialized),				\
+    TCACHE_ENABLED_ZERO_INITIALIZER,					\
+    false,								\
+    0,									\
+    0,									\
+    0,									\
+    0,									\
+    0,									\
+    0,									\
+    NULL,								\
+    RTREE_CTX_ZERO_INITIALIZER,						\
+    NULL,								\
+    NULL,								\
+    NULL,								\
+    TSD_BINSHARDS_ZERO_INITIALIZER,					\
+    TCACHE_ZERO_INITIALIZER,						\
+    WITNESS_TSD_INITIALIZER						\
+    MALLOC_TEST_TSD_INITIALIZER						\
+}
+
+void *malloc_tsd_malloc(size_t size);
+void malloc_tsd_dalloc(void *wrapper);
+void malloc_tsd_cleanup_register(bool (*f)(void));
+tsd_t *malloc_tsd_boot0(void);
+void malloc_tsd_boot1(void);
+void tsd_cleanup(void *arg);
+tsd_t *tsd_fetch_slow(tsd_t *tsd, bool internal);
+void tsd_state_set(tsd_t *tsd, uint8_t new_state);
+void tsd_slow_update(tsd_t *tsd);
+void tsd_prefork(tsd_t *tsd);
+void tsd_postfork_parent(tsd_t *tsd);
+void tsd_postfork_child(tsd_t *tsd);
+
+/*
+ * Call ..._inc when your module wants to take all threads down the slow paths,
+ * and ..._dec when it no longer needs to.
+ */
+void tsd_global_slow_inc(tsdn_t *tsdn);
+void tsd_global_slow_dec(tsdn_t *tsdn);
+bool tsd_global_slow();
+
+enum {
+	/* Common case --> jnz. */
+	tsd_state_nominal = 0,
+	/* Initialized but on slow path. */
+	tsd_state_nominal_slow = 1,
+	/*
+	 * Some thread has changed global state in such a way that all nominal
+	 * threads need to recompute their fast / slow status the next time they
+	 * get a chance.
+	 *
+	 * Any thread can change another thread's status *to* recompute, but
+	 * threads are the only ones who can change their status *from*
+	 * recompute.
+	 */
+	tsd_state_nominal_recompute = 2,
+	/*
+	 * The above nominal states should be lower values.  We use
+	 * tsd_nominal_max to separate nominal states from threads in the
+	 * process of being born / dying.
+	 */
+	tsd_state_nominal_max = 2,
+
+	/*
+	 * A thread might free() during its death as its only allocator action;
+	 * in such scenarios, we need tsd, but set up in such a way that no
+	 * cleanup is necessary.
+	 */
+	tsd_state_minimal_initialized = 3,
+	/* States during which we know we're in thread death. */
+	tsd_state_purgatory = 4,
+	tsd_state_reincarnated = 5,
+	/*
+	 * What it says on the tin; tsd that hasn't been initialized.  Note
+	 * that even when the tsd struct lives in TLS, when need to keep track
+	 * of stuff like whether or not our pthread destructors have been
+	 * scheduled, so this really truly is different than the nominal state.
+	 */
+	tsd_state_uninitialized = 6
+};
+
+/*
+ * Some TSD accesses can only be done in a nominal state.  To enforce this, we
+ * wrap TSD member access in a function that asserts on TSD state, and mangle
+ * field names to prevent touching them accidentally.
+ */
+#define TSD_MANGLE(n) cant_access_tsd_items_directly_use_a_getter_or_setter_##n
+
+#ifdef JEMALLOC_U8_ATOMICS
+#  define tsd_state_t atomic_u8_t
+#  define tsd_atomic_load atomic_load_u8
+#  define tsd_atomic_store atomic_store_u8
+#  define tsd_atomic_exchange atomic_exchange_u8
+#else
+#  define tsd_state_t atomic_u32_t
+#  define tsd_atomic_load atomic_load_u32
+#  define tsd_atomic_store atomic_store_u32
+#  define tsd_atomic_exchange atomic_exchange_u32
+#endif
+
+/* The actual tsd. */
+struct tsd_s {
+	/*
+	 * The contents should be treated as totally opaque outside the tsd
+	 * module.  Access any thread-local state through the getters and
+	 * setters below.
+	 */
+
+	/*
+	 * We manually limit the state to just a single byte.  Unless the 8-bit
+	 * atomics are unavailable (which is rare).
+	 */
+	tsd_state_t state;
+#define O(n, t, nt)							\
+	t TSD_MANGLE(n);
+MALLOC_TSD
+#undef O
+};
+
+JEMALLOC_ALWAYS_INLINE uint8_t
+tsd_state_get(tsd_t *tsd) {
+	/*
+	 * This should be atomic.  Unfortunately, compilers right now can't tell
+	 * that this can be done as a memory comparison, and forces a load into
+	 * a register that hurts fast-path performance.
+	 */
+	/* return atomic_load_u8(&tsd->state, ATOMIC_RELAXED); */
+	return *(uint8_t *)&tsd->state;
+}
+
+/*
+ * Wrapper around tsd_t that makes it possible to avoid implicit conversion
+ * between tsd_t and tsdn_t, where tsdn_t is "nullable" and has to be
+ * explicitly converted to tsd_t, which is non-nullable.
+ */
+struct tsdn_s {
+	tsd_t tsd;
+};
+#define TSDN_NULL ((tsdn_t *)0)
+JEMALLOC_ALWAYS_INLINE tsdn_t *
+tsd_tsdn(tsd_t *tsd) {
+	return (tsdn_t *)tsd;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsdn_null(const tsdn_t *tsdn) {
+	return tsdn == NULL;
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsdn_tsd(tsdn_t *tsdn) {
+	assert(!tsdn_null(tsdn));
+
+	return &tsdn->tsd;
+}
+
+/*
+ * We put the platform-specific data declarations and inlines into their own
+ * header files to avoid cluttering this file.  They define tsd_boot0,
+ * tsd_boot1, tsd_boot, tsd_booted_get, tsd_get_allocates, tsd_get, and tsd_set.
+ */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#include "jemalloc/internal/tsd_malloc_thread_cleanup.h"
+#elif (defined(JEMALLOC_TLS))
+#include "jemalloc/internal/tsd_tls.h"
+#elif (defined(_WIN32))
+#include "jemalloc/internal/tsd_win.h"
+#else
+#include "jemalloc/internal/tsd_generic.h"
+#endif
+
+/*
+ * tsd_foop_get_unsafe(tsd) returns a pointer to the thread-local instance of
+ * foo.  This omits some safety checks, and so can be used during tsd
+ * initialization and cleanup.
+ */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE t *						\
+tsd_##n##p_get_unsafe(tsd_t *tsd) {					\
+	return &tsd->TSD_MANGLE(n);					\
+}
+MALLOC_TSD
+#undef O
+
+/* tsd_foop_get(tsd) returns a pointer to the thread-local instance of foo. */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE t *						\
+tsd_##n##p_get(tsd_t *tsd) {						\
+	/*								\
+	 * Because the state might change asynchronously if it's	\
+	 * nominal, we need to make sure that we only read it once.	\
+	 */								\
+	uint8_t state = tsd_state_get(tsd);				\
+	assert(state == tsd_state_nominal ||				\
+	    state == tsd_state_nominal_slow ||				\
+	    state == tsd_state_nominal_recompute ||			\
+	    state == tsd_state_reincarnated ||				\
+	    state == tsd_state_minimal_initialized);			\
+	return tsd_##n##p_get_unsafe(tsd);				\
+}
+MALLOC_TSD
+#undef O
+
+/*
+ * tsdn_foop_get(tsdn) returns either the thread-local instance of foo (if tsdn
+ * isn't NULL), or NULL (if tsdn is NULL), cast to the nullable pointer type.
+ */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE nt *						\
+tsdn_##n##p_get(tsdn_t *tsdn) {						\
+	if (tsdn_null(tsdn)) {						\
+		return NULL;						\
+	}								\
+	tsd_t *tsd = tsdn_tsd(tsdn);					\
+	return (nt *)tsd_##n##p_get(tsd);				\
+}
+MALLOC_TSD
+#undef O
+
+/* tsd_foo_get(tsd) returns the value of the thread-local instance of foo. */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE t						\
+tsd_##n##_get(tsd_t *tsd) {						\
+	return *tsd_##n##p_get(tsd);					\
+}
+MALLOC_TSD
+#undef O
+
+/* tsd_foo_set(tsd, val) updates the thread-local instance of foo to be val. */
+#define O(n, t, nt)							\
+JEMALLOC_ALWAYS_INLINE void						\
+tsd_##n##_set(tsd_t *tsd, t val) {					\
+	assert(tsd_state_get(tsd) != tsd_state_reincarnated &&		\
+	    tsd_state_get(tsd) != tsd_state_minimal_initialized);	\
+	*tsd_##n##p_get(tsd) = val;					\
+}
+MALLOC_TSD
+#undef O
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_assert_fast(tsd_t *tsd) {
+	/*
+	 * Note that our fastness assertion does *not* include global slowness
+	 * counters; it's not in general possible to ensure that they won't
+	 * change asynchronously from underneath us.
+	 */
+	assert(!malloc_slow && tsd_tcache_enabled_get(tsd) &&
+	    tsd_reentrancy_level_get(tsd) == 0);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_fast(tsd_t *tsd) {
+	bool fast = (tsd_state_get(tsd) == tsd_state_nominal);
+	if (fast) {
+		tsd_assert_fast(tsd);
+	}
+
+	return fast;
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_fetch_impl(bool init, bool minimal) {
+	tsd_t *tsd = tsd_get(init);
+
+	if (!init && tsd_get_allocates() && tsd == NULL) {
+		return NULL;
+	}
+	assert(tsd != NULL);
+
+	if (unlikely(tsd_state_get(tsd) != tsd_state_nominal)) {
+		return tsd_fetch_slow(tsd, minimal);
+	}
+	assert(tsd_fast(tsd));
+	tsd_assert_fast(tsd);
+
+	return tsd;
+}
+
+/* Get a minimal TSD that requires no cleanup.  See comments in free(). */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_fetch_min(void) {
+	return tsd_fetch_impl(true, true);
+}
+
+/* For internal background threads use only. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_internal_fetch(void) {
+	tsd_t *tsd = tsd_fetch_min();
+	/* Use reincarnated state to prevent full initialization. */
+	tsd_state_set(tsd, tsd_state_reincarnated);
+
+	return tsd;
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_fetch(void) {
+	return tsd_fetch_impl(true, false);
+}
+
+static inline bool
+tsd_nominal(tsd_t *tsd) {
+	return (tsd_state_get(tsd) <= tsd_state_nominal_max);
+}
+
+JEMALLOC_ALWAYS_INLINE tsdn_t *
+tsdn_fetch(void) {
+	if (!tsd_booted_get()) {
+		return NULL;
+	}
+
+	return tsd_tsdn(tsd_fetch_impl(false, false));
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_ctx_t *
+tsd_rtree_ctx(tsd_t *tsd) {
+	return tsd_rtree_ctxp_get(tsd);
+}
+
+JEMALLOC_ALWAYS_INLINE rtree_ctx_t *
+tsdn_rtree_ctx(tsdn_t *tsdn, rtree_ctx_t *fallback) {
+	/*
+	 * If tsd cannot be accessed, initialize the fallback rtree_ctx and
+	 * return a pointer to it.
+	 */
+	if (unlikely(tsdn_null(tsdn))) {
+		rtree_ctx_data_init(fallback);
+		return fallback;
+	}
+	return tsd_rtree_ctx(tsdn_tsd(tsdn));
+}
+
+#endif /* JEMALLOC_INTERNAL_TSD_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_generic.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_generic.h
new file mode 100644
index 0000000..cf73c0c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_generic.h
@@ -0,0 +1,163 @@
+#ifdef JEMALLOC_INTERNAL_TSD_GENERIC_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_GENERIC_H
+
+typedef struct tsd_init_block_s tsd_init_block_t;
+struct tsd_init_block_s {
+	ql_elm(tsd_init_block_t) link;
+	pthread_t thread;
+	void *data;
+};
+
+/* Defined in tsd.c, to allow the mutex headers to have tsd dependencies. */
+typedef struct tsd_init_head_s tsd_init_head_t;
+
+typedef struct {
+	bool initialized;
+	tsd_t val;
+} tsd_wrapper_t;
+
+void *tsd_init_check_recursion(tsd_init_head_t *head,
+    tsd_init_block_t *block);
+void tsd_init_finish(tsd_init_head_t *head, tsd_init_block_t *block);
+
+extern pthread_key_t tsd_tsd;
+extern tsd_init_head_t tsd_init_head;
+extern tsd_wrapper_t tsd_boot_wrapper;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE void
+tsd_cleanup_wrapper(void *arg) {
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *)arg;
+
+	if (wrapper->initialized) {
+		wrapper->initialized = false;
+		tsd_cleanup(&wrapper->val);
+		if (wrapper->initialized) {
+			/* Trigger another cleanup round. */
+			if (pthread_setspecific(tsd_tsd, (void *)wrapper) != 0)
+			{
+				malloc_write("<jemalloc>: Error setting TSD\n");
+				if (opt_abort) {
+					abort();
+				}
+			}
+			return;
+		}
+	}
+	malloc_tsd_dalloc(wrapper);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_wrapper_set(tsd_wrapper_t *wrapper) {
+	if (pthread_setspecific(tsd_tsd, (void *)wrapper) != 0) {
+		malloc_write("<jemalloc>: Error setting TSD\n");
+		abort();
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_wrapper_t *
+tsd_wrapper_get(bool init) {
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *)pthread_getspecific(tsd_tsd);
+
+	if (init && unlikely(wrapper == NULL)) {
+		tsd_init_block_t block;
+		wrapper = (tsd_wrapper_t *)
+		    tsd_init_check_recursion(&tsd_init_head, &block);
+		if (wrapper) {
+			return wrapper;
+		}
+		wrapper = (tsd_wrapper_t *)
+		    malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+		block.data = (void *)wrapper;
+		if (wrapper == NULL) {
+			malloc_write("<jemalloc>: Error allocating TSD\n");
+			abort();
+		} else {
+			wrapper->initialized = false;
+      JEMALLOC_DIAGNOSTIC_PUSH
+      JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+			tsd_t initializer = TSD_INITIALIZER;
+      JEMALLOC_DIAGNOSTIC_POP
+			wrapper->val = initializer;
+		}
+		tsd_wrapper_set(wrapper);
+		tsd_init_finish(&tsd_init_head, &block);
+	}
+	return wrapper;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	if (pthread_key_create(&tsd_tsd, tsd_cleanup_wrapper) != 0) {
+		return true;
+	}
+	tsd_wrapper_set(&tsd_boot_wrapper);
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	tsd_wrapper_t *wrapper;
+	wrapper = (tsd_wrapper_t *)malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+	if (wrapper == NULL) {
+		malloc_write("<jemalloc>: Error allocating TSD\n");
+		abort();
+	}
+	tsd_boot_wrapper.initialized = false;
+	tsd_cleanup(&tsd_boot_wrapper.val);
+	wrapper->initialized = false;
+  JEMALLOC_DIAGNOSTIC_PUSH
+  JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+	tsd_t initializer = TSD_INITIALIZER;
+  JEMALLOC_DIAGNOSTIC_POP
+	wrapper->val = initializer;
+	tsd_wrapper_set(wrapper);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	if (tsd_boot0()) {
+		return true;
+	}
+	tsd_boot1();
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return true;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(init);
+	if (tsd_get_allocates() && !init && wrapper == NULL) {
+		return NULL;
+	}
+	return &wrapper->val;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(true);
+	if (likely(&wrapper->val != val)) {
+		wrapper->val = *(val);
+	}
+	wrapper->initialized = true;
+}
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
new file mode 100644
index 0000000..65852d5
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_malloc_thread_cleanup.h
@@ -0,0 +1,61 @@
+#ifdef JEMALLOC_INTERNAL_TSD_MALLOC_THREAD_CLEANUP_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_MALLOC_THREAD_CLEANUP_H
+
+#define JEMALLOC_TSD_TYPE_ATTR(type) __thread type JEMALLOC_TLS_MODEL
+
+extern JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls;
+extern JEMALLOC_TSD_TYPE_ATTR(bool) tsd_initialized;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE bool
+tsd_cleanup_wrapper(void) {
+	if (tsd_initialized) {
+		tsd_initialized = false;
+		tsd_cleanup(&tsd_tls);
+	}
+	return tsd_initialized;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	malloc_tsd_cleanup_register(&tsd_cleanup_wrapper);
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	/* Do nothing. */
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	return tsd_boot0();
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return false;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	return &tsd_tls;
+}
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	assert(tsd_booted);
+	if (likely(&tsd_tls != val)) {
+		tsd_tls = (*val);
+	}
+	tsd_initialized = true;
+}
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_tls.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_tls.h
new file mode 100644
index 0000000..7d6c805
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_tls.h
@@ -0,0 +1,60 @@
+#ifdef JEMALLOC_INTERNAL_TSD_TLS_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_TLS_H
+
+#define JEMALLOC_TSD_TYPE_ATTR(type) __thread type JEMALLOC_TLS_MODEL
+
+extern JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls;
+extern pthread_key_t tsd_tsd;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	if (pthread_key_create(&tsd_tsd, &tsd_cleanup) != 0) {
+		return true;
+	}
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	/* Do nothing. */
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	return tsd_boot0();
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return false;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	return &tsd_tls;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	assert(tsd_booted);
+	if (likely(&tsd_tls != val)) {
+		tsd_tls = (*val);
+	}
+	if (pthread_setspecific(tsd_tsd, (void *)(&tsd_tls)) != 0) {
+		malloc_write("<jemalloc>: Error setting tsd.\n");
+		if (opt_abort) {
+			abort();
+		}
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_types.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_types.h
new file mode 100644
index 0000000..6200af6
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_types.h
@@ -0,0 +1,10 @@
+#ifndef JEMALLOC_INTERNAL_TSD_TYPES_H
+#define JEMALLOC_INTERNAL_TSD_TYPES_H
+
+#define MALLOC_TSD_CLEANUPS_MAX	2
+
+typedef struct tsd_s tsd_t;
+typedef struct tsdn_s tsdn_t;
+typedef bool (*malloc_tsd_cleanup_t)(void);
+
+#endif /* JEMALLOC_INTERNAL_TSD_TYPES_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_win.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_win.h
new file mode 100644
index 0000000..cf30d18
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/tsd_win.h
@@ -0,0 +1,139 @@
+#ifdef JEMALLOC_INTERNAL_TSD_WIN_H
+#error This file should be included only once, by tsd.h.
+#endif
+#define JEMALLOC_INTERNAL_TSD_WIN_H
+
+typedef struct {
+	bool initialized;
+	tsd_t val;
+} tsd_wrapper_t;
+
+extern DWORD tsd_tsd;
+extern tsd_wrapper_t tsd_boot_wrapper;
+extern bool tsd_booted;
+
+/* Initialization/cleanup. */
+JEMALLOC_ALWAYS_INLINE bool
+tsd_cleanup_wrapper(void) {
+	DWORD error = GetLastError();
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *)TlsGetValue(tsd_tsd);
+	SetLastError(error);
+
+	if (wrapper == NULL) {
+		return false;
+	}
+
+	if (wrapper->initialized) {
+		wrapper->initialized = false;
+		tsd_cleanup(&wrapper->val);
+		if (wrapper->initialized) {
+			/* Trigger another cleanup round. */
+			return true;
+		}
+	}
+	malloc_tsd_dalloc(wrapper);
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_wrapper_set(tsd_wrapper_t *wrapper) {
+	if (!TlsSetValue(tsd_tsd, (void *)wrapper)) {
+		malloc_write("<jemalloc>: Error setting TSD\n");
+		abort();
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE tsd_wrapper_t *
+tsd_wrapper_get(bool init) {
+	DWORD error = GetLastError();
+	tsd_wrapper_t *wrapper = (tsd_wrapper_t *) TlsGetValue(tsd_tsd);
+	SetLastError(error);
+
+	if (init && unlikely(wrapper == NULL)) {
+		wrapper = (tsd_wrapper_t *)
+		    malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+		if (wrapper == NULL) {
+			malloc_write("<jemalloc>: Error allocating TSD\n");
+			abort();
+		} else {
+			wrapper->initialized = false;
+			/* MSVC is finicky about aggregate initialization. */
+			tsd_t tsd_initializer = TSD_INITIALIZER;
+			wrapper->val = tsd_initializer;
+		}
+		tsd_wrapper_set(wrapper);
+	}
+	return wrapper;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot0(void) {
+	tsd_tsd = TlsAlloc();
+	if (tsd_tsd == TLS_OUT_OF_INDEXES) {
+		return true;
+	}
+	malloc_tsd_cleanup_register(&tsd_cleanup_wrapper);
+	tsd_wrapper_set(&tsd_boot_wrapper);
+	tsd_booted = true;
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_boot1(void) {
+	tsd_wrapper_t *wrapper;
+	wrapper = (tsd_wrapper_t *)
+	    malloc_tsd_malloc(sizeof(tsd_wrapper_t));
+	if (wrapper == NULL) {
+		malloc_write("<jemalloc>: Error allocating TSD\n");
+		abort();
+	}
+	tsd_boot_wrapper.initialized = false;
+	tsd_cleanup(&tsd_boot_wrapper.val);
+	wrapper->initialized = false;
+	tsd_t initializer = TSD_INITIALIZER;
+	wrapper->val = initializer;
+	tsd_wrapper_set(wrapper);
+}
+JEMALLOC_ALWAYS_INLINE bool
+tsd_boot(void) {
+	if (tsd_boot0()) {
+		return true;
+	}
+	tsd_boot1();
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_booted_get(void) {
+	return tsd_booted;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tsd_get_allocates(void) {
+	return true;
+}
+
+/* Get/set. */
+JEMALLOC_ALWAYS_INLINE tsd_t *
+tsd_get(bool init) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(init);
+	if (tsd_get_allocates() && !init && wrapper == NULL) {
+		return NULL;
+	}
+	return &wrapper->val;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+tsd_set(tsd_t *val) {
+	tsd_wrapper_t *wrapper;
+
+	assert(tsd_booted);
+	wrapper = tsd_wrapper_get(true);
+	if (likely(&wrapper->val != val)) {
+		wrapper->val = *(val);
+	}
+	wrapper->initialized = true;
+}
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/util.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/util.h
new file mode 100644
index 0000000..304cb54
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/util.h
@@ -0,0 +1,67 @@
+#ifndef JEMALLOC_INTERNAL_UTIL_H
+#define JEMALLOC_INTERNAL_UTIL_H
+
+#define UTIL_INLINE static inline
+
+/* Junk fill patterns. */
+#ifndef JEMALLOC_ALLOC_JUNK
+#  define JEMALLOC_ALLOC_JUNK	((uint8_t)0xa5)
+#endif
+#ifndef JEMALLOC_FREE_JUNK
+#  define JEMALLOC_FREE_JUNK	((uint8_t)0x5a)
+#endif
+
+/*
+ * Wrap a cpp argument that contains commas such that it isn't broken up into
+ * multiple arguments.
+ */
+#define JEMALLOC_ARG_CONCAT(...) __VA_ARGS__
+
+/* cpp macro definition stringification. */
+#define STRINGIFY_HELPER(x) #x
+#define STRINGIFY(x) STRINGIFY_HELPER(x)
+
+/*
+ * Silence compiler warnings due to uninitialized values.  This is used
+ * wherever the compiler fails to recognize that the variable is never used
+ * uninitialized.
+ */
+#define JEMALLOC_CC_SILENCE_INIT(v) = v
+
+#ifdef __GNUC__
+#  define likely(x)   __builtin_expect(!!(x), 1)
+#  define unlikely(x) __builtin_expect(!!(x), 0)
+#else
+#  define likely(x)   !!(x)
+#  define unlikely(x) !!(x)
+#endif
+
+#if !defined(JEMALLOC_INTERNAL_UNREACHABLE)
+#  error JEMALLOC_INTERNAL_UNREACHABLE should have been defined by configure
+#endif
+
+#define unreachable() JEMALLOC_INTERNAL_UNREACHABLE()
+
+/* Set error code. */
+UTIL_INLINE void
+set_errno(int errnum) {
+#ifdef _WIN32
+	SetLastError(errnum);
+#else
+	errno = errnum;
+#endif
+}
+
+/* Get last error code. */
+UTIL_INLINE int
+get_errno(void) {
+#ifdef _WIN32
+	return GetLastError();
+#else
+	return errno;
+#endif
+}
+
+#undef UTIL_INLINE
+
+#endif /* JEMALLOC_INTERNAL_UTIL_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/internal/witness.h b/memkind-1.10.0/jemalloc/include/jemalloc/internal/witness.h
new file mode 100644
index 0000000..fff9e98
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/internal/witness.h
@@ -0,0 +1,347 @@
+#ifndef JEMALLOC_INTERNAL_WITNESS_H
+#define JEMALLOC_INTERNAL_WITNESS_H
+
+#include "jemalloc/internal/ql.h"
+
+/******************************************************************************/
+/* LOCK RANKS */
+/******************************************************************************/
+
+/*
+ * Witnesses with rank WITNESS_RANK_OMIT are completely ignored by the witness
+ * machinery.
+ */
+
+#define WITNESS_RANK_OMIT		0U
+
+#define WITNESS_RANK_MIN		1U
+
+#define WITNESS_RANK_INIT		1U
+#define WITNESS_RANK_CTL		1U
+#define WITNESS_RANK_TCACHES		2U
+#define WITNESS_RANK_ARENAS		3U
+
+#define WITNESS_RANK_BACKGROUND_THREAD_GLOBAL	4U
+
+#define WITNESS_RANK_PROF_DUMP		5U
+#define WITNESS_RANK_PROF_BT2GCTX	6U
+#define WITNESS_RANK_PROF_TDATAS	7U
+#define WITNESS_RANK_PROF_TDATA		8U
+#define WITNESS_RANK_PROF_LOG		9U
+#define WITNESS_RANK_PROF_GCTX		10U
+#define WITNESS_RANK_BACKGROUND_THREAD	11U
+
+/*
+ * Used as an argument to witness_assert_depth_to_rank() in order to validate
+ * depth excluding non-core locks with lower ranks.  Since the rank argument to
+ * witness_assert_depth_to_rank() is inclusive rather than exclusive, this
+ * definition can have the same value as the minimally ranked core lock.
+ */
+#define WITNESS_RANK_CORE		12U
+
+#define WITNESS_RANK_DECAY		12U
+#define WITNESS_RANK_TCACHE_QL		13U
+#define WITNESS_RANK_EXTENT_GROW	14U
+#define WITNESS_RANK_EXTENTS		15U
+#define WITNESS_RANK_EXTENT_AVAIL	16U
+
+#define WITNESS_RANK_EXTENT_POOL	17U
+#define WITNESS_RANK_RTREE		18U
+#define WITNESS_RANK_BASE		19U
+#define WITNESS_RANK_ARENA_LARGE	20U
+#define WITNESS_RANK_HOOK		21U
+
+#define WITNESS_RANK_LEAF		0xffffffffU
+#define WITNESS_RANK_BIN		WITNESS_RANK_LEAF
+#define WITNESS_RANK_ARENA_STATS	WITNESS_RANK_LEAF
+#define WITNESS_RANK_DSS		WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_ACTIVE	WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_ACCUM		WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_DUMP_SEQ	WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_GDUMP		WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_NEXT_THR_UID	WITNESS_RANK_LEAF
+#define WITNESS_RANK_PROF_THREAD_ACTIVE_INIT	WITNESS_RANK_LEAF
+
+/******************************************************************************/
+/* PER-WITNESS DATA */
+/******************************************************************************/
+#if defined(JEMALLOC_DEBUG)
+#  define WITNESS_INITIALIZER(name, rank) {name, rank, NULL, NULL, {NULL, NULL}}
+#else
+#  define WITNESS_INITIALIZER(name, rank)
+#endif
+
+typedef struct witness_s witness_t;
+typedef unsigned witness_rank_t;
+typedef ql_head(witness_t) witness_list_t;
+typedef int witness_comp_t (const witness_t *, void *, const witness_t *,
+    void *);
+
+struct witness_s {
+	/* Name, used for printing lock order reversal messages. */
+	const char		*name;
+
+	/*
+	 * Witness rank, where 0 is lowest and UINT_MAX is highest.  Witnesses
+	 * must be acquired in order of increasing rank.
+	 */
+	witness_rank_t		rank;
+
+	/*
+	 * If two witnesses are of equal rank and they have the samp comp
+	 * function pointer, it is called as a last attempt to differentiate
+	 * between witnesses of equal rank.
+	 */
+	witness_comp_t		*comp;
+
+	/* Opaque data, passed to comp(). */
+	void			*opaque;
+
+	/* Linkage for thread's currently owned locks. */
+	ql_elm(witness_t)	link;
+};
+
+/******************************************************************************/
+/* PER-THREAD DATA */
+/******************************************************************************/
+typedef struct witness_tsd_s witness_tsd_t;
+struct witness_tsd_s {
+	witness_list_t witnesses;
+	bool forking;
+};
+
+#define WITNESS_TSD_INITIALIZER { ql_head_initializer(witnesses), false }
+#define WITNESS_TSDN_NULL ((witness_tsdn_t *)0)
+
+/******************************************************************************/
+/* (PER-THREAD) NULLABILITY HELPERS */
+/******************************************************************************/
+typedef struct witness_tsdn_s witness_tsdn_t;
+struct witness_tsdn_s {
+	witness_tsd_t witness_tsd;
+};
+
+JEMALLOC_ALWAYS_INLINE witness_tsdn_t *
+witness_tsd_tsdn(witness_tsd_t *witness_tsd) {
+	return (witness_tsdn_t *)witness_tsd;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+witness_tsdn_null(witness_tsdn_t *witness_tsdn) {
+	return witness_tsdn == NULL;
+}
+
+JEMALLOC_ALWAYS_INLINE witness_tsd_t *
+witness_tsdn_tsd(witness_tsdn_t *witness_tsdn) {
+	assert(!witness_tsdn_null(witness_tsdn));
+	return &witness_tsdn->witness_tsd;
+}
+
+/******************************************************************************/
+/* API */
+/******************************************************************************/
+void witness_init(witness_t *witness, const char *name, witness_rank_t rank,
+    witness_comp_t *comp, void *opaque);
+
+typedef void (witness_lock_error_t)(const witness_list_t *, const witness_t *);
+extern witness_lock_error_t *JET_MUTABLE witness_lock_error;
+
+typedef void (witness_owner_error_t)(const witness_t *);
+extern witness_owner_error_t *JET_MUTABLE witness_owner_error;
+
+typedef void (witness_not_owner_error_t)(const witness_t *);
+extern witness_not_owner_error_t *JET_MUTABLE witness_not_owner_error;
+
+typedef void (witness_depth_error_t)(const witness_list_t *,
+    witness_rank_t rank_inclusive, unsigned depth);
+extern witness_depth_error_t *JET_MUTABLE witness_depth_error;
+
+void witnesses_cleanup(witness_tsd_t *witness_tsd);
+void witness_prefork(witness_tsd_t *witness_tsd);
+void witness_postfork_parent(witness_tsd_t *witness_tsd);
+void witness_postfork_child(witness_tsd_t *witness_tsd);
+
+/* Helper, not intended for direct use. */
+static inline bool
+witness_owner(witness_tsd_t *witness_tsd, const witness_t *witness) {
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	cassert(config_debug);
+
+	witnesses = &witness_tsd->witnesses;
+	ql_foreach(w, witnesses, link) {
+		if (w == witness) {
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static inline void
+witness_assert_owner(witness_tsdn_t *witness_tsdn, const witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	if (witness_owner(witness_tsd, witness)) {
+		return;
+	}
+	witness_owner_error(witness);
+}
+
+static inline void
+witness_assert_not_owner(witness_tsdn_t *witness_tsdn,
+    const witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	witnesses = &witness_tsd->witnesses;
+	ql_foreach(w, witnesses, link) {
+		if (w == witness) {
+			witness_not_owner_error(witness);
+		}
+	}
+}
+
+static inline void
+witness_assert_depth_to_rank(witness_tsdn_t *witness_tsdn,
+    witness_rank_t rank_inclusive, unsigned depth) {
+	witness_tsd_t *witness_tsd;
+	unsigned d;
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+
+	d = 0;
+	witnesses = &witness_tsd->witnesses;
+	w = ql_last(witnesses, link);
+	if (w != NULL) {
+		ql_reverse_foreach(w, witnesses, link) {
+			if (w->rank < rank_inclusive) {
+				break;
+			}
+			d++;
+		}
+	}
+	if (d != depth) {
+		witness_depth_error(witnesses, rank_inclusive, depth);
+	}
+}
+
+static inline void
+witness_assert_depth(witness_tsdn_t *witness_tsdn, unsigned depth) {
+	witness_assert_depth_to_rank(witness_tsdn, WITNESS_RANK_MIN, depth);
+}
+
+static inline void
+witness_assert_lockless(witness_tsdn_t *witness_tsdn) {
+	witness_assert_depth(witness_tsdn, 0);
+}
+
+static inline void
+witness_lock(witness_tsdn_t *witness_tsdn, witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+	witness_list_t *witnesses;
+	witness_t *w;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	witness_assert_not_owner(witness_tsdn, witness);
+
+	witnesses = &witness_tsd->witnesses;
+	w = ql_last(witnesses, link);
+	if (w == NULL) {
+		/* No other locks; do nothing. */
+	} else if (witness_tsd->forking && w->rank <= witness->rank) {
+		/* Forking, and relaxed ranking satisfied. */
+	} else if (w->rank > witness->rank) {
+		/* Not forking, rank order reversal. */
+		witness_lock_error(witnesses, witness);
+	} else if (w->rank == witness->rank && (w->comp == NULL || w->comp !=
+	    witness->comp || w->comp(w, w->opaque, witness, witness->opaque) >
+	    0)) {
+		/*
+		 * Missing/incompatible comparison function, or comparison
+		 * function indicates rank order reversal.
+		 */
+		witness_lock_error(witnesses, witness);
+	}
+
+	ql_elm_new(witness, link);
+	ql_tail_insert(witnesses, witness, link);
+}
+
+static inline void
+witness_unlock(witness_tsdn_t *witness_tsdn, witness_t *witness) {
+	witness_tsd_t *witness_tsd;
+	witness_list_t *witnesses;
+
+	if (!config_debug) {
+		return;
+	}
+
+	if (witness_tsdn_null(witness_tsdn)) {
+		return;
+	}
+	witness_tsd = witness_tsdn_tsd(witness_tsdn);
+	if (witness->rank == WITNESS_RANK_OMIT) {
+		return;
+	}
+
+	/*
+	 * Check whether owner before removal, rather than relying on
+	 * witness_assert_owner() to abort, so that unit tests can test this
+	 * function's failure mode without causing undefined behavior.
+	 */
+	if (witness_owner(witness_tsd, witness)) {
+		witnesses = &witness_tsd->witnesses;
+		ql_remove(witnesses, witness, link);
+	} else {
+		witness_assert_owner(witness_tsdn, witness);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_WITNESS_H */
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc.sh b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc.sh
new file mode 100755
index 0000000..b19b154
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc.sh
@@ -0,0 +1,27 @@
+#!/bin/sh
+
+objroot=$1
+
+cat <<EOF
+#ifndef JEMALLOC_H_
+#define JEMALLOC_H_
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+EOF
+
+for hdr in jemalloc_defs.h jemalloc_rename.h jemalloc_macros.h \
+           jemalloc_protos.h jemalloc_typedefs.h jemalloc_mangle.h ; do
+  cat "${objroot}include/jemalloc/${hdr}" \
+      | grep -v 'Generated from .* by configure\.' \
+      | sed -e 's/ $//g'
+  echo
+done
+
+cat <<EOF
+#ifdef __cplusplus
+}
+#endif
+#endif /* JEMALLOC_H_ */
+EOF
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_defs.h.in b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_defs.h.in
new file mode 100644
index 0000000..11c3918
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_defs.h.in
@@ -0,0 +1,48 @@
+/* Defined if __attribute__((...)) syntax is supported. */
+#undef JEMALLOC_HAVE_ATTR
+
+/* Defined if alloc_size attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_ALLOC_SIZE
+
+/* Defined if format_arg(...) attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_FORMAT_ARG
+
+/* Defined if format(gnu_printf, ...) attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF
+
+/* Defined if format(printf, ...) attribute is supported. */
+#undef JEMALLOC_HAVE_ATTR_FORMAT_PRINTF
+
+/*
+ * Define overrides for non-standard allocator-related functions if they are
+ * present on the system.
+ */
+#undef JEMALLOC_OVERRIDE_MEMALIGN
+#undef JEMALLOC_OVERRIDE_VALLOC
+
+/*
+ * At least Linux omits the "const" in:
+ *
+ *   size_t malloc_usable_size(const void *ptr);
+ *
+ * Match the operating system's prototype.
+ */
+#undef JEMALLOC_USABLE_SIZE_CONST
+
+/*
+ * If defined, specify throw() for the public function prototypes when compiling
+ * with C++.  The only justification for this is to match the prototypes that
+ * glibc defines.
+ */
+#undef JEMALLOC_USE_CXX_THROW
+
+#ifdef _MSC_VER
+#  ifdef _WIN64
+#    define LG_SIZEOF_PTR_WIN 3
+#  else
+#    define LG_SIZEOF_PTR_WIN 2
+#  endif
+#endif
+
+/* sizeof(void *) == 2^LG_SIZEOF_PTR. */
+#undef LG_SIZEOF_PTR
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_macros.h.in b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_macros.h.in
new file mode 100644
index 0000000..59e2955
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_macros.h.in
@@ -0,0 +1,129 @@
+#include <stdlib.h>
+#include <stdbool.h>
+#include <stdint.h>
+#include <limits.h>
+#include <strings.h>
+
+#define JEMALLOC_VERSION "@jemalloc_version@"
+#define JEMALLOC_VERSION_MAJOR @jemalloc_version_major@
+#define JEMALLOC_VERSION_MINOR @jemalloc_version_minor@
+#define JEMALLOC_VERSION_BUGFIX @jemalloc_version_bugfix@
+#define JEMALLOC_VERSION_NREV @jemalloc_version_nrev@
+#define JEMALLOC_VERSION_GID "@jemalloc_version_gid@"
+#define JEMALLOC_VERSION_GID_IDENT @jemalloc_version_gid@
+
+#define MALLOCX_LG_ALIGN(la)	((int)(la))
+#if LG_SIZEOF_PTR == 2
+#  define MALLOCX_ALIGN(a)	((int)(ffs((int)(a))-1))
+#else
+#  define MALLOCX_ALIGN(a)						\
+     ((int)(((size_t)(a) < (size_t)INT_MAX) ? ffs((int)(a))-1 :	\
+     ffs((int)(((size_t)(a))>>32))+31))
+#endif
+#define MALLOCX_ZERO	((int)0x40)
+/*
+ * Bias tcache index bits so that 0 encodes "automatic tcache management", and 1
+ * encodes MALLOCX_TCACHE_NONE.
+ */
+#define MALLOCX_TCACHE(tc)	((int)(((tc)+2) << 8))
+#define MALLOCX_TCACHE_NONE	MALLOCX_TCACHE(-1)
+/*
+ * Bias arena index bits so that 0 encodes "use an automatically chosen arena".
+ */
+#define MALLOCX_ARENA(a)	((((int)(a))+1) << 20)
+
+/*
+ * Use as arena index in "arena.<i>.{purge,decay,dss}" and
+ * "stats.arenas.<i>.*" mallctl interfaces to select all arenas.  This
+ * definition is intentionally specified in raw decimal format to support
+ * cpp-based string concatenation, e.g.
+ *
+ *   #define STRINGIFY_HELPER(x) #x
+ *   #define STRINGIFY(x) STRINGIFY_HELPER(x)
+ *
+ *   mallctl("arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".purge", NULL, NULL, NULL,
+ *       0);
+ */
+#define MALLCTL_ARENAS_ALL	4096
+/*
+ * Use as arena index in "stats.arenas.<i>.*" mallctl interfaces to select
+ * destroyed arenas.
+ */
+#define MALLCTL_ARENAS_DESTROYED	4097
+
+#if defined(__cplusplus) && defined(JEMALLOC_USE_CXX_THROW)
+#  define JEMALLOC_CXX_THROW throw()
+#else
+#  define JEMALLOC_CXX_THROW
+#endif
+
+#if defined(_MSC_VER)
+#  define JEMALLOC_ATTR(s)
+#  define JEMALLOC_ALIGNED(s) __declspec(align(s))
+#  define JEMALLOC_ALLOC_SIZE(s)
+#  define JEMALLOC_ALLOC_SIZE2(s1, s2)
+#  ifndef JEMALLOC_EXPORT
+#    ifdef DLLEXPORT
+#      define JEMALLOC_EXPORT __declspec(dllexport)
+#    else
+#      define JEMALLOC_EXPORT __declspec(dllimport)
+#    endif
+#  endif
+#  define JEMALLOC_FORMAT_ARG(i)
+#  define JEMALLOC_FORMAT_PRINTF(s, i)
+#  define JEMALLOC_NOINLINE __declspec(noinline)
+#  ifdef __cplusplus
+#    define JEMALLOC_NOTHROW __declspec(nothrow)
+#  else
+#    define JEMALLOC_NOTHROW
+#  endif
+#  define JEMALLOC_SECTION(s) __declspec(allocate(s))
+#  define JEMALLOC_RESTRICT_RETURN __declspec(restrict)
+#  if _MSC_VER >= 1900 && !defined(__EDG__)
+#    define JEMALLOC_ALLOCATOR __declspec(allocator)
+#  else
+#    define JEMALLOC_ALLOCATOR
+#  endif
+#elif defined(JEMALLOC_HAVE_ATTR)
+#  define JEMALLOC_ATTR(s) __attribute__((s))
+#  define JEMALLOC_ALIGNED(s) JEMALLOC_ATTR(aligned(s))
+#  ifdef JEMALLOC_HAVE_ATTR_ALLOC_SIZE
+#    define JEMALLOC_ALLOC_SIZE(s) JEMALLOC_ATTR(alloc_size(s))
+#    define JEMALLOC_ALLOC_SIZE2(s1, s2) JEMALLOC_ATTR(alloc_size(s1, s2))
+#  else
+#    define JEMALLOC_ALLOC_SIZE(s)
+#    define JEMALLOC_ALLOC_SIZE2(s1, s2)
+#  endif
+#  ifndef JEMALLOC_EXPORT
+#    define JEMALLOC_EXPORT JEMALLOC_ATTR(visibility("default"))
+#  endif
+#  ifdef JEMALLOC_HAVE_ATTR_FORMAT_ARG
+#    define JEMALLOC_FORMAT_ARG(i) JEMALLOC_ATTR(__format_arg__(3))
+#  else
+#    define JEMALLOC_FORMAT_ARG(i)
+#  endif
+#  ifdef JEMALLOC_HAVE_ATTR_FORMAT_GNU_PRINTF
+#    define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(gnu_printf, s, i))
+#  elif defined(JEMALLOC_HAVE_ATTR_FORMAT_PRINTF)
+#    define JEMALLOC_FORMAT_PRINTF(s, i) JEMALLOC_ATTR(format(printf, s, i))
+#  else
+#    define JEMALLOC_FORMAT_PRINTF(s, i)
+#  endif
+#  define JEMALLOC_NOINLINE JEMALLOC_ATTR(noinline)
+#  define JEMALLOC_NOTHROW JEMALLOC_ATTR(nothrow)
+#  define JEMALLOC_SECTION(s) JEMALLOC_ATTR(section(s))
+#  define JEMALLOC_RESTRICT_RETURN
+#  define JEMALLOC_ALLOCATOR
+#else
+#  define JEMALLOC_ATTR(s)
+#  define JEMALLOC_ALIGNED(s)
+#  define JEMALLOC_ALLOC_SIZE(s)
+#  define JEMALLOC_ALLOC_SIZE2(s1, s2)
+#  define JEMALLOC_EXPORT
+#  define JEMALLOC_FORMAT_PRINTF(s, i)
+#  define JEMALLOC_NOINLINE
+#  define JEMALLOC_NOTHROW
+#  define JEMALLOC_SECTION(s)
+#  define JEMALLOC_RESTRICT_RETURN
+#  define JEMALLOC_ALLOCATOR
+#endif
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_mangle.sh b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_mangle.sh
new file mode 100755
index 0000000..c675bb4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_mangle.sh
@@ -0,0 +1,45 @@
+#!/bin/sh -eu
+
+public_symbols_txt=$1
+symbol_prefix=$2
+
+cat <<EOF
+/*
+ * By default application code must explicitly refer to mangled symbol names,
+ * so that it is possible to use jemalloc in conjunction with another allocator
+ * in the same application.  Define JEMALLOC_MANGLE in order to cause automatic
+ * name mangling that matches the API prefixing that happened as a result of
+ * --with-mangling and/or --with-jemalloc-prefix configuration settings.
+ */
+#ifdef JEMALLOC_MANGLE
+#  ifndef JEMALLOC_NO_DEMANGLE
+#    define JEMALLOC_NO_DEMANGLE
+#  endif
+EOF
+
+for nm in `cat ${public_symbols_txt}` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#  define ${n} ${symbol_prefix}${n}"
+done
+
+cat <<EOF
+#endif
+
+/*
+ * The ${symbol_prefix}* macros can be used as stable alternative names for the
+ * public jemalloc API if JEMALLOC_NO_DEMANGLE is defined.  This is primarily
+ * meant for use in jemalloc itself, but it can be used by application code to
+ * provide isolation from the name mangling specified via --with-mangling
+ * and/or --with-jemalloc-prefix.
+ */
+#ifndef JEMALLOC_NO_DEMANGLE
+EOF
+
+for nm in `cat ${public_symbols_txt}` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  echo "#  undef ${symbol_prefix}${n}"
+done
+
+cat <<EOF
+#endif
+EOF
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_protos.h.in b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_protos.h.in
new file mode 100644
index 0000000..a78414b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_protos.h.in
@@ -0,0 +1,66 @@
+/*
+ * The @je_@ prefix on the following public symbol declarations is an artifact
+ * of namespace management, and should be omitted in application code unless
+ * JEMALLOC_NO_DEMANGLE is defined (see jemalloc_mangle@install_suffix@.h).
+ */
+extern JEMALLOC_EXPORT const char	*@je_@malloc_conf;
+extern JEMALLOC_EXPORT void		(*@je_@malloc_message)(void *cbopaque,
+    const char *s);
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@malloc(size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1);
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@calloc(size_t num, size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE2(1, 2);
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@posix_memalign(void **memptr,
+    size_t alignment, size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(nonnull(1));
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@aligned_alloc(size_t alignment,
+    size_t size) JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc)
+    JEMALLOC_ALLOC_SIZE(2);
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@realloc(void *ptr, size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ALLOC_SIZE(2);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@free(void *ptr)
+    JEMALLOC_CXX_THROW;
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@mallocx(size_t size, int flags)
+    JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1);
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@rallocx(void *ptr, size_t size,
+    int flags) JEMALLOC_ALLOC_SIZE(2);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@xallocx(void *ptr, size_t size,
+    size_t extra, int flags);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@sallocx(const void *ptr,
+    int flags) JEMALLOC_ATTR(pure);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@dallocx(void *ptr, int flags);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@sdallocx(void *ptr, size_t size,
+    int flags);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@nallocx(size_t size, int flags)
+    JEMALLOC_ATTR(pure);
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@mallctl(const char *name,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen);
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@mallctlnametomib(const char *name,
+    size_t *mibp, size_t *miblenp);
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW	@je_@mallctlbymib(const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen);
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW	@je_@malloc_stats_print(
+    void (*write_cb)(void *, const char *), void *@je_@cbopaque,
+    const char *opts);
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW	@je_@malloc_usable_size(
+    JEMALLOC_USABLE_SIZE_CONST void *ptr) JEMALLOC_CXX_THROW;
+
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@memalign(size_t alignment, size_t size)
+    JEMALLOC_CXX_THROW JEMALLOC_ATTR(malloc);
+#endif
+
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+    void JEMALLOC_NOTHROW	*@je_@valloc(size_t size) JEMALLOC_CXX_THROW
+    JEMALLOC_ATTR(malloc);
+#endif
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_rename.sh b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_rename.sh
new file mode 100755
index 0000000..f943891
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_rename.sh
@@ -0,0 +1,22 @@
+#!/bin/sh
+
+public_symbols_txt=$1
+
+cat <<EOF
+/*
+ * Name mangling for public symbols is controlled by --with-mangling and
+ * --with-jemalloc-prefix.  With default settings the je_ prefix is stripped by
+ * these macro definitions.
+ */
+#ifndef JEMALLOC_NO_RENAME
+EOF
+
+for nm in `cat ${public_symbols_txt}` ; do
+  n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
+  m=`echo ${nm} |tr ':' ' ' |awk '{print $2}'`
+  echo "#  define je_${n} ${m}"
+done
+
+cat <<EOF
+#endif
+EOF
diff --git a/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_typedefs.h.in b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_typedefs.h.in
new file mode 100644
index 0000000..1a58874
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/jemalloc/jemalloc_typedefs.h.in
@@ -0,0 +1,77 @@
+typedef struct extent_hooks_s extent_hooks_t;
+
+/*
+ * void *
+ * extent_alloc(extent_hooks_t *extent_hooks, void *new_addr, size_t size,
+ *     size_t alignment, bool *zero, bool *commit, unsigned arena_ind);
+ */
+typedef void *(extent_alloc_t)(extent_hooks_t *, void *, size_t, size_t, bool *,
+    bool *, unsigned);
+
+/*
+ * bool
+ * extent_dalloc(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     bool committed, unsigned arena_ind);
+ */
+typedef bool (extent_dalloc_t)(extent_hooks_t *, void *, size_t, bool,
+    unsigned);
+
+/*
+ * void
+ * extent_destroy(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     bool committed, unsigned arena_ind);
+ */
+typedef void (extent_destroy_t)(extent_hooks_t *, void *, size_t, bool,
+    unsigned);
+
+/*
+ * bool
+ * extent_commit(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t offset, size_t length, unsigned arena_ind);
+ */
+typedef bool (extent_commit_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
+    unsigned);
+
+/*
+ * bool
+ * extent_decommit(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t offset, size_t length, unsigned arena_ind);
+ */
+typedef bool (extent_decommit_t)(extent_hooks_t *, void *, size_t, size_t,
+    size_t, unsigned);
+
+/*
+ * bool
+ * extent_purge(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t offset, size_t length, unsigned arena_ind);
+ */
+typedef bool (extent_purge_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
+    unsigned);
+
+/*
+ * bool
+ * extent_split(extent_hooks_t *extent_hooks, void *addr, size_t size,
+ *     size_t size_a, size_t size_b, bool committed, unsigned arena_ind);
+ */
+typedef bool (extent_split_t)(extent_hooks_t *, void *, size_t, size_t, size_t,
+    bool, unsigned);
+
+/*
+ * bool
+ * extent_merge(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a,
+ *     void *addr_b, size_t size_b, bool committed, unsigned arena_ind);
+ */
+typedef bool (extent_merge_t)(extent_hooks_t *, void *, size_t, void *, size_t,
+    bool, unsigned);
+
+struct extent_hooks_s {
+	extent_alloc_t		*alloc;
+	extent_dalloc_t		*dalloc;
+	extent_destroy_t	*destroy;
+	extent_commit_t		*commit;
+	extent_decommit_t	*decommit;
+	extent_purge_t		*purge_lazy;
+	extent_purge_t		*purge_forced;
+	extent_split_t		*split;
+	extent_merge_t		*merge;
+};
diff --git a/memkind-1.10.0/jemalloc/include/msvc_compat/C99/stdbool.h b/memkind-1.10.0/jemalloc/include/msvc_compat/C99/stdbool.h
new file mode 100644
index 0000000..d92160e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/msvc_compat/C99/stdbool.h
@@ -0,0 +1,20 @@
+#ifndef stdbool_h
+#define stdbool_h
+
+#include <wtypes.h>
+
+/* MSVC doesn't define _Bool or bool in C, but does have BOOL */
+/* Note this doesn't pass autoconf's test because (bool) 0.5 != true */
+/* Clang-cl uses MSVC headers, so needs msvc_compat, but has _Bool as
+ * a built-in type. */
+#ifndef __clang__
+typedef BOOL _Bool;
+#endif
+
+#define bool _Bool
+#define true 1
+#define false 0
+
+#define __bool_true_false_are_defined 1
+
+#endif /* stdbool_h */
diff --git a/memkind-1.10.0/jemalloc/include/msvc_compat/C99/stdint.h b/memkind-1.10.0/jemalloc/include/msvc_compat/C99/stdint.h
new file mode 100644
index 0000000..d02608a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/msvc_compat/C99/stdint.h
@@ -0,0 +1,247 @@
+// ISO C9x  compliant stdint.h for Microsoft Visual Studio
+// Based on ISO/IEC 9899:TC2 Committee draft (May 6, 2005) WG14/N1124 
+// 
+//  Copyright (c) 2006-2008 Alexander Chemeris
+// 
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are met:
+// 
+//   1. Redistributions of source code must retain the above copyright notice,
+//      this list of conditions and the following disclaimer.
+// 
+//   2. Redistributions in binary form must reproduce the above copyright
+//      notice, this list of conditions and the following disclaimer in the
+//      documentation and/or other materials provided with the distribution.
+// 
+//   3. The name of the author may be used to endorse or promote products
+//      derived from this software without specific prior written permission.
+// 
+// THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED
+// WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+// MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO
+// EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
+// OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, 
+// WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
+// OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+// ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+// 
+///////////////////////////////////////////////////////////////////////////////
+
+#ifndef _MSC_VER // [
+#error "Use this header only with Microsoft Visual C++ compilers!"
+#endif // _MSC_VER ]
+
+#ifndef _MSC_STDINT_H_ // [
+#define _MSC_STDINT_H_
+
+#if _MSC_VER > 1000
+#pragma once
+#endif
+
+#include <limits.h>
+
+// For Visual Studio 6 in C++ mode and for many Visual Studio versions when
+// compiling for ARM we should wrap <wchar.h> include with 'extern "C++" {}'
+// or compiler give many errors like this:
+//   error C2733: second C linkage of overloaded function 'wmemchr' not allowed
+#ifdef __cplusplus
+extern "C" {
+#endif
+#  include <wchar.h>
+#ifdef __cplusplus
+}
+#endif
+
+// Define _W64 macros to mark types changing their size, like intptr_t.
+#ifndef _W64
+#  if !defined(__midl) && (defined(_X86_) || defined(_M_IX86)) && _MSC_VER >= 1300
+#     define _W64 __w64
+#  else
+#     define _W64
+#  endif
+#endif
+
+
+// 7.18.1 Integer types
+
+// 7.18.1.1 Exact-width integer types
+
+// Visual Studio 6 and Embedded Visual C++ 4 doesn't
+// realize that, e.g. char has the same size as __int8
+// so we give up on __intX for them.
+#if (_MSC_VER < 1300)
+   typedef signed char       int8_t;
+   typedef signed short      int16_t;
+   typedef signed int        int32_t;
+   typedef unsigned char     uint8_t;
+   typedef unsigned short    uint16_t;
+   typedef unsigned int      uint32_t;
+#else
+   typedef signed __int8     int8_t;
+   typedef signed __int16    int16_t;
+   typedef signed __int32    int32_t;
+   typedef unsigned __int8   uint8_t;
+   typedef unsigned __int16  uint16_t;
+   typedef unsigned __int32  uint32_t;
+#endif
+typedef signed __int64       int64_t;
+typedef unsigned __int64     uint64_t;
+
+
+// 7.18.1.2 Minimum-width integer types
+typedef int8_t    int_least8_t;
+typedef int16_t   int_least16_t;
+typedef int32_t   int_least32_t;
+typedef int64_t   int_least64_t;
+typedef uint8_t   uint_least8_t;
+typedef uint16_t  uint_least16_t;
+typedef uint32_t  uint_least32_t;
+typedef uint64_t  uint_least64_t;
+
+// 7.18.1.3 Fastest minimum-width integer types
+typedef int8_t    int_fast8_t;
+typedef int16_t   int_fast16_t;
+typedef int32_t   int_fast32_t;
+typedef int64_t   int_fast64_t;
+typedef uint8_t   uint_fast8_t;
+typedef uint16_t  uint_fast16_t;
+typedef uint32_t  uint_fast32_t;
+typedef uint64_t  uint_fast64_t;
+
+// 7.18.1.4 Integer types capable of holding object pointers
+#ifdef _WIN64 // [
+   typedef signed __int64    intptr_t;
+   typedef unsigned __int64  uintptr_t;
+#else // _WIN64 ][
+   typedef _W64 signed int   intptr_t;
+   typedef _W64 unsigned int uintptr_t;
+#endif // _WIN64 ]
+
+// 7.18.1.5 Greatest-width integer types
+typedef int64_t   intmax_t;
+typedef uint64_t  uintmax_t;
+
+
+// 7.18.2 Limits of specified-width integer types
+
+#if !defined(__cplusplus) || defined(__STDC_LIMIT_MACROS) // [   See footnote 220 at page 257 and footnote 221 at page 259
+
+// 7.18.2.1 Limits of exact-width integer types
+#define INT8_MIN     ((int8_t)_I8_MIN)
+#define INT8_MAX     _I8_MAX
+#define INT16_MIN    ((int16_t)_I16_MIN)
+#define INT16_MAX    _I16_MAX
+#define INT32_MIN    ((int32_t)_I32_MIN)
+#define INT32_MAX    _I32_MAX
+#define INT64_MIN    ((int64_t)_I64_MIN)
+#define INT64_MAX    _I64_MAX
+#define UINT8_MAX    _UI8_MAX
+#define UINT16_MAX   _UI16_MAX
+#define UINT32_MAX   _UI32_MAX
+#define UINT64_MAX   _UI64_MAX
+
+// 7.18.2.2 Limits of minimum-width integer types
+#define INT_LEAST8_MIN    INT8_MIN
+#define INT_LEAST8_MAX    INT8_MAX
+#define INT_LEAST16_MIN   INT16_MIN
+#define INT_LEAST16_MAX   INT16_MAX
+#define INT_LEAST32_MIN   INT32_MIN
+#define INT_LEAST32_MAX   INT32_MAX
+#define INT_LEAST64_MIN   INT64_MIN
+#define INT_LEAST64_MAX   INT64_MAX
+#define UINT_LEAST8_MAX   UINT8_MAX
+#define UINT_LEAST16_MAX  UINT16_MAX
+#define UINT_LEAST32_MAX  UINT32_MAX
+#define UINT_LEAST64_MAX  UINT64_MAX
+
+// 7.18.2.3 Limits of fastest minimum-width integer types
+#define INT_FAST8_MIN    INT8_MIN
+#define INT_FAST8_MAX    INT8_MAX
+#define INT_FAST16_MIN   INT16_MIN
+#define INT_FAST16_MAX   INT16_MAX
+#define INT_FAST32_MIN   INT32_MIN
+#define INT_FAST32_MAX   INT32_MAX
+#define INT_FAST64_MIN   INT64_MIN
+#define INT_FAST64_MAX   INT64_MAX
+#define UINT_FAST8_MAX   UINT8_MAX
+#define UINT_FAST16_MAX  UINT16_MAX
+#define UINT_FAST32_MAX  UINT32_MAX
+#define UINT_FAST64_MAX  UINT64_MAX
+
+// 7.18.2.4 Limits of integer types capable of holding object pointers
+#ifdef _WIN64 // [
+#  define INTPTR_MIN   INT64_MIN
+#  define INTPTR_MAX   INT64_MAX
+#  define UINTPTR_MAX  UINT64_MAX
+#else // _WIN64 ][
+#  define INTPTR_MIN   INT32_MIN
+#  define INTPTR_MAX   INT32_MAX
+#  define UINTPTR_MAX  UINT32_MAX
+#endif // _WIN64 ]
+
+// 7.18.2.5 Limits of greatest-width integer types
+#define INTMAX_MIN   INT64_MIN
+#define INTMAX_MAX   INT64_MAX
+#define UINTMAX_MAX  UINT64_MAX
+
+// 7.18.3 Limits of other integer types
+
+#ifdef _WIN64 // [
+#  define PTRDIFF_MIN  _I64_MIN
+#  define PTRDIFF_MAX  _I64_MAX
+#else  // _WIN64 ][
+#  define PTRDIFF_MIN  _I32_MIN
+#  define PTRDIFF_MAX  _I32_MAX
+#endif  // _WIN64 ]
+
+#define SIG_ATOMIC_MIN  INT_MIN
+#define SIG_ATOMIC_MAX  INT_MAX
+
+#ifndef SIZE_MAX // [
+#  ifdef _WIN64 // [
+#     define SIZE_MAX  _UI64_MAX
+#  else // _WIN64 ][
+#     define SIZE_MAX  _UI32_MAX
+#  endif // _WIN64 ]
+#endif // SIZE_MAX ]
+
+// WCHAR_MIN and WCHAR_MAX are also defined in <wchar.h>
+#ifndef WCHAR_MIN // [
+#  define WCHAR_MIN  0
+#endif  // WCHAR_MIN ]
+#ifndef WCHAR_MAX // [
+#  define WCHAR_MAX  _UI16_MAX
+#endif  // WCHAR_MAX ]
+
+#define WINT_MIN  0
+#define WINT_MAX  _UI16_MAX
+
+#endif // __STDC_LIMIT_MACROS ]
+
+
+// 7.18.4 Limits of other integer types
+
+#if !defined(__cplusplus) || defined(__STDC_CONSTANT_MACROS) // [   See footnote 224 at page 260
+
+// 7.18.4.1 Macros for minimum-width integer constants
+
+#define INT8_C(val)  val##i8
+#define INT16_C(val) val##i16
+#define INT32_C(val) val##i32
+#define INT64_C(val) val##i64
+
+#define UINT8_C(val)  val##ui8
+#define UINT16_C(val) val##ui16
+#define UINT32_C(val) val##ui32
+#define UINT64_C(val) val##ui64
+
+// 7.18.4.2 Macros for greatest-width integer constants
+#define INTMAX_C   INT64_C
+#define UINTMAX_C  UINT64_C
+
+#endif // __STDC_CONSTANT_MACROS ]
+
+
+#endif // _MSC_STDINT_H_ ]
diff --git a/memkind-1.10.0/jemalloc/include/msvc_compat/strings.h b/memkind-1.10.0/jemalloc/include/msvc_compat/strings.h
new file mode 100644
index 0000000..996f256
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/msvc_compat/strings.h
@@ -0,0 +1,58 @@
+#ifndef strings_h
+#define strings_h
+
+/* MSVC doesn't define ffs/ffsl. This dummy strings.h header is provided
+ * for both */
+#ifdef _MSC_VER
+#  include <intrin.h>
+#  pragma intrinsic(_BitScanForward)
+static __forceinline int ffsl(long x) {
+	unsigned long i;
+
+	if (_BitScanForward(&i, x)) {
+		return i + 1;
+	}
+	return 0;
+}
+
+static __forceinline int ffs(int x) {
+	return ffsl(x);
+}
+
+#  ifdef  _M_X64
+#    pragma intrinsic(_BitScanForward64)
+#  endif
+
+static __forceinline int ffsll(unsigned __int64 x) {
+	unsigned long i;
+#ifdef  _M_X64
+	if (_BitScanForward64(&i, x)) {
+		return i + 1;
+	}
+	return 0;
+#else
+// Fallback for 32-bit build where 64-bit version not available
+// assuming little endian
+	union {
+		unsigned __int64 ll;
+		unsigned   long l[2];
+	} s;
+
+	s.ll = x;
+
+	if (_BitScanForward(&i, s.l[0])) {
+		return i + 1;
+	} else if(_BitScanForward(&i, s.l[1])) {
+		return i + 33;
+	}
+	return 0;
+#endif
+}
+
+#else
+#  define ffsll(x) __builtin_ffsll(x)
+#  define ffsl(x) __builtin_ffsl(x)
+#  define ffs(x) __builtin_ffs(x)
+#endif
+
+#endif /* strings_h */
diff --git a/memkind-1.10.0/jemalloc/include/msvc_compat/windows_extra.h b/memkind-1.10.0/jemalloc/include/msvc_compat/windows_extra.h
new file mode 100644
index 0000000..a6ebb93
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/include/msvc_compat/windows_extra.h
@@ -0,0 +1,6 @@
+#ifndef MSVC_COMPAT_WINDOWS_EXTRA_H
+#define MSVC_COMPAT_WINDOWS_EXTRA_H
+
+#include <errno.h>
+
+#endif /* MSVC_COMPAT_WINDOWS_EXTRA_H */
diff --git a/memkind-1.10.0/jemalloc/jemalloc.pc.in b/memkind-1.10.0/jemalloc/jemalloc.pc.in
new file mode 100644
index 0000000..c428a86
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/jemalloc.pc.in
@@ -0,0 +1,12 @@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+install_suffix=@install_suffix@
+
+Name: jemalloc
+Description: A general purpose malloc(3) implementation that emphasizes fragmentation avoidance and scalable concurrency support.
+URL: http://jemalloc.net/
+Version: @jemalloc_version_major@.@jemalloc_version_minor@.@jemalloc_version_bugfix@_@jemalloc_version_nrev@
+Cflags: -I${includedir}
+Libs: -L${libdir} -ljemalloc${install_suffix}
diff --git a/memkind-1.10.0/jemalloc/m4/ax_cxx_compile_stdcxx.m4 b/memkind-1.10.0/jemalloc/m4/ax_cxx_compile_stdcxx.m4
new file mode 100644
index 0000000..2c18e49
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/m4/ax_cxx_compile_stdcxx.m4
@@ -0,0 +1,562 @@
+# ===========================================================================
+#   http://www.gnu.org/software/autoconf-archive/ax_cxx_compile_stdcxx.html
+# ===========================================================================
+#
+# SYNOPSIS
+#
+#   AX_CXX_COMPILE_STDCXX(VERSION, [ext|noext], [mandatory|optional])
+#
+# DESCRIPTION
+#
+#   Check for baseline language coverage in the compiler for the specified
+#   version of the C++ standard.  If necessary, add switches to CXX and
+#   CXXCPP to enable support.  VERSION may be '11' (for the C++11 standard)
+#   or '14' (for the C++14 standard).
+#
+#   The second argument, if specified, indicates whether you insist on an
+#   extended mode (e.g. -std=gnu++11) or a strict conformance mode (e.g.
+#   -std=c++11).  If neither is specified, you get whatever works, with
+#   preference for an extended mode.
+#
+#   The third argument, if specified 'mandatory' or if left unspecified,
+#   indicates that baseline support for the specified C++ standard is
+#   required and that the macro should error out if no mode with that
+#   support is found.  If specified 'optional', then configuration proceeds
+#   regardless, after defining HAVE_CXX${VERSION} if and only if a
+#   supporting mode is found.
+#
+# LICENSE
+#
+#   Copyright (c) 2008 Benjamin Kosnik <bkoz@redhat.com>
+#   Copyright (c) 2012 Zack Weinberg <zackw@panix.com>
+#   Copyright (c) 2013 Roy Stogner <roystgnr@ices.utexas.edu>
+#   Copyright (c) 2014, 2015 Google Inc.; contributed by Alexey Sokolov <sokolov@google.com>
+#   Copyright (c) 2015 Paul Norman <penorman@mac.com>
+#   Copyright (c) 2015 Moritz Klammler <moritz@klammler.eu>
+#
+#   Copying and distribution of this file, with or without modification, are
+#   permitted in any medium without royalty provided the copyright notice
+#   and this notice are preserved.  This file is offered as-is, without any
+#   warranty.
+
+#serial 4
+
+dnl  This macro is based on the code from the AX_CXX_COMPILE_STDCXX_11 macro
+dnl  (serial version number 13).
+
+AC_DEFUN([AX_CXX_COMPILE_STDCXX], [dnl
+  m4_if([$1], [11], [],
+        [$1], [14], [],
+        [$1], [17], [m4_fatal([support for C++17 not yet implemented in AX_CXX_COMPILE_STDCXX])],
+        [m4_fatal([invalid first argument `$1' to AX_CXX_COMPILE_STDCXX])])dnl
+  m4_if([$2], [], [],
+        [$2], [ext], [],
+        [$2], [noext], [],
+        [m4_fatal([invalid second argument `$2' to AX_CXX_COMPILE_STDCXX])])dnl
+  m4_if([$3], [], [ax_cxx_compile_cxx$1_required=true],
+        [$3], [mandatory], [ax_cxx_compile_cxx$1_required=true],
+        [$3], [optional], [ax_cxx_compile_cxx$1_required=false],
+        [m4_fatal([invalid third argument `$3' to AX_CXX_COMPILE_STDCXX])])
+  AC_LANG_PUSH([C++])dnl
+  ac_success=no
+  AC_CACHE_CHECK(whether $CXX supports C++$1 features by default,
+  ax_cv_cxx_compile_cxx$1,
+  [AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_testbody_$1])],
+    [ax_cv_cxx_compile_cxx$1=yes],
+    [ax_cv_cxx_compile_cxx$1=no])])
+  if test x$ax_cv_cxx_compile_cxx$1 = xyes; then
+    ac_success=yes
+  fi
+
+  m4_if([$2], [noext], [], [dnl
+  if test x$ac_success = xno; then
+    for switch in -std=gnu++$1 -std=gnu++0x; do
+      cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx$1_$switch])
+      AC_CACHE_CHECK(whether $CXX supports C++$1 features with $switch,
+                     $cachevar,
+        [ac_save_CXX="$CXX"
+         CXX="$CXX $switch"
+         AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_testbody_$1])],
+          [eval $cachevar=yes],
+          [eval $cachevar=no])
+         CXX="$ac_save_CXX"])
+      if eval test x\$$cachevar = xyes; then
+        CXX="$CXX $switch"
+        if test -n "$CXXCPP" ; then
+          CXXCPP="$CXXCPP $switch"
+        fi
+        ac_success=yes
+        break
+      fi
+    done
+  fi])
+
+  m4_if([$2], [ext], [], [dnl
+  if test x$ac_success = xno; then
+    dnl HP's aCC needs +std=c++11 according to:
+    dnl http://h21007.www2.hp.com/portal/download/files/unprot/aCxx/PDF_Release_Notes/769149-001.pdf
+    dnl Cray's crayCC needs "-h std=c++11"
+    for switch in -std=c++$1 -std=c++0x +std=c++$1 "-h std=c++$1"; do
+      cachevar=AS_TR_SH([ax_cv_cxx_compile_cxx$1_$switch])
+      AC_CACHE_CHECK(whether $CXX supports C++$1 features with $switch,
+                     $cachevar,
+        [ac_save_CXX="$CXX"
+         CXX="$CXX $switch"
+         AC_COMPILE_IFELSE([AC_LANG_SOURCE([_AX_CXX_COMPILE_STDCXX_testbody_$1])],
+          [eval $cachevar=yes],
+          [eval $cachevar=no])
+         CXX="$ac_save_CXX"])
+      if eval test x\$$cachevar = xyes; then
+        CXX="$CXX $switch"
+        if test -n "$CXXCPP" ; then
+          CXXCPP="$CXXCPP $switch"
+        fi
+        ac_success=yes
+        break
+      fi
+    done
+  fi])
+  AC_LANG_POP([C++])
+  if test x$ax_cxx_compile_cxx$1_required = xtrue; then
+    if test x$ac_success = xno; then
+      AC_MSG_ERROR([*** A compiler with support for C++$1 language features is required.])
+    fi
+  fi
+  if test x$ac_success = xno; then
+    HAVE_CXX$1=0
+    AC_MSG_NOTICE([No compiler with C++$1 support was found])
+  else
+    HAVE_CXX$1=1
+    AC_DEFINE(HAVE_CXX$1,1,
+              [define if the compiler supports basic C++$1 syntax])
+  fi
+  AC_SUBST(HAVE_CXX$1)
+])
+
+
+dnl  Test body for checking C++11 support
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_11],
+  _AX_CXX_COMPILE_STDCXX_testbody_new_in_11
+)
+
+
+dnl  Test body for checking C++14 support
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_14],
+  _AX_CXX_COMPILE_STDCXX_testbody_new_in_11
+  _AX_CXX_COMPILE_STDCXX_testbody_new_in_14
+)
+
+
+dnl  Tests for new features in C++11
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_new_in_11], [[
+
+// If the compiler admits that it is not ready for C++11, why torture it?
+// Hopefully, this will speed up the test.
+
+#ifndef __cplusplus
+
+#error "This is not a C++ compiler"
+
+#elif __cplusplus < 201103L
+
+#error "This is not a C++11 compiler"
+
+#else
+
+namespace cxx11
+{
+
+  namespace test_static_assert
+  {
+
+    template <typename T>
+    struct check
+    {
+      static_assert(sizeof(int) <= sizeof(T), "not big enough");
+    };
+
+  }
+
+  namespace test_final_override
+  {
+
+    struct Base
+    {
+      virtual void f() {}
+    };
+
+    struct Derived : public Base
+    {
+      virtual void f() override {}
+    };
+
+  }
+
+  namespace test_double_right_angle_brackets
+  {
+
+    template < typename T >
+    struct check {};
+
+    typedef check<void> single_type;
+    typedef check<check<void>> double_type;
+    typedef check<check<check<void>>> triple_type;
+    typedef check<check<check<check<void>>>> quadruple_type;
+
+  }
+
+  namespace test_decltype
+  {
+
+    int
+    f()
+    {
+      int a = 1;
+      decltype(a) b = 2;
+      return a + b;
+    }
+
+  }
+
+  namespace test_type_deduction
+  {
+
+    template < typename T1, typename T2 >
+    struct is_same
+    {
+      static const bool value = false;
+    };
+
+    template < typename T >
+    struct is_same<T, T>
+    {
+      static const bool value = true;
+    };
+
+    template < typename T1, typename T2 >
+    auto
+    add(T1 a1, T2 a2) -> decltype(a1 + a2)
+    {
+      return a1 + a2;
+    }
+
+    int
+    test(const int c, volatile int v)
+    {
+      static_assert(is_same<int, decltype(0)>::value == true, "");
+      static_assert(is_same<int, decltype(c)>::value == false, "");
+      static_assert(is_same<int, decltype(v)>::value == false, "");
+      auto ac = c;
+      auto av = v;
+      auto sumi = ac + av + 'x';
+      auto sumf = ac + av + 1.0;
+      static_assert(is_same<int, decltype(ac)>::value == true, "");
+      static_assert(is_same<int, decltype(av)>::value == true, "");
+      static_assert(is_same<int, decltype(sumi)>::value == true, "");
+      static_assert(is_same<int, decltype(sumf)>::value == false, "");
+      static_assert(is_same<int, decltype(add(c, v))>::value == true, "");
+      return (sumf > 0.0) ? sumi : add(c, v);
+    }
+
+  }
+
+  namespace test_noexcept
+  {
+
+    int f() { return 0; }
+    int g() noexcept { return 0; }
+
+    static_assert(noexcept(f()) == false, "");
+    static_assert(noexcept(g()) == true, "");
+
+  }
+
+  namespace test_constexpr
+  {
+
+    template < typename CharT >
+    unsigned long constexpr
+    strlen_c_r(const CharT *const s, const unsigned long acc) noexcept
+    {
+      return *s ? strlen_c_r(s + 1, acc + 1) : acc;
+    }
+
+    template < typename CharT >
+    unsigned long constexpr
+    strlen_c(const CharT *const s) noexcept
+    {
+      return strlen_c_r(s, 0UL);
+    }
+
+    static_assert(strlen_c("") == 0UL, "");
+    static_assert(strlen_c("1") == 1UL, "");
+    static_assert(strlen_c("example") == 7UL, "");
+    static_assert(strlen_c("another\0example") == 7UL, "");
+
+  }
+
+  namespace test_rvalue_references
+  {
+
+    template < int N >
+    struct answer
+    {
+      static constexpr int value = N;
+    };
+
+    answer<1> f(int&)       { return answer<1>(); }
+    answer<2> f(const int&) { return answer<2>(); }
+    answer<3> f(int&&)      { return answer<3>(); }
+
+    void
+    test()
+    {
+      int i = 0;
+      const int c = 0;
+      static_assert(decltype(f(i))::value == 1, "");
+      static_assert(decltype(f(c))::value == 2, "");
+      static_assert(decltype(f(0))::value == 3, "");
+    }
+
+  }
+
+  namespace test_uniform_initialization
+  {
+
+    struct test
+    {
+      static const int zero {};
+      static const int one {1};
+    };
+
+    static_assert(test::zero == 0, "");
+    static_assert(test::one == 1, "");
+
+  }
+
+  namespace test_lambdas
+  {
+
+    void
+    test1()
+    {
+      auto lambda1 = [](){};
+      auto lambda2 = lambda1;
+      lambda1();
+      lambda2();
+    }
+
+    int
+    test2()
+    {
+      auto a = [](int i, int j){ return i + j; }(1, 2);
+      auto b = []() -> int { return '0'; }();
+      auto c = [=](){ return a + b; }();
+      auto d = [&](){ return c; }();
+      auto e = [a, &b](int x) mutable {
+        const auto identity = [](int y){ return y; };
+        for (auto i = 0; i < a; ++i)
+          a += b--;
+        return x + identity(a + b);
+      }(0);
+      return a + b + c + d + e;
+    }
+
+    int
+    test3()
+    {
+      const auto nullary = [](){ return 0; };
+      const auto unary = [](int x){ return x; };
+      using nullary_t = decltype(nullary);
+      using unary_t = decltype(unary);
+      const auto higher1st = [](nullary_t f){ return f(); };
+      const auto higher2nd = [unary](nullary_t f1){
+        return [unary, f1](unary_t f2){ return f2(unary(f1())); };
+      };
+      return higher1st(nullary) + higher2nd(nullary)(unary);
+    }
+
+  }
+
+  namespace test_variadic_templates
+  {
+
+    template <int...>
+    struct sum;
+
+    template <int N0, int... N1toN>
+    struct sum<N0, N1toN...>
+    {
+      static constexpr auto value = N0 + sum<N1toN...>::value;
+    };
+
+    template <>
+    struct sum<>
+    {
+      static constexpr auto value = 0;
+    };
+
+    static_assert(sum<>::value == 0, "");
+    static_assert(sum<1>::value == 1, "");
+    static_assert(sum<23>::value == 23, "");
+    static_assert(sum<1, 2>::value == 3, "");
+    static_assert(sum<5, 5, 11>::value == 21, "");
+    static_assert(sum<2, 3, 5, 7, 11, 13>::value == 41, "");
+
+  }
+
+  // http://stackoverflow.com/questions/13728184/template-aliases-and-sfinae
+  // Clang 3.1 fails with headers of libstd++ 4.8.3 when using std::function
+  // because of this.
+  namespace test_template_alias_sfinae
+  {
+
+    struct foo {};
+
+    template<typename T>
+    using member = typename T::member_type;
+
+    template<typename T>
+    void func(...) {}
+
+    template<typename T>
+    void func(member<T>*) {}
+
+    void test();
+
+    void test() { func<foo>(0); }
+
+  }
+
+}  // namespace cxx11
+
+#endif  // __cplusplus >= 201103L
+
+]])
+
+
+dnl  Tests for new features in C++14
+
+m4_define([_AX_CXX_COMPILE_STDCXX_testbody_new_in_14], [[
+
+// If the compiler admits that it is not ready for C++14, why torture it?
+// Hopefully, this will speed up the test.
+
+#ifndef __cplusplus
+
+#error "This is not a C++ compiler"
+
+#elif __cplusplus < 201402L
+
+#error "This is not a C++14 compiler"
+
+#else
+
+namespace cxx14
+{
+
+  namespace test_polymorphic_lambdas
+  {
+
+    int
+    test()
+    {
+      const auto lambda = [](auto&&... args){
+        const auto istiny = [](auto x){
+          return (sizeof(x) == 1UL) ? 1 : 0;
+        };
+        const int aretiny[] = { istiny(args)... };
+        return aretiny[0];
+      };
+      return lambda(1, 1L, 1.0f, '1');
+    }
+
+  }
+
+  namespace test_binary_literals
+  {
+
+    constexpr auto ivii = 0b0000000000101010;
+    static_assert(ivii == 42, "wrong value");
+
+  }
+
+  namespace test_generalized_constexpr
+  {
+
+    template < typename CharT >
+    constexpr unsigned long
+    strlen_c(const CharT *const s) noexcept
+    {
+      auto length = 0UL;
+      for (auto p = s; *p; ++p)
+        ++length;
+      return length;
+    }
+
+    static_assert(strlen_c("") == 0UL, "");
+    static_assert(strlen_c("x") == 1UL, "");
+    static_assert(strlen_c("test") == 4UL, "");
+    static_assert(strlen_c("another\0test") == 7UL, "");
+
+  }
+
+  namespace test_lambda_init_capture
+  {
+
+    int
+    test()
+    {
+      auto x = 0;
+      const auto lambda1 = [a = x](int b){ return a + b; };
+      const auto lambda2 = [a = lambda1(x)](){ return a; };
+      return lambda2();
+    }
+
+  }
+
+  namespace test_digit_seperators
+  {
+
+    constexpr auto ten_million = 100'000'000;
+    static_assert(ten_million == 100000000, "");
+
+  }
+
+  namespace test_return_type_deduction
+  {
+
+    auto f(int& x) { return x; }
+    decltype(auto) g(int& x) { return x; }
+
+    template < typename T1, typename T2 >
+    struct is_same
+    {
+      static constexpr auto value = false;
+    };
+
+    template < typename T >
+    struct is_same<T, T>
+    {
+      static constexpr auto value = true;
+    };
+
+    int
+    test()
+    {
+      auto x = 0;
+      static_assert(is_same<int, decltype(f(x))>::value, "");
+      static_assert(is_same<int&, decltype(g(x))>::value, "");
+      return x;
+    }
+
+  }
+
+}  // namespace cxx14
+
+#endif  // __cplusplus >= 201402L
+
+]])
diff --git a/memkind-1.10.0/jemalloc/msvc/ReadMe.txt b/memkind-1.10.0/jemalloc/msvc/ReadMe.txt
new file mode 100644
index 0000000..633a7d4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/ReadMe.txt
@@ -0,0 +1,23 @@
+
+How to build jemalloc for Windows
+=================================
+
+1. Install Cygwin with at least the following packages:
+   * autoconf
+   * autogen
+   * gawk
+   * grep
+   * sed
+
+2. Install Visual Studio 2015 or 2017 with Visual C++
+
+3. Add Cygwin\bin to the PATH environment variable
+
+4. Open "x64 Native Tools Command Prompt for VS 2017"
+   (note: x86/x64 doesn't matter at this point)
+
+5. Generate header files:
+   sh -c "CC=cl ./autogen.sh"
+
+6. Now the project can be opened and built in Visual Studio:
+   msvc\jemalloc_vc2017.sln
diff --git a/memkind-1.10.0/jemalloc/msvc/jemalloc_vc2015.sln b/memkind-1.10.0/jemalloc/msvc/jemalloc_vc2015.sln
new file mode 100644
index 0000000..aedd5e5
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/jemalloc_vc2015.sln
@@ -0,0 +1,63 @@
+
+Microsoft Visual Studio Solution File, Format Version 12.00
+# Visual Studio 14
+VisualStudioVersion = 14.0.24720.0
+MinimumVisualStudioVersion = 10.0.40219.1
+Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "Solution Items", "Solution Items", "{70A99006-6DE9-472B-8F83-4CEE6C616DF3}"
+	ProjectSection(SolutionItems) = preProject
+		ReadMe.txt = ReadMe.txt
+	EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "jemalloc", "projects\vc2015\jemalloc\jemalloc.vcxproj", "{8D6BB292-9E1C-413D-9F98-4864BDC1514A}"
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "test_threads", "projects\vc2015\test_threads\test_threads.vcxproj", "{09028CFD-4EB7-491D-869C-0708DB97ED44}"
+EndProject
+Global
+	GlobalSection(SolutionConfigurationPlatforms) = preSolution
+		Debug|x64 = Debug|x64
+		Debug|x86 = Debug|x86
+		Debug-static|x64 = Debug-static|x64
+		Debug-static|x86 = Debug-static|x86
+		Release|x64 = Release|x64
+		Release|x86 = Release|x86
+		Release-static|x64 = Release-static|x64
+		Release-static|x86 = Release-static|x86
+	EndGlobalSection
+	GlobalSection(ProjectConfigurationPlatforms) = postSolution
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.ActiveCfg = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.Build.0 = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.ActiveCfg = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.Build.0 = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.Build.0 = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.ActiveCfg = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.Build.0 = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.ActiveCfg = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.Build.0 = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.ActiveCfg = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.Build.0 = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.Build.0 = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.ActiveCfg = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.Build.0 = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.ActiveCfg = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.Build.0 = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.Build.0 = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.ActiveCfg = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.Build.0 = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.ActiveCfg = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.Build.0 = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.ActiveCfg = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.Build.0 = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.Build.0 = Release-static|Win32
+	EndGlobalSection
+	GlobalSection(SolutionProperties) = preSolution
+		HideSolutionNode = FALSE
+	EndGlobalSection
+EndGlobal
diff --git a/memkind-1.10.0/jemalloc/msvc/jemalloc_vc2017.sln b/memkind-1.10.0/jemalloc/msvc/jemalloc_vc2017.sln
new file mode 100644
index 0000000..c22fcb4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/jemalloc_vc2017.sln
@@ -0,0 +1,63 @@
+
+Microsoft Visual Studio Solution File, Format Version 12.00
+# Visual Studio 14
+VisualStudioVersion = 14.0.24720.0
+MinimumVisualStudioVersion = 10.0.40219.1
+Project("{2150E333-8FDC-42A3-9474-1A3956D46DE8}") = "Solution Items", "Solution Items", "{70A99006-6DE9-472B-8F83-4CEE6C616DF3}"
+	ProjectSection(SolutionItems) = preProject
+		ReadMe.txt = ReadMe.txt
+	EndProjectSection
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "jemalloc", "projects\vc2017\jemalloc\jemalloc.vcxproj", "{8D6BB292-9E1C-413D-9F98-4864BDC1514A}"
+EndProject
+Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "test_threads", "projects\vc2017\test_threads\test_threads.vcxproj", "{09028CFD-4EB7-491D-869C-0708DB97ED44}"
+EndProject
+Global
+	GlobalSection(SolutionConfigurationPlatforms) = preSolution
+		Debug|x64 = Debug|x64
+		Debug|x86 = Debug|x86
+		Debug-static|x64 = Debug-static|x64
+		Debug-static|x86 = Debug-static|x86
+		Release|x64 = Release|x64
+		Release|x86 = Release|x86
+		Release-static|x64 = Release-static|x64
+		Release-static|x86 = Release-static|x86
+	EndGlobalSection
+	GlobalSection(ProjectConfigurationPlatforms) = postSolution
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.ActiveCfg = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x64.Build.0 = Debug|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.ActiveCfg = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug|x86.Build.0 = Debug|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x64.Build.0 = Debug-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.ActiveCfg = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x64.Build.0 = Release|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.ActiveCfg = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release|x86.Build.0 = Release|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.ActiveCfg = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x64.Build.0 = Release-static|x64
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{8D6BB292-9E1C-413D-9F98-4864BDC1514A}.Release-static|x86.Build.0 = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.ActiveCfg = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x64.Build.0 = Debug|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.ActiveCfg = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug|x86.Build.0 = Debug|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.ActiveCfg = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x64.Build.0 = Debug-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.ActiveCfg = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Debug-static|x86.Build.0 = Debug-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.ActiveCfg = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x64.Build.0 = Release|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.ActiveCfg = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release|x86.Build.0 = Release|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.ActiveCfg = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x64.Build.0 = Release-static|x64
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.ActiveCfg = Release-static|Win32
+		{09028CFD-4EB7-491D-869C-0708DB97ED44}.Release-static|x86.Build.0 = Release-static|Win32
+	EndGlobalSection
+	GlobalSection(SolutionProperties) = preSolution
+		HideSolutionNode = FALSE
+	EndGlobalSection
+EndGlobal
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
new file mode 100644
index 0000000..228e8be
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj
@@ -0,0 +1,350 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="14.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c" />
+    <ClCompile Include="..\..\..\..\src\background_thread.c" />
+    <ClCompile Include="..\..\..\..\src\base.c" />
+    <ClCompile Include="..\..\..\..\src\bin.c" />
+    <ClCompile Include="..\..\..\..\src\bitmap.c" />
+    <ClCompile Include="..\..\..\..\src\ckh.c" />
+    <ClCompile Include="..\..\..\..\src\ctl.c" />
+    <ClCompile Include="..\..\..\..\src\div.c" />
+    <ClCompile Include="..\..\..\..\src\extent.c" />
+    <ClCompile Include="..\..\..\..\src\extent_dss.c" />
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c" />
+    <ClCompile Include="..\..\..\..\src\hash.c" />
+    <ClCompile Include="..\..\..\..\src\hook.c" />
+    <ClCompile Include="..\..\..\..\src\jemalloc.c" />
+    <ClCompile Include="..\..\..\..\src\large.c" />
+    <ClCompile Include="..\..\..\..\src\log.c" />
+    <ClCompile Include="..\..\..\..\src\malloc_io.c" />
+    <ClCompile Include="..\..\..\..\src\mutex.c" />
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c" />
+    <ClCompile Include="..\..\..\..\src\nstime.c" />
+    <ClCompile Include="..\..\..\..\src\pages.c" />
+    <ClCompile Include="..\..\..\..\src\prng.c" />
+    <ClCompile Include="..\..\..\..\src\prof.c" />
+    <ClCompile Include="..\..\..\..\src\rtree.c" />
+    <ClCompile Include="..\..\..\..\src\sc.c" />
+    <ClCompile Include="..\..\..\..\src\stats.c" />
+    <ClCompile Include="..\..\..\..\src\sz.c" />
+    <ClCompile Include="..\..\..\..\src\tcache.c" />
+    <ClCompile Include="..\..\..\..\src\ticker.c" />
+    <ClCompile Include="..\..\..\..\src\tsd.c" />
+    <ClCompile Include="..\..\..\..\src\witness.c" />
+    <ClCompile Include="..\..\..\..\src\safety_check.c" />
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{8D6BB292-9E1C-413D-9F98-4864BDC1514A}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>jemalloc</RootNamespace>
+    <WindowsTargetPlatformVersion>8.1</WindowsTargetPlatformVersion>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+      <MinimalRebuild>false</MinimalRebuild>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
new file mode 100644
index 0000000..d839515
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/jemalloc/jemalloc.vcxproj.filters
@@ -0,0 +1,107 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\background_thread.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\base.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bitmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ckh.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ctl.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_dss.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hash.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hook.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\jemalloc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\large.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\malloc_io.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\nstime.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\pages.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prng.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prof.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\rtree.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\stats.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sz.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tcache.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ticker.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tsd.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\witness.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\log.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bin.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\div.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\safety_check.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+</Project>
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
new file mode 100644
index 0000000..325876d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj
@@ -0,0 +1,327 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="14.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{09028CFD-4EB7-491D-869C-0708DB97ED44}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>test_threads</RootNamespace>
+    <WindowsTargetPlatformVersion>8.1</WindowsTargetPlatformVersion>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v140</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp" />
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="..\jemalloc\jemalloc.vcxproj">
+      <Project>{8d6bb292-9e1c-413d-9f98-4864bdc1514a}</Project>
+    </ProjectReference>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h" />
+  </ItemGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
\ No newline at end of file
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
new file mode 100644
index 0000000..fa4588f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2015/test_threads/test_threads.vcxproj.filters
@@ -0,0 +1,26 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+    <Filter Include="Header Files">
+      <UniqueIdentifier>{93995380-89BD-4b04-88EB-625FBE52EBFB}</UniqueIdentifier>
+      <Extensions>h;hh;hpp;hxx;hm;inl;inc;xsd</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+  </ItemGroup>
+</Project>
\ No newline at end of file
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
new file mode 100644
index 0000000..edcceed
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj
@@ -0,0 +1,350 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="15.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c" />
+    <ClCompile Include="..\..\..\..\src\background_thread.c" />
+    <ClCompile Include="..\..\..\..\src\base.c" />
+    <ClCompile Include="..\..\..\..\src\bin.c" />
+    <ClCompile Include="..\..\..\..\src\bitmap.c" />
+    <ClCompile Include="..\..\..\..\src\ckh.c" />
+    <ClCompile Include="..\..\..\..\src\ctl.c" />
+    <ClCompile Include="..\..\..\..\src\div.c" />
+    <ClCompile Include="..\..\..\..\src\extent.c" />
+    <ClCompile Include="..\..\..\..\src\extent_dss.c" />
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c" />
+    <ClCompile Include="..\..\..\..\src\hash.c" />
+    <ClCompile Include="..\..\..\..\src\hook.c" />
+    <ClCompile Include="..\..\..\..\src\jemalloc.c" />
+    <ClCompile Include="..\..\..\..\src\large.c" />
+    <ClCompile Include="..\..\..\..\src\log.c" />
+    <ClCompile Include="..\..\..\..\src\malloc_io.c" />
+    <ClCompile Include="..\..\..\..\src\mutex.c" />
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c" />
+    <ClCompile Include="..\..\..\..\src\nstime.c" />
+    <ClCompile Include="..\..\..\..\src\pages.c" />
+    <ClCompile Include="..\..\..\..\src\prng.c" />
+    <ClCompile Include="..\..\..\..\src\prof.c" />
+    <ClCompile Include="..\..\..\..\src\rtree.c" />
+    <ClCompile Include="..\..\..\..\src\sc.c" />
+    <ClCompile Include="..\..\..\..\src\stats.c" />
+    <ClCompile Include="..\..\..\..\src\sz.c" />
+    <ClCompile Include="..\..\..\..\src\tcache.c" />
+    <ClCompile Include="..\..\..\..\src\test_hooks.c" />
+    <ClCompile Include="..\..\..\..\src\ticker.c" />
+    <ClCompile Include="..\..\..\..\src\tsd.c" />
+    <ClCompile Include="..\..\..\..\src\witness.c" />
+    <ClCompile Include="..\..\..\..\src\safety_check.c" />
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{8D6BB292-9E1C-413D-9F98-4864BDC1514A}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>jemalloc</RootNamespace>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>DynamicLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>StaticLibrary</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-$(PlatformToolset)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)d</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <TargetName>$(ProjectName)-vc$(PlatformToolsetVersion)-$(Configuration)</TargetName>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;JEMALLOC_DEBUG;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;JEMALLOC_DEBUG;_REENTRANT;JEMALLOC_EXPORT=;_DEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+      <MinimalRebuild>false</MinimalRebuild>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;_WINDLL;DLLEXPORT;NDEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <ProgramDataBaseFileName>$(OutputPath)$(TargetName).pdb</ProgramDataBaseFileName>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_NO_PRIVATE_NAMESPACE;_REENTRANT;JEMALLOC_EXPORT=;NDEBUG;_LIB;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+      <DisableSpecificWarnings>4090;4146;4267;4334</DisableSpecificWarnings>
+      <DebugInformationFormat>OldStyle</DebugInformationFormat>
+    </ClCompile>
+    <Link>
+      <SubSystem>Windows</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+    </Link>
+  </ItemDefinitionGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
new file mode 100644
index 0000000..6df7260
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/jemalloc/jemalloc.vcxproj.filters
@@ -0,0 +1,110 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\..\src\arena.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\background_thread.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\base.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bitmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ckh.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ctl.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_dss.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\extent_mmap.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hash.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\hook.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\jemalloc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\large.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\malloc_io.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\mutex_pool.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\nstime.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\pages.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prng.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\prof.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\rtree.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sc.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\stats.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\sz.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tcache.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\ticker.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\tsd.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\witness.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\log.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\bin.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\div.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\test_hooks.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\..\src\safety_check.c">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+</Project>
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
new file mode 100644
index 0000000..c35b0f5
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj
@@ -0,0 +1,326 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project DefaultTargets="Build" ToolsVersion="15.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup Label="ProjectConfigurations">
+    <ProjectConfiguration Include="Debug-static|Win32">
+      <Configuration>Debug-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug-static|x64">
+      <Configuration>Debug-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|Win32">
+      <Configuration>Debug</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|Win32">
+      <Configuration>Release-static</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release-static|x64">
+      <Configuration>Release-static</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|Win32">
+      <Configuration>Release</Configuration>
+      <Platform>Win32</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Debug|x64">
+      <Configuration>Debug</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+    <ProjectConfiguration Include="Release|x64">
+      <Configuration>Release</Configuration>
+      <Platform>x64</Platform>
+    </ProjectConfiguration>
+  </ItemGroup>
+  <PropertyGroup Label="Globals">
+    <ProjectGuid>{09028CFD-4EB7-491D-869C-0708DB97ED44}</ProjectGuid>
+    <Keyword>Win32Proj</Keyword>
+    <RootNamespace>test_threads</RootNamespace>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>true</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="Configuration">
+    <ConfigurationType>Application</ConfigurationType>
+    <UseDebugLibraries>false</UseDebugLibraries>
+    <PlatformToolset>v141</PlatformToolset>
+    <WholeProgramOptimization>true</WholeProgramOptimization>
+    <CharacterSet>MultiByte</CharacterSet>
+  </PropertyGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.props" />
+  <ImportGroup Label="ExtensionSettings">
+  </ImportGroup>
+  <ImportGroup Label="Shared">
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Label="PropertySheets" Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <ImportGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'" Label="PropertySheets">
+    <Import Project="$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props" Condition="exists('$(UserRootDir)\Microsoft.Cpp.$(Platform).user.props')" Label="LocalAppDataPlatform" />
+  </ImportGroup>
+  <PropertyGroup Label="UserMacros" />
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>true</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <LinkIncremental>true</LinkIncremental>
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <OutDir>$(SolutionDir)$(Platform)\$(Configuration)\</OutDir>
+    <IntDir>$(Platform)\$(Configuration)\</IntDir>
+    <LinkIncremental>false</LinkIncremental>
+  </PropertyGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>WIN32;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|Win32'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemallocd.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Debug-static|x64'">
+    <ClCompile>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <WarningLevel>Level3</WarningLevel>
+      <Optimization>Disabled</Optimization>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;_DEBUG;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreadedDebug</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>WIN32;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|Win32'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-$(PlatformToolset)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc.lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemDefinitionGroup Condition="'$(Configuration)|$(Platform)'=='Release-static|x64'">
+    <ClCompile>
+      <WarningLevel>Level3</WarningLevel>
+      <PrecompiledHeader>
+      </PrecompiledHeader>
+      <Optimization>MaxSpeed</Optimization>
+      <FunctionLevelLinking>true</FunctionLevelLinking>
+      <IntrinsicFunctions>true</IntrinsicFunctions>
+      <PreprocessorDefinitions>JEMALLOC_EXPORT=;JEMALLOC_STATIC;NDEBUG;_CONSOLE;%(PreprocessorDefinitions)</PreprocessorDefinitions>
+      <AdditionalIncludeDirectories>..\..\..\..\test\include;..\..\..\..\include;..\..\..\..\include\msvc_compat;%(AdditionalIncludeDirectories)</AdditionalIncludeDirectories>
+      <RuntimeLibrary>MultiThreaded</RuntimeLibrary>
+    </ClCompile>
+    <Link>
+      <SubSystem>Console</SubSystem>
+      <GenerateDebugInformation>true</GenerateDebugInformation>
+      <EnableCOMDATFolding>true</EnableCOMDATFolding>
+      <OptimizeReferences>true</OptimizeReferences>
+      <AdditionalLibraryDirectories>$(SolutionDir)$(Platform)\$(Configuration)</AdditionalLibraryDirectories>
+      <AdditionalDependencies>jemalloc-vc$(PlatformToolsetVersion)-$(Configuration).lib;kernel32.lib;user32.lib;gdi32.lib;winspool.lib;comdlg32.lib;advapi32.lib;shell32.lib;ole32.lib;oleaut32.lib;uuid.lib;odbc32.lib;odbccp32.lib;%(AdditionalDependencies)</AdditionalDependencies>
+    </Link>
+  </ItemDefinitionGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp" />
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="..\jemalloc\jemalloc.vcxproj">
+      <Project>{8d6bb292-9e1c-413d-9f98-4864bdc1514a}</Project>
+    </ProjectReference>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h" />
+  </ItemGroup>
+  <Import Project="$(VCTargetsPath)\Microsoft.Cpp.targets" />
+  <ImportGroup Label="ExtensionTargets">
+  </ImportGroup>
+</Project>
\ No newline at end of file
diff --git a/memkind-1.10.0/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
new file mode 100644
index 0000000..fa4588f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/projects/vc2017/test_threads/test_threads.vcxproj.filters
@@ -0,0 +1,26 @@
+<?xml version="1.0" encoding="utf-8"?>
+<Project ToolsVersion="4.0" xmlns="http://schemas.microsoft.com/developer/msbuild/2003">
+  <ItemGroup>
+    <Filter Include="Source Files">
+      <UniqueIdentifier>{4FC737F1-C7A5-4376-A066-2A32D752A2FF}</UniqueIdentifier>
+      <Extensions>cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx</Extensions>
+    </Filter>
+    <Filter Include="Header Files">
+      <UniqueIdentifier>{93995380-89BD-4b04-88EB-625FBE52EBFB}</UniqueIdentifier>
+      <Extensions>h;hh;hpp;hxx;hm;inl;inc;xsd</Extensions>
+    </Filter>
+  </ItemGroup>
+  <ItemGroup>
+    <ClCompile Include="..\..\..\test_threads\test_threads.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+    <ClCompile Include="..\..\..\test_threads\test_threads_main.cpp">
+      <Filter>Source Files</Filter>
+    </ClCompile>
+  </ItemGroup>
+  <ItemGroup>
+    <ClInclude Include="..\..\..\test_threads\test_threads.h">
+      <Filter>Header Files</Filter>
+    </ClInclude>
+  </ItemGroup>
+</Project>
\ No newline at end of file
diff --git a/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads.cpp b/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads.cpp
new file mode 100644
index 0000000..92e3162
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads.cpp
@@ -0,0 +1,88 @@
+// jemalloc C++ threaded test
+// Author: Rustam Abdullaev
+// Public Domain
+
+#include <atomic>
+#include <functional>
+#include <future>
+#include <random>
+#include <thread>
+#include <vector>
+#include <stdio.h>
+#include <jemalloc/jemalloc.h>
+
+using std::vector;
+using std::thread;
+using std::uniform_int_distribution;
+using std::minstd_rand;
+
+int test_threads() {
+  je_malloc_conf = "narenas:3";
+  int narenas = 0;
+  size_t sz = sizeof(narenas);
+  je_mallctl("opt.narenas", (void *)&narenas, &sz, NULL, 0);
+  if (narenas != 3) {
+    printf("Error: unexpected number of arenas: %d\n", narenas);
+    return 1;
+  }
+  static const int sizes[] = { 7, 16, 32, 60, 91, 100, 120, 144, 169, 199, 255, 400, 670, 900, 917, 1025, 3333, 5190, 13131, 49192, 99999, 123123, 255265, 2333111 };
+  static const int numSizes = (int)(sizeof(sizes) / sizeof(sizes[0]));
+  vector<thread> workers;
+  static const int numThreads = narenas + 1, numAllocsMax = 25, numIter1 = 50, numIter2 = 50;
+  je_malloc_stats_print(NULL, NULL, NULL);
+  size_t allocated1;
+  size_t sz1 = sizeof(allocated1);
+  je_mallctl("stats.active", (void *)&allocated1, &sz1, NULL, 0);
+  printf("\nPress Enter to start threads...\n");
+  getchar();
+  printf("Starting %d threads x %d x %d iterations...\n", numThreads, numIter1, numIter2);
+  for (int i = 0; i < numThreads; i++) {
+    workers.emplace_back([tid=i]() {
+      uniform_int_distribution<int> sizeDist(0, numSizes - 1);
+      minstd_rand rnd(tid * 17);
+      uint8_t* ptrs[numAllocsMax];
+      int ptrsz[numAllocsMax];
+      for (int i = 0; i < numIter1; ++i) {
+        thread t([&]() {
+          for (int i = 0; i < numIter2; ++i) {
+            const int numAllocs = numAllocsMax - sizeDist(rnd);
+            for (int j = 0; j < numAllocs; j += 64) {
+              const int x = sizeDist(rnd);
+              const int sz = sizes[x];
+              ptrsz[j] = sz;
+              ptrs[j] = (uint8_t*)je_malloc(sz);
+              if (!ptrs[j]) {
+                printf("Unable to allocate %d bytes in thread %d, iter %d, alloc %d. %d\n", sz, tid, i, j, x);
+                exit(1);
+              }
+              for (int k = 0; k < sz; k++)
+                ptrs[j][k] = tid + k;
+            }
+            for (int j = 0; j < numAllocs; j += 64) {
+              for (int k = 0, sz = ptrsz[j]; k < sz; k++)
+                if (ptrs[j][k] != (uint8_t)(tid + k)) {
+                  printf("Memory error in thread %d, iter %d, alloc %d @ %d : %02X!=%02X\n", tid, i, j, k, ptrs[j][k], (uint8_t)(tid + k));
+                  exit(1);
+                }
+              je_free(ptrs[j]);
+            }
+          }
+        });
+        t.join();
+      }
+    });
+  }
+  for (thread& t : workers) {
+    t.join();
+  }
+  je_malloc_stats_print(NULL, NULL, NULL);
+  size_t allocated2;
+  je_mallctl("stats.active", (void *)&allocated2, &sz1, NULL, 0);
+  size_t leaked = allocated2 - allocated1;
+  printf("\nDone. Leaked: %zd bytes\n", leaked);
+  bool failed = leaked > 65536; // in case C++ runtime allocated something (e.g. iostream locale or facet)
+  printf("\nTest %s!\n", (failed ? "FAILED" : "successful"));
+  printf("\nPress Enter to continue...\n");
+  getchar();
+  return failed ? 1 : 0;
+}
diff --git a/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads.h b/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads.h
new file mode 100644
index 0000000..64d0cdb
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads.h
@@ -0,0 +1,3 @@
+#pragma once
+
+int test_threads();
diff --git a/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads_main.cpp b/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads_main.cpp
new file mode 100644
index 0000000..0a022fb
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/msvc/test_threads/test_threads_main.cpp
@@ -0,0 +1,11 @@
+#include "test_threads.h"
+#include <future>
+#include <functional>
+#include <chrono>
+
+using namespace std::chrono_literals;
+
+int main(int argc, char** argv) {
+  int rc = test_threads();
+  return rc;
+}
diff --git a/memkind-1.10.0/jemalloc/run_tests.sh b/memkind-1.10.0/jemalloc/run_tests.sh
new file mode 100755
index 0000000..b434f15
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/run_tests.sh
@@ -0,0 +1 @@
+$(dirname "$)")/scripts/gen_run_tests.py | bash
diff --git a/memkind-1.10.0/jemalloc/scripts/gen_run_tests.py b/memkind-1.10.0/jemalloc/scripts/gen_run_tests.py
new file mode 100755
index 0000000..a414f81
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/scripts/gen_run_tests.py
@@ -0,0 +1,126 @@
+#!/usr/bin/env python
+
+import sys
+from itertools import combinations
+from os import uname
+from multiprocessing import cpu_count
+from subprocess import call
+
+# Later, we want to test extended vaddr support.  Apparently, the "real" way of
+# checking this is flaky on OS X.
+bits_64 = sys.maxsize > 2**32
+
+nparallel = cpu_count() * 2
+
+uname = uname()[0]
+
+if "BSD" in uname:
+    make_cmd = 'gmake'
+else:
+    make_cmd = 'make'
+
+def powerset(items):
+    result = []
+    for i in xrange(len(items) + 1):
+        result += combinations(items, i)
+    return result
+
+possible_compilers = []
+for cc, cxx in (['gcc', 'g++'], ['clang', 'clang++']):
+    try:
+        cmd_ret = call([cc, "-v"])
+        if cmd_ret == 0:
+            possible_compilers.append((cc, cxx))
+    except:
+        pass
+possible_compiler_opts = [
+    '-m32',
+]
+possible_config_opts = [
+    '--enable-debug',
+    '--enable-prof',
+    '--disable-stats',
+    '--enable-opt-safety-checks',
+]
+if bits_64:
+    possible_config_opts.append('--with-lg-vaddr=56')
+
+possible_malloc_conf_opts = [
+    'tcache:false',
+    'dss:primary',
+    'percpu_arena:percpu',
+    'background_thread:true',
+]
+
+print 'set -e'
+print 'if [ -f Makefile ] ; then %(make_cmd)s relclean ; fi' % {'make_cmd': make_cmd}
+print 'autoconf'
+print 'rm -rf run_tests.out'
+print 'mkdir run_tests.out'
+print 'cd run_tests.out'
+
+ind = 0
+for cc, cxx in possible_compilers:
+    for compiler_opts in powerset(possible_compiler_opts):
+        for config_opts in powerset(possible_config_opts):
+            for malloc_conf_opts in powerset(possible_malloc_conf_opts):
+                if cc is 'clang' \
+                  and '-m32' in possible_compiler_opts \
+                  and '--enable-prof' in config_opts:
+                    continue
+                config_line = (
+                    'EXTRA_CFLAGS=-Werror EXTRA_CXXFLAGS=-Werror '
+                    + 'CC="{} {}" '.format(cc, " ".join(compiler_opts))
+                    + 'CXX="{} {}" '.format(cxx, " ".join(compiler_opts))
+                    + '../../configure '
+                    + " ".join(config_opts) + (' --with-malloc-conf=' +
+                    ",".join(malloc_conf_opts) if len(malloc_conf_opts) > 0
+                    else '')
+                )
+
+                # We don't want to test large vaddr spaces in 32-bit mode.
+		if ('-m32' in compiler_opts and '--with-lg-vaddr=56' in
+                  config_opts):
+		    continue
+
+                # Per CPU arenas are only supported on Linux.
+                linux_supported = ('percpu_arena:percpu' in malloc_conf_opts \
+                  or 'background_thread:true' in malloc_conf_opts)
+                # Heap profiling and dss are not supported on OS X.
+                darwin_unsupported = ('--enable-prof' in config_opts or \
+                  'dss:primary' in malloc_conf_opts)
+                if (uname == 'Linux' and linux_supported) \
+                  or (not linux_supported and (uname != 'Darwin' or \
+                  not darwin_unsupported)):
+                    print """cat <<EOF > run_test_%(ind)d.sh
+#!/bin/sh
+
+set -e
+
+abort() {
+    echo "==> Error" >> run_test.log
+    echo "Error; see run_tests.out/run_test_%(ind)d.out/run_test.log"
+    exit 255 # Special exit code tells xargs to terminate.
+}
+
+# Environment variables are not supported.
+run_cmd() {
+    echo "==> \$@" >> run_test.log
+    \$@ >> run_test.log 2>&1 || abort
+}
+
+echo "=> run_test_%(ind)d: %(config_line)s"
+mkdir run_test_%(ind)d.out
+cd run_test_%(ind)d.out
+
+echo "==> %(config_line)s" >> run_test.log
+%(config_line)s >> run_test.log 2>&1 || abort
+
+run_cmd %(make_cmd)s all tests
+run_cmd %(make_cmd)s check
+run_cmd %(make_cmd)s distclean
+EOF
+chmod 755 run_test_%(ind)d.sh""" % {'ind': ind, 'config_line': config_line, 'make_cmd': make_cmd}
+                    ind += 1
+
+print 'for i in `seq 0 %(last_ind)d` ; do echo run_test_${i}.sh ; done | xargs -P %(nparallel)d -n 1 sh' % {'last_ind': ind-1, 'nparallel': nparallel}
diff --git a/memkind-1.10.0/jemalloc/scripts/gen_travis.py b/memkind-1.10.0/jemalloc/scripts/gen_travis.py
new file mode 100755
index 0000000..f1478c6
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/scripts/gen_travis.py
@@ -0,0 +1,149 @@
+#!/usr/bin/env python
+
+from itertools import combinations
+
+travis_template = """\
+language: generic
+dist: precise
+
+matrix:
+  include:
+%s
+
+before_script:
+  - autoconf
+  - scripts/gen_travis.py > travis_script && diff .travis.yml travis_script
+  - ./configure ${COMPILER_FLAGS:+ \
+      CC="$CC $COMPILER_FLAGS" \
+      CXX="$CXX $COMPILER_FLAGS" } \
+      $CONFIGURE_FLAGS
+  - make -j3
+  - make -j3 tests
+
+script:
+  - make check
+"""
+
+# The 'default' configuration is gcc, on linux, with no compiler or configure
+# flags.  We also test with clang, -m32, --enable-debug, --enable-prof,
+# --disable-stats, and --with-malloc-conf=tcache:false.  To avoid abusing
+# travis though, we don't test all 2**7 = 128 possible combinations of these;
+# instead, we only test combinations of up to 2 'unusual' settings, under the
+# hope that bugs involving interactions of such settings are rare.
+# Things at once, for C(7, 0) + C(7, 1) + C(7, 2) = 29
+MAX_UNUSUAL_OPTIONS = 2
+
+os_default = 'linux'
+os_unusual = 'osx'
+
+compilers_default = 'CC=gcc CXX=g++'
+compilers_unusual = 'CC=clang CXX=clang++'
+
+compiler_flag_unusuals = ['-m32']
+
+configure_flag_unusuals = [
+    '--enable-debug',
+    '--enable-prof',
+    '--disable-stats',
+    '--disable-libdl',
+    '--enable-opt-safety-checks',
+]
+
+malloc_conf_unusuals = [
+    'tcache:false',
+    'dss:primary',
+    'percpu_arena:percpu',
+    'background_thread:true',
+]
+
+all_unusuals = (
+    [os_unusual] + [compilers_unusual] + compiler_flag_unusuals
+    + configure_flag_unusuals + malloc_conf_unusuals
+)
+
+unusual_combinations_to_test = []
+for i in xrange(MAX_UNUSUAL_OPTIONS + 1):
+    unusual_combinations_to_test += combinations(all_unusuals, i)
+
+gcc_multilib_set = False
+# Formats a job from a combination of flags
+def format_job(combination):
+    global gcc_multilib_set
+
+    os = os_unusual if os_unusual in combination else os_default
+    compilers = compilers_unusual if compilers_unusual in combination else compilers_default
+
+    compiler_flags = [x for x in combination if x in compiler_flag_unusuals]
+    configure_flags = [x for x in combination if x in configure_flag_unusuals]
+    malloc_conf = [x for x in combination if x in malloc_conf_unusuals]
+
+    # Filter out unsupported configurations on OS X.
+    if os == 'osx' and ('dss:primary' in malloc_conf or \
+      'percpu_arena:percpu' in malloc_conf or 'background_thread:true' \
+      in malloc_conf):
+        return ""
+    if len(malloc_conf) > 0:
+        configure_flags.append('--with-malloc-conf=' + ",".join(malloc_conf))
+
+    # Filter out an unsupported configuration - heap profiling on OS X.
+    if os == 'osx' and '--enable-prof' in configure_flags:
+        return ""
+
+    # We get some spurious errors when -Warray-bounds is enabled.
+    env_string = ('{} COMPILER_FLAGS="{}" CONFIGURE_FLAGS="{}" '
+	'EXTRA_CFLAGS="-Werror -Wno-array-bounds"').format(
+        compilers, " ".join(compiler_flags), " ".join(configure_flags))
+
+    job = ""
+    job += '    - os: %s\n' % os
+    job += '      env: %s\n' % env_string
+    if '-m32' in combination and os == 'linux':
+        job += '      addons:'
+        if gcc_multilib_set:
+            job += ' *gcc_multilib\n'
+        else:
+            job += ' &gcc_multilib\n'
+            job += '        apt:\n'
+            job += '          packages:\n'
+            job += '            - gcc-multilib\n'
+            gcc_multilib_set = True
+    return job
+
+include_rows = ""
+for combination in unusual_combinations_to_test:
+    include_rows += format_job(combination)
+
+# Development build
+include_rows += '''\
+    # Development build
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-cache-oblivious --enable-stats --enable-log --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+'''
+
+# Enable-expermental-smallocx
+include_rows += '''\
+    # --enable-expermental-smallocx:
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-experimental-smallocx --enable-stats --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+'''
+
+# Valgrind build bots
+include_rows += '''
+    # Valgrind
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds" JEMALLOC_TEST_PREFIX="valgrind"
+      addons:
+        apt:
+          packages:
+            - valgrind
+'''
+
+# To enable valgrind on macosx add:
+#
+#  - os: osx
+#    env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds" JEMALLOC_TEST_PREFIX="valgrind"
+#    install: brew install valgrind
+#
+# It currently fails due to: https://github.com/jemalloc/jemalloc/issues/1274
+
+print travis_template % include_rows
diff --git a/memkind-1.10.0/jemalloc/src/arena.c b/memkind-1.10.0/jemalloc/src/arena.c
new file mode 100644
index 0000000..ba50e41
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/arena.c
@@ -0,0 +1,2298 @@
+#define JEMALLOC_ARENA_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/div.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/util.h"
+
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+/******************************************************************************/
+/* Data. */
+
+/*
+ * Define names for both unininitialized and initialized phases, so that
+ * options and mallctl processing are straightforward.
+ */
+const char *percpu_arena_mode_names[] = {
+	"percpu",
+	"phycpu",
+	"disabled",
+	"percpu",
+	"phycpu"
+};
+percpu_arena_mode_t opt_percpu_arena = PERCPU_ARENA_DEFAULT;
+
+ssize_t opt_dirty_decay_ms = DIRTY_DECAY_MS_DEFAULT;
+ssize_t opt_muzzy_decay_ms = MUZZY_DECAY_MS_DEFAULT;
+
+static atomic_zd_t dirty_decay_ms_default;
+static atomic_zd_t muzzy_decay_ms_default;
+
+const uint64_t h_steps[SMOOTHSTEP_NSTEPS] = {
+#define STEP(step, h, x, y)			\
+		h,
+		SMOOTHSTEP
+#undef STEP
+};
+
+static div_info_t arena_binind_div_info[SC_NBINS];
+
+size_t opt_oversize_threshold = OVERSIZE_THRESHOLD_DEFAULT;
+size_t oversize_threshold = OVERSIZE_THRESHOLD_DEFAULT;
+static unsigned huge_arena_ind;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static void arena_decay_to_limit(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, extents_t *extents, bool all, size_t npages_limit,
+    size_t npages_decay_max, bool is_background_thread);
+static bool arena_decay_dirty(tsdn_t *tsdn, arena_t *arena,
+    bool is_background_thread, bool all);
+static void arena_dalloc_bin_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin);
+static void arena_bin_lower_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin);
+
+/******************************************************************************/
+
+void
+arena_basic_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
+    const char **dss, ssize_t *dirty_decay_ms, ssize_t *muzzy_decay_ms,
+    size_t *nactive, size_t *ndirty, size_t *nmuzzy) {
+	*nthreads += arena_nthreads_get(arena, false);
+	*dss = dss_prec_names[arena_dss_prec_get(arena)];
+	*dirty_decay_ms = arena_dirty_decay_ms_get(arena);
+	*muzzy_decay_ms = arena_muzzy_decay_ms_get(arena);
+	*nactive += atomic_load_zu(&arena->nactive, ATOMIC_RELAXED);
+	*ndirty += extents_npages_get(&arena->extents_dirty);
+	*nmuzzy += extents_npages_get(&arena->extents_muzzy);
+}
+
+void
+arena_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
+    const char **dss, ssize_t *dirty_decay_ms, ssize_t *muzzy_decay_ms,
+    size_t *nactive, size_t *ndirty, size_t *nmuzzy, arena_stats_t *astats,
+    bin_stats_t *bstats, arena_stats_large_t *lstats,
+    arena_stats_extents_t *estats) {
+	cassert(config_stats);
+
+	arena_basic_stats_merge(tsdn, arena, nthreads, dss, dirty_decay_ms,
+	    muzzy_decay_ms, nactive, ndirty, nmuzzy);
+
+	size_t base_allocated, base_resident, base_mapped, metadata_thp;
+	base_stats_get(tsdn, arena->base, &base_allocated, &base_resident,
+	    &base_mapped, &metadata_thp);
+
+	arena_stats_lock(tsdn, &arena->stats);
+
+	arena_stats_accum_zu(&astats->mapped, base_mapped
+	    + arena_stats_read_zu(tsdn, &arena->stats, &arena->stats.mapped));
+	arena_stats_accum_zu(&astats->retained,
+	    extents_npages_get(&arena->extents_retained) << LG_PAGE);
+
+	atomic_store_zu(&astats->extent_avail,
+	    atomic_load_zu(&arena->extent_avail_cnt, ATOMIC_RELAXED),
+	    ATOMIC_RELAXED);
+
+	arena_stats_accum_u64(&astats->decay_dirty.npurge,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_dirty.npurge));
+	arena_stats_accum_u64(&astats->decay_dirty.nmadvise,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_dirty.nmadvise));
+	arena_stats_accum_u64(&astats->decay_dirty.purged,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_dirty.purged));
+
+	arena_stats_accum_u64(&astats->decay_muzzy.npurge,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_muzzy.npurge));
+	arena_stats_accum_u64(&astats->decay_muzzy.nmadvise,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_muzzy.nmadvise));
+	arena_stats_accum_u64(&astats->decay_muzzy.purged,
+	    arena_stats_read_u64(tsdn, &arena->stats,
+	    &arena->stats.decay_muzzy.purged));
+
+	arena_stats_accum_zu(&astats->base, base_allocated);
+	arena_stats_accum_zu(&astats->internal, arena_internal_get(arena));
+	arena_stats_accum_zu(&astats->metadata_thp, metadata_thp);
+	arena_stats_accum_zu(&astats->resident, base_resident +
+	    (((atomic_load_zu(&arena->nactive, ATOMIC_RELAXED) +
+	    extents_npages_get(&arena->extents_dirty) +
+	    extents_npages_get(&arena->extents_muzzy)) << LG_PAGE)));
+	arena_stats_accum_zu(&astats->abandoned_vm, atomic_load_zu(
+	    &arena->stats.abandoned_vm, ATOMIC_RELAXED));
+
+	for (szind_t i = 0; i < SC_NSIZES - SC_NBINS; i++) {
+		uint64_t nmalloc = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].nmalloc);
+		arena_stats_accum_u64(&lstats[i].nmalloc, nmalloc);
+		arena_stats_accum_u64(&astats->nmalloc_large, nmalloc);
+
+		uint64_t ndalloc = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].ndalloc);
+		arena_stats_accum_u64(&lstats[i].ndalloc, ndalloc);
+		arena_stats_accum_u64(&astats->ndalloc_large, ndalloc);
+
+		uint64_t nrequests = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].nrequests);
+		arena_stats_accum_u64(&lstats[i].nrequests,
+		    nmalloc + nrequests);
+		arena_stats_accum_u64(&astats->nrequests_large,
+		    nmalloc + nrequests);
+
+		/* nfill == nmalloc for large currently. */
+		arena_stats_accum_u64(&lstats[i].nfills, nmalloc);
+		arena_stats_accum_u64(&astats->nfills_large, nmalloc);
+
+		uint64_t nflush = arena_stats_read_u64(tsdn, &arena->stats,
+		    &arena->stats.lstats[i].nflushes);
+		arena_stats_accum_u64(&lstats[i].nflushes, nflush);
+		arena_stats_accum_u64(&astats->nflushes_large, nflush);
+
+		assert(nmalloc >= ndalloc);
+		assert(nmalloc - ndalloc <= SIZE_T_MAX);
+		size_t curlextents = (size_t)(nmalloc - ndalloc);
+		lstats[i].curlextents += curlextents;
+		arena_stats_accum_zu(&astats->allocated_large,
+		    curlextents * sz_index2size(SC_NBINS + i));
+	}
+
+	for (pszind_t i = 0; i < SC_NPSIZES; i++) {
+		size_t dirty, muzzy, retained, dirty_bytes, muzzy_bytes,
+		    retained_bytes;
+		dirty = extents_nextents_get(&arena->extents_dirty, i);
+		muzzy = extents_nextents_get(&arena->extents_muzzy, i);
+		retained = extents_nextents_get(&arena->extents_retained, i);
+		dirty_bytes = extents_nbytes_get(&arena->extents_dirty, i);
+		muzzy_bytes = extents_nbytes_get(&arena->extents_muzzy, i);
+		retained_bytes =
+		    extents_nbytes_get(&arena->extents_retained, i);
+
+		atomic_store_zu(&estats[i].ndirty, dirty, ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].nmuzzy, muzzy, ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].nretained, retained, ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].dirty_bytes, dirty_bytes,
+		    ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].muzzy_bytes, muzzy_bytes,
+		    ATOMIC_RELAXED);
+		atomic_store_zu(&estats[i].retained_bytes, retained_bytes,
+		    ATOMIC_RELAXED);
+	}
+
+	arena_stats_unlock(tsdn, &arena->stats);
+
+	/* tcache_bytes counts currently cached bytes. */
+	atomic_store_zu(&astats->tcache_bytes, 0, ATOMIC_RELAXED);
+	malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+	cache_bin_array_descriptor_t *descriptor;
+	ql_foreach(descriptor, &arena->cache_bin_array_descriptor_ql, link) {
+		szind_t i = 0;
+		for (; i < SC_NBINS; i++) {
+			cache_bin_t *tbin = &descriptor->bins_small[i];
+			arena_stats_accum_zu(&astats->tcache_bytes,
+			    tbin->ncached * sz_index2size(i));
+		}
+		for (; i < nhbins; i++) {
+			cache_bin_t *tbin = &descriptor->bins_large[i];
+			arena_stats_accum_zu(&astats->tcache_bytes,
+			    tbin->ncached * sz_index2size(i));
+		}
+	}
+	malloc_mutex_prof_read(tsdn,
+	    &astats->mutex_prof_data[arena_prof_mutex_tcache_list],
+	    &arena->tcache_ql_mtx);
+	malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx);
+
+#define READ_ARENA_MUTEX_PROF_DATA(mtx, ind)				\
+    malloc_mutex_lock(tsdn, &arena->mtx);				\
+    malloc_mutex_prof_read(tsdn, &astats->mutex_prof_data[ind],		\
+        &arena->mtx);							\
+    malloc_mutex_unlock(tsdn, &arena->mtx);
+
+	/* Gather per arena mutex profiling data. */
+	READ_ARENA_MUTEX_PROF_DATA(large_mtx, arena_prof_mutex_large);
+	READ_ARENA_MUTEX_PROF_DATA(extent_avail_mtx,
+	    arena_prof_mutex_extent_avail)
+	READ_ARENA_MUTEX_PROF_DATA(extents_dirty.mtx,
+	    arena_prof_mutex_extents_dirty)
+	READ_ARENA_MUTEX_PROF_DATA(extents_muzzy.mtx,
+	    arena_prof_mutex_extents_muzzy)
+	READ_ARENA_MUTEX_PROF_DATA(extents_retained.mtx,
+	    arena_prof_mutex_extents_retained)
+	READ_ARENA_MUTEX_PROF_DATA(decay_dirty.mtx,
+	    arena_prof_mutex_decay_dirty)
+	READ_ARENA_MUTEX_PROF_DATA(decay_muzzy.mtx,
+	    arena_prof_mutex_decay_muzzy)
+	READ_ARENA_MUTEX_PROF_DATA(base->mtx,
+	    arena_prof_mutex_base)
+#undef READ_ARENA_MUTEX_PROF_DATA
+
+	nstime_copy(&astats->uptime, &arena->create_time);
+	nstime_update(&astats->uptime);
+	nstime_subtract(&astats->uptime, &arena->create_time);
+
+	for (szind_t i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_stats_merge(tsdn, &bstats[i],
+			    &arena->bins[i].bin_shards[j]);
+		}
+	}
+}
+
+void
+arena_extents_dirty_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extents_dalloc(tsdn, arena, r_extent_hooks, &arena->extents_dirty,
+	    extent);
+	if (arena_dirty_decay_ms_get(arena) == 0) {
+		arena_decay_dirty(tsdn, arena, false, true);
+	} else {
+		arena_background_thread_inactivity_check(tsdn, arena, false);
+	}
+}
+
+static void *
+arena_slab_reg_alloc(extent_t *slab, const bin_info_t *bin_info) {
+	void *ret;
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+	size_t regind;
+
+	assert(extent_nfree_get(slab) > 0);
+	assert(!bitmap_full(slab_data->bitmap, &bin_info->bitmap_info));
+
+	regind = bitmap_sfu(slab_data->bitmap, &bin_info->bitmap_info);
+	ret = (void *)((uintptr_t)extent_addr_get(slab) +
+	    (uintptr_t)(bin_info->reg_size * regind));
+	extent_nfree_dec(slab);
+	return ret;
+}
+
+static void
+arena_slab_reg_alloc_batch(extent_t *slab, const bin_info_t *bin_info,
+			   unsigned cnt, void** ptrs) {
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+
+	assert(extent_nfree_get(slab) >= cnt);
+	assert(!bitmap_full(slab_data->bitmap, &bin_info->bitmap_info));
+
+#if (! defined JEMALLOC_INTERNAL_POPCOUNTL) || (defined BITMAP_USE_TREE)
+	for (unsigned i = 0; i < cnt; i++) {
+		size_t regind = bitmap_sfu(slab_data->bitmap,
+					   &bin_info->bitmap_info);
+		*(ptrs + i) = (void *)((uintptr_t)extent_addr_get(slab) +
+		    (uintptr_t)(bin_info->reg_size * regind));
+	}
+#else
+	unsigned group = 0;
+	bitmap_t g = slab_data->bitmap[group];
+	unsigned i = 0;
+	while (i < cnt) {
+		while (g == 0) {
+			g = slab_data->bitmap[++group];
+		}
+		size_t shift = group << LG_BITMAP_GROUP_NBITS;
+		size_t pop = popcount_lu(g);
+		if (pop > (cnt - i)) {
+			pop = cnt - i;
+		}
+
+		/*
+		 * Load from memory locations only once, outside the
+		 * hot loop below.
+		 */
+		uintptr_t base = (uintptr_t)extent_addr_get(slab);
+		uintptr_t regsize = (uintptr_t)bin_info->reg_size;
+		while (pop--) {
+			size_t bit = cfs_lu(&g);
+			size_t regind = shift + bit;
+			*(ptrs + i) = (void *)(base + regsize * regind);
+
+			i++;
+		}
+		slab_data->bitmap[group] = g;
+	}
+#endif
+	extent_nfree_sub(slab, cnt);
+}
+
+#ifndef JEMALLOC_JET
+static
+#endif
+size_t
+arena_slab_regind(extent_t *slab, szind_t binind, const void *ptr) {
+	size_t diff, regind;
+
+	/* Freeing a pointer outside the slab can cause assertion failure. */
+	assert((uintptr_t)ptr >= (uintptr_t)extent_addr_get(slab));
+	assert((uintptr_t)ptr < (uintptr_t)extent_past_get(slab));
+	/* Freeing an interior pointer can cause assertion failure. */
+	assert(((uintptr_t)ptr - (uintptr_t)extent_addr_get(slab)) %
+	    (uintptr_t)bin_infos[binind].reg_size == 0);
+
+	diff = (size_t)((uintptr_t)ptr - (uintptr_t)extent_addr_get(slab));
+
+	/* Avoid doing division with a variable divisor. */
+	regind = div_compute(&arena_binind_div_info[binind], diff);
+
+	assert(regind < bin_infos[binind].nregs);
+
+	return regind;
+}
+
+static void
+arena_slab_reg_dalloc(extent_t *slab, arena_slab_data_t *slab_data, void *ptr) {
+	szind_t binind = extent_szind_get(slab);
+	const bin_info_t *bin_info = &bin_infos[binind];
+	size_t regind = arena_slab_regind(slab, binind, ptr);
+
+	assert(extent_nfree_get(slab) < bin_info->nregs);
+	/* Freeing an unallocated pointer can cause assertion failure. */
+	assert(bitmap_get(slab_data->bitmap, &bin_info->bitmap_info, regind));
+
+	bitmap_unset(slab_data->bitmap, &bin_info->bitmap_info, regind);
+	extent_nfree_inc(slab);
+}
+
+static void
+arena_nactive_add(arena_t *arena, size_t add_pages) {
+	atomic_fetch_add_zu(&arena->nactive, add_pages, ATOMIC_RELAXED);
+}
+
+static void
+arena_nactive_sub(arena_t *arena, size_t sub_pages) {
+	assert(atomic_load_zu(&arena->nactive, ATOMIC_RELAXED) >= sub_pages);
+	atomic_fetch_sub_zu(&arena->nactive, sub_pages, ATOMIC_RELAXED);
+}
+
+static void
+arena_large_malloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t usize) {
+	szind_t index, hindex;
+
+	cassert(config_stats);
+
+	if (usize < SC_LARGE_MINCLASS) {
+		usize = SC_LARGE_MINCLASS;
+	}
+	index = sz_size2index(usize);
+	hindex = (index >= SC_NBINS) ? index - SC_NBINS : 0;
+
+	arena_stats_add_u64(tsdn, &arena->stats,
+	    &arena->stats.lstats[hindex].nmalloc, 1);
+}
+
+static void
+arena_large_dalloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t usize) {
+	szind_t index, hindex;
+
+	cassert(config_stats);
+
+	if (usize < SC_LARGE_MINCLASS) {
+		usize = SC_LARGE_MINCLASS;
+	}
+	index = sz_size2index(usize);
+	hindex = (index >= SC_NBINS) ? index - SC_NBINS : 0;
+
+	arena_stats_add_u64(tsdn, &arena->stats,
+	    &arena->stats.lstats[hindex].ndalloc, 1);
+}
+
+static void
+arena_large_ralloc_stats_update(tsdn_t *tsdn, arena_t *arena, size_t oldusize,
+    size_t usize) {
+	arena_large_dalloc_stats_update(tsdn, arena, oldusize);
+	arena_large_malloc_stats_update(tsdn, arena, usize);
+}
+
+static bool
+arena_may_have_muzzy(arena_t *arena) {
+	return (pages_can_purge_lazy && (arena_muzzy_decay_ms_get(arena) != 0));
+}
+
+extent_t *
+arena_extent_alloc_large(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool *zero) {
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	szind_t szind = sz_size2index(usize);
+	size_t mapped_add;
+	bool commit = true;
+	extent_t *extent = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_dirty, NULL, usize, sz_large_pad, alignment, false,
+	    szind, zero, &commit);
+	if (extent == NULL && arena_may_have_muzzy(arena)) {
+		extent = extents_alloc(tsdn, arena, &extent_hooks,
+		    &arena->extents_muzzy, NULL, usize, sz_large_pad, alignment,
+		    false, szind, zero, &commit);
+	}
+	size_t size = usize + sz_large_pad;
+	if (extent == NULL) {
+		extent = extent_alloc_wrapper(tsdn, arena, &extent_hooks, NULL,
+		    usize, sz_large_pad, alignment, false, szind, zero,
+		    &commit);
+		if (config_stats) {
+			/*
+			 * extent may be NULL on OOM, but in that case
+			 * mapped_add isn't used below, so there's no need to
+			 * conditionlly set it to 0 here.
+			 */
+			mapped_add = size;
+		}
+	} else if (config_stats) {
+		mapped_add = 0;
+	}
+
+	if (extent != NULL) {
+		if (config_stats) {
+			arena_stats_lock(tsdn, &arena->stats);
+			arena_large_malloc_stats_update(tsdn, arena, usize);
+			if (mapped_add != 0) {
+				arena_stats_add_zu(tsdn, &arena->stats,
+				    &arena->stats.mapped, mapped_add);
+			}
+			arena_stats_unlock(tsdn, &arena->stats);
+		}
+		arena_nactive_add(arena, size >> LG_PAGE);
+	}
+
+	return extent;
+}
+
+void
+arena_extent_dalloc_large_prep(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_large_dalloc_stats_update(tsdn, arena,
+		    extent_usize_get(extent));
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+	arena_nactive_sub(arena, extent_size_get(extent) >> LG_PAGE);
+}
+
+void
+arena_extent_ralloc_large_shrink(tsdn_t *tsdn, arena_t *arena, extent_t *extent,
+    size_t oldusize) {
+	size_t usize = extent_usize_get(extent);
+	size_t udiff = oldusize - usize;
+
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_large_ralloc_stats_update(tsdn, arena, oldusize, usize);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+	arena_nactive_sub(arena, udiff >> LG_PAGE);
+}
+
+void
+arena_extent_ralloc_large_expand(tsdn_t *tsdn, arena_t *arena, extent_t *extent,
+    size_t oldusize) {
+	size_t usize = extent_usize_get(extent);
+	size_t udiff = usize - oldusize;
+
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_large_ralloc_stats_update(tsdn, arena, oldusize, usize);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+	arena_nactive_add(arena, udiff >> LG_PAGE);
+}
+
+static ssize_t
+arena_decay_ms_read(arena_decay_t *decay) {
+	return atomic_load_zd(&decay->time_ms, ATOMIC_RELAXED);
+}
+
+static void
+arena_decay_ms_write(arena_decay_t *decay, ssize_t decay_ms) {
+	atomic_store_zd(&decay->time_ms, decay_ms, ATOMIC_RELAXED);
+}
+
+static void
+arena_decay_deadline_init(arena_decay_t *decay) {
+	/*
+	 * Generate a new deadline that is uniformly random within the next
+	 * epoch after the current one.
+	 */
+	nstime_copy(&decay->deadline, &decay->epoch);
+	nstime_add(&decay->deadline, &decay->interval);
+	if (arena_decay_ms_read(decay) > 0) {
+		nstime_t jitter;
+
+		nstime_init(&jitter, prng_range_u64(&decay->jitter_state,
+		    nstime_ns(&decay->interval)));
+		nstime_add(&decay->deadline, &jitter);
+	}
+}
+
+static bool
+arena_decay_deadline_reached(const arena_decay_t *decay, const nstime_t *time) {
+	return (nstime_compare(&decay->deadline, time) <= 0);
+}
+
+static size_t
+arena_decay_backlog_npages_limit(const arena_decay_t *decay) {
+	uint64_t sum;
+	size_t npages_limit_backlog;
+	unsigned i;
+
+	/*
+	 * For each element of decay_backlog, multiply by the corresponding
+	 * fixed-point smoothstep decay factor.  Sum the products, then divide
+	 * to round down to the nearest whole number of pages.
+	 */
+	sum = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+		sum += decay->backlog[i] * h_steps[i];
+	}
+	npages_limit_backlog = (size_t)(sum >> SMOOTHSTEP_BFP);
+
+	return npages_limit_backlog;
+}
+
+static void
+arena_decay_backlog_update_last(arena_decay_t *decay, size_t current_npages) {
+	size_t npages_delta = (current_npages > decay->nunpurged) ?
+	    current_npages - decay->nunpurged : 0;
+	decay->backlog[SMOOTHSTEP_NSTEPS-1] = npages_delta;
+
+	if (config_debug) {
+		if (current_npages > decay->ceil_npages) {
+			decay->ceil_npages = current_npages;
+		}
+		size_t npages_limit = arena_decay_backlog_npages_limit(decay);
+		assert(decay->ceil_npages >= npages_limit);
+		if (decay->ceil_npages > npages_limit) {
+			decay->ceil_npages = npages_limit;
+		}
+	}
+}
+
+static void
+arena_decay_backlog_update(arena_decay_t *decay, uint64_t nadvance_u64,
+    size_t current_npages) {
+	if (nadvance_u64 >= SMOOTHSTEP_NSTEPS) {
+		memset(decay->backlog, 0, (SMOOTHSTEP_NSTEPS-1) *
+		    sizeof(size_t));
+	} else {
+		size_t nadvance_z = (size_t)nadvance_u64;
+
+		assert((uint64_t)nadvance_z == nadvance_u64);
+
+		memmove(decay->backlog, &decay->backlog[nadvance_z],
+		    (SMOOTHSTEP_NSTEPS - nadvance_z) * sizeof(size_t));
+		if (nadvance_z > 1) {
+			memset(&decay->backlog[SMOOTHSTEP_NSTEPS -
+			    nadvance_z], 0, (nadvance_z-1) * sizeof(size_t));
+		}
+	}
+
+	arena_decay_backlog_update_last(decay, current_npages);
+}
+
+static void
+arena_decay_try_purge(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, size_t current_npages, size_t npages_limit,
+    bool is_background_thread) {
+	if (current_npages > npages_limit) {
+		arena_decay_to_limit(tsdn, arena, decay, extents, false,
+		    npages_limit, current_npages - npages_limit,
+		    is_background_thread);
+	}
+}
+
+static void
+arena_decay_epoch_advance_helper(arena_decay_t *decay, const nstime_t *time,
+    size_t current_npages) {
+	assert(arena_decay_deadline_reached(decay, time));
+
+	nstime_t delta;
+	nstime_copy(&delta, time);
+	nstime_subtract(&delta, &decay->epoch);
+
+	uint64_t nadvance_u64 = nstime_divide(&delta, &decay->interval);
+	assert(nadvance_u64 > 0);
+
+	/* Add nadvance_u64 decay intervals to epoch. */
+	nstime_copy(&delta, &decay->interval);
+	nstime_imultiply(&delta, nadvance_u64);
+	nstime_add(&decay->epoch, &delta);
+
+	/* Set a new deadline. */
+	arena_decay_deadline_init(decay);
+
+	/* Update the backlog. */
+	arena_decay_backlog_update(decay, nadvance_u64, current_npages);
+}
+
+static void
+arena_decay_epoch_advance(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, const nstime_t *time, bool is_background_thread) {
+	size_t current_npages = extents_npages_get(extents);
+	arena_decay_epoch_advance_helper(decay, time, current_npages);
+
+	size_t npages_limit = arena_decay_backlog_npages_limit(decay);
+	/* We may unlock decay->mtx when try_purge(). Finish logging first. */
+	decay->nunpurged = (npages_limit > current_npages) ? npages_limit :
+	    current_npages;
+
+	if (!background_thread_enabled() || is_background_thread) {
+		arena_decay_try_purge(tsdn, arena, decay, extents,
+		    current_npages, npages_limit, is_background_thread);
+	}
+}
+
+static void
+arena_decay_reinit(arena_decay_t *decay, ssize_t decay_ms) {
+	arena_decay_ms_write(decay, decay_ms);
+	if (decay_ms > 0) {
+		nstime_init(&decay->interval, (uint64_t)decay_ms *
+		    KQU(1000000));
+		nstime_idivide(&decay->interval, SMOOTHSTEP_NSTEPS);
+	}
+
+	nstime_init(&decay->epoch, 0);
+	nstime_update(&decay->epoch);
+	decay->jitter_state = (uint64_t)(uintptr_t)decay;
+	arena_decay_deadline_init(decay);
+	decay->nunpurged = 0;
+	memset(decay->backlog, 0, SMOOTHSTEP_NSTEPS * sizeof(size_t));
+}
+
+static bool
+arena_decay_init(arena_decay_t *decay, ssize_t decay_ms,
+    arena_stats_decay_t *stats) {
+	if (config_debug) {
+		for (size_t i = 0; i < sizeof(arena_decay_t); i++) {
+			assert(((char *)decay)[i] == 0);
+		}
+		decay->ceil_npages = 0;
+	}
+	if (malloc_mutex_init(&decay->mtx, "decay", WITNESS_RANK_DECAY,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	decay->purging = false;
+	arena_decay_reinit(decay, decay_ms);
+	/* Memory is zeroed, so there is no need to clear stats. */
+	if (config_stats) {
+		decay->stats = stats;
+	}
+	return false;
+}
+
+static bool
+arena_decay_ms_valid(ssize_t decay_ms) {
+	if (decay_ms < -1) {
+		return false;
+	}
+	if (decay_ms == -1 || (uint64_t)decay_ms <= NSTIME_SEC_MAX *
+	    KQU(1000)) {
+		return true;
+	}
+	return false;
+}
+
+static bool
+arena_maybe_decay(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, bool is_background_thread) {
+	malloc_mutex_assert_owner(tsdn, &decay->mtx);
+
+	/* Purge all or nothing if the option is disabled. */
+	ssize_t decay_ms = arena_decay_ms_read(decay);
+	if (decay_ms <= 0) {
+		if (decay_ms == 0) {
+			arena_decay_to_limit(tsdn, arena, decay, extents, false,
+			    0, extents_npages_get(extents),
+			    is_background_thread);
+		}
+		return false;
+	}
+
+	nstime_t time;
+	nstime_init(&time, 0);
+	nstime_update(&time);
+	if (unlikely(!nstime_monotonic() && nstime_compare(&decay->epoch, &time)
+	    > 0)) {
+		/*
+		 * Time went backwards.  Move the epoch back in time and
+		 * generate a new deadline, with the expectation that time
+		 * typically flows forward for long enough periods of time that
+		 * epochs complete.  Unfortunately, this strategy is susceptible
+		 * to clock jitter triggering premature epoch advances, but
+		 * clock jitter estimation and compensation isn't feasible here
+		 * because calls into this code are event-driven.
+		 */
+		nstime_copy(&decay->epoch, &time);
+		arena_decay_deadline_init(decay);
+	} else {
+		/* Verify that time does not go backwards. */
+		assert(nstime_compare(&decay->epoch, &time) <= 0);
+	}
+
+	/*
+	 * If the deadline has been reached, advance to the current epoch and
+	 * purge to the new limit if necessary.  Note that dirty pages created
+	 * during the current epoch are not subject to purge until a future
+	 * epoch, so as a result purging only happens during epoch advances, or
+	 * being triggered by background threads (scheduled event).
+	 */
+	bool advance_epoch = arena_decay_deadline_reached(decay, &time);
+	if (advance_epoch) {
+		arena_decay_epoch_advance(tsdn, arena, decay, extents, &time,
+		    is_background_thread);
+	} else if (is_background_thread) {
+		arena_decay_try_purge(tsdn, arena, decay, extents,
+		    extents_npages_get(extents),
+		    arena_decay_backlog_npages_limit(decay),
+		    is_background_thread);
+	}
+
+	return advance_epoch;
+}
+
+static ssize_t
+arena_decay_ms_get(arena_decay_t *decay) {
+	return arena_decay_ms_read(decay);
+}
+
+ssize_t
+arena_dirty_decay_ms_get(arena_t *arena) {
+	return arena_decay_ms_get(&arena->decay_dirty);
+}
+
+ssize_t
+arena_muzzy_decay_ms_get(arena_t *arena) {
+	return arena_decay_ms_get(&arena->decay_muzzy);
+}
+
+static bool
+arena_decay_ms_set(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, ssize_t decay_ms) {
+	if (!arena_decay_ms_valid(decay_ms)) {
+		return true;
+	}
+
+	malloc_mutex_lock(tsdn, &decay->mtx);
+	/*
+	 * Restart decay backlog from scratch, which may cause many dirty pages
+	 * to be immediately purged.  It would conceptually be possible to map
+	 * the old backlog onto the new backlog, but there is no justification
+	 * for such complexity since decay_ms changes are intended to be
+	 * infrequent, either between the {-1, 0, >0} states, or a one-time
+	 * arbitrary change during initial arena configuration.
+	 */
+	arena_decay_reinit(decay, decay_ms);
+	arena_maybe_decay(tsdn, arena, decay, extents, false);
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	return false;
+}
+
+bool
+arena_dirty_decay_ms_set(tsdn_t *tsdn, arena_t *arena,
+    ssize_t decay_ms) {
+	return arena_decay_ms_set(tsdn, arena, &arena->decay_dirty,
+	    &arena->extents_dirty, decay_ms);
+}
+
+bool
+arena_muzzy_decay_ms_set(tsdn_t *tsdn, arena_t *arena,
+    ssize_t decay_ms) {
+	return arena_decay_ms_set(tsdn, arena, &arena->decay_muzzy,
+	    &arena->extents_muzzy, decay_ms);
+}
+
+static size_t
+arena_stash_decayed(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, size_t npages_limit,
+	size_t npages_decay_max, extent_list_t *decay_extents) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	/* Stash extents according to npages_limit. */
+	size_t nstashed = 0;
+	extent_t *extent;
+	while (nstashed < npages_decay_max &&
+	    (extent = extents_evict(tsdn, arena, r_extent_hooks, extents,
+	    npages_limit)) != NULL) {
+		extent_list_append(decay_extents, extent);
+		nstashed += extent_size_get(extent) >> LG_PAGE;
+	}
+	return nstashed;
+}
+
+static size_t
+arena_decay_stashed(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, arena_decay_t *decay, extents_t *extents,
+    bool all, extent_list_t *decay_extents, bool is_background_thread) {
+	size_t nmadvise, nunmapped;
+	size_t npurged;
+
+	if (config_stats) {
+		nmadvise = 0;
+		nunmapped = 0;
+	}
+	npurged = 0;
+
+	ssize_t muzzy_decay_ms = arena_muzzy_decay_ms_get(arena);
+	for (extent_t *extent = extent_list_first(decay_extents); extent !=
+	    NULL; extent = extent_list_first(decay_extents)) {
+		if (config_stats) {
+			nmadvise++;
+		}
+		size_t npages = extent_size_get(extent) >> LG_PAGE;
+		npurged += npages;
+		extent_list_remove(decay_extents, extent);
+		switch (extents_state_get(extents)) {
+		case extent_state_active:
+			not_reached();
+		case extent_state_dirty:
+			if (!all && muzzy_decay_ms != 0 &&
+			    !extent_purge_lazy_wrapper(tsdn, arena,
+			    r_extent_hooks, extent, 0,
+			    extent_size_get(extent))) {
+				extents_dalloc(tsdn, arena, r_extent_hooks,
+				    &arena->extents_muzzy, extent);
+				arena_background_thread_inactivity_check(tsdn,
+				    arena, is_background_thread);
+				break;
+			}
+			/* Fall through. */
+		case extent_state_muzzy:
+			extent_dalloc_wrapper(tsdn, arena, r_extent_hooks,
+			    extent);
+			if (config_stats) {
+				nunmapped += npages;
+			}
+			break;
+		case extent_state_retained:
+		default:
+			not_reached();
+		}
+	}
+
+	if (config_stats) {
+		arena_stats_lock(tsdn, &arena->stats);
+		arena_stats_add_u64(tsdn, &arena->stats, &decay->stats->npurge,
+		    1);
+		arena_stats_add_u64(tsdn, &arena->stats,
+		    &decay->stats->nmadvise, nmadvise);
+		arena_stats_add_u64(tsdn, &arena->stats, &decay->stats->purged,
+		    npurged);
+		arena_stats_sub_zu(tsdn, &arena->stats, &arena->stats.mapped,
+		    nunmapped << LG_PAGE);
+		arena_stats_unlock(tsdn, &arena->stats);
+	}
+
+	return npurged;
+}
+
+/*
+ * npages_limit: Decay at most npages_decay_max pages without violating the
+ * invariant: (extents_npages_get(extents) >= npages_limit).  We need an upper
+ * bound on number of pages in order to prevent unbounded growth (namely in
+ * stashed), otherwise unbounded new pages could be added to extents during the
+ * current decay run, so that the purging thread never finishes.
+ */
+static void
+arena_decay_to_limit(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, bool all, size_t npages_limit, size_t npages_decay_max,
+    bool is_background_thread) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 1);
+	malloc_mutex_assert_owner(tsdn, &decay->mtx);
+
+	if (decay->purging) {
+		return;
+	}
+	decay->purging = true;
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+
+	extent_list_t decay_extents;
+	extent_list_init(&decay_extents);
+
+	size_t npurge = arena_stash_decayed(tsdn, arena, &extent_hooks, extents,
+	    npages_limit, npages_decay_max, &decay_extents);
+	if (npurge != 0) {
+		size_t npurged = arena_decay_stashed(tsdn, arena,
+		    &extent_hooks, decay, extents, all, &decay_extents,
+		    is_background_thread);
+		assert(npurged == npurge);
+	}
+
+	malloc_mutex_lock(tsdn, &decay->mtx);
+	decay->purging = false;
+}
+
+static bool
+arena_decay_impl(tsdn_t *tsdn, arena_t *arena, arena_decay_t *decay,
+    extents_t *extents, bool is_background_thread, bool all) {
+	if (all) {
+		malloc_mutex_lock(tsdn, &decay->mtx);
+		arena_decay_to_limit(tsdn, arena, decay, extents, all, 0,
+		    extents_npages_get(extents), is_background_thread);
+		malloc_mutex_unlock(tsdn, &decay->mtx);
+
+		return false;
+	}
+
+	if (malloc_mutex_trylock(tsdn, &decay->mtx)) {
+		/* No need to wait if another thread is in progress. */
+		return true;
+	}
+
+	bool epoch_advanced = arena_maybe_decay(tsdn, arena, decay, extents,
+	    is_background_thread);
+	size_t npages_new;
+	if (epoch_advanced) {
+		/* Backlog is updated on epoch advance. */
+		npages_new = decay->backlog[SMOOTHSTEP_NSTEPS-1];
+	}
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	if (have_background_thread && background_thread_enabled() &&
+	    epoch_advanced && !is_background_thread) {
+		background_thread_interval_check(tsdn, arena, decay,
+		    npages_new);
+	}
+
+	return false;
+}
+
+static bool
+arena_decay_dirty(tsdn_t *tsdn, arena_t *arena, bool is_background_thread,
+    bool all) {
+	return arena_decay_impl(tsdn, arena, &arena->decay_dirty,
+	    &arena->extents_dirty, is_background_thread, all);
+}
+
+static bool
+arena_decay_muzzy(tsdn_t *tsdn, arena_t *arena, bool is_background_thread,
+    bool all) {
+	return arena_decay_impl(tsdn, arena, &arena->decay_muzzy,
+	    &arena->extents_muzzy, is_background_thread, all);
+}
+
+void
+arena_decay(tsdn_t *tsdn, arena_t *arena, bool is_background_thread, bool all) {
+	if (arena_decay_dirty(tsdn, arena, is_background_thread, all)) {
+		return;
+	}
+	arena_decay_muzzy(tsdn, arena, is_background_thread, all);
+}
+
+static void
+arena_slab_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *slab) {
+	arena_nactive_sub(arena, extent_size_get(slab) >> LG_PAGE);
+
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+	arena_extents_dirty_dalloc(tsdn, arena, &extent_hooks, slab);
+}
+
+static void
+arena_bin_slabs_nonfull_insert(bin_t *bin, extent_t *slab) {
+	assert(extent_nfree_get(slab) > 0);
+	extent_heap_insert(&bin->slabs_nonfull, slab);
+	if (config_stats) {
+		bin->stats.nonfull_slabs++;
+	}
+}
+
+static void
+arena_bin_slabs_nonfull_remove(bin_t *bin, extent_t *slab) {
+	extent_heap_remove(&bin->slabs_nonfull, slab);
+	if (config_stats) {
+		bin->stats.nonfull_slabs--;
+	}
+}
+
+static extent_t *
+arena_bin_slabs_nonfull_tryget(bin_t *bin) {
+	extent_t *slab = extent_heap_remove_first(&bin->slabs_nonfull);
+	if (slab == NULL) {
+		return NULL;
+	}
+	if (config_stats) {
+		bin->stats.reslabs++;
+		bin->stats.nonfull_slabs--;
+	}
+	return slab;
+}
+
+static void
+arena_bin_slabs_full_insert(arena_t *arena, bin_t *bin, extent_t *slab) {
+	assert(extent_nfree_get(slab) == 0);
+	/*
+	 *  Tracking extents is required by arena_reset, which is not allowed
+	 *  for auto arenas.  Bypass this step to avoid touching the extent
+	 *  linkage (often results in cache misses) for auto arenas.
+	 */
+	if (arena_is_auto(arena)) {
+		return;
+	}
+	extent_list_append(&bin->slabs_full, slab);
+}
+
+static void
+arena_bin_slabs_full_remove(arena_t *arena, bin_t *bin, extent_t *slab) {
+	if (arena_is_auto(arena)) {
+		return;
+	}
+	extent_list_remove(&bin->slabs_full, slab);
+}
+
+static void
+arena_bin_reset(tsd_t *tsd, arena_t *arena, bin_t *bin) {
+	extent_t *slab;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	if (bin->slabcur != NULL) {
+		slab = bin->slabcur;
+		bin->slabcur = NULL;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_slab_dalloc(tsd_tsdn(tsd), arena, slab);
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	}
+	while ((slab = extent_heap_remove_first(&bin->slabs_nonfull)) != NULL) {
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_slab_dalloc(tsd_tsdn(tsd), arena, slab);
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	}
+	for (slab = extent_list_first(&bin->slabs_full); slab != NULL;
+	     slab = extent_list_first(&bin->slabs_full)) {
+		arena_bin_slabs_full_remove(arena, bin, slab);
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_slab_dalloc(tsd_tsdn(tsd), arena, slab);
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+	}
+	if (config_stats) {
+		bin->stats.curregs = 0;
+		bin->stats.curslabs = 0;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+}
+
+void
+arena_reset(tsd_t *tsd, arena_t *arena) {
+	/*
+	 * Locking in this function is unintuitive.  The caller guarantees that
+	 * no concurrent operations are happening in this arena, but there are
+	 * still reasons that some locking is necessary:
+	 *
+	 * - Some of the functions in the transitive closure of calls assume
+	 *   appropriate locks are held, and in some cases these locks are
+	 *   temporarily dropped to avoid lock order reversal or deadlock due to
+	 *   reentry.
+	 * - mallctl("epoch", ...) may concurrently refresh stats.  While
+	 *   strictly speaking this is a "concurrent operation", disallowing
+	 *   stats refreshes would impose an inconvenient burden.
+	 */
+
+	/* Large allocations. */
+	malloc_mutex_lock(tsd_tsdn(tsd), &arena->large_mtx);
+
+	for (extent_t *extent = extent_list_first(&arena->large); extent !=
+	    NULL; extent = extent_list_first(&arena->large)) {
+		void *ptr = extent_base_get(extent);
+		size_t usize;
+
+		malloc_mutex_unlock(tsd_tsdn(tsd), &arena->large_mtx);
+		alloc_ctx_t alloc_ctx;
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+		assert(alloc_ctx.szind != SC_NSIZES);
+
+		if (config_stats || (config_prof && opt_prof)) {
+			usize = sz_index2size(alloc_ctx.szind);
+			assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+		}
+		/* Remove large allocation from prof sample set. */
+		if (config_prof && opt_prof) {
+			prof_free(tsd, ptr, usize, &alloc_ctx);
+		}
+		large_dalloc(tsd_tsdn(tsd), extent);
+		malloc_mutex_lock(tsd_tsdn(tsd), &arena->large_mtx);
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &arena->large_mtx);
+
+	/* Bins. */
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			arena_bin_reset(tsd, arena,
+			    &arena->bins[i].bin_shards[j]);
+		}
+	}
+
+	atomic_store_zu(&arena->nactive, 0, ATOMIC_RELAXED);
+}
+
+static void
+arena_destroy_retained(tsdn_t *tsdn, arena_t *arena) {
+	/*
+	 * Iterate over the retained extents and destroy them.  This gives the
+	 * extent allocator underlying the extent hooks an opportunity to unmap
+	 * all retained memory without having to keep its own metadata
+	 * structures.  In practice, virtual memory for dss-allocated extents is
+	 * leaked here, so best practice is to avoid dss for arenas to be
+	 * destroyed, or provide custom extent hooks that track retained
+	 * dss-based extents for later reuse.
+	 */
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+	extent_t *extent;
+	while ((extent = extents_evict(tsdn, arena, &extent_hooks,
+	    &arena->extents_retained, 0)) != NULL) {
+		extent_destroy_wrapper(tsdn, arena, &extent_hooks, extent);
+	}
+}
+
+void
+arena_destroy(tsd_t *tsd, arena_t *arena) {
+	assert(base_ind_get(arena->base) >= narenas_auto);
+	assert(arena_nthreads_get(arena, false) == 0);
+	assert(arena_nthreads_get(arena, true) == 0);
+
+	/*
+	 * No allocations have occurred since arena_reset() was called.
+	 * Furthermore, the caller (arena_i_destroy_ctl()) purged all cached
+	 * extents, so only retained extents may remain.
+	 */
+	assert(extents_npages_get(&arena->extents_dirty) == 0);
+	assert(extents_npages_get(&arena->extents_muzzy) == 0);
+
+	/* Deallocate retained memory. */
+	arena_destroy_retained(tsd_tsdn(tsd), arena);
+
+	/*
+	 * Remove the arena pointer from the arenas array.  We rely on the fact
+	 * that there is no way for the application to get a dirty read from the
+	 * arenas array unless there is an inherent race in the application
+	 * involving access of an arena being concurrently destroyed.  The
+	 * application must synchronize knowledge of the arena's validity, so as
+	 * long as we use an atomic write to update the arenas array, the
+	 * application will get a clean read any time after it synchronizes
+	 * knowledge that the arena is no longer valid.
+	 */
+	arena_set(base_ind_get(arena->base), NULL);
+
+	/*
+	 * Destroy the base allocator, which manages all metadata ever mapped by
+	 * this arena.
+	 */
+	base_delete(tsd_tsdn(tsd), arena->base);
+}
+
+static extent_t *
+arena_slab_alloc_hard(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, const bin_info_t *bin_info,
+    szind_t szind) {
+	extent_t *slab;
+	bool zero, commit;
+
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	zero = false;
+	commit = true;
+	slab = extent_alloc_wrapper(tsdn, arena, r_extent_hooks, NULL,
+	    bin_info->slab_size, 0, PAGE, true, szind, &zero, &commit);
+
+	if (config_stats && slab != NULL) {
+		arena_stats_mapped_add(tsdn, &arena->stats,
+		    bin_info->slab_size);
+	}
+
+	return slab;
+}
+
+static extent_t *
+arena_slab_alloc(tsdn_t *tsdn, arena_t *arena, szind_t binind, unsigned binshard,
+    const bin_info_t *bin_info) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+	szind_t szind = sz_size2index(bin_info->reg_size);
+	bool zero = false;
+	bool commit = true;
+	extent_t *slab = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_dirty, NULL, bin_info->slab_size, 0, PAGE, true,
+	    binind, &zero, &commit);
+	if (slab == NULL && arena_may_have_muzzy(arena)) {
+		slab = extents_alloc(tsdn, arena, &extent_hooks,
+		    &arena->extents_muzzy, NULL, bin_info->slab_size, 0, PAGE,
+		    true, binind, &zero, &commit);
+	}
+	if (slab == NULL) {
+		slab = arena_slab_alloc_hard(tsdn, arena, &extent_hooks,
+		    bin_info, szind);
+		if (slab == NULL) {
+			return NULL;
+		}
+	}
+	assert(extent_slab_get(slab));
+
+	/* Initialize slab internals. */
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+	extent_nfree_binshard_set(slab, bin_info->nregs, binshard);
+	bitmap_init(slab_data->bitmap, &bin_info->bitmap_info, false);
+
+	arena_nactive_add(arena, extent_size_get(slab) >> LG_PAGE);
+
+	return slab;
+}
+
+static extent_t *
+arena_bin_nonfull_slab_get(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, unsigned binshard) {
+	extent_t *slab;
+	const bin_info_t *bin_info;
+
+	/* Look for a usable slab. */
+	slab = arena_bin_slabs_nonfull_tryget(bin);
+	if (slab != NULL) {
+		return slab;
+	}
+	/* No existing slabs have any space available. */
+
+	bin_info = &bin_infos[binind];
+
+	/* Allocate a new slab. */
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	/******************************/
+	slab = arena_slab_alloc(tsdn, arena, binind, binshard, bin_info);
+	/********************************/
+	malloc_mutex_lock(tsdn, &bin->lock);
+	if (slab != NULL) {
+		if (config_stats) {
+			bin->stats.nslabs++;
+			bin->stats.curslabs++;
+		}
+		return slab;
+	}
+
+	/*
+	 * arena_slab_alloc() failed, but another thread may have made
+	 * sufficient memory available while this one dropped bin->lock above,
+	 * so search one more time.
+	 */
+	slab = arena_bin_slabs_nonfull_tryget(bin);
+	if (slab != NULL) {
+		return slab;
+	}
+
+	return NULL;
+}
+
+/* Re-fill bin->slabcur, then call arena_slab_reg_alloc(). */
+static void *
+arena_bin_malloc_hard(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, unsigned binshard) {
+	const bin_info_t *bin_info;
+	extent_t *slab;
+
+	bin_info = &bin_infos[binind];
+	if (!arena_is_auto(arena) && bin->slabcur != NULL) {
+		arena_bin_slabs_full_insert(arena, bin, bin->slabcur);
+		bin->slabcur = NULL;
+	}
+	slab = arena_bin_nonfull_slab_get(tsdn, arena, bin, binind, binshard);
+	if (bin->slabcur != NULL) {
+		/*
+		 * Another thread updated slabcur while this one ran without the
+		 * bin lock in arena_bin_nonfull_slab_get().
+		 */
+		if (extent_nfree_get(bin->slabcur) > 0) {
+			void *ret = arena_slab_reg_alloc(bin->slabcur,
+			    bin_info);
+			if (slab != NULL) {
+				/*
+				 * arena_slab_alloc() may have allocated slab,
+				 * or it may have been pulled from
+				 * slabs_nonfull.  Therefore it is unsafe to
+				 * make any assumptions about how slab has
+				 * previously been used, and
+				 * arena_bin_lower_slab() must be called, as if
+				 * a region were just deallocated from the slab.
+				 */
+				if (extent_nfree_get(slab) == bin_info->nregs) {
+					arena_dalloc_bin_slab(tsdn, arena, slab,
+					    bin);
+				} else {
+					arena_bin_lower_slab(tsdn, arena, slab,
+					    bin);
+				}
+			}
+			return ret;
+		}
+
+		arena_bin_slabs_full_insert(arena, bin, bin->slabcur);
+		bin->slabcur = NULL;
+	}
+
+	if (slab == NULL) {
+		return NULL;
+	}
+	bin->slabcur = slab;
+
+	assert(extent_nfree_get(bin->slabcur) > 0);
+
+	return arena_slab_reg_alloc(slab, bin_info);
+}
+
+/* Choose a bin shard and return the locked bin. */
+bin_t *
+arena_bin_choose_lock(tsdn_t *tsdn, arena_t *arena, szind_t binind,
+    unsigned *binshard) {
+	bin_t *bin;
+	if (tsdn_null(tsdn) || tsd_arena_get(tsdn_tsd(tsdn)) == NULL) {
+		*binshard = 0;
+	} else {
+		*binshard = tsd_binshardsp_get(tsdn_tsd(tsdn))->binshard[binind];
+	}
+	assert(*binshard < bin_infos[binind].n_shards);
+	bin = &arena->bins[binind].bin_shards[*binshard];
+	malloc_mutex_lock(tsdn, &bin->lock);
+
+	return bin;
+}
+
+void
+arena_tcache_fill_small(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, uint64_t prof_accumbytes) {
+	unsigned i, nfill, cnt;
+
+	assert(tbin->ncached == 0);
+
+	if (config_prof && arena_prof_accum(tsdn, arena, prof_accumbytes)) {
+		prof_idump(tsdn);
+	}
+
+	unsigned binshard;
+	bin_t *bin = arena_bin_choose_lock(tsdn, arena, binind, &binshard);
+
+	for (i = 0, nfill = (tcache_bin_info[binind].ncached_max >>
+	    tcache->lg_fill_div[binind]); i < nfill; i += cnt) {
+		extent_t *slab;
+		if ((slab = bin->slabcur) != NULL && extent_nfree_get(slab) >
+		    0) {
+			unsigned tofill = nfill - i;
+			cnt = tofill < extent_nfree_get(slab) ?
+				tofill : extent_nfree_get(slab);
+			arena_slab_reg_alloc_batch(
+			   slab, &bin_infos[binind], cnt,
+			   tbin->avail - nfill + i);
+		} else {
+			cnt = 1;
+			void *ptr = arena_bin_malloc_hard(tsdn, arena, bin,
+			    binind, binshard);
+			/*
+			 * OOM.  tbin->avail isn't yet filled down to its first
+			 * element, so the successful allocations (if any) must
+			 * be moved just before tbin->avail before bailing out.
+			 */
+			if (ptr == NULL) {
+				if (i > 0) {
+					memmove(tbin->avail - i,
+						tbin->avail - nfill,
+						i * sizeof(void *));
+				}
+				break;
+			}
+			/* Insert such that low regions get used first. */
+			*(tbin->avail - nfill + i) = ptr;
+		}
+		if (config_fill && unlikely(opt_junk_alloc)) {
+			for (unsigned j = 0; j < cnt; j++) {
+				void* ptr = *(tbin->avail - nfill + i + j);
+				arena_alloc_junk_small(ptr, &bin_infos[binind],
+							true);
+			}
+		}
+	}
+	if (config_stats) {
+		bin->stats.nmalloc += i;
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		bin->stats.curregs += i;
+		bin->stats.nfills++;
+		tbin->tstats.nrequests = 0;
+	}
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	tbin->ncached = i;
+	arena_decay_tick(tsdn, arena);
+}
+
+void
+arena_alloc_junk_small(void *ptr, const bin_info_t *bin_info, bool zero) {
+	if (!zero) {
+		memset(ptr, JEMALLOC_ALLOC_JUNK, bin_info->reg_size);
+	}
+}
+
+static void
+arena_dalloc_junk_small_impl(void *ptr, const bin_info_t *bin_info) {
+	memset(ptr, JEMALLOC_FREE_JUNK, bin_info->reg_size);
+}
+arena_dalloc_junk_small_t *JET_MUTABLE arena_dalloc_junk_small =
+    arena_dalloc_junk_small_impl;
+
+static void *
+arena_malloc_small(tsdn_t *tsdn, arena_t *arena, szind_t binind, bool zero) {
+	void *ret;
+	bin_t *bin;
+	size_t usize;
+	extent_t *slab;
+
+	assert(binind < SC_NBINS);
+	usize = sz_index2size(binind);
+	unsigned binshard;
+	bin = arena_bin_choose_lock(tsdn, arena, binind, &binshard);
+
+	if ((slab = bin->slabcur) != NULL && extent_nfree_get(slab) > 0) {
+		ret = arena_slab_reg_alloc(slab, &bin_infos[binind]);
+	} else {
+		ret = arena_bin_malloc_hard(tsdn, arena, bin, binind, binshard);
+	}
+
+	if (ret == NULL) {
+		malloc_mutex_unlock(tsdn, &bin->lock);
+		return NULL;
+	}
+
+	if (config_stats) {
+		bin->stats.nmalloc++;
+		bin->stats.nrequests++;
+		bin->stats.curregs++;
+	}
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	if (config_prof && arena_prof_accum(tsdn, arena, usize)) {
+		prof_idump(tsdn);
+	}
+
+	if (!zero) {
+		if (config_fill) {
+			if (unlikely(opt_junk_alloc)) {
+				arena_alloc_junk_small(ret,
+				    &bin_infos[binind], false);
+			} else if (unlikely(opt_zero)) {
+				memset(ret, 0, usize);
+			}
+		}
+	} else {
+		if (config_fill && unlikely(opt_junk_alloc)) {
+			arena_alloc_junk_small(ret, &bin_infos[binind],
+			    true);
+		}
+		memset(ret, 0, usize);
+	}
+
+	arena_decay_tick(tsdn, arena);
+	return ret;
+}
+
+void *
+arena_malloc_hard(tsdn_t *tsdn, arena_t *arena, size_t size, szind_t ind,
+    bool zero) {
+	assert(!tsdn_null(tsdn) || arena != NULL);
+
+	if (likely(!tsdn_null(tsdn))) {
+		arena = arena_choose_maybe_huge(tsdn_tsd(tsdn), arena, size);
+	}
+	if (unlikely(arena == NULL)) {
+		return NULL;
+	}
+
+	if (likely(size <= SC_SMALL_MAXCLASS)) {
+		return arena_malloc_small(tsdn, arena, ind, zero);
+	}
+	return large_malloc(tsdn, arena, sz_index2size(ind), zero);
+}
+
+void *
+arena_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
+    bool zero, tcache_t *tcache) {
+	void *ret;
+
+	if (usize <= SC_SMALL_MAXCLASS
+	    && (alignment < PAGE
+	    || (alignment == PAGE && (usize & PAGE_MASK) == 0))) {
+		/* Small; alignment doesn't require special slab placement. */
+		ret = arena_malloc(tsdn, arena, usize, sz_size2index(usize),
+		    zero, tcache, true);
+	} else {
+		if (likely(alignment <= CACHELINE)) {
+			ret = large_malloc(tsdn, arena, usize, zero);
+		} else {
+			ret = large_palloc(tsdn, arena, usize, alignment, zero);
+		}
+	}
+	return ret;
+}
+
+void
+arena_prof_promote(tsdn_t *tsdn, void *ptr, size_t usize) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(isalloc(tsdn, ptr) == SC_LARGE_MINCLASS);
+	assert(usize <= SC_SMALL_MAXCLASS);
+
+	if (config_opt_safety_checks) {
+		safety_check_set_redzone(ptr, usize, SC_LARGE_MINCLASS);
+	}
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent = rtree_extent_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+	arena_t *arena = extent_arena_get(extent);
+
+	szind_t szind = sz_size2index(usize);
+	extent_szind_set(extent, szind);
+	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
+	    szind, false);
+
+	prof_accum_cancel(tsdn, &arena->prof_accum, usize);
+
+	assert(isalloc(tsdn, ptr) == usize);
+}
+
+static size_t
+arena_prof_demote(tsdn_t *tsdn, extent_t *extent, const void *ptr) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_szind_set(extent, SC_NBINS);
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
+	    SC_NBINS, false);
+
+	assert(isalloc(tsdn, ptr) == SC_LARGE_MINCLASS);
+
+	return SC_LARGE_MINCLASS;
+}
+
+void
+arena_dalloc_promoted(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    bool slow_path) {
+	cassert(config_prof);
+	assert(opt_prof);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	size_t usize = extent_usize_get(extent);
+	size_t bumped_usize = arena_prof_demote(tsdn, extent, ptr);
+	if (config_opt_safety_checks && usize < SC_LARGE_MINCLASS) {
+		/*
+		 * Currently, we only do redzoning for small sampled
+		 * allocations.
+		 */
+		assert(bumped_usize == SC_LARGE_MINCLASS);
+		safety_check_verify_redzone(ptr, usize, bumped_usize);
+	}
+	if (bumped_usize <= tcache_maxclass && tcache != NULL) {
+		tcache_dalloc_large(tsdn_tsd(tsdn), tcache, ptr,
+		    sz_size2index(bumped_usize), slow_path);
+	} else {
+		large_dalloc(tsdn, extent);
+	}
+}
+
+static void
+arena_dissociate_bin_slab(arena_t *arena, extent_t *slab, bin_t *bin) {
+	/* Dissociate slab from bin. */
+	if (slab == bin->slabcur) {
+		bin->slabcur = NULL;
+	} else {
+		szind_t binind = extent_szind_get(slab);
+		const bin_info_t *bin_info = &bin_infos[binind];
+
+		/*
+		 * The following block's conditional is necessary because if the
+		 * slab only contains one region, then it never gets inserted
+		 * into the non-full slabs heap.
+		 */
+		if (bin_info->nregs == 1) {
+			arena_bin_slabs_full_remove(arena, bin, slab);
+		} else {
+			arena_bin_slabs_nonfull_remove(bin, slab);
+		}
+	}
+}
+
+static void
+arena_dalloc_bin_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin) {
+	assert(slab != bin->slabcur);
+
+	malloc_mutex_unlock(tsdn, &bin->lock);
+	/******************************/
+	arena_slab_dalloc(tsdn, arena, slab);
+	/****************************/
+	malloc_mutex_lock(tsdn, &bin->lock);
+	if (config_stats) {
+		bin->stats.curslabs--;
+	}
+}
+
+static void
+arena_bin_lower_slab(tsdn_t *tsdn, arena_t *arena, extent_t *slab,
+    bin_t *bin) {
+	assert(extent_nfree_get(slab) > 0);
+
+	/*
+	 * Make sure that if bin->slabcur is non-NULL, it refers to the
+	 * oldest/lowest non-full slab.  It is okay to NULL slabcur out rather
+	 * than proactively keeping it pointing at the oldest/lowest non-full
+	 * slab.
+	 */
+	if (bin->slabcur != NULL && extent_snad_comp(bin->slabcur, slab) > 0) {
+		/* Switch slabcur. */
+		if (extent_nfree_get(bin->slabcur) > 0) {
+			arena_bin_slabs_nonfull_insert(bin, bin->slabcur);
+		} else {
+			arena_bin_slabs_full_insert(arena, bin, bin->slabcur);
+		}
+		bin->slabcur = slab;
+		if (config_stats) {
+			bin->stats.reslabs++;
+		}
+	} else {
+		arena_bin_slabs_nonfull_insert(bin, slab);
+	}
+}
+
+static void
+arena_dalloc_bin_locked_impl(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, extent_t *slab, void *ptr, bool junked) {
+	arena_slab_data_t *slab_data = extent_slab_data_get(slab);
+	const bin_info_t *bin_info = &bin_infos[binind];
+
+	if (!junked && config_fill && unlikely(opt_junk_free)) {
+		arena_dalloc_junk_small(ptr, bin_info);
+	}
+
+	arena_slab_reg_dalloc(slab, slab_data, ptr);
+	unsigned nfree = extent_nfree_get(slab);
+	if (nfree == bin_info->nregs) {
+		arena_dissociate_bin_slab(arena, slab, bin);
+		arena_dalloc_bin_slab(tsdn, arena, slab, bin);
+	} else if (nfree == 1 && slab != bin->slabcur) {
+		arena_bin_slabs_full_remove(arena, bin, slab);
+		arena_bin_lower_slab(tsdn, arena, slab, bin);
+	}
+
+	if (config_stats) {
+		bin->stats.ndalloc++;
+		bin->stats.curregs--;
+	}
+}
+
+void
+arena_dalloc_bin_junked_locked(tsdn_t *tsdn, arena_t *arena, bin_t *bin,
+    szind_t binind, extent_t *extent, void *ptr) {
+	arena_dalloc_bin_locked_impl(tsdn, arena, bin, binind, extent, ptr,
+	    true);
+}
+
+static void
+arena_dalloc_bin(tsdn_t *tsdn, arena_t *arena, extent_t *extent, void *ptr) {
+	szind_t binind = extent_szind_get(extent);
+	unsigned binshard = extent_binshard_get(extent);
+	bin_t *bin = &arena->bins[binind].bin_shards[binshard];
+
+	malloc_mutex_lock(tsdn, &bin->lock);
+	arena_dalloc_bin_locked_impl(tsdn, arena, bin, binind, extent, ptr,
+	    false);
+	malloc_mutex_unlock(tsdn, &bin->lock);
+}
+
+void
+arena_dalloc_small(tsdn_t *tsdn, void *ptr) {
+	extent_t *extent = iealloc(tsdn, ptr);
+	arena_t *arena = extent_arena_get(extent);
+
+	arena_dalloc_bin(tsdn, arena, extent, ptr);
+	arena_decay_tick(tsdn, arena);
+}
+
+bool
+arena_ralloc_no_move(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero, size_t *newsize) {
+	bool ret;
+	/* Calls with non-zero extra had to clamp extra. */
+	assert(extra == 0 || size + extra <= SC_LARGE_MAXCLASS);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		ret = true;
+		goto done;
+	}
+
+	size_t usize_min = sz_s2u(size);
+	size_t usize_max = sz_s2u(size + extra);
+	if (likely(oldsize <= SC_SMALL_MAXCLASS && usize_min
+	    <= SC_SMALL_MAXCLASS)) {
+		/*
+		 * Avoid moving the allocation if the size class can be left the
+		 * same.
+		 */
+		assert(bin_infos[sz_size2index(oldsize)].reg_size ==
+		    oldsize);
+		if ((usize_max > SC_SMALL_MAXCLASS
+		    || sz_size2index(usize_max) != sz_size2index(oldsize))
+		    && (size > oldsize || usize_max < oldsize)) {
+			ret = true;
+			goto done;
+		}
+
+		arena_decay_tick(tsdn, extent_arena_get(extent));
+		ret = false;
+	} else if (oldsize >= SC_LARGE_MINCLASS
+	    && usize_max >= SC_LARGE_MINCLASS) {
+		ret = large_ralloc_no_move(tsdn, extent, usize_min, usize_max,
+		    zero);
+	} else {
+		ret = true;
+	}
+done:
+	assert(extent == iealloc(tsdn, ptr));
+	*newsize = extent_usize_get(extent);
+
+	return ret;
+}
+
+static void *
+arena_ralloc_move_helper(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache) {
+	if (alignment == 0) {
+		return arena_malloc(tsdn, arena, usize, sz_size2index(usize),
+		    zero, tcache, true);
+	}
+	usize = sz_sa2u(usize, alignment);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+	return ipalloct(tsdn, usize, alignment, zero, tcache, arena);
+}
+
+void *
+arena_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t oldsize,
+    size_t size, size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args) {
+	size_t usize = sz_s2u(size);
+	if (unlikely(usize == 0 || size > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+
+	if (likely(usize <= SC_SMALL_MAXCLASS)) {
+		/* Try to avoid moving the allocation. */
+		UNUSED size_t newsize;
+		if (!arena_ralloc_no_move(tsdn, ptr, oldsize, usize, 0, zero,
+		    &newsize)) {
+			hook_invoke_expand(hook_args->is_realloc
+			    ? hook_expand_realloc : hook_expand_rallocx,
+			    ptr, oldsize, usize, (uintptr_t)ptr,
+			    hook_args->args);
+			return ptr;
+		}
+	}
+
+	if (oldsize >= SC_LARGE_MINCLASS
+	    && usize >= SC_LARGE_MINCLASS) {
+		return large_ralloc(tsdn, arena, ptr, usize,
+		    alignment, zero, tcache, hook_args);
+	}
+
+	/*
+	 * size and oldsize are different enough that we need to move the
+	 * object.  In that case, fall back to allocating new space and copying.
+	 */
+	void *ret = arena_ralloc_move_helper(tsdn, arena, usize, alignment,
+	    zero, tcache);
+	if (ret == NULL) {
+		return NULL;
+	}
+
+	hook_invoke_alloc(hook_args->is_realloc
+	    ? hook_alloc_realloc : hook_alloc_rallocx, ret, (uintptr_t)ret,
+	    hook_args->args);
+	hook_invoke_dalloc(hook_args->is_realloc
+	    ? hook_dalloc_realloc : hook_dalloc_rallocx, ptr, hook_args->args);
+
+	/*
+	 * Junk/zero-filling were already done by
+	 * ipalloc()/arena_malloc().
+	 */
+	size_t copysize = (usize < oldsize) ? usize : oldsize;
+	memcpy(ret, ptr, copysize);
+	isdalloct(tsdn, ptr, oldsize, tcache, NULL, true);
+	return ret;
+}
+
+dss_prec_t
+arena_dss_prec_get(arena_t *arena) {
+	return (dss_prec_t)atomic_load_u(&arena->dss_prec, ATOMIC_ACQUIRE);
+}
+
+bool
+arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec) {
+	if (!have_dss) {
+		return (dss_prec != dss_prec_disabled);
+	}
+	atomic_store_u(&arena->dss_prec, (unsigned)dss_prec, ATOMIC_RELEASE);
+	return false;
+}
+
+ssize_t
+arena_dirty_decay_ms_default_get(void) {
+	return atomic_load_zd(&dirty_decay_ms_default, ATOMIC_RELAXED);
+}
+
+bool
+arena_dirty_decay_ms_default_set(ssize_t decay_ms) {
+	if (!arena_decay_ms_valid(decay_ms)) {
+		return true;
+	}
+	atomic_store_zd(&dirty_decay_ms_default, decay_ms, ATOMIC_RELAXED);
+	return false;
+}
+
+ssize_t
+arena_muzzy_decay_ms_default_get(void) {
+	return atomic_load_zd(&muzzy_decay_ms_default, ATOMIC_RELAXED);
+}
+
+bool
+arena_muzzy_decay_ms_default_set(ssize_t decay_ms) {
+	if (!arena_decay_ms_valid(decay_ms)) {
+		return true;
+	}
+	atomic_store_zd(&muzzy_decay_ms_default, decay_ms, ATOMIC_RELAXED);
+	return false;
+}
+
+bool
+arena_retain_grow_limit_get_set(tsd_t *tsd, arena_t *arena, size_t *old_limit,
+    size_t *new_limit) {
+	assert(opt_retain);
+
+	pszind_t new_ind JEMALLOC_CC_SILENCE_INIT(0);
+	if (new_limit != NULL) {
+		size_t limit = *new_limit;
+		/* Grow no more than the new limit. */
+		if ((new_ind = sz_psz2ind(limit + 1) - 1) >= SC_NPSIZES) {
+			return true;
+		}
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &arena->extent_grow_mtx);
+	if (old_limit != NULL) {
+		*old_limit = sz_pind2sz(arena->retain_grow_limit);
+	}
+	if (new_limit != NULL) {
+		arena->retain_grow_limit = new_ind;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &arena->extent_grow_mtx);
+
+	return false;
+}
+
+unsigned
+arena_nthreads_get(arena_t *arena, bool internal) {
+	return atomic_load_u(&arena->nthreads[internal], ATOMIC_RELAXED);
+}
+
+void
+arena_nthreads_inc(arena_t *arena, bool internal) {
+	atomic_fetch_add_u(&arena->nthreads[internal], 1, ATOMIC_RELAXED);
+}
+
+void
+arena_nthreads_dec(arena_t *arena, bool internal) {
+	atomic_fetch_sub_u(&arena->nthreads[internal], 1, ATOMIC_RELAXED);
+}
+
+size_t
+arena_extent_sn_next(arena_t *arena) {
+	return atomic_fetch_add_zu(&arena->extent_sn_next, 1, ATOMIC_RELAXED);
+}
+
+arena_t *
+arena_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	arena_t *arena;
+	base_t *base;
+	unsigned i;
+
+	if (ind == 0) {
+		base = b0get();
+	} else {
+		base = base_new(tsdn, ind, extent_hooks);
+		if (base == NULL) {
+			return NULL;
+		}
+	}
+
+	unsigned nbins_total = 0;
+	for (i = 0; i < SC_NBINS; i++) {
+		nbins_total += bin_infos[i].n_shards;
+	}
+	size_t arena_size = sizeof(arena_t) + sizeof(bin_t) * nbins_total;
+	arena = (arena_t *)base_alloc(tsdn, base, arena_size, CACHELINE);
+	if (arena == NULL) {
+		goto label_error;
+	}
+
+	atomic_store_u(&arena->nthreads[0], 0, ATOMIC_RELAXED);
+	atomic_store_u(&arena->nthreads[1], 0, ATOMIC_RELAXED);
+	arena->last_thd = NULL;
+
+	if (config_stats) {
+		if (arena_stats_init(tsdn, &arena->stats)) {
+			goto label_error;
+		}
+
+		ql_new(&arena->tcache_ql);
+		ql_new(&arena->cache_bin_array_descriptor_ql);
+		if (malloc_mutex_init(&arena->tcache_ql_mtx, "tcache_ql",
+		    WITNESS_RANK_TCACHE_QL, malloc_mutex_rank_exclusive)) {
+			goto label_error;
+		}
+	}
+
+	if (config_prof) {
+		if (prof_accum_init(tsdn, &arena->prof_accum)) {
+			goto label_error;
+		}
+	}
+
+	if (config_cache_oblivious) {
+		/*
+		 * A nondeterministic seed based on the address of arena reduces
+		 * the likelihood of lockstep non-uniform cache index
+		 * utilization among identical concurrent processes, but at the
+		 * cost of test repeatability.  For debug builds, instead use a
+		 * deterministic seed.
+		 */
+		atomic_store_zu(&arena->offset_state, config_debug ? ind :
+		    (size_t)(uintptr_t)arena, ATOMIC_RELAXED);
+	}
+
+	atomic_store_zu(&arena->extent_sn_next, 0, ATOMIC_RELAXED);
+
+	atomic_store_u(&arena->dss_prec, (unsigned)extent_dss_prec_get(),
+	    ATOMIC_RELAXED);
+
+	atomic_store_zu(&arena->nactive, 0, ATOMIC_RELAXED);
+
+	extent_list_init(&arena->large);
+	if (malloc_mutex_init(&arena->large_mtx, "arena_large",
+	    WITNESS_RANK_ARENA_LARGE, malloc_mutex_rank_exclusive)) {
+		goto label_error;
+	}
+
+	/*
+	 * Delay coalescing for dirty extents despite the disruptive effect on
+	 * memory layout for best-fit extent allocation, since cached extents
+	 * are likely to be reused soon after deallocation, and the cost of
+	 * merging/splitting extents is non-trivial.
+	 */
+	if (extents_init(tsdn, &arena->extents_dirty, extent_state_dirty,
+	    true)) {
+		goto label_error;
+	}
+	/*
+	 * Coalesce muzzy extents immediately, because operations on them are in
+	 * the critical path much less often than for dirty extents.
+	 */
+	if (extents_init(tsdn, &arena->extents_muzzy, extent_state_muzzy,
+	    false)) {
+		goto label_error;
+	}
+	/*
+	 * Coalesce retained extents immediately, in part because they will
+	 * never be evicted (and therefore there's no opportunity for delayed
+	 * coalescing), but also because operations on retained extents are not
+	 * in the critical path.
+	 */
+	if (extents_init(tsdn, &arena->extents_retained, extent_state_retained,
+	    false)) {
+		goto label_error;
+	}
+
+	if (arena_decay_init(&arena->decay_dirty,
+	    arena_dirty_decay_ms_default_get(), &arena->stats.decay_dirty)) {
+		goto label_error;
+	}
+	if (arena_decay_init(&arena->decay_muzzy,
+	    arena_muzzy_decay_ms_default_get(), &arena->stats.decay_muzzy)) {
+		goto label_error;
+	}
+
+	arena->extent_grow_next = sz_psz2ind(HUGEPAGE);
+	arena->retain_grow_limit = sz_psz2ind(SC_LARGE_MAXCLASS);
+	if (malloc_mutex_init(&arena->extent_grow_mtx, "extent_grow",
+	    WITNESS_RANK_EXTENT_GROW, malloc_mutex_rank_exclusive)) {
+		goto label_error;
+	}
+
+	extent_avail_new(&arena->extent_avail);
+	if (malloc_mutex_init(&arena->extent_avail_mtx, "extent_avail",
+	    WITNESS_RANK_EXTENT_AVAIL, malloc_mutex_rank_exclusive)) {
+		goto label_error;
+	}
+
+	/* Initialize bins. */
+	uintptr_t bin_addr = (uintptr_t)arena + sizeof(arena_t);
+	atomic_store_u(&arena->binshard_next, 0, ATOMIC_RELEASE);
+	for (i = 0; i < SC_NBINS; i++) {
+		unsigned nshards = bin_infos[i].n_shards;
+		arena->bins[i].bin_shards = (bin_t *)bin_addr;
+		bin_addr += nshards * sizeof(bin_t);
+		for (unsigned j = 0; j < nshards; j++) {
+			bool err = bin_init(&arena->bins[i].bin_shards[j]);
+			if (err) {
+				goto label_error;
+			}
+		}
+	}
+	assert(bin_addr == (uintptr_t)arena + arena_size);
+
+	arena->base = base;
+	/* Set arena before creating background threads. */
+	arena_set(ind, arena);
+
+	nstime_init(&arena->create_time, 0);
+	nstime_update(&arena->create_time);
+
+	/* We don't support reentrancy for arena 0 bootstrapping. */
+	if (ind != 0) {
+		/*
+		 * If we're here, then arena 0 already exists, so bootstrapping
+		 * is done enough that we should have tsd.
+		 */
+		assert(!tsdn_null(tsdn));
+		pre_reentrancy(tsdn_tsd(tsdn), arena);
+		if (test_hooks_arena_new_hook) {
+			test_hooks_arena_new_hook();
+		}
+		post_reentrancy(tsdn_tsd(tsdn));
+	}
+
+	return arena;
+label_error:
+	if (ind != 0) {
+		base_delete(tsdn, base);
+	}
+	return NULL;
+}
+
+arena_t *
+arena_choose_huge(tsd_t *tsd) {
+	/* huge_arena_ind can be 0 during init (will use a0). */
+	if (huge_arena_ind == 0) {
+		assert(!malloc_initialized());
+	}
+
+	arena_t *huge_arena = arena_get(tsd_tsdn(tsd), huge_arena_ind, false);
+	if (huge_arena == NULL) {
+		/* Create the huge arena on demand. */
+		assert(huge_arena_ind != 0);
+		huge_arena = arena_get(tsd_tsdn(tsd), huge_arena_ind, true);
+		if (huge_arena == NULL) {
+			return NULL;
+		}
+		/*
+		 * Purge eagerly for huge allocations, because: 1) number of
+		 * huge allocations is usually small, which means ticker based
+		 * decay is not reliable; and 2) less immediate reuse is
+		 * expected for huge allocations.
+		 */
+		if (arena_dirty_decay_ms_default_get() > 0) {
+			arena_dirty_decay_ms_set(tsd_tsdn(tsd), huge_arena, 0);
+		}
+		if (arena_muzzy_decay_ms_default_get() > 0) {
+			arena_muzzy_decay_ms_set(tsd_tsdn(tsd), huge_arena, 0);
+		}
+	}
+
+	return huge_arena;
+}
+
+bool
+arena_init_huge(void) {
+	bool huge_enabled;
+
+	/* The threshold should be large size class. */
+	if (opt_oversize_threshold > SC_LARGE_MAXCLASS ||
+	    opt_oversize_threshold < SC_LARGE_MINCLASS) {
+		opt_oversize_threshold = 0;
+		oversize_threshold = SC_LARGE_MAXCLASS + PAGE;
+		huge_enabled = false;
+	} else {
+		/* Reserve the index for the huge arena. */
+		huge_arena_ind = narenas_total_get();
+		oversize_threshold = opt_oversize_threshold;
+		huge_enabled = true;
+	}
+
+	return huge_enabled;
+}
+
+bool
+arena_is_huge(unsigned arena_ind) {
+	if (huge_arena_ind == 0) {
+		return false;
+	}
+	return (arena_ind == huge_arena_ind);
+}
+
+void
+arena_boot(sc_data_t *sc_data) {
+	arena_dirty_decay_ms_default_set(opt_dirty_decay_ms);
+	arena_muzzy_decay_ms_default_set(opt_muzzy_decay_ms);
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		sc_t *sc = &sc_data->sc[i];
+		div_init(&arena_binind_div_info[i],
+		    (1U << sc->lg_base) + (sc->ndelta << sc->lg_delta));
+	}
+}
+
+void
+arena_prefork0(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_prefork(tsdn, &arena->decay_muzzy.mtx);
+}
+
+void
+arena_prefork1(tsdn_t *tsdn, arena_t *arena) {
+	if (config_stats) {
+		malloc_mutex_prefork(tsdn, &arena->tcache_ql_mtx);
+	}
+}
+
+void
+arena_prefork2(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->extent_grow_mtx);
+}
+
+void
+arena_prefork3(tsdn_t *tsdn, arena_t *arena) {
+	extents_prefork(tsdn, &arena->extents_dirty);
+	extents_prefork(tsdn, &arena->extents_muzzy);
+	extents_prefork(tsdn, &arena->extents_retained);
+}
+
+void
+arena_prefork4(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->extent_avail_mtx);
+}
+
+void
+arena_prefork5(tsdn_t *tsdn, arena_t *arena) {
+	base_prefork(tsdn, arena->base);
+}
+
+void
+arena_prefork6(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_prefork(tsdn, &arena->large_mtx);
+}
+
+void
+arena_prefork7(tsdn_t *tsdn, arena_t *arena) {
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_prefork(tsdn, &arena->bins[i].bin_shards[j]);
+		}
+	}
+}
+
+void
+arena_postfork_parent(tsdn_t *tsdn, arena_t *arena) {
+	unsigned i;
+
+	for (i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_postfork_parent(tsdn,
+			    &arena->bins[i].bin_shards[j]);
+		}
+	}
+	malloc_mutex_postfork_parent(tsdn, &arena->large_mtx);
+	base_postfork_parent(tsdn, arena->base);
+	malloc_mutex_postfork_parent(tsdn, &arena->extent_avail_mtx);
+	extents_postfork_parent(tsdn, &arena->extents_dirty);
+	extents_postfork_parent(tsdn, &arena->extents_muzzy);
+	extents_postfork_parent(tsdn, &arena->extents_retained);
+	malloc_mutex_postfork_parent(tsdn, &arena->extent_grow_mtx);
+	malloc_mutex_postfork_parent(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_postfork_parent(tsdn, &arena->decay_muzzy.mtx);
+	if (config_stats) {
+		malloc_mutex_postfork_parent(tsdn, &arena->tcache_ql_mtx);
+	}
+}
+
+void
+arena_postfork_child(tsdn_t *tsdn, arena_t *arena) {
+	unsigned i;
+
+	atomic_store_u(&arena->nthreads[0], 0, ATOMIC_RELAXED);
+	atomic_store_u(&arena->nthreads[1], 0, ATOMIC_RELAXED);
+	if (tsd_arena_get(tsdn_tsd(tsdn)) == arena) {
+		arena_nthreads_inc(arena, false);
+	}
+	if (tsd_iarena_get(tsdn_tsd(tsdn)) == arena) {
+		arena_nthreads_inc(arena, true);
+	}
+	if (config_stats) {
+		ql_new(&arena->tcache_ql);
+		ql_new(&arena->cache_bin_array_descriptor_ql);
+		tcache_t *tcache = tcache_get(tsdn_tsd(tsdn));
+		if (tcache != NULL && tcache->arena == arena) {
+			ql_elm_new(tcache, link);
+			ql_tail_insert(&arena->tcache_ql, tcache, link);
+			cache_bin_array_descriptor_init(
+			    &tcache->cache_bin_array_descriptor,
+			    tcache->bins_small, tcache->bins_large);
+			ql_tail_insert(&arena->cache_bin_array_descriptor_ql,
+			    &tcache->cache_bin_array_descriptor, link);
+		}
+	}
+
+	for (i = 0; i < SC_NBINS; i++) {
+		for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+			bin_postfork_child(tsdn, &arena->bins[i].bin_shards[j]);
+		}
+	}
+	malloc_mutex_postfork_child(tsdn, &arena->large_mtx);
+	base_postfork_child(tsdn, arena->base);
+	malloc_mutex_postfork_child(tsdn, &arena->extent_avail_mtx);
+	extents_postfork_child(tsdn, &arena->extents_dirty);
+	extents_postfork_child(tsdn, &arena->extents_muzzy);
+	extents_postfork_child(tsdn, &arena->extents_retained);
+	malloc_mutex_postfork_child(tsdn, &arena->extent_grow_mtx);
+	malloc_mutex_postfork_child(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_postfork_child(tsdn, &arena->decay_muzzy.mtx);
+	if (config_stats) {
+		malloc_mutex_postfork_child(tsdn, &arena->tcache_ql_mtx);
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/src/background_thread.c b/memkind-1.10.0/jemalloc/src/background_thread.c
new file mode 100644
index 0000000..57b9b25
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/background_thread.c
@@ -0,0 +1,939 @@
+#define JEMALLOC_BACKGROUND_THREAD_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+
+JEMALLOC_DIAGNOSTIC_DISABLE_SPURIOUS
+
+/******************************************************************************/
+/* Data. */
+
+/* This option should be opt-in only. */
+#define BACKGROUND_THREAD_DEFAULT false
+/* Read-only after initialization. */
+bool opt_background_thread = BACKGROUND_THREAD_DEFAULT;
+size_t opt_max_background_threads = MAX_BACKGROUND_THREAD_LIMIT + 1;
+
+/* Used for thread creation, termination and stats. */
+malloc_mutex_t background_thread_lock;
+/* Indicates global state.  Atomic because decay reads this w/o locking. */
+atomic_b_t background_thread_enabled_state;
+size_t n_background_threads;
+size_t max_background_threads;
+/* Thread info per-index. */
+background_thread_info_t *background_thread_info;
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+
+static int (*pthread_create_fptr)(pthread_t *__restrict, const pthread_attr_t *,
+    void *(*)(void *), void *__restrict);
+
+static void
+pthread_create_wrapper_init(void) {
+#ifdef JEMALLOC_LAZY_LOCK
+	if (!isthreaded) {
+		isthreaded = true;
+	}
+#endif
+}
+
+int
+pthread_create_wrapper(pthread_t *__restrict thread, const pthread_attr_t *attr,
+    void *(*start_routine)(void *), void *__restrict arg) {
+	pthread_create_wrapper_init();
+
+	return pthread_create_fptr(thread, attr, start_routine, arg);
+}
+#endif /* JEMALLOC_PTHREAD_CREATE_WRAPPER */
+
+#ifndef JEMALLOC_BACKGROUND_THREAD
+#define NOT_REACHED { not_reached(); }
+bool background_thread_create(tsd_t *tsd, unsigned arena_ind) NOT_REACHED
+bool background_threads_enable(tsd_t *tsd) NOT_REACHED
+bool background_threads_disable(tsd_t *tsd) NOT_REACHED
+void background_thread_interval_check(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, size_t npages_new) NOT_REACHED
+void background_thread_prefork0(tsdn_t *tsdn) NOT_REACHED
+void background_thread_prefork1(tsdn_t *tsdn) NOT_REACHED
+void background_thread_postfork_parent(tsdn_t *tsdn) NOT_REACHED
+void background_thread_postfork_child(tsdn_t *tsdn) NOT_REACHED
+bool background_thread_stats_read(tsdn_t *tsdn,
+    background_thread_stats_t *stats) NOT_REACHED
+void background_thread_ctl_init(tsdn_t *tsdn) NOT_REACHED
+#undef NOT_REACHED
+#else
+
+static bool background_thread_enabled_at_fork;
+
+static void
+background_thread_info_init(tsdn_t *tsdn, background_thread_info_t *info) {
+	background_thread_wakeup_time_set(tsdn, info, 0);
+	info->npages_to_purge_new = 0;
+	if (config_stats) {
+		info->tot_n_runs = 0;
+		nstime_init(&info->tot_sleep_time, 0);
+	}
+}
+
+static inline bool
+set_current_thread_affinity(int cpu) {
+#if defined(JEMALLOC_HAVE_SCHED_SETAFFINITY)
+	cpu_set_t cpuset;
+	CPU_ZERO(&cpuset);
+	CPU_SET(cpu, &cpuset);
+	int ret = sched_setaffinity(0, sizeof(cpu_set_t), &cpuset);
+
+	return (ret != 0);
+#else
+	return false;
+#endif
+}
+
+/* Threshold for determining when to wake up the background thread. */
+#define BACKGROUND_THREAD_NPAGES_THRESHOLD UINT64_C(1024)
+#define BILLION UINT64_C(1000000000)
+/* Minimal sleep interval 100 ms. */
+#define BACKGROUND_THREAD_MIN_INTERVAL_NS (BILLION / 10)
+
+static inline size_t
+decay_npurge_after_interval(arena_decay_t *decay, size_t interval) {
+	size_t i;
+	uint64_t sum = 0;
+	for (i = 0; i < interval; i++) {
+		sum += decay->backlog[i] * h_steps[i];
+	}
+	for (; i < SMOOTHSTEP_NSTEPS; i++) {
+		sum += decay->backlog[i] * (h_steps[i] - h_steps[i - interval]);
+	}
+
+	return (size_t)(sum >> SMOOTHSTEP_BFP);
+}
+
+static uint64_t
+arena_decay_compute_purge_interval_impl(tsdn_t *tsdn, arena_decay_t *decay,
+    extents_t *extents) {
+	if (malloc_mutex_trylock(tsdn, &decay->mtx)) {
+		/* Use minimal interval if decay is contended. */
+		return BACKGROUND_THREAD_MIN_INTERVAL_NS;
+	}
+
+	uint64_t interval;
+	ssize_t decay_time = atomic_load_zd(&decay->time_ms, ATOMIC_RELAXED);
+	if (decay_time <= 0) {
+		/* Purging is eagerly done or disabled currently. */
+		interval = BACKGROUND_THREAD_INDEFINITE_SLEEP;
+		goto label_done;
+	}
+
+	uint64_t decay_interval_ns = nstime_ns(&decay->interval);
+	assert(decay_interval_ns > 0);
+	size_t npages = extents_npages_get(extents);
+	if (npages == 0) {
+		unsigned i;
+		for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+			if (decay->backlog[i] > 0) {
+				break;
+			}
+		}
+		if (i == SMOOTHSTEP_NSTEPS) {
+			/* No dirty pages recorded.  Sleep indefinitely. */
+			interval = BACKGROUND_THREAD_INDEFINITE_SLEEP;
+			goto label_done;
+		}
+	}
+	if (npages <= BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		/* Use max interval. */
+		interval = decay_interval_ns * SMOOTHSTEP_NSTEPS;
+		goto label_done;
+	}
+
+	size_t lb = BACKGROUND_THREAD_MIN_INTERVAL_NS / decay_interval_ns;
+	size_t ub = SMOOTHSTEP_NSTEPS;
+	/* Minimal 2 intervals to ensure reaching next epoch deadline. */
+	lb = (lb < 2) ? 2 : lb;
+	if ((decay_interval_ns * ub <= BACKGROUND_THREAD_MIN_INTERVAL_NS) ||
+	    (lb + 2 > ub)) {
+		interval = BACKGROUND_THREAD_MIN_INTERVAL_NS;
+		goto label_done;
+	}
+
+	assert(lb + 2 <= ub);
+	size_t npurge_lb, npurge_ub;
+	npurge_lb = decay_npurge_after_interval(decay, lb);
+	if (npurge_lb > BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		interval = decay_interval_ns * lb;
+		goto label_done;
+	}
+	npurge_ub = decay_npurge_after_interval(decay, ub);
+	if (npurge_ub < BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		interval = decay_interval_ns * ub;
+		goto label_done;
+	}
+
+	unsigned n_search = 0;
+	size_t target, npurge;
+	while ((npurge_lb + BACKGROUND_THREAD_NPAGES_THRESHOLD < npurge_ub)
+	    && (lb + 2 < ub)) {
+		target = (lb + ub) / 2;
+		npurge = decay_npurge_after_interval(decay, target);
+		if (npurge > BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+			ub = target;
+			npurge_ub = npurge;
+		} else {
+			lb = target;
+			npurge_lb = npurge;
+		}
+		assert(n_search++ < lg_floor(SMOOTHSTEP_NSTEPS) + 1);
+	}
+	interval = decay_interval_ns * (ub + lb) / 2;
+label_done:
+	interval = (interval < BACKGROUND_THREAD_MIN_INTERVAL_NS) ?
+	    BACKGROUND_THREAD_MIN_INTERVAL_NS : interval;
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+
+	return interval;
+}
+
+/* Compute purge interval for background threads. */
+static uint64_t
+arena_decay_compute_purge_interval(tsdn_t *tsdn, arena_t *arena) {
+	uint64_t i1, i2;
+	i1 = arena_decay_compute_purge_interval_impl(tsdn, &arena->decay_dirty,
+	    &arena->extents_dirty);
+	if (i1 == BACKGROUND_THREAD_MIN_INTERVAL_NS) {
+		return i1;
+	}
+	i2 = arena_decay_compute_purge_interval_impl(tsdn, &arena->decay_muzzy,
+	    &arena->extents_muzzy);
+
+	return i1 < i2 ? i1 : i2;
+}
+
+static void
+background_thread_sleep(tsdn_t *tsdn, background_thread_info_t *info,
+    uint64_t interval) {
+	if (config_stats) {
+		info->tot_n_runs++;
+	}
+	info->npages_to_purge_new = 0;
+
+	struct timeval tv;
+	/* Specific clock required by timedwait. */
+	gettimeofday(&tv, NULL);
+	nstime_t before_sleep;
+	nstime_init2(&before_sleep, tv.tv_sec, tv.tv_usec * 1000);
+
+	int ret;
+	if (interval == BACKGROUND_THREAD_INDEFINITE_SLEEP) {
+		assert(background_thread_indefinite_sleep(info));
+		ret = pthread_cond_wait(&info->cond, &info->mtx.lock);
+		assert(ret == 0);
+	} else {
+		assert(interval >= BACKGROUND_THREAD_MIN_INTERVAL_NS &&
+		    interval <= BACKGROUND_THREAD_INDEFINITE_SLEEP);
+		/* We need malloc clock (can be different from tv). */
+		nstime_t next_wakeup;
+		nstime_init(&next_wakeup, 0);
+		nstime_update(&next_wakeup);
+		nstime_iadd(&next_wakeup, interval);
+		assert(nstime_ns(&next_wakeup) <
+		    BACKGROUND_THREAD_INDEFINITE_SLEEP);
+		background_thread_wakeup_time_set(tsdn, info,
+		    nstime_ns(&next_wakeup));
+
+		nstime_t ts_wakeup;
+		nstime_copy(&ts_wakeup, &before_sleep);
+		nstime_iadd(&ts_wakeup, interval);
+		struct timespec ts;
+		ts.tv_sec = (size_t)nstime_sec(&ts_wakeup);
+		ts.tv_nsec = (size_t)nstime_nsec(&ts_wakeup);
+
+		assert(!background_thread_indefinite_sleep(info));
+		ret = pthread_cond_timedwait(&info->cond, &info->mtx.lock, &ts);
+		assert(ret == ETIMEDOUT || ret == 0);
+		background_thread_wakeup_time_set(tsdn, info,
+		    BACKGROUND_THREAD_INDEFINITE_SLEEP);
+	}
+	if (config_stats) {
+		gettimeofday(&tv, NULL);
+		nstime_t after_sleep;
+		nstime_init2(&after_sleep, tv.tv_sec, tv.tv_usec * 1000);
+		if (nstime_compare(&after_sleep, &before_sleep) > 0) {
+			nstime_subtract(&after_sleep, &before_sleep);
+			nstime_add(&info->tot_sleep_time, &after_sleep);
+		}
+	}
+}
+
+static bool
+background_thread_pause_check(tsdn_t *tsdn, background_thread_info_t *info) {
+	if (unlikely(info->state == background_thread_paused)) {
+		malloc_mutex_unlock(tsdn, &info->mtx);
+		/* Wait on global lock to update status. */
+		malloc_mutex_lock(tsdn, &background_thread_lock);
+		malloc_mutex_unlock(tsdn, &background_thread_lock);
+		malloc_mutex_lock(tsdn, &info->mtx);
+		return true;
+	}
+
+	return false;
+}
+
+static inline void
+background_work_sleep_once(tsdn_t *tsdn, background_thread_info_t *info, unsigned ind) {
+	uint64_t min_interval = BACKGROUND_THREAD_INDEFINITE_SLEEP;
+	unsigned narenas = narenas_total_get();
+
+	for (unsigned i = ind; i < narenas; i += max_background_threads) {
+		arena_t *arena = arena_get(tsdn, i, false);
+		if (!arena) {
+			continue;
+		}
+		arena_decay(tsdn, arena, true, false);
+		if (min_interval == BACKGROUND_THREAD_MIN_INTERVAL_NS) {
+			/* Min interval will be used. */
+			continue;
+		}
+		uint64_t interval = arena_decay_compute_purge_interval(tsdn,
+		    arena);
+		assert(interval >= BACKGROUND_THREAD_MIN_INTERVAL_NS);
+		if (min_interval > interval) {
+			min_interval = interval;
+		}
+	}
+	background_thread_sleep(tsdn, info, min_interval);
+}
+
+static bool
+background_threads_disable_single(tsd_t *tsd, background_thread_info_t *info) {
+	if (info == &background_thread_info[0]) {
+		malloc_mutex_assert_owner(tsd_tsdn(tsd),
+		    &background_thread_lock);
+	} else {
+		malloc_mutex_assert_not_owner(tsd_tsdn(tsd),
+		    &background_thread_lock);
+	}
+
+	pre_reentrancy(tsd, NULL);
+	malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	bool has_thread;
+	assert(info->state != background_thread_paused);
+	if (info->state == background_thread_started) {
+		has_thread = true;
+		info->state = background_thread_stopped;
+		pthread_cond_signal(&info->cond);
+	} else {
+		has_thread = false;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+
+	if (!has_thread) {
+		post_reentrancy(tsd);
+		return false;
+	}
+	void *ret;
+	if (pthread_join(info->thread, &ret)) {
+		post_reentrancy(tsd);
+		return true;
+	}
+	assert(ret == NULL);
+	n_background_threads--;
+	post_reentrancy(tsd);
+
+	return false;
+}
+
+static void *background_thread_entry(void *ind_arg);
+
+static int
+background_thread_create_signals_masked(pthread_t *thread,
+    const pthread_attr_t *attr, void *(*start_routine)(void *), void *arg) {
+	/*
+	 * Mask signals during thread creation so that the thread inherits
+	 * an empty signal set.
+	 */
+	sigset_t set;
+	sigfillset(&set);
+	sigset_t oldset;
+	int mask_err = pthread_sigmask(SIG_SETMASK, &set, &oldset);
+	if (mask_err != 0) {
+		return mask_err;
+	}
+	int create_err = pthread_create_wrapper(thread, attr, start_routine,
+	    arg);
+	/*
+	 * Restore the signal mask.  Failure to restore the signal mask here
+	 * changes program behavior.
+	 */
+	int restore_err = pthread_sigmask(SIG_SETMASK, &oldset, NULL);
+	if (restore_err != 0) {
+		malloc_printf("<jemalloc>: background thread creation "
+		    "failed (%d), and signal mask restoration failed "
+		    "(%d)\n", create_err, restore_err);
+		if (opt_abort) {
+			abort();
+		}
+	}
+	return create_err;
+}
+
+static bool
+check_background_thread_creation(tsd_t *tsd, unsigned *n_created,
+    bool *created_threads) {
+	bool ret = false;
+	if (likely(*n_created == n_background_threads)) {
+		return ret;
+	}
+
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+	malloc_mutex_unlock(tsdn, &background_thread_info[0].mtx);
+	for (unsigned i = 1; i < max_background_threads; i++) {
+		if (created_threads[i]) {
+			continue;
+		}
+		background_thread_info_t *info = &background_thread_info[i];
+		malloc_mutex_lock(tsdn, &info->mtx);
+		/*
+		 * In case of the background_thread_paused state because of
+		 * arena reset, delay the creation.
+		 */
+		bool create = (info->state == background_thread_started);
+		malloc_mutex_unlock(tsdn, &info->mtx);
+		if (!create) {
+			continue;
+		}
+
+		pre_reentrancy(tsd, NULL);
+		int err = background_thread_create_signals_masked(&info->thread,
+		    NULL, background_thread_entry, (void *)(uintptr_t)i);
+		post_reentrancy(tsd);
+
+		if (err == 0) {
+			(*n_created)++;
+			created_threads[i] = true;
+		} else {
+			malloc_printf("<jemalloc>: background thread "
+			    "creation failed (%d)\n", err);
+			if (opt_abort) {
+				abort();
+			}
+		}
+		/* Return to restart the loop since we unlocked. */
+		ret = true;
+		break;
+	}
+	malloc_mutex_lock(tsdn, &background_thread_info[0].mtx);
+
+	return ret;
+}
+
+static void
+background_thread0_work(tsd_t *tsd) {
+	/* Thread0 is also responsible for launching / terminating threads. */
+	VARIABLE_ARRAY(bool, created_threads, max_background_threads);
+	unsigned i;
+	for (i = 1; i < max_background_threads; i++) {
+		created_threads[i] = false;
+	}
+	/* Start working, and create more threads when asked. */
+	unsigned n_created = 1;
+	while (background_thread_info[0].state != background_thread_stopped) {
+		if (background_thread_pause_check(tsd_tsdn(tsd),
+		    &background_thread_info[0])) {
+			continue;
+		}
+		if (check_background_thread_creation(tsd, &n_created,
+		    (bool *)&created_threads)) {
+			continue;
+		}
+		background_work_sleep_once(tsd_tsdn(tsd),
+		    &background_thread_info[0], 0);
+	}
+
+	/*
+	 * Shut down other threads at exit.  Note that the ctl thread is holding
+	 * the global background_thread mutex (and is waiting) for us.
+	 */
+	assert(!background_thread_enabled());
+	for (i = 1; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		assert(info->state != background_thread_paused);
+		if (created_threads[i]) {
+			background_threads_disable_single(tsd, info);
+		} else {
+			malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+			if (info->state != background_thread_stopped) {
+				/* The thread was not created. */
+				assert(info->state ==
+				    background_thread_started);
+				n_background_threads--;
+				info->state = background_thread_stopped;
+			}
+			malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		}
+	}
+	background_thread_info[0].state = background_thread_stopped;
+	assert(n_background_threads == 1);
+}
+
+static void
+background_work(tsd_t *tsd, unsigned ind) {
+	background_thread_info_t *info = &background_thread_info[ind];
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	background_thread_wakeup_time_set(tsd_tsdn(tsd), info,
+	    BACKGROUND_THREAD_INDEFINITE_SLEEP);
+	if (ind == 0) {
+		background_thread0_work(tsd);
+	} else {
+		while (info->state != background_thread_stopped) {
+			if (background_thread_pause_check(tsd_tsdn(tsd),
+			    info)) {
+				continue;
+			}
+			background_work_sleep_once(tsd_tsdn(tsd), info, ind);
+		}
+	}
+	assert(info->state == background_thread_stopped);
+	background_thread_wakeup_time_set(tsd_tsdn(tsd), info, 0);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+}
+
+static void *
+background_thread_entry(void *ind_arg) {
+	unsigned thread_ind = (unsigned)(uintptr_t)ind_arg;
+	assert(thread_ind < max_background_threads);
+#ifdef JEMALLOC_HAVE_PTHREAD_SETNAME_NP
+	pthread_setname_np(pthread_self(), "jemalloc_bg_thd");
+#elif defined(__FreeBSD__)
+	pthread_set_name_np(pthread_self(), "jemalloc_bg_thd");
+#endif
+	if (opt_percpu_arena != percpu_arena_disabled) {
+		set_current_thread_affinity((int)thread_ind);
+	}
+	/*
+	 * Start periodic background work.  We use internal tsd which avoids
+	 * side effects, for example triggering new arena creation (which in
+	 * turn triggers another background thread creation).
+	 */
+	background_work(tsd_internal_fetch(), thread_ind);
+	assert(pthread_equal(pthread_self(),
+	    background_thread_info[thread_ind].thread));
+
+	return NULL;
+}
+
+static void
+background_thread_init(tsd_t *tsd, background_thread_info_t *info) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+	info->state = background_thread_started;
+	background_thread_info_init(tsd_tsdn(tsd), info);
+	n_background_threads++;
+}
+
+static bool
+background_thread_create_locked(tsd_t *tsd, unsigned arena_ind) {
+	assert(have_background_thread);
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+
+	/* We create at most NCPUs threads. */
+	size_t thread_ind = arena_ind % max_background_threads;
+	background_thread_info_t *info = &background_thread_info[thread_ind];
+
+	bool need_new_thread;
+	malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	need_new_thread = background_thread_enabled() &&
+	    (info->state == background_thread_stopped);
+	if (need_new_thread) {
+		background_thread_init(tsd, info);
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+	if (!need_new_thread) {
+		return false;
+	}
+	if (arena_ind != 0) {
+		/* Threads are created asynchronously by Thread 0. */
+		background_thread_info_t *t0 = &background_thread_info[0];
+		malloc_mutex_lock(tsd_tsdn(tsd), &t0->mtx);
+		assert(t0->state == background_thread_started);
+		pthread_cond_signal(&t0->cond);
+		malloc_mutex_unlock(tsd_tsdn(tsd), &t0->mtx);
+
+		return false;
+	}
+
+	pre_reentrancy(tsd, NULL);
+	/*
+	 * To avoid complications (besides reentrancy), create internal
+	 * background threads with the underlying pthread_create.
+	 */
+	int err = background_thread_create_signals_masked(&info->thread, NULL,
+	    background_thread_entry, (void *)thread_ind);
+	post_reentrancy(tsd);
+
+	if (err != 0) {
+		malloc_printf("<jemalloc>: arena 0 background thread creation "
+		    "failed (%d)\n", err);
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+		info->state = background_thread_stopped;
+		n_background_threads--;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+
+		return true;
+	}
+
+	return false;
+}
+
+/* Create a new background thread if needed. */
+bool
+background_thread_create(tsd_t *tsd, unsigned arena_ind) {
+	assert(have_background_thread);
+
+	bool ret;
+	malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+	ret = background_thread_create_locked(tsd, arena_ind);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+
+	return ret;
+}
+
+bool
+background_threads_enable(tsd_t *tsd) {
+	assert(n_background_threads == 0);
+	assert(background_thread_enabled());
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+
+	VARIABLE_ARRAY(bool, marked, max_background_threads);
+	unsigned i, nmarked;
+	for (i = 0; i < max_background_threads; i++) {
+		marked[i] = false;
+	}
+	nmarked = 0;
+	/* Thread 0 is required and created at the end. */
+	marked[0] = true;
+	/* Mark the threads we need to create for thread 0. */
+	unsigned n = narenas_total_get();
+	for (i = 1; i < n; i++) {
+		if (marked[i % max_background_threads] ||
+		    arena_get(tsd_tsdn(tsd), i, false) == NULL) {
+			continue;
+		}
+		background_thread_info_t *info = &background_thread_info[
+		    i % max_background_threads];
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+		assert(info->state == background_thread_stopped);
+		background_thread_init(tsd, info);
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		marked[i % max_background_threads] = true;
+		if (++nmarked == max_background_threads) {
+			break;
+		}
+	}
+
+	return background_thread_create_locked(tsd, 0);
+}
+
+bool
+background_threads_disable(tsd_t *tsd) {
+	assert(!background_thread_enabled());
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &background_thread_lock);
+
+	/* Thread 0 will be responsible for terminating other threads. */
+	if (background_threads_disable_single(tsd,
+	    &background_thread_info[0])) {
+		return true;
+	}
+	assert(n_background_threads == 0);
+
+	return false;
+}
+
+/* Check if we need to signal the background thread early. */
+void
+background_thread_interval_check(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, size_t npages_new) {
+	background_thread_info_t *info = arena_background_thread_info_get(
+	    arena);
+	if (malloc_mutex_trylock(tsdn, &info->mtx)) {
+		/*
+		 * Background thread may hold the mutex for a long period of
+		 * time.  We'd like to avoid the variance on application
+		 * threads.  So keep this non-blocking, and leave the work to a
+		 * future epoch.
+		 */
+		return;
+	}
+
+	if (info->state != background_thread_started) {
+		goto label_done;
+	}
+	if (malloc_mutex_trylock(tsdn, &decay->mtx)) {
+		goto label_done;
+	}
+
+	ssize_t decay_time = atomic_load_zd(&decay->time_ms, ATOMIC_RELAXED);
+	if (decay_time <= 0) {
+		/* Purging is eagerly done or disabled currently. */
+		goto label_done_unlock2;
+	}
+	uint64_t decay_interval_ns = nstime_ns(&decay->interval);
+	assert(decay_interval_ns > 0);
+
+	nstime_t diff;
+	nstime_init(&diff, background_thread_wakeup_time_get(info));
+	if (nstime_compare(&diff, &decay->epoch) <= 0) {
+		goto label_done_unlock2;
+	}
+	nstime_subtract(&diff, &decay->epoch);
+	if (nstime_ns(&diff) < BACKGROUND_THREAD_MIN_INTERVAL_NS) {
+		goto label_done_unlock2;
+	}
+
+	if (npages_new > 0) {
+		size_t n_epoch = (size_t)(nstime_ns(&diff) / decay_interval_ns);
+		/*
+		 * Compute how many new pages we would need to purge by the next
+		 * wakeup, which is used to determine if we should signal the
+		 * background thread.
+		 */
+		uint64_t npurge_new;
+		if (n_epoch >= SMOOTHSTEP_NSTEPS) {
+			npurge_new = npages_new;
+		} else {
+			uint64_t h_steps_max = h_steps[SMOOTHSTEP_NSTEPS - 1];
+			assert(h_steps_max >=
+			    h_steps[SMOOTHSTEP_NSTEPS - 1 - n_epoch]);
+			npurge_new = npages_new * (h_steps_max -
+			    h_steps[SMOOTHSTEP_NSTEPS - 1 - n_epoch]);
+			npurge_new >>= SMOOTHSTEP_BFP;
+		}
+		info->npages_to_purge_new += npurge_new;
+	}
+
+	bool should_signal;
+	if (info->npages_to_purge_new > BACKGROUND_THREAD_NPAGES_THRESHOLD) {
+		should_signal = true;
+	} else if (unlikely(background_thread_indefinite_sleep(info)) &&
+	    (extents_npages_get(&arena->extents_dirty) > 0 ||
+	    extents_npages_get(&arena->extents_muzzy) > 0 ||
+	    info->npages_to_purge_new > 0)) {
+		should_signal = true;
+	} else {
+		should_signal = false;
+	}
+
+	if (should_signal) {
+		info->npages_to_purge_new = 0;
+		pthread_cond_signal(&info->cond);
+	}
+label_done_unlock2:
+	malloc_mutex_unlock(tsdn, &decay->mtx);
+label_done:
+	malloc_mutex_unlock(tsdn, &info->mtx);
+}
+
+void
+background_thread_prefork0(tsdn_t *tsdn) {
+	malloc_mutex_prefork(tsdn, &background_thread_lock);
+	background_thread_enabled_at_fork = background_thread_enabled();
+}
+
+void
+background_thread_prefork1(tsdn_t *tsdn) {
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		malloc_mutex_prefork(tsdn, &background_thread_info[i].mtx);
+	}
+}
+
+void
+background_thread_postfork_parent(tsdn_t *tsdn) {
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		malloc_mutex_postfork_parent(tsdn,
+		    &background_thread_info[i].mtx);
+	}
+	malloc_mutex_postfork_parent(tsdn, &background_thread_lock);
+}
+
+void
+background_thread_postfork_child(tsdn_t *tsdn) {
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		malloc_mutex_postfork_child(tsdn,
+		    &background_thread_info[i].mtx);
+	}
+	malloc_mutex_postfork_child(tsdn, &background_thread_lock);
+	if (!background_thread_enabled_at_fork) {
+		return;
+	}
+
+	/* Clear background_thread state (reset to disabled for child). */
+	malloc_mutex_lock(tsdn, &background_thread_lock);
+	n_background_threads = 0;
+	background_thread_enabled_set(tsdn, false);
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		malloc_mutex_lock(tsdn, &info->mtx);
+		info->state = background_thread_stopped;
+		int ret = pthread_cond_init(&info->cond, NULL);
+		assert(ret == 0);
+		background_thread_info_init(tsdn, info);
+		malloc_mutex_unlock(tsdn, &info->mtx);
+	}
+	malloc_mutex_unlock(tsdn, &background_thread_lock);
+}
+
+bool
+background_thread_stats_read(tsdn_t *tsdn, background_thread_stats_t *stats) {
+	assert(config_stats);
+	malloc_mutex_lock(tsdn, &background_thread_lock);
+	if (!background_thread_enabled()) {
+		malloc_mutex_unlock(tsdn, &background_thread_lock);
+		return true;
+	}
+
+	stats->num_threads = n_background_threads;
+	uint64_t num_runs = 0;
+	nstime_init(&stats->run_interval, 0);
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		if (malloc_mutex_trylock(tsdn, &info->mtx)) {
+			/*
+			 * Each background thread run may take a long time;
+			 * avoid waiting on the stats if the thread is active.
+			 */
+			continue;
+		}
+		if (info->state != background_thread_stopped) {
+			num_runs += info->tot_n_runs;
+			nstime_add(&stats->run_interval, &info->tot_sleep_time);
+		}
+		malloc_mutex_unlock(tsdn, &info->mtx);
+	}
+	stats->num_runs = num_runs;
+	if (num_runs > 0) {
+		nstime_idivide(&stats->run_interval, num_runs);
+	}
+	malloc_mutex_unlock(tsdn, &background_thread_lock);
+
+	return false;
+}
+
+#undef BACKGROUND_THREAD_NPAGES_THRESHOLD
+#undef BILLION
+#undef BACKGROUND_THREAD_MIN_INTERVAL_NS
+
+#ifdef JEMALLOC_HAVE_DLSYM
+#include <dlfcn.h>
+#endif
+
+static bool
+pthread_create_fptr_init(void) {
+	if (pthread_create_fptr != NULL) {
+		return false;
+	}
+	/*
+	 * Try the next symbol first, because 1) when use lazy_lock we have a
+	 * wrapper for pthread_create; and 2) application may define its own
+	 * wrapper as well (and can call malloc within the wrapper).
+	 */
+#ifdef JEMALLOC_HAVE_DLSYM
+	pthread_create_fptr = dlsym(RTLD_NEXT, "pthread_create");
+#else
+	pthread_create_fptr = NULL;
+#endif
+	if (pthread_create_fptr == NULL) {
+		if (config_lazy_lock) {
+			malloc_write("<jemalloc>: Error in dlsym(RTLD_NEXT, "
+			    "\"pthread_create\")\n");
+			abort();
+		} else {
+			/* Fall back to the default symbol. */
+			pthread_create_fptr = pthread_create;
+		}
+	}
+
+	return false;
+}
+
+/*
+ * When lazy lock is enabled, we need to make sure setting isthreaded before
+ * taking any background_thread locks.  This is called early in ctl (instead of
+ * wait for the pthread_create calls to trigger) because the mutex is required
+ * before creating background threads.
+ */
+void
+background_thread_ctl_init(tsdn_t *tsdn) {
+	malloc_mutex_assert_not_owner(tsdn, &background_thread_lock);
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+	pthread_create_fptr_init();
+	pthread_create_wrapper_init();
+#endif
+}
+
+#endif /* defined(JEMALLOC_BACKGROUND_THREAD) */
+
+bool
+background_thread_boot0(void) {
+	if (!have_background_thread && opt_background_thread) {
+		malloc_printf("<jemalloc>: option background_thread currently "
+		    "supports pthread only\n");
+		return true;
+	}
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+	if ((config_lazy_lock || opt_background_thread) &&
+	    pthread_create_fptr_init()) {
+		return true;
+	}
+#endif
+	return false;
+}
+
+bool
+background_thread_boot1(tsdn_t *tsdn) {
+#ifdef JEMALLOC_BACKGROUND_THREAD
+	assert(have_background_thread);
+	assert(narenas_total_get() > 0);
+
+	if (opt_max_background_threads > MAX_BACKGROUND_THREAD_LIMIT) {
+		opt_max_background_threads = DEFAULT_NUM_BACKGROUND_THREAD;
+	}
+	max_background_threads = opt_max_background_threads;
+
+	background_thread_enabled_set(tsdn, opt_background_thread);
+	if (malloc_mutex_init(&background_thread_lock,
+	    "background_thread_global",
+	    WITNESS_RANK_BACKGROUND_THREAD_GLOBAL,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	background_thread_info = (background_thread_info_t *)base_alloc(tsdn,
+	    b0get(), opt_max_background_threads *
+	    sizeof(background_thread_info_t), CACHELINE);
+	if (background_thread_info == NULL) {
+		return true;
+	}
+
+	for (unsigned i = 0; i < max_background_threads; i++) {
+		background_thread_info_t *info = &background_thread_info[i];
+		/* Thread mutex is rank_inclusive because of thread0. */
+		if (malloc_mutex_init(&info->mtx, "background_thread",
+		    WITNESS_RANK_BACKGROUND_THREAD,
+		    malloc_mutex_address_ordered)) {
+			return true;
+		}
+		if (pthread_cond_init(&info->cond, NULL)) {
+			return true;
+		}
+		malloc_mutex_lock(tsdn, &info->mtx);
+		info->state = background_thread_stopped;
+		background_thread_info_init(tsdn, info);
+		malloc_mutex_unlock(tsdn, &info->mtx);
+	}
+#endif
+
+	return false;
+}
diff --git a/memkind-1.10.0/jemalloc/src/base.c b/memkind-1.10.0/jemalloc/src/base.c
new file mode 100644
index 0000000..f3c6166
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/base.c
@@ -0,0 +1,514 @@
+#define JEMALLOC_BASE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/sz.h"
+
+/******************************************************************************/
+/* Data. */
+
+static base_t *b0;
+
+metadata_thp_mode_t opt_metadata_thp = METADATA_THP_DEFAULT;
+
+const char *metadata_thp_mode_names[] = {
+	"disabled",
+	"auto",
+	"always"
+};
+
+/******************************************************************************/
+
+static inline bool
+metadata_thp_madvise(void) {
+	return (metadata_thp_enabled() &&
+	    (init_system_thp_mode == thp_mode_default));
+}
+
+static void *
+base_map(tsdn_t *tsdn, extent_hooks_t *extent_hooks, unsigned ind, size_t size) {
+	void *addr;
+	bool zero = true;
+	bool commit = true;
+
+	/* Use huge page sizes and alignment regardless of opt_metadata_thp. */
+	assert(size == HUGEPAGE_CEILING(size));
+	size_t alignment = HUGEPAGE;
+	if (extent_hooks == &extent_hooks_default) {
+		addr = extent_alloc_mmap(NULL, size, alignment, &zero, &commit);
+	} else {
+		/* No arena context as we are creating new arenas. */
+		tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+		pre_reentrancy(tsd, NULL);
+		addr = extent_hooks->alloc(extent_hooks, NULL, size, alignment,
+		    &zero, &commit, ind);
+		post_reentrancy(tsd);
+	}
+
+	return addr;
+}
+
+static void
+base_unmap(tsdn_t *tsdn, extent_hooks_t *extent_hooks, unsigned ind, void *addr,
+    size_t size) {
+	/*
+	 * Cascade through dalloc, decommit, purge_forced, and purge_lazy,
+	 * stopping at first success.  This cascade is performed for consistency
+	 * with the cascade in extent_dalloc_wrapper() because an application's
+	 * custom hooks may not support e.g. dalloc.  This function is only ever
+	 * called as a side effect of arena destruction, so although it might
+	 * seem pointless to do anything besides dalloc here, the application
+	 * may in fact want the end state of all associated virtual memory to be
+	 * in some consistent-but-allocated state.
+	 */
+	if (extent_hooks == &extent_hooks_default) {
+		if (!extent_dalloc_mmap(addr, size)) {
+			goto label_done;
+		}
+		if (!pages_decommit(addr, size)) {
+			goto label_done;
+		}
+		if (!pages_purge_forced(addr, size)) {
+			goto label_done;
+		}
+		if (!pages_purge_lazy(addr, size)) {
+			goto label_done;
+		}
+		/* Nothing worked.  This should never happen. */
+		not_reached();
+	} else {
+		tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+		pre_reentrancy(tsd, NULL);
+		if (extent_hooks->dalloc != NULL &&
+		    !extent_hooks->dalloc(extent_hooks, addr, size, true,
+		    ind)) {
+			goto label_post_reentrancy;
+		}
+		if (extent_hooks->decommit != NULL &&
+		    !extent_hooks->decommit(extent_hooks, addr, size, 0, size,
+		    ind)) {
+			goto label_post_reentrancy;
+		}
+		if (extent_hooks->purge_forced != NULL &&
+		    !extent_hooks->purge_forced(extent_hooks, addr, size, 0,
+		    size, ind)) {
+			goto label_post_reentrancy;
+		}
+		if (extent_hooks->purge_lazy != NULL &&
+		    !extent_hooks->purge_lazy(extent_hooks, addr, size, 0, size,
+		    ind)) {
+			goto label_post_reentrancy;
+		}
+		/* Nothing worked.  That's the application's problem. */
+	label_post_reentrancy:
+		post_reentrancy(tsd);
+	}
+label_done:
+	if (metadata_thp_madvise()) {
+		/* Set NOHUGEPAGE after unmap to avoid kernel defrag. */
+		assert(((uintptr_t)addr & HUGEPAGE_MASK) == 0 &&
+		    (size & HUGEPAGE_MASK) == 0);
+		pages_nohuge(addr, size);
+	}
+}
+
+static void
+base_extent_init(size_t *extent_sn_next, extent_t *extent, void *addr,
+    size_t size) {
+	size_t sn;
+
+	sn = *extent_sn_next;
+	(*extent_sn_next)++;
+
+	extent_binit(extent, addr, size, sn);
+}
+
+static size_t
+base_get_num_blocks(base_t *base, bool with_new_block) {
+	base_block_t *b = base->blocks;
+	assert(b != NULL);
+
+	size_t n_blocks = with_new_block ? 2 : 1;
+	while (b->next != NULL) {
+		n_blocks++;
+		b = b->next;
+	}
+
+	return n_blocks;
+}
+
+static void
+base_auto_thp_switch(tsdn_t *tsdn, base_t *base) {
+	assert(opt_metadata_thp == metadata_thp_auto);
+	malloc_mutex_assert_owner(tsdn, &base->mtx);
+	if (base->auto_thp_switched) {
+		return;
+	}
+	/* Called when adding a new block. */
+	bool should_switch;
+	if (base_ind_get(base) != 0) {
+		should_switch = (base_get_num_blocks(base, true) ==
+		    BASE_AUTO_THP_THRESHOLD);
+	} else {
+		should_switch = (base_get_num_blocks(base, true) ==
+		    BASE_AUTO_THP_THRESHOLD_A0);
+	}
+	if (!should_switch) {
+		return;
+	}
+
+	base->auto_thp_switched = true;
+	assert(!config_stats || base->n_thp == 0);
+	/* Make the initial blocks THP lazily. */
+	base_block_t *block = base->blocks;
+	while (block != NULL) {
+		assert((block->size & HUGEPAGE_MASK) == 0);
+		pages_huge(block, block->size);
+		if (config_stats) {
+			base->n_thp += HUGEPAGE_CEILING(block->size -
+			    extent_bsize_get(&block->extent)) >> LG_HUGEPAGE;
+		}
+		block = block->next;
+		assert(block == NULL || (base_ind_get(base) == 0));
+	}
+}
+
+static void *
+base_extent_bump_alloc_helper(extent_t *extent, size_t *gap_size, size_t size,
+    size_t alignment) {
+	void *ret;
+
+	assert(alignment == ALIGNMENT_CEILING(alignment, QUANTUM));
+	assert(size == ALIGNMENT_CEILING(size, alignment));
+
+	*gap_size = ALIGNMENT_CEILING((uintptr_t)extent_addr_get(extent),
+	    alignment) - (uintptr_t)extent_addr_get(extent);
+	ret = (void *)((uintptr_t)extent_addr_get(extent) + *gap_size);
+	assert(extent_bsize_get(extent) >= *gap_size + size);
+	extent_binit(extent, (void *)((uintptr_t)extent_addr_get(extent) +
+	    *gap_size + size), extent_bsize_get(extent) - *gap_size - size,
+	    extent_sn_get(extent));
+	return ret;
+}
+
+static void
+base_extent_bump_alloc_post(base_t *base, extent_t *extent, size_t gap_size,
+    void *addr, size_t size) {
+	if (extent_bsize_get(extent) > 0) {
+		/*
+		 * Compute the index for the largest size class that does not
+		 * exceed extent's size.
+		 */
+		szind_t index_floor =
+		    sz_size2index(extent_bsize_get(extent) + 1) - 1;
+		extent_heap_insert(&base->avail[index_floor], extent);
+	}
+
+	if (config_stats) {
+		base->allocated += size;
+		/*
+		 * Add one PAGE to base_resident for every page boundary that is
+		 * crossed by the new allocation. Adjust n_thp similarly when
+		 * metadata_thp is enabled.
+		 */
+		base->resident += PAGE_CEILING((uintptr_t)addr + size) -
+		    PAGE_CEILING((uintptr_t)addr - gap_size);
+		assert(base->allocated <= base->resident);
+		assert(base->resident <= base->mapped);
+		if (metadata_thp_madvise() && (opt_metadata_thp ==
+		    metadata_thp_always || base->auto_thp_switched)) {
+			base->n_thp += (HUGEPAGE_CEILING((uintptr_t)addr + size)
+			    - HUGEPAGE_CEILING((uintptr_t)addr - gap_size)) >>
+			    LG_HUGEPAGE;
+			assert(base->mapped >= base->n_thp << LG_HUGEPAGE);
+		}
+	}
+}
+
+static void *
+base_extent_bump_alloc(base_t *base, extent_t *extent, size_t size,
+    size_t alignment) {
+	void *ret;
+	size_t gap_size;
+
+	ret = base_extent_bump_alloc_helper(extent, &gap_size, size, alignment);
+	base_extent_bump_alloc_post(base, extent, gap_size, ret, size);
+	return ret;
+}
+
+/*
+ * Allocate a block of virtual memory that is large enough to start with a
+ * base_block_t header, followed by an object of specified size and alignment.
+ * On success a pointer to the initialized base_block_t header is returned.
+ */
+static base_block_t *
+base_block_alloc(tsdn_t *tsdn, base_t *base, extent_hooks_t *extent_hooks,
+    unsigned ind, pszind_t *pind_last, size_t *extent_sn_next, size_t size,
+    size_t alignment) {
+	alignment = ALIGNMENT_CEILING(alignment, QUANTUM);
+	size_t usize = ALIGNMENT_CEILING(size, alignment);
+	size_t header_size = sizeof(base_block_t);
+	size_t gap_size = ALIGNMENT_CEILING(header_size, alignment) -
+	    header_size;
+	/*
+	 * Create increasingly larger blocks in order to limit the total number
+	 * of disjoint virtual memory ranges.  Choose the next size in the page
+	 * size class series (skipping size classes that are not a multiple of
+	 * HUGEPAGE), or a size large enough to satisfy the requested size and
+	 * alignment, whichever is larger.
+	 */
+	size_t min_block_size = HUGEPAGE_CEILING(sz_psz2u(header_size + gap_size
+	    + usize));
+	pszind_t pind_next = (*pind_last + 1 < sz_psz2ind(SC_LARGE_MAXCLASS)) ?
+	    *pind_last + 1 : *pind_last;
+	size_t next_block_size = HUGEPAGE_CEILING(sz_pind2sz(pind_next));
+	size_t block_size = (min_block_size > next_block_size) ? min_block_size
+	    : next_block_size;
+	base_block_t *block = (base_block_t *)base_map(tsdn, extent_hooks, ind,
+	    block_size);
+	if (block == NULL) {
+		return NULL;
+	}
+
+	if (metadata_thp_madvise()) {
+		void *addr = (void *)block;
+		assert(((uintptr_t)addr & HUGEPAGE_MASK) == 0 &&
+		    (block_size & HUGEPAGE_MASK) == 0);
+		if (opt_metadata_thp == metadata_thp_always) {
+			pages_huge(addr, block_size);
+		} else if (opt_metadata_thp == metadata_thp_auto &&
+		    base != NULL) {
+			/* base != NULL indicates this is not a new base. */
+			malloc_mutex_lock(tsdn, &base->mtx);
+			base_auto_thp_switch(tsdn, base);
+			if (base->auto_thp_switched) {
+				pages_huge(addr, block_size);
+			}
+			malloc_mutex_unlock(tsdn, &base->mtx);
+		}
+	}
+
+	*pind_last = sz_psz2ind(block_size);
+	block->size = block_size;
+	block->next = NULL;
+	assert(block_size >= header_size);
+	base_extent_init(extent_sn_next, &block->extent,
+	    (void *)((uintptr_t)block + header_size), block_size - header_size);
+	return block;
+}
+
+/*
+ * Allocate an extent that is at least as large as specified size, with
+ * specified alignment.
+ */
+static extent_t *
+base_extent_alloc(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment) {
+	malloc_mutex_assert_owner(tsdn, &base->mtx);
+
+	extent_hooks_t *extent_hooks = base_extent_hooks_get(base);
+	/*
+	 * Drop mutex during base_block_alloc(), because an extent hook will be
+	 * called.
+	 */
+	malloc_mutex_unlock(tsdn, &base->mtx);
+	base_block_t *block = base_block_alloc(tsdn, base, extent_hooks,
+	    base_ind_get(base), &base->pind_last, &base->extent_sn_next, size,
+	    alignment);
+	malloc_mutex_lock(tsdn, &base->mtx);
+	if (block == NULL) {
+		return NULL;
+	}
+	block->next = base->blocks;
+	base->blocks = block;
+	if (config_stats) {
+		base->allocated += sizeof(base_block_t);
+		base->resident += PAGE_CEILING(sizeof(base_block_t));
+		base->mapped += block->size;
+		if (metadata_thp_madvise() &&
+		    !(opt_metadata_thp == metadata_thp_auto
+		      && !base->auto_thp_switched)) {
+			assert(base->n_thp > 0);
+			base->n_thp += HUGEPAGE_CEILING(sizeof(base_block_t)) >>
+			    LG_HUGEPAGE;
+		}
+		assert(base->allocated <= base->resident);
+		assert(base->resident <= base->mapped);
+		assert(base->n_thp << LG_HUGEPAGE <= base->mapped);
+	}
+	return &block->extent;
+}
+
+base_t *
+b0get(void) {
+	return b0;
+}
+
+base_t *
+base_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	pszind_t pind_last = 0;
+	size_t extent_sn_next = 0;
+	base_block_t *block = base_block_alloc(tsdn, NULL, extent_hooks, ind,
+	    &pind_last, &extent_sn_next, sizeof(base_t), QUANTUM);
+	if (block == NULL) {
+		return NULL;
+	}
+
+	size_t gap_size;
+	size_t base_alignment = CACHELINE;
+	size_t base_size = ALIGNMENT_CEILING(sizeof(base_t), base_alignment);
+	base_t *base = (base_t *)base_extent_bump_alloc_helper(&block->extent,
+	    &gap_size, base_size, base_alignment);
+	base->ind = ind;
+	atomic_store_p(&base->extent_hooks, extent_hooks, ATOMIC_RELAXED);
+	if (malloc_mutex_init(&base->mtx, "base", WITNESS_RANK_BASE,
+	    malloc_mutex_rank_exclusive)) {
+		base_unmap(tsdn, extent_hooks, ind, block, block->size);
+		return NULL;
+	}
+	base->pind_last = pind_last;
+	base->extent_sn_next = extent_sn_next;
+	base->blocks = block;
+	base->auto_thp_switched = false;
+	for (szind_t i = 0; i < SC_NSIZES; i++) {
+		extent_heap_new(&base->avail[i]);
+	}
+	if (config_stats) {
+		base->allocated = sizeof(base_block_t);
+		base->resident = PAGE_CEILING(sizeof(base_block_t));
+		base->mapped = block->size;
+		base->n_thp = (opt_metadata_thp == metadata_thp_always) &&
+		    metadata_thp_madvise() ? HUGEPAGE_CEILING(sizeof(base_block_t))
+		    >> LG_HUGEPAGE : 0;
+		assert(base->allocated <= base->resident);
+		assert(base->resident <= base->mapped);
+		assert(base->n_thp << LG_HUGEPAGE <= base->mapped);
+	}
+	base_extent_bump_alloc_post(base, &block->extent, gap_size, base,
+	    base_size);
+
+	return base;
+}
+
+void
+base_delete(tsdn_t *tsdn, base_t *base) {
+	extent_hooks_t *extent_hooks = base_extent_hooks_get(base);
+	base_block_t *next = base->blocks;
+	do {
+		base_block_t *block = next;
+		next = block->next;
+		base_unmap(tsdn, extent_hooks, base_ind_get(base), block,
+		    block->size);
+	} while (next != NULL);
+}
+
+extent_hooks_t *
+base_extent_hooks_get(base_t *base) {
+	return (extent_hooks_t *)atomic_load_p(&base->extent_hooks,
+	    ATOMIC_ACQUIRE);
+}
+
+extent_hooks_t *
+base_extent_hooks_set(base_t *base, extent_hooks_t *extent_hooks) {
+	extent_hooks_t *old_extent_hooks = base_extent_hooks_get(base);
+	atomic_store_p(&base->extent_hooks, extent_hooks, ATOMIC_RELEASE);
+	return old_extent_hooks;
+}
+
+static void *
+base_alloc_impl(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment,
+    size_t *esn) {
+	alignment = QUANTUM_CEILING(alignment);
+	size_t usize = ALIGNMENT_CEILING(size, alignment);
+	size_t asize = usize + alignment - QUANTUM;
+
+	extent_t *extent = NULL;
+	malloc_mutex_lock(tsdn, &base->mtx);
+	for (szind_t i = sz_size2index(asize); i < SC_NSIZES; i++) {
+		extent = extent_heap_remove_first(&base->avail[i]);
+		if (extent != NULL) {
+			/* Use existing space. */
+			break;
+		}
+	}
+	if (extent == NULL) {
+		/* Try to allocate more space. */
+		extent = base_extent_alloc(tsdn, base, usize, alignment);
+	}
+	void *ret;
+	if (extent == NULL) {
+		ret = NULL;
+		goto label_return;
+	}
+
+	ret = base_extent_bump_alloc(base, extent, usize, alignment);
+	if (esn != NULL) {
+		*esn = extent_sn_get(extent);
+	}
+label_return:
+	malloc_mutex_unlock(tsdn, &base->mtx);
+	return ret;
+}
+
+/*
+ * base_alloc() returns zeroed memory, which is always demand-zeroed for the
+ * auto arenas, in order to make multi-page sparse data structures such as radix
+ * tree nodes efficient with respect to physical memory usage.  Upon success a
+ * pointer to at least size bytes with specified alignment is returned.  Note
+ * that size is rounded up to the nearest multiple of alignment to avoid false
+ * sharing.
+ */
+void *
+base_alloc(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment) {
+	return base_alloc_impl(tsdn, base, size, alignment, NULL);
+}
+
+extent_t *
+base_alloc_extent(tsdn_t *tsdn, base_t *base) {
+	size_t esn;
+	extent_t *extent = base_alloc_impl(tsdn, base, sizeof(extent_t),
+	    CACHELINE, &esn);
+	if (extent == NULL) {
+		return NULL;
+	}
+	extent_esn_set(extent, esn);
+	return extent;
+}
+
+void
+base_stats_get(tsdn_t *tsdn, base_t *base, size_t *allocated, size_t *resident,
+    size_t *mapped, size_t *n_thp) {
+	cassert(config_stats);
+
+	malloc_mutex_lock(tsdn, &base->mtx);
+	assert(base->allocated <= base->resident);
+	assert(base->resident <= base->mapped);
+	*allocated = base->allocated;
+	*resident = base->resident;
+	*mapped = base->mapped;
+	*n_thp = base->n_thp;
+	malloc_mutex_unlock(tsdn, &base->mtx);
+}
+
+void
+base_prefork(tsdn_t *tsdn, base_t *base) {
+	malloc_mutex_prefork(tsdn, &base->mtx);
+}
+
+void
+base_postfork_parent(tsdn_t *tsdn, base_t *base) {
+	malloc_mutex_postfork_parent(tsdn, &base->mtx);
+}
+
+void
+base_postfork_child(tsdn_t *tsdn, base_t *base) {
+	malloc_mutex_postfork_child(tsdn, &base->mtx);
+}
+
+bool
+base_boot(tsdn_t *tsdn) {
+	b0 = base_new(tsdn, 0, (extent_hooks_t *)&extent_hooks_default);
+	return (b0 == NULL);
+}
diff --git a/memkind-1.10.0/jemalloc/src/bin.c b/memkind-1.10.0/jemalloc/src/bin.c
new file mode 100644
index 0000000..bca6b12
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/bin.c
@@ -0,0 +1,95 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/witness.h"
+
+bin_info_t bin_infos[SC_NBINS];
+
+static void
+bin_infos_init(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS],
+    bin_info_t bin_infos[SC_NBINS]) {
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		bin_info_t *bin_info = &bin_infos[i];
+		sc_t *sc = &sc_data->sc[i];
+		bin_info->reg_size = ((size_t)1U << sc->lg_base)
+		    + ((size_t)sc->ndelta << sc->lg_delta);
+		bin_info->slab_size = (sc->pgs << LG_PAGE);
+		bin_info->nregs =
+		    (uint32_t)(bin_info->slab_size / bin_info->reg_size);
+		bin_info->n_shards = bin_shard_sizes[i];
+		bitmap_info_t bitmap_info = BITMAP_INFO_INITIALIZER(
+		    bin_info->nregs);
+		bin_info->bitmap_info = bitmap_info;
+	}
+}
+
+bool
+bin_update_shard_size(unsigned bin_shard_sizes[SC_NBINS], size_t start_size,
+    size_t end_size, size_t nshards) {
+	if (nshards > BIN_SHARDS_MAX || nshards == 0) {
+		return true;
+	}
+
+	if (start_size > SC_SMALL_MAXCLASS) {
+		return false;
+	}
+	if (end_size > SC_SMALL_MAXCLASS) {
+		end_size = SC_SMALL_MAXCLASS;
+	}
+
+	/* Compute the index since this may happen before sz init. */
+	szind_t ind1 = sz_size2index_compute(start_size);
+	szind_t ind2 = sz_size2index_compute(end_size);
+	for (unsigned i = ind1; i <= ind2; i++) {
+		bin_shard_sizes[i] = (unsigned)nshards;
+	}
+
+	return false;
+}
+
+void
+bin_shard_sizes_boot(unsigned bin_shard_sizes[SC_NBINS]) {
+	/* Load the default number of shards. */
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		bin_shard_sizes[i] = N_BIN_SHARDS_DEFAULT;
+	}
+}
+
+void
+bin_boot(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS]) {
+	assert(sc_data->initialized);
+	bin_infos_init(sc_data, bin_shard_sizes, bin_infos);
+}
+
+bool
+bin_init(bin_t *bin) {
+	if (malloc_mutex_init(&bin->lock, "bin", WITNESS_RANK_BIN,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	bin->slabcur = NULL;
+	extent_heap_new(&bin->slabs_nonfull);
+	extent_list_init(&bin->slabs_full);
+	if (config_stats) {
+		memset(&bin->stats, 0, sizeof(bin_stats_t));
+	}
+	return false;
+}
+
+void
+bin_prefork(tsdn_t *tsdn, bin_t *bin) {
+	malloc_mutex_prefork(tsdn, &bin->lock);
+}
+
+void
+bin_postfork_parent(tsdn_t *tsdn, bin_t *bin) {
+	malloc_mutex_postfork_parent(tsdn, &bin->lock);
+}
+
+void
+bin_postfork_child(tsdn_t *tsdn, bin_t *bin) {
+	malloc_mutex_postfork_child(tsdn, &bin->lock);
+}
diff --git a/memkind-1.10.0/jemalloc/src/bitmap.c b/memkind-1.10.0/jemalloc/src/bitmap.c
new file mode 100644
index 0000000..468b317
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/bitmap.c
@@ -0,0 +1,121 @@
+#define JEMALLOC_BITMAP_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+
+/******************************************************************************/
+
+#ifdef BITMAP_USE_TREE
+
+void
+bitmap_info_init(bitmap_info_t *binfo, size_t nbits) {
+	unsigned i;
+	size_t group_count;
+
+	assert(nbits > 0);
+	assert(nbits <= (ZU(1) << LG_BITMAP_MAXBITS));
+
+	/*
+	 * Compute the number of groups necessary to store nbits bits, and
+	 * progressively work upward through the levels until reaching a level
+	 * that requires only one group.
+	 */
+	binfo->levels[0].group_offset = 0;
+	group_count = BITMAP_BITS2GROUPS(nbits);
+	for (i = 1; group_count > 1; i++) {
+		assert(i < BITMAP_MAX_LEVELS);
+		binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+		    + group_count;
+		group_count = BITMAP_BITS2GROUPS(group_count);
+	}
+	binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+	    + group_count;
+	assert(binfo->levels[i].group_offset <= BITMAP_GROUPS_MAX);
+	binfo->nlevels = i;
+	binfo->nbits = nbits;
+}
+
+static size_t
+bitmap_info_ngroups(const bitmap_info_t *binfo) {
+	return binfo->levels[binfo->nlevels].group_offset;
+}
+
+void
+bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo, bool fill) {
+	size_t extra;
+	unsigned i;
+
+	/*
+	 * Bits are actually inverted with regard to the external bitmap
+	 * interface.
+	 */
+
+	if (fill) {
+		/* The "filled" bitmap starts out with all 0 bits. */
+		memset(bitmap, 0, bitmap_size(binfo));
+		return;
+	}
+
+	/*
+	 * The "empty" bitmap starts out with all 1 bits, except for trailing
+	 * unused bits (if any).  Note that each group uses bit 0 to correspond
+	 * to the first logical bit in the group, so extra bits are the most
+	 * significant bits of the last group.
+	 */
+	memset(bitmap, 0xffU, bitmap_size(binfo));
+	extra = (BITMAP_GROUP_NBITS - (binfo->nbits & BITMAP_GROUP_NBITS_MASK))
+	    & BITMAP_GROUP_NBITS_MASK;
+	if (extra != 0) {
+		bitmap[binfo->levels[1].group_offset - 1] >>= extra;
+	}
+	for (i = 1; i < binfo->nlevels; i++) {
+		size_t group_count = binfo->levels[i].group_offset -
+		    binfo->levels[i-1].group_offset;
+		extra = (BITMAP_GROUP_NBITS - (group_count &
+		    BITMAP_GROUP_NBITS_MASK)) & BITMAP_GROUP_NBITS_MASK;
+		if (extra != 0) {
+			bitmap[binfo->levels[i+1].group_offset - 1] >>= extra;
+		}
+	}
+}
+
+#else /* BITMAP_USE_TREE */
+
+void
+bitmap_info_init(bitmap_info_t *binfo, size_t nbits) {
+	assert(nbits > 0);
+	assert(nbits <= (ZU(1) << LG_BITMAP_MAXBITS));
+
+	binfo->ngroups = BITMAP_BITS2GROUPS(nbits);
+	binfo->nbits = nbits;
+}
+
+static size_t
+bitmap_info_ngroups(const bitmap_info_t *binfo) {
+	return binfo->ngroups;
+}
+
+void
+bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo, bool fill) {
+	size_t extra;
+
+	if (fill) {
+		memset(bitmap, 0, bitmap_size(binfo));
+		return;
+	}
+
+	memset(bitmap, 0xffU, bitmap_size(binfo));
+	extra = (BITMAP_GROUP_NBITS - (binfo->nbits & BITMAP_GROUP_NBITS_MASK))
+	    & BITMAP_GROUP_NBITS_MASK;
+	if (extra != 0) {
+		bitmap[binfo->ngroups - 1] >>= extra;
+	}
+}
+
+#endif /* BITMAP_USE_TREE */
+
+size_t
+bitmap_size(const bitmap_info_t *binfo) {
+	return (bitmap_info_ngroups(binfo) << LG_SIZEOF_BITMAP);
+}
diff --git a/memkind-1.10.0/jemalloc/src/ckh.c b/memkind-1.10.0/jemalloc/src/ckh.c
new file mode 100644
index 0000000..1bf6df5
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/ckh.c
@@ -0,0 +1,570 @@
+/*
+ *******************************************************************************
+ * Implementation of (2^1+,2) cuckoo hashing, where 2^1+ indicates that each
+ * hash bucket contains 2^n cells, for n >= 1, and 2 indicates that two hash
+ * functions are employed.  The original cuckoo hashing algorithm was described
+ * in:
+ *
+ *   Pagh, R., F.F. Rodler (2004) Cuckoo Hashing.  Journal of Algorithms
+ *     51(2):122-144.
+ *
+ * Generalization of cuckoo hashing was discussed in:
+ *
+ *   Erlingsson, U., M. Manasse, F. McSherry (2006) A cool and practical
+ *     alternative to traditional hash tables.  In Proceedings of the 7th
+ *     Workshop on Distributed Data and Structures (WDAS'06), Santa Clara, CA,
+ *     January 2006.
+ *
+ * This implementation uses precisely two hash functions because that is the
+ * fewest that can work, and supporting multiple hashes is an implementation
+ * burden.  Here is a reproduction of Figure 1 from Erlingsson et al. (2006)
+ * that shows approximate expected maximum load factors for various
+ * configurations:
+ *
+ *           |         #cells/bucket         |
+ *   #hashes |   1   |   2   |   4   |   8   |
+ *   --------+-------+-------+-------+-------+
+ *         1 | 0.006 | 0.006 | 0.03  | 0.12  |
+ *         2 | 0.49  | 0.86  |>0.93< |>0.96< |
+ *         3 | 0.91  | 0.97  | 0.98  | 0.999 |
+ *         4 | 0.97  | 0.99  | 0.999 |       |
+ *
+ * The number of cells per bucket is chosen such that a bucket fits in one cache
+ * line.  So, on 32- and 64-bit systems, we use (8,2) and (4,2) cuckoo hashing,
+ * respectively.
+ *
+ ******************************************************************************/
+#define JEMALLOC_CKH_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/ckh.h"
+
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static bool	ckh_grow(tsd_t *tsd, ckh_t *ckh);
+static void	ckh_shrink(tsd_t *tsd, ckh_t *ckh);
+
+/******************************************************************************/
+
+/*
+ * Search bucket for key and return the cell number if found; SIZE_T_MAX
+ * otherwise.
+ */
+static size_t
+ckh_bucket_search(ckh_t *ckh, size_t bucket, const void *key) {
+	ckhc_t *cell;
+	unsigned i;
+
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		if (cell->key != NULL && ckh->keycomp(key, cell->key)) {
+			return (bucket << LG_CKH_BUCKET_CELLS) + i;
+		}
+	}
+
+	return SIZE_T_MAX;
+}
+
+/*
+ * Search table for key and return cell number if found; SIZE_T_MAX otherwise.
+ */
+static size_t
+ckh_isearch(ckh_t *ckh, const void *key) {
+	size_t hashes[2], bucket, cell;
+
+	assert(ckh != NULL);
+
+	ckh->hash(key, hashes);
+
+	/* Search primary bucket. */
+	bucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	if (cell != SIZE_T_MAX) {
+		return cell;
+	}
+
+	/* Search secondary bucket. */
+	bucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	return cell;
+}
+
+static bool
+ckh_try_bucket_insert(ckh_t *ckh, size_t bucket, const void *key,
+    const void *data) {
+	ckhc_t *cell;
+	unsigned offset, i;
+
+	/*
+	 * Cycle through the cells in the bucket, starting at a random position.
+	 * The randomness avoids worst-case search overhead as buckets fill up.
+	 */
+	offset = (unsigned)prng_lg_range_u64(&ckh->prng_state,
+	    LG_CKH_BUCKET_CELLS);
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) +
+		    ((i + offset) & ((ZU(1) << LG_CKH_BUCKET_CELLS) - 1))];
+		if (cell->key == NULL) {
+			cell->key = key;
+			cell->data = data;
+			ckh->count++;
+			return false;
+		}
+	}
+
+	return true;
+}
+
+/*
+ * No space is available in bucket.  Randomly evict an item, then try to find an
+ * alternate location for that item.  Iteratively repeat this
+ * eviction/relocation procedure until either success or detection of an
+ * eviction/relocation bucket cycle.
+ */
+static bool
+ckh_evict_reloc_insert(ckh_t *ckh, size_t argbucket, void const **argkey,
+    void const **argdata) {
+	const void *key, *data, *tkey, *tdata;
+	ckhc_t *cell;
+	size_t hashes[2], bucket, tbucket;
+	unsigned i;
+
+	bucket = argbucket;
+	key = *argkey;
+	data = *argdata;
+	while (true) {
+		/*
+		 * Choose a random item within the bucket to evict.  This is
+		 * critical to correct function, because without (eventually)
+		 * evicting all items within a bucket during iteration, it
+		 * would be possible to get stuck in an infinite loop if there
+		 * were an item for which both hashes indicated the same
+		 * bucket.
+		 */
+		i = (unsigned)prng_lg_range_u64(&ckh->prng_state,
+		    LG_CKH_BUCKET_CELLS);
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		assert(cell->key != NULL);
+
+		/* Swap cell->{key,data} and {key,data} (evict). */
+		tkey = cell->key; tdata = cell->data;
+		cell->key = key; cell->data = data;
+		key = tkey; data = tdata;
+
+#ifdef CKH_COUNT
+		ckh->nrelocs++;
+#endif
+
+		/* Find the alternate bucket for the evicted item. */
+		ckh->hash(key, hashes);
+		tbucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+		if (tbucket == bucket) {
+			tbucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets)
+			    - 1);
+			/*
+			 * It may be that (tbucket == bucket) still, if the
+			 * item's hashes both indicate this bucket.  However,
+			 * we are guaranteed to eventually escape this bucket
+			 * during iteration, assuming pseudo-random item
+			 * selection (true randomness would make infinite
+			 * looping a remote possibility).  The reason we can
+			 * never get trapped forever is that there are two
+			 * cases:
+			 *
+			 * 1) This bucket == argbucket, so we will quickly
+			 *    detect an eviction cycle and terminate.
+			 * 2) An item was evicted to this bucket from another,
+			 *    which means that at least one item in this bucket
+			 *    has hashes that indicate distinct buckets.
+			 */
+		}
+		/* Check for a cycle. */
+		if (tbucket == argbucket) {
+			*argkey = key;
+			*argdata = data;
+			return true;
+		}
+
+		bucket = tbucket;
+		if (!ckh_try_bucket_insert(ckh, bucket, key, data)) {
+			return false;
+		}
+	}
+}
+
+static bool
+ckh_try_insert(ckh_t *ckh, void const**argkey, void const**argdata) {
+	size_t hashes[2], bucket;
+	const void *key = *argkey;
+	const void *data = *argdata;
+
+	ckh->hash(key, hashes);
+
+	/* Try to insert in primary bucket. */
+	bucket = hashes[0] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (!ckh_try_bucket_insert(ckh, bucket, key, data)) {
+		return false;
+	}
+
+	/* Try to insert in secondary bucket. */
+	bucket = hashes[1] & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (!ckh_try_bucket_insert(ckh, bucket, key, data)) {
+		return false;
+	}
+
+	/*
+	 * Try to find a place for this item via iterative eviction/relocation.
+	 */
+	return ckh_evict_reloc_insert(ckh, bucket, argkey, argdata);
+}
+
+/*
+ * Try to rebuild the hash table from scratch by inserting all items from the
+ * old table into the new.
+ */
+static bool
+ckh_rebuild(ckh_t *ckh, ckhc_t *aTab) {
+	size_t count, i, nins;
+	const void *key, *data;
+
+	count = ckh->count;
+	ckh->count = 0;
+	for (i = nins = 0; nins < count; i++) {
+		if (aTab[i].key != NULL) {
+			key = aTab[i].key;
+			data = aTab[i].data;
+			if (ckh_try_insert(ckh, &key, &data)) {
+				ckh->count = count;
+				return true;
+			}
+			nins++;
+		}
+	}
+
+	return false;
+}
+
+static bool
+ckh_grow(tsd_t *tsd, ckh_t *ckh) {
+	bool ret;
+	ckhc_t *tab, *ttab;
+	unsigned lg_prevbuckets, lg_curcells;
+
+#ifdef CKH_COUNT
+	ckh->ngrows++;
+#endif
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table will have to be doubled more than once in order to create a
+	 * usable table.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS;
+	while (true) {
+		size_t usize;
+
+		lg_curcells++;
+		usize = sz_sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+		if (unlikely(usize == 0
+		    || usize > SC_LARGE_MAXCLASS)) {
+			ret = true;
+			goto label_return;
+		}
+		tab = (ckhc_t *)ipallocztm(tsd_tsdn(tsd), usize, CACHELINE,
+		    true, NULL, true, arena_ichoose(tsd, NULL));
+		if (tab == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		/* Swap in new table. */
+		ttab = ckh->tab;
+		ckh->tab = tab;
+		tab = ttab;
+		ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+		if (!ckh_rebuild(ckh, tab)) {
+			idalloctm(tsd_tsdn(tsd), tab, NULL, NULL, true, true);
+			break;
+		}
+
+		/* Rebuilding failed, so back out partially rebuilt table. */
+		idalloctm(tsd_tsdn(tsd), ckh->tab, NULL, NULL, true, true);
+		ckh->tab = tab;
+		ckh->lg_curbuckets = lg_prevbuckets;
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+static void
+ckh_shrink(tsd_t *tsd, ckh_t *ckh) {
+	ckhc_t *tab, *ttab;
+	size_t usize;
+	unsigned lg_prevbuckets, lg_curcells;
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table rebuild will fail.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS - 1;
+	usize = sz_sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		return;
+	}
+	tab = (ckhc_t *)ipallocztm(tsd_tsdn(tsd), usize, CACHELINE, true, NULL,
+	    true, arena_ichoose(tsd, NULL));
+	if (tab == NULL) {
+		/*
+		 * An OOM error isn't worth propagating, since it doesn't
+		 * prevent this or future operations from proceeding.
+		 */
+		return;
+	}
+	/* Swap in new table. */
+	ttab = ckh->tab;
+	ckh->tab = tab;
+	tab = ttab;
+	ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+	if (!ckh_rebuild(ckh, tab)) {
+		idalloctm(tsd_tsdn(tsd), tab, NULL, NULL, true, true);
+#ifdef CKH_COUNT
+		ckh->nshrinks++;
+#endif
+		return;
+	}
+
+	/* Rebuilding failed, so back out partially rebuilt table. */
+	idalloctm(tsd_tsdn(tsd), ckh->tab, NULL, NULL, true, true);
+	ckh->tab = tab;
+	ckh->lg_curbuckets = lg_prevbuckets;
+#ifdef CKH_COUNT
+	ckh->nshrinkfails++;
+#endif
+}
+
+bool
+ckh_new(tsd_t *tsd, ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
+    ckh_keycomp_t *keycomp) {
+	bool ret;
+	size_t mincells, usize;
+	unsigned lg_mincells;
+
+	assert(minitems > 0);
+	assert(hash != NULL);
+	assert(keycomp != NULL);
+
+#ifdef CKH_COUNT
+	ckh->ngrows = 0;
+	ckh->nshrinks = 0;
+	ckh->nshrinkfails = 0;
+	ckh->ninserts = 0;
+	ckh->nrelocs = 0;
+#endif
+	ckh->prng_state = 42; /* Value doesn't really matter. */
+	ckh->count = 0;
+
+	/*
+	 * Find the minimum power of 2 that is large enough to fit minitems
+	 * entries.  We are using (2+,2) cuckoo hashing, which has an expected
+	 * maximum load factor of at least ~0.86, so 0.75 is a conservative load
+	 * factor that will typically allow mincells items to fit without ever
+	 * growing the table.
+	 */
+	assert(LG_CKH_BUCKET_CELLS > 0);
+	mincells = ((minitems + (3 - (minitems % 3))) / 3) << 2;
+	for (lg_mincells = LG_CKH_BUCKET_CELLS;
+	    (ZU(1) << lg_mincells) < mincells;
+	    lg_mincells++) {
+		/* Do nothing. */
+	}
+	ckh->lg_minbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->lg_curbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->hash = hash;
+	ckh->keycomp = keycomp;
+
+	usize = sz_sa2u(sizeof(ckhc_t) << lg_mincells, CACHELINE);
+	if (unlikely(usize == 0 || usize > SC_LARGE_MAXCLASS)) {
+		ret = true;
+		goto label_return;
+	}
+	ckh->tab = (ckhc_t *)ipallocztm(tsd_tsdn(tsd), usize, CACHELINE, true,
+	    NULL, true, arena_ichoose(tsd, NULL));
+	if (ckh->tab == NULL) {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+void
+ckh_delete(tsd_t *tsd, ckh_t *ckh) {
+	assert(ckh != NULL);
+
+#ifdef CKH_VERBOSE
+	malloc_printf(
+	    "%s(%p): ngrows: %"FMTu64", nshrinks: %"FMTu64","
+	    " nshrinkfails: %"FMTu64", ninserts: %"FMTu64","
+	    " nrelocs: %"FMTu64"\n", __func__, ckh,
+	    (unsigned long long)ckh->ngrows,
+	    (unsigned long long)ckh->nshrinks,
+	    (unsigned long long)ckh->nshrinkfails,
+	    (unsigned long long)ckh->ninserts,
+	    (unsigned long long)ckh->nrelocs);
+#endif
+
+	idalloctm(tsd_tsdn(tsd), ckh->tab, NULL, NULL, true, true);
+	if (config_debug) {
+		memset(ckh, JEMALLOC_FREE_JUNK, sizeof(ckh_t));
+	}
+}
+
+size_t
+ckh_count(ckh_t *ckh) {
+	assert(ckh != NULL);
+
+	return ckh->count;
+}
+
+bool
+ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data) {
+	size_t i, ncells;
+
+	for (i = *tabind, ncells = (ZU(1) << (ckh->lg_curbuckets +
+	    LG_CKH_BUCKET_CELLS)); i < ncells; i++) {
+		if (ckh->tab[i].key != NULL) {
+			if (key != NULL) {
+				*key = (void *)ckh->tab[i].key;
+			}
+			if (data != NULL) {
+				*data = (void *)ckh->tab[i].data;
+			}
+			*tabind = i + 1;
+			return false;
+		}
+	}
+
+	return true;
+}
+
+bool
+ckh_insert(tsd_t *tsd, ckh_t *ckh, const void *key, const void *data) {
+	bool ret;
+
+	assert(ckh != NULL);
+	assert(ckh_search(ckh, key, NULL, NULL));
+
+#ifdef CKH_COUNT
+	ckh->ninserts++;
+#endif
+
+	while (ckh_try_insert(ckh, &key, &data)) {
+		if (ckh_grow(tsd, ckh)) {
+			ret = true;
+			goto label_return;
+		}
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+bool
+ckh_remove(tsd_t *tsd, ckh_t *ckh, const void *searchkey, void **key,
+    void **data) {
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL) {
+			*key = (void *)ckh->tab[cell].key;
+		}
+		if (data != NULL) {
+			*data = (void *)ckh->tab[cell].data;
+		}
+		ckh->tab[cell].key = NULL;
+		ckh->tab[cell].data = NULL; /* Not necessary. */
+
+		ckh->count--;
+		/* Try to halve the table if it is less than 1/4 full. */
+		if (ckh->count < (ZU(1) << (ckh->lg_curbuckets
+		    + LG_CKH_BUCKET_CELLS - 2)) && ckh->lg_curbuckets
+		    > ckh->lg_minbuckets) {
+			/* Ignore error due to OOM. */
+			ckh_shrink(tsd, ckh);
+		}
+
+		return false;
+	}
+
+	return true;
+}
+
+bool
+ckh_search(ckh_t *ckh, const void *searchkey, void **key, void **data) {
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL) {
+			*key = (void *)ckh->tab[cell].key;
+		}
+		if (data != NULL) {
+			*data = (void *)ckh->tab[cell].data;
+		}
+		return false;
+	}
+
+	return true;
+}
+
+void
+ckh_string_hash(const void *key, size_t r_hash[2]) {
+	hash(key, strlen((const char *)key), 0x94122f33U, r_hash);
+}
+
+bool
+ckh_string_keycomp(const void *k1, const void *k2) {
+	assert(k1 != NULL);
+	assert(k2 != NULL);
+
+	return !strcmp((char *)k1, (char *)k2);
+}
+
+void
+ckh_pointer_hash(const void *key, size_t r_hash[2]) {
+	union {
+		const void	*v;
+		size_t		i;
+	} u;
+
+	assert(sizeof(u.v) == sizeof(u.i));
+	u.v = key;
+	hash(&u.i, sizeof(u.i), 0xd983396eU, r_hash);
+}
+
+bool
+ckh_pointer_keycomp(const void *k1, const void *k2) {
+	return (k1 == k2);
+}
diff --git a/memkind-1.10.0/jemalloc/src/ctl.c b/memkind-1.10.0/jemalloc/src/ctl.c
new file mode 100644
index 0000000..37d4610
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/ctl.c
@@ -0,0 +1,3436 @@
+#define JEMALLOC_CTL_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+/* Data. */
+
+/*
+ * ctl_mtx protects the following:
+ * - ctl_stats->*
+ */
+static malloc_mutex_t	ctl_mtx;
+static bool		ctl_initialized;
+static ctl_stats_t	*ctl_stats;
+static ctl_arenas_t	*ctl_arenas;
+
+/******************************************************************************/
+/* Helpers for named and indexed nodes. */
+
+static const ctl_named_node_t *
+ctl_named_node(const ctl_node_t *node) {
+	return ((node->named) ? (const ctl_named_node_t *)node : NULL);
+}
+
+static const ctl_named_node_t *
+ctl_named_children(const ctl_named_node_t *node, size_t index) {
+	const ctl_named_node_t *children = ctl_named_node(node->children);
+
+	return (children ? &children[index] : NULL);
+}
+
+static const ctl_indexed_node_t *
+ctl_indexed_node(const ctl_node_t *node) {
+	return (!node->named ? (const ctl_indexed_node_t *)node : NULL);
+}
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+#define CTL_PROTO(n)							\
+static int	n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,	\
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen);
+
+#define INDEX_PROTO(n)							\
+static const ctl_named_node_t	*n##_index(tsdn_t *tsdn,		\
+    const size_t *mib, size_t miblen, size_t i);
+
+CTL_PROTO(version)
+CTL_PROTO(epoch)
+CTL_PROTO(background_thread)
+CTL_PROTO(max_background_threads)
+CTL_PROTO(thread_tcache_enabled)
+CTL_PROTO(thread_tcache_flush)
+CTL_PROTO(thread_prof_name)
+CTL_PROTO(thread_prof_active)
+CTL_PROTO(thread_arena)
+CTL_PROTO(thread_allocated)
+CTL_PROTO(thread_allocatedp)
+CTL_PROTO(thread_deallocated)
+CTL_PROTO(thread_deallocatedp)
+CTL_PROTO(config_cache_oblivious)
+CTL_PROTO(config_debug)
+CTL_PROTO(config_fill)
+CTL_PROTO(config_lazy_lock)
+CTL_PROTO(config_malloc_conf)
+CTL_PROTO(config_opt_safety_checks)
+CTL_PROTO(config_prof)
+CTL_PROTO(config_prof_libgcc)
+CTL_PROTO(config_prof_libunwind)
+CTL_PROTO(config_stats)
+CTL_PROTO(config_utrace)
+CTL_PROTO(config_xmalloc)
+CTL_PROTO(opt_abort)
+CTL_PROTO(opt_abort_conf)
+CTL_PROTO(opt_confirm_conf)
+CTL_PROTO(opt_metadata_thp)
+CTL_PROTO(opt_retain)
+CTL_PROTO(opt_dss)
+CTL_PROTO(opt_narenas)
+CTL_PROTO(opt_percpu_arena)
+CTL_PROTO(opt_oversize_threshold)
+CTL_PROTO(opt_background_thread)
+CTL_PROTO(opt_max_background_threads)
+CTL_PROTO(opt_dirty_decay_ms)
+CTL_PROTO(opt_muzzy_decay_ms)
+CTL_PROTO(opt_stats_print)
+CTL_PROTO(opt_stats_print_opts)
+CTL_PROTO(opt_junk)
+CTL_PROTO(opt_zero)
+CTL_PROTO(opt_utrace)
+CTL_PROTO(opt_xmalloc)
+CTL_PROTO(opt_tcache)
+CTL_PROTO(opt_thp)
+CTL_PROTO(opt_lg_extent_max_active_fit)
+CTL_PROTO(opt_lg_tcache_max)
+CTL_PROTO(opt_prof)
+CTL_PROTO(opt_prof_prefix)
+CTL_PROTO(opt_prof_active)
+CTL_PROTO(opt_prof_thread_active_init)
+CTL_PROTO(opt_lg_prof_sample)
+CTL_PROTO(opt_lg_prof_interval)
+CTL_PROTO(opt_prof_gdump)
+CTL_PROTO(opt_prof_final)
+CTL_PROTO(opt_prof_leak)
+CTL_PROTO(opt_prof_accum)
+CTL_PROTO(tcache_create)
+CTL_PROTO(tcache_flush)
+CTL_PROTO(tcache_destroy)
+CTL_PROTO(arena_i_initialized)
+CTL_PROTO(arena_i_decay)
+CTL_PROTO(arena_i_purge)
+CTL_PROTO(arena_i_reset)
+CTL_PROTO(arena_i_destroy)
+CTL_PROTO(arena_i_dss)
+CTL_PROTO(arena_i_dirty_decay_ms)
+CTL_PROTO(arena_i_muzzy_decay_ms)
+CTL_PROTO(arena_i_extent_hooks)
+CTL_PROTO(arena_i_retain_grow_limit)
+INDEX_PROTO(arena_i)
+CTL_PROTO(arenas_bin_i_size)
+CTL_PROTO(arenas_bin_i_nregs)
+CTL_PROTO(arenas_bin_i_slab_size)
+CTL_PROTO(arenas_bin_i_nshards)
+INDEX_PROTO(arenas_bin_i)
+CTL_PROTO(arenas_lextent_i_size)
+INDEX_PROTO(arenas_lextent_i)
+CTL_PROTO(arenas_narenas)
+CTL_PROTO(arenas_dirty_decay_ms)
+CTL_PROTO(arenas_muzzy_decay_ms)
+CTL_PROTO(arenas_quantum)
+CTL_PROTO(arenas_page)
+CTL_PROTO(arenas_tcache_max)
+CTL_PROTO(arenas_nbins)
+CTL_PROTO(arenas_nhbins)
+CTL_PROTO(arenas_nlextents)
+CTL_PROTO(arenas_create)
+CTL_PROTO(arenas_lookup)
+CTL_PROTO(prof_thread_active_init)
+CTL_PROTO(prof_active)
+CTL_PROTO(prof_dump)
+CTL_PROTO(prof_gdump)
+CTL_PROTO(prof_reset)
+CTL_PROTO(prof_interval)
+CTL_PROTO(lg_prof_sample)
+CTL_PROTO(prof_log_start)
+CTL_PROTO(prof_log_stop)
+CTL_PROTO(stats_arenas_i_small_allocated)
+CTL_PROTO(stats_arenas_i_small_nmalloc)
+CTL_PROTO(stats_arenas_i_small_ndalloc)
+CTL_PROTO(stats_arenas_i_small_nrequests)
+CTL_PROTO(stats_arenas_i_small_nfills)
+CTL_PROTO(stats_arenas_i_small_nflushes)
+CTL_PROTO(stats_arenas_i_large_allocated)
+CTL_PROTO(stats_arenas_i_large_nmalloc)
+CTL_PROTO(stats_arenas_i_large_ndalloc)
+CTL_PROTO(stats_arenas_i_large_nrequests)
+CTL_PROTO(stats_arenas_i_large_nfills)
+CTL_PROTO(stats_arenas_i_large_nflushes)
+CTL_PROTO(stats_arenas_i_bins_j_nmalloc)
+CTL_PROTO(stats_arenas_i_bins_j_ndalloc)
+CTL_PROTO(stats_arenas_i_bins_j_nrequests)
+CTL_PROTO(stats_arenas_i_bins_j_curregs)
+CTL_PROTO(stats_arenas_i_bins_j_nfills)
+CTL_PROTO(stats_arenas_i_bins_j_nflushes)
+CTL_PROTO(stats_arenas_i_bins_j_nslabs)
+CTL_PROTO(stats_arenas_i_bins_j_nreslabs)
+CTL_PROTO(stats_arenas_i_bins_j_curslabs)
+CTL_PROTO(stats_arenas_i_bins_j_nonfull_slabs)
+INDEX_PROTO(stats_arenas_i_bins_j)
+CTL_PROTO(stats_arenas_i_lextents_j_nmalloc)
+CTL_PROTO(stats_arenas_i_lextents_j_ndalloc)
+CTL_PROTO(stats_arenas_i_lextents_j_nrequests)
+CTL_PROTO(stats_arenas_i_lextents_j_curlextents)
+INDEX_PROTO(stats_arenas_i_lextents_j)
+CTL_PROTO(stats_arenas_i_extents_j_ndirty)
+CTL_PROTO(stats_arenas_i_extents_j_nmuzzy)
+CTL_PROTO(stats_arenas_i_extents_j_nretained)
+CTL_PROTO(stats_arenas_i_extents_j_dirty_bytes)
+CTL_PROTO(stats_arenas_i_extents_j_muzzy_bytes)
+CTL_PROTO(stats_arenas_i_extents_j_retained_bytes)
+INDEX_PROTO(stats_arenas_i_extents_j)
+CTL_PROTO(stats_arenas_i_nthreads)
+CTL_PROTO(stats_arenas_i_uptime)
+CTL_PROTO(stats_arenas_i_dss)
+CTL_PROTO(stats_arenas_i_dirty_decay_ms)
+CTL_PROTO(stats_arenas_i_muzzy_decay_ms)
+CTL_PROTO(stats_arenas_i_pactive)
+CTL_PROTO(stats_arenas_i_pdirty)
+CTL_PROTO(stats_arenas_i_pmuzzy)
+CTL_PROTO(stats_arenas_i_mapped)
+CTL_PROTO(stats_arenas_i_retained)
+CTL_PROTO(stats_arenas_i_extent_avail)
+CTL_PROTO(stats_arenas_i_dirty_npurge)
+CTL_PROTO(stats_arenas_i_dirty_nmadvise)
+CTL_PROTO(stats_arenas_i_dirty_purged)
+CTL_PROTO(stats_arenas_i_muzzy_npurge)
+CTL_PROTO(stats_arenas_i_muzzy_nmadvise)
+CTL_PROTO(stats_arenas_i_muzzy_purged)
+CTL_PROTO(stats_arenas_i_base)
+CTL_PROTO(stats_arenas_i_internal)
+CTL_PROTO(stats_arenas_i_metadata_thp)
+CTL_PROTO(stats_arenas_i_tcache_bytes)
+CTL_PROTO(stats_arenas_i_resident)
+CTL_PROTO(stats_arenas_i_abandoned_vm)
+INDEX_PROTO(stats_arenas_i)
+CTL_PROTO(stats_allocated)
+CTL_PROTO(stats_active)
+CTL_PROTO(stats_background_thread_num_threads)
+CTL_PROTO(stats_background_thread_num_runs)
+CTL_PROTO(stats_background_thread_run_interval)
+CTL_PROTO(stats_metadata)
+CTL_PROTO(stats_metadata_thp)
+CTL_PROTO(stats_resident)
+CTL_PROTO(stats_mapped)
+CTL_PROTO(stats_retained)
+CTL_PROTO(experimental_hooks_install)
+CTL_PROTO(experimental_hooks_remove)
+CTL_PROTO(experimental_utilization_query)
+CTL_PROTO(experimental_utilization_batch_query)
+CTL_PROTO(experimental_arenas_i_pactivep)
+INDEX_PROTO(experimental_arenas_i)
+
+#define MUTEX_STATS_CTL_PROTO_GEN(n)					\
+CTL_PROTO(stats_##n##_num_ops)						\
+CTL_PROTO(stats_##n##_num_wait)						\
+CTL_PROTO(stats_##n##_num_spin_acq)					\
+CTL_PROTO(stats_##n##_num_owner_switch)					\
+CTL_PROTO(stats_##n##_total_wait_time)					\
+CTL_PROTO(stats_##n##_max_wait_time)					\
+CTL_PROTO(stats_##n##_max_num_thds)
+
+/* Global mutexes. */
+#define OP(mtx) MUTEX_STATS_CTL_PROTO_GEN(mutexes_##mtx)
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+
+/* Per arena mutexes. */
+#define OP(mtx) MUTEX_STATS_CTL_PROTO_GEN(arenas_i_mutexes_##mtx)
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+
+/* Arena bin mutexes. */
+MUTEX_STATS_CTL_PROTO_GEN(arenas_i_bins_j_mutex)
+#undef MUTEX_STATS_CTL_PROTO_GEN
+
+CTL_PROTO(stats_mutexes_reset)
+
+/******************************************************************************/
+/* mallctl tree. */
+
+#define NAME(n)	{true},	n
+#define CHILD(t, c)							\
+	sizeof(c##_node) / sizeof(ctl_##t##_node_t),			\
+	(ctl_node_t *)c##_node,						\
+	NULL
+#define CTL(c)	0, NULL, c##_ctl
+
+/*
+ * Only handles internal indexed nodes, since there are currently no external
+ * ones.
+ */
+#define INDEX(i)	{false},	i##_index
+
+static const ctl_named_node_t	thread_tcache_node[] = {
+	{NAME("enabled"),	CTL(thread_tcache_enabled)},
+	{NAME("flush"),		CTL(thread_tcache_flush)}
+};
+
+static const ctl_named_node_t	thread_prof_node[] = {
+	{NAME("name"),		CTL(thread_prof_name)},
+	{NAME("active"),	CTL(thread_prof_active)}
+};
+
+static const ctl_named_node_t	thread_node[] = {
+	{NAME("arena"),		CTL(thread_arena)},
+	{NAME("allocated"),	CTL(thread_allocated)},
+	{NAME("allocatedp"),	CTL(thread_allocatedp)},
+	{NAME("deallocated"),	CTL(thread_deallocated)},
+	{NAME("deallocatedp"),	CTL(thread_deallocatedp)},
+	{NAME("tcache"),	CHILD(named, thread_tcache)},
+	{NAME("prof"),		CHILD(named, thread_prof)}
+};
+
+static const ctl_named_node_t	config_node[] = {
+	{NAME("cache_oblivious"), CTL(config_cache_oblivious)},
+	{NAME("debug"),		CTL(config_debug)},
+	{NAME("fill"),		CTL(config_fill)},
+	{NAME("lazy_lock"),	CTL(config_lazy_lock)},
+	{NAME("malloc_conf"),	CTL(config_malloc_conf)},
+	{NAME("opt_safety_checks"),	CTL(config_opt_safety_checks)},
+	{NAME("prof"),		CTL(config_prof)},
+	{NAME("prof_libgcc"),	CTL(config_prof_libgcc)},
+	{NAME("prof_libunwind"), CTL(config_prof_libunwind)},
+	{NAME("stats"),		CTL(config_stats)},
+	{NAME("utrace"),	CTL(config_utrace)},
+	{NAME("xmalloc"),	CTL(config_xmalloc)}
+};
+
+static const ctl_named_node_t opt_node[] = {
+	{NAME("abort"),		CTL(opt_abort)},
+	{NAME("abort_conf"),	CTL(opt_abort_conf)},
+	{NAME("confirm_conf"),	CTL(opt_confirm_conf)},
+	{NAME("metadata_thp"),	CTL(opt_metadata_thp)},
+	{NAME("retain"),	CTL(opt_retain)},
+	{NAME("dss"),		CTL(opt_dss)},
+	{NAME("narenas"),	CTL(opt_narenas)},
+	{NAME("percpu_arena"),	CTL(opt_percpu_arena)},
+	{NAME("oversize_threshold"),	CTL(opt_oversize_threshold)},
+	{NAME("background_thread"),	CTL(opt_background_thread)},
+	{NAME("max_background_threads"),	CTL(opt_max_background_threads)},
+	{NAME("dirty_decay_ms"), CTL(opt_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(opt_muzzy_decay_ms)},
+	{NAME("stats_print"),	CTL(opt_stats_print)},
+	{NAME("stats_print_opts"),	CTL(opt_stats_print_opts)},
+	{NAME("junk"),		CTL(opt_junk)},
+	{NAME("zero"),		CTL(opt_zero)},
+	{NAME("utrace"),	CTL(opt_utrace)},
+	{NAME("xmalloc"),	CTL(opt_xmalloc)},
+	{NAME("tcache"),	CTL(opt_tcache)},
+	{NAME("thp"),		CTL(opt_thp)},
+	{NAME("lg_extent_max_active_fit"), CTL(opt_lg_extent_max_active_fit)},
+	{NAME("lg_tcache_max"),	CTL(opt_lg_tcache_max)},
+	{NAME("prof"),		CTL(opt_prof)},
+	{NAME("prof_prefix"),	CTL(opt_prof_prefix)},
+	{NAME("prof_active"),	CTL(opt_prof_active)},
+	{NAME("prof_thread_active_init"), CTL(opt_prof_thread_active_init)},
+	{NAME("lg_prof_sample"), CTL(opt_lg_prof_sample)},
+	{NAME("lg_prof_interval"), CTL(opt_lg_prof_interval)},
+	{NAME("prof_gdump"),	CTL(opt_prof_gdump)},
+	{NAME("prof_final"),	CTL(opt_prof_final)},
+	{NAME("prof_leak"),	CTL(opt_prof_leak)},
+	{NAME("prof_accum"),	CTL(opt_prof_accum)}
+};
+
+static const ctl_named_node_t	tcache_node[] = {
+	{NAME("create"),	CTL(tcache_create)},
+	{NAME("flush"),		CTL(tcache_flush)},
+	{NAME("destroy"),	CTL(tcache_destroy)}
+};
+
+static const ctl_named_node_t arena_i_node[] = {
+	{NAME("initialized"),	CTL(arena_i_initialized)},
+	{NAME("decay"),		CTL(arena_i_decay)},
+	{NAME("purge"),		CTL(arena_i_purge)},
+	{NAME("reset"),		CTL(arena_i_reset)},
+	{NAME("destroy"),	CTL(arena_i_destroy)},
+	{NAME("dss"),		CTL(arena_i_dss)},
+	{NAME("dirty_decay_ms"), CTL(arena_i_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(arena_i_muzzy_decay_ms)},
+	{NAME("extent_hooks"),	CTL(arena_i_extent_hooks)},
+	{NAME("retain_grow_limit"),	CTL(arena_i_retain_grow_limit)}
+};
+static const ctl_named_node_t super_arena_i_node[] = {
+	{NAME(""),		CHILD(named, arena_i)}
+};
+
+static const ctl_indexed_node_t arena_node[] = {
+	{INDEX(arena_i)}
+};
+
+static const ctl_named_node_t arenas_bin_i_node[] = {
+	{NAME("size"),		CTL(arenas_bin_i_size)},
+	{NAME("nregs"),		CTL(arenas_bin_i_nregs)},
+	{NAME("slab_size"),	CTL(arenas_bin_i_slab_size)},
+	{NAME("nshards"),	CTL(arenas_bin_i_nshards)}
+};
+static const ctl_named_node_t super_arenas_bin_i_node[] = {
+	{NAME(""),		CHILD(named, arenas_bin_i)}
+};
+
+static const ctl_indexed_node_t arenas_bin_node[] = {
+	{INDEX(arenas_bin_i)}
+};
+
+static const ctl_named_node_t arenas_lextent_i_node[] = {
+	{NAME("size"),		CTL(arenas_lextent_i_size)}
+};
+static const ctl_named_node_t super_arenas_lextent_i_node[] = {
+	{NAME(""),		CHILD(named, arenas_lextent_i)}
+};
+
+static const ctl_indexed_node_t arenas_lextent_node[] = {
+	{INDEX(arenas_lextent_i)}
+};
+
+static const ctl_named_node_t arenas_node[] = {
+	{NAME("narenas"),	CTL(arenas_narenas)},
+	{NAME("dirty_decay_ms"), CTL(arenas_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(arenas_muzzy_decay_ms)},
+	{NAME("quantum"),	CTL(arenas_quantum)},
+	{NAME("page"),		CTL(arenas_page)},
+	{NAME("tcache_max"),	CTL(arenas_tcache_max)},
+	{NAME("nbins"),		CTL(arenas_nbins)},
+	{NAME("nhbins"),	CTL(arenas_nhbins)},
+	{NAME("bin"),		CHILD(indexed, arenas_bin)},
+	{NAME("nlextents"),	CTL(arenas_nlextents)},
+	{NAME("lextent"),	CHILD(indexed, arenas_lextent)},
+	{NAME("create"),	CTL(arenas_create)},
+	{NAME("lookup"),	CTL(arenas_lookup)}
+};
+
+static const ctl_named_node_t	prof_node[] = {
+	{NAME("thread_active_init"), CTL(prof_thread_active_init)},
+	{NAME("active"),	CTL(prof_active)},
+	{NAME("dump"),		CTL(prof_dump)},
+	{NAME("gdump"),		CTL(prof_gdump)},
+	{NAME("reset"),		CTL(prof_reset)},
+	{NAME("interval"),	CTL(prof_interval)},
+	{NAME("lg_sample"),	CTL(lg_prof_sample)},
+	{NAME("log_start"),	CTL(prof_log_start)},
+	{NAME("log_stop"),	CTL(prof_log_stop)}
+};
+static const ctl_named_node_t stats_arenas_i_small_node[] = {
+	{NAME("allocated"),	CTL(stats_arenas_i_small_allocated)},
+	{NAME("nmalloc"),	CTL(stats_arenas_i_small_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_small_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_small_nrequests)},
+	{NAME("nfills"),	CTL(stats_arenas_i_small_nfills)},
+	{NAME("nflushes"),	CTL(stats_arenas_i_small_nflushes)}
+};
+
+static const ctl_named_node_t stats_arenas_i_large_node[] = {
+	{NAME("allocated"),	CTL(stats_arenas_i_large_allocated)},
+	{NAME("nmalloc"),	CTL(stats_arenas_i_large_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_large_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_large_nrequests)},
+	{NAME("nfills"),	CTL(stats_arenas_i_large_nfills)},
+	{NAME("nflushes"),	CTL(stats_arenas_i_large_nflushes)}
+};
+
+#define MUTEX_PROF_DATA_NODE(prefix)					\
+static const ctl_named_node_t stats_##prefix##_node[] = {		\
+	{NAME("num_ops"),						\
+	 CTL(stats_##prefix##_num_ops)},				\
+	{NAME("num_wait"),						\
+	 CTL(stats_##prefix##_num_wait)},				\
+	{NAME("num_spin_acq"),						\
+	 CTL(stats_##prefix##_num_spin_acq)},				\
+	{NAME("num_owner_switch"),					\
+	 CTL(stats_##prefix##_num_owner_switch)},			\
+	{NAME("total_wait_time"),					\
+	 CTL(stats_##prefix##_total_wait_time)},			\
+	{NAME("max_wait_time"),						\
+	 CTL(stats_##prefix##_max_wait_time)},				\
+	{NAME("max_num_thds"),						\
+	 CTL(stats_##prefix##_max_num_thds)}				\
+	/* Note that # of current waiting thread not provided. */	\
+};
+
+MUTEX_PROF_DATA_NODE(arenas_i_bins_j_mutex)
+
+static const ctl_named_node_t stats_arenas_i_bins_j_node[] = {
+	{NAME("nmalloc"),	CTL(stats_arenas_i_bins_j_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_bins_j_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_bins_j_nrequests)},
+	{NAME("curregs"),	CTL(stats_arenas_i_bins_j_curregs)},
+	{NAME("nfills"),	CTL(stats_arenas_i_bins_j_nfills)},
+	{NAME("nflushes"),	CTL(stats_arenas_i_bins_j_nflushes)},
+	{NAME("nslabs"),	CTL(stats_arenas_i_bins_j_nslabs)},
+	{NAME("nreslabs"),	CTL(stats_arenas_i_bins_j_nreslabs)},
+	{NAME("curslabs"),	CTL(stats_arenas_i_bins_j_curslabs)},
+	{NAME("nonfull_slabs"),	CTL(stats_arenas_i_bins_j_nonfull_slabs)},
+	{NAME("mutex"),		CHILD(named, stats_arenas_i_bins_j_mutex)}
+};
+
+static const ctl_named_node_t super_stats_arenas_i_bins_j_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i_bins_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_bins_node[] = {
+	{INDEX(stats_arenas_i_bins_j)}
+};
+
+static const ctl_named_node_t stats_arenas_i_lextents_j_node[] = {
+	{NAME("nmalloc"),	CTL(stats_arenas_i_lextents_j_nmalloc)},
+	{NAME("ndalloc"),	CTL(stats_arenas_i_lextents_j_ndalloc)},
+	{NAME("nrequests"),	CTL(stats_arenas_i_lextents_j_nrequests)},
+	{NAME("curlextents"),	CTL(stats_arenas_i_lextents_j_curlextents)}
+};
+static const ctl_named_node_t super_stats_arenas_i_lextents_j_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i_lextents_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_lextents_node[] = {
+	{INDEX(stats_arenas_i_lextents_j)}
+};
+
+static const ctl_named_node_t stats_arenas_i_extents_j_node[] = {
+	{NAME("ndirty"),	CTL(stats_arenas_i_extents_j_ndirty)},
+	{NAME("nmuzzy"),	CTL(stats_arenas_i_extents_j_nmuzzy)},
+	{NAME("nretained"),	CTL(stats_arenas_i_extents_j_nretained)},
+	{NAME("dirty_bytes"),	CTL(stats_arenas_i_extents_j_dirty_bytes)},
+	{NAME("muzzy_bytes"),	CTL(stats_arenas_i_extents_j_muzzy_bytes)},
+	{NAME("retained_bytes"), CTL(stats_arenas_i_extents_j_retained_bytes)}
+};
+
+static const ctl_named_node_t super_stats_arenas_i_extents_j_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i_extents_j)}
+};
+
+static const ctl_indexed_node_t stats_arenas_i_extents_node[] = {
+	{INDEX(stats_arenas_i_extents_j)}
+};
+
+#define OP(mtx)  MUTEX_PROF_DATA_NODE(arenas_i_mutexes_##mtx)
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+
+static const ctl_named_node_t stats_arenas_i_mutexes_node[] = {
+#define OP(mtx) {NAME(#mtx), CHILD(named, stats_arenas_i_mutexes_##mtx)},
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+};
+
+static const ctl_named_node_t stats_arenas_i_node[] = {
+	{NAME("nthreads"),	CTL(stats_arenas_i_nthreads)},
+	{NAME("uptime"),	CTL(stats_arenas_i_uptime)},
+	{NAME("dss"),		CTL(stats_arenas_i_dss)},
+	{NAME("dirty_decay_ms"), CTL(stats_arenas_i_dirty_decay_ms)},
+	{NAME("muzzy_decay_ms"), CTL(stats_arenas_i_muzzy_decay_ms)},
+	{NAME("pactive"),	CTL(stats_arenas_i_pactive)},
+	{NAME("pdirty"),	CTL(stats_arenas_i_pdirty)},
+	{NAME("pmuzzy"),	CTL(stats_arenas_i_pmuzzy)},
+	{NAME("mapped"),	CTL(stats_arenas_i_mapped)},
+	{NAME("retained"),	CTL(stats_arenas_i_retained)},
+	{NAME("extent_avail"),	CTL(stats_arenas_i_extent_avail)},
+	{NAME("dirty_npurge"),	CTL(stats_arenas_i_dirty_npurge)},
+	{NAME("dirty_nmadvise"), CTL(stats_arenas_i_dirty_nmadvise)},
+	{NAME("dirty_purged"),	CTL(stats_arenas_i_dirty_purged)},
+	{NAME("muzzy_npurge"),	CTL(stats_arenas_i_muzzy_npurge)},
+	{NAME("muzzy_nmadvise"), CTL(stats_arenas_i_muzzy_nmadvise)},
+	{NAME("muzzy_purged"),	CTL(stats_arenas_i_muzzy_purged)},
+	{NAME("base"),		CTL(stats_arenas_i_base)},
+	{NAME("internal"),	CTL(stats_arenas_i_internal)},
+	{NAME("metadata_thp"),	CTL(stats_arenas_i_metadata_thp)},
+	{NAME("tcache_bytes"),	CTL(stats_arenas_i_tcache_bytes)},
+	{NAME("resident"),	CTL(stats_arenas_i_resident)},
+	{NAME("abandoned_vm"),	CTL(stats_arenas_i_abandoned_vm)},
+	{NAME("small"),		CHILD(named, stats_arenas_i_small)},
+	{NAME("large"),		CHILD(named, stats_arenas_i_large)},
+	{NAME("bins"),		CHILD(indexed, stats_arenas_i_bins)},
+	{NAME("lextents"),	CHILD(indexed, stats_arenas_i_lextents)},
+	{NAME("extents"),	CHILD(indexed, stats_arenas_i_extents)},
+	{NAME("mutexes"),	CHILD(named, stats_arenas_i_mutexes)}
+};
+static const ctl_named_node_t super_stats_arenas_i_node[] = {
+	{NAME(""),		CHILD(named, stats_arenas_i)}
+};
+
+static const ctl_indexed_node_t stats_arenas_node[] = {
+	{INDEX(stats_arenas_i)}
+};
+
+static const ctl_named_node_t stats_background_thread_node[] = {
+	{NAME("num_threads"),	CTL(stats_background_thread_num_threads)},
+	{NAME("num_runs"),	CTL(stats_background_thread_num_runs)},
+	{NAME("run_interval"),	CTL(stats_background_thread_run_interval)}
+};
+
+#define OP(mtx) MUTEX_PROF_DATA_NODE(mutexes_##mtx)
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+
+static const ctl_named_node_t stats_mutexes_node[] = {
+#define OP(mtx) {NAME(#mtx), CHILD(named, stats_mutexes_##mtx)},
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+	{NAME("reset"),		CTL(stats_mutexes_reset)}
+};
+#undef MUTEX_PROF_DATA_NODE
+
+static const ctl_named_node_t stats_node[] = {
+	{NAME("allocated"),	CTL(stats_allocated)},
+	{NAME("active"),	CTL(stats_active)},
+	{NAME("metadata"),	CTL(stats_metadata)},
+	{NAME("metadata_thp"),	CTL(stats_metadata_thp)},
+	{NAME("resident"),	CTL(stats_resident)},
+	{NAME("mapped"),	CTL(stats_mapped)},
+	{NAME("retained"),	CTL(stats_retained)},
+	{NAME("background_thread"),
+	 CHILD(named, stats_background_thread)},
+	{NAME("mutexes"),	CHILD(named, stats_mutexes)},
+	{NAME("arenas"),	CHILD(indexed, stats_arenas)}
+};
+
+static const ctl_named_node_t experimental_hooks_node[] = {
+	{NAME("install"),	CTL(experimental_hooks_install)},
+	{NAME("remove"),	CTL(experimental_hooks_remove)}
+};
+
+static const ctl_named_node_t experimental_utilization_node[] = {
+	{NAME("query"),		CTL(experimental_utilization_query)},
+	{NAME("batch_query"),	CTL(experimental_utilization_batch_query)}
+};
+
+static const ctl_named_node_t experimental_arenas_i_node[] = {
+	{NAME("pactivep"),	CTL(experimental_arenas_i_pactivep)}
+};
+static const ctl_named_node_t super_experimental_arenas_i_node[] = {
+	{NAME(""),		CHILD(named, experimental_arenas_i)}
+};
+
+static const ctl_indexed_node_t experimental_arenas_node[] = {
+	{INDEX(experimental_arenas_i)}
+};
+
+static const ctl_named_node_t experimental_node[] = {
+	{NAME("hooks"),		CHILD(named, experimental_hooks)},
+	{NAME("utilization"),	CHILD(named, experimental_utilization)},
+	{NAME("arenas"),	CHILD(indexed, experimental_arenas)}
+};
+
+static const ctl_named_node_t	root_node[] = {
+	{NAME("version"),	CTL(version)},
+	{NAME("epoch"),		CTL(epoch)},
+	{NAME("background_thread"),	CTL(background_thread)},
+	{NAME("max_background_threads"),	CTL(max_background_threads)},
+	{NAME("thread"),	CHILD(named, thread)},
+	{NAME("config"),	CHILD(named, config)},
+	{NAME("opt"),		CHILD(named, opt)},
+	{NAME("tcache"),	CHILD(named, tcache)},
+	{NAME("arena"),		CHILD(indexed, arena)},
+	{NAME("arenas"),	CHILD(named, arenas)},
+	{NAME("prof"),		CHILD(named, prof)},
+	{NAME("stats"),		CHILD(named, stats)},
+	{NAME("experimental"),	CHILD(named, experimental)}
+};
+static const ctl_named_node_t super_root_node[] = {
+	{NAME(""),		CHILD(named, root)}
+};
+
+#undef NAME
+#undef CHILD
+#undef CTL
+#undef INDEX
+
+/******************************************************************************/
+
+/*
+ * Sets *dst + *src non-atomically.  This is safe, since everything is
+ * synchronized by the ctl mutex.
+ */
+static void
+ctl_accum_arena_stats_u64(arena_stats_u64_t *dst, arena_stats_u64_t *src) {
+#ifdef JEMALLOC_ATOMIC_U64
+	uint64_t cur_dst = atomic_load_u64(dst, ATOMIC_RELAXED);
+	uint64_t cur_src = atomic_load_u64(src, ATOMIC_RELAXED);
+	atomic_store_u64(dst, cur_dst + cur_src, ATOMIC_RELAXED);
+#else
+	*dst += *src;
+#endif
+}
+
+/* Likewise: with ctl mutex synchronization, reading is simple. */
+static uint64_t
+ctl_arena_stats_read_u64(arena_stats_u64_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_u64(p, ATOMIC_RELAXED);
+#else
+	return *p;
+#endif
+}
+
+static void
+accum_atomic_zu(atomic_zu_t *dst, atomic_zu_t *src) {
+	size_t cur_dst = atomic_load_zu(dst, ATOMIC_RELAXED);
+	size_t cur_src = atomic_load_zu(src, ATOMIC_RELAXED);
+	atomic_store_zu(dst, cur_dst + cur_src, ATOMIC_RELAXED);
+}
+
+/******************************************************************************/
+
+static unsigned
+arenas_i2a_impl(size_t i, bool compat, bool validate) {
+	unsigned a;
+
+	switch (i) {
+	case MALLCTL_ARENAS_ALL:
+		a = 0;
+		break;
+	case MALLCTL_ARENAS_DESTROYED:
+		a = 1;
+		break;
+	default:
+		if (compat && i == ctl_arenas->narenas) {
+			/*
+			 * Provide deprecated backward compatibility for
+			 * accessing the merged stats at index narenas rather
+			 * than via MALLCTL_ARENAS_ALL.  This is scheduled for
+			 * removal in 6.0.0.
+			 */
+			a = 0;
+		} else if (validate && i >= ctl_arenas->narenas) {
+			a = UINT_MAX;
+		} else {
+			/*
+			 * This function should never be called for an index
+			 * more than one past the range of indices that have
+			 * initialized ctl data.
+			 */
+			assert(i < ctl_arenas->narenas || (!validate && i ==
+			    ctl_arenas->narenas));
+			a = (unsigned)i + 2;
+		}
+		break;
+	}
+
+	return a;
+}
+
+static unsigned
+arenas_i2a(size_t i) {
+	return arenas_i2a_impl(i, true, false);
+}
+
+static ctl_arena_t *
+arenas_i_impl(tsd_t *tsd, size_t i, bool compat, bool init) {
+	ctl_arena_t *ret;
+
+	assert(!compat || !init);
+
+	ret = ctl_arenas->arenas[arenas_i2a_impl(i, compat, false)];
+	if (init && ret == NULL) {
+		if (config_stats) {
+			struct container_s {
+				ctl_arena_t		ctl_arena;
+				ctl_arena_stats_t	astats;
+			};
+			struct container_s *cont =
+			    (struct container_s *)base_alloc(tsd_tsdn(tsd),
+			    b0get(), sizeof(struct container_s), QUANTUM);
+			if (cont == NULL) {
+				return NULL;
+			}
+			ret = &cont->ctl_arena;
+			ret->astats = &cont->astats;
+		} else {
+			ret = (ctl_arena_t *)base_alloc(tsd_tsdn(tsd), b0get(),
+			    sizeof(ctl_arena_t), QUANTUM);
+			if (ret == NULL) {
+				return NULL;
+			}
+		}
+		ret->arena_ind = (unsigned)i;
+		ctl_arenas->arenas[arenas_i2a_impl(i, compat, false)] = ret;
+	}
+
+	assert(ret == NULL || arenas_i2a(ret->arena_ind) == arenas_i2a(i));
+	return ret;
+}
+
+static ctl_arena_t *
+arenas_i(size_t i) {
+	ctl_arena_t *ret = arenas_i_impl(tsd_fetch(), i, true, false);
+	assert(ret != NULL);
+	return ret;
+}
+
+static void
+ctl_arena_clear(ctl_arena_t *ctl_arena) {
+	ctl_arena->nthreads = 0;
+	ctl_arena->dss = dss_prec_names[dss_prec_limit];
+	ctl_arena->dirty_decay_ms = -1;
+	ctl_arena->muzzy_decay_ms = -1;
+	ctl_arena->pactive = 0;
+	ctl_arena->pdirty = 0;
+	ctl_arena->pmuzzy = 0;
+	if (config_stats) {
+		memset(&ctl_arena->astats->astats, 0, sizeof(arena_stats_t));
+		ctl_arena->astats->allocated_small = 0;
+		ctl_arena->astats->nmalloc_small = 0;
+		ctl_arena->astats->ndalloc_small = 0;
+		ctl_arena->astats->nrequests_small = 0;
+		ctl_arena->astats->nfills_small = 0;
+		ctl_arena->astats->nflushes_small = 0;
+		memset(ctl_arena->astats->bstats, 0, SC_NBINS *
+		    sizeof(bin_stats_t));
+		memset(ctl_arena->astats->lstats, 0, (SC_NSIZES - SC_NBINS) *
+		    sizeof(arena_stats_large_t));
+		memset(ctl_arena->astats->estats, 0, SC_NPSIZES *
+		    sizeof(arena_stats_extents_t));
+	}
+}
+
+static void
+ctl_arena_stats_amerge(tsdn_t *tsdn, ctl_arena_t *ctl_arena, arena_t *arena) {
+	unsigned i;
+
+	if (config_stats) {
+		arena_stats_merge(tsdn, arena, &ctl_arena->nthreads,
+		    &ctl_arena->dss, &ctl_arena->dirty_decay_ms,
+		    &ctl_arena->muzzy_decay_ms, &ctl_arena->pactive,
+		    &ctl_arena->pdirty, &ctl_arena->pmuzzy,
+		    &ctl_arena->astats->astats, ctl_arena->astats->bstats,
+		    ctl_arena->astats->lstats, ctl_arena->astats->estats);
+
+		for (i = 0; i < SC_NBINS; i++) {
+			ctl_arena->astats->allocated_small +=
+			    ctl_arena->astats->bstats[i].curregs *
+			    sz_index2size(i);
+			ctl_arena->astats->nmalloc_small +=
+			    ctl_arena->astats->bstats[i].nmalloc;
+			ctl_arena->astats->ndalloc_small +=
+			    ctl_arena->astats->bstats[i].ndalloc;
+			ctl_arena->astats->nrequests_small +=
+			    ctl_arena->astats->bstats[i].nrequests;
+			ctl_arena->astats->nfills_small +=
+			    ctl_arena->astats->bstats[i].nfills;
+			ctl_arena->astats->nflushes_small +=
+			    ctl_arena->astats->bstats[i].nflushes;
+		}
+	} else {
+		arena_basic_stats_merge(tsdn, arena, &ctl_arena->nthreads,
+		    &ctl_arena->dss, &ctl_arena->dirty_decay_ms,
+		    &ctl_arena->muzzy_decay_ms, &ctl_arena->pactive,
+		    &ctl_arena->pdirty, &ctl_arena->pmuzzy);
+	}
+}
+
+static void
+ctl_arena_stats_sdmerge(ctl_arena_t *ctl_sdarena, ctl_arena_t *ctl_arena,
+    bool destroyed) {
+	unsigned i;
+
+	if (!destroyed) {
+		ctl_sdarena->nthreads += ctl_arena->nthreads;
+		ctl_sdarena->pactive += ctl_arena->pactive;
+		ctl_sdarena->pdirty += ctl_arena->pdirty;
+		ctl_sdarena->pmuzzy += ctl_arena->pmuzzy;
+	} else {
+		assert(ctl_arena->nthreads == 0);
+		assert(ctl_arena->pactive == 0);
+		assert(ctl_arena->pdirty == 0);
+		assert(ctl_arena->pmuzzy == 0);
+	}
+
+	if (config_stats) {
+		ctl_arena_stats_t *sdstats = ctl_sdarena->astats;
+		ctl_arena_stats_t *astats = ctl_arena->astats;
+
+		if (!destroyed) {
+			accum_atomic_zu(&sdstats->astats.mapped,
+			    &astats->astats.mapped);
+			accum_atomic_zu(&sdstats->astats.retained,
+			    &astats->astats.retained);
+			accum_atomic_zu(&sdstats->astats.extent_avail,
+			    &astats->astats.extent_avail);
+		}
+
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_dirty.npurge,
+		    &astats->astats.decay_dirty.npurge);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_dirty.nmadvise,
+		    &astats->astats.decay_dirty.nmadvise);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_dirty.purged,
+		    &astats->astats.decay_dirty.purged);
+
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_muzzy.npurge,
+		    &astats->astats.decay_muzzy.npurge);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_muzzy.nmadvise,
+		    &astats->astats.decay_muzzy.nmadvise);
+		ctl_accum_arena_stats_u64(&sdstats->astats.decay_muzzy.purged,
+		    &astats->astats.decay_muzzy.purged);
+
+#define OP(mtx) malloc_mutex_prof_merge(				\
+		    &(sdstats->astats.mutex_prof_data[			\
+		        arena_prof_mutex_##mtx]),			\
+		    &(astats->astats.mutex_prof_data[			\
+		        arena_prof_mutex_##mtx]));
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+		if (!destroyed) {
+			accum_atomic_zu(&sdstats->astats.base,
+			    &astats->astats.base);
+			accum_atomic_zu(&sdstats->astats.internal,
+			    &astats->astats.internal);
+			accum_atomic_zu(&sdstats->astats.resident,
+			    &astats->astats.resident);
+			accum_atomic_zu(&sdstats->astats.metadata_thp,
+			    &astats->astats.metadata_thp);
+		} else {
+			assert(atomic_load_zu(
+			    &astats->astats.internal, ATOMIC_RELAXED) == 0);
+		}
+
+		if (!destroyed) {
+			sdstats->allocated_small += astats->allocated_small;
+		} else {
+			assert(astats->allocated_small == 0);
+		}
+		sdstats->nmalloc_small += astats->nmalloc_small;
+		sdstats->ndalloc_small += astats->ndalloc_small;
+		sdstats->nrequests_small += astats->nrequests_small;
+		sdstats->nfills_small += astats->nfills_small;
+		sdstats->nflushes_small += astats->nflushes_small;
+
+		if (!destroyed) {
+			accum_atomic_zu(&sdstats->astats.allocated_large,
+			    &astats->astats.allocated_large);
+		} else {
+			assert(atomic_load_zu(&astats->astats.allocated_large,
+			    ATOMIC_RELAXED) == 0);
+		}
+		ctl_accum_arena_stats_u64(&sdstats->astats.nmalloc_large,
+		    &astats->astats.nmalloc_large);
+		ctl_accum_arena_stats_u64(&sdstats->astats.ndalloc_large,
+		    &astats->astats.ndalloc_large);
+		ctl_accum_arena_stats_u64(&sdstats->astats.nrequests_large,
+		    &astats->astats.nrequests_large);
+		accum_atomic_zu(&sdstats->astats.abandoned_vm,
+		    &astats->astats.abandoned_vm);
+
+		accum_atomic_zu(&sdstats->astats.tcache_bytes,
+		    &astats->astats.tcache_bytes);
+
+		if (ctl_arena->arena_ind == 0) {
+			sdstats->astats.uptime = astats->astats.uptime;
+		}
+
+		/* Merge bin stats. */
+		for (i = 0; i < SC_NBINS; i++) {
+			sdstats->bstats[i].nmalloc += astats->bstats[i].nmalloc;
+			sdstats->bstats[i].ndalloc += astats->bstats[i].ndalloc;
+			sdstats->bstats[i].nrequests +=
+			    astats->bstats[i].nrequests;
+			if (!destroyed) {
+				sdstats->bstats[i].curregs +=
+				    astats->bstats[i].curregs;
+			} else {
+				assert(astats->bstats[i].curregs == 0);
+			}
+			sdstats->bstats[i].nfills += astats->bstats[i].nfills;
+			sdstats->bstats[i].nflushes +=
+			    astats->bstats[i].nflushes;
+			sdstats->bstats[i].nslabs += astats->bstats[i].nslabs;
+			sdstats->bstats[i].reslabs += astats->bstats[i].reslabs;
+			if (!destroyed) {
+				sdstats->bstats[i].curslabs +=
+				    astats->bstats[i].curslabs;
+				sdstats->bstats[i].nonfull_slabs +=
+				    astats->bstats[i].nonfull_slabs;
+			} else {
+				assert(astats->bstats[i].curslabs == 0);
+				assert(astats->bstats[i].nonfull_slabs == 0);
+			}
+			malloc_mutex_prof_merge(&sdstats->bstats[i].mutex_data,
+			    &astats->bstats[i].mutex_data);
+		}
+
+		/* Merge stats for large allocations. */
+		for (i = 0; i < SC_NSIZES - SC_NBINS; i++) {
+			ctl_accum_arena_stats_u64(&sdstats->lstats[i].nmalloc,
+			    &astats->lstats[i].nmalloc);
+			ctl_accum_arena_stats_u64(&sdstats->lstats[i].ndalloc,
+			    &astats->lstats[i].ndalloc);
+			ctl_accum_arena_stats_u64(&sdstats->lstats[i].nrequests,
+			    &astats->lstats[i].nrequests);
+			if (!destroyed) {
+				sdstats->lstats[i].curlextents +=
+				    astats->lstats[i].curlextents;
+			} else {
+				assert(astats->lstats[i].curlextents == 0);
+			}
+		}
+
+		/* Merge extents stats. */
+		for (i = 0; i < SC_NPSIZES; i++) {
+			accum_atomic_zu(&sdstats->estats[i].ndirty,
+			    &astats->estats[i].ndirty);
+			accum_atomic_zu(&sdstats->estats[i].nmuzzy,
+			    &astats->estats[i].nmuzzy);
+			accum_atomic_zu(&sdstats->estats[i].nretained,
+			    &astats->estats[i].nretained);
+			accum_atomic_zu(&sdstats->estats[i].dirty_bytes,
+			    &astats->estats[i].dirty_bytes);
+			accum_atomic_zu(&sdstats->estats[i].muzzy_bytes,
+			    &astats->estats[i].muzzy_bytes);
+			accum_atomic_zu(&sdstats->estats[i].retained_bytes,
+			    &astats->estats[i].retained_bytes);
+		}
+	}
+}
+
+static void
+ctl_arena_refresh(tsdn_t *tsdn, arena_t *arena, ctl_arena_t *ctl_sdarena,
+    unsigned i, bool destroyed) {
+	ctl_arena_t *ctl_arena = arenas_i(i);
+
+	ctl_arena_clear(ctl_arena);
+	ctl_arena_stats_amerge(tsdn, ctl_arena, arena);
+	/* Merge into sum stats as well. */
+	ctl_arena_stats_sdmerge(ctl_sdarena, ctl_arena, destroyed);
+}
+
+static unsigned
+ctl_arena_init(tsd_t *tsd, extent_hooks_t *extent_hooks) {
+	unsigned arena_ind;
+	ctl_arena_t *ctl_arena;
+
+	if ((ctl_arena = ql_last(&ctl_arenas->destroyed, destroyed_link)) !=
+	    NULL) {
+		ql_remove(&ctl_arenas->destroyed, ctl_arena, destroyed_link);
+		arena_ind = ctl_arena->arena_ind;
+	} else {
+		arena_ind = ctl_arenas->narenas;
+	}
+
+	/* Trigger stats allocation. */
+	if (arenas_i_impl(tsd, arena_ind, false, true) == NULL) {
+		return UINT_MAX;
+	}
+
+	/* Initialize new arena. */
+	if (arena_init(tsd_tsdn(tsd), arena_ind, extent_hooks) == NULL) {
+		return UINT_MAX;
+	}
+
+	if (arena_ind == ctl_arenas->narenas) {
+		ctl_arenas->narenas++;
+	}
+
+	return arena_ind;
+}
+
+static void
+ctl_background_thread_stats_read(tsdn_t *tsdn) {
+	background_thread_stats_t *stats = &ctl_stats->background_thread;
+	if (!have_background_thread ||
+	    background_thread_stats_read(tsdn, stats)) {
+		memset(stats, 0, sizeof(background_thread_stats_t));
+		nstime_init(&stats->run_interval, 0);
+	}
+}
+
+static void
+ctl_refresh(tsdn_t *tsdn) {
+	unsigned i;
+	ctl_arena_t *ctl_sarena = arenas_i(MALLCTL_ARENAS_ALL);
+	VARIABLE_ARRAY(arena_t *, tarenas, ctl_arenas->narenas);
+
+	/*
+	 * Clear sum stats, since they will be merged into by
+	 * ctl_arena_refresh().
+	 */
+	ctl_arena_clear(ctl_sarena);
+
+	for (i = 0; i < ctl_arenas->narenas; i++) {
+		tarenas[i] = arena_get(tsdn, i, false);
+	}
+
+	for (i = 0; i < ctl_arenas->narenas; i++) {
+		ctl_arena_t *ctl_arena = arenas_i(i);
+		bool initialized = (tarenas[i] != NULL);
+
+		ctl_arena->initialized = initialized;
+		if (initialized) {
+			ctl_arena_refresh(tsdn, tarenas[i], ctl_sarena, i,
+			    false);
+		}
+	}
+
+	if (config_stats) {
+		ctl_stats->allocated = ctl_sarena->astats->allocated_small +
+		    atomic_load_zu(&ctl_sarena->astats->astats.allocated_large,
+			ATOMIC_RELAXED);
+		ctl_stats->active = (ctl_sarena->pactive << LG_PAGE);
+		ctl_stats->metadata = atomic_load_zu(
+		    &ctl_sarena->astats->astats.base, ATOMIC_RELAXED) +
+		    atomic_load_zu(&ctl_sarena->astats->astats.internal,
+			ATOMIC_RELAXED);
+		ctl_stats->metadata_thp = atomic_load_zu(
+		    &ctl_sarena->astats->astats.metadata_thp, ATOMIC_RELAXED);
+		ctl_stats->resident = atomic_load_zu(
+		    &ctl_sarena->astats->astats.resident, ATOMIC_RELAXED);
+		ctl_stats->mapped = atomic_load_zu(
+		    &ctl_sarena->astats->astats.mapped, ATOMIC_RELAXED);
+		ctl_stats->retained = atomic_load_zu(
+		    &ctl_sarena->astats->astats.retained, ATOMIC_RELAXED);
+
+		ctl_background_thread_stats_read(tsdn);
+
+#define READ_GLOBAL_MUTEX_PROF_DATA(i, mtx)				\
+    malloc_mutex_lock(tsdn, &mtx);					\
+    malloc_mutex_prof_read(tsdn, &ctl_stats->mutex_prof_data[i], &mtx);	\
+    malloc_mutex_unlock(tsdn, &mtx);
+
+		if (config_prof && opt_prof) {
+			READ_GLOBAL_MUTEX_PROF_DATA(global_prof_mutex_prof,
+			    bt2gctx_mtx);
+		}
+		if (have_background_thread) {
+			READ_GLOBAL_MUTEX_PROF_DATA(
+			    global_prof_mutex_background_thread,
+			    background_thread_lock);
+		} else {
+			memset(&ctl_stats->mutex_prof_data[
+			    global_prof_mutex_background_thread], 0,
+			    sizeof(mutex_prof_data_t));
+		}
+		/* We own ctl mutex already. */
+		malloc_mutex_prof_read(tsdn,
+		    &ctl_stats->mutex_prof_data[global_prof_mutex_ctl],
+		    &ctl_mtx);
+#undef READ_GLOBAL_MUTEX_PROF_DATA
+	}
+	ctl_arenas->epoch++;
+}
+
+static bool
+ctl_init(tsd_t *tsd) {
+	bool ret;
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	if (!ctl_initialized) {
+		ctl_arena_t *ctl_sarena, *ctl_darena;
+		unsigned i;
+
+		/*
+		 * Allocate demand-zeroed space for pointers to the full
+		 * range of supported arena indices.
+		 */
+		if (ctl_arenas == NULL) {
+			ctl_arenas = (ctl_arenas_t *)base_alloc(tsdn,
+			    b0get(), sizeof(ctl_arenas_t), QUANTUM);
+			if (ctl_arenas == NULL) {
+				ret = true;
+				goto label_return;
+			}
+		}
+
+		if (config_stats && ctl_stats == NULL) {
+			ctl_stats = (ctl_stats_t *)base_alloc(tsdn, b0get(),
+			    sizeof(ctl_stats_t), QUANTUM);
+			if (ctl_stats == NULL) {
+				ret = true;
+				goto label_return;
+			}
+		}
+
+		/*
+		 * Allocate space for the current full range of arenas
+		 * here rather than doing it lazily elsewhere, in order
+		 * to limit when OOM-caused errors can occur.
+		 */
+		if ((ctl_sarena = arenas_i_impl(tsd, MALLCTL_ARENAS_ALL, false,
+		    true)) == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		ctl_sarena->initialized = true;
+
+		if ((ctl_darena = arenas_i_impl(tsd, MALLCTL_ARENAS_DESTROYED,
+		    false, true)) == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		ctl_arena_clear(ctl_darena);
+		/*
+		 * Don't toggle ctl_darena to initialized until an arena is
+		 * actually destroyed, so that arena.<i>.initialized can be used
+		 * to query whether the stats are relevant.
+		 */
+
+		ctl_arenas->narenas = narenas_total_get();
+		for (i = 0; i < ctl_arenas->narenas; i++) {
+			if (arenas_i_impl(tsd, i, false, true) == NULL) {
+				ret = true;
+				goto label_return;
+			}
+		}
+
+		ql_new(&ctl_arenas->destroyed);
+		ctl_refresh(tsdn);
+
+		ctl_initialized = true;
+	}
+
+	ret = false;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+static int
+ctl_lookup(tsdn_t *tsdn, const char *name, ctl_node_t const **nodesp,
+    size_t *mibp, size_t *depthp) {
+	int ret;
+	const char *elm, *tdot, *dot;
+	size_t elen, i, j;
+	const ctl_named_node_t *node;
+
+	elm = name;
+	/* Equivalent to strchrnul(). */
+	dot = ((tdot = strchr(elm, '.')) != NULL) ? tdot : strchr(elm, '\0');
+	elen = (size_t)((uintptr_t)dot - (uintptr_t)elm);
+	if (elen == 0) {
+		ret = ENOENT;
+		goto label_return;
+	}
+	node = super_root_node;
+	for (i = 0; i < *depthp; i++) {
+		assert(node);
+		assert(node->nchildren > 0);
+		if (ctl_named_node(node->children) != NULL) {
+			const ctl_named_node_t *pnode = node;
+
+			/* Children are named. */
+			for (j = 0; j < node->nchildren; j++) {
+				const ctl_named_node_t *child =
+				    ctl_named_children(node, j);
+				if (strlen(child->name) == elen &&
+				    strncmp(elm, child->name, elen) == 0) {
+					node = child;
+					if (nodesp != NULL) {
+						nodesp[i] =
+						    (const ctl_node_t *)node;
+					}
+					mibp[i] = j;
+					break;
+				}
+			}
+			if (node == pnode) {
+				ret = ENOENT;
+				goto label_return;
+			}
+		} else {
+			uintmax_t index;
+			const ctl_indexed_node_t *inode;
+
+			/* Children are indexed. */
+			index = malloc_strtoumax(elm, NULL, 10);
+			if (index == UINTMAX_MAX || index > SIZE_T_MAX) {
+				ret = ENOENT;
+				goto label_return;
+			}
+
+			inode = ctl_indexed_node(node->children);
+			node = inode->index(tsdn, mibp, *depthp, (size_t)index);
+			if (node == NULL) {
+				ret = ENOENT;
+				goto label_return;
+			}
+
+			if (nodesp != NULL) {
+				nodesp[i] = (const ctl_node_t *)node;
+			}
+			mibp[i] = (size_t)index;
+		}
+
+		if (node->ctl != NULL) {
+			/* Terminal node. */
+			if (*dot != '\0') {
+				/*
+				 * The name contains more elements than are
+				 * in this path through the tree.
+				 */
+				ret = ENOENT;
+				goto label_return;
+			}
+			/* Complete lookup successful. */
+			*depthp = i + 1;
+			break;
+		}
+
+		/* Update elm. */
+		if (*dot == '\0') {
+			/* No more elements. */
+			ret = ENOENT;
+			goto label_return;
+		}
+		elm = &dot[1];
+		dot = ((tdot = strchr(elm, '.')) != NULL) ? tdot :
+		    strchr(elm, '\0');
+		elen = (size_t)((uintptr_t)dot - (uintptr_t)elm);
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+int
+ctl_byname(tsd_t *tsd, const char *name, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen) {
+	int ret;
+	size_t depth;
+	ctl_node_t const *nodes[CTL_MAX_DEPTH];
+	size_t mib[CTL_MAX_DEPTH];
+	const ctl_named_node_t *node;
+
+	if (!ctl_initialized && ctl_init(tsd)) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	depth = CTL_MAX_DEPTH;
+	ret = ctl_lookup(tsd_tsdn(tsd), name, nodes, mib, &depth);
+	if (ret != 0) {
+		goto label_return;
+	}
+
+	node = ctl_named_node(nodes[depth-1]);
+	if (node != NULL && node->ctl) {
+		ret = node->ctl(tsd, mib, depth, oldp, oldlenp, newp, newlen);
+	} else {
+		/* The name refers to a partial path through the ctl tree. */
+		ret = ENOENT;
+	}
+
+label_return:
+	return(ret);
+}
+
+int
+ctl_nametomib(tsd_t *tsd, const char *name, size_t *mibp, size_t *miblenp) {
+	int ret;
+
+	if (!ctl_initialized && ctl_init(tsd)) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	ret = ctl_lookup(tsd_tsdn(tsd), name, NULL, mibp, miblenp);
+label_return:
+	return(ret);
+}
+
+int
+ctl_bymib(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	const ctl_named_node_t *node;
+	size_t i;
+
+	if (!ctl_initialized && ctl_init(tsd)) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+
+	/* Iterate down the tree. */
+	node = super_root_node;
+	for (i = 0; i < miblen; i++) {
+		assert(node);
+		assert(node->nchildren > 0);
+		if (ctl_named_node(node->children) != NULL) {
+			/* Children are named. */
+			if (node->nchildren <= mib[i]) {
+				ret = ENOENT;
+				goto label_return;
+			}
+			node = ctl_named_children(node, mib[i]);
+		} else {
+			const ctl_indexed_node_t *inode;
+
+			/* Indexed element. */
+			inode = ctl_indexed_node(node->children);
+			node = inode->index(tsd_tsdn(tsd), mib, miblen, mib[i]);
+			if (node == NULL) {
+				ret = ENOENT;
+				goto label_return;
+			}
+		}
+	}
+
+	/* Call the ctl function. */
+	if (node && node->ctl) {
+		ret = node->ctl(tsd, mib, miblen, oldp, oldlenp, newp, newlen);
+	} else {
+		/* Partial MIB. */
+		ret = ENOENT;
+	}
+
+label_return:
+	return(ret);
+}
+
+bool
+ctl_boot(void) {
+	if (malloc_mutex_init(&ctl_mtx, "ctl", WITNESS_RANK_CTL,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	ctl_initialized = false;
+
+	return false;
+}
+
+void
+ctl_prefork(tsdn_t *tsdn) {
+	malloc_mutex_prefork(tsdn, &ctl_mtx);
+}
+
+void
+ctl_postfork_parent(tsdn_t *tsdn) {
+	malloc_mutex_postfork_parent(tsdn, &ctl_mtx);
+}
+
+void
+ctl_postfork_child(tsdn_t *tsdn) {
+	malloc_mutex_postfork_child(tsdn, &ctl_mtx);
+}
+
+/******************************************************************************/
+/* *_ctl() functions. */
+
+#define READONLY()	do {						\
+	if (newp != NULL || newlen != 0) {				\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define WRITEONLY()	do {						\
+	if (oldp != NULL || oldlenp != NULL) {				\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define READ_XOR_WRITE()	do {					\
+	if ((oldp != NULL && oldlenp != NULL) && (newp != NULL ||	\
+	    newlen != 0)) {						\
+		ret = EPERM;						\
+		goto label_return;					\
+	}								\
+} while (0)
+
+#define READ(v, t)	do {						\
+	if (oldp != NULL && oldlenp != NULL) {				\
+		if (*oldlenp != sizeof(t)) {				\
+			size_t	copylen = (sizeof(t) <= *oldlenp)	\
+			    ? sizeof(t) : *oldlenp;			\
+			memcpy(oldp, (void *)&(v), copylen);		\
+			ret = EINVAL;					\
+			goto label_return;				\
+		}							\
+		*(t *)oldp = (v);					\
+	}								\
+} while (0)
+
+#define WRITE(v, t)	do {						\
+	if (newp != NULL) {						\
+		if (newlen != sizeof(t)) {				\
+			ret = EINVAL;					\
+			goto label_return;				\
+		}							\
+		(v) = *(t *)newp;					\
+	}								\
+} while (0)
+
+#define MIB_UNSIGNED(v, i) do {						\
+	if (mib[i] > UINT_MAX) {					\
+		ret = EFAULT;						\
+		goto label_return;					\
+	}								\
+	v = (unsigned)mib[i];						\
+} while (0)
+
+/*
+ * There's a lot of code duplication in the following macros due to limitations
+ * in how nested cpp macros are expanded.
+ */
+#define CTL_RO_CLGEN(c, l, n, v, t)					\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,	\
+    size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	if (l) {							\
+		malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);		\
+	}								\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	if (l) {							\
+		malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);		\
+	}								\
+	return ret;							\
+}
+
+#define CTL_RO_CGEN(c, n, v, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);			\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);			\
+	return ret;							\
+}
+
+#define CTL_RO_GEN(n, v, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,	\
+    size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);			\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);			\
+	return ret;							\
+}
+
+/*
+ * ctl_mtx is not acquired, under the assumption that no pertinent data will
+ * mutate during the call.
+ */
+#define CTL_RO_NL_CGEN(c, n, v, t)					\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+#define CTL_RO_NL_GEN(n, v, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	READONLY();							\
+	oldval = (v);							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+#define CTL_TSD_RO_NL_CGEN(c, n, m, t)					\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,	\
+    size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	if (!(c)) {							\
+		return ENOENT;						\
+	}								\
+	READONLY();							\
+	oldval = (m(tsd));						\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+#define CTL_RO_CONFIG_GEN(n, t)						\
+static int								\
+n##_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, \
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {			\
+	int ret;							\
+	t oldval;							\
+									\
+	READONLY();							\
+	oldval = n;							\
+	READ(oldval, t);						\
+									\
+	ret = 0;							\
+label_return:								\
+	return ret;							\
+}
+
+/******************************************************************************/
+
+CTL_RO_NL_GEN(version, JEMALLOC_VERSION, const char *)
+
+static int
+epoch_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	UNUSED uint64_t newval;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	WRITE(newval, uint64_t);
+	if (newp != NULL) {
+		ctl_refresh(tsd_tsdn(tsd));
+	}
+	READ(ctl_arenas->epoch, uint64_t);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+background_thread_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!have_background_thread) {
+		return ENOENT;
+	}
+	background_thread_ctl_init(tsd_tsdn(tsd));
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+	if (newp == NULL) {
+		oldval = background_thread_enabled();
+		READ(oldval, bool);
+	} else {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = background_thread_enabled();
+		READ(oldval, bool);
+
+		bool newval = *(bool *)newp;
+		if (newval == oldval) {
+			ret = 0;
+			goto label_return;
+		}
+
+		background_thread_enabled_set(tsd_tsdn(tsd), newval);
+		if (newval) {
+			if (background_threads_enable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		} else {
+			if (background_threads_disable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		}
+	}
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+
+	return ret;
+}
+
+static int
+max_background_threads_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	size_t oldval;
+
+	if (!have_background_thread) {
+		return ENOENT;
+	}
+	background_thread_ctl_init(tsd_tsdn(tsd));
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+	if (newp == NULL) {
+		oldval = max_background_threads;
+		READ(oldval, size_t);
+	} else {
+		if (newlen != sizeof(size_t)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = max_background_threads;
+		READ(oldval, size_t);
+
+		size_t newval = *(size_t *)newp;
+		if (newval == oldval) {
+			ret = 0;
+			goto label_return;
+		}
+		if (newval > opt_max_background_threads) {
+			ret = EINVAL;
+			goto label_return;
+		}
+
+		if (background_thread_enabled()) {
+			background_thread_enabled_set(tsd_tsdn(tsd), false);
+			if (background_threads_disable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+			max_background_threads = newval;
+			background_thread_enabled_set(tsd_tsdn(tsd), true);
+			if (background_threads_enable(tsd)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		} else {
+			max_background_threads = newval;
+		}
+	}
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+
+	return ret;
+}
+
+/******************************************************************************/
+
+CTL_RO_CONFIG_GEN(config_cache_oblivious, bool)
+CTL_RO_CONFIG_GEN(config_debug, bool)
+CTL_RO_CONFIG_GEN(config_fill, bool)
+CTL_RO_CONFIG_GEN(config_lazy_lock, bool)
+CTL_RO_CONFIG_GEN(config_malloc_conf, const char *)
+CTL_RO_CONFIG_GEN(config_opt_safety_checks, bool)
+CTL_RO_CONFIG_GEN(config_prof, bool)
+CTL_RO_CONFIG_GEN(config_prof_libgcc, bool)
+CTL_RO_CONFIG_GEN(config_prof_libunwind, bool)
+CTL_RO_CONFIG_GEN(config_stats, bool)
+CTL_RO_CONFIG_GEN(config_utrace, bool)
+CTL_RO_CONFIG_GEN(config_xmalloc, bool)
+
+/******************************************************************************/
+
+CTL_RO_NL_GEN(opt_abort, opt_abort, bool)
+CTL_RO_NL_GEN(opt_abort_conf, opt_abort_conf, bool)
+CTL_RO_NL_GEN(opt_confirm_conf, opt_confirm_conf, bool)
+CTL_RO_NL_GEN(opt_metadata_thp, metadata_thp_mode_names[opt_metadata_thp],
+    const char *)
+CTL_RO_NL_GEN(opt_retain, opt_retain, bool)
+CTL_RO_NL_GEN(opt_dss, opt_dss, const char *)
+CTL_RO_NL_GEN(opt_narenas, opt_narenas, unsigned)
+CTL_RO_NL_GEN(opt_percpu_arena, percpu_arena_mode_names[opt_percpu_arena],
+    const char *)
+CTL_RO_NL_GEN(opt_oversize_threshold, opt_oversize_threshold, size_t)
+CTL_RO_NL_GEN(opt_background_thread, opt_background_thread, bool)
+CTL_RO_NL_GEN(opt_max_background_threads, opt_max_background_threads, size_t)
+CTL_RO_NL_GEN(opt_dirty_decay_ms, opt_dirty_decay_ms, ssize_t)
+CTL_RO_NL_GEN(opt_muzzy_decay_ms, opt_muzzy_decay_ms, ssize_t)
+CTL_RO_NL_GEN(opt_stats_print, opt_stats_print, bool)
+CTL_RO_NL_GEN(opt_stats_print_opts, opt_stats_print_opts, const char *)
+CTL_RO_NL_CGEN(config_fill, opt_junk, opt_junk, const char *)
+CTL_RO_NL_CGEN(config_fill, opt_zero, opt_zero, bool)
+CTL_RO_NL_CGEN(config_utrace, opt_utrace, opt_utrace, bool)
+CTL_RO_NL_CGEN(config_xmalloc, opt_xmalloc, opt_xmalloc, bool)
+CTL_RO_NL_GEN(opt_tcache, opt_tcache, bool)
+CTL_RO_NL_GEN(opt_thp, thp_mode_names[opt_thp], const char *)
+CTL_RO_NL_GEN(opt_lg_extent_max_active_fit, opt_lg_extent_max_active_fit,
+    size_t)
+CTL_RO_NL_GEN(opt_lg_tcache_max, opt_lg_tcache_max, ssize_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof, opt_prof, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_prefix, opt_prof_prefix, const char *)
+CTL_RO_NL_CGEN(config_prof, opt_prof_active, opt_prof_active, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_thread_active_init,
+    opt_prof_thread_active_init, bool)
+CTL_RO_NL_CGEN(config_prof, opt_lg_prof_sample, opt_lg_prof_sample, size_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof_accum, opt_prof_accum, bool)
+CTL_RO_NL_CGEN(config_prof, opt_lg_prof_interval, opt_lg_prof_interval, ssize_t)
+CTL_RO_NL_CGEN(config_prof, opt_prof_gdump, opt_prof_gdump, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_final, opt_prof_final, bool)
+CTL_RO_NL_CGEN(config_prof, opt_prof_leak, opt_prof_leak, bool)
+
+/******************************************************************************/
+
+static int
+thread_arena_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	arena_t *oldarena;
+	unsigned newind, oldind;
+
+	oldarena = arena_choose(tsd, NULL);
+	if (oldarena == NULL) {
+		return EAGAIN;
+	}
+	newind = oldind = arena_ind_get(oldarena);
+	WRITE(newind, unsigned);
+	READ(oldind, unsigned);
+
+	if (newind != oldind) {
+		arena_t *newarena;
+
+		if (newind >= narenas_total_get()) {
+			/* New arena index is out of range. */
+			ret = EFAULT;
+			goto label_return;
+		}
+
+		if (have_percpu_arena &&
+		    PERCPU_ARENA_ENABLED(opt_percpu_arena)) {
+			if (newind < percpu_arena_ind_limit(opt_percpu_arena)) {
+				/*
+				 * If perCPU arena is enabled, thread_arena
+				 * control is not allowed for the auto arena
+				 * range.
+				 */
+				ret = EPERM;
+				goto label_return;
+			}
+		}
+
+		/* Initialize arena if necessary. */
+		newarena = arena_get(tsd_tsdn(tsd), newind, true);
+		if (newarena == NULL) {
+			ret = EAGAIN;
+			goto label_return;
+		}
+		/* Set new arena/tcache associations. */
+		arena_migrate(tsd, oldind, newind);
+		if (tcache_available(tsd)) {
+			tcache_arena_reassociate(tsd_tsdn(tsd),
+			    tsd_tcachep_get(tsd), newarena);
+		}
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+CTL_TSD_RO_NL_CGEN(config_stats, thread_allocated, tsd_thread_allocated_get,
+    uint64_t)
+CTL_TSD_RO_NL_CGEN(config_stats, thread_allocatedp, tsd_thread_allocatedp_get,
+    uint64_t *)
+CTL_TSD_RO_NL_CGEN(config_stats, thread_deallocated, tsd_thread_deallocated_get,
+    uint64_t)
+CTL_TSD_RO_NL_CGEN(config_stats, thread_deallocatedp,
+    tsd_thread_deallocatedp_get, uint64_t *)
+
+static int
+thread_tcache_enabled_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	bool oldval;
+
+	oldval = tcache_enabled_get(tsd);
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		tcache_enabled_set(tsd, *(bool *)newp);
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+thread_tcache_flush_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+
+	if (!tcache_available(tsd)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	READONLY();
+	WRITEONLY();
+
+	tcache_flush(tsd);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+thread_prof_name_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	READ_XOR_WRITE();
+
+	if (newp != NULL) {
+		if (newlen != sizeof(const char *)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+
+		if ((ret = prof_thread_name_set(tsd, *(const char **)newp)) !=
+		    0) {
+			goto label_return;
+		}
+	} else {
+		const char *oldname = prof_thread_name_get(tsd);
+		READ(oldname, const char *);
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+thread_prof_active_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	oldval = prof_thread_active_get(tsd);
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		if (prof_thread_active_set(tsd, *(bool *)newp)) {
+			ret = EAGAIN;
+			goto label_return;
+		}
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+tcache_create_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned tcache_ind;
+
+	READONLY();
+	if (tcaches_create(tsd, &tcache_ind)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+	READ(tcache_ind, unsigned);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+tcache_flush_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned tcache_ind;
+
+	WRITEONLY();
+	tcache_ind = UINT_MAX;
+	WRITE(tcache_ind, unsigned);
+	if (tcache_ind == UINT_MAX) {
+		ret = EFAULT;
+		goto label_return;
+	}
+	tcaches_flush(tsd, tcache_ind);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+tcache_destroy_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned tcache_ind;
+
+	WRITEONLY();
+	tcache_ind = UINT_MAX;
+	WRITE(tcache_ind, unsigned);
+	if (tcache_ind == UINT_MAX) {
+		ret = EFAULT;
+		goto label_return;
+	}
+	tcaches_destroy(tsd, tcache_ind);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+arena_i_initialized_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+	unsigned arena_ind;
+	bool initialized;
+
+	READONLY();
+	MIB_UNSIGNED(arena_ind, 1);
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	initialized = arenas_i(arena_ind)->initialized;
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+
+	READ(initialized, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static void
+arena_i_decay(tsdn_t *tsdn, unsigned arena_ind, bool all) {
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	{
+		unsigned narenas = ctl_arenas->narenas;
+
+		/*
+		 * Access via index narenas is deprecated, and scheduled for
+		 * removal in 6.0.0.
+		 */
+		if (arena_ind == MALLCTL_ARENAS_ALL || arena_ind == narenas) {
+			unsigned i;
+			VARIABLE_ARRAY(arena_t *, tarenas, narenas);
+
+			for (i = 0; i < narenas; i++) {
+				tarenas[i] = arena_get(tsdn, i, false);
+			}
+
+			/*
+			 * No further need to hold ctl_mtx, since narenas and
+			 * tarenas contain everything needed below.
+			 */
+			malloc_mutex_unlock(tsdn, &ctl_mtx);
+
+			for (i = 0; i < narenas; i++) {
+				if (tarenas[i] != NULL) {
+					arena_decay(tsdn, tarenas[i], false,
+					    all);
+				}
+			}
+		} else {
+			arena_t *tarena;
+
+			assert(arena_ind < narenas);
+
+			tarena = arena_get(tsdn, arena_ind, false);
+
+			/* No further need to hold ctl_mtx. */
+			malloc_mutex_unlock(tsdn, &ctl_mtx);
+
+			if (tarena != NULL) {
+				arena_decay(tsdn, tarena, false, all);
+			}
+		}
+	}
+}
+
+static int
+arena_i_decay_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+
+	READONLY();
+	WRITEONLY();
+	MIB_UNSIGNED(arena_ind, 1);
+	arena_i_decay(tsd_tsdn(tsd), arena_ind, false);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arena_i_purge_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+
+	READONLY();
+	WRITEONLY();
+	MIB_UNSIGNED(arena_ind, 1);
+	arena_i_decay(tsd_tsdn(tsd), arena_ind, true);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arena_i_reset_destroy_helper(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen, unsigned *arena_ind,
+    arena_t **arena) {
+	int ret;
+
+	READONLY();
+	WRITEONLY();
+	MIB_UNSIGNED(*arena_ind, 1);
+
+	*arena = arena_get(tsd_tsdn(tsd), *arena_ind, false);
+	if (*arena == NULL || arena_is_auto(*arena)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static void
+arena_reset_prepare_background_thread(tsd_t *tsd, unsigned arena_ind) {
+	/* Temporarily disable the background thread during arena reset. */
+	if (have_background_thread) {
+		malloc_mutex_lock(tsd_tsdn(tsd), &background_thread_lock);
+		if (background_thread_enabled()) {
+			background_thread_info_t *info =
+			    background_thread_info_get(arena_ind);
+			assert(info->state == background_thread_started);
+			malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+			info->state = background_thread_paused;
+			malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		}
+	}
+}
+
+static void
+arena_reset_finish_background_thread(tsd_t *tsd, unsigned arena_ind) {
+	if (have_background_thread) {
+		if (background_thread_enabled()) {
+			background_thread_info_t *info =
+			    background_thread_info_get(arena_ind);
+			assert(info->state == background_thread_paused);
+			malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+			info->state = background_thread_started;
+			malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &background_thread_lock);
+	}
+}
+
+static int
+arena_i_reset_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	ret = arena_i_reset_destroy_helper(tsd, mib, miblen, oldp, oldlenp,
+	    newp, newlen, &arena_ind, &arena);
+	if (ret != 0) {
+		return ret;
+	}
+
+	arena_reset_prepare_background_thread(tsd, arena_ind);
+	arena_reset(tsd, arena);
+	arena_reset_finish_background_thread(tsd, arena_ind);
+
+	return ret;
+}
+
+static int
+arena_i_destroy_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+	ctl_arena_t *ctl_darena, *ctl_arena;
+
+	ret = arena_i_reset_destroy_helper(tsd, mib, miblen, oldp, oldlenp,
+	    newp, newlen, &arena_ind, &arena);
+	if (ret != 0) {
+		goto label_return;
+	}
+
+	if (arena_nthreads_get(arena, false) != 0 || arena_nthreads_get(arena,
+	    true) != 0) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	arena_reset_prepare_background_thread(tsd, arena_ind);
+	/* Merge stats after resetting and purging arena. */
+	arena_reset(tsd, arena);
+	arena_decay(tsd_tsdn(tsd), arena, false, true);
+	ctl_darena = arenas_i(MALLCTL_ARENAS_DESTROYED);
+	ctl_darena->initialized = true;
+	ctl_arena_refresh(tsd_tsdn(tsd), arena, ctl_darena, arena_ind, true);
+	/* Destroy arena. */
+	arena_destroy(tsd, arena);
+	ctl_arena = arenas_i(arena_ind);
+	ctl_arena->initialized = false;
+	/* Record arena index for later recycling via arenas.create. */
+	ql_elm_new(ctl_arena, destroyed_link);
+	ql_tail_insert(&ctl_arenas->destroyed, ctl_arena, destroyed_link);
+	arena_reset_finish_background_thread(tsd, arena_ind);
+
+	assert(ret == 0);
+label_return:
+	return ret;
+}
+
+static int
+arena_i_dss_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	const char *dss = NULL;
+	unsigned arena_ind;
+	dss_prec_t dss_prec_old = dss_prec_limit;
+	dss_prec_t dss_prec = dss_prec_limit;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	WRITE(dss, const char *);
+	MIB_UNSIGNED(arena_ind, 1);
+	if (dss != NULL) {
+		int i;
+		bool match = false;
+
+		for (i = 0; i < dss_prec_limit; i++) {
+			if (strcmp(dss_prec_names[i], dss) == 0) {
+				dss_prec = i;
+				match = true;
+				break;
+			}
+		}
+
+		if (!match) {
+			ret = EINVAL;
+			goto label_return;
+		}
+	}
+
+	/*
+	 * Access via index narenas is deprecated, and scheduled for removal in
+	 * 6.0.0.
+	 */
+	if (arena_ind == MALLCTL_ARENAS_ALL || arena_ind ==
+	    ctl_arenas->narenas) {
+		if (dss_prec != dss_prec_limit &&
+		    extent_dss_prec_set(dss_prec)) {
+			ret = EFAULT;
+			goto label_return;
+		}
+		dss_prec_old = extent_dss_prec_get();
+	} else {
+		arena_t *arena = arena_get(tsd_tsdn(tsd), arena_ind, false);
+		if (arena == NULL || (dss_prec != dss_prec_limit &&
+		    arena_dss_prec_set(arena, dss_prec))) {
+			ret = EFAULT;
+			goto label_return;
+		}
+		dss_prec_old = arena_dss_prec_get(arena);
+	}
+
+	dss = dss_prec_names[dss_prec_old];
+	READ(dss, const char *);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arena_i_decay_ms_ctl_impl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen, bool dirty) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	MIB_UNSIGNED(arena_ind, 1);
+	arena = arena_get(tsd_tsdn(tsd), arena_ind, false);
+	if (arena == NULL) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	if (oldp != NULL && oldlenp != NULL) {
+		size_t oldval = dirty ? arena_dirty_decay_ms_get(arena) :
+		    arena_muzzy_decay_ms_get(arena);
+		READ(oldval, ssize_t);
+	}
+	if (newp != NULL) {
+		if (newlen != sizeof(ssize_t)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		if (arena_is_huge(arena_ind) && *(ssize_t *)newp > 0) {
+			/*
+			 * By default the huge arena purges eagerly.  If it is
+			 * set to non-zero decay time afterwards, background
+			 * thread might be needed.
+			 */
+			if (background_thread_create(tsd, arena_ind)) {
+				ret = EFAULT;
+				goto label_return;
+			}
+		}
+		if (dirty ? arena_dirty_decay_ms_set(tsd_tsdn(tsd), arena,
+		    *(ssize_t *)newp) : arena_muzzy_decay_ms_set(tsd_tsdn(tsd),
+		    arena, *(ssize_t *)newp)) {
+			ret = EFAULT;
+			goto label_return;
+		}
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arena_i_dirty_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arena_i_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, true);
+}
+
+static int
+arena_i_muzzy_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arena_i_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, false);
+}
+
+static int
+arena_i_extent_hooks_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	MIB_UNSIGNED(arena_ind, 1);
+	if (arena_ind < narenas_total_get()) {
+		extent_hooks_t *old_extent_hooks;
+		arena = arena_get(tsd_tsdn(tsd), arena_ind, false);
+		if (arena == NULL) {
+			if (arena_ind >= narenas_auto) {
+				ret = EFAULT;
+				goto label_return;
+			}
+			old_extent_hooks =
+			    (extent_hooks_t *)&extent_hooks_default;
+			READ(old_extent_hooks, extent_hooks_t *);
+			if (newp != NULL) {
+				/* Initialize a new arena as a side effect. */
+				extent_hooks_t *new_extent_hooks
+				    JEMALLOC_CC_SILENCE_INIT(NULL);
+				WRITE(new_extent_hooks, extent_hooks_t *);
+				arena = arena_init(tsd_tsdn(tsd), arena_ind,
+				    new_extent_hooks);
+				if (arena == NULL) {
+					ret = EFAULT;
+					goto label_return;
+				}
+			}
+		} else {
+			if (newp != NULL) {
+				extent_hooks_t *new_extent_hooks
+				    JEMALLOC_CC_SILENCE_INIT(NULL);
+				WRITE(new_extent_hooks, extent_hooks_t *);
+				old_extent_hooks = extent_hooks_set(tsd, arena,
+				    new_extent_hooks);
+				READ(old_extent_hooks, extent_hooks_t *);
+			} else {
+				old_extent_hooks = extent_hooks_get(arena);
+				READ(old_extent_hooks, extent_hooks_t *);
+			}
+		}
+	} else {
+		ret = EFAULT;
+		goto label_return;
+	}
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arena_i_retain_grow_limit_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	arena_t *arena;
+
+	if (!opt_retain) {
+		/* Only relevant when retain is enabled. */
+		return ENOENT;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	MIB_UNSIGNED(arena_ind, 1);
+	if (arena_ind < narenas_total_get() && (arena =
+	    arena_get(tsd_tsdn(tsd), arena_ind, false)) != NULL) {
+		size_t old_limit, new_limit;
+		if (newp != NULL) {
+			WRITE(new_limit, size_t);
+		}
+		bool err = arena_retain_grow_limit_get_set(tsd, arena,
+		    &old_limit, newp != NULL ? &new_limit : NULL);
+		if (!err) {
+			READ(old_limit, size_t);
+			ret = 0;
+		} else {
+			ret = EFAULT;
+		}
+	} else {
+		ret = EFAULT;
+	}
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static const ctl_named_node_t *
+arena_i_index(tsdn_t *tsdn, const size_t *mib, size_t miblen,
+    size_t i) {
+	const ctl_named_node_t *ret;
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	switch (i) {
+	case MALLCTL_ARENAS_ALL:
+	case MALLCTL_ARENAS_DESTROYED:
+		break;
+	default:
+		if (i > ctl_arenas->narenas) {
+			ret = NULL;
+			goto label_return;
+		}
+		break;
+	}
+
+	ret = super_arena_i_node;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+arenas_narenas_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	unsigned narenas;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	READONLY();
+	if (*oldlenp != sizeof(unsigned)) {
+		ret = EINVAL;
+		goto label_return;
+	}
+	narenas = ctl_arenas->narenas;
+	READ(narenas, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arenas_decay_ms_ctl_impl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen, bool dirty) {
+	int ret;
+
+	if (oldp != NULL && oldlenp != NULL) {
+		size_t oldval = (dirty ? arena_dirty_decay_ms_default_get() :
+		    arena_muzzy_decay_ms_default_get());
+		READ(oldval, ssize_t);
+	}
+	if (newp != NULL) {
+		if (newlen != sizeof(ssize_t)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		if (dirty ? arena_dirty_decay_ms_default_set(*(ssize_t *)newp)
+		    : arena_muzzy_decay_ms_default_set(*(ssize_t *)newp)) {
+			ret = EFAULT;
+			goto label_return;
+		}
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+arenas_dirty_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arenas_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, true);
+}
+
+static int
+arenas_muzzy_decay_ms_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	return arenas_decay_ms_ctl_impl(tsd, mib, miblen, oldp, oldlenp, newp,
+	    newlen, false);
+}
+
+CTL_RO_NL_GEN(arenas_quantum, QUANTUM, size_t)
+CTL_RO_NL_GEN(arenas_page, PAGE, size_t)
+CTL_RO_NL_GEN(arenas_tcache_max, tcache_maxclass, size_t)
+CTL_RO_NL_GEN(arenas_nbins, SC_NBINS, unsigned)
+CTL_RO_NL_GEN(arenas_nhbins, nhbins, unsigned)
+CTL_RO_NL_GEN(arenas_bin_i_size, bin_infos[mib[2]].reg_size, size_t)
+CTL_RO_NL_GEN(arenas_bin_i_nregs, bin_infos[mib[2]].nregs, uint32_t)
+CTL_RO_NL_GEN(arenas_bin_i_slab_size, bin_infos[mib[2]].slab_size, size_t)
+CTL_RO_NL_GEN(arenas_bin_i_nshards, bin_infos[mib[2]].n_shards, uint32_t)
+static const ctl_named_node_t *
+arenas_bin_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	if (i > SC_NBINS) {
+		return NULL;
+	}
+	return super_arenas_bin_i_node;
+}
+
+CTL_RO_NL_GEN(arenas_nlextents, SC_NSIZES - SC_NBINS, unsigned)
+CTL_RO_NL_GEN(arenas_lextent_i_size, sz_index2size(SC_NBINS+(szind_t)mib[2]),
+    size_t)
+static const ctl_named_node_t *
+arenas_lextent_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	if (i > SC_NSIZES - SC_NBINS) {
+		return NULL;
+	}
+	return super_arenas_lextent_i_node;
+}
+
+static int
+arenas_create_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	extent_hooks_t *extent_hooks;
+	unsigned arena_ind;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+
+	extent_hooks = (extent_hooks_t *)&extent_hooks_default;
+	WRITE(extent_hooks, extent_hooks_t *);
+	if ((arena_ind = ctl_arena_init(tsd, extent_hooks)) == UINT_MAX) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+	READ(arena_ind, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+static int
+arenas_lookup_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	unsigned arena_ind;
+	void *ptr;
+	extent_t *extent;
+	arena_t *arena;
+
+	ptr = NULL;
+	ret = EINVAL;
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	WRITE(ptr, void *);
+	extent = iealloc(tsd_tsdn(tsd), ptr);
+	if (extent == NULL)
+		goto label_return;
+
+	arena = extent_arena_get(extent);
+	if (arena == NULL)
+		goto label_return;
+
+	arena_ind = arena_ind_get(arena);
+	READ(arena_ind, unsigned);
+
+	ret = 0;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
+
+/******************************************************************************/
+
+static int
+prof_thread_active_init_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = prof_thread_active_init_set(tsd_tsdn(tsd),
+		    *(bool *)newp);
+	} else {
+		oldval = prof_thread_active_init_get(tsd_tsdn(tsd));
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_active_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = prof_active_set(tsd_tsdn(tsd), *(bool *)newp);
+	} else {
+		oldval = prof_active_get(tsd_tsdn(tsd));
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_dump_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	const char *filename = NULL;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	WRITEONLY();
+	WRITE(filename, const char *);
+
+	if (prof_mdump(tsd, filename)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_gdump_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	bool oldval;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (newp != NULL) {
+		if (newlen != sizeof(bool)) {
+			ret = EINVAL;
+			goto label_return;
+		}
+		oldval = prof_gdump_set(tsd_tsdn(tsd), *(bool *)newp);
+	} else {
+		oldval = prof_gdump_get(tsd_tsdn(tsd));
+	}
+	READ(oldval, bool);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_reset_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	size_t lg_sample = lg_prof_sample;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	WRITEONLY();
+	WRITE(lg_sample, size_t);
+	if (lg_sample >= (sizeof(uint64_t) << 3)) {
+		lg_sample = (sizeof(uint64_t) << 3) - 1;
+	}
+
+	prof_reset(tsd, lg_sample);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+CTL_RO_NL_CGEN(config_prof, prof_interval, prof_interval, uint64_t)
+CTL_RO_NL_CGEN(config_prof, lg_prof_sample, lg_prof_sample, size_t)
+
+static int
+prof_log_start_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+
+	const char *filename = NULL;
+
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	WRITEONLY();
+	WRITE(filename, const char *);
+
+	if (prof_log_start(tsd_tsdn(tsd), filename)) {
+		ret = EFAULT;
+		goto label_return;
+	}
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+prof_log_stop_ctl(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen) {
+	if (!config_prof) {
+		return ENOENT;
+	}
+
+	if (prof_log_stop(tsd_tsdn(tsd))) {
+		return EFAULT;
+	}
+
+	return 0;
+}
+
+/******************************************************************************/
+
+CTL_RO_CGEN(config_stats, stats_allocated, ctl_stats->allocated, size_t)
+CTL_RO_CGEN(config_stats, stats_active, ctl_stats->active, size_t)
+CTL_RO_CGEN(config_stats, stats_metadata, ctl_stats->metadata, size_t)
+CTL_RO_CGEN(config_stats, stats_metadata_thp, ctl_stats->metadata_thp, size_t)
+CTL_RO_CGEN(config_stats, stats_resident, ctl_stats->resident, size_t)
+CTL_RO_CGEN(config_stats, stats_mapped, ctl_stats->mapped, size_t)
+CTL_RO_CGEN(config_stats, stats_retained, ctl_stats->retained, size_t)
+
+CTL_RO_CGEN(config_stats, stats_background_thread_num_threads,
+    ctl_stats->background_thread.num_threads, size_t)
+CTL_RO_CGEN(config_stats, stats_background_thread_num_runs,
+    ctl_stats->background_thread.num_runs, uint64_t)
+CTL_RO_CGEN(config_stats, stats_background_thread_run_interval,
+    nstime_ns(&ctl_stats->background_thread.run_interval), uint64_t)
+
+CTL_RO_GEN(stats_arenas_i_dss, arenas_i(mib[2])->dss, const char *)
+CTL_RO_GEN(stats_arenas_i_dirty_decay_ms, arenas_i(mib[2])->dirty_decay_ms,
+    ssize_t)
+CTL_RO_GEN(stats_arenas_i_muzzy_decay_ms, arenas_i(mib[2])->muzzy_decay_ms,
+    ssize_t)
+CTL_RO_GEN(stats_arenas_i_nthreads, arenas_i(mib[2])->nthreads, unsigned)
+CTL_RO_GEN(stats_arenas_i_uptime,
+    nstime_ns(&arenas_i(mib[2])->astats->astats.uptime), uint64_t)
+CTL_RO_GEN(stats_arenas_i_pactive, arenas_i(mib[2])->pactive, size_t)
+CTL_RO_GEN(stats_arenas_i_pdirty, arenas_i(mib[2])->pdirty, size_t)
+CTL_RO_GEN(stats_arenas_i_pmuzzy, arenas_i(mib[2])->pmuzzy, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_mapped,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.mapped, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_retained,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.retained, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_extent_avail,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.extent_avail,
+        ATOMIC_RELAXED),
+    size_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_dirty_npurge,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_dirty.npurge), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_dirty_nmadvise,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_dirty.nmadvise), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_dirty_purged,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_dirty.purged), uint64_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_muzzy_npurge,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_muzzy.npurge), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_muzzy_nmadvise,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_muzzy.nmadvise), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_muzzy_purged,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.decay_muzzy.purged), uint64_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_base,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.base, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_internal,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.internal, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_metadata_thp,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.metadata_thp,
+    ATOMIC_RELAXED), size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_tcache_bytes,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.tcache_bytes,
+    ATOMIC_RELAXED), size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_resident,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.resident, ATOMIC_RELAXED),
+    size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_abandoned_vm,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.abandoned_vm,
+    ATOMIC_RELAXED), size_t)
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_allocated,
+    arenas_i(mib[2])->astats->allocated_small, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nmalloc,
+    arenas_i(mib[2])->astats->nmalloc_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_ndalloc,
+    arenas_i(mib[2])->astats->ndalloc_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nrequests,
+    arenas_i(mib[2])->astats->nrequests_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nfills,
+    arenas_i(mib[2])->astats->nfills_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_small_nflushes,
+    arenas_i(mib[2])->astats->nflushes_small, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_allocated,
+    atomic_load_zu(&arenas_i(mib[2])->astats->astats.allocated_large,
+    ATOMIC_RELAXED), size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nmalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nmalloc_large), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_ndalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.ndalloc_large), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nrequests,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nrequests_large), uint64_t)
+/*
+ * Note: "nmalloc_large" here instead of "nfills" in the read.  This is
+ * intentional (large has no batch fill).
+ */
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nfills,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nmalloc_large), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_large_nflushes,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->astats.nflushes_large), uint64_t)
+
+/* Lock profiling related APIs below. */
+#define RO_MUTEX_CTL_GEN(n, l)						\
+CTL_RO_CGEN(config_stats, stats_##n##_num_ops,				\
+    l.n_lock_ops, uint64_t)						\
+CTL_RO_CGEN(config_stats, stats_##n##_num_wait,				\
+    l.n_wait_times, uint64_t)						\
+CTL_RO_CGEN(config_stats, stats_##n##_num_spin_acq,			\
+    l.n_spin_acquired, uint64_t)					\
+CTL_RO_CGEN(config_stats, stats_##n##_num_owner_switch,			\
+    l.n_owner_switches, uint64_t) 					\
+CTL_RO_CGEN(config_stats, stats_##n##_total_wait_time,			\
+    nstime_ns(&l.tot_wait_time), uint64_t)				\
+CTL_RO_CGEN(config_stats, stats_##n##_max_wait_time,			\
+    nstime_ns(&l.max_wait_time), uint64_t)				\
+CTL_RO_CGEN(config_stats, stats_##n##_max_num_thds,			\
+    l.max_n_thds, uint32_t)
+
+/* Global mutexes. */
+#define OP(mtx)								\
+    RO_MUTEX_CTL_GEN(mutexes_##mtx,					\
+        ctl_stats->mutex_prof_data[global_prof_mutex_##mtx])
+MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+
+/* Per arena mutexes */
+#define OP(mtx) RO_MUTEX_CTL_GEN(arenas_i_mutexes_##mtx,		\
+    arenas_i(mib[2])->astats->astats.mutex_prof_data[arena_prof_mutex_##mtx])
+MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+
+/* tcache bin mutex */
+RO_MUTEX_CTL_GEN(arenas_i_bins_j_mutex,
+    arenas_i(mib[2])->astats->bstats[mib[4]].mutex_data)
+#undef RO_MUTEX_CTL_GEN
+
+/* Resets all mutex stats, including global, arena and bin mutexes. */
+static int
+stats_mutexes_reset_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen) {
+	if (!config_stats) {
+		return ENOENT;
+	}
+
+	tsdn_t *tsdn = tsd_tsdn(tsd);
+
+#define MUTEX_PROF_RESET(mtx)						\
+    malloc_mutex_lock(tsdn, &mtx);					\
+    malloc_mutex_prof_data_reset(tsdn, &mtx);				\
+    malloc_mutex_unlock(tsdn, &mtx);
+
+	/* Global mutexes: ctl and prof. */
+	MUTEX_PROF_RESET(ctl_mtx);
+	if (have_background_thread) {
+		MUTEX_PROF_RESET(background_thread_lock);
+	}
+	if (config_prof && opt_prof) {
+		MUTEX_PROF_RESET(bt2gctx_mtx);
+	}
+
+
+	/* Per arena mutexes. */
+	unsigned n = narenas_total_get();
+
+	for (unsigned i = 0; i < n; i++) {
+		arena_t *arena = arena_get(tsdn, i, false);
+		if (!arena) {
+			continue;
+		}
+		MUTEX_PROF_RESET(arena->large_mtx);
+		MUTEX_PROF_RESET(arena->extent_avail_mtx);
+		MUTEX_PROF_RESET(arena->extents_dirty.mtx);
+		MUTEX_PROF_RESET(arena->extents_muzzy.mtx);
+		MUTEX_PROF_RESET(arena->extents_retained.mtx);
+		MUTEX_PROF_RESET(arena->decay_dirty.mtx);
+		MUTEX_PROF_RESET(arena->decay_muzzy.mtx);
+		MUTEX_PROF_RESET(arena->tcache_ql_mtx);
+		MUTEX_PROF_RESET(arena->base->mtx);
+
+		for (szind_t i = 0; i < SC_NBINS; i++) {
+			for (unsigned j = 0; j < bin_infos[i].n_shards; j++) {
+				bin_t *bin = &arena->bins[i].bin_shards[j];
+				MUTEX_PROF_RESET(bin->lock);
+			}
+		}
+	}
+#undef MUTEX_PROF_RESET
+	return 0;
+}
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nmalloc,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nmalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_ndalloc,
+    arenas_i(mib[2])->astats->bstats[mib[4]].ndalloc, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nrequests,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nrequests, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_curregs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].curregs, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nfills,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nfills, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nflushes,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nflushes, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nslabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nslabs, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nreslabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].reslabs, uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_curslabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].curslabs, size_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_bins_j_nonfull_slabs,
+    arenas_i(mib[2])->astats->bstats[mib[4]].nonfull_slabs, size_t)
+
+static const ctl_named_node_t *
+stats_arenas_i_bins_j_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t j) {
+	if (j > SC_NBINS) {
+		return NULL;
+	}
+	return super_stats_arenas_i_bins_j_node;
+}
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_nmalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->lstats[mib[4]].nmalloc), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_ndalloc,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->lstats[mib[4]].ndalloc), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_nrequests,
+    ctl_arena_stats_read_u64(
+    &arenas_i(mib[2])->astats->lstats[mib[4]].nrequests), uint64_t)
+CTL_RO_CGEN(config_stats, stats_arenas_i_lextents_j_curlextents,
+    arenas_i(mib[2])->astats->lstats[mib[4]].curlextents, size_t)
+
+static const ctl_named_node_t *
+stats_arenas_i_lextents_j_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t j) {
+	if (j > SC_NSIZES - SC_NBINS) {
+		return NULL;
+	}
+	return super_stats_arenas_i_lextents_j_node;
+}
+
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_ndirty,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].ndirty,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_nmuzzy,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].nmuzzy,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_nretained,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].nretained,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_dirty_bytes,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].dirty_bytes,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_muzzy_bytes,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].muzzy_bytes,
+	ATOMIC_RELAXED), size_t);
+CTL_RO_CGEN(config_stats, stats_arenas_i_extents_j_retained_bytes,
+    atomic_load_zu(
+        &arenas_i(mib[2])->astats->estats[mib[4]].retained_bytes,
+	ATOMIC_RELAXED), size_t);
+
+static const ctl_named_node_t *
+stats_arenas_i_extents_j_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t j) {
+	if (j >= SC_NPSIZES) {
+		return NULL;
+	}
+	return super_stats_arenas_i_extents_j_node;
+}
+
+static bool
+ctl_arenas_i_verify(size_t i) {
+	size_t a = arenas_i2a_impl(i, true, true);
+	if (a == UINT_MAX || !ctl_arenas->arenas[a]->initialized) {
+		return true;
+	}
+
+	return false;
+}
+
+static const ctl_named_node_t *
+stats_arenas_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	const ctl_named_node_t *ret;
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	if (ctl_arenas_i_verify(i)) {
+		ret = NULL;
+		goto label_return;
+	}
+
+	ret = super_stats_arenas_i_node;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+static int
+experimental_hooks_install_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	if (oldp == NULL || oldlenp == NULL|| newp == NULL) {
+		ret = EINVAL;
+		goto label_return;
+	}
+	/*
+	 * Note: this is a *private* struct.  This is an experimental interface;
+	 * forcing the user to know the jemalloc internals well enough to
+	 * extract the ABI hopefully ensures nobody gets too comfortable with
+	 * this API, which can change at a moment's notice.
+	 */
+	hooks_t hooks;
+	WRITE(hooks, hooks_t);
+	void *handle = hook_install(tsd_tsdn(tsd), &hooks);
+	if (handle == NULL) {
+		ret = EAGAIN;
+		goto label_return;
+	}
+	READ(handle, void *);
+
+	ret = 0;
+label_return:
+	return ret;
+}
+
+static int
+experimental_hooks_remove_ctl(tsd_t *tsd, const size_t *mib, size_t miblen,
+    void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+	WRITEONLY();
+	void *handle = NULL;
+	WRITE(handle, void *);
+	if (handle == NULL) {
+		ret = EINVAL;
+		goto label_return;
+	}
+	hook_remove(tsd_tsdn(tsd), handle);
+	ret = 0;
+label_return:
+	return ret;
+}
+
+/*
+ * Output six memory utilization entries for an input pointer, the first one of
+ * type (void *) and the remaining five of type size_t, describing the following
+ * (in the same order):
+ *
+ * (a) memory address of the extent a potential reallocation would go into,
+ * == the five fields below describe about the extent the pointer resides in ==
+ * (b) number of free regions in the extent,
+ * (c) number of regions in the extent,
+ * (d) size of the extent in terms of bytes,
+ * (e) total number of free regions in the bin the extent belongs to, and
+ * (f) total number of regions in the bin the extent belongs to.
+ *
+ * Note that "(e)" and "(f)" are only available when stats are enabled;
+ * otherwise their values are undefined.
+ *
+ * This API is mainly intended for small class allocations, where extents are
+ * used as slab.  Note that if the bin the extent belongs to is completely
+ * full, "(a)" will be NULL.
+ *
+ * In case of large class allocations, "(a)" will be NULL, and "(e)" and "(f)"
+ * will be zero (if stats are enabled; otherwise undefined).  The other three
+ * fields will be properly set though the values are trivial: "(b)" will be 0,
+ * "(c)" will be 1, and "(d)" will be the usable size.
+ *
+ * The input pointer and size are respectively passed in by newp and newlen,
+ * and the output fields and size are respectively oldp and *oldlenp.
+ *
+ * It can be beneficial to define the following macros to make it easier to
+ * access the output:
+ *
+ * #define SLABCUR_READ(out) (*(void **)out)
+ * #define COUNTS(out) ((size_t *)((void **)out + 1))
+ * #define NFREE_READ(out) COUNTS(out)[0]
+ * #define NREGS_READ(out) COUNTS(out)[1]
+ * #define SIZE_READ(out) COUNTS(out)[2]
+ * #define BIN_NFREE_READ(out) COUNTS(out)[3]
+ * #define BIN_NREGS_READ(out) COUNTS(out)[4]
+ *
+ * and then write e.g. NFREE_READ(oldp) to fetch the output.  See the unit test
+ * test_query in test/unit/extent_util.c for an example.
+ *
+ * For a typical defragmentation workflow making use of this API for
+ * understanding the fragmentation level, please refer to the comment for
+ * experimental_utilization_batch_query_ctl.
+ *
+ * It's up to the application how to determine the significance of
+ * fragmentation relying on the outputs returned.  Possible choices are:
+ *
+ * (a) if extent utilization ratio is below certain threshold,
+ * (b) if extent memory consumption is above certain threshold,
+ * (c) if extent utilization ratio is significantly below bin utilization ratio,
+ * (d) if input pointer deviates a lot from potential reallocation address, or
+ * (e) some selection/combination of the above.
+ *
+ * The caller needs to make sure that the input/output arguments are valid,
+ * in particular, that the size of the output is correct, i.e.:
+ *
+ *     *oldlenp = sizeof(void *) + sizeof(size_t) * 5
+ *
+ * Otherwise, the function immediately returns EINVAL without touching anything.
+ *
+ * In the rare case where there's no associated extent found for the input
+ * pointer, the function zeros out all output fields and return.  Please refer
+ * to the comment for experimental_utilization_batch_query_ctl to understand the
+ * motivation from C++.
+ */
+static int
+experimental_utilization_query_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+
+	assert(sizeof(extent_util_stats_verbose_t)
+	    == sizeof(void *) + sizeof(size_t) * 5);
+
+	if (oldp == NULL || oldlenp == NULL
+	    || *oldlenp != sizeof(extent_util_stats_verbose_t)
+	    || newp == NULL) {
+		ret = EINVAL;
+		goto label_return;
+	}
+
+	void *ptr = NULL;
+	WRITE(ptr, void *);
+	extent_util_stats_verbose_t *util_stats
+	    = (extent_util_stats_verbose_t *)oldp;
+	extent_util_stats_verbose_get(tsd_tsdn(tsd), ptr,
+	    &util_stats->nfree, &util_stats->nregs, &util_stats->size,
+	    &util_stats->bin_nfree, &util_stats->bin_nregs,
+	    &util_stats->slabcur_addr);
+	ret = 0;
+
+label_return:
+	return ret;
+}
+
+/*
+ * Given an input array of pointers, output three memory utilization entries of
+ * type size_t for each input pointer about the extent it resides in:
+ *
+ * (a) number of free regions in the extent,
+ * (b) number of regions in the extent, and
+ * (c) size of the extent in terms of bytes.
+ *
+ * This API is mainly intended for small class allocations, where extents are
+ * used as slab.  In case of large class allocations, the outputs are trivial:
+ * "(a)" will be 0, "(b)" will be 1, and "(c)" will be the usable size.
+ *
+ * Note that multiple input pointers may reside on a same extent so the output
+ * fields may contain duplicates.
+ *
+ * The format of the input/output looks like:
+ *
+ * input[0]:  1st_pointer_to_query	|  output[0]: 1st_extent_n_free_regions
+ *					|  output[1]: 1st_extent_n_regions
+ *					|  output[2]: 1st_extent_size
+ * input[1]:  2nd_pointer_to_query	|  output[3]: 2nd_extent_n_free_regions
+ *					|  output[4]: 2nd_extent_n_regions
+ *					|  output[5]: 2nd_extent_size
+ * ...					|  ...
+ *
+ * The input array and size are respectively passed in by newp and newlen, and
+ * the output array and size are respectively oldp and *oldlenp.
+ *
+ * It can be beneficial to define the following macros to make it easier to
+ * access the output:
+ *
+ * #define NFREE_READ(out, i) out[(i) * 3]
+ * #define NREGS_READ(out, i) out[(i) * 3 + 1]
+ * #define SIZE_READ(out, i) out[(i) * 3 + 2]
+ *
+ * and then write e.g. NFREE_READ(oldp, i) to fetch the output.  See the unit
+ * test test_batch in test/unit/extent_util.c for a concrete example.
+ *
+ * A typical workflow would be composed of the following steps:
+ *
+ * (1) flush tcache: mallctl("thread.tcache.flush", ...)
+ * (2) initialize input array of pointers to query fragmentation
+ * (3) allocate output array to hold utilization statistics
+ * (4) query utilization: mallctl("experimental.utilization.batch_query", ...)
+ * (5) (optional) decide if it's worthwhile to defragment; otherwise stop here
+ * (6) disable tcache: mallctl("thread.tcache.enabled", ...)
+ * (7) defragment allocations with significant fragmentation, e.g.:
+ *         for each allocation {
+ *             if it's fragmented {
+ *                 malloc(...);
+ *                 memcpy(...);
+ *                 free(...);
+ *             }
+ *         }
+ * (8) enable tcache: mallctl("thread.tcache.enabled", ...)
+ *
+ * The application can determine the significance of fragmentation themselves
+ * relying on the statistics returned, both at the overall level i.e. step "(5)"
+ * and at individual allocation level i.e. within step "(7)".  Possible choices
+ * are:
+ *
+ * (a) whether memory utilization ratio is below certain threshold,
+ * (b) whether memory consumption is above certain threshold, or
+ * (c) some combination of the two.
+ *
+ * The caller needs to make sure that the input/output arrays are valid and
+ * their sizes are proper as well as matched, meaning:
+ *
+ * (a) newlen = n_pointers * sizeof(const void *)
+ * (b) *oldlenp = n_pointers * sizeof(size_t) * 3
+ * (c) n_pointers > 0
+ *
+ * Otherwise, the function immediately returns EINVAL without touching anything.
+ *
+ * In the rare case where there's no associated extent found for some pointers,
+ * rather than immediately terminating the computation and raising an error,
+ * the function simply zeros out the corresponding output fields and continues
+ * the computation until all input pointers are handled.  The motivations of
+ * such a design are as follows:
+ *
+ * (a) The function always either processes nothing or processes everything, and
+ * never leaves the output half touched and half untouched.
+ *
+ * (b) It facilitates usage needs especially common in C++.  A vast variety of
+ * C++ objects are instantiated with multiple dynamic memory allocations.  For
+ * example, std::string and std::vector typically use at least two allocations,
+ * one for the metadata and one for the actual content.  Other types may use
+ * even more allocations.  When inquiring about utilization statistics, the
+ * caller often wants to examine into all such allocations, especially internal
+ * one(s), rather than just the topmost one.  The issue comes when some
+ * implementations do certain optimizations to reduce/aggregate some internal
+ * allocations, e.g. putting short strings directly into the metadata, and such
+ * decisions are not known to the caller.  Therefore, we permit pointers to
+ * memory usages that may not be returned by previous malloc calls, and we
+ * provide the caller a convenient way to identify such cases.
+ */
+static int
+experimental_utilization_batch_query_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	int ret;
+
+	assert(sizeof(extent_util_stats_t) == sizeof(size_t) * 3);
+
+	const size_t len = newlen / sizeof(const void *);
+	if (oldp == NULL || oldlenp == NULL || newp == NULL || newlen == 0
+	    || newlen != len * sizeof(const void *)
+	    || *oldlenp != len * sizeof(extent_util_stats_t)) {
+		ret = EINVAL;
+		goto label_return;
+	}
+
+	void **ptrs = (void **)newp;
+	extent_util_stats_t *util_stats = (extent_util_stats_t *)oldp;
+	size_t i;
+	for (i = 0; i < len; ++i) {
+		extent_util_stats_get(tsd_tsdn(tsd), ptrs[i],
+		    &util_stats[i].nfree, &util_stats[i].nregs,
+		    &util_stats[i].size);
+	}
+	ret = 0;
+
+label_return:
+	return ret;
+}
+
+static const ctl_named_node_t *
+experimental_arenas_i_index(tsdn_t *tsdn, const size_t *mib,
+    size_t miblen, size_t i) {
+	const ctl_named_node_t *ret;
+
+	malloc_mutex_lock(tsdn, &ctl_mtx);
+	if (ctl_arenas_i_verify(i)) {
+		ret = NULL;
+		goto label_return;
+	}
+	ret = super_experimental_arenas_i_node;
+label_return:
+	malloc_mutex_unlock(tsdn, &ctl_mtx);
+	return ret;
+}
+
+static int
+experimental_arenas_i_pactivep_ctl(tsd_t *tsd, const size_t *mib,
+    size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen) {
+	if (!config_stats) {
+		return ENOENT;
+	}
+	if (oldp == NULL || oldlenp == NULL || *oldlenp != sizeof(size_t *)) {
+		return EINVAL;
+	}
+
+	unsigned arena_ind;
+	arena_t *arena;
+	int ret;
+	size_t *pactivep;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &ctl_mtx);
+	READONLY();
+	MIB_UNSIGNED(arena_ind, 2);
+	if (arena_ind < narenas_total_get() && (arena =
+	    arena_get(tsd_tsdn(tsd), arena_ind, false)) != NULL) {
+#if defined(JEMALLOC_GCC_ATOMIC_ATOMICS) ||				\
+    defined(JEMALLOC_GCC_SYNC_ATOMICS) || defined(_MSC_VER)
+		/* Expose the underlying counter for fast read. */
+		pactivep = (size_t *)&(arena->nactive.repr);
+		READ(pactivep, size_t *);
+		ret = 0;
+#else
+		ret = EFAULT;
+#endif
+	} else {
+		ret = EFAULT;
+	}
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &ctl_mtx);
+	return ret;
+}
diff --git a/memkind-1.10.0/jemalloc/src/div.c b/memkind-1.10.0/jemalloc/src/div.c
new file mode 100644
index 0000000..808892a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/div.c
@@ -0,0 +1,55 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/div.h"
+
+#include "jemalloc/internal/assert.h"
+
+/*
+ * Suppose we have n = q * d, all integers. We know n and d, and want q = n / d.
+ *
+ * For any k, we have (here, all division is exact; not C-style rounding):
+ * floor(ceil(2^k / d) * n / 2^k) = floor((2^k + r) / d * n / 2^k), where
+ * r = (-2^k) mod d.
+ *
+ * Expanding this out:
+ * ... = floor(2^k / d * n / 2^k + r / d * n / 2^k)
+ *     = floor(n / d + (r / d) * (n / 2^k)).
+ *
+ * The fractional part of n / d is 0 (because of the assumption that d divides n
+ * exactly), so we have:
+ * ... = n / d + floor((r / d) * (n / 2^k))
+ *
+ * So that our initial expression is equal to the quantity we seek, so long as
+ * (r / d) * (n / 2^k) < 1.
+ *
+ * r is a remainder mod d, so r < d and r / d < 1 always. We can make
+ * n / 2 ^ k < 1 by setting k = 32. This gets us a value of magic that works.
+ */
+
+void
+div_init(div_info_t *div_info, size_t d) {
+	/* Nonsensical. */
+	assert(d != 0);
+	/*
+	 * This would make the value of magic too high to fit into a uint32_t
+	 * (we would want magic = 2^32 exactly). This would mess with code gen
+	 * on 32-bit machines.
+	 */
+	assert(d != 1);
+
+	uint64_t two_to_k = ((uint64_t)1 << 32);
+	uint32_t magic = (uint32_t)(two_to_k / d);
+
+	/*
+	 * We want magic = ceil(2^k / d), but C gives us floor. We have to
+	 * increment it unless the result was exact (i.e. unless d is a power of
+	 * two).
+	 */
+	if (two_to_k % d != 0) {
+		magic++;
+	}
+	div_info->magic = magic;
+#ifdef JEMALLOC_DEBUG
+	div_info->d = d;
+#endif
+}
diff --git a/memkind-1.10.0/jemalloc/src/extent.c b/memkind-1.10.0/jemalloc/src/extent.c
new file mode 100644
index 0000000..efdcdc0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/extent.c
@@ -0,0 +1,2408 @@
+#define JEMALLOC_EXTENT_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+
+/******************************************************************************/
+/* Data. */
+
+rtree_t		extents_rtree;
+/* Keyed by the address of the extent_t being protected. */
+mutex_pool_t	extent_mutex_pool;
+
+size_t opt_lg_extent_max_active_fit = LG_EXTENT_MAX_ACTIVE_FIT_DEFAULT;
+
+static const bitmap_info_t extents_bitmap_info =
+    BITMAP_INFO_INITIALIZER(SC_NPSIZES+1);
+
+static void *extent_alloc_default(extent_hooks_t *extent_hooks, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit,
+    unsigned arena_ind);
+static bool extent_dalloc_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static void extent_destroy_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static bool extent_commit_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool extent_commit_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained);
+static bool extent_decommit_default(extent_hooks_t *extent_hooks,
+    void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);
+#ifdef PAGES_CAN_PURGE_LAZY
+static bool extent_purge_lazy_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+#endif
+static bool extent_purge_lazy_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained);
+#ifdef PAGES_CAN_PURGE_FORCED
+static bool extent_purge_forced_default(extent_hooks_t *extent_hooks,
+    void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);
+#endif
+static bool extent_purge_forced_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained);
+static bool extent_split_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t size_a, size_t size_b, bool committed,
+    unsigned arena_ind);
+static extent_t *extent_split_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b,
+    bool growing_retained);
+static bool extent_merge_default(extent_hooks_t *extent_hooks, void *addr_a,
+    size_t size_a, void *addr_b, size_t size_b, bool committed,
+    unsigned arena_ind);
+static bool extent_merge_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b,
+    bool growing_retained);
+
+const extent_hooks_t	extent_hooks_default = {
+	extent_alloc_default,
+	extent_dalloc_default,
+	extent_destroy_default,
+	extent_commit_default,
+	extent_decommit_default
+#ifdef PAGES_CAN_PURGE_LAZY
+	,
+	extent_purge_lazy_default
+#else
+	,
+	NULL
+#endif
+#ifdef PAGES_CAN_PURGE_FORCED
+	,
+	extent_purge_forced_default
+#else
+	,
+	NULL
+#endif
+	,
+	extent_split_default,
+	extent_merge_default
+};
+
+/* Used exclusively for gdump triggering. */
+static atomic_zu_t curpages;
+static atomic_zu_t highpages;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static void extent_deregister(tsdn_t *tsdn, extent_t *extent);
+static extent_t *extent_recycle(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, void *new_addr,
+    size_t usize, size_t pad, size_t alignment, bool slab, szind_t szind,
+    bool *zero, bool *commit, bool growing_retained);
+static extent_t *extent_try_coalesce(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained);
+static void extent_record(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, extent_t *extent,
+    bool growing_retained);
+
+/******************************************************************************/
+
+#define ATTR_NONE /* does nothing */
+
+ph_gen(ATTR_NONE, extent_avail_, extent_tree_t, extent_t, ph_link,
+    extent_esnead_comp)
+
+#undef ATTR_NONE
+
+typedef enum {
+	lock_result_success,
+	lock_result_failure,
+	lock_result_no_extent
+} lock_result_t;
+
+static lock_result_t
+extent_rtree_leaf_elm_try_lock(tsdn_t *tsdn, rtree_leaf_elm_t *elm,
+    extent_t **result, bool inactive_only) {
+	extent_t *extent1 = rtree_leaf_elm_extent_read(tsdn, &extents_rtree,
+	    elm, true);
+
+	/* Slab implies active extents and should be skipped. */
+	if (extent1 == NULL || (inactive_only && rtree_leaf_elm_slab_read(tsdn,
+	    &extents_rtree, elm, true))) {
+		return lock_result_no_extent;
+	}
+
+	/*
+	 * It's possible that the extent changed out from under us, and with it
+	 * the leaf->extent mapping.  We have to recheck while holding the lock.
+	 */
+	extent_lock(tsdn, extent1);
+	extent_t *extent2 = rtree_leaf_elm_extent_read(tsdn,
+	    &extents_rtree, elm, true);
+
+	if (extent1 == extent2) {
+		*result = extent1;
+		return lock_result_success;
+	} else {
+		extent_unlock(tsdn, extent1);
+		return lock_result_failure;
+	}
+}
+
+/*
+ * Returns a pool-locked extent_t * if there's one associated with the given
+ * address, and NULL otherwise.
+ */
+static extent_t *
+extent_lock_from_addr(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx, void *addr,
+    bool inactive_only) {
+	extent_t *ret = NULL;
+	rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, &extents_rtree,
+	    rtree_ctx, (uintptr_t)addr, false, false);
+	if (elm == NULL) {
+		return NULL;
+	}
+	lock_result_t lock_result;
+	do {
+		lock_result = extent_rtree_leaf_elm_try_lock(tsdn, elm, &ret,
+		    inactive_only);
+	} while (lock_result == lock_result_failure);
+	return ret;
+}
+
+extent_t *
+extent_alloc(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_lock(tsdn, &arena->extent_avail_mtx);
+	extent_t *extent = extent_avail_first(&arena->extent_avail);
+	if (extent == NULL) {
+		malloc_mutex_unlock(tsdn, &arena->extent_avail_mtx);
+		return base_alloc_extent(tsdn, arena->base);
+	}
+	extent_avail_remove(&arena->extent_avail, extent);
+	atomic_fetch_sub_zu(&arena->extent_avail_cnt, 1, ATOMIC_RELAXED);
+	malloc_mutex_unlock(tsdn, &arena->extent_avail_mtx);
+	return extent;
+}
+
+void
+extent_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	malloc_mutex_lock(tsdn, &arena->extent_avail_mtx);
+	extent_avail_insert(&arena->extent_avail, extent);
+	atomic_fetch_add_zu(&arena->extent_avail_cnt, 1, ATOMIC_RELAXED);
+	malloc_mutex_unlock(tsdn, &arena->extent_avail_mtx);
+}
+
+extent_hooks_t *
+extent_hooks_get(arena_t *arena) {
+	return base_extent_hooks_get(arena->base);
+}
+
+extent_hooks_t *
+extent_hooks_set(tsd_t *tsd, arena_t *arena, extent_hooks_t *extent_hooks) {
+	background_thread_info_t *info;
+	if (have_background_thread) {
+		info = arena_background_thread_info_get(arena);
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+	}
+	extent_hooks_t *ret = base_extent_hooks_set(arena->base, extent_hooks);
+	if (have_background_thread) {
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+	}
+
+	return ret;
+}
+
+static void
+extent_hooks_assure_initialized(arena_t *arena,
+    extent_hooks_t **r_extent_hooks) {
+	if (*r_extent_hooks == EXTENT_HOOKS_INITIALIZER) {
+		*r_extent_hooks = extent_hooks_get(arena);
+	}
+}
+
+#ifndef JEMALLOC_JET
+static
+#endif
+size_t
+extent_size_quantize_floor(size_t size) {
+	size_t ret;
+	pszind_t pind;
+
+	assert(size > 0);
+	assert((size & PAGE_MASK) == 0);
+
+	pind = sz_psz2ind(size - sz_large_pad + 1);
+	if (pind == 0) {
+		/*
+		 * Avoid underflow.  This short-circuit would also do the right
+		 * thing for all sizes in the range for which there are
+		 * PAGE-spaced size classes, but it's simplest to just handle
+		 * the one case that would cause erroneous results.
+		 */
+		return size;
+	}
+	ret = sz_pind2sz(pind - 1) + sz_large_pad;
+	assert(ret <= size);
+	return ret;
+}
+
+#ifndef JEMALLOC_JET
+static
+#endif
+size_t
+extent_size_quantize_ceil(size_t size) {
+	size_t ret;
+
+	assert(size > 0);
+	assert(size - sz_large_pad <= SC_LARGE_MAXCLASS);
+	assert((size & PAGE_MASK) == 0);
+
+	ret = extent_size_quantize_floor(size);
+	if (ret < size) {
+		/*
+		 * Skip a quantization that may have an adequately large extent,
+		 * because under-sized extents may be mixed in.  This only
+		 * happens when an unusual size is requested, i.e. for aligned
+		 * allocation, and is just one of several places where linear
+		 * search would potentially find sufficiently aligned available
+		 * memory somewhere lower.
+		 */
+		ret = sz_pind2sz(sz_psz2ind(ret - sz_large_pad + 1)) +
+		    sz_large_pad;
+	}
+	return ret;
+}
+
+/* Generate pairing heap functions. */
+ph_gen(, extent_heap_, extent_heap_t, extent_t, ph_link, extent_snad_comp)
+
+bool
+extents_init(tsdn_t *tsdn, extents_t *extents, extent_state_t state,
+    bool delay_coalesce) {
+	if (malloc_mutex_init(&extents->mtx, "extents", WITNESS_RANK_EXTENTS,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	for (unsigned i = 0; i < SC_NPSIZES + 1; i++) {
+		extent_heap_new(&extents->heaps[i]);
+	}
+	bitmap_init(extents->bitmap, &extents_bitmap_info, true);
+	extent_list_init(&extents->lru);
+	atomic_store_zu(&extents->npages, 0, ATOMIC_RELAXED);
+	extents->state = state;
+	extents->delay_coalesce = delay_coalesce;
+	return false;
+}
+
+extent_state_t
+extents_state_get(const extents_t *extents) {
+	return extents->state;
+}
+
+size_t
+extents_npages_get(extents_t *extents) {
+	return atomic_load_zu(&extents->npages, ATOMIC_RELAXED);
+}
+
+size_t
+extents_nextents_get(extents_t *extents, pszind_t pind) {
+	return atomic_load_zu(&extents->nextents[pind], ATOMIC_RELAXED);
+}
+
+size_t
+extents_nbytes_get(extents_t *extents, pszind_t pind) {
+	return atomic_load_zu(&extents->nbytes[pind], ATOMIC_RELAXED);
+}
+
+static void
+extents_stats_add(extents_t *extent, pszind_t pind, size_t sz) {
+	size_t cur = atomic_load_zu(&extent->nextents[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nextents[pind], cur + 1, ATOMIC_RELAXED);
+	cur = atomic_load_zu(&extent->nbytes[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nbytes[pind], cur + sz, ATOMIC_RELAXED);
+}
+
+static void
+extents_stats_sub(extents_t *extent, pszind_t pind, size_t sz) {
+	size_t cur = atomic_load_zu(&extent->nextents[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nextents[pind], cur - 1, ATOMIC_RELAXED);
+	cur = atomic_load_zu(&extent->nbytes[pind], ATOMIC_RELAXED);
+	atomic_store_zu(&extent->nbytes[pind], cur - sz, ATOMIC_RELAXED);
+}
+
+static void
+extents_insert_locked(tsdn_t *tsdn, extents_t *extents, extent_t *extent) {
+	malloc_mutex_assert_owner(tsdn, &extents->mtx);
+	assert(extent_state_get(extent) == extents->state);
+
+	size_t size = extent_size_get(extent);
+	size_t psz = extent_size_quantize_floor(size);
+	pszind_t pind = sz_psz2ind(psz);
+	if (extent_heap_empty(&extents->heaps[pind])) {
+		bitmap_unset(extents->bitmap, &extents_bitmap_info,
+		    (size_t)pind);
+	}
+	extent_heap_insert(&extents->heaps[pind], extent);
+
+	if (config_stats) {
+		extents_stats_add(extents, pind, size);
+	}
+
+	extent_list_append(&extents->lru, extent);
+	size_t npages = size >> LG_PAGE;
+	/*
+	 * All modifications to npages hold the mutex (as asserted above), so we
+	 * don't need an atomic fetch-add; we can get by with a load followed by
+	 * a store.
+	 */
+	size_t cur_extents_npages =
+	    atomic_load_zu(&extents->npages, ATOMIC_RELAXED);
+	atomic_store_zu(&extents->npages, cur_extents_npages + npages,
+	    ATOMIC_RELAXED);
+}
+
+static void
+extents_remove_locked(tsdn_t *tsdn, extents_t *extents, extent_t *extent) {
+	malloc_mutex_assert_owner(tsdn, &extents->mtx);
+	assert(extent_state_get(extent) == extents->state);
+
+	size_t size = extent_size_get(extent);
+	size_t psz = extent_size_quantize_floor(size);
+	pszind_t pind = sz_psz2ind(psz);
+	extent_heap_remove(&extents->heaps[pind], extent);
+
+	if (config_stats) {
+		extents_stats_sub(extents, pind, size);
+	}
+
+	if (extent_heap_empty(&extents->heaps[pind])) {
+		bitmap_set(extents->bitmap, &extents_bitmap_info,
+		    (size_t)pind);
+	}
+	extent_list_remove(&extents->lru, extent);
+	size_t npages = size >> LG_PAGE;
+	/*
+	 * As in extents_insert_locked, we hold extents->mtx and so don't need
+	 * atomic operations for updating extents->npages.
+	 */
+	size_t cur_extents_npages =
+	    atomic_load_zu(&extents->npages, ATOMIC_RELAXED);
+	assert(cur_extents_npages >= npages);
+	atomic_store_zu(&extents->npages,
+	    cur_extents_npages - (size >> LG_PAGE), ATOMIC_RELAXED);
+}
+
+/*
+ * Find an extent with size [min_size, max_size) to satisfy the alignment
+ * requirement.  For each size, try only the first extent in the heap.
+ */
+static extent_t *
+extents_fit_alignment(extents_t *extents, size_t min_size, size_t max_size,
+    size_t alignment) {
+        pszind_t pind = sz_psz2ind(extent_size_quantize_ceil(min_size));
+        pszind_t pind_max = sz_psz2ind(extent_size_quantize_ceil(max_size));
+
+	for (pszind_t i = (pszind_t)bitmap_ffu(extents->bitmap,
+	    &extents_bitmap_info, (size_t)pind); i < pind_max; i =
+	    (pszind_t)bitmap_ffu(extents->bitmap, &extents_bitmap_info,
+	    (size_t)i+1)) {
+		assert(i < SC_NPSIZES);
+		assert(!extent_heap_empty(&extents->heaps[i]));
+		extent_t *extent = extent_heap_first(&extents->heaps[i]);
+		uintptr_t base = (uintptr_t)extent_base_get(extent);
+		size_t candidate_size = extent_size_get(extent);
+		assert(candidate_size >= min_size);
+
+		uintptr_t next_align = ALIGNMENT_CEILING((uintptr_t)base,
+		    PAGE_CEILING(alignment));
+		if (base > next_align || base + candidate_size <= next_align) {
+			/* Overflow or not crossing the next alignment. */
+			continue;
+		}
+
+		size_t leadsize = next_align - base;
+		if (candidate_size - leadsize >= min_size) {
+			return extent;
+		}
+	}
+
+	return NULL;
+}
+
+/*
+ * Do first-fit extent selection, i.e. select the oldest/lowest extent that is
+ * large enough.
+ */
+static extent_t *
+extents_first_fit_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    size_t size) {
+	extent_t *ret = NULL;
+
+	pszind_t pind = sz_psz2ind(extent_size_quantize_ceil(size));
+
+	if (!maps_coalesce && !opt_retain) {
+		/*
+		 * No split / merge allowed (Windows w/o retain). Try exact fit
+		 * only.
+		 */
+		return extent_heap_empty(&extents->heaps[pind]) ? NULL :
+		    extent_heap_first(&extents->heaps[pind]);
+	}
+
+	for (pszind_t i = (pszind_t)bitmap_ffu(extents->bitmap,
+	    &extents_bitmap_info, (size_t)pind);
+	    i < SC_NPSIZES + 1;
+	    i = (pszind_t)bitmap_ffu(extents->bitmap, &extents_bitmap_info,
+	    (size_t)i+1)) {
+		assert(!extent_heap_empty(&extents->heaps[i]));
+		extent_t *extent = extent_heap_first(&extents->heaps[i]);
+		assert(extent_size_get(extent) >= size);
+		/*
+		 * In order to reduce fragmentation, avoid reusing and splitting
+		 * large extents for much smaller sizes.
+		 *
+		 * Only do check for dirty extents (delay_coalesce).
+		 */
+		if (extents->delay_coalesce &&
+		    (sz_pind2sz(i) >> opt_lg_extent_max_active_fit) > size) {
+			break;
+		}
+		if (ret == NULL || extent_snad_comp(extent, ret) < 0) {
+			ret = extent;
+		}
+		if (i == SC_NPSIZES) {
+			break;
+		}
+		assert(i < SC_NPSIZES);
+	}
+
+	return ret;
+}
+
+/*
+ * Do first-fit extent selection, where the selection policy choice is
+ * based on extents->delay_coalesce.
+ */
+static extent_t *
+extents_fit_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    size_t esize, size_t alignment) {
+	malloc_mutex_assert_owner(tsdn, &extents->mtx);
+
+	size_t max_size = esize + PAGE_CEILING(alignment) - PAGE;
+	/* Beware size_t wrap-around. */
+	if (max_size < esize) {
+		return NULL;
+	}
+
+	extent_t *extent =
+	    extents_first_fit_locked(tsdn, arena, extents, max_size);
+
+	if (alignment > PAGE && extent == NULL) {
+		/*
+		 * max_size guarantees the alignment requirement but is rather
+		 * pessimistic.  Next we try to satisfy the aligned allocation
+		 * with sizes in [esize, max_size).
+		 */
+		extent = extents_fit_alignment(extents, esize, max_size,
+		    alignment);
+	}
+
+	return extent;
+}
+
+static bool
+extent_try_delayed_coalesce(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent) {
+	extent_state_set(extent, extent_state_active);
+	bool coalesced;
+	extent = extent_try_coalesce(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, extent, &coalesced, false);
+	extent_state_set(extent, extents_state_get(extents));
+
+	if (!coalesced) {
+		return true;
+	}
+	extents_insert_locked(tsdn, extents, extent);
+	return false;
+}
+
+extent_t *
+extents_alloc(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	assert(size + pad != 0);
+	assert(alignment != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_t *extent = extent_recycle(tsdn, arena, r_extent_hooks, extents,
+	    new_addr, size, pad, alignment, slab, szind, zero, commit, false);
+	assert(extent == NULL || extent_dumpable_get(extent));
+	return extent;
+}
+
+void
+extents_dalloc(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *extent) {
+	assert(extent_base_get(extent) != NULL);
+	assert(extent_size_get(extent) != 0);
+	assert(extent_dumpable_get(extent));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_addr_set(extent, extent_base_get(extent));
+	extent_zeroed_set(extent, false);
+
+	extent_record(tsdn, arena, r_extent_hooks, extents, extent, false);
+}
+
+extent_t *
+extents_evict(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, size_t npages_min) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	malloc_mutex_lock(tsdn, &extents->mtx);
+
+	/*
+	 * Get the LRU coalesced extent, if any.  If coalescing was delayed,
+	 * the loop will iterate until the LRU extent is fully coalesced.
+	 */
+	extent_t *extent;
+	while (true) {
+		/* Get the LRU extent, if any. */
+		extent = extent_list_first(&extents->lru);
+		if (extent == NULL) {
+			goto label_return;
+		}
+		/* Check the eviction limit. */
+		size_t extents_npages = atomic_load_zu(&extents->npages,
+		    ATOMIC_RELAXED);
+		if (extents_npages <= npages_min) {
+			extent = NULL;
+			goto label_return;
+		}
+		extents_remove_locked(tsdn, extents, extent);
+		if (!extents->delay_coalesce) {
+			break;
+		}
+		/* Try to coalesce. */
+		if (extent_try_delayed_coalesce(tsdn, arena, r_extent_hooks,
+		    rtree_ctx, extents, extent)) {
+			break;
+		}
+		/*
+		 * The LRU extent was just coalesced and the result placed in
+		 * the LRU at its neighbor's position.  Start over.
+		 */
+	}
+
+	/*
+	 * Either mark the extent active or deregister it to protect against
+	 * concurrent operations.
+	 */
+	switch (extents_state_get(extents)) {
+	case extent_state_active:
+		not_reached();
+	case extent_state_dirty:
+	case extent_state_muzzy:
+		extent_state_set(extent, extent_state_active);
+		break;
+	case extent_state_retained:
+		extent_deregister(tsdn, extent);
+		break;
+	default:
+		not_reached();
+	}
+
+label_return:
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+	return extent;
+}
+
+/*
+ * This can only happen when we fail to allocate a new extent struct (which
+ * indicates OOM), e.g. when trying to split an existing extent.
+ */
+static void
+extents_abandon_vm(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *extent, bool growing_retained) {
+	size_t sz = extent_size_get(extent);
+	if (config_stats) {
+		arena_stats_accum_zu(&arena->stats.abandoned_vm, sz);
+	}
+	/*
+	 * Leak extent after making sure its pages have already been purged, so
+	 * that this is only a virtual memory leak.
+	 */
+	if (extents_state_get(extents) == extent_state_dirty) {
+		if (extent_purge_lazy_impl(tsdn, arena, r_extent_hooks,
+		    extent, 0, sz, growing_retained)) {
+			extent_purge_forced_impl(tsdn, arena, r_extent_hooks,
+			    extent, 0, extent_size_get(extent),
+			    growing_retained);
+		}
+	}
+	extent_dalloc(tsdn, arena, extent);
+}
+
+void
+extents_prefork(tsdn_t *tsdn, extents_t *extents) {
+	malloc_mutex_prefork(tsdn, &extents->mtx);
+}
+
+void
+extents_postfork_parent(tsdn_t *tsdn, extents_t *extents) {
+	malloc_mutex_postfork_parent(tsdn, &extents->mtx);
+}
+
+void
+extents_postfork_child(tsdn_t *tsdn, extents_t *extents) {
+	malloc_mutex_postfork_child(tsdn, &extents->mtx);
+}
+
+static void
+extent_deactivate_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    extent_t *extent) {
+	assert(extent_arena_get(extent) == arena);
+	assert(extent_state_get(extent) == extent_state_active);
+
+	extent_state_set(extent, extents_state_get(extents));
+	extents_insert_locked(tsdn, extents, extent);
+}
+
+static void
+extent_deactivate(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    extent_t *extent) {
+	malloc_mutex_lock(tsdn, &extents->mtx);
+	extent_deactivate_locked(tsdn, arena, extents, extent);
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+}
+
+static void
+extent_activate_locked(tsdn_t *tsdn, arena_t *arena, extents_t *extents,
+    extent_t *extent) {
+	assert(extent_arena_get(extent) == arena);
+	assert(extent_state_get(extent) == extents_state_get(extents));
+
+	extents_remove_locked(tsdn, extents, extent);
+	extent_state_set(extent, extent_state_active);
+}
+
+static bool
+extent_rtree_leaf_elms_lookup(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx,
+    const extent_t *extent, bool dependent, bool init_missing,
+    rtree_leaf_elm_t **r_elm_a, rtree_leaf_elm_t **r_elm_b) {
+	*r_elm_a = rtree_leaf_elm_lookup(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_base_get(extent), dependent, init_missing);
+	if (!dependent && *r_elm_a == NULL) {
+		return true;
+	}
+	assert(*r_elm_a != NULL);
+
+	*r_elm_b = rtree_leaf_elm_lookup(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_last_get(extent), dependent, init_missing);
+	if (!dependent && *r_elm_b == NULL) {
+		return true;
+	}
+	assert(*r_elm_b != NULL);
+
+	return false;
+}
+
+static void
+extent_rtree_write_acquired(tsdn_t *tsdn, rtree_leaf_elm_t *elm_a,
+    rtree_leaf_elm_t *elm_b, extent_t *extent, szind_t szind, bool slab) {
+	rtree_leaf_elm_write(tsdn, &extents_rtree, elm_a, extent, szind, slab);
+	if (elm_b != NULL) {
+		rtree_leaf_elm_write(tsdn, &extents_rtree, elm_b, extent, szind,
+		    slab);
+	}
+}
+
+static void
+extent_interior_register(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx, extent_t *extent,
+    szind_t szind) {
+	assert(extent_slab_get(extent));
+
+	/* Register interior. */
+	for (size_t i = 1; i < (extent_size_get(extent) >> LG_PAGE) - 1; i++) {
+		rtree_write(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)extent_base_get(extent) + (uintptr_t)(i <<
+		    LG_PAGE), extent, szind, true);
+	}
+}
+
+static void
+extent_gdump_add(tsdn_t *tsdn, const extent_t *extent) {
+	cassert(config_prof);
+	/* prof_gdump() requirement. */
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (opt_prof && extent_state_get(extent) == extent_state_active) {
+		size_t nadd = extent_size_get(extent) >> LG_PAGE;
+		size_t cur = atomic_fetch_add_zu(&curpages, nadd,
+		    ATOMIC_RELAXED) + nadd;
+		size_t high = atomic_load_zu(&highpages, ATOMIC_RELAXED);
+		while (cur > high && !atomic_compare_exchange_weak_zu(
+		    &highpages, &high, cur, ATOMIC_RELAXED, ATOMIC_RELAXED)) {
+			/*
+			 * Don't refresh cur, because it may have decreased
+			 * since this thread lost the highpages update race.
+			 * Note that high is updated in case of CAS failure.
+			 */
+		}
+		if (cur > high && prof_gdump_get_unlocked()) {
+			prof_gdump(tsdn);
+		}
+	}
+}
+
+static void
+extent_gdump_sub(tsdn_t *tsdn, const extent_t *extent) {
+	cassert(config_prof);
+
+	if (opt_prof && extent_state_get(extent) == extent_state_active) {
+		size_t nsub = extent_size_get(extent) >> LG_PAGE;
+		assert(atomic_load_zu(&curpages, ATOMIC_RELAXED) >= nsub);
+		atomic_fetch_sub_zu(&curpages, nsub, ATOMIC_RELAXED);
+	}
+}
+
+static bool
+extent_register_impl(tsdn_t *tsdn, extent_t *extent, bool gdump_add) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *elm_a, *elm_b;
+
+	/*
+	 * We need to hold the lock to protect against a concurrent coalesce
+	 * operation that sees us in a partial state.
+	 */
+	extent_lock(tsdn, extent);
+
+	if (extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, extent, false, true,
+	    &elm_a, &elm_b)) {
+		extent_unlock(tsdn, extent);
+		return true;
+	}
+
+	szind_t szind = extent_szind_get_maybe_invalid(extent);
+	bool slab = extent_slab_get(extent);
+	extent_rtree_write_acquired(tsdn, elm_a, elm_b, extent, szind, slab);
+	if (slab) {
+		extent_interior_register(tsdn, rtree_ctx, extent, szind);
+	}
+
+	extent_unlock(tsdn, extent);
+
+	if (config_prof && gdump_add) {
+		extent_gdump_add(tsdn, extent);
+	}
+
+	return false;
+}
+
+static bool
+extent_register(tsdn_t *tsdn, extent_t *extent) {
+	return extent_register_impl(tsdn, extent, true);
+}
+
+static bool
+extent_register_no_gdump_add(tsdn_t *tsdn, extent_t *extent) {
+	return extent_register_impl(tsdn, extent, false);
+}
+
+static void
+extent_reregister(tsdn_t *tsdn, extent_t *extent) {
+	bool err = extent_register(tsdn, extent);
+	assert(!err);
+}
+
+/*
+ * Removes all pointers to the given extent from the global rtree indices for
+ * its interior.  This is relevant for slab extents, for which we need to do
+ * metadata lookups at places other than the head of the extent.  We deregister
+ * on the interior, then, when an extent moves from being an active slab to an
+ * inactive state.
+ */
+static void
+extent_interior_deregister(tsdn_t *tsdn, rtree_ctx_t *rtree_ctx,
+    extent_t *extent) {
+	size_t i;
+
+	assert(extent_slab_get(extent));
+
+	for (i = 1; i < (extent_size_get(extent) >> LG_PAGE) - 1; i++) {
+		rtree_clear(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)extent_base_get(extent) + (uintptr_t)(i <<
+		    LG_PAGE));
+	}
+}
+
+/*
+ * Removes all pointers to the given extent from the global rtree.
+ */
+static void
+extent_deregister_impl(tsdn_t *tsdn, extent_t *extent, bool gdump) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *elm_a, *elm_b;
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, extent, true, false,
+	    &elm_a, &elm_b);
+
+	extent_lock(tsdn, extent);
+
+	extent_rtree_write_acquired(tsdn, elm_a, elm_b, NULL, SC_NSIZES, false);
+	if (extent_slab_get(extent)) {
+		extent_interior_deregister(tsdn, rtree_ctx, extent);
+		extent_slab_set(extent, false);
+	}
+
+	extent_unlock(tsdn, extent);
+
+	if (config_prof && gdump) {
+		extent_gdump_sub(tsdn, extent);
+	}
+}
+
+static void
+extent_deregister(tsdn_t *tsdn, extent_t *extent) {
+	extent_deregister_impl(tsdn, extent, true);
+}
+
+static void
+extent_deregister_no_gdump_sub(tsdn_t *tsdn, extent_t *extent) {
+	extent_deregister_impl(tsdn, extent, false);
+}
+
+/*
+ * Tries to find and remove an extent from extents that can be used for the
+ * given allocation request.
+ */
+static extent_t *
+extent_recycle_extract(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    void *new_addr, size_t size, size_t pad, size_t alignment, bool slab,
+    bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+	assert(alignment > 0);
+	if (config_debug && new_addr != NULL) {
+		/*
+		 * Non-NULL new_addr has two use cases:
+		 *
+		 *   1) Recycle a known-extant extent, e.g. during purging.
+		 *   2) Perform in-place expanding reallocation.
+		 *
+		 * Regardless of use case, new_addr must either refer to a
+		 * non-existing extent, or to the base of an extant extent,
+		 * since only active slabs support interior lookups (which of
+		 * course cannot be recycled).
+		 */
+		assert(PAGE_ADDR2BASE(new_addr) == new_addr);
+		assert(pad == 0);
+		assert(alignment <= PAGE);
+	}
+
+	size_t esize = size + pad;
+	malloc_mutex_lock(tsdn, &extents->mtx);
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	extent_t *extent;
+	if (new_addr != NULL) {
+		extent = extent_lock_from_addr(tsdn, rtree_ctx, new_addr,
+		    false);
+		if (extent != NULL) {
+			/*
+			 * We might null-out extent to report an error, but we
+			 * still need to unlock the associated mutex after.
+			 */
+			extent_t *unlock_extent = extent;
+			assert(extent_base_get(extent) == new_addr);
+			if (extent_arena_get(extent) != arena ||
+			    extent_size_get(extent) < esize ||
+			    extent_state_get(extent) !=
+			    extents_state_get(extents)) {
+				extent = NULL;
+			}
+			extent_unlock(tsdn, unlock_extent);
+		}
+	} else {
+		extent = extents_fit_locked(tsdn, arena, extents, esize,
+		    alignment);
+	}
+	if (extent == NULL) {
+		malloc_mutex_unlock(tsdn, &extents->mtx);
+		return NULL;
+	}
+
+	extent_activate_locked(tsdn, arena, extents, extent);
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+
+	return extent;
+}
+
+/*
+ * Given an allocation request and an extent guaranteed to be able to satisfy
+ * it, this splits off lead and trail extents, leaving extent pointing to an
+ * extent satisfying the allocation.
+ * This function doesn't put lead or trail into any extents_t; it's the caller's
+ * job to ensure that they can be reused.
+ */
+typedef enum {
+	/*
+	 * Split successfully.  lead, extent, and trail, are modified to extents
+	 * describing the ranges before, in, and after the given allocation.
+	 */
+	extent_split_interior_ok,
+	/*
+	 * The extent can't satisfy the given allocation request.  None of the
+	 * input extent_t *s are touched.
+	 */
+	extent_split_interior_cant_alloc,
+	/*
+	 * In a potentially invalid state.  Must leak (if *to_leak is non-NULL),
+	 * and salvage what's still salvageable (if *to_salvage is non-NULL).
+	 * None of lead, extent, or trail are valid.
+	 */
+	extent_split_interior_error
+} extent_split_interior_result_t;
+
+static extent_split_interior_result_t
+extent_split_interior(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx,
+    /* The result of splitting, in case of success. */
+    extent_t **extent, extent_t **lead, extent_t **trail,
+    /* The mess to clean up, in case of error. */
+    extent_t **to_leak, extent_t **to_salvage,
+    void *new_addr, size_t size, size_t pad, size_t alignment, bool slab,
+    szind_t szind, bool growing_retained) {
+	size_t esize = size + pad;
+	size_t leadsize = ALIGNMENT_CEILING((uintptr_t)extent_base_get(*extent),
+	    PAGE_CEILING(alignment)) - (uintptr_t)extent_base_get(*extent);
+	assert(new_addr == NULL || leadsize == 0);
+	if (extent_size_get(*extent) < leadsize + esize) {
+		return extent_split_interior_cant_alloc;
+	}
+	size_t trailsize = extent_size_get(*extent) - leadsize - esize;
+
+	*lead = NULL;
+	*trail = NULL;
+	*to_leak = NULL;
+	*to_salvage = NULL;
+
+	/* Split the lead. */
+	if (leadsize != 0) {
+		*lead = *extent;
+		*extent = extent_split_impl(tsdn, arena, r_extent_hooks,
+		    *lead, leadsize, SC_NSIZES, false, esize + trailsize, szind,
+		    slab, growing_retained);
+		if (*extent == NULL) {
+			*to_leak = *lead;
+			*lead = NULL;
+			return extent_split_interior_error;
+		}
+	}
+
+	/* Split the trail. */
+	if (trailsize != 0) {
+		*trail = extent_split_impl(tsdn, arena, r_extent_hooks, *extent,
+		    esize, szind, slab, trailsize, SC_NSIZES, false,
+		    growing_retained);
+		if (*trail == NULL) {
+			*to_leak = *extent;
+			*to_salvage = *lead;
+			*lead = NULL;
+			*extent = NULL;
+			return extent_split_interior_error;
+		}
+	}
+
+	if (leadsize == 0 && trailsize == 0) {
+		/*
+		 * Splitting causes szind to be set as a side effect, but no
+		 * splitting occurred.
+		 */
+		extent_szind_set(*extent, szind);
+		if (szind != SC_NSIZES) {
+			rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx,
+			    (uintptr_t)extent_addr_get(*extent), szind, slab);
+			if (slab && extent_size_get(*extent) > PAGE) {
+				rtree_szind_slab_update(tsdn, &extents_rtree,
+				    rtree_ctx,
+				    (uintptr_t)extent_past_get(*extent) -
+				    (uintptr_t)PAGE, szind, slab);
+			}
+		}
+	}
+
+	return extent_split_interior_ok;
+}
+
+/*
+ * This fulfills the indicated allocation request out of the given extent (which
+ * the caller should have ensured was big enough).  If there's any unused space
+ * before or after the resulting allocation, that space is given its own extent
+ * and put back into extents.
+ */
+static extent_t *
+extent_recycle_split(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    void *new_addr, size_t size, size_t pad, size_t alignment, bool slab,
+    szind_t szind, extent_t *extent, bool growing_retained) {
+	extent_t *lead;
+	extent_t *trail;
+	extent_t *to_leak;
+	extent_t *to_salvage;
+
+	extent_split_interior_result_t result = extent_split_interior(
+	    tsdn, arena, r_extent_hooks, rtree_ctx, &extent, &lead, &trail,
+	    &to_leak, &to_salvage, new_addr, size, pad, alignment, slab, szind,
+	    growing_retained);
+
+	if (!maps_coalesce && result != extent_split_interior_ok
+	    && !opt_retain) {
+		/*
+		 * Split isn't supported (implies Windows w/o retain).  Avoid
+		 * leaking the extents.
+		 */
+		assert(to_leak != NULL && lead == NULL && trail == NULL);
+		extent_deactivate(tsdn, arena, extents, to_leak);
+		return NULL;
+	}
+
+	if (result == extent_split_interior_ok) {
+		if (lead != NULL) {
+			extent_deactivate(tsdn, arena, extents, lead);
+		}
+		if (trail != NULL) {
+			extent_deactivate(tsdn, arena, extents, trail);
+		}
+		return extent;
+	} else {
+		/*
+		 * We should have picked an extent that was large enough to
+		 * fulfill our allocation request.
+		 */
+		assert(result == extent_split_interior_error);
+		if (to_salvage != NULL) {
+			extent_deregister(tsdn, to_salvage);
+		}
+		if (to_leak != NULL) {
+			void *leak = extent_base_get(to_leak);
+			extent_deregister_no_gdump_sub(tsdn, to_leak);
+			extents_abandon_vm(tsdn, arena, r_extent_hooks, extents,
+			    to_leak, growing_retained);
+			assert(extent_lock_from_addr(tsdn, rtree_ctx, leak,
+			    false) == NULL);
+		}
+		return NULL;
+	}
+	unreachable();
+}
+
+static bool
+extent_need_manual_zero(arena_t *arena) {
+	/*
+	 * Need to manually zero the extent on repopulating if either; 1) non
+	 * default extent hooks installed (in which case the purge semantics may
+	 * change); or 2) transparent huge pages enabled.
+	 */
+	return (!arena_has_default_hooks(arena) ||
+		(opt_thp == thp_mode_always));
+}
+
+/*
+ * Tries to satisfy the given allocation request by reusing one of the extents
+ * in the given extents_t.
+ */
+static extent_t *
+extent_recycle(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit,
+    bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+	assert(new_addr == NULL || !slab);
+	assert(pad == 0 || !slab);
+	assert(!*zero || !slab);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent = extent_recycle_extract(tsdn, arena, r_extent_hooks,
+	    rtree_ctx, extents, new_addr, size, pad, alignment, slab,
+	    growing_retained);
+	if (extent == NULL) {
+		return NULL;
+	}
+
+	extent = extent_recycle_split(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, new_addr, size, pad, alignment, slab, szind, extent,
+	    growing_retained);
+	if (extent == NULL) {
+		return NULL;
+	}
+
+	if (*commit && !extent_committed_get(extent)) {
+		if (extent_commit_impl(tsdn, arena, r_extent_hooks, extent,
+		    0, extent_size_get(extent), growing_retained)) {
+			extent_record(tsdn, arena, r_extent_hooks, extents,
+			    extent, growing_retained);
+			return NULL;
+		}
+		if (!extent_need_manual_zero(arena)) {
+			extent_zeroed_set(extent, true);
+		}
+	}
+
+	if (extent_committed_get(extent)) {
+		*commit = true;
+	}
+	if (extent_zeroed_get(extent)) {
+		*zero = true;
+	}
+
+	if (pad != 0) {
+		extent_addr_randomize(tsdn, extent, alignment);
+	}
+	assert(extent_state_get(extent) == extent_state_active);
+	if (slab) {
+		extent_slab_set(extent, slab);
+		extent_interior_register(tsdn, rtree_ctx, extent, szind);
+	}
+
+	if (*zero) {
+		void *addr = extent_base_get(extent);
+		if (!extent_zeroed_get(extent)) {
+			size_t size = extent_size_get(extent);
+			if (extent_need_manual_zero(arena) ||
+			    pages_purge_forced(addr, size)) {
+				memset(addr, 0, size);
+			}
+		} else if (config_debug) {
+			size_t *p = (size_t *)(uintptr_t)addr;
+			/* Check the first page only. */
+			for (size_t i = 0; i < PAGE / sizeof(size_t); i++) {
+				assert(p[i] == 0);
+			}
+		}
+	}
+	return extent;
+}
+
+/*
+ * If the caller specifies (!*zero), it is still possible to receive zeroed
+ * memory, in which case *zero is toggled to true.  arena_extent_alloc() takes
+ * advantage of this to avoid demanding zeroed extents, but taking advantage of
+ * them if they are returned.
+ */
+static void *
+extent_alloc_core(tsdn_t *tsdn, arena_t *arena, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit, dss_prec_t dss_prec) {
+	void *ret;
+
+	assert(size != 0);
+	assert(alignment != 0);
+
+	/* "primary" dss. */
+	if (have_dss && dss_prec == dss_prec_primary && (ret =
+	    extent_alloc_dss(tsdn, arena, new_addr, size, alignment, zero,
+	    commit)) != NULL) {
+		return ret;
+	}
+	/* mmap. */
+	if ((ret = extent_alloc_mmap(new_addr, size, alignment, zero, commit))
+	    != NULL) {
+		return ret;
+	}
+	/* "secondary" dss. */
+	if (have_dss && dss_prec == dss_prec_secondary && (ret =
+	    extent_alloc_dss(tsdn, arena, new_addr, size, alignment, zero,
+	    commit)) != NULL) {
+		return ret;
+	}
+
+	/* All strategies for allocation failed. */
+	return NULL;
+}
+
+static void *
+extent_alloc_default_impl(tsdn_t *tsdn, arena_t *arena, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit) {
+	void *ret = extent_alloc_core(tsdn, arena, new_addr, size, alignment, zero,
+	    commit, (dss_prec_t)atomic_load_u(&arena->dss_prec,
+	    ATOMIC_RELAXED));
+	if (have_madvise_huge && ret) {
+		pages_set_thp_state(ret, size);
+	}
+	return ret;
+}
+
+static void *
+extent_alloc_default(extent_hooks_t *extent_hooks, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit, unsigned arena_ind) {
+	tsdn_t *tsdn;
+	arena_t *arena;
+
+	tsdn = tsdn_fetch();
+	arena = arena_get(tsdn, arena_ind, false);
+	/*
+	 * The arena we're allocating on behalf of must have been initialized
+	 * already.
+	 */
+	assert(arena != NULL);
+
+	return extent_alloc_default_impl(tsdn, arena, new_addr, size,
+	    ALIGNMENT_CEILING(alignment, PAGE), zero, commit);
+}
+
+static void
+extent_hook_pre_reentrancy(tsdn_t *tsdn, arena_t *arena) {
+	tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+	if (arena == arena_get(tsd_tsdn(tsd), 0, false)) {
+		/*
+		 * The only legitimate case of customized extent hooks for a0 is
+		 * hooks with no allocation activities.  One such example is to
+		 * place metadata on pre-allocated resources such as huge pages.
+		 * In that case, rely on reentrancy_level checks to catch
+		 * infinite recursions.
+		 */
+		pre_reentrancy(tsd, NULL);
+	} else {
+		pre_reentrancy(tsd, arena);
+	}
+}
+
+static void
+extent_hook_post_reentrancy(tsdn_t *tsdn) {
+	tsd_t *tsd = tsdn_null(tsdn) ? tsd_fetch() : tsdn_tsd(tsdn);
+	post_reentrancy(tsd);
+}
+
+/*
+ * If virtual memory is retained, create increasingly larger extents from which
+ * to split requested extents in order to limit the total number of disjoint
+ * virtual memory ranges retained by each arena.
+ */
+static extent_t *
+extent_grow_retained(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, size_t size, size_t pad, size_t alignment,
+    bool slab, szind_t szind, bool *zero, bool *commit) {
+	malloc_mutex_assert_owner(tsdn, &arena->extent_grow_mtx);
+	assert(pad == 0 || !slab);
+	assert(!*zero || !slab);
+
+	size_t esize = size + pad;
+	size_t alloc_size_min = esize + PAGE_CEILING(alignment) - PAGE;
+	/* Beware size_t wrap-around. */
+	if (alloc_size_min < esize) {
+		goto label_err;
+	}
+	/*
+	 * Find the next extent size in the series that would be large enough to
+	 * satisfy this request.
+	 */
+	pszind_t egn_skip = 0;
+	size_t alloc_size = sz_pind2sz(arena->extent_grow_next + egn_skip);
+	while (alloc_size < alloc_size_min) {
+		egn_skip++;
+		if (arena->extent_grow_next + egn_skip >=
+		    sz_psz2ind(SC_LARGE_MAXCLASS)) {
+			/* Outside legal range. */
+			goto label_err;
+		}
+		alloc_size = sz_pind2sz(arena->extent_grow_next + egn_skip);
+	}
+
+	extent_t *extent = extent_alloc(tsdn, arena);
+	if (extent == NULL) {
+		goto label_err;
+	}
+	bool zeroed = false;
+	bool committed = false;
+
+	void *ptr;
+	if (*r_extent_hooks == &extent_hooks_default) {
+		ptr = extent_alloc_default_impl(tsdn, arena, NULL,
+		    alloc_size, PAGE, &zeroed, &committed);
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		ptr = (*r_extent_hooks)->alloc(*r_extent_hooks, NULL,
+		    alloc_size, PAGE, &zeroed, &committed,
+		    arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	extent_init(extent, arena, ptr, alloc_size, false, SC_NSIZES,
+	    arena_extent_sn_next(arena), extent_state_active, zeroed,
+	    committed, true, EXTENT_IS_HEAD);
+	if (ptr == NULL) {
+		extent_dalloc(tsdn, arena, extent);
+		goto label_err;
+	}
+
+	if (extent_register_no_gdump_add(tsdn, extent)) {
+		extent_dalloc(tsdn, arena, extent);
+		goto label_err;
+	}
+
+	if (extent_zeroed_get(extent) && extent_committed_get(extent)) {
+		*zero = true;
+	}
+	if (extent_committed_get(extent)) {
+		*commit = true;
+	}
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *lead;
+	extent_t *trail;
+	extent_t *to_leak;
+	extent_t *to_salvage;
+	extent_split_interior_result_t result = extent_split_interior(
+	    tsdn, arena, r_extent_hooks, rtree_ctx, &extent, &lead, &trail,
+	    &to_leak, &to_salvage, NULL, size, pad, alignment, slab, szind,
+	    true);
+
+	if (result == extent_split_interior_ok) {
+		if (lead != NULL) {
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, lead, true);
+		}
+		if (trail != NULL) {
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, trail, true);
+		}
+	} else {
+		/*
+		 * We should have allocated a sufficiently large extent; the
+		 * cant_alloc case should not occur.
+		 */
+		assert(result == extent_split_interior_error);
+		if (to_salvage != NULL) {
+			if (config_prof) {
+				extent_gdump_add(tsdn, to_salvage);
+			}
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, to_salvage, true);
+		}
+		if (to_leak != NULL) {
+			extent_deregister_no_gdump_sub(tsdn, to_leak);
+			extents_abandon_vm(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, to_leak, true);
+		}
+		goto label_err;
+	}
+
+	if (*commit && !extent_committed_get(extent)) {
+		if (extent_commit_impl(tsdn, arena, r_extent_hooks, extent, 0,
+		    extent_size_get(extent), true)) {
+			extent_record(tsdn, arena, r_extent_hooks,
+			    &arena->extents_retained, extent, true);
+			goto label_err;
+		}
+		if (!extent_need_manual_zero(arena)) {
+			extent_zeroed_set(extent, true);
+		}
+	}
+
+	/*
+	 * Increment extent_grow_next if doing so wouldn't exceed the allowed
+	 * range.
+	 */
+	if (arena->extent_grow_next + egn_skip + 1 <=
+	    arena->retain_grow_limit) {
+		arena->extent_grow_next += egn_skip + 1;
+	} else {
+		arena->extent_grow_next = arena->retain_grow_limit;
+	}
+	/* All opportunities for failure are past. */
+	malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+
+	if (config_prof) {
+		/* Adjust gdump stats now that extent is final size. */
+		extent_gdump_add(tsdn, extent);
+	}
+	if (pad != 0) {
+		extent_addr_randomize(tsdn, extent, alignment);
+	}
+	if (slab) {
+		rtree_ctx_t rtree_ctx_fallback;
+		rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+		    &rtree_ctx_fallback);
+
+		extent_slab_set(extent, true);
+		extent_interior_register(tsdn, rtree_ctx, extent, szind);
+	}
+	if (*zero && !extent_zeroed_get(extent)) {
+		void *addr = extent_base_get(extent);
+		size_t size = extent_size_get(extent);
+		if (extent_need_manual_zero(arena) ||
+		    pages_purge_forced(addr, size)) {
+			memset(addr, 0, size);
+		}
+	}
+
+	return extent;
+label_err:
+	malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+	return NULL;
+}
+
+static extent_t *
+extent_alloc_retained(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	assert(size != 0);
+	assert(alignment != 0);
+
+	malloc_mutex_lock(tsdn, &arena->extent_grow_mtx);
+
+	extent_t *extent = extent_recycle(tsdn, arena, r_extent_hooks,
+	    &arena->extents_retained, new_addr, size, pad, alignment, slab,
+	    szind, zero, commit, true);
+	if (extent != NULL) {
+		malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+		if (config_prof) {
+			extent_gdump_add(tsdn, extent);
+		}
+	} else if (opt_retain && new_addr == NULL) {
+		extent = extent_grow_retained(tsdn, arena, r_extent_hooks, size,
+		    pad, alignment, slab, szind, zero, commit);
+		/* extent_grow_retained() always releases extent_grow_mtx. */
+	} else {
+		malloc_mutex_unlock(tsdn, &arena->extent_grow_mtx);
+	}
+	malloc_mutex_assert_not_owner(tsdn, &arena->extent_grow_mtx);
+
+	return extent;
+}
+
+static extent_t *
+extent_alloc_wrapper_hard(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	size_t esize = size + pad;
+	extent_t *extent = extent_alloc(tsdn, arena);
+	if (extent == NULL) {
+		return NULL;
+	}
+	void *addr;
+	size_t palignment = ALIGNMENT_CEILING(alignment, PAGE);
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		addr = extent_alloc_default_impl(tsdn, arena, new_addr, esize,
+		    palignment, zero, commit);
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		addr = (*r_extent_hooks)->alloc(*r_extent_hooks, new_addr,
+		    esize, palignment, zero, commit, arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+	if (addr == NULL) {
+		extent_dalloc(tsdn, arena, extent);
+		return NULL;
+	}
+	extent_init(extent, arena, addr, esize, slab, szind,
+	    arena_extent_sn_next(arena), extent_state_active, *zero, *commit,
+	    true, EXTENT_NOT_HEAD);
+	if (pad != 0) {
+		extent_addr_randomize(tsdn, extent, alignment);
+	}
+	if (extent_register(tsdn, extent)) {
+		extent_dalloc(tsdn, arena, extent);
+		return NULL;
+	}
+
+	return extent;
+}
+
+extent_t *
+extent_alloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	extent_t *extent = extent_alloc_retained(tsdn, arena, r_extent_hooks,
+	    new_addr, size, pad, alignment, slab, szind, zero, commit);
+	if (extent == NULL) {
+		if (opt_retain && new_addr != NULL) {
+			/*
+			 * When retain is enabled and new_addr is set, we do not
+			 * attempt extent_alloc_wrapper_hard which does mmap
+			 * that is very unlikely to succeed (unless it happens
+			 * to be at the end).
+			 */
+			return NULL;
+		}
+		extent = extent_alloc_wrapper_hard(tsdn, arena, r_extent_hooks,
+		    new_addr, size, pad, alignment, slab, szind, zero, commit);
+	}
+
+	assert(extent == NULL || extent_dumpable_get(extent));
+	return extent;
+}
+
+static bool
+extent_can_coalesce(arena_t *arena, extents_t *extents, const extent_t *inner,
+    const extent_t *outer) {
+	assert(extent_arena_get(inner) == arena);
+	if (extent_arena_get(outer) != arena) {
+		return false;
+	}
+
+	assert(extent_state_get(inner) == extent_state_active);
+	if (extent_state_get(outer) != extents->state) {
+		return false;
+	}
+
+	if (extent_committed_get(inner) != extent_committed_get(outer)) {
+		return false;
+	}
+
+	return true;
+}
+
+static bool
+extent_coalesce(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *inner, extent_t *outer, bool forward,
+    bool growing_retained) {
+	assert(extent_can_coalesce(arena, extents, inner, outer));
+
+	extent_activate_locked(tsdn, arena, extents, outer);
+
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+	bool err = extent_merge_impl(tsdn, arena, r_extent_hooks,
+	    forward ? inner : outer, forward ? outer : inner, growing_retained);
+	malloc_mutex_lock(tsdn, &extents->mtx);
+
+	if (err) {
+		extent_deactivate_locked(tsdn, arena, extents, outer);
+	}
+
+	return err;
+}
+
+static extent_t *
+extent_try_coalesce_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained,
+    bool inactive_only) {
+	/*
+	 * We avoid checking / locking inactive neighbors for large size
+	 * classes, since they are eagerly coalesced on deallocation which can
+	 * cause lock contention.
+	 */
+	/*
+	 * Continue attempting to coalesce until failure, to protect against
+	 * races with other threads that are thwarted by this one.
+	 */
+	bool again;
+	do {
+		again = false;
+
+		/* Try to coalesce forward. */
+		extent_t *next = extent_lock_from_addr(tsdn, rtree_ctx,
+		    extent_past_get(extent), inactive_only);
+		if (next != NULL) {
+			/*
+			 * extents->mtx only protects against races for
+			 * like-state extents, so call extent_can_coalesce()
+			 * before releasing next's pool lock.
+			 */
+			bool can_coalesce = extent_can_coalesce(arena, extents,
+			    extent, next);
+
+			extent_unlock(tsdn, next);
+
+			if (can_coalesce && !extent_coalesce(tsdn, arena,
+			    r_extent_hooks, extents, extent, next, true,
+			    growing_retained)) {
+				if (extents->delay_coalesce) {
+					/* Do minimal coalescing. */
+					*coalesced = true;
+					return extent;
+				}
+				again = true;
+			}
+		}
+
+		/* Try to coalesce backward. */
+		extent_t *prev = extent_lock_from_addr(tsdn, rtree_ctx,
+		    extent_before_get(extent), inactive_only);
+		if (prev != NULL) {
+			bool can_coalesce = extent_can_coalesce(arena, extents,
+			    extent, prev);
+			extent_unlock(tsdn, prev);
+
+			if (can_coalesce && !extent_coalesce(tsdn, arena,
+			    r_extent_hooks, extents, extent, prev, false,
+			    growing_retained)) {
+				extent = prev;
+				if (extents->delay_coalesce) {
+					/* Do minimal coalescing. */
+					*coalesced = true;
+					return extent;
+				}
+				again = true;
+			}
+		}
+	} while (again);
+
+	if (extents->delay_coalesce) {
+		*coalesced = false;
+	}
+	return extent;
+}
+
+static extent_t *
+extent_try_coalesce(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained) {
+	return extent_try_coalesce_impl(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, extent, coalesced, growing_retained, false);
+}
+
+static extent_t *
+extent_try_coalesce_large(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, rtree_ctx_t *rtree_ctx, extents_t *extents,
+    extent_t *extent, bool *coalesced, bool growing_retained) {
+	return extent_try_coalesce_impl(tsdn, arena, r_extent_hooks, rtree_ctx,
+	    extents, extent, coalesced, growing_retained, true);
+}
+
+/*
+ * Does the metadata management portions of putting an unused extent into the
+ * given extents_t (coalesces, deregisters slab interiors, the heap operations).
+ */
+static void
+extent_record(tsdn_t *tsdn, arena_t *arena, extent_hooks_t **r_extent_hooks,
+    extents_t *extents, extent_t *extent, bool growing_retained) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	assert((extents_state_get(extents) != extent_state_dirty &&
+	    extents_state_get(extents) != extent_state_muzzy) ||
+	    !extent_zeroed_get(extent));
+
+	malloc_mutex_lock(tsdn, &extents->mtx);
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	extent_szind_set(extent, SC_NSIZES);
+	if (extent_slab_get(extent)) {
+		extent_interior_deregister(tsdn, rtree_ctx, extent);
+		extent_slab_set(extent, false);
+	}
+
+	assert(rtree_extent_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_base_get(extent), true) == extent);
+
+	if (!extents->delay_coalesce) {
+		extent = extent_try_coalesce(tsdn, arena, r_extent_hooks,
+		    rtree_ctx, extents, extent, NULL, growing_retained);
+	} else if (extent_size_get(extent) >= SC_LARGE_MINCLASS) {
+		assert(extents == &arena->extents_dirty);
+		/* Always coalesce large extents eagerly. */
+		bool coalesced;
+		do {
+			assert(extent_state_get(extent) == extent_state_active);
+			extent = extent_try_coalesce_large(tsdn, arena,
+			    r_extent_hooks, rtree_ctx, extents, extent,
+			    &coalesced, growing_retained);
+		} while (coalesced);
+		if (extent_size_get(extent) >= oversize_threshold) {
+			/* Shortcut to purge the oversize extent eagerly. */
+			malloc_mutex_unlock(tsdn, &extents->mtx);
+			arena_decay_extent(tsdn, arena, r_extent_hooks, extent);
+			return;
+		}
+	}
+	extent_deactivate_locked(tsdn, arena, extents, extent);
+
+	malloc_mutex_unlock(tsdn, &extents->mtx);
+}
+
+void
+extent_dalloc_gap(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (extent_register(tsdn, extent)) {
+		extent_dalloc(tsdn, arena, extent);
+		return;
+	}
+	extent_dalloc_wrapper(tsdn, arena, &extent_hooks, extent);
+}
+
+static bool
+extent_may_dalloc(void) {
+	/* With retain enabled, the default dalloc always fails. */
+	return !opt_retain;
+}
+
+static bool
+extent_dalloc_default_impl(void *addr, size_t size) {
+	if (!have_dss || !extent_in_dss(addr)) {
+		return extent_dalloc_mmap(addr, size);
+	}
+	return true;
+}
+
+static bool
+extent_dalloc_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	return extent_dalloc_default_impl(addr, size);
+}
+
+static bool
+extent_dalloc_wrapper_try(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	bool err;
+
+	assert(extent_base_get(extent) != NULL);
+	assert(extent_size_get(extent) != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_addr_set(extent, extent_base_get(extent));
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	/* Try to deallocate. */
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		err = extent_dalloc_default_impl(extent_base_get(extent),
+		    extent_size_get(extent));
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		err = ((*r_extent_hooks)->dalloc == NULL ||
+		    (*r_extent_hooks)->dalloc(*r_extent_hooks,
+		    extent_base_get(extent), extent_size_get(extent),
+		    extent_committed_get(extent), arena_ind_get(arena)));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	if (!err) {
+		extent_dalloc(tsdn, arena, extent);
+	}
+
+	return err;
+}
+
+void
+extent_dalloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	assert(extent_dumpable_get(extent));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	/* Avoid calling the default extent_dalloc unless have to. */
+	if (*r_extent_hooks != &extent_hooks_default || extent_may_dalloc()) {
+		/*
+		 * Deregister first to avoid a race with other allocating
+		 * threads, and reregister if deallocation fails.
+		 */
+		extent_deregister(tsdn, extent);
+		if (!extent_dalloc_wrapper_try(tsdn, arena, r_extent_hooks,
+		    extent)) {
+			return;
+		}
+		extent_reregister(tsdn, extent);
+	}
+
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	/* Try to decommit; purge if that fails. */
+	bool zeroed;
+	if (!extent_committed_get(extent)) {
+		zeroed = true;
+	} else if (!extent_decommit_wrapper(tsdn, arena, r_extent_hooks, extent,
+	    0, extent_size_get(extent))) {
+		zeroed = true;
+	} else if ((*r_extent_hooks)->purge_forced != NULL &&
+	    !(*r_extent_hooks)->purge_forced(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), 0,
+	    extent_size_get(extent), arena_ind_get(arena))) {
+		zeroed = true;
+	} else if (extent_state_get(extent) == extent_state_muzzy ||
+	    ((*r_extent_hooks)->purge_lazy != NULL &&
+	    !(*r_extent_hooks)->purge_lazy(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), 0,
+	    extent_size_get(extent), arena_ind_get(arena)))) {
+		zeroed = false;
+	} else {
+		zeroed = false;
+	}
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	extent_zeroed_set(extent, zeroed);
+
+	if (config_prof) {
+		extent_gdump_sub(tsdn, extent);
+	}
+
+	extent_record(tsdn, arena, r_extent_hooks, &arena->extents_retained,
+	    extent, false);
+}
+
+static void
+extent_destroy_default_impl(void *addr, size_t size) {
+	if (!have_dss || !extent_in_dss(addr)) {
+		pages_unmap(addr, size);
+	}
+}
+
+static void
+extent_destroy_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	extent_destroy_default_impl(addr, size);
+}
+
+void
+extent_destroy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent) {
+	assert(extent_base_get(extent) != NULL);
+	assert(extent_size_get(extent) != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	/* Deregister first to avoid a race with other allocating threads. */
+	extent_deregister(tsdn, extent);
+
+	extent_addr_set(extent, extent_base_get(extent));
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	/* Try to destroy; silently fail otherwise. */
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		extent_destroy_default_impl(extent_base_get(extent),
+		    extent_size_get(extent));
+	} else if ((*r_extent_hooks)->destroy != NULL) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		(*r_extent_hooks)->destroy(*r_extent_hooks,
+		    extent_base_get(extent), extent_size_get(extent),
+		    extent_committed_get(extent), arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	extent_dalloc(tsdn, arena, extent);
+}
+
+static bool
+extent_commit_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	return pages_commit((void *)((uintptr_t)addr + (uintptr_t)offset),
+	    length);
+}
+
+static bool
+extent_commit_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = ((*r_extent_hooks)->commit == NULL ||
+	    (*r_extent_hooks)->commit(*r_extent_hooks, extent_base_get(extent),
+	    extent_size_get(extent), offset, length, arena_ind_get(arena)));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	extent_committed_set(extent, extent_committed_get(extent) || !err);
+	return err;
+}
+
+bool
+extent_commit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	return extent_commit_impl(tsdn, arena, r_extent_hooks, extent, offset,
+	    length, false);
+}
+
+static bool
+extent_decommit_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	return pages_decommit((void *)((uintptr_t)addr + (uintptr_t)offset),
+	    length);
+}
+
+bool
+extent_decommit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = ((*r_extent_hooks)->decommit == NULL ||
+	    (*r_extent_hooks)->decommit(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), offset, length,
+	    arena_ind_get(arena)));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	extent_committed_set(extent, extent_committed_get(extent) && err);
+	return err;
+}
+
+#ifdef PAGES_CAN_PURGE_LAZY
+static bool
+extent_purge_lazy_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	assert(addr != NULL);
+	assert((offset & PAGE_MASK) == 0);
+	assert(length != 0);
+	assert((length & PAGE_MASK) == 0);
+
+	return pages_purge_lazy((void *)((uintptr_t)addr + (uintptr_t)offset),
+	    length);
+}
+#endif
+
+static bool
+extent_purge_lazy_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->purge_lazy == NULL) {
+		return true;
+	}
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = (*r_extent_hooks)->purge_lazy(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), offset, length,
+	    arena_ind_get(arena));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	return err;
+}
+
+bool
+extent_purge_lazy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	return extent_purge_lazy_impl(tsdn, arena, r_extent_hooks, extent,
+	    offset, length, false);
+}
+
+#ifdef PAGES_CAN_PURGE_FORCED
+static bool
+extent_purge_forced_default(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind) {
+	assert(addr != NULL);
+	assert((offset & PAGE_MASK) == 0);
+	assert(length != 0);
+	assert((length & PAGE_MASK) == 0);
+
+	return pages_purge_forced((void *)((uintptr_t)addr +
+	    (uintptr_t)offset), length);
+}
+#endif
+
+static bool
+extent_purge_forced_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length, bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->purge_forced == NULL) {
+		return true;
+	}
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = (*r_extent_hooks)->purge_forced(*r_extent_hooks,
+	    extent_base_get(extent), extent_size_get(extent), offset, length,
+	    arena_ind_get(arena));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	return err;
+}
+
+bool
+extent_purge_forced_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length) {
+	return extent_purge_forced_impl(tsdn, arena, r_extent_hooks, extent,
+	    offset, length, false);
+}
+
+static bool
+extent_split_default(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t size_a, size_t size_b, bool committed, unsigned arena_ind) {
+	if (!maps_coalesce) {
+		/*
+		 * Without retain, only whole regions can be purged (required by
+		 * MEM_RELEASE on Windows) -- therefore disallow splitting.  See
+		 * comments in extent_head_no_merge().
+		 */
+		return !opt_retain;
+	}
+
+	return false;
+}
+
+/*
+ * Accepts the extent to split, and the characteristics of each side of the
+ * split.  The 'a' parameters go with the 'lead' of the resulting pair of
+ * extents (the lower addressed portion of the split), and the 'b' parameters go
+ * with the trail (the higher addressed portion).  This makes 'extent' the lead,
+ * and returns the trail (except in case of error).
+ */
+static extent_t *
+extent_split_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b,
+    bool growing_retained) {
+	assert(extent_size_get(extent) == size_a + size_b);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->split == NULL) {
+		return NULL;
+	}
+
+	extent_t *trail = extent_alloc(tsdn, arena);
+	if (trail == NULL) {
+		goto label_error_a;
+	}
+
+	extent_init(trail, arena, (void *)((uintptr_t)extent_base_get(extent) +
+	    size_a), size_b, slab_b, szind_b, extent_sn_get(extent),
+	    extent_state_get(extent), extent_zeroed_get(extent),
+	    extent_committed_get(extent), extent_dumpable_get(extent),
+	    EXTENT_NOT_HEAD);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *lead_elm_a, *lead_elm_b;
+	{
+		extent_t lead;
+
+		extent_init(&lead, arena, extent_addr_get(extent), size_a,
+		    slab_a, szind_a, extent_sn_get(extent),
+		    extent_state_get(extent), extent_zeroed_get(extent),
+		    extent_committed_get(extent), extent_dumpable_get(extent),
+		    EXTENT_NOT_HEAD);
+
+		extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, &lead, false,
+		    true, &lead_elm_a, &lead_elm_b);
+	}
+	rtree_leaf_elm_t *trail_elm_a, *trail_elm_b;
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, trail, false, true,
+	    &trail_elm_a, &trail_elm_b);
+
+	if (lead_elm_a == NULL || lead_elm_b == NULL || trail_elm_a == NULL
+	    || trail_elm_b == NULL) {
+		goto label_error_b;
+	}
+
+	extent_lock2(tsdn, extent, trail);
+
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_pre_reentrancy(tsdn, arena);
+	}
+	bool err = (*r_extent_hooks)->split(*r_extent_hooks, extent_base_get(extent),
+	    size_a + size_b, size_a, size_b, extent_committed_get(extent),
+	    arena_ind_get(arena));
+	if (*r_extent_hooks != &extent_hooks_default) {
+		extent_hook_post_reentrancy(tsdn);
+	}
+	if (err) {
+		goto label_error_c;
+	}
+
+	extent_size_set(extent, size_a);
+	extent_szind_set(extent, szind_a);
+
+	extent_rtree_write_acquired(tsdn, lead_elm_a, lead_elm_b, extent,
+	    szind_a, slab_a);
+	extent_rtree_write_acquired(tsdn, trail_elm_a, trail_elm_b, trail,
+	    szind_b, slab_b);
+
+	extent_unlock2(tsdn, extent, trail);
+
+	return trail;
+label_error_c:
+	extent_unlock2(tsdn, extent, trail);
+label_error_b:
+	extent_dalloc(tsdn, arena, trail);
+label_error_a:
+	return NULL;
+}
+
+extent_t *
+extent_split_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b) {
+	return extent_split_impl(tsdn, arena, r_extent_hooks, extent, size_a,
+	    szind_a, slab_a, size_b, szind_b, slab_b, false);
+}
+
+static bool
+extent_merge_default_impl(void *addr_a, void *addr_b) {
+	if (!maps_coalesce && !opt_retain) {
+		return true;
+	}
+	if (have_dss && !extent_dss_mergeable(addr_a, addr_b)) {
+		return true;
+	}
+
+	return false;
+}
+
+/*
+ * Returns true if the given extents can't be merged because of their head bit
+ * settings.  Assumes the second extent has the higher address.
+ */
+static bool
+extent_head_no_merge(extent_t *a, extent_t *b) {
+	assert(extent_base_get(a) < extent_base_get(b));
+	/*
+	 * When coalesce is not always allowed (Windows), only merge extents
+	 * from the same VirtualAlloc region under opt.retain (in which case
+	 * MEM_DECOMMIT is utilized for purging).
+	 */
+	if (maps_coalesce) {
+		return false;
+	}
+	if (!opt_retain) {
+		return true;
+	}
+	/* If b is a head extent, disallow the cross-region merge. */
+	if (extent_is_head_get(b)) {
+		/*
+		 * Additionally, sn should not overflow with retain; sanity
+		 * check that different regions have unique sn.
+		 */
+		assert(extent_sn_comp(a, b) != 0);
+		return true;
+	}
+	assert(extent_sn_comp(a, b) == 0);
+
+	return false;
+}
+
+static bool
+extent_merge_default(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a,
+    void *addr_b, size_t size_b, bool committed, unsigned arena_ind) {
+	if (!maps_coalesce) {
+		tsdn_t *tsdn = tsdn_fetch();
+		extent_t *a = iealloc(tsdn, addr_a);
+		extent_t *b = iealloc(tsdn, addr_b);
+		if (extent_head_no_merge(a, b)) {
+			return true;
+		}
+	}
+	return extent_merge_default_impl(addr_a, addr_b);
+}
+
+static bool
+extent_merge_impl(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b,
+    bool growing_retained) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, growing_retained ? 1 : 0);
+	assert(extent_base_get(a) < extent_base_get(b));
+
+	extent_hooks_assure_initialized(arena, r_extent_hooks);
+
+	if ((*r_extent_hooks)->merge == NULL || extent_head_no_merge(a, b)) {
+		return true;
+	}
+
+	bool err;
+	if (*r_extent_hooks == &extent_hooks_default) {
+		/* Call directly to propagate tsdn. */
+		err = extent_merge_default_impl(extent_base_get(a),
+		    extent_base_get(b));
+	} else {
+		extent_hook_pre_reentrancy(tsdn, arena);
+		err = (*r_extent_hooks)->merge(*r_extent_hooks,
+		    extent_base_get(a), extent_size_get(a), extent_base_get(b),
+		    extent_size_get(b), extent_committed_get(a),
+		    arena_ind_get(arena));
+		extent_hook_post_reentrancy(tsdn);
+	}
+
+	if (err) {
+		return true;
+	}
+
+	/*
+	 * The rtree writes must happen while all the relevant elements are
+	 * owned, so the following code uses decomposed helper functions rather
+	 * than extent_{,de}register() to do things in the right order.
+	 */
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	rtree_leaf_elm_t *a_elm_a, *a_elm_b, *b_elm_a, *b_elm_b;
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, a, true, false, &a_elm_a,
+	    &a_elm_b);
+	extent_rtree_leaf_elms_lookup(tsdn, rtree_ctx, b, true, false, &b_elm_a,
+	    &b_elm_b);
+
+	extent_lock2(tsdn, a, b);
+
+	if (a_elm_b != NULL) {
+		rtree_leaf_elm_write(tsdn, &extents_rtree, a_elm_b, NULL,
+		    SC_NSIZES, false);
+	}
+	if (b_elm_b != NULL) {
+		rtree_leaf_elm_write(tsdn, &extents_rtree, b_elm_a, NULL,
+		    SC_NSIZES, false);
+	} else {
+		b_elm_b = b_elm_a;
+	}
+
+	extent_size_set(a, extent_size_get(a) + extent_size_get(b));
+	extent_szind_set(a, SC_NSIZES);
+	extent_sn_set(a, (extent_sn_get(a) < extent_sn_get(b)) ?
+	    extent_sn_get(a) : extent_sn_get(b));
+	extent_zeroed_set(a, extent_zeroed_get(a) && extent_zeroed_get(b));
+
+	extent_rtree_write_acquired(tsdn, a_elm_a, b_elm_b, a, SC_NSIZES,
+	    false);
+
+	extent_unlock2(tsdn, a, b);
+
+	extent_dalloc(tsdn, extent_arena_get(b), b);
+
+	return false;
+}
+
+bool
+extent_merge_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b) {
+	return extent_merge_impl(tsdn, arena, r_extent_hooks, a, b, false);
+}
+
+bool
+extent_boot(void) {
+	if (rtree_new(&extents_rtree, true)) {
+		return true;
+	}
+
+	if (mutex_pool_init(&extent_mutex_pool, "extent_mutex_pool",
+	    WITNESS_RANK_EXTENT_POOL)) {
+		return true;
+	}
+
+	if (have_dss) {
+		extent_dss_boot();
+	}
+
+	return false;
+}
+
+void
+extent_util_stats_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size) {
+	assert(ptr != NULL && nfree != NULL && nregs != NULL && size != NULL);
+
+	const extent_t *extent = iealloc(tsdn, ptr);
+	if (unlikely(extent == NULL)) {
+		*nfree = *nregs = *size = 0;
+		return;
+	}
+
+	*size = extent_size_get(extent);
+	if (!extent_slab_get(extent)) {
+		*nfree = 0;
+		*nregs = 1;
+	} else {
+		*nfree = extent_nfree_get(extent);
+		*nregs = bin_infos[extent_szind_get(extent)].nregs;
+		assert(*nfree <= *nregs);
+		assert(*nfree * extent_usize_get(extent) <= *size);
+	}
+}
+
+void
+extent_util_stats_verbose_get(tsdn_t *tsdn, const void *ptr,
+    size_t *nfree, size_t *nregs, size_t *size,
+    size_t *bin_nfree, size_t *bin_nregs, void **slabcur_addr) {
+	assert(ptr != NULL && nfree != NULL && nregs != NULL && size != NULL
+	    && bin_nfree != NULL && bin_nregs != NULL && slabcur_addr != NULL);
+
+	const extent_t *extent = iealloc(tsdn, ptr);
+	if (unlikely(extent == NULL)) {
+		*nfree = *nregs = *size = *bin_nfree = *bin_nregs = 0;
+		*slabcur_addr = NULL;
+		return;
+	}
+
+	*size = extent_size_get(extent);
+	if (!extent_slab_get(extent)) {
+		*nfree = *bin_nfree = *bin_nregs = 0;
+		*nregs = 1;
+		*slabcur_addr = NULL;
+		return;
+	}
+
+	*nfree = extent_nfree_get(extent);
+	const szind_t szind = extent_szind_get(extent);
+	*nregs = bin_infos[szind].nregs;
+	assert(*nfree <= *nregs);
+	assert(*nfree * extent_usize_get(extent) <= *size);
+
+	const arena_t *arena = extent_arena_get(extent);
+	assert(arena != NULL);
+	const unsigned binshard = extent_binshard_get(extent);
+	bin_t *bin = &arena->bins[szind].bin_shards[binshard];
+
+	malloc_mutex_lock(tsdn, &bin->lock);
+	if (config_stats) {
+		*bin_nregs = *nregs * bin->stats.curslabs;
+		assert(*bin_nregs >= bin->stats.curregs);
+		*bin_nfree = *bin_nregs - bin->stats.curregs;
+	} else {
+		*bin_nfree = *bin_nregs = 0;
+	}
+	extent_t *slab;
+	if (bin->slabcur != NULL) {
+		slab = bin->slabcur;
+	} else {
+		slab = extent_heap_first(&bin->slabs_nonfull);
+	}
+	*slabcur_addr = slab != NULL ? extent_addr_get(slab) : NULL;
+	malloc_mutex_unlock(tsdn, &bin->lock);
+}
diff --git a/memkind-1.10.0/jemalloc/src/extent_dss.c b/memkind-1.10.0/jemalloc/src/extent_dss.c
new file mode 100644
index 0000000..8581789
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/extent_dss.c
@@ -0,0 +1,271 @@
+#define JEMALLOC_EXTENT_DSS_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/spin.h"
+
+/******************************************************************************/
+/* Data. */
+
+const char	*opt_dss = DSS_DEFAULT;
+
+const char	*dss_prec_names[] = {
+	"disabled",
+	"primary",
+	"secondary",
+	"N/A"
+};
+
+/*
+ * Current dss precedence default, used when creating new arenas.  NB: This is
+ * stored as unsigned rather than dss_prec_t because in principle there's no
+ * guarantee that sizeof(dss_prec_t) is the same as sizeof(unsigned), and we use
+ * atomic operations to synchronize the setting.
+ */
+static atomic_u_t	dss_prec_default = ATOMIC_INIT(
+    (unsigned)DSS_PREC_DEFAULT);
+
+/* Base address of the DSS. */
+static void		*dss_base;
+/* Atomic boolean indicating whether a thread is currently extending DSS. */
+static atomic_b_t	dss_extending;
+/* Atomic boolean indicating whether the DSS is exhausted. */
+static atomic_b_t	dss_exhausted;
+/* Atomic current upper limit on DSS addresses. */
+static atomic_p_t	dss_max;
+
+/******************************************************************************/
+
+static void *
+extent_dss_sbrk(intptr_t increment) {
+#ifdef JEMALLOC_DSS
+	return sbrk(increment);
+#else
+	not_implemented();
+	return NULL;
+#endif
+}
+
+dss_prec_t
+extent_dss_prec_get(void) {
+	dss_prec_t ret;
+
+	if (!have_dss) {
+		return dss_prec_disabled;
+	}
+	ret = (dss_prec_t)atomic_load_u(&dss_prec_default, ATOMIC_ACQUIRE);
+	return ret;
+}
+
+bool
+extent_dss_prec_set(dss_prec_t dss_prec) {
+	if (!have_dss) {
+		return (dss_prec != dss_prec_disabled);
+	}
+	atomic_store_u(&dss_prec_default, (unsigned)dss_prec, ATOMIC_RELEASE);
+	return false;
+}
+
+static void
+extent_dss_extending_start(void) {
+	spin_t spinner = SPIN_INITIALIZER;
+	while (true) {
+		bool expected = false;
+		if (atomic_compare_exchange_weak_b(&dss_extending, &expected,
+		    true, ATOMIC_ACQ_REL, ATOMIC_RELAXED)) {
+			break;
+		}
+		spin_adaptive(&spinner);
+	}
+}
+
+static void
+extent_dss_extending_finish(void) {
+	assert(atomic_load_b(&dss_extending, ATOMIC_RELAXED));
+
+	atomic_store_b(&dss_extending, false, ATOMIC_RELEASE);
+}
+
+static void *
+extent_dss_max_update(void *new_addr) {
+	/*
+	 * Get the current end of the DSS as max_cur and assure that dss_max is
+	 * up to date.
+	 */
+	void *max_cur = extent_dss_sbrk(0);
+	if (max_cur == (void *)-1) {
+		return NULL;
+	}
+	atomic_store_p(&dss_max, max_cur, ATOMIC_RELEASE);
+	/* Fixed new_addr can only be supported if it is at the edge of DSS. */
+	if (new_addr != NULL && max_cur != new_addr) {
+		return NULL;
+	}
+	return max_cur;
+}
+
+void *
+extent_alloc_dss(tsdn_t *tsdn, arena_t *arena, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit) {
+	extent_t *gap;
+
+	cassert(have_dss);
+	assert(size > 0);
+	assert(alignment == ALIGNMENT_CEILING(alignment, PAGE));
+
+	/*
+	 * sbrk() uses a signed increment argument, so take care not to
+	 * interpret a large allocation request as a negative increment.
+	 */
+	if ((intptr_t)size < 0) {
+		return NULL;
+	}
+
+	gap = extent_alloc(tsdn, arena);
+	if (gap == NULL) {
+		return NULL;
+	}
+
+	extent_dss_extending_start();
+	if (!atomic_load_b(&dss_exhausted, ATOMIC_ACQUIRE)) {
+		/*
+		 * The loop is necessary to recover from races with other
+		 * threads that are using the DSS for something other than
+		 * malloc.
+		 */
+		while (true) {
+			void *max_cur = extent_dss_max_update(new_addr);
+			if (max_cur == NULL) {
+				goto label_oom;
+			}
+
+			/*
+			 * Compute how much page-aligned gap space (if any) is
+			 * necessary to satisfy alignment.  This space can be
+			 * recycled for later use.
+			 */
+			void *gap_addr_page = (void *)(PAGE_CEILING(
+			    (uintptr_t)max_cur));
+			void *ret = (void *)ALIGNMENT_CEILING(
+			    (uintptr_t)gap_addr_page, alignment);
+			size_t gap_size_page = (uintptr_t)ret -
+			    (uintptr_t)gap_addr_page;
+			if (gap_size_page != 0) {
+				extent_init(gap, arena, gap_addr_page,
+				    gap_size_page, false, SC_NSIZES,
+				    arena_extent_sn_next(arena),
+				    extent_state_active, false, true, true,
+				    EXTENT_NOT_HEAD);
+			}
+			/*
+			 * Compute the address just past the end of the desired
+			 * allocation space.
+			 */
+			void *dss_next = (void *)((uintptr_t)ret + size);
+			if ((uintptr_t)ret < (uintptr_t)max_cur ||
+			    (uintptr_t)dss_next < (uintptr_t)max_cur) {
+				goto label_oom; /* Wrap-around. */
+			}
+			/* Compute the increment, including subpage bytes. */
+			void *gap_addr_subpage = max_cur;
+			size_t gap_size_subpage = (uintptr_t)ret -
+			    (uintptr_t)gap_addr_subpage;
+			intptr_t incr = gap_size_subpage + size;
+
+			assert((uintptr_t)max_cur + incr == (uintptr_t)ret +
+			    size);
+
+			/* Try to allocate. */
+			void *dss_prev = extent_dss_sbrk(incr);
+			if (dss_prev == max_cur) {
+				/* Success. */
+				atomic_store_p(&dss_max, dss_next,
+				    ATOMIC_RELEASE);
+				extent_dss_extending_finish();
+
+				if (gap_size_page != 0) {
+					extent_dalloc_gap(tsdn, arena, gap);
+				} else {
+					extent_dalloc(tsdn, arena, gap);
+				}
+				if (!*commit) {
+					*commit = pages_decommit(ret, size);
+				}
+				if (*zero && *commit) {
+					extent_hooks_t *extent_hooks =
+					    EXTENT_HOOKS_INITIALIZER;
+					extent_t extent;
+
+					extent_init(&extent, arena, ret, size,
+					    size, false, SC_NSIZES,
+					    extent_state_active, false, true,
+					    true, EXTENT_NOT_HEAD);
+					if (extent_purge_forced_wrapper(tsdn,
+					    arena, &extent_hooks, &extent, 0,
+					    size)) {
+						memset(ret, 0, size);
+					}
+				}
+				return ret;
+			}
+			/*
+			 * Failure, whether due to OOM or a race with a raw
+			 * sbrk() call from outside the allocator.
+			 */
+			if (dss_prev == (void *)-1) {
+				/* OOM. */
+				atomic_store_b(&dss_exhausted, true,
+				    ATOMIC_RELEASE);
+				goto label_oom;
+			}
+		}
+	}
+label_oom:
+	extent_dss_extending_finish();
+	extent_dalloc(tsdn, arena, gap);
+	return NULL;
+}
+
+static bool
+extent_in_dss_helper(void *addr, void *max) {
+	return ((uintptr_t)addr >= (uintptr_t)dss_base && (uintptr_t)addr <
+	    (uintptr_t)max);
+}
+
+bool
+extent_in_dss(void *addr) {
+	cassert(have_dss);
+
+	return extent_in_dss_helper(addr, atomic_load_p(&dss_max,
+	    ATOMIC_ACQUIRE));
+}
+
+bool
+extent_dss_mergeable(void *addr_a, void *addr_b) {
+	void *max;
+
+	cassert(have_dss);
+
+	if ((uintptr_t)addr_a < (uintptr_t)dss_base && (uintptr_t)addr_b <
+	    (uintptr_t)dss_base) {
+		return true;
+	}
+
+	max = atomic_load_p(&dss_max, ATOMIC_ACQUIRE);
+	return (extent_in_dss_helper(addr_a, max) ==
+	    extent_in_dss_helper(addr_b, max));
+}
+
+void
+extent_dss_boot(void) {
+	cassert(have_dss);
+
+	dss_base = extent_dss_sbrk(0);
+	atomic_store_b(&dss_extending, false, ATOMIC_RELAXED);
+	atomic_store_b(&dss_exhausted, dss_base == (void *)-1, ATOMIC_RELAXED);
+	atomic_store_p(&dss_max, dss_base, ATOMIC_RELAXED);
+}
+
+/******************************************************************************/
diff --git a/memkind-1.10.0/jemalloc/src/extent_mmap.c b/memkind-1.10.0/jemalloc/src/extent_mmap.c
new file mode 100644
index 0000000..17fd1c8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/extent_mmap.c
@@ -0,0 +1,42 @@
+#define JEMALLOC_EXTENT_MMAP_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_mmap.h"
+
+/******************************************************************************/
+/* Data. */
+
+bool	opt_retain =
+#ifdef JEMALLOC_RETAIN
+    true
+#else
+    false
+#endif
+    ;
+
+/******************************************************************************/
+
+void *
+extent_alloc_mmap(void *new_addr, size_t size, size_t alignment, bool *zero,
+    bool *commit) {
+	assert(alignment == ALIGNMENT_CEILING(alignment, PAGE));
+	void *ret = pages_map(new_addr, size, alignment, commit);
+	if (ret == NULL) {
+		return NULL;
+	}
+	assert(ret != NULL);
+	if (*commit) {
+		*zero = true;
+	}
+	return ret;
+}
+
+bool
+extent_dalloc_mmap(void *addr, size_t size) {
+	if (!opt_retain) {
+		pages_unmap(addr, size);
+	}
+	return opt_retain;
+}
diff --git a/memkind-1.10.0/jemalloc/src/hash.c b/memkind-1.10.0/jemalloc/src/hash.c
new file mode 100644
index 0000000..7b2bdc2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/hash.c
@@ -0,0 +1,3 @@
+#define JEMALLOC_HASH_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
diff --git a/memkind-1.10.0/jemalloc/src/hook.c b/memkind-1.10.0/jemalloc/src/hook.c
new file mode 100644
index 0000000..9ac703c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/hook.c
@@ -0,0 +1,195 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/hook.h"
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/seq.h"
+
+typedef struct hooks_internal_s hooks_internal_t;
+struct hooks_internal_s {
+	hooks_t hooks;
+	bool in_use;
+};
+
+seq_define(hooks_internal_t, hooks)
+
+static atomic_u_t nhooks = ATOMIC_INIT(0);
+static seq_hooks_t hooks[HOOK_MAX];
+static malloc_mutex_t hooks_mu;
+
+bool
+hook_boot() {
+	return malloc_mutex_init(&hooks_mu, "hooks", WITNESS_RANK_HOOK,
+	    malloc_mutex_rank_exclusive);
+}
+
+static void *
+hook_install_locked(hooks_t *to_install) {
+	hooks_internal_t hooks_internal;
+	for (int i = 0; i < HOOK_MAX; i++) {
+		bool success = seq_try_load_hooks(&hooks_internal, &hooks[i]);
+		/* We hold mu; no concurrent access. */
+		assert(success);
+		if (!hooks_internal.in_use) {
+			hooks_internal.hooks = *to_install;
+			hooks_internal.in_use = true;
+			seq_store_hooks(&hooks[i], &hooks_internal);
+			atomic_store_u(&nhooks,
+			    atomic_load_u(&nhooks, ATOMIC_RELAXED) + 1,
+			    ATOMIC_RELAXED);
+			return &hooks[i];
+		}
+	}
+	return NULL;
+}
+
+void *
+hook_install(tsdn_t *tsdn, hooks_t *to_install) {
+	malloc_mutex_lock(tsdn, &hooks_mu);
+	void *ret = hook_install_locked(to_install);
+	if (ret != NULL) {
+		tsd_global_slow_inc(tsdn);
+	}
+	malloc_mutex_unlock(tsdn, &hooks_mu);
+	return ret;
+}
+
+static void
+hook_remove_locked(seq_hooks_t *to_remove) {
+	hooks_internal_t hooks_internal;
+	bool success = seq_try_load_hooks(&hooks_internal, to_remove);
+	/* We hold mu; no concurrent access. */
+	assert(success);
+	/* Should only remove hooks that were added. */
+	assert(hooks_internal.in_use);
+	hooks_internal.in_use = false;
+	seq_store_hooks(to_remove, &hooks_internal);
+	atomic_store_u(&nhooks, atomic_load_u(&nhooks, ATOMIC_RELAXED) - 1,
+	    ATOMIC_RELAXED);
+}
+
+void
+hook_remove(tsdn_t *tsdn, void *opaque) {
+	if (config_debug) {
+		char *hooks_begin = (char *)&hooks[0];
+		char *hooks_end = (char *)&hooks[HOOK_MAX];
+		char *hook = (char *)opaque;
+		assert(hooks_begin <= hook && hook < hooks_end
+		    && (hook - hooks_begin) % sizeof(seq_hooks_t) == 0);
+	}
+	malloc_mutex_lock(tsdn, &hooks_mu);
+	hook_remove_locked((seq_hooks_t *)opaque);
+	tsd_global_slow_dec(tsdn);
+	malloc_mutex_unlock(tsdn, &hooks_mu);
+}
+
+#define FOR_EACH_HOOK_BEGIN(hooks_internal_ptr)				\
+for (int for_each_hook_counter = 0;					\
+    for_each_hook_counter < HOOK_MAX;					\
+    for_each_hook_counter++) {						\
+	bool for_each_hook_success = seq_try_load_hooks(		\
+	    (hooks_internal_ptr), &hooks[for_each_hook_counter]);	\
+	if (!for_each_hook_success) {					\
+		continue;						\
+	}								\
+	if (!(hooks_internal_ptr)->in_use) {				\
+		continue;						\
+	}
+#define FOR_EACH_HOOK_END						\
+}
+
+static bool *
+hook_reentrantp() {
+	/*
+	 * We prevent user reentrancy within hooks.  This is basically just a
+	 * thread-local bool that triggers an early-exit.
+	 *
+	 * We don't fold in_hook into reentrancy.  There are two reasons for
+	 * this:
+	 * - Right now, we turn on reentrancy during things like extent hook
+	 *   execution.  Allocating during extent hooks is not officially
+	 *   supported, but we don't want to break it for the time being.  These
+	 *   sorts of allocations should probably still be hooked, though.
+	 * - If a hook allocates, we may want it to be relatively fast (after
+	 *   all, it executes on every allocator operation).  Turning on
+	 *   reentrancy is a fairly heavyweight mode (disabling tcache,
+	 *   redirecting to arena 0, etc.).  It's possible we may one day want
+	 *   to turn on reentrant mode here, if it proves too difficult to keep
+	 *   this working.  But that's fairly easy for us to see; OTOH, people
+	 *   not using hooks because they're too slow is easy for us to miss.
+	 *
+	 * The tricky part is
+	 * that this code might get invoked even if we don't have access to tsd.
+	 * This function mimics getting a pointer to thread-local data, except
+	 * that it might secretly return a pointer to some global data if we
+	 * know that the caller will take the early-exit path.
+	 * If we return a bool that indicates that we are reentrant, then the
+	 * caller will go down the early exit path, leaving the global
+	 * untouched.
+	 */
+	static bool in_hook_global = true;
+	tsdn_t *tsdn = tsdn_fetch();
+	tcache_t *tcache = tsdn_tcachep_get(tsdn);
+	if (tcache != NULL) {
+		return &tcache->in_hook;
+	}
+	return &in_hook_global;
+}
+
+#define HOOK_PROLOGUE							\
+	if (likely(atomic_load_u(&nhooks, ATOMIC_RELAXED) == 0)) {	\
+		return;							\
+	}								\
+	bool *in_hook = hook_reentrantp();				\
+	if (*in_hook) {							\
+		return;							\
+	}								\
+	*in_hook = true;
+
+#define HOOK_EPILOGUE							\
+	*in_hook = false;
+
+void
+hook_invoke_alloc(hook_alloc_t type, void *result, uintptr_t result_raw,
+    uintptr_t args_raw[3]) {
+	HOOK_PROLOGUE
+
+	hooks_internal_t hook;
+	FOR_EACH_HOOK_BEGIN(&hook)
+		hook_alloc h = hook.hooks.alloc_hook;
+		if (h != NULL) {
+			h(hook.hooks.extra, type, result, result_raw, args_raw);
+		}
+	FOR_EACH_HOOK_END
+
+	HOOK_EPILOGUE
+}
+
+void
+hook_invoke_dalloc(hook_dalloc_t type, void *address, uintptr_t args_raw[3]) {
+	HOOK_PROLOGUE
+	hooks_internal_t hook;
+	FOR_EACH_HOOK_BEGIN(&hook)
+		hook_dalloc h = hook.hooks.dalloc_hook;
+		if (h != NULL) {
+			h(hook.hooks.extra, type, address, args_raw);
+		}
+	FOR_EACH_HOOK_END
+	HOOK_EPILOGUE
+}
+
+void
+hook_invoke_expand(hook_expand_t type, void *address, size_t old_usize,
+    size_t new_usize, uintptr_t result_raw, uintptr_t args_raw[4]) {
+	HOOK_PROLOGUE
+	hooks_internal_t hook;
+	FOR_EACH_HOOK_BEGIN(&hook)
+		hook_expand h = hook.hooks.expand_hook;
+		if (h != NULL) {
+			h(hook.hooks.extra, type, address, old_usize, new_usize,
+			    result_raw, args_raw);
+		}
+	FOR_EACH_HOOK_END
+	HOOK_EPILOGUE
+}
diff --git a/memkind-1.10.0/jemalloc/src/jemalloc.c b/memkind-1.10.0/jemalloc/src/jemalloc.c
new file mode 100644
index 0000000..ed13718
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/jemalloc.c
@@ -0,0 +1,3922 @@
+#define JEMALLOC_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/log.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/sc.h"
+#include "jemalloc/internal/spin.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/ticker.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+/* Data. */
+
+/* Runtime configuration options. */
+const char	*je_malloc_conf
+#ifndef _WIN32
+    JEMALLOC_ATTR(weak)
+#endif
+    ;
+bool	opt_abort =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+bool	opt_abort_conf =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+/* Intentionally default off, even with debug builds. */
+bool	opt_confirm_conf = false;
+const char	*opt_junk =
+#if (defined(JEMALLOC_DEBUG) && defined(JEMALLOC_FILL))
+    "true"
+#else
+    "false"
+#endif
+    ;
+bool	opt_junk_alloc =
+#if (defined(JEMALLOC_DEBUG) && defined(JEMALLOC_FILL))
+    true
+#else
+    false
+#endif
+    ;
+bool	opt_junk_free =
+#if (defined(JEMALLOC_DEBUG) && defined(JEMALLOC_FILL))
+    true
+#else
+    false
+#endif
+    ;
+
+bool	opt_utrace = false;
+bool	opt_xmalloc = false;
+bool	opt_zero = false;
+unsigned	opt_narenas = 0;
+
+unsigned	ncpus;
+
+/* Protects arenas initialization. */
+malloc_mutex_t arenas_lock;
+/*
+ * Arenas that are used to service external requests.  Not all elements of the
+ * arenas array are necessarily used; arenas are created lazily as needed.
+ *
+ * arenas[0..narenas_auto) are used for automatic multiplexing of threads and
+ * arenas.  arenas[narenas_auto..narenas_total) are only used if the application
+ * takes some action to create them and allocate from them.
+ *
+ * Points to an arena_t.
+ */
+JEMALLOC_ALIGNED(CACHELINE)
+atomic_p_t		arenas[MALLOCX_ARENA_LIMIT];
+static atomic_u_t	narenas_total; /* Use narenas_total_*(). */
+/* Below three are read-only after initialization. */
+static arena_t		*a0; /* arenas[0]. */
+unsigned		narenas_auto;
+unsigned		manual_arena_base;
+
+typedef enum {
+	malloc_init_uninitialized	= 3,
+	malloc_init_a0_initialized	= 2,
+	malloc_init_recursible		= 1,
+	malloc_init_initialized		= 0 /* Common case --> jnz. */
+} malloc_init_t;
+static malloc_init_t	malloc_init_state = malloc_init_uninitialized;
+
+/* False should be the common case.  Set to true to trigger initialization. */
+bool			malloc_slow = true;
+
+/* When malloc_slow is true, set the corresponding bits for sanity check. */
+enum {
+	flag_opt_junk_alloc	= (1U),
+	flag_opt_junk_free	= (1U << 1),
+	flag_opt_zero		= (1U << 2),
+	flag_opt_utrace		= (1U << 3),
+	flag_opt_xmalloc	= (1U << 4)
+};
+static uint8_t	malloc_slow_flags;
+
+#ifdef JEMALLOC_THREADED_INIT
+/* Used to let the initializing thread recursively allocate. */
+#  define NO_INITIALIZER	((unsigned long)0)
+#  define INITIALIZER		pthread_self()
+#  define IS_INITIALIZER	(malloc_initializer == pthread_self())
+static pthread_t		malloc_initializer = NO_INITIALIZER;
+#else
+#  define NO_INITIALIZER	false
+#  define INITIALIZER		true
+#  define IS_INITIALIZER	malloc_initializer
+static bool			malloc_initializer = NO_INITIALIZER;
+#endif
+
+/* Used to avoid initialization races. */
+#ifdef _WIN32
+#if _WIN32_WINNT >= 0x0600
+static malloc_mutex_t	init_lock = SRWLOCK_INIT;
+#else
+static malloc_mutex_t	init_lock;
+static bool init_lock_initialized = false;
+
+JEMALLOC_ATTR(constructor)
+static void WINAPI
+_init_init_lock(void) {
+	/*
+	 * If another constructor in the same binary is using mallctl to e.g.
+	 * set up extent hooks, it may end up running before this one, and
+	 * malloc_init_hard will crash trying to lock the uninitialized lock. So
+	 * we force an initialization of the lock in malloc_init_hard as well.
+	 * We don't try to care about atomicity of the accessed to the
+	 * init_lock_initialized boolean, since it really only matters early in
+	 * the process creation, before any separate thread normally starts
+	 * doing anything.
+	 */
+	if (!init_lock_initialized) {
+		malloc_mutex_init(&init_lock, "init", WITNESS_RANK_INIT,
+		    malloc_mutex_rank_exclusive);
+	}
+	init_lock_initialized = true;
+}
+
+#ifdef _MSC_VER
+#  pragma section(".CRT$XCU", read)
+JEMALLOC_SECTION(".CRT$XCU") JEMALLOC_ATTR(used)
+static const void (WINAPI *init_init_lock)(void) = _init_init_lock;
+#endif
+#endif
+#else
+static malloc_mutex_t	init_lock = MALLOC_MUTEX_INITIALIZER;
+#endif
+
+typedef struct {
+	void	*p;	/* Input pointer (as in realloc(p, s)). */
+	size_t	s;	/* Request size. */
+	void	*r;	/* Result pointer. */
+} malloc_utrace_t;
+
+#ifdef JEMALLOC_UTRACE
+#  define UTRACE(a, b, c) do {						\
+	if (unlikely(opt_utrace)) {					\
+		int utrace_serrno = errno;				\
+		malloc_utrace_t ut;					\
+		ut.p = (a);						\
+		ut.s = (b);						\
+		ut.r = (c);						\
+		utrace(&ut, sizeof(ut));				\
+		errno = utrace_serrno;					\
+	}								\
+} while (0)
+#else
+#  define UTRACE(a, b, c)
+#endif
+
+/* Whether encountered any invalid config options. */
+static bool had_conf_error = false;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static bool	malloc_init_hard_a0(void);
+static bool	malloc_init_hard(void);
+
+/******************************************************************************/
+/*
+ * Begin miscellaneous support functions.
+ */
+
+bool
+malloc_initialized(void) {
+	return (malloc_init_state == malloc_init_initialized);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+malloc_init_a0(void) {
+	if (unlikely(malloc_init_state == malloc_init_uninitialized)) {
+		return malloc_init_hard_a0();
+	}
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+malloc_init(void) {
+	if (unlikely(!malloc_initialized()) && malloc_init_hard()) {
+		return true;
+	}
+	return false;
+}
+
+/*
+ * The a0*() functions are used instead of i{d,}alloc() in situations that
+ * cannot tolerate TLS variable access.
+ */
+
+static void *
+a0ialloc(size_t size, bool zero, bool is_internal) {
+	if (unlikely(malloc_init_a0())) {
+		return NULL;
+	}
+
+	return iallocztm(TSDN_NULL, size, sz_size2index(size), zero, NULL,
+	    is_internal, arena_get(TSDN_NULL, 0, true), true);
+}
+
+static void
+a0idalloc(void *ptr, bool is_internal) {
+	idalloctm(TSDN_NULL, ptr, NULL, NULL, is_internal, true);
+}
+
+void *
+a0malloc(size_t size) {
+	return a0ialloc(size, false, true);
+}
+
+void
+a0dalloc(void *ptr) {
+	a0idalloc(ptr, true);
+}
+
+/*
+ * FreeBSD's libc uses the bootstrap_*() functions in bootstrap-senstive
+ * situations that cannot tolerate TLS variable access (TLS allocation and very
+ * early internal data structure initialization).
+ */
+
+void *
+bootstrap_malloc(size_t size) {
+	if (unlikely(size == 0)) {
+		size = 1;
+	}
+
+	return a0ialloc(size, false, false);
+}
+
+void *
+bootstrap_calloc(size_t num, size_t size) {
+	size_t num_size;
+
+	num_size = num * size;
+	if (unlikely(num_size == 0)) {
+		assert(num == 0 || size == 0);
+		num_size = 1;
+	}
+
+	return a0ialloc(num_size, true, false);
+}
+
+void
+bootstrap_free(void *ptr) {
+	if (unlikely(ptr == NULL)) {
+		return;
+	}
+
+	a0idalloc(ptr, false);
+}
+
+void
+arena_set(unsigned ind, arena_t *arena) {
+	atomic_store_p(&arenas[ind], arena, ATOMIC_RELEASE);
+}
+
+static void
+narenas_total_set(unsigned narenas) {
+	atomic_store_u(&narenas_total, narenas, ATOMIC_RELEASE);
+}
+
+static void
+narenas_total_inc(void) {
+	atomic_fetch_add_u(&narenas_total, 1, ATOMIC_RELEASE);
+}
+
+unsigned
+narenas_total_get(void) {
+	return atomic_load_u(&narenas_total, ATOMIC_ACQUIRE);
+}
+
+/* Create a new arena and insert it into the arenas array at index ind. */
+static arena_t *
+arena_init_locked(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	arena_t *arena;
+
+	assert(ind <= narenas_total_get());
+	if (ind >= MALLOCX_ARENA_LIMIT) {
+		return NULL;
+	}
+	if (ind == narenas_total_get()) {
+		narenas_total_inc();
+	}
+
+	/*
+	 * Another thread may have already initialized arenas[ind] if it's an
+	 * auto arena.
+	 */
+	arena = arena_get(tsdn, ind, false);
+	if (arena != NULL) {
+		assert(arena_is_auto(arena));
+		return arena;
+	}
+
+	/* Actually initialize the arena. */
+	arena = arena_new(tsdn, ind, extent_hooks);
+
+	return arena;
+}
+
+static void
+arena_new_create_background_thread(tsdn_t *tsdn, unsigned ind) {
+	if (ind == 0) {
+		return;
+	}
+	/*
+	 * Avoid creating a new background thread just for the huge arena, which
+	 * purges eagerly by default.
+	 */
+	if (have_background_thread && !arena_is_huge(ind)) {
+		if (background_thread_create(tsdn_tsd(tsdn), ind)) {
+			malloc_printf("<jemalloc>: error in background thread "
+				      "creation for arena %u. Abort.\n", ind);
+			abort();
+		}
+	}
+}
+
+arena_t *
+arena_init(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks) {
+	arena_t *arena;
+
+	malloc_mutex_lock(tsdn, &arenas_lock);
+	arena = arena_init_locked(tsdn, ind, extent_hooks);
+	malloc_mutex_unlock(tsdn, &arenas_lock);
+
+	arena_new_create_background_thread(tsdn, ind);
+
+	return arena;
+}
+
+static void
+arena_bind(tsd_t *tsd, unsigned ind, bool internal) {
+	arena_t *arena = arena_get(tsd_tsdn(tsd), ind, false);
+	arena_nthreads_inc(arena, internal);
+
+	if (internal) {
+		tsd_iarena_set(tsd, arena);
+	} else {
+		tsd_arena_set(tsd, arena);
+		unsigned shard = atomic_fetch_add_u(&arena->binshard_next, 1,
+		    ATOMIC_RELAXED);
+		tsd_binshards_t *bins = tsd_binshardsp_get(tsd);
+		for (unsigned i = 0; i < SC_NBINS; i++) {
+			assert(bin_infos[i].n_shards > 0 &&
+			    bin_infos[i].n_shards <= BIN_SHARDS_MAX);
+			bins->binshard[i] = shard % bin_infos[i].n_shards;
+		}
+	}
+}
+
+void
+arena_migrate(tsd_t *tsd, unsigned oldind, unsigned newind) {
+	arena_t *oldarena, *newarena;
+
+	oldarena = arena_get(tsd_tsdn(tsd), oldind, false);
+	newarena = arena_get(tsd_tsdn(tsd), newind, false);
+	arena_nthreads_dec(oldarena, false);
+	arena_nthreads_inc(newarena, false);
+	tsd_arena_set(tsd, newarena);
+}
+
+static void
+arena_unbind(tsd_t *tsd, unsigned ind, bool internal) {
+	arena_t *arena;
+
+	arena = arena_get(tsd_tsdn(tsd), ind, false);
+	arena_nthreads_dec(arena, internal);
+
+	if (internal) {
+		tsd_iarena_set(tsd, NULL);
+	} else {
+		tsd_arena_set(tsd, NULL);
+	}
+}
+
+arena_tdata_t *
+arena_tdata_get_hard(tsd_t *tsd, unsigned ind) {
+	arena_tdata_t *tdata, *arenas_tdata_old;
+	arena_tdata_t *arenas_tdata = tsd_arenas_tdata_get(tsd);
+	unsigned narenas_tdata_old, i;
+	unsigned narenas_tdata = tsd_narenas_tdata_get(tsd);
+	unsigned narenas_actual = narenas_total_get();
+
+	/*
+	 * Dissociate old tdata array (and set up for deallocation upon return)
+	 * if it's too small.
+	 */
+	if (arenas_tdata != NULL && narenas_tdata < narenas_actual) {
+		arenas_tdata_old = arenas_tdata;
+		narenas_tdata_old = narenas_tdata;
+		arenas_tdata = NULL;
+		narenas_tdata = 0;
+		tsd_arenas_tdata_set(tsd, arenas_tdata);
+		tsd_narenas_tdata_set(tsd, narenas_tdata);
+	} else {
+		arenas_tdata_old = NULL;
+		narenas_tdata_old = 0;
+	}
+
+	/* Allocate tdata array if it's missing. */
+	if (arenas_tdata == NULL) {
+		bool *arenas_tdata_bypassp = tsd_arenas_tdata_bypassp_get(tsd);
+		narenas_tdata = (ind < narenas_actual) ? narenas_actual : ind+1;
+
+		if (tsd_nominal(tsd) && !*arenas_tdata_bypassp) {
+			*arenas_tdata_bypassp = true;
+			arenas_tdata = (arena_tdata_t *)a0malloc(
+			    sizeof(arena_tdata_t) * narenas_tdata);
+			*arenas_tdata_bypassp = false;
+		}
+		if (arenas_tdata == NULL) {
+			tdata = NULL;
+			goto label_return;
+		}
+		assert(tsd_nominal(tsd) && !*arenas_tdata_bypassp);
+		tsd_arenas_tdata_set(tsd, arenas_tdata);
+		tsd_narenas_tdata_set(tsd, narenas_tdata);
+	}
+
+	/*
+	 * Copy to tdata array.  It's possible that the actual number of arenas
+	 * has increased since narenas_total_get() was called above, but that
+	 * causes no correctness issues unless two threads concurrently execute
+	 * the arenas.create mallctl, which we trust mallctl synchronization to
+	 * prevent.
+	 */
+
+	/* Copy/initialize tickers. */
+	for (i = 0; i < narenas_actual; i++) {
+		if (i < narenas_tdata_old) {
+			ticker_copy(&arenas_tdata[i].decay_ticker,
+			    &arenas_tdata_old[i].decay_ticker);
+		} else {
+			ticker_init(&arenas_tdata[i].decay_ticker,
+			    DECAY_NTICKS_PER_UPDATE);
+		}
+	}
+	if (narenas_tdata > narenas_actual) {
+		memset(&arenas_tdata[narenas_actual], 0, sizeof(arena_tdata_t)
+		    * (narenas_tdata - narenas_actual));
+	}
+
+	/* Read the refreshed tdata array. */
+	tdata = &arenas_tdata[ind];
+label_return:
+	if (arenas_tdata_old != NULL) {
+		a0dalloc(arenas_tdata_old);
+	}
+	return tdata;
+}
+
+/* Slow path, called only by arena_choose(). */
+arena_t *
+arena_choose_hard(tsd_t *tsd, bool internal) {
+	arena_t *ret JEMALLOC_CC_SILENCE_INIT(NULL);
+
+	if (have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena)) {
+		unsigned choose = percpu_arena_choose();
+		ret = arena_get(tsd_tsdn(tsd), choose, true);
+		assert(ret != NULL);
+		arena_bind(tsd, arena_ind_get(ret), false);
+		arena_bind(tsd, arena_ind_get(ret), true);
+
+		return ret;
+	}
+
+	if (narenas_auto > 1) {
+		unsigned i, j, choose[2], first_null;
+		bool is_new_arena[2];
+
+		/*
+		 * Determine binding for both non-internal and internal
+		 * allocation.
+		 *
+		 *   choose[0]: For application allocation.
+		 *   choose[1]: For internal metadata allocation.
+		 */
+
+		for (j = 0; j < 2; j++) {
+			choose[j] = 0;
+			is_new_arena[j] = false;
+		}
+
+		first_null = narenas_auto;
+		malloc_mutex_lock(tsd_tsdn(tsd), &arenas_lock);
+		assert(arena_get(tsd_tsdn(tsd), 0, false) != NULL);
+		for (i = 1; i < narenas_auto; i++) {
+			if (arena_get(tsd_tsdn(tsd), i, false) != NULL) {
+				/*
+				 * Choose the first arena that has the lowest
+				 * number of threads assigned to it.
+				 */
+				for (j = 0; j < 2; j++) {
+					if (arena_nthreads_get(arena_get(
+					    tsd_tsdn(tsd), i, false), !!j) <
+					    arena_nthreads_get(arena_get(
+					    tsd_tsdn(tsd), choose[j], false),
+					    !!j)) {
+						choose[j] = i;
+					}
+				}
+			} else if (first_null == narenas_auto) {
+				/*
+				 * Record the index of the first uninitialized
+				 * arena, in case all extant arenas are in use.
+				 *
+				 * NB: It is possible for there to be
+				 * discontinuities in terms of initialized
+				 * versus uninitialized arenas, due to the
+				 * "thread.arena" mallctl.
+				 */
+				first_null = i;
+			}
+		}
+
+		for (j = 0; j < 2; j++) {
+			if (arena_nthreads_get(arena_get(tsd_tsdn(tsd),
+			    choose[j], false), !!j) == 0 || first_null ==
+			    narenas_auto) {
+				/*
+				 * Use an unloaded arena, or the least loaded
+				 * arena if all arenas are already initialized.
+				 */
+				if (!!j == internal) {
+					ret = arena_get(tsd_tsdn(tsd),
+					    choose[j], false);
+				}
+			} else {
+				arena_t *arena;
+
+				/* Initialize a new arena. */
+				choose[j] = first_null;
+				arena = arena_init_locked(tsd_tsdn(tsd),
+				    choose[j],
+				    (extent_hooks_t *)&extent_hooks_default);
+				if (arena == NULL) {
+					malloc_mutex_unlock(tsd_tsdn(tsd),
+					    &arenas_lock);
+					return NULL;
+				}
+				is_new_arena[j] = true;
+				if (!!j == internal) {
+					ret = arena;
+				}
+			}
+			arena_bind(tsd, choose[j], !!j);
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &arenas_lock);
+
+		for (j = 0; j < 2; j++) {
+			if (is_new_arena[j]) {
+				assert(choose[j] > 0);
+				arena_new_create_background_thread(
+				    tsd_tsdn(tsd), choose[j]);
+			}
+		}
+
+	} else {
+		ret = arena_get(tsd_tsdn(tsd), 0, false);
+		arena_bind(tsd, 0, false);
+		arena_bind(tsd, 0, true);
+	}
+
+	return ret;
+}
+
+void
+iarena_cleanup(tsd_t *tsd) {
+	arena_t *iarena;
+
+	iarena = tsd_iarena_get(tsd);
+	if (iarena != NULL) {
+		arena_unbind(tsd, arena_ind_get(iarena), true);
+	}
+}
+
+void
+arena_cleanup(tsd_t *tsd) {
+	arena_t *arena;
+
+	arena = tsd_arena_get(tsd);
+	if (arena != NULL) {
+		arena_unbind(tsd, arena_ind_get(arena), false);
+	}
+}
+
+void
+arenas_tdata_cleanup(tsd_t *tsd) {
+	arena_tdata_t *arenas_tdata;
+
+	/* Prevent tsd->arenas_tdata from being (re)created. */
+	*tsd_arenas_tdata_bypassp_get(tsd) = true;
+
+	arenas_tdata = tsd_arenas_tdata_get(tsd);
+	if (arenas_tdata != NULL) {
+		tsd_arenas_tdata_set(tsd, NULL);
+		a0dalloc(arenas_tdata);
+	}
+}
+
+static void
+stats_print_atexit(void) {
+	if (config_stats) {
+		tsdn_t *tsdn;
+		unsigned narenas, i;
+
+		tsdn = tsdn_fetch();
+
+		/*
+		 * Merge stats from extant threads.  This is racy, since
+		 * individual threads do not lock when recording tcache stats
+		 * events.  As a consequence, the final stats may be slightly
+		 * out of date by the time they are reported, if other threads
+		 * continue to allocate.
+		 */
+		for (i = 0, narenas = narenas_total_get(); i < narenas; i++) {
+			arena_t *arena = arena_get(tsdn, i, false);
+			if (arena != NULL) {
+				tcache_t *tcache;
+
+				malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+				ql_foreach(tcache, &arena->tcache_ql, link) {
+					tcache_stats_merge(tsdn, tcache, arena);
+				}
+				malloc_mutex_unlock(tsdn,
+				    &arena->tcache_ql_mtx);
+			}
+		}
+	}
+	je_malloc_stats_print(NULL, NULL, opt_stats_print_opts);
+}
+
+/*
+ * Ensure that we don't hold any locks upon entry to or exit from allocator
+ * code (in a "broad" sense that doesn't count a reentrant allocation as an
+ * entrance or exit).
+ */
+JEMALLOC_ALWAYS_INLINE void
+check_entry_exit_locking(tsdn_t *tsdn) {
+	if (!config_debug) {
+		return;
+	}
+	if (tsdn_null(tsdn)) {
+		return;
+	}
+	tsd_t *tsd = tsdn_tsd(tsdn);
+	/*
+	 * It's possible we hold locks at entry/exit if we're in a nested
+	 * allocation.
+	 */
+	int8_t reentrancy_level = tsd_reentrancy_level_get(tsd);
+	if (reentrancy_level != 0) {
+		return;
+	}
+	witness_assert_lockless(tsdn_witness_tsdp_get(tsdn));
+}
+
+/*
+ * End miscellaneous support functions.
+ */
+/******************************************************************************/
+/*
+ * Begin initialization functions.
+ */
+
+static char *
+jemalloc_secure_getenv(const char *name) {
+#ifdef JEMALLOC_HAVE_SECURE_GETENV
+	return secure_getenv(name);
+#else
+#  ifdef JEMALLOC_HAVE_ISSETUGID
+	if (issetugid() != 0) {
+		return NULL;
+	}
+#  endif
+	return getenv(name);
+#endif
+}
+
+static unsigned
+malloc_ncpus(void) {
+	long result;
+
+#ifdef _WIN32
+	SYSTEM_INFO si;
+	GetSystemInfo(&si);
+	result = si.dwNumberOfProcessors;
+#elif defined(JEMALLOC_GLIBC_MALLOC_HOOK) && defined(CPU_COUNT)
+	/*
+	 * glibc >= 2.6 has the CPU_COUNT macro.
+	 *
+	 * glibc's sysconf() uses isspace().  glibc allocates for the first time
+	 * *before* setting up the isspace tables.  Therefore we need a
+	 * different method to get the number of CPUs.
+	 */
+	{
+		cpu_set_t set;
+
+		pthread_getaffinity_np(pthread_self(), sizeof(set), &set);
+		result = CPU_COUNT(&set);
+	}
+#else
+	result = sysconf(_SC_NPROCESSORS_ONLN);
+#endif
+	return ((result == -1) ? 1 : (unsigned)result);
+}
+
+static void
+init_opt_stats_print_opts(const char *v, size_t vlen) {
+	size_t opts_len = strlen(opt_stats_print_opts);
+	assert(opts_len <= stats_print_tot_num_options);
+
+	for (size_t i = 0; i < vlen; i++) {
+		switch (v[i]) {
+#define OPTION(o, v, d, s) case o: break;
+			STATS_PRINT_OPTIONS
+#undef OPTION
+		default: continue;
+		}
+
+		if (strchr(opt_stats_print_opts, v[i]) != NULL) {
+			/* Ignore repeated. */
+			continue;
+		}
+
+		opt_stats_print_opts[opts_len++] = v[i];
+		opt_stats_print_opts[opts_len] = '\0';
+		assert(opts_len <= stats_print_tot_num_options);
+	}
+	assert(opts_len == strlen(opt_stats_print_opts));
+}
+
+/* Reads the next size pair in a multi-sized option. */
+static bool
+malloc_conf_multi_sizes_next(const char **slab_size_segment_cur,
+    size_t *vlen_left, size_t *slab_start, size_t *slab_end, size_t *new_size) {
+	const char *cur = *slab_size_segment_cur;
+	char *end;
+	uintmax_t um;
+
+	set_errno(0);
+
+	/* First number, then '-' */
+	um = malloc_strtoumax(cur, &end, 0);
+	if (get_errno() != 0 || *end != '-') {
+		return true;
+	}
+	*slab_start = (size_t)um;
+	cur = end + 1;
+
+	/* Second number, then ':' */
+	um = malloc_strtoumax(cur, &end, 0);
+	if (get_errno() != 0 || *end != ':') {
+		return true;
+	}
+	*slab_end = (size_t)um;
+	cur = end + 1;
+
+	/* Last number */
+	um = malloc_strtoumax(cur, &end, 0);
+	if (get_errno() != 0) {
+		return true;
+	}
+	*new_size = (size_t)um;
+
+	/* Consume the separator if there is one. */
+	if (*end == '|') {
+		end++;
+	}
+
+	*vlen_left -= end - *slab_size_segment_cur;
+	*slab_size_segment_cur = end;
+
+	return false;
+}
+
+static bool
+malloc_conf_next(char const **opts_p, char const **k_p, size_t *klen_p,
+    char const **v_p, size_t *vlen_p) {
+	bool accept;
+	const char *opts = *opts_p;
+
+	*k_p = opts;
+
+	for (accept = false; !accept;) {
+		switch (*opts) {
+		case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
+		case 'G': case 'H': case 'I': case 'J': case 'K': case 'L':
+		case 'M': case 'N': case 'O': case 'P': case 'Q': case 'R':
+		case 'S': case 'T': case 'U': case 'V': case 'W': case 'X':
+		case 'Y': case 'Z':
+		case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
+		case 'g': case 'h': case 'i': case 'j': case 'k': case 'l':
+		case 'm': case 'n': case 'o': case 'p': case 'q': case 'r':
+		case 's': case 't': case 'u': case 'v': case 'w': case 'x':
+		case 'y': case 'z':
+		case '0': case '1': case '2': case '3': case '4': case '5':
+		case '6': case '7': case '8': case '9':
+		case '_':
+			opts++;
+			break;
+		case ':':
+			opts++;
+			*klen_p = (uintptr_t)opts - 1 - (uintptr_t)*k_p;
+			*v_p = opts;
+			accept = true;
+			break;
+		case '\0':
+			if (opts != *opts_p) {
+				malloc_write("<jemalloc>: Conf string ends "
+				    "with key\n");
+			}
+			return true;
+		default:
+			malloc_write("<jemalloc>: Malformed conf string\n");
+			return true;
+		}
+	}
+
+	for (accept = false; !accept;) {
+		switch (*opts) {
+		case ',':
+			opts++;
+			/*
+			 * Look ahead one character here, because the next time
+			 * this function is called, it will assume that end of
+			 * input has been cleanly reached if no input remains,
+			 * but we have optimistically already consumed the
+			 * comma if one exists.
+			 */
+			if (*opts == '\0') {
+				malloc_write("<jemalloc>: Conf string ends "
+				    "with comma\n");
+			}
+			*vlen_p = (uintptr_t)opts - 1 - (uintptr_t)*v_p;
+			accept = true;
+			break;
+		case '\0':
+			*vlen_p = (uintptr_t)opts - (uintptr_t)*v_p;
+			accept = true;
+			break;
+		default:
+			opts++;
+			break;
+		}
+	}
+
+	*opts_p = opts;
+	return false;
+}
+
+static void
+malloc_abort_invalid_conf(void) {
+	assert(opt_abort_conf);
+	malloc_printf("<jemalloc>: Abort (abort_conf:true) on invalid conf "
+	    "value (see above).\n");
+	abort();
+}
+
+static void
+malloc_conf_error(const char *msg, const char *k, size_t klen, const char *v,
+    size_t vlen) {
+	malloc_printf("<jemalloc>: %s: %.*s:%.*s\n", msg, (int)klen, k,
+	    (int)vlen, v);
+	/* If abort_conf is set, error out after processing all options. */
+	const char *experimental = "experimental_";
+	if (strncmp(k, experimental, strlen(experimental)) == 0) {
+		/* However, tolerate experimental features. */
+		return;
+	}
+	had_conf_error = true;
+}
+
+static void
+malloc_slow_flag_init(void) {
+	/*
+	 * Combine the runtime options into malloc_slow for fast path.  Called
+	 * after processing all the options.
+	 */
+	malloc_slow_flags |= (opt_junk_alloc ? flag_opt_junk_alloc : 0)
+	    | (opt_junk_free ? flag_opt_junk_free : 0)
+	    | (opt_zero ? flag_opt_zero : 0)
+	    | (opt_utrace ? flag_opt_utrace : 0)
+	    | (opt_xmalloc ? flag_opt_xmalloc : 0);
+
+	malloc_slow = (malloc_slow_flags != 0);
+}
+
+/* Number of sources for initializing malloc_conf */
+#define MALLOC_CONF_NSOURCES 4
+
+static const char *
+obtain_malloc_conf(unsigned which_source, char buf[PATH_MAX + 1]) {
+	if (config_debug) {
+		static unsigned read_source = 0;
+		/*
+		 * Each source should only be read once, to minimize # of
+		 * syscalls on init.
+		 */
+		assert(read_source++ == which_source);
+	}
+	assert(which_source < MALLOC_CONF_NSOURCES);
+
+	const char *ret;
+	switch (which_source) {
+	case 0:
+		ret = config_malloc_conf;
+		break;
+	case 1:
+		if (je_malloc_conf != NULL) {
+			/* Use options that were compiled into the program. */
+			ret = je_malloc_conf;
+		} else {
+			/* No configuration specified. */
+			ret = NULL;
+		}
+		break;
+	case 2: {
+		ssize_t linklen = 0;
+#ifndef _WIN32
+		int saved_errno = errno;
+		const char *linkname =
+#  ifdef JEMALLOC_PREFIX
+		    "/etc/"JEMALLOC_PREFIX"malloc.conf"
+#  else
+		    "/etc/malloc.conf"
+#  endif
+		    ;
+
+		/*
+		 * Try to use the contents of the "/etc/malloc.conf" symbolic
+		 * link's name.
+		 */
+#ifndef JEMALLOC_READLINKAT
+		linklen = readlink(linkname, buf, PATH_MAX);
+#else
+		linklen = readlinkat(AT_FDCWD, linkname, buf, PATH_MAX);
+#endif
+		if (linklen == -1) {
+			/* No configuration specified. */
+			linklen = 0;
+			/* Restore errno. */
+			set_errno(saved_errno);
+		}
+#endif
+		buf[linklen] = '\0';
+		ret = buf;
+		break;
+	} case 3: {
+		const char *envname =
+#ifdef JEMALLOC_PREFIX
+		    JEMALLOC_CPREFIX"MALLOC_CONF"
+#else
+		    "MALLOC_CONF"
+#endif
+		    ;
+
+		if ((ret = jemalloc_secure_getenv(envname)) != NULL) {
+			/*
+			 * Do nothing; opts is already initialized to the value
+			 * of the MALLOC_CONF environment variable.
+			 */
+		} else {
+			/* No configuration specified. */
+			ret = NULL;
+		}
+		break;
+	} default:
+		not_reached();
+		ret = NULL;
+	}
+	return ret;
+}
+
+static void
+malloc_conf_init_helper(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS],
+    bool initial_call, const char *opts_cache[MALLOC_CONF_NSOURCES],
+    char buf[PATH_MAX + 1]) {
+	static const char *opts_explain[MALLOC_CONF_NSOURCES] = {
+		"string specified via --with-malloc-conf",
+		"string pointed to by the global variable malloc_conf",
+		"\"name\" of the file referenced by the symbolic link named "
+		    "/etc/malloc.conf",
+		"value of the environment variable MALLOC_CONF"
+	};
+	unsigned i;
+	const char *opts, *k, *v;
+	size_t klen, vlen;
+
+	for (i = 0; i < MALLOC_CONF_NSOURCES; i++) {
+		/* Get runtime configuration. */
+		if (initial_call) {
+			opts_cache[i] = obtain_malloc_conf(i, buf);
+		}
+		opts = opts_cache[i];
+		if (!initial_call && opt_confirm_conf) {
+			malloc_printf(
+			    "<jemalloc>: malloc_conf #%u (%s): \"%s\"\n",
+			    i + 1, opts_explain[i], opts != NULL ? opts : "");
+		}
+		if (opts == NULL) {
+			continue;
+		}
+
+		while (*opts != '\0' && !malloc_conf_next(&opts, &k, &klen, &v,
+		    &vlen)) {
+
+#define CONF_ERROR(msg, k, klen, v, vlen)				\
+			if (!initial_call) {				\
+				malloc_conf_error(			\
+				    msg, k, klen, v, vlen);		\
+				cur_opt_valid = false;			\
+			}
+#define CONF_CONTINUE	{						\
+				if (!initial_call && opt_confirm_conf	\
+				    && cur_opt_valid) {			\
+					malloc_printf("<jemalloc>: -- "	\
+					    "Set conf value: %.*s:%.*s"	\
+					    "\n", (int)klen, k,		\
+					    (int)vlen, v);		\
+				}					\
+				continue;				\
+			}
+#define CONF_MATCH(n)							\
+	(sizeof(n)-1 == klen && strncmp(n, k, klen) == 0)
+#define CONF_MATCH_VALUE(n)						\
+	(sizeof(n)-1 == vlen && strncmp(n, v, vlen) == 0)
+#define CONF_HANDLE_BOOL(o, n)						\
+			if (CONF_MATCH(n)) {				\
+				if (CONF_MATCH_VALUE("true")) {		\
+					o = true;			\
+				} else if (CONF_MATCH_VALUE("false")) {	\
+					o = false;			\
+				} else {				\
+					CONF_ERROR("Invalid conf value",\
+					    k, klen, v, vlen);		\
+				}					\
+				CONF_CONTINUE;				\
+			}
+      /*
+       * One of the CONF_MIN macros below expands, in one of the use points,
+       * to "unsigned integer < 0", which is always false, triggering the
+       * GCC -Wtype-limits warning, which we disable here and re-enable below.
+       */
+      JEMALLOC_DIAGNOSTIC_PUSH
+      JEMALLOC_DIAGNOSTIC_IGNORE_TYPE_LIMITS
+
+#define CONF_DONT_CHECK_MIN(um, min)	false
+#define CONF_CHECK_MIN(um, min)	((um) < (min))
+#define CONF_DONT_CHECK_MAX(um, max)	false
+#define CONF_CHECK_MAX(um, max)	((um) > (max))
+#define CONF_HANDLE_T_U(t, o, n, min, max, check_min, check_max, clip)	\
+			if (CONF_MATCH(n)) {				\
+				uintmax_t um;				\
+				char *end;				\
+									\
+				set_errno(0);				\
+				um = malloc_strtoumax(v, &end, 0);	\
+				if (get_errno() != 0 || (uintptr_t)end -\
+				    (uintptr_t)v != vlen) {		\
+					CONF_ERROR("Invalid conf value",\
+					    k, klen, v, vlen);		\
+				} else if (clip) {			\
+					if (check_min(um, (t)(min))) {	\
+						o = (t)(min);		\
+					} else if (			\
+					    check_max(um, (t)(max))) {	\
+						o = (t)(max);		\
+					} else {			\
+						o = (t)um;		\
+					}				\
+				} else {				\
+					if (check_min(um, (t)(min)) ||	\
+					    check_max(um, (t)(max))) {	\
+						CONF_ERROR(		\
+						    "Out-of-range "	\
+						    "conf value",	\
+						    k, klen, v, vlen);	\
+					} else {			\
+						o = (t)um;		\
+					}				\
+				}					\
+				CONF_CONTINUE;				\
+			}
+#define CONF_HANDLE_UNSIGNED(o, n, min, max, check_min, check_max,	\
+    clip)								\
+			CONF_HANDLE_T_U(unsigned, o, n, min, max,	\
+			    check_min, check_max, clip)
+#define CONF_HANDLE_SIZE_T(o, n, min, max, check_min, check_max, clip)	\
+			CONF_HANDLE_T_U(size_t, o, n, min, max,		\
+			    check_min, check_max, clip)
+#define CONF_HANDLE_SSIZE_T(o, n, min, max)				\
+			if (CONF_MATCH(n)) {				\
+				long l;					\
+				char *end;				\
+									\
+				set_errno(0);				\
+				l = strtol(v, &end, 0);			\
+				if (get_errno() != 0 || (uintptr_t)end -\
+				    (uintptr_t)v != vlen) {		\
+					CONF_ERROR("Invalid conf value",\
+					    k, klen, v, vlen);		\
+				} else if (l < (ssize_t)(min) || l >	\
+				    (ssize_t)(max)) {			\
+					CONF_ERROR(			\
+					    "Out-of-range conf value",	\
+					    k, klen, v, vlen);		\
+				} else {				\
+					o = l;				\
+				}					\
+				CONF_CONTINUE;				\
+			}
+#define CONF_HANDLE_CHAR_P(o, n, d)					\
+			if (CONF_MATCH(n)) {				\
+				size_t cpylen = (vlen <=		\
+				    sizeof(o)-1) ? vlen :		\
+				    sizeof(o)-1;			\
+				strncpy(o, v, cpylen);			\
+				o[cpylen] = '\0';			\
+				CONF_CONTINUE;				\
+			}
+
+			bool cur_opt_valid = true;
+
+			CONF_HANDLE_BOOL(opt_confirm_conf, "confirm_conf")
+			if (initial_call) {
+				continue;
+			}
+
+			CONF_HANDLE_BOOL(opt_abort, "abort")
+			CONF_HANDLE_BOOL(opt_abort_conf, "abort_conf")
+			if (strncmp("metadata_thp", k, klen) == 0) {
+				int i;
+				bool match = false;
+				for (i = 0; i < metadata_thp_mode_limit; i++) {
+					if (strncmp(metadata_thp_mode_names[i],
+					    v, vlen) == 0) {
+						opt_metadata_thp = i;
+						match = true;
+						break;
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_BOOL(opt_retain, "retain")
+			if (strncmp("dss", k, klen) == 0) {
+				int i;
+				bool match = false;
+				for (i = 0; i < dss_prec_limit; i++) {
+					if (strncmp(dss_prec_names[i], v, vlen)
+					    == 0) {
+						if (extent_dss_prec_set(i)) {
+							CONF_ERROR(
+							    "Error setting dss",
+							    k, klen, v, vlen);
+						} else {
+							opt_dss =
+							    dss_prec_names[i];
+							match = true;
+							break;
+						}
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_UNSIGNED(opt_narenas, "narenas", 1,
+			    UINT_MAX, CONF_CHECK_MIN, CONF_DONT_CHECK_MAX,
+			    false)
+			if (CONF_MATCH("bin_shards")) {
+				const char *bin_shards_segment_cur = v;
+				size_t vlen_left = vlen;
+				do {
+					size_t size_start;
+					size_t size_end;
+					size_t nshards;
+					bool err = malloc_conf_multi_sizes_next(
+					    &bin_shards_segment_cur, &vlen_left,
+					    &size_start, &size_end, &nshards);
+					if (err || bin_update_shard_size(
+					    bin_shard_sizes, size_start,
+					    size_end, nshards)) {
+						CONF_ERROR(
+						    "Invalid settings for "
+						    "bin_shards", k, klen, v,
+						    vlen);
+						break;
+					}
+				} while (vlen_left > 0);
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_SSIZE_T(opt_dirty_decay_ms,
+			    "dirty_decay_ms", -1, NSTIME_SEC_MAX * KQU(1000) <
+			    QU(SSIZE_MAX) ? NSTIME_SEC_MAX * KQU(1000) :
+			    SSIZE_MAX);
+			CONF_HANDLE_SSIZE_T(opt_muzzy_decay_ms,
+			    "muzzy_decay_ms", -1, NSTIME_SEC_MAX * KQU(1000) <
+			    QU(SSIZE_MAX) ? NSTIME_SEC_MAX * KQU(1000) :
+			    SSIZE_MAX);
+			CONF_HANDLE_BOOL(opt_stats_print, "stats_print")
+			if (CONF_MATCH("stats_print_opts")) {
+				init_opt_stats_print_opts(v, vlen);
+				CONF_CONTINUE;
+			}
+			if (config_fill) {
+				if (CONF_MATCH("junk")) {
+					if (CONF_MATCH_VALUE("true")) {
+						opt_junk = "true";
+						opt_junk_alloc = opt_junk_free =
+						    true;
+					} else if (CONF_MATCH_VALUE("false")) {
+						opt_junk = "false";
+						opt_junk_alloc = opt_junk_free =
+						    false;
+					} else if (CONF_MATCH_VALUE("alloc")) {
+						opt_junk = "alloc";
+						opt_junk_alloc = true;
+						opt_junk_free = false;
+					} else if (CONF_MATCH_VALUE("free")) {
+						opt_junk = "free";
+						opt_junk_alloc = false;
+						opt_junk_free = true;
+					} else {
+						CONF_ERROR(
+						    "Invalid conf value",
+						    k, klen, v, vlen);
+					}
+					CONF_CONTINUE;
+				}
+				CONF_HANDLE_BOOL(opt_zero, "zero")
+			}
+			if (config_utrace) {
+				CONF_HANDLE_BOOL(opt_utrace, "utrace")
+			}
+			if (config_xmalloc) {
+				CONF_HANDLE_BOOL(opt_xmalloc, "xmalloc")
+			}
+			CONF_HANDLE_BOOL(opt_tcache, "tcache")
+			CONF_HANDLE_SSIZE_T(opt_lg_tcache_max, "lg_tcache_max",
+			    -1, (sizeof(size_t) << 3) - 1)
+
+			/*
+			 * The runtime option of oversize_threshold remains
+			 * undocumented.  It may be tweaked in the next major
+			 * release (6.0).  The default value 8M is rather
+			 * conservative / safe.  Tuning it further down may
+			 * improve fragmentation a bit more, but may also cause
+			 * contention on the huge arena.
+			 */
+			CONF_HANDLE_SIZE_T(opt_oversize_threshold,
+			    "oversize_threshold", 0, SC_LARGE_MAXCLASS,
+			    CONF_DONT_CHECK_MIN, CONF_CHECK_MAX, false)
+			CONF_HANDLE_SIZE_T(opt_lg_extent_max_active_fit,
+			    "lg_extent_max_active_fit", 0,
+			    (sizeof(size_t) << 3), CONF_DONT_CHECK_MIN,
+			    CONF_CHECK_MAX, false)
+
+			if (strncmp("percpu_arena", k, klen) == 0) {
+				bool match = false;
+				for (int i = percpu_arena_mode_names_base; i <
+				    percpu_arena_mode_names_limit; i++) {
+					if (strncmp(percpu_arena_mode_names[i],
+					    v, vlen) == 0) {
+						if (!have_percpu_arena) {
+							CONF_ERROR(
+							    "No getcpu support",
+							    k, klen, v, vlen);
+						}
+						opt_percpu_arena = i;
+						match = true;
+						break;
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_HANDLE_BOOL(opt_background_thread,
+			    "background_thread");
+			CONF_HANDLE_SIZE_T(opt_max_background_threads,
+					   "max_background_threads", 1,
+					   opt_max_background_threads,
+					   CONF_CHECK_MIN, CONF_CHECK_MAX,
+					   true);
+			if (CONF_MATCH("slab_sizes")) {
+				bool err;
+				const char *slab_size_segment_cur = v;
+				size_t vlen_left = vlen;
+				do {
+					size_t slab_start;
+					size_t slab_end;
+					size_t pgs;
+					err = malloc_conf_multi_sizes_next(
+					    &slab_size_segment_cur,
+					    &vlen_left, &slab_start, &slab_end,
+					    &pgs);
+					if (!err) {
+						sc_data_update_slab_size(
+						    sc_data, slab_start,
+						    slab_end, (int)pgs);
+					} else {
+						CONF_ERROR("Invalid settings "
+						    "for slab_sizes",
+						    k, klen, v, vlen);
+					}
+				} while (!err && vlen_left > 0);
+				CONF_CONTINUE;
+			}
+			if (config_prof) {
+				CONF_HANDLE_BOOL(opt_prof, "prof")
+				CONF_HANDLE_CHAR_P(opt_prof_prefix,
+				    "prof_prefix", "jeprof")
+				CONF_HANDLE_BOOL(opt_prof_active, "prof_active")
+				CONF_HANDLE_BOOL(opt_prof_thread_active_init,
+				    "prof_thread_active_init")
+				CONF_HANDLE_SIZE_T(opt_lg_prof_sample,
+				    "lg_prof_sample", 0, (sizeof(uint64_t) << 3)
+				    - 1, CONF_DONT_CHECK_MIN, CONF_CHECK_MAX,
+				    true)
+				CONF_HANDLE_BOOL(opt_prof_accum, "prof_accum")
+				CONF_HANDLE_SSIZE_T(opt_lg_prof_interval,
+				    "lg_prof_interval", -1,
+				    (sizeof(uint64_t) << 3) - 1)
+				CONF_HANDLE_BOOL(opt_prof_gdump, "prof_gdump")
+				CONF_HANDLE_BOOL(opt_prof_final, "prof_final")
+				CONF_HANDLE_BOOL(opt_prof_leak, "prof_leak")
+				CONF_HANDLE_BOOL(opt_prof_log, "prof_log")
+			}
+			if (config_log) {
+				if (CONF_MATCH("log")) {
+					size_t cpylen = (
+					    vlen <= sizeof(log_var_names) ?
+					    vlen : sizeof(log_var_names) - 1);
+					strncpy(log_var_names, v, cpylen);
+					log_var_names[cpylen] = '\0';
+					CONF_CONTINUE;
+				}
+			}
+			if (CONF_MATCH("thp")) {
+				bool match = false;
+				for (int i = 0; i < thp_mode_names_limit; i++) {
+					if (strncmp(thp_mode_names[i],v, vlen)
+					    == 0) {
+						if (!have_madvise_huge) {
+							CONF_ERROR(
+							    "No THP support",
+							    k, klen, v, vlen);
+						}
+						opt_thp = i;
+						match = true;
+						break;
+					}
+				}
+				if (!match) {
+					CONF_ERROR("Invalid conf value",
+					    k, klen, v, vlen);
+				}
+				CONF_CONTINUE;
+			}
+			CONF_ERROR("Invalid conf pair", k, klen, v, vlen);
+#undef CONF_ERROR
+#undef CONF_CONTINUE
+#undef CONF_MATCH
+#undef CONF_MATCH_VALUE
+#undef CONF_HANDLE_BOOL
+#undef CONF_DONT_CHECK_MIN
+#undef CONF_CHECK_MIN
+#undef CONF_DONT_CHECK_MAX
+#undef CONF_CHECK_MAX
+#undef CONF_HANDLE_T_U
+#undef CONF_HANDLE_UNSIGNED
+#undef CONF_HANDLE_SIZE_T
+#undef CONF_HANDLE_SSIZE_T
+#undef CONF_HANDLE_CHAR_P
+    /* Re-enable diagnostic "-Wtype-limits" */
+    JEMALLOC_DIAGNOSTIC_POP
+		}
+		if (opt_abort_conf && had_conf_error) {
+			malloc_abort_invalid_conf();
+		}
+	}
+	atomic_store_b(&log_init_done, true, ATOMIC_RELEASE);
+}
+
+static void
+malloc_conf_init(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS]) {
+	const char *opts_cache[MALLOC_CONF_NSOURCES] = {NULL, NULL, NULL, NULL};
+	char buf[PATH_MAX + 1];
+
+	/* The first call only set the confirm_conf option and opts_cache */
+	malloc_conf_init_helper(NULL, NULL, true, opts_cache, buf);
+	malloc_conf_init_helper(sc_data, bin_shard_sizes, false, opts_cache,
+	    NULL);
+}
+
+#undef MALLOC_CONF_NSOURCES
+
+static bool
+malloc_init_hard_needed(void) {
+	if (malloc_initialized() || (IS_INITIALIZER && malloc_init_state ==
+	    malloc_init_recursible)) {
+		/*
+		 * Another thread initialized the allocator before this one
+		 * acquired init_lock, or this thread is the initializing
+		 * thread, and it is recursively allocating.
+		 */
+		return false;
+	}
+#ifdef JEMALLOC_THREADED_INIT
+	if (malloc_initializer != NO_INITIALIZER && !IS_INITIALIZER) {
+		/* Busy-wait until the initializing thread completes. */
+		spin_t spinner = SPIN_INITIALIZER;
+		do {
+			malloc_mutex_unlock(TSDN_NULL, &init_lock);
+			spin_adaptive(&spinner);
+			malloc_mutex_lock(TSDN_NULL, &init_lock);
+		} while (!malloc_initialized());
+		return false;
+	}
+#endif
+	return true;
+}
+
+static bool
+malloc_init_hard_a0_locked() {
+	malloc_initializer = INITIALIZER;
+
+	JEMALLOC_DIAGNOSTIC_PUSH
+	JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+	sc_data_t sc_data = {0};
+	JEMALLOC_DIAGNOSTIC_POP
+
+	/*
+	 * Ordering here is somewhat tricky; we need sc_boot() first, since that
+	 * determines what the size classes will be, and then
+	 * malloc_conf_init(), since any slab size tweaking will need to be done
+	 * before sz_boot and bin_boot, which assume that the values they read
+	 * out of sc_data_global are final.
+	 */
+	sc_boot(&sc_data);
+	unsigned bin_shard_sizes[SC_NBINS];
+	bin_shard_sizes_boot(bin_shard_sizes);
+	/*
+	 * prof_boot0 only initializes opt_prof_prefix.  We need to do it before
+	 * we parse malloc_conf options, in case malloc_conf parsing overwrites
+	 * it.
+	 */
+	if (config_prof) {
+		prof_boot0();
+	}
+	malloc_conf_init(&sc_data, bin_shard_sizes);
+	sz_boot(&sc_data);
+	bin_boot(&sc_data, bin_shard_sizes);
+
+	if (opt_stats_print) {
+		/* Print statistics at exit. */
+		if (atexit(stats_print_atexit) != 0) {
+			malloc_write("<jemalloc>: Error in atexit()\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+	}
+	if (pages_boot()) {
+		return true;
+	}
+	if (base_boot(TSDN_NULL)) {
+		return true;
+	}
+	if (extent_boot()) {
+		return true;
+	}
+	if (ctl_boot()) {
+		return true;
+	}
+	if (config_prof) {
+		prof_boot1();
+	}
+	arena_boot(&sc_data);
+	if (tcache_boot(TSDN_NULL)) {
+		return true;
+	}
+	if (malloc_mutex_init(&arenas_lock, "arenas", WITNESS_RANK_ARENAS,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	hook_boot();
+	/*
+	 * Create enough scaffolding to allow recursive allocation in
+	 * malloc_ncpus().
+	 */
+	narenas_auto = 1;
+	manual_arena_base = narenas_auto + 1;
+	memset(arenas, 0, sizeof(arena_t *) * narenas_auto);
+	/*
+	 * Initialize one arena here.  The rest are lazily created in
+	 * arena_choose_hard().
+	 */
+	if (arena_init(TSDN_NULL, 0, (extent_hooks_t *)&extent_hooks_default)
+	    == NULL) {
+		return true;
+	}
+	a0 = arena_get(TSDN_NULL, 0, false);
+	malloc_init_state = malloc_init_a0_initialized;
+
+	return false;
+}
+
+static bool
+malloc_init_hard_a0(void) {
+	bool ret;
+
+	malloc_mutex_lock(TSDN_NULL, &init_lock);
+	ret = malloc_init_hard_a0_locked();
+	malloc_mutex_unlock(TSDN_NULL, &init_lock);
+	return ret;
+}
+
+/* Initialize data structures which may trigger recursive allocation. */
+static bool
+malloc_init_hard_recursible(void) {
+	malloc_init_state = malloc_init_recursible;
+
+	ncpus = malloc_ncpus();
+
+#if (defined(JEMALLOC_HAVE_PTHREAD_ATFORK) && !defined(JEMALLOC_MUTEX_INIT_CB) \
+    && !defined(JEMALLOC_ZONE) && !defined(_WIN32) && \
+    !defined(__native_client__))
+	/* LinuxThreads' pthread_atfork() allocates. */
+	if (pthread_atfork(jemalloc_prefork, jemalloc_postfork_parent,
+	    jemalloc_postfork_child) != 0) {
+		malloc_write("<jemalloc>: Error in pthread_atfork()\n");
+		if (opt_abort) {
+			abort();
+		}
+		return true;
+	}
+#endif
+
+	if (background_thread_boot0()) {
+		return true;
+	}
+
+	return false;
+}
+
+static unsigned
+malloc_narenas_default(void) {
+	assert(ncpus > 0);
+	/*
+	 * For SMP systems, create more than one arena per CPU by
+	 * default.
+	 */
+	if (ncpus > 1) {
+		return ncpus << 2;
+	} else {
+		return 1;
+	}
+}
+
+static percpu_arena_mode_t
+percpu_arena_as_initialized(percpu_arena_mode_t mode) {
+	assert(!malloc_initialized());
+	assert(mode <= percpu_arena_disabled);
+
+	if (mode != percpu_arena_disabled) {
+		mode += percpu_arena_mode_enabled_base;
+	}
+
+	return mode;
+}
+
+static bool
+malloc_init_narenas(void) {
+	assert(ncpus > 0);
+
+	if (opt_percpu_arena != percpu_arena_disabled) {
+		if (!have_percpu_arena || malloc_getcpu() < 0) {
+			opt_percpu_arena = percpu_arena_disabled;
+			malloc_printf("<jemalloc>: perCPU arena getcpu() not "
+			    "available. Setting narenas to %u.\n", opt_narenas ?
+			    opt_narenas : malloc_narenas_default());
+			if (opt_abort) {
+				abort();
+			}
+		} else {
+			if (ncpus >= MALLOCX_ARENA_LIMIT) {
+				malloc_printf("<jemalloc>: narenas w/ percpu"
+				    "arena beyond limit (%d)\n", ncpus);
+				if (opt_abort) {
+					abort();
+				}
+				return true;
+			}
+			/* NB: opt_percpu_arena isn't fully initialized yet. */
+			if (percpu_arena_as_initialized(opt_percpu_arena) ==
+			    per_phycpu_arena && ncpus % 2 != 0) {
+				malloc_printf("<jemalloc>: invalid "
+				    "configuration -- per physical CPU arena "
+				    "with odd number (%u) of CPUs (no hyper "
+				    "threading?).\n", ncpus);
+				if (opt_abort)
+					abort();
+			}
+			unsigned n = percpu_arena_ind_limit(
+			    percpu_arena_as_initialized(opt_percpu_arena));
+			if (opt_narenas < n) {
+				/*
+				 * If narenas is specified with percpu_arena
+				 * enabled, actual narenas is set as the greater
+				 * of the two. percpu_arena_choose will be free
+				 * to use any of the arenas based on CPU
+				 * id. This is conservative (at a small cost)
+				 * but ensures correctness.
+				 *
+				 * If for some reason the ncpus determined at
+				 * boot is not the actual number (e.g. because
+				 * of affinity setting from numactl), reserving
+				 * narenas this way provides a workaround for
+				 * percpu_arena.
+				 */
+				opt_narenas = n;
+			}
+		}
+	}
+	if (opt_narenas == 0) {
+		opt_narenas = malloc_narenas_default();
+	}
+	assert(opt_narenas > 0);
+
+	narenas_auto = opt_narenas;
+	/*
+	 * Limit the number of arenas to the indexing range of MALLOCX_ARENA().
+	 */
+	if (narenas_auto >= MALLOCX_ARENA_LIMIT) {
+		narenas_auto = MALLOCX_ARENA_LIMIT - 1;
+		malloc_printf("<jemalloc>: Reducing narenas to limit (%d)\n",
+		    narenas_auto);
+	}
+	narenas_total_set(narenas_auto);
+	if (arena_init_huge()) {
+		narenas_total_inc();
+	}
+	manual_arena_base = narenas_total_get();
+
+	return false;
+}
+
+static void
+malloc_init_percpu(void) {
+	opt_percpu_arena = percpu_arena_as_initialized(opt_percpu_arena);
+}
+
+static bool
+malloc_init_hard_finish(void) {
+	if (malloc_mutex_boot()) {
+		return true;
+	}
+
+	malloc_init_state = malloc_init_initialized;
+	malloc_slow_flag_init();
+
+	return false;
+}
+
+static void
+malloc_init_hard_cleanup(tsdn_t *tsdn, bool reentrancy_set) {
+	malloc_mutex_assert_owner(tsdn, &init_lock);
+	malloc_mutex_unlock(tsdn, &init_lock);
+	if (reentrancy_set) {
+		assert(!tsdn_null(tsdn));
+		tsd_t *tsd = tsdn_tsd(tsdn);
+		assert(tsd_reentrancy_level_get(tsd) > 0);
+		post_reentrancy(tsd);
+	}
+}
+
+static bool
+malloc_init_hard(void) {
+	tsd_t *tsd;
+
+#if defined(_WIN32) && _WIN32_WINNT < 0x0600
+	_init_init_lock();
+#endif
+	malloc_mutex_lock(TSDN_NULL, &init_lock);
+
+#define UNLOCK_RETURN(tsdn, ret, reentrancy)		\
+	malloc_init_hard_cleanup(tsdn, reentrancy);	\
+	return ret;
+
+	if (!malloc_init_hard_needed()) {
+		UNLOCK_RETURN(TSDN_NULL, false, false)
+	}
+
+	if (malloc_init_state != malloc_init_a0_initialized &&
+	    malloc_init_hard_a0_locked()) {
+		UNLOCK_RETURN(TSDN_NULL, true, false)
+	}
+
+	malloc_mutex_unlock(TSDN_NULL, &init_lock);
+	/* Recursive allocation relies on functional tsd. */
+	tsd = malloc_tsd_boot0();
+	if (tsd == NULL) {
+		return true;
+	}
+	if (malloc_init_hard_recursible()) {
+		return true;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &init_lock);
+	/* Set reentrancy level to 1 during init. */
+	pre_reentrancy(tsd, NULL);
+	/* Initialize narenas before prof_boot2 (for allocation). */
+	if (malloc_init_narenas() || background_thread_boot1(tsd_tsdn(tsd))) {
+		UNLOCK_RETURN(tsd_tsdn(tsd), true, true)
+	}
+	if (config_prof && prof_boot2(tsd)) {
+		UNLOCK_RETURN(tsd_tsdn(tsd), true, true)
+	}
+
+	malloc_init_percpu();
+
+	if (malloc_init_hard_finish()) {
+		UNLOCK_RETURN(tsd_tsdn(tsd), true, true)
+	}
+	post_reentrancy(tsd);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &init_lock);
+
+	witness_assert_lockless(witness_tsd_tsdn(
+	    tsd_witness_tsdp_get_unsafe(tsd)));
+	malloc_tsd_boot1();
+	/* Update TSD after tsd_boot1. */
+	tsd = tsd_fetch();
+	if (opt_background_thread) {
+		assert(have_background_thread);
+		/*
+		 * Need to finish init & unlock first before creating background
+		 * threads (pthread_create depends on malloc).  ctl_init (which
+		 * sets isthreaded) needs to be called without holding any lock.
+		 */
+		background_thread_ctl_init(tsd_tsdn(tsd));
+		if (background_thread_create(tsd, 0)) {
+			return true;
+		}
+	}
+#undef UNLOCK_RETURN
+	return false;
+}
+
+/*
+ * End initialization functions.
+ */
+/******************************************************************************/
+/*
+ * Begin allocation-path internal functions and data structures.
+ */
+
+/*
+ * Settings determined by the documented behavior of the allocation functions.
+ */
+typedef struct static_opts_s static_opts_t;
+struct static_opts_s {
+	/* Whether or not allocation size may overflow. */
+	bool may_overflow;
+
+	/*
+	 * Whether or not allocations (with alignment) of size 0 should be
+	 * treated as size 1.
+	 */
+	bool bump_empty_aligned_alloc;
+	/*
+	 * Whether to assert that allocations are not of size 0 (after any
+	 * bumping).
+	 */
+	bool assert_nonempty_alloc;
+
+	/*
+	 * Whether or not to modify the 'result' argument to malloc in case of
+	 * error.
+	 */
+	bool null_out_result_on_error;
+	/* Whether to set errno when we encounter an error condition. */
+	bool set_errno_on_error;
+
+	/*
+	 * The minimum valid alignment for functions requesting aligned storage.
+	 */
+	size_t min_alignment;
+
+	/* The error string to use if we oom. */
+	const char *oom_string;
+	/* The error string to use if the passed-in alignment is invalid. */
+	const char *invalid_alignment_string;
+
+	/*
+	 * False if we're configured to skip some time-consuming operations.
+	 *
+	 * This isn't really a malloc "behavior", but it acts as a useful
+	 * summary of several other static (or at least, static after program
+	 * initialization) options.
+	 */
+	bool slow;
+	/*
+	 * Return size.
+	 */
+	bool usize;
+};
+
+JEMALLOC_ALWAYS_INLINE void
+static_opts_init(static_opts_t *static_opts) {
+	static_opts->may_overflow = false;
+	static_opts->bump_empty_aligned_alloc = false;
+	static_opts->assert_nonempty_alloc = false;
+	static_opts->null_out_result_on_error = false;
+	static_opts->set_errno_on_error = false;
+	static_opts->min_alignment = 0;
+	static_opts->oom_string = "";
+	static_opts->invalid_alignment_string = "";
+	static_opts->slow = false;
+	static_opts->usize = false;
+}
+
+/*
+ * These correspond to the macros in jemalloc/jemalloc_macros.h.  Broadly, we
+ * should have one constant here per magic value there.  Note however that the
+ * representations need not be related.
+ */
+#define TCACHE_IND_NONE ((unsigned)-1)
+#define TCACHE_IND_AUTOMATIC ((unsigned)-2)
+#define ARENA_IND_AUTOMATIC ((unsigned)-1)
+
+typedef struct dynamic_opts_s dynamic_opts_t;
+struct dynamic_opts_s {
+	void **result;
+	size_t usize;
+	size_t num_items;
+	size_t item_size;
+	size_t alignment;
+	bool zero;
+	unsigned tcache_ind;
+	unsigned arena_ind;
+};
+
+JEMALLOC_ALWAYS_INLINE void
+dynamic_opts_init(dynamic_opts_t *dynamic_opts) {
+	dynamic_opts->result = NULL;
+	dynamic_opts->usize = 0;
+	dynamic_opts->num_items = 0;
+	dynamic_opts->item_size = 0;
+	dynamic_opts->alignment = 0;
+	dynamic_opts->zero = false;
+	dynamic_opts->tcache_ind = TCACHE_IND_AUTOMATIC;
+	dynamic_opts->arena_ind = ARENA_IND_AUTOMATIC;
+}
+
+/* ind is ignored if dopts->alignment > 0. */
+JEMALLOC_ALWAYS_INLINE void *
+imalloc_no_sample(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd,
+    size_t size, size_t usize, szind_t ind) {
+	tcache_t *tcache;
+	arena_t *arena;
+
+	/* Fill in the tcache. */
+	if (dopts->tcache_ind == TCACHE_IND_AUTOMATIC) {
+		if (likely(!sopts->slow)) {
+			/* Getting tcache ptr unconditionally. */
+			tcache = tsd_tcachep_get(tsd);
+			assert(tcache == tcache_get(tsd));
+		} else {
+			tcache = tcache_get(tsd);
+		}
+	} else if (dopts->tcache_ind == TCACHE_IND_NONE) {
+		tcache = NULL;
+	} else {
+		tcache = tcaches_get(tsd, dopts->tcache_ind);
+	}
+
+	/* Fill in the arena. */
+	if (dopts->arena_ind == ARENA_IND_AUTOMATIC) {
+		/*
+		 * In case of automatic arena management, we defer arena
+		 * computation until as late as we can, hoping to fill the
+		 * allocation out of the tcache.
+		 */
+		arena = NULL;
+	} else {
+		arena = arena_get(tsd_tsdn(tsd), dopts->arena_ind, true);
+	}
+
+	if (unlikely(dopts->alignment != 0)) {
+		return ipalloct(tsd_tsdn(tsd), usize, dopts->alignment,
+		    dopts->zero, tcache, arena);
+	}
+
+	return iallocztm(tsd_tsdn(tsd), size, ind, dopts->zero, tcache, false,
+	    arena, sopts->slow);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+imalloc_sample(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd,
+    size_t usize, szind_t ind) {
+	void *ret;
+
+	/*
+	 * For small allocations, sampling bumps the usize.  If so, we allocate
+	 * from the ind_large bucket.
+	 */
+	szind_t ind_large;
+	size_t bumped_usize = usize;
+
+	if (usize <= SC_SMALL_MAXCLASS) {
+		assert(((dopts->alignment == 0) ?
+		    sz_s2u(SC_LARGE_MINCLASS) :
+		    sz_sa2u(SC_LARGE_MINCLASS, dopts->alignment))
+			== SC_LARGE_MINCLASS);
+		ind_large = sz_size2index(SC_LARGE_MINCLASS);
+		bumped_usize = sz_s2u(SC_LARGE_MINCLASS);
+		ret = imalloc_no_sample(sopts, dopts, tsd, bumped_usize,
+		    bumped_usize, ind_large);
+		if (unlikely(ret == NULL)) {
+			return NULL;
+		}
+		arena_prof_promote(tsd_tsdn(tsd), ret, usize);
+	} else {
+		ret = imalloc_no_sample(sopts, dopts, tsd, usize, usize, ind);
+	}
+
+	return ret;
+}
+
+/*
+ * Returns true if the allocation will overflow, and false otherwise.  Sets
+ * *size to the product either way.
+ */
+JEMALLOC_ALWAYS_INLINE bool
+compute_size_with_overflow(bool may_overflow, dynamic_opts_t *dopts,
+    size_t *size) {
+	/*
+	 * This function is just num_items * item_size, except that we may have
+	 * to check for overflow.
+	 */
+
+	if (!may_overflow) {
+		assert(dopts->num_items == 1);
+		*size = dopts->item_size;
+		return false;
+	}
+
+	/* A size_t with its high-half bits all set to 1. */
+	static const size_t high_bits = SIZE_T_MAX << (sizeof(size_t) * 8 / 2);
+
+	*size = dopts->item_size * dopts->num_items;
+
+	if (unlikely(*size == 0)) {
+		return (dopts->num_items != 0 && dopts->item_size != 0);
+	}
+
+	/*
+	 * We got a non-zero size, but we don't know if we overflowed to get
+	 * there.  To avoid having to do a divide, we'll be clever and note that
+	 * if both A and B can be represented in N/2 bits, then their product
+	 * can be represented in N bits (without the possibility of overflow).
+	 */
+	if (likely((high_bits & (dopts->num_items | dopts->item_size)) == 0)) {
+		return false;
+	}
+	if (likely(*size / dopts->item_size == dopts->num_items)) {
+		return false;
+	}
+	return true;
+}
+
+JEMALLOC_ALWAYS_INLINE int
+imalloc_body(static_opts_t *sopts, dynamic_opts_t *dopts, tsd_t *tsd) {
+	/* Where the actual allocated memory will live. */
+	void *allocation = NULL;
+	/* Filled in by compute_size_with_overflow below. */
+	size_t size = 0;
+	/*
+	 * For unaligned allocations, we need only ind.  For aligned
+	 * allocations, or in case of stats or profiling we need usize.
+	 *
+	 * These are actually dead stores, in that their values are reset before
+	 * any branch on their value is taken.  Sometimes though, it's
+	 * convenient to pass them as arguments before this point.  To avoid
+	 * undefined behavior then, we initialize them with dummy stores.
+	 */
+	szind_t ind = 0;
+	size_t usize = 0;
+
+	/* Reentrancy is only checked on slow path. */
+	int8_t reentrancy_level;
+
+	/* Compute the amount of memory the user wants. */
+	if (unlikely(compute_size_with_overflow(sopts->may_overflow, dopts,
+	    &size))) {
+		goto label_oom;
+	}
+
+	if (unlikely(dopts->alignment < sopts->min_alignment
+	    || (dopts->alignment & (dopts->alignment - 1)) != 0)) {
+		goto label_invalid_alignment;
+	}
+
+	/* This is the beginning of the "core" algorithm. */
+
+	if (dopts->alignment == 0) {
+		ind = sz_size2index(size);
+		if (unlikely(ind >= SC_NSIZES)) {
+			goto label_oom;
+		}
+		if (config_stats || (config_prof && opt_prof) || sopts->usize) {
+			usize = sz_index2size(ind);
+			dopts->usize = usize;
+			assert(usize > 0 && usize
+			    <= SC_LARGE_MAXCLASS);
+		}
+	} else {
+		if (sopts->bump_empty_aligned_alloc) {
+			if (unlikely(size == 0)) {
+				size = 1;
+			}
+		}
+		usize = sz_sa2u(size, dopts->alignment);
+		dopts->usize = usize;
+		if (unlikely(usize == 0
+		    || usize > SC_LARGE_MAXCLASS)) {
+			goto label_oom;
+		}
+	}
+	/* Validate the user input. */
+	if (sopts->assert_nonempty_alloc) {
+		assert (size != 0);
+	}
+
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	/*
+	 * If we need to handle reentrancy, we can do it out of a
+	 * known-initialized arena (i.e. arena 0).
+	 */
+	reentrancy_level = tsd_reentrancy_level_get(tsd);
+	if (sopts->slow && unlikely(reentrancy_level > 0)) {
+		/*
+		 * We should never specify particular arenas or tcaches from
+		 * within our internal allocations.
+		 */
+		assert(dopts->tcache_ind == TCACHE_IND_AUTOMATIC ||
+		    dopts->tcache_ind == TCACHE_IND_NONE);
+		assert(dopts->arena_ind == ARENA_IND_AUTOMATIC);
+		dopts->tcache_ind = TCACHE_IND_NONE;
+		/* We know that arena 0 has already been initialized. */
+		dopts->arena_ind = 0;
+	}
+
+	/* If profiling is on, get our profiling context. */
+	if (config_prof && opt_prof) {
+		/*
+		 * Note that if we're going down this path, usize must have been
+		 * initialized in the previous if statement.
+		 */
+		prof_tctx_t *tctx = prof_alloc_prep(
+		    tsd, usize, prof_active_get_unlocked(), true);
+
+		alloc_ctx_t alloc_ctx;
+		if (likely((uintptr_t)tctx == (uintptr_t)1U)) {
+			alloc_ctx.slab = (usize
+			    <= SC_SMALL_MAXCLASS);
+			allocation = imalloc_no_sample(
+			    sopts, dopts, tsd, usize, usize, ind);
+		} else if ((uintptr_t)tctx > (uintptr_t)1U) {
+			/*
+			 * Note that ind might still be 0 here.  This is fine;
+			 * imalloc_sample ignores ind if dopts->alignment > 0.
+			 */
+			allocation = imalloc_sample(
+			    sopts, dopts, tsd, usize, ind);
+			alloc_ctx.slab = false;
+		} else {
+			allocation = NULL;
+		}
+
+		if (unlikely(allocation == NULL)) {
+			prof_alloc_rollback(tsd, tctx, true);
+			goto label_oom;
+		}
+		prof_malloc(tsd_tsdn(tsd), allocation, usize, &alloc_ctx, tctx);
+	} else {
+		/*
+		 * If dopts->alignment > 0, then ind is still 0, but usize was
+		 * computed in the previous if statement.  Down the positive
+		 * alignment path, imalloc_no_sample ignores ind and size
+		 * (relying only on usize).
+		 */
+		allocation = imalloc_no_sample(sopts, dopts, tsd, size, usize,
+		    ind);
+		if (unlikely(allocation == NULL)) {
+			goto label_oom;
+		}
+	}
+
+	/*
+	 * Allocation has been done at this point.  We still have some
+	 * post-allocation work to do though.
+	 */
+	assert(dopts->alignment == 0
+	    || ((uintptr_t)allocation & (dopts->alignment - 1)) == ZU(0));
+
+	if (config_stats) {
+		assert(usize == isalloc(tsd_tsdn(tsd), allocation));
+		*tsd_thread_allocatedp_get(tsd) += usize;
+	}
+
+	if (sopts->slow) {
+		UTRACE(0, size, allocation);
+	}
+
+	/* Success! */
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	*dopts->result = allocation;
+	return 0;
+
+label_oom:
+	if (unlikely(sopts->slow) && config_xmalloc && unlikely(opt_xmalloc)) {
+		malloc_write(sopts->oom_string);
+		abort();
+	}
+
+	if (sopts->slow) {
+		UTRACE(NULL, size, NULL);
+	}
+
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	if (sopts->set_errno_on_error) {
+		set_errno(ENOMEM);
+	}
+
+	if (sopts->null_out_result_on_error) {
+		*dopts->result = NULL;
+	}
+
+	return ENOMEM;
+
+	/*
+	 * This label is only jumped to by one goto; we move it out of line
+	 * anyways to avoid obscuring the non-error paths, and for symmetry with
+	 * the oom case.
+	 */
+label_invalid_alignment:
+	if (config_xmalloc && unlikely(opt_xmalloc)) {
+		malloc_write(sopts->invalid_alignment_string);
+		abort();
+	}
+
+	if (sopts->set_errno_on_error) {
+		set_errno(EINVAL);
+	}
+
+	if (sopts->slow) {
+		UTRACE(NULL, size, NULL);
+	}
+
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	if (sopts->null_out_result_on_error) {
+		*dopts->result = NULL;
+	}
+
+	return EINVAL;
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+imalloc_init_check(static_opts_t *sopts, dynamic_opts_t *dopts) {
+	if (unlikely(!malloc_initialized()) && unlikely(malloc_init())) {
+		if (config_xmalloc && unlikely(opt_xmalloc)) {
+			malloc_write(sopts->oom_string);
+			abort();
+		}
+		UTRACE(NULL, dopts->num_items * dopts->item_size, NULL);
+		set_errno(ENOMEM);
+		*dopts->result = NULL;
+
+		return false;
+	}
+
+	return true;
+}
+
+/* Returns the errno-style error code of the allocation. */
+JEMALLOC_ALWAYS_INLINE int
+imalloc(static_opts_t *sopts, dynamic_opts_t *dopts) {
+	if (tsd_get_allocates() && !imalloc_init_check(sopts, dopts)) {
+		return ENOMEM;
+	}
+
+	/* We always need the tsd.  Let's grab it right away. */
+	tsd_t *tsd = tsd_fetch();
+	assert(tsd);
+	if (likely(tsd_fast(tsd))) {
+		/* Fast and common path. */
+		tsd_assert_fast(tsd);
+		sopts->slow = false;
+		return imalloc_body(sopts, dopts, tsd);
+	} else {
+		if (!tsd_get_allocates() && !imalloc_init_check(sopts, dopts)) {
+			return ENOMEM;
+		}
+
+		sopts->slow = true;
+		return imalloc_body(sopts, dopts, tsd);
+	}
+}
+
+JEMALLOC_NOINLINE
+void *
+malloc_default(size_t size) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.malloc.entry", "size: %zu", size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.null_out_result_on_error = true;
+	sopts.set_errno_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in malloc(): out of memory\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+
+	imalloc(&sopts, &dopts);
+	/*
+	 * Note that this branch gets optimized away -- it immediately follows
+	 * the check on tsd_fast that sets sopts.slow.
+	 */
+	if (sopts.slow) {
+		uintptr_t args[3] = {size};
+		hook_invoke_alloc(hook_alloc_malloc, ret, (uintptr_t)ret, args);
+	}
+
+	LOG("core.malloc.exit", "result: %p", ret);
+
+	return ret;
+}
+
+/******************************************************************************/
+/*
+ * Begin malloc(3)-compatible functions.
+ */
+
+/*
+ * malloc() fastpath.
+ *
+ * Fastpath assumes size <= SC_LOOKUP_MAXCLASS, and that we hit
+ * tcache.  If either of these is false, we tail-call to the slowpath,
+ * malloc_default().  Tail-calling is used to avoid any caller-saved
+ * registers.
+ *
+ * fastpath supports ticker and profiling, both of which will also
+ * tail-call to the slowpath if they fire.
+ */
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1)
+je_malloc(size_t size) {
+	LOG("core.malloc.entry", "size: %zu", size);
+
+	if (tsd_get_allocates() && unlikely(!malloc_initialized())) {
+		return malloc_default(size);
+	}
+
+	tsd_t *tsd = tsd_get(false);
+	if (unlikely(!tsd || !tsd_fast(tsd) || (size > SC_LOOKUP_MAXCLASS))) {
+		return malloc_default(size);
+	}
+
+	tcache_t *tcache = tsd_tcachep_get(tsd);
+
+	if (unlikely(ticker_trytick(&tcache->gc_ticker))) {
+		return malloc_default(size);
+	}
+
+	szind_t ind = sz_size2index_lookup(size);
+	size_t usize;
+	if (config_stats || config_prof) {
+		usize = sz_index2size(ind);
+	}
+	/* Fast path relies on size being a bin. I.e. SC_LOOKUP_MAXCLASS < SC_SMALL_MAXCLASS */
+	assert(ind < SC_NBINS);
+	assert(size <= SC_SMALL_MAXCLASS);
+
+	if (config_prof) {
+		int64_t bytes_until_sample = tsd_bytes_until_sample_get(tsd);
+		bytes_until_sample -= usize;
+		tsd_bytes_until_sample_set(tsd, bytes_until_sample);
+
+		if (unlikely(bytes_until_sample < 0)) {
+			/*
+			 * Avoid a prof_active check on the fastpath.
+			 * If prof_active is false, set bytes_until_sample to
+			 * a large value.  If prof_active is set to true,
+			 * bytes_until_sample will be reset.
+			 */
+			if (!prof_active) {
+				tsd_bytes_until_sample_set(tsd, SSIZE_MAX);
+			}
+			return malloc_default(size);
+		}
+	}
+
+	cache_bin_t *bin = tcache_small_bin_get(tcache, ind);
+	bool tcache_success;
+	void* ret = cache_bin_alloc_easy(bin, &tcache_success);
+
+	if (tcache_success) {
+		if (config_stats) {
+			*tsd_thread_allocatedp_get(tsd) += usize;
+			bin->tstats.nrequests++;
+		}
+		if (config_prof) {
+			tcache->prof_accumbytes += usize;
+		}
+
+		LOG("core.malloc.exit", "result: %p", ret);
+
+		/* Fastpath success */
+		return ret;
+	}
+
+	return malloc_default(size);
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+JEMALLOC_ATTR(nonnull(1))
+je_posix_memalign(void **memptr, size_t alignment, size_t size) {
+	int ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.posix_memalign.entry", "mem ptr: %p, alignment: %zu, "
+	    "size: %zu", memptr, alignment, size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.bump_empty_aligned_alloc = true;
+	sopts.min_alignment = sizeof(void *);
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+
+	dopts.result = memptr;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = alignment;
+
+	ret = imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {(uintptr_t)memptr, (uintptr_t)alignment,
+			(uintptr_t)size};
+		hook_invoke_alloc(hook_alloc_posix_memalign, *memptr,
+		    (uintptr_t)ret, args);
+	}
+
+	LOG("core.posix_memalign.exit", "result: %d, alloc ptr: %p", ret,
+	    *memptr);
+
+	return ret;
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(2)
+je_aligned_alloc(size_t alignment, size_t size) {
+	void *ret;
+
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.aligned_alloc.entry", "alignment: %zu, size: %zu\n",
+	    alignment, size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.bump_empty_aligned_alloc = true;
+	sopts.null_out_result_on_error = true;
+	sopts.set_errno_on_error = true;
+	sopts.min_alignment = 1;
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = alignment;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {(uintptr_t)alignment, (uintptr_t)size};
+		hook_invoke_alloc(hook_alloc_aligned_alloc, ret,
+		    (uintptr_t)ret, args);
+	}
+
+	LOG("core.aligned_alloc.exit", "result: %p", ret);
+
+	return ret;
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE2(1, 2)
+je_calloc(size_t num, size_t size) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.calloc.entry", "num: %zu, size: %zu\n", num, size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.may_overflow = true;
+	sopts.null_out_result_on_error = true;
+	sopts.set_errno_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in calloc(): out of memory\n";
+
+	dopts.result = &ret;
+	dopts.num_items = num;
+	dopts.item_size = size;
+	dopts.zero = true;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {(uintptr_t)num, (uintptr_t)size};
+		hook_invoke_alloc(hook_alloc_calloc, ret, (uintptr_t)ret, args);
+	}
+
+	LOG("core.calloc.exit", "result: %p", ret);
+
+	return ret;
+}
+
+static void *
+irealloc_prof_sample(tsd_t *tsd, void *old_ptr, size_t old_usize, size_t usize,
+    prof_tctx_t *tctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+
+	if (tctx == NULL) {
+		return NULL;
+	}
+	if (usize <= SC_SMALL_MAXCLASS) {
+		p = iralloc(tsd, old_ptr, old_usize,
+		    SC_LARGE_MINCLASS, 0, false, hook_args);
+		if (p == NULL) {
+			return NULL;
+		}
+		arena_prof_promote(tsd_tsdn(tsd), p, usize);
+	} else {
+		p = iralloc(tsd, old_ptr, old_usize, usize, 0, false,
+		    hook_args);
+	}
+
+	return p;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+irealloc_prof(tsd_t *tsd, void *old_ptr, size_t old_usize, size_t usize,
+   alloc_ctx_t *alloc_ctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+	bool prof_active;
+	prof_tctx_t *old_tctx, *tctx;
+
+	prof_active = prof_active_get_unlocked();
+	old_tctx = prof_tctx_get(tsd_tsdn(tsd), old_ptr, alloc_ctx);
+	tctx = prof_alloc_prep(tsd, usize, prof_active, true);
+	if (unlikely((uintptr_t)tctx != (uintptr_t)1U)) {
+		p = irealloc_prof_sample(tsd, old_ptr, old_usize, usize, tctx,
+		    hook_args);
+	} else {
+		p = iralloc(tsd, old_ptr, old_usize, usize, 0, false,
+		    hook_args);
+	}
+	if (unlikely(p == NULL)) {
+		prof_alloc_rollback(tsd, tctx, true);
+		return NULL;
+	}
+	prof_realloc(tsd, p, usize, tctx, prof_active, true, old_ptr, old_usize,
+	    old_tctx);
+
+	return p;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+ifree(tsd_t *tsd, void *ptr, tcache_t *tcache, bool slow_path) {
+	if (!slow_path) {
+		tsd_assert_fast(tsd);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	if (tsd_reentrancy_level_get(tsd) != 0) {
+		assert(slow_path);
+	}
+
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	alloc_ctx_t alloc_ctx;
+	rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+	assert(alloc_ctx.szind != SC_NSIZES);
+
+	size_t usize;
+	if (config_prof && opt_prof) {
+		usize = sz_index2size(alloc_ctx.szind);
+		prof_free(tsd, ptr, usize, &alloc_ctx);
+	} else if (config_stats) {
+		usize = sz_index2size(alloc_ctx.szind);
+	}
+	if (config_stats) {
+		*tsd_thread_deallocatedp_get(tsd) += usize;
+	}
+
+	if (likely(!slow_path)) {
+		idalloctm(tsd_tsdn(tsd), ptr, tcache, &alloc_ctx, false,
+		    false);
+	} else {
+		idalloctm(tsd_tsdn(tsd), ptr, tcache, &alloc_ctx, false,
+		    true);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+isfree(tsd_t *tsd, void *ptr, size_t usize, tcache_t *tcache, bool slow_path) {
+	if (!slow_path) {
+		tsd_assert_fast(tsd);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	if (tsd_reentrancy_level_get(tsd) != 0) {
+		assert(slow_path);
+	}
+
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	alloc_ctx_t alloc_ctx, *ctx;
+	if (!config_cache_oblivious && ((uintptr_t)ptr & PAGE_MASK) != 0) {
+		/*
+		 * When cache_oblivious is disabled and ptr is not page aligned,
+		 * the allocation was not sampled -- usize can be used to
+		 * determine szind directly.
+		 */
+		alloc_ctx.szind = sz_size2index(usize);
+		alloc_ctx.slab = true;
+		ctx = &alloc_ctx;
+		if (config_debug) {
+			alloc_ctx_t dbg_ctx;
+			rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+			rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree,
+			    rtree_ctx, (uintptr_t)ptr, true, &dbg_ctx.szind,
+			    &dbg_ctx.slab);
+			assert(dbg_ctx.szind == alloc_ctx.szind);
+			assert(dbg_ctx.slab == alloc_ctx.slab);
+		}
+	} else if (config_prof && opt_prof) {
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+		assert(alloc_ctx.szind == sz_size2index(usize));
+		ctx = &alloc_ctx;
+	} else {
+		ctx = NULL;
+	}
+
+	if (config_prof && opt_prof) {
+		prof_free(tsd, ptr, usize, ctx);
+	}
+	if (config_stats) {
+		*tsd_thread_deallocatedp_get(tsd) += usize;
+	}
+
+	if (likely(!slow_path)) {
+		isdalloct(tsd_tsdn(tsd), ptr, usize, tcache, ctx, false);
+	} else {
+		isdalloct(tsd_tsdn(tsd), ptr, usize, tcache, ctx, true);
+	}
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ALLOC_SIZE(2)
+je_realloc(void *ptr, size_t arg_size) {
+	void *ret;
+	tsdn_t *tsdn JEMALLOC_CC_SILENCE_INIT(NULL);
+	size_t usize JEMALLOC_CC_SILENCE_INIT(0);
+	size_t old_usize = 0;
+	size_t size = arg_size;
+
+	LOG("core.realloc.entry", "ptr: %p, size: %zu\n", ptr, size);
+
+	if (unlikely(size == 0)) {
+		if (ptr != NULL) {
+			/* realloc(ptr, 0) is equivalent to free(ptr). */
+			UTRACE(ptr, 0, 0);
+			tcache_t *tcache;
+			tsd_t *tsd = tsd_fetch();
+			if (tsd_reentrancy_level_get(tsd) == 0) {
+				tcache = tcache_get(tsd);
+			} else {
+				tcache = NULL;
+			}
+
+			uintptr_t args[3] = {(uintptr_t)ptr, size};
+			hook_invoke_dalloc(hook_dalloc_realloc, ptr, args);
+
+			ifree(tsd, ptr, tcache, true);
+
+			LOG("core.realloc.exit", "result: %p", NULL);
+			return NULL;
+		}
+		size = 1;
+	}
+
+	if (likely(ptr != NULL)) {
+		assert(malloc_initialized() || IS_INITIALIZER);
+		tsd_t *tsd = tsd_fetch();
+
+		check_entry_exit_locking(tsd_tsdn(tsd));
+
+
+		hook_ralloc_args_t hook_args = {true, {(uintptr_t)ptr,
+			(uintptr_t)arg_size, 0, 0}};
+
+		alloc_ctx_t alloc_ctx;
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+		assert(alloc_ctx.szind != SC_NSIZES);
+		old_usize = sz_index2size(alloc_ctx.szind);
+		assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
+		if (config_prof && opt_prof) {
+			usize = sz_s2u(size);
+			if (unlikely(usize == 0
+			    || usize > SC_LARGE_MAXCLASS)) {
+				ret = NULL;
+			} else {
+				ret = irealloc_prof(tsd, ptr, old_usize, usize,
+				    &alloc_ctx, &hook_args);
+			}
+		} else {
+			if (config_stats) {
+				usize = sz_s2u(size);
+			}
+			ret = iralloc(tsd, ptr, old_usize, size, 0, false,
+			    &hook_args);
+		}
+		tsdn = tsd_tsdn(tsd);
+	} else {
+		/* realloc(NULL, size) is equivalent to malloc(size). */
+		static_opts_t sopts;
+		dynamic_opts_t dopts;
+
+		static_opts_init(&sopts);
+		dynamic_opts_init(&dopts);
+
+		sopts.null_out_result_on_error = true;
+		sopts.set_errno_on_error = true;
+		sopts.oom_string =
+		    "<jemalloc>: Error in realloc(): out of memory\n";
+
+		dopts.result = &ret;
+		dopts.num_items = 1;
+		dopts.item_size = size;
+
+		imalloc(&sopts, &dopts);
+		if (sopts.slow) {
+			uintptr_t args[3] = {(uintptr_t)ptr, arg_size};
+			hook_invoke_alloc(hook_alloc_realloc, ret,
+			    (uintptr_t)ret, args);
+		}
+
+		return ret;
+	}
+
+	if (unlikely(ret == NULL)) {
+		if (config_xmalloc && unlikely(opt_xmalloc)) {
+			malloc_write("<jemalloc>: Error in realloc(): "
+			    "out of memory\n");
+			abort();
+		}
+		set_errno(ENOMEM);
+	}
+	if (config_stats && likely(ret != NULL)) {
+		tsd_t *tsd;
+
+		assert(usize == isalloc(tsdn, ret));
+		tsd = tsdn_tsd(tsdn);
+		*tsd_thread_allocatedp_get(tsd) += usize;
+		*tsd_thread_deallocatedp_get(tsd) += old_usize;
+	}
+	UTRACE(ptr, size, ret);
+	check_entry_exit_locking(tsdn);
+
+	LOG("core.realloc.exit", "result: %p", ret);
+	return ret;
+}
+
+JEMALLOC_NOINLINE
+void
+free_default(void *ptr) {
+	UTRACE(ptr, 0, 0);
+	if (likely(ptr != NULL)) {
+		/*
+		 * We avoid setting up tsd fully (e.g. tcache, arena binding)
+		 * based on only free() calls -- other activities trigger the
+		 * minimal to full transition.  This is because free() may
+		 * happen during thread shutdown after tls deallocation: if a
+		 * thread never had any malloc activities until then, a
+		 * fully-setup tsd won't be destructed properly.
+		 */
+		tsd_t *tsd = tsd_fetch_min();
+		check_entry_exit_locking(tsd_tsdn(tsd));
+
+		tcache_t *tcache;
+		if (likely(tsd_fast(tsd))) {
+			tsd_assert_fast(tsd);
+			/* Unconditionally get tcache ptr on fast path. */
+			tcache = tsd_tcachep_get(tsd);
+			ifree(tsd, ptr, tcache, false);
+		} else {
+			if (likely(tsd_reentrancy_level_get(tsd) == 0)) {
+				tcache = tcache_get(tsd);
+			} else {
+				tcache = NULL;
+			}
+			uintptr_t args_raw[3] = {(uintptr_t)ptr};
+			hook_invoke_dalloc(hook_dalloc_free, ptr, args_raw);
+			ifree(tsd, ptr, tcache, true);
+		}
+		check_entry_exit_locking(tsd_tsdn(tsd));
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE
+bool free_fastpath(void *ptr, size_t size, bool size_hint) {
+	tsd_t *tsd = tsd_get(false);
+	if (unlikely(!tsd || !tsd_fast(tsd))) {
+		return false;
+	}
+
+	tcache_t *tcache = tsd_tcachep_get(tsd);
+
+	alloc_ctx_t alloc_ctx;
+	/*
+	 * If !config_cache_oblivious, we can check PAGE alignment to
+	 * detect sampled objects.  Otherwise addresses are
+	 * randomized, and we have to look it up in the rtree anyway.
+	 * See also isfree().
+	 */
+	if (!size_hint || config_cache_oblivious) {
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+		bool res = rtree_szind_slab_read_fast(tsd_tsdn(tsd), &extents_rtree,
+						      rtree_ctx, (uintptr_t)ptr,
+						      &alloc_ctx.szind, &alloc_ctx.slab);
+
+		/* Note: profiled objects will have alloc_ctx.slab set */
+		if (!res || !alloc_ctx.slab) {
+			return false;
+		}
+		assert(alloc_ctx.szind != SC_NSIZES);
+	} else {
+		/*
+		 * Check for both sizes that are too large, and for sampled objects.
+		 * Sampled objects are always page-aligned.  The sampled object check
+		 * will also check for null ptr.
+		 */
+		if (size > SC_LOOKUP_MAXCLASS || (((uintptr_t)ptr & PAGE_MASK) == 0)) {
+			return false;
+		}
+		alloc_ctx.szind = sz_size2index_lookup(size);
+	}
+
+	if (unlikely(ticker_trytick(&tcache->gc_ticker))) {
+		return false;
+	}
+
+	cache_bin_t *bin = tcache_small_bin_get(tcache, alloc_ctx.szind);
+	cache_bin_info_t *bin_info = &tcache_bin_info[alloc_ctx.szind];
+	if (!cache_bin_dalloc_easy(bin, bin_info, ptr)) {
+		return false;
+	}
+
+	if (config_stats) {
+		size_t usize = sz_index2size(alloc_ctx.szind);
+		*tsd_thread_deallocatedp_get(tsd) += usize;
+	}
+
+	return true;
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_free(void *ptr) {
+	LOG("core.free.entry", "ptr: %p", ptr);
+
+	if (!free_fastpath(ptr, 0, false)) {
+		free_default(ptr);
+	}
+
+	LOG("core.free.exit", "");
+}
+
+/*
+ * End malloc(3)-compatible functions.
+ */
+/******************************************************************************/
+/*
+ * Begin non-standard override functions.
+ */
+
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc)
+je_memalign(size_t alignment, size_t size) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.memalign.entry", "alignment: %zu, size: %zu\n", alignment,
+	    size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.min_alignment = 1;
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+	sopts.null_out_result_on_error = true;
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = alignment;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {alignment, size};
+		hook_invoke_alloc(hook_alloc_memalign, ret, (uintptr_t)ret,
+		    args);
+	}
+
+	LOG("core.memalign.exit", "result: %p", ret);
+	return ret;
+}
+#endif
+
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc)
+je_valloc(size_t size) {
+	void *ret;
+
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.valloc.entry", "size: %zu\n", size);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.null_out_result_on_error = true;
+	sopts.min_alignment = PAGE;
+	sopts.oom_string =
+	    "<jemalloc>: Error allocating aligned memory: out of memory\n";
+	sopts.invalid_alignment_string =
+	    "<jemalloc>: Error allocating aligned memory: invalid alignment\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	dopts.alignment = PAGE;
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {size};
+		hook_invoke_alloc(hook_alloc_valloc, ret, (uintptr_t)ret, args);
+	}
+
+	LOG("core.valloc.exit", "result: %p\n", ret);
+	return ret;
+}
+#endif
+
+#if defined(JEMALLOC_IS_MALLOC) && defined(JEMALLOC_GLIBC_MALLOC_HOOK)
+/*
+ * glibc provides the RTLD_DEEPBIND flag for dlopen which can make it possible
+ * to inconsistently reference libc's malloc(3)-compatible functions
+ * (https://bugzilla.mozilla.org/show_bug.cgi?id=493541).
+ *
+ * These definitions interpose hooks in glibc.  The functions are actually
+ * passed an extra argument for the caller return address, which will be
+ * ignored.
+ */
+JEMALLOC_EXPORT void (*__free_hook)(void *ptr) = je_free;
+JEMALLOC_EXPORT void *(*__malloc_hook)(size_t size) = je_malloc;
+JEMALLOC_EXPORT void *(*__realloc_hook)(void *ptr, size_t size) = je_realloc;
+#  ifdef JEMALLOC_GLIBC_MEMALIGN_HOOK
+JEMALLOC_EXPORT void *(*__memalign_hook)(size_t alignment, size_t size) =
+    je_memalign;
+#  endif
+
+#  ifdef CPU_COUNT
+/*
+ * To enable static linking with glibc, the libc specific malloc interface must
+ * be implemented also, so none of glibc's malloc.o functions are added to the
+ * link.
+ */
+#    define ALIAS(je_fn)	__attribute__((alias (#je_fn), used))
+/* To force macro expansion of je_ prefix before stringification. */
+#    define PREALIAS(je_fn)	ALIAS(je_fn)
+#    ifdef JEMALLOC_OVERRIDE___LIBC_CALLOC
+void *__libc_calloc(size_t n, size_t size) PREALIAS(je_calloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_FREE
+void __libc_free(void* ptr) PREALIAS(je_free);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_MALLOC
+void *__libc_malloc(size_t size) PREALIAS(je_malloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_MEMALIGN
+void *__libc_memalign(size_t align, size_t s) PREALIAS(je_memalign);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_REALLOC
+void *__libc_realloc(void* ptr, size_t size) PREALIAS(je_realloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___LIBC_VALLOC
+void *__libc_valloc(size_t size) PREALIAS(je_valloc);
+#    endif
+#    ifdef JEMALLOC_OVERRIDE___POSIX_MEMALIGN
+int __posix_memalign(void** r, size_t a, size_t s) PREALIAS(je_posix_memalign);
+#    endif
+#    undef PREALIAS
+#    undef ALIAS
+#  endif
+#endif
+
+/*
+ * End non-standard override functions.
+ */
+/******************************************************************************/
+/*
+ * Begin non-standard functions.
+ */
+
+#ifdef JEMALLOC_EXPERIMENTAL_SMALLOCX_API
+
+#define JEMALLOC_SMALLOCX_CONCAT_HELPER(x, y) x ## y
+#define JEMALLOC_SMALLOCX_CONCAT_HELPER2(x, y)  \
+  JEMALLOC_SMALLOCX_CONCAT_HELPER(x, y)
+
+typedef struct {
+	void *ptr;
+	size_t size;
+} smallocx_return_t;
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+smallocx_return_t JEMALLOC_NOTHROW
+/*
+ * The attribute JEMALLOC_ATTR(malloc) cannot be used due to:
+ *  - https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86488
+ */
+JEMALLOC_SMALLOCX_CONCAT_HELPER2(je_smallocx_, JEMALLOC_VERSION_GID_IDENT)
+  (size_t size, int flags) {
+	/*
+	 * Note: the attribute JEMALLOC_ALLOC_SIZE(1) cannot be
+	 * used here because it makes writing beyond the `size`
+	 * of the `ptr` undefined behavior, but the objective
+	 * of this function is to allow writing beyond `size`
+	 * up to `smallocx_return_t::size`.
+	 */
+	smallocx_return_t ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.smallocx.entry", "size: %zu, flags: %d", size, flags);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.assert_nonempty_alloc = true;
+	sopts.null_out_result_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in mallocx(): out of memory\n";
+	sopts.usize = true;
+
+	dopts.result = &ret.ptr;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	if (unlikely(flags != 0)) {
+		if ((flags & MALLOCX_LG_ALIGN_MASK) != 0) {
+			dopts.alignment = MALLOCX_ALIGN_GET_SPECIFIED(flags);
+		}
+
+		dopts.zero = MALLOCX_ZERO_GET(flags);
+
+		if ((flags & MALLOCX_TCACHE_MASK) != 0) {
+			if ((flags & MALLOCX_TCACHE_MASK)
+			    == MALLOCX_TCACHE_NONE) {
+				dopts.tcache_ind = TCACHE_IND_NONE;
+			} else {
+				dopts.tcache_ind = MALLOCX_TCACHE_GET(flags);
+			}
+		} else {
+			dopts.tcache_ind = TCACHE_IND_AUTOMATIC;
+		}
+
+		if ((flags & MALLOCX_ARENA_MASK) != 0)
+			dopts.arena_ind = MALLOCX_ARENA_GET(flags);
+	}
+
+	imalloc(&sopts, &dopts);
+	assert(dopts.usize == je_nallocx(size, flags));
+	ret.size = dopts.usize;
+
+	LOG("core.smallocx.exit", "result: %p, size: %zu", ret.ptr, ret.size);
+	return ret;
+}
+#undef JEMALLOC_SMALLOCX_CONCAT_HELPER
+#undef JEMALLOC_SMALLOCX_CONCAT_HELPER2
+#endif
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ATTR(malloc) JEMALLOC_ALLOC_SIZE(1)
+je_mallocx(size_t size, int flags) {
+	void *ret;
+	static_opts_t sopts;
+	dynamic_opts_t dopts;
+
+	LOG("core.mallocx.entry", "size: %zu, flags: %d", size, flags);
+
+	static_opts_init(&sopts);
+	dynamic_opts_init(&dopts);
+
+	sopts.assert_nonempty_alloc = true;
+	sopts.null_out_result_on_error = true;
+	sopts.oom_string = "<jemalloc>: Error in mallocx(): out of memory\n";
+
+	dopts.result = &ret;
+	dopts.num_items = 1;
+	dopts.item_size = size;
+	if (unlikely(flags != 0)) {
+		if ((flags & MALLOCX_LG_ALIGN_MASK) != 0) {
+			dopts.alignment = MALLOCX_ALIGN_GET_SPECIFIED(flags);
+		}
+
+		dopts.zero = MALLOCX_ZERO_GET(flags);
+
+		if ((flags & MALLOCX_TCACHE_MASK) != 0) {
+			if ((flags & MALLOCX_TCACHE_MASK)
+			    == MALLOCX_TCACHE_NONE) {
+				dopts.tcache_ind = TCACHE_IND_NONE;
+			} else {
+				dopts.tcache_ind = MALLOCX_TCACHE_GET(flags);
+			}
+		} else {
+			dopts.tcache_ind = TCACHE_IND_AUTOMATIC;
+		}
+
+		if ((flags & MALLOCX_ARENA_MASK) != 0)
+			dopts.arena_ind = MALLOCX_ARENA_GET(flags);
+	}
+
+	imalloc(&sopts, &dopts);
+	if (sopts.slow) {
+		uintptr_t args[3] = {size, flags};
+		hook_invoke_alloc(hook_alloc_mallocx, ret, (uintptr_t)ret,
+		    args);
+	}
+
+	LOG("core.mallocx.exit", "result: %p", ret);
+	return ret;
+}
+
+static void *
+irallocx_prof_sample(tsdn_t *tsdn, void *old_ptr, size_t old_usize,
+    size_t usize, size_t alignment, bool zero, tcache_t *tcache, arena_t *arena,
+    prof_tctx_t *tctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+
+	if (tctx == NULL) {
+		return NULL;
+	}
+	if (usize <= SC_SMALL_MAXCLASS) {
+		p = iralloct(tsdn, old_ptr, old_usize,
+		    SC_LARGE_MINCLASS, alignment, zero, tcache,
+		    arena, hook_args);
+		if (p == NULL) {
+			return NULL;
+		}
+		arena_prof_promote(tsdn, p, usize);
+	} else {
+		p = iralloct(tsdn, old_ptr, old_usize, usize, alignment, zero,
+		    tcache, arena, hook_args);
+	}
+
+	return p;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+irallocx_prof(tsd_t *tsd, void *old_ptr, size_t old_usize, size_t size,
+    size_t alignment, size_t *usize, bool zero, tcache_t *tcache,
+    arena_t *arena, alloc_ctx_t *alloc_ctx, hook_ralloc_args_t *hook_args) {
+	void *p;
+	bool prof_active;
+	prof_tctx_t *old_tctx, *tctx;
+
+	prof_active = prof_active_get_unlocked();
+	old_tctx = prof_tctx_get(tsd_tsdn(tsd), old_ptr, alloc_ctx);
+	tctx = prof_alloc_prep(tsd, *usize, prof_active, false);
+	if (unlikely((uintptr_t)tctx != (uintptr_t)1U)) {
+		p = irallocx_prof_sample(tsd_tsdn(tsd), old_ptr, old_usize,
+		    *usize, alignment, zero, tcache, arena, tctx, hook_args);
+	} else {
+		p = iralloct(tsd_tsdn(tsd), old_ptr, old_usize, size, alignment,
+		    zero, tcache, arena, hook_args);
+	}
+	if (unlikely(p == NULL)) {
+		prof_alloc_rollback(tsd, tctx, false);
+		return NULL;
+	}
+
+	if (p == old_ptr && alignment != 0) {
+		/*
+		 * The allocation did not move, so it is possible that the size
+		 * class is smaller than would guarantee the requested
+		 * alignment, and that the alignment constraint was
+		 * serendipitously satisfied.  Additionally, old_usize may not
+		 * be the same as the current usize because of in-place large
+		 * reallocation.  Therefore, query the actual value of usize.
+		 */
+		*usize = isalloc(tsd_tsdn(tsd), p);
+	}
+	prof_realloc(tsd, p, *usize, tctx, prof_active, false, old_ptr,
+	    old_usize, old_tctx);
+
+	return p;
+}
+
+JEMALLOC_EXPORT JEMALLOC_ALLOCATOR JEMALLOC_RESTRICT_RETURN
+void JEMALLOC_NOTHROW *
+JEMALLOC_ALLOC_SIZE(2)
+je_rallocx(void *ptr, size_t size, int flags) {
+	void *p;
+	tsd_t *tsd;
+	size_t usize;
+	size_t old_usize;
+	size_t alignment = MALLOCX_ALIGN_GET(flags);
+	bool zero = flags & MALLOCX_ZERO;
+	arena_t *arena;
+	tcache_t *tcache;
+
+	LOG("core.rallocx.entry", "ptr: %p, size: %zu, flags: %d", ptr,
+	    size, flags);
+
+
+	assert(ptr != NULL);
+	assert(size != 0);
+	assert(malloc_initialized() || IS_INITIALIZER);
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	if (unlikely((flags & MALLOCX_ARENA_MASK) != 0)) {
+		unsigned arena_ind = MALLOCX_ARENA_GET(flags);
+		arena = arena_get(tsd_tsdn(tsd), arena_ind, true);
+		if (unlikely(arena == NULL)) {
+			goto label_oom;
+		}
+	} else {
+		arena = NULL;
+	}
+
+	if (unlikely((flags & MALLOCX_TCACHE_MASK) != 0)) {
+		if ((flags & MALLOCX_TCACHE_MASK) == MALLOCX_TCACHE_NONE) {
+			tcache = NULL;
+		} else {
+			tcache = tcaches_get(tsd, MALLOCX_TCACHE_GET(flags));
+		}
+	} else {
+		tcache = tcache_get(tsd);
+	}
+
+	alloc_ctx_t alloc_ctx;
+	rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+	assert(alloc_ctx.szind != SC_NSIZES);
+	old_usize = sz_index2size(alloc_ctx.szind);
+	assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
+
+	hook_ralloc_args_t hook_args = {false, {(uintptr_t)ptr, size, flags,
+		0}};
+	if (config_prof && opt_prof) {
+		usize = (alignment == 0) ?
+		    sz_s2u(size) : sz_sa2u(size, alignment);
+		if (unlikely(usize == 0
+		    || usize > SC_LARGE_MAXCLASS)) {
+			goto label_oom;
+		}
+		p = irallocx_prof(tsd, ptr, old_usize, size, alignment, &usize,
+		    zero, tcache, arena, &alloc_ctx, &hook_args);
+		if (unlikely(p == NULL)) {
+			goto label_oom;
+		}
+	} else {
+		p = iralloct(tsd_tsdn(tsd), ptr, old_usize, size, alignment,
+		    zero, tcache, arena, &hook_args);
+		if (unlikely(p == NULL)) {
+			goto label_oom;
+		}
+		if (config_stats) {
+			usize = isalloc(tsd_tsdn(tsd), p);
+		}
+	}
+	assert(alignment == 0 || ((uintptr_t)p & (alignment - 1)) == ZU(0));
+
+	if (config_stats) {
+		*tsd_thread_allocatedp_get(tsd) += usize;
+		*tsd_thread_deallocatedp_get(tsd) += old_usize;
+	}
+	UTRACE(ptr, size, p);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.rallocx.exit", "result: %p", p);
+	return p;
+label_oom:
+	if (config_xmalloc && unlikely(opt_xmalloc)) {
+		malloc_write("<jemalloc>: Error in rallocx(): out of memory\n");
+		abort();
+	}
+	UTRACE(ptr, size, 0);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.rallocx.exit", "result: %p", NULL);
+	return NULL;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ixallocx_helper(tsdn_t *tsdn, void *ptr, size_t old_usize, size_t size,
+    size_t extra, size_t alignment, bool zero) {
+	size_t newsize;
+
+	if (ixalloc(tsdn, ptr, old_usize, size, extra, alignment, zero,
+	    &newsize)) {
+		return old_usize;
+	}
+
+	return newsize;
+}
+
+static size_t
+ixallocx_prof_sample(tsdn_t *tsdn, void *ptr, size_t old_usize, size_t size,
+    size_t extra, size_t alignment, bool zero, prof_tctx_t *tctx) {
+	size_t usize;
+
+	if (tctx == NULL) {
+		return old_usize;
+	}
+	usize = ixallocx_helper(tsdn, ptr, old_usize, size, extra, alignment,
+	    zero);
+
+	return usize;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ixallocx_prof(tsd_t *tsd, void *ptr, size_t old_usize, size_t size,
+    size_t extra, size_t alignment, bool zero, alloc_ctx_t *alloc_ctx) {
+	size_t usize_max, usize;
+	bool prof_active;
+	prof_tctx_t *old_tctx, *tctx;
+
+	prof_active = prof_active_get_unlocked();
+	old_tctx = prof_tctx_get(tsd_tsdn(tsd), ptr, alloc_ctx);
+	/*
+	 * usize isn't knowable before ixalloc() returns when extra is non-zero.
+	 * Therefore, compute its maximum possible value and use that in
+	 * prof_alloc_prep() to decide whether to capture a backtrace.
+	 * prof_realloc() will use the actual usize to decide whether to sample.
+	 */
+	if (alignment == 0) {
+		usize_max = sz_s2u(size+extra);
+		assert(usize_max > 0
+		    && usize_max <= SC_LARGE_MAXCLASS);
+	} else {
+		usize_max = sz_sa2u(size+extra, alignment);
+		if (unlikely(usize_max == 0
+		    || usize_max > SC_LARGE_MAXCLASS)) {
+			/*
+			 * usize_max is out of range, and chances are that
+			 * allocation will fail, but use the maximum possible
+			 * value and carry on with prof_alloc_prep(), just in
+			 * case allocation succeeds.
+			 */
+			usize_max = SC_LARGE_MAXCLASS;
+		}
+	}
+	tctx = prof_alloc_prep(tsd, usize_max, prof_active, false);
+
+	if (unlikely((uintptr_t)tctx != (uintptr_t)1U)) {
+		usize = ixallocx_prof_sample(tsd_tsdn(tsd), ptr, old_usize,
+		    size, extra, alignment, zero, tctx);
+	} else {
+		usize = ixallocx_helper(tsd_tsdn(tsd), ptr, old_usize, size,
+		    extra, alignment, zero);
+	}
+	if (usize == old_usize) {
+		prof_alloc_rollback(tsd, tctx, false);
+		return usize;
+	}
+	prof_realloc(tsd, ptr, usize, tctx, prof_active, false, ptr, old_usize,
+	    old_tctx);
+
+	return usize;
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+je_xallocx(void *ptr, size_t size, size_t extra, int flags) {
+	tsd_t *tsd;
+	size_t usize, old_usize;
+	size_t alignment = MALLOCX_ALIGN_GET(flags);
+	bool zero = flags & MALLOCX_ZERO;
+
+	LOG("core.xallocx.entry", "ptr: %p, size: %zu, extra: %zu, "
+	    "flags: %d", ptr, size, extra, flags);
+
+	assert(ptr != NULL);
+	assert(size != 0);
+	assert(SIZE_T_MAX - size >= extra);
+	assert(malloc_initialized() || IS_INITIALIZER);
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	alloc_ctx_t alloc_ctx;
+	rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsd);
+	rtree_szind_slab_read(tsd_tsdn(tsd), &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true, &alloc_ctx.szind, &alloc_ctx.slab);
+	assert(alloc_ctx.szind != SC_NSIZES);
+	old_usize = sz_index2size(alloc_ctx.szind);
+	assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
+	/*
+	 * The API explicitly absolves itself of protecting against (size +
+	 * extra) numerical overflow, but we may need to clamp extra to avoid
+	 * exceeding SC_LARGE_MAXCLASS.
+	 *
+	 * Ordinarily, size limit checking is handled deeper down, but here we
+	 * have to check as part of (size + extra) clamping, since we need the
+	 * clamped value in the above helper functions.
+	 */
+	if (unlikely(size > SC_LARGE_MAXCLASS)) {
+		usize = old_usize;
+		goto label_not_resized;
+	}
+	if (unlikely(SC_LARGE_MAXCLASS - size < extra)) {
+		extra = SC_LARGE_MAXCLASS - size;
+	}
+
+	if (config_prof && opt_prof) {
+		usize = ixallocx_prof(tsd, ptr, old_usize, size, extra,
+		    alignment, zero, &alloc_ctx);
+	} else {
+		usize = ixallocx_helper(tsd_tsdn(tsd), ptr, old_usize, size,
+		    extra, alignment, zero);
+	}
+	if (unlikely(usize == old_usize)) {
+		goto label_not_resized;
+	}
+
+	if (config_stats) {
+		*tsd_thread_allocatedp_get(tsd) += usize;
+		*tsd_thread_deallocatedp_get(tsd) += old_usize;
+	}
+label_not_resized:
+	if (unlikely(!tsd_fast(tsd))) {
+		uintptr_t args[4] = {(uintptr_t)ptr, size, extra, flags};
+		hook_invoke_expand(hook_expand_xallocx, ptr, old_usize,
+		    usize, (uintptr_t)usize, args);
+	}
+
+	UTRACE(ptr, size, ptr);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.xallocx.exit", "result: %zu", usize);
+	return usize;
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+JEMALLOC_ATTR(pure)
+je_sallocx(const void *ptr, int flags) {
+	size_t usize;
+	tsdn_t *tsdn;
+
+	LOG("core.sallocx.entry", "ptr: %p, flags: %d", ptr, flags);
+
+	assert(malloc_initialized() || IS_INITIALIZER);
+	assert(ptr != NULL);
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+
+	if (config_debug || force_ivsalloc) {
+		usize = ivsalloc(tsdn, ptr);
+		assert(force_ivsalloc || usize != 0);
+	} else {
+		usize = isalloc(tsdn, ptr);
+	}
+
+	check_entry_exit_locking(tsdn);
+
+	LOG("core.sallocx.exit", "result: %zu", usize);
+	return usize;
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_dallocx(void *ptr, int flags) {
+	LOG("core.dallocx.entry", "ptr: %p, flags: %d", ptr, flags);
+
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	tsd_t *tsd = tsd_fetch();
+	bool fast = tsd_fast(tsd);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	tcache_t *tcache;
+	if (unlikely((flags & MALLOCX_TCACHE_MASK) != 0)) {
+		/* Not allowed to be reentrant and specify a custom tcache. */
+		assert(tsd_reentrancy_level_get(tsd) == 0);
+		if ((flags & MALLOCX_TCACHE_MASK) == MALLOCX_TCACHE_NONE) {
+			tcache = NULL;
+		} else {
+			tcache = tcaches_get(tsd, MALLOCX_TCACHE_GET(flags));
+		}
+	} else {
+		if (likely(fast)) {
+			tcache = tsd_tcachep_get(tsd);
+			assert(tcache == tcache_get(tsd));
+		} else {
+			if (likely(tsd_reentrancy_level_get(tsd) == 0)) {
+				tcache = tcache_get(tsd);
+			}  else {
+				tcache = NULL;
+			}
+		}
+	}
+
+	UTRACE(ptr, 0, 0);
+	if (likely(fast)) {
+		tsd_assert_fast(tsd);
+		ifree(tsd, ptr, tcache, false);
+	} else {
+		uintptr_t args_raw[3] = {(uintptr_t)ptr, flags};
+		hook_invoke_dalloc(hook_dalloc_dallocx, ptr, args_raw);
+		ifree(tsd, ptr, tcache, true);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.dallocx.exit", "");
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+inallocx(tsdn_t *tsdn, size_t size, int flags) {
+	check_entry_exit_locking(tsdn);
+
+	size_t usize;
+	if (likely((flags & MALLOCX_LG_ALIGN_MASK) == 0)) {
+		usize = sz_s2u(size);
+	} else {
+		usize = sz_sa2u(size, MALLOCX_ALIGN_GET_SPECIFIED(flags));
+	}
+	check_entry_exit_locking(tsdn);
+	return usize;
+}
+
+JEMALLOC_NOINLINE void
+sdallocx_default(void *ptr, size_t size, int flags) {
+	assert(ptr != NULL);
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	tsd_t *tsd = tsd_fetch();
+	bool fast = tsd_fast(tsd);
+	size_t usize = inallocx(tsd_tsdn(tsd), size, flags);
+	assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	tcache_t *tcache;
+	if (unlikely((flags & MALLOCX_TCACHE_MASK) != 0)) {
+		/* Not allowed to be reentrant and specify a custom tcache. */
+		assert(tsd_reentrancy_level_get(tsd) == 0);
+		if ((flags & MALLOCX_TCACHE_MASK) == MALLOCX_TCACHE_NONE) {
+			tcache = NULL;
+		} else {
+			tcache = tcaches_get(tsd, MALLOCX_TCACHE_GET(flags));
+		}
+	} else {
+		if (likely(fast)) {
+			tcache = tsd_tcachep_get(tsd);
+			assert(tcache == tcache_get(tsd));
+		} else {
+			if (likely(tsd_reentrancy_level_get(tsd) == 0)) {
+				tcache = tcache_get(tsd);
+			} else {
+				tcache = NULL;
+			}
+		}
+	}
+
+	UTRACE(ptr, 0, 0);
+	if (likely(fast)) {
+		tsd_assert_fast(tsd);
+		isfree(tsd, ptr, usize, tcache, false);
+	} else {
+		uintptr_t args_raw[3] = {(uintptr_t)ptr, size, flags};
+		hook_invoke_dalloc(hook_dalloc_sdallocx, ptr, args_raw);
+		isfree(tsd, ptr, usize, tcache, true);
+	}
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_sdallocx(void *ptr, size_t size, int flags) {
+	LOG("core.sdallocx.entry", "ptr: %p, size: %zu, flags: %d", ptr,
+		size, flags);
+
+	if (flags !=0 || !free_fastpath(ptr, size, true)) {
+		sdallocx_default(ptr, size, flags);
+	}
+
+	LOG("core.sdallocx.exit", "");
+}
+
+void JEMALLOC_NOTHROW
+je_sdallocx_noflags(void *ptr, size_t size) {
+	LOG("core.sdallocx.entry", "ptr: %p, size: %zu, flags: 0", ptr,
+		size);
+
+	if (!free_fastpath(ptr, size, true)) {
+		sdallocx_default(ptr, size, 0);
+	}
+
+	LOG("core.sdallocx.exit", "");
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+JEMALLOC_ATTR(pure)
+je_nallocx(size_t size, int flags) {
+	size_t usize;
+	tsdn_t *tsdn;
+
+	assert(size != 0);
+
+	if (unlikely(malloc_init())) {
+		LOG("core.nallocx.exit", "result: %zu", ZU(0));
+		return 0;
+	}
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+
+	usize = inallocx(tsdn, size, flags);
+	if (unlikely(usize > SC_LARGE_MAXCLASS)) {
+		LOG("core.nallocx.exit", "result: %zu", ZU(0));
+		return 0;
+	}
+
+	check_entry_exit_locking(tsdn);
+	LOG("core.nallocx.exit", "result: %zu", usize);
+	return usize;
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+je_mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen) {
+	int ret;
+	tsd_t *tsd;
+
+	LOG("core.mallctl.entry", "name: %s", name);
+
+	if (unlikely(malloc_init())) {
+		LOG("core.mallctl.exit", "result: %d", EAGAIN);
+		return EAGAIN;
+	}
+
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	ret = ctl_byname(tsd, name, oldp, oldlenp, newp, newlen);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.mallctl.exit", "result: %d", ret);
+	return ret;
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+je_mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp) {
+	int ret;
+
+	LOG("core.mallctlnametomib.entry", "name: %s", name);
+
+	if (unlikely(malloc_init())) {
+		LOG("core.mallctlnametomib.exit", "result: %d", EAGAIN);
+		return EAGAIN;
+	}
+
+	tsd_t *tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	ret = ctl_nametomib(tsd, name, mibp, miblenp);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+
+	LOG("core.mallctlnametomib.exit", "result: %d", ret);
+	return ret;
+}
+
+JEMALLOC_EXPORT int JEMALLOC_NOTHROW
+je_mallctlbymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+  void *newp, size_t newlen) {
+	int ret;
+	tsd_t *tsd;
+
+	LOG("core.mallctlbymib.entry", "");
+
+	if (unlikely(malloc_init())) {
+		LOG("core.mallctlbymib.exit", "result: %d", EAGAIN);
+		return EAGAIN;
+	}
+
+	tsd = tsd_fetch();
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	ret = ctl_bymib(tsd, mib, miblen, oldp, oldlenp, newp, newlen);
+	check_entry_exit_locking(tsd_tsdn(tsd));
+	LOG("core.mallctlbymib.exit", "result: %d", ret);
+	return ret;
+}
+
+JEMALLOC_EXPORT void JEMALLOC_NOTHROW
+je_malloc_stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts) {
+	tsdn_t *tsdn;
+
+	LOG("core.malloc_stats_print.entry", "");
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+	stats_print(write_cb, cbopaque, opts);
+	check_entry_exit_locking(tsdn);
+	LOG("core.malloc_stats_print.exit", "");
+}
+
+JEMALLOC_EXPORT size_t JEMALLOC_NOTHROW
+je_malloc_usable_size(JEMALLOC_USABLE_SIZE_CONST void *ptr) {
+	size_t ret;
+	tsdn_t *tsdn;
+
+	LOG("core.malloc_usable_size.entry", "ptr: %p", ptr);
+
+	assert(malloc_initialized() || IS_INITIALIZER);
+
+	tsdn = tsdn_fetch();
+	check_entry_exit_locking(tsdn);
+
+	if (unlikely(ptr == NULL)) {
+		ret = 0;
+	} else {
+		if (config_debug || force_ivsalloc) {
+			ret = ivsalloc(tsdn, ptr);
+			assert(force_ivsalloc || ret != 0);
+		} else {
+			ret = isalloc(tsdn, ptr);
+		}
+	}
+
+	check_entry_exit_locking(tsdn);
+	LOG("core.malloc_usable_size.exit", "result: %zu", ret);
+	return ret;
+}
+
+/*
+ * End non-standard functions.
+ */
+/******************************************************************************/
+/*
+ * The following functions are used by threading libraries for protection of
+ * malloc during fork().
+ */
+
+/*
+ * If an application creates a thread before doing any allocation in the main
+ * thread, then calls fork(2) in the main thread followed by memory allocation
+ * in the child process, a race can occur that results in deadlock within the
+ * child: the main thread may have forked while the created thread had
+ * partially initialized the allocator.  Ordinarily jemalloc prevents
+ * fork/malloc races via the following functions it registers during
+ * initialization using pthread_atfork(), but of course that does no good if
+ * the allocator isn't fully initialized at fork time.  The following library
+ * constructor is a partial solution to this problem.  It may still be possible
+ * to trigger the deadlock described above, but doing so would involve forking
+ * via a library constructor that runs before jemalloc's runs.
+ */
+#ifndef JEMALLOC_JET
+JEMALLOC_ATTR(constructor)
+static void
+jemalloc_constructor(void) {
+	malloc_init();
+}
+#endif
+
+#ifndef JEMALLOC_MUTEX_INIT_CB
+void
+jemalloc_prefork(void)
+#else
+JEMALLOC_EXPORT void
+_malloc_prefork(void)
+#endif
+{
+	tsd_t *tsd;
+	unsigned i, j, narenas;
+	arena_t *arena;
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	if (!malloc_initialized()) {
+		return;
+	}
+#endif
+	assert(malloc_initialized());
+
+	tsd = tsd_fetch();
+
+	narenas = narenas_total_get();
+
+	witness_prefork(tsd_witness_tsdp_get(tsd));
+	/* Acquire all mutexes in a safe order. */
+	ctl_prefork(tsd_tsdn(tsd));
+	tcache_prefork(tsd_tsdn(tsd));
+	malloc_mutex_prefork(tsd_tsdn(tsd), &arenas_lock);
+	if (have_background_thread) {
+		background_thread_prefork0(tsd_tsdn(tsd));
+	}
+	prof_prefork0(tsd_tsdn(tsd));
+	if (have_background_thread) {
+		background_thread_prefork1(tsd_tsdn(tsd));
+	}
+	/* Break arena prefork into stages to preserve lock order. */
+	for (i = 0; i < 8; i++) {
+		for (j = 0; j < narenas; j++) {
+			if ((arena = arena_get(tsd_tsdn(tsd), j, false)) !=
+			    NULL) {
+				switch (i) {
+				case 0:
+					arena_prefork0(tsd_tsdn(tsd), arena);
+					break;
+				case 1:
+					arena_prefork1(tsd_tsdn(tsd), arena);
+					break;
+				case 2:
+					arena_prefork2(tsd_tsdn(tsd), arena);
+					break;
+				case 3:
+					arena_prefork3(tsd_tsdn(tsd), arena);
+					break;
+				case 4:
+					arena_prefork4(tsd_tsdn(tsd), arena);
+					break;
+				case 5:
+					arena_prefork5(tsd_tsdn(tsd), arena);
+					break;
+				case 6:
+					arena_prefork6(tsd_tsdn(tsd), arena);
+					break;
+				case 7:
+					arena_prefork7(tsd_tsdn(tsd), arena);
+					break;
+				default: not_reached();
+				}
+			}
+		}
+	}
+	prof_prefork1(tsd_tsdn(tsd));
+	tsd_prefork(tsd);
+}
+
+#ifndef JEMALLOC_MUTEX_INIT_CB
+void
+jemalloc_postfork_parent(void)
+#else
+JEMALLOC_EXPORT void
+_malloc_postfork(void)
+#endif
+{
+	tsd_t *tsd;
+	unsigned i, narenas;
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	if (!malloc_initialized()) {
+		return;
+	}
+#endif
+	assert(malloc_initialized());
+
+	tsd = tsd_fetch();
+
+	tsd_postfork_parent(tsd);
+
+	witness_postfork_parent(tsd_witness_tsdp_get(tsd));
+	/* Release all mutexes, now that fork() has completed. */
+	for (i = 0, narenas = narenas_total_get(); i < narenas; i++) {
+		arena_t *arena;
+
+		if ((arena = arena_get(tsd_tsdn(tsd), i, false)) != NULL) {
+			arena_postfork_parent(tsd_tsdn(tsd), arena);
+		}
+	}
+	prof_postfork_parent(tsd_tsdn(tsd));
+	if (have_background_thread) {
+		background_thread_postfork_parent(tsd_tsdn(tsd));
+	}
+	malloc_mutex_postfork_parent(tsd_tsdn(tsd), &arenas_lock);
+	tcache_postfork_parent(tsd_tsdn(tsd));
+	ctl_postfork_parent(tsd_tsdn(tsd));
+}
+
+void
+jemalloc_postfork_child(void) {
+	tsd_t *tsd;
+	unsigned i, narenas;
+
+	assert(malloc_initialized());
+
+	tsd = tsd_fetch();
+
+	tsd_postfork_child(tsd);
+
+	witness_postfork_child(tsd_witness_tsdp_get(tsd));
+	/* Release all mutexes, now that fork() has completed. */
+	for (i = 0, narenas = narenas_total_get(); i < narenas; i++) {
+		arena_t *arena;
+
+		if ((arena = arena_get(tsd_tsdn(tsd), i, false)) != NULL) {
+			arena_postfork_child(tsd_tsdn(tsd), arena);
+		}
+	}
+	prof_postfork_child(tsd_tsdn(tsd));
+	if (have_background_thread) {
+		background_thread_postfork_child(tsd_tsdn(tsd));
+	}
+	malloc_mutex_postfork_child(tsd_tsdn(tsd), &arenas_lock);
+	tcache_postfork_child(tsd_tsdn(tsd));
+	ctl_postfork_child(tsd_tsdn(tsd));
+}
+
+/******************************************************************************/
diff --git a/memkind-1.10.0/jemalloc/src/jemalloc_cpp.cpp b/memkind-1.10.0/jemalloc/src/jemalloc_cpp.cpp
new file mode 100644
index 0000000..da0441a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/jemalloc_cpp.cpp
@@ -0,0 +1,141 @@
+#include <mutex>
+#include <new>
+
+#define JEMALLOC_CPP_CPP_
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#ifdef __cplusplus
+}
+#endif
+
+// All operators in this file are exported.
+
+// Possibly alias hidden versions of malloc and sdallocx to avoid an extra plt
+// thunk?
+//
+// extern __typeof (sdallocx) sdallocx_int
+//  __attribute ((alias ("sdallocx"),
+//		visibility ("hidden")));
+//
+// ... but it needs to work with jemalloc namespaces.
+
+void	*operator new(std::size_t size);
+void	*operator new[](std::size_t size);
+void	*operator new(std::size_t size, const std::nothrow_t &) noexcept;
+void	*operator new[](std::size_t size, const std::nothrow_t &) noexcept;
+void	operator delete(void *ptr) noexcept;
+void	operator delete[](void *ptr) noexcept;
+void	operator delete(void *ptr, const std::nothrow_t &) noexcept;
+void	operator delete[](void *ptr, const std::nothrow_t &) noexcept;
+
+#if __cpp_sized_deallocation >= 201309
+/* C++14's sized-delete operators. */
+void	operator delete(void *ptr, std::size_t size) noexcept;
+void	operator delete[](void *ptr, std::size_t size) noexcept;
+#endif
+
+JEMALLOC_NOINLINE
+static void *
+handleOOM(std::size_t size, bool nothrow) {
+	void *ptr = nullptr;
+
+	while (ptr == nullptr) {
+		std::new_handler handler;
+		// GCC-4.8 and clang 4.0 do not have std::get_new_handler.
+		{
+			static std::mutex mtx;
+			std::lock_guard<std::mutex> lock(mtx);
+
+			handler = std::set_new_handler(nullptr);
+			std::set_new_handler(handler);
+		}
+		if (handler == nullptr)
+			break;
+
+		try {
+			handler();
+		} catch (const std::bad_alloc &) {
+			break;
+		}
+
+		ptr = je_malloc(size);
+	}
+
+	if (ptr == nullptr && !nothrow)
+		std::__throw_bad_alloc();
+	return ptr;
+}
+
+template <bool IsNoExcept>
+JEMALLOC_ALWAYS_INLINE
+void *
+newImpl(std::size_t size) noexcept(IsNoExcept) {
+	void *ptr = je_malloc(size);
+	if (likely(ptr != nullptr))
+		return ptr;
+
+	return handleOOM(size, IsNoExcept);
+}
+
+void *
+operator new(std::size_t size) {
+	return newImpl<false>(size);
+}
+
+void *
+operator new[](std::size_t size) {
+	return newImpl<false>(size);
+}
+
+void *
+operator new(std::size_t size, const std::nothrow_t &) noexcept {
+	return newImpl<true>(size);
+}
+
+void *
+operator new[](std::size_t size, const std::nothrow_t &) noexcept {
+	return newImpl<true>(size);
+}
+
+void
+operator delete(void *ptr) noexcept {
+	je_free(ptr);
+}
+
+void
+operator delete[](void *ptr) noexcept {
+	je_free(ptr);
+}
+
+void
+operator delete(void *ptr, const std::nothrow_t &) noexcept {
+	je_free(ptr);
+}
+
+void operator delete[](void *ptr, const std::nothrow_t &) noexcept {
+	je_free(ptr);
+}
+
+#if __cpp_sized_deallocation >= 201309
+
+void
+operator delete(void *ptr, std::size_t size) noexcept {
+	if (unlikely(ptr == nullptr)) {
+		return;
+	}
+	je_sdallocx_noflags(ptr, size);
+}
+
+void operator delete[](void *ptr, std::size_t size) noexcept {
+	if (unlikely(ptr == nullptr)) {
+		return;
+	}
+	je_sdallocx_noflags(ptr, size);
+}
+
+#endif  // __cpp_sized_deallocation
diff --git a/memkind-1.10.0/jemalloc/src/large.c b/memkind-1.10.0/jemalloc/src/large.c
new file mode 100644
index 0000000..8e7a781
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/large.c
@@ -0,0 +1,395 @@
+#define JEMALLOC_LARGE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/util.h"
+
+/******************************************************************************/
+
+void *
+large_malloc(tsdn_t *tsdn, arena_t *arena, size_t usize, bool zero) {
+	assert(usize == sz_s2u(usize));
+
+	return large_palloc(tsdn, arena, usize, CACHELINE, zero);
+}
+
+void *
+large_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
+    bool zero) {
+	size_t ausize;
+	extent_t *extent;
+	bool is_zeroed;
+	UNUSED bool idump JEMALLOC_CC_SILENCE_INIT(false);
+
+	assert(!tsdn_null(tsdn) || arena != NULL);
+
+	ausize = sz_sa2u(usize, alignment);
+	if (unlikely(ausize == 0 || ausize > SC_LARGE_MAXCLASS)) {
+		return NULL;
+	}
+
+	if (config_fill && unlikely(opt_zero)) {
+		zero = true;
+	}
+	/*
+	 * Copy zero into is_zeroed and pass the copy when allocating the
+	 * extent, so that it is possible to make correct junk/zero fill
+	 * decisions below, even if is_zeroed ends up true when zero is false.
+	 */
+	is_zeroed = zero;
+	if (likely(!tsdn_null(tsdn))) {
+		arena = arena_choose_maybe_huge(tsdn_tsd(tsdn), arena, usize);
+	}
+	if (unlikely(arena == NULL) || (extent = arena_extent_alloc_large(tsdn,
+	    arena, usize, alignment, &is_zeroed)) == NULL) {
+		return NULL;
+	}
+
+	/* See comments in arena_bin_slabs_full_insert(). */
+	if (!arena_is_auto(arena)) {
+		/* Insert extent into large. */
+		malloc_mutex_lock(tsdn, &arena->large_mtx);
+		extent_list_append(&arena->large, extent);
+		malloc_mutex_unlock(tsdn, &arena->large_mtx);
+	}
+	if (config_prof && arena_prof_accum(tsdn, arena, usize)) {
+		prof_idump(tsdn);
+	}
+
+	if (zero) {
+		assert(is_zeroed);
+	} else if (config_fill && unlikely(opt_junk_alloc)) {
+		memset(extent_addr_get(extent), JEMALLOC_ALLOC_JUNK,
+		    extent_usize_get(extent));
+	}
+
+	arena_decay_tick(tsdn, arena);
+	return extent_addr_get(extent);
+}
+
+static void
+large_dalloc_junk_impl(void *ptr, size_t size) {
+	memset(ptr, JEMALLOC_FREE_JUNK, size);
+}
+large_dalloc_junk_t *JET_MUTABLE large_dalloc_junk = large_dalloc_junk_impl;
+
+static void
+large_dalloc_maybe_junk_impl(void *ptr, size_t size) {
+	if (config_fill && have_dss && unlikely(opt_junk_free)) {
+		/*
+		 * Only bother junk filling if the extent isn't about to be
+		 * unmapped.
+		 */
+		if (opt_retain || (have_dss && extent_in_dss(ptr))) {
+			large_dalloc_junk(ptr, size);
+		}
+	}
+}
+large_dalloc_maybe_junk_t *JET_MUTABLE large_dalloc_maybe_junk =
+    large_dalloc_maybe_junk_impl;
+
+static bool
+large_ralloc_no_move_shrink(tsdn_t *tsdn, extent_t *extent, size_t usize) {
+	arena_t *arena = extent_arena_get(extent);
+	size_t oldusize = extent_usize_get(extent);
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+	size_t diff = extent_size_get(extent) - (usize + sz_large_pad);
+
+	assert(oldusize > usize);
+
+	if (extent_hooks->split == NULL) {
+		return true;
+	}
+
+	/* Split excess pages. */
+	if (diff != 0) {
+		extent_t *trail = extent_split_wrapper(tsdn, arena,
+		    &extent_hooks, extent, usize + sz_large_pad,
+		    sz_size2index(usize), false, diff, SC_NSIZES, false);
+		if (trail == NULL) {
+			return true;
+		}
+
+		if (config_fill && unlikely(opt_junk_free)) {
+			large_dalloc_maybe_junk(extent_addr_get(trail),
+			    extent_size_get(trail));
+		}
+
+		arena_extents_dirty_dalloc(tsdn, arena, &extent_hooks, trail);
+	}
+
+	arena_extent_ralloc_large_shrink(tsdn, arena, extent, oldusize);
+
+	return false;
+}
+
+static bool
+large_ralloc_no_move_expand(tsdn_t *tsdn, extent_t *extent, size_t usize,
+    bool zero) {
+	arena_t *arena = extent_arena_get(extent);
+	size_t oldusize = extent_usize_get(extent);
+	extent_hooks_t *extent_hooks = extent_hooks_get(arena);
+	size_t trailsize = usize - oldusize;
+
+	if (extent_hooks->merge == NULL) {
+		return true;
+	}
+
+	if (config_fill && unlikely(opt_zero)) {
+		zero = true;
+	}
+	/*
+	 * Copy zero into is_zeroed_trail and pass the copy when allocating the
+	 * extent, so that it is possible to make correct junk/zero fill
+	 * decisions below, even if is_zeroed_trail ends up true when zero is
+	 * false.
+	 */
+	bool is_zeroed_trail = zero;
+	bool commit = true;
+	extent_t *trail;
+	bool new_mapping;
+	if ((trail = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_dirty, extent_past_get(extent), trailsize, 0,
+	    CACHELINE, false, SC_NSIZES, &is_zeroed_trail, &commit)) != NULL
+	    || (trail = extents_alloc(tsdn, arena, &extent_hooks,
+	    &arena->extents_muzzy, extent_past_get(extent), trailsize, 0,
+	    CACHELINE, false, SC_NSIZES, &is_zeroed_trail, &commit)) != NULL) {
+		if (config_stats) {
+			new_mapping = false;
+		}
+	} else {
+		if ((trail = extent_alloc_wrapper(tsdn, arena, &extent_hooks,
+		    extent_past_get(extent), trailsize, 0, CACHELINE, false,
+		    SC_NSIZES, &is_zeroed_trail, &commit)) == NULL) {
+			return true;
+		}
+		if (config_stats) {
+			new_mapping = true;
+		}
+	}
+
+	if (extent_merge_wrapper(tsdn, arena, &extent_hooks, extent, trail)) {
+		extent_dalloc_wrapper(tsdn, arena, &extent_hooks, trail);
+		return true;
+	}
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+	szind_t szind = sz_size2index(usize);
+	extent_szind_set(extent, szind);
+	rtree_szind_slab_update(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)extent_addr_get(extent), szind, false);
+
+	if (config_stats && new_mapping) {
+		arena_stats_mapped_add(tsdn, &arena->stats, trailsize);
+	}
+
+	if (zero) {
+		if (config_cache_oblivious) {
+			/*
+			 * Zero the trailing bytes of the original allocation's
+			 * last page, since they are in an indeterminate state.
+			 * There will always be trailing bytes, because ptr's
+			 * offset from the beginning of the extent is a multiple
+			 * of CACHELINE in [0 .. PAGE).
+			 */
+			void *zbase = (void *)
+			    ((uintptr_t)extent_addr_get(extent) + oldusize);
+			void *zpast = PAGE_ADDR2BASE((void *)((uintptr_t)zbase +
+			    PAGE));
+			size_t nzero = (uintptr_t)zpast - (uintptr_t)zbase;
+			assert(nzero > 0);
+			memset(zbase, 0, nzero);
+		}
+		assert(is_zeroed_trail);
+	} else if (config_fill && unlikely(opt_junk_alloc)) {
+		memset((void *)((uintptr_t)extent_addr_get(extent) + oldusize),
+		    JEMALLOC_ALLOC_JUNK, usize - oldusize);
+	}
+
+	arena_extent_ralloc_large_expand(tsdn, arena, extent, oldusize);
+
+	return false;
+}
+
+bool
+large_ralloc_no_move(tsdn_t *tsdn, extent_t *extent, size_t usize_min,
+    size_t usize_max, bool zero) {
+	size_t oldusize = extent_usize_get(extent);
+
+	/* The following should have been caught by callers. */
+	assert(usize_min > 0 && usize_max <= SC_LARGE_MAXCLASS);
+	/* Both allocation sizes must be large to avoid a move. */
+	assert(oldusize >= SC_LARGE_MINCLASS
+	    && usize_max >= SC_LARGE_MINCLASS);
+
+	if (usize_max > oldusize) {
+		/* Attempt to expand the allocation in-place. */
+		if (!large_ralloc_no_move_expand(tsdn, extent, usize_max,
+		    zero)) {
+			arena_decay_tick(tsdn, extent_arena_get(extent));
+			return false;
+		}
+		/* Try again, this time with usize_min. */
+		if (usize_min < usize_max && usize_min > oldusize &&
+		    large_ralloc_no_move_expand(tsdn, extent, usize_min,
+		    zero)) {
+			arena_decay_tick(tsdn, extent_arena_get(extent));
+			return false;
+		}
+	}
+
+	/*
+	 * Avoid moving the allocation if the existing extent size accommodates
+	 * the new size.
+	 */
+	if (oldusize >= usize_min && oldusize <= usize_max) {
+		arena_decay_tick(tsdn, extent_arena_get(extent));
+		return false;
+	}
+
+	/* Attempt to shrink the allocation in-place. */
+	if (oldusize > usize_max) {
+		if (!large_ralloc_no_move_shrink(tsdn, extent, usize_max)) {
+			arena_decay_tick(tsdn, extent_arena_get(extent));
+			return false;
+		}
+	}
+	return true;
+}
+
+static void *
+large_ralloc_move_helper(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool zero) {
+	if (alignment <= CACHELINE) {
+		return large_malloc(tsdn, arena, usize, zero);
+	}
+	return large_palloc(tsdn, arena, usize, alignment, zero);
+}
+
+void *
+large_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache,
+    hook_ralloc_args_t *hook_args) {
+	extent_t *extent = iealloc(tsdn, ptr);
+
+	size_t oldusize = extent_usize_get(extent);
+	/* The following should have been caught by callers. */
+	assert(usize > 0 && usize <= SC_LARGE_MAXCLASS);
+	/* Both allocation sizes must be large to avoid a move. */
+	assert(oldusize >= SC_LARGE_MINCLASS
+	    && usize >= SC_LARGE_MINCLASS);
+
+	/* Try to avoid moving the allocation. */
+	if (!large_ralloc_no_move(tsdn, extent, usize, usize, zero)) {
+		hook_invoke_expand(hook_args->is_realloc
+		    ? hook_expand_realloc : hook_expand_rallocx, ptr, oldusize,
+		    usize, (uintptr_t)ptr, hook_args->args);
+		return extent_addr_get(extent);
+	}
+
+	/*
+	 * usize and old size are different enough that we need to use a
+	 * different size class.  In that case, fall back to allocating new
+	 * space and copying.
+	 */
+	void *ret = large_ralloc_move_helper(tsdn, arena, usize, alignment,
+	    zero);
+	if (ret == NULL) {
+		return NULL;
+	}
+
+	hook_invoke_alloc(hook_args->is_realloc
+	    ? hook_alloc_realloc : hook_alloc_rallocx, ret, (uintptr_t)ret,
+	    hook_args->args);
+	hook_invoke_dalloc(hook_args->is_realloc
+	    ? hook_dalloc_realloc : hook_dalloc_rallocx, ptr, hook_args->args);
+
+	size_t copysize = (usize < oldusize) ? usize : oldusize;
+	memcpy(ret, extent_addr_get(extent), copysize);
+	isdalloct(tsdn, extent_addr_get(extent), oldusize, tcache, NULL, true);
+	return ret;
+}
+
+/*
+ * junked_locked indicates whether the extent's data have been junk-filled, and
+ * whether the arena's large_mtx is currently held.
+ */
+static void
+large_dalloc_prep_impl(tsdn_t *tsdn, arena_t *arena, extent_t *extent,
+    bool junked_locked) {
+	if (!junked_locked) {
+		/* See comments in arena_bin_slabs_full_insert(). */
+		if (!arena_is_auto(arena)) {
+			malloc_mutex_lock(tsdn, &arena->large_mtx);
+			extent_list_remove(&arena->large, extent);
+			malloc_mutex_unlock(tsdn, &arena->large_mtx);
+		}
+		large_dalloc_maybe_junk(extent_addr_get(extent),
+		    extent_usize_get(extent));
+	} else {
+		/* Only hold the large_mtx if necessary. */
+		if (!arena_is_auto(arena)) {
+			malloc_mutex_assert_owner(tsdn, &arena->large_mtx);
+			extent_list_remove(&arena->large, extent);
+		}
+	}
+	arena_extent_dalloc_large_prep(tsdn, arena, extent);
+}
+
+static void
+large_dalloc_finish_impl(tsdn_t *tsdn, arena_t *arena, extent_t *extent) {
+	extent_hooks_t *extent_hooks = EXTENT_HOOKS_INITIALIZER;
+	arena_extents_dirty_dalloc(tsdn, arena, &extent_hooks, extent);
+}
+
+void
+large_dalloc_prep_junked_locked(tsdn_t *tsdn, extent_t *extent) {
+	large_dalloc_prep_impl(tsdn, extent_arena_get(extent), extent, true);
+}
+
+void
+large_dalloc_finish(tsdn_t *tsdn, extent_t *extent) {
+	large_dalloc_finish_impl(tsdn, extent_arena_get(extent), extent);
+}
+
+void
+large_dalloc(tsdn_t *tsdn, extent_t *extent) {
+	arena_t *arena = extent_arena_get(extent);
+	large_dalloc_prep_impl(tsdn, arena, extent, false);
+	large_dalloc_finish_impl(tsdn, arena, extent);
+	arena_decay_tick(tsdn, arena);
+}
+
+size_t
+large_salloc(tsdn_t *tsdn, const extent_t *extent) {
+	return extent_usize_get(extent);
+}
+
+prof_tctx_t *
+large_prof_tctx_get(tsdn_t *tsdn, const extent_t *extent) {
+	return extent_prof_tctx_get(extent);
+}
+
+void
+large_prof_tctx_set(tsdn_t *tsdn, extent_t *extent, prof_tctx_t *tctx) {
+	extent_prof_tctx_set(extent, tctx);
+}
+
+void
+large_prof_tctx_reset(tsdn_t *tsdn, extent_t *extent) {
+	large_prof_tctx_set(tsdn, extent, (prof_tctx_t *)(uintptr_t)1U);
+}
+
+nstime_t
+large_prof_alloc_time_get(const extent_t *extent) {
+	return extent_prof_alloc_time_get(extent);
+}
+
+void
+large_prof_alloc_time_set(extent_t *extent, nstime_t t) {
+	extent_prof_alloc_time_set(extent, t);
+}
diff --git a/memkind-1.10.0/jemalloc/src/log.c b/memkind-1.10.0/jemalloc/src/log.c
new file mode 100644
index 0000000..778902f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/log.c
@@ -0,0 +1,78 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/log.h"
+
+char log_var_names[JEMALLOC_LOG_VAR_BUFSIZE];
+atomic_b_t log_init_done = ATOMIC_INIT(false);
+
+/*
+ * Returns true if we were able to pick out a segment.  Fills in r_segment_end
+ * with a pointer to the first character after the end of the string.
+ */
+static const char *
+log_var_extract_segment(const char* segment_begin) {
+	const char *end;
+	for (end = segment_begin; *end != '\0' && *end != '|'; end++) {
+	}
+	return end;
+}
+
+static bool
+log_var_matches_segment(const char *segment_begin, const char *segment_end,
+    const char *log_var_begin, const char *log_var_end) {
+	assert(segment_begin <= segment_end);
+	assert(log_var_begin < log_var_end);
+
+	ptrdiff_t segment_len = segment_end - segment_begin;
+	ptrdiff_t log_var_len = log_var_end - log_var_begin;
+	/* The special '.' segment matches everything. */
+	if (segment_len == 1 && *segment_begin == '.') {
+		return true;
+	}
+        if (segment_len == log_var_len) {
+		return strncmp(segment_begin, log_var_begin, segment_len) == 0;
+	} else if (segment_len < log_var_len) {
+		return strncmp(segment_begin, log_var_begin, segment_len) == 0
+		    && log_var_begin[segment_len] == '.';
+        } else {
+		return false;
+	}
+}
+
+unsigned
+log_var_update_state(log_var_t *log_var) {
+	const char *log_var_begin = log_var->name;
+	const char *log_var_end = log_var->name + strlen(log_var->name);
+
+	/* Pointer to one before the beginning of the current segment. */
+	const char *segment_begin = log_var_names;
+
+	/*
+	 * If log_init done is false, we haven't parsed the malloc conf yet.  To
+	 * avoid log-spew, we default to not displaying anything.
+	 */
+	if (!atomic_load_b(&log_init_done, ATOMIC_ACQUIRE)) {
+		return LOG_INITIALIZED_NOT_ENABLED;
+	}
+
+	while (true) {
+		const char *segment_end = log_var_extract_segment(
+		    segment_begin);
+		assert(segment_end < log_var_names + JEMALLOC_LOG_VAR_BUFSIZE);
+		if (log_var_matches_segment(segment_begin, segment_end,
+		    log_var_begin, log_var_end)) {
+			atomic_store_u(&log_var->state, LOG_ENABLED,
+			    ATOMIC_RELAXED);
+			return LOG_ENABLED;
+		}
+		if (*segment_end == '\0') {
+			/* Hit the end of the segment string with no match. */
+			atomic_store_u(&log_var->state,
+			    LOG_INITIALIZED_NOT_ENABLED, ATOMIC_RELAXED);
+			return LOG_INITIALIZED_NOT_ENABLED;
+		}
+		/* Otherwise, skip the delimiter and continue. */
+		segment_begin = segment_end + 1;
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/src/malloc_io.c b/memkind-1.10.0/jemalloc/src/malloc_io.c
new file mode 100644
index 0000000..d7cb0f5
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/malloc_io.c
@@ -0,0 +1,675 @@
+#define JEMALLOC_MALLOC_IO_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/util.h"
+
+#ifdef assert
+#  undef assert
+#endif
+#ifdef not_reached
+#  undef not_reached
+#endif
+#ifdef not_implemented
+#  undef not_implemented
+#endif
+#ifdef assert_not_implemented
+#  undef assert_not_implemented
+#endif
+
+/*
+ * Define simple versions of assertion macros that won't recurse in case
+ * of assertion failures in malloc_*printf().
+ */
+#define assert(e) do {							\
+	if (config_debug && !(e)) {					\
+		malloc_write("<jemalloc>: Failed assertion\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define not_reached() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Unreachable code reached\n");	\
+		abort();						\
+	}								\
+	unreachable();							\
+} while (0)
+
+#define not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Not implemented\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define assert_not_implemented(e) do {					\
+	if (unlikely(config_debug && !(e))) {				\
+		not_implemented();					\
+	}								\
+} while (0)
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void wrtmessage(void *cbopaque, const char *s);
+#define U2S_BUFSIZE ((1U << (LG_SIZEOF_INTMAX_T + 3)) + 1)
+static char *u2s(uintmax_t x, unsigned base, bool uppercase, char *s,
+    size_t *slen_p);
+#define D2S_BUFSIZE (1 + U2S_BUFSIZE)
+static char *d2s(intmax_t x, char sign, char *s, size_t *slen_p);
+#define O2S_BUFSIZE (1 + U2S_BUFSIZE)
+static char *o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p);
+#define X2S_BUFSIZE (2 + U2S_BUFSIZE)
+static char *x2s(uintmax_t x, bool alt_form, bool uppercase, char *s,
+    size_t *slen_p);
+
+/******************************************************************************/
+
+/* malloc_message() setup. */
+static void
+wrtmessage(void *cbopaque, const char *s) {
+	malloc_write_fd(STDERR_FILENO, s, strlen(s));
+}
+
+JEMALLOC_EXPORT void	(*je_malloc_message)(void *, const char *s);
+
+/*
+ * Wrapper around malloc_message() that avoids the need for
+ * je_malloc_message(...) throughout the code.
+ */
+void
+malloc_write(const char *s) {
+	if (je_malloc_message != NULL) {
+		je_malloc_message(NULL, s);
+	} else {
+		wrtmessage(NULL, s);
+	}
+}
+
+/*
+ * glibc provides a non-standard strerror_r() when _GNU_SOURCE is defined, so
+ * provide a wrapper.
+ */
+int
+buferror(int err, char *buf, size_t buflen) {
+#ifdef _WIN32
+	FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM, NULL, err, 0,
+	    (LPSTR)buf, (DWORD)buflen, NULL);
+	return 0;
+#elif defined(JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE) && defined(_GNU_SOURCE)
+	char *b = strerror_r(err, buf, buflen);
+	if (b != buf) {
+		strncpy(buf, b, buflen);
+		buf[buflen-1] = '\0';
+	}
+	return 0;
+#else
+	return strerror_r(err, buf, buflen);
+#endif
+}
+
+uintmax_t
+malloc_strtoumax(const char *restrict nptr, char **restrict endptr, int base) {
+	uintmax_t ret, digit;
+	unsigned b;
+	bool neg;
+	const char *p, *ns;
+
+	p = nptr;
+	if (base < 0 || base == 1 || base > 36) {
+		ns = p;
+		set_errno(EINVAL);
+		ret = UINTMAX_MAX;
+		goto label_return;
+	}
+	b = base;
+
+	/* Swallow leading whitespace and get sign, if any. */
+	neg = false;
+	while (true) {
+		switch (*p) {
+		case '\t': case '\n': case '\v': case '\f': case '\r': case ' ':
+			p++;
+			break;
+		case '-':
+			neg = true;
+			/* Fall through. */
+		case '+':
+			p++;
+			/* Fall through. */
+		default:
+			goto label_prefix;
+		}
+	}
+
+	/* Get prefix, if any. */
+	label_prefix:
+	/*
+	 * Note where the first non-whitespace/sign character is so that it is
+	 * possible to tell whether any digits are consumed (e.g., "  0" vs.
+	 * "  -x").
+	 */
+	ns = p;
+	if (*p == '0') {
+		switch (p[1]) {
+		case '0': case '1': case '2': case '3': case '4': case '5':
+		case '6': case '7':
+			if (b == 0) {
+				b = 8;
+			}
+			if (b == 8) {
+				p++;
+			}
+			break;
+		case 'X': case 'x':
+			switch (p[2]) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+				if (b == 0) {
+					b = 16;
+				}
+				if (b == 16) {
+					p += 2;
+				}
+				break;
+			default:
+				break;
+			}
+			break;
+		default:
+			p++;
+			ret = 0;
+			goto label_return;
+		}
+	}
+	if (b == 0) {
+		b = 10;
+	}
+
+	/* Convert. */
+	ret = 0;
+	while ((*p >= '0' && *p <= '9' && (digit = *p - '0') < b)
+	    || (*p >= 'A' && *p <= 'Z' && (digit = 10 + *p - 'A') < b)
+	    || (*p >= 'a' && *p <= 'z' && (digit = 10 + *p - 'a') < b)) {
+		uintmax_t pret = ret;
+		ret *= b;
+		ret += digit;
+		if (ret < pret) {
+			/* Overflow. */
+			set_errno(ERANGE);
+			ret = UINTMAX_MAX;
+			goto label_return;
+		}
+		p++;
+	}
+	if (neg) {
+		ret = (uintmax_t)(-((intmax_t)ret));
+	}
+
+	if (p == ns) {
+		/* No conversion performed. */
+		set_errno(EINVAL);
+		ret = UINTMAX_MAX;
+		goto label_return;
+	}
+
+label_return:
+	if (endptr != NULL) {
+		if (p == ns) {
+			/* No characters were converted. */
+			*endptr = (char *)nptr;
+		} else {
+			*endptr = (char *)p;
+		}
+	}
+	return ret;
+}
+
+static char *
+u2s(uintmax_t x, unsigned base, bool uppercase, char *s, size_t *slen_p) {
+	unsigned i;
+
+	i = U2S_BUFSIZE - 1;
+	s[i] = '\0';
+	switch (base) {
+	case 10:
+		do {
+			i--;
+			s[i] = "0123456789"[x % (uint64_t)10];
+			x /= (uint64_t)10;
+		} while (x > 0);
+		break;
+	case 16: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEF"
+		    : "0123456789abcdef";
+
+		do {
+			i--;
+			s[i] = digits[x & 0xf];
+			x >>= 4;
+		} while (x > 0);
+		break;
+	} default: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+		    : "0123456789abcdefghijklmnopqrstuvwxyz";
+
+		assert(base >= 2 && base <= 36);
+		do {
+			i--;
+			s[i] = digits[x % (uint64_t)base];
+			x /= (uint64_t)base;
+		} while (x > 0);
+	}}
+
+	*slen_p = U2S_BUFSIZE - 1 - i;
+	return &s[i];
+}
+
+static char *
+d2s(intmax_t x, char sign, char *s, size_t *slen_p) {
+	bool neg;
+
+	if ((neg = (x < 0))) {
+		x = -x;
+	}
+	s = u2s(x, 10, false, s, slen_p);
+	if (neg) {
+		sign = '-';
+	}
+	switch (sign) {
+	case '-':
+		if (!neg) {
+			break;
+		}
+		/* Fall through. */
+	case ' ':
+	case '+':
+		s--;
+		(*slen_p)++;
+		*s = sign;
+		break;
+	default: not_reached();
+	}
+	return s;
+}
+
+static char *
+o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p) {
+	s = u2s(x, 8, false, s, slen_p);
+	if (alt_form && *s != '0') {
+		s--;
+		(*slen_p)++;
+		*s = '0';
+	}
+	return s;
+}
+
+static char *
+x2s(uintmax_t x, bool alt_form, bool uppercase, char *s, size_t *slen_p) {
+	s = u2s(x, 16, uppercase, s, slen_p);
+	if (alt_form) {
+		s -= 2;
+		(*slen_p) += 2;
+		memcpy(s, uppercase ? "0X" : "0x", 2);
+	}
+	return s;
+}
+
+size_t
+malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) {
+	size_t i;
+	const char *f;
+
+#define APPEND_C(c) do {						\
+	if (i < size) {							\
+		str[i] = (c);						\
+	}								\
+	i++;								\
+} while (0)
+#define APPEND_S(s, slen) do {						\
+	if (i < size) {							\
+		size_t cpylen = (slen <= size - i) ? slen : size - i;	\
+		memcpy(&str[i], s, cpylen);				\
+	}								\
+	i += slen;							\
+} while (0)
+#define APPEND_PADDED_S(s, slen, width, left_justify) do {		\
+	/* Left padding. */						\
+	size_t pad_len = (width == -1) ? 0 : ((slen < (size_t)width) ?	\
+	    (size_t)width - slen : 0);					\
+	if (!left_justify && pad_len != 0) {				\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++) {				\
+			APPEND_C(' ');					\
+		}							\
+	}								\
+	/* Value. */							\
+	APPEND_S(s, slen);						\
+	/* Right padding. */						\
+	if (left_justify && pad_len != 0) {				\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++) {				\
+			APPEND_C(' ');					\
+		}							\
+	}								\
+} while (0)
+#define GET_ARG_NUMERIC(val, len) do {					\
+	switch ((unsigned char)len) {					\
+	case '?':							\
+		val = va_arg(ap, int);					\
+		break;							\
+	case '?' | 0x80:						\
+		val = va_arg(ap, unsigned int);				\
+		break;							\
+	case 'l':							\
+		val = va_arg(ap, long);					\
+		break;							\
+	case 'l' | 0x80:						\
+		val = va_arg(ap, unsigned long);			\
+		break;							\
+	case 'q':							\
+		val = va_arg(ap, long long);				\
+		break;							\
+	case 'q' | 0x80:						\
+		val = va_arg(ap, unsigned long long);			\
+		break;							\
+	case 'j':							\
+		val = va_arg(ap, intmax_t);				\
+		break;							\
+	case 'j' | 0x80:						\
+		val = va_arg(ap, uintmax_t);				\
+		break;							\
+	case 't':							\
+		val = va_arg(ap, ptrdiff_t);				\
+		break;							\
+	case 'z':							\
+		val = va_arg(ap, ssize_t);				\
+		break;							\
+	case 'z' | 0x80:						\
+		val = va_arg(ap, size_t);				\
+		break;							\
+	case 'p': /* Synthetic; used for %p. */				\
+		val = va_arg(ap, uintptr_t);				\
+		break;							\
+	default:							\
+		not_reached();						\
+		val = 0;						\
+	}								\
+} while (0)
+
+	i = 0;
+	f = format;
+	while (true) {
+		switch (*f) {
+		case '\0': goto label_out;
+		case '%': {
+			bool alt_form = false;
+			bool left_justify = false;
+			bool plus_space = false;
+			bool plus_plus = false;
+			int prec = -1;
+			int width = -1;
+			unsigned char len = '?';
+			char *s;
+			size_t slen;
+
+			f++;
+			/* Flags. */
+			while (true) {
+				switch (*f) {
+				case '#':
+					assert(!alt_form);
+					alt_form = true;
+					break;
+				case '-':
+					assert(!left_justify);
+					left_justify = true;
+					break;
+				case ' ':
+					assert(!plus_space);
+					plus_space = true;
+					break;
+				case '+':
+					assert(!plus_plus);
+					plus_plus = true;
+					break;
+				default: goto label_width;
+				}
+				f++;
+			}
+			/* Width. */
+			label_width:
+			switch (*f) {
+			case '*':
+				width = va_arg(ap, int);
+				f++;
+				if (width < 0) {
+					left_justify = true;
+					width = -width;
+				}
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uwidth;
+				set_errno(0);
+				uwidth = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uwidth != UINTMAX_MAX || get_errno() !=
+				    ERANGE);
+				width = (int)uwidth;
+				break;
+			} default:
+				break;
+			}
+			/* Width/precision separator. */
+			if (*f == '.') {
+				f++;
+			} else {
+				goto label_length;
+			}
+			/* Precision. */
+			switch (*f) {
+			case '*':
+				prec = va_arg(ap, int);
+				f++;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uprec;
+				set_errno(0);
+				uprec = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uprec != UINTMAX_MAX || get_errno() !=
+				    ERANGE);
+				prec = (int)uprec;
+				break;
+			}
+			default: break;
+			}
+			/* Length. */
+			label_length:
+			switch (*f) {
+			case 'l':
+				f++;
+				if (*f == 'l') {
+					len = 'q';
+					f++;
+				} else {
+					len = 'l';
+				}
+				break;
+			case 'q': case 'j': case 't': case 'z':
+				len = *f;
+				f++;
+				break;
+			default: break;
+			}
+			/* Conversion specifier. */
+			switch (*f) {
+			case '%':
+				/* %% */
+				APPEND_C(*f);
+				f++;
+				break;
+			case 'd': case 'i': {
+				intmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[D2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len);
+				s = d2s(val, (plus_plus ? '+' : (plus_space ?
+				    ' ' : '-')), buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'o': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[O2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = o2s(val, alt_form, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'u': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[U2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = u2s(val, 10, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'x': case 'X': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len | 0x80);
+				s = x2s(val, alt_form, *f == 'X', buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'c': {
+				unsigned char val;
+				char buf[2];
+
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				val = va_arg(ap, int);
+				buf[0] = val;
+				buf[1] = '\0';
+				APPEND_PADDED_S(buf, 1, width, left_justify);
+				f++;
+				break;
+			} case 's':
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				s = va_arg(ap, char *);
+				slen = (prec < 0) ? strlen(s) : (size_t)prec;
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			case 'p': {
+				uintmax_t val;
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, 'p');
+				s = x2s(val, true, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} default: not_reached();
+			}
+			break;
+		} default: {
+			APPEND_C(*f);
+			f++;
+			break;
+		}}
+	}
+	label_out:
+	if (i < size) {
+		str[i] = '\0';
+	} else {
+		str[size - 1] = '\0';
+	}
+
+#undef APPEND_C
+#undef APPEND_S
+#undef APPEND_PADDED_S
+#undef GET_ARG_NUMERIC
+	return i;
+}
+
+JEMALLOC_FORMAT_PRINTF(3, 4)
+size_t
+malloc_snprintf(char *str, size_t size, const char *format, ...) {
+	size_t ret;
+	va_list ap;
+
+	va_start(ap, format);
+	ret = malloc_vsnprintf(str, size, format, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+void
+malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap) {
+	char buf[MALLOC_PRINTF_BUFSIZE];
+
+	if (write_cb == NULL) {
+		/*
+		 * The caller did not provide an alternate write_cb callback
+		 * function, so use the default one.  malloc_write() is an
+		 * inline function, so use malloc_message() directly here.
+		 */
+		write_cb = (je_malloc_message != NULL) ? je_malloc_message :
+		    wrtmessage;
+	}
+
+	malloc_vsnprintf(buf, sizeof(buf), format, ap);
+	write_cb(cbopaque, buf);
+}
+
+/*
+ * Print to a callback function in such a way as to (hopefully) avoid memory
+ * allocation.
+ */
+JEMALLOC_FORMAT_PRINTF(3, 4)
+void
+malloc_cprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(write_cb, cbopaque, format, ap);
+	va_end(ap);
+}
+
+/* Print to stderr in such a way as to avoid memory allocation. */
+JEMALLOC_FORMAT_PRINTF(1, 2)
+void
+malloc_printf(const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+}
+
+/*
+ * Restore normal assertion macros, in order to make it possible to compile all
+ * C files as a single concatenation.
+ */
+#undef assert
+#undef not_reached
+#undef not_implemented
+#undef assert_not_implemented
+#include "jemalloc/internal/assert.h"
diff --git a/memkind-1.10.0/jemalloc/src/mutex.c b/memkind-1.10.0/jemalloc/src/mutex.c
new file mode 100644
index 0000000..3f920f5
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/mutex.c
@@ -0,0 +1,223 @@
+#define JEMALLOC_MUTEX_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/spin.h"
+
+#ifndef _CRT_SPINCOUNT
+#define _CRT_SPINCOUNT 4000
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+#ifdef JEMALLOC_LAZY_LOCK
+bool isthreaded = false;
+#endif
+#ifdef JEMALLOC_MUTEX_INIT_CB
+static bool		postpone_init = true;
+static malloc_mutex_t	*postponed_mutexes = NULL;
+#endif
+
+/******************************************************************************/
+/*
+ * We intercept pthread_create() calls in order to toggle isthreaded if the
+ * process goes multi-threaded.
+ */
+
+#if defined(JEMALLOC_LAZY_LOCK) && !defined(_WIN32)
+JEMALLOC_EXPORT int
+pthread_create(pthread_t *__restrict thread,
+    const pthread_attr_t *__restrict attr, void *(*start_routine)(void *),
+    void *__restrict arg) {
+	return pthread_create_wrapper(thread, attr, start_routine, arg);
+}
+#endif
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+JEMALLOC_EXPORT int	_pthread_mutex_init_calloc_cb(pthread_mutex_t *mutex,
+    void *(calloc_cb)(size_t, size_t));
+#endif
+
+void
+malloc_mutex_lock_slow(malloc_mutex_t *mutex) {
+	mutex_prof_data_t *data = &mutex->prof_data;
+	nstime_t before = NSTIME_ZERO_INITIALIZER;
+
+	if (ncpus == 1) {
+		goto label_spin_done;
+	}
+
+	int cnt = 0, max_cnt = MALLOC_MUTEX_MAX_SPIN;
+	do {
+		spin_cpu_spinwait();
+		if (!atomic_load_b(&mutex->locked, ATOMIC_RELAXED)
+                    && !malloc_mutex_trylock_final(mutex)) {
+			data->n_spin_acquired++;
+			return;
+		}
+	} while (cnt++ < max_cnt);
+
+	if (!config_stats) {
+		/* Only spin is useful when stats is off. */
+		malloc_mutex_lock_final(mutex);
+		return;
+	}
+label_spin_done:
+	nstime_update(&before);
+	/* Copy before to after to avoid clock skews. */
+	nstime_t after;
+	nstime_copy(&after, &before);
+	uint32_t n_thds = atomic_fetch_add_u32(&data->n_waiting_thds, 1,
+	    ATOMIC_RELAXED) + 1;
+	/* One last try as above two calls may take quite some cycles. */
+	if (!malloc_mutex_trylock_final(mutex)) {
+		atomic_fetch_sub_u32(&data->n_waiting_thds, 1, ATOMIC_RELAXED);
+		data->n_spin_acquired++;
+		return;
+	}
+
+	/* True slow path. */
+	malloc_mutex_lock_final(mutex);
+	/* Update more slow-path only counters. */
+	atomic_fetch_sub_u32(&data->n_waiting_thds, 1, ATOMIC_RELAXED);
+	nstime_update(&after);
+
+	nstime_t delta;
+	nstime_copy(&delta, &after);
+	nstime_subtract(&delta, &before);
+
+	data->n_wait_times++;
+	nstime_add(&data->tot_wait_time, &delta);
+	if (nstime_compare(&data->max_wait_time, &delta) < 0) {
+		nstime_copy(&data->max_wait_time, &delta);
+	}
+	if (n_thds > data->max_n_thds) {
+		data->max_n_thds = n_thds;
+	}
+}
+
+static void
+mutex_prof_data_init(mutex_prof_data_t *data) {
+	memset(data, 0, sizeof(mutex_prof_data_t));
+	nstime_init(&data->max_wait_time, 0);
+	nstime_init(&data->tot_wait_time, 0);
+	data->prev_owner = NULL;
+}
+
+void
+malloc_mutex_prof_data_reset(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	malloc_mutex_assert_owner(tsdn, mutex);
+	mutex_prof_data_init(&mutex->prof_data);
+}
+
+static int
+mutex_addr_comp(const witness_t *witness1, void *mutex1,
+    const witness_t *witness2, void *mutex2) {
+	assert(mutex1 != NULL);
+	assert(mutex2 != NULL);
+	uintptr_t mu1int = (uintptr_t)mutex1;
+	uintptr_t mu2int = (uintptr_t)mutex2;
+	if (mu1int < mu2int) {
+		return -1;
+	} else if (mu1int == mu2int) {
+		return 0;
+	} else {
+		return 1;
+	}
+}
+
+bool
+malloc_mutex_init(malloc_mutex_t *mutex, const char *name,
+    witness_rank_t rank, malloc_mutex_lock_order_t lock_order) {
+	mutex_prof_data_init(&mutex->prof_data);
+#ifdef _WIN32
+#  if _WIN32_WINNT >= 0x0600
+	InitializeSRWLock(&mutex->lock);
+#  else
+	if (!InitializeCriticalSectionAndSpinCount(&mutex->lock,
+	    _CRT_SPINCOUNT)) {
+		return true;
+	}
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+       mutex->lock = OS_UNFAIR_LOCK_INIT;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+	if (postpone_init) {
+		mutex->postponed_next = postponed_mutexes;
+		postponed_mutexes = mutex;
+	} else {
+		if (_pthread_mutex_init_calloc_cb(&mutex->lock,
+		    bootstrap_calloc) != 0) {
+			return true;
+		}
+	}
+#else
+	pthread_mutexattr_t attr;
+
+	if (pthread_mutexattr_init(&attr) != 0) {
+		return true;
+	}
+	pthread_mutexattr_settype(&attr, MALLOC_MUTEX_TYPE);
+	if (pthread_mutex_init(&mutex->lock, &attr) != 0) {
+		pthread_mutexattr_destroy(&attr);
+		return true;
+	}
+	pthread_mutexattr_destroy(&attr);
+#endif
+	if (config_debug) {
+		mutex->lock_order = lock_order;
+		if (lock_order == malloc_mutex_address_ordered) {
+			witness_init(&mutex->witness, name, rank,
+			    mutex_addr_comp, mutex);
+		} else {
+			witness_init(&mutex->witness, name, rank, NULL, NULL);
+		}
+	}
+	return false;
+}
+
+void
+malloc_mutex_prefork(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	malloc_mutex_lock(tsdn, mutex);
+}
+
+void
+malloc_mutex_postfork_parent(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	malloc_mutex_unlock(tsdn, mutex);
+}
+
+void
+malloc_mutex_postfork_child(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	malloc_mutex_unlock(tsdn, mutex);
+#else
+	if (malloc_mutex_init(mutex, mutex->witness.name,
+	    mutex->witness.rank, mutex->lock_order)) {
+		malloc_printf("<jemalloc>: Error re-initializing mutex in "
+		    "child\n");
+		if (opt_abort) {
+			abort();
+		}
+	}
+#endif
+}
+
+bool
+malloc_mutex_boot(void) {
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	postpone_init = false;
+	while (postponed_mutexes != NULL) {
+		if (_pthread_mutex_init_calloc_cb(&postponed_mutexes->lock,
+		    bootstrap_calloc) != 0) {
+			return true;
+		}
+		postponed_mutexes = postponed_mutexes->postponed_next;
+	}
+#endif
+	return false;
+}
diff --git a/memkind-1.10.0/jemalloc/src/mutex_pool.c b/memkind-1.10.0/jemalloc/src/mutex_pool.c
new file mode 100644
index 0000000..f24d10e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/mutex_pool.c
@@ -0,0 +1,18 @@
+#define JEMALLOC_MUTEX_POOL_C_
+
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+
+bool
+mutex_pool_init(mutex_pool_t *pool, const char *name, witness_rank_t rank) {
+	for (int i = 0; i < MUTEX_POOL_SIZE; ++i) {
+		if (malloc_mutex_init(&pool->mutexes[i], name, rank,
+		    malloc_mutex_address_ordered)) {
+			return true;
+		}
+	}
+	return false;
+}
diff --git a/memkind-1.10.0/jemalloc/src/nstime.c b/memkind-1.10.0/jemalloc/src/nstime.c
new file mode 100644
index 0000000..71db353
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/nstime.c
@@ -0,0 +1,170 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/nstime.h"
+
+#include "jemalloc/internal/assert.h"
+
+#define BILLION	UINT64_C(1000000000)
+#define MILLION	UINT64_C(1000000)
+
+void
+nstime_init(nstime_t *time, uint64_t ns) {
+	time->ns = ns;
+}
+
+void
+nstime_init2(nstime_t *time, uint64_t sec, uint64_t nsec) {
+	time->ns = sec * BILLION + nsec;
+}
+
+uint64_t
+nstime_ns(const nstime_t *time) {
+	return time->ns;
+}
+
+uint64_t
+nstime_msec(const nstime_t *time) {
+	return time->ns / MILLION;
+}
+
+uint64_t
+nstime_sec(const nstime_t *time) {
+	return time->ns / BILLION;
+}
+
+uint64_t
+nstime_nsec(const nstime_t *time) {
+	return time->ns % BILLION;
+}
+
+void
+nstime_copy(nstime_t *time, const nstime_t *source) {
+	*time = *source;
+}
+
+int
+nstime_compare(const nstime_t *a, const nstime_t *b) {
+	return (a->ns > b->ns) - (a->ns < b->ns);
+}
+
+void
+nstime_add(nstime_t *time, const nstime_t *addend) {
+	assert(UINT64_MAX - time->ns >= addend->ns);
+
+	time->ns += addend->ns;
+}
+
+void
+nstime_iadd(nstime_t *time, uint64_t addend) {
+	assert(UINT64_MAX - time->ns >= addend);
+
+	time->ns += addend;
+}
+
+void
+nstime_subtract(nstime_t *time, const nstime_t *subtrahend) {
+	assert(nstime_compare(time, subtrahend) >= 0);
+
+	time->ns -= subtrahend->ns;
+}
+
+void
+nstime_isubtract(nstime_t *time, uint64_t subtrahend) {
+	assert(time->ns >= subtrahend);
+
+	time->ns -= subtrahend;
+}
+
+void
+nstime_imultiply(nstime_t *time, uint64_t multiplier) {
+	assert((((time->ns | multiplier) & (UINT64_MAX << (sizeof(uint64_t) <<
+	    2))) == 0) || ((time->ns * multiplier) / multiplier == time->ns));
+
+	time->ns *= multiplier;
+}
+
+void
+nstime_idivide(nstime_t *time, uint64_t divisor) {
+	assert(divisor != 0);
+
+	time->ns /= divisor;
+}
+
+uint64_t
+nstime_divide(const nstime_t *time, const nstime_t *divisor) {
+	assert(divisor->ns != 0);
+
+	return time->ns / divisor->ns;
+}
+
+#ifdef _WIN32
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	FILETIME ft;
+	uint64_t ticks_100ns;
+
+	GetSystemTimeAsFileTime(&ft);
+	ticks_100ns = (((uint64_t)ft.dwHighDateTime) << 32) | ft.dwLowDateTime;
+
+	nstime_init(time, ticks_100ns * 100);
+}
+#elif defined(JEMALLOC_HAVE_CLOCK_MONOTONIC_COARSE)
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC_COARSE, &ts);
+	nstime_init2(time, ts.tv_sec, ts.tv_nsec);
+}
+#elif defined(JEMALLOC_HAVE_CLOCK_MONOTONIC)
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	struct timespec ts;
+
+	clock_gettime(CLOCK_MONOTONIC, &ts);
+	nstime_init2(time, ts.tv_sec, ts.tv_nsec);
+}
+#elif defined(JEMALLOC_HAVE_MACH_ABSOLUTE_TIME)
+#  define NSTIME_MONOTONIC true
+static void
+nstime_get(nstime_t *time) {
+	nstime_init(time, mach_absolute_time());
+}
+#else
+#  define NSTIME_MONOTONIC false
+static void
+nstime_get(nstime_t *time) {
+	struct timeval tv;
+
+	gettimeofday(&tv, NULL);
+	nstime_init2(time, tv.tv_sec, tv.tv_usec * 1000);
+}
+#endif
+
+static bool
+nstime_monotonic_impl(void) {
+	return NSTIME_MONOTONIC;
+#undef NSTIME_MONOTONIC
+}
+nstime_monotonic_t *JET_MUTABLE nstime_monotonic = nstime_monotonic_impl;
+
+static bool
+nstime_update_impl(nstime_t *time) {
+	nstime_t old_time;
+
+	nstime_copy(&old_time, time);
+	nstime_get(time);
+
+	/* Handle non-monotonic clocks. */
+	if (unlikely(nstime_compare(&old_time, time) > 0)) {
+		nstime_copy(time, &old_time);
+		return true;
+	}
+
+	return false;
+}
+nstime_update_t *JET_MUTABLE nstime_update = nstime_update_impl;
diff --git a/memkind-1.10.0/jemalloc/src/pages.c b/memkind-1.10.0/jemalloc/src/pages.c
new file mode 100644
index 0000000..b601578
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/pages.c
@@ -0,0 +1,649 @@
+#define JEMALLOC_PAGES_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/pages.h"
+
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/malloc_io.h"
+
+#ifdef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+#include <sys/sysctl.h>
+#ifdef __FreeBSD__
+#include <vm/vm_param.h>
+#endif
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+/* Actual operating system page size, detected during bootstrap, <= PAGE. */
+static size_t	os_page;
+
+#ifndef _WIN32
+#  define PAGES_PROT_COMMIT (PROT_READ | PROT_WRITE)
+#  define PAGES_PROT_DECOMMIT (PROT_NONE)
+static int	mmap_flags;
+#endif
+static bool	os_overcommits;
+
+const char *thp_mode_names[] = {
+	"default",
+	"always",
+	"never",
+	"not supported"
+};
+thp_mode_t opt_thp = THP_MODE_DEFAULT;
+thp_mode_t init_system_thp_mode;
+
+/* Runtime support for lazy purge. Irrelevant when !pages_can_purge_lazy. */
+static bool pages_can_purge_lazy_runtime = true;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static void os_pages_unmap(void *addr, size_t size);
+
+/******************************************************************************/
+
+static void *
+os_pages_map(void *addr, size_t size, size_t alignment, bool *commit) {
+	assert(ALIGNMENT_ADDR2BASE(addr, os_page) == addr);
+	assert(ALIGNMENT_CEILING(size, os_page) == size);
+	assert(size != 0);
+
+	if (os_overcommits) {
+		*commit = true;
+	}
+
+	void *ret;
+#ifdef _WIN32
+	/*
+	 * If VirtualAlloc can't allocate at the given address when one is
+	 * given, it fails and returns NULL.
+	 */
+	ret = VirtualAlloc(addr, size, MEM_RESERVE | (*commit ? MEM_COMMIT : 0),
+	    PAGE_READWRITE);
+#else
+	/*
+	 * We don't use MAP_FIXED here, because it can cause the *replacement*
+	 * of existing mappings, and we only want to create new mappings.
+	 */
+	{
+		int prot = *commit ? PAGES_PROT_COMMIT : PAGES_PROT_DECOMMIT;
+
+		ret = mmap(addr, size, prot, mmap_flags, -1, 0);
+	}
+	assert(ret != NULL);
+
+	if (ret == MAP_FAILED) {
+		ret = NULL;
+	} else if (addr != NULL && ret != addr) {
+		/*
+		 * We succeeded in mapping memory, but not in the right place.
+		 */
+		os_pages_unmap(ret, size);
+		ret = NULL;
+	}
+#endif
+	assert(ret == NULL || (addr == NULL && ret != addr) || (addr != NULL &&
+	    ret == addr));
+	return ret;
+}
+
+static void *
+os_pages_trim(void *addr, size_t alloc_size, size_t leadsize, size_t size,
+    bool *commit) {
+	void *ret = (void *)((uintptr_t)addr + leadsize);
+
+	assert(alloc_size >= leadsize + size);
+#ifdef _WIN32
+	os_pages_unmap(addr, alloc_size);
+	void *new_addr = os_pages_map(ret, size, PAGE, commit);
+	if (new_addr == ret) {
+		return ret;
+	}
+	if (new_addr != NULL) {
+		os_pages_unmap(new_addr, size);
+	}
+	return NULL;
+#else
+	size_t trailsize = alloc_size - leadsize - size;
+
+	if (leadsize != 0) {
+		os_pages_unmap(addr, leadsize);
+	}
+	if (trailsize != 0) {
+		os_pages_unmap((void *)((uintptr_t)ret + size), trailsize);
+	}
+	return ret;
+#endif
+}
+
+static void
+os_pages_unmap(void *addr, size_t size) {
+	assert(ALIGNMENT_ADDR2BASE(addr, os_page) == addr);
+	assert(ALIGNMENT_CEILING(size, os_page) == size);
+
+#ifdef _WIN32
+	if (VirtualFree(addr, 0, MEM_RELEASE) == 0)
+#else
+	if (munmap(addr, size) == -1)
+#endif
+	{
+		char buf[BUFERROR_BUF];
+
+		buferror(get_errno(), buf, sizeof(buf));
+		malloc_printf("<jemalloc>: Error in "
+#ifdef _WIN32
+		    "VirtualFree"
+#else
+		    "munmap"
+#endif
+		    "(): %s\n", buf);
+		if (opt_abort) {
+			abort();
+		}
+	}
+}
+
+static void *
+pages_map_slow(size_t size, size_t alignment, bool *commit) {
+	size_t alloc_size = size + alignment - os_page;
+	/* Beware size_t wrap-around. */
+	if (alloc_size < size) {
+		return NULL;
+	}
+
+	void *ret;
+	do {
+		void *pages = os_pages_map(NULL, alloc_size, alignment, commit);
+		if (pages == NULL) {
+			return NULL;
+		}
+		size_t leadsize = ALIGNMENT_CEILING((uintptr_t)pages, alignment)
+		    - (uintptr_t)pages;
+		ret = os_pages_trim(pages, alloc_size, leadsize, size, commit);
+	} while (ret == NULL);
+
+	assert(ret != NULL);
+	assert(PAGE_ADDR2BASE(ret) == ret);
+	return ret;
+}
+
+void *
+pages_map(void *addr, size_t size, size_t alignment, bool *commit) {
+	assert(alignment >= PAGE);
+	assert(ALIGNMENT_ADDR2BASE(addr, alignment) == addr);
+
+#if defined(__FreeBSD__) && defined(MAP_EXCL)
+	/*
+	 * FreeBSD has mechanisms both to mmap at specific address without
+	 * touching existing mappings, and to mmap with specific alignment.
+	 */
+	{
+		if (os_overcommits) {
+			*commit = true;
+		}
+
+		int prot = *commit ? PAGES_PROT_COMMIT : PAGES_PROT_DECOMMIT;
+		int flags = mmap_flags;
+
+		if (addr != NULL) {
+			flags |= MAP_FIXED | MAP_EXCL;
+		} else {
+			unsigned alignment_bits = ffs_zu(alignment);
+			assert(alignment_bits > 1);
+			flags |= MAP_ALIGNED(alignment_bits - 1);
+		}
+
+		void *ret = mmap(addr, size, prot, flags, -1, 0);
+		if (ret == MAP_FAILED) {
+			ret = NULL;
+		}
+
+		return ret;
+	}
+#endif
+	/*
+	 * Ideally, there would be a way to specify alignment to mmap() (like
+	 * NetBSD has), but in the absence of such a feature, we have to work
+	 * hard to efficiently create aligned mappings.  The reliable, but
+	 * slow method is to create a mapping that is over-sized, then trim the
+	 * excess.  However, that always results in one or two calls to
+	 * os_pages_unmap(), and it can leave holes in the process's virtual
+	 * memory map if memory grows downward.
+	 *
+	 * Optimistically try mapping precisely the right amount before falling
+	 * back to the slow method, with the expectation that the optimistic
+	 * approach works most of the time.
+	 */
+
+	void *ret = os_pages_map(addr, size, os_page, commit);
+	if (ret == NULL || ret == addr) {
+		return ret;
+	}
+	assert(addr == NULL);
+	if (ALIGNMENT_ADDR2OFFSET(ret, alignment) != 0) {
+		os_pages_unmap(ret, size);
+		return pages_map_slow(size, alignment, commit);
+	}
+
+	assert(PAGE_ADDR2BASE(ret) == ret);
+	return ret;
+}
+
+void
+pages_unmap(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	os_pages_unmap(addr, size);
+}
+
+static bool
+pages_commit_impl(void *addr, size_t size, bool commit) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	if (os_overcommits) {
+		return true;
+	}
+
+#ifdef _WIN32
+	return (commit ? (addr != VirtualAlloc(addr, size, MEM_COMMIT,
+	    PAGE_READWRITE)) : (!VirtualFree(addr, size, MEM_DECOMMIT)));
+#else
+	{
+		int prot = commit ? PAGES_PROT_COMMIT : PAGES_PROT_DECOMMIT;
+		void *result = mmap(addr, size, prot, mmap_flags | MAP_FIXED,
+		    -1, 0);
+		if (result == MAP_FAILED) {
+			return true;
+		}
+		if (result != addr) {
+			/*
+			 * We succeeded in mapping memory, but not in the right
+			 * place.
+			 */
+			os_pages_unmap(result, size);
+			return true;
+		}
+		return false;
+	}
+#endif
+}
+
+bool
+pages_commit(void *addr, size_t size) {
+	return pages_commit_impl(addr, size, true);
+}
+
+bool
+pages_decommit(void *addr, size_t size) {
+	return pages_commit_impl(addr, size, false);
+}
+
+bool
+pages_purge_lazy(void *addr, size_t size) {
+	assert(ALIGNMENT_ADDR2BASE(addr, os_page) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	if (!pages_can_purge_lazy) {
+		return true;
+	}
+	if (!pages_can_purge_lazy_runtime) {
+		/*
+		 * Built with lazy purge enabled, but detected it was not
+		 * supported on the current system.
+		 */
+		return true;
+	}
+
+#ifdef _WIN32
+	VirtualAlloc(addr, size, MEM_RESET, PAGE_READWRITE);
+	return false;
+#elif defined(JEMALLOC_PURGE_MADVISE_FREE)
+	return (madvise(addr, size,
+#  ifdef MADV_FREE
+	    MADV_FREE
+#  else
+	    JEMALLOC_MADV_FREE
+#  endif
+	    ) != 0);
+#elif defined(JEMALLOC_PURGE_MADVISE_DONTNEED) && \
+    !defined(JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS)
+	return (madvise(addr, size, MADV_DONTNEED) != 0);
+#else
+	not_reached();
+#endif
+}
+
+bool
+pages_purge_forced(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+
+	if (!pages_can_purge_forced) {
+		return true;
+	}
+
+#if defined(JEMALLOC_PURGE_MADVISE_DONTNEED) && \
+    defined(JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS)
+	return (madvise(addr, size, MADV_DONTNEED) != 0);
+#elif defined(JEMALLOC_MAPS_COALESCE)
+	/* Try to overlay a new demand-zeroed mapping. */
+	return pages_commit(addr, size);
+#else
+	not_reached();
+#endif
+}
+
+static bool
+pages_huge_impl(void *addr, size_t size, bool aligned) {
+	if (aligned) {
+		assert(HUGEPAGE_ADDR2BASE(addr) == addr);
+		assert(HUGEPAGE_CEILING(size) == size);
+	}
+#ifdef JEMALLOC_HAVE_MADVISE_HUGE
+	return (madvise(addr, size, MADV_HUGEPAGE) != 0);
+#else
+	return true;
+#endif
+}
+
+bool
+pages_huge(void *addr, size_t size) {
+	return pages_huge_impl(addr, size, true);
+}
+
+static bool
+pages_huge_unaligned(void *addr, size_t size) {
+	return pages_huge_impl(addr, size, false);
+}
+
+static bool
+pages_nohuge_impl(void *addr, size_t size, bool aligned) {
+	if (aligned) {
+		assert(HUGEPAGE_ADDR2BASE(addr) == addr);
+		assert(HUGEPAGE_CEILING(size) == size);
+	}
+
+#ifdef JEMALLOC_HAVE_MADVISE_HUGE
+	return (madvise(addr, size, MADV_NOHUGEPAGE) != 0);
+#else
+	return false;
+#endif
+}
+
+bool
+pages_nohuge(void *addr, size_t size) {
+	return pages_nohuge_impl(addr, size, true);
+}
+
+static bool
+pages_nohuge_unaligned(void *addr, size_t size) {
+	return pages_nohuge_impl(addr, size, false);
+}
+
+bool
+pages_dontdump(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+#ifdef JEMALLOC_MADVISE_DONTDUMP
+	return madvise(addr, size, MADV_DONTDUMP) != 0;
+#else
+	return false;
+#endif
+}
+
+bool
+pages_dodump(void *addr, size_t size) {
+	assert(PAGE_ADDR2BASE(addr) == addr);
+	assert(PAGE_CEILING(size) == size);
+#ifdef JEMALLOC_MADVISE_DONTDUMP
+	return madvise(addr, size, MADV_DODUMP) != 0;
+#else
+	return false;
+#endif
+}
+
+
+static size_t
+os_page_detect(void) {
+#ifdef _WIN32
+	SYSTEM_INFO si;
+	GetSystemInfo(&si);
+	return si.dwPageSize;
+#elif defined(__FreeBSD__)
+	/*
+	 * This returns the value obtained from
+	 * the auxv vector, avoiding a syscall.
+	 */
+	return getpagesize();
+#else
+	long result = sysconf(_SC_PAGESIZE);
+	if (result == -1) {
+		return LG_PAGE;
+	}
+	return (size_t)result;
+#endif
+}
+
+#ifdef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+static bool
+os_overcommits_sysctl(void) {
+	int vm_overcommit;
+	size_t sz;
+
+	sz = sizeof(vm_overcommit);
+#if defined(__FreeBSD__) && defined(VM_OVERCOMMIT)
+	int mib[2];
+
+	mib[0] = CTL_VM;
+	mib[1] = VM_OVERCOMMIT;
+	if (sysctl(mib, 2, &vm_overcommit, &sz, NULL, 0) != 0) {
+		return false; /* Error. */
+	}
+#else
+	if (sysctlbyname("vm.overcommit", &vm_overcommit, &sz, NULL, 0) != 0) {
+		return false; /* Error. */
+	}
+#endif
+
+	return ((vm_overcommit & 0x3) == 0);
+}
+#endif
+
+#ifdef JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY
+/*
+ * Use syscall(2) rather than {open,read,close}(2) when possible to avoid
+ * reentry during bootstrapping if another library has interposed system call
+ * wrappers.
+ */
+static bool
+os_overcommits_proc(void) {
+	int fd;
+	char buf[1];
+
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_open)
+	#if defined(O_CLOEXEC)
+		fd = (int)syscall(SYS_open, "/proc/sys/vm/overcommit_memory", O_RDONLY |
+			O_CLOEXEC);
+	#else
+		fd = (int)syscall(SYS_open, "/proc/sys/vm/overcommit_memory", O_RDONLY);
+		if (fd != -1) {
+			fcntl(fd, F_SETFD, fcntl(fd, F_GETFD) | FD_CLOEXEC);
+		}
+	#endif
+#elif defined(JEMALLOC_USE_SYSCALL) && defined(SYS_openat)
+	#if defined(O_CLOEXEC)
+		fd = (int)syscall(SYS_openat,
+			AT_FDCWD, "/proc/sys/vm/overcommit_memory", O_RDONLY | O_CLOEXEC);
+	#else
+		fd = (int)syscall(SYS_openat,
+			AT_FDCWD, "/proc/sys/vm/overcommit_memory", O_RDONLY);
+		if (fd != -1) {
+			fcntl(fd, F_SETFD, fcntl(fd, F_GETFD) | FD_CLOEXEC);
+		}
+	#endif
+#else
+	#if defined(O_CLOEXEC)
+		fd = open("/proc/sys/vm/overcommit_memory", O_RDONLY | O_CLOEXEC);
+	#else
+		fd = open("/proc/sys/vm/overcommit_memory", O_RDONLY);
+		if (fd != -1) {
+			fcntl(fd, F_SETFD, fcntl(fd, F_GETFD) | FD_CLOEXEC);
+		}
+	#endif
+#endif
+
+	if (fd == -1) {
+		return false; /* Error. */
+	}
+
+	ssize_t nread = malloc_read_fd(fd, &buf, sizeof(buf));
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_close)
+	syscall(SYS_close, fd);
+#else
+	close(fd);
+#endif
+
+	if (nread < 1) {
+		return false; /* Error. */
+	}
+	/*
+	 * /proc/sys/vm/overcommit_memory meanings:
+	 * 0: Heuristic overcommit.
+	 * 1: Always overcommit.
+	 * 2: Never overcommit.
+	 */
+	return (buf[0] == '0' || buf[0] == '1');
+}
+#endif
+
+void
+pages_set_thp_state (void *ptr, size_t size) {
+	if (opt_thp == thp_mode_default || opt_thp == init_system_thp_mode) {
+		return;
+	}
+	assert(opt_thp != thp_mode_not_supported &&
+	    init_system_thp_mode != thp_mode_not_supported);
+
+	if (opt_thp == thp_mode_always
+	    && init_system_thp_mode != thp_mode_never) {
+		assert(init_system_thp_mode == thp_mode_default);
+		pages_huge_unaligned(ptr, size);
+	} else if (opt_thp == thp_mode_never) {
+		assert(init_system_thp_mode == thp_mode_default ||
+		    init_system_thp_mode == thp_mode_always);
+		pages_nohuge_unaligned(ptr, size);
+	}
+}
+
+static void
+init_thp_state(void) {
+	if (!have_madvise_huge) {
+		if (metadata_thp_enabled() && opt_abort) {
+			malloc_write("<jemalloc>: no MADV_HUGEPAGE support\n");
+			abort();
+		}
+		goto label_error;
+	}
+
+	static const char sys_state_madvise[] = "always [madvise] never\n";
+	static const char sys_state_always[] = "[always] madvise never\n";
+	static const char sys_state_never[] = "always madvise [never]\n";
+	char buf[sizeof(sys_state_madvise)];
+
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_open)
+	int fd = (int)syscall(SYS_open,
+	    "/sys/kernel/mm/transparent_hugepage/enabled", O_RDONLY);
+#else
+	int fd = open("/sys/kernel/mm/transparent_hugepage/enabled", O_RDONLY);
+#endif
+	if (fd == -1) {
+		goto label_error;
+	}
+
+	ssize_t nread = malloc_read_fd(fd, &buf, sizeof(buf));
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_close)
+	syscall(SYS_close, fd);
+#else
+	close(fd);
+#endif
+
+        if (nread < 0) {
+		goto label_error;
+        }
+
+	if (strncmp(buf, sys_state_madvise, (size_t)nread) == 0) {
+		init_system_thp_mode = thp_mode_default;
+	} else if (strncmp(buf, sys_state_always, (size_t)nread) == 0) {
+		init_system_thp_mode = thp_mode_always;
+	} else if (strncmp(buf, sys_state_never, (size_t)nread) == 0) {
+		init_system_thp_mode = thp_mode_never;
+	} else {
+		goto label_error;
+	}
+	return;
+label_error:
+	opt_thp = init_system_thp_mode = thp_mode_not_supported;
+}
+
+bool
+pages_boot(void) {
+	os_page = os_page_detect();
+	if (os_page > PAGE) {
+		malloc_write("<jemalloc>: Unsupported system page size\n");
+		if (opt_abort) {
+			abort();
+		}
+		return true;
+	}
+
+#ifndef _WIN32
+	mmap_flags = MAP_PRIVATE | MAP_ANON;
+#endif
+
+#ifdef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+	os_overcommits = os_overcommits_sysctl();
+#elif defined(JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY)
+	os_overcommits = os_overcommits_proc();
+#  ifdef MAP_NORESERVE
+	if (os_overcommits) {
+		mmap_flags |= MAP_NORESERVE;
+	}
+#  endif
+#else
+	os_overcommits = false;
+#endif
+
+	init_thp_state();
+
+#ifdef __FreeBSD__
+	/*
+	 * FreeBSD doesn't need the check; madvise(2) is known to work.
+	 */
+#else
+	/* Detect lazy purge runtime support. */
+	if (pages_can_purge_lazy) {
+		bool committed = false;
+		void *madv_free_page = os_pages_map(NULL, PAGE, PAGE, &committed);
+		if (madv_free_page == NULL) {
+			return true;
+		}
+		assert(pages_can_purge_lazy_runtime);
+		if (pages_purge_lazy(madv_free_page, PAGE)) {
+			pages_can_purge_lazy_runtime = false;
+		}
+		os_pages_unmap(madv_free_page, PAGE);
+	}
+#endif
+
+	return false;
+}
diff --git a/memkind-1.10.0/jemalloc/src/prng.c b/memkind-1.10.0/jemalloc/src/prng.c
new file mode 100644
index 0000000..83c04bf
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/prng.c
@@ -0,0 +1,3 @@
+#define JEMALLOC_PRNG_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
diff --git a/memkind-1.10.0/jemalloc/src/prof.c b/memkind-1.10.0/jemalloc/src/prof.c
new file mode 100644
index 0000000..13334cb
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/prof.c
@@ -0,0 +1,3160 @@
+#define JEMALLOC_PROF_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/emitter.h"
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_PROF_LIBUNWIND
+#define UNW_LOCAL_ONLY
+#include <libunwind.h>
+#endif
+
+#ifdef JEMALLOC_PROF_LIBGCC
+/*
+ * We have a circular dependency -- jemalloc_internal.h tells us if we should
+ * use libgcc's unwinding functionality, but after we've included that, we've
+ * already hooked _Unwind_Backtrace.  We'll temporarily disable hooking.
+ */
+#undef _Unwind_Backtrace
+#include <unwind.h>
+#define _Unwind_Backtrace JEMALLOC_HOOK(_Unwind_Backtrace, test_hooks_libc_hook)
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+bool		opt_prof = false;
+bool		opt_prof_active = true;
+bool		opt_prof_thread_active_init = true;
+size_t		opt_lg_prof_sample = LG_PROF_SAMPLE_DEFAULT;
+ssize_t		opt_lg_prof_interval = LG_PROF_INTERVAL_DEFAULT;
+bool		opt_prof_gdump = false;
+bool		opt_prof_final = false;
+bool		opt_prof_leak = false;
+bool		opt_prof_accum = false;
+bool		opt_prof_log = false;
+char		opt_prof_prefix[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PATH_MAX +
+#endif
+    1];
+
+/*
+ * Initialized as opt_prof_active, and accessed via
+ * prof_active_[gs]et{_unlocked,}().
+ */
+bool			prof_active;
+static malloc_mutex_t	prof_active_mtx;
+
+/*
+ * Initialized as opt_prof_thread_active_init, and accessed via
+ * prof_thread_active_init_[gs]et().
+ */
+static bool		prof_thread_active_init;
+static malloc_mutex_t	prof_thread_active_init_mtx;
+
+/*
+ * Initialized as opt_prof_gdump, and accessed via
+ * prof_gdump_[gs]et{_unlocked,}().
+ */
+bool			prof_gdump_val;
+static malloc_mutex_t	prof_gdump_mtx;
+
+uint64_t	prof_interval = 0;
+
+size_t		lg_prof_sample;
+
+typedef enum prof_logging_state_e prof_logging_state_t;
+enum prof_logging_state_e {
+	prof_logging_state_stopped,
+	prof_logging_state_started,
+	prof_logging_state_dumping
+};
+
+/*
+ * - stopped: log_start never called, or previous log_stop has completed.
+ * - started: log_start called, log_stop not called yet. Allocations are logged.
+ * - dumping: log_stop called but not finished; samples are not logged anymore.
+ */
+prof_logging_state_t prof_logging_state = prof_logging_state_stopped;
+
+#ifdef JEMALLOC_JET
+static bool prof_log_dummy = false;
+#endif
+
+/* Incremented for every log file that is output. */
+static uint64_t log_seq = 0;
+static char log_filename[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PATH_MAX +
+#endif
+    1];
+
+/* Timestamp for most recent call to log_start(). */
+static nstime_t log_start_timestamp = NSTIME_ZERO_INITIALIZER;
+
+/* Increment these when adding to the log_bt and log_thr linked lists. */
+static size_t log_bt_index = 0;
+static size_t log_thr_index = 0;
+
+/* Linked list node definitions. These are only used in prof.c. */
+typedef struct prof_bt_node_s prof_bt_node_t;
+
+struct prof_bt_node_s {
+	prof_bt_node_t *next;
+	size_t index;
+	prof_bt_t bt;
+	/* Variable size backtrace vector pointed to by bt. */
+	void *vec[1];
+};
+
+typedef struct prof_thr_node_s prof_thr_node_t;
+
+struct prof_thr_node_s {
+	prof_thr_node_t *next;
+	size_t index;
+	uint64_t thr_uid;
+	/* Variable size based on thr_name_sz. */
+	char name[1];
+};
+
+typedef struct prof_alloc_node_s prof_alloc_node_t;
+
+/* This is output when logging sampled allocations. */
+struct prof_alloc_node_s {
+	prof_alloc_node_t *next;
+	/* Indices into an array of thread data. */
+	size_t alloc_thr_ind;
+	size_t free_thr_ind;
+
+	/* Indices into an array of backtraces. */
+	size_t alloc_bt_ind;
+	size_t free_bt_ind;
+
+	uint64_t alloc_time_ns;
+	uint64_t free_time_ns;
+
+	size_t usize;
+};
+
+/*
+ * Created on the first call to prof_log_start and deleted on prof_log_stop.
+ * These are the backtraces and threads that have already been logged by an
+ * allocation.
+ */
+static bool log_tables_initialized = false;
+static ckh_t log_bt_node_set;
+static ckh_t log_thr_node_set;
+
+/* Store linked lists for logged data. */
+static prof_bt_node_t *log_bt_first = NULL;
+static prof_bt_node_t *log_bt_last = NULL;
+static prof_thr_node_t *log_thr_first = NULL;
+static prof_thr_node_t *log_thr_last = NULL;
+static prof_alloc_node_t *log_alloc_first = NULL;
+static prof_alloc_node_t *log_alloc_last = NULL;
+
+/* Protects the prof_logging_state and any log_{...} variable. */
+static malloc_mutex_t log_mtx;
+
+/*
+ * Table of mutexes that are shared among gctx's.  These are leaf locks, so
+ * there is no problem with using them for more than one gctx at the same time.
+ * The primary motivation for this sharing though is that gctx's are ephemeral,
+ * and destroying mutexes causes complications for systems that allocate when
+ * creating/destroying mutexes.
+ */
+static malloc_mutex_t	*gctx_locks;
+static atomic_u_t	cum_gctxs; /* Atomic counter. */
+
+/*
+ * Table of mutexes that are shared among tdata's.  No operations require
+ * holding multiple tdata locks, so there is no problem with using them for more
+ * than one tdata at the same time, even though a gctx lock may be acquired
+ * while holding a tdata lock.
+ */
+static malloc_mutex_t	*tdata_locks;
+
+/*
+ * Global hash of (prof_bt_t *)-->(prof_gctx_t *).  This is the master data
+ * structure that knows about all backtraces currently captured.
+ */
+static ckh_t		bt2gctx;
+/* Non static to enable profiling. */
+malloc_mutex_t		bt2gctx_mtx;
+
+/*
+ * Tree of all extant prof_tdata_t structures, regardless of state,
+ * {attached,detached,expired}.
+ */
+static prof_tdata_tree_t	tdatas;
+static malloc_mutex_t	tdatas_mtx;
+
+static uint64_t		next_thr_uid;
+static malloc_mutex_t	next_thr_uid_mtx;
+
+static malloc_mutex_t	prof_dump_seq_mtx;
+static uint64_t		prof_dump_seq;
+static uint64_t		prof_dump_iseq;
+static uint64_t		prof_dump_mseq;
+static uint64_t		prof_dump_useq;
+
+/*
+ * This buffer is rather large for stack allocation, so use a single buffer for
+ * all profile dumps.
+ */
+static malloc_mutex_t	prof_dump_mtx;
+static char		prof_dump_buf[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PROF_DUMP_BUFSIZE
+#else
+    1
+#endif
+];
+static size_t		prof_dump_buf_end;
+static int		prof_dump_fd;
+
+/* Do not dump any profiles until bootstrapping is complete. */
+static bool		prof_booted = false;
+
+/******************************************************************************/
+/*
+ * Function prototypes for static functions that are referenced prior to
+ * definition.
+ */
+
+static bool	prof_tctx_should_destroy(tsdn_t *tsdn, prof_tctx_t *tctx);
+static void	prof_tctx_destroy(tsd_t *tsd, prof_tctx_t *tctx);
+static bool	prof_tdata_should_destroy(tsdn_t *tsdn, prof_tdata_t *tdata,
+    bool even_if_attached);
+static void	prof_tdata_destroy(tsd_t *tsd, prof_tdata_t *tdata,
+    bool even_if_attached);
+static char	*prof_thread_name_alloc(tsdn_t *tsdn, const char *thread_name);
+
+/* Hashtable functions for log_bt_node_set and log_thr_node_set. */
+static void prof_thr_node_hash(const void *key, size_t r_hash[2]);
+static bool prof_thr_node_keycomp(const void *k1, const void *k2);
+static void prof_bt_node_hash(const void *key, size_t r_hash[2]);
+static bool prof_bt_node_keycomp(const void *k1, const void *k2);
+
+/******************************************************************************/
+/* Red-black trees. */
+
+static int
+prof_tctx_comp(const prof_tctx_t *a, const prof_tctx_t *b) {
+	uint64_t a_thr_uid = a->thr_uid;
+	uint64_t b_thr_uid = b->thr_uid;
+	int ret = (a_thr_uid > b_thr_uid) - (a_thr_uid < b_thr_uid);
+	if (ret == 0) {
+		uint64_t a_thr_discrim = a->thr_discrim;
+		uint64_t b_thr_discrim = b->thr_discrim;
+		ret = (a_thr_discrim > b_thr_discrim) - (a_thr_discrim <
+		    b_thr_discrim);
+		if (ret == 0) {
+			uint64_t a_tctx_uid = a->tctx_uid;
+			uint64_t b_tctx_uid = b->tctx_uid;
+			ret = (a_tctx_uid > b_tctx_uid) - (a_tctx_uid <
+			    b_tctx_uid);
+		}
+	}
+	return ret;
+}
+
+rb_gen(static UNUSED, tctx_tree_, prof_tctx_tree_t, prof_tctx_t,
+    tctx_link, prof_tctx_comp)
+
+static int
+prof_gctx_comp(const prof_gctx_t *a, const prof_gctx_t *b) {
+	unsigned a_len = a->bt.len;
+	unsigned b_len = b->bt.len;
+	unsigned comp_len = (a_len < b_len) ? a_len : b_len;
+	int ret = memcmp(a->bt.vec, b->bt.vec, comp_len * sizeof(void *));
+	if (ret == 0) {
+		ret = (a_len > b_len) - (a_len < b_len);
+	}
+	return ret;
+}
+
+rb_gen(static UNUSED, gctx_tree_, prof_gctx_tree_t, prof_gctx_t, dump_link,
+    prof_gctx_comp)
+
+static int
+prof_tdata_comp(const prof_tdata_t *a, const prof_tdata_t *b) {
+	int ret;
+	uint64_t a_uid = a->thr_uid;
+	uint64_t b_uid = b->thr_uid;
+
+	ret = ((a_uid > b_uid) - (a_uid < b_uid));
+	if (ret == 0) {
+		uint64_t a_discrim = a->thr_discrim;
+		uint64_t b_discrim = b->thr_discrim;
+
+		ret = ((a_discrim > b_discrim) - (a_discrim < b_discrim));
+	}
+	return ret;
+}
+
+rb_gen(static UNUSED, tdata_tree_, prof_tdata_tree_t, prof_tdata_t, tdata_link,
+    prof_tdata_comp)
+
+/******************************************************************************/
+
+void
+prof_alloc_rollback(tsd_t *tsd, prof_tctx_t *tctx, bool updated) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	if (updated) {
+		/*
+		 * Compute a new sample threshold.  This isn't very important in
+		 * practice, because this function is rarely executed, so the
+		 * potential for sample bias is minimal except in contrived
+		 * programs.
+		 */
+		tdata = prof_tdata_get(tsd, true);
+		if (tdata != NULL) {
+			prof_sample_threshold_update(tdata);
+		}
+	}
+
+	if ((uintptr_t)tctx > (uintptr_t)1U) {
+		malloc_mutex_lock(tsd_tsdn(tsd), tctx->tdata->lock);
+		tctx->prepared = false;
+		if (prof_tctx_should_destroy(tsd_tsdn(tsd), tctx)) {
+			prof_tctx_destroy(tsd, tctx);
+		} else {
+			malloc_mutex_unlock(tsd_tsdn(tsd), tctx->tdata->lock);
+		}
+	}
+}
+
+void
+prof_malloc_sample_object(tsdn_t *tsdn, const void *ptr, size_t usize,
+    prof_tctx_t *tctx) {
+	prof_tctx_set(tsdn, ptr, usize, NULL, tctx);
+
+	/* Get the current time and set this in the extent_t. We'll read this
+	 * when free() is called. */
+	nstime_t t = NSTIME_ZERO_INITIALIZER;
+	nstime_update(&t);
+	prof_alloc_time_set(tsdn, ptr, NULL, t);
+
+	malloc_mutex_lock(tsdn, tctx->tdata->lock);
+	tctx->cnts.curobjs++;
+	tctx->cnts.curbytes += usize;
+	if (opt_prof_accum) {
+		tctx->cnts.accumobjs++;
+		tctx->cnts.accumbytes += usize;
+	}
+	tctx->prepared = false;
+	malloc_mutex_unlock(tsdn, tctx->tdata->lock);
+}
+
+static size_t
+prof_log_bt_index(tsd_t *tsd, prof_bt_t *bt) {
+	assert(prof_logging_state == prof_logging_state_started);
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &log_mtx);
+
+	prof_bt_node_t dummy_node;
+	dummy_node.bt = *bt;
+	prof_bt_node_t *node;
+
+	/* See if this backtrace is already cached in the table. */
+	if (ckh_search(&log_bt_node_set, (void *)(&dummy_node),
+	    (void **)(&node), NULL)) {
+		size_t sz = offsetof(prof_bt_node_t, vec) +
+			        (bt->len * sizeof(void *));
+		prof_bt_node_t *new_node = (prof_bt_node_t *)
+		    iallocztm(tsd_tsdn(tsd), sz, sz_size2index(sz), false, NULL,
+		    true, arena_get(TSDN_NULL, 0, true), true);
+		if (log_bt_first == NULL) {
+			log_bt_first = new_node;
+			log_bt_last = new_node;
+		} else {
+			log_bt_last->next = new_node;
+			log_bt_last = new_node;
+		}
+
+		new_node->next = NULL;
+		new_node->index = log_bt_index;
+		/*
+		 * Copy the backtrace: bt is inside a tdata or gctx, which
+		 * might die before prof_log_stop is called.
+		 */
+		new_node->bt.len = bt->len;
+		memcpy(new_node->vec, bt->vec, bt->len * sizeof(void *));
+		new_node->bt.vec = new_node->vec;
+
+		log_bt_index++;
+		ckh_insert(tsd, &log_bt_node_set, (void *)new_node, NULL);
+		return new_node->index;
+	} else {
+		return node->index;
+	}
+}
+static size_t
+prof_log_thr_index(tsd_t *tsd, uint64_t thr_uid, const char *name) {
+	assert(prof_logging_state == prof_logging_state_started);
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &log_mtx);
+
+	prof_thr_node_t dummy_node;
+	dummy_node.thr_uid = thr_uid;
+	prof_thr_node_t *node;
+
+	/* See if this thread is already cached in the table. */
+	if (ckh_search(&log_thr_node_set, (void *)(&dummy_node),
+	    (void **)(&node), NULL)) {
+		size_t sz = offsetof(prof_thr_node_t, name) + strlen(name) + 1;
+		prof_thr_node_t *new_node = (prof_thr_node_t *)
+		    iallocztm(tsd_tsdn(tsd), sz, sz_size2index(sz), false, NULL,
+		    true, arena_get(TSDN_NULL, 0, true), true);
+		if (log_thr_first == NULL) {
+			log_thr_first = new_node;
+			log_thr_last = new_node;
+		} else {
+			log_thr_last->next = new_node;
+			log_thr_last = new_node;
+		}
+
+		new_node->next = NULL;
+		new_node->index = log_thr_index;
+		new_node->thr_uid = thr_uid;
+		strcpy(new_node->name, name);
+
+		log_thr_index++;
+		ckh_insert(tsd, &log_thr_node_set, (void *)new_node, NULL);
+		return new_node->index;
+	} else {
+		return node->index;
+	}
+}
+
+static void
+prof_try_log(tsd_t *tsd, const void *ptr, size_t usize, prof_tctx_t *tctx) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	prof_tdata_t *cons_tdata = prof_tdata_get(tsd, false);
+	if (cons_tdata == NULL) {
+		/*
+		 * We decide not to log these allocations. cons_tdata will be
+		 * NULL only when the current thread is in a weird state (e.g.
+		 * it's being destroyed).
+		 */
+		return;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &log_mtx);
+
+	if (prof_logging_state != prof_logging_state_started) {
+		goto label_done;
+	}
+
+	if (!log_tables_initialized) {
+		bool err1 = ckh_new(tsd, &log_bt_node_set, PROF_CKH_MINITEMS,
+				prof_bt_node_hash, prof_bt_node_keycomp);
+		bool err2 = ckh_new(tsd, &log_thr_node_set, PROF_CKH_MINITEMS,
+				prof_thr_node_hash, prof_thr_node_keycomp);
+		if (err1 || err2) {
+			goto label_done;
+		}
+		log_tables_initialized = true;
+	}
+
+	nstime_t alloc_time = prof_alloc_time_get(tsd_tsdn(tsd), ptr,
+			          (alloc_ctx_t *)NULL);
+	nstime_t free_time = NSTIME_ZERO_INITIALIZER;
+	nstime_update(&free_time);
+
+	size_t sz = sizeof(prof_alloc_node_t);
+	prof_alloc_node_t *new_node = (prof_alloc_node_t *)
+	    iallocztm(tsd_tsdn(tsd), sz, sz_size2index(sz), false, NULL, true,
+	    arena_get(TSDN_NULL, 0, true), true);
+
+	const char *prod_thr_name = (tctx->tdata->thread_name == NULL)?
+				        "" : tctx->tdata->thread_name;
+	const char *cons_thr_name = prof_thread_name_get(tsd);
+
+	prof_bt_t bt;
+	/* Initialize the backtrace, using the buffer in tdata to store it. */
+	bt_init(&bt, cons_tdata->vec);
+	prof_backtrace(&bt);
+	prof_bt_t *cons_bt = &bt;
+
+	/* We haven't destroyed tctx yet, so gctx should be good to read. */
+	prof_bt_t *prod_bt = &tctx->gctx->bt;
+
+	new_node->next = NULL;
+	new_node->alloc_thr_ind = prof_log_thr_index(tsd, tctx->tdata->thr_uid,
+				      prod_thr_name);
+	new_node->free_thr_ind = prof_log_thr_index(tsd, cons_tdata->thr_uid,
+				     cons_thr_name);
+	new_node->alloc_bt_ind = prof_log_bt_index(tsd, prod_bt);
+	new_node->free_bt_ind = prof_log_bt_index(tsd, cons_bt);
+	new_node->alloc_time_ns = nstime_ns(&alloc_time);
+	new_node->free_time_ns = nstime_ns(&free_time);
+	new_node->usize = usize;
+
+	if (log_alloc_first == NULL) {
+		log_alloc_first = new_node;
+		log_alloc_last = new_node;
+	} else {
+		log_alloc_last->next = new_node;
+		log_alloc_last = new_node;
+	}
+
+label_done:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &log_mtx);
+}
+
+void
+prof_free_sampled_object(tsd_t *tsd, const void *ptr, size_t usize,
+    prof_tctx_t *tctx) {
+	malloc_mutex_lock(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	assert(tctx->cnts.curobjs > 0);
+	assert(tctx->cnts.curbytes >= usize);
+	tctx->cnts.curobjs--;
+	tctx->cnts.curbytes -= usize;
+
+	prof_try_log(tsd, ptr, usize, tctx);
+
+	if (prof_tctx_should_destroy(tsd_tsdn(tsd), tctx)) {
+		prof_tctx_destroy(tsd, tctx);
+	} else {
+		malloc_mutex_unlock(tsd_tsdn(tsd), tctx->tdata->lock);
+	}
+}
+
+void
+bt_init(prof_bt_t *bt, void **vec) {
+	cassert(config_prof);
+
+	bt->vec = vec;
+	bt->len = 0;
+}
+
+static void
+prof_enter(tsd_t *tsd, prof_tdata_t *tdata) {
+	cassert(config_prof);
+	assert(tdata == prof_tdata_get(tsd, false));
+
+	if (tdata != NULL) {
+		assert(!tdata->enq);
+		tdata->enq = true;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &bt2gctx_mtx);
+}
+
+static void
+prof_leave(tsd_t *tsd, prof_tdata_t *tdata) {
+	cassert(config_prof);
+	assert(tdata == prof_tdata_get(tsd, false));
+
+	malloc_mutex_unlock(tsd_tsdn(tsd), &bt2gctx_mtx);
+
+	if (tdata != NULL) {
+		bool idump, gdump;
+
+		assert(tdata->enq);
+		tdata->enq = false;
+		idump = tdata->enq_idump;
+		tdata->enq_idump = false;
+		gdump = tdata->enq_gdump;
+		tdata->enq_gdump = false;
+
+		if (idump) {
+			prof_idump(tsd_tsdn(tsd));
+		}
+		if (gdump) {
+			prof_gdump(tsd_tsdn(tsd));
+		}
+	}
+}
+
+#ifdef JEMALLOC_PROF_LIBUNWIND
+void
+prof_backtrace(prof_bt_t *bt) {
+	int nframes;
+
+	cassert(config_prof);
+	assert(bt->len == 0);
+	assert(bt->vec != NULL);
+
+	nframes = unw_backtrace(bt->vec, PROF_BT_MAX);
+	if (nframes <= 0) {
+		return;
+	}
+	bt->len = nframes;
+}
+#elif (defined(JEMALLOC_PROF_LIBGCC))
+static _Unwind_Reason_Code
+prof_unwind_init_callback(struct _Unwind_Context *context, void *arg) {
+	cassert(config_prof);
+
+	return _URC_NO_REASON;
+}
+
+static _Unwind_Reason_Code
+prof_unwind_callback(struct _Unwind_Context *context, void *arg) {
+	prof_unwind_data_t *data = (prof_unwind_data_t *)arg;
+	void *ip;
+
+	cassert(config_prof);
+
+	ip = (void *)_Unwind_GetIP(context);
+	if (ip == NULL) {
+		return _URC_END_OF_STACK;
+	}
+	data->bt->vec[data->bt->len] = ip;
+	data->bt->len++;
+	if (data->bt->len == data->max) {
+		return _URC_END_OF_STACK;
+	}
+
+	return _URC_NO_REASON;
+}
+
+void
+prof_backtrace(prof_bt_t *bt) {
+	prof_unwind_data_t data = {bt, PROF_BT_MAX};
+
+	cassert(config_prof);
+
+	_Unwind_Backtrace(prof_unwind_callback, &data);
+}
+#elif (defined(JEMALLOC_PROF_GCC))
+void
+prof_backtrace(prof_bt_t *bt) {
+#define BT_FRAME(i)							\
+	if ((i) < PROF_BT_MAX) {					\
+		void *p;						\
+		if (__builtin_frame_address(i) == 0) {			\
+			return;						\
+		}							\
+		p = __builtin_return_address(i);			\
+		if (p == NULL) {					\
+			return;						\
+		}							\
+		bt->vec[(i)] = p;					\
+		bt->len = (i) + 1;					\
+	} else {							\
+		return;							\
+	}
+
+	cassert(config_prof);
+
+	BT_FRAME(0)
+	BT_FRAME(1)
+	BT_FRAME(2)
+	BT_FRAME(3)
+	BT_FRAME(4)
+	BT_FRAME(5)
+	BT_FRAME(6)
+	BT_FRAME(7)
+	BT_FRAME(8)
+	BT_FRAME(9)
+
+	BT_FRAME(10)
+	BT_FRAME(11)
+	BT_FRAME(12)
+	BT_FRAME(13)
+	BT_FRAME(14)
+	BT_FRAME(15)
+	BT_FRAME(16)
+	BT_FRAME(17)
+	BT_FRAME(18)
+	BT_FRAME(19)
+
+	BT_FRAME(20)
+	BT_FRAME(21)
+	BT_FRAME(22)
+	BT_FRAME(23)
+	BT_FRAME(24)
+	BT_FRAME(25)
+	BT_FRAME(26)
+	BT_FRAME(27)
+	BT_FRAME(28)
+	BT_FRAME(29)
+
+	BT_FRAME(30)
+	BT_FRAME(31)
+	BT_FRAME(32)
+	BT_FRAME(33)
+	BT_FRAME(34)
+	BT_FRAME(35)
+	BT_FRAME(36)
+	BT_FRAME(37)
+	BT_FRAME(38)
+	BT_FRAME(39)
+
+	BT_FRAME(40)
+	BT_FRAME(41)
+	BT_FRAME(42)
+	BT_FRAME(43)
+	BT_FRAME(44)
+	BT_FRAME(45)
+	BT_FRAME(46)
+	BT_FRAME(47)
+	BT_FRAME(48)
+	BT_FRAME(49)
+
+	BT_FRAME(50)
+	BT_FRAME(51)
+	BT_FRAME(52)
+	BT_FRAME(53)
+	BT_FRAME(54)
+	BT_FRAME(55)
+	BT_FRAME(56)
+	BT_FRAME(57)
+	BT_FRAME(58)
+	BT_FRAME(59)
+
+	BT_FRAME(60)
+	BT_FRAME(61)
+	BT_FRAME(62)
+	BT_FRAME(63)
+	BT_FRAME(64)
+	BT_FRAME(65)
+	BT_FRAME(66)
+	BT_FRAME(67)
+	BT_FRAME(68)
+	BT_FRAME(69)
+
+	BT_FRAME(70)
+	BT_FRAME(71)
+	BT_FRAME(72)
+	BT_FRAME(73)
+	BT_FRAME(74)
+	BT_FRAME(75)
+	BT_FRAME(76)
+	BT_FRAME(77)
+	BT_FRAME(78)
+	BT_FRAME(79)
+
+	BT_FRAME(80)
+	BT_FRAME(81)
+	BT_FRAME(82)
+	BT_FRAME(83)
+	BT_FRAME(84)
+	BT_FRAME(85)
+	BT_FRAME(86)
+	BT_FRAME(87)
+	BT_FRAME(88)
+	BT_FRAME(89)
+
+	BT_FRAME(90)
+	BT_FRAME(91)
+	BT_FRAME(92)
+	BT_FRAME(93)
+	BT_FRAME(94)
+	BT_FRAME(95)
+	BT_FRAME(96)
+	BT_FRAME(97)
+	BT_FRAME(98)
+	BT_FRAME(99)
+
+	BT_FRAME(100)
+	BT_FRAME(101)
+	BT_FRAME(102)
+	BT_FRAME(103)
+	BT_FRAME(104)
+	BT_FRAME(105)
+	BT_FRAME(106)
+	BT_FRAME(107)
+	BT_FRAME(108)
+	BT_FRAME(109)
+
+	BT_FRAME(110)
+	BT_FRAME(111)
+	BT_FRAME(112)
+	BT_FRAME(113)
+	BT_FRAME(114)
+	BT_FRAME(115)
+	BT_FRAME(116)
+	BT_FRAME(117)
+	BT_FRAME(118)
+	BT_FRAME(119)
+
+	BT_FRAME(120)
+	BT_FRAME(121)
+	BT_FRAME(122)
+	BT_FRAME(123)
+	BT_FRAME(124)
+	BT_FRAME(125)
+	BT_FRAME(126)
+	BT_FRAME(127)
+#undef BT_FRAME
+}
+#else
+void
+prof_backtrace(prof_bt_t *bt) {
+	cassert(config_prof);
+	not_reached();
+}
+#endif
+
+static malloc_mutex_t *
+prof_gctx_mutex_choose(void) {
+	unsigned ngctxs = atomic_fetch_add_u(&cum_gctxs, 1, ATOMIC_RELAXED);
+
+	return &gctx_locks[(ngctxs - 1) % PROF_NCTX_LOCKS];
+}
+
+static malloc_mutex_t *
+prof_tdata_mutex_choose(uint64_t thr_uid) {
+	return &tdata_locks[thr_uid % PROF_NTDATA_LOCKS];
+}
+
+static prof_gctx_t *
+prof_gctx_create(tsdn_t *tsdn, prof_bt_t *bt) {
+	/*
+	 * Create a single allocation that has space for vec of length bt->len.
+	 */
+	size_t size = offsetof(prof_gctx_t, vec) + (bt->len * sizeof(void *));
+	prof_gctx_t *gctx = (prof_gctx_t *)iallocztm(tsdn, size,
+	    sz_size2index(size), false, NULL, true, arena_get(TSDN_NULL, 0, true),
+	    true);
+	if (gctx == NULL) {
+		return NULL;
+	}
+	gctx->lock = prof_gctx_mutex_choose();
+	/*
+	 * Set nlimbo to 1, in order to avoid a race condition with
+	 * prof_tctx_destroy()/prof_gctx_try_destroy().
+	 */
+	gctx->nlimbo = 1;
+	tctx_tree_new(&gctx->tctxs);
+	/* Duplicate bt. */
+	memcpy(gctx->vec, bt->vec, bt->len * sizeof(void *));
+	gctx->bt.vec = gctx->vec;
+	gctx->bt.len = bt->len;
+	return gctx;
+}
+
+static void
+prof_gctx_try_destroy(tsd_t *tsd, prof_tdata_t *tdata_self, prof_gctx_t *gctx,
+    prof_tdata_t *tdata) {
+	cassert(config_prof);
+
+	/*
+	 * Check that gctx is still unused by any thread cache before destroying
+	 * it.  prof_lookup() increments gctx->nlimbo in order to avoid a race
+	 * condition with this function, as does prof_tctx_destroy() in order to
+	 * avoid a race between the main body of prof_tctx_destroy() and entry
+	 * into this function.
+	 */
+	prof_enter(tsd, tdata_self);
+	malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+	assert(gctx->nlimbo != 0);
+	if (tctx_tree_empty(&gctx->tctxs) && gctx->nlimbo == 1) {
+		/* Remove gctx from bt2gctx. */
+		if (ckh_remove(tsd, &bt2gctx, &gctx->bt, NULL, NULL)) {
+			not_reached();
+		}
+		prof_leave(tsd, tdata_self);
+		/* Destroy gctx. */
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+		idalloctm(tsd_tsdn(tsd), gctx, NULL, NULL, true, true);
+	} else {
+		/*
+		 * Compensate for increment in prof_tctx_destroy() or
+		 * prof_lookup().
+		 */
+		gctx->nlimbo--;
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+		prof_leave(tsd, tdata_self);
+	}
+}
+
+static bool
+prof_tctx_should_destroy(tsdn_t *tsdn, prof_tctx_t *tctx) {
+	malloc_mutex_assert_owner(tsdn, tctx->tdata->lock);
+
+	if (opt_prof_accum) {
+		return false;
+	}
+	if (tctx->cnts.curobjs != 0) {
+		return false;
+	}
+	if (tctx->prepared) {
+		return false;
+	}
+	return true;
+}
+
+static bool
+prof_gctx_should_destroy(prof_gctx_t *gctx) {
+	if (opt_prof_accum) {
+		return false;
+	}
+	if (!tctx_tree_empty(&gctx->tctxs)) {
+		return false;
+	}
+	if (gctx->nlimbo != 0) {
+		return false;
+	}
+	return true;
+}
+
+static void
+prof_tctx_destroy(tsd_t *tsd, prof_tctx_t *tctx) {
+	prof_tdata_t *tdata = tctx->tdata;
+	prof_gctx_t *gctx = tctx->gctx;
+	bool destroy_tdata, destroy_tctx, destroy_gctx;
+
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	assert(tctx->cnts.curobjs == 0);
+	assert(tctx->cnts.curbytes == 0);
+	assert(!opt_prof_accum);
+	assert(tctx->cnts.accumobjs == 0);
+	assert(tctx->cnts.accumbytes == 0);
+
+	ckh_remove(tsd, &tdata->bt2tctx, &gctx->bt, NULL, NULL);
+	destroy_tdata = prof_tdata_should_destroy(tsd_tsdn(tsd), tdata, false);
+	malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+
+	malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+	switch (tctx->state) {
+	case prof_tctx_state_nominal:
+		tctx_tree_remove(&gctx->tctxs, tctx);
+		destroy_tctx = true;
+		if (prof_gctx_should_destroy(gctx)) {
+			/*
+			 * Increment gctx->nlimbo in order to keep another
+			 * thread from winning the race to destroy gctx while
+			 * this one has gctx->lock dropped.  Without this, it
+			 * would be possible for another thread to:
+			 *
+			 * 1) Sample an allocation associated with gctx.
+			 * 2) Deallocate the sampled object.
+			 * 3) Successfully prof_gctx_try_destroy(gctx).
+			 *
+			 * The result would be that gctx no longer exists by the
+			 * time this thread accesses it in
+			 * prof_gctx_try_destroy().
+			 */
+			gctx->nlimbo++;
+			destroy_gctx = true;
+		} else {
+			destroy_gctx = false;
+		}
+		break;
+	case prof_tctx_state_dumping:
+		/*
+		 * A dumping thread needs tctx to remain valid until dumping
+		 * has finished.  Change state such that the dumping thread will
+		 * complete destruction during a late dump iteration phase.
+		 */
+		tctx->state = prof_tctx_state_purgatory;
+		destroy_tctx = false;
+		destroy_gctx = false;
+		break;
+	default:
+		not_reached();
+		destroy_tctx = false;
+		destroy_gctx = false;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+	if (destroy_gctx) {
+		prof_gctx_try_destroy(tsd, prof_tdata_get(tsd, false), gctx,
+		    tdata);
+	}
+
+	malloc_mutex_assert_not_owner(tsd_tsdn(tsd), tctx->tdata->lock);
+
+	if (destroy_tdata) {
+		prof_tdata_destroy(tsd, tdata, false);
+	}
+
+	if (destroy_tctx) {
+		idalloctm(tsd_tsdn(tsd), tctx, NULL, NULL, true, true);
+	}
+}
+
+static bool
+prof_lookup_global(tsd_t *tsd, prof_bt_t *bt, prof_tdata_t *tdata,
+    void **p_btkey, prof_gctx_t **p_gctx, bool *p_new_gctx) {
+	union {
+		prof_gctx_t	*p;
+		void		*v;
+	} gctx, tgctx;
+	union {
+		prof_bt_t	*p;
+		void		*v;
+	} btkey;
+	bool new_gctx;
+
+	prof_enter(tsd, tdata);
+	if (ckh_search(&bt2gctx, bt, &btkey.v, &gctx.v)) {
+		/* bt has never been seen before.  Insert it. */
+		prof_leave(tsd, tdata);
+		tgctx.p = prof_gctx_create(tsd_tsdn(tsd), bt);
+		if (tgctx.v == NULL) {
+			return true;
+		}
+		prof_enter(tsd, tdata);
+		if (ckh_search(&bt2gctx, bt, &btkey.v, &gctx.v)) {
+			gctx.p = tgctx.p;
+			btkey.p = &gctx.p->bt;
+			if (ckh_insert(tsd, &bt2gctx, btkey.v, gctx.v)) {
+				/* OOM. */
+				prof_leave(tsd, tdata);
+				idalloctm(tsd_tsdn(tsd), gctx.v, NULL, NULL,
+				    true, true);
+				return true;
+			}
+			new_gctx = true;
+		} else {
+			new_gctx = false;
+		}
+	} else {
+		tgctx.v = NULL;
+		new_gctx = false;
+	}
+
+	if (!new_gctx) {
+		/*
+		 * Increment nlimbo, in order to avoid a race condition with
+		 * prof_tctx_destroy()/prof_gctx_try_destroy().
+		 */
+		malloc_mutex_lock(tsd_tsdn(tsd), gctx.p->lock);
+		gctx.p->nlimbo++;
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx.p->lock);
+		new_gctx = false;
+
+		if (tgctx.v != NULL) {
+			/* Lost race to insert. */
+			idalloctm(tsd_tsdn(tsd), tgctx.v, NULL, NULL, true,
+			    true);
+		}
+	}
+	prof_leave(tsd, tdata);
+
+	*p_btkey = btkey.v;
+	*p_gctx = gctx.p;
+	*p_new_gctx = new_gctx;
+	return false;
+}
+
+prof_tctx_t *
+prof_lookup(tsd_t *tsd, prof_bt_t *bt) {
+	union {
+		prof_tctx_t	*p;
+		void		*v;
+	} ret;
+	prof_tdata_t *tdata;
+	bool not_found;
+
+	cassert(config_prof);
+
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return NULL;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), tdata->lock);
+	not_found = ckh_search(&tdata->bt2tctx, bt, NULL, &ret.v);
+	if (!not_found) { /* Note double negative! */
+		ret.p->prepared = true;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+	if (not_found) {
+		void *btkey;
+		prof_gctx_t *gctx;
+		bool new_gctx, error;
+
+		/*
+		 * This thread's cache lacks bt.  Look for it in the global
+		 * cache.
+		 */
+		if (prof_lookup_global(tsd, bt, tdata, &btkey, &gctx,
+		    &new_gctx)) {
+			return NULL;
+		}
+
+		/* Link a prof_tctx_t into gctx for this thread. */
+		ret.v = iallocztm(tsd_tsdn(tsd), sizeof(prof_tctx_t),
+		    sz_size2index(sizeof(prof_tctx_t)), false, NULL, true,
+		    arena_ichoose(tsd, NULL), true);
+		if (ret.p == NULL) {
+			if (new_gctx) {
+				prof_gctx_try_destroy(tsd, tdata, gctx, tdata);
+			}
+			return NULL;
+		}
+		ret.p->tdata = tdata;
+		ret.p->thr_uid = tdata->thr_uid;
+		ret.p->thr_discrim = tdata->thr_discrim;
+		memset(&ret.p->cnts, 0, sizeof(prof_cnt_t));
+		ret.p->gctx = gctx;
+		ret.p->tctx_uid = tdata->tctx_uid_next++;
+		ret.p->prepared = true;
+		ret.p->state = prof_tctx_state_initializing;
+		malloc_mutex_lock(tsd_tsdn(tsd), tdata->lock);
+		error = ckh_insert(tsd, &tdata->bt2tctx, btkey, ret.v);
+		malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+		if (error) {
+			if (new_gctx) {
+				prof_gctx_try_destroy(tsd, tdata, gctx, tdata);
+			}
+			idalloctm(tsd_tsdn(tsd), ret.v, NULL, NULL, true, true);
+			return NULL;
+		}
+		malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+		ret.p->state = prof_tctx_state_nominal;
+		tctx_tree_insert(&gctx->tctxs, ret.p);
+		gctx->nlimbo--;
+		malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+	}
+
+	return ret.p;
+}
+
+/*
+ * The bodies of this function and prof_leakcheck() are compiled out unless heap
+ * profiling is enabled, so that it is possible to compile jemalloc with
+ * floating point support completely disabled.  Avoiding floating point code is
+ * important on memory-constrained systems, but it also enables a workaround for
+ * versions of glibc that don't properly save/restore floating point registers
+ * during dynamic lazy symbol loading (which internally calls into whatever
+ * malloc implementation happens to be integrated into the application).  Note
+ * that some compilers (e.g.  gcc 4.8) may use floating point registers for fast
+ * memory moves, so jemalloc must be compiled with such optimizations disabled
+ * (e.g.
+ * -mno-sse) in order for the workaround to be complete.
+ */
+void
+prof_sample_threshold_update(prof_tdata_t *tdata) {
+#ifdef JEMALLOC_PROF
+	if (!config_prof) {
+		return;
+	}
+
+	if (lg_prof_sample == 0) {
+		tsd_bytes_until_sample_set(tsd_fetch(), 0);
+		return;
+	}
+
+	/*
+	 * Compute sample interval as a geometrically distributed random
+	 * variable with mean (2^lg_prof_sample).
+	 *
+	 *                             __        __
+	 *                             |  log(u)  |                     1
+	 * tdata->bytes_until_sample = | -------- |, where p = ---------------
+	 *                             | log(1-p) |             lg_prof_sample
+	 *                                                     2
+	 *
+	 * For more information on the math, see:
+	 *
+	 *   Non-Uniform Random Variate Generation
+	 *   Luc Devroye
+	 *   Springer-Verlag, New York, 1986
+	 *   pp 500
+	 *   (http://luc.devroye.org/rnbookindex.html)
+	 */
+	uint64_t r = prng_lg_range_u64(&tdata->prng_state, 53);
+	double u = (double)r * (1.0/9007199254740992.0L);
+	uint64_t bytes_until_sample = (uint64_t)(log(u) /
+	    log(1.0 - (1.0 / (double)((uint64_t)1U << lg_prof_sample))))
+	    + (uint64_t)1U;
+	if (bytes_until_sample > SSIZE_MAX) {
+		bytes_until_sample = SSIZE_MAX;
+	}
+	tsd_bytes_until_sample_set(tsd_fetch(), bytes_until_sample);
+
+#endif
+}
+
+#ifdef JEMALLOC_JET
+static prof_tdata_t *
+prof_tdata_count_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *arg) {
+	size_t *tdata_count = (size_t *)arg;
+
+	(*tdata_count)++;
+
+	return NULL;
+}
+
+size_t
+prof_tdata_count(void) {
+	size_t tdata_count = 0;
+	tsdn_t *tsdn;
+
+	tsdn = tsdn_fetch();
+	malloc_mutex_lock(tsdn, &tdatas_mtx);
+	tdata_tree_iter(&tdatas, NULL, prof_tdata_count_iter,
+	    (void *)&tdata_count);
+	malloc_mutex_unlock(tsdn, &tdatas_mtx);
+
+	return tdata_count;
+}
+
+size_t
+prof_bt_count(void) {
+	size_t bt_count;
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+
+	tsd = tsd_fetch();
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return 0;
+	}
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &bt2gctx_mtx);
+	bt_count = ckh_count(&bt2gctx);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &bt2gctx_mtx);
+
+	return bt_count;
+}
+#endif
+
+static int
+prof_dump_open_impl(bool propagate_err, const char *filename) {
+	int fd;
+
+	fd = creat(filename, 0644);
+	if (fd == -1 && !propagate_err) {
+		malloc_printf("<jemalloc>: creat(\"%s\"), 0644) failed\n",
+		    filename);
+		if (opt_abort) {
+			abort();
+		}
+	}
+
+	return fd;
+}
+prof_dump_open_t *JET_MUTABLE prof_dump_open = prof_dump_open_impl;
+
+static bool
+prof_dump_flush(bool propagate_err) {
+	bool ret = false;
+	ssize_t err;
+
+	cassert(config_prof);
+
+	err = malloc_write_fd(prof_dump_fd, prof_dump_buf, prof_dump_buf_end);
+	if (err == -1) {
+		if (!propagate_err) {
+			malloc_write("<jemalloc>: write() failed during heap "
+			    "profile flush\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+		ret = true;
+	}
+	prof_dump_buf_end = 0;
+
+	return ret;
+}
+
+static bool
+prof_dump_close(bool propagate_err) {
+	bool ret;
+
+	assert(prof_dump_fd != -1);
+	ret = prof_dump_flush(propagate_err);
+	close(prof_dump_fd);
+	prof_dump_fd = -1;
+
+	return ret;
+}
+
+static bool
+prof_dump_write(bool propagate_err, const char *s) {
+	size_t i, slen, n;
+
+	cassert(config_prof);
+
+	i = 0;
+	slen = strlen(s);
+	while (i < slen) {
+		/* Flush the buffer if it is full. */
+		if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) {
+			if (prof_dump_flush(propagate_err) && propagate_err) {
+				return true;
+			}
+		}
+
+		if (prof_dump_buf_end + slen - i <= PROF_DUMP_BUFSIZE) {
+			/* Finish writing. */
+			n = slen - i;
+		} else {
+			/* Write as much of s as will fit. */
+			n = PROF_DUMP_BUFSIZE - prof_dump_buf_end;
+		}
+		memcpy(&prof_dump_buf[prof_dump_buf_end], &s[i], n);
+		prof_dump_buf_end += n;
+		i += n;
+	}
+	assert(i == slen);
+
+	return false;
+}
+
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static bool
+prof_dump_printf(bool propagate_err, const char *format, ...) {
+	bool ret;
+	va_list ap;
+	char buf[PROF_PRINTF_BUFSIZE];
+
+	va_start(ap, format);
+	malloc_vsnprintf(buf, sizeof(buf), format, ap);
+	va_end(ap);
+	ret = prof_dump_write(propagate_err, buf);
+
+	return ret;
+}
+
+static void
+prof_tctx_merge_tdata(tsdn_t *tsdn, prof_tctx_t *tctx, prof_tdata_t *tdata) {
+	malloc_mutex_assert_owner(tsdn, tctx->tdata->lock);
+
+	malloc_mutex_lock(tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_initializing:
+		malloc_mutex_unlock(tsdn, tctx->gctx->lock);
+		return;
+	case prof_tctx_state_nominal:
+		tctx->state = prof_tctx_state_dumping;
+		malloc_mutex_unlock(tsdn, tctx->gctx->lock);
+
+		memcpy(&tctx->dump_cnts, &tctx->cnts, sizeof(prof_cnt_t));
+
+		tdata->cnt_summed.curobjs += tctx->dump_cnts.curobjs;
+		tdata->cnt_summed.curbytes += tctx->dump_cnts.curbytes;
+		if (opt_prof_accum) {
+			tdata->cnt_summed.accumobjs +=
+			    tctx->dump_cnts.accumobjs;
+			tdata->cnt_summed.accumbytes +=
+			    tctx->dump_cnts.accumbytes;
+		}
+		break;
+	case prof_tctx_state_dumping:
+	case prof_tctx_state_purgatory:
+		not_reached();
+	}
+}
+
+static void
+prof_tctx_merge_gctx(tsdn_t *tsdn, prof_tctx_t *tctx, prof_gctx_t *gctx) {
+	malloc_mutex_assert_owner(tsdn, gctx->lock);
+
+	gctx->cnt_summed.curobjs += tctx->dump_cnts.curobjs;
+	gctx->cnt_summed.curbytes += tctx->dump_cnts.curbytes;
+	if (opt_prof_accum) {
+		gctx->cnt_summed.accumobjs += tctx->dump_cnts.accumobjs;
+		gctx->cnt_summed.accumbytes += tctx->dump_cnts.accumbytes;
+	}
+}
+
+static prof_tctx_t *
+prof_tctx_merge_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *arg) {
+	tsdn_t *tsdn = (tsdn_t *)arg;
+
+	malloc_mutex_assert_owner(tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_nominal:
+		/* New since dumping started; ignore. */
+		break;
+	case prof_tctx_state_dumping:
+	case prof_tctx_state_purgatory:
+		prof_tctx_merge_gctx(tsdn, tctx, tctx->gctx);
+		break;
+	default:
+		not_reached();
+	}
+
+	return NULL;
+}
+
+struct prof_tctx_dump_iter_arg_s {
+	tsdn_t	*tsdn;
+	bool	propagate_err;
+};
+
+static prof_tctx_t *
+prof_tctx_dump_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *opaque) {
+	struct prof_tctx_dump_iter_arg_s *arg =
+	    (struct prof_tctx_dump_iter_arg_s *)opaque;
+
+	malloc_mutex_assert_owner(arg->tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_initializing:
+	case prof_tctx_state_nominal:
+		/* Not captured by this dump. */
+		break;
+	case prof_tctx_state_dumping:
+	case prof_tctx_state_purgatory:
+		if (prof_dump_printf(arg->propagate_err,
+		    "  t%"FMTu64": %"FMTu64": %"FMTu64" [%"FMTu64": "
+		    "%"FMTu64"]\n", tctx->thr_uid, tctx->dump_cnts.curobjs,
+		    tctx->dump_cnts.curbytes, tctx->dump_cnts.accumobjs,
+		    tctx->dump_cnts.accumbytes)) {
+			return tctx;
+		}
+		break;
+	default:
+		not_reached();
+	}
+	return NULL;
+}
+
+static prof_tctx_t *
+prof_tctx_finish_iter(prof_tctx_tree_t *tctxs, prof_tctx_t *tctx, void *arg) {
+	tsdn_t *tsdn = (tsdn_t *)arg;
+	prof_tctx_t *ret;
+
+	malloc_mutex_assert_owner(tsdn, tctx->gctx->lock);
+
+	switch (tctx->state) {
+	case prof_tctx_state_nominal:
+		/* New since dumping started; ignore. */
+		break;
+	case prof_tctx_state_dumping:
+		tctx->state = prof_tctx_state_nominal;
+		break;
+	case prof_tctx_state_purgatory:
+		ret = tctx;
+		goto label_return;
+	default:
+		not_reached();
+	}
+
+	ret = NULL;
+label_return:
+	return ret;
+}
+
+static void
+prof_dump_gctx_prep(tsdn_t *tsdn, prof_gctx_t *gctx, prof_gctx_tree_t *gctxs) {
+	cassert(config_prof);
+
+	malloc_mutex_lock(tsdn, gctx->lock);
+
+	/*
+	 * Increment nlimbo so that gctx won't go away before dump.
+	 * Additionally, link gctx into the dump list so that it is included in
+	 * prof_dump()'s second pass.
+	 */
+	gctx->nlimbo++;
+	gctx_tree_insert(gctxs, gctx);
+
+	memset(&gctx->cnt_summed, 0, sizeof(prof_cnt_t));
+
+	malloc_mutex_unlock(tsdn, gctx->lock);
+}
+
+struct prof_gctx_merge_iter_arg_s {
+	tsdn_t	*tsdn;
+	size_t	leak_ngctx;
+};
+
+static prof_gctx_t *
+prof_gctx_merge_iter(prof_gctx_tree_t *gctxs, prof_gctx_t *gctx, void *opaque) {
+	struct prof_gctx_merge_iter_arg_s *arg =
+	    (struct prof_gctx_merge_iter_arg_s *)opaque;
+
+	malloc_mutex_lock(arg->tsdn, gctx->lock);
+	tctx_tree_iter(&gctx->tctxs, NULL, prof_tctx_merge_iter,
+	    (void *)arg->tsdn);
+	if (gctx->cnt_summed.curobjs != 0) {
+		arg->leak_ngctx++;
+	}
+	malloc_mutex_unlock(arg->tsdn, gctx->lock);
+
+	return NULL;
+}
+
+static void
+prof_gctx_finish(tsd_t *tsd, prof_gctx_tree_t *gctxs) {
+	prof_tdata_t *tdata = prof_tdata_get(tsd, false);
+	prof_gctx_t *gctx;
+
+	/*
+	 * Standard tree iteration won't work here, because as soon as we
+	 * decrement gctx->nlimbo and unlock gctx, another thread can
+	 * concurrently destroy it, which will corrupt the tree.  Therefore,
+	 * tear down the tree one node at a time during iteration.
+	 */
+	while ((gctx = gctx_tree_first(gctxs)) != NULL) {
+		gctx_tree_remove(gctxs, gctx);
+		malloc_mutex_lock(tsd_tsdn(tsd), gctx->lock);
+		{
+			prof_tctx_t *next;
+
+			next = NULL;
+			do {
+				prof_tctx_t *to_destroy =
+				    tctx_tree_iter(&gctx->tctxs, next,
+				    prof_tctx_finish_iter,
+				    (void *)tsd_tsdn(tsd));
+				if (to_destroy != NULL) {
+					next = tctx_tree_next(&gctx->tctxs,
+					    to_destroy);
+					tctx_tree_remove(&gctx->tctxs,
+					    to_destroy);
+					idalloctm(tsd_tsdn(tsd), to_destroy,
+					    NULL, NULL, true, true);
+				} else {
+					next = NULL;
+				}
+			} while (next != NULL);
+		}
+		gctx->nlimbo--;
+		if (prof_gctx_should_destroy(gctx)) {
+			gctx->nlimbo++;
+			malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+			prof_gctx_try_destroy(tsd, tdata, gctx, tdata);
+		} else {
+			malloc_mutex_unlock(tsd_tsdn(tsd), gctx->lock);
+		}
+	}
+}
+
+struct prof_tdata_merge_iter_arg_s {
+	tsdn_t		*tsdn;
+	prof_cnt_t	cnt_all;
+};
+
+static prof_tdata_t *
+prof_tdata_merge_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *opaque) {
+	struct prof_tdata_merge_iter_arg_s *arg =
+	    (struct prof_tdata_merge_iter_arg_s *)opaque;
+
+	malloc_mutex_lock(arg->tsdn, tdata->lock);
+	if (!tdata->expired) {
+		size_t tabind;
+		union {
+			prof_tctx_t	*p;
+			void		*v;
+		} tctx;
+
+		tdata->dumping = true;
+		memset(&tdata->cnt_summed, 0, sizeof(prof_cnt_t));
+		for (tabind = 0; !ckh_iter(&tdata->bt2tctx, &tabind, NULL,
+		    &tctx.v);) {
+			prof_tctx_merge_tdata(arg->tsdn, tctx.p, tdata);
+		}
+
+		arg->cnt_all.curobjs += tdata->cnt_summed.curobjs;
+		arg->cnt_all.curbytes += tdata->cnt_summed.curbytes;
+		if (opt_prof_accum) {
+			arg->cnt_all.accumobjs += tdata->cnt_summed.accumobjs;
+			arg->cnt_all.accumbytes += tdata->cnt_summed.accumbytes;
+		}
+	} else {
+		tdata->dumping = false;
+	}
+	malloc_mutex_unlock(arg->tsdn, tdata->lock);
+
+	return NULL;
+}
+
+static prof_tdata_t *
+prof_tdata_dump_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *arg) {
+	bool propagate_err = *(bool *)arg;
+
+	if (!tdata->dumping) {
+		return NULL;
+	}
+
+	if (prof_dump_printf(propagate_err,
+	    "  t%"FMTu64": %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]%s%s\n",
+	    tdata->thr_uid, tdata->cnt_summed.curobjs,
+	    tdata->cnt_summed.curbytes, tdata->cnt_summed.accumobjs,
+	    tdata->cnt_summed.accumbytes,
+	    (tdata->thread_name != NULL) ? " " : "",
+	    (tdata->thread_name != NULL) ? tdata->thread_name : "")) {
+		return tdata;
+	}
+	return NULL;
+}
+
+static bool
+prof_dump_header_impl(tsdn_t *tsdn, bool propagate_err,
+    const prof_cnt_t *cnt_all) {
+	bool ret;
+
+	if (prof_dump_printf(propagate_err,
+	    "heap_v2/%"FMTu64"\n"
+	    "  t*: %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]\n",
+	    ((uint64_t)1U << lg_prof_sample), cnt_all->curobjs,
+	    cnt_all->curbytes, cnt_all->accumobjs, cnt_all->accumbytes)) {
+		return true;
+	}
+
+	malloc_mutex_lock(tsdn, &tdatas_mtx);
+	ret = (tdata_tree_iter(&tdatas, NULL, prof_tdata_dump_iter,
+	    (void *)&propagate_err) != NULL);
+	malloc_mutex_unlock(tsdn, &tdatas_mtx);
+	return ret;
+}
+prof_dump_header_t *JET_MUTABLE prof_dump_header = prof_dump_header_impl;
+
+static bool
+prof_dump_gctx(tsdn_t *tsdn, bool propagate_err, prof_gctx_t *gctx,
+    const prof_bt_t *bt, prof_gctx_tree_t *gctxs) {
+	bool ret;
+	unsigned i;
+	struct prof_tctx_dump_iter_arg_s prof_tctx_dump_iter_arg;
+
+	cassert(config_prof);
+	malloc_mutex_assert_owner(tsdn, gctx->lock);
+
+	/* Avoid dumping such gctx's that have no useful data. */
+	if ((!opt_prof_accum && gctx->cnt_summed.curobjs == 0) ||
+	    (opt_prof_accum && gctx->cnt_summed.accumobjs == 0)) {
+		assert(gctx->cnt_summed.curobjs == 0);
+		assert(gctx->cnt_summed.curbytes == 0);
+		assert(gctx->cnt_summed.accumobjs == 0);
+		assert(gctx->cnt_summed.accumbytes == 0);
+		ret = false;
+		goto label_return;
+	}
+
+	if (prof_dump_printf(propagate_err, "@")) {
+		ret = true;
+		goto label_return;
+	}
+	for (i = 0; i < bt->len; i++) {
+		if (prof_dump_printf(propagate_err, " %#"FMTxPTR,
+		    (uintptr_t)bt->vec[i])) {
+			ret = true;
+			goto label_return;
+		}
+	}
+
+	if (prof_dump_printf(propagate_err,
+	    "\n"
+	    "  t*: %"FMTu64": %"FMTu64" [%"FMTu64": %"FMTu64"]\n",
+	    gctx->cnt_summed.curobjs, gctx->cnt_summed.curbytes,
+	    gctx->cnt_summed.accumobjs, gctx->cnt_summed.accumbytes)) {
+		ret = true;
+		goto label_return;
+	}
+
+	prof_tctx_dump_iter_arg.tsdn = tsdn;
+	prof_tctx_dump_iter_arg.propagate_err = propagate_err;
+	if (tctx_tree_iter(&gctx->tctxs, NULL, prof_tctx_dump_iter,
+	    (void *)&prof_tctx_dump_iter_arg) != NULL) {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	return ret;
+}
+
+#ifndef _WIN32
+JEMALLOC_FORMAT_PRINTF(1, 2)
+static int
+prof_open_maps(const char *format, ...) {
+	int mfd;
+	va_list ap;
+	char filename[PATH_MAX + 1];
+
+	va_start(ap, format);
+	malloc_vsnprintf(filename, sizeof(filename), format, ap);
+	va_end(ap);
+
+#if defined(O_CLOEXEC)
+	mfd = open(filename, O_RDONLY | O_CLOEXEC);
+#else
+	mfd = open(filename, O_RDONLY);
+	if (mfd != -1) {
+		fcntl(mfd, F_SETFD, fcntl(mfd, F_GETFD) | FD_CLOEXEC);
+	}
+#endif
+
+	return mfd;
+}
+#endif
+
+static int
+prof_getpid(void) {
+#ifdef _WIN32
+	return GetCurrentProcessId();
+#else
+	return getpid();
+#endif
+}
+
+static bool
+prof_dump_maps(bool propagate_err) {
+	bool ret;
+	int mfd;
+
+	cassert(config_prof);
+#ifdef __FreeBSD__
+	mfd = prof_open_maps("/proc/curproc/map");
+#elif defined(_WIN32)
+	mfd = -1; // Not implemented
+#else
+	{
+		int pid = prof_getpid();
+
+		mfd = prof_open_maps("/proc/%d/task/%d/maps", pid, pid);
+		if (mfd == -1) {
+			mfd = prof_open_maps("/proc/%d/maps", pid);
+		}
+	}
+#endif
+	if (mfd != -1) {
+		ssize_t nread;
+
+		if (prof_dump_write(propagate_err, "\nMAPPED_LIBRARIES:\n") &&
+		    propagate_err) {
+			ret = true;
+			goto label_return;
+		}
+		nread = 0;
+		do {
+			prof_dump_buf_end += nread;
+			if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) {
+				/* Make space in prof_dump_buf before read(). */
+				if (prof_dump_flush(propagate_err) &&
+				    propagate_err) {
+					ret = true;
+					goto label_return;
+				}
+			}
+			nread = malloc_read_fd(mfd,
+			    &prof_dump_buf[prof_dump_buf_end], PROF_DUMP_BUFSIZE
+			    - prof_dump_buf_end);
+		} while (nread > 0);
+	} else {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	if (mfd != -1) {
+		close(mfd);
+	}
+	return ret;
+}
+
+/*
+ * See prof_sample_threshold_update() comment for why the body of this function
+ * is conditionally compiled.
+ */
+static void
+prof_leakcheck(const prof_cnt_t *cnt_all, size_t leak_ngctx,
+    const char *filename) {
+#ifdef JEMALLOC_PROF
+	/*
+	 * Scaling is equivalent AdjustSamples() in jeprof, but the result may
+	 * differ slightly from what jeprof reports, because here we scale the
+	 * summary values, whereas jeprof scales each context individually and
+	 * reports the sums of the scaled values.
+	 */
+	if (cnt_all->curbytes != 0) {
+		double sample_period = (double)((uint64_t)1 << lg_prof_sample);
+		double ratio = (((double)cnt_all->curbytes) /
+		    (double)cnt_all->curobjs) / sample_period;
+		double scale_factor = 1.0 / (1.0 - exp(-ratio));
+		uint64_t curbytes = (uint64_t)round(((double)cnt_all->curbytes)
+		    * scale_factor);
+		uint64_t curobjs = (uint64_t)round(((double)cnt_all->curobjs) *
+		    scale_factor);
+
+		malloc_printf("<jemalloc>: Leak approximation summary: ~%"FMTu64
+		    " byte%s, ~%"FMTu64" object%s, >= %zu context%s\n",
+		    curbytes, (curbytes != 1) ? "s" : "", curobjs, (curobjs !=
+		    1) ? "s" : "", leak_ngctx, (leak_ngctx != 1) ? "s" : "");
+		malloc_printf(
+		    "<jemalloc>: Run jeprof on \"%s\" for leak detail\n",
+		    filename);
+	}
+#endif
+}
+
+struct prof_gctx_dump_iter_arg_s {
+	tsdn_t	*tsdn;
+	bool	propagate_err;
+};
+
+static prof_gctx_t *
+prof_gctx_dump_iter(prof_gctx_tree_t *gctxs, prof_gctx_t *gctx, void *opaque) {
+	prof_gctx_t *ret;
+	struct prof_gctx_dump_iter_arg_s *arg =
+	    (struct prof_gctx_dump_iter_arg_s *)opaque;
+
+	malloc_mutex_lock(arg->tsdn, gctx->lock);
+
+	if (prof_dump_gctx(arg->tsdn, arg->propagate_err, gctx, &gctx->bt,
+	    gctxs)) {
+		ret = gctx;
+		goto label_return;
+	}
+
+	ret = NULL;
+label_return:
+	malloc_mutex_unlock(arg->tsdn, gctx->lock);
+	return ret;
+}
+
+static void
+prof_dump_prep(tsd_t *tsd, prof_tdata_t *tdata,
+    struct prof_tdata_merge_iter_arg_s *prof_tdata_merge_iter_arg,
+    struct prof_gctx_merge_iter_arg_s *prof_gctx_merge_iter_arg,
+    prof_gctx_tree_t *gctxs) {
+	size_t tabind;
+	union {
+		prof_gctx_t	*p;
+		void		*v;
+	} gctx;
+
+	prof_enter(tsd, tdata);
+
+	/*
+	 * Put gctx's in limbo and clear their counters in preparation for
+	 * summing.
+	 */
+	gctx_tree_new(gctxs);
+	for (tabind = 0; !ckh_iter(&bt2gctx, &tabind, NULL, &gctx.v);) {
+		prof_dump_gctx_prep(tsd_tsdn(tsd), gctx.p, gctxs);
+	}
+
+	/*
+	 * Iterate over tdatas, and for the non-expired ones snapshot their tctx
+	 * stats and merge them into the associated gctx's.
+	 */
+	prof_tdata_merge_iter_arg->tsdn = tsd_tsdn(tsd);
+	memset(&prof_tdata_merge_iter_arg->cnt_all, 0, sizeof(prof_cnt_t));
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+	tdata_tree_iter(&tdatas, NULL, prof_tdata_merge_iter,
+	    (void *)prof_tdata_merge_iter_arg);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+
+	/* Merge tctx stats into gctx's. */
+	prof_gctx_merge_iter_arg->tsdn = tsd_tsdn(tsd);
+	prof_gctx_merge_iter_arg->leak_ngctx = 0;
+	gctx_tree_iter(gctxs, NULL, prof_gctx_merge_iter,
+	    (void *)prof_gctx_merge_iter_arg);
+
+	prof_leave(tsd, tdata);
+}
+
+static bool
+prof_dump_file(tsd_t *tsd, bool propagate_err, const char *filename,
+    bool leakcheck, prof_tdata_t *tdata,
+    struct prof_tdata_merge_iter_arg_s *prof_tdata_merge_iter_arg,
+    struct prof_gctx_merge_iter_arg_s *prof_gctx_merge_iter_arg,
+    struct prof_gctx_dump_iter_arg_s *prof_gctx_dump_iter_arg,
+    prof_gctx_tree_t *gctxs) {
+	/* Create dump file. */
+	if ((prof_dump_fd = prof_dump_open(propagate_err, filename)) == -1) {
+		return true;
+	}
+
+	/* Dump profile header. */
+	if (prof_dump_header(tsd_tsdn(tsd), propagate_err,
+	    &prof_tdata_merge_iter_arg->cnt_all)) {
+		goto label_write_error;
+	}
+
+	/* Dump per gctx profile stats. */
+	prof_gctx_dump_iter_arg->tsdn = tsd_tsdn(tsd);
+	prof_gctx_dump_iter_arg->propagate_err = propagate_err;
+	if (gctx_tree_iter(gctxs, NULL, prof_gctx_dump_iter,
+	    (void *)prof_gctx_dump_iter_arg) != NULL) {
+		goto label_write_error;
+	}
+
+	/* Dump /proc/<pid>/maps if possible. */
+	if (prof_dump_maps(propagate_err)) {
+		goto label_write_error;
+	}
+
+	if (prof_dump_close(propagate_err)) {
+		return true;
+	}
+
+	return false;
+label_write_error:
+	prof_dump_close(propagate_err);
+	return true;
+}
+
+static bool
+prof_dump(tsd_t *tsd, bool propagate_err, const char *filename,
+    bool leakcheck) {
+	cassert(config_prof);
+	assert(tsd_reentrancy_level_get(tsd) == 0);
+
+	prof_tdata_t * tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return true;
+	}
+
+	pre_reentrancy(tsd, NULL);
+	malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_mtx);
+
+	prof_gctx_tree_t gctxs;
+	struct prof_tdata_merge_iter_arg_s prof_tdata_merge_iter_arg;
+	struct prof_gctx_merge_iter_arg_s prof_gctx_merge_iter_arg;
+	struct prof_gctx_dump_iter_arg_s prof_gctx_dump_iter_arg;
+	prof_dump_prep(tsd, tdata, &prof_tdata_merge_iter_arg,
+	    &prof_gctx_merge_iter_arg, &gctxs);
+	bool err = prof_dump_file(tsd, propagate_err, filename, leakcheck, tdata,
+	    &prof_tdata_merge_iter_arg, &prof_gctx_merge_iter_arg,
+	    &prof_gctx_dump_iter_arg, &gctxs);
+	prof_gctx_finish(tsd, &gctxs);
+
+	malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_mtx);
+	post_reentrancy(tsd);
+
+	if (err) {
+		return true;
+	}
+
+	if (leakcheck) {
+		prof_leakcheck(&prof_tdata_merge_iter_arg.cnt_all,
+		    prof_gctx_merge_iter_arg.leak_ngctx, filename);
+	}
+	return false;
+}
+
+#ifdef JEMALLOC_JET
+void
+prof_cnt_all(uint64_t *curobjs, uint64_t *curbytes, uint64_t *accumobjs,
+    uint64_t *accumbytes) {
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+	struct prof_tdata_merge_iter_arg_s prof_tdata_merge_iter_arg;
+	struct prof_gctx_merge_iter_arg_s prof_gctx_merge_iter_arg;
+	prof_gctx_tree_t gctxs;
+
+	tsd = tsd_fetch();
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		if (curobjs != NULL) {
+			*curobjs = 0;
+		}
+		if (curbytes != NULL) {
+			*curbytes = 0;
+		}
+		if (accumobjs != NULL) {
+			*accumobjs = 0;
+		}
+		if (accumbytes != NULL) {
+			*accumbytes = 0;
+		}
+		return;
+	}
+
+	prof_dump_prep(tsd, tdata, &prof_tdata_merge_iter_arg,
+	    &prof_gctx_merge_iter_arg, &gctxs);
+	prof_gctx_finish(tsd, &gctxs);
+
+	if (curobjs != NULL) {
+		*curobjs = prof_tdata_merge_iter_arg.cnt_all.curobjs;
+	}
+	if (curbytes != NULL) {
+		*curbytes = prof_tdata_merge_iter_arg.cnt_all.curbytes;
+	}
+	if (accumobjs != NULL) {
+		*accumobjs = prof_tdata_merge_iter_arg.cnt_all.accumobjs;
+	}
+	if (accumbytes != NULL) {
+		*accumbytes = prof_tdata_merge_iter_arg.cnt_all.accumbytes;
+	}
+}
+#endif
+
+#define DUMP_FILENAME_BUFSIZE	(PATH_MAX + 1)
+#define VSEQ_INVALID		UINT64_C(0xffffffffffffffff)
+static void
+prof_dump_filename(char *filename, char v, uint64_t vseq) {
+	cassert(config_prof);
+
+	if (vseq != VSEQ_INVALID) {
+	        /* "<prefix>.<pid>.<seq>.v<vseq>.heap" */
+		malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE,
+		    "%s.%d.%"FMTu64".%c%"FMTu64".heap",
+		    opt_prof_prefix, prof_getpid(), prof_dump_seq, v, vseq);
+	} else {
+	        /* "<prefix>.<pid>.<seq>.<v>.heap" */
+		malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE,
+		    "%s.%d.%"FMTu64".%c.heap",
+		    opt_prof_prefix, prof_getpid(), prof_dump_seq, v);
+	}
+	prof_dump_seq++;
+}
+
+static void
+prof_fdump(void) {
+	tsd_t *tsd;
+	char filename[DUMP_FILENAME_BUFSIZE];
+
+	cassert(config_prof);
+	assert(opt_prof_final);
+	assert(opt_prof_prefix[0] != '\0');
+
+	if (!prof_booted) {
+		return;
+	}
+	tsd = tsd_fetch();
+	assert(tsd_reentrancy_level_get(tsd) == 0);
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+	prof_dump_filename(filename, 'f', VSEQ_INVALID);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+	prof_dump(tsd, false, filename, opt_prof_leak);
+}
+
+bool
+prof_accum_init(tsdn_t *tsdn, prof_accum_t *prof_accum) {
+	cassert(config_prof);
+
+#ifndef JEMALLOC_ATOMIC_U64
+	if (malloc_mutex_init(&prof_accum->mtx, "prof_accum",
+	    WITNESS_RANK_PROF_ACCUM, malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+	prof_accum->accumbytes = 0;
+#else
+	atomic_store_u64(&prof_accum->accumbytes, 0, ATOMIC_RELAXED);
+#endif
+	return false;
+}
+
+void
+prof_idump(tsdn_t *tsdn) {
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	if (!prof_booted || tsdn_null(tsdn) || !prof_active_get_unlocked()) {
+		return;
+	}
+	tsd = tsdn_tsd(tsdn);
+	if (tsd_reentrancy_level_get(tsd) > 0) {
+		return;
+	}
+
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return;
+	}
+	if (tdata->enq) {
+		tdata->enq_idump = true;
+		return;
+	}
+
+	if (opt_prof_prefix[0] != '\0') {
+		char filename[PATH_MAX + 1];
+		malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		prof_dump_filename(filename, 'i', prof_dump_iseq);
+		prof_dump_iseq++;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		prof_dump(tsd, false, filename, false);
+	}
+}
+
+bool
+prof_mdump(tsd_t *tsd, const char *filename) {
+	cassert(config_prof);
+	assert(tsd_reentrancy_level_get(tsd) == 0);
+
+	if (!opt_prof || !prof_booted) {
+		return true;
+	}
+	char filename_buf[DUMP_FILENAME_BUFSIZE];
+	if (filename == NULL) {
+		/* No filename specified, so automatically generate one. */
+		if (opt_prof_prefix[0] == '\0') {
+			return true;
+		}
+		malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		prof_dump_filename(filename_buf, 'm', prof_dump_mseq);
+		prof_dump_mseq++;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_seq_mtx);
+		filename = filename_buf;
+	}
+	return prof_dump(tsd, true, filename, false);
+}
+
+void
+prof_gdump(tsdn_t *tsdn) {
+	tsd_t *tsd;
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	if (!prof_booted || tsdn_null(tsdn) || !prof_active_get_unlocked()) {
+		return;
+	}
+	tsd = tsdn_tsd(tsdn);
+	if (tsd_reentrancy_level_get(tsd) > 0) {
+		return;
+	}
+
+	tdata = prof_tdata_get(tsd, false);
+	if (tdata == NULL) {
+		return;
+	}
+	if (tdata->enq) {
+		tdata->enq_gdump = true;
+		return;
+	}
+
+	if (opt_prof_prefix[0] != '\0') {
+		char filename[DUMP_FILENAME_BUFSIZE];
+		malloc_mutex_lock(tsdn, &prof_dump_seq_mtx);
+		prof_dump_filename(filename, 'u', prof_dump_useq);
+		prof_dump_useq++;
+		malloc_mutex_unlock(tsdn, &prof_dump_seq_mtx);
+		prof_dump(tsd, false, filename, false);
+	}
+}
+
+static void
+prof_bt_hash(const void *key, size_t r_hash[2]) {
+	prof_bt_t *bt = (prof_bt_t *)key;
+
+	cassert(config_prof);
+
+	hash(bt->vec, bt->len * sizeof(void *), 0x94122f33U, r_hash);
+}
+
+static bool
+prof_bt_keycomp(const void *k1, const void *k2) {
+	const prof_bt_t *bt1 = (prof_bt_t *)k1;
+	const prof_bt_t *bt2 = (prof_bt_t *)k2;
+
+	cassert(config_prof);
+
+	if (bt1->len != bt2->len) {
+		return false;
+	}
+	return (memcmp(bt1->vec, bt2->vec, bt1->len * sizeof(void *)) == 0);
+}
+
+static void
+prof_bt_node_hash(const void *key, size_t r_hash[2]) {
+	const prof_bt_node_t *bt_node = (prof_bt_node_t *)key;
+	prof_bt_hash((void *)(&bt_node->bt), r_hash);
+}
+
+static bool
+prof_bt_node_keycomp(const void *k1, const void *k2) {
+	const prof_bt_node_t *bt_node1 = (prof_bt_node_t *)k1;
+	const prof_bt_node_t *bt_node2 = (prof_bt_node_t *)k2;
+	return prof_bt_keycomp((void *)(&bt_node1->bt),
+	    (void *)(&bt_node2->bt));
+}
+
+static void
+prof_thr_node_hash(const void *key, size_t r_hash[2]) {
+	const prof_thr_node_t *thr_node = (prof_thr_node_t *)key;
+	hash(&thr_node->thr_uid, sizeof(uint64_t), 0x94122f35U, r_hash);
+}
+
+static bool
+prof_thr_node_keycomp(const void *k1, const void *k2) {
+	const prof_thr_node_t *thr_node1 = (prof_thr_node_t *)k1;
+	const prof_thr_node_t *thr_node2 = (prof_thr_node_t *)k2;
+	return thr_node1->thr_uid == thr_node2->thr_uid;
+}
+
+static uint64_t
+prof_thr_uid_alloc(tsdn_t *tsdn) {
+	uint64_t thr_uid;
+
+	malloc_mutex_lock(tsdn, &next_thr_uid_mtx);
+	thr_uid = next_thr_uid;
+	next_thr_uid++;
+	malloc_mutex_unlock(tsdn, &next_thr_uid_mtx);
+
+	return thr_uid;
+}
+
+static prof_tdata_t *
+prof_tdata_init_impl(tsd_t *tsd, uint64_t thr_uid, uint64_t thr_discrim,
+    char *thread_name, bool active) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	/* Initialize an empty cache for this thread. */
+	tdata = (prof_tdata_t *)iallocztm(tsd_tsdn(tsd), sizeof(prof_tdata_t),
+	    sz_size2index(sizeof(prof_tdata_t)), false, NULL, true,
+	    arena_get(TSDN_NULL, 0, true), true);
+	if (tdata == NULL) {
+		return NULL;
+	}
+
+	tdata->lock = prof_tdata_mutex_choose(thr_uid);
+	tdata->thr_uid = thr_uid;
+	tdata->thr_discrim = thr_discrim;
+	tdata->thread_name = thread_name;
+	tdata->attached = true;
+	tdata->expired = false;
+	tdata->tctx_uid_next = 0;
+
+	if (ckh_new(tsd, &tdata->bt2tctx, PROF_CKH_MINITEMS, prof_bt_hash,
+	    prof_bt_keycomp)) {
+		idalloctm(tsd_tsdn(tsd), tdata, NULL, NULL, true, true);
+		return NULL;
+	}
+
+	tdata->prng_state = (uint64_t)(uintptr_t)tdata;
+	prof_sample_threshold_update(tdata);
+
+	tdata->enq = false;
+	tdata->enq_idump = false;
+	tdata->enq_gdump = false;
+
+	tdata->dumping = false;
+	tdata->active = active;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+	tdata_tree_insert(&tdatas, tdata);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+
+	return tdata;
+}
+
+prof_tdata_t *
+prof_tdata_init(tsd_t *tsd) {
+	return prof_tdata_init_impl(tsd, prof_thr_uid_alloc(tsd_tsdn(tsd)), 0,
+	    NULL, prof_thread_active_init_get(tsd_tsdn(tsd)));
+}
+
+static bool
+prof_tdata_should_destroy_unlocked(prof_tdata_t *tdata, bool even_if_attached) {
+	if (tdata->attached && !even_if_attached) {
+		return false;
+	}
+	if (ckh_count(&tdata->bt2tctx) != 0) {
+		return false;
+	}
+	return true;
+}
+
+static bool
+prof_tdata_should_destroy(tsdn_t *tsdn, prof_tdata_t *tdata,
+    bool even_if_attached) {
+	malloc_mutex_assert_owner(tsdn, tdata->lock);
+
+	return prof_tdata_should_destroy_unlocked(tdata, even_if_attached);
+}
+
+static void
+prof_tdata_destroy_locked(tsd_t *tsd, prof_tdata_t *tdata,
+    bool even_if_attached) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &tdatas_mtx);
+
+	tdata_tree_remove(&tdatas, tdata);
+
+	assert(prof_tdata_should_destroy_unlocked(tdata, even_if_attached));
+
+	if (tdata->thread_name != NULL) {
+		idalloctm(tsd_tsdn(tsd), tdata->thread_name, NULL, NULL, true,
+		    true);
+	}
+	ckh_delete(tsd, &tdata->bt2tctx);
+	idalloctm(tsd_tsdn(tsd), tdata, NULL, NULL, true, true);
+}
+
+static void
+prof_tdata_destroy(tsd_t *tsd, prof_tdata_t *tdata, bool even_if_attached) {
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+	prof_tdata_destroy_locked(tsd, tdata, even_if_attached);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+}
+
+static void
+prof_tdata_detach(tsd_t *tsd, prof_tdata_t *tdata) {
+	bool destroy_tdata;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), tdata->lock);
+	if (tdata->attached) {
+		destroy_tdata = prof_tdata_should_destroy(tsd_tsdn(tsd), tdata,
+		    true);
+		/*
+		 * Only detach if !destroy_tdata, because detaching would allow
+		 * another thread to win the race to destroy tdata.
+		 */
+		if (!destroy_tdata) {
+			tdata->attached = false;
+		}
+		tsd_prof_tdata_set(tsd, NULL);
+	} else {
+		destroy_tdata = false;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), tdata->lock);
+	if (destroy_tdata) {
+		prof_tdata_destroy(tsd, tdata, true);
+	}
+}
+
+prof_tdata_t *
+prof_tdata_reinit(tsd_t *tsd, prof_tdata_t *tdata) {
+	uint64_t thr_uid = tdata->thr_uid;
+	uint64_t thr_discrim = tdata->thr_discrim + 1;
+	char *thread_name = (tdata->thread_name != NULL) ?
+	    prof_thread_name_alloc(tsd_tsdn(tsd), tdata->thread_name) : NULL;
+	bool active = tdata->active;
+
+	prof_tdata_detach(tsd, tdata);
+	return prof_tdata_init_impl(tsd, thr_uid, thr_discrim, thread_name,
+	    active);
+}
+
+static bool
+prof_tdata_expire(tsdn_t *tsdn, prof_tdata_t *tdata) {
+	bool destroy_tdata;
+
+	malloc_mutex_lock(tsdn, tdata->lock);
+	if (!tdata->expired) {
+		tdata->expired = true;
+		destroy_tdata = tdata->attached ? false :
+		    prof_tdata_should_destroy(tsdn, tdata, false);
+	} else {
+		destroy_tdata = false;
+	}
+	malloc_mutex_unlock(tsdn, tdata->lock);
+
+	return destroy_tdata;
+}
+
+static prof_tdata_t *
+prof_tdata_reset_iter(prof_tdata_tree_t *tdatas, prof_tdata_t *tdata,
+    void *arg) {
+	tsdn_t *tsdn = (tsdn_t *)arg;
+
+	return (prof_tdata_expire(tsdn, tdata) ? tdata : NULL);
+}
+
+void
+prof_reset(tsd_t *tsd, size_t lg_sample) {
+	prof_tdata_t *next;
+
+	assert(lg_sample < (sizeof(uint64_t) << 3));
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &prof_dump_mtx);
+	malloc_mutex_lock(tsd_tsdn(tsd), &tdatas_mtx);
+
+	lg_prof_sample = lg_sample;
+
+	next = NULL;
+	do {
+		prof_tdata_t *to_destroy = tdata_tree_iter(&tdatas, next,
+		    prof_tdata_reset_iter, (void *)tsd);
+		if (to_destroy != NULL) {
+			next = tdata_tree_next(&tdatas, to_destroy);
+			prof_tdata_destroy_locked(tsd, to_destroy, false);
+		} else {
+			next = NULL;
+		}
+	} while (next != NULL);
+
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tdatas_mtx);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &prof_dump_mtx);
+}
+
+void
+prof_tdata_cleanup(tsd_t *tsd) {
+	prof_tdata_t *tdata;
+
+	if (!config_prof) {
+		return;
+	}
+
+	tdata = tsd_prof_tdata_get(tsd);
+	if (tdata != NULL) {
+		prof_tdata_detach(tsd, tdata);
+	}
+}
+
+bool
+prof_active_get(tsdn_t *tsdn) {
+	bool prof_active_current;
+
+	malloc_mutex_lock(tsdn, &prof_active_mtx);
+	prof_active_current = prof_active;
+	malloc_mutex_unlock(tsdn, &prof_active_mtx);
+	return prof_active_current;
+}
+
+bool
+prof_active_set(tsdn_t *tsdn, bool active) {
+	bool prof_active_old;
+
+	malloc_mutex_lock(tsdn, &prof_active_mtx);
+	prof_active_old = prof_active;
+	prof_active = active;
+	malloc_mutex_unlock(tsdn, &prof_active_mtx);
+	return prof_active_old;
+}
+
+#ifdef JEMALLOC_JET
+size_t
+prof_log_bt_count(void) {
+	size_t cnt = 0;
+	prof_bt_node_t *node = log_bt_first;
+	while (node != NULL) {
+		cnt++;
+		node = node->next;
+	}
+	return cnt;
+}
+
+size_t
+prof_log_alloc_count(void) {
+	size_t cnt = 0;
+	prof_alloc_node_t *node = log_alloc_first;
+	while (node != NULL) {
+		cnt++;
+		node = node->next;
+	}
+	return cnt;
+}
+
+size_t
+prof_log_thr_count(void) {
+	size_t cnt = 0;
+	prof_thr_node_t *node = log_thr_first;
+	while (node != NULL) {
+		cnt++;
+		node = node->next;
+	}
+	return cnt;
+}
+
+bool
+prof_log_is_logging(void) {
+	return prof_logging_state == prof_logging_state_started;
+}
+
+bool
+prof_log_rep_check(void) {
+	if (prof_logging_state == prof_logging_state_stopped
+	    && log_tables_initialized) {
+		return true;
+	}
+
+	if (log_bt_last != NULL && log_bt_last->next != NULL) {
+		return true;
+	}
+	if (log_thr_last != NULL && log_thr_last->next != NULL) {
+		return true;
+	}
+	if (log_alloc_last != NULL && log_alloc_last->next != NULL) {
+		return true;
+	}
+
+	size_t bt_count = prof_log_bt_count();
+	size_t thr_count = prof_log_thr_count();
+	size_t alloc_count = prof_log_alloc_count();
+
+
+	if (prof_logging_state == prof_logging_state_stopped) {
+		if (bt_count != 0 || thr_count != 0 || alloc_count || 0) {
+			return true;
+		}
+	}
+
+	prof_alloc_node_t *node = log_alloc_first;
+	while (node != NULL) {
+		if (node->alloc_bt_ind >= bt_count) {
+			return true;
+		}
+		if (node->free_bt_ind >= bt_count) {
+			return true;
+		}
+		if (node->alloc_thr_ind >= thr_count) {
+			return true;
+		}
+		if (node->free_thr_ind >= thr_count) {
+			return true;
+		}
+		if (node->alloc_time_ns > node->free_time_ns) {
+			return true;
+		}
+		node = node->next;
+	}
+
+	return false;
+}
+
+void
+prof_log_dummy_set(bool new_value) {
+	prof_log_dummy = new_value;
+}
+#endif
+
+bool
+prof_log_start(tsdn_t *tsdn, const char *filename) {
+	if (!opt_prof || !prof_booted) {
+		return true;
+	}
+
+	bool ret = false;
+	size_t buf_size = PATH_MAX + 1;
+
+	malloc_mutex_lock(tsdn, &log_mtx);
+
+	if (prof_logging_state != prof_logging_state_stopped) {
+		ret = true;
+	} else if (filename == NULL) {
+		/* Make default name. */
+		malloc_snprintf(log_filename, buf_size, "%s.%d.%"FMTu64".json",
+		    opt_prof_prefix, prof_getpid(), log_seq);
+		log_seq++;
+		prof_logging_state = prof_logging_state_started;
+	} else if (strlen(filename) >= buf_size) {
+		ret = true;
+	} else {
+		strcpy(log_filename, filename);
+		prof_logging_state = prof_logging_state_started;
+	}
+
+	if (!ret) {
+		nstime_update(&log_start_timestamp);
+	}
+
+	malloc_mutex_unlock(tsdn, &log_mtx);
+
+	return ret;
+}
+
+/* Used as an atexit function to stop logging on exit. */
+static void
+prof_log_stop_final(void) {
+	tsd_t *tsd = tsd_fetch();
+	prof_log_stop(tsd_tsdn(tsd));
+}
+
+struct prof_emitter_cb_arg_s {
+	int fd;
+	ssize_t ret;
+};
+
+static void
+prof_emitter_write_cb(void *opaque, const char *to_write) {
+	struct prof_emitter_cb_arg_s *arg =
+	    (struct prof_emitter_cb_arg_s *)opaque;
+	size_t bytes = strlen(to_write);
+#ifdef JEMALLOC_JET
+	if (prof_log_dummy) {
+		return;
+	}
+#endif
+	arg->ret = write(arg->fd, (void *)to_write, bytes);
+}
+
+/*
+ * prof_log_emit_{...} goes through the appropriate linked list, emitting each
+ * node to the json and deallocating it.
+ */
+static void
+prof_log_emit_threads(tsd_t *tsd, emitter_t *emitter) {
+	emitter_json_array_kv_begin(emitter, "threads");
+	prof_thr_node_t *thr_node = log_thr_first;
+	prof_thr_node_t *thr_old_node;
+	while (thr_node != NULL) {
+		emitter_json_object_begin(emitter);
+
+		emitter_json_kv(emitter, "thr_uid", emitter_type_uint64,
+		    &thr_node->thr_uid);
+
+		char *thr_name = thr_node->name;
+
+		emitter_json_kv(emitter, "thr_name", emitter_type_string,
+		    &thr_name);
+
+		emitter_json_object_end(emitter);
+		thr_old_node = thr_node;
+		thr_node = thr_node->next;
+		idalloc(tsd, thr_old_node);
+	}
+	emitter_json_array_end(emitter);
+}
+
+static void
+prof_log_emit_traces(tsd_t *tsd, emitter_t *emitter) {
+	emitter_json_array_kv_begin(emitter, "stack_traces");
+	prof_bt_node_t *bt_node = log_bt_first;
+	prof_bt_node_t *bt_old_node;
+	/*
+	 * Calculate how many hex digits we need: twice number of bytes, two for
+	 * "0x", and then one more for terminating '\0'.
+	 */
+	char buf[2 * sizeof(intptr_t) + 3];
+	size_t buf_sz = sizeof(buf);
+	while (bt_node != NULL) {
+		emitter_json_array_begin(emitter);
+		size_t i;
+		for (i = 0; i < bt_node->bt.len; i++) {
+			malloc_snprintf(buf, buf_sz, "%p", bt_node->bt.vec[i]);
+			char *trace_str = buf;
+			emitter_json_value(emitter, emitter_type_string,
+			    &trace_str);
+		}
+		emitter_json_array_end(emitter);
+
+		bt_old_node = bt_node;
+		bt_node = bt_node->next;
+		idalloc(tsd, bt_old_node);
+	}
+	emitter_json_array_end(emitter);
+}
+
+static void
+prof_log_emit_allocs(tsd_t *tsd, emitter_t *emitter) {
+	emitter_json_array_kv_begin(emitter, "allocations");
+	prof_alloc_node_t *alloc_node = log_alloc_first;
+	prof_alloc_node_t *alloc_old_node;
+	while (alloc_node != NULL) {
+		emitter_json_object_begin(emitter);
+
+		emitter_json_kv(emitter, "alloc_thread", emitter_type_size,
+		    &alloc_node->alloc_thr_ind);
+
+		emitter_json_kv(emitter, "free_thread", emitter_type_size,
+		    &alloc_node->free_thr_ind);
+
+		emitter_json_kv(emitter, "alloc_trace", emitter_type_size,
+		    &alloc_node->alloc_bt_ind);
+
+		emitter_json_kv(emitter, "free_trace", emitter_type_size,
+		    &alloc_node->free_bt_ind);
+
+		emitter_json_kv(emitter, "alloc_timestamp",
+		    emitter_type_uint64, &alloc_node->alloc_time_ns);
+
+		emitter_json_kv(emitter, "free_timestamp", emitter_type_uint64,
+		    &alloc_node->free_time_ns);
+
+		emitter_json_kv(emitter, "usize", emitter_type_uint64,
+		    &alloc_node->usize);
+
+		emitter_json_object_end(emitter);
+
+		alloc_old_node = alloc_node;
+		alloc_node = alloc_node->next;
+		idalloc(tsd, alloc_old_node);
+	}
+	emitter_json_array_end(emitter);
+}
+
+static void
+prof_log_emit_metadata(emitter_t *emitter) {
+	emitter_json_object_kv_begin(emitter, "info");
+
+	nstime_t now = NSTIME_ZERO_INITIALIZER;
+
+	nstime_update(&now);
+	uint64_t ns = nstime_ns(&now) - nstime_ns(&log_start_timestamp);
+	emitter_json_kv(emitter, "duration", emitter_type_uint64, &ns);
+
+	char *vers = JEMALLOC_VERSION;
+	emitter_json_kv(emitter, "version",
+	    emitter_type_string, &vers);
+
+	emitter_json_kv(emitter, "lg_sample_rate",
+	    emitter_type_int, &lg_prof_sample);
+
+	int pid = prof_getpid();
+	emitter_json_kv(emitter, "pid", emitter_type_int, &pid);
+
+	emitter_json_object_end(emitter);
+}
+
+
+bool
+prof_log_stop(tsdn_t *tsdn) {
+	if (!opt_prof || !prof_booted) {
+		return true;
+	}
+
+	tsd_t *tsd = tsdn_tsd(tsdn);
+	malloc_mutex_lock(tsdn, &log_mtx);
+
+	if (prof_logging_state != prof_logging_state_started) {
+		malloc_mutex_unlock(tsdn, &log_mtx);
+		return true;
+	}
+
+	/*
+	 * Set the state to dumping. We'll set it to stopped when we're done.
+	 * Since other threads won't be able to start/stop/log when the state is
+	 * dumping, we don't have to hold the lock during the whole method.
+	 */
+	prof_logging_state = prof_logging_state_dumping;
+	malloc_mutex_unlock(tsdn, &log_mtx);
+
+
+	emitter_t emitter;
+
+	/* Create a file. */
+
+	int fd;
+#ifdef JEMALLOC_JET
+	if (prof_log_dummy) {
+		fd = 0;
+	} else {
+		fd = creat(log_filename, 0644);
+	}
+#else
+	fd = creat(log_filename, 0644);
+#endif
+
+	if (fd == -1) {
+		malloc_printf("<jemalloc>: creat() for log file \"%s\" "
+			      " failed with %d\n", log_filename, errno);
+		if (opt_abort) {
+			abort();
+		}
+		return true;
+	}
+
+	/* Emit to json. */
+	struct prof_emitter_cb_arg_s arg;
+	arg.fd = fd;
+	emitter_init(&emitter, emitter_output_json, &prof_emitter_write_cb,
+	    (void *)(&arg));
+
+	emitter_begin(&emitter);
+	prof_log_emit_metadata(&emitter);
+	prof_log_emit_threads(tsd, &emitter);
+	prof_log_emit_traces(tsd, &emitter);
+	prof_log_emit_allocs(tsd, &emitter);
+	emitter_end(&emitter);
+
+	/* Reset global state. */
+	if (log_tables_initialized) {
+		ckh_delete(tsd, &log_bt_node_set);
+		ckh_delete(tsd, &log_thr_node_set);
+	}
+	log_tables_initialized = false;
+	log_bt_index = 0;
+	log_thr_index = 0;
+	log_bt_first = NULL;
+	log_bt_last = NULL;
+	log_thr_first = NULL;
+	log_thr_last = NULL;
+	log_alloc_first = NULL;
+	log_alloc_last = NULL;
+
+	malloc_mutex_lock(tsdn, &log_mtx);
+	prof_logging_state = prof_logging_state_stopped;
+	malloc_mutex_unlock(tsdn, &log_mtx);
+
+#ifdef JEMALLOC_JET
+	if (prof_log_dummy) {
+		return false;
+	}
+#endif
+	return close(fd);
+}
+
+const char *
+prof_thread_name_get(tsd_t *tsd) {
+	prof_tdata_t *tdata;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return "";
+	}
+	return (tdata->thread_name != NULL ? tdata->thread_name : "");
+}
+
+static char *
+prof_thread_name_alloc(tsdn_t *tsdn, const char *thread_name) {
+	char *ret;
+	size_t size;
+
+	if (thread_name == NULL) {
+		return NULL;
+	}
+
+	size = strlen(thread_name) + 1;
+	if (size == 1) {
+		return "";
+	}
+
+	ret = iallocztm(tsdn, size, sz_size2index(size), false, NULL, true,
+	    arena_get(TSDN_NULL, 0, true), true);
+	if (ret == NULL) {
+		return NULL;
+	}
+	memcpy(ret, thread_name, size);
+	return ret;
+}
+
+int
+prof_thread_name_set(tsd_t *tsd, const char *thread_name) {
+	prof_tdata_t *tdata;
+	unsigned i;
+	char *s;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return EAGAIN;
+	}
+
+	/* Validate input. */
+	if (thread_name == NULL) {
+		return EFAULT;
+	}
+	for (i = 0; thread_name[i] != '\0'; i++) {
+		char c = thread_name[i];
+		if (!isgraph(c) && !isblank(c)) {
+			return EFAULT;
+		}
+	}
+
+	s = prof_thread_name_alloc(tsd_tsdn(tsd), thread_name);
+	if (s == NULL) {
+		return EAGAIN;
+	}
+
+	if (tdata->thread_name != NULL) {
+		idalloctm(tsd_tsdn(tsd), tdata->thread_name, NULL, NULL, true,
+		    true);
+		tdata->thread_name = NULL;
+	}
+	if (strlen(s) > 0) {
+		tdata->thread_name = s;
+	}
+	return 0;
+}
+
+bool
+prof_thread_active_get(tsd_t *tsd) {
+	prof_tdata_t *tdata;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return false;
+	}
+	return tdata->active;
+}
+
+bool
+prof_thread_active_set(tsd_t *tsd, bool active) {
+	prof_tdata_t *tdata;
+
+	tdata = prof_tdata_get(tsd, true);
+	if (tdata == NULL) {
+		return true;
+	}
+	tdata->active = active;
+	return false;
+}
+
+bool
+prof_thread_active_init_get(tsdn_t *tsdn) {
+	bool active_init;
+
+	malloc_mutex_lock(tsdn, &prof_thread_active_init_mtx);
+	active_init = prof_thread_active_init;
+	malloc_mutex_unlock(tsdn, &prof_thread_active_init_mtx);
+	return active_init;
+}
+
+bool
+prof_thread_active_init_set(tsdn_t *tsdn, bool active_init) {
+	bool active_init_old;
+
+	malloc_mutex_lock(tsdn, &prof_thread_active_init_mtx);
+	active_init_old = prof_thread_active_init;
+	prof_thread_active_init = active_init;
+	malloc_mutex_unlock(tsdn, &prof_thread_active_init_mtx);
+	return active_init_old;
+}
+
+bool
+prof_gdump_get(tsdn_t *tsdn) {
+	bool prof_gdump_current;
+
+	malloc_mutex_lock(tsdn, &prof_gdump_mtx);
+	prof_gdump_current = prof_gdump_val;
+	malloc_mutex_unlock(tsdn, &prof_gdump_mtx);
+	return prof_gdump_current;
+}
+
+bool
+prof_gdump_set(tsdn_t *tsdn, bool gdump) {
+	bool prof_gdump_old;
+
+	malloc_mutex_lock(tsdn, &prof_gdump_mtx);
+	prof_gdump_old = prof_gdump_val;
+	prof_gdump_val = gdump;
+	malloc_mutex_unlock(tsdn, &prof_gdump_mtx);
+	return prof_gdump_old;
+}
+
+void
+prof_boot0(void) {
+	cassert(config_prof);
+
+	memcpy(opt_prof_prefix, PROF_PREFIX_DEFAULT,
+	    sizeof(PROF_PREFIX_DEFAULT));
+}
+
+void
+prof_boot1(void) {
+	cassert(config_prof);
+
+	/*
+	 * opt_prof must be in its final state before any arenas are
+	 * initialized, so this function must be executed early.
+	 */
+
+	if (opt_prof_leak && !opt_prof) {
+		/*
+		 * Enable opt_prof, but in such a way that profiles are never
+		 * automatically dumped.
+		 */
+		opt_prof = true;
+		opt_prof_gdump = false;
+	} else if (opt_prof) {
+		if (opt_lg_prof_interval >= 0) {
+			prof_interval = (((uint64_t)1U) <<
+			    opt_lg_prof_interval);
+		}
+	}
+}
+
+bool
+prof_boot2(tsd_t *tsd) {
+	cassert(config_prof);
+
+	if (opt_prof) {
+		unsigned i;
+
+		lg_prof_sample = opt_lg_prof_sample;
+
+		prof_active = opt_prof_active;
+		if (malloc_mutex_init(&prof_active_mtx, "prof_active",
+		    WITNESS_RANK_PROF_ACTIVE, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		prof_gdump_val = opt_prof_gdump;
+		if (malloc_mutex_init(&prof_gdump_mtx, "prof_gdump",
+		    WITNESS_RANK_PROF_GDUMP, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		prof_thread_active_init = opt_prof_thread_active_init;
+		if (malloc_mutex_init(&prof_thread_active_init_mtx,
+		    "prof_thread_active_init",
+		    WITNESS_RANK_PROF_THREAD_ACTIVE_INIT,
+		    malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (ckh_new(tsd, &bt2gctx, PROF_CKH_MINITEMS, prof_bt_hash,
+		    prof_bt_keycomp)) {
+			return true;
+		}
+		if (malloc_mutex_init(&bt2gctx_mtx, "prof_bt2gctx",
+		    WITNESS_RANK_PROF_BT2GCTX, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		tdata_tree_new(&tdatas);
+		if (malloc_mutex_init(&tdatas_mtx, "prof_tdatas",
+		    WITNESS_RANK_PROF_TDATAS, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		next_thr_uid = 0;
+		if (malloc_mutex_init(&next_thr_uid_mtx, "prof_next_thr_uid",
+		    WITNESS_RANK_PROF_NEXT_THR_UID, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (malloc_mutex_init(&prof_dump_seq_mtx, "prof_dump_seq",
+		    WITNESS_RANK_PROF_DUMP_SEQ, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+		if (malloc_mutex_init(&prof_dump_mtx, "prof_dump",
+		    WITNESS_RANK_PROF_DUMP, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (opt_prof_final && opt_prof_prefix[0] != '\0' &&
+		    atexit(prof_fdump) != 0) {
+			malloc_write("<jemalloc>: Error in atexit()\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+
+		if (opt_prof_log) {
+			prof_log_start(tsd_tsdn(tsd), NULL);
+		}
+
+		if (atexit(prof_log_stop_final) != 0) {
+			malloc_write("<jemalloc>: Error in atexit() "
+				     "for logging\n");
+			if (opt_abort) {
+				abort();
+			}
+		}
+
+		if (malloc_mutex_init(&log_mtx, "prof_log",
+		    WITNESS_RANK_PROF_LOG, malloc_mutex_rank_exclusive)) {
+			return true;
+		}
+
+		if (ckh_new(tsd, &log_bt_node_set, PROF_CKH_MINITEMS,
+		    prof_bt_node_hash, prof_bt_node_keycomp)) {
+			return true;
+		}
+
+		if (ckh_new(tsd, &log_thr_node_set, PROF_CKH_MINITEMS,
+		    prof_thr_node_hash, prof_thr_node_keycomp)) {
+			return true;
+		}
+
+		log_tables_initialized = true;
+
+		gctx_locks = (malloc_mutex_t *)base_alloc(tsd_tsdn(tsd),
+		    b0get(), PROF_NCTX_LOCKS * sizeof(malloc_mutex_t),
+		    CACHELINE);
+		if (gctx_locks == NULL) {
+			return true;
+		}
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			if (malloc_mutex_init(&gctx_locks[i], "prof_gctx",
+			    WITNESS_RANK_PROF_GCTX,
+			    malloc_mutex_rank_exclusive)) {
+				return true;
+			}
+		}
+
+		tdata_locks = (malloc_mutex_t *)base_alloc(tsd_tsdn(tsd),
+		    b0get(), PROF_NTDATA_LOCKS * sizeof(malloc_mutex_t),
+		    CACHELINE);
+		if (tdata_locks == NULL) {
+			return true;
+		}
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			if (malloc_mutex_init(&tdata_locks[i], "prof_tdata",
+			    WITNESS_RANK_PROF_TDATA,
+			    malloc_mutex_rank_exclusive)) {
+				return true;
+			}
+		}
+#ifdef JEMALLOC_PROF_LIBGCC
+		/*
+		 * Cause the backtracing machinery to allocate its internal
+		 * state before enabling profiling.
+		 */
+		_Unwind_Backtrace(prof_unwind_init_callback, NULL);
+#endif
+	}
+	prof_booted = true;
+
+	return false;
+}
+
+void
+prof_prefork0(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		unsigned i;
+
+		malloc_mutex_prefork(tsdn, &prof_dump_mtx);
+		malloc_mutex_prefork(tsdn, &bt2gctx_mtx);
+		malloc_mutex_prefork(tsdn, &tdatas_mtx);
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			malloc_mutex_prefork(tsdn, &tdata_locks[i]);
+		}
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			malloc_mutex_prefork(tsdn, &gctx_locks[i]);
+		}
+	}
+}
+
+void
+prof_prefork1(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		malloc_mutex_prefork(tsdn, &prof_active_mtx);
+		malloc_mutex_prefork(tsdn, &prof_dump_seq_mtx);
+		malloc_mutex_prefork(tsdn, &prof_gdump_mtx);
+		malloc_mutex_prefork(tsdn, &next_thr_uid_mtx);
+		malloc_mutex_prefork(tsdn, &prof_thread_active_init_mtx);
+	}
+}
+
+void
+prof_postfork_parent(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		unsigned i;
+
+		malloc_mutex_postfork_parent(tsdn,
+		    &prof_thread_active_init_mtx);
+		malloc_mutex_postfork_parent(tsdn, &next_thr_uid_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_gdump_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_dump_seq_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_active_mtx);
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			malloc_mutex_postfork_parent(tsdn, &gctx_locks[i]);
+		}
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			malloc_mutex_postfork_parent(tsdn, &tdata_locks[i]);
+		}
+		malloc_mutex_postfork_parent(tsdn, &tdatas_mtx);
+		malloc_mutex_postfork_parent(tsdn, &bt2gctx_mtx);
+		malloc_mutex_postfork_parent(tsdn, &prof_dump_mtx);
+	}
+}
+
+void
+prof_postfork_child(tsdn_t *tsdn) {
+	if (config_prof && opt_prof) {
+		unsigned i;
+
+		malloc_mutex_postfork_child(tsdn, &prof_thread_active_init_mtx);
+		malloc_mutex_postfork_child(tsdn, &next_thr_uid_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_gdump_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_dump_seq_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_active_mtx);
+		for (i = 0; i < PROF_NCTX_LOCKS; i++) {
+			malloc_mutex_postfork_child(tsdn, &gctx_locks[i]);
+		}
+		for (i = 0; i < PROF_NTDATA_LOCKS; i++) {
+			malloc_mutex_postfork_child(tsdn, &tdata_locks[i]);
+		}
+		malloc_mutex_postfork_child(tsdn, &tdatas_mtx);
+		malloc_mutex_postfork_child(tsdn, &bt2gctx_mtx);
+		malloc_mutex_postfork_child(tsdn, &prof_dump_mtx);
+	}
+}
+
+/******************************************************************************/
diff --git a/memkind-1.10.0/jemalloc/src/rtree.c b/memkind-1.10.0/jemalloc/src/rtree.c
new file mode 100644
index 0000000..4ae41fe
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/rtree.c
@@ -0,0 +1,320 @@
+#define JEMALLOC_RTREE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/mutex.h"
+
+/*
+ * Only the most significant bits of keys passed to rtree_{read,write}() are
+ * used.
+ */
+bool
+rtree_new(rtree_t *rtree, bool zeroed) {
+#ifdef JEMALLOC_JET
+	if (!zeroed) {
+		memset(rtree, 0, sizeof(rtree_t)); /* Clear root. */
+	}
+#else
+	assert(zeroed);
+#endif
+
+	if (malloc_mutex_init(&rtree->init_lock, "rtree", WITNESS_RANK_RTREE,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	return false;
+}
+
+static rtree_node_elm_t *
+rtree_node_alloc_impl(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	return (rtree_node_elm_t *)base_alloc(tsdn, b0get(), nelms *
+	    sizeof(rtree_node_elm_t), CACHELINE);
+}
+rtree_node_alloc_t *JET_MUTABLE rtree_node_alloc = rtree_node_alloc_impl;
+
+static void
+rtree_node_dalloc_impl(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *node) {
+	/* Nodes are never deleted during normal operation. */
+	not_reached();
+}
+rtree_node_dalloc_t *JET_MUTABLE rtree_node_dalloc =
+    rtree_node_dalloc_impl;
+
+static rtree_leaf_elm_t *
+rtree_leaf_alloc_impl(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	return (rtree_leaf_elm_t *)base_alloc(tsdn, b0get(), nelms *
+	    sizeof(rtree_leaf_elm_t), CACHELINE);
+}
+rtree_leaf_alloc_t *JET_MUTABLE rtree_leaf_alloc = rtree_leaf_alloc_impl;
+
+static void
+rtree_leaf_dalloc_impl(tsdn_t *tsdn, rtree_t *rtree, rtree_leaf_elm_t *leaf) {
+	/* Leaves are never deleted during normal operation. */
+	not_reached();
+}
+rtree_leaf_dalloc_t *JET_MUTABLE rtree_leaf_dalloc =
+    rtree_leaf_dalloc_impl;
+
+#ifdef JEMALLOC_JET
+#  if RTREE_HEIGHT > 1
+static void
+rtree_delete_subtree(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *subtree,
+    unsigned level) {
+	size_t nchildren = ZU(1) << rtree_levels[level].bits;
+	if (level + 2 < RTREE_HEIGHT) {
+		for (size_t i = 0; i < nchildren; i++) {
+			rtree_node_elm_t *node =
+			    (rtree_node_elm_t *)atomic_load_p(&subtree[i].child,
+			    ATOMIC_RELAXED);
+			if (node != NULL) {
+				rtree_delete_subtree(tsdn, rtree, node, level +
+				    1);
+			}
+		}
+	} else {
+		for (size_t i = 0; i < nchildren; i++) {
+			rtree_leaf_elm_t *leaf =
+			    (rtree_leaf_elm_t *)atomic_load_p(&subtree[i].child,
+			    ATOMIC_RELAXED);
+			if (leaf != NULL) {
+				rtree_leaf_dalloc(tsdn, rtree, leaf);
+			}
+		}
+	}
+
+	if (subtree != rtree->root) {
+		rtree_node_dalloc(tsdn, rtree, subtree);
+	}
+}
+#  endif
+
+void
+rtree_delete(tsdn_t *tsdn, rtree_t *rtree) {
+#  if RTREE_HEIGHT > 1
+	rtree_delete_subtree(tsdn, rtree, rtree->root, 0);
+#  endif
+}
+#endif
+
+static rtree_node_elm_t *
+rtree_node_init(tsdn_t *tsdn, rtree_t *rtree, unsigned level,
+    atomic_p_t *elmp) {
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+	/*
+	 * If *elmp is non-null, then it was initialized with the init lock
+	 * held, so we can get by with 'relaxed' here.
+	 */
+	rtree_node_elm_t *node = atomic_load_p(elmp, ATOMIC_RELAXED);
+	if (node == NULL) {
+		node = rtree_node_alloc(tsdn, rtree, ZU(1) <<
+		    rtree_levels[level].bits);
+		if (node == NULL) {
+			malloc_mutex_unlock(tsdn, &rtree->init_lock);
+			return NULL;
+		}
+		/*
+		 * Even though we hold the lock, a later reader might not; we
+		 * need release semantics.
+		 */
+		atomic_store_p(elmp, node, ATOMIC_RELEASE);
+	}
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+
+	return node;
+}
+
+static rtree_leaf_elm_t *
+rtree_leaf_init(tsdn_t *tsdn, rtree_t *rtree, atomic_p_t *elmp) {
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+	/*
+	 * If *elmp is non-null, then it was initialized with the init lock
+	 * held, so we can get by with 'relaxed' here.
+	 */
+	rtree_leaf_elm_t *leaf = atomic_load_p(elmp, ATOMIC_RELAXED);
+	if (leaf == NULL) {
+		leaf = rtree_leaf_alloc(tsdn, rtree, ZU(1) <<
+		    rtree_levels[RTREE_HEIGHT-1].bits);
+		if (leaf == NULL) {
+			malloc_mutex_unlock(tsdn, &rtree->init_lock);
+			return NULL;
+		}
+		/*
+		 * Even though we hold the lock, a later reader might not; we
+		 * need release semantics.
+		 */
+		atomic_store_p(elmp, leaf, ATOMIC_RELEASE);
+	}
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+
+	return leaf;
+}
+
+static bool
+rtree_node_valid(rtree_node_elm_t *node) {
+	return ((uintptr_t)node != (uintptr_t)0);
+}
+
+static bool
+rtree_leaf_valid(rtree_leaf_elm_t *leaf) {
+	return ((uintptr_t)leaf != (uintptr_t)0);
+}
+
+static rtree_node_elm_t *
+rtree_child_node_tryread(rtree_node_elm_t *elm, bool dependent) {
+	rtree_node_elm_t *node;
+
+	if (dependent) {
+		node = (rtree_node_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_RELAXED);
+	} else {
+		node = (rtree_node_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_ACQUIRE);
+	}
+
+	assert(!dependent || node != NULL);
+	return node;
+}
+
+static rtree_node_elm_t *
+rtree_child_node_read(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *elm,
+    unsigned level, bool dependent) {
+	rtree_node_elm_t *node;
+
+	node = rtree_child_node_tryread(elm, dependent);
+	if (!dependent && unlikely(!rtree_node_valid(node))) {
+		node = rtree_node_init(tsdn, rtree, level + 1, &elm->child);
+	}
+	assert(!dependent || node != NULL);
+	return node;
+}
+
+static rtree_leaf_elm_t *
+rtree_child_leaf_tryread(rtree_node_elm_t *elm, bool dependent) {
+	rtree_leaf_elm_t *leaf;
+
+	if (dependent) {
+		leaf = (rtree_leaf_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_RELAXED);
+	} else {
+		leaf = (rtree_leaf_elm_t *)atomic_load_p(&elm->child,
+		    ATOMIC_ACQUIRE);
+	}
+
+	assert(!dependent || leaf != NULL);
+	return leaf;
+}
+
+static rtree_leaf_elm_t *
+rtree_child_leaf_read(tsdn_t *tsdn, rtree_t *rtree, rtree_node_elm_t *elm,
+    unsigned level, bool dependent) {
+	rtree_leaf_elm_t *leaf;
+
+	leaf = rtree_child_leaf_tryread(elm, dependent);
+	if (!dependent && unlikely(!rtree_leaf_valid(leaf))) {
+		leaf = rtree_leaf_init(tsdn, rtree, &elm->child);
+	}
+	assert(!dependent || leaf != NULL);
+	return leaf;
+}
+
+rtree_leaf_elm_t *
+rtree_leaf_elm_lookup_hard(tsdn_t *tsdn, rtree_t *rtree, rtree_ctx_t *rtree_ctx,
+    uintptr_t key, bool dependent, bool init_missing) {
+	rtree_node_elm_t *node;
+	rtree_leaf_elm_t *leaf;
+#if RTREE_HEIGHT > 1
+	node = rtree->root;
+#else
+	leaf = rtree->root;
+#endif
+
+	if (config_debug) {
+		uintptr_t leafkey = rtree_leafkey(key);
+		for (unsigned i = 0; i < RTREE_CTX_NCACHE; i++) {
+			assert(rtree_ctx->cache[i].leafkey != leafkey);
+		}
+		for (unsigned i = 0; i < RTREE_CTX_NCACHE_L2; i++) {
+			assert(rtree_ctx->l2_cache[i].leafkey != leafkey);
+		}
+	}
+
+#define RTREE_GET_CHILD(level) {					\
+		assert(level < RTREE_HEIGHT-1);				\
+		if (level != 0 && !dependent &&				\
+		    unlikely(!rtree_node_valid(node))) {		\
+			return NULL;					\
+		}							\
+		uintptr_t subkey = rtree_subkey(key, level);		\
+		if (level + 2 < RTREE_HEIGHT) {				\
+			node = init_missing ?				\
+			    rtree_child_node_read(tsdn, rtree,		\
+			    &node[subkey], level, dependent) :		\
+			    rtree_child_node_tryread(&node[subkey],	\
+			    dependent);					\
+		} else {						\
+			leaf = init_missing ?				\
+			    rtree_child_leaf_read(tsdn, rtree,		\
+			    &node[subkey], level, dependent) :		\
+			    rtree_child_leaf_tryread(&node[subkey],	\
+			    dependent);					\
+		}							\
+	}
+	/*
+	 * Cache replacement upon hard lookup (i.e. L1 & L2 rtree cache miss):
+	 * (1) evict last entry in L2 cache; (2) move the collision slot from L1
+	 * cache down to L2; and 3) fill L1.
+	 */
+#define RTREE_GET_LEAF(level) {						\
+		assert(level == RTREE_HEIGHT-1);			\
+		if (!dependent && unlikely(!rtree_leaf_valid(leaf))) {	\
+			return NULL;					\
+		}							\
+		if (RTREE_CTX_NCACHE_L2 > 1) {				\
+			memmove(&rtree_ctx->l2_cache[1],		\
+			    &rtree_ctx->l2_cache[0],			\
+			    sizeof(rtree_ctx_cache_elm_t) *		\
+			    (RTREE_CTX_NCACHE_L2 - 1));			\
+		}							\
+		size_t slot = rtree_cache_direct_map(key);		\
+		rtree_ctx->l2_cache[0].leafkey =			\
+		    rtree_ctx->cache[slot].leafkey;			\
+		rtree_ctx->l2_cache[0].leaf =				\
+		    rtree_ctx->cache[slot].leaf;			\
+		uintptr_t leafkey = rtree_leafkey(key);			\
+		rtree_ctx->cache[slot].leafkey = leafkey;		\
+		rtree_ctx->cache[slot].leaf = leaf;			\
+		uintptr_t subkey = rtree_subkey(key, level);		\
+		return &leaf[subkey];					\
+	}
+	if (RTREE_HEIGHT > 1) {
+		RTREE_GET_CHILD(0)
+	}
+	if (RTREE_HEIGHT > 2) {
+		RTREE_GET_CHILD(1)
+	}
+	if (RTREE_HEIGHT > 3) {
+		for (unsigned i = 2; i < RTREE_HEIGHT-1; i++) {
+			RTREE_GET_CHILD(i)
+		}
+	}
+	RTREE_GET_LEAF(RTREE_HEIGHT-1)
+#undef RTREE_GET_CHILD
+#undef RTREE_GET_LEAF
+	not_reached();
+}
+
+void
+rtree_ctx_data_init(rtree_ctx_t *ctx) {
+	for (unsigned i = 0; i < RTREE_CTX_NCACHE; i++) {
+		rtree_ctx_cache_elm_t *cache = &ctx->cache[i];
+		cache->leafkey = RTREE_LEAFKEY_INVALID;
+		cache->leaf = NULL;
+	}
+	for (unsigned i = 0; i < RTREE_CTX_NCACHE_L2; i++) {
+		rtree_ctx_cache_elm_t *cache = &ctx->l2_cache[i];
+		cache->leafkey = RTREE_LEAFKEY_INVALID;
+		cache->leaf = NULL;
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/src/safety_check.c b/memkind-1.10.0/jemalloc/src/safety_check.c
new file mode 100644
index 0000000..804155d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/safety_check.c
@@ -0,0 +1,24 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+static void (*safety_check_abort)(const char *message);
+
+void safety_check_set_abort(void (*abort_fn)(const char *)) {
+	safety_check_abort = abort_fn;
+}
+
+void safety_check_fail(const char *format, ...) {
+	char buf[MALLOC_PRINTF_BUFSIZE];
+
+	va_list ap;
+	va_start(ap, format);
+	malloc_vsnprintf(buf, MALLOC_PRINTF_BUFSIZE, format, ap);
+	va_end(ap);
+
+	if (safety_check_abort == NULL) {
+		malloc_write(buf);
+		abort();
+	} else {
+		safety_check_abort(buf);
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/src/sc.c b/memkind-1.10.0/jemalloc/src/sc.c
new file mode 100644
index 0000000..6cfcc1c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/sc.c
@@ -0,0 +1,313 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/sc.h"
+
+/*
+ * This module computes the size classes used to satisfy allocations.  The logic
+ * here was ported more or less line-by-line from a shell script, and because of
+ * that is not the most idiomatic C.  Eventually we should fix this, but for now
+ * at least the damage is compartmentalized to this file.
+ */
+
+sc_data_t sc_data_global;
+
+static size_t
+reg_size_compute(int lg_base, int lg_delta, int ndelta) {
+	return (ZU(1) << lg_base) + (ZU(ndelta) << lg_delta);
+}
+
+/* Returns the number of pages in the slab. */
+static int
+slab_size(int lg_page, int lg_base, int lg_delta, int ndelta) {
+	size_t page = (ZU(1) << lg_page);
+	size_t reg_size = reg_size_compute(lg_base, lg_delta, ndelta);
+
+	size_t try_slab_size = page;
+	size_t try_nregs = try_slab_size / reg_size;
+	size_t perfect_slab_size = 0;
+	bool perfect = false;
+	/*
+	 * This loop continues until we find the least common multiple of the
+	 * page size and size class size.  Size classes are all of the form
+	 * base + ndelta * delta == (ndelta + base/ndelta) * delta, which is
+	 * (ndelta + ngroup) * delta.  The way we choose slabbing strategies
+	 * means that delta is at most the page size and ndelta < ngroup.  So
+	 * the loop executes for at most 2 * ngroup - 1 iterations, which is
+	 * also the bound on the number of pages in a slab chosen by default.
+	 * With the current default settings, this is at most 7.
+	 */
+	while (!perfect) {
+		perfect_slab_size = try_slab_size;
+		size_t perfect_nregs = try_nregs;
+		try_slab_size += page;
+		try_nregs = try_slab_size / reg_size;
+		if (perfect_slab_size == perfect_nregs * reg_size) {
+			perfect = true;
+		}
+	}
+	return (int)(perfect_slab_size / page);
+}
+
+static void
+size_class(
+    /* Output. */
+    sc_t *sc,
+    /* Configuration decisions. */
+    int lg_max_lookup, int lg_page, int lg_ngroup,
+    /* Inputs specific to the size class. */
+    int index, int lg_base, int lg_delta, int ndelta) {
+	sc->index = index;
+	sc->lg_base = lg_base;
+	sc->lg_delta = lg_delta;
+	sc->ndelta = ndelta;
+	sc->psz = (reg_size_compute(lg_base, lg_delta, ndelta)
+	    % (ZU(1) << lg_page) == 0);
+	size_t size = (ZU(1) << lg_base) + (ZU(ndelta) << lg_delta);
+	if (index == 0) {
+		assert(!sc->psz);
+	}
+	if (size < (ZU(1) << (lg_page + lg_ngroup))) {
+		sc->bin = true;
+		sc->pgs = slab_size(lg_page, lg_base, lg_delta, ndelta);
+	} else {
+		sc->bin = false;
+		sc->pgs = 0;
+	}
+	if (size <= (ZU(1) << lg_max_lookup)) {
+		sc->lg_delta_lookup = lg_delta;
+	} else {
+		sc->lg_delta_lookup = 0;
+	}
+}
+
+static void
+size_classes(
+    /* Output. */
+    sc_data_t *sc_data,
+    /* Determined by the system. */
+    size_t lg_ptr_size, int lg_quantum,
+    /* Configuration decisions. */
+    int lg_tiny_min, int lg_max_lookup, int lg_page, int lg_ngroup) {
+	int ptr_bits = (1 << lg_ptr_size) * 8;
+	int ngroup = (1 << lg_ngroup);
+	int ntiny = 0;
+	int nlbins = 0;
+	int lg_tiny_maxclass = (unsigned)-1;
+	int nbins = 0;
+	int npsizes = 0;
+
+	int index = 0;
+
+	int ndelta = 0;
+	int lg_base = lg_tiny_min;
+	int lg_delta = lg_base;
+
+	/* Outputs that we update as we go. */
+	size_t lookup_maxclass = 0;
+	size_t small_maxclass = 0;
+	int lg_large_minclass = 0;
+	size_t large_maxclass = 0;
+
+	/* Tiny size classes. */
+	while (lg_base < lg_quantum) {
+		sc_t *sc = &sc_data->sc[index];
+		size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+		    lg_base, lg_delta, ndelta);
+		if (sc->lg_delta_lookup != 0) {
+			nlbins = index + 1;
+		}
+		if (sc->psz) {
+			npsizes++;
+		}
+		if (sc->bin) {
+			nbins++;
+		}
+		ntiny++;
+		/* Final written value is correct. */
+		lg_tiny_maxclass = lg_base;
+		index++;
+		lg_delta = lg_base;
+		lg_base++;
+	}
+
+	/* First non-tiny (pseudo) group. */
+	if (ntiny != 0) {
+		sc_t *sc = &sc_data->sc[index];
+		/*
+		 * See the note in sc.h; the first non-tiny size class has an
+		 * unusual encoding.
+		 */
+		lg_base--;
+		ndelta = 1;
+		size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+		    lg_base, lg_delta, ndelta);
+		index++;
+		lg_base++;
+		lg_delta++;
+		if (sc->psz) {
+			npsizes++;
+		}
+		if (sc->bin) {
+			nbins++;
+		}
+	}
+	while (ndelta < ngroup) {
+		sc_t *sc = &sc_data->sc[index];
+		size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+		    lg_base, lg_delta, ndelta);
+		index++;
+		ndelta++;
+		if (sc->psz) {
+			npsizes++;
+		}
+		if (sc->bin) {
+			nbins++;
+		}
+	}
+
+	/* All remaining groups. */
+	lg_base = lg_base + lg_ngroup;
+	while (lg_base < ptr_bits - 1) {
+		ndelta = 1;
+		int ndelta_limit;
+		if (lg_base == ptr_bits - 2) {
+			ndelta_limit = ngroup - 1;
+		} else {
+			ndelta_limit = ngroup;
+		}
+		while (ndelta <= ndelta_limit) {
+			sc_t *sc = &sc_data->sc[index];
+			size_class(sc, lg_max_lookup, lg_page, lg_ngroup, index,
+			    lg_base, lg_delta, ndelta);
+			if (sc->lg_delta_lookup != 0) {
+				nlbins = index + 1;
+				/* Final written value is correct. */
+				lookup_maxclass = (ZU(1) << lg_base)
+				    + (ZU(ndelta) << lg_delta);
+			}
+			if (sc->psz) {
+				npsizes++;
+			}
+			if (sc->bin) {
+				nbins++;
+				/* Final written value is correct. */
+				small_maxclass = (ZU(1) << lg_base)
+				    + (ZU(ndelta) << lg_delta);
+				if (lg_ngroup > 0) {
+					lg_large_minclass = lg_base + 1;
+				} else {
+					lg_large_minclass = lg_base + 2;
+				}
+			}
+			large_maxclass = (ZU(1) << lg_base)
+			    + (ZU(ndelta) << lg_delta);
+			index++;
+			ndelta++;
+		}
+		lg_base++;
+		lg_delta++;
+	}
+	/* Additional outputs. */
+	int nsizes = index;
+	unsigned lg_ceil_nsizes = lg_ceil(nsizes);
+
+	/* Fill in the output data. */
+	sc_data->ntiny = ntiny;
+	sc_data->nlbins = nlbins;
+	sc_data->nbins = nbins;
+	sc_data->nsizes = nsizes;
+	sc_data->lg_ceil_nsizes = lg_ceil_nsizes;
+	sc_data->npsizes = npsizes;
+	sc_data->lg_tiny_maxclass = lg_tiny_maxclass;
+	sc_data->lookup_maxclass = lookup_maxclass;
+	sc_data->small_maxclass = small_maxclass;
+	sc_data->lg_large_minclass = lg_large_minclass;
+	sc_data->large_minclass = (ZU(1) << lg_large_minclass);
+	sc_data->large_maxclass = large_maxclass;
+
+	/*
+	 * We compute these values in two ways:
+	 *   - Incrementally, as above.
+	 *   - In macros, in sc.h.
+	 * The computation is easier when done incrementally, but putting it in
+	 * a constant makes it available to the fast paths without having to
+	 * touch the extra global cacheline.  We assert, however, that the two
+	 * computations are equivalent.
+	 */
+	assert(sc_data->npsizes == SC_NPSIZES);
+	assert(sc_data->lg_tiny_maxclass == SC_LG_TINY_MAXCLASS);
+	assert(sc_data->small_maxclass == SC_SMALL_MAXCLASS);
+	assert(sc_data->large_minclass == SC_LARGE_MINCLASS);
+	assert(sc_data->lg_large_minclass == SC_LG_LARGE_MINCLASS);
+	assert(sc_data->large_maxclass == SC_LARGE_MAXCLASS);
+
+	/*
+	 * In the allocation fastpath, we want to assume that we can
+	 * unconditionally subtract the requested allocation size from
+	 * a ssize_t, and detect passing through 0 correctly.  This
+	 * results in optimal generated code.  For this to work, the
+	 * maximum allocation size must be less than SSIZE_MAX.
+	 */
+	assert(SC_LARGE_MAXCLASS < SSIZE_MAX);
+}
+
+void
+sc_data_init(sc_data_t *sc_data) {
+	assert(!sc_data->initialized);
+
+	int lg_max_lookup = 12;
+
+	size_classes(sc_data, LG_SIZEOF_PTR, LG_QUANTUM, SC_LG_TINY_MIN,
+	    lg_max_lookup, LG_PAGE, 2);
+
+	sc_data->initialized = true;
+}
+
+static void
+sc_data_update_sc_slab_size(sc_t *sc, size_t reg_size, size_t pgs_guess) {
+	size_t min_pgs = reg_size / PAGE;
+	if (reg_size % PAGE != 0) {
+		min_pgs++;
+	}
+	/*
+	 * BITMAP_MAXBITS is actually determined by putting the smallest
+	 * possible size-class on one page, so this can never be 0.
+	 */
+	size_t max_pgs = BITMAP_MAXBITS * reg_size / PAGE;
+
+	assert(min_pgs <= max_pgs);
+	assert(min_pgs > 0);
+	assert(max_pgs >= 1);
+	if (pgs_guess < min_pgs) {
+		sc->pgs = (int)min_pgs;
+	} else if (pgs_guess > max_pgs) {
+		sc->pgs = (int)max_pgs;
+	} else {
+		sc->pgs = (int)pgs_guess;
+	}
+}
+
+void
+sc_data_update_slab_size(sc_data_t *data, size_t begin, size_t end, int pgs) {
+	assert(data->initialized);
+	for (int i = 0; i < data->nsizes; i++) {
+		sc_t *sc = &data->sc[i];
+		if (!sc->bin) {
+			break;
+		}
+		size_t reg_size = reg_size_compute(sc->lg_base, sc->lg_delta,
+		    sc->ndelta);
+		if (begin <= reg_size && reg_size <= end) {
+			sc_data_update_sc_slab_size(sc, reg_size, pgs);
+		}
+	}
+}
+
+void
+sc_boot(sc_data_t *data) {
+	sc_data_init(data);
+}
diff --git a/memkind-1.10.0/jemalloc/src/stats.c b/memkind-1.10.0/jemalloc/src/stats.c
new file mode 100644
index 0000000..118e05d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/stats.c
@@ -0,0 +1,1457 @@
+#define JEMALLOC_STATS_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/emitter.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_prof.h"
+
+const char *global_mutex_names[mutex_prof_num_global_mutexes] = {
+#define OP(mtx) #mtx,
+	MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+};
+
+const char *arena_mutex_names[mutex_prof_num_arena_mutexes] = {
+#define OP(mtx) #mtx,
+	MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+};
+
+#define CTL_GET(n, v, t) do {						\
+	size_t sz = sizeof(t);						\
+	xmallctl(n, (void *)v, &sz, NULL, 0);				\
+} while (0)
+
+#define CTL_M2_GET(n, i, v, t) do {					\
+	size_t mib[CTL_MAX_DEPTH];					\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = (i);							\
+	xmallctlbymib(mib, miblen, (void *)v, &sz, NULL, 0);		\
+} while (0)
+
+#define CTL_M2_M4_GET(n, i, j, v, t) do {				\
+	size_t mib[CTL_MAX_DEPTH];					\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = (i);							\
+	mib[4] = (j);							\
+	xmallctlbymib(mib, miblen, (void *)v, &sz, NULL, 0);		\
+} while (0)
+
+/******************************************************************************/
+/* Data. */
+
+bool opt_stats_print = false;
+char opt_stats_print_opts[stats_print_tot_num_options+1] = "";
+
+/******************************************************************************/
+
+static uint64_t
+rate_per_second(uint64_t value, uint64_t uptime_ns) {
+	uint64_t billion = 1000000000;
+	if (uptime_ns == 0 || value == 0) {
+		return 0;
+	}
+	if (uptime_ns < billion) {
+		return value;
+	} else {
+		uint64_t uptime_s = uptime_ns / billion;
+		return value / uptime_s;
+	}
+}
+
+/* Calculate x.yyy and output a string (takes a fixed sized char array). */
+static bool
+get_rate_str(uint64_t dividend, uint64_t divisor, char str[6]) {
+	if (divisor == 0 || dividend > divisor) {
+		/* The rate is not supposed to be greater than 1. */
+		return true;
+	}
+	if (dividend > 0) {
+		assert(UINT64_MAX / dividend >= 1000);
+	}
+
+	unsigned n = (unsigned)((dividend * 1000) / divisor);
+	if (n < 10) {
+		malloc_snprintf(str, 6, "0.00%u", n);
+	} else if (n < 100) {
+		malloc_snprintf(str, 6, "0.0%u", n);
+	} else if (n < 1000) {
+		malloc_snprintf(str, 6, "0.%u", n);
+	} else {
+		malloc_snprintf(str, 6, "1");
+	}
+
+	return false;
+}
+
+#define MUTEX_CTL_STR_MAX_LENGTH 128
+static void
+gen_mutex_ctl_str(char *str, size_t buf_len, const char *prefix,
+    const char *mutex, const char *counter) {
+	malloc_snprintf(str, buf_len, "stats.%s.%s.%s", prefix, mutex, counter);
+}
+
+static void
+mutex_stats_init_cols(emitter_row_t *row, const char *table_name,
+    emitter_col_t *name,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters]) {
+	mutex_prof_uint64_t_counter_ind_t k_uint64_t = 0;
+	mutex_prof_uint32_t_counter_ind_t k_uint32_t = 0;
+
+	emitter_col_t *col;
+
+	if (name != NULL) {
+		emitter_col_init(name, row);
+		name->justify = emitter_justify_left;
+		name->width = 21;
+		name->type = emitter_type_title;
+		name->str_val = table_name;
+	}
+
+#define WIDTH_uint32_t 12
+#define WIDTH_uint64_t 16
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	col = &col_##counter_type[k_##counter_type];			\
+	++k_##counter_type;						\
+	emitter_col_init(col, row);					\
+	col->justify = emitter_justify_right;				\
+	col->width = derived ? 8 : WIDTH_##counter_type;		\
+	col->type = emitter_type_title;					\
+	col->str_val = human;
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef WIDTH_uint32_t
+#undef WIDTH_uint64_t
+	col_uint64_t[mutex_counter_total_wait_time_ps].width = 10;
+}
+
+static void
+mutex_stats_read_global(const char *name, emitter_col_t *col_name,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters],
+    uint64_t uptime) {
+	char cmd[MUTEX_CTL_STR_MAX_LENGTH];
+
+	col_name->str_val = name;
+
+	emitter_col_t *dst;
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	dst = &col_##counter_type[mutex_counter_##counter];		\
+	dst->type = EMITTER_TYPE_##counter_type;			\
+	if (!derived) {							\
+		gen_mutex_ctl_str(cmd, MUTEX_CTL_STR_MAX_LENGTH,	\
+		    "mutexes", name, #counter);				\
+		CTL_GET(cmd, (counter_type *)&dst->bool_val, counter_type);	\
+	} else { \
+	    emitter_col_t *base = &col_##counter_type[mutex_counter_##base_counter];	\
+	    dst->counter_type##_val = rate_per_second(base->counter_type##_val, uptime); \
+	}
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+static void
+mutex_stats_read_arena(unsigned arena_ind, mutex_prof_arena_ind_t mutex_ind,
+    const char *name, emitter_col_t *col_name,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters],
+    uint64_t uptime) {
+	char cmd[MUTEX_CTL_STR_MAX_LENGTH];
+
+	col_name->str_val = name;
+
+	emitter_col_t *dst;
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	dst = &col_##counter_type[mutex_counter_##counter];		\
+	dst->type = EMITTER_TYPE_##counter_type;			\
+	if (!derived) {                                   \
+		gen_mutex_ctl_str(cmd, MUTEX_CTL_STR_MAX_LENGTH,        \
+		    "arenas.0.mutexes", arena_mutex_names[mutex_ind], #counter);\
+		CTL_M2_GET(cmd, arena_ind, (counter_type *)&dst->bool_val, counter_type); \
+	} else {                      \
+		emitter_col_t *base = &col_##counter_type[mutex_counter_##base_counter];	\
+		dst->counter_type##_val = rate_per_second(base->counter_type##_val, uptime); \
+	}
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+static void
+mutex_stats_read_arena_bin(unsigned arena_ind, unsigned bin_ind,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters],
+    uint64_t uptime) {
+	char cmd[MUTEX_CTL_STR_MAX_LENGTH];
+	emitter_col_t *dst;
+
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, counter_type, human, derived, base_counter)	\
+	dst = &col_##counter_type[mutex_counter_##counter];		\
+	dst->type = EMITTER_TYPE_##counter_type;			\
+	if (!derived) {                                   \
+		gen_mutex_ctl_str(cmd, MUTEX_CTL_STR_MAX_LENGTH,        \
+		    "arenas.0.bins.0","mutex", #counter);            \
+		CTL_M2_M4_GET(cmd, arena_ind, bin_ind,                \
+		    (counter_type *)&dst->bool_val, counter_type);  \
+	} else {                      \
+		emitter_col_t *base = &col_##counter_type[mutex_counter_##base_counter]; \
+		dst->counter_type##_val = rate_per_second(base->counter_type##_val, uptime); \
+	}
+	MUTEX_PROF_COUNTERS
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+/* "row" can be NULL to avoid emitting in table mode. */
+static void
+mutex_stats_emit(emitter_t *emitter, emitter_row_t *row,
+    emitter_col_t col_uint64_t[mutex_prof_num_uint64_t_counters],
+    emitter_col_t col_uint32_t[mutex_prof_num_uint32_t_counters]) {
+	if (row != NULL) {
+		emitter_table_row(emitter, row);
+	}
+
+	mutex_prof_uint64_t_counter_ind_t k_uint64_t = 0;
+	mutex_prof_uint32_t_counter_ind_t k_uint32_t = 0;
+
+	emitter_col_t *col;
+
+#define EMITTER_TYPE_uint32_t emitter_type_uint32
+#define EMITTER_TYPE_uint64_t emitter_type_uint64
+#define OP(counter, type, human, derived, base_counter)		\
+	if (!derived) {                    \
+		col = &col_##type[k_##type];                        \
+		++k_##type;                            \
+		emitter_json_kv(emitter, #counter, EMITTER_TYPE_##type,        \
+		    (const void *)&col->bool_val); \
+	}
+	MUTEX_PROF_COUNTERS;
+#undef OP
+#undef EMITTER_TYPE_uint32_t
+#undef EMITTER_TYPE_uint64_t
+}
+
+#define COL(row_name, column_name, left_or_right, col_width, etype)      \
+	emitter_col_t col_##column_name;                                     \
+	emitter_col_init(&col_##column_name, &row_name);                     \
+	col_##column_name.justify = emitter_justify_##left_or_right;         \
+	col_##column_name.width = col_width;                                 \
+	col_##column_name.type = emitter_type_##etype;
+
+#define COL_HDR(row_name, column_name, human, left_or_right, col_width, etype)  \
+	COL(row_name, column_name, left_or_right, col_width, etype)	         \
+	emitter_col_t header_##column_name;                                  \
+	emitter_col_init(&header_##column_name, &header_##row_name);         \
+	header_##column_name.justify = emitter_justify_##left_or_right;      \
+	header_##column_name.width = col_width;                              \
+	header_##column_name.type = emitter_type_title;                      \
+	header_##column_name.str_val = human ? human : #column_name;
+
+
+static void
+stats_arena_bins_print(emitter_t *emitter, bool mutex, unsigned i, uint64_t uptime) {
+	size_t page;
+	bool in_gap, in_gap_prev;
+	unsigned nbins, j;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+
+	emitter_row_t header_row;
+	emitter_row_init(&header_row);
+
+	emitter_row_t row;
+	emitter_row_init(&row);
+
+	COL_HDR(row, size, NULL, right, 20, size)
+	COL_HDR(row, ind, NULL, right, 4, unsigned)
+	COL_HDR(row, allocated, NULL, right, 13, uint64)
+	COL_HDR(row, nmalloc, NULL, right, 13, uint64)
+	COL_HDR(row, nmalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, ndalloc, NULL, right, 13, uint64)
+	COL_HDR(row, ndalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nrequests, NULL, right, 13, uint64)
+	COL_HDR(row, nrequests_ps, "(#/sec)", right, 10, uint64)
+	COL_HDR(row, nshards, NULL, right, 9, unsigned)
+	COL_HDR(row, curregs, NULL, right, 13, size)
+	COL_HDR(row, curslabs, NULL, right, 13, size)
+	COL_HDR(row, nonfull_slabs, NULL, right, 15, size)
+	COL_HDR(row, regs, NULL, right, 5, unsigned)
+	COL_HDR(row, pgs, NULL, right, 4, size)
+	/* To buffer a right- and left-justified column. */
+	COL_HDR(row, justify_spacer, NULL, right, 1, title)
+	COL_HDR(row, util, NULL, right, 6, title)
+	COL_HDR(row, nfills, NULL, right, 13, uint64)
+	COL_HDR(row, nfills_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nflushes, NULL, right, 13, uint64)
+	COL_HDR(row, nflushes_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nslabs, NULL, right, 13, uint64)
+	COL_HDR(row, nreslabs, NULL, right, 13, uint64)
+	COL_HDR(row, nreslabs_ps, "(#/sec)", right, 8, uint64)
+
+	/* Don't want to actually print the name. */
+	header_justify_spacer.str_val = " ";
+	col_justify_spacer.str_val = " ";
+
+	emitter_col_t col_mutex64[mutex_prof_num_uint64_t_counters];
+	emitter_col_t col_mutex32[mutex_prof_num_uint32_t_counters];
+
+	emitter_col_t header_mutex64[mutex_prof_num_uint64_t_counters];
+	emitter_col_t header_mutex32[mutex_prof_num_uint32_t_counters];
+
+	if (mutex) {
+		mutex_stats_init_cols(&row, NULL, NULL, col_mutex64,
+		    col_mutex32);
+		mutex_stats_init_cols(&header_row, NULL, NULL, header_mutex64,
+		    header_mutex32);
+	}
+
+	/*
+	 * We print a "bins:" header as part of the table row; we need to adjust
+	 * the header size column to compensate.
+	 */
+	header_size.width -=5;
+	emitter_table_printf(emitter, "bins:");
+	emitter_table_row(emitter, &header_row);
+	emitter_json_array_kv_begin(emitter, "bins");
+
+	for (j = 0, in_gap = false; j < nbins; j++) {
+		uint64_t nslabs;
+		size_t reg_size, slab_size, curregs;
+		size_t curslabs;
+		size_t nonfull_slabs;
+		uint32_t nregs, nshards;
+		uint64_t nmalloc, ndalloc, nrequests, nfills, nflushes;
+		uint64_t nreslabs;
+
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nslabs", i, j, &nslabs,
+		    uint64_t);
+		in_gap_prev = in_gap;
+		in_gap = (nslabs == 0);
+
+		if (in_gap_prev && !in_gap) {
+			emitter_table_printf(emitter,
+			    "                     ---\n");
+		}
+
+		CTL_M2_GET("arenas.bin.0.size", j, &reg_size, size_t);
+		CTL_M2_GET("arenas.bin.0.nregs", j, &nregs, uint32_t);
+		CTL_M2_GET("arenas.bin.0.slab_size", j, &slab_size, size_t);
+		CTL_M2_GET("arenas.bin.0.nshards", j, &nshards, uint32_t);
+
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nmalloc", i, j, &nmalloc,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.ndalloc", i, j, &ndalloc,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.curregs", i, j, &curregs,
+		    size_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nrequests", i, j,
+		    &nrequests, uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nfills", i, j, &nfills,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nflushes", i, j, &nflushes,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nreslabs", i, j, &nreslabs,
+		    uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.curslabs", i, j, &curslabs,
+		    size_t);
+		CTL_M2_M4_GET("stats.arenas.0.bins.0.nonfull_slabs", i, j, &nonfull_slabs,
+		    size_t);
+
+		if (mutex) {
+			mutex_stats_read_arena_bin(i, j, col_mutex64,
+			    col_mutex32, uptime);
+		}
+
+		emitter_json_object_begin(emitter);
+		emitter_json_kv(emitter, "nmalloc", emitter_type_uint64,
+		    &nmalloc);
+		emitter_json_kv(emitter, "ndalloc", emitter_type_uint64,
+		    &ndalloc);
+		emitter_json_kv(emitter, "curregs", emitter_type_size,
+		    &curregs);
+		emitter_json_kv(emitter, "nrequests", emitter_type_uint64,
+		    &nrequests);
+		emitter_json_kv(emitter, "nfills", emitter_type_uint64,
+		    &nfills);
+		emitter_json_kv(emitter, "nflushes", emitter_type_uint64,
+		    &nflushes);
+		emitter_json_kv(emitter, "nreslabs", emitter_type_uint64,
+		    &nreslabs);
+		emitter_json_kv(emitter, "curslabs", emitter_type_size,
+		    &curslabs);
+		emitter_json_kv(emitter, "nonfull_slabs", emitter_type_size,
+		    &nonfull_slabs);
+		if (mutex) {
+			emitter_json_object_kv_begin(emitter, "mutex");
+			mutex_stats_emit(emitter, NULL, col_mutex64,
+			    col_mutex32);
+			emitter_json_object_end(emitter);
+		}
+		emitter_json_object_end(emitter);
+
+		size_t availregs = nregs * curslabs;
+		char util[6];
+		if (get_rate_str((uint64_t)curregs, (uint64_t)availregs, util))
+		{
+			if (availregs == 0) {
+				malloc_snprintf(util, sizeof(util), "1");
+			} else if (curregs > availregs) {
+				/*
+				 * Race detected: the counters were read in
+				 * separate mallctl calls and concurrent
+				 * operations happened in between.  In this case
+				 * no meaningful utilization can be computed.
+				 */
+				malloc_snprintf(util, sizeof(util), " race");
+			} else {
+				not_reached();
+			}
+		}
+
+		col_size.size_val = reg_size;
+		col_ind.unsigned_val = j;
+		col_allocated.size_val = curregs * reg_size;
+		col_nmalloc.uint64_val = nmalloc;
+		col_nmalloc_ps.uint64_val = rate_per_second(nmalloc, uptime);
+		col_ndalloc.uint64_val = ndalloc;
+		col_ndalloc_ps.uint64_val = rate_per_second(ndalloc, uptime);
+		col_nrequests.uint64_val = nrequests;
+		col_nrequests_ps.uint64_val = rate_per_second(nrequests, uptime);
+		col_nshards.unsigned_val = nshards;
+		col_curregs.size_val = curregs;
+		col_curslabs.size_val = curslabs;
+		col_nonfull_slabs.size_val = nonfull_slabs;
+		col_regs.unsigned_val = nregs;
+		col_pgs.size_val = slab_size / page;
+		col_util.str_val = util;
+		col_nfills.uint64_val = nfills;
+		col_nfills_ps.uint64_val = rate_per_second(nfills, uptime);
+		col_nflushes.uint64_val = nflushes;
+		col_nflushes_ps.uint64_val = rate_per_second(nflushes, uptime);
+		col_nslabs.uint64_val = nslabs;
+		col_nreslabs.uint64_val = nreslabs;
+		col_nreslabs_ps.uint64_val = rate_per_second(nreslabs, uptime);
+
+		/*
+		 * Note that mutex columns were initialized above, if mutex ==
+		 * true.
+		 */
+
+		emitter_table_row(emitter, &row);
+	}
+	emitter_json_array_end(emitter); /* Close "bins". */
+
+	if (in_gap) {
+		emitter_table_printf(emitter, "                     ---\n");
+	}
+}
+
+static void
+stats_arena_lextents_print(emitter_t *emitter, unsigned i, uint64_t uptime) {
+	unsigned nbins, nlextents, j;
+	bool in_gap, in_gap_prev;
+
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+	CTL_GET("arenas.nlextents", &nlextents, unsigned);
+
+	emitter_row_t header_row;
+	emitter_row_init(&header_row);
+	emitter_row_t row;
+	emitter_row_init(&row);
+
+	COL_HDR(row, size, NULL, right, 20, size)
+	COL_HDR(row, ind, NULL, right, 4, unsigned)
+	COL_HDR(row, allocated, NULL, right, 13, size)
+	COL_HDR(row, nmalloc, NULL, right, 13, uint64)
+	COL_HDR(row, nmalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, ndalloc, NULL, right, 13, uint64)
+	COL_HDR(row, ndalloc_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, nrequests, NULL, right, 13, uint64)
+	COL_HDR(row, nrequests_ps, "(#/sec)", right, 8, uint64)
+	COL_HDR(row, curlextents, NULL, right, 13, size)
+
+	/* As with bins, we label the large extents table. */
+	header_size.width -= 6;
+	emitter_table_printf(emitter, "large:");
+	emitter_table_row(emitter, &header_row);
+	emitter_json_array_kv_begin(emitter, "lextents");
+
+	for (j = 0, in_gap = false; j < nlextents; j++) {
+		uint64_t nmalloc, ndalloc, nrequests;
+		size_t lextent_size, curlextents;
+
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.nmalloc", i, j,
+		    &nmalloc, uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.ndalloc", i, j,
+		    &ndalloc, uint64_t);
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.nrequests", i, j,
+		    &nrequests, uint64_t);
+		in_gap_prev = in_gap;
+		in_gap = (nrequests == 0);
+
+		if (in_gap_prev && !in_gap) {
+			emitter_table_printf(emitter,
+			    "                     ---\n");
+		}
+
+		CTL_M2_GET("arenas.lextent.0.size", j, &lextent_size, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.lextents.0.curlextents", i, j,
+		    &curlextents, size_t);
+
+		emitter_json_object_begin(emitter);
+		emitter_json_kv(emitter, "curlextents", emitter_type_size,
+		    &curlextents);
+		emitter_json_object_end(emitter);
+
+		col_size.size_val = lextent_size;
+		col_ind.unsigned_val = nbins + j;
+		col_allocated.size_val = curlextents * lextent_size;
+		col_nmalloc.uint64_val = nmalloc;
+		col_nmalloc_ps.uint64_val = rate_per_second(nmalloc, uptime);
+		col_ndalloc.uint64_val = ndalloc;
+		col_ndalloc_ps.uint64_val = rate_per_second(ndalloc, uptime);
+		col_nrequests.uint64_val = nrequests;
+		col_nrequests_ps.uint64_val = rate_per_second(nrequests, uptime);
+		col_curlextents.size_val = curlextents;
+
+		if (!in_gap) {
+			emitter_table_row(emitter, &row);
+		}
+	}
+	emitter_json_array_end(emitter); /* Close "lextents". */
+	if (in_gap) {
+		emitter_table_printf(emitter, "                     ---\n");
+	}
+}
+
+static void
+stats_arena_extents_print(emitter_t *emitter, unsigned i) {
+	unsigned j;
+	bool in_gap, in_gap_prev;
+	emitter_row_t header_row;
+	emitter_row_init(&header_row);
+	emitter_row_t row;
+	emitter_row_init(&row);
+
+	COL_HDR(row, size, NULL, right, 20, size)
+	COL_HDR(row, ind, NULL, right, 4, unsigned)
+	COL_HDR(row, ndirty, NULL, right, 13, size)
+	COL_HDR(row, dirty, NULL, right, 13, size)
+	COL_HDR(row, nmuzzy, NULL, right, 13, size)
+	COL_HDR(row, muzzy, NULL, right, 13, size)
+	COL_HDR(row, nretained, NULL, right, 13, size)
+	COL_HDR(row, retained, NULL, right, 13, size)
+	COL_HDR(row, ntotal, NULL, right, 13, size)
+	COL_HDR(row, total, NULL, right, 13, size)
+
+	/* Label this section. */
+	header_size.width -= 8;
+	emitter_table_printf(emitter, "extents:");
+	emitter_table_row(emitter, &header_row);
+	emitter_json_array_kv_begin(emitter, "extents");
+
+	in_gap = false;
+	for (j = 0; j < SC_NPSIZES; j++) {
+		size_t ndirty, nmuzzy, nretained, total, dirty_bytes,
+		    muzzy_bytes, retained_bytes, total_bytes;
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.ndirty", i, j,
+		    &ndirty, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.nmuzzy", i, j,
+		    &nmuzzy, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.nretained", i, j,
+		    &nretained, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.dirty_bytes", i, j,
+		    &dirty_bytes, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.muzzy_bytes", i, j,
+		    &muzzy_bytes, size_t);
+		CTL_M2_M4_GET("stats.arenas.0.extents.0.retained_bytes", i, j,
+		    &retained_bytes, size_t);
+		total = ndirty + nmuzzy + nretained;
+		total_bytes = dirty_bytes + muzzy_bytes + retained_bytes;
+
+		in_gap_prev = in_gap;
+		in_gap = (total == 0);
+
+		if (in_gap_prev && !in_gap) {
+			emitter_table_printf(emitter,
+			    "                     ---\n");
+		}
+
+		emitter_json_object_begin(emitter);
+		emitter_json_kv(emitter, "ndirty", emitter_type_size, &ndirty);
+		emitter_json_kv(emitter, "nmuzzy", emitter_type_size, &nmuzzy);
+		emitter_json_kv(emitter, "nretained", emitter_type_size,
+		    &nretained);
+
+		emitter_json_kv(emitter, "dirty_bytes", emitter_type_size,
+		    &dirty_bytes);
+		emitter_json_kv(emitter, "muzzy_bytes", emitter_type_size,
+		    &muzzy_bytes);
+		emitter_json_kv(emitter, "retained_bytes", emitter_type_size,
+		    &retained_bytes);
+		emitter_json_object_end(emitter);
+
+		col_size.size_val = sz_pind2sz(j);
+		col_ind.size_val = j;
+		col_ndirty.size_val = ndirty;
+		col_dirty.size_val = dirty_bytes;
+		col_nmuzzy.size_val = nmuzzy;
+		col_muzzy.size_val = muzzy_bytes;
+		col_nretained.size_val = nretained;
+		col_retained.size_val = retained_bytes;
+		col_ntotal.size_val = total;
+		col_total.size_val = total_bytes;
+
+		if (!in_gap) {
+			emitter_table_row(emitter, &row);
+		}
+	}
+	emitter_json_array_end(emitter); /* Close "extents". */
+	if (in_gap) {
+		emitter_table_printf(emitter, "                     ---\n");
+	}
+}
+
+static void
+stats_arena_mutexes_print(emitter_t *emitter, unsigned arena_ind, uint64_t uptime) {
+	emitter_row_t row;
+	emitter_col_t col_name;
+	emitter_col_t col64[mutex_prof_num_uint64_t_counters];
+	emitter_col_t col32[mutex_prof_num_uint32_t_counters];
+
+	emitter_row_init(&row);
+	mutex_stats_init_cols(&row, "", &col_name, col64, col32);
+
+	emitter_json_object_kv_begin(emitter, "mutexes");
+	emitter_table_row(emitter, &row);
+
+	for (mutex_prof_arena_ind_t i = 0; i < mutex_prof_num_arena_mutexes;
+	    i++) {
+		const char *name = arena_mutex_names[i];
+		emitter_json_object_kv_begin(emitter, name);
+		mutex_stats_read_arena(arena_ind, i, name, &col_name, col64,
+		    col32, uptime);
+		mutex_stats_emit(emitter, &row, col64, col32);
+		emitter_json_object_end(emitter); /* Close the mutex dict. */
+	}
+	emitter_json_object_end(emitter); /* End "mutexes". */
+}
+
+static void
+stats_arena_print(emitter_t *emitter, unsigned i, bool bins, bool large,
+    bool mutex, bool extents) {
+	unsigned nthreads;
+	const char *dss;
+	ssize_t dirty_decay_ms, muzzy_decay_ms;
+	size_t page, pactive, pdirty, pmuzzy, mapped, retained;
+	size_t base, internal, resident, metadata_thp, extent_avail;
+	uint64_t dirty_npurge, dirty_nmadvise, dirty_purged;
+	uint64_t muzzy_npurge, muzzy_nmadvise, muzzy_purged;
+	size_t small_allocated;
+	uint64_t small_nmalloc, small_ndalloc, small_nrequests, small_nfills,
+	    small_nflushes;
+	size_t large_allocated;
+	uint64_t large_nmalloc, large_ndalloc, large_nrequests, large_nfills,
+	    large_nflushes;
+	size_t tcache_bytes, abandoned_vm;
+	uint64_t uptime;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_M2_GET("stats.arenas.0.nthreads", i, &nthreads, unsigned);
+	emitter_kv(emitter, "nthreads", "assigned threads",
+	    emitter_type_unsigned, &nthreads);
+
+	CTL_M2_GET("stats.arenas.0.uptime", i, &uptime, uint64_t);
+	emitter_kv(emitter, "uptime_ns", "uptime", emitter_type_uint64,
+	    &uptime);
+
+	CTL_M2_GET("stats.arenas.0.dss", i, &dss, const char *);
+	emitter_kv(emitter, "dss", "dss allocation precedence",
+	    emitter_type_string, &dss);
+
+	CTL_M2_GET("stats.arenas.0.dirty_decay_ms", i, &dirty_decay_ms,
+	    ssize_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_decay_ms", i, &muzzy_decay_ms,
+	    ssize_t);
+	CTL_M2_GET("stats.arenas.0.pactive", i, &pactive, size_t);
+	CTL_M2_GET("stats.arenas.0.pdirty", i, &pdirty, size_t);
+	CTL_M2_GET("stats.arenas.0.pmuzzy", i, &pmuzzy, size_t);
+	CTL_M2_GET("stats.arenas.0.dirty_npurge", i, &dirty_npurge, uint64_t);
+	CTL_M2_GET("stats.arenas.0.dirty_nmadvise", i, &dirty_nmadvise,
+	    uint64_t);
+	CTL_M2_GET("stats.arenas.0.dirty_purged", i, &dirty_purged, uint64_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_npurge", i, &muzzy_npurge, uint64_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_nmadvise", i, &muzzy_nmadvise,
+	    uint64_t);
+	CTL_M2_GET("stats.arenas.0.muzzy_purged", i, &muzzy_purged, uint64_t);
+
+	emitter_row_t decay_row;
+	emitter_row_init(&decay_row);
+
+	/* JSON-style emission. */
+	emitter_json_kv(emitter, "dirty_decay_ms", emitter_type_ssize,
+	    &dirty_decay_ms);
+	emitter_json_kv(emitter, "muzzy_decay_ms", emitter_type_ssize,
+	    &muzzy_decay_ms);
+
+	emitter_json_kv(emitter, "pactive", emitter_type_size, &pactive);
+	emitter_json_kv(emitter, "pdirty", emitter_type_size, &pdirty);
+	emitter_json_kv(emitter, "pmuzzy", emitter_type_size, &pmuzzy);
+
+	emitter_json_kv(emitter, "dirty_npurge", emitter_type_uint64,
+	    &dirty_npurge);
+	emitter_json_kv(emitter, "dirty_nmadvise", emitter_type_uint64,
+	    &dirty_nmadvise);
+	emitter_json_kv(emitter, "dirty_purged", emitter_type_uint64,
+	    &dirty_purged);
+
+	emitter_json_kv(emitter, "muzzy_npurge", emitter_type_uint64,
+	    &muzzy_npurge);
+	emitter_json_kv(emitter, "muzzy_nmadvise", emitter_type_uint64,
+	    &muzzy_nmadvise);
+	emitter_json_kv(emitter, "muzzy_purged", emitter_type_uint64,
+	    &muzzy_purged);
+
+	/* Table-style emission. */
+	COL(decay_row, decay_type, right, 9, title);
+	col_decay_type.str_val = "decaying:";
+
+	COL(decay_row, decay_time, right, 6, title);
+	col_decay_time.str_val = "time";
+
+	COL(decay_row, decay_npages, right, 13, title);
+	col_decay_npages.str_val = "npages";
+
+	COL(decay_row, decay_sweeps, right, 13, title);
+	col_decay_sweeps.str_val = "sweeps";
+
+	COL(decay_row, decay_madvises, right, 13, title);
+	col_decay_madvises.str_val = "madvises";
+
+	COL(decay_row, decay_purged, right, 13, title);
+	col_decay_purged.str_val = "purged";
+
+	/* Title row. */
+	emitter_table_row(emitter, &decay_row);
+
+	/* Dirty row. */
+	col_decay_type.str_val = "dirty:";
+
+	if (dirty_decay_ms >= 0) {
+		col_decay_time.type = emitter_type_ssize;
+		col_decay_time.ssize_val = dirty_decay_ms;
+	} else {
+		col_decay_time.type = emitter_type_title;
+		col_decay_time.str_val = "N/A";
+	}
+
+	col_decay_npages.type = emitter_type_size;
+	col_decay_npages.size_val = pdirty;
+
+	col_decay_sweeps.type = emitter_type_uint64;
+	col_decay_sweeps.uint64_val = dirty_npurge;
+
+	col_decay_madvises.type = emitter_type_uint64;
+	col_decay_madvises.uint64_val = dirty_nmadvise;
+
+	col_decay_purged.type = emitter_type_uint64;
+	col_decay_purged.uint64_val = dirty_purged;
+
+	emitter_table_row(emitter, &decay_row);
+
+	/* Muzzy row. */
+	col_decay_type.str_val = "muzzy:";
+
+	if (muzzy_decay_ms >= 0) {
+		col_decay_time.type = emitter_type_ssize;
+		col_decay_time.ssize_val = muzzy_decay_ms;
+	} else {
+		col_decay_time.type = emitter_type_title;
+		col_decay_time.str_val = "N/A";
+	}
+
+	col_decay_npages.type = emitter_type_size;
+	col_decay_npages.size_val = pmuzzy;
+
+	col_decay_sweeps.type = emitter_type_uint64;
+	col_decay_sweeps.uint64_val = muzzy_npurge;
+
+	col_decay_madvises.type = emitter_type_uint64;
+	col_decay_madvises.uint64_val = muzzy_nmadvise;
+
+	col_decay_purged.type = emitter_type_uint64;
+	col_decay_purged.uint64_val = muzzy_purged;
+
+	emitter_table_row(emitter, &decay_row);
+
+	/* Small / large / total allocation counts. */
+	emitter_row_t alloc_count_row;
+	emitter_row_init(&alloc_count_row);
+
+	COL(alloc_count_row, count_title, left, 21, title);
+	col_count_title.str_val = "";
+
+	COL(alloc_count_row, count_allocated, right, 16, title);
+	col_count_allocated.str_val = "allocated";
+
+	COL(alloc_count_row, count_nmalloc, right, 16, title);
+	col_count_nmalloc.str_val = "nmalloc";
+	COL(alloc_count_row, count_nmalloc_ps, right, 8, title);
+	col_count_nmalloc_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_ndalloc, right, 16, title);
+	col_count_ndalloc.str_val = "ndalloc";
+	COL(alloc_count_row, count_ndalloc_ps, right, 8, title);
+	col_count_ndalloc_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_nrequests, right, 16, title);
+	col_count_nrequests.str_val = "nrequests";
+	COL(alloc_count_row, count_nrequests_ps, right, 10, title);
+	col_count_nrequests_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_nfills, right, 16, title);
+	col_count_nfills.str_val = "nfill";
+	COL(alloc_count_row, count_nfills_ps, right, 10, title);
+	col_count_nfills_ps.str_val = "(#/sec)";
+
+	COL(alloc_count_row, count_nflushes, right, 16, title);
+	col_count_nflushes.str_val = "nflush";
+	COL(alloc_count_row, count_nflushes_ps, right, 10, title);
+	col_count_nflushes_ps.str_val = "(#/sec)";
+
+	emitter_table_row(emitter, &alloc_count_row);
+
+	col_count_nmalloc_ps.type = emitter_type_uint64;
+	col_count_ndalloc_ps.type = emitter_type_uint64;
+	col_count_nrequests_ps.type = emitter_type_uint64;
+	col_count_nfills_ps.type = emitter_type_uint64;
+	col_count_nflushes_ps.type = emitter_type_uint64;
+
+#define GET_AND_EMIT_ALLOC_STAT(small_or_large, name, valtype)		\
+	CTL_M2_GET("stats.arenas.0." #small_or_large "." #name, i,	\
+	    &small_or_large##_##name, valtype##_t);			\
+	emitter_json_kv(emitter, #name, emitter_type_##valtype,		\
+	    &small_or_large##_##name);					\
+	col_count_##name.type = emitter_type_##valtype;		\
+	col_count_##name.valtype##_val = small_or_large##_##name;
+
+	emitter_json_object_kv_begin(emitter, "small");
+	col_count_title.str_val = "small:";
+
+	GET_AND_EMIT_ALLOC_STAT(small, allocated, size)
+	GET_AND_EMIT_ALLOC_STAT(small, nmalloc, uint64)
+	col_count_nmalloc_ps.uint64_val =
+	    rate_per_second(col_count_nmalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, ndalloc, uint64)
+	col_count_ndalloc_ps.uint64_val =
+	    rate_per_second(col_count_ndalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, nrequests, uint64)
+	col_count_nrequests_ps.uint64_val =
+	    rate_per_second(col_count_nrequests.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, nfills, uint64)
+	col_count_nfills_ps.uint64_val =
+	    rate_per_second(col_count_nfills.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(small, nflushes, uint64)
+	col_count_nflushes_ps.uint64_val =
+	    rate_per_second(col_count_nflushes.uint64_val, uptime);
+
+	emitter_table_row(emitter, &alloc_count_row);
+	emitter_json_object_end(emitter); /* Close "small". */
+
+	emitter_json_object_kv_begin(emitter, "large");
+	col_count_title.str_val = "large:";
+
+	GET_AND_EMIT_ALLOC_STAT(large, allocated, size)
+	GET_AND_EMIT_ALLOC_STAT(large, nmalloc, uint64)
+	col_count_nmalloc_ps.uint64_val =
+	    rate_per_second(col_count_nmalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, ndalloc, uint64)
+	col_count_ndalloc_ps.uint64_val =
+	    rate_per_second(col_count_ndalloc.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, nrequests, uint64)
+	col_count_nrequests_ps.uint64_val =
+	    rate_per_second(col_count_nrequests.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, nfills, uint64)
+	col_count_nfills_ps.uint64_val =
+	    rate_per_second(col_count_nfills.uint64_val, uptime);
+	GET_AND_EMIT_ALLOC_STAT(large, nflushes, uint64)
+	col_count_nflushes_ps.uint64_val =
+	    rate_per_second(col_count_nflushes.uint64_val, uptime);
+
+	emitter_table_row(emitter, &alloc_count_row);
+	emitter_json_object_end(emitter); /* Close "large". */
+
+#undef GET_AND_EMIT_ALLOC_STAT
+
+	/* Aggregated small + large stats are emitter only in table mode. */
+	col_count_title.str_val = "total:";
+	col_count_allocated.size_val = small_allocated + large_allocated;
+	col_count_nmalloc.uint64_val = small_nmalloc + large_nmalloc;
+	col_count_ndalloc.uint64_val = small_ndalloc + large_ndalloc;
+	col_count_nrequests.uint64_val = small_nrequests + large_nrequests;
+	col_count_nfills.uint64_val = small_nfills + large_nfills;
+	col_count_nflushes.uint64_val = small_nflushes + large_nflushes;
+	col_count_nmalloc_ps.uint64_val =
+	    rate_per_second(col_count_nmalloc.uint64_val, uptime);
+	col_count_ndalloc_ps.uint64_val =
+	    rate_per_second(col_count_ndalloc.uint64_val, uptime);
+	col_count_nrequests_ps.uint64_val =
+	    rate_per_second(col_count_nrequests.uint64_val, uptime);
+	col_count_nfills_ps.uint64_val =
+	    rate_per_second(col_count_nfills.uint64_val, uptime);
+	col_count_nflushes_ps.uint64_val =
+	    rate_per_second(col_count_nflushes.uint64_val, uptime);
+	emitter_table_row(emitter, &alloc_count_row);
+
+	emitter_row_t mem_count_row;
+	emitter_row_init(&mem_count_row);
+
+	emitter_col_t mem_count_title;
+	emitter_col_init(&mem_count_title, &mem_count_row);
+	mem_count_title.justify = emitter_justify_left;
+	mem_count_title.width = 21;
+	mem_count_title.type = emitter_type_title;
+	mem_count_title.str_val = "";
+
+	emitter_col_t mem_count_val;
+	emitter_col_init(&mem_count_val, &mem_count_row);
+	mem_count_val.justify = emitter_justify_right;
+	mem_count_val.width = 16;
+	mem_count_val.type = emitter_type_title;
+	mem_count_val.str_val = "";
+
+	emitter_table_row(emitter, &mem_count_row);
+	mem_count_val.type = emitter_type_size;
+
+	/* Active count in bytes is emitted only in table mode. */
+	mem_count_title.str_val = "active:";
+	mem_count_val.size_val = pactive * page;
+	emitter_table_row(emitter, &mem_count_row);
+
+#define GET_AND_EMIT_MEM_STAT(stat)					\
+	CTL_M2_GET("stats.arenas.0."#stat, i, &stat, size_t);		\
+	emitter_json_kv(emitter, #stat, emitter_type_size, &stat);	\
+	mem_count_title.str_val = #stat":";				\
+	mem_count_val.size_val = stat;					\
+	emitter_table_row(emitter, &mem_count_row);
+
+	GET_AND_EMIT_MEM_STAT(mapped)
+	GET_AND_EMIT_MEM_STAT(retained)
+	GET_AND_EMIT_MEM_STAT(base)
+	GET_AND_EMIT_MEM_STAT(internal)
+	GET_AND_EMIT_MEM_STAT(metadata_thp)
+	GET_AND_EMIT_MEM_STAT(tcache_bytes)
+	GET_AND_EMIT_MEM_STAT(resident)
+	GET_AND_EMIT_MEM_STAT(abandoned_vm)
+	GET_AND_EMIT_MEM_STAT(extent_avail)
+#undef GET_AND_EMIT_MEM_STAT
+
+	if (mutex) {
+		stats_arena_mutexes_print(emitter, i, uptime);
+	}
+	if (bins) {
+		stats_arena_bins_print(emitter, mutex, i, uptime);
+	}
+	if (large) {
+		stats_arena_lextents_print(emitter, i, uptime);
+	}
+	if (extents) {
+		stats_arena_extents_print(emitter, i);
+	}
+}
+
+static void
+stats_general_print(emitter_t *emitter) {
+	const char *cpv;
+	bool bv, bv2;
+	unsigned uv;
+	uint32_t u32v;
+	uint64_t u64v;
+	ssize_t ssv, ssv2;
+	size_t sv, bsz, usz, ssz, sssz, cpsz;
+
+	bsz = sizeof(bool);
+	usz = sizeof(unsigned);
+	ssz = sizeof(size_t);
+	sssz = sizeof(ssize_t);
+	cpsz = sizeof(const char *);
+
+	CTL_GET("version", &cpv, const char *);
+	emitter_kv(emitter, "version", "Version", emitter_type_string, &cpv);
+
+	/* config. */
+	emitter_dict_begin(emitter, "config", "Build-time option settings");
+#define CONFIG_WRITE_BOOL(name)						\
+	do {								\
+		CTL_GET("config."#name, &bv, bool);			\
+		emitter_kv(emitter, #name, "config."#name,		\
+		    emitter_type_bool, &bv);				\
+	} while (0)
+
+	CONFIG_WRITE_BOOL(cache_oblivious);
+	CONFIG_WRITE_BOOL(debug);
+	CONFIG_WRITE_BOOL(fill);
+	CONFIG_WRITE_BOOL(lazy_lock);
+	emitter_kv(emitter, "malloc_conf", "config.malloc_conf",
+	    emitter_type_string, &config_malloc_conf);
+
+	CONFIG_WRITE_BOOL(opt_safety_checks);
+	CONFIG_WRITE_BOOL(prof);
+	CONFIG_WRITE_BOOL(prof_libgcc);
+	CONFIG_WRITE_BOOL(prof_libunwind);
+	CONFIG_WRITE_BOOL(stats);
+	CONFIG_WRITE_BOOL(utrace);
+	CONFIG_WRITE_BOOL(xmalloc);
+#undef CONFIG_WRITE_BOOL
+	emitter_dict_end(emitter); /* Close "config" dict. */
+
+	/* opt. */
+#define OPT_WRITE(name, var, size, emitter_type)			\
+	if (je_mallctl("opt."name, (void *)&var, &size, NULL, 0) ==	\
+	    0) {							\
+		emitter_kv(emitter, name, "opt."name, emitter_type,	\
+		    &var);						\
+	}
+
+#define OPT_WRITE_MUTABLE(name, var1, var2, size, emitter_type,		\
+    altname)								\
+	if (je_mallctl("opt."name, (void *)&var1, &size, NULL, 0) ==	\
+	    0 && je_mallctl(altname, (void *)&var2, &size, NULL, 0)	\
+	    == 0) {							\
+		emitter_kv_note(emitter, name, "opt."name,		\
+		    emitter_type, &var1, altname, emitter_type,		\
+		    &var2);						\
+	}
+
+#define OPT_WRITE_BOOL(name) OPT_WRITE(name, bv, bsz, emitter_type_bool)
+#define OPT_WRITE_BOOL_MUTABLE(name, altname)				\
+	OPT_WRITE_MUTABLE(name, bv, bv2, bsz, emitter_type_bool, altname)
+
+#define OPT_WRITE_UNSIGNED(name)					\
+	OPT_WRITE(name, uv, usz, emitter_type_unsigned)
+
+#define OPT_WRITE_SIZE_T(name)						\
+	OPT_WRITE(name, sv, ssz, emitter_type_size)
+#define OPT_WRITE_SSIZE_T(name)						\
+	OPT_WRITE(name, ssv, sssz, emitter_type_ssize)
+#define OPT_WRITE_SSIZE_T_MUTABLE(name, altname)			\
+	OPT_WRITE_MUTABLE(name, ssv, ssv2, sssz, emitter_type_ssize,	\
+	    altname)
+
+#define OPT_WRITE_CHAR_P(name)						\
+	OPT_WRITE(name, cpv, cpsz, emitter_type_string)
+
+	emitter_dict_begin(emitter, "opt", "Run-time option settings");
+
+	OPT_WRITE_BOOL("abort")
+	OPT_WRITE_BOOL("abort_conf")
+	OPT_WRITE_BOOL("confirm_conf")
+	OPT_WRITE_BOOL("retain")
+	OPT_WRITE_CHAR_P("dss")
+	OPT_WRITE_UNSIGNED("narenas")
+	OPT_WRITE_CHAR_P("percpu_arena")
+	OPT_WRITE_SIZE_T("oversize_threshold")
+	OPT_WRITE_CHAR_P("metadata_thp")
+	OPT_WRITE_BOOL_MUTABLE("background_thread", "background_thread")
+	OPT_WRITE_SSIZE_T_MUTABLE("dirty_decay_ms", "arenas.dirty_decay_ms")
+	OPT_WRITE_SSIZE_T_MUTABLE("muzzy_decay_ms", "arenas.muzzy_decay_ms")
+	OPT_WRITE_SIZE_T("lg_extent_max_active_fit")
+	OPT_WRITE_CHAR_P("junk")
+	OPT_WRITE_BOOL("zero")
+	OPT_WRITE_BOOL("utrace")
+	OPT_WRITE_BOOL("xmalloc")
+	OPT_WRITE_BOOL("tcache")
+	OPT_WRITE_SSIZE_T("lg_tcache_max")
+	OPT_WRITE_CHAR_P("thp")
+	OPT_WRITE_BOOL("prof")
+	OPT_WRITE_CHAR_P("prof_prefix")
+	OPT_WRITE_BOOL_MUTABLE("prof_active", "prof.active")
+	OPT_WRITE_BOOL_MUTABLE("prof_thread_active_init",
+	    "prof.thread_active_init")
+	OPT_WRITE_SSIZE_T_MUTABLE("lg_prof_sample", "prof.lg_sample")
+	OPT_WRITE_BOOL("prof_accum")
+	OPT_WRITE_SSIZE_T("lg_prof_interval")
+	OPT_WRITE_BOOL("prof_gdump")
+	OPT_WRITE_BOOL("prof_final")
+	OPT_WRITE_BOOL("prof_leak")
+	OPT_WRITE_BOOL("stats_print")
+	OPT_WRITE_CHAR_P("stats_print_opts")
+
+	emitter_dict_end(emitter);
+
+#undef OPT_WRITE
+#undef OPT_WRITE_MUTABLE
+#undef OPT_WRITE_BOOL
+#undef OPT_WRITE_BOOL_MUTABLE
+#undef OPT_WRITE_UNSIGNED
+#undef OPT_WRITE_SSIZE_T
+#undef OPT_WRITE_SSIZE_T_MUTABLE
+#undef OPT_WRITE_CHAR_P
+
+	/* prof. */
+	if (config_prof) {
+		emitter_dict_begin(emitter, "prof", "Profiling settings");
+
+		CTL_GET("prof.thread_active_init", &bv, bool);
+		emitter_kv(emitter, "thread_active_init",
+		    "prof.thread_active_init", emitter_type_bool, &bv);
+
+		CTL_GET("prof.active", &bv, bool);
+		emitter_kv(emitter, "active", "prof.active", emitter_type_bool,
+		    &bv);
+
+		CTL_GET("prof.gdump", &bv, bool);
+		emitter_kv(emitter, "gdump", "prof.gdump", emitter_type_bool,
+		    &bv);
+
+		CTL_GET("prof.interval", &u64v, uint64_t);
+		emitter_kv(emitter, "interval", "prof.interval",
+		    emitter_type_uint64, &u64v);
+
+		CTL_GET("prof.lg_sample", &ssv, ssize_t);
+		emitter_kv(emitter, "lg_sample", "prof.lg_sample",
+		    emitter_type_ssize, &ssv);
+
+		emitter_dict_end(emitter); /* Close "prof". */
+	}
+
+	/* arenas. */
+	/*
+	 * The json output sticks arena info into an "arenas" dict; the table
+	 * output puts them at the top-level.
+	 */
+	emitter_json_object_kv_begin(emitter, "arenas");
+
+	CTL_GET("arenas.narenas", &uv, unsigned);
+	emitter_kv(emitter, "narenas", "Arenas", emitter_type_unsigned, &uv);
+
+	/*
+	 * Decay settings are emitted only in json mode; in table mode, they're
+	 * emitted as notes with the opt output, above.
+	 */
+	CTL_GET("arenas.dirty_decay_ms", &ssv, ssize_t);
+	emitter_json_kv(emitter, "dirty_decay_ms", emitter_type_ssize, &ssv);
+
+	CTL_GET("arenas.muzzy_decay_ms", &ssv, ssize_t);
+	emitter_json_kv(emitter, "muzzy_decay_ms", emitter_type_ssize, &ssv);
+
+	CTL_GET("arenas.quantum", &sv, size_t);
+	emitter_kv(emitter, "quantum", "Quantum size", emitter_type_size, &sv);
+
+	CTL_GET("arenas.page", &sv, size_t);
+	emitter_kv(emitter, "page", "Page size", emitter_type_size, &sv);
+
+	if (je_mallctl("arenas.tcache_max", (void *)&sv, &ssz, NULL, 0) == 0) {
+		emitter_kv(emitter, "tcache_max",
+		    "Maximum thread-cached size class", emitter_type_size, &sv);
+	}
+
+	unsigned nbins;
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+	emitter_kv(emitter, "nbins", "Number of bin size classes",
+	    emitter_type_unsigned, &nbins);
+
+	unsigned nhbins;
+	CTL_GET("arenas.nhbins", &nhbins, unsigned);
+	emitter_kv(emitter, "nhbins", "Number of thread-cache bin size classes",
+	    emitter_type_unsigned, &nhbins);
+
+	/*
+	 * We do enough mallctls in a loop that we actually want to omit them
+	 * (not just omit the printing).
+	 */
+	if (emitter->output == emitter_output_json) {
+		emitter_json_array_kv_begin(emitter, "bin");
+		for (unsigned i = 0; i < nbins; i++) {
+			emitter_json_object_begin(emitter);
+
+			CTL_M2_GET("arenas.bin.0.size", i, &sv, size_t);
+			emitter_json_kv(emitter, "size", emitter_type_size,
+			    &sv);
+
+			CTL_M2_GET("arenas.bin.0.nregs", i, &u32v, uint32_t);
+			emitter_json_kv(emitter, "nregs", emitter_type_uint32,
+			    &u32v);
+
+			CTL_M2_GET("arenas.bin.0.slab_size", i, &sv, size_t);
+			emitter_json_kv(emitter, "slab_size", emitter_type_size,
+			    &sv);
+
+			CTL_M2_GET("arenas.bin.0.nshards", i, &u32v, uint32_t);
+			emitter_json_kv(emitter, "nshards", emitter_type_uint32,
+			    &u32v);
+
+			emitter_json_object_end(emitter);
+		}
+		emitter_json_array_end(emitter); /* Close "bin". */
+	}
+
+	unsigned nlextents;
+	CTL_GET("arenas.nlextents", &nlextents, unsigned);
+	emitter_kv(emitter, "nlextents", "Number of large size classes",
+	    emitter_type_unsigned, &nlextents);
+
+	if (emitter->output == emitter_output_json) {
+		emitter_json_array_kv_begin(emitter, "lextent");
+		for (unsigned i = 0; i < nlextents; i++) {
+			emitter_json_object_begin(emitter);
+
+			CTL_M2_GET("arenas.lextent.0.size", i, &sv, size_t);
+			emitter_json_kv(emitter, "size", emitter_type_size,
+			    &sv);
+
+			emitter_json_object_end(emitter);
+		}
+		emitter_json_array_end(emitter); /* Close "lextent". */
+	}
+
+	emitter_json_object_end(emitter); /* Close "arenas" */
+}
+
+static void
+stats_print_helper(emitter_t *emitter, bool merged, bool destroyed,
+    bool unmerged, bool bins, bool large, bool mutex, bool extents) {
+	/*
+	 * These should be deleted.  We keep them around for a while, to aid in
+	 * the transition to the emitter code.
+	 */
+	size_t allocated, active, metadata, metadata_thp, resident, mapped,
+	    retained;
+	size_t num_background_threads;
+	uint64_t background_thread_num_runs, background_thread_run_interval;
+
+	CTL_GET("stats.allocated", &allocated, size_t);
+	CTL_GET("stats.active", &active, size_t);
+	CTL_GET("stats.metadata", &metadata, size_t);
+	CTL_GET("stats.metadata_thp", &metadata_thp, size_t);
+	CTL_GET("stats.resident", &resident, size_t);
+	CTL_GET("stats.mapped", &mapped, size_t);
+	CTL_GET("stats.retained", &retained, size_t);
+
+	if (have_background_thread) {
+		CTL_GET("stats.background_thread.num_threads",
+		    &num_background_threads, size_t);
+		CTL_GET("stats.background_thread.num_runs",
+		    &background_thread_num_runs, uint64_t);
+		CTL_GET("stats.background_thread.run_interval",
+		    &background_thread_run_interval, uint64_t);
+	} else {
+		num_background_threads = 0;
+		background_thread_num_runs = 0;
+		background_thread_run_interval = 0;
+	}
+
+	/* Generic global stats. */
+	emitter_json_object_kv_begin(emitter, "stats");
+	emitter_json_kv(emitter, "allocated", emitter_type_size, &allocated);
+	emitter_json_kv(emitter, "active", emitter_type_size, &active);
+	emitter_json_kv(emitter, "metadata", emitter_type_size, &metadata);
+	emitter_json_kv(emitter, "metadata_thp", emitter_type_size,
+	    &metadata_thp);
+	emitter_json_kv(emitter, "resident", emitter_type_size, &resident);
+	emitter_json_kv(emitter, "mapped", emitter_type_size, &mapped);
+	emitter_json_kv(emitter, "retained", emitter_type_size, &retained);
+
+	emitter_table_printf(emitter, "Allocated: %zu, active: %zu, "
+	    "metadata: %zu (n_thp %zu), resident: %zu, mapped: %zu, "
+	    "retained: %zu\n", allocated, active, metadata, metadata_thp,
+	    resident, mapped, retained);
+
+	/* Background thread stats. */
+	emitter_json_object_kv_begin(emitter, "background_thread");
+	emitter_json_kv(emitter, "num_threads", emitter_type_size,
+	    &num_background_threads);
+	emitter_json_kv(emitter, "num_runs", emitter_type_uint64,
+	    &background_thread_num_runs);
+	emitter_json_kv(emitter, "run_interval", emitter_type_uint64,
+	    &background_thread_run_interval);
+	emitter_json_object_end(emitter); /* Close "background_thread". */
+
+	emitter_table_printf(emitter, "Background threads: %zu, "
+	    "num_runs: %"FMTu64", run_interval: %"FMTu64" ns\n",
+	    num_background_threads, background_thread_num_runs,
+	    background_thread_run_interval);
+
+	if (mutex) {
+		emitter_row_t row;
+		emitter_col_t name;
+		emitter_col_t col64[mutex_prof_num_uint64_t_counters];
+		emitter_col_t col32[mutex_prof_num_uint32_t_counters];
+		uint64_t uptime;
+
+		emitter_row_init(&row);
+		mutex_stats_init_cols(&row, "", &name, col64, col32);
+
+		emitter_table_row(emitter, &row);
+		emitter_json_object_kv_begin(emitter, "mutexes");
+
+		CTL_M2_GET("stats.arenas.0.uptime", 0, &uptime, uint64_t);
+
+		for (int i = 0; i < mutex_prof_num_global_mutexes; i++) {
+			mutex_stats_read_global(global_mutex_names[i], &name,
+			    col64, col32, uptime);
+			emitter_json_object_kv_begin(emitter, global_mutex_names[i]);
+			mutex_stats_emit(emitter, &row, col64, col32);
+			emitter_json_object_end(emitter);
+		}
+
+		emitter_json_object_end(emitter); /* Close "mutexes". */
+	}
+
+	emitter_json_object_end(emitter); /* Close "stats". */
+
+	if (merged || destroyed || unmerged) {
+		unsigned narenas;
+
+		emitter_json_object_kv_begin(emitter, "stats.arenas");
+
+		CTL_GET("arenas.narenas", &narenas, unsigned);
+		size_t mib[3];
+		size_t miblen = sizeof(mib) / sizeof(size_t);
+		size_t sz;
+		VARIABLE_ARRAY(bool, initialized, narenas);
+		bool destroyed_initialized;
+		unsigned i, j, ninitialized;
+
+		xmallctlnametomib("arena.0.initialized", mib, &miblen);
+		for (i = ninitialized = 0; i < narenas; i++) {
+			mib[1] = i;
+			sz = sizeof(bool);
+			xmallctlbymib(mib, miblen, &initialized[i], &sz,
+			    NULL, 0);
+			if (initialized[i]) {
+				ninitialized++;
+			}
+		}
+		mib[1] = MALLCTL_ARENAS_DESTROYED;
+		sz = sizeof(bool);
+		xmallctlbymib(mib, miblen, &destroyed_initialized, &sz,
+		    NULL, 0);
+
+		/* Merged stats. */
+		if (merged && (ninitialized > 1 || !unmerged)) {
+			/* Print merged arena stats. */
+			emitter_table_printf(emitter, "Merged arenas stats:\n");
+			emitter_json_object_kv_begin(emitter, "merged");
+			stats_arena_print(emitter, MALLCTL_ARENAS_ALL, bins,
+			    large, mutex, extents);
+			emitter_json_object_end(emitter); /* Close "merged". */
+		}
+
+		/* Destroyed stats. */
+		if (destroyed_initialized && destroyed) {
+			/* Print destroyed arena stats. */
+			emitter_table_printf(emitter,
+			    "Destroyed arenas stats:\n");
+			emitter_json_object_kv_begin(emitter, "destroyed");
+			stats_arena_print(emitter, MALLCTL_ARENAS_DESTROYED,
+			    bins, large, mutex, extents);
+			emitter_json_object_end(emitter); /* Close "destroyed". */
+		}
+
+		/* Unmerged stats. */
+		if (unmerged) {
+			for (i = j = 0; i < narenas; i++) {
+				if (initialized[i]) {
+					char arena_ind_str[20];
+					malloc_snprintf(arena_ind_str,
+					    sizeof(arena_ind_str), "%u", i);
+					emitter_json_object_kv_begin(emitter,
+					    arena_ind_str);
+					emitter_table_printf(emitter,
+					    "arenas[%s]:\n", arena_ind_str);
+					stats_arena_print(emitter, i, bins,
+					    large, mutex, extents);
+					/* Close "<arena-ind>". */
+					emitter_json_object_end(emitter);
+				}
+			}
+		}
+		emitter_json_object_end(emitter); /* Close "stats.arenas". */
+	}
+}
+
+void
+stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts) {
+	int err;
+	uint64_t epoch;
+	size_t u64sz;
+#define OPTION(o, v, d, s) bool v = d;
+	STATS_PRINT_OPTIONS
+#undef OPTION
+
+	/*
+	 * Refresh stats, in case mallctl() was called by the application.
+	 *
+	 * Check for OOM here, since refreshing the ctl cache can trigger
+	 * allocation.  In practice, none of the subsequent mallctl()-related
+	 * calls in this function will cause OOM if this one succeeds.
+	 * */
+	epoch = 1;
+	u64sz = sizeof(uint64_t);
+	err = je_mallctl("epoch", (void *)&epoch, &u64sz, (void *)&epoch,
+	    sizeof(uint64_t));
+	if (err != 0) {
+		if (err == EAGAIN) {
+			malloc_write("<jemalloc>: Memory allocation failure in "
+			    "mallctl(\"epoch\", ...)\n");
+			return;
+		}
+		malloc_write("<jemalloc>: Failure in mallctl(\"epoch\", "
+		    "...)\n");
+		abort();
+	}
+
+	if (opts != NULL) {
+		for (unsigned i = 0; opts[i] != '\0'; i++) {
+			switch (opts[i]) {
+#define OPTION(o, v, d, s) case o: v = s; break;
+				STATS_PRINT_OPTIONS
+#undef OPTION
+			default:;
+			}
+		}
+	}
+
+	emitter_t emitter;
+	emitter_init(&emitter,
+	    json ? emitter_output_json : emitter_output_table, write_cb,
+	    cbopaque);
+	emitter_begin(&emitter);
+	emitter_table_printf(&emitter, "___ Begin jemalloc statistics ___\n");
+	emitter_json_object_kv_begin(&emitter, "jemalloc");
+
+	if (general) {
+		stats_general_print(&emitter);
+	}
+	if (config_stats) {
+		stats_print_helper(&emitter, merged, destroyed, unmerged,
+		    bins, large, mutex, extents);
+	}
+
+	emitter_json_object_end(&emitter); /* Closes the "jemalloc" dict. */
+	emitter_table_printf(&emitter, "--- End jemalloc statistics ---\n");
+	emitter_end(&emitter);
+}
diff --git a/memkind-1.10.0/jemalloc/src/sz.c b/memkind-1.10.0/jemalloc/src/sz.c
new file mode 100644
index 0000000..8633fb0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/sz.c
@@ -0,0 +1,64 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/sz.h"
+
+JEMALLOC_ALIGNED(CACHELINE)
+size_t sz_pind2sz_tab[SC_NPSIZES+1];
+
+static void
+sz_boot_pind2sz_tab(const sc_data_t *sc_data) {
+	int pind = 0;
+	for (unsigned i = 0; i < SC_NSIZES; i++) {
+		const sc_t *sc = &sc_data->sc[i];
+		if (sc->psz) {
+			sz_pind2sz_tab[pind] = (ZU(1) << sc->lg_base)
+			    + (ZU(sc->ndelta) << sc->lg_delta);
+			pind++;
+		}
+	}
+	for (int i = pind; i <= (int)SC_NPSIZES; i++) {
+		sz_pind2sz_tab[pind] = sc_data->large_maxclass + PAGE;
+	}
+}
+
+JEMALLOC_ALIGNED(CACHELINE)
+size_t sz_index2size_tab[SC_NSIZES];
+
+static void
+sz_boot_index2size_tab(const sc_data_t *sc_data) {
+	for (unsigned i = 0; i < SC_NSIZES; i++) {
+		const sc_t *sc = &sc_data->sc[i];
+		sz_index2size_tab[i] = (ZU(1) << sc->lg_base)
+		    + (ZU(sc->ndelta) << (sc->lg_delta));
+	}
+}
+
+/*
+ * To keep this table small, we divide sizes by the tiny min size, which gives
+ * the smallest interval for which the result can change.
+ */
+JEMALLOC_ALIGNED(CACHELINE)
+uint8_t sz_size2index_tab[(SC_LOOKUP_MAXCLASS >> SC_LG_TINY_MIN) + 1];
+
+static void
+sz_boot_size2index_tab(const sc_data_t *sc_data) {
+	size_t dst_max = (SC_LOOKUP_MAXCLASS >> SC_LG_TINY_MIN) + 1;
+	size_t dst_ind = 0;
+	for (unsigned sc_ind = 0; sc_ind < SC_NSIZES && dst_ind < dst_max;
+	    sc_ind++) {
+		const sc_t *sc = &sc_data->sc[sc_ind];
+		size_t sz = (ZU(1) << sc->lg_base)
+		    + (ZU(sc->ndelta) << sc->lg_delta);
+		size_t max_ind = ((sz + (ZU(1) << SC_LG_TINY_MIN) - 1)
+				   >> SC_LG_TINY_MIN);
+		for (; dst_ind <= max_ind && dst_ind < dst_max; dst_ind++) {
+			sz_size2index_tab[dst_ind] = sc_ind;
+		}
+	}
+}
+
+void
+sz_boot(const sc_data_t *sc_data) {
+	sz_boot_pind2sz_tab(sc_data);
+	sz_boot_index2size_tab(sc_data);
+	sz_boot_size2index_tab(sc_data);
+}
diff --git a/memkind-1.10.0/jemalloc/src/tcache.c b/memkind-1.10.0/jemalloc/src/tcache.c
new file mode 100644
index 0000000..50099a9
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/tcache.c
@@ -0,0 +1,798 @@
+#define JEMALLOC_TCACHE_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/safety_check.h"
+#include "jemalloc/internal/sc.h"
+
+/******************************************************************************/
+/* Data. */
+
+bool	opt_tcache = true;
+ssize_t	opt_lg_tcache_max = LG_TCACHE_MAXCLASS_DEFAULT;
+
+cache_bin_info_t	*tcache_bin_info;
+static unsigned		stack_nelms; /* Total stack elms per tcache. */
+
+unsigned		nhbins;
+size_t			tcache_maxclass;
+
+tcaches_t		*tcaches;
+
+/* Index of first element within tcaches that has never been used. */
+static unsigned		tcaches_past;
+
+/* Head of singly linked list tracking available tcaches elements. */
+static tcaches_t	*tcaches_avail;
+
+/* Protects tcaches{,_past,_avail}. */
+static malloc_mutex_t	tcaches_mtx;
+
+/******************************************************************************/
+
+size_t
+tcache_salloc(tsdn_t *tsdn, const void *ptr) {
+	return arena_salloc(tsdn, ptr);
+}
+
+void
+tcache_event_hard(tsd_t *tsd, tcache_t *tcache) {
+	szind_t binind = tcache->next_gc_bin;
+
+	cache_bin_t *tbin;
+	if (binind < SC_NBINS) {
+		tbin = tcache_small_bin_get(tcache, binind);
+	} else {
+		tbin = tcache_large_bin_get(tcache, binind);
+	}
+	if (tbin->low_water > 0) {
+		/*
+		 * Flush (ceiling) 3/4 of the objects below the low water mark.
+		 */
+		if (binind < SC_NBINS) {
+			tcache_bin_flush_small(tsd, tcache, tbin, binind,
+			    tbin->ncached - tbin->low_water + (tbin->low_water
+			    >> 2));
+			/*
+			 * Reduce fill count by 2X.  Limit lg_fill_div such that
+			 * the fill count is always at least 1.
+			 */
+			cache_bin_info_t *tbin_info = &tcache_bin_info[binind];
+			if ((tbin_info->ncached_max >>
+			     (tcache->lg_fill_div[binind] + 1)) >= 1) {
+				tcache->lg_fill_div[binind]++;
+			}
+		} else {
+			tcache_bin_flush_large(tsd, tbin, binind, tbin->ncached
+			    - tbin->low_water + (tbin->low_water >> 2), tcache);
+		}
+	} else if (tbin->low_water < 0) {
+		/*
+		 * Increase fill count by 2X for small bins.  Make sure
+		 * lg_fill_div stays greater than 0.
+		 */
+		if (binind < SC_NBINS && tcache->lg_fill_div[binind] > 1) {
+			tcache->lg_fill_div[binind]--;
+		}
+	}
+	tbin->low_water = tbin->ncached;
+
+	tcache->next_gc_bin++;
+	if (tcache->next_gc_bin == nhbins) {
+		tcache->next_gc_bin = 0;
+	}
+}
+
+void *
+tcache_alloc_small_hard(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, bool *tcache_success) {
+	void *ret;
+
+	assert(tcache->arena != NULL);
+	arena_tcache_fill_small(tsdn, arena, tcache, tbin, binind,
+	    config_prof ? tcache->prof_accumbytes : 0);
+	if (config_prof) {
+		tcache->prof_accumbytes = 0;
+	}
+	ret = cache_bin_alloc_easy(tbin, tcache_success);
+
+	return ret;
+}
+
+/* Enabled with --enable-extra-size-check. */
+static void
+tbin_extents_lookup_size_check(tsdn_t *tsdn, cache_bin_t *tbin, szind_t binind,
+    size_t nflush, extent_t **extents){
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	/*
+	 * Verify that the items in the tcache all have the correct size; this
+	 * is useful for catching sized deallocation bugs, also to fail early
+	 * instead of corrupting metadata.  Since this can be turned on for opt
+	 * builds, avoid the branch in the loop.
+	 */
+	szind_t szind;
+	size_t sz_sum = binind * nflush;
+	for (unsigned i = 0 ; i < nflush; i++) {
+		rtree_extent_szind_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)*(tbin->avail - 1 - i), true,
+		    &extents[i], &szind);
+		sz_sum -= szind;
+	}
+	if (sz_sum != 0) {
+		safety_check_fail("<jemalloc>: size mismatch in thread cache "
+		    "detected, likely caused by sized deallocation bugs by "
+		    "application. Abort.\n");
+		abort();
+	}
+}
+
+void
+tcache_bin_flush_small(tsd_t *tsd, tcache_t *tcache, cache_bin_t *tbin,
+    szind_t binind, unsigned rem) {
+	bool merged_stats = false;
+
+	assert(binind < SC_NBINS);
+	assert((cache_bin_sz_t)rem <= tbin->ncached);
+
+	arena_t *arena = tcache->arena;
+	assert(arena != NULL);
+	unsigned nflush = tbin->ncached - rem;
+	VARIABLE_ARRAY(extent_t *, item_extent, nflush);
+
+	/* Look up extent once per item. */
+	if (config_opt_safety_checks) {
+		tbin_extents_lookup_size_check(tsd_tsdn(tsd), tbin, binind,
+		    nflush, item_extent);
+	} else {
+		for (unsigned i = 0 ; i < nflush; i++) {
+			item_extent[i] = iealloc(tsd_tsdn(tsd),
+			    *(tbin->avail - 1 - i));
+		}
+	}
+	while (nflush > 0) {
+		/* Lock the arena bin associated with the first object. */
+		extent_t *extent = item_extent[0];
+		unsigned bin_arena_ind = extent_arena_ind_get(extent);
+		arena_t *bin_arena = arena_get(tsd_tsdn(tsd), bin_arena_ind,
+		    false);
+		unsigned binshard = extent_binshard_get(extent);
+		assert(binshard < bin_infos[binind].n_shards);
+		bin_t *bin = &bin_arena->bins[binind].bin_shards[binshard];
+
+		if (config_prof && bin_arena == arena) {
+			if (arena_prof_accum(tsd_tsdn(tsd), arena,
+			    tcache->prof_accumbytes)) {
+				prof_idump(tsd_tsdn(tsd));
+			}
+			tcache->prof_accumbytes = 0;
+		}
+
+		malloc_mutex_lock(tsd_tsdn(tsd), &bin->lock);
+		if (config_stats && bin_arena == arena && !merged_stats) {
+			merged_stats = true;
+			bin->stats.nflushes++;
+			bin->stats.nrequests += tbin->tstats.nrequests;
+			tbin->tstats.nrequests = 0;
+		}
+		unsigned ndeferred = 0;
+		for (unsigned i = 0; i < nflush; i++) {
+			void *ptr = *(tbin->avail - 1 - i);
+			extent = item_extent[i];
+			assert(ptr != NULL && extent != NULL);
+
+			if (extent_arena_ind_get(extent) == bin_arena_ind
+			    && extent_binshard_get(extent) == binshard) {
+				arena_dalloc_bin_junked_locked(tsd_tsdn(tsd),
+				    bin_arena, bin, binind, extent, ptr);
+			} else {
+				/*
+				 * This object was allocated via a different
+				 * arena bin than the one that is currently
+				 * locked.  Stash the object, so that it can be
+				 * handled in a future pass.
+				 */
+				*(tbin->avail - 1 - ndeferred) = ptr;
+				item_extent[ndeferred] = extent;
+				ndeferred++;
+			}
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+		arena_decay_ticks(tsd_tsdn(tsd), bin_arena, nflush - ndeferred);
+		nflush = ndeferred;
+	}
+	if (config_stats && !merged_stats) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		unsigned binshard;
+		bin_t *bin = arena_bin_choose_lock(tsd_tsdn(tsd), arena, binind,
+		    &binshard);
+		bin->stats.nflushes++;
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+		malloc_mutex_unlock(tsd_tsdn(tsd), &bin->lock);
+	}
+
+	memmove(tbin->avail - rem, tbin->avail - tbin->ncached, rem *
+	    sizeof(void *));
+	tbin->ncached = rem;
+	if (tbin->ncached < tbin->low_water) {
+		tbin->low_water = tbin->ncached;
+	}
+}
+
+void
+tcache_bin_flush_large(tsd_t *tsd, cache_bin_t *tbin, szind_t binind,
+    unsigned rem, tcache_t *tcache) {
+	bool merged_stats = false;
+
+	assert(binind < nhbins);
+	assert((cache_bin_sz_t)rem <= tbin->ncached);
+
+	arena_t *tcache_arena = tcache->arena;
+	assert(tcache_arena != NULL);
+	unsigned nflush = tbin->ncached - rem;
+	VARIABLE_ARRAY(extent_t *, item_extent, nflush);
+
+#ifndef JEMALLOC_EXTRA_SIZE_CHECK
+	/* Look up extent once per item. */
+	for (unsigned i = 0 ; i < nflush; i++) {
+		item_extent[i] = iealloc(tsd_tsdn(tsd), *(tbin->avail - 1 - i));
+	}
+#else
+	tbin_extents_lookup_size_check(tsd_tsdn(tsd), tbin, binind, nflush,
+	    item_extent);
+#endif
+	while (nflush > 0) {
+		/* Lock the arena associated with the first object. */
+		extent_t *extent = item_extent[0];
+		unsigned locked_arena_ind = extent_arena_ind_get(extent);
+		arena_t *locked_arena = arena_get(tsd_tsdn(tsd),
+		    locked_arena_ind, false);
+		bool idump;
+
+		if (config_prof) {
+			idump = false;
+		}
+
+		bool lock_large = !arena_is_auto(locked_arena);
+		if (lock_large) {
+			malloc_mutex_lock(tsd_tsdn(tsd), &locked_arena->large_mtx);
+		}
+		for (unsigned i = 0; i < nflush; i++) {
+			void *ptr = *(tbin->avail - 1 - i);
+			assert(ptr != NULL);
+			extent = item_extent[i];
+			if (extent_arena_ind_get(extent) == locked_arena_ind) {
+				large_dalloc_prep_junked_locked(tsd_tsdn(tsd),
+				    extent);
+			}
+		}
+		if ((config_prof || config_stats) &&
+		    (locked_arena == tcache_arena)) {
+			if (config_prof) {
+				idump = arena_prof_accum(tsd_tsdn(tsd),
+				    tcache_arena, tcache->prof_accumbytes);
+				tcache->prof_accumbytes = 0;
+			}
+			if (config_stats) {
+				merged_stats = true;
+				arena_stats_large_flush_nrequests_add(
+				    tsd_tsdn(tsd), &tcache_arena->stats, binind,
+				    tbin->tstats.nrequests);
+				tbin->tstats.nrequests = 0;
+			}
+		}
+		if (lock_large) {
+			malloc_mutex_unlock(tsd_tsdn(tsd), &locked_arena->large_mtx);
+		}
+
+		unsigned ndeferred = 0;
+		for (unsigned i = 0; i < nflush; i++) {
+			void *ptr = *(tbin->avail - 1 - i);
+			extent = item_extent[i];
+			assert(ptr != NULL && extent != NULL);
+
+			if (extent_arena_ind_get(extent) == locked_arena_ind) {
+				large_dalloc_finish(tsd_tsdn(tsd), extent);
+			} else {
+				/*
+				 * This object was allocated via a different
+				 * arena than the one that is currently locked.
+				 * Stash the object, so that it can be handled
+				 * in a future pass.
+				 */
+				*(tbin->avail - 1 - ndeferred) = ptr;
+				item_extent[ndeferred] = extent;
+				ndeferred++;
+			}
+		}
+		if (config_prof && idump) {
+			prof_idump(tsd_tsdn(tsd));
+		}
+		arena_decay_ticks(tsd_tsdn(tsd), locked_arena, nflush -
+		    ndeferred);
+		nflush = ndeferred;
+	}
+	if (config_stats && !merged_stats) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		arena_stats_large_flush_nrequests_add(tsd_tsdn(tsd),
+		    &tcache_arena->stats, binind, tbin->tstats.nrequests);
+		tbin->tstats.nrequests = 0;
+	}
+
+	memmove(tbin->avail - rem, tbin->avail - tbin->ncached, rem *
+	    sizeof(void *));
+	tbin->ncached = rem;
+	if (tbin->ncached < tbin->low_water) {
+		tbin->low_water = tbin->ncached;
+	}
+}
+
+void
+tcache_arena_associate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) {
+	assert(tcache->arena == NULL);
+	tcache->arena = arena;
+
+	if (config_stats) {
+		/* Link into list of extant tcaches. */
+		malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+
+		ql_elm_new(tcache, link);
+		ql_tail_insert(&arena->tcache_ql, tcache, link);
+		cache_bin_array_descriptor_init(
+		    &tcache->cache_bin_array_descriptor, tcache->bins_small,
+		    tcache->bins_large);
+		ql_tail_insert(&arena->cache_bin_array_descriptor_ql,
+		    &tcache->cache_bin_array_descriptor, link);
+
+		malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx);
+	}
+}
+
+static void
+tcache_arena_dissociate(tsdn_t *tsdn, tcache_t *tcache) {
+	arena_t *arena = tcache->arena;
+	assert(arena != NULL);
+	if (config_stats) {
+		/* Unlink from list of extant tcaches. */
+		malloc_mutex_lock(tsdn, &arena->tcache_ql_mtx);
+		if (config_debug) {
+			bool in_ql = false;
+			tcache_t *iter;
+			ql_foreach(iter, &arena->tcache_ql, link) {
+				if (iter == tcache) {
+					in_ql = true;
+					break;
+				}
+			}
+			assert(in_ql);
+		}
+		ql_remove(&arena->tcache_ql, tcache, link);
+		ql_remove(&arena->cache_bin_array_descriptor_ql,
+		    &tcache->cache_bin_array_descriptor, link);
+		tcache_stats_merge(tsdn, tcache, arena);
+		malloc_mutex_unlock(tsdn, &arena->tcache_ql_mtx);
+	}
+	tcache->arena = NULL;
+}
+
+void
+tcache_arena_reassociate(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) {
+	tcache_arena_dissociate(tsdn, tcache);
+	tcache_arena_associate(tsdn, tcache, arena);
+}
+
+bool
+tsd_tcache_enabled_data_init(tsd_t *tsd) {
+	/* Called upon tsd initialization. */
+	tsd_tcache_enabled_set(tsd, opt_tcache);
+	tsd_slow_update(tsd);
+
+	if (opt_tcache) {
+		/* Trigger tcache init. */
+		tsd_tcache_data_init(tsd);
+	}
+
+	return false;
+}
+
+/* Initialize auto tcache (embedded in TSD). */
+static void
+tcache_init(tsd_t *tsd, tcache_t *tcache, void *avail_stack) {
+	memset(&tcache->link, 0, sizeof(ql_elm(tcache_t)));
+	tcache->prof_accumbytes = 0;
+	tcache->next_gc_bin = 0;
+	tcache->arena = NULL;
+
+	ticker_init(&tcache->gc_ticker, TCACHE_GC_INCR);
+
+	size_t stack_offset = 0;
+	assert((TCACHE_NSLOTS_SMALL_MAX & 1U) == 0);
+	memset(tcache->bins_small, 0, sizeof(cache_bin_t) * SC_NBINS);
+	memset(tcache->bins_large, 0, sizeof(cache_bin_t) * (nhbins - SC_NBINS));
+	unsigned i = 0;
+	for (; i < SC_NBINS; i++) {
+		tcache->lg_fill_div[i] = 1;
+		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *);
+		/*
+		 * avail points past the available space.  Allocations will
+		 * access the slots toward higher addresses (for the benefit of
+		 * prefetch).
+		 */
+		tcache_small_bin_get(tcache, i)->avail =
+		    (void **)((uintptr_t)avail_stack + (uintptr_t)stack_offset);
+	}
+	for (; i < nhbins; i++) {
+		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *);
+		tcache_large_bin_get(tcache, i)->avail =
+		    (void **)((uintptr_t)avail_stack + (uintptr_t)stack_offset);
+	}
+	assert(stack_offset == stack_nelms * sizeof(void *));
+}
+
+/* Initialize auto tcache (embedded in TSD). */
+bool
+tsd_tcache_data_init(tsd_t *tsd) {
+	tcache_t *tcache = tsd_tcachep_get_unsafe(tsd);
+	assert(tcache_small_bin_get(tcache, 0)->avail == NULL);
+	size_t size = stack_nelms * sizeof(void *);
+	/* Avoid false cacheline sharing. */
+	size = sz_sa2u(size, CACHELINE);
+
+	void *avail_array = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true,
+	    NULL, true, arena_get(TSDN_NULL, 0, true));
+	if (avail_array == NULL) {
+		return true;
+	}
+
+	tcache_init(tsd, tcache, avail_array);
+	/*
+	 * Initialization is a bit tricky here.  After malloc init is done, all
+	 * threads can rely on arena_choose and associate tcache accordingly.
+	 * However, the thread that does actual malloc bootstrapping relies on
+	 * functional tsd, and it can only rely on a0.  In that case, we
+	 * associate its tcache to a0 temporarily, and later on
+	 * arena_choose_hard() will re-associate properly.
+	 */
+	tcache->arena = NULL;
+	arena_t *arena;
+	if (!malloc_initialized()) {
+		/* If in initialization, assign to a0. */
+		arena = arena_get(tsd_tsdn(tsd), 0, false);
+		tcache_arena_associate(tsd_tsdn(tsd), tcache, arena);
+	} else {
+		arena = arena_choose(tsd, NULL);
+		/* This may happen if thread.tcache.enabled is used. */
+		if (tcache->arena == NULL) {
+			tcache_arena_associate(tsd_tsdn(tsd), tcache, arena);
+		}
+	}
+	assert(arena == tcache->arena);
+
+	return false;
+}
+
+/* Created manual tcache for tcache.create mallctl. */
+tcache_t *
+tcache_create_explicit(tsd_t *tsd) {
+	tcache_t *tcache;
+	size_t size, stack_offset;
+
+	size = sizeof(tcache_t);
+	/* Naturally align the pointer stacks. */
+	size = PTR_CEILING(size);
+	stack_offset = size;
+	size += stack_nelms * sizeof(void *);
+	/* Avoid false cacheline sharing. */
+	size = sz_sa2u(size, CACHELINE);
+
+	tcache = ipallocztm(tsd_tsdn(tsd), size, CACHELINE, true, NULL, true,
+	    arena_get(TSDN_NULL, 0, true));
+	if (tcache == NULL) {
+		return NULL;
+	}
+
+	tcache_init(tsd, tcache,
+	    (void *)((uintptr_t)tcache + (uintptr_t)stack_offset));
+	tcache_arena_associate(tsd_tsdn(tsd), tcache, arena_ichoose(tsd, NULL));
+
+	return tcache;
+}
+
+static void
+tcache_flush_cache(tsd_t *tsd, tcache_t *tcache) {
+	assert(tcache->arena != NULL);
+
+	for (unsigned i = 0; i < SC_NBINS; i++) {
+		cache_bin_t *tbin = tcache_small_bin_get(tcache, i);
+		tcache_bin_flush_small(tsd, tcache, tbin, i, 0);
+
+		if (config_stats) {
+			assert(tbin->tstats.nrequests == 0);
+		}
+	}
+	for (unsigned i = SC_NBINS; i < nhbins; i++) {
+		cache_bin_t *tbin = tcache_large_bin_get(tcache, i);
+		tcache_bin_flush_large(tsd, tbin, i, 0, tcache);
+
+		if (config_stats) {
+			assert(tbin->tstats.nrequests == 0);
+		}
+	}
+
+	if (config_prof && tcache->prof_accumbytes > 0 &&
+	    arena_prof_accum(tsd_tsdn(tsd), tcache->arena,
+	    tcache->prof_accumbytes)) {
+		prof_idump(tsd_tsdn(tsd));
+	}
+}
+
+void
+tcache_flush(tsd_t *tsd) {
+	assert(tcache_available(tsd));
+	tcache_flush_cache(tsd, tsd_tcachep_get(tsd));
+}
+
+static void
+tcache_destroy(tsd_t *tsd, tcache_t *tcache, bool tsd_tcache) {
+	tcache_flush_cache(tsd, tcache);
+	arena_t *arena = tcache->arena;
+	tcache_arena_dissociate(tsd_tsdn(tsd), tcache);
+
+	if (tsd_tcache) {
+		/* Release the avail array for the TSD embedded auto tcache. */
+		void *avail_array =
+		    (void *)((uintptr_t)tcache_small_bin_get(tcache, 0)->avail -
+		    (uintptr_t)tcache_bin_info[0].ncached_max * sizeof(void *));
+		idalloctm(tsd_tsdn(tsd), avail_array, NULL, NULL, true, true);
+	} else {
+		/* Release both the tcache struct and avail array. */
+		idalloctm(tsd_tsdn(tsd), tcache, NULL, NULL, true, true);
+	}
+
+	/*
+	 * The deallocation and tcache flush above may not trigger decay since
+	 * we are on the tcache shutdown path (potentially with non-nominal
+	 * tsd).  Manually trigger decay to avoid pathological cases.  Also
+	 * include arena 0 because the tcache array is allocated from it.
+	 */
+	arena_decay(tsd_tsdn(tsd), arena_get(tsd_tsdn(tsd), 0, false),
+	    false, false);
+
+	if (arena_nthreads_get(arena, false) == 0 &&
+	    !background_thread_enabled()) {
+		/* Force purging when no threads assigned to the arena anymore. */
+		arena_decay(tsd_tsdn(tsd), arena, false, true);
+	} else {
+		arena_decay(tsd_tsdn(tsd), arena, false, false);
+	}
+}
+
+/* For auto tcache (embedded in TSD) only. */
+void
+tcache_cleanup(tsd_t *tsd) {
+	tcache_t *tcache = tsd_tcachep_get(tsd);
+	if (!tcache_available(tsd)) {
+		assert(tsd_tcache_enabled_get(tsd) == false);
+		if (config_debug) {
+			assert(tcache_small_bin_get(tcache, 0)->avail == NULL);
+		}
+		return;
+	}
+	assert(tsd_tcache_enabled_get(tsd));
+	assert(tcache_small_bin_get(tcache, 0)->avail != NULL);
+
+	tcache_destroy(tsd, tcache, true);
+	if (config_debug) {
+		tcache_small_bin_get(tcache, 0)->avail = NULL;
+	}
+}
+
+void
+tcache_stats_merge(tsdn_t *tsdn, tcache_t *tcache, arena_t *arena) {
+	unsigned i;
+
+	cassert(config_stats);
+
+	/* Merge and reset tcache stats. */
+	for (i = 0; i < SC_NBINS; i++) {
+		cache_bin_t *tbin = tcache_small_bin_get(tcache, i);
+		unsigned binshard;
+		bin_t *bin = arena_bin_choose_lock(tsdn, arena, i, &binshard);
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		malloc_mutex_unlock(tsdn, &bin->lock);
+		tbin->tstats.nrequests = 0;
+	}
+
+	for (; i < nhbins; i++) {
+		cache_bin_t *tbin = tcache_large_bin_get(tcache, i);
+		arena_stats_large_flush_nrequests_add(tsdn, &arena->stats, i,
+		    tbin->tstats.nrequests);
+		tbin->tstats.nrequests = 0;
+	}
+}
+
+static bool
+tcaches_create_prep(tsd_t *tsd) {
+	bool err;
+
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+
+	if (tcaches == NULL) {
+		tcaches = base_alloc(tsd_tsdn(tsd), b0get(), sizeof(tcache_t *)
+		    * (MALLOCX_TCACHE_MAX+1), CACHELINE);
+		if (tcaches == NULL) {
+			err = true;
+			goto label_return;
+		}
+	}
+
+	if (tcaches_avail == NULL && tcaches_past > MALLOCX_TCACHE_MAX) {
+		err = true;
+		goto label_return;
+	}
+
+	err = false;
+label_return:
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+	return err;
+}
+
+bool
+tcaches_create(tsd_t *tsd, unsigned *r_ind) {
+	witness_assert_depth(tsdn_witness_tsdp_get(tsd_tsdn(tsd)), 0);
+
+	bool err;
+
+	if (tcaches_create_prep(tsd)) {
+		err = true;
+		goto label_return;
+	}
+
+	tcache_t *tcache = tcache_create_explicit(tsd);
+	if (tcache == NULL) {
+		err = true;
+		goto label_return;
+	}
+
+	tcaches_t *elm;
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+	if (tcaches_avail != NULL) {
+		elm = tcaches_avail;
+		tcaches_avail = tcaches_avail->next;
+		elm->tcache = tcache;
+		*r_ind = (unsigned)(elm - tcaches);
+	} else {
+		elm = &tcaches[tcaches_past];
+		elm->tcache = tcache;
+		*r_ind = tcaches_past;
+		tcaches_past++;
+	}
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+
+	err = false;
+label_return:
+	witness_assert_depth(tsdn_witness_tsdp_get(tsd_tsdn(tsd)), 0);
+	return err;
+}
+
+static tcache_t *
+tcaches_elm_remove(tsd_t *tsd, tcaches_t *elm, bool allow_reinit) {
+	malloc_mutex_assert_owner(tsd_tsdn(tsd), &tcaches_mtx);
+
+	if (elm->tcache == NULL) {
+		return NULL;
+	}
+	tcache_t *tcache = elm->tcache;
+	if (allow_reinit) {
+		elm->tcache = TCACHES_ELM_NEED_REINIT;
+	} else {
+		elm->tcache = NULL;
+	}
+
+	if (tcache == TCACHES_ELM_NEED_REINIT) {
+		return NULL;
+	}
+	return tcache;
+}
+
+void
+tcaches_flush(tsd_t *tsd, unsigned ind) {
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+	tcache_t *tcache = tcaches_elm_remove(tsd, &tcaches[ind], true);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+	if (tcache != NULL) {
+		/* Destroy the tcache; recreate in tcaches_get() if needed. */
+		tcache_destroy(tsd, tcache, false);
+	}
+}
+
+void
+tcaches_destroy(tsd_t *tsd, unsigned ind) {
+	malloc_mutex_lock(tsd_tsdn(tsd), &tcaches_mtx);
+	tcaches_t *elm = &tcaches[ind];
+	tcache_t *tcache = tcaches_elm_remove(tsd, elm, false);
+	elm->next = tcaches_avail;
+	tcaches_avail = elm;
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tcaches_mtx);
+	if (tcache != NULL) {
+		tcache_destroy(tsd, tcache, false);
+	}
+}
+
+bool
+tcache_boot(tsdn_t *tsdn) {
+	/* If necessary, clamp opt_lg_tcache_max. */
+	if (opt_lg_tcache_max < 0 || (ZU(1) << opt_lg_tcache_max) <
+	    SC_SMALL_MAXCLASS) {
+		tcache_maxclass = SC_SMALL_MAXCLASS;
+	} else {
+		tcache_maxclass = (ZU(1) << opt_lg_tcache_max);
+	}
+
+	if (malloc_mutex_init(&tcaches_mtx, "tcaches", WITNESS_RANK_TCACHES,
+	    malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+
+	nhbins = sz_size2index(tcache_maxclass) + 1;
+
+	/* Initialize tcache_bin_info. */
+	tcache_bin_info = (cache_bin_info_t *)base_alloc(tsdn, b0get(), nhbins
+	    * sizeof(cache_bin_info_t), CACHELINE);
+	if (tcache_bin_info == NULL) {
+		return true;
+	}
+	stack_nelms = 0;
+	unsigned i;
+	for (i = 0; i < SC_NBINS; i++) {
+		if ((bin_infos[i].nregs << 1) <= TCACHE_NSLOTS_SMALL_MIN) {
+			tcache_bin_info[i].ncached_max =
+			    TCACHE_NSLOTS_SMALL_MIN;
+		} else if ((bin_infos[i].nregs << 1) <=
+		    TCACHE_NSLOTS_SMALL_MAX) {
+			tcache_bin_info[i].ncached_max =
+			    (bin_infos[i].nregs << 1);
+		} else {
+			tcache_bin_info[i].ncached_max =
+			    TCACHE_NSLOTS_SMALL_MAX;
+		}
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+	for (; i < nhbins; i++) {
+		tcache_bin_info[i].ncached_max = TCACHE_NSLOTS_LARGE;
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+
+	return false;
+}
+
+void
+tcache_prefork(tsdn_t *tsdn) {
+	if (!config_prof && opt_tcache) {
+		malloc_mutex_prefork(tsdn, &tcaches_mtx);
+	}
+}
+
+void
+tcache_postfork_parent(tsdn_t *tsdn) {
+	if (!config_prof && opt_tcache) {
+		malloc_mutex_postfork_parent(tsdn, &tcaches_mtx);
+	}
+}
+
+void
+tcache_postfork_child(tsdn_t *tsdn) {
+	if (!config_prof && opt_tcache) {
+		malloc_mutex_postfork_child(tsdn, &tcaches_mtx);
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/src/test_hooks.c b/memkind-1.10.0/jemalloc/src/test_hooks.c
new file mode 100644
index 0000000..ace00d9
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/test_hooks.c
@@ -0,0 +1,12 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+
+/*
+ * The hooks are a little bit screwy -- they're not genuinely exported in the
+ * sense that we want them available to end-users, but we do want them visible
+ * from outside the generated library, so that we can use them in test code.
+ */
+JEMALLOC_EXPORT
+void (*test_hooks_arena_new_hook)() = NULL;
+
+JEMALLOC_EXPORT
+void (*test_hooks_libc_hook)() = NULL;
diff --git a/memkind-1.10.0/jemalloc/src/ticker.c b/memkind-1.10.0/jemalloc/src/ticker.c
new file mode 100644
index 0000000..d7b8cd2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/ticker.c
@@ -0,0 +1,3 @@
+#define JEMALLOC_TICKER_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
diff --git a/memkind-1.10.0/jemalloc/src/tsd.c b/memkind-1.10.0/jemalloc/src/tsd.c
new file mode 100644
index 0000000..a31f6b9
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/tsd.c
@@ -0,0 +1,534 @@
+#define JEMALLOC_TSD_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+
+/******************************************************************************/
+/* Data. */
+
+static unsigned ncleanups;
+static malloc_tsd_cleanup_t cleanups[MALLOC_TSD_CLEANUPS_MAX];
+
+/* TSD_INITIALIZER triggers "-Wmissing-field-initializer" */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_MISSING_STRUCT_FIELD_INITIALIZERS
+
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls = TSD_INITIALIZER;
+JEMALLOC_TSD_TYPE_ATTR(bool) JEMALLOC_TLS_MODEL tsd_initialized = false;
+bool tsd_booted = false;
+#elif (defined(JEMALLOC_TLS))
+JEMALLOC_TSD_TYPE_ATTR(tsd_t) tsd_tls = TSD_INITIALIZER;
+pthread_key_t tsd_tsd;
+bool tsd_booted = false;
+#elif (defined(_WIN32))
+DWORD tsd_tsd;
+tsd_wrapper_t tsd_boot_wrapper = {false, TSD_INITIALIZER};
+bool tsd_booted = false;
+#else
+
+/*
+ * This contains a mutex, but it's pretty convenient to allow the mutex code to
+ * have a dependency on tsd.  So we define the struct here, and only refer to it
+ * by pointer in the header.
+ */
+struct tsd_init_head_s {
+	ql_head(tsd_init_block_t) blocks;
+	malloc_mutex_t lock;
+};
+
+pthread_key_t tsd_tsd;
+tsd_init_head_t	tsd_init_head = {
+	ql_head_initializer(blocks),
+	MALLOC_MUTEX_INITIALIZER
+};
+
+tsd_wrapper_t tsd_boot_wrapper = {
+	false,
+	TSD_INITIALIZER
+};
+bool tsd_booted = false;
+#endif
+
+JEMALLOC_DIAGNOSTIC_POP
+
+/******************************************************************************/
+
+/* A list of all the tsds in the nominal state. */
+typedef ql_head(tsd_t) tsd_list_t;
+static tsd_list_t tsd_nominal_tsds = ql_head_initializer(tsd_nominal_tsds);
+static malloc_mutex_t tsd_nominal_tsds_lock;
+
+/* How many slow-path-enabling features are turned on. */
+static atomic_u32_t tsd_global_slow_count = ATOMIC_INIT(0);
+
+static bool
+tsd_in_nominal_list(tsd_t *tsd) {
+	tsd_t *tsd_list;
+	bool found = false;
+	/*
+	 * We don't know that tsd is nominal; it might not be safe to get data
+	 * out of it here.
+	 */
+	malloc_mutex_lock(TSDN_NULL, &tsd_nominal_tsds_lock);
+	ql_foreach(tsd_list, &tsd_nominal_tsds, TSD_MANGLE(tcache).tsd_link) {
+		if (tsd == tsd_list) {
+			found = true;
+			break;
+		}
+	}
+	malloc_mutex_unlock(TSDN_NULL, &tsd_nominal_tsds_lock);
+	return found;
+}
+
+static void
+tsd_add_nominal(tsd_t *tsd) {
+	assert(!tsd_in_nominal_list(tsd));
+	assert(tsd_state_get(tsd) <= tsd_state_nominal_max);
+	ql_elm_new(tsd, TSD_MANGLE(tcache).tsd_link);
+	malloc_mutex_lock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+	ql_tail_insert(&tsd_nominal_tsds, tsd, TSD_MANGLE(tcache).tsd_link);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+static void
+tsd_remove_nominal(tsd_t *tsd) {
+	assert(tsd_in_nominal_list(tsd));
+	assert(tsd_state_get(tsd) <= tsd_state_nominal_max);
+	malloc_mutex_lock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+	ql_remove(&tsd_nominal_tsds, tsd, TSD_MANGLE(tcache).tsd_link);
+	malloc_mutex_unlock(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+static void
+tsd_force_recompute(tsdn_t *tsdn) {
+	/*
+	 * The stores to tsd->state here need to synchronize with the exchange
+	 * in tsd_slow_update.
+	 */
+	atomic_fence(ATOMIC_RELEASE);
+	malloc_mutex_lock(tsdn, &tsd_nominal_tsds_lock);
+	tsd_t *remote_tsd;
+	ql_foreach(remote_tsd, &tsd_nominal_tsds, TSD_MANGLE(tcache).tsd_link) {
+		assert(tsd_atomic_load(&remote_tsd->state, ATOMIC_RELAXED)
+		    <= tsd_state_nominal_max);
+		tsd_atomic_store(&remote_tsd->state, tsd_state_nominal_recompute,
+		    ATOMIC_RELAXED);
+	}
+	malloc_mutex_unlock(tsdn, &tsd_nominal_tsds_lock);
+}
+
+void
+tsd_global_slow_inc(tsdn_t *tsdn) {
+	atomic_fetch_add_u32(&tsd_global_slow_count, 1, ATOMIC_RELAXED);
+	/*
+	 * We unconditionally force a recompute, even if the global slow count
+	 * was already positive.  If we didn't, then it would be possible for us
+	 * to return to the user, have the user synchronize externally with some
+	 * other thread, and then have that other thread not have picked up the
+	 * update yet (since the original incrementing thread might still be
+	 * making its way through the tsd list).
+	 */
+	tsd_force_recompute(tsdn);
+}
+
+void tsd_global_slow_dec(tsdn_t *tsdn) {
+	atomic_fetch_sub_u32(&tsd_global_slow_count, 1, ATOMIC_RELAXED);
+	/* See the note in ..._inc(). */
+	tsd_force_recompute(tsdn);
+}
+
+static bool
+tsd_local_slow(tsd_t *tsd) {
+	return !tsd_tcache_enabled_get(tsd)
+	    || tsd_reentrancy_level_get(tsd) > 0;
+}
+
+bool
+tsd_global_slow() {
+	return atomic_load_u32(&tsd_global_slow_count, ATOMIC_RELAXED) > 0;
+}
+
+/******************************************************************************/
+
+static uint8_t
+tsd_state_compute(tsd_t *tsd) {
+	if (!tsd_nominal(tsd)) {
+		return tsd_state_get(tsd);
+	}
+	/* We're in *a* nominal state; but which one? */
+	if (malloc_slow || tsd_local_slow(tsd) || tsd_global_slow()) {
+		return tsd_state_nominal_slow;
+	} else {
+		return tsd_state_nominal;
+	}
+}
+
+void
+tsd_slow_update(tsd_t *tsd) {
+	uint8_t old_state;
+	do {
+		uint8_t new_state = tsd_state_compute(tsd);
+		old_state = tsd_atomic_exchange(&tsd->state, new_state,
+		    ATOMIC_ACQUIRE);
+	} while (old_state == tsd_state_nominal_recompute);
+}
+
+void
+tsd_state_set(tsd_t *tsd, uint8_t new_state) {
+	/* Only the tsd module can change the state *to* recompute. */
+	assert(new_state != tsd_state_nominal_recompute);
+	uint8_t old_state = tsd_atomic_load(&tsd->state, ATOMIC_RELAXED);
+	if (old_state > tsd_state_nominal_max) {
+		/*
+		 * Not currently in the nominal list, but it might need to be
+		 * inserted there.
+		 */
+		assert(!tsd_in_nominal_list(tsd));
+		tsd_atomic_store(&tsd->state, new_state, ATOMIC_RELAXED);
+		if (new_state <= tsd_state_nominal_max) {
+			tsd_add_nominal(tsd);
+		}
+	} else {
+		/*
+		 * We're currently nominal.  If the new state is non-nominal,
+		 * great; we take ourselves off the list and just enter the new
+		 * state.
+		 */
+		assert(tsd_in_nominal_list(tsd));
+		if (new_state > tsd_state_nominal_max) {
+			tsd_remove_nominal(tsd);
+			tsd_atomic_store(&tsd->state, new_state,
+			    ATOMIC_RELAXED);
+		} else {
+			/*
+			 * This is the tricky case.  We're transitioning from
+			 * one nominal state to another.  The caller can't know
+			 * about any races that are occuring at the same time,
+			 * so we always have to recompute no matter what.
+			 */
+			tsd_slow_update(tsd);
+		}
+	}
+}
+
+static bool
+tsd_data_init(tsd_t *tsd) {
+	/*
+	 * We initialize the rtree context first (before the tcache), since the
+	 * tcache initialization depends on it.
+	 */
+	rtree_ctx_data_init(tsd_rtree_ctxp_get_unsafe(tsd));
+
+	/*
+	 * A nondeterministic seed based on the address of tsd reduces
+	 * the likelihood of lockstep non-uniform cache index
+	 * utilization among identical concurrent processes, but at the
+	 * cost of test repeatability.  For debug builds, instead use a
+	 * deterministic seed.
+	 */
+	*tsd_offset_statep_get(tsd) = config_debug ? 0 :
+	    (uint64_t)(uintptr_t)tsd;
+
+	return tsd_tcache_enabled_data_init(tsd);
+}
+
+static void
+assert_tsd_data_cleanup_done(tsd_t *tsd) {
+	assert(!tsd_nominal(tsd));
+	assert(!tsd_in_nominal_list(tsd));
+	assert(*tsd_arenap_get_unsafe(tsd) == NULL);
+	assert(*tsd_iarenap_get_unsafe(tsd) == NULL);
+	assert(*tsd_arenas_tdata_bypassp_get_unsafe(tsd) == true);
+	assert(*tsd_arenas_tdatap_get_unsafe(tsd) == NULL);
+	assert(*tsd_tcache_enabledp_get_unsafe(tsd) == false);
+	assert(*tsd_prof_tdatap_get_unsafe(tsd) == NULL);
+}
+
+static bool
+tsd_data_init_nocleanup(tsd_t *tsd) {
+	assert(tsd_state_get(tsd) == tsd_state_reincarnated ||
+	    tsd_state_get(tsd) == tsd_state_minimal_initialized);
+	/*
+	 * During reincarnation, there is no guarantee that the cleanup function
+	 * will be called (deallocation may happen after all tsd destructors).
+	 * We set up tsd in a way that no cleanup is needed.
+	 */
+	rtree_ctx_data_init(tsd_rtree_ctxp_get_unsafe(tsd));
+	*tsd_arenas_tdata_bypassp_get(tsd) = true;
+	*tsd_tcache_enabledp_get_unsafe(tsd) = false;
+	*tsd_reentrancy_levelp_get(tsd) = 1;
+	assert_tsd_data_cleanup_done(tsd);
+
+	return false;
+}
+
+tsd_t *
+tsd_fetch_slow(tsd_t *tsd, bool minimal) {
+	assert(!tsd_fast(tsd));
+
+	if (tsd_state_get(tsd) == tsd_state_nominal_slow) {
+		/*
+		 * On slow path but no work needed.  Note that we can't
+		 * necessarily *assert* that we're slow, because we might be
+		 * slow because of an asynchronous modification to global state,
+		 * which might be asynchronously modified *back*.
+		 */
+	} else if (tsd_state_get(tsd) == tsd_state_nominal_recompute) {
+		tsd_slow_update(tsd);
+	} else if (tsd_state_get(tsd) == tsd_state_uninitialized) {
+		if (!minimal) {
+			if (tsd_booted) {
+				tsd_state_set(tsd, tsd_state_nominal);
+				tsd_slow_update(tsd);
+				/* Trigger cleanup handler registration. */
+				tsd_set(tsd);
+				tsd_data_init(tsd);
+			}
+		} else {
+			tsd_state_set(tsd, tsd_state_minimal_initialized);
+			tsd_set(tsd);
+			tsd_data_init_nocleanup(tsd);
+		}
+	} else if (tsd_state_get(tsd) == tsd_state_minimal_initialized) {
+		if (!minimal) {
+			/* Switch to fully initialized. */
+			tsd_state_set(tsd, tsd_state_nominal);
+			assert(*tsd_reentrancy_levelp_get(tsd) >= 1);
+			(*tsd_reentrancy_levelp_get(tsd))--;
+			tsd_slow_update(tsd);
+			tsd_data_init(tsd);
+		} else {
+			assert_tsd_data_cleanup_done(tsd);
+		}
+	} else if (tsd_state_get(tsd) == tsd_state_purgatory) {
+		tsd_state_set(tsd, tsd_state_reincarnated);
+		tsd_set(tsd);
+		tsd_data_init_nocleanup(tsd);
+	} else {
+		assert(tsd_state_get(tsd) == tsd_state_reincarnated);
+	}
+
+	return tsd;
+}
+
+void *
+malloc_tsd_malloc(size_t size) {
+	return a0malloc(CACHELINE_CEILING(size));
+}
+
+void
+malloc_tsd_dalloc(void *wrapper) {
+	a0dalloc(wrapper);
+}
+
+#if defined(JEMALLOC_MALLOC_THREAD_CLEANUP) || defined(_WIN32)
+#ifndef _WIN32
+JEMALLOC_EXPORT
+#endif
+void
+_malloc_thread_cleanup(void) {
+	bool pending[MALLOC_TSD_CLEANUPS_MAX], again;
+	unsigned i;
+
+	for (i = 0; i < ncleanups; i++) {
+		pending[i] = true;
+	}
+
+	do {
+		again = false;
+		for (i = 0; i < ncleanups; i++) {
+			if (pending[i]) {
+				pending[i] = cleanups[i]();
+				if (pending[i]) {
+					again = true;
+				}
+			}
+		}
+	} while (again);
+}
+#endif
+
+void
+malloc_tsd_cleanup_register(bool (*f)(void)) {
+	assert(ncleanups < MALLOC_TSD_CLEANUPS_MAX);
+	cleanups[ncleanups] = f;
+	ncleanups++;
+}
+
+static void
+tsd_do_data_cleanup(tsd_t *tsd) {
+	prof_tdata_cleanup(tsd);
+	iarena_cleanup(tsd);
+	arena_cleanup(tsd);
+	arenas_tdata_cleanup(tsd);
+	tcache_cleanup(tsd);
+	witnesses_cleanup(tsd_witness_tsdp_get_unsafe(tsd));
+}
+
+void
+tsd_cleanup(void *arg) {
+	tsd_t *tsd = (tsd_t *)arg;
+
+	switch (tsd_state_get(tsd)) {
+	case tsd_state_uninitialized:
+		/* Do nothing. */
+		break;
+	case tsd_state_minimal_initialized:
+		/* This implies the thread only did free() in its life time. */
+		/* Fall through. */
+	case tsd_state_reincarnated:
+		/*
+		 * Reincarnated means another destructor deallocated memory
+		 * after the destructor was called.  Cleanup isn't required but
+		 * is still called for testing and completeness.
+		 */
+		assert_tsd_data_cleanup_done(tsd);
+		/* Fall through. */
+	case tsd_state_nominal:
+	case tsd_state_nominal_slow:
+		tsd_do_data_cleanup(tsd);
+		tsd_state_set(tsd, tsd_state_purgatory);
+		tsd_set(tsd);
+		break;
+	case tsd_state_purgatory:
+		/*
+		 * The previous time this destructor was called, we set the
+		 * state to tsd_state_purgatory so that other destructors
+		 * wouldn't cause re-creation of the tsd.  This time, do
+		 * nothing, and do not request another callback.
+		 */
+		break;
+	default:
+		not_reached();
+	}
+#ifdef JEMALLOC_JET
+	test_callback_t test_callback = *tsd_test_callbackp_get_unsafe(tsd);
+	int *data = tsd_test_datap_get_unsafe(tsd);
+	if (test_callback != NULL) {
+		test_callback(data);
+	}
+#endif
+}
+
+tsd_t *
+malloc_tsd_boot0(void) {
+	tsd_t *tsd;
+
+	ncleanups = 0;
+	if (malloc_mutex_init(&tsd_nominal_tsds_lock, "tsd_nominal_tsds_lock",
+	    WITNESS_RANK_OMIT, malloc_mutex_rank_exclusive)) {
+		return NULL;
+	}
+	if (tsd_boot0()) {
+		return NULL;
+	}
+	tsd = tsd_fetch();
+	*tsd_arenas_tdata_bypassp_get(tsd) = true;
+	return tsd;
+}
+
+void
+malloc_tsd_boot1(void) {
+	tsd_boot1();
+	tsd_t *tsd = tsd_fetch();
+	/* malloc_slow has been set properly.  Update tsd_slow. */
+	tsd_slow_update(tsd);
+	*tsd_arenas_tdata_bypassp_get(tsd) = false;
+}
+
+#ifdef _WIN32
+static BOOL WINAPI
+_tls_callback(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved) {
+	switch (fdwReason) {
+#ifdef JEMALLOC_LAZY_LOCK
+	case DLL_THREAD_ATTACH:
+		isthreaded = true;
+		break;
+#endif
+	case DLL_THREAD_DETACH:
+		_malloc_thread_cleanup();
+		break;
+	default:
+		break;
+	}
+	return true;
+}
+
+/*
+ * We need to be able to say "read" here (in the "pragma section"), but have
+ * hooked "read". We won't read for the rest of the file, so we can get away
+ * with unhooking.
+ */
+#ifdef read
+#  undef read
+#endif
+
+#ifdef _MSC_VER
+#  ifdef _M_IX86
+#    pragma comment(linker, "/INCLUDE:__tls_used")
+#    pragma comment(linker, "/INCLUDE:_tls_callback")
+#  else
+#    pragma comment(linker, "/INCLUDE:_tls_used")
+#    pragma comment(linker, "/INCLUDE:" STRINGIFY(tls_callback) )
+#  endif
+#  pragma section(".CRT$XLY",long,read)
+#endif
+JEMALLOC_SECTION(".CRT$XLY") JEMALLOC_ATTR(used)
+BOOL	(WINAPI *const tls_callback)(HINSTANCE hinstDLL,
+    DWORD fdwReason, LPVOID lpvReserved) = _tls_callback;
+#endif
+
+#if (!defined(JEMALLOC_MALLOC_THREAD_CLEANUP) && !defined(JEMALLOC_TLS) && \
+    !defined(_WIN32))
+void *
+tsd_init_check_recursion(tsd_init_head_t *head, tsd_init_block_t *block) {
+	pthread_t self = pthread_self();
+	tsd_init_block_t *iter;
+
+	/* Check whether this thread has already inserted into the list. */
+	malloc_mutex_lock(TSDN_NULL, &head->lock);
+	ql_foreach(iter, &head->blocks, link) {
+		if (iter->thread == self) {
+			malloc_mutex_unlock(TSDN_NULL, &head->lock);
+			return iter->data;
+		}
+	}
+	/* Insert block into list. */
+	ql_elm_new(block, link);
+	block->thread = self;
+	ql_tail_insert(&head->blocks, block, link);
+	malloc_mutex_unlock(TSDN_NULL, &head->lock);
+	return NULL;
+}
+
+void
+tsd_init_finish(tsd_init_head_t *head, tsd_init_block_t *block) {
+	malloc_mutex_lock(TSDN_NULL, &head->lock);
+	ql_remove(&head->blocks, block, link);
+	malloc_mutex_unlock(TSDN_NULL, &head->lock);
+}
+#endif
+
+void
+tsd_prefork(tsd_t *tsd) {
+	malloc_mutex_prefork(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+void
+tsd_postfork_parent(tsd_t *tsd) {
+	malloc_mutex_postfork_parent(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+}
+
+void
+tsd_postfork_child(tsd_t *tsd) {
+	malloc_mutex_postfork_child(tsd_tsdn(tsd), &tsd_nominal_tsds_lock);
+	ql_new(&tsd_nominal_tsds);
+
+	if (tsd_state_get(tsd) <= tsd_state_nominal_max) {
+		tsd_add_nominal(tsd);
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/src/witness.c b/memkind-1.10.0/jemalloc/src/witness.c
new file mode 100644
index 0000000..f42b72a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/witness.c
@@ -0,0 +1,100 @@
+#define JEMALLOC_WITNESS_C_
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+#include "jemalloc/internal/malloc_io.h"
+
+void
+witness_init(witness_t *witness, const char *name, witness_rank_t rank,
+    witness_comp_t *comp, void *opaque) {
+	witness->name = name;
+	witness->rank = rank;
+	witness->comp = comp;
+	witness->opaque = opaque;
+}
+
+static void
+witness_lock_error_impl(const witness_list_t *witnesses,
+    const witness_t *witness) {
+	witness_t *w;
+
+	malloc_printf("<jemalloc>: Lock rank order reversal:");
+	ql_foreach(w, witnesses, link) {
+		malloc_printf(" %s(%u)", w->name, w->rank);
+	}
+	malloc_printf(" %s(%u)\n", witness->name, witness->rank);
+	abort();
+}
+witness_lock_error_t *JET_MUTABLE witness_lock_error = witness_lock_error_impl;
+
+static void
+witness_owner_error_impl(const witness_t *witness) {
+	malloc_printf("<jemalloc>: Should own %s(%u)\n", witness->name,
+	    witness->rank);
+	abort();
+}
+witness_owner_error_t *JET_MUTABLE witness_owner_error =
+    witness_owner_error_impl;
+
+static void
+witness_not_owner_error_impl(const witness_t *witness) {
+	malloc_printf("<jemalloc>: Should not own %s(%u)\n", witness->name,
+	    witness->rank);
+	abort();
+}
+witness_not_owner_error_t *JET_MUTABLE witness_not_owner_error =
+    witness_not_owner_error_impl;
+
+static void
+witness_depth_error_impl(const witness_list_t *witnesses,
+    witness_rank_t rank_inclusive, unsigned depth) {
+	witness_t *w;
+
+	malloc_printf("<jemalloc>: Should own %u lock%s of rank >= %u:", depth,
+	    (depth != 1) ?  "s" : "", rank_inclusive);
+	ql_foreach(w, witnesses, link) {
+		malloc_printf(" %s(%u)", w->name, w->rank);
+	}
+	malloc_printf("\n");
+	abort();
+}
+witness_depth_error_t *JET_MUTABLE witness_depth_error =
+    witness_depth_error_impl;
+
+void
+witnesses_cleanup(witness_tsd_t *witness_tsd) {
+	witness_assert_lockless(witness_tsd_tsdn(witness_tsd));
+
+	/* Do nothing. */
+}
+
+void
+witness_prefork(witness_tsd_t *witness_tsd) {
+	if (!config_debug) {
+		return;
+	}
+	witness_tsd->forking = true;
+}
+
+void
+witness_postfork_parent(witness_tsd_t *witness_tsd) {
+	if (!config_debug) {
+		return;
+	}
+	witness_tsd->forking = false;
+}
+
+void
+witness_postfork_child(witness_tsd_t *witness_tsd) {
+	if (!config_debug) {
+		return;
+	}
+#ifndef JEMALLOC_MUTEX_INIT_CB
+	witness_list_t *witnesses;
+
+	witnesses = &witness_tsd->witnesses;
+	ql_new(witnesses);
+#endif
+	witness_tsd->forking = false;
+}
diff --git a/memkind-1.10.0/jemalloc/src/zone.c b/memkind-1.10.0/jemalloc/src/zone.c
new file mode 100644
index 0000000..23dfdd0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/src/zone.c
@@ -0,0 +1,469 @@
+#include "jemalloc/internal/jemalloc_preamble.h"
+#include "jemalloc/internal/jemalloc_internal_includes.h"
+
+#include "jemalloc/internal/assert.h"
+
+#ifndef JEMALLOC_ZONE
+#  error "This source file is for zones on Darwin (OS X)."
+#endif
+
+/* Definitions of the following structs in malloc/malloc.h might be too old
+ * for the built binary to run on newer versions of OSX. So use the newest
+ * possible version of those structs.
+ */
+typedef struct _malloc_zone_t {
+	void *reserved1;
+	void *reserved2;
+	size_t (*size)(struct _malloc_zone_t *, const void *);
+	void *(*malloc)(struct _malloc_zone_t *, size_t);
+	void *(*calloc)(struct _malloc_zone_t *, size_t, size_t);
+	void *(*valloc)(struct _malloc_zone_t *, size_t);
+	void (*free)(struct _malloc_zone_t *, void *);
+	void *(*realloc)(struct _malloc_zone_t *, void *, size_t);
+	void (*destroy)(struct _malloc_zone_t *);
+	const char *zone_name;
+	unsigned (*batch_malloc)(struct _malloc_zone_t *, size_t, void **, unsigned);
+	void (*batch_free)(struct _malloc_zone_t *, void **, unsigned);
+	struct malloc_introspection_t *introspect;
+	unsigned version;
+	void *(*memalign)(struct _malloc_zone_t *, size_t, size_t);
+	void (*free_definite_size)(struct _malloc_zone_t *, void *, size_t);
+	size_t (*pressure_relief)(struct _malloc_zone_t *, size_t);
+} malloc_zone_t;
+
+typedef struct {
+	vm_address_t address;
+	vm_size_t size;
+} vm_range_t;
+
+typedef struct malloc_statistics_t {
+	unsigned blocks_in_use;
+	size_t size_in_use;
+	size_t max_size_in_use;
+	size_t size_allocated;
+} malloc_statistics_t;
+
+typedef kern_return_t memory_reader_t(task_t, vm_address_t, vm_size_t, void **);
+
+typedef void vm_range_recorder_t(task_t, void *, unsigned type, vm_range_t *, unsigned);
+
+typedef struct malloc_introspection_t {
+	kern_return_t (*enumerator)(task_t, void *, unsigned, vm_address_t, memory_reader_t, vm_range_recorder_t);
+	size_t (*good_size)(malloc_zone_t *, size_t);
+	boolean_t (*check)(malloc_zone_t *);
+	void (*print)(malloc_zone_t *, boolean_t);
+	void (*log)(malloc_zone_t *, void *);
+	void (*force_lock)(malloc_zone_t *);
+	void (*force_unlock)(malloc_zone_t *);
+	void (*statistics)(malloc_zone_t *, malloc_statistics_t *);
+	boolean_t (*zone_locked)(malloc_zone_t *);
+	boolean_t (*enable_discharge_checking)(malloc_zone_t *);
+	boolean_t (*disable_discharge_checking)(malloc_zone_t *);
+	void (*discharge)(malloc_zone_t *, void *);
+#ifdef __BLOCKS__
+	void (*enumerate_discharged_pointers)(malloc_zone_t *, void (^)(void *, void *));
+#else
+	void *enumerate_unavailable_without_blocks;
+#endif
+	void (*reinit_lock)(malloc_zone_t *);
+} malloc_introspection_t;
+
+extern kern_return_t malloc_get_all_zones(task_t, memory_reader_t, vm_address_t **, unsigned *);
+
+extern malloc_zone_t *malloc_default_zone(void);
+
+extern void malloc_zone_register(malloc_zone_t *zone);
+
+extern void malloc_zone_unregister(malloc_zone_t *zone);
+
+/*
+ * The malloc_default_purgeable_zone() function is only available on >= 10.6.
+ * We need to check whether it is present at runtime, thus the weak_import.
+ */
+extern malloc_zone_t *malloc_default_purgeable_zone(void)
+JEMALLOC_ATTR(weak_import);
+
+/******************************************************************************/
+/* Data. */
+
+static malloc_zone_t *default_zone, *purgeable_zone;
+static malloc_zone_t jemalloc_zone;
+static struct malloc_introspection_t jemalloc_zone_introspect;
+static pid_t zone_force_lock_pid = -1;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static size_t	zone_size(malloc_zone_t *zone, const void *ptr);
+static void	*zone_malloc(malloc_zone_t *zone, size_t size);
+static void	*zone_calloc(malloc_zone_t *zone, size_t num, size_t size);
+static void	*zone_valloc(malloc_zone_t *zone, size_t size);
+static void	zone_free(malloc_zone_t *zone, void *ptr);
+static void	*zone_realloc(malloc_zone_t *zone, void *ptr, size_t size);
+static void	*zone_memalign(malloc_zone_t *zone, size_t alignment,
+    size_t size);
+static void	zone_free_definite_size(malloc_zone_t *zone, void *ptr,
+    size_t size);
+static void	zone_destroy(malloc_zone_t *zone);
+static unsigned	zone_batch_malloc(struct _malloc_zone_t *zone, size_t size,
+    void **results, unsigned num_requested);
+static void	zone_batch_free(struct _malloc_zone_t *zone,
+    void **to_be_freed, unsigned num_to_be_freed);
+static size_t	zone_pressure_relief(struct _malloc_zone_t *zone, size_t goal);
+static size_t	zone_good_size(malloc_zone_t *zone, size_t size);
+static kern_return_t	zone_enumerator(task_t task, void *data, unsigned type_mask,
+    vm_address_t zone_address, memory_reader_t reader,
+    vm_range_recorder_t recorder);
+static boolean_t	zone_check(malloc_zone_t *zone);
+static void	zone_print(malloc_zone_t *zone, boolean_t verbose);
+static void	zone_log(malloc_zone_t *zone, void *address);
+static void	zone_force_lock(malloc_zone_t *zone);
+static void	zone_force_unlock(malloc_zone_t *zone);
+static void	zone_statistics(malloc_zone_t *zone,
+    malloc_statistics_t *stats);
+static boolean_t	zone_locked(malloc_zone_t *zone);
+static void	zone_reinit_lock(malloc_zone_t *zone);
+
+/******************************************************************************/
+/*
+ * Functions.
+ */
+
+static size_t
+zone_size(malloc_zone_t *zone, const void *ptr) {
+	/*
+	 * There appear to be places within Darwin (such as setenv(3)) that
+	 * cause calls to this function with pointers that *no* zone owns.  If
+	 * we knew that all pointers were owned by *some* zone, we could split
+	 * our zone into two parts, and use one as the default allocator and
+	 * the other as the default deallocator/reallocator.  Since that will
+	 * not work in practice, we must check all pointers to assure that they
+	 * reside within a mapped extent before determining size.
+	 */
+	return ivsalloc(tsdn_fetch(), ptr);
+}
+
+static void *
+zone_malloc(malloc_zone_t *zone, size_t size) {
+	return je_malloc(size);
+}
+
+static void *
+zone_calloc(malloc_zone_t *zone, size_t num, size_t size) {
+	return je_calloc(num, size);
+}
+
+static void *
+zone_valloc(malloc_zone_t *zone, size_t size) {
+	void *ret = NULL; /* Assignment avoids useless compiler warning. */
+
+	je_posix_memalign(&ret, PAGE, size);
+
+	return ret;
+}
+
+static void
+zone_free(malloc_zone_t *zone, void *ptr) {
+	if (ivsalloc(tsdn_fetch(), ptr) != 0) {
+		je_free(ptr);
+		return;
+	}
+
+	free(ptr);
+}
+
+static void *
+zone_realloc(malloc_zone_t *zone, void *ptr, size_t size) {
+	if (ivsalloc(tsdn_fetch(), ptr) != 0) {
+		return je_realloc(ptr, size);
+	}
+
+	return realloc(ptr, size);
+}
+
+static void *
+zone_memalign(malloc_zone_t *zone, size_t alignment, size_t size) {
+	void *ret = NULL; /* Assignment avoids useless compiler warning. */
+
+	je_posix_memalign(&ret, alignment, size);
+
+	return ret;
+}
+
+static void
+zone_free_definite_size(malloc_zone_t *zone, void *ptr, size_t size) {
+	size_t alloc_size;
+
+	alloc_size = ivsalloc(tsdn_fetch(), ptr);
+	if (alloc_size != 0) {
+		assert(alloc_size == size);
+		je_free(ptr);
+		return;
+	}
+
+	free(ptr);
+}
+
+static void
+zone_destroy(malloc_zone_t *zone) {
+	/* This function should never be called. */
+	not_reached();
+}
+
+static unsigned
+zone_batch_malloc(struct _malloc_zone_t *zone, size_t size, void **results,
+    unsigned num_requested) {
+	unsigned i;
+
+	for (i = 0; i < num_requested; i++) {
+		results[i] = je_malloc(size);
+		if (!results[i])
+			break;
+	}
+
+	return i;
+}
+
+static void
+zone_batch_free(struct _malloc_zone_t *zone, void **to_be_freed,
+    unsigned num_to_be_freed) {
+	unsigned i;
+
+	for (i = 0; i < num_to_be_freed; i++) {
+		zone_free(zone, to_be_freed[i]);
+		to_be_freed[i] = NULL;
+	}
+}
+
+static size_t
+zone_pressure_relief(struct _malloc_zone_t *zone, size_t goal) {
+	return 0;
+}
+
+static size_t
+zone_good_size(malloc_zone_t *zone, size_t size) {
+	if (size == 0) {
+		size = 1;
+	}
+	return sz_s2u(size);
+}
+
+static kern_return_t
+zone_enumerator(task_t task, void *data, unsigned type_mask,
+    vm_address_t zone_address, memory_reader_t reader,
+    vm_range_recorder_t recorder) {
+	return KERN_SUCCESS;
+}
+
+static boolean_t
+zone_check(malloc_zone_t *zone) {
+	return true;
+}
+
+static void
+zone_print(malloc_zone_t *zone, boolean_t verbose) {
+}
+
+static void
+zone_log(malloc_zone_t *zone, void *address) {
+}
+
+static void
+zone_force_lock(malloc_zone_t *zone) {
+	if (isthreaded) {
+		/*
+		 * See the note in zone_force_unlock, below, to see why we need
+		 * this.
+		 */
+		assert(zone_force_lock_pid == -1);
+		zone_force_lock_pid = getpid();
+		jemalloc_prefork();
+	}
+}
+
+static void
+zone_force_unlock(malloc_zone_t *zone) {
+	/*
+	 * zone_force_lock and zone_force_unlock are the entry points to the
+	 * forking machinery on OS X.  The tricky thing is, the child is not
+	 * allowed to unlock mutexes locked in the parent, even if owned by the
+	 * forking thread (and the mutex type we use in OS X will fail an assert
+	 * if we try).  In the child, we can get away with reinitializing all
+	 * the mutexes, which has the effect of unlocking them.  In the parent,
+	 * doing this would mean we wouldn't wake any waiters blocked on the
+	 * mutexes we unlock.  So, we record the pid of the current thread in
+	 * zone_force_lock, and use that to detect if we're in the parent or
+	 * child here, to decide which unlock logic we need.
+	 */
+	if (isthreaded) {
+		assert(zone_force_lock_pid != -1);
+		if (getpid() == zone_force_lock_pid) {
+			jemalloc_postfork_parent();
+		} else {
+			jemalloc_postfork_child();
+		}
+		zone_force_lock_pid = -1;
+	}
+}
+
+static void
+zone_statistics(malloc_zone_t *zone, malloc_statistics_t *stats) {
+	/* We make no effort to actually fill the values */
+	stats->blocks_in_use = 0;
+	stats->size_in_use = 0;
+	stats->max_size_in_use = 0;
+	stats->size_allocated = 0;
+}
+
+static boolean_t
+zone_locked(malloc_zone_t *zone) {
+	/* Pretend no lock is being held */
+	return false;
+}
+
+static void
+zone_reinit_lock(malloc_zone_t *zone) {
+	/* As of OSX 10.12, this function is only used when force_unlock would
+	 * be used if the zone version were < 9. So just use force_unlock. */
+	zone_force_unlock(zone);
+}
+
+static void
+zone_init(void) {
+	jemalloc_zone.size = zone_size;
+	jemalloc_zone.malloc = zone_malloc;
+	jemalloc_zone.calloc = zone_calloc;
+	jemalloc_zone.valloc = zone_valloc;
+	jemalloc_zone.free = zone_free;
+	jemalloc_zone.realloc = zone_realloc;
+	jemalloc_zone.destroy = zone_destroy;
+	jemalloc_zone.zone_name = "jemalloc_zone";
+	jemalloc_zone.batch_malloc = zone_batch_malloc;
+	jemalloc_zone.batch_free = zone_batch_free;
+	jemalloc_zone.introspect = &jemalloc_zone_introspect;
+	jemalloc_zone.version = 9;
+	jemalloc_zone.memalign = zone_memalign;
+	jemalloc_zone.free_definite_size = zone_free_definite_size;
+	jemalloc_zone.pressure_relief = zone_pressure_relief;
+
+	jemalloc_zone_introspect.enumerator = zone_enumerator;
+	jemalloc_zone_introspect.good_size = zone_good_size;
+	jemalloc_zone_introspect.check = zone_check;
+	jemalloc_zone_introspect.print = zone_print;
+	jemalloc_zone_introspect.log = zone_log;
+	jemalloc_zone_introspect.force_lock = zone_force_lock;
+	jemalloc_zone_introspect.force_unlock = zone_force_unlock;
+	jemalloc_zone_introspect.statistics = zone_statistics;
+	jemalloc_zone_introspect.zone_locked = zone_locked;
+	jemalloc_zone_introspect.enable_discharge_checking = NULL;
+	jemalloc_zone_introspect.disable_discharge_checking = NULL;
+	jemalloc_zone_introspect.discharge = NULL;
+#ifdef __BLOCKS__
+	jemalloc_zone_introspect.enumerate_discharged_pointers = NULL;
+#else
+	jemalloc_zone_introspect.enumerate_unavailable_without_blocks = NULL;
+#endif
+	jemalloc_zone_introspect.reinit_lock = zone_reinit_lock;
+}
+
+static malloc_zone_t *
+zone_default_get(void) {
+	malloc_zone_t **zones = NULL;
+	unsigned int num_zones = 0;
+
+	/*
+	 * On OSX 10.12, malloc_default_zone returns a special zone that is not
+	 * present in the list of registered zones. That zone uses a "lite zone"
+	 * if one is present (apparently enabled when malloc stack logging is
+	 * enabled), or the first registered zone otherwise. In practice this
+	 * means unless malloc stack logging is enabled, the first registered
+	 * zone is the default.  So get the list of zones to get the first one,
+	 * instead of relying on malloc_default_zone.
+	 */
+	if (KERN_SUCCESS != malloc_get_all_zones(0, NULL,
+	    (vm_address_t**)&zones, &num_zones)) {
+		/*
+		 * Reset the value in case the failure happened after it was
+		 * set.
+		 */
+		num_zones = 0;
+	}
+
+	if (num_zones) {
+		return zones[0];
+	}
+
+	return malloc_default_zone();
+}
+
+/* As written, this function can only promote jemalloc_zone. */
+static void
+zone_promote(void) {
+	malloc_zone_t *zone;
+
+	do {
+		/*
+		 * Unregister and reregister the default zone.  On OSX >= 10.6,
+		 * unregistering takes the last registered zone and places it
+		 * at the location of the specified zone.  Unregistering the
+		 * default zone thus makes the last registered one the default.
+		 * On OSX < 10.6, unregistering shifts all registered zones.
+		 * The first registered zone then becomes the default.
+		 */
+		malloc_zone_unregister(default_zone);
+		malloc_zone_register(default_zone);
+
+		/*
+		 * On OSX 10.6, having the default purgeable zone appear before
+		 * the default zone makes some things crash because it thinks it
+		 * owns the default zone allocated pointers.  We thus
+		 * unregister/re-register it in order to ensure it's always
+		 * after the default zone.  On OSX < 10.6, there is no purgeable
+		 * zone, so this does nothing.  On OSX >= 10.6, unregistering
+		 * replaces the purgeable zone with the last registered zone
+		 * above, i.e. the default zone.  Registering it again then puts
+		 * it at the end, obviously after the default zone.
+		 */
+		if (purgeable_zone != NULL) {
+			malloc_zone_unregister(purgeable_zone);
+			malloc_zone_register(purgeable_zone);
+		}
+
+		zone = zone_default_get();
+	} while (zone != &jemalloc_zone);
+}
+
+JEMALLOC_ATTR(constructor)
+void
+zone_register(void) {
+	/*
+	 * If something else replaced the system default zone allocator, don't
+	 * register jemalloc's.
+	 */
+	default_zone = zone_default_get();
+	if (!default_zone->zone_name || strcmp(default_zone->zone_name,
+	    "DefaultMallocZone") != 0) {
+		return;
+	}
+
+	/*
+	 * The default purgeable zone is created lazily by OSX's libc.  It uses
+	 * the default zone when it is created for "small" allocations
+	 * (< 15 KiB), but assumes the default zone is a scalable_zone.  This
+	 * obviously fails when the default zone is the jemalloc zone, so
+	 * malloc_default_purgeable_zone() is called beforehand so that the
+	 * default purgeable zone is created when the default zone is still
+	 * a scalable_zone.  As purgeable zones only exist on >= 10.6, we need
+	 * to check for the existence of malloc_default_purgeable_zone() at
+	 * run time.
+	 */
+	purgeable_zone = (malloc_default_purgeable_zone == NULL) ? NULL :
+	    malloc_default_purgeable_zone();
+
+	/* Register the custom zone.  At this point it won't be the default. */
+	zone_init();
+	malloc_zone_register(&jemalloc_zone);
+
+	/* Promote the custom zone to be default. */
+	zone_promote();
+}
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-alti.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-alti.h
new file mode 100644
index 0000000..a1885db
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-alti.h
@@ -0,0 +1,186 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/**
+ * @file SFMT-alti.h
+ *
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT)
+ * pseudorandom number generator
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * Copyright (C) 2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software.
+ * see LICENSE.txt
+ */
+
+#ifndef SFMT_ALTI_H
+#define SFMT_ALTI_H
+
+/**
+ * This function represents the recursion formula in AltiVec and BIG ENDIAN.
+ * @param a a 128-bit part of the interal state array
+ * @param b a 128-bit part of the interal state array
+ * @param c a 128-bit part of the interal state array
+ * @param d a 128-bit part of the interal state array
+ * @return output
+ */
+JEMALLOC_ALWAYS_INLINE
+vector unsigned int vec_recursion(vector unsigned int a,
+						vector unsigned int b,
+						vector unsigned int c,
+						vector unsigned int d) {
+
+    const vector unsigned int sl1 = ALTI_SL1;
+    const vector unsigned int sr1 = ALTI_SR1;
+#ifdef ONLY64
+    const vector unsigned int mask = ALTI_MSK64;
+    const vector unsigned char perm_sl = ALTI_SL2_PERM64;
+    const vector unsigned char perm_sr = ALTI_SR2_PERM64;
+#else
+    const vector unsigned int mask = ALTI_MSK;
+    const vector unsigned char perm_sl = ALTI_SL2_PERM;
+    const vector unsigned char perm_sr = ALTI_SR2_PERM;
+#endif
+    vector unsigned int v, w, x, y, z;
+    x = vec_perm(a, (vector unsigned int)perm_sl, perm_sl);
+    v = a;
+    y = vec_sr(b, sr1);
+    z = vec_perm(c, (vector unsigned int)perm_sr, perm_sr);
+    w = vec_sl(d, sl1);
+    z = vec_xor(z, w);
+    y = vec_and(y, mask);
+    v = vec_xor(v, x);
+    z = vec_xor(z, y);
+    z = vec_xor(z, v);
+    return z;
+}
+
+/**
+ * This function fills the internal state array with pseudorandom
+ * integers.
+ */
+static inline void gen_rand_all(sfmt_t *ctx) {
+    int i;
+    vector unsigned int r, r1, r2;
+
+    r1 = ctx->sfmt[N - 2].s;
+    r2 = ctx->sfmt[N - 1].s;
+    for (i = 0; i < N - POS1; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, ctx->sfmt[i + POS1].s, r1, r2);
+	ctx->sfmt[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, ctx->sfmt[i + POS1 - N].s, r1, r2);
+	ctx->sfmt[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+/**
+ * This function fills the user-specified array with pseudorandom
+ * integers.
+ *
+ * @param array an 128-bit array to be filled by pseudorandom numbers.
+ * @param size number of 128-bit pesudorandom numbers to be generated.
+ */
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size) {
+    int i, j;
+    vector unsigned int r, r1, r2;
+
+    r1 = ctx->sfmt[N - 2].s;
+    r2 = ctx->sfmt[N - 1].s;
+    for (i = 0; i < N - POS1; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, ctx->sfmt[i + POS1].s, r1, r2);
+	array[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = vec_recursion(ctx->sfmt[i].s, array[i + POS1 - N].s, r1, r2);
+	array[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    /* main loop */
+    for (; i < size - N; i++) {
+	r = vec_recursion(array[i - N].s, array[i + POS1 - N].s, r1, r2);
+	array[i].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+    for (j = 0; j < 2 * N - size; j++) {
+	ctx->sfmt[j].s = array[j + size - N].s;
+    }
+    for (; i < size; i++) {
+	r = vec_recursion(array[i - N].s, array[i + POS1 - N].s, r1, r2);
+	array[i].s = r;
+	ctx->sfmt[j++].s = r;
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+#ifndef ONLY64
+#if defined(__APPLE__)
+#define ALTI_SWAP (vector unsigned char) \
+	(4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11)
+#else
+#define ALTI_SWAP {4, 5, 6, 7, 0, 1, 2, 3, 12, 13, 14, 15, 8, 9, 10, 11}
+#endif
+/**
+ * This function swaps high and low 32-bit of 64-bit integers in user
+ * specified array.
+ *
+ * @param array an 128-bit array to be swaped.
+ * @param size size of 128-bit array.
+ */
+static inline void swap(w128_t *array, int size) {
+    int i;
+    const vector unsigned char perm = ALTI_SWAP;
+
+    for (i = 0; i < size; i++) {
+	array[i].s = vec_perm(array[i].s, (vector unsigned int)perm, perm);
+    }
+}
+#endif
+
+#endif
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params.h
new file mode 100644
index 0000000..ade6622
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params.h
@@ -0,0 +1,132 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS_H
+#define SFMT_PARAMS_H
+
+#if !defined(MEXP)
+#ifdef __GNUC__
+  #warning "MEXP is not defined. I assume MEXP is 19937."
+#endif
+  #define MEXP 19937
+#endif
+/*-----------------
+  BASIC DEFINITIONS
+  -----------------*/
+/** Mersenne Exponent. The period of the sequence 
+ *  is a multiple of 2^MEXP-1.
+ * #define MEXP 19937 */
+/** SFMT generator has an internal state array of 128-bit integers,
+ * and N is its size. */
+#define N (MEXP / 128 + 1)
+/** N32 is the size of internal state array when regarded as an array
+ * of 32-bit integers.*/
+#define N32 (N * 4)
+/** N64 is the size of internal state array when regarded as an array
+ * of 64-bit integers.*/
+#define N64 (N * 2)
+
+/*----------------------
+  the parameters of SFMT
+  following definitions are in paramsXXXX.h file.
+  ----------------------*/
+/** the pick up position of the array.
+#define POS1 122 
+*/
+
+/** the parameter of shift left as four 32-bit registers.
+#define SL1 18
+ */
+
+/** the parameter of shift left as one 128-bit register. 
+ * The 128-bit integer is shifted by (SL2 * 8) bits. 
+#define SL2 1 
+*/
+
+/** the parameter of shift right as four 32-bit registers.
+#define SR1 11
+*/
+
+/** the parameter of shift right as one 128-bit register. 
+ * The 128-bit integer is shifted by (SL2 * 8) bits. 
+#define SR2 1 
+*/
+
+/** A bitmask, used in the recursion.  These parameters are introduced
+ * to break symmetry of SIMD.
+#define MSK1 0xdfffffefU
+#define MSK2 0xddfecb7fU
+#define MSK3 0xbffaffffU
+#define MSK4 0xbffffff6U 
+*/
+
+/** These definitions are part of a 128-bit period certification vector.
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0xc98e126aU
+*/
+
+#if MEXP == 607
+  #include "test/SFMT-params607.h"
+#elif MEXP == 1279
+  #include "test/SFMT-params1279.h"
+#elif MEXP == 2281
+  #include "test/SFMT-params2281.h"
+#elif MEXP == 4253
+  #include "test/SFMT-params4253.h"
+#elif MEXP == 11213
+  #include "test/SFMT-params11213.h"
+#elif MEXP == 19937
+  #include "test/SFMT-params19937.h"
+#elif MEXP == 44497
+  #include "test/SFMT-params44497.h"
+#elif MEXP == 86243
+  #include "test/SFMT-params86243.h"
+#elif MEXP == 132049
+  #include "test/SFMT-params132049.h"
+#elif MEXP == 216091
+  #include "test/SFMT-params216091.h"
+#else
+#ifdef __GNUC__
+  #error "MEXP is not valid."
+  #undef MEXP
+#else
+  #undef MEXP
+#endif
+
+#endif
+
+#endif /* SFMT_PARAMS_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params11213.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params11213.h
new file mode 100644
index 0000000..2994bd2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params11213.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS11213_H
+#define SFMT_PARAMS11213_H
+
+#define POS1	68
+#define SL1	14
+#define SL2	3
+#define SR1	7
+#define SR2	3
+#define MSK1	0xeffff7fbU
+#define MSK2	0xffffffefU
+#define MSK3	0xdfdfbfffU
+#define MSK4	0x7fffdbfdU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0xe8148000U
+#define PARITY4	0xd0c7afa3U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12}
+    #define ALTI_SR2_PERM64	{13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-11213:68-14-3-7-3:effff7fb-ffffffef-dfdfbfff-7fffdbfd"
+
+#endif /* SFMT_PARAMS11213_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params1279.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params1279.h
new file mode 100644
index 0000000..d7959f9
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params1279.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS1279_H
+#define SFMT_PARAMS1279_H
+
+#define POS1	7
+#define SL1	14
+#define SL2	3
+#define SR1	5
+#define SR2	1
+#define MSK1	0xf7fefffdU
+#define MSK2	0x7fefcfffU
+#define MSK3	0xaff3ef3fU
+#define MSK4	0xb5ffff7fU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x20000000U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-1279:7-14-3-5-1:f7fefffd-7fefcfff-aff3ef3f-b5ffff7f"
+
+#endif /* SFMT_PARAMS1279_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params132049.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params132049.h
new file mode 100644
index 0000000..a1dcec3
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params132049.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS132049_H
+#define SFMT_PARAMS132049_H
+
+#define POS1	110
+#define SL1	19
+#define SL2	1
+#define SR1	21
+#define SR2	1
+#define MSK1	0xffffbb5fU
+#define MSK2	0xfb6ebf95U
+#define MSK3	0xfffefffaU
+#define MSK4	0xcff77fffU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0xcb520000U
+#define PARITY4	0xc7e91c7dU
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-132049:110-19-1-21-1:ffffbb5f-fb6ebf95-fffefffa-cff77fff"
+
+#endif /* SFMT_PARAMS132049_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params19937.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params19937.h
new file mode 100644
index 0000000..fb92b4c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params19937.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS19937_H
+#define SFMT_PARAMS19937_H
+
+#define POS1	122
+#define SL1	18
+#define SL2	1
+#define SR1	11
+#define SR2	1
+#define MSK1	0xdfffffefU
+#define MSK2	0xddfecb7fU
+#define MSK3	0xbffaffffU
+#define MSK4	0xbffffff6U
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x13c9e684U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-19937:122-18-1-11-1:dfffffef-ddfecb7f-bffaffff-bffffff6"
+
+#endif /* SFMT_PARAMS19937_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params216091.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params216091.h
new file mode 100644
index 0000000..125ce28
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params216091.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS216091_H
+#define SFMT_PARAMS216091_H
+
+#define POS1	627
+#define SL1	11
+#define SL2	3
+#define SR1	10
+#define SR2	1
+#define MSK1	0xbff7bff7U
+#define MSK2	0xbfffffffU
+#define MSK3	0xbffffa7fU
+#define MSK4	0xffddfbfbU
+#define PARITY1	0xf8000001U
+#define PARITY2	0x89e80709U
+#define PARITY3	0x3bd2b64bU
+#define PARITY4	0x0c64b1e4U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-216091:627-11-3-10-1:bff7bff7-bfffffff-bffffa7f-ffddfbfb"
+
+#endif /* SFMT_PARAMS216091_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params2281.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params2281.h
new file mode 100644
index 0000000..0ef85c4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params2281.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS2281_H
+#define SFMT_PARAMS2281_H
+
+#define POS1	12
+#define SL1	19
+#define SL2	1
+#define SR1	5
+#define SR2	1
+#define MSK1	0xbff7ffbfU
+#define MSK2	0xfdfffffeU
+#define MSK3	0xf7ffef7fU
+#define MSK4	0xf2f7cbbfU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x41dfa600U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-2281:12-19-1-5-1:bff7ffbf-fdfffffe-f7ffef7f-f2f7cbbf"
+
+#endif /* SFMT_PARAMS2281_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params4253.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params4253.h
new file mode 100644
index 0000000..9f07bc6
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params4253.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS4253_H
+#define SFMT_PARAMS4253_H
+
+#define POS1	17
+#define SL1	20
+#define SL2	1
+#define SR1	7
+#define SR2	1
+#define MSK1	0x9f7bffffU
+#define MSK2	0x9fffff5fU
+#define MSK3	0x3efffffbU
+#define MSK4	0xfffff7bbU
+#define PARITY1	0xa8000001U
+#define PARITY2	0xaf5390a3U
+#define PARITY3	0xb740b3f8U
+#define PARITY4	0x6c11486dU
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{1,2,3,23,5,6,7,0,9,10,11,4,13,14,15,8}
+    #define ALTI_SL2_PERM64	{1,2,3,4,5,6,7,31,9,10,11,12,13,14,15,0}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-4253:17-20-1-7-1:9f7bffff-9fffff5f-3efffffb-fffff7bb"
+
+#endif /* SFMT_PARAMS4253_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params44497.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params44497.h
new file mode 100644
index 0000000..85598fe
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params44497.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS44497_H
+#define SFMT_PARAMS44497_H
+
+#define POS1	330
+#define SL1	5
+#define SL2	3
+#define SR1	9
+#define SR2	3
+#define MSK1	0xeffffffbU
+#define MSK2	0xdfbebfffU
+#define MSK3	0xbfbf7befU
+#define MSK4	0x9ffd7bffU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0xa3ac4000U
+#define PARITY4	0xecc1327aU
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12}
+    #define ALTI_SR2_PERM64	{13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-44497:330-5-3-9-3:effffffb-dfbebfff-bfbf7bef-9ffd7bff"
+
+#endif /* SFMT_PARAMS44497_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params607.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params607.h
new file mode 100644
index 0000000..bc76485
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params607.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS607_H
+#define SFMT_PARAMS607_H
+
+#define POS1	2
+#define SL1	15
+#define SL2	3
+#define SR1	13
+#define SR2	3
+#define MSK1	0xfdff37ffU
+#define MSK2	0xef7f3f7dU
+#define MSK3	0xff777b7dU
+#define MSK4	0x7ff7fb2fU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0x5986f054U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{3,21,21,21,7,0,1,2,11,4,5,6,15,8,9,10}
+    #define ALTI_SL2_PERM64	{3,4,5,6,7,29,29,29,11,12,13,14,15,0,1,2}
+    #define ALTI_SR2_PERM	{5,6,7,0,9,10,11,4,13,14,15,8,19,19,19,12}
+    #define ALTI_SR2_PERM64	{13,14,15,0,1,2,3,4,19,19,19,8,9,10,11,12}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-607:2-15-3-13-3:fdff37ff-ef7f3f7d-ff777b7d-7ff7fb2f"
+
+#endif /* SFMT_PARAMS607_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-params86243.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params86243.h
new file mode 100644
index 0000000..5e4d783
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-params86243.h
@@ -0,0 +1,81 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#ifndef SFMT_PARAMS86243_H
+#define SFMT_PARAMS86243_H
+
+#define POS1	366
+#define SL1	6
+#define SL2	7
+#define SR1	19
+#define SR2	1
+#define MSK1	0xfdbffbffU
+#define MSK2	0xbff7ff3fU
+#define MSK3	0xfd77efffU
+#define MSK4	0xbf9ff3ffU
+#define PARITY1	0x00000001U
+#define PARITY2	0x00000000U
+#define PARITY3	0x00000000U
+#define PARITY4	0xe9528d85U
+
+
+/* PARAMETERS FOR ALTIVEC */
+#if defined(__APPLE__)	/* For OSX */
+    #define ALTI_SL1	(vector unsigned int)(SL1, SL1, SL1, SL1)
+    #define ALTI_SR1	(vector unsigned int)(SR1, SR1, SR1, SR1)
+    #define ALTI_MSK	(vector unsigned int)(MSK1, MSK2, MSK3, MSK4)
+    #define ALTI_MSK64 \
+	(vector unsigned int)(MSK2, MSK1, MSK4, MSK3)
+    #define ALTI_SL2_PERM \
+	(vector unsigned char)(25,25,25,25,3,25,25,25,7,0,1,2,11,4,5,6)
+    #define ALTI_SL2_PERM64 \
+	(vector unsigned char)(7,25,25,25,25,25,25,25,15,0,1,2,3,4,5,6)
+    #define ALTI_SR2_PERM \
+	(vector unsigned char)(7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14)
+    #define ALTI_SR2_PERM64 \
+	(vector unsigned char)(15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14)
+#else	/* For OTHER OSs(Linux?) */
+    #define ALTI_SL1	{SL1, SL1, SL1, SL1}
+    #define ALTI_SR1	{SR1, SR1, SR1, SR1}
+    #define ALTI_MSK	{MSK1, MSK2, MSK3, MSK4}
+    #define ALTI_MSK64	{MSK2, MSK1, MSK4, MSK3}
+    #define ALTI_SL2_PERM	{25,25,25,25,3,25,25,25,7,0,1,2,11,4,5,6}
+    #define ALTI_SL2_PERM64	{7,25,25,25,25,25,25,25,15,0,1,2,3,4,5,6}
+    #define ALTI_SR2_PERM	{7,0,1,2,11,4,5,6,15,8,9,10,17,12,13,14}
+    #define ALTI_SR2_PERM64	{15,0,1,2,3,4,5,6,17,8,9,10,11,12,13,14}
+#endif	/* For OSX */
+#define IDSTR	"SFMT-86243:366-6-7-19-1:fdbffbff-bff7ff3f-fd77efff-bf9ff3ff"
+
+#endif /* SFMT_PARAMS86243_H */
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT-sse2.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT-sse2.h
new file mode 100644
index 0000000..169ad55
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT-sse2.h
@@ -0,0 +1,157 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/**
+ * @file  SFMT-sse2.h
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT) for Intel SSE2
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * @note We assume LITTLE ENDIAN in this file
+ *
+ * Copyright (C) 2006, 2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software, see LICENSE.txt
+ */
+
+#ifndef SFMT_SSE2_H
+#define SFMT_SSE2_H
+
+/**
+ * This function represents the recursion formula.
+ * @param a a 128-bit part of the interal state array
+ * @param b a 128-bit part of the interal state array
+ * @param c a 128-bit part of the interal state array
+ * @param d a 128-bit part of the interal state array
+ * @param mask 128-bit mask
+ * @return output
+ */
+JEMALLOC_ALWAYS_INLINE __m128i mm_recursion(__m128i *a, __m128i *b,
+				   __m128i c, __m128i d, __m128i mask) {
+    __m128i v, x, y, z;
+
+    x = _mm_load_si128(a);
+    y = _mm_srli_epi32(*b, SR1);
+    z = _mm_srli_si128(c, SR2);
+    v = _mm_slli_epi32(d, SL1);
+    z = _mm_xor_si128(z, x);
+    z = _mm_xor_si128(z, v);
+    x = _mm_slli_si128(x, SL2);
+    y = _mm_and_si128(y, mask);
+    z = _mm_xor_si128(z, x);
+    z = _mm_xor_si128(z, y);
+    return z;
+}
+
+/**
+ * This function fills the internal state array with pseudorandom
+ * integers.
+ */
+static inline void gen_rand_all(sfmt_t *ctx) {
+    int i;
+    __m128i r, r1, r2, mask;
+    mask = _mm_set_epi32(MSK4, MSK3, MSK2, MSK1);
+
+    r1 = _mm_load_si128(&ctx->sfmt[N - 2].si);
+    r2 = _mm_load_si128(&ctx->sfmt[N - 1].si);
+    for (i = 0; i < N - POS1; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &ctx->sfmt[i + POS1].si, r1, r2,
+	  mask);
+	_mm_store_si128(&ctx->sfmt[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &ctx->sfmt[i + POS1 - N].si, r1, r2,
+	  mask);
+	_mm_store_si128(&ctx->sfmt[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+/**
+ * This function fills the user-specified array with pseudorandom
+ * integers.
+ *
+ * @param array an 128-bit array to be filled by pseudorandom numbers.
+ * @param size number of 128-bit pesudorandom numbers to be generated.
+ */
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size) {
+    int i, j;
+    __m128i r, r1, r2, mask;
+    mask = _mm_set_epi32(MSK4, MSK3, MSK2, MSK1);
+
+    r1 = _mm_load_si128(&ctx->sfmt[N - 2].si);
+    r2 = _mm_load_si128(&ctx->sfmt[N - 1].si);
+    for (i = 0; i < N - POS1; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &ctx->sfmt[i + POS1].si, r1, r2,
+	  mask);
+	_mm_store_si128(&array[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    for (; i < N; i++) {
+	r = mm_recursion(&ctx->sfmt[i].si, &array[i + POS1 - N].si, r1, r2,
+	  mask);
+	_mm_store_si128(&array[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    /* main loop */
+    for (; i < size - N; i++) {
+	r = mm_recursion(&array[i - N].si, &array[i + POS1 - N].si, r1, r2,
+			 mask);
+	_mm_store_si128(&array[i].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+    for (j = 0; j < 2 * N - size; j++) {
+	r = _mm_load_si128(&array[j + size - N].si);
+	_mm_store_si128(&ctx->sfmt[j].si, r);
+    }
+    for (; i < size; i++) {
+	r = mm_recursion(&array[i - N].si, &array[i + POS1 - N].si, r1, r2,
+			 mask);
+	_mm_store_si128(&array[i].si, r);
+	_mm_store_si128(&ctx->sfmt[j++].si, r);
+	r1 = r2;
+	r2 = r;
+    }
+}
+
+#endif
diff --git a/memkind-1.10.0/jemalloc/test/include/test/SFMT.h b/memkind-1.10.0/jemalloc/test/include/test/SFMT.h
new file mode 100644
index 0000000..863fc55
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/SFMT.h
@@ -0,0 +1,146 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/** 
+ * @file SFMT.h 
+ *
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT) pseudorandom
+ * number generator
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * Copyright (C) 2006, 2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software.
+ * see LICENSE.txt
+ *
+ * @note We assume that your system has inttypes.h.  If your system
+ * doesn't have inttypes.h, you have to typedef uint32_t and uint64_t,
+ * and you have to define PRIu64 and PRIx64 in this file as follows:
+ * @verbatim
+ typedef unsigned int uint32_t
+ typedef unsigned long long uint64_t  
+ #define PRIu64 "llu"
+ #define PRIx64 "llx"
+@endverbatim
+ * uint32_t must be exactly 32-bit unsigned integer type (no more, no
+ * less), and uint64_t must be exactly 64-bit unsigned integer type.
+ * PRIu64 and PRIx64 are used for printf function to print 64-bit
+ * unsigned int and 64-bit unsigned int in hexadecimal format.
+ */
+
+#ifndef SFMT_H
+#define SFMT_H
+
+typedef struct sfmt_s sfmt_t;
+
+uint32_t gen_rand32(sfmt_t *ctx);
+uint32_t gen_rand32_range(sfmt_t *ctx, uint32_t limit);
+uint64_t gen_rand64(sfmt_t *ctx);
+uint64_t gen_rand64_range(sfmt_t *ctx, uint64_t limit);
+void fill_array32(sfmt_t *ctx, uint32_t *array, int size);
+void fill_array64(sfmt_t *ctx, uint64_t *array, int size);
+sfmt_t *init_gen_rand(uint32_t seed);
+sfmt_t *init_by_array(uint32_t *init_key, int key_length);
+void fini_gen_rand(sfmt_t *ctx);
+const char *get_idstring(void);
+int get_min_array_size32(void);
+int get_min_array_size64(void);
+
+/* These real versions are due to Isaku Wada */
+/** generates a random number on [0,1]-real-interval */
+static inline double to_real1(uint32_t v) {
+    return v * (1.0/4294967295.0); 
+    /* divided by 2^32-1 */ 
+}
+
+/** generates a random number on [0,1]-real-interval */
+static inline double genrand_real1(sfmt_t *ctx) {
+    return to_real1(gen_rand32(ctx));
+}
+
+/** generates a random number on [0,1)-real-interval */
+static inline double to_real2(uint32_t v) {
+    return v * (1.0/4294967296.0); 
+    /* divided by 2^32 */
+}
+
+/** generates a random number on [0,1)-real-interval */
+static inline double genrand_real2(sfmt_t *ctx) {
+    return to_real2(gen_rand32(ctx));
+}
+
+/** generates a random number on (0,1)-real-interval */
+static inline double to_real3(uint32_t v) {
+    return (((double)v) + 0.5)*(1.0/4294967296.0); 
+    /* divided by 2^32 */
+}
+
+/** generates a random number on (0,1)-real-interval */
+static inline double genrand_real3(sfmt_t *ctx) {
+    return to_real3(gen_rand32(ctx));
+}
+/** These real versions are due to Isaku Wada */
+
+/** generates a random number on [0,1) with 53-bit resolution*/
+static inline double to_res53(uint64_t v) {
+    return v * (1.0/18446744073709551616.0L);
+}
+
+/** generates a random number on [0,1) with 53-bit resolution from two
+ * 32 bit integers */
+static inline double to_res53_mix(uint32_t x, uint32_t y) {
+    return to_res53(x | ((uint64_t)y << 32));
+}
+
+/** generates a random number on [0,1) with 53-bit resolution
+ */
+static inline double genrand_res53(sfmt_t *ctx) {
+    return to_res53(gen_rand64(ctx));
+}
+
+/** generates a random number on [0,1) with 53-bit resolution
+    using 32bit integer.
+ */
+static inline double genrand_res53_mix(sfmt_t *ctx) {
+    uint32_t x, y;
+
+    x = gen_rand32(ctx);
+    y = gen_rand32(ctx);
+    return to_res53_mix(x, y);
+}
+#endif
diff --git a/memkind-1.10.0/jemalloc/test/include/test/btalloc.h b/memkind-1.10.0/jemalloc/test/include/test/btalloc.h
new file mode 100644
index 0000000..5877ea7
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/btalloc.h
@@ -0,0 +1,30 @@
+/* btalloc() provides a mechanism for allocating via permuted backtraces. */
+void	*btalloc(size_t size, unsigned bits);
+
+#define btalloc_n_proto(n)						\
+void	*btalloc_##n(size_t size, unsigned bits);
+btalloc_n_proto(0)
+btalloc_n_proto(1)
+
+#define btalloc_n_gen(n)						\
+void *									\
+btalloc_##n(size_t size, unsigned bits) {				\
+	void *p;							\
+									\
+	if (bits == 0) {						\
+		p = mallocx(size, 0);					\
+	} else {							\
+		switch (bits & 0x1U) {					\
+		case 0:							\
+			p = (btalloc_0(size, bits >> 1));		\
+			break;						\
+		case 1:							\
+			p = (btalloc_1(size, bits >> 1));		\
+			break;						\
+		default: not_reached();					\
+		}							\
+	}								\
+	/* Intentionally sabotage tail call optimization. */		\
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");		\
+	return p;							\
+}
diff --git a/memkind-1.10.0/jemalloc/test/include/test/extent_hooks.h b/memkind-1.10.0/jemalloc/test/include/test/extent_hooks.h
new file mode 100644
index 0000000..1f06201
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/extent_hooks.h
@@ -0,0 +1,289 @@
+/*
+ * Boilerplate code used for testing extent hooks via interception and
+ * passthrough.
+ */
+
+static void	*extent_alloc_hook(extent_hooks_t *extent_hooks, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit,
+    unsigned arena_ind);
+static bool	extent_dalloc_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static void	extent_destroy_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, bool committed, unsigned arena_ind);
+static bool	extent_commit_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_decommit_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_purge_lazy_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_purge_forced_hook(extent_hooks_t *extent_hooks,
+    void *addr, size_t size, size_t offset, size_t length, unsigned arena_ind);
+static bool	extent_split_hook(extent_hooks_t *extent_hooks, void *addr,
+    size_t size, size_t size_a, size_t size_b, bool committed,
+    unsigned arena_ind);
+static bool	extent_merge_hook(extent_hooks_t *extent_hooks, void *addr_a,
+    size_t size_a, void *addr_b, size_t size_b, bool committed,
+    unsigned arena_ind);
+
+static extent_hooks_t *default_hooks;
+static extent_hooks_t hooks = {
+	extent_alloc_hook,
+	extent_dalloc_hook,
+	extent_destroy_hook,
+	extent_commit_hook,
+	extent_decommit_hook,
+	extent_purge_lazy_hook,
+	extent_purge_forced_hook,
+	extent_split_hook,
+	extent_merge_hook
+};
+
+/* Control whether hook functions pass calls through to default hooks. */
+static bool try_alloc = true;
+static bool try_dalloc = true;
+static bool try_destroy = true;
+static bool try_commit = true;
+static bool try_decommit = true;
+static bool try_purge_lazy = true;
+static bool try_purge_forced = true;
+static bool try_split = true;
+static bool try_merge = true;
+
+/* Set to false prior to operations, then introspect after operations. */
+static bool called_alloc;
+static bool called_dalloc;
+static bool called_destroy;
+static bool called_commit;
+static bool called_decommit;
+static bool called_purge_lazy;
+static bool called_purge_forced;
+static bool called_split;
+static bool called_merge;
+
+/* Set to false prior to operations, then introspect after operations. */
+static bool did_alloc;
+static bool did_dalloc;
+static bool did_destroy;
+static bool did_commit;
+static bool did_decommit;
+static bool did_purge_lazy;
+static bool did_purge_forced;
+static bool did_split;
+static bool did_merge;
+
+#if 0
+#  define TRACE_HOOK(fmt, ...) malloc_printf(fmt, __VA_ARGS__)
+#else
+#  define TRACE_HOOK(fmt, ...)
+#endif
+
+static void *
+extent_alloc_hook(extent_hooks_t *extent_hooks, void *new_addr, size_t size,
+    size_t alignment, bool *zero, bool *commit, unsigned arena_ind) {
+	void *ret;
+
+	TRACE_HOOK("%s(extent_hooks=%p, new_addr=%p, size=%zu, alignment=%zu, "
+	    "*zero=%s, *commit=%s, arena_ind=%u)\n", __func__, extent_hooks,
+	    new_addr, size, alignment, *zero ?  "true" : "false", *commit ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->alloc, extent_alloc_hook,
+	    "Wrong hook function");
+	called_alloc = true;
+	if (!try_alloc) {
+		return NULL;
+	}
+	ret = default_hooks->alloc(default_hooks, new_addr, size, alignment,
+	    zero, commit, 0);
+	did_alloc = (ret != NULL);
+	return ret;
+}
+
+static bool
+extent_dalloc_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, committed=%s, "
+	    "arena_ind=%u)\n", __func__, extent_hooks, addr, size, committed ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->dalloc, extent_dalloc_hook,
+	    "Wrong hook function");
+	called_dalloc = true;
+	if (!try_dalloc) {
+		return true;
+	}
+	err = default_hooks->dalloc(default_hooks, addr, size, committed, 0);
+	did_dalloc = !err;
+	return err;
+}
+
+static void
+extent_destroy_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, committed=%s, "
+	    "arena_ind=%u)\n", __func__, extent_hooks, addr, size, committed ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->destroy, extent_destroy_hook,
+	    "Wrong hook function");
+	called_destroy = true;
+	if (!try_destroy) {
+		return;
+	}
+	default_hooks->destroy(default_hooks, addr, size, committed, 0);
+	did_destroy = true;
+}
+
+static bool
+extent_commit_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu, arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->commit, extent_commit_hook,
+	    "Wrong hook function");
+	called_commit = true;
+	if (!try_commit) {
+		return true;
+	}
+	err = default_hooks->commit(default_hooks, addr, size, offset, length,
+	    0);
+	did_commit = !err;
+	return err;
+}
+
+static bool
+extent_decommit_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu, arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->decommit, extent_decommit_hook,
+	    "Wrong hook function");
+	called_decommit = true;
+	if (!try_decommit) {
+		return true;
+	}
+	err = default_hooks->decommit(default_hooks, addr, size, offset, length,
+	    0);
+	did_decommit = !err;
+	return err;
+}
+
+static bool
+extent_purge_lazy_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->purge_lazy, extent_purge_lazy_hook,
+	    "Wrong hook function");
+	called_purge_lazy = true;
+	if (!try_purge_lazy) {
+		return true;
+	}
+	err = default_hooks->purge_lazy == NULL ||
+	    default_hooks->purge_lazy(default_hooks, addr, size, offset, length,
+	    0);
+	did_purge_lazy = !err;
+	return err;
+}
+
+static bool
+extent_purge_forced_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t offset, size_t length, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, offset=%zu, "
+	    "length=%zu arena_ind=%u)\n", __func__, extent_hooks, addr, size,
+	    offset, length, arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->purge_forced, extent_purge_forced_hook,
+	    "Wrong hook function");
+	called_purge_forced = true;
+	if (!try_purge_forced) {
+		return true;
+	}
+	err = default_hooks->purge_forced == NULL ||
+	    default_hooks->purge_forced(default_hooks, addr, size, offset,
+	    length, 0);
+	did_purge_forced = !err;
+	return err;
+}
+
+static bool
+extent_split_hook(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    size_t size_a, size_t size_b, bool committed, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, size_a=%zu, "
+	    "size_b=%zu, committed=%s, arena_ind=%u)\n", __func__, extent_hooks,
+	    addr, size, size_a, size_b, committed ? "true" : "false",
+	    arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->split, extent_split_hook,
+	    "Wrong hook function");
+	called_split = true;
+	if (!try_split) {
+		return true;
+	}
+	err = (default_hooks->split == NULL ||
+	    default_hooks->split(default_hooks, addr, size, size_a, size_b,
+	    committed, 0));
+	did_split = !err;
+	return err;
+}
+
+static bool
+extent_merge_hook(extent_hooks_t *extent_hooks, void *addr_a, size_t size_a,
+    void *addr_b, size_t size_b, bool committed, unsigned arena_ind) {
+	bool err;
+
+	TRACE_HOOK("%s(extent_hooks=%p, addr_a=%p, size_a=%zu, addr_b=%p "
+	    "size_b=%zu, committed=%s, arena_ind=%u)\n", __func__, extent_hooks,
+	    addr_a, size_a, addr_b, size_b, committed ? "true" : "false",
+	    arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->merge, extent_merge_hook,
+	    "Wrong hook function");
+	assert_ptr_eq((void *)((uintptr_t)addr_a + size_a), addr_b,
+	    "Extents not mergeable");
+	called_merge = true;
+	if (!try_merge) {
+		return true;
+	}
+	err = (default_hooks->merge == NULL ||
+	    default_hooks->merge(default_hooks, addr_a, size_a, addr_b, size_b,
+	    committed, 0));
+	did_merge = !err;
+	return err;
+}
+
+static void
+extent_hooks_prep(void) {
+	size_t sz;
+
+	sz = sizeof(default_hooks);
+	assert_d_eq(mallctl("arena.0.extent_hooks", (void *)&default_hooks, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() error");
+}
diff --git a/memkind-1.10.0/jemalloc/test/include/test/jemalloc_test.h.in b/memkind-1.10.0/jemalloc/test/include/test/jemalloc_test.h.in
new file mode 100644
index 0000000..c46af5d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/jemalloc_test.h.in
@@ -0,0 +1,173 @@
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <limits.h>
+#ifndef SIZE_T_MAX
+#  define SIZE_T_MAX	SIZE_MAX
+#endif
+#include <stdlib.h>
+#include <stdarg.h>
+#include <stdbool.h>
+#include <errno.h>
+#include <math.h>
+#include <string.h>
+#ifdef _WIN32
+#  include "msvc_compat/strings.h"
+#endif
+
+#ifdef _WIN32
+#  include <windows.h>
+#  include "msvc_compat/windows_extra.h"
+#else
+#  include <pthread.h>
+#endif
+
+#include "test/jemalloc_test_defs.h"
+
+#if defined(JEMALLOC_OSATOMIC)
+#  include <libkern/OSAtomic.h>
+#endif
+
+#if defined(HAVE_ALTIVEC) && !defined(__APPLE__)
+#  include <altivec.h>
+#endif
+#ifdef HAVE_SSE2
+#  include <emmintrin.h>
+#endif
+
+/******************************************************************************/
+/*
+ * For unit tests, expose all public and private interfaces.
+ */
+#ifdef JEMALLOC_UNIT_TEST
+#  define JEMALLOC_JET
+#  define JEMALLOC_MANGLE
+#  include "jemalloc/internal/jemalloc_preamble.h"
+#  include "jemalloc/internal/jemalloc_internal_includes.h"
+
+/******************************************************************************/
+/*
+ * For integration tests, expose the public jemalloc interfaces, but only
+ * expose the minimum necessary internal utility code (to avoid re-implementing
+ * essentially identical code within the test infrastructure).
+ */
+#elif defined(JEMALLOC_INTEGRATION_TEST) || \
+    defined(JEMALLOC_INTEGRATION_CPP_TEST)
+#  define JEMALLOC_MANGLE
+#  include "jemalloc/jemalloc@install_suffix@.h"
+#  include "jemalloc/internal/jemalloc_internal_defs.h"
+#  include "jemalloc/internal/jemalloc_internal_macros.h"
+
+static const bool config_debug =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+
+#  define JEMALLOC_N(n) @private_namespace@##n
+#  include "jemalloc/internal/private_namespace.h"
+#  include "jemalloc/internal/test_hooks.h"
+
+/* Hermetic headers. */
+#  include "jemalloc/internal/assert.h"
+#  include "jemalloc/internal/malloc_io.h"
+#  include "jemalloc/internal/nstime.h"
+#  include "jemalloc/internal/util.h"
+
+/* Non-hermetic headers. */
+#  include "jemalloc/internal/qr.h"
+#  include "jemalloc/internal/ql.h"
+
+/******************************************************************************/
+/*
+ * For stress tests, expose the public jemalloc interfaces with name mangling
+ * so that they can be tested as e.g. malloc() and free().  Also expose the
+ * public jemalloc interfaces with jet_ prefixes, so that stress tests can use
+ * a separate allocator for their internal data structures.
+ */
+#elif defined(JEMALLOC_STRESS_TEST)
+#  include "jemalloc/jemalloc@install_suffix@.h"
+
+#  include "jemalloc/jemalloc_protos_jet.h"
+
+#  define JEMALLOC_JET
+#  include "jemalloc/internal/jemalloc_preamble.h"
+#  include "jemalloc/internal/jemalloc_internal_includes.h"
+#  include "jemalloc/internal/public_unnamespace.h"
+#  undef JEMALLOC_JET
+
+#  include "jemalloc/jemalloc_rename.h"
+#  define JEMALLOC_MANGLE
+#  ifdef JEMALLOC_STRESS_TESTLIB
+#    include "jemalloc/jemalloc_mangle_jet.h"
+#  else
+#    include "jemalloc/jemalloc_mangle.h"
+#  endif
+
+/******************************************************************************/
+/*
+ * This header does dangerous things, the effects of which only test code
+ * should be subject to.
+ */
+#else
+#  error "This header cannot be included outside a testing context"
+#endif
+
+/******************************************************************************/
+/*
+ * Common test utilities.
+ */
+#include "test/btalloc.h"
+#include "test/math.h"
+#include "test/mtx.h"
+#include "test/mq.h"
+#include "test/test.h"
+#include "test/timer.h"
+#include "test/thd.h"
+#define MEXP 19937
+#include "test/SFMT.h"
+
+/******************************************************************************/
+/*
+ * Define always-enabled assertion macros, so that test assertions execute even
+ * if assertions are disabled in the library code.
+ */
+#undef assert
+#undef not_reached
+#undef not_implemented
+#undef assert_not_implemented
+
+#define assert(e) do {							\
+	if (!(e)) {							\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Failed assertion: \"%s\"\n",	\
+		    __FILE__, __LINE__, #e);				\
+		abort();						\
+	}								\
+} while (0)
+
+#define not_reached() do {						\
+	malloc_printf(							\
+	    "<jemalloc>: %s:%d: Unreachable code reached\n",		\
+	    __FILE__, __LINE__);					\
+	abort();							\
+} while (0)
+
+#define not_implemented() do {						\
+	malloc_printf("<jemalloc>: %s:%d: Not implemented\n",		\
+	    __FILE__, __LINE__);					\
+	abort();							\
+} while (0)
+
+#define assert_not_implemented(e) do {					\
+	if (!(e)) {							\
+		not_implemented();					\
+	}								\
+} while (0)
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/memkind-1.10.0/jemalloc/test/include/test/jemalloc_test_defs.h.in b/memkind-1.10.0/jemalloc/test/include/test/jemalloc_test_defs.h.in
new file mode 100644
index 0000000..5cc8532
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/jemalloc_test_defs.h.in
@@ -0,0 +1,9 @@
+#include "jemalloc/internal/jemalloc_internal_defs.h"
+#include "jemalloc/internal/jemalloc_internal_decls.h"
+
+/*
+ * For use by SFMT.  configure.ac doesn't actually define HAVE_SSE2 because its
+ * dependencies are notoriously unportable in practice.
+ */
+#undef HAVE_SSE2
+#undef HAVE_ALTIVEC
diff --git a/memkind-1.10.0/jemalloc/test/include/test/math.h b/memkind-1.10.0/jemalloc/test/include/test/math.h
new file mode 100644
index 0000000..efba086
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/math.h
@@ -0,0 +1,306 @@
+/*
+ * Compute the natural log of Gamma(x), accurate to 10 decimal places.
+ *
+ * This implementation is based on:
+ *
+ *   Pike, M.C., I.D. Hill (1966) Algorithm 291: Logarithm of Gamma function
+ *   [S14].  Communications of the ACM 9(9):684.
+ */
+static inline double
+ln_gamma(double x) {
+	double f, z;
+
+	assert(x > 0.0);
+
+	if (x < 7.0) {
+		f = 1.0;
+		z = x;
+		while (z < 7.0) {
+			f *= z;
+			z += 1.0;
+		}
+		x = z;
+		f = -log(f);
+	} else {
+		f = 0.0;
+	}
+
+	z = 1.0 / (x * x);
+
+	return f + (x-0.5) * log(x) - x + 0.918938533204673 +
+	    (((-0.000595238095238 * z + 0.000793650793651) * z -
+	    0.002777777777778) * z + 0.083333333333333) / x;
+}
+
+/*
+ * Compute the incomplete Gamma ratio for [0..x], where p is the shape
+ * parameter, and ln_gamma_p is ln_gamma(p).
+ *
+ * This implementation is based on:
+ *
+ *   Bhattacharjee, G.P. (1970) Algorithm AS 32: The incomplete Gamma integral.
+ *   Applied Statistics 19:285-287.
+ */
+static inline double
+i_gamma(double x, double p, double ln_gamma_p) {
+	double acu, factor, oflo, gin, term, rn, a, b, an, dif;
+	double pn[6];
+	unsigned i;
+
+	assert(p > 0.0);
+	assert(x >= 0.0);
+
+	if (x == 0.0) {
+		return 0.0;
+	}
+
+	acu = 1.0e-10;
+	oflo = 1.0e30;
+	gin = 0.0;
+	factor = exp(p * log(x) - x - ln_gamma_p);
+
+	if (x <= 1.0 || x < p) {
+		/* Calculation by series expansion. */
+		gin = 1.0;
+		term = 1.0;
+		rn = p;
+
+		while (true) {
+			rn += 1.0;
+			term *= x / rn;
+			gin += term;
+			if (term <= acu) {
+				gin *= factor / p;
+				return gin;
+			}
+		}
+	} else {
+		/* Calculation by continued fraction. */
+		a = 1.0 - p;
+		b = a + x + 1.0;
+		term = 0.0;
+		pn[0] = 1.0;
+		pn[1] = x;
+		pn[2] = x + 1.0;
+		pn[3] = x * b;
+		gin = pn[2] / pn[3];
+
+		while (true) {
+			a += 1.0;
+			b += 2.0;
+			term += 1.0;
+			an = a * term;
+			for (i = 0; i < 2; i++) {
+				pn[i+4] = b * pn[i+2] - an * pn[i];
+			}
+			if (pn[5] != 0.0) {
+				rn = pn[4] / pn[5];
+				dif = fabs(gin - rn);
+				if (dif <= acu && dif <= acu * rn) {
+					gin = 1.0 - factor * gin;
+					return gin;
+				}
+				gin = rn;
+			}
+			for (i = 0; i < 4; i++) {
+				pn[i] = pn[i+2];
+			}
+
+			if (fabs(pn[4]) >= oflo) {
+				for (i = 0; i < 4; i++) {
+					pn[i] /= oflo;
+				}
+			}
+		}
+	}
+}
+
+/*
+ * Given a value p in [0..1] of the lower tail area of the normal distribution,
+ * compute the limit on the definite integral from [-inf..z] that satisfies p,
+ * accurate to 16 decimal places.
+ *
+ * This implementation is based on:
+ *
+ *   Wichura, M.J. (1988) Algorithm AS 241: The percentage points of the normal
+ *   distribution.  Applied Statistics 37(3):477-484.
+ */
+static inline double
+pt_norm(double p) {
+	double q, r, ret;
+
+	assert(p > 0.0 && p < 1.0);
+
+	q = p - 0.5;
+	if (fabs(q) <= 0.425) {
+		/* p close to 1/2. */
+		r = 0.180625 - q * q;
+		return q * (((((((2.5090809287301226727e3 * r +
+		    3.3430575583588128105e4) * r + 6.7265770927008700853e4) * r
+		    + 4.5921953931549871457e4) * r + 1.3731693765509461125e4) *
+		    r + 1.9715909503065514427e3) * r + 1.3314166789178437745e2)
+		    * r + 3.3871328727963666080e0) /
+		    (((((((5.2264952788528545610e3 * r +
+		    2.8729085735721942674e4) * r + 3.9307895800092710610e4) * r
+		    + 2.1213794301586595867e4) * r + 5.3941960214247511077e3) *
+		    r + 6.8718700749205790830e2) * r + 4.2313330701600911252e1)
+		    * r + 1.0);
+	} else {
+		if (q < 0.0) {
+			r = p;
+		} else {
+			r = 1.0 - p;
+		}
+		assert(r > 0.0);
+
+		r = sqrt(-log(r));
+		if (r <= 5.0) {
+			/* p neither close to 1/2 nor 0 or 1. */
+			r -= 1.6;
+			ret = ((((((((7.74545014278341407640e-4 * r +
+			    2.27238449892691845833e-2) * r +
+			    2.41780725177450611770e-1) * r +
+			    1.27045825245236838258e0) * r +
+			    3.64784832476320460504e0) * r +
+			    5.76949722146069140550e0) * r +
+			    4.63033784615654529590e0) * r +
+			    1.42343711074968357734e0) /
+			    (((((((1.05075007164441684324e-9 * r +
+			    5.47593808499534494600e-4) * r +
+			    1.51986665636164571966e-2)
+			    * r + 1.48103976427480074590e-1) * r +
+			    6.89767334985100004550e-1) * r +
+			    1.67638483018380384940e0) * r +
+			    2.05319162663775882187e0) * r + 1.0));
+		} else {
+			/* p near 0 or 1. */
+			r -= 5.0;
+			ret = ((((((((2.01033439929228813265e-7 * r +
+			    2.71155556874348757815e-5) * r +
+			    1.24266094738807843860e-3) * r +
+			    2.65321895265761230930e-2) * r +
+			    2.96560571828504891230e-1) * r +
+			    1.78482653991729133580e0) * r +
+			    5.46378491116411436990e0) * r +
+			    6.65790464350110377720e0) /
+			    (((((((2.04426310338993978564e-15 * r +
+			    1.42151175831644588870e-7) * r +
+			    1.84631831751005468180e-5) * r +
+			    7.86869131145613259100e-4) * r +
+			    1.48753612908506148525e-2) * r +
+			    1.36929880922735805310e-1) * r +
+			    5.99832206555887937690e-1)
+			    * r + 1.0));
+		}
+		if (q < 0.0) {
+			ret = -ret;
+		}
+		return ret;
+	}
+}
+
+/*
+ * Given a value p in [0..1] of the lower tail area of the Chi^2 distribution
+ * with df degrees of freedom, where ln_gamma_df_2 is ln_gamma(df/2.0), compute
+ * the upper limit on the definite integral from [0..z] that satisfies p,
+ * accurate to 12 decimal places.
+ *
+ * This implementation is based on:
+ *
+ *   Best, D.J., D.E. Roberts (1975) Algorithm AS 91: The percentage points of
+ *   the Chi^2 distribution.  Applied Statistics 24(3):385-388.
+ *
+ *   Shea, B.L. (1991) Algorithm AS R85: A remark on AS 91: The percentage
+ *   points of the Chi^2 distribution.  Applied Statistics 40(1):233-235.
+ */
+static inline double
+pt_chi2(double p, double df, double ln_gamma_df_2) {
+	double e, aa, xx, c, ch, a, q, p1, p2, t, x, b, s1, s2, s3, s4, s5, s6;
+	unsigned i;
+
+	assert(p >= 0.0 && p < 1.0);
+	assert(df > 0.0);
+
+	e = 5.0e-7;
+	aa = 0.6931471805;
+
+	xx = 0.5 * df;
+	c = xx - 1.0;
+
+	if (df < -1.24 * log(p)) {
+		/* Starting approximation for small Chi^2. */
+		ch = pow(p * xx * exp(ln_gamma_df_2 + xx * aa), 1.0 / xx);
+		if (ch - e < 0.0) {
+			return ch;
+		}
+	} else {
+		if (df > 0.32) {
+			x = pt_norm(p);
+			/*
+			 * Starting approximation using Wilson and Hilferty
+			 * estimate.
+			 */
+			p1 = 0.222222 / df;
+			ch = df * pow(x * sqrt(p1) + 1.0 - p1, 3.0);
+			/* Starting approximation for p tending to 1. */
+			if (ch > 2.2 * df + 6.0) {
+				ch = -2.0 * (log(1.0 - p) - c * log(0.5 * ch) +
+				    ln_gamma_df_2);
+			}
+		} else {
+			ch = 0.4;
+			a = log(1.0 - p);
+			while (true) {
+				q = ch;
+				p1 = 1.0 + ch * (4.67 + ch);
+				p2 = ch * (6.73 + ch * (6.66 + ch));
+				t = -0.5 + (4.67 + 2.0 * ch) / p1 - (6.73 + ch
+				    * (13.32 + 3.0 * ch)) / p2;
+				ch -= (1.0 - exp(a + ln_gamma_df_2 + 0.5 * ch +
+				    c * aa) * p2 / p1) / t;
+				if (fabs(q / ch - 1.0) - 0.01 <= 0.0) {
+					break;
+				}
+			}
+		}
+	}
+
+	for (i = 0; i < 20; i++) {
+		/* Calculation of seven-term Taylor series. */
+		q = ch;
+		p1 = 0.5 * ch;
+		if (p1 < 0.0) {
+			return -1.0;
+		}
+		p2 = p - i_gamma(p1, xx, ln_gamma_df_2);
+		t = p2 * exp(xx * aa + ln_gamma_df_2 + p1 - c * log(ch));
+		b = t / ch;
+		a = 0.5 * t - b * c;
+		s1 = (210.0 + a * (140.0 + a * (105.0 + a * (84.0 + a * (70.0 +
+		    60.0 * a))))) / 420.0;
+		s2 = (420.0 + a * (735.0 + a * (966.0 + a * (1141.0 + 1278.0 *
+		    a)))) / 2520.0;
+		s3 = (210.0 + a * (462.0 + a * (707.0 + 932.0 * a))) / 2520.0;
+		s4 = (252.0 + a * (672.0 + 1182.0 * a) + c * (294.0 + a *
+		    (889.0 + 1740.0 * a))) / 5040.0;
+		s5 = (84.0 + 264.0 * a + c * (175.0 + 606.0 * a)) / 2520.0;
+		s6 = (120.0 + c * (346.0 + 127.0 * c)) / 5040.0;
+		ch += t * (1.0 + 0.5 * t * s1 - b * c * (s1 - b * (s2 - b * (s3
+		    - b * (s4 - b * (s5 - b * s6))))));
+		if (fabs(q / ch - 1.0) <= e) {
+			break;
+		}
+	}
+
+	return ch;
+}
+
+/*
+ * Given a value p in [0..1] and Gamma distribution shape and scale parameters,
+ * compute the upper limit on the definite integral from [0..z] that satisfies
+ * p.
+ */
+static inline double
+pt_gamma(double p, double shape, double scale, double ln_gamma_shape) {
+	return pt_chi2(p, shape * 2.0, ln_gamma_shape) * 0.5 * scale;
+}
diff --git a/memkind-1.10.0/jemalloc/test/include/test/mq.h b/memkind-1.10.0/jemalloc/test/include/test/mq.h
new file mode 100644
index 0000000..af2c078
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/mq.h
@@ -0,0 +1,107 @@
+void	mq_nanosleep(unsigned ns);
+
+/*
+ * Simple templated message queue implementation that relies on only mutexes for
+ * synchronization (which reduces portability issues).  Given the following
+ * setup:
+ *
+ *   typedef struct mq_msg_s mq_msg_t;
+ *   struct mq_msg_s {
+ *           mq_msg(mq_msg_t) link;
+ *           [message data]
+ *   };
+ *   mq_gen(, mq_, mq_t, mq_msg_t, link)
+ *
+ * The API is as follows:
+ *
+ *   bool mq_init(mq_t *mq);
+ *   void mq_fini(mq_t *mq);
+ *   unsigned mq_count(mq_t *mq);
+ *   mq_msg_t *mq_tryget(mq_t *mq);
+ *   mq_msg_t *mq_get(mq_t *mq);
+ *   void mq_put(mq_t *mq, mq_msg_t *msg);
+ *
+ * The message queue linkage embedded in each message is to be treated as
+ * externally opaque (no need to initialize or clean up externally).  mq_fini()
+ * does not perform any cleanup of messages, since it knows nothing of their
+ * payloads.
+ */
+#define mq_msg(a_mq_msg_type)	ql_elm(a_mq_msg_type)
+
+#define mq_gen(a_attr, a_prefix, a_mq_type, a_mq_msg_type, a_field)	\
+typedef struct {							\
+	mtx_t			lock;					\
+	ql_head(a_mq_msg_type)	msgs;					\
+	unsigned		count;					\
+} a_mq_type;								\
+a_attr bool								\
+a_prefix##init(a_mq_type *mq) {						\
+									\
+	if (mtx_init(&mq->lock)) {					\
+		return true;						\
+	}								\
+	ql_new(&mq->msgs);						\
+	mq->count = 0;							\
+	return false;							\
+}									\
+a_attr void								\
+a_prefix##fini(a_mq_type *mq) {						\
+	mtx_fini(&mq->lock);						\
+}									\
+a_attr unsigned								\
+a_prefix##count(a_mq_type *mq) {					\
+	unsigned count;							\
+									\
+	mtx_lock(&mq->lock);						\
+	count = mq->count;						\
+	mtx_unlock(&mq->lock);						\
+	return count;							\
+}									\
+a_attr a_mq_msg_type *							\
+a_prefix##tryget(a_mq_type *mq) {					\
+	a_mq_msg_type *msg;						\
+									\
+	mtx_lock(&mq->lock);						\
+	msg = ql_first(&mq->msgs);					\
+	if (msg != NULL) {						\
+		ql_head_remove(&mq->msgs, a_mq_msg_type, a_field);	\
+		mq->count--;						\
+	}								\
+	mtx_unlock(&mq->lock);						\
+	return msg;							\
+}									\
+a_attr a_mq_msg_type *							\
+a_prefix##get(a_mq_type *mq) {						\
+	a_mq_msg_type *msg;						\
+	unsigned ns;							\
+									\
+	msg = a_prefix##tryget(mq);					\
+	if (msg != NULL) {						\
+		return msg;						\
+	}								\
+									\
+	ns = 1;								\
+	while (true) {							\
+		mq_nanosleep(ns);					\
+		msg = a_prefix##tryget(mq);				\
+		if (msg != NULL) {					\
+			return msg;					\
+		}							\
+		if (ns < 1000*1000*1000) {				\
+			/* Double sleep time, up to max 1 second. */	\
+			ns <<= 1;					\
+			if (ns > 1000*1000*1000) {			\
+				ns = 1000*1000*1000;			\
+			}						\
+		}							\
+	}								\
+}									\
+a_attr void								\
+a_prefix##put(a_mq_type *mq, a_mq_msg_type *msg) {			\
+									\
+	mtx_lock(&mq->lock);						\
+	ql_elm_new(msg, a_field);					\
+	ql_tail_insert(&mq->msgs, msg, a_field);			\
+	mq->count++;							\
+	mtx_unlock(&mq->lock);						\
+}
diff --git a/memkind-1.10.0/jemalloc/test/include/test/mtx.h b/memkind-1.10.0/jemalloc/test/include/test/mtx.h
new file mode 100644
index 0000000..066a213
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/mtx.h
@@ -0,0 +1,21 @@
+/*
+ * mtx is a slightly simplified version of malloc_mutex.  This code duplication
+ * is unfortunate, but there are allocator bootstrapping considerations that
+ * would leak into the test infrastructure if malloc_mutex were used directly
+ * in tests.
+ */
+
+typedef struct {
+#ifdef _WIN32
+	CRITICAL_SECTION	lock;
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	os_unfair_lock		lock;
+#else
+	pthread_mutex_t		lock;
+#endif
+} mtx_t;
+
+bool	mtx_init(mtx_t *mtx);
+void	mtx_fini(mtx_t *mtx);
+void	mtx_lock(mtx_t *mtx);
+void	mtx_unlock(mtx_t *mtx);
diff --git a/memkind-1.10.0/jemalloc/test/include/test/test.h b/memkind-1.10.0/jemalloc/test/include/test/test.h
new file mode 100644
index 0000000..fd0e526
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/test.h
@@ -0,0 +1,338 @@
+#define ASSERT_BUFSIZE	256
+
+#define assert_cmp(t, a, b, cmp, neg_cmp, pri, ...) do {		\
+	t a_ = (a);							\
+	t b_ = (b);							\
+	if (!(a_ cmp b_)) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) " #cmp " (%s) --> "				\
+		    "%" pri " " #neg_cmp " %" pri ": ",			\
+		    __func__, __FILE__, __LINE__,			\
+		    #a, #b, a_, b_);					\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+
+#define assert_ptr_eq(a, b, ...)	assert_cmp(void *, a, b, ==,	\
+    !=, "p", __VA_ARGS__)
+#define assert_ptr_ne(a, b, ...)	assert_cmp(void *, a, b, !=,	\
+    ==, "p", __VA_ARGS__)
+#define assert_ptr_null(a, ...)		assert_cmp(void *, a, NULL, ==,	\
+    !=, "p", __VA_ARGS__)
+#define assert_ptr_not_null(a, ...)	assert_cmp(void *, a, NULL, !=,	\
+    ==, "p", __VA_ARGS__)
+
+#define assert_c_eq(a, b, ...)	assert_cmp(char, a, b, ==, !=, "c", __VA_ARGS__)
+#define assert_c_ne(a, b, ...)	assert_cmp(char, a, b, !=, ==, "c", __VA_ARGS__)
+#define assert_c_lt(a, b, ...)	assert_cmp(char, a, b, <, >=, "c", __VA_ARGS__)
+#define assert_c_le(a, b, ...)	assert_cmp(char, a, b, <=, >, "c", __VA_ARGS__)
+#define assert_c_ge(a, b, ...)	assert_cmp(char, a, b, >=, <, "c", __VA_ARGS__)
+#define assert_c_gt(a, b, ...)	assert_cmp(char, a, b, >, <=, "c", __VA_ARGS__)
+
+#define assert_x_eq(a, b, ...)	assert_cmp(int, a, b, ==, !=, "#x", __VA_ARGS__)
+#define assert_x_ne(a, b, ...)	assert_cmp(int, a, b, !=, ==, "#x", __VA_ARGS__)
+#define assert_x_lt(a, b, ...)	assert_cmp(int, a, b, <, >=, "#x", __VA_ARGS__)
+#define assert_x_le(a, b, ...)	assert_cmp(int, a, b, <=, >, "#x", __VA_ARGS__)
+#define assert_x_ge(a, b, ...)	assert_cmp(int, a, b, >=, <, "#x", __VA_ARGS__)
+#define assert_x_gt(a, b, ...)	assert_cmp(int, a, b, >, <=, "#x", __VA_ARGS__)
+
+#define assert_d_eq(a, b, ...)	assert_cmp(int, a, b, ==, !=, "d", __VA_ARGS__)
+#define assert_d_ne(a, b, ...)	assert_cmp(int, a, b, !=, ==, "d", __VA_ARGS__)
+#define assert_d_lt(a, b, ...)	assert_cmp(int, a, b, <, >=, "d", __VA_ARGS__)
+#define assert_d_le(a, b, ...)	assert_cmp(int, a, b, <=, >, "d", __VA_ARGS__)
+#define assert_d_ge(a, b, ...)	assert_cmp(int, a, b, >=, <, "d", __VA_ARGS__)
+#define assert_d_gt(a, b, ...)	assert_cmp(int, a, b, >, <=, "d", __VA_ARGS__)
+
+#define assert_u_eq(a, b, ...)	assert_cmp(int, a, b, ==, !=, "u", __VA_ARGS__)
+#define assert_u_ne(a, b, ...)	assert_cmp(int, a, b, !=, ==, "u", __VA_ARGS__)
+#define assert_u_lt(a, b, ...)	assert_cmp(int, a, b, <, >=, "u", __VA_ARGS__)
+#define assert_u_le(a, b, ...)	assert_cmp(int, a, b, <=, >, "u", __VA_ARGS__)
+#define assert_u_ge(a, b, ...)	assert_cmp(int, a, b, >=, <, "u", __VA_ARGS__)
+#define assert_u_gt(a, b, ...)	assert_cmp(int, a, b, >, <=, "u", __VA_ARGS__)
+
+#define assert_ld_eq(a, b, ...)	assert_cmp(long, a, b, ==,	\
+    !=, "ld", __VA_ARGS__)
+#define assert_ld_ne(a, b, ...)	assert_cmp(long, a, b, !=,	\
+    ==, "ld", __VA_ARGS__)
+#define assert_ld_lt(a, b, ...)	assert_cmp(long, a, b, <,	\
+    >=, "ld", __VA_ARGS__)
+#define assert_ld_le(a, b, ...)	assert_cmp(long, a, b, <=,	\
+    >, "ld", __VA_ARGS__)
+#define assert_ld_ge(a, b, ...)	assert_cmp(long, a, b, >=,	\
+    <, "ld", __VA_ARGS__)
+#define assert_ld_gt(a, b, ...)	assert_cmp(long, a, b, >,	\
+    <=, "ld", __VA_ARGS__)
+
+#define assert_lu_eq(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, ==, !=, "lu", __VA_ARGS__)
+#define assert_lu_ne(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, !=, ==, "lu", __VA_ARGS__)
+#define assert_lu_lt(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, <, >=, "lu", __VA_ARGS__)
+#define assert_lu_le(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, <=, >, "lu", __VA_ARGS__)
+#define assert_lu_ge(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, >=, <, "lu", __VA_ARGS__)
+#define assert_lu_gt(a, b, ...)	assert_cmp(unsigned long,	\
+    a, b, >, <=, "lu", __VA_ARGS__)
+
+#define assert_qd_eq(a, b, ...)	assert_cmp(long long, a, b, ==,	\
+    !=, "qd", __VA_ARGS__)
+#define assert_qd_ne(a, b, ...)	assert_cmp(long long, a, b, !=,	\
+    ==, "qd", __VA_ARGS__)
+#define assert_qd_lt(a, b, ...)	assert_cmp(long long, a, b, <,	\
+    >=, "qd", __VA_ARGS__)
+#define assert_qd_le(a, b, ...)	assert_cmp(long long, a, b, <=,	\
+    >, "qd", __VA_ARGS__)
+#define assert_qd_ge(a, b, ...)	assert_cmp(long long, a, b, >=,	\
+    <, "qd", __VA_ARGS__)
+#define assert_qd_gt(a, b, ...)	assert_cmp(long long, a, b, >,	\
+    <=, "qd", __VA_ARGS__)
+
+#define assert_qu_eq(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, ==, !=, "qu", __VA_ARGS__)
+#define assert_qu_ne(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, !=, ==, "qu", __VA_ARGS__)
+#define assert_qu_lt(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, <, >=, "qu", __VA_ARGS__)
+#define assert_qu_le(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, <=, >, "qu", __VA_ARGS__)
+#define assert_qu_ge(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, >=, <, "qu", __VA_ARGS__)
+#define assert_qu_gt(a, b, ...)	assert_cmp(unsigned long long,	\
+    a, b, >, <=, "qu", __VA_ARGS__)
+
+#define assert_jd_eq(a, b, ...)	assert_cmp(intmax_t, a, b, ==,	\
+    !=, "jd", __VA_ARGS__)
+#define assert_jd_ne(a, b, ...)	assert_cmp(intmax_t, a, b, !=,	\
+    ==, "jd", __VA_ARGS__)
+#define assert_jd_lt(a, b, ...)	assert_cmp(intmax_t, a, b, <,	\
+    >=, "jd", __VA_ARGS__)
+#define assert_jd_le(a, b, ...)	assert_cmp(intmax_t, a, b, <=,	\
+    >, "jd", __VA_ARGS__)
+#define assert_jd_ge(a, b, ...)	assert_cmp(intmax_t, a, b, >=,	\
+    <, "jd", __VA_ARGS__)
+#define assert_jd_gt(a, b, ...)	assert_cmp(intmax_t, a, b, >,	\
+    <=, "jd", __VA_ARGS__)
+
+#define assert_ju_eq(a, b, ...)	assert_cmp(uintmax_t, a, b, ==,	\
+    !=, "ju", __VA_ARGS__)
+#define assert_ju_ne(a, b, ...)	assert_cmp(uintmax_t, a, b, !=,	\
+    ==, "ju", __VA_ARGS__)
+#define assert_ju_lt(a, b, ...)	assert_cmp(uintmax_t, a, b, <,	\
+    >=, "ju", __VA_ARGS__)
+#define assert_ju_le(a, b, ...)	assert_cmp(uintmax_t, a, b, <=,	\
+    >, "ju", __VA_ARGS__)
+#define assert_ju_ge(a, b, ...)	assert_cmp(uintmax_t, a, b, >=,	\
+    <, "ju", __VA_ARGS__)
+#define assert_ju_gt(a, b, ...)	assert_cmp(uintmax_t, a, b, >,	\
+    <=, "ju", __VA_ARGS__)
+
+#define assert_zd_eq(a, b, ...)	assert_cmp(ssize_t, a, b, ==,	\
+    !=, "zd", __VA_ARGS__)
+#define assert_zd_ne(a, b, ...)	assert_cmp(ssize_t, a, b, !=,	\
+    ==, "zd", __VA_ARGS__)
+#define assert_zd_lt(a, b, ...)	assert_cmp(ssize_t, a, b, <,	\
+    >=, "zd", __VA_ARGS__)
+#define assert_zd_le(a, b, ...)	assert_cmp(ssize_t, a, b, <=,	\
+    >, "zd", __VA_ARGS__)
+#define assert_zd_ge(a, b, ...)	assert_cmp(ssize_t, a, b, >=,	\
+    <, "zd", __VA_ARGS__)
+#define assert_zd_gt(a, b, ...)	assert_cmp(ssize_t, a, b, >,	\
+    <=, "zd", __VA_ARGS__)
+
+#define assert_zu_eq(a, b, ...)	assert_cmp(size_t, a, b, ==,	\
+    !=, "zu", __VA_ARGS__)
+#define assert_zu_ne(a, b, ...)	assert_cmp(size_t, a, b, !=,	\
+    ==, "zu", __VA_ARGS__)
+#define assert_zu_lt(a, b, ...)	assert_cmp(size_t, a, b, <,	\
+    >=, "zu", __VA_ARGS__)
+#define assert_zu_le(a, b, ...)	assert_cmp(size_t, a, b, <=,	\
+    >, "zu", __VA_ARGS__)
+#define assert_zu_ge(a, b, ...)	assert_cmp(size_t, a, b, >=,	\
+    <, "zu", __VA_ARGS__)
+#define assert_zu_gt(a, b, ...)	assert_cmp(size_t, a, b, >,	\
+    <=, "zu", __VA_ARGS__)
+
+#define assert_d32_eq(a, b, ...)	assert_cmp(int32_t, a, b, ==,	\
+    !=, FMTd32, __VA_ARGS__)
+#define assert_d32_ne(a, b, ...)	assert_cmp(int32_t, a, b, !=,	\
+    ==, FMTd32, __VA_ARGS__)
+#define assert_d32_lt(a, b, ...)	assert_cmp(int32_t, a, b, <,	\
+    >=, FMTd32, __VA_ARGS__)
+#define assert_d32_le(a, b, ...)	assert_cmp(int32_t, a, b, <=,	\
+    >, FMTd32, __VA_ARGS__)
+#define assert_d32_ge(a, b, ...)	assert_cmp(int32_t, a, b, >=,	\
+    <, FMTd32, __VA_ARGS__)
+#define assert_d32_gt(a, b, ...)	assert_cmp(int32_t, a, b, >,	\
+    <=, FMTd32, __VA_ARGS__)
+
+#define assert_u32_eq(a, b, ...)	assert_cmp(uint32_t, a, b, ==,	\
+    !=, FMTu32, __VA_ARGS__)
+#define assert_u32_ne(a, b, ...)	assert_cmp(uint32_t, a, b, !=,	\
+    ==, FMTu32, __VA_ARGS__)
+#define assert_u32_lt(a, b, ...)	assert_cmp(uint32_t, a, b, <,	\
+    >=, FMTu32, __VA_ARGS__)
+#define assert_u32_le(a, b, ...)	assert_cmp(uint32_t, a, b, <=,	\
+    >, FMTu32, __VA_ARGS__)
+#define assert_u32_ge(a, b, ...)	assert_cmp(uint32_t, a, b, >=,	\
+    <, FMTu32, __VA_ARGS__)
+#define assert_u32_gt(a, b, ...)	assert_cmp(uint32_t, a, b, >,	\
+    <=, FMTu32, __VA_ARGS__)
+
+#define assert_d64_eq(a, b, ...)	assert_cmp(int64_t, a, b, ==,	\
+    !=, FMTd64, __VA_ARGS__)
+#define assert_d64_ne(a, b, ...)	assert_cmp(int64_t, a, b, !=,	\
+    ==, FMTd64, __VA_ARGS__)
+#define assert_d64_lt(a, b, ...)	assert_cmp(int64_t, a, b, <,	\
+    >=, FMTd64, __VA_ARGS__)
+#define assert_d64_le(a, b, ...)	assert_cmp(int64_t, a, b, <=,	\
+    >, FMTd64, __VA_ARGS__)
+#define assert_d64_ge(a, b, ...)	assert_cmp(int64_t, a, b, >=,	\
+    <, FMTd64, __VA_ARGS__)
+#define assert_d64_gt(a, b, ...)	assert_cmp(int64_t, a, b, >,	\
+    <=, FMTd64, __VA_ARGS__)
+
+#define assert_u64_eq(a, b, ...)	assert_cmp(uint64_t, a, b, ==,	\
+    !=, FMTu64, __VA_ARGS__)
+#define assert_u64_ne(a, b, ...)	assert_cmp(uint64_t, a, b, !=,	\
+    ==, FMTu64, __VA_ARGS__)
+#define assert_u64_lt(a, b, ...)	assert_cmp(uint64_t, a, b, <,	\
+    >=, FMTu64, __VA_ARGS__)
+#define assert_u64_le(a, b, ...)	assert_cmp(uint64_t, a, b, <=,	\
+    >, FMTu64, __VA_ARGS__)
+#define assert_u64_ge(a, b, ...)	assert_cmp(uint64_t, a, b, >=,	\
+    <, FMTu64, __VA_ARGS__)
+#define assert_u64_gt(a, b, ...)	assert_cmp(uint64_t, a, b, >,	\
+    <=, FMTu64, __VA_ARGS__)
+
+#define assert_b_eq(a, b, ...) do {					\
+	bool a_ = (a);							\
+	bool b_ = (b);							\
+	if (!(a_ == b_)) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) == (%s) --> %s != %s: ",			\
+		    __func__, __FILE__, __LINE__,			\
+		    #a, #b, a_ ? "true" : "false",			\
+		    b_ ? "true" : "false");				\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+#define assert_b_ne(a, b, ...) do {					\
+	bool a_ = (a);							\
+	bool b_ = (b);							\
+	if (!(a_ != b_)) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) != (%s) --> %s == %s: ",			\
+		    __func__, __FILE__, __LINE__,			\
+		    #a, #b, a_ ? "true" : "false",			\
+		    b_ ? "true" : "false");				\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+#define assert_true(a, ...)	assert_b_eq(a, true, __VA_ARGS__)
+#define assert_false(a, ...)	assert_b_eq(a, false, __VA_ARGS__)
+
+#define assert_str_eq(a, b, ...) do {				\
+	if (strcmp((a), (b))) {						\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) same as (%s) --> "				\
+		    "\"%s\" differs from \"%s\": ",			\
+		    __func__, __FILE__, __LINE__, #a, #b, a, b);	\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+#define assert_str_ne(a, b, ...) do {				\
+	if (!strcmp((a), (b))) {					\
+		char prefix[ASSERT_BUFSIZE];				\
+		char message[ASSERT_BUFSIZE];				\
+		malloc_snprintf(prefix, sizeof(prefix),			\
+		    "%s:%s:%d: Failed assertion: "			\
+		    "(%s) differs from (%s) --> "			\
+		    "\"%s\" same as \"%s\": ",				\
+		    __func__, __FILE__, __LINE__, #a, #b, a, b);	\
+		malloc_snprintf(message, sizeof(message), __VA_ARGS__);	\
+		p_test_fail(prefix, message);				\
+	}								\
+} while (0)
+
+#define assert_not_reached(...) do {					\
+	char prefix[ASSERT_BUFSIZE];					\
+	char message[ASSERT_BUFSIZE];					\
+	malloc_snprintf(prefix, sizeof(prefix),				\
+	    "%s:%s:%d: Unreachable code reached: ",			\
+	    __func__, __FILE__, __LINE__);				\
+	malloc_snprintf(message, sizeof(message), __VA_ARGS__);		\
+	p_test_fail(prefix, message);					\
+} while (0)
+
+/*
+ * If this enum changes, corresponding changes in test/test.sh.in are also
+ * necessary.
+ */
+typedef enum {
+	test_status_pass = 0,
+	test_status_skip = 1,
+	test_status_fail = 2,
+
+	test_status_count = 3
+} test_status_t;
+
+typedef void (test_t)(void);
+
+#define TEST_BEGIN(f)							\
+static void								\
+f(void) {								\
+	p_test_init(#f);
+
+#define TEST_END							\
+	goto label_test_end;						\
+label_test_end:								\
+	p_test_fini();							\
+}
+
+#define test(...)							\
+	p_test(__VA_ARGS__, NULL)
+
+#define test_no_reentrancy(...)							\
+	p_test_no_reentrancy(__VA_ARGS__, NULL)
+
+#define test_no_malloc_init(...)					\
+	p_test_no_malloc_init(__VA_ARGS__, NULL)
+
+#define test_skip_if(e) do {						\
+	if (e) {							\
+		test_skip("%s:%s:%d: Test skipped: (%s)",		\
+		    __func__, __FILE__, __LINE__, #e);			\
+		goto label_test_end;					\
+	}								\
+} while (0)
+
+bool test_is_reentrant();
+
+void	test_skip(const char *format, ...) JEMALLOC_FORMAT_PRINTF(1, 2);
+void	test_fail(const char *format, ...) JEMALLOC_FORMAT_PRINTF(1, 2);
+
+/* For private use by macros. */
+test_status_t	p_test(test_t *t, ...);
+test_status_t	p_test_no_reentrancy(test_t *t, ...);
+test_status_t	p_test_no_malloc_init(test_t *t, ...);
+void	p_test_init(const char *name);
+void	p_test_fini(void);
+void	p_test_fail(const char *prefix, const char *message);
diff --git a/memkind-1.10.0/jemalloc/test/include/test/thd.h b/memkind-1.10.0/jemalloc/test/include/test/thd.h
new file mode 100644
index 0000000..47a5126
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/thd.h
@@ -0,0 +1,9 @@
+/* Abstraction layer for threading in tests. */
+#ifdef _WIN32
+typedef HANDLE thd_t;
+#else
+typedef pthread_t thd_t;
+#endif
+
+void	thd_create(thd_t *thd, void *(*proc)(void *), void *arg);
+void	thd_join(thd_t thd, void **ret);
diff --git a/memkind-1.10.0/jemalloc/test/include/test/timer.h b/memkind-1.10.0/jemalloc/test/include/test/timer.h
new file mode 100644
index 0000000..ace6191
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/include/test/timer.h
@@ -0,0 +1,11 @@
+/* Simple timer, for use in benchmark reporting. */
+
+typedef struct {
+	nstime_t t0;
+	nstime_t t1;
+} timedelta_t;
+
+void	timer_start(timedelta_t *timer);
+void	timer_stop(timedelta_t *timer);
+uint64_t	timer_usec(const timedelta_t *timer);
+void	timer_ratio(timedelta_t *a, timedelta_t *b, char *buf, size_t buflen);
diff --git a/memkind-1.10.0/jemalloc/test/integration/MALLOCX_ARENA.c b/memkind-1.10.0/jemalloc/test/integration/MALLOCX_ARENA.c
new file mode 100644
index 0000000..222164d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/MALLOCX_ARENA.c
@@ -0,0 +1,66 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS 10
+
+static bool have_dss =
+#ifdef JEMALLOC_DSS
+    true
+#else
+    false
+#endif
+    ;
+
+void *
+thd_start(void *arg) {
+	unsigned thread_ind = (unsigned)(uintptr_t)arg;
+	unsigned arena_ind;
+	void *p;
+	size_t sz;
+
+	sz = sizeof(arena_ind);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Error in arenas.create");
+
+	if (thread_ind % 4 != 3) {
+		size_t mib[3];
+		size_t miblen = sizeof(mib) / sizeof(size_t);
+		const char *dss_precs[] = {"disabled", "primary", "secondary"};
+		unsigned prec_ind = thread_ind %
+		    (sizeof(dss_precs)/sizeof(char*));
+		const char *dss = dss_precs[prec_ind];
+		int expected_err = (have_dss || prec_ind == 0) ? 0 : EFAULT;
+		assert_d_eq(mallctlnametomib("arena.0.dss", mib, &miblen), 0,
+		    "Error in mallctlnametomib()");
+		mib[1] = arena_ind;
+		assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, (void *)&dss,
+		    sizeof(const char *)), expected_err,
+		    "Error in mallctlbymib()");
+	}
+
+	p = mallocx(1, MALLOCX_ARENA(arena_ind));
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	dallocx(p, 0);
+
+	return NULL;
+}
+
+TEST_BEGIN(test_MALLOCX_ARENA) {
+	thd_t thds[NTHREADS];
+	unsigned i;
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start,
+		    (void *)(uintptr_t)i);
+	}
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_MALLOCX_ARENA);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/aligned_alloc.c b/memkind-1.10.0/jemalloc/test/integration/aligned_alloc.c
new file mode 100644
index 0000000..4375b17
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/aligned_alloc.c
@@ -0,0 +1,157 @@
+#include "test/jemalloc_test.h"
+
+#define MAXALIGN (((size_t)1) << 23)
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+TEST_BEGIN(test_alignment_errors) {
+	size_t alignment;
+	void *p;
+
+	alignment = 0;
+	set_errno(0);
+	p = aligned_alloc(alignment, 1);
+	assert_false(p != NULL || get_errno() != EINVAL,
+	    "Expected error for invalid alignment %zu", alignment);
+
+	for (alignment = sizeof(size_t); alignment < MAXALIGN;
+	    alignment <<= 1) {
+		set_errno(0);
+		p = aligned_alloc(alignment + 1, 1);
+		assert_false(p != NULL || get_errno() != EINVAL,
+		    "Expected error for invalid alignment %zu",
+		    alignment + 1);
+	}
+}
+TEST_END
+
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_oom_errors) {
+	size_t alignment, size;
+	void *p;
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x8000000000000000);
+	size      = UINT64_C(0x8000000000000000);
+#else
+	alignment = 0x80000000LU;
+	size      = 0x80000000LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	assert_false(p != NULL || get_errno() != ENOMEM,
+	    "Expected error for aligned_alloc(%zu, %zu)",
+	    alignment, size);
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x4000000000000000);
+	size      = UINT64_C(0xc000000000000001);
+#else
+	alignment = 0x40000000LU;
+	size      = 0xc0000001LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	assert_false(p != NULL || get_errno() != ENOMEM,
+	    "Expected error for aligned_alloc(%zu, %zu)",
+	    alignment, size);
+
+	alignment = 0x10LU;
+#if LG_SIZEOF_PTR == 3
+	size = UINT64_C(0xfffffffffffffff0);
+#else
+	size = 0xfffffff0LU;
+#endif
+	set_errno(0);
+	p = aligned_alloc(alignment, size);
+	assert_false(p != NULL || get_errno() != ENOMEM,
+	    "Expected error for aligned_alloc(&p, %zu, %zu)",
+	    alignment, size);
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+TEST_BEGIN(test_alignment_and_size) {
+#define NITER 4
+	size_t alignment, size, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (size = 1;
+		    size < 3 * alignment && size < (1U << 31);
+		    size += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				ps[i] = aligned_alloc(alignment, size);
+				if (ps[i] == NULL) {
+					char buf[BUFERROR_BUF];
+
+					buferror(get_errno(), buf, sizeof(buf));
+					test_fail(
+					    "Error for alignment=%zu, "
+					    "size=%zu (%#zx): %s",
+					    alignment, size, size, buf);
+				}
+				total += malloc_usable_size(ps[i]);
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					free(ps[i]);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef NITER
+}
+TEST_END
+
+TEST_BEGIN(test_zero_alloc) {
+	void *res = aligned_alloc(8, 0);
+	assert(res);
+	size_t usable = malloc_usable_size(res);
+	assert(usable > 0);
+	free(res);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_alignment_errors,
+	    test_oom_errors,
+	    test_alignment_and_size,
+	    test_zero_alloc);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/allocated.c b/memkind-1.10.0/jemalloc/test/integration/allocated.c
new file mode 100644
index 0000000..1425fd0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/allocated.c
@@ -0,0 +1,124 @@
+#include "test/jemalloc_test.h"
+
+static const bool config_stats =
+#ifdef JEMALLOC_STATS
+    true
+#else
+    false
+#endif
+    ;
+
+void *
+thd_start(void *arg) {
+	int err;
+	void *p;
+	uint64_t a0, a1, d0, d1;
+	uint64_t *ap0, *ap1, *dp0, *dp1;
+	size_t sz, usize;
+
+	sz = sizeof(a0);
+	if ((err = mallctl("thread.allocated", (void *)&a0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	sz = sizeof(ap0);
+	if ((err = mallctl("thread.allocatedp", (void *)&ap0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	assert_u64_eq(*ap0, a0,
+	    "\"thread.allocatedp\" should provide a pointer to internal "
+	    "storage");
+
+	sz = sizeof(d0);
+	if ((err = mallctl("thread.deallocated", (void *)&d0, &sz, NULL, 0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	sz = sizeof(dp0);
+	if ((err = mallctl("thread.deallocatedp", (void *)&dp0, &sz, NULL,
+	    0))) {
+		if (err == ENOENT) {
+			goto label_ENOENT;
+		}
+		test_fail("%s(): Error in mallctl(): %s", __func__,
+		    strerror(err));
+	}
+	assert_u64_eq(*dp0, d0,
+	    "\"thread.deallocatedp\" should provide a pointer to internal "
+	    "storage");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() error");
+
+	sz = sizeof(a1);
+	mallctl("thread.allocated", (void *)&a1, &sz, NULL, 0);
+	sz = sizeof(ap1);
+	mallctl("thread.allocatedp", (void *)&ap1, &sz, NULL, 0);
+	assert_u64_eq(*ap1, a1,
+	    "Dereferenced \"thread.allocatedp\" value should equal "
+	    "\"thread.allocated\" value");
+	assert_ptr_eq(ap0, ap1,
+	    "Pointer returned by \"thread.allocatedp\" should not change");
+
+	usize = malloc_usable_size(p);
+	assert_u64_le(a0 + usize, a1,
+	    "Allocated memory counter should increase by at least the amount "
+	    "explicitly allocated");
+
+	free(p);
+
+	sz = sizeof(d1);
+	mallctl("thread.deallocated", (void *)&d1, &sz, NULL, 0);
+	sz = sizeof(dp1);
+	mallctl("thread.deallocatedp", (void *)&dp1, &sz, NULL, 0);
+	assert_u64_eq(*dp1, d1,
+	    "Dereferenced \"thread.deallocatedp\" value should equal "
+	    "\"thread.deallocated\" value");
+	assert_ptr_eq(dp0, dp1,
+	    "Pointer returned by \"thread.deallocatedp\" should not change");
+
+	assert_u64_le(d0 + usize, d1,
+	    "Deallocated memory counter should increase by at least the amount "
+	    "explicitly deallocated");
+
+	return NULL;
+label_ENOENT:
+	assert_false(config_stats,
+	    "ENOENT should only be returned if stats are disabled");
+	test_skip("\"thread.allocated\" mallctl not available");
+	return NULL;
+}
+
+TEST_BEGIN(test_main_thread) {
+	thd_start(NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_subthread) {
+	thd_t thd;
+
+	thd_create(&thd, thd_start, NULL);
+	thd_join(thd, NULL);
+}
+TEST_END
+
+int
+main(void) {
+	/* Run tests multiple times to check for bad interactions. */
+	return test(
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/extent.c b/memkind-1.10.0/jemalloc/test/integration/extent.c
new file mode 100644
index 0000000..b5db087
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/extent.c
@@ -0,0 +1,248 @@
+#include "test/jemalloc_test.h"
+
+#include "test/extent_hooks.h"
+
+static bool
+check_background_thread_enabled(void) {
+	bool enabled;
+	size_t sz = sizeof(bool);
+	int ret = mallctl("background_thread", (void *)&enabled, &sz, NULL,0);
+	if (ret == ENOENT) {
+		return false;
+	}
+	assert_d_eq(ret, 0, "Unexpected mallctl error");
+	return enabled;
+}
+
+static void
+test_extent_body(unsigned arena_ind) {
+	void *p;
+	size_t large0, large1, large2, sz;
+	size_t purge_mib[3];
+	size_t purge_miblen;
+	int flags;
+	bool xallocx_success_a, xallocx_success_b, xallocx_success_c;
+
+	flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+
+	/* Get large size classes. */
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large0, &sz, NULL,
+	    0), 0, "Unexpected arenas.lextent.0.size failure");
+	assert_d_eq(mallctl("arenas.lextent.1.size", (void *)&large1, &sz, NULL,
+	    0), 0, "Unexpected arenas.lextent.1.size failure");
+	assert_d_eq(mallctl("arenas.lextent.2.size", (void *)&large2, &sz, NULL,
+	    0), 0, "Unexpected arenas.lextent.2.size failure");
+
+	/* Test dalloc/decommit/purge cascade. */
+	purge_miblen = sizeof(purge_mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.purge", purge_mib, &purge_miblen),
+	    0, "Unexpected mallctlnametomib() failure");
+	purge_mib[1] = (size_t)arena_ind;
+	called_alloc = false;
+	try_alloc = true;
+	try_dalloc = false;
+	try_decommit = false;
+	p = mallocx(large0 * 2, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	assert_true(called_alloc, "Expected alloc call");
+	called_dalloc = false;
+	called_decommit = false;
+	did_purge_lazy = false;
+	did_purge_forced = false;
+	called_split = false;
+	xallocx_success_a = (xallocx(p, large0, 0, flags) == large0);
+	assert_d_eq(mallctlbymib(purge_mib, purge_miblen, NULL, NULL, NULL, 0),
+	    0, "Unexpected arena.%u.purge error", arena_ind);
+	if (xallocx_success_a) {
+		assert_true(called_dalloc, "Expected dalloc call");
+		assert_true(called_decommit, "Expected decommit call");
+		assert_true(did_purge_lazy || did_purge_forced,
+		    "Expected purge");
+	}
+	assert_true(called_split, "Expected split call");
+	dallocx(p, flags);
+	try_dalloc = true;
+
+	/* Test decommit/commit and observe split/merge. */
+	try_dalloc = false;
+	try_decommit = true;
+	p = mallocx(large0 * 2, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	did_decommit = false;
+	did_commit = false;
+	called_split = false;
+	did_split = false;
+	did_merge = false;
+	xallocx_success_b = (xallocx(p, large0, 0, flags) == large0);
+	assert_d_eq(mallctlbymib(purge_mib, purge_miblen, NULL, NULL, NULL, 0),
+	    0, "Unexpected arena.%u.purge error", arena_ind);
+	if (xallocx_success_b) {
+		assert_true(did_split, "Expected split");
+	}
+	xallocx_success_c = (xallocx(p, large0 * 2, 0, flags) == large0 * 2);
+	if (did_split) {
+		assert_b_eq(did_decommit, did_commit,
+		    "Expected decommit/commit match");
+	}
+	if (xallocx_success_b && xallocx_success_c) {
+		assert_true(did_merge, "Expected merge");
+	}
+	dallocx(p, flags);
+	try_dalloc = true;
+	try_decommit = false;
+
+	/* Make sure non-large allocation succeeds. */
+	p = mallocx(42, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	dallocx(p, flags);
+}
+
+static void
+test_manual_hook_auto_arena(void) {
+	unsigned narenas;
+	size_t old_size, new_size, sz;
+	size_t hooks_mib[3];
+	size_t hooks_miblen;
+	extent_hooks_t *new_hooks, *old_hooks;
+
+	extent_hooks_prep();
+
+	sz = sizeof(unsigned);
+	/* Get number of auto arenas. */
+	assert_d_eq(mallctl("opt.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	if (narenas == 1) {
+		return;
+	}
+
+	/* Install custom extent hooks on arena 1 (might not be initialized). */
+	hooks_miblen = sizeof(hooks_mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.extent_hooks", hooks_mib,
+	    &hooks_miblen), 0, "Unexpected mallctlnametomib() failure");
+	hooks_mib[1] = 1;
+	old_size = sizeof(extent_hooks_t *);
+	new_hooks = &hooks;
+	new_size = sizeof(extent_hooks_t *);
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, (void *)&old_hooks,
+	    &old_size, (void *)&new_hooks, new_size), 0,
+	    "Unexpected extent_hooks error");
+	static bool auto_arena_created = false;
+	if (old_hooks != &hooks) {
+		assert_b_eq(auto_arena_created, false,
+		    "Expected auto arena 1 created only once.");
+		auto_arena_created = true;
+	}
+}
+
+static void
+test_manual_hook_body(void) {
+	unsigned arena_ind;
+	size_t old_size, new_size, sz;
+	size_t hooks_mib[3];
+	size_t hooks_miblen;
+	extent_hooks_t *new_hooks, *old_hooks;
+
+	extent_hooks_prep();
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+
+	/* Install custom extent hooks. */
+	hooks_miblen = sizeof(hooks_mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.extent_hooks", hooks_mib,
+	    &hooks_miblen), 0, "Unexpected mallctlnametomib() failure");
+	hooks_mib[1] = (size_t)arena_ind;
+	old_size = sizeof(extent_hooks_t *);
+	new_hooks = &hooks;
+	new_size = sizeof(extent_hooks_t *);
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, (void *)&old_hooks,
+	    &old_size, (void *)&new_hooks, new_size), 0,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->alloc, extent_alloc_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->dalloc, extent_dalloc_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->commit, extent_commit_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->decommit, extent_decommit_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->purge_lazy, extent_purge_lazy_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->purge_forced, extent_purge_forced_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->split, extent_split_hook,
+	    "Unexpected extent_hooks error");
+	assert_ptr_ne(old_hooks->merge, extent_merge_hook,
+	    "Unexpected extent_hooks error");
+
+	if (!check_background_thread_enabled()) {
+		test_extent_body(arena_ind);
+	}
+
+	/* Restore extent hooks. */
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, NULL, NULL,
+	    (void *)&old_hooks, new_size), 0, "Unexpected extent_hooks error");
+	assert_d_eq(mallctlbymib(hooks_mib, hooks_miblen, (void *)&old_hooks,
+	    &old_size, NULL, 0), 0, "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks, default_hooks, "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->alloc, default_hooks->alloc,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->dalloc, default_hooks->dalloc,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->commit, default_hooks->commit,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->decommit, default_hooks->decommit,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->purge_lazy, default_hooks->purge_lazy,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->purge_forced, default_hooks->purge_forced,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->split, default_hooks->split,
+	    "Unexpected extent_hooks error");
+	assert_ptr_eq(old_hooks->merge, default_hooks->merge,
+	    "Unexpected extent_hooks error");
+}
+
+TEST_BEGIN(test_extent_manual_hook) {
+	test_manual_hook_auto_arena();
+	test_manual_hook_body();
+
+	/* Test failure paths. */
+	try_split = false;
+	test_manual_hook_body();
+	try_merge = false;
+	test_manual_hook_body();
+	try_purge_lazy = false;
+	try_purge_forced = false;
+	test_manual_hook_body();
+
+	try_split = try_merge = try_purge_lazy = try_purge_forced = true;
+}
+TEST_END
+
+TEST_BEGIN(test_extent_auto_hook) {
+	unsigned arena_ind;
+	size_t new_size, sz;
+	extent_hooks_t *new_hooks;
+
+	extent_hooks_prep();
+
+	sz = sizeof(unsigned);
+	new_hooks = &hooks;
+	new_size = sizeof(extent_hooks_t *);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz,
+	    (void *)&new_hooks, new_size), 0, "Unexpected mallctl() failure");
+
+	test_skip_if(check_background_thread_enabled());
+	test_extent_body(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_extent_manual_hook,
+	    test_extent_auto_hook);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/extent.sh b/memkind-1.10.0/jemalloc/test/integration/extent.sh
new file mode 100644
index 0000000..0cc2187
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/extent.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="junk:false"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/integration/malloc.c b/memkind-1.10.0/jemalloc/test/integration/malloc.c
new file mode 100644
index 0000000..8b33bc8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/malloc.c
@@ -0,0 +1,16 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_zero_alloc) {
+	void *res = malloc(0);
+	assert(res);
+	size_t usable = malloc_usable_size(res);
+	assert(usable > 0);
+	free(res);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_zero_alloc);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/mallocx.c b/memkind-1.10.0/jemalloc/test/integration/mallocx.c
new file mode 100644
index 0000000..645d4db
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/mallocx.c
@@ -0,0 +1,274 @@
+#include "test/jemalloc_test.h"
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	size_t largemax;
+
+	largemax = get_large_size(get_nlarge()-1);
+
+	assert_ptr_null(mallocx(largemax+1, 0),
+	    "Expected OOM for mallocx(size=%#zx, 0)", largemax+1);
+
+	assert_ptr_null(mallocx(ZU(PTRDIFF_MAX)+1, 0),
+	    "Expected OOM for mallocx(size=%#zx, 0)", ZU(PTRDIFF_MAX)+1);
+
+	assert_ptr_null(mallocx(SIZE_T_MAX, 0),
+	    "Expected OOM for mallocx(size=%#zx, 0)", SIZE_T_MAX);
+
+	assert_ptr_null(mallocx(1, MALLOCX_ALIGN(ZU(PTRDIFF_MAX)+1)),
+	    "Expected OOM for mallocx(size=1, MALLOCX_ALIGN(%#zx))",
+	    ZU(PTRDIFF_MAX)+1);
+}
+TEST_END
+
+static void *
+remote_alloc(void *arg) {
+	unsigned arena;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	size_t large_sz;
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large_sz, &sz,
+	    NULL, 0), 0, "Unexpected mallctl failure");
+
+	void *ptr = mallocx(large_sz, MALLOCX_ARENA(arena)
+	    | MALLOCX_TCACHE_NONE);
+	void **ret = (void **)arg;
+	*ret = ptr;
+
+	return NULL;
+}
+
+TEST_BEGIN(test_remote_free) {
+	thd_t thd;
+	void *ret;
+	thd_create(&thd, remote_alloc, (void *)&ret);
+	thd_join(thd, NULL);
+	assert_ptr_not_null(ret, "Unexpected mallocx failure");
+
+	/* Avoid TCACHE_NONE to explicitly test tcache_flush(). */
+	dallocx(ret, 0);
+	mallctl("thread.tcache.flush", NULL, NULL, NULL, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_oom) {
+	size_t largemax;
+	bool oom;
+	void *ptrs[3];
+	unsigned i;
+
+	/*
+	 * It should be impossible to allocate three objects that each consume
+	 * nearly half the virtual address space.
+	 */
+	largemax = get_large_size(get_nlarge()-1);
+	oom = false;
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		ptrs[i] = mallocx(largemax, MALLOCX_ARENA(0));
+		if (ptrs[i] == NULL) {
+			oom = true;
+		}
+	}
+	assert_true(oom,
+	    "Expected OOM during series of calls to mallocx(size=%zu, 0)",
+	    largemax);
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		if (ptrs[i] != NULL) {
+			dallocx(ptrs[i], 0);
+		}
+	}
+	purge();
+
+#if LG_SIZEOF_PTR == 3
+	assert_ptr_null(mallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x8000000000000000ULL)),
+	    "Expected OOM for mallocx()");
+	assert_ptr_null(mallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x80000000)),
+	    "Expected OOM for mallocx()");
+#else
+	assert_ptr_null(mallocx(0x80000000UL, MALLOCX_ALIGN(0x80000000UL)),
+	    "Expected OOM for mallocx()");
+#endif
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+TEST_BEGIN(test_basic) {
+#define MAXSZ (((size_t)1) << 23)
+	size_t sz;
+
+	for (sz = 1; sz < MAXSZ; sz = nallocx(sz, 0) + 1) {
+		size_t nsz, rsz;
+		void *p;
+		nsz = nallocx(sz, 0);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		p = mallocx(sz, 0);
+		assert_ptr_not_null(p,
+		    "Unexpected mallocx(size=%zx, flags=0) error", sz);
+		rsz = sallocx(p, 0);
+		assert_zu_ge(rsz, sz, "Real size smaller than expected");
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() size mismatch");
+		dallocx(p, 0);
+
+		p = mallocx(sz, 0);
+		assert_ptr_not_null(p,
+		    "Unexpected mallocx(size=%zx, flags=0) error", sz);
+		dallocx(p, 0);
+
+		nsz = nallocx(sz, MALLOCX_ZERO);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		p = mallocx(sz, MALLOCX_ZERO);
+		assert_ptr_not_null(p,
+		    "Unexpected mallocx(size=%zx, flags=MALLOCX_ZERO) error",
+		    nsz);
+		rsz = sallocx(p, 0);
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() rsize mismatch");
+		dallocx(p, 0);
+		purge();
+	}
+#undef MAXSZ
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+	const char *percpu_arena;
+	size_t sz = sizeof(percpu_arena);
+
+	if(mallctl("opt.percpu_arena", (void *)&percpu_arena, &sz, NULL, 0) ||
+	    strcmp(percpu_arena, "disabled") != 0) {
+		test_skip("test_alignment_and_size skipped: "
+		    "not working with percpu arena.");
+	};
+#define MAXALIGN (((size_t)1) << 23)
+#define NITER 4
+	size_t nsz, rsz, alignment, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (sz = 1;
+		    sz < 3 * alignment && sz < (1U << 31);
+		    sz += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				nsz = nallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO | MALLOCX_ARENA(0));
+				assert_zu_ne(nsz, 0,
+				    "nallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				ps[i] = mallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO | MALLOCX_ARENA(0));
+				assert_ptr_not_null(ps[i],
+				    "mallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				rsz = sallocx(ps[i], 0);
+				assert_zu_ge(rsz, sz,
+				    "Real size smaller than expected for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_zu_eq(nsz, rsz,
+				    "nallocx()/sallocx() size mismatch for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_ptr_null(
+				    (void *)((uintptr_t)ps[i] & (alignment-1)),
+				    "%p inadequately aligned for"
+				    " alignment=%zu, size=%zu", ps[i],
+				    alignment, sz);
+				total += rsz;
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					dallocx(ps[i], 0);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef MAXALIGN
+#undef NITER
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_overflow,
+	    test_oom,
+	    test_remote_free,
+	    test_basic,
+	    test_alignment_and_size);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/mallocx.sh b/memkind-1.10.0/jemalloc/test/integration/mallocx.sh
new file mode 100644
index 0000000..0cc2187
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/mallocx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="junk:false"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/integration/overflow.c b/memkind-1.10.0/jemalloc/test/integration/overflow.c
new file mode 100644
index 0000000..748ebb6
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/overflow.c
@@ -0,0 +1,59 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	unsigned nlextents;
+	size_t mib[4];
+	size_t sz, miblen, max_size_class;
+	void *p;
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nlextents", (void *)&nlextents, &sz, NULL,
+	    0), 0, "Unexpected mallctl() error");
+
+	miblen = sizeof(mib) / sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arenas.lextent.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() error");
+	mib[2] = nlextents - 1;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&max_size_class, &sz,
+	    NULL, 0), 0, "Unexpected mallctlbymib() error");
+
+	assert_ptr_null(malloc(max_size_class + 1),
+	    "Expected OOM due to over-sized allocation request");
+	assert_ptr_null(malloc(SIZE_T_MAX),
+	    "Expected OOM due to over-sized allocation request");
+
+	assert_ptr_null(calloc(1, max_size_class + 1),
+	    "Expected OOM due to over-sized allocation request");
+	assert_ptr_null(calloc(1, SIZE_T_MAX),
+	    "Expected OOM due to over-sized allocation request");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() OOM");
+	assert_ptr_null(realloc(p, max_size_class + 1),
+	    "Expected OOM due to over-sized allocation request");
+	assert_ptr_null(realloc(p, SIZE_T_MAX),
+	    "Expected OOM due to over-sized allocation request");
+	free(p);
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+int
+main(void) {
+	return test(
+	    test_overflow);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/posix_memalign.c b/memkind-1.10.0/jemalloc/test/integration/posix_memalign.c
new file mode 100644
index 0000000..d992260
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/posix_memalign.c
@@ -0,0 +1,128 @@
+#include "test/jemalloc_test.h"
+
+#define MAXALIGN (((size_t)1) << 23)
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+TEST_BEGIN(test_alignment_errors) {
+	size_t alignment;
+	void *p;
+
+	for (alignment = 0; alignment < sizeof(void *); alignment++) {
+		assert_d_eq(posix_memalign(&p, alignment, 1), EINVAL,
+		    "Expected error for invalid alignment %zu",
+		    alignment);
+	}
+
+	for (alignment = sizeof(size_t); alignment < MAXALIGN;
+	    alignment <<= 1) {
+		assert_d_ne(posix_memalign(&p, alignment + 1, 1), 0,
+		    "Expected error for invalid alignment %zu",
+		    alignment + 1);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_oom_errors) {
+	size_t alignment, size;
+	void *p;
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x8000000000000000);
+	size      = UINT64_C(0x8000000000000000);
+#else
+	alignment = 0x80000000LU;
+	size      = 0x80000000LU;
+#endif
+	assert_d_ne(posix_memalign(&p, alignment, size), 0,
+	    "Expected error for posix_memalign(&p, %zu, %zu)",
+	    alignment, size);
+
+#if LG_SIZEOF_PTR == 3
+	alignment = UINT64_C(0x4000000000000000);
+	size      = UINT64_C(0xc000000000000001);
+#else
+	alignment = 0x40000000LU;
+	size      = 0xc0000001LU;
+#endif
+	assert_d_ne(posix_memalign(&p, alignment, size), 0,
+	    "Expected error for posix_memalign(&p, %zu, %zu)",
+	    alignment, size);
+
+	alignment = 0x10LU;
+#if LG_SIZEOF_PTR == 3
+	size = UINT64_C(0xfffffffffffffff0);
+#else
+	size = 0xfffffff0LU;
+#endif
+	assert_d_ne(posix_memalign(&p, alignment, size), 0,
+	    "Expected error for posix_memalign(&p, %zu, %zu)",
+	    alignment, size);
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+#define NITER 4
+	size_t alignment, size, total;
+	unsigned i;
+	int err;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (size = 0;
+		    size < 3 * alignment && size < (1U << 31);
+		    size += ((size == 0) ? 1 :
+		    (alignment >> (LG_SIZEOF_PTR-1)) - 1)) {
+			for (i = 0; i < NITER; i++) {
+				err = posix_memalign(&ps[i],
+				    alignment, size);
+				if (err) {
+					char buf[BUFERROR_BUF];
+
+					buferror(get_errno(), buf, sizeof(buf));
+					test_fail(
+					    "Error for alignment=%zu, "
+					    "size=%zu (%#zx): %s",
+					    alignment, size, size, buf);
+				}
+				total += malloc_usable_size(ps[i]);
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					free(ps[i]);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef NITER
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_alignment_errors,
+	    test_oom_errors,
+	    test_alignment_and_size);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/rallocx.c b/memkind-1.10.0/jemalloc/test/integration/rallocx.c
new file mode 100644
index 0000000..08ed08d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/rallocx.c
@@ -0,0 +1,258 @@
+#include "test/jemalloc_test.h"
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+TEST_BEGIN(test_grow_and_shrink) {
+	void *p, *q;
+	size_t tsz;
+#define NCYCLES 3
+	unsigned i, j;
+#define NSZS 1024
+	size_t szs[NSZS];
+#define MAXSZ ZU(12 * 1024 * 1024)
+
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	szs[0] = sallocx(p, 0);
+
+	for (i = 0; i < NCYCLES; i++) {
+		for (j = 1; j < NSZS && szs[j-1] < MAXSZ; j++) {
+			q = rallocx(p, szs[j-1]+1, 0);
+			assert_ptr_not_null(q,
+			    "Unexpected rallocx() error for size=%zu-->%zu",
+			    szs[j-1], szs[j-1]+1);
+			szs[j] = sallocx(q, 0);
+			assert_zu_ne(szs[j], szs[j-1]+1,
+			    "Expected size to be at least: %zu", szs[j-1]+1);
+			p = q;
+		}
+
+		for (j--; j > 0; j--) {
+			q = rallocx(p, szs[j-1], 0);
+			assert_ptr_not_null(q,
+			    "Unexpected rallocx() error for size=%zu-->%zu",
+			    szs[j], szs[j-1]);
+			tsz = sallocx(q, 0);
+			assert_zu_eq(tsz, szs[j-1],
+			    "Expected size=%zu, got size=%zu", szs[j-1], tsz);
+			p = q;
+		}
+	}
+
+	dallocx(p, 0);
+#undef MAXSZ
+#undef NSZS
+#undef NCYCLES
+}
+TEST_END
+
+static bool
+validate_fill(const void *p, uint8_t c, size_t offset, size_t len) {
+	bool ret = false;
+	const uint8_t *buf = (const uint8_t *)p;
+	size_t i;
+
+	for (i = 0; i < len; i++) {
+		uint8_t b = buf[offset+i];
+		if (b != c) {
+			test_fail("Allocation at %p (len=%zu) contains %#x "
+			    "rather than %#x at offset %zu", p, len, b, c,
+			    offset+i);
+			ret = true;
+		}
+	}
+
+	return ret;
+}
+
+TEST_BEGIN(test_zero) {
+	void *p, *q;
+	size_t psz, qsz, i, j;
+	size_t start_sizes[] = {1, 3*1024, 63*1024, 4095*1024};
+#define FILL_BYTE 0xaaU
+#define RANGE 2048
+
+	for (i = 0; i < sizeof(start_sizes)/sizeof(size_t); i++) {
+		size_t start_size = start_sizes[i];
+		p = mallocx(start_size, MALLOCX_ZERO);
+		assert_ptr_not_null(p, "Unexpected mallocx() error");
+		psz = sallocx(p, 0);
+
+		assert_false(validate_fill(p, 0, 0, psz),
+		    "Expected zeroed memory");
+		memset(p, FILL_BYTE, psz);
+		assert_false(validate_fill(p, FILL_BYTE, 0, psz),
+		    "Expected filled memory");
+
+		for (j = 1; j < RANGE; j++) {
+			q = rallocx(p, start_size+j, MALLOCX_ZERO);
+			assert_ptr_not_null(q, "Unexpected rallocx() error");
+			qsz = sallocx(q, 0);
+			if (q != p || qsz != psz) {
+				assert_false(validate_fill(q, FILL_BYTE, 0,
+				    psz), "Expected filled memory");
+				assert_false(validate_fill(q, 0, psz, qsz-psz),
+				    "Expected zeroed memory");
+			}
+			if (psz != qsz) {
+				memset((void *)((uintptr_t)q+psz), FILL_BYTE,
+				    qsz-psz);
+				psz = qsz;
+			}
+			p = q;
+		}
+		assert_false(validate_fill(p, FILL_BYTE, 0, psz),
+		    "Expected filled memory");
+		dallocx(p, 0);
+	}
+#undef FILL_BYTE
+}
+TEST_END
+
+TEST_BEGIN(test_align) {
+	void *p, *q;
+	size_t align;
+#define MAX_ALIGN (ZU(1) << 25)
+
+	align = ZU(1);
+	p = mallocx(1, MALLOCX_ALIGN(align));
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	for (align <<= 1; align <= MAX_ALIGN; align <<= 1) {
+		q = rallocx(p, 1, MALLOCX_ALIGN(align));
+		assert_ptr_not_null(q,
+		    "Unexpected rallocx() error for align=%zu", align);
+		assert_ptr_null(
+		    (void *)((uintptr_t)q & (align-1)),
+		    "%p inadequately aligned for align=%zu",
+		    q, align);
+		p = q;
+	}
+	dallocx(p, 0);
+#undef MAX_ALIGN
+}
+TEST_END
+
+TEST_BEGIN(test_lg_align_and_zero) {
+	void *p, *q;
+	unsigned lg_align;
+	size_t sz;
+#define MAX_LG_ALIGN 25
+#define MAX_VALIDATE (ZU(1) << 22)
+
+	lg_align = 0;
+	p = mallocx(1, MALLOCX_LG_ALIGN(lg_align)|MALLOCX_ZERO);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	for (lg_align++; lg_align <= MAX_LG_ALIGN; lg_align++) {
+		q = rallocx(p, 1, MALLOCX_LG_ALIGN(lg_align)|MALLOCX_ZERO);
+		assert_ptr_not_null(q,
+		    "Unexpected rallocx() error for lg_align=%u", lg_align);
+		assert_ptr_null(
+		    (void *)((uintptr_t)q & ((ZU(1) << lg_align)-1)),
+		    "%p inadequately aligned for lg_align=%u", q, lg_align);
+		sz = sallocx(q, 0);
+		if ((sz << 1) <= MAX_VALIDATE) {
+			assert_false(validate_fill(q, 0, 0, sz),
+			    "Expected zeroed memory");
+		} else {
+			assert_false(validate_fill(q, 0, 0, MAX_VALIDATE),
+			    "Expected zeroed memory");
+			assert_false(validate_fill(
+			    (void *)((uintptr_t)q+sz-MAX_VALIDATE),
+			    0, 0, MAX_VALIDATE), "Expected zeroed memory");
+		}
+		p = q;
+	}
+	dallocx(p, 0);
+#undef MAX_VALIDATE
+#undef MAX_LG_ALIGN
+}
+TEST_END
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	size_t largemax;
+	void *p;
+
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_ptr_null(rallocx(p, largemax+1, 0),
+	    "Expected OOM for rallocx(p, size=%#zx, 0)", largemax+1);
+
+	assert_ptr_null(rallocx(p, ZU(PTRDIFF_MAX)+1, 0),
+	    "Expected OOM for rallocx(p, size=%#zx, 0)", ZU(PTRDIFF_MAX)+1);
+
+	assert_ptr_null(rallocx(p, SIZE_T_MAX, 0),
+	    "Expected OOM for rallocx(p, size=%#zx, 0)", SIZE_T_MAX);
+
+	assert_ptr_null(rallocx(p, 1, MALLOCX_ALIGN(ZU(PTRDIFF_MAX)+1)),
+	    "Expected OOM for rallocx(p, size=1, MALLOCX_ALIGN(%#zx))",
+	    ZU(PTRDIFF_MAX)+1);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+int
+main(void) {
+	return test(
+	    test_grow_and_shrink,
+	    test_zero,
+	    test_align,
+	    test_lg_align_and_zero,
+	    test_overflow);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/sdallocx.c b/memkind-1.10.0/jemalloc/test/integration/sdallocx.c
new file mode 100644
index 0000000..ca01448
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/sdallocx.c
@@ -0,0 +1,55 @@
+#include "test/jemalloc_test.h"
+
+#define MAXALIGN (((size_t)1) << 22)
+#define NITER 3
+
+TEST_BEGIN(test_basic) {
+	void *ptr = mallocx(64, 0);
+	sdallocx(ptr, 64, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+	size_t nsz, sz, alignment, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (sz = 1;
+		    sz < 3 * alignment && sz < (1U << 31);
+		    sz += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				nsz = nallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO);
+				ps[i] = mallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO);
+				total += nsz;
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					sdallocx(ps[i], sz,
+					    MALLOCX_ALIGN(alignment));
+					ps[i] = NULL;
+				}
+			}
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_basic,
+	    test_alignment_and_size);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/slab_sizes.c b/memkind-1.10.0/jemalloc/test/integration/slab_sizes.c
new file mode 100644
index 0000000..af250c3
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/slab_sizes.c
@@ -0,0 +1,80 @@
+#include "test/jemalloc_test.h"
+
+/* Note that this test relies on the unusual slab sizes set in slab_sizes.sh. */
+
+TEST_BEGIN(test_slab_sizes) {
+	unsigned nbins;
+	size_t page;
+	size_t sizemib[4];
+	size_t slabmib[4];
+	size_t len;
+
+	len = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", &nbins, &len, NULL, 0), 0,
+	    "nbins mallctl failure");
+
+	len = sizeof(page);
+	assert_d_eq(mallctl("arenas.page", &page, &len, NULL, 0), 0,
+	    "page mallctl failure");
+
+	len = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.size", sizemib, &len), 0,
+	    "bin size mallctlnametomib failure");
+
+	len = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.slab_size", slabmib, &len),
+	    0, "slab size mallctlnametomib failure");
+
+	size_t biggest_slab_seen = 0;
+
+	for (unsigned i = 0; i < nbins; i++) {
+		size_t bin_size;
+		size_t slab_size;
+		len = sizeof(size_t);
+		sizemib[2] = i;
+		slabmib[2] = i;
+		assert_d_eq(mallctlbymib(sizemib, 4, (void *)&bin_size, &len,
+		    NULL, 0), 0, "bin size mallctlbymib failure");
+
+		len = sizeof(size_t);
+		assert_d_eq(mallctlbymib(slabmib, 4, (void *)&slab_size, &len,
+		    NULL, 0), 0, "slab size mallctlbymib failure");
+
+		if (bin_size < 100) {
+			/*
+			 * Then we should be as close to 17 as possible.  Since
+			 * not all page sizes are valid (because of bitmap
+			 * limitations on the number of items in a slab), we
+			 * should at least make sure that the number of pages
+			 * goes up.
+			 */
+			assert_zu_ge(slab_size, biggest_slab_seen,
+			    "Slab sizes should go up");
+			biggest_slab_seen = slab_size;
+		} else if (
+		    (100 <= bin_size && bin_size < 128)
+		    || (128 < bin_size && bin_size <= 200)) {
+			assert_zu_eq(slab_size, page,
+			    "Forced-small slabs should be small");
+		} else if (bin_size == 128) {
+			assert_zu_eq(slab_size, 2 * page,
+			    "Forced-2-page slab should be 2 pages");
+		} else if (200 < bin_size && bin_size <= 4096) {
+			assert_zu_ge(slab_size, biggest_slab_seen,
+			    "Slab sizes should go up");
+			biggest_slab_seen = slab_size;
+		}
+	}
+	/*
+	 * For any reasonable configuration, 17 pages should be a valid slab
+	 * size for 4096-byte items.
+	 */
+	assert_zu_eq(biggest_slab_seen, 17 * page, "Didn't hit page target");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_slab_sizes);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/slab_sizes.sh b/memkind-1.10.0/jemalloc/test/integration/slab_sizes.sh
new file mode 100644
index 0000000..07e3db8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/slab_sizes.sh
@@ -0,0 +1,4 @@
+#!/bin/sh
+
+# Some screwy-looking slab sizes.
+export MALLOC_CONF="slab_sizes:1-4096:17|100-200:1|128-128:2"
diff --git a/memkind-1.10.0/jemalloc/test/integration/smallocx.c b/memkind-1.10.0/jemalloc/test/integration/smallocx.c
new file mode 100644
index 0000000..2486752
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/smallocx.c
@@ -0,0 +1,312 @@
+#include "test/jemalloc_test.h"
+#include "jemalloc/jemalloc_macros.h"
+
+#define STR_HELPER(x) #x
+#define STR(x) STR_HELPER(x)
+
+#ifndef JEMALLOC_VERSION_GID_IDENT
+  #error "JEMALLOC_VERSION_GID_IDENT not defined"
+#endif
+
+#define JOIN(x, y) x ## y
+#define JOIN2(x, y) JOIN(x, y)
+#define smallocx JOIN2(smallocx_, JEMALLOC_VERSION_GID_IDENT)
+
+typedef struct {
+	void *ptr;
+	size_t size;
+} smallocx_return_t;
+
+extern smallocx_return_t
+smallocx(size_t size, int flags);
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+/*
+ * On systems which can't merge extents, tests that call this function generate
+ * a lot of dirty memory very quickly.  Purging between cycles mitigates
+ * potential OOM on e.g. 32-bit Windows.
+ */
+static void
+purge(void) {
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl error");
+}
+
+/*
+ * GCC "-Walloc-size-larger-than" warning detects when one of the memory
+ * allocation functions is called with a size larger than the maximum size that
+ * they support. Here we want to explicitly test that the allocation functions
+ * do indeed fail properly when this is the case, which triggers the warning.
+ * Therefore we disable the warning for these tests.
+ */
+JEMALLOC_DIAGNOSTIC_PUSH
+JEMALLOC_DIAGNOSTIC_IGNORE_ALLOC_SIZE_LARGER_THAN
+
+TEST_BEGIN(test_overflow) {
+	size_t largemax;
+
+	largemax = get_large_size(get_nlarge()-1);
+
+	assert_ptr_null(smallocx(largemax+1, 0).ptr,
+	    "Expected OOM for smallocx(size=%#zx, 0)", largemax+1);
+
+	assert_ptr_null(smallocx(ZU(PTRDIFF_MAX)+1, 0).ptr,
+	    "Expected OOM for smallocx(size=%#zx, 0)", ZU(PTRDIFF_MAX)+1);
+
+	assert_ptr_null(smallocx(SIZE_T_MAX, 0).ptr,
+	    "Expected OOM for smallocx(size=%#zx, 0)", SIZE_T_MAX);
+
+	assert_ptr_null(smallocx(1, MALLOCX_ALIGN(ZU(PTRDIFF_MAX)+1)).ptr,
+	    "Expected OOM for smallocx(size=1, MALLOCX_ALIGN(%#zx))",
+	    ZU(PTRDIFF_MAX)+1);
+}
+TEST_END
+
+static void *
+remote_alloc(void *arg) {
+	unsigned arena;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	size_t large_sz;
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large_sz, &sz,
+	    NULL, 0), 0, "Unexpected mallctl failure");
+
+	smallocx_return_t r
+	    = smallocx(large_sz, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE);
+	void *ptr = r.ptr;
+	assert_zu_eq(r.size,
+	    nallocx(large_sz, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE),
+	    "Expected smalloc(size,flags).size == nallocx(size,flags)");
+	void **ret = (void **)arg;
+	*ret = ptr;
+
+	return NULL;
+}
+
+TEST_BEGIN(test_remote_free) {
+	thd_t thd;
+	void *ret;
+	thd_create(&thd, remote_alloc, (void *)&ret);
+	thd_join(thd, NULL);
+	assert_ptr_not_null(ret, "Unexpected smallocx failure");
+
+	/* Avoid TCACHE_NONE to explicitly test tcache_flush(). */
+	dallocx(ret, 0);
+	mallctl("thread.tcache.flush", NULL, NULL, NULL, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_oom) {
+	size_t largemax;
+	bool oom;
+	void *ptrs[3];
+	unsigned i;
+
+	/*
+	 * It should be impossible to allocate three objects that each consume
+	 * nearly half the virtual address space.
+	 */
+	largemax = get_large_size(get_nlarge()-1);
+	oom = false;
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		ptrs[i] = smallocx(largemax, 0).ptr;
+		if (ptrs[i] == NULL) {
+			oom = true;
+		}
+	}
+	assert_true(oom,
+	    "Expected OOM during series of calls to smallocx(size=%zu, 0)",
+	    largemax);
+	for (i = 0; i < sizeof(ptrs) / sizeof(void *); i++) {
+		if (ptrs[i] != NULL) {
+			dallocx(ptrs[i], 0);
+		}
+	}
+	purge();
+
+#if LG_SIZEOF_PTR == 3
+	assert_ptr_null(smallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x8000000000000000ULL)).ptr,
+	    "Expected OOM for smallocx()");
+	assert_ptr_null(smallocx(0x8000000000000000ULL,
+	    MALLOCX_ALIGN(0x80000000)).ptr,
+	    "Expected OOM for smallocx()");
+#else
+	assert_ptr_null(smallocx(0x80000000UL, MALLOCX_ALIGN(0x80000000UL)).ptr,
+	    "Expected OOM for smallocx()");
+#endif
+}
+TEST_END
+
+/* Re-enable the "-Walloc-size-larger-than=" warning */
+JEMALLOC_DIAGNOSTIC_POP
+
+TEST_BEGIN(test_basic) {
+#define MAXSZ (((size_t)1) << 23)
+	size_t sz;
+
+	for (sz = 1; sz < MAXSZ; sz = nallocx(sz, 0) + 1) {
+		smallocx_return_t ret;
+		size_t nsz, rsz, smz;
+		void *p;
+		nsz = nallocx(sz, 0);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		ret = smallocx(sz, 0);
+		p = ret.ptr;
+		smz = ret.size;
+		assert_ptr_not_null(p,
+		    "Unexpected smallocx(size=%zx, flags=0) error", sz);
+		rsz = sallocx(p, 0);
+		assert_zu_ge(rsz, sz, "Real size smaller than expected");
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() size mismatch");
+		assert_zu_eq(nsz, smz, "nallocx()/smallocx() size mismatch");
+		dallocx(p, 0);
+
+		ret = smallocx(sz, 0);
+		p = ret.ptr;
+		smz = ret.size;
+		assert_ptr_not_null(p,
+		    "Unexpected smallocx(size=%zx, flags=0) error", sz);
+		dallocx(p, 0);
+
+		nsz = nallocx(sz, MALLOCX_ZERO);
+		assert_zu_ne(nsz, 0, "Unexpected nallocx() error");
+		assert_zu_ne(smz, 0, "Unexpected smallocx() error");
+		ret = smallocx(sz, MALLOCX_ZERO);
+		p = ret.ptr;
+		assert_ptr_not_null(p,
+		    "Unexpected smallocx(size=%zx, flags=MALLOCX_ZERO) error",
+		    nsz);
+		rsz = sallocx(p, 0);
+		assert_zu_eq(nsz, rsz, "nallocx()/sallocx() rsize mismatch");
+		assert_zu_eq(nsz, smz, "nallocx()/smallocx() size mismatch");
+		dallocx(p, 0);
+		purge();
+	}
+#undef MAXSZ
+}
+TEST_END
+
+TEST_BEGIN(test_alignment_and_size) {
+	const char *percpu_arena;
+	size_t sz = sizeof(percpu_arena);
+
+	if(mallctl("opt.percpu_arena", (void *)&percpu_arena, &sz, NULL, 0) ||
+	    strcmp(percpu_arena, "disabled") != 0) {
+		test_skip("test_alignment_and_size skipped: "
+		    "not working with percpu arena.");
+	};
+#define MAXALIGN (((size_t)1) << 23)
+#define NITER 4
+	size_t nsz, rsz, smz, alignment, total;
+	unsigned i;
+	void *ps[NITER];
+
+	for (i = 0; i < NITER; i++) {
+		ps[i] = NULL;
+	}
+
+	for (alignment = 8;
+	    alignment <= MAXALIGN;
+	    alignment <<= 1) {
+		total = 0;
+		for (sz = 1;
+		    sz < 3 * alignment && sz < (1U << 31);
+		    sz += (alignment >> (LG_SIZEOF_PTR-1)) - 1) {
+			for (i = 0; i < NITER; i++) {
+				nsz = nallocx(sz, MALLOCX_ALIGN(alignment) |
+				    MALLOCX_ZERO);
+				assert_zu_ne(nsz, 0,
+				    "nallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				smallocx_return_t ret
+				    = smallocx(sz, MALLOCX_ALIGN(alignment) | MALLOCX_ZERO);
+				ps[i] = ret.ptr;
+				assert_ptr_not_null(ps[i],
+				    "smallocx() error for alignment=%zu, "
+				    "size=%zu (%#zx)", alignment, sz, sz);
+				rsz = sallocx(ps[i], 0);
+				smz = ret.size;
+				assert_zu_ge(rsz, sz,
+				    "Real size smaller than expected for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_zu_eq(nsz, rsz,
+				    "nallocx()/sallocx() size mismatch for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_zu_eq(nsz, smz,
+				    "nallocx()/smallocx() size mismatch for "
+				    "alignment=%zu, size=%zu", alignment, sz);
+				assert_ptr_null(
+				    (void *)((uintptr_t)ps[i] & (alignment-1)),
+				    "%p inadequately aligned for"
+				    " alignment=%zu, size=%zu", ps[i],
+				    alignment, sz);
+				total += rsz;
+				if (total >= (MAXALIGN << 1)) {
+					break;
+				}
+			}
+			for (i = 0; i < NITER; i++) {
+				if (ps[i] != NULL) {
+					dallocx(ps[i], 0);
+					ps[i] = NULL;
+				}
+			}
+		}
+		purge();
+	}
+#undef MAXALIGN
+#undef NITER
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_overflow,
+	    test_oom,
+	    test_remote_free,
+	    test_basic,
+	    test_alignment_and_size);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/smallocx.sh b/memkind-1.10.0/jemalloc/test/integration/smallocx.sh
new file mode 100644
index 0000000..d07f10f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/smallocx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+    export MALLOC_CONF="junk:false"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/integration/thread_arena.c b/memkind-1.10.0/jemalloc/test/integration/thread_arena.c
new file mode 100644
index 0000000..1e5ec05
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/thread_arena.c
@@ -0,0 +1,86 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS 10
+
+void *
+thd_start(void *arg) {
+	unsigned main_arena_ind = *(unsigned *)arg;
+	void *p;
+	unsigned arena_ind;
+	size_t size;
+	int err;
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Error in malloc()");
+	free(p);
+
+	size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", (void *)&arena_ind, &size,
+	    (void *)&main_arena_ind, sizeof(main_arena_ind)))) {
+		char buf[BUFERROR_BUF];
+
+		buferror(err, buf, sizeof(buf));
+		test_fail("Error in mallctl(): %s", buf);
+	}
+
+	size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", (void *)&arena_ind, &size, NULL,
+	    0))) {
+		char buf[BUFERROR_BUF];
+
+		buferror(err, buf, sizeof(buf));
+		test_fail("Error in mallctl(): %s", buf);
+	}
+	assert_u_eq(arena_ind, main_arena_ind,
+	    "Arena index should be same as for main thread");
+
+	return NULL;
+}
+
+static void
+mallctl_failure(int err) {
+	char buf[BUFERROR_BUF];
+
+	buferror(err, buf, sizeof(buf));
+	test_fail("Error in mallctl(): %s", buf);
+}
+
+TEST_BEGIN(test_thread_arena) {
+	void *p;
+	int err;
+	thd_t thds[NTHREADS];
+	unsigned i;
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Error in malloc()");
+
+	unsigned arena_ind, old_arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Arena creation failure");
+
+	size_t size = sizeof(arena_ind);
+	if ((err = mallctl("thread.arena", (void *)&old_arena_ind, &size,
+	    (void *)&arena_ind, sizeof(arena_ind))) != 0) {
+		mallctl_failure(err);
+	}
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start,
+		    (void *)&arena_ind);
+	}
+
+	for (i = 0; i < NTHREADS; i++) {
+		intptr_t join_ret;
+		thd_join(thds[i], (void *)&join_ret);
+		assert_zd_eq(join_ret, 0, "Unexpected thread join error");
+	}
+	free(p);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_thread_arena);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/thread_tcache_enabled.c b/memkind-1.10.0/jemalloc/test/integration/thread_tcache_enabled.c
new file mode 100644
index 0000000..95c9acc
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/thread_tcache_enabled.c
@@ -0,0 +1,87 @@
+#include "test/jemalloc_test.h"
+
+void *
+thd_start(void *arg) {
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	if (e0) {
+		e1 = false;
+		assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+		    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+		assert_true(e0, "tcache should be enabled");
+	}
+
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	free(malloc(1));
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	free(malloc(1));
+	e1 = true;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	free(malloc(1));
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_true(e0, "tcache should be enabled");
+
+	free(malloc(1));
+	e1 = false;
+	assert_d_eq(mallctl("thread.tcache.enabled", (void *)&e0, &sz,
+	    (void *)&e1, sz), 0, "Unexpected mallctl() error");
+	assert_false(e0, "tcache should be disabled");
+
+	free(malloc(1));
+	return NULL;
+}
+
+TEST_BEGIN(test_main_thread) {
+	thd_start(NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_subthread) {
+	thd_t thd;
+
+	thd_create(&thd, thd_start, NULL);
+	thd_join(thd, NULL);
+}
+TEST_END
+
+int
+main(void) {
+	/* Run tests multiple times to check for bad interactions. */
+	return test(
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread,
+	    test_subthread,
+	    test_main_thread);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/xallocx.c b/memkind-1.10.0/jemalloc/test/integration/xallocx.c
new file mode 100644
index 0000000..cd0ca04
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/xallocx.c
@@ -0,0 +1,384 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * Use a separate arena for xallocx() extension/contraction tests so that
+ * internal allocation e.g. by heap profiling can't interpose allocations where
+ * xallocx() would ordinarily be able to extend.
+ */
+static unsigned
+arena_ind(void) {
+	static unsigned ind = 0;
+
+	if (ind == 0) {
+		size_t sz = sizeof(ind);
+		assert_d_eq(mallctl("arenas.create", (void *)&ind, &sz, NULL,
+		    0), 0, "Unexpected mallctl failure creating arena");
+	}
+
+	return ind;
+}
+
+TEST_BEGIN(test_same_size) {
+	void *p;
+	size_t sz, tsz;
+
+	p = mallocx(42, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	sz = sallocx(p, 0);
+
+	tsz = xallocx(p, sz, 0, 0);
+	assert_zu_eq(tsz, sz, "Unexpected size change: %zu --> %zu", sz, tsz);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_extra_no_move) {
+	void *p;
+	size_t sz, tsz;
+
+	p = mallocx(42, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	sz = sallocx(p, 0);
+
+	tsz = xallocx(p, sz, sz-42, 0);
+	assert_zu_eq(tsz, sz, "Unexpected size change: %zu --> %zu", sz, tsz);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_no_move_fail) {
+	void *p;
+	size_t sz, tsz;
+
+	p = mallocx(42, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	sz = sallocx(p, 0);
+
+	tsz = xallocx(p, sz + 5, 0, 0);
+	assert_zu_eq(tsz, sz, "Unexpected size change: %zu --> %zu", sz, tsz);
+
+	dallocx(p, 0);
+}
+TEST_END
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nsmall(void) {
+	return get_nsizes_impl("arenas.nbins");
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_small_size(size_t ind) {
+	return get_size_impl("arenas.bin.0.size", ind);
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+TEST_BEGIN(test_size) {
+	size_t small0, largemax;
+	void *p;
+
+	/* Get size classes. */
+	small0 = get_small_size(0);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(small0, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	/* Test smallest supported size. */
+	assert_zu_eq(xallocx(p, 1, 0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	/* Test largest supported size. */
+	assert_zu_le(xallocx(p, largemax, 0, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	/* Test size overflow. */
+	assert_zu_le(xallocx(p, largemax+1, 0, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, SIZE_T_MAX, 0, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_size_extra_overflow) {
+	size_t small0, largemax;
+	void *p;
+
+	/* Get size classes. */
+	small0 = get_small_size(0);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(small0, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	/* Test overflows that can be resolved by clamping extra. */
+	assert_zu_le(xallocx(p, largemax-1, 2, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, largemax, 1, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	/* Test overflow such that largemax-size underflows. */
+	assert_zu_le(xallocx(p, largemax+1, 2, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, largemax+2, 3, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, SIZE_T_MAX-2, 2, 0), largemax,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, SIZE_T_MAX-1, 1, 0), largemax,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_extra_small) {
+	size_t small0, small1, largemax;
+	void *p;
+
+	/* Get size classes. */
+	small0 = get_small_size(0);
+	small1 = get_small_size(1);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(small0, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	assert_zu_eq(xallocx(p, small1, 0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_eq(xallocx(p, small1, 0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_eq(xallocx(p, small0, small1 - small0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	/* Test size+extra overflow. */
+	assert_zu_eq(xallocx(p, small0, largemax - small0 + 1, 0), small0,
+	    "Unexpected xallocx() behavior");
+	assert_zu_eq(xallocx(p, small0, SIZE_T_MAX - small0, 0), small0,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_extra_large) {
+	int flags = MALLOCX_ARENA(arena_ind());
+	size_t smallmax, large1, large2, large3, largemax;
+	void *p;
+
+	/* Get size classes. */
+	smallmax = get_small_size(get_nsmall()-1);
+	large1 = get_large_size(1);
+	large2 = get_large_size(2);
+	large3 = get_large_size(3);
+	largemax = get_large_size(get_nlarge()-1);
+
+	p = mallocx(large3, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+
+	assert_zu_eq(xallocx(p, large3, 0, flags), large3,
+	    "Unexpected xallocx() behavior");
+	/* Test size decrease with zero extra. */
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	assert_zu_ge(xallocx(p, smallmax, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+
+	if (xallocx(p, large3, 0, flags) != large3) {
+		p = rallocx(p, large3, flags);
+		assert_ptr_not_null(p, "Unexpected rallocx() failure");
+	}
+	/* Test size decrease with non-zero extra. */
+	assert_zu_eq(xallocx(p, large1, large3 - large1, flags), large3,
+	    "Unexpected xallocx() behavior");
+	assert_zu_eq(xallocx(p, large2, large3 - large2, flags), large3,
+	    "Unexpected xallocx() behavior");
+	assert_zu_ge(xallocx(p, large1, large2 - large1, flags), large2,
+	    "Unexpected xallocx() behavior");
+	assert_zu_ge(xallocx(p, smallmax, large1 - smallmax, flags), large1,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	/* Test size increase with zero extra. */
+	assert_zu_le(xallocx(p, large3, 0, flags), large3,
+	    "Unexpected xallocx() behavior");
+	assert_zu_le(xallocx(p, largemax+1, 0, flags), large3,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	/* Test size increase with non-zero extra. */
+	assert_zu_le(xallocx(p, large1, SIZE_T_MAX - large1, flags), largemax,
+	    "Unexpected xallocx() behavior");
+
+	assert_zu_ge(xallocx(p, large1, 0, flags), large1,
+	    "Unexpected xallocx() behavior");
+	/* Test size increase with non-zero extra. */
+	assert_zu_le(xallocx(p, large1, large3 - large1, flags), large3,
+	    "Unexpected xallocx() behavior");
+
+	if (xallocx(p, large3, 0, flags) != large3) {
+		p = rallocx(p, large3, flags);
+		assert_ptr_not_null(p, "Unexpected rallocx() failure");
+	}
+	/* Test size+extra overflow. */
+	assert_zu_le(xallocx(p, large3, largemax - large3 + 1, flags), largemax,
+	    "Unexpected xallocx() behavior");
+
+	dallocx(p, flags);
+}
+TEST_END
+
+static void
+print_filled_extents(const void *p, uint8_t c, size_t len) {
+	const uint8_t *pc = (const uint8_t *)p;
+	size_t i, range0;
+	uint8_t c0;
+
+	malloc_printf("  p=%p, c=%#x, len=%zu:", p, c, len);
+	range0 = 0;
+	c0 = pc[0];
+	for (i = 0; i < len; i++) {
+		if (pc[i] != c0) {
+			malloc_printf(" %#x[%zu..%zu)", c0, range0, i);
+			range0 = i;
+			c0 = pc[i];
+		}
+	}
+	malloc_printf(" %#x[%zu..%zu)\n", c0, range0, i);
+}
+
+static bool
+validate_fill(const void *p, uint8_t c, size_t offset, size_t len) {
+	const uint8_t *pc = (const uint8_t *)p;
+	bool err;
+	size_t i;
+
+	for (i = offset, err = false; i < offset+len; i++) {
+		if (pc[i] != c) {
+			err = true;
+		}
+	}
+
+	if (err) {
+		print_filled_extents(p, c, offset + len);
+	}
+
+	return err;
+}
+
+static void
+test_zero(size_t szmin, size_t szmax) {
+	int flags = MALLOCX_ARENA(arena_ind()) | MALLOCX_ZERO;
+	size_t sz, nsz;
+	void *p;
+#define FILL_BYTE 0x7aU
+
+	sz = szmax;
+	p = mallocx(sz, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() error");
+	assert_false(validate_fill(p, 0x00, 0, sz), "Memory not filled: sz=%zu",
+	    sz);
+
+	/*
+	 * Fill with non-zero so that non-debug builds are more likely to detect
+	 * errors.
+	 */
+	memset(p, FILL_BYTE, sz);
+	assert_false(validate_fill(p, FILL_BYTE, 0, sz),
+	    "Memory not filled: sz=%zu", sz);
+
+	/* Shrink in place so that we can expect growing in place to succeed. */
+	sz = szmin;
+	if (xallocx(p, sz, 0, flags) != sz) {
+		p = rallocx(p, sz, flags);
+		assert_ptr_not_null(p, "Unexpected rallocx() failure");
+	}
+	assert_false(validate_fill(p, FILL_BYTE, 0, sz),
+	    "Memory not filled: sz=%zu", sz);
+
+	for (sz = szmin; sz < szmax; sz = nsz) {
+		nsz = nallocx(sz+1, flags);
+		if (xallocx(p, sz+1, 0, flags) != nsz) {
+			p = rallocx(p, sz+1, flags);
+			assert_ptr_not_null(p, "Unexpected rallocx() failure");
+		}
+		assert_false(validate_fill(p, FILL_BYTE, 0, sz),
+		    "Memory not filled: sz=%zu", sz);
+		assert_false(validate_fill(p, 0x00, sz, nsz-sz),
+		    "Memory not filled: sz=%zu, nsz-sz=%zu", sz, nsz-sz);
+		memset((void *)((uintptr_t)p + sz), FILL_BYTE, nsz-sz);
+		assert_false(validate_fill(p, FILL_BYTE, 0, nsz),
+		    "Memory not filled: nsz=%zu", nsz);
+	}
+
+	dallocx(p, flags);
+}
+
+TEST_BEGIN(test_zero_large) {
+	size_t large0, large1;
+
+	/* Get size classes. */
+	large0 = get_large_size(0);
+	large1 = get_large_size(1);
+
+	test_zero(large1, large0 * 2);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_same_size,
+	    test_extra_no_move,
+	    test_no_move_fail,
+	    test_size,
+	    test_size_extra_overflow,
+	    test_extra_small,
+	    test_extra_large,
+	    test_zero_large);
+}
diff --git a/memkind-1.10.0/jemalloc/test/integration/xallocx.sh b/memkind-1.10.0/jemalloc/test/integration/xallocx.sh
new file mode 100644
index 0000000..0cc2187
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/integration/xallocx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="junk:false"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/src/SFMT.c b/memkind-1.10.0/jemalloc/test/src/SFMT.c
new file mode 100644
index 0000000..c05e218
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/SFMT.c
@@ -0,0 +1,719 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+/**
+ * @file  SFMT.c
+ * @brief SIMD oriented Fast Mersenne Twister(SFMT)
+ *
+ * @author Mutsuo Saito (Hiroshima University)
+ * @author Makoto Matsumoto (Hiroshima University)
+ *
+ * Copyright (C) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ * University. All rights reserved.
+ *
+ * The new BSD License is applied to this software, see LICENSE.txt
+ */
+#define SFMT_C_
+#include "test/jemalloc_test.h"
+#include "test/SFMT-params.h"
+
+#if defined(JEMALLOC_BIG_ENDIAN) && !defined(BIG_ENDIAN64)
+#define BIG_ENDIAN64 1
+#endif
+#if defined(__BIG_ENDIAN__) && !defined(__amd64) && !defined(BIG_ENDIAN64)
+#define BIG_ENDIAN64 1
+#endif
+#if defined(HAVE_ALTIVEC) && !defined(BIG_ENDIAN64)
+#define BIG_ENDIAN64 1
+#endif
+#if defined(ONLY64) && !defined(BIG_ENDIAN64)
+  #if defined(__GNUC__)
+    #error "-DONLY64 must be specified with -DBIG_ENDIAN64"
+  #endif
+#undef ONLY64
+#endif
+/*------------------------------------------------------
+  128-bit SIMD data type for Altivec, SSE2 or standard C
+  ------------------------------------------------------*/
+#if defined(HAVE_ALTIVEC)
+/** 128-bit data structure */
+union W128_T {
+    vector unsigned int s;
+    uint32_t u[4];
+};
+/** 128-bit data type */
+typedef union W128_T w128_t;
+
+#elif defined(HAVE_SSE2)
+/** 128-bit data structure */
+union W128_T {
+    __m128i si;
+    uint32_t u[4];
+};
+/** 128-bit data type */
+typedef union W128_T w128_t;
+
+#else
+
+/** 128-bit data structure */
+struct W128_T {
+    uint32_t u[4];
+};
+/** 128-bit data type */
+typedef struct W128_T w128_t;
+
+#endif
+
+struct sfmt_s {
+    /** the 128-bit internal state array */
+    w128_t sfmt[N];
+    /** index counter to the 32-bit internal state array */
+    int idx;
+    /** a flag: it is 0 if and only if the internal state is not yet
+     * initialized. */
+    int initialized;
+};
+
+/*--------------------------------------
+  FILE GLOBAL VARIABLES
+  internal state, index counter and flag
+  --------------------------------------*/
+
+/** a parity check vector which certificate the period of 2^{MEXP} */
+static uint32_t parity[4] = {PARITY1, PARITY2, PARITY3, PARITY4};
+
+/*----------------
+  STATIC FUNCTIONS
+  ----------------*/
+static inline int idxof(int i);
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+static inline void rshift128(w128_t *out,  w128_t const *in, int shift);
+static inline void lshift128(w128_t *out,  w128_t const *in, int shift);
+#endif
+static inline void gen_rand_all(sfmt_t *ctx);
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size);
+static inline uint32_t func1(uint32_t x);
+static inline uint32_t func2(uint32_t x);
+static void period_certification(sfmt_t *ctx);
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+static inline void swap(w128_t *array, int size);
+#endif
+
+#if defined(HAVE_ALTIVEC)
+  #include "test/SFMT-alti.h"
+#elif defined(HAVE_SSE2)
+  #include "test/SFMT-sse2.h"
+#endif
+
+/**
+ * This function simulate a 64-bit index of LITTLE ENDIAN
+ * in BIG ENDIAN machine.
+ */
+#ifdef ONLY64
+static inline int idxof(int i) {
+    return i ^ 1;
+}
+#else
+static inline int idxof(int i) {
+    return i;
+}
+#endif
+/**
+ * This function simulates SIMD 128-bit right shift by the standard C.
+ * The 128-bit integer given in in is shifted by (shift * 8) bits.
+ * This function simulates the LITTLE ENDIAN SIMD.
+ * @param out the output of this function
+ * @param in the 128-bit data to be shifted
+ * @param shift the shift value
+ */
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+#ifdef ONLY64
+static inline void rshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[2] << 32) | ((uint64_t)in->u[3]);
+    tl = ((uint64_t)in->u[0] << 32) | ((uint64_t)in->u[1]);
+
+    oh = th >> (shift * 8);
+    ol = tl >> (shift * 8);
+    ol |= th << (64 - shift * 8);
+    out->u[0] = (uint32_t)(ol >> 32);
+    out->u[1] = (uint32_t)ol;
+    out->u[2] = (uint32_t)(oh >> 32);
+    out->u[3] = (uint32_t)oh;
+}
+#else
+static inline void rshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[3] << 32) | ((uint64_t)in->u[2]);
+    tl = ((uint64_t)in->u[1] << 32) | ((uint64_t)in->u[0]);
+
+    oh = th >> (shift * 8);
+    ol = tl >> (shift * 8);
+    ol |= th << (64 - shift * 8);
+    out->u[1] = (uint32_t)(ol >> 32);
+    out->u[0] = (uint32_t)ol;
+    out->u[3] = (uint32_t)(oh >> 32);
+    out->u[2] = (uint32_t)oh;
+}
+#endif
+/**
+ * This function simulates SIMD 128-bit left shift by the standard C.
+ * The 128-bit integer given in in is shifted by (shift * 8) bits.
+ * This function simulates the LITTLE ENDIAN SIMD.
+ * @param out the output of this function
+ * @param in the 128-bit data to be shifted
+ * @param shift the shift value
+ */
+#ifdef ONLY64
+static inline void lshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[2] << 32) | ((uint64_t)in->u[3]);
+    tl = ((uint64_t)in->u[0] << 32) | ((uint64_t)in->u[1]);
+
+    oh = th << (shift * 8);
+    ol = tl << (shift * 8);
+    oh |= tl >> (64 - shift * 8);
+    out->u[0] = (uint32_t)(ol >> 32);
+    out->u[1] = (uint32_t)ol;
+    out->u[2] = (uint32_t)(oh >> 32);
+    out->u[3] = (uint32_t)oh;
+}
+#else
+static inline void lshift128(w128_t *out, w128_t const *in, int shift) {
+    uint64_t th, tl, oh, ol;
+
+    th = ((uint64_t)in->u[3] << 32) | ((uint64_t)in->u[2]);
+    tl = ((uint64_t)in->u[1] << 32) | ((uint64_t)in->u[0]);
+
+    oh = th << (shift * 8);
+    ol = tl << (shift * 8);
+    oh |= tl >> (64 - shift * 8);
+    out->u[1] = (uint32_t)(ol >> 32);
+    out->u[0] = (uint32_t)ol;
+    out->u[3] = (uint32_t)(oh >> 32);
+    out->u[2] = (uint32_t)oh;
+}
+#endif
+#endif
+
+/**
+ * This function represents the recursion formula.
+ * @param r output
+ * @param a a 128-bit part of the internal state array
+ * @param b a 128-bit part of the internal state array
+ * @param c a 128-bit part of the internal state array
+ * @param d a 128-bit part of the internal state array
+ */
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+#ifdef ONLY64
+static inline void do_recursion(w128_t *r, w128_t *a, w128_t *b, w128_t *c,
+				w128_t *d) {
+    w128_t x;
+    w128_t y;
+
+    lshift128(&x, a, SL2);
+    rshift128(&y, c, SR2);
+    r->u[0] = a->u[0] ^ x.u[0] ^ ((b->u[0] >> SR1) & MSK2) ^ y.u[0]
+	^ (d->u[0] << SL1);
+    r->u[1] = a->u[1] ^ x.u[1] ^ ((b->u[1] >> SR1) & MSK1) ^ y.u[1]
+	^ (d->u[1] << SL1);
+    r->u[2] = a->u[2] ^ x.u[2] ^ ((b->u[2] >> SR1) & MSK4) ^ y.u[2]
+	^ (d->u[2] << SL1);
+    r->u[3] = a->u[3] ^ x.u[3] ^ ((b->u[3] >> SR1) & MSK3) ^ y.u[3]
+	^ (d->u[3] << SL1);
+}
+#else
+static inline void do_recursion(w128_t *r, w128_t *a, w128_t *b, w128_t *c,
+				w128_t *d) {
+    w128_t x;
+    w128_t y;
+
+    lshift128(&x, a, SL2);
+    rshift128(&y, c, SR2);
+    r->u[0] = a->u[0] ^ x.u[0] ^ ((b->u[0] >> SR1) & MSK1) ^ y.u[0]
+	^ (d->u[0] << SL1);
+    r->u[1] = a->u[1] ^ x.u[1] ^ ((b->u[1] >> SR1) & MSK2) ^ y.u[1]
+	^ (d->u[1] << SL1);
+    r->u[2] = a->u[2] ^ x.u[2] ^ ((b->u[2] >> SR1) & MSK3) ^ y.u[2]
+	^ (d->u[2] << SL1);
+    r->u[3] = a->u[3] ^ x.u[3] ^ ((b->u[3] >> SR1) & MSK4) ^ y.u[3]
+	^ (d->u[3] << SL1);
+}
+#endif
+#endif
+
+#if (!defined(HAVE_ALTIVEC)) && (!defined(HAVE_SSE2))
+/**
+ * This function fills the internal state array with pseudorandom
+ * integers.
+ */
+static inline void gen_rand_all(sfmt_t *ctx) {
+    int i;
+    w128_t *r1, *r2;
+
+    r1 = &ctx->sfmt[N - 2];
+    r2 = &ctx->sfmt[N - 1];
+    for (i = 0; i < N - POS1; i++) {
+	do_recursion(&ctx->sfmt[i], &ctx->sfmt[i], &ctx->sfmt[i + POS1], r1,
+	  r2);
+	r1 = r2;
+	r2 = &ctx->sfmt[i];
+    }
+    for (; i < N; i++) {
+	do_recursion(&ctx->sfmt[i], &ctx->sfmt[i], &ctx->sfmt[i + POS1 - N], r1,
+	  r2);
+	r1 = r2;
+	r2 = &ctx->sfmt[i];
+    }
+}
+
+/**
+ * This function fills the user-specified array with pseudorandom
+ * integers.
+ *
+ * @param array an 128-bit array to be filled by pseudorandom numbers.
+ * @param size number of 128-bit pseudorandom numbers to be generated.
+ */
+static inline void gen_rand_array(sfmt_t *ctx, w128_t *array, int size) {
+    int i, j;
+    w128_t *r1, *r2;
+
+    r1 = &ctx->sfmt[N - 2];
+    r2 = &ctx->sfmt[N - 1];
+    for (i = 0; i < N - POS1; i++) {
+	do_recursion(&array[i], &ctx->sfmt[i], &ctx->sfmt[i + POS1], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+    }
+    for (; i < N; i++) {
+	do_recursion(&array[i], &ctx->sfmt[i], &array[i + POS1 - N], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+    }
+    for (; i < size - N; i++) {
+	do_recursion(&array[i], &array[i - N], &array[i + POS1 - N], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+    }
+    for (j = 0; j < 2 * N - size; j++) {
+	ctx->sfmt[j] = array[j + size - N];
+    }
+    for (; i < size; i++, j++) {
+	do_recursion(&array[i], &array[i - N], &array[i + POS1 - N], r1, r2);
+	r1 = r2;
+	r2 = &array[i];
+	ctx->sfmt[j] = array[i];
+    }
+}
+#endif
+
+#if defined(BIG_ENDIAN64) && !defined(ONLY64) && !defined(HAVE_ALTIVEC)
+static inline void swap(w128_t *array, int size) {
+    int i;
+    uint32_t x, y;
+
+    for (i = 0; i < size; i++) {
+	x = array[i].u[0];
+	y = array[i].u[2];
+	array[i].u[0] = array[i].u[1];
+	array[i].u[2] = array[i].u[3];
+	array[i].u[1] = x;
+	array[i].u[3] = y;
+    }
+}
+#endif
+/**
+ * This function represents a function used in the initialization
+ * by init_by_array
+ * @param x 32-bit integer
+ * @return 32-bit integer
+ */
+static uint32_t func1(uint32_t x) {
+    return (x ^ (x >> 27)) * (uint32_t)1664525UL;
+}
+
+/**
+ * This function represents a function used in the initialization
+ * by init_by_array
+ * @param x 32-bit integer
+ * @return 32-bit integer
+ */
+static uint32_t func2(uint32_t x) {
+    return (x ^ (x >> 27)) * (uint32_t)1566083941UL;
+}
+
+/**
+ * This function certificate the period of 2^{MEXP}
+ */
+static void period_certification(sfmt_t *ctx) {
+    int inner = 0;
+    int i, j;
+    uint32_t work;
+    uint32_t *psfmt32 = &ctx->sfmt[0].u[0];
+
+    for (i = 0; i < 4; i++)
+	inner ^= psfmt32[idxof(i)] & parity[i];
+    for (i = 16; i > 0; i >>= 1)
+	inner ^= inner >> i;
+    inner &= 1;
+    /* check OK */
+    if (inner == 1) {
+	return;
+    }
+    /* check NG, and modification */
+    for (i = 0; i < 4; i++) {
+	work = 1;
+	for (j = 0; j < 32; j++) {
+	    if ((work & parity[i]) != 0) {
+		psfmt32[idxof(i)] ^= work;
+		return;
+	    }
+	    work = work << 1;
+	}
+    }
+}
+
+/*----------------
+  PUBLIC FUNCTIONS
+  ----------------*/
+/**
+ * This function returns the identification string.
+ * The string shows the word size, the Mersenne exponent,
+ * and all parameters of this generator.
+ */
+const char *get_idstring(void) {
+    return IDSTR;
+}
+
+/**
+ * This function returns the minimum size of array used for \b
+ * fill_array32() function.
+ * @return minimum size of array used for fill_array32() function.
+ */
+int get_min_array_size32(void) {
+    return N32;
+}
+
+/**
+ * This function returns the minimum size of array used for \b
+ * fill_array64() function.
+ * @return minimum size of array used for fill_array64() function.
+ */
+int get_min_array_size64(void) {
+    return N64;
+}
+
+#ifndef ONLY64
+/**
+ * This function generates and returns 32-bit pseudorandom number.
+ * init_gen_rand or init_by_array must be called before this function.
+ * @return 32-bit pseudorandom number
+ */
+uint32_t gen_rand32(sfmt_t *ctx) {
+    uint32_t r;
+    uint32_t *psfmt32 = &ctx->sfmt[0].u[0];
+
+    assert(ctx->initialized);
+    if (ctx->idx >= N32) {
+	gen_rand_all(ctx);
+	ctx->idx = 0;
+    }
+    r = psfmt32[ctx->idx++];
+    return r;
+}
+
+/* Generate a random integer in [0..limit). */
+uint32_t gen_rand32_range(sfmt_t *ctx, uint32_t limit) {
+    uint32_t ret, above;
+
+    above = 0xffffffffU - (0xffffffffU % limit);
+    while (1) {
+	ret = gen_rand32(ctx);
+	if (ret < above) {
+	    ret %= limit;
+	    break;
+	}
+    }
+    return ret;
+}
+#endif
+/**
+ * This function generates and returns 64-bit pseudorandom number.
+ * init_gen_rand or init_by_array must be called before this function.
+ * The function gen_rand64 should not be called after gen_rand32,
+ * unless an initialization is again executed.
+ * @return 64-bit pseudorandom number
+ */
+uint64_t gen_rand64(sfmt_t *ctx) {
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+    uint32_t r1, r2;
+    uint32_t *psfmt32 = &ctx->sfmt[0].u[0];
+#else
+    uint64_t r;
+    uint64_t *psfmt64 = (uint64_t *)&ctx->sfmt[0].u[0];
+#endif
+
+    assert(ctx->initialized);
+    assert(ctx->idx % 2 == 0);
+
+    if (ctx->idx >= N32) {
+	gen_rand_all(ctx);
+	ctx->idx = 0;
+    }
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+    r1 = psfmt32[ctx->idx];
+    r2 = psfmt32[ctx->idx + 1];
+    ctx->idx += 2;
+    return ((uint64_t)r2 << 32) | r1;
+#else
+    r = psfmt64[ctx->idx / 2];
+    ctx->idx += 2;
+    return r;
+#endif
+}
+
+/* Generate a random integer in [0..limit). */
+uint64_t gen_rand64_range(sfmt_t *ctx, uint64_t limit) {
+    uint64_t ret, above;
+
+    above = KQU(0xffffffffffffffff) - (KQU(0xffffffffffffffff) % limit);
+    while (1) {
+	ret = gen_rand64(ctx);
+	if (ret < above) {
+	    ret %= limit;
+	    break;
+	}
+    }
+    return ret;
+}
+
+#ifndef ONLY64
+/**
+ * This function generates pseudorandom 32-bit integers in the
+ * specified array[] by one call. The number of pseudorandom integers
+ * is specified by the argument size, which must be at least 624 and a
+ * multiple of four.  The generation by this function is much faster
+ * than the following gen_rand function.
+ *
+ * For initialization, init_gen_rand or init_by_array must be called
+ * before the first call of this function. This function can not be
+ * used after calling gen_rand function, without initialization.
+ *
+ * @param array an array where pseudorandom 32-bit integers are filled
+ * by this function.  The pointer to the array must be \b "aligned"
+ * (namely, must be a multiple of 16) in the SIMD version, since it
+ * refers to the address of a 128-bit integer.  In the standard C
+ * version, the pointer is arbitrary.
+ *
+ * @param size the number of 32-bit pseudorandom integers to be
+ * generated.  size must be a multiple of 4, and greater than or equal
+ * to (MEXP / 128 + 1) * 4.
+ *
+ * @note \b memalign or \b posix_memalign is available to get aligned
+ * memory. Mac OSX doesn't have these functions, but \b malloc of OSX
+ * returns the pointer to the aligned memory block.
+ */
+void fill_array32(sfmt_t *ctx, uint32_t *array, int size) {
+    assert(ctx->initialized);
+    assert(ctx->idx == N32);
+    assert(size % 4 == 0);
+    assert(size >= N32);
+
+    gen_rand_array(ctx, (w128_t *)array, size / 4);
+    ctx->idx = N32;
+}
+#endif
+
+/**
+ * This function generates pseudorandom 64-bit integers in the
+ * specified array[] by one call. The number of pseudorandom integers
+ * is specified by the argument size, which must be at least 312 and a
+ * multiple of two.  The generation by this function is much faster
+ * than the following gen_rand function.
+ *
+ * For initialization, init_gen_rand or init_by_array must be called
+ * before the first call of this function. This function can not be
+ * used after calling gen_rand function, without initialization.
+ *
+ * @param array an array where pseudorandom 64-bit integers are filled
+ * by this function.  The pointer to the array must be "aligned"
+ * (namely, must be a multiple of 16) in the SIMD version, since it
+ * refers to the address of a 128-bit integer.  In the standard C
+ * version, the pointer is arbitrary.
+ *
+ * @param size the number of 64-bit pseudorandom integers to be
+ * generated.  size must be a multiple of 2, and greater than or equal
+ * to (MEXP / 128 + 1) * 2
+ *
+ * @note \b memalign or \b posix_memalign is available to get aligned
+ * memory. Mac OSX doesn't have these functions, but \b malloc of OSX
+ * returns the pointer to the aligned memory block.
+ */
+void fill_array64(sfmt_t *ctx, uint64_t *array, int size) {
+    assert(ctx->initialized);
+    assert(ctx->idx == N32);
+    assert(size % 2 == 0);
+    assert(size >= N64);
+
+    gen_rand_array(ctx, (w128_t *)array, size / 2);
+    ctx->idx = N32;
+
+#if defined(BIG_ENDIAN64) && !defined(ONLY64)
+    swap((w128_t *)array, size /2);
+#endif
+}
+
+/**
+ * This function initializes the internal state array with a 32-bit
+ * integer seed.
+ *
+ * @param seed a 32-bit integer used as the seed.
+ */
+sfmt_t *init_gen_rand(uint32_t seed) {
+    void *p;
+    sfmt_t *ctx;
+    int i;
+    uint32_t *psfmt32;
+
+    if (posix_memalign(&p, sizeof(w128_t), sizeof(sfmt_t)) != 0) {
+	return NULL;
+    }
+    ctx = (sfmt_t *)p;
+    psfmt32 = &ctx->sfmt[0].u[0];
+
+    psfmt32[idxof(0)] = seed;
+    for (i = 1; i < N32; i++) {
+	psfmt32[idxof(i)] = 1812433253UL * (psfmt32[idxof(i - 1)]
+					    ^ (psfmt32[idxof(i - 1)] >> 30))
+	    + i;
+    }
+    ctx->idx = N32;
+    period_certification(ctx);
+    ctx->initialized = 1;
+
+    return ctx;
+}
+
+/**
+ * This function initializes the internal state array,
+ * with an array of 32-bit integers used as the seeds
+ * @param init_key the array of 32-bit integers, used as a seed.
+ * @param key_length the length of init_key.
+ */
+sfmt_t *init_by_array(uint32_t *init_key, int key_length) {
+    void *p;
+    sfmt_t *ctx;
+    int i, j, count;
+    uint32_t r;
+    int lag;
+    int mid;
+    int size = N * 4;
+    uint32_t *psfmt32;
+
+    if (posix_memalign(&p, sizeof(w128_t), sizeof(sfmt_t)) != 0) {
+	return NULL;
+    }
+    ctx = (sfmt_t *)p;
+    psfmt32 = &ctx->sfmt[0].u[0];
+
+    if (size >= 623) {
+	lag = 11;
+    } else if (size >= 68) {
+	lag = 7;
+    } else if (size >= 39) {
+	lag = 5;
+    } else {
+	lag = 3;
+    }
+    mid = (size - lag) / 2;
+
+    memset(ctx->sfmt, 0x8b, sizeof(ctx->sfmt));
+    if (key_length + 1 > N32) {
+	count = key_length + 1;
+    } else {
+	count = N32;
+    }
+    r = func1(psfmt32[idxof(0)] ^ psfmt32[idxof(mid)]
+	      ^ psfmt32[idxof(N32 - 1)]);
+    psfmt32[idxof(mid)] += r;
+    r += key_length;
+    psfmt32[idxof(mid + lag)] += r;
+    psfmt32[idxof(0)] = r;
+
+    count--;
+    for (i = 1, j = 0; (j < count) && (j < key_length); j++) {
+	r = func1(psfmt32[idxof(i)] ^ psfmt32[idxof((i + mid) % N32)]
+		  ^ psfmt32[idxof((i + N32 - 1) % N32)]);
+	psfmt32[idxof((i + mid) % N32)] += r;
+	r += init_key[j] + i;
+	psfmt32[idxof((i + mid + lag) % N32)] += r;
+	psfmt32[idxof(i)] = r;
+	i = (i + 1) % N32;
+    }
+    for (; j < count; j++) {
+	r = func1(psfmt32[idxof(i)] ^ psfmt32[idxof((i + mid) % N32)]
+		  ^ psfmt32[idxof((i + N32 - 1) % N32)]);
+	psfmt32[idxof((i + mid) % N32)] += r;
+	r += i;
+	psfmt32[idxof((i + mid + lag) % N32)] += r;
+	psfmt32[idxof(i)] = r;
+	i = (i + 1) % N32;
+    }
+    for (j = 0; j < N32; j++) {
+	r = func2(psfmt32[idxof(i)] + psfmt32[idxof((i + mid) % N32)]
+		  + psfmt32[idxof((i + N32 - 1) % N32)]);
+	psfmt32[idxof((i + mid) % N32)] ^= r;
+	r -= i;
+	psfmt32[idxof((i + mid + lag) % N32)] ^= r;
+	psfmt32[idxof(i)] = r;
+	i = (i + 1) % N32;
+    }
+
+    ctx->idx = N32;
+    period_certification(ctx);
+    ctx->initialized = 1;
+
+    return ctx;
+}
+
+void fini_gen_rand(sfmt_t *ctx) {
+    assert(ctx != NULL);
+
+    ctx->initialized = 0;
+    free(ctx);
+}
diff --git a/memkind-1.10.0/jemalloc/test/src/btalloc.c b/memkind-1.10.0/jemalloc/test/src/btalloc.c
new file mode 100644
index 0000000..d570952
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/btalloc.c
@@ -0,0 +1,6 @@
+#include "test/jemalloc_test.h"
+
+void *
+btalloc(size_t size, unsigned bits) {
+	return btalloc_0(size, bits);
+}
diff --git a/memkind-1.10.0/jemalloc/test/src/btalloc_0.c b/memkind-1.10.0/jemalloc/test/src/btalloc_0.c
new file mode 100644
index 0000000..77d8904
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/btalloc_0.c
@@ -0,0 +1,3 @@
+#include "test/jemalloc_test.h"
+
+btalloc_n_gen(0)
diff --git a/memkind-1.10.0/jemalloc/test/src/btalloc_1.c b/memkind-1.10.0/jemalloc/test/src/btalloc_1.c
new file mode 100644
index 0000000..4c126c3
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/btalloc_1.c
@@ -0,0 +1,3 @@
+#include "test/jemalloc_test.h"
+
+btalloc_n_gen(1)
diff --git a/memkind-1.10.0/jemalloc/test/src/math.c b/memkind-1.10.0/jemalloc/test/src/math.c
new file mode 100644
index 0000000..1758c67
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/math.c
@@ -0,0 +1,2 @@
+#define MATH_C_
+#include "test/jemalloc_test.h"
diff --git a/memkind-1.10.0/jemalloc/test/src/mq.c b/memkind-1.10.0/jemalloc/test/src/mq.c
new file mode 100644
index 0000000..9b5f672
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/mq.c
@@ -0,0 +1,27 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * Sleep for approximately ns nanoseconds.  No lower *nor* upper bound on sleep
+ * time is guaranteed.
+ */
+void
+mq_nanosleep(unsigned ns) {
+	assert(ns <= 1000*1000*1000);
+
+#ifdef _WIN32
+	Sleep(ns / 1000);
+#else
+	{
+		struct timespec timeout;
+
+		if (ns < 1000*1000*1000) {
+			timeout.tv_sec = 0;
+			timeout.tv_nsec = ns;
+		} else {
+			timeout.tv_sec = 1;
+			timeout.tv_nsec = 0;
+		}
+		nanosleep(&timeout, NULL);
+	}
+#endif
+}
diff --git a/memkind-1.10.0/jemalloc/test/src/mtx.c b/memkind-1.10.0/jemalloc/test/src/mtx.c
new file mode 100644
index 0000000..d9ce375
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/mtx.c
@@ -0,0 +1,61 @@
+#include "test/jemalloc_test.h"
+
+#ifndef _CRT_SPINCOUNT
+#define _CRT_SPINCOUNT 4000
+#endif
+
+bool
+mtx_init(mtx_t *mtx) {
+#ifdef _WIN32
+	if (!InitializeCriticalSectionAndSpinCount(&mtx->lock,
+	    _CRT_SPINCOUNT)) {
+		return true;
+	}
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	mtx->lock = OS_UNFAIR_LOCK_INIT;
+#else
+	pthread_mutexattr_t attr;
+
+	if (pthread_mutexattr_init(&attr) != 0) {
+		return true;
+	}
+	pthread_mutexattr_settype(&attr, PTHREAD_MUTEX_DEFAULT);
+	if (pthread_mutex_init(&mtx->lock, &attr) != 0) {
+		pthread_mutexattr_destroy(&attr);
+		return true;
+	}
+	pthread_mutexattr_destroy(&attr);
+#endif
+	return false;
+}
+
+void
+mtx_fini(mtx_t *mtx) {
+#ifdef _WIN32
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#else
+	pthread_mutex_destroy(&mtx->lock);
+#endif
+}
+
+void
+mtx_lock(mtx_t *mtx) {
+#ifdef _WIN32
+	EnterCriticalSection(&mtx->lock);
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	os_unfair_lock_lock(&mtx->lock);
+#else
+	pthread_mutex_lock(&mtx->lock);
+#endif
+}
+
+void
+mtx_unlock(mtx_t *mtx) {
+#ifdef _WIN32
+	LeaveCriticalSection(&mtx->lock);
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+	os_unfair_lock_unlock(&mtx->lock);
+#else
+	pthread_mutex_unlock(&mtx->lock);
+#endif
+}
diff --git a/memkind-1.10.0/jemalloc/test/src/test.c b/memkind-1.10.0/jemalloc/test/src/test.c
new file mode 100644
index 0000000..f97ce4d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/test.c
@@ -0,0 +1,234 @@
+#include "test/jemalloc_test.h"
+
+/* Test status state. */
+
+static unsigned		test_count = 0;
+static test_status_t	test_counts[test_status_count] = {0, 0, 0};
+static test_status_t	test_status = test_status_pass;
+static const char *	test_name = "";
+
+/* Reentrancy testing helpers. */
+
+#define NUM_REENTRANT_ALLOCS 20
+typedef enum {
+	non_reentrant = 0,
+	libc_reentrant = 1,
+	arena_new_reentrant = 2
+} reentrancy_t;
+static reentrancy_t reentrancy;
+
+static bool libc_hook_ran = false;
+static bool arena_new_hook_ran = false;
+
+static const char *
+reentrancy_t_str(reentrancy_t r) {
+	switch (r) {
+	case non_reentrant:
+		return "non-reentrant";
+	case libc_reentrant:
+		return "libc-reentrant";
+	case arena_new_reentrant:
+		return "arena_new-reentrant";
+	default:
+		unreachable();
+	}
+}
+
+static void
+do_hook(bool *hook_ran, void (**hook)()) {
+	*hook_ran = true;
+	*hook = NULL;
+
+	size_t alloc_size = 1;
+	for (int i = 0; i < NUM_REENTRANT_ALLOCS; i++) {
+		free(malloc(alloc_size));
+		alloc_size *= 2;
+	}
+}
+
+static void
+libc_reentrancy_hook() {
+	do_hook(&libc_hook_ran, &test_hooks_libc_hook);
+}
+
+static void
+arena_new_reentrancy_hook() {
+	do_hook(&arena_new_hook_ran, &test_hooks_arena_new_hook);
+}
+
+/* Actual test infrastructure. */
+bool
+test_is_reentrant() {
+	return reentrancy != non_reentrant;
+}
+
+JEMALLOC_FORMAT_PRINTF(1, 2)
+void
+test_skip(const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+	malloc_printf("\n");
+	test_status = test_status_skip;
+}
+
+JEMALLOC_FORMAT_PRINTF(1, 2)
+void
+test_fail(const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+	malloc_printf("\n");
+	test_status = test_status_fail;
+}
+
+static const char *
+test_status_string(test_status_t test_status) {
+	switch (test_status) {
+	case test_status_pass: return "pass";
+	case test_status_skip: return "skip";
+	case test_status_fail: return "fail";
+	default: not_reached();
+	}
+}
+
+void
+p_test_init(const char *name) {
+	test_count++;
+	test_status = test_status_pass;
+	test_name = name;
+}
+
+void
+p_test_fini(void) {
+	test_counts[test_status]++;
+	malloc_printf("%s (%s): %s\n", test_name, reentrancy_t_str(reentrancy),
+	    test_status_string(test_status));
+}
+
+static void
+check_global_slow(test_status_t *status) {
+#ifdef JEMALLOC_UNIT_TEST
+	/*
+	 * This check needs to peek into tsd internals, which is why it's only
+	 * exposed in unit tests.
+	 */
+	if (tsd_global_slow()) {
+		malloc_printf("Testing increased global slow count\n");
+		*status = test_status_fail;
+	}
+#endif
+}
+
+static test_status_t
+p_test_impl(bool do_malloc_init, bool do_reentrant, test_t *t, va_list ap) {
+	test_status_t ret;
+
+	if (do_malloc_init) {
+		/*
+		 * Make sure initialization occurs prior to running tests.
+		 * Tests are special because they may use internal facilities
+		 * prior to triggering initialization as a side effect of
+		 * calling into the public API.
+		 */
+		if (nallocx(1, 0) == 0) {
+			malloc_printf("Initialization error");
+			return test_status_fail;
+		}
+	}
+
+	ret = test_status_pass;
+	for (; t != NULL; t = va_arg(ap, test_t *)) {
+		/* Non-reentrant run. */
+		reentrancy = non_reentrant;
+		test_hooks_arena_new_hook = test_hooks_libc_hook = NULL;
+		t();
+		if (test_status > ret) {
+			ret = test_status;
+		}
+		check_global_slow(&ret);
+		/* Reentrant run. */
+		if (do_reentrant) {
+			reentrancy = libc_reentrant;
+			test_hooks_arena_new_hook = NULL;
+			test_hooks_libc_hook = &libc_reentrancy_hook;
+			t();
+			if (test_status > ret) {
+				ret = test_status;
+			}
+			check_global_slow(&ret);
+
+			reentrancy = arena_new_reentrant;
+			test_hooks_libc_hook = NULL;
+			test_hooks_arena_new_hook = &arena_new_reentrancy_hook;
+			t();
+			if (test_status > ret) {
+				ret = test_status;
+			}
+			check_global_slow(&ret);
+		}
+	}
+
+	malloc_printf("--- %s: %u/%u, %s: %u/%u, %s: %u/%u ---\n",
+	    test_status_string(test_status_pass),
+	    test_counts[test_status_pass], test_count,
+	    test_status_string(test_status_skip),
+	    test_counts[test_status_skip], test_count,
+	    test_status_string(test_status_fail),
+	    test_counts[test_status_fail], test_count);
+
+	return ret;
+}
+
+test_status_t
+p_test(test_t *t, ...) {
+	test_status_t ret;
+	va_list ap;
+
+	ret = test_status_pass;
+	va_start(ap, t);
+	ret = p_test_impl(true, true, t, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+test_status_t
+p_test_no_reentrancy(test_t *t, ...) {
+	test_status_t ret;
+	va_list ap;
+
+	ret = test_status_pass;
+	va_start(ap, t);
+	ret = p_test_impl(true, false, t, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+test_status_t
+p_test_no_malloc_init(test_t *t, ...) {
+	test_status_t ret;
+	va_list ap;
+
+	ret = test_status_pass;
+	va_start(ap, t);
+	/*
+	 * We also omit reentrancy from bootstrapping tests, since we don't
+	 * (yet) care about general reentrancy during bootstrapping.
+	 */
+	ret = p_test_impl(false, false, t, ap);
+	va_end(ap);
+
+	return ret;
+}
+
+void
+p_test_fail(const char *prefix, const char *message) {
+	malloc_cprintf(NULL, NULL, "%s%s\n", prefix, message);
+	test_status = test_status_fail;
+}
diff --git a/memkind-1.10.0/jemalloc/test/src/thd.c b/memkind-1.10.0/jemalloc/test/src/thd.c
new file mode 100644
index 0000000..9a15eab
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/thd.c
@@ -0,0 +1,34 @@
+#include "test/jemalloc_test.h"
+
+#ifdef _WIN32
+void
+thd_create(thd_t *thd, void *(*proc)(void *), void *arg) {
+	LPTHREAD_START_ROUTINE routine = (LPTHREAD_START_ROUTINE)proc;
+	*thd = CreateThread(NULL, 0, routine, arg, 0, NULL);
+	if (*thd == NULL) {
+		test_fail("Error in CreateThread()\n");
+	}
+}
+
+void
+thd_join(thd_t thd, void **ret) {
+	if (WaitForSingleObject(thd, INFINITE) == WAIT_OBJECT_0 && ret) {
+		DWORD exit_code;
+		GetExitCodeThread(thd, (LPDWORD) &exit_code);
+		*ret = (void *)(uintptr_t)exit_code;
+	}
+}
+
+#else
+void
+thd_create(thd_t *thd, void *(*proc)(void *), void *arg) {
+	if (pthread_create(thd, NULL, proc, arg) != 0) {
+		test_fail("Error in pthread_create()\n");
+	}
+}
+
+void
+thd_join(thd_t thd, void **ret) {
+	pthread_join(thd, ret);
+}
+#endif
diff --git a/memkind-1.10.0/jemalloc/test/src/timer.c b/memkind-1.10.0/jemalloc/test/src/timer.c
new file mode 100644
index 0000000..c451c63
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/src/timer.c
@@ -0,0 +1,56 @@
+#include "test/jemalloc_test.h"
+
+void
+timer_start(timedelta_t *timer) {
+	nstime_init(&timer->t0, 0);
+	nstime_update(&timer->t0);
+}
+
+void
+timer_stop(timedelta_t *timer) {
+	nstime_copy(&timer->t1, &timer->t0);
+	nstime_update(&timer->t1);
+}
+
+uint64_t
+timer_usec(const timedelta_t *timer) {
+	nstime_t delta;
+
+	nstime_copy(&delta, &timer->t1);
+	nstime_subtract(&delta, &timer->t0);
+	return nstime_ns(&delta) / 1000;
+}
+
+void
+timer_ratio(timedelta_t *a, timedelta_t *b, char *buf, size_t buflen) {
+	uint64_t t0 = timer_usec(a);
+	uint64_t t1 = timer_usec(b);
+	uint64_t mult;
+	size_t i = 0;
+	size_t j, n;
+
+	/* Whole. */
+	n = malloc_snprintf(&buf[i], buflen-i, "%"FMTu64, t0 / t1);
+	i += n;
+	if (i >= buflen) {
+		return;
+	}
+	mult = 1;
+	for (j = 0; j < n; j++) {
+		mult *= 10;
+	}
+
+	/* Decimal. */
+	n = malloc_snprintf(&buf[i], buflen-i, ".");
+	i += n;
+
+	/* Fraction. */
+	while (i < buflen-1) {
+		uint64_t round = (i+1 == buflen-1 && ((t0 * mult * 10 / t1) % 10
+		    >= 5)) ? 1 : 0;
+		n = malloc_snprintf(&buf[i], buflen-i,
+		    "%"FMTu64, (t0 * mult / t1) % 10 + round);
+		i += n;
+		mult *= 10;
+	}
+}
diff --git a/memkind-1.10.0/jemalloc/test/stress/hookbench.c b/memkind-1.10.0/jemalloc/test/stress/hookbench.c
new file mode 100644
index 0000000..97e90b0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/stress/hookbench.c
@@ -0,0 +1,73 @@
+#include "test/jemalloc_test.h"
+
+static void
+noop_alloc_hook(void *extra, hook_alloc_t type, void *result,
+    uintptr_t result_raw, uintptr_t args_raw[3]) {
+}
+
+static void
+noop_dalloc_hook(void *extra, hook_dalloc_t type, void *address,
+    uintptr_t args_raw[3]) {
+}
+
+static void
+noop_expand_hook(void *extra, hook_expand_t type, void *address,
+    size_t old_usize, size_t new_usize, uintptr_t result_raw,
+    uintptr_t args_raw[4]) {
+}
+
+static void
+malloc_free_loop(int iters) {
+	for (int i = 0; i < iters; i++) {
+		void *p = mallocx(1, 0);
+		free(p);
+	}
+}
+
+static void
+test_hooked(int iters) {
+	hooks_t hooks = {&noop_alloc_hook, &noop_dalloc_hook, &noop_expand_hook,
+		NULL};
+
+	int err;
+	void *handles[HOOK_MAX];
+	size_t sz = sizeof(handles[0]);
+
+	for (int i = 0; i < HOOK_MAX; i++) {
+		err = mallctl("experimental.hooks.install", &handles[i],
+		    &sz, &hooks, sizeof(hooks));
+		assert(err == 0);
+
+		timedelta_t timer;
+		timer_start(&timer);
+		malloc_free_loop(iters);
+		timer_stop(&timer);
+		malloc_printf("With %d hook%s: %"FMTu64"us\n", i + 1,
+		    i + 1 == 1 ? "" : "s", timer_usec(&timer));
+	}
+	for (int i = 0; i < HOOK_MAX; i++) {
+		err = mallctl("experimental.hooks.remove", NULL, NULL,
+		    &handles[i], sizeof(handles[i]));
+		assert(err == 0);
+	}
+}
+
+static void
+test_unhooked(int iters) {
+	timedelta_t timer;
+	timer_start(&timer);
+	malloc_free_loop(iters);
+	timer_stop(&timer);
+
+	malloc_printf("Without hooks: %"FMTu64"us\n", timer_usec(&timer));
+}
+
+int
+main(void) {
+	/* Initialize */
+	free(mallocx(1, 0));
+	int iters = 10 * 1000 * 1000;
+	malloc_printf("Benchmarking hooks with %d iterations:\n", iters);
+	test_hooked(iters);
+	test_unhooked(iters);
+}
diff --git a/memkind-1.10.0/jemalloc/test/stress/microbench.c b/memkind-1.10.0/jemalloc/test/stress/microbench.c
new file mode 100644
index 0000000..988b793
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/stress/microbench.c
@@ -0,0 +1,165 @@
+#include "test/jemalloc_test.h"
+
+static inline void
+time_func(timedelta_t *timer, uint64_t nwarmup, uint64_t niter,
+    void (*func)(void)) {
+	uint64_t i;
+
+	for (i = 0; i < nwarmup; i++) {
+		func();
+	}
+	timer_start(timer);
+	for (i = 0; i < niter; i++) {
+		func();
+	}
+	timer_stop(timer);
+}
+
+void
+compare_funcs(uint64_t nwarmup, uint64_t niter, const char *name_a,
+    void (*func_a), const char *name_b, void (*func_b)) {
+	timedelta_t timer_a, timer_b;
+	char ratio_buf[6];
+	void *p;
+
+	p = mallocx(1, 0);
+	if (p == NULL) {
+		test_fail("Unexpected mallocx() failure");
+		return;
+	}
+
+	time_func(&timer_a, nwarmup, niter, func_a);
+	time_func(&timer_b, nwarmup, niter, func_b);
+
+	timer_ratio(&timer_a, &timer_b, ratio_buf, sizeof(ratio_buf));
+	malloc_printf("%"FMTu64" iterations, %s=%"FMTu64"us, "
+	    "%s=%"FMTu64"us, ratio=1:%s\n",
+	    niter, name_a, timer_usec(&timer_a), name_b, timer_usec(&timer_b),
+	    ratio_buf);
+
+	dallocx(p, 0);
+}
+
+static void
+malloc_free(void) {
+	/* The compiler can optimize away free(malloc(1))! */
+	void *p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	free(p);
+}
+
+static void
+mallocx_free(void) {
+	void *p = mallocx(1, 0);
+	if (p == NULL) {
+		test_fail("Unexpected mallocx() failure");
+		return;
+	}
+	free(p);
+}
+
+TEST_BEGIN(test_malloc_vs_mallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "malloc",
+	    malloc_free, "mallocx", mallocx_free);
+}
+TEST_END
+
+static void
+malloc_dallocx(void) {
+	void *p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	dallocx(p, 0);
+}
+
+static void
+malloc_sdallocx(void) {
+	void *p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	sdallocx(p, 1, 0);
+}
+
+TEST_BEGIN(test_free_vs_dallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "free", malloc_free,
+	    "dallocx", malloc_dallocx);
+}
+TEST_END
+
+TEST_BEGIN(test_dallocx_vs_sdallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "dallocx", malloc_dallocx,
+	    "sdallocx", malloc_sdallocx);
+}
+TEST_END
+
+static void
+malloc_mus_free(void) {
+	void *p;
+
+	p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	malloc_usable_size(p);
+	free(p);
+}
+
+static void
+malloc_sallocx_free(void) {
+	void *p;
+
+	p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	if (sallocx(p, 0) < 1) {
+		test_fail("Unexpected sallocx() failure");
+	}
+	free(p);
+}
+
+TEST_BEGIN(test_mus_vs_sallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "malloc_usable_size",
+	    malloc_mus_free, "sallocx", malloc_sallocx_free);
+}
+TEST_END
+
+static void
+malloc_nallocx_free(void) {
+	void *p;
+
+	p = malloc(1);
+	if (p == NULL) {
+		test_fail("Unexpected malloc() failure");
+		return;
+	}
+	if (nallocx(1, 0) < 1) {
+		test_fail("Unexpected nallocx() failure");
+	}
+	free(p);
+}
+
+TEST_BEGIN(test_sallocx_vs_nallocx) {
+	compare_funcs(10*1000*1000, 100*1000*1000, "sallocx",
+	    malloc_sallocx_free, "nallocx", malloc_nallocx_free);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_malloc_vs_mallocx,
+	    test_free_vs_dallocx,
+	    test_dallocx_vs_sdallocx,
+	    test_mus_vs_sallocx,
+	    test_sallocx_vs_nallocx);
+}
diff --git a/memkind-1.10.0/jemalloc/test/test.sh.in b/memkind-1.10.0/jemalloc/test/test.sh.in
new file mode 100644
index 0000000..39302ff
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/test.sh.in
@@ -0,0 +1,80 @@
+#!/bin/sh
+
+case @abi@ in
+  macho)
+    export DYLD_FALLBACK_LIBRARY_PATH="@objroot@lib"
+    ;;
+  pecoff)
+    export PATH="${PATH}:@objroot@lib"
+    ;;
+  *)
+    ;;
+esac
+
+# Make a copy of the @JEMALLOC_CPREFIX@MALLOC_CONF passed in to this script, so
+# it can be repeatedly concatenated with per test settings.
+export MALLOC_CONF_ALL=${@JEMALLOC_CPREFIX@MALLOC_CONF}
+# Concatenate the individual test's MALLOC_CONF and MALLOC_CONF_ALL.
+export_malloc_conf() {
+  if [ "x${MALLOC_CONF}" != "x" -a "x${MALLOC_CONF_ALL}" != "x" ] ; then
+    export @JEMALLOC_CPREFIX@MALLOC_CONF="${MALLOC_CONF},${MALLOC_CONF_ALL}"
+  else
+    export @JEMALLOC_CPREFIX@MALLOC_CONF="${MALLOC_CONF}${MALLOC_CONF_ALL}"
+  fi
+}
+
+# Corresponds to test_status_t.
+pass_code=0
+skip_code=1
+fail_code=2
+
+pass_count=0
+skip_count=0
+fail_count=0
+for t in $@; do
+  if [ $pass_count -ne 0 -o $skip_count -ne 0 -o $fail_count != 0 ] ; then
+    echo
+  fi
+  echo "=== ${t} ==="
+  if [ -e "@srcroot@${t}.sh" ] ; then
+    # Source the shell script corresponding to the test in a subshell and
+    # execute the test.  This allows the shell script to set MALLOC_CONF, which
+    # is then used to set @JEMALLOC_CPREFIX@MALLOC_CONF (thus allowing the
+    # per test shell script to ignore the @JEMALLOC_CPREFIX@ detail).
+    enable_fill=@enable_fill@ \
+    enable_prof=@enable_prof@ \
+    . @srcroot@${t}.sh && \
+    export_malloc_conf && \
+    $JEMALLOC_TEST_PREFIX ${t}@exe@ @abs_srcroot@ @abs_objroot@
+  else
+    export MALLOC_CONF= && \
+    export_malloc_conf && \
+    $JEMALLOC_TEST_PREFIX ${t}@exe@ @abs_srcroot@ @abs_objroot@
+  fi
+  result_code=$?
+  case ${result_code} in
+    ${pass_code})
+      pass_count=$((pass_count+1))
+      ;;
+    ${skip_code})
+      skip_count=$((skip_count+1))
+      ;;
+    ${fail_code})
+      fail_count=$((fail_count+1))
+      ;;
+    *)
+      echo "Test harness error: ${t} w/ MALLOC_CONF=\"${MALLOC_CONF}\"" 1>&2
+      echo "Use prefix to debug, e.g. JEMALLOC_TEST_PREFIX=\"gdb --args\" sh test/test.sh ${t}" 1>&2
+      exit 1
+  esac
+done
+
+total_count=`expr ${pass_count} + ${skip_count} + ${fail_count}`
+echo
+echo "Test suite summary: pass: ${pass_count}/${total_count}, skip: ${skip_count}/${total_count}, fail: ${fail_count}/${total_count}"
+
+if [ ${fail_count} -eq 0 ] ; then
+  exit 0
+else
+  exit 1
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/SFMT.c b/memkind-1.10.0/jemalloc/test/unit/SFMT.c
new file mode 100644
index 0000000..1fc8cf1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/SFMT.c
@@ -0,0 +1,1599 @@
+/*
+ * This file derives from SFMT 1.3.3
+ * (http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/SFMT/index.html), which was
+ * released under the terms of the following license:
+ *
+ *   Copyright (c) 2006,2007 Mutsuo Saito, Makoto Matsumoto and Hiroshima
+ *   University. All rights reserved.
+ *
+ *   Redistribution and use in source and binary forms, with or without
+ *   modification, are permitted provided that the following conditions are
+ *   met:
+ *
+ *       * Redistributions of source code must retain the above copyright
+ *         notice, this list of conditions and the following disclaimer.
+ *       * Redistributions in binary form must reproduce the above
+ *         copyright notice, this list of conditions and the following
+ *         disclaimer in the documentation and/or other materials provided
+ *         with the distribution.
+ *       * Neither the name of the Hiroshima University nor the names of
+ *         its contributors may be used to endorse or promote products
+ *         derived from this software without specific prior written
+ *         permission.
+ *
+ *   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ *   "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ *   LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ *   A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ *   OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ *   SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ *   LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ *   DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ *   THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ *   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ *   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "test/jemalloc_test.h"
+
+#define BLOCK_SIZE 10000
+#define BLOCK_SIZE64 (BLOCK_SIZE / 2)
+#define COUNT_1 1000
+#define COUNT_2 700
+
+static const uint32_t init_gen_rand_32_expected[] = {
+	3440181298U, 1564997079U, 1510669302U, 2930277156U, 1452439940U,
+	3796268453U,  423124208U, 2143818589U, 3827219408U, 2987036003U,
+	2674978610U, 1536842514U, 2027035537U, 2534897563U, 1686527725U,
+	 545368292U, 1489013321U, 1370534252U, 4231012796U, 3994803019U,
+	1764869045U,  824597505U,  862581900U, 2469764249U,  812862514U,
+	 359318673U,  116957936U, 3367389672U, 2327178354U, 1898245200U,
+	3206507879U, 2378925033U, 1040214787U, 2524778605U, 3088428700U,
+	1417665896U,  964324147U, 2282797708U, 2456269299U,  313400376U,
+	2245093271U, 1015729427U, 2694465011U, 3246975184U, 1992793635U,
+	 463679346U, 3721104591U, 3475064196U,  856141236U, 1499559719U,
+	3522818941U, 3721533109U, 1954826617U, 1282044024U, 1543279136U,
+	1301863085U, 2669145051U, 4221477354U, 3896016841U, 3392740262U,
+	 462466863U, 1037679449U, 1228140306U,  922298197U, 1205109853U,
+	1872938061U, 3102547608U, 2742766808U, 1888626088U, 4028039414U,
+	 157593879U, 1136901695U, 4038377686U, 3572517236U, 4231706728U,
+	2997311961U, 1189931652U, 3981543765U, 2826166703U,   87159245U,
+	1721379072U, 3897926942U, 1790395498U, 2569178939U, 1047368729U,
+	2340259131U, 3144212906U, 2301169789U, 2442885464U, 3034046771U,
+	3667880593U, 3935928400U, 2372805237U, 1666397115U, 2460584504U,
+	 513866770U, 3810869743U, 2147400037U, 2792078025U, 2941761810U,
+	3212265810U,  984692259U,  346590253U, 1804179199U, 3298543443U,
+	 750108141U, 2880257022U,  243310542U, 1869036465U, 1588062513U,
+	2983949551U, 1931450364U, 4034505847U, 2735030199U, 1628461061U,
+	2539522841U,  127965585U, 3992448871U,  913388237U,  559130076U,
+	1202933193U, 4087643167U, 2590021067U, 2256240196U, 1746697293U,
+	1013913783U, 1155864921U, 2715773730U,  915061862U, 1948766573U,
+	2322882854U, 3761119102U, 1343405684U, 3078711943U, 3067431651U,
+	3245156316U, 3588354584U, 3484623306U, 3899621563U, 4156689741U,
+	3237090058U, 3880063844U,  862416318U, 4039923869U, 2303788317U,
+	3073590536U,  701653667U, 2131530884U, 3169309950U, 2028486980U,
+	 747196777U, 3620218225U,  432016035U, 1449580595U, 2772266392U,
+	 444224948U, 1662832057U, 3184055582U, 3028331792U, 1861686254U,
+	1104864179U,  342430307U, 1350510923U, 3024656237U, 1028417492U,
+	2870772950U,  290847558U, 3675663500U,  508431529U, 4264340390U,
+	2263569913U, 1669302976U,  519511383U, 2706411211U, 3764615828U,
+	3883162495U, 4051445305U, 2412729798U, 3299405164U, 3991911166U,
+	2348767304U, 2664054906U, 3763609282U,  593943581U, 3757090046U,
+	2075338894U, 2020550814U, 4287452920U, 4290140003U, 1422957317U,
+	2512716667U, 2003485045U, 2307520103U, 2288472169U, 3940751663U,
+	4204638664U, 2892583423U, 1710068300U, 3904755993U, 2363243951U,
+	3038334120U,  547099465U,  771105860U, 3199983734U, 4282046461U,
+	2298388363U,  934810218U, 2837827901U, 3952500708U, 2095130248U,
+	3083335297U,   26885281U, 3932155283U, 1531751116U, 1425227133U,
+	 495654159U, 3279634176U, 3855562207U, 3957195338U, 4159985527U,
+	 893375062U, 1875515536U, 1327247422U, 3754140693U, 1028923197U,
+	1729880440U,  805571298U,  448971099U, 2726757106U, 2749436461U,
+	2485987104U,  175337042U, 3235477922U, 3882114302U, 2020970972U,
+	 943926109U, 2762587195U, 1904195558U, 3452650564U,  108432281U,
+	3893463573U, 3977583081U, 2636504348U, 1110673525U, 3548479841U,
+	4258854744U,  980047703U, 4057175418U, 3890008292U,  145653646U,
+	3141868989U, 3293216228U, 1194331837U, 1254570642U, 3049934521U,
+	2868313360U, 2886032750U, 1110873820U,  279553524U, 3007258565U,
+	1104807822U, 3186961098U,  315764646U, 2163680838U, 3574508994U,
+	3099755655U,  191957684U, 3642656737U, 3317946149U, 3522087636U,
+	 444526410U,  779157624U, 1088229627U, 1092460223U, 1856013765U,
+	3659877367U,  368270451U,  503570716U, 3000984671U, 2742789647U,
+	 928097709U, 2914109539U,  308843566U, 2816161253U, 3667192079U,
+	2762679057U, 3395240989U, 2928925038U, 1491465914U, 3458702834U,
+	3787782576U, 2894104823U, 1296880455U, 1253636503U,  989959407U,
+	2291560361U, 2776790436U, 1913178042U, 1584677829U,  689637520U,
+	1898406878U,  688391508U, 3385234998U,  845493284U, 1943591856U,
+	2720472050U,  222695101U, 1653320868U, 2904632120U, 4084936008U,
+	1080720688U, 3938032556U,  387896427U, 2650839632U,   99042991U,
+	1720913794U, 1047186003U, 1877048040U, 2090457659U,  517087501U,
+	4172014665U, 2129713163U, 2413533132U, 2760285054U, 4129272496U,
+	1317737175U, 2309566414U, 2228873332U, 3889671280U, 1110864630U,
+	3576797776U, 2074552772U,  832002644U, 3097122623U, 2464859298U,
+	2679603822U, 1667489885U, 3237652716U, 1478413938U, 1719340335U,
+	2306631119U,  639727358U, 3369698270U,  226902796U, 2099920751U,
+	1892289957U, 2201594097U, 3508197013U, 3495811856U, 3900381493U,
+	 841660320U, 3974501451U, 3360949056U, 1676829340U,  728899254U,
+	2047809627U, 2390948962U,  670165943U, 3412951831U, 4189320049U,
+	1911595255U, 2055363086U,  507170575U,  418219594U, 4141495280U,
+	2692088692U, 4203630654U, 3540093932U,  791986533U, 2237921051U,
+	2526864324U, 2956616642U, 1394958700U, 1983768223U, 1893373266U,
+	 591653646U,  228432437U, 1611046598U, 3007736357U, 1040040725U,
+	2726180733U, 2789804360U, 4263568405U,  829098158U, 3847722805U,
+	1123578029U, 1804276347U,  997971319U, 4203797076U, 4185199713U,
+	2811733626U, 2343642194U, 2985262313U, 1417930827U, 3759587724U,
+	1967077982U, 1585223204U, 1097475516U, 1903944948U,  740382444U,
+	1114142065U, 1541796065U, 1718384172U, 1544076191U, 1134682254U,
+	3519754455U, 2866243923U,  341865437U,  645498576U, 2690735853U,
+	1046963033U, 2493178460U, 1187604696U, 1619577821U,  488503634U,
+	3255768161U, 2306666149U, 1630514044U, 2377698367U, 2751503746U,
+	3794467088U, 1796415981U, 3657173746U,  409136296U, 1387122342U,
+	1297726519U,  219544855U, 4270285558U,  437578827U, 1444698679U,
+	2258519491U,  963109892U, 3982244073U, 3351535275U,  385328496U,
+	1804784013U,  698059346U, 3920535147U,  708331212U,  784338163U,
+	 785678147U, 1238376158U, 1557298846U, 2037809321U,  271576218U,
+	4145155269U, 1913481602U, 2763691931U,  588981080U, 1201098051U,
+	3717640232U, 1509206239U,  662536967U, 3180523616U, 1133105435U,
+	2963500837U, 2253971215U, 3153642623U, 1066925709U, 2582781958U,
+	3034720222U, 1090798544U, 2942170004U, 4036187520U,  686972531U,
+	2610990302U, 2641437026U, 1837562420U,  722096247U, 1315333033U,
+	2102231203U, 3402389208U, 3403698140U, 1312402831U, 2898426558U,
+	 814384596U,  385649582U, 1916643285U, 1924625106U, 2512905582U,
+	2501170304U, 4275223366U, 2841225246U, 1467663688U, 3563567847U,
+	2969208552U,  884750901U,  102992576U,  227844301U, 3681442994U,
+	3502881894U, 4034693299U, 1166727018U, 1697460687U, 1737778332U,
+	1787161139U, 1053003655U, 1215024478U, 2791616766U, 2525841204U,
+	1629323443U,    3233815U, 2003823032U, 3083834263U, 2379264872U,
+	3752392312U, 1287475550U, 3770904171U, 3004244617U, 1502117784U,
+	 918698423U, 2419857538U, 3864502062U, 1751322107U, 2188775056U,
+	4018728324U,  983712955U,  440071928U, 3710838677U, 2001027698U,
+	3994702151U,   22493119U, 3584400918U, 3446253670U, 4254789085U,
+	1405447860U, 1240245579U, 1800644159U, 1661363424U, 3278326132U,
+	3403623451U,   67092802U, 2609352193U, 3914150340U, 1814842761U,
+	3610830847U,  591531412U, 3880232807U, 1673505890U, 2585326991U,
+	1678544474U, 3148435887U, 3457217359U, 1193226330U, 2816576908U,
+	 154025329U,  121678860U, 1164915738U,  973873761U,  269116100U,
+	  52087970U,  744015362U,  498556057U,   94298882U, 1563271621U,
+	2383059628U, 4197367290U, 3958472990U, 2592083636U, 2906408439U,
+	1097742433U, 3924840517U,  264557272U, 2292287003U, 3203307984U,
+	4047038857U, 3820609705U, 2333416067U, 1839206046U, 3600944252U,
+	3412254904U,  583538222U, 2390557166U, 4140459427U, 2810357445U,
+	 226777499U, 2496151295U, 2207301712U, 3283683112U,  611630281U,
+	1933218215U, 3315610954U, 3889441987U, 3719454256U, 3957190521U,
+	1313998161U, 2365383016U, 3146941060U, 1801206260U,  796124080U,
+	2076248581U, 1747472464U, 3254365145U,  595543130U, 3573909503U,
+	3758250204U, 2020768540U, 2439254210U,   93368951U, 3155792250U,
+	2600232980U, 3709198295U, 3894900440U, 2971850836U, 1578909644U,
+	1443493395U, 2581621665U, 3086506297U, 2443465861U,  558107211U,
+	1519367835U,  249149686U,  908102264U, 2588765675U, 1232743965U,
+	1001330373U, 3561331654U, 2259301289U, 1564977624U, 3835077093U,
+	 727244906U, 4255738067U, 1214133513U, 2570786021U, 3899704621U,
+	1633861986U, 1636979509U, 1438500431U,   58463278U, 2823485629U,
+	2297430187U, 2926781924U, 3371352948U, 1864009023U, 2722267973U,
+	1444292075U,  437703973U, 1060414512U,  189705863U,  910018135U,
+	4077357964U,  884213423U, 2644986052U, 3973488374U, 1187906116U,
+	2331207875U,  780463700U, 3713351662U, 3854611290U,  412805574U,
+	2978462572U, 2176222820U,  829424696U, 2790788332U, 2750819108U,
+	1594611657U, 3899878394U, 3032870364U, 1702887682U, 1948167778U,
+	  14130042U,  192292500U,  947227076U,   90719497U, 3854230320U,
+	 784028434U, 2142399787U, 1563449646U, 2844400217U,  819143172U,
+	2883302356U, 2328055304U, 1328532246U, 2603885363U, 3375188924U,
+	 933941291U, 3627039714U, 2129697284U, 2167253953U, 2506905438U,
+	1412424497U, 2981395985U, 1418359660U, 2925902456U,   52752784U,
+	3713667988U, 3924669405U,  648975707U, 1145520213U, 4018650664U,
+	3805915440U, 2380542088U, 2013260958U, 3262572197U, 2465078101U,
+	1114540067U, 3728768081U, 2396958768U,  590672271U,  904818725U,
+	4263660715U,  700754408U, 1042601829U, 4094111823U, 4274838909U,
+	2512692617U, 2774300207U, 2057306915U, 3470942453U,   99333088U,
+	1142661026U, 2889931380U,   14316674U, 2201179167U,  415289459U,
+	 448265759U, 3515142743U, 3254903683U,  246633281U, 1184307224U,
+	2418347830U, 2092967314U, 2682072314U, 2558750234U, 2000352263U,
+	1544150531U,  399010405U, 1513946097U,  499682937U,  461167460U,
+	3045570638U, 1633669705U,  851492362U, 4052801922U, 2055266765U,
+	 635556996U,  368266356U, 2385737383U, 3218202352U, 2603772408U,
+	 349178792U,  226482567U, 3102426060U, 3575998268U, 2103001871U,
+	3243137071U,  225500688U, 1634718593U, 4283311431U, 4292122923U,
+	3842802787U,  811735523U,  105712518U,  663434053U, 1855889273U,
+	2847972595U, 1196355421U, 2552150115U, 4254510614U, 3752181265U,
+	3430721819U, 3828705396U, 3436287905U, 3441964937U, 4123670631U,
+	 353001539U,  459496439U, 3799690868U, 1293777660U, 2761079737U,
+	 498096339U, 3398433374U, 4080378380U, 2304691596U, 2995729055U,
+	4134660419U, 3903444024U, 3576494993U,  203682175U, 3321164857U,
+	2747963611U,   79749085U, 2992890370U, 1240278549U, 1772175713U,
+	2111331972U, 2655023449U, 1683896345U, 2836027212U, 3482868021U,
+	2489884874U,  756853961U, 2298874501U, 4013448667U, 4143996022U,
+	2948306858U, 4132920035U, 1283299272U,  995592228U, 3450508595U,
+	1027845759U, 1766942720U, 3861411826U, 1446861231U,   95974993U,
+	3502263554U, 1487532194U,  601502472U, 4129619129U,  250131773U,
+	2050079547U, 3198903947U, 3105589778U, 4066481316U, 3026383978U,
+	2276901713U,  365637751U, 2260718426U, 1394775634U, 1791172338U,
+	2690503163U, 2952737846U, 1568710462U,  732623190U, 2980358000U,
+	1053631832U, 1432426951U, 3229149635U, 1854113985U, 3719733532U,
+	3204031934U,  735775531U,  107468620U, 3734611984U,  631009402U,
+	3083622457U, 4109580626U,  159373458U, 1301970201U, 4132389302U,
+	1293255004U,  847182752U, 4170022737U,   96712900U, 2641406755U,
+	1381727755U,  405608287U, 4287919625U, 1703554290U, 3589580244U,
+	2911403488U,    2166565U, 2647306451U, 2330535117U, 1200815358U,
+	1165916754U,  245060911U, 4040679071U, 3684908771U, 2452834126U,
+	2486872773U, 2318678365U, 2940627908U, 1837837240U, 3447897409U,
+	4270484676U, 1495388728U, 3754288477U, 4204167884U, 1386977705U,
+	2692224733U, 3076249689U, 4109568048U, 4170955115U, 4167531356U,
+	4020189950U, 4261855038U, 3036907575U, 3410399885U, 3076395737U,
+	1046178638U,  144496770U,  230725846U, 3349637149U,   17065717U,
+	2809932048U, 2054581785U, 3608424964U, 3259628808U,  134897388U,
+	3743067463U,  257685904U, 3795656590U, 1562468719U, 3589103904U,
+	3120404710U,  254684547U, 2653661580U, 3663904795U, 2631942758U,
+	1063234347U, 2609732900U, 2332080715U, 3521125233U, 1180599599U,
+	1935868586U, 4110970440U,  296706371U, 2128666368U, 1319875791U,
+	1570900197U, 3096025483U, 1799882517U, 1928302007U, 1163707758U,
+	1244491489U, 3533770203U,  567496053U, 2757924305U, 2781639343U,
+	2818420107U,  560404889U, 2619609724U, 4176035430U, 2511289753U,
+	2521842019U, 3910553502U, 2926149387U, 3302078172U, 4237118867U,
+	 330725126U,  367400677U,  888239854U,  545570454U, 4259590525U,
+	 134343617U, 1102169784U, 1647463719U, 3260979784U, 1518840883U,
+	3631537963U, 3342671457U, 1301549147U, 2083739356U,  146593792U,
+	3217959080U,  652755743U, 2032187193U, 3898758414U, 1021358093U,
+	4037409230U, 2176407931U, 3427391950U, 2883553603U,  985613827U,
+	3105265092U, 3423168427U, 3387507672U,  467170288U, 2141266163U,
+	3723870208U,  916410914U, 1293987799U, 2652584950U,  769160137U,
+	3205292896U, 1561287359U, 1684510084U, 3136055621U, 3765171391U,
+	 639683232U, 2639569327U, 1218546948U, 4263586685U, 3058215773U,
+	2352279820U,  401870217U, 2625822463U, 1529125296U, 2981801895U,
+	1191285226U, 4027725437U, 3432700217U, 4098835661U,  971182783U,
+	2443861173U, 3881457123U, 3874386651U,  457276199U, 2638294160U,
+	4002809368U,  421169044U, 1112642589U, 3076213779U, 3387033971U,
+	2499610950U, 3057240914U, 1662679783U,  461224431U, 1168395933U
+};
+static const uint32_t init_by_array_32_expected[] = {
+	2920711183U, 3885745737U, 3501893680U,  856470934U, 1421864068U,
+	 277361036U, 1518638004U, 2328404353U, 3355513634U,   64329189U,
+	1624587673U, 3508467182U, 2481792141U, 3706480799U, 1925859037U,
+	2913275699U,  882658412U,  384641219U,  422202002U, 1873384891U,
+	2006084383U, 3924929912U, 1636718106U, 3108838742U, 1245465724U,
+	4195470535U,  779207191U, 1577721373U, 1390469554U, 2928648150U,
+	 121399709U, 3170839019U, 4044347501U,  953953814U, 3821710850U,
+	3085591323U, 3666535579U, 3577837737U, 2012008410U, 3565417471U,
+	4044408017U,  433600965U, 1637785608U, 1798509764U,  860770589U,
+	3081466273U, 3982393409U, 2451928325U, 3437124742U, 4093828739U,
+	3357389386U, 2154596123U,  496568176U, 2650035164U, 2472361850U,
+	   3438299U, 2150366101U, 1577256676U, 3802546413U, 1787774626U,
+	4078331588U, 3706103141U,  170391138U, 3806085154U, 1680970100U,
+	1961637521U, 3316029766U,  890610272U, 1453751581U, 1430283664U,
+	3051057411U, 3597003186U,  542563954U, 3796490244U, 1690016688U,
+	3448752238U,  440702173U,  347290497U, 1121336647U, 2540588620U,
+	 280881896U, 2495136428U,  213707396U,   15104824U, 2946180358U,
+	 659000016U,  566379385U, 2614030979U, 2855760170U,  334526548U,
+	2315569495U, 2729518615U,  564745877U, 1263517638U, 3157185798U,
+	1604852056U, 1011639885U, 2950579535U, 2524219188U,  312951012U,
+	1528896652U, 1327861054U, 2846910138U, 3966855905U, 2536721582U,
+	 855353911U, 1685434729U, 3303978929U, 1624872055U, 4020329649U,
+	3164802143U, 1642802700U, 1957727869U, 1792352426U, 3334618929U,
+	2631577923U, 3027156164U,  842334259U, 3353446843U, 1226432104U,
+	1742801369U, 3552852535U, 3471698828U, 1653910186U, 3380330939U,
+	2313782701U, 3351007196U, 2129839995U, 1800682418U, 4085884420U,
+	1625156629U, 3669701987U,  615211810U, 3294791649U, 4131143784U,
+	2590843588U, 3207422808U, 3275066464U,  561592872U, 3957205738U,
+	3396578098U,   48410678U, 3505556445U, 1005764855U, 3920606528U,
+	2936980473U, 2378918600U, 2404449845U, 1649515163U,  701203563U,
+	3705256349U,   83714199U, 3586854132U,  922978446U, 2863406304U,
+	3523398907U, 2606864832U, 2385399361U, 3171757816U, 4262841009U,
+	3645837721U, 1169579486U, 3666433897U, 3174689479U, 1457866976U,
+	3803895110U, 3346639145U, 1907224409U, 1978473712U, 1036712794U,
+	 980754888U, 1302782359U, 1765252468U,  459245755U, 3728923860U,
+	1512894209U, 2046491914U,  207860527U,  514188684U, 2288713615U,
+	1597354672U, 3349636117U, 2357291114U, 3995796221U,  945364213U,
+	1893326518U, 3770814016U, 1691552714U, 2397527410U,  967486361U,
+	 776416472U, 4197661421U,  951150819U, 1852770983U, 4044624181U,
+	1399439738U, 4194455275U, 2284037669U, 1550734958U, 3321078108U,
+	1865235926U, 2912129961U, 2664980877U, 1357572033U, 2600196436U,
+	2486728200U, 2372668724U, 1567316966U, 2374111491U, 1839843570U,
+	  20815612U, 3727008608U, 3871996229U,  824061249U, 1932503978U,
+	3404541726U,  758428924U, 2609331364U, 1223966026U, 1299179808U,
+	 648499352U, 2180134401U,  880821170U, 3781130950U,  113491270U,
+	1032413764U, 4185884695U, 2490396037U, 1201932817U, 4060951446U,
+	4165586898U, 1629813212U, 2887821158U,  415045333U,  628926856U,
+	2193466079U, 3391843445U, 2227540681U, 1907099846U, 2848448395U,
+	1717828221U, 1372704537U, 1707549841U, 2294058813U, 2101214437U,
+	2052479531U, 1695809164U, 3176587306U, 2632770465U,   81634404U,
+	1603220563U,  644238487U,  302857763U,  897352968U, 2613146653U,
+	1391730149U, 4245717312U, 4191828749U, 1948492526U, 2618174230U,
+	3992984522U, 2178852787U, 3596044509U, 3445573503U, 2026614616U,
+	 915763564U, 3415689334U, 2532153403U, 3879661562U, 2215027417U,
+	3111154986U, 2929478371U,  668346391U, 1152241381U, 2632029711U,
+	3004150659U, 2135025926U,  948690501U, 2799119116U, 4228829406U,
+	1981197489U, 4209064138U,  684318751U, 3459397845U,  201790843U,
+	4022541136U, 3043635877U,  492509624U, 3263466772U, 1509148086U,
+	 921459029U, 3198857146U,  705479721U, 3835966910U, 3603356465U,
+	 576159741U, 1742849431U,  594214882U, 2055294343U, 3634861861U,
+	 449571793U, 3246390646U, 3868232151U, 1479156585U, 2900125656U,
+	2464815318U, 3960178104U, 1784261920U,   18311476U, 3627135050U,
+	 644609697U,  424968996U,  919890700U, 2986824110U,  816423214U,
+	4003562844U, 1392714305U, 1757384428U, 2569030598U,  995949559U,
+	3875659880U, 2933807823U, 2752536860U, 2993858466U, 4030558899U,
+	2770783427U, 2775406005U, 2777781742U, 1931292655U,  472147933U,
+	3865853827U, 2726470545U, 2668412860U, 2887008249U,  408979190U,
+	3578063323U, 3242082049U, 1778193530U,   27981909U, 2362826515U,
+	 389875677U, 1043878156U,  581653903U, 3830568952U,  389535942U,
+	3713523185U, 2768373359U, 2526101582U, 1998618197U, 1160859704U,
+	3951172488U, 1098005003U,  906275699U, 3446228002U, 2220677963U,
+	2059306445U,  132199571U,  476838790U, 1868039399U, 3097344807U,
+	 857300945U,  396345050U, 2835919916U, 1782168828U, 1419519470U,
+	4288137521U,  819087232U,  596301494U,  872823172U, 1526888217U,
+	 805161465U, 1116186205U, 2829002754U, 2352620120U,  620121516U,
+	 354159268U, 3601949785U,  209568138U, 1352371732U, 2145977349U,
+	4236871834U, 1539414078U, 3558126206U, 3224857093U, 4164166682U,
+	3817553440U, 3301780278U, 2682696837U, 3734994768U, 1370950260U,
+	1477421202U, 2521315749U, 1330148125U, 1261554731U, 2769143688U,
+	3554756293U, 4235882678U, 3254686059U, 3530579953U, 1215452615U,
+	3574970923U, 4057131421U,  589224178U, 1000098193U,  171190718U,
+	2521852045U, 2351447494U, 2284441580U, 2646685513U, 3486933563U,
+	3789864960U, 1190528160U, 1702536782U, 1534105589U, 4262946827U,
+	2726686826U, 3584544841U, 2348270128U, 2145092281U, 2502718509U,
+	1027832411U, 3571171153U, 1287361161U, 4011474411U, 3241215351U,
+	2419700818U,  971242709U, 1361975763U, 1096842482U, 3271045537U,
+	  81165449U,  612438025U, 3912966678U, 1356929810U,  733545735U,
+	 537003843U, 1282953084U,  884458241U,  588930090U, 3930269801U,
+	2961472450U, 1219535534U, 3632251943U,  268183903U, 1441240533U,
+	3653903360U, 3854473319U, 2259087390U, 2548293048U, 2022641195U,
+	2105543911U, 1764085217U, 3246183186U,  482438805U,  888317895U,
+	2628314765U, 2466219854U,  717546004U, 2322237039U,  416725234U,
+	1544049923U, 1797944973U, 3398652364U, 3111909456U,  485742908U,
+	2277491072U, 1056355088U, 3181001278U,  129695079U, 2693624550U,
+	1764438564U, 3797785470U,  195503713U, 3266519725U, 2053389444U,
+	1961527818U, 3400226523U, 3777903038U, 2597274307U, 4235851091U,
+	4094406648U, 2171410785U, 1781151386U, 1378577117U,  654643266U,
+	3424024173U, 3385813322U,  679385799U,  479380913U,  681715441U,
+	3096225905U,  276813409U, 3854398070U, 2721105350U,  831263315U,
+	3276280337U, 2628301522U, 3984868494U, 1466099834U, 2104922114U,
+	1412672743U,  820330404U, 3491501010U,  942735832U,  710652807U,
+	3972652090U,  679881088U,   40577009U, 3705286397U, 2815423480U,
+	3566262429U,  663396513U, 3777887429U, 4016670678U,  404539370U,
+	1142712925U, 1140173408U, 2913248352U, 2872321286U,  263751841U,
+	3175196073U, 3162557581U, 2878996619U,   75498548U, 3836833140U,
+	3284664959U, 1157523805U,  112847376U,  207855609U, 1337979698U,
+	1222578451U,  157107174U,  901174378U, 3883717063U, 1618632639U,
+	1767889440U, 4264698824U, 1582999313U,  884471997U, 2508825098U,
+	3756370771U, 2457213553U, 3565776881U, 3709583214U,  915609601U,
+	 460833524U, 1091049576U,   85522880U,    2553251U,  132102809U,
+	2429882442U, 2562084610U, 1386507633U, 4112471229U,   21965213U,
+	1981516006U, 2418435617U, 3054872091U, 4251511224U, 2025783543U,
+	1916911512U, 2454491136U, 3938440891U, 3825869115U, 1121698605U,
+	3463052265U,  802340101U, 1912886800U, 4031997367U, 3550640406U,
+	1596096923U,  610150600U,  431464457U, 2541325046U,  486478003U,
+	 739704936U, 2862696430U, 3037903166U, 1129749694U, 2611481261U,
+	1228993498U,  510075548U, 3424962587U, 2458689681U,  818934833U,
+	4233309125U, 1608196251U, 3419476016U, 1858543939U, 2682166524U,
+	3317854285U,  631986188U, 3008214764U,  613826412U, 3567358221U,
+	3512343882U, 1552467474U, 3316162670U, 1275841024U, 4142173454U,
+	 565267881U,  768644821U,  198310105U, 2396688616U, 1837659011U,
+	 203429334U,  854539004U, 4235811518U, 3338304926U, 3730418692U,
+	3852254981U, 3032046452U, 2329811860U, 2303590566U, 2696092212U,
+	3894665932U,  145835667U,  249563655U, 1932210840U, 2431696407U,
+	3312636759U,  214962629U, 2092026914U, 3020145527U, 4073039873U,
+	2739105705U, 1308336752U,  855104522U, 2391715321U,   67448785U,
+	 547989482U,  854411802U, 3608633740U,  431731530U,  537375589U,
+	3888005760U,  696099141U,  397343236U, 1864511780U,   44029739U,
+	1729526891U, 1993398655U, 2010173426U, 2591546756U,  275223291U,
+	1503900299U, 4217765081U, 2185635252U, 1122436015U, 3550155364U,
+	 681707194U, 3260479338U,  933579397U, 2983029282U, 2505504587U,
+	2667410393U, 2962684490U, 4139721708U, 2658172284U, 2452602383U,
+	2607631612U, 1344296217U, 3075398709U, 2949785295U, 1049956168U,
+	3917185129U, 2155660174U, 3280524475U, 1503827867U,  674380765U,
+	1918468193U, 3843983676U,  634358221U, 2538335643U, 1873351298U,
+	3368723763U, 2129144130U, 3203528633U, 3087174986U, 2691698871U,
+	2516284287U,   24437745U, 1118381474U, 2816314867U, 2448576035U,
+	4281989654U,  217287825U,  165872888U, 2628995722U, 3533525116U,
+	2721669106U,  872340568U, 3429930655U, 3309047304U, 3916704967U,
+	3270160355U, 1348884255U, 1634797670U,  881214967U, 4259633554U,
+	 174613027U, 1103974314U, 1625224232U, 2678368291U, 1133866707U,
+	3853082619U, 4073196549U, 1189620777U,  637238656U,  930241537U,
+	4042750792U, 3842136042U, 2417007212U, 2524907510U, 1243036827U,
+	1282059441U, 3764588774U, 1394459615U, 2323620015U, 1166152231U,
+	3307479609U, 3849322257U, 3507445699U, 4247696636U,  758393720U,
+	 967665141U, 1095244571U, 1319812152U,  407678762U, 2640605208U,
+	2170766134U, 3663594275U, 4039329364U, 2512175520U,  725523154U,
+	2249807004U, 3312617979U, 2414634172U, 1278482215U,  349206484U,
+	1573063308U, 1196429124U, 3873264116U, 2400067801U,  268795167U,
+	 226175489U, 2961367263U, 1968719665U,   42656370U, 1010790699U,
+	 561600615U, 2422453992U, 3082197735U, 1636700484U, 3977715296U,
+	3125350482U, 3478021514U, 2227819446U, 1540868045U, 3061908980U,
+	1087362407U, 3625200291U,  361937537U,  580441897U, 1520043666U,
+	2270875402U, 1009161260U, 2502355842U, 4278769785U,  473902412U,
+	1057239083U, 1905829039U, 1483781177U, 2080011417U, 1207494246U,
+	1806991954U, 2194674403U, 3455972205U,  807207678U, 3655655687U,
+	 674112918U,  195425752U, 3917890095U, 1874364234U, 1837892715U,
+	3663478166U, 1548892014U, 2570748714U, 2049929836U, 2167029704U,
+	 697543767U, 3499545023U, 3342496315U, 1725251190U, 3561387469U,
+	2905606616U, 1580182447U, 3934525927U, 4103172792U, 1365672522U,
+	1534795737U, 3308667416U, 2841911405U, 3943182730U, 4072020313U,
+	3494770452U, 3332626671U,   55327267U,  478030603U,  411080625U,
+	3419529010U, 1604767823U, 3513468014U,  570668510U,  913790824U,
+	2283967995U,  695159462U, 3825542932U, 4150698144U, 1829758699U,
+	 202895590U, 1609122645U, 1267651008U, 2910315509U, 2511475445U,
+	2477423819U, 3932081579U,  900879979U, 2145588390U, 2670007504U,
+	 580819444U, 1864996828U, 2526325979U, 1019124258U,  815508628U,
+	2765933989U, 1277301341U, 3006021786U,  855540956U,  288025710U,
+	1919594237U, 2331223864U,  177452412U, 2475870369U, 2689291749U,
+	 865194284U,  253432152U, 2628531804U, 2861208555U, 2361597573U,
+	1653952120U, 1039661024U, 2159959078U, 3709040440U, 3564718533U,
+	2596878672U, 2041442161U,   31164696U, 2662962485U, 3665637339U,
+	1678115244U, 2699839832U, 3651968520U, 3521595541U,  458433303U,
+	2423096824U,   21831741U,  380011703U, 2498168716U,  861806087U,
+	1673574843U, 4188794405U, 2520563651U, 2632279153U, 2170465525U,
+	4171949898U, 3886039621U, 1661344005U, 3424285243U,  992588372U,
+	2500984144U, 2993248497U, 3590193895U, 1535327365U,  515645636U,
+	 131633450U, 3729760261U, 1613045101U, 3254194278U,   15889678U,
+	1493590689U,  244148718U, 2991472662U, 1401629333U,  777349878U,
+	2501401703U, 4285518317U, 3794656178U,  955526526U, 3442142820U,
+	3970298374U,  736025417U, 2737370764U, 1271509744U,  440570731U,
+	 136141826U, 1596189518U,  923399175U,  257541519U, 3505774281U,
+	2194358432U, 2518162991U, 1379893637U, 2667767062U, 3748146247U,
+	1821712620U, 3923161384U, 1947811444U, 2392527197U, 4127419685U,
+	1423694998U, 4156576871U, 1382885582U, 3420127279U, 3617499534U,
+	2994377493U, 4038063986U, 1918458672U, 2983166794U, 4200449033U,
+	 353294540U, 1609232588U,  243926648U, 2332803291U,  507996832U,
+	2392838793U, 4075145196U, 2060984340U, 4287475136U,   88232602U,
+	2491531140U, 4159725633U, 2272075455U,  759298618U,  201384554U,
+	 838356250U, 1416268324U,  674476934U,   90795364U,  141672229U,
+	3660399588U, 4196417251U, 3249270244U, 3774530247U,   59587265U,
+	3683164208U,   19392575U, 1463123697U, 1882205379U,  293780489U,
+	2553160622U, 2933904694U,  675638239U, 2851336944U, 1435238743U,
+	2448730183U,  804436302U, 2119845972U,  322560608U, 4097732704U,
+	2987802540U,  641492617U, 2575442710U, 4217822703U, 3271835300U,
+	2836418300U, 3739921620U, 2138378768U, 2879771855U, 4294903423U,
+	3121097946U, 2603440486U, 2560820391U, 1012930944U, 2313499967U,
+	 584489368U, 3431165766U,  897384869U, 2062537737U, 2847889234U,
+	3742362450U, 2951174585U, 4204621084U, 1109373893U, 3668075775U,
+	2750138839U, 3518055702U,  733072558U, 4169325400U,  788493625U
+};
+static const uint64_t init_gen_rand_64_expected[] = {
+	KQU(16924766246869039260), KQU( 8201438687333352714),
+	KQU( 2265290287015001750), KQU(18397264611805473832),
+	KQU( 3375255223302384358), KQU( 6345559975416828796),
+	KQU(18229739242790328073), KQU( 7596792742098800905),
+	KQU(  255338647169685981), KQU( 2052747240048610300),
+	KQU(18328151576097299343), KQU(12472905421133796567),
+	KQU(11315245349717600863), KQU(16594110197775871209),
+	KQU(15708751964632456450), KQU(10452031272054632535),
+	KQU(11097646720811454386), KQU( 4556090668445745441),
+	KQU(17116187693090663106), KQU(14931526836144510645),
+	KQU( 9190752218020552591), KQU( 9625800285771901401),
+	KQU(13995141077659972832), KQU( 5194209094927829625),
+	KQU( 4156788379151063303), KQU( 8523452593770139494),
+	KQU(14082382103049296727), KQU( 2462601863986088483),
+	KQU( 3030583461592840678), KQU( 5221622077872827681),
+	KQU( 3084210671228981236), KQU(13956758381389953823),
+	KQU(13503889856213423831), KQU(15696904024189836170),
+	KQU( 4612584152877036206), KQU( 6231135538447867881),
+	KQU(10172457294158869468), KQU( 6452258628466708150),
+	KQU(14044432824917330221), KQU(  370168364480044279),
+	KQU(10102144686427193359), KQU(  667870489994776076),
+	KQU( 2732271956925885858), KQU(18027788905977284151),
+	KQU(15009842788582923859), KQU( 7136357960180199542),
+	KQU(15901736243475578127), KQU(16951293785352615701),
+	KQU(10551492125243691632), KQU(17668869969146434804),
+	KQU(13646002971174390445), KQU( 9804471050759613248),
+	KQU( 5511670439655935493), KQU(18103342091070400926),
+	KQU(17224512747665137533), KQU(15534627482992618168),
+	KQU( 1423813266186582647), KQU(15821176807932930024),
+	KQU(   30323369733607156), KQU(11599382494723479403),
+	KQU(  653856076586810062), KQU( 3176437395144899659),
+	KQU(14028076268147963917), KQU(16156398271809666195),
+	KQU( 3166955484848201676), KQU( 5746805620136919390),
+	KQU(17297845208891256593), KQU(11691653183226428483),
+	KQU(17900026146506981577), KQU(15387382115755971042),
+	KQU(16923567681040845943), KQU( 8039057517199388606),
+	KQU(11748409241468629263), KQU(  794358245539076095),
+	KQU(13438501964693401242), KQU(14036803236515618962),
+	KQU( 5252311215205424721), KQU(17806589612915509081),
+	KQU( 6802767092397596006), KQU(14212120431184557140),
+	KQU( 1072951366761385712), KQU(13098491780722836296),
+	KQU( 9466676828710797353), KQU(12673056849042830081),
+	KQU(12763726623645357580), KQU(16468961652999309493),
+	KQU(15305979875636438926), KQU(17444713151223449734),
+	KQU( 5692214267627883674), KQU(13049589139196151505),
+	KQU(  880115207831670745), KQU( 1776529075789695498),
+	KQU(16695225897801466485), KQU(10666901778795346845),
+	KQU( 6164389346722833869), KQU( 2863817793264300475),
+	KQU( 9464049921886304754), KQU( 3993566636740015468),
+	KQU( 9983749692528514136), KQU(16375286075057755211),
+	KQU(16042643417005440820), KQU(11445419662923489877),
+	KQU( 7999038846885158836), KQU( 6721913661721511535),
+	KQU( 5363052654139357320), KQU( 1817788761173584205),
+	KQU(13290974386445856444), KQU( 4650350818937984680),
+	KQU( 8219183528102484836), KQU( 1569862923500819899),
+	KQU( 4189359732136641860), KQU(14202822961683148583),
+	KQU( 4457498315309429058), KQU(13089067387019074834),
+	KQU(11075517153328927293), KQU(10277016248336668389),
+	KQU( 7070509725324401122), KQU(17808892017780289380),
+	KQU(13143367339909287349), KQU( 1377743745360085151),
+	KQU( 5749341807421286485), KQU(14832814616770931325),
+	KQU( 7688820635324359492), KQU(10960474011539770045),
+	KQU(   81970066653179790), KQU(12619476072607878022),
+	KQU( 4419566616271201744), KQU(15147917311750568503),
+	KQU( 5549739182852706345), KQU( 7308198397975204770),
+	KQU(13580425496671289278), KQU(17070764785210130301),
+	KQU( 8202832846285604405), KQU( 6873046287640887249),
+	KQU( 6927424434308206114), KQU( 6139014645937224874),
+	KQU(10290373645978487639), KQU(15904261291701523804),
+	KQU( 9628743442057826883), KQU(18383429096255546714),
+	KQU( 4977413265753686967), KQU( 7714317492425012869),
+	KQU( 9025232586309926193), KQU(14627338359776709107),
+	KQU(14759849896467790763), KQU(10931129435864423252),
+	KQU( 4588456988775014359), KQU(10699388531797056724),
+	KQU(  468652268869238792), KQU( 5755943035328078086),
+	KQU( 2102437379988580216), KQU( 9986312786506674028),
+	KQU( 2654207180040945604), KQU( 8726634790559960062),
+	KQU(  100497234871808137), KQU( 2800137176951425819),
+	KQU( 6076627612918553487), KQU( 5780186919186152796),
+	KQU( 8179183595769929098), KQU( 6009426283716221169),
+	KQU( 2796662551397449358), KQU( 1756961367041986764),
+	KQU( 6972897917355606205), KQU(14524774345368968243),
+	KQU( 2773529684745706940), KQU( 4853632376213075959),
+	KQU( 4198177923731358102), KQU( 8271224913084139776),
+	KQU( 2741753121611092226), KQU(16782366145996731181),
+	KQU(15426125238972640790), KQU(13595497100671260342),
+	KQU( 3173531022836259898), KQU( 6573264560319511662),
+	KQU(18041111951511157441), KQU( 2351433581833135952),
+	KQU( 3113255578908173487), KQU( 1739371330877858784),
+	KQU(16046126562789165480), KQU( 8072101652214192925),
+	KQU(15267091584090664910), KQU( 9309579200403648940),
+	KQU( 5218892439752408722), KQU(14492477246004337115),
+	KQU(17431037586679770619), KQU( 7385248135963250480),
+	KQU( 9580144956565560660), KQU( 4919546228040008720),
+	KQU(15261542469145035584), KQU(18233297270822253102),
+	KQU( 5453248417992302857), KQU( 9309519155931460285),
+	KQU(10342813012345291756), KQU(15676085186784762381),
+	KQU(15912092950691300645), KQU( 9371053121499003195),
+	KQU( 9897186478226866746), KQU(14061858287188196327),
+	KQU(  122575971620788119), KQU(12146750969116317754),
+	KQU( 4438317272813245201), KQU( 8332576791009527119),
+	KQU(13907785691786542057), KQU(10374194887283287467),
+	KQU( 2098798755649059566), KQU( 3416235197748288894),
+	KQU( 8688269957320773484), KQU( 7503964602397371571),
+	KQU(16724977015147478236), KQU( 9461512855439858184),
+	KQU(13259049744534534727), KQU( 3583094952542899294),
+	KQU( 8764245731305528292), KQU(13240823595462088985),
+	KQU(13716141617617910448), KQU(18114969519935960955),
+	KQU( 2297553615798302206), KQU( 4585521442944663362),
+	KQU(17776858680630198686), KQU( 4685873229192163363),
+	KQU(  152558080671135627), KQU(15424900540842670088),
+	KQU(13229630297130024108), KQU(17530268788245718717),
+	KQU(16675633913065714144), KQU( 3158912717897568068),
+	KQU(15399132185380087288), KQU( 7401418744515677872),
+	KQU(13135412922344398535), KQU( 6385314346100509511),
+	KQU(13962867001134161139), KQU(10272780155442671999),
+	KQU(12894856086597769142), KQU(13340877795287554994),
+	KQU(12913630602094607396), KQU(12543167911119793857),
+	KQU(17343570372251873096), KQU(10959487764494150545),
+	KQU( 6966737953093821128), KQU(13780699135496988601),
+	KQU( 4405070719380142046), KQU(14923788365607284982),
+	KQU( 2869487678905148380), KQU( 6416272754197188403),
+	KQU(15017380475943612591), KQU( 1995636220918429487),
+	KQU( 3402016804620122716), KQU(15800188663407057080),
+	KQU(11362369990390932882), KQU(15262183501637986147),
+	KQU(10239175385387371494), KQU( 9352042420365748334),
+	KQU( 1682457034285119875), KQU( 1724710651376289644),
+	KQU( 2038157098893817966), KQU( 9897825558324608773),
+	KQU( 1477666236519164736), KQU(16835397314511233640),
+	KQU(10370866327005346508), KQU(10157504370660621982),
+	KQU(12113904045335882069), KQU(13326444439742783008),
+	KQU(11302769043000765804), KQU(13594979923955228484),
+	KQU(11779351762613475968), KQU( 3786101619539298383),
+	KQU( 8021122969180846063), KQU(15745904401162500495),
+	KQU(10762168465993897267), KQU(13552058957896319026),
+	KQU(11200228655252462013), KQU( 5035370357337441226),
+	KQU( 7593918984545500013), KQU( 5418554918361528700),
+	KQU( 4858270799405446371), KQU( 9974659566876282544),
+	KQU(18227595922273957859), KQU( 2772778443635656220),
+	KQU(14285143053182085385), KQU( 9939700992429600469),
+	KQU(12756185904545598068), KQU( 2020783375367345262),
+	KQU(   57026775058331227), KQU(  950827867930065454),
+	KQU( 6602279670145371217), KQU( 2291171535443566929),
+	KQU( 5832380724425010313), KQU( 1220343904715982285),
+	KQU(17045542598598037633), KQU(15460481779702820971),
+	KQU(13948388779949365130), KQU(13975040175430829518),
+	KQU(17477538238425541763), KQU(11104663041851745725),
+	KQU(15860992957141157587), KQU(14529434633012950138),
+	KQU( 2504838019075394203), KQU( 7512113882611121886),
+	KQU( 4859973559980886617), KQU( 1258601555703250219),
+	KQU(15594548157514316394), KQU( 4516730171963773048),
+	KQU(11380103193905031983), KQU( 6809282239982353344),
+	KQU(18045256930420065002), KQU( 2453702683108791859),
+	KQU(  977214582986981460), KQU( 2006410402232713466),
+	KQU( 6192236267216378358), KQU( 3429468402195675253),
+	KQU(18146933153017348921), KQU(17369978576367231139),
+	KQU( 1246940717230386603), KQU(11335758870083327110),
+	KQU(14166488801730353682), KQU( 9008573127269635732),
+	KQU(10776025389820643815), KQU(15087605441903942962),
+	KQU( 1359542462712147922), KQU(13898874411226454206),
+	KQU(17911176066536804411), KQU( 9435590428600085274),
+	KQU(  294488509967864007), KQU( 8890111397567922046),
+	KQU( 7987823476034328778), KQU(13263827582440967651),
+	KQU( 7503774813106751573), KQU(14974747296185646837),
+	KQU( 8504765037032103375), KQU(17340303357444536213),
+	KQU( 7704610912964485743), KQU( 8107533670327205061),
+	KQU( 9062969835083315985), KQU(16968963142126734184),
+	KQU(12958041214190810180), KQU( 2720170147759570200),
+	KQU( 2986358963942189566), KQU(14884226322219356580),
+	KQU(  286224325144368520), KQU(11313800433154279797),
+	KQU(18366849528439673248), KQU(17899725929482368789),
+	KQU( 3730004284609106799), KQU( 1654474302052767205),
+	KQU( 5006698007047077032), KQU( 8196893913601182838),
+	KQU(15214541774425211640), KQU(17391346045606626073),
+	KQU( 8369003584076969089), KQU( 3939046733368550293),
+	KQU(10178639720308707785), KQU( 2180248669304388697),
+	KQU(   62894391300126322), KQU( 9205708961736223191),
+	KQU( 6837431058165360438), KQU( 3150743890848308214),
+	KQU(17849330658111464583), KQU(12214815643135450865),
+	KQU(13410713840519603402), KQU( 3200778126692046802),
+	KQU(13354780043041779313), KQU(  800850022756886036),
+	KQU(15660052933953067433), KQU( 6572823544154375676),
+	KQU(11030281857015819266), KQU(12682241941471433835),
+	KQU(11654136407300274693), KQU( 4517795492388641109),
+	KQU( 9757017371504524244), KQU(17833043400781889277),
+	KQU(12685085201747792227), KQU(10408057728835019573),
+	KQU(   98370418513455221), KQU( 6732663555696848598),
+	KQU(13248530959948529780), KQU( 3530441401230622826),
+	KQU(18188251992895660615), KQU( 1847918354186383756),
+	KQU( 1127392190402660921), KQU(11293734643143819463),
+	KQU( 3015506344578682982), KQU(13852645444071153329),
+	KQU( 2121359659091349142), KQU( 1294604376116677694),
+	KQU( 5616576231286352318), KQU( 7112502442954235625),
+	KQU(11676228199551561689), KQU(12925182803007305359),
+	KQU( 7852375518160493082), KQU( 1136513130539296154),
+	KQU( 5636923900916593195), KQU( 3221077517612607747),
+	KQU(17784790465798152513), KQU( 3554210049056995938),
+	KQU(17476839685878225874), KQU( 3206836372585575732),
+	KQU( 2765333945644823430), KQU(10080070903718799528),
+	KQU( 5412370818878286353), KQU( 9689685887726257728),
+	KQU( 8236117509123533998), KQU( 1951139137165040214),
+	KQU( 4492205209227980349), KQU(16541291230861602967),
+	KQU( 1424371548301437940), KQU( 9117562079669206794),
+	KQU(14374681563251691625), KQU(13873164030199921303),
+	KQU( 6680317946770936731), KQU(15586334026918276214),
+	KQU(10896213950976109802), KQU( 9506261949596413689),
+	KQU( 9903949574308040616), KQU( 6038397344557204470),
+	KQU(  174601465422373648), KQU(15946141191338238030),
+	KQU(17142225620992044937), KQU( 7552030283784477064),
+	KQU( 2947372384532947997), KQU(  510797021688197711),
+	KQU( 4962499439249363461), KQU(   23770320158385357),
+	KQU(  959774499105138124), KQU( 1468396011518788276),
+	KQU( 2015698006852312308), KQU( 4149400718489980136),
+	KQU( 5992916099522371188), KQU(10819182935265531076),
+	KQU(16189787999192351131), KQU(  342833961790261950),
+	KQU(12470830319550495336), KQU(18128495041912812501),
+	KQU( 1193600899723524337), KQU( 9056793666590079770),
+	KQU( 2154021227041669041), KQU( 4963570213951235735),
+	KQU( 4865075960209211409), KQU( 2097724599039942963),
+	KQU( 2024080278583179845), KQU(11527054549196576736),
+	KQU(10650256084182390252), KQU( 4808408648695766755),
+	KQU( 1642839215013788844), KQU(10607187948250398390),
+	KQU( 7076868166085913508), KQU(  730522571106887032),
+	KQU(12500579240208524895), KQU( 4484390097311355324),
+	KQU(15145801330700623870), KQU( 8055827661392944028),
+	KQU( 5865092976832712268), KQU(15159212508053625143),
+	KQU( 3560964582876483341), KQU( 4070052741344438280),
+	KQU( 6032585709886855634), KQU(15643262320904604873),
+	KQU( 2565119772293371111), KQU(  318314293065348260),
+	KQU(15047458749141511872), KQU( 7772788389811528730),
+	KQU( 7081187494343801976), KQU( 6465136009467253947),
+	KQU(10425940692543362069), KQU(  554608190318339115),
+	KQU(14796699860302125214), KQU( 1638153134431111443),
+	KQU(10336967447052276248), KQU( 8412308070396592958),
+	KQU( 4004557277152051226), KQU( 8143598997278774834),
+	KQU(16413323996508783221), KQU(13139418758033994949),
+	KQU( 9772709138335006667), KQU( 2818167159287157659),
+	KQU(17091740573832523669), KQU(14629199013130751608),
+	KQU(18268322711500338185), KQU( 8290963415675493063),
+	KQU( 8830864907452542588), KQU( 1614839084637494849),
+	KQU(14855358500870422231), KQU( 3472996748392519937),
+	KQU(15317151166268877716), KQU( 5825895018698400362),
+	KQU(16730208429367544129), KQU(10481156578141202800),
+	KQU( 4746166512382823750), KQU(12720876014472464998),
+	KQU( 8825177124486735972), KQU(13733447296837467838),
+	KQU( 6412293741681359625), KQU( 8313213138756135033),
+	KQU(11421481194803712517), KQU( 7997007691544174032),
+	KQU( 6812963847917605930), KQU( 9683091901227558641),
+	KQU(14703594165860324713), KQU( 1775476144519618309),
+	KQU( 2724283288516469519), KQU(  717642555185856868),
+	KQU( 8736402192215092346), KQU(11878800336431381021),
+	KQU( 4348816066017061293), KQU( 6115112756583631307),
+	KQU( 9176597239667142976), KQU(12615622714894259204),
+	KQU(10283406711301385987), KQU( 5111762509485379420),
+	KQU( 3118290051198688449), KQU( 7345123071632232145),
+	KQU( 9176423451688682359), KQU( 4843865456157868971),
+	KQU(12008036363752566088), KQU(12058837181919397720),
+	KQU( 2145073958457347366), KQU( 1526504881672818067),
+	KQU( 3488830105567134848), KQU(13208362960674805143),
+	KQU( 4077549672899572192), KQU( 7770995684693818365),
+	KQU( 1398532341546313593), KQU(12711859908703927840),
+	KQU( 1417561172594446813), KQU(17045191024194170604),
+	KQU( 4101933177604931713), KQU(14708428834203480320),
+	KQU(17447509264469407724), KQU(14314821973983434255),
+	KQU(17990472271061617265), KQU( 5087756685841673942),
+	KQU(12797820586893859939), KQU( 1778128952671092879),
+	KQU( 3535918530508665898), KQU( 9035729701042481301),
+	KQU(14808661568277079962), KQU(14587345077537747914),
+	KQU(11920080002323122708), KQU( 6426515805197278753),
+	KQU( 3295612216725984831), KQU(11040722532100876120),
+	KQU(12305952936387598754), KQU(16097391899742004253),
+	KQU( 4908537335606182208), KQU(12446674552196795504),
+	KQU(16010497855816895177), KQU( 9194378874788615551),
+	KQU( 3382957529567613384), KQU( 5154647600754974077),
+	KQU( 9801822865328396141), KQU( 9023662173919288143),
+	KQU(17623115353825147868), KQU( 8238115767443015816),
+	KQU(15811444159859002560), KQU( 9085612528904059661),
+	KQU( 6888601089398614254), KQU(  258252992894160189),
+	KQU( 6704363880792428622), KQU( 6114966032147235763),
+	KQU(11075393882690261875), KQU( 8797664238933620407),
+	KQU( 5901892006476726920), KQU( 5309780159285518958),
+	KQU(14940808387240817367), KQU(14642032021449656698),
+	KQU( 9808256672068504139), KQU( 3670135111380607658),
+	KQU(11211211097845960152), KQU( 1474304506716695808),
+	KQU(15843166204506876239), KQU( 7661051252471780561),
+	KQU(10170905502249418476), KQU( 7801416045582028589),
+	KQU( 2763981484737053050), KQU( 9491377905499253054),
+	KQU(16201395896336915095), KQU( 9256513756442782198),
+	KQU( 5411283157972456034), KQU( 5059433122288321676),
+	KQU( 4327408006721123357), KQU( 9278544078834433377),
+	KQU( 7601527110882281612), KQU(11848295896975505251),
+	KQU(12096998801094735560), KQU(14773480339823506413),
+	KQU(15586227433895802149), KQU(12786541257830242872),
+	KQU( 6904692985140503067), KQU( 5309011515263103959),
+	KQU(12105257191179371066), KQU(14654380212442225037),
+	KQU( 2556774974190695009), KQU( 4461297399927600261),
+	KQU(14888225660915118646), KQU(14915459341148291824),
+	KQU( 2738802166252327631), KQU( 6047155789239131512),
+	KQU(12920545353217010338), KQU(10697617257007840205),
+	KQU( 2751585253158203504), KQU(13252729159780047496),
+	KQU(14700326134672815469), KQU(14082527904374600529),
+	KQU(16852962273496542070), KQU(17446675504235853907),
+	KQU(15019600398527572311), KQU(12312781346344081551),
+	KQU(14524667935039810450), KQU( 5634005663377195738),
+	KQU(11375574739525000569), KQU( 2423665396433260040),
+	KQU( 5222836914796015410), KQU( 4397666386492647387),
+	KQU( 4619294441691707638), KQU(  665088602354770716),
+	KQU(13246495665281593610), KQU( 6564144270549729409),
+	KQU(10223216188145661688), KQU( 3961556907299230585),
+	KQU(11543262515492439914), KQU(16118031437285993790),
+	KQU( 7143417964520166465), KQU(13295053515909486772),
+	KQU(   40434666004899675), KQU(17127804194038347164),
+	KQU( 8599165966560586269), KQU( 8214016749011284903),
+	KQU(13725130352140465239), KQU( 5467254474431726291),
+	KQU( 7748584297438219877), KQU(16933551114829772472),
+	KQU( 2169618439506799400), KQU( 2169787627665113463),
+	KQU(17314493571267943764), KQU(18053575102911354912),
+	KQU(11928303275378476973), KQU(11593850925061715550),
+	KQU(17782269923473589362), KQU( 3280235307704747039),
+	KQU( 6145343578598685149), KQU(17080117031114086090),
+	KQU(18066839902983594755), KQU( 6517508430331020706),
+	KQU( 8092908893950411541), KQU(12558378233386153732),
+	KQU( 4476532167973132976), KQU(16081642430367025016),
+	KQU( 4233154094369139361), KQU( 8693630486693161027),
+	KQU(11244959343027742285), KQU(12273503967768513508),
+	KQU(14108978636385284876), KQU( 7242414665378826984),
+	KQU( 6561316938846562432), KQU( 8601038474994665795),
+	KQU(17532942353612365904), KQU(17940076637020912186),
+	KQU( 7340260368823171304), KQU( 7061807613916067905),
+	KQU(10561734935039519326), KQU(17990796503724650862),
+	KQU( 6208732943911827159), KQU(  359077562804090617),
+	KQU(14177751537784403113), KQU(10659599444915362902),
+	KQU(15081727220615085833), KQU(13417573895659757486),
+	KQU(15513842342017811524), KQU(11814141516204288231),
+	KQU( 1827312513875101814), KQU( 2804611699894603103),
+	KQU(17116500469975602763), KQU(12270191815211952087),
+	KQU(12256358467786024988), KQU(18435021722453971267),
+	KQU(  671330264390865618), KQU(  476504300460286050),
+	KQU(16465470901027093441), KQU( 4047724406247136402),
+	KQU( 1322305451411883346), KQU( 1388308688834322280),
+	KQU( 7303989085269758176), KQU( 9323792664765233642),
+	KQU( 4542762575316368936), KQU(17342696132794337618),
+	KQU( 4588025054768498379), KQU(13415475057390330804),
+	KQU(17880279491733405570), KQU(10610553400618620353),
+	KQU( 3180842072658960139), KQU(13002966655454270120),
+	KQU( 1665301181064982826), KQU( 7083673946791258979),
+	KQU(  190522247122496820), KQU(17388280237250677740),
+	KQU( 8430770379923642945), KQU(12987180971921668584),
+	KQU( 2311086108365390642), KQU( 2870984383579822345),
+	KQU(14014682609164653318), KQU(14467187293062251484),
+	KQU(  192186361147413298), KQU(15171951713531796524),
+	KQU( 9900305495015948728), KQU(17958004775615466344),
+	KQU(14346380954498606514), KQU(18040047357617407096),
+	KQU( 5035237584833424532), KQU(15089555460613972287),
+	KQU( 4131411873749729831), KQU( 1329013581168250330),
+	KQU(10095353333051193949), KQU(10749518561022462716),
+	KQU( 9050611429810755847), KQU(15022028840236655649),
+	KQU( 8775554279239748298), KQU(13105754025489230502),
+	KQU(15471300118574167585), KQU(   89864764002355628),
+	KQU( 8776416323420466637), KQU( 5280258630612040891),
+	KQU( 2719174488591862912), KQU( 7599309137399661994),
+	KQU(15012887256778039979), KQU(14062981725630928925),
+	KQU(12038536286991689603), KQU( 7089756544681775245),
+	KQU(10376661532744718039), KQU( 1265198725901533130),
+	KQU(13807996727081142408), KQU( 2935019626765036403),
+	KQU( 7651672460680700141), KQU( 3644093016200370795),
+	KQU( 2840982578090080674), KQU(17956262740157449201),
+	KQU(18267979450492880548), KQU(11799503659796848070),
+	KQU( 9942537025669672388), KQU(11886606816406990297),
+	KQU( 5488594946437447576), KQU( 7226714353282744302),
+	KQU( 3784851653123877043), KQU(  878018453244803041),
+	KQU(12110022586268616085), KQU(  734072179404675123),
+	KQU(11869573627998248542), KQU(  469150421297783998),
+	KQU(  260151124912803804), KQU(11639179410120968649),
+	KQU( 9318165193840846253), KQU(12795671722734758075),
+	KQU(15318410297267253933), KQU(  691524703570062620),
+	KQU( 5837129010576994601), KQU(15045963859726941052),
+	KQU( 5850056944932238169), KQU(12017434144750943807),
+	KQU( 7447139064928956574), KQU( 3101711812658245019),
+	KQU(16052940704474982954), KQU(18195745945986994042),
+	KQU( 8932252132785575659), KQU(13390817488106794834),
+	KQU(11582771836502517453), KQU( 4964411326683611686),
+	KQU( 2195093981702694011), KQU(14145229538389675669),
+	KQU(16459605532062271798), KQU(  866316924816482864),
+	KQU( 4593041209937286377), KQU( 8415491391910972138),
+	KQU( 4171236715600528969), KQU(16637569303336782889),
+	KQU( 2002011073439212680), KQU(17695124661097601411),
+	KQU( 4627687053598611702), KQU( 7895831936020190403),
+	KQU( 8455951300917267802), KQU( 2923861649108534854),
+	KQU( 8344557563927786255), KQU( 6408671940373352556),
+	KQU(12210227354536675772), KQU(14294804157294222295),
+	KQU(10103022425071085127), KQU(10092959489504123771),
+	KQU( 6554774405376736268), KQU(12629917718410641774),
+	KQU( 6260933257596067126), KQU( 2460827021439369673),
+	KQU( 2541962996717103668), KQU(  597377203127351475),
+	KQU( 5316984203117315309), KQU( 4811211393563241961),
+	KQU(13119698597255811641), KQU( 8048691512862388981),
+	KQU(10216818971194073842), KQU( 4612229970165291764),
+	KQU(10000980798419974770), KQU( 6877640812402540687),
+	KQU( 1488727563290436992), KQU( 2227774069895697318),
+	KQU(11237754507523316593), KQU(13478948605382290972),
+	KQU( 1963583846976858124), KQU( 5512309205269276457),
+	KQU( 3972770164717652347), KQU( 3841751276198975037),
+	KQU(10283343042181903117), KQU( 8564001259792872199),
+	KQU(16472187244722489221), KQU( 8953493499268945921),
+	KQU( 3518747340357279580), KQU( 4003157546223963073),
+	KQU( 3270305958289814590), KQU( 3966704458129482496),
+	KQU( 8122141865926661939), KQU(14627734748099506653),
+	KQU(13064426990862560568), KQU( 2414079187889870829),
+	KQU( 5378461209354225306), KQU(10841985740128255566),
+	KQU(  538582442885401738), KQU( 7535089183482905946),
+	KQU(16117559957598879095), KQU( 8477890721414539741),
+	KQU( 1459127491209533386), KQU(17035126360733620462),
+	KQU( 8517668552872379126), KQU(10292151468337355014),
+	KQU(17081267732745344157), KQU(13751455337946087178),
+	KQU(14026945459523832966), KQU( 6653278775061723516),
+	KQU(10619085543856390441), KQU( 2196343631481122885),
+	KQU(10045966074702826136), KQU(10082317330452718282),
+	KQU( 5920859259504831242), KQU( 9951879073426540617),
+	KQU( 7074696649151414158), KQU(15808193543879464318),
+	KQU( 7385247772746953374), KQU( 3192003544283864292),
+	KQU(18153684490917593847), KQU(12423498260668568905),
+	KQU(10957758099756378169), KQU(11488762179911016040),
+	KQU( 2099931186465333782), KQU(11180979581250294432),
+	KQU( 8098916250668367933), KQU( 3529200436790763465),
+	KQU(12988418908674681745), KQU( 6147567275954808580),
+	KQU( 3207503344604030989), KQU(10761592604898615360),
+	KQU(  229854861031893504), KQU( 8809853962667144291),
+	KQU(13957364469005693860), KQU( 7634287665224495886),
+	KQU(12353487366976556874), KQU( 1134423796317152034),
+	KQU( 2088992471334107068), KQU( 7393372127190799698),
+	KQU( 1845367839871058391), KQU(  207922563987322884),
+	KQU(11960870813159944976), KQU(12182120053317317363),
+	KQU(17307358132571709283), KQU(13871081155552824936),
+	KQU(18304446751741566262), KQU( 7178705220184302849),
+	KQU(10929605677758824425), KQU(16446976977835806844),
+	KQU(13723874412159769044), KQU( 6942854352100915216),
+	KQU( 1726308474365729390), KQU( 2150078766445323155),
+	KQU(15345558947919656626), KQU(12145453828874527201),
+	KQU( 2054448620739726849), KQU( 2740102003352628137),
+	KQU(11294462163577610655), KQU(  756164283387413743),
+	KQU(17841144758438810880), KQU(10802406021185415861),
+	KQU( 8716455530476737846), KQU( 6321788834517649606),
+	KQU(14681322910577468426), KQU(17330043563884336387),
+	KQU(12701802180050071614), KQU(14695105111079727151),
+	KQU( 5112098511654172830), KQU( 4957505496794139973),
+	KQU( 8270979451952045982), KQU(12307685939199120969),
+	KQU(12425799408953443032), KQU( 8376410143634796588),
+	KQU(16621778679680060464), KQU( 3580497854566660073),
+	KQU( 1122515747803382416), KQU(  857664980960597599),
+	KQU( 6343640119895925918), KQU(12878473260854462891),
+	KQU(10036813920765722626), KQU(14451335468363173812),
+	KQU( 5476809692401102807), KQU(16442255173514366342),
+	KQU(13060203194757167104), KQU(14354124071243177715),
+	KQU(15961249405696125227), KQU(13703893649690872584),
+	KQU(  363907326340340064), KQU( 6247455540491754842),
+	KQU(12242249332757832361), KQU(  156065475679796717),
+	KQU( 9351116235749732355), KQU( 4590350628677701405),
+	KQU( 1671195940982350389), KQU(13501398458898451905),
+	KQU( 6526341991225002255), KQU( 1689782913778157592),
+	KQU( 7439222350869010334), KQU(13975150263226478308),
+	KQU(11411961169932682710), KQU(17204271834833847277),
+	KQU(  541534742544435367), KQU( 6591191931218949684),
+	KQU( 2645454775478232486), KQU( 4322857481256485321),
+	KQU( 8477416487553065110), KQU(12902505428548435048),
+	KQU(  971445777981341415), KQU(14995104682744976712),
+	KQU( 4243341648807158063), KQU( 8695061252721927661),
+	KQU( 5028202003270177222), KQU( 2289257340915567840),
+	KQU(13870416345121866007), KQU(13994481698072092233),
+	KQU( 6912785400753196481), KQU( 2278309315841980139),
+	KQU( 4329765449648304839), KQU( 5963108095785485298),
+	KQU( 4880024847478722478), KQU(16015608779890240947),
+	KQU( 1866679034261393544), KQU(  914821179919731519),
+	KQU( 9643404035648760131), KQU( 2418114953615593915),
+	KQU(  944756836073702374), KQU(15186388048737296834),
+	KQU( 7723355336128442206), KQU( 7500747479679599691),
+	KQU(18013961306453293634), KQU( 2315274808095756456),
+	KQU(13655308255424029566), KQU(17203800273561677098),
+	KQU( 1382158694422087756), KQU( 5090390250309588976),
+	KQU(  517170818384213989), KQU( 1612709252627729621),
+	KQU( 1330118955572449606), KQU(  300922478056709885),
+	KQU(18115693291289091987), KQU(13491407109725238321),
+	KQU(15293714633593827320), KQU( 5151539373053314504),
+	KQU( 5951523243743139207), KQU(14459112015249527975),
+	KQU( 5456113959000700739), KQU( 3877918438464873016),
+	KQU(12534071654260163555), KQU(15871678376893555041),
+	KQU(11005484805712025549), KQU(16353066973143374252),
+	KQU( 4358331472063256685), KQU( 8268349332210859288),
+	KQU(12485161590939658075), KQU(13955993592854471343),
+	KQU( 5911446886848367039), KQU(14925834086813706974),
+	KQU( 6590362597857994805), KQU( 1280544923533661875),
+	KQU( 1637756018947988164), KQU( 4734090064512686329),
+	KQU(16693705263131485912), KQU( 6834882340494360958),
+	KQU( 8120732176159658505), KQU( 2244371958905329346),
+	KQU(10447499707729734021), KQU( 7318742361446942194),
+	KQU( 8032857516355555296), KQU(14023605983059313116),
+	KQU( 1032336061815461376), KQU( 9840995337876562612),
+	KQU( 9869256223029203587), KQU(12227975697177267636),
+	KQU(12728115115844186033), KQU( 7752058479783205470),
+	KQU(  729733219713393087), KQU(12954017801239007622)
+};
+static const uint64_t init_by_array_64_expected[] = {
+	KQU( 2100341266307895239), KQU( 8344256300489757943),
+	KQU(15687933285484243894), KQU( 8268620370277076319),
+	KQU(12371852309826545459), KQU( 8800491541730110238),
+	KQU(18113268950100835773), KQU( 2886823658884438119),
+	KQU( 3293667307248180724), KQU( 9307928143300172731),
+	KQU( 7688082017574293629), KQU(  900986224735166665),
+	KQU( 9977972710722265039), KQU( 6008205004994830552),
+	KQU(  546909104521689292), KQU( 7428471521869107594),
+	KQU(14777563419314721179), KQU(16116143076567350053),
+	KQU( 5322685342003142329), KQU( 4200427048445863473),
+	KQU( 4693092150132559146), KQU(13671425863759338582),
+	KQU( 6747117460737639916), KQU( 4732666080236551150),
+	KQU( 5912839950611941263), KQU( 3903717554504704909),
+	KQU( 2615667650256786818), KQU(10844129913887006352),
+	KQU(13786467861810997820), KQU(14267853002994021570),
+	KQU(13767807302847237439), KQU(16407963253707224617),
+	KQU( 4802498363698583497), KQU( 2523802839317209764),
+	KQU( 3822579397797475589), KQU( 8950320572212130610),
+	KQU( 3745623504978342534), KQU(16092609066068482806),
+	KQU( 9817016950274642398), KQU(10591660660323829098),
+	KQU(11751606650792815920), KQU( 5122873818577122211),
+	KQU(17209553764913936624), KQU( 6249057709284380343),
+	KQU(15088791264695071830), KQU(15344673071709851930),
+	KQU( 4345751415293646084), KQU( 2542865750703067928),
+	KQU(13520525127852368784), KQU(18294188662880997241),
+	KQU( 3871781938044881523), KQU( 2873487268122812184),
+	KQU(15099676759482679005), KQU(15442599127239350490),
+	KQU( 6311893274367710888), KQU( 3286118760484672933),
+	KQU( 4146067961333542189), KQU(13303942567897208770),
+	KQU( 8196013722255630418), KQU( 4437815439340979989),
+	KQU(15433791533450605135), KQU( 4254828956815687049),
+	KQU( 1310903207708286015), KQU(10529182764462398549),
+	KQU(14900231311660638810), KQU( 9727017277104609793),
+	KQU( 1821308310948199033), KQU(11628861435066772084),
+	KQU( 9469019138491546924), KQU( 3145812670532604988),
+	KQU( 9938468915045491919), KQU( 1562447430672662142),
+	KQU(13963995266697989134), KQU( 3356884357625028695),
+	KQU( 4499850304584309747), KQU( 8456825817023658122),
+	KQU(10859039922814285279), KQU( 8099512337972526555),
+	KQU(  348006375109672149), KQU(11919893998241688603),
+	KQU( 1104199577402948826), KQU(16689191854356060289),
+	KQU(10992552041730168078), KQU( 7243733172705465836),
+	KQU( 5668075606180319560), KQU(18182847037333286970),
+	KQU( 4290215357664631322), KQU( 4061414220791828613),
+	KQU(13006291061652989604), KQU( 7140491178917128798),
+	KQU(12703446217663283481), KQU( 5500220597564558267),
+	KQU(10330551509971296358), KQU(15958554768648714492),
+	KQU( 5174555954515360045), KQU( 1731318837687577735),
+	KQU( 3557700801048354857), KQU(13764012341928616198),
+	KQU(13115166194379119043), KQU( 7989321021560255519),
+	KQU( 2103584280905877040), KQU( 9230788662155228488),
+	KQU(16396629323325547654), KQU(  657926409811318051),
+	KQU(15046700264391400727), KQU( 5120132858771880830),
+	KQU( 7934160097989028561), KQU( 6963121488531976245),
+	KQU(17412329602621742089), KQU(15144843053931774092),
+	KQU(17204176651763054532), KQU(13166595387554065870),
+	KQU( 8590377810513960213), KQU( 5834365135373991938),
+	KQU( 7640913007182226243), KQU( 3479394703859418425),
+	KQU(16402784452644521040), KQU( 4993979809687083980),
+	KQU(13254522168097688865), KQU(15643659095244365219),
+	KQU( 5881437660538424982), KQU(11174892200618987379),
+	KQU(  254409966159711077), KQU(17158413043140549909),
+	KQU( 3638048789290376272), KQU( 1376816930299489190),
+	KQU( 4622462095217761923), KQU(15086407973010263515),
+	KQU(13253971772784692238), KQU( 5270549043541649236),
+	KQU(11182714186805411604), KQU(12283846437495577140),
+	KQU( 5297647149908953219), KQU(10047451738316836654),
+	KQU( 4938228100367874746), KQU(12328523025304077923),
+	KQU( 3601049438595312361), KQU( 9313624118352733770),
+	KQU(13322966086117661798), KQU(16660005705644029394),
+	KQU(11337677526988872373), KQU(13869299102574417795),
+	KQU(15642043183045645437), KQU( 3021755569085880019),
+	KQU( 4979741767761188161), KQU(13679979092079279587),
+	KQU( 3344685842861071743), KQU(13947960059899588104),
+	KQU(  305806934293368007), KQU( 5749173929201650029),
+	KQU(11123724852118844098), KQU(15128987688788879802),
+	KQU(15251651211024665009), KQU( 7689925933816577776),
+	KQU(16732804392695859449), KQU(17087345401014078468),
+	KQU(14315108589159048871), KQU( 4820700266619778917),
+	KQU(16709637539357958441), KQU( 4936227875177351374),
+	KQU( 2137907697912987247), KQU(11628565601408395420),
+	KQU( 2333250549241556786), KQU( 5711200379577778637),
+	KQU( 5170680131529031729), KQU(12620392043061335164),
+	KQU(   95363390101096078), KQU( 5487981914081709462),
+	KQU( 1763109823981838620), KQU( 3395861271473224396),
+	KQU( 1300496844282213595), KQU( 6894316212820232902),
+	KQU(10673859651135576674), KQU( 5911839658857903252),
+	KQU(17407110743387299102), KQU( 8257427154623140385),
+	KQU(11389003026741800267), KQU( 4070043211095013717),
+	KQU(11663806997145259025), KQU(15265598950648798210),
+	KQU(  630585789434030934), KQU( 3524446529213587334),
+	KQU( 7186424168495184211), KQU(10806585451386379021),
+	KQU(11120017753500499273), KQU( 1586837651387701301),
+	KQU(17530454400954415544), KQU( 9991670045077880430),
+	KQU( 7550997268990730180), KQU( 8640249196597379304),
+	KQU( 3522203892786893823), KQU(10401116549878854788),
+	KQU(13690285544733124852), KQU( 8295785675455774586),
+	KQU(15535716172155117603), KQU( 3112108583723722511),
+	KQU(17633179955339271113), KQU(18154208056063759375),
+	KQU( 1866409236285815666), KQU(13326075895396412882),
+	KQU( 8756261842948020025), KQU( 6281852999868439131),
+	KQU(15087653361275292858), KQU(10333923911152949397),
+	KQU( 5265567645757408500), KQU(12728041843210352184),
+	KQU( 6347959327507828759), KQU(  154112802625564758),
+	KQU(18235228308679780218), KQU( 3253805274673352418),
+	KQU( 4849171610689031197), KQU(17948529398340432518),
+	KQU(13803510475637409167), KQU(13506570190409883095),
+	KQU(15870801273282960805), KQU( 8451286481299170773),
+	KQU( 9562190620034457541), KQU( 8518905387449138364),
+	KQU(12681306401363385655), KQU( 3788073690559762558),
+	KQU( 5256820289573487769), KQU( 2752021372314875467),
+	KQU( 6354035166862520716), KQU( 4328956378309739069),
+	KQU(  449087441228269600), KQU( 5533508742653090868),
+	KQU( 1260389420404746988), KQU(18175394473289055097),
+	KQU( 1535467109660399420), KQU( 8818894282874061442),
+	KQU(12140873243824811213), KQU(15031386653823014946),
+	KQU( 1286028221456149232), KQU( 6329608889367858784),
+	KQU( 9419654354945132725), KQU( 6094576547061672379),
+	KQU(17706217251847450255), KQU( 1733495073065878126),
+	KQU(16918923754607552663), KQU( 8881949849954945044),
+	KQU(12938977706896313891), KQU(14043628638299793407),
+	KQU(18393874581723718233), KQU( 6886318534846892044),
+	KQU(14577870878038334081), KQU(13541558383439414119),
+	KQU(13570472158807588273), KQU(18300760537910283361),
+	KQU(  818368572800609205), KQU( 1417000585112573219),
+	KQU(12337533143867683655), KQU(12433180994702314480),
+	KQU(  778190005829189083), KQU(13667356216206524711),
+	KQU( 9866149895295225230), KQU(11043240490417111999),
+	KQU( 1123933826541378598), KQU( 6469631933605123610),
+	KQU(14508554074431980040), KQU(13918931242962026714),
+	KQU( 2870785929342348285), KQU(14786362626740736974),
+	KQU(13176680060902695786), KQU( 9591778613541679456),
+	KQU( 9097662885117436706), KQU(  749262234240924947),
+	KQU( 1944844067793307093), KQU( 4339214904577487742),
+	KQU( 8009584152961946551), KQU(16073159501225501777),
+	KQU( 3335870590499306217), KQU(17088312653151202847),
+	KQU( 3108893142681931848), KQU(16636841767202792021),
+	KQU(10423316431118400637), KQU( 8008357368674443506),
+	KQU(11340015231914677875), KQU(17687896501594936090),
+	KQU(15173627921763199958), KQU(  542569482243721959),
+	KQU(15071714982769812975), KQU( 4466624872151386956),
+	KQU( 1901780715602332461), KQU( 9822227742154351098),
+	KQU( 1479332892928648780), KQU( 6981611948382474400),
+	KQU( 7620824924456077376), KQU(14095973329429406782),
+	KQU( 7902744005696185404), KQU(15830577219375036920),
+	KQU(10287076667317764416), KQU(12334872764071724025),
+	KQU( 4419302088133544331), KQU(14455842851266090520),
+	KQU(12488077416504654222), KQU( 7953892017701886766),
+	KQU( 6331484925529519007), KQU( 4902145853785030022),
+	KQU(17010159216096443073), KQU(11945354668653886087),
+	KQU(15112022728645230829), KQU(17363484484522986742),
+	KQU( 4423497825896692887), KQU( 8155489510809067471),
+	KQU(  258966605622576285), KQU( 5462958075742020534),
+	KQU( 6763710214913276228), KQU( 2368935183451109054),
+	KQU(14209506165246453811), KQU( 2646257040978514881),
+	KQU( 3776001911922207672), KQU( 1419304601390147631),
+	KQU(14987366598022458284), KQU( 3977770701065815721),
+	KQU(  730820417451838898), KQU( 3982991703612885327),
+	KQU( 2803544519671388477), KQU(17067667221114424649),
+	KQU( 2922555119737867166), KQU( 1989477584121460932),
+	KQU(15020387605892337354), KQU( 9293277796427533547),
+	KQU(10722181424063557247), KQU(16704542332047511651),
+	KQU( 5008286236142089514), KQU(16174732308747382540),
+	KQU(17597019485798338402), KQU(13081745199110622093),
+	KQU( 8850305883842258115), KQU(12723629125624589005),
+	KQU( 8140566453402805978), KQU(15356684607680935061),
+	KQU(14222190387342648650), KQU(11134610460665975178),
+	KQU( 1259799058620984266), KQU(13281656268025610041),
+	KQU(  298262561068153992), KQU(12277871700239212922),
+	KQU(13911297774719779438), KQU(16556727962761474934),
+	KQU(17903010316654728010), KQU( 9682617699648434744),
+	KQU(14757681836838592850), KQU( 1327242446558524473),
+	KQU(11126645098780572792), KQU( 1883602329313221774),
+	KQU( 2543897783922776873), KQU(15029168513767772842),
+	KQU(12710270651039129878), KQU(16118202956069604504),
+	KQU(15010759372168680524), KQU( 2296827082251923948),
+	KQU(10793729742623518101), KQU(13829764151845413046),
+	KQU(17769301223184451213), KQU( 3118268169210783372),
+	KQU(17626204544105123127), KQU( 7416718488974352644),
+	KQU(10450751996212925994), KQU( 9352529519128770586),
+	KQU(  259347569641110140), KQU( 8048588892269692697),
+	KQU( 1774414152306494058), KQU(10669548347214355622),
+	KQU(13061992253816795081), KQU(18432677803063861659),
+	KQU( 8879191055593984333), KQU(12433753195199268041),
+	KQU(14919392415439730602), KQU( 6612848378595332963),
+	KQU( 6320986812036143628), KQU(10465592420226092859),
+	KQU( 4196009278962570808), KQU( 3747816564473572224),
+	KQU(17941203486133732898), KQU( 2350310037040505198),
+	KQU( 5811779859134370113), KQU(10492109599506195126),
+	KQU( 7699650690179541274), KQU( 1954338494306022961),
+	KQU(14095816969027231152), KQU( 5841346919964852061),
+	KQU(14945969510148214735), KQU( 3680200305887550992),
+	KQU( 6218047466131695792), KQU( 8242165745175775096),
+	KQU(11021371934053307357), KQU( 1265099502753169797),
+	KQU( 4644347436111321718), KQU( 3609296916782832859),
+	KQU( 8109807992218521571), KQU(18387884215648662020),
+	KQU(14656324896296392902), KQU(17386819091238216751),
+	KQU(17788300878582317152), KQU( 7919446259742399591),
+	KQU( 4466613134576358004), KQU(12928181023667938509),
+	KQU(13147446154454932030), KQU(16552129038252734620),
+	KQU( 8395299403738822450), KQU(11313817655275361164),
+	KQU(  434258809499511718), KQU( 2074882104954788676),
+	KQU( 7929892178759395518), KQU( 9006461629105745388),
+	KQU( 5176475650000323086), KQU(11128357033468341069),
+	KQU(12026158851559118955), KQU(14699716249471156500),
+	KQU(  448982497120206757), KQU( 4156475356685519900),
+	KQU( 6063816103417215727), KQU(10073289387954971479),
+	KQU( 8174466846138590962), KQU( 2675777452363449006),
+	KQU( 9090685420572474281), KQU( 6659652652765562060),
+	KQU(12923120304018106621), KQU(11117480560334526775),
+	KQU(  937910473424587511), KQU( 1838692113502346645),
+	KQU(11133914074648726180), KQU( 7922600945143884053),
+	KQU(13435287702700959550), KQU( 5287964921251123332),
+	KQU(11354875374575318947), KQU(17955724760748238133),
+	KQU(13728617396297106512), KQU( 4107449660118101255),
+	KQU( 1210269794886589623), KQU(11408687205733456282),
+	KQU( 4538354710392677887), KQU(13566803319341319267),
+	KQU(17870798107734050771), KQU( 3354318982568089135),
+	KQU( 9034450839405133651), KQU(13087431795753424314),
+	KQU(  950333102820688239), KQU( 1968360654535604116),
+	KQU(16840551645563314995), KQU( 8867501803892924995),
+	KQU(11395388644490626845), KQU( 1529815836300732204),
+	KQU(13330848522996608842), KQU( 1813432878817504265),
+	KQU( 2336867432693429560), KQU(15192805445973385902),
+	KQU( 2528593071076407877), KQU(  128459777936689248),
+	KQU( 9976345382867214866), KQU( 6208885766767996043),
+	KQU(14982349522273141706), KQU( 3099654362410737822),
+	KQU(13776700761947297661), KQU( 8806185470684925550),
+	KQU( 8151717890410585321), KQU(  640860591588072925),
+	KQU(14592096303937307465), KQU( 9056472419613564846),
+	KQU(14861544647742266352), KQU(12703771500398470216),
+	KQU( 3142372800384138465), KQU( 6201105606917248196),
+	KQU(18337516409359270184), KQU(15042268695665115339),
+	KQU(15188246541383283846), KQU(12800028693090114519),
+	KQU( 5992859621101493472), KQU(18278043971816803521),
+	KQU( 9002773075219424560), KQU( 7325707116943598353),
+	KQU( 7930571931248040822), KQU( 5645275869617023448),
+	KQU( 7266107455295958487), KQU( 4363664528273524411),
+	KQU(14313875763787479809), KQU(17059695613553486802),
+	KQU( 9247761425889940932), KQU(13704726459237593128),
+	KQU( 2701312427328909832), KQU(17235532008287243115),
+	KQU(14093147761491729538), KQU( 6247352273768386516),
+	KQU( 8268710048153268415), KQU( 7985295214477182083),
+	KQU(15624495190888896807), KQU( 3772753430045262788),
+	KQU( 9133991620474991698), KQU( 5665791943316256028),
+	KQU( 7551996832462193473), KQU(13163729206798953877),
+	KQU( 9263532074153846374), KQU( 1015460703698618353),
+	KQU(17929874696989519390), KQU(18257884721466153847),
+	KQU(16271867543011222991), KQU( 3905971519021791941),
+	KQU(16814488397137052085), KQU( 1321197685504621613),
+	KQU( 2870359191894002181), KQU(14317282970323395450),
+	KQU(13663920845511074366), KQU( 2052463995796539594),
+	KQU(14126345686431444337), KQU( 1727572121947022534),
+	KQU(17793552254485594241), KQU( 6738857418849205750),
+	KQU( 1282987123157442952), KQU(16655480021581159251),
+	KQU( 6784587032080183866), KQU(14726758805359965162),
+	KQU( 7577995933961987349), KQU(12539609320311114036),
+	KQU(10789773033385439494), KQU( 8517001497411158227),
+	KQU(10075543932136339710), KQU(14838152340938811081),
+	KQU( 9560840631794044194), KQU(17445736541454117475),
+	KQU(10633026464336393186), KQU(15705729708242246293),
+	KQU( 1117517596891411098), KQU( 4305657943415886942),
+	KQU( 4948856840533979263), KQU(16071681989041789593),
+	KQU(13723031429272486527), KQU( 7639567622306509462),
+	KQU(12670424537483090390), KQU( 9715223453097197134),
+	KQU( 5457173389992686394), KQU(  289857129276135145),
+	KQU(17048610270521972512), KQU(  692768013309835485),
+	KQU(14823232360546632057), KQU(18218002361317895936),
+	KQU( 3281724260212650204), KQU(16453957266549513795),
+	KQU( 8592711109774511881), KQU(  929825123473369579),
+	KQU(15966784769764367791), KQU( 9627344291450607588),
+	KQU(10849555504977813287), KQU( 9234566913936339275),
+	KQU( 6413807690366911210), KQU(10862389016184219267),
+	KQU(13842504799335374048), KQU( 1531994113376881174),
+	KQU( 2081314867544364459), KQU(16430628791616959932),
+	KQU( 8314714038654394368), KQU( 9155473892098431813),
+	KQU(12577843786670475704), KQU( 4399161106452401017),
+	KQU( 1668083091682623186), KQU( 1741383777203714216),
+	KQU( 2162597285417794374), KQU(15841980159165218736),
+	KQU( 1971354603551467079), KQU( 1206714764913205968),
+	KQU( 4790860439591272330), KQU(14699375615594055799),
+	KQU( 8374423871657449988), KQU(10950685736472937738),
+	KQU(  697344331343267176), KQU(10084998763118059810),
+	KQU(12897369539795983124), KQU(12351260292144383605),
+	KQU( 1268810970176811234), KQU( 7406287800414582768),
+	KQU(  516169557043807831), KQU( 5077568278710520380),
+	KQU( 3828791738309039304), KQU( 7721974069946943610),
+	KQU( 3534670260981096460), KQU( 4865792189600584891),
+	KQU(16892578493734337298), KQU( 9161499464278042590),
+	KQU(11976149624067055931), KQU(13219479887277343990),
+	KQU(14161556738111500680), KQU(14670715255011223056),
+	KQU( 4671205678403576558), KQU(12633022931454259781),
+	KQU(14821376219869187646), KQU(  751181776484317028),
+	KQU( 2192211308839047070), KQU(11787306362361245189),
+	KQU(10672375120744095707), KQU( 4601972328345244467),
+	KQU(15457217788831125879), KQU( 8464345256775460809),
+	KQU(10191938789487159478), KQU( 6184348739615197613),
+	KQU(11425436778806882100), KQU( 2739227089124319793),
+	KQU(  461464518456000551), KQU( 4689850170029177442),
+	KQU( 6120307814374078625), KQU(11153579230681708671),
+	KQU( 7891721473905347926), KQU(10281646937824872400),
+	KQU( 3026099648191332248), KQU( 8666750296953273818),
+	KQU(14978499698844363232), KQU(13303395102890132065),
+	KQU( 8182358205292864080), KQU(10560547713972971291),
+	KQU(11981635489418959093), KQU( 3134621354935288409),
+	KQU(11580681977404383968), KQU(14205530317404088650),
+	KQU( 5997789011854923157), KQU(13659151593432238041),
+	KQU(11664332114338865086), KQU( 7490351383220929386),
+	KQU( 7189290499881530378), KQU(15039262734271020220),
+	KQU( 2057217285976980055), KQU(  555570804905355739),
+	KQU(11235311968348555110), KQU(13824557146269603217),
+	KQU(16906788840653099693), KQU( 7222878245455661677),
+	KQU( 5245139444332423756), KQU( 4723748462805674292),
+	KQU(12216509815698568612), KQU(17402362976648951187),
+	KQU(17389614836810366768), KQU( 4880936484146667711),
+	KQU( 9085007839292639880), KQU(13837353458498535449),
+	KQU(11914419854360366677), KQU(16595890135313864103),
+	KQU( 6313969847197627222), KQU(18296909792163910431),
+	KQU(10041780113382084042), KQU( 2499478551172884794),
+	KQU(11057894246241189489), KQU( 9742243032389068555),
+	KQU(12838934582673196228), KQU(13437023235248490367),
+	KQU(13372420669446163240), KQU( 6752564244716909224),
+	KQU( 7157333073400313737), KQU(12230281516370654308),
+	KQU( 1182884552219419117), KQU( 2955125381312499218),
+	KQU(10308827097079443249), KQU( 1337648572986534958),
+	KQU(16378788590020343939), KQU(  108619126514420935),
+	KQU( 3990981009621629188), KQU( 5460953070230946410),
+	KQU( 9703328329366531883), KQU(13166631489188077236),
+	KQU( 1104768831213675170), KQU( 3447930458553877908),
+	KQU( 8067172487769945676), KQU( 5445802098190775347),
+	KQU( 3244840981648973873), KQU(17314668322981950060),
+	KQU( 5006812527827763807), KQU(18158695070225526260),
+	KQU( 2824536478852417853), KQU(13974775809127519886),
+	KQU( 9814362769074067392), KQU(17276205156374862128),
+	KQU(11361680725379306967), KQU( 3422581970382012542),
+	KQU(11003189603753241266), KQU(11194292945277862261),
+	KQU( 6839623313908521348), KQU(11935326462707324634),
+	KQU( 1611456788685878444), KQU(13112620989475558907),
+	KQU(  517659108904450427), KQU(13558114318574407624),
+	KQU(15699089742731633077), KQU( 4988979278862685458),
+	KQU( 8111373583056521297), KQU( 3891258746615399627),
+	KQU( 8137298251469718086), KQU(12748663295624701649),
+	KQU( 4389835683495292062), KQU( 5775217872128831729),
+	KQU( 9462091896405534927), KQU( 8498124108820263989),
+	KQU( 8059131278842839525), KQU(10503167994254090892),
+	KQU(11613153541070396656), KQU(18069248738504647790),
+	KQU(  570657419109768508), KQU( 3950574167771159665),
+	KQU( 5514655599604313077), KQU( 2908460854428484165),
+	KQU(10777722615935663114), KQU(12007363304839279486),
+	KQU( 9800646187569484767), KQU( 8795423564889864287),
+	KQU(14257396680131028419), KQU( 6405465117315096498),
+	KQU( 7939411072208774878), KQU(17577572378528990006),
+	KQU(14785873806715994850), KQU(16770572680854747390),
+	KQU(18127549474419396481), KQU(11637013449455757750),
+	KQU(14371851933996761086), KQU( 3601181063650110280),
+	KQU( 4126442845019316144), KQU(10198287239244320669),
+	KQU(18000169628555379659), KQU(18392482400739978269),
+	KQU( 6219919037686919957), KQU( 3610085377719446052),
+	KQU( 2513925039981776336), KQU(16679413537926716955),
+	KQU(12903302131714909434), KQU( 5581145789762985009),
+	KQU(12325955044293303233), KQU(17216111180742141204),
+	KQU( 6321919595276545740), KQU( 3507521147216174501),
+	KQU( 9659194593319481840), KQU(11473976005975358326),
+	KQU(14742730101435987026), KQU(  492845897709954780),
+	KQU(16976371186162599676), KQU(17712703422837648655),
+	KQU( 9881254778587061697), KQU( 8413223156302299551),
+	KQU( 1563841828254089168), KQU( 9996032758786671975),
+	KQU(  138877700583772667), KQU(13003043368574995989),
+	KQU( 4390573668650456587), KQU( 8610287390568126755),
+	KQU(15126904974266642199), KQU( 6703637238986057662),
+	KQU( 2873075592956810157), KQU( 6035080933946049418),
+	KQU(13382846581202353014), KQU( 7303971031814642463),
+	KQU(18418024405307444267), KQU( 5847096731675404647),
+	KQU( 4035880699639842500), KQU(11525348625112218478),
+	KQU( 3041162365459574102), KQU( 2604734487727986558),
+	KQU(15526341771636983145), KQU(14556052310697370254),
+	KQU(12997787077930808155), KQU( 9601806501755554499),
+	KQU(11349677952521423389), KQU(14956777807644899350),
+	KQU(16559736957742852721), KQU(12360828274778140726),
+	KQU( 6685373272009662513), KQU(16932258748055324130),
+	KQU(15918051131954158508), KQU( 1692312913140790144),
+	KQU(  546653826801637367), KQU( 5341587076045986652),
+	KQU(14975057236342585662), KQU(12374976357340622412),
+	KQU(10328833995181940552), KQU(12831807101710443149),
+	KQU(10548514914382545716), KQU( 2217806727199715993),
+	KQU(12627067369242845138), KQU( 4598965364035438158),
+	KQU(  150923352751318171), KQU(14274109544442257283),
+	KQU( 4696661475093863031), KQU( 1505764114384654516),
+	KQU(10699185831891495147), KQU( 2392353847713620519),
+	KQU( 3652870166711788383), KQU( 8640653276221911108),
+	KQU( 3894077592275889704), KQU( 4918592872135964845),
+	KQU(16379121273281400789), KQU(12058465483591683656),
+	KQU(11250106829302924945), KQU( 1147537556296983005),
+	KQU( 6376342756004613268), KQU(14967128191709280506),
+	KQU(18007449949790627628), KQU( 9497178279316537841),
+	KQU( 7920174844809394893), KQU(10037752595255719907),
+	KQU(15875342784985217697), KQU(15311615921712850696),
+	KQU( 9552902652110992950), KQU(14054979450099721140),
+	KQU( 5998709773566417349), KQU(18027910339276320187),
+	KQU( 8223099053868585554), KQU( 7842270354824999767),
+	KQU( 4896315688770080292), KQU(12969320296569787895),
+	KQU( 2674321489185759961), KQU( 4053615936864718439),
+	KQU(11349775270588617578), KQU( 4743019256284553975),
+	KQU( 5602100217469723769), KQU(14398995691411527813),
+	KQU( 7412170493796825470), KQU(  836262406131744846),
+	KQU( 8231086633845153022), KQU( 5161377920438552287),
+	KQU( 8828731196169924949), KQU(16211142246465502680),
+	KQU( 3307990879253687818), KQU( 5193405406899782022),
+	KQU( 8510842117467566693), KQU( 6070955181022405365),
+	KQU(14482950231361409799), KQU(12585159371331138077),
+	KQU( 3511537678933588148), KQU( 2041849474531116417),
+	KQU(10944936685095345792), KQU(18303116923079107729),
+	KQU( 2720566371239725320), KQU( 4958672473562397622),
+	KQU( 3032326668253243412), KQU(13689418691726908338),
+	KQU( 1895205511728843996), KQU( 8146303515271990527),
+	KQU(16507343500056113480), KQU(  473996939105902919),
+	KQU( 9897686885246881481), KQU(14606433762712790575),
+	KQU( 6732796251605566368), KQU( 1399778120855368916),
+	KQU(  935023885182833777), KQU(16066282816186753477),
+	KQU( 7291270991820612055), KQU(17530230393129853844),
+	KQU(10223493623477451366), KQU(15841725630495676683),
+	KQU(17379567246435515824), KQU( 8588251429375561971),
+	KQU(18339511210887206423), KQU(17349587430725976100),
+	KQU(12244876521394838088), KQU( 6382187714147161259),
+	KQU(12335807181848950831), KQU(16948885622305460665),
+	KQU(13755097796371520506), KQU(14806740373324947801),
+	KQU( 4828699633859287703), KQU( 8209879281452301604),
+	KQU(12435716669553736437), KQU(13970976859588452131),
+	KQU( 6233960842566773148), KQU(12507096267900505759),
+	KQU( 1198713114381279421), KQU(14989862731124149015),
+	KQU(15932189508707978949), KQU( 2526406641432708722),
+	KQU(   29187427817271982), KQU( 1499802773054556353),
+	KQU(10816638187021897173), KQU( 5436139270839738132),
+	KQU( 6659882287036010082), KQU( 2154048955317173697),
+	KQU(10887317019333757642), KQU(16281091802634424955),
+	KQU(10754549879915384901), KQU(10760611745769249815),
+	KQU( 2161505946972504002), KQU( 5243132808986265107),
+	KQU(10129852179873415416), KQU(  710339480008649081),
+	KQU( 7802129453068808528), KQU(17967213567178907213),
+	KQU(15730859124668605599), KQU(13058356168962376502),
+	KQU( 3701224985413645909), KQU(14464065869149109264),
+	KQU( 9959272418844311646), KQU(10157426099515958752),
+	KQU(14013736814538268528), KQU(17797456992065653951),
+	KQU(17418878140257344806), KQU(15457429073540561521),
+	KQU( 2184426881360949378), KQU( 2062193041154712416),
+	KQU( 8553463347406931661), KQU( 4913057625202871854),
+	KQU( 2668943682126618425), KQU(17064444737891172288),
+	KQU( 4997115903913298637), KQU(12019402608892327416),
+	KQU(17603584559765897352), KQU(11367529582073647975),
+	KQU( 8211476043518436050), KQU( 8676849804070323674),
+	KQU(18431829230394475730), KQU(10490177861361247904),
+	KQU( 9508720602025651349), KQU( 7409627448555722700),
+	KQU( 5804047018862729008), KQU(11943858176893142594),
+	KQU(11908095418933847092), KQU( 5415449345715887652),
+	KQU( 1554022699166156407), KQU( 9073322106406017161),
+	KQU( 7080630967969047082), KQU(18049736940860732943),
+	KQU(12748714242594196794), KQU( 1226992415735156741),
+	KQU(17900981019609531193), KQU(11720739744008710999),
+	KQU( 3006400683394775434), KQU(11347974011751996028),
+	KQU( 3316999628257954608), KQU( 8384484563557639101),
+	KQU(18117794685961729767), KQU( 1900145025596618194),
+	KQU(17459527840632892676), KQU( 5634784101865710994),
+	KQU( 7918619300292897158), KQU( 3146577625026301350),
+	KQU( 9955212856499068767), KQU( 1873995843681746975),
+	KQU( 1561487759967972194), KQU( 8322718804375878474),
+	KQU(11300284215327028366), KQU( 4667391032508998982),
+	KQU( 9820104494306625580), KQU(17922397968599970610),
+	KQU( 1784690461886786712), KQU(14940365084341346821),
+	KQU( 5348719575594186181), KQU(10720419084507855261),
+	KQU(14210394354145143274), KQU( 2426468692164000131),
+	KQU(16271062114607059202), KQU(14851904092357070247),
+	KQU( 6524493015693121897), KQU( 9825473835127138531),
+	KQU(14222500616268569578), KQU(15521484052007487468),
+	KQU(14462579404124614699), KQU(11012375590820665520),
+	KQU(11625327350536084927), KQU(14452017765243785417),
+	KQU( 9989342263518766305), KQU( 3640105471101803790),
+	KQU( 4749866455897513242), KQU(13963064946736312044),
+	KQU(10007416591973223791), KQU(18314132234717431115),
+	KQU( 3286596588617483450), KQU( 7726163455370818765),
+	KQU( 7575454721115379328), KQU( 5308331576437663422),
+	KQU(18288821894903530934), KQU( 8028405805410554106),
+	KQU(15744019832103296628), KQU(  149765559630932100),
+	KQU( 6137705557200071977), KQU(14513416315434803615),
+	KQU(11665702820128984473), KQU(  218926670505601386),
+	KQU( 6868675028717769519), KQU(15282016569441512302),
+	KQU( 5707000497782960236), KQU( 6671120586555079567),
+	KQU( 2194098052618985448), KQU(16849577895477330978),
+	KQU(12957148471017466283), KQU( 1997805535404859393),
+	KQU( 1180721060263860490), KQU(13206391310193756958),
+	KQU(12980208674461861797), KQU( 3825967775058875366),
+	KQU(17543433670782042631), KQU( 1518339070120322730),
+	KQU(16344584340890991669), KQU( 2611327165318529819),
+	KQU(11265022723283422529), KQU( 4001552800373196817),
+	KQU(14509595890079346161), KQU( 3528717165416234562),
+	KQU(18153222571501914072), KQU( 9387182977209744425),
+	KQU(10064342315985580021), KQU(11373678413215253977),
+	KQU( 2308457853228798099), KQU( 9729042942839545302),
+	KQU( 7833785471140127746), KQU( 6351049900319844436),
+	KQU(14454610627133496067), KQU(12533175683634819111),
+	KQU(15570163926716513029), KQU(13356980519185762498)
+};
+
+TEST_BEGIN(test_gen_rand_32) {
+	uint32_t array32[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	uint32_t array32_2[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint32_t r32;
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size32(), BLOCK_SIZE,
+	    "Array size too small");
+	ctx = init_gen_rand(1234);
+	fill_array32(ctx, array32, BLOCK_SIZE);
+	fill_array32(ctx, array32_2, BLOCK_SIZE);
+	fini_gen_rand(ctx);
+
+	ctx = init_gen_rand(1234);
+	for (i = 0; i < BLOCK_SIZE; i++) {
+		if (i < COUNT_1) {
+			assert_u32_eq(array32[i], init_gen_rand_32_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32[i],
+		    "Mismatch at array32[%d]=%x, gen=%x", i, array32[i], r32);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32_2[i],
+		    "Mismatch at array32_2[%d]=%x, gen=%x", i, array32_2[i],
+		    r32);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+TEST_BEGIN(test_by_array_32) {
+	uint32_t array32[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	uint32_t array32_2[BLOCK_SIZE] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint32_t ini[4] = {0x1234, 0x5678, 0x9abc, 0xdef0};
+	uint32_t r32;
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size32(), BLOCK_SIZE,
+	    "Array size too small");
+	ctx = init_by_array(ini, 4);
+	fill_array32(ctx, array32, BLOCK_SIZE);
+	fill_array32(ctx, array32_2, BLOCK_SIZE);
+	fini_gen_rand(ctx);
+
+	ctx = init_by_array(ini, 4);
+	for (i = 0; i < BLOCK_SIZE; i++) {
+		if (i < COUNT_1) {
+			assert_u32_eq(array32[i], init_by_array_32_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32[i],
+		    "Mismatch at array32[%d]=%x, gen=%x", i, array32[i], r32);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r32 = gen_rand32(ctx);
+		assert_u32_eq(r32, array32_2[i],
+		    "Mismatch at array32_2[%d]=%x, gen=%x", i, array32_2[i],
+		    r32);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+TEST_BEGIN(test_gen_rand_64) {
+	uint64_t array64[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	uint64_t array64_2[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint64_t r;
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size64(), BLOCK_SIZE64,
+	    "Array size too small");
+	ctx = init_gen_rand(4321);
+	fill_array64(ctx, array64, BLOCK_SIZE64);
+	fill_array64(ctx, array64_2, BLOCK_SIZE64);
+	fini_gen_rand(ctx);
+
+	ctx = init_gen_rand(4321);
+	for (i = 0; i < BLOCK_SIZE64; i++) {
+		if (i < COUNT_1) {
+			assert_u64_eq(array64[i], init_gen_rand_64_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64[i],
+		    "Mismatch at array64[%d]=%"FMTx64", gen=%"FMTx64, i,
+		    array64[i], r);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64_2[i],
+		    "Mismatch at array64_2[%d]=%"FMTx64" gen=%"FMTx64"", i,
+		    array64_2[i], r);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+TEST_BEGIN(test_by_array_64) {
+	uint64_t array64[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	uint64_t array64_2[BLOCK_SIZE64] JEMALLOC_ATTR(aligned(16));
+	int i;
+	uint64_t r;
+	uint32_t ini[] = {5, 4, 3, 2, 1};
+	sfmt_t *ctx;
+
+	assert_d_le(get_min_array_size64(), BLOCK_SIZE64,
+	    "Array size too small");
+	ctx = init_by_array(ini, 5);
+	fill_array64(ctx, array64, BLOCK_SIZE64);
+	fill_array64(ctx, array64_2, BLOCK_SIZE64);
+	fini_gen_rand(ctx);
+
+	ctx = init_by_array(ini, 5);
+	for (i = 0; i < BLOCK_SIZE64; i++) {
+		if (i < COUNT_1) {
+			assert_u64_eq(array64[i], init_by_array_64_expected[i],
+			    "Output mismatch for i=%d", i);
+		}
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64[i],
+		    "Mismatch at array64[%d]=%"FMTx64" gen=%"FMTx64, i,
+		    array64[i], r);
+	}
+	for (i = 0; i < COUNT_2; i++) {
+		r = gen_rand64(ctx);
+		assert_u64_eq(r, array64_2[i],
+		    "Mismatch at array64_2[%d]=%"FMTx64" gen=%"FMTx64, i,
+		    array64_2[i], r);
+	}
+	fini_gen_rand(ctx);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_gen_rand_32,
+	    test_by_array_32,
+	    test_gen_rand_64,
+	    test_by_array_64);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/a0.c b/memkind-1.10.0/jemalloc/test/unit/a0.c
new file mode 100644
index 0000000..a27ab3f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/a0.c
@@ -0,0 +1,16 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_a0) {
+	void *p;
+
+	p = a0malloc(1);
+	assert_ptr_not_null(p, "Unexpected a0malloc() error");
+	a0dalloc(p);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_malloc_init(
+	    test_a0);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/arena_reset.c b/memkind-1.10.0/jemalloc/test/unit/arena_reset.c
new file mode 100644
index 0000000..b182f31
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/arena_reset.c
@@ -0,0 +1,349 @@
+#ifndef ARENA_RESET_PROF_C_
+#include "test/jemalloc_test.h"
+#endif
+
+#include "jemalloc/internal/extent_mmap.h"
+#include "jemalloc/internal/rtree.h"
+
+#include "test/extent_hooks.h"
+
+static unsigned
+get_nsizes_impl(const char *cmd) {
+	unsigned ret;
+	size_t z;
+
+	z = sizeof(unsigned);
+	assert_d_eq(mallctl(cmd, (void *)&ret, &z, NULL, 0), 0,
+	    "Unexpected mallctl(\"%s\", ...) failure", cmd);
+
+	return ret;
+}
+
+static unsigned
+get_nsmall(void) {
+	return get_nsizes_impl("arenas.nbins");
+}
+
+static unsigned
+get_nlarge(void) {
+	return get_nsizes_impl("arenas.nlextents");
+}
+
+static size_t
+get_size_impl(const char *cmd, size_t ind) {
+	size_t ret;
+	size_t z;
+	size_t mib[4];
+	size_t miblen = 4;
+
+	z = sizeof(size_t);
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = ind;
+	z = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&ret, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\", %zu], ...) failure", cmd, ind);
+
+	return ret;
+}
+
+static size_t
+get_small_size(size_t ind) {
+	return get_size_impl("arenas.bin.0.size", ind);
+}
+
+static size_t
+get_large_size(size_t ind) {
+	return get_size_impl("arenas.lextent.0.size", ind);
+}
+
+/* Like ivsalloc(), but safe to call on discarded allocations. */
+static size_t
+vsalloc(tsdn_t *tsdn, const void *ptr) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent;
+	szind_t szind;
+	if (rtree_extent_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, false, &extent, &szind)) {
+		return 0;
+	}
+
+	if (extent == NULL) {
+		return 0;
+	}
+	if (extent_state_get(extent) != extent_state_active) {
+		return 0;
+	}
+
+	if (szind == SC_NSIZES) {
+		return 0;
+	}
+
+	return sz_index2size(szind);
+}
+
+static unsigned
+do_arena_create(extent_hooks_t *h) {
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz,
+	    (void *)(h != NULL ? &h : NULL), (h != NULL ? sizeof(h) : 0)), 0,
+	    "Unexpected mallctl() failure");
+	return arena_ind;
+}
+
+static void
+do_arena_reset_pre(unsigned arena_ind, void ***ptrs, unsigned *nptrs) {
+#define NLARGE	32
+	unsigned nsmall, nlarge, i;
+	size_t sz;
+	int flags;
+	tsdn_t *tsdn;
+
+	flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+
+	nsmall = get_nsmall();
+	nlarge = get_nlarge() > NLARGE ? NLARGE : get_nlarge();
+	*nptrs = nsmall + nlarge;
+	*ptrs = (void **)malloc(*nptrs * sizeof(void *));
+	assert_ptr_not_null(*ptrs, "Unexpected malloc() failure");
+
+	/* Allocate objects with a wide range of sizes. */
+	for (i = 0; i < nsmall; i++) {
+		sz = get_small_size(i);
+		(*ptrs)[i] = mallocx(sz, flags);
+		assert_ptr_not_null((*ptrs)[i],
+		    "Unexpected mallocx(%zu, %#x) failure", sz, flags);
+	}
+	for (i = 0; i < nlarge; i++) {
+		sz = get_large_size(i);
+		(*ptrs)[nsmall + i] = mallocx(sz, flags);
+		assert_ptr_not_null((*ptrs)[i],
+		    "Unexpected mallocx(%zu, %#x) failure", sz, flags);
+	}
+
+	tsdn = tsdn_fetch();
+
+	/* Verify allocations. */
+	for (i = 0; i < *nptrs; i++) {
+		assert_zu_gt(ivsalloc(tsdn, (*ptrs)[i]), 0,
+		    "Allocation should have queryable size");
+	}
+}
+
+static void
+do_arena_reset_post(void **ptrs, unsigned nptrs, unsigned arena_ind) {
+	tsdn_t *tsdn;
+	unsigned i;
+
+	tsdn = tsdn_fetch();
+
+	if (have_background_thread) {
+		malloc_mutex_lock(tsdn,
+		    &background_thread_info_get(arena_ind)->mtx);
+	}
+	/* Verify allocations no longer exist. */
+	for (i = 0; i < nptrs; i++) {
+		assert_zu_eq(vsalloc(tsdn, ptrs[i]), 0,
+		    "Allocation should no longer exist");
+	}
+	if (have_background_thread) {
+		malloc_mutex_unlock(tsdn,
+		    &background_thread_info_get(arena_ind)->mtx);
+	}
+
+	free(ptrs);
+}
+
+static void
+do_arena_reset_destroy(const char *name, unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib(name, mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+static void
+do_arena_reset(unsigned arena_ind) {
+	do_arena_reset_destroy("arena.0.reset", arena_ind);
+}
+
+static void
+do_arena_destroy(unsigned arena_ind) {
+	do_arena_reset_destroy("arena.0.destroy", arena_ind);
+}
+
+TEST_BEGIN(test_arena_reset) {
+	unsigned arena_ind;
+	void **ptrs;
+	unsigned nptrs;
+
+	arena_ind = do_arena_create(NULL);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+	do_arena_reset(arena_ind);
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+}
+TEST_END
+
+static bool
+arena_i_initialized(unsigned arena_ind, bool refresh) {
+	bool initialized;
+	size_t mib[3];
+	size_t miblen, sz;
+
+	if (refresh) {
+		uint64_t epoch = 1;
+		assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+		    sizeof(epoch)), 0, "Unexpected mallctl() failure");
+	}
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.initialized", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	sz = sizeof(initialized);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&initialized, &sz, NULL,
+	    0), 0, "Unexpected mallctlbymib() failure");
+
+	return initialized;
+}
+
+TEST_BEGIN(test_arena_destroy_initial) {
+	assert_false(arena_i_initialized(MALLCTL_ARENAS_DESTROYED, false),
+	    "Destroyed arena stats should not be initialized");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_destroy_hooks_default) {
+	unsigned arena_ind, arena_ind_another, arena_ind_prev;
+	void **ptrs;
+	unsigned nptrs;
+
+	arena_ind = do_arena_create(NULL);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+
+	assert_false(arena_i_initialized(arena_ind, false),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(arena_ind, true),
+	    "Arena stats should be initialized");
+
+	/*
+	 * Create another arena before destroying one, to better verify arena
+	 * index reuse.
+	 */
+	arena_ind_another = do_arena_create(NULL);
+
+	do_arena_destroy(arena_ind);
+
+	assert_false(arena_i_initialized(arena_ind, true),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(MALLCTL_ARENAS_DESTROYED, false),
+	    "Destroyed arena stats should be initialized");
+
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+
+	arena_ind_prev = arena_ind;
+	arena_ind = do_arena_create(NULL);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+	assert_u_eq(arena_ind, arena_ind_prev,
+	    "Arena index should have been recycled");
+	do_arena_destroy(arena_ind);
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+
+	do_arena_destroy(arena_ind_another);
+}
+TEST_END
+
+/*
+ * Actually unmap extents, regardless of opt_retain, so that attempts to access
+ * a destroyed arena's memory will segfault.
+ */
+static bool
+extent_dalloc_unmap(extent_hooks_t *extent_hooks, void *addr, size_t size,
+    bool committed, unsigned arena_ind) {
+	TRACE_HOOK("%s(extent_hooks=%p, addr=%p, size=%zu, committed=%s, "
+	    "arena_ind=%u)\n", __func__, extent_hooks, addr, size, committed ?
+	    "true" : "false", arena_ind);
+	assert_ptr_eq(extent_hooks, &hooks,
+	    "extent_hooks should be same as pointer used to set hooks");
+	assert_ptr_eq(extent_hooks->dalloc, extent_dalloc_unmap,
+	    "Wrong hook function");
+	called_dalloc = true;
+	if (!try_dalloc) {
+		return true;
+	}
+	did_dalloc = true;
+	if (!maps_coalesce && opt_retain) {
+		return true;
+	}
+	pages_unmap(addr, size);
+	return false;
+}
+
+static extent_hooks_t hooks_orig;
+
+static extent_hooks_t hooks_unmap = {
+	extent_alloc_hook,
+	extent_dalloc_unmap, /* dalloc */
+	extent_destroy_hook,
+	extent_commit_hook,
+	extent_decommit_hook,
+	extent_purge_lazy_hook,
+	extent_purge_forced_hook,
+	extent_split_hook,
+	extent_merge_hook
+};
+
+TEST_BEGIN(test_arena_destroy_hooks_unmap) {
+	unsigned arena_ind;
+	void **ptrs;
+	unsigned nptrs;
+
+	extent_hooks_prep();
+	if (maps_coalesce) {
+		try_decommit = false;
+	}
+	memcpy(&hooks_orig, &hooks, sizeof(extent_hooks_t));
+	memcpy(&hooks, &hooks_unmap, sizeof(extent_hooks_t));
+
+	did_alloc = false;
+	arena_ind = do_arena_create(&hooks);
+	do_arena_reset_pre(arena_ind, &ptrs, &nptrs);
+
+	assert_true(did_alloc, "Expected alloc");
+
+	assert_false(arena_i_initialized(arena_ind, false),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(arena_ind, true),
+	    "Arena stats should be initialized");
+
+	did_dalloc = false;
+	do_arena_destroy(arena_ind);
+	assert_true(did_dalloc, "Expected dalloc");
+
+	assert_false(arena_i_initialized(arena_ind, true),
+	    "Arena stats should not be initialized");
+	assert_true(arena_i_initialized(MALLCTL_ARENAS_DESTROYED, false),
+	    "Destroyed arena stats should be initialized");
+
+	do_arena_reset_post(ptrs, nptrs, arena_ind);
+
+	memcpy(&hooks, &hooks_orig, sizeof(extent_hooks_t));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_arena_reset,
+	    test_arena_destroy_initial,
+	    test_arena_destroy_hooks_default,
+	    test_arena_destroy_hooks_unmap);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/arena_reset_prof.c b/memkind-1.10.0/jemalloc/test/unit/arena_reset_prof.c
new file mode 100644
index 0000000..38d8012
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/arena_reset_prof.c
@@ -0,0 +1,4 @@
+#include "test/jemalloc_test.h"
+#define ARENA_RESET_PROF_C_
+
+#include "arena_reset.c"
diff --git a/memkind-1.10.0/jemalloc/test/unit/arena_reset_prof.sh b/memkind-1.10.0/jemalloc/test/unit/arena_reset_prof.sh
new file mode 100644
index 0000000..041dc1c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/arena_reset_prof.sh
@@ -0,0 +1,3 @@
+#!/bin/sh
+
+export MALLOC_CONF="prof:true,lg_prof_sample:0"
diff --git a/memkind-1.10.0/jemalloc/test/unit/atomic.c b/memkind-1.10.0/jemalloc/test/unit/atomic.c
new file mode 100644
index 0000000..572d8d2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/atomic.c
@@ -0,0 +1,229 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * We *almost* have consistent short names (e.g. "u32" for uint32_t, "b" for
+ * bool, etc.  The one exception is that the short name for void * is "p" in
+ * some places and "ptr" in others.  In the long run it would be nice to unify
+ * these, but in the short run we'll use this shim.
+ */
+#define assert_p_eq assert_ptr_eq
+
+/*
+ * t: the non-atomic type, like "uint32_t".
+ * ta: the short name for the type, like "u32".
+ * val[1,2,3]: Values of the given type.  The CAS tests use val2 for expected,
+ * and val3 for desired.
+ */
+
+#define DO_TESTS(t, ta, val1, val2, val3) do {				\
+	t val;								\
+	t expected;							\
+	bool success;							\
+	/* This (along with the load below) also tests ATOMIC_LOAD. */	\
+	atomic_##ta##_t atom = ATOMIC_INIT(val1);			\
+									\
+	/* ATOMIC_INIT and load. */					\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1, val, "Load or init failed");		\
+									\
+	/* Store. */							\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	atomic_store_##ta(&atom, val2, ATOMIC_RELAXED);			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val2, val, "Store failed");			\
+									\
+	/* Exchange. */							\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_exchange_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val, "Exchange returned invalid value");	\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val2, val, "Exchange store invalid value");	\
+									\
+	/* 								\
+	 * Weak CAS.  Spurious failures are allowed, so we loop a few	\
+	 * times.							\
+	 */								\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	success = false;						\
+	for (int i = 0; i < 10 && !success; i++) {			\
+		expected = val2;					\
+		success = atomic_compare_exchange_weak_##ta(&atom,	\
+		    &expected, val3, ATOMIC_RELAXED, ATOMIC_RELAXED);	\
+		assert_##ta##_eq(val1, expected, 			\
+		    "CAS should update expected");			\
+	}								\
+	assert_b_eq(val1 == val2, success,				\
+	    "Weak CAS did the wrong state update");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	if (success) {							\
+		assert_##ta##_eq(val3, val,				\
+		    "Successful CAS should update atomic");		\
+	} else {							\
+		assert_##ta##_eq(val1, val,				\
+		    "Unsuccessful CAS should not update atomic");	\
+	}								\
+									\
+	/* Strong CAS. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	expected = val2;						\
+	success = atomic_compare_exchange_strong_##ta(&atom, &expected,	\
+	    val3, ATOMIC_RELAXED, ATOMIC_RELAXED);			\
+	assert_b_eq(val1 == val2, success,				\
+	    "Strong CAS did the wrong state update");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	if (success) {							\
+		assert_##ta##_eq(val3, val,				\
+		    "Successful CAS should update atomic");		\
+	} else {							\
+		assert_##ta##_eq(val1, val,				\
+		    "Unsuccessful CAS should not update atomic");	\
+	}								\
+									\
+									\
+} while (0)
+
+#define DO_INTEGER_TESTS(t, ta, val1, val2) do {			\
+	atomic_##ta##_t atom;						\
+	t val;								\
+									\
+	/* Fetch-add. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_add_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-add should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 + val2, val,				\
+	    "Fetch-add should update atomic");				\
+									\
+	/* Fetch-sub. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_sub_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-sub should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 - val2, val,				\
+	    "Fetch-sub should update atomic");				\
+									\
+	/* Fetch-and. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_and_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-and should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 & val2, val,				\
+	    "Fetch-and should update atomic");				\
+									\
+	/* Fetch-or. */							\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_or_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-or should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 | val2, val,				\
+	    "Fetch-or should update atomic");				\
+									\
+	/* Fetch-xor. */						\
+	atomic_store_##ta(&atom, val1, ATOMIC_RELAXED);			\
+	val = atomic_fetch_xor_##ta(&atom, val2, ATOMIC_RELAXED);	\
+	assert_##ta##_eq(val1, val,					\
+	    "Fetch-xor should return previous value");			\
+	val = atomic_load_##ta(&atom, ATOMIC_RELAXED);			\
+	assert_##ta##_eq(val1 ^ val2, val,				\
+	    "Fetch-xor should update atomic");				\
+} while (0)
+
+#define TEST_STRUCT(t, ta)						\
+typedef struct {							\
+	t val1;								\
+	t val2;								\
+	t val3;								\
+} ta##_test_t;
+
+#define TEST_CASES(t) {							\
+	{(t)-1, (t)-1, (t)-2},						\
+	{(t)-1, (t) 0, (t)-2},						\
+	{(t)-1, (t) 1, (t)-2},						\
+									\
+	{(t) 0, (t)-1, (t)-2},						\
+	{(t) 0, (t) 0, (t)-2},						\
+	{(t) 0, (t) 1, (t)-2},						\
+									\
+	{(t) 1, (t)-1, (t)-2},						\
+	{(t) 1, (t) 0, (t)-2},						\
+	{(t) 1, (t) 1, (t)-2},						\
+									\
+	{(t)0, (t)-(1 << 22), (t)-2},					\
+	{(t)0, (t)(1 << 22), (t)-2},					\
+	{(t)(1 << 22), (t)-(1 << 22), (t)-2},				\
+	{(t)(1 << 22), (t)(1 << 22), (t)-2}				\
+}
+
+#define TEST_BODY(t, ta) do {						\
+	const ta##_test_t tests[] = TEST_CASES(t);			\
+	for (unsigned i = 0; i < sizeof(tests)/sizeof(tests[0]); i++) {	\
+		ta##_test_t test = tests[i];				\
+		DO_TESTS(t, ta, test.val1, test.val2, test.val3);	\
+	}								\
+} while (0)
+
+#define INTEGER_TEST_BODY(t, ta) do {					\
+	const ta##_test_t tests[] = TEST_CASES(t);			\
+	for (unsigned i = 0; i < sizeof(tests)/sizeof(tests[0]); i++) {	\
+		ta##_test_t test = tests[i];				\
+		DO_TESTS(t, ta, test.val1, test.val2, test.val3);	\
+		DO_INTEGER_TESTS(t, ta, test.val1, test.val2);		\
+	}								\
+} while (0)
+
+TEST_STRUCT(uint64_t, u64);
+TEST_BEGIN(test_atomic_u64) {
+#if !(LG_SIZEOF_PTR == 3 || LG_SIZEOF_INT == 3)
+	test_skip("64-bit atomic operations not supported");
+#else
+	INTEGER_TEST_BODY(uint64_t, u64);
+#endif
+}
+TEST_END
+
+
+TEST_STRUCT(uint32_t, u32);
+TEST_BEGIN(test_atomic_u32) {
+	INTEGER_TEST_BODY(uint32_t, u32);
+}
+TEST_END
+
+TEST_STRUCT(void *, p);
+TEST_BEGIN(test_atomic_p) {
+	TEST_BODY(void *, p);
+}
+TEST_END
+
+TEST_STRUCT(size_t, zu);
+TEST_BEGIN(test_atomic_zu) {
+	INTEGER_TEST_BODY(size_t, zu);
+}
+TEST_END
+
+TEST_STRUCT(ssize_t, zd);
+TEST_BEGIN(test_atomic_zd) {
+	INTEGER_TEST_BODY(ssize_t, zd);
+}
+TEST_END
+
+
+TEST_STRUCT(unsigned, u);
+TEST_BEGIN(test_atomic_u) {
+	INTEGER_TEST_BODY(unsigned, u);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_atomic_u64,
+	    test_atomic_u32,
+	    test_atomic_p,
+	    test_atomic_zu,
+	    test_atomic_zd,
+	    test_atomic_u);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/background_thread.c b/memkind-1.10.0/jemalloc/test/unit/background_thread.c
new file mode 100644
index 0000000..f7bd37c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/background_thread.c
@@ -0,0 +1,119 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/util.h"
+
+static void
+test_switch_background_thread_ctl(bool new_val) {
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+
+	e1 = new_val;
+	assert_d_eq(mallctl("background_thread", (void *)&e0, &sz,
+	    &e1, sz), 0, "Unexpected mallctl() failure");
+	assert_b_eq(e0, !e1,
+	    "background_thread should be %d before.\n", !e1);
+	if (e1) {
+		assert_zu_gt(n_background_threads, 0,
+		    "Number of background threads should be non zero.\n");
+	} else {
+		assert_zu_eq(n_background_threads, 0,
+		    "Number of background threads should be zero.\n");
+	}
+}
+
+static void
+test_repeat_background_thread_ctl(bool before) {
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+
+	e1 = before;
+	assert_d_eq(mallctl("background_thread", (void *)&e0, &sz,
+	    &e1, sz), 0, "Unexpected mallctl() failure");
+	assert_b_eq(e0, before,
+	    "background_thread should be %d.\n", before);
+	if (e1) {
+		assert_zu_gt(n_background_threads, 0,
+		    "Number of background threads should be non zero.\n");
+	} else {
+		assert_zu_eq(n_background_threads, 0,
+		    "Number of background threads should be zero.\n");
+	}
+}
+
+TEST_BEGIN(test_background_thread_ctl) {
+	test_skip_if(!have_background_thread);
+
+	bool e0, e1;
+	size_t sz = sizeof(bool);
+
+	assert_d_eq(mallctl("opt.background_thread", (void *)&e0, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctl("background_thread", (void *)&e1, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() failure");
+	assert_b_eq(e0, e1,
+	    "Default and opt.background_thread does not match.\n");
+	if (e0) {
+		test_switch_background_thread_ctl(false);
+	}
+	assert_zu_eq(n_background_threads, 0,
+	    "Number of background threads should be 0.\n");
+
+	for (unsigned i = 0; i < 4; i++) {
+		test_switch_background_thread_ctl(true);
+		test_repeat_background_thread_ctl(true);
+		test_repeat_background_thread_ctl(true);
+
+		test_switch_background_thread_ctl(false);
+		test_repeat_background_thread_ctl(false);
+		test_repeat_background_thread_ctl(false);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_background_thread_running) {
+	test_skip_if(!have_background_thread);
+	test_skip_if(!config_stats);
+
+#if defined(JEMALLOC_BACKGROUND_THREAD)
+	tsd_t *tsd = tsd_fetch();
+	background_thread_info_t *info = &background_thread_info[0];
+
+	test_repeat_background_thread_ctl(false);
+	test_switch_background_thread_ctl(true);
+	assert_b_eq(info->state, background_thread_started,
+	    "Background_thread did not start.\n");
+
+	nstime_t start, now;
+	nstime_init(&start, 0);
+	nstime_update(&start);
+
+	bool ran = false;
+	while (true) {
+		malloc_mutex_lock(tsd_tsdn(tsd), &info->mtx);
+		if (info->tot_n_runs > 0) {
+			ran = true;
+		}
+		malloc_mutex_unlock(tsd_tsdn(tsd), &info->mtx);
+		if (ran) {
+			break;
+		}
+
+		nstime_init(&now, 0);
+		nstime_update(&now);
+		nstime_subtract(&now, &start);
+		assert_u64_lt(nstime_sec(&now), 1000,
+		    "Background threads did not run for 1000 seconds.");
+		sleep(1);
+	}
+	test_switch_background_thread_ctl(false);
+#endif
+}
+TEST_END
+
+int
+main(void) {
+	/* Background_thread creation tests reentrancy naturally. */
+	return test_no_reentrancy(
+	    test_background_thread_ctl,
+	    test_background_thread_running);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/background_thread_enable.c b/memkind-1.10.0/jemalloc/test/unit/background_thread_enable.c
new file mode 100644
index 0000000..d894e93
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/background_thread_enable.c
@@ -0,0 +1,85 @@
+#include "test/jemalloc_test.h"
+
+const char *malloc_conf = "background_thread:false,narenas:1,max_background_threads:20";
+
+TEST_BEGIN(test_deferred) {
+	test_skip_if(!have_background_thread);
+
+	unsigned id;
+	size_t sz_u = sizeof(unsigned);
+
+	/*
+	 * 10 here is somewhat arbitrary, except insofar as we want to ensure
+	 * that the number of background threads is smaller than the number of
+	 * arenas.  I'll ragequit long before we have to spin up 10 threads per
+	 * cpu to handle background purging, so this is a conservative
+	 * approximation.
+	 */
+	for (unsigned i = 0; i < 10 * ncpus; i++) {
+		assert_d_eq(mallctl("arenas.create", &id, &sz_u, NULL, 0), 0,
+		    "Failed to create arena");
+	}
+
+	bool enable = true;
+	size_t sz_b = sizeof(bool);
+	assert_d_eq(mallctl("background_thread", NULL, NULL, &enable, sz_b), 0,
+	    "Failed to enable background threads");
+	enable = false;
+	assert_d_eq(mallctl("background_thread", NULL, NULL, &enable, sz_b), 0,
+	    "Failed to disable background threads");
+}
+TEST_END
+
+TEST_BEGIN(test_max_background_threads) {
+	test_skip_if(!have_background_thread);
+
+	size_t max_n_thds;
+	size_t opt_max_n_thds;
+	size_t sz_m = sizeof(max_n_thds);
+	assert_d_eq(mallctl("opt.max_background_threads",
+	    &opt_max_n_thds, &sz_m, NULL, 0), 0,
+	    "Failed to get opt.max_background_threads");
+	assert_d_eq(mallctl("max_background_threads", &max_n_thds, &sz_m, NULL,
+	    0), 0, "Failed to get max background threads");
+	assert_zu_eq(opt_max_n_thds, max_n_thds,
+	    "max_background_threads and "
+	    "opt.max_background_threads should match");
+	assert_d_eq(mallctl("max_background_threads", NULL, NULL, &max_n_thds,
+	    sz_m), 0, "Failed to set max background threads");
+
+	unsigned id;
+	size_t sz_u = sizeof(unsigned);
+
+	for (unsigned i = 0; i < 10 * ncpus; i++) {
+		assert_d_eq(mallctl("arenas.create", &id, &sz_u, NULL, 0), 0,
+		    "Failed to create arena");
+	}
+
+	bool enable = true;
+	size_t sz_b = sizeof(bool);
+	assert_d_eq(mallctl("background_thread", NULL, NULL, &enable, sz_b), 0,
+	    "Failed to enable background threads");
+	assert_zu_eq(n_background_threads, max_n_thds,
+	    "Number of background threads should not change.\n");
+	size_t new_max_thds = max_n_thds - 1;
+	if (new_max_thds > 0) {
+		assert_d_eq(mallctl("max_background_threads", NULL, NULL,
+		    &new_max_thds, sz_m), 0,
+		    "Failed to set max background threads");
+		assert_zu_eq(n_background_threads, new_max_thds,
+		    "Number of background threads should decrease by 1.\n");
+	}
+	new_max_thds = 1;
+	assert_d_eq(mallctl("max_background_threads", NULL, NULL, &new_max_thds,
+	    sz_m), 0, "Failed to set max background threads");
+	assert_zu_eq(n_background_threads, new_max_thds,
+	    "Number of background threads should be 1.\n");
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+		test_deferred,
+		test_max_background_threads);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/base.c b/memkind-1.10.0/jemalloc/test/unit/base.c
new file mode 100644
index 0000000..6b792cf
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/base.c
@@ -0,0 +1,234 @@
+#include "test/jemalloc_test.h"
+
+#include "test/extent_hooks.h"
+
+static extent_hooks_t hooks_null = {
+	extent_alloc_hook,
+	NULL, /* dalloc */
+	NULL, /* destroy */
+	NULL, /* commit */
+	NULL, /* decommit */
+	NULL, /* purge_lazy */
+	NULL, /* purge_forced */
+	NULL, /* split */
+	NULL /* merge */
+};
+
+static extent_hooks_t hooks_not_null = {
+	extent_alloc_hook,
+	extent_dalloc_hook,
+	extent_destroy_hook,
+	NULL, /* commit */
+	extent_decommit_hook,
+	extent_purge_lazy_hook,
+	extent_purge_forced_hook,
+	NULL, /* split */
+	NULL /* merge */
+};
+
+TEST_BEGIN(test_base_hooks_default) {
+	base_t *base;
+	size_t allocated0, allocated1, resident, mapped, n_thp;
+
+	tsdn_t *tsdn = tsd_tsdn(tsd_fetch());
+	base = base_new(tsdn, 0, (extent_hooks_t *)&extent_hooks_default);
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated0, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated0, sizeof(base_t),
+		    "Base header should count as allocated");
+		if (opt_metadata_thp == metadata_thp_always) {
+			assert_zu_gt(n_thp, 0,
+			    "Base should have 1 THP at least.");
+		}
+	}
+
+	assert_ptr_not_null(base_alloc(tsdn, base, 42, 1),
+	    "Unexpected base_alloc() failure");
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated1, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated1 - allocated0, 42,
+		    "At least 42 bytes were allocated by base_alloc()");
+	}
+
+	base_delete(tsdn, base);
+}
+TEST_END
+
+TEST_BEGIN(test_base_hooks_null) {
+	extent_hooks_t hooks_orig;
+	base_t *base;
+	size_t allocated0, allocated1, resident, mapped, n_thp;
+
+	extent_hooks_prep();
+	try_dalloc = false;
+	try_destroy = true;
+	try_decommit = false;
+	try_purge_lazy = false;
+	try_purge_forced = false;
+	memcpy(&hooks_orig, &hooks, sizeof(extent_hooks_t));
+	memcpy(&hooks, &hooks_null, sizeof(extent_hooks_t));
+
+	tsdn_t *tsdn = tsd_tsdn(tsd_fetch());
+	base = base_new(tsdn, 0, &hooks);
+	assert_ptr_not_null(base, "Unexpected base_new() failure");
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated0, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated0, sizeof(base_t),
+		    "Base header should count as allocated");
+		if (opt_metadata_thp == metadata_thp_always) {
+			assert_zu_gt(n_thp, 0,
+			    "Base should have 1 THP at least.");
+		}
+	}
+
+	assert_ptr_not_null(base_alloc(tsdn, base, 42, 1),
+	    "Unexpected base_alloc() failure");
+
+	if (config_stats) {
+		base_stats_get(tsdn, base, &allocated1, &resident, &mapped,
+		    &n_thp);
+		assert_zu_ge(allocated1 - allocated0, 42,
+		    "At least 42 bytes were allocated by base_alloc()");
+	}
+
+	base_delete(tsdn, base);
+
+	memcpy(&hooks, &hooks_orig, sizeof(extent_hooks_t));
+}
+TEST_END
+
+TEST_BEGIN(test_base_hooks_not_null) {
+	extent_hooks_t hooks_orig;
+	base_t *base;
+	void *p, *q, *r, *r_exp;
+
+	extent_hooks_prep();
+	try_dalloc = false;
+	try_destroy = true;
+	try_decommit = false;
+	try_purge_lazy = false;
+	try_purge_forced = false;
+	memcpy(&hooks_orig, &hooks, sizeof(extent_hooks_t));
+	memcpy(&hooks, &hooks_not_null, sizeof(extent_hooks_t));
+
+	tsdn_t *tsdn = tsd_tsdn(tsd_fetch());
+	did_alloc = false;
+	base = base_new(tsdn, 0, &hooks);
+	assert_ptr_not_null(base, "Unexpected base_new() failure");
+	assert_true(did_alloc, "Expected alloc");
+
+	/*
+	 * Check for tight packing at specified alignment under simple
+	 * conditions.
+	 */
+	{
+		const size_t alignments[] = {
+			1,
+			QUANTUM,
+			QUANTUM << 1,
+			CACHELINE,
+			CACHELINE << 1,
+		};
+		unsigned i;
+
+		for (i = 0; i < sizeof(alignments) / sizeof(size_t); i++) {
+			size_t alignment = alignments[i];
+			size_t align_ceil = ALIGNMENT_CEILING(alignment,
+			    QUANTUM);
+			p = base_alloc(tsdn, base, 1, alignment);
+			assert_ptr_not_null(p,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq(p,
+			    (void *)(ALIGNMENT_CEILING((uintptr_t)p,
+			    alignment)), "Expected quantum alignment");
+			q = base_alloc(tsdn, base, alignment, alignment);
+			assert_ptr_not_null(q,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq((void *)((uintptr_t)p + align_ceil), q,
+			    "Minimal allocation should take up %zu bytes",
+			    align_ceil);
+			r = base_alloc(tsdn, base, 1, alignment);
+			assert_ptr_not_null(r,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq((void *)((uintptr_t)q + align_ceil), r,
+			    "Minimal allocation should take up %zu bytes",
+			    align_ceil);
+		}
+	}
+
+	/*
+	 * Allocate an object that cannot fit in the first block, then verify
+	 * that the first block's remaining space is considered for subsequent
+	 * allocation.
+	 */
+	assert_zu_ge(extent_bsize_get(&base->blocks->extent), QUANTUM,
+	    "Remainder insufficient for test");
+	/* Use up all but one quantum of block. */
+	while (extent_bsize_get(&base->blocks->extent) > QUANTUM) {
+		p = base_alloc(tsdn, base, QUANTUM, QUANTUM);
+		assert_ptr_not_null(p, "Unexpected base_alloc() failure");
+	}
+	r_exp = extent_addr_get(&base->blocks->extent);
+	assert_zu_eq(base->extent_sn_next, 1, "One extant block expected");
+	q = base_alloc(tsdn, base, QUANTUM + 1, QUANTUM);
+	assert_ptr_not_null(q, "Unexpected base_alloc() failure");
+	assert_ptr_ne(q, r_exp, "Expected allocation from new block");
+	assert_zu_eq(base->extent_sn_next, 2, "Two extant blocks expected");
+	r = base_alloc(tsdn, base, QUANTUM, QUANTUM);
+	assert_ptr_not_null(r, "Unexpected base_alloc() failure");
+	assert_ptr_eq(r, r_exp, "Expected allocation from first block");
+	assert_zu_eq(base->extent_sn_next, 2, "Two extant blocks expected");
+
+	/*
+	 * Check for proper alignment support when normal blocks are too small.
+	 */
+	{
+		const size_t alignments[] = {
+			HUGEPAGE,
+			HUGEPAGE << 1
+		};
+		unsigned i;
+
+		for (i = 0; i < sizeof(alignments) / sizeof(size_t); i++) {
+			size_t alignment = alignments[i];
+			p = base_alloc(tsdn, base, QUANTUM, alignment);
+			assert_ptr_not_null(p,
+			    "Unexpected base_alloc() failure");
+			assert_ptr_eq(p,
+			    (void *)(ALIGNMENT_CEILING((uintptr_t)p,
+			    alignment)), "Expected %zu-byte alignment",
+			    alignment);
+		}
+	}
+
+	called_dalloc = called_destroy = called_decommit = called_purge_lazy =
+	    called_purge_forced = false;
+	base_delete(tsdn, base);
+	assert_true(called_dalloc, "Expected dalloc call");
+	assert_true(!called_destroy, "Unexpected destroy call");
+	assert_true(called_decommit, "Expected decommit call");
+	assert_true(called_purge_lazy, "Expected purge_lazy call");
+	assert_true(called_purge_forced, "Expected purge_forced call");
+
+	try_dalloc = true;
+	try_destroy = true;
+	try_decommit = true;
+	try_purge_lazy = true;
+	try_purge_forced = true;
+	memcpy(&hooks, &hooks_orig, sizeof(extent_hooks_t));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_base_hooks_default,
+	    test_base_hooks_null,
+	    test_base_hooks_not_null);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/binshard.c b/memkind-1.10.0/jemalloc/test/unit/binshard.c
new file mode 100644
index 0000000..d7a8df8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/binshard.c
@@ -0,0 +1,154 @@
+#include "test/jemalloc_test.h"
+
+/* Config -- "narenas:1,bin_shards:1-160:16|129-512:4|256-256:8" */
+
+#define NTHREADS 16
+#define REMOTE_NALLOC 256
+
+static void *
+thd_producer(void *varg) {
+	void **mem = varg;
+	unsigned arena, i;
+	size_t sz;
+
+	sz = sizeof(arena);
+	/* Remote arena. */
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	for (i = 0; i < REMOTE_NALLOC / 2; i++) {
+		mem[i] = mallocx(1, MALLOCX_TCACHE_NONE | MALLOCX_ARENA(arena));
+	}
+
+	/* Remote bin. */
+	for (; i < REMOTE_NALLOC; i++) {
+		mem[i] = mallocx(1, MALLOCX_TCACHE_NONE | MALLOCX_ARENA(0));
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_producer_consumer) {
+	thd_t thds[NTHREADS];
+	void *mem[NTHREADS][REMOTE_NALLOC];
+	unsigned i;
+
+	/* Create producer threads to allocate. */
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_producer, mem[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+	/* Remote deallocation by the current thread. */
+	for (i = 0; i < NTHREADS; i++) {
+		for (unsigned j = 0; j < REMOTE_NALLOC; j++) {
+			assert_ptr_not_null(mem[i][j],
+			    "Unexpected remote allocation failure");
+			dallocx(mem[i][j], 0);
+		}
+	}
+}
+TEST_END
+
+static void *
+thd_start(void *varg) {
+	void *ptr, *ptr2;
+	extent_t *extent;
+	unsigned shard1, shard2;
+
+	tsdn_t *tsdn = tsdn_fetch();
+	/* Try triggering allocations from sharded bins. */
+	for (unsigned i = 0; i < 1024; i++) {
+		ptr = mallocx(1, MALLOCX_TCACHE_NONE);
+		ptr2 = mallocx(129, MALLOCX_TCACHE_NONE);
+
+		extent = iealloc(tsdn, ptr);
+		shard1 = extent_binshard_get(extent);
+		dallocx(ptr, 0);
+		assert_u_lt(shard1, 16, "Unexpected bin shard used");
+
+		extent = iealloc(tsdn, ptr2);
+		shard2 = extent_binshard_get(extent);
+		dallocx(ptr2, 0);
+		assert_u_lt(shard2, 4, "Unexpected bin shard used");
+
+		if (shard1 > 0 || shard2 > 0) {
+			/* Triggered sharded bin usage. */
+			return (void *)(uintptr_t)shard1;
+		}
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_bin_shard_mt) {
+	test_skip_if(have_percpu_arena &&
+	    PERCPU_ARENA_ENABLED(opt_percpu_arena));
+
+	thd_t thds[NTHREADS];
+	unsigned i;
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start, NULL);
+	}
+	bool sharded = false;
+	for (i = 0; i < NTHREADS; i++) {
+		void *ret;
+		thd_join(thds[i], &ret);
+		if (ret != NULL) {
+			sharded = true;
+		}
+	}
+	assert_b_eq(sharded, true, "Did not find sharded bins");
+}
+TEST_END
+
+TEST_BEGIN(test_bin_shard) {
+	unsigned nbins, i;
+	size_t mib[4], mib2[4];
+	size_t miblen, miblen2, len;
+
+	len = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &len, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	miblen = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.nshards", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	miblen2 = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.size", mib2, &miblen2), 0,
+	    "Unexpected mallctlnametomib() failure");
+
+	for (i = 0; i < nbins; i++) {
+		uint32_t nshards;
+		size_t size, sz1, sz2;
+
+		mib[2] = i;
+		sz1 = sizeof(nshards);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&nshards, &sz1,
+		    NULL, 0), 0, "Unexpected mallctlbymib() failure");
+
+		mib2[2] = i;
+		sz2 = sizeof(size);
+		assert_d_eq(mallctlbymib(mib2, miblen2, (void *)&size, &sz2,
+		    NULL, 0), 0, "Unexpected mallctlbymib() failure");
+
+		if (size >= 1 && size <= 128) {
+			assert_u_eq(nshards, 16, "Unexpected nshards");
+		} else if (size == 256) {
+			assert_u_eq(nshards, 8, "Unexpected nshards");
+		} else if (size > 128 && size <= 512) {
+			assert_u_eq(nshards, 4, "Unexpected nshards");
+		} else {
+			assert_u_eq(nshards, 1, "Unexpected nshards");
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_bin_shard,
+	    test_bin_shard_mt,
+	    test_producer_consumer);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/binshard.sh b/memkind-1.10.0/jemalloc/test/unit/binshard.sh
new file mode 100644
index 0000000..c1d58c8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/binshard.sh
@@ -0,0 +1,3 @@
+#!/bin/sh
+
+export MALLOC_CONF="narenas:1,bin_shards:1-160:16|129-512:4|256-256:8"
diff --git a/memkind-1.10.0/jemalloc/test/unit/bit_util.c b/memkind-1.10.0/jemalloc/test/unit/bit_util.c
new file mode 100644
index 0000000..b747deb
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/bit_util.c
@@ -0,0 +1,111 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/bit_util.h"
+
+#define TEST_POW2_CEIL(t, suf, pri) do {				\
+	unsigned i, pow2;						\
+	t x;								\
+									\
+	assert_##suf##_eq(pow2_ceil_##suf(0), 0, "Unexpected result");	\
+									\
+	for (i = 0; i < sizeof(t) * 8; i++) {				\
+		assert_##suf##_eq(pow2_ceil_##suf(((t)1) << i), ((t)1)	\
+		    << i, "Unexpected result");				\
+	}								\
+									\
+	for (i = 2; i < sizeof(t) * 8; i++) {				\
+		assert_##suf##_eq(pow2_ceil_##suf((((t)1) << i) - 1),	\
+		    ((t)1) << i, "Unexpected result");			\
+	}								\
+									\
+	for (i = 0; i < sizeof(t) * 8 - 1; i++) {			\
+		assert_##suf##_eq(pow2_ceil_##suf((((t)1) << i) + 1),	\
+		    ((t)1) << (i+1), "Unexpected result");		\
+	}								\
+									\
+	for (pow2 = 1; pow2 < 25; pow2++) {				\
+		for (x = (((t)1) << (pow2-1)) + 1; x <= ((t)1) << pow2;	\
+		    x++) {						\
+			assert_##suf##_eq(pow2_ceil_##suf(x),		\
+			    ((t)1) << pow2,				\
+			    "Unexpected result, x=%"pri, x);		\
+		}							\
+	}								\
+} while (0)
+
+TEST_BEGIN(test_pow2_ceil_u64) {
+	TEST_POW2_CEIL(uint64_t, u64, FMTu64);
+}
+TEST_END
+
+TEST_BEGIN(test_pow2_ceil_u32) {
+	TEST_POW2_CEIL(uint32_t, u32, FMTu32);
+}
+TEST_END
+
+TEST_BEGIN(test_pow2_ceil_zu) {
+	TEST_POW2_CEIL(size_t, zu, "zu");
+}
+TEST_END
+
+void
+assert_lg_ceil_range(size_t input, unsigned answer) {
+	if (input == 1) {
+		assert_u_eq(0, answer, "Got %u as lg_ceil of 1", answer);
+		return;
+	}
+	assert_zu_le(input, (ZU(1) << answer),
+	    "Got %u as lg_ceil of %zu", answer, input);
+	assert_zu_gt(input, (ZU(1) << (answer - 1)),
+	    "Got %u as lg_ceil of %zu", answer, input);
+}
+
+void
+assert_lg_floor_range(size_t input, unsigned answer) {
+	if (input == 1) {
+		assert_u_eq(0, answer, "Got %u as lg_floor of 1", answer);
+		return;
+	}
+	assert_zu_ge(input, (ZU(1) << answer),
+	    "Got %u as lg_floor of %zu", answer, input);
+	assert_zu_lt(input, (ZU(1) << (answer + 1)),
+	    "Got %u as lg_floor of %zu", answer, input);
+}
+
+TEST_BEGIN(test_lg_ceil_floor) {
+	for (size_t i = 1; i < 10 * 1000 * 1000; i++) {
+		assert_lg_ceil_range(i, lg_ceil(i));
+		assert_lg_ceil_range(i, LG_CEIL(i));
+		assert_lg_floor_range(i, lg_floor(i));
+		assert_lg_floor_range(i, LG_FLOOR(i));
+	}
+	for (int i = 10; i < 8 * (1 << LG_SIZEOF_PTR) - 5; i++) {
+		for (size_t j = 0; j < (1 << 4); j++) {
+			size_t num1 = ((size_t)1 << i)
+			    - j * ((size_t)1 << (i - 4));
+			size_t num2 = ((size_t)1 << i)
+			    + j * ((size_t)1 << (i - 4));
+			assert_zu_ne(num1, 0, "Invalid lg argument");
+			assert_zu_ne(num2, 0, "Invalid lg argument");
+			assert_lg_ceil_range(num1, lg_ceil(num1));
+			assert_lg_ceil_range(num1, LG_CEIL(num1));
+			assert_lg_ceil_range(num2, lg_ceil(num2));
+			assert_lg_ceil_range(num2, LG_CEIL(num2));
+
+			assert_lg_floor_range(num1, lg_floor(num1));
+			assert_lg_floor_range(num1, LG_FLOOR(num1));
+			assert_lg_floor_range(num2, lg_floor(num2));
+			assert_lg_floor_range(num2, LG_FLOOR(num2));
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_pow2_ceil_u64,
+	    test_pow2_ceil_u32,
+	    test_pow2_ceil_zu,
+	    test_lg_ceil_floor);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/bitmap.c b/memkind-1.10.0/jemalloc/test/unit/bitmap.c
new file mode 100644
index 0000000..cafb203
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/bitmap.c
@@ -0,0 +1,431 @@
+#include "test/jemalloc_test.h"
+
+#define NBITS_TAB \
+    NB( 1) \
+    NB( 2) \
+    NB( 3) \
+    NB( 4) \
+    NB( 5) \
+    NB( 6) \
+    NB( 7) \
+    NB( 8) \
+    NB( 9) \
+    NB(10) \
+    NB(11) \
+    NB(12) \
+    NB(13) \
+    NB(14) \
+    NB(15) \
+    NB(16) \
+    NB(17) \
+    NB(18) \
+    NB(19) \
+    NB(20) \
+    NB(21) \
+    NB(22) \
+    NB(23) \
+    NB(24) \
+    NB(25) \
+    NB(26) \
+    NB(27) \
+    NB(28) \
+    NB(29) \
+    NB(30) \
+    NB(31) \
+    NB(32) \
+    \
+    NB(33) \
+    NB(34) \
+    NB(35) \
+    NB(36) \
+    NB(37) \
+    NB(38) \
+    NB(39) \
+    NB(40) \
+    NB(41) \
+    NB(42) \
+    NB(43) \
+    NB(44) \
+    NB(45) \
+    NB(46) \
+    NB(47) \
+    NB(48) \
+    NB(49) \
+    NB(50) \
+    NB(51) \
+    NB(52) \
+    NB(53) \
+    NB(54) \
+    NB(55) \
+    NB(56) \
+    NB(57) \
+    NB(58) \
+    NB(59) \
+    NB(60) \
+    NB(61) \
+    NB(62) \
+    NB(63) \
+    NB(64) \
+    NB(65) \
+    \
+    NB(126) \
+    NB(127) \
+    NB(128) \
+    NB(129) \
+    NB(130) \
+    \
+    NB(254) \
+    NB(255) \
+    NB(256) \
+    NB(257) \
+    NB(258) \
+    \
+    NB(510) \
+    NB(511) \
+    NB(512) \
+    NB(513) \
+    NB(514) \
+    \
+    NB(1024) \
+    NB(2048) \
+    NB(4096) \
+    NB(8192) \
+    NB(16384) \
+
+static void
+test_bitmap_initializer_body(const bitmap_info_t *binfo, size_t nbits) {
+	bitmap_info_t binfo_dyn;
+	bitmap_info_init(&binfo_dyn, nbits);
+
+	assert_zu_eq(bitmap_size(binfo), bitmap_size(&binfo_dyn),
+	    "Unexpected difference between static and dynamic initialization, "
+	    "nbits=%zu", nbits);
+	assert_zu_eq(binfo->nbits, binfo_dyn.nbits,
+	    "Unexpected difference between static and dynamic initialization, "
+	    "nbits=%zu", nbits);
+#ifdef BITMAP_USE_TREE
+	assert_u_eq(binfo->nlevels, binfo_dyn.nlevels,
+	    "Unexpected difference between static and dynamic initialization, "
+	    "nbits=%zu", nbits);
+	{
+		unsigned i;
+
+		for (i = 0; i < binfo->nlevels; i++) {
+			assert_zu_eq(binfo->levels[i].group_offset,
+			    binfo_dyn.levels[i].group_offset,
+			    "Unexpected difference between static and dynamic "
+			    "initialization, nbits=%zu, level=%u", nbits, i);
+		}
+	}
+#else
+	assert_zu_eq(binfo->ngroups, binfo_dyn.ngroups,
+	    "Unexpected difference between static and dynamic initialization");
+#endif
+}
+
+TEST_BEGIN(test_bitmap_initializer) {
+#define NB(nbits) {							\
+		if (nbits <= BITMAP_MAXBITS) {				\
+			bitmap_info_t binfo =				\
+			    BITMAP_INFO_INITIALIZER(nbits);		\
+			test_bitmap_initializer_body(&binfo, nbits);	\
+		}							\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static size_t
+test_bitmap_size_body(const bitmap_info_t *binfo, size_t nbits,
+    size_t prev_size) {
+	size_t size = bitmap_size(binfo);
+	assert_zu_ge(size, (nbits >> 3),
+	    "Bitmap size is smaller than expected");
+	assert_zu_ge(size, prev_size, "Bitmap size is smaller than expected");
+	return size;
+}
+
+TEST_BEGIN(test_bitmap_size) {
+	size_t nbits, prev_size;
+
+	prev_size = 0;
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		prev_size = test_bitmap_size_body(&binfo, nbits, prev_size);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		prev_size = test_bitmap_size_body(&binfo, nbits,	\
+		    prev_size);						\
+	}
+	prev_size = 0;
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_init_body(const bitmap_info_t *binfo, size_t nbits) {
+	size_t i;
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+
+	bitmap_init(bitmap, binfo, false);
+	for (i = 0; i < nbits; i++) {
+		assert_false(bitmap_get(bitmap, binfo, i),
+		    "Bit should be unset");
+	}
+
+	bitmap_init(bitmap, binfo, true);
+	for (i = 0; i < nbits; i++) {
+		assert_true(bitmap_get(bitmap, binfo, i), "Bit should be set");
+	}
+
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_init) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_init_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_init_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_set_body(const bitmap_info_t *binfo, size_t nbits) {
+	size_t i;
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+	bitmap_init(bitmap, binfo, false);
+
+	for (i = 0; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i);
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_set) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_set_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_set_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_unset_body(const bitmap_info_t *binfo, size_t nbits) {
+	size_t i;
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+	bitmap_init(bitmap, binfo, false);
+
+	for (i = 0; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i);
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+	for (i = 0; i < nbits; i++) {
+		bitmap_unset(bitmap, binfo, i);
+	}
+	for (i = 0; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i);
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_unset) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_unset_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_unset_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+static void
+test_bitmap_xfu_body(const bitmap_info_t *binfo, size_t nbits) {
+	bitmap_t *bitmap = (bitmap_t *)malloc(bitmap_size(binfo));
+	assert_ptr_not_null(bitmap, "Unexpected malloc() failure");
+	bitmap_init(bitmap, binfo, false);
+
+	/* Iteratively set bits starting at the beginning. */
+	for (size_t i = 0; i < nbits; i++) {
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, (i > 0) ? i-1 : i), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+		    "First unset bit should be just after previous first unset "
+		    "bit");
+	}
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+
+	/*
+	 * Iteratively unset bits starting at the end, and verify that
+	 * bitmap_sfu() reaches the unset bits.
+	 */
+	for (size_t i = nbits - 1; i < nbits; i--) { /* (nbits..0] */
+		bitmap_unset(bitmap, binfo, i);
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), i,
+		    "First unset bit should the bit previously unset");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, (i > 0) ? i-1 : i), i,
+		    "First unset bit should the bit previously unset");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+		    "First unset bit should the bit previously unset");
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+		    "First unset bit should the bit previously unset");
+		bitmap_unset(bitmap, binfo, i);
+	}
+	assert_false(bitmap_get(bitmap, binfo, 0), "Bit should be unset");
+
+	/*
+	 * Iteratively set bits starting at the beginning, and verify that
+	 * bitmap_sfu() looks past them.
+	 */
+	for (size_t i = 1; i < nbits; i++) {
+		bitmap_set(bitmap, binfo, i - 1);
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, (i > 0) ? i-1 : i), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+		    "First unset bit should be just after the bit previously "
+		    "set");
+		bitmap_unset(bitmap, binfo, i);
+	}
+	assert_zu_eq(bitmap_ffu(bitmap, binfo, 0), nbits - 1,
+	    "First unset bit should be the last bit");
+	assert_zu_eq(bitmap_ffu(bitmap, binfo, (nbits > 1) ? nbits-2 : nbits-1),
+	    nbits - 1, "First unset bit should be the last bit");
+	assert_zu_eq(bitmap_ffu(bitmap, binfo, nbits - 1), nbits - 1,
+	    "First unset bit should be the last bit");
+	assert_zu_eq(bitmap_sfu(bitmap, binfo), nbits - 1,
+	    "First unset bit should be the last bit");
+	assert_true(bitmap_full(bitmap, binfo), "All bits should be set");
+
+	/*
+	 * Bubble a "usu" pattern through the bitmap and verify that
+	 * bitmap_ffu() finds the correct bit for all five min_bit cases.
+	 */
+	if (nbits >= 3) {
+		for (size_t i = 0; i < nbits-2; i++) {
+			bitmap_unset(bitmap, binfo, i);
+			bitmap_unset(bitmap, binfo, i+2);
+			if (i > 0) {
+				assert_zu_eq(bitmap_ffu(bitmap, binfo, i-1), i,
+				    "Unexpected first unset bit");
+			}
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i+1), i+2,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i+2), i+2,
+			    "Unexpected first unset bit");
+			if (i + 3 < nbits) {
+				assert_zu_eq(bitmap_ffu(bitmap, binfo, i+3),
+				    nbits, "Unexpected first unset bit");
+			}
+			assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_sfu(bitmap, binfo), i+2,
+			    "Unexpected first unset bit");
+		}
+	}
+
+	/*
+	 * Unset the last bit, bubble another unset bit through the bitmap, and
+	 * verify that bitmap_ffu() finds the correct bit for all four min_bit
+	 * cases.
+	 */
+	if (nbits >= 3) {
+		bitmap_unset(bitmap, binfo, nbits-1);
+		for (size_t i = 0; i < nbits-1; i++) {
+			bitmap_unset(bitmap, binfo, i);
+			if (i > 0) {
+				assert_zu_eq(bitmap_ffu(bitmap, binfo, i-1), i,
+				    "Unexpected first unset bit");
+			}
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i), i,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, i+1), nbits-1,
+			    "Unexpected first unset bit");
+			assert_zu_eq(bitmap_ffu(bitmap, binfo, nbits-1),
+			    nbits-1, "Unexpected first unset bit");
+
+			assert_zu_eq(bitmap_sfu(bitmap, binfo), i,
+			    "Unexpected first unset bit");
+		}
+		assert_zu_eq(bitmap_sfu(bitmap, binfo), nbits-1,
+		    "Unexpected first unset bit");
+	}
+
+	free(bitmap);
+}
+
+TEST_BEGIN(test_bitmap_xfu) {
+	size_t nbits;
+
+	for (nbits = 1; nbits <= BITMAP_MAXBITS; nbits++) {
+		bitmap_info_t binfo;
+		bitmap_info_init(&binfo, nbits);
+		test_bitmap_xfu_body(&binfo, nbits);
+	}
+#define NB(nbits) {							\
+		bitmap_info_t binfo = BITMAP_INFO_INITIALIZER(nbits);	\
+		test_bitmap_xfu_body(&binfo, nbits);			\
+	}
+	NBITS_TAB
+#undef NB
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_bitmap_initializer,
+	    test_bitmap_size,
+	    test_bitmap_init,
+	    test_bitmap_set,
+	    test_bitmap_unset,
+	    test_bitmap_xfu);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/ckh.c b/memkind-1.10.0/jemalloc/test/unit/ckh.c
new file mode 100644
index 0000000..707ea5f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/ckh.c
@@ -0,0 +1,211 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_new_delete) {
+	tsd_t *tsd;
+	ckh_t ckh;
+
+	tsd = tsd_fetch();
+
+	assert_false(ckh_new(tsd, &ckh, 2, ckh_string_hash,
+	    ckh_string_keycomp), "Unexpected ckh_new() error");
+	ckh_delete(tsd, &ckh);
+
+	assert_false(ckh_new(tsd, &ckh, 3, ckh_pointer_hash,
+	    ckh_pointer_keycomp), "Unexpected ckh_new() error");
+	ckh_delete(tsd, &ckh);
+}
+TEST_END
+
+TEST_BEGIN(test_count_insert_search_remove) {
+	tsd_t *tsd;
+	ckh_t ckh;
+	const char *strs[] = {
+	    "a string",
+	    "A string",
+	    "a string.",
+	    "A string."
+	};
+	const char *missing = "A string not in the hash table.";
+	size_t i;
+
+	tsd = tsd_fetch();
+
+	assert_false(ckh_new(tsd, &ckh, 2, ckh_string_hash,
+	    ckh_string_keycomp), "Unexpected ckh_new() error");
+	assert_zu_eq(ckh_count(&ckh), 0,
+	    "ckh_count() should return %zu, but it returned %zu", ZU(0),
+	    ckh_count(&ckh));
+
+	/* Insert. */
+	for (i = 0; i < sizeof(strs)/sizeof(const char *); i++) {
+		ckh_insert(tsd, &ckh, strs[i], strs[i]);
+		assert_zu_eq(ckh_count(&ckh), i+1,
+		    "ckh_count() should return %zu, but it returned %zu", i+1,
+		    ckh_count(&ckh));
+	}
+
+	/* Search. */
+	for (i = 0; i < sizeof(strs)/sizeof(const char *); i++) {
+		union {
+			void *p;
+			const char *s;
+		} k, v;
+		void **kp, **vp;
+		const char *ks, *vs;
+
+		kp = (i & 1) ? &k.p : NULL;
+		vp = (i & 2) ? &v.p : NULL;
+		k.p = NULL;
+		v.p = NULL;
+		assert_false(ckh_search(&ckh, strs[i], kp, vp),
+		    "Unexpected ckh_search() error");
+
+		ks = (i & 1) ? strs[i] : (const char *)NULL;
+		vs = (i & 2) ? strs[i] : (const char *)NULL;
+		assert_ptr_eq((void *)ks, (void *)k.s, "Key mismatch, i=%zu",
+		    i);
+		assert_ptr_eq((void *)vs, (void *)v.s, "Value mismatch, i=%zu",
+		    i);
+	}
+	assert_true(ckh_search(&ckh, missing, NULL, NULL),
+	    "Unexpected ckh_search() success");
+
+	/* Remove. */
+	for (i = 0; i < sizeof(strs)/sizeof(const char *); i++) {
+		union {
+			void *p;
+			const char *s;
+		} k, v;
+		void **kp, **vp;
+		const char *ks, *vs;
+
+		kp = (i & 1) ? &k.p : NULL;
+		vp = (i & 2) ? &v.p : NULL;
+		k.p = NULL;
+		v.p = NULL;
+		assert_false(ckh_remove(tsd, &ckh, strs[i], kp, vp),
+		    "Unexpected ckh_remove() error");
+
+		ks = (i & 1) ? strs[i] : (const char *)NULL;
+		vs = (i & 2) ? strs[i] : (const char *)NULL;
+		assert_ptr_eq((void *)ks, (void *)k.s, "Key mismatch, i=%zu",
+		    i);
+		assert_ptr_eq((void *)vs, (void *)v.s, "Value mismatch, i=%zu",
+		    i);
+		assert_zu_eq(ckh_count(&ckh),
+		    sizeof(strs)/sizeof(const char *) - i - 1,
+		    "ckh_count() should return %zu, but it returned %zu",
+		        sizeof(strs)/sizeof(const char *) - i - 1,
+		    ckh_count(&ckh));
+	}
+
+	ckh_delete(tsd, &ckh);
+}
+TEST_END
+
+TEST_BEGIN(test_insert_iter_remove) {
+#define NITEMS ZU(1000)
+	tsd_t *tsd;
+	ckh_t ckh;
+	void **p[NITEMS];
+	void *q, *r;
+	size_t i;
+
+	tsd = tsd_fetch();
+
+	assert_false(ckh_new(tsd, &ckh, 2, ckh_pointer_hash,
+	    ckh_pointer_keycomp), "Unexpected ckh_new() error");
+
+	for (i = 0; i < NITEMS; i++) {
+		p[i] = mallocx(i+1, 0);
+		assert_ptr_not_null(p[i], "Unexpected mallocx() failure");
+	}
+
+	for (i = 0; i < NITEMS; i++) {
+		size_t j;
+
+		for (j = i; j < NITEMS; j++) {
+			assert_false(ckh_insert(tsd, &ckh, p[j], p[j]),
+			    "Unexpected ckh_insert() failure");
+			assert_false(ckh_search(&ckh, p[j], &q, &r),
+			    "Unexpected ckh_search() failure");
+			assert_ptr_eq(p[j], q, "Key pointer mismatch");
+			assert_ptr_eq(p[j], r, "Value pointer mismatch");
+		}
+
+		assert_zu_eq(ckh_count(&ckh), NITEMS,
+		    "ckh_count() should return %zu, but it returned %zu",
+		    NITEMS, ckh_count(&ckh));
+
+		for (j = i + 1; j < NITEMS; j++) {
+			assert_false(ckh_search(&ckh, p[j], NULL, NULL),
+			    "Unexpected ckh_search() failure");
+			assert_false(ckh_remove(tsd, &ckh, p[j], &q, &r),
+			    "Unexpected ckh_remove() failure");
+			assert_ptr_eq(p[j], q, "Key pointer mismatch");
+			assert_ptr_eq(p[j], r, "Value pointer mismatch");
+			assert_true(ckh_search(&ckh, p[j], NULL, NULL),
+			    "Unexpected ckh_search() success");
+			assert_true(ckh_remove(tsd, &ckh, p[j], &q, &r),
+			    "Unexpected ckh_remove() success");
+		}
+
+		{
+			bool seen[NITEMS];
+			size_t tabind;
+
+			memset(seen, 0, sizeof(seen));
+
+			for (tabind = 0; !ckh_iter(&ckh, &tabind, &q, &r);) {
+				size_t k;
+
+				assert_ptr_eq(q, r, "Key and val not equal");
+
+				for (k = 0; k < NITEMS; k++) {
+					if (p[k] == q) {
+						assert_false(seen[k],
+						    "Item %zu already seen", k);
+						seen[k] = true;
+						break;
+					}
+				}
+			}
+
+			for (j = 0; j < i + 1; j++) {
+				assert_true(seen[j], "Item %zu not seen", j);
+			}
+			for (; j < NITEMS; j++) {
+				assert_false(seen[j], "Item %zu seen", j);
+			}
+		}
+	}
+
+	for (i = 0; i < NITEMS; i++) {
+		assert_false(ckh_search(&ckh, p[i], NULL, NULL),
+		    "Unexpected ckh_search() failure");
+		assert_false(ckh_remove(tsd, &ckh, p[i], &q, &r),
+		    "Unexpected ckh_remove() failure");
+		assert_ptr_eq(p[i], q, "Key pointer mismatch");
+		assert_ptr_eq(p[i], r, "Value pointer mismatch");
+		assert_true(ckh_search(&ckh, p[i], NULL, NULL),
+		    "Unexpected ckh_search() success");
+		assert_true(ckh_remove(tsd, &ckh, p[i], &q, &r),
+		    "Unexpected ckh_remove() success");
+		dallocx(p[i], 0);
+	}
+
+	assert_zu_eq(ckh_count(&ckh), 0,
+	    "ckh_count() should return %zu, but it returned %zu",
+	    ZU(0), ckh_count(&ckh));
+	ckh_delete(tsd, &ckh);
+#undef NITEMS
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_new_delete,
+	    test_count_insert_search_remove,
+	    test_insert_iter_remove);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/decay.c b/memkind-1.10.0/jemalloc/test/unit/decay.c
new file mode 100644
index 0000000..cf3c079
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/decay.c
@@ -0,0 +1,605 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ticker.h"
+
+static nstime_monotonic_t *nstime_monotonic_orig;
+static nstime_update_t *nstime_update_orig;
+
+static unsigned nupdates_mock;
+static nstime_t time_mock;
+static bool monotonic_mock;
+
+static bool
+check_background_thread_enabled(void) {
+	bool enabled;
+	size_t sz = sizeof(bool);
+	int ret = mallctl("background_thread", (void *)&enabled, &sz, NULL,0);
+	if (ret == ENOENT) {
+		return false;
+	}
+	assert_d_eq(ret, 0, "Unexpected mallctl error");
+	return enabled;
+}
+
+static bool
+nstime_monotonic_mock(void) {
+	return monotonic_mock;
+}
+
+static bool
+nstime_update_mock(nstime_t *time) {
+	nupdates_mock++;
+	if (monotonic_mock) {
+		nstime_copy(time, &time_mock);
+	}
+	return !monotonic_mock;
+}
+
+static unsigned
+do_arena_create(ssize_t dirty_decay_ms, ssize_t muzzy_decay_ms) {
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib("arena.0.dirty_decay_ms", mib, &miblen),
+	    0, "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(dirty_decay_ms)), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	assert_d_eq(mallctlnametomib("arena.0.muzzy_decay_ms", mib, &miblen),
+	    0, "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(muzzy_decay_ms)), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	return arena_ind;
+}
+
+static void
+do_arena_destroy(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.destroy", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+void
+do_epoch(void) {
+	uint64_t epoch = 1;
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+}
+
+void
+do_purge(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.purge", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+void
+do_decay(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.decay", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+static uint64_t
+get_arena_npurge_impl(const char *mibname, unsigned arena_ind) {
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib(mibname, mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[2] = (size_t)arena_ind;
+	uint64_t npurge = 0;
+	size_t sz = sizeof(npurge);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&npurge, &sz, NULL, 0),
+	    config_stats ? 0 : ENOENT, "Unexpected mallctlbymib() failure");
+	return npurge;
+}
+
+static uint64_t
+get_arena_dirty_npurge(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.dirty_npurge", arena_ind);
+}
+
+static uint64_t
+get_arena_dirty_purged(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.dirty_purged", arena_ind);
+}
+
+static uint64_t
+get_arena_muzzy_npurge(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.muzzy_npurge", arena_ind);
+}
+
+static uint64_t
+get_arena_npurge(unsigned arena_ind) {
+	do_epoch();
+	return get_arena_npurge_impl("stats.arenas.0.dirty_npurge", arena_ind) +
+	    get_arena_npurge_impl("stats.arenas.0.muzzy_npurge", arena_ind);
+}
+
+static size_t
+get_arena_pdirty(unsigned arena_ind) {
+	do_epoch();
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("stats.arenas.0.pdirty", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[2] = (size_t)arena_ind;
+	size_t pdirty;
+	size_t sz = sizeof(pdirty);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&pdirty, &sz, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+	return pdirty;
+}
+
+static size_t
+get_arena_pmuzzy(unsigned arena_ind) {
+	do_epoch();
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("stats.arenas.0.pmuzzy", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[2] = (size_t)arena_ind;
+	size_t pmuzzy;
+	size_t sz = sizeof(pmuzzy);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&pmuzzy, &sz, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+	return pmuzzy;
+}
+
+static void *
+do_mallocx(size_t size, int flags) {
+	void *p = mallocx(size, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	return p;
+}
+
+static void
+generate_dirty(unsigned arena_ind, size_t size) {
+	int flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+	void *p = do_mallocx(size, flags);
+	dallocx(p, flags);
+}
+
+TEST_BEGIN(test_decay_ticks) {
+	test_skip_if(check_background_thread_enabled());
+
+	ticker_t *decay_ticker;
+	unsigned tick0, tick1, arena_ind;
+	size_t sz, large0;
+	void *p;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large0, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	/* Set up a manually managed arena for test. */
+	arena_ind = do_arena_create(0, 0);
+
+	/* Migrate to the new arena, and get the ticker. */
+	unsigned old_arena_ind;
+	size_t sz_arena_ind = sizeof(old_arena_ind);
+	assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind,
+	    &sz_arena_ind, (void *)&arena_ind, sizeof(arena_ind)), 0,
+	    "Unexpected mallctl() failure");
+	decay_ticker = decay_ticker_get(tsd_fetch(), arena_ind);
+	assert_ptr_not_null(decay_ticker,
+	    "Unexpected failure getting decay ticker");
+
+	/*
+	 * Test the standard APIs using a large size class, since we can't
+	 * control tcache interactions for small size classes (except by
+	 * completely disabling tcache for the entire test program).
+	 */
+
+	/* malloc(). */
+	tick0 = ticker_read(decay_ticker);
+	p = malloc(large0);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during malloc()");
+	/* free(). */
+	tick0 = ticker_read(decay_ticker);
+	free(p);
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during free()");
+
+	/* calloc(). */
+	tick0 = ticker_read(decay_ticker);
+	p = calloc(1, large0);
+	assert_ptr_not_null(p, "Unexpected calloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during calloc()");
+	free(p);
+
+	/* posix_memalign(). */
+	tick0 = ticker_read(decay_ticker);
+	assert_d_eq(posix_memalign(&p, sizeof(size_t), large0), 0,
+	    "Unexpected posix_memalign() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0,
+	    "Expected ticker to tick during posix_memalign()");
+	free(p);
+
+	/* aligned_alloc(). */
+	tick0 = ticker_read(decay_ticker);
+	p = aligned_alloc(sizeof(size_t), large0);
+	assert_ptr_not_null(p, "Unexpected aligned_alloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0,
+	    "Expected ticker to tick during aligned_alloc()");
+	free(p);
+
+	/* realloc(). */
+	/* Allocate. */
+	tick0 = ticker_read(decay_ticker);
+	p = realloc(NULL, large0);
+	assert_ptr_not_null(p, "Unexpected realloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during realloc()");
+	/* Reallocate. */
+	tick0 = ticker_read(decay_ticker);
+	p = realloc(p, large0);
+	assert_ptr_not_null(p, "Unexpected realloc() failure");
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during realloc()");
+	/* Deallocate. */
+	tick0 = ticker_read(decay_ticker);
+	realloc(p, 0);
+	tick1 = ticker_read(decay_ticker);
+	assert_u32_ne(tick1, tick0, "Expected ticker to tick during realloc()");
+
+	/*
+	 * Test the *allocx() APIs using large and small size classes, with
+	 * tcache explicitly disabled.
+	 */
+	{
+		unsigned i;
+		size_t allocx_sizes[2];
+		allocx_sizes[0] = large0;
+		allocx_sizes[1] = 1;
+
+		for (i = 0; i < sizeof(allocx_sizes) / sizeof(size_t); i++) {
+			sz = allocx_sizes[i];
+
+			/* mallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			p = mallocx(sz, MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p, "Unexpected mallocx() failure");
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during mallocx() (sz=%zu)",
+			    sz);
+			/* rallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			p = rallocx(p, sz, MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p, "Unexpected rallocx() failure");
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during rallocx() (sz=%zu)",
+			    sz);
+			/* xallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			xallocx(p, sz, 0, MALLOCX_TCACHE_NONE);
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during xallocx() (sz=%zu)",
+			    sz);
+			/* dallocx(). */
+			tick0 = ticker_read(decay_ticker);
+			dallocx(p, MALLOCX_TCACHE_NONE);
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during dallocx() (sz=%zu)",
+			    sz);
+			/* sdallocx(). */
+			p = mallocx(sz, MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p, "Unexpected mallocx() failure");
+			tick0 = ticker_read(decay_ticker);
+			sdallocx(p, sz, MALLOCX_TCACHE_NONE);
+			tick1 = ticker_read(decay_ticker);
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during sdallocx() "
+			    "(sz=%zu)", sz);
+		}
+	}
+
+	/*
+	 * Test tcache fill/flush interactions for large and small size classes,
+	 * using an explicit tcache.
+	 */
+	unsigned tcache_ind, i;
+	size_t tcache_sizes[2];
+	tcache_sizes[0] = large0;
+	tcache_sizes[1] = 1;
+
+	size_t tcache_max, sz_tcache_max;
+	sz_tcache_max = sizeof(tcache_max);
+	assert_d_eq(mallctl("arenas.tcache_max", (void *)&tcache_max,
+	    &sz_tcache_max, NULL, 0), 0, "Unexpected mallctl() failure");
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("tcache.create", (void *)&tcache_ind, &sz,
+	    NULL, 0), 0, "Unexpected mallctl failure");
+
+	for (i = 0; i < sizeof(tcache_sizes) / sizeof(size_t); i++) {
+		sz = tcache_sizes[i];
+
+		/* tcache fill. */
+		tick0 = ticker_read(decay_ticker);
+		p = mallocx(sz, MALLOCX_TCACHE(tcache_ind));
+		assert_ptr_not_null(p, "Unexpected mallocx() failure");
+		tick1 = ticker_read(decay_ticker);
+		assert_u32_ne(tick1, tick0,
+		    "Expected ticker to tick during tcache fill "
+		    "(sz=%zu)", sz);
+		/* tcache flush. */
+		dallocx(p, MALLOCX_TCACHE(tcache_ind));
+		tick0 = ticker_read(decay_ticker);
+		assert_d_eq(mallctl("tcache.flush", NULL, NULL,
+		    (void *)&tcache_ind, sizeof(unsigned)), 0,
+		    "Unexpected mallctl failure");
+		tick1 = ticker_read(decay_ticker);
+
+		/* Will only tick if it's in tcache. */
+		if (sz <= tcache_max) {
+			assert_u32_ne(tick1, tick0,
+			    "Expected ticker to tick during tcache "
+			    "flush (sz=%zu)", sz);
+		} else {
+			assert_u32_eq(tick1, tick0,
+			    "Unexpected ticker tick during tcache "
+			    "flush (sz=%zu)", sz);
+		}
+	}
+}
+TEST_END
+
+static void
+decay_ticker_helper(unsigned arena_ind, int flags, bool dirty, ssize_t dt,
+    uint64_t dirty_npurge0, uint64_t muzzy_npurge0, bool terminate_asap) {
+#define NINTERVALS 101
+	nstime_t time, update_interval, decay_ms, deadline;
+
+	nstime_init(&time, 0);
+	nstime_update(&time);
+
+	nstime_init2(&decay_ms, dt, 0);
+	nstime_copy(&deadline, &time);
+	nstime_add(&deadline, &decay_ms);
+
+	nstime_init2(&update_interval, dt, 0);
+	nstime_idivide(&update_interval, NINTERVALS);
+
+	/*
+	 * Keep q's slab from being deallocated during the looping below.  If a
+	 * cached slab were to repeatedly come and go during looping, it could
+	 * prevent the decay backlog ever becoming empty.
+	 */
+	void *p = do_mallocx(1, flags);
+	uint64_t dirty_npurge1, muzzy_npurge1;
+	do {
+		for (unsigned i = 0; i < DECAY_NTICKS_PER_UPDATE / 2;
+		    i++) {
+			void *q = do_mallocx(1, flags);
+			dallocx(q, flags);
+		}
+		dirty_npurge1 = get_arena_dirty_npurge(arena_ind);
+		muzzy_npurge1 = get_arena_muzzy_npurge(arena_ind);
+
+		nstime_add(&time_mock, &update_interval);
+		nstime_update(&time);
+	} while (nstime_compare(&time, &deadline) <= 0 && ((dirty_npurge1 ==
+	    dirty_npurge0 && muzzy_npurge1 == muzzy_npurge0) ||
+	    !terminate_asap));
+	dallocx(p, flags);
+
+	if (config_stats) {
+		assert_u64_gt(dirty_npurge1 + muzzy_npurge1, dirty_npurge0 +
+		    muzzy_npurge0, "Expected purging to occur");
+	}
+#undef NINTERVALS
+}
+
+TEST_BEGIN(test_decay_ticker) {
+	test_skip_if(check_background_thread_enabled());
+#define NPS 2048
+	ssize_t ddt = opt_dirty_decay_ms;
+	ssize_t mdt = opt_muzzy_decay_ms;
+	unsigned arena_ind = do_arena_create(ddt, mdt);
+	int flags = (MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE);
+	void *ps[NPS];
+	size_t large;
+
+	/*
+	 * Allocate a bunch of large objects, pause the clock, deallocate every
+	 * other object (to fragment virtual memory), restore the clock, then
+	 * [md]allocx() in a tight loop while advancing time rapidly to verify
+	 * the ticker triggers purging.
+	 */
+
+	size_t tcache_max;
+	size_t sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.tcache_max", (void *)&tcache_max, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+	large = nallocx(tcache_max + 1, flags);
+
+	do_purge(arena_ind);
+	uint64_t dirty_npurge0 = get_arena_dirty_npurge(arena_ind);
+	uint64_t muzzy_npurge0 = get_arena_muzzy_npurge(arena_ind);
+
+	for (unsigned i = 0; i < NPS; i++) {
+		ps[i] = do_mallocx(large, flags);
+	}
+
+	nupdates_mock = 0;
+	nstime_init(&time_mock, 0);
+	nstime_update(&time_mock);
+	monotonic_mock = true;
+
+	nstime_monotonic_orig = nstime_monotonic;
+	nstime_update_orig = nstime_update;
+	nstime_monotonic = nstime_monotonic_mock;
+	nstime_update = nstime_update_mock;
+
+	for (unsigned i = 0; i < NPS; i += 2) {
+		dallocx(ps[i], flags);
+		unsigned nupdates0 = nupdates_mock;
+		do_decay(arena_ind);
+		assert_u_gt(nupdates_mock, nupdates0,
+		    "Expected nstime_update() to be called");
+	}
+
+	decay_ticker_helper(arena_ind, flags, true, ddt, dirty_npurge0,
+	    muzzy_npurge0, true);
+	decay_ticker_helper(arena_ind, flags, false, ddt+mdt, dirty_npurge0,
+	    muzzy_npurge0, false);
+
+	do_arena_destroy(arena_ind);
+
+	nstime_monotonic = nstime_monotonic_orig;
+	nstime_update = nstime_update_orig;
+#undef NPS
+}
+TEST_END
+
+TEST_BEGIN(test_decay_nonmonotonic) {
+	test_skip_if(check_background_thread_enabled());
+#define NPS (SMOOTHSTEP_NSTEPS + 1)
+	int flags = (MALLOCX_ARENA(0) | MALLOCX_TCACHE_NONE);
+	void *ps[NPS];
+	uint64_t npurge0 = 0;
+	uint64_t npurge1 = 0;
+	size_t sz, large0;
+	unsigned i, nupdates0;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&large0, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure");
+	do_epoch();
+	sz = sizeof(uint64_t);
+	npurge0 = get_arena_npurge(0);
+
+	nupdates_mock = 0;
+	nstime_init(&time_mock, 0);
+	nstime_update(&time_mock);
+	monotonic_mock = false;
+
+	nstime_monotonic_orig = nstime_monotonic;
+	nstime_update_orig = nstime_update;
+	nstime_monotonic = nstime_monotonic_mock;
+	nstime_update = nstime_update_mock;
+
+	for (i = 0; i < NPS; i++) {
+		ps[i] = mallocx(large0, flags);
+		assert_ptr_not_null(ps[i], "Unexpected mallocx() failure");
+	}
+
+	for (i = 0; i < NPS; i++) {
+		dallocx(ps[i], flags);
+		nupdates0 = nupdates_mock;
+		assert_d_eq(mallctl("arena.0.decay", NULL, NULL, NULL, 0), 0,
+		    "Unexpected arena.0.decay failure");
+		assert_u_gt(nupdates_mock, nupdates0,
+		    "Expected nstime_update() to be called");
+	}
+
+	do_epoch();
+	sz = sizeof(uint64_t);
+	npurge1 = get_arena_npurge(0);
+
+	if (config_stats) {
+		assert_u64_eq(npurge0, npurge1, "Unexpected purging occurred");
+	}
+
+	nstime_monotonic = nstime_monotonic_orig;
+	nstime_update = nstime_update_orig;
+#undef NPS
+}
+TEST_END
+
+TEST_BEGIN(test_decay_now) {
+	test_skip_if(check_background_thread_enabled());
+
+	unsigned arena_ind = do_arena_create(0, 0);
+	assert_zu_eq(get_arena_pdirty(arena_ind), 0, "Unexpected dirty pages");
+	assert_zu_eq(get_arena_pmuzzy(arena_ind), 0, "Unexpected muzzy pages");
+	size_t sizes[] = {16, PAGE<<2, HUGEPAGE<<2};
+	/* Verify that dirty/muzzy pages never linger after deallocation. */
+	for (unsigned i = 0; i < sizeof(sizes)/sizeof(size_t); i++) {
+		size_t size = sizes[i];
+		generate_dirty(arena_ind, size);
+		assert_zu_eq(get_arena_pdirty(arena_ind), 0,
+		    "Unexpected dirty pages");
+		assert_zu_eq(get_arena_pmuzzy(arena_ind), 0,
+		    "Unexpected muzzy pages");
+	}
+	do_arena_destroy(arena_ind);
+}
+TEST_END
+
+TEST_BEGIN(test_decay_never) {
+	test_skip_if(check_background_thread_enabled() || !config_stats);
+
+	unsigned arena_ind = do_arena_create(-1, -1);
+	int flags = MALLOCX_ARENA(arena_ind) | MALLOCX_TCACHE_NONE;
+	assert_zu_eq(get_arena_pdirty(arena_ind), 0, "Unexpected dirty pages");
+	assert_zu_eq(get_arena_pmuzzy(arena_ind), 0, "Unexpected muzzy pages");
+	size_t sizes[] = {16, PAGE<<2, HUGEPAGE<<2};
+	void *ptrs[sizeof(sizes)/sizeof(size_t)];
+	for (unsigned i = 0; i < sizeof(sizes)/sizeof(size_t); i++) {
+		ptrs[i] = do_mallocx(sizes[i], flags);
+	}
+	/* Verify that each deallocation generates additional dirty pages. */
+	size_t pdirty_prev = get_arena_pdirty(arena_ind);
+	size_t pmuzzy_prev = get_arena_pmuzzy(arena_ind);
+	assert_zu_eq(pdirty_prev, 0, "Unexpected dirty pages");
+	assert_zu_eq(pmuzzy_prev, 0, "Unexpected muzzy pages");
+	for (unsigned i = 0; i < sizeof(sizes)/sizeof(size_t); i++) {
+		dallocx(ptrs[i], flags);
+		size_t pdirty = get_arena_pdirty(arena_ind);
+		size_t pmuzzy = get_arena_pmuzzy(arena_ind);
+		assert_zu_gt(pdirty + (size_t)get_arena_dirty_purged(arena_ind),
+		    pdirty_prev, "Expected dirty pages to increase.");
+		assert_zu_eq(pmuzzy, 0, "Unexpected muzzy pages");
+		pdirty_prev = pdirty;
+	}
+	do_arena_destroy(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_decay_ticks,
+	    test_decay_ticker,
+	    test_decay_nonmonotonic,
+	    test_decay_now,
+	    test_decay_never);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/decay.sh b/memkind-1.10.0/jemalloc/test/unit/decay.sh
new file mode 100644
index 0000000..45aeccf
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/decay.sh
@@ -0,0 +1,3 @@
+#!/bin/sh
+
+export MALLOC_CONF="dirty_decay_ms:1000,muzzy_decay_ms:1000,lg_tcache_max:0"
diff --git a/memkind-1.10.0/jemalloc/test/unit/div.c b/memkind-1.10.0/jemalloc/test/unit/div.c
new file mode 100644
index 0000000..b47f10b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/div.c
@@ -0,0 +1,29 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/div.h"
+
+TEST_BEGIN(test_div_exhaustive) {
+	for (size_t divisor = 2; divisor < 1000 * 1000; ++divisor) {
+		div_info_t div_info;
+		div_init(&div_info, divisor);
+		size_t max = 1000 * divisor;
+		if (max < 1000 * 1000) {
+			max = 1000 * 1000;
+		}
+		for (size_t dividend = 0; dividend < 1000 * divisor;
+		    dividend += divisor) {
+			size_t quotient = div_compute(
+			    &div_info, dividend);
+			assert_zu_eq(dividend, quotient * divisor,
+			    "With divisor = %zu, dividend = %zu, "
+			    "got quotient %zu", divisor, dividend, quotient);
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_div_exhaustive);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/emitter.c b/memkind-1.10.0/jemalloc/test/unit/emitter.c
new file mode 100644
index 0000000..b4a693f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/emitter.c
@@ -0,0 +1,469 @@
+#include "test/jemalloc_test.h"
+#include "jemalloc/internal/emitter.h"
+
+/*
+ * This is so useful for debugging and feature work, we'll leave printing
+ * functionality committed but disabled by default.
+ */
+/* Print the text as it will appear. */
+static bool print_raw = false;
+/* Print the text escaped, so it can be copied back into the test case. */
+static bool print_escaped = false;
+
+typedef struct buf_descriptor_s buf_descriptor_t;
+struct buf_descriptor_s {
+	char *buf;
+	size_t len;
+	bool mid_quote;
+};
+
+/*
+ * Forwards all writes to the passed-in buf_v (which should be cast from a
+ * buf_descriptor_t *).
+ */
+static void
+forwarding_cb(void *buf_descriptor_v, const char *str) {
+	buf_descriptor_t *buf_descriptor = (buf_descriptor_t *)buf_descriptor_v;
+
+	if (print_raw) {
+		malloc_printf("%s", str);
+	}
+	if (print_escaped) {
+		const char *it = str;
+		while (*it != '\0') {
+			if (!buf_descriptor->mid_quote) {
+				malloc_printf("\"");
+				buf_descriptor->mid_quote = true;
+			}
+			switch (*it) {
+			case '\\':
+				malloc_printf("\\");
+				break;
+			case '\"':
+				malloc_printf("\\\"");
+				break;
+			case '\t':
+				malloc_printf("\\t");
+				break;
+			case '\n':
+				malloc_printf("\\n\"\n");
+				buf_descriptor->mid_quote = false;
+				break;
+			default:
+				malloc_printf("%c", *it);
+			}
+			it++;
+		}
+	}
+
+	size_t written = malloc_snprintf(buf_descriptor->buf,
+	    buf_descriptor->len, "%s", str);
+	assert_zu_eq(written, strlen(str), "Buffer overflow!");
+	buf_descriptor->buf += written;
+	buf_descriptor->len -= written;
+	assert_zu_gt(buf_descriptor->len, 0, "Buffer out of space!");
+}
+
+static void
+assert_emit_output(void (*emit_fn)(emitter_t *),
+    const char *expected_json_output, const char *expected_table_output) {
+	emitter_t emitter;
+	char buf[MALLOC_PRINTF_BUFSIZE];
+	buf_descriptor_t buf_descriptor;
+
+	buf_descriptor.buf = buf;
+	buf_descriptor.len = MALLOC_PRINTF_BUFSIZE;
+	buf_descriptor.mid_quote = false;
+
+	emitter_init(&emitter, emitter_output_json, &forwarding_cb,
+	    &buf_descriptor);
+	(*emit_fn)(&emitter);
+	assert_str_eq(expected_json_output, buf, "json output failure");
+
+	buf_descriptor.buf = buf;
+	buf_descriptor.len = MALLOC_PRINTF_BUFSIZE;
+	buf_descriptor.mid_quote = false;
+
+	emitter_init(&emitter, emitter_output_table, &forwarding_cb,
+	    &buf_descriptor);
+	(*emit_fn)(&emitter);
+	assert_str_eq(expected_table_output, buf, "table output failure");
+}
+
+static void
+emit_dict(emitter_t *emitter) {
+	bool b_false = false;
+	bool b_true = true;
+	int i_123 = 123;
+	const char *str = "a string";
+
+	emitter_begin(emitter);
+	emitter_dict_begin(emitter, "foo", "This is the foo table:");
+	emitter_kv(emitter, "abc", "ABC", emitter_type_bool, &b_false);
+	emitter_kv(emitter, "def", "DEF", emitter_type_bool, &b_true);
+	emitter_kv_note(emitter, "ghi", "GHI", emitter_type_int, &i_123,
+	    "note_key1", emitter_type_string, &str);
+	emitter_kv_note(emitter, "jkl", "JKL", emitter_type_string, &str,
+	    "note_key2", emitter_type_bool, &b_false);
+	emitter_dict_end(emitter);
+	emitter_end(emitter);
+}
+static const char *dict_json =
+"{\n"
+"\t\"foo\": {\n"
+"\t\t\"abc\": false,\n"
+"\t\t\"def\": true,\n"
+"\t\t\"ghi\": 123,\n"
+"\t\t\"jkl\": \"a string\"\n"
+"\t}\n"
+"}\n";
+static const char *dict_table =
+"This is the foo table:\n"
+"  ABC: false\n"
+"  DEF: true\n"
+"  GHI: 123 (note_key1: \"a string\")\n"
+"  JKL: \"a string\" (note_key2: false)\n";
+
+TEST_BEGIN(test_dict) {
+	assert_emit_output(&emit_dict, dict_json, dict_table);
+}
+TEST_END
+
+static void
+emit_table_printf(emitter_t *emitter) {
+	emitter_begin(emitter);
+	emitter_table_printf(emitter, "Table note 1\n");
+	emitter_table_printf(emitter, "Table note 2 %s\n",
+	    "with format string");
+	emitter_end(emitter);
+}
+
+static const char *table_printf_json =
+"{\n"
+"}\n";
+
+static const char *table_printf_table =
+"Table note 1\n"
+"Table note 2 with format string\n";
+
+TEST_BEGIN(test_table_printf) {
+	assert_emit_output(&emit_table_printf, table_printf_json,
+	    table_printf_table);
+}
+TEST_END
+
+static void emit_nested_dict(emitter_t *emitter) {
+	int val = 123;
+	emitter_begin(emitter);
+	emitter_dict_begin(emitter, "json1", "Dict 1");
+	emitter_dict_begin(emitter, "json2", "Dict 2");
+	emitter_kv(emitter, "primitive", "A primitive", emitter_type_int, &val);
+	emitter_dict_end(emitter); /* Close 2 */
+	emitter_dict_begin(emitter, "json3", "Dict 3");
+	emitter_dict_end(emitter); /* Close 3 */
+	emitter_dict_end(emitter); /* Close 1 */
+	emitter_dict_begin(emitter, "json4", "Dict 4");
+	emitter_kv(emitter, "primitive", "Another primitive",
+	    emitter_type_int, &val);
+	emitter_dict_end(emitter); /* Close 4 */
+	emitter_end(emitter);
+}
+
+static const char *nested_object_json =
+"{\n"
+"\t\"json1\": {\n"
+"\t\t\"json2\": {\n"
+"\t\t\t\"primitive\": 123\n"
+"\t\t},\n"
+"\t\t\"json3\": {\n"
+"\t\t}\n"
+"\t},\n"
+"\t\"json4\": {\n"
+"\t\t\"primitive\": 123\n"
+"\t}\n"
+"}\n";
+
+static const char *nested_object_table =
+"Dict 1\n"
+"  Dict 2\n"
+"    A primitive: 123\n"
+"  Dict 3\n"
+"Dict 4\n"
+"  Another primitive: 123\n";
+
+TEST_BEGIN(test_nested_dict) {
+	assert_emit_output(&emit_nested_dict, nested_object_json,
+	    nested_object_table);
+}
+TEST_END
+
+static void
+emit_types(emitter_t *emitter) {
+	bool b = false;
+	int i = -123;
+	unsigned u = 123;
+	ssize_t zd = -456;
+	size_t zu = 456;
+	const char *str = "string";
+	uint32_t u32 = 789;
+	uint64_t u64 = 10000000000ULL;
+
+	emitter_begin(emitter);
+	emitter_kv(emitter, "k1", "K1", emitter_type_bool, &b);
+	emitter_kv(emitter, "k2", "K2", emitter_type_int, &i);
+	emitter_kv(emitter, "k3", "K3", emitter_type_unsigned, &u);
+	emitter_kv(emitter, "k4", "K4", emitter_type_ssize, &zd);
+	emitter_kv(emitter, "k5", "K5", emitter_type_size, &zu);
+	emitter_kv(emitter, "k6", "K6", emitter_type_string, &str);
+	emitter_kv(emitter, "k7", "K7", emitter_type_uint32, &u32);
+	emitter_kv(emitter, "k8", "K8", emitter_type_uint64, &u64);
+	/*
+	 * We don't test the title type, since it's only used for tables.  It's
+	 * tested in the emitter_table_row tests.
+	 */
+	emitter_end(emitter);
+}
+
+static const char *types_json =
+"{\n"
+"\t\"k1\": false,\n"
+"\t\"k2\": -123,\n"
+"\t\"k3\": 123,\n"
+"\t\"k4\": -456,\n"
+"\t\"k5\": 456,\n"
+"\t\"k6\": \"string\",\n"
+"\t\"k7\": 789,\n"
+"\t\"k8\": 10000000000\n"
+"}\n";
+
+static const char *types_table =
+"K1: false\n"
+"K2: -123\n"
+"K3: 123\n"
+"K4: -456\n"
+"K5: 456\n"
+"K6: \"string\"\n"
+"K7: 789\n"
+"K8: 10000000000\n";
+
+TEST_BEGIN(test_types) {
+	assert_emit_output(&emit_types, types_json, types_table);
+}
+TEST_END
+
+static void
+emit_modal(emitter_t *emitter) {
+	int val = 123;
+	emitter_begin(emitter);
+	emitter_dict_begin(emitter, "j0", "T0");
+	emitter_json_key(emitter, "j1");
+	emitter_json_object_begin(emitter);
+	emitter_kv(emitter, "i1", "I1", emitter_type_int, &val);
+	emitter_json_kv(emitter, "i2", emitter_type_int, &val);
+	emitter_table_kv(emitter, "I3", emitter_type_int, &val);
+	emitter_table_dict_begin(emitter, "T1");
+	emitter_kv(emitter, "i4", "I4", emitter_type_int, &val);
+	emitter_json_object_end(emitter); /* Close j1 */
+	emitter_kv(emitter, "i5", "I5", emitter_type_int, &val);
+	emitter_table_dict_end(emitter); /* Close T1 */
+	emitter_kv(emitter, "i6", "I6", emitter_type_int, &val);
+	emitter_dict_end(emitter); /* Close j0 / T0 */
+	emitter_end(emitter);
+}
+
+const char *modal_json =
+"{\n"
+"\t\"j0\": {\n"
+"\t\t\"j1\": {\n"
+"\t\t\t\"i1\": 123,\n"
+"\t\t\t\"i2\": 123,\n"
+"\t\t\t\"i4\": 123\n"
+"\t\t},\n"
+"\t\t\"i5\": 123,\n"
+"\t\t\"i6\": 123\n"
+"\t}\n"
+"}\n";
+
+const char *modal_table =
+"T0\n"
+"  I1: 123\n"
+"  I3: 123\n"
+"  T1\n"
+"    I4: 123\n"
+"    I5: 123\n"
+"  I6: 123\n";
+
+TEST_BEGIN(test_modal) {
+	assert_emit_output(&emit_modal, modal_json, modal_table);
+}
+TEST_END
+
+static void
+emit_json_arr(emitter_t *emitter) {
+	int ival = 123;
+
+	emitter_begin(emitter);
+	emitter_json_key(emitter, "dict");
+	emitter_json_object_begin(emitter);
+	emitter_json_key(emitter, "arr");
+	emitter_json_array_begin(emitter);
+	emitter_json_object_begin(emitter);
+	emitter_json_kv(emitter, "foo", emitter_type_int, &ival);
+	emitter_json_object_end(emitter); /* Close arr[0] */
+	/* arr[1] and arr[2] are primitives. */
+	emitter_json_value(emitter, emitter_type_int, &ival);
+	emitter_json_value(emitter, emitter_type_int, &ival);
+	emitter_json_object_begin(emitter);
+	emitter_json_kv(emitter, "bar", emitter_type_int, &ival);
+	emitter_json_kv(emitter, "baz", emitter_type_int, &ival);
+	emitter_json_object_end(emitter); /* Close arr[3]. */
+	emitter_json_array_end(emitter); /* Close arr. */
+	emitter_json_object_end(emitter); /* Close dict. */
+	emitter_end(emitter);
+}
+
+static const char *json_array_json =
+"{\n"
+"\t\"dict\": {\n"
+"\t\t\"arr\": [\n"
+"\t\t\t{\n"
+"\t\t\t\t\"foo\": 123\n"
+"\t\t\t},\n"
+"\t\t\t123,\n"
+"\t\t\t123,\n"
+"\t\t\t{\n"
+"\t\t\t\t\"bar\": 123,\n"
+"\t\t\t\t\"baz\": 123\n"
+"\t\t\t}\n"
+"\t\t]\n"
+"\t}\n"
+"}\n";
+
+static const char *json_array_table = "";
+
+TEST_BEGIN(test_json_arr) {
+	assert_emit_output(&emit_json_arr, json_array_json, json_array_table);
+}
+TEST_END
+
+static void
+emit_json_nested_array(emitter_t *emitter) {
+	int ival = 123;
+	char *sval = "foo";
+	emitter_begin(emitter);
+	emitter_json_array_begin(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_value(emitter, emitter_type_string, &sval);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_value(emitter, emitter_type_string, &sval);
+		emitter_json_array_end(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_array_end(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_value(emitter, emitter_type_string, &sval);
+		emitter_json_value(emitter, emitter_type_int, &ival);
+		emitter_json_array_end(emitter);
+		emitter_json_array_begin(emitter);
+		emitter_json_array_end(emitter);
+	emitter_json_array_end(emitter);
+	emitter_end(emitter);
+}
+
+static const char *json_nested_array_json =
+"{\n"
+"\t[\n"
+"\t\t[\n"
+"\t\t\t123,\n"
+"\t\t\t\"foo\",\n"
+"\t\t\t123,\n"
+"\t\t\t\"foo\"\n"
+"\t\t],\n"
+"\t\t[\n"
+"\t\t\t123\n"
+"\t\t],\n"
+"\t\t[\n"
+"\t\t\t\"foo\",\n"
+"\t\t\t123\n"
+"\t\t],\n"
+"\t\t[\n"
+"\t\t]\n"
+"\t]\n"
+"}\n";
+
+TEST_BEGIN(test_json_nested_arr) {
+	assert_emit_output(&emit_json_nested_array, json_nested_array_json,
+	    json_array_table);
+}
+TEST_END
+
+static void
+emit_table_row(emitter_t *emitter) {
+	emitter_begin(emitter);
+	emitter_row_t row;
+	emitter_col_t abc = {emitter_justify_left, 10, emitter_type_title, {0}, {0, 0}};
+	abc.str_val = "ABC title";
+	emitter_col_t def = {emitter_justify_right, 15, emitter_type_title, {0}, {0, 0}};
+	def.str_val = "DEF title";
+	emitter_col_t ghi = {emitter_justify_right, 5, emitter_type_title, {0}, {0, 0}};
+	ghi.str_val = "GHI";
+
+	emitter_row_init(&row);
+	emitter_col_init(&abc, &row);
+	emitter_col_init(&def, &row);
+	emitter_col_init(&ghi, &row);
+
+	emitter_table_row(emitter, &row);
+
+	abc.type = emitter_type_int;
+	def.type = emitter_type_bool;
+	ghi.type = emitter_type_int;
+
+	abc.int_val = 123;
+	def.bool_val = true;
+	ghi.int_val = 456;
+	emitter_table_row(emitter, &row);
+
+	abc.int_val = 789;
+	def.bool_val = false;
+	ghi.int_val = 1011;
+	emitter_table_row(emitter, &row);
+
+	abc.type = emitter_type_string;
+	abc.str_val = "a string";
+	def.bool_val = false;
+	ghi.type = emitter_type_title;
+	ghi.str_val = "ghi";
+	emitter_table_row(emitter, &row);
+
+	emitter_end(emitter);
+}
+
+static const char *table_row_json =
+"{\n"
+"}\n";
+
+static const char *table_row_table =
+"ABC title       DEF title  GHI\n"
+"123                  true  456\n"
+"789                 false 1011\n"
+"\"a string\"          false  ghi\n";
+
+TEST_BEGIN(test_table_row) {
+	assert_emit_output(&emit_table_row, table_row_json, table_row_table);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_dict,
+	    test_table_printf,
+	    test_nested_dict,
+	    test_types,
+	    test_modal,
+	    test_json_arr,
+	    test_json_nested_arr,
+	    test_table_row);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/extent_quantize.c b/memkind-1.10.0/jemalloc/test/unit/extent_quantize.c
new file mode 100644
index 0000000..0ca7a75
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/extent_quantize.c
@@ -0,0 +1,141 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_small_extent_size) {
+	unsigned nbins, i;
+	size_t sz, extent_size;
+	size_t mib[4];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+
+	/*
+	 * Iterate over all small size classes, get their extent sizes, and
+	 * verify that the quantized size is the same as the extent size.
+	 */
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &sz, NULL, 0), 0,
+	    "Unexpected mallctl failure");
+
+	assert_d_eq(mallctlnametomib("arenas.bin.0.slab_size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib failure");
+	for (i = 0; i < nbins; i++) {
+		mib[2] = i;
+		sz = sizeof(size_t);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&extent_size, &sz,
+		    NULL, 0), 0, "Unexpected mallctlbymib failure");
+		assert_zu_eq(extent_size,
+		    extent_size_quantize_floor(extent_size),
+		    "Small extent quantization should be a no-op "
+		    "(extent_size=%zu)", extent_size);
+		assert_zu_eq(extent_size,
+		    extent_size_quantize_ceil(extent_size),
+		    "Small extent quantization should be a no-op "
+		    "(extent_size=%zu)", extent_size);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_large_extent_size) {
+	bool cache_oblivious;
+	unsigned nlextents, i;
+	size_t sz, extent_size_prev, ceil_prev;
+	size_t mib[4];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+
+	/*
+	 * Iterate over all large size classes, get their extent sizes, and
+	 * verify that the quantized size is the same as the extent size.
+	 */
+
+	sz = sizeof(bool);
+	assert_d_eq(mallctl("config.cache_oblivious", (void *)&cache_oblivious,
+	    &sz, NULL, 0), 0, "Unexpected mallctl failure");
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nlextents", (void *)&nlextents, &sz, NULL,
+	    0), 0, "Unexpected mallctl failure");
+
+	assert_d_eq(mallctlnametomib("arenas.lextent.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib failure");
+	for (i = 0; i < nlextents; i++) {
+		size_t lextent_size, extent_size, floor, ceil;
+
+		mib[2] = i;
+		sz = sizeof(size_t);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&lextent_size,
+		    &sz, NULL, 0), 0, "Unexpected mallctlbymib failure");
+		extent_size = cache_oblivious ? lextent_size + PAGE :
+		    lextent_size;
+		floor = extent_size_quantize_floor(extent_size);
+		ceil = extent_size_quantize_ceil(extent_size);
+
+		assert_zu_eq(extent_size, floor,
+		    "Extent quantization should be a no-op for precise size "
+		    "(lextent_size=%zu, extent_size=%zu)", lextent_size,
+		    extent_size);
+		assert_zu_eq(extent_size, ceil,
+		    "Extent quantization should be a no-op for precise size "
+		    "(lextent_size=%zu, extent_size=%zu)", lextent_size,
+		    extent_size);
+
+		if (i > 0) {
+			assert_zu_eq(extent_size_prev,
+			    extent_size_quantize_floor(extent_size - PAGE),
+			    "Floor should be a precise size");
+			if (extent_size_prev < ceil_prev) {
+				assert_zu_eq(ceil_prev, extent_size,
+				    "Ceiling should be a precise size "
+				    "(extent_size_prev=%zu, ceil_prev=%zu, "
+				    "extent_size=%zu)", extent_size_prev,
+				    ceil_prev, extent_size);
+			}
+		}
+		if (i + 1 < nlextents) {
+			extent_size_prev = floor;
+			ceil_prev = extent_size_quantize_ceil(extent_size +
+			    PAGE);
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_monotonic) {
+#define SZ_MAX	ZU(4 * 1024 * 1024)
+	unsigned i;
+	size_t floor_prev, ceil_prev;
+
+	floor_prev = 0;
+	ceil_prev = 0;
+	for (i = 1; i <= SZ_MAX >> LG_PAGE; i++) {
+		size_t extent_size, floor, ceil;
+
+		extent_size = i << LG_PAGE;
+		floor = extent_size_quantize_floor(extent_size);
+		ceil = extent_size_quantize_ceil(extent_size);
+
+		assert_zu_le(floor, extent_size,
+		    "Floor should be <= (floor=%zu, extent_size=%zu, ceil=%zu)",
+		    floor, extent_size, ceil);
+		assert_zu_ge(ceil, extent_size,
+		    "Ceiling should be >= (floor=%zu, extent_size=%zu, "
+		    "ceil=%zu)", floor, extent_size, ceil);
+
+		assert_zu_le(floor_prev, floor, "Floor should be monotonic "
+		    "(floor_prev=%zu, floor=%zu, extent_size=%zu, ceil=%zu)",
+		    floor_prev, floor, extent_size, ceil);
+		assert_zu_le(ceil_prev, ceil, "Ceiling should be monotonic "
+		    "(floor=%zu, extent_size=%zu, ceil_prev=%zu, ceil=%zu)",
+		    floor, extent_size, ceil_prev, ceil);
+
+		floor_prev = floor;
+		ceil_prev = ceil;
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_small_extent_size,
+	    test_large_extent_size,
+	    test_monotonic);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/extent_util.c b/memkind-1.10.0/jemalloc/test/unit/extent_util.c
new file mode 100644
index 0000000..4de0b04
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/extent_util.c
@@ -0,0 +1,267 @@
+#include "test/jemalloc_test.h"
+
+#define TEST_UTIL_EINVAL(node, a, b, c, d, why_inval) do {		\
+	assert_d_eq(mallctl("experimental.utilization." node,		\
+	    a, b, c, d), EINVAL, "Should fail when " why_inval);	\
+	assert_zu_eq(out_sz, out_sz_ref,				\
+	    "Output size touched when given invalid arguments");	\
+	assert_d_eq(memcmp(out, out_ref, out_sz_ref), 0,		\
+	    "Output content touched when given invalid arguments");	\
+} while (0)
+
+#define TEST_UTIL_QUERY_EINVAL(a, b, c, d, why_inval)			\
+	TEST_UTIL_EINVAL("query", a, b, c, d, why_inval)
+#define TEST_UTIL_BATCH_EINVAL(a, b, c, d, why_inval)			\
+	TEST_UTIL_EINVAL("batch_query", a, b, c, d, why_inval)
+
+#define TEST_UTIL_VALID(node) do {					\
+        assert_d_eq(mallctl("experimental.utilization." node,		\
+	    out, &out_sz, in, in_sz), 0,				\
+	    "Should return 0 on correct arguments");			\
+        assert_zu_eq(out_sz, out_sz_ref, "incorrect output size");	\
+	assert_d_ne(memcmp(out, out_ref, out_sz_ref), 0,		\
+	    "Output content should be changed");			\
+} while (0)
+
+#define TEST_UTIL_BATCH_VALID TEST_UTIL_VALID("batch_query")
+
+#define TEST_MAX_SIZE (1 << 20)
+
+TEST_BEGIN(test_query) {
+	size_t sz;
+	/*
+	 * Select some sizes that can span both small and large sizes, and are
+	 * numerically unrelated to any size boundaries.
+	 */
+	for (sz = 7; sz <= TEST_MAX_SIZE && sz <= SC_LARGE_MAXCLASS;
+	    sz += (sz <= SC_SMALL_MAXCLASS ? 1009 : 99989)) {
+		void *p = mallocx(sz, 0);
+		void **in = &p;
+		size_t in_sz = sizeof(const void *);
+		size_t out_sz = sizeof(void *) + sizeof(size_t) * 5;
+		void *out = mallocx(out_sz, 0);
+		void *out_ref = mallocx(out_sz, 0);
+		size_t out_sz_ref = out_sz;
+
+		assert_ptr_not_null(p,
+		    "test pointer allocation failed");
+		assert_ptr_not_null(out,
+		    "test output allocation failed");
+		assert_ptr_not_null(out_ref,
+		    "test reference output allocation failed");
+
+#define SLABCUR_READ(out) (*(void **)out)
+#define COUNTS(out) ((size_t *)((void **)out + 1))
+#define NFREE_READ(out) COUNTS(out)[0]
+#define NREGS_READ(out) COUNTS(out)[1]
+#define SIZE_READ(out) COUNTS(out)[2]
+#define BIN_NFREE_READ(out) COUNTS(out)[3]
+#define BIN_NREGS_READ(out) COUNTS(out)[4]
+
+		SLABCUR_READ(out) = NULL;
+		NFREE_READ(out) = NREGS_READ(out) = SIZE_READ(out) = -1;
+		BIN_NFREE_READ(out) = BIN_NREGS_READ(out) = -1;
+		memcpy(out_ref, out, out_sz);
+
+		/* Test invalid argument(s) errors */
+		TEST_UTIL_QUERY_EINVAL(NULL, &out_sz, in, in_sz,
+		    "old is NULL");
+		TEST_UTIL_QUERY_EINVAL(out, NULL, in, in_sz,
+		    "oldlenp is NULL");
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, NULL, in_sz,
+		    "newp is NULL");
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, in, 0,
+		    "newlen is zero");
+		in_sz -= 1;
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, in, in_sz,
+		    "invalid newlen");
+		in_sz += 1;
+		out_sz_ref = out_sz -= 2 * sizeof(size_t);
+		TEST_UTIL_QUERY_EINVAL(out, &out_sz, in, in_sz,
+		    "invalid *oldlenp");
+		out_sz_ref = out_sz += 2 * sizeof(size_t);
+
+		/* Examine output for valid call */
+		TEST_UTIL_VALID("query");
+		assert_zu_le(sz, SIZE_READ(out),
+		    "Extent size should be at least allocation size");
+		assert_zu_eq(SIZE_READ(out) & (PAGE - 1), 0,
+		    "Extent size should be a multiple of page size");
+		if (sz <= SC_SMALL_MAXCLASS) {
+			assert_zu_le(NFREE_READ(out), NREGS_READ(out),
+			    "Extent free count exceeded region count");
+			assert_zu_le(NREGS_READ(out), SIZE_READ(out),
+			    "Extent region count exceeded size");
+			assert_zu_ne(NREGS_READ(out), 0,
+			    "Extent region count must be positive");
+			assert_true(NFREE_READ(out) == 0 || (SLABCUR_READ(out)
+			    != NULL && SLABCUR_READ(out) <= p),
+			    "Allocation should follow first fit principle");
+			if (config_stats) {
+				assert_zu_le(BIN_NFREE_READ(out),
+				    BIN_NREGS_READ(out),
+				    "Bin free count exceeded region count");
+				assert_zu_ne(BIN_NREGS_READ(out), 0,
+				    "Bin region count must be positive");
+				assert_zu_le(NFREE_READ(out),
+				    BIN_NFREE_READ(out),
+				    "Extent free count exceeded bin free count");
+				assert_zu_le(NREGS_READ(out),
+				    BIN_NREGS_READ(out),
+				    "Extent region count exceeded "
+				    "bin region count");
+				assert_zu_eq(BIN_NREGS_READ(out)
+				    % NREGS_READ(out), 0,
+				    "Bin region count isn't a multiple of "
+				    "extent region count");
+				assert_zu_le(
+				    BIN_NFREE_READ(out) - NFREE_READ(out),
+				    BIN_NREGS_READ(out) - NREGS_READ(out),
+				    "Free count in other extents in the bin "
+				    "exceeded region count in other extents "
+				    "in the bin");
+				assert_zu_le(NREGS_READ(out) - NFREE_READ(out),
+				    BIN_NREGS_READ(out) - BIN_NFREE_READ(out),
+				    "Extent utilized count exceeded "
+				    "bin utilized count");
+			}
+		} else {
+			assert_zu_eq(NFREE_READ(out), 0,
+			    "Extent free count should be zero");
+			assert_zu_eq(NREGS_READ(out), 1,
+			    "Extent region count should be one");
+			assert_ptr_null(SLABCUR_READ(out),
+			    "Current slab must be null for large size classes");
+			if (config_stats) {
+				assert_zu_eq(BIN_NFREE_READ(out), 0,
+				    "Bin free count must be zero for "
+				    "large sizes");
+				assert_zu_eq(BIN_NREGS_READ(out), 0,
+				    "Bin region count must be zero for "
+				    "large sizes");
+			}
+		}
+
+#undef BIN_NREGS_READ
+#undef BIN_NFREE_READ
+#undef SIZE_READ
+#undef NREGS_READ
+#undef NFREE_READ
+#undef COUNTS
+#undef SLABCUR_READ
+
+		free(out_ref);
+		free(out);
+		free(p);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_batch) {
+	size_t sz;
+	/*
+	 * Select some sizes that can span both small and large sizes, and are
+	 * numerically unrelated to any size boundaries.
+	 */
+	for (sz = 17; sz <= TEST_MAX_SIZE && sz <= SC_LARGE_MAXCLASS;
+	    sz += (sz <= SC_SMALL_MAXCLASS ? 1019 : 99991)) {
+		void *p = mallocx(sz, 0);
+		void *q = mallocx(sz, 0);
+		void *in[] = {p, q};
+		size_t in_sz = sizeof(const void *) * 2;
+		size_t out[] = {-1, -1, -1, -1, -1, -1};
+		size_t out_sz = sizeof(size_t) * 6;
+		size_t out_ref[] = {-1, -1, -1, -1, -1, -1};
+		size_t out_sz_ref = out_sz;
+
+		assert_ptr_not_null(p, "test pointer allocation failed");
+		assert_ptr_not_null(q, "test pointer allocation failed");
+
+		/* Test invalid argument(s) errors */
+		TEST_UTIL_BATCH_EINVAL(NULL, &out_sz, in, in_sz,
+		    "old is NULL");
+		TEST_UTIL_BATCH_EINVAL(out, NULL, in, in_sz,
+		    "oldlenp is NULL");
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, NULL, in_sz,
+		    "newp is NULL");
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, 0,
+		    "newlen is zero");
+		in_sz -= 1;
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, in_sz,
+		    "newlen is not an exact multiple");
+		in_sz += 1;
+		out_sz_ref = out_sz -= 2 * sizeof(size_t);
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, in_sz,
+		    "*oldlenp is not an exact multiple");
+		out_sz_ref = out_sz += 2 * sizeof(size_t);
+		in_sz -= sizeof(const void *);
+		TEST_UTIL_BATCH_EINVAL(out, &out_sz, in, in_sz,
+		    "*oldlenp and newlen do not match");
+		in_sz += sizeof(const void *);
+
+	/* Examine output for valid calls */
+#define TEST_EQUAL_REF(i, message) \
+	assert_d_eq(memcmp(out + (i) * 3, out_ref + (i) * 3, 3), 0, message)
+
+#define NFREE_READ(out, i) out[(i) * 3]
+#define NREGS_READ(out, i) out[(i) * 3 + 1]
+#define SIZE_READ(out, i) out[(i) * 3 + 2]
+
+		out_sz_ref = out_sz /= 2;
+		in_sz /= 2;
+		TEST_UTIL_BATCH_VALID;
+		assert_zu_le(sz, SIZE_READ(out, 0),
+		    "Extent size should be at least allocation size");
+		assert_zu_eq(SIZE_READ(out, 0) & (PAGE - 1), 0,
+		    "Extent size should be a multiple of page size");
+		if (sz <= SC_SMALL_MAXCLASS) {
+			assert_zu_le(NFREE_READ(out, 0), NREGS_READ(out, 0),
+			    "Extent free count exceeded region count");
+			assert_zu_le(NREGS_READ(out, 0), SIZE_READ(out, 0),
+			    "Extent region count exceeded size");
+			assert_zu_ne(NREGS_READ(out, 0), 0,
+			    "Extent region count must be positive");
+		} else {
+			assert_zu_eq(NFREE_READ(out, 0), 0,
+			    "Extent free count should be zero");
+			assert_zu_eq(NREGS_READ(out, 0), 1,
+			    "Extent region count should be one");
+		}
+		TEST_EQUAL_REF(1,
+		    "Should not overwrite content beyond what's needed");
+		in_sz *= 2;
+		out_sz_ref = out_sz *= 2;
+
+		memcpy(out_ref, out, 3 * sizeof(size_t));
+		TEST_UTIL_BATCH_VALID;
+		TEST_EQUAL_REF(0, "Statistics should be stable across calls");
+		if (sz <= SC_SMALL_MAXCLASS) {
+			assert_zu_le(NFREE_READ(out, 1), NREGS_READ(out, 1),
+			    "Extent free count exceeded region count");
+		} else {
+			assert_zu_eq(NFREE_READ(out, 0), 0,
+			    "Extent free count should be zero");
+		}
+		assert_zu_eq(NREGS_READ(out, 0), NREGS_READ(out, 1),
+		    "Extent region count should be same for same region size");
+		assert_zu_eq(SIZE_READ(out, 0), SIZE_READ(out, 1),
+		    "Extent size should be same for same region size");
+
+#undef SIZE_READ
+#undef NREGS_READ
+#undef NFREE_READ
+
+#undef TEST_EQUAL_REF
+
+		free(q);
+		free(p);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	assert_zu_lt(SC_SMALL_MAXCLASS, TEST_MAX_SIZE,
+	    "Test case cannot cover large classes");
+	return test(test_query, test_batch);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/fork.c b/memkind-1.10.0/jemalloc/test/unit/fork.c
new file mode 100644
index 0000000..b169075
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/fork.c
@@ -0,0 +1,141 @@
+#include "test/jemalloc_test.h"
+
+#ifndef _WIN32
+#include <sys/wait.h>
+#endif
+
+#ifndef _WIN32
+static void
+wait_for_child_exit(int pid) {
+	int status;
+	while (true) {
+		if (waitpid(pid, &status, 0) == -1) {
+			test_fail("Unexpected waitpid() failure.");
+		}
+		if (WIFSIGNALED(status)) {
+			test_fail("Unexpected child termination due to "
+			    "signal %d", WTERMSIG(status));
+			break;
+		}
+		if (WIFEXITED(status)) {
+			if (WEXITSTATUS(status) != 0) {
+				test_fail("Unexpected child exit value %d",
+				    WEXITSTATUS(status));
+			}
+			break;
+		}
+	}
+}
+#endif
+
+TEST_BEGIN(test_fork) {
+#ifndef _WIN32
+	void *p;
+	pid_t pid;
+
+	/* Set up a manually managed arena for test. */
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+
+	/* Migrate to the new arena. */
+	unsigned old_arena_ind;
+	sz = sizeof(old_arena_ind);
+	assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+	    (void *)&arena_ind, sizeof(arena_ind)), 0,
+	    "Unexpected mallctl() failure");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	pid = fork();
+
+	free(p);
+
+	p = malloc(64);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+	free(p);
+
+	if (pid == -1) {
+		/* Error. */
+		test_fail("Unexpected fork() failure");
+	} else if (pid == 0) {
+		/* Child. */
+		_exit(0);
+	} else {
+		wait_for_child_exit(pid);
+	}
+#else
+	test_skip("fork(2) is irrelevant to Windows");
+#endif
+}
+TEST_END
+
+#ifndef _WIN32
+static void *
+do_fork_thd(void *arg) {
+	malloc(1);
+	int pid = fork();
+	if (pid == -1) {
+		/* Error. */
+		test_fail("Unexpected fork() failure");
+	} else if (pid == 0) {
+		/* Child. */
+		char *args[] = {"true", NULL};
+		execvp(args[0], args);
+		test_fail("Exec failed");
+	} else {
+		/* Parent */
+		wait_for_child_exit(pid);
+	}
+	return NULL;
+}
+#endif
+
+#ifndef _WIN32
+static void
+do_test_fork_multithreaded() {
+	thd_t child;
+	thd_create(&child, do_fork_thd, NULL);
+	do_fork_thd(NULL);
+	thd_join(child, NULL);
+}
+#endif
+
+TEST_BEGIN(test_fork_multithreaded) {
+#ifndef _WIN32
+	/*
+	 * We've seen bugs involving hanging on arenas_lock (though the same
+	 * class of bugs can happen on any mutex).  The bugs are intermittent
+	 * though, so we want to run the test multiple times.  Since we hold the
+	 * arenas lock only early in the process lifetime, we can't just run
+	 * this test in a loop (since, after all the arenas are initialized, we
+	 * won't acquire arenas_lock any further).  We therefore repeat the test
+	 * with multiple processes.
+	 */
+	for (int i = 0; i < 100; i++) {
+		int pid = fork();
+		if (pid == -1) {
+			/* Error. */
+			test_fail("Unexpected fork() failure,");
+		} else if (pid == 0) {
+			/* Child. */
+			do_test_fork_multithreaded();
+			_exit(0);
+		} else {
+			wait_for_child_exit(pid);
+		}
+	}
+#else
+	test_skip("fork(2) is irrelevant to Windows");
+#endif
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_fork,
+	    test_fork_multithreaded);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/hash.c b/memkind-1.10.0/jemalloc/test/unit/hash.c
new file mode 100644
index 0000000..7cc034f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/hash.c
@@ -0,0 +1,173 @@
+/*
+ * This file is based on code that is part of SMHasher
+ * (https://code.google.com/p/smhasher/), and is subject to the MIT license
+ * (http://www.opensource.org/licenses/mit-license.php).  Both email addresses
+ * associated with the source code's revision history belong to Austin Appleby,
+ * and the revision history ranges from 2010 to 2012.  Therefore the copyright
+ * and license are here taken to be:
+ *
+ * Copyright (c) 2010-2012 Austin Appleby
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ */
+
+#include "test/jemalloc_test.h"
+#include "jemalloc/internal/hash.h"
+
+typedef enum {
+	hash_variant_x86_32,
+	hash_variant_x86_128,
+	hash_variant_x64_128
+} hash_variant_t;
+
+static int
+hash_variant_bits(hash_variant_t variant) {
+	switch (variant) {
+	case hash_variant_x86_32: return 32;
+	case hash_variant_x86_128: return 128;
+	case hash_variant_x64_128: return 128;
+	default: not_reached();
+	}
+}
+
+static const char *
+hash_variant_string(hash_variant_t variant) {
+	switch (variant) {
+	case hash_variant_x86_32: return "hash_x86_32";
+	case hash_variant_x86_128: return "hash_x86_128";
+	case hash_variant_x64_128: return "hash_x64_128";
+	default: not_reached();
+	}
+}
+
+#define KEY_SIZE	256
+static void
+hash_variant_verify_key(hash_variant_t variant, uint8_t *key) {
+	const int hashbytes = hash_variant_bits(variant) / 8;
+	const int hashes_size = hashbytes * 256;
+	VARIABLE_ARRAY(uint8_t, hashes, hashes_size);
+	VARIABLE_ARRAY(uint8_t, final, hashbytes);
+	unsigned i;
+	uint32_t computed, expected;
+
+	memset(key, 0, KEY_SIZE);
+	memset(hashes, 0, hashes_size);
+	memset(final, 0, hashbytes);
+
+	/*
+	 * Hash keys of the form {0}, {0,1}, {0,1,2}, ..., {0,1,...,255} as the
+	 * seed.
+	 */
+	for (i = 0; i < 256; i++) {
+		key[i] = (uint8_t)i;
+		switch (variant) {
+		case hash_variant_x86_32: {
+			uint32_t out;
+			out = hash_x86_32(key, i, 256-i);
+			memcpy(&hashes[i*hashbytes], &out, hashbytes);
+			break;
+		} case hash_variant_x86_128: {
+			uint64_t out[2];
+			hash_x86_128(key, i, 256-i, out);
+			memcpy(&hashes[i*hashbytes], out, hashbytes);
+			break;
+		} case hash_variant_x64_128: {
+			uint64_t out[2];
+			hash_x64_128(key, i, 256-i, out);
+			memcpy(&hashes[i*hashbytes], out, hashbytes);
+			break;
+		} default: not_reached();
+		}
+	}
+
+	/* Hash the result array. */
+	switch (variant) {
+	case hash_variant_x86_32: {
+		uint32_t out = hash_x86_32(hashes, hashes_size, 0);
+		memcpy(final, &out, sizeof(out));
+		break;
+	} case hash_variant_x86_128: {
+		uint64_t out[2];
+		hash_x86_128(hashes, hashes_size, 0, out);
+		memcpy(final, out, sizeof(out));
+		break;
+	} case hash_variant_x64_128: {
+		uint64_t out[2];
+		hash_x64_128(hashes, hashes_size, 0, out);
+		memcpy(final, out, sizeof(out));
+		break;
+	} default: not_reached();
+	}
+
+	computed = (final[0] << 0) | (final[1] << 8) | (final[2] << 16) |
+	    (final[3] << 24);
+
+	switch (variant) {
+#ifdef JEMALLOC_BIG_ENDIAN
+	case hash_variant_x86_32: expected = 0x6213303eU; break;
+	case hash_variant_x86_128: expected = 0x266820caU; break;
+	case hash_variant_x64_128: expected = 0xcc622b6fU; break;
+#else
+	case hash_variant_x86_32: expected = 0xb0f57ee3U; break;
+	case hash_variant_x86_128: expected = 0xb3ece62aU; break;
+	case hash_variant_x64_128: expected = 0x6384ba69U; break;
+#endif
+	default: not_reached();
+	}
+
+	assert_u32_eq(computed, expected,
+	    "Hash mismatch for %s(): expected %#x but got %#x",
+	    hash_variant_string(variant), expected, computed);
+}
+
+static void
+hash_variant_verify(hash_variant_t variant) {
+#define MAX_ALIGN	16
+	uint8_t key[KEY_SIZE + (MAX_ALIGN - 1)];
+	unsigned i;
+
+	for (i = 0; i < MAX_ALIGN; i++) {
+		hash_variant_verify_key(variant, &key[i]);
+	}
+#undef MAX_ALIGN
+}
+#undef KEY_SIZE
+
+TEST_BEGIN(test_hash_x86_32) {
+	hash_variant_verify(hash_variant_x86_32);
+}
+TEST_END
+
+TEST_BEGIN(test_hash_x86_128) {
+	hash_variant_verify(hash_variant_x86_128);
+}
+TEST_END
+
+TEST_BEGIN(test_hash_x64_128) {
+	hash_variant_verify(hash_variant_x64_128);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_hash_x86_32,
+	    test_hash_x86_128,
+	    test_hash_x64_128);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/hook.c b/memkind-1.10.0/jemalloc/test/unit/hook.c
new file mode 100644
index 0000000..72fcc43
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/hook.c
@@ -0,0 +1,580 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/hook.h"
+
+static void *arg_extra;
+static int arg_type;
+static void *arg_result;
+static void *arg_address;
+static size_t arg_old_usize;
+static size_t arg_new_usize;
+static uintptr_t arg_result_raw;
+static uintptr_t arg_args_raw[4];
+
+static int call_count = 0;
+
+static void
+reset_args() {
+	arg_extra = NULL;
+	arg_type = 12345;
+	arg_result = NULL;
+	arg_address = NULL;
+	arg_old_usize = 0;
+	arg_new_usize = 0;
+	arg_result_raw = 0;
+	memset(arg_args_raw, 77, sizeof(arg_args_raw));
+}
+
+static void
+alloc_free_size(size_t sz) {
+	void *ptr = mallocx(1, 0);
+	free(ptr);
+	ptr = mallocx(1, 0);
+	free(ptr);
+	ptr = mallocx(1, MALLOCX_TCACHE_NONE);
+	dallocx(ptr, MALLOCX_TCACHE_NONE);
+}
+
+/*
+ * We want to support a degree of user reentrancy.  This tests a variety of
+ * allocation scenarios.
+ */
+static void
+be_reentrant() {
+	/* Let's make sure the tcache is non-empty if enabled. */
+	alloc_free_size(1);
+	alloc_free_size(1024);
+	alloc_free_size(64 * 1024);
+	alloc_free_size(256 * 1024);
+	alloc_free_size(1024 * 1024);
+
+	/* Some reallocation. */
+	void *ptr = mallocx(129, 0);
+	ptr = rallocx(ptr, 130, 0);
+	free(ptr);
+
+	ptr = mallocx(2 * 1024 * 1024, 0);
+	free(ptr);
+	ptr = mallocx(1 * 1024 * 1024, 0);
+	ptr = rallocx(ptr, 2 * 1024 * 1024, 0);
+	free(ptr);
+
+	ptr = mallocx(1, 0);
+	ptr = rallocx(ptr, 1000, 0);
+	free(ptr);
+}
+
+static void
+set_args_raw(uintptr_t *args_raw, int nargs) {
+	memcpy(arg_args_raw, args_raw, sizeof(uintptr_t) * nargs);
+}
+
+static void
+assert_args_raw(uintptr_t *args_raw_expected, int nargs) {
+	int cmp = memcmp(args_raw_expected, arg_args_raw,
+	    sizeof(uintptr_t) * nargs);
+	assert_d_eq(cmp, 0, "Raw args mismatch");
+}
+
+static void
+reset() {
+	call_count = 0;
+	reset_args();
+}
+
+static void
+test_alloc_hook(void *extra, hook_alloc_t type, void *result,
+    uintptr_t result_raw, uintptr_t args_raw[3]) {
+	call_count++;
+	arg_extra = extra;
+	arg_type = (int)type;
+	arg_result = result;
+	arg_result_raw = result_raw;
+	set_args_raw(args_raw, 3);
+	be_reentrant();
+}
+
+static void
+test_dalloc_hook(void *extra, hook_dalloc_t type, void *address,
+    uintptr_t args_raw[3]) {
+	call_count++;
+	arg_extra = extra;
+	arg_type = (int)type;
+	arg_address = address;
+	set_args_raw(args_raw, 3);
+	be_reentrant();
+}
+
+static void
+test_expand_hook(void *extra, hook_expand_t type, void *address,
+    size_t old_usize, size_t new_usize, uintptr_t result_raw,
+    uintptr_t args_raw[4]) {
+	call_count++;
+	arg_extra = extra;
+	arg_type = (int)type;
+	arg_address = address;
+	arg_old_usize = old_usize;
+	arg_new_usize = new_usize;
+	arg_result_raw = result_raw;
+	set_args_raw(args_raw, 4);
+	be_reentrant();
+}
+
+TEST_BEGIN(test_hooks_basic) {
+	/* Just verify that the record their arguments correctly. */
+	hooks_t hooks = {
+		&test_alloc_hook, &test_dalloc_hook, &test_expand_hook,
+		(void *)111};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	uintptr_t args_raw[4] = {10, 20, 30, 40};
+
+	/* Alloc */
+	reset_args();
+	hook_invoke_alloc(hook_alloc_posix_memalign, (void *)222, 333,
+	    args_raw);
+	assert_ptr_eq(arg_extra, (void *)111, "Passed wrong user pointer");
+	assert_d_eq((int)hook_alloc_posix_memalign, arg_type,
+	    "Passed wrong alloc type");
+	assert_ptr_eq((void *)222, arg_result, "Passed wrong result address");
+	assert_u64_eq(333, arg_result_raw, "Passed wrong result");
+	assert_args_raw(args_raw, 3);
+
+	/* Dalloc */
+	reset_args();
+	hook_invoke_dalloc(hook_dalloc_sdallocx, (void *)222, args_raw);
+	assert_d_eq((int)hook_dalloc_sdallocx, arg_type,
+	    "Passed wrong dalloc type");
+	assert_ptr_eq((void *)111, arg_extra, "Passed wrong user pointer");
+	assert_ptr_eq((void *)222, arg_address, "Passed wrong address");
+	assert_args_raw(args_raw, 3);
+
+	/* Expand */
+	reset_args();
+	hook_invoke_expand(hook_expand_xallocx, (void *)222, 333, 444, 555,
+	    args_raw);
+	assert_d_eq((int)hook_expand_xallocx, arg_type,
+	    "Passed wrong expand type");
+	assert_ptr_eq((void *)111, arg_extra, "Passed wrong user pointer");
+	assert_ptr_eq((void *)222, arg_address, "Passed wrong address");
+	assert_zu_eq(333, arg_old_usize, "Passed wrong old usize");
+	assert_zu_eq(444, arg_new_usize, "Passed wrong new usize");
+	assert_zu_eq(555, arg_result_raw, "Passed wrong result");
+	assert_args_raw(args_raw, 4);
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_null) {
+	/* Null hooks should be ignored, not crash. */
+	hooks_t hooks1 = {NULL, NULL, NULL, NULL};
+	hooks_t hooks2 = {&test_alloc_hook, NULL, NULL, NULL};
+	hooks_t hooks3 = {NULL, &test_dalloc_hook, NULL, NULL};
+	hooks_t hooks4 = {NULL, NULL, &test_expand_hook, NULL};
+
+	void *handle1 = hook_install(TSDN_NULL, &hooks1);
+	void *handle2 = hook_install(TSDN_NULL, &hooks2);
+	void *handle3 = hook_install(TSDN_NULL, &hooks3);
+	void *handle4 = hook_install(TSDN_NULL, &hooks4);
+
+	assert_ptr_ne(handle1, NULL, "Hook installation failed");
+	assert_ptr_ne(handle2, NULL, "Hook installation failed");
+	assert_ptr_ne(handle3, NULL, "Hook installation failed");
+	assert_ptr_ne(handle4, NULL, "Hook installation failed");
+
+	uintptr_t args_raw[4] = {10, 20, 30, 40};
+
+	call_count = 0;
+	hook_invoke_alloc(hook_alloc_malloc, NULL, 0, args_raw);
+	assert_d_eq(call_count, 1, "Called wrong number of times");
+
+	call_count = 0;
+	hook_invoke_dalloc(hook_dalloc_free, NULL, args_raw);
+	assert_d_eq(call_count, 1, "Called wrong number of times");
+
+	call_count = 0;
+	hook_invoke_expand(hook_expand_realloc, NULL, 0, 0, 0, args_raw);
+	assert_d_eq(call_count, 1, "Called wrong number of times");
+
+	hook_remove(TSDN_NULL, handle1);
+	hook_remove(TSDN_NULL, handle2);
+	hook_remove(TSDN_NULL, handle3);
+	hook_remove(TSDN_NULL, handle4);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_remove) {
+	hooks_t hooks = {&test_alloc_hook, NULL, NULL, NULL};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+	call_count = 0;
+	uintptr_t args_raw[4] = {10, 20, 30, 40};
+	hook_invoke_alloc(hook_alloc_malloc, NULL, 0, args_raw);
+	assert_d_eq(call_count, 1, "Hook not invoked");
+
+	call_count = 0;
+	hook_remove(TSDN_NULL, handle);
+	hook_invoke_alloc(hook_alloc_malloc, NULL, 0, NULL);
+	assert_d_eq(call_count, 0, "Hook invoked after removal");
+
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_alloc_simple) {
+	/* "Simple" in the sense that we're not in a realloc variant. */
+	hooks_t hooks = {&test_alloc_hook, NULL, NULL, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	/* Stop malloc from being optimized away. */
+	volatile int err;
+	void *volatile ptr;
+
+	/* malloc */
+	reset();
+	ptr = malloc(1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_malloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[0], "Wrong argument");
+	free(ptr);
+
+	/* posix_memalign */
+	reset();
+	err = posix_memalign((void **)&ptr, 1024, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_posix_memalign,
+	    "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)err, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)&ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1024, arg_args_raw[1], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[2], "Wrong argument");
+	free(ptr);
+
+	/* aligned_alloc */
+	reset();
+	ptr = aligned_alloc(1024, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_aligned_alloc,
+	    "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1024, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/* calloc */
+	reset();
+	ptr = calloc(11, 13);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_calloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)11, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)13, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/* memalign */
+#ifdef JEMALLOC_OVERRIDE_MEMALIGN
+	reset();
+	ptr = memalign(1024, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_memalign, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1024, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+#endif /* JEMALLOC_OVERRIDE_MEMALIGN */
+
+	/* valloc */
+#ifdef JEMALLOC_OVERRIDE_VALLOC
+	reset();
+	ptr = valloc(1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_valloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[0], "Wrong argument");
+	free(ptr);
+#endif /* JEMALLOC_OVERRIDE_VALLOC */
+
+	/* mallocx */
+	reset();
+	ptr = mallocx(1, MALLOCX_LG_ALIGN(10));
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_mallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)MALLOCX_LG_ALIGN(10), arg_args_raw[1],
+	    "Wrong flags");
+	free(ptr);
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_dalloc_simple) {
+	/* "Simple" in the sense that we're not in a realloc variant. */
+	hooks_t hooks = {NULL, &test_dalloc_hook, NULL, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+
+	/* free() */
+	reset();
+	ptr = malloc(1);
+	free(ptr);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_free, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+
+	/* dallocx() */
+	reset();
+	ptr = malloc(1);
+	dallocx(ptr, MALLOCX_TCACHE_NONE);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_dallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)MALLOCX_TCACHE_NONE, arg_args_raw[1],
+	    "Wrong raw arg");
+
+	/* sdallocx() */
+	reset();
+	ptr = malloc(1);
+	sdallocx(ptr, 1, MALLOCX_TCACHE_NONE);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_sdallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)MALLOCX_TCACHE_NONE, arg_args_raw[2],
+	    "Wrong raw arg");
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_expand_simple) {
+	/* "Simple" in the sense that we're not in a realloc variant. */
+	hooks_t hooks = {NULL, NULL, &test_expand_hook, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+
+	/* xallocx() */
+	reset();
+	ptr = malloc(1);
+	size_t new_usize = xallocx(ptr, 100, 200, MALLOCX_TCACHE_NONE);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_expand_xallocx, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer expanded");
+	assert_u64_eq(arg_old_usize, nallocx(1, 0), "Wrong old usize");
+	assert_u64_eq(arg_new_usize, sallocx(ptr, 0), "Wrong new usize");
+	assert_u64_eq(new_usize, arg_result_raw, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong arg");
+	assert_u64_eq(100, arg_args_raw[1], "Wrong arg");
+	assert_u64_eq(200, arg_args_raw[2], "Wrong arg");
+	assert_u64_eq(MALLOCX_TCACHE_NONE, arg_args_raw[3], "Wrong arg");
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_realloc_as_malloc_or_free) {
+	hooks_t hooks = {&test_alloc_hook, &test_dalloc_hook,
+		&test_expand_hook, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+
+	/* realloc(NULL, size) as malloc */
+	reset();
+	ptr = realloc(NULL, 1);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_realloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)NULL, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)1, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/* realloc(ptr, 0) as free */
+	ptr = malloc(1);
+	reset();
+	realloc(ptr, 0);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_dalloc_realloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong pointer freed");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong raw arg");
+	assert_u64_eq((uintptr_t)0, arg_args_raw[1], "Wrong raw arg");
+
+	/* realloc(NULL, 0) as malloc(0) */
+	reset();
+	ptr = realloc(NULL, 0);
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, (int)hook_alloc_realloc, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_result, "Wrong result");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)NULL, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)0, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	hook_remove(TSDN_NULL, handle);
+}
+TEST_END
+
+static void
+do_realloc_test(void *(*ralloc)(void *, size_t, int), int flags,
+    int expand_type, int dalloc_type) {
+	hooks_t hooks = {&test_alloc_hook, &test_dalloc_hook,
+		&test_expand_hook, (void *)123};
+	void *handle = hook_install(TSDN_NULL, &hooks);
+	assert_ptr_ne(handle, NULL, "Hook installation failed");
+
+	void *volatile ptr;
+	void *volatile ptr2;
+
+	/* Realloc in-place, small. */
+	ptr = malloc(129);
+	reset();
+	ptr2 = ralloc(ptr, 130, flags);
+	assert_ptr_eq(ptr, ptr2, "Small realloc moved");
+
+	assert_d_eq(call_count, 1, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, expand_type, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)130, arg_args_raw[1], "Wrong argument");
+	free(ptr);
+
+	/*
+	 * Realloc in-place, large.  Since we can't guarantee the large case
+	 * across all platforms, we stay resilient to moving results.
+	 */
+	ptr = malloc(2 * 1024 * 1024);
+	free(ptr);
+	ptr2 = malloc(1 * 1024 * 1024);
+	reset();
+	ptr = ralloc(ptr2, 2 * 1024 * 1024, flags);
+	/* ptr is the new address, ptr2 is the old address. */
+	if (ptr == ptr2) {
+		assert_d_eq(call_count, 1, "Hook not called");
+		assert_d_eq(arg_type, expand_type, "Wrong hook type");
+	} else {
+		assert_d_eq(call_count, 2, "Wrong hooks called");
+		assert_ptr_eq(ptr, arg_result, "Wrong address");
+		assert_d_eq(arg_type, dalloc_type, "Wrong hook type");
+	}
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_ptr_eq(ptr2, arg_address, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr2, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)2 * 1024 * 1024, arg_args_raw[1],
+	    "Wrong argument");
+	free(ptr);
+
+	/* Realloc with move, small. */
+	ptr = malloc(8);
+	reset();
+	ptr2 = ralloc(ptr, 128, flags);
+	assert_ptr_ne(ptr, ptr2, "Small realloc didn't move");
+
+	assert_d_eq(call_count, 2, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, dalloc_type, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong address");
+	assert_ptr_eq(ptr2, arg_result, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr2, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)128, arg_args_raw[1], "Wrong argument");
+	free(ptr2);
+
+	/* Realloc with move, large. */
+	ptr = malloc(1);
+	reset();
+	ptr2 = ralloc(ptr, 2 * 1024 * 1024, flags);
+	assert_ptr_ne(ptr, ptr2, "Large realloc didn't move");
+
+	assert_d_eq(call_count, 2, "Hook not called");
+	assert_ptr_eq(arg_extra, (void *)123, "Wrong extra");
+	assert_d_eq(arg_type, dalloc_type, "Wrong hook type");
+	assert_ptr_eq(ptr, arg_address, "Wrong address");
+	assert_ptr_eq(ptr2, arg_result, "Wrong address");
+	assert_u64_eq((uintptr_t)ptr2, (uintptr_t)arg_result_raw,
+	    "Wrong raw result");
+	assert_u64_eq((uintptr_t)ptr, arg_args_raw[0], "Wrong argument");
+	assert_u64_eq((uintptr_t)2 * 1024 * 1024, arg_args_raw[1],
+	    "Wrong argument");
+	free(ptr2);
+
+	hook_remove(TSDN_NULL, handle);
+}
+
+static void *
+realloc_wrapper(void *ptr, size_t size, UNUSED int flags) {
+	return realloc(ptr, size);
+}
+
+TEST_BEGIN(test_hooks_realloc) {
+	do_realloc_test(&realloc_wrapper, 0, hook_expand_realloc,
+	    hook_dalloc_realloc);
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_rallocx) {
+	do_realloc_test(&rallocx, MALLOCX_TCACHE_NONE, hook_expand_rallocx,
+	    hook_dalloc_rallocx);
+}
+TEST_END
+
+int
+main(void) {
+	/* We assert on call counts. */
+	return test_no_reentrancy(
+	    test_hooks_basic,
+	    test_hooks_null,
+	    test_hooks_remove,
+	    test_hooks_alloc_simple,
+	    test_hooks_dalloc_simple,
+	    test_hooks_expand_simple,
+	    test_hooks_realloc_as_malloc_or_free,
+	    test_hooks_realloc,
+	    test_hooks_rallocx);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/huge.c b/memkind-1.10.0/jemalloc/test/unit/huge.c
new file mode 100644
index 0000000..ab72cf0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/huge.c
@@ -0,0 +1,108 @@
+#include "test/jemalloc_test.h"
+
+/* Threshold: 2 << 20 = 2097152. */
+const char *malloc_conf = "oversize_threshold:2097152";
+
+#define HUGE_SZ (2 << 20)
+#define SMALL_SZ (8)
+
+TEST_BEGIN(huge_bind_thread) {
+	unsigned arena1, arena2;
+	size_t sz = sizeof(unsigned);
+
+	/* Bind to a manual arena. */
+	assert_d_eq(mallctl("arenas.create", &arena1, &sz, NULL, 0), 0,
+	    "Failed to create arena");
+	assert_d_eq(mallctl("thread.arena", NULL, NULL, &arena1,
+	    sizeof(arena1)), 0, "Fail to bind thread");
+
+	void *ptr = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_eq(arena1, arena2, "Wrong arena used after binding");
+	dallocx(ptr, 0);
+
+	/* Switch back to arena 0. */
+	test_skip_if(have_percpu_arena &&
+	    PERCPU_ARENA_ENABLED(opt_percpu_arena));
+	arena2 = 0;
+	assert_d_eq(mallctl("thread.arena", NULL, NULL, &arena2,
+	    sizeof(arena2)), 0, "Fail to bind thread");
+	ptr = mallocx(SMALL_SZ, MALLOCX_TCACHE_NONE);
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_eq(arena2, 0, "Wrong arena used after binding");
+	dallocx(ptr, MALLOCX_TCACHE_NONE);
+
+	/* Then huge allocation should use the huge arena. */
+	ptr = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena2, 0, "Wrong arena used after binding");
+	assert_u_ne(arena1, arena2, "Wrong arena used after binding");
+	dallocx(ptr, 0);
+}
+TEST_END
+
+TEST_BEGIN(huge_mallocx) {
+	unsigned arena1, arena2;
+	size_t sz = sizeof(unsigned);
+
+	assert_d_eq(mallctl("arenas.create", &arena1, &sz, NULL, 0), 0,
+	    "Failed to create arena");
+	void *huge = mallocx(HUGE_SZ, MALLOCX_ARENA(arena1));
+	assert_ptr_not_null(huge, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &huge,
+	    sizeof(huge)), 0, "Unexpected mallctl() failure");
+	assert_u_eq(arena1, arena2, "Wrong arena used for mallocx");
+	dallocx(huge, MALLOCX_ARENA(arena1));
+
+	void *huge2 = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(huge, "Fail to allocate huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &huge2,
+	    sizeof(huge2)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena1, arena2,
+	    "Huge allocation should not come from the manual arena.");
+	assert_u_ne(arena2, 0,
+	    "Huge allocation should not come from the arena 0.");
+	dallocx(huge2, 0);
+}
+TEST_END
+
+TEST_BEGIN(huge_allocation) {
+	unsigned arena1, arena2;
+
+	void *ptr = mallocx(HUGE_SZ, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate huge size");
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.lookup", &arena1, &sz, &ptr, sizeof(ptr)),
+	    0, "Unexpected mallctl() failure");
+	assert_u_gt(arena1, 0, "Huge allocation should not come from arena 0");
+	dallocx(ptr, 0);
+
+	ptr = mallocx(HUGE_SZ >> 1, 0);
+	assert_ptr_not_null(ptr, "Fail to allocate half huge size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena1, arena2, "Wrong arena used for half huge");
+	dallocx(ptr, 0);
+
+	ptr = mallocx(SMALL_SZ, MALLOCX_TCACHE_NONE);
+	assert_ptr_not_null(ptr, "Fail to allocate small size");
+	assert_d_eq(mallctl("arenas.lookup", &arena2, &sz, &ptr,
+	    sizeof(ptr)), 0, "Unexpected mallctl() failure");
+	assert_u_ne(arena1, arena2,
+	    "Huge and small should be from different arenas");
+	dallocx(ptr, 0);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    huge_allocation,
+	    huge_mallocx,
+	    huge_bind_thread);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/junk.c b/memkind-1.10.0/jemalloc/test/unit/junk.c
new file mode 100644
index 0000000..57e3ad4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/junk.c
@@ -0,0 +1,141 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/util.h"
+
+static arena_dalloc_junk_small_t *arena_dalloc_junk_small_orig;
+static large_dalloc_junk_t *large_dalloc_junk_orig;
+static large_dalloc_maybe_junk_t *large_dalloc_maybe_junk_orig;
+static void *watch_for_junking;
+static bool saw_junking;
+
+static void
+watch_junking(void *p) {
+	watch_for_junking = p;
+	saw_junking = false;
+}
+
+static void
+arena_dalloc_junk_small_intercept(void *ptr, const bin_info_t *bin_info) {
+	size_t i;
+
+	arena_dalloc_junk_small_orig(ptr, bin_info);
+	for (i = 0; i < bin_info->reg_size; i++) {
+		assert_u_eq(((uint8_t *)ptr)[i], JEMALLOC_FREE_JUNK,
+		    "Missing junk fill for byte %zu/%zu of deallocated region",
+		    i, bin_info->reg_size);
+	}
+	if (ptr == watch_for_junking) {
+		saw_junking = true;
+	}
+}
+
+static void
+large_dalloc_junk_intercept(void *ptr, size_t usize) {
+	size_t i;
+
+	large_dalloc_junk_orig(ptr, usize);
+	for (i = 0; i < usize; i++) {
+		assert_u_eq(((uint8_t *)ptr)[i], JEMALLOC_FREE_JUNK,
+		    "Missing junk fill for byte %zu/%zu of deallocated region",
+		    i, usize);
+	}
+	if (ptr == watch_for_junking) {
+		saw_junking = true;
+	}
+}
+
+static void
+large_dalloc_maybe_junk_intercept(void *ptr, size_t usize) {
+	large_dalloc_maybe_junk_orig(ptr, usize);
+	if (ptr == watch_for_junking) {
+		saw_junking = true;
+	}
+}
+
+static void
+test_junk(size_t sz_min, size_t sz_max) {
+	uint8_t *s;
+	size_t sz_prev, sz, i;
+
+	if (opt_junk_free) {
+		arena_dalloc_junk_small_orig = arena_dalloc_junk_small;
+		arena_dalloc_junk_small = arena_dalloc_junk_small_intercept;
+		large_dalloc_junk_orig = large_dalloc_junk;
+		large_dalloc_junk = large_dalloc_junk_intercept;
+		large_dalloc_maybe_junk_orig = large_dalloc_maybe_junk;
+		large_dalloc_maybe_junk = large_dalloc_maybe_junk_intercept;
+	}
+
+	sz_prev = 0;
+	s = (uint8_t *)mallocx(sz_min, 0);
+	assert_ptr_not_null((void *)s, "Unexpected mallocx() failure");
+
+	for (sz = sallocx(s, 0); sz <= sz_max;
+	    sz_prev = sz, sz = sallocx(s, 0)) {
+		if (sz_prev > 0) {
+			assert_u_eq(s[0], 'a',
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    ZU(0), sz_prev);
+			assert_u_eq(s[sz_prev-1], 'a',
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    sz_prev-1, sz_prev);
+		}
+
+		for (i = sz_prev; i < sz; i++) {
+			if (opt_junk_alloc) {
+				assert_u_eq(s[i], JEMALLOC_ALLOC_JUNK,
+				    "Newly allocated byte %zu/%zu isn't "
+				    "junk-filled", i, sz);
+			}
+			s[i] = 'a';
+		}
+
+		if (xallocx(s, sz+1, 0, 0) == sz) {
+			uint8_t *t;
+			watch_junking(s);
+			t = (uint8_t *)rallocx(s, sz+1, 0);
+			assert_ptr_not_null((void *)t,
+			    "Unexpected rallocx() failure");
+			assert_zu_ge(sallocx(t, 0), sz+1,
+			    "Unexpectedly small rallocx() result");
+			if (!background_thread_enabled()) {
+				assert_ptr_ne(s, t,
+				    "Unexpected in-place rallocx()");
+				assert_true(!opt_junk_free || saw_junking,
+				    "Expected region of size %zu to be "
+				    "junk-filled", sz);
+			}
+			s = t;
+		}
+	}
+
+	watch_junking(s);
+	dallocx(s, 0);
+	assert_true(!opt_junk_free || saw_junking,
+	    "Expected region of size %zu to be junk-filled", sz);
+
+	if (opt_junk_free) {
+		arena_dalloc_junk_small = arena_dalloc_junk_small_orig;
+		large_dalloc_junk = large_dalloc_junk_orig;
+		large_dalloc_maybe_junk = large_dalloc_maybe_junk_orig;
+	}
+}
+
+TEST_BEGIN(test_junk_small) {
+	test_skip_if(!config_fill);
+	test_junk(1, SC_SMALL_MAXCLASS - 1);
+}
+TEST_END
+
+TEST_BEGIN(test_junk_large) {
+	test_skip_if(!config_fill);
+	test_junk(SC_SMALL_MAXCLASS + 1, (1U << (SC_LG_LARGE_MINCLASS + 1)));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_junk_small,
+	    test_junk_large);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/junk.sh b/memkind-1.10.0/jemalloc/test/unit/junk.sh
new file mode 100644
index 0000000..97cd8ca
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/junk.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,zero:false,junk:true"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/junk_alloc.c b/memkind-1.10.0/jemalloc/test/unit/junk_alloc.c
new file mode 100644
index 0000000..a442a0c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/junk_alloc.c
@@ -0,0 +1 @@
+#include "junk.c"
diff --git a/memkind-1.10.0/jemalloc/test/unit/junk_alloc.sh b/memkind-1.10.0/jemalloc/test/unit/junk_alloc.sh
new file mode 100644
index 0000000..e1008c2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/junk_alloc.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,zero:false,junk:alloc"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/junk_free.c b/memkind-1.10.0/jemalloc/test/unit/junk_free.c
new file mode 100644
index 0000000..a442a0c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/junk_free.c
@@ -0,0 +1 @@
+#include "junk.c"
diff --git a/memkind-1.10.0/jemalloc/test/unit/junk_free.sh b/memkind-1.10.0/jemalloc/test/unit/junk_free.sh
new file mode 100644
index 0000000..402196c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/junk_free.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,zero:false,junk:free"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/log.c b/memkind-1.10.0/jemalloc/test/unit/log.c
new file mode 100644
index 0000000..a52bd73
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/log.c
@@ -0,0 +1,193 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/log.h"
+
+static void
+expect_no_logging(const char *names) {
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	log_var_t log_l2 = LOG_VAR_INIT("l2");
+	log_var_t log_l2_a = LOG_VAR_INIT("l2.a");
+
+	strcpy(log_var_names, names);
+
+	int count = 0;
+
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+			count++;
+		log_do_end(log_l1)
+
+		log_do_begin(log_l2)
+			count++;
+		log_do_end(log_l2)
+
+		log_do_begin(log_l2_a)
+			count++;
+		log_do_end(log_l2_a)
+	}
+	assert_d_eq(count, 0, "Disabled logging not ignored!");
+}
+
+TEST_BEGIN(test_log_disabled) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	expect_no_logging("");
+	expect_no_logging("abc");
+	expect_no_logging("a.b.c");
+	expect_no_logging("l12");
+	expect_no_logging("l123|a456|b789");
+	expect_no_logging("|||");
+}
+TEST_END
+
+TEST_BEGIN(test_log_enabled_direct) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	log_var_t log_l1_a = LOG_VAR_INIT("l1.a");
+	log_var_t log_l2 = LOG_VAR_INIT("l2");
+
+	int count;
+
+	count = 0;
+	strcpy(log_var_names, "l1");
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+			count++;
+		log_do_end(log_l1)
+	}
+	assert_d_eq(count, 10, "Mis-logged!");
+
+	count = 0;
+	strcpy(log_var_names, "l1.a");
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1_a)
+			count++;
+		log_do_end(log_l1_a)
+	}
+	assert_d_eq(count, 10, "Mis-logged!");
+
+	count = 0;
+	strcpy(log_var_names, "l1.a|abc|l2|def");
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1_a)
+			count++;
+		log_do_end(log_l1_a)
+
+		log_do_begin(log_l2)
+			count++;
+		log_do_end(log_l2)
+	}
+	assert_d_eq(count, 20, "Mis-logged!");
+}
+TEST_END
+
+TEST_BEGIN(test_log_enabled_indirect) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	strcpy(log_var_names, "l0|l1|abc|l2.b|def");
+
+	/* On. */
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	/* Off. */
+	log_var_t log_l1a = LOG_VAR_INIT("l1a");
+	/* On. */
+	log_var_t log_l1_a = LOG_VAR_INIT("l1.a");
+	/* Off. */
+	log_var_t log_l2_a = LOG_VAR_INIT("l2.a");
+	/* On. */
+	log_var_t log_l2_b_a = LOG_VAR_INIT("l2.b.a");
+	/* On. */
+	log_var_t log_l2_b_b = LOG_VAR_INIT("l2.b.b");
+
+	/* 4 are on total, so should sum to 40. */
+	int count = 0;
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+			count++;
+		log_do_end(log_l1)
+
+		log_do_begin(log_l1a)
+			count++;
+		log_do_end(log_l1a)
+
+		log_do_begin(log_l1_a)
+			count++;
+		log_do_end(log_l1_a)
+
+		log_do_begin(log_l2_a)
+			count++;
+		log_do_end(log_l2_a)
+
+		log_do_begin(log_l2_b_a)
+			count++;
+		log_do_end(log_l2_b_a)
+
+		log_do_begin(log_l2_b_b)
+			count++;
+		log_do_end(log_l2_b_b)
+	}
+
+	assert_d_eq(count, 40, "Mis-logged!");
+}
+TEST_END
+
+TEST_BEGIN(test_log_enabled_global) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, true, ATOMIC_RELAXED);
+	strcpy(log_var_names, "abc|.|def");
+
+	log_var_t log_l1 = LOG_VAR_INIT("l1");
+	log_var_t log_l2_a_a = LOG_VAR_INIT("l2.a.a");
+
+	int count = 0;
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(log_l1)
+		    count++;
+		log_do_end(log_l1)
+
+		log_do_begin(log_l2_a_a)
+		    count++;
+		log_do_end(log_l2_a_a)
+	}
+	assert_d_eq(count, 20, "Mis-logged!");
+}
+TEST_END
+
+TEST_BEGIN(test_logs_if_no_init) {
+	test_skip_if(!config_log);
+	atomic_store_b(&log_init_done, false, ATOMIC_RELAXED);
+
+	log_var_t l = LOG_VAR_INIT("definitely.not.enabled");
+
+	int count = 0;
+	for (int i = 0; i < 10; i++) {
+		log_do_begin(l)
+			count++;
+		log_do_end(l)
+	}
+	assert_d_eq(count, 0, "Logging shouldn't happen if not initialized.");
+}
+TEST_END
+
+/*
+ * This really just checks to make sure that this usage compiles; we don't have
+ * any test code to run.
+ */
+TEST_BEGIN(test_log_only_format_string) {
+	if (false) {
+		LOG("log_str", "No arguments follow this format string.");
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_log_disabled,
+	    test_log_enabled_direct,
+	    test_log_enabled_indirect,
+	    test_log_enabled_global,
+	    test_logs_if_no_init,
+	    test_log_only_format_string);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/mallctl.c b/memkind-1.10.0/jemalloc/test/unit/mallctl.c
new file mode 100644
index 0000000..3a75ac0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/mallctl.c
@@ -0,0 +1,888 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/hook.h"
+#include "jemalloc/internal/util.h"
+
+TEST_BEGIN(test_mallctl_errors) {
+	uint64_t epoch;
+	size_t sz;
+
+	assert_d_eq(mallctl("no_such_name", NULL, NULL, NULL, 0), ENOENT,
+	    "mallctl() should return ENOENT for non-existent names");
+
+	assert_d_eq(mallctl("version", NULL, NULL, "0.0.0", strlen("0.0.0")),
+	    EPERM, "mallctl() should return EPERM on attempt to write "
+	    "read-only value");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)-1), EINVAL,
+	    "mallctl() should return EINVAL for input size mismatch");
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)+1), EINVAL,
+	    "mallctl() should return EINVAL for input size mismatch");
+
+	sz = sizeof(epoch)-1;
+	assert_d_eq(mallctl("epoch", (void *)&epoch, &sz, NULL, 0), EINVAL,
+	    "mallctl() should return EINVAL for output size mismatch");
+	sz = sizeof(epoch)+1;
+	assert_d_eq(mallctl("epoch", (void *)&epoch, &sz, NULL, 0), EINVAL,
+	    "mallctl() should return EINVAL for output size mismatch");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctlnametomib_errors) {
+	size_t mib[1];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("no_such_name", mib, &miblen), ENOENT,
+	    "mallctlnametomib() should return ENOENT for non-existent names");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctlbymib_errors) {
+	uint64_t epoch;
+	size_t sz;
+	size_t mib[1];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("version", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, "0.0.0",
+	    strlen("0.0.0")), EPERM, "mallctl() should return EPERM on "
+	    "attempt to write read-only value");
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("epoch", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)-1), EINVAL,
+	    "mallctlbymib() should return EINVAL for input size mismatch");
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)+1), EINVAL,
+	    "mallctlbymib() should return EINVAL for input size mismatch");
+
+	sz = sizeof(epoch)-1;
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&epoch, &sz, NULL, 0),
+	    EINVAL,
+	    "mallctlbymib() should return EINVAL for output size mismatch");
+	sz = sizeof(epoch)+1;
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&epoch, &sz, NULL, 0),
+	    EINVAL,
+	    "mallctlbymib() should return EINVAL for output size mismatch");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctl_read_write) {
+	uint64_t old_epoch, new_epoch;
+	size_t sz = sizeof(old_epoch);
+
+	/* Blind. */
+	assert_d_eq(mallctl("epoch", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+
+	/* Read. */
+	assert_d_eq(mallctl("epoch", (void *)&old_epoch, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+
+	/* Write. */
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&new_epoch,
+	    sizeof(new_epoch)), 0, "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+
+	/* Read+write. */
+	assert_d_eq(mallctl("epoch", (void *)&old_epoch, &sz,
+	    (void *)&new_epoch, sizeof(new_epoch)), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(sz, sizeof(old_epoch), "Unexpected output size");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctlnametomib_short_mib) {
+	size_t mib[4];
+	size_t miblen;
+
+	miblen = 3;
+	mib[3] = 42;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.nregs", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	assert_zu_eq(miblen, 3, "Unexpected mib output length");
+	assert_zu_eq(mib[3], 42,
+	    "mallctlnametomib() wrote past the end of the input mib");
+}
+TEST_END
+
+TEST_BEGIN(test_mallctl_config) {
+#define TEST_MALLCTL_CONFIG(config, t) do {				\
+	t oldval;							\
+	size_t sz = sizeof(oldval);					\
+	assert_d_eq(mallctl("config."#config, (void *)&oldval, &sz,	\
+	    NULL, 0), 0, "Unexpected mallctl() failure");		\
+	assert_b_eq(oldval, config_##config, "Incorrect config value");	\
+	assert_zu_eq(sz, sizeof(oldval), "Unexpected output size");	\
+} while (0)
+
+	TEST_MALLCTL_CONFIG(cache_oblivious, bool);
+	TEST_MALLCTL_CONFIG(debug, bool);
+	TEST_MALLCTL_CONFIG(fill, bool);
+	TEST_MALLCTL_CONFIG(lazy_lock, bool);
+	TEST_MALLCTL_CONFIG(malloc_conf, const char *);
+	TEST_MALLCTL_CONFIG(prof, bool);
+	TEST_MALLCTL_CONFIG(prof_libgcc, bool);
+	TEST_MALLCTL_CONFIG(prof_libunwind, bool);
+	TEST_MALLCTL_CONFIG(stats, bool);
+	TEST_MALLCTL_CONFIG(utrace, bool);
+	TEST_MALLCTL_CONFIG(xmalloc, bool);
+
+#undef TEST_MALLCTL_CONFIG
+}
+TEST_END
+
+TEST_BEGIN(test_mallctl_opt) {
+	bool config_always = true;
+
+#define TEST_MALLCTL_OPT(t, opt, config) do {				\
+	t oldval;							\
+	size_t sz = sizeof(oldval);					\
+	int expected = config_##config ? 0 : ENOENT;			\
+	int result = mallctl("opt."#opt, (void *)&oldval, &sz, NULL,	\
+	    0);								\
+	assert_d_eq(result, expected,					\
+	    "Unexpected mallctl() result for opt."#opt);		\
+	assert_zu_eq(sz, sizeof(oldval), "Unexpected output size");	\
+} while (0)
+
+	TEST_MALLCTL_OPT(bool, abort, always);
+	TEST_MALLCTL_OPT(bool, abort_conf, always);
+	TEST_MALLCTL_OPT(bool, confirm_conf, always);
+	TEST_MALLCTL_OPT(const char *, metadata_thp, always);
+	TEST_MALLCTL_OPT(bool, retain, always);
+	TEST_MALLCTL_OPT(const char *, dss, always);
+	TEST_MALLCTL_OPT(unsigned, narenas, always);
+	TEST_MALLCTL_OPT(const char *, percpu_arena, always);
+	TEST_MALLCTL_OPT(size_t, oversize_threshold, always);
+	TEST_MALLCTL_OPT(bool, background_thread, always);
+	TEST_MALLCTL_OPT(ssize_t, dirty_decay_ms, always);
+	TEST_MALLCTL_OPT(ssize_t, muzzy_decay_ms, always);
+	TEST_MALLCTL_OPT(bool, stats_print, always);
+	TEST_MALLCTL_OPT(const char *, junk, fill);
+	TEST_MALLCTL_OPT(bool, zero, fill);
+	TEST_MALLCTL_OPT(bool, utrace, utrace);
+	TEST_MALLCTL_OPT(bool, xmalloc, xmalloc);
+	TEST_MALLCTL_OPT(bool, tcache, always);
+	TEST_MALLCTL_OPT(size_t, lg_extent_max_active_fit, always);
+	TEST_MALLCTL_OPT(size_t, lg_tcache_max, always);
+	TEST_MALLCTL_OPT(const char *, thp, always);
+	TEST_MALLCTL_OPT(bool, prof, prof);
+	TEST_MALLCTL_OPT(const char *, prof_prefix, prof);
+	TEST_MALLCTL_OPT(bool, prof_active, prof);
+	TEST_MALLCTL_OPT(ssize_t, lg_prof_sample, prof);
+	TEST_MALLCTL_OPT(bool, prof_accum, prof);
+	TEST_MALLCTL_OPT(ssize_t, lg_prof_interval, prof);
+	TEST_MALLCTL_OPT(bool, prof_gdump, prof);
+	TEST_MALLCTL_OPT(bool, prof_final, prof);
+	TEST_MALLCTL_OPT(bool, prof_leak, prof);
+
+#undef TEST_MALLCTL_OPT
+}
+TEST_END
+
+TEST_BEGIN(test_manpage_example) {
+	unsigned nbins, i;
+	size_t mib[4];
+	size_t len, miblen;
+
+	len = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &len, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	miblen = 4;
+	assert_d_eq(mallctlnametomib("arenas.bin.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	for (i = 0; i < nbins; i++) {
+		size_t bin_size;
+
+		mib[2] = i;
+		len = sizeof(bin_size);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&bin_size, &len,
+		    NULL, 0), 0, "Unexpected mallctlbymib() failure");
+		/* Do something with bin_size... */
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_tcache_none) {
+	test_skip_if(!opt_tcache);
+
+	/* Allocate p and q. */
+	void *p0 = mallocx(42, 0);
+	assert_ptr_not_null(p0, "Unexpected mallocx() failure");
+	void *q = mallocx(42, 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+
+	/* Deallocate p and q, but bypass the tcache for q. */
+	dallocx(p0, 0);
+	dallocx(q, MALLOCX_TCACHE_NONE);
+
+	/* Make sure that tcache-based allocation returns p, not q. */
+	void *p1 = mallocx(42, 0);
+	assert_ptr_not_null(p1, "Unexpected mallocx() failure");
+	assert_ptr_eq(p0, p1, "Expected tcache to allocate cached region");
+
+	/* Clean up. */
+	dallocx(p1, MALLOCX_TCACHE_NONE);
+}
+TEST_END
+
+TEST_BEGIN(test_tcache) {
+#define NTCACHES	10
+	unsigned tis[NTCACHES];
+	void *ps[NTCACHES];
+	void *qs[NTCACHES];
+	unsigned i;
+	size_t sz, psz, qsz;
+
+	psz = 42;
+	qsz = nallocx(psz, 0) + 1;
+
+	/* Create tcaches. */
+	for (i = 0; i < NTCACHES; i++) {
+		sz = sizeof(unsigned);
+		assert_d_eq(mallctl("tcache.create", (void *)&tis[i], &sz, NULL,
+		    0), 0, "Unexpected mallctl() failure, i=%u", i);
+	}
+
+	/* Exercise tcache ID recycling. */
+	for (i = 0; i < NTCACHES; i++) {
+		assert_d_eq(mallctl("tcache.destroy", NULL, NULL,
+		    (void *)&tis[i], sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure, i=%u", i);
+	}
+	for (i = 0; i < NTCACHES; i++) {
+		sz = sizeof(unsigned);
+		assert_d_eq(mallctl("tcache.create", (void *)&tis[i], &sz, NULL,
+		    0), 0, "Unexpected mallctl() failure, i=%u", i);
+	}
+
+	/* Flush empty tcaches. */
+	for (i = 0; i < NTCACHES; i++) {
+		assert_d_eq(mallctl("tcache.flush", NULL, NULL, (void *)&tis[i],
+		    sizeof(unsigned)), 0, "Unexpected mallctl() failure, i=%u",
+		    i);
+	}
+
+	/* Cache some allocations. */
+	for (i = 0; i < NTCACHES; i++) {
+		ps[i] = mallocx(psz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(ps[i], "Unexpected mallocx() failure, i=%u",
+		    i);
+		dallocx(ps[i], MALLOCX_TCACHE(tis[i]));
+
+		qs[i] = mallocx(qsz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(qs[i], "Unexpected mallocx() failure, i=%u",
+		    i);
+		dallocx(qs[i], MALLOCX_TCACHE(tis[i]));
+	}
+
+	/* Verify that tcaches allocate cached regions. */
+	for (i = 0; i < NTCACHES; i++) {
+		void *p0 = ps[i];
+		ps[i] = mallocx(psz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(ps[i], "Unexpected mallocx() failure, i=%u",
+		    i);
+		assert_ptr_eq(ps[i], p0,
+		    "Expected mallocx() to allocate cached region, i=%u", i);
+	}
+
+	/* Verify that reallocation uses cached regions. */
+	for (i = 0; i < NTCACHES; i++) {
+		void *q0 = qs[i];
+		qs[i] = rallocx(ps[i], qsz, MALLOCX_TCACHE(tis[i]));
+		assert_ptr_not_null(qs[i], "Unexpected rallocx() failure, i=%u",
+		    i);
+		assert_ptr_eq(qs[i], q0,
+		    "Expected rallocx() to allocate cached region, i=%u", i);
+		/* Avoid undefined behavior in case of test failure. */
+		if (qs[i] == NULL) {
+			qs[i] = ps[i];
+		}
+	}
+	for (i = 0; i < NTCACHES; i++) {
+		dallocx(qs[i], MALLOCX_TCACHE(tis[i]));
+	}
+
+	/* Flush some non-empty tcaches. */
+	for (i = 0; i < NTCACHES/2; i++) {
+		assert_d_eq(mallctl("tcache.flush", NULL, NULL, (void *)&tis[i],
+		    sizeof(unsigned)), 0, "Unexpected mallctl() failure, i=%u",
+		    i);
+	}
+
+	/* Destroy tcaches. */
+	for (i = 0; i < NTCACHES; i++) {
+		assert_d_eq(mallctl("tcache.destroy", NULL, NULL,
+		    (void *)&tis[i], sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure, i=%u", i);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_thread_arena) {
+	unsigned old_arena_ind, new_arena_ind, narenas;
+
+	const char *opa;
+	size_t sz = sizeof(opa);
+	assert_d_eq(mallctl("opt.percpu_arena", (void *)&opa, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	if (opt_oversize_threshold != 0) {
+		narenas--;
+	}
+	assert_u_eq(narenas, opt_narenas, "Number of arenas incorrect");
+
+	if (strcmp(opa, "disabled") == 0) {
+		new_arena_ind = narenas - 1;
+		assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+		    (void *)&new_arena_ind, sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure");
+		new_arena_ind = 0;
+		assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+		    (void *)&new_arena_ind, sizeof(unsigned)), 0,
+		    "Unexpected mallctl() failure");
+	} else {
+		assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+		    NULL, 0), 0, "Unexpected mallctl() failure");
+		new_arena_ind = percpu_arena_ind_limit(opt_percpu_arena) - 1;
+		if (old_arena_ind != new_arena_ind) {
+			assert_d_eq(mallctl("thread.arena",
+			    (void *)&old_arena_ind, &sz, (void *)&new_arena_ind,
+			    sizeof(unsigned)), EPERM, "thread.arena ctl "
+			    "should not be allowed with percpu arena");
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_initialized) {
+	unsigned narenas, i;
+	size_t sz;
+	size_t mib[3];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+	bool initialized;
+
+	sz = sizeof(narenas);
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctlnametomib("arena.0.initialized", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	for (i = 0; i < narenas; i++) {
+		mib[1] = i;
+		sz = sizeof(initialized);
+		assert_d_eq(mallctlbymib(mib, miblen, &initialized, &sz, NULL,
+		    0), 0, "Unexpected mallctl() failure");
+	}
+
+	mib[1] = MALLCTL_ARENAS_ALL;
+	sz = sizeof(initialized);
+	assert_d_eq(mallctlbymib(mib, miblen, &initialized, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_true(initialized,
+	    "Merged arena statistics should always be initialized");
+
+	/* Equivalent to the above but using mallctl() directly. */
+	sz = sizeof(initialized);
+	assert_d_eq(mallctl(
+	    "arena." STRINGIFY(MALLCTL_ARENAS_ALL) ".initialized",
+	    (void *)&initialized, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_true(initialized,
+	    "Merged arena statistics should always be initialized");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_dirty_decay_ms) {
+	ssize_t dirty_decay_ms, orig_dirty_decay_ms, prev_dirty_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arena.0.dirty_decay_ms",
+	    (void *)&orig_dirty_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	dirty_decay_ms = -2;
+	assert_d_eq(mallctl("arena.0.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	dirty_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arena.0.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), 0,
+	    "Unexpected mallctl() failure");
+
+	for (prev_dirty_decay_ms = dirty_decay_ms, dirty_decay_ms = -1;
+	    dirty_decay_ms < 20; prev_dirty_decay_ms = dirty_decay_ms,
+	    dirty_decay_ms++) {
+		ssize_t old_dirty_decay_ms;
+
+		assert_d_eq(mallctl("arena.0.dirty_decay_ms",
+		    (void *)&old_dirty_decay_ms, &sz, (void *)&dirty_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_dirty_decay_ms, prev_dirty_decay_ms,
+		    "Unexpected old arena.0.dirty_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_muzzy_decay_ms) {
+	ssize_t muzzy_decay_ms, orig_muzzy_decay_ms, prev_muzzy_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arena.0.muzzy_decay_ms",
+	    (void *)&orig_muzzy_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	muzzy_decay_ms = -2;
+	assert_d_eq(mallctl("arena.0.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	muzzy_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arena.0.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), 0,
+	    "Unexpected mallctl() failure");
+
+	for (prev_muzzy_decay_ms = muzzy_decay_ms, muzzy_decay_ms = -1;
+	    muzzy_decay_ms < 20; prev_muzzy_decay_ms = muzzy_decay_ms,
+	    muzzy_decay_ms++) {
+		ssize_t old_muzzy_decay_ms;
+
+		assert_d_eq(mallctl("arena.0.muzzy_decay_ms",
+		    (void *)&old_muzzy_decay_ms, &sz, (void *)&muzzy_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_muzzy_decay_ms, prev_muzzy_decay_ms,
+		    "Unexpected old arena.0.muzzy_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_purge) {
+	unsigned narenas;
+	size_t sz = sizeof(unsigned);
+	size_t mib[3];
+	size_t miblen = 3;
+
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlnametomib("arena.0.purge", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = narenas;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	mib[1] = MALLCTL_ARENAS_ALL;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_decay) {
+	unsigned narenas;
+	size_t sz = sizeof(unsigned);
+	size_t mib[3];
+	size_t miblen = 3;
+
+	assert_d_eq(mallctl("arena.0.decay", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlnametomib("arena.0.decay", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = narenas;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+
+	mib[1] = MALLCTL_ARENAS_ALL;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_dss) {
+	const char *dss_prec_old, *dss_prec_new;
+	size_t sz = sizeof(dss_prec_old);
+	size_t mib[3];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.dss", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() error");
+
+	dss_prec_new = "disabled";
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz,
+	    (void *)&dss_prec_new, sizeof(dss_prec_new)), 0,
+	    "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected default for dss precedence");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_new, &sz,
+	    (void *)&dss_prec_old, sizeof(dss_prec_old)), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected value for dss precedence");
+
+	mib[1] = narenas_total_get();
+	dss_prec_new = "disabled";
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz,
+	    (void *)&dss_prec_new, sizeof(dss_prec_new)), 0,
+	    "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected default for dss precedence");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_new, &sz,
+	    (void *)&dss_prec_old, sizeof(dss_prec_new)), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&dss_prec_old, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+	assert_str_ne(dss_prec_old, "primary",
+	    "Unexpected value for dss precedence");
+}
+TEST_END
+
+TEST_BEGIN(test_arena_i_retain_grow_limit) {
+	size_t old_limit, new_limit, default_limit;
+	size_t mib[3];
+	size_t miblen;
+
+	bool retain_enabled;
+	size_t sz = sizeof(retain_enabled);
+	assert_d_eq(mallctl("opt.retain", &retain_enabled, &sz, NULL, 0),
+	    0, "Unexpected mallctl() failure");
+	test_skip_if(!retain_enabled);
+
+	sz = sizeof(default_limit);
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.retain_grow_limit", mib, &miblen),
+	    0, "Unexpected mallctlnametomib() error");
+
+	assert_d_eq(mallctlbymib(mib, miblen, &default_limit, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(default_limit, SC_LARGE_MAXCLASS,
+	    "Unexpected default for retain_grow_limit");
+
+	new_limit = PAGE - 1;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &new_limit,
+	    sizeof(new_limit)), EFAULT, "Unexpected mallctl() success");
+
+	new_limit = PAGE + 1;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &new_limit,
+	    sizeof(new_limit)), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlbymib(mib, miblen, &old_limit, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(old_limit, PAGE,
+	    "Unexpected value for retain_grow_limit");
+
+	/* Expect grow less than psize class 10. */
+	new_limit = sz_pind2sz(10) - 1;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &new_limit,
+	    sizeof(new_limit)), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctlbymib(mib, miblen, &old_limit, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_zu_eq(old_limit, sz_pind2sz(9),
+	    "Unexpected value for retain_grow_limit");
+
+	/* Restore to default. */
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, &default_limit,
+	    sizeof(default_limit)), 0, "Unexpected mallctl() failure");
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_dirty_decay_ms) {
+	ssize_t dirty_decay_ms, orig_dirty_decay_ms, prev_dirty_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arenas.dirty_decay_ms",
+	    (void *)&orig_dirty_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	dirty_decay_ms = -2;
+	assert_d_eq(mallctl("arenas.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	dirty_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arenas.dirty_decay_ms", NULL, NULL,
+	    (void *)&dirty_decay_ms, sizeof(ssize_t)), 0,
+	    "Expected mallctl() failure");
+
+	for (prev_dirty_decay_ms = dirty_decay_ms, dirty_decay_ms = -1;
+	    dirty_decay_ms < 20; prev_dirty_decay_ms = dirty_decay_ms,
+	    dirty_decay_ms++) {
+		ssize_t old_dirty_decay_ms;
+
+		assert_d_eq(mallctl("arenas.dirty_decay_ms",
+		    (void *)&old_dirty_decay_ms, &sz, (void *)&dirty_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_dirty_decay_ms, prev_dirty_decay_ms,
+		    "Unexpected old arenas.dirty_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_muzzy_decay_ms) {
+	ssize_t muzzy_decay_ms, orig_muzzy_decay_ms, prev_muzzy_decay_ms;
+	size_t sz = sizeof(ssize_t);
+
+	assert_d_eq(mallctl("arenas.muzzy_decay_ms",
+	    (void *)&orig_muzzy_decay_ms, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	muzzy_decay_ms = -2;
+	assert_d_eq(mallctl("arenas.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), EFAULT,
+	    "Unexpected mallctl() success");
+
+	muzzy_decay_ms = 0x7fffffff;
+	assert_d_eq(mallctl("arenas.muzzy_decay_ms", NULL, NULL,
+	    (void *)&muzzy_decay_ms, sizeof(ssize_t)), 0,
+	    "Expected mallctl() failure");
+
+	for (prev_muzzy_decay_ms = muzzy_decay_ms, muzzy_decay_ms = -1;
+	    muzzy_decay_ms < 20; prev_muzzy_decay_ms = muzzy_decay_ms,
+	    muzzy_decay_ms++) {
+		ssize_t old_muzzy_decay_ms;
+
+		assert_d_eq(mallctl("arenas.muzzy_decay_ms",
+		    (void *)&old_muzzy_decay_ms, &sz, (void *)&muzzy_decay_ms,
+		    sizeof(ssize_t)), 0, "Unexpected mallctl() failure");
+		assert_zd_eq(old_muzzy_decay_ms, prev_muzzy_decay_ms,
+		    "Unexpected old arenas.muzzy_decay_ms");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_constants) {
+#define TEST_ARENAS_CONSTANT(t, name, expected) do {			\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("arenas."#name, (void *)&name, &sz, NULL,	\
+	    0), 0, "Unexpected mallctl() failure");			\
+	assert_zu_eq(name, expected, "Incorrect "#name" size");		\
+} while (0)
+
+	TEST_ARENAS_CONSTANT(size_t, quantum, QUANTUM);
+	TEST_ARENAS_CONSTANT(size_t, page, PAGE);
+	TEST_ARENAS_CONSTANT(unsigned, nbins, SC_NBINS);
+	TEST_ARENAS_CONSTANT(unsigned, nlextents, SC_NSIZES - SC_NBINS);
+
+#undef TEST_ARENAS_CONSTANT
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_bin_constants) {
+#define TEST_ARENAS_BIN_CONSTANT(t, name, expected) do {		\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("arenas.bin.0."#name, (void *)&name, &sz,	\
+	    NULL, 0), 0, "Unexpected mallctl() failure");		\
+	assert_zu_eq(name, expected, "Incorrect "#name" size");		\
+} while (0)
+
+	TEST_ARENAS_BIN_CONSTANT(size_t, size, bin_infos[0].reg_size);
+	TEST_ARENAS_BIN_CONSTANT(uint32_t, nregs, bin_infos[0].nregs);
+	TEST_ARENAS_BIN_CONSTANT(size_t, slab_size,
+	    bin_infos[0].slab_size);
+	TEST_ARENAS_BIN_CONSTANT(uint32_t, nshards, bin_infos[0].n_shards);
+
+#undef TEST_ARENAS_BIN_CONSTANT
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_lextent_constants) {
+#define TEST_ARENAS_LEXTENT_CONSTANT(t, name, expected) do {		\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("arenas.lextent.0."#name, (void *)&name,	\
+	    &sz, NULL, 0), 0, "Unexpected mallctl() failure");		\
+	assert_zu_eq(name, expected, "Incorrect "#name" size");		\
+} while (0)
+
+	TEST_ARENAS_LEXTENT_CONSTANT(size_t, size,
+	    SC_LARGE_MINCLASS);
+
+#undef TEST_ARENAS_LEXTENT_CONSTANT
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_create) {
+	unsigned narenas_before, arena, narenas_after;
+	size_t sz = sizeof(unsigned);
+
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas_before, &sz,
+	    NULL, 0), 0, "Unexpected mallctl() failure");
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	assert_d_eq(mallctl("arenas.narenas", (void *)&narenas_after, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+
+	assert_u_eq(narenas_before+1, narenas_after,
+	    "Unexpected number of arenas before versus after extension");
+	assert_u_eq(arena, narenas_after-1, "Unexpected arena index");
+}
+TEST_END
+
+TEST_BEGIN(test_arenas_lookup) {
+	unsigned arena, arena1;
+	void *ptr;
+	size_t sz = sizeof(unsigned);
+
+	assert_d_eq(mallctl("arenas.create", (void *)&arena, &sz, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+	ptr = mallocx(42, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE);
+	assert_ptr_not_null(ptr, "Unexpected mallocx() failure");
+	assert_d_eq(mallctl("arenas.lookup", &arena1, &sz, &ptr, sizeof(ptr)),
+	    0, "Unexpected mallctl() failure");
+	assert_u_eq(arena, arena1, "Unexpected arena index");
+	dallocx(ptr, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas) {
+#define TEST_STATS_ARENAS(t, name) do {					\
+	t name;								\
+	size_t sz = sizeof(t);						\
+	assert_d_eq(mallctl("stats.arenas.0."#name, (void *)&name, &sz,	\
+	    NULL, 0), 0, "Unexpected mallctl() failure");		\
+} while (0)
+
+	TEST_STATS_ARENAS(unsigned, nthreads);
+	TEST_STATS_ARENAS(const char *, dss);
+	TEST_STATS_ARENAS(ssize_t, dirty_decay_ms);
+	TEST_STATS_ARENAS(ssize_t, muzzy_decay_ms);
+	TEST_STATS_ARENAS(size_t, pactive);
+	TEST_STATS_ARENAS(size_t, pdirty);
+
+#undef TEST_STATS_ARENAS
+}
+TEST_END
+
+static void
+alloc_hook(void *extra, UNUSED hook_alloc_t type, UNUSED void *result,
+    UNUSED uintptr_t result_raw, UNUSED uintptr_t args_raw[3]) {
+	*(bool *)extra = true;
+}
+
+static void
+dalloc_hook(void *extra, UNUSED hook_dalloc_t type,
+    UNUSED void *address, UNUSED uintptr_t args_raw[3]) {
+	*(bool *)extra = true;
+}
+
+TEST_BEGIN(test_hooks) {
+	bool hook_called = false;
+	hooks_t hooks = {&alloc_hook, &dalloc_hook, NULL, &hook_called};
+	void *handle = NULL;
+	size_t sz = sizeof(handle);
+	int err = mallctl("experimental.hooks.install", &handle, &sz, &hooks,
+	    sizeof(hooks));
+	assert_d_eq(err, 0, "Hook installation failed");
+	assert_ptr_ne(handle, NULL, "Hook installation gave null handle");
+	void *ptr = mallocx(1, 0);
+	assert_true(hook_called, "Alloc hook not called");
+	hook_called = false;
+	free(ptr);
+	assert_true(hook_called, "Free hook not called");
+
+	err = mallctl("experimental.hooks.remove", NULL, NULL, &handle,
+	    sizeof(handle));
+	assert_d_eq(err, 0, "Hook removal failed");
+	hook_called = false;
+	ptr = mallocx(1, 0);
+	free(ptr);
+	assert_false(hook_called, "Hook called after removal");
+}
+TEST_END
+
+TEST_BEGIN(test_hooks_exhaustion) {
+	bool hook_called = false;
+	hooks_t hooks = {&alloc_hook, &dalloc_hook, NULL, &hook_called};
+
+	void *handle;
+	void *handles[HOOK_MAX];
+	size_t sz = sizeof(handle);
+	int err;
+	for (int i = 0; i < HOOK_MAX; i++) {
+		handle = NULL;
+		err = mallctl("experimental.hooks.install", &handle, &sz,
+		    &hooks, sizeof(hooks));
+		assert_d_eq(err, 0, "Error installation hooks");
+		assert_ptr_ne(handle, NULL, "Got NULL handle");
+		handles[i] = handle;
+	}
+	err = mallctl("experimental.hooks.install", &handle, &sz, &hooks,
+	    sizeof(hooks));
+	assert_d_eq(err, EAGAIN, "Should have failed hook installation");
+	for (int i = 0; i < HOOK_MAX; i++) {
+		err = mallctl("experimental.hooks.remove", NULL, NULL,
+		    &handles[i], sizeof(handles[i]));
+		assert_d_eq(err, 0, "Hook removal failed");
+	}
+	/* Insertion failed, but then we removed some; it should work now. */
+	handle = NULL;
+	err = mallctl("experimental.hooks.install", &handle, &sz, &hooks,
+	    sizeof(hooks));
+	assert_d_eq(err, 0, "Hook insertion failed");
+	assert_ptr_ne(handle, NULL, "Got NULL handle");
+	err = mallctl("experimental.hooks.remove", NULL, NULL, &handle,
+	    sizeof(handle));
+	assert_d_eq(err, 0, "Hook removal failed");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_mallctl_errors,
+	    test_mallctlnametomib_errors,
+	    test_mallctlbymib_errors,
+	    test_mallctl_read_write,
+	    test_mallctlnametomib_short_mib,
+	    test_mallctl_config,
+	    test_mallctl_opt,
+	    test_manpage_example,
+	    test_tcache_none,
+	    test_tcache,
+	    test_thread_arena,
+	    test_arena_i_initialized,
+	    test_arena_i_dirty_decay_ms,
+	    test_arena_i_muzzy_decay_ms,
+	    test_arena_i_purge,
+	    test_arena_i_decay,
+	    test_arena_i_dss,
+	    test_arena_i_retain_grow_limit,
+	    test_arenas_dirty_decay_ms,
+	    test_arenas_muzzy_decay_ms,
+	    test_arenas_constants,
+	    test_arenas_bin_constants,
+	    test_arenas_lextent_constants,
+	    test_arenas_create,
+	    test_arenas_lookup,
+	    test_stats_arenas,
+	    test_hooks,
+	    test_hooks_exhaustion);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/malloc_io.c b/memkind-1.10.0/jemalloc/test/unit/malloc_io.c
new file mode 100644
index 0000000..79ba7fc
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/malloc_io.c
@@ -0,0 +1,258 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_malloc_strtoumax_no_endptr) {
+	int err;
+
+	set_errno(0);
+	assert_ju_eq(malloc_strtoumax("0", NULL, 0), 0, "Unexpected result");
+	err = get_errno();
+	assert_d_eq(err, 0, "Unexpected failure");
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_strtoumax) {
+	struct test_s {
+		const char *input;
+		const char *expected_remainder;
+		int base;
+		int expected_errno;
+		const char *expected_errno_name;
+		uintmax_t expected_x;
+	};
+#define ERR(e)		e, #e
+#define KUMAX(x)	((uintmax_t)x##ULL)
+#define KSMAX(x)	((uintmax_t)(intmax_t)x##LL)
+	struct test_s tests[] = {
+		{"0",		"0",	-1,	ERR(EINVAL),	UINTMAX_MAX},
+		{"0",		"0",	1,	ERR(EINVAL),	UINTMAX_MAX},
+		{"0",		"0",	37,	ERR(EINVAL),	UINTMAX_MAX},
+
+		{"",		"",	0,	ERR(EINVAL),	UINTMAX_MAX},
+		{"+",		"+",	0,	ERR(EINVAL),	UINTMAX_MAX},
+		{"++3",		"++3",	0,	ERR(EINVAL),	UINTMAX_MAX},
+		{"-",		"-",	0,	ERR(EINVAL),	UINTMAX_MAX},
+
+		{"42",		"",	0,	ERR(0),		KUMAX(42)},
+		{"+42",		"",	0,	ERR(0),		KUMAX(42)},
+		{"-42",		"",	0,	ERR(0),		KSMAX(-42)},
+		{"042",		"",	0,	ERR(0),		KUMAX(042)},
+		{"+042",	"",	0,	ERR(0),		KUMAX(042)},
+		{"-042",	"",	0,	ERR(0),		KSMAX(-042)},
+		{"0x42",	"",	0,	ERR(0),		KUMAX(0x42)},
+		{"+0x42",	"",	0,	ERR(0),		KUMAX(0x42)},
+		{"-0x42",	"",	0,	ERR(0),		KSMAX(-0x42)},
+
+		{"0",		"",	0,	ERR(0),		KUMAX(0)},
+		{"1",		"",	0,	ERR(0),		KUMAX(1)},
+
+		{"42",		"",	0,	ERR(0),		KUMAX(42)},
+		{" 42",		"",	0,	ERR(0),		KUMAX(42)},
+		{"42 ",		" ",	0,	ERR(0),		KUMAX(42)},
+		{"0x",		"x",	0,	ERR(0),		KUMAX(0)},
+		{"42x",		"x",	0,	ERR(0),		KUMAX(42)},
+
+		{"07",		"",	0,	ERR(0),		KUMAX(7)},
+		{"010",		"",	0,	ERR(0),		KUMAX(8)},
+		{"08",		"8",	0,	ERR(0),		KUMAX(0)},
+		{"0_",		"_",	0,	ERR(0),		KUMAX(0)},
+
+		{"0x",		"x",	0,	ERR(0),		KUMAX(0)},
+		{"0X",		"X",	0,	ERR(0),		KUMAX(0)},
+		{"0xg",		"xg",	0,	ERR(0),		KUMAX(0)},
+		{"0XA",		"",	0,	ERR(0),		KUMAX(10)},
+
+		{"010",		"",	10,	ERR(0),		KUMAX(10)},
+		{"0x3",		"x3",	10,	ERR(0),		KUMAX(0)},
+
+		{"12",		"2",	2,	ERR(0),		KUMAX(1)},
+		{"78",		"8",	8,	ERR(0),		KUMAX(7)},
+		{"9a",		"a",	10,	ERR(0),		KUMAX(9)},
+		{"9A",		"A",	10,	ERR(0),		KUMAX(9)},
+		{"fg",		"g",	16,	ERR(0),		KUMAX(15)},
+		{"FG",		"G",	16,	ERR(0),		KUMAX(15)},
+		{"0xfg",	"g",	16,	ERR(0),		KUMAX(15)},
+		{"0XFG",	"G",	16,	ERR(0),		KUMAX(15)},
+		{"z_",		"_",	36,	ERR(0),		KUMAX(35)},
+		{"Z_",		"_",	36,	ERR(0),		KUMAX(35)}
+	};
+#undef ERR
+#undef KUMAX
+#undef KSMAX
+	unsigned i;
+
+	for (i = 0; i < sizeof(tests)/sizeof(struct test_s); i++) {
+		struct test_s *test = &tests[i];
+		int err;
+		uintmax_t result;
+		char *remainder;
+
+		set_errno(0);
+		result = malloc_strtoumax(test->input, &remainder, test->base);
+		err = get_errno();
+		assert_d_eq(err, test->expected_errno,
+		    "Expected errno %s for \"%s\", base %d",
+		    test->expected_errno_name, test->input, test->base);
+		assert_str_eq(remainder, test->expected_remainder,
+		    "Unexpected remainder for \"%s\", base %d",
+		    test->input, test->base);
+		if (err == 0) {
+			assert_ju_eq(result, test->expected_x,
+			    "Unexpected result for \"%s\", base %d",
+			    test->input, test->base);
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_snprintf_truncated) {
+#define BUFLEN	15
+	char buf[BUFLEN];
+	size_t result;
+	size_t len;
+#define TEST(expected_str_untruncated, ...) do {			\
+	result = malloc_snprintf(buf, len, __VA_ARGS__);		\
+	assert_d_eq(strncmp(buf, expected_str_untruncated, len-1), 0,	\
+	    "Unexpected string inequality (\"%s\" vs \"%s\")",		\
+	    buf, expected_str_untruncated);				\
+	assert_zu_eq(result, strlen(expected_str_untruncated),		\
+	    "Unexpected result");					\
+} while (0)
+
+	for (len = 1; len < BUFLEN; len++) {
+		TEST("012346789",	"012346789");
+		TEST("a0123b",		"a%sb", "0123");
+		TEST("a01234567",	"a%s%s", "0123", "4567");
+		TEST("a0123  ",		"a%-6s", "0123");
+		TEST("a  0123",		"a%6s", "0123");
+		TEST("a   012",		"a%6.3s", "0123");
+		TEST("a   012",		"a%*.*s", 6, 3, "0123");
+		TEST("a 123b",		"a% db", 123);
+		TEST("a123b",		"a%-db", 123);
+		TEST("a-123b",		"a%-db", -123);
+		TEST("a+123b",		"a%+db", 123);
+	}
+#undef BUFLEN
+#undef TEST
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_snprintf) {
+#define BUFLEN	128
+	char buf[BUFLEN];
+	size_t result;
+#define TEST(expected_str, ...) do {					\
+	result = malloc_snprintf(buf, sizeof(buf), __VA_ARGS__);	\
+	assert_str_eq(buf, expected_str, "Unexpected output");		\
+	assert_zu_eq(result, strlen(expected_str), "Unexpected result");\
+} while (0)
+
+	TEST("hello", "hello");
+
+	TEST("50%, 100%", "50%%, %d%%", 100);
+
+	TEST("a0123b", "a%sb", "0123");
+
+	TEST("a 0123b", "a%5sb", "0123");
+	TEST("a 0123b", "a%*sb", 5, "0123");
+
+	TEST("a0123 b", "a%-5sb", "0123");
+	TEST("a0123b", "a%*sb", -1, "0123");
+	TEST("a0123 b", "a%*sb", -5, "0123");
+	TEST("a0123 b", "a%-*sb", -5, "0123");
+
+	TEST("a012b", "a%.3sb", "0123");
+	TEST("a012b", "a%.*sb", 3, "0123");
+	TEST("a0123b", "a%.*sb", -3, "0123");
+
+	TEST("a  012b", "a%5.3sb", "0123");
+	TEST("a  012b", "a%5.*sb", 3, "0123");
+	TEST("a  012b", "a%*.3sb", 5, "0123");
+	TEST("a  012b", "a%*.*sb", 5, 3, "0123");
+	TEST("a 0123b", "a%*.*sb", 5, -3, "0123");
+
+	TEST("_abcd_", "_%x_", 0xabcd);
+	TEST("_0xabcd_", "_%#x_", 0xabcd);
+	TEST("_1234_", "_%o_", 01234);
+	TEST("_01234_", "_%#o_", 01234);
+	TEST("_1234_", "_%u_", 1234);
+
+	TEST("_1234_", "_%d_", 1234);
+	TEST("_ 1234_", "_% d_", 1234);
+	TEST("_+1234_", "_%+d_", 1234);
+	TEST("_-1234_", "_%d_", -1234);
+	TEST("_-1234_", "_% d_", -1234);
+	TEST("_-1234_", "_%+d_", -1234);
+
+	TEST("_-1234_", "_%d_", -1234);
+	TEST("_1234_", "_%d_", 1234);
+	TEST("_-1234_", "_%i_", -1234);
+	TEST("_1234_", "_%i_", 1234);
+	TEST("_01234_", "_%#o_", 01234);
+	TEST("_1234_", "_%u_", 1234);
+	TEST("_0x1234abc_", "_%#x_", 0x1234abc);
+	TEST("_0X1234ABC_", "_%#X_", 0x1234abc);
+	TEST("_c_", "_%c_", 'c');
+	TEST("_string_", "_%s_", "string");
+	TEST("_0x42_", "_%p_", ((void *)0x42));
+
+	TEST("_-1234_", "_%ld_", ((long)-1234));
+	TEST("_1234_", "_%ld_", ((long)1234));
+	TEST("_-1234_", "_%li_", ((long)-1234));
+	TEST("_1234_", "_%li_", ((long)1234));
+	TEST("_01234_", "_%#lo_", ((long)01234));
+	TEST("_1234_", "_%lu_", ((long)1234));
+	TEST("_0x1234abc_", "_%#lx_", ((long)0x1234abc));
+	TEST("_0X1234ABC_", "_%#lX_", ((long)0x1234ABC));
+
+	TEST("_-1234_", "_%lld_", ((long long)-1234));
+	TEST("_1234_", "_%lld_", ((long long)1234));
+	TEST("_-1234_", "_%lli_", ((long long)-1234));
+	TEST("_1234_", "_%lli_", ((long long)1234));
+	TEST("_01234_", "_%#llo_", ((long long)01234));
+	TEST("_1234_", "_%llu_", ((long long)1234));
+	TEST("_0x1234abc_", "_%#llx_", ((long long)0x1234abc));
+	TEST("_0X1234ABC_", "_%#llX_", ((long long)0x1234ABC));
+
+	TEST("_-1234_", "_%qd_", ((long long)-1234));
+	TEST("_1234_", "_%qd_", ((long long)1234));
+	TEST("_-1234_", "_%qi_", ((long long)-1234));
+	TEST("_1234_", "_%qi_", ((long long)1234));
+	TEST("_01234_", "_%#qo_", ((long long)01234));
+	TEST("_1234_", "_%qu_", ((long long)1234));
+	TEST("_0x1234abc_", "_%#qx_", ((long long)0x1234abc));
+	TEST("_0X1234ABC_", "_%#qX_", ((long long)0x1234ABC));
+
+	TEST("_-1234_", "_%jd_", ((intmax_t)-1234));
+	TEST("_1234_", "_%jd_", ((intmax_t)1234));
+	TEST("_-1234_", "_%ji_", ((intmax_t)-1234));
+	TEST("_1234_", "_%ji_", ((intmax_t)1234));
+	TEST("_01234_", "_%#jo_", ((intmax_t)01234));
+	TEST("_1234_", "_%ju_", ((intmax_t)1234));
+	TEST("_0x1234abc_", "_%#jx_", ((intmax_t)0x1234abc));
+	TEST("_0X1234ABC_", "_%#jX_", ((intmax_t)0x1234ABC));
+
+	TEST("_1234_", "_%td_", ((ptrdiff_t)1234));
+	TEST("_-1234_", "_%td_", ((ptrdiff_t)-1234));
+	TEST("_1234_", "_%ti_", ((ptrdiff_t)1234));
+	TEST("_-1234_", "_%ti_", ((ptrdiff_t)-1234));
+
+	TEST("_-1234_", "_%zd_", ((ssize_t)-1234));
+	TEST("_1234_", "_%zd_", ((ssize_t)1234));
+	TEST("_-1234_", "_%zi_", ((ssize_t)-1234));
+	TEST("_1234_", "_%zi_", ((ssize_t)1234));
+	TEST("_01234_", "_%#zo_", ((ssize_t)01234));
+	TEST("_1234_", "_%zu_", ((ssize_t)1234));
+	TEST("_0x1234abc_", "_%#zx_", ((ssize_t)0x1234abc));
+	TEST("_0X1234ABC_", "_%#zX_", ((ssize_t)0x1234ABC));
+#undef BUFLEN
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_malloc_strtoumax_no_endptr,
+	    test_malloc_strtoumax,
+	    test_malloc_snprintf_truncated,
+	    test_malloc_snprintf);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/math.c b/memkind-1.10.0/jemalloc/test/unit/math.c
new file mode 100644
index 0000000..09ef20c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/math.c
@@ -0,0 +1,390 @@
+#include "test/jemalloc_test.h"
+
+#define MAX_REL_ERR 1.0e-9
+#define MAX_ABS_ERR 1.0e-9
+
+#include <float.h>
+
+#ifdef __PGI
+#undef INFINITY
+#endif
+
+#ifndef INFINITY
+#define INFINITY (DBL_MAX + DBL_MAX)
+#endif
+
+static bool
+double_eq_rel(double a, double b, double max_rel_err, double max_abs_err) {
+	double rel_err;
+
+	if (fabs(a - b) < max_abs_err) {
+		return true;
+	}
+	rel_err = (fabs(b) > fabs(a)) ? fabs((a-b)/b) : fabs((a-b)/a);
+	return (rel_err < max_rel_err);
+}
+
+static uint64_t
+factorial(unsigned x) {
+	uint64_t ret = 1;
+	unsigned i;
+
+	for (i = 2; i <= x; i++) {
+		ret *= (uint64_t)i;
+	}
+
+	return ret;
+}
+
+TEST_BEGIN(test_ln_gamma_factorial) {
+	unsigned x;
+
+	/* exp(ln_gamma(x)) == (x-1)! for integer x. */
+	for (x = 1; x <= 21; x++) {
+		assert_true(double_eq_rel(exp(ln_gamma(x)),
+		    (double)factorial(x-1), MAX_REL_ERR, MAX_ABS_ERR),
+		    "Incorrect factorial result for x=%u", x);
+	}
+}
+TEST_END
+
+/* Expected ln_gamma([0.0..100.0] increment=0.25). */
+static const double ln_gamma_misc_expected[] = {
+	INFINITY,
+	1.28802252469807743, 0.57236494292470008, 0.20328095143129538,
+	0.00000000000000000, -0.09827183642181320, -0.12078223763524518,
+	-0.08440112102048555, 0.00000000000000000, 0.12487171489239651,
+	0.28468287047291918, 0.47521466691493719, 0.69314718055994529,
+	0.93580193110872523, 1.20097360234707429, 1.48681557859341718,
+	1.79175946922805496, 2.11445692745037128, 2.45373657084244234,
+	2.80857141857573644, 3.17805383034794575, 3.56137591038669710,
+	3.95781396761871651, 4.36671603662228680, 4.78749174278204581,
+	5.21960398699022932, 5.66256205985714178, 6.11591589143154568,
+	6.57925121201010121, 7.05218545073853953, 7.53436423675873268,
+	8.02545839631598312, 8.52516136106541467, 9.03318691960512332,
+	9.54926725730099690, 10.07315123968123949, 10.60460290274525086,
+	11.14340011995171231, 11.68933342079726856, 12.24220494005076176,
+	12.80182748008146909, 13.36802367147604720, 13.94062521940376342,
+	14.51947222506051816, 15.10441257307551943, 15.69530137706046524,
+	16.29200047656724237, 16.89437797963419285, 17.50230784587389010,
+	18.11566950571089407, 18.73434751193644843, 19.35823122022435427,
+	19.98721449566188468, 20.62119544270163018, 21.26007615624470048,
+	21.90376249182879320, 22.55216385312342098, 23.20519299513386002,
+	23.86276584168908954, 24.52480131594137802, 25.19122118273868338,
+	25.86194990184851861, 26.53691449111561340, 27.21604439872720604,
+	27.89927138384089389, 28.58652940490193828, 29.27775451504081516,
+	29.97288476399884871, 30.67186010608067548, 31.37462231367769050,
+	32.08111489594735843, 32.79128302226991565, 33.50507345013689076,
+	34.22243445715505317, 34.94331577687681545, 35.66766853819134298,
+	36.39544520803305261, 37.12659953718355865, 37.86108650896109395,
+	38.59886229060776230, 39.33988418719949465, 40.08411059791735198,
+	40.83150097453079752, 41.58201578195490100, 42.33561646075348506,
+	43.09226539146988699, 43.85192586067515208, 44.61456202863158893,
+	45.38013889847690052, 46.14862228684032885, 46.91997879580877395,
+	47.69417578616628361, 48.47118135183522014, 49.25096429545256882,
+	50.03349410501914463, 50.81874093156324790, 51.60667556776436982,
+	52.39726942748592364, 53.19049452616926743, 53.98632346204390586,
+	54.78472939811231157, 55.58568604486942633, 56.38916764371992940,
+	57.19514895105859864, 58.00360522298051080, 58.81451220059079787,
+	59.62784609588432261, 60.44358357816834371, 61.26170176100199427,
+	62.08217818962842927, 62.90499082887649962, 63.73011805151035958,
+	64.55753862700632340, 65.38723171073768015, 66.21917683354901385,
+	67.05335389170279825, 67.88974313718154008, 68.72832516833013017,
+	69.56908092082363737, 70.41199165894616385, 71.25703896716800045,
+	72.10420474200799390, 72.95347118416940191, 73.80482079093779646,
+	74.65823634883015814, 75.51370092648485866, 76.37119786778275454,
+	77.23071078519033961, 78.09222355331530707, 78.95572030266725960,
+	79.82118541361435859, 80.68860351052903468, 81.55795945611502873,
+	82.42923834590904164, 83.30242550295004378, 84.17750647261028973,
+	85.05446701758152983, 85.93329311301090456, 86.81397094178107920,
+	87.69648688992882057, 88.58082754219766741, 89.46697967771913795,
+	90.35493026581838194, 91.24466646193963015, 92.13617560368709292,
+	93.02944520697742803, 93.92446296229978486, 94.82121673107967297,
+	95.71969454214321615, 96.61988458827809723, 97.52177522288820910,
+	98.42535495673848800, 99.33061245478741341, 100.23753653310367895,
+	101.14611615586458981, 102.05634043243354370, 102.96819861451382394,
+	103.88168009337621811, 104.79677439715833032, 105.71347118823287303,
+	106.63176026064346047, 107.55163153760463501, 108.47307506906540198,
+	109.39608102933323153, 110.32063971475740516, 111.24674154146920557,
+	112.17437704317786995, 113.10353686902013237, 114.03421178146170689,
+	114.96639265424990128, 115.90007047041454769, 116.83523632031698014,
+	117.77188139974506953, 118.70999700805310795, 119.64957454634490830,
+	120.59060551569974962, 121.53308151543865279, 122.47699424143097247,
+	123.42233548443955726, 124.36909712850338394, 125.31727114935689826,
+	126.26684961288492559, 127.21782467361175861, 128.17018857322420899,
+	129.12393363912724453, 130.07905228303084755, 131.03553699956862033,
+	131.99338036494577864, 132.95257503561629164, 133.91311374698926784,
+	134.87498931216194364, 135.83819462068046846, 136.80272263732638294,
+	137.76856640092901785, 138.73571902320256299, 139.70417368760718091,
+	140.67392364823425055, 141.64496222871400732, 142.61728282114600574,
+	143.59087888505104047, 144.56574394634486680, 145.54187159633210058,
+	146.51925549072063859, 147.49788934865566148, 148.47776695177302031,
+	149.45888214327129617, 150.44122882700193600, 151.42480096657754984,
+	152.40959258449737490, 153.39559776128982094, 154.38281063467164245,
+	155.37122539872302696, 156.36083630307879844, 157.35163765213474107,
+	158.34362380426921391, 159.33678917107920370, 160.33112821663092973,
+	161.32663545672428995, 162.32330545817117695, 163.32113283808695314,
+	164.32011226319519892, 165.32023844914485267, 166.32150615984036790,
+	167.32391020678358018, 168.32744544842768164, 169.33210678954270634,
+	170.33788918059275375, 171.34478761712384198, 172.35279713916281707,
+	173.36191283062726143, 174.37212981874515094, 175.38344327348534080,
+	176.39584840699734514, 177.40934047306160437, 178.42391476654847793,
+	179.43956662288721304, 180.45629141754378111, 181.47408456550741107,
+	182.49294152078630304, 183.51285777591152737, 184.53382886144947861,
+	185.55585034552262869, 186.57891783333786861, 187.60302696672312095,
+	188.62817342367162610, 189.65435291789341932, 190.68156119837468054,
+	191.70979404894376330, 192.73904728784492590, 193.76931676731820176,
+	194.80059837318714244, 195.83288802445184729, 196.86618167288995096,
+	197.90047530266301123, 198.93576492992946214, 199.97204660246373464,
+	201.00931639928148797, 202.04757043027063901, 203.08680483582807597,
+	204.12701578650228385, 205.16819948264117102, 206.21035215404597807,
+	207.25347005962987623, 208.29754948708190909, 209.34258675253678916,
+	210.38857820024875878, 211.43552020227099320, 212.48340915813977858,
+	213.53224149456323744, 214.58201366511514152, 215.63272214993284592,
+	216.68436345542014010, 217.73693411395422004, 218.79043068359703739,
+	219.84484974781133815, 220.90018791517996988, 221.95644181913033322,
+	223.01360811766215875, 224.07168349307951871, 225.13066465172661879,
+	226.19054832372759734, 227.25133126272962159, 228.31301024565024704,
+	229.37558207242807384, 230.43904356577689896, 231.50339157094342113,
+	232.56862295546847008, 233.63473460895144740, 234.70172344281823484,
+	235.76958639009222907, 236.83832040516844586, 237.90792246359117712,
+	238.97838956183431947, 240.04971871708477238, 241.12190696702904802,
+	242.19495136964280846, 243.26884900298270509, 244.34359696498191283,
+	245.41919237324782443, 246.49563236486270057, 247.57291409618682110,
+	248.65103474266476269, 249.72999149863338175, 250.80978157713354904,
+	251.89040220972316320, 252.97185064629374551, 254.05412415488834199,
+	255.13722002152300661, 256.22113555000953511, 257.30586806178126835,
+	258.39141489572085675, 259.47777340799029844, 260.56494097186322279,
+	261.65291497755913497, 262.74169283208021852, 263.83127195904967266,
+	264.92164979855277807, 266.01282380697938379, 267.10479145686849733,
+	268.19755023675537586, 269.29109765101975427, 270.38543121973674488,
+	271.48054847852881721, 272.57644697842033565, 273.67312428569374561,
+	274.77057798174683967, 275.86880566295326389, 276.96780494052313770,
+	278.06757344036617496, 279.16810880295668085, 280.26940868320008349,
+	281.37147075030043197, 282.47429268763045229, 283.57787219260217171,
+	284.68220697654078322, 285.78729476455760050, 286.89313329542699194,
+	287.99972032146268930, 289.10705360839756395, 290.21513093526289140,
+	291.32395009427028754, 292.43350889069523646, 293.54380514276073200,
+	294.65483668152336350, 295.76660135076059532, 296.87909700685889902,
+	297.99232151870342022, 299.10627276756946458, 300.22094864701409733,
+	301.33634706277030091, 302.45246593264130297, 303.56930318639643929,
+	304.68685676566872189, 305.80512462385280514, 306.92410472600477078,
+	308.04379504874236773, 309.16419358014690033, 310.28529831966631036,
+	311.40710727801865687, 312.52961847709792664, 313.65282994987899201,
+	314.77673974032603610, 315.90134590329950015, 317.02664650446632777,
+	318.15263962020929966, 319.27932333753892635, 320.40669575400545455,
+	321.53475497761127144, 322.66349912672620803, 323.79292633000159185,
+	324.92303472628691452, 326.05382246454587403, 327.18528770377525916,
+	328.31742861292224234, 329.45024337080525356, 330.58373016603343331,
+	331.71788719692847280, 332.85271267144611329, 333.98820480709991898,
+	335.12436183088397001, 336.26118197919845443, 337.39866349777429377,
+	338.53680464159958774, 339.67560367484657036, 340.81505887079896411,
+	341.95516851178109619, 343.09593088908627578, 344.23734430290727460,
+	345.37940706226686416, 346.52211748494903532, 347.66547389743118401,
+	348.80947463481720661, 349.95411804077025408, 351.09940246744753267,
+	352.24532627543504759, 353.39188783368263103, 354.53908551944078908,
+	355.68691771819692349, 356.83538282361303118, 357.98447923746385868,
+	359.13420536957539753
+};
+
+TEST_BEGIN(test_ln_gamma_misc) {
+	unsigned i;
+
+	for (i = 1; i < sizeof(ln_gamma_misc_expected)/sizeof(double); i++) {
+		double x = (double)i * 0.25;
+		assert_true(double_eq_rel(ln_gamma(x),
+		    ln_gamma_misc_expected[i], MAX_REL_ERR, MAX_ABS_ERR),
+		    "Incorrect ln_gamma result for i=%u", i);
+	}
+}
+TEST_END
+
+/* Expected pt_norm([0.01..0.99] increment=0.01). */
+static const double pt_norm_expected[] = {
+	-INFINITY,
+	-2.32634787404084076, -2.05374891063182252, -1.88079360815125085,
+	-1.75068607125216946, -1.64485362695147264, -1.55477359459685305,
+	-1.47579102817917063, -1.40507156030963221, -1.34075503369021654,
+	-1.28155156554460081, -1.22652812003661049, -1.17498679206608991,
+	-1.12639112903880045, -1.08031934081495606, -1.03643338949378938,
+	-0.99445788320975281, -0.95416525314619416, -0.91536508784281390,
+	-0.87789629505122846, -0.84162123357291418, -0.80642124701824025,
+	-0.77219321418868492, -0.73884684918521371, -0.70630256284008752,
+	-0.67448975019608171, -0.64334540539291685, -0.61281299101662701,
+	-0.58284150727121620, -0.55338471955567281, -0.52440051270804067,
+	-0.49585034734745320, -0.46769879911450812, -0.43991316567323380,
+	-0.41246312944140462, -0.38532046640756751, -0.35845879325119373,
+	-0.33185334643681652, -0.30548078809939738, -0.27931903444745404,
+	-0.25334710313579978, -0.22754497664114931, -0.20189347914185077,
+	-0.17637416478086135, -0.15096921549677725, -0.12566134685507399,
+	-0.10043372051146975, -0.07526986209982976, -0.05015358346473352,
+	-0.02506890825871106, 0.00000000000000000, 0.02506890825871106,
+	0.05015358346473366, 0.07526986209982990, 0.10043372051146990,
+	0.12566134685507413, 0.15096921549677739, 0.17637416478086146,
+	0.20189347914185105, 0.22754497664114931, 0.25334710313579978,
+	0.27931903444745404, 0.30548078809939738, 0.33185334643681652,
+	0.35845879325119373, 0.38532046640756762, 0.41246312944140484,
+	0.43991316567323391, 0.46769879911450835, 0.49585034734745348,
+	0.52440051270804111, 0.55338471955567303, 0.58284150727121620,
+	0.61281299101662701, 0.64334540539291685, 0.67448975019608171,
+	0.70630256284008752, 0.73884684918521371, 0.77219321418868492,
+	0.80642124701824036, 0.84162123357291441, 0.87789629505122879,
+	0.91536508784281423, 0.95416525314619460, 0.99445788320975348,
+	1.03643338949378938, 1.08031934081495606, 1.12639112903880045,
+	1.17498679206608991, 1.22652812003661049, 1.28155156554460081,
+	1.34075503369021654, 1.40507156030963265, 1.47579102817917085,
+	1.55477359459685394, 1.64485362695147308, 1.75068607125217102,
+	1.88079360815125041, 2.05374891063182208, 2.32634787404084076
+};
+
+TEST_BEGIN(test_pt_norm) {
+	unsigned i;
+
+	for (i = 1; i < sizeof(pt_norm_expected)/sizeof(double); i++) {
+		double p = (double)i * 0.01;
+		assert_true(double_eq_rel(pt_norm(p), pt_norm_expected[i],
+		    MAX_REL_ERR, MAX_ABS_ERR),
+		    "Incorrect pt_norm result for i=%u", i);
+	}
+}
+TEST_END
+
+/*
+ * Expected pt_chi2(p=[0.01..0.99] increment=0.07,
+ *                  df={0.1, 1.1, 10.1, 100.1, 1000.1}).
+ */
+static const double pt_chi2_df[] = {0.1, 1.1, 10.1, 100.1, 1000.1};
+static const double pt_chi2_expected[] = {
+	1.168926411457320e-40, 1.347680397072034e-22, 3.886980416666260e-17,
+	8.245951724356564e-14, 2.068936347497604e-11, 1.562561743309233e-09,
+	5.459543043426564e-08, 1.114775688149252e-06, 1.532101202364371e-05,
+	1.553884683726585e-04, 1.239396954915939e-03, 8.153872320255721e-03,
+	4.631183739647523e-02, 2.473187311701327e-01, 2.175254800183617e+00,
+
+	0.0003729887888876379, 0.0164409238228929513, 0.0521523015190650113,
+	0.1064701372271216612, 0.1800913735793082115, 0.2748704281195626931,
+	0.3939246282787986497, 0.5420727552260817816, 0.7267265822221973259,
+	0.9596554296000253670, 1.2607440376386165326, 1.6671185084541604304,
+	2.2604828984738705167, 3.2868613342148607082, 6.9298574921692139839,
+
+	2.606673548632508, 4.602913725294877, 5.646152813924212,
+	6.488971315540869, 7.249823275816285, 7.977314231410841,
+	8.700354939944047, 9.441728024225892, 10.224338321374127,
+	11.076435368801061, 12.039320937038386, 13.183878752697167,
+	14.657791935084575, 16.885728216339373, 23.361991680031817,
+
+	70.14844087392152, 80.92379498849355, 85.53325420085891,
+	88.94433120715347, 91.83732712857017, 94.46719943606301,
+	96.96896479994635, 99.43412843510363, 101.94074719829733,
+	104.57228644307247, 107.43900093448734, 110.71844673417287,
+	114.76616819871325, 120.57422505959563, 135.92318818757556,
+
+	899.0072447849649, 937.9271278858220, 953.8117189560207,
+	965.3079371501154, 974.8974061207954, 983.4936235182347,
+	991.5691170518946, 999.4334123954690, 1007.3391826856553,
+	1015.5445154999951, 1024.3777075619569, 1034.3538789836223,
+	1046.4872561869577, 1063.5717461999654, 1107.0741966053859
+};
+
+TEST_BEGIN(test_pt_chi2) {
+	unsigned i, j;
+	unsigned e = 0;
+
+	for (i = 0; i < sizeof(pt_chi2_df)/sizeof(double); i++) {
+		double df = pt_chi2_df[i];
+		double ln_gamma_df = ln_gamma(df * 0.5);
+		for (j = 1; j < 100; j += 7) {
+			double p = (double)j * 0.01;
+			assert_true(double_eq_rel(pt_chi2(p, df, ln_gamma_df),
+			    pt_chi2_expected[e], MAX_REL_ERR, MAX_ABS_ERR),
+			    "Incorrect pt_chi2 result for i=%u, j=%u", i, j);
+			e++;
+		}
+	}
+}
+TEST_END
+
+/*
+ * Expected pt_gamma(p=[0.1..0.99] increment=0.07,
+ *                   shape=[0.5..3.0] increment=0.5).
+ */
+static const double pt_gamma_shape[] = {0.5, 1.0, 1.5, 2.0, 2.5, 3.0};
+static const double pt_gamma_expected[] = {
+	7.854392895485103e-05, 5.043466107888016e-03, 1.788288957794883e-02,
+	3.900956150232906e-02, 6.913847560638034e-02, 1.093710833465766e-01,
+	1.613412523825817e-01, 2.274682115597864e-01, 3.114117323127083e-01,
+	4.189466220207417e-01, 5.598106789059246e-01, 7.521856146202706e-01,
+	1.036125427911119e+00, 1.532450860038180e+00, 3.317448300510606e+00,
+
+	0.01005033585350144, 0.08338160893905107, 0.16251892949777497,
+	0.24846135929849966, 0.34249030894677596, 0.44628710262841947,
+	0.56211891815354142, 0.69314718055994529, 0.84397007029452920,
+	1.02165124753198167, 1.23787435600161766, 1.51412773262977574,
+	1.89711998488588196, 2.52572864430825783, 4.60517018598809091,
+
+	0.05741590094955853, 0.24747378084860744, 0.39888572212236084,
+	0.54394139997444901, 0.69048812513915159, 0.84311389861296104,
+	1.00580622221479898, 1.18298694218766931, 1.38038096305861213,
+	1.60627736383027453, 1.87396970522337947, 2.20749220408081070,
+	2.65852391865854942, 3.37934630984842244, 5.67243336507218476,
+
+	0.1485547402532659, 0.4657458011640391, 0.6832386130709406,
+	0.8794297834672100, 1.0700752852474524, 1.2629614217350744,
+	1.4638400448580779, 1.6783469900166610, 1.9132338090606940,
+	2.1778589228618777, 2.4868823970010991, 2.8664695666264195,
+	3.3724415436062114, 4.1682658512758071, 6.6383520679938108,
+
+	0.2771490383641385, 0.7195001279643727, 0.9969081732265243,
+	1.2383497880608061, 1.4675206597269927, 1.6953064251816552,
+	1.9291243435606809, 2.1757300955477641, 2.4428032131216391,
+	2.7406534569230616, 3.0851445039665513, 3.5043101122033367,
+	4.0575997065264637, 4.9182956424675286, 7.5431362346944937,
+
+	0.4360451650782932, 0.9983600902486267, 1.3306365880734528,
+	1.6129750834753802, 1.8767241606994294, 2.1357032436097660,
+	2.3988853336865565, 2.6740603137235603, 2.9697561737517959,
+	3.2971457713883265, 3.6731795898504660, 4.1275751617770631,
+	4.7230515633946677, 5.6417477865306020, 8.4059469148854635
+};
+
+TEST_BEGIN(test_pt_gamma_shape) {
+	unsigned i, j;
+	unsigned e = 0;
+
+	for (i = 0; i < sizeof(pt_gamma_shape)/sizeof(double); i++) {
+		double shape = pt_gamma_shape[i];
+		double ln_gamma_shape = ln_gamma(shape);
+		for (j = 1; j < 100; j += 7) {
+			double p = (double)j * 0.01;
+			assert_true(double_eq_rel(pt_gamma(p, shape, 1.0,
+			    ln_gamma_shape), pt_gamma_expected[e], MAX_REL_ERR,
+			    MAX_ABS_ERR),
+			    "Incorrect pt_gamma result for i=%u, j=%u", i, j);
+			e++;
+		}
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_pt_gamma_scale) {
+	double shape = 1.0;
+	double ln_gamma_shape = ln_gamma(shape);
+
+	assert_true(double_eq_rel(
+	    pt_gamma(0.5, shape, 1.0, ln_gamma_shape) * 10.0,
+	    pt_gamma(0.5, shape, 10.0, ln_gamma_shape), MAX_REL_ERR,
+	    MAX_ABS_ERR),
+	    "Scale should be trivially equivalent to external multiplication");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ln_gamma_factorial,
+	    test_ln_gamma_misc,
+	    test_pt_norm,
+	    test_pt_chi2,
+	    test_pt_gamma_shape,
+	    test_pt_gamma_scale);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/mq.c b/memkind-1.10.0/jemalloc/test/unit/mq.c
new file mode 100644
index 0000000..57a4d54
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/mq.c
@@ -0,0 +1,89 @@
+#include "test/jemalloc_test.h"
+
+#define NSENDERS	3
+#define NMSGS		100000
+
+typedef struct mq_msg_s mq_msg_t;
+struct mq_msg_s {
+	mq_msg(mq_msg_t)	link;
+};
+mq_gen(static, mq_, mq_t, mq_msg_t, link)
+
+TEST_BEGIN(test_mq_basic) {
+	mq_t mq;
+	mq_msg_t msg;
+
+	assert_false(mq_init(&mq), "Unexpected mq_init() failure");
+	assert_u_eq(mq_count(&mq), 0, "mq should be empty");
+	assert_ptr_null(mq_tryget(&mq),
+	    "mq_tryget() should fail when the queue is empty");
+
+	mq_put(&mq, &msg);
+	assert_u_eq(mq_count(&mq), 1, "mq should contain one message");
+	assert_ptr_eq(mq_tryget(&mq), &msg, "mq_tryget() should return msg");
+
+	mq_put(&mq, &msg);
+	assert_ptr_eq(mq_get(&mq), &msg, "mq_get() should return msg");
+
+	mq_fini(&mq);
+}
+TEST_END
+
+static void *
+thd_receiver_start(void *arg) {
+	mq_t *mq = (mq_t *)arg;
+	unsigned i;
+
+	for (i = 0; i < (NSENDERS * NMSGS); i++) {
+		mq_msg_t *msg = mq_get(mq);
+		assert_ptr_not_null(msg, "mq_get() should never return NULL");
+		dallocx(msg, 0);
+	}
+	return NULL;
+}
+
+static void *
+thd_sender_start(void *arg) {
+	mq_t *mq = (mq_t *)arg;
+	unsigned i;
+
+	for (i = 0; i < NMSGS; i++) {
+		mq_msg_t *msg;
+		void *p;
+		p = mallocx(sizeof(mq_msg_t), 0);
+		assert_ptr_not_null(p, "Unexpected mallocx() failure");
+		msg = (mq_msg_t *)p;
+		mq_put(mq, msg);
+	}
+	return NULL;
+}
+
+TEST_BEGIN(test_mq_threaded) {
+	mq_t mq;
+	thd_t receiver;
+	thd_t senders[NSENDERS];
+	unsigned i;
+
+	assert_false(mq_init(&mq), "Unexpected mq_init() failure");
+
+	thd_create(&receiver, thd_receiver_start, (void *)&mq);
+	for (i = 0; i < NSENDERS; i++) {
+		thd_create(&senders[i], thd_sender_start, (void *)&mq);
+	}
+
+	thd_join(receiver, NULL);
+	for (i = 0; i < NSENDERS; i++) {
+		thd_join(senders[i], NULL);
+	}
+
+	mq_fini(&mq);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_mq_basic,
+	    test_mq_threaded);
+}
+
diff --git a/memkind-1.10.0/jemalloc/test/unit/mtx.c b/memkind-1.10.0/jemalloc/test/unit/mtx.c
new file mode 100644
index 0000000..424587b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/mtx.c
@@ -0,0 +1,57 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS	2
+#define NINCRS		2000000
+
+TEST_BEGIN(test_mtx_basic) {
+	mtx_t mtx;
+
+	assert_false(mtx_init(&mtx), "Unexpected mtx_init() failure");
+	mtx_lock(&mtx);
+	mtx_unlock(&mtx);
+	mtx_fini(&mtx);
+}
+TEST_END
+
+typedef struct {
+	mtx_t		mtx;
+	unsigned	x;
+} thd_start_arg_t;
+
+static void *
+thd_start(void *varg) {
+	thd_start_arg_t *arg = (thd_start_arg_t *)varg;
+	unsigned i;
+
+	for (i = 0; i < NINCRS; i++) {
+		mtx_lock(&arg->mtx);
+		arg->x++;
+		mtx_unlock(&arg->mtx);
+	}
+	return NULL;
+}
+
+TEST_BEGIN(test_mtx_race) {
+	thd_start_arg_t arg;
+	thd_t thds[NTHREADS];
+	unsigned i;
+
+	assert_false(mtx_init(&arg.mtx), "Unexpected mtx_init() failure");
+	arg.x = 0;
+	for (i = 0; i < NTHREADS; i++) {
+		thd_create(&thds[i], thd_start, (void *)&arg);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+	assert_u_eq(arg.x, NTHREADS * NINCRS,
+	    "Race-related counter corruption");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_mtx_basic,
+	    test_mtx_race);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/nstime.c b/memkind-1.10.0/jemalloc/test/unit/nstime.c
new file mode 100644
index 0000000..f313780
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/nstime.c
@@ -0,0 +1,249 @@
+#include "test/jemalloc_test.h"
+
+#define BILLION	UINT64_C(1000000000)
+
+TEST_BEGIN(test_nstime_init) {
+	nstime_t nst;
+
+	nstime_init(&nst, 42000000043);
+	assert_u64_eq(nstime_ns(&nst), 42000000043, "ns incorrectly read");
+	assert_u64_eq(nstime_sec(&nst), 42, "sec incorrectly read");
+	assert_u64_eq(nstime_nsec(&nst), 43, "nsec incorrectly read");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_init2) {
+	nstime_t nst;
+
+	nstime_init2(&nst, 42, 43);
+	assert_u64_eq(nstime_sec(&nst), 42, "sec incorrectly read");
+	assert_u64_eq(nstime_nsec(&nst), 43, "nsec incorrectly read");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_copy) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_init(&nstb, 0);
+	nstime_copy(&nstb, &nsta);
+	assert_u64_eq(nstime_sec(&nstb), 42, "sec incorrectly copied");
+	assert_u64_eq(nstime_nsec(&nstb), 43, "nsec incorrectly copied");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_compare) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0, "Times should be equal");
+	assert_d_eq(nstime_compare(&nstb, &nsta), 0, "Times should be equal");
+
+	nstime_init2(&nstb, 42, 42);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 1,
+	    "nsta should be greater than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), -1,
+	    "nstb should be less than nsta");
+
+	nstime_init2(&nstb, 42, 44);
+	assert_d_eq(nstime_compare(&nsta, &nstb), -1,
+	    "nsta should be less than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), 1,
+	    "nstb should be greater than nsta");
+
+	nstime_init2(&nstb, 41, BILLION - 1);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 1,
+	    "nsta should be greater than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), -1,
+	    "nstb should be less than nsta");
+
+	nstime_init2(&nstb, 43, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), -1,
+	    "nsta should be less than nstb");
+	assert_d_eq(nstime_compare(&nstb, &nsta), 1,
+	    "nstb should be greater than nsta");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_add) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_add(&nsta, &nstb);
+	nstime_init2(&nstb, 84, 86);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+
+	nstime_init2(&nsta, 42, BILLION - 1);
+	nstime_copy(&nstb, &nsta);
+	nstime_add(&nsta, &nstb);
+	nstime_init2(&nstb, 85, BILLION - 2);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_iadd) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, BILLION - 1);
+	nstime_iadd(&nsta, 1);
+	nstime_init2(&nstb, 43, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+
+	nstime_init2(&nsta, 42, 1);
+	nstime_iadd(&nsta, BILLION + 1);
+	nstime_init2(&nstb, 43, 2);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect addition result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_subtract) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_subtract(&nsta, &nstb);
+	nstime_init(&nstb, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_init2(&nstb, 41, 44);
+	nstime_subtract(&nsta, &nstb);
+	nstime_init2(&nstb, 0, BILLION - 1);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_isubtract) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_isubtract(&nsta, 42*BILLION + 43);
+	nstime_init(&nstb, 0);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_isubtract(&nsta, 41*BILLION + 44);
+	nstime_init2(&nstb, 0, BILLION - 1);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect subtraction result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_imultiply) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_imultiply(&nsta, 10);
+	nstime_init2(&nstb, 420, 430);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect multiplication result");
+
+	nstime_init2(&nsta, 42, 666666666);
+	nstime_imultiply(&nsta, 3);
+	nstime_init2(&nstb, 127, 999999998);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect multiplication result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_idivide) {
+	nstime_t nsta, nstb;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	nstime_idivide(&nsta, 10);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect division result");
+
+	nstime_init2(&nsta, 42, 666666666);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 3);
+	nstime_idivide(&nsta, 3);
+	assert_d_eq(nstime_compare(&nsta, &nstb), 0,
+	    "Incorrect division result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_divide) {
+	nstime_t nsta, nstb, nstc;
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	assert_u64_eq(nstime_divide(&nsta, &nstb), 10,
+	    "Incorrect division result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	nstime_init(&nstc, 1);
+	nstime_add(&nsta, &nstc);
+	assert_u64_eq(nstime_divide(&nsta, &nstb), 10,
+	    "Incorrect division result");
+
+	nstime_init2(&nsta, 42, 43);
+	nstime_copy(&nstb, &nsta);
+	nstime_imultiply(&nsta, 10);
+	nstime_init(&nstc, 1);
+	nstime_subtract(&nsta, &nstc);
+	assert_u64_eq(nstime_divide(&nsta, &nstb), 9,
+	    "Incorrect division result");
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_monotonic) {
+	nstime_monotonic();
+}
+TEST_END
+
+TEST_BEGIN(test_nstime_update) {
+	nstime_t nst;
+
+	nstime_init(&nst, 0);
+
+	assert_false(nstime_update(&nst), "Basic time update failed.");
+
+	/* Only Rip Van Winkle sleeps this long. */
+	{
+		nstime_t addend;
+		nstime_init2(&addend, 631152000, 0);
+		nstime_add(&nst, &addend);
+	}
+	{
+		nstime_t nst0;
+		nstime_copy(&nst0, &nst);
+		assert_true(nstime_update(&nst),
+		    "Update should detect time roll-back.");
+		assert_d_eq(nstime_compare(&nst, &nst0), 0,
+		    "Time should not have been modified");
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_nstime_init,
+	    test_nstime_init2,
+	    test_nstime_copy,
+	    test_nstime_compare,
+	    test_nstime_add,
+	    test_nstime_iadd,
+	    test_nstime_subtract,
+	    test_nstime_isubtract,
+	    test_nstime_imultiply,
+	    test_nstime_idivide,
+	    test_nstime_divide,
+	    test_nstime_monotonic,
+	    test_nstime_update);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/pack.c b/memkind-1.10.0/jemalloc/test/unit/pack.c
new file mode 100644
index 0000000..fc188b0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/pack.c
@@ -0,0 +1,166 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * Size class that is a divisor of the page size, ideally 4+ regions per run.
+ */
+#if LG_PAGE <= 14
+#define SZ	(ZU(1) << (LG_PAGE - 2))
+#else
+#define SZ	ZU(4096)
+#endif
+
+/*
+ * Number of slabs to consume at high water mark.  Should be at least 2 so that
+ * if mmap()ed memory grows downward, downward growth of mmap()ed memory is
+ * tested.
+ */
+#define NSLABS	8
+
+static unsigned
+binind_compute(void) {
+	size_t sz;
+	unsigned nbins, i;
+
+	sz = sizeof(nbins);
+	assert_d_eq(mallctl("arenas.nbins", (void *)&nbins, &sz, NULL, 0), 0,
+	    "Unexpected mallctl failure");
+
+	for (i = 0; i < nbins; i++) {
+		size_t mib[4];
+		size_t miblen = sizeof(mib)/sizeof(size_t);
+		size_t size;
+
+		assert_d_eq(mallctlnametomib("arenas.bin.0.size", mib,
+		    &miblen), 0, "Unexpected mallctlnametomb failure");
+		mib[2] = (size_t)i;
+
+		sz = sizeof(size);
+		assert_d_eq(mallctlbymib(mib, miblen, (void *)&size, &sz, NULL,
+		    0), 0, "Unexpected mallctlbymib failure");
+		if (size == SZ) {
+			return i;
+		}
+	}
+
+	test_fail("Unable to compute nregs_per_run");
+	return 0;
+}
+
+static size_t
+nregs_per_run_compute(void) {
+	uint32_t nregs;
+	size_t sz;
+	unsigned binind = binind_compute();
+	size_t mib[4];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib("arenas.bin.0.nregs", mib, &miblen), 0,
+	    "Unexpected mallctlnametomb failure");
+	mib[2] = (size_t)binind;
+	sz = sizeof(nregs);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&nregs, &sz, NULL,
+	    0), 0, "Unexpected mallctlbymib failure");
+	return nregs;
+}
+
+static unsigned
+arenas_create_mallctl(void) {
+	unsigned arena_ind;
+	size_t sz;
+
+	sz = sizeof(arena_ind);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Error in arenas.create");
+
+	return arena_ind;
+}
+
+static void
+arena_reset_mallctl(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen = sizeof(mib)/sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib("arena.0.reset", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+TEST_BEGIN(test_pack) {
+	bool prof_enabled;
+	size_t sz = sizeof(prof_enabled);
+	if (mallctl("opt.prof", (void *)&prof_enabled, &sz, NULL, 0) == 0) {
+		test_skip_if(prof_enabled);
+	}
+
+	unsigned arena_ind = arenas_create_mallctl();
+	size_t nregs_per_run = nregs_per_run_compute();
+	size_t nregs = nregs_per_run * NSLABS;
+	VARIABLE_ARRAY(void *, ptrs, nregs);
+	size_t i, j, offset;
+
+	/* Fill matrix. */
+	for (i = offset = 0; i < NSLABS; i++) {
+		for (j = 0; j < nregs_per_run; j++) {
+			void *p = mallocx(SZ, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE);
+			assert_ptr_not_null(p,
+			    "Unexpected mallocx(%zu, MALLOCX_ARENA(%u) |"
+			    " MALLOCX_TCACHE_NONE) failure, run=%zu, reg=%zu",
+			    SZ, arena_ind, i, j);
+			ptrs[(i * nregs_per_run) + j] = p;
+		}
+	}
+
+	/*
+	 * Free all but one region of each run, but rotate which region is
+	 * preserved, so that subsequent allocations exercise the within-run
+	 * layout policy.
+	 */
+	offset = 0;
+	for (i = offset = 0;
+	    i < NSLABS;
+	    i++, offset = (offset + 1) % nregs_per_run) {
+		for (j = 0; j < nregs_per_run; j++) {
+			void *p = ptrs[(i * nregs_per_run) + j];
+			if (offset == j) {
+				continue;
+			}
+			dallocx(p, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE);
+		}
+	}
+
+	/*
+	 * Logically refill matrix, skipping preserved regions and verifying
+	 * that the matrix is unmodified.
+	 */
+	offset = 0;
+	for (i = offset = 0;
+	    i < NSLABS;
+	    i++, offset = (offset + 1) % nregs_per_run) {
+		for (j = 0; j < nregs_per_run; j++) {
+			void *p;
+
+			if (offset == j) {
+				continue;
+			}
+			p = mallocx(SZ, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE);
+			assert_ptr_eq(p, ptrs[(i * nregs_per_run) + j],
+			    "Unexpected refill discrepancy, run=%zu, reg=%zu\n",
+			    i, j);
+		}
+	}
+
+	/* Clean up. */
+	arena_reset_mallctl(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_pack);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/pack.sh b/memkind-1.10.0/jemalloc/test/unit/pack.sh
new file mode 100644
index 0000000..6f45148
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/pack.sh
@@ -0,0 +1,4 @@
+#!/bin/sh
+
+# Immediately purge to minimize fragmentation.
+export MALLOC_CONF="dirty_decay_ms:0,muzzy_decay_ms:0"
diff --git a/memkind-1.10.0/jemalloc/test/unit/pages.c b/memkind-1.10.0/jemalloc/test/unit/pages.c
new file mode 100644
index 0000000..ee729ee
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/pages.c
@@ -0,0 +1,29 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_pages_huge) {
+	size_t alloc_size;
+	bool commit;
+	void *pages, *hugepage;
+
+	alloc_size = HUGEPAGE * 2 - PAGE;
+	commit = true;
+	pages = pages_map(NULL, alloc_size, PAGE, &commit);
+	assert_ptr_not_null(pages, "Unexpected pages_map() error");
+
+	if (init_system_thp_mode == thp_mode_default) {
+	    hugepage = (void *)(ALIGNMENT_CEILING((uintptr_t)pages, HUGEPAGE));
+	    assert_b_ne(pages_huge(hugepage, HUGEPAGE), have_madvise_huge,
+	        "Unexpected pages_huge() result");
+	    assert_false(pages_nohuge(hugepage, HUGEPAGE),
+	        "Unexpected pages_nohuge() result");
+	}
+
+	pages_unmap(pages, alloc_size);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_pages_huge);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/ph.c b/memkind-1.10.0/jemalloc/test/unit/ph.c
new file mode 100644
index 0000000..88bf56f
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/ph.c
@@ -0,0 +1,318 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ph.h"
+
+typedef struct node_s node_t;
+
+struct node_s {
+#define NODE_MAGIC 0x9823af7e
+	uint32_t magic;
+	phn(node_t) link;
+	uint64_t key;
+};
+
+static int
+node_cmp(const node_t *a, const node_t *b) {
+	int ret;
+
+	ret = (a->key > b->key) - (a->key < b->key);
+	if (ret == 0) {
+		/*
+		 * Duplicates are not allowed in the heap, so force an
+		 * arbitrary ordering for non-identical items with equal keys.
+		 */
+		ret = (((uintptr_t)a) > ((uintptr_t)b))
+		    - (((uintptr_t)a) < ((uintptr_t)b));
+	}
+	return ret;
+}
+
+static int
+node_cmp_magic(const node_t *a, const node_t *b) {
+
+	assert_u32_eq(a->magic, NODE_MAGIC, "Bad magic");
+	assert_u32_eq(b->magic, NODE_MAGIC, "Bad magic");
+
+	return node_cmp(a, b);
+}
+
+typedef ph(node_t) heap_t;
+ph_gen(static, heap_, heap_t, node_t, link, node_cmp_magic);
+
+static void
+node_print(const node_t *node, unsigned depth) {
+	unsigned i;
+	node_t *leftmost_child, *sibling;
+
+	for (i = 0; i < depth; i++) {
+		malloc_printf("\t");
+	}
+	malloc_printf("%2"FMTu64"\n", node->key);
+
+	leftmost_child = phn_lchild_get(node_t, link, node);
+	if (leftmost_child == NULL) {
+		return;
+	}
+	node_print(leftmost_child, depth + 1);
+
+	for (sibling = phn_next_get(node_t, link, leftmost_child); sibling !=
+	    NULL; sibling = phn_next_get(node_t, link, sibling)) {
+		node_print(sibling, depth + 1);
+	}
+}
+
+static void
+heap_print(const heap_t *heap) {
+	node_t *auxelm;
+
+	malloc_printf("vvv heap %p vvv\n", heap);
+	if (heap->ph_root == NULL) {
+		goto label_return;
+	}
+
+	node_print(heap->ph_root, 0);
+
+	for (auxelm = phn_next_get(node_t, link, heap->ph_root); auxelm != NULL;
+	    auxelm = phn_next_get(node_t, link, auxelm)) {
+		assert_ptr_eq(phn_next_get(node_t, link, phn_prev_get(node_t,
+		    link, auxelm)), auxelm,
+		    "auxelm's prev doesn't link to auxelm");
+		node_print(auxelm, 0);
+	}
+
+label_return:
+	malloc_printf("^^^ heap %p ^^^\n", heap);
+}
+
+static unsigned
+node_validate(const node_t *node, const node_t *parent) {
+	unsigned nnodes = 1;
+	node_t *leftmost_child, *sibling;
+
+	if (parent != NULL) {
+		assert_d_ge(node_cmp_magic(node, parent), 0,
+		    "Child is less than parent");
+	}
+
+	leftmost_child = phn_lchild_get(node_t, link, node);
+	if (leftmost_child == NULL) {
+		return nnodes;
+	}
+	assert_ptr_eq((void *)phn_prev_get(node_t, link, leftmost_child),
+	    (void *)node, "Leftmost child does not link to node");
+	nnodes += node_validate(leftmost_child, node);
+
+	for (sibling = phn_next_get(node_t, link, leftmost_child); sibling !=
+	    NULL; sibling = phn_next_get(node_t, link, sibling)) {
+		assert_ptr_eq(phn_next_get(node_t, link, phn_prev_get(node_t,
+		    link, sibling)), sibling,
+		    "sibling's prev doesn't link to sibling");
+		nnodes += node_validate(sibling, node);
+	}
+	return nnodes;
+}
+
+static unsigned
+heap_validate(const heap_t *heap) {
+	unsigned nnodes = 0;
+	node_t *auxelm;
+
+	if (heap->ph_root == NULL) {
+		goto label_return;
+	}
+
+	nnodes += node_validate(heap->ph_root, NULL);
+
+	for (auxelm = phn_next_get(node_t, link, heap->ph_root); auxelm != NULL;
+	    auxelm = phn_next_get(node_t, link, auxelm)) {
+		assert_ptr_eq(phn_next_get(node_t, link, phn_prev_get(node_t,
+		    link, auxelm)), auxelm,
+		    "auxelm's prev doesn't link to auxelm");
+		nnodes += node_validate(auxelm, NULL);
+	}
+
+label_return:
+	if (false) {
+		heap_print(heap);
+	}
+	return nnodes;
+}
+
+TEST_BEGIN(test_ph_empty) {
+	heap_t heap;
+
+	heap_new(&heap);
+	assert_true(heap_empty(&heap), "Heap should be empty");
+	assert_ptr_null(heap_first(&heap), "Unexpected node");
+	assert_ptr_null(heap_any(&heap), "Unexpected node");
+}
+TEST_END
+
+static void
+node_remove(heap_t *heap, node_t *node) {
+	heap_remove(heap, node);
+
+	node->magic = 0;
+}
+
+static node_t *
+node_remove_first(heap_t *heap) {
+	node_t *node = heap_remove_first(heap);
+	node->magic = 0;
+	return node;
+}
+
+static node_t *
+node_remove_any(heap_t *heap) {
+	node_t *node = heap_remove_any(heap);
+	node->magic = 0;
+	return node;
+}
+
+TEST_BEGIN(test_ph_random) {
+#define NNODES 25
+#define NBAGS 250
+#define SEED 42
+	sfmt_t *sfmt;
+	uint64_t bag[NNODES];
+	heap_t heap;
+	node_t nodes[NNODES];
+	unsigned i, j, k;
+
+	sfmt = init_gen_rand(SEED);
+	for (i = 0; i < NBAGS; i++) {
+		switch (i) {
+		case 0:
+			/* Insert in order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = j;
+			}
+			break;
+		case 1:
+			/* Insert in reverse order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = NNODES - j - 1;
+			}
+			break;
+		default:
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = gen_rand64_range(sfmt, NNODES);
+			}
+		}
+
+		for (j = 1; j <= NNODES; j++) {
+			/* Initialize heap and nodes. */
+			heap_new(&heap);
+			assert_u_eq(heap_validate(&heap), 0,
+			    "Incorrect node count");
+			for (k = 0; k < j; k++) {
+				nodes[k].magic = NODE_MAGIC;
+				nodes[k].key = bag[k];
+			}
+
+			/* Insert nodes. */
+			for (k = 0; k < j; k++) {
+				heap_insert(&heap, &nodes[k]);
+				if (i % 13 == 12) {
+					assert_ptr_not_null(heap_any(&heap),
+					    "Heap should not be empty");
+					/* Trigger merging. */
+					assert_ptr_not_null(heap_first(&heap),
+					    "Heap should not be empty");
+				}
+				assert_u_eq(heap_validate(&heap), k + 1,
+				    "Incorrect node count");
+			}
+
+			assert_false(heap_empty(&heap),
+			    "Heap should not be empty");
+
+			/* Remove nodes. */
+			switch (i % 6) {
+			case 0:
+				for (k = 0; k < j; k++) {
+					assert_u_eq(heap_validate(&heap), j - k,
+					    "Incorrect node count");
+					node_remove(&heap, &nodes[k]);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+				}
+				break;
+			case 1:
+				for (k = j; k > 0; k--) {
+					node_remove(&heap, &nodes[k-1]);
+					assert_u_eq(heap_validate(&heap), k - 1,
+					    "Incorrect node count");
+				}
+				break;
+			case 2: {
+				node_t *prev = NULL;
+				for (k = 0; k < j; k++) {
+					node_t *node = node_remove_first(&heap);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+					if (prev != NULL) {
+						assert_d_ge(node_cmp(node,
+						    prev), 0,
+						    "Bad removal order");
+					}
+					prev = node;
+				}
+				break;
+			} case 3: {
+				node_t *prev = NULL;
+				for (k = 0; k < j; k++) {
+					node_t *node = heap_first(&heap);
+					assert_u_eq(heap_validate(&heap), j - k,
+					    "Incorrect node count");
+					if (prev != NULL) {
+						assert_d_ge(node_cmp(node,
+						    prev), 0,
+						    "Bad removal order");
+					}
+					node_remove(&heap, node);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+					prev = node;
+				}
+				break;
+			} case 4: {
+				for (k = 0; k < j; k++) {
+					node_remove_any(&heap);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+				}
+				break;
+			} case 5: {
+				for (k = 0; k < j; k++) {
+					node_t *node = heap_any(&heap);
+					assert_u_eq(heap_validate(&heap), j - k,
+					    "Incorrect node count");
+					node_remove(&heap, node);
+					assert_u_eq(heap_validate(&heap), j - k
+					    - 1, "Incorrect node count");
+				}
+				break;
+			} default:
+				not_reached();
+			}
+
+			assert_ptr_null(heap_first(&heap),
+			    "Heap should be empty");
+			assert_ptr_null(heap_any(&heap),
+			    "Heap should be empty");
+			assert_true(heap_empty(&heap), "Heap should be empty");
+		}
+	}
+	fini_gen_rand(sfmt);
+#undef NNODES
+#undef SEED
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ph_empty,
+	    test_ph_random);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prng.c b/memkind-1.10.0/jemalloc/test/unit/prng.c
new file mode 100644
index 0000000..b5795c2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prng.c
@@ -0,0 +1,237 @@
+#include "test/jemalloc_test.h"
+
+static void
+test_prng_lg_range_u32(bool atomic) {
+	atomic_u32_t sa, sb;
+	uint32_t ra, rb;
+	unsigned lg_range;
+
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_u32(&sa, 32, atomic);
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_u32(&sa, 32, atomic);
+	assert_u32_eq(ra, rb,
+	    "Repeated generation should produce repeated results");
+
+	atomic_store_u32(&sb, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_u32(&sb, 32, atomic);
+	assert_u32_eq(ra, rb,
+	    "Equivalent generation should produce equivalent results");
+
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_u32(&sa, 32, atomic);
+	rb = prng_lg_range_u32(&sa, 32, atomic);
+	assert_u32_ne(ra, rb,
+	    "Full-width results must not immediately repeat");
+
+	atomic_store_u32(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_u32(&sa, 32, atomic);
+	for (lg_range = 31; lg_range > 0; lg_range--) {
+		atomic_store_u32(&sb, 42, ATOMIC_RELAXED);
+		rb = prng_lg_range_u32(&sb, lg_range, atomic);
+		assert_u32_eq((rb & (UINT32_C(0xffffffff) << lg_range)),
+		    0, "High order bits should be 0, lg_range=%u", lg_range);
+		assert_u32_eq(rb, (ra >> (32 - lg_range)),
+		    "Expected high order bits of full-width result, "
+		    "lg_range=%u", lg_range);
+	}
+}
+
+static void
+test_prng_lg_range_u64(void) {
+	uint64_t sa, sb, ra, rb;
+	unsigned lg_range;
+
+	sa = 42;
+	ra = prng_lg_range_u64(&sa, 64);
+	sa = 42;
+	rb = prng_lg_range_u64(&sa, 64);
+	assert_u64_eq(ra, rb,
+	    "Repeated generation should produce repeated results");
+
+	sb = 42;
+	rb = prng_lg_range_u64(&sb, 64);
+	assert_u64_eq(ra, rb,
+	    "Equivalent generation should produce equivalent results");
+
+	sa = 42;
+	ra = prng_lg_range_u64(&sa, 64);
+	rb = prng_lg_range_u64(&sa, 64);
+	assert_u64_ne(ra, rb,
+	    "Full-width results must not immediately repeat");
+
+	sa = 42;
+	ra = prng_lg_range_u64(&sa, 64);
+	for (lg_range = 63; lg_range > 0; lg_range--) {
+		sb = 42;
+		rb = prng_lg_range_u64(&sb, lg_range);
+		assert_u64_eq((rb & (UINT64_C(0xffffffffffffffff) << lg_range)),
+		    0, "High order bits should be 0, lg_range=%u", lg_range);
+		assert_u64_eq(rb, (ra >> (64 - lg_range)),
+		    "Expected high order bits of full-width result, "
+		    "lg_range=%u", lg_range);
+	}
+}
+
+static void
+test_prng_lg_range_zu(bool atomic) {
+	atomic_zu_t sa, sb;
+	size_t ra, rb;
+	unsigned lg_range;
+
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	assert_zu_eq(ra, rb,
+	    "Repeated generation should produce repeated results");
+
+	atomic_store_zu(&sb, 42, ATOMIC_RELAXED);
+	rb = prng_lg_range_zu(&sb, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	assert_zu_eq(ra, rb,
+	    "Equivalent generation should produce equivalent results");
+
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	rb = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	assert_zu_ne(ra, rb,
+	    "Full-width results must not immediately repeat");
+
+	atomic_store_zu(&sa, 42, ATOMIC_RELAXED);
+	ra = prng_lg_range_zu(&sa, ZU(1) << (3 + LG_SIZEOF_PTR), atomic);
+	for (lg_range = (ZU(1) << (3 + LG_SIZEOF_PTR)) - 1; lg_range > 0;
+	    lg_range--) {
+		atomic_store_zu(&sb, 42, ATOMIC_RELAXED);
+		rb = prng_lg_range_zu(&sb, lg_range, atomic);
+		assert_zu_eq((rb & (SIZE_T_MAX << lg_range)),
+		    0, "High order bits should be 0, lg_range=%u", lg_range);
+		assert_zu_eq(rb, (ra >> ((ZU(1) << (3 + LG_SIZEOF_PTR)) -
+		    lg_range)), "Expected high order bits of full-width "
+		    "result, lg_range=%u", lg_range);
+	}
+}
+
+TEST_BEGIN(test_prng_lg_range_u32_nonatomic) {
+	test_prng_lg_range_u32(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_u32_atomic) {
+	test_prng_lg_range_u32(true);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_u64_nonatomic) {
+	test_prng_lg_range_u64();
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_zu_nonatomic) {
+	test_prng_lg_range_zu(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_lg_range_zu_atomic) {
+	test_prng_lg_range_zu(true);
+}
+TEST_END
+
+static void
+test_prng_range_u32(bool atomic) {
+	uint32_t range;
+#define MAX_RANGE	10000000
+#define RANGE_STEP	97
+#define NREPS		10
+
+	for (range = 2; range < MAX_RANGE; range += RANGE_STEP) {
+		atomic_u32_t s;
+		unsigned rep;
+
+		atomic_store_u32(&s, range, ATOMIC_RELAXED);
+		for (rep = 0; rep < NREPS; rep++) {
+			uint32_t r = prng_range_u32(&s, range, atomic);
+
+			assert_u32_lt(r, range, "Out of range");
+		}
+	}
+}
+
+static void
+test_prng_range_u64(void) {
+	uint64_t range;
+#define MAX_RANGE	10000000
+#define RANGE_STEP	97
+#define NREPS		10
+
+	for (range = 2; range < MAX_RANGE; range += RANGE_STEP) {
+		uint64_t s;
+		unsigned rep;
+
+		s = range;
+		for (rep = 0; rep < NREPS; rep++) {
+			uint64_t r = prng_range_u64(&s, range);
+
+			assert_u64_lt(r, range, "Out of range");
+		}
+	}
+}
+
+static void
+test_prng_range_zu(bool atomic) {
+	size_t range;
+#define MAX_RANGE	10000000
+#define RANGE_STEP	97
+#define NREPS		10
+
+	for (range = 2; range < MAX_RANGE; range += RANGE_STEP) {
+		atomic_zu_t s;
+		unsigned rep;
+
+		atomic_store_zu(&s, range, ATOMIC_RELAXED);
+		for (rep = 0; rep < NREPS; rep++) {
+			size_t r = prng_range_zu(&s, range, atomic);
+
+			assert_zu_lt(r, range, "Out of range");
+		}
+	}
+}
+
+TEST_BEGIN(test_prng_range_u32_nonatomic) {
+	test_prng_range_u32(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_u32_atomic) {
+	test_prng_range_u32(true);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_u64_nonatomic) {
+	test_prng_range_u64();
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_zu_nonatomic) {
+	test_prng_range_zu(false);
+}
+TEST_END
+
+TEST_BEGIN(test_prng_range_zu_atomic) {
+	test_prng_range_zu(true);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_prng_lg_range_u32_nonatomic,
+	    test_prng_lg_range_u32_atomic,
+	    test_prng_lg_range_u64_nonatomic,
+	    test_prng_lg_range_zu_nonatomic,
+	    test_prng_lg_range_zu_atomic,
+	    test_prng_range_u32_nonatomic,
+	    test_prng_range_u32_atomic,
+	    test_prng_range_u64_nonatomic,
+	    test_prng_range_zu_nonatomic,
+	    test_prng_range_zu_atomic);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_accum.c b/memkind-1.10.0/jemalloc/test/unit/prof_accum.c
new file mode 100644
index 0000000..2522006
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_accum.c
@@ -0,0 +1,81 @@
+#include "test/jemalloc_test.h"
+
+#define NTHREADS		4
+#define NALLOCS_PER_THREAD	50
+#define DUMP_INTERVAL		1
+#define BT_COUNT_CHECK_INTERVAL	5
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+static void *
+alloc_from_permuted_backtrace(unsigned thd_ind, unsigned iteration) {
+	return btalloc(1, thd_ind*NALLOCS_PER_THREAD + iteration);
+}
+
+static void *
+thd_start(void *varg) {
+	unsigned thd_ind = *(unsigned *)varg;
+	size_t bt_count_prev, bt_count;
+	unsigned i_prev, i;
+
+	i_prev = 0;
+	bt_count_prev = 0;
+	for (i = 0; i < NALLOCS_PER_THREAD; i++) {
+		void *p = alloc_from_permuted_backtrace(thd_ind, i);
+		dallocx(p, 0);
+		if (i % DUMP_INTERVAL == 0) {
+			assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+			    0, "Unexpected error while dumping heap profile");
+		}
+
+		if (i % BT_COUNT_CHECK_INTERVAL == 0 ||
+		    i+1 == NALLOCS_PER_THREAD) {
+			bt_count = prof_bt_count();
+			assert_zu_le(bt_count_prev+(i-i_prev), bt_count,
+			    "Expected larger backtrace count increase");
+			i_prev = i;
+			bt_count_prev = bt_count;
+		}
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_idump) {
+	bool active;
+	thd_t thds[NTHREADS];
+	unsigned thd_args[NTHREADS];
+	unsigned i;
+
+	test_skip_if(!config_prof);
+
+	active = true;
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0,
+	    "Unexpected mallctl failure while activating profiling");
+
+	prof_dump_open = prof_dump_open_intercept;
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_args[i] = i;
+		thd_create(&thds[i], thd_start, (void *)&thd_args[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_idump);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_accum.sh b/memkind-1.10.0/jemalloc/test/unit/prof_accum.sh
new file mode 100644
index 0000000..b3e13fc
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_accum.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_accum:true,prof_active:false,lg_prof_sample:0"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_active.c b/memkind-1.10.0/jemalloc/test/unit/prof_active.c
new file mode 100644
index 0000000..850a24a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_active.c
@@ -0,0 +1,117 @@
+#include "test/jemalloc_test.h"
+
+static void
+mallctl_bool_get(const char *name, bool expected, const char *func, int line) {
+	bool old;
+	size_t sz;
+
+	sz = sizeof(old);
+	assert_d_eq(mallctl(name, (void *)&old, &sz, NULL, 0), 0,
+	    "%s():%d: Unexpected mallctl failure reading %s", func, line, name);
+	assert_b_eq(old, expected, "%s():%d: Unexpected %s value", func, line,
+	    name);
+}
+
+static void
+mallctl_bool_set(const char *name, bool old_expected, bool val_new,
+    const char *func, int line) {
+	bool old;
+	size_t sz;
+
+	sz = sizeof(old);
+	assert_d_eq(mallctl(name, (void *)&old, &sz, (void *)&val_new,
+	    sizeof(val_new)), 0,
+	    "%s():%d: Unexpected mallctl failure reading/writing %s", func,
+	    line, name);
+	assert_b_eq(old, old_expected, "%s():%d: Unexpected %s value", func,
+	    line, name);
+}
+
+static void
+mallctl_prof_active_get_impl(bool prof_active_old_expected, const char *func,
+    int line) {
+	mallctl_bool_get("prof.active", prof_active_old_expected, func, line);
+}
+#define mallctl_prof_active_get(a)					\
+	mallctl_prof_active_get_impl(a, __func__, __LINE__)
+
+static void
+mallctl_prof_active_set_impl(bool prof_active_old_expected,
+    bool prof_active_new, const char *func, int line) {
+	mallctl_bool_set("prof.active", prof_active_old_expected,
+	    prof_active_new, func, line);
+}
+#define mallctl_prof_active_set(a, b)					\
+	mallctl_prof_active_set_impl(a, b, __func__, __LINE__)
+
+static void
+mallctl_thread_prof_active_get_impl(bool thread_prof_active_old_expected,
+    const char *func, int line) {
+	mallctl_bool_get("thread.prof.active", thread_prof_active_old_expected,
+	    func, line);
+}
+#define mallctl_thread_prof_active_get(a)				\
+	mallctl_thread_prof_active_get_impl(a, __func__, __LINE__)
+
+static void
+mallctl_thread_prof_active_set_impl(bool thread_prof_active_old_expected,
+    bool thread_prof_active_new, const char *func, int line) {
+	mallctl_bool_set("thread.prof.active", thread_prof_active_old_expected,
+	    thread_prof_active_new, func, line);
+}
+#define mallctl_thread_prof_active_set(a, b)				\
+	mallctl_thread_prof_active_set_impl(a, b, __func__, __LINE__)
+
+static void
+prof_sampling_probe_impl(bool expect_sample, const char *func, int line) {
+	void *p;
+	size_t expected_backtraces = expect_sample ? 1 : 0;
+
+	assert_zu_eq(prof_bt_count(), 0, "%s():%d: Expected 0 backtraces", func,
+	    line);
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	assert_zu_eq(prof_bt_count(), expected_backtraces,
+	    "%s():%d: Unexpected backtrace count", func, line);
+	dallocx(p, 0);
+}
+#define prof_sampling_probe(a)						\
+	prof_sampling_probe_impl(a, __func__, __LINE__)
+
+TEST_BEGIN(test_prof_active) {
+	test_skip_if(!config_prof);
+
+	mallctl_prof_active_get(true);
+	mallctl_thread_prof_active_get(false);
+
+	mallctl_prof_active_set(true, true);
+	mallctl_thread_prof_active_set(false, false);
+	/* prof.active, !thread.prof.active. */
+	prof_sampling_probe(false);
+
+	mallctl_prof_active_set(true, false);
+	mallctl_thread_prof_active_set(false, false);
+	/* !prof.active, !thread.prof.active. */
+	prof_sampling_probe(false);
+
+	mallctl_prof_active_set(false, false);
+	mallctl_thread_prof_active_set(false, true);
+	/* !prof.active, thread.prof.active. */
+	prof_sampling_probe(false);
+
+	mallctl_prof_active_set(false, true);
+	mallctl_thread_prof_active_set(true, true);
+	/* prof.active, thread.prof.active. */
+	prof_sampling_probe(true);
+
+	/* Restore settings. */
+	mallctl_prof_active_set(true, true);
+	mallctl_thread_prof_active_set(true, false);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_prof_active);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_active.sh b/memkind-1.10.0/jemalloc/test/unit/prof_active.sh
new file mode 100644
index 0000000..0167cb1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_active.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_thread_active_init:false,lg_prof_sample:0"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_gdump.c b/memkind-1.10.0/jemalloc/test/unit/prof_gdump.c
new file mode 100644
index 0000000..f7e0aac
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_gdump.c
@@ -0,0 +1,74 @@
+#include "test/jemalloc_test.h"
+
+static bool did_prof_dump_open;
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	did_prof_dump_open = true;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+TEST_BEGIN(test_gdump) {
+	bool active, gdump, gdump_old;
+	void *p, *q, *r, *s;
+	size_t sz;
+
+	test_skip_if(!config_prof);
+
+	active = true;
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0,
+	    "Unexpected mallctl failure while activating profiling");
+
+	prof_dump_open = prof_dump_open_intercept;
+
+	did_prof_dump_open = false;
+	p = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+
+	did_prof_dump_open = false;
+	q = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+
+	gdump = false;
+	sz = sizeof(gdump_old);
+	assert_d_eq(mallctl("prof.gdump", (void *)&gdump_old, &sz,
+	    (void *)&gdump, sizeof(gdump)), 0,
+	    "Unexpected mallctl failure while disabling prof.gdump");
+	assert(gdump_old);
+	did_prof_dump_open = false;
+	r = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+	assert_false(did_prof_dump_open, "Unexpected profile dump");
+
+	gdump = true;
+	sz = sizeof(gdump_old);
+	assert_d_eq(mallctl("prof.gdump", (void *)&gdump_old, &sz,
+	    (void *)&gdump, sizeof(gdump)), 0,
+	    "Unexpected mallctl failure while enabling prof.gdump");
+	assert(!gdump_old);
+	did_prof_dump_open = false;
+	s = mallocx((1U << SC_LG_LARGE_MINCLASS), 0);
+	assert_ptr_not_null(q, "Unexpected mallocx() failure");
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+
+	dallocx(p, 0);
+	dallocx(q, 0);
+	dallocx(r, 0);
+	dallocx(s, 0);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_gdump);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_gdump.sh b/memkind-1.10.0/jemalloc/test/unit/prof_gdump.sh
new file mode 100644
index 0000000..3f600d2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_gdump.sh
@@ -0,0 +1,6 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_active:false,prof_gdump:true"
+fi
+
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_idump.c b/memkind-1.10.0/jemalloc/test/unit/prof_idump.c
new file mode 100644
index 0000000..1cc6c98
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_idump.c
@@ -0,0 +1,42 @@
+#include "test/jemalloc_test.h"
+
+static bool did_prof_dump_open;
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	did_prof_dump_open = true;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+TEST_BEGIN(test_idump) {
+	bool active;
+	void *p;
+
+	test_skip_if(!config_prof);
+
+	active = true;
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0,
+	    "Unexpected mallctl failure while activating profiling");
+
+	prof_dump_open = prof_dump_open_intercept;
+
+	did_prof_dump_open = false;
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	dallocx(p, 0);
+	assert_true(did_prof_dump_open, "Expected a profile dump");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_idump);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_idump.sh b/memkind-1.10.0/jemalloc/test/unit/prof_idump.sh
new file mode 100644
index 0000000..4dc599a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_idump.sh
@@ -0,0 +1,8 @@
+#!/bin/sh
+
+export MALLOC_CONF="tcache:false"
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="${MALLOC_CONF},prof:true,prof_accum:true,prof_active:false,lg_prof_sample:0,lg_prof_interval:0"
+fi
+
+
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_log.c b/memkind-1.10.0/jemalloc/test/unit/prof_log.c
new file mode 100644
index 0000000..92fbd7c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_log.c
@@ -0,0 +1,148 @@
+#include "test/jemalloc_test.h"
+
+#define N_PARAM 100
+#define N_THREADS 10
+
+static void assert_rep() {
+	assert_b_eq(prof_log_rep_check(), false, "Rep check failed");
+}
+
+static void assert_log_empty() {
+	assert_zu_eq(prof_log_bt_count(), 0,
+	    "The log has backtraces; it isn't empty");
+	assert_zu_eq(prof_log_thr_count(), 0,
+	    "The log has threads; it isn't empty");
+	assert_zu_eq(prof_log_alloc_count(), 0,
+	    "The log has allocations; it isn't empty");
+}
+
+void *buf[N_PARAM];
+
+static void f() {
+	int i;
+	for (i = 0; i < N_PARAM; i++) {
+		buf[i] = malloc(100);
+	}
+	for (i = 0; i < N_PARAM; i++) {
+		free(buf[i]);
+	}
+}
+
+TEST_BEGIN(test_prof_log_many_logs) {
+	int i;
+
+	test_skip_if(!config_prof);
+
+	for (i = 0; i < N_PARAM; i++) {
+		assert_b_eq(prof_log_is_logging(), false,
+		    "Logging shouldn't have started yet");
+		assert_d_eq(mallctl("prof.log_start", NULL, NULL, NULL, 0), 0,
+		    "Unexpected mallctl failure when starting logging");
+		assert_b_eq(prof_log_is_logging(), true,
+		    "Logging should be started by now");
+		assert_log_empty();
+		assert_rep();
+		f();
+		assert_zu_eq(prof_log_thr_count(), 1, "Wrong thread count");
+		assert_rep();
+		assert_b_eq(prof_log_is_logging(), true,
+		    "Logging should still be on");
+		assert_d_eq(mallctl("prof.log_stop", NULL, NULL, NULL, 0), 0,
+		    "Unexpected mallctl failure when stopping logging");
+		assert_b_eq(prof_log_is_logging(), false,
+		    "Logging should have turned off");
+	}
+}
+TEST_END
+
+thd_t thr_buf[N_THREADS];
+
+static void *f_thread(void *unused) {
+	int i;
+	for (i = 0; i < N_PARAM; i++) {
+		void *p = malloc(100);
+		memset(p, 100, sizeof(char));
+		free(p);
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_prof_log_many_threads) {
+
+	test_skip_if(!config_prof);
+
+	int i;
+	assert_d_eq(mallctl("prof.log_start", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when starting logging");
+	for (i = 0; i < N_THREADS; i++) {
+		thd_create(&thr_buf[i], &f_thread, NULL);
+	}
+
+	for (i = 0; i < N_THREADS; i++) {
+		thd_join(thr_buf[i], NULL);
+	}
+	assert_zu_eq(prof_log_thr_count(), N_THREADS,
+	    "Wrong number of thread entries");
+	assert_rep();
+	assert_d_eq(mallctl("prof.log_stop", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when stopping logging");
+}
+TEST_END
+
+static void f3() {
+	void *p = malloc(100);
+	free(p);
+}
+
+static void f1() {
+	void *p = malloc(100);
+	f3();
+	free(p);
+}
+
+static void f2() {
+	void *p = malloc(100);
+	free(p);
+}
+
+TEST_BEGIN(test_prof_log_many_traces) {
+
+	test_skip_if(!config_prof);
+
+	assert_d_eq(mallctl("prof.log_start", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when starting logging");
+	int i;
+	assert_rep();
+	assert_log_empty();
+	for (i = 0; i < N_PARAM; i++) {
+		assert_rep();
+		f1();
+		assert_rep();
+		f2();
+		assert_rep();
+		f3();
+		assert_rep();
+	}
+	/*
+	 * There should be 8 total backtraces: two for malloc/free in f1(), two
+	 * for malloc/free in f2(), two for malloc/free in f3(), and then two
+	 * for malloc/free in f1()'s call to f3().  However compiler
+	 * optimizations such as loop unrolling might generate more call sites.
+	 * So >= 8 traces are expected.
+	 */
+	assert_zu_ge(prof_log_bt_count(), 8,
+	    "Expect at least 8 backtraces given sample workload");
+	assert_d_eq(mallctl("prof.log_stop", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl failure when stopping logging");
+}
+TEST_END
+
+int
+main(void) {
+	prof_log_dummy_set(true);
+	return test_no_reentrancy(
+	    test_prof_log_many_logs,
+	    test_prof_log_many_traces,
+	    test_prof_log_many_threads);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_log.sh b/memkind-1.10.0/jemalloc/test/unit/prof_log.sh
new file mode 100644
index 0000000..8fcc7d8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_log.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,lg_prof_sample:0"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_reset.c b/memkind-1.10.0/jemalloc/test/unit/prof_reset.c
new file mode 100644
index 0000000..7cce42d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_reset.c
@@ -0,0 +1,286 @@
+#include "test/jemalloc_test.h"
+
+static int
+prof_dump_open_intercept(bool propagate_err, const char *filename) {
+	int fd;
+
+	fd = open("/dev/null", O_WRONLY);
+	assert_d_ne(fd, -1, "Unexpected open() failure");
+
+	return fd;
+}
+
+static void
+set_prof_active(bool active) {
+	assert_d_eq(mallctl("prof.active", NULL, NULL, (void *)&active,
+	    sizeof(active)), 0, "Unexpected mallctl failure");
+}
+
+static size_t
+get_lg_prof_sample(void) {
+	size_t lg_prof_sample;
+	size_t sz = sizeof(size_t);
+
+	assert_d_eq(mallctl("prof.lg_sample", (void *)&lg_prof_sample, &sz,
+	    NULL, 0), 0,
+	    "Unexpected mallctl failure while reading profiling sample rate");
+	return lg_prof_sample;
+}
+
+static void
+do_prof_reset(size_t lg_prof_sample) {
+	assert_d_eq(mallctl("prof.reset", NULL, NULL,
+	    (void *)&lg_prof_sample, sizeof(size_t)), 0,
+	    "Unexpected mallctl failure while resetting profile data");
+	assert_zu_eq(lg_prof_sample, get_lg_prof_sample(),
+	    "Expected profile sample rate change");
+}
+
+TEST_BEGIN(test_prof_reset_basic) {
+	size_t lg_prof_sample_orig, lg_prof_sample, lg_prof_sample_next;
+	size_t sz;
+	unsigned i;
+
+	test_skip_if(!config_prof);
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("opt.lg_prof_sample", (void *)&lg_prof_sample_orig,
+	    &sz, NULL, 0), 0,
+	    "Unexpected mallctl failure while reading profiling sample rate");
+	assert_zu_eq(lg_prof_sample_orig, 0,
+	    "Unexpected profiling sample rate");
+	lg_prof_sample = get_lg_prof_sample();
+	assert_zu_eq(lg_prof_sample_orig, lg_prof_sample,
+	    "Unexpected disagreement between \"opt.lg_prof_sample\" and "
+	    "\"prof.lg_sample\"");
+
+	/* Test simple resets. */
+	for (i = 0; i < 2; i++) {
+		assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0), 0,
+		    "Unexpected mallctl failure while resetting profile data");
+		lg_prof_sample = get_lg_prof_sample();
+		assert_zu_eq(lg_prof_sample_orig, lg_prof_sample,
+		    "Unexpected profile sample rate change");
+	}
+
+	/* Test resets with prof.lg_sample changes. */
+	lg_prof_sample_next = 1;
+	for (i = 0; i < 2; i++) {
+		do_prof_reset(lg_prof_sample_next);
+		lg_prof_sample = get_lg_prof_sample();
+		assert_zu_eq(lg_prof_sample, lg_prof_sample_next,
+		    "Expected profile sample rate change");
+		lg_prof_sample_next = lg_prof_sample_orig;
+	}
+
+	/* Make sure the test code restored prof.lg_sample. */
+	lg_prof_sample = get_lg_prof_sample();
+	assert_zu_eq(lg_prof_sample_orig, lg_prof_sample,
+	    "Unexpected disagreement between \"opt.lg_prof_sample\" and "
+	    "\"prof.lg_sample\"");
+}
+TEST_END
+
+bool prof_dump_header_intercepted = false;
+prof_cnt_t cnt_all_copy = {0, 0, 0, 0};
+static bool
+prof_dump_header_intercept(tsdn_t *tsdn, bool propagate_err,
+    const prof_cnt_t *cnt_all) {
+	prof_dump_header_intercepted = true;
+	memcpy(&cnt_all_copy, cnt_all, sizeof(prof_cnt_t));
+
+	return false;
+}
+
+TEST_BEGIN(test_prof_reset_cleanup) {
+	void *p;
+	prof_dump_header_t *prof_dump_header_orig;
+
+	test_skip_if(!config_prof);
+
+	set_prof_active(true);
+
+	assert_zu_eq(prof_bt_count(), 0, "Expected 0 backtraces");
+	p = mallocx(1, 0);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	assert_zu_eq(prof_bt_count(), 1, "Expected 1 backtrace");
+
+	prof_dump_header_orig = prof_dump_header;
+	prof_dump_header = prof_dump_header_intercept;
+	assert_false(prof_dump_header_intercepted, "Unexpected intercept");
+
+	assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+	    0, "Unexpected error while dumping heap profile");
+	assert_true(prof_dump_header_intercepted, "Expected intercept");
+	assert_u64_eq(cnt_all_copy.curobjs, 1, "Expected 1 allocation");
+
+	assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0), 0,
+	    "Unexpected error while resetting heap profile data");
+	assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+	    0, "Unexpected error while dumping heap profile");
+	assert_u64_eq(cnt_all_copy.curobjs, 0, "Expected 0 allocations");
+	assert_zu_eq(prof_bt_count(), 1, "Expected 1 backtrace");
+
+	prof_dump_header = prof_dump_header_orig;
+
+	dallocx(p, 0);
+	assert_zu_eq(prof_bt_count(), 0, "Expected 0 backtraces");
+
+	set_prof_active(false);
+}
+TEST_END
+
+#define NTHREADS		4
+#define NALLOCS_PER_THREAD	(1U << 13)
+#define OBJ_RING_BUF_COUNT	1531
+#define RESET_INTERVAL		(1U << 10)
+#define DUMP_INTERVAL		3677
+static void *
+thd_start(void *varg) {
+	unsigned thd_ind = *(unsigned *)varg;
+	unsigned i;
+	void *objs[OBJ_RING_BUF_COUNT];
+
+	memset(objs, 0, sizeof(objs));
+
+	for (i = 0; i < NALLOCS_PER_THREAD; i++) {
+		if (i % RESET_INTERVAL == 0) {
+			assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0),
+			    0, "Unexpected error while resetting heap profile "
+			    "data");
+		}
+
+		if (i % DUMP_INTERVAL == 0) {
+			assert_d_eq(mallctl("prof.dump", NULL, NULL, NULL, 0),
+			    0, "Unexpected error while dumping heap profile");
+		}
+
+		{
+			void **pp = &objs[i % OBJ_RING_BUF_COUNT];
+			if (*pp != NULL) {
+				dallocx(*pp, 0);
+				*pp = NULL;
+			}
+			*pp = btalloc(1, thd_ind*NALLOCS_PER_THREAD + i);
+			assert_ptr_not_null(*pp,
+			    "Unexpected btalloc() failure");
+		}
+	}
+
+	/* Clean up any remaining objects. */
+	for (i = 0; i < OBJ_RING_BUF_COUNT; i++) {
+		void **pp = &objs[i % OBJ_RING_BUF_COUNT];
+		if (*pp != NULL) {
+			dallocx(*pp, 0);
+			*pp = NULL;
+		}
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_prof_reset) {
+	size_t lg_prof_sample_orig;
+	thd_t thds[NTHREADS];
+	unsigned thd_args[NTHREADS];
+	unsigned i;
+	size_t bt_count, tdata_count;
+
+	test_skip_if(!config_prof);
+
+	bt_count = prof_bt_count();
+	assert_zu_eq(bt_count, 0,
+	    "Unexpected pre-existing tdata structures");
+	tdata_count = prof_tdata_count();
+
+	lg_prof_sample_orig = get_lg_prof_sample();
+	do_prof_reset(5);
+
+	set_prof_active(true);
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_args[i] = i;
+		thd_create(&thds[i], thd_start, (void *)&thd_args[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+
+	assert_zu_eq(prof_bt_count(), bt_count,
+	    "Unexpected bactrace count change");
+	assert_zu_eq(prof_tdata_count(), tdata_count,
+	    "Unexpected remaining tdata structures");
+
+	set_prof_active(false);
+
+	do_prof_reset(lg_prof_sample_orig);
+}
+TEST_END
+#undef NTHREADS
+#undef NALLOCS_PER_THREAD
+#undef OBJ_RING_BUF_COUNT
+#undef RESET_INTERVAL
+#undef DUMP_INTERVAL
+
+/* Test sampling at the same allocation site across resets. */
+#define NITER 10
+TEST_BEGIN(test_xallocx) {
+	size_t lg_prof_sample_orig;
+	unsigned i;
+	void *ptrs[NITER];
+
+	test_skip_if(!config_prof);
+
+	lg_prof_sample_orig = get_lg_prof_sample();
+	set_prof_active(true);
+
+	/* Reset profiling. */
+	do_prof_reset(0);
+
+	for (i = 0; i < NITER; i++) {
+		void *p;
+		size_t sz, nsz;
+
+		/* Reset profiling. */
+		do_prof_reset(0);
+
+		/* Allocate small object (which will be promoted). */
+		p = ptrs[i] = mallocx(1, 0);
+		assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+		/* Reset profiling. */
+		do_prof_reset(0);
+
+		/* Perform successful xallocx(). */
+		sz = sallocx(p, 0);
+		assert_zu_eq(xallocx(p, sz, 0, 0), sz,
+		    "Unexpected xallocx() failure");
+
+		/* Perform unsuccessful xallocx(). */
+		nsz = nallocx(sz+1, 0);
+		assert_zu_eq(xallocx(p, nsz, 0, 0), sz,
+		    "Unexpected xallocx() success");
+	}
+
+	for (i = 0; i < NITER; i++) {
+		/* dallocx. */
+		dallocx(ptrs[i], 0);
+	}
+
+	set_prof_active(false);
+	do_prof_reset(lg_prof_sample_orig);
+}
+TEST_END
+#undef NITER
+
+int
+main(void) {
+	/* Intercept dumping prior to running any tests. */
+	prof_dump_open = prof_dump_open_intercept;
+
+	return test_no_reentrancy(
+	    test_prof_reset_basic,
+	    test_prof_reset_cleanup,
+	    test_prof_reset,
+	    test_xallocx);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_reset.sh b/memkind-1.10.0/jemalloc/test/unit/prof_reset.sh
new file mode 100644
index 0000000..43c516a
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_reset.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_active:false,lg_prof_sample:0"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_tctx.c b/memkind-1.10.0/jemalloc/test/unit/prof_tctx.c
new file mode 100644
index 0000000..ff3b2b0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_tctx.c
@@ -0,0 +1,46 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_prof_realloc) {
+	tsdn_t *tsdn;
+	int flags;
+	void *p, *q;
+	prof_tctx_t *tctx_p, *tctx_q;
+	uint64_t curobjs_0, curobjs_1, curobjs_2, curobjs_3;
+
+	test_skip_if(!config_prof);
+
+	tsdn = tsdn_fetch();
+	flags = MALLOCX_TCACHE_NONE;
+
+	prof_cnt_all(&curobjs_0, NULL, NULL, NULL);
+	p = mallocx(1024, flags);
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+	tctx_p = prof_tctx_get(tsdn, p, NULL);
+	assert_ptr_ne(tctx_p, (prof_tctx_t *)(uintptr_t)1U,
+	    "Expected valid tctx");
+	prof_cnt_all(&curobjs_1, NULL, NULL, NULL);
+	assert_u64_eq(curobjs_0 + 1, curobjs_1,
+	    "Allocation should have increased sample size");
+
+	q = rallocx(p, 2048, flags);
+	assert_ptr_ne(p, q, "Expected move");
+	assert_ptr_not_null(p, "Unexpected rmallocx() failure");
+	tctx_q = prof_tctx_get(tsdn, q, NULL);
+	assert_ptr_ne(tctx_q, (prof_tctx_t *)(uintptr_t)1U,
+	    "Expected valid tctx");
+	prof_cnt_all(&curobjs_2, NULL, NULL, NULL);
+	assert_u64_eq(curobjs_1, curobjs_2,
+	    "Reallocation should not have changed sample size");
+
+	dallocx(q, flags);
+	prof_cnt_all(&curobjs_3, NULL, NULL, NULL);
+	assert_u64_eq(curobjs_0, curobjs_3,
+	    "Sample size should have returned to base level");
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_prof_realloc);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_tctx.sh b/memkind-1.10.0/jemalloc/test/unit/prof_tctx.sh
new file mode 100644
index 0000000..8fcc7d8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_tctx.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,lg_prof_sample:0"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_thread_name.c b/memkind-1.10.0/jemalloc/test/unit/prof_thread_name.c
new file mode 100644
index 0000000..c9c2a2b
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_thread_name.c
@@ -0,0 +1,120 @@
+#include "test/jemalloc_test.h"
+
+static void
+mallctl_thread_name_get_impl(const char *thread_name_expected, const char *func,
+    int line) {
+	const char *thread_name_old;
+	size_t sz;
+
+	sz = sizeof(thread_name_old);
+	assert_d_eq(mallctl("thread.prof.name", (void *)&thread_name_old, &sz,
+	    NULL, 0), 0,
+	    "%s():%d: Unexpected mallctl failure reading thread.prof.name",
+	    func, line);
+	assert_str_eq(thread_name_old, thread_name_expected,
+	    "%s():%d: Unexpected thread.prof.name value", func, line);
+}
+#define mallctl_thread_name_get(a)					\
+	mallctl_thread_name_get_impl(a, __func__, __LINE__)
+
+static void
+mallctl_thread_name_set_impl(const char *thread_name, const char *func,
+    int line) {
+	assert_d_eq(mallctl("thread.prof.name", NULL, NULL,
+	    (void *)&thread_name, sizeof(thread_name)), 0,
+	    "%s():%d: Unexpected mallctl failure reading thread.prof.name",
+	    func, line);
+	mallctl_thread_name_get_impl(thread_name, func, line);
+}
+#define mallctl_thread_name_set(a)					\
+	mallctl_thread_name_set_impl(a, __func__, __LINE__)
+
+TEST_BEGIN(test_prof_thread_name_validation) {
+	const char *thread_name;
+
+	test_skip_if(!config_prof);
+
+	mallctl_thread_name_get("");
+	mallctl_thread_name_set("hi there");
+
+	/* NULL input shouldn't be allowed. */
+	thread_name = NULL;
+	assert_d_eq(mallctl("thread.prof.name", NULL, NULL,
+	    (void *)&thread_name, sizeof(thread_name)), EFAULT,
+	    "Unexpected mallctl result writing \"%s\" to thread.prof.name",
+	    thread_name);
+
+	/* '\n' shouldn't be allowed. */
+	thread_name = "hi\nthere";
+	assert_d_eq(mallctl("thread.prof.name", NULL, NULL,
+	    (void *)&thread_name, sizeof(thread_name)), EFAULT,
+	    "Unexpected mallctl result writing \"%s\" to thread.prof.name",
+	    thread_name);
+
+	/* Simultaneous read/write shouldn't be allowed. */
+	{
+		const char *thread_name_old;
+		size_t sz;
+
+		sz = sizeof(thread_name_old);
+		assert_d_eq(mallctl("thread.prof.name",
+		    (void *)&thread_name_old, &sz, (void *)&thread_name,
+		    sizeof(thread_name)), EPERM,
+		    "Unexpected mallctl result writing \"%s\" to "
+		    "thread.prof.name", thread_name);
+	}
+
+	mallctl_thread_name_set("");
+}
+TEST_END
+
+#define NTHREADS	4
+#define NRESET		25
+static void *
+thd_start(void *varg) {
+	unsigned thd_ind = *(unsigned *)varg;
+	char thread_name[16] = "";
+	unsigned i;
+
+	malloc_snprintf(thread_name, sizeof(thread_name), "thread %u", thd_ind);
+
+	mallctl_thread_name_get("");
+	mallctl_thread_name_set(thread_name);
+
+	for (i = 0; i < NRESET; i++) {
+		assert_d_eq(mallctl("prof.reset", NULL, NULL, NULL, 0), 0,
+		    "Unexpected error while resetting heap profile data");
+		mallctl_thread_name_get(thread_name);
+	}
+
+	mallctl_thread_name_set(thread_name);
+	mallctl_thread_name_set("");
+
+	return NULL;
+}
+
+TEST_BEGIN(test_prof_thread_name_threaded) {
+	thd_t thds[NTHREADS];
+	unsigned thd_args[NTHREADS];
+	unsigned i;
+
+	test_skip_if(!config_prof);
+
+	for (i = 0; i < NTHREADS; i++) {
+		thd_args[i] = i;
+		thd_create(&thds[i], thd_start, (void *)&thd_args[i]);
+	}
+	for (i = 0; i < NTHREADS; i++) {
+		thd_join(thds[i], NULL);
+	}
+}
+TEST_END
+#undef NTHREADS
+#undef NRESET
+
+int
+main(void) {
+	return test(
+	    test_prof_thread_name_validation,
+	    test_prof_thread_name_threaded);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/prof_thread_name.sh b/memkind-1.10.0/jemalloc/test/unit/prof_thread_name.sh
new file mode 100644
index 0000000..298c105
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/prof_thread_name.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,prof_active:false"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/ql.c b/memkind-1.10.0/jemalloc/test/unit/ql.c
new file mode 100644
index 0000000..b76c24c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/ql.c
@@ -0,0 +1,204 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ql.h"
+
+/* Number of ring entries, in [2..26]. */
+#define NENTRIES 9
+
+typedef struct list_s list_t;
+typedef ql_head(list_t) list_head_t;
+
+struct list_s {
+	ql_elm(list_t) link;
+	char id;
+};
+
+static void
+test_empty_list(list_head_t *head) {
+	list_t *t;
+	unsigned i;
+
+	assert_ptr_null(ql_first(head), "Unexpected element for empty list");
+	assert_ptr_null(ql_last(head, link),
+	    "Unexpected element for empty list");
+
+	i = 0;
+	ql_foreach(t, head, link) {
+		i++;
+	}
+	assert_u_eq(i, 0, "Unexpected element for empty list");
+
+	i = 0;
+	ql_reverse_foreach(t, head, link) {
+		i++;
+	}
+	assert_u_eq(i, 0, "Unexpected element for empty list");
+}
+
+TEST_BEGIN(test_ql_empty) {
+	list_head_t head;
+
+	ql_new(&head);
+	test_empty_list(&head);
+}
+TEST_END
+
+static void
+init_entries(list_t *entries, unsigned nentries) {
+	unsigned i;
+
+	for (i = 0; i < nentries; i++) {
+		entries[i].id = 'a' + i;
+		ql_elm_new(&entries[i], link);
+	}
+}
+
+static void
+test_entries_list(list_head_t *head, list_t *entries, unsigned nentries) {
+	list_t *t;
+	unsigned i;
+
+	assert_c_eq(ql_first(head)->id, entries[0].id, "Element id mismatch");
+	assert_c_eq(ql_last(head, link)->id, entries[nentries-1].id,
+	    "Element id mismatch");
+
+	i = 0;
+	ql_foreach(t, head, link) {
+		assert_c_eq(t->id, entries[i].id, "Element id mismatch");
+		i++;
+	}
+
+	i = 0;
+	ql_reverse_foreach(t, head, link) {
+		assert_c_eq(t->id, entries[nentries-i-1].id,
+		    "Element id mismatch");
+		i++;
+	}
+
+	for (i = 0; i < nentries-1; i++) {
+		t = ql_next(head, &entries[i], link);
+		assert_c_eq(t->id, entries[i+1].id, "Element id mismatch");
+	}
+	assert_ptr_null(ql_next(head, &entries[nentries-1], link),
+	    "Unexpected element");
+
+	assert_ptr_null(ql_prev(head, &entries[0], link), "Unexpected element");
+	for (i = 1; i < nentries; i++) {
+		t = ql_prev(head, &entries[i], link);
+		assert_c_eq(t->id, entries[i-1].id, "Element id mismatch");
+	}
+}
+
+TEST_BEGIN(test_ql_tail_insert) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_tail_insert(&head, &entries[i], link);
+	}
+
+	test_entries_list(&head, entries, NENTRIES);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_tail_remove) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_tail_insert(&head, &entries[i], link);
+	}
+
+	for (i = 0; i < NENTRIES; i++) {
+		test_entries_list(&head, entries, NENTRIES-i);
+		ql_tail_remove(&head, list_t, link);
+	}
+	test_empty_list(&head);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_head_insert) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_head_insert(&head, &entries[NENTRIES-i-1], link);
+	}
+
+	test_entries_list(&head, entries, NENTRIES);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_head_remove) {
+	list_head_t head;
+	list_t entries[NENTRIES];
+	unsigned i;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	for (i = 0; i < NENTRIES; i++) {
+		ql_head_insert(&head, &entries[NENTRIES-i-1], link);
+	}
+
+	for (i = 0; i < NENTRIES; i++) {
+		test_entries_list(&head, &entries[i], NENTRIES-i);
+		ql_head_remove(&head, list_t, link);
+	}
+	test_empty_list(&head);
+}
+TEST_END
+
+TEST_BEGIN(test_ql_insert) {
+	list_head_t head;
+	list_t entries[8];
+	list_t *a, *b, *c, *d, *e, *f, *g, *h;
+
+	ql_new(&head);
+	init_entries(entries, sizeof(entries)/sizeof(list_t));
+	a = &entries[0];
+	b = &entries[1];
+	c = &entries[2];
+	d = &entries[3];
+	e = &entries[4];
+	f = &entries[5];
+	g = &entries[6];
+	h = &entries[7];
+
+	/*
+	 * ql_remove(), ql_before_insert(), and ql_after_insert() are used
+	 * internally by other macros that are already tested, so there's no
+	 * need to test them completely.  However, insertion/deletion from the
+	 * middle of lists is not otherwise tested; do so here.
+	 */
+	ql_tail_insert(&head, f, link);
+	ql_before_insert(&head, f, b, link);
+	ql_before_insert(&head, f, c, link);
+	ql_after_insert(f, h, link);
+	ql_after_insert(f, g, link);
+	ql_before_insert(&head, b, a, link);
+	ql_after_insert(c, d, link);
+	ql_before_insert(&head, f, e, link);
+
+	test_entries_list(&head, entries, sizeof(entries)/sizeof(list_t));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ql_empty,
+	    test_ql_tail_insert,
+	    test_ql_tail_remove,
+	    test_ql_head_insert,
+	    test_ql_head_remove,
+	    test_ql_insert);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/qr.c b/memkind-1.10.0/jemalloc/test/unit/qr.c
new file mode 100644
index 0000000..271a109
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/qr.c
@@ -0,0 +1,243 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/qr.h"
+
+/* Number of ring entries, in [2..26]. */
+#define NENTRIES 9
+/* Split index, in [1..NENTRIES). */
+#define SPLIT_INDEX 5
+
+typedef struct ring_s ring_t;
+
+struct ring_s {
+	qr(ring_t) link;
+	char id;
+};
+
+static void
+init_entries(ring_t *entries) {
+	unsigned i;
+
+	for (i = 0; i < NENTRIES; i++) {
+		qr_new(&entries[i], link);
+		entries[i].id = 'a' + i;
+	}
+}
+
+static void
+test_independent_entries(ring_t *entries) {
+	ring_t *t;
+	unsigned i, j;
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			j++;
+		}
+		assert_u_eq(j, 1,
+		    "Iteration over single-element ring should visit precisely "
+		    "one element");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			j++;
+		}
+		assert_u_eq(j, 1,
+		    "Iteration over single-element ring should visit precisely "
+		    "one element");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_next(&entries[i], link);
+		assert_ptr_eq(t, &entries[i],
+		    "Next element in single-element ring should be same as "
+		    "current element");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_prev(&entries[i], link);
+		assert_ptr_eq(t, &entries[i],
+		    "Previous element in single-element ring should be same as "
+		    "current element");
+	}
+}
+
+TEST_BEGIN(test_qr_one) {
+	ring_t entries[NENTRIES];
+
+	init_entries(entries);
+	test_independent_entries(entries);
+}
+TEST_END
+
+static void
+test_entries_ring(ring_t *entries) {
+	ring_t *t;
+	unsigned i, j;
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(i+j) % NENTRIES].id,
+			    "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(NENTRIES+i-j-1) %
+			    NENTRIES].id, "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_next(&entries[i], link);
+		assert_c_eq(t->id, entries[(i+1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_prev(&entries[i], link);
+		assert_c_eq(t->id, entries[(NENTRIES+i-1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+}
+
+TEST_BEGIN(test_qr_after_insert) {
+	ring_t entries[NENTRIES];
+	unsigned i;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_after_insert(&entries[i - 1], &entries[i], link);
+	}
+	test_entries_ring(entries);
+}
+TEST_END
+
+TEST_BEGIN(test_qr_remove) {
+	ring_t entries[NENTRIES];
+	ring_t *t;
+	unsigned i, j;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_after_insert(&entries[i - 1], &entries[i], link);
+	}
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[i+j].id,
+			    "Element id mismatch");
+			j++;
+		}
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[NENTRIES - 1 - j].id,
+			"Element id mismatch");
+			j++;
+		}
+		qr_remove(&entries[i], link);
+	}
+	test_independent_entries(entries);
+}
+TEST_END
+
+TEST_BEGIN(test_qr_before_insert) {
+	ring_t entries[NENTRIES];
+	ring_t *t;
+	unsigned i, j;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_before_insert(&entries[i - 1], &entries[i], link);
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(NENTRIES+i-j) %
+			    NENTRIES].id, "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_reverse_foreach(t, &entries[i], link) {
+			assert_c_eq(t->id, entries[(i+j+1) % NENTRIES].id,
+			    "Element id mismatch");
+			j++;
+		}
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_next(&entries[i], link);
+		assert_c_eq(t->id, entries[(NENTRIES+i-1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+	for (i = 0; i < NENTRIES; i++) {
+		t = qr_prev(&entries[i], link);
+		assert_c_eq(t->id, entries[(i+1) % NENTRIES].id,
+		    "Element id mismatch");
+	}
+}
+TEST_END
+
+static void
+test_split_entries(ring_t *entries) {
+	ring_t *t;
+	unsigned i, j;
+
+	for (i = 0; i < NENTRIES; i++) {
+		j = 0;
+		qr_foreach(t, &entries[i], link) {
+			if (i < SPLIT_INDEX) {
+				assert_c_eq(t->id,
+				    entries[(i+j) % SPLIT_INDEX].id,
+				    "Element id mismatch");
+			} else {
+				assert_c_eq(t->id, entries[(i+j-SPLIT_INDEX) %
+				    (NENTRIES-SPLIT_INDEX) + SPLIT_INDEX].id,
+				    "Element id mismatch");
+			}
+			j++;
+		}
+	}
+}
+
+TEST_BEGIN(test_qr_meld_split) {
+	ring_t entries[NENTRIES];
+	unsigned i;
+
+	init_entries(entries);
+	for (i = 1; i < NENTRIES; i++) {
+		qr_after_insert(&entries[i - 1], &entries[i], link);
+	}
+
+	qr_split(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_split_entries(entries);
+
+	qr_meld(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_entries_ring(entries);
+
+	qr_meld(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_split_entries(entries);
+
+	qr_split(&entries[0], &entries[SPLIT_INDEX], ring_t, link);
+	test_entries_ring(entries);
+
+	qr_split(&entries[0], &entries[0], ring_t, link);
+	test_entries_ring(entries);
+
+	qr_meld(&entries[0], &entries[0], ring_t, link);
+	test_entries_ring(entries);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_qr_one,
+	    test_qr_after_insert,
+	    test_qr_remove,
+	    test_qr_before_insert,
+	    test_qr_meld_split);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/rb.c b/memkind-1.10.0/jemalloc/test/unit/rb.c
new file mode 100644
index 0000000..65c0492
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/rb.c
@@ -0,0 +1,355 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/rb.h"
+
+#define rbtn_black_height(a_type, a_field, a_rbt, r_height) do {	\
+	a_type *rbp_bh_t;						\
+	for (rbp_bh_t = (a_rbt)->rbt_root, (r_height) = 0; rbp_bh_t !=	\
+	    NULL; rbp_bh_t = rbtn_left_get(a_type, a_field,		\
+	    rbp_bh_t)) {						\
+		if (!rbtn_red_get(a_type, a_field, rbp_bh_t)) {		\
+		(r_height)++;						\
+		}							\
+	}								\
+} while (0)
+
+typedef struct node_s node_t;
+
+struct node_s {
+#define NODE_MAGIC 0x9823af7e
+	uint32_t magic;
+	rb_node(node_t) link;
+	uint64_t key;
+};
+
+static int
+node_cmp(const node_t *a, const node_t *b) {
+	int ret;
+
+	assert_u32_eq(a->magic, NODE_MAGIC, "Bad magic");
+	assert_u32_eq(b->magic, NODE_MAGIC, "Bad magic");
+
+	ret = (a->key > b->key) - (a->key < b->key);
+	if (ret == 0) {
+		/*
+		 * Duplicates are not allowed in the tree, so force an
+		 * arbitrary ordering for non-identical items with equal keys.
+		 */
+		ret = (((uintptr_t)a) > ((uintptr_t)b))
+		    - (((uintptr_t)a) < ((uintptr_t)b));
+	}
+	return ret;
+}
+
+typedef rb_tree(node_t) tree_t;
+rb_gen(static, tree_, tree_t, node_t, link, node_cmp);
+
+TEST_BEGIN(test_rb_empty) {
+	tree_t tree;
+	node_t key;
+
+	tree_new(&tree);
+
+	assert_true(tree_empty(&tree), "Tree should be empty");
+	assert_ptr_null(tree_first(&tree), "Unexpected node");
+	assert_ptr_null(tree_last(&tree), "Unexpected node");
+
+	key.key = 0;
+	key.magic = NODE_MAGIC;
+	assert_ptr_null(tree_search(&tree, &key), "Unexpected node");
+
+	key.key = 0;
+	key.magic = NODE_MAGIC;
+	assert_ptr_null(tree_nsearch(&tree, &key), "Unexpected node");
+
+	key.key = 0;
+	key.magic = NODE_MAGIC;
+	assert_ptr_null(tree_psearch(&tree, &key), "Unexpected node");
+}
+TEST_END
+
+static unsigned
+tree_recurse(node_t *node, unsigned black_height, unsigned black_depth) {
+	unsigned ret = 0;
+	node_t *left_node;
+	node_t *right_node;
+
+	if (node == NULL) {
+		return ret;
+	}
+
+	left_node = rbtn_left_get(node_t, link, node);
+	right_node = rbtn_right_get(node_t, link, node);
+
+	if (!rbtn_red_get(node_t, link, node)) {
+		black_depth++;
+	}
+
+	/* Red nodes must be interleaved with black nodes. */
+	if (rbtn_red_get(node_t, link, node)) {
+		if (left_node != NULL) {
+			assert_false(rbtn_red_get(node_t, link, left_node),
+				"Node should be black");
+		}
+		if (right_node != NULL) {
+			assert_false(rbtn_red_get(node_t, link, right_node),
+			    "Node should be black");
+		}
+	}
+
+	/* Self. */
+	assert_u32_eq(node->magic, NODE_MAGIC, "Bad magic");
+
+	/* Left subtree. */
+	if (left_node != NULL) {
+		ret += tree_recurse(left_node, black_height, black_depth);
+	} else {
+		ret += (black_depth != black_height);
+	}
+
+	/* Right subtree. */
+	if (right_node != NULL) {
+		ret += tree_recurse(right_node, black_height, black_depth);
+	} else {
+		ret += (black_depth != black_height);
+	}
+
+	return ret;
+}
+
+static node_t *
+tree_iterate_cb(tree_t *tree, node_t *node, void *data) {
+	unsigned *i = (unsigned *)data;
+	node_t *search_node;
+
+	assert_u32_eq(node->magic, NODE_MAGIC, "Bad magic");
+
+	/* Test rb_search(). */
+	search_node = tree_search(tree, node);
+	assert_ptr_eq(search_node, node,
+	    "tree_search() returned unexpected node");
+
+	/* Test rb_nsearch(). */
+	search_node = tree_nsearch(tree, node);
+	assert_ptr_eq(search_node, node,
+	    "tree_nsearch() returned unexpected node");
+
+	/* Test rb_psearch(). */
+	search_node = tree_psearch(tree, node);
+	assert_ptr_eq(search_node, node,
+	    "tree_psearch() returned unexpected node");
+
+	(*i)++;
+
+	return NULL;
+}
+
+static unsigned
+tree_iterate(tree_t *tree) {
+	unsigned i;
+
+	i = 0;
+	tree_iter(tree, NULL, tree_iterate_cb, (void *)&i);
+
+	return i;
+}
+
+static unsigned
+tree_iterate_reverse(tree_t *tree) {
+	unsigned i;
+
+	i = 0;
+	tree_reverse_iter(tree, NULL, tree_iterate_cb, (void *)&i);
+
+	return i;
+}
+
+static void
+node_remove(tree_t *tree, node_t *node, unsigned nnodes) {
+	node_t *search_node;
+	unsigned black_height, imbalances;
+
+	tree_remove(tree, node);
+
+	/* Test rb_nsearch(). */
+	search_node = tree_nsearch(tree, node);
+	if (search_node != NULL) {
+		assert_u64_ge(search_node->key, node->key,
+		    "Key ordering error");
+	}
+
+	/* Test rb_psearch(). */
+	search_node = tree_psearch(tree, node);
+	if (search_node != NULL) {
+		assert_u64_le(search_node->key, node->key,
+		    "Key ordering error");
+	}
+
+	node->magic = 0;
+
+	rbtn_black_height(node_t, link, tree, black_height);
+	imbalances = tree_recurse(tree->rbt_root, black_height, 0);
+	assert_u_eq(imbalances, 0, "Tree is unbalanced");
+	assert_u_eq(tree_iterate(tree), nnodes-1,
+	    "Unexpected node iteration count");
+	assert_u_eq(tree_iterate_reverse(tree), nnodes-1,
+	    "Unexpected node iteration count");
+}
+
+static node_t *
+remove_iterate_cb(tree_t *tree, node_t *node, void *data) {
+	unsigned *nnodes = (unsigned *)data;
+	node_t *ret = tree_next(tree, node);
+
+	node_remove(tree, node, *nnodes);
+
+	return ret;
+}
+
+static node_t *
+remove_reverse_iterate_cb(tree_t *tree, node_t *node, void *data) {
+	unsigned *nnodes = (unsigned *)data;
+	node_t *ret = tree_prev(tree, node);
+
+	node_remove(tree, node, *nnodes);
+
+	return ret;
+}
+
+static void
+destroy_cb(node_t *node, void *data) {
+	unsigned *nnodes = (unsigned *)data;
+
+	assert_u_gt(*nnodes, 0, "Destruction removed too many nodes");
+	(*nnodes)--;
+}
+
+TEST_BEGIN(test_rb_random) {
+#define NNODES 25
+#define NBAGS 250
+#define SEED 42
+	sfmt_t *sfmt;
+	uint64_t bag[NNODES];
+	tree_t tree;
+	node_t nodes[NNODES];
+	unsigned i, j, k, black_height, imbalances;
+
+	sfmt = init_gen_rand(SEED);
+	for (i = 0; i < NBAGS; i++) {
+		switch (i) {
+		case 0:
+			/* Insert in order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = j;
+			}
+			break;
+		case 1:
+			/* Insert in reverse order. */
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = NNODES - j - 1;
+			}
+			break;
+		default:
+			for (j = 0; j < NNODES; j++) {
+				bag[j] = gen_rand64_range(sfmt, NNODES);
+			}
+		}
+
+		for (j = 1; j <= NNODES; j++) {
+			/* Initialize tree and nodes. */
+			tree_new(&tree);
+			for (k = 0; k < j; k++) {
+				nodes[k].magic = NODE_MAGIC;
+				nodes[k].key = bag[k];
+			}
+
+			/* Insert nodes. */
+			for (k = 0; k < j; k++) {
+				tree_insert(&tree, &nodes[k]);
+
+				rbtn_black_height(node_t, link, &tree,
+				    black_height);
+				imbalances = tree_recurse(tree.rbt_root,
+				    black_height, 0);
+				assert_u_eq(imbalances, 0,
+				    "Tree is unbalanced");
+
+				assert_u_eq(tree_iterate(&tree), k+1,
+				    "Unexpected node iteration count");
+				assert_u_eq(tree_iterate_reverse(&tree), k+1,
+				    "Unexpected node iteration count");
+
+				assert_false(tree_empty(&tree),
+				    "Tree should not be empty");
+				assert_ptr_not_null(tree_first(&tree),
+				    "Tree should not be empty");
+				assert_ptr_not_null(tree_last(&tree),
+				    "Tree should not be empty");
+
+				tree_next(&tree, &nodes[k]);
+				tree_prev(&tree, &nodes[k]);
+			}
+
+			/* Remove nodes. */
+			switch (i % 5) {
+			case 0:
+				for (k = 0; k < j; k++) {
+					node_remove(&tree, &nodes[k], j - k);
+				}
+				break;
+			case 1:
+				for (k = j; k > 0; k--) {
+					node_remove(&tree, &nodes[k-1], k);
+				}
+				break;
+			case 2: {
+				node_t *start;
+				unsigned nnodes = j;
+
+				start = NULL;
+				do {
+					start = tree_iter(&tree, start,
+					    remove_iterate_cb, (void *)&nnodes);
+					nnodes--;
+				} while (start != NULL);
+				assert_u_eq(nnodes, 0,
+				    "Removal terminated early");
+				break;
+			} case 3: {
+				node_t *start;
+				unsigned nnodes = j;
+
+				start = NULL;
+				do {
+					start = tree_reverse_iter(&tree, start,
+					    remove_reverse_iterate_cb,
+					    (void *)&nnodes);
+					nnodes--;
+				} while (start != NULL);
+				assert_u_eq(nnodes, 0,
+				    "Removal terminated early");
+				break;
+			} case 4: {
+				unsigned nnodes = j;
+				tree_destroy(&tree, destroy_cb, &nnodes);
+				assert_u_eq(nnodes, 0,
+				    "Destruction terminated early");
+				break;
+			} default:
+				not_reached();
+			}
+		}
+	}
+	fini_gen_rand(sfmt);
+#undef NNODES
+#undef NBAGS
+#undef SEED
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_rb_empty,
+	    test_rb_random);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/retained.c b/memkind-1.10.0/jemalloc/test/unit/retained.c
new file mode 100644
index 0000000..7993fd3
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/retained.c
@@ -0,0 +1,184 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/spin.h"
+
+static unsigned		arena_ind;
+static size_t		sz;
+static size_t		esz;
+#define NEPOCHS		8
+#define PER_THD_NALLOCS	1
+static atomic_u_t	epoch;
+static atomic_u_t	nfinished;
+
+static unsigned
+do_arena_create(extent_hooks_t *h) {
+	unsigned arena_ind;
+	size_t sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz,
+	    (void *)(h != NULL ? &h : NULL), (h != NULL ? sizeof(h) : 0)), 0,
+	    "Unexpected mallctl() failure");
+	return arena_ind;
+}
+
+static void
+do_arena_destroy(unsigned arena_ind) {
+	size_t mib[3];
+	size_t miblen;
+
+	miblen = sizeof(mib)/sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arena.0.destroy", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() failure");
+	mib[1] = (size_t)arena_ind;
+	assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctlbymib() failure");
+}
+
+static void
+do_refresh(void) {
+	uint64_t epoch = 1;
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch,
+	    sizeof(epoch)), 0, "Unexpected mallctl() failure");
+}
+
+static size_t
+do_get_size_impl(const char *cmd, unsigned arena_ind) {
+	size_t mib[4];
+	size_t miblen = sizeof(mib) / sizeof(size_t);
+	size_t z = sizeof(size_t);
+
+	assert_d_eq(mallctlnametomib(cmd, mib, &miblen),
+	    0, "Unexpected mallctlnametomib(\"%s\", ...) failure", cmd);
+	mib[2] = arena_ind;
+	size_t size;
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&size, &z, NULL, 0),
+	    0, "Unexpected mallctlbymib([\"%s\"], ...) failure", cmd);
+
+	return size;
+}
+
+static size_t
+do_get_active(unsigned arena_ind) {
+	return do_get_size_impl("stats.arenas.0.pactive", arena_ind) * PAGE;
+}
+
+static size_t
+do_get_mapped(unsigned arena_ind) {
+	return do_get_size_impl("stats.arenas.0.mapped", arena_ind);
+}
+
+static void *
+thd_start(void *arg) {
+	for (unsigned next_epoch = 1; next_epoch < NEPOCHS; next_epoch++) {
+		/* Busy-wait for next epoch. */
+		unsigned cur_epoch;
+		spin_t spinner = SPIN_INITIALIZER;
+		while ((cur_epoch = atomic_load_u(&epoch, ATOMIC_ACQUIRE)) !=
+		    next_epoch) {
+			spin_adaptive(&spinner);
+		}
+		assert_u_eq(cur_epoch, next_epoch, "Unexpected epoch");
+
+		/*
+		 * Allocate.  The main thread will reset the arena, so there's
+		 * no need to deallocate.
+		 */
+		for (unsigned i = 0; i < PER_THD_NALLOCS; i++) {
+			void *p = mallocx(sz, MALLOCX_ARENA(arena_ind) |
+			    MALLOCX_TCACHE_NONE
+			    );
+			assert_ptr_not_null(p,
+			    "Unexpected mallocx() failure\n");
+		}
+
+		/* Let the main thread know we've finished this iteration. */
+		atomic_fetch_add_u(&nfinished, 1, ATOMIC_RELEASE);
+	}
+
+	return NULL;
+}
+
+TEST_BEGIN(test_retained) {
+	test_skip_if(!config_stats);
+
+	arena_ind = do_arena_create(NULL);
+	sz = nallocx(HUGEPAGE, 0);
+	esz = sz + sz_large_pad;
+
+	atomic_store_u(&epoch, 0, ATOMIC_RELAXED);
+
+	unsigned nthreads = ncpus * 2;
+	if (LG_SIZEOF_PTR < 3 && nthreads > 16) {
+		nthreads = 16; /* 32-bit platform could run out of vaddr. */
+	}
+	VARIABLE_ARRAY(thd_t, threads, nthreads);
+	for (unsigned i = 0; i < nthreads; i++) {
+		thd_create(&threads[i], thd_start, NULL);
+	}
+
+	for (unsigned e = 1; e < NEPOCHS; e++) {
+		atomic_store_u(&nfinished, 0, ATOMIC_RELEASE);
+		atomic_store_u(&epoch, e, ATOMIC_RELEASE);
+
+		/* Wait for threads to finish allocating. */
+		spin_t spinner = SPIN_INITIALIZER;
+		while (atomic_load_u(&nfinished, ATOMIC_ACQUIRE) < nthreads) {
+			spin_adaptive(&spinner);
+		}
+
+		/*
+		 * Assert that retained is no more than the sum of size classes
+		 * that should have been used to satisfy the worker threads'
+		 * requests, discounting per growth fragmentation.
+		 */
+		do_refresh();
+
+		size_t allocated = esz * nthreads * PER_THD_NALLOCS;
+		size_t active = do_get_active(arena_ind);
+		assert_zu_le(allocated, active, "Unexpected active memory");
+		size_t mapped = do_get_mapped(arena_ind);
+		assert_zu_le(active, mapped, "Unexpected mapped memory");
+
+		arena_t *arena = arena_get(tsdn_fetch(), arena_ind, false);
+		size_t usable = 0;
+		size_t fragmented = 0;
+		for (pszind_t pind = sz_psz2ind(HUGEPAGE); pind <
+		    arena->extent_grow_next; pind++) {
+			size_t psz = sz_pind2sz(pind);
+			size_t psz_fragmented = psz % esz;
+			size_t psz_usable = psz - psz_fragmented;
+			/*
+			 * Only consider size classes that wouldn't be skipped.
+			 */
+			if (psz_usable > 0) {
+				assert_zu_lt(usable, allocated,
+				    "Excessive retained memory "
+				    "(%#zx[+%#zx] > %#zx)", usable, psz_usable,
+				    allocated);
+				fragmented += psz_fragmented;
+				usable += psz_usable;
+			}
+		}
+
+		/*
+		 * Clean up arena.  Destroying and recreating the arena
+		 * is simpler that specifying extent hooks that deallocate
+		 * (rather than retaining) during reset.
+		 */
+		do_arena_destroy(arena_ind);
+		assert_u_eq(do_arena_create(NULL), arena_ind,
+		    "Unexpected arena index");
+	}
+
+	for (unsigned i = 0; i < nthreads; i++) {
+		thd_join(threads[i], NULL);
+	}
+
+	do_arena_destroy(arena_ind);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_retained);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/rtree.c b/memkind-1.10.0/jemalloc/test/unit/rtree.c
new file mode 100644
index 0000000..90adca1
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/rtree.c
@@ -0,0 +1,228 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/rtree.h"
+
+rtree_node_alloc_t *rtree_node_alloc_orig;
+rtree_node_dalloc_t *rtree_node_dalloc_orig;
+rtree_leaf_alloc_t *rtree_leaf_alloc_orig;
+rtree_leaf_dalloc_t *rtree_leaf_dalloc_orig;
+
+/* Potentially too large to safely place on the stack. */
+rtree_t test_rtree;
+
+static rtree_node_elm_t *
+rtree_node_alloc_intercept(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	rtree_node_elm_t *node;
+
+	if (rtree != &test_rtree) {
+		return rtree_node_alloc_orig(tsdn, rtree, nelms);
+	}
+
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+	node = (rtree_node_elm_t *)calloc(nelms, sizeof(rtree_node_elm_t));
+	assert_ptr_not_null(node, "Unexpected calloc() failure");
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+
+	return node;
+}
+
+static void
+rtree_node_dalloc_intercept(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_node_elm_t *node) {
+	if (rtree != &test_rtree) {
+		rtree_node_dalloc_orig(tsdn, rtree, node);
+		return;
+	}
+
+	free(node);
+}
+
+static rtree_leaf_elm_t *
+rtree_leaf_alloc_intercept(tsdn_t *tsdn, rtree_t *rtree, size_t nelms) {
+	rtree_leaf_elm_t *leaf;
+
+	if (rtree != &test_rtree) {
+		return rtree_leaf_alloc_orig(tsdn, rtree, nelms);
+	}
+
+	malloc_mutex_unlock(tsdn, &rtree->init_lock);
+	leaf = (rtree_leaf_elm_t *)calloc(nelms, sizeof(rtree_leaf_elm_t));
+	assert_ptr_not_null(leaf, "Unexpected calloc() failure");
+	malloc_mutex_lock(tsdn, &rtree->init_lock);
+
+	return leaf;
+}
+
+static void
+rtree_leaf_dalloc_intercept(tsdn_t *tsdn, rtree_t *rtree,
+    rtree_leaf_elm_t *leaf) {
+	if (rtree != &test_rtree) {
+		rtree_leaf_dalloc_orig(tsdn, rtree, leaf);
+		return;
+	}
+
+	free(leaf);
+}
+
+TEST_BEGIN(test_rtree_read_empty) {
+	tsdn_t *tsdn;
+
+	tsdn = tsdn_fetch();
+
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+	assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx, PAGE,
+	    false), "rtree_extent_read() should return NULL for empty tree");
+	rtree_delete(tsdn, rtree);
+}
+TEST_END
+
+#undef NTHREADS
+#undef NITERS
+#undef SEED
+
+TEST_BEGIN(test_rtree_extrema) {
+	extent_t extent_a, extent_b;
+	extent_init(&extent_a, NULL, NULL, SC_LARGE_MINCLASS, false,
+	    sz_size2index(SC_LARGE_MINCLASS), 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+	extent_init(&extent_b, NULL, NULL, 0, false, SC_NSIZES, 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+
+	tsdn_t *tsdn = tsdn_fetch();
+
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+
+	assert_false(rtree_write(tsdn, rtree, &rtree_ctx, PAGE, &extent_a,
+	    extent_szind_get(&extent_a), extent_slab_get(&extent_a)),
+	    "Unexpected rtree_write() failure");
+	rtree_szind_slab_update(tsdn, rtree, &rtree_ctx, PAGE,
+	    extent_szind_get(&extent_a), extent_slab_get(&extent_a));
+	assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx, PAGE, true),
+	    &extent_a,
+	    "rtree_extent_read() should return previously set value");
+
+	assert_false(rtree_write(tsdn, rtree, &rtree_ctx, ~((uintptr_t)0),
+	    &extent_b, extent_szind_get_maybe_invalid(&extent_b),
+	    extent_slab_get(&extent_b)), "Unexpected rtree_write() failure");
+	assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+	    ~((uintptr_t)0), true), &extent_b,
+	    "rtree_extent_read() should return previously set value");
+
+	rtree_delete(tsdn, rtree);
+}
+TEST_END
+
+TEST_BEGIN(test_rtree_bits) {
+	tsdn_t *tsdn = tsdn_fetch();
+
+	uintptr_t keys[] = {PAGE, PAGE + 1,
+	    PAGE + (((uintptr_t)1) << LG_PAGE) - 1};
+
+	extent_t extent;
+	extent_init(&extent, NULL, NULL, 0, false, SC_NSIZES, 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+
+	for (unsigned i = 0; i < sizeof(keys)/sizeof(uintptr_t); i++) {
+		assert_false(rtree_write(tsdn, rtree, &rtree_ctx, keys[i],
+		    &extent, SC_NSIZES, false),
+		    "Unexpected rtree_write() failure");
+		for (unsigned j = 0; j < sizeof(keys)/sizeof(uintptr_t); j++) {
+			assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+			    keys[j], true), &extent,
+			    "rtree_extent_read() should return previously set "
+			    "value and ignore insignificant key bits; i=%u, "
+			    "j=%u, set key=%#"FMTxPTR", get key=%#"FMTxPTR, i,
+			    j, keys[i], keys[j]);
+		}
+		assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    (((uintptr_t)2) << LG_PAGE), false),
+		    "Only leftmost rtree leaf should be set; i=%u", i);
+		rtree_clear(tsdn, rtree, &rtree_ctx, keys[i]);
+	}
+
+	rtree_delete(tsdn, rtree);
+}
+TEST_END
+
+TEST_BEGIN(test_rtree_random) {
+#define NSET 16
+#define SEED 42
+	sfmt_t *sfmt = init_gen_rand(SEED);
+	tsdn_t *tsdn = tsdn_fetch();
+	uintptr_t keys[NSET];
+	rtree_t *rtree = &test_rtree;
+	rtree_ctx_t rtree_ctx;
+	rtree_ctx_data_init(&rtree_ctx);
+
+	extent_t extent;
+	extent_init(&extent, NULL, NULL, 0, false, SC_NSIZES, 0,
+	    extent_state_active, false, false, true, EXTENT_NOT_HEAD);
+
+	assert_false(rtree_new(rtree, false), "Unexpected rtree_new() failure");
+
+	for (unsigned i = 0; i < NSET; i++) {
+		keys[i] = (uintptr_t)gen_rand64(sfmt);
+		rtree_leaf_elm_t *elm = rtree_leaf_elm_lookup(tsdn, rtree,
+		    &rtree_ctx, keys[i], false, true);
+		assert_ptr_not_null(elm,
+		    "Unexpected rtree_leaf_elm_lookup() failure");
+		rtree_leaf_elm_write(tsdn, rtree, elm, &extent, SC_NSIZES,
+		    false);
+		assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true), &extent,
+		    "rtree_extent_read() should return previously set value");
+	}
+	for (unsigned i = 0; i < NSET; i++) {
+		assert_ptr_eq(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true), &extent,
+		    "rtree_extent_read() should return previously set value, "
+		    "i=%u", i);
+	}
+
+	for (unsigned i = 0; i < NSET; i++) {
+		rtree_clear(tsdn, rtree, &rtree_ctx, keys[i]);
+		assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true),
+		   "rtree_extent_read() should return previously set value");
+	}
+	for (unsigned i = 0; i < NSET; i++) {
+		assert_ptr_null(rtree_extent_read(tsdn, rtree, &rtree_ctx,
+		    keys[i], true),
+		    "rtree_extent_read() should return previously set value");
+	}
+
+	rtree_delete(tsdn, rtree);
+	fini_gen_rand(sfmt);
+#undef NSET
+#undef SEED
+}
+TEST_END
+
+int
+main(void) {
+	rtree_node_alloc_orig = rtree_node_alloc;
+	rtree_node_alloc = rtree_node_alloc_intercept;
+	rtree_node_dalloc_orig = rtree_node_dalloc;
+	rtree_node_dalloc = rtree_node_dalloc_intercept;
+	rtree_leaf_alloc_orig = rtree_leaf_alloc;
+	rtree_leaf_alloc = rtree_leaf_alloc_intercept;
+	rtree_leaf_dalloc_orig = rtree_leaf_dalloc;
+	rtree_leaf_dalloc = rtree_leaf_dalloc_intercept;
+
+	return test(
+	    test_rtree_read_empty,
+	    test_rtree_extrema,
+	    test_rtree_bits,
+	    test_rtree_random);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/safety_check.c b/memkind-1.10.0/jemalloc/test/unit/safety_check.c
new file mode 100644
index 0000000..bf4bd86
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/safety_check.c
@@ -0,0 +1,156 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/safety_check.h"
+
+/*
+ * Note that we get called through safety_check.sh, which turns on sampling for
+ * everything.
+ */
+
+bool fake_abort_called;
+void fake_abort(const char *message) {
+	(void)message;
+	fake_abort_called = true;
+}
+
+TEST_BEGIN(test_malloc_free_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	free(ptr);
+	safety_check_set_abort(NULL);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_mallocx_dallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = mallocx(128, 0);
+	ptr[128] = 0;
+	dallocx(ptr, 0);
+	safety_check_set_abort(NULL);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_malloc_sdallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	sdallocx(ptr, 128, 0);
+	safety_check_set_abort(NULL);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_realloc_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	ptr = realloc(ptr, 129);
+	safety_check_set_abort(NULL);
+	free(ptr);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_rallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	ptr = rallocx(ptr, 129, 0);
+	safety_check_set_abort(NULL);
+	free(ptr);
+
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+}
+TEST_END
+
+TEST_BEGIN(test_xallocx_overflow) {
+	test_skip_if(!config_prof);
+	test_skip_if(!config_opt_safety_checks);
+
+	safety_check_set_abort(&fake_abort);
+	/* Buffer overflow! */
+	char* ptr = malloc(128);
+	ptr[128] = 0;
+	size_t result = xallocx(ptr, 129, 0, 0);
+	assert_zu_eq(result, 128, "");
+	free(ptr);
+	assert_b_eq(fake_abort_called, true, "Redzone check didn't fire.");
+	fake_abort_called = false;
+	safety_check_set_abort(NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_realloc_no_overflow) {
+	char* ptr = malloc(128);
+	ptr = realloc(ptr, 256);
+	ptr[128] = 0;
+	ptr[255] = 0;
+	free(ptr);
+
+	ptr = malloc(128);
+	ptr = realloc(ptr, 64);
+	ptr[63] = 0;
+	ptr[0] = 0;
+	free(ptr);
+}
+TEST_END
+
+TEST_BEGIN(test_rallocx_no_overflow) {
+	char* ptr = malloc(128);
+	ptr = rallocx(ptr, 256, 0);
+	ptr[128] = 0;
+	ptr[255] = 0;
+	free(ptr);
+
+	ptr = malloc(128);
+	ptr = rallocx(ptr, 64, 0);
+	ptr[63] = 0;
+	ptr[0] = 0;
+	free(ptr);
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_malloc_free_overflow,
+	    test_mallocx_dallocx_overflow,
+	    test_malloc_sdallocx_overflow,
+	    test_realloc_overflow,
+	    test_rallocx_overflow,
+	    test_xallocx_overflow,
+	    test_realloc_no_overflow,
+	    test_rallocx_no_overflow);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/safety_check.sh b/memkind-1.10.0/jemalloc/test/unit/safety_check.sh
new file mode 100644
index 0000000..8fcc7d8
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/safety_check.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_prof}" = "x1" ] ; then
+  export MALLOC_CONF="prof:true,lg_prof_sample:0"
+fi
diff --git a/memkind-1.10.0/jemalloc/test/unit/sc.c b/memkind-1.10.0/jemalloc/test/unit/sc.c
new file mode 100644
index 0000000..bf51d8e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/sc.c
@@ -0,0 +1,33 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_update_slab_size) {
+	sc_data_t data;
+	memset(&data, 0, sizeof(data));
+	sc_data_init(&data);
+	sc_t *tiny = &data.sc[0];
+	size_t tiny_size = (ZU(1) << tiny->lg_base)
+	    + (ZU(tiny->ndelta) << tiny->lg_delta);
+	size_t pgs_too_big = (tiny_size * BITMAP_MAXBITS + PAGE - 1) / PAGE + 1;
+	sc_data_update_slab_size(&data, tiny_size, tiny_size, (int)pgs_too_big);
+	assert_zu_lt((size_t)tiny->pgs, pgs_too_big, "Allowed excessive pages");
+
+	sc_data_update_slab_size(&data, 1, 10 * PAGE, 1);
+	for (int i = 0; i < data.nbins; i++) {
+		sc_t *sc = &data.sc[i];
+		size_t reg_size = (ZU(1) << sc->lg_base)
+		    + (ZU(sc->ndelta) << sc->lg_delta);
+		if (reg_size <= PAGE) {
+			assert_d_eq(sc->pgs, 1, "Ignored valid page size hint");
+		} else {
+			assert_d_gt(sc->pgs, 1,
+			    "Allowed invalid page size hint");
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_update_slab_size);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/seq.c b/memkind-1.10.0/jemalloc/test/unit/seq.c
new file mode 100644
index 0000000..19613b0
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/seq.c
@@ -0,0 +1,95 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/seq.h"
+
+typedef struct data_s data_t;
+struct data_s {
+	int arr[10];
+};
+
+static void
+set_data(data_t *data, int num) {
+	for (int i = 0; i < 10; i++) {
+		data->arr[i] = num;
+	}
+}
+
+static void
+assert_data(data_t *data) {
+	int num = data->arr[0];
+	for (int i = 0; i < 10; i++) {
+		assert_d_eq(num, data->arr[i], "Data consistency error");
+	}
+}
+
+seq_define(data_t, data)
+
+typedef struct thd_data_s thd_data_t;
+struct thd_data_s {
+	seq_data_t data;
+};
+
+static void *
+seq_reader_thd(void *arg) {
+	thd_data_t *thd_data = (thd_data_t *)arg;
+	int iter = 0;
+	data_t local_data;
+	while (iter < 1000 * 1000 - 1) {
+		bool success = seq_try_load_data(&local_data, &thd_data->data);
+		if (success) {
+			assert_data(&local_data);
+			assert_d_le(iter, local_data.arr[0],
+			    "Seq read went back in time.");
+			iter = local_data.arr[0];
+		}
+	}
+	return NULL;
+}
+
+static void *
+seq_writer_thd(void *arg) {
+	thd_data_t *thd_data = (thd_data_t *)arg;
+	data_t local_data;
+	memset(&local_data, 0, sizeof(local_data));
+	for (int i = 0; i < 1000 * 1000; i++) {
+		set_data(&local_data, i);
+		seq_store_data(&thd_data->data, &local_data);
+	}
+	return NULL;
+}
+
+TEST_BEGIN(test_seq_threaded) {
+	thd_data_t thd_data;
+	memset(&thd_data, 0, sizeof(thd_data));
+
+	thd_t reader;
+	thd_t writer;
+
+	thd_create(&reader, seq_reader_thd, &thd_data);
+	thd_create(&writer, seq_writer_thd, &thd_data);
+
+	thd_join(reader, NULL);
+	thd_join(writer, NULL);
+}
+TEST_END
+
+TEST_BEGIN(test_seq_simple) {
+	data_t data;
+	seq_data_t seq;
+	memset(&seq, 0, sizeof(seq));
+	for (int i = 0; i < 1000 * 1000; i++) {
+		set_data(&data, i);
+		seq_store_data(&seq, &data);
+		set_data(&data, 0);
+		bool success = seq_try_load_data(&data, &seq);
+		assert_b_eq(success, true, "Failed non-racing read");
+		assert_data(&data);
+	}
+}
+TEST_END
+
+int main(void) {
+	return test_no_reentrancy(
+	    test_seq_simple,
+	    test_seq_threaded);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/size_classes.c b/memkind-1.10.0/jemalloc/test/unit/size_classes.c
new file mode 100644
index 0000000..6947336
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/size_classes.c
@@ -0,0 +1,188 @@
+#include "test/jemalloc_test.h"
+
+static size_t
+get_max_size_class(void) {
+	unsigned nlextents;
+	size_t mib[4];
+	size_t sz, miblen, max_size_class;
+
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.nlextents", (void *)&nlextents, &sz, NULL,
+	    0), 0, "Unexpected mallctl() error");
+
+	miblen = sizeof(mib) / sizeof(size_t);
+	assert_d_eq(mallctlnametomib("arenas.lextent.0.size", mib, &miblen), 0,
+	    "Unexpected mallctlnametomib() error");
+	mib[2] = nlextents - 1;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctlbymib(mib, miblen, (void *)&max_size_class, &sz,
+	    NULL, 0), 0, "Unexpected mallctlbymib() error");
+
+	return max_size_class;
+}
+
+TEST_BEGIN(test_size_classes) {
+	size_t size_class, max_size_class;
+	szind_t index, max_index;
+
+	max_size_class = get_max_size_class();
+	max_index = sz_size2index(max_size_class);
+
+	for (index = 0, size_class = sz_index2size(index); index < max_index ||
+	    size_class < max_size_class; index++, size_class =
+	    sz_index2size(index)) {
+		assert_true(index < max_index,
+		    "Loop conditionals should be equivalent; index=%u, "
+		    "size_class=%zu (%#zx)", index, size_class, size_class);
+		assert_true(size_class < max_size_class,
+		    "Loop conditionals should be equivalent; index=%u, "
+		    "size_class=%zu (%#zx)", index, size_class, size_class);
+
+		assert_u_eq(index, sz_size2index(size_class),
+		    "sz_size2index() does not reverse sz_index2size(): index=%u"
+		    " --> size_class=%zu --> index=%u --> size_class=%zu",
+		    index, size_class, sz_size2index(size_class),
+		    sz_index2size(sz_size2index(size_class)));
+		assert_zu_eq(size_class,
+		    sz_index2size(sz_size2index(size_class)),
+		    "sz_index2size() does not reverse sz_size2index(): index=%u"
+		    " --> size_class=%zu --> index=%u --> size_class=%zu",
+		    index, size_class, sz_size2index(size_class),
+		    sz_index2size(sz_size2index(size_class)));
+
+		assert_u_eq(index+1, sz_size2index(size_class+1),
+		    "Next size_class does not round up properly");
+
+		assert_zu_eq(size_class, (index > 0) ?
+		    sz_s2u(sz_index2size(index-1)+1) : sz_s2u(1),
+		    "sz_s2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_s2u(size_class-1),
+		    "sz_s2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_s2u(size_class),
+		    "sz_s2u() does not compute same size class");
+		assert_zu_eq(sz_s2u(size_class+1), sz_index2size(index+1),
+		    "sz_s2u() does not round up to next size class");
+	}
+
+	assert_u_eq(index, sz_size2index(sz_index2size(index)),
+	    "sz_size2index() does not reverse sz_index2size()");
+	assert_zu_eq(max_size_class, sz_index2size(
+	    sz_size2index(max_size_class)),
+	    "sz_index2size() does not reverse sz_size2index()");
+
+	assert_zu_eq(size_class, sz_s2u(sz_index2size(index-1)+1),
+	    "sz_s2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_s2u(size_class-1),
+	    "sz_s2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_s2u(size_class),
+	    "sz_s2u() does not compute same size class");
+}
+TEST_END
+
+TEST_BEGIN(test_psize_classes) {
+	size_t size_class, max_psz;
+	pszind_t pind, max_pind;
+
+	max_psz = get_max_size_class() + PAGE;
+	max_pind = sz_psz2ind(max_psz);
+
+	for (pind = 0, size_class = sz_pind2sz(pind);
+	    pind < max_pind || size_class < max_psz;
+	    pind++, size_class = sz_pind2sz(pind)) {
+		assert_true(pind < max_pind,
+		    "Loop conditionals should be equivalent; pind=%u, "
+		    "size_class=%zu (%#zx)", pind, size_class, size_class);
+		assert_true(size_class < max_psz,
+		    "Loop conditionals should be equivalent; pind=%u, "
+		    "size_class=%zu (%#zx)", pind, size_class, size_class);
+
+		assert_u_eq(pind, sz_psz2ind(size_class),
+		    "sz_psz2ind() does not reverse sz_pind2sz(): pind=%u -->"
+		    " size_class=%zu --> pind=%u --> size_class=%zu", pind,
+		    size_class, sz_psz2ind(size_class),
+		    sz_pind2sz(sz_psz2ind(size_class)));
+		assert_zu_eq(size_class, sz_pind2sz(sz_psz2ind(size_class)),
+		    "sz_pind2sz() does not reverse sz_psz2ind(): pind=%u -->"
+		    " size_class=%zu --> pind=%u --> size_class=%zu", pind,
+		    size_class, sz_psz2ind(size_class),
+		    sz_pind2sz(sz_psz2ind(size_class)));
+
+		if (size_class == SC_LARGE_MAXCLASS) {
+			assert_u_eq(SC_NPSIZES, sz_psz2ind(size_class + 1),
+			    "Next size_class does not round up properly");
+		} else {
+			assert_u_eq(pind + 1, sz_psz2ind(size_class + 1),
+			    "Next size_class does not round up properly");
+		}
+
+		assert_zu_eq(size_class, (pind > 0) ?
+		    sz_psz2u(sz_pind2sz(pind-1)+1) : sz_psz2u(1),
+		    "sz_psz2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_psz2u(size_class-1),
+		    "sz_psz2u() does not round up to size class");
+		assert_zu_eq(size_class, sz_psz2u(size_class),
+		    "sz_psz2u() does not compute same size class");
+		assert_zu_eq(sz_psz2u(size_class+1), sz_pind2sz(pind+1),
+		    "sz_psz2u() does not round up to next size class");
+	}
+
+	assert_u_eq(pind, sz_psz2ind(sz_pind2sz(pind)),
+	    "sz_psz2ind() does not reverse sz_pind2sz()");
+	assert_zu_eq(max_psz, sz_pind2sz(sz_psz2ind(max_psz)),
+	    "sz_pind2sz() does not reverse sz_psz2ind()");
+
+	assert_zu_eq(size_class, sz_psz2u(sz_pind2sz(pind-1)+1),
+	    "sz_psz2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_psz2u(size_class-1),
+	    "sz_psz2u() does not round up to size class");
+	assert_zu_eq(size_class, sz_psz2u(size_class),
+	    "sz_psz2u() does not compute same size class");
+}
+TEST_END
+
+TEST_BEGIN(test_overflow) {
+	size_t max_size_class, max_psz;
+
+	max_size_class = get_max_size_class();
+	max_psz = max_size_class + PAGE;
+
+	assert_u_eq(sz_size2index(max_size_class+1), SC_NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+	assert_u_eq(sz_size2index(ZU(PTRDIFF_MAX)+1), SC_NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+	assert_u_eq(sz_size2index(SIZE_T_MAX), SC_NSIZES,
+	    "sz_size2index() should return NSIZES on overflow");
+
+	assert_zu_eq(sz_s2u(max_size_class+1), 0,
+	    "sz_s2u() should return 0 for unsupported size");
+	assert_zu_eq(sz_s2u(ZU(PTRDIFF_MAX)+1), 0,
+	    "sz_s2u() should return 0 for unsupported size");
+	assert_zu_eq(sz_s2u(SIZE_T_MAX), 0,
+	    "sz_s2u() should return 0 on overflow");
+
+	assert_u_eq(sz_psz2ind(max_size_class+1), SC_NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+	assert_u_eq(sz_psz2ind(ZU(PTRDIFF_MAX)+1), SC_NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+	assert_u_eq(sz_psz2ind(SIZE_T_MAX), SC_NPSIZES,
+	    "sz_psz2ind() should return NPSIZES on overflow");
+
+	assert_zu_eq(sz_psz2u(max_size_class+1), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported"
+	    " size");
+	assert_zu_eq(sz_psz2u(ZU(PTRDIFF_MAX)+1), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) for unsupported "
+	    "size");
+	assert_zu_eq(sz_psz2u(SIZE_T_MAX), max_psz,
+	    "sz_psz2u() should return (LARGE_MAXCLASS + PAGE) on overflow");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_size_classes,
+	    test_psize_classes,
+	    test_overflow);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/slab.c b/memkind-1.10.0/jemalloc/test/unit/slab.c
new file mode 100644
index 0000000..c56af25
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/slab.c
@@ -0,0 +1,33 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_arena_slab_regind) {
+	szind_t binind;
+
+	for (binind = 0; binind < SC_NBINS; binind++) {
+		size_t regind;
+		extent_t slab;
+		const bin_info_t *bin_info = &bin_infos[binind];
+		extent_init(&slab, NULL, mallocx(bin_info->slab_size,
+		    MALLOCX_LG_ALIGN(LG_PAGE)), bin_info->slab_size, true,
+		    binind, 0, extent_state_active, false, true, true,
+		    EXTENT_NOT_HEAD);
+		assert_ptr_not_null(extent_addr_get(&slab),
+		    "Unexpected malloc() failure");
+		for (regind = 0; regind < bin_info->nregs; regind++) {
+			void *reg = (void *)((uintptr_t)extent_addr_get(&slab) +
+			    (bin_info->reg_size * regind));
+			assert_zu_eq(arena_slab_regind(&slab, binind, reg),
+			    regind,
+			    "Incorrect region index computed for size %zu",
+			    bin_info->reg_size);
+		}
+		free(extent_addr_get(&slab));
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_arena_slab_regind);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/smoothstep.c b/memkind-1.10.0/jemalloc/test/unit/smoothstep.c
new file mode 100644
index 0000000..7c5dbb7
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/smoothstep.c
@@ -0,0 +1,102 @@
+#include "test/jemalloc_test.h"
+
+static const uint64_t smoothstep_tab[] = {
+#define STEP(step, h, x, y)			\
+	h,
+	SMOOTHSTEP
+#undef STEP
+};
+
+TEST_BEGIN(test_smoothstep_integral) {
+	uint64_t sum, min, max;
+	unsigned i;
+
+	/*
+	 * The integral of smoothstep in the [0..1] range equals 1/2.  Verify
+	 * that the fixed point representation's integral is no more than
+	 * rounding error distant from 1/2.  Regarding rounding, each table
+	 * element is rounded down to the nearest fixed point value, so the
+	 * integral may be off by as much as SMOOTHSTEP_NSTEPS ulps.
+	 */
+	sum = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+		sum += smoothstep_tab[i];
+	}
+
+	max = (KQU(1) << (SMOOTHSTEP_BFP-1)) * (SMOOTHSTEP_NSTEPS+1);
+	min = max - SMOOTHSTEP_NSTEPS;
+
+	assert_u64_ge(sum, min,
+	    "Integral too small, even accounting for truncation");
+	assert_u64_le(sum, max, "Integral exceeds 1/2");
+	if (false) {
+		malloc_printf("%"FMTu64" ulps under 1/2 (limit %d)\n",
+		    max - sum, SMOOTHSTEP_NSTEPS);
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_smoothstep_monotonic) {
+	uint64_t prev_h;
+	unsigned i;
+
+	/*
+	 * The smoothstep function is monotonic in [0..1], i.e. its slope is
+	 * non-negative.  In practice we want to parametrize table generation
+	 * such that piecewise slope is greater than zero, but do not require
+	 * that here.
+	 */
+	prev_h = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS; i++) {
+		uint64_t h = smoothstep_tab[i];
+		assert_u64_ge(h, prev_h, "Piecewise non-monotonic, i=%u", i);
+		prev_h = h;
+	}
+	assert_u64_eq(smoothstep_tab[SMOOTHSTEP_NSTEPS-1],
+	    (KQU(1) << SMOOTHSTEP_BFP), "Last step must equal 1");
+}
+TEST_END
+
+TEST_BEGIN(test_smoothstep_slope) {
+	uint64_t prev_h, prev_delta;
+	unsigned i;
+
+	/*
+	 * The smoothstep slope strictly increases until x=0.5, and then
+	 * strictly decreases until x=1.0.  Verify the slightly weaker
+	 * requirement of monotonicity, so that inadequate table precision does
+	 * not cause false test failures.
+	 */
+	prev_h = 0;
+	prev_delta = 0;
+	for (i = 0; i < SMOOTHSTEP_NSTEPS / 2 + SMOOTHSTEP_NSTEPS % 2; i++) {
+		uint64_t h = smoothstep_tab[i];
+		uint64_t delta = h - prev_h;
+		assert_u64_ge(delta, prev_delta,
+		    "Slope must monotonically increase in 0.0 <= x <= 0.5, "
+		    "i=%u", i);
+		prev_h = h;
+		prev_delta = delta;
+	}
+
+	prev_h = KQU(1) << SMOOTHSTEP_BFP;
+	prev_delta = 0;
+	for (i = SMOOTHSTEP_NSTEPS-1; i >= SMOOTHSTEP_NSTEPS / 2; i--) {
+		uint64_t h = smoothstep_tab[i];
+		uint64_t delta = prev_h - h;
+		assert_u64_ge(delta, prev_delta,
+		    "Slope must monotonically decrease in 0.5 <= x <= 1.0, "
+		    "i=%u", i);
+		prev_h = h;
+		prev_delta = delta;
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_smoothstep_integral,
+	    test_smoothstep_monotonic,
+	    test_smoothstep_slope);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/spin.c b/memkind-1.10.0/jemalloc/test/unit/spin.c
new file mode 100644
index 0000000..b965f74
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/spin.c
@@ -0,0 +1,18 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/spin.h"
+
+TEST_BEGIN(test_spin) {
+	spin_t spinner = SPIN_INITIALIZER;
+
+	for (unsigned i = 0; i < 100; i++) {
+		spin_adaptive(&spinner);
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_spin);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/stats.c b/memkind-1.10.0/jemalloc/test/unit/stats.c
new file mode 100644
index 0000000..646768e
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/stats.c
@@ -0,0 +1,374 @@
+#include "test/jemalloc_test.h"
+
+TEST_BEGIN(test_stats_summary) {
+	size_t sz, allocated, active, resident, mapped;
+	int expected = config_stats ? 0 : ENOENT;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.allocated", (void *)&allocated, &sz, NULL,
+	    0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.active", (void *)&active, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.resident", (void *)&resident, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.mapped", (void *)&mapped, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_le(allocated, active,
+		    "allocated should be no larger than active");
+		assert_zu_lt(active, resident,
+		    "active should be less than resident");
+		assert_zu_lt(active, mapped,
+		    "active should be less than mapped");
+	}
+}
+TEST_END
+
+TEST_BEGIN(test_stats_large) {
+	void *p;
+	uint64_t epoch;
+	size_t allocated;
+	uint64_t nmalloc, ndalloc, nrequests;
+	size_t sz;
+	int expected = config_stats ? 0 : ENOENT;
+
+	p = mallocx(SC_SMALL_MAXCLASS + 1, MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.allocated",
+	    (void *)&allocated, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.nmalloc", (void *)&nmalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.large.ndalloc", (void *)&ndalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.large.nrequests",
+	    (void *)&nrequests, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_gt(allocated, 0,
+		    "allocated should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_le(nmalloc, nrequests,
+		    "nmalloc should no larger than nrequests");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas_summary) {
+	void *little, *large;
+	uint64_t epoch;
+	size_t sz;
+	int expected = config_stats ? 0 : ENOENT;
+	size_t mapped;
+	uint64_t dirty_npurge, dirty_nmadvise, dirty_purged;
+	uint64_t muzzy_npurge, muzzy_nmadvise, muzzy_purged;
+
+	little = mallocx(SC_SMALL_MAXCLASS, MALLOCX_ARENA(0));
+	assert_ptr_not_null(little, "Unexpected mallocx() failure");
+	large = mallocx((1U << SC_LG_LARGE_MINCLASS),
+	    MALLOCX_ARENA(0));
+	assert_ptr_not_null(large, "Unexpected mallocx() failure");
+
+	dallocx(little, 0);
+	dallocx(large, 0);
+
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("arena.0.purge", NULL, NULL, NULL, 0), 0,
+	    "Unexpected mallctl() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.mapped", (void *)&mapped, &sz, NULL,
+	    0), expected, "Unexepected mallctl() result");
+
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.dirty_npurge",
+	    (void *)&dirty_npurge, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.dirty_nmadvise",
+	    (void *)&dirty_nmadvise, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.dirty_purged",
+	    (void *)&dirty_purged, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.muzzy_npurge",
+	    (void *)&muzzy_npurge, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.muzzy_nmadvise",
+	    (void *)&muzzy_nmadvise, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.muzzy_purged",
+	    (void *)&muzzy_purged, &sz, NULL, 0), expected,
+	    "Unexepected mallctl() result");
+
+	if (config_stats) {
+		if (!background_thread_enabled()) {
+			assert_u64_gt(dirty_npurge + muzzy_npurge, 0,
+			    "At least one purge should have occurred");
+		}
+		assert_u64_le(dirty_nmadvise, dirty_purged,
+		    "dirty_nmadvise should be no greater than dirty_purged");
+		assert_u64_le(muzzy_nmadvise, muzzy_purged,
+		    "muzzy_nmadvise should be no greater than muzzy_purged");
+	}
+}
+TEST_END
+
+void *
+thd_start(void *arg) {
+	return NULL;
+}
+
+static void
+no_lazy_lock(void) {
+	thd_t thd;
+
+	thd_create(&thd, thd_start, NULL);
+	thd_join(thd, NULL);
+}
+
+TEST_BEGIN(test_stats_arenas_small) {
+	void *p;
+	size_t sz, allocated;
+	uint64_t epoch, nmalloc, ndalloc, nrequests;
+	int expected = config_stats ? 0 : ENOENT;
+
+	no_lazy_lock(); /* Lazy locking would dodge tcache testing. */
+
+	p = mallocx(SC_SMALL_MAXCLASS, MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.small.allocated",
+	    (void *)&allocated, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.small.nmalloc", (void *)&nmalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.small.ndalloc", (void *)&ndalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.small.nrequests",
+	    (void *)&nrequests, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_gt(allocated, 0,
+		    "allocated should be greater than zero");
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be no greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_gt(nrequests, 0,
+		    "nrequests should be greater than zero");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas_large) {
+	void *p;
+	size_t sz, allocated;
+	uint64_t epoch, nmalloc, ndalloc;
+	int expected = config_stats ? 0 : ENOENT;
+
+	p = mallocx((1U << SC_LG_LARGE_MINCLASS), MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.allocated",
+	    (void *)&allocated, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.large.nmalloc", (void *)&nmalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.large.ndalloc", (void *)&ndalloc,
+	    &sz, NULL, 0), expected, "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_zu_gt(allocated, 0,
+		    "allocated should be greater than zero");
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+static void
+gen_mallctl_str(char *cmd, char *name, unsigned arena_ind) {
+	sprintf(cmd, "stats.arenas.%u.bins.0.%s", arena_ind, name);
+}
+
+TEST_BEGIN(test_stats_arenas_bins) {
+	void *p;
+	size_t sz, curslabs, curregs, nonfull_slabs;
+	uint64_t epoch, nmalloc, ndalloc, nrequests, nfills, nflushes;
+	uint64_t nslabs, nreslabs;
+	int expected = config_stats ? 0 : ENOENT;
+
+	/* Make sure allocation below isn't satisfied by tcache. */
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+
+	unsigned arena_ind, old_arena_ind;
+	sz = sizeof(unsigned);
+	assert_d_eq(mallctl("arenas.create", (void *)&arena_ind, &sz, NULL, 0),
+	    0, "Arena creation failure");
+	sz = sizeof(arena_ind);
+	assert_d_eq(mallctl("thread.arena", (void *)&old_arena_ind, &sz,
+	    (void *)&arena_ind, sizeof(arena_ind)), 0,
+	    "Unexpected mallctl() failure");
+
+	p = malloc(bin_infos[0].reg_size);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	assert_d_eq(mallctl("thread.tcache.flush", NULL, NULL, NULL, 0),
+	    opt_tcache ? 0 : EFAULT, "Unexpected mallctl() result");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	char cmd[128];
+	sz = sizeof(uint64_t);
+	gen_mallctl_str(cmd, "nmalloc", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nmalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "ndalloc", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&ndalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nrequests", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nrequests, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(size_t);
+	gen_mallctl_str(cmd, "curregs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&curregs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	sz = sizeof(uint64_t);
+	gen_mallctl_str(cmd, "nfills", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nfills, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nflushes", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nflushes, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	gen_mallctl_str(cmd, "nslabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nslabs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nreslabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nreslabs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(size_t);
+	gen_mallctl_str(cmd, "curslabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&curslabs, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	gen_mallctl_str(cmd, "nonfull_slabs", arena_ind);
+	assert_d_eq(mallctl(cmd, (void *)&nonfull_slabs, &sz, NULL, 0),
+	    expected, "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_gt(nrequests, 0,
+		    "nrequests should be greater than zero");
+		assert_zu_gt(curregs, 0,
+		    "allocated should be greater than zero");
+		if (opt_tcache) {
+			assert_u64_gt(nfills, 0,
+			    "At least one fill should have occurred");
+			assert_u64_gt(nflushes, 0,
+			    "At least one flush should have occurred");
+		}
+		assert_u64_gt(nslabs, 0,
+		    "At least one slab should have been allocated");
+		assert_zu_gt(curslabs, 0,
+		    "At least one slab should be currently allocated");
+		assert_zu_eq(nonfull_slabs, 0,
+		    "slabs_nonfull should be empty");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_stats_arenas_lextents) {
+	void *p;
+	uint64_t epoch, nmalloc, ndalloc;
+	size_t curlextents, sz, hsize;
+	int expected = config_stats ? 0 : ENOENT;
+
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("arenas.lextent.0.size", (void *)&hsize, &sz, NULL,
+	    0), 0, "Unexpected mallctl() failure");
+
+	p = mallocx(hsize, MALLOCX_ARENA(0));
+	assert_ptr_not_null(p, "Unexpected mallocx() failure");
+
+	assert_d_eq(mallctl("epoch", NULL, NULL, (void *)&epoch, sizeof(epoch)),
+	    0, "Unexpected mallctl() failure");
+
+	sz = sizeof(uint64_t);
+	assert_d_eq(mallctl("stats.arenas.0.lextents.0.nmalloc",
+	    (void *)&nmalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	assert_d_eq(mallctl("stats.arenas.0.lextents.0.ndalloc",
+	    (void *)&ndalloc, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+	sz = sizeof(size_t);
+	assert_d_eq(mallctl("stats.arenas.0.lextents.0.curlextents",
+	    (void *)&curlextents, &sz, NULL, 0), expected,
+	    "Unexpected mallctl() result");
+
+	if (config_stats) {
+		assert_u64_gt(nmalloc, 0,
+		    "nmalloc should be greater than zero");
+		assert_u64_ge(nmalloc, ndalloc,
+		    "nmalloc should be at least as large as ndalloc");
+		assert_u64_gt(curlextents, 0,
+		    "At least one extent should be currently allocated");
+	}
+
+	dallocx(p, 0);
+}
+TEST_END
+
+int
+main(void) {
+	return test_no_reentrancy(
+	    test_stats_summary,
+	    test_stats_large,
+	    test_stats_arenas_summary,
+	    test_stats_arenas_small,
+	    test_stats_arenas_large,
+	    test_stats_arenas_bins,
+	    test_stats_arenas_lextents);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/stats_print.c b/memkind-1.10.0/jemalloc/test/unit/stats_print.c
new file mode 100644
index 0000000..014d002
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/stats_print.c
@@ -0,0 +1,999 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/util.h"
+
+typedef enum {
+	TOKEN_TYPE_NONE,
+	TOKEN_TYPE_ERROR,
+	TOKEN_TYPE_EOI,
+	TOKEN_TYPE_NULL,
+	TOKEN_TYPE_FALSE,
+	TOKEN_TYPE_TRUE,
+	TOKEN_TYPE_LBRACKET,
+	TOKEN_TYPE_RBRACKET,
+	TOKEN_TYPE_LBRACE,
+	TOKEN_TYPE_RBRACE,
+	TOKEN_TYPE_COLON,
+	TOKEN_TYPE_COMMA,
+	TOKEN_TYPE_STRING,
+	TOKEN_TYPE_NUMBER
+} token_type_t;
+
+typedef struct parser_s parser_t;
+typedef struct {
+	parser_t	*parser;
+	token_type_t	token_type;
+	size_t		pos;
+	size_t		len;
+	size_t		line;
+	size_t		col;
+} token_t;
+
+struct parser_s {
+	bool verbose;
+	char	*buf; /* '\0'-terminated. */
+	size_t	len; /* Number of characters preceding '\0' in buf. */
+	size_t	pos;
+	size_t	line;
+	size_t	col;
+	token_t	token;
+};
+
+static void
+token_init(token_t *token, parser_t *parser, token_type_t token_type,
+    size_t pos, size_t len, size_t line, size_t col) {
+	token->parser = parser;
+	token->token_type = token_type;
+	token->pos = pos;
+	token->len = len;
+	token->line = line;
+	token->col = col;
+}
+
+static void
+token_error(token_t *token) {
+	if (!token->parser->verbose) {
+		return;
+	}
+	switch (token->token_type) {
+	case TOKEN_TYPE_NONE:
+		not_reached();
+	case TOKEN_TYPE_ERROR:
+		malloc_printf("%zu:%zu: Unexpected character in token: ",
+		    token->line, token->col);
+		break;
+	default:
+		malloc_printf("%zu:%zu: Unexpected token: ", token->line,
+		    token->col);
+		break;
+	}
+	UNUSED ssize_t err = malloc_write_fd(STDERR_FILENO,
+	    &token->parser->buf[token->pos], token->len);
+	malloc_printf("\n");
+}
+
+static void
+parser_init(parser_t *parser, bool verbose) {
+	parser->verbose = verbose;
+	parser->buf = NULL;
+	parser->len = 0;
+	parser->pos = 0;
+	parser->line = 1;
+	parser->col = 0;
+}
+
+static void
+parser_fini(parser_t *parser) {
+	if (parser->buf != NULL) {
+		dallocx(parser->buf, MALLOCX_TCACHE_NONE);
+	}
+}
+
+static bool
+parser_append(parser_t *parser, const char *str) {
+	size_t len = strlen(str);
+	char *buf = (parser->buf == NULL) ? mallocx(len + 1,
+	    MALLOCX_TCACHE_NONE) : rallocx(parser->buf, parser->len + len + 1,
+	    MALLOCX_TCACHE_NONE);
+	if (buf == NULL) {
+		return true;
+	}
+	memcpy(&buf[parser->len], str, len + 1);
+	parser->buf = buf;
+	parser->len += len;
+	return false;
+}
+
+static bool
+parser_tokenize(parser_t *parser) {
+	enum {
+		STATE_START,
+		STATE_EOI,
+		STATE_N, STATE_NU, STATE_NUL, STATE_NULL,
+		STATE_F, STATE_FA, STATE_FAL, STATE_FALS, STATE_FALSE,
+		STATE_T, STATE_TR, STATE_TRU, STATE_TRUE,
+		STATE_LBRACKET,
+		STATE_RBRACKET,
+		STATE_LBRACE,
+		STATE_RBRACE,
+		STATE_COLON,
+		STATE_COMMA,
+		STATE_CHARS,
+		STATE_CHAR_ESCAPE,
+		STATE_CHAR_U, STATE_CHAR_UD, STATE_CHAR_UDD, STATE_CHAR_UDDD,
+		STATE_STRING,
+		STATE_MINUS,
+		STATE_LEADING_ZERO,
+		STATE_DIGITS,
+		STATE_DECIMAL,
+		STATE_FRAC_DIGITS,
+		STATE_EXP,
+		STATE_EXP_SIGN,
+		STATE_EXP_DIGITS,
+		STATE_ACCEPT
+	} state = STATE_START;
+	size_t token_pos JEMALLOC_CC_SILENCE_INIT(0);
+	size_t token_line JEMALLOC_CC_SILENCE_INIT(1);
+	size_t token_col JEMALLOC_CC_SILENCE_INIT(0);
+
+	assert_zu_le(parser->pos, parser->len,
+	    "Position is past end of buffer");
+
+	while (state != STATE_ACCEPT) {
+		char c = parser->buf[parser->pos];
+
+		switch (state) {
+		case STATE_START:
+			token_pos = parser->pos;
+			token_line = parser->line;
+			token_col = parser->col;
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+				break;
+			case '\0':
+				state = STATE_EOI;
+				break;
+			case 'n':
+				state = STATE_N;
+				break;
+			case 'f':
+				state = STATE_F;
+				break;
+			case 't':
+				state = STATE_T;
+				break;
+			case '[':
+				state = STATE_LBRACKET;
+				break;
+			case ']':
+				state = STATE_RBRACKET;
+				break;
+			case '{':
+				state = STATE_LBRACE;
+				break;
+			case '}':
+				state = STATE_RBRACE;
+				break;
+			case ':':
+				state = STATE_COLON;
+				break;
+			case ',':
+				state = STATE_COMMA;
+				break;
+			case '"':
+				state = STATE_CHARS;
+				break;
+			case '-':
+				state = STATE_MINUS;
+				break;
+			case '0':
+				state = STATE_LEADING_ZERO;
+				break;
+			case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_EOI:
+			token_init(&parser->token, parser,
+			    TOKEN_TYPE_EOI, token_pos, parser->pos -
+			    token_pos, token_line, token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_N:
+			switch (c) {
+			case 'u':
+				state = STATE_NU;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_NU:
+			switch (c) {
+			case 'l':
+				state = STATE_NUL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_NUL:
+			switch (c) {
+			case 'l':
+				state = STATE_NULL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_NULL:
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+			case '\0':
+			case '[': case ']': case '{': case '}': case ':':
+			case ',':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			token_init(&parser->token, parser, TOKEN_TYPE_NULL,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_F:
+			switch (c) {
+			case 'a':
+				state = STATE_FA;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FA:
+			switch (c) {
+			case 'l':
+				state = STATE_FAL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FAL:
+			switch (c) {
+			case 's':
+				state = STATE_FALS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FALS:
+			switch (c) {
+			case 'e':
+				state = STATE_FALSE;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FALSE:
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+			case '\0':
+			case '[': case ']': case '{': case '}': case ':':
+			case ',':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			token_init(&parser->token, parser,
+			    TOKEN_TYPE_FALSE, token_pos, parser->pos -
+			    token_pos, token_line, token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_T:
+			switch (c) {
+			case 'r':
+				state = STATE_TR;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_TR:
+			switch (c) {
+			case 'u':
+				state = STATE_TRU;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_TRU:
+			switch (c) {
+			case 'e':
+				state = STATE_TRUE;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_TRUE:
+			switch (c) {
+			case ' ': case '\b': case '\n': case '\r': case '\t':
+			case '\0':
+			case '[': case ']': case '{': case '}': case ':':
+			case ',':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			token_init(&parser->token, parser, TOKEN_TYPE_TRUE,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_LBRACKET:
+			token_init(&parser->token, parser, TOKEN_TYPE_LBRACKET,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_RBRACKET:
+			token_init(&parser->token, parser, TOKEN_TYPE_RBRACKET,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_LBRACE:
+			token_init(&parser->token, parser, TOKEN_TYPE_LBRACE,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_RBRACE:
+			token_init(&parser->token, parser, TOKEN_TYPE_RBRACE,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_COLON:
+			token_init(&parser->token, parser, TOKEN_TYPE_COLON,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_COMMA:
+			token_init(&parser->token, parser, TOKEN_TYPE_COMMA,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_CHARS:
+			switch (c) {
+			case '\\':
+				state = STATE_CHAR_ESCAPE;
+				break;
+			case '"':
+				state = STATE_STRING;
+				break;
+			case 0x00: case 0x01: case 0x02: case 0x03: case 0x04:
+			case 0x05: case 0x06: case 0x07: case 0x08: case 0x09:
+			case 0x0a: case 0x0b: case 0x0c: case 0x0d: case 0x0e:
+			case 0x0f: case 0x10: case 0x11: case 0x12: case 0x13:
+			case 0x14: case 0x15: case 0x16: case 0x17: case 0x18:
+			case 0x19: case 0x1a: case 0x1b: case 0x1c: case 0x1d:
+			case 0x1e: case 0x1f:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			default:
+				break;
+			}
+			break;
+		case STATE_CHAR_ESCAPE:
+			switch (c) {
+			case '"': case '\\': case '/': case 'b': case 'n':
+			case 'r': case 't':
+				state = STATE_CHARS;
+				break;
+			case 'u':
+				state = STATE_CHAR_U;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_U:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHAR_UD;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_UD:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHAR_UDD;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_UDD:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHAR_UDDD;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_CHAR_UDDD:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+				state = STATE_CHARS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_STRING:
+			token_init(&parser->token, parser, TOKEN_TYPE_STRING,
+			    token_pos, parser->pos - token_pos, token_line,
+			    token_col);
+			state = STATE_ACCEPT;
+			break;
+		case STATE_MINUS:
+			switch (c) {
+			case '0':
+				state = STATE_LEADING_ZERO;
+				break;
+			case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_LEADING_ZERO:
+			switch (c) {
+			case '.':
+				state = STATE_DECIMAL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		case STATE_DIGITS:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				break;
+			case '.':
+				state = STATE_DECIMAL;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		case STATE_DECIMAL:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_FRAC_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_FRAC_DIGITS:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				break;
+			case 'e': case 'E':
+				state = STATE_EXP;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		case STATE_EXP:
+			switch (c) {
+			case '-': case '+':
+				state = STATE_EXP_SIGN;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_EXP_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_EXP_SIGN:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				state = STATE_EXP_DIGITS;
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_ERROR, token_pos, parser->pos + 1
+				    - token_pos, token_line, token_col);
+				return true;
+			}
+			break;
+		case STATE_EXP_DIGITS:
+			switch (c) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+				break;
+			default:
+				token_init(&parser->token, parser,
+				    TOKEN_TYPE_NUMBER, token_pos, parser->pos -
+				    token_pos, token_line, token_col);
+				state = STATE_ACCEPT;
+				break;
+			}
+			break;
+		default:
+			not_reached();
+		}
+
+		if (state != STATE_ACCEPT) {
+			if (c == '\n') {
+				parser->line++;
+				parser->col = 0;
+			} else {
+				parser->col++;
+			}
+			parser->pos++;
+		}
+	}
+	return false;
+}
+
+static bool	parser_parse_array(parser_t *parser);
+static bool	parser_parse_object(parser_t *parser);
+
+static bool
+parser_parse_value(parser_t *parser) {
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_NULL:
+	case TOKEN_TYPE_FALSE:
+	case TOKEN_TYPE_TRUE:
+	case TOKEN_TYPE_STRING:
+	case TOKEN_TYPE_NUMBER:
+		return false;
+	case TOKEN_TYPE_LBRACE:
+		return parser_parse_object(parser);
+	case TOKEN_TYPE_LBRACKET:
+		return parser_parse_array(parser);
+	default:
+		return true;
+	}
+	not_reached();
+}
+
+static bool
+parser_parse_pair(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_STRING,
+	    "Pair should start with string");
+	if (parser_tokenize(parser)) {
+		return true;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_COLON:
+		if (parser_tokenize(parser)) {
+			return true;
+		}
+		return parser_parse_value(parser);
+	default:
+		return true;
+	}
+}
+
+static bool
+parser_parse_values(parser_t *parser) {
+	if (parser_parse_value(parser)) {
+		return true;
+	}
+
+	while (true) {
+		if (parser_tokenize(parser)) {
+			return true;
+		}
+		switch (parser->token.token_type) {
+		case TOKEN_TYPE_COMMA:
+			if (parser_tokenize(parser)) {
+				return true;
+			}
+			if (parser_parse_value(parser)) {
+				return true;
+			}
+			break;
+		case TOKEN_TYPE_RBRACKET:
+			return false;
+		default:
+			return true;
+		}
+	}
+}
+
+static bool
+parser_parse_array(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_LBRACKET,
+	    "Array should start with [");
+	if (parser_tokenize(parser)) {
+		return true;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_RBRACKET:
+		return false;
+	default:
+		return parser_parse_values(parser);
+	}
+	not_reached();
+}
+
+static bool
+parser_parse_pairs(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_STRING,
+	    "Object should start with string");
+	if (parser_parse_pair(parser)) {
+		return true;
+	}
+
+	while (true) {
+		if (parser_tokenize(parser)) {
+			return true;
+		}
+		switch (parser->token.token_type) {
+		case TOKEN_TYPE_COMMA:
+			if (parser_tokenize(parser)) {
+				return true;
+			}
+			switch (parser->token.token_type) {
+			case TOKEN_TYPE_STRING:
+				if (parser_parse_pair(parser)) {
+					return true;
+				}
+				break;
+			default:
+				return true;
+			}
+			break;
+		case TOKEN_TYPE_RBRACE:
+			return false;
+		default:
+			return true;
+		}
+	}
+}
+
+static bool
+parser_parse_object(parser_t *parser) {
+	assert_d_eq(parser->token.token_type, TOKEN_TYPE_LBRACE,
+	    "Object should start with {");
+	if (parser_tokenize(parser)) {
+		return true;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_STRING:
+		return parser_parse_pairs(parser);
+	case TOKEN_TYPE_RBRACE:
+		return false;
+	default:
+		return true;
+	}
+	not_reached();
+}
+
+static bool
+parser_parse(parser_t *parser) {
+	if (parser_tokenize(parser)) {
+		goto label_error;
+	}
+	if (parser_parse_value(parser)) {
+		goto label_error;
+	}
+
+	if (parser_tokenize(parser)) {
+		goto label_error;
+	}
+	switch (parser->token.token_type) {
+	case TOKEN_TYPE_EOI:
+		return false;
+	default:
+		goto label_error;
+	}
+	not_reached();
+
+label_error:
+	token_error(&parser->token);
+	return true;
+}
+
+TEST_BEGIN(test_json_parser) {
+	size_t i;
+	const char *invalid_inputs[] = {
+		/* Tokenizer error case tests. */
+		"{ \"string\": X }",
+		"{ \"string\": nXll }",
+		"{ \"string\": nuXl }",
+		"{ \"string\": nulX }",
+		"{ \"string\": nullX }",
+		"{ \"string\": fXlse }",
+		"{ \"string\": faXse }",
+		"{ \"string\": falXe }",
+		"{ \"string\": falsX }",
+		"{ \"string\": falseX }",
+		"{ \"string\": tXue }",
+		"{ \"string\": trXe }",
+		"{ \"string\": truX }",
+		"{ \"string\": trueX }",
+		"{ \"string\": \"\n\" }",
+		"{ \"string\": \"\\z\" }",
+		"{ \"string\": \"\\uX000\" }",
+		"{ \"string\": \"\\u0X00\" }",
+		"{ \"string\": \"\\u00X0\" }",
+		"{ \"string\": \"\\u000X\" }",
+		"{ \"string\": -X }",
+		"{ \"string\": 0.X }",
+		"{ \"string\": 0.0eX }",
+		"{ \"string\": 0.0e+X }",
+
+		/* Parser error test cases. */
+		"{\"string\": }",
+		"{\"string\" }",
+		"{\"string\": [ 0 }",
+		"{\"string\": {\"a\":0, 1 } }",
+		"{\"string\": {\"a\":0: } }",
+		"{",
+		"{}{",
+	};
+	const char *valid_inputs[] = {
+		/* Token tests. */
+		"null",
+		"false",
+		"true",
+		"{}",
+		"{\"a\": 0}",
+		"[]",
+		"[0, 1]",
+		"0",
+		"1",
+		"10",
+		"-10",
+		"10.23",
+		"10.23e4",
+		"10.23e-4",
+		"10.23e+4",
+		"10.23E4",
+		"10.23E-4",
+		"10.23E+4",
+		"-10.23",
+		"-10.23e4",
+		"-10.23e-4",
+		"-10.23e+4",
+		"-10.23E4",
+		"-10.23E-4",
+		"-10.23E+4",
+		"\"value\"",
+		"\" \\\" \\/ \\b \\n \\r \\t \\u0abc \\u1DEF \"",
+
+		/* Parser test with various nesting. */
+		"{\"a\":null, \"b\":[1,[{\"c\":2},3]], \"d\":{\"e\":true}}",
+	};
+
+	for (i = 0; i < sizeof(invalid_inputs)/sizeof(const char *); i++) {
+		const char *input = invalid_inputs[i];
+		parser_t parser;
+		parser_init(&parser, false);
+		assert_false(parser_append(&parser, input),
+		    "Unexpected input appending failure");
+		assert_true(parser_parse(&parser),
+		    "Unexpected parse success for input: %s", input);
+		parser_fini(&parser);
+	}
+
+	for (i = 0; i < sizeof(valid_inputs)/sizeof(const char *); i++) {
+		const char *input = valid_inputs[i];
+		parser_t parser;
+		parser_init(&parser, true);
+		assert_false(parser_append(&parser, input),
+		    "Unexpected input appending failure");
+		assert_false(parser_parse(&parser),
+		    "Unexpected parse error for input: %s", input);
+		parser_fini(&parser);
+	}
+}
+TEST_END
+
+void
+write_cb(void *opaque, const char *str) {
+	parser_t *parser = (parser_t *)opaque;
+	if (parser_append(parser, str)) {
+		test_fail("Unexpected input appending failure");
+	}
+}
+
+TEST_BEGIN(test_stats_print_json) {
+	const char *opts[] = {
+		"J",
+		"Jg",
+		"Jm",
+		"Jd",
+		"Jmd",
+		"Jgd",
+		"Jgm",
+		"Jgmd",
+		"Ja",
+		"Jb",
+		"Jl",
+		"Jx",
+		"Jbl",
+		"Jal",
+		"Jab",
+		"Jabl",
+		"Jax",
+		"Jbx",
+		"Jlx",
+		"Jablx",
+		"Jgmdablx",
+	};
+	unsigned arena_ind, i;
+
+	for (i = 0; i < 3; i++) {
+		unsigned j;
+
+		switch (i) {
+		case 0:
+			break;
+		case 1: {
+			size_t sz = sizeof(arena_ind);
+			assert_d_eq(mallctl("arenas.create", (void *)&arena_ind,
+			    &sz, NULL, 0), 0, "Unexpected mallctl failure");
+			break;
+		} case 2: {
+			size_t mib[3];
+			size_t miblen = sizeof(mib)/sizeof(size_t);
+			assert_d_eq(mallctlnametomib("arena.0.destroy",
+			    mib, &miblen), 0,
+			    "Unexpected mallctlnametomib failure");
+			mib[1] = arena_ind;
+			assert_d_eq(mallctlbymib(mib, miblen, NULL, NULL, NULL,
+			    0), 0, "Unexpected mallctlbymib failure");
+			break;
+		} default:
+			not_reached();
+		}
+
+		for (j = 0; j < sizeof(opts)/sizeof(const char *); j++) {
+			parser_t parser;
+
+			parser_init(&parser, true);
+			malloc_stats_print(write_cb, (void *)&parser, opts[j]);
+			assert_false(parser_parse(&parser),
+			    "Unexpected parse error, opts=\"%s\"", opts[j]);
+			parser_fini(&parser);
+		}
+	}
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_json_parser,
+	    test_stats_print_json);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/test_hooks.c b/memkind-1.10.0/jemalloc/test/unit/test_hooks.c
new file mode 100644
index 0000000..ded8698
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/test_hooks.c
@@ -0,0 +1,38 @@
+#include "test/jemalloc_test.h"
+
+static bool hook_called = false;
+
+static void
+hook() {
+	hook_called = true;
+}
+
+static int
+func_to_hook(int arg1, int arg2) {
+	return arg1 + arg2;
+}
+
+#define func_to_hook JEMALLOC_HOOK(func_to_hook, test_hooks_libc_hook)
+
+TEST_BEGIN(unhooked_call) {
+	test_hooks_libc_hook = NULL;
+	hook_called = false;
+	assert_d_eq(3, func_to_hook(1, 2), "Hooking changed return value.");
+	assert_false(hook_called, "Nulling out hook didn't take.");
+}
+TEST_END
+
+TEST_BEGIN(hooked_call) {
+	test_hooks_libc_hook = &hook;
+	hook_called = false;
+	assert_d_eq(3, func_to_hook(1, 2), "Hooking changed return value.");
+	assert_true(hook_called, "Hook should have executed.");
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    unhooked_call,
+	    hooked_call);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/ticker.c b/memkind-1.10.0/jemalloc/test/unit/ticker.c
new file mode 100644
index 0000000..e5790a3
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/ticker.c
@@ -0,0 +1,73 @@
+#include "test/jemalloc_test.h"
+
+#include "jemalloc/internal/ticker.h"
+
+TEST_BEGIN(test_ticker_tick) {
+#define NREPS 2
+#define NTICKS 3
+	ticker_t ticker;
+	int32_t i, j;
+
+	ticker_init(&ticker, NTICKS);
+	for (i = 0; i < NREPS; i++) {
+		for (j = 0; j < NTICKS; j++) {
+			assert_u_eq(ticker_read(&ticker), NTICKS - j,
+			    "Unexpected ticker value (i=%d, j=%d)", i, j);
+			assert_false(ticker_tick(&ticker),
+			    "Unexpected ticker fire (i=%d, j=%d)", i, j);
+		}
+		assert_u32_eq(ticker_read(&ticker), 0,
+		    "Expected ticker depletion");
+		assert_true(ticker_tick(&ticker),
+		    "Expected ticker fire (i=%d)", i);
+		assert_u32_eq(ticker_read(&ticker), NTICKS,
+		    "Expected ticker reset");
+	}
+#undef NTICKS
+}
+TEST_END
+
+TEST_BEGIN(test_ticker_ticks) {
+#define NTICKS 3
+	ticker_t ticker;
+
+	ticker_init(&ticker, NTICKS);
+
+	assert_u_eq(ticker_read(&ticker), NTICKS, "Unexpected ticker value");
+	assert_false(ticker_ticks(&ticker, NTICKS), "Unexpected ticker fire");
+	assert_u_eq(ticker_read(&ticker), 0, "Unexpected ticker value");
+	assert_true(ticker_ticks(&ticker, NTICKS), "Expected ticker fire");
+	assert_u_eq(ticker_read(&ticker), NTICKS, "Unexpected ticker value");
+
+	assert_true(ticker_ticks(&ticker, NTICKS + 1), "Expected ticker fire");
+	assert_u_eq(ticker_read(&ticker), NTICKS, "Unexpected ticker value");
+#undef NTICKS
+}
+TEST_END
+
+TEST_BEGIN(test_ticker_copy) {
+#define NTICKS 3
+	ticker_t ta, tb;
+
+	ticker_init(&ta, NTICKS);
+	ticker_copy(&tb, &ta);
+	assert_u_eq(ticker_read(&tb), NTICKS, "Unexpected ticker value");
+	assert_true(ticker_ticks(&tb, NTICKS + 1), "Expected ticker fire");
+	assert_u_eq(ticker_read(&tb), NTICKS, "Unexpected ticker value");
+
+	ticker_tick(&ta);
+	ticker_copy(&tb, &ta);
+	assert_u_eq(ticker_read(&tb), NTICKS - 1, "Unexpected ticker value");
+	assert_true(ticker_ticks(&tb, NTICKS), "Expected ticker fire");
+	assert_u_eq(ticker_read(&tb), NTICKS, "Unexpected ticker value");
+#undef NTICKS
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_ticker_tick,
+	    test_ticker_ticks,
+	    test_ticker_copy);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/tsd.c b/memkind-1.10.0/jemalloc/test/unit/tsd.c
new file mode 100644
index 0000000..917884d
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/tsd.c
@@ -0,0 +1,267 @@
+#include "test/jemalloc_test.h"
+
+/*
+ * If we're e.g. in debug mode, we *never* enter the fast path, and so shouldn't
+ * be asserting that we're on one.
+ */
+static bool originally_fast;
+static int data_cleanup_count;
+
+void
+data_cleanup(int *data) {
+	if (data_cleanup_count == 0) {
+		assert_x_eq(*data, MALLOC_TSD_TEST_DATA_INIT,
+		    "Argument passed into cleanup function should match tsd "
+		    "value");
+	}
+	++data_cleanup_count;
+
+	/*
+	 * Allocate during cleanup for two rounds, in order to assure that
+	 * jemalloc's internal tsd reinitialization happens.
+	 */
+	bool reincarnate = false;
+	switch (*data) {
+	case MALLOC_TSD_TEST_DATA_INIT:
+		*data = 1;
+		reincarnate = true;
+		break;
+	case 1:
+		*data = 2;
+		reincarnate = true;
+		break;
+	case 2:
+		return;
+	default:
+		not_reached();
+	}
+
+	if (reincarnate) {
+		void *p = mallocx(1, 0);
+		assert_ptr_not_null(p, "Unexpeced mallocx() failure");
+		dallocx(p, 0);
+	}
+}
+
+static void *
+thd_start(void *arg) {
+	int d = (int)(uintptr_t)arg;
+	void *p;
+
+	tsd_t *tsd = tsd_fetch();
+	assert_x_eq(tsd_test_data_get(tsd), MALLOC_TSD_TEST_DATA_INIT,
+	    "Initial tsd get should return initialization value");
+
+	p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	tsd_test_data_set(tsd, d);
+	assert_x_eq(tsd_test_data_get(tsd), d,
+	    "After tsd set, tsd get should return value that was set");
+
+	d = 0;
+	assert_x_eq(tsd_test_data_get(tsd), (int)(uintptr_t)arg,
+	    "Resetting local data should have no effect on tsd");
+
+	tsd_test_callback_set(tsd, &data_cleanup);
+
+	free(p);
+	return NULL;
+}
+
+TEST_BEGIN(test_tsd_main_thread) {
+	thd_start((void *)(uintptr_t)0xa5f3e329);
+}
+TEST_END
+
+TEST_BEGIN(test_tsd_sub_thread) {
+	thd_t thd;
+
+	data_cleanup_count = 0;
+	thd_create(&thd, thd_start, (void *)MALLOC_TSD_TEST_DATA_INIT);
+	thd_join(thd, NULL);
+	/*
+	 * We reincarnate twice in the data cleanup, so it should execute at
+	 * least 3 times.
+	 */
+	assert_x_ge(data_cleanup_count, 3,
+	    "Cleanup function should have executed multiple times.");
+}
+TEST_END
+
+static void *
+thd_start_reincarnated(void *arg) {
+	tsd_t *tsd = tsd_fetch();
+	assert(tsd);
+
+	void *p = malloc(1);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+
+	/* Manually trigger reincarnation. */
+	assert_ptr_not_null(tsd_arena_get(tsd),
+	    "Should have tsd arena set.");
+	tsd_cleanup((void *)tsd);
+	assert_ptr_null(*tsd_arenap_get_unsafe(tsd),
+	    "TSD arena should have been cleared.");
+	assert_u_eq(tsd_state_get(tsd), tsd_state_purgatory,
+	    "TSD state should be purgatory\n");
+
+	free(p);
+	assert_u_eq(tsd_state_get(tsd), tsd_state_reincarnated,
+	    "TSD state should be reincarnated\n");
+	p = mallocx(1, MALLOCX_TCACHE_NONE);
+	assert_ptr_not_null(p, "Unexpected malloc() failure");
+	assert_ptr_null(*tsd_arenap_get_unsafe(tsd),
+	    "Should not have tsd arena set after reincarnation.");
+
+	free(p);
+	tsd_cleanup((void *)tsd);
+	assert_ptr_null(*tsd_arenap_get_unsafe(tsd),
+	    "TSD arena should have been cleared after 2nd cleanup.");
+
+	return NULL;
+}
+
+TEST_BEGIN(test_tsd_reincarnation) {
+	thd_t thd;
+	thd_create(&thd, thd_start_reincarnated, NULL);
+	thd_join(thd, NULL);
+}
+TEST_END
+
+typedef struct {
+	atomic_u32_t phase;
+	atomic_b_t error;
+} global_slow_data_t;
+
+static void *
+thd_start_global_slow(void *arg) {
+	/* PHASE 0 */
+	global_slow_data_t *data = (global_slow_data_t *)arg;
+	free(mallocx(1, 0));
+
+	tsd_t *tsd = tsd_fetch();
+	/*
+	 * No global slowness has happened yet; there was an error if we were
+	 * originally fast but aren't now.
+	 */
+	atomic_store_b(&data->error, originally_fast && !tsd_fast(tsd),
+	    ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 1, ATOMIC_SEQ_CST);
+
+	/* PHASE 2 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 2) {
+	}
+	free(mallocx(1, 0));
+	atomic_store_b(&data->error, tsd_fast(tsd), ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 3, ATOMIC_SEQ_CST);
+
+	/* PHASE 4 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 4) {
+	}
+	free(mallocx(1, 0));
+	atomic_store_b(&data->error, tsd_fast(tsd), ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 5, ATOMIC_SEQ_CST);
+
+	/* PHASE 6 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 6) {
+	}
+	free(mallocx(1, 0));
+	/* Only one decrement so far. */
+	atomic_store_b(&data->error, tsd_fast(tsd), ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 7, ATOMIC_SEQ_CST);
+
+	/* PHASE 8 */
+	while (atomic_load_u32(&data->phase, ATOMIC_SEQ_CST) != 8) {
+	}
+	free(mallocx(1, 0));
+	/*
+	 * Both decrements happened; we should be fast again (if we ever
+	 * were)
+	 */
+	atomic_store_b(&data->error, originally_fast && !tsd_fast(tsd),
+	    ATOMIC_SEQ_CST);
+	atomic_store_u32(&data->phase, 9, ATOMIC_SEQ_CST);
+
+	return NULL;
+}
+
+TEST_BEGIN(test_tsd_global_slow) {
+	global_slow_data_t data = {ATOMIC_INIT(0), ATOMIC_INIT(false)};
+	/*
+	 * Note that the "mallocx" here (vs. malloc) is important, since the
+	 * compiler is allowed to optimize away free(malloc(1)) but not
+	 * free(mallocx(1)).
+	 */
+	free(mallocx(1, 0));
+	tsd_t *tsd = tsd_fetch();
+	originally_fast = tsd_fast(tsd);
+
+	thd_t thd;
+	thd_create(&thd, thd_start_global_slow, (void *)&data.phase);
+	/* PHASE 1 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 1) {
+		/*
+		 * We don't have a portable condvar/semaphore mechanism.
+		 * Spin-wait.
+		 */
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	tsd_global_slow_inc(tsd_tsdn(tsd));
+	free(mallocx(1, 0));
+	assert_false(tsd_fast(tsd), "");
+	atomic_store_u32(&data.phase, 2, ATOMIC_SEQ_CST);
+
+	/* PHASE 3 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 3) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	/* Increase again, so that we can test multiple fast/slow changes. */
+	tsd_global_slow_inc(tsd_tsdn(tsd));
+	atomic_store_u32(&data.phase, 4, ATOMIC_SEQ_CST);
+	free(mallocx(1, 0));
+	assert_false(tsd_fast(tsd), "");
+
+	/* PHASE 5 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 5) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	tsd_global_slow_dec(tsd_tsdn(tsd));
+	atomic_store_u32(&data.phase, 6, ATOMIC_SEQ_CST);
+	/* We only decreased once; things should still be slow. */
+	free(mallocx(1, 0));
+	assert_false(tsd_fast(tsd), "");
+
+	/* PHASE 7 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 7) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+	tsd_global_slow_dec(tsd_tsdn(tsd));
+	atomic_store_u32(&data.phase, 8, ATOMIC_SEQ_CST);
+	/* We incremented and then decremented twice; we should be fast now. */
+	free(mallocx(1, 0));
+	assert_true(!originally_fast || tsd_fast(tsd), "");
+
+	/* PHASE 9 */
+	while (atomic_load_u32(&data.phase, ATOMIC_SEQ_CST) != 9) {
+	}
+	assert_false(atomic_load_b(&data.error, ATOMIC_SEQ_CST), "");
+
+	thd_join(thd, NULL);
+}
+TEST_END
+
+int
+main(void) {
+	/* Ensure tsd bootstrapped. */
+	if (nallocx(1, 0) == 0) {
+		malloc_printf("Initialization error");
+		return test_status_fail;
+	}
+
+	return test_no_reentrancy(
+	    test_tsd_main_thread,
+	    test_tsd_sub_thread,
+	    test_tsd_reincarnation,
+	    test_tsd_global_slow);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/witness.c b/memkind-1.10.0/jemalloc/test/unit/witness.c
new file mode 100644
index 0000000..5986da4
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/witness.c
@@ -0,0 +1,280 @@
+#include "test/jemalloc_test.h"
+
+static witness_lock_error_t *witness_lock_error_orig;
+static witness_owner_error_t *witness_owner_error_orig;
+static witness_not_owner_error_t *witness_not_owner_error_orig;
+static witness_depth_error_t *witness_depth_error_orig;
+
+static bool saw_lock_error;
+static bool saw_owner_error;
+static bool saw_not_owner_error;
+static bool saw_depth_error;
+
+static void
+witness_lock_error_intercept(const witness_list_t *witnesses,
+    const witness_t *witness) {
+	saw_lock_error = true;
+}
+
+static void
+witness_owner_error_intercept(const witness_t *witness) {
+	saw_owner_error = true;
+}
+
+static void
+witness_not_owner_error_intercept(const witness_t *witness) {
+	saw_not_owner_error = true;
+}
+
+static void
+witness_depth_error_intercept(const witness_list_t *witnesses,
+    witness_rank_t rank_inclusive, unsigned depth) {
+	saw_depth_error = true;
+}
+
+static int
+witness_comp(const witness_t *a, void *oa, const witness_t *b, void *ob) {
+	assert_u_eq(a->rank, b->rank, "Witnesses should have equal rank");
+
+	assert(oa == (void *)a);
+	assert(ob == (void *)b);
+
+	return strcmp(a->name, b->name);
+}
+
+static int
+witness_comp_reverse(const witness_t *a, void *oa, const witness_t *b,
+    void *ob) {
+	assert_u_eq(a->rank, b->rank, "Witnesses should have equal rank");
+
+	assert(oa == (void *)a);
+	assert(ob == (void *)b);
+
+	return -strcmp(a->name, b->name);
+}
+
+TEST_BEGIN(test_witness) {
+	witness_t a, b;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 0);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+	witness_assert_not_owner(&witness_tsdn, &a);
+	witness_lock(&witness_tsdn, &a);
+	witness_assert_owner(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)2U, 0);
+
+	witness_init(&b, "b", 2, NULL, NULL);
+	witness_assert_not_owner(&witness_tsdn, &b);
+	witness_lock(&witness_tsdn, &b);
+	witness_assert_owner(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 2);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 2);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)2U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)3U, 0);
+
+	witness_unlock(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)2U, 1);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)3U, 0);
+	witness_unlock(&witness_tsdn, &b);
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+	witness_assert_depth_to_rank(&witness_tsdn, (witness_rank_t)1U, 0);
+}
+TEST_END
+
+TEST_BEGIN(test_witness_comp) {
+	witness_t a, b, c, d;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, witness_comp, &a);
+	witness_assert_not_owner(&witness_tsdn, &a);
+	witness_lock(&witness_tsdn, &a);
+	witness_assert_owner(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	witness_init(&b, "b", 1, witness_comp, &b);
+	witness_assert_not_owner(&witness_tsdn, &b);
+	witness_lock(&witness_tsdn, &b);
+	witness_assert_owner(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 2);
+	witness_unlock(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	witness_lock_error_orig = witness_lock_error;
+	witness_lock_error = witness_lock_error_intercept;
+	saw_lock_error = false;
+
+	witness_init(&c, "c", 1, witness_comp_reverse, &c);
+	witness_assert_not_owner(&witness_tsdn, &c);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	witness_lock(&witness_tsdn, &c);
+	assert_true(saw_lock_error, "Expected witness lock error");
+	witness_unlock(&witness_tsdn, &c);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	saw_lock_error = false;
+
+	witness_init(&d, "d", 1, NULL, NULL);
+	witness_assert_not_owner(&witness_tsdn, &d);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	witness_lock(&witness_tsdn, &d);
+	assert_true(saw_lock_error, "Expected witness lock error");
+	witness_unlock(&witness_tsdn, &d);
+	witness_assert_depth(&witness_tsdn, 1);
+
+	witness_unlock(&witness_tsdn, &a);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_lock_error = witness_lock_error_orig;
+}
+TEST_END
+
+TEST_BEGIN(test_witness_reversal) {
+	witness_t a, b;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_lock_error_orig = witness_lock_error;
+	witness_lock_error = witness_lock_error_intercept;
+	saw_lock_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+	witness_init(&b, "b", 2, NULL, NULL);
+
+	witness_lock(&witness_tsdn, &b);
+	witness_assert_depth(&witness_tsdn, 1);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	witness_lock(&witness_tsdn, &a);
+	assert_true(saw_lock_error, "Expected witness lock error");
+
+	witness_unlock(&witness_tsdn, &a);
+	witness_assert_depth(&witness_tsdn, 1);
+	witness_unlock(&witness_tsdn, &b);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_lock_error = witness_lock_error_orig;
+}
+TEST_END
+
+TEST_BEGIN(test_witness_recursive) {
+	witness_t a;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_not_owner_error_orig = witness_not_owner_error;
+	witness_not_owner_error = witness_not_owner_error_intercept;
+	saw_not_owner_error = false;
+
+	witness_lock_error_orig = witness_lock_error;
+	witness_lock_error = witness_lock_error_intercept;
+	saw_lock_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+
+	witness_lock(&witness_tsdn, &a);
+	assert_false(saw_lock_error, "Unexpected witness lock error");
+	assert_false(saw_not_owner_error, "Unexpected witness not owner error");
+	witness_lock(&witness_tsdn, &a);
+	assert_true(saw_lock_error, "Expected witness lock error");
+	assert_true(saw_not_owner_error, "Expected witness not owner error");
+
+	witness_unlock(&witness_tsdn, &a);
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_owner_error = witness_owner_error_orig;
+	witness_lock_error = witness_lock_error_orig;
+
+}
+TEST_END
+
+TEST_BEGIN(test_witness_unlock_not_owned) {
+	witness_t a;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_owner_error_orig = witness_owner_error;
+	witness_owner_error = witness_owner_error_intercept;
+	saw_owner_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+
+	assert_false(saw_owner_error, "Unexpected owner error");
+	witness_unlock(&witness_tsdn, &a);
+	assert_true(saw_owner_error, "Expected owner error");
+
+	witness_assert_lockless(&witness_tsdn);
+
+	witness_owner_error = witness_owner_error_orig;
+}
+TEST_END
+
+TEST_BEGIN(test_witness_depth) {
+	witness_t a;
+	witness_tsdn_t witness_tsdn = { WITNESS_TSD_INITIALIZER };
+
+	test_skip_if(!config_debug);
+
+	witness_depth_error_orig = witness_depth_error;
+	witness_depth_error = witness_depth_error_intercept;
+	saw_depth_error = false;
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+
+	witness_init(&a, "a", 1, NULL, NULL);
+
+	assert_false(saw_depth_error, "Unexpected depth error");
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+
+	witness_lock(&witness_tsdn, &a);
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+	assert_true(saw_depth_error, "Expected depth error");
+
+	witness_unlock(&witness_tsdn, &a);
+
+	witness_assert_lockless(&witness_tsdn);
+	witness_assert_depth(&witness_tsdn, 0);
+
+	witness_depth_error = witness_depth_error_orig;
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_witness,
+	    test_witness_comp,
+	    test_witness_reversal,
+	    test_witness_recursive,
+	    test_witness_unlock_not_owned,
+	    test_witness_depth);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/zero.c b/memkind-1.10.0/jemalloc/test/unit/zero.c
new file mode 100644
index 0000000..271fd5c
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/zero.c
@@ -0,0 +1,59 @@
+#include "test/jemalloc_test.h"
+
+static void
+test_zero(size_t sz_min, size_t sz_max) {
+	uint8_t *s;
+	size_t sz_prev, sz, i;
+#define MAGIC	((uint8_t)0x61)
+
+	sz_prev = 0;
+	s = (uint8_t *)mallocx(sz_min, 0);
+	assert_ptr_not_null((void *)s, "Unexpected mallocx() failure");
+
+	for (sz = sallocx(s, 0); sz <= sz_max;
+	    sz_prev = sz, sz = sallocx(s, 0)) {
+		if (sz_prev > 0) {
+			assert_u_eq(s[0], MAGIC,
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    ZU(0), sz_prev);
+			assert_u_eq(s[sz_prev-1], MAGIC,
+			    "Previously allocated byte %zu/%zu is corrupted",
+			    sz_prev-1, sz_prev);
+		}
+
+		for (i = sz_prev; i < sz; i++) {
+			assert_u_eq(s[i], 0x0,
+			    "Newly allocated byte %zu/%zu isn't zero-filled",
+			    i, sz);
+			s[i] = MAGIC;
+		}
+
+		if (xallocx(s, sz+1, 0, 0) == sz) {
+			s = (uint8_t *)rallocx(s, sz+1, 0);
+			assert_ptr_not_null((void *)s,
+			    "Unexpected rallocx() failure");
+		}
+	}
+
+	dallocx(s, 0);
+#undef MAGIC
+}
+
+TEST_BEGIN(test_zero_small) {
+	test_skip_if(!config_fill);
+	test_zero(1, SC_SMALL_MAXCLASS - 1);
+}
+TEST_END
+
+TEST_BEGIN(test_zero_large) {
+	test_skip_if(!config_fill);
+	test_zero(SC_SMALL_MAXCLASS + 1, 1U << (SC_LG_LARGE_MINCLASS + 1));
+}
+TEST_END
+
+int
+main(void) {
+	return test(
+	    test_zero_small,
+	    test_zero_large);
+}
diff --git a/memkind-1.10.0/jemalloc/test/unit/zero.sh b/memkind-1.10.0/jemalloc/test/unit/zero.sh
new file mode 100644
index 0000000..b4540b2
--- /dev/null
+++ b/memkind-1.10.0/jemalloc/test/unit/zero.sh
@@ -0,0 +1,5 @@
+#!/bin/sh
+
+if [ "x${enable_fill}" = "x1" ] ; then
+  export MALLOC_CONF="abort:false,junk:false,zero:true"
+fi
diff --git a/memkind-1.10.0/make_rpm.mk b/memkind-1.10.0/make_rpm.mk
new file mode 100644
index 0000000..cfafd8e
--- /dev/null
+++ b/memkind-1.10.0/make_rpm.mk
@@ -0,0 +1,106 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+package_prefix ?=
+name = $(package_prefix)memkind
+arch = $(shell uname -p)
+version = 0.0.0
+release = 1
+
+src = $(shell cat MANIFEST)
+
+topdir = $(HOME)/rpmbuild
+rpm = $(topdir)/RPMS/$(arch)/$(name)-devel-$(version)-$(release).$(arch).rpm
+trpm = $(topdir)/TRPMS/$(arch)/$(name)-tests-$(version)-$(release).$(arch).rpm
+srpm = $(topdir)/SRPMS/$(name)-$(version)-$(release).src.rpm
+specfile = $(topdir)/SPECS/$(name)-$(version).spec
+source_tar = $(topdir)/SOURCES/memkind-$(version).tar.gz
+source_tmp_dir := $(topdir)/SOURCES/$(name)-tmp-$(shell date +%s)
+rpmbuild_flags = -E '%define _topdir $(topdir)'
+rpmclean_flags = $(rpmbuild_flags) --clean --rmsource --rmspec
+memkind_test_dir = $(MPSS_TEST_BASEDIR)/memkind-dt
+exclude_source_files = test/memkind-afts.ts \
+test/memkind-afts-ext.ts \
+test/memkind-slts.ts \
+test/memkind-perf.ts \
+test/memkind-perf-ext.ts \
+test/memkind-pytests.ts \
+test/python_framework/cmd_helper.py \
+test/python_framework/huge_page_organizer.py \
+test/python_framework/__init__.py \
+test/hbw_detection_test.py \
+test/dax_kmem_env_var_test.py \
+test/autohbw_test.py \
+test/trace_mechanism_test.py \
+test/draw_plots.py \
+VERSION
+
+all: $(rpm)
+
+$(rpm): $(specfile) $(source_tar)
+	rpmbuild $(rpmbuild_flags) $(specfile) -ba
+	mkdir -p $(topdir)/TRPMS/$(arch)
+	mv $(topdir)/RPMS/$(arch)/$(name)-tests-$(version)-$(release).$(arch).rpm $(trpm)
+
+$(source_tar): $(topdir)/.setup $(src) MANIFEST
+	mkdir -p $(source_tmp_dir)
+	set -e ; \
+	for f in $(exclude_source_files); do \
+		echo $$f >> $(source_tmp_dir)/EXCLUDE ; \
+	done
+	tar cf $(source_tmp_dir)/tmp.tar -T MANIFEST --transform="s|^|$(name)-$(version)/|" -X $(source_tmp_dir)/EXCLUDE
+	cd $(source_tmp_dir) && tar xf tmp.tar
+	set -e ; \
+	for f in $(exclude_source_files); do \
+		mkdir -p `dirname $(source_tmp_dir)/$(name)-$(version)/$$f` ; \
+		cp $$f $(source_tmp_dir)/$(name)-$(version)/$$f ; \
+	done
+	cd $(source_tmp_dir)/$(name)-$(version) && ./autogen.sh && ./configure && make dist; \
+	# tar.gz produced by "make dist" from above produces memkind-$(version).tar.gz
+	# If $(package_prefix) is not empty, then need to repackage that tar.gz to $(name)-$(version)
+	# thus below command. Otherwise, rpmbuild will fail.
+	if [ -n "$(package_prefix)" ]; then \
+	    cd $(source_tmp_dir)/$(name)-$(version) && \
+	    tar xf memkind-$(version).tar.gz && \
+	    rm -rf memkind-$(version).tar.gz && \
+	    mv memkind-$(version) $(name)-$(version) && \
+	    tar cfz memkind-$(version).tar.gz $(name)-$(version); \
+	fi
+	mv $(source_tmp_dir)/$(name)-$(version)/memkind-$(version).tar.gz $@
+	rm -rf $(source_tmp_dir)
+
+$(specfile): $(topdir)/.setup memkind.spec.mk
+	@echo "$$memkind_spec" > $@
+
+$(topdir)/.setup:
+	mkdir -p $(topdir)/{SOURCES,SPECS}
+	touch $@
+
+clean:
+	-rpmbuild $(rpmclean_flags) $(specfile)
+
+.PHONY: all clean
+
+include memkind.spec.mk
+
diff --git a/memkind-1.10.0/man/autohbw.7 b/memkind-1.10.0/man/autohbw.7
new file mode 100644
index 0000000..a1a4232
--- /dev/null
+++ b/memkind-1.10.0/man/autohbw.7
@@ -0,0 +1,195 @@
+.\"
+.\" Copyright (C) 2014-2016 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "AUTOHBW" 7 "2016-07-28" "Intel Corporation" "AUTOHBW" \" -*- nroff -*-
+.SH "NAME"
+libautohbw.so \- An interposer library for redirecting heap allocations
+.SH "SYNOPSIS"
+.BR LD_PRELOAD=libautohbw.so
+command {arguments ...}
+.SH "DESCRIPTION"
+.B AutoHBW
+library
+.BR (libautohbw.so)
+is an interposer library for redirecting heap allocations
+.B (malloc, calloc, realloc, valloc, posix_memalign, memlign)
+to high-bandwidth
+.B (HBW)
+memory. Consequently, AutoHBW library can be used to
+automatically allocate high-bandwidth memory without any modification to
+source code of an application.
+
+.br
+For instance, the following command-line runs existing binary /bin/ls with
+AutoHBW library, automatically redirecting heap allocations (larger than a given
+threshold) to high-bandwidth memory.
+.IP
+.B LD_PRELOAD=libautohbw.so
+/bin/ls
+
+.SH "ENVIRONMENT"
+
+The behavior of AutoHBW library is controlled by the following environment
+variables.
+
+.PP
+.B AUTO_HBW_SIZE=x:[y]
+.br
+Indicates that any allocation larger than
+.I x
+and smaller than
+.I y
+should be
+allocated in HBW memory.
+.I x
+and
+.I y
+can be followed by a K, M, or G to indicate
+the size in Kilo/Memga/Giga bytes (e.g., 4K, 3M, 2G).
+.br
+
+Examples:
+.IP AUTO_HBW_SIZE=4K
+# allocations larger than 4K allocated in HBW
+.IP AUTO_HBW_SIZE=1M:5M
+# allocations between 1M and 5M allocated in HBW
+
+.PP
+.B AUTO_HBW_LOG=level
+.br
+Sets the value of logging (printing)
+.I level.
+If
+.I level
+is:
+.br
+.IP -1
+no messages are printed
+.br
+.IP 0
+no allocations messages are printed but INFO messages are printed
+.br
+.IP 1
+a log message is printed for each allocation (Default)
+.br
+.IP 2
+a log message is printed for each allocation with a backtrace.
+Redirect this output and use
+.B autohbw_get_src_lines.pl
+to find source lines for each allocation. Your application must
+be compiled with
+.B -g
+to see source lines.
+.PP
+Notes:
+.IP
+1. Logging adds extra overhead. Therefore, for performance
+critical runs, logging level should be 0
+.IP
+2. The amount of free memory printed with log messages is only
+approximate -- e.g. pages that are not touched yet are excluded
+.PP
+Examples:
+.IP AUTO_HBW_LOG=1
+
+.PP
+.B AUTO_HBW_MEM_TYPE=memory_type
+.br
+Sets the type of memory type that should be automatically allocated. By
+default, this type is MEMKIND_HBW_PREFERRED, if MCDRAM is found in your
+system; otherwise, the default is MEMKIND_DEFAULT. The names of memory
+types are defined in
+.B memkind(3)
+man page.
+.B memory_type
+has to be one of
+.B MEMKIND_DEFAULT, MEMKIND_HUGETLB, MEMKIND_INTERLEAVE, MEMKIND_HBW,
+.B MEMKIND_HBW_PREFERRED, MEMKIND_HBW_HUGETLB, MEMKIND_HBW_PREFERRED_HUGETLB,
+.B MEMKIND_HBW_GBTLB (DEPRECATED), MEMKIND_HBW_PREFERRED_GBTLB (DEPRECATED), MEMKIND_GBTLB (DEPRECATED),
+.B MEMKIND_HBW_INTERLEAVE
+
+If you are requesting any huge
+TLB pages, please make sure that the requested type is currently enabled
+in your OS.
+
+Examples:
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_HBW_PREFERRED
+# (default, if MCDRAM present)
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_DEFAULT
+# (default, if MCDRAM absent)
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_HBW_HUGETLB
+.IP AUTO_HBW_MEM_TYPE=MEMKIND_HUGETLB
+
+.PP
+.B AUTO_HBW_DEBUG=0|1|2
+.br
+Set the debug message printing level. Default is 0. This is mainly for
+development.
+
+.SH "NOTES"
+It is possible to temporarily disable/enable automatic HBW allocations by
+calling disableAutoHBW() and enableAutoHBW() in source code. To call
+these routines, please include
+.B autohbw_api.h
+header file and link with -lautohbw.
+
+
+.br
+If high-bandwidth memory is not physically present in your system,
+the environment variable
+.B MEMKIND_HBW_NODES
+must be set to indicate the high-bandwidth node as indicated in
+.B memkind(3).
+
+
+.SH "EXAMPLES"
+.br
+The following will run /bin/ls with AutoHBW library. Make sure that paths to
+both libautohbw.so and libmemkind.so are included in
+.B LD_LIBRARY_PATH.
+.IP
+.B LD_PRELOAD=libautohbw.so
+/bin/ls -l
+.PP
+To run with MPI, a shell script must be created, with the correct LD_PRELOAD
+command for each rank. For example, if we put
+.B LD_PRELOAD=libautohbw.so /bin/ls
+in a shell script named autohbw_test.sh, it can be executed with 2 MPI ranks as:
+.br
+.IP
+.B mpirun -n 2 ./autohbw_test.sh
+
+.SH "COPYRIGHT"
+Copyright (C) 2014, 2015, 2016 Intel Corporation. All rights reserved.
+
+.SH "SEE ALSO"
+.BR memkind(3)
+.BR malloc (3),
+.BR numactl (8),
+
+
+
+
+
+
+
diff --git a/memkind-1.10.0/man/hbwallocator.3 b/memkind-1.10.0/man/hbwallocator.3
new file mode 100644
index 0000000..031c577
--- /dev/null
+++ b/memkind-1.10.0/man/hbwallocator.3
@@ -0,0 +1,125 @@
+.\"
+.\" Copyright (C) 2015 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "HBWALLOCATOR" 3 "2015-11-02" "Intel Corporation" "HBWALLOCATOR" \" -*- nroff -*-
+.SH "NAME"
+hbw::allocator<T> \- The C++ allocator compatible with the C++ standard library allocator concepts
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.nf
+.B #include <hbw_allocator.h>
+.sp
+.B Link with -lmemkind
+.sp
+.BI "hbw::allocator();"
+.br
+.BI "template <class U>hbw::allocator<T>::allocator(const hbw::allocator<U>&);
+.br
+.BI "hbw::allocator<T>::~allocator()'"
+.br
+.BI "hbw::allocator<T>::pointer hbw::allocator<T>::address(hbw::allocator<T>::reference " "x" );
+.br
+.BI "hbw::allocator<T>::const_pointer hbw::allocator<T>::address(hbw::allocator<T>::const_reference " "x" );
+.br
+.BI "hbw::allocator<T>::pointer hbw::allocator<T>::allocate(hbw::allocator<T>::size_type " "n" ", const void * = 0);
+.br
+.BI "void hbw::allocator<T>::deallocate(hbw::allocator<T>::pointer " "p" ", hbw::allocator<T>::size_type " "n" );
+.br
+.BI "hbw::allocator<T>::size_type  hbw::allocator<T>::max_size();
+.br
+.BI "void hbw::allocator<T>::construct(hbw::allocator<T>::pointer " "p" ", const hbw::allocator<T>::value_type& " "val" );
+.br
+.BI "void hbw::allocator<T>::destroy(hbw::allocator<T>::pointer " "p" );
+.fi
+.SH "DESCRIPTION"
+The
+.BR hbw::allocator<T>
+is intended to be used with STL containers to allocate high bandwidth memory. Memory management is based on hbwmalloc (memkind library), enabling users to gain performance in multithreaded applications. Refer
+.BR hbwmalloc (3)
+and
+.BR memkind (3)
+man page for more details.
+.PP
+All public member types and functions corresponds to standard library allocator concepts and definitions. The current implementation supports C++03 standard.
+.PP
+Template arguments:
+.br
+.I T
+is an object type aliased by value_type.
+.br
+.I U
+is an object type.
+.PP
+.B Note:
+.br
+.BI "hbw::allocator<T>::pointer hbw::allocator<T>::allocate(hbw::allocator<T>::size_type " "n" ", const void * = 0)"
+allocates high bandwidth memory using
+.BR hbw_malloc ().
+Throw
+.I std::bad_alloc
+when:
+.br
+.RS
+.I n
+= 0,
+.RE
+.br
+.RS
+.I n
+>
+.BR max_size ()
+.RE
+.br
+.RS
+or there is not enough memory to satisfy the request.
+.RE
+.PP
+.BI "hbw::allocator<T>::deallocate(hbw::allocator<T>::pointer " "p" ", hbw::allocator<T>::size_type " "n")
+deallocates memory associated with pointer returned by
+.BR allocate ()
+using
+.BR hbw_free ().
+
+.SH ERRORS
+The same as described in
+.B ERRORS
+section of
+.BR hbwmalloc (3)
+man page.
+.SH "NOTES"
+The
+.BR hbw::allocator<T>
+behavior depends on hbwmalloc heap management policy. To get and set the policy please use
+.BR hbw_get_policy ()
+and
+.BR hbw_set_policy ()
+respectively.
+.SH "COPYRIGHT"
+Copyright (C) 2015 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR hbwmalloc (3),
+.BR numa (3),
+.BR numactl (8),
+.BR memkind (3)
diff --git a/memkind-1.10.0/man/hbwmalloc.3 b/memkind-1.10.0/man/hbwmalloc.3
new file mode 100644
index 0000000..4c7329a
--- /dev/null
+++ b/memkind-1.10.0/man/hbwmalloc.3
@@ -0,0 +1,515 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "HBWMALLOC" 3 "2015-03-31" "Intel Corporation" "HBWMALLOC" \" -*- nroff -*-
+.SH "NAME"
+hbwmalloc \- The high bandwidth memory interface
+.br
+.BR Note:
+.I hbwmalloc.h
+functionality is considered as stable API (STANDARD API).
+.SH "SYNOPSIS"
+.nf
+.B #include <hbwmalloc.h>
+.sp
+.B Link with -lmemkind
+.sp
+.B int hbw_check_available(void);
+.br
+.BI "void* hbw_malloc(size_t " "size" );
+.br
+.BI "void* hbw_calloc(size_t " "nmemb" ", size_t " "size" );
+.br
+.BI "void* hbw_realloc (void " "*ptr" ", size_t " "size" );
+.br
+.BI "void hbw_free(void " "*ptr" );
+.br
+.BI "size_t hbw_malloc_usable_size(void " "*ptr" );
+.br
+.BI "int hbw_posix_memalign(void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.br
+.BI "int hbw_posix_memalign_psize(void " "**memptr" ", size_t " "alignment" ", size_t " "size" ", hbw_pagesize_t " "pagesize" );
+.br
+.B hbw_policy_t hbw_get_policy(void);
+.br
+.BI "int hbw_set_policy(hbw_policy_t " "mode" );
+.br
+.BI "int hbw_verify_memory_region(void " "*addr" ", size_t " "size" ", int " "flags" );
+.fi
+.SH "DESCRIPTION"
+.BR hbw_check_available ()
+returns zero if high bandwidth memory is available or an error code
+described in the
+.B ERRORS
+section if not.
+.PP
+.BR hbw_malloc ()
+allocates
+.I size
+bytes of uninitialized high bandwidth memory. The allocated space is
+suitably aligned (after possible pointer coercion) for storage of any
+type of object. If
+.I size
+is zero then
+.BR hbw_malloc ()
+returns
+.IR "NULL" .
+.PP
+.BR hbw_calloc ()
+allocates space for
+.I nmemb
+objects in high bandwidth memory, each
+.I size
+bytes in length. The result is identical to calling
+.BR hbw_malloc ()
+with an argument of
+.IR "nmemb"
+*
+.IR "size" ,
+with the exception that the allocated memory is explicitly
+initialized to zero bytes.  If
+.I nmemb
+or
+.I size
+is 0, then
+.BR hbw_calloc ()
+returns
+.IR "NULL" .
+.PP
+.BR hbw_realloc ()
+changes the size of the previously allocated high bandwidth memory
+referenced by
+.I ptr
+to
+.I size
+bytes. The contents of the memory remain unchanged up to the lesser of
+the new and old sizes. If the new size is larger, the contents of the
+newly allocated portion of the memory are undefined. Upon success, the
+memory referenced by
+.I ptr
+is freed and a pointer to the newly allocated high bandwidth memory is
+returned.
+
+.B Note:
+.BR hbw_realloc ()
+may move the memory allocation, resulting in a different return value
+than
+.IR "ptr" .
+
+If
+.I ptr
+is
+.IR "NULL" ,
+the
+.BR hbw_realloc ()
+function behaves identically to
+.BR hbw_malloc ()
+for the specified size.
+If
+.I size
+is equal to zero, and
+.I ptr
+is not
+.IR "NULL" ,
+then the call is equivalent to
+.I hbw_free(ptr)
+and
+.I NULL
+is returned. The address
+.IR "ptr" ,
+if not
+.IR "NULL" ,
+was returned by a previous call to
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc ()
+or
+.BR hbw_posix_memalign ().
+Otherwise, or if
+.I hbw_free(ptr)
+was called before, undefined behavior occurs.
+
+.B Note:
+.BR hbw_realloc ()
+cannot be used with a pointer returned by
+.BR hbw_posix_memalign_psize ().
+
+.PP
+.BR hbw_free ()
+causes the allocated memory referenced by
+.I ptr
+to be made available for future allocations. If
+.I ptr
+is
+.IR "NULL" ,
+no action occurs.
+The address
+.IR "ptr" ,
+if not
+.IR "NULL" ,
+must have been returned by a previous call to
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc (),
+.BR hbw_posix_memalign ()
+or
+.BR hbw_posix_memalign_psize ().
+Otherwise, if
+.I hbw_free(ptr)
+was called before, undefined behavior occurs.
+.PP
+.BR hbw_malloc_usable_size ()
+returns the number of usable bytes in the block pointed to by
+.IR "ptr" ,
+a pointer to a block of memory allocated by
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc (),
+.BR hbw_posix_memalign (),
+or
+.BR hbw_posix_memalign_psize ().
+.PP
+.BR hbw_posix_memalign ()
+allocates
+.I size
+bytes of high bandwidth memory such that the allocation's base address
+is an even multiple of
+.IR "alignment" ,
+and returns the allocation in the value pointed to by
+.IR "memptr" .
+The requested
+.I alignment
+must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+If
+.I size
+is 0, then
+.BR hbw_posix_memalign ()
+returns 0, with a
+.I NULL
+returned in
+.IR "memptr" .
+.PP
+.BR hbw_posix_memalign_psize ()
+allocates
+.I size
+bytes of high bandwidth memory such that the allocation's base address
+is an even multiple of
+.IR "alignment" ,
+and returns the allocation in the value pointed to by
+.IR "memptr" .
+The requested
+.I alignment
+must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+The memory will be allocated using pages determined by the
+.IR "pagesize"
+variable which may be one of the following enumerated values:
+.TP
+.B HBW_PAGESIZE_4KB
+The four kilobyte page size option. Note that with transparent huge
+pages enabled these allocations may be promoted by the operating
+system to two megabyte pages.
+.TP
+.B HBW_PAGESIZE_2MB
+The two megabyte page size option.
+.BR Note:
+This page size requires huge pages configuration described in
+.BR "SYSTEM CONFIGURATION"
+section.
+.TP
+.B HBW_PAGESIZE_1GB (DEPRECATED)
+This option allows the user to specify arbitrary sizes backed by
+1GB chunks of huge pages. Huge pages are allocated even if the
+size is not a modulo of 1GB.
+.BR Note:
+This page size requires huge pages configuration described in
+.BR "SYSTEM CONFIGURATION"
+section.
+.TP
+.B HBW_PAGESIZE_1GB_STRICT (DEPRECATED)
+The total size of the allocation must be a multiple of 1GB with
+this option, otherwise the allocation will fail.
+.BR Note:
+This page size requires huge pages configuration described in
+.BR "SYSTEM CONFIGURATION"
+section.
+.PP
+.BR Note:
+.BR HBW_PAGESIZE_2MB ,
+.BR HBW_PAGESIZE_1GB
+and
+.BR HBW_PAGESIZE_1GB_STRICT
+options are not supported with
+.BR HBW_POLICY_INTERLEAVE
+policy.
+.PP
+.BR hbw_get_policy ()
+returns the current fallback policy when insufficient high bandwidth
+memory is available.
+.PP
+.BR hbw_set_policy ()
+sets the current fallback policy. The policy can be modified only once in the lifetime of an application and before calling
+.BR hbw_malloc (),
+.BR hbw_calloc (),
+.BR hbw_realloc (),
+.BR hbw_posix_memalign (),
+or
+.BR hbw_posix_memalign_psize ()
+function.
+.br
+.BR Note:
+If the policy is not set, than
+.B HBW_POLICY_PREFERRED
+will be used by default.
+.TP
+.B HBW_POLICY_BIND
+If insufficient high bandwidth memory from the nearest NUMA node is
+available to satisfy a request, the allocated pointer is set to
+.I NULL
+and
+.I errno
+is set to
+.BR ENOMEM .
+If insufficient high bandwidth memory pages are
+available at fault time the Out Of Memory (OOM) Killer is triggered.
+Note that pages are faulted exclusively from the high bandwidth NUMA
+node nearest at time of allocation, not at time of fault.
+.TP
+.B HBW_POLICY_BIND_ALL
+If insufficient high bandwidth memory is available to satisfy a request,
+the allocated pointer is set to
+.I NULL
+and
+.I errno
+is set to
+.BR ENOMEM .
+If insufficient high bandwidth memory pages are
+available at fault time the Out Of Memory (OOM) Killer is triggered.
+Note that pages are faulted from the high bandwidth NUMA nodes.
+Nearest NUMA node is selected at time of page fault.
+.TP
+.B HBW_POLICY_PREFERRED
+If insufficient memory is available from the high bandwidth NUMA node
+closest at allocation time, fall back to standard memory (default)
+with the smallest NUMA distance.
+.TP
+.B HBW_POLICY_INTERLEAVE
+Interleave faulted pages from across all high bandwidth NUMA nodes
+using standard size pages (the Transparent Huge Page feature is
+disabled).
+.PP
+.BR hbw_verify_memory_region ()
+verifies if memory region fully falls into high bandwidth memory. Returns
+0 if memory address range from
+.IR "addr"
+to
+.IR "addr"
++
+.IR "size"
+is allocated in high bandwidth memory,
+-1 if any fragment of memory was not backed by high bandwidth memory (e.g. when memory is not initialized)
+or one of error codes described in
+.B ERRORS
+section.
+
+Using this function in production code may result in serious performance penalty.
+
+The
+.IR Flags
+argument may include optional flags that modify function behavior:
+.TP
+.B HBW_TOUCH_PAGES
+Before checking pages, function will touch first byte of all pages in address range starting from
+.IR "addr"
+to
+.IR "addr"
++
+.IR "size"
+by read and write (so the content will be overwritten by the same data as it was read).
+Using this option may trigger Out Of Memory Killer.
+.SH "RETURN VALUE"
+.BR hbw_get_policy ()
+returns
+.BR HBW_POLICY_BIND,
+.BR HBW_POLICY_BIND_ALL,
+.BR HBW_POLICY_PREFERRED
+or
+.BR HBW_POLICY_INTERLEAVE
+which represents the current high bandwidth policy.
+.BR hbw_free ()
+do not have return value.
+.BR hbw_malloc ()
+.BR hbw_calloc ()
+and
+.BR hbw_realloc ()
+return the pointer to the allocated memory, or
+.I NULL
+if the request fails.
+.BR hbw_posix_memalign (),
+.BR hbw_posix_memalign_psize ()
+and
+.BR hbw_set_policy ()
+return zero on success and return an error code
+as described in the
+.B ERRORS
+section below on failure.
+.SH ERRORS
+.TP
+Error codes described here are the POSIX standard error codes as defined in
+.I <errno.h>
+.TP
+.BR hbw_check_available ()
+returns
+.BR ENODEV
+if high-bandwidth memory is unavailable.
+.TP
+.BR "hbw_posix_memalign" "() and " "hbw_posix_memalign_psize" "()"
+If the
+.I alignment
+parameter is not a power of two, or was not a multiple of
+.IR "sizeof(void*)" ,
+then
+.B EINVAL
+is returned.
+If the policy and
+.I pagesize
+combination is unsupported then
+.B EINVAL
+is returned.
+If there was insufficient memory to satisfy the request then
+.B ENOMEM
+is returned.
+.TP
+.BR hbw_set_policy ()
+returns
+.B EPERM
+if
+.BR hbw_set_policy ()
+was called more than once, or
+.B EINVAL
+if
+.I mode
+argument was neither
+.BR HBW_POLICY_PREFERRED ,
+.BR HBW_POLICY_BIND ,
+.BR HBW_POLICY_BIND_ALL
+nor
+.BR HBW_POLICY_INTERLEAVE .
+.TP
+.BR hbw_verify_memory_region ()
+returns
+.B EINVAL
+if
+.IR "addr"
+is
+.IR "NULL" ,
+.IR "size"
+equals 0 or
+.IR "flags"
+contained unsupported bit set. If memory pointed by
+.IR "addr"
+could not be verified then
+.B EFAULT
+is returned.
+.SH "NOTES"
+The
+.I <hbwmalloc.h>
+file defines the external functions and enumerations for the hbwmalloc
+library. These interfaces define a heap manager that targets high
+bandwidth memory numa nodes.
+.SH "FILES"
+.TP
+.I /usr/bin/memkind-hbw-nodes
+Prints a comma-separated list of high bandwidth nodes.
+.SH "ENVIRONMENT"
+.TP
+.B MEMKIND_HBW_NODES
+This environment variable is a comma-separated list of NUMA nodes that
+are treated as high bandwidth. Uses the
+.I libnuma
+routine
+.BR numa_parse_nodestring ()
+for parsing, so the syntax described in the
+.BR numa (3)
+man page for this routine applies for example: 1-3,5 is a valid setting.
+.TP
+.B MEMKIND_ARENA_NUM_PER_KIND
+This environment variable allows leveraging internal mechanism of
+the library for setting number of arenas per kind. Value should be
+a positive integer (not greater than
+.B INT_MAX
+defined in
+.IR <limits.h> ).
+The user should set the value based on the characteristics
+of application that is using the library. Higher value can
+provide better performance in extremely multithreaded applications at
+the cost of memory overhead. See section
+.BR "IMPLEMENTATION NOTES"
+of
+.BR jemalloc (3)
+for more details about arenas.
+.TP
+.B MEMKIND_HEAP_MANAGER
+Controls heap management behavior in memkind library by switching to one of the available heap managers.
+.br
+Values:
+.br
+    JEMALLOC - sets the jemalloc heap manager
+.br
+    TBB - sets the Intel Threading Building Blocks heap manager. This option requires installed
+    Intel Threading Building Blocks library.
+.PP
+.BR Note:
+If the
+.B MEMKIND_HEAP_MANAGER
+is not set then the jemalloc heap manager will be used by default.
+.SH "SYSTEM CONFIGURATION"
+Interfaces for obtaining 2MB (HUGETLB) memory need allocated
+huge pages in the kernel's huge page pool.
+.TP
+.B HUGETLB (huge pages)
+Current number of "persistent" huge pages can be read from
+.I /proc/sys/vm/nr_hugepages
+file.
+Proposed way of setting hugepages is:
+.BR "sudo sysctl vm.nr_hugepages=<number_of_hugepages>" .
+More information can be found here:
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.SH "KNOWN ISSUES"
+.TP
+.B HUGETLB (huge pages)
+There might be some overhead in huge pages consumption caused by heap management.
+If your allocation fails because of OOM, please try to allocate extra huge pages (e.g. 8 huge pages).
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR malloc (3),
+.BR numa (3),
+.BR numactl (8),
+.BR mbind (2),
+.BR mmap (2),
+.BR move_pages (2),
+.BR jemalloc (3),
+.BR memkind (3)
diff --git a/memkind-1.10.0/man/memkind-auto-dax-kmem-nodes.1 b/memkind-1.10.0/man/memkind-auto-dax-kmem-nodes.1
new file mode 100644
index 0000000..4ec49ae
--- /dev/null
+++ b/memkind-1.10.0/man/memkind-auto-dax-kmem-nodes.1
@@ -0,0 +1,56 @@
+.\"
+.\" Copyright (C) 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND-AUTO-DAX-KMEM-NODES" 1 "2019-08-29" "Intel Corporation" "MEMKIND-AUTO-DAX-KMEM-NODES" \" -*- nroff -*-
+.SH "NAME"
+memkind-auto-dax-kmem-nodes - prints comma-separated list of automatically detected persistent memory NUMA nodes.
+.br
+Note: This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes), or removed.
+.SH "SYNOPSIS"
+memkind-auto-dax-kmem-nodes [-h|--help]
+.SH "DESCRIPTION"
+.BR memkind-auto-dax-kmem-nodes
+prints a comma-separated list of automatically detected persistent memory NUMA nodes that can be used with the numactl --membind option.
+.SH "OPTIONS"
+.BR " -h, --help"
+              Displays help text and exit.
+.SH "EXIT STATUS"
+.BR 0
+on success
+.br
+.BR 1
+if automatic detection is disabled (memkind wasn't build with suitable libdaxctl-devel version)
+.br
+.BR 2
+if argument is invalid
+.br
+.BR 3
+on other failure
+.SH "COPYRIGHT"
+Copyright (C) 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR daxctl (3),
+.BR memkind (3),
+.BR numactl (8)
+.SH "AUTHOR"
+Michal Biesek <michal.biesek@intel.com>
diff --git a/memkind-1.10.0/man/memkind-hbw-nodes.1 b/memkind-1.10.0/man/memkind-hbw-nodes.1
new file mode 100644
index 0000000..1d4fd34
--- /dev/null
+++ b/memkind-1.10.0/man/memkind-hbw-nodes.1
@@ -0,0 +1,47 @@
+.\"
+.\" Copyright (C) 2018 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND-HBW_NODES" 1 "2018-03-12" "Intel Corporation" "MEMKIND-HBW_NODES" \" -*- nroff -*-
+.SH "NAME"
+memkind-hbw-nodes - print comma-separated list of high bandwidth nodes.
+.br
+Note: This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes), or removed.
+.SH "SYNOPSIS"
+memkind-hbw-nodes [-h|--help]
+.SH "DESCRIPTION"
+.BR memkind-hbw-nodes
+prints a comma-separated list of high bandwidth NUMA nodes that can be used with the numactl --membind option.
+.SH "OPTIONS"
+.BR " -h, --help"
+              Displays help text and exit.
+.SH "EXIT STATUS"
+.BR memkind-hbw-nodes
+exits with status 0 on success, 1 on failure, 2 on invalid argument.
+.SH "COPYRIGHT"
+Copyright (C) 2016 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR hbwmalloc (3),
+.BR memkind (3),
+.BR numactl (8)
+.SH "AUTHOR"
+Krzysztof Kulakowski <krzysztof.kulakowski@intel.com>
diff --git a/memkind-1.10.0/man/memkind.3 b/memkind-1.10.0/man/memkind.3
new file mode 100644
index 0000000..2658de3
--- /dev/null
+++ b/memkind-1.10.0/man/memkind.3
@@ -0,0 +1,1092 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND" 3 "2015-03-31" "Intel Corporation" "MEMKIND" \" -*- nroff -*-
+.SH "NAME"
+memkind \- Heap manager that enables allocations to memory with different properties.
+.br
+This header expose STANDARD and EXPERIMENTAL API. API Standards are described below in this man page.
+.SH "SYNOPSIS"
+.nf
+.B #include <memkind.h>
+.sp
+.B Link with -lmemkind
+.br
+.SS "EXPERIMENTAL API:"
+.sp
+.B "HEAP MANAGEMENT:"
+.br
+.BI "int memkind_posix_memalign(memkind_t " "kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.sp
+.B "KIND MANAGEMENT:"
+.br
+.BI "int memkind_create_kind(memkind_memtype_t " "memtype_flags" ", memkind_policy_t " "policy" ", memkind_bits_t " "flags" ", memkind_t " "*kind" );
+.sp
+.SS "STANDARD API:"
+.sp
+.B "ERROR HANDLING:"
+.br
+.BI "void memkind_error_message(int " "err" ", char " "*msg" ", size_t " "size" );
+.sp
+.B "LIBRARY VERSION:"
+.br
+.BI "int memkind_get_version();"
+.sp
+.B "HEAP MANAGEMENT:"
+.br
+.BI "void *memkind_malloc(memkind_t " "kind" ", size_t " "size" );
+.br
+.BI "void *memkind_calloc(memkind_t " "kind" ", size_t " "num" ", size_t " "size" );
+.br
+.BI "void *memkind_realloc(memkind_t " "kind" ", void " "*ptr" ", size_t " "size" );
+.br
+.BI "void memkind_free(memkind_t " "kind" ", void " "*ptr" );
+.br
+.BI "size_t memkind_malloc_usable_size(memkind_t " "kind" ", void " "*ptr" );
+.br
+.BI "void *memkind_defrag_reallocate(memkind_t " "kind" ", void " "*ptr" );
+.br
+.BI "memkind_t memkind_detect_kind(void " "*ptr" );
+.sp
+.B "KIND CONFIGURATION MANAGEMENT:"
+.br
+.BI "struct memkind_config *memkind_config_new();"
+.br
+.BI "void memkind_config_delete(struct memkind_config " "*cfg" );
+.br
+.BI "void memkind_config_set_path(struct memkind_config " "*cfg" ", const char " "*pmem_dir" );
+.br
+.BI "void memkind_config_set_size(struct memkind_config " "*cfg" ", size_t " "pmem_size" );
+.br
+.BI "void memkind_config_set_memory_usage_policy(struct memkind_config " "*cfg" ", memkind_mem_usage_policy " "policy" );
+.sp
+.B "KIND MANAGEMENT:"
+.br
+.BI "int memkind_create_pmem(const char " "*dir" ", size_t " "max_size" ", memkind_t " "*kind" );
+.br
+.BI "int memkind_create_pmem_with_config(struct memkind_config " "*cfg" ", memkind_t " "*kind" );
+.br
+.BI "int memkind_destroy_kind(memkind_t " "kind" );
+.br
+.BI "int memkind_check_available(memkind_t " "kind" );
+.sp
+.B "STATISTICS:"
+.br
+.BI "int memkind_update_cached_stats(void);"
+.br
+.BI "int memkind_get_stat(memkind_t " "kind" ", memkind_stat " "stat" ", size_t " "*value" );
+.sp
+.B "DECORATORS:"
+.br
+.BI "void memkind_malloc_pre(memkind_t " "*kind" ", size_t " "*size" );
+.br
+.BI "void memkind_malloc_post(memkind_t " "kind" ", size_t " "size" ", void " "**result" );
+.br
+.BI "void memkind_calloc_pre(memkind_t " "*kind" ", size_t " "*nmemb" ", size_t " "*size" );
+.br
+.BI "void memkind_calloc_post(memkind_t " "kind" ", size_t " "nmemb" ", size_t " "size" ", void " "**result" );
+.br
+.BI "void memkind_posix_memalign_pre(memkind_t " "*kind" ", void " "**memptr" ", size_t " "*alignment" ", size_t " "*size" );
+.br
+.BI "void memkind_posix_memalign_post(memkind_t " "kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" ", int " "*err" );
+.br
+.BI "void memkind_realloc_pre(memkind_t " "*kind" ", void " "**ptr" ", size_t " "*size" );
+.br
+.BI "void memkind_realloc_post(memkind_t " "*kind" ", void " "*ptr" ", size_t " "size" ", void " "**result" );
+.br
+.BI "void memkind_free_pre(memkind_t " "*kind" ", void " "**ptr" );
+.br
+.BI "void memkind_free_post(memkind_t " "kind" ", void " "*ptr" );
+.sp
+.sp
+.br
+.SH "DESCRIPTION"
+.PP
+.BR memkind_error_message ()
+converts an error number
+.I err
+returned by a member of the memkind
+interface to an error message
+.I msg
+where the maximum size of the message is passed by the
+.I size
+parameter.
+
+.B "HEAP MANAGEMENT:"
+.br
+The functions described in this section define a heap manager with an
+interface modeled on the ISO C standard API's, except that the user
+must specify the
+.I kind
+of memory with the first argument to each function. See the
+.B KINDS
+section below for a full description of the implemented kinds.
+For file-backed kind of memory see
+.BR memkind_create_pmem ()
+or
+.BR memkind_create_pmem_with_config ().
+.PP
+.BR memkind_malloc ()
+allocates
+.I size
+bytes of uninitialized memory of the specified
+.IR "kind" .
+The allocated space is suitably aligned (after possible pointer
+coercion) for storage of any type of object. If
+.I size
+is 0, then
+.BR memkind_malloc ()
+returns
+.IR "NULL" .
+.PP
+.BR memkind_calloc ()
+allocates space for
+.I num
+objects each
+.I size
+bytes in length in memory of the specified
+.IR "kind" .
+The result is identical to calling
+.BR memkind_malloc ()
+with an argument of
+.IR num
+*
+.IR "size" ,
+with the exception that the allocated memory is explicitly
+initialized to zero bytes.
+If
+.I num
+or
+.I size
+is 0, then
+.BR memkind_calloc ()
+returns
+.IR "NULL" .
+.PP
+.BR memkind_realloc ()
+changes the size of the previously allocated memory referenced by
+.I ptr
+to
+.I size
+bytes of the specified
+.IR "kind" .
+The contents of the memory remain unchanged up to the lesser of
+the new and old sizes. If the new size is larger, the contents of the
+newly allocated portion of the memory are undefined. Upon success, the
+memory referenced by
+.I ptr
+is freed and a pointer to the newly allocated memory is returned.
+
+.BR Note:
+.BR memkind_realloc ()
+may move the memory allocation, resulting in a different return value
+than
+.IR "ptr" .
+
+If
+.I ptr
+is
+.IR "NULL" ,
+the
+.BR memkind_realloc ()
+function behaves identically to
+.BR memkind_malloc ()
+for the specified size.
+If
+.I size
+is equal to zero, and
+.I ptr
+is not
+.IR "NULL" ,
+then the call is equivalent to
+.IR "memkind_free(kind, ptr)"
+and
+.I NULL
+is returned. The address
+.IR "ptr" ,
+if not
+.IR "NULL" ,
+must have been returned by a previous call to
+.BR memkind_malloc (),
+.BR memkind_calloc (),
+.BR memkind_realloc (),
+.BR memkind_defrag_reallocate ()
+or
+.BR memkind_posix_memalign ()
+with the same
+.I kind
+as specified to the call to
+.BR memkind_realloc ().
+Otherwise, if
+.I memkind_free(kind, ptr)
+was called before, undefined behavior occurs.
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_realloc ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_realloc (),
+but this will require an internal look up for correct kind.
+.BR Note:
+The look up for
+.I kind
+could result in serious performance penalty,
+which can be avoided by specifying a correct
+.IR kind .
+If
+.I kind
+is
+.I NULL
+and
+.I ptr
+is
+.IR "NULL" ,
+then
+.BR memkind_realloc ()
+returns
+.I NULL
+and sets
+.I errno
+to
+.BR EINVAL .
+.PP
+.BR memkind_posix_memalign ()
+allocates
+.I size
+bytes of memory of a specified
+.I kind
+such that the allocation's base address
+is an even multiple of
+.IR "alignment" ,
+and returns the allocation in the value pointed to by
+.IR "memptr" .
+The requested
+.I alignment
+must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+If
+.I size
+is 0, then
+.BR memkind_posix_memalign ()
+returns 0, with a
+.I NULL
+returned in
+.IR "memptr" .
+.PP
+.BR memkind_malloc_usable_size ()
+function provides the same semantics as
+.BR malloc_usable_size (3),
+but operates on specified
+.IR "kind" .
+.br
+.BR Note:
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_malloc_usable_size ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_malloc_usable_size (),
+but this could require a internal look up for correct kind.
+.BR memkind_malloc_usable_size ()
+is supported by TBB heap manager described in
+.B ENVIRONMENT
+section since Intel TBB 2019 Update 4.
+.PP
+.BR memkind_defrag_reallocate ()
+reallocates the object conditionally inside specific
+.IR "kind" .
+Function determines if it's worthwhile to move allocation to reduce degree of external fragmentation of the heap.
+In case of failure function returns
+.IR "NULL" ,
+otherwise function returns a pointer to reallocated memory and memory referenced by
+.I ptr
+was released and should not be accessed.
+If
+.I ptr
+is
+.IR "NULL" ,
+then
+.BR memkind_defrag_reallocate ()
+returns
+.IR "NULL" .
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_defrag_reallocate ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_defrag_reallocate (),
+but this will require an internal look up for correct kind.
+.BR Note:
+The look up for
+.I kind
+could result in serious performance penalty,
+which can be avoided by specifying a correct
+.IR kind .
+.PP
+.BR memkind_detect_kind ()
+returns the kind associated with allocated memory referenced by
+.IR ptr .
+This pointer
+must have been returned by a previous call to
+.BR memkind_malloc (),
+.BR memkind_calloc (),
+.BR memkind_realloc (),
+.BR memkind_defrag_reallocate ()
+or
+.BR memkind_posix_memalign ().
+If
+.I ptr
+is
+.IR "NULL" ,
+then
+.BR memkind_detect_kind ()
+returns
+.IR "NULL" .
+.BR Note:
+This function has non-trivial performance overhead.
+.PP
+.BR memkind_free ()
+causes the allocated memory referenced by
+.I ptr
+to be made available for future allocations. This pointer
+must have been returned by a previous call to
+.BR memkind_malloc (),
+.BR memkind_calloc (),
+.BR memkind_realloc (),
+.BR memkind_defrag_reallocate ()
+or
+.BR memkind_posix_memalign ().
+Otherwise, if
+.I memkind_free(kind, ptr)
+was already called before, undefined behavior occurs.
+If
+.I ptr
+is
+.IR "NULL" ,
+no operation is performed.
+In cases where the kind is unknown in the
+context of the call to
+.BR memkind_free ()
+.I NULL
+can be given as the
+.I kind
+specified to
+.BR memkind_free (),
+but this will require an internal look up for correct kind.
+.BR Note:
+The look up for
+.I kind
+could result in serious performance penalty,
+which can be avoided by specifying a correct
+.IR kind .
+.sp
+.B "KIND CONFIGURATION MANAGEMENT:"
+.br
+The functions described in this section define a way to create, delete and update kind specific configuration.
+Except of
+.BR memkind_config_new (),
+user must specify the memkind configuration with the first argument to each function.
+API described here is most useful with file-backed kind of memory, e.g.
+.BR memkind_create_pmem_with_config ()
+method.
+.PP
+.BR memkind_config_new ()
+creates the memkind configuration.
+.PP
+.BR memkind_config_delete ()
+deletes previously created memkind configuration, which must have been returned by a previous call to
+.BR memkind_config_new ().
+.PP
+.BR memkind_config_set_path ()
+updates the memkind
+.IR pmem_dir
+configuration parameter, which specifies directory path, where file-backed kind of memory will be created.
+.BR Note:
+This function does not validate that
+.I pmem_dir
+specifies a valid path.
+.PP
+.BR memkind_config_set_size ()
+updates the memkind
+.IR pmem_size
+configuration parameter, which allows to limit the file-backed kind memory partition.
+.BR Note:
+This function does not validate that
+.I pmem_size
+is in valid range.
+.PP
+.BR memkind_config_set_memory_usage_policy ()
+updates the memkind
+.IR policy
+configuration parameter, which allows to tune up memory utilization.
+The user should set the value based on the characteristics of application that is using
+the library (e.g. prioritize memory usage, CPU utilization), for more details about
+.IR policy
+see the
+.BR "MEMORY USAGE POLICY"
+section below.
+.BR Note:
+This function does not validate that
+.I policy
+is in valid range.
+.sp
+.B "KIND MANAGEMENT:"
+.br
+There are built-in kinds that are always available and these are enumerated in the
+.B KINDS
+section. The user can also create their own kinds of memory. This
+section describes the API's that enable the tracking of the different
+kinds of memory and determining their properties.
+.PP
+.BR memkind_create_pmem ()
+is a convenient function used to create a file-backed kind of memory.
+It allocates a temporary file in the given directory
+.IR dir .
+The file is created in a fashion similar to
+.BR tmpfile (3),
+so that the file name does not appear when the directory is listed and
+the space is automatically freed when the program terminates.
+The file is truncated to a size of
+.I max_size
+bytes and the resulting space is memory-mapped.
+.br
+Note that the actual file system space is not allocated immediately, but only
+on a call to
+.BR memkind_pmem_mmap ()
+(see
+.BR memkind_pmem (3)).
+This allows to create a pmem memkind of a pretty large size without the
+need to reserve in advance the corresponding file system space for the entire
+heap. If the value of
+.I max_size
+equals 0, pmem memkind is only limited by the capacity of the file system mounted under
+.I dir
+argument.
+The minimum
+.I max_size
+value which allows to limit the size of kind by the library is defined as
+.BR MEMKIND_PMEM_MIN_SIZE .
+Calling
+.BR memkind_create_pmem ()
+with a size smaller than that and different than 0 will return an error.
+The maximum allowed size is not limited by
+.BR memkind ,
+but by the file system specified by the
+.I dir
+argument.
+The
+.I max_size
+passed in is the raw size of the memory pool and
+.B jemalloc
+will use some of that space for its own metadata.
+Returns zero if the pmem memkind is created successfully or an error code from the
+.B ERRORS
+section if not.
+.PP
+.BR memkind_create_pmem_with_config ()
+is a second function used to create a file-backed kind of memory.
+Function behaves simillar to
+.BR memkind_create_pmem ()
+but instead of passing
+.I dir
+and
+.I max_size
+arguments,
+it uses
+.I config
+param to specify characteristics of created file-backed kind of memory (see
+.B KIND CONFIGURATION MANAGEMENT
+section).
+.PP
+.BR memkind_create_kind ()
+creates kind that allocates memory with specific memory type, memory binding policy and flags (see
+.B MEMORY FLAGS
+section).
+The
+.IR memtype_flags
+(see
+.B MEMORY TYPES
+section) determine memory types to allocate,
+.IR policy
+argument is policy for specifying page binding to memory types selected by
+.IR memtype_flags .
+Returns zero if the specified kind is created successfully or an error code from the
+.B ERRORS
+section if not.
+.PP
+.BR memkind_destroy_kind ()
+destroys previously created kind object, which must have been returned by a previous call to
+.BR memkind_create_pmem (),
+.BR memkind_create_pmem_with_config ()
+or
+.BR memkind_create_kind ().
+Otherwise, or if
+.I memkind_destroy_kind(kind)
+was already called before, undefined behavior occurs.
+Note that, when the kind was returned by
+.BR memkind_create_kind ()
+all allocated memory must be freed before kind is destroyed,
+otherwise this will cause memory leak. When the kind was returned by
+.BR memkind_create_pmem ()
+or
+.BR memkind_create_pmem_with_config ()
+all allocated memory will be freed after kind will be destroyed.
+.PP
+.BR memkind_check_available ()
+returns zero if the specified
+.I kind
+is available or an error code from the
+.B ERRORS
+section if it is not.
+.PP
+.BR MEMKIND_PMEM_MIN_SIZE
+The minimum size which allows to limit the file-backed memory partition.
+.sp
+.B "STATISTICS:"
+.br
+The functions described in this section define a way to get specific memory allocation statistics.
+.PP
+.BR memkind_update_cached_stats ()
+is used to force an update of cached dynamic allocator statistics.
+Statistics are not updated real-time by memkind library and this method allows to force its update.
+.PP
+.BR memkind_get_stat ()
+retrieves statistic of the specified type
+and returns it in
+.IR "value".
+For more details about
+.I stat
+see the
+.B "MEMORY STATISTICS TYPE"
+section below.
+Measured statistic applies
+to specific
+.IR "kind",
+when
+.I NULL
+is given as
+.I kind
+then statistic applies to memory used by the whole memkind library.
+.BR Note:
+You need to call
+.BR memkind_update_cached_stats ()
+before calling
+.BR memkind_get_stat ()
+because statistics are cached by memkind library.
+.sp
+.B "DECORATORS:"
+.br
+The memkind library enables the user to define decorator functions that
+can be called before and after each memkind heap management API. The
+decorators that are called at the beginning of the function end are named
+after that function with
+.I _pre
+appended to the name and those that are called at the end of the
+function are named after that function with
+.I _post
+appended to the name. These are weak symbols and if they are not
+present at link time they are not called. The memkind library does
+not define these symbols which are reserved for user definition.
+These decorators can be used to track calls to the heap management
+interface or to modify parameters. The decorators that are called at
+the beginning of the allocator pass all inputs by reference and the
+decorators that are called at the end of the allocator pass the output
+by reference. This enables the modification of the input and output
+of each heap management function by the decorators.
+.sp
+.B "LIBRARY VERSION:"
+.br
+The memkind library version scheme consist major, minor and patch numbers separated by dot. Combining those numbers, we got the following representation:
+.br
+major.minor.patch, where:
+.br
+	-major number is incremented whenever API is changed (loss of backward compatibility),
+.br
+	-minor number is incremented whenever additional extensions are introduced or behavior has been changed,
+.br
+	-patch number is incremented whenever small bug fixes are added.
+.sp
+memkind library provide numeric representation of the version by exposing the following API:
+.PP
+.BR memkind_get_version ()
+returns version number represented by a single integer number, obtained from the formula:
+.br
+major * 1000000 + minor * 1000 + patch
+.sp
+.BR Note:
+major < 1 means unstable API.
+.sp
+API standards:
+.br
+-STANDARD API, API is considered as stable
+.br
+-NON-STANDARD API, API is considered as stable, however this is not a standard way to use memkind
+.br
+-EXPERIMENTAL API, API is considered as unstable and the subject to change
+.br
+.sp
+.SH "RETURN VALUE"
+.BR memkind_calloc (),
+.BR memkind_malloc (),
+.BR memkind_realloc ()
+and
+.BR memkind_defrag_reallocate ()
+returns the pointer to the allocated memory or
+.I NULL
+if the request fails.
+.BR memkind_malloc_usable_size ()
+returns the number of usable bytes in the block of allocated memory pointed to by
+.IR "ptr" ,
+a pointer to a block of memory allocated by
+.BR memkind_malloc ()
+or a related function. If
+.I ptr
+is
+.IR "NULL" ,
+0 is returned.
+.BR memkind_free ()
+and
+.BR memkind_error_message ()
+do not have return values.
+All other memkind API's return 0 upon
+success and an error code defined in the
+.B ERRORS
+section upon failure.
+The memkind library avoids setting
+.I errno
+directly, but calls to underlying libraries and system calls may set
+.IR errno
+(e.g.
+.BR memkind_create_pmem ()).
+.SH "KINDS"
+The available kinds of memory:
+.TP
+.B MEMKIND_DEFAULT
+Default allocation using standard memory and default page size.
+.TP
+.B MEMKIND_HUGETLB
+Allocate from standard memory using huge pages.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_GBTLB (DEPRECATED)
+Allocate from standard memory using 1GB chunks backed by huge pages.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_INTERLEAVE
+Allocate pages interleaved across all NUMA nodes with transparent huge
+pages disabled.
+.TP
+.B MEMKIND_HBW
+Allocate from the closest high bandwidth memory NUMA node at the time
+of allocation. If there is not enough high bandwidth memory to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_HBW_ALL
+Same as
+.B MEMKIND_HBW
+except decision regarding closest NUMA node is postponed until the time of first write.
+.TP
+.B MEMKIND_HBW_HUGETLB
+Same as
+.B MEMKIND_HBW
+except the allocation is backed by huge pages.
+.BR Note:
+This kind requires
+huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_ALL_HUGETLB
+Combination of
+.B MEMKIND_HBW_ALL
+and
+.B MEMKIND_HBW_HUGETLB
+properties.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_PREFERRED
+Same as
+.B MEMKIND_HBW
+except that if there is not enough high bandwidth memory to satisfy
+the request, the allocation will fall back on standard memory.
+.TP
+.B MEMKIND_HBW_PREFERRED_HUGETLB
+Same as
+.B MEMKIND_HBW_PREFERRED
+except the allocation is backed by huge pages.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_GBTLB (DEPRECATED)
+Same as
+.B MEMKIND_HBW
+except the allocation is backed by 1GB chunks of huge pages. Note that
+.I size
+can take on any value, but full gigabyte pages will allocated for each
+request, so remainder of the last page will be wasted.
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_PREFERRED_GBTLB (DEPRECATED)
+Same as
+.B MEMKIND_HBW_GBTLB
+except that if there is not enough high bandwidth memory to satisfy
+the request, the allocation will fall back on standard memory.
+.BR Note:
+This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.TP
+.B MEMKIND_HBW_INTERLEAVE
+Same as
+.B MEMKIND_HBW
+except that the pages that support the allocation are interleaved
+across all high bandwidth nodes and transparent huge pages are
+disabled.
+.TP
+.B MEMKIND_DAX_KMEM
+Allocate from the closest persistent memory NUMA node at the time
+of allocation. If there is not enough memory in the closest persistent memory NUMA node to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_DAX_KMEM_ALL
+Allocate from the closest persistent memory NUMA node available at the time
+of allocation. If there is not enough memory on any of persistent memory NUMA nodes to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_DAX_KMEM_PREFERRED
+Same as
+.B MEMKIND_DAX_KMEM
+except that if there is not enough memory in the closest persistent memory NUMA node to satisfy
+the request, the allocation will fall back on other memory NUMA nodes.
+.BR Note:
+For this kind, the allocation will not succeed if two or more
+persistent memory NUMA nodes are in the same shortest distance to the same CPU on which process is eligible to run.
+Check on that eligibility is done upon starting the application.
+.TP
+.B MEMKIND_REGULAR
+Allocate from regular memory using the default page size. Regular means general purpose memory
+from the NUMA nodes containing CPUs.
+.SH "MEMORY TYPES"
+The available types of memory:
+.TP
+.B MEMKIND_MEMTYPE_DEFAULT
+Standard memory, the same as process uses.
+.TP
+.B MEMKIND_MEMTYPE_HIGH_BANDWIDTH
+High bandwidth memory (HBM). There must be at least two memory types with different bandwidth to determine which is the HBM.
+.SH "MEMORY BINDING POLICY"
+The available types of memory binding policy:
+.TP
+.B MEMKIND_POLICY_BIND_LOCAL
+Allocate local memory. If there is not enough memory to satisfy the request
+.I errno
+is set to
+.BR ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_POLICY_BIND_ALL
+Memory locality is ignored. If there is not enough memory to satisfy the request
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.TP
+.B MEMKIND_POLICY_PREFERRED_LOCAL
+Allocate preferred memory that is local.
+If there is not enough preferred memory to satisfy the request or
+preferred memory is not available, the allocation will fall back on any other memory.
+.TP
+.B MEMKIND_POLICY_INTERLEAVE_LOCAL
+Interleave allocation across local memory.
+For n memory types the allocation will be interleaved across all of them.
+.TP
+.B MEMKIND_POLICY_INTERLEAVE_ALL
+Interleave allocation. Locality is ignored.
+For n memory types the allocation will be interleaved across all of them.
+.TP
+.B MEMKIND_POLICY_MAX_VALUE
+Max policy value.
+.SH "MEMORY FLAGS"
+The available types of memory flags:
+.TP
+.B MEMKIND_MASK_PAGE_SIZE_2MB
+Allocation backed by 2MB page size.
+.SH "MEMORY USAGE POLICY"
+The available types of memory usage policy:
+.TP
+.B MEMKIND_MEM_USAGE_POLICY_DEFAULT
+Default memory usage policy.
+.TP
+.B MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE
+Conservative memory usage policy - prioritize memory usage at cost of performance.
+.BR Note:
+Memory usage policies have no effect for TBB heap manager described in
+.B ENVIRONMENT
+section.
+.SH "MEMORY STATISTICS TYPE"
+The available types of memory statistics:
+.TP
+.B MEMKIND_STAT_TYPE_RESIDENT
+Maximum number of bytes in physically resident data pages mapped.
+.TP
+.B MEMKIND_STAT_TYPE_ACTIVE
+Total number of bytes in active pages.
+.TP
+.B MEMKIND_STAT_TYPE_ALLOCATED
+Total number of allocated bytes.
+.SH "ERRORS"
+.TP
+.BR memkind_posix_memalign ()
+returns the one of the POSIX standard error codes
+.B EINVAL
+or
+.B ENOMEM
+as defined in
+.I <errno.h>
+if an error occurs (these have positive values).
+If the
+.I alignment
+parameter is not a power of two or is not a multiple of
+.IR "sizeof(void*)" ,
+then
+.B EINVAL
+is returned. If there is insufficient memory to satisfy the request then
+.B ENOMEM
+is returned.
+.PP
+All functions other than
+.BR memkind_posix_memalign ()
+which have an integer return type return one of the negative error
+codes as defined in
+.I <memkind.h>
+and described below.
+.TP
+.B MEMKIND_ERROR_UNAVAILABLE
+Requested memory kind is not available
+.TP
+.B MEMKIND_ERROR_MBIND
+Call to
+.BR mbind (2)
+failed
+.TP
+.B MEMKIND_ERROR_MMAP
+Call to
+.BR mmap (2)
+failed
+.TP
+.B MEMKIND_ERROR_MALLOC
+Call to jemalloc's
+.BR malloc ()
+failed
+.TP
+.B MEMKIND_ERROR_ENVIRON
+Error parsing environment variable
+.I MEMKIND_*
+.TP
+.B MEMKIND_ERROR_INVALID
+Invalid input arguments to memkind routine
+.TP
+.B MEMKIND_ERROR_TOOMANY
+Error trying to initialize more than maximum
+.B MEMKIND_MAX_KIND
+number of kinds
+.TP
+.B MEMKIND_ERROR_BADOPS
+Error memkind operation structure is missing or invalid
+.TP
+.B MEMKIND_ERROR_HUGETLB
+Unable to allocate huge pages
+.TP
+.B MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE
+Error requested memory type is not available
+.TP
+.B MEMKIND_ERROR_OPERATION_FAILED
+Error memkind operation failed
+.TP
+.B MEMKIND_ERROR_ARENAS_CREATE
+Call to jemalloc's
+.BR arenas.create ()
+failed
+.TP
+.B MEMKIND_ERROR_RUNTIME
+Unspecified run-time error
+.SH "FILES"
+.TP
+.I /usr/bin/memkind-hbw-nodes
+Prints a comma-separated list of high bandwidth nodes.
+.TP
+.I /usr/bin/memkind-auto-dax-kmem-nodes
+Prints a comma-separated list of persistent memory NUMA nodes, which are automatically detected.
+.SH "ENVIRONMENT"
+.TP
+.B MEMKIND_HBW_NODES
+This environment variable is a comma-separated list of NUMA nodes that
+are treated as high bandwidth. Uses the
+.I libnuma
+routine
+.BR numa_parse_nodestring ()
+for parsing, so the syntax described in the
+.BR numa (3)
+man page for this routine applies: e.g. 1-3,5 is a valid setting.
+.TP
+.B MEMKIND_DAX_KMEM_NODES
+This environment variable is a comma-separated list of NUMA nodes that
+are treated as PMEM memory. Uses the
+.I libnuma
+routine
+.BR numa_parse_nodestring ()
+for parsing, so the syntax described in the
+.BR numa (3)
+man page for this routine applies: e.g. 1-3,5 is a valid setting.
+.TP
+.B MEMKIND_ARENA_NUM_PER_KIND
+This environment variable allows leveraging internal mechanism of
+the library for setting number of arenas per kind. Value should be
+a positive integer (not greater than
+.B INT_MAX
+defined in
+.IR <limits.h> ).
+The user should set the value based on the characteristics
+of application that is using the library. Higher value can
+provide better performance in extremely multithreaded applications at
+the cost of memory overhead. See section
+.BR "IMPLEMENTATION NOTES"
+of
+.BR jemalloc (3)
+for more details about arenas.
+.TP
+.B MEMKIND_HOG_MEMORY
+Controls behavior of memkind with regards to returning memory to underlying OS. Setting
+.B MEMKIND_HOG_MEMORY
+to 1 causes memkind to not release memory to OS in anticipation of memory reuse soon. This will
+improve latency of 'free' operations but increase memory usage.
+.TP
+.B MEMKIND_DEBUG
+Controls logging mechanism in memkind. Setting
+.B MEMKIND_DEBUG
+to 1 enables printing messages like errors and general information about environment to
+.IR stderr .
+.TP
+.B MEMKIND_BACKGROUND_THREAD_LIMIT
+Enable background worker threads.
+Value should be from range 0 to maximum number of cpus.
+Setting
+.B MEMKIND_BACKGROUND_THREAD_LIMIT
+to specific value will limit maximum number of background worker threads to this value.
+0 means maximum number of background worker threads will be limited to maximum number of cpus.
+.TP
+.B MEMKIND_HEAP_MANAGER
+Controls heap management behavior in memkind library by switching to one of the available heap managers.
+.br
+Values:
+.br
+    JEMALLOC - sets the jemalloc heap manager
+.br
+    TBB - sets the Intel Threading Building Blocks heap manager. This option requires installed
+    Intel Threading Building Blocks library.
+.PP
+If the
+.B MEMKIND_HEAP_MANAGER
+is not set then the jemalloc heap manager will be used by default.
+.SH "SYSTEM CONFIGURATION"
+Interfaces for obtaining 2MB (HUGETLB) memory need allocated
+huge pages in the kernel's huge page pool.
+.TP
+.B HUGETLB (huge pages)
+Current number of "persistent" huge pages can be read from
+.I /proc/sys/vm/nr_hugepages
+file.
+Proposed way of setting hugepages is:
+.BR "sudo sysctl vm.nr_hugepages=<number_of_hugepages>" .
+More information can be found here:
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.SH "STATIC LINKING"
+When linking statically against memkind,
+.I libmemkind.a
+should be used together with its dependencies
+.I libnuma
+and pthread. Pthread can be linked by adding
+.I /usr/lib64/libpthread.a
+as a dependency (exact path may vary). Typically
+.I libnuma
+will need to be compiled from sources to use it as a static dependency.
+.I libnuma
+can be reached on GitHub:
+.UR https://github.com/numactl/numactl
+.UE
+.SH "KNOWN ISSUES"
+.TP
+.B HUGETLB (huge pages)
+There might be some overhead in huge pages consumption caused by heap management.
+If your allocation fails because of OOM, please try to allocate extra huge pages (e.g. 8 huge pages).
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR malloc (3),
+.BR malloc_usable_size (3),
+.BR numa (3),
+.BR numactl (8),
+.BR mbind (2),
+.BR mmap (2),
+.BR move_pages (2),
+.BR jemalloc (3),
+.BR memkind_dax_kmem (3),
+.BR memkind_default (3),
+.BR memkind_arena (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3)
diff --git a/memkind-1.10.0/man/memkind_arena.3 b/memkind-1.10.0/man/memkind_arena.3
new file mode 100644
index 0000000..45a4d1a
--- /dev/null
+++ b/memkind-1.10.0/man/memkind_arena.3
@@ -0,0 +1,222 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_ARENA" 3 "2015-04-21" "Intel Corporation" "MEMKIND_ARENA" \" -*- nroff -*-
+.SH "NAME"
+memkind_arena \- jemalloc arena allocation memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_arena_create(struct memkind " "*kind" ", struct memkind_ops " "*ops" ", const char " "*name" );
+.BI "int memkind_arena_create_map(struct memkind " "*kind" ", extent_hooks_t " "*hooks" );
+.BI "int memkind_arena_destroy(struct memkind " "*kind" );
+.BI "void *memkind_arena_malloc(struct memkind " "*kind" ", size_t " "size" );
+.BI "void *memkind_arena_calloc(struct memkind " "*kind" ", size_t " "num" ", size_t " "size" );
+.BI "int memkind_arena_posix_memalign(struct memkind " "*kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.BI "void *memkind_arena_realloc(struct memkind " "*kind" ", void " "*ptr" ", size_t " "size" );
+.BI "void *memkind_arena_realloc_with_kind_detect(void " "*ptr" ", size_t " "size" );
+.BI "int memkind_thread_get_arena(struct memkind " "*kind" ", unsigned int " "*arena" ", size_t " "size" );
+.BI "int memkind_bijective_get_arena(struct memkind " "*kind" ", unsigned int " "*arena" ", size_t " "size" );
+.BI "struct memkind *get_kind_by_arena(unsigned " "arena_ind" );
+.BI "struct memkind *memkind_arena_detect_kind(void " "*ptr" );
+.BI "int memkind_arena_finalize(struct memkind " "*kind" );
+.BI "void memkind_arena_init(struct memkind " "*kind" );
+.BI "void memkind_arena_free(struct memkind " "*kind" ", void " "*ptr" );
+.BI "void memkind_arena_free_with_kind_detect(void " "*ptr" );
+.BI "size_t memkind_arena_malloc_usable_size(void " "*ptr" );
+.BI "int memkind_arena_update_memory_usage_policy(struct memkind " "*kind" ", memkind_mem_usage_policy " "policy" );
+.BI "int memkind_arena_enable_background_threads(size_t threads_limit);"
+.br
+.SH DESCRIPTION
+This header file is a collection of functions can be used to populate
+the memkind operations structure for memory kinds that use jemalloc.
+.PP
+.BR memkind_arena_create ()
+is an implementation of the memkind "create" operation for memory
+kinds that use jemalloc.  This calls
+.BR memkind_default_create ()
+(see
+.BR memkind_default (3))
+followed by
+.BR memkind_arena_create_map ()
+described below.
+.PP
+.BR memkind_arena_create_map ()
+creates the
+.I arena_map
+array for the memkind structure pointed to by
+.IR kind
+which can be indexed by the
+.BR ops.get_arena ()
+function from the kind's operations.  If get_arena points
+.BR memkind_thread_get_arena ()
+then there will be four arenas created for each processor,
+and if get_arena points to
+.BR memkind_bijective_get_arena ()
+then just one arena is created.
+.PP
+.BR memkind_arena_destroy ()
+is an implementation of the memkind "destroy" operation for memory
+kinds that use jemalloc.  This releases all of the resources
+allocated by
+.BR memkind_arena_create ().
+.PP
+.BR memkind_arena_malloc ()
+is an implementation of the memkind "malloc" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.
+.PP
+.BR memkind_arena_calloc ()
+is an implementation of the memkind "calloc" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.
+.PP
+.BR memkind_arena_posix_memalign ()
+is an implementation of the memkind "posix_memalign" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.  The POSIX standard requires that
+.BR posix_memalign (3)
+may not set
+.I errno
+however the jemalloc's
+.BR mallocx ()
+routine may.  In an attempt to abide by the standard
+.I errno
+is recorded before calling jemalloc's
+.BR mallocx ()
+and then reset after the call.
+.PP
+.BR memkind_arena_realloc ()
+is an implementation of the memkind "realloc" operation for memory
+kinds that use jemalloc.  This allocates memory using the arenas
+created by
+.BR memkind_arena_create ()
+through the jemalloc's
+.BR mallocx ()
+interface.  It uses the memkind "get_arena" operation to select the
+arena.
+.PP
+.BR memkind_arena_realloc_with_kind_detect ()
+function will look up for kind associated to the allocated memory referenced by
+.I ptr
+and call (depending on kind value)
+.BR memkind_arena_realloc ()
+or
+.BR memkind_default_realloc ().
+.PP
+.BR memkind_thread_get_arena ()
+retrieves the
+.I arena
+index that is bound to to the calling thread based on a hash of its
+thread ID.  The
+.I arena
+index can be used with the MALLOCX_ARENA macro to set flags for jemalloc's
+.BR mallocx ().
+.PP
+.BR memkind_bijective_arena_get_arena ()
+retrieves the
+.I arena
+index to be used with the MALLOCX_ARENA macro to set flags for jemalloc's
+.BR mallocx ().
+Use of this operation implies that only one arena is used for the
+.IR kind .
+.PP
+.BR memkind_arena_free ()
+is an implementation of the memkind "free" operation for memory
+kinds that use jemalloc.  It causes the allocated memory referenced by
+.IR ptr ,
+which must have been returned by a previous call to
+.BR memkind_arena_malloc (),
+.BR memkind_arena_calloc ()
+or
+.BR memkind_arena_realloc ()
+to be made available for future allocations.
+It uses the memkind "get_arena" operation to select the arena.
+.PP
+.BR memkind_arena_free_with_kind_detect ()
+function will look up for kind associated to the allocated memory referenced by
+.I ptr
+and call
+.BR memkind_arena_free ().
+.PP
+.BR memkind_arena_detect_kind ()
+returns pointer to memory kind structure associated with given allocated memory referenced by
+.IR ptr .
+.PP
+.BR get_kind_by_arena ()
+returns pointer to memory kind structure associated with given arena.
+.PP
+.BR memkind_arena_malloc_usable_size ()
+is an implementation of the memkind "usable_size" operation for memory kinds that use jemalloc.
+.PP
+.BR memkind_arena_finalize ()
+is an implementation of the memkind "finalize" operation for memory kinds that
+use jemalloc. This function releases all resources allocated by
+.BR memkind_arena_create ()
+and it's called when
+.BR main ()
+finishes or after calling
+.BR exit ()
+function.
+.PP
+.BR memkind_arena_init ()
+creates arena map with proper hooks per specified kind.
+.PP
+.BR memkind_arena_update_memory_usage_policy ()
+function changes time, which determine how fast jemalloc returns unused pages back to
+the operating system, in other words how fast it deallocates file space.
+.PP
+.BR memkind_arena_enable_background_threads ()
+enables internal background worker threads in jemalloc.
+The
+.I threads_limit
+specify limit of background threads which will be enabled ( 0 means no limit).
+.PP
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/memkind-1.10.0/man/memkind_dax_kmem.3 b/memkind-1.10.0/man/memkind_dax_kmem.3
new file mode 100644
index 0000000..99300a9
--- /dev/null
+++ b/memkind-1.10.0/man/memkind_dax_kmem.3
@@ -0,0 +1,60 @@
+.\"
+.\" Copyright (C) 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_DAX_KMEM" 3 "2019-07-19" "Intel Corporation" "MEMKIND_DAX_KMEM" \" -*- nroff -*-
+.SH "NAME"
+memkind_dax_kmem \- DAX KMEM memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_dax_kmem_all_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.SH DESCRIPTION
+.PP
+DAX KMEM memory memkind operations.
+.PP
+.BR memkind_dax_kmem_all_get_mbind_nodemask ()
+sets the
+.I nodemask
+bits to one that correspond to the all persistent memory NUMA nodes in
+the system. All other bits up to
+.I maxnode
+are set to zero.
+The
+.I nodemask
+can be used in conjunction with the
+.BR mbind (2)
+system call.
+.SH "COPYRIGHT"
+Copyright (C) 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/memkind-1.10.0/man/memkind_default.3 b/memkind-1.10.0/man/memkind_default.3
new file mode 100644
index 0000000..3880f18
--- /dev/null
+++ b/memkind-1.10.0/man/memkind_default.3
@@ -0,0 +1,226 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_DEFAULT" 3 "2015-04-21" "Intel Corporation" "MEMKIND_DEFAULT" \" -*- nroff -*-
+.SH "NAME"
+memkind_default \- default implementations for memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes), or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_default_create(struct memkind " "*kind" ", struct memkind_ops " "*ops" ", const char " "*name" );
+.br
+.BI "int memkind_default_destroy(struct memkind " "*kind" );
+.br
+.BI "void *memkind_default_malloc(struct memkind " "*kind" ", size_t " "size" );
+.br
+.BI "void *memkind_default_calloc(struct memkind " "*kind" ", size_t " "num" ", size_t " "size" );
+.br
+.BI "int memkind_default_posix_memalign(struct memkind " "*kind" ", void " "**memptr" ", size_t " "alignment" ", size_t " "size" );
+.br
+.BI "void *memkind_default_realloc(struct memkind " "*kind" ", void " "*ptr" ", size_t " "size" );
+.br
+.BI "void memkind_default_free(struct memkind " "*kind" ", void " "*ptr" );
+.br
+.BI "void *memkind_default_mmap(struct memkind " "*kind" ", void " "*addr" ", size_t " "size" );
+.br
+.BI "int memkind_default_mbind(struct memkind " "*kind" ", void " "*ptr" ", size_t " "len" );
+.br
+.BI "int memkind_default_get_mmap_flags(struct memkind " "*kind" ", int " "*flags" );
+.br
+.BI "int memkind_default_get_mbind_mode(struct memkind " "*kind" ", int " "*mode" );
+.br
+.BI "size_t memkind_default_malloc_usable_size(struct memkind " "*kind" ", void " "*ptr" );
+.br
+.BI "int memkind_preferred_get_mbind_mode(struct memkind " "*kind" ", int " "*mode" );
+.br
+.BI "int memkind_interleave_get_mbind_mode(struct memkind " "*kind" ", int " "*mode" );
+.br
+.BI "int memkind_nohugepage_madvise(struct memkind " "*kind" ", void " "*addr" ", size_t " "size" );
+.br
+.BI "int memkind_posix_check_alignment(struct memkind " "*kind" ", size_t " "alignment" );
+.br
+.BI "int memkind_default_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.BI "void memkind_default_init_once(void);"
+.br
+.BI "bool size_out_of_bounds(size_t " "size" );
+.br
+.SH DESCRIPTION
+.PP
+Default implementations for memkind operations which include a several
+useful methods that are not part of the
+.B MEMKIND_DEFAULT
+kind which is a fall through to the jemalloc implementation.
+.PP
+.BR memkind_default_create ()
+implements the required start up for every kind. If a kind does not
+point to this function directly for its
+.BR ops.create ()
+operation, then the function that it points to must call
+.BR memkind_default_create ()
+at its start.
+.PP
+.BR memkind_default_destroy ()
+implements the required shutdown for every kind. If a kind does not
+point to this function directly for its
+.BR ops.destroy ()
+operation, then the function that it points to must call
+.BR memkind_default_destroy ()
+at its end.
+.PP
+.BR memkind_default_malloc ()
+is a direct call through the jemalloc's
+.BR malloc ().
+.PP
+.BR memkind_default_calloc ()
+is a direct call through the jemalloc's
+.BR calloc ().
+.PP
+.BR memkind_default_posix_memalign ()
+is a direct call through the jemalloc's
+.BR posix_memalign ().
+.PP
+.BR memkind_default_realloc ()
+is a direct call through the jemalloc's
+.BR realloc ().
+.PP
+.BR memkind_default_free ()
+is a direct call through the jemalloc's
+.BR free ().
+Note that this method can be called on any pointer returned by a
+jemalloc allocation, and in particular, all of the arena
+allocations described in
+.BR memkind_arena (3)
+can use this function for freeing.
+.PP
+.BR memkind_default_mmap ()
+This calls the ops->get_mmap_flags()
+operations for the kind, or falls back on the default implementations
+if the function pointers are
+.IR "NULL" .
+The results of these calls are
+passed to the
+.BR mmap (2)
+call to allocate pages from the operating system.  The
+.I addr
+is the hint passed through to
+.BR mmap (2)
+and
+.I size
+is the size of the buffer to be allocated.  The return value is the
+allocated buffer or
+.B MAP_FAILED
+in the case of an error.
+.PP
+.BR memkind_default_mbind ()
+makes calls the kind's
+.BR ops.get_mbind_nodemask ()
+and
+.BR ops.get_mbind_mode ()
+operations to gather inputs and then calls the
+.BR mbind (2)
+system call using the results along with and user input
+.I ptr
+and
+.IR len .
+.PP
+.BR memkind_default_get_mmap_flags ()
+sets
+.I flags
+to
+.BR "MAP_PRIVATE | MAP_ANONYMOUS" .
+See
+.BR mmap (2)
+for more information about these flags.
+.PP
+.BR memkind_default_get_mbind_mode ()
+sets
+.I mode
+to
+.BR MPOL_BIND .
+See
+.BR mbind (2)
+for more information about this flag.
+.PP
+.BR memkind_default_malloc_usable_size ()
+is a direct call through the jemalloc's
+.BR malloc_usable_size ().
+.PP
+.BR memkind_preferred_get_mbind_mode ()
+sets
+.I mode
+to
+.BR MPOL_PREFERRED .
+See
+.BR mbind (2)
+for more information about this flag.
+.PP
+.BR memkind_interleave_get_mbind_mode ()
+sets
+.I mode
+to
+.BR MPOL_INTERLEAVE .
+See
+.BR mbind (2)
+for more information about this flag.
+.PP
+.BR memkind_nohugepage_madvise ()
+calls
+.BR madvise (2)
+with the
+.B MADV_NOHUGEPAGE
+advice.
+See
+.BR madvise (2)
+for more information about this option.
+.PP
+.BR memkind_posix_check_alignment ()
+can be used to check the alignment value for
+.BR memkind_posix_memalign ()
+to ensure that is abides by the POSIX requirements:
+alignment must be a power of 2 at least as large as
+.IR "sizeof(void*)" .
+.PP
+.BR memkind_default_get_mbind_nodemask ()
+wraps jemalloc's
+.BR copy_bitmask_to_bitmask .
+This function copies body of the bitmask structure into passed pointer.
+.PP
+.BR memkind_default_init_once ()
+initializes heap manager.
+.PP
+.BR size_out_of_bounds ()
+returns true if given size is out of bounds, otherwise will return false.
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/memkind-1.10.0/man/memkind_hbw.3 b/memkind-1.10.0/man/memkind_hbw.3
new file mode 100644
index 0000000..afd9c72
--- /dev/null
+++ b/memkind-1.10.0/man/memkind_hbw.3
@@ -0,0 +1,160 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_HBW" 3 "2016-06-13" "Intel Corporation" "MEMKIND_HBW" \" -*- nroff -*-
+.SH "NAME"
+memkind_hbw \- high bandwidth memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.nf
+.sp
+.BI "int memkind_hbw_check_available(struct memkind " "*kind" );
+.br
+.BI "int memkind_hbw_hugetlb_check_available(struct memkind " "*kind" );
+.br
+.BI "int memkind_hbw_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.BI "int memkind_hbw_all_get_mbind_nodemask(struct memkind " "*kind" ", unsigned long " "*nodemask" ", unsigned long " "maxnode" );
+.br
+.BI "void memkind_hbw_init_once(void);"
+.br
+.BI "void memkind_hbw_hugetlb_init_once(void);"
+.br
+.BI "void memkind_hbw_all_hugetlb_init_once(void);"
+.br
+.BI "void memkind_hbw_preferred_init_once(void);"
+.br
+.BI "void memkind_hbw_preferred_hugetlb_init_once(void);"
+.br
+.BI "void memkind_hbw_interleave_init_once(void);"
+.br
+.SH DESCRIPTION
+.PP
+High bandwidth memory memkind operations.
+.PP
+.BR memkind_hbw_check_available ()
+returns zero if library was able to detect heterogeneous NUMA node
+bandwidths.   Returns
+.B MEMKIND_ERROR_UNAVAILABLE
+if the detection mechanism failed.
+Detection mechanism can be also overridden by the
+environment variable
+.B MEMKIND_HBW_NODES
+as described in the
+.BR memkind (3)
+man page.
+.PP
+.BR memkind_hbw_hugetlb_check_available ()
+In addition to checking for high bandwidth memory as is done by
+.BR memkind_hbw_check_available (),
+this also checks for 2MB huge pages as
+is done by
+.BR memkind_hugetlb_check_available_2mb ().
+.PP
+.BR memkind_hbw_get_mbind_nodemask ()
+sets the
+.I nodemask
+bit to one that corresponds to the high bandwidth NUMA node that has
+the closest NUMA distance to the CPU of the calling process.
+All other bits up to
+.I maxnode
+are set to zero.
+The
+.I nodemask
+can be used in conjunction with the
+.BR mbind (2)
+system call.
+.PP
+.BR memkind_hbw_all_get_mbind_nodemask ()
+sets the
+.I nodemask
+bits to one that correspond to the all high bandwidth NUMA nodes in
+the system. All other bits up to
+.I maxnode
+are set to zero.
+The
+.I nodemask
+can be used in conjunction with the
+.BR mbind (2)
+system call.
+.PP
+.BR memkind_hbw_init_once ()
+This function initializes
+.B MEMKIND_HBW
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_hugetlb_init_once ()
+This function initializes
+.B MEMKIND_HBW_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_hugetlb_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_preferred_init_once ()
+This function initializes
+.B MEMKIND_HBW_PREFERRED
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_preferred_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_preferred_hugetlb_init_once ()
+This function initializes
+.B MEMKIND_HBW_PREFERRED_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_preferred_hugetlb_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_all_hugetlb_init_once ()
+This function initializes
+.B MEMKIND_HBW_ALL_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_all_hugetlb_init_once ()
+may reserve some extra memory.
+.PP
+.BR memkind_hbw_interleave_init_once ()
+This function initializes
+.B MEMKIND_HBW_INTERLEAVE
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hbw_interleave_init_once ()
+may reserve some extra memory.
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hugetlb (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/memkind-1.10.0/man/memkind_hugetlb.3 b/memkind-1.10.0/man/memkind_hugetlb.3
new file mode 100644
index 0000000..52c42b1
--- /dev/null
+++ b/memkind-1.10.0/man/memkind_hugetlb.3
@@ -0,0 +1,77 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_HUGETLB" 3 "2015-04-21" "Intel Corporation" "MEMKIND_HUGETLB" \" -*- nroff -*-
+.SH "NAME"
+memkind_hugetlb \- hugetlb memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_hugetlb_check_available_2mb(struct memkind " "*kind" );
+.br
+.BI "int memkind_hugetlb_get_mmap_flags(struct memkind " "*kind" ", int " "*flags" );
+.br
+.BI "void memkind_hugetlb_init_once(void);"
+.br
+
+.SH DESCRIPTION
+.PP
+The hugetlb memory memkind operations enable memory kinds which use
+the Linux hugetlbfs file system.  For more information about the
+hugetlbfs see link below.
+.br
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.PP
+.BR memkind_hugetlb_check_available_2mb ()
+check if there are 2MB pages reserved in the default hugetlbfs.  If
+the kind implements
+.BR ops.get_mbind_nodemask (),
+then only the NUMA nodes
+set by the nodemask are checked, otherwise every NUMA node is checked.
+.PP
+.BR memkind_hugetlb_get_mmap_flags ()
+sets the flags for the
+.BR mmap ()
+system call such that the hugetlbfs is utilized for allocations.
+.PP
+.BR memkind_hugetlb_init_once ()
+this function initializes
+.B MEMKIND_HUGETLB
+kind and it should not be called more than once.
+.BR Note:
+.BR memkind_hugetlb_init_once ()
+may reserve some extra memory.
+.SH "COPYRIGHT"
+Copyright (C) 2014 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/memkind-1.10.0/man/memkind_interleave.3 b/memkind-1.10.0/man/memkind_interleave.3
new file mode 100644
index 0000000..9e110f2
--- /dev/null
+++ b/memkind-1.10.0/man/memkind_interleave.3
@@ -0,0 +1,51 @@
+.\"
+.\" Copyright (C) 2016 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_INTERLEAVE" 3 "2016-02-19" "Intel Corporation" "MEMKIND_INTERLEAVE" \" -*- nroff -*-
+.SH "NAME"
+memkind_interleave \- interleave memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "void memkind_interleave_init_once(void);"
+.br
+.SH DESCRIPTION
+.PP
+.BR memkind_interleave_init_once ()
+initializes MEMKIND_INTERLEAVE kind and it should not be called more than once.
+.BR Note:
+.BR memkind_interleave_init_once ()
+may reserve some extra memory.
+.SH "COPYRIGHT"
+Copyright (C) 2016 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_pmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/memkind-1.10.0/man/memkind_pmem.3 b/memkind-1.10.0/man/memkind_pmem.3
new file mode 100644
index 0000000..93d02a1
--- /dev/null
+++ b/memkind-1.10.0/man/memkind_pmem.3
@@ -0,0 +1,114 @@
+.\"
+.\" Copyright (C) 2014 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKIND_PMEM" 3 "2015-04-21" "Intel Corporation" "MEMKIND_PMEM" \" -*- nroff -*-
+.SH "NAME"
+memkind_pmem \- file-backed memory memkind operations.
+.br
+.BR Note:
+This is EXPERIMENTAL API. The functionality and the header file itself can be changed (including non-backward compatible changes) or removed.
+.SH "SYNOPSIS"
+.sp
+.BI "int memkind_pmem_create(struct memkind " "*kind" ", struct memkind_ops " "*ops" ", const char " "*name" );
+.br
+.BI "int memkind_pmem_destroy(struct memkind " "*kind" );
+.br
+.BI "void *memkind_pmem_mmap(struct memkind " "*kind" ", void " "*addr" ", size_t " "size" );
+.br
+.BI "int memkind_pmem_get_mmap_flags(struct memkind " "*kind" ", int " "*flags" );
+.br
+.SH DESCRIPTION
+.PP
+The pmem memory memkind operations enable memory kinds built on memory-mapped
+files.  These support traditional
+.B volatile
+memory allocation in a fashion similar to
+.BR libvmem (3)
+library.  It uses the
+.BR mmap (2)
+system call to create a pool of volatile memory.  Such memory may have different
+attributes, depending on the file system containing the memory-mapped files.
+(See also
+.IR http://pmem.io/pmdk/libvmem ).
+.PP
+The pmem memkinds are most useful when used with
+.I Direct Access
+storage (DAX), which is memory-addressable persistent storage
+that supports load/store access without being paged via the system page cache.
+A Persistent Memory-aware file system is typically used to provide this
+type of access.
+.PP
+The most convenient way to create pmem memkinds is to use
+.BR memkind_create_pmem ()
+or
+.BR memkind_create_pmem_with_config ()
+(see
+.BR memkind (3)).
+.PP
+.BR memkind_pmem_create ()
+is an implementation of the memkind "create" operation for file-backed memory
+kinds.  This allocates a space for some pmem-specific metadata, then calls
+.BR memkind_arena_create ()
+(see
+.BR memkind_arena (3))
+.PP
+.BR memkind_pmem_destroy ()
+is an implementation of the memkind "destroy" operation for file-backed memory
+kinds.  This releases all of the resources
+allocated by
+.BR memkind_pmem_create ()
+and allows the file system space to be reclaimed.
+.PP
+.BR memkind_pmem_mmap ()
+allocates the file system space for a block of
+.I size
+bytes in the memory-mapped file associated with given kind.
+The
+.I addr
+hint is ignored.  The return value is the address of mapped memory region or
+.B MAP_FAILED
+in the case of an error.
+.PP
+.BR memkind_pmem_get_mmap_flags ()
+sets
+.I flags
+to
+.BR "MAP_SHARED" .
+See
+.BR mmap (2)
+for more information about these flags.
+.TP
+.B MEMKIND_PMEM_CHUNK_SIZE
+The size of the PMEM chunk size.
+.SH "COPYRIGHT"
+Copyright (C) 2015 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3),
+.BR memkind_arena (3),
+.BR memkind_default (3),
+.BR memkind_hbw (3),
+.BR memkind_hugetlb (3),
+.BR libvmem (3),
+.BR jemalloc (3),
+.BR mbind (2),
+.BR mmap (2)
diff --git a/memkind-1.10.0/man/memkindallocator.3 b/memkind-1.10.0/man/memkindallocator.3
new file mode 100644
index 0000000..9533519
--- /dev/null
+++ b/memkind-1.10.0/man/memkindallocator.3
@@ -0,0 +1,194 @@
+.\"
+.\" Copyright (C) 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "MEMKINDALLOCATOR" 3 "2019-09-24" "Intel Corporation" "MEMKINDALLOCATOR" \" -*- nroff -*-
+.SH "NAME"
+libmemkind::static_kind::allocator<T> \- The C++ allocator compatible with the C++ standard library allocator concepts
+.br
+.BR Note:
+.I memkind_allocator.h
+functionality is considered as stable API (STANDARD API).
+.SH "SYNOPSIS"
+.nf
+.B #include <memkind_allocator.h>
+.sp
+.B Link with -lmemkind
+.sp
+.BI "libmemkind::static_kind::allocator(libmemkind::kinds " "kind" );
+.br
+.BI "template <typename U> libmemkind::static_kind::allocator<T>::allocator(const libmemkind::static_kind::allocator<U>&)" " "noexcept;
+.br
+.BI "template <typename U> libmemkind::static_kind::allocator(const allocator<U>&& " "other" ) " "noexcept;
+.br
+.BI "libmemkind::static_kind::allocator<T>::~allocator();
+.br
+.BI "T *libmemkind::static_kind::allocator<T>::allocate(std::size_t " "n" ) " "const;
+.br
+.BI "void libmemkind::static_kind::allocator<T>::deallocate(T " "*p" ", std::size_t " "n" ) " "const;
+.br
+.BI "template <class U, class... Args> void libmemkind::static_kind::allocator<T>::construct(U " "*p" ", Args... " "args" ) " "const;
+.br
+.BI "void libmemkind::static_kind::allocator<T>::destroy(T " "*p" ) " "const;
+.fi
+.SH "DESCRIPTION"
+The
+.BR libmemkind::static_kind::allocator<T>
+is intended to be used with STL containers to allocate from static kinds memory. Memory management is based on memkind library. Refer
+.BR memkind (3)
+man page for more details.
+.PP
+The
+.BR libmemkind::kinds
+specifies allocator static kinds of memory, representing type of memory which offers different characteristics. The available types of allocator kinds of memory:
+.PP
+.B libmemkind::kinds::DEFAULT
+Default allocation using standard memory and default page size.
+.PP
+.B libmemkind::kinds::HUGETLB
+Allocate from standard memory using huge pages. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::INTERLEAVE
+Allocate pages interleaved across all NUMA nodes with transparent huge pages disabled.
+.PP
+.B libmemkind::kinds::HBW
+Allocate from the closest high bandwidth memory NUMA node at the time of allocation. If there is not enough high bandwidth memory to satisfy the request, errno is set to ENOMEM and the allocated pointer is set to
+.IR "NULL" .
+.PP
+.B libmemkind::kinds::HBW_ALL
+Same as
+.B libmemkind::kinds::HBW
+except decision regarding closest NUMA node is postponed until the time of first write.
+.PP
+.B libmemkind::kinds::HBW_HUGETLB
+Same as
+.B libmemkind::kinds::HBW
+except the allocation is backed by huge pages. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::HBW_ALL_HUGETLB
+Combination of
+.B libmemkind::kinds::HBW_ALL
+and
+.B libmemkind::kinds::HBW_HUGETLB
+properties. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::HBW_PREFERRED
+Same as
+.B libmemkind::kinds::HBW
+except that if there is not enough high bandwidth memory to satisfy the request, the allocation will fall back on standard memory.
+.PP
+.B libmemkind::kinds::HBW_PREFERRED_HUGETLB
+Same as
+.B libmemkind::kinds::HBW_PREFERRED
+except the allocation is backed by huge pages. Note: This kind requires huge pages configuration described in
+.B SYSTEM CONFIGURATION
+section.
+.PP
+.B libmemkind::kinds::HBW_INTERLEAVE
+Same as
+.B libmemkind::kinds::HBW
+except that the pages that support the allocation are interleaved across all high bandwidth nodes and transparent huge pages are disabled.
+.PP
+.B libmemkind::kinds::REGULAR
+Allocate from regular memory using the default page size. Regular means general purpose memory from the NUMA nodes containing CPUs.
+.PP
+.B libmemkind::kinds::DAX_KMEM
+Allocate from the closest persistent memory NUMA node at the time of allocation. If there is not enough memory in the closest persistent memory NUMA node to satisfy the request,
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.PP
+.B libmemkind::kinds::DAX_KMEM_ALL
+Allocate from the closest persistent memory NUMA node available at the time of allocation. If there is not enough memory on any of persistent memory NUMA nodes to satisfy the request,
+.I errno
+is set to
+.B ENOMEM
+and the allocated pointer is set to
+.IR "NULL" .
+.PP
+.B libmemkind::kinds::DAX_KMEM_PREFERRED
+Same as
+.B libmemkind::kinds::DAX_KMEM
+except that if there is not enough memory in the closest persistent memory NUMA node to satisfy the request, the allocation will fall back on other memory NUMA nodes.
+.BR Note:
+For this kind, the allocation will not succeed if two or more
+persistent memory NUMA nodes are in the same shortest distance to the same CPU on which process is eligible to run.
+Check on that eligibility is done upon starting the application.
+.PP
+All public member types and functions correspond to standard library allocator concepts and definitions. The current implementation supports C++11 standard.
+.PP
+Template arguments:
+.br
+.I T
+is an object type aliased by value_type.
+.br
+.I U
+is an object type.
+.PP
+.BR Note:
+.br
+.BI "T *libmemkind::static_kind::allocator<T>::allocate(std::size_t " "n")
+allocates memory using
+.BR memkind_malloc ().
+Throw
+.I std::bad_alloc
+when:
+.RS
+.I n
+= 0
+.RE
+.br
+.RS
+.br
+or there is not enough memory to satisfy the request.
+.RE
+.PP
+.BI "libmemkind::static_kind::allocator<T>::deallocate(T " "*p" ", std::size_t " "n")
+deallocates memory associated with pointer returned by
+.BR allocate ()
+using
+.BR memkind_free ().
+.SH "SYSTEM CONFIGURATION"
+Interfaces for obtaining 2MB (HUGETLB) memory need allocated
+huge pages in the kernel's huge page pool.
+.TP
+.B HUGETLB (huge pages)
+Current number of "persistent" huge pages can be read from
+.I /proc/sys/vm/nr_hugepages
+file.
+Proposed way of setting hugepages is:
+.BR "sudo sysctl vm.nr_hugepages=<number_of_hugepages>" .
+More information can be found here:
+.UR https://www.kernel.org/doc/Documentation/vm/hugetlbpage.txt
+.UE
+.SH "COPYRIGHT"
+Copyright (C) 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind (3)
diff --git a/memkind-1.10.0/man/pmemallocator.3 b/memkind-1.10.0/man/pmemallocator.3
new file mode 100644
index 0000000..3c07a4c
--- /dev/null
+++ b/memkind-1.10.0/man/pmemallocator.3
@@ -0,0 +1,116 @@
+.\"
+.\" Copyright (C) 2018 - 2019 Intel Corporation.
+.\" All rights reserved.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions are met:
+.\" 1. Redistributions of source code must retain the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright notice(s),
+.\"    this list of conditions and the following disclaimer in the documentation
+.\"    and/or other materials provided with the distribution.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+.\" OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+.\" MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+.\" EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+.\" INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+.\" LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+.\" PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+.\" LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+.\" ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+.\"
+.TH "PMEMALLOCATOR" 3 "2018-09-13" "Intel Corporation" "PMEMALLOCATOR" \" -*- nroff -*-
+.SH "NAME"
+libmemkind::pmem::allocator<T> \- The C++ allocator compatible with the C++ standard library allocator concepts
+.br
+.BR Note:
+.I pmem_allocator.h
+functionality is considered as stable API (STANDARD API).
+.SH "SYNOPSIS"
+.nf
+.B #include <pmem_allocator.h>
+.sp
+.B Link with -lmemkind
+.sp
+.BI "libmemkind::pmem::allocator(const char " "*dir" ", size_t " "max_size" );
+.br
+.BI "libmemkind::pmem::llocator(const char " "&dir" ", size_t " "max_size" ", libmemkind::allocation_policy " "alloc_policy" );
+.br
+.BI "libmemkind::pmem::allocator(const std::string " "&dir" ", size_t " "max_size" );
+.br
+.BI "libmemkind::pmem::allocator(const std::string " "&dir" ", size_t " "max_size" ", libmemkind::allocation_policy " "alloc_policy" );
+.br
+.BI "template <typename U> libmemkind::pmem::allocator<T>::allocator(const libmemkind::pmem::allocator<U>&)" " "noexcept;
+.br
+.BI "template <typename U> libmemkind::pmem::allocator(const allocator<U>&& " "other" ) " "noexcept;
+.br
+.BI "libmemkind::pmem::allocator<T>::~allocator();
+.br
+.BI "T *libmemkind::pmem::allocator<T>::allocate(std::size_t " "n" ) " "const;
+.br
+.BI "void libmemkind::pmem::allocator<T>::deallocate(T " "*p" ", std::size_t " "n" ) " "const;
+.br
+.BI "template <class U, class... Args> void libmemkind::pmem::allocator<T>::construct(U " "*p" ", Args... " "args" ) " "const;
+.br
+.BI "void libmemkind::pmem::allocator<T>::destroy(T " "*p" ) " "const;
+.fi
+.SH "DESCRIPTION"
+The
+.BR libmemkind::pmem::allocator<T>
+is intended to be used with STL containers to allocate persistent memory. Memory management is based on memkind_pmem (memkind library). Refer
+.BR memkind_pmem (3)
+and
+.BR memkind (3)
+man page for more details.
+.PP
+The
+.BR libmemkind::allocation_policy
+specifies allocator memory usage policy, which allows to tune up memory utilization. The available types of allocator usage policy:
+.PP
+.B libmemkind::allocation_policy::DEFAULT
+Default allocator memory usage policy.
+.PP
+.B libmemkind::allocation_policy::CONSERVATIVE
+Conservative allocator memory usage policy - prioritize memory usage at cost of performance.
+.PP
+All public member types and functions corresponds to standard library allocator concepts and definitions. The current implementation supports C++11 standard.
+.PP
+Template arguments:
+.br
+.I T
+is an object type aliased by value_type.
+.br
+.I U
+is an object type.
+.PP
+.BR Note:
+.br
+.BI "T *libmemkind::pmem::allocator<T>::allocate(std::size_t " "n")
+allocates persistent memory using
+.BR memkind_malloc ().
+Throw
+.I std::bad_alloc
+when:
+.br
+.RS
+.I n
+= 0
+.RE
+.br
+.RS
+or there is not enough memory to satisfy the request.
+.RE
+.PP
+.BI "libmemkind::pmem::allocator<T>::deallocate(T " "*p" ", std::size_t " "n")
+deallocates memory associated with pointer returned by
+.BR allocate ()
+using
+.BR memkind_free ().
+
+.SH "COPYRIGHT"
+Copyright (C) 2018 - 2019 Intel Corporation. All rights reserved.
+.SH "SEE ALSO"
+.BR memkind_pmem (3),
+.BR memkind (3)
diff --git a/memkind-1.10.0/memkind.pc.in b/memkind-1.10.0/memkind.pc.in
new file mode 100644
index 0000000..4441acc
--- /dev/null
+++ b/memkind-1.10.0/memkind.pc.in
@@ -0,0 +1,13 @@
+version=@VERSION@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: memkind
+Description: User Extensible Heap Manager
+URL: http://memkind.github.io/memkind
+Version: @VERSION@
+Cflags: -I${includedir}
+Libs: -L${libdir} -lmemkind
+Libs.private: -ldl -lnuma -lpthread @DAXCTL_LIBS@
diff --git a/memkind-1.10.0/memkind.spec.mk b/memkind-1.10.0/memkind.spec.mk
new file mode 100644
index 0000000..49a29e1
--- /dev/null
+++ b/memkind-1.10.0/memkind.spec.mk
@@ -0,0 +1,262 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+# targets for building rpm
+version ?= 0.0.0
+release ?= 1
+arch = $(shell uname -p)
+name ?= memkind
+
+rpm: $(name)-$(version).tar.gz
+	rpmbuild $(rpmbuild_flags) $^ -ta
+
+memkind-$(version).spec:
+	@echo "$$memkind_spec" > $@
+	cat ChangeLog >> $@
+
+.PHONY: rpm
+
+define memkind_spec
+Summary: User Extensible Heap Manager
+Name: $(name)
+Version: $(version)
+Release: $(release)
+License: BSD-2-Clause
+Group: System Environment/Libraries
+URL: http://memkind.github.io/memkind
+BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
+BuildRequires: automake libtool gcc-c++ unzip
+%if %{defined suse_version}
+BuildRequires: libnuma-devel
+%else
+BuildRequires: numactl-devel
+%endif
+
+%define daxctl_min_version 66
+%if %{defined suse_version}
+BuildRequires: libdaxctl-devel >= %{daxctl_min_version}
+%else
+BuildRequires: daxctl-devel >= %{daxctl_min_version}
+%endif
+
+Prefix: %{_prefix}
+Prefix: %{_unitdir}
+%if %{undefined suse_version}
+Obsoletes: memkind
+Provides: memkind libmemkind0
+%endif
+
+%define namespace memkind
+
+%if %{defined suse_version}
+%define docdir %{_defaultdocdir}/%{namespace}
+%else
+%define docdir %{_defaultdocdir}/%{namespace}-%{version}
+%endif
+
+# Upstream testing of memkind is done exclusively on x86_64; other archs
+# are unsupported but may work.
+ExclusiveArch: x86_64 ppc64 ppc64le s390x aarch64
+
+# default values if version is a tagged release on github
+%{!?commit: %define commit %{version}}
+%{!?buildsubdir: %define buildsubdir %{namespace}-%{commit}}
+Source0: https://github.com/%{namespace}/%{namespace}/archive/v%{commit}/%{buildsubdir}.tar.gz
+
+%description
+The memkind library is an user extensible heap manager built on top of
+jemalloc which enables control of memory characteristics and a
+partitioning of the heap between kinds of memory. The kinds of memory
+are defined by operating system memory policies that have been applied
+to virtual address ranges. Memory characteristics supported by
+memkind without user extension include control of NUMA and page size
+features. The jemalloc non-standard interface has been extended to
+enable specialized arenas to make requests for virtual memory from the
+operating system through the memkind partition interface. Through the
+other memkind interfaces the user can control and extend memory
+partition features and allocate memory while selecting enabled
+features. This software is being made available for early evaluation.
+Feedback on design or implementation is greatly appreciated.
+
+%package devel
+Summary: Memkind User Extensible Heap Manager development lib and tools
+Group: Development/Libraries
+Requires: %{name} = %{version}-%{release}
+%if %{undefined suse_version}
+Obsoletes: memkind-devel
+Provides: memkind-devel
+%endif
+
+%description devel
+Install header files and development aids to link memkind library into
+applications.
+
+%package tests
+Summary: Extension to libnuma for kinds of memory - validation
+Group: Validation/Libraries
+Requires: %{name} = %{version}-%{release}
+
+%description tests
+memkind functional tests
+
+%prep
+%setup -q -a 0 -n $(name)-%{version}
+
+%build
+# It is required that we configure and build the jemalloc subdirectory
+# before we configure and start building the top level memkind directory.
+# To ensure the memkind build step is able to discover the output
+# of the jemalloc build we must create an 'obj' directory, and build
+# from within that directory.
+
+cd %{_builddir}/%{buildsubdir}
+echo %{version} > %{_builddir}/%{buildsubdir}/VERSION
+./build.sh --prefix=%{_prefix} --includedir=%{_includedir} --libdir=%{_libdir} \
+           --bindir=%{_bindir} --docdir=%{_docdir}/%{namespace} --mandir=%{_mandir} --sbindir=%{_sbindir}
+
+%install
+cd %{_builddir}/%{buildsubdir}
+%{__make} DESTDIR=%{buildroot} install
+%{__install} -d %{buildroot}$(memkind_test_dir)
+%{__install} -d %{buildroot}/%{_unitdir}
+%{__install} -d %{buildroot}/$(memkind_test_dir)/python_framework
+%{__install} test/.libs/* test/*.sh test/*.ts test/*.py %{buildroot}$(memkind_test_dir)
+%{__install} test/python_framework/*.py %{buildroot}/$(memkind_test_dir)/python_framework
+rm -f %{buildroot}$(memkind_test_dir)/libautohbw.*
+rm -f %{buildroot}/%{_libdir}/lib%{namespace}.{l,}a
+rm -f %{buildroot}/%{_libdir}/libautohbw.{l,}a
+
+%pre
+
+%post
+/sbin/ldconfig
+
+%preun
+
+%postun
+/sbin/ldconfig
+
+%files
+%defattr(-,root,root,-)
+%license %{_docdir}/%{namespace}/COPYING
+%doc %{_docdir}/%{namespace}/README
+%doc %{_docdir}/%{namespace}/VERSION
+%dir %{_docdir}/%{namespace}
+%{_libdir}/lib%{namespace}.so.*
+%{_libdir}/libautohbw.so.*
+%{_bindir}/%{namespace}-auto-dax-kmem-nodes
+%{_bindir}/%{namespace}-hbw-nodes
+
+%files devel
+%defattr(-,root,root,-)
+%{_includedir}
+%{_includedir}/hbwmalloc.h
+%{_includedir}/hbw_allocator.h
+%{_includedir}/memkind_allocator.h
+%{_includedir}/pmem_allocator.h
+%{_libdir}/lib%{namespace}.so
+%{_libdir}/libautohbw.so
+%{_libdir}/pkgconfig/memkind.pc
+%{_includedir}/%{namespace}.h
+%{_mandir}/man1/memkind-auto-dax-kmem-nodes.1.*
+%{_mandir}/man1/memkind-hbw-nodes.1.*
+%{_mandir}/man3/hbwmalloc.3.*
+%{_mandir}/man3/hbwallocator.3.*
+%{_mandir}/man3/pmemallocator.3.*
+%{_mandir}/man3/%{namespace}*.3.*
+%{_mandir}/man7/autohbw.7.*
+
+%files tests
+%defattr(-,root,root,-)
+$(memkind_test_dir)/all_tests
+${memkind_test_dir}/environ_err_dax_kmem_malloc_positive_test
+${memkind_test_dir}/environ_err_dax_kmem_malloc_test
+$(memkind_test_dir)/environ_err_hbw_malloc_test
+${memkind_test_dir}/dax_kmem_test
+$(memkind_test_dir)/decorator_test
+$(memkind_test_dir)/locality_test
+$(memkind_test_dir)/freeing_memory_segfault_test
+$(memkind_test_dir)/gb_page_tests_bind_policy
+$(memkind_test_dir)/filter_memkind
+$(memkind_test_dir)/hello_hbw
+$(memkind_test_dir)/hello_memkind
+$(memkind_test_dir)/hello_memkind_debug
+$(memkind_test_dir)/memkind_allocated
+$(memkind_test_dir)/memkind_cpp_allocator
+$(memkind_test_dir)/memkind_get_stat
+$(memkind_test_dir)/memkind_stat_test
+$(memkind_test_dir)/autohbw_candidates
+${memkind_test_dir}/pmem_kinds
+${memkind_test_dir}/pmem_malloc
+${memkind_test_dir}/pmem_malloc_unlimited
+${memkind_test_dir}/pmem_usable_size
+${memkind_test_dir}/pmem_alignment
+${memkind_test_dir}/pmem_and_dax_kmem_kind
+${memkind_test_dir}/pmem_and_default_kind
+${memkind_test_dir}/pmem_config
+$(memkind_test_dir)/pmem_detect_kind
+${memkind_test_dir}/pmem_multithreads
+${memkind_test_dir}/pmem_multithreads_onekind
+$(memkind_test_dir)/pmem_free_with_unknown_kind
+${memkind_test_dir}/pmem_cpp_allocator
+${memkind_test_dir}/pmem_test
+${memkind_test_dir}/allocator_perf_tool_tests
+${memkind_test_dir}/perf_tool
+${memkind_test_dir}/autohbw_test_helper
+${memkind_test_dir}/trace_mechanism_test_helper
+$(memkind_test_dir)/memkind-afts.ts
+$(memkind_test_dir)/memkind-afts-ext.ts
+$(memkind_test_dir)/memkind-slts.ts
+$(memkind_test_dir)/memkind-perf.ts
+$(memkind_test_dir)/memkind-perf-ext.ts
+$(memkind_test_dir)/memkind-pytests.ts
+$(memkind_test_dir)/performance_test
+$(memkind_test_dir)/test.sh
+$(memkind_test_dir)/test_dax_kmem.sh
+$(memkind_test_dir)/hbw_detection_test.py
+$(memkind_test_dir)/dax_kmem_env_var_test.py
+$(memkind_test_dir)/autohbw_test.py
+$(memkind_test_dir)/trace_mechanism_test.py
+$(memkind_test_dir)/python_framework
+$(memkind_test_dir)/python_framework/cmd_helper.py
+$(memkind_test_dir)/python_framework/huge_page_organizer.py
+$(memkind_test_dir)/python_framework/__init__.py
+$(memkind_test_dir)/draw_plots.py
+$(memkind_test_dir)/run_alloc_benchmark.sh
+$(memkind_test_dir)/alloc_benchmark_hbw
+$(memkind_test_dir)/alloc_benchmark_glibc
+$(memkind_test_dir)/alloc_benchmark_tbb
+$(memkind_test_dir)/alloc_benchmark_pmem
+$(memkind_test_dir)/fragmentation_benchmark_pmem
+$(memkind_test_dir)/defrag_reallocate
+
+%exclude $(memkind_test_dir)/*.pyo
+%exclude $(memkind_test_dir)/*.pyc
+%exclude $(memkind_test_dir)/python_framework/*.pyo
+%exclude $(memkind_test_dir)/python_framework/*.pyc
+
+%changelog
+endef
+
+export memkind_spec
diff --git a/memkind-1.10.0/src/Makefile.mk b/memkind-1.10.0/src/Makefile.mk
new file mode 100644
index 0000000..714a8dc
--- /dev/null
+++ b/memkind-1.10.0/src/Makefile.mk
@@ -0,0 +1,28 @@
+#
+#  Copyright (C) 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+clean-local: src-clean
+
+src-clean:
+	rm -f src/*.gcno
diff --git a/memkind-1.10.0/src/hbwmalloc.c b/memkind-1.10.0/src/hbwmalloc.c
new file mode 100644
index 0000000..16d8c29
--- /dev/null
+++ b/memkind-1.10.0/src/hbwmalloc.c
@@ -0,0 +1,302 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbwmalloc.h>
+#include <memkind.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <pthread.h>
+#include <errno.h>
+#include <numa.h>
+#include <numaif.h>
+#include <unistd.h>
+#include <stdint.h>
+
+static hbw_policy_t hbw_policy_g = HBW_POLICY_PREFERRED;
+static pthread_once_t hbw_policy_once_g = PTHREAD_ONCE_INIT;
+
+static void hbw_policy_bind_init(void)
+{
+    hbw_policy_g = HBW_POLICY_BIND;
+}
+
+static void hbw_policy_bind_all_init(void)
+{
+    hbw_policy_g = HBW_POLICY_BIND_ALL;
+}
+
+static void hbw_policy_preferred_init(void)
+{
+    hbw_policy_g = HBW_POLICY_PREFERRED;
+}
+
+static void hbw_policy_interleave_init(void)
+{
+    hbw_policy_g = HBW_POLICY_INTERLEAVE;
+}
+
+// This function is intended to be called once per pagesize
+// Getting kind should be done using hbw_get_kind() defined below
+static memkind_t hbw_choose_kind(hbw_pagesize_t pagesize)
+{
+    memkind_t result = NULL;
+
+    hbw_set_policy(hbw_policy_g);
+
+    int policy = hbw_get_policy();
+
+    // PREFERRED policy have separate handling cause it can fallback
+    // to non-HBW kinds in case of HBW absence
+    if (policy != HBW_POLICY_PREFERRED ) {
+        switch (pagesize) {
+            case HBW_PAGESIZE_2MB:
+                if(policy == HBW_POLICY_BIND_ALL) {
+                    result = MEMKIND_HBW_ALL_HUGETLB;
+                } else {
+                    result = MEMKIND_HBW_HUGETLB;
+                }
+                break;
+            case HBW_PAGESIZE_1GB:
+            case HBW_PAGESIZE_1GB_STRICT:
+                result = MEMKIND_HBW_GBTLB;
+                break;
+            default:
+                if (policy == HBW_POLICY_BIND) {
+                    result = MEMKIND_HBW;
+                } else if (policy == HBW_POLICY_BIND_ALL) {
+                    result = MEMKIND_HBW_ALL;
+                } else {
+                    result = MEMKIND_HBW_INTERLEAVE;
+                }
+                break;
+        }
+    } else if (memkind_check_available(MEMKIND_HBW) == 0) {
+        switch (pagesize) {
+            case HBW_PAGESIZE_2MB:
+                result = MEMKIND_HBW_PREFERRED_HUGETLB;
+                break;
+            case HBW_PAGESIZE_1GB:
+            case HBW_PAGESIZE_1GB_STRICT:
+                result = MEMKIND_HBW_PREFERRED_GBTLB;
+                break;
+            default:
+                result = MEMKIND_HBW_PREFERRED;
+                break;
+        }
+    } else {
+        switch (pagesize) {
+            case HBW_PAGESIZE_2MB:
+                result = MEMKIND_HUGETLB;
+                break;
+            case HBW_PAGESIZE_1GB:
+            case HBW_PAGESIZE_1GB_STRICT:
+                result = MEMKIND_GBTLB;
+                break;
+            default:
+                result = MEMKIND_DEFAULT;
+                break;
+        }
+    }
+    return result;
+}
+
+static memkind_t pagesize_kind[HBW_PAGESIZE_MAX_VALUE];
+static inline memkind_t hbw_get_kind(hbw_pagesize_t pagesize)
+{
+    if(pagesize_kind[pagesize] == NULL) {
+        pagesize_kind[pagesize] = hbw_choose_kind(pagesize);
+    }
+    return pagesize_kind[pagesize];
+}
+
+
+MEMKIND_EXPORT hbw_policy_t hbw_get_policy(void)
+{
+    return hbw_policy_g;
+}
+
+MEMKIND_EXPORT int hbw_set_policy(hbw_policy_t mode)
+{
+    switch(mode) {
+        case HBW_POLICY_PREFERRED:
+            pthread_once(&hbw_policy_once_g, hbw_policy_preferred_init);
+            break;
+        case HBW_POLICY_BIND:
+            pthread_once(&hbw_policy_once_g, hbw_policy_bind_init);
+            break;
+        case HBW_POLICY_BIND_ALL:
+            pthread_once(&hbw_policy_once_g, hbw_policy_bind_all_init);
+            break;
+        case HBW_POLICY_INTERLEAVE:
+            pthread_once(&hbw_policy_once_g, hbw_policy_interleave_init);
+            break;
+        default:
+            return EINVAL;
+    }
+
+    if (mode != hbw_policy_g) {
+        return EPERM;
+    }
+
+    return 0;
+}
+
+MEMKIND_EXPORT int hbw_check_available(void)
+{
+    return  (memkind_check_available(MEMKIND_HBW) == 0) ? 0 : ENODEV;
+}
+
+static inline void hbw_touch_page(void *addr)
+{
+    volatile char *temp_ptr = (volatile char *) addr;
+    char value = temp_ptr[0];
+    temp_ptr[0] = value;
+}
+
+MEMKIND_EXPORT int hbw_verify_memory_region(void *addr, size_t size, int flags)
+{
+    /*
+     * if size is invalid, flags have unsupported bit set or if addr is NULL.
+     */
+    if (addr == NULL || size == 0 || flags & ~HBW_TOUCH_PAGES) {
+        return EINVAL;
+    }
+
+    /*
+     * 4KB is the smallest pagesize. When pagesize is bigger, pages are verified more than once
+     */
+    const size_t page_size = sysconf(_SC_PAGESIZE);
+    const size_t page_mask = ~(page_size-1);
+
+    /*
+     * block size should be power of two to enable compiler optimizations
+     */
+    const unsigned block_size = 64;
+
+    char *end = addr + size;
+    char *aligned_beg = (char *)((uintptr_t)addr & page_mask);
+    nodemask_t nodemask;
+    struct bitmask expected_nodemask = {NUMA_NUM_NODES, nodemask.n};
+
+    memkind_hbw_all_get_mbind_nodemask(NULL, expected_nodemask.maskp,
+                                       expected_nodemask.size);
+
+    while(aligned_beg < end) {
+        int nodes[block_size];
+        void *pages[block_size];
+        int i = 0, page_count = 0;
+        char *iter_end = aligned_beg + block_size*page_size;
+
+        if (iter_end > end) {
+            iter_end = end;
+        }
+
+        while (aligned_beg < iter_end) {
+            if (flags & HBW_TOUCH_PAGES) {
+                hbw_touch_page(aligned_beg);
+            }
+            pages[page_count++] = aligned_beg;
+            aligned_beg += page_size;
+        }
+
+        if (move_pages(0, page_count, pages, NULL, nodes, MPOL_MF_MOVE)) {
+            return EFAULT;
+        }
+
+        for (i = 0; i < page_count; i++) {
+            /*
+             * negative value of nodes[i] indicates that move_pages could not establish
+             * page location, e.g. addr is not pointing to valid virtual mapping
+             */
+            if(nodes[i] < 0) {
+                return -1;
+            }
+            /*
+             * if nodes[i] is not present in expected_nodemask then
+             * physical memory backing page is not hbw
+             */
+            if (!numa_bitmask_isbitset(&expected_nodemask, nodes[i])) {
+                return -1;
+            }
+        }
+    }
+
+    return 0;
+}
+
+MEMKIND_EXPORT void *hbw_malloc(size_t size)
+{
+    return memkind_malloc(hbw_get_kind(HBW_PAGESIZE_4KB), size);
+}
+
+MEMKIND_EXPORT void *hbw_calloc(size_t num, size_t size)
+{
+    return memkind_calloc(hbw_get_kind(HBW_PAGESIZE_4KB), num, size);
+}
+
+MEMKIND_EXPORT int hbw_posix_memalign(void **memptr, size_t alignment,
+                                      size_t size)
+{
+    return memkind_posix_memalign(hbw_get_kind(HBW_PAGESIZE_4KB), memptr, alignment,
+                                  size);
+}
+
+MEMKIND_EXPORT int hbw_posix_memalign_psize(void **memptr, size_t alignment,
+                                            size_t size,
+                                            hbw_pagesize_t pagesize)
+{
+    if (pagesize == HBW_PAGESIZE_1GB_STRICT && size % (1 << 30)) {
+        return EINVAL;
+    }
+
+    if((pagesize == HBW_PAGESIZE_2MB ||
+        pagesize == HBW_PAGESIZE_1GB_STRICT ||
+        pagesize == HBW_PAGESIZE_1GB) &&
+       hbw_get_policy() == HBW_POLICY_INTERLEAVE) {
+
+        log_err("HBW_POLICY_INTERLEAVE is unsupported with used page size!");
+        return EINVAL;
+    }
+
+    return memkind_posix_memalign(hbw_get_kind(pagesize), memptr, alignment, size);
+}
+
+MEMKIND_EXPORT void *hbw_realloc(void *ptr, size_t size)
+{
+    return memkind_realloc(hbw_get_kind(HBW_PAGESIZE_4KB), ptr, size);
+}
+
+MEMKIND_EXPORT void hbw_free(void *ptr)
+{
+    memkind_free(0, ptr);
+}
+
+MEMKIND_EXPORT size_t hbw_malloc_usable_size(void *ptr)
+{
+    return memkind_malloc_usable_size(0, ptr);
+}
diff --git a/memkind-1.10.0/src/heap_manager.c b/memkind-1.10.0/src/heap_manager.c
new file mode 100644
index 0000000..f6c8678
--- /dev/null
+++ b/memkind-1.10.0/src/heap_manager.c
@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/heap_manager.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/memkind_arena.h>
+
+#include <stdio.h>
+#include <pthread.h>
+#include <string.h>
+
+static struct heap_manager_ops *heap_manager_g;
+
+static pthread_once_t heap_manager_init_once_g = PTHREAD_ONCE_INIT;
+
+struct heap_manager_ops {
+    void (*init)(struct memkind *kind);
+    size_t (*heap_manager_malloc_usable_size)(void *ptr);
+    void (*heap_manager_free)(void *ptr);
+    void *(*heap_manager_realloc)(void *ptr, size_t size);
+    struct memkind *(*heap_manager_detect_kind)(void *ptr);
+    int (*heap_manager_update_cached_stats)(void);
+    int (*heap_manager_get_stat)(memkind_stat_type stat, size_t *value);
+    void *(*heap_manager_defrag_reallocate)(void *ptr);
+};
+
+static struct heap_manager_ops arena_heap_manager_g = {
+    .init = memkind_arena_init,
+    .heap_manager_malloc_usable_size = memkind_arena_malloc_usable_size,
+    .heap_manager_free = memkind_arena_free_with_kind_detect,
+    .heap_manager_realloc = memkind_arena_realloc_with_kind_detect,
+    .heap_manager_detect_kind = memkind_arena_detect_kind,
+    .heap_manager_update_cached_stats = memkind_arena_update_cached_stats,
+    .heap_manager_get_stat = memkind_arena_get_global_stat,
+    .heap_manager_defrag_reallocate = memkind_arena_defrag_reallocate_with_kind_detect
+};
+
+static struct heap_manager_ops tbb_heap_manager_g = {
+    .init = tbb_initialize,
+    .heap_manager_malloc_usable_size = tbb_pool_malloc_usable_size_with_kind_detect,
+    .heap_manager_free = tbb_pool_free_with_kind_detect,
+    .heap_manager_realloc = tbb_pool_realloc_with_kind_detect,
+    .heap_manager_detect_kind = tbb_detect_kind,
+    .heap_manager_update_cached_stats = tbb_update_cached_stats,
+    .heap_manager_get_stat = tbb_get_global_stat,
+    .heap_manager_defrag_reallocate = tbb_pool_defrag_reallocate_with_kind_detect
+};
+
+static void set_heap_manager()
+{
+    heap_manager_g = &arena_heap_manager_g;
+    const char *env = secure_getenv("MEMKIND_HEAP_MANAGER");
+    if(env && strcmp(env, "TBB") == 0) {
+        heap_manager_g = &tbb_heap_manager_g;
+    }
+}
+
+static inline struct heap_manager_ops *get_heap_manager()
+{
+    pthread_once(&heap_manager_init_once_g, set_heap_manager);
+    return heap_manager_g;
+}
+
+void heap_manager_init(struct memkind *kind)
+{
+    get_heap_manager()->init(kind);
+}
+
+size_t heap_manager_malloc_usable_size(void *ptr)
+{
+    return get_heap_manager()->heap_manager_malloc_usable_size(ptr);
+}
+
+void heap_manager_free(void *ptr)
+{
+    get_heap_manager()->heap_manager_free(ptr);
+}
+
+void *heap_manager_realloc(void *ptr, size_t size)
+{
+    return get_heap_manager()->heap_manager_realloc(ptr, size);
+}
+
+struct memkind *heap_manager_detect_kind(void *ptr)
+{
+    return get_heap_manager()->heap_manager_detect_kind(ptr);
+}
+
+int heap_manager_update_cached_stats(void)
+{
+    return get_heap_manager()->heap_manager_update_cached_stats();
+}
+
+int heap_manager_get_stat(memkind_stat_type stat, size_t *value)
+{
+    return get_heap_manager()->heap_manager_get_stat(stat, value);
+}
+
+void *heap_manager_defrag_reallocate(void *ptr)
+{
+    return get_heap_manager()->heap_manager_defrag_reallocate(ptr);
+}
diff --git a/memkind-1.10.0/src/memkind-auto-dax-kmem-nodes.c b/memkind-1.10.0/src/memkind-auto-dax-kmem-nodes.c
new file mode 100644
index 0000000..06c30dc
--- /dev/null
+++ b/memkind-1.10.0/src/memkind-auto-dax-kmem-nodes.c
@@ -0,0 +1,108 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_dax_kmem.h>
+#include <memkind.h>
+#include <numa.h>
+#include <stdio.h>
+
+#define MAX_ARG_LEN 8
+
+static const char *const help_message =
+    "\n"
+    "NAME\n"
+    "    memkind-auto-dax-kmem-nodes - prints comma-separated list of automatically detected persistent memory NUMA nodes.\n"
+    "\n"
+    "SYNOPSIS\n"
+    "    memkind-auto-dax-kmem-nodes -h | --help\n"
+    "        Prints this help message.\n"
+    "\n"
+    "DESCRIPTION\n"
+    "    Prints a comma-separated list of automatically detected persistent memory NUMA nodes\n"
+    "    that can be used with the numactl --membind option.\n"
+    "\n"
+    "EXIT STATUS\n"
+    "    Return code is :\n"
+    "        0 on success\n"
+    "        1 if automatic detection is disabled (memkind wasn't build with suitable libdaxctl-devel version)\n"
+    "        2 on invalid argument\n"
+    "        3 on other failure\n"
+    "\n"
+    "COPYRIGHT\n"
+    "    Copyright 2019 Intel Corporation All Rights Reserved.\n"
+    "\n"
+    "AUTHORS\n"
+    "    Michal Biesek\n"
+    "\n"
+    "SEE ALSO\n"
+    "    memkind(3)\n"
+    "\n";
+
+
+static int print_dax_kmem_nodes()
+{
+    unsigned i, j = 0;
+
+    nodemask_t nodemask;
+    struct bitmask nodemask_dax_kmem = {NUMA_NUM_NODES, nodemask.n};
+
+    // ensuring functions in numa library are defined
+    if (numa_available() == -1) {
+        return 3;
+    }
+    numa_bitmask_clearall(&nodemask_dax_kmem);
+
+    //WARNING: code below is usage of memkind experimental API which may be changed in future
+    int status = memkind_dax_kmem_all_get_mbind_nodemask(NULL, nodemask.n,
+                                                         NUMA_NUM_NODES);
+
+    if (status == MEMKIND_ERROR_OPERATION_FAILED ) {
+        return 1;
+    } else if (status != MEMKIND_SUCCESS ) {
+        return 3;
+    }
+
+    for(i = 0; i<NUMA_NUM_NODES; i++) {
+        if(numa_bitmask_isbitset(&nodemask_dax_kmem, i)) {
+            printf("%u%s", i, (++j == numa_bitmask_weight(&nodemask_dax_kmem)) ? "" : ",");
+        }
+    }
+    printf("\n");
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+    if(argc == 1) {
+        return print_dax_kmem_nodes();
+    } else if ((argc == 2) && (strncmp(argv[1], "-h", MAX_ARG_LEN) == 0 ||
+                               strncmp(argv[1], "--help", MAX_ARG_LEN) == 0)) {
+        printf("%s", help_message);
+        return 0;
+    }
+
+    printf("ERROR: Unknown option %s. More info with \"%s --help\".\n", argv[1],
+           argv[0]);
+    return 2;
+}
diff --git a/memkind-1.10.0/src/memkind-hbw-nodes.c b/memkind-1.10.0/src/memkind-hbw-nodes.c
new file mode 100644
index 0000000..9b3fac4
--- /dev/null
+++ b/memkind-1.10.0/src/memkind-hbw-nodes.c
@@ -0,0 +1,103 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+
+#include <numa.h>
+#include <stdio.h>
+
+#define MAX_ARG_LEN 8
+
+const char *help_message =
+    "\n"
+    "NAME\n"
+    "    memkind-hbw-nodes - Prints comma-separated list of high bandwidth nodes.\n"
+    "\n"
+    "SYNOPSIS\n"
+    "    memkind-hbw-nodes -h | --help\n"
+    "        Prints this help message.\n"
+    "\n"
+    "DESCRIPTION\n"
+    "    Prints a comma-separated list of high bandwidth NUMA nodes\n"
+    "    that can be used with the numactl --membind option.\n"
+    "\n"
+    "EXIT STATUS\n"
+    "    Return code is :\n"
+    "        0 on success\n"
+    "        1 on failure\n"
+    "        2 on invalid argument\n"
+    "\n"
+    "COPYRIGHT\n"
+    "    Copyright 2016 Intel Corporation All Rights Reserved.\n"
+    "\n"
+    "AUTHORS\n"
+    "    Krzysztof Kulakowski\n"
+    "\n"
+    "SEE ALSO\n"
+    "    hbwmalloc(3), memkind(3)\n"
+    "\n";
+
+extern unsigned int numa_bitmask_weight(const struct bitmask *bmp );
+
+int print_hbw_nodes()
+{
+    int i, j = 0;
+
+    nodemask_t nodemask;
+    struct bitmask nodemask_bm = {NUMA_NUM_NODES, nodemask.n};
+
+    // ensuring functions in numa library are defined
+    if (numa_available() == -1) {
+        return 3;
+    }
+    numa_bitmask_clearall(&nodemask_bm);
+
+    //WARNING: code below is usage of memkind experimental API which may be changed in future
+    if(memkind_hbw_all_get_mbind_nodemask(NULL, nodemask.n, NUMA_NUM_NODES) != 0) {
+        return 1;
+    }
+
+    for(i=0; i<NUMA_NUM_NODES; i++) {
+        if(numa_bitmask_isbitset(&nodemask_bm, i)) {
+            printf("%d%s", i, (++j == numa_bitmask_weight(&nodemask_bm)) ? "" : ",");
+        }
+    }
+    printf("\n");
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+    if(argc == 1) {
+        return print_hbw_nodes();
+    } else if ((argc == 2) && (strncmp(argv[1], "-h", MAX_ARG_LEN) == 0 ||
+                               strncmp(argv[1], "--help", MAX_ARG_LEN) == 0)) {
+        printf("%s", help_message);
+        return 2;
+    }
+
+    printf("ERROR: Unknown option %s. More info with \"%s --help\".\n", argv[1],
+           argv[0]);
+    return 2;
+}
diff --git a/memkind-1.10.0/src/memkind.c b/memkind-1.10.0/src/memkind.c
new file mode 100644
index 0000000..d3f169d
--- /dev/null
+++ b/memkind-1.10.0/src/memkind.c
@@ -0,0 +1,924 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#define MEMKIND_VERSION_MAJOR 1
+#define MEMKIND_VERSION_MINOR 10
+#define MEMKIND_VERSION_PATCH 0
+
+#include <memkind.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_regular.h>
+#include <memkind/internal/memkind_gbtlb.h>
+#include <memkind/internal/memkind_dax_kmem.h>
+#include <memkind/internal/memkind_pmem.h>
+#include <memkind/internal/memkind_interleave.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/heap_manager.h>
+
+#include "config.h"
+
+#include <numa.h>
+#include <sys/param.h>
+#include <sys/mman.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+#include <pthread.h>
+#include <errno.h>
+#include <signal.h>
+#include <fcntl.h>
+#include <unistd.h>
+
+/* Clear bits in x, but only this specified in mask. */
+#define CLEAR_BIT(x, mask) ((x) &= (~(mask)))
+
+extern struct memkind_ops MEMKIND_HBW_GBTLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_PREFERRED_GBTLB_OPS;
+extern struct memkind_ops MEMKIND_GBTLB_OPS;
+
+static struct memkind MEMKIND_DEFAULT_STATIC = {
+    .ops =  &MEMKIND_DEFAULT_OPS,
+    .partition = MEMKIND_PARTITION_DEFAULT,
+    .name = "memkind_default",
+    .init_once = PTHREAD_ONCE_INIT,
+    .arena_zero = 0,
+    .arena_map_len = ARENA_LIMIT_DEFAULT_KIND - 1,
+};
+
+static struct memkind MEMKIND_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HUGETLB,
+    .name = "memkind_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_INTERLEAVE_STATIC = {
+    .ops = &MEMKIND_INTERLEAVE_OPS,
+    .partition = MEMKIND_PARTITION_INTERLEAVE,
+    .name = "memkind_interleave",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_STATIC = {
+    .ops = &MEMKIND_HBW_OPS,
+    .partition = MEMKIND_PARTITION_HBW,
+    .name = "memkind_hbw",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_ALL_STATIC = {
+    .ops = &MEMKIND_HBW_ALL_OPS,
+    .partition = MEMKIND_PARTITION_HBW_ALL,
+    .name = "memkind_hbw_all",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_PREFERRED_STATIC = {
+    .ops = &MEMKIND_HBW_PREFERRED_OPS,
+    .partition = MEMKIND_PARTITION_HBW_PREFERRED,
+    .name = "memkind_hbw_preferred",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HBW_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_HUGETLB,
+    .name = "memkind_hbw_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_ALL_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HBW_ALL_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_ALL_HUGETLB,
+    .name = "memkind_hbw_all_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_PREFERRED_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HBW_PREFERRED_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_PREFERRED_HUGETLB,
+    .name = "memkind_hbw_preferred_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_GBTLB_STATIC = {
+    .ops = &MEMKIND_HBW_GBTLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_GBTLB,
+    .name = "memkind_hbw_gbtlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_PREFERRED_GBTLB_STATIC = {
+    .ops = &MEMKIND_HBW_PREFERRED_GBTLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_PREFERRED_GBTLB,
+    .name = "memkind_hbw_preferred_gbtlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_GBTLB_STATIC = {
+    .ops = &MEMKIND_GBTLB_OPS,
+    .partition = MEMKIND_PARTITION_GBTLB,
+    .name = "memkind_gbtlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_INTERLEAVE_STATIC = {
+    .ops = &MEMKIND_HBW_INTERLEAVE_OPS,
+    .partition = MEMKIND_PARTITION_HBW_INTERLEAVE,
+    .name = "memkind_hbw_interleave",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_REGULAR_STATIC = {
+    .ops = &MEMKIND_REGULAR_OPS,
+    .partition = MEMKIND_PARTITION_REGULAR,
+    .name = "memkind_regular",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_DAX_KMEM_STATIC = {
+    .ops = &MEMKIND_DAX_KMEM_OPS,
+    .partition = MEMKIND_PARTITION_DAX_KMEM,
+    .name = "memkind_dax_kmem",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_DAX_KMEM_ALL_STATIC = {
+    .ops = &MEMKIND_DAX_KMEM_ALL_OPS,
+    .partition = MEMKIND_PARTITION_DAX_KMEM_ALL,
+    .name = "memkind_dax_kmem_all",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_DAX_KMEM_PREFERRED_STATIC = {
+    .ops = &MEMKIND_DAX_KMEM_PREFERRED_OPS,
+    .partition = MEMKIND_PARTITION_DAX_KMEM_PREFERRED,
+    .name = "memkind_dax_kmem_preferred",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+MEMKIND_EXPORT struct memkind *MEMKIND_DEFAULT = &MEMKIND_DEFAULT_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HUGETLB = &MEMKIND_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_INTERLEAVE = &MEMKIND_INTERLEAVE_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW = &MEMKIND_HBW_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_ALL = &MEMKIND_HBW_ALL_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_PREFERRED =
+        &MEMKIND_HBW_PREFERRED_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_HUGETLB =
+        &MEMKIND_HBW_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_ALL_HUGETLB =
+        &MEMKIND_HBW_ALL_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_PREFERRED_HUGETLB =
+        &MEMKIND_HBW_PREFERRED_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_GBTLB = &MEMKIND_HBW_GBTLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_PREFERRED_GBTLB =
+        &MEMKIND_HBW_PREFERRED_GBTLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_GBTLB = &MEMKIND_GBTLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_INTERLEAVE =
+        &MEMKIND_HBW_INTERLEAVE_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_REGULAR = &MEMKIND_REGULAR_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_DAX_KMEM = &MEMKIND_DAX_KMEM_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_DAX_KMEM_ALL =
+        &MEMKIND_DAX_KMEM_ALL_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_DAX_KMEM_PREFERRED =
+        &MEMKIND_DAX_KMEM_PREFERRED_STATIC;
+
+struct memkind_registry {
+    struct memkind *partition_map[MEMKIND_MAX_KIND];
+    int num_kind;
+    pthread_mutex_t lock;
+};
+
+static struct memkind_registry memkind_registry_g = {
+    {
+        [MEMKIND_PARTITION_DEFAULT] = &MEMKIND_DEFAULT_STATIC,
+        [MEMKIND_PARTITION_HBW] = &MEMKIND_HBW_STATIC,
+        [MEMKIND_PARTITION_HBW_PREFERRED] = &MEMKIND_HBW_PREFERRED_STATIC,
+        [MEMKIND_PARTITION_HBW_HUGETLB] = &MEMKIND_HBW_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_HBW_PREFERRED_HUGETLB] = &MEMKIND_HBW_PREFERRED_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_HUGETLB] = &MEMKIND_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_HBW_GBTLB] = &MEMKIND_HBW_GBTLB_STATIC,
+        [MEMKIND_PARTITION_HBW_PREFERRED_GBTLB] = &MEMKIND_HBW_PREFERRED_GBTLB_STATIC,
+        [MEMKIND_PARTITION_GBTLB] = &MEMKIND_GBTLB_STATIC,
+        [MEMKIND_PARTITION_HBW_INTERLEAVE] = &MEMKIND_HBW_INTERLEAVE_STATIC,
+        [MEMKIND_PARTITION_INTERLEAVE] = &MEMKIND_INTERLEAVE_STATIC,
+        [MEMKIND_PARTITION_REGULAR] = &MEMKIND_REGULAR_STATIC,
+        [MEMKIND_PARTITION_HBW_ALL] = &MEMKIND_HBW_ALL_STATIC,
+        [MEMKIND_PARTITION_HBW_ALL_HUGETLB] = &MEMKIND_HBW_ALL_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_DAX_KMEM] = &MEMKIND_DAX_KMEM_STATIC,
+        [MEMKIND_PARTITION_DAX_KMEM_ALL] = &MEMKIND_DAX_KMEM_ALL_STATIC,
+        [MEMKIND_PARTITION_DAX_KMEM_PREFERRED] = &MEMKIND_DAX_KMEM_PREFERRED_STATIC,
+    },
+    MEMKIND_NUM_BASE_KIND,
+    PTHREAD_MUTEX_INITIALIZER
+};
+
+void *kind_mmap(struct memkind *kind, void *addr, size_t size)
+{
+    if (MEMKIND_LIKELY(kind->ops->mmap == NULL)) {
+        return memkind_default_mmap(kind, addr, size);
+    } else {
+        return kind->ops->mmap(kind, addr, size);
+    }
+}
+
+
+static int validate_memtype_bits(memkind_memtype_t memtype)
+{
+    if(memtype == 0) return -1;
+
+    CLEAR_BIT(memtype, MEMKIND_MEMTYPE_DEFAULT);
+    CLEAR_BIT(memtype, MEMKIND_MEMTYPE_HIGH_BANDWIDTH);
+
+    if(memtype != 0) return -1;
+    return 0;
+}
+
+static int validate_flags_bits(memkind_bits_t flags)
+{
+    CLEAR_BIT(flags, MEMKIND_MASK_PAGE_SIZE_2MB);
+
+    if(flags != 0) return -1;
+    return 0;
+}
+
+static int validate_policy(memkind_policy_t policy)
+{
+    if((policy >= 0) && (policy < MEMKIND_POLICY_MAX_VALUE)) return 0;
+    return -1;
+}
+
+struct create_args {
+    memkind_t kind;
+    memkind_policy_t policy;
+    memkind_bits_t flags;
+    memkind_memtype_t memtype_flags;
+};
+
+static struct create_args supported_args[] = {
+
+    {&MEMKIND_HBW_STATIC,                    MEMKIND_POLICY_BIND_LOCAL,      0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_HUGETLB_STATIC,            MEMKIND_POLICY_BIND_LOCAL,      MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_ALL_STATIC,                MEMKIND_POLICY_BIND_ALL,        0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_ALL_HUGETLB_STATIC,        MEMKIND_POLICY_BIND_ALL,        MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_PREFERRED_STATIC,          MEMKIND_POLICY_PREFERRED_LOCAL, 0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_PREFERRED_HUGETLB_STATIC,  MEMKIND_POLICY_PREFERRED_LOCAL, MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_INTERLEAVE_STATIC,         MEMKIND_POLICY_INTERLEAVE_ALL,  0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_DEFAULT_STATIC,                MEMKIND_POLICY_PREFERRED_LOCAL, 0,                          MEMKIND_MEMTYPE_DEFAULT},
+    {&MEMKIND_HUGETLB_STATIC,                MEMKIND_POLICY_PREFERRED_LOCAL, MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_DEFAULT},
+    {&MEMKIND_INTERLEAVE_STATIC,             MEMKIND_POLICY_INTERLEAVE_ALL,  0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH | MEMKIND_MEMTYPE_DEFAULT},
+};
+
+/* Kind creation */
+MEMKIND_EXPORT int memkind_create_kind(memkind_memtype_t memtype_flags,
+                                       memkind_policy_t policy,
+                                       memkind_bits_t flags,
+                                       memkind_t *kind)
+{
+    if(validate_memtype_bits(memtype_flags) != 0) {
+        log_err("Cannot create kind: incorrect memtype_flags.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if(validate_flags_bits(flags) != 0) {
+        log_err("Cannot create kind: incorrect flags.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if(validate_policy(policy) != 0) {
+        log_err("Cannot create kind: incorrect policy.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if(kind == NULL) {
+        log_err("Cannot create kind: 'kind' is NULL pointer.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    int i, num_supported_args = sizeof(supported_args) / sizeof(struct create_args);
+    for(i = 0; i < num_supported_args; i++) {
+        if((supported_args[i].memtype_flags == memtype_flags) &&
+           (supported_args[i].policy == policy) &&
+           (supported_args[i].flags == flags)) {
+
+            if(memkind_check_available(supported_args[i].kind) == 0) {
+                *kind = supported_args[i].kind;
+                return MEMKIND_SUCCESS;
+            } else if(policy == MEMKIND_POLICY_PREFERRED_LOCAL) {
+                *kind = MEMKIND_DEFAULT;
+                return MEMKIND_SUCCESS;
+            }
+            log_err("Cannot create kind: requested memory type is not available.");
+            return MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE;
+        }
+    }
+
+    log_err("Cannot create kind: unsupported set of capabilities.");
+    return MEMKIND_ERROR_INVALID;
+}
+
+static void memkind_destroy_dynamic_kind_from_register(unsigned int i,
+                                                       memkind_t kind)
+{
+    if (i >= MEMKIND_NUM_BASE_KIND) {
+        memkind_registry_g.partition_map[i] = NULL;
+        --memkind_registry_g.num_kind;
+        free(kind);
+    }
+}
+
+/* Kind destruction. */
+MEMKIND_EXPORT int memkind_destroy_kind(memkind_t kind)
+{
+    if (pthread_mutex_lock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to acquire mutex");
+    unsigned i;
+    int err = kind->ops->destroy(kind);
+    for (i = MEMKIND_NUM_BASE_KIND; i < MEMKIND_MAX_KIND; ++i) {
+        if (memkind_registry_g.partition_map[i] &&
+            strcmp(kind->name, memkind_registry_g.partition_map[i]->name) == 0) {
+            memkind_destroy_dynamic_kind_from_register(i, kind);
+            break;
+        }
+    }
+    if (pthread_mutex_unlock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to release mutex");
+    return err;
+}
+
+MEMKIND_EXPORT memkind_t memkind_detect_kind(void *ptr)
+{
+    return heap_manager_detect_kind(ptr);
+}
+
+/* Declare weak symbols for allocator decorators */
+extern void memkind_malloc_pre(struct memkind **,
+                               size_t *) __attribute__((weak));
+extern void memkind_malloc_post(struct memkind *, size_t,
+                                void **) __attribute__((weak));
+extern void memkind_calloc_pre(struct memkind **, size_t *,
+                               size_t *) __attribute__((weak));
+extern void memkind_calloc_post(struct memkind *, size_t, size_t,
+                                void **) __attribute__((weak));
+extern void memkind_posix_memalign_pre(struct memkind **, void **, size_t *,
+                                       size_t *) __attribute__((weak));
+extern void memkind_posix_memalign_post(struct memkind *, void **, size_t,
+                                        size_t, int *) __attribute__((weak));
+extern void memkind_realloc_pre(struct memkind **, void **,
+                                size_t *) __attribute__((weak));
+extern void memkind_realloc_post(struct memkind *, void *, size_t,
+                                 void **) __attribute__((weak));
+extern void memkind_free_pre(struct memkind **, void **) __attribute__((weak));
+extern void memkind_free_post(struct memkind *, void *) __attribute__((weak));
+
+MEMKIND_EXPORT int memkind_get_version()
+{
+    return MEMKIND_VERSION_MAJOR * 1000000 + MEMKIND_VERSION_MINOR * 1000 +
+           MEMKIND_VERSION_PATCH;
+}
+
+MEMKIND_EXPORT void memkind_error_message(int err, char *msg, size_t size)
+{
+    switch (err) {
+        case MEMKIND_ERROR_UNAVAILABLE:
+            strncpy(msg, "<memkind> Requested memory kind is not available", size);
+            break;
+        case MEMKIND_ERROR_MBIND:
+            strncpy(msg, "<memkind> Call to mbind() failed", size);
+            break;
+        case MEMKIND_ERROR_MMAP:
+            strncpy(msg, "<memkind> Call to mmap() failed", size);
+            break;
+        case MEMKIND_ERROR_MALLOC:
+            strncpy(msg, "<memkind> Call to malloc() failed", size);
+            break;
+        case MEMKIND_ERROR_ENVIRON:
+            strncpy(msg, "<memkind> Error parsing environment variable (MEMKIND_*)", size);
+            break;
+        case MEMKIND_ERROR_INVALID:
+            strncpy(msg, "<memkind> Invalid input arguments to memkind routine", size);
+            break;
+        case MEMKIND_ERROR_TOOMANY:
+            snprintf(msg, size,
+                     "<memkind> Attempted to initialize more than maximum (%i) number of kinds",
+                     MEMKIND_MAX_KIND);
+            break;
+        case MEMKIND_ERROR_RUNTIME:
+            strncpy(msg, "<memkind> Unspecified run-time error", size);
+            break;
+        case EINVAL:
+            strncpy(msg,
+                    "<memkind> Alignment must be a power of two and larger than sizeof(void *)",
+                    size);
+            break;
+        case ENOMEM:
+            strncpy(msg, "<memkind> Call to jemk_mallocx() failed", size);
+            break;
+        case MEMKIND_ERROR_HUGETLB:
+            strncpy(msg, "<memkind> unable to allocate huge pages", size);
+            break;
+        case MEMKIND_ERROR_BADOPS:
+            strncpy(msg,
+                    "<memkind> memkind_ops structure is poorly formed (missing or incorrect functions)",
+                    size);
+            break;
+        case MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE:
+            strncpy(msg, "<memkind> Requested memory type is not available", size);
+            break;
+        case MEMKIND_ERROR_OPERATION_FAILED:
+            strncpy(msg, "<memkind> Operation failed", size);
+            break;
+        case MEMKIND_ERROR_ARENAS_CREATE:
+            strncpy(msg, "<memkind> Call to jemalloc's arenas.create () failed", size);
+            break;
+        default:
+            snprintf(msg, size, "<memkind> Undefined error number: %i", err);
+            break;
+    }
+    if (size > 0) {
+        msg[size-1] = '\0';
+    }
+}
+
+void memkind_init(memkind_t kind, bool check_numa)
+{
+    log_info("Initializing kind %s.", kind->name);
+    heap_manager_init(kind);
+    if (check_numa) {
+        int err = numa_available();
+        if (err) {
+            log_fatal("[%s] NUMA not available (error code:%d).", kind->name, err);
+            abort();
+        }
+    }
+}
+
+static void nop(void) {}
+
+static int memkind_create(struct memkind_ops *ops, const char *name,
+                          struct memkind **kind)
+{
+    int err;
+    unsigned i;
+    unsigned id_kind = 0;
+
+    *kind = NULL;
+    if (pthread_mutex_lock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to acquire mutex");
+
+    if (memkind_registry_g.num_kind == MEMKIND_MAX_KIND) {
+        log_err("Attempted to initialize more than maximum (%i) number of kinds.",
+                MEMKIND_MAX_KIND);
+        err = MEMKIND_ERROR_TOOMANY;
+        goto exit;
+    }
+    if (ops->create == NULL ||
+        ops->destroy == NULL ||
+        ops->malloc == NULL ||
+        ops->calloc == NULL ||
+        ops->realloc == NULL ||
+        ops->posix_memalign == NULL ||
+        ops->free == NULL ||
+        ops->init_once != NULL) {
+        err = MEMKIND_ERROR_BADOPS;
+        goto exit;
+    }
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        if (memkind_registry_g.partition_map[i] == NULL) {
+            id_kind = i;
+            break;
+        } else if (strcmp(name, memkind_registry_g.partition_map[i]->name) == 0) {
+            log_err("Kind with the name %s already exists", name);
+            err = MEMKIND_ERROR_INVALID;
+            goto exit;
+        }
+    }
+    *kind = (struct memkind *)calloc(1, sizeof(struct memkind));
+    if (!*kind) {
+        err = MEMKIND_ERROR_MALLOC;
+        log_err("calloc() failed.");
+        goto exit;
+    }
+
+    (*kind)->partition = memkind_registry_g.num_kind;
+    err = ops->create(*kind, ops, name);
+    if (err) {
+        free(*kind);
+        goto exit;
+    }
+    memkind_registry_g.partition_map[id_kind] = *kind;
+    ++memkind_registry_g.num_kind;
+
+    (*kind)->init_once = PTHREAD_ONCE_INIT;
+    pthread_once(&(*kind)->init_once,
+                 nop); //this is done to avoid init_once for dynamic kinds
+exit:
+    if (pthread_mutex_unlock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to release mutex");
+
+    return err;
+}
+
+#ifdef __GNUC__
+__attribute__((constructor))
+#endif
+static void memkind_construct(void)
+{
+    const char *env = secure_getenv("MEMKIND_HEAP_MANAGER");
+    if (env && strcmp(env, "TBB") == 0) {
+        load_tbb_symbols();
+    } else {
+        env = secure_getenv("MEMKIND_BACKGROUND_THREAD_LIMIT");
+        if (env) {
+            char *end;
+            errno = 0;
+            size_t thread_limit = strtoull(env, &end, 10);
+            if (*end != '\0' || errno != 0 ) {
+                log_err("Error on parsing value MEMKIND_BACKGROUND_THREAD_LIMIT");
+                return;
+            }
+            memkind_arena_enable_background_threads(thread_limit);
+        }
+    }
+}
+
+#ifdef __GNUC__
+__attribute__((destructor))
+#endif
+static int memkind_finalize(void)
+{
+    struct memkind *kind;
+    unsigned i;
+    int err = MEMKIND_SUCCESS;
+
+    if (pthread_mutex_lock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to acquire mutex");
+
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        kind = memkind_registry_g.partition_map[i];
+        if (kind && kind->ops->finalize) {
+            err = kind->ops->finalize(kind);
+            if (err) {
+                goto exit;
+            }
+            memkind_destroy_dynamic_kind_from_register(i, kind);
+        }
+    }
+    assert(memkind_registry_g.num_kind == MEMKIND_NUM_BASE_KIND);
+
+exit:
+    if (pthread_mutex_unlock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to release mutex");
+
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_check_available(struct memkind *kind)
+{
+    int err = MEMKIND_SUCCESS;
+
+    if (MEMKIND_LIKELY(kind->ops->check_available)) {
+        err = kind->ops->check_available(kind);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT size_t memkind_malloc_usable_size(struct memkind *kind,
+                                                 void *ptr)
+{
+    if (!kind) {
+        return heap_manager_malloc_usable_size(ptr);
+    } else {
+        return kind->ops->malloc_usable_size(kind, ptr);
+    }
+}
+
+MEMKIND_EXPORT void *memkind_malloc(struct memkind *kind, size_t size)
+{
+    void *result;
+
+    pthread_once(&kind->init_once, kind->ops->init_once);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_malloc_pre) {
+        memkind_malloc_pre(&kind, &size);
+    }
+#endif
+
+    result = kind->ops->malloc(kind, size);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_malloc_post) {
+        memkind_malloc_post(kind, size, &result);
+    }
+#endif
+
+    return result;
+}
+
+MEMKIND_EXPORT void *memkind_calloc(struct memkind *kind, size_t num,
+                                    size_t size)
+{
+    void *result;
+
+    pthread_once(&kind->init_once, kind->ops->init_once);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_calloc_pre) {
+        memkind_calloc_pre(&kind, &num, &size);
+    }
+#endif
+
+    result = kind->ops->calloc(kind, num, size);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_calloc_post) {
+        memkind_calloc_post(kind, num, size, &result);
+    }
+#endif
+
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_posix_memalign(struct memkind *kind, void **memptr,
+                                          size_t alignment,
+                                          size_t size)
+{
+    int err;
+
+    pthread_once(&kind->init_once, kind->ops->init_once);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_posix_memalign_pre) {
+        memkind_posix_memalign_pre(&kind, memptr, &alignment, &size);
+    }
+#endif
+
+    err = kind->ops->posix_memalign(kind, memptr, alignment, size);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_posix_memalign_post) {
+        memkind_posix_memalign_post(kind, memptr, alignment, size, &err);
+    }
+#endif
+
+    return err;
+}
+
+MEMKIND_EXPORT void *memkind_realloc(struct memkind *kind, void *ptr,
+                                     size_t size)
+{
+    void *result;
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_realloc_pre) {
+        memkind_realloc_pre(&kind, &ptr, &size);
+    }
+#endif
+
+    if (!kind) {
+        result = heap_manager_realloc(ptr, size);
+    } else {
+        pthread_once(&kind->init_once, kind->ops->init_once);
+        result = kind->ops->realloc(kind, ptr, size);
+    }
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_realloc_post) {
+        memkind_realloc_post(kind, ptr, size, &result);
+    }
+#endif
+
+    return result;
+}
+
+MEMKIND_EXPORT void memkind_free(struct memkind *kind, void *ptr)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_free_pre) {
+        memkind_free_pre(&kind, &ptr);
+    }
+#endif
+    if (!kind) {
+        heap_manager_free(ptr);
+    } else {
+        pthread_once(&kind->init_once, kind->ops->init_once);
+        kind->ops->free(kind, ptr);
+    }
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_free_post) {
+        memkind_free_post(kind, ptr);
+    }
+#endif
+}
+
+static int memkind_tmpfile(const char *dir, int *fd)
+{
+    static char template[] = "/memkind.XXXXXX";
+    int err = MEMKIND_SUCCESS;
+    int oerrno;
+    int dir_len = strlen(dir);
+
+    if (dir_len > PATH_MAX) {
+        log_err("Could not create temporary file: too long path.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    char fullname[dir_len + sizeof (template)];
+    (void) strcpy(fullname, dir);
+    (void) strcat(fullname, template);
+
+    sigset_t set, oldset;
+    sigfillset(&set);
+    (void) sigprocmask(SIG_BLOCK, &set, &oldset);
+
+    if ((*fd = mkstemp(fullname)) < 0) {
+        log_err("Could not create temporary file: errno=%d.", errno);
+        err = MEMKIND_ERROR_INVALID;
+        goto exit;
+    }
+
+    (void) unlink(fullname);
+    (void) sigprocmask(SIG_SETMASK, &oldset, NULL);
+
+    return err;
+
+exit:
+    oerrno = errno;
+    (void) sigprocmask(SIG_SETMASK, &oldset, NULL);
+    if (*fd != -1) {
+        (void) close(*fd);
+    }
+    *fd = -1;
+    errno = oerrno;
+    return err;
+}
+
+MEMKIND_EXPORT struct memkind_config *memkind_config_new(void)
+{
+    struct memkind_config *cfg = (struct memkind_config *)malloc(sizeof(
+                                                                     struct memkind_config));
+    return cfg;
+}
+
+MEMKIND_EXPORT void memkind_config_delete(struct memkind_config *cfg)
+{
+    free(cfg);
+}
+
+MEMKIND_EXPORT void memkind_config_set_path(struct memkind_config *cfg,
+                                            const char *pmem_dir)
+{
+    cfg->pmem_dir = pmem_dir;
+}
+
+MEMKIND_EXPORT void memkind_config_set_size(struct memkind_config *cfg,
+                                            size_t pmem_size)
+{
+    cfg->pmem_size = pmem_size;
+}
+
+MEMKIND_EXPORT void memkind_config_set_memory_usage_policy(
+    struct memkind_config *cfg, memkind_mem_usage_policy policy)
+{
+    cfg->policy = policy;
+}
+
+MEMKIND_EXPORT int memkind_create_pmem(const char *dir, size_t max_size,
+                                       struct memkind **kind)
+{
+    int oerrno;
+
+    if (max_size && max_size < MEMKIND_PMEM_MIN_SIZE) {
+        log_err("Cannot create pmem: invalid size.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if (max_size) {
+        /* round up to a multiple of jemalloc chunk size */
+        max_size = roundup(max_size, MEMKIND_PMEM_CHUNK_SIZE);
+    }
+
+    int fd = -1;
+    char name[16];
+
+    int err = memkind_tmpfile(dir, &fd);
+    if (err) {
+        goto exit;
+    }
+
+    snprintf(name, sizeof (name), "pmem%08x", fd);
+
+    err = memkind_create(&MEMKIND_PMEM_OPS, name, kind);
+    if (err) {
+        goto exit;
+    }
+
+    struct memkind_pmem *priv = (*kind)->priv;
+
+    priv->fd = fd;
+    priv->offset = 0;
+    priv->current_size = 0;
+    priv->max_size = max_size;
+
+    return err;
+
+exit:
+    oerrno = errno;
+    if (fd != -1) {
+        (void) close(fd);
+    }
+    errno = oerrno;
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_create_pmem_with_config(struct memkind_config *cfg,
+                                                   struct memkind **kind)
+{
+    int status = memkind_create_pmem(cfg->pmem_dir, cfg->pmem_size, kind);
+    if (MEMKIND_LIKELY(!status)) {
+        status = (*kind)->ops->update_memory_usage_policy(*kind, cfg->policy);
+    }
+
+    return status;
+}
+
+
+static int memkind_get_kind_by_partition_internal(int partition,
+                                                  struct memkind **kind)
+{
+    int err = MEMKIND_SUCCESS;
+
+    if (MEMKIND_LIKELY(partition >= 0 &&
+                       partition < MEMKIND_MAX_KIND &&
+                       memkind_registry_g.partition_map[partition] != NULL)) {
+        *kind = memkind_registry_g.partition_map[partition];
+    } else {
+        *kind = NULL;
+        err = MEMKIND_ERROR_UNAVAILABLE;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_get_kind_by_partition(int partition,
+                                                 struct memkind **kind)
+{
+    return memkind_get_kind_by_partition_internal(partition, kind);
+}
+
+MEMKIND_EXPORT int memkind_update_cached_stats(void)
+{
+    return heap_manager_update_cached_stats();
+}
+
+MEMKIND_EXPORT void *memkind_defrag_reallocate(memkind_t kind, void *ptr)
+{
+    if (!kind) {
+        return heap_manager_defrag_reallocate(ptr);
+    } else {
+        return kind->ops->defrag_reallocate(kind, ptr);
+    }
+}
+
+MEMKIND_EXPORT int memkind_get_stat(memkind_t kind, memkind_stat_type stat,
+                                    size_t *value)
+{
+    if (MEMKIND_UNLIKELY(stat >= MEMKIND_STAT_TYPE_MAX_VALUE)) {
+        log_err("Unrecognized type of memory statistic %d", stat);
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if (!kind) {
+        return heap_manager_get_stat(stat, value);
+    } else {
+        return kind->ops->get_stat(kind, stat, value);
+    }
+}
diff --git a/memkind-1.10.0/src/memkind_arena.c b/memkind-1.10.0/src/memkind_arena.c
new file mode 100644
index 0000000..8b525cc
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_arena.c
@@ -0,0 +1,938 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <limits.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <pthread.h>
+#include <errno.h>
+#include <numa.h>
+#include <numaif.h>
+#include <jemalloc/jemalloc.h>
+#include <utmpx.h>
+#include <sched.h>
+#include <limits.h>
+#include <sys/mman.h>
+#include <sys/param.h>
+#include <assert.h>
+
+#include "config.h"
+
+#define HUGE_PAGE_SIZE (1ull << MEMKIND_MASK_PAGE_SIZE_2MB)
+#define PAGE_2_BYTES(x) ((x) << 12)
+
+static const char *const global_stats[MEMKIND_STAT_TYPE_MAX_VALUE] = {
+    "stats.resident",
+    "stats.active",
+    "stats.allocated"
+};
+
+#define ARENA_STAT_MAX 2
+
+struct stats_arena {
+    const char *stats[ARENA_STAT_MAX];
+    unsigned stats_no;
+};
+
+static const struct stats_arena arena_stats [MEMKIND_STAT_TYPE_MAX_VALUE] = {
+    { .stats = {"stats.arenas.%u.resident", NULL}, .stats_no = 1},
+    { .stats = {"stats.arenas.%u.pactive", NULL}, .stats_no = 1},
+    { .stats = {"stats.arenas.%u.small.allocated", "stats.arenas.%u.large.allocated"}, .stats_no = 2},
+};
+
+static void *jemk_mallocx_check(size_t size, int flags);
+static void *jemk_rallocx_check(void *ptr, size_t size, int flags);
+static void tcache_finalize(void *args);
+
+static unsigned integer_log2(unsigned v)
+{
+    return (sizeof(unsigned) * 8) - (__builtin_clz(v) + 1);
+}
+
+static unsigned round_pow2_up(unsigned v)
+{
+    unsigned v_log2 = integer_log2(v);
+
+    if (v != 1 << v_log2) {
+        v = 1 << (v_log2 + 1);
+    }
+    return v;
+}
+
+MEMKIND_EXPORT int memkind_set_arena_map_len(struct memkind *kind)
+{
+    if (kind->ops->get_arena == memkind_bijective_get_arena) {
+        kind->arena_map_len = 1;
+    } else if (kind->ops->get_arena == memkind_thread_get_arena) {
+        char *arena_num_env = secure_getenv("MEMKIND_ARENA_NUM_PER_KIND");
+
+        if (arena_num_env) {
+            unsigned long int arena_num_value = strtoul(arena_num_env, NULL, 10);
+
+            if ((arena_num_value == 0) || (arena_num_value > INT_MAX)) {
+                log_err("Wrong MEMKIND_ARENA_NUM_PER_KIND environment value: %lu.",
+                        arena_num_value);
+                return MEMKIND_ERROR_ENVIRON;
+            }
+
+            kind->arena_map_len = arena_num_value;
+        } else {
+            int calculated_arena_num = numa_num_configured_cpus() * 4;
+
+#if ARENA_LIMIT_PER_KIND != 0
+            calculated_arena_num = MIN((ARENA_LIMIT_PER_KIND), calculated_arena_num);
+#endif
+            kind->arena_map_len = calculated_arena_num;
+        }
+
+        kind->arena_map_len = round_pow2_up(kind->arena_map_len);
+    }
+
+    kind->arena_map_mask = kind->arena_map_len - 1;
+    return 0;
+}
+
+static pthread_once_t arena_config_once = PTHREAD_ONCE_INIT;
+static int arena_init_status;
+
+static pthread_key_t tcache_key;
+static bool memkind_hog_memory;
+
+static void arena_config_init()
+{
+    const char *str = secure_getenv("MEMKIND_HOG_MEMORY");
+    memkind_hog_memory = str && str[0] == '1';
+
+    arena_init_status = pthread_key_create(&tcache_key, tcache_finalize);
+}
+
+#define MALLOCX_ARENA_MAX           0xffe        // copy-pasted from jemalloc/internal/jemalloc_internal.h
+#define DIRTY_DECAY_MS_DEFAULT      10000        // copy-pasted from jemalloc/internal/arena_types.h DIRTY_DECAY_MS_DEFAULT
+#define DIRTY_DECAY_MS_CONSERVATIVE     0
+
+static struct memkind *arena_registry_g[MALLOCX_ARENA_MAX];
+static pthread_mutex_t arena_registry_write_lock;
+
+struct memkind *get_kind_by_arena(unsigned arena_ind)
+{
+    // there is no way to obtain MALLOCX_ARENA_MAX from jemalloc
+    // so this checks if arena_ind does not exceed assumed range
+    assert(arena_ind < MALLOCX_ARENA_MAX);
+
+    return arena_registry_g[arena_ind];
+}
+
+// Allocates size bytes aligned to alignment. Returns NULL if allocation fails.
+static void *alloc_aligned_slow(size_t size, size_t alignment,
+                                struct memkind *kind)
+{
+    size_t extended_size = size + alignment;
+    void *ptr;
+
+    ptr = kind_mmap(kind,  NULL, extended_size);
+
+    if(ptr == MAP_FAILED) {
+        return NULL;
+    }
+
+    uintptr_t addr = (uintptr_t)ptr;
+    uintptr_t aligned_addr = (addr + alignment) & ~(alignment - 1);
+
+    size_t head_len = aligned_addr - addr;
+    if (head_len > 0) {
+        munmap(ptr, head_len);
+    }
+
+    uintptr_t tail = aligned_addr + size;
+    size_t tail_len = (addr + extended_size) - (aligned_addr + size);
+    if (tail_len > 0) {
+        munmap((void *)tail, tail_len);
+    }
+
+    return (void *)aligned_addr;
+}
+
+
+void *arena_extent_alloc(extent_hooks_t *extent_hooks,
+                         void *new_addr,
+                         size_t size,
+                         size_t alignment,
+                         bool *zero,
+                         bool *commit,
+                         unsigned arena_ind)
+{
+    struct memkind *kind = get_kind_by_arena(arena_ind);
+
+    int err = memkind_check_available(kind);
+    if (err) {
+        return NULL;
+    }
+
+    void *addr = kind_mmap(kind, new_addr, size);
+    if (addr == MAP_FAILED) {
+        return NULL;
+    }
+
+    if (new_addr != NULL && addr != new_addr) {
+        /* wrong place */
+        munmap(addr, size);
+        return NULL;
+    }
+
+    if ((uintptr_t)addr & (alignment-1)) {
+        munmap(addr, size);
+        addr = alloc_aligned_slow(size, alignment, kind);
+        if(addr == NULL) {
+            return NULL;
+        }
+    }
+
+    *zero = true;
+    *commit = true;
+
+    return addr;
+}
+
+void *arena_extent_alloc_hugetlb(extent_hooks_t *extent_hooks,
+                                 void *new_addr,
+                                 size_t size,
+                                 size_t alignment,
+                                 bool *zero,
+                                 bool *commit,
+                                 unsigned arena_ind)
+{
+    //round up to huge page size
+    size = (size + (HUGE_PAGE_SIZE - 1)) & ~(HUGE_PAGE_SIZE - 1);
+    return arena_extent_alloc(extent_hooks, new_addr, size, alignment, zero, commit,
+                              arena_ind);
+}
+
+bool arena_extent_dalloc(extent_hooks_t *extent_hooks,
+                         void *addr,
+                         size_t size,
+                         bool committed,
+                         unsigned arena_ind)
+{
+    return true;
+}
+
+bool arena_extent_commit(extent_hooks_t *extent_hooks,
+                         void *addr,
+                         size_t size,
+                         size_t offset,
+                         size_t length,
+                         unsigned arena_ind)
+{
+    return false;
+}
+
+bool arena_extent_decommit(extent_hooks_t *extent_hooks,
+                           void *addr,
+                           size_t size,
+                           size_t offset,
+                           size_t length,
+                           unsigned arena_ind)
+{
+    return true;
+}
+
+bool arena_extent_purge(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        size_t offset,
+                        size_t length,
+                        unsigned arena_ind)
+{
+    if (memkind_hog_memory) {
+        return true;
+    }
+
+    int err = madvise(addr + offset, length, MADV_DONTNEED);
+    return (err != 0);
+}
+
+bool arena_extent_split(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        size_t size_a,
+                        size_t size_b,
+                        bool committed,
+                        unsigned arena_ind)
+{
+    return false;
+}
+
+bool arena_extent_merge(extent_hooks_t *extent_hooks,
+                        void *addr_a,
+                        size_t size_a,
+                        void *addr_b,
+                        size_t size_b,
+                        bool committed,
+                        unsigned arena_ind)
+{
+    return false;
+}
+
+extent_hooks_t arena_extent_hooks = {
+    .alloc = arena_extent_alloc,
+    .dalloc = arena_extent_dalloc,
+    .commit = arena_extent_commit,
+    .decommit = arena_extent_decommit,
+    .purge_lazy = arena_extent_purge,
+    .split = arena_extent_split,
+    .merge = arena_extent_merge
+};
+
+extent_hooks_t arena_extent_hooks_hugetlb = {
+    .alloc = arena_extent_alloc_hugetlb,
+    .dalloc = arena_extent_dalloc,
+    .commit = arena_extent_commit,
+    .decommit = arena_extent_decommit,
+    .purge_lazy = arena_extent_purge,
+    .split = arena_extent_split,
+    .merge = arena_extent_merge
+};
+
+extent_hooks_t *get_extent_hooks_by_kind(struct memkind *kind)
+{
+    if (kind == MEMKIND_HUGETLB
+        || kind == MEMKIND_HBW_HUGETLB
+        || kind == MEMKIND_HBW_ALL_HUGETLB
+        || kind == MEMKIND_HBW_PREFERRED_HUGETLB) {
+        return &arena_extent_hooks_hugetlb;
+    } else {
+        return &arena_extent_hooks;
+    }
+}
+
+MEMKIND_EXPORT int memkind_arena_create_map(struct memkind *kind,
+                                            extent_hooks_t *hooks)
+{
+    int err;
+
+    pthread_once(&arena_config_once, arena_config_init);
+    if(arena_init_status) {
+        return arena_init_status;
+    }
+
+    err = memkind_set_arena_map_len(kind);
+    if(err) {
+        return err;
+    }
+#ifdef MEMKIND_TLS
+    if (kind->ops->get_arena == memkind_thread_get_arena) {
+        pthread_key_create(&(kind->arena_key), free);
+    }
+#endif
+
+    if (pthread_mutex_lock(&arena_registry_write_lock) != 0)
+        assert(0 && "failed to acquire mutex");
+    unsigned i;
+    size_t unsigned_size = sizeof(unsigned int);
+    kind->arena_zero = UINT_MAX;
+    for(i = 0; i<kind->arena_map_len; i++) {
+        unsigned arena_index;
+        //create new arena with consecutive index
+        err = jemk_mallctl("arenas.create", (void *)&arena_index, &unsigned_size, NULL,
+                           0);
+        if(err) {
+            log_err("Could not create arena.");
+            err = MEMKIND_ERROR_ARENAS_CREATE;
+            goto exit;
+        }
+        //store arena with lowest index (arenas could be created in descending/ascending order)
+        if(kind->arena_zero > arena_index) {
+            kind->arena_zero = arena_index;
+        }
+        //setup extent_hooks for newly created arena
+        char cmd[64];
+        snprintf(cmd, sizeof(cmd), "arena.%u.extent_hooks", arena_index);
+        err = jemk_mallctl(cmd, NULL, NULL, (void *)&hooks, sizeof(extent_hooks_t *));
+        if(err) {
+            goto exit;
+        }
+        arena_registry_g[arena_index] = kind;
+    }
+
+exit:
+    if (pthread_mutex_unlock(&arena_registry_write_lock) != 0)
+        assert(0 && "failed to release mutex");
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_arena_create(struct memkind *kind,
+                                        struct memkind_ops *ops, const char *name)
+{
+    int err = memkind_default_create(kind, ops, name);
+    if (!err) {
+        err = memkind_arena_create_map(kind, get_extent_hooks_by_kind(kind));
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_arena_destroy(struct memkind *kind)
+{
+    if (kind->arena_map_len) {
+        char cmd[128];
+        unsigned i;
+
+        if (pthread_mutex_lock(&arena_registry_write_lock) != 0)
+            assert(0 && "failed to acquire mutex");
+
+        for (i = 0; i < kind->arena_map_len; ++i) {
+            snprintf(cmd, 128, "arena.%u.destroy", kind->arena_zero + i);
+            jemk_mallctl(cmd, NULL, NULL, NULL, 0);
+            arena_registry_g[kind->arena_zero + i] = NULL;
+        }
+
+        if (pthread_mutex_unlock(&arena_registry_write_lock) != 0)
+            assert(0 && "failed to release mutex");
+
+#ifdef MEMKIND_TLS
+        if (kind->ops->get_arena == memkind_thread_get_arena) {
+            pthread_key_delete(kind->arena_key);
+        }
+#endif
+    }
+
+    memkind_default_destroy(kind);
+    return 0;
+}
+
+int memkind_arena_finalize(struct memkind *kind)
+{
+    return memkind_arena_destroy(kind);
+}
+
+// max allocation size to be cached by tcache mechanism
+#define TCACHE_MAX (1<<(JEMALLOC_TCACHE_CLASS))
+
+static void tcache_finalize(void *args)
+{
+    int i;
+    unsigned *tcache_map = args;
+    for(i = 0; i<MEMKIND_NUM_BASE_KIND; i++) {
+        if(tcache_map[i] != 0) {
+            jemk_mallctl("tcache.destroy", NULL, NULL, (void *)&tcache_map[i],
+                         sizeof(unsigned));
+        }
+    }
+}
+
+static inline int memkind_lookup_arena(void *ptr, unsigned int *arena)
+{
+    size_t sz = sizeof(unsigned);
+    unsigned temp_arena;
+    int err = jemk_mallctl("arenas.lookup", &temp_arena, &sz, &ptr, sizeof(ptr));
+
+    if (err) {
+        log_err("Could not found arena, err=%d", err);
+        return 1;
+    }
+
+    *arena = temp_arena;
+    return 0;
+}
+
+MEMKIND_EXPORT struct memkind *memkind_arena_detect_kind(void *ptr)
+{
+    if (!ptr) {
+        return NULL;
+    }
+    struct memkind *kind = NULL;
+    unsigned arena;
+    int err = memkind_lookup_arena(ptr, &arena);
+    if (MEMKIND_LIKELY(!err)) {
+        kind = get_kind_by_arena(arena);
+    }
+
+    /* if no kind was associated with arena it means that allocation doesn't come from
+       jemk_*allocx API - it is jemk_*alloc API (MEMKIND_DEFAULT) */
+
+    return (kind) ? kind : MEMKIND_DEFAULT;
+}
+
+static inline int get_tcache_flag(unsigned partition, size_t size)
+{
+
+    // do not cache allocation larger than tcache_max nor those coming from non-static kinds
+    if(size > TCACHE_MAX || partition >= MEMKIND_NUM_BASE_KIND) {
+        return MALLOCX_TCACHE_NONE;
+    }
+
+    unsigned *tcache_map = pthread_getspecific(tcache_key);
+    if(tcache_map == NULL) {
+        tcache_map = calloc(MEMKIND_NUM_BASE_KIND, sizeof(unsigned));
+        if(tcache_map == NULL) {
+            return MALLOCX_TCACHE_NONE;
+        }
+        pthread_setspecific(tcache_key, (void *)tcache_map);
+    }
+
+    if(MEMKIND_UNLIKELY(tcache_map[partition] == 0)) {
+        size_t unsigned_size = sizeof(unsigned);
+        int err = jemk_mallctl("tcache.create", (void *)&tcache_map[partition],
+                               &unsigned_size, NULL, 0);
+        if(err) {
+            log_err("Could not acquire tcache, err=%d", err);
+            return MALLOCX_TCACHE_NONE;
+        }
+    }
+    return MALLOCX_TCACHE(tcache_map[partition]);
+}
+
+MEMKIND_EXPORT void *memkind_arena_malloc(struct memkind *kind, size_t size)
+{
+    void *result = NULL;
+    unsigned arena;
+
+    int err = kind->ops->get_arena(kind, &arena, size);
+    if (MEMKIND_LIKELY(!err)) {
+        result = jemk_mallocx_check(size,
+                                    MALLOCX_ARENA(arena) | get_tcache_flag(kind->partition, size));
+    }
+    return result;
+}
+
+static void *memkind_arena_malloc_no_tcache(struct memkind *kind, size_t size)
+{
+    void *result = NULL;
+    if (kind == MEMKIND_DEFAULT) {
+        result = jemk_mallocx_check(size, MALLOCX_TCACHE_NONE);
+    } else {
+        unsigned arena;
+        int err = kind->ops->get_arena(kind, &arena, size);
+        if (MEMKIND_LIKELY(!err)) {
+            result = jemk_mallocx_check(size, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE);
+        }
+    }
+    return result;
+}
+
+MEMKIND_EXPORT void memkind_arena_free(struct memkind *kind, void *ptr)
+{
+    if (kind == MEMKIND_DEFAULT) {
+        jemk_free(ptr);
+    } else if (ptr != NULL) {
+        jemk_dallocx(ptr, get_tcache_flag(kind->partition, 0));
+    }
+}
+
+MEMKIND_EXPORT void memkind_arena_free_with_kind_detect(void *ptr)
+{
+    struct memkind *kind = memkind_arena_detect_kind(ptr);
+
+    memkind_arena_free(kind, ptr);
+}
+
+MEMKIND_EXPORT size_t memkind_arena_malloc_usable_size(void *ptr)
+{
+    return memkind_default_malloc_usable_size(NULL, ptr);
+}
+
+MEMKIND_EXPORT void *memkind_arena_realloc(struct memkind *kind, void *ptr,
+                                           size_t size)
+{
+    unsigned arena;
+
+    if (size == 0 && ptr != NULL) {
+        memkind_arena_free(kind, ptr);
+        ptr = NULL;
+    } else {
+        int err = kind->ops->get_arena(kind, &arena, size);
+        if (MEMKIND_LIKELY(!err)) {
+            if (ptr == NULL) {
+                ptr = jemk_mallocx_check(size,
+                                         MALLOCX_ARENA(arena) | get_tcache_flag(kind->partition, size));
+            } else {
+                ptr = jemk_rallocx_check(ptr, size,
+                                         MALLOCX_ARENA(arena) | get_tcache_flag(kind->partition, size));
+            }
+        }
+    }
+    return ptr;
+}
+
+int memkind_arena_update_cached_stats(void)
+{
+    uint64_t epoch = 1;
+    return jemk_mallctl("epoch", NULL, NULL, &epoch, sizeof(epoch));
+}
+
+MEMKIND_EXPORT void *memkind_arena_realloc_with_kind_detect(void *ptr,
+                                                            size_t size)
+{
+    if (!ptr) {
+        errno = EINVAL;
+        return NULL;
+    }
+    struct memkind *kind = memkind_arena_detect_kind(ptr);
+    if (kind == MEMKIND_DEFAULT) {
+        return memkind_default_realloc(kind, ptr, size);
+    } else {
+        return memkind_arena_realloc(kind, ptr, size);
+    }
+}
+
+MEMKIND_EXPORT int memkind_arena_update_memory_usage_policy(
+    struct memkind *kind, memkind_mem_usage_policy policy)
+{
+    int err = MEMKIND_SUCCESS;
+    unsigned i;
+    ssize_t dirty_decay_val = 0;
+    switch ( policy ) {
+        case MEMKIND_MEM_USAGE_POLICY_DEFAULT:
+            dirty_decay_val = DIRTY_DECAY_MS_DEFAULT;
+            break;
+        case MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE:
+            dirty_decay_val = DIRTY_DECAY_MS_CONSERVATIVE;
+            break;
+        default:
+            log_err("Unrecognized memory policy %d", policy);
+            return MEMKIND_ERROR_INVALID;
+    }
+
+    for (i = 0; i < kind->arena_map_len; ++i) {
+        char cmd[64];
+
+        snprintf(cmd, sizeof(cmd), "arena.%u.dirty_decay_ms", kind->arena_zero + i);
+        err = jemk_mallctl(cmd, NULL, NULL, (void *)&dirty_decay_val, sizeof(ssize_t));
+        if ( err ) {
+            log_err("Incorrect dirty_decay_ms value %zu", dirty_decay_val);
+            return MEMKIND_ERROR_INVALID;
+        }
+    }
+
+    return err;
+}
+
+MEMKIND_EXPORT void *memkind_arena_calloc(struct memkind *kind, size_t num,
+                                          size_t size)
+{
+    void *result = NULL;
+    unsigned arena;
+
+    int err = kind->ops->get_arena(kind, &arena, size);
+    if (MEMKIND_LIKELY(!err)) {
+        result = jemk_mallocx_check(num * size,
+                                    MALLOCX_ARENA(arena) | MALLOCX_ZERO | get_tcache_flag(kind->partition, size));
+    }
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_arena_posix_memalign(struct memkind *kind,
+                                                void **memptr, size_t alignment,
+                                                size_t size)
+{
+    int err;
+    unsigned arena;
+
+    *memptr = NULL;
+    err = kind->ops->get_arena(kind, &arena, size);
+    if (MEMKIND_LIKELY(!err)) {
+        err = memkind_posix_check_alignment(kind, alignment);
+    }
+    if (MEMKIND_LIKELY(!err)) {
+        if (MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+            return 0;
+        }
+        /* posix_memalign should not change errno.
+           Set it to its previous value after calling jemalloc */
+        int errno_before = errno;
+        *memptr = jemk_mallocx_check(size,
+                                     MALLOCX_ALIGN(alignment) | MALLOCX_ARENA(arena) | get_tcache_flag(
+                                         kind->partition, size));
+        errno = errno_before;
+        err = *memptr ? 0 : ENOMEM;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_bijective_get_arena(struct memkind *kind,
+                                               unsigned int *arena, size_t size)
+{
+    *arena = kind->arena_zero;
+    return 0;
+}
+
+#ifdef MEMKIND_TLS
+MEMKIND_EXPORT int memkind_thread_get_arena(struct memkind *kind,
+                                            unsigned int *arena, size_t size)
+{
+    int err = 0;
+    unsigned int *arena_tsd;
+    arena_tsd = pthread_getspecific(kind->arena_key);
+
+    if (MEMKIND_UNLIKELY(arena_tsd == NULL)) {
+        arena_tsd = malloc(sizeof(unsigned int));
+        if (arena_tsd == NULL) {
+            err = MEMKIND_ERROR_MALLOC;
+            log_err("malloc() failed.");
+        }
+        if (!err) {
+            // On glibc pthread_self() is incremented by 0x801000 for every
+            // thread (no matter the arch's word width).  This might change
+            // in the future, but even in the worst case the hash will
+            // degenerate to a single bucket with no loss of correctness.
+            *arena_tsd = ((uint64_t)pthread_self() >> 12) %
+                         kind->arena_map_len;
+            err = pthread_setspecific(kind->arena_key, arena_tsd) ?
+                  MEMKIND_ERROR_RUNTIME : 0;
+        }
+    }
+    *arena = kind->arena_zero + *arena_tsd;
+    return err;
+}
+
+#else
+
+/*
+ *
+ * We use thread control block as unique thread identifier
+ * For more read: https://www.akkadia.org/drepper/tls.pdf
+ * We could consider using rdfsbase when it will arrive to linux kernel
+ *
+ * This approach works only on glibc (and possibly similar implementations)
+ * but that covers our current needs.
+ *
+ */
+static uintptr_t get_fs_base()
+{
+    return (uintptr_t)pthread_self();
+}
+
+MEMKIND_EXPORT int memkind_thread_get_arena(struct memkind *kind,
+                                            unsigned int *arena, size_t size)
+{
+    unsigned int arena_idx;
+    // it's likely that each thread control block lies on different page
+    // so we extracting page number with >> 12 to improve hashing
+    arena_idx = (get_fs_base() >> 12) & kind->arena_map_mask;
+    *arena = kind->arena_zero + arena_idx;
+    return 0;
+}
+#endif //MEMKIND_TLS
+
+static void *jemk_mallocx_check(size_t size, int flags)
+{
+    /*
+     * Checking for out of range size due to unhandled error in
+     * jemk_mallocx().  Size invalid for the range
+     * LLONG_MAX <= size <= ULLONG_MAX
+     * which is the result of passing a negative signed number as size
+     */
+    void *result = NULL;
+
+    if (MEMKIND_UNLIKELY(size >= LLONG_MAX)) {
+        errno = ENOMEM;
+    } else if (size != 0) {
+        result = jemk_mallocx(size, flags);
+    }
+    return result;
+}
+
+static void *jemk_rallocx_check(void *ptr, size_t size, int flags)
+{
+    /*
+     * Checking for out of range size due to unhandled error in
+     * jemk_mallocx().  Size invalid for the range
+     * LLONG_MAX <= size <= ULLONG_MAX
+     * which is the result of passing a negative signed number as size
+     */
+    void *result = NULL;
+
+    if (MEMKIND_UNLIKELY(size >= LLONG_MAX)) {
+        errno = ENOMEM;
+    } else {
+        result = jemk_rallocx(ptr, size, flags);
+    }
+    return result;
+
+}
+
+void memkind_arena_init(struct memkind *kind)
+{
+    if (kind != MEMKIND_DEFAULT) {
+        int err = memkind_arena_create_map(kind, get_extent_hooks_by_kind(kind));
+        if (err) {
+            log_fatal("[%s] Failed to create arena map (error code:%d).", kind->name, err);
+            abort();
+        }
+    }
+}
+
+static int memkind_arena_get_stat(struct memkind *kind, memkind_stat_type stat,
+                                  bool check_init, size_t *value)
+{
+    size_t sz = sizeof(size_t);
+    size_t temp_stat;
+    int err = MEMKIND_SUCCESS;
+    unsigned i, j;
+    char cmd[128];
+
+    *value = 0;
+    for (i = 0; i < kind->arena_map_len; ++i) {
+        if (check_init) {
+            bool is_init;
+            size_t sz_b_state = sizeof(is_init);
+            snprintf(cmd, 128, "arena.%u.initialized", kind->arena_zero + i);
+            err =  jemk_mallctl(cmd, (void *)&is_init, &sz_b_state, NULL, 0);
+            if (err) {
+                log_err("Error on getting initialized state of arena.");
+                return MEMKIND_ERROR_INVALID;
+            }
+            if (!is_init) {
+                continue;
+            }
+        }
+
+        for (j = 0; j < arena_stats[stat].stats_no; ++j) {
+            snprintf(cmd, 128, arena_stats[stat].stats[j], kind->arena_zero + i);
+            err = jemk_mallctl(cmd, &temp_stat, &sz, NULL, 0);
+            if (err) {
+                log_err("Error on getting arena statistic.");
+                return MEMKIND_ERROR_INVALID;
+            }
+            *value += temp_stat;
+        }
+    }
+    return err;
+}
+
+int memkind_arena_get_stat_with_check_init(struct memkind *kind,
+                                           memkind_stat_type stat, bool check_init, size_t *value)
+{
+    int status;
+    switch (stat) {
+        case MEMKIND_STAT_TYPE_RESIDENT:
+        case MEMKIND_STAT_TYPE_ALLOCATED:
+            status = memkind_arena_get_stat(kind, stat, check_init, value);
+            break;
+        case MEMKIND_STAT_TYPE_ACTIVE:
+            status = memkind_arena_get_stat(kind, stat, check_init, value);
+            *value = PAGE_2_BYTES(*value);
+            break;
+        default:
+            //not reached
+            return MEMKIND_ERROR_INVALID;
+            break;
+    }
+    return status;
+}
+
+int memkind_arena_get_kind_stat(struct memkind *kind, memkind_stat_type stat,
+                                size_t *value)
+{
+    return memkind_arena_get_stat_with_check_init(kind, stat, false, value);
+}
+
+int memkind_arena_get_global_stat(memkind_stat_type stat, size_t *value)
+{
+    size_t sz = sizeof(size_t);
+    int err = jemk_mallctl(global_stats[stat], value, &sz, NULL, 0);
+    if (err) {
+        log_err("Error on getting global statistic.");
+        return MEMKIND_ERROR_INVALID;
+    }
+    return err;
+}
+
+int memkind_arena_enable_background_threads(size_t threads_limit)
+{
+    bool background_thread_val = true;
+    int err;
+
+    if (threads_limit) {
+        err = jemk_mallctl("max_background_threads", NULL, NULL, &threads_limit,
+                           sizeof(size_t));
+        if (err) {
+            log_err("Error on setting threads limit");
+            return MEMKIND_ERROR_INVALID;
+        }
+    }
+    err = jemk_mallctl("background_thread", NULL, NULL, &background_thread_val,
+                       sizeof(bool));
+    if (err) {
+        log_err("Error on activating background thread");
+        return MEMKIND_ERROR_INVALID;
+    }
+    return err;
+}
+
+#define DEST_SLAB_END(begin, size) ((uintptr_t)begin+ (uintptr_t)size)
+
+struct mem_util_stats {
+    void *target_slab;      // address of the slab of a potential realloaction would go to ( NULL in case of large/huge allocation)
+    size_t nfree;           // number of free regions in the slab
+    size_t nregs;           // number of regions in the slab
+    size_t slab_size;       // size of the slab in bytes
+    size_t bin_nfree;       // total number of free regions in the bin the slab belongs to
+    size_t bin_nregs;       // total number of regions in the bin the slab belongs to
+};
+
+void *memkind_arena_defrag_reallocate_with_kind_detect (void *ptr)
+{
+    return memkind_arena_defrag_reallocate(memkind_detect_kind(ptr), ptr);
+}
+
+void *memkind_arena_defrag_reallocate(struct memkind *kind, void *ptr)
+{
+    if (!ptr) {
+        return NULL;
+    }
+
+    size_t out_sz = sizeof(struct mem_util_stats);
+    struct mem_util_stats out;
+    int err = jemk_mallctl("experimental.utilization.query", &out, &out_sz, &ptr,
+                           sizeof(ptr));
+    if (err) {
+        log_err("Error on get utilization query");
+        return NULL;
+    }
+
+    // Check if input pointer resides outside of potential reallocation slab
+    // Check if occupied regions inside the slab are below average occupied regions inside bin
+    // Check if there are some free regions in the destination slab
+    if (out.target_slab &&
+        ((ptr < out.target_slab) ||
+         (uintptr_t)ptr > DEST_SLAB_END(out.target_slab, out.slab_size)) &&
+        out.nfree * out.bin_nregs >= out.nregs * out.bin_nfree &&
+        out.nfree != 0) {
+        size_t size = memkind_malloc_usable_size(kind, ptr);
+        void *ptr_new = memkind_arena_malloc_no_tcache(kind, size);
+        if (MEMKIND_UNLIKELY(!ptr_new)) return NULL;
+        memcpy(ptr_new, ptr, size);
+        memkind_free(kind, ptr);
+        return ptr_new;
+    }
+    return NULL;
+}
diff --git a/memkind-1.10.0/src/memkind_bandwidth.c b/memkind-1.10.0/src/memkind_bandwidth.c
new file mode 100644
index 0000000..8c99dae
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_bandwidth.c
@@ -0,0 +1,372 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <errno.h>
+#include <limits.h>
+#include <stdint.h>
+
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_bandwidth.h>
+
+struct numanode_bandwidth_t {
+    int numanode;
+    int bandwidth;
+};
+
+struct bandwidth_nodes_t {
+    int bandwidth;
+    int num_numanodes;
+    int *numanodes;
+};
+
+VEC(vec_cpu_node, int);
+
+#define BANDWIDTH_NODE_HIGH_VAL       2
+#define BANDWIDTH_NODE_LOW_VAL        1
+#define BANDWIDTH_NODE_NOT_PRESENT    0
+
+static void bandwidth_assign_arbitrary_values(int *bandwidth,
+                                              struct bitmask *numa_nodes_bm)
+{
+    unsigned i;
+    int nodes_num = numa_num_configured_nodes();
+
+    for (i = 0; i<NUMA_NUM_NODES; i++) {
+        if (numa_bitmask_isbitset(numa_nodes_bm, i)) {
+            bandwidth[i] = BANDWIDTH_NODE_HIGH_VAL;
+            nodes_num--;
+        }
+    }
+
+    for (i = 0; i<NUMA_NUM_NODES; i++) {
+        if (!numa_bitmask_isbitset(numa_nodes_bm, i)) {
+            bandwidth[i] = BANDWIDTH_NODE_LOW_VAL;
+            nodes_num--;
+        }
+        if (nodes_num==0) break;
+    }
+}
+
+int bandwidth_fill(int *bandwidth, get_node_bitmask get_bitmask)
+{
+    struct bitmask *high_val_node_mask = numa_allocate_nodemask();
+    int ret = get_bitmask(high_val_node_mask);
+    if(ret == 0) {
+        bandwidth_assign_arbitrary_values(bandwidth, high_val_node_mask);
+    }
+    numa_bitmask_free(high_val_node_mask);
+
+    return ret;
+}
+
+static int bandwidth_fill_values_from_enviroment(int *bandwidth,
+                                                 char *nodes_env)
+{
+    struct bitmask *numa_nodes_bm = numa_parse_nodestring(nodes_env);
+
+    if (!numa_nodes_bm) {
+        log_err("Invalid value of environment variable.");
+        return MEMKIND_ERROR_ENVIRON;
+    } else {
+        bandwidth_assign_arbitrary_values(bandwidth, numa_nodes_bm);
+        numa_bitmask_free(numa_nodes_bm);
+    }
+
+    return 0;
+}
+
+static int bandwidth_fill_nodes(int *bandwidth, fill_bandwidth_values fill,
+                                const char *env)
+{
+    char *high_value_nodes_env = secure_getenv(env);
+    if (high_value_nodes_env) {
+        log_info("Environment variable %s detected: %s.", env, high_value_nodes_env);
+        return bandwidth_fill_values_from_enviroment(bandwidth, high_value_nodes_env);
+    }
+    return fill(bandwidth);
+}
+
+static int bandwidth_compare_numanode(const void *a, const void *b)
+{
+    /***************************************************************************
+    *  qsort comparison function for numa_node_bandwidth structures.  Sorts in *
+    *  order of bandwidth and then numanode.                                   *
+    ***************************************************************************/
+    int result;
+    struct numanode_bandwidth_t *aa = (struct numanode_bandwidth_t *)(a);
+    struct numanode_bandwidth_t *bb = (struct numanode_bandwidth_t *)(b);
+    result = (aa->bandwidth > bb->bandwidth) - (aa->bandwidth < bb->bandwidth);
+    if (result == 0) {
+        result = (aa->numanode > bb->numanode) - (aa->numanode < bb->numanode);
+    }
+    return result;
+}
+
+static int bandwidth_create_nodes(const int *bandwidth, int *num_unique,
+                                  struct bandwidth_nodes_t **bandwidth_nodes)
+{
+    /***************************************************************************
+    *   bandwidth (IN):                                                        *
+    *       A vector of length equal NUMA_NUM_NODES that gives bandwidth for   *
+    *       each numa node, zero if numa node has unknown bandwidth.           *
+    *   num_unique (OUT):                                                      *
+    *       number of unique non-zero bandwidth values in bandwidth            *
+    *       vector.                                                            *
+    *   bandwidth_nodes (OUT):                                                 *
+    *       A list of length num_unique sorted by bandwidth value where        *
+    *       each element gives a list of the numa nodes that have the          *
+    *       given bandwidth.                                                   *
+    *   RETURNS MEMKIND_SUCCESS on success, error code on failure              *
+    ***************************************************************************/
+    int err = MEMKIND_SUCCESS;
+    int i, last_bandwidth;
+    *bandwidth_nodes = NULL;
+    /* allocate space for sorting array */
+    int nodes_num = numa_num_configured_nodes();
+    struct numanode_bandwidth_t *numanode_bandwidth = malloc(sizeof(
+                                                                 struct numanode_bandwidth_t) * nodes_num);
+    if (!numanode_bandwidth) {
+        err = MEMKIND_ERROR_MALLOC;
+        log_err("malloc() failed.");
+    }
+    if (!err) {
+        /* set sorting array */
+        int j = 0;
+        for (i = 0; i < NUMA_NUM_NODES; ++i) {
+            if (bandwidth[i] != BANDWIDTH_NODE_NOT_PRESENT) {
+                numanode_bandwidth[j].numanode = i;
+                numanode_bandwidth[j].bandwidth = bandwidth[i];
+                ++j;
+            }
+        }
+
+        if (j == 0) {
+            err = MEMKIND_ERROR_UNAVAILABLE;
+        }
+    }
+    if (!err) {
+        qsort(numanode_bandwidth, nodes_num, sizeof(struct numanode_bandwidth_t),
+              bandwidth_compare_numanode);
+        /* calculate the number of unique bandwidths */
+        *num_unique = 1;
+        last_bandwidth = numanode_bandwidth[0].bandwidth;
+        for (i = 1; i < nodes_num; ++i) {
+            if (numanode_bandwidth[i].bandwidth != last_bandwidth) {
+                last_bandwidth = numanode_bandwidth[i].bandwidth;
+                ++*num_unique;
+            }
+        }
+        /* allocate output array */
+        *bandwidth_nodes = (struct bandwidth_nodes_t *)malloc(sizeof(
+                                                                  struct bandwidth_nodes_t) * (*num_unique) + sizeof(int) * nodes_num);
+        if (!*bandwidth_nodes) {
+            err = MEMKIND_ERROR_MALLOC;
+            log_err("malloc() failed.");
+        }
+    }
+    if (!err) {
+        /* populate output */
+        (*bandwidth_nodes)[0].numanodes = (int *)(*bandwidth_nodes + *num_unique);
+        last_bandwidth = numanode_bandwidth[0].bandwidth;
+        int k = 0;
+        int l = 0;
+        for (i = 0; i < nodes_num; ++i, ++l) {
+            (*bandwidth_nodes)[0].numanodes[i] = numanode_bandwidth[i].numanode;
+            if (numanode_bandwidth[i].bandwidth != last_bandwidth) {
+                (*bandwidth_nodes)[k].num_numanodes = l;
+                (*bandwidth_nodes)[k].bandwidth = last_bandwidth;
+                l = 0;
+                ++k;
+                (*bandwidth_nodes)[k].numanodes = (*bandwidth_nodes)[0].numanodes + i;
+                last_bandwidth = numanode_bandwidth[i].bandwidth;
+            }
+        }
+        (*bandwidth_nodes)[k].num_numanodes = l;
+        (*bandwidth_nodes)[k].bandwidth = last_bandwidth;
+    }
+    if (numanode_bandwidth) {
+        free(numanode_bandwidth);
+    }
+    if (err) {
+        if (*bandwidth_nodes) {
+            free(*bandwidth_nodes);
+        }
+    }
+    return err;
+}
+
+static int bandwidth_set_closest_numanode(int num_unique,
+                                          const struct bandwidth_nodes_t *bandwidth_nodes, bool is_single_node,
+                                          int num_cpunode, struct vec_cpu_node **closest_numanode)
+{
+    /***************************************************************************
+    *   num_unique (IN):                                                       *
+    *       Length of bandwidth_nodes vector.                                  *
+    *   bandwidth_nodes (IN):                                                  *
+    *       Output vector from bandwidth_create_nodes().                       *
+    *   is_single_node (IN):                                                   *
+    *       Flag used to mark is only single closest numanode                  *
+    *       is a valid configuration                                           *
+    *   num_cpunode (IN):                                                      *
+    *       Number of cpu's and length of closest_numanode.                    *
+    *   closest_numanode (OUT):                                                *
+    *       Vector that maps cpu index to closest numa node(s)                 *
+    *       of the specified bandwidth.                                        *
+    *   RETURNS zero on success, error code on failure                         *
+    ***************************************************************************/
+    int err = MEMKIND_SUCCESS;
+    int min_distance, distance, i, j, old_errno;
+    struct bandwidth_nodes_t match;
+    match.bandwidth = -1;
+    int target_bandwidth = bandwidth_nodes[num_unique-1].bandwidth;
+
+    struct vec_cpu_node *node_arr = (struct vec_cpu_node *) calloc(num_cpunode,
+                                                                   sizeof(struct vec_cpu_node));
+
+    if (!node_arr) {
+        log_err("calloc() failed.");
+        return MEMKIND_ERROR_MALLOC;
+    }
+
+    for (i = 0; i < num_unique; ++i) {
+        if (bandwidth_nodes[i].bandwidth == target_bandwidth) {
+            match = bandwidth_nodes[i];
+            break;
+        }
+    }
+    if (match.bandwidth == -1) {
+        err = MEMKIND_ERROR_UNAVAILABLE;
+    } else {
+        for (i = 0; i < num_cpunode; ++i) {
+            min_distance = INT_MAX;
+            for (j = 0; j < match.num_numanodes; ++j) {
+                old_errno = errno;
+                distance = numa_distance(numa_node_of_cpu(i), match.numanodes[j]);
+                errno = old_errno;
+                if (distance < min_distance) {
+                    min_distance = distance;
+                    VEC_CLEAR(&node_arr[i]);
+                    VEC_PUSH_BACK(&node_arr[i], match.numanodes[j]);
+                } else if (distance == min_distance) {
+                    VEC_PUSH_BACK(&node_arr[i], match.numanodes[j]);
+                }
+            }
+            if (VEC_SIZE(&node_arr[i]) > 1 && is_single_node &&
+                numa_bitmask_isbitset(numa_all_cpus_ptr, i)) {
+                log_err("Invalid Numa Configuration for cpu %d", i);
+                int node = -1;
+                VEC_FOREACH(node, &node_arr[i]) {
+                    log_err("Node %d", node);
+                }
+                err = MEMKIND_ERROR_RUNTIME;
+            }
+        }
+    }
+
+    if (err) {
+        for (i = 0; i < num_cpunode; ++i) {
+            VEC_DELETE(&node_arr[i]);
+        }
+        free(node_arr);
+        node_arr = NULL;
+    }
+
+    *closest_numanode = node_arr;
+
+    return err;
+}
+
+int set_closest_numanode(fill_bandwidth_values fill_values, const char *env,
+                         struct vec_cpu_node **closest_numanode, int num_cpu, bool is_single_node)
+{
+    int status;
+    int num_unique = 0;
+    int i;
+    struct bandwidth_nodes_t *bandwidth_nodes = NULL;
+    int *bandwidth = (int *)calloc(NUMA_NUM_NODES, sizeof(int));
+
+    if (!bandwidth) {
+        log_err("calloc() failed.");
+        return MEMKIND_ERROR_MALLOC;
+    }
+
+    status = bandwidth_fill_nodes(bandwidth, fill_values, env);
+    if (status)
+        goto exit;
+
+    status = bandwidth_create_nodes(bandwidth, &num_unique, &bandwidth_nodes);
+    if (status)
+        goto exit;
+
+    status = bandwidth_set_closest_numanode(num_unique, bandwidth_nodes,
+                                            is_single_node, num_cpu, closest_numanode);
+
+    for(i = 0; i < bandwidth_nodes[num_unique-1].num_numanodes; ++i) {
+        log_info("NUMA node %d is high-bandwidth/dax-kmem memory.",
+                 bandwidth_nodes[num_unique-1].numanodes[i]);
+    }
+
+exit:
+
+    free(bandwidth_nodes);
+    free(bandwidth);
+
+    return status;
+}
+
+void set_bitmask_for_all_closest_numanodes(unsigned long *nodemask,
+                                           unsigned long maxnode, const struct vec_cpu_node *closest_numanode, int num_cpu)
+{
+    if (MEMKIND_LIKELY(nodemask)) {
+        struct bitmask nodemask_bm = {maxnode, nodemask};
+        int cpu;
+        int node = -1;
+        numa_bitmask_clearall(&nodemask_bm);
+        for (cpu = 0; cpu < num_cpu; ++cpu) {
+            VEC_FOREACH(node, &closest_numanode[cpu]) {
+                numa_bitmask_setbit(&nodemask_bm, node);
+            }
+        }
+    }
+}
+
+int set_bitmask_for_current_closest_numanode(unsigned long *nodemask,
+                                             unsigned long maxnode, const struct vec_cpu_node *closest_numanode, int num_cpu)
+{
+    if (MEMKIND_LIKELY(nodemask)) {
+        struct bitmask nodemask_bm = {maxnode, nodemask};
+        numa_bitmask_clearall(&nodemask_bm);
+        int cpu = sched_getcpu();
+        if (MEMKIND_LIKELY(cpu < num_cpu)) {
+            int node = -1;
+            VEC_FOREACH(node, &closest_numanode[cpu]) {
+                numa_bitmask_setbit(&nodemask_bm, node);
+            }
+        } else {
+            return MEMKIND_ERROR_RUNTIME;
+        }
+    }
+    return MEMKIND_SUCCESS;
+}
diff --git a/memkind-1.10.0/src/memkind_dax_kmem.c b/memkind-1.10.0/src/memkind_dax_kmem.c
new file mode 100644
index 0000000..cf9b46a
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_dax_kmem.c
@@ -0,0 +1,257 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_bandwidth.h>
+#include <memkind/internal/memkind_dax_kmem.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/heap_manager.h>
+
+#include "config.h"
+#include <pthread.h>
+#include <numa.h>
+#include <errno.h>
+
+struct dax_closest_numanode_t {
+    int init_err;
+    int num_cpu;
+    struct vec_cpu_node *closest_numanode;
+};
+
+#define NODE_VARIANT_MULTIPLE 0
+#define NODE_VARIANT_SINGLE   1
+#define NODE_VARIANT_MAX      2
+
+static struct dax_closest_numanode_t
+    memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MAX];
+static pthread_once_t memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_MAX]
+    = {PTHREAD_ONCE_INIT};
+
+static void memkind_dax_kmem_closest_numanode_init(void);
+static void memkind_dax_kmem_preferred_closest_numanode_init(void);
+#ifdef MEMKIND_DAXCTL_KMEM
+#include <daxctl/libdaxctl.h>
+#ifndef daxctl_region_foreach_safe
+#define daxctl_region_foreach_safe(ctx, region, _region) \
+    for (region = daxctl_region_get_first(ctx), \
+         _region = region ? daxctl_region_get_next(region) : NULL; \
+         region != NULL; \
+         region = _region, \
+        _region = _region ? daxctl_region_get_next(_region) : NULL)
+#endif
+
+#ifndef daxctl_dev_foreach_safe
+#define daxctl_dev_foreach_safe(region, dev, _dev) \
+    for (dev = daxctl_dev_get_first(region), \
+         _dev = dev ? daxctl_dev_get_next(dev) : NULL; \
+         dev != NULL; \
+         dev = _dev, \
+        _dev = _dev ? daxctl_dev_get_next(_dev) : NULL)
+#endif
+
+static int get_dax_kmem_nodes(struct bitmask *dax_kmem_node_mask)
+{
+    struct daxctl_region *region, *_region;
+    struct daxctl_dev *dev, *_dev;
+    struct daxctl_ctx *ctx;
+
+    int rc = daxctl_new(&ctx);
+    if (rc < 0)
+        return MEMKIND_ERROR_UNAVAILABLE;
+
+    daxctl_region_foreach_safe(ctx, region, _region) {
+        daxctl_dev_foreach_safe(region, dev, _dev) {
+            struct daxctl_memory *mem = daxctl_dev_get_memory(dev);
+            if (mem) {
+                numa_bitmask_setbit(dax_kmem_node_mask,
+                                    (unsigned)daxctl_dev_get_target_node(dev));
+            }
+        }
+    }
+
+    daxctl_unref(ctx);
+
+    return (numa_bitmask_weight(dax_kmem_node_mask) != 0) ? MEMKIND_SUCCESS :
+           MEMKIND_ERROR_UNAVAILABLE;
+}
+
+static int fill_dax_kmem_values_automatic(int *bandwidth)
+{
+    return bandwidth_fill(bandwidth, get_dax_kmem_nodes);
+}
+#else
+static int fill_dax_kmem_values_automatic(int *bandwidth)
+{
+    log_err("DAX KMEM nodes cannot be automatically detected.");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+#endif
+
+static int memkind_dax_kmem_check_available(struct memkind *kind)
+{
+    return kind->ops->get_mbind_nodemask(kind, NULL, 0);
+}
+
+static int memkind_dax_kmem_get_mbind_nodemask(struct memkind *kind,
+                                               unsigned long *nodemask, unsigned long maxnode)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MULTIPLE];
+    pthread_once(&memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_MULTIPLE],
+                 memkind_dax_kmem_closest_numanode_init);
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        g->init_err = set_bitmask_for_current_closest_numanode(nodemask, maxnode,
+                                                               g->closest_numanode, g->num_cpu);
+    }
+    return g->init_err;
+}
+
+static int memkind_dax_kmem_get_preferred_mbind_nodemask(struct memkind *kind,
+                                                         unsigned long *nodemask, unsigned long maxnode)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_SINGLE];
+    pthread_once(&memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_SINGLE],
+                 memkind_dax_kmem_preferred_closest_numanode_init);
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        g->init_err = set_bitmask_for_current_closest_numanode(nodemask, maxnode,
+                                                               g->closest_numanode, g->num_cpu);
+    }
+    return g->init_err;
+}
+
+MEMKIND_EXPORT int memkind_dax_kmem_all_get_mbind_nodemask(struct memkind *kind,
+                                                           unsigned long *nodemask, unsigned long maxnode)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MULTIPLE];
+    pthread_once(&memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_MULTIPLE],
+                 memkind_dax_kmem_closest_numanode_init);
+
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        set_bitmask_for_all_closest_numanodes(nodemask, maxnode, g->closest_numanode,
+                                              g->num_cpu);
+    }
+    return g->init_err;
+}
+
+static void memkind_dax_kmem_closest_numanode_init(void)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MULTIPLE];
+    g->num_cpu = numa_num_configured_cpus();
+    g->closest_numanode = NULL;
+    g->init_err = set_closest_numanode(fill_dax_kmem_values_automatic,
+                                       "MEMKIND_DAX_KMEM_NODES", &g->closest_numanode, g->num_cpu, false);
+}
+
+static void memkind_dax_kmem_preferred_closest_numanode_init(void)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_SINGLE];
+    g->num_cpu = numa_num_configured_cpus();
+    g->closest_numanode = NULL;
+    g->init_err = set_closest_numanode(fill_dax_kmem_values_automatic,
+                                       "MEMKIND_DAX_KMEM_NODES", &g->closest_numanode, g->num_cpu, true);
+}
+
+static void memkind_dax_kmem_init_once(void)
+{
+    memkind_init(MEMKIND_DAX_KMEM, true);
+}
+
+static void memkind_dax_kmem_all_init_once(void)
+{
+    memkind_init(MEMKIND_DAX_KMEM_ALL, true);
+}
+
+static void memkind_dax_kmem_preferred_init_once(void)
+{
+    memkind_init(MEMKIND_DAX_KMEM_PREFERRED, true);
+}
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DAX_KMEM_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_dax_kmem_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_dax_kmem_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_dax_kmem_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DAX_KMEM_ALL_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_dax_kmem_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_dax_kmem_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_dax_kmem_all_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DAX_KMEM_PREFERRED_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_dax_kmem_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_dax_kmem_get_preferred_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_dax_kmem_preferred_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
diff --git a/memkind-1.10.0/src/memkind_default.c b/memkind-1.10.0/src/memkind_default.c
new file mode 100644
index 0000000..9a3b96a
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_default.c
@@ -0,0 +1,265 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/heap_manager.h>
+
+#include <numa.h>
+#include <numaif.h>
+#include <sys/mman.h>
+#include <errno.h>
+#include <jemalloc/jemalloc.h>
+#include <stdint.h>
+
+#include "config.h"
+
+#ifndef MADV_NOHUGEPAGE
+#define MADV_NOHUGEPAGE 15
+#endif
+
+static int memkind_default_get_kind_stat(struct memkind *kind,
+                                         memkind_stat_type stat, size_t *value)
+{
+    return memkind_arena_get_stat_with_check_init(kind, stat, true, value);
+}
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DEFAULT_OPS = {
+    .create = memkind_default_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_default_malloc,
+    .calloc = memkind_default_calloc,
+    .posix_memalign = memkind_default_posix_memalign,
+    .realloc = memkind_default_realloc,
+    .free = memkind_default_free,
+    .init_once = memkind_default_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_default_destroy,
+    .get_stat = memkind_default_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT int memkind_default_create(struct memkind *kind,
+                                          struct memkind_ops *ops, const char *name)
+{
+    int err = 0;
+
+    kind->ops = ops;
+    if (strlen(name) >= MEMKIND_NAME_LENGTH_PRIV) {
+        kind->name[0] = '\0';
+        err = MEMKIND_ERROR_INVALID;
+    } else {
+        strcpy(kind->name, name);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_default_destroy(struct memkind *kind)
+{
+    return 0;
+}
+
+MEMKIND_EXPORT void *memkind_default_malloc(struct memkind *kind, size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+        return NULL;
+    }
+    return jemk_malloc(size);
+}
+
+MEMKIND_EXPORT void *memkind_default_calloc(struct memkind *kind, size_t num,
+                                            size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(num) || size_out_of_bounds(size))) {
+        return NULL;
+    }
+    return jemk_calloc(num, size);
+}
+
+MEMKIND_EXPORT int memkind_default_posix_memalign(struct memkind *kind,
+                                                  void **memptr, size_t alignment, size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+        *memptr = NULL;
+        return 0;
+    }
+    return jemk_posix_memalign(memptr, alignment, size);
+}
+
+MEMKIND_EXPORT void *memkind_default_realloc(struct memkind *kind, void *ptr,
+                                             size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+        jemk_free(ptr);
+        return NULL;
+    }
+    return jemk_realloc(ptr, size);
+}
+
+MEMKIND_EXPORT void memkind_default_free(struct memkind *kind, void *ptr)
+{
+    jemk_free(ptr);
+}
+
+MEMKIND_EXPORT size_t memkind_default_malloc_usable_size(struct memkind *kind,
+                                                         void *ptr)
+{
+    return jemk_malloc_usable_size(ptr);
+}
+
+MEMKIND_EXPORT void *memkind_default_mmap(struct memkind *kind, void *addr,
+                                          size_t size)
+{
+    void *result = MAP_FAILED;
+    int err = 0;
+    int flags;
+
+    if (kind->ops->get_mmap_flags) {
+        err = kind->ops->get_mmap_flags(kind, &flags);
+    } else {
+        err = memkind_default_get_mmap_flags(kind, &flags);
+    }
+    if (MEMKIND_LIKELY(!err)) {
+        result = mmap(addr, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+        if (result == MAP_FAILED) {
+            log_err("syscall mmap() returned: %p", result);
+            return result;
+        }
+    }
+    if (kind->ops->mbind) {
+        err = kind->ops->mbind(kind, result, size);
+        if (err) {
+            munmap(result, size);
+            result = MAP_FAILED;
+        }
+    }
+    if (kind->ops->madvise) {
+        err = kind->ops->madvise(kind, result, size);
+        if (err) {
+            munmap(result, size);
+            result = MAP_FAILED;
+        }
+    }
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_nohugepage_madvise(struct memkind *kind, void *addr,
+                                              size_t size)
+{
+    int err = madvise(addr, size, MADV_NOHUGEPAGE);
+
+    //checking if EINVAL was returned due to lack of THP support in kernel
+    if ((err == EINVAL) && (((uintptr_t) addr & (uintptr_t) 0xfff) == 0) &&
+        (size > 0)) {
+        return 0;
+    }
+    if (MEMKIND_UNLIKELY(err)) {
+        log_err("syscall madvise() returned: %d", err);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_default_mbind(struct memkind *kind, void *ptr,
+                                         size_t size)
+{
+    nodemask_t nodemask;
+    int err = 0;
+    int mode;
+
+    if (MEMKIND_UNLIKELY(kind->ops->get_mbind_nodemask == NULL ||
+                         kind->ops->get_mbind_mode == NULL)) {
+        log_err("memkind_ops->mbind_mode or memkind_ops->bind_nodemask is NULL.");
+        return MEMKIND_ERROR_BADOPS;
+    }
+    err = kind->ops->get_mbind_nodemask(kind, nodemask.n, NUMA_NUM_NODES);
+    if (MEMKIND_UNLIKELY(err)) {
+        return err;
+    }
+    err = kind->ops->get_mbind_mode(kind, &mode);
+    if (MEMKIND_UNLIKELY(err)) {
+        return err;
+    }
+    err = mbind(ptr, size, mode, nodemask.n, NUMA_NUM_NODES, 0);
+    if (MEMKIND_UNLIKELY(err)) {
+        log_err("syscall mbind() returned: %d", err);
+        return MEMKIND_ERROR_MBIND;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_default_get_mmap_flags(struct memkind *kind,
+                                                  int *flags)
+{
+    *flags = MAP_PRIVATE | MAP_ANONYMOUS;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_default_get_mbind_nodemask(struct memkind *kind,
+                                                      unsigned long *nodemask,
+                                                      unsigned long maxnode)
+{
+    struct bitmask nodemask_bm = {maxnode, nodemask};
+    copy_bitmask_to_bitmask(numa_all_nodes_ptr, &nodemask_bm);
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_default_get_mbind_mode(struct memkind *kind,
+                                                  int *mode)
+{
+    *mode = MPOL_BIND;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_preferred_get_mbind_mode(struct memkind *kind,
+                                                    int *mode)
+{
+    *mode = MPOL_PREFERRED;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_interleave_get_mbind_mode(struct memkind *kind,
+                                                     int *mode)
+{
+    *mode = MPOL_INTERLEAVE;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_posix_check_alignment(struct memkind *kind,
+                                                 size_t alignment)
+{
+    int err = 0;
+    if ((alignment < sizeof(void *)) ||
+        (((alignment - 1) & alignment) != 0)) {
+        err = EINVAL;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT void memkind_default_init_once(void)
+{
+    heap_manager_init(MEMKIND_DEFAULT);
+}
diff --git a/memkind-1.10.0/src/memkind_gbtlb.c b/memkind-1.10.0/src/memkind_gbtlb.c
new file mode 100644
index 0000000..0470517
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_gbtlb.c
@@ -0,0 +1,128 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_gbtlb.h>
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/memkind_arena.h>
+
+static void memkind_hbw_gbtlb_init_once(void);
+static void memkind_hbw_preferred_gbtlb_init_once(void);
+static void memkind_gbtlb_init_once(void);
+
+static void *gbtlb_mmap(struct memkind *kind, void *addr, size_t size);
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_GBTLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = gbtlb_mmap,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_gbtlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_PREFERRED_GBTLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = gbtlb_mmap,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_preferred_gbtlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_GBTLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = gbtlb_mmap,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_gbtlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+#define ONE_GB 1073741824ULL
+
+static void memkind_gbtlb_ceil_size(size_t *size)
+{
+    *size = (*size % ONE_GB) ? ((*size / ONE_GB) + 1) * ONE_GB : *size;
+}
+
+static void *gbtlb_mmap(struct memkind *kind, void *addr, size_t size)
+{
+    memkind_gbtlb_ceil_size(&size);
+    return memkind_default_mmap(kind, addr, size);
+}
+
+static void memkind_hbw_gbtlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_GBTLB, false);
+}
+
+static void memkind_hbw_preferred_gbtlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_PREFERRED_GBTLB, false);
+}
+
+static void memkind_gbtlb_init_once(void)
+{
+    memkind_init(MEMKIND_GBTLB, false);
+}
diff --git a/memkind-1.10.0/src/memkind_hbw.c b/memkind-1.10.0/src/memkind_hbw.c
new file mode 100644
index 0000000..93d0da6
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_hbw.c
@@ -0,0 +1,404 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_bandwidth.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/heap_manager.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <unistd.h>
+#include <limits.h>
+#include <pthread.h>
+#include <numa.h>
+#include <numaif.h>
+#include <errno.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <utmpx.h>
+#include <sched.h>
+#include <stdint.h>
+#include <assert.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_ALL_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_all_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_hugetlb_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_ALL_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_hugetlb_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_all_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_PREFERRED_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_preferred_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_PREFERRED_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_hugetlb_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_preferred_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_INTERLEAVE_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .madvise = memkind_nohugepage_madvise,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_interleave_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_interleave_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+struct hbw_closest_numanode_t {
+    int init_err;
+    int num_cpu;
+    struct vec_cpu_node *closest_numanode;
+};
+
+static struct hbw_closest_numanode_t memkind_hbw_closest_numanode_g;
+static pthread_once_t memkind_hbw_closest_numanode_once_g = PTHREAD_ONCE_INIT;
+
+static void memkind_hbw_closest_numanode_init(void);
+
+// This declaration is necessary, cause it's missing in headers from libnuma 2.0.8
+extern unsigned int numa_bitmask_weight(const struct bitmask *bmp );
+
+static int fill_bandwidth_values_heuristically (int *bandwidth);
+
+MEMKIND_EXPORT int memkind_hbw_check_available(struct memkind *kind)
+{
+    return kind->ops->get_mbind_nodemask(kind, NULL, 0);
+}
+
+MEMKIND_EXPORT int memkind_hbw_hugetlb_check_available(struct memkind *kind)
+{
+    int err = memkind_hbw_check_available(kind);
+    if (!err) {
+        err = memkind_hugetlb_check_available_2mb(kind);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_hbw_get_mbind_nodemask(struct memkind *kind,
+                                                  unsigned long *nodemask,
+                                                  unsigned long maxnode)
+{
+    struct hbw_closest_numanode_t *g = &memkind_hbw_closest_numanode_g;
+    pthread_once(&memkind_hbw_closest_numanode_once_g,
+                 memkind_hbw_closest_numanode_init);
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        g->init_err = set_bitmask_for_current_closest_numanode(nodemask, maxnode,
+                                                               g->closest_numanode, g->num_cpu);
+    }
+    return g->init_err;
+}
+
+MEMKIND_EXPORT int memkind_hbw_all_get_mbind_nodemask(struct memkind *kind,
+                                                      unsigned long *nodemask,
+                                                      unsigned long maxnode)
+{
+    struct hbw_closest_numanode_t *g = &memkind_hbw_closest_numanode_g;
+    pthread_once(&memkind_hbw_closest_numanode_once_g,
+                 memkind_hbw_closest_numanode_init);
+
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        set_bitmask_for_all_closest_numanodes(nodemask, maxnode, g->closest_numanode,
+                                              g->num_cpu);
+    }
+    return g->init_err;
+}
+
+typedef struct registers_t {
+    uint32_t eax;
+    uint32_t ebx;
+    uint32_t ecx;
+    uint32_t edx;
+} registers_t;
+
+inline static void cpuid_asm(int leaf, int subleaf, registers_t *registers)
+{
+#ifdef __x86_64__
+    asm volatile("cpuid":"=a"(registers->eax),
+                 "=b"(registers->ebx),
+                 "=c"(registers->ecx),
+                 "=d"(registers->edx):"0"(leaf), "2"(subleaf));
+#else
+    // Non-x86 can't possibly be Knight's Landing nor Knight's Mill.
+    registers->eax = 0;
+#endif
+}
+
+#define CPUID_MODEL_SHIFT       (4)
+#define CPUID_MODEL_MASK        (0xf)
+#define CPUID_EXT_MODEL_MASK    (0xf)
+#define CPUID_EXT_MODEL_SHIFT   (16)
+#define CPUID_FAMILY_MASK       (0xf)
+#define CPUID_FAMILY_SHIFT      (8)
+#define CPU_MODEL_KNL           (0x57)
+#define CPU_MODEL_KNM           (0x85)
+#define CPU_FAMILY_INTEL        (0x06)
+
+typedef struct {
+    uint32_t model;
+    uint32_t family;
+} cpu_model_data_t;
+
+static cpu_model_data_t get_cpu_model_data()
+{
+    registers_t registers;
+    cpuid_asm(1, 0, &registers);
+    uint32_t model = (registers.eax >> CPUID_MODEL_SHIFT) & CPUID_MODEL_MASK;
+    uint32_t model_ext = (registers.eax >> CPUID_EXT_MODEL_SHIFT) &
+                         CPUID_EXT_MODEL_MASK;
+
+    cpu_model_data_t data;
+    data.model = model | (model_ext << 4);
+    data.family = (registers.eax >> CPUID_FAMILY_SHIFT) & CPUID_FAMILY_MASK;
+    return data;
+}
+
+static bool is_hbm_supported(cpu_model_data_t cpu)
+{
+    return cpu.family == CPU_FAMILY_INTEL &&
+           (cpu.model == CPU_MODEL_KNL || cpu.model == CPU_MODEL_KNM);
+}
+
+static int get_high_bandwidth_nodes(struct bitmask *hbw_node_mask)
+{
+    int nodes_num = numa_num_configured_nodes();
+    // Check if NUMA configuration is supported.
+    if(nodes_num == 2 || nodes_num == 4 || nodes_num == 8) {
+        struct bitmask *node_cpus = numa_allocate_cpumask();
+
+        assert(hbw_node_mask->size >= nodes_num);
+        assert(node_cpus->size >= nodes_num);
+        int i;
+        for(i=0; i<nodes_num; i++) {
+            numa_node_to_cpus(i, node_cpus);
+            if(numa_bitmask_weight(node_cpus) == 0) {
+                //NUMA nodes without CPU are HBW nodes.
+                numa_bitmask_setbit(hbw_node_mask, i);
+            }
+        }
+
+        numa_bitmask_free(node_cpus);
+
+        if(2*numa_bitmask_weight(hbw_node_mask) == nodes_num) {
+            return 0;
+        }
+    }
+
+    return MEMKIND_ERROR_UNAVAILABLE;
+}
+
+///This function tries to fill bandwidth array based on knowledge about known CPU models
+static int fill_bandwidth_values_heuristically(int *bandwidth)
+{
+    cpu_model_data_t cpu = get_cpu_model_data();
+
+    if(!is_hbm_supported(cpu)) {
+        log_err("High Bandwidth Memory is not supported by this CPU.");
+        return MEMKIND_ERROR_UNAVAILABLE;
+    }
+
+    switch(cpu.model) {
+        case CPU_MODEL_KNL:
+        case CPU_MODEL_KNM: {
+            int ret = bandwidth_fill(bandwidth, get_high_bandwidth_nodes);
+            if(ret == 0) {
+                log_info("Detected High Bandwidth Memory.");
+            }
+            return ret;
+        }
+        default:
+            return MEMKIND_ERROR_UNAVAILABLE;
+    }
+}
+static void memkind_hbw_closest_numanode_init(void)
+{
+    struct hbw_closest_numanode_t *g = &memkind_hbw_closest_numanode_g;
+    g->num_cpu = numa_num_configured_cpus();
+    g->closest_numanode = NULL;
+    g->init_err = set_closest_numanode(fill_bandwidth_values_heuristically,
+                                       "MEMKIND_HBW_NODES", &g->closest_numanode, g->num_cpu, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_init_once(void)
+{
+    memkind_init(MEMKIND_HBW, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_all_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_ALL, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_HUGETLB, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_all_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_ALL_HUGETLB, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_preferred_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_PREFERRED, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_preferred_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_PREFERRED_HUGETLB, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_interleave_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_INTERLEAVE, true);
+}
diff --git a/memkind-1.10.0/src/memkind_hugetlb.c b/memkind-1.10.0/src/memkind_hugetlb.c
new file mode 100644
index 0000000..dab2baa
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_hugetlb.c
@@ -0,0 +1,379 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <sys/mman.h>
+#ifndef MAP_HUGETLB
+#define MAP_HUGETLB 0x40000
+#endif
+#ifndef MAP_HUGE_2MB
+#define MAP_HUGE_2MB (21 << 26)
+#endif
+
+#include <stdio.h>
+#include <errno.h>
+#include <numa.h>
+#include <pthread.h>
+#include <dirent.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+static int get_nr_overcommit_hugepages_cached(size_t pagesize, size_t *out);
+static int get_nr_hugepages_cached(size_t pagesize, struct bitmask *nodemask,
+                                   size_t *out);
+
+static int memkind_hugetlb_check_available(struct memkind *kind,
+                                           size_t huge_size);
+
+MEMKIND_EXPORT int memkind_hugetlb_get_mmap_flags(struct memkind *kind,
+                                                  int *flags)
+{
+    *flags = MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB | MAP_HUGE_2MB;
+    return 0;
+}
+
+MEMKIND_EXPORT void memkind_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HUGETLB, true);
+}
+
+MEMKIND_EXPORT int memkind_hugetlb_check_available_2mb(struct memkind *kind)
+{
+    return memkind_hugetlb_check_available(kind, 2097152);
+}
+
+/* huge_size: the huge page size in bytes */
+static int memkind_hugetlb_check_available(struct memkind *kind,
+                                           size_t huge_size)
+{
+    int err = 0;
+    nodemask_t nodemask;
+    struct bitmask nodemask_bm = {NUMA_NUM_NODES, nodemask.n};
+
+    /* on x86_64 default huge page size is 2MB */
+    if (huge_size == 0) {
+        huge_size = 2097152;
+    }
+
+    if (kind->ops->get_mbind_nodemask) {
+        err = kind->ops->get_mbind_nodemask(kind, nodemask.n, NUMA_NUM_NODES);
+    } else {
+        numa_bitmask_setall(&nodemask_bm);
+    }
+
+    size_t nr_persistent_hugepages, nr_overcommit_hugepages;
+
+    err = get_nr_hugepages_cached(huge_size, &nodemask_bm,
+                                  &nr_persistent_hugepages);
+    if(err) {
+        return err;
+    }
+
+    err = get_nr_overcommit_hugepages_cached(huge_size, &nr_overcommit_hugepages);
+    if(err) {
+        return err;
+    }
+
+    if (!nr_overcommit_hugepages && !nr_persistent_hugepages) {
+        log_err("Persistent hugepages and overcommit hugepages are not available.");
+        return MEMKIND_ERROR_HUGETLB;
+    }
+
+    return err;
+}
+
+struct hugepage_size_info {
+    size_t size;
+    size_t *nr_hugepages_per_node_array;
+    size_t  nr_overcommit;
+};
+
+struct memkind_hugepages_config_t {
+    struct hugepage_size_info **hugepages_info_array;
+    int hugepages_info_array_len;
+    int err; // 0 if sysfs parsing successful, appropriate memkind_error otherwise
+} memkind_hugepages_config;
+
+static pthread_once_t memkind_hugepages_config_once_g = PTHREAD_ONCE_INIT;
+
+static struct hugepage_size_info *allocate_hugepage_size_info()
+{
+    struct hugepage_size_info *newInfo = malloc(sizeof(struct hugepage_size_info));
+    if(newInfo == NULL) {
+        log_err("malloc() failed.");
+        return NULL;
+    }
+
+    newInfo->nr_hugepages_per_node_array = calloc(NUMA_NUM_NODES, sizeof(size_t));
+    if(newInfo->nr_hugepages_per_node_array == NULL) {
+        free(newInfo);
+        log_err("calloc() failed.");
+        return NULL;
+    }
+
+    return newInfo;
+}
+
+static size_t get_sysfs_entry_value(const char *entry_path)
+{
+    int errno_before;
+    FILE *fid;
+    int num_read;
+    size_t value_read, ret = 0;
+
+    errno_before = errno;
+    fid = fopen(entry_path, "r");
+    if (fid) {
+        num_read = fscanf(fid, "%zud", &value_read);
+        if(num_read) {
+            ret  = value_read;
+        }
+        fclose(fid);
+    } else {
+        errno = errno_before;
+    }
+    return ret;
+}
+
+// construct hugepage_size_info object and fill it with data for provided pagesize
+static void init_hugepage_size_info(size_t pagesize,
+                                    struct hugepage_size_info *newInfo)
+{
+    char formatted_path[128];
+    const char *nr_path_fmt =
+        "/sys/devices/system/node/node%u/hugepages/hugepages-%zukB/nr_hugepages";
+    const char *nr_overcommit_path_fmt =
+        "/sys/kernel/mm/hugepages/hugepages-%zukB/nr_overcommit_hugepages";
+    int snprintf_ret = 0;
+    size_t node;
+
+    size_t pagesize_kb = pagesize >> 10;
+
+    newInfo->size = pagesize;
+
+    //read overcommit hugepages limit for this pagesize
+    snprintf_ret = snprintf(formatted_path, sizeof(formatted_path),
+                            nr_overcommit_path_fmt, pagesize_kb);
+    if (snprintf_ret > 0 && snprintf_ret < sizeof(formatted_path)) {
+        newInfo->nr_overcommit = get_sysfs_entry_value(formatted_path);
+        log_info("Overcommit limit for %zu kB hugepages is %zu.", pagesize,
+                 newInfo->nr_overcommit);
+    }
+
+    //read every node nr_hugepages for this pagesize
+    for (node = 0; node < NUMA_NUM_NODES; ++node) {
+        snprintf_ret = snprintf(formatted_path, sizeof(formatted_path), nr_path_fmt,
+                                node, pagesize_kb);
+        if(snprintf_ret > 0 && snprintf_ret < sizeof(formatted_path)) {
+            newInfo->nr_hugepages_per_node_array[node] = get_sysfs_entry_value(
+                                                             formatted_path);
+            if(node < numa_num_configured_nodes()) {
+                log_info("Number of %zu kB hugepages on node %zu equals %zu.", pagesize, node,
+                         newInfo->nr_hugepages_per_node_array[node]);
+            }
+        }
+    }
+}
+
+// get hugepage size in bytes out of sysfs dir name
+static int parse_pagesize_from_sysfs_entry(const char *entry, size_t *out)
+{
+    size_t pagesize;
+    int ret = sscanf(entry, "hugepages-%zukB", &pagesize);
+
+    if(ret == 1) {
+        *out = pagesize << 10; //we are using bytes but kernel is using kB
+        return 0;
+    }
+
+    return -1;
+}
+
+
+static void hugepages_config_init_once()
+{
+    unsigned j, i = 0;
+    size_t pagesize;
+    struct hugepage_size_info **hugepages_info_array = NULL;
+    struct dirent *dir;
+    DIR *hugepages_sysfs = opendir("/sys/kernel/mm/hugepages");
+    if(hugepages_sysfs == NULL) {
+        memkind_hugepages_config.err = MEMKIND_ERROR_HUGETLB;
+        log_err("/sys/kernel/mm/hugepages directory is not available.");
+        return;
+    }
+
+    unsigned hugepages_info_array_len = 2; //initial size of array
+    hugepages_info_array = malloc(hugepages_info_array_len * sizeof(
+                                      struct hugepage_size_info *));
+    if (hugepages_info_array == NULL) {
+        memkind_hugepages_config.err = MEMKIND_ERROR_MALLOC;
+        closedir(hugepages_sysfs);
+        log_err("malloc() failed.");
+        return;
+    }
+
+    while ((dir = readdir(hugepages_sysfs)) != NULL) {
+        if(dir->d_type == DT_DIR &&
+           parse_pagesize_from_sysfs_entry(dir->d_name, &pagesize) == 0) {
+            struct hugepage_size_info *new_hugepage_info = allocate_hugepage_size_info();
+            if(new_hugepage_info == NULL) {
+                memkind_hugepages_config.err = MEMKIND_ERROR_MALLOC;
+                break;
+            }
+
+            init_hugepage_size_info(pagesize, new_hugepage_info);
+
+            //there is more hugepage sizes than expected, reallocation of array needed
+            if(i == hugepages_info_array_len) {
+                hugepages_info_array_len *= 2;
+                struct hugepage_size_info **swap_tmp = realloc(hugepages_info_array,
+                                                               hugepages_info_array_len * sizeof(struct hugepage_size_info *));
+                if(swap_tmp == NULL) {
+                    free(new_hugepage_info);
+                    memkind_hugepages_config.err = MEMKIND_ERROR_MALLOC;
+                    log_err("realloc() failed.");
+                    break;
+                }
+                hugepages_info_array = swap_tmp;
+
+            }
+            hugepages_info_array[i] = new_hugepage_info;
+            i++;
+        }
+    }
+
+    closedir(hugepages_sysfs);
+
+    if(memkind_hugepages_config.err == 0) {
+        memkind_hugepages_config.hugepages_info_array = hugepages_info_array;
+        memkind_hugepages_config.hugepages_info_array_len = i;
+    } else {
+        for(j=0; j<i; j++) {
+            free(hugepages_info_array[i]);
+        }
+        free(hugepages_info_array);
+    }
+
+    return;
+}
+
+#ifdef __GNUC__
+__attribute__((destructor))
+#endif
+static void destroy_hugepages_per_node()
+{
+    int i;
+    for(i=0; i<memkind_hugepages_config.hugepages_info_array_len; i++) {
+        free(memkind_hugepages_config.hugepages_info_array[i]);
+    }
+    free(memkind_hugepages_config.hugepages_info_array);
+}
+
+// helper function that find and return hugepage_size_info object for specified pagesize
+static struct hugepage_size_info *get_hugepage_info_for_pagesize(
+    size_t pagesize)
+{
+    int i;
+
+    for(i=0; i<memkind_hugepages_config.hugepages_info_array_len; i++) {
+        if(memkind_hugepages_config.hugepages_info_array[i]->size == pagesize) {
+            return memkind_hugepages_config.hugepages_info_array[i];
+        }
+    }
+    return NULL;
+}
+
+// returns sum of pre-allocated hugepage for specified pagesize and set of nodes
+static int get_nr_hugepages_cached(size_t pagesize, struct bitmask *nodemask,
+                                   size_t *out)
+{
+    int i;
+    size_t nr_hugepages = 0;
+    int num_node = numa_num_configured_nodes();
+    pthread_once(&memkind_hugepages_config_once_g,
+                 hugepages_config_init_once);
+
+
+    if(memkind_hugepages_config.err != 0) {
+        return memkind_hugepages_config.err;
+    }
+
+    struct hugepage_size_info *info = get_hugepage_info_for_pagesize(pagesize);
+    if(info == NULL) {
+        log_err("Unable to allocate hugepages, because info about pre-allocated hugepages is not available.");
+        return MEMKIND_ERROR_HUGETLB;
+    }
+
+    for(i=0; i<num_node; i++) {
+        if(numa_bitmask_isbitset(nodemask, i)) {
+            nr_hugepages += info->nr_hugepages_per_node_array[i];
+        }
+    }
+
+    *out = nr_hugepages;
+    return 0;
+}
+
+// returns hugepages overcommit limit for specified pagesize
+static int get_nr_overcommit_hugepages_cached(size_t pagesize, size_t *out)
+{
+    pthread_once(&memkind_hugepages_config_once_g,
+                 hugepages_config_init_once);
+
+    if(memkind_hugepages_config.err != 0) {
+        return memkind_hugepages_config.err;
+    }
+
+    struct hugepage_size_info *info = get_hugepage_info_for_pagesize(pagesize);
+    if(info == NULL) {
+        log_err("Unable to allocate hugepages, because info about overcommit hugepages is not available.");
+        return MEMKIND_ERROR_HUGETLB;
+    }
+
+    *out = info->nr_overcommit;
+    return 0;
+}
+
diff --git a/memkind-1.10.0/src/memkind_interleave.c b/memkind-1.10.0/src/memkind_interleave.c
new file mode 100644
index 0000000..79d56f8
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_interleave.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_interleave.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_INTERLEAVE_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mbind = memkind_default_mbind,
+    .madvise = memkind_nohugepage_madvise,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_interleave_get_mbind_mode,
+    .get_mbind_nodemask = memkind_default_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_interleave_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT void memkind_interleave_init_once(void)
+{
+    memkind_init(MEMKIND_INTERLEAVE, true);
+}
diff --git a/memkind-1.10.0/src/memkind_log.c b/memkind-1.10.0/src/memkind_log.c
new file mode 100644
index 0000000..86aa8c1
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_log.c
@@ -0,0 +1,104 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_log.h>
+
+#include <assert.h>
+#include <stdio.h>
+#include <pthread.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdbool.h>
+
+typedef enum {
+    MESSAGE_TYPE_INFO,
+    MESSAGE_TYPE_ERROR,
+    MESSAGE_TYPE_FATAL,
+    MESSAGE_TYPE_MAX_VALUE,
+} message_type_t;
+
+static char *message_prefixes[MESSAGE_TYPE_MAX_VALUE] = {
+    [MESSAGE_TYPE_INFO]     = "MEMKIND_INFO",
+    [MESSAGE_TYPE_ERROR]    = "MEMKIND_ERROR",
+    [MESSAGE_TYPE_FATAL]    = "MEMKIND_FATAL",
+};
+
+static bool log_enabled;
+
+static pthread_once_t init_once = PTHREAD_ONCE_INIT;
+static void log_init_once(void)
+{
+    char *memkind_debug_env = secure_getenv("MEMKIND_DEBUG");
+
+    if (memkind_debug_env) {
+        if(strcmp(memkind_debug_env, "1") == 0) {
+            log_enabled = true;
+        } else {
+            fprintf(stderr,
+                    "MEMKIND_WARNING: debug option \"%s\" unknown; Try man memkind for available options.\n",
+                    memkind_debug_env);
+        }
+    }
+}
+
+static pthread_mutex_t log_lock = PTHREAD_MUTEX_INITIALIZER;
+static void log_generic(message_type_t type, const char *format, va_list args)
+{
+    pthread_once(&init_once, log_init_once);
+    if(log_enabled || (type == MESSAGE_TYPE_FATAL)) {
+        if (pthread_mutex_lock(&log_lock) != 0)
+            assert(0 && "failed to acquire mutex");
+        fprintf(stderr, "%s: ", message_prefixes[type]);
+        vfprintf(stderr, format, args);
+        fprintf(stderr, "\n");
+        if (pthread_mutex_unlock(&log_lock) != 0)
+            assert(0 && "failed to release mutex");
+    }
+}
+
+void log_info(const char *format, ...)
+{
+    va_list args;
+    va_start(args, format);
+    log_generic(MESSAGE_TYPE_INFO, format, args);
+    va_end(args);
+}
+
+void log_err(const char *format, ...)
+{
+    va_list args;
+    va_start(args, format);
+    log_generic(MESSAGE_TYPE_ERROR, format, args);
+    va_end(args);
+}
+
+void log_fatal(const char *format, ...)
+{
+    va_list args;
+    va_start(args, format);
+    log_generic(MESSAGE_TYPE_FATAL, format, args);
+    va_end(args);
+}
+
diff --git a/memkind-1.10.0/src/memkind_pmem.c b/memkind-1.10.0/src/memkind_pmem.c
new file mode 100644
index 0000000..deeef18
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_pmem.c
@@ -0,0 +1,292 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_pmem.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <sys/mman.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <assert.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_PMEM_OPS = {
+    .create = memkind_pmem_create,
+    .destroy = memkind_pmem_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = memkind_pmem_mmap,
+    .get_mmap_flags = memkind_pmem_get_mmap_flags,
+    .get_arena = memkind_thread_get_arena,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_pmem_destroy,
+    .update_memory_usage_policy = memkind_arena_update_memory_usage_policy,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+void *pmem_extent_alloc(extent_hooks_t *extent_hooks,
+                        void *new_addr,
+                        size_t size,
+                        size_t alignment,
+                        bool *zero,
+                        bool *commit,
+                        unsigned arena_ind)
+{
+    int err;
+    void *addr = NULL;
+
+    if (new_addr != NULL) {
+        /* not supported */
+        goto exit;
+    }
+
+    struct memkind *kind = get_kind_by_arena(arena_ind);
+    if (kind == NULL) {
+        return NULL;
+    }
+
+    err = memkind_check_available(kind);
+    if (err) {
+        goto exit;
+    }
+
+    addr = memkind_pmem_mmap(kind, new_addr, size);
+
+    if (addr != MAP_FAILED) {
+        *zero = true;
+        *commit = true;
+
+        /* XXX - check alignment */
+    } else {
+        addr = NULL;
+    }
+exit:
+    return addr;
+}
+
+bool pmem_extent_dalloc(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        bool committed,
+                        unsigned arena_ind)
+{
+    // if madvise fail, it means that addr isn't mapped shared (doesn't come from pmem)
+    // and it should be unmapped to avoid space exhaustion when calling large number of
+    // operations like memkind_create_pmem and memkind_destroy_kind
+    errno = 0;
+    int status = madvise(addr, size, MADV_REMOVE);
+    if (!status) {
+        struct memkind *kind = get_kind_by_arena(arena_ind);
+        struct memkind_pmem *priv = kind->priv;
+        if (pthread_mutex_lock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to acquire mutex");
+        priv->current_size -= size;
+        if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to release mutex");
+    } else {
+        if (errno == EOPNOTSUPP) {
+            log_fatal("Filesystem doesn't support FALLOC_FL_PUNCH_HOLE.");
+            abort();
+        }
+        if (munmap(addr, size) == -1) {
+            log_err("munmap failed!");
+        }
+    }
+    return true;
+}
+
+bool pmem_extent_commit(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        size_t offset,
+                        size_t length,
+                        unsigned arena_ind)
+{
+    /* do nothing - report success */
+    return false;
+}
+
+bool pmem_extent_decommit(extent_hooks_t *extent_hooks,
+                          void *addr,
+                          size_t size,
+                          size_t offset,
+                          size_t length,
+                          unsigned arena_ind)
+{
+    /* do nothing - report failure (opt-out) */
+    return true;
+}
+
+bool pmem_extent_purge(extent_hooks_t *extent_hooks,
+                       void *addr,
+                       size_t size,
+                       size_t offset,
+                       size_t length,
+                       unsigned arena_ind)
+{
+    /* do nothing - report failure (opt-out) */
+    return true;
+}
+
+bool pmem_extent_split(extent_hooks_t *extent_hooks,
+                       void *addr,
+                       size_t size,
+                       size_t size_a,
+                       size_t size_b,
+                       bool committed,
+                       unsigned arena_ind)
+{
+    /* do nothing - report success */
+    return false;
+}
+
+bool pmem_extent_merge(extent_hooks_t *extent_hooks,
+                       void *addr_a,
+                       size_t size_a,
+                       void *addr_b,
+                       size_t size_b,
+                       bool committed,
+                       unsigned arena_ind)
+{
+    /* do nothing - report success */
+    return false;
+}
+
+void pmem_extent_destroy(extent_hooks_t *extent_hooks,
+                         void *addr,
+                         size_t size,
+                         bool committed,
+                         unsigned arena_ind)
+{
+    if (munmap(addr, size) == -1) {
+        log_err("munmap failed!");
+    }
+}
+
+static extent_hooks_t pmem_extent_hooks = {
+    .alloc = pmem_extent_alloc,
+    .dalloc = pmem_extent_dalloc,
+    .commit = pmem_extent_commit,
+    .decommit = pmem_extent_decommit,
+    .purge_lazy = pmem_extent_purge,
+    .split = pmem_extent_split,
+    .merge = pmem_extent_merge,
+    .destroy = pmem_extent_destroy
+};
+
+MEMKIND_EXPORT int memkind_pmem_create(struct memkind *kind,
+                                       struct memkind_ops *ops, const char *name)
+{
+    struct memkind_pmem *priv;
+    int err;
+
+    priv = (struct memkind_pmem *)malloc(sizeof(struct memkind_pmem));
+    if (!priv) {
+        log_err("malloc() failed.");
+        return MEMKIND_ERROR_MALLOC;
+    }
+
+    if (pthread_mutex_init(&priv->pmem_lock, NULL) != 0) {
+        err = MEMKIND_ERROR_RUNTIME;
+        goto exit;
+    }
+
+    err = memkind_default_create(kind, ops, name);
+    if (err) {
+        goto exit;
+    }
+
+    err = memkind_arena_create_map(kind, &pmem_extent_hooks);
+    if (err) {
+        goto exit;
+    }
+
+    kind->priv = priv;
+    return 0;
+
+exit:
+    /* err is set, please don't overwrite it with result of pthread_mutex_destroy */
+    pthread_mutex_destroy(&priv->pmem_lock);
+    free(priv);
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_pmem_destroy(struct memkind *kind)
+{
+    struct memkind_pmem *priv = kind->priv;
+
+    memkind_arena_destroy(kind);
+
+    pthread_mutex_destroy(&priv->pmem_lock);
+
+    (void) close(priv->fd);
+    free(priv);
+
+    return 0;
+}
+
+MEMKIND_EXPORT void *memkind_pmem_mmap(struct memkind *kind, void *addr,
+                                       size_t size)
+{
+    struct memkind_pmem *priv = kind->priv;
+    void *result;
+
+    if (pthread_mutex_lock(&priv->pmem_lock) != 0)
+        assert(0 && "failed to acquire mutex");
+
+    if (priv->max_size != 0 && priv->current_size + size > priv->max_size) {
+        if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to release mutex");
+        return MAP_FAILED;
+    }
+
+    if ((errno = posix_fallocate(priv->fd, priv->offset, (off_t)size)) != 0) {
+        if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to release mutex");
+        return MAP_FAILED;
+    }
+
+    if ((result = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, priv->fd,
+                       priv->offset)) != MAP_FAILED) {
+        priv->offset += size;
+        priv->current_size += size;
+    }
+
+    if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+        assert(0 && "failed to release mutex");
+
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_pmem_get_mmap_flags(struct memkind *kind, int *flags)
+{
+    *flags = MAP_SHARED;
+    return 0;
+}
diff --git a/memkind-1.10.0/src/memkind_regular.c b/memkind-1.10.0/src/memkind_regular.c
new file mode 100644
index 0000000..4cd1887
--- /dev/null
+++ b/memkind-1.10.0/src/memkind_regular.c
@@ -0,0 +1,109 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/heap_manager.h>
+
+#include <numa.h>
+
+static struct bitmask *regular_nodes_mask = NULL;
+
+static void regular_nodes_init(void)
+{
+    unsigned i;
+    unsigned nodes_num = (unsigned)numa_num_configured_nodes();
+    struct bitmask *node_cpus = numa_allocate_cpumask();
+
+    regular_nodes_mask = numa_allocate_nodemask();
+
+    for (i = 0; i < nodes_num; i++) {
+        numa_node_to_cpus(i, node_cpus);
+        if (numa_bitmask_weight(node_cpus))
+            numa_bitmask_setbit(regular_nodes_mask, i);
+    }
+    numa_bitmask_free(node_cpus);
+}
+
+static void memkind_regular_init_once(void)
+{
+    regular_nodes_init();
+    memkind_init(MEMKIND_REGULAR, true);
+}
+
+static int memkind_regular_check_available(struct memkind *kind)
+{
+    /* init_once method is called in memkind_malloc function
+     * when memkind malloc is not called this function will fail.
+     * Call pthread_once to be sure that situation mentioned
+     * above will never happen */
+    pthread_once(&kind->init_once, kind->ops->init_once);
+    return regular_nodes_mask != NULL ? MEMKIND_SUCCESS : MEMKIND_ERROR_UNAVAILABLE;
+}
+
+MEMKIND_EXPORT int memkind_regular_all_get_mbind_nodemask(struct memkind *kind,
+                                                          unsigned long *nodemask,
+                                                          unsigned long maxnode)
+{
+    struct bitmask nodemask_bm = {maxnode, nodemask};
+
+    if(!regular_nodes_mask) {
+        return MEMKIND_ERROR_UNAVAILABLE;
+    }
+
+    copy_bitmask_to_bitmask(regular_nodes_mask, &nodemask_bm);
+    return MEMKIND_SUCCESS;
+}
+
+static int memkind_regular_finalize(memkind_t kind)
+{
+    if(regular_nodes_mask)
+        numa_bitmask_free(regular_nodes_mask);
+
+    return memkind_arena_finalize(kind);
+}
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_REGULAR_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_regular_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_regular_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_regular_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_regular_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+
diff --git a/memkind-1.10.0/src/tbb_wrapper.c b/memkind-1.10.0/src/tbb_wrapper.c
new file mode 100644
index 0000000..869a5d4
--- /dev/null
+++ b/memkind-1.10.0/src/tbb_wrapper.c
@@ -0,0 +1,308 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/tbb_mem_pool_policy.h>
+#include <limits.h>
+
+#include <stdint.h>
+#include <errno.h>
+#include <stdio.h>
+#include <sys/mman.h>
+#include <stdlib.h>
+#include <dlfcn.h>
+#include <string.h>
+
+void *(*pool_malloc)(void *, size_t);
+void *(*pool_realloc)(void *, void *, size_t);
+void *(*pool_aligned_malloc)(void *, size_t, size_t);
+bool (*pool_free)(void *, void *);
+int (*pool_create_v1)(intptr_t, const struct MemPoolPolicy *, void **);
+bool (*pool_destroy)(void *);
+void *(*pool_identify)(void *object);
+size_t (*pool_msize)(void *, void *);
+
+static void *tbb_handle = NULL;
+static bool TBBInitDone = false;
+
+void load_tbb_symbols(void)
+{
+    const char so_name[]="libtbbmalloc.so.2";
+    tbb_handle = dlopen(so_name, RTLD_LAZY);
+    if(!tbb_handle) {
+        log_fatal("%s not found.", so_name);
+        abort();
+    }
+
+    pool_malloc = dlsym(tbb_handle, "_ZN3rml11pool_mallocEPNS_10MemoryPoolEm");
+    pool_realloc = dlsym(tbb_handle, "_ZN3rml12pool_reallocEPNS_10MemoryPoolEPvm");
+    pool_aligned_malloc = dlsym(tbb_handle,
+                                "_ZN3rml19pool_aligned_mallocEPNS_10MemoryPoolEmm");
+    pool_free = dlsym(tbb_handle, "_ZN3rml9pool_freeEPNS_10MemoryPoolEPv");
+    pool_create_v1 = dlsym(tbb_handle,
+                           "_ZN3rml14pool_create_v1ElPKNS_13MemPoolPolicyEPPNS_10MemoryPoolE");
+    pool_destroy = dlsym(tbb_handle, "_ZN3rml12pool_destroyEPNS_10MemoryPoolE");
+    pool_identify = dlsym(tbb_handle, "_ZN3rml13pool_identifyEPv");
+    pool_msize = dlsym(tbb_handle, "_ZN3rml10pool_msizeEPNS_10MemoryPoolEPv");
+
+    if(!pool_malloc ||
+       !pool_realloc ||
+       !pool_aligned_malloc ||
+       !pool_free ||
+       !pool_create_v1 ||
+       !pool_destroy ||
+       !pool_identify ) {
+        log_fatal("Could not find symbols in %s.", so_name);
+        dlclose(tbb_handle);
+        abort();
+    }
+    TBBInitDone = true;
+}
+
+//Granularity of raw_alloc allocations
+#define GRANULARITY 2*1024*1024
+static void *raw_alloc(intptr_t pool_id, size_t *bytes/*=n*GRANULARITY*/)
+{
+    void *ptr = kind_mmap((struct memkind *)pool_id, NULL, *bytes);
+    return (ptr==MAP_FAILED) ? NULL : ptr;
+}
+
+static int raw_free(intptr_t pool_id, void *raw_ptr, size_t raw_bytes)
+{
+    return munmap(raw_ptr, raw_bytes);
+}
+
+static void *tbb_pool_malloc(struct memkind *kind, size_t size)
+{
+    if(size_out_of_bounds(size)) return NULL;
+    void *result = pool_malloc(kind->priv, size);
+    if (!result)
+        errno = ENOMEM;
+    return result;
+}
+
+static void *tbb_pool_calloc(struct memkind *kind, size_t num, size_t size)
+{
+    if (size_out_of_bounds(num) || size_out_of_bounds(size)) return NULL;
+
+    const size_t array_size = num*size;
+    if (array_size/num != size) {
+        errno = ENOMEM;
+        return NULL;
+    }
+    void *result = pool_malloc(kind->priv, array_size);
+    if (result) {
+        memset(result, 0, array_size);
+    } else {
+        errno = ENOMEM;
+    }
+    return result;
+}
+
+static void *tbb_pool_common_realloc(void *pool, void *ptr, size_t size)
+{
+    if (size_out_of_bounds(size)) {
+        pool_free(pool, ptr);
+        return NULL;
+    }
+    void *result = pool_realloc(pool, ptr, size);
+    if (!result)
+        errno = ENOMEM;
+    return result;
+}
+
+static void *tbb_pool_realloc(struct memkind *kind, void *ptr, size_t size)
+{
+    return tbb_pool_common_realloc(kind->priv, ptr, size);
+}
+
+void *tbb_pool_realloc_with_kind_detect(void *ptr, size_t size)
+{
+    if (!ptr) {
+        errno = EINVAL;
+        return NULL;
+    }
+    return tbb_pool_common_realloc(pool_identify(ptr), ptr, size);
+}
+
+int tbb_get_global_stat(memkind_stat_type stat, size_t *value)
+{
+    log_err("Get allocator statistic is not supported by TBB");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+
+static int tbb_get_kind_stat(memkind_t kind, memkind_stat_type stat,
+                             size_t *value)
+{
+    log_err("Get kind statistic is not supported by TBB");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+
+int tbb_update_cached_stats(void)
+{
+    log_err("Update cached statistics is not supported by TBB");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+
+void *tbb_pool_defrag_reallocate_with_kind_detect(void *ptr)
+{
+    log_err("Defrag reallocate method is not supported by TBB");
+    return NULL;
+}
+
+static void *tbb_defrag_reallocate(struct memkind *kind, void *ptr)
+{
+    log_err("Defrag reallocate method is not supported by TBB");
+    return NULL;
+}
+
+struct memkind *tbb_detect_kind(void *ptr)
+{
+    if (!ptr) {
+        return NULL;
+    }
+    struct memkind *kind = NULL;
+    unsigned i;
+    void *pool = pool_identify(ptr);
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_get_kind_by_partition(i, &kind);
+        if (!err && kind->priv == pool) {
+            break;
+        }
+    }
+
+    return kind;
+}
+
+static int tbb_pool_posix_memalign(struct memkind *kind, void **memptr,
+                                   size_t alignment, size_t size)
+{
+    //Check if alignment is "at least as large as sizeof(void *)".
+    if(!alignment && (0 != (alignment & (alignment-sizeof(void *))))) return EINVAL;
+    //Check if alignment is "a power of 2".
+    if(alignment & (alignment-1)) return EINVAL;
+    if(size_out_of_bounds(size)) {
+        *memptr = NULL;
+        return 0;
+    }
+    void *result = pool_aligned_malloc(kind->priv, size, alignment);
+    if (!result) {
+        return ENOMEM;
+    }
+    *memptr = result;
+    return 0;
+}
+
+void tbb_pool_free_with_kind_detect(void *ptr)
+{
+    if (ptr) {
+        pool_free(pool_identify(ptr), ptr);
+    }
+}
+
+static void tbb_pool_free(struct memkind *kind, void *ptr)
+{
+    pool_free(kind->priv, ptr);
+}
+
+static size_t tbb_pool_common_malloc_usable_size(void *pool, void *ptr)
+{
+    if (pool_msize) {
+        return pool_msize(pool, ptr);
+    } else {
+        log_err("memkind_malloc_usable_size() is not supported by this TBB version.");
+        return 0;
+    }
+}
+
+static size_t tbb_pool_malloc_usable_size(struct memkind *kind, void *ptr)
+{
+    return tbb_pool_common_malloc_usable_size(kind->priv, ptr);
+}
+
+size_t tbb_pool_malloc_usable_size_with_kind_detect(void *ptr)
+{
+    size_t size = 0;
+    if (ptr) {
+        size = tbb_pool_common_malloc_usable_size(pool_identify(ptr), ptr);
+    }
+    return size;
+}
+
+static int tbb_update_memory_usage_policy(struct memkind *kind,
+                                          memkind_mem_usage_policy policy)
+{
+    log_info("Memkind memory usage policies have no effect in TBB manager.");
+    return MEMKIND_SUCCESS;
+}
+
+static int tbb_destroy(struct memkind *kind)
+{
+    bool pool_destroy_ret = pool_destroy(kind->priv);
+    dlclose(tbb_handle);
+
+    if(!pool_destroy_ret) {
+        log_err("TBB pool destroy failure.");
+        return MEMKIND_ERROR_OPERATION_FAILED;
+    }
+    return MEMKIND_SUCCESS;
+}
+
+void tbb_initialize(struct memkind *kind)
+{
+    if(!kind || !TBBInitDone) {
+        log_fatal("Failed to initialize TBB.");
+        abort();
+    }
+
+    struct MemPoolPolicy policy = {
+        .pAlloc = raw_alloc,
+        .pFree = raw_free,
+        .granularity = GRANULARITY,
+        .version = 1,
+        .fixedPool = false,
+        .keepAllMemory = false,
+        .reserved = 0
+    };
+
+    pool_create_v1((intptr_t)kind, &policy, &kind->priv);
+    if (!kind->priv) {
+        log_fatal("Unable to create TBB memory pool.");
+        abort();
+    }
+
+    kind->ops->malloc = tbb_pool_malloc;
+    kind->ops->calloc = tbb_pool_calloc;
+    kind->ops->posix_memalign = tbb_pool_posix_memalign;
+    kind->ops->realloc = tbb_pool_realloc;
+    kind->ops->free = tbb_pool_free;
+    kind->ops->finalize = tbb_destroy;
+    kind->ops->malloc_usable_size = tbb_pool_malloc_usable_size;
+    kind->ops->update_memory_usage_policy = tbb_update_memory_usage_policy;
+    kind->ops->get_stat = tbb_get_kind_stat;
+    kind->ops->defrag_reallocate = tbb_defrag_reallocate;
+}
diff --git a/memkind-1.10.0/test/Allocator.hpp b/memkind-1.10.0/test/Allocator.hpp
new file mode 100644
index 0000000..5156efc
--- /dev/null
+++ b/memkind-1.10.0/test/Allocator.hpp
@@ -0,0 +1,222 @@
+/*
+* Copyright (C) 2017 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <stdio.h>
+#include <assert.h>
+#include <map>
+#include <iterator>
+
+class Allocator
+{
+public:
+
+    virtual void *malloc(size_t size) = 0;
+    virtual void *calloc(size_t num, size_t size) = 0;
+    virtual void *realloc(void *ptr, size_t size) = 0;
+    virtual int memalign(void **ptr, size_t alignment, size_t size) = 0;
+    virtual void free(void *ptr) = 0;
+
+    /* get_numa_policy() returns MPOL_INTERLEAVE, MPOL_BIND, MPOL_PREFERRED
+     or -1 when the allocator is not providing NUMA policy */
+    virtual int get_numa_policy() = 0;
+    virtual bool is_high_bandwidth() = 0;
+    virtual size_t get_page_size() = 0;
+
+    virtual ~Allocator(void) {}
+};
+
+class MemkindAllocator : public Allocator
+{
+public:
+
+    MemkindAllocator(memkind_memtype_t memtype, memkind_policy_t policy,
+                     memkind_bits_t flags) :
+        memtype(memtype),
+        flags(flags)
+    {
+        int ret = memkind_create_kind(memtype, policy, flags, &kind);
+        assert(ret == MEMKIND_SUCCESS);
+        assert(kind != NULL);
+    }
+
+    MemkindAllocator(memkind_t kind) : kind(kind)
+    {
+        assert(kind != NULL);
+    }
+
+    virtual void *malloc(size_t size)
+    {
+        return memkind_malloc(kind, size);
+    }
+
+    virtual void *calloc(size_t num, size_t size)
+    {
+        return memkind_calloc(kind, num, size);
+    }
+
+    virtual void *realloc(void *ptr, size_t size)
+    {
+        return memkind_realloc(kind, ptr, size);
+    }
+
+    virtual int memalign(void **ptr, size_t alignment, size_t size)
+    {
+        return memkind_posix_memalign(kind, ptr, alignment, size);
+    }
+
+    virtual void free(void *ptr)
+    {
+        memkind_free(kind, ptr);
+    }
+
+    virtual int get_numa_policy()
+    {
+        static std::map<memkind_t, int> kind_policy = {
+            std::make_pair(MEMKIND_HBW_INTERLEAVE, MPOL_INTERLEAVE),
+            std::make_pair(MEMKIND_INTERLEAVE, MPOL_INTERLEAVE),
+            std::make_pair(MEMKIND_HBW_PREFERRED, MPOL_PREFERRED),
+            std::make_pair(MEMKIND_HBW_PREFERRED_HUGETLB, MPOL_PREFERRED),
+            std::make_pair(MEMKIND_HBW, MPOL_BIND),
+            std::make_pair(MEMKIND_HBW_HUGETLB, MPOL_BIND),
+            std::make_pair(MEMKIND_REGULAR, MPOL_BIND),
+            std::make_pair(MEMKIND_DEFAULT, MPOL_DEFAULT),
+            std::make_pair(MEMKIND_HUGETLB, MPOL_DEFAULT),
+            std::make_pair(MEMKIND_HBW_ALL_HUGETLB, MPOL_BIND),
+            std::make_pair(MEMKIND_HBW_ALL, MPOL_BIND)
+        };
+        auto it = kind_policy.find(kind);
+
+        if(it != std::end(kind_policy)) {
+            return it->second;
+        }
+        return -1;
+    }
+
+    virtual bool is_high_bandwidth()
+    {
+        return  memtype == MEMKIND_MEMTYPE_HIGH_BANDWIDTH ||
+                kind == MEMKIND_HBW ||
+                kind == MEMKIND_HBW_HUGETLB ||
+                kind == MEMKIND_HBW_PREFERRED ||
+                kind == MEMKIND_HBW_INTERLEAVE;
+    }
+
+    virtual size_t get_page_size()
+    {
+        if (kind == MEMKIND_HUGETLB ||
+            kind == MEMKIND_HBW_HUGETLB ||
+            kind == MEMKIND_HBW_PREFERRED_HUGETLB ||
+            (flags & MEMKIND_MASK_PAGE_SIZE_2MB)) {
+            return 2*MB;
+        }
+        return 4*KB;
+    }
+
+    virtual ~MemkindAllocator()
+    {
+        int ret = memkind_destroy_kind(kind);
+        assert(ret == MEMKIND_SUCCESS);
+        kind = NULL;
+    }
+
+private:
+    memkind_t kind = NULL;
+    memkind_memtype_t memtype = memkind_memtype_t();
+    memkind_bits_t flags = memkind_bits_t();
+};
+
+class HbwmallocAllocator : public Allocator
+{
+public:
+
+    HbwmallocAllocator(hbw_policy_t hbw_policy)
+    {
+        hbw_set_policy(hbw_policy);
+    }
+
+    virtual void *malloc(size_t size)
+    {
+        return hbw_malloc(size);
+    }
+
+    virtual void *calloc(size_t num, size_t size)
+    {
+        return hbw_calloc(num, size);
+    }
+
+    virtual void *realloc(void *ptr, size_t size)
+    {
+        return hbw_realloc(ptr, size);
+    }
+
+    virtual void set_memalign_page_size(hbw_pagesize_t psize)
+    {
+        page_size = psize;
+    }
+
+    virtual int memalign(void **ptr, size_t alignment, size_t size)
+    {
+        if (page_size == HBW_PAGESIZE_4KB) {
+            return hbw_posix_memalign(ptr, alignment, size);
+        } else {
+            return hbw_posix_memalign_psize(ptr, alignment, size, page_size);
+        }
+    }
+
+    virtual void free(void *ptr)
+    {
+        hbw_free(ptr);
+    }
+
+    virtual int get_numa_policy()
+    {
+        if (hbw_get_policy() == HBW_POLICY_INTERLEAVE) {
+            return MPOL_INTERLEAVE;
+        } else if (hbw_get_policy() == HBW_POLICY_PREFERRED) {
+            return MPOL_PREFERRED;
+        } else if (hbw_get_policy() == HBW_POLICY_BIND ||
+                   hbw_get_policy() == HBW_POLICY_BIND_ALL) {
+            return MPOL_BIND;
+        }
+        return -1;
+    }
+
+    virtual bool is_high_bandwidth()
+    {
+        return  hbw_check_available() == 0;
+    }
+
+    virtual size_t get_page_size()
+    {
+        if (page_size == HBW_PAGESIZE_2MB) {
+            return 2*MB;
+        }
+        return 4*KB;
+    }
+
+private:
+    hbw_pagesize_t page_size = HBW_PAGESIZE_4KB;
+};
+
diff --git a/memkind-1.10.0/test/Makefile.mk b/memkind-1.10.0/test/Makefile.mk
new file mode 100644
index 0000000..03a5621
--- /dev/null
+++ b/memkind-1.10.0/test/Makefile.mk
@@ -0,0 +1,341 @@
+#
+#  Copyright (C) 2014 - 2020 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+AM_CPPFLAGS += -Itest/gtest_fused -DMEMKIND_DEPRECATED\(x\)=x
+
+check_PROGRAMS += test/all_tests \
+                  test/allocator_perf_tool_tests \
+                  test/autohbw_test_helper \
+                  test/dax_kmem_test \
+                  test/decorator_test \
+                  test/environ_err_dax_kmem_malloc_test \
+                  test/environ_err_dax_kmem_malloc_positive_test \
+                  test/environ_err_hbw_malloc_test \
+                  test/freeing_memory_segfault_test \
+                  test/gb_page_tests_bind_policy \
+                  test/locality_test \
+                  test/memkind_stat_test \
+                  test/performance_test \
+                  test/trace_mechanism_test_helper \
+                  # end
+if HAVE_CXX11
+check_PROGRAMS += test/pmem_test \
+                  test/defrag_reallocate
+endif
+
+TESTS += test/test.sh
+
+EXTRA_DIST += test/autohbw_test.py \
+              test/draw_plots.py \
+              test/gtest_fused/gtest/gtest-all.cc \
+              test/gtest_fused/gtest/gtest.h \
+              test/hbw_detection_test.py \
+              test/dax_kmem_env_var_test.py \
+              test/memkind-afts-ext.ts \
+              test/memkind-afts.ts \
+              test/memkind-perf-ext.ts \
+              test/memkind-perf.ts \
+              test/memkind-pytests.ts \
+              test/memkind-slts.ts \
+              test/python_framework/__init__.py \
+              test/python_framework/cmd_helper.py \
+              test/python_framework/huge_page_organizer.py \
+              test/run_alloc_benchmark.sh \
+              test/trace_mechanism_test.py \
+              # end
+
+test_all_tests_LDADD = libmemkind.la
+test_allocator_perf_tool_tests_LDADD = libmemkind.la
+test_autohbw_test_helper_LDADD = libmemkind.la
+test_dax_kmem_test_LDADD = libmemkind.la
+test_decorator_test_LDADD = libmemkind.la
+test_environ_err_hbw_malloc_test_LDADD = libmemkind.la
+test_environ_err_dax_kmem_malloc_test_LDADD = libmemkind.la
+test_environ_err_dax_kmem_malloc_positive_test_LDADD = libmemkind.la
+test_freeing_memory_segfault_test_LDADD = libmemkind.la
+test_gb_page_tests_bind_policy_LDADD = libmemkind.la
+test_memkind_stat_test_LDADD = libmemkind.la
+test_trace_mechanism_test_helper_LDADD = libmemkind.la
+
+if HAVE_CXX11
+test_pmem_test_SOURCES = $(fused_gtest) test/memkind_pmem_config_tests.cpp test/memkind_pmem_long_time_tests.cpp test/memkind_pmem_tests.cpp
+test_pmem_test_LDADD = libmemkind.la
+test_defrag_reallocate_SOURCES = $(fused_gtest) test/memkind_defrag_reallocate.cpp
+test_defrag_reallocate_LDADD = libmemkind.la
+endif
+
+fused_gtest = test/gtest_fused/gtest/gtest-all.cc \
+              test/main.cpp \
+              # end
+
+test_all_tests_SOURCES = $(fused_gtest) \
+                         test/Allocator.hpp \
+                         test/TestPolicy.hpp \
+                         test/bat_tests.cpp \
+                         test/check.cpp \
+                         test/check.h \
+                         test/common.h \
+                         test/dlopen_test.cpp \
+                         test/error_message_tests.cpp \
+                         test/get_arena_test.cpp \
+                         test/hbw_allocator_tests.cpp \
+                         test/hbw_verify_function_test.cpp \
+                         test/memkind_allocator_tests.cpp \
+                         test/memkind_detect_kind_tests.cpp \
+                         test/memkind_null_kind_test.cpp \
+                         test/memkind_versioning_tests.cpp \
+                         test/multithreaded_tests.cpp \
+                         test/negative_tests.cpp \
+                         test/pmem_allocator_tests.cpp \
+                         test/static_kinds_list.h \
+                         test/static_kinds_tests.cpp \
+                         test/trial_generator.cpp \
+                         test/trial_generator.h \
+                         #end
+
+test_performance_test_SOURCES = $(fused_gtest) test/performance/perf_tests.cpp \
+                                test/performance/perf_tests.hpp \
+                                test/performance/framework.cpp \
+                                test/performance/framework.hpp \
+                                test/performance/operations.hpp \
+                                test/performance/perf_tests.hpp
+
+test_performance_test_LDADD = libmemkind.la
+
+test_locality_test_SOURCES = $(fused_gtest) test/allocator_perf_tool/Allocation_info.cpp test/locality_test.cpp
+test_locality_test_LDADD = libmemkind.la
+
+test_locality_test_CPPFLAGS = $(OPENMP_CFLAGS) -O0 -Wno-error $(AM_CPPFLAGS)
+test_locality_test_CXXFLAGS = $(OPENMP_CFLAGS) -O0 -Wno-error $(AM_CPPFLAGS)
+
+test_autohbw_test_helper_SOURCES = test/autohbw_test_helper.c
+test_decorator_test_SOURCES = $(fused_gtest) test/decorator_test.cpp test/decorator_test.h
+test_dax_kmem_test_SOURCES = $(fused_gtest) test/dax_kmem_nodes.h test/dax_kmem_nodes.cpp test/memkind_dax_kmem_test.cpp
+test_environ_err_hbw_malloc_test_SOURCES = test/environ_err_hbw_malloc_test.cpp
+test_environ_err_dax_kmem_malloc_test_SOURCES = test/environ_err_dax_kmem_malloc_test.cpp
+test_environ_err_dax_kmem_malloc_positive_test_SOURCES = test/environ_err_dax_kmem_malloc_positive_test.cpp
+test_freeing_memory_segfault_test_SOURCES = $(fused_gtest) test/freeing_memory_segfault_test.cpp
+test_gb_page_tests_bind_policy_SOURCES = $(fused_gtest) test/gb_page_tests_bind_policy.cpp test/trial_generator.cpp test/check.cpp
+test_memkind_stat_test_SOURCES = $(fused_gtest) test/memkind_stat_test.cpp
+test_trace_mechanism_test_helper_SOURCES = test/trace_mechanism_test_helper.c
+
+#Tests based on Allocator Perf Tool
+allocator_perf_tool_library_sources = test/allocator_perf_tool/AllocationSizes.hpp \
+                                      test/allocator_perf_tool/Allocation_info.hpp \
+                                      test/allocator_perf_tool/Allocation_info.cpp \
+                                      test/allocator_perf_tool/Allocator.hpp \
+                                      test/allocator_perf_tool/AllocatorFactory.hpp \
+                                      test/allocator_perf_tool/CSVLogger.hpp \
+                                      test/allocator_perf_tool/CommandLine.hpp \
+                                      test/allocator_perf_tool/Configuration.hpp \
+                                      test/allocator_perf_tool/ConsoleLog.hpp \
+                                      test/allocator_perf_tool/FunctionCalls.hpp \
+                                      test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp \
+                                      test/allocator_perf_tool/FunctionCallsPerformanceTask.h \
+                                      test/allocator_perf_tool/GTestAdapter.hpp \
+                                      test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/HugePageOrganizer.hpp \
+                                      test/allocator_perf_tool/HugePageUnmap.hpp \
+                                      test/allocator_perf_tool/Iterator.hpp \
+                                      test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/Numastat.hpp \
+                                      test/allocator_perf_tool/PmemMockup.cpp \
+                                      test/allocator_perf_tool/PmemMockup.hpp \
+                                      test/allocator_perf_tool/Runnable.hpp \
+                                      test/allocator_perf_tool/ScenarioWorkload.cpp \
+                                      test/allocator_perf_tool/ScenarioWorkload.h \
+                                      test/allocator_perf_tool/StandardAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/Stats.hpp \
+                                      test/allocator_perf_tool/StressIncreaseToMax.cpp \
+                                      test/allocator_perf_tool/StressIncreaseToMax.h \
+                                      test/allocator_perf_tool/Task.hpp \
+                                      test/allocator_perf_tool/TaskFactory.hpp \
+                                      test/allocator_perf_tool/Tests.hpp \
+                                      test/allocator_perf_tool/Thread.hpp \
+                                      test/allocator_perf_tool/TimerSysTime.hpp \
+                                      test/allocator_perf_tool/VectorIterator.hpp \
+                                      test/allocator_perf_tool/Workload.hpp \
+                                      test/allocator_perf_tool/WrappersMacros.hpp \
+                                      test/memory_manager.h \
+                                      test/proc_stat.h \
+                                      test/random_sizes_allocator.h \
+                                      # end
+
+
+test_allocator_perf_tool_tests_SOURCES = $(allocator_perf_tool_library_sources) \
+                                         $(fused_gtest) \
+                                         test/alloc_performance_tests.cpp \
+                                         test/allocate_to_max_stress_test.cpp \
+                                         test/hbw_allocator_performance_tests.cpp \
+                                         test/heap_manager_init_perf_test.cpp \
+                                         test/huge_page_test.cpp \
+                                         test/memory_footprint_test.cpp \
+                                         test/pmem_alloc_performance_tests.cpp \
+                                         # end
+
+
+test_allocator_perf_tool_tests_CPPFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_allocator_perf_tool_tests_CXXFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_allocator_perf_tool_tests_LDFLAGS = -lpthread -lnuma
+
+NUMAKIND_MAX = 2048
+test_all_tests_CXXFLAGS = $(AM_CXXFLAGS) $(CXXFLAGS) $(OPENMP_CFLAGS) -DNUMAKIND_MAX=$(NUMAKIND_MAX)
+test_all_tests_LDFLAGS = -ldl
+
+#Allocator Perf Tool stand alone app
+check_PROGRAMS += test/perf_tool
+test_perf_tool_LDADD = libmemkind.la
+test_perf_tool_SOURCES = $(allocator_perf_tool_library_sources) \
+                         test/allocator_perf_tool/main.cpp \
+                         # end
+
+
+test_perf_tool_CPPFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_perf_tool_CXXFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_perf_tool_LDFLAGS = -lpthread -lnuma
+if HAVE_CXX11
+test_perf_tool_CPPFLAGS += -std=c++11
+test_perf_tool_CXXFLAGS += -std=c++11
+endif
+
+#Alloc benchmark
+check_PROGRAMS += test/alloc_benchmark_glibc \
+                  test/alloc_benchmark_hbw \
+                  test/alloc_benchmark_pmem \
+                  test/alloc_benchmark_tbb \
+                  # end
+
+test_alloc_benchmark_glibc_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall
+test_alloc_benchmark_glibc_LDADD = libmemkind.la
+test_alloc_benchmark_glibc_SOURCES = test/alloc_benchmark.c
+
+test_alloc_benchmark_hbw_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall -DHBWMALLOC
+test_alloc_benchmark_hbw_LDADD = libmemkind.la
+test_alloc_benchmark_hbw_LDFLAGS = -lmemkind
+test_alloc_benchmark_hbw_SOURCES = test/alloc_benchmark.c
+
+test_alloc_benchmark_pmem_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall -DPMEMMALLOC
+test_alloc_benchmark_pmem_LDADD = libmemkind.la
+test_alloc_benchmark_pmem_LDFLAGS = -ldl
+test_alloc_benchmark_pmem_SOURCES = test/alloc_benchmark.c
+
+test_alloc_benchmark_tbb_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall -DTBBMALLOC
+test_alloc_benchmark_tbb_LDADD = libmemkind.la
+test_alloc_benchmark_tbb_LDFLAGS = -ldl
+test_alloc_benchmark_tbb_SOURCES = test/alloc_benchmark.c \
+                                   test/load_tbbmalloc_symbols.c \
+                                   test/tbbmalloc.h \
+                                   # end
+
+# Pmem fragmentation benchmark
+check_PROGRAMS += test/fragmentation_benchmark_pmem
+test_fragmentation_benchmark_pmem_LDADD = libmemkind.la
+test_fragmentation_benchmark_pmem_SOURCES = test/fragmentation_benchmark_pmem.cpp
+test_fragmentation_benchmark_pmem_CXXFLAGS = -O0 -Wall
+if HAVE_CXX11
+test_fragmentation_benchmark_pmem_CXXFLAGS += -std=c++11
+endif
+
+# Examples as tests
+check_PROGRAMS += test/autohbw_candidates \
+                  test/filter_memkind \
+                  test/hello_hbw \
+                  test/hello_memkind \
+                  test/hello_memkind_debug \
+                  test/memkind_get_stat \
+                  test/pmem_alignment \
+                  test/pmem_and_dax_kmem_kind \
+                  test/pmem_and_default_kind \
+                  test/pmem_config \
+                  test/pmem_detect_kind \
+                  test/pmem_free_with_unknown_kind \
+                  test/pmem_kinds \
+                  test/pmem_malloc \
+                  test/pmem_malloc_unlimited \
+                  test/pmem_multithreads \
+                  test/pmem_multithreads_onekind \
+                  test/pmem_usable_size \
+                  # end
+if HAVE_CXX11
+check_PROGRAMS += test/memkind_allocated \
+                  test/memkind_cpp_allocator \
+                  test/pmem_cpp_allocator
+endif
+
+test_autohbw_candidates_LDADD = libmemkind.la
+test_filter_memkind_LDADD = libmemkind.la
+test_hello_hbw_LDADD = libmemkind.la
+test_hello_memkind_LDADD = libmemkind.la
+test_hello_memkind_debug_LDADD = libmemkind.la
+test_memkind_get_stat_LDADD = libmemkind.la
+test_pmem_alignment_LDADD = libmemkind.la
+test_pmem_and_dax_kmem_kind_LDADD = libmemkind.la
+test_pmem_and_default_kind_LDADD = libmemkind.la
+test_pmem_config_LDADD = libmemkind.la
+test_pmem_detect_kind_LDADD = libmemkind.la
+test_pmem_free_with_unknown_kind_LDADD = libmemkind.la
+test_pmem_kinds_LDADD = libmemkind.la
+test_pmem_malloc_LDADD = libmemkind.la
+test_pmem_malloc_unlimited_LDADD = libmemkind.la
+test_pmem_multithreads_LDADD = libmemkind.la
+test_pmem_multithreads_onekind_LDADD = libmemkind.la
+test_pmem_usable_size_LDADD = libmemkind.la
+if HAVE_CXX11
+test_memkind_allocated_LDADD = libmemkind.la
+test_memkind_cpp_allocator_LDADD = libmemkind.la
+test_pmem_cpp_allocator_LDADD = libmemkind.la
+endif
+
+test_autohbw_candidates_SOURCES = examples/autohbw_candidates.c
+test_filter_memkind_SOURCES = examples/filter_example.c
+test_hello_hbw_SOURCES = examples/hello_hbw_example.c
+test_hello_memkind_SOURCES = examples/hello_memkind_example.c
+test_hello_memkind_debug_SOURCES = examples/hello_memkind_example.c examples/memkind_decorator_debug.c
+test_memkind_get_stat_SOURCES = examples/memkind_get_stat.c
+test_pmem_alignment_SOURCES = examples/pmem_alignment.c
+test_pmem_and_dax_kmem_kind_SOURCES = examples/pmem_and_dax_kmem_kind.c
+test_pmem_and_default_kind_SOURCES = examples/pmem_and_default_kind.c
+test_pmem_config_SOURCES = examples/pmem_config.c
+test_pmem_detect_kind_SOURCES = examples/pmem_detect_kind.c
+test_pmem_free_with_unknown_kind_SOURCES = examples/pmem_free_with_unknown_kind.c
+test_pmem_kinds_SOURCES = examples/pmem_kinds.c
+test_pmem_malloc_SOURCES = examples/pmem_malloc.c
+test_pmem_malloc_unlimited_SOURCES = examples/pmem_malloc_unlimited.c
+test_pmem_multithreads_SOURCES = examples/pmem_multithreads.c
+test_pmem_multithreads_onekind_SOURCES = examples/pmem_multithreads_onekind.c
+test_pmem_usable_size_SOURCES = examples/pmem_usable_size.c
+
+test_libautohbw_la_SOURCES = autohbw/autohbw.c
+noinst_LTLIBRARIES += test/libautohbw.la
+if HAVE_CXX11
+test_memkind_allocated_SOURCES = examples/memkind_allocated_example.cpp examples/memkind_allocated.hpp
+test_memkind_cpp_allocator_SOURCES = examples/memkind_cpp_allocator.cpp
+test_pmem_cpp_allocator_SOURCES = examples/pmem_cpp_allocator.cpp
+endif
+
+clean-local: test-clean
+
+test-clean:
+	find test \( -name "*.gcda" -o -name "*.gcno" \) -type f -delete
diff --git a/memkind-1.10.0/test/README b/memkind-1.10.0/test/README
new file mode 100644
index 0000000..2f7bbcd
--- /dev/null
+++ b/memkind-1.10.0/test/README
@@ -0,0 +1,85 @@
+MEMKIND TESTS
+=============
+
+
+DISCLAIMER
+==========
+SEE COPYING FILE FOR LICENSE INFORMATION.
+
+THIS SOFTWARE IS PROVIDED AS A DEVELOPMENT SNAPSHOT TO AID
+COLLABORATION AND WAS NOT ISSUED AS A RELEASED PRODUCT BY INTEL.
+
+LAST UPDATE
+===========
+Agata Wozniak <agata.wozniak@intel.com>
+Wednesday May 11 2016
+
+TESTS SCENARIOS
+===============
+
+memkind-afts
+------------
+This is scenario for basic functional tests. Maximum number of allocations per numa node:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+
+memkind-afts-ext
+----------------
+The same tests as in memkind-afts, however this tests are more ravenous and needs more memory
+available. Special test cases prepared only for this scenario are described as ravenous
+and needs at least:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+memory available per numa node.
+
+memkind-perf
+------------
+This is scenario for performance tests, excluding tests that allocate more than:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+
+memkind-perf-ext
+----------------
+This test scenario extend memkind-perf for tests that allocate more than:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+AllocPerformanceTest is a group of performance tests for *alloc() execution. TC properties:
+- 6 kinds (MEMKIND_DEFAULT, MEMKIND_HBW, MEMKIND_INTERLEAVE, MEMKIND_HBW_INTERLEAVE, MEMKIND_HBW_PREFERRED, PMEM kind)
+- 3 calls (malloc, calloc, realloc)
+- 5 allocation sizes (100bytes, 4096bytes, 1000bytes, 1001bytes, 1.5MB - 1572864bytes)
+- single thread and multithreaded (10threads and 72threads)
+All tests record three values:
+- total_time_spend_on_alloc (total time spend on *alloc() call during the test - less is better)
+- alloc_operations_per_thread (number of done allocations per thread - this value is constant)
+- ref_delta_time_percent (percent delta between standard allocator and current tested kind - less is better)
+
+memkind-slts
+------------
+This is scenario for all stress and longevity tests.
+
+BENCHMARKS
+===============
+
+fragmentation_benchmark_pmem
+------------
+The program logging memory fragmentation of specific workload in file
+with date and timestamps ("bench_single_thread_yyyymmdd-hhmm_mem_<pmem_memory_usage_policy>.log").
+
+The program takes, as arguments:
+* pmem_path - path where file-backed memory kind should be created
+* pmem_size - file-backed memory kind size limit, given in megabytes
+* pmem_memory_usage_policy - 0 - MEMKIND_MEM_USAGE_POLICY_DEFAULT, 1 - MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE
+* test_time - time how long program runs, given in seconds
+
+For example:
+
+    ./fragmentation_benchmark_pmem /mnt/pmem 17 0 60
+
+This will create PMEM kind limited with 17 MB in /mnt/pmem/ with default memory usage policy
+and save results in log file after 60 seconds.
+
+MANUAL TESTING
+==============
+In order to run tests on local machine, test.sh should be executed.
+When test.sh is run, all tests are executed as long as -d (skip MCDRAM nodes detection), -m (skip
+tests that require 2MB pages) parameters are not used.
diff --git a/memkind-1.10.0/test/TestPolicy.hpp b/memkind-1.10.0/test/TestPolicy.hpp
new file mode 100644
index 0000000..91d534f
--- /dev/null
+++ b/memkind-1.10.0/test/TestPolicy.hpp
@@ -0,0 +1,177 @@
+/*
+ * Copyright (C) 2017 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include <memkind/internal/memkind_hbw.h>
+
+#include "common.h"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+#include "numa.h"
+
+#include <memory>
+
+namespace TestPolicy
+{
+    typedef std::unique_ptr<struct bitmask, decltype(&numa_free_nodemask)>
+    unique_bitmask_ptr;
+
+    unique_bitmask_ptr make_nodemask_ptr()
+    {
+        return unique_bitmask_ptr(numa_allocate_nodemask(), numa_free_nodemask);
+    }
+
+    unique_bitmask_ptr make_cpumask_ptr()
+    {
+        return unique_bitmask_ptr(numa_allocate_cpumask(), numa_free_nodemask);
+
+    }
+
+    int get_num_of_pages(const size_t size, const size_t page_size)
+    {
+        size_t pages_number = size / page_size;
+        pages_number += size % page_size ? 1 : 0;
+        return pages_number;
+    }
+
+    std::vector<void *> get_address_of_pages(const void *ptr,
+                                             const size_t pages_number, const size_t page_size)
+    {
+        std::vector<void *> address(pages_number);
+        const size_t page_mask = ~(page_size-1);
+        address[0] = (void *)((uintptr_t)ptr &
+                              page_mask); //aligned address of first page
+        for (size_t page_num = 1; page_num < pages_number; page_num++) {
+            address[page_num] = (char *)address[page_num-1] + page_size;
+        }
+        return address;
+    }
+
+    void record_page_association(const void *ptr, const size_t size,
+                                 const size_t page_size)
+    {
+        size_t pages_number = get_num_of_pages(size, page_size);
+        std::vector<void *> address = get_address_of_pages(ptr, pages_number,
+                                                           page_size);
+
+        int max_node_id = numa_max_node();
+        std::vector<int> nodes(pages_number);
+        std::vector<int> pages_on_node(max_node_id+1);
+
+        if (move_pages(0, pages_number, address.data(), NULL, nodes.data(),
+                       MPOL_MF_MOVE)) {
+            fprintf(stderr, "Error: move_pages() returned %s\n", strerror(errno));
+            return;
+        }
+
+        for (size_t i = 0; i < pages_number; i++) {
+            if (nodes[i] < 0) {
+                fprintf(stderr,"Error: status of page %p is %d\n", address[i], nodes[i]);
+                return;
+            } else {
+                pages_on_node[nodes[i]]++;
+            }
+        }
+
+        for (size_t i = 0; i < (size_t)max_node_id + 1; i++) {
+            if (pages_on_node[i] > 0) {
+                char buffer[1024];
+                snprintf(buffer, sizeof(buffer), "Node%zd", i);
+                GTestAdapter::RecordProperty(buffer, pages_on_node[i]);
+            }
+        }
+    }
+
+    void check_numa_nodes(unique_bitmask_ptr &expected_bitmask, int policy,
+                          void *ptr, size_t size)
+    {
+
+        const size_t page_size = sysconf(_SC_PAGESIZE);
+        size_t pages_number = get_num_of_pages(size, page_size);
+        std::vector<void *> address = get_address_of_pages(ptr, pages_number,
+                                                           page_size);
+        unique_bitmask_ptr returned_bitmask = make_nodemask_ptr();
+        int status = -1;
+
+        for (size_t page_num = 0; page_num < address.size(); page_num++) {
+            ASSERT_EQ(0, get_mempolicy(&status, returned_bitmask->maskp,
+                                       returned_bitmask->size, address[page_num], MPOL_F_ADDR));
+            ASSERT_EQ(policy, status);
+            switch(policy) {
+                case MPOL_INTERLEAVE:
+                    EXPECT_TRUE(numa_bitmask_equal(expected_bitmask.get(), returned_bitmask.get()));
+                    break;
+                case MPOL_DEFAULT:
+                    break;
+                case MPOL_BIND:
+                case MPOL_PREFERRED:
+                    for(int i=0; i < numa_num_possible_nodes(); i++) {
+                        if(numa_bitmask_isbitset(returned_bitmask.get(), i)) {
+                            EXPECT_TRUE(numa_bitmask_isbitset(expected_bitmask.get(), i));
+                        }
+                    }
+                    break;
+                default:
+                    assert(!"Unknown policy\n");
+            }
+        }
+    }
+
+    void check_hbw_numa_nodes(int policy, void *ptr, size_t size)
+    {
+        unique_bitmask_ptr expected_bitmask = make_nodemask_ptr();
+
+        memkind_hbw_all_get_mbind_nodemask(NULL, expected_bitmask->maskp,
+                                           expected_bitmask->size);
+        check_numa_nodes(expected_bitmask, policy, ptr, size);
+    }
+
+    void check_all_numa_nodes(int policy, void *ptr, size_t size)
+    {
+        if (policy != MPOL_INTERLEAVE && policy != MPOL_DEFAULT) return;
+
+        unique_bitmask_ptr expected_bitmask = make_nodemask_ptr();
+
+        for(int i=0; i < numa_num_configured_nodes(); i++) {
+            numa_bitmask_setbit(expected_bitmask.get(), i);
+        }
+
+        check_numa_nodes(expected_bitmask, policy, ptr, size);
+    }
+
+    std::set<int> get_regular_numa_nodes(void)
+    {
+        struct bitmask *cpu_mask = numa_allocate_cpumask();
+        std::set<int> regular_nodes;
+
+        const int MAXNODE_ID = numa_max_node();
+        for (int id = 0; id <= MAXNODE_ID; ++id) {
+            numa_node_to_cpus(id, cpu_mask);
+
+            if (numa_bitmask_weight(cpu_mask) != 0) {
+                regular_nodes.insert(id);
+            }
+        }
+        numa_free_cpumask(cpu_mask);
+
+        return regular_nodes;
+    }
+}
diff --git a/memkind-1.10.0/test/alloc_benchmark.c b/memkind-1.10.0/test/alloc_benchmark.c
new file mode 100644
index 0000000..698974c
--- /dev/null
+++ b/memkind-1.10.0/test/alloc_benchmark.c
@@ -0,0 +1,213 @@
+/*
+ * Copyright (C) 2016 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <sys/time.h>
+#include <unistd.h>
+#include <stdint.h>
+#include <limits.h>
+#ifdef _OPENMP
+#include <omp.h>
+#endif
+#if defined(HBWMALLOC)
+#include <hbwmalloc.h>
+#define MALLOC_FN hbw_malloc
+#define FREE_FN hbw_free
+#elif defined (TBBMALLOC)
+#include "tbbmalloc.h"
+void *(*scalable_malloc)(size_t);
+void *(*scalable_realloc)(void *, size_t);
+void *(*scalable_calloc)(size_t, size_t);
+void  (*scalable_free)(void *);
+#define MALLOC_FN scalable_malloc
+#define FREE_FN scalable_free
+#elif defined (PMEMMALLOC)
+#include <sys/stat.h>
+#include "memkind.h"
+#define MALLOC_FN(x) memkind_malloc(pmem_bench_kind, (x))
+#define FREE_FN(x) memkind_free(pmem_bench_kind, (x))
+
+static const size_t PMEM_PART_SIZE = 0;
+static const char *PMEM_DIR = "/tmp/";
+static memkind_t pmem_bench_kind;
+#else
+#define MALLOC_FN malloc
+#define FREE_FN free
+#endif
+
+double ctimer(void);
+void usage(char *name);
+
+int main(int argc, char *argv[])
+{
+#ifdef _OPENMP
+    int nthr = omp_get_max_threads();
+#else
+    int nthr = 1;
+#endif
+    long n, size;
+    size_t alloc_size;
+    unsigned long i;
+    double dt, t_start, t_end, t_malloc, t_free, t_first_malloc, t_first_free,
+           malloc_time = 0.0, free_time = 0.0, first_malloc_time, first_free_time;
+    void *ptr;
+#ifdef TBBMALLOC
+    int ret;
+
+    ret = load_tbbmalloc_symbols();
+    if (ret) {
+        printf("Error: TBB symbols not loaded (ret: %d)\n", ret);
+        return EXIT_FAILURE;
+    }
+#endif
+#ifdef PMEMMALLOC
+    struct stat st;
+
+    /* Handle command line arguments */
+    if (argc == 3 || argc == 4) {
+        n = atol(argv[1]);
+        size = atol(argv[2]);
+
+        if (argc == 4) {
+            if (stat(argv[3], &st) != 0 || !S_ISDIR(st.st_mode)) {
+                usage(argv[0]);
+                return EXIT_FAILURE;
+            } else {
+                PMEM_DIR = argv[3];
+            }
+        }
+    }
+    if ((argc != 3 && argc != 4) || n < 0 || size < 0 || size > (LONG_MAX >> 10)) {
+        usage(argv[0]);
+        return EXIT_FAILURE;
+    }
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_bench_kind);
+    if (err) {
+        printf("Error: memkind_create_pmem failed %d\n", err);
+        return EXIT_FAILURE;
+    }
+#else
+    /* Handle command line arguments */
+    if (argc == 3) {
+        n = atol(argv[1]);
+        size = atol(argv[2]);
+    }
+    if (argc != 3 || n < 0 || size < 0 || size > (LONG_MAX >> 10)) {
+        usage(argv[0]);
+        return EXIT_FAILURE;
+    }
+#endif
+    alloc_size = (size_t) size * 1024;
+
+    /* Get pagesize and compute page_mask */
+    const size_t page_size = sysconf(_SC_PAGESIZE);
+    const size_t page_mask = ~(page_size-1);
+
+    /* Warm up */
+    t_first_malloc = ctimer();
+    ptr = MALLOC_FN(alloc_size);
+    first_malloc_time = ctimer() - t_first_malloc;
+    if (ptr == NULL) {
+        printf("Error: first allocation failed\n");
+        return EXIT_FAILURE;
+    }
+    t_first_free = ctimer();
+    FREE_FN(ptr);
+    first_free_time = ctimer() - t_first_free;
+    ptr = NULL;
+
+    t_start = ctimer();
+    #pragma omp parallel private(i,t_malloc,t_free,ptr) reduction(max:malloc_time,free_time)
+    {
+        malloc_time = 0.0;
+        free_time = 0.0;
+        for (i=0; i<n; i++) {
+            t_malloc = ctimer();
+            ptr = (void *) MALLOC_FN(alloc_size);
+            malloc_time += ctimer() - t_malloc;
+            #pragma omp critical
+            {
+                if (ptr == NULL) {
+                    printf("Error: allocation failed\n");
+                    exit(EXIT_FAILURE);
+                }
+            }
+
+            /* Make sure to touch every page */
+            char *end = ptr + alloc_size;
+            char *aligned_beg = (char *)((uintptr_t)ptr & page_mask);
+            while(aligned_beg < end) {
+                char *temp_ptr = (char *) aligned_beg;
+                char value = temp_ptr[0];
+                temp_ptr[0] = value;
+                aligned_beg += page_size;
+            }
+
+            t_free = ctimer();
+            FREE_FN(ptr);
+            free_time += ctimer() - t_free;
+            ptr = NULL;
+        }
+    }
+    t_end = ctimer();
+    dt = t_end - t_start;
+
+    printf("%d %lu %8.6f %8.6f  %8.6f  %8.6f  %8.6f\n",
+           nthr, size, dt/n, malloc_time/n, free_time/n, first_malloc_time,
+           first_free_time);
+#ifdef PMEMMALLOC
+    err = memkind_destroy_kind(pmem_bench_kind);
+    if (err) {
+        printf("Error: memkind_destroy_kind failed %d\n", err);
+        return EXIT_FAILURE;
+    }
+#endif
+    return EXIT_SUCCESS;
+}
+
+void usage(char *name)
+{
+#ifdef PMEMMALLOC
+    printf("Usage: %s <N> <SIZE> [DIR], where \n"
+           "N is an number of repetitions \n"
+           "SIZE is an allocation size in kbytes\n"
+           "DIR is a custom path for PMEM kind, (default: \"/tmp/\")\n",
+           name);
+#else
+    printf("Usage: %s <N> <SIZE>, where \n"
+           "N is an number of repetitions \n"
+           "SIZE is an allocation size in kbytes\n", name);
+#endif
+}
+
+inline double ctimer()
+{
+    struct timeval tmr;
+    gettimeofday(&tmr, NULL);
+    /* Return time in ms */
+    return (tmr.tv_sec + tmr.tv_usec/1000000.0)*1000;
+}
diff --git a/memkind-1.10.0/test/alloc_performance_tests.cpp b/memkind-1.10.0/test/alloc_performance_tests.cpp
new file mode 100644
index 0000000..35e9cdc
--- /dev/null
+++ b/memkind-1.10.0/test/alloc_performance_tests.cpp
@@ -0,0 +1,1641 @@
+/*
+* Copyright (C) 2016 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/TaskFactory.hpp"
+#include "allocator_perf_tool/Stats.hpp"
+#include "allocator_perf_tool/Thread.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+
+class AllocPerformanceTest: public :: testing::Test
+{
+private:
+    AllocatorFactory allocator_factory;
+
+protected:
+    void SetUp()
+    {
+        allocator_factory.initialize_allocator(AllocatorTypes::STANDARD_ALLOCATOR);
+    }
+
+    void TearDown()
+    {}
+
+    float run(unsigned kind, unsigned call, size_t threads_number,
+              size_t alloc_size, unsigned mem_operations_num)
+    {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+        TypesConf func_calls;
+        TypesConf allocator_types;
+
+        func_calls.enable_type(FunctionCalls::FREE);
+        func_calls.enable_type(call);
+        allocator_types.enable_type(kind);
+
+        TaskConf conf = {
+            mem_operations_num, //number of memory operations
+            {
+                mem_operations_num, //number of memory operations
+                alloc_size, //min. size of single allocation
+                alloc_size //max. size of single allocatioion
+            },
+            func_calls, //enable function calls
+            allocator_types, //enable allocators
+            11, //random seed
+            false, //do not log memory operations and statistics to csv file
+        };
+
+        for (int i=0; i<threads_number; i++) {
+            Task *task = task_factory.create(conf);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats time_stats;
+        for (int i=0; i<tasks.size(); i++) {
+            time_stats += tasks[i]->get_results();
+        }
+
+        return time_stats.stats[kind][call].total_time;
+    }
+
+    void run_test(unsigned kind, unsigned call, size_t threads_number,
+                  size_t alloc_size, unsigned mem_operations_num)
+    {
+        allocator_factory.initialize_allocator(kind);
+        float ref_time = run(AllocatorTypes::STANDARD_ALLOCATOR, call, threads_number,
+                             alloc_size, mem_operations_num);
+        float perf_time = run(kind, call, threads_number, alloc_size,
+                              mem_operations_num);
+        float ref_delta_time_percent = allocator_factory.calc_ref_delta(ref_time,
+                                                                        perf_time);
+
+        GTestAdapter::RecordProperty("total_time_spend_on_alloc", perf_time);
+        GTestAdapter::RecordProperty("alloc_operations_per_thread", mem_operations_num);
+        GTestAdapter::RecordProperty("ref_delta_time_percent", ref_delta_time_percent);
+    }
+
+};
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_ext_MEMKIND_HBW_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
diff --git a/memkind-1.10.0/test/allocate_to_max_stress_test.cpp b/memkind-1.10.0/test/allocate_to_max_stress_test.cpp
new file mode 100644
index 0000000..4895ab7
--- /dev/null
+++ b/memkind-1.10.0/test/allocate_to_max_stress_test.cpp
@@ -0,0 +1,176 @@
+/*
+ * Copyright (C) 2015 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <chrono>
+#include "common.h"
+#include "allocator_perf_tool/Configuration.hpp"
+#include "allocator_perf_tool/StressIncreaseToMax.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+//memkind stress and longevity tests using Allocatr Perf Tool.
+class AllocateToMaxStressTests: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+    //Allocates memory up to 'memory_request_limit'.
+    void run(TypesConf kinds, TypesConf func_calls, unsigned operations,
+             size_t size_from, size_t size_to, size_t memory_request_limit,
+             bool touch_memory)
+    {
+        RecordProperty("memory_operations", operations);
+        RecordProperty("size_from", size_from);
+        RecordProperty("size_to", size_to);
+
+        TaskConf task_conf = {
+            .n = operations, //number of memory operations
+            .allocation_sizes_conf = {
+                operations, //number of memory operations
+                size_from, //no random sizes.
+                size_to
+            },
+            .func_calls = func_calls, //enable allocator function call
+            .allocators_types = kinds, //enable allocator
+            .seed = 11, //random seed
+            .touch_memory = touch_memory //enable or disable touching memory
+        };
+
+        std::chrono::time_point<std::chrono::system_clock> start, end;
+        start = std::chrono::system_clock::now();
+
+        //Execute test iterations.
+        std::vector<iteration_result> results =
+            StressIncreaseToMax::execute_test_iterations(task_conf, 120,
+                                                         memory_request_limit);
+
+        end = std::chrono::system_clock::now();
+
+        std::chrono::duration<double> elapsed_time = end - start;
+
+        RecordProperty("elapsed_time", elapsed_time.count());
+
+        //Check finish status.
+        EXPECT_EQ(check_allocation_errors(results, task_conf), 0);
+    }
+
+    //Check true allocation errors over all iterations.
+    //Return iteration number (>0) when error occurs, or zero
+    int check_allocation_errors(std::vector<iteration_result> &results,
+                                const TaskConf &task_conf)
+    {
+        for (size_t i=0; i<results.size(); i++) {
+            //Check if test ends with allocation error.
+            if(results[i].is_allocation_error) {
+                return i+1;
+            }
+        }
+
+        return 0;
+    }
+};
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_MEMKIND_HBW)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW), TypesConf(FunctionCalls::MALLOC),
+        1024, MB, MB, GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_MEMKIND_INTERLEAVE)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_INTERLEAVE),
+        TypesConf(FunctionCalls::MALLOC), 4096, MB, MB, 4*GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_MEMKIND_HBW_PREFERRED)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW_PREFERRED),
+        TypesConf(FunctionCalls::MALLOC), 17408, MB, MB, 17*GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_2MBPages_slts_ext_ALLOCATE_TO_MAX_MEMKIND_HBW_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(2250);
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW_HUGETLB),
+        TypesConf(FunctionCalls::MALLOC), 1024, 4*MB, 4*MB, GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_DIFFERENT_SIZES)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW), TypesConf(FunctionCalls::MALLOC),
+        2500, 1, 8*MB, GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_AND_FREE_MEMKIND_DEFAULT)
+{
+    TypesConf func_calls;
+    func_calls.enable_type(FunctionCalls::MALLOC);
+    func_calls.enable_type(FunctionCalls::FREE);
+    run(TypesConf(AllocatorTypes::MEMKIND_DEFAULT), func_calls, 2500, 500*MB, 8*GB,
+        16*GB, false);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_AND_FREE_MEMKIND_REGULAR)
+{
+    TypesConf func_calls;
+    func_calls.enable_type(FunctionCalls::MALLOC);
+    func_calls.enable_type(FunctionCalls::FREE);
+    run(TypesConf(AllocatorTypes::MEMKIND_REGULAR), func_calls, 2500, 500*MB, 8*GB,
+        16*GB, false);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_DIFFERENT_KINDS)
+{
+    TypesConf kinds;
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_PREFERRED);
+    kinds.enable_type(AllocatorTypes::MEMKIND_DEFAULT);
+    kinds.enable_type(AllocatorTypes::MEMKIND_INTERLEAVE);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_INTERLEAVE);
+    kinds.enable_type(AllocatorTypes::MEMKIND_REGULAR);
+    run(kinds, TypesConf(FunctionCalls::MALLOC), 2048, MB, MB, 2*GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ext_ALLOCATE_TO_MAX_DIFFERENT_KINDS_WITH_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(2250);
+    TypesConf kinds;
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_HUGETLB);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB);
+    run(kinds, TypesConf(FunctionCalls::MALLOC), 2048, MB, MB, 2*GB, true);
+}
diff --git a/memkind-1.10.0/test/allocator_perf_tool/AllocationSizes.hpp b/memkind-1.10.0/test/allocator_perf_tool/AllocationSizes.hpp
new file mode 100644
index 0000000..59b780d
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/AllocationSizes.hpp
@@ -0,0 +1,60 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <vector>
+#include <cstdlib>
+
+#include "VectorIterator.hpp"
+
+class AllocationSizes
+{
+public:
+
+    static VectorIterator<size_t> generate_random_sizes(int sizes_num, size_t from,
+                                                        size_t to, int seed)
+    {
+        srand(seed);
+        std::vector<size_t> sizes;
+        size_t range = to - from;
+
+        if(from == to) {
+            for (int i=0; i<sizes_num; i++)
+                sizes.push_back(from);
+        } else {
+            for (int i=0; i<sizes_num; i++)
+                sizes.push_back(rand() % (range-1)+from);
+        }
+
+        return VectorIterator<size_t>::create(sizes);
+    }
+
+    static VectorIterator<size_t> generate_random_sizes(AllocationSizesConf conf,
+                                                        int seed)
+    {
+        return generate_random_sizes(conf.n, conf.size_from, conf.size_to, seed);
+    }
+
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Allocation_info.cpp b/memkind-1.10.0/test/allocator_perf_tool/Allocation_info.cpp
new file mode 100644
index 0000000..8f345c5
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Allocation_info.cpp
@@ -0,0 +1,39 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "Allocation_info.hpp"
+#include <numaif.h>
+
+double convert_bytes_to_mb(uint64_t bytes)
+{
+    return bytes / (1024.0 * 1024.0);
+}
+
+int get_numa_node_id(void *ptr)
+{
+    int status = -1;
+
+    get_mempolicy(&status, NULL, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+    return status;
+}
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Allocation_info.hpp b/memkind-1.10.0/test/allocator_perf_tool/Allocation_info.hpp
new file mode 100644
index 0000000..c33c49a
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Allocation_info.hpp
@@ -0,0 +1,44 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <stdint.h>
+#include <stddef.h>
+
+
+
+//This structure is responsible to store information about single memory operation.
+struct memory_operation {
+    void *ptr;
+    double total_time;
+    size_t size_of_allocation;
+    unsigned allocator_type;
+    unsigned allocation_method;
+    bool is_allocated;
+    int error_code;
+};
+
+double convert_bytes_to_mb(uint64_t bytes);
+
+int get_numa_node_id(void *ptr);
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Allocator.hpp b/memkind-1.10.0/test/allocator_perf_tool/Allocator.hpp
new file mode 100644
index 0000000..41eaf99
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Allocator.hpp
@@ -0,0 +1,41 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <stdio.h>
+#include "Allocation_info.hpp"
+
+class Allocator
+{
+public:
+
+    virtual memory_operation wrapped_malloc(size_t size) = 0;
+    virtual memory_operation wrapped_calloc(size_t num, size_t size) = 0;
+    virtual memory_operation wrapped_realloc(void *ptr, size_t size) = 0;
+    virtual void wrapped_free(void *ptr) = 0;
+    virtual unsigned type() = 0;
+
+    virtual ~Allocator(void) {}
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/AllocatorFactory.hpp b/memkind-1.10.0/test/allocator_perf_tool/AllocatorFactory.hpp
new file mode 100644
index 0000000..4dcd81b
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/AllocatorFactory.hpp
@@ -0,0 +1,212 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "memkind.h"
+
+#include "Allocator.hpp"
+#include "StandardAllocatorWithTimer.hpp"
+#include "VectorIterator.hpp"
+#include "Configuration.hpp"
+#include "JemallocAllocatorWithTimer.hpp"
+#include "MemkindAllocatorWithTimer.hpp"
+#include "HBWmallocAllocatorWithTimer.hpp"
+#include "Numastat.hpp"
+#include "PmemMockup.hpp"
+
+#include <vector>
+#include <assert.h>
+#include <string.h>
+#include <map>
+
+
+class AllocatorFactory
+{
+public:
+    //Allocator initialization statistics
+    struct initialization_stat {
+        float total_time; //Total time of initialization.
+        float ref_delta_time; //Delta Time.
+        unsigned allocator_type;
+        std::vector<float> memory_overhead; //Memory overhead per numa node.
+    };
+
+    AllocatorFactory()
+    {
+        memkind_allocators[AllocatorTypes::MEMKIND_DEFAULT] = MemkindAllocatorWithTimer(
+                                                                  MEMKIND_DEFAULT, AllocatorTypes::MEMKIND_DEFAULT);
+        memkind_allocators[AllocatorTypes::MEMKIND_REGULAR] = MemkindAllocatorWithTimer(
+                                                                  MEMKIND_REGULAR, AllocatorTypes::MEMKIND_REGULAR);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW] = MemkindAllocatorWithTimer(
+                                                              MEMKIND_HBW, AllocatorTypes::MEMKIND_HBW);
+        memkind_allocators[AllocatorTypes::MEMKIND_INTERLEAVE] =
+            MemkindAllocatorWithTimer(MEMKIND_INTERLEAVE,
+                                      AllocatorTypes::MEMKIND_INTERLEAVE);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_INTERLEAVE] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_INTERLEAVE,
+                                      AllocatorTypes::MEMKIND_HBW_INTERLEAVE);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_PREFERRED] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_PREFERRED,
+                                      AllocatorTypes::MEMKIND_HBW_PREFERRED);
+        memkind_allocators[AllocatorTypes::MEMKIND_HUGETLB] = MemkindAllocatorWithTimer(
+                                                                  MEMKIND_HUGETLB, AllocatorTypes::MEMKIND_HUGETLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_GBTLB] = MemkindAllocatorWithTimer(
+                                                                MEMKIND_GBTLB, AllocatorTypes::MEMKIND_GBTLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_HUGETLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_HUGETLB,
+                                      AllocatorTypes::MEMKIND_HBW_HUGETLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_PREFERRED_HUGETLB,
+                                      AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_GBTLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_GBTLB, AllocatorTypes::MEMKIND_HBW_GBTLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_PREFERRED_GBTLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_PREFERRED_GBTLB,
+                                      AllocatorTypes::MEMKIND_HBW_PREFERRED_GBTLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_PMEM] = MemkindAllocatorWithTimer(
+                                                               MEMKIND_PMEM_MOCKUP, AllocatorTypes::MEMKIND_PMEM);
+    }
+
+    //Get existing allocator without creating new.
+    //The owner of existing allocator is AllocatorFactory object.
+    Allocator *get_existing(unsigned type)
+    {
+        switch(type) {
+            case AllocatorTypes::STANDARD_ALLOCATOR:
+                return &standard_allocator;
+
+            case AllocatorTypes::JEMALLOC:
+                return &jemalloc;
+
+            case AllocatorTypes::HBWMALLOC_ALLOCATOR:
+                return &hbwmalloc_allocator;
+
+            default: {
+                if(memkind_allocators.count(type))
+                    return &memkind_allocators[type];
+
+                assert(!"'type' out of range!");
+            }
+        }
+    }
+
+    initialization_stat initialize_allocator(Allocator &allocator)
+    {
+        size_t initial_size = 512;
+        float before_node1 = Numastat::get_total_memory(0);
+        float before_node2 = Numastat::get_total_memory(1);
+        initialization_stat stat = {0};
+
+        //malloc
+        memory_operation malloc_data = allocator.wrapped_malloc(initial_size);
+        stat.total_time += malloc_data.total_time;
+
+        //realloc
+        memory_operation realloc_data = allocator.wrapped_realloc(malloc_data.ptr, 256);
+        allocator.wrapped_free(realloc_data.ptr);
+
+        stat.total_time += realloc_data.total_time;
+
+        //calloc
+        memory_operation calloc_data = allocator.wrapped_calloc(initial_size, 1);
+        allocator.wrapped_free(calloc_data.ptr);
+
+        stat.total_time += calloc_data.total_time;
+
+        stat.allocator_type = allocator.type();
+
+        //calc memory overhead
+        stat.memory_overhead.push_back(Numastat::get_total_memory(0) - before_node1);
+        stat.memory_overhead.push_back(Numastat::get_total_memory(1) - before_node2);
+
+        return stat;
+    }
+
+    initialization_stat initialize_allocator(unsigned type)
+    {
+        return initialize_allocator(*get_existing(type));
+    }
+
+    //Calc percent delta between reference value and current value.
+    float calc_ref_delta(float ref_value, float value)
+    {
+        return ((value / ref_value)  - 1.0) * 100.0;
+    }
+
+    //Test initialization performance over available allocators.
+    //Return statistics.
+    std::vector<initialization_stat> initialization_test()
+    {
+        std::vector<initialization_stat> stats;
+        initialization_stat stat;
+
+        stat = initialize_allocator(standard_allocator);
+        float ref_time = stat.total_time;
+        stats.push_back(stat);
+
+        //Loop over available allocators to call initializer and compute stats.
+        for (unsigned i=1; i<AllocatorTypes::NUM_OF_ALLOCATOR_TYPES; i++) {
+            stat = initialize_allocator(*get_existing(i));
+            stat.ref_delta_time = calc_ref_delta(ref_time, stat.total_time);
+            stats.push_back(stat);
+        }
+
+        return stats;
+    }
+
+    VectorIterator<Allocator *> generate_random_allocator_calls(int num, int seed,
+                                                                TypesConf allocator_calls)
+    {
+        srand(seed);
+        std::vector<Allocator *> allocators_calls;
+
+        for (int i=0; i<num; i++) {
+            int index;
+
+            do {
+                index = (rand() % (AllocatorTypes::NUM_OF_ALLOCATOR_TYPES));
+            } while(!allocator_calls.is_enabled(index));
+
+            allocators_calls.push_back(get_existing(index));
+        }
+
+        return VectorIterator<Allocator *>::create(allocators_calls);
+    }
+
+    //Return kind to the corresponding AllocatorTypes enum specified in argument.
+    memkind_t get_kind_by_type(unsigned type)
+    {
+        if(memkind_allocators.count(type))
+            return memkind_allocators[type].get_kind();
+
+        assert(!"'type' out of range!");
+    }
+
+private:
+    StandardAllocatorWithTimer standard_allocator;
+    JemallocAllocatorWithTimer jemalloc;
+    HBWmallocAllocatorWithTimer hbwmalloc_allocator;
+
+    std::map<unsigned, MemkindAllocatorWithTimer> memkind_allocators;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/CSVLogger.hpp b/memkind-1.10.0/test/allocator_perf_tool/CSVLogger.hpp
new file mode 100644
index 0000000..fd60731
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/CSVLogger.hpp
@@ -0,0 +1,61 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <iostream>
+#include <string>
+#include <sstream>
+#include <fstream>
+#include <ios>
+
+namespace csv
+{
+
+    class Row
+    {
+    public:
+        Row()
+        {
+            row << std::fixed;
+            row.precision(6);
+        }
+
+        template<class T>
+        void append(const T &e)
+        {
+            row << "," << e;
+        }
+
+        std::string export_row() const
+        {
+            std::stringstream ss(row.str());
+            ss << std::endl;
+            return ss.str();
+        }
+
+    private:
+        std::stringstream row;
+    };
+
+}
diff --git a/memkind-1.10.0/test/allocator_perf_tool/CommandLine.hpp b/memkind-1.10.0/test/allocator_perf_tool/CommandLine.hpp
new file mode 100644
index 0000000..1d04f6b
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/CommandLine.hpp
@@ -0,0 +1,85 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <map>
+#include <vector>
+#include <string>
+
+class CommandLine
+{
+public:
+
+    //Parse and write to val when option exist and strtol(...) > 0, otherwise val is not changed.
+    //T should be an integer type.
+    template<class T>
+    void parse_with_strtol(const std::string &option, T &val)
+    {
+        if(args.count(option)) {
+            T tmp = static_cast<T>(strtol(args[option].c_str(), NULL, 10));
+            if(tmp > 0)
+                val = tmp;
+            else
+                printf("Warning! Option '%s' may not be set.\n", option.c_str());
+        } else {
+            //Do not modify val.
+            printf("Warning! Option '%s' is not present.\n", option.c_str());
+        }
+    }
+
+    bool is_option_present(const std::string &option) const
+    {
+        return args.count(option);
+    }
+
+    bool is_option_set(const std::string option, std::string val)
+    {
+        if(is_option_present(option))
+            return (args[option] == val);
+        return false;
+    }
+
+    const std::string &get_option_value(const std::string &option)
+    {
+        return args[option];
+    }
+
+    CommandLine(int argc, char *argv[])
+    {
+        for (int i=0; i<argc; i++) {
+            std::string arg(argv[i]);
+            size_t found = arg.find("=");
+
+            if(found != std::string::npos) {
+                std::string option = arg.substr(0, found);
+                std::string val = arg.substr(found+1, arg.length());
+
+                args[option] = val;
+            }
+        }
+    }
+
+private:
+    std::map<std::string, std::string> args;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Configuration.hpp b/memkind-1.10.0/test/allocator_perf_tool/Configuration.hpp
new file mode 100644
index 0000000..c3ce905
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Configuration.hpp
@@ -0,0 +1,155 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <string>
+#include <map>
+#include <assert.h>
+
+//AllocatorTypes class represent allocator types and names related to this types.
+class AllocatorTypes
+{
+public:
+    enum {
+        STANDARD_ALLOCATOR,
+        JEMALLOC,
+        MEMKIND_DEFAULT,
+        MEMKIND_REGULAR,
+        MEMKIND_HBW,
+        MEMKIND_INTERLEAVE,
+        MEMKIND_HBW_INTERLEAVE,
+        MEMKIND_HBW_PREFERRED,
+        MEMKIND_HUGETLB,
+        MEMKIND_GBTLB,
+        MEMKIND_HBW_HUGETLB,
+        MEMKIND_HBW_PREFERRED_HUGETLB,
+        MEMKIND_HBW_GBTLB,
+        MEMKIND_HBW_PREFERRED_GBTLB,
+        HBWMALLOC_ALLOCATOR,
+        MEMKIND_PMEM,
+        NUM_OF_ALLOCATOR_TYPES
+    };
+
+    static const std::string &allocator_name(unsigned type)
+    {
+        static const std::string names[] = {
+            "STANDARD_ALLOCATOR",
+            "JEMALLOC",
+            "MEMKIND_DEFAULT",
+            "MEMKIND_REGULAR",
+            "MEMKIND_HBW",
+            "MEMKIND_INTERLEAVE",
+            "MEMKIND_HBW_INTERLEAVE",
+            "MEMKIND_HBW_PREFERRED",
+            "MEMKIND_HUGETLB",
+            "MEMKIND_GBTLB",
+            "MEMKIND_HBW_HUGETLB",
+            "MEMKIND_HBW_PREFERRED_HUGETLB",
+            "MEMKIND_HBW_GBTLB",
+            "MEMKIND_HBW_PREFERRED_GBTLB",
+            "HBWMALLOC_ALLOCATOR",
+            "MEMKIND_PMEM"
+        };
+
+        if(type >= NUM_OF_ALLOCATOR_TYPES) assert(!"Invalid input argument!");
+
+        return names[type];
+    }
+
+    static unsigned allocator_type(const std::string &name)
+    {
+        for (unsigned i=0; i<NUM_OF_ALLOCATOR_TYPES; i++) {
+            if(allocator_name(i) == name)
+                return i;
+        }
+
+        assert(!"Invalid input argument!");
+    }
+
+    static bool is_valid_memkind(unsigned type)
+    {
+        return (type >= MEMKIND_DEFAULT) && (type < NUM_OF_ALLOCATOR_TYPES);
+    }
+};
+
+//Enable or disable enum values (types).
+class TypesConf
+{
+public:
+    TypesConf() {}
+
+    TypesConf(unsigned type)
+    {
+        enable_type(type);
+    }
+
+    void enable_type(unsigned type)
+    {
+        types[type] = true;
+    }
+
+    void disable_type(unsigned type)
+    {
+        if(types.count(type))
+            types[type] = false;
+    }
+
+    bool is_enabled(unsigned type) const
+    {
+        return (types.count(type) ? types.find(type)->second : false);
+    }
+
+private:
+    std::map<unsigned, bool> types;
+};
+
+//AllocationSizesConf class represents allocation sizes configuration.
+//This data is needed to generate "n" sizes in range from "size_from" to "size_to".
+class AllocationSizesConf
+{
+public:
+    unsigned n;
+    size_t size_from;
+    size_t size_to;
+};
+
+//TaskConf class contain configuration data for task,
+//where:
+// - "n" - number of iterations,
+// - "allocation_sizes_conf" - allocation sizes configuration,
+// - "func_calls" - enabled or disabled function calls,
+// - "allocators_types" - enable allocators,
+// - "seed" - random seed.
+// - "touch_memory" - enable or disable touching memory
+class TaskConf
+{
+public:
+    unsigned n;
+    AllocationSizesConf allocation_sizes_conf;
+    TypesConf func_calls;
+    TypesConf allocators_types;
+    unsigned seed;
+    bool touch_memory;
+    bool is_csv_log_enabled;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/ConsoleLog.hpp b/memkind-1.10.0/test/allocator_perf_tool/ConsoleLog.hpp
new file mode 100644
index 0000000..396bcca
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/ConsoleLog.hpp
@@ -0,0 +1,77 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include "Configuration.hpp"
+#include "Stats.hpp"
+#include "FunctionCalls.hpp"
+
+#include <stdio.h>
+
+class ConsoleLog
+{
+public:
+    static void print_stats(TimeStats &stats, unsigned allocator_type,
+                            unsigned func_calls)
+    {
+        if(stats.stats.count(allocator_type)) {
+            if(stats.stats[allocator_type].count(func_calls)) {
+                MethodStats method_stats = stats.stats[allocator_type][func_calls];
+                printf(" %20s (%u) | %7s | %10f.s | %10f.s  | %zu bytes/%f MB \n",
+                       AllocatorTypes::allocator_name(allocator_type).c_str(),
+                       allocator_type,
+                       FunctionCalls::function_name(func_calls).c_str(),
+                       method_stats.total_time,
+                       method_stats.average_time,
+                       method_stats.allocation_size,
+                       convert_bytes_to_mb(method_stats.allocation_size)
+                      );
+            }
+        }
+    }
+
+    static void print_table(TimeStats &stats)
+    {
+        printf("\n====== Allocators function calls performance =================================================\n");
+        printf(" %20s Id:   Method:    Total time:    Average time:  Allocated memory bytes/MB: \n",
+               "Allocator:");
+        for (unsigned i=0; i<=AllocatorTypes::MEMKIND_HBW_PREFERRED; i++) {
+            for (unsigned func_call=FunctionCalls::FREE+1;
+                 func_call<FunctionCalls::NUM_OF_FUNCTIONS; func_call++) {
+                print_stats(stats, i, func_call);
+            }
+        }
+        printf("==============================================================================================\n");
+    }
+
+    static void print_requested_memory(TimeStats &stats, std::string test_name)
+    {
+        printf("\n====== Requested memory stats for %s =================\n",
+               test_name.c_str());
+        printf("Total requested allocations: %zu bytes/%f MB. \n",stats.get_allocated(),
+               convert_bytes_to_mb(stats.get_allocated()));
+        printf("Total requested deallocations: %zu bytes/%f MB. \n",
+               stats.get_deallocated(), convert_bytes_to_mb(stats.get_deallocated()));
+        printf("=====================================================================\n");
+    }
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/FunctionCalls.hpp b/memkind-1.10.0/test/allocator_perf_tool/FunctionCalls.hpp
new file mode 100644
index 0000000..ef2d7da
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/FunctionCalls.hpp
@@ -0,0 +1,75 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "VectorIterator.hpp"
+#include "Configuration.hpp"
+#include <vector>
+#include <cstdlib>
+
+class FunctionCalls
+{
+public:
+    enum {FREE, MALLOC, CALLOC, REALLOC, NUM_OF_FUNCTIONS};
+
+    static const std::string function_name(unsigned type)
+    {
+        static std::string names[] = {"free", "malloc", "calloc", "realloc"};
+
+        if(type >= NUM_OF_FUNCTIONS) assert(!"Invalidate input argument!");
+
+        return names[type];
+    }
+
+    static unsigned function_type(const std::string &name)
+    {
+        for (unsigned i=0; i<NUM_OF_FUNCTIONS; i++) {
+            if(function_name(i) == name)
+                return i;
+        }
+
+        assert(!"Invalid input argument!");
+    }
+
+    static VectorIterator<int> generate_random_allocator_func_calls(int call_num,
+                                                                    int seed, TypesConf func_calls)
+    {
+        std::vector<unsigned> avail_types;
+
+        std::srand(seed);
+        std::vector<int> calls;
+
+        for (int i=0; i<call_num; i++) {
+            int index;
+            do {
+                index = rand() % (NUM_OF_FUNCTIONS);
+            } while(!func_calls.is_enabled(index));
+
+            calls.push_back(index);
+        }
+
+        return VectorIterator<int>::create(calls);
+    }
+
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp b/memkind-1.10.0/test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
new file mode 100644
index 0000000..bdfe5c3
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
@@ -0,0 +1,51 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#include "FunctionCallsPerformanceTask.h"
+
+
+void FunctionCallsPerformanceTask::run()
+{
+    VectorIterator<size_t> allocation_sizes =
+        AllocationSizes::generate_random_sizes(task_conf.allocation_sizes_conf,
+                                               task_conf.seed);
+
+    VectorIterator<int> func_calls =
+        FunctionCalls::generate_random_allocator_func_calls(task_conf.n, task_conf.seed,
+                                                            task_conf.func_calls);
+
+    AllocatorFactory allocator_types;
+    VectorIterator<Allocator *> allocators_calls =
+        allocator_types.generate_random_allocator_calls(task_conf.n, task_conf.seed,
+                                                        task_conf.allocators_types);
+
+    ScenarioWorkload scenario_workload = ScenarioWorkload(
+                                             &allocators_calls,
+                                             &allocation_sizes,
+                                             &func_calls
+                                         );
+
+    while (scenario_workload.run());
+
+    results = scenario_workload.get_allocations_info();
+}
diff --git a/memkind-1.10.0/test/allocator_perf_tool/FunctionCallsPerformanceTask.h b/memkind-1.10.0/test/allocator_perf_tool/FunctionCallsPerformanceTask.h
new file mode 100644
index 0000000..1add585
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/FunctionCallsPerformanceTask.h
@@ -0,0 +1,54 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Task.hpp"
+#include "AllocatorFactory.hpp"
+#include "FunctionCalls.hpp"
+#include "AllocationSizes.hpp"
+#include "VectorIterator.hpp"
+#include "ScenarioWorkload.h"
+#include "Configuration.hpp"
+
+
+class FunctionCallsPerformanceTask :
+    public Task
+{
+public:
+    FunctionCallsPerformanceTask(TaskConf conf)
+    {
+        task_conf = conf;
+    }
+
+    void run();
+
+    std::vector<memory_operation> get_results()
+    {
+        return results;
+    }
+
+private:
+    TaskConf task_conf;
+    std::vector<memory_operation> results;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/GTestAdapter.hpp b/memkind-1.10.0/test/allocator_perf_tool/GTestAdapter.hpp
new file mode 100644
index 0000000..30dc9ce
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/GTestAdapter.hpp
@@ -0,0 +1,49 @@
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <iostream>
+#include <sstream>
+#include <gtest/gtest.h>
+
+/*
+ * The GTestAdapter class is an adapter for GTest framework.
+ * All methods from GTest framework that need to be adapted should be located here.
+ */
+
+class GTestAdapter
+{
+public:
+    template<class T>
+    static void RecordProperty(const std::string &key, const T &value)
+    {
+        std::ostringstream tmp_value;
+        tmp_value << value;
+        testing::Test::RecordProperty(key, tmp_value.str());
+        std::ios::fmtflags flags(std::cout.flags());
+        std::cout << key << ": " << std::fixed  <<  std::setprecision(
+                      5) << value << std::endl;
+        std::cout.flags(flags);
+    }
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp b/memkind-1.10.0/test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
new file mode 100644
index 0000000..4ef7463
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
@@ -0,0 +1,74 @@
+/*
+* Copyright (C) 2017 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <hbwmalloc.h>
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+#include <cerrno>
+
+#include <stdlib.h>
+
+
+class HBWmallocAllocatorWithTimer
+    : public Allocator
+{
+public:
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC)
+        data.ptr = hbw_malloc(size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC)
+        data.ptr = hbw_calloc(num, size);
+        END_TEST
+    }
+
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC)
+        data.ptr = hbw_realloc(ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        hbw_free(ptr);
+    }
+
+    unsigned type()
+    {
+        return AllocatorTypes::HBWMALLOC_ALLOCATOR;
+    }
+
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/HugePageOrganizer.hpp b/memkind-1.10.0/test/allocator_perf_tool/HugePageOrganizer.hpp
new file mode 100644
index 0000000..64180c9
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/HugePageOrganizer.hpp
@@ -0,0 +1,94 @@
+
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <fstream>
+#include <string.h>
+#include <numa.h>
+#include <vector>
+#include <iostream>
+#include <stdexcept>
+
+/*
+ * HugePageOrganizer sets hugepages per NUMA node and restore initial setup of hugepages.
+ * It writes and reads from the same file, so using HugePageOrganizer with parallel execution may cause undefined behaviour.
+ */
+
+class HugePageOrganizer
+{
+public:
+    HugePageOrganizer(int nr_hugepages_per_node)
+    {
+        for (int node_id = 0; node_id < numa_num_configured_nodes(); node_id++) {
+            initial_nr_hugepages_per_nodes.push_back(get_nr_hugepages(node_id));
+            if (set_nr_hugepages(nr_hugepages_per_node, node_id)) {
+                restore();
+                throw std::runtime_error("Error: Could not set the requested amount of huge pages.");
+            }
+        }
+    }
+
+    ~HugePageOrganizer()
+    {
+        restore();
+    }
+private:
+    std::vector<int> initial_nr_hugepages_per_nodes;
+
+    int get_nr_hugepages(int node_number)
+    {
+        std::string line;
+        char path[128];
+        sprintf(path,
+                "/sys/devices/system/node/node%d/hugepages/hugepages-2048kB/nr_hugepages",
+                node_number);
+        std::ifstream file(path);
+        if (!file) {
+            return -1;
+        }
+        std::getline(file, line);
+        return strtol(line.c_str(), 0, 10);
+    }
+
+    int set_nr_hugepages(int nr_hugepages, int node_number)
+    {
+        char cmd[128];
+        sprintf(cmd, "sudo sh -c \"echo %d > \
+            /sys/devices/system/node/node%d/hugepages/hugepages-2048kB/nr_hugepages\"",
+                nr_hugepages, node_number);
+        if (system(cmd) || (get_nr_hugepages(node_number) != nr_hugepages)) {
+            return -1;
+        }
+        return 0;
+    }
+
+    void restore()
+    {
+        for(size_t node_id = 0; node_id < initial_nr_hugepages_per_nodes.size();
+            node_id++) {
+            set_nr_hugepages(initial_nr_hugepages_per_nodes[node_id], node_id);
+        }
+    }
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/HugePageUnmap.hpp b/memkind-1.10.0/test/allocator_perf_tool/HugePageUnmap.hpp
new file mode 100644
index 0000000..237c357
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/HugePageUnmap.hpp
@@ -0,0 +1,82 @@
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <vector>
+#include "Task.hpp"
+
+class HugePageUnmap: public Task
+{
+
+public:
+    HugePageUnmap(int operations, bool touch_memory, size_t alignment_size,
+                  size_t alloc_size, hbw_pagesize_t page_size) :
+        mem_operations_num(operations),
+        touch_memory(touch_memory),
+        alignment_size(alignment_size),
+        alloc_size(alloc_size),
+        page_size(page_size)
+    {}
+
+    ~HugePageUnmap()
+    {
+        for(int i=0; i<mem_operations_num; i++) {
+            hbw_free(results[i].ptr);
+        }
+    };
+
+    void run()
+    {
+        void *ptr = NULL;
+
+        for(int i=0; i<mem_operations_num; i++) {
+            int ret = hbw_posix_memalign_psize(&ptr, alignment_size, alloc_size, page_size);
+
+            ASSERT_EQ(ret, 0);
+            ASSERT_FALSE(ptr == NULL);
+
+            if(touch_memory) {
+                memset(ptr, alignment_size, alignment_size);
+            }
+
+            memory_operation data;
+            data.ptr = ptr;
+            results.push_back(data);
+        }
+    }
+
+    std::vector<memory_operation> get_results()
+    {
+        return results;
+    }
+
+private:
+    int mem_operations_num;
+    std::vector<memory_operation> results;
+
+    bool touch_memory;
+    size_t alignment_size;
+    size_t alloc_size;
+    hbw_pagesize_t page_size;
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Iterator.hpp b/memkind-1.10.0/test/allocator_perf_tool/Iterator.hpp
new file mode 100644
index 0000000..1891ef4
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Iterator.hpp
@@ -0,0 +1,34 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+template<class T>
+class Iterator
+{
+public:
+    virtual bool has_next() const = 0;
+    virtual T next() = 0;
+    virtual size_t size() const = 0;
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp b/memkind-1.10.0/test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
new file mode 100644
index 0000000..466af7b
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
@@ -0,0 +1,68 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+
+#include <jemalloc/jemalloc.h>
+
+class JemallocAllocatorWithTimer
+    : public Allocator
+{
+public:
+
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(AllocatorTypes::JEMALLOC, FunctionCalls::MALLOC)
+        data.ptr = malloc(size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(AllocatorTypes::JEMALLOC, FunctionCalls::CALLOC)
+        data.ptr = calloc(num, size);
+        END_TEST
+    }
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(AllocatorTypes::JEMALLOC, FunctionCalls::REALLOC)
+        data.ptr = realloc(ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        free(ptr);
+    }
+
+    unsigned type()
+    {
+        return AllocatorTypes::JEMALLOC;
+    }
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Makefile b/memkind-1.10.0/test/allocator_perf_tool/Makefile
new file mode 100644
index 0000000..e53248e
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Makefile
@@ -0,0 +1,26 @@
+#
+#  Copyright (C) 2015 - 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+perf_tool: main.cpp ScenarioWorkload.cpp FunctionCallsPerformanceTask.cpp StressIncreaseToMax.cpp Allocation_info.cpp PmemMockup.cpp
+	g++ -o perf_tool $^ -O0 -lmemkind -std=c++11 -lpthread -lnuma -g
diff --git a/memkind-1.10.0/test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp b/memkind-1.10.0/test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
new file mode 100644
index 0000000..7265b18
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
@@ -0,0 +1,97 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <memkind.h>
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+#include <cerrno>
+
+#include <stdlib.h>
+
+
+class MemkindAllocatorWithTimer
+    : public Allocator
+{
+public:
+    MemkindAllocatorWithTimer() : kind(&MEMKIND_DEFAULT) {}
+
+    MemkindAllocatorWithTimer(memkind_t &memory_kind, unsigned kind_type_id)
+    {
+        kind = &memory_kind;
+        type_id = kind_type_id;
+    }
+    ~MemkindAllocatorWithTimer(void) {}
+
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(type_id, FunctionCalls::MALLOC)
+        data.ptr = memkind_malloc(*kind, size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(type_id, FunctionCalls::CALLOC)
+        data.ptr = memkind_calloc(*kind, num, size);
+        END_TEST
+    }
+
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(type_id, FunctionCalls::REALLOC)
+        data.ptr = memkind_realloc(*kind, ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        memkind_free(*kind, ptr);
+    }
+
+    void change_kind(memkind_t &memory_kind, unsigned kind_type_id)
+    {
+        kind = &memory_kind;
+        type_id = kind_type_id;
+    }
+
+    unsigned type()
+    {
+        return type_id;
+    }
+    memkind_t get_kind()
+    {
+        return *kind;
+    }
+
+private:
+    memkind_t *kind;
+    unsigned type_id;
+
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Numastat.hpp b/memkind-1.10.0/test/allocator_perf_tool/Numastat.hpp
new file mode 100644
index 0000000..d8ecb19
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Numastat.hpp
@@ -0,0 +1,85 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include <assert.h>
+
+#include <sys/types.h>
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <iostream>
+#include <string>
+#include <sstream>
+
+
+class Numastat
+{
+public:
+    //Returns numastat memory usage per node.
+    static float get_total_memory(unsigned node)
+    {
+        static pid_t pid = getpid();
+
+        std::stringstream cmd;
+        cmd << "numastat " << pid;
+
+        FILE *file;
+        char buff[256];
+        float result = -1.0;
+
+        if((file = popen(cmd.str().c_str(), "r"))) {
+            while(fgets(buff, 256, file));
+            pclose(file);
+
+            //We got: "Total                     1181.90            2.00         1183.90".
+            //2.00 is our memory usage from Node 1.
+            std::string last_line(buff);
+
+            size_t dot_pos = 0;
+            //Now we search in: "           2.00         1183.90".
+            for(unsigned i=0; i<=node; i++) {
+                dot_pos = last_line.find(".", dot_pos+1);
+                assert(dot_pos != std::string::npos);
+            }
+
+            //We are at: " 2.00         1183.90".
+            size_t number_begin = last_line.rfind(" ", dot_pos);
+            assert(number_begin != std::string::npos);
+
+            number_begin += 1;
+            buff[dot_pos+3] = '\0';
+
+#if __cplusplus > 201100L
+            result = strtod(&buff[number_begin], NULL);
+#else
+            result = atof(&buff[number_begin]);
+#endif
+        }
+
+        assert(result > -0.01);
+
+        return result;
+    }
+
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/PmemMockup.cpp b/memkind-1.10.0/test/allocator_perf_tool/PmemMockup.cpp
new file mode 100644
index 0000000..619af5d
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/PmemMockup.cpp
@@ -0,0 +1,26 @@
+/*
+* Copyright (C) 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#include "PmemMockup.hpp"
+
+struct memkind *MEMKIND_PMEM_MOCKUP;
diff --git a/memkind-1.10.0/test/allocator_perf_tool/PmemMockup.hpp b/memkind-1.10.0/test/allocator_perf_tool/PmemMockup.hpp
new file mode 100644
index 0000000..a096ebe
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/PmemMockup.hpp
@@ -0,0 +1,30 @@
+/*
+* Copyright (C) 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include "memkind.h"
+
+///
+/// \brief Mockup structure for PMEM kind to use allocator_perf_tool engine
+///
+extern memkind_t MEMKIND_PMEM_MOCKUP;
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Runnable.hpp b/memkind-1.10.0/test/allocator_perf_tool/Runnable.hpp
new file mode 100644
index 0000000..57455cc
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Runnable.hpp
@@ -0,0 +1,31 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+class Runnable
+{
+public:
+    virtual void run() = 0;
+    virtual ~Runnable() {};
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/ScenarioWorkload.cpp b/memkind-1.10.0/test/allocator_perf_tool/ScenarioWorkload.cpp
new file mode 100644
index 0000000..e4ad796
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/ScenarioWorkload.cpp
@@ -0,0 +1,113 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#include "ScenarioWorkload.h"
+
+ScenarioWorkload::ScenarioWorkload(VectorIterator<Allocator *> *a,
+                                   VectorIterator<size_t> *as,  VectorIterator<int> *fc)
+{
+    allocators = a;
+    func_calls = fc;
+    alloc_sizes = as;
+}
+
+bool ScenarioWorkload::run()
+{
+    if(func_calls->has_next() && allocators->has_next() &&
+       alloc_sizes->has_next()) {
+        switch(func_calls->next()) {
+            case FunctionCalls::MALLOC: {
+                memory_operation data = allocators->next()->wrapped_malloc(alloc_sizes->next());
+                post_allocation_check(data);
+                break;
+            }
+            case FunctionCalls::CALLOC: {
+                memory_operation data = allocators->next()->wrapped_calloc(1,
+                                                                           alloc_sizes->next());
+                post_allocation_check(data);
+                break;
+            }
+            case FunctionCalls::REALLOC: {
+                //Guarantee the memory for realloc.
+                Allocator *allocator = allocators->next();
+                memory_operation to_realloc = allocator->wrapped_malloc(512);
+
+                memory_operation data = allocator->wrapped_realloc(to_realloc.ptr,
+                                                                   alloc_sizes->next());
+                post_allocation_check(data);
+                break;
+            }
+            case FunctionCalls::FREE: {
+                memory_operation *data = get_allocated_memory();
+
+                if(!allocations.empty() && (data != NULL)) {
+                    allocator_factory.get_existing(data->allocator_type)->wrapped_free(data->ptr);
+                    data->is_allocated = false;
+
+                    memory_operation free_op = *data;
+                    free_op.allocation_method = FunctionCalls::FREE;
+                    allocations.push_back(free_op);
+                }
+
+                break;
+            }
+            default:
+                assert(!"Function call identifier out of range.");
+                break;
+        }
+
+        return true;
+    }
+
+    return false;
+}
+
+ScenarioWorkload::~ScenarioWorkload(void)
+{
+    for (int i=0; i<allocations.size(); i++) {
+        memory_operation data = allocations[i];
+        if(data.is_allocated && (data.allocation_method != FunctionCalls::FREE))
+            allocator_factory.get_existing(data.allocator_type)->wrapped_free(data.ptr);
+    }
+}
+
+memory_operation *ScenarioWorkload::get_allocated_memory()
+{
+    for (int i=allocations.size()-1; i>=0; i--) {
+        memory_operation *data = &allocations[i];
+        if(data->is_allocated)
+            return data;
+    }
+
+    return NULL;
+}
+
+void ScenarioWorkload::post_allocation_check(const memory_operation &data)
+{
+    allocations.push_back(data);
+    if(touch_memory_on_allocation && (data.ptr != NULL) &&
+       (data.error_code != ENOMEM)) {
+        //Write memory to ensure physical allocation.
+        memset(data.ptr, 1, data.size_of_allocation);
+    }
+}
diff --git a/memkind-1.10.0/test/allocator_perf_tool/ScenarioWorkload.h b/memkind-1.10.0/test/allocator_perf_tool/ScenarioWorkload.h
new file mode 100644
index 0000000..1d80072
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/ScenarioWorkload.h
@@ -0,0 +1,68 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include "Workload.hpp"
+#include "FunctionCalls.hpp"
+#include "Allocator.hpp"
+#include "AllocatorFactory.hpp"
+
+#include <string.h>
+
+class ScenarioWorkload :
+    public Workload
+{
+public:
+    ScenarioWorkload(VectorIterator<Allocator *> *a, VectorIterator<size_t> *as,
+                     VectorIterator<int> *fc);
+    ~ScenarioWorkload(void);
+
+    double get_time_costs();
+
+    const std::vector<memory_operation> &get_allocations_info() const
+    {
+        return allocations;
+    }
+
+    bool run();
+
+    memory_operation *get_allocated_memory();
+
+    void enable_touch_memory_on_allocation(bool enable)
+    {
+        touch_memory_on_allocation = enable;
+    }
+
+    void post_allocation_check(const memory_operation &data);
+
+private:
+    AllocatorFactory allocator_factory;
+    std::vector<memory_operation> allocations;
+
+    bool touch_memory_on_allocation;
+
+    VectorIterator<int> *func_calls;
+    VectorIterator<size_t> *alloc_sizes;
+    VectorIterator<Allocator *> *allocators;
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/StandardAllocatorWithTimer.hpp b/memkind-1.10.0/test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
new file mode 100644
index 0000000..0c53b95
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
@@ -0,0 +1,68 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+
+#include <stdlib.h>
+
+class StandardAllocatorWithTimer
+    : public Allocator
+{
+public:
+
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(AllocatorTypes::STANDARD_ALLOCATOR, FunctionCalls::MALLOC)
+        data.ptr = malloc(size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(AllocatorTypes::STANDARD_ALLOCATOR, FunctionCalls::CALLOC)
+        data.ptr = calloc(num, size);
+        END_TEST
+    }
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(AllocatorTypes::STANDARD_ALLOCATOR, FunctionCalls::REALLOC)
+        data.ptr = realloc(ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        free(ptr);
+    }
+
+    unsigned type()
+    {
+        return AllocatorTypes::STANDARD_ALLOCATOR;
+    }
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Stats.hpp b/memkind-1.10.0/test/allocator_perf_tool/Stats.hpp
new file mode 100644
index 0000000..c474f04
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Stats.hpp
@@ -0,0 +1,101 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <map>
+#include <vector>
+
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "FunctionCalls.hpp"
+
+class MethodStats
+{
+public:
+    MethodStats()
+    {
+        total_time = 0.0;
+        average_time = 0;
+        allocation_size = 0;
+        samples_num = 0;
+    }
+
+    double total_time;
+    double average_time;
+    unsigned samples_num;
+    size_t allocation_size;
+};
+
+
+class TimeStats
+{
+public:
+
+    TimeStats()
+    {
+        allocated = 0;
+        deallocated = 0;
+    }
+
+    std::map<unsigned, std::map<unsigned, MethodStats> > stats;
+
+    TimeStats &operator+=(const std::vector<memory_operation> &data)
+    {
+        for (size_t i=0; i<data.size(); i++) {
+            memory_operation tmp = data[i];
+            MethodStats &method_stats = stats[tmp.allocator_type][tmp.allocation_method];
+            method_stats.allocation_size += tmp.size_of_allocation;
+            method_stats.total_time += tmp.total_time;
+            method_stats.samples_num++;
+
+            //Update average.
+            double total_time = method_stats.total_time;
+            double samples_num = method_stats.samples_num;
+            double avg = total_time/samples_num;
+            method_stats.average_time = avg;
+
+            if(tmp.allocation_method != FunctionCalls::FREE) {
+                allocated += tmp.size_of_allocation;
+            } else {
+                deallocated += tmp.size_of_allocation;
+            }
+        }
+
+        return *this;
+    }
+
+    size_t get_allocated() const
+    {
+        return allocated;
+    }
+    size_t get_deallocated() const
+    {
+        return deallocated;
+    }
+
+private:
+    size_t allocated;
+    size_t deallocated;
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/StressIncreaseToMax.cpp b/memkind-1.10.0/test/allocator_perf_tool/StressIncreaseToMax.cpp
new file mode 100644
index 0000000..908eef5
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/StressIncreaseToMax.cpp
@@ -0,0 +1,138 @@
+/*
+* Copyright (C) 2015 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "StressIncreaseToMax.h"
+
+
+void StressIncreaseToMax::run()
+{
+    //Generate constant allocation sizes.
+    VectorIterator<size_t> allocation_sizes =
+        AllocationSizes::generate_random_sizes(task_conf.allocation_sizes_conf,
+                                               task_conf.seed);
+    //Generate only mallocs.
+    VectorIterator<int> func_calls =
+        FunctionCalls::generate_random_allocator_func_calls(task_conf.n, task_conf.seed,
+                                                            task_conf.func_calls);
+
+    unsigned type;
+    for (type = 0; type < AllocatorTypes::NUM_OF_ALLOCATOR_TYPES; type++) {
+        if(task_conf.allocators_types.is_enabled(type))
+            break; //Assume that there is only one type.
+    }
+
+
+    AllocatorFactory allocator_factory;
+    VectorIterator<Allocator *> allocators_calls =
+        allocator_factory.generate_random_allocator_calls(task_conf.n, task_conf.seed,
+                                                          task_conf.allocators_types);
+
+    ScenarioWorkload scenario_workload(
+        &allocators_calls,
+        &allocation_sizes,
+        &func_calls
+    );
+
+    scenario_workload.enable_touch_memory_on_allocation(task_conf.touch_memory);
+
+    test_status.is_allocation_error = false;
+
+    size_t requested_memory = 0;
+    bool has_reach_memory_request_limit = false;
+
+    while (!has_reach_memory_request_limit &&
+           !test_status.is_allocation_error &&
+           (test_status.has_next_memory_operation = scenario_workload.run())) {
+        memory_operation data = scenario_workload.get_allocations_info().back();
+        test_status.is_allocation_error = (data.error_code == ENOMEM) ||
+                                          (data.ptr == NULL);
+
+        if(data.allocation_method != FunctionCalls::FREE) {
+            requested_memory += data.size_of_allocation;
+            has_reach_memory_request_limit = requested_memory >= req_mem_limit;
+        }
+    }
+
+    if(!(scenario_workload.get_allocations_info().size() < task_conf.n) &&
+       !has_reach_memory_request_limit)
+        printf("\nWARNING: Too few memory operations to reach the limit.\n");
+    if(test_status.is_allocation_error) printf("\nWARNING: Allocation error. \n");
+
+    results = scenario_workload.get_allocations_info();
+}
+
+std::vector<iteration_result> StressIncreaseToMax::execute_test_iterations(
+    const TaskConf &task_conf,
+    unsigned time,
+    size_t requested_memory_limit)
+{
+    TimerSysTime timer;
+    unsigned itr = 0;
+    std::vector<iteration_result> results;
+
+    std::ofstream csv_file;
+
+    csv::Row row;
+    row.append("Iteration");
+    row.append("Allocated memory (MB)");
+    row.append("Elapsed time (seconds)");
+    if(task_conf.is_csv_log_enabled) {
+        csv_file.open("stress_test_increase_to_max.csv");
+        csv_file << row.export_row();
+    }
+    printf("%s", row.export_row().c_str());
+
+    timer.start();
+
+    while (timer.getElapsedTime() < time) {
+        StressIncreaseToMax stress_test(task_conf, requested_memory_limit);
+        stress_test.run();
+        float elapsed_time = timer.getElapsedTime();
+
+        TimeStats stats;
+        stats += stress_test.get_results();
+
+        results.push_back(stress_test.get_test_status());
+
+        //Log every iteration of StressIncreaseToMax test.
+        csv::Row row;
+        row.append(itr);
+        row.append(convert_bytes_to_mb(stats.get_allocated()));
+        row.append(elapsed_time);
+        if(task_conf.is_csv_log_enabled) {
+            csv_file << row.export_row();
+        }
+        printf("%s", row.export_row().c_str());
+        fflush(stdout);
+
+        itr++;
+    }
+
+    printf("\nStress test (StressIncreaseToMax) finish in time %f.\n",
+           timer.getElapsedTime());
+
+    csv_file.close();
+
+    return results;
+}
diff --git a/memkind-1.10.0/test/allocator_perf_tool/StressIncreaseToMax.h b/memkind-1.10.0/test/allocator_perf_tool/StressIncreaseToMax.h
new file mode 100644
index 0000000..867d8fa
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/StressIncreaseToMax.h
@@ -0,0 +1,78 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Configuration.hpp"
+#include "ScenarioWorkload.h"
+#include "FunctionCalls.hpp"
+#include "AllocationSizes.hpp"
+#include "VectorIterator.hpp"
+#include "AllocatorFactory.hpp"
+#include "Task.hpp"
+#include "Numastat.hpp"
+#include "CSVLogger.hpp"
+#include "TimerSysTime.hpp"
+#include "Stats.hpp"
+
+#include <vector>
+
+struct iteration_result {
+    bool has_next_memory_operation;
+    bool is_allocation_error;
+};
+
+class StressIncreaseToMax
+    : public Task
+{
+public:
+    StressIncreaseToMax(const TaskConf &conf, size_t requested_memory_limit)
+        : task_conf(conf),
+          req_mem_limit(requested_memory_limit)
+    {}
+
+    void run();
+
+    //Return memory operations from the last run.
+    std::vector<memory_operation> get_results()
+    {
+        return results;
+    }
+    iteration_result get_test_status()
+    {
+        return test_status;
+    }
+
+    static std::vector<iteration_result> execute_test_iterations(
+        const TaskConf &task_conf, unsigned time, size_t requested_memory_limit);
+
+private:
+    size_t req_mem_limit;
+    ScenarioWorkload *scenario_workload;
+    std::vector<memory_operation> results;
+    const TaskConf &task_conf;
+
+    //Test status
+    iteration_result test_status;
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Task.hpp b/memkind-1.10.0/test/allocator_perf_tool/Task.hpp
new file mode 100644
index 0000000..0631669
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Task.hpp
@@ -0,0 +1,39 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <vector>
+
+#include "Runnable.hpp"
+#include "Allocation_info.hpp"
+
+class Task
+    : public Runnable
+{
+public:
+    virtual ~Task() {}
+
+    virtual std::vector<memory_operation> get_results() = 0;
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/TaskFactory.hpp b/memkind-1.10.0/test/allocator_perf_tool/TaskFactory.hpp
new file mode 100644
index 0000000..ea9ff28
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/TaskFactory.hpp
@@ -0,0 +1,58 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <cstdlib>
+#include <vector>
+#include <assert.h>
+
+#include "Task.hpp"
+#include "FunctionCallsPerformanceTask.h"
+#include "Configuration.hpp"
+
+
+class TaskFactory
+{
+public:
+
+    Task *create(TaskConf conf)
+    {
+        Task *task = NULL;
+        task = new FunctionCallsPerformanceTask(conf);
+
+        tasks.push_back(task);
+
+        return task;
+    }
+
+    ~TaskFactory()
+    {
+        for (int i=0; i<tasks.size(); i++) {
+            delete tasks[i];
+        }
+    }
+
+private:
+    std::vector<Task *> tasks;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Tests.hpp b/memkind-1.10.0/test/allocator_perf_tool/Tests.hpp
new file mode 100644
index 0000000..d7f20d4
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Tests.hpp
@@ -0,0 +1,191 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+/*
+This file contain self-tests for Allocator Perf Tool.
+*/
+#pragma once
+
+#include <unistd.h>
+
+#include "Configuration.hpp"
+#include "AllocatorFactory.hpp"
+#include "AllocationSizes.hpp"
+#include "VectorIterator.hpp"
+#include "StandardAllocatorWithTimer.hpp"
+#include "ScenarioWorkload.h"
+#include "TaskFactory.hpp"
+#include "Task.hpp"
+
+//Basic test for allocators. Allocate 32 bytes by malloc, then deallocate it by free.
+void test_allocator(Allocator &allocator, const char *allocator_name)
+{
+    printf("\n============= Allocator (%s) test ============= \n",allocator_name);
+
+    memory_operation data = allocator.wrapped_malloc(32);
+    printf("malloc: %p \n",data.ptr);
+
+    assert(data.ptr!=NULL);
+
+    allocator.wrapped_free(data.ptr);
+    printf("free: %p \n",data.ptr);
+
+    printf("\n==================================================== \n");
+}
+
+//Test for the workload classes. This test count and validate the number or runs.
+void test_workload(Workload &workload, const int N, const char *workload_name)
+{
+    printf("\n============= Work load (%s) test ============= \n",workload_name);
+
+    assert(workload.run()); // Do "single" memory operation (allocate or deallocate).
+
+    int i;
+    for (i=1; workload.run(); i++); // Do the rest of operations.
+    assert(i == N);
+
+    printf("\n==================================================== \n");
+}
+
+//Simulate time consuming memory operation.
+memory_operation time_consuming_memory_operation(unsigned int seconds)
+{
+    size_t size;
+
+    START_TEST(0,0)
+    sleep(seconds);
+    END_TEST
+}
+
+//This test check if timer has measured time.
+void test_timer()
+{
+    float total_time = time_consuming_memory_operation(2).total_time;
+
+    bool test_res = (total_time >=  2.0) && (total_time < 2.2);
+    if(!test_res)
+        printf("test_timer(): unexpected timing %f", total_time);
+    assert(test_res);
+}
+
+//Test for iterators. Check the number of iterations, such as:
+//N == number of iterations.
+template<class T>
+void test_iterator(Iterator<T> &it, const int N)
+{
+    printf("\n================= Iteartor test ==================== \n");
+
+    assert(it.size() == N);
+    int i;
+    for (i=0; it.has_next(); i++) {
+        it.next();
+    }
+    printf("iterations=%d/%d\n",i,N);
+    assert(i == N);
+
+    printf("\n==================================================== \n");
+}
+
+//This test validate range of values generated by random generators.
+template<class T, class C>
+void test_iterator_values(VectorIterator<T> &it, const C from, const C to)
+{
+    for (int i=0; it.has_next(); i++) {
+        T val = it.next();
+
+        if(val < from) std::cout << "ivalid value: actual=" << val << ", expected= >="
+                                     << from << std::endl;
+
+        if(val > to) std::cout << "ivalid value: actual=" << val << ", expected= <=" <<
+                                   to << std::endl;
+    }
+}
+
+//Test time counting instrumentation in StandardAllocatorWithTimer.
+//The instrumentation is made with START_TEST and END_TEST macros from WrapperMacros.h.
+void test_allocator_with_timer(int N, int seed)
+{
+    printf("\n============= Allocator with timer test ============= \n");
+
+    StandardAllocatorWithTimer allocator;
+    VectorIterator<size_t> allocation_sizes =
+        AllocationSizes::generate_random_sizes(N, 32, 2048, seed);
+    memory_operation data;
+
+    double elaspsed_time = 0;
+    for (int i=0; i<N; i++) {
+        data = allocator.wrapped_malloc(allocation_sizes.next());
+        elaspsed_time += data.total_time;
+        allocator.wrapped_free(data.ptr);
+    }
+
+    printf("%d allocations and frees done in time: %f \n", N, elaspsed_time);
+
+    printf("\n==================================================== \n");
+}
+
+//Test behavior of TypesConfiguration class, by enabling and disabling types.
+void test_types_conf()
+{
+    TypesConf types;
+
+    for (unsigned i=0; i<FunctionCalls::NUM_OF_FUNCTIONS; i++) {
+        types.enable_type(i);
+        assert(types.is_enabled(i));
+        types.disable_type(i);
+        assert(!types.is_enabled(i));
+    }
+}
+
+void execute_self_tests()
+{
+    const int N = 10000;
+    const size_t size_from = 32, size_to = 2048;
+    const unsigned seed = 11;
+
+    test_types_conf();
+
+    {
+        VectorIterator<size_t> it = AllocationSizes::generate_random_sizes(N, size_from,
+                                                                           size_to,seed);
+        test_iterator_values(it, size_from, size_to);
+    }
+
+    {
+        TypesConf enable_func_calls;
+        enable_func_calls.enable_type(FunctionCalls::MALLOC);
+
+        VectorIterator<int> it = FunctionCalls::generate_random_allocator_func_calls(N,
+                                                                                     seed, enable_func_calls);
+        test_iterator(it, N);
+    }
+
+//Timer implementation __cplusplus < 201100L is based on CPU clocks counting and it will fail on this test.
+#if __cplusplus > 201100L
+    test_timer();
+#endif
+
+    printf("Test completed! (press ENTER to continue)\n");
+}
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Thread.hpp b/memkind-1.10.0/test/allocator_perf_tool/Thread.hpp
new file mode 100644
index 0000000..3b83ccc
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Thread.hpp
@@ -0,0 +1,99 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include <pthread.h>
+#include <assert.h>
+
+#include "Runnable.hpp"
+
+class Thread
+{
+public:
+    Thread(Runnable *runnable) : runnable_task(runnable) {}
+
+    void start()
+    {
+        int err = pthread_create(&thread_handle, NULL, execute_thread,
+                                 static_cast<void *>(runnable_task));
+        assert(!err);
+    };
+
+    void wait()
+    {
+        pthread_join(thread_handle, NULL);
+    };
+
+    Runnable *get_runnable_task()
+    {
+        return runnable_task;
+    }
+
+private:
+    static void *execute_thread(void *ptr)
+    {
+        Runnable *runnable = static_cast<Runnable *>(ptr);
+        assert(runnable);
+        runnable->run();
+        pthread_exit(NULL);
+    }
+
+    pthread_t thread_handle;
+    Runnable *runnable_task;
+};
+
+
+class ThreadsManager
+{
+public:
+    ThreadsManager(std::vector<Thread *> &threads_vec) : threads(threads_vec) {}
+    ~ThreadsManager()
+    {
+        release();
+    }
+
+    void start()
+    {
+        for (int i=0; i<threads.size(); i++) {
+            threads[i]->start();
+        }
+    }
+
+    void barrier()
+    {
+        for (int i=0; i<threads.size(); i++) {
+            threads[i]->wait();
+        }
+    }
+
+    void release()
+    {
+        for (int i=0; i<threads.size(); i++) {
+            delete threads[i];
+        }
+        threads.clear();
+    }
+
+private:
+    std::vector<Thread *> &threads;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/TimerSysTime.hpp b/memkind-1.10.0/test/allocator_perf_tool/TimerSysTime.hpp
new file mode 100644
index 0000000..55e69dd
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/TimerSysTime.hpp
@@ -0,0 +1,48 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <sys/time.h>
+
+class TimerSysTime
+{
+public:
+    void start()
+    {
+        gettimeofday(&last, 0);
+    }
+
+    double getElapsedTime() const
+    {
+        struct timeval now;
+        gettimeofday(&now, 0);
+        double time_delta_sec = ((double)now.tv_sec - (double)last.tv_sec);
+        double time_delta_usec = ((double)now.tv_usec - (double)last.tv_usec) /
+                                 1000000.0;
+        return time_delta_sec + time_delta_usec;
+    }
+
+private:
+    struct timeval last;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/VectorIterator.hpp b/memkind-1.10.0/test/allocator_perf_tool/VectorIterator.hpp
new file mode 100644
index 0000000..3bb1863
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/VectorIterator.hpp
@@ -0,0 +1,69 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Iterator.hpp"
+
+#include <vector>
+
+template<class T>
+class VectorIterator
+    : public Iterator<T>
+{
+public:
+
+    static VectorIterator create(int init_size)
+    {
+        return VectorIterator(init_size);
+    }
+    static VectorIterator create(std::vector<T> v)
+    {
+        return VectorIterator(v);
+    }
+
+    bool has_next() const
+    {
+        return !vec.empty();
+    }
+
+    T next()
+    {
+        T val = vec.back();
+        vec.pop_back();
+        return val;
+    }
+
+    size_t size() const
+    {
+        return vec.size();
+    }
+
+private:
+    VectorIterator(std::vector<T> v)
+    {
+        vec = v;
+    }
+
+    std::vector<T> vec;
+};
diff --git a/memkind-1.10.0/test/allocator_perf_tool/Workload.hpp b/memkind-1.10.0/test/allocator_perf_tool/Workload.hpp
new file mode 100644
index 0000000..96de5a3
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/Workload.hpp
@@ -0,0 +1,37 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Allocator.hpp"
+
+class Workload
+{
+public:
+    virtual bool run() = 0;
+    virtual ~Workload(void) {}
+
+protected:
+    Allocator *allocator;
+};
+
diff --git a/memkind-1.10.0/test/allocator_perf_tool/WrappersMacros.hpp b/memkind-1.10.0/test/allocator_perf_tool/WrappersMacros.hpp
new file mode 100644
index 0000000..7069f59
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/WrappersMacros.hpp
@@ -0,0 +1,43 @@
+/*
+* Copyright (C) 2015 - 2017 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <ctime>
+#include <sys/time.h>
+
+#include <chrono>
+#define START_TEST(ALLOCATOR, METHOD) \
+	memory_operation data; \
+	data.allocator_type = ALLOCATOR; \
+	data.allocation_method = METHOD;\
+	data.error_code = 0; \
+	std::chrono::system_clock::time_point last = std::chrono::high_resolution_clock::now();
+#define END_TEST \
+	std::chrono::system_clock::time_point now = std::chrono::high_resolution_clock::now(); \
+	std::chrono::duration<double,std::milli> elapsedTime(now - last); \
+	data.total_time = elapsedTime.count() / 1000.0; \
+	data.size_of_allocation = size; \
+	data.error_code = errno; \
+	data.is_allocated = true; \
+	return data;
diff --git a/memkind-1.10.0/test/allocator_perf_tool/main.cpp b/memkind-1.10.0/test/allocator_perf_tool/main.cpp
new file mode 100644
index 0000000..3aa3057
--- /dev/null
+++ b/memkind-1.10.0/test/allocator_perf_tool/main.cpp
@@ -0,0 +1,231 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <stdio.h>
+#include <assert.h>
+#include <iostream>
+#include <vector>
+
+#include "Configuration.hpp"
+#include "AllocatorFactory.hpp"
+#include "TaskFactory.hpp"
+#include "Task.hpp"
+#include "ConsoleLog.hpp"
+#include "Stats.hpp"
+#include "Thread.hpp"
+#include "Tests.hpp"
+#include "CommandLine.hpp"
+#include "FunctionCallsPerformanceTask.h"
+#include "StressIncreaseToMax.h"
+
+/*
+Command line description.
+Syntax:
+	key=value
+Options:
+	- 'test' - specify the test case. This option can be used with the following values: 'calls', 'all' or 'self',
+	where:
+		'calls' - function calls performance test,
+		'all' - execute both above ('footprint' and 'calls') tests,
+		'self' - execute self tests
+		's1' - stress tests
+		(perform allocations until the maximum amount of allocated memory has been reached, than frees allocated memory.
+		If the time interval has not been exceed, than repeat the test),
+	- 'operations' - the number of memory operations per thread
+	- 'size_from' - lower bound for the random sizes of allocation
+	- 'size_to' - upper bound for the random sizes of allocation
+	- 'seed' - random seed
+	- 'threads_num' - the number of threads per test case
+	- 'time' - minimum execution time interval
+	- 'kind' - the kind to test
+	- 'csv_log' - if 'true' then log to csv file memory operations and statistics
+	- 'call' specify the allocation function call. This option can be used with the following values: 'malloc' (default), 'calloc', 'realloc',
+	- 'requested_memory_limit' test stops when the requested memory limit has been reached
+* - maximum of available memory in OS, or maximum memory based 'operations' parameter
+Example:
+1. Performance test:
+./perf_tool test=all operations=1000 size_from=32 size_to=20480 seed=11 threads_num=200
+2. Stress test
+./perf_tool test=s1 time=120 kind=MEMKIND_HBW size_from=1048576 csv_log=true requested_memory_limit=1048576
+*/
+
+int main(int argc, char *argv[])
+{
+    unsigned mem_operations_num = 1000;
+    size_t size_from = 32, size_to = 2048*1024;
+    unsigned seed = 11;
+    //should be at least one
+    size_t threads_number = 10;
+
+    CommandLine cmd_line(argc, argv);
+
+    if((argc >= 1) && cmd_line.is_option_set("test", "self")) {
+        execute_self_tests();
+        getchar();
+    }
+
+    cmd_line.parse_with_strtol("operations", mem_operations_num);
+    cmd_line.parse_with_strtol("size_from", size_from);
+    cmd_line.parse_with_strtol("size_to", size_to);
+    cmd_line.parse_with_strtol("seed", seed);
+    cmd_line.parse_with_strtol("threads_num", threads_number);
+
+    bool is_csv_log_enabled = cmd_line.is_option_set("csv_log", "true");
+
+    //Heap Manager initialization
+    std::vector<AllocatorFactory::initialization_stat> stats =
+        AllocatorFactory().initialization_test();
+
+    if(!cmd_line.is_option_set("print_init_stats", "false")) {
+        printf("\nInitialization overhead:\n");
+        for (int i=0; i<stats.size(); i++) {
+            AllocatorFactory::initialization_stat stat = stats[i];
+            printf("%32s : time=%7.7f.s, ref_delta_time=%15f, node0=%10fMB, node1=%7.7fMB\n",
+                   AllocatorTypes::allocator_name(stat.allocator_type).c_str(),
+                   stat.total_time,
+                   stat.ref_delta_time,
+                   stat.memory_overhead[0],
+                   stat.memory_overhead[1]);
+        }
+    }
+
+    //Stress test by repeatedly increasing memory (to maximum), until given time interval has been exceed.
+    if(cmd_line.is_option_set("test", "s1")) {
+        printf("Stress test (StressIncreaseToMax) start. \n");
+
+        if(!cmd_line.is_option_present("operations"))
+            mem_operations_num = 1000000;
+
+        unsigned time = 120; //Default time interval.
+        cmd_line.parse_with_strtol("time", time);
+
+        size_t requested_memory_limit = 1024*1024;
+        cmd_line.parse_with_strtol("requested_memory_limit", requested_memory_limit);
+
+        unsigned allocator = AllocatorTypes::MEMKIND_HBW;
+        if(cmd_line.is_option_present("kind")) {
+            //Enable memkind allocator and specify kind.
+            allocator = AllocatorTypes::allocator_type(cmd_line.get_option_value("kind"));
+        }
+        TypesConf allocator_types;
+        allocator_types.enable_type(allocator);
+
+        TypesConf enable_func_calls;
+        enable_func_calls.enable_type(FunctionCalls::MALLOC);
+
+        TaskConf task_conf = {
+            mem_operations_num,
+            {
+                mem_operations_num,
+                size_from, //No random sizes.
+                size_from
+            },
+            enable_func_calls,
+            allocator_types,
+            11,
+            is_csv_log_enabled,
+        };
+
+        StressIncreaseToMax::execute_test_iterations(task_conf, time,
+                                                     requested_memory_limit);
+        return 0;
+    }
+
+    printf("\nTest configuration: \n");
+    printf("\t memory operations per thread = %u \n", mem_operations_num);
+    printf("\t seed = %d\n", seed);
+    printf("\t number of threads = %zu\n", threads_number);
+    printf("\t size from-to = %zu-%zu\n\n", size_from, size_to);
+
+    assert(size_from <= size_to);
+
+
+    TypesConf func_calls;
+    func_calls.enable_type(FunctionCalls::FREE);
+
+    if(cmd_line.is_option_present("call")) {
+        //Enable heap manager function call.
+        func_calls.enable_type(FunctionCalls::function_type(
+                                   cmd_line.get_option_value("call")));
+    } else {
+        func_calls.enable_type(FunctionCalls::MALLOC);
+    }
+
+    TypesConf allocator_types;
+    if(cmd_line.is_option_present("allocator")) {
+        allocator_types.enable_type(AllocatorTypes::allocator_type(
+                                        cmd_line.get_option_value("allocator")));
+    } else {
+        for(unsigned i = 0; i <= AllocatorTypes::MEMKIND_HBW_PREFERRED; i++) {
+            allocator_types.enable_type(i);
+        }
+    }
+
+    TaskConf conf = {
+        mem_operations_num, //number memory operations
+        {
+            mem_operations_num, //number of memory operations
+            size_from, //min. size of single allocation
+            size_to //max. size of single allocatioion
+        },
+        func_calls, //enable function calls
+        allocator_types, //enable allocators
+        seed, //random seed
+        is_csv_log_enabled,
+    };
+
+    //Function calls test
+    if(cmd_line.is_option_set("test", "calls") ||
+       cmd_line.is_option_set("test", "all")) {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+
+        for (int i=0; i<threads_number; i++) {
+            FunctionCallsPerformanceTask *task =
+                static_cast<FunctionCallsPerformanceTask *>(
+                    task_factory.create(conf)
+                );
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats stats;
+        for (int i=0; i<tasks.size(); i++) {
+            stats += tasks[i]->get_results();
+        }
+
+        ConsoleLog::print_table(stats);
+        ConsoleLog::print_requested_memory(stats, "func. calls test");
+
+        threads_manager.release();
+    }
+
+    return 0;
+}
diff --git a/memkind-1.10.0/test/autohbw_test.py b/memkind-1.10.0/test/autohbw_test.py
new file mode 100755
index 0000000..661ec62
--- /dev/null
+++ b/memkind-1.10.0/test/autohbw_test.py
@@ -0,0 +1,80 @@
+#
+#  Copyright (C) 2016 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+import pytest
+import os
+from python_framework import CMD_helper
+
+def _get_lib_path():
+    for p in ["/usr/lib64", "/usr/lib"]:
+        if os.path.isdir(p):
+            return p
+    raise Exception("Cannot find library path in OS")
+
+class Test_autohbw(object):
+    binary = "../autohbw_test_helper"
+    fail_msg = "Test failed with:\n {0}"
+    test_prefix = "AUTO_HBW_LOG=2 LD_PRELOAD=%s/libautohbw.so.0 " % _get_lib_path()
+    memkind_malloc_log = "In my memkind malloc"
+    memkind_calloc_log = "In my memkind calloc"
+    memkind_realloc_log = "In my memkind realloc"
+    memkind_posix_memalign_log = "In my memkind align"
+    memkind_free_log = "In my memkind free"
+    cmd_helper = CMD_helper()
+
+    def test_TC_MEMKIND_autohbw_malloc_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding malloc() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " malloc"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_malloc_log in output, self.fail_msg.format("\nError: malloc was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("\nError: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_autohbw_calloc_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding calloc() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " calloc"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_calloc_log in output, self.fail_msg.format("\nError: calloc was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("Error: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_autohbw_realloc_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding realloc() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " realloc"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_realloc_log in output, self.fail_msg.format("\nError: realloc was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("\nError: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_autohbw_posix_memalign_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding posix_memalign() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " posix_memalign"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_posix_memalign_log in output, self.fail_msg.format("\nError: posix_memalign was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("\nError: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
diff --git a/memkind-1.10.0/test/autohbw_test_helper.c b/memkind-1.10.0/test/autohbw_test_helper.c
new file mode 100644
index 0000000..c04d9ef
--- /dev/null
+++ b/memkind-1.10.0/test/autohbw_test_helper.c
@@ -0,0 +1,80 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+int main(int argc, char *argv[])
+{
+    int err = 0;
+    const size_t size = 1024 * 1024;
+    void *buf = NULL;
+
+    //It is expected that "malloc", "calloc", "realloc" or "posix_memalign" argument is passed
+    if (argc != 2) {
+        printf("Error: Wrong number of parameters\n");
+        err = -1;
+        return err;
+    }
+
+    if (strcmp(argv[1], "malloc") == 0) {
+        buf = malloc(size);
+        if (buf == NULL) {
+            printf("Error: malloc returned NULL\n");
+            err = -1;
+        }
+        free(buf);
+    } else if (strcmp(argv[1], "calloc") == 0) {
+        buf = calloc(size, 1);
+        if (buf == NULL) {
+            printf("Error: calloc returned NULL\n");
+            err = -1;
+        }
+        free(buf);
+    } else if (strcmp(argv[1], "realloc") == 0) {
+        buf = malloc(size);
+        if (buf == NULL) {
+            printf("Error: malloc before realloc returned NULL\n");
+            err = -1;
+        }
+        buf = realloc(buf, size * 2);
+        if (buf == NULL) {
+            printf("Error: realloc returned NULL\n");
+            err = -1;
+        }
+        free(buf);
+    } else if (strcmp(argv[1], "posix_memalign") == 0) {
+        err = posix_memalign(&buf, 64, size);
+        if (err != 0) {
+            printf("Error: posix_memalign returned %d\n", err);
+        }
+        free(buf);
+    } else {
+        printf("Error: unknown parameter\n");
+        err = -1;
+    }
+
+    return err;
+}
diff --git a/memkind-1.10.0/test/bat_tests.cpp b/memkind-1.10.0/test/bat_tests.cpp
new file mode 100644
index 0000000..9d4d318
--- /dev/null
+++ b/memkind-1.10.0/test/bat_tests.cpp
@@ -0,0 +1,532 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include <fstream>
+#include <algorithm>
+#include <numaif.h>
+#include <memkind/internal/memkind_regular.h>
+
+#include "common.h"
+#include "check.h"
+#include "omp.h"
+#include "trial_generator.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+#include "Allocator.hpp"
+#include "TestPolicy.hpp"
+
+#define TEST_PREFIX "test_TC_MEMKIND_BAT_"
+
+typedef void(*test_function)(Allocator *, size_t);
+
+typedef std::tuple<test_function, memkind_memtype_t, memkind_policy_t, memkind_bits_t, size_t>
+memtype_policy_test_params;
+
+test_function get_function(memtype_policy_test_params params)
+{
+    return std::get<0>(params);
+}
+
+memkind_memtype_t get_memtype(memtype_policy_test_params params)
+{
+    return std::get<1>(params);
+}
+
+memkind_policy_t get_policy(memtype_policy_test_params params)
+{
+    return std::get<2>(params);
+}
+
+memkind_bits_t get_flags(memtype_policy_test_params params)
+{
+    return std::get<3>(params);
+}
+
+size_t get_size(memtype_policy_test_params params)
+{
+    return std::get<4>(params);
+}
+
+typedef std::tuple<test_function, hbw_policy_t, size_t> hbw_policy_test_params;
+
+test_function get_function(hbw_policy_test_params params)
+{
+    return std::get<0>(params);
+}
+
+hbw_policy_t get_hbw_policy(hbw_policy_test_params params)
+{
+    return std::get<1>(params);
+}
+
+size_t get_size(hbw_policy_test_params params)
+{
+    return std::get<2>(params);
+}
+
+typedef std::tuple<hbw_policy_t, size_t> hbw_policy_huge_page_test_params;
+
+hbw_policy_t get_hbw_policy(hbw_policy_huge_page_test_params params)
+{
+    return std::get<0>(params);
+}
+
+size_t get_size(hbw_policy_huge_page_test_params params)
+{
+    return std::get<1>(params);
+}
+
+
+
+/*
+ * Set of basic acceptance tests.
+ */
+class BATest: public TGTest {};
+class BasicAllocTest
+{
+public:
+
+    BasicAllocTest(Allocator *allocator) : allocator(allocator) {}
+
+    void check_policy_and_numa_node(void *ptr, size_t size)
+    {
+        int policy = allocator->get_numa_policy();
+
+        EXPECT_NE(-1, policy);
+
+        if (allocator->is_high_bandwidth()) {
+            TestPolicy::check_hbw_numa_nodes(policy, ptr, size);
+        } else {
+            TestPolicy::check_all_numa_nodes(policy, ptr, size);
+        }
+    }
+
+    void record_page_association(void *ptr, size_t size)
+    {
+        TestPolicy::record_page_association(ptr, size, allocator->get_page_size());
+    }
+
+    void malloc(size_t size)
+    {
+        void *ptr = allocator->malloc(size);
+        ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+        void *memset_ret = memset(ptr, 3, size);
+        EXPECT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void calloc(size_t num, size_t size)
+    {
+        void *ptr = allocator->calloc(num, size);
+        ASSERT_TRUE(ptr != NULL) << "calloc() returns NULL";
+        void *memset_ret = memset(ptr, 3, size);
+        ASSERT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void realloc(size_t size)
+    {
+        void *ptr = allocator->malloc(size);
+        ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+        size_t realloc_size = size+128;
+        ptr = allocator->realloc(ptr, realloc_size);
+        ASSERT_TRUE(ptr != NULL) << "realloc() returns NULL";
+        void *memset_ret = memset(ptr, 3, realloc_size);
+        ASSERT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void memalign(size_t alignment, size_t size)
+    {
+        void *ptr = NULL;
+        int ret = allocator->memalign(&ptr, alignment, size);
+        ASSERT_EQ(0, ret) << "posix_memalign() != 0";
+        ASSERT_TRUE(ptr != NULL) << "posix_memalign() returns NULL pointer";
+        void *memset_ret = memset(ptr, 3, size);
+        ASSERT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void free(size_t size)
+    {
+        void *ptr = allocator->malloc(size);
+        ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+        allocator->free(ptr);
+    }
+
+    virtual ~BasicAllocTest() {}
+private:
+    Allocator *allocator;
+};
+
+
+static void test_malloc(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).malloc(size);
+}
+
+static void test_calloc(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).calloc(1, size);
+}
+
+static void test_realloc(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).realloc(size);
+}
+
+static void test_memalign(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).memalign(4096, size);
+}
+
+static void test_free(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).free(size);
+}
+
+
+template <typename T>
+std::vector<T> GetKeys(std::map<T, std::string> dict)
+{
+    std::vector<T> keys;
+    for (auto const &item: dict) {
+        keys.push_back(item.first);
+    }
+    return keys;
+}
+
+struct TestParameters {
+    static const std::map<test_function, std::string> functions;
+    static const std::map<memkind_memtype_t, std::string> memtypes;
+    static const std::map<memkind_policy_t, std::string> policies;
+    static const std::map<hbw_policy_t, std::string> hbw_policies;
+    static const std::vector<size_t> sizes;
+};
+
+const map<test_function, std::string> TestParameters::functions = {
+    {&test_malloc, "malloc"},
+    {&test_calloc, "calloc"},
+    {&test_realloc, "realloc"},
+    {&test_memalign, "memalign"},
+    {&test_free, "free"}
+};
+const std::map<memkind_memtype_t, std::string> TestParameters::memtypes = {{MEMKIND_MEMTYPE_DEFAULT, "DEFAULT"},
+    {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, "HIGH_BANDWIDTH"}
+};
+const std::map<memkind_policy_t, std::string> TestParameters::policies = {{MEMKIND_POLICY_PREFERRED_LOCAL, "PREFERRED_LOCAL"},
+    {MEMKIND_POLICY_BIND_LOCAL, "BIND_LOCAL"},
+    {MEMKIND_POLICY_BIND_ALL, "BIND_ALL"},
+    {MEMKIND_POLICY_INTERLEAVE_ALL, "MEMKIND_POLICY_INTERLEAVE_ALL"}
+};
+const std::map<hbw_policy_t, std::string> TestParameters::hbw_policies = {{HBW_POLICY_PREFERRED, "HBW_POLICY_PREFERRED"},
+    {HBW_POLICY_BIND, "HBW_POLICY_BIND"},
+    {HBW_POLICY_BIND_ALL, "HBW_POLICY_BIND_ALL"},
+    {HBW_POLICY_INTERLEAVE, "HBW_POLICY_INTERLEAVE"}
+};
+const std::vector<size_t> TestParameters::sizes = {4096, 4194305};
+
+
+class MemtypePolicyTest: public TGTest,
+    public ::testing::WithParamInterface<memtype_policy_test_params>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<memtype_policy_test_params> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::functions.at(get_function(info.param)) + "_" +
+               TestParameters::memtypes.at(get_memtype(info.param)) + "_" +
+               TestParameters::policies.at(get_policy(info.param)) + "_" +
+               (get_flags(info.param) == MEMKIND_MASK_PAGE_SIZE_2MB ? "PAGE_SIZE_2MB_" : "") +
+               std::to_string(get_size(info.param)) + "_bytes";
+    }
+};
+
+TEST_P(MemtypePolicyTest, memkind_heap_mgmt)
+{
+    HugePageOrganizer huge_page_organizer(8);
+    auto params = GetParam();
+    auto flags = get_flags(params);
+    MemkindAllocator allocator(get_memtype(GetParam()), get_policy(GetParam()),
+                               flags);
+    get_function(GetParam())(&allocator, get_size(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(DEFAULT,
+                        MemtypePolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::Values(MEMKIND_MEMTYPE_DEFAULT),
+                                           ::testing::Values(MEMKIND_POLICY_PREFERRED_LOCAL),
+                                           ::testing::Values(MEMKIND_MASK_PAGE_SIZE_2MB, memkind_bits_t()),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemtypePolicyTest::get_test_name_suffix
+                       );
+
+INSTANTIATE_TEST_CASE_P(HBW,
+                        MemtypePolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::Values(MEMKIND_MEMTYPE_HIGH_BANDWIDTH),
+                                           ::testing::ValuesIn(GetKeys(TestParameters::policies)),
+                                           ::testing::Values(memkind_bits_t()),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemtypePolicyTest::get_test_name_suffix
+                       );
+
+INSTANTIATE_TEST_CASE_P(HBW_HUGE,
+                        MemtypePolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::Values(MEMKIND_MEMTYPE_HIGH_BANDWIDTH),
+                                           ::testing::Values(MEMKIND_POLICY_PREFERRED_LOCAL,
+                                                             MEMKIND_POLICY_BIND_LOCAL,
+                                                             MEMKIND_POLICY_BIND_ALL),
+                                           ::testing::Values(MEMKIND_MASK_PAGE_SIZE_2MB),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemtypePolicyTest::get_test_name_suffix
+                       );
+
+
+class HBWPolicyHugePageTest: public TGTest,
+    public ::testing::WithParamInterface<hbw_policy_huge_page_test_params>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<hbw_policy_huge_page_test_params> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::hbw_policies.at(get_hbw_policy(info.param)) + "_" +
+               std::to_string(get_size(info.param)) + "_bytes";
+    }
+};
+
+TEST_P(HBWPolicyHugePageTest, memalign)
+{
+    HugePageOrganizer huge_page_organizer(8);
+    HbwmallocAllocator hbwmalloc_allocator(get_hbw_policy(GetParam()));
+    hbwmalloc_allocator.set_memalign_page_size(HBW_PAGESIZE_2MB);
+    BasicAllocTest(&hbwmalloc_allocator).memalign(4096, get_size(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(HBW_memalign,
+                        HBWPolicyHugePageTest,
+                        ::testing::Combine(::testing::Values(HBW_POLICY_PREFERRED, HBW_POLICY_BIND,
+                                                             HBW_POLICY_BIND_ALL),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        HBWPolicyHugePageTest::get_test_name_suffix
+                       );
+
+
+class HBWPolicyTest: public TGTest,
+    public ::testing::WithParamInterface<hbw_policy_test_params>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<hbw_policy_test_params> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::functions.at(get_function(info.param)) + "_" +
+               TestParameters::hbw_policies.at(get_hbw_policy(info.param)) + "_" +
+               std::to_string(get_size(info.param)) + "_bytes";
+    }
+};
+
+TEST_P(HBWPolicyTest, memkind_heap_mgmt)
+{
+    HbwmallocAllocator allocator(get_hbw_policy(GetParam()));
+    get_function(GetParam())(&allocator, get_size(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(HBWPolicy,
+                        HBWPolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::ValuesIn(GetKeys(TestParameters::hbw_policies)),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        HBWPolicyTest::get_test_name_suffix
+                       );
+
+
+class MemkindRegularTest: public TGTest,
+    public ::testing::WithParamInterface<tuple<test_function, size_t>>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<std::tuple<test_function, size_t>> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::functions.at(std::get<0>(info.param)) + "_" +
+               std::to_string(std::get<1>(info.param)) + "_bytes";
+    }
+};
+
+
+TEST_P(MemkindRegularTest, memkind_heap_mgmt)
+{
+    MemkindAllocator allocator(MEMKIND_REGULAR);
+    std::get<0>(GetParam())(&allocator, std::get<1>(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(Regular,
+                        MemkindRegularTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemkindRegularTest::get_test_name_suffix
+                       );
+
+
+TEST_F(BATest,
+       test_TC_MEMKIND_malloc_DEFAULT_HIGH_BANDWIDTH_INTERLEAVE_ALL_4194305_bytes)
+{
+    MemkindAllocator allocator((memkind_memtype_t)(MEMKIND_MEMTYPE_DEFAULT |
+                                                   MEMKIND_MEMTYPE_HIGH_BANDWIDTH), MEMKIND_POLICY_INTERLEAVE_ALL,
+                               memkind_bits_t());
+    test_malloc(&allocator, 4194305);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_free_MEMKIND_DEFAULT_free_with_NULL_kind_4096_bytes)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 4096);
+    ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+    memkind_free(0, ptr);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_free_ext_MEMKIND_GBTLB_4096_bytes)
+{
+    HugePageOrganizer huge_page_organizer(1000);
+    MemkindAllocator memkind_allocator(MEMKIND_GBTLB);
+    BasicAllocTest(&memkind_allocator).free(4096);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbwmalloc_Pref_CheckAvailable)
+{
+    ASSERT_EQ(0, hbw_check_available());
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbw_malloc_usable_size_nullptr_0bytes)
+{
+    ASSERT_EQ(0U, hbw_malloc_usable_size(nullptr));
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbw_malloc_usable_size_hbw_malloc_16bytes)
+{
+    void *ptr = hbw_malloc(16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 16U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_malloc_16bytes_def_kind)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 16U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_hbw_malloc_usable_size_hbw_calloc_16bytes_16bytes)
+{
+    void *ptr = hbw_calloc(16, 16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 16U*16U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_calloc_16bytes_16bytes_def_kind)
+{
+    void *ptr = memkind_calloc(MEMKIND_DEFAULT, 16, 16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 16U*16U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbw_malloc_usable_size_hbw_realloc_1024bytes)
+{
+    void *ptr = hbw_realloc(nullptr, 1024);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 1024U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_realloc_1024bytes_def_kind)
+{
+    void *ptr = memkind_realloc(MEMKIND_DEFAULT, nullptr, 1024);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 1024U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_hbw_malloc_usable_size_hbw_posix_memalign_32bytes)
+{
+    void *ptr = nullptr;
+    int res = hbw_posix_memalign(&ptr, 64, 32);
+    ASSERT_EQ(0, res);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 32U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_posix_memalign_32bytes_def_kind)
+{
+    void *ptr = nullptr;
+    int res = memkind_posix_memalign(MEMKIND_DEFAULT, &ptr, 64, 32);
+    ASSERT_EQ(0, res);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 32U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbwmalloc_Pref_Policy)
+{
+    hbw_set_policy(HBW_POLICY_PREFERRED);
+    EXPECT_EQ(HBW_POLICY_PREFERRED, hbw_get_policy());
+}
+
+TEST_F(BATest, test_TC_MEMKIND_REGULAR_nodemask)
+{
+    using namespace TestPolicy;
+    void *mem = memkind_malloc(MEMKIND_REGULAR, 1234567);
+    unique_bitmask_ptr kind_nodemask = make_nodemask_ptr();
+    ASSERT_EQ(0, memkind_regular_all_get_mbind_nodemask(MEMKIND_REGULAR,
+                                                        kind_nodemask->maskp,
+                                                        kind_nodemask->size));
+
+    check_numa_nodes(kind_nodemask, MPOL_BIND, mem, 1234567);
+    memkind_free(MEMKIND_REGULAR, mem);
+}
+
diff --git a/memkind-1.10.0/test/check.cpp b/memkind-1.10.0/test/check.cpp
new file mode 100644
index 0000000..a038d1e
--- /dev/null
+++ b/memkind-1.10.0/test/check.cpp
@@ -0,0 +1,269 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <numaif.h>
+#include <errno.h>
+#include <fstream>
+#include <sstream>
+#include <iostream>
+#include <fstream>
+#include <string>
+#include <cstring>
+#include <numa.h>
+#include <errno.h>
+
+#include "check.h"
+
+
+using namespace std;
+
+
+/*Check each page between the start
+and the end address additionally also
+check the end address for pagesize*/
+Check::Check(const void *p, const trial_t &trial): Check(p, trial.size,
+                                                             trial.page_size)
+{
+}
+
+Check::Check(const void *p, const size_t size, const size_t page_size)
+{
+    const size_t min_page_size = 4096;
+    this->ptr = p;
+    this->size = size;
+    size_t psize = (page_size >= min_page_size ? page_size : min_page_size);
+    if (p && size && psize) {
+        num_address = size / psize;
+        num_address += size % psize ? 1 : 0;
+
+        address = new void *[num_address];
+        size_t i;
+        for (i = 0; i < num_address - 1; ++i) {
+            address[i] = (char *)ptr + i * psize;
+        }
+        address[i] = (char *)p + size - 1;
+    } else {
+        address = NULL;
+        num_address = 0;
+    }
+}
+
+Check::~Check()
+{
+    delete[] address;
+}
+
+Check::Check(const Check &other)
+{
+    num_address = other.num_address;
+
+    address = new void *[num_address];
+    for (size_t i = 0; i < num_address; ++i) {
+        address[i] = other.address[i];
+    }
+}
+
+int Check::check_zero(void)
+{
+    size_t i;
+    const char *cptr = (char *)ptr;
+    for (i = 0; i < size; ++i) {
+        if (cptr[i] != '\0') {
+            return -1;
+        }
+    }
+    return 0;
+}
+
+int Check::check_align(size_t align)
+{
+    return (size_t)ptr % align;
+}
+
+string Check::skip_to_next_entry (ifstream &ip)
+{
+    string temp, token;
+    size_t found = 0;
+    string empty ="";
+
+    while (!ip.eof()) {
+        getline (ip, temp);
+        found = temp.find("-");
+        if (found != string::npos) {
+            istringstream iss(temp);
+            getline(iss, token, ' ');
+            return token;
+        }
+    }
+    return empty;
+}
+
+string Check::skip_to_next_kpage(ifstream &ip)
+{
+    string temp, token;
+    size_t found = 0;
+    string empty ="";
+
+    while (!ip.eof()) {
+        getline (ip, temp);
+        found = temp.find("KernelPageSize:");
+        if (found != string::npos) {
+            return temp;
+        }
+    }
+    return empty;
+}
+
+
+void Check::get_address_range(string &line,
+                              unsigned long long *start_addr,
+                              unsigned long long *end_addr)
+{
+    stringstream ss(line);
+    string token;
+
+    getline(ss, token, '-');
+    *start_addr = strtoul(token.c_str(),
+                          NULL,
+                          16);
+    getline(ss, token, '-');
+    *end_addr = strtoul(token.c_str(),
+                        NULL,
+                        16);
+}
+
+size_t Check::get_kpagesize(string line)
+{
+    stringstream ss(line);
+    string token;
+    size_t pagesize;
+
+
+    ss  >> token;
+    ss  >> token;
+
+    pagesize = atol(token.c_str());
+
+    return (size_t)pagesize;
+}
+
+int Check::check_page_size(size_t page_size)
+{
+    int err = 0;
+    size_t i;
+
+    ip.open ("/proc/self/smaps");
+
+    populate_smaps_table();
+    if (check_page_size(page_size, address[0])) {
+        err = -1;
+    }
+    for (i = 1; i < num_address && !err; ++i) {
+        if (check_page_size(page_size, address[i])) {
+            err = i;
+        }
+    }
+    return err;
+}
+
+int Check::populate_smaps_table ()
+{
+    string read;
+    size_t lpagesize;
+    smaps_entry_t lentry;
+    unsigned long long start_addr;
+    unsigned long long end_addr;
+
+    ip >> read;
+    while (!ip.eof()) {
+
+        start_addr = end_addr = 0;
+        get_address_range (read,
+                           &start_addr,
+                           &end_addr);
+        read = skip_to_next_kpage(ip);
+        getline(ip, read);
+        lpagesize = get_kpagesize(read);
+        lpagesize *= 1024;
+        lentry.start_addr = start_addr;
+        lentry.end_addr = end_addr;
+        lentry.pagesize = lpagesize;
+        smaps_table.push_back(lentry);
+        read = skip_to_next_entry(ip);
+        if (read.empty()) {
+            break;
+        }
+    }
+
+    if (0 == smaps_table.size()) {
+        fprintf(stderr,"Empty smaps table\n");
+        return -1;
+    } else {
+        return 0;
+    }
+
+}
+
+int Check::check_page_size(size_t page_size, void *vaddr)
+{
+    string read;
+    unsigned long long virt_addr;
+    size_t lpagesize;
+    list<smaps_entry_t>::iterator it;
+    unsigned long long start_addr;
+    unsigned long long end_addr;
+
+    virt_addr = (unsigned long long)(vaddr);
+
+    for (it = smaps_table.begin();
+         it != smaps_table.end();
+         it++) {
+
+        start_addr = it->start_addr;
+        end_addr = it->end_addr;
+
+        if ((virt_addr >= start_addr) &&
+            (virt_addr < end_addr)) {
+            lpagesize = it->pagesize;
+            if (lpagesize == page_size) {
+                return 0;
+            } else {
+                /*The pagesize of allocation and req don't match*/
+                fprintf(stderr,"%zd does not match entry in SMAPS (%zd)\n",
+                        page_size, lpagesize);
+                return -1;
+            }
+        }
+    }
+    /*Never found a match!*/
+    return 1;
+}
diff --git a/memkind-1.10.0/test/check.h b/memkind-1.10.0/test/check.h
new file mode 100644
index 0000000..f9935f3
--- /dev/null
+++ b/memkind-1.10.0/test/check.h
@@ -0,0 +1,68 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef check_include_h
+#define check_include_h
+
+#include <list>
+#include "trial_generator.h"
+
+
+typedef struct {
+    unsigned long start_addr;
+    unsigned long end_addr;
+    size_t pagesize;
+} smaps_entry_t;
+
+
+
+using namespace std;
+
+class Check
+{
+public:
+    Check(const void *p, const trial_t &trial);
+    Check(const void *p, const size_t size, const size_t page_size);
+    Check(const Check &);
+    ~Check();
+    int check_page_size(size_t page_size);
+    int check_zero(void);
+    int check_align(size_t align);
+private:
+    const void *ptr;
+    size_t size;
+    void **address;
+    size_t num_address;
+    list<smaps_entry_t>smaps_table;
+    ifstream ip;
+    string skip_to_next_entry(ifstream &);
+    string skip_to_next_kpage(ifstream &);
+    void get_address_range(string &line, unsigned long long *start_addr,
+                           unsigned long long *end_addr);
+    size_t get_kpagesize(string line);
+    int check_page_size(size_t page_size, void *vaddr);
+    int populate_smaps_table();
+};
+
+#endif
diff --git a/memkind-1.10.0/test/common.h b/memkind-1.10.0/test/common.h
new file mode 100644
index 0000000..bca4ff6
--- /dev/null
+++ b/memkind-1.10.0/test/common.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef COMMON_H
+#define COMMON_H
+
+#include <hbwmalloc.h>
+
+#include <iostream>
+#include <stdint.h>
+#include <stdlib.h>
+#include <gtest/gtest.h>
+
+#define MB 1048576ULL
+#define GB 1073741824ULL
+#define KB 1024ULL
+
+#define HBW_SUCCESS 0
+#define HBW_ERROR -1
+
+#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
+
+#endif
diff --git a/memkind-1.10.0/test/dax_kmem_env_var_test.py b/memkind-1.10.0/test/dax_kmem_env_var_test.py
new file mode 100644
index 0000000..ef314fc
--- /dev/null
+++ b/memkind-1.10.0/test/dax_kmem_env_var_test.py
@@ -0,0 +1,67 @@
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+from distutils.spawn import find_executable
+from python_framework import CMD_helper
+import os
+
+class Test_dax_kmem_env_var(object):
+    os.environ["PATH"] += os.pathsep + os.path.dirname(os.path.dirname(__file__))
+    binary_path = find_executable("memkind-auto-dax-kmem-nodes")
+    environ_err_test = "../environ_err_dax_kmem_malloc_test"
+    environ_err_positive_test = "../environ_err_dax_kmem_malloc_positive_test"
+    expected_libnuma_warning = "libnuma: Warning: node argument -1 is out of range\n\n"
+    fail_msg = "Test failed with:\n {0}"
+    cmd_helper = CMD_helper()
+
+    def get_dax_kmem_nodes(self, nodemask=None):
+        """ This function executes memkind function 'get_mbind_nodemask' and returns its output - comma-separated DAX_KMEM nodes """
+        command = self.binary_path
+        if (nodemask):
+            command = "MEMKIND_DAX_KMEM_NODES={}".format(nodemask) + command
+        output, retcode = self.cmd_helper.execute_cmd(command)
+        assert retcode == 0, self.fail_msg.format("\nError: Execution of \'{0}\' returns {1}, \noutput: {2}".format(command, retcode, output))
+        print("\nExecution of {} returns output {}".format(command, output))
+        return output
+
+    def test_TC_MEMKIND_dax_kmem_env_var_compare_nodemask_default_and_env_variable(self):
+        """ This test checks whether dax_kmem_nodemask_default and dax_kmem_nodemask_env_variable have the same value """
+        dax_kmem_nodemask_default = self.get_dax_kmem_nodes()
+        dax_kmem_nodemask_env_variable = self.get_dax_kmem_nodes(dax_kmem_nodemask_default)
+        assert dax_kmem_nodemask_default == dax_kmem_nodemask_env_variable, self.fail_msg.format("Error: Nodemask dax_kmem_nodemask_default ({0}) " \
+               "is not the same as nodemask dax_kmem_nodemask_env_variable ({1})".format(dax_kmem_nodemask_default, dax_kmem_nodemask_env_variable))
+
+    def test_TC_MEMKIND_dax_kmem_env_var_negative_memkind_malloc(self):
+        """ This test sets usupported value of MEMKIND_DAX_KMEM_NODES, then tries to perform a successful allocation from DRAM using memkind_malloc() """
+        command = "MEMKIND_DAX_KMEM_NODES=-1 " + self.cmd_helper.get_command_path(self.environ_err_test)
+        output, retcode = self.cmd_helper.execute_cmd(command)
+        assert retcode != 0, self.fail_msg.format("\nError: Execution of: \'{0}\' returns: {1} \noutput: {2}".format(command, retcode, output))
+        assert self.expected_libnuma_warning == output, self.fail_msg.format("Error: expected libnuma warning ({0}) " \
+               "was not found (output: {1})").format(self.expected_libnuma_warning, output)
+
+    def test_TC_MEMKIND_dax_kmem_env_var_proper_memkind_malloc(self):
+        """ This test checks if allocation is performed on persistent memory correctly """
+        dax_kmem_nodemask_default = self.get_dax_kmem_nodes().strip()
+        command = "MEMKIND_DAX_KMEM_NODES={} ".format(dax_kmem_nodemask_default) + self.cmd_helper.get_command_path(self.environ_err_positive_test)
+        output, retcode = self.cmd_helper.execute_cmd(command)
+        assert retcode == 0, self.fail_msg.format("\nError: Execution of: \'{0}\' returns: {1} \noutput: {2}".format(command, retcode, output))
diff --git a/memkind-1.10.0/test/dax_kmem_nodes.cpp b/memkind-1.10.0/test/dax_kmem_nodes.cpp
new file mode 100644
index 0000000..f810d77
--- /dev/null
+++ b/memkind-1.10.0/test/dax_kmem_nodes.cpp
@@ -0,0 +1,97 @@
+/*
+* Copyright (C) 2020 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <climits>
+#include <numa.h>
+#include <numaif.h>
+
+#include "dax_kmem_nodes.h"
+
+DaxKmem::DaxKmem()
+{
+    dax_kmem_nodes = get_dax_kmem_nodes();
+}
+
+size_t DaxKmem::get_free_space(void)
+{
+    size_t sum_of_dax_kmem_free_space = 0;
+    long long free_space;
+
+    for(auto const &node: dax_kmem_nodes) {
+        numa_node_size64(node, &free_space);
+        sum_of_dax_kmem_free_space += free_space;
+    }
+
+    return sum_of_dax_kmem_free_space;
+}
+
+std::set<int> DaxKmem::get_closest_numa_nodes(int regular_node)
+{
+    int min_distance = INT_MAX;
+    std::set<int> closest_numa_ids;
+
+    for (auto const &node: dax_kmem_nodes) {
+        int distance_to_i_node = numa_distance(regular_node, node);
+
+        if (distance_to_i_node < min_distance) {
+            min_distance = distance_to_i_node;
+            closest_numa_ids.clear();
+            closest_numa_ids.insert(node);
+        } else if (distance_to_i_node == min_distance) {
+            closest_numa_ids.insert(node);
+        }
+    }
+
+    return closest_numa_ids;
+}
+
+std::set<int> DaxKmem::get_dax_kmem_nodes(void)
+{
+    struct bitmask *cpu_mask = numa_allocate_cpumask();
+    std::set<int> dax_kmem_nodes;
+
+    const int MAXNODE_ID = numa_max_node();
+    for (int id = 0; id <= MAXNODE_ID; ++id) {
+        numa_node_to_cpus(id, cpu_mask);
+
+        // Check if NUMA node exists and if it is NUMA node created from persistent memory
+        if (numa_node_size64(id, nullptr) > 0 &&
+            numa_bitmask_weight(cpu_mask) == 0) {
+            dax_kmem_nodes.insert(id);
+        }
+    }
+    numa_free_cpumask(cpu_mask);
+
+    return dax_kmem_nodes;
+}
+
+size_t DaxKmem::size(void)
+{
+    return dax_kmem_nodes.size();
+}
+
+bool DaxKmem::contains(int node)
+{
+    return (dax_kmem_nodes.find(node) != dax_kmem_nodes.end());
+}
diff --git a/memkind-1.10.0/test/dax_kmem_nodes.h b/memkind-1.10.0/test/dax_kmem_nodes.h
new file mode 100644
index 0000000..058e0bc
--- /dev/null
+++ b/memkind-1.10.0/test/dax_kmem_nodes.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright (C) 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <set>
+
+class DaxKmem
+{
+public:
+    DaxKmem();
+    size_t get_free_space(void);
+    std::set<int> get_closest_numa_nodes(int regular_node);
+    size_t size(void);
+    bool contains(int node);
+
+private:
+    std::set<int> dax_kmem_nodes;
+    std::set<int> get_dax_kmem_nodes(void);
+}; // DaxKmem
diff --git a/memkind-1.10.0/test/decorator_test.cpp b/memkind-1.10.0/test/decorator_test.cpp
new file mode 100644
index 0000000..152111a
--- /dev/null
+++ b/memkind-1.10.0/test/decorator_test.cpp
@@ -0,0 +1,116 @@
+/*
+ * Copyright (C) 2014 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include "common.h"
+#include "decorator_test.h"
+#include "config.h"
+
+size_t size = 16;
+memkind_t kind = MEMKIND_DEFAULT;
+
+class DecoratorTest: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {
+        decorators_state = new decorators_flags();
+    }
+
+    void TearDown()
+    {
+        free(decorators_state);
+    }
+};
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_malloc)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_malloc(kind, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    EXPECT_EQ(1, decorators_state->malloc_pre);
+    EXPECT_EQ(1, decorators_state->malloc_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_calloc)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_calloc(kind, 1, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    EXPECT_EQ(1, decorators_state->calloc_pre);
+    EXPECT_EQ(1, decorators_state->calloc_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_posix_memalign)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer;
+
+    int res = memkind_posix_memalign(kind, &buffer, 8, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    ASSERT_EQ(0, res);
+    EXPECT_EQ(1, decorators_state->posix_memalign_pre);
+    EXPECT_EQ(1, decorators_state->posix_memalign_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_realloc)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_realloc(kind, NULL, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    EXPECT_EQ(1, decorators_state->realloc_pre);
+    EXPECT_EQ(1, decorators_state->realloc_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_free)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_malloc(kind, size);
+
+    ASSERT_TRUE(buffer != NULL);
+
+    memkind_free(0, buffer);
+
+    EXPECT_EQ(1, decorators_state->free_pre);
+    EXPECT_EQ(1, decorators_state->free_post);
+#endif
+}
\ No newline at end of file
diff --git a/memkind-1.10.0/test/decorator_test.h b/memkind-1.10.0/test/decorator_test.h
new file mode 100644
index 0000000..d713b73
--- /dev/null
+++ b/memkind-1.10.0/test/decorator_test.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <memkind.h>
+
+struct decorators_flags {
+    int malloc_pre;
+    int malloc_post;
+    int calloc_pre;
+    int calloc_post;
+    int posix_memalign_pre;
+    int posix_memalign_post;
+    int realloc_pre;
+    int realloc_post;
+    int free_pre;
+    int free_post;
+};
+
+struct decorators_flags *decorators_state;
+
+extern "C" {
+
+    void memkind_malloc_pre(struct memkind *kind, size_t size)
+    {
+        decorators_state->malloc_pre++;
+    }
+
+    void memkind_malloc_post(struct memkind *kind, size_t size, void **result)
+    {
+        decorators_state->malloc_post++;
+    }
+
+    void memkind_calloc_pre(struct memkind *kind, size_t nmemb, size_t size)
+    {
+        decorators_state->calloc_pre++;
+    }
+
+    void memkind_calloc_post(struct memkind *kind, size_t nmemb, size_t size,
+                             void **result)
+    {
+        decorators_state->calloc_post++;
+    }
+
+    void memkind_posix_memalign_pre(struct memkind *kind, void **memptr,
+                                    size_t alignment, size_t size)
+    {
+        decorators_state->posix_memalign_pre++;
+    }
+
+    void memkind_posix_memalign_post(struct memkind *kind, void **memptr,
+                                     size_t alignment, size_t size, int *err)
+    {
+        decorators_state->posix_memalign_post++;
+    }
+
+    void memkind_realloc_pre(struct memkind *kind, void *ptr, size_t size)
+    {
+        decorators_state->realloc_pre++;
+    }
+
+    void memkind_realloc_post(struct memkind *kind, void *ptr, size_t size,
+                              void **result)
+    {
+        decorators_state->realloc_post++;
+    }
+
+    void memkind_free_pre(struct memkind **kind, void **ptr)
+    {
+        decorators_state->free_pre++;
+    }
+
+    void memkind_free_post(struct memkind **kind, void **ptr)
+    {
+        decorators_state->free_post++;
+    }
+
+}
diff --git a/memkind-1.10.0/test/dlopen_test.cpp b/memkind-1.10.0/test/dlopen_test.cpp
new file mode 100644
index 0000000..8968f13
--- /dev/null
+++ b/memkind-1.10.0/test/dlopen_test.cpp
@@ -0,0 +1,108 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+#include <dlfcn.h>
+
+#include "common.h"
+
+class DlopenTest: public :: testing::Test
+{
+protected:
+    DlopenTest()
+    {
+        const char *path = "/usr/lib64/libmemkind.so";
+        if (!pathExists(path)) {
+            path = "/usr/lib/libmemkind.so";
+        }
+        dlerror();
+        handle = dlopen(path, RTLD_LAZY);
+        assert((handle != NULL && dlerror() == NULL) && "Couldn't open libmemkind.so");
+        memkind_malloc = (memkind_malloc_t)dlsym(handle, "memkind_malloc");
+        assert(dlerror() == NULL && "Couldn't get memkind_malloc from memkind library");
+        memkind_free = (memkind_free_t)dlsym(handle, "memkind_free");
+        assert(dlerror() == NULL && "Couldn't get memkind_free from memkind library");
+    }
+
+    ~DlopenTest()
+    {
+        dlclose(handle);
+    }
+
+    void test(const char *kind_name, size_t alloc_size)
+    {
+        void **kind_ptr = (void **)dlsym(handle, kind_name);
+        EXPECT_TRUE(dlerror() == NULL) << "Couldn't get kind from memkind library";
+        EXPECT_TRUE(kind_ptr != NULL) << "Kind ptr to memkind library is NULL";
+
+        void *allocation_ptr = memkind_malloc((*kind_ptr), alloc_size);
+        EXPECT_TRUE(allocation_ptr != NULL) << "Allocation with memkind_malloc failed";
+
+        memset(allocation_ptr, 0, alloc_size);
+
+        memkind_free((*kind_ptr), allocation_ptr);
+    }
+
+    bool pathExists(const char *p)
+    {
+        struct stat info;
+        if (0 != stat(p, &info)) {
+            return false;
+        }
+        return true;
+    }
+
+private:
+    void *handle;
+    typedef void *(*memkind_malloc_t)(void *, size_t);
+    typedef void (*memkind_free_t)(void *, void *);
+    memkind_malloc_t memkind_malloc;
+    memkind_free_t memkind_free;
+};
+
+TEST_F(DlopenTest, test_TC_MEMKIND_DEFAULT_4194305_bytes)
+{
+    test("MEMKIND_DEFAULT", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_4194305_bytes)
+{
+    test("MEMKIND_HBW", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_HUGETLB_4194305_bytes)
+{
+    HugePageOrganizer huge_page_organizer(8);
+    test("MEMKIND_HBW_HUGETLB", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_PREFERRED_4194305_bytes)
+{
+    test("MEMKIND_HBW_PREFERRED", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_INTERLEAVE_4194305_bytes)
+{
+    test("MEMKIND_HBW_INTERLEAVE", 4194305);
+}
diff --git a/memkind-1.10.0/test/draw_plots.py b/memkind-1.10.0/test/draw_plots.py
new file mode 100644
index 0000000..1d26f60
--- /dev/null
+++ b/memkind-1.10.0/test/draw_plots.py
@@ -0,0 +1,271 @@
+#
+#  Copyright (C) 2016 - 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+from mpl_toolkits.mplot3d import Axes3D
+import matplotlib.pyplot as plt
+import numpy as np
+import numpy.ma as ma
+import os
+from shutil import rmtree
+
+files = ('alloctest_hbw.txt', 'alloctest_glibc.txt', 'alloctest_tbb.txt', 'alloctest_pmem.txt')
+legend = ('avg hbw', 'avg glibc', 'avg tbb', 'avg pmem', 'first operation')
+colors = ('red', 'green', 'blue', 'black')
+first_operation_color = 'yellow'
+
+threads_values = ('1', '2', '4', '8', '16', '18', '36')
+
+small_sizes_values = ('1', '4', '16')
+medium_sizes_values = ('64', '256')
+big_sizes_values = ('1024', '4096', '16384')
+sizes_values = (small_sizes_values, medium_sizes_values, big_sizes_values)
+sizes_display = ('small', 'medium', 'big')
+
+operations = ('Allocation', 'Free', 'Total')
+
+iterations = 1000
+output_directory = './plots/'
+
+threads_index      = np.arange(len(threads_values))
+small_sizes_index  = np.arange(len(small_sizes_values))
+medium_sizes_index = np.arange(len(medium_sizes_values))
+big_sizes_index    = np.arange(len(big_sizes_values))
+sizes_index = (small_sizes_index, medium_sizes_index, big_sizes_index)
+
+# 3D and 2D plots needs different width, so that bars do not cover each other in 3D
+bar_width_2D = 0.2
+bar_width_3D = 0.1
+
+# return times (total, allocation, free, first allocation, first free) as columns 2 - 6
+def return_times(entry):
+    return entry[:,2], entry[:,3], entry[:,4], entry[:,5], entry[:,6]
+
+# initialize axis with labels, ticks and values
+def init_axis(fig, suptitle, subplot_projection, x_label, x_ticks, x_values, y_label, y_ticks, y_values, z_label):
+    assert fig is not None
+    fig.clf()
+    fig.suptitle(suptitle)
+    if subplot_projection:
+        ax = fig.add_subplot(111, projection=subplot_projection)
+    else:
+        ax = fig.add_subplot(111)
+    assert ax is not None
+    if x_label:
+        ax.set_xlabel(x_label)
+        if x_ticks is not None and x_values is not None:
+            ax.set_xticks(x_ticks)
+            ax.set_xticklabels(x_values)
+    if y_label:
+        ax.set_ylabel(y_label)
+        if y_ticks is not None and y_values is not None:
+            ax.set_yticks(y_ticks)
+            ax.set_yticklabels(y_values)
+    if z_label:
+        ax.set_zlabel(z_label)
+    return ax
+
+def draw_bar(ax, show_first_operation, x, y, time, init_time, draw_color):
+    assert ax is not None
+    if y is None:
+        if show_first_operation is True:
+            mask_time = ma.where(time>=init_time)
+            mask_init_time = ma.where(init_time>=time)
+            ax.bar(x[mask_time], time[mask_time], width=bar_width, color=draw_color)
+            ax.bar(x, init_time, width=bar_width, color=first_operation_color)
+            ax.bar(x[mask_init_time], time[mask_init_time], width=bar_width, color=draw_color)
+        else:
+            ax.bar(x, time, width=bar_width, color=draw_color)
+    else:
+        if show_first_operation is True:
+            mask_time = ma.where(time>=init_time)
+            mask_init_time = ma.where(init_time>=time)
+            ax.bar(x[mask_time], (time - init_time)[mask_time], y[mask_time], bottom=init_time[mask_time], zdir='y', width=bar_width, color=draw_color)
+            ax.bar(x[mask_init_time], (init_time - time)[mask_init_time], y[mask_init_time], bottom=time[mask_init_time], zdir='y', width=bar_width, color=first_operation_color)
+            ax.bar(x[mask_init_time], time[mask_init_time], y[mask_init_time], zdir='y', width=bar_width, color=draw_color)
+            ax.bar(x[mask_time], init_time[mask_time], y[mask_time], zdir='y', width=bar_width, color=first_operation_color)
+        else:
+            ax.bar(x, time, y, zdir='y', width=bar_width, color=draw_color)
+
+def save_plot(show_first_operation, subfolder, filename):
+    if show_first_operation:
+        path = os.path.join(output_directory, "with_first_operation")
+    else:
+        path = os.path.join(output_directory, "without_first_operation")
+    if subfolder:
+        path = os.path.join(path, subfolder)
+    if not os.path.exists(path):
+        os.mkdir(path)
+    plt.savefig(os.path.join(path, filename))
+    print "Saved file %s" % filename
+
+def load_data_from_files():
+    data = []
+    # load all files into data
+    for f in files:
+        assert os.path.isfile(f) is True
+        data.append(np.loadtxt(f, comments='#'))
+    return data
+
+def set_bar_width_and_offsets(requested_width):
+    bar_width = requested_width
+    offsets = (0, bar_width, 2*bar_width, 3*bar_width)
+    return bar_width, offsets
+
+if os.path.exists(output_directory):
+    rmtree(output_directory)
+os.mkdir(output_directory)
+
+data = load_data_from_files()
+
+fig = plt.figure()
+
+########################################
+# Draw 3D plots (time, sizes, threads) #
+########################################
+
+bar_width, offsets = set_bar_width_and_offsets(bar_width_3D)
+
+# for each size range (small, medium, big)
+for size_values, size_index, size_display in zip(sizes_values, sizes_index, sizes_display):
+    # show plots with and without first operation
+    for show_first_operation in (True, False):
+        # for each operation (allocation, free, total)
+        for operation in operations:
+            # add bar_width to each element of size_index
+            ax = init_axis(fig, "%s time of %s sizes (%s iterations)" % (operation, size_display, iterations), '3d',
+                           'size [kB]', size_index + (bar_width,) * len(size_index), size_values,
+                           'threads', threads_index, threads_values,
+                           'time [ms]')
+            legend_data = []
+            # for each allocator (hbw, glibc, tbb, pmem)
+            for entry, offset, draw_color in zip(data, offsets, colors):
+                # remove all rows where column 1 (size) is not in size_values (current size range)
+                entry = entry[np.in1d(entry[:,1], np.array(size_values).astype(np.int))]
+                # convert column 0 (threads values) to thread index
+                threads_col = np.array([threads_values.index(str(n)) for n in map(int, entry[:,0])])
+                # convert column 1 (sizes values) to size index
+                size_col = [size_values.index(str(n)) for n in map(int, entry[:,1])]
+                # add offset to size index so that bars display near each other
+                size_col = np.array(size_col) + offset
+                total_time_col, alloc_time_col, free_time_col, first_alloc_time_col, first_free_time_col = return_times(entry)
+                if operation == 'Allocation':
+                    draw_bar(ax, show_first_operation, size_col, threads_col, alloc_time_col, first_alloc_time_col, draw_color)
+                elif operation == 'Free':
+                    draw_bar(ax, show_first_operation, size_col, threads_col, free_time_col, first_free_time_col, draw_color)
+                elif operation == 'Total':
+                    draw_bar(ax, show_first_operation, size_col, threads_col, total_time_col, first_alloc_time_col+first_free_time_col, draw_color)
+                legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=draw_color))
+            if show_first_operation is True:
+                legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=first_operation_color))
+            ax.legend(legend_data,legend,loc='best')
+            plt.grid()
+            save_plot(show_first_operation, None, "%s_time_of_%s_sizes_%s_iterations.png" % (operation, size_display, iterations))
+
+################################################
+# Draw 2D plots (time, sizes, constant thread) #
+################################################
+
+bar_width, offsets = set_bar_width_and_offsets(bar_width_2D)
+
+# for each size range (small, medium, big)
+for size_values, size_index, size_display in zip(sizes_values, sizes_index, sizes_display):
+    # show plots with and without first operation
+    for show_first_operation in (True, False):
+        for thread in threads_values:
+            # for each operation (allocation, free, total)
+            for operation in operations:
+                # add bar_width to each element of size_index
+                ax = init_axis(fig, "%s time of %s sizes with %s threads (%s operations)" % (operation, size_display, thread, iterations), None,
+                               'size [kB]', size_index + (bar_width,) * len(size_index), size_values,
+                               'time [ms]', None, None,
+                               None)
+                legend_data = []
+                # for each allocator (hbw, glibc, tbb, pmem)
+                for entry, offset, draw_color in zip(data, offsets, colors):
+                    # remove all rows where column 1 (size) is not in size_values (current size range)
+                    entry = entry[np.in1d(entry[:,1], np.array(size_values).astype(np.int))]
+                    # remove all rows where column 0 (threads) is not equal to currently analyzed thread value
+                    entry = entry[entry[:,0] == int(thread)]
+                    # convert column 0 (threads values) to thread index
+                    threads_col = np.array([threads_values.index(str(n)) for n in map(int, entry[:,0])])
+                    # convert column 1 (size values) to size index
+                    size_col = [size_values.index(str(n)) for n in map(int, entry[:,1])]
+                    # add offset to size index so that bars display near each other
+                    size_col = np.array(size_col) + offset
+                    total_time_col, alloc_time_col, free_time_col, first_alloc_time_col, first_free_time_col = return_times(entry)
+                    if operation == 'Allocation':
+                        draw_bar(ax, show_first_operation, size_col, None, alloc_time_col, first_alloc_time_col, draw_color)
+                    elif operation == 'Free':
+                        draw_bar(ax, show_first_operation, size_col, None, free_time_col, first_free_time_col, draw_color)
+                    elif operation == 'Total':
+                        draw_bar(ax, show_first_operation, size_col, None, total_time_col, first_alloc_time_col+first_free_time_col, draw_color)
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=draw_color))
+                if show_first_operation is True:
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=first_operation_color))
+                ax.legend(legend_data,legend,loc='best')
+                plt.grid()
+                save_plot(show_first_operation, "constant_thread", "%s_time_of_%s_sizes_with_%s_threads_%s_iterations.png" % (operation, size_display, thread, iterations))
+
+################################################
+# Draw 2D plots (time, threads, constant size) #
+################################################
+
+# for each size range (small, medium, big)
+for size_values, size_index, size_display in zip(sizes_values, sizes_index, sizes_display):
+    # show plots with and without first operation
+    for show_first_operation in (True, False):
+        for size in size_values:
+            # for each operation (allocation, free, total)
+            for operation in operations:
+                # add bar_width to each element of threads_index
+                ax = init_axis(fig, "%s time of %s kB (%s operations)" % (operation, size, iterations), None,
+                               'threads', threads_index + (bar_width,) * len(threads_index), threads_values,
+                               'time [ms]', None, None,
+                               None)
+                legend_data = []
+                # for each allocator (hbw, glibc, tbb, pmem)
+                for entry, offset, draw_color in zip(data, offsets, colors):
+                    # remove all rows where column 1 (size) is not in size_values (current size range)
+                    entry = entry[np.in1d(entry[:,1], np.array(size_values).astype(np.int))]
+                    # remove all rows where column 1 (size) is not equal to currently analyzed size value
+                    entry = entry[entry[:,1] == int(size)]
+                    # convert column 0 (threads values) to thread index
+                    threads_col = np.array([threads_values.index(str(n)) for n in map(int, entry[:,0])])
+                    # add offset to thread index so that bars display near each other
+                    threads_col = np.array(threads_col) + offset
+                    # convert column 1 (size values) to size index
+                    size_col = [size_values.index(str(n)) for n in map(int, entry[:,1])]
+                    total_time_col, alloc_time_col, free_time_col, first_alloc_time_col, first_free_time_col = return_times(entry)
+                    if operation == 'Allocation':
+                        draw_bar(ax, show_first_operation, threads_col, None, alloc_time_col, first_alloc_time_col, draw_color)
+                    elif operation == 'Free':
+                        draw_bar(ax, show_first_operation, threads_col, None, free_time_col, first_free_time_col, draw_color)
+                    elif operation == 'Total':
+                        draw_bar(ax, show_first_operation, threads_col, None, total_time_col, first_alloc_time_col+first_free_time_col, draw_color)
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=draw_color))
+                if show_first_operation is True:
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=first_operation_color))
+                ax.legend(legend_data,legend,loc='best')
+                plt.grid()
+                save_plot(show_first_operation, "constant_size", "%s_time_of_%s_kB_%s_operations.png" % (operation, size, iterations))
diff --git a/memkind-1.10.0/test/environ_err_dax_kmem_malloc_positive_test.cpp b/memkind-1.10.0/test/environ_err_dax_kmem_malloc_positive_test.cpp
new file mode 100644
index 0000000..ca17f04
--- /dev/null
+++ b/memkind-1.10.0/test/environ_err_dax_kmem_malloc_positive_test.cpp
@@ -0,0 +1,103 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "memkind/internal/memkind_dax_kmem.h"
+
+#include <numa.h>
+#include <numaif.h>
+
+#include "common.h"
+
+#include <climits>
+#include <vector>
+
+/* This test is run with overridden MEMKIND_DAX_KMEM_NODES environment variable
+ * and tries to perform allocation on persistent memory using memkind_malloc()
+ */
+int main()
+{
+    struct bitmask *all_dax_kmem_nodes_nodemask = nullptr;
+    void *ptr = nullptr;
+    int ret = 1;
+    std::vector<int> closest_numa_ids;
+    int expected_numa_id = -1;
+    int returned_numa_id = -1;
+    char *env_value_str = secure_getenv("MEMKIND_DAX_KMEM_NODES");
+    int process_cpu = sched_getcpu();
+    int process_node = numa_node_of_cpu(process_cpu);
+    unsigned numa_max_id = (unsigned) numa_max_node();
+    int min_distance;
+
+    if (!env_value_str) {
+        printf("Error: MEMKIND_DAX_KMEM environment variable is empty.\n");
+        goto exit;
+    }
+
+    ptr = memkind_malloc(MEMKIND_DAX_KMEM, KB);
+    if (ptr == nullptr) {
+        printf("Error: allocation failed\n");
+        goto exit;
+    }
+    memset(ptr, 'A', KB/2);
+
+    all_dax_kmem_nodes_nodemask = numa_parse_nodestring(env_value_str);
+    min_distance = INT_MAX;
+    for (unsigned i = 0; i <= numa_max_id; ++i) {
+        if (numa_bitmask_isbitset(all_dax_kmem_nodes_nodemask, i)) {
+            int distance_to_i_node = numa_distance(process_node, i);
+            if (distance_to_i_node < min_distance) {
+                min_distance = distance_to_i_node;
+                closest_numa_ids.clear();
+                closest_numa_ids.push_back(i);
+            } else if (distance_to_i_node == min_distance) {
+                closest_numa_ids.push_back(i);
+            }
+        }
+    }
+
+    get_mempolicy(&returned_numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+
+    for (auto const &closest_numa : closest_numa_ids) {
+        if (closest_numa == returned_numa_id) {
+            expected_numa_id = closest_numa;
+            ret = 0;
+            break;
+        }
+    }
+    if (ret) {
+        printf("Error: Memkind dax kmem and allocated pointer node id's are not equal: %d!=%d\n",
+               expected_numa_id, returned_numa_id);
+    }
+
+exit:
+    if (all_dax_kmem_nodes_nodemask) {
+        numa_free_nodemask(all_dax_kmem_nodes_nodemask);
+    }
+    if (ptr) {
+        memkind_free(MEMKIND_DAX_KMEM, ptr);
+    }
+
+    return ret;
+}
diff --git a/memkind-1.10.0/test/environ_err_dax_kmem_malloc_test.cpp b/memkind-1.10.0/test/environ_err_dax_kmem_malloc_test.cpp
new file mode 100644
index 0000000..e5c709f
--- /dev/null
+++ b/memkind-1.10.0/test/environ_err_dax_kmem_malloc_test.cpp
@@ -0,0 +1,93 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "memkind/internal/memkind_dax_kmem.h"
+
+#include <numa.h>
+#include <numaif.h>
+
+#include "common.h"
+
+/* This test is run with overridden MEMKIND_DAX_KMEM_NODES environment variable
+ * and tries to perform allocation on DRAM using memkind_malloc()
+ */
+int main()
+{
+    struct bitmask *expected_nodemask = nullptr;
+    struct bitmask *returned_nodemask = nullptr;
+    void *ptr = nullptr;
+    int ret = 0;
+    int status = 0;
+
+    ptr = memkind_malloc(MEMKIND_DAX_KMEM, KB);
+    if (ptr != nullptr) {
+        printf("Error: allocation to DAX_KMEM should fail\n");
+        goto exit;
+    }
+
+    ptr = memkind_malloc(MEMKIND_DEFAULT, KB);
+    if (ptr == nullptr) {
+        printf("Error: allocation failed\n");
+        goto exit;
+    }
+
+    expected_nodemask = numa_allocate_nodemask();
+    status = memkind_dax_kmem_all_get_mbind_nodemask(nullptr,
+                                                     expected_nodemask->maskp,
+                                                     expected_nodemask->size);
+    if (status != MEMKIND_ERROR_ENVIRON) {
+        printf("Error: wrong return value from memkind_dax_kmem_all_get_mbind_nodemask()\n");
+        printf("Expected: %d\n", MEMKIND_ERROR_ENVIRON);
+        printf("Actual: %d\n", status);
+        goto exit;
+    }
+
+    returned_nodemask = numa_allocate_nodemask();
+    status = get_mempolicy(nullptr, returned_nodemask->maskp,
+                           returned_nodemask->size,
+                           ptr, MPOL_F_ADDR);
+    if (status) {
+        printf("Error: get_mempolicy() returned %d\n", status);
+        goto exit;
+    }
+
+    ret = numa_bitmask_equal(returned_nodemask, expected_nodemask);
+    if (!ret) {
+        printf("Error: Memkind dax kmem and allocated pointer nodemasks are not equal\n");
+    }
+
+exit:
+    if (expected_nodemask) {
+        numa_free_nodemask(expected_nodemask);
+    }
+    if (returned_nodemask) {
+        numa_free_nodemask(returned_nodemask);
+    }
+    if (ptr) {
+        memkind_free(nullptr, ptr);
+    }
+
+    return ret;
+}
diff --git a/memkind-1.10.0/test/environ_err_hbw_malloc_test.cpp b/memkind-1.10.0/test/environ_err_hbw_malloc_test.cpp
new file mode 100644
index 0000000..9a954cc
--- /dev/null
+++ b/memkind-1.10.0/test/environ_err_hbw_malloc_test.cpp
@@ -0,0 +1,87 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbwmalloc.h>
+#include <memkind.h>
+#include "memkind/internal/memkind_hbw.h"
+
+#include <numa.h>
+#include <numaif.h>
+
+#include "common.h"
+
+/* This test is run with overridden MEMKIND_HBW_NODES environment variable
+ * and tries to perform allocation from DRAM using hbw_malloc() using
+ * default HBW_POLICY_PREFERRED policy.
+ */
+int main()
+{
+    struct bitmask *expected_nodemask = NULL;
+    struct bitmask *returned_nodemask = NULL;
+    void *ptr = NULL;
+    int ret = 0;
+    int status = 0;
+
+    ptr = hbw_malloc(KB);
+    if (ptr == NULL) {
+        printf("Error: allocation failed\n");
+        goto exit;
+    }
+
+    expected_nodemask = numa_allocate_nodemask();
+    status = memkind_hbw_all_get_mbind_nodemask(NULL, expected_nodemask->maskp,
+                                                expected_nodemask->size);
+    if (status != MEMKIND_ERROR_ENVIRON) {
+        printf("Error: wrong return value from memkind_hbw_all_get_mbind_nodemask()\n");
+        printf("Expected: %d\n", MEMKIND_ERROR_ENVIRON);
+        printf("Actual: %d\n", status);
+        goto exit;
+    }
+
+    returned_nodemask = numa_allocate_nodemask();
+    status = get_mempolicy(NULL, returned_nodemask->maskp, returned_nodemask->size,
+                           ptr, MPOL_F_ADDR);
+    if (status) {
+        printf("Error: get_mempolicy() returned %d\n", status);
+        goto exit;
+    }
+
+    ret = numa_bitmask_equal(returned_nodemask, expected_nodemask);
+    if (!ret) {
+        printf("Error: Memkind hbw and allocated pointer nodemasks are not equal\n");
+    }
+
+exit:
+    if (expected_nodemask) {
+        numa_free_nodemask(expected_nodemask);
+    }
+    if (returned_nodemask) {
+        numa_free_nodemask(returned_nodemask);
+    }
+    if (ptr) {
+        hbw_free(ptr);
+    }
+
+    return ret;
+}
diff --git a/memkind-1.10.0/test/error_message_tests.cpp b/memkind-1.10.0/test/error_message_tests.cpp
new file mode 100644
index 0000000..72f8acc
--- /dev/null
+++ b/memkind-1.10.0/test/error_message_tests.cpp
@@ -0,0 +1,94 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <errno.h>
+#include <gtest/gtest.h>
+
+/* Tests which calls APIS in wrong ways to generate Error Messages thrown by the
+ * the memkind library
+ */
+
+const int all_error_code[] = {
+    MEMKIND_ERROR_UNAVAILABLE,
+    MEMKIND_ERROR_MBIND,
+    MEMKIND_ERROR_MMAP,
+    MEMKIND_ERROR_MALLOC,
+    MEMKIND_ERROR_RUNTIME,
+    MEMKIND_ERROR_ENVIRON,
+    MEMKIND_ERROR_INVALID,
+    MEMKIND_ERROR_TOOMANY,
+    MEMKIND_ERROR_BADOPS,
+    MEMKIND_ERROR_HUGETLB,
+    MEMKIND_ERROR_ARENAS_CREATE,
+    MEMKIND_ERROR_OPERATION_FAILED,
+    MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE,
+    EINVAL,
+    ENOMEM
+};
+
+class ErrorMessage: public :: testing :: Test
+{
+protected:
+
+    void SetUp()
+    {
+    }
+    void TearDown()
+    {
+    }
+};
+
+TEST_F(ErrorMessage, test_TC_MEMKIND_ErrorMsgLength)
+{
+    size_t i;
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    for (i = 0; i < sizeof(all_error_code)/sizeof(all_error_code[0]); ++i) {
+        memkind_error_message(all_error_code[i], error_message,
+                              MEMKIND_ERROR_MESSAGE_SIZE);
+        EXPECT_TRUE(strlen(error_message) < MEMKIND_ERROR_MESSAGE_SIZE - 1);
+    }
+    memkind_error_message(MEMKIND_ERROR_UNAVAILABLE, NULL, 0);
+}
+
+TEST_F(ErrorMessage, test_TC_MEMKIND_ErrorMsgFormat)
+{
+    size_t i;
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    for (i = 0; i < sizeof(all_error_code)/sizeof(all_error_code[0]); ++i) {
+        memkind_error_message(all_error_code[i], error_message,
+                              MEMKIND_ERROR_MESSAGE_SIZE);
+        EXPECT_TRUE(strncmp(error_message, "<memkind>", strlen("<memkind>")) == 0);
+    }
+}
+
+TEST_F(ErrorMessage, test_TC_MEMKIND_ErrorMsgUndefMesg)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(-0xdeadbeef, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    EXPECT_TRUE(strncmp(error_message, "<memkind> Undefined error number:",
+                        strlen("<memkind> Undefined error number:")) == 0);
+}
+
diff --git a/memkind-1.10.0/test/fragmentation_benchmark_pmem.cpp b/memkind-1.10.0/test/fragmentation_benchmark_pmem.cpp
new file mode 100644
index 0000000..a0e9a4d
--- /dev/null
+++ b/memkind-1.10.0/test/fragmentation_benchmark_pmem.cpp
@@ -0,0 +1,199 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <iostream>
+#include <chrono>
+#include <cstdio>
+#include <cstring>
+#include <ctime>
+#include <random>
+#include <vector>
+
+#define MB (1024 * 1024)
+#define PRINT_FREQ 100000
+#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
+#define BENCHMARK_LOG "bench_single_thread_%Y%m%d-%H%M"
+
+static size_t block_size [] =
+{8519680, 4325376, 8519680, 4325376, 8519680, 4325376, 8519680, 4325376, 432517, 608478};
+
+static struct memkind *pmem_kind;
+static FILE *log_file;
+
+static const char *const log_tag[] = {
+    "_mem_default.log",
+    "_mem_conservative.log"
+};
+
+static void usage(char *name)
+{
+    fprintf(stderr,
+            "Usage: %s pmem_kind_dir_path pmem_size pmem_policy test_time_limit_in_sec\n",
+            name);
+}
+
+static int fragmentatation_test(size_t pmem_max_size, double test_time)
+{
+    std::random_device rd;
+    std::mt19937 mt(rd());
+    std::uniform_int_distribution<> m_size(0, ARRAY_SIZE(block_size) - 1);
+    char *pmem_str = nullptr;
+    std::vector<char *> pmem_strs;
+    size_t total_size = pmem_max_size;
+    size_t print_iter = 0;
+    size_t total_allocated = 0;
+    double elapsed_seconds;
+    auto start = std::chrono::steady_clock::now();
+
+    do {
+        print_iter++;
+        int index = m_size(mt);
+        size_t size = block_size[index];
+        size_t length = pmem_strs.size() / 2;
+        std::uniform_int_distribution<size_t> m_index(0, length - 1);
+
+        while ((pmem_str = static_cast<char *>(memkind_malloc(pmem_kind,
+                                                              size))) == nullptr) {
+            size_t to_evict = m_index(mt);
+            char *str_to_evict = pmem_strs[to_evict];
+            size_t evict_size = memkind_malloc_usable_size(pmem_kind, str_to_evict);
+            total_allocated -= evict_size;
+
+            if (total_allocated < total_size * 0.1) {
+                fprintf(stderr,"allocated less than 10 percent.\n");
+                return 1;
+            }
+            memkind_free(pmem_kind, str_to_evict);
+            pmem_strs.erase(pmem_strs.begin() + to_evict);
+        }
+        pmem_strs.push_back(pmem_str);
+        total_allocated += memkind_malloc_usable_size(pmem_kind, pmem_str);
+
+        if (print_iter % PRINT_FREQ == 0) {
+            fprintf(log_file,"%f\n", static_cast<double>(total_allocated) / total_size);
+            fflush(stdout);
+        }
+
+        auto finish = std::chrono::steady_clock::now();
+        elapsed_seconds = std::chrono::duration_cast<std::chrono::duration<double>>
+                          (finish - start).count();
+    } while (elapsed_seconds < test_time);
+    return 0;
+}
+
+static int create_pmem(const char *pmem_dir, size_t pmem_size,
+                       memkind_mem_usage_policy policy)
+{
+    int err = 0;
+    if (pmem_size == 0 ) {
+        fprintf(stderr, "Invalid size to pmem kind must be not equal zero.\n");
+        return 1;
+    }
+
+    if (policy > MEMKIND_MEM_USAGE_POLICY_MAX_VALUE) {
+        fprintf(stderr, "Invalid memory usage policy param %u.\n", policy);
+        return 1;
+    }
+
+    memkind_config *pmem_cfg = memkind_config_new();
+    if (!pmem_cfg) {
+        fprintf(stderr, "Unable to create pmem configuration.\n");
+        return 1;
+    }
+
+    memkind_config_set_path(pmem_cfg, pmem_dir);
+    memkind_config_set_size(pmem_cfg, pmem_size);
+    memkind_config_set_memory_usage_policy(pmem_cfg, policy);
+
+    err = memkind_create_pmem_with_config(pmem_cfg, &pmem_kind);
+    memkind_config_delete(pmem_cfg);
+    if (err) {
+        fprintf(stderr, "Unable to create pmem kind.\n");
+        return 1;
+    }
+
+    return 0;
+}
+
+static int create_log_file(memkind_mem_usage_policy policy)
+{
+    char file_name[100] = {'\0'};
+    auto result = std::time(nullptr);
+
+    strftime(file_name, 100, BENCHMARK_LOG, std::localtime(&result));
+    std::strcat(file_name, log_tag[policy]);
+
+    if ((log_file = fopen(file_name, "w+")) == nullptr) {
+        fprintf(stderr, "Cannot create output file %s.\n", file_name);
+        return 1;
+    }
+
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+    char *pmem_dir;
+    size_t pmem_size;
+    memkind_mem_usage_policy pmem_policy;
+    double test_time_limit_in_sec;
+    int status = 0;
+    int err = 0;
+
+    if (argc != 5) {
+        usage(argv[0]);
+        return 1;
+    } else {
+        pmem_dir = argv[1];
+        pmem_size = std::stoull(argv[2]) * MB;
+        pmem_policy = static_cast<memkind_mem_usage_policy>(std::stoul(argv[3]));
+        test_time_limit_in_sec = std::stod(argv[4]);
+    }
+
+    err = create_pmem(pmem_dir, pmem_size, pmem_policy);
+    if (err) {
+        fprintf(stderr, "Cannot create pmem.\n");
+        return 1;
+    }
+
+    err = create_log_file(pmem_policy);
+    if (err) {
+        memkind_destroy_kind(pmem_kind);
+        fprintf(stderr, "Cannot create log file.\n");
+        return 1;
+    }
+
+    status = fragmentatation_test(pmem_size, test_time_limit_in_sec);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        fprintf(stderr, "Unable to destroy pmem kind.\n");
+    }
+
+    fclose(log_file);
+    return status;
+
+}
diff --git a/memkind-1.10.0/test/freeing_memory_segfault_test.cpp b/memkind-1.10.0/test/freeing_memory_segfault_test.cpp
new file mode 100644
index 0000000..a40cba8
--- /dev/null
+++ b/memkind-1.10.0/test/freeing_memory_segfault_test.cpp
@@ -0,0 +1,50 @@
+/*
+* Copyright (C) 2017 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "memkind.h"
+#include "common.h"
+#include <thread>
+
+//This test reproduce segfault when using TBB library.
+class FreeingMemorySegfault: public :: testing::Test
+{
+protected:
+    void SetUp() {}
+    void TearDown() {}
+};
+
+TEST_F(FreeingMemorySegfault,
+       test_TC_MEMKIND_freeing_memory_after_thread_finish)
+{
+    void *ptr = nullptr;
+
+    std::thread t([&] {
+        ptr = memkind_malloc(MEMKIND_DEFAULT, 32);
+        ASSERT_TRUE(ptr != NULL);
+    });
+    t.join();
+
+    memkind_free(0, ptr);
+    SUCCEED();
+}
diff --git a/memkind-1.10.0/test/gb_page_tests_bind_policy.cpp b/memkind-1.10.0/test/gb_page_tests_bind_policy.cpp
new file mode 100644
index 0000000..9af41bb
--- /dev/null
+++ b/memkind-1.10.0/test/gb_page_tests_bind_policy.cpp
@@ -0,0 +1,92 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "hbwmalloc.h"
+#include "common.h"
+#include "trial_generator.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+/* This set of tests are intended to test HBW_PAGESIZE_1GB and
+ * HBW_PAGESIZE_1GB_STRICT flags.
+ */
+class GBPagesTestBindPolicy : public TGTest
+{
+protected:
+    void run(int iterations, size_t alignment, bool psize_strict)
+    {
+        std::vector<void *> addr_to_free;
+        hbw_set_policy(HBW_POLICY_BIND);
+        EXPECT_EQ(HBW_POLICY_BIND, hbw_get_policy());
+
+        for (int i = 0; i < iterations; i++) {
+            void *ptr = NULL;
+            int ret;
+            if (psize_strict) {
+                ret = hbw_posix_memalign_psize(&ptr, alignment, GB, HBW_PAGESIZE_1GB_STRICT);
+            } else {
+                ret = hbw_posix_memalign_psize(&ptr, alignment, GB, HBW_PAGESIZE_1GB);
+            }
+            ASSERT_EQ(0, ret);
+            ASSERT_FALSE(ptr == NULL);
+            ASSERT_EQ(0, hbw_verify_memory_region(ptr, GB, HBW_TOUCH_PAGES));
+            addr_to_free.push_back(ptr);
+        }
+
+        for (int i = 0; i < iterations; i++) {
+            hbw_free(addr_to_free[i]);
+        }
+    }
+
+private:
+    HugePageOrganizer huge_page_organizer = HugePageOrganizer(2250);
+};
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Misalign_Preferred_Bind_Strict_1GB)
+{
+    run(1, 2*GB, true);
+}
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Memalign_Psize_Bind_1GB)
+{
+    run(1, GB, true);
+}
+
+/*
+ * Below tests allocate GB pages incrementally.
+*/
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Memalign_Psize_Bind_2GB)
+{
+    run(2, GB, false);
+}
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Memalign_Psize_Strict_Bind_3GB)
+{
+    run(3, GB, true);
+}
diff --git a/memkind-1.10.0/test/get_arena_test.cpp b/memkind-1.10.0/test/get_arena_test.cpp
new file mode 100644
index 0000000..e1c125b
--- /dev/null
+++ b/memkind-1.10.0/test/get_arena_test.cpp
@@ -0,0 +1,91 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_arena.h>
+
+#include <algorithm>
+#include <vector>
+#include <gtest/gtest.h>
+#include <omp.h>
+#include <pthread.h>
+
+class GetArenaTest: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+};
+
+bool uint_comp(unsigned int a, unsigned int b)
+{
+    return (a < b);
+}
+
+TEST_F(GetArenaTest, test_TC_MEMKIND_ThreadHash)
+{
+#ifdef _OPENMP
+    int num_threads = omp_get_max_threads();
+    std::vector<unsigned int> arena_idx(num_threads);
+    unsigned int thread_idx, idx;
+    int err = 0;
+    size_t size = 0;
+    int i;
+    unsigned max_collisions, collisions;
+    const unsigned collisions_limit = 5;
+
+    //Initialize kind
+    memkind_malloc(MEMKIND_HBW, 0);
+
+    #pragma omp parallel shared(arena_idx) private(thread_idx)
+    {
+        thread_idx = omp_get_thread_num();
+        err = memkind_thread_get_arena(MEMKIND_HBW, &(arena_idx[thread_idx]), size);
+    }
+    ASSERT_TRUE(err == 0);
+    std::sort(arena_idx.begin(), arena_idx.end(), uint_comp);
+    idx = arena_idx[0];
+    collisions = 0;
+    max_collisions = 0;
+    for (i = 1; i < num_threads; ++i) {
+        if (arena_idx[i] == idx) {
+            collisions++;
+        } else {
+            if (collisions > max_collisions) {
+                max_collisions = collisions;
+            }
+            idx = arena_idx[i];
+            collisions = 0;
+        }
+    }
+    EXPECT_LE(max_collisions, collisions_limit);
+    RecordProperty("max_collisions", max_collisions);
+#else
+    std::cout << "[ SKIPPED ] Feature OPENMP not supported" << std::endl;
+#endif
+}
diff --git a/memkind-1.10.0/test/gtest_fused/gtest/gtest-all.cc b/memkind-1.10.0/test/gtest_fused/gtest/gtest-all.cc
new file mode 100644
index 0000000..1120ce8
--- /dev/null
+++ b/memkind-1.10.0/test/gtest_fused/gtest/gtest-all.cc
@@ -0,0 +1,10495 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+// Google C++ Testing Framework (Google Test)
+//
+// Sometimes it's desirable to build Google Test by compiling a single file.
+// This file serves this purpose.
+
+// This line ensures that gtest.h can be compiled on its own, even
+// when it's fused.
+#include "gtest.h"
+
+// The following lines pull in the real gtest *.cc files.
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Utilities for testing Google Test itself and code that uses Google Test
+// (e.g. frameworks built on top of Google Test).
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+#define GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+
+
+namespace testing {
+
+// This helper class can be used to mock out Google Test failure reporting
+// so that we can test Google Test or code that builds on Google Test.
+//
+// An object of this class appends a TestPartResult object to the
+// TestPartResultArray object given in the constructor whenever a Google Test
+// failure is reported. It can either intercept only failures that are
+// generated in the same thread that created this object or it can intercept
+// all generated failures. The scope of this mock object can be controlled with
+// the second argument to the two arguments constructor.
+class GTEST_API_ ScopedFakeTestPartResultReporter
+    : public TestPartResultReporterInterface {
+ public:
+  // The two possible mocking modes of this object.
+  enum InterceptMode {
+    INTERCEPT_ONLY_CURRENT_THREAD,  // Intercepts only thread local failures.
+    INTERCEPT_ALL_THREADS           // Intercepts all failures.
+  };
+
+  // The c'tor sets this object as the test part result reporter used
+  // by Google Test.  The 'result' parameter specifies where to report the
+  // results. This reporter will only catch failures generated in the current
+  // thread. DEPRECATED
+  explicit ScopedFakeTestPartResultReporter(TestPartResultArray* result);
+
+  // Same as above, but you can choose the interception scope of this object.
+  ScopedFakeTestPartResultReporter(InterceptMode intercept_mode,
+                                   TestPartResultArray* result);
+
+  // The d'tor restores the previous test part result reporter.
+  virtual ~ScopedFakeTestPartResultReporter();
+
+  // Appends the TestPartResult object to the TestPartResultArray
+  // received in the constructor.
+  //
+  // This method is from the TestPartResultReporterInterface
+  // interface.
+  virtual void ReportTestPartResult(const TestPartResult& result);
+ private:
+  void Init();
+
+  const InterceptMode intercept_mode_;
+  TestPartResultReporterInterface* old_reporter_;
+  TestPartResultArray* const result_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedFakeTestPartResultReporter);
+};
+
+namespace internal {
+
+// A helper class for implementing EXPECT_FATAL_FAILURE() and
+// EXPECT_NONFATAL_FAILURE().  Its destructor verifies that the given
+// TestPartResultArray contains exactly one failure that has the given
+// type and contains the given substring.  If that's not the case, a
+// non-fatal failure will be generated.
+class GTEST_API_ SingleFailureChecker {
+ public:
+  // The constructor remembers the arguments.
+  SingleFailureChecker(const TestPartResultArray* results,
+                       TestPartResult::Type type, const std::string& substr);
+  ~SingleFailureChecker();
+ private:
+  const TestPartResultArray* const results_;
+  const TestPartResult::Type type_;
+  const std::string substr_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(SingleFailureChecker);
+};
+
+}  // namespace internal
+
+}  // namespace testing
+
+// A set of macros for testing Google Test assertions or code that's expected
+// to generate Google Test fatal failures.  It verifies that the given
+// statement will cause exactly one fatal Google Test failure with 'substr'
+// being part of the failure message.
+//
+// There are two different versions of this macro. EXPECT_FATAL_FAILURE only
+// affects and considers failures generated in the current thread and
+// EXPECT_FATAL_FAILURE_ON_ALL_THREADS does the same but for all threads.
+//
+// The verification of the assertion is done correctly even when the statement
+// throws an exception or aborts the current function.
+//
+// Known restrictions:
+//   - 'statement' cannot reference local non-static variables or
+//     non-static members of the current object.
+//   - 'statement' cannot return a value.
+//   - You cannot stream a failure message to this macro.
+//
+// Note that even though the implementations of the following two
+// macros are much alike, we cannot refactor them to use a common
+// helper macro, due to some peculiarity in how the preprocessor
+// works.  The AcceptsMacroThatExpandsToUnprotectedComma test in
+// gtest_unittest.cc will fail to compile if we do that.
+#define EXPECT_FATAL_FAILURE(statement, substr) \
+  do { \
+    class GTestExpectFatalFailureHelper {\
+     public:\
+      static void Execute() { statement; }\
+    };\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kFatalFailure, (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter:: \
+          INTERCEPT_ONLY_CURRENT_THREAD, &gtest_failures);\
+      GTestExpectFatalFailureHelper::Execute();\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+#define EXPECT_FATAL_FAILURE_ON_ALL_THREADS(statement, substr) \
+  do { \
+    class GTestExpectFatalFailureHelper {\
+     public:\
+      static void Execute() { statement; }\
+    };\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kFatalFailure, (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter:: \
+          INTERCEPT_ALL_THREADS, &gtest_failures);\
+      GTestExpectFatalFailureHelper::Execute();\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+// A macro for testing Google Test assertions or code that's expected to
+// generate Google Test non-fatal failures.  It asserts that the given
+// statement will cause exactly one non-fatal Google Test failure with 'substr'
+// being part of the failure message.
+//
+// There are two different versions of this macro. EXPECT_NONFATAL_FAILURE only
+// affects and considers failures generated in the current thread and
+// EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS does the same but for all threads.
+//
+// 'statement' is allowed to reference local variables and members of
+// the current object.
+//
+// The verification of the assertion is done correctly even when the statement
+// throws an exception or aborts the current function.
+//
+// Known restrictions:
+//   - You cannot stream a failure message to this macro.
+//
+// Note that even though the implementations of the following two
+// macros are much alike, we cannot refactor them to use a common
+// helper macro, due to some peculiarity in how the preprocessor
+// works.  If we do that, the code won't compile when the user gives
+// EXPECT_NONFATAL_FAILURE() a statement that contains a macro that
+// expands to code containing an unprotected comma.  The
+// AcceptsMacroThatExpandsToUnprotectedComma test in gtest_unittest.cc
+// catches that.
+//
+// For the same reason, we have to write
+//   if (::testing::internal::AlwaysTrue()) { statement; }
+// instead of
+//   GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement)
+// to avoid an MSVC warning on unreachable code.
+#define EXPECT_NONFATAL_FAILURE(statement, substr) \
+  do {\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kNonFatalFailure, \
+        (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter:: \
+          INTERCEPT_ONLY_CURRENT_THREAD, &gtest_failures);\
+      if (::testing::internal::AlwaysTrue()) { statement; }\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+#define EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(statement, substr) \
+  do {\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kNonFatalFailure, \
+        (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter::INTERCEPT_ALL_THREADS, \
+          &gtest_failures);\
+      if (::testing::internal::AlwaysTrue()) { statement; }\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+
+#include <ctype.h>
+#include <math.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <wchar.h>
+#include <wctype.h>
+
+#include <algorithm>
+#include <iomanip>
+#include <limits>
+#include <list>
+#include <map>
+#include <ostream>  // NOLINT
+#include <sstream>
+#include <vector>
+
+#if GTEST_OS_LINUX
+
+// TODO(kenton@google.com): Use autoconf to detect availability of
+// gettimeofday().
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+
+# include <fcntl.h>  // NOLINT
+# include <limits.h>  // NOLINT
+# include <sched.h>  // NOLINT
+// Declares vsnprintf().  This header is not available on Windows.
+# include <strings.h>  // NOLINT
+# include <sys/mman.h>  // NOLINT
+# include <sys/time.h>  // NOLINT
+# include <unistd.h>  // NOLINT
+# include <string>
+
+#elif GTEST_OS_SYMBIAN
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+# include <sys/time.h>  // NOLINT
+
+#elif GTEST_OS_ZOS
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+# include <sys/time.h>  // NOLINT
+
+// On z/OS we additionally need strings.h for strcasecmp.
+# include <strings.h>  // NOLINT
+
+#elif GTEST_OS_WINDOWS_MOBILE  // We are on Windows CE.
+
+# include <windows.h>  // NOLINT
+# undef min
+
+#elif GTEST_OS_WINDOWS  // We are on Windows proper.
+
+# include <io.h>  // NOLINT
+# include <sys/timeb.h>  // NOLINT
+# include <sys/types.h>  // NOLINT
+# include <sys/stat.h>  // NOLINT
+
+# if GTEST_OS_WINDOWS_MINGW
+// MinGW has gettimeofday() but not _ftime64().
+// TODO(kenton@google.com): Use autoconf to detect availability of
+//   gettimeofday().
+// TODO(kenton@google.com): There are other ways to get the time on
+//   Windows, like GetTickCount() or GetSystemTimeAsFileTime().  MinGW
+//   supports these.  consider using them instead.
+#  define GTEST_HAS_GETTIMEOFDAY_ 1
+#  include <sys/time.h>  // NOLINT
+# endif  // GTEST_OS_WINDOWS_MINGW
+
+// cpplint thinks that the header is already included, so we want to
+// silence it.
+# include <windows.h>  // NOLINT
+# undef min
+
+#else
+
+// Assume other platforms have gettimeofday().
+// TODO(kenton@google.com): Use autoconf to detect availability of
+//   gettimeofday().
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+
+// cpplint thinks that the header is already included, so we want to
+// silence it.
+# include <sys/time.h>  // NOLINT
+# include <unistd.h>  // NOLINT
+
+#endif  // GTEST_OS_LINUX
+
+#if GTEST_HAS_EXCEPTIONS
+# include <stdexcept>
+#endif
+
+#if GTEST_CAN_STREAM_RESULTS_
+# include <arpa/inet.h>  // NOLINT
+# include <netdb.h>  // NOLINT
+# include <sys/socket.h>  // NOLINT
+# include <sys/types.h>  // NOLINT
+#endif
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// Utility functions and classes used by the Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// This file contains purely Google Test's internal implementation.  Please
+// DO NOT #INCLUDE IT IN A USER PROGRAM.
+
+#ifndef GTEST_SRC_GTEST_INTERNAL_INL_H_
+#define GTEST_SRC_GTEST_INTERNAL_INL_H_
+
+// GTEST_IMPLEMENTATION_ is defined to 1 iff the current translation unit is
+// part of Google Test's implementation; otherwise it's undefined.
+#if !GTEST_IMPLEMENTATION_
+// If this file is included from the user's code, just say no.
+# error "gtest-internal-inl.h is part of Google Test's internal implementation."
+# error "It must not be included except by Google Test itself."
+#endif  // GTEST_IMPLEMENTATION_
+
+#ifndef _WIN32_WCE
+# include <errno.h>
+#endif  // !_WIN32_WCE
+#include <stddef.h>
+#include <stdlib.h>  // For strtoll/_strtoul64/malloc/free.
+#include <string.h>  // For memmove.
+
+#include <algorithm>
+#include <string>
+#include <vector>
+
+
+#if GTEST_CAN_STREAM_RESULTS_
+# include <arpa/inet.h>  // NOLINT
+# include <netdb.h>  // NOLINT
+#endif
+
+#if GTEST_OS_WINDOWS
+# include <windows.h>  // NOLINT
+#endif  // GTEST_OS_WINDOWS
+
+
+namespace testing {
+
+// Declares the flags.
+//
+// We don't want the users to modify this flag in the code, but want
+// Google Test's own unit tests to be able to access it. Therefore we
+// declare it here as opposed to in gtest.h.
+GTEST_DECLARE_bool_(death_test_use_fork);
+
+namespace internal {
+
+// The value of GetTestTypeId() as seen from within the Google Test
+// library.  This is solely for testing GetTestTypeId().
+GTEST_API_ extern const TypeId kTestTypeIdInGoogleTest;
+
+// Names of the flags (needed for parsing Google Test flags).
+const char kAlsoRunDisabledTestsFlag[] = "also_run_disabled_tests";
+const char kBreakOnFailureFlag[] = "break_on_failure";
+const char kCatchExceptionsFlag[] = "catch_exceptions";
+const char kColorFlag[] = "color";
+const char kFilterFlag[] = "filter";
+const char kListTestsFlag[] = "list_tests";
+const char kOutputFlag[] = "output";
+const char kPrintTimeFlag[] = "print_time";
+const char kRandomSeedFlag[] = "random_seed";
+const char kRepeatFlag[] = "repeat";
+const char kShuffleFlag[] = "shuffle";
+const char kStackTraceDepthFlag[] = "stack_trace_depth";
+const char kStreamResultToFlag[] = "stream_result_to";
+const char kThrowOnFailureFlag[] = "throw_on_failure";
+const char kFlagfileFlag[] = "flagfile";
+
+// A valid random seed must be in [1, kMaxRandomSeed].
+const int kMaxRandomSeed = 99999;
+
+// g_help_flag is true iff the --help flag or an equivalent form is
+// specified on the command line.
+GTEST_API_ extern bool g_help_flag;
+
+// Returns the current time in milliseconds.
+GTEST_API_ TimeInMillis GetTimeInMillis();
+
+// Returns true iff Google Test should use colors in the output.
+GTEST_API_ bool ShouldUseColor(bool stdout_is_tty);
+
+// Formats the given time in milliseconds as seconds.
+GTEST_API_ std::string FormatTimeInMillisAsSeconds(TimeInMillis ms);
+
+// Converts the given time in milliseconds to a date string in the ISO 8601
+// format, without the timezone information.  N.B.: due to the use the
+// non-reentrant localtime() function, this function is not thread safe.  Do
+// not use it in any code that can be called from multiple threads.
+GTEST_API_ std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms);
+
+// Parses a string for an Int32 flag, in the form of "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+GTEST_API_ bool ParseInt32Flag(
+    const char* str, const char* flag, Int32* value);
+
+// Returns a random seed in range [1, kMaxRandomSeed] based on the
+// given --gtest_random_seed flag value.
+inline int GetRandomSeedFromFlag(Int32 random_seed_flag) {
+  const unsigned int raw_seed = (random_seed_flag == 0) ?
+      static_cast<unsigned int>(GetTimeInMillis()) :
+      static_cast<unsigned int>(random_seed_flag);
+
+  // Normalizes the actual seed to range [1, kMaxRandomSeed] such that
+  // it's easy to type.
+  const int normalized_seed =
+      static_cast<int>((raw_seed - 1U) %
+                       static_cast<unsigned int>(kMaxRandomSeed)) + 1;
+  return normalized_seed;
+}
+
+// Returns the first valid random seed after 'seed'.  The behavior is
+// undefined if 'seed' is invalid.  The seed after kMaxRandomSeed is
+// considered to be 1.
+inline int GetNextRandomSeed(int seed) {
+  GTEST_CHECK_(1 <= seed && seed <= kMaxRandomSeed)
+      << "Invalid random seed " << seed << " - must be in [1, "
+      << kMaxRandomSeed << "].";
+  const int next_seed = seed + 1;
+  return (next_seed > kMaxRandomSeed) ? 1 : next_seed;
+}
+
+// This class saves the values of all Google Test flags in its c'tor, and
+// restores them in its d'tor.
+class GTestFlagSaver {
+ public:
+  // The c'tor.
+  GTestFlagSaver() {
+    also_run_disabled_tests_ = GTEST_FLAG(also_run_disabled_tests);
+    break_on_failure_ = GTEST_FLAG(break_on_failure);
+    catch_exceptions_ = GTEST_FLAG(catch_exceptions);
+    color_ = GTEST_FLAG(color);
+    death_test_style_ = GTEST_FLAG(death_test_style);
+    death_test_use_fork_ = GTEST_FLAG(death_test_use_fork);
+    filter_ = GTEST_FLAG(filter);
+    internal_run_death_test_ = GTEST_FLAG(internal_run_death_test);
+    list_tests_ = GTEST_FLAG(list_tests);
+    output_ = GTEST_FLAG(output);
+    print_time_ = GTEST_FLAG(print_time);
+    random_seed_ = GTEST_FLAG(random_seed);
+    repeat_ = GTEST_FLAG(repeat);
+    shuffle_ = GTEST_FLAG(shuffle);
+    stack_trace_depth_ = GTEST_FLAG(stack_trace_depth);
+    stream_result_to_ = GTEST_FLAG(stream_result_to);
+    throw_on_failure_ = GTEST_FLAG(throw_on_failure);
+  }
+
+  // The d'tor is not virtual.  DO NOT INHERIT FROM THIS CLASS.
+  ~GTestFlagSaver() {
+    GTEST_FLAG(also_run_disabled_tests) = also_run_disabled_tests_;
+    GTEST_FLAG(break_on_failure) = break_on_failure_;
+    GTEST_FLAG(catch_exceptions) = catch_exceptions_;
+    GTEST_FLAG(color) = color_;
+    GTEST_FLAG(death_test_style) = death_test_style_;
+    GTEST_FLAG(death_test_use_fork) = death_test_use_fork_;
+    GTEST_FLAG(filter) = filter_;
+    GTEST_FLAG(internal_run_death_test) = internal_run_death_test_;
+    GTEST_FLAG(list_tests) = list_tests_;
+    GTEST_FLAG(output) = output_;
+    GTEST_FLAG(print_time) = print_time_;
+    GTEST_FLAG(random_seed) = random_seed_;
+    GTEST_FLAG(repeat) = repeat_;
+    GTEST_FLAG(shuffle) = shuffle_;
+    GTEST_FLAG(stack_trace_depth) = stack_trace_depth_;
+    GTEST_FLAG(stream_result_to) = stream_result_to_;
+    GTEST_FLAG(throw_on_failure) = throw_on_failure_;
+  }
+
+ private:
+  // Fields for saving the original values of flags.
+  bool also_run_disabled_tests_;
+  bool break_on_failure_;
+  bool catch_exceptions_;
+  std::string color_;
+  std::string death_test_style_;
+  bool death_test_use_fork_;
+  std::string filter_;
+  std::string internal_run_death_test_;
+  bool list_tests_;
+  std::string output_;
+  bool print_time_;
+  internal::Int32 random_seed_;
+  internal::Int32 repeat_;
+  bool shuffle_;
+  internal::Int32 stack_trace_depth_;
+  std::string stream_result_to_;
+  bool throw_on_failure_;
+} GTEST_ATTRIBUTE_UNUSED_;
+
+// Converts a Unicode code point to a narrow string in UTF-8 encoding.
+// code_point parameter is of type UInt32 because wchar_t may not be
+// wide enough to contain a code point.
+// If the code_point is not a valid Unicode code point
+// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted
+// to "(Invalid Unicode 0xXXXXXXXX)".
+GTEST_API_ std::string CodePointToUtf8(UInt32 code_point);
+
+// Converts a wide string to a narrow string in UTF-8 encoding.
+// The wide string is assumed to have the following encoding:
+//   UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS)
+//   UTF-32 if sizeof(wchar_t) == 4 (on Linux)
+// Parameter str points to a null-terminated wide string.
+// Parameter num_chars may additionally limit the number
+// of wchar_t characters processed. -1 is used when the entire string
+// should be processed.
+// If the string contains code points that are not valid Unicode code points
+// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding
+// and contains invalid UTF-16 surrogate pairs, values in those pairs
+// will be encoded as individual Unicode characters from Basic Normal Plane.
+GTEST_API_ std::string WideStringToUtf8(const wchar_t* str, int num_chars);
+
+// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file
+// if the variable is present. If a file already exists at this location, this
+// function will write over it. If the variable is present, but the file cannot
+// be created, prints an error and exits.
+void WriteToShardStatusFileIfNeeded();
+
+// Checks whether sharding is enabled by examining the relevant
+// environment variable values. If the variables are present,
+// but inconsistent (e.g., shard_index >= total_shards), prints
+// an error and exits. If in_subprocess_for_death_test, sharding is
+// disabled because it must only be applied to the original test
+// process. Otherwise, we could filter out death tests we intended to execute.
+GTEST_API_ bool ShouldShard(const char* total_shards_str,
+                            const char* shard_index_str,
+                            bool in_subprocess_for_death_test);
+
+// Parses the environment variable var as an Int32. If it is unset,
+// returns default_val. If it is not an Int32, prints an error and
+// and aborts.
+GTEST_API_ Int32 Int32FromEnvOrDie(const char* env_var, Int32 default_val);
+
+// Given the total number of shards, the shard index, and the test id,
+// returns true iff the test should be run on this shard. The test id is
+// some arbitrary but unique non-negative integer assigned to each test
+// method. Assumes that 0 <= shard_index < total_shards.
+GTEST_API_ bool ShouldRunTestOnShard(
+    int total_shards, int shard_index, int test_id);
+
+// STL container utilities.
+
+// Returns the number of elements in the given container that satisfy
+// the given predicate.
+template <class Container, typename Predicate>
+inline int CountIf(const Container& c, Predicate predicate) {
+  // Implemented as an explicit loop since std::count_if() in libCstd on
+  // Solaris has a non-standard signature.
+  int count = 0;
+  for (typename Container::const_iterator it = c.begin(); it != c.end(); ++it) {
+    if (predicate(*it))
+      ++count;
+  }
+  return count;
+}
+
+// Applies a function/functor to each element in the container.
+template <class Container, typename Functor>
+void ForEach(const Container& c, Functor functor) {
+  std::for_each(c.begin(), c.end(), functor);
+}
+
+// Returns the i-th element of the vector, or default_value if i is not
+// in range [0, v.size()).
+template <typename E>
+inline E GetElementOr(const std::vector<E>& v, int i, E default_value) {
+  return (i < 0 || i >= static_cast<int>(v.size())) ? default_value : v[i];
+}
+
+// Performs an in-place shuffle of a range of the vector's elements.
+// 'begin' and 'end' are element indices as an STL-style range;
+// i.e. [begin, end) are shuffled, where 'end' == size() means to
+// shuffle to the end of the vector.
+template <typename E>
+void ShuffleRange(internal::Random* random, int begin, int end,
+                  std::vector<E>* v) {
+  const int size = static_cast<int>(v->size());
+  GTEST_CHECK_(0 <= begin && begin <= size)
+      << "Invalid shuffle range start " << begin << ": must be in range [0, "
+      << size << "].";
+  GTEST_CHECK_(begin <= end && end <= size)
+      << "Invalid shuffle range finish " << end << ": must be in range ["
+      << begin << ", " << size << "].";
+
+  // Fisher-Yates shuffle, from
+  // http://en.wikipedia.org/wiki/Fisher-Yates_shuffle
+  for (int range_width = end - begin; range_width >= 2; range_width--) {
+    const int last_in_range = begin + range_width - 1;
+    const int selected = begin + random->Generate(range_width);
+    std::swap((*v)[selected], (*v)[last_in_range]);
+  }
+}
+
+// Performs an in-place shuffle of the vector's elements.
+template <typename E>
+inline void Shuffle(internal::Random* random, std::vector<E>* v) {
+  ShuffleRange(random, 0, static_cast<int>(v->size()), v);
+}
+
+// A function for deleting an object.  Handy for being used as a
+// functor.
+template <typename T>
+static void Delete(T* x) {
+  delete x;
+}
+
+// A predicate that checks the key of a TestProperty against a known key.
+//
+// TestPropertyKeyIs is copyable.
+class TestPropertyKeyIs {
+ public:
+  // Constructor.
+  //
+  // TestPropertyKeyIs has NO default constructor.
+  explicit TestPropertyKeyIs(const std::string& key) : key_(key) {}
+
+  // Returns true iff the test name of test property matches on key_.
+  bool operator()(const TestProperty& test_property) const {
+    return test_property.key() == key_;
+  }
+
+ private:
+  std::string key_;
+};
+
+// Class UnitTestOptions.
+//
+// This class contains functions for processing options the user
+// specifies when running the tests.  It has only static members.
+//
+// In most cases, the user can specify an option using either an
+// environment variable or a command line flag.  E.g. you can set the
+// test filter using either GTEST_FILTER or --gtest_filter.  If both
+// the variable and the flag are present, the latter overrides the
+// former.
+class GTEST_API_ UnitTestOptions {
+ public:
+  // Functions for processing the gtest_output flag.
+
+  // Returns the output format, or "" for normal printed output.
+  static std::string GetOutputFormat();
+
+  // Returns the absolute path of the requested output file, or the
+  // default (test_detail.xml in the original working directory) if
+  // none was explicitly specified.
+  static std::string GetAbsolutePathToOutputFile();
+
+  // Functions for processing the gtest_filter flag.
+
+  // Returns true iff the wildcard pattern matches the string.  The
+  // first ':' or '\0' character in pattern marks the end of it.
+  //
+  // This recursive algorithm isn't very efficient, but is clear and
+  // works well enough for matching test names, which are short.
+  static bool PatternMatchesString(const char *pattern, const char *str);
+
+  // Returns true iff the user-specified filter matches the test case
+  // name and the test name.
+  static bool FilterMatchesTest(const std::string &test_case_name,
+                                const std::string &test_name);
+
+#if GTEST_OS_WINDOWS
+  // Function for supporting the gtest_catch_exception flag.
+
+  // Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the
+  // given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise.
+  // This function is useful as an __except condition.
+  static int GTestShouldProcessSEH(DWORD exception_code);
+#endif  // GTEST_OS_WINDOWS
+
+  // Returns true if "name" matches the ':' separated list of glob-style
+  // filters in "filter".
+  static bool MatchesFilter(const std::string& name, const char* filter);
+};
+
+// Returns the current application's name, removing directory path if that
+// is present.  Used by UnitTestOptions::GetOutputFile.
+GTEST_API_ FilePath GetCurrentExecutableName();
+
+// The role interface for getting the OS stack trace as a string.
+class OsStackTraceGetterInterface {
+ public:
+  OsStackTraceGetterInterface() {}
+  virtual ~OsStackTraceGetterInterface() {}
+
+  // Returns the current OS stack trace as an std::string.  Parameters:
+  //
+  //   max_depth  - the maximum number of stack frames to be included
+  //                in the trace.
+  //   skip_count - the number of top frames to be skipped; doesn't count
+  //                against max_depth.
+  virtual std::string CurrentStackTrace(int max_depth, int skip_count) = 0;
+
+  // UponLeavingGTest() should be called immediately before Google Test calls
+  // user code. It saves some information about the current stack that
+  // CurrentStackTrace() will use to find and hide Google Test stack frames.
+  virtual void UponLeavingGTest() = 0;
+
+  // This string is inserted in place of stack frames that are part of
+  // Google Test's implementation.
+  static const char* const kElidedFramesMarker;
+
+ private:
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetterInterface);
+};
+
+// A working implementation of the OsStackTraceGetterInterface interface.
+class OsStackTraceGetter : public OsStackTraceGetterInterface {
+ public:
+  OsStackTraceGetter() {}
+
+  virtual std::string CurrentStackTrace(int max_depth, int skip_count);
+  virtual void UponLeavingGTest();
+
+ private:
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetter);
+};
+
+// Information about a Google Test trace point.
+struct TraceInfo {
+  const char* file;
+  int line;
+  std::string message;
+};
+
+// This is the default global test part result reporter used in UnitTestImpl.
+// This class should only be used by UnitTestImpl.
+class DefaultGlobalTestPartResultReporter
+  : public TestPartResultReporterInterface {
+ public:
+  explicit DefaultGlobalTestPartResultReporter(UnitTestImpl* unit_test);
+  // Implements the TestPartResultReporterInterface. Reports the test part
+  // result in the current test.
+  virtual void ReportTestPartResult(const TestPartResult& result);
+
+ private:
+  UnitTestImpl* const unit_test_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultGlobalTestPartResultReporter);
+};
+
+// This is the default per thread test part result reporter used in
+// UnitTestImpl. This class should only be used by UnitTestImpl.
+class DefaultPerThreadTestPartResultReporter
+    : public TestPartResultReporterInterface {
+ public:
+  explicit DefaultPerThreadTestPartResultReporter(UnitTestImpl* unit_test);
+  // Implements the TestPartResultReporterInterface. The implementation just
+  // delegates to the current global test part result reporter of *unit_test_.
+  virtual void ReportTestPartResult(const TestPartResult& result);
+
+ private:
+  UnitTestImpl* const unit_test_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultPerThreadTestPartResultReporter);
+};
+
+// The private implementation of the UnitTest class.  We don't protect
+// the methods under a mutex, as this class is not accessible by a
+// user and the UnitTest class that delegates work to this class does
+// proper locking.
+class GTEST_API_ UnitTestImpl {
+ public:
+  explicit UnitTestImpl(UnitTest* parent);
+  virtual ~UnitTestImpl();
+
+  // There are two different ways to register your own TestPartResultReporter.
+  // You can register your own repoter to listen either only for test results
+  // from the current thread or for results from all threads.
+  // By default, each per-thread test result repoter just passes a new
+  // TestPartResult to the global test result reporter, which registers the
+  // test part result for the currently running test.
+
+  // Returns the global test part result reporter.
+  TestPartResultReporterInterface* GetGlobalTestPartResultReporter();
+
+  // Sets the global test part result reporter.
+  void SetGlobalTestPartResultReporter(
+      TestPartResultReporterInterface* reporter);
+
+  // Returns the test part result reporter for the current thread.
+  TestPartResultReporterInterface* GetTestPartResultReporterForCurrentThread();
+
+  // Sets the test part result reporter for the current thread.
+  void SetTestPartResultReporterForCurrentThread(
+      TestPartResultReporterInterface* reporter);
+
+  // Gets the number of successful test cases.
+  int successful_test_case_count() const;
+
+  // Gets the number of failed test cases.
+  int failed_test_case_count() const;
+
+  // Gets the number of all test cases.
+  int total_test_case_count() const;
+
+  // Gets the number of all test cases that contain at least one test
+  // that should run.
+  int test_case_to_run_count() const;
+
+  // Gets the number of successful tests.
+  int successful_test_count() const;
+
+  // Gets the number of skipped tests.
+  int skipped_test_count() const;
+
+  // Gets the number of failed tests.
+  int failed_test_count() const;
+
+  // Gets the number of disabled tests that will be reported in the XML report.
+  int reportable_disabled_test_count() const;
+
+  // Gets the number of disabled tests.
+  int disabled_test_count() const;
+
+  // Gets the number of tests to be printed in the XML report.
+  int reportable_test_count() const;
+
+  // Gets the number of all tests.
+  int total_test_count() const;
+
+  // Gets the number of tests that should run.
+  int test_to_run_count() const;
+
+  // Gets the time of the test program start, in ms from the start of the
+  // UNIX epoch.
+  TimeInMillis start_timestamp() const { return start_timestamp_; }
+
+  // Gets the elapsed time, in milliseconds.
+  TimeInMillis elapsed_time() const { return elapsed_time_; }
+
+  // Returns true iff the unit test passed (i.e. all test cases passed).
+  bool Passed() const { return !Failed(); }
+
+  // Returns true iff the unit test failed (i.e. some test case failed
+  // or something outside of all tests failed).
+  bool Failed() const {
+    return failed_test_case_count() > 0 || ad_hoc_test_result()->Failed();
+  }
+
+  // Gets the i-th test case among all the test cases. i can range from 0 to
+  // total_test_case_count() - 1. If i is not in that range, returns NULL.
+  const TestCase* GetTestCase(int i) const {
+    const int index = GetElementOr(test_case_indices_, i, -1);
+    return index < 0 ? NULL : test_cases_[i];
+  }
+
+  // Gets the i-th test case among all the test cases. i can range from 0 to
+  // total_test_case_count() - 1. If i is not in that range, returns NULL.
+  TestCase* GetMutableTestCase(int i) {
+    const int index = GetElementOr(test_case_indices_, i, -1);
+    return index < 0 ? NULL : test_cases_[index];
+  }
+
+  // Provides access to the event listener list.
+  TestEventListeners* listeners() { return &listeners_; }
+
+  // Returns the TestResult for the test that's currently running, or
+  // the TestResult for the ad hoc test if no test is running.
+  TestResult* current_test_result();
+
+  // Returns the TestResult for the ad hoc test.
+  const TestResult* ad_hoc_test_result() const { return &ad_hoc_test_result_; }
+
+  // Sets the OS stack trace getter.
+  //
+  // Does nothing if the input and the current OS stack trace getter
+  // are the same; otherwise, deletes the old getter and makes the
+  // input the current getter.
+  void set_os_stack_trace_getter(OsStackTraceGetterInterface* getter);
+
+  // Returns the current OS stack trace getter if it is not NULL;
+  // otherwise, creates an OsStackTraceGetter, makes it the current
+  // getter, and returns it.
+  OsStackTraceGetterInterface* os_stack_trace_getter();
+
+  // Returns the current OS stack trace as an std::string.
+  //
+  // The maximum number of stack frames to be included is specified by
+  // the gtest_stack_trace_depth flag.  The skip_count parameter
+  // specifies the number of top frames to be skipped, which doesn't
+  // count against the number of frames to be included.
+  //
+  // For example, if Foo() calls Bar(), which in turn calls
+  // CurrentOsStackTraceExceptTop(1), Foo() will be included in the
+  // trace but Bar() and CurrentOsStackTraceExceptTop() won't.
+  std::string CurrentOsStackTraceExceptTop(int skip_count) GTEST_NO_INLINE_;
+
+  // Finds and returns a TestCase with the given name.  If one doesn't
+  // exist, creates one and returns it.
+  //
+  // Arguments:
+  //
+  //   test_case_name: name of the test case
+  //   type_param:     the name of the test's type parameter, or NULL if
+  //                   this is not a typed or a type-parameterized test.
+  //   set_up_tc:      pointer to the function that sets up the test case
+  //   tear_down_tc:   pointer to the function that tears down the test case
+  TestCase* GetTestCase(const char* test_case_name,
+                        const char* type_param,
+                        Test::SetUpTestCaseFunc set_up_tc,
+                        Test::TearDownTestCaseFunc tear_down_tc);
+
+  // Adds a TestInfo to the unit test.
+  //
+  // Arguments:
+  //
+  //   set_up_tc:    pointer to the function that sets up the test case
+  //   tear_down_tc: pointer to the function that tears down the test case
+  //   test_info:    the TestInfo object
+  void AddTestInfo(Test::SetUpTestCaseFunc set_up_tc,
+                   Test::TearDownTestCaseFunc tear_down_tc,
+                   TestInfo* test_info) {
+    // In order to support thread-safe death tests, we need to
+    // remember the original working directory when the test program
+    // was first invoked.  We cannot do this in RUN_ALL_TESTS(), as
+    // the user may have changed the current directory before calling
+    // RUN_ALL_TESTS().  Therefore we capture the current directory in
+    // AddTestInfo(), which is called to register a TEST or TEST_F
+    // before main() is reached.
+    if (original_working_dir_.IsEmpty()) {
+      original_working_dir_.Set(FilePath::GetCurrentDir());
+      GTEST_CHECK_(!original_working_dir_.IsEmpty())
+          << "Failed to get the current working directory.";
+    }
+
+    GetTestCase(test_info->test_case_name(),
+                test_info->type_param(),
+                set_up_tc,
+                tear_down_tc)->AddTestInfo(test_info);
+  }
+
+#if GTEST_HAS_PARAM_TEST
+  // Returns ParameterizedTestCaseRegistry object used to keep track of
+  // value-parameterized tests and instantiate and register them.
+  internal::ParameterizedTestCaseRegistry& parameterized_test_registry() {
+    return parameterized_test_registry_;
+  }
+#endif  // GTEST_HAS_PARAM_TEST
+
+  // Sets the TestCase object for the test that's currently running.
+  void set_current_test_case(TestCase* a_current_test_case) {
+    current_test_case_ = a_current_test_case;
+  }
+
+  // Sets the TestInfo object for the test that's currently running.  If
+  // current_test_info is NULL, the assertion results will be stored in
+  // ad_hoc_test_result_.
+  void set_current_test_info(TestInfo* a_current_test_info) {
+    current_test_info_ = a_current_test_info;
+  }
+
+  // Registers all parameterized tests defined using TEST_P and
+  // INSTANTIATE_TEST_CASE_P, creating regular tests for each test/parameter
+  // combination. This method can be called more then once; it has guards
+  // protecting from registering the tests more then once.  If
+  // value-parameterized tests are disabled, RegisterParameterizedTests is
+  // present but does nothing.
+  void RegisterParameterizedTests();
+
+  // Runs all tests in this UnitTest object, prints the result, and
+  // returns true if all tests are successful.  If any exception is
+  // thrown during a test, this test is considered to be failed, but
+  // the rest of the tests will still be run.
+  bool RunAllTests();
+
+  // Clears the results of all tests, except the ad hoc tests.
+  void ClearNonAdHocTestResult() {
+    ForEach(test_cases_, TestCase::ClearTestCaseResult);
+  }
+
+  // Clears the results of ad-hoc test assertions.
+  void ClearAdHocTestResult() {
+    ad_hoc_test_result_.Clear();
+  }
+
+  // Adds a TestProperty to the current TestResult object when invoked in a
+  // context of a test or a test case, or to the global property set. If the
+  // result already contains a property with the same key, the value will be
+  // updated.
+  void RecordProperty(const TestProperty& test_property);
+
+  enum ReactionToSharding {
+    HONOR_SHARDING_PROTOCOL,
+    IGNORE_SHARDING_PROTOCOL
+  };
+
+  // Matches the full name of each test against the user-specified
+  // filter to decide whether the test should run, then records the
+  // result in each TestCase and TestInfo object.
+  // If shard_tests == HONOR_SHARDING_PROTOCOL, further filters tests
+  // based on sharding variables in the environment.
+  // Returns the number of tests that should run.
+  int FilterTests(ReactionToSharding shard_tests);
+
+  // Prints the names of the tests matching the user-specified filter flag.
+  void ListTestsMatchingFilter();
+
+  const TestCase* current_test_case() const { return current_test_case_; }
+  TestInfo* current_test_info() { return current_test_info_; }
+  const TestInfo* current_test_info() const { return current_test_info_; }
+
+  // Returns the vector of environments that need to be set-up/torn-down
+  // before/after the tests are run.
+  std::vector<Environment*>& environments() { return environments_; }
+
+  // Getters for the per-thread Google Test trace stack.
+  std::vector<TraceInfo>& gtest_trace_stack() {
+    return *(gtest_trace_stack_.pointer());
+  }
+  const std::vector<TraceInfo>& gtest_trace_stack() const {
+    return gtest_trace_stack_.get();
+  }
+
+#if GTEST_HAS_DEATH_TEST
+  void InitDeathTestSubprocessControlInfo() {
+    internal_run_death_test_flag_.reset(ParseInternalRunDeathTestFlag());
+  }
+  // Returns a pointer to the parsed --gtest_internal_run_death_test
+  // flag, or NULL if that flag was not specified.
+  // This information is useful only in a death test child process.
+  // Must not be called before a call to InitGoogleTest.
+  const InternalRunDeathTestFlag* internal_run_death_test_flag() const {
+    return internal_run_death_test_flag_.get();
+  }
+
+  // Returns a pointer to the current death test factory.
+  internal::DeathTestFactory* death_test_factory() {
+    return death_test_factory_.get();
+  }
+
+  void SuppressTestEventsIfInSubprocess();
+
+  friend class ReplaceDeathTestFactory;
+#endif  // GTEST_HAS_DEATH_TEST
+
+  // Initializes the event listener performing XML output as specified by
+  // UnitTestOptions. Must not be called before InitGoogleTest.
+  void ConfigureXmlOutput();
+
+#if GTEST_CAN_STREAM_RESULTS_
+  // Initializes the event listener for streaming test results to a socket.
+  // Must not be called before InitGoogleTest.
+  void ConfigureStreamingOutput();
+#endif
+
+  // Performs initialization dependent upon flag values obtained in
+  // ParseGoogleTestFlagsOnly.  Is called from InitGoogleTest after the call to
+  // ParseGoogleTestFlagsOnly.  In case a user neglects to call InitGoogleTest
+  // this function is also called from RunAllTests.  Since this function can be
+  // called more than once, it has to be idempotent.
+  void PostFlagParsingInit();
+
+  // Gets the random seed used at the start of the current test iteration.
+  int random_seed() const { return random_seed_; }
+
+  // Gets the random number generator.
+  internal::Random* random() { return &random_; }
+
+  // Shuffles all test cases, and the tests within each test case,
+  // making sure that death tests are still run first.
+  void ShuffleTests();
+
+  // Restores the test cases and tests to their order before the first shuffle.
+  void UnshuffleTests();
+
+  // Returns the value of GTEST_FLAG(catch_exceptions) at the moment
+  // UnitTest::Run() starts.
+  bool catch_exceptions() const { return catch_exceptions_; }
+
+ private:
+  friend class ::testing::UnitTest;
+
+  // Used by UnitTest::Run() to capture the state of
+  // GTEST_FLAG(catch_exceptions) at the moment it starts.
+  void set_catch_exceptions(bool value) { catch_exceptions_ = value; }
+
+  // The UnitTest object that owns this implementation object.
+  UnitTest* const parent_;
+
+  // The working directory when the first TEST() or TEST_F() was
+  // executed.
+  internal::FilePath original_working_dir_;
+
+  // The default test part result reporters.
+  DefaultGlobalTestPartResultReporter default_global_test_part_result_reporter_;
+  DefaultPerThreadTestPartResultReporter
+      default_per_thread_test_part_result_reporter_;
+
+  // Points to (but doesn't own) the global test part result reporter.
+  TestPartResultReporterInterface* global_test_part_result_repoter_;
+
+  // Protects read and write access to global_test_part_result_reporter_.
+  internal::Mutex global_test_part_result_reporter_mutex_;
+
+  // Points to (but doesn't own) the per-thread test part result reporter.
+  internal::ThreadLocal<TestPartResultReporterInterface*>
+      per_thread_test_part_result_reporter_;
+
+  // The vector of environments that need to be set-up/torn-down
+  // before/after the tests are run.
+  std::vector<Environment*> environments_;
+
+  // The vector of TestCases in their original order.  It owns the
+  // elements in the vector.
+  std::vector<TestCase*> test_cases_;
+
+  // Provides a level of indirection for the test case list to allow
+  // easy shuffling and restoring the test case order.  The i-th
+  // element of this vector is the index of the i-th test case in the
+  // shuffled order.
+  std::vector<int> test_case_indices_;
+
+#if GTEST_HAS_PARAM_TEST
+  // ParameterizedTestRegistry object used to register value-parameterized
+  // tests.
+  internal::ParameterizedTestCaseRegistry parameterized_test_registry_;
+
+  // Indicates whether RegisterParameterizedTests() has been called already.
+  bool parameterized_tests_registered_;
+#endif  // GTEST_HAS_PARAM_TEST
+
+  // Index of the last death test case registered.  Initially -1.
+  int last_death_test_case_;
+
+  // This points to the TestCase for the currently running test.  It
+  // changes as Google Test goes through one test case after another.
+  // When no test is running, this is set to NULL and Google Test
+  // stores assertion results in ad_hoc_test_result_.  Initially NULL.
+  TestCase* current_test_case_;
+
+  // This points to the TestInfo for the currently running test.  It
+  // changes as Google Test goes through one test after another.  When
+  // no test is running, this is set to NULL and Google Test stores
+  // assertion results in ad_hoc_test_result_.  Initially NULL.
+  TestInfo* current_test_info_;
+
+  // Normally, a user only writes assertions inside a TEST or TEST_F,
+  // or inside a function called by a TEST or TEST_F.  Since Google
+  // Test keeps track of which test is current running, it can
+  // associate such an assertion with the test it belongs to.
+  //
+  // If an assertion is encountered when no TEST or TEST_F is running,
+  // Google Test attributes the assertion result to an imaginary "ad hoc"
+  // test, and records the result in ad_hoc_test_result_.
+  TestResult ad_hoc_test_result_;
+
+  // The list of event listeners that can be used to track events inside
+  // Google Test.
+  TestEventListeners listeners_;
+
+  // The OS stack trace getter.  Will be deleted when the UnitTest
+  // object is destructed.  By default, an OsStackTraceGetter is used,
+  // but the user can set this field to use a custom getter if that is
+  // desired.
+  OsStackTraceGetterInterface* os_stack_trace_getter_;
+
+  // True iff PostFlagParsingInit() has been called.
+  bool post_flag_parse_init_performed_;
+
+  // The random number seed used at the beginning of the test run.
+  int random_seed_;
+
+  // Our random number generator.
+  internal::Random random_;
+
+  // The time of the test program start, in ms from the start of the
+  // UNIX epoch.
+  TimeInMillis start_timestamp_;
+
+  // How long the test took to run, in milliseconds.
+  TimeInMillis elapsed_time_;
+
+#if GTEST_HAS_DEATH_TEST
+  // The decomposed components of the gtest_internal_run_death_test flag,
+  // parsed when RUN_ALL_TESTS is called.
+  internal::scoped_ptr<InternalRunDeathTestFlag> internal_run_death_test_flag_;
+  internal::scoped_ptr<internal::DeathTestFactory> death_test_factory_;
+#endif  // GTEST_HAS_DEATH_TEST
+
+  // A per-thread stack of traces created by the SCOPED_TRACE() macro.
+  internal::ThreadLocal<std::vector<TraceInfo> > gtest_trace_stack_;
+
+  // The value of GTEST_FLAG(catch_exceptions) at the moment RunAllTests()
+  // starts.
+  bool catch_exceptions_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTestImpl);
+};  // class UnitTestImpl
+
+// Convenience function for accessing the global UnitTest
+// implementation object.
+inline UnitTestImpl* GetUnitTestImpl() {
+  return UnitTest::GetInstance()->impl();
+}
+
+#if GTEST_USES_SIMPLE_RE
+
+// Internal helper functions for implementing the simple regular
+// expression matcher.
+GTEST_API_ bool IsInSet(char ch, const char* str);
+GTEST_API_ bool IsAsciiDigit(char ch);
+GTEST_API_ bool IsAsciiPunct(char ch);
+GTEST_API_ bool IsRepeat(char ch);
+GTEST_API_ bool IsAsciiWhiteSpace(char ch);
+GTEST_API_ bool IsAsciiWordChar(char ch);
+GTEST_API_ bool IsValidEscape(char ch);
+GTEST_API_ bool AtomMatchesChar(bool escaped, char pattern, char ch);
+GTEST_API_ bool ValidateRegex(const char* regex);
+GTEST_API_ bool MatchRegexAtHead(const char* regex, const char* str);
+GTEST_API_ bool MatchRepetitionAndRegexAtHead(
+    bool escaped, char ch, char repeat, const char* regex, const char* str);
+GTEST_API_ bool MatchRegexAnywhere(const char* regex, const char* str);
+
+#endif  // GTEST_USES_SIMPLE_RE
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.
+GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, char** argv);
+GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv);
+
+#if GTEST_HAS_DEATH_TEST
+
+// Returns the message describing the last system error, regardless of the
+// platform.
+GTEST_API_ std::string GetLastErrnoDescription();
+
+// Attempts to parse a string into a positive integer pointed to by the
+// number parameter.  Returns true if that is possible.
+// GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we can use
+// it here.
+template <typename Integer>
+bool ParseNaturalNumber(const ::std::string& str, Integer* number) {
+  // Fail fast if the given string does not begin with a digit;
+  // this bypasses strtoXXX's "optional leading whitespace and plus
+  // or minus sign" semantics, which are undesirable here.
+  if (str.empty() || !IsDigit(str[0])) {
+    return false;
+  }
+  errno = 0;
+
+  char* end;
+  // BiggestConvertible is the largest integer type that system-provided
+  // string-to-number conversion routines can return.
+
+# if GTEST_OS_WINDOWS && !defined(__GNUC__)
+
+  // MSVC and C++ Builder define __int64 instead of the standard long long.
+  typedef unsigned __int64 BiggestConvertible;
+  const BiggestConvertible parsed = _strtoui64(str.c_str(), &end, 10);
+
+# else
+
+  typedef unsigned long long BiggestConvertible;  // NOLINT
+  const BiggestConvertible parsed = strtoull(str.c_str(), &end, 10);
+
+# endif  // GTEST_OS_WINDOWS && !defined(__GNUC__)
+
+  const bool parse_success = *end == '\0' && errno == 0;
+
+  // TODO(vladl@google.com): Convert this to compile time assertion when it is
+  // available.
+  GTEST_CHECK_(sizeof(Integer) <= sizeof(parsed));
+
+  const Integer result = static_cast<Integer>(parsed);
+  if (parse_success && static_cast<BiggestConvertible>(result) == parsed) {
+    *number = result;
+    return true;
+  }
+  return false;
+}
+#endif  // GTEST_HAS_DEATH_TEST
+
+// TestResult contains some private methods that should be hidden from
+// Google Test user but are required for testing. This class allow our tests
+// to access them.
+//
+// This class is supplied only for the purpose of testing Google Test's own
+// constructs. Do not use it in user tests, either directly or indirectly.
+class TestResultAccessor {
+ public:
+  static void RecordProperty(TestResult* test_result,
+                             const std::string& xml_element,
+                             const TestProperty& property) {
+    test_result->RecordProperty(xml_element, property);
+  }
+
+  static void ClearTestPartResults(TestResult* test_result) {
+    test_result->ClearTestPartResults();
+  }
+
+  static const std::vector<testing::TestPartResult>& test_part_results(
+      const TestResult& test_result) {
+    return test_result.test_part_results();
+  }
+};
+
+#if GTEST_CAN_STREAM_RESULTS_
+
+// Streams test results to the given port on the given host machine.
+class GTEST_API_ StreamingListener : public EmptyTestEventListener {
+ public:
+  // Abstract base class for writing strings to a socket.
+  class AbstractSocketWriter {
+   public:
+    virtual ~AbstractSocketWriter() {}
+
+    // Sends a string to the socket.
+    virtual void Send(const std::string& message) = 0;
+
+    // Closes the socket.
+    virtual void CloseConnection() {}
+
+    // Sends a string and a newline to the socket.
+    void SendLn(const std::string& message) { Send(message + "\n"); }
+  };
+
+  // Concrete class for actually writing strings to a socket.
+  class SocketWriter : public AbstractSocketWriter {
+   public:
+    SocketWriter(const std::string& host, const std::string& port)
+        : sockfd_(-1), host_name_(host), port_num_(port) {
+      MakeConnection();
+    }
+
+    virtual ~SocketWriter() {
+      if (sockfd_ != -1)
+        CloseConnection();
+    }
+
+    // Sends a string to the socket.
+    virtual void Send(const std::string& message) {
+      GTEST_CHECK_(sockfd_ != -1)
+          << "Send() can be called only when there is a connection.";
+
+      const int len = static_cast<int>(message.length());
+      if (write(sockfd_, message.c_str(), len) != len) {
+        GTEST_LOG_(WARNING)
+            << "stream_result_to: failed to stream to "
+            << host_name_ << ":" << port_num_;
+      }
+    }
+
+   private:
+    // Creates a client socket and connects to the server.
+    void MakeConnection();
+
+    // Closes the socket.
+    void CloseConnection() {
+      GTEST_CHECK_(sockfd_ != -1)
+          << "CloseConnection() can be called only when there is a connection.";
+
+      close(sockfd_);
+      sockfd_ = -1;
+    }
+
+    int sockfd_;  // socket file descriptor
+    const std::string host_name_;
+    const std::string port_num_;
+
+    GTEST_DISALLOW_COPY_AND_ASSIGN_(SocketWriter);
+  };  // class SocketWriter
+
+  // Escapes '=', '&', '%', and '\n' characters in str as "%xx".
+  static std::string UrlEncode(const char* str);
+
+  StreamingListener(const std::string& host, const std::string& port)
+      : socket_writer_(new SocketWriter(host, port)) {
+    Start();
+  }
+
+  explicit StreamingListener(AbstractSocketWriter* socket_writer)
+      : socket_writer_(socket_writer) { Start(); }
+
+  void OnTestProgramStart(const UnitTest& /* unit_test */) {
+    SendLn("event=TestProgramStart");
+  }
+
+  void OnTestProgramEnd(const UnitTest& unit_test) {
+    // Note that Google Test current only report elapsed time for each
+    // test iteration, not for the entire test program.
+    SendLn("event=TestProgramEnd&passed=" + FormatBool(unit_test.Passed()));
+
+    // Notify the streaming server to stop.
+    socket_writer_->CloseConnection();
+  }
+
+  void OnTestIterationStart(const UnitTest& /* unit_test */, int iteration) {
+    SendLn("event=TestIterationStart&iteration=" +
+           StreamableToString(iteration));
+  }
+
+  void OnTestIterationEnd(const UnitTest& unit_test, int /* iteration */) {
+    SendLn("event=TestIterationEnd&passed=" +
+           FormatBool(unit_test.Passed()) + "&elapsed_time=" +
+           StreamableToString(unit_test.elapsed_time()) + "ms");
+  }
+
+  void OnTestCaseStart(const TestCase& test_case) {
+    SendLn(std::string("event=TestCaseStart&name=") + test_case.name());
+  }
+
+  void OnTestCaseEnd(const TestCase& test_case) {
+    SendLn("event=TestCaseEnd&passed=" + FormatBool(test_case.Passed())
+           + "&elapsed_time=" + StreamableToString(test_case.elapsed_time())
+           + "ms");
+  }
+
+  void OnTestStart(const TestInfo& test_info) {
+    SendLn(std::string("event=TestStart&name=") + test_info.name());
+  }
+
+  void OnTestEnd(const TestInfo& test_info) {
+    SendLn("event=TestEnd&passed=" +
+           FormatBool((test_info.result())->Passed()) +
+           "&elapsed_time=" +
+           StreamableToString((test_info.result())->elapsed_time()) + "ms");
+  }
+
+  void OnTestPartResult(const TestPartResult& test_part_result) {
+    const char* file_name = test_part_result.file_name();
+    if (file_name == NULL)
+      file_name = "";
+    SendLn("event=TestPartResult&file=" + UrlEncode(file_name) +
+           "&line=" + StreamableToString(test_part_result.line_number()) +
+           "&message=" + UrlEncode(test_part_result.message()));
+  }
+
+ private:
+  // Sends the given message and a newline to the socket.
+  void SendLn(const std::string& message) { socket_writer_->SendLn(message); }
+
+  // Called at the start of streaming to notify the receiver what
+  // protocol we are using.
+  void Start() { SendLn("gtest_streaming_protocol_version=1.0"); }
+
+  std::string FormatBool(bool value) { return value ? "1" : "0"; }
+
+  const scoped_ptr<AbstractSocketWriter> socket_writer_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(StreamingListener);
+};  // class StreamingListener
+
+#endif  // GTEST_CAN_STREAM_RESULTS_
+
+}  // namespace internal
+
+// Returns the value of g_help_flag.
+bool GetGtestHelpFlag() {
+  return internal::g_help_flag;
+}
+}  // namespace testing
+
+#endif  // GTEST_SRC_GTEST_INTERNAL_INL_H_
+#undef GTEST_IMPLEMENTATION_
+
+#if GTEST_OS_WINDOWS
+# define vsnprintf _vsnprintf
+#endif  // GTEST_OS_WINDOWS
+
+namespace testing {
+
+using internal::CountIf;
+using internal::ForEach;
+using internal::GetElementOr;
+using internal::Shuffle;
+
+// Constants.
+
+// A test whose test case name or test name matches this filter is
+// disabled and not run.
+static const char kDisableTestFilter[] = "DISABLED_*:*/DISABLED_*";
+
+// A test case whose name matches this filter is considered a death
+// test case and will be run before test cases whose name doesn't
+// match this filter.
+static const char kDeathTestCaseFilter[] = "*DeathTest:*DeathTest/*";
+
+// A test filter that matches everything.
+static const char kUniversalFilter[] = "*";
+
+// The default output file for XML output.
+static const char kDefaultOutputFile[] = "test_detail.xml";
+
+// The environment variable name for the test shard index.
+static const char kTestShardIndex[] = "GTEST_SHARD_INDEX";
+// The environment variable name for the total number of test shards.
+static const char kTestTotalShards[] = "GTEST_TOTAL_SHARDS";
+// The environment variable name for the test shard status file.
+static const char kTestShardStatusFile[] = "GTEST_SHARD_STATUS_FILE";
+
+namespace internal {
+
+// The text used in failure messages to indicate the start of the
+// stack trace.
+const char kStackTraceMarker[] = "\nStack trace:\n";
+
+// g_help_flag is true iff the --help flag or an equivalent form is
+// specified on the command line.
+bool g_help_flag = false;
+
+}  // namespace internal
+
+static const char* GetDefaultFilter() {
+#ifdef GTEST_TEST_FILTER_ENV_VAR_
+  const char* const testbridge_test_only = getenv(GTEST_TEST_FILTER_ENV_VAR_);
+  if (testbridge_test_only != NULL) {
+    return testbridge_test_only;
+  }
+#endif  // GTEST_TEST_FILTER_ENV_VAR_
+  return kUniversalFilter;
+}
+
+GTEST_DEFINE_bool_(
+    also_run_disabled_tests,
+    internal::BoolFromGTestEnv("also_run_disabled_tests", false),
+    "Run disabled tests too, in addition to the tests normally being run.");
+
+GTEST_DEFINE_bool_(
+    break_on_failure,
+    internal::BoolFromGTestEnv("break_on_failure", false),
+    "True iff a failed assertion should be a debugger break-point.");
+
+GTEST_DEFINE_bool_(
+    catch_exceptions,
+    internal::BoolFromGTestEnv("catch_exceptions", true),
+    "True iff " GTEST_NAME_
+    " should catch exceptions and treat them as test failures.");
+
+GTEST_DEFINE_string_(
+    color,
+    internal::StringFromGTestEnv("color", "auto"),
+    "Whether to use colors in the output.  Valid values: yes, no, "
+    "and auto.  'auto' means to use colors if the output is "
+    "being sent to a terminal and the TERM environment variable "
+    "is set to a terminal type that supports colors.");
+
+GTEST_DEFINE_string_(
+    filter,
+    internal::StringFromGTestEnv("filter", GetDefaultFilter()),
+    "A colon-separated list of glob (not regex) patterns "
+    "for filtering the tests to run, optionally followed by a "
+    "'-' and a : separated list of negative patterns (tests to "
+    "exclude).  A test is run if it matches one of the positive "
+    "patterns and does not match any of the negative patterns.");
+
+GTEST_DEFINE_bool_(list_tests, false,
+                   "List all tests without running them.");
+
+GTEST_DEFINE_string_(
+    output,
+    internal::StringFromGTestEnv("output", ""),
+    "A format (currently must be \"xml\"), optionally followed "
+    "by a colon and an output file name or directory. A directory "
+    "is indicated by a trailing pathname separator. "
+    "Examples: \"xml:filename.xml\", \"xml::directoryname/\". "
+    "If a directory is specified, output files will be created "
+    "within that directory, with file-names based on the test "
+    "executable's name and, if necessary, made unique by adding "
+    "digits.");
+
+GTEST_DEFINE_bool_(
+    print_time,
+    internal::BoolFromGTestEnv("print_time", true),
+    "True iff " GTEST_NAME_
+    " should display elapsed time in text output.");
+
+GTEST_DEFINE_int32_(
+    random_seed,
+    internal::Int32FromGTestEnv("random_seed", 0),
+    "Random number seed to use when shuffling test orders.  Must be in range "
+    "[1, 99999], or 0 to use a seed based on the current time.");
+
+GTEST_DEFINE_int32_(
+    repeat,
+    internal::Int32FromGTestEnv("repeat", 1),
+    "How many times to repeat each test.  Specify a negative number "
+    "for repeating forever.  Useful for shaking out flaky tests.");
+
+GTEST_DEFINE_bool_(
+    show_internal_stack_frames, false,
+    "True iff " GTEST_NAME_ " should include internal stack frames when "
+    "printing test failure stack traces.");
+
+GTEST_DEFINE_bool_(
+    shuffle,
+    internal::BoolFromGTestEnv("shuffle", false),
+    "True iff " GTEST_NAME_
+    " should randomize tests' order on every run.");
+
+GTEST_DEFINE_int32_(
+    stack_trace_depth,
+    internal::Int32FromGTestEnv("stack_trace_depth", kMaxStackTraceDepth),
+    "The maximum number of stack frames to print when an "
+    "assertion fails.  The valid range is 0 through 100, inclusive.");
+
+GTEST_DEFINE_string_(
+    stream_result_to,
+    internal::StringFromGTestEnv("stream_result_to", ""),
+    "This flag specifies the host name and the port number on which to stream "
+    "test results. Example: \"localhost:555\". The flag is effective only on "
+    "Linux.");
+
+GTEST_DEFINE_bool_(
+    throw_on_failure,
+    internal::BoolFromGTestEnv("throw_on_failure", false),
+    "When this flag is specified, a failed assertion will throw an exception "
+    "if exceptions are enabled or exit the program with a non-zero code "
+    "otherwise.");
+
+#if GTEST_USE_OWN_FLAGFILE_FLAG_
+GTEST_DEFINE_string_(
+    flagfile,
+    internal::StringFromGTestEnv("flagfile", ""),
+    "This flag specifies the flagfile to read command-line flags from.");
+#endif  // GTEST_USE_OWN_FLAGFILE_FLAG_
+
+namespace internal {
+
+// Generates a random number from [0, range), using a Linear
+// Congruential Generator (LCG).  Crashes if 'range' is 0 or greater
+// than kMaxRange.
+UInt32 Random::Generate(UInt32 range) {
+  // These constants are the same as are used in glibc's rand(3).
+  state_ = (1103515245U*state_ + 12345U) % kMaxRange;
+
+  GTEST_CHECK_(range > 0)
+      << "Cannot generate a number in the range [0, 0).";
+  GTEST_CHECK_(range <= kMaxRange)
+      << "Generation of a number in [0, " << range << ") was requested, "
+      << "but this can only generate numbers in [0, " << kMaxRange << ").";
+
+  // Converting via modulus introduces a bit of downward bias, but
+  // it's simple, and a linear congruential generator isn't too good
+  // to begin with.
+  return state_ % range;
+}
+
+// GTestIsInitialized() returns true iff the user has initialized
+// Google Test.  Useful for catching the user mistake of not initializing
+// Google Test before calling RUN_ALL_TESTS().
+static bool GTestIsInitialized() { return GetArgvs().size() > 0; }
+
+// Iterates over a vector of TestCases, keeping a running sum of the
+// results of calling a given int-returning method on each.
+// Returns the sum.
+static int SumOverTestCaseList(const std::vector<TestCase*>& case_list,
+                               int (TestCase::*method)() const) {
+  int sum = 0;
+  for (size_t i = 0; i < case_list.size(); i++) {
+    sum += (case_list[i]->*method)();
+  }
+  return sum;
+}
+
+// Returns true iff the test case passed.
+static bool TestCasePassed(const TestCase* test_case) {
+  return test_case->should_run() && test_case->Passed();
+}
+
+// Returns true iff the test case failed.
+static bool TestCaseFailed(const TestCase* test_case) {
+  return test_case->should_run() && test_case->Failed();
+}
+
+// Returns true iff test_case contains at least one test that should
+// run.
+static bool ShouldRunTestCase(const TestCase* test_case) {
+  return test_case->should_run();
+}
+
+// AssertHelper constructor.
+AssertHelper::AssertHelper(TestPartResult::Type type,
+                           const char* file,
+                           int line,
+                           const char* message)
+    : data_(new AssertHelperData(type, file, line, message)) {
+}
+
+AssertHelper::~AssertHelper() {
+  delete data_;
+}
+
+// Message assignment, for assertion streaming support.
+void AssertHelper::operator=(const Message& message) const {
+  UnitTest::GetInstance()->
+    AddTestPartResult(data_->type, data_->file, data_->line,
+                      AppendUserMessage(data_->message, message),
+                      UnitTest::GetInstance()->impl()
+                      ->CurrentOsStackTraceExceptTop(1)
+                      // Skips the stack frame for this function itself.
+                      );  // NOLINT
+}
+
+// Mutex for linked pointers.
+GTEST_API_ GTEST_DEFINE_STATIC_MUTEX_(g_linked_ptr_mutex);
+
+// A copy of all command line arguments.  Set by InitGoogleTest().
+::std::vector<testing::internal::string> g_argvs;
+
+const ::std::vector<testing::internal::string>& GetArgvs() {
+#if defined(GTEST_CUSTOM_GET_ARGVS_)
+  return GTEST_CUSTOM_GET_ARGVS_();
+#else  // defined(GTEST_CUSTOM_GET_ARGVS_)
+  return g_argvs;
+#endif  // defined(GTEST_CUSTOM_GET_ARGVS_)
+}
+
+// Returns the current application's name, removing directory path if that
+// is present.
+FilePath GetCurrentExecutableName() {
+  FilePath result;
+
+#if GTEST_OS_WINDOWS
+  result.Set(FilePath(GetArgvs()[0]).RemoveExtension("exe"));
+#else
+  result.Set(FilePath(GetArgvs()[0]));
+#endif  // GTEST_OS_WINDOWS
+
+  return result.RemoveDirectoryName();
+}
+
+// Functions for processing the gtest_output flag.
+
+// Returns the output format, or "" for normal printed output.
+std::string UnitTestOptions::GetOutputFormat() {
+  const char* const gtest_output_flag = GTEST_FLAG(output).c_str();
+  if (gtest_output_flag == NULL) return std::string("");
+
+  const char* const colon = strchr(gtest_output_flag, ':');
+  return (colon == NULL) ?
+      std::string(gtest_output_flag) :
+      std::string(gtest_output_flag, colon - gtest_output_flag);
+}
+
+// Returns the name of the requested output file, or the default if none
+// was explicitly specified.
+std::string UnitTestOptions::GetAbsolutePathToOutputFile() {
+  const char* const gtest_output_flag = GTEST_FLAG(output).c_str();
+  if (gtest_output_flag == NULL)
+    return "";
+
+  const char* const colon = strchr(gtest_output_flag, ':');
+  if (colon == NULL)
+    return internal::FilePath::ConcatPaths(
+        internal::FilePath(
+            UnitTest::GetInstance()->original_working_dir()),
+        internal::FilePath(kDefaultOutputFile)).string();
+
+  internal::FilePath output_name(colon + 1);
+  if (!output_name.IsAbsolutePath())
+    // TODO(wan@google.com): on Windows \some\path is not an absolute
+    // path (as its meaning depends on the current drive), yet the
+    // following logic for turning it into an absolute path is wrong.
+    // Fix it.
+    output_name = internal::FilePath::ConcatPaths(
+        internal::FilePath(UnitTest::GetInstance()->original_working_dir()),
+        internal::FilePath(colon + 1));
+
+  if (!output_name.IsDirectory())
+    return output_name.string();
+
+  internal::FilePath result(internal::FilePath::GenerateUniqueFileName(
+      output_name, internal::GetCurrentExecutableName(),
+      GetOutputFormat().c_str()));
+  return result.string();
+}
+
+// Returns true iff the wildcard pattern matches the string.  The
+// first ':' or '\0' character in pattern marks the end of it.
+//
+// This recursive algorithm isn't very efficient, but is clear and
+// works well enough for matching test names, which are short.
+bool UnitTestOptions::PatternMatchesString(const char *pattern,
+                                           const char *str) {
+  switch (*pattern) {
+    case '\0':
+    case ':':  // Either ':' or '\0' marks the end of the pattern.
+      return *str == '\0';
+    case '?':  // Matches any single character.
+      return *str != '\0' && PatternMatchesString(pattern + 1, str + 1);
+    case '*':  // Matches any string (possibly empty) of characters.
+      return (*str != '\0' && PatternMatchesString(pattern, str + 1)) ||
+          PatternMatchesString(pattern + 1, str);
+    default:  // Non-special character.  Matches itself.
+      return *pattern == *str &&
+          PatternMatchesString(pattern + 1, str + 1);
+  }
+}
+
+bool UnitTestOptions::MatchesFilter(
+    const std::string& name, const char* filter) {
+  const char *cur_pattern = filter;
+  for (;;) {
+    if (PatternMatchesString(cur_pattern, name.c_str())) {
+      return true;
+    }
+
+    // Finds the next pattern in the filter.
+    cur_pattern = strchr(cur_pattern, ':');
+
+    // Returns if no more pattern can be found.
+    if (cur_pattern == NULL) {
+      return false;
+    }
+
+    // Skips the pattern separator (the ':' character).
+    cur_pattern++;
+  }
+}
+
+// Returns true iff the user-specified filter matches the test case
+// name and the test name.
+bool UnitTestOptions::FilterMatchesTest(const std::string &test_case_name,
+                                        const std::string &test_name) {
+  const std::string& full_name = test_case_name + "." + test_name.c_str();
+
+  // Split --gtest_filter at '-', if there is one, to separate into
+  // positive filter and negative filter portions
+  const char* const p = GTEST_FLAG(filter).c_str();
+  const char* const dash = strchr(p, '-');
+  std::string positive;
+  std::string negative;
+  if (dash == NULL) {
+    positive = GTEST_FLAG(filter).c_str();  // Whole string is a positive filter
+    negative = "";
+  } else {
+    positive = std::string(p, dash);   // Everything up to the dash
+    negative = std::string(dash + 1);  // Everything after the dash
+    if (positive.empty()) {
+      // Treat '-test1' as the same as '*-test1'
+      positive = kUniversalFilter;
+    }
+  }
+
+  // A filter is a colon-separated list of patterns.  It matches a
+  // test if any pattern in it matches the test.
+  return (MatchesFilter(full_name, positive.c_str()) &&
+          !MatchesFilter(full_name, negative.c_str()));
+}
+
+#if GTEST_HAS_SEH
+// Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the
+// given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise.
+// This function is useful as an __except condition.
+int UnitTestOptions::GTestShouldProcessSEH(DWORD exception_code) {
+  // Google Test should handle a SEH exception if:
+  //   1. the user wants it to, AND
+  //   2. this is not a breakpoint exception, AND
+  //   3. this is not a C++ exception (VC++ implements them via SEH,
+  //      apparently).
+  //
+  // SEH exception code for C++ exceptions.
+  // (see http://support.microsoft.com/kb/185294 for more information).
+  const DWORD kCxxExceptionCode = 0xe06d7363;
+
+  bool should_handle = true;
+
+  if (!GTEST_FLAG(catch_exceptions))
+    should_handle = false;
+  else if (exception_code == EXCEPTION_BREAKPOINT)
+    should_handle = false;
+  else if (exception_code == kCxxExceptionCode)
+    should_handle = false;
+
+  return should_handle ? EXCEPTION_EXECUTE_HANDLER : EXCEPTION_CONTINUE_SEARCH;
+}
+#endif  // GTEST_HAS_SEH
+
+}  // namespace internal
+
+// The c'tor sets this object as the test part result reporter used by
+// Google Test.  The 'result' parameter specifies where to report the
+// results. Intercepts only failures from the current thread.
+ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter(
+    TestPartResultArray* result)
+    : intercept_mode_(INTERCEPT_ONLY_CURRENT_THREAD),
+      result_(result) {
+  Init();
+}
+
+// The c'tor sets this object as the test part result reporter used by
+// Google Test.  The 'result' parameter specifies where to report the
+// results.
+ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter(
+    InterceptMode intercept_mode, TestPartResultArray* result)
+    : intercept_mode_(intercept_mode),
+      result_(result) {
+  Init();
+}
+
+void ScopedFakeTestPartResultReporter::Init() {
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  if (intercept_mode_ == INTERCEPT_ALL_THREADS) {
+    old_reporter_ = impl->GetGlobalTestPartResultReporter();
+    impl->SetGlobalTestPartResultReporter(this);
+  } else {
+    old_reporter_ = impl->GetTestPartResultReporterForCurrentThread();
+    impl->SetTestPartResultReporterForCurrentThread(this);
+  }
+}
+
+// The d'tor restores the test part result reporter used by Google Test
+// before.
+ScopedFakeTestPartResultReporter::~ScopedFakeTestPartResultReporter() {
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  if (intercept_mode_ == INTERCEPT_ALL_THREADS) {
+    impl->SetGlobalTestPartResultReporter(old_reporter_);
+  } else {
+    impl->SetTestPartResultReporterForCurrentThread(old_reporter_);
+  }
+}
+
+// Increments the test part result count and remembers the result.
+// This method is from the TestPartResultReporterInterface interface.
+void ScopedFakeTestPartResultReporter::ReportTestPartResult(
+    const TestPartResult& result) {
+  result_->Append(result);
+}
+
+namespace internal {
+
+// Returns the type ID of ::testing::Test.  We should always call this
+// instead of GetTypeId< ::testing::Test>() to get the type ID of
+// testing::Test.  This is to work around a suspected linker bug when
+// using Google Test as a framework on Mac OS X.  The bug causes
+// GetTypeId< ::testing::Test>() to return different values depending
+// on whether the call is from the Google Test framework itself or
+// from user test code.  GetTestTypeId() is guaranteed to always
+// return the same value, as it always calls GetTypeId<>() from the
+// gtest.cc, which is within the Google Test framework.
+TypeId GetTestTypeId() {
+  return GetTypeId<Test>();
+}
+
+// The value of GetTestTypeId() as seen from within the Google Test
+// library.  This is solely for testing GetTestTypeId().
+extern const TypeId kTestTypeIdInGoogleTest = GetTestTypeId();
+
+// This predicate-formatter checks that 'results' contains a test part
+// failure of the given type and that the failure message contains the
+// given substring.
+AssertionResult HasOneFailure(const char* /* results_expr */,
+                              const char* /* type_expr */,
+                              const char* /* substr_expr */,
+                              const TestPartResultArray& results,
+                              TestPartResult::Type type,
+                              const std::string& substr) {
+  const std::string expected(type == TestPartResult::kFatalFailure ?
+                        "1 fatal failure" :
+                        "1 non-fatal failure");
+  Message msg;
+  if (results.size() != 1) {
+    msg << "Expected: " << expected << "\n"
+        << "  Actual: " << results.size() << " failures";
+    for (int i = 0; i < results.size(); i++) {
+      msg << "\n" << results.GetTestPartResult(i);
+    }
+    return AssertionFailure() << msg;
+  }
+
+  const TestPartResult& r = results.GetTestPartResult(0);
+  if (r.type() != type) {
+    return AssertionFailure() << "Expected: " << expected << "\n"
+                              << "  Actual:\n"
+                              << r;
+  }
+
+  if (strstr(r.message(), substr.c_str()) == NULL) {
+    return AssertionFailure() << "Expected: " << expected << " containing \""
+                              << substr << "\"\n"
+                              << "  Actual:\n"
+                              << r;
+  }
+
+  return AssertionSuccess();
+}
+
+// The constructor of SingleFailureChecker remembers where to look up
+// test part results, what type of failure we expect, and what
+// substring the failure message should contain.
+SingleFailureChecker::SingleFailureChecker(const TestPartResultArray* results,
+                                           TestPartResult::Type type,
+                                           const std::string& substr)
+    : results_(results), type_(type), substr_(substr) {}
+
+// The destructor of SingleFailureChecker verifies that the given
+// TestPartResultArray contains exactly one failure that has the given
+// type and contains the given substring.  If that's not the case, a
+// non-fatal failure will be generated.
+SingleFailureChecker::~SingleFailureChecker() {
+  EXPECT_PRED_FORMAT3(HasOneFailure, *results_, type_, substr_);
+}
+
+DefaultGlobalTestPartResultReporter::DefaultGlobalTestPartResultReporter(
+    UnitTestImpl* unit_test) : unit_test_(unit_test) {}
+
+void DefaultGlobalTestPartResultReporter::ReportTestPartResult(
+    const TestPartResult& result) {
+  unit_test_->current_test_result()->AddTestPartResult(result);
+  unit_test_->listeners()->repeater()->OnTestPartResult(result);
+}
+
+DefaultPerThreadTestPartResultReporter::DefaultPerThreadTestPartResultReporter(
+    UnitTestImpl* unit_test) : unit_test_(unit_test) {}
+
+void DefaultPerThreadTestPartResultReporter::ReportTestPartResult(
+    const TestPartResult& result) {
+  unit_test_->GetGlobalTestPartResultReporter()->ReportTestPartResult(result);
+}
+
+// Returns the global test part result reporter.
+TestPartResultReporterInterface*
+UnitTestImpl::GetGlobalTestPartResultReporter() {
+  internal::MutexLock lock(&global_test_part_result_reporter_mutex_);
+  return global_test_part_result_repoter_;
+}
+
+// Sets the global test part result reporter.
+void UnitTestImpl::SetGlobalTestPartResultReporter(
+    TestPartResultReporterInterface* reporter) {
+  internal::MutexLock lock(&global_test_part_result_reporter_mutex_);
+  global_test_part_result_repoter_ = reporter;
+}
+
+// Returns the test part result reporter for the current thread.
+TestPartResultReporterInterface*
+UnitTestImpl::GetTestPartResultReporterForCurrentThread() {
+  return per_thread_test_part_result_reporter_.get();
+}
+
+// Sets the test part result reporter for the current thread.
+void UnitTestImpl::SetTestPartResultReporterForCurrentThread(
+    TestPartResultReporterInterface* reporter) {
+  per_thread_test_part_result_reporter_.set(reporter);
+}
+
+// Gets the number of successful test cases.
+int UnitTestImpl::successful_test_case_count() const {
+  return CountIf(test_cases_, TestCasePassed);
+}
+
+// Gets the number of failed test cases.
+int UnitTestImpl::failed_test_case_count() const {
+  return CountIf(test_cases_, TestCaseFailed);
+}
+
+// Gets the number of all test cases.
+int UnitTestImpl::total_test_case_count() const {
+  return static_cast<int>(test_cases_.size());
+}
+
+// Gets the number of all test cases that contain at least one test
+// that should run.
+int UnitTestImpl::test_case_to_run_count() const {
+  return CountIf(test_cases_, ShouldRunTestCase);
+}
+
+// Gets the number of successful tests.
+int UnitTestImpl::successful_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::successful_test_count);
+}
+
+// Gets the number of skipped tests.
+int UnitTestImpl::skipped_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::skipped_test_count);
+}
+
+// Gets the number of failed tests.
+int UnitTestImpl::failed_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::failed_test_count);
+}
+
+// Gets the number of disabled tests that will be reported in the XML report.
+int UnitTestImpl::reportable_disabled_test_count() const {
+  return SumOverTestCaseList(test_cases_,
+                             &TestCase::reportable_disabled_test_count);
+}
+
+// Gets the number of disabled tests.
+int UnitTestImpl::disabled_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::disabled_test_count);
+}
+
+// Gets the number of tests to be printed in the XML report.
+int UnitTestImpl::reportable_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::reportable_test_count);
+}
+
+// Gets the number of all tests.
+int UnitTestImpl::total_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::total_test_count);
+}
+
+// Gets the number of tests that should run.
+int UnitTestImpl::test_to_run_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::test_to_run_count);
+}
+
+// Returns the current OS stack trace as an std::string.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag.  The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// CurrentOsStackTraceExceptTop(1), Foo() will be included in the
+// trace but Bar() and CurrentOsStackTraceExceptTop() won't.
+std::string UnitTestImpl::CurrentOsStackTraceExceptTop(int skip_count) {
+  return os_stack_trace_getter()->CurrentStackTrace(
+      static_cast<int>(GTEST_FLAG(stack_trace_depth)),
+      skip_count + 1
+      // Skips the user-specified number of frames plus this function
+      // itself.
+      );  // NOLINT
+}
+
+// Returns the current time in milliseconds.
+TimeInMillis GetTimeInMillis() {
+#if GTEST_OS_WINDOWS_MOBILE || defined(__BORLANDC__)
+  // Difference between 1970-01-01 and 1601-01-01 in milliseconds.
+  // http://analogous.blogspot.com/2005/04/epoch.html
+  const TimeInMillis kJavaEpochToWinFileTimeDelta =
+    static_cast<TimeInMillis>(116444736UL) * 100000UL;
+  const DWORD kTenthMicrosInMilliSecond = 10000;
+
+  SYSTEMTIME now_systime;
+  FILETIME now_filetime;
+  ULARGE_INTEGER now_int64;
+  // TODO(kenton@google.com): Shouldn't this just use
+  //   GetSystemTimeAsFileTime()?
+  GetSystemTime(&now_systime);
+  if (SystemTimeToFileTime(&now_systime, &now_filetime)) {
+    now_int64.LowPart = now_filetime.dwLowDateTime;
+    now_int64.HighPart = now_filetime.dwHighDateTime;
+    now_int64.QuadPart = (now_int64.QuadPart / kTenthMicrosInMilliSecond) -
+      kJavaEpochToWinFileTimeDelta;
+    return now_int64.QuadPart;
+  }
+  return 0;
+#elif GTEST_OS_WINDOWS && !GTEST_HAS_GETTIMEOFDAY_
+  __timeb64 now;
+
+  // MSVC 8 deprecates _ftime64(), so we want to suppress warning 4996
+  // (deprecated function) there.
+  // TODO(kenton@google.com): Use GetTickCount()?  Or use
+  //   SystemTimeToFileTime()
+  GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996)
+  _ftime64(&now);
+  GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+  return static_cast<TimeInMillis>(now.time) * 1000 + now.millitm;
+#elif GTEST_HAS_GETTIMEOFDAY_
+  struct timeval now;
+  gettimeofday(&now, NULL);
+  return static_cast<TimeInMillis>(now.tv_sec) * 1000 + now.tv_usec / 1000;
+#else
+# error "Don't know how to get the current time on your system."
+#endif
+}
+
+// Utilities
+
+// class String.
+
+#if GTEST_OS_WINDOWS_MOBILE
+// Creates a UTF-16 wide string from the given ANSI string, allocating
+// memory using new. The caller is responsible for deleting the return
+// value using delete[]. Returns the wide string, or NULL if the
+// input is NULL.
+LPCWSTR String::AnsiToUtf16(const char* ansi) {
+  if (!ansi) return NULL;
+  const int length = strlen(ansi);
+  const int unicode_length =
+      MultiByteToWideChar(CP_ACP, 0, ansi, length,
+                          NULL, 0);
+  WCHAR* unicode = new WCHAR[unicode_length + 1];
+  MultiByteToWideChar(CP_ACP, 0, ansi, length,
+                      unicode, unicode_length);
+  unicode[unicode_length] = 0;
+  return unicode;
+}
+
+// Creates an ANSI string from the given wide string, allocating
+// memory using new. The caller is responsible for deleting the return
+// value using delete[]. Returns the ANSI string, or NULL if the
+// input is NULL.
+const char* String::Utf16ToAnsi(LPCWSTR utf16_str)  {
+  if (!utf16_str) return NULL;
+  const int ansi_length =
+      WideCharToMultiByte(CP_ACP, 0, utf16_str, -1,
+                          NULL, 0, NULL, NULL);
+  char* ansi = new char[ansi_length + 1];
+  WideCharToMultiByte(CP_ACP, 0, utf16_str, -1,
+                      ansi, ansi_length, NULL, NULL);
+  ansi[ansi_length] = 0;
+  return ansi;
+}
+
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+// Compares two C strings.  Returns true iff they have the same content.
+//
+// Unlike strcmp(), this function can handle NULL argument(s).  A NULL
+// C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::CStringEquals(const char * lhs, const char * rhs) {
+  if ( lhs == NULL ) return rhs == NULL;
+
+  if ( rhs == NULL ) return false;
+
+  return strcmp(lhs, rhs) == 0;
+}
+
+#if GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING
+
+// Converts an array of wide chars to a narrow string using the UTF-8
+// encoding, and streams the result to the given Message object.
+static void StreamWideCharsToMessage(const wchar_t* wstr, size_t length,
+                                     Message* msg) {
+  for (size_t i = 0; i != length; ) {  // NOLINT
+    if (wstr[i] != L'\0') {
+      *msg << WideStringToUtf8(wstr + i, static_cast<int>(length - i));
+      while (i != length && wstr[i] != L'\0')
+        i++;
+    } else {
+      *msg << '\0';
+      i++;
+    }
+  }
+}
+
+#endif  // GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING
+
+void SplitString(const ::std::string& str, char delimiter,
+                 ::std::vector< ::std::string>* dest) {
+  ::std::vector< ::std::string> parsed;
+  ::std::string::size_type pos = 0;
+  while (::testing::internal::AlwaysTrue()) {
+    const ::std::string::size_type colon = str.find(delimiter, pos);
+    if (colon == ::std::string::npos) {
+      parsed.push_back(str.substr(pos));
+      break;
+    } else {
+      parsed.push_back(str.substr(pos, colon - pos));
+      pos = colon + 1;
+    }
+  }
+  dest->swap(parsed);
+}
+
+}  // namespace internal
+
+// Constructs an empty Message.
+// We allocate the stringstream separately because otherwise each use of
+// ASSERT/EXPECT in a procedure adds over 200 bytes to the procedure's
+// stack frame leading to huge stack frames in some cases; gcc does not reuse
+// the stack space.
+Message::Message() : ss_(new ::std::stringstream) {
+  // By default, we want there to be enough precision when printing
+  // a double to a Message.
+  *ss_ << std::setprecision(std::numeric_limits<double>::digits10 + 2);
+}
+
+// These two overloads allow streaming a wide C string to a Message
+// using the UTF-8 encoding.
+Message& Message::operator <<(const wchar_t* wide_c_str) {
+  return *this << internal::String::ShowWideCString(wide_c_str);
+}
+Message& Message::operator <<(wchar_t* wide_c_str) {
+  return *this << internal::String::ShowWideCString(wide_c_str);
+}
+
+#if GTEST_HAS_STD_WSTRING
+// Converts the given wide string to a narrow string using the UTF-8
+// encoding, and streams the result to this Message object.
+Message& Message::operator <<(const ::std::wstring& wstr) {
+  internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this);
+  return *this;
+}
+#endif  // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+// Converts the given wide string to a narrow string using the UTF-8
+// encoding, and streams the result to this Message object.
+Message& Message::operator <<(const ::wstring& wstr) {
+  internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this);
+  return *this;
+}
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+// Gets the text streamed to this object so far as an std::string.
+// Each '\0' character in the buffer is replaced with "\\0".
+std::string Message::GetString() const {
+  return internal::StringStreamToString(ss_.get());
+}
+
+// AssertionResult constructors.
+// Used in EXPECT_TRUE/FALSE(assertion_result).
+AssertionResult::AssertionResult(const AssertionResult& other)
+    : success_(other.success_),
+      message_(other.message_.get() != NULL ?
+               new ::std::string(*other.message_) :
+               static_cast< ::std::string*>(NULL)) {
+}
+
+// Swaps two AssertionResults.
+void AssertionResult::swap(AssertionResult& other) {
+  using std::swap;
+  swap(success_, other.success_);
+  swap(message_, other.message_);
+}
+
+// Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE.
+AssertionResult AssertionResult::operator!() const {
+  AssertionResult negation(!success_);
+  if (message_.get() != NULL)
+    negation << *message_;
+  return negation;
+}
+
+// Makes a successful assertion result.
+AssertionResult AssertionSuccess() {
+  return AssertionResult(true);
+}
+
+// Makes a failed assertion result.
+AssertionResult AssertionFailure() {
+  return AssertionResult(false);
+}
+
+// Makes a failed assertion result with the given failure message.
+// Deprecated; use AssertionFailure() << message.
+AssertionResult AssertionFailure(const Message& message) {
+  return AssertionFailure() << message;
+}
+
+namespace internal {
+
+namespace edit_distance {
+std::vector<EditType> CalculateOptimalEdits(const std::vector<size_t>& left,
+                                            const std::vector<size_t>& right) {
+  std::vector<std::vector<double> > costs(
+      left.size() + 1, std::vector<double>(right.size() + 1));
+  std::vector<std::vector<EditType> > best_move(
+      left.size() + 1, std::vector<EditType>(right.size() + 1));
+
+  // Populate for empty right.
+  for (size_t l_i = 0; l_i < costs.size(); ++l_i) {
+    costs[l_i][0] = static_cast<double>(l_i);
+    best_move[l_i][0] = kRemove;
+  }
+  // Populate for empty left.
+  for (size_t r_i = 1; r_i < costs[0].size(); ++r_i) {
+    costs[0][r_i] = static_cast<double>(r_i);
+    best_move[0][r_i] = kAdd;
+  }
+
+  for (size_t l_i = 0; l_i < left.size(); ++l_i) {
+    for (size_t r_i = 0; r_i < right.size(); ++r_i) {
+      if (left[l_i] == right[r_i]) {
+        // Found a match. Consume it.
+        costs[l_i + 1][r_i + 1] = costs[l_i][r_i];
+        best_move[l_i + 1][r_i + 1] = kMatch;
+        continue;
+      }
+
+      const double add = costs[l_i + 1][r_i];
+      const double remove = costs[l_i][r_i + 1];
+      const double replace = costs[l_i][r_i];
+      if (add < remove && add < replace) {
+        costs[l_i + 1][r_i + 1] = add + 1;
+        best_move[l_i + 1][r_i + 1] = kAdd;
+      } else if (remove < add && remove < replace) {
+        costs[l_i + 1][r_i + 1] = remove + 1;
+        best_move[l_i + 1][r_i + 1] = kRemove;
+      } else {
+        // We make replace a little more expensive than add/remove to lower
+        // their priority.
+        costs[l_i + 1][r_i + 1] = replace + 1.00001;
+        best_move[l_i + 1][r_i + 1] = kReplace;
+      }
+    }
+  }
+
+  // Reconstruct the best path. We do it in reverse order.
+  std::vector<EditType> best_path;
+  for (size_t l_i = left.size(), r_i = right.size(); l_i > 0 || r_i > 0;) {
+    EditType move = best_move[l_i][r_i];
+    best_path.push_back(move);
+    l_i -= move != kAdd;
+    r_i -= move != kRemove;
+  }
+  std::reverse(best_path.begin(), best_path.end());
+  return best_path;
+}
+
+namespace {
+
+// Helper class to convert string into ids with deduplication.
+class InternalStrings {
+ public:
+  size_t GetId(const std::string& str) {
+    IdMap::iterator it = ids_.find(str);
+    if (it != ids_.end()) return it->second;
+    size_t id = ids_.size();
+    return ids_[str] = id;
+  }
+
+ private:
+  typedef std::map<std::string, size_t> IdMap;
+  IdMap ids_;
+};
+
+}  // namespace
+
+std::vector<EditType> CalculateOptimalEdits(
+    const std::vector<std::string>& left,
+    const std::vector<std::string>& right) {
+  std::vector<size_t> left_ids, right_ids;
+  {
+    InternalStrings intern_table;
+    for (size_t i = 0; i < left.size(); ++i) {
+      left_ids.push_back(intern_table.GetId(left[i]));
+    }
+    for (size_t i = 0; i < right.size(); ++i) {
+      right_ids.push_back(intern_table.GetId(right[i]));
+    }
+  }
+  return CalculateOptimalEdits(left_ids, right_ids);
+}
+
+namespace {
+
+// Helper class that holds the state for one hunk and prints it out to the
+// stream.
+// It reorders adds/removes when possible to group all removes before all
+// adds. It also adds the hunk header before printint into the stream.
+class Hunk {
+ public:
+  Hunk(size_t left_start, size_t right_start)
+      : left_start_(left_start),
+        right_start_(right_start),
+        adds_(),
+        removes_(),
+        common_() {}
+
+  void PushLine(char edit, const char* line) {
+    switch (edit) {
+      case ' ':
+        ++common_;
+        FlushEdits();
+        hunk_.push_back(std::make_pair(' ', line));
+        break;
+      case '-':
+        ++removes_;
+        hunk_removes_.push_back(std::make_pair('-', line));
+        break;
+      case '+':
+        ++adds_;
+        hunk_adds_.push_back(std::make_pair('+', line));
+        break;
+    }
+  }
+
+  void PrintTo(std::ostream* os) {
+    PrintHeader(os);
+    FlushEdits();
+    for (std::list<std::pair<char, const char*> >::const_iterator it =
+             hunk_.begin();
+         it != hunk_.end(); ++it) {
+      *os << it->first << it->second << "\n";
+    }
+  }
+
+  bool has_edits() const { return adds_ || removes_; }
+
+ private:
+  void FlushEdits() {
+    hunk_.splice(hunk_.end(), hunk_removes_);
+    hunk_.splice(hunk_.end(), hunk_adds_);
+  }
+
+  // Print a unified diff header for one hunk.
+  // The format is
+  //   "@@ -<left_start>,<left_length> +<right_start>,<right_length> @@"
+  // where the left/right parts are omitted if unnecessary.
+  void PrintHeader(std::ostream* ss) const {
+    *ss << "@@ ";
+    if (removes_) {
+      *ss << "-" << left_start_ << "," << (removes_ + common_);
+    }
+    if (removes_ && adds_) {
+      *ss << " ";
+    }
+    if (adds_) {
+      *ss << "+" << right_start_ << "," << (adds_ + common_);
+    }
+    *ss << " @@\n";
+  }
+
+  size_t left_start_, right_start_;
+  size_t adds_, removes_, common_;
+  std::list<std::pair<char, const char*> > hunk_, hunk_adds_, hunk_removes_;
+};
+
+}  // namespace
+
+// Create a list of diff hunks in Unified diff format.
+// Each hunk has a header generated by PrintHeader above plus a body with
+// lines prefixed with ' ' for no change, '-' for deletion and '+' for
+// addition.
+// 'context' represents the desired unchanged prefix/suffix around the diff.
+// If two hunks are close enough that their contexts overlap, then they are
+// joined into one hunk.
+std::string CreateUnifiedDiff(const std::vector<std::string>& left,
+                              const std::vector<std::string>& right,
+                              size_t context) {
+  const std::vector<EditType> edits = CalculateOptimalEdits(left, right);
+
+  size_t l_i = 0, r_i = 0, edit_i = 0;
+  std::stringstream ss;
+  while (edit_i < edits.size()) {
+    // Find first edit.
+    while (edit_i < edits.size() && edits[edit_i] == kMatch) {
+      ++l_i;
+      ++r_i;
+      ++edit_i;
+    }
+
+    // Find the first line to include in the hunk.
+    const size_t prefix_context = std::min(l_i, context);
+    Hunk hunk(l_i - prefix_context + 1, r_i - prefix_context + 1);
+    for (size_t i = prefix_context; i > 0; --i) {
+      hunk.PushLine(' ', left[l_i - i].c_str());
+    }
+
+    // Iterate the edits until we found enough suffix for the hunk or the input
+    // is over.
+    size_t n_suffix = 0;
+    for (; edit_i < edits.size(); ++edit_i) {
+      if (n_suffix >= context) {
+        // Continue only if the next hunk is very close.
+        std::vector<EditType>::const_iterator it = edits.begin() + edit_i;
+        while (it != edits.end() && *it == kMatch) ++it;
+        if (it == edits.end() || (it - edits.begin()) - edit_i >= context) {
+          // There is no next edit or it is too far away.
+          break;
+        }
+      }
+
+      EditType edit = edits[edit_i];
+      // Reset count when a non match is found.
+      n_suffix = edit == kMatch ? n_suffix + 1 : 0;
+
+      if (edit == kMatch || edit == kRemove || edit == kReplace) {
+        hunk.PushLine(edit == kMatch ? ' ' : '-', left[l_i].c_str());
+      }
+      if (edit == kAdd || edit == kReplace) {
+        hunk.PushLine('+', right[r_i].c_str());
+      }
+
+      // Advance indices, depending on edit type.
+      l_i += edit != kAdd;
+      r_i += edit != kRemove;
+    }
+
+    if (!hunk.has_edits()) {
+      // We are done. We don't want this hunk.
+      break;
+    }
+
+    hunk.PrintTo(&ss);
+  }
+  return ss.str();
+}
+
+}  // namespace edit_distance
+
+namespace {
+
+// The string representation of the values received in EqFailure() are already
+// escaped. Split them on escaped '\n' boundaries. Leave all other escaped
+// characters the same.
+std::vector<std::string> SplitEscapedString(const std::string& str) {
+  std::vector<std::string> lines;
+  size_t start = 0, end = str.size();
+  if (end > 2 && str[0] == '"' && str[end - 1] == '"') {
+    ++start;
+    --end;
+  }
+  bool escaped = false;
+  for (size_t i = start; i + 1 < end; ++i) {
+    if (escaped) {
+      escaped = false;
+      if (str[i] == 'n') {
+        lines.push_back(str.substr(start, i - start - 1));
+        start = i + 1;
+      }
+    } else {
+      escaped = str[i] == '\\';
+    }
+  }
+  lines.push_back(str.substr(start, end - start));
+  return lines;
+}
+
+}  // namespace
+
+// Constructs and returns the message for an equality assertion
+// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure.
+//
+// The first four parameters are the expressions used in the assertion
+// and their values, as strings.  For example, for ASSERT_EQ(foo, bar)
+// where foo is 5 and bar is 6, we have:
+//
+//   lhs_expression: "foo"
+//   rhs_expression: "bar"
+//   lhs_value:      "5"
+//   rhs_value:      "6"
+//
+// The ignoring_case parameter is true iff the assertion is a
+// *_STRCASEEQ*.  When it's true, the string "Ignoring case" will
+// be inserted into the message.
+AssertionResult EqFailure(const char* lhs_expression,
+                          const char* rhs_expression,
+                          const std::string& lhs_value,
+                          const std::string& rhs_value,
+                          bool ignoring_case) {
+  Message msg;
+  msg << "      Expected: " << lhs_expression;
+  if (lhs_value != lhs_expression) {
+    msg << "\n      Which is: " << lhs_value;
+  }
+  msg << "\nTo be equal to: " << rhs_expression;
+  if (rhs_value != rhs_expression) {
+    msg << "\n      Which is: " << rhs_value;
+  }
+
+  if (ignoring_case) {
+    msg << "\nIgnoring case";
+  }
+
+  if (!lhs_value.empty() && !rhs_value.empty()) {
+    const std::vector<std::string> lhs_lines =
+        SplitEscapedString(lhs_value);
+    const std::vector<std::string> rhs_lines =
+        SplitEscapedString(rhs_value);
+    if (lhs_lines.size() > 1 || rhs_lines.size() > 1) {
+      msg << "\nWith diff:\n"
+          << edit_distance::CreateUnifiedDiff(lhs_lines, rhs_lines);
+    }
+  }
+
+  return AssertionFailure() << msg;
+}
+
+// Constructs a failure message for Boolean assertions such as EXPECT_TRUE.
+std::string GetBoolAssertionFailureMessage(
+    const AssertionResult& assertion_result,
+    const char* expression_text,
+    const char* actual_predicate_value,
+    const char* expected_predicate_value) {
+  const char* actual_message = assertion_result.message();
+  Message msg;
+  msg << "Value of: " << expression_text
+      << "\n  Actual: " << actual_predicate_value;
+  if (actual_message[0] != '\0')
+    msg << " (" << actual_message << ")";
+  msg << "\nExpected: " << expected_predicate_value;
+  return msg.GetString();
+}
+
+// Helper function for implementing ASSERT_NEAR.
+AssertionResult DoubleNearPredFormat(const char* expr1,
+                                     const char* expr2,
+                                     const char* abs_error_expr,
+                                     double val1,
+                                     double val2,
+                                     double abs_error) {
+  const double diff = fabs(val1 - val2);
+  if (diff <= abs_error) return AssertionSuccess();
+
+  // TODO(wan): do not print the value of an expression if it's
+  // already a literal.
+  return AssertionFailure()
+      << "The difference between " << expr1 << " and " << expr2
+      << " is " << diff << ", which exceeds " << abs_error_expr << ", where\n"
+      << expr1 << " evaluates to " << val1 << ",\n"
+      << expr2 << " evaluates to " << val2 << ", and\n"
+      << abs_error_expr << " evaluates to " << abs_error << ".";
+}
+
+
+// Helper template for implementing FloatLE() and DoubleLE().
+template <typename RawType>
+AssertionResult FloatingPointLE(const char* expr1,
+                                const char* expr2,
+                                RawType val1,
+                                RawType val2) {
+  // Returns success if val1 is less than val2,
+  if (val1 < val2) {
+    return AssertionSuccess();
+  }
+
+  // or if val1 is almost equal to val2.
+  const FloatingPoint<RawType> lhs(val1), rhs(val2);
+  if (lhs.AlmostEquals(rhs)) {
+    return AssertionSuccess();
+  }
+
+  // Note that the above two checks will both fail if either val1 or
+  // val2 is NaN, as the IEEE floating-point standard requires that
+  // any predicate involving a NaN must return false.
+
+  ::std::stringstream val1_ss;
+  val1_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+          << val1;
+
+  ::std::stringstream val2_ss;
+  val2_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+          << val2;
+
+  return AssertionFailure()
+      << "Expected: (" << expr1 << ") <= (" << expr2 << ")\n"
+      << "  Actual: " << StringStreamToString(&val1_ss) << " vs "
+      << StringStreamToString(&val2_ss);
+}
+
+}  // namespace internal
+
+// Asserts that val1 is less than, or almost equal to, val2.  Fails
+// otherwise.  In particular, it fails if either val1 or val2 is NaN.
+AssertionResult FloatLE(const char* expr1, const char* expr2,
+                        float val1, float val2) {
+  return internal::FloatingPointLE<float>(expr1, expr2, val1, val2);
+}
+
+// Asserts that val1 is less than, or almost equal to, val2.  Fails
+// otherwise.  In particular, it fails if either val1 or val2 is NaN.
+AssertionResult DoubleLE(const char* expr1, const char* expr2,
+                         double val1, double val2) {
+  return internal::FloatingPointLE<double>(expr1, expr2, val1, val2);
+}
+
+namespace internal {
+
+// The helper function for {ASSERT|EXPECT}_EQ with int or enum
+// arguments.
+AssertionResult CmpHelperEQ(const char* lhs_expression,
+                            const char* rhs_expression,
+                            BiggestInt lhs,
+                            BiggestInt rhs) {
+  if (lhs == rhs) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   FormatForComparisonFailureMessage(lhs, rhs),
+                   FormatForComparisonFailureMessage(rhs, lhs),
+                   false);
+}
+
+// A macro for implementing the helper functions needed to implement
+// ASSERT_?? and EXPECT_?? with integer or enum arguments.  It is here
+// just to avoid copy-and-paste of similar code.
+#define GTEST_IMPL_CMP_HELPER_(op_name, op)\
+AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \
+                                   BiggestInt val1, BiggestInt val2) {\
+  if (val1 op val2) {\
+    return AssertionSuccess();\
+  } else {\
+    return AssertionFailure() \
+        << "Expected: (" << expr1 << ") " #op " (" << expr2\
+        << "), actual: " << FormatForComparisonFailureMessage(val1, val2)\
+        << " vs " << FormatForComparisonFailureMessage(val2, val1);\
+  }\
+}
+
+// Implements the helper function for {ASSERT|EXPECT}_NE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(NE, !=)
+// Implements the helper function for {ASSERT|EXPECT}_LE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(LE, <=)
+// Implements the helper function for {ASSERT|EXPECT}_LT with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(LT, < )
+// Implements the helper function for {ASSERT|EXPECT}_GE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(GE, >=)
+// Implements the helper function for {ASSERT|EXPECT}_GT with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(GT, > )
+
+#undef GTEST_IMPL_CMP_HELPER_
+
+// The helper function for {ASSERT|EXPECT}_STREQ.
+AssertionResult CmpHelperSTREQ(const char* lhs_expression,
+                               const char* rhs_expression,
+                               const char* lhs,
+                               const char* rhs) {
+  if (String::CStringEquals(lhs, rhs)) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   PrintToString(lhs),
+                   PrintToString(rhs),
+                   false);
+}
+
+// The helper function for {ASSERT|EXPECT}_STRCASEEQ.
+AssertionResult CmpHelperSTRCASEEQ(const char* lhs_expression,
+                                   const char* rhs_expression,
+                                   const char* lhs,
+                                   const char* rhs) {
+  if (String::CaseInsensitiveCStringEquals(lhs, rhs)) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   PrintToString(lhs),
+                   PrintToString(rhs),
+                   true);
+}
+
+// The helper function for {ASSERT|EXPECT}_STRNE.
+AssertionResult CmpHelperSTRNE(const char* s1_expression,
+                               const char* s2_expression,
+                               const char* s1,
+                               const char* s2) {
+  if (!String::CStringEquals(s1, s2)) {
+    return AssertionSuccess();
+  } else {
+    return AssertionFailure() << "Expected: (" << s1_expression << ") != ("
+                              << s2_expression << "), actual: \""
+                              << s1 << "\" vs \"" << s2 << "\"";
+  }
+}
+
+// The helper function for {ASSERT|EXPECT}_STRCASENE.
+AssertionResult CmpHelperSTRCASENE(const char* s1_expression,
+                                   const char* s2_expression,
+                                   const char* s1,
+                                   const char* s2) {
+  if (!String::CaseInsensitiveCStringEquals(s1, s2)) {
+    return AssertionSuccess();
+  } else {
+    return AssertionFailure()
+        << "Expected: (" << s1_expression << ") != ("
+        << s2_expression << ") (ignoring case), actual: \""
+        << s1 << "\" vs \"" << s2 << "\"";
+  }
+}
+
+}  // namespace internal
+
+namespace {
+
+// Helper functions for implementing IsSubString() and IsNotSubstring().
+
+// This group of overloaded functions return true iff needle is a
+// substring of haystack.  NULL is considered a substring of itself
+// only.
+
+bool IsSubstringPred(const char* needle, const char* haystack) {
+  if (needle == NULL || haystack == NULL)
+    return needle == haystack;
+
+  return strstr(haystack, needle) != NULL;
+}
+
+bool IsSubstringPred(const wchar_t* needle, const wchar_t* haystack) {
+  if (needle == NULL || haystack == NULL)
+    return needle == haystack;
+
+  return wcsstr(haystack, needle) != NULL;
+}
+
+// StringType here can be either ::std::string or ::std::wstring.
+template <typename StringType>
+bool IsSubstringPred(const StringType& needle,
+                     const StringType& haystack) {
+  return haystack.find(needle) != StringType::npos;
+}
+
+// This function implements either IsSubstring() or IsNotSubstring(),
+// depending on the value of the expected_to_be_substring parameter.
+// StringType here can be const char*, const wchar_t*, ::std::string,
+// or ::std::wstring.
+template <typename StringType>
+AssertionResult IsSubstringImpl(
+    bool expected_to_be_substring,
+    const char* needle_expr, const char* haystack_expr,
+    const StringType& needle, const StringType& haystack) {
+  if (IsSubstringPred(needle, haystack) == expected_to_be_substring)
+    return AssertionSuccess();
+
+  const bool is_wide_string = sizeof(needle[0]) > 1;
+  const char* const begin_string_quote = is_wide_string ? "L\"" : "\"";
+  return AssertionFailure()
+      << "Value of: " << needle_expr << "\n"
+      << "  Actual: " << begin_string_quote << needle << "\"\n"
+      << "Expected: " << (expected_to_be_substring ? "" : "not ")
+      << "a substring of " << haystack_expr << "\n"
+      << "Which is: " << begin_string_quote << haystack << "\"";
+}
+
+}  // namespace
+
+// IsSubstring() and IsNotSubstring() check whether needle is a
+// substring of haystack (NULL is considered a substring of itself
+// only), and return an appropriate error message when they fail.
+
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const char* needle, const char* haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const wchar_t* needle, const wchar_t* haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const char* needle, const char* haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const wchar_t* needle, const wchar_t* haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::string& needle, const ::std::string& haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::string& needle, const ::std::string& haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+#if GTEST_HAS_STD_WSTRING
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::wstring& needle, const ::std::wstring& haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::wstring& needle, const ::std::wstring& haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+#endif  // GTEST_HAS_STD_WSTRING
+
+namespace internal {
+
+#if GTEST_OS_WINDOWS
+
+namespace {
+
+// Helper function for IsHRESULT{SuccessFailure} predicates
+AssertionResult HRESULTFailureHelper(const char* expr,
+                                     const char* expected,
+                                     long hr) {  // NOLINT
+# if GTEST_OS_WINDOWS_MOBILE
+
+  // Windows CE doesn't support FormatMessage.
+  const char error_text[] = "";
+
+# else
+
+  // Looks up the human-readable system message for the HRESULT code
+  // and since we're not passing any params to FormatMessage, we don't
+  // want inserts expanded.
+  const DWORD kFlags = FORMAT_MESSAGE_FROM_SYSTEM |
+                       FORMAT_MESSAGE_IGNORE_INSERTS;
+  const DWORD kBufSize = 4096;
+  // Gets the system's human readable message string for this HRESULT.
+  char error_text[kBufSize] = { '\0' };
+  DWORD message_length = ::FormatMessageA(kFlags,
+                                          0,  // no source, we're asking system
+                                          hr,  // the error
+                                          0,  // no line width restrictions
+                                          error_text,  // output buffer
+                                          kBufSize,  // buf size
+                                          NULL);  // no arguments for inserts
+  // Trims tailing white space (FormatMessage leaves a trailing CR-LF)
+  for (; message_length && IsSpace(error_text[message_length - 1]);
+          --message_length) {
+    error_text[message_length - 1] = '\0';
+  }
+
+# endif  // GTEST_OS_WINDOWS_MOBILE
+
+  const std::string error_hex("0x" + String::FormatHexInt(hr));
+  return ::testing::AssertionFailure()
+      << "Expected: " << expr << " " << expected << ".\n"
+      << "  Actual: " << error_hex << " " << error_text << "\n";
+}
+
+}  // namespace
+
+AssertionResult IsHRESULTSuccess(const char* expr, long hr) {  // NOLINT
+  if (SUCCEEDED(hr)) {
+    return AssertionSuccess();
+  }
+  return HRESULTFailureHelper(expr, "succeeds", hr);
+}
+
+AssertionResult IsHRESULTFailure(const char* expr, long hr) {  // NOLINT
+  if (FAILED(hr)) {
+    return AssertionSuccess();
+  }
+  return HRESULTFailureHelper(expr, "fails", hr);
+}
+
+#endif  // GTEST_OS_WINDOWS
+
+// Utility functions for encoding Unicode text (wide strings) in
+// UTF-8.
+
+// A Unicode code-point can have up to 21 bits, and is encoded in UTF-8
+// like this:
+//
+// Code-point length   Encoding
+//   0 -  7 bits       0xxxxxxx
+//   8 - 11 bits       110xxxxx 10xxxxxx
+//  12 - 16 bits       1110xxxx 10xxxxxx 10xxxxxx
+//  17 - 21 bits       11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+
+// The maximum code-point a one-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint1 = (static_cast<UInt32>(1) <<  7) - 1;
+
+// The maximum code-point a two-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint2 = (static_cast<UInt32>(1) << (5 + 6)) - 1;
+
+// The maximum code-point a three-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint3 = (static_cast<UInt32>(1) << (4 + 2*6)) - 1;
+
+// The maximum code-point a four-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint4 = (static_cast<UInt32>(1) << (3 + 3*6)) - 1;
+
+// Chops off the n lowest bits from a bit pattern.  Returns the n
+// lowest bits.  As a side effect, the original bit pattern will be
+// shifted to the right by n bits.
+inline UInt32 ChopLowBits(UInt32* bits, int n) {
+  const UInt32 low_bits = *bits & ((static_cast<UInt32>(1) << n) - 1);
+  *bits >>= n;
+  return low_bits;
+}
+
+// Converts a Unicode code point to a narrow string in UTF-8 encoding.
+// code_point parameter is of type UInt32 because wchar_t may not be
+// wide enough to contain a code point.
+// If the code_point is not a valid Unicode code point
+// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted
+// to "(Invalid Unicode 0xXXXXXXXX)".
+std::string CodePointToUtf8(UInt32 code_point) {
+  if (code_point > kMaxCodePoint4) {
+    return "(Invalid Unicode 0x" + String::FormatHexInt(code_point) + ")";
+  }
+
+  char str[5];  // Big enough for the largest valid code point.
+  if (code_point <= kMaxCodePoint1) {
+    str[1] = '\0';
+    str[0] = static_cast<char>(code_point);                          // 0xxxxxxx
+  } else if (code_point <= kMaxCodePoint2) {
+    str[2] = '\0';
+    str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[0] = static_cast<char>(0xC0 | code_point);                   // 110xxxxx
+  } else if (code_point <= kMaxCodePoint3) {
+    str[3] = '\0';
+    str[2] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[0] = static_cast<char>(0xE0 | code_point);                   // 1110xxxx
+  } else {  // code_point <= kMaxCodePoint4
+    str[4] = '\0';
+    str[3] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[2] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[0] = static_cast<char>(0xF0 | code_point);                   // 11110xxx
+  }
+  return str;
+}
+
+// The following two functions only make sense if the the system
+// uses UTF-16 for wide string encoding. All supported systems
+// with 16 bit wchar_t (Windows, Cygwin, Symbian OS) do use UTF-16.
+
+// Determines if the arguments constitute UTF-16 surrogate pair
+// and thus should be combined into a single Unicode code point
+// using CreateCodePointFromUtf16SurrogatePair.
+inline bool IsUtf16SurrogatePair(wchar_t first, wchar_t second) {
+  return sizeof(wchar_t) == 2 &&
+      (first & 0xFC00) == 0xD800 && (second & 0xFC00) == 0xDC00;
+}
+
+// Creates a Unicode code point from UTF16 surrogate pair.
+inline UInt32 CreateCodePointFromUtf16SurrogatePair(wchar_t first,
+                                                    wchar_t second) {
+  const UInt32 mask = (1 << 10) - 1;
+  return (sizeof(wchar_t) == 2) ?
+      (((first & mask) << 10) | (second & mask)) + 0x10000 :
+      // This function should not be called when the condition is
+      // false, but we provide a sensible default in case it is.
+      static_cast<UInt32>(first);
+}
+
+// Converts a wide string to a narrow string in UTF-8 encoding.
+// The wide string is assumed to have the following encoding:
+//   UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS)
+//   UTF-32 if sizeof(wchar_t) == 4 (on Linux)
+// Parameter str points to a null-terminated wide string.
+// Parameter num_chars may additionally limit the number
+// of wchar_t characters processed. -1 is used when the entire string
+// should be processed.
+// If the string contains code points that are not valid Unicode code points
+// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding
+// and contains invalid UTF-16 surrogate pairs, values in those pairs
+// will be encoded as individual Unicode characters from Basic Normal Plane.
+std::string WideStringToUtf8(const wchar_t* str, int num_chars) {
+  if (num_chars == -1)
+    num_chars = static_cast<int>(wcslen(str));
+
+  ::std::stringstream stream;
+  for (int i = 0; i < num_chars; ++i) {
+    UInt32 unicode_code_point;
+
+    if (str[i] == L'\0') {
+      break;
+    } else if (i + 1 < num_chars && IsUtf16SurrogatePair(str[i], str[i + 1])) {
+      unicode_code_point = CreateCodePointFromUtf16SurrogatePair(str[i],
+                                                                 str[i + 1]);
+      i++;
+    } else {
+      unicode_code_point = static_cast<UInt32>(str[i]);
+    }
+
+    stream << CodePointToUtf8(unicode_code_point);
+  }
+  return StringStreamToString(&stream);
+}
+
+// Converts a wide C string to an std::string using the UTF-8 encoding.
+// NULL will be converted to "(null)".
+std::string String::ShowWideCString(const wchar_t * wide_c_str) {
+  if (wide_c_str == NULL)  return "(null)";
+
+  return internal::WideStringToUtf8(wide_c_str, -1);
+}
+
+// Compares two wide C strings.  Returns true iff they have the same
+// content.
+//
+// Unlike wcscmp(), this function can handle NULL argument(s).  A NULL
+// C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::WideCStringEquals(const wchar_t * lhs, const wchar_t * rhs) {
+  if (lhs == NULL) return rhs == NULL;
+
+  if (rhs == NULL) return false;
+
+  return wcscmp(lhs, rhs) == 0;
+}
+
+// Helper function for *_STREQ on wide strings.
+AssertionResult CmpHelperSTREQ(const char* lhs_expression,
+                               const char* rhs_expression,
+                               const wchar_t* lhs,
+                               const wchar_t* rhs) {
+  if (String::WideCStringEquals(lhs, rhs)) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   PrintToString(lhs),
+                   PrintToString(rhs),
+                   false);
+}
+
+// Helper function for *_STRNE on wide strings.
+AssertionResult CmpHelperSTRNE(const char* s1_expression,
+                               const char* s2_expression,
+                               const wchar_t* s1,
+                               const wchar_t* s2) {
+  if (!String::WideCStringEquals(s1, s2)) {
+    return AssertionSuccess();
+  }
+
+  return AssertionFailure() << "Expected: (" << s1_expression << ") != ("
+                            << s2_expression << "), actual: "
+                            << PrintToString(s1)
+                            << " vs " << PrintToString(s2);
+}
+
+// Compares two C strings, ignoring case.  Returns true iff they have
+// the same content.
+//
+// Unlike strcasecmp(), this function can handle NULL argument(s).  A
+// NULL C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::CaseInsensitiveCStringEquals(const char * lhs, const char * rhs) {
+  if (lhs == NULL)
+    return rhs == NULL;
+  if (rhs == NULL)
+    return false;
+  return posix::StrCaseCmp(lhs, rhs) == 0;
+}
+
+  // Compares two wide C strings, ignoring case.  Returns true iff they
+  // have the same content.
+  //
+  // Unlike wcscasecmp(), this function can handle NULL argument(s).
+  // A NULL C string is considered different to any non-NULL wide C string,
+  // including the empty string.
+  // NB: The implementations on different platforms slightly differ.
+  // On windows, this method uses _wcsicmp which compares according to LC_CTYPE
+  // environment variable. On GNU platform this method uses wcscasecmp
+  // which compares according to LC_CTYPE category of the current locale.
+  // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the
+  // current locale.
+bool String::CaseInsensitiveWideCStringEquals(const wchar_t* lhs,
+                                              const wchar_t* rhs) {
+  if (lhs == NULL) return rhs == NULL;
+
+  if (rhs == NULL) return false;
+
+#if GTEST_OS_WINDOWS
+  return _wcsicmp(lhs, rhs) == 0;
+#elif GTEST_OS_LINUX && !GTEST_OS_LINUX_ANDROID
+  return wcscasecmp(lhs, rhs) == 0;
+#else
+  // Android, Mac OS X and Cygwin don't define wcscasecmp.
+  // Other unknown OSes may not define it either.
+  wint_t left, right;
+  do {
+    left = towlower(*lhs++);
+    right = towlower(*rhs++);
+  } while (left && left == right);
+  return left == right;
+#endif  // OS selector
+}
+
+// Returns true iff str ends with the given suffix, ignoring case.
+// Any string is considered to end with an empty suffix.
+bool String::EndsWithCaseInsensitive(
+    const std::string& str, const std::string& suffix) {
+  const size_t str_len = str.length();
+  const size_t suffix_len = suffix.length();
+  return (str_len >= suffix_len) &&
+         CaseInsensitiveCStringEquals(str.c_str() + str_len - suffix_len,
+                                      suffix.c_str());
+}
+
+// Formats an int value as "%02d".
+std::string String::FormatIntWidth2(int value) {
+  std::stringstream ss;
+  ss << std::setfill('0') << std::setw(2) << value;
+  return ss.str();
+}
+
+// Formats an int value as "%X".
+std::string String::FormatHexInt(int value) {
+  std::stringstream ss;
+  ss << std::hex << std::uppercase << value;
+  return ss.str();
+}
+
+// Formats a byte as "%02X".
+std::string String::FormatByte(unsigned char value) {
+  std::stringstream ss;
+  ss << std::setfill('0') << std::setw(2) << std::hex << std::uppercase
+     << static_cast<unsigned int>(value);
+  return ss.str();
+}
+
+// Converts the buffer in a stringstream to an std::string, converting NUL
+// bytes to "\\0" along the way.
+std::string StringStreamToString(::std::stringstream* ss) {
+  const ::std::string& str = ss->str();
+  const char* const start = str.c_str();
+  const char* const end = start + str.length();
+
+  std::string result;
+  result.reserve(2 * (end - start));
+  for (const char* ch = start; ch != end; ++ch) {
+    if (*ch == '\0') {
+      result += "\\0";  // Replaces NUL with "\\0";
+    } else {
+      result += *ch;
+    }
+  }
+
+  return result;
+}
+
+// Appends the user-supplied message to the Google-Test-generated message.
+std::string AppendUserMessage(const std::string& gtest_msg,
+                              const Message& user_msg) {
+  // Appends the user message if it's non-empty.
+  const std::string user_msg_string = user_msg.GetString();
+  if (user_msg_string.empty()) {
+    return gtest_msg;
+  }
+
+  return gtest_msg + "\n" + user_msg_string;
+}
+
+}  // namespace internal
+
+// class TestResult
+
+// Creates an empty TestResult.
+TestResult::TestResult()
+    : death_test_count_(0),
+      elapsed_time_(0) {
+}
+
+// D'tor.
+TestResult::~TestResult() {
+}
+
+// Returns the i-th test part result among all the results. i can
+// range from 0 to total_part_count() - 1. If i is not in that range,
+// aborts the program.
+const TestPartResult& TestResult::GetTestPartResult(int i) const {
+  if (i < 0 || i >= total_part_count())
+    internal::posix::Abort();
+  return test_part_results_.at(i);
+}
+
+// Returns the i-th test property. i can range from 0 to
+// test_property_count() - 1. If i is not in that range, aborts the
+// program.
+const TestProperty& TestResult::GetTestProperty(int i) const {
+  if (i < 0 || i >= test_property_count())
+    internal::posix::Abort();
+  return test_properties_.at(i);
+}
+
+// Clears the test part results.
+void TestResult::ClearTestPartResults() {
+  test_part_results_.clear();
+}
+
+// Adds a test part result to the list.
+void TestResult::AddTestPartResult(const TestPartResult& test_part_result) {
+  test_part_results_.push_back(test_part_result);
+}
+
+// Adds a test property to the list. If a property with the same key as the
+// supplied property is already represented, the value of this test_property
+// replaces the old value for that key.
+void TestResult::RecordProperty(const std::string& xml_element,
+                                const TestProperty& test_property) {
+  if (!ValidateTestProperty(xml_element, test_property)) {
+    return;
+  }
+  internal::MutexLock lock(&test_properites_mutex_);
+  const std::vector<TestProperty>::iterator property_with_matching_key =
+      std::find_if(test_properties_.begin(), test_properties_.end(),
+                   internal::TestPropertyKeyIs(test_property.key()));
+  if (property_with_matching_key == test_properties_.end()) {
+    test_properties_.push_back(test_property);
+    return;
+  }
+  property_with_matching_key->SetValue(test_property.value());
+}
+
+// The list of reserved attributes used in the <testsuites> element of XML
+// output.
+static const char* const kReservedTestSuitesAttributes[] = {
+  "disabled",
+  "errors",
+  "failures",
+  "name",
+  "random_seed",
+  "tests",
+  "time",
+  "timestamp"
+};
+
+// The list of reserved attributes used in the <testsuite> element of XML
+// output.
+static const char* const kReservedTestSuiteAttributes[] = {
+  "disabled",
+  "errors",
+  "failures",
+  "name",
+  "tests",
+  "time"
+};
+
+// The list of reserved attributes used in the <testcase> element of XML output.
+static const char* const kReservedTestCaseAttributes[] = {
+  "classname",
+  "name",
+  "status",
+  "time",
+  "type_param",
+  "value_param"
+};
+
+template <int kSize>
+std::vector<std::string> ArrayAsVector(const char* const (&array)[kSize]) {
+  return std::vector<std::string>(array, array + kSize);
+}
+
+static std::vector<std::string> GetReservedAttributesForElement(
+    const std::string& xml_element) {
+  if (xml_element == "testsuites") {
+    return ArrayAsVector(kReservedTestSuitesAttributes);
+  } else if (xml_element == "testsuite") {
+    return ArrayAsVector(kReservedTestSuiteAttributes);
+  } else if (xml_element == "testcase") {
+    return ArrayAsVector(kReservedTestCaseAttributes);
+  } else {
+    GTEST_CHECK_(false) << "Unrecognized xml_element provided: " << xml_element;
+  }
+  // This code is unreachable but some compilers may not realizes that.
+  return std::vector<std::string>();
+}
+
+static std::string FormatWordList(const std::vector<std::string>& words) {
+  Message word_list;
+  for (size_t i = 0; i < words.size(); ++i) {
+    if (i > 0 && words.size() > 2) {
+      word_list << ", ";
+    }
+    if (i == words.size() - 1) {
+      word_list << "and ";
+    }
+    word_list << "'" << words[i] << "'";
+  }
+  return word_list.GetString();
+}
+
+bool ValidateTestPropertyName(const std::string& property_name,
+                              const std::vector<std::string>& reserved_names) {
+  if (std::find(reserved_names.begin(), reserved_names.end(), property_name) !=
+          reserved_names.end()) {
+    ADD_FAILURE() << "Reserved key used in RecordProperty(): " << property_name
+                  << " (" << FormatWordList(reserved_names)
+                  << " are reserved by " << GTEST_NAME_ << ")";
+    return false;
+  }
+  return true;
+}
+
+// Adds a failure if the key is a reserved attribute of the element named
+// xml_element.  Returns true if the property is valid.
+bool TestResult::ValidateTestProperty(const std::string& xml_element,
+                                      const TestProperty& test_property) {
+  return ValidateTestPropertyName(test_property.key(),
+                                  GetReservedAttributesForElement(xml_element));
+}
+
+// Clears the object.
+void TestResult::Clear() {
+  test_part_results_.clear();
+  test_properties_.clear();
+  death_test_count_ = 0;
+  elapsed_time_ = 0;
+}
+
+// Returns true off the test part was skipped.
+static bool TestPartSkipped(const TestPartResult& result) {
+  return result.skipped();
+}
+
+// Returns true if and only if the test was skipped.
+bool TestResult::Skipped() const {
+  return !Failed() && CountIf(test_part_results_, TestPartSkipped) > 0;
+}
+
+// Returns true iff the test failed.
+bool TestResult::Failed() const {
+  for (int i = 0; i < total_part_count(); ++i) {
+    if (GetTestPartResult(i).failed())
+      return true;
+  }
+  return false;
+}
+
+// Returns true iff the test part fatally failed.
+static bool TestPartFatallyFailed(const TestPartResult& result) {
+  return result.fatally_failed();
+}
+
+// Returns true iff the test fatally failed.
+bool TestResult::HasFatalFailure() const {
+  return CountIf(test_part_results_, TestPartFatallyFailed) > 0;
+}
+
+// Returns true iff the test part non-fatally failed.
+static bool TestPartNonfatallyFailed(const TestPartResult& result) {
+  return result.nonfatally_failed();
+}
+
+// Returns true iff the test has a non-fatal failure.
+bool TestResult::HasNonfatalFailure() const {
+  return CountIf(test_part_results_, TestPartNonfatallyFailed) > 0;
+}
+
+// Gets the number of all test parts.  This is the sum of the number
+// of successful test parts and the number of failed test parts.
+int TestResult::total_part_count() const {
+  return static_cast<int>(test_part_results_.size());
+}
+
+// Returns the number of the test properties.
+int TestResult::test_property_count() const {
+  return static_cast<int>(test_properties_.size());
+}
+
+// class Test
+
+// Creates a Test object.
+
+// The c'tor saves the states of all flags.
+Test::Test()
+    : gtest_flag_saver_(new GTEST_FLAG_SAVER_) {
+}
+
+// The d'tor restores the states of all flags.  The actual work is
+// done by the d'tor of the gtest_flag_saver_ field, and thus not
+// visible here.
+Test::~Test() {
+}
+
+// Sets up the test fixture.
+//
+// A sub-class may override this.
+void Test::SetUp() {
+}
+
+// Tears down the test fixture.
+//
+// A sub-class may override this.
+void Test::TearDown() {
+}
+
+// Allows user supplied key value pairs to be recorded for later output.
+void Test::RecordProperty(const std::string& key, const std::string& value) {
+  UnitTest::GetInstance()->RecordProperty(key, value);
+}
+
+// Allows user supplied key value pairs to be recorded for later output.
+void Test::RecordProperty(const std::string& key, int value) {
+  Message value_message;
+  value_message << value;
+  RecordProperty(key, value_message.GetString().c_str());
+}
+
+namespace internal {
+
+void ReportFailureInUnknownLocation(TestPartResult::Type result_type,
+                                    const std::string& message) {
+  // This function is a friend of UnitTest and as such has access to
+  // AddTestPartResult.
+  UnitTest::GetInstance()->AddTestPartResult(
+      result_type,
+      NULL,  // No info about the source file where the exception occurred.
+      -1,    // We have no info on which line caused the exception.
+      message,
+      "");   // No stack trace, either.
+}
+
+}  // namespace internal
+
+// Google Test requires all tests in the same test case to use the same test
+// fixture class.  This function checks if the current test has the
+// same fixture class as the first test in the current test case.  If
+// yes, it returns true; otherwise it generates a Google Test failure and
+// returns false.
+bool Test::HasSameFixtureClass() {
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  const TestCase* const test_case = impl->current_test_case();
+
+  // Info about the first test in the current test case.
+  const TestInfo* const first_test_info = test_case->test_info_list()[0];
+  const internal::TypeId first_fixture_id = first_test_info->fixture_class_id_;
+  const char* const first_test_name = first_test_info->name();
+
+  // Info about the current test.
+  const TestInfo* const this_test_info = impl->current_test_info();
+  const internal::TypeId this_fixture_id = this_test_info->fixture_class_id_;
+  const char* const this_test_name = this_test_info->name();
+
+  if (this_fixture_id != first_fixture_id) {
+    // Is the first test defined using TEST?
+    const bool first_is_TEST = first_fixture_id == internal::GetTestTypeId();
+    // Is this test defined using TEST?
+    const bool this_is_TEST = this_fixture_id == internal::GetTestTypeId();
+
+    if (first_is_TEST || this_is_TEST) {
+      // Both TEST and TEST_F appear in same test case, which is incorrect.
+      // Tell the user how to fix this.
+
+      // Gets the name of the TEST and the name of the TEST_F.  Note
+      // that first_is_TEST and this_is_TEST cannot both be true, as
+      // the fixture IDs are different for the two tests.
+      const char* const TEST_name =
+          first_is_TEST ? first_test_name : this_test_name;
+      const char* const TEST_F_name =
+          first_is_TEST ? this_test_name : first_test_name;
+
+      ADD_FAILURE()
+          << "All tests in the same test case must use the same test fixture\n"
+          << "class, so mixing TEST_F and TEST in the same test case is\n"
+          << "illegal.  In test case " << this_test_info->test_case_name()
+          << ",\n"
+          << "test " << TEST_F_name << " is defined using TEST_F but\n"
+          << "test " << TEST_name << " is defined using TEST.  You probably\n"
+          << "want to change the TEST to TEST_F or move it to another test\n"
+          << "case.";
+    } else {
+      // Two fixture classes with the same name appear in two different
+      // namespaces, which is not allowed. Tell the user how to fix this.
+      ADD_FAILURE()
+          << "All tests in the same test case must use the same test fixture\n"
+          << "class.  However, in test case "
+          << this_test_info->test_case_name() << ",\n"
+          << "you defined test " << first_test_name
+          << " and test " << this_test_name << "\n"
+          << "using two different test fixture classes.  This can happen if\n"
+          << "the two classes are from different namespaces or translation\n"
+          << "units and have the same name.  You should probably rename one\n"
+          << "of the classes to put the tests into different test cases.";
+    }
+    return false;
+  }
+
+  return true;
+}
+
+#if GTEST_HAS_SEH
+
+// Adds an "exception thrown" fatal failure to the current test.  This
+// function returns its result via an output parameter pointer because VC++
+// prohibits creation of objects with destructors on stack in functions
+// using __try (see error C2712).
+static std::string* FormatSehExceptionMessage(DWORD exception_code,
+                                              const char* location) {
+  Message message;
+  message << "SEH exception with code 0x" << std::setbase(16) <<
+    exception_code << std::setbase(10) << " thrown in " << location << ".";
+
+  return new std::string(message.GetString());
+}
+
+#endif  // GTEST_HAS_SEH
+
+namespace internal {
+
+#if GTEST_HAS_EXCEPTIONS
+
+// Adds an "exception thrown" fatal failure to the current test.
+static std::string FormatCxxExceptionMessage(const char* description,
+                                             const char* location) {
+  Message message;
+  if (description != NULL) {
+    message << "C++ exception with description \"" << description << "\"";
+  } else {
+    message << "Unknown C++ exception";
+  }
+  message << " thrown in " << location << ".";
+
+  return message.GetString();
+}
+
+static std::string PrintTestPartResultToString(
+    const TestPartResult& test_part_result);
+
+GoogleTestFailureException::GoogleTestFailureException(
+    const TestPartResult& failure)
+    : ::std::runtime_error(PrintTestPartResultToString(failure).c_str()) {}
+
+#endif  // GTEST_HAS_EXCEPTIONS
+
+// We put these helper functions in the internal namespace as IBM's xlC
+// compiler rejects the code if they were declared static.
+
+// Runs the given method and handles SEH exceptions it throws, when
+// SEH is supported; returns the 0-value for type Result in case of an
+// SEH exception.  (Microsoft compilers cannot handle SEH and C++
+// exceptions in the same function.  Therefore, we provide a separate
+// wrapper function for handling SEH exceptions.)
+template <class T, typename Result>
+Result HandleSehExceptionsInMethodIfSupported(
+    T* object, Result (T::*method)(), const char* location) {
+#if GTEST_HAS_SEH
+  __try {
+    return (object->*method)();
+  } __except (internal::UnitTestOptions::GTestShouldProcessSEH(  // NOLINT
+      GetExceptionCode())) {
+    // We create the exception message on the heap because VC++ prohibits
+    // creation of objects with destructors on stack in functions using __try
+    // (see error C2712).
+    std::string* exception_message = FormatSehExceptionMessage(
+        GetExceptionCode(), location);
+    internal::ReportFailureInUnknownLocation(TestPartResult::kFatalFailure,
+                                             *exception_message);
+    delete exception_message;
+    return static_cast<Result>(0);
+  }
+#else
+  (void)location;
+  return (object->*method)();
+#endif  // GTEST_HAS_SEH
+}
+
+// Runs the given method and catches and reports C++ and/or SEH-style
+// exceptions, if they are supported; returns the 0-value for type
+// Result in case of an SEH exception.
+template <class T, typename Result>
+Result HandleExceptionsInMethodIfSupported(
+    T* object, Result (T::*method)(), const char* location) {
+  // NOTE: The user code can affect the way in which Google Test handles
+  // exceptions by setting GTEST_FLAG(catch_exceptions), but only before
+  // RUN_ALL_TESTS() starts. It is technically possible to check the flag
+  // after the exception is caught and either report or re-throw the
+  // exception based on the flag's value:
+  //
+  // try {
+  //   // Perform the test method.
+  // } catch (...) {
+  //   if (GTEST_FLAG(catch_exceptions))
+  //     // Report the exception as failure.
+  //   else
+  //     throw;  // Re-throws the original exception.
+  // }
+  //
+  // However, the purpose of this flag is to allow the program to drop into
+  // the debugger when the exception is thrown. On most platforms, once the
+  // control enters the catch block, the exception origin information is
+  // lost and the debugger will stop the program at the point of the
+  // re-throw in this function -- instead of at the point of the original
+  // throw statement in the code under test.  For this reason, we perform
+  // the check early, sacrificing the ability to affect Google Test's
+  // exception handling in the method where the exception is thrown.
+  if (internal::GetUnitTestImpl()->catch_exceptions()) {
+#if GTEST_HAS_EXCEPTIONS
+    try {
+      return HandleSehExceptionsInMethodIfSupported(object, method, location);
+    } catch (const internal::GoogleTestFailureException&) {  // NOLINT
+      // This exception type can only be thrown by a failed Google
+      // Test assertion with the intention of letting another testing
+      // framework catch it.  Therefore we just re-throw it.
+      throw;
+    } catch (const std::exception& e) {  // NOLINT
+      internal::ReportFailureInUnknownLocation(
+          TestPartResult::kFatalFailure,
+          FormatCxxExceptionMessage(e.what(), location));
+    } catch (...) {  // NOLINT
+      internal::ReportFailureInUnknownLocation(
+          TestPartResult::kFatalFailure,
+          FormatCxxExceptionMessage(NULL, location));
+    }
+    return static_cast<Result>(0);
+#else
+    return HandleSehExceptionsInMethodIfSupported(object, method, location);
+#endif  // GTEST_HAS_EXCEPTIONS
+  } else {
+    return (object->*method)();
+  }
+}
+
+}  // namespace internal
+
+// Runs the test and updates the test result.
+void Test::Run() {
+  if (!HasSameFixtureClass()) return;
+
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(this, &Test::SetUp, "SetUp()");
+  // We will run the test only if SetUp() was successful and didn't call
+  // GTEST_SKIP.
+  if (!HasFatalFailure() && !IsSkipped()) {
+    impl->os_stack_trace_getter()->UponLeavingGTest();
+    internal::HandleExceptionsInMethodIfSupported(
+        this, &Test::TestBody, "the test body");
+  }
+
+  // However, we want to clean up as much as possible.  Hence we will
+  // always call TearDown(), even if SetUp() or the test body has
+  // failed.
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      this, &Test::TearDown, "TearDown()");
+}
+
+// Returns true iff the current test has a fatal failure.
+bool Test::HasFatalFailure() {
+  return internal::GetUnitTestImpl()->current_test_result()->HasFatalFailure();
+}
+
+// Returns true iff the current test has a non-fatal failure.
+bool Test::HasNonfatalFailure() {
+  return internal::GetUnitTestImpl()->current_test_result()->
+      HasNonfatalFailure();
+}
+
+// Returns true if and only if the current test was skipped.
+bool Test::IsSkipped() {
+  return internal::GetUnitTestImpl()->current_test_result()->Skipped();
+}
+
+// class TestInfo
+
+// Constructs a TestInfo object. It assumes ownership of the test factory
+// object.
+TestInfo::TestInfo(const std::string& a_test_case_name,
+                   const std::string& a_name,
+                   const char* a_type_param,
+                   const char* a_value_param,
+                   internal::CodeLocation a_code_location,
+                   internal::TypeId fixture_class_id,
+                   internal::TestFactoryBase* factory)
+    : test_case_name_(a_test_case_name),
+      name_(a_name),
+      type_param_(a_type_param ? new std::string(a_type_param) : NULL),
+      value_param_(a_value_param ? new std::string(a_value_param) : NULL),
+      location_(a_code_location),
+      fixture_class_id_(fixture_class_id),
+      should_run_(false),
+      is_disabled_(false),
+      matches_filter_(false),
+      factory_(factory),
+      result_() {}
+
+// Destructs a TestInfo object.
+TestInfo::~TestInfo() { delete factory_; }
+
+namespace internal {
+
+// Creates a new TestInfo object and registers it with Google Test;
+// returns the created object.
+//
+// Arguments:
+//
+//   test_case_name:   name of the test case
+//   name:             name of the test
+//   type_param:       the name of the test's type parameter, or NULL if
+//                     this is not a typed or a type-parameterized test.
+//   value_param:      text representation of the test's value parameter,
+//                     or NULL if this is not a value-parameterized test.
+//   code_location:    code location where the test is defined
+//   fixture_class_id: ID of the test fixture class
+//   set_up_tc:        pointer to the function that sets up the test case
+//   tear_down_tc:     pointer to the function that tears down the test case
+//   factory:          pointer to the factory that creates a test object.
+//                     The newly created TestInfo instance will assume
+//                     ownership of the factory object.
+TestInfo* MakeAndRegisterTestInfo(
+    const char* test_case_name,
+    const char* name,
+    const char* type_param,
+    const char* value_param,
+    CodeLocation code_location,
+    TypeId fixture_class_id,
+    SetUpTestCaseFunc set_up_tc,
+    TearDownTestCaseFunc tear_down_tc,
+    TestFactoryBase* factory) {
+  TestInfo* const test_info =
+      new TestInfo(test_case_name, name, type_param, value_param,
+                   code_location, fixture_class_id, factory);
+  GetUnitTestImpl()->AddTestInfo(set_up_tc, tear_down_tc, test_info);
+  return test_info;
+}
+
+#if GTEST_HAS_PARAM_TEST
+void ReportInvalidTestCaseType(const char* test_case_name,
+                               CodeLocation code_location) {
+  Message errors;
+  errors
+      << "Attempted redefinition of test case " << test_case_name << ".\n"
+      << "All tests in the same test case must use the same test fixture\n"
+      << "class.  However, in test case " << test_case_name << ", you tried\n"
+      << "to define a test using a fixture class different from the one\n"
+      << "used earlier. This can happen if the two fixture classes are\n"
+      << "from different namespaces and have the same name. You should\n"
+      << "probably rename one of the classes to put the tests into different\n"
+      << "test cases.";
+
+  fprintf(stderr, "%s %s",
+          FormatFileLocation(code_location.file.c_str(),
+                             code_location.line).c_str(),
+          errors.GetString().c_str());
+}
+#endif  // GTEST_HAS_PARAM_TEST
+
+}  // namespace internal
+
+namespace {
+
+// A predicate that checks the test name of a TestInfo against a known
+// value.
+//
+// This is used for implementation of the TestCase class only.  We put
+// it in the anonymous namespace to prevent polluting the outer
+// namespace.
+//
+// TestNameIs is copyable.
+class TestNameIs {
+ public:
+  // Constructor.
+  //
+  // TestNameIs has NO default constructor.
+  explicit TestNameIs(const char* name)
+      : name_(name) {}
+
+  // Returns true iff the test name of test_info matches name_.
+  bool operator()(const TestInfo * test_info) const {
+    return test_info && test_info->name() == name_;
+  }
+
+ private:
+  std::string name_;
+};
+
+}  // namespace
+
+namespace internal {
+
+// This method expands all parameterized tests registered with macros TEST_P
+// and INSTANTIATE_TEST_CASE_P into regular tests and registers those.
+// This will be done just once during the program runtime.
+void UnitTestImpl::RegisterParameterizedTests() {
+#if GTEST_HAS_PARAM_TEST
+  if (!parameterized_tests_registered_) {
+    parameterized_test_registry_.RegisterTests();
+    parameterized_tests_registered_ = true;
+  }
+#endif
+}
+
+}  // namespace internal
+
+// Creates the test object, runs it, records its result, and then
+// deletes it.
+void TestInfo::Run() {
+  if (!should_run_) return;
+
+  // Tells UnitTest where to store test result.
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  impl->set_current_test_info(this);
+
+  TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater();
+
+  // Notifies the unit test event listeners that a test is about to start.
+  repeater->OnTestStart(*this);
+
+  const TimeInMillis start = internal::GetTimeInMillis();
+
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+
+  // Creates the test object.
+  Test* const test = internal::HandleExceptionsInMethodIfSupported(
+      factory_, &internal::TestFactoryBase::CreateTest,
+      "the test fixture's constructor");
+
+  // Runs the test only if the test object was created and its
+  // constructor didn't generate a fatal failure or invoke GTEST_SKIP().
+  if ((test != NULL) && !Test::HasFatalFailure() && !Test::IsSkipped()) {
+    // This doesn't throw as all user code that can throw are wrapped into
+    // exception handling code.
+    test->Run();
+  }
+
+  // Deletes the test object.
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      test, &Test::DeleteSelf_, "the test fixture's destructor");
+
+  result_.set_elapsed_time(internal::GetTimeInMillis() - start);
+
+  // Notifies the unit test event listener that a test has just finished.
+  repeater->OnTestEnd(*this);
+
+  // Tells UnitTest to stop associating assertion results to this
+  // test.
+  impl->set_current_test_info(NULL);
+}
+
+// class TestCase
+
+// Gets the number of successful tests in this test case.
+int TestCase::successful_test_count() const {
+  return CountIf(test_info_list_, TestPassed);
+}
+
+// Gets the number of successful tests in this test suite.
+int TestCase::skipped_test_count() const {
+  return CountIf(test_info_list_, TestSkipped);
+}
+
+// Gets the number of failed tests in this test case.
+int TestCase::failed_test_count() const {
+  return CountIf(test_info_list_, TestFailed);
+}
+
+// Gets the number of disabled tests that will be reported in the XML report.
+int TestCase::reportable_disabled_test_count() const {
+  return CountIf(test_info_list_, TestReportableDisabled);
+}
+
+// Gets the number of disabled tests in this test case.
+int TestCase::disabled_test_count() const {
+  return CountIf(test_info_list_, TestDisabled);
+}
+
+// Gets the number of tests to be printed in the XML report.
+int TestCase::reportable_test_count() const {
+  return CountIf(test_info_list_, TestReportable);
+}
+
+// Get the number of tests in this test case that should run.
+int TestCase::test_to_run_count() const {
+  return CountIf(test_info_list_, ShouldRunTest);
+}
+
+// Gets the number of all tests.
+int TestCase::total_test_count() const {
+  return static_cast<int>(test_info_list_.size());
+}
+
+// Creates a TestCase with the given name.
+//
+// Arguments:
+//
+//   name:         name of the test case
+//   a_type_param: the name of the test case's type parameter, or NULL if
+//                 this is not a typed or a type-parameterized test case.
+//   set_up_tc:    pointer to the function that sets up the test case
+//   tear_down_tc: pointer to the function that tears down the test case
+TestCase::TestCase(const char* a_name, const char* a_type_param,
+                   Test::SetUpTestCaseFunc set_up_tc,
+                   Test::TearDownTestCaseFunc tear_down_tc)
+    : name_(a_name),
+      type_param_(a_type_param ? new std::string(a_type_param) : NULL),
+      set_up_tc_(set_up_tc),
+      tear_down_tc_(tear_down_tc),
+      should_run_(false),
+      elapsed_time_(0) {
+}
+
+// Destructor of TestCase.
+TestCase::~TestCase() {
+  // Deletes every Test in the collection.
+  ForEach(test_info_list_, internal::Delete<TestInfo>);
+}
+
+// Returns the i-th test among all the tests. i can range from 0 to
+// total_test_count() - 1. If i is not in that range, returns NULL.
+const TestInfo* TestCase::GetTestInfo(int i) const {
+  const int index = GetElementOr(test_indices_, i, -1);
+  return index < 0 ? NULL : test_info_list_[index];
+}
+
+// Returns the i-th test among all the tests. i can range from 0 to
+// total_test_count() - 1. If i is not in that range, returns NULL.
+TestInfo* TestCase::GetMutableTestInfo(int i) {
+  const int index = GetElementOr(test_indices_, i, -1);
+  return index < 0 ? NULL : test_info_list_[index];
+}
+
+// Adds a test to this test case.  Will delete the test upon
+// destruction of the TestCase object.
+void TestCase::AddTestInfo(TestInfo * test_info) {
+  test_info_list_.push_back(test_info);
+  test_indices_.push_back(static_cast<int>(test_indices_.size()));
+}
+
+// Runs every test in this TestCase.
+void TestCase::Run() {
+  if (!should_run_) return;
+
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  impl->set_current_test_case(this);
+
+  TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater();
+
+  repeater->OnTestCaseStart(*this);
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      this, &TestCase::RunSetUpTestCase, "SetUpTestCase()");
+
+  const internal::TimeInMillis start = internal::GetTimeInMillis();
+  for (int i = 0; i < total_test_count(); i++) {
+    GetMutableTestInfo(i)->Run();
+  }
+  elapsed_time_ = internal::GetTimeInMillis() - start;
+
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      this, &TestCase::RunTearDownTestCase, "TearDownTestCase()");
+
+  repeater->OnTestCaseEnd(*this);
+  impl->set_current_test_case(NULL);
+}
+
+// Clears the results of all tests in this test case.
+void TestCase::ClearResult() {
+  ad_hoc_test_result_.Clear();
+  ForEach(test_info_list_, TestInfo::ClearTestResult);
+}
+
+// Shuffles the tests in this test case.
+void TestCase::ShuffleTests(internal::Random* random) {
+  Shuffle(random, &test_indices_);
+}
+
+// Restores the test order to before the first shuffle.
+void TestCase::UnshuffleTests() {
+  for (size_t i = 0; i < test_indices_.size(); i++) {
+    test_indices_[i] = static_cast<int>(i);
+  }
+}
+
+// Formats a countable noun.  Depending on its quantity, either the
+// singular form or the plural form is used. e.g.
+//
+// FormatCountableNoun(1, "formula", "formuli") returns "1 formula".
+// FormatCountableNoun(5, "book", "books") returns "5 books".
+static std::string FormatCountableNoun(int count,
+                                       const char * singular_form,
+                                       const char * plural_form) {
+  return internal::StreamableToString(count) + " " +
+      (count == 1 ? singular_form : plural_form);
+}
+
+// Formats the count of tests.
+static std::string FormatTestCount(int test_count) {
+  return FormatCountableNoun(test_count, "test", "tests");
+}
+
+// Formats the count of test cases.
+static std::string FormatTestCaseCount(int test_case_count) {
+  return FormatCountableNoun(test_case_count, "test case", "test cases");
+}
+
+// Converts a TestPartResult::Type enum to human-friendly string
+// representation.  Both kNonFatalFailure and kFatalFailure are translated
+// to "Failure", as the user usually doesn't care about the difference
+// between the two when viewing the test result.
+static const char * TestPartResultTypeToString(TestPartResult::Type type) {
+  switch (type) {
+    case TestPartResult::kSkip:
+      return "Skipped";
+    case TestPartResult::kSuccess:
+      return "Success";
+
+    case TestPartResult::kNonFatalFailure:
+    case TestPartResult::kFatalFailure:
+#ifdef _MSC_VER
+      return "error: ";
+#else
+      return "Failure\n";
+#endif
+    default:
+      return "Unknown result type";
+  }
+}
+
+namespace internal {
+
+// Prints a TestPartResult to an std::string.
+static std::string PrintTestPartResultToString(
+    const TestPartResult& test_part_result) {
+  return (Message()
+          << internal::FormatFileLocation(test_part_result.file_name(),
+                                          test_part_result.line_number())
+          << " " << TestPartResultTypeToString(test_part_result.type())
+          << test_part_result.message()).GetString();
+}
+
+// Prints a TestPartResult.
+static void PrintTestPartResult(const TestPartResult& test_part_result) {
+  const std::string& result =
+      PrintTestPartResultToString(test_part_result);
+  printf("%s\n", result.c_str());
+  fflush(stdout);
+  // If the test program runs in Visual Studio or a debugger, the
+  // following statements add the test part result message to the Output
+  // window such that the user can double-click on it to jump to the
+  // corresponding source code location; otherwise they do nothing.
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+  // We don't call OutputDebugString*() on Windows Mobile, as printing
+  // to stdout is done by OutputDebugString() there already - we don't
+  // want the same message printed twice.
+  ::OutputDebugStringA(result.c_str());
+  ::OutputDebugStringA("\n");
+#endif
+}
+
+// class PrettyUnitTestResultPrinter
+
+enum GTestColor {
+  COLOR_DEFAULT,
+  COLOR_RED,
+  COLOR_GREEN,
+  COLOR_YELLOW
+};
+
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \
+    !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+
+// Returns the character attribute for the given color.
+WORD GetColorAttribute(GTestColor color) {
+  switch (color) {
+    case COLOR_RED:    return FOREGROUND_RED;
+    case COLOR_GREEN:  return FOREGROUND_GREEN;
+    case COLOR_YELLOW: return FOREGROUND_RED | FOREGROUND_GREEN;
+    default:           return 0;
+  }
+}
+
+#else
+
+// Returns the ANSI color code for the given color.  COLOR_DEFAULT is
+// an invalid input.
+const char* GetAnsiColorCode(GTestColor color) {
+  switch (color) {
+    case COLOR_RED:     return "1";
+    case COLOR_GREEN:   return "2";
+    case COLOR_YELLOW:  return "3";
+    default:            return NULL;
+  };
+}
+
+#endif  // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+
+// Returns true iff Google Test should use colors in the output.
+bool ShouldUseColor(bool stdout_is_tty) {
+  const char* const gtest_color = GTEST_FLAG(color).c_str();
+
+  if (String::CaseInsensitiveCStringEquals(gtest_color, "auto")) {
+#if GTEST_OS_WINDOWS
+    // On Windows the TERM variable is usually not set, but the
+    // console there does support colors.
+    return stdout_is_tty;
+#else
+    // On non-Windows platforms, we rely on the TERM variable.
+    const char* const term = posix::GetEnv("TERM");
+    const bool term_supports_color =
+        String::CStringEquals(term, "xterm") ||
+        String::CStringEquals(term, "xterm-color") ||
+        String::CStringEquals(term, "xterm-256color") ||
+        String::CStringEquals(term, "screen") ||
+        String::CStringEquals(term, "screen-256color") ||
+        String::CStringEquals(term, "tmux") ||
+        String::CStringEquals(term, "tmux-256color") ||
+        String::CStringEquals(term, "rxvt-unicode") ||
+        String::CStringEquals(term, "rxvt-unicode-256color") ||
+        String::CStringEquals(term, "linux") ||
+        String::CStringEquals(term, "cygwin");
+    return stdout_is_tty && term_supports_color;
+#endif  // GTEST_OS_WINDOWS
+  }
+
+  return String::CaseInsensitiveCStringEquals(gtest_color, "yes") ||
+      String::CaseInsensitiveCStringEquals(gtest_color, "true") ||
+      String::CaseInsensitiveCStringEquals(gtest_color, "t") ||
+      String::CStringEquals(gtest_color, "1");
+  // We take "yes", "true", "t", and "1" as meaning "yes".  If the
+  // value is neither one of these nor "auto", we treat it as "no" to
+  // be conservative.
+}
+
+// Helpers for printing colored strings to stdout. Note that on Windows, we
+// cannot simply emit special characters and have the terminal change colors.
+// This routine must actually emit the characters rather than return a string
+// that would be colored when printed, as can be done on Linux.
+GTEST_ATTRIBUTE_PRINTF_(2, 3)
+void ColoredPrintf(GTestColor color, const char* fmt, ...) {
+  va_list args;
+  va_start(args, fmt);
+
+#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS || \
+    GTEST_OS_IOS || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT
+  const bool use_color = AlwaysFalse();
+#else
+  static const bool in_color_mode =
+      ShouldUseColor(posix::IsATTY(posix::FileNo(stdout)) != 0);
+  const bool use_color = in_color_mode && (color != COLOR_DEFAULT);
+#endif  // GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS
+  // The '!= 0' comparison is necessary to satisfy MSVC 7.1.
+
+  if (!use_color) {
+    vprintf(fmt, args);
+    va_end(args);
+    return;
+  }
+
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \
+    !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+  const HANDLE stdout_handle = GetStdHandle(STD_OUTPUT_HANDLE);
+
+  // Gets the current text color.
+  CONSOLE_SCREEN_BUFFER_INFO buffer_info;
+  GetConsoleScreenBufferInfo(stdout_handle, &buffer_info);
+  const WORD old_color_attrs = buffer_info.wAttributes;
+
+  // We need to flush the stream buffers into the console before each
+  // SetConsoleTextAttribute call lest it affect the text that is already
+  // printed but has not yet reached the console.
+  fflush(stdout);
+  SetConsoleTextAttribute(stdout_handle,
+                          GetColorAttribute(color) | FOREGROUND_INTENSITY);
+  vprintf(fmt, args);
+
+  fflush(stdout);
+  // Restores the text color.
+  SetConsoleTextAttribute(stdout_handle, old_color_attrs);
+#else
+  printf("\033[0;3%sm", GetAnsiColorCode(color));
+  vprintf(fmt, args);
+  printf("\033[m");  // Resets the terminal to default.
+#endif  // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+  va_end(args);
+}
+
+// Text printed in Google Test's text output and --gunit_list_tests
+// output to label the type parameter and value parameter for a test.
+static const char kTypeParamLabel[] = "TypeParam";
+static const char kValueParamLabel[] = "GetParam()";
+
+void PrintFullTestCommentIfPresent(const TestInfo& test_info) {
+  const char* const type_param = test_info.type_param();
+  const char* const value_param = test_info.value_param();
+
+  if (type_param != NULL || value_param != NULL) {
+    printf(", where ");
+    if (type_param != NULL) {
+      printf("%s = %s", kTypeParamLabel, type_param);
+      if (value_param != NULL)
+        printf(" and ");
+    }
+    if (value_param != NULL) {
+      printf("%s = %s", kValueParamLabel, value_param);
+    }
+  }
+}
+
+// This class implements the TestEventListener interface.
+//
+// Class PrettyUnitTestResultPrinter is copyable.
+class PrettyUnitTestResultPrinter : public TestEventListener {
+ public:
+  PrettyUnitTestResultPrinter() {}
+  static void PrintTestName(const char * test_case, const char * test) {
+    printf("%s.%s", test_case, test);
+  }
+
+  // The following methods override what's in the TestEventListener class.
+  virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {}
+  virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration);
+  virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) {}
+  virtual void OnTestCaseStart(const TestCase& test_case);
+  virtual void OnTestStart(const TestInfo& test_info);
+  virtual void OnTestPartResult(const TestPartResult& result);
+  virtual void OnTestEnd(const TestInfo& test_info);
+  virtual void OnTestCaseEnd(const TestCase& test_case);
+  virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) {}
+  virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+  virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {}
+
+ private:
+  static void PrintFailedTests(const UnitTest& unit_test);
+  static void PrintSkippedTests(const UnitTest& unit_test);
+};
+
+  // Fired before each iteration of tests starts.
+void PrettyUnitTestResultPrinter::OnTestIterationStart(
+    const UnitTest& unit_test, int iteration) {
+  if (GTEST_FLAG(repeat) != 1)
+    printf("\nRepeating all tests (iteration %d) . . .\n\n", iteration + 1);
+
+  const char* const filter = GTEST_FLAG(filter).c_str();
+
+  // Prints the filter if it's not *.  This reminds the user that some
+  // tests may be skipped.
+  if (!String::CStringEquals(filter, kUniversalFilter)) {
+    ColoredPrintf(COLOR_YELLOW,
+                  "Note: %s filter = %s\n", GTEST_NAME_, filter);
+  }
+
+  if (internal::ShouldShard(kTestTotalShards, kTestShardIndex, false)) {
+    const Int32 shard_index = Int32FromEnvOrDie(kTestShardIndex, -1);
+    ColoredPrintf(COLOR_YELLOW,
+                  "Note: This is test shard %d of %s.\n",
+                  static_cast<int>(shard_index) + 1,
+                  internal::posix::GetEnv(kTestTotalShards));
+  }
+
+  if (GTEST_FLAG(shuffle)) {
+    ColoredPrintf(COLOR_YELLOW,
+                  "Note: Randomizing tests' orders with a seed of %d .\n",
+                  unit_test.random_seed());
+  }
+
+  ColoredPrintf(COLOR_GREEN,  "[==========] ");
+  printf("Running %s from %s.\n",
+         FormatTestCount(unit_test.test_to_run_count()).c_str(),
+         FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str());
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnEnvironmentsSetUpStart(
+    const UnitTest& /*unit_test*/) {
+  ColoredPrintf(COLOR_GREEN,  "[----------] ");
+  printf("Global test environment set-up.\n");
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestCaseStart(const TestCase& test_case) {
+  const std::string counts =
+      FormatCountableNoun(test_case.test_to_run_count(), "test", "tests");
+  ColoredPrintf(COLOR_GREEN, "[----------] ");
+  printf("%s from %s", counts.c_str(), test_case.name());
+  if (test_case.type_param() == NULL) {
+    printf("\n");
+  } else {
+    printf(", where %s = %s\n", kTypeParamLabel, test_case.type_param());
+  }
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestStart(const TestInfo& test_info) {
+  ColoredPrintf(COLOR_GREEN,  "[ RUN      ] ");
+  PrintTestName(test_info.test_case_name(), test_info.name());
+  printf("\n");
+  fflush(stdout);
+}
+
+// Called after an assertion failure.
+void PrettyUnitTestResultPrinter::OnTestPartResult(
+    const TestPartResult& result) {
+  switch(result.type()) {
+    // If the test part succeeded,
+    // we don't need to do anything.
+    case TestPartResult::kSuccess:
+      return;
+    default:
+      // Print failure message from the assertion
+      // (e.g. expected this and got that).
+      PrintTestPartResult(result);
+      fflush(stdout);
+  }
+}
+
+void PrettyUnitTestResultPrinter::OnTestEnd(const TestInfo& test_info) {
+  if (test_info.result()->Passed()) {
+    ColoredPrintf(COLOR_GREEN, "[       OK ] ");
+  } else if (test_info.result()->Skipped()) {
+    ColoredPrintf(COLOR_GREEN, "[  SKIPPED ] ");
+  } else {
+    ColoredPrintf(COLOR_RED, "[  FAILED  ] ");
+  }
+  PrintTestName(test_info.test_case_name(), test_info.name());
+  if (test_info.result()->Failed())
+    PrintFullTestCommentIfPresent(test_info);
+
+  if (GTEST_FLAG(print_time)) {
+    printf(" (%s ms)\n", internal::StreamableToString(
+           test_info.result()->elapsed_time()).c_str());
+  } else {
+    printf("\n");
+  }
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestCaseEnd(const TestCase& test_case) {
+  if (!GTEST_FLAG(print_time)) return;
+
+  const std::string counts =
+      FormatCountableNoun(test_case.test_to_run_count(), "test", "tests");
+  ColoredPrintf(COLOR_GREEN, "[----------] ");
+  printf("%s from %s (%s ms total)\n\n",
+         counts.c_str(), test_case.name(),
+         internal::StreamableToString(test_case.elapsed_time()).c_str());
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnEnvironmentsTearDownStart(
+    const UnitTest& /*unit_test*/) {
+  ColoredPrintf(COLOR_GREEN,  "[----------] ");
+  printf("Global test environment tear-down\n");
+  fflush(stdout);
+}
+
+// Internal helper for printing the list of failed tests.
+void PrettyUnitTestResultPrinter::PrintFailedTests(const UnitTest& unit_test) {
+  const int failed_test_count = unit_test.failed_test_count();
+  if (failed_test_count == 0) {
+    return;
+  }
+
+  for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+    const TestCase& test_case = *unit_test.GetTestCase(i);
+    if (!test_case.should_run() || (test_case.failed_test_count() == 0)) {
+      continue;
+    }
+    for (int j = 0; j < test_case.total_test_count(); ++j) {
+      const TestInfo& test_info = *test_case.GetTestInfo(j);
+      if (!test_info.should_run() || !test_info.result()->Failed()) {
+        continue;
+      }
+      ColoredPrintf(COLOR_RED, "[  FAILED  ] ");
+      printf("%s.%s", test_case.name(), test_info.name());
+      PrintFullTestCommentIfPresent(test_info);
+      printf("\n");
+    }
+  }
+}
+
+// Internal helper for printing the list of skipped tests.
+void PrettyUnitTestResultPrinter::PrintSkippedTests(const UnitTest& unit_test) {
+  const int skipped_test_count = unit_test.skipped_test_count();
+  if (skipped_test_count == 0) {
+    return;
+  }
+
+  for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+    const TestCase& test_suite = *unit_test.GetTestCase(i);
+    if (!test_suite.should_run() || (test_suite.skipped_test_count() == 0)) {
+      continue;
+    }
+    for (int j = 0; j < test_suite.total_test_count(); ++j) {
+      const TestInfo& test_info = *test_suite.GetTestInfo(j);
+      if (!test_info.should_run() || !test_info.result()->Skipped()) {
+        continue;
+      }
+      ColoredPrintf(COLOR_GREEN, "[  SKIPPED ] ");
+      printf("%s.%s", test_suite.name(), test_info.name());
+      printf("\n");
+    }
+  }
+}
+
+void PrettyUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test,
+                                                     int /*iteration*/) {
+  ColoredPrintf(COLOR_GREEN,  "[==========] ");
+  printf("%s from %s ran.",
+         FormatTestCount(unit_test.test_to_run_count()).c_str(),
+         FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str());
+  if (GTEST_FLAG(print_time)) {
+    printf(" (%s ms total)",
+           internal::StreamableToString(unit_test.elapsed_time()).c_str());
+  }
+  printf("\n");
+  ColoredPrintf(COLOR_GREEN,  "[  PASSED  ] ");
+  printf("%s.\n", FormatTestCount(unit_test.successful_test_count()).c_str());
+
+  const int skipped_test_count = unit_test.skipped_test_count();
+  if (skipped_test_count > 0) {
+    ColoredPrintf(COLOR_GREEN, "[  SKIPPED ] ");
+    printf("%s, listed below:\n", FormatTestCount(skipped_test_count).c_str());
+    PrintSkippedTests(unit_test);
+  }
+
+  int num_failures = unit_test.failed_test_count();
+  if (!unit_test.Passed()) {
+    const int failed_test_count = unit_test.failed_test_count();
+    ColoredPrintf(COLOR_RED,  "[  FAILED  ] ");
+    printf("%s, listed below:\n", FormatTestCount(failed_test_count).c_str());
+    PrintFailedTests(unit_test);
+    printf("\n%2d FAILED %s\n", num_failures,
+                        num_failures == 1 ? "TEST" : "TESTS");
+  }
+
+  int num_disabled = unit_test.reportable_disabled_test_count();
+  if (num_disabled && !GTEST_FLAG(also_run_disabled_tests)) {
+    if (!num_failures) {
+      printf("\n");  // Add a spacer if no FAILURE banner is displayed.
+    }
+    ColoredPrintf(COLOR_YELLOW,
+                  "  YOU HAVE %d DISABLED %s\n\n",
+                  num_disabled,
+                  num_disabled == 1 ? "TEST" : "TESTS");
+  }
+  // Ensure that Google Test output is printed before, e.g., heapchecker output.
+  fflush(stdout);
+}
+
+// End PrettyUnitTestResultPrinter
+
+// class TestEventRepeater
+//
+// This class forwards events to other event listeners.
+class TestEventRepeater : public TestEventListener {
+ public:
+  TestEventRepeater() : forwarding_enabled_(true) {}
+  virtual ~TestEventRepeater();
+  void Append(TestEventListener *listener);
+  TestEventListener* Release(TestEventListener* listener);
+
+  // Controls whether events will be forwarded to listeners_. Set to false
+  // in death test child processes.
+  bool forwarding_enabled() const { return forwarding_enabled_; }
+  void set_forwarding_enabled(bool enable) { forwarding_enabled_ = enable; }
+
+  virtual void OnTestProgramStart(const UnitTest& unit_test);
+  virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration);
+  virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsSetUpEnd(const UnitTest& unit_test);
+  virtual void OnTestCaseStart(const TestCase& test_case);
+  virtual void OnTestStart(const TestInfo& test_info);
+  virtual void OnTestPartResult(const TestPartResult& result);
+  virtual void OnTestEnd(const TestInfo& test_info);
+  virtual void OnTestCaseEnd(const TestCase& test_case);
+  virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsTearDownEnd(const UnitTest& unit_test);
+  virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+  virtual void OnTestProgramEnd(const UnitTest& unit_test);
+
+ private:
+  // Controls whether events will be forwarded to listeners_. Set to false
+  // in death test child processes.
+  bool forwarding_enabled_;
+  // The list of listeners that receive events.
+  std::vector<TestEventListener*> listeners_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventRepeater);
+};
+
+TestEventRepeater::~TestEventRepeater() {
+  ForEach(listeners_, Delete<TestEventListener>);
+}
+
+void TestEventRepeater::Append(TestEventListener *listener) {
+  listeners_.push_back(listener);
+}
+
+// TODO(vladl@google.com): Factor the search functionality into Vector::Find.
+TestEventListener* TestEventRepeater::Release(TestEventListener *listener) {
+  for (size_t i = 0; i < listeners_.size(); ++i) {
+    if (listeners_[i] == listener) {
+      listeners_.erase(listeners_.begin() + i);
+      return listener;
+    }
+  }
+
+  return NULL;
+}
+
+// Since most methods are very similar, use macros to reduce boilerplate.
+// This defines a member that forwards the call to all listeners.
+#define GTEST_REPEATER_METHOD_(Name, Type) \
+void TestEventRepeater::Name(const Type& parameter) { \
+  if (forwarding_enabled_) { \
+    for (size_t i = 0; i < listeners_.size(); i++) { \
+      listeners_[i]->Name(parameter); \
+    } \
+  } \
+}
+// This defines a member that forwards the call to all listeners in reverse
+// order.
+#define GTEST_REVERSE_REPEATER_METHOD_(Name, Type) \
+void TestEventRepeater::Name(const Type& parameter) { \
+  if (forwarding_enabled_) { \
+    for (int i = static_cast<int>(listeners_.size()) - 1; i >= 0; i--) { \
+      listeners_[i]->Name(parameter); \
+    } \
+  } \
+}
+
+GTEST_REPEATER_METHOD_(OnTestProgramStart, UnitTest)
+GTEST_REPEATER_METHOD_(OnEnvironmentsSetUpStart, UnitTest)
+GTEST_REPEATER_METHOD_(OnTestCaseStart, TestCase)
+GTEST_REPEATER_METHOD_(OnTestStart, TestInfo)
+GTEST_REPEATER_METHOD_(OnTestPartResult, TestPartResult)
+GTEST_REPEATER_METHOD_(OnEnvironmentsTearDownStart, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsSetUpEnd, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsTearDownEnd, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestEnd, TestInfo)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestCaseEnd, TestCase)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestProgramEnd, UnitTest)
+
+#undef GTEST_REPEATER_METHOD_
+#undef GTEST_REVERSE_REPEATER_METHOD_
+
+void TestEventRepeater::OnTestIterationStart(const UnitTest& unit_test,
+                                             int iteration) {
+  if (forwarding_enabled_) {
+    for (size_t i = 0; i < listeners_.size(); i++) {
+      listeners_[i]->OnTestIterationStart(unit_test, iteration);
+    }
+  }
+}
+
+void TestEventRepeater::OnTestIterationEnd(const UnitTest& unit_test,
+                                           int iteration) {
+  if (forwarding_enabled_) {
+    for (int i = static_cast<int>(listeners_.size()) - 1; i >= 0; i--) {
+      listeners_[i]->OnTestIterationEnd(unit_test, iteration);
+    }
+  }
+}
+
+// End TestEventRepeater
+
+// This class generates an XML output file.
+class XmlUnitTestResultPrinter : public EmptyTestEventListener {
+ public:
+  explicit XmlUnitTestResultPrinter(const char* output_file);
+
+  virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+
+ private:
+  // Is c a whitespace character that is normalized to a space character
+  // when it appears in an XML attribute value?
+  static bool IsNormalizableWhitespace(char c) {
+    return c == 0x9 || c == 0xA || c == 0xD;
+  }
+
+  // May c appear in a well-formed XML document?
+  static bool IsValidXmlCharacter(char c) {
+    return IsNormalizableWhitespace(c) || c >= 0x20;
+  }
+
+  // Returns an XML-escaped copy of the input string str.  If
+  // is_attribute is true, the text is meant to appear as an attribute
+  // value, and normalizable whitespace is preserved by replacing it
+  // with character references.
+  static std::string EscapeXml(const std::string& str, bool is_attribute);
+
+  // Returns the given string with all characters invalid in XML removed.
+  static std::string RemoveInvalidXmlCharacters(const std::string& str);
+
+  // Convenience wrapper around EscapeXml when str is an attribute value.
+  static std::string EscapeXmlAttribute(const std::string& str) {
+    return EscapeXml(str, true);
+  }
+
+  // Convenience wrapper around EscapeXml when str is not an attribute value.
+  static std::string EscapeXmlText(const char* str) {
+    return EscapeXml(str, false);
+  }
+
+  // Verifies that the given attribute belongs to the given element and
+  // streams the attribute as XML.
+  static void OutputXmlAttribute(std::ostream* stream,
+                                 const std::string& element_name,
+                                 const std::string& name,
+                                 const std::string& value);
+
+  // Streams an XML CDATA section, escaping invalid CDATA sequences as needed.
+  static void OutputXmlCDataSection(::std::ostream* stream, const char* data);
+
+  // Streams an XML representation of a TestInfo object.
+  static void OutputXmlTestInfo(::std::ostream* stream,
+                                const char* test_case_name,
+                                const TestInfo& test_info);
+
+  // Prints an XML representation of a TestCase object
+  static void PrintXmlTestCase(::std::ostream* stream,
+                               const TestCase& test_case);
+
+  // Prints an XML summary of unit_test to output stream out.
+  static void PrintXmlUnitTest(::std::ostream* stream,
+                               const UnitTest& unit_test);
+
+  // Produces a string representing the test properties in a result as space
+  // delimited XML attributes based on the property key="value" pairs.
+  // When the std::string is not empty, it includes a space at the beginning,
+  // to delimit this attribute from prior attributes.
+  static std::string TestPropertiesAsXmlAttributes(const TestResult& result);
+
+  // The output file.
+  const std::string output_file_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(XmlUnitTestResultPrinter);
+};
+
+// Creates a new XmlUnitTestResultPrinter.
+XmlUnitTestResultPrinter::XmlUnitTestResultPrinter(const char* output_file)
+    : output_file_(output_file) {
+  if (output_file_.c_str() == NULL || output_file_.empty()) {
+    fprintf(stderr, "XML output file may not be null\n");
+    fflush(stderr);
+    exit(EXIT_FAILURE);
+  }
+}
+
+// Called after the unit test ends.
+void XmlUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test,
+                                                  int /*iteration*/) {
+  FILE* xmlout = NULL;
+  FilePath output_file(output_file_);
+  FilePath output_dir(output_file.RemoveFileName());
+
+  if (output_dir.CreateDirectoriesRecursively()) {
+    xmlout = posix::FOpen(output_file_.c_str(), "w");
+  }
+  if (xmlout == NULL) {
+    // TODO(wan): report the reason of the failure.
+    //
+    // We don't do it for now as:
+    //
+    //   1. There is no urgent need for it.
+    //   2. It's a bit involved to make the errno variable thread-safe on
+    //      all three operating systems (Linux, Windows, and Mac OS).
+    //   3. To interpret the meaning of errno in a thread-safe way,
+    //      we need the strerror_r() function, which is not available on
+    //      Windows.
+    fprintf(stderr,
+            "Unable to open file \"%s\"\n",
+            output_file_.c_str());
+    fflush(stderr);
+    exit(EXIT_FAILURE);
+  }
+  std::stringstream stream;
+  PrintXmlUnitTest(&stream, unit_test);
+  fprintf(xmlout, "%s", StringStreamToString(&stream).c_str());
+  fclose(xmlout);
+}
+
+// Returns an XML-escaped copy of the input string str.  If is_attribute
+// is true, the text is meant to appear as an attribute value, and
+// normalizable whitespace is preserved by replacing it with character
+// references.
+//
+// Invalid XML characters in str, if any, are stripped from the output.
+// It is expected that most, if not all, of the text processed by this
+// module will consist of ordinary English text.
+// If this module is ever modified to produce version 1.1 XML output,
+// most invalid characters can be retained using character references.
+// TODO(wan): It might be nice to have a minimally invasive, human-readable
+// escaping scheme for invalid characters, rather than dropping them.
+std::string XmlUnitTestResultPrinter::EscapeXml(
+    const std::string& str, bool is_attribute) {
+  Message m;
+
+  for (size_t i = 0; i < str.size(); ++i) {
+    const char ch = str[i];
+    switch (ch) {
+      case '<':
+        m << "&lt;";
+        break;
+      case '>':
+        m << "&gt;";
+        break;
+      case '&':
+        m << "&amp;";
+        break;
+      case '\'':
+        if (is_attribute)
+          m << "&apos;";
+        else
+          m << '\'';
+        break;
+      case '"':
+        if (is_attribute)
+          m << "&quot;";
+        else
+          m << '"';
+        break;
+      default:
+        if (IsValidXmlCharacter(ch)) {
+          if (is_attribute && IsNormalizableWhitespace(ch))
+            m << "&#x" << String::FormatByte(static_cast<unsigned char>(ch))
+              << ";";
+          else
+            m << ch;
+        }
+        break;
+    }
+  }
+
+  return m.GetString();
+}
+
+// Returns the given string with all characters invalid in XML removed.
+// Currently invalid characters are dropped from the string. An
+// alternative is to replace them with certain characters such as . or ?.
+std::string XmlUnitTestResultPrinter::RemoveInvalidXmlCharacters(
+    const std::string& str) {
+  std::string output;
+  output.reserve(str.size());
+  for (std::string::const_iterator it = str.begin(); it != str.end(); ++it)
+    if (IsValidXmlCharacter(*it))
+      output.push_back(*it);
+
+  return output;
+}
+
+// The following routines generate an XML representation of a UnitTest
+// object.
+//
+// This is how Google Test concepts map to the DTD:
+//
+// <testsuites name="AllTests">        <-- corresponds to a UnitTest object
+//   <testsuite name="testcase-name">  <-- corresponds to a TestCase object
+//     <testcase name="test-name">     <-- corresponds to a TestInfo object
+//       <failure message="...">...</failure>
+//       <failure message="...">...</failure>
+//       <failure message="...">...</failure>
+//                                     <-- individual assertion failures
+//     </testcase>
+//   </testsuite>
+// </testsuites>
+
+// Formats the given time in milliseconds as seconds.
+std::string FormatTimeInMillisAsSeconds(TimeInMillis ms) {
+  ::std::stringstream ss;
+  ss << (static_cast<double>(ms) * 1e-3);
+  return ss.str();
+}
+
+static bool PortableLocaltime(time_t seconds, struct tm* out) {
+#if defined(_MSC_VER)
+  return localtime_s(out, &seconds) == 0;
+#elif defined(__MINGW32__) || defined(__MINGW64__)
+  // MINGW <time.h> provides neither localtime_r nor localtime_s, but uses
+  // Windows' localtime(), which has a thread-local tm buffer.
+  struct tm* tm_ptr = localtime(&seconds);  // NOLINT
+  if (tm_ptr == NULL)
+    return false;
+  *out = *tm_ptr;
+  return true;
+#else
+  return localtime_r(&seconds, out) != NULL;
+#endif
+}
+
+// Converts the given epoch time in milliseconds to a date string in the ISO
+// 8601 format, without the timezone information.
+std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms) {
+  struct tm time_struct;
+  if (!PortableLocaltime(static_cast<time_t>(ms / 1000), &time_struct))
+    return "";
+  // YYYY-MM-DDThh:mm:ss
+  return StreamableToString(time_struct.tm_year + 1900) + "-" +
+      String::FormatIntWidth2(time_struct.tm_mon + 1) + "-" +
+      String::FormatIntWidth2(time_struct.tm_mday) + "T" +
+      String::FormatIntWidth2(time_struct.tm_hour) + ":" +
+      String::FormatIntWidth2(time_struct.tm_min) + ":" +
+      String::FormatIntWidth2(time_struct.tm_sec);
+}
+
+// Streams an XML CDATA section, escaping invalid CDATA sequences as needed.
+void XmlUnitTestResultPrinter::OutputXmlCDataSection(::std::ostream* stream,
+                                                     const char* data) {
+  const char* segment = data;
+  *stream << "<![CDATA[";
+  for (;;) {
+    const char* const next_segment = strstr(segment, "]]>");
+    if (next_segment != NULL) {
+      stream->write(
+          segment, static_cast<std::streamsize>(next_segment - segment));
+      *stream << "]]>]]&gt;<![CDATA[";
+      segment = next_segment + strlen("]]>");
+    } else {
+      *stream << segment;
+      break;
+    }
+  }
+  *stream << "]]>";
+}
+
+void XmlUnitTestResultPrinter::OutputXmlAttribute(
+    std::ostream* stream,
+    const std::string& element_name,
+    const std::string& name,
+    const std::string& value) {
+  const std::vector<std::string>& allowed_names =
+      GetReservedAttributesForElement(element_name);
+
+  GTEST_CHECK_(std::find(allowed_names.begin(), allowed_names.end(), name) !=
+                   allowed_names.end())
+      << "Attribute " << name << " is not allowed for element <" << element_name
+      << ">.";
+
+  *stream << " " << name << "=\"" << EscapeXmlAttribute(value) << "\"";
+}
+
+// Prints an XML representation of a TestInfo object.
+// TODO(wan): There is also value in printing properties with the plain printer.
+void XmlUnitTestResultPrinter::OutputXmlTestInfo(::std::ostream* stream,
+                                                 const char* test_case_name,
+                                                 const TestInfo& test_info) {
+  const TestResult& result = *test_info.result();
+  const std::string kTestcase = "testcase";
+
+  *stream << "    <testcase";
+  OutputXmlAttribute(stream, kTestcase, "name", test_info.name());
+
+  if (test_info.value_param() != NULL) {
+    OutputXmlAttribute(stream, kTestcase, "value_param",
+                       test_info.value_param());
+  }
+  if (test_info.type_param() != NULL) {
+    OutputXmlAttribute(stream, kTestcase, "type_param", test_info.type_param());
+  }
+
+  OutputXmlAttribute(stream, kTestcase, "status",
+                     test_info.should_run() ? "run" : "notrun");
+  OutputXmlAttribute(stream, kTestcase, "result",
+                     test_info.should_run()
+                         ? (result.Skipped() ? "skipped" : "completed")
+                         : "suppressed");
+  OutputXmlAttribute(stream, kTestcase, "time",
+                     FormatTimeInMillisAsSeconds(result.elapsed_time()));
+  OutputXmlAttribute(stream, kTestcase, "classname", test_case_name);
+  *stream << TestPropertiesAsXmlAttributes(result);
+
+  int failures = 0;
+  for (int i = 0; i < result.total_part_count(); ++i) {
+    const TestPartResult& part = result.GetTestPartResult(i);
+    if (part.failed()) {
+      if (++failures == 1) {
+        *stream << ">\n";
+      }
+      const std::string location =
+          internal::FormatCompilerIndependentFileLocation(part.file_name(),
+                                                          part.line_number());
+      const std::string summary = location + "\n" + part.summary();
+      *stream << "      <failure message=\""
+              << EscapeXmlAttribute(summary.c_str())
+              << "\" type=\"\">";
+      const std::string detail = location + "\n" + part.message();
+      OutputXmlCDataSection(stream, RemoveInvalidXmlCharacters(detail).c_str());
+      *stream << "</failure>\n";
+    }
+  }
+
+  if (failures == 0)
+    *stream << " />\n";
+  else
+    *stream << "    </testcase>\n";
+}
+
+// Prints an XML representation of a TestCase object
+void XmlUnitTestResultPrinter::PrintXmlTestCase(std::ostream* stream,
+                                                const TestCase& test_case) {
+  const std::string kTestsuite = "testsuite";
+  *stream << "  <" << kTestsuite;
+  OutputXmlAttribute(stream, kTestsuite, "name", test_case.name());
+  OutputXmlAttribute(stream, kTestsuite, "tests",
+                     StreamableToString(test_case.reportable_test_count()));
+  OutputXmlAttribute(stream, kTestsuite, "failures",
+                     StreamableToString(test_case.failed_test_count()));
+  OutputXmlAttribute(
+      stream, kTestsuite, "disabled",
+      StreamableToString(test_case.reportable_disabled_test_count()));
+  OutputXmlAttribute(stream, kTestsuite, "errors", "0");
+  OutputXmlAttribute(stream, kTestsuite, "time",
+                     FormatTimeInMillisAsSeconds(test_case.elapsed_time()));
+  *stream << TestPropertiesAsXmlAttributes(test_case.ad_hoc_test_result())
+          << ">\n";
+
+  for (int i = 0; i < test_case.total_test_count(); ++i) {
+    if (test_case.GetTestInfo(i)->is_reportable())
+      OutputXmlTestInfo(stream, test_case.name(), *test_case.GetTestInfo(i));
+  }
+  *stream << "  </" << kTestsuite << ">\n";
+}
+
+// Prints an XML summary of unit_test to output stream out.
+void XmlUnitTestResultPrinter::PrintXmlUnitTest(std::ostream* stream,
+                                                const UnitTest& unit_test) {
+  const std::string kTestsuites = "testsuites";
+
+  *stream << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
+  *stream << "<" << kTestsuites;
+
+  OutputXmlAttribute(stream, kTestsuites, "tests",
+                     StreamableToString(unit_test.reportable_test_count()));
+  OutputXmlAttribute(stream, kTestsuites, "failures",
+                     StreamableToString(unit_test.failed_test_count()));
+  OutputXmlAttribute(
+      stream, kTestsuites, "disabled",
+      StreamableToString(unit_test.reportable_disabled_test_count()));
+  OutputXmlAttribute(stream, kTestsuites, "errors", "0");
+  OutputXmlAttribute(
+      stream, kTestsuites, "timestamp",
+      FormatEpochTimeInMillisAsIso8601(unit_test.start_timestamp()));
+  OutputXmlAttribute(stream, kTestsuites, "time",
+                     FormatTimeInMillisAsSeconds(unit_test.elapsed_time()));
+
+  if (GTEST_FLAG(shuffle)) {
+    OutputXmlAttribute(stream, kTestsuites, "random_seed",
+                       StreamableToString(unit_test.random_seed()));
+  }
+
+  *stream << TestPropertiesAsXmlAttributes(unit_test.ad_hoc_test_result());
+
+  OutputXmlAttribute(stream, kTestsuites, "name", "AllTests");
+  *stream << ">\n";
+
+  for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+    if (unit_test.GetTestCase(i)->reportable_test_count() > 0)
+      PrintXmlTestCase(stream, *unit_test.GetTestCase(i));
+  }
+  *stream << "</" << kTestsuites << ">\n";
+}
+
+// Produces a string representing the test properties in a result as space
+// delimited XML attributes based on the property key="value" pairs.
+std::string XmlUnitTestResultPrinter::TestPropertiesAsXmlAttributes(
+    const TestResult& result) {
+  Message attributes;
+  for (int i = 0; i < result.test_property_count(); ++i) {
+    const TestProperty& property = result.GetTestProperty(i);
+    attributes << " " << property.key() << "="
+        << "\"" << EscapeXmlAttribute(property.value()) << "\"";
+  }
+  return attributes.GetString();
+}
+
+// End XmlUnitTestResultPrinter
+
+#if GTEST_CAN_STREAM_RESULTS_
+
+// Checks if str contains '=', '&', '%' or '\n' characters. If yes,
+// replaces them by "%xx" where xx is their hexadecimal value. For
+// example, replaces "=" with "%3D".  This algorithm is O(strlen(str))
+// in both time and space -- important as the input str may contain an
+// arbitrarily long test failure message and stack trace.
+std::string StreamingListener::UrlEncode(const char* str) {
+  std::string result;
+  result.reserve(strlen(str) + 1);
+  for (char ch = *str; ch != '\0'; ch = *++str) {
+    switch (ch) {
+      case '%':
+      case '=':
+      case '&':
+      case '\n':
+        result.append("%" + String::FormatByte(static_cast<unsigned char>(ch)));
+        break;
+      default:
+        result.push_back(ch);
+        break;
+    }
+  }
+  return result;
+}
+
+void StreamingListener::SocketWriter::MakeConnection() {
+  GTEST_CHECK_(sockfd_ == -1)
+      << "MakeConnection() can't be called when there is already a connection.";
+
+  addrinfo hints;
+  memset(&hints, 0, sizeof(hints));
+  hints.ai_family = AF_UNSPEC;    // To allow both IPv4 and IPv6 addresses.
+  hints.ai_socktype = SOCK_STREAM;
+  addrinfo* servinfo = NULL;
+
+  // Use the getaddrinfo() to get a linked list of IP addresses for
+  // the given host name.
+  const int error_num = getaddrinfo(
+      host_name_.c_str(), port_num_.c_str(), &hints, &servinfo);
+  if (error_num != 0) {
+    GTEST_LOG_(WARNING) << "stream_result_to: getaddrinfo() failed: "
+                        << gai_strerror(error_num);
+  }
+
+  // Loop through all the results and connect to the first we can.
+  for (addrinfo* cur_addr = servinfo; sockfd_ == -1 && cur_addr != NULL;
+       cur_addr = cur_addr->ai_next) {
+    sockfd_ = socket(
+        cur_addr->ai_family, cur_addr->ai_socktype, cur_addr->ai_protocol);
+    if (sockfd_ != -1) {
+      // Connect the client socket to the server socket.
+      if (connect(sockfd_, cur_addr->ai_addr, cur_addr->ai_addrlen) == -1) {
+        close(sockfd_);
+        sockfd_ = -1;
+      }
+    }
+  }
+
+  freeaddrinfo(servinfo);  // all done with this structure
+
+  if (sockfd_ == -1) {
+    GTEST_LOG_(WARNING) << "stream_result_to: failed to connect to "
+                        << host_name_ << ":" << port_num_;
+  }
+}
+
+// End of class Streaming Listener
+#endif  // GTEST_CAN_STREAM_RESULTS__
+
+// Class ScopedTrace
+
+// Pushes the given source file location and message onto a per-thread
+// trace stack maintained by Google Test.
+ScopedTrace::ScopedTrace(const char* file, int line, const Message& message)
+    GTEST_LOCK_EXCLUDED_(&UnitTest::mutex_) {
+  TraceInfo trace;
+  trace.file = file;
+  trace.line = line;
+  trace.message = message.GetString();
+
+  UnitTest::GetInstance()->PushGTestTrace(trace);
+}
+
+// Pops the info pushed by the c'tor.
+ScopedTrace::~ScopedTrace()
+    GTEST_LOCK_EXCLUDED_(&UnitTest::mutex_) {
+  UnitTest::GetInstance()->PopGTestTrace();
+}
+
+
+// class OsStackTraceGetter
+
+const char* const OsStackTraceGetterInterface::kElidedFramesMarker =
+    "... " GTEST_NAME_ " internal frames ...";
+
+std::string OsStackTraceGetter::CurrentStackTrace(int /*max_depth*/,
+                                                  int /*skip_count*/) {
+  return "";
+}
+
+void OsStackTraceGetter::UponLeavingGTest() {}
+
+// A helper class that creates the premature-exit file in its
+// constructor and deletes the file in its destructor.
+class ScopedPrematureExitFile {
+ public:
+  explicit ScopedPrematureExitFile(const char* premature_exit_filepath)
+      : premature_exit_filepath_(premature_exit_filepath) {
+    // If a path to the premature-exit file is specified...
+    if (premature_exit_filepath != NULL && *premature_exit_filepath != '\0') {
+      // create the file with a single "0" character in it.  I/O
+      // errors are ignored as there's nothing better we can do and we
+      // don't want to fail the test because of this.
+      FILE* pfile = posix::FOpen(premature_exit_filepath, "w");
+      fwrite("0", 1, 1, pfile);
+      fclose(pfile);
+    }
+  }
+
+  ~ScopedPrematureExitFile() {
+    if (premature_exit_filepath_ != NULL && *premature_exit_filepath_ != '\0') {
+      remove(premature_exit_filepath_);
+    }
+  }
+
+ private:
+  const char* const premature_exit_filepath_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedPrematureExitFile);
+};
+
+}  // namespace internal
+
+// class TestEventListeners
+
+TestEventListeners::TestEventListeners()
+    : repeater_(new internal::TestEventRepeater()),
+      default_result_printer_(NULL),
+      default_xml_generator_(NULL) {
+}
+
+TestEventListeners::~TestEventListeners() { delete repeater_; }
+
+// Returns the standard listener responsible for the default console
+// output.  Can be removed from the listeners list to shut down default
+// console output.  Note that removing this object from the listener list
+// with Release transfers its ownership to the user.
+void TestEventListeners::Append(TestEventListener* listener) {
+  repeater_->Append(listener);
+}
+
+// Removes the given event listener from the list and returns it.  It then
+// becomes the caller's responsibility to delete the listener. Returns
+// NULL if the listener is not found in the list.
+TestEventListener* TestEventListeners::Release(TestEventListener* listener) {
+  if (listener == default_result_printer_)
+    default_result_printer_ = NULL;
+  else if (listener == default_xml_generator_)
+    default_xml_generator_ = NULL;
+  return repeater_->Release(listener);
+}
+
+// Returns repeater that broadcasts the TestEventListener events to all
+// subscribers.
+TestEventListener* TestEventListeners::repeater() { return repeater_; }
+
+// Sets the default_result_printer attribute to the provided listener.
+// The listener is also added to the listener list and previous
+// default_result_printer is removed from it and deleted. The listener can
+// also be NULL in which case it will not be added to the list. Does
+// nothing if the previous and the current listener objects are the same.
+void TestEventListeners::SetDefaultResultPrinter(TestEventListener* listener) {
+  if (default_result_printer_ != listener) {
+    // It is an error to pass this method a listener that is already in the
+    // list.
+    delete Release(default_result_printer_);
+    default_result_printer_ = listener;
+    if (listener != NULL)
+      Append(listener);
+  }
+}
+
+// Sets the default_xml_generator attribute to the provided listener.  The
+// listener is also added to the listener list and previous
+// default_xml_generator is removed from it and deleted. The listener can
+// also be NULL in which case it will not be added to the list. Does
+// nothing if the previous and the current listener objects are the same.
+void TestEventListeners::SetDefaultXmlGenerator(TestEventListener* listener) {
+  if (default_xml_generator_ != listener) {
+    // It is an error to pass this method a listener that is already in the
+    // list.
+    delete Release(default_xml_generator_);
+    default_xml_generator_ = listener;
+    if (listener != NULL)
+      Append(listener);
+  }
+}
+
+// Controls whether events will be forwarded by the repeater to the
+// listeners in the list.
+bool TestEventListeners::EventForwardingEnabled() const {
+  return repeater_->forwarding_enabled();
+}
+
+void TestEventListeners::SuppressEventForwarding() {
+  repeater_->set_forwarding_enabled(false);
+}
+
+// class UnitTest
+
+// Gets the singleton UnitTest object.  The first time this method is
+// called, a UnitTest object is constructed and returned.  Consecutive
+// calls will return the same object.
+//
+// We don't protect this under mutex_ as a user is not supposed to
+// call this before main() starts, from which point on the return
+// value will never change.
+UnitTest* UnitTest::GetInstance() {
+  // When compiled with MSVC 7.1 in optimized mode, destroying the
+  // UnitTest object upon exiting the program messes up the exit code,
+  // causing successful tests to appear failed.  We have to use a
+  // different implementation in this case to bypass the compiler bug.
+  // This implementation makes the compiler happy, at the cost of
+  // leaking the UnitTest object.
+
+  // CodeGear C++Builder insists on a public destructor for the
+  // default implementation.  Use this implementation to keep good OO
+  // design with private destructor.
+
+#if (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__)
+  static UnitTest* const instance = new UnitTest;
+  return instance;
+#else
+  static UnitTest instance;
+  return &instance;
+#endif  // (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__)
+}
+
+// Gets the number of successful test cases.
+int UnitTest::successful_test_case_count() const {
+  return impl()->successful_test_case_count();
+}
+
+// Gets the number of failed test cases.
+int UnitTest::failed_test_case_count() const {
+  return impl()->failed_test_case_count();
+}
+
+// Gets the number of all test cases.
+int UnitTest::total_test_case_count() const {
+  return impl()->total_test_case_count();
+}
+
+// Gets the number of all test cases that contain at least one test
+// that should run.
+int UnitTest::test_case_to_run_count() const {
+  return impl()->test_case_to_run_count();
+}
+
+// Gets the number of successful tests.
+int UnitTest::successful_test_count() const {
+  return impl()->successful_test_count();
+}
+
+// Gets the number of skipped tests.
+int UnitTest::skipped_test_count() const {
+  return impl()->skipped_test_count();
+}
+
+// Gets the number of failed tests.
+int UnitTest::failed_test_count() const { return impl()->failed_test_count(); }
+
+// Gets the number of disabled tests that will be reported in the XML report.
+int UnitTest::reportable_disabled_test_count() const {
+  return impl()->reportable_disabled_test_count();
+}
+
+// Gets the number of disabled tests.
+int UnitTest::disabled_test_count() const {
+  return impl()->disabled_test_count();
+}
+
+// Gets the number of tests to be printed in the XML report.
+int UnitTest::reportable_test_count() const {
+  return impl()->reportable_test_count();
+}
+
+// Gets the number of all tests.
+int UnitTest::total_test_count() const { return impl()->total_test_count(); }
+
+// Gets the number of tests that should run.
+int UnitTest::test_to_run_count() const { return impl()->test_to_run_count(); }
+
+// Gets the time of the test program start, in ms from the start of the
+// UNIX epoch.
+internal::TimeInMillis UnitTest::start_timestamp() const {
+    return impl()->start_timestamp();
+}
+
+// Gets the elapsed time, in milliseconds.
+internal::TimeInMillis UnitTest::elapsed_time() const {
+  return impl()->elapsed_time();
+}
+
+// Returns true iff the unit test passed (i.e. all test cases passed).
+bool UnitTest::Passed() const { return impl()->Passed(); }
+
+// Returns true iff the unit test failed (i.e. some test case failed
+// or something outside of all tests failed).
+bool UnitTest::Failed() const { return impl()->Failed(); }
+
+// Gets the i-th test case among all the test cases. i can range from 0 to
+// total_test_case_count() - 1. If i is not in that range, returns NULL.
+const TestCase* UnitTest::GetTestCase(int i) const {
+  return impl()->GetTestCase(i);
+}
+
+// Returns the TestResult containing information on test failures and
+// properties logged outside of individual test cases.
+const TestResult& UnitTest::ad_hoc_test_result() const {
+  return *impl()->ad_hoc_test_result();
+}
+
+// Gets the i-th test case among all the test cases. i can range from 0 to
+// total_test_case_count() - 1. If i is not in that range, returns NULL.
+TestCase* UnitTest::GetMutableTestCase(int i) {
+  return impl()->GetMutableTestCase(i);
+}
+
+// Returns the list of event listeners that can be used to track events
+// inside Google Test.
+TestEventListeners& UnitTest::listeners() {
+  return *impl()->listeners();
+}
+
+// Registers and returns a global test environment.  When a test
+// program is run, all global test environments will be set-up in the
+// order they were registered.  After all tests in the program have
+// finished, all global test environments will be torn-down in the
+// *reverse* order they were registered.
+//
+// The UnitTest object takes ownership of the given environment.
+//
+// We don't protect this under mutex_, as we only support calling it
+// from the main thread.
+Environment* UnitTest::AddEnvironment(Environment* env) {
+  if (env == NULL) {
+    return NULL;
+  }
+
+  impl_->environments().push_back(env);
+  return env;
+}
+
+// Adds a TestPartResult to the current TestResult object.  All Google Test
+// assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc) eventually call
+// this to report their results.  The user code should use the
+// assertion macros instead of calling this directly.
+void UnitTest::AddTestPartResult(
+    TestPartResult::Type result_type,
+    const char* file_name,
+    int line_number,
+    const std::string& message,
+    const std::string& os_stack_trace) GTEST_LOCK_EXCLUDED_(mutex_) {
+  Message msg;
+  msg << message;
+
+  internal::MutexLock lock(&mutex_);
+  if (impl_->gtest_trace_stack().size() > 0) {
+    msg << "\n" << GTEST_NAME_ << " trace:";
+
+    for (int i = static_cast<int>(impl_->gtest_trace_stack().size());
+         i > 0; --i) {
+      const internal::TraceInfo& trace = impl_->gtest_trace_stack()[i - 1];
+      msg << "\n" << internal::FormatFileLocation(trace.file, trace.line)
+          << " " << trace.message;
+    }
+  }
+
+  if (os_stack_trace.c_str() != NULL && !os_stack_trace.empty()) {
+    msg << internal::kStackTraceMarker << os_stack_trace;
+  }
+
+  const TestPartResult result =
+    TestPartResult(result_type, file_name, line_number,
+                   msg.GetString().c_str());
+  impl_->GetTestPartResultReporterForCurrentThread()->
+      ReportTestPartResult(result);
+
+  if (result_type != TestPartResult::kSuccess &&
+      result_type != TestPartResult::kSkip) {
+    // gtest_break_on_failure takes precedence over
+    // gtest_throw_on_failure.  This allows a user to set the latter
+    // in the code (perhaps in order to use Google Test assertions
+    // with another testing framework) and specify the former on the
+    // command line for debugging.
+    if (GTEST_FLAG(break_on_failure)) {
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+      // Using DebugBreak on Windows allows gtest to still break into a debugger
+      // when a failure happens and both the --gtest_break_on_failure and
+      // the --gtest_catch_exceptions flags are specified.
+      DebugBreak();
+#else
+      // Dereference NULL through a volatile pointer to prevent the compiler
+      // from removing. We use this rather than abort() or __builtin_trap() for
+      // portability: Symbian doesn't implement abort() well, and some debuggers
+      // don't correctly trap abort().
+      *static_cast<volatile int*>(NULL) = 1;
+#endif  // GTEST_OS_WINDOWS
+    } else if (GTEST_FLAG(throw_on_failure)) {
+#if GTEST_HAS_EXCEPTIONS
+      throw internal::GoogleTestFailureException(result);
+#else
+      // We cannot call abort() as it generates a pop-up in debug mode
+      // that cannot be suppressed in VC 7.1 or below.
+      exit(1);
+#endif
+    }
+  }
+}
+
+// Adds a TestProperty to the current TestResult object when invoked from
+// inside a test, to current TestCase's ad_hoc_test_result_ when invoked
+// from SetUpTestCase or TearDownTestCase, or to the global property set
+// when invoked elsewhere.  If the result already contains a property with
+// the same key, the value will be updated.
+void UnitTest::RecordProperty(const std::string& key,
+                              const std::string& value) {
+  impl_->RecordProperty(TestProperty(key, value));
+}
+
+// Runs all tests in this UnitTest object and prints the result.
+// Returns 0 if successful, or 1 otherwise.
+//
+// We don't protect this under mutex_, as we only support calling it
+// from the main thread.
+int UnitTest::Run() {
+  const bool in_death_test_child_process =
+      internal::GTEST_FLAG(internal_run_death_test).length() > 0;
+
+  // Google Test implements this protocol for catching that a test
+  // program exits before returning control to Google Test:
+  //
+  //   1. Upon start, Google Test creates a file whose absolute path
+  //      is specified by the environment variable
+  //      TEST_PREMATURE_EXIT_FILE.
+  //   2. When Google Test has finished its work, it deletes the file.
+  //
+  // This allows a test runner to set TEST_PREMATURE_EXIT_FILE before
+  // running a Google-Test-based test program and check the existence
+  // of the file at the end of the test execution to see if it has
+  // exited prematurely.
+
+  // If we are in the child process of a death test, don't
+  // create/delete the premature exit file, as doing so is unnecessary
+  // and will confuse the parent process.  Otherwise, create/delete
+  // the file upon entering/leaving this function.  If the program
+  // somehow exits before this function has a chance to return, the
+  // premature-exit file will be left undeleted, causing a test runner
+  // that understands the premature-exit-file protocol to report the
+  // test as having failed.
+  const internal::ScopedPrematureExitFile premature_exit_file(
+      in_death_test_child_process ?
+      NULL : internal::posix::GetEnv("TEST_PREMATURE_EXIT_FILE"));
+
+  // Captures the value of GTEST_FLAG(catch_exceptions).  This value will be
+  // used for the duration of the program.
+  impl()->set_catch_exceptions(GTEST_FLAG(catch_exceptions));
+
+#if GTEST_HAS_SEH
+  // Either the user wants Google Test to catch exceptions thrown by the
+  // tests or this is executing in the context of death test child
+  // process. In either case the user does not want to see pop-up dialogs
+  // about crashes - they are expected.
+  if (impl()->catch_exceptions() || in_death_test_child_process) {
+# if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+    // SetErrorMode doesn't exist on CE.
+    SetErrorMode(SEM_FAILCRITICALERRORS | SEM_NOALIGNMENTFAULTEXCEPT |
+                 SEM_NOGPFAULTERRORBOX | SEM_NOOPENFILEERRORBOX);
+# endif  // !GTEST_OS_WINDOWS_MOBILE
+
+# if (defined(_MSC_VER) || GTEST_OS_WINDOWS_MINGW) && !GTEST_OS_WINDOWS_MOBILE
+    // Death test children can be terminated with _abort().  On Windows,
+    // _abort() can show a dialog with a warning message.  This forces the
+    // abort message to go to stderr instead.
+    _set_error_mode(_OUT_TO_STDERR);
+# endif
+
+# if _MSC_VER >= 1400 && !GTEST_OS_WINDOWS_MOBILE
+    // In the debug version, Visual Studio pops up a separate dialog
+    // offering a choice to debug the aborted program. We need to suppress
+    // this dialog or it will pop up for every EXPECT/ASSERT_DEATH statement
+    // executed. Google Test will notify the user of any unexpected
+    // failure via stderr.
+    //
+    // VC++ doesn't define _set_abort_behavior() prior to the version 8.0.
+    // Users of prior VC versions shall suffer the agony and pain of
+    // clicking through the countless debug dialogs.
+    // TODO(vladl@google.com): find a way to suppress the abort dialog() in the
+    // debug mode when compiled with VC 7.1 or lower.
+    if (!GTEST_FLAG(break_on_failure))
+      _set_abort_behavior(
+          0x0,                                    // Clear the following flags:
+          _WRITE_ABORT_MSG | _CALL_REPORTFAULT);  // pop-up window, core dump.
+# endif
+  }
+#endif  // GTEST_HAS_SEH
+
+  return internal::HandleExceptionsInMethodIfSupported(
+      impl(),
+      &internal::UnitTestImpl::RunAllTests,
+      "auxiliary test code (environments or event listeners)") ? 0 : 1;
+}
+
+// Returns the working directory when the first TEST() or TEST_F() was
+// executed.
+const char* UnitTest::original_working_dir() const {
+  return impl_->original_working_dir_.c_str();
+}
+
+// Returns the TestCase object for the test that's currently running,
+// or NULL if no test is running.
+const TestCase* UnitTest::current_test_case() const
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  return impl_->current_test_case();
+}
+
+// Returns the TestInfo object for the test that's currently running,
+// or NULL if no test is running.
+const TestInfo* UnitTest::current_test_info() const
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  return impl_->current_test_info();
+}
+
+// Returns the random seed used at the start of the current test run.
+int UnitTest::random_seed() const { return impl_->random_seed(); }
+
+#if GTEST_HAS_PARAM_TEST
+// Returns ParameterizedTestCaseRegistry object used to keep track of
+// value-parameterized tests and instantiate and register them.
+internal::ParameterizedTestCaseRegistry&
+    UnitTest::parameterized_test_registry()
+        GTEST_LOCK_EXCLUDED_(mutex_) {
+  return impl_->parameterized_test_registry();
+}
+#endif  // GTEST_HAS_PARAM_TEST
+
+// Creates an empty UnitTest.
+UnitTest::UnitTest() {
+  impl_ = new internal::UnitTestImpl(this);
+}
+
+// Destructor of UnitTest.
+UnitTest::~UnitTest() {
+  delete impl_;
+}
+
+// Pushes a trace defined by SCOPED_TRACE() on to the per-thread
+// Google Test trace stack.
+void UnitTest::PushGTestTrace(const internal::TraceInfo& trace)
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  impl_->gtest_trace_stack().push_back(trace);
+}
+
+// Pops a trace from the per-thread Google Test trace stack.
+void UnitTest::PopGTestTrace()
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  impl_->gtest_trace_stack().pop_back();
+}
+
+namespace internal {
+
+UnitTestImpl::UnitTestImpl(UnitTest* parent)
+    : parent_(parent),
+      GTEST_DISABLE_MSC_WARNINGS_PUSH_(4355 /* using this in initializer */)
+      default_global_test_part_result_reporter_(this),
+      default_per_thread_test_part_result_reporter_(this),
+      GTEST_DISABLE_MSC_WARNINGS_POP_()
+      global_test_part_result_repoter_(
+          &default_global_test_part_result_reporter_),
+      per_thread_test_part_result_reporter_(
+          &default_per_thread_test_part_result_reporter_),
+#if GTEST_HAS_PARAM_TEST
+      parameterized_test_registry_(),
+      parameterized_tests_registered_(false),
+#endif  // GTEST_HAS_PARAM_TEST
+      last_death_test_case_(-1),
+      current_test_case_(NULL),
+      current_test_info_(NULL),
+      ad_hoc_test_result_(),
+      os_stack_trace_getter_(NULL),
+      post_flag_parse_init_performed_(false),
+      random_seed_(0),  // Will be overridden by the flag before first use.
+      random_(0),  // Will be reseeded before first use.
+      start_timestamp_(0),
+      elapsed_time_(0),
+#if GTEST_HAS_DEATH_TEST
+      death_test_factory_(new DefaultDeathTestFactory),
+#endif
+      // Will be overridden by the flag before first use.
+      catch_exceptions_(false) {
+  listeners()->SetDefaultResultPrinter(new PrettyUnitTestResultPrinter);
+}
+
+UnitTestImpl::~UnitTestImpl() {
+  // Deletes every TestCase.
+  ForEach(test_cases_, internal::Delete<TestCase>);
+
+  // Deletes every Environment.
+  ForEach(environments_, internal::Delete<Environment>);
+
+  delete os_stack_trace_getter_;
+}
+
+// Adds a TestProperty to the current TestResult object when invoked in a
+// context of a test, to current test case's ad_hoc_test_result when invoke
+// from SetUpTestCase/TearDownTestCase, or to the global property set
+// otherwise.  If the result already contains a property with the same key,
+// the value will be updated.
+void UnitTestImpl::RecordProperty(const TestProperty& test_property) {
+  std::string xml_element;
+  TestResult* test_result;  // TestResult appropriate for property recording.
+
+  if (current_test_info_ != NULL) {
+    xml_element = "testcase";
+    test_result = &(current_test_info_->result_);
+  } else if (current_test_case_ != NULL) {
+    xml_element = "testsuite";
+    test_result = &(current_test_case_->ad_hoc_test_result_);
+  } else {
+    xml_element = "testsuites";
+    test_result = &ad_hoc_test_result_;
+  }
+  test_result->RecordProperty(xml_element, test_property);
+}
+
+#if GTEST_HAS_DEATH_TEST
+// Disables event forwarding if the control is currently in a death test
+// subprocess. Must not be called before InitGoogleTest.
+void UnitTestImpl::SuppressTestEventsIfInSubprocess() {
+  if (internal_run_death_test_flag_.get() != NULL)
+    listeners()->SuppressEventForwarding();
+}
+#endif  // GTEST_HAS_DEATH_TEST
+
+// Initializes event listeners performing XML output as specified by
+// UnitTestOptions. Must not be called before InitGoogleTest.
+void UnitTestImpl::ConfigureXmlOutput() {
+  const std::string& output_format = UnitTestOptions::GetOutputFormat();
+  if (output_format == "xml") {
+    listeners()->SetDefaultXmlGenerator(new XmlUnitTestResultPrinter(
+        UnitTestOptions::GetAbsolutePathToOutputFile().c_str()));
+  } else if (output_format != "") {
+    printf("WARNING: unrecognized output format \"%s\" ignored.\n",
+           output_format.c_str());
+    fflush(stdout);
+  }
+}
+
+#if GTEST_CAN_STREAM_RESULTS_
+// Initializes event listeners for streaming test results in string form.
+// Must not be called before InitGoogleTest.
+void UnitTestImpl::ConfigureStreamingOutput() {
+  const std::string& target = GTEST_FLAG(stream_result_to);
+  if (!target.empty()) {
+    const size_t pos = target.find(':');
+    if (pos != std::string::npos) {
+      listeners()->Append(new StreamingListener(target.substr(0, pos),
+                                                target.substr(pos+1)));
+    } else {
+      printf("WARNING: unrecognized streaming target \"%s\" ignored.\n",
+             target.c_str());
+      fflush(stdout);
+    }
+  }
+}
+#endif  // GTEST_CAN_STREAM_RESULTS_
+
+// Performs initialization dependent upon flag values obtained in
+// ParseGoogleTestFlagsOnly.  Is called from InitGoogleTest after the call to
+// ParseGoogleTestFlagsOnly.  In case a user neglects to call InitGoogleTest
+// this function is also called from RunAllTests.  Since this function can be
+// called more than once, it has to be idempotent.
+void UnitTestImpl::PostFlagParsingInit() {
+  // Ensures that this function does not execute more than once.
+  if (!post_flag_parse_init_performed_) {
+    post_flag_parse_init_performed_ = true;
+
+#if defined(GTEST_CUSTOM_TEST_EVENT_LISTENER_)
+    // Register to send notifications about key process state changes.
+    listeners()->Append(new GTEST_CUSTOM_TEST_EVENT_LISTENER_());
+#endif  // defined(GTEST_CUSTOM_TEST_EVENT_LISTENER_)
+
+#if GTEST_HAS_DEATH_TEST
+    InitDeathTestSubprocessControlInfo();
+    SuppressTestEventsIfInSubprocess();
+#endif  // GTEST_HAS_DEATH_TEST
+
+    // Registers parameterized tests. This makes parameterized tests
+    // available to the UnitTest reflection API without running
+    // RUN_ALL_TESTS.
+    RegisterParameterizedTests();
+
+    // Configures listeners for XML output. This makes it possible for users
+    // to shut down the default XML output before invoking RUN_ALL_TESTS.
+    ConfigureXmlOutput();
+
+#if GTEST_CAN_STREAM_RESULTS_
+    // Configures listeners for streaming test results to the specified server.
+    ConfigureStreamingOutput();
+#endif  // GTEST_CAN_STREAM_RESULTS_
+  }
+}
+
+// A predicate that checks the name of a TestCase against a known
+// value.
+//
+// This is used for implementation of the UnitTest class only.  We put
+// it in the anonymous namespace to prevent polluting the outer
+// namespace.
+//
+// TestCaseNameIs is copyable.
+class TestCaseNameIs {
+ public:
+  // Constructor.
+  explicit TestCaseNameIs(const std::string& name)
+      : name_(name) {}
+
+  // Returns true iff the name of test_case matches name_.
+  bool operator()(const TestCase* test_case) const {
+    return test_case != NULL && strcmp(test_case->name(), name_.c_str()) == 0;
+  }
+
+ private:
+  std::string name_;
+};
+
+// Finds and returns a TestCase with the given name.  If one doesn't
+// exist, creates one and returns it.  It's the CALLER'S
+// RESPONSIBILITY to ensure that this function is only called WHEN THE
+// TESTS ARE NOT SHUFFLED.
+//
+// Arguments:
+//
+//   test_case_name: name of the test case
+//   type_param:     the name of the test case's type parameter, or NULL if
+//                   this is not a typed or a type-parameterized test case.
+//   set_up_tc:      pointer to the function that sets up the test case
+//   tear_down_tc:   pointer to the function that tears down the test case
+TestCase* UnitTestImpl::GetTestCase(const char* test_case_name,
+                                    const char* type_param,
+                                    Test::SetUpTestCaseFunc set_up_tc,
+                                    Test::TearDownTestCaseFunc tear_down_tc) {
+  // Can we find a TestCase with the given name?
+  const std::vector<TestCase*>::const_iterator test_case =
+      std::find_if(test_cases_.begin(), test_cases_.end(),
+                   TestCaseNameIs(test_case_name));
+
+  if (test_case != test_cases_.end())
+    return *test_case;
+
+  // No.  Let's create one.
+  TestCase* const new_test_case =
+      new TestCase(test_case_name, type_param, set_up_tc, tear_down_tc);
+
+  // Is this a death test case?
+  if (internal::UnitTestOptions::MatchesFilter(test_case_name,
+                                               kDeathTestCaseFilter)) {
+    // Yes.  Inserts the test case after the last death test case
+    // defined so far.  This only works when the test cases haven't
+    // been shuffled.  Otherwise we may end up running a death test
+    // after a non-death test.
+    ++last_death_test_case_;
+    test_cases_.insert(test_cases_.begin() + last_death_test_case_,
+                       new_test_case);
+  } else {
+    // No.  Appends to the end of the list.
+    test_cases_.push_back(new_test_case);
+  }
+
+  test_case_indices_.push_back(static_cast<int>(test_case_indices_.size()));
+  return new_test_case;
+}
+
+// Helpers for setting up / tearing down the given environment.  They
+// are for use in the ForEach() function.
+static void SetUpEnvironment(Environment* env) { env->SetUp(); }
+static void TearDownEnvironment(Environment* env) { env->TearDown(); }
+
+// Runs all tests in this UnitTest object, prints the result, and
+// returns true if all tests are successful.  If any exception is
+// thrown during a test, the test is considered to be failed, but the
+// rest of the tests will still be run.
+//
+// When parameterized tests are enabled, it expands and registers
+// parameterized tests first in RegisterParameterizedTests().
+// All other functions called from RunAllTests() may safely assume that
+// parameterized tests are ready to be counted and run.
+bool UnitTestImpl::RunAllTests() {
+  // Makes sure InitGoogleTest() was called.
+  if (!GTestIsInitialized()) {
+    printf("%s",
+           "\nThis test program did NOT call ::testing::InitGoogleTest "
+           "before calling RUN_ALL_TESTS().  Please fix it.\n");
+    return false;
+  }
+
+  // Do not run any test if the --help flag was specified.
+  if (g_help_flag)
+    return true;
+
+  // Repeats the call to the post-flag parsing initialization in case the
+  // user didn't call InitGoogleTest.
+  PostFlagParsingInit();
+
+  // Even if sharding is not on, test runners may want to use the
+  // GTEST_SHARD_STATUS_FILE to query whether the test supports the sharding
+  // protocol.
+  internal::WriteToShardStatusFileIfNeeded();
+
+  // True iff we are in a subprocess for running a thread-safe-style
+  // death test.
+  bool in_subprocess_for_death_test = false;
+
+#if GTEST_HAS_DEATH_TEST
+  in_subprocess_for_death_test = (internal_run_death_test_flag_.get() != NULL);
+# if defined(GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_)
+  if (in_subprocess_for_death_test) {
+    GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_();
+  }
+# endif  // defined(GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_)
+#endif  // GTEST_HAS_DEATH_TEST
+
+  const bool should_shard = ShouldShard(kTestTotalShards, kTestShardIndex,
+                                        in_subprocess_for_death_test);
+
+  // Compares the full test names with the filter to decide which
+  // tests to run.
+  const bool has_tests_to_run = FilterTests(should_shard
+                                              ? HONOR_SHARDING_PROTOCOL
+                                              : IGNORE_SHARDING_PROTOCOL) > 0;
+
+  // Lists the tests and exits if the --gtest_list_tests flag was specified.
+  if (GTEST_FLAG(list_tests)) {
+    // This must be called *after* FilterTests() has been called.
+    ListTestsMatchingFilter();
+    return true;
+  }
+
+  random_seed_ = GTEST_FLAG(shuffle) ?
+      GetRandomSeedFromFlag(GTEST_FLAG(random_seed)) : 0;
+
+  // True iff at least one test has failed.
+  bool failed = false;
+
+  TestEventListener* repeater = listeners()->repeater();
+
+  start_timestamp_ = GetTimeInMillis();
+  repeater->OnTestProgramStart(*parent_);
+
+  // How many times to repeat the tests?  We don't want to repeat them
+  // when we are inside the subprocess of a death test.
+  const int repeat = in_subprocess_for_death_test ? 1 : GTEST_FLAG(repeat);
+  // Repeats forever if the repeat count is negative.
+  const bool forever = repeat < 0;
+  for (int i = 0; forever || i != repeat; i++) {
+    // We want to preserve failures generated by ad-hoc test
+    // assertions executed before RUN_ALL_TESTS().
+    ClearNonAdHocTestResult();
+
+    const TimeInMillis start = GetTimeInMillis();
+
+    // Shuffles test cases and tests if requested.
+    if (has_tests_to_run && GTEST_FLAG(shuffle)) {
+      random()->Reseed(random_seed_);
+      // This should be done before calling OnTestIterationStart(),
+      // such that a test event listener can see the actual test order
+      // in the event.
+      ShuffleTests();
+    }
+
+    // Tells the unit test event listeners that the tests are about to start.
+    repeater->OnTestIterationStart(*parent_, i);
+
+    // Runs each test case if there is at least one test to run.
+    if (has_tests_to_run) {
+      // Sets up all environments beforehand.
+      repeater->OnEnvironmentsSetUpStart(*parent_);
+      ForEach(environments_, SetUpEnvironment);
+      repeater->OnEnvironmentsSetUpEnd(*parent_);
+
+      // Runs the tests only if there was no fatal failure or skip triggered
+      //during global set-up.
+      if (Test::IsSkipped()) {
+        // Emit diagnostics when global set-up calls skip, as it will not be
+        // emitted by default.
+        TestResult& test_result =
+            *internal::GetUnitTestImpl()->current_test_result();
+        for (int j = 0; j < test_result.total_part_count(); ++j) {
+          const TestPartResult& test_part_result =
+              test_result.GetTestPartResult(j);
+          if (test_part_result.type() == TestPartResult::kSkip) {
+            const std::string& result = test_part_result.message();
+            printf("%s\n", result.c_str());
+          }
+        }
+        fflush(stdout);
+      } else if (!Test::HasFatalFailure()) {
+        for (int test_index = 0; test_index < total_test_case_count();
+             test_index++) {
+          GetMutableTestCase(test_index)->Run();
+        }
+      }
+
+      // Tears down all environments in reverse order afterwards.
+      repeater->OnEnvironmentsTearDownStart(*parent_);
+      std::for_each(environments_.rbegin(), environments_.rend(),
+                    TearDownEnvironment);
+      repeater->OnEnvironmentsTearDownEnd(*parent_);
+    }
+
+    elapsed_time_ = GetTimeInMillis() - start;
+
+    // Tells the unit test event listener that the tests have just finished.
+    repeater->OnTestIterationEnd(*parent_, i);
+
+    // Gets the result and clears it.
+    if (!Passed()) {
+      failed = true;
+    }
+
+    // Restores the original test order after the iteration.  This
+    // allows the user to quickly repro a failure that happens in the
+    // N-th iteration without repeating the first (N - 1) iterations.
+    // This is not enclosed in "if (GTEST_FLAG(shuffle)) { ... }", in
+    // case the user somehow changes the value of the flag somewhere
+    // (it's always safe to unshuffle the tests).
+    UnshuffleTests();
+
+    if (GTEST_FLAG(shuffle)) {
+      // Picks a new random seed for each iteration.
+      random_seed_ = GetNextRandomSeed(random_seed_);
+    }
+  }
+
+  repeater->OnTestProgramEnd(*parent_);
+
+  return !failed;
+}
+
+// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file
+// if the variable is present. If a file already exists at this location, this
+// function will write over it. If the variable is present, but the file cannot
+// be created, prints an error and exits.
+void WriteToShardStatusFileIfNeeded() {
+  const char* const test_shard_file = posix::GetEnv(kTestShardStatusFile);
+  if (test_shard_file != NULL) {
+    FILE* const file = posix::FOpen(test_shard_file, "w");
+    if (file == NULL) {
+      ColoredPrintf(COLOR_RED,
+                    "Could not write to the test shard status file \"%s\" "
+                    "specified by the %s environment variable.\n",
+                    test_shard_file, kTestShardStatusFile);
+      fflush(stdout);
+      exit(EXIT_FAILURE);
+    }
+    fclose(file);
+  }
+}
+
+// Checks whether sharding is enabled by examining the relevant
+// environment variable values. If the variables are present,
+// but inconsistent (i.e., shard_index >= total_shards), prints
+// an error and exits. If in_subprocess_for_death_test, sharding is
+// disabled because it must only be applied to the original test
+// process. Otherwise, we could filter out death tests we intended to execute.
+bool ShouldShard(const char* total_shards_env,
+                 const char* shard_index_env,
+                 bool in_subprocess_for_death_test) {
+  if (in_subprocess_for_death_test) {
+    return false;
+  }
+
+  const Int32 total_shards = Int32FromEnvOrDie(total_shards_env, -1);
+  const Int32 shard_index = Int32FromEnvOrDie(shard_index_env, -1);
+
+  if (total_shards == -1 && shard_index == -1) {
+    return false;
+  } else if (total_shards == -1 && shard_index != -1) {
+    const Message msg = Message()
+      << "Invalid environment variables: you have "
+      << kTestShardIndex << " = " << shard_index
+      << ", but have left " << kTestTotalShards << " unset.\n";
+    ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str());
+    fflush(stdout);
+    exit(EXIT_FAILURE);
+  } else if (total_shards != -1 && shard_index == -1) {
+    const Message msg = Message()
+      << "Invalid environment variables: you have "
+      << kTestTotalShards << " = " << total_shards
+      << ", but have left " << kTestShardIndex << " unset.\n";
+    ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str());
+    fflush(stdout);
+    exit(EXIT_FAILURE);
+  } else if (shard_index < 0 || shard_index >= total_shards) {
+    const Message msg = Message()
+      << "Invalid environment variables: we require 0 <= "
+      << kTestShardIndex << " < " << kTestTotalShards
+      << ", but you have " << kTestShardIndex << "=" << shard_index
+      << ", " << kTestTotalShards << "=" << total_shards << ".\n";
+    ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str());
+    fflush(stdout);
+    exit(EXIT_FAILURE);
+  }
+
+  return total_shards > 1;
+}
+
+// Parses the environment variable var as an Int32. If it is unset,
+// returns default_val. If it is not an Int32, prints an error
+// and aborts.
+Int32 Int32FromEnvOrDie(const char* var, Int32 default_val) {
+  const char* str_val = posix::GetEnv(var);
+  if (str_val == NULL) {
+    return default_val;
+  }
+
+  Int32 result;
+  if (!ParseInt32(Message() << "The value of environment variable " << var,
+                  str_val, &result)) {
+    exit(EXIT_FAILURE);
+  }
+  return result;
+}
+
+// Given the total number of shards, the shard index, and the test id,
+// returns true iff the test should be run on this shard. The test id is
+// some arbitrary but unique non-negative integer assigned to each test
+// method. Assumes that 0 <= shard_index < total_shards.
+bool ShouldRunTestOnShard(int total_shards, int shard_index, int test_id) {
+  return (test_id % total_shards) == shard_index;
+}
+
+// Compares the name of each test with the user-specified filter to
+// decide whether the test should be run, then records the result in
+// each TestCase and TestInfo object.
+// If shard_tests == true, further filters tests based on sharding
+// variables in the environment - see
+// http://code.google.com/p/googletest/wiki/GoogleTestAdvancedGuide.
+// Returns the number of tests that should run.
+int UnitTestImpl::FilterTests(ReactionToSharding shard_tests) {
+  const Int32 total_shards = shard_tests == HONOR_SHARDING_PROTOCOL ?
+      Int32FromEnvOrDie(kTestTotalShards, -1) : -1;
+  const Int32 shard_index = shard_tests == HONOR_SHARDING_PROTOCOL ?
+      Int32FromEnvOrDie(kTestShardIndex, -1) : -1;
+
+  // num_runnable_tests are the number of tests that will
+  // run across all shards (i.e., match filter and are not disabled).
+  // num_selected_tests are the number of tests to be run on
+  // this shard.
+  int num_runnable_tests = 0;
+  int num_selected_tests = 0;
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    TestCase* const test_case = test_cases_[i];
+    const std::string &test_case_name = test_case->name();
+    test_case->set_should_run(false);
+
+    for (size_t j = 0; j < test_case->test_info_list().size(); j++) {
+      TestInfo* const test_info = test_case->test_info_list()[j];
+      const std::string test_name(test_info->name());
+      // A test is disabled if test case name or test name matches
+      // kDisableTestFilter.
+      const bool is_disabled =
+          internal::UnitTestOptions::MatchesFilter(test_case_name,
+                                                   kDisableTestFilter) ||
+          internal::UnitTestOptions::MatchesFilter(test_name,
+                                                   kDisableTestFilter);
+      test_info->is_disabled_ = is_disabled;
+
+      const bool matches_filter =
+          internal::UnitTestOptions::FilterMatchesTest(test_case_name,
+                                                       test_name);
+      test_info->matches_filter_ = matches_filter;
+
+      const bool is_runnable =
+          (GTEST_FLAG(also_run_disabled_tests) || !is_disabled) &&
+          matches_filter;
+
+      const bool is_selected = is_runnable &&
+          (shard_tests == IGNORE_SHARDING_PROTOCOL ||
+           ShouldRunTestOnShard(total_shards, shard_index,
+                                num_runnable_tests));
+
+      num_runnable_tests += is_runnable;
+      num_selected_tests += is_selected;
+
+      test_info->should_run_ = is_selected;
+      test_case->set_should_run(test_case->should_run() || is_selected);
+    }
+  }
+  return num_selected_tests;
+}
+
+// Prints the given C-string on a single line by replacing all '\n'
+// characters with string "\\n".  If the output takes more than
+// max_length characters, only prints the first max_length characters
+// and "...".
+static void PrintOnOneLine(const char* str, int max_length) {
+  if (str != NULL) {
+    for (int i = 0; *str != '\0'; ++str) {
+      if (i >= max_length) {
+        printf("...");
+        break;
+      }
+      if (*str == '\n') {
+        printf("\\n");
+        i += 2;
+      } else {
+        printf("%c", *str);
+        ++i;
+      }
+    }
+  }
+}
+
+// Prints the names of the tests matching the user-specified filter flag.
+void UnitTestImpl::ListTestsMatchingFilter() {
+  // Print at most this many characters for each type/value parameter.
+  const int kMaxParamLength = 250;
+
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    const TestCase* const test_case = test_cases_[i];
+    bool printed_test_case_name = false;
+
+    for (size_t j = 0; j < test_case->test_info_list().size(); j++) {
+      const TestInfo* const test_info =
+          test_case->test_info_list()[j];
+      if (test_info->matches_filter_) {
+        if (!printed_test_case_name) {
+          printed_test_case_name = true;
+          printf("%s.", test_case->name());
+          if (test_case->type_param() != NULL) {
+            printf("  # %s = ", kTypeParamLabel);
+            // We print the type parameter on a single line to make
+            // the output easy to parse by a program.
+            PrintOnOneLine(test_case->type_param(), kMaxParamLength);
+          }
+          printf("\n");
+        }
+        printf("  %s", test_info->name());
+        printf("\n");
+      }
+    }
+  }
+  fflush(stdout);
+}
+
+// Sets the OS stack trace getter.
+//
+// Does nothing if the input and the current OS stack trace getter are
+// the same; otherwise, deletes the old getter and makes the input the
+// current getter.
+void UnitTestImpl::set_os_stack_trace_getter(
+    OsStackTraceGetterInterface* getter) {
+  if (os_stack_trace_getter_ != getter) {
+    delete os_stack_trace_getter_;
+    os_stack_trace_getter_ = getter;
+  }
+}
+
+// Returns the current OS stack trace getter if it is not NULL;
+// otherwise, creates an OsStackTraceGetter, makes it the current
+// getter, and returns it.
+OsStackTraceGetterInterface* UnitTestImpl::os_stack_trace_getter() {
+  if (os_stack_trace_getter_ == NULL) {
+#ifdef GTEST_OS_STACK_TRACE_GETTER_
+    os_stack_trace_getter_ = new GTEST_OS_STACK_TRACE_GETTER_;
+#else
+    os_stack_trace_getter_ = new OsStackTraceGetter;
+#endif  // GTEST_OS_STACK_TRACE_GETTER_
+  }
+
+  return os_stack_trace_getter_;
+}
+
+// Returns the TestResult for the test that's currently running, or
+// the TestResult for the ad hoc test if no test is running.
+TestResult* UnitTestImpl::current_test_result() {
+  return current_test_info_ ?
+      &(current_test_info_->result_) : &ad_hoc_test_result_;
+}
+
+// Shuffles all test cases, and the tests within each test case,
+// making sure that death tests are still run first.
+void UnitTestImpl::ShuffleTests() {
+  // Shuffles the death test cases.
+  ShuffleRange(random(), 0, last_death_test_case_ + 1, &test_case_indices_);
+
+  // Shuffles the non-death test cases.
+  ShuffleRange(random(), last_death_test_case_ + 1,
+               static_cast<int>(test_cases_.size()), &test_case_indices_);
+
+  // Shuffles the tests inside each test case.
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    test_cases_[i]->ShuffleTests(random());
+  }
+}
+
+// Restores the test cases and tests to their order before the first shuffle.
+void UnitTestImpl::UnshuffleTests() {
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    // Unshuffles the tests in each test case.
+    test_cases_[i]->UnshuffleTests();
+    // Resets the index of each test case.
+    test_case_indices_[i] = static_cast<int>(i);
+  }
+}
+
+// Returns the current OS stack trace as an std::string.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag.  The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in
+// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't.
+std::string GetCurrentOsStackTraceExceptTop(UnitTest* /*unit_test*/,
+                                            int skip_count) {
+  // We pass skip_count + 1 to skip this wrapper function in addition
+  // to what the user really wants to skip.
+  return GetUnitTestImpl()->CurrentOsStackTraceExceptTop(skip_count + 1);
+}
+
+// Used by the GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_ macro to
+// suppress unreachable code warnings.
+namespace {
+class ClassUniqueToAlwaysTrue {};
+}
+
+bool IsTrue(bool condition) { return condition; }
+
+bool AlwaysTrue() {
+#if GTEST_HAS_EXCEPTIONS
+  // This condition is always false so AlwaysTrue() never actually throws,
+  // but it makes the compiler think that it may throw.
+  if (IsTrue(false))
+    throw ClassUniqueToAlwaysTrue();
+#endif  // GTEST_HAS_EXCEPTIONS
+  return true;
+}
+
+// If *pstr starts with the given prefix, modifies *pstr to be right
+// past the prefix and returns true; otherwise leaves *pstr unchanged
+// and returns false.  None of pstr, *pstr, and prefix can be NULL.
+bool SkipPrefix(const char* prefix, const char** pstr) {
+  const size_t prefix_len = strlen(prefix);
+  if (strncmp(*pstr, prefix, prefix_len) == 0) {
+    *pstr += prefix_len;
+    return true;
+  }
+  return false;
+}
+
+// Parses a string as a command line flag.  The string should have
+// the format "--flag=value".  When def_optional is true, the "=value"
+// part can be omitted.
+//
+// Returns the value of the flag, or NULL if the parsing failed.
+const char* ParseFlagValue(const char* str,
+                           const char* flag,
+                           bool def_optional) {
+  // str and flag must not be NULL.
+  if (str == NULL || flag == NULL) return NULL;
+
+  // The flag must start with "--" followed by GTEST_FLAG_PREFIX_.
+  const std::string flag_str = std::string("--") + GTEST_FLAG_PREFIX_ + flag;
+  const size_t flag_len = flag_str.length();
+  if (strncmp(str, flag_str.c_str(), flag_len) != 0) return NULL;
+
+  // Skips the flag name.
+  const char* flag_end = str + flag_len;
+
+  // When def_optional is true, it's OK to not have a "=value" part.
+  if (def_optional && (flag_end[0] == '\0')) {
+    return flag_end;
+  }
+
+  // If def_optional is true and there are more characters after the
+  // flag name, or if def_optional is false, there must be a '=' after
+  // the flag name.
+  if (flag_end[0] != '=') return NULL;
+
+  // Returns the string after "=".
+  return flag_end + 1;
+}
+
+// Parses a string for a bool flag, in the form of either
+// "--flag=value" or "--flag".
+//
+// In the former case, the value is taken as true as long as it does
+// not start with '0', 'f', or 'F'.
+//
+// In the latter case, the value is taken as true.
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+bool ParseBoolFlag(const char* str, const char* flag, bool* value) {
+  // Gets the value of the flag as a string.
+  const char* const value_str = ParseFlagValue(str, flag, true);
+
+  // Aborts if the parsing failed.
+  if (value_str == NULL) return false;
+
+  // Converts the string value to a bool.
+  *value = !(*value_str == '0' || *value_str == 'f' || *value_str == 'F');
+  return true;
+}
+
+// Parses a string for an Int32 flag, in the form of
+// "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+bool ParseInt32Flag(const char* str, const char* flag, Int32* value) {
+  // Gets the value of the flag as a string.
+  const char* const value_str = ParseFlagValue(str, flag, false);
+
+  // Aborts if the parsing failed.
+  if (value_str == NULL) return false;
+
+  // Sets *value to the value of the flag.
+  return ParseInt32(Message() << "The value of flag --" << flag,
+                    value_str, value);
+}
+
+// Parses a string for a string flag, in the form of
+// "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+bool ParseStringFlag(const char* str, const char* flag, std::string* value) {
+  // Gets the value of the flag as a string.
+  const char* const value_str = ParseFlagValue(str, flag, false);
+
+  // Aborts if the parsing failed.
+  if (value_str == NULL) return false;
+
+  // Sets *value to the value of the flag.
+  *value = value_str;
+  return true;
+}
+
+// Determines whether a string has a prefix that Google Test uses for its
+// flags, i.e., starts with GTEST_FLAG_PREFIX_ or GTEST_FLAG_PREFIX_DASH_.
+// If Google Test detects that a command line flag has its prefix but is not
+// recognized, it will print its help message. Flags starting with
+// GTEST_INTERNAL_PREFIX_ followed by "internal_" are considered Google Test
+// internal flags and do not trigger the help message.
+static bool HasGoogleTestFlagPrefix(const char* str) {
+  return (SkipPrefix("--", &str) ||
+          SkipPrefix("-", &str) ||
+          SkipPrefix("/", &str)) &&
+         !SkipPrefix(GTEST_FLAG_PREFIX_ "internal_", &str) &&
+         (SkipPrefix(GTEST_FLAG_PREFIX_, &str) ||
+          SkipPrefix(GTEST_FLAG_PREFIX_DASH_, &str));
+}
+
+// Prints a string containing code-encoded text.  The following escape
+// sequences can be used in the string to control the text color:
+//
+//   @@    prints a single '@' character.
+//   @R    changes the color to red.
+//   @G    changes the color to green.
+//   @Y    changes the color to yellow.
+//   @D    changes to the default terminal text color.
+//
+// TODO(wan@google.com): Write tests for this once we add stdout
+// capturing to Google Test.
+static void PrintColorEncoded(const char* str) {
+  GTestColor color = COLOR_DEFAULT;  // The current color.
+
+  // Conceptually, we split the string into segments divided by escape
+  // sequences.  Then we print one segment at a time.  At the end of
+  // each iteration, the str pointer advances to the beginning of the
+  // next segment.
+  for (;;) {
+    const char* p = strchr(str, '@');
+    if (p == NULL) {
+      ColoredPrintf(color, "%s", str);
+      return;
+    }
+
+    ColoredPrintf(color, "%s", std::string(str, p).c_str());
+
+    const char ch = p[1];
+    str = p + 2;
+    if (ch == '@') {
+      ColoredPrintf(color, "@");
+    } else if (ch == 'D') {
+      color = COLOR_DEFAULT;
+    } else if (ch == 'R') {
+      color = COLOR_RED;
+    } else if (ch == 'G') {
+      color = COLOR_GREEN;
+    } else if (ch == 'Y') {
+      color = COLOR_YELLOW;
+    } else {
+      --str;
+    }
+  }
+}
+
+static const char kColorEncodedHelpMessage[] =
+"This program contains tests written using " GTEST_NAME_ ". You can use the\n"
+"following command line flags to control its behavior:\n"
+"\n"
+"Test Selection:\n"
+"  @G--" GTEST_FLAG_PREFIX_ "list_tests@D\n"
+"      List the names of all tests instead of running them. The name of\n"
+"      TEST(Foo, Bar) is \"Foo.Bar\".\n"
+"  @G--" GTEST_FLAG_PREFIX_ "filter=@YPOSTIVE_PATTERNS"
+    "[@G-@YNEGATIVE_PATTERNS]@D\n"
+"      Run only the tests whose name matches one of the positive patterns but\n"
+"      none of the negative patterns. '?' matches any single character; '*'\n"
+"      matches any substring; ':' separates two patterns.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "also_run_disabled_tests@D\n"
+"      Run all disabled tests too.\n"
+"\n"
+"Test Execution:\n"
+"  @G--" GTEST_FLAG_PREFIX_ "repeat=@Y[COUNT]@D\n"
+"      Run the tests repeatedly; use a negative count to repeat forever.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "shuffle@D\n"
+"      Randomize tests' orders on every iteration.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "random_seed=@Y[NUMBER]@D\n"
+"      Random number seed to use for shuffling test orders (between 1 and\n"
+"      99999, or 0 to use a seed based on the current time).\n"
+"\n"
+"Test Output:\n"
+"  @G--" GTEST_FLAG_PREFIX_ "color=@Y(@Gyes@Y|@Gno@Y|@Gauto@Y)@D\n"
+"      Enable/disable colored output. The default is @Gauto@D.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "print_time=0@D\n"
+"      Don't print the elapsed time of each test.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "output=xml@Y[@G:@YDIRECTORY_PATH@G"
+    GTEST_PATH_SEP_ "@Y|@G:@YFILE_PATH]@D\n"
+"      Generate an XML report in the given directory or with the given file\n"
+"      name. @YFILE_PATH@D defaults to @Gtest_details.xml@D.\n"
+#if GTEST_CAN_STREAM_RESULTS_
+"  @G--" GTEST_FLAG_PREFIX_ "stream_result_to=@YHOST@G:@YPORT@D\n"
+"      Stream test results to the given server.\n"
+#endif  // GTEST_CAN_STREAM_RESULTS_
+"\n"
+"Assertion Behavior:\n"
+#if GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS
+"  @G--" GTEST_FLAG_PREFIX_ "death_test_style=@Y(@Gfast@Y|@Gthreadsafe@Y)@D\n"
+"      Set the default death test style.\n"
+#endif  // GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS
+"  @G--" GTEST_FLAG_PREFIX_ "break_on_failure@D\n"
+"      Turn assertion failures into debugger break-points.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "throw_on_failure@D\n"
+"      Turn assertion failures into C++ exceptions.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "catch_exceptions=0@D\n"
+"      Do not report exceptions as test failures. Instead, allow them\n"
+"      to crash the program or throw a pop-up (on Windows).\n"
+"\n"
+"Except for @G--" GTEST_FLAG_PREFIX_ "list_tests@D, you can alternatively set "
+    "the corresponding\n"
+"environment variable of a flag (all letters in upper-case). For example, to\n"
+"disable colored text output, you can either specify @G--" GTEST_FLAG_PREFIX_
+    "color=no@D or set\n"
+"the @G" GTEST_FLAG_PREFIX_UPPER_ "COLOR@D environment variable to @Gno@D.\n"
+"\n"
+"For more information, please read the " GTEST_NAME_ " documentation at\n"
+"@G" GTEST_PROJECT_URL_ "@D. If you find a bug in " GTEST_NAME_ "\n"
+"(not one in your own code or tests), please report it to\n"
+"@G<" GTEST_DEV_EMAIL_ ">@D.\n";
+
+bool ParseGoogleTestFlag(const char* const arg) {
+  return ParseBoolFlag(arg, kAlsoRunDisabledTestsFlag,
+                       &GTEST_FLAG(also_run_disabled_tests)) ||
+      ParseBoolFlag(arg, kBreakOnFailureFlag,
+                    &GTEST_FLAG(break_on_failure)) ||
+      ParseBoolFlag(arg, kCatchExceptionsFlag,
+                    &GTEST_FLAG(catch_exceptions)) ||
+      ParseStringFlag(arg, kColorFlag, &GTEST_FLAG(color)) ||
+      ParseStringFlag(arg, kDeathTestStyleFlag,
+                      &GTEST_FLAG(death_test_style)) ||
+      ParseBoolFlag(arg, kDeathTestUseFork,
+                    &GTEST_FLAG(death_test_use_fork)) ||
+      ParseStringFlag(arg, kFilterFlag, &GTEST_FLAG(filter)) ||
+      ParseStringFlag(arg, kInternalRunDeathTestFlag,
+                      &GTEST_FLAG(internal_run_death_test)) ||
+      ParseBoolFlag(arg, kListTestsFlag, &GTEST_FLAG(list_tests)) ||
+      ParseStringFlag(arg, kOutputFlag, &GTEST_FLAG(output)) ||
+      ParseBoolFlag(arg, kPrintTimeFlag, &GTEST_FLAG(print_time)) ||
+      ParseInt32Flag(arg, kRandomSeedFlag, &GTEST_FLAG(random_seed)) ||
+      ParseInt32Flag(arg, kRepeatFlag, &GTEST_FLAG(repeat)) ||
+      ParseBoolFlag(arg, kShuffleFlag, &GTEST_FLAG(shuffle)) ||
+      ParseInt32Flag(arg, kStackTraceDepthFlag,
+                     &GTEST_FLAG(stack_trace_depth)) ||
+      ParseStringFlag(arg, kStreamResultToFlag,
+                      &GTEST_FLAG(stream_result_to)) ||
+      ParseBoolFlag(arg, kThrowOnFailureFlag,
+                    &GTEST_FLAG(throw_on_failure));
+}
+
+#if GTEST_USE_OWN_FLAGFILE_FLAG_
+void LoadFlagsFromFile(const std::string& path) {
+  FILE* flagfile = posix::FOpen(path.c_str(), "r");
+  if (!flagfile) {
+    fprintf(stderr,
+            "Unable to open file \"%s\"\n",
+            GTEST_FLAG(flagfile).c_str());
+    fflush(stderr);
+    exit(EXIT_FAILURE);
+  }
+  std::string contents(ReadEntireFile(flagfile));
+  posix::FClose(flagfile);
+  std::vector<std::string> lines;
+  SplitString(contents, '\n', &lines);
+  for (size_t i = 0; i < lines.size(); ++i) {
+    if (lines[i].empty())
+      continue;
+    if (!ParseGoogleTestFlag(lines[i].c_str()))
+      g_help_flag = true;
+  }
+}
+#endif  // GTEST_USE_OWN_FLAGFILE_FLAG_
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.  The type parameter CharType can be
+// instantiated to either char or wchar_t.
+template <typename CharType>
+void ParseGoogleTestFlagsOnlyImpl(int* argc, CharType** argv) {
+  for (int i = 1; i < *argc; i++) {
+    const std::string arg_string = StreamableToString(argv[i]);
+    const char* const arg = arg_string.c_str();
+
+    using internal::ParseBoolFlag;
+    using internal::ParseInt32Flag;
+    using internal::ParseStringFlag;
+
+    bool remove_flag = false;
+    if (ParseGoogleTestFlag(arg)) {
+      remove_flag = true;
+#if GTEST_USE_OWN_FLAGFILE_FLAG_
+    } else if (ParseStringFlag(arg, kFlagfileFlag, &GTEST_FLAG(flagfile))) {
+      LoadFlagsFromFile(GTEST_FLAG(flagfile));
+      remove_flag = true;
+#endif  // GTEST_USE_OWN_FLAGFILE_FLAG_
+    } else if (arg_string == "--help" || arg_string == "-h" ||
+               arg_string == "-?" || arg_string == "/?" ||
+               HasGoogleTestFlagPrefix(arg)) {
+      // Both help flag and unrecognized Google Test flags (excluding
+      // internal ones) trigger help display.
+      g_help_flag = true;
+    }
+
+    if (remove_flag) {
+      // Shift the remainder of the argv list left by one.  Note
+      // that argv has (*argc + 1) elements, the last one always being
+      // NULL.  The following loop moves the trailing NULL element as
+      // well.
+      for (int j = i; j != *argc; j++) {
+        argv[j] = argv[j + 1];
+      }
+
+      // Decrements the argument count.
+      (*argc)--;
+
+      // We also need to decrement the iterator as we just removed
+      // an element.
+      i--;
+    }
+  }
+
+  if (g_help_flag) {
+    // We print the help here instead of in RUN_ALL_TESTS(), as the
+    // latter may not be called at all if the user is using Google
+    // Test with another testing framework.
+    PrintColorEncoded(kColorEncodedHelpMessage);
+  }
+}
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.
+void ParseGoogleTestFlagsOnly(int* argc, char** argv) {
+  ParseGoogleTestFlagsOnlyImpl(argc, argv);
+}
+void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv) {
+  ParseGoogleTestFlagsOnlyImpl(argc, argv);
+}
+
+// The internal implementation of InitGoogleTest().
+//
+// The type parameter CharType can be instantiated to either char or
+// wchar_t.
+template <typename CharType>
+void InitGoogleTestImpl(int* argc, CharType** argv) {
+  // We don't want to run the initialization code twice.
+  if (GTestIsInitialized()) return;
+
+  if (*argc <= 0) return;
+
+  g_argvs.clear();
+  for (int i = 0; i != *argc; i++) {
+    g_argvs.push_back(StreamableToString(argv[i]));
+  }
+
+  ParseGoogleTestFlagsOnly(argc, argv);
+  GetUnitTestImpl()->PostFlagParsingInit();
+}
+
+}  // namespace internal
+
+// Initializes Google Test.  This must be called before calling
+// RUN_ALL_TESTS().  In particular, it parses a command line for the
+// flags that Google Test recognizes.  Whenever a Google Test flag is
+// seen, it is removed from argv, and *argc is decremented.
+//
+// No value is returned.  Instead, the Google Test flag variables are
+// updated.
+//
+// Calling the function for the second time has no user-visible effect.
+void InitGoogleTest(int* argc, char** argv) {
+#if defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_(argc, argv);
+#else  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  internal::InitGoogleTestImpl(argc, argv);
+#endif  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+}
+
+// This overloaded version can be used in Windows programs compiled in
+// UNICODE mode.
+void InitGoogleTest(int* argc, wchar_t** argv) {
+#if defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_(argc, argv);
+#else  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  internal::InitGoogleTestImpl(argc, argv);
+#endif  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+}
+
+std::string TempDir() {
+#if defined(GTEST_CUSTOM_TEMPDIR_FUNCTION_)
+    return GTEST_CUSTOM_TEMPDIR_FUNCTION_();
+#endif
+#if GTEST_OS_WINDOWS_MOBILE
+  return "\\temp\\";
+#elif GTEST_OS_WINDOWS
+  const char* temp_dir = internal::posix::GetEnv("TEMP");
+  if (temp_dir == NULL || temp_dir[0] == '\0')
+    return "\\temp\\";
+  else if (temp_dir[strlen(temp_dir) - 1] == '\\')
+    return temp_dir;
+  else
+    return std::string(temp_dir) + "\\";
+#elif GTEST_OS_LINUX_ANDROID
+  return "/sdcard/";
+#else
+  return "/tmp/";
+#endif  // GTEST_OS_WINDOWS_MOBILE
+}
+
+}  // namespace testing
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan), vladl@google.com (Vlad Losev)
+//
+// This file implements death tests.
+
+
+#if GTEST_HAS_DEATH_TEST
+
+# if GTEST_OS_MAC
+#  include <crt_externs.h>
+# endif  // GTEST_OS_MAC
+
+# include <errno.h>
+# include <fcntl.h>
+# include <limits.h>
+
+# if GTEST_OS_LINUX
+#  include <signal.h>
+# endif  // GTEST_OS_LINUX
+
+# include <stdarg.h>
+
+# if GTEST_OS_WINDOWS
+#  include <windows.h>
+# else
+#  include <sys/mman.h>
+#  include <sys/wait.h>
+# endif  // GTEST_OS_WINDOWS
+
+# if GTEST_OS_QNX
+#  include <spawn.h>
+# endif  // GTEST_OS_QNX
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick exists to
+// prevent the accidental inclusion of gtest-internal-inl.h in the
+// user's code.
+#define GTEST_IMPLEMENTATION_ 1
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+
+// Constants.
+
+// The default death test style.
+static const char kDefaultDeathTestStyle[] = "fast";
+
+GTEST_DEFINE_string_(
+    death_test_style,
+    internal::StringFromGTestEnv("death_test_style", kDefaultDeathTestStyle),
+    "Indicates how to run a death test in a forked child process: "
+    "\"threadsafe\" (child process re-executes the test binary "
+    "from the beginning, running only the specific death test) or "
+    "\"fast\" (child process runs the death test immediately "
+    "after forking).");
+
+GTEST_DEFINE_bool_(
+    death_test_use_fork,
+    internal::BoolFromGTestEnv("death_test_use_fork", false),
+    "Instructs to use fork()/_exit() instead of clone() in death tests. "
+    "Ignored and always uses fork() on POSIX systems where clone() is not "
+    "implemented. Useful when running under valgrind or similar tools if "
+    "those do not support clone(). Valgrind 3.3.1 will just fail if "
+    "it sees an unsupported combination of clone() flags. "
+    "It is not recommended to use this flag w/o valgrind though it will "
+    "work in 99% of the cases. Once valgrind is fixed, this flag will "
+    "most likely be removed.");
+
+namespace internal {
+GTEST_DEFINE_string_(
+    internal_run_death_test, "",
+    "Indicates the file, line number, temporal index of "
+    "the single death test to run, and a file descriptor to "
+    "which a success code may be sent, all separated by "
+    "the '|' characters.  This flag is specified if and only if the current "
+    "process is a sub-process launched for running a thread-safe "
+    "death test.  FOR INTERNAL USE ONLY.");
+}  // namespace internal
+
+#if GTEST_HAS_DEATH_TEST
+
+namespace internal {
+
+// Valid only for fast death tests. Indicates the code is running in the
+// child process of a fast style death test.
+# if !GTEST_OS_WINDOWS
+static bool g_in_fast_death_test_child = false;
+# endif
+
+// Returns a Boolean value indicating whether the caller is currently
+// executing in the context of the death test child process.  Tools such as
+// Valgrind heap checkers may need this to modify their behavior in death
+// tests.  IMPORTANT: This is an internal utility.  Using it may break the
+// implementation of death tests.  User code MUST NOT use it.
+bool InDeathTestChild() {
+# if GTEST_OS_WINDOWS
+
+  // On Windows, death tests are thread-safe regardless of the value of the
+  // death_test_style flag.
+  return !GTEST_FLAG(internal_run_death_test).empty();
+
+# else
+
+  if (GTEST_FLAG(death_test_style) == "threadsafe")
+    return !GTEST_FLAG(internal_run_death_test).empty();
+  else
+    return g_in_fast_death_test_child;
+#endif
+}
+
+}  // namespace internal
+
+// ExitedWithCode constructor.
+ExitedWithCode::ExitedWithCode(int exit_code) : exit_code_(exit_code) {
+}
+
+// ExitedWithCode function-call operator.
+bool ExitedWithCode::operator()(int exit_status) const {
+# if GTEST_OS_WINDOWS
+
+  return exit_status == exit_code_;
+
+# else
+
+  return WIFEXITED(exit_status) && WEXITSTATUS(exit_status) == exit_code_;
+
+# endif  // GTEST_OS_WINDOWS
+}
+
+# if !GTEST_OS_WINDOWS
+// KilledBySignal constructor.
+KilledBySignal::KilledBySignal(int signum) : signum_(signum) {
+}
+
+// KilledBySignal function-call operator.
+bool KilledBySignal::operator()(int exit_status) const {
+#  if defined(GTEST_KILLED_BY_SIGNAL_OVERRIDE_)
+  {
+    bool result;
+    if (GTEST_KILLED_BY_SIGNAL_OVERRIDE_(signum_, exit_status, &result)) {
+      return result;
+    }
+  }
+#  endif  // defined(GTEST_KILLED_BY_SIGNAL_OVERRIDE_)
+  return WIFSIGNALED(exit_status) && WTERMSIG(exit_status) == signum_;
+}
+# endif  // !GTEST_OS_WINDOWS
+
+namespace internal {
+
+// Utilities needed for death tests.
+
+// Generates a textual description of a given exit code, in the format
+// specified by wait(2).
+static std::string ExitSummary(int exit_code) {
+  Message m;
+
+# if GTEST_OS_WINDOWS
+
+  m << "Exited with exit status " << exit_code;
+
+# else
+
+  if (WIFEXITED(exit_code)) {
+    m << "Exited with exit status " << WEXITSTATUS(exit_code);
+  } else if (WIFSIGNALED(exit_code)) {
+    m << "Terminated by signal " << WTERMSIG(exit_code);
+  }
+#  ifdef WCOREDUMP
+  if (WCOREDUMP(exit_code)) {
+    m << " (core dumped)";
+  }
+#  endif
+# endif  // GTEST_OS_WINDOWS
+
+  return m.GetString();
+}
+
+// Returns true if exit_status describes a process that was terminated
+// by a signal, or exited normally with a nonzero exit code.
+bool ExitedUnsuccessfully(int exit_status) {
+  return !ExitedWithCode(0)(exit_status);
+}
+
+# if !GTEST_OS_WINDOWS
+// Generates a textual failure message when a death test finds more than
+// one thread running, or cannot determine the number of threads, prior
+// to executing the given statement.  It is the responsibility of the
+// caller not to pass a thread_count of 1.
+static std::string DeathTestThreadWarning(size_t thread_count) {
+  Message msg;
+  msg << "Death tests use fork(), which is unsafe particularly"
+      << " in a threaded context. For this test, " << GTEST_NAME_ << " ";
+  if (thread_count == 0)
+    msg << "couldn't detect the number of threads.";
+  else
+    msg << "detected " << thread_count << " threads.";
+  return msg.GetString();
+}
+# endif  // !GTEST_OS_WINDOWS
+
+// Flag characters for reporting a death test that did not die.
+static const char kDeathTestLived = 'L';
+static const char kDeathTestReturned = 'R';
+static const char kDeathTestThrew = 'T';
+static const char kDeathTestInternalError = 'I';
+
+// An enumeration describing all of the possible ways that a death test can
+// conclude.  DIED means that the process died while executing the test
+// code; LIVED means that process lived beyond the end of the test code;
+// RETURNED means that the test statement attempted to execute a return
+// statement, which is not allowed; THREW means that the test statement
+// returned control by throwing an exception.  IN_PROGRESS means the test
+// has not yet concluded.
+// TODO(vladl@google.com): Unify names and possibly values for
+// AbortReason, DeathTestOutcome, and flag characters above.
+enum DeathTestOutcome { IN_PROGRESS, DIED, LIVED, RETURNED, THREW };
+
+// Routine for aborting the program which is safe to call from an
+// exec-style death test child process, in which case the error
+// message is propagated back to the parent process.  Otherwise, the
+// message is simply printed to stderr.  In either case, the program
+// then exits with status 1.
+void DeathTestAbort(const std::string& message) {
+  // On a POSIX system, this function may be called from a threadsafe-style
+  // death test child process, which operates on a very small stack.  Use
+  // the heap for any additional non-minuscule memory requirements.
+  const InternalRunDeathTestFlag* const flag =
+      GetUnitTestImpl()->internal_run_death_test_flag();
+  if (flag != NULL) {
+    FILE* parent = posix::FDOpen(flag->write_fd(), "w");
+    fputc(kDeathTestInternalError, parent);
+    fprintf(parent, "%s", message.c_str());
+    fflush(parent);
+    _exit(1);
+  } else {
+    fprintf(stderr, "%s", message.c_str());
+    fflush(stderr);
+    posix::Abort();
+  }
+}
+
+// A replacement for CHECK that calls DeathTestAbort if the assertion
+// fails.
+# define GTEST_DEATH_TEST_CHECK_(expression) \
+  do { \
+    if (!::testing::internal::IsTrue(expression)) { \
+      DeathTestAbort( \
+          ::std::string("CHECK failed: File ") + __FILE__ +  ", line " \
+          + ::testing::internal::StreamableToString(__LINE__) + ": " \
+          + #expression); \
+    } \
+  } while (::testing::internal::AlwaysFalse())
+
+// This macro is similar to GTEST_DEATH_TEST_CHECK_, but it is meant for
+// evaluating any system call that fulfills two conditions: it must return
+// -1 on failure, and set errno to EINTR when it is interrupted and
+// should be tried again.  The macro expands to a loop that repeatedly
+// evaluates the expression as long as it evaluates to -1 and sets
+// errno to EINTR.  If the expression evaluates to -1 but errno is
+// something other than EINTR, DeathTestAbort is called.
+# define GTEST_DEATH_TEST_CHECK_SYSCALL_(expression) \
+  do { \
+    int gtest_retval; \
+    do { \
+      gtest_retval = (expression); \
+    } while (gtest_retval == -1 && errno == EINTR); \
+    if (gtest_retval == -1) { \
+      DeathTestAbort( \
+          ::std::string("CHECK failed: File ") + __FILE__ + ", line " \
+          + ::testing::internal::StreamableToString(__LINE__) + ": " \
+          + #expression + " != -1"); \
+    } \
+  } while (::testing::internal::AlwaysFalse())
+
+// Returns the message describing the last system error in errno.
+std::string GetLastErrnoDescription() {
+    return errno == 0 ? "" : posix::StrError(errno);
+}
+
+// This is called from a death test parent process to read a failure
+// message from the death test child process and log it with the FATAL
+// severity. On Windows, the message is read from a pipe handle. On other
+// platforms, it is read from a file descriptor.
+static void FailFromInternalError(int fd) {
+  Message error;
+  char buffer[256];
+  int num_read;
+
+  do {
+    while ((num_read = posix::Read(fd, buffer, 255)) > 0) {
+      buffer[num_read] = '\0';
+      error << buffer;
+    }
+  } while (num_read == -1 && errno == EINTR);
+
+  if (num_read == 0) {
+    GTEST_LOG_(FATAL) << error.GetString();
+  } else {
+    const int last_error = errno;
+    GTEST_LOG_(FATAL) << "Error while reading death test internal: "
+                      << GetLastErrnoDescription() << " [" << last_error << "]";
+  }
+}
+
+// Death test constructor.  Increments the running death test count
+// for the current test.
+DeathTest::DeathTest() {
+  TestInfo* const info = GetUnitTestImpl()->current_test_info();
+  if (info == NULL) {
+    DeathTestAbort("Cannot run a death test outside of a TEST or "
+                   "TEST_F construct");
+  }
+}
+
+// Creates and returns a death test by dispatching to the current
+// death test factory.
+bool DeathTest::Create(const char* statement, const RE* regex,
+                       const char* file, int line, DeathTest** test) {
+  return GetUnitTestImpl()->death_test_factory()->Create(
+      statement, regex, file, line, test);
+}
+
+const char* DeathTest::LastMessage() {
+  return last_death_test_message_.c_str();
+}
+
+void DeathTest::set_last_death_test_message(const std::string& message) {
+  last_death_test_message_ = message;
+}
+
+std::string DeathTest::last_death_test_message_;
+
+// Provides cross platform implementation for some death functionality.
+class DeathTestImpl : public DeathTest {
+ protected:
+  DeathTestImpl(const char* a_statement, const RE* a_regex)
+      : statement_(a_statement),
+        regex_(a_regex),
+        spawned_(false),
+        status_(-1),
+        outcome_(IN_PROGRESS),
+        read_fd_(-1),
+        write_fd_(-1) {}
+
+  // read_fd_ is expected to be closed and cleared by a derived class.
+  ~DeathTestImpl() { GTEST_DEATH_TEST_CHECK_(read_fd_ == -1); }
+
+  void Abort(AbortReason reason);
+  virtual bool Passed(bool status_ok);
+
+  const char* statement() const { return statement_; }
+  const RE* regex() const { return regex_; }
+  bool spawned() const { return spawned_; }
+  void set_spawned(bool is_spawned) { spawned_ = is_spawned; }
+  int status() const { return status_; }
+  void set_status(int a_status) { status_ = a_status; }
+  DeathTestOutcome outcome() const { return outcome_; }
+  void set_outcome(DeathTestOutcome an_outcome) { outcome_ = an_outcome; }
+  int read_fd() const { return read_fd_; }
+  void set_read_fd(int fd) { read_fd_ = fd; }
+  int write_fd() const { return write_fd_; }
+  void set_write_fd(int fd) { write_fd_ = fd; }
+
+  // Called in the parent process only. Reads the result code of the death
+  // test child process via a pipe, interprets it to set the outcome_
+  // member, and closes read_fd_.  Outputs diagnostics and terminates in
+  // case of unexpected codes.
+  void ReadAndInterpretStatusByte();
+
+ private:
+  // The textual content of the code this object is testing.  This class
+  // doesn't own this string and should not attempt to delete it.
+  const char* const statement_;
+  // The regular expression which test output must match.  DeathTestImpl
+  // doesn't own this object and should not attempt to delete it.
+  const RE* const regex_;
+  // True if the death test child process has been successfully spawned.
+  bool spawned_;
+  // The exit status of the child process.
+  int status_;
+  // How the death test concluded.
+  DeathTestOutcome outcome_;
+  // Descriptor to the read end of the pipe to the child process.  It is
+  // always -1 in the child process.  The child keeps its write end of the
+  // pipe in write_fd_.
+  int read_fd_;
+  // Descriptor to the child's write end of the pipe to the parent process.
+  // It is always -1 in the parent process.  The parent keeps its end of the
+  // pipe in read_fd_.
+  int write_fd_;
+};
+
+// Called in the parent process only. Reads the result code of the death
+// test child process via a pipe, interprets it to set the outcome_
+// member, and closes read_fd_.  Outputs diagnostics and terminates in
+// case of unexpected codes.
+void DeathTestImpl::ReadAndInterpretStatusByte() {
+  char flag;
+  int bytes_read;
+
+  // The read() here blocks until data is available (signifying the
+  // failure of the death test) or until the pipe is closed (signifying
+  // its success), so it's okay to call this in the parent before
+  // the child process has exited.
+  do {
+    bytes_read = posix::Read(read_fd(), &flag, 1);
+  } while (bytes_read == -1 && errno == EINTR);
+
+  if (bytes_read == 0) {
+    set_outcome(DIED);
+  } else if (bytes_read == 1) {
+    switch (flag) {
+      case kDeathTestReturned:
+        set_outcome(RETURNED);
+        break;
+      case kDeathTestThrew:
+        set_outcome(THREW);
+        break;
+      case kDeathTestLived:
+        set_outcome(LIVED);
+        break;
+      case kDeathTestInternalError:
+        FailFromInternalError(read_fd());  // Does not return.
+        break;
+      default:
+        GTEST_LOG_(FATAL) << "Death test child process reported "
+                          << "unexpected status byte ("
+                          << static_cast<unsigned int>(flag) << ")";
+    }
+  } else {
+    GTEST_LOG_(FATAL) << "Read from death test child process failed: "
+                      << GetLastErrnoDescription();
+  }
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Close(read_fd()));
+  set_read_fd(-1);
+}
+
+// Signals that the death test code which should have exited, didn't.
+// Should be called only in a death test child process.
+// Writes a status byte to the child's status file descriptor, then
+// calls _exit(1).
+void DeathTestImpl::Abort(AbortReason reason) {
+  // The parent process considers the death test to be a failure if
+  // it finds any data in our pipe.  So, here we write a single flag byte
+  // to the pipe, then exit.
+  const char status_ch =
+      reason == TEST_DID_NOT_DIE ? kDeathTestLived :
+      reason == TEST_THREW_EXCEPTION ? kDeathTestThrew : kDeathTestReturned;
+
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Write(write_fd(), &status_ch, 1));
+  // We are leaking the descriptor here because on some platforms (i.e.,
+  // when built as Windows DLL), destructors of global objects will still
+  // run after calling _exit(). On such systems, write_fd_ will be
+  // indirectly closed from the destructor of UnitTestImpl, causing double
+  // close if it is also closed here. On debug configurations, double close
+  // may assert. As there are no in-process buffers to flush here, we are
+  // relying on the OS to close the descriptor after the process terminates
+  // when the destructors are not run.
+  _exit(1);  // Exits w/o any normal exit hooks (we were supposed to crash)
+}
+
+// Returns an indented copy of stderr output for a death test.
+// This makes distinguishing death test output lines from regular log lines
+// much easier.
+static ::std::string FormatDeathTestOutput(const ::std::string& output) {
+  ::std::string ret;
+  for (size_t at = 0; ; ) {
+    const size_t line_end = output.find('\n', at);
+    ret += "[  DEATH   ] ";
+    if (line_end == ::std::string::npos) {
+      ret += output.substr(at);
+      break;
+    }
+    ret += output.substr(at, line_end + 1 - at);
+    at = line_end + 1;
+  }
+  return ret;
+}
+
+// Assesses the success or failure of a death test, using both private
+// members which have previously been set, and one argument:
+//
+// Private data members:
+//   outcome:  An enumeration describing how the death test
+//             concluded: DIED, LIVED, THREW, or RETURNED.  The death test
+//             fails in the latter three cases.
+//   status:   The exit status of the child process. On *nix, it is in the
+//             in the format specified by wait(2). On Windows, this is the
+//             value supplied to the ExitProcess() API or a numeric code
+//             of the exception that terminated the program.
+//   regex:    A regular expression object to be applied to
+//             the test's captured standard error output; the death test
+//             fails if it does not match.
+//
+// Argument:
+//   status_ok: true if exit_status is acceptable in the context of
+//              this particular death test, which fails if it is false
+//
+// Returns true iff all of the above conditions are met.  Otherwise, the
+// first failing condition, in the order given above, is the one that is
+// reported. Also sets the last death test message string.
+bool DeathTestImpl::Passed(bool status_ok) {
+  if (!spawned())
+    return false;
+
+  const std::string error_message = GetCapturedStderr();
+
+  bool success = false;
+  Message buffer;
+
+  buffer << "Death test: " << statement() << "\n";
+  switch (outcome()) {
+    case LIVED:
+      buffer << "    Result: failed to die.\n"
+             << " Error msg:\n" << FormatDeathTestOutput(error_message);
+      break;
+    case THREW:
+      buffer << "    Result: threw an exception.\n"
+             << " Error msg:\n" << FormatDeathTestOutput(error_message);
+      break;
+    case RETURNED:
+      buffer << "    Result: illegal return in test statement.\n"
+             << " Error msg:\n" << FormatDeathTestOutput(error_message);
+      break;
+    case DIED:
+      if (status_ok) {
+        const bool matched = RE::PartialMatch(error_message.c_str(), *regex());
+        if (matched) {
+          success = true;
+        } else {
+          buffer << "    Result: died but not with expected error.\n"
+                 << "  Expected: " << regex()->pattern() << "\n"
+                 << "Actual msg:\n" << FormatDeathTestOutput(error_message);
+        }
+      } else {
+        buffer << "    Result: died but not with expected exit code:\n"
+               << "            " << ExitSummary(status()) << "\n"
+               << "Actual msg:\n" << FormatDeathTestOutput(error_message);
+      }
+      break;
+    case IN_PROGRESS:
+    default:
+      GTEST_LOG_(FATAL)
+          << "DeathTest::Passed somehow called before conclusion of test";
+  }
+
+  DeathTest::set_last_death_test_message(buffer.GetString());
+  return success;
+}
+
+# if GTEST_OS_WINDOWS
+// WindowsDeathTest implements death tests on Windows. Due to the
+// specifics of starting new processes on Windows, death tests there are
+// always threadsafe, and Google Test considers the
+// --gtest_death_test_style=fast setting to be equivalent to
+// --gtest_death_test_style=threadsafe there.
+//
+// A few implementation notes:  Like the Linux version, the Windows
+// implementation uses pipes for child-to-parent communication. But due to
+// the specifics of pipes on Windows, some extra steps are required:
+//
+// 1. The parent creates a communication pipe and stores handles to both
+//    ends of it.
+// 2. The parent starts the child and provides it with the information
+//    necessary to acquire the handle to the write end of the pipe.
+// 3. The child acquires the write end of the pipe and signals the parent
+//    using a Windows event.
+// 4. Now the parent can release the write end of the pipe on its side. If
+//    this is done before step 3, the object's reference count goes down to
+//    0 and it is destroyed, preventing the child from acquiring it. The
+//    parent now has to release it, or read operations on the read end of
+//    the pipe will not return when the child terminates.
+// 5. The parent reads child's output through the pipe (outcome code and
+//    any possible error messages) from the pipe, and its stderr and then
+//    determines whether to fail the test.
+//
+// Note: to distinguish Win32 API calls from the local method and function
+// calls, the former are explicitly resolved in the global namespace.
+//
+class WindowsDeathTest : public DeathTestImpl {
+ public:
+  WindowsDeathTest(const char* a_statement,
+                   const RE* a_regex,
+                   const char* file,
+                   int line)
+      : DeathTestImpl(a_statement, a_regex), file_(file), line_(line) {}
+
+  // All of these virtual functions are inherited from DeathTest.
+  virtual int Wait();
+  virtual TestRole AssumeRole();
+
+ private:
+  // The name of the file in which the death test is located.
+  const char* const file_;
+  // The line number on which the death test is located.
+  const int line_;
+  // Handle to the write end of the pipe to the child process.
+  AutoHandle write_handle_;
+  // Child process handle.
+  AutoHandle child_handle_;
+  // Event the child process uses to signal the parent that it has
+  // acquired the handle to the write end of the pipe. After seeing this
+  // event the parent can release its own handles to make sure its
+  // ReadFile() calls return when the child terminates.
+  AutoHandle event_handle_;
+};
+
+// Waits for the child in a death test to exit, returning its exit
+// status, or 0 if no child process exists.  As a side effect, sets the
+// outcome data member.
+int WindowsDeathTest::Wait() {
+  if (!spawned())
+    return 0;
+
+  // Wait until the child either signals that it has acquired the write end
+  // of the pipe or it dies.
+  const HANDLE wait_handles[2] = { child_handle_.Get(), event_handle_.Get() };
+  switch (::WaitForMultipleObjects(2,
+                                   wait_handles,
+                                   FALSE,  // Waits for any of the handles.
+                                   INFINITE)) {
+    case WAIT_OBJECT_0:
+    case WAIT_OBJECT_0 + 1:
+      break;
+    default:
+      GTEST_DEATH_TEST_CHECK_(false);  // Should not get here.
+  }
+
+  // The child has acquired the write end of the pipe or exited.
+  // We release the handle on our side and continue.
+  write_handle_.Reset();
+  event_handle_.Reset();
+
+  ReadAndInterpretStatusByte();
+
+  // Waits for the child process to exit if it haven't already. This
+  // returns immediately if the child has already exited, regardless of
+  // whether previous calls to WaitForMultipleObjects synchronized on this
+  // handle or not.
+  GTEST_DEATH_TEST_CHECK_(
+      WAIT_OBJECT_0 == ::WaitForSingleObject(child_handle_.Get(),
+                                             INFINITE));
+  DWORD status_code;
+  GTEST_DEATH_TEST_CHECK_(
+      ::GetExitCodeProcess(child_handle_.Get(), &status_code) != FALSE);
+  child_handle_.Reset();
+  set_status(static_cast<int>(status_code));
+  return status();
+}
+
+// The AssumeRole process for a Windows death test.  It creates a child
+// process with the same executable as the current process to run the
+// death test.  The child process is given the --gtest_filter and
+// --gtest_internal_run_death_test flags such that it knows to run the
+// current death test only.
+DeathTest::TestRole WindowsDeathTest::AssumeRole() {
+  const UnitTestImpl* const impl = GetUnitTestImpl();
+  const InternalRunDeathTestFlag* const flag =
+      impl->internal_run_death_test_flag();
+  const TestInfo* const info = impl->current_test_info();
+  const int death_test_index = info->result()->death_test_count();
+
+  if (flag != NULL) {
+    // ParseInternalRunDeathTestFlag() has performed all the necessary
+    // processing.
+    set_write_fd(flag->write_fd());
+    return EXECUTE_TEST;
+  }
+
+  // WindowsDeathTest uses an anonymous pipe to communicate results of
+  // a death test.
+  SECURITY_ATTRIBUTES handles_are_inheritable = {
+    sizeof(SECURITY_ATTRIBUTES), NULL, TRUE };
+  HANDLE read_handle, write_handle;
+  GTEST_DEATH_TEST_CHECK_(
+      ::CreatePipe(&read_handle, &write_handle, &handles_are_inheritable,
+                   0)  // Default buffer size.
+      != FALSE);
+  set_read_fd(::_open_osfhandle(reinterpret_cast<intptr_t>(read_handle),
+                                O_RDONLY));
+  write_handle_.Reset(write_handle);
+  event_handle_.Reset(::CreateEvent(
+      &handles_are_inheritable,
+      TRUE,    // The event will automatically reset to non-signaled state.
+      FALSE,   // The initial state is non-signalled.
+      NULL));  // The even is unnamed.
+  GTEST_DEATH_TEST_CHECK_(event_handle_.Get() != NULL);
+  const std::string filter_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kFilterFlag + "=" +
+      info->test_case_name() + "." + info->name();
+  const std::string internal_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag +
+      "=" + file_ + "|" + StreamableToString(line_) + "|" +
+      StreamableToString(death_test_index) + "|" +
+      StreamableToString(static_cast<unsigned int>(::GetCurrentProcessId())) +
+      // size_t has the same width as pointers on both 32-bit and 64-bit
+      // Windows platforms.
+      // See http://msdn.microsoft.com/en-us/library/tcxf1dw6.aspx.
+      "|" + StreamableToString(reinterpret_cast<size_t>(write_handle)) +
+      "|" + StreamableToString(reinterpret_cast<size_t>(event_handle_.Get()));
+
+  char executable_path[_MAX_PATH + 1];  // NOLINT
+  GTEST_DEATH_TEST_CHECK_(
+      _MAX_PATH + 1 != ::GetModuleFileNameA(NULL,
+                                            executable_path,
+                                            _MAX_PATH));
+
+  std::string command_line =
+      std::string(::GetCommandLineA()) + " " + filter_flag + " \"" +
+      internal_flag + "\"";
+
+  DeathTest::set_last_death_test_message("");
+
+  CaptureStderr();
+  // Flush the log buffers since the log streams are shared with the child.
+  FlushInfoLog();
+
+  // The child process will share the standard handles with the parent.
+  STARTUPINFOA startup_info;
+  memset(&startup_info, 0, sizeof(STARTUPINFO));
+  startup_info.dwFlags = STARTF_USESTDHANDLES;
+  startup_info.hStdInput = ::GetStdHandle(STD_INPUT_HANDLE);
+  startup_info.hStdOutput = ::GetStdHandle(STD_OUTPUT_HANDLE);
+  startup_info.hStdError = ::GetStdHandle(STD_ERROR_HANDLE);
+
+  PROCESS_INFORMATION process_info;
+  GTEST_DEATH_TEST_CHECK_(::CreateProcessA(
+      executable_path,
+      const_cast<char*>(command_line.c_str()),
+      NULL,   // Returned process handle is not inheritable.
+      NULL,   // Returned thread handle is not inheritable.
+      TRUE,   // Child inherits all inheritable handles (for write_handle_).
+      0x0,    // Default creation flags.
+      NULL,   // Inherit the parent's environment.
+      UnitTest::GetInstance()->original_working_dir(),
+      &startup_info,
+      &process_info) != FALSE);
+  child_handle_.Reset(process_info.hProcess);
+  ::CloseHandle(process_info.hThread);
+  set_spawned(true);
+  return OVERSEE_TEST;
+}
+# else  // We are not on Windows.
+
+// ForkingDeathTest provides implementations for most of the abstract
+// methods of the DeathTest interface.  Only the AssumeRole method is
+// left undefined.
+class ForkingDeathTest : public DeathTestImpl {
+ public:
+  ForkingDeathTest(const char* statement, const RE* regex);
+
+  // All of these virtual functions are inherited from DeathTest.
+  virtual int Wait();
+
+ protected:
+  void set_child_pid(pid_t child_pid) { child_pid_ = child_pid; }
+
+ private:
+  // PID of child process during death test; 0 in the child process itself.
+  pid_t child_pid_;
+};
+
+// Constructs a ForkingDeathTest.
+ForkingDeathTest::ForkingDeathTest(const char* a_statement, const RE* a_regex)
+    : DeathTestImpl(a_statement, a_regex),
+      child_pid_(-1) {}
+
+// Waits for the child in a death test to exit, returning its exit
+// status, or 0 if no child process exists.  As a side effect, sets the
+// outcome data member.
+int ForkingDeathTest::Wait() {
+  if (!spawned())
+    return 0;
+
+  ReadAndInterpretStatusByte();
+
+  int status_value;
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(waitpid(child_pid_, &status_value, 0));
+  set_status(status_value);
+  return status_value;
+}
+
+// A concrete death test class that forks, then immediately runs the test
+// in the child process.
+class NoExecDeathTest : public ForkingDeathTest {
+ public:
+  NoExecDeathTest(const char* a_statement, const RE* a_regex) :
+      ForkingDeathTest(a_statement, a_regex) { }
+  virtual TestRole AssumeRole();
+};
+
+// The AssumeRole process for a fork-and-run death test.  It implements a
+// straightforward fork, with a simple pipe to transmit the status byte.
+DeathTest::TestRole NoExecDeathTest::AssumeRole() {
+  const size_t thread_count = GetThreadCount();
+  if (thread_count != 1) {
+    GTEST_LOG_(WARNING) << DeathTestThreadWarning(thread_count);
+  }
+
+  int pipe_fd[2];
+  GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1);
+
+  DeathTest::set_last_death_test_message("");
+  CaptureStderr();
+  // When we fork the process below, the log file buffers are copied, but the
+  // file descriptors are shared.  We flush all log files here so that closing
+  // the file descriptors in the child process doesn't throw off the
+  // synchronization between descriptors and buffers in the parent process.
+  // This is as close to the fork as possible to avoid a race condition in case
+  // there are multiple threads running before the death test, and another
+  // thread writes to the log file.
+  FlushInfoLog();
+
+  const pid_t child_pid = fork();
+  GTEST_DEATH_TEST_CHECK_(child_pid != -1);
+  set_child_pid(child_pid);
+  if (child_pid == 0) {
+    GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[0]));
+    set_write_fd(pipe_fd[1]);
+    // Redirects all logging to stderr in the child process to prevent
+    // concurrent writes to the log files.  We capture stderr in the parent
+    // process and append the child process' output to a log.
+    LogToStderr();
+    // Event forwarding to the listeners of event listener API mush be shut
+    // down in death test subprocesses.
+    GetUnitTestImpl()->listeners()->SuppressEventForwarding();
+    g_in_fast_death_test_child = true;
+    return EXECUTE_TEST;
+  } else {
+    GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1]));
+    set_read_fd(pipe_fd[0]);
+    set_spawned(true);
+    return OVERSEE_TEST;
+  }
+}
+
+// A concrete death test class that forks and re-executes the main
+// program from the beginning, with command-line flags set that cause
+// only this specific death test to be run.
+class ExecDeathTest : public ForkingDeathTest {
+ public:
+  ExecDeathTest(const char* a_statement, const RE* a_regex,
+                const char* file, int line) :
+      ForkingDeathTest(a_statement, a_regex), file_(file), line_(line) { }
+  virtual TestRole AssumeRole();
+ private:
+  static ::std::vector<std::string> GetArgvsForDeathTestChildProcess() {
+    ::std::vector<std::string> args = GetInjectableArgvs();
+#  if defined(GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_)
+    ::std::vector<std::string> extra_args =
+        GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_();
+    args.insert(args.end(), extra_args.begin(), extra_args.end());
+#  endif  // defined(GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_)
+    return args;
+  }
+  // The name of the file in which the death test is located.
+  const char* const file_;
+  // The line number on which the death test is located.
+  const int line_;
+};
+
+// Utility class for accumulating command-line arguments.
+class Arguments {
+ public:
+  Arguments() {
+    args_.push_back(NULL);
+  }
+
+  ~Arguments() {
+    for (std::vector<char*>::iterator i = args_.begin(); i != args_.end();
+         ++i) {
+      free(*i);
+    }
+  }
+  void AddArgument(const char* argument) {
+    args_.insert(args_.end() - 1, posix::StrDup(argument));
+  }
+
+  template <typename Str>
+  void AddArguments(const ::std::vector<Str>& arguments) {
+    for (typename ::std::vector<Str>::const_iterator i = arguments.begin();
+         i != arguments.end();
+         ++i) {
+      args_.insert(args_.end() - 1, posix::StrDup(i->c_str()));
+    }
+  }
+  char* const* Argv() {
+    return &args_[0];
+  }
+
+ private:
+  std::vector<char*> args_;
+};
+
+// A struct that encompasses the arguments to the child process of a
+// threadsafe-style death test process.
+struct ExecDeathTestArgs {
+  char* const* argv;  // Command-line arguments for the child's call to exec
+  int close_fd;       // File descriptor to close; the read end of a pipe
+};
+
+#  if GTEST_OS_MAC
+inline char** GetEnviron() {
+  // When Google Test is built as a framework on MacOS X, the environ variable
+  // is unavailable. Apple's documentation (man environ) recommends using
+  // _NSGetEnviron() instead.
+  return *_NSGetEnviron();
+}
+#  else
+// Some POSIX platforms expect you to declare environ. extern "C" makes
+// it reside in the global namespace.
+extern "C" char** environ;
+inline char** GetEnviron() { return environ; }
+#  endif  // GTEST_OS_MAC
+
+#  if !GTEST_OS_QNX
+// The main function for a threadsafe-style death test child process.
+// This function is called in a clone()-ed process and thus must avoid
+// any potentially unsafe operations like malloc or libc functions.
+static int ExecDeathTestChildMain(void* child_arg) {
+  ExecDeathTestArgs* const args = static_cast<ExecDeathTestArgs*>(child_arg);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(close(args->close_fd));
+
+  // We need to execute the test program in the same environment where
+  // it was originally invoked.  Therefore we change to the original
+  // working directory first.
+  const char* const original_dir =
+      UnitTest::GetInstance()->original_working_dir();
+  // We can safely call chdir() as it's a direct system call.
+  if (chdir(original_dir) != 0) {
+    DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " +
+                   GetLastErrnoDescription());
+    return EXIT_FAILURE;
+  }
+
+  // We can safely call execve() as it's a direct system call.  We
+  // cannot use execvp() as it's a libc function and thus potentially
+  // unsafe.  Since execve() doesn't search the PATH, the user must
+  // invoke the test program via a valid path that contains at least
+  // one path separator.
+  execve(args->argv[0], args->argv, GetEnviron());
+  DeathTestAbort(std::string("execve(") + args->argv[0] + ", ...) in " +
+                 original_dir + " failed: " +
+                 GetLastErrnoDescription());
+  return EXIT_FAILURE;
+}
+#  endif  // !GTEST_OS_QNX
+
+// Two utility routines that together determine the direction the stack
+// grows.
+// This could be accomplished more elegantly by a single recursive
+// function, but we want to guard against the unlikely possibility of
+// a smart compiler optimizing the recursion away.
+//
+// GTEST_NO_INLINE_ is required to prevent GCC 4.6 from inlining
+// StackLowerThanAddress into StackGrowsDown, which then doesn't give
+// correct answer.
+void StackLowerThanAddress(const void* ptr, bool* result) GTEST_NO_INLINE_;
+void StackLowerThanAddress(const void* ptr, bool* result) {
+  int dummy;
+  *result = (&dummy < ptr);
+}
+
+// Make sure AddressSanitizer does not tamper with the stack here.
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+bool StackGrowsDown() {
+  int dummy;
+  bool result;
+  StackLowerThanAddress(&dummy, &result);
+  return result;
+}
+
+// Spawns a child process with the same executable as the current process in
+// a thread-safe manner and instructs it to run the death test.  The
+// implementation uses fork(2) + exec.  On systems where clone(2) is
+// available, it is used instead, being slightly more thread-safe.  On QNX,
+// fork supports only single-threaded environments, so this function uses
+// spawn(2) there instead.  The function dies with an error message if
+// anything goes wrong.
+static pid_t ExecDeathTestSpawnChild(char* const* argv, int close_fd) {
+  ExecDeathTestArgs args = { argv, close_fd };
+  pid_t child_pid = -1;
+
+#  if GTEST_OS_QNX
+  // Obtains the current directory and sets it to be closed in the child
+  // process.
+  const int cwd_fd = open(".", O_RDONLY);
+  GTEST_DEATH_TEST_CHECK_(cwd_fd != -1);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(cwd_fd, F_SETFD, FD_CLOEXEC));
+  // We need to execute the test program in the same environment where
+  // it was originally invoked.  Therefore we change to the original
+  // working directory first.
+  const char* const original_dir =
+      UnitTest::GetInstance()->original_working_dir();
+  // We can safely call chdir() as it's a direct system call.
+  if (chdir(original_dir) != 0) {
+    DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " +
+                   GetLastErrnoDescription());
+    return EXIT_FAILURE;
+  }
+
+  int fd_flags;
+  // Set close_fd to be closed after spawn.
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(fd_flags = fcntl(close_fd, F_GETFD));
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(close_fd, F_SETFD,
+                                        fd_flags | FD_CLOEXEC));
+  struct inheritance inherit = {0};
+  // spawn is a system call.
+  child_pid = spawn(args.argv[0], 0, NULL, &inherit, args.argv, GetEnviron());
+  // Restores the current working directory.
+  GTEST_DEATH_TEST_CHECK_(fchdir(cwd_fd) != -1);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(close(cwd_fd));
+
+#  else   // GTEST_OS_QNX
+#   if GTEST_OS_LINUX
+  // When a SIGPROF signal is received while fork() or clone() are executing,
+  // the process may hang. To avoid this, we ignore SIGPROF here and re-enable
+  // it after the call to fork()/clone() is complete.
+  struct sigaction saved_sigprof_action;
+  struct sigaction ignore_sigprof_action;
+  memset(&ignore_sigprof_action, 0, sizeof(ignore_sigprof_action));
+  sigemptyset(&ignore_sigprof_action.sa_mask);
+  ignore_sigprof_action.sa_handler = SIG_IGN;
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(sigaction(
+      SIGPROF, &ignore_sigprof_action, &saved_sigprof_action));
+#   endif  // GTEST_OS_LINUX
+
+#   if GTEST_HAS_CLONE
+  const bool use_fork = GTEST_FLAG(death_test_use_fork);
+
+  if (!use_fork) {
+    static const bool stack_grows_down = StackGrowsDown();
+    const size_t stack_size = getpagesize();
+    // MMAP_ANONYMOUS is not defined on Mac, so we use MAP_ANON instead.
+    void* const stack = mmap(NULL, stack_size, PROT_READ | PROT_WRITE,
+                             MAP_ANON | MAP_PRIVATE, -1, 0);
+    GTEST_DEATH_TEST_CHECK_(stack != MAP_FAILED);
+
+    // Maximum stack alignment in bytes:  For a downward-growing stack, this
+    // amount is subtracted from size of the stack space to get an address
+    // that is within the stack space and is aligned on all systems we care
+    // about.  As far as I know there is no ABI with stack alignment greater
+    // than 64.  We assume stack and stack_size already have alignment of
+    // kMaxStackAlignment.
+    const size_t kMaxStackAlignment = 64;
+    void* const stack_top =
+        static_cast<char*>(stack) +
+            (stack_grows_down ? stack_size - kMaxStackAlignment : 0);
+    GTEST_DEATH_TEST_CHECK_(stack_size > kMaxStackAlignment &&
+        reinterpret_cast<intptr_t>(stack_top) % kMaxStackAlignment == 0);
+
+    child_pid = clone(&ExecDeathTestChildMain, stack_top, SIGCHLD, &args);
+
+    GTEST_DEATH_TEST_CHECK_(munmap(stack, stack_size) != -1);
+  }
+#   else
+  const bool use_fork = true;
+#   endif  // GTEST_HAS_CLONE
+
+  if (use_fork && (child_pid = fork()) == 0) {
+      ExecDeathTestChildMain(&args);
+      _exit(0);
+  }
+#  endif  // GTEST_OS_QNX
+#  if GTEST_OS_LINUX
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(
+      sigaction(SIGPROF, &saved_sigprof_action, NULL));
+#  endif  // GTEST_OS_LINUX
+
+  GTEST_DEATH_TEST_CHECK_(child_pid != -1);
+  return child_pid;
+}
+
+// The AssumeRole process for a fork-and-exec death test.  It re-executes the
+// main program from the beginning, setting the --gtest_filter
+// and --gtest_internal_run_death_test flags to cause only the current
+// death test to be re-run.
+DeathTest::TestRole ExecDeathTest::AssumeRole() {
+  const UnitTestImpl* const impl = GetUnitTestImpl();
+  const InternalRunDeathTestFlag* const flag =
+      impl->internal_run_death_test_flag();
+  const TestInfo* const info = impl->current_test_info();
+  const int death_test_index = info->result()->death_test_count();
+
+  if (flag != NULL) {
+    set_write_fd(flag->write_fd());
+    return EXECUTE_TEST;
+  }
+
+  int pipe_fd[2];
+  GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1);
+  // Clear the close-on-exec flag on the write end of the pipe, lest
+  // it be closed when the child process does an exec:
+  GTEST_DEATH_TEST_CHECK_(fcntl(pipe_fd[1], F_SETFD, 0) != -1);
+
+  const std::string filter_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kFilterFlag + "="
+      + info->test_case_name() + "." + info->name();
+  const std::string internal_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag + "="
+      + file_ + "|" + StreamableToString(line_) + "|"
+      + StreamableToString(death_test_index) + "|"
+      + StreamableToString(pipe_fd[1]);
+  Arguments args;
+  args.AddArguments(GetArgvsForDeathTestChildProcess());
+  args.AddArgument(filter_flag.c_str());
+  args.AddArgument(internal_flag.c_str());
+
+  DeathTest::set_last_death_test_message("");
+
+  CaptureStderr();
+  // See the comment in NoExecDeathTest::AssumeRole for why the next line
+  // is necessary.
+  FlushInfoLog();
+
+  const pid_t child_pid = ExecDeathTestSpawnChild(args.Argv(), pipe_fd[0]);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1]));
+  set_child_pid(child_pid);
+  set_read_fd(pipe_fd[0]);
+  set_spawned(true);
+  return OVERSEE_TEST;
+}
+
+# endif  // !GTEST_OS_WINDOWS
+
+// Creates a concrete DeathTest-derived class that depends on the
+// --gtest_death_test_style flag, and sets the pointer pointed to
+// by the "test" argument to its address.  If the test should be
+// skipped, sets that pointer to NULL.  Returns true, unless the
+// flag is set to an invalid value.
+bool DefaultDeathTestFactory::Create(const char* statement, const RE* regex,
+                                     const char* file, int line,
+                                     DeathTest** test) {
+  UnitTestImpl* const impl = GetUnitTestImpl();
+  const InternalRunDeathTestFlag* const flag =
+      impl->internal_run_death_test_flag();
+  const int death_test_index = impl->current_test_info()
+      ->increment_death_test_count();
+
+  if (flag != NULL) {
+    if (death_test_index > flag->index()) {
+      DeathTest::set_last_death_test_message(
+          "Death test count (" + StreamableToString(death_test_index)
+          + ") somehow exceeded expected maximum ("
+          + StreamableToString(flag->index()) + ")");
+      return false;
+    }
+
+    if (!(flag->file() == file && flag->line() == line &&
+          flag->index() == death_test_index)) {
+      *test = NULL;
+      return true;
+    }
+  }
+
+# if GTEST_OS_WINDOWS
+
+  if (GTEST_FLAG(death_test_style) == "threadsafe" ||
+      GTEST_FLAG(death_test_style) == "fast") {
+    *test = new WindowsDeathTest(statement, regex, file, line);
+  }
+
+# else
+
+  if (GTEST_FLAG(death_test_style) == "threadsafe") {
+    *test = new ExecDeathTest(statement, regex, file, line);
+  } else if (GTEST_FLAG(death_test_style) == "fast") {
+    *test = new NoExecDeathTest(statement, regex);
+  }
+
+# endif  // GTEST_OS_WINDOWS
+
+  else {  // NOLINT - this is more readable than unbalanced brackets inside #if.
+    DeathTest::set_last_death_test_message(
+        "Unknown death test style \"" + GTEST_FLAG(death_test_style)
+        + "\" encountered");
+    return false;
+  }
+
+  return true;
+}
+
+# if GTEST_OS_WINDOWS
+// Recreates the pipe and event handles from the provided parameters,
+// signals the event, and returns a file descriptor wrapped around the pipe
+// handle. This function is called in the child process only.
+int GetStatusFileDescriptor(unsigned int parent_process_id,
+                            size_t write_handle_as_size_t,
+                            size_t event_handle_as_size_t) {
+  AutoHandle parent_process_handle(::OpenProcess(PROCESS_DUP_HANDLE,
+                                                   FALSE,  // Non-inheritable.
+                                                   parent_process_id));
+  if (parent_process_handle.Get() == INVALID_HANDLE_VALUE) {
+    DeathTestAbort("Unable to open parent process " +
+                   StreamableToString(parent_process_id));
+  }
+
+  // TODO(vladl@google.com): Replace the following check with a
+  // compile-time assertion when available.
+  GTEST_CHECK_(sizeof(HANDLE) <= sizeof(size_t));
+
+  const HANDLE write_handle =
+      reinterpret_cast<HANDLE>(write_handle_as_size_t);
+  HANDLE dup_write_handle;
+
+  // The newly initialized handle is accessible only in in the parent
+  // process. To obtain one accessible within the child, we need to use
+  // DuplicateHandle.
+  if (!::DuplicateHandle(parent_process_handle.Get(), write_handle,
+                         ::GetCurrentProcess(), &dup_write_handle,
+                         0x0,    // Requested privileges ignored since
+                                 // DUPLICATE_SAME_ACCESS is used.
+                         FALSE,  // Request non-inheritable handler.
+                         DUPLICATE_SAME_ACCESS)) {
+    DeathTestAbort("Unable to duplicate the pipe handle " +
+                   StreamableToString(write_handle_as_size_t) +
+                   " from the parent process " +
+                   StreamableToString(parent_process_id));
+  }
+
+  const HANDLE event_handle = reinterpret_cast<HANDLE>(event_handle_as_size_t);
+  HANDLE dup_event_handle;
+
+  if (!::DuplicateHandle(parent_process_handle.Get(), event_handle,
+                         ::GetCurrentProcess(), &dup_event_handle,
+                         0x0,
+                         FALSE,
+                         DUPLICATE_SAME_ACCESS)) {
+    DeathTestAbort("Unable to duplicate the event handle " +
+                   StreamableToString(event_handle_as_size_t) +
+                   " from the parent process " +
+                   StreamableToString(parent_process_id));
+  }
+
+  const int write_fd =
+      ::_open_osfhandle(reinterpret_cast<intptr_t>(dup_write_handle), O_APPEND);
+  if (write_fd == -1) {
+    DeathTestAbort("Unable to convert pipe handle " +
+                   StreamableToString(write_handle_as_size_t) +
+                   " to a file descriptor");
+  }
+
+  // Signals the parent that the write end of the pipe has been acquired
+  // so the parent can release its own write end.
+  ::SetEvent(dup_event_handle);
+
+  return write_fd;
+}
+# endif  // GTEST_OS_WINDOWS
+
+// Returns a newly created InternalRunDeathTestFlag object with fields
+// initialized from the GTEST_FLAG(internal_run_death_test) flag if
+// the flag is specified; otherwise returns NULL.
+InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag() {
+  if (GTEST_FLAG(internal_run_death_test) == "") return NULL;
+
+  // GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we
+  // can use it here.
+  int line = -1;
+  int index = -1;
+  ::std::vector< ::std::string> fields;
+  SplitString(GTEST_FLAG(internal_run_death_test).c_str(), '|', &fields);
+  int write_fd = -1;
+
+# if GTEST_OS_WINDOWS
+
+  unsigned int parent_process_id = 0;
+  size_t write_handle_as_size_t = 0;
+  size_t event_handle_as_size_t = 0;
+
+  if (fields.size() != 6
+      || !ParseNaturalNumber(fields[1], &line)
+      || !ParseNaturalNumber(fields[2], &index)
+      || !ParseNaturalNumber(fields[3], &parent_process_id)
+      || !ParseNaturalNumber(fields[4], &write_handle_as_size_t)
+      || !ParseNaturalNumber(fields[5], &event_handle_as_size_t)) {
+    DeathTestAbort("Bad --gtest_internal_run_death_test flag: " +
+                   GTEST_FLAG(internal_run_death_test));
+  }
+  write_fd = GetStatusFileDescriptor(parent_process_id,
+                                     write_handle_as_size_t,
+                                     event_handle_as_size_t);
+# else
+
+  if (fields.size() != 4
+      || !ParseNaturalNumber(fields[1], &line)
+      || !ParseNaturalNumber(fields[2], &index)
+      || !ParseNaturalNumber(fields[3], &write_fd)) {
+    DeathTestAbort("Bad --gtest_internal_run_death_test flag: "
+        + GTEST_FLAG(internal_run_death_test));
+  }
+
+# endif  // GTEST_OS_WINDOWS
+
+  return new InternalRunDeathTestFlag(fields[0], line, index, write_fd);
+}
+
+}  // namespace internal
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+}  // namespace testing
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: keith.ray@gmail.com (Keith Ray)
+
+
+#include <stdlib.h>
+
+#if GTEST_OS_WINDOWS_MOBILE
+# include <windows.h>
+#elif GTEST_OS_WINDOWS
+# include <direct.h>
+# include <io.h>
+#elif GTEST_OS_SYMBIAN
+// Symbian OpenC has PATH_MAX in sys/syslimits.h
+# include <sys/syslimits.h>
+#else
+# include <limits.h>
+# include <climits>  // Some Linux distributions define PATH_MAX here.
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+#if GTEST_OS_WINDOWS
+# define GTEST_PATH_MAX_ _MAX_PATH
+#elif defined(PATH_MAX)
+# define GTEST_PATH_MAX_ PATH_MAX
+#elif defined(_XOPEN_PATH_MAX)
+# define GTEST_PATH_MAX_ _XOPEN_PATH_MAX
+#else
+# define GTEST_PATH_MAX_ _POSIX_PATH_MAX
+#endif  // GTEST_OS_WINDOWS
+
+
+namespace testing {
+namespace internal {
+
+#if GTEST_OS_WINDOWS
+// On Windows, '\\' is the standard path separator, but many tools and the
+// Windows API also accept '/' as an alternate path separator. Unless otherwise
+// noted, a file path can contain either kind of path separators, or a mixture
+// of them.
+const char kPathSeparator = '\\';
+const char kAlternatePathSeparator = '/';
+const char kAlternatePathSeparatorString[] = "/";
+# if GTEST_OS_WINDOWS_MOBILE
+// Windows CE doesn't have a current directory. You should not use
+// the current directory in tests on Windows CE, but this at least
+// provides a reasonable fallback.
+const char kCurrentDirectoryString[] = "\\";
+// Windows CE doesn't define INVALID_FILE_ATTRIBUTES
+const DWORD kInvalidFileAttributes = 0xffffffff;
+# else
+const char kCurrentDirectoryString[] = ".\\";
+# endif  // GTEST_OS_WINDOWS_MOBILE
+#else
+const char kPathSeparator = '/';
+const char kCurrentDirectoryString[] = "./";
+#endif  // GTEST_OS_WINDOWS
+
+// Returns whether the given character is a valid path separator.
+static bool IsPathSeparator(char c) {
+#if GTEST_HAS_ALT_PATH_SEP_
+  return (c == kPathSeparator) || (c == kAlternatePathSeparator);
+#else
+  return c == kPathSeparator;
+#endif
+}
+
+// Returns the current working directory, or "" if unsuccessful.
+FilePath FilePath::GetCurrentDir() {
+#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT
+  // Windows CE doesn't have a current directory, so we just return
+  // something reasonable.
+  return FilePath(kCurrentDirectoryString);
+#elif GTEST_OS_WINDOWS
+  char cwd[GTEST_PATH_MAX_ + 1] = { '\0' };
+  return FilePath(_getcwd(cwd, sizeof(cwd)) == NULL ? "" : cwd);
+#else
+  char cwd[GTEST_PATH_MAX_ + 1] = { '\0' };
+  char* result = getcwd(cwd, sizeof(cwd));
+# if GTEST_OS_NACL
+  // getcwd will likely fail in NaCl due to the sandbox, so return something
+  // reasonable. The user may have provided a shim implementation for getcwd,
+  // however, so fallback only when failure is detected.
+  return FilePath(result == NULL ? kCurrentDirectoryString : cwd);
+# endif  // GTEST_OS_NACL
+  return FilePath(result == NULL ? "" : cwd);
+#endif  // GTEST_OS_WINDOWS_MOBILE
+}
+
+// Returns a copy of the FilePath with the case-insensitive extension removed.
+// Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns
+// FilePath("dir/file"). If a case-insensitive extension is not
+// found, returns a copy of the original FilePath.
+FilePath FilePath::RemoveExtension(const char* extension) const {
+  const std::string dot_extension = std::string(".") + extension;
+  if (String::EndsWithCaseInsensitive(pathname_, dot_extension)) {
+    return FilePath(pathname_.substr(
+        0, pathname_.length() - dot_extension.length()));
+  }
+  return *this;
+}
+
+// Returns a pointer to the last occurrence of a valid path separator in
+// the FilePath. On Windows, for example, both '/' and '\' are valid path
+// separators. Returns NULL if no path separator was found.
+const char* FilePath::FindLastPathSeparator() const {
+  const char* const last_sep = strrchr(c_str(), kPathSeparator);
+#if GTEST_HAS_ALT_PATH_SEP_
+  const char* const last_alt_sep = strrchr(c_str(), kAlternatePathSeparator);
+  // Comparing two pointers of which only one is NULL is undefined.
+  if (last_alt_sep != NULL &&
+      (last_sep == NULL || last_alt_sep > last_sep)) {
+    return last_alt_sep;
+  }
+#endif
+  return last_sep;
+}
+
+// Returns a copy of the FilePath with the directory part removed.
+// Example: FilePath("path/to/file").RemoveDirectoryName() returns
+// FilePath("file"). If there is no directory part ("just_a_file"), it returns
+// the FilePath unmodified. If there is no file part ("just_a_dir/") it
+// returns an empty FilePath ("").
+// On Windows platform, '\' is the path separator, otherwise it is '/'.
+FilePath FilePath::RemoveDirectoryName() const {
+  const char* const last_sep = FindLastPathSeparator();
+  return last_sep ? FilePath(last_sep + 1) : *this;
+}
+
+// RemoveFileName returns the directory path with the filename removed.
+// Example: FilePath("path/to/file").RemoveFileName() returns "path/to/".
+// If the FilePath is "a_file" or "/a_file", RemoveFileName returns
+// FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does
+// not have a file, like "just/a/dir/", it returns the FilePath unmodified.
+// On Windows platform, '\' is the path separator, otherwise it is '/'.
+FilePath FilePath::RemoveFileName() const {
+  const char* const last_sep = FindLastPathSeparator();
+  std::string dir;
+  if (last_sep) {
+    dir = std::string(c_str(), last_sep + 1 - c_str());
+  } else {
+    dir = kCurrentDirectoryString;
+  }
+  return FilePath(dir);
+}
+
+// Helper functions for naming files in a directory for xml output.
+
+// Given directory = "dir", base_name = "test", number = 0,
+// extension = "xml", returns "dir/test.xml". If number is greater
+// than zero (e.g., 12), returns "dir/test_12.xml".
+// On Windows platform, uses \ as the separator rather than /.
+FilePath FilePath::MakeFileName(const FilePath& directory,
+                                const FilePath& base_name,
+                                int number,
+                                const char* extension) {
+  std::string file;
+  if (number == 0) {
+    file = base_name.string() + "." + extension;
+  } else {
+    file = base_name.string() + "_" + StreamableToString(number)
+        + "." + extension;
+  }
+  return ConcatPaths(directory, FilePath(file));
+}
+
+// Given directory = "dir", relative_path = "test.xml", returns "dir/test.xml".
+// On Windows, uses \ as the separator rather than /.
+FilePath FilePath::ConcatPaths(const FilePath& directory,
+                               const FilePath& relative_path) {
+  if (directory.IsEmpty())
+    return relative_path;
+  const FilePath dir(directory.RemoveTrailingPathSeparator());
+  return FilePath(dir.string() + kPathSeparator + relative_path.string());
+}
+
+// Returns true if pathname describes something findable in the file-system,
+// either a file, directory, or whatever.
+bool FilePath::FileOrDirectoryExists() const {
+#if GTEST_OS_WINDOWS_MOBILE
+  LPCWSTR unicode = String::AnsiToUtf16(pathname_.c_str());
+  const DWORD attributes = GetFileAttributes(unicode);
+  delete [] unicode;
+  return attributes != kInvalidFileAttributes;
+#else
+  posix::StatStruct file_stat;
+  return posix::Stat(pathname_.c_str(), &file_stat) == 0;
+#endif  // GTEST_OS_WINDOWS_MOBILE
+}
+
+// Returns true if pathname describes a directory in the file-system
+// that exists.
+bool FilePath::DirectoryExists() const {
+  bool result = false;
+#if GTEST_OS_WINDOWS
+  // Don't strip off trailing separator if path is a root directory on
+  // Windows (like "C:\\").
+  const FilePath& path(IsRootDirectory() ? *this :
+                                           RemoveTrailingPathSeparator());
+#else
+  const FilePath& path(*this);
+#endif
+
+#if GTEST_OS_WINDOWS_MOBILE
+  LPCWSTR unicode = String::AnsiToUtf16(path.c_str());
+  const DWORD attributes = GetFileAttributes(unicode);
+  delete [] unicode;
+  if ((attributes != kInvalidFileAttributes) &&
+      (attributes & FILE_ATTRIBUTE_DIRECTORY)) {
+    result = true;
+  }
+#else
+  posix::StatStruct file_stat;
+  result = posix::Stat(path.c_str(), &file_stat) == 0 &&
+      posix::IsDir(file_stat);
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+  return result;
+}
+
+// Returns true if pathname describes a root directory. (Windows has one
+// root directory per disk drive.)
+bool FilePath::IsRootDirectory() const {
+#if GTEST_OS_WINDOWS
+  // TODO(wan@google.com): on Windows a network share like
+  // \\server\share can be a root directory, although it cannot be the
+  // current directory.  Handle this properly.
+  return pathname_.length() == 3 && IsAbsolutePath();
+#else
+  return pathname_.length() == 1 && IsPathSeparator(pathname_.c_str()[0]);
+#endif
+}
+
+// Returns true if pathname describes an absolute path.
+bool FilePath::IsAbsolutePath() const {
+  const char* const name = pathname_.c_str();
+#if GTEST_OS_WINDOWS
+  return pathname_.length() >= 3 &&
+     ((name[0] >= 'a' && name[0] <= 'z') ||
+      (name[0] >= 'A' && name[0] <= 'Z')) &&
+     name[1] == ':' &&
+     IsPathSeparator(name[2]);
+#else
+  return IsPathSeparator(name[0]);
+#endif
+}
+
+// Returns a pathname for a file that does not currently exist. The pathname
+// will be directory/base_name.extension or
+// directory/base_name_<number>.extension if directory/base_name.extension
+// already exists. The number will be incremented until a pathname is found
+// that does not already exist.
+// Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'.
+// There could be a race condition if two or more processes are calling this
+// function at the same time -- they could both pick the same filename.
+FilePath FilePath::GenerateUniqueFileName(const FilePath& directory,
+                                          const FilePath& base_name,
+                                          const char* extension) {
+  FilePath full_pathname;
+  int number = 0;
+  do {
+    full_pathname.Set(MakeFileName(directory, base_name, number++, extension));
+  } while (full_pathname.FileOrDirectoryExists());
+  return full_pathname;
+}
+
+// Returns true if FilePath ends with a path separator, which indicates that
+// it is intended to represent a directory. Returns false otherwise.
+// This does NOT check that a directory (or file) actually exists.
+bool FilePath::IsDirectory() const {
+  return !pathname_.empty() &&
+         IsPathSeparator(pathname_.c_str()[pathname_.length() - 1]);
+}
+
+// Create directories so that path exists. Returns true if successful or if
+// the directories already exist; returns false if unable to create directories
+// for any reason.
+bool FilePath::CreateDirectoriesRecursively() const {
+  if (!this->IsDirectory()) {
+    return false;
+  }
+
+  if (pathname_.length() == 0 || this->DirectoryExists()) {
+    return true;
+  }
+
+  const FilePath parent(this->RemoveTrailingPathSeparator().RemoveFileName());
+  return parent.CreateDirectoriesRecursively() && this->CreateFolder();
+}
+
+// Create the directory so that path exists. Returns true if successful or
+// if the directory already exists; returns false if unable to create the
+// directory for any reason, including if the parent directory does not
+// exist. Not named "CreateDirectory" because that's a macro on Windows.
+bool FilePath::CreateFolder() const {
+#if GTEST_OS_WINDOWS_MOBILE
+  FilePath removed_sep(this->RemoveTrailingPathSeparator());
+  LPCWSTR unicode = String::AnsiToUtf16(removed_sep.c_str());
+  int result = CreateDirectory(unicode, NULL) ? 0 : -1;
+  delete [] unicode;
+#elif GTEST_OS_WINDOWS
+  int result = _mkdir(pathname_.c_str());
+#else
+  int result = mkdir(pathname_.c_str(), 0777);
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+  if (result == -1) {
+    return this->DirectoryExists();  // An error is OK if the directory exists.
+  }
+  return true;  // No error.
+}
+
+// If input name has a trailing separator character, remove it and return the
+// name, otherwise return the name string unmodified.
+// On Windows platform, uses \ as the separator, other platforms use /.
+FilePath FilePath::RemoveTrailingPathSeparator() const {
+  return IsDirectory()
+      ? FilePath(pathname_.substr(0, pathname_.length() - 1))
+      : *this;
+}
+
+// Removes any redundant separators that might be in the pathname.
+// For example, "bar///foo" becomes "bar/foo". Does not eliminate other
+// redundancies that might be in a pathname involving "." or "..".
+// TODO(wan@google.com): handle Windows network shares (e.g. \\server\share).
+void FilePath::Normalize() {
+  if (pathname_.c_str() == NULL) {
+    pathname_ = "";
+    return;
+  }
+  const char* src = pathname_.c_str();
+  char* const dest = new char[pathname_.length() + 1];
+  char* dest_ptr = dest;
+  memset(dest_ptr, 0, pathname_.length() + 1);
+
+  while (*src != '\0') {
+    *dest_ptr = *src;
+    if (!IsPathSeparator(*src)) {
+      src++;
+    } else {
+#if GTEST_HAS_ALT_PATH_SEP_
+      if (*dest_ptr == kAlternatePathSeparator) {
+        *dest_ptr = kPathSeparator;
+      }
+#endif
+      while (IsPathSeparator(*src))
+        src++;
+    }
+    dest_ptr++;
+  }
+  *dest_ptr = '\0';
+  pathname_ = dest;
+  delete[] dest;
+}
+
+}  // namespace internal
+}  // namespace testing
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+
+#include <limits.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <fstream>
+
+#if GTEST_OS_WINDOWS
+# include <windows.h>
+# include <io.h>
+# include <sys/stat.h>
+# include <map>  // Used in ThreadLocal.
+#else
+# include <unistd.h>
+#endif  // GTEST_OS_WINDOWS
+
+#if GTEST_OS_MAC
+# include <mach/mach_init.h>
+# include <mach/task.h>
+# include <mach/vm_map.h>
+#endif  // GTEST_OS_MAC
+
+#if GTEST_OS_QNX
+# include <devctl.h>
+# include <fcntl.h>
+# include <sys/procfs.h>
+#endif  // GTEST_OS_QNX
+
+#if GTEST_OS_AIX
+# include <procinfo.h>
+# include <sys/types.h>
+#endif  // GTEST_OS_AIX
+
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick exists to
+// prevent the accidental inclusion of gtest-internal-inl.h in the
+// user's code.
+#define GTEST_IMPLEMENTATION_ 1
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+namespace internal {
+
+#if defined(_MSC_VER) || defined(__BORLANDC__)
+// MSVC and C++Builder do not provide a definition of STDERR_FILENO.
+const int kStdOutFileno = 1;
+const int kStdErrFileno = 2;
+#else
+const int kStdOutFileno = STDOUT_FILENO;
+const int kStdErrFileno = STDERR_FILENO;
+#endif  // _MSC_VER
+
+#if GTEST_OS_LINUX
+
+namespace {
+template <typename T>
+T ReadProcFileField(const std::string& filename, int field) {
+  std::string dummy;
+  std::ifstream file(filename.c_str());
+  while (field-- > 0) {
+    file >> dummy;
+  }
+  T output = 0;
+  file >> output;
+  return output;
+}
+}  // namespace
+
+// Returns the number of active threads, or 0 when there is an error.
+size_t GetThreadCount() {
+  const std::string filename =
+      (Message() << "/proc/" << getpid() << "/stat").GetString();
+  return ReadProcFileField<int>(filename, 19);
+}
+
+#elif GTEST_OS_MAC
+
+size_t GetThreadCount() {
+  const task_t task = mach_task_self();
+  mach_msg_type_number_t thread_count;
+  thread_act_array_t thread_list;
+  const kern_return_t status = task_threads(task, &thread_list, &thread_count);
+  if (status == KERN_SUCCESS) {
+    // task_threads allocates resources in thread_list and we need to free them
+    // to avoid leaks.
+    vm_deallocate(task,
+                  reinterpret_cast<vm_address_t>(thread_list),
+                  sizeof(thread_t) * thread_count);
+    return static_cast<size_t>(thread_count);
+  } else {
+    return 0;
+  }
+}
+
+#elif GTEST_OS_QNX
+
+// Returns the number of threads running in the process, or 0 to indicate that
+// we cannot detect it.
+size_t GetThreadCount() {
+  const int fd = open("/proc/self/as", O_RDONLY);
+  if (fd < 0) {
+    return 0;
+  }
+  procfs_info process_info;
+  const int status =
+      devctl(fd, DCMD_PROC_INFO, &process_info, sizeof(process_info), NULL);
+  close(fd);
+  if (status == EOK) {
+    return static_cast<size_t>(process_info.num_threads);
+  } else {
+    return 0;
+  }
+}
+
+#elif GTEST_OS_AIX
+
+size_t GetThreadCount() {
+  struct procentry64 entry;
+  pid_t pid = getpid();
+  int status = getprocs64(&entry, sizeof(entry), NULL, 0, &pid, 1);
+  if (status == 1) {
+    return entry.pi_thcount;
+  } else {
+    return 0;
+  }
+}
+
+#else
+
+size_t GetThreadCount() {
+  // There's no portable way to detect the number of threads, so we just
+  // return 0 to indicate that we cannot detect it.
+  return 0;
+}
+
+#endif  // GTEST_OS_LINUX
+
+#if GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS
+
+void SleepMilliseconds(int n) {
+  ::Sleep(n);
+}
+
+AutoHandle::AutoHandle()
+    : handle_(INVALID_HANDLE_VALUE) {}
+
+AutoHandle::AutoHandle(Handle handle)
+    : handle_(handle) {}
+
+AutoHandle::~AutoHandle() {
+  Reset();
+}
+
+AutoHandle::Handle AutoHandle::Get() const {
+  return handle_;
+}
+
+void AutoHandle::Reset() {
+  Reset(INVALID_HANDLE_VALUE);
+}
+
+void AutoHandle::Reset(HANDLE handle) {
+  // Resetting with the same handle we already own is invalid.
+  if (handle_ != handle) {
+    if (IsCloseable()) {
+      ::CloseHandle(handle_);
+    }
+    handle_ = handle;
+  } else {
+    GTEST_CHECK_(!IsCloseable())
+        << "Resetting a valid handle to itself is likely a programmer error "
+            "and thus not allowed.";
+  }
+}
+
+bool AutoHandle::IsCloseable() const {
+  // Different Windows APIs may use either of these values to represent an
+  // invalid handle.
+  return handle_ != NULL && handle_ != INVALID_HANDLE_VALUE;
+}
+
+Notification::Notification()
+    : event_(::CreateEvent(NULL,   // Default security attributes.
+                           TRUE,   // Do not reset automatically.
+                           FALSE,  // Initially unset.
+                           NULL)) {  // Anonymous event.
+  GTEST_CHECK_(event_.Get() != NULL);
+}
+
+void Notification::Notify() {
+  GTEST_CHECK_(::SetEvent(event_.Get()) != FALSE);
+}
+
+void Notification::WaitForNotification() {
+  GTEST_CHECK_(
+      ::WaitForSingleObject(event_.Get(), INFINITE) == WAIT_OBJECT_0);
+}
+
+Mutex::Mutex()
+    : owner_thread_id_(0),
+      type_(kDynamic),
+      critical_section_init_phase_(0),
+      critical_section_(new CRITICAL_SECTION) {
+  ::InitializeCriticalSection(critical_section_);
+}
+
+Mutex::~Mutex() {
+  // Static mutexes are leaked intentionally. It is not thread-safe to try
+  // to clean them up.
+  // TODO(yukawa): Switch to Slim Reader/Writer (SRW) Locks, which requires
+  // nothing to clean it up but is available only on Vista and later.
+  // http://msdn.microsoft.com/en-us/library/windows/desktop/aa904937.aspx
+  if (type_ == kDynamic) {
+    ::DeleteCriticalSection(critical_section_);
+    delete critical_section_;
+    critical_section_ = NULL;
+  }
+}
+
+void Mutex::Lock() {
+  ThreadSafeLazyInit();
+  ::EnterCriticalSection(critical_section_);
+  owner_thread_id_ = ::GetCurrentThreadId();
+}
+
+void Mutex::Unlock() {
+  ThreadSafeLazyInit();
+  // We don't protect writing to owner_thread_id_ here, as it's the
+  // caller's responsibility to ensure that the current thread holds the
+  // mutex when this is called.
+  owner_thread_id_ = 0;
+  ::LeaveCriticalSection(critical_section_);
+}
+
+// Does nothing if the current thread holds the mutex. Otherwise, crashes
+// with high probability.
+void Mutex::AssertHeld() {
+  ThreadSafeLazyInit();
+  GTEST_CHECK_(owner_thread_id_ == ::GetCurrentThreadId())
+      << "The current thread is not holding the mutex @" << this;
+}
+
+// Initializes owner_thread_id_ and critical_section_ in static mutexes.
+void Mutex::ThreadSafeLazyInit() {
+  // Dynamic mutexes are initialized in the constructor.
+  if (type_ == kStatic) {
+    switch (
+        ::InterlockedCompareExchange(&critical_section_init_phase_, 1L, 0L)) {
+      case 0:
+        // If critical_section_init_phase_ was 0 before the exchange, we
+        // are the first to test it and need to perform the initialization.
+        owner_thread_id_ = 0;
+        critical_section_ = new CRITICAL_SECTION;
+        ::InitializeCriticalSection(critical_section_);
+        // Updates the critical_section_init_phase_ to 2 to signal
+        // initialization complete.
+        GTEST_CHECK_(::InterlockedCompareExchange(
+                          &critical_section_init_phase_, 2L, 1L) ==
+                      1L);
+        break;
+      case 1:
+        // Somebody else is already initializing the mutex; spin until they
+        // are done.
+        while (::InterlockedCompareExchange(&critical_section_init_phase_,
+                                            2L,
+                                            2L) != 2L) {
+          // Possibly yields the rest of the thread's time slice to other
+          // threads.
+          ::Sleep(0);
+        }
+        break;
+
+      case 2:
+        break;  // The mutex is already initialized and ready for use.
+
+      default:
+        GTEST_CHECK_(false)
+            << "Unexpected value of critical_section_init_phase_ "
+            << "while initializing a static mutex.";
+    }
+  }
+}
+
+namespace {
+
+class ThreadWithParamSupport : public ThreadWithParamBase {
+ public:
+  static HANDLE CreateThread(Runnable* runnable,
+                             Notification* thread_can_start) {
+    ThreadMainParam* param = new ThreadMainParam(runnable, thread_can_start);
+    DWORD thread_id;
+    // TODO(yukawa): Consider to use _beginthreadex instead.
+    HANDLE thread_handle = ::CreateThread(
+        NULL,    // Default security.
+        0,       // Default stack size.
+        &ThreadWithParamSupport::ThreadMain,
+        param,   // Parameter to ThreadMainStatic
+        0x0,     // Default creation flags.
+        &thread_id);  // Need a valid pointer for the call to work under Win98.
+    GTEST_CHECK_(thread_handle != NULL) << "CreateThread failed with error "
+                                        << ::GetLastError() << ".";
+    if (thread_handle == NULL) {
+      delete param;
+    }
+    return thread_handle;
+  }
+
+ private:
+  struct ThreadMainParam {
+    ThreadMainParam(Runnable* runnable, Notification* thread_can_start)
+        : runnable_(runnable),
+          thread_can_start_(thread_can_start) {
+    }
+    scoped_ptr<Runnable> runnable_;
+    // Does not own.
+    Notification* thread_can_start_;
+  };
+
+  static DWORD WINAPI ThreadMain(void* ptr) {
+    // Transfers ownership.
+    scoped_ptr<ThreadMainParam> param(static_cast<ThreadMainParam*>(ptr));
+    if (param->thread_can_start_ != NULL)
+      param->thread_can_start_->WaitForNotification();
+    param->runnable_->Run();
+    return 0;
+  }
+
+  // Prohibit instantiation.
+  ThreadWithParamSupport();
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParamSupport);
+};
+
+}  // namespace
+
+ThreadWithParamBase::ThreadWithParamBase(Runnable *runnable,
+                                         Notification* thread_can_start)
+      : thread_(ThreadWithParamSupport::CreateThread(runnable,
+                                                     thread_can_start)) {
+}
+
+ThreadWithParamBase::~ThreadWithParamBase() {
+  Join();
+}
+
+void ThreadWithParamBase::Join() {
+  GTEST_CHECK_(::WaitForSingleObject(thread_.Get(), INFINITE) == WAIT_OBJECT_0)
+      << "Failed to join the thread with error " << ::GetLastError() << ".";
+}
+
+// Maps a thread to a set of ThreadIdToThreadLocals that have values
+// instantiated on that thread and notifies them when the thread exits.  A
+// ThreadLocal instance is expected to persist until all threads it has
+// values on have terminated.
+class ThreadLocalRegistryImpl {
+ public:
+  // Registers thread_local_instance as having value on the current thread.
+  // Returns a value that can be used to identify the thread from other threads.
+  static ThreadLocalValueHolderBase* GetValueOnCurrentThread(
+      const ThreadLocalBase* thread_local_instance) {
+    DWORD current_thread = ::GetCurrentThreadId();
+    MutexLock lock(&mutex_);
+    ThreadIdToThreadLocals* const thread_to_thread_locals =
+        GetThreadLocalsMapLocked();
+    ThreadIdToThreadLocals::iterator thread_local_pos =
+        thread_to_thread_locals->find(current_thread);
+    if (thread_local_pos == thread_to_thread_locals->end()) {
+      thread_local_pos = thread_to_thread_locals->insert(
+          std::make_pair(current_thread, ThreadLocalValues())).first;
+      StartWatcherThreadFor(current_thread);
+    }
+    ThreadLocalValues& thread_local_values = thread_local_pos->second;
+    ThreadLocalValues::iterator value_pos =
+        thread_local_values.find(thread_local_instance);
+    if (value_pos == thread_local_values.end()) {
+      value_pos =
+          thread_local_values
+              .insert(std::make_pair(
+                  thread_local_instance,
+                  linked_ptr<ThreadLocalValueHolderBase>(
+                      thread_local_instance->NewValueForCurrentThread())))
+              .first;
+    }
+    return value_pos->second.get();
+  }
+
+  static void OnThreadLocalDestroyed(
+      const ThreadLocalBase* thread_local_instance) {
+    std::vector<linked_ptr<ThreadLocalValueHolderBase> > value_holders;
+    // Clean up the ThreadLocalValues data structure while holding the lock, but
+    // defer the destruction of the ThreadLocalValueHolderBases.
+    {
+      MutexLock lock(&mutex_);
+      ThreadIdToThreadLocals* const thread_to_thread_locals =
+          GetThreadLocalsMapLocked();
+      for (ThreadIdToThreadLocals::iterator it =
+          thread_to_thread_locals->begin();
+          it != thread_to_thread_locals->end();
+          ++it) {
+        ThreadLocalValues& thread_local_values = it->second;
+        ThreadLocalValues::iterator value_pos =
+            thread_local_values.find(thread_local_instance);
+        if (value_pos != thread_local_values.end()) {
+          value_holders.push_back(value_pos->second);
+          thread_local_values.erase(value_pos);
+          // This 'if' can only be successful at most once, so theoretically we
+          // could break out of the loop here, but we don't bother doing so.
+        }
+      }
+    }
+    // Outside the lock, let the destructor for 'value_holders' deallocate the
+    // ThreadLocalValueHolderBases.
+  }
+
+  static void OnThreadExit(DWORD thread_id) {
+    GTEST_CHECK_(thread_id != 0) << ::GetLastError();
+    std::vector<linked_ptr<ThreadLocalValueHolderBase> > value_holders;
+    // Clean up the ThreadIdToThreadLocals data structure while holding the
+    // lock, but defer the destruction of the ThreadLocalValueHolderBases.
+    {
+      MutexLock lock(&mutex_);
+      ThreadIdToThreadLocals* const thread_to_thread_locals =
+          GetThreadLocalsMapLocked();
+      ThreadIdToThreadLocals::iterator thread_local_pos =
+          thread_to_thread_locals->find(thread_id);
+      if (thread_local_pos != thread_to_thread_locals->end()) {
+        ThreadLocalValues& thread_local_values = thread_local_pos->second;
+        for (ThreadLocalValues::iterator value_pos =
+            thread_local_values.begin();
+            value_pos != thread_local_values.end();
+            ++value_pos) {
+          value_holders.push_back(value_pos->second);
+        }
+        thread_to_thread_locals->erase(thread_local_pos);
+      }
+    }
+    // Outside the lock, let the destructor for 'value_holders' deallocate the
+    // ThreadLocalValueHolderBases.
+  }
+
+ private:
+  // In a particular thread, maps a ThreadLocal object to its value.
+  typedef std::map<const ThreadLocalBase*,
+                   linked_ptr<ThreadLocalValueHolderBase> > ThreadLocalValues;
+  // Stores all ThreadIdToThreadLocals having values in a thread, indexed by
+  // thread's ID.
+  typedef std::map<DWORD, ThreadLocalValues> ThreadIdToThreadLocals;
+
+  // Holds the thread id and thread handle that we pass from
+  // StartWatcherThreadFor to WatcherThreadFunc.
+  typedef std::pair<DWORD, HANDLE> ThreadIdAndHandle;
+
+  static void StartWatcherThreadFor(DWORD thread_id) {
+    // The returned handle will be kept in thread_map and closed by
+    // watcher_thread in WatcherThreadFunc.
+    HANDLE thread = ::OpenThread(SYNCHRONIZE | THREAD_QUERY_INFORMATION,
+                                 FALSE,
+                                 thread_id);
+    GTEST_CHECK_(thread != NULL);
+    // We need to to pass a valid thread ID pointer into CreateThread for it
+    // to work correctly under Win98.
+    DWORD watcher_thread_id;
+    HANDLE watcher_thread = ::CreateThread(
+        NULL,   // Default security.
+        0,      // Default stack size
+        &ThreadLocalRegistryImpl::WatcherThreadFunc,
+        reinterpret_cast<LPVOID>(new ThreadIdAndHandle(thread_id, thread)),
+        CREATE_SUSPENDED,
+        &watcher_thread_id);
+    GTEST_CHECK_(watcher_thread != NULL);
+    // Give the watcher thread the same priority as ours to avoid being
+    // blocked by it.
+    ::SetThreadPriority(watcher_thread,
+                        ::GetThreadPriority(::GetCurrentThread()));
+    ::ResumeThread(watcher_thread);
+    ::CloseHandle(watcher_thread);
+  }
+
+  // Monitors exit from a given thread and notifies those
+  // ThreadIdToThreadLocals about thread termination.
+  static DWORD WINAPI WatcherThreadFunc(LPVOID param) {
+    const ThreadIdAndHandle* tah =
+        reinterpret_cast<const ThreadIdAndHandle*>(param);
+    GTEST_CHECK_(
+        ::WaitForSingleObject(tah->second, INFINITE) == WAIT_OBJECT_0);
+    OnThreadExit(tah->first);
+    ::CloseHandle(tah->second);
+    delete tah;
+    return 0;
+  }
+
+  // Returns map of thread local instances.
+  static ThreadIdToThreadLocals* GetThreadLocalsMapLocked() {
+    mutex_.AssertHeld();
+    static ThreadIdToThreadLocals* map = new ThreadIdToThreadLocals;
+    return map;
+  }
+
+  // Protects access to GetThreadLocalsMapLocked() and its return value.
+  static Mutex mutex_;
+  // Protects access to GetThreadMapLocked() and its return value.
+  static Mutex thread_map_mutex_;
+};
+
+Mutex ThreadLocalRegistryImpl::mutex_(Mutex::kStaticMutex);
+Mutex ThreadLocalRegistryImpl::thread_map_mutex_(Mutex::kStaticMutex);
+
+ThreadLocalValueHolderBase* ThreadLocalRegistry::GetValueOnCurrentThread(
+      const ThreadLocalBase* thread_local_instance) {
+  return ThreadLocalRegistryImpl::GetValueOnCurrentThread(
+      thread_local_instance);
+}
+
+void ThreadLocalRegistry::OnThreadLocalDestroyed(
+      const ThreadLocalBase* thread_local_instance) {
+  ThreadLocalRegistryImpl::OnThreadLocalDestroyed(thread_local_instance);
+}
+
+#endif  // GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS
+
+#if GTEST_USES_POSIX_RE
+
+// Implements RE.  Currently only needed for death tests.
+
+RE::~RE() {
+  if (is_valid_) {
+    // regfree'ing an invalid regex might crash because the content
+    // of the regex is undefined. Since the regex's are essentially
+    // the same, one cannot be valid (or invalid) without the other
+    // being so too.
+    regfree(&partial_regex_);
+    regfree(&full_regex_);
+  }
+  free(const_cast<char*>(pattern_));
+}
+
+// Returns true iff regular expression re matches the entire str.
+bool RE::FullMatch(const char* str, const RE& re) {
+  if (!re.is_valid_) return false;
+
+  regmatch_t match;
+  return regexec(&re.full_regex_, str, 1, &match, 0) == 0;
+}
+
+// Returns true iff regular expression re matches a substring of str
+// (including str itself).
+bool RE::PartialMatch(const char* str, const RE& re) {
+  if (!re.is_valid_) return false;
+
+  regmatch_t match;
+  return regexec(&re.partial_regex_, str, 1, &match, 0) == 0;
+}
+
+// Initializes an RE from its string representation.
+void RE::Init(const char* regex) {
+  pattern_ = posix::StrDup(regex);
+
+  // Reserves enough bytes to hold the regular expression used for a
+  // full match.
+  const size_t full_regex_len = strlen(regex) + 10;
+  char* const full_pattern = new char[full_regex_len];
+
+  snprintf(full_pattern, full_regex_len, "^(%s)$", regex);
+  is_valid_ = regcomp(&full_regex_, full_pattern, REG_EXTENDED) == 0;
+  // We want to call regcomp(&partial_regex_, ...) even if the
+  // previous expression returns false.  Otherwise partial_regex_ may
+  // not be properly initialized can may cause trouble when it's
+  // freed.
+  //
+  // Some implementation of POSIX regex (e.g. on at least some
+  // versions of Cygwin) doesn't accept the empty string as a valid
+  // regex.  We change it to an equivalent form "()" to be safe.
+  if (is_valid_) {
+    const char* const partial_regex = (*regex == '\0') ? "()" : regex;
+    is_valid_ = regcomp(&partial_regex_, partial_regex, REG_EXTENDED) == 0;
+  }
+  EXPECT_TRUE(is_valid_)
+      << "Regular expression \"" << regex
+      << "\" is not a valid POSIX Extended regular expression.";
+
+  delete[] full_pattern;
+}
+
+#elif GTEST_USES_SIMPLE_RE
+
+// Returns true iff ch appears anywhere in str (excluding the
+// terminating '\0' character).
+bool IsInSet(char ch, const char* str) {
+  return ch != '\0' && strchr(str, ch) != NULL;
+}
+
+// Returns true iff ch belongs to the given classification.  Unlike
+// similar functions in <ctype.h>, these aren't affected by the
+// current locale.
+bool IsAsciiDigit(char ch) { return '0' <= ch && ch <= '9'; }
+bool IsAsciiPunct(char ch) {
+  return IsInSet(ch, "^-!\"#$%&'()*+,./:;<=>?@[\\]_`{|}~");
+}
+bool IsRepeat(char ch) { return IsInSet(ch, "?*+"); }
+bool IsAsciiWhiteSpace(char ch) { return IsInSet(ch, " \f\n\r\t\v"); }
+bool IsAsciiWordChar(char ch) {
+  return ('a' <= ch && ch <= 'z') || ('A' <= ch && ch <= 'Z') ||
+      ('0' <= ch && ch <= '9') || ch == '_';
+}
+
+// Returns true iff "\\c" is a supported escape sequence.
+bool IsValidEscape(char c) {
+  return (IsAsciiPunct(c) || IsInSet(c, "dDfnrsStvwW"));
+}
+
+// Returns true iff the given atom (specified by escaped and pattern)
+// matches ch.  The result is undefined if the atom is invalid.
+bool AtomMatchesChar(bool escaped, char pattern_char, char ch) {
+  if (escaped) {  // "\\p" where p is pattern_char.
+    switch (pattern_char) {
+      case 'd': return IsAsciiDigit(ch);
+      case 'D': return !IsAsciiDigit(ch);
+      case 'f': return ch == '\f';
+      case 'n': return ch == '\n';
+      case 'r': return ch == '\r';
+      case 's': return IsAsciiWhiteSpace(ch);
+      case 'S': return !IsAsciiWhiteSpace(ch);
+      case 't': return ch == '\t';
+      case 'v': return ch == '\v';
+      case 'w': return IsAsciiWordChar(ch);
+      case 'W': return !IsAsciiWordChar(ch);
+    }
+    return IsAsciiPunct(pattern_char) && pattern_char == ch;
+  }
+
+  return (pattern_char == '.' && ch != '\n') || pattern_char == ch;
+}
+
+// Helper function used by ValidateRegex() to format error messages.
+std::string FormatRegexSyntaxError(const char* regex, int index) {
+  return (Message() << "Syntax error at index " << index
+          << " in simple regular expression \"" << regex << "\": ").GetString();
+}
+
+// Generates non-fatal failures and returns false if regex is invalid;
+// otherwise returns true.
+bool ValidateRegex(const char* regex) {
+  if (regex == NULL) {
+    // TODO(wan@google.com): fix the source file location in the
+    // assertion failures to match where the regex is used in user
+    // code.
+    ADD_FAILURE() << "NULL is not a valid simple regular expression.";
+    return false;
+  }
+
+  bool is_valid = true;
+
+  // True iff ?, *, or + can follow the previous atom.
+  bool prev_repeatable = false;
+  for (int i = 0; regex[i]; i++) {
+    if (regex[i] == '\\') {  // An escape sequence
+      i++;
+      if (regex[i] == '\0') {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
+                      << "'\\' cannot appear at the end.";
+        return false;
+      }
+
+      if (!IsValidEscape(regex[i])) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
+                      << "invalid escape sequence \"\\" << regex[i] << "\".";
+        is_valid = false;
+      }
+      prev_repeatable = true;
+    } else {  // Not an escape sequence.
+      const char ch = regex[i];
+
+      if (ch == '^' && i > 0) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'^' can only appear at the beginning.";
+        is_valid = false;
+      } else if (ch == '$' && regex[i + 1] != '\0') {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'$' can only appear at the end.";
+        is_valid = false;
+      } else if (IsInSet(ch, "()[]{}|")) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'" << ch << "' is unsupported.";
+        is_valid = false;
+      } else if (IsRepeat(ch) && !prev_repeatable) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'" << ch << "' can only follow a repeatable token.";
+        is_valid = false;
+      }
+
+      prev_repeatable = !IsInSet(ch, "^$?*+");
+    }
+  }
+
+  return is_valid;
+}
+
+// Matches a repeated regex atom followed by a valid simple regular
+// expression.  The regex atom is defined as c if escaped is false,
+// or \c otherwise.  repeat is the repetition meta character (?, *,
+// or +).  The behavior is undefined if str contains too many
+// characters to be indexable by size_t, in which case the test will
+// probably time out anyway.  We are fine with this limitation as
+// std::string has it too.
+bool MatchRepetitionAndRegexAtHead(
+    bool escaped, char c, char repeat, const char* regex,
+    const char* str) {
+  const size_t min_count = (repeat == '+') ? 1 : 0;
+  const size_t max_count = (repeat == '?') ? 1 :
+      static_cast<size_t>(-1) - 1;
+  // We cannot call numeric_limits::max() as it conflicts with the
+  // max() macro on Windows.
+
+  for (size_t i = 0; i <= max_count; ++i) {
+    // We know that the atom matches each of the first i characters in str.
+    if (i >= min_count && MatchRegexAtHead(regex, str + i)) {
+      // We have enough matches at the head, and the tail matches too.
+      // Since we only care about *whether* the pattern matches str
+      // (as opposed to *how* it matches), there is no need to find a
+      // greedy match.
+      return true;
+    }
+    if (str[i] == '\0' || !AtomMatchesChar(escaped, c, str[i]))
+      return false;
+  }
+  return false;
+}
+
+// Returns true iff regex matches a prefix of str.  regex must be a
+// valid simple regular expression and not start with "^", or the
+// result is undefined.
+bool MatchRegexAtHead(const char* regex, const char* str) {
+  if (*regex == '\0')  // An empty regex matches a prefix of anything.
+    return true;
+
+  // "$" only matches the end of a string.  Note that regex being
+  // valid guarantees that there's nothing after "$" in it.
+  if (*regex == '$')
+    return *str == '\0';
+
+  // Is the first thing in regex an escape sequence?
+  const bool escaped = *regex == '\\';
+  if (escaped)
+    ++regex;
+  if (IsRepeat(regex[1])) {
+    // MatchRepetitionAndRegexAtHead() calls MatchRegexAtHead(), so
+    // here's an indirect recursion.  It terminates as the regex gets
+    // shorter in each recursion.
+    return MatchRepetitionAndRegexAtHead(
+        escaped, regex[0], regex[1], regex + 2, str);
+  } else {
+    // regex isn't empty, isn't "$", and doesn't start with a
+    // repetition.  We match the first atom of regex with the first
+    // character of str and recurse.
+    return (*str != '\0') && AtomMatchesChar(escaped, *regex, *str) &&
+        MatchRegexAtHead(regex + 1, str + 1);
+  }
+}
+
+// Returns true iff regex matches any substring of str.  regex must be
+// a valid simple regular expression, or the result is undefined.
+//
+// The algorithm is recursive, but the recursion depth doesn't exceed
+// the regex length, so we won't need to worry about running out of
+// stack space normally.  In rare cases the time complexity can be
+// exponential with respect to the regex length + the string length,
+// but usually it's must faster (often close to linear).
+bool MatchRegexAnywhere(const char* regex, const char* str) {
+  if (regex == NULL || str == NULL)
+    return false;
+
+  if (*regex == '^')
+    return MatchRegexAtHead(regex + 1, str);
+
+  // A successful match can be anywhere in str.
+  do {
+    if (MatchRegexAtHead(regex, str))
+      return true;
+  } while (*str++ != '\0');
+  return false;
+}
+
+// Implements the RE class.
+
+RE::~RE() {
+  free(const_cast<char*>(pattern_));
+  free(const_cast<char*>(full_pattern_));
+}
+
+// Returns true iff regular expression re matches the entire str.
+bool RE::FullMatch(const char* str, const RE& re) {
+  return re.is_valid_ && MatchRegexAnywhere(re.full_pattern_, str);
+}
+
+// Returns true iff regular expression re matches a substring of str
+// (including str itself).
+bool RE::PartialMatch(const char* str, const RE& re) {
+  return re.is_valid_ && MatchRegexAnywhere(re.pattern_, str);
+}
+
+// Initializes an RE from its string representation.
+void RE::Init(const char* regex) {
+  pattern_ = full_pattern_ = NULL;
+  if (regex != NULL) {
+    pattern_ = posix::StrDup(regex);
+  }
+
+  is_valid_ = ValidateRegex(regex);
+  if (!is_valid_) {
+    // No need to calculate the full pattern when the regex is invalid.
+    return;
+  }
+
+  const size_t len = strlen(regex);
+  // Reserves enough bytes to hold the regular expression used for a
+  // full match: we need space to prepend a '^', append a '$', and
+  // terminate the string with '\0'.
+  char* buffer = static_cast<char*>(malloc(len + 3));
+  full_pattern_ = buffer;
+
+  if (*regex != '^')
+    *buffer++ = '^';  // Makes sure full_pattern_ starts with '^'.
+
+  // We don't use snprintf or strncpy, as they trigger a warning when
+  // compiled with VC++ 8.0.
+  memcpy(buffer, regex, len);
+  buffer += len;
+
+  if (len == 0 || regex[len - 1] != '$')
+    *buffer++ = '$';  // Makes sure full_pattern_ ends with '$'.
+
+  *buffer = '\0';
+}
+
+#endif  // GTEST_USES_POSIX_RE
+
+const char kUnknownFile[] = "unknown file";
+
+// Formats a source file path and a line number as they would appear
+// in an error message from the compiler used to compile this code.
+GTEST_API_ ::std::string FormatFileLocation(const char* file, int line) {
+  const std::string file_name(file == NULL ? kUnknownFile : file);
+
+  if (line < 0) {
+    return file_name + ":";
+  }
+#ifdef _MSC_VER
+  return file_name + "(" + StreamableToString(line) + "):";
+#else
+  return file_name + ":" + StreamableToString(line) + ":";
+#endif  // _MSC_VER
+}
+
+// Formats a file location for compiler-independent XML output.
+// Although this function is not platform dependent, we put it next to
+// FormatFileLocation in order to contrast the two functions.
+// Note that FormatCompilerIndependentFileLocation() does NOT append colon
+// to the file location it produces, unlike FormatFileLocation().
+GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(
+    const char* file, int line) {
+  const std::string file_name(file == NULL ? kUnknownFile : file);
+
+  if (line < 0)
+    return file_name;
+  else
+    return file_name + ":" + StreamableToString(line);
+}
+
+GTestLog::GTestLog(GTestLogSeverity severity, const char* file, int line)
+    : severity_(severity) {
+  const char* const marker =
+      severity == GTEST_INFO ?    "[  INFO ]" :
+      severity == GTEST_WARNING ? "[WARNING]" :
+      severity == GTEST_ERROR ?   "[ ERROR ]" : "[ FATAL ]";
+  GetStream() << ::std::endl << marker << " "
+              << FormatFileLocation(file, line).c_str() << ": ";
+}
+
+// Flushes the buffers and, if severity is GTEST_FATAL, aborts the program.
+GTestLog::~GTestLog() {
+  GetStream() << ::std::endl;
+  if (severity_ == GTEST_FATAL) {
+    fflush(stderr);
+    posix::Abort();
+  }
+}
+// Disable Microsoft deprecation warnings for POSIX functions called from
+// this class (creat, dup, dup2, and close)
+GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996)
+
+#if GTEST_HAS_STREAM_REDIRECTION
+
+// Object that captures an output stream (stdout/stderr).
+class CapturedStream {
+ public:
+  // The ctor redirects the stream to a temporary file.
+  explicit CapturedStream(int fd) : fd_(fd), uncaptured_fd_(dup(fd)) {
+# if GTEST_OS_WINDOWS
+    char temp_dir_path[MAX_PATH + 1] = { '\0' };  // NOLINT
+    char temp_file_path[MAX_PATH + 1] = { '\0' };  // NOLINT
+
+    ::GetTempPathA(sizeof(temp_dir_path), temp_dir_path);
+    const UINT success = ::GetTempFileNameA(temp_dir_path,
+                                            "gtest_redir",
+                                            0,  // Generate unique file name.
+                                            temp_file_path);
+    GTEST_CHECK_(success != 0)
+        << "Unable to create a temporary file in " << temp_dir_path;
+    const int captured_fd = creat(temp_file_path, _S_IREAD | _S_IWRITE);
+    GTEST_CHECK_(captured_fd != -1) << "Unable to open temporary file "
+                                    << temp_file_path;
+    filename_ = temp_file_path;
+# else
+    // There's no guarantee that a test has write access to the current
+    // directory, so we create the temporary file in the /tmp directory
+    // instead. We use /tmp on most systems, and /sdcard on Android.
+    // That's because Android doesn't have /tmp.
+#  if GTEST_OS_LINUX_ANDROID
+    // Note: Android applications are expected to call the framework's
+    // Context.getExternalStorageDirectory() method through JNI to get
+    // the location of the world-writable SD Card directory. However,
+    // this requires a Context handle, which cannot be retrieved
+    // globally from native code. Doing so also precludes running the
+    // code as part of a regular standalone executable, which doesn't
+    // run in a Dalvik process (e.g. when running it through 'adb shell').
+    //
+    // The location /sdcard is directly accessible from native code
+    // and is the only location (unofficially) supported by the Android
+    // team. It's generally a symlink to the real SD Card mount point
+    // which can be /mnt/sdcard, /mnt/sdcard0, /system/media/sdcard, or
+    // other OEM-customized locations. Never rely on these, and always
+    // use /sdcard.
+    char name_template[] = "/sdcard/gtest_captured_stream.XXXXXX";
+#  else
+    char name_template[] = "/tmp/captured_stream.XXXXXX";
+#  endif  // GTEST_OS_LINUX_ANDROID
+    const int captured_fd = mkstemp(name_template);
+    filename_ = name_template;
+# endif  // GTEST_OS_WINDOWS
+    fflush(NULL);
+    dup2(captured_fd, fd_);
+    close(captured_fd);
+  }
+
+  ~CapturedStream() {
+    remove(filename_.c_str());
+  }
+
+  std::string GetCapturedString() {
+    if (uncaptured_fd_ != -1) {
+      // Restores the original stream.
+      fflush(NULL);
+      dup2(uncaptured_fd_, fd_);
+      close(uncaptured_fd_);
+      uncaptured_fd_ = -1;
+    }
+
+    FILE* const file = posix::FOpen(filename_.c_str(), "r");
+    const std::string content = ReadEntireFile(file);
+    posix::FClose(file);
+    return content;
+  }
+
+ private:
+  const int fd_;  // A stream to capture.
+  int uncaptured_fd_;
+  // Name of the temporary file holding the stderr output.
+  ::std::string filename_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(CapturedStream);
+};
+
+GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+static CapturedStream* g_captured_stderr = NULL;
+static CapturedStream* g_captured_stdout = NULL;
+
+// Starts capturing an output stream (stdout/stderr).
+void CaptureStream(int fd, const char* stream_name, CapturedStream** stream) {
+  if (*stream != NULL) {
+    GTEST_LOG_(FATAL) << "Only one " << stream_name
+                      << " capturer can exist at a time.";
+  }
+  *stream = new CapturedStream(fd);
+}
+
+// Stops capturing the output stream and returns the captured string.
+std::string GetCapturedStream(CapturedStream** captured_stream) {
+  const std::string content = (*captured_stream)->GetCapturedString();
+
+  delete *captured_stream;
+  *captured_stream = NULL;
+
+  return content;
+}
+
+// Starts capturing stdout.
+void CaptureStdout() {
+  CaptureStream(kStdOutFileno, "stdout", &g_captured_stdout);
+}
+
+// Starts capturing stderr.
+void CaptureStderr() {
+  CaptureStream(kStdErrFileno, "stderr", &g_captured_stderr);
+}
+
+// Stops capturing stdout and returns the captured string.
+std::string GetCapturedStdout() {
+  return GetCapturedStream(&g_captured_stdout);
+}
+
+// Stops capturing stderr and returns the captured string.
+std::string GetCapturedStderr() {
+  return GetCapturedStream(&g_captured_stderr);
+}
+
+#endif  // GTEST_HAS_STREAM_REDIRECTION
+
+size_t GetFileSize(FILE* file) {
+  fseek(file, 0, SEEK_END);
+  return static_cast<size_t>(ftell(file));
+}
+
+std::string ReadEntireFile(FILE* file) {
+  const size_t file_size = GetFileSize(file);
+  char* const buffer = new char[file_size];
+
+  size_t bytes_last_read = 0;  // # of bytes read in the last fread()
+  size_t bytes_read = 0;       // # of bytes read so far
+
+  fseek(file, 0, SEEK_SET);
+
+  // Keeps reading the file until we cannot read further or the
+  // pre-determined file size is reached.
+  do {
+    bytes_last_read = fread(buffer+bytes_read, 1, file_size-bytes_read, file);
+    bytes_read += bytes_last_read;
+  } while (bytes_last_read > 0 && bytes_read < file_size);
+
+  const std::string content(buffer, bytes_read);
+  delete[] buffer;
+
+  return content;
+}
+
+#if GTEST_HAS_DEATH_TEST
+
+static const ::std::vector<testing::internal::string>* g_injected_test_argvs =
+                                        NULL;  // Owned.
+
+void SetInjectableArgvs(const ::std::vector<testing::internal::string>* argvs) {
+  if (g_injected_test_argvs != argvs)
+    delete g_injected_test_argvs;
+  g_injected_test_argvs = argvs;
+}
+
+const ::std::vector<testing::internal::string>& GetInjectableArgvs() {
+  if (g_injected_test_argvs != NULL) {
+    return *g_injected_test_argvs;
+  }
+  return GetArgvs();
+}
+#endif  // GTEST_HAS_DEATH_TEST
+
+#if GTEST_OS_WINDOWS_MOBILE
+namespace posix {
+void Abort() {
+  DebugBreak();
+  TerminateProcess(GetCurrentProcess(), 1);
+}
+}  // namespace posix
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+// Returns the name of the environment variable corresponding to the
+// given flag.  For example, FlagToEnvVar("foo") will return
+// "GTEST_FOO" in the open-source version.
+static std::string FlagToEnvVar(const char* flag) {
+  const std::string full_flag =
+      (Message() << GTEST_FLAG_PREFIX_ << flag).GetString();
+
+  Message env_var;
+  for (size_t i = 0; i != full_flag.length(); i++) {
+    env_var << ToUpper(full_flag.c_str()[i]);
+  }
+
+  return env_var.GetString();
+}
+
+// Parses 'str' for a 32-bit signed integer.  If successful, writes
+// the result to *value and returns true; otherwise leaves *value
+// unchanged and returns false.
+bool ParseInt32(const Message& src_text, const char* str, Int32* value) {
+  // Parses the environment variable as a decimal integer.
+  char* end = NULL;
+  const long long_value = strtol(str, &end, 10);  // NOLINT
+
+  // Has strtol() consumed all characters in the string?
+  if (*end != '\0') {
+    // No - an invalid character was encountered.
+    Message msg;
+    msg << "WARNING: " << src_text
+        << " is expected to be a 32-bit integer, but actually"
+        << " has value \"" << str << "\".\n";
+    printf("%s", msg.GetString().c_str());
+    fflush(stdout);
+    return false;
+  }
+
+  // Is the parsed value in the range of an Int32?
+  const Int32 result = static_cast<Int32>(long_value);
+  if (long_value == LONG_MAX || long_value == LONG_MIN ||
+      // The parsed value overflows as a long.  (strtol() returns
+      // LONG_MAX or LONG_MIN when the input overflows.)
+      result != long_value
+      // The parsed value overflows as an Int32.
+      ) {
+    Message msg;
+    msg << "WARNING: " << src_text
+        << " is expected to be a 32-bit integer, but actually"
+        << " has value " << str << ", which overflows.\n";
+    printf("%s", msg.GetString().c_str());
+    fflush(stdout);
+    return false;
+  }
+
+  *value = result;
+  return true;
+}
+
+// Reads and returns the Boolean environment variable corresponding to
+// the given flag; if it's not set, returns default_value.
+//
+// The value is considered true iff it's not "0".
+bool BoolFromGTestEnv(const char* flag, bool default_value) {
+#if defined(GTEST_GET_BOOL_FROM_ENV_)
+  return GTEST_GET_BOOL_FROM_ENV_(flag, default_value);
+#endif  // defined(GTEST_GET_BOOL_FROM_ENV_)
+  const std::string env_var = FlagToEnvVar(flag);
+  const char* const string_value = posix::GetEnv(env_var.c_str());
+  return string_value == NULL ?
+      default_value : strcmp(string_value, "0") != 0;
+}
+
+// Reads and returns a 32-bit integer stored in the environment
+// variable corresponding to the given flag; if it isn't set or
+// doesn't represent a valid 32-bit integer, returns default_value.
+Int32 Int32FromGTestEnv(const char* flag, Int32 default_value) {
+#if defined(GTEST_GET_INT32_FROM_ENV_)
+  return GTEST_GET_INT32_FROM_ENV_(flag, default_value);
+#endif  // defined(GTEST_GET_INT32_FROM_ENV_)
+  const std::string env_var = FlagToEnvVar(flag);
+  const char* const string_value = posix::GetEnv(env_var.c_str());
+  if (string_value == NULL) {
+    // The environment variable is not set.
+    return default_value;
+  }
+
+  Int32 result = default_value;
+  if (!ParseInt32(Message() << "Environment variable " << env_var,
+                  string_value, &result)) {
+    printf("The default value %s is used.\n",
+           (Message() << default_value).GetString().c_str());
+    fflush(stdout);
+    return default_value;
+  }
+
+  return result;
+}
+
+// Reads and returns the string environment variable corresponding to
+// the given flag; if it's not set, returns default_value.
+std::string StringFromGTestEnv(const char* flag, const char* default_value) {
+#if defined(GTEST_GET_STRING_FROM_ENV_)
+  return GTEST_GET_STRING_FROM_ENV_(flag, default_value);
+#endif  // defined(GTEST_GET_STRING_FROM_ENV_)
+  const std::string env_var = FlagToEnvVar(flag);
+  const char* value = posix::GetEnv(env_var.c_str());
+  if (value != NULL) {
+    return value;
+  }
+
+  // As a special case for the 'output' flag, if GTEST_OUTPUT is not
+  // set, we look for XML_OUTPUT_FILE, which is set by the Bazel build
+  // system.  The value of XML_OUTPUT_FILE is a filename without the
+  // "xml:" prefix of GTEST_OUTPUT.
+  //
+  // The net priority order after flag processing is thus:
+  //   --gtest_output command line flag
+  //   GTEST_OUTPUT environment variable
+  //   XML_OUTPUT_FILE environment variable
+  //   'default_value'
+  if (strcmp(flag, "output") == 0) {
+    value = posix::GetEnv("XML_OUTPUT_FILE");
+    if (value != NULL) {
+      return std::string("xml:") + value;
+    }
+  }
+  return default_value;
+}
+
+}  // namespace internal
+}  // namespace testing
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Google Test - The Google C++ Testing Framework
+//
+// This file implements a universal value printer that can print a
+// value of any type T:
+//
+//   void ::testing::internal::UniversalPrinter<T>::Print(value, ostream_ptr);
+//
+// It uses the << operator when possible, and prints the bytes in the
+// object otherwise.  A user can override its behavior for a class
+// type Foo by defining either operator<<(::std::ostream&, const Foo&)
+// or void PrintTo(const Foo&, ::std::ostream*) in the namespace that
+// defines Foo.
+
+#include <ctype.h>
+#include <stdio.h>
+#include <cwchar>
+#include <ostream>  // NOLINT
+#include <string>
+
+namespace testing {
+
+namespace {
+
+using ::std::ostream;
+
+// Prints a segment of bytes in the given object.
+GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+void PrintByteSegmentInObjectTo(const unsigned char* obj_bytes, size_t start,
+                                size_t count, ostream* os) {
+  char text[5] = "";
+  for (size_t i = 0; i != count; i++) {
+    const size_t j = start + i;
+    if (i != 0) {
+      // Organizes the bytes into groups of 2 for easy parsing by
+      // human.
+      if ((j % 2) == 0)
+        *os << ' ';
+      else
+        *os << '-';
+    }
+    GTEST_SNPRINTF_(text, sizeof(text), "%02X", obj_bytes[j]);
+    *os << text;
+  }
+}
+
+// Prints the bytes in the given value to the given ostream.
+void PrintBytesInObjectToImpl(const unsigned char* obj_bytes, size_t count,
+                              ostream* os) {
+  // Tells the user how big the object is.
+  *os << count << "-byte object <";
+
+  const size_t kThreshold = 132;
+  const size_t kChunkSize = 64;
+  // If the object size is bigger than kThreshold, we'll have to omit
+  // some details by printing only the first and the last kChunkSize
+  // bytes.
+  // TODO(wan): let the user control the threshold using a flag.
+  if (count < kThreshold) {
+    PrintByteSegmentInObjectTo(obj_bytes, 0, count, os);
+  } else {
+    PrintByteSegmentInObjectTo(obj_bytes, 0, kChunkSize, os);
+    *os << " ... ";
+    // Rounds up to 2-byte boundary.
+    const size_t resume_pos = (count - kChunkSize + 1)/2*2;
+    PrintByteSegmentInObjectTo(obj_bytes, resume_pos, count - resume_pos, os);
+  }
+  *os << ">";
+}
+
+}  // namespace
+
+namespace internal2 {
+
+// Delegates to PrintBytesInObjectToImpl() to print the bytes in the
+// given object.  The delegation simplifies the implementation, which
+// uses the << operator and thus is easier done outside of the
+// ::testing::internal namespace, which contains a << operator that
+// sometimes conflicts with the one in STL.
+void PrintBytesInObjectTo(const unsigned char* obj_bytes, size_t count,
+                          ostream* os) {
+  PrintBytesInObjectToImpl(obj_bytes, count, os);
+}
+
+}  // namespace internal2
+
+namespace internal {
+
+// Depending on the value of a char (or wchar_t), we print it in one
+// of three formats:
+//   - as is if it's a printable ASCII (e.g. 'a', '2', ' '),
+//   - as a hexadecimal escape sequence (e.g. '\x7F'), or
+//   - as a special escape sequence (e.g. '\r', '\n').
+enum CharFormat {
+  kAsIs,
+  kHexEscape,
+  kSpecialEscape
+};
+
+// Returns true if c is a printable ASCII character.  We test the
+// value of c directly instead of calling isprint(), which is buggy on
+// Windows Mobile.
+inline bool IsPrintableAscii(wchar_t c) {
+  return 0x20 <= c && c <= 0x7E;
+}
+
+// Prints a wide or narrow char c as a character literal without the
+// quotes, escaping it when necessary; returns how c was formatted.
+// The template argument UnsignedChar is the unsigned version of Char,
+// which is the type of c.
+template <typename UnsignedChar, typename Char>
+static CharFormat PrintAsCharLiteralTo(Char c, ostream* os) {
+  switch (static_cast<wchar_t>(c)) {
+    case L'\0':
+      *os << "\\0";
+      break;
+    case L'\'':
+      *os << "\\'";
+      break;
+    case L'\\':
+      *os << "\\\\";
+      break;
+    case L'\a':
+      *os << "\\a";
+      break;
+    case L'\b':
+      *os << "\\b";
+      break;
+    case L'\f':
+      *os << "\\f";
+      break;
+    case L'\n':
+      *os << "\\n";
+      break;
+    case L'\r':
+      *os << "\\r";
+      break;
+    case L'\t':
+      *os << "\\t";
+      break;
+    case L'\v':
+      *os << "\\v";
+      break;
+    default:
+      if (IsPrintableAscii(c)) {
+        *os << static_cast<char>(c);
+        return kAsIs;
+      } else {
+        *os << "\\x" + String::FormatHexInt(static_cast<UnsignedChar>(c));
+        return kHexEscape;
+      }
+  }
+  return kSpecialEscape;
+}
+
+// Prints a wchar_t c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsStringLiteralTo(wchar_t c, ostream* os) {
+  switch (c) {
+    case L'\'':
+      *os << "'";
+      return kAsIs;
+    case L'"':
+      *os << "\\\"";
+      return kSpecialEscape;
+    default:
+      return PrintAsCharLiteralTo<wchar_t>(c, os);
+  }
+}
+
+// Prints a char c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsStringLiteralTo(char c, ostream* os) {
+  return PrintAsStringLiteralTo(
+      static_cast<wchar_t>(static_cast<unsigned char>(c)), os);
+}
+
+// Prints a wide or narrow character c and its code.  '\0' is printed
+// as "'\\0'", other unprintable characters are also properly escaped
+// using the standard C++ escape sequence.  The template argument
+// UnsignedChar is the unsigned version of Char, which is the type of c.
+template <typename UnsignedChar, typename Char>
+void PrintCharAndCodeTo(Char c, ostream* os) {
+  // First, print c as a literal in the most readable form we can find.
+  *os << ((sizeof(c) > 1) ? "L'" : "'");
+  const CharFormat format = PrintAsCharLiteralTo<UnsignedChar>(c, os);
+  *os << "'";
+
+  // To aid user debugging, we also print c's code in decimal, unless
+  // it's 0 (in which case c was printed as '\\0', making the code
+  // obvious).
+  if (c == 0)
+    return;
+  *os << " (" << static_cast<int>(c);
+
+  // For more convenience, we print c's code again in hexadecimal,
+  // unless c was already printed in the form '\x##' or the code is in
+  // [1, 9].
+  if (format == kHexEscape || (1 <= c && c <= 9)) {
+    // Do nothing.
+  } else {
+    *os << ", 0x" << String::FormatHexInt(static_cast<UnsignedChar>(c));
+  }
+  *os << ")";
+}
+
+void PrintTo(unsigned char c, ::std::ostream* os) {
+  PrintCharAndCodeTo<unsigned char>(c, os);
+}
+void PrintTo(signed char c, ::std::ostream* os) {
+  PrintCharAndCodeTo<unsigned char>(c, os);
+}
+
+// Prints a wchar_t as a symbol if it is printable or as its internal
+// code otherwise and also as its code.  L'\0' is printed as "L'\\0'".
+void PrintTo(wchar_t wc, ostream* os) {
+  PrintCharAndCodeTo<wchar_t>(wc, os);
+}
+
+// Prints the given array of characters to the ostream.  CharType must be either
+// char or wchar_t.
+// The array starts at begin, the length is len, it may include '\0' characters
+// and may not be NUL-terminated.
+template <typename CharType>
+GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+static void PrintCharsAsStringTo(
+    const CharType* begin, size_t len, ostream* os) {
+  const char* const kQuoteBegin = sizeof(CharType) == 1 ? "\"" : "L\"";
+  *os << kQuoteBegin;
+  bool is_previous_hex = false;
+  for (size_t index = 0; index < len; ++index) {
+    const CharType cur = begin[index];
+    if (is_previous_hex && IsXDigit(cur)) {
+      // Previous character is of '\x..' form and this character can be
+      // interpreted as another hexadecimal digit in its number. Break string to
+      // disambiguate.
+      *os << "\" " << kQuoteBegin;
+    }
+    is_previous_hex = PrintAsStringLiteralTo(cur, os) == kHexEscape;
+  }
+  *os << "\"";
+}
+
+// Prints a (const) char/wchar_t array of 'len' elements, starting at address
+// 'begin'.  CharType must be either char or wchar_t.
+template <typename CharType>
+GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+static void UniversalPrintCharArray(
+    const CharType* begin, size_t len, ostream* os) {
+  // The code
+  //   const char kFoo[] = "foo";
+  // generates an array of 4, not 3, elements, with the last one being '\0'.
+  //
+  // Therefore when printing a char array, we don't print the last element if
+  // it's '\0', such that the output matches the string literal as it's
+  // written in the source code.
+  if (len > 0 && begin[len - 1] == '\0') {
+    PrintCharsAsStringTo(begin, len - 1, os);
+    return;
+  }
+
+  // If, however, the last element in the array is not '\0', e.g.
+  //    const char kFoo[] = { 'f', 'o', 'o' };
+  // we must print the entire array.  We also print a message to indicate
+  // that the array is not NUL-terminated.
+  PrintCharsAsStringTo(begin, len, os);
+  *os << " (no terminating NUL)";
+}
+
+// Prints a (const) char array of 'len' elements, starting at address 'begin'.
+void UniversalPrintArray(const char* begin, size_t len, ostream* os) {
+  UniversalPrintCharArray(begin, len, os);
+}
+
+// Prints a (const) wchar_t array of 'len' elements, starting at address
+// 'begin'.
+void UniversalPrintArray(const wchar_t* begin, size_t len, ostream* os) {
+  UniversalPrintCharArray(begin, len, os);
+}
+
+// Prints the given C string to the ostream.
+void PrintTo(const char* s, ostream* os) {
+  if (s == NULL) {
+    *os << "NULL";
+  } else {
+    *os << ImplicitCast_<const void*>(s) << " pointing to ";
+    PrintCharsAsStringTo(s, strlen(s), os);
+  }
+}
+
+// MSVC compiler can be configured to define whar_t as a typedef
+// of unsigned short. Defining an overload for const wchar_t* in that case
+// would cause pointers to unsigned shorts be printed as wide strings,
+// possibly accessing more memory than intended and causing invalid
+// memory accesses. MSVC defines _NATIVE_WCHAR_T_DEFINED symbol when
+// wchar_t is implemented as a native type.
+#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED)
+// Prints the given wide C string to the ostream.
+void PrintTo(const wchar_t* s, ostream* os) {
+  if (s == NULL) {
+    *os << "NULL";
+  } else {
+    *os << ImplicitCast_<const void*>(s) << " pointing to ";
+    PrintCharsAsStringTo(s, std::wcslen(s), os);
+  }
+}
+#endif  // wchar_t is native
+
+// Prints a ::string object.
+#if GTEST_HAS_GLOBAL_STRING
+void PrintStringTo(const ::string& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+void PrintStringTo(const ::std::string& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+
+// Prints a ::wstring object.
+#if GTEST_HAS_GLOBAL_WSTRING
+void PrintWideStringTo(const ::wstring& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+#if GTEST_HAS_STD_WSTRING
+void PrintWideStringTo(const ::std::wstring& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif  // GTEST_HAS_STD_WSTRING
+
+}  // namespace internal
+
+}  // namespace testing
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+// The Google C++ Testing Framework (Google Test)
+
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick exists to
+// prevent the accidental inclusion of gtest-internal-inl.h in the
+// user's code.
+#define GTEST_IMPLEMENTATION_ 1
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+
+using internal::GetUnitTestImpl;
+
+// Gets the summary of the failure message by omitting the stack trace
+// in it.
+std::string TestPartResult::ExtractSummary(const char* message) {
+  const char* const stack_trace = strstr(message, internal::kStackTraceMarker);
+  return stack_trace == NULL ? message :
+      std::string(message, stack_trace);
+}
+
+// Prints a TestPartResult object.
+std::ostream& operator<<(std::ostream& os, const TestPartResult& result) {
+  return os
+      << result.file_name() << ":" << result.line_number() << ": "
+      << (result.type() == TestPartResult::kSuccess ? "Success" :
+          result.type() == TestPartResult::kFatalFailure ? "Fatal failure" :
+          "Non-fatal failure") << ":\n"
+      << result.message() << std::endl;
+}
+
+// Appends a TestPartResult to the array.
+void TestPartResultArray::Append(const TestPartResult& result) {
+  array_.push_back(result);
+}
+
+// Returns the TestPartResult at the given index (0-based).
+const TestPartResult& TestPartResultArray::GetTestPartResult(int index) const {
+  if (index < 0 || index >= size()) {
+    printf("\nInvalid index (%d) into TestPartResultArray.\n", index);
+    internal::posix::Abort();
+  }
+
+  return array_[index];
+}
+
+// Returns the number of TestPartResult objects in the array.
+int TestPartResultArray::size() const {
+  return static_cast<int>(array_.size());
+}
+
+namespace internal {
+
+HasNewFatalFailureHelper::HasNewFatalFailureHelper()
+    : has_new_fatal_failure_(false),
+      original_reporter_(GetUnitTestImpl()->
+                         GetTestPartResultReporterForCurrentThread()) {
+  GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(this);
+}
+
+HasNewFatalFailureHelper::~HasNewFatalFailureHelper() {
+  GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(
+      original_reporter_);
+}
+
+void HasNewFatalFailureHelper::ReportTestPartResult(
+    const TestPartResult& result) {
+  if (result.fatally_failed())
+    has_new_fatal_failure_ = true;
+  original_reporter_->ReportTestPartResult(result);
+}
+
+}  // namespace internal
+
+}  // namespace testing
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+
+namespace testing {
+namespace internal {
+
+#if GTEST_HAS_TYPED_TEST_P
+
+// Skips to the first non-space char in str. Returns an empty string if str
+// contains only whitespace characters.
+static const char* SkipSpaces(const char* str) {
+  while (IsSpace(*str))
+    str++;
+  return str;
+}
+
+static std::vector<std::string> SplitIntoTestNames(const char* src) {
+  std::vector<std::string> name_vec;
+  src = SkipSpaces(src);
+  for (; src != NULL; src = SkipComma(src)) {
+    name_vec.push_back(StripTrailingSpaces(GetPrefixUntilComma(src)));
+  }
+  return name_vec;
+}
+
+// Verifies that registered_tests match the test names in
+// registered_tests_; returns registered_tests if successful, or
+// aborts the program otherwise.
+const char* TypedTestCasePState::VerifyRegisteredTestNames(
+    const char* file, int line, const char* registered_tests) {
+  typedef RegisteredTestsMap::const_iterator RegisteredTestIter;
+  registered_ = true;
+
+  std::vector<std::string> name_vec = SplitIntoTestNames(registered_tests);
+
+  Message errors;
+
+  std::set<std::string> tests;
+  for (std::vector<std::string>::const_iterator name_it = name_vec.begin();
+       name_it != name_vec.end(); ++name_it) {
+    const std::string& name = *name_it;
+    if (tests.count(name) != 0) {
+      errors << "Test " << name << " is listed more than once.\n";
+      continue;
+    }
+
+    bool found = false;
+    for (RegisteredTestIter it = registered_tests_.begin();
+         it != registered_tests_.end();
+         ++it) {
+      if (name == it->first) {
+        found = true;
+        break;
+      }
+    }
+
+    if (found) {
+      tests.insert(name);
+    } else {
+      errors << "No test named " << name
+             << " can be found in this test case.\n";
+    }
+  }
+
+  for (RegisteredTestIter it = registered_tests_.begin();
+       it != registered_tests_.end();
+       ++it) {
+    if (tests.count(it->first) == 0) {
+      errors << "You forgot to list test " << it->first << ".\n";
+    }
+  }
+
+  const std::string& errors_str = errors.GetString();
+  if (errors_str != "") {
+    fprintf(stderr, "%s %s", FormatFileLocation(file, line).c_str(),
+            errors_str.c_str());
+    fflush(stderr);
+    posix::Abort();
+  }
+
+  return registered_tests;
+}
+
+#endif  // GTEST_HAS_TYPED_TEST_P
+
+}  // namespace internal
+}  // namespace testing
diff --git a/memkind-1.10.0/test/gtest_fused/gtest/gtest.h b/memkind-1.10.0/test/gtest_fused/gtest/gtest.h
new file mode 100644
index 0000000..55c43d6
--- /dev/null
+++ b/memkind-1.10.0/test/gtest_fused/gtest/gtest.h
@@ -0,0 +1,22635 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the public API for Google Test.  It should be
+// included by any test program that uses Google Test.
+//
+// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
+// leave some internal implementation details in this header file.
+// They are clearly marked by comments like this:
+//
+//   // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+//
+// Such code is NOT meant to be used by a user directly, and is subject
+// to CHANGE WITHOUT NOTICE.  Therefore DO NOT DEPEND ON IT in a user
+// program!
+//
+// Acknowledgment: Google Test borrowed the idea of automatic test
+// registration from Barthelemy Dagenais' (barthelemy@prologique.com)
+// easyUnit framework.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_H_
+
+#include <limits>
+#include <ostream>
+#include <vector>
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file declares functions and macros used internally by
+// Google Test.  They are subject to change without notice.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan)
+//
+// Low-level types and utilities for porting Google Test to various
+// platforms.  All macros ending with _ and symbols defined in an
+// internal namespace are subject to change without notice.  Code
+// outside Google Test MUST NOT USE THEM DIRECTLY.  Macros that don't
+// end with _ are part of Google Test's public API and can be used by
+// code outside Google Test.
+//
+// This file is fundamental to Google Test.  All other Google Test source
+// files are expected to #include this.  Therefore, it cannot #include
+// any other Google Test header.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+
+// Environment-describing macros
+// -----------------------------
+//
+// Google Test can be used in many different environments.  Macros in
+// this section tell Google Test what kind of environment it is being
+// used in, such that Google Test can provide environment-specific
+// features and implementations.
+//
+// Google Test tries to automatically detect the properties of its
+// environment, so users usually don't need to worry about these
+// macros.  However, the automatic detection is not perfect.
+// Sometimes it's necessary for a user to define some of the following
+// macros in the build script to override Google Test's decisions.
+//
+// If the user doesn't define a macro in the list, Google Test will
+// provide a default definition.  After this header is #included, all
+// macros in this list will be defined to either 1 or 0.
+//
+// Notes to maintainers:
+//   - Each macro here is a user-tweakable knob; do not grow the list
+//     lightly.
+//   - Use #if to key off these macros.  Don't use #ifdef or "#if
+//     defined(...)", which will not work as these macros are ALWAYS
+//     defined.
+//
+//   GTEST_HAS_CLONE          - Define it to 1/0 to indicate that clone(2)
+//                              is/isn't available.
+//   GTEST_HAS_EXCEPTIONS     - Define it to 1/0 to indicate that exceptions
+//                              are enabled.
+//   GTEST_HAS_GLOBAL_STRING  - Define it to 1/0 to indicate that ::string
+//                              is/isn't available (some systems define
+//                              ::string, which is different to std::string).
+//   GTEST_HAS_GLOBAL_WSTRING - Define it to 1/0 to indicate that ::string
+//                              is/isn't available (some systems define
+//                              ::wstring, which is different to std::wstring).
+//   GTEST_HAS_POSIX_RE       - Define it to 1/0 to indicate that POSIX regular
+//                              expressions are/aren't available.
+//   GTEST_HAS_PTHREAD        - Define it to 1/0 to indicate that <pthread.h>
+//                              is/isn't available.
+//   GTEST_HAS_RTTI           - Define it to 1/0 to indicate that RTTI is/isn't
+//                              enabled.
+//   GTEST_HAS_STD_WSTRING    - Define it to 1/0 to indicate that
+//                              std::wstring does/doesn't work (Google Test can
+//                              be used where std::wstring is unavailable).
+//   GTEST_HAS_TR1_TUPLE      - Define it to 1/0 to indicate tr1::tuple
+//                              is/isn't available.
+//   GTEST_HAS_SEH            - Define it to 1/0 to indicate whether the
+//                              compiler supports Microsoft's "Structured
+//                              Exception Handling".
+//   GTEST_HAS_STREAM_REDIRECTION
+//                            - Define it to 1/0 to indicate whether the
+//                              platform supports I/O stream redirection using
+//                              dup() and dup2().
+//   GTEST_USE_OWN_TR1_TUPLE  - Define it to 1/0 to indicate whether Google
+//                              Test's own tr1 tuple implementation should be
+//                              used.  Unused when the user sets
+//                              GTEST_HAS_TR1_TUPLE to 0.
+//   GTEST_LANG_CXX11         - Define it to 1/0 to indicate that Google Test
+//                              is building in C++11/C++98 mode.
+//   GTEST_LINKED_AS_SHARED_LIBRARY
+//                            - Define to 1 when compiling tests that use
+//                              Google Test as a shared library (known as
+//                              DLL on Windows).
+//   GTEST_CREATE_SHARED_LIBRARY
+//                            - Define to 1 when compiling Google Test itself
+//                              as a shared library.
+
+// Platform-indicating macros
+// --------------------------
+//
+// Macros indicating the platform on which Google Test is being used
+// (a macro is defined to 1 if compiled on the given platform;
+// otherwise UNDEFINED -- it's never defined to 0.).  Google Test
+// defines these macros automatically.  Code outside Google Test MUST
+// NOT define them.
+//
+//   GTEST_OS_AIX      - IBM AIX
+//   GTEST_OS_CYGWIN   - Cygwin
+//   GTEST_OS_FREEBSD  - FreeBSD
+//   GTEST_OS_HPUX     - HP-UX
+//   GTEST_OS_LINUX    - Linux
+//     GTEST_OS_LINUX_ANDROID - Google Android
+//   GTEST_OS_MAC      - Mac OS X
+//     GTEST_OS_IOS    - iOS
+//   GTEST_OS_NACL     - Google Native Client (NaCl)
+//   GTEST_OS_NETBSD   - NetBSD
+//   GTEST_OS_OPENBSD  - OpenBSD
+//   GTEST_OS_QNX      - QNX
+//   GTEST_OS_SOLARIS  - Sun Solaris
+//   GTEST_OS_SYMBIAN  - Symbian
+//   GTEST_OS_WINDOWS  - Windows (Desktop, MinGW, or Mobile)
+//     GTEST_OS_WINDOWS_DESKTOP  - Windows Desktop
+//     GTEST_OS_WINDOWS_MINGW    - MinGW
+//     GTEST_OS_WINDOWS_MOBILE   - Windows Mobile
+//     GTEST_OS_WINDOWS_PHONE    - Windows Phone
+//     GTEST_OS_WINDOWS_RT       - Windows Store App/WinRT
+//   GTEST_OS_ZOS      - z/OS
+//
+// Among the platforms, Cygwin, Linux, Max OS X, and Windows have the
+// most stable support.  Since core members of the Google Test project
+// don't have access to other platforms, support for them may be less
+// stable.  If you notice any problems on your platform, please notify
+// googletestframework@googlegroups.com (patches for fixing them are
+// even more welcome!).
+//
+// It is possible that none of the GTEST_OS_* macros are defined.
+
+// Feature-indicating macros
+// -------------------------
+//
+// Macros indicating which Google Test features are available (a macro
+// is defined to 1 if the corresponding feature is supported;
+// otherwise UNDEFINED -- it's never defined to 0.).  Google Test
+// defines these macros automatically.  Code outside Google Test MUST
+// NOT define them.
+//
+// These macros are public so that portable tests can be written.
+// Such tests typically surround code using a feature with an #if
+// which controls that code.  For example:
+//
+// #if GTEST_HAS_DEATH_TEST
+//   EXPECT_DEATH(DoSomethingDeadly());
+// #endif
+//
+//   GTEST_HAS_COMBINE      - the Combine() function (for value-parameterized
+//                            tests)
+//   GTEST_HAS_DEATH_TEST   - death tests
+//   GTEST_HAS_PARAM_TEST   - value-parameterized tests
+//   GTEST_HAS_TYPED_TEST   - typed tests
+//   GTEST_HAS_TYPED_TEST_P - type-parameterized tests
+//   GTEST_IS_THREADSAFE    - Google Test is thread-safe.
+//   GTEST_USES_POSIX_RE    - enhanced POSIX regex is used. Do not confuse with
+//                            GTEST_HAS_POSIX_RE (see above) which users can
+//                            define themselves.
+//   GTEST_USES_SIMPLE_RE   - our own simple regex is used;
+//                            the above two are mutually exclusive.
+//   GTEST_CAN_COMPARE_NULL - accepts untyped NULL in EXPECT_EQ().
+
+// Misc public macros
+// ------------------
+//
+//   GTEST_FLAG(flag_name)  - references the variable corresponding to
+//                            the given Google Test flag.
+
+// Internal utilities
+// ------------------
+//
+// The following macros and utilities are for Google Test's INTERNAL
+// use only.  Code outside Google Test MUST NOT USE THEM DIRECTLY.
+//
+// Macros for basic C++ coding:
+//   GTEST_AMBIGUOUS_ELSE_BLOCKER_ - for disabling a gcc warning.
+//   GTEST_ATTRIBUTE_UNUSED_  - declares that a class' instances or a
+//                              variable don't have to be used.
+//   GTEST_DISALLOW_ASSIGN_   - disables operator=.
+//   GTEST_DISALLOW_COPY_AND_ASSIGN_ - disables copy ctor and operator=.
+//   GTEST_MUST_USE_RESULT_   - declares that a function's result must be used.
+//   GTEST_INTENTIONAL_CONST_COND_PUSH_ - start code section where MSVC C4127 is
+//                                        suppressed (constant conditional).
+//   GTEST_INTENTIONAL_CONST_COND_POP_  - finish code section where MSVC C4127
+//                                        is suppressed.
+//
+// C++11 feature wrappers:
+//
+//   testing::internal::move  - portability wrapper for std::move.
+//
+// Synchronization:
+//   Mutex, MutexLock, ThreadLocal, GetThreadCount()
+//                            - synchronization primitives.
+//
+// Template meta programming:
+//   is_pointer     - as in TR1; needed on Symbian and IBM XL C/C++ only.
+//   IteratorTraits - partial implementation of std::iterator_traits, which
+//                    is not available in libCstd when compiled with Sun C++.
+//
+// Smart pointers:
+//   scoped_ptr     - as in TR2.
+//
+// Regular expressions:
+//   RE             - a simple regular expression class using the POSIX
+//                    Extended Regular Expression syntax on UNIX-like
+//                    platforms, or a reduced regular exception syntax on
+//                    other platforms, including Windows.
+//
+// Logging:
+//   GTEST_LOG_()   - logs messages at the specified severity level.
+//   LogToStderr()  - directs all log messages to stderr.
+//   FlushInfoLog() - flushes informational log messages.
+//
+// Stdout and stderr capturing:
+//   CaptureStdout()     - starts capturing stdout.
+//   GetCapturedStdout() - stops capturing stdout and returns the captured
+//                         string.
+//   CaptureStderr()     - starts capturing stderr.
+//   GetCapturedStderr() - stops capturing stderr and returns the captured
+//                         string.
+//
+// Integer types:
+//   TypeWithSize   - maps an integer to a int type.
+//   Int32, UInt32, Int64, UInt64, TimeInMillis
+//                  - integers of known sizes.
+//   BiggestInt     - the biggest signed integer type.
+//
+// Command-line utilities:
+//   GTEST_DECLARE_*()  - declares a flag.
+//   GTEST_DEFINE_*()   - defines a flag.
+//   GetInjectableArgvs() - returns the command line as a vector of strings.
+//
+// Environment variable utilities:
+//   GetEnv()             - gets the value of an environment variable.
+//   BoolFromGTestEnv()   - parses a bool environment variable.
+//   Int32FromGTestEnv()  - parses an Int32 environment variable.
+//   StringFromGTestEnv() - parses a string environment variable.
+
+#include <ctype.h>   // for isspace, etc
+#include <stddef.h>  // for ptrdiff_t
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#ifndef _WIN32_WCE
+# include <sys/types.h>
+# include <sys/stat.h>
+#endif  // !_WIN32_WCE
+
+#if defined __APPLE__
+# include <AvailabilityMacros.h>
+# include <TargetConditionals.h>
+#endif
+
+#include <algorithm>  // NOLINT
+#include <iostream>  // NOLINT
+#include <sstream>  // NOLINT
+#include <string>  // NOLINT
+#include <utility>
+#include <vector>  // NOLINT
+
+// Copyright 2015, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the GTEST_OS_* macro.
+// It is separate from gtest-port.h so that custom/gtest-port.h can include it.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_
+
+// Determines the platform on which Google Test is compiled.
+#ifdef __CYGWIN__
+# define GTEST_OS_CYGWIN 1
+#elif defined __SYMBIAN32__
+# define GTEST_OS_SYMBIAN 1
+#elif defined _WIN32
+# define GTEST_OS_WINDOWS 1
+# ifdef _WIN32_WCE
+#  define GTEST_OS_WINDOWS_MOBILE 1
+# elif defined(__MINGW__) || defined(__MINGW32__)
+#  define GTEST_OS_WINDOWS_MINGW 1
+# elif defined(WINAPI_FAMILY)
+#  include <winapifamily.h>
+#  if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
+#   define GTEST_OS_WINDOWS_DESKTOP 1
+#  elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_PHONE_APP)
+#   define GTEST_OS_WINDOWS_PHONE 1
+#  elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_APP)
+#   define GTEST_OS_WINDOWS_RT 1
+#  else
+// WINAPI_FAMILY defined but no known partition matched.
+// Default to desktop.
+#   define GTEST_OS_WINDOWS_DESKTOP 1
+#  endif
+# else
+#  define GTEST_OS_WINDOWS_DESKTOP 1
+# endif  // _WIN32_WCE
+#elif defined __APPLE__
+# define GTEST_OS_MAC 1
+# if TARGET_OS_IPHONE
+#  define GTEST_OS_IOS 1
+# endif
+#elif defined __FreeBSD__
+# define GTEST_OS_FREEBSD 1
+#elif defined __linux__
+# define GTEST_OS_LINUX 1
+# if defined __ANDROID__
+#  define GTEST_OS_LINUX_ANDROID 1
+# endif
+#elif defined __MVS__
+# define GTEST_OS_ZOS 1
+#elif defined(__sun) && defined(__SVR4)
+# define GTEST_OS_SOLARIS 1
+#elif defined(_AIX)
+# define GTEST_OS_AIX 1
+#elif defined(__hpux)
+# define GTEST_OS_HPUX 1
+#elif defined __native_client__
+# define GTEST_OS_NACL 1
+#elif defined __NetBSD__
+# define GTEST_OS_NETBSD 1
+#elif defined __OpenBSD__
+# define GTEST_OS_OPENBSD 1
+#elif defined __QNX__
+# define GTEST_OS_QNX 1
+#endif  // __CYGWIN__
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_
+// Copyright 2015, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Injection point for custom user configurations.
+// The following macros can be defined:
+//
+//   Flag related macros:
+//     GTEST_FLAG(flag_name)
+//     GTEST_USE_OWN_FLAGFILE_FLAG_  - Define to 0 when the system provides its
+//                                     own flagfile flag parsing.
+//     GTEST_DECLARE_bool_(name)
+//     GTEST_DECLARE_int32_(name)
+//     GTEST_DECLARE_string_(name)
+//     GTEST_DEFINE_bool_(name, default_val, doc)
+//     GTEST_DEFINE_int32_(name, default_val, doc)
+//     GTEST_DEFINE_string_(name, default_val, doc)
+//
+//   Test filtering:
+//     GTEST_TEST_FILTER_ENV_VAR_ - The name of an environment variable that
+//                                  will be used if --GTEST_FLAG(test_filter)
+//                                  is not provided.
+//
+//   Logging:
+//     GTEST_LOG_(severity)
+//     GTEST_CHECK_(condition)
+//     Functions LogToStderr() and FlushInfoLog() have to be provided too.
+//
+//   Threading:
+//     GTEST_HAS_NOTIFICATION_ - Enabled if Notification is already provided.
+//     GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ - Enabled if Mutex and ThreadLocal are
+//                                         already provided.
+//     Must also provide GTEST_DECLARE_STATIC_MUTEX_(mutex) and
+//     GTEST_DEFINE_STATIC_MUTEX_(mutex)
+//
+//     GTEST_EXCLUSIVE_LOCK_REQUIRED_(locks)
+//     GTEST_LOCK_EXCLUDED_(locks)
+//
+//   Exporting API symbols:
+//     GTEST_API_ - Specifier for exported symbols.
+//
+// ** Custom implementation starts here **
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_
+
+#if !defined(GTEST_DEV_EMAIL_)
+# define GTEST_DEV_EMAIL_ "googletestframework@@googlegroups.com"
+# define GTEST_FLAG_PREFIX_ "gtest_"
+# define GTEST_FLAG_PREFIX_DASH_ "gtest-"
+# define GTEST_FLAG_PREFIX_UPPER_ "GTEST_"
+# define GTEST_NAME_ "Google Test"
+# define GTEST_PROJECT_URL_ "https://github.com/google/googletest/"
+#endif  // !defined(GTEST_DEV_EMAIL_)
+
+#if !defined(GTEST_INIT_GOOGLE_TEST_NAME_)
+# define GTEST_INIT_GOOGLE_TEST_NAME_ "testing::InitGoogleTest"
+#endif  // !defined(GTEST_INIT_GOOGLE_TEST_NAME_)
+
+// Determines the version of gcc that is used to compile this.
+#ifdef __GNUC__
+// 40302 means version 4.3.2.
+# define GTEST_GCC_VER_ \
+    (__GNUC__*10000 + __GNUC_MINOR__*100 + __GNUC_PATCHLEVEL__)
+#endif  // __GNUC__
+
+// Macros for disabling Microsoft Visual C++ warnings.
+//
+//   GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 4385)
+//   /* code that triggers warnings C4800 and C4385 */
+//   GTEST_DISABLE_MSC_WARNINGS_POP_()
+#if _MSC_VER >= 1500
+# define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings) \
+    __pragma(warning(push))                        \
+    __pragma(warning(disable: warnings))
+# define GTEST_DISABLE_MSC_WARNINGS_POP_()          \
+    __pragma(warning(pop))
+#else
+// Older versions of MSVC don't have __pragma.
+# define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings)
+# define GTEST_DISABLE_MSC_WARNINGS_POP_()
+#endif
+
+#ifndef GTEST_LANG_CXX11
+// gcc and clang define __GXX_EXPERIMENTAL_CXX0X__ when
+// -std={c,gnu}++{0x,11} is passed.  The C++11 standard specifies a
+// value for __cplusplus, and recent versions of clang, gcc, and
+// probably other compilers set that too in C++11 mode.
+# if __GXX_EXPERIMENTAL_CXX0X__ || __cplusplus >= 201103L
+// Compiling in at least C++11 mode.
+#  define GTEST_LANG_CXX11 1
+# else
+#  define GTEST_LANG_CXX11 0
+# endif
+#endif
+
+// Distinct from C++11 language support, some environments don't provide
+// proper C++11 library support. Notably, it's possible to build in
+// C++11 mode when targeting Mac OS X 10.6, which has an old libstdc++
+// with no C++11 support.
+//
+// libstdc++ has sufficient C++11 support as of GCC 4.6.0, __GLIBCXX__
+// 20110325, but maintenance releases in the 4.4 and 4.5 series followed
+// this date, so check for those versions by their date stamps.
+// https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html#abi.versioning
+#if GTEST_LANG_CXX11 && \
+    (!defined(__GLIBCXX__) || ( \
+        __GLIBCXX__ >= 20110325ul &&  /* GCC >= 4.6.0 */ \
+        /* Blacklist of patch releases of older branches: */ \
+        __GLIBCXX__ != 20110416ul &&  /* GCC 4.4.6 */ \
+        __GLIBCXX__ != 20120313ul &&  /* GCC 4.4.7 */ \
+        __GLIBCXX__ != 20110428ul &&  /* GCC 4.5.3 */ \
+        __GLIBCXX__ != 20120702ul))   /* GCC 4.5.4 */
+# define GTEST_STDLIB_CXX11 1
+#endif
+
+// Only use C++11 library features if the library provides them.
+#if GTEST_STDLIB_CXX11
+# define GTEST_HAS_STD_BEGIN_AND_END_ 1
+# define GTEST_HAS_STD_FORWARD_LIST_ 1
+# define GTEST_HAS_STD_FUNCTION_ 1
+# define GTEST_HAS_STD_INITIALIZER_LIST_ 1
+# define GTEST_HAS_STD_MOVE_ 1
+# define GTEST_HAS_STD_SHARED_PTR_ 1
+# define GTEST_HAS_STD_TYPE_TRAITS_ 1
+# define GTEST_HAS_STD_UNIQUE_PTR_ 1
+#endif
+
+// C++11 specifies that <tuple> provides std::tuple.
+// Some platforms still might not have it, however.
+#if GTEST_LANG_CXX11
+# define GTEST_HAS_STD_TUPLE_ 1
+# if defined(__clang__)
+// Inspired by http://clang.llvm.org/docs/LanguageExtensions.html#__has_include
+#  if defined(__has_include) && !__has_include(<tuple>)
+#   undef GTEST_HAS_STD_TUPLE_
+#  endif
+# elif defined(_MSC_VER)
+// Inspired by boost/config/stdlib/dinkumware.hpp
+#  if defined(_CPPLIB_VER) && _CPPLIB_VER < 520
+#   undef GTEST_HAS_STD_TUPLE_
+#  endif
+# elif defined(__GLIBCXX__)
+// Inspired by boost/config/stdlib/libstdcpp3.hpp,
+// http://gcc.gnu.org/gcc-4.2/changes.html and
+// http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#manual.intro.status.standard.200x
+#  if __GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 2)
+#   undef GTEST_HAS_STD_TUPLE_
+#  endif
+# endif
+#endif
+
+// Brings in definitions for functions used in the testing::internal::posix
+// namespace (read, write, close, chdir, isatty, stat). We do not currently
+// use them on Windows Mobile.
+#if GTEST_OS_WINDOWS
+# if !GTEST_OS_WINDOWS_MOBILE
+#  include <direct.h>
+#  include <io.h>
+# endif
+// In order to avoid having to include <windows.h>, use forward declaration
+#if GTEST_OS_WINDOWS_MINGW && !defined(__MINGW64_VERSION_MAJOR)
+// MinGW defined _CRITICAL_SECTION and _RTL_CRITICAL_SECTION as two
+// separate (equivalent) structs, instead of using typedef
+typedef struct _CRITICAL_SECTION GTEST_CRITICAL_SECTION;
+#else
+// Assume CRITICAL_SECTION is a typedef of _RTL_CRITICAL_SECTION.
+// This assumption is verified by
+// WindowsTypesTest.CRITICAL_SECTIONIs_RTL_CRITICAL_SECTION.
+typedef struct _RTL_CRITICAL_SECTION GTEST_CRITICAL_SECTION;
+#endif
+#else
+// This assumes that non-Windows OSes provide unistd.h. For OSes where this
+// is not the case, we need to include headers that provide the functions
+// mentioned above.
+# include <unistd.h>
+# include <strings.h>
+#endif  // GTEST_OS_WINDOWS
+
+#if GTEST_OS_LINUX_ANDROID
+// Used to define __ANDROID_API__ matching the target NDK API level.
+#  include <android/api-level.h>  // NOLINT
+#endif
+
+// Defines this to true iff Google Test can use POSIX regular expressions.
+#ifndef GTEST_HAS_POSIX_RE
+# if GTEST_OS_LINUX_ANDROID
+// On Android, <regex.h> is only available starting with Gingerbread.
+#  define GTEST_HAS_POSIX_RE (__ANDROID_API__ >= 9)
+# else
+#  define GTEST_HAS_POSIX_RE (!GTEST_OS_WINDOWS)
+# endif
+#endif
+
+#if GTEST_USES_PCRE
+// The appropriate headers have already been included.
+
+#elif GTEST_HAS_POSIX_RE
+
+// On some platforms, <regex.h> needs someone to define size_t, and
+// won't compile otherwise.  We can #include it here as we already
+// included <stdlib.h>, which is guaranteed to define size_t through
+// <stddef.h>.
+# include <regex.h>  // NOLINT
+
+# define GTEST_USES_POSIX_RE 1
+
+#elif GTEST_OS_WINDOWS
+
+// <regex.h> is not available on Windows.  Use our own simple regex
+// implementation instead.
+# define GTEST_USES_SIMPLE_RE 1
+
+#else
+
+// <regex.h> may not be available on this platform.  Use our own
+// simple regex implementation instead.
+# define GTEST_USES_SIMPLE_RE 1
+
+#endif  // GTEST_USES_PCRE
+
+#ifndef GTEST_HAS_EXCEPTIONS
+// The user didn't tell us whether exceptions are enabled, so we need
+// to figure it out.
+# if defined(_MSC_VER) || defined(__BORLANDC__)
+// MSVC's and C++Builder's implementations of the STL use the _HAS_EXCEPTIONS
+// macro to enable exceptions, so we'll do the same.
+// Assumes that exceptions are enabled by default.
+#  ifndef _HAS_EXCEPTIONS
+#   define _HAS_EXCEPTIONS 1
+#  endif  // _HAS_EXCEPTIONS
+#  define GTEST_HAS_EXCEPTIONS _HAS_EXCEPTIONS
+# elif defined(__clang__)
+// clang defines __EXCEPTIONS iff exceptions are enabled before clang 220714,
+// but iff cleanups are enabled after that. In Obj-C++ files, there can be
+// cleanups for ObjC exceptions which also need cleanups, even if C++ exceptions
+// are disabled. clang has __has_feature(cxx_exceptions) which checks for C++
+// exceptions starting at clang r206352, but which checked for cleanups prior to
+// that. To reliably check for C++ exception availability with clang, check for
+// __EXCEPTIONS && __has_feature(cxx_exceptions).
+#  define GTEST_HAS_EXCEPTIONS (__EXCEPTIONS && __has_feature(cxx_exceptions))
+# elif defined(__GNUC__) && __EXCEPTIONS
+// gcc defines __EXCEPTIONS to 1 iff exceptions are enabled.
+#  define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__SUNPRO_CC)
+// Sun Pro CC supports exceptions.  However, there is no compile-time way of
+// detecting whether they are enabled or not.  Therefore, we assume that
+// they are enabled unless the user tells us otherwise.
+#  define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__IBMCPP__) && __EXCEPTIONS
+// xlC defines __EXCEPTIONS to 1 iff exceptions are enabled.
+#  define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__HP_aCC)
+// Exception handling is in effect by default in HP aCC compiler. It has to
+// be turned of by +noeh compiler option if desired.
+#  define GTEST_HAS_EXCEPTIONS 1
+# else
+// For other compilers, we assume exceptions are disabled to be
+// conservative.
+#  define GTEST_HAS_EXCEPTIONS 0
+# endif  // defined(_MSC_VER) || defined(__BORLANDC__)
+#endif  // GTEST_HAS_EXCEPTIONS
+
+#if !defined(GTEST_HAS_STD_STRING)
+// Even though we don't use this macro any longer, we keep it in case
+// some clients still depend on it.
+# define GTEST_HAS_STD_STRING 1
+#elif !GTEST_HAS_STD_STRING
+// The user told us that ::std::string isn't available.
+# error "Google Test cannot be used where ::std::string isn't available."
+#endif  // !defined(GTEST_HAS_STD_STRING)
+
+#ifndef GTEST_HAS_GLOBAL_STRING
+// The user didn't tell us whether ::string is available, so we need
+// to figure it out.
+
+# define GTEST_HAS_GLOBAL_STRING 0
+
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+#ifndef GTEST_HAS_STD_WSTRING
+// The user didn't tell us whether ::std::wstring is available, so we need
+// to figure it out.
+// TODO(wan@google.com): uses autoconf to detect whether ::std::wstring
+//   is available.
+
+// Cygwin 1.7 and below doesn't support ::std::wstring.
+// Solaris' libc++ doesn't support it either.  Android has
+// no support for it at least as recent as Froyo (2.2).
+# define GTEST_HAS_STD_WSTRING \
+    (!(GTEST_OS_LINUX_ANDROID || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS))
+
+#endif  // GTEST_HAS_STD_WSTRING
+
+#ifndef GTEST_HAS_GLOBAL_WSTRING
+// The user didn't tell us whether ::wstring is available, so we need
+// to figure it out.
+# define GTEST_HAS_GLOBAL_WSTRING \
+    (GTEST_HAS_STD_WSTRING && GTEST_HAS_GLOBAL_STRING)
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+// Determines whether RTTI is available.
+#ifndef GTEST_HAS_RTTI
+// The user didn't tell us whether RTTI is enabled, so we need to
+// figure it out.
+
+# ifdef _MSC_VER
+
+#  ifdef _CPPRTTI  // MSVC defines this macro iff RTTI is enabled.
+#   define GTEST_HAS_RTTI 1
+#  else
+#   define GTEST_HAS_RTTI 0
+#  endif
+
+// Starting with version 4.3.2, gcc defines __GXX_RTTI iff RTTI is enabled.
+# elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40302)
+
+#  ifdef __GXX_RTTI
+// When building against STLport with the Android NDK and with
+// -frtti -fno-exceptions, the build fails at link time with undefined
+// references to __cxa_bad_typeid. Note sure if STL or toolchain bug,
+// so disable RTTI when detected.
+#   if GTEST_OS_LINUX_ANDROID && defined(_STLPORT_MAJOR) && \
+       !defined(__EXCEPTIONS)
+#    define GTEST_HAS_RTTI 0
+#   else
+#    define GTEST_HAS_RTTI 1
+#   endif  // GTEST_OS_LINUX_ANDROID && __STLPORT_MAJOR && !__EXCEPTIONS
+#  else
+#   define GTEST_HAS_RTTI 0
+#  endif  // __GXX_RTTI
+
+// Clang defines __GXX_RTTI starting with version 3.0, but its manual recommends
+// using has_feature instead. has_feature(cxx_rtti) is supported since 2.7, the
+// first version with C++ support.
+# elif defined(__clang__)
+
+#  define GTEST_HAS_RTTI __has_feature(cxx_rtti)
+
+// Starting with version 9.0 IBM Visual Age defines __RTTI_ALL__ to 1 if
+// both the typeid and dynamic_cast features are present.
+# elif defined(__IBMCPP__) && (__IBMCPP__ >= 900)
+
+#  ifdef __RTTI_ALL__
+#   define GTEST_HAS_RTTI 1
+#  else
+#   define GTEST_HAS_RTTI 0
+#  endif
+
+# else
+
+// For all other compilers, we assume RTTI is enabled.
+#  define GTEST_HAS_RTTI 1
+
+# endif  // _MSC_VER
+
+#endif  // GTEST_HAS_RTTI
+
+// It's this header's responsibility to #include <typeinfo> when RTTI
+// is enabled.
+#if GTEST_HAS_RTTI
+# include <typeinfo>
+#endif
+
+// Determines whether Google Test can use the pthreads library.
+#ifndef GTEST_HAS_PTHREAD
+// The user didn't tell us explicitly, so we make reasonable assumptions about
+// which platforms have pthreads support.
+//
+// To disable threading support in Google Test, add -DGTEST_HAS_PTHREAD=0
+// to your compiler flags.
+# define GTEST_HAS_PTHREAD (GTEST_OS_LINUX || GTEST_OS_MAC || GTEST_OS_HPUX \
+    || GTEST_OS_QNX || GTEST_OS_FREEBSD || GTEST_OS_NACL || GTEST_OS_NETBSD)
+#endif  // GTEST_HAS_PTHREAD
+
+#if GTEST_HAS_PTHREAD
+// gtest-port.h guarantees to #include <pthread.h> when GTEST_HAS_PTHREAD is
+// true.
+# include <pthread.h>  // NOLINT
+
+// For timespec and nanosleep, used below.
+# include <time.h>  // NOLINT
+#endif
+
+// Determines if hash_map/hash_set are available.
+// Only used for testing against those containers.
+#if !defined(GTEST_HAS_HASH_MAP_)
+# if defined(_MSC_VER) && (_MSC_VER < 1900)
+#  define GTEST_HAS_HASH_MAP_ 1  // Indicates that hash_map is available.
+#  define GTEST_HAS_HASH_SET_ 1  // Indicates that hash_set is available.
+# endif  // _MSC_VER
+#endif  // !defined(GTEST_HAS_HASH_MAP_)
+
+// Determines whether Google Test can use tr1/tuple.  You can define
+// this macro to 0 to prevent Google Test from using tuple (any
+// feature depending on tuple with be disabled in this mode).
+#ifndef GTEST_HAS_TR1_TUPLE
+# if GTEST_OS_LINUX_ANDROID && defined(_STLPORT_MAJOR)
+// STLport, provided with the Android NDK, has neither <tr1/tuple> or <tuple>.
+#  define GTEST_HAS_TR1_TUPLE 0
+# else
+// The user didn't tell us not to do it, so we assume it's OK.
+#  define GTEST_HAS_TR1_TUPLE 1
+# endif
+#endif  // GTEST_HAS_TR1_TUPLE
+
+// Determines whether Google Test's own tr1 tuple implementation
+// should be used.
+#ifndef GTEST_USE_OWN_TR1_TUPLE
+// The user didn't tell us, so we need to figure it out.
+
+// We use our own TR1 tuple if we aren't sure the user has an
+// implementation of it already.  At this time, libstdc++ 4.0.0+ and
+// MSVC 2010 are the only mainstream standard libraries that come
+// with a TR1 tuple implementation.  NVIDIA's CUDA NVCC compiler
+// pretends to be GCC by defining __GNUC__ and friends, but cannot
+// compile GCC's tuple implementation.  MSVC 2008 (9.0) provides TR1
+// tuple in a 323 MB Feature Pack download, which we cannot assume the
+// user has.  QNX's QCC compiler is a modified GCC but it doesn't
+// support TR1 tuple.  libc++ only provides std::tuple, in C++11 mode,
+// and it can be used with some compilers that define __GNUC__.
+# if (defined(__GNUC__) && !defined(__CUDACC__) && (GTEST_GCC_VER_ >= 40000) \
+      && !GTEST_OS_QNX && !defined(_LIBCPP_VERSION)) || _MSC_VER >= 1600
+#  define GTEST_ENV_HAS_TR1_TUPLE_ 1
+# endif
+
+// C++11 specifies that <tuple> provides std::tuple. Use that if gtest is used
+// in C++11 mode and libstdc++ isn't very old (binaries targeting OS X 10.6
+// can build with clang but need to use gcc4.2's libstdc++).
+# if GTEST_LANG_CXX11 && (!defined(__GLIBCXX__) || __GLIBCXX__ > 20110325)
+#  define GTEST_ENV_HAS_STD_TUPLE_ 1
+# endif
+
+# if GTEST_ENV_HAS_TR1_TUPLE_ || GTEST_ENV_HAS_STD_TUPLE_
+#  define GTEST_USE_OWN_TR1_TUPLE 0
+# else
+#  define GTEST_USE_OWN_TR1_TUPLE 1
+# endif
+
+#endif  // GTEST_USE_OWN_TR1_TUPLE
+
+// To avoid conditional compilation everywhere, we make it
+// gtest-port.h's responsibility to #include the header implementing
+// tuple.
+#if GTEST_HAS_STD_TUPLE_
+# include <tuple>  // IWYU pragma: export
+# define GTEST_TUPLE_NAMESPACE_ ::std
+#endif  // GTEST_HAS_STD_TUPLE_
+
+// We include tr1::tuple even if std::tuple is available to define printers for
+// them.
+#if GTEST_HAS_TR1_TUPLE
+# ifndef GTEST_TUPLE_NAMESPACE_
+#  define GTEST_TUPLE_NAMESPACE_ ::std::tr1
+# endif  // GTEST_TUPLE_NAMESPACE_
+
+# if GTEST_USE_OWN_TR1_TUPLE
+// This file was GENERATED by command:
+//     pump.py gtest-tuple.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2009 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Implements a subset of TR1 tuple needed by Google Test and Google Mock.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+
+#include <utility>  // For ::std::pair.
+
+// The compiler used in Symbian has a bug that prevents us from declaring the
+// tuple template as a friend (it complains that tuple is redefined).  This
+// hack bypasses the bug by declaring the members that should otherwise be
+// private as public.
+// Sun Studio versions < 12 also have the above bug.
+#if defined(__SYMBIAN32__) || (defined(__SUNPRO_CC) && __SUNPRO_CC < 0x590)
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ public:
+#else
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ \
+    template <GTEST_10_TYPENAMES_(U)> friend class tuple; \
+   private:
+#endif
+
+// Visual Studio 2010, 2012, and 2013 define symbols in std::tr1 that conflict
+// with our own definitions. Therefore using our own tuple does not work on
+// those compilers.
+#if defined(_MSC_VER) && _MSC_VER >= 1600  /* 1600 is Visual Studio 2010 */
+# error "gtest's tuple doesn't compile on Visual Studio 2010 or later. \
+GTEST_USE_OWN_TR1_TUPLE must be set to 0 on those compilers."
+#endif
+
+// GTEST_n_TUPLE_(T) is the type of an n-tuple.
+#define GTEST_0_TUPLE_(T) tuple<>
+#define GTEST_1_TUPLE_(T) tuple<T##0, void, void, void, void, void, void, \
+    void, void, void>
+#define GTEST_2_TUPLE_(T) tuple<T##0, T##1, void, void, void, void, void, \
+    void, void, void>
+#define GTEST_3_TUPLE_(T) tuple<T##0, T##1, T##2, void, void, void, void, \
+    void, void, void>
+#define GTEST_4_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, void, void, void, \
+    void, void, void>
+#define GTEST_5_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, void, void, \
+    void, void, void>
+#define GTEST_6_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, void, \
+    void, void, void>
+#define GTEST_7_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    void, void, void>
+#define GTEST_8_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    T##7, void, void>
+#define GTEST_9_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    T##7, T##8, void>
+#define GTEST_10_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    T##7, T##8, T##9>
+
+// GTEST_n_TYPENAMES_(T) declares a list of n typenames.
+#define GTEST_0_TYPENAMES_(T)
+#define GTEST_1_TYPENAMES_(T) typename T##0
+#define GTEST_2_TYPENAMES_(T) typename T##0, typename T##1
+#define GTEST_3_TYPENAMES_(T) typename T##0, typename T##1, typename T##2
+#define GTEST_4_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3
+#define GTEST_5_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4
+#define GTEST_6_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5
+#define GTEST_7_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6
+#define GTEST_8_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6, typename T##7
+#define GTEST_9_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6, \
+    typename T##7, typename T##8
+#define GTEST_10_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6, \
+    typename T##7, typename T##8, typename T##9
+
+// In theory, defining stuff in the ::std namespace is undefined
+// behavior.  We can do this as we are playing the role of a standard
+// library vendor.
+namespace std
+{
+    namespace tr1
+    {
+
+        template <typename T0 = void, typename T1 = void, typename T2 = void,
+                  typename T3 = void, typename T4 = void, typename T5 = void,
+                  typename T6 = void, typename T7 = void, typename T8 = void,
+                  typename T9 = void>
+        class tuple;
+
+// Anything in namespace gtest_internal is Google Test's INTERNAL
+// IMPLEMENTATION DETAIL and MUST NOT BE USED DIRECTLY in user code.
+        namespace gtest_internal
+        {
+
+// ByRef<T>::type is T if T is a reference; otherwise it's const T&.
+            template <typename T>
+            struct ByRef {
+                typedef const T &type;
+            };  // NOLINT
+            template <typename T>
+            struct ByRef<T &> {
+                typedef T &type;
+            };  // NOLINT
+
+// A handy wrapper for ByRef.
+#define GTEST_BY_REF_(T) typename ::std::tr1::gtest_internal::ByRef<T>::type
+
+// AddRef<T>::type is T if T is a reference; otherwise it's T&.  This
+// is the same as tr1::add_reference<T>::type.
+            template <typename T>
+            struct AddRef {
+                typedef T &type;
+            };  // NOLINT
+            template <typename T>
+            struct AddRef<T &> {
+                typedef T &type;
+            };  // NOLINT
+
+// A handy wrapper for AddRef.
+#define GTEST_ADD_REF_(T) typename ::std::tr1::gtest_internal::AddRef<T>::type
+
+// A helper for implementing get<k>().
+            template <int k> class Get;
+
+// A helper for implementing tuple_element<k, T>.  kIndexValid is true
+// iff k < the number of fields in tuple type T.
+            template <bool kIndexValid, int kIndex, class Tuple>
+            struct TupleElement;
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 0, GTEST_10_TUPLE_(T) > {
+                typedef T0 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 1, GTEST_10_TUPLE_(T) > {
+                typedef T1 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 2, GTEST_10_TUPLE_(T) > {
+                typedef T2 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 3, GTEST_10_TUPLE_(T) > {
+                typedef T3 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 4, GTEST_10_TUPLE_(T) > {
+                typedef T4 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 5, GTEST_10_TUPLE_(T) > {
+                typedef T5 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 6, GTEST_10_TUPLE_(T) > {
+                typedef T6 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 7, GTEST_10_TUPLE_(T) > {
+                typedef T7 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 8, GTEST_10_TUPLE_(T) > {
+                typedef T8 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 9, GTEST_10_TUPLE_(T) > {
+                typedef T9 type;
+            };
+
+        }  // namespace gtest_internal
+
+        template <>
+        class tuple<>
+        {
+        public:
+            tuple() {}
+            tuple(const tuple & /* t */)  {}
+            tuple &operator=(const tuple & /* t */)
+            {
+                return *this;
+            }
+        };
+
+        template <GTEST_1_TYPENAMES_(T)>
+        class GTEST_1_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0) : f0_(f0) {}
+
+            tuple(const tuple& t) : f0_(t.f0_) {}
+
+            template <GTEST_1_TYPENAMES_(U)>
+            tuple(const GTEST_1_TUPLE_(U)& t) : f0_(t.f0_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_1_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_1_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_1_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_1_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                return *this;
+            }
+
+            T0 f0_;
+        };
+
+        template <GTEST_2_TYPENAMES_(T)>
+        class GTEST_2_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1) : f0_(f0),
+                f1_(f1) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_) {}
+
+            template <GTEST_2_TYPENAMES_(U)>
+            tuple(const GTEST_2_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_) {}
+            template <typename U0, typename U1>
+            tuple(const ::std::pair<U0, U1> &p) : f0_(p.first), f1_(p.second) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_2_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_2_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+            template <typename U0, typename U1>
+            tuple &operator=(const ::std::pair<U0, U1> &p) {
+                f0_ = p.first;
+                f1_ = p.second;
+                return *this;
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_2_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_2_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+        };
+
+        template <GTEST_3_TYPENAMES_(T)>
+        class GTEST_3_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2) : f0_(f0), f1_(f1), f2_(f2) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {}
+
+            template <GTEST_3_TYPENAMES_(U)>
+            tuple(const GTEST_3_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_3_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_3_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_3_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_3_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+        };
+
+        template <GTEST_4_TYPENAMES_(T)>
+        class GTEST_4_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3) : f0_(f0), f1_(f1), f2_(f2),
+                f3_(f3) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_) {}
+
+            template <GTEST_4_TYPENAMES_(U)>
+            tuple(const GTEST_4_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_4_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_4_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_4_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_4_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+        };
+
+        template <GTEST_5_TYPENAMES_(T)>
+        class GTEST_5_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3,
+                           GTEST_BY_REF_(T4) f4) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_) {}
+
+            template <GTEST_5_TYPENAMES_(U)>
+            tuple(const GTEST_5_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_5_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_5_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_5_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_5_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+        };
+
+        template <GTEST_6_TYPENAMES_(T)>
+        class GTEST_6_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+                f5_(f5) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_) {}
+
+            template <GTEST_6_TYPENAMES_(U)>
+            tuple(const GTEST_6_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_6_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_6_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_6_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_6_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+        };
+
+        template <GTEST_7_TYPENAMES_(T)>
+        class GTEST_7_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6) : f0_(f0), f1_(f1), f2_(f2),
+                f3_(f3), f4_(f4), f5_(f5), f6_(f6) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {}
+
+            template <GTEST_7_TYPENAMES_(U)>
+            tuple(const GTEST_7_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_7_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_7_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_7_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_7_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+        };
+
+        template <GTEST_8_TYPENAMES_(T)>
+        class GTEST_8_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6,
+                           GTEST_BY_REF_(T7) f7) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+                f5_(f5), f6_(f6), f7_(f7) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {}
+
+            template <GTEST_8_TYPENAMES_(U)>
+            tuple(const GTEST_8_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_8_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_8_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_8_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_8_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                f7_ = t.f7_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+            T7 f7_;
+        };
+
+        template <GTEST_9_TYPENAMES_(T)>
+        class GTEST_9_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7,
+                           GTEST_BY_REF_(T8) f8) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+                f5_(f5), f6_(f6), f7_(f7), f8_(f8) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {}
+
+            template <GTEST_9_TYPENAMES_(U)>
+            tuple(const GTEST_9_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_9_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_9_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_9_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_9_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                f7_ = t.f7_;
+                f8_ = t.f8_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+            T7 f7_;
+            T8 f8_;
+        };
+
+        template <GTEST_10_TYPENAMES_(T)>
+        class tuple
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_(),
+                f9_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7,
+                           GTEST_BY_REF_(T8) f8, GTEST_BY_REF_(T9) f9) : f0_(f0), f1_(f1), f2_(f2),
+                f3_(f3), f4_(f4), f5_(f5), f6_(f6), f7_(f7), f8_(f8), f9_(f9) {}
+
+            tuple(const tuple &t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_), f9_(t.f9_) {}
+
+            template <GTEST_10_TYPENAMES_(U)>
+            tuple(const GTEST_10_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_),
+                f9_(t.f9_) {}
+
+            tuple &operator=(const tuple &t)
+            {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_10_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_10_TUPLE_(U)& t)
+            {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_10_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_10_TUPLE_(U)& t)
+            {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                f7_ = t.f7_;
+                f8_ = t.f8_;
+                f9_ = t.f9_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+            T7 f7_;
+            T8 f8_;
+            T9 f9_;
+        };
+
+// 6.1.3.2 Tuple creation functions.
+
+// Known limitations: we don't support passing an
+// std::tr1::reference_wrapper<T> to make_tuple().  And we don't
+// implement tie().
+
+        inline tuple<> make_tuple()
+        {
+            return tuple<>();
+        }
+
+        template <GTEST_1_TYPENAMES_(T)>
+        inline GTEST_1_TUPLE_(T) make_tuple(const T0 &f0)
+        {
+            return GTEST_1_TUPLE_(T)(f0);
+        }
+
+        template <GTEST_2_TYPENAMES_(T)>
+        inline GTEST_2_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1)
+        {
+            return GTEST_2_TUPLE_(T)(f0, f1);
+        }
+
+        template <GTEST_3_TYPENAMES_(T)>
+        inline GTEST_3_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2)
+        {
+            return GTEST_3_TUPLE_(T)(f0, f1, f2);
+        }
+
+        template <GTEST_4_TYPENAMES_(T)>
+        inline GTEST_4_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3)
+        {
+            return GTEST_4_TUPLE_(T)(f0, f1, f2, f3);
+        }
+
+        template <GTEST_5_TYPENAMES_(T)>
+        inline GTEST_5_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4)
+        {
+            return GTEST_5_TUPLE_(T)(f0, f1, f2, f3, f4);
+        }
+
+        template <GTEST_6_TYPENAMES_(T)>
+        inline GTEST_6_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5)
+        {
+            return GTEST_6_TUPLE_(T)(f0, f1, f2, f3, f4, f5);
+        }
+
+        template <GTEST_7_TYPENAMES_(T)>
+        inline GTEST_7_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6)
+        {
+            return GTEST_7_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6);
+        }
+
+        template <GTEST_8_TYPENAMES_(T)>
+        inline GTEST_8_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6, const T7 &f7)
+        {
+            return GTEST_8_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7);
+        }
+
+        template <GTEST_9_TYPENAMES_(T)>
+        inline GTEST_9_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6, const T7 &f7,
+                                            const T8 &f8)
+        {
+            return GTEST_9_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8);
+        }
+
+        template <GTEST_10_TYPENAMES_(T)>
+        inline GTEST_10_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                             const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6, const T7 &f7,
+                                             const T8 &f8, const T9 &f9)
+        {
+            return GTEST_10_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8, f9);
+        }
+
+// 6.1.3.3 Tuple helper classes.
+
+        template <typename Tuple> struct tuple_size;
+
+        template <GTEST_0_TYPENAMES_(T)>
+        struct tuple_size<GTEST_0_TUPLE_(T) > {
+            static const int value = 0;
+        };
+
+        template <GTEST_1_TYPENAMES_(T)>
+        struct tuple_size<GTEST_1_TUPLE_(T) > {
+            static const int value = 1;
+        };
+
+        template <GTEST_2_TYPENAMES_(T)>
+        struct tuple_size<GTEST_2_TUPLE_(T) > {
+            static const int value = 2;
+        };
+
+        template <GTEST_3_TYPENAMES_(T)>
+        struct tuple_size<GTEST_3_TUPLE_(T) > {
+            static const int value = 3;
+        };
+
+        template <GTEST_4_TYPENAMES_(T)>
+        struct tuple_size<GTEST_4_TUPLE_(T) > {
+            static const int value = 4;
+        };
+
+        template <GTEST_5_TYPENAMES_(T)>
+        struct tuple_size<GTEST_5_TUPLE_(T) > {
+            static const int value = 5;
+        };
+
+        template <GTEST_6_TYPENAMES_(T)>
+        struct tuple_size<GTEST_6_TUPLE_(T) > {
+            static const int value = 6;
+        };
+
+        template <GTEST_7_TYPENAMES_(T)>
+        struct tuple_size<GTEST_7_TUPLE_(T) > {
+            static const int value = 7;
+        };
+
+        template <GTEST_8_TYPENAMES_(T)>
+        struct tuple_size<GTEST_8_TUPLE_(T) > {
+            static const int value = 8;
+        };
+
+        template <GTEST_9_TYPENAMES_(T)>
+        struct tuple_size<GTEST_9_TUPLE_(T) > {
+            static const int value = 9;
+        };
+
+        template <GTEST_10_TYPENAMES_(T)>
+        struct tuple_size<GTEST_10_TUPLE_(T) > {
+            static const int value = 10;
+        };
+
+        template <int k, class Tuple>
+        struct tuple_element {
+            typedef typename gtest_internal::TupleElement<
+            k < (tuple_size<Tuple>::value), k, Tuple>::type type;
+        };
+
+#define GTEST_TUPLE_ELEMENT_(k, Tuple) typename tuple_element<k, Tuple >::type
+
+// 6.1.3.4 Element access.
+
+        namespace gtest_internal
+        {
+
+            template <>
+            class Get<0>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f0_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f0_;
+                }
+            };
+
+            template <>
+            class Get<1>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f1_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f1_;
+                }
+            };
+
+            template <>
+            class Get<2>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f2_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f2_;
+                }
+            };
+
+            template <>
+            class Get<3>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f3_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f3_;
+                }
+            };
+
+            template <>
+            class Get<4>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f4_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f4_;
+                }
+            };
+
+            template <>
+            class Get<5>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f5_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f5_;
+                }
+            };
+
+            template <>
+            class Get<6>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f6_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f6_;
+                }
+            };
+
+            template <>
+            class Get<7>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f7_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f7_;
+                }
+            };
+
+            template <>
+            class Get<8>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f8_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f8_;
+                }
+            };
+
+            template <>
+            class Get<9>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f9_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f9_;
+                }
+            };
+
+        }  // namespace gtest_internal
+
+        template <int k, GTEST_10_TYPENAMES_(T)>
+        GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_10_TUPLE_(T)))
+        get(GTEST_10_TUPLE_(T)& t)
+        {
+            return gtest_internal::Get<k>::Field(t);
+        }
+
+        template <int k, GTEST_10_TYPENAMES_(T)>
+        GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(k,  GTEST_10_TUPLE_(T)))
+        get(const GTEST_10_TUPLE_(T)& t)
+        {
+            return gtest_internal::Get<k>::ConstField(t);
+        }
+
+// 6.1.3.5 Relational operators
+
+// We only implement == and !=, as we don't have a need for the rest yet.
+
+        namespace gtest_internal
+        {
+
+// SameSizeTuplePrefixComparator<k, k>::Eq(t1, t2) returns true if the
+// first k fields of t1 equals the first k fields of t2.
+// SameSizeTuplePrefixComparator(k1, k2) would be a compiler error if
+// k1 != k2.
+            template <int kSize1, int kSize2>
+            struct SameSizeTuplePrefixComparator;
+
+            template <>
+            struct SameSizeTuplePrefixComparator<0, 0> {
+                template <class Tuple1, class Tuple2>
+                static bool Eq(const Tuple1 & /* t1 */, const Tuple2 & /* t2 */)
+                {
+                    return true;
+                }
+            };
+
+            template <int k>
+            struct SameSizeTuplePrefixComparator<k, k> {
+                template <class Tuple1, class Tuple2>
+                static bool Eq(const Tuple1 &t1, const Tuple2 &t2)
+                {
+                    return SameSizeTuplePrefixComparator<k - 1, k - 1>::Eq(t1, t2) &&
+                           ::std::tr1::get<k - 1>(t1) == ::std::tr1::get<k - 1>(t2);
+                }
+            };
+
+        }  // namespace gtest_internal
+
+        template <GTEST_10_TYPENAMES_(T), GTEST_10_TYPENAMES_(U)>
+        inline bool operator==(const GTEST_10_TUPLE_(T)& t,
+                               const GTEST_10_TUPLE_(U)& u)
+        {
+            return gtest_internal::SameSizeTuplePrefixComparator<
+                   tuple_size<GTEST_10_TUPLE_(T) >::value,
+                   tuple_size<GTEST_10_TUPLE_(U) >::value>::Eq(t, u);
+        }
+
+        template <GTEST_10_TYPENAMES_(T), GTEST_10_TYPENAMES_(U)>
+        inline bool operator!=(const GTEST_10_TUPLE_(T)& t,
+                               const GTEST_10_TUPLE_(U)& u)
+        {
+            return !(t == u);
+        }
+
+// 6.1.4 Pairs.
+// Unimplemented.
+
+    }  // namespace tr1
+}  // namespace std
+
+#undef GTEST_0_TUPLE_
+#undef GTEST_1_TUPLE_
+#undef GTEST_2_TUPLE_
+#undef GTEST_3_TUPLE_
+#undef GTEST_4_TUPLE_
+#undef GTEST_5_TUPLE_
+#undef GTEST_6_TUPLE_
+#undef GTEST_7_TUPLE_
+#undef GTEST_8_TUPLE_
+#undef GTEST_9_TUPLE_
+#undef GTEST_10_TUPLE_
+
+#undef GTEST_0_TYPENAMES_
+#undef GTEST_1_TYPENAMES_
+#undef GTEST_2_TYPENAMES_
+#undef GTEST_3_TYPENAMES_
+#undef GTEST_4_TYPENAMES_
+#undef GTEST_5_TYPENAMES_
+#undef GTEST_6_TYPENAMES_
+#undef GTEST_7_TYPENAMES_
+#undef GTEST_8_TYPENAMES_
+#undef GTEST_9_TYPENAMES_
+#undef GTEST_10_TYPENAMES_
+
+#undef GTEST_DECLARE_TUPLE_AS_FRIEND_
+#undef GTEST_BY_REF_
+#undef GTEST_ADD_REF_
+#undef GTEST_TUPLE_ELEMENT_
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+# elif GTEST_ENV_HAS_STD_TUPLE_
+#  include <tuple>
+// C++11 puts its tuple into the ::std namespace rather than
+// ::std::tr1.  gtest expects tuple to live in ::std::tr1, so put it there.
+// This causes undefined behavior, but supported compilers react in
+// the way we intend.
+namespace std
+{
+    namespace tr1
+    {
+        using ::std::get;
+        using ::std::make_tuple;
+        using ::std::tuple;
+        using ::std::tuple_element;
+        using ::std::tuple_size;
+    }
+}
+
+# elif GTEST_OS_SYMBIAN
+
+// On Symbian, BOOST_HAS_TR1_TUPLE causes Boost's TR1 tuple library to
+// use STLport's tuple implementation, which unfortunately doesn't
+// work as the copy of STLport distributed with Symbian is incomplete.
+// By making sure BOOST_HAS_TR1_TUPLE is undefined, we force Boost to
+// use its own tuple implementation.
+#  ifdef BOOST_HAS_TR1_TUPLE
+#   undef BOOST_HAS_TR1_TUPLE
+#  endif  // BOOST_HAS_TR1_TUPLE
+
+// This prevents <boost/tr1/detail/config.hpp>, which defines
+// BOOST_HAS_TR1_TUPLE, from being #included by Boost's <tuple>.
+#  define BOOST_TR1_DETAIL_CONFIG_HPP_INCLUDED
+#  include <tuple>  // IWYU pragma: export  // NOLINT
+
+# elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40000)
+// GCC 4.0+ implements tr1/tuple in the <tr1/tuple> header.  This does
+// not conform to the TR1 spec, which requires the header to be <tuple>.
+
+#  if !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302
+// Until version 4.3.2, gcc has a bug that causes <tr1/functional>,
+// which is #included by <tr1/tuple>, to not compile when RTTI is
+// disabled.  _TR1_FUNCTIONAL is the header guard for
+// <tr1/functional>.  Hence the following #define is a hack to prevent
+// <tr1/functional> from being included.
+#   define _TR1_FUNCTIONAL 1
+#   include <tr1/tuple>
+#   undef _TR1_FUNCTIONAL  // Allows the user to #include
+// <tr1/functional> if he chooses to.
+#  else
+#   include <tr1/tuple>  // NOLINT
+#  endif  // !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302
+
+# else
+// If the compiler is not GCC 4.0+, we assume the user is using a
+// spec-conforming TR1 implementation.
+#  include <tuple>  // IWYU pragma: export  // NOLINT
+# endif  // GTEST_USE_OWN_TR1_TUPLE
+
+#endif  // GTEST_HAS_TR1_TUPLE
+
+// Determines whether clone(2) is supported.
+// Usually it will only be available on Linux, excluding
+// Linux on the Itanium architecture.
+// Also see http://linux.die.net/man/2/clone.
+#ifndef GTEST_HAS_CLONE
+// The user didn't tell us, so we need to figure it out.
+
+# if GTEST_OS_LINUX && !defined(__ia64__)
+#  if GTEST_OS_LINUX_ANDROID
+// On Android, clone() became available at different API levels for each 32-bit
+// architecture.
+#    if defined(__LP64__) || \
+        (defined(__arm__) && __ANDROID_API__ >= 9) || \
+        (defined(__mips__) && __ANDROID_API__ >= 12) || \
+        (defined(__i386__) && __ANDROID_API__ >= 17)
+#     define GTEST_HAS_CLONE 1
+#    else
+#     define GTEST_HAS_CLONE 0
+#    endif
+#  else
+#   define GTEST_HAS_CLONE 1
+#  endif
+# else
+#  define GTEST_HAS_CLONE 0
+# endif  // GTEST_OS_LINUX && !defined(__ia64__)
+
+#endif  // GTEST_HAS_CLONE
+
+// Determines whether to support stream redirection. This is used to test
+// output correctness and to implement death tests.
+#ifndef GTEST_HAS_STREAM_REDIRECTION
+// By default, we assume that stream redirection is supported on all
+// platforms except known mobile ones.
+# if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || \
+    GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT
+#  define GTEST_HAS_STREAM_REDIRECTION 0
+# else
+#  define GTEST_HAS_STREAM_REDIRECTION 1
+# endif  // !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_SYMBIAN
+#endif  // GTEST_HAS_STREAM_REDIRECTION
+
+// Determines whether to support death tests.
+// Google Test does not support death tests for VC 7.1 and earlier as
+// abort() in a VC 7.1 application compiled as GUI in debug config
+// pops up a dialog window that cannot be suppressed programmatically.
+#if (GTEST_OS_LINUX || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS || \
+     (GTEST_OS_MAC && !GTEST_OS_IOS) || \
+     (GTEST_OS_WINDOWS_DESKTOP && _MSC_VER >= 1400) || \
+     GTEST_OS_WINDOWS_MINGW || GTEST_OS_AIX || GTEST_OS_HPUX || \
+     GTEST_OS_OPENBSD || GTEST_OS_QNX || GTEST_OS_FREEBSD || GTEST_OS_NETBSD)
+# define GTEST_HAS_DEATH_TEST 1
+#endif
+
+// We don't support MSVC 7.1 with exceptions disabled now.  Therefore
+// all the compilers we care about are adequate for supporting
+// value-parameterized tests.
+#define GTEST_HAS_PARAM_TEST 1
+
+// Determines whether to support type-driven tests.
+
+// Typed tests need <typeinfo> and variadic macros, which GCC, VC++ 8.0,
+// Sun Pro CC, IBM Visual Age, and HP aCC support.
+#if defined(__GNUC__) || (_MSC_VER >= 1400) || defined(__SUNPRO_CC) || \
+    defined(__IBMCPP__) || defined(__HP_aCC)
+# define GTEST_HAS_TYPED_TEST 1
+# define GTEST_HAS_TYPED_TEST_P 1
+#endif
+
+// Determines whether to support Combine(). This only makes sense when
+// value-parameterized tests are enabled.  The implementation doesn't
+// work on Sun Studio since it doesn't understand templated conversion
+// operators.
+#if GTEST_HAS_PARAM_TEST && GTEST_HAS_TR1_TUPLE && !defined(__SUNPRO_CC)
+# define GTEST_HAS_COMBINE 1
+#endif
+
+// Determines whether the system compiler uses UTF-16 for encoding wide strings.
+#define GTEST_WIDE_STRING_USES_UTF16_ \
+    (GTEST_OS_WINDOWS || GTEST_OS_CYGWIN || GTEST_OS_SYMBIAN || GTEST_OS_AIX)
+
+// Determines whether test results can be streamed to a socket.
+#if GTEST_OS_LINUX
+# define GTEST_CAN_STREAM_RESULTS_ 1
+#endif
+
+// Defines some utility macros.
+
+// The GNU compiler emits a warning if nested "if" statements are followed by
+// an "else" statement and braces are not used to explicitly disambiguate the
+// "else" binding.  This leads to problems with code like:
+//
+//   if (gate)
+//     ASSERT_*(condition) << "Some message";
+//
+// The "switch (0) case 0:" idiom is used to suppress this.
+#ifdef __INTEL_COMPILER
+# define GTEST_AMBIGUOUS_ELSE_BLOCKER_
+#else
+# define GTEST_AMBIGUOUS_ELSE_BLOCKER_ switch (0) case 0: default:  // NOLINT
+#endif
+
+// Use this annotation at the end of a struct/class definition to
+// prevent the compiler from optimizing away instances that are never
+// used.  This is useful when all interesting logic happens inside the
+// c'tor and / or d'tor.  Example:
+//
+//   struct Foo {
+//     Foo() { ... }
+//   } GTEST_ATTRIBUTE_UNUSED_;
+//
+// Also use it after a variable or parameter declaration to tell the
+// compiler the variable/parameter does not have to be used.
+#if defined(__GNUC__) && !defined(COMPILER_ICC)
+# define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused))
+#elif defined(__clang__)
+# if __has_attribute(unused)
+#  define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused))
+# endif
+#endif
+#ifndef GTEST_ATTRIBUTE_UNUSED_
+# define GTEST_ATTRIBUTE_UNUSED_
+#endif
+
+// Use this annotation before a function that takes a printf format string.
+#if defined(__GNUC__) && !defined(COMPILER_ICC)
+# if defined(__MINGW_PRINTF_FORMAT)
+// MinGW has two different printf implementations. Ensure the format macro
+// matches the selected implementation. See
+// https://sourceforge.net/p/mingw-w64/wiki2/gnu%20printf/.
+#  define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check) \
+       __attribute__((__format__(__MINGW_PRINTF_FORMAT, string_index, \
+                                 first_to_check)))
+# else
+#  define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check) \
+       __attribute__((__format__(__printf__, string_index, first_to_check)))
+# endif
+#else
+# define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check)
+#endif
+
+// A macro to disallow operator=
+// This should be used in the private: declarations for a class.
+#define GTEST_DISALLOW_ASSIGN_(type)\
+  void operator=(type const &)
+
+// A macro to disallow copy constructor and operator=
+// This should be used in the private: declarations for a class.
+#define GTEST_DISALLOW_COPY_AND_ASSIGN_(type)\
+  type(type const &);\
+  GTEST_DISALLOW_ASSIGN_(type)
+
+// Tell the compiler to warn about unused return values for functions declared
+// with this macro.  The macro should be used on function declarations
+// following the argument list:
+//
+//   Sprocket* AllocateSprocket() GTEST_MUST_USE_RESULT_;
+#if defined(__GNUC__) && (GTEST_GCC_VER_ >= 30400) && !defined(COMPILER_ICC)
+# define GTEST_MUST_USE_RESULT_ __attribute__ ((warn_unused_result))
+#else
+# define GTEST_MUST_USE_RESULT_
+#endif  // __GNUC__ && (GTEST_GCC_VER_ >= 30400) && !COMPILER_ICC
+
+// MS C++ compiler emits warning when a conditional expression is compile time
+// constant. In some contexts this warning is false positive and needs to be
+// suppressed. Use the following two macros in such cases:
+//
+// GTEST_INTENTIONAL_CONST_COND_PUSH_()
+// while (true) {
+// GTEST_INTENTIONAL_CONST_COND_POP_()
+// }
+# define GTEST_INTENTIONAL_CONST_COND_PUSH_() \
+    GTEST_DISABLE_MSC_WARNINGS_PUSH_(4127)
+# define GTEST_INTENTIONAL_CONST_COND_POP_() \
+    GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+// Determine whether the compiler supports Microsoft's Structured Exception
+// Handling.  This is supported by several Windows compilers but generally
+// does not exist on any other system.
+#ifndef GTEST_HAS_SEH
+// The user didn't tell us, so we need to figure it out.
+
+# if defined(_MSC_VER) || defined(__BORLANDC__)
+// These two compilers are known to support SEH.
+#  define GTEST_HAS_SEH 1
+# else
+// Assume no SEH.
+#  define GTEST_HAS_SEH 0
+# endif
+
+#define GTEST_IS_THREADSAFE \
+    (GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ \
+     || (GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT) \
+     || GTEST_HAS_PTHREAD)
+
+#endif  // GTEST_HAS_SEH
+
+// GTEST_API_ qualifies all symbols that must be exported. The definitions below
+// are guarded by #ifndef to give embedders a chance to define GTEST_API_ in
+// gtest/internal/custom/gtest-port.h
+#ifndef GTEST_API_
+
+#ifdef _MSC_VER
+# if GTEST_LINKED_AS_SHARED_LIBRARY
+#  define GTEST_API_ __declspec(dllimport)
+# elif GTEST_CREATE_SHARED_LIBRARY
+#  define GTEST_API_ __declspec(dllexport)
+# endif
+#elif __GNUC__ >= 4 || defined(__clang__)
+# define GTEST_API_ __attribute__((visibility ("default")))
+#endif // _MSC_VER
+
+#endif // GTEST_API_
+
+#ifndef GTEST_API_
+# define GTEST_API_
+#endif // GTEST_API_
+
+#ifdef __GNUC__
+// Ask the compiler to never inline a given function.
+# define GTEST_NO_INLINE_ __attribute__((noinline))
+#else
+# define GTEST_NO_INLINE_
+#endif
+
+// _LIBCPP_VERSION is defined by the libc++ library from the LLVM project.
+#if defined(__GLIBCXX__) || defined(_LIBCPP_VERSION)
+# define GTEST_HAS_CXXABI_H_ 1
+#else
+# define GTEST_HAS_CXXABI_H_ 0
+#endif
+
+// A function level attribute to disable checking for use of uninitialized
+// memory when built with MemorySanitizer.
+#if defined(__clang__)
+# if __has_feature(memory_sanitizer)
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ \
+       __attribute__((no_sanitize_memory))
+# else
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+# endif  // __has_feature(memory_sanitizer)
+#else
+# define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+#endif  // __clang__
+
+// A function level attribute to disable AddressSanitizer instrumentation.
+#if defined(__clang__)
+# if __has_feature(address_sanitizer)
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ \
+       __attribute__((no_sanitize_address))
+# else
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+# endif  // __has_feature(address_sanitizer)
+#else
+# define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+#endif  // __clang__
+
+// A function level attribute to disable ThreadSanitizer instrumentation.
+#if defined(__clang__)
+# if __has_feature(thread_sanitizer)
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ \
+       __attribute__((no_sanitize_thread))
+# else
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+# endif  // __has_feature(thread_sanitizer)
+#else
+# define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+#endif  // __clang__
+
+namespace testing
+{
+
+    class Message;
+
+#if defined(GTEST_TUPLE_NAMESPACE_)
+// Import tuple and friends into the ::testing namespace.
+// It is part of our interface, having them in ::testing allows us to change
+// their types as needed.
+    using GTEST_TUPLE_NAMESPACE_::get;
+    using GTEST_TUPLE_NAMESPACE_::make_tuple;
+    using GTEST_TUPLE_NAMESPACE_::tuple;
+    using GTEST_TUPLE_NAMESPACE_::tuple_size;
+    using GTEST_TUPLE_NAMESPACE_::tuple_element;
+#endif  // defined(GTEST_TUPLE_NAMESPACE_)
+
+    namespace internal
+    {
+
+// A secret type that Google Test users don't know about.  It has no
+// definition on purpose.  Therefore it's impossible to create a
+// Secret object, which is what we want.
+        class Secret;
+
+// The GTEST_COMPILE_ASSERT_ macro can be used to verify that a compile time
+// expression is true. For example, you could use it to verify the
+// size of a static array:
+//
+//   GTEST_COMPILE_ASSERT_(GTEST_ARRAY_SIZE_(names) == NUM_NAMES,
+//                         names_incorrect_size);
+//
+// or to make sure a struct is smaller than a certain size:
+//
+//   GTEST_COMPILE_ASSERT_(sizeof(foo) < 128, foo_too_large);
+//
+// The second argument to the macro is the name of the variable. If
+// the expression is false, most compilers will issue a warning/error
+// containing the name of the variable.
+
+#if GTEST_LANG_CXX11
+# define GTEST_COMPILE_ASSERT_(expr, msg) static_assert(expr, #msg)
+#else  // !GTEST_LANG_CXX11
+        template <bool>
+        struct CompileAssert {
+        };
+
+# define GTEST_COMPILE_ASSERT_(expr, msg) \
+  typedef ::testing::internal::CompileAssert<(static_cast<bool>(expr))> \
+      msg[static_cast<bool>(expr) ? 1 : -1] GTEST_ATTRIBUTE_UNUSED_
+#endif  // !GTEST_LANG_CXX11
+
+// Implementation details of GTEST_COMPILE_ASSERT_:
+//
+// (In C++11, we simply use static_assert instead of the following)
+//
+// - GTEST_COMPILE_ASSERT_ works by defining an array type that has -1
+//   elements (and thus is invalid) when the expression is false.
+//
+// - The simpler definition
+//
+//    #define GTEST_COMPILE_ASSERT_(expr, msg) typedef char msg[(expr) ? 1 : -1]
+//
+//   does not work, as gcc supports variable-length arrays whose sizes
+//   are determined at run-time (this is gcc's extension and not part
+//   of the C++ standard).  As a result, gcc fails to reject the
+//   following code with the simple definition:
+//
+//     int foo;
+//     GTEST_COMPILE_ASSERT_(foo, msg); // not supposed to compile as foo is
+//                                      // not a compile-time constant.
+//
+// - By using the type CompileAssert<(bool(expr))>, we ensures that
+//   expr is a compile-time constant.  (Template arguments must be
+//   determined at compile-time.)
+//
+// - The outter parentheses in CompileAssert<(bool(expr))> are necessary
+//   to work around a bug in gcc 3.4.4 and 4.0.1.  If we had written
+//
+//     CompileAssert<bool(expr)>
+//
+//   instead, these compilers will refuse to compile
+//
+//     GTEST_COMPILE_ASSERT_(5 > 0, some_message);
+//
+//   (They seem to think the ">" in "5 > 0" marks the end of the
+//   template argument list.)
+//
+// - The array size is (bool(expr) ? 1 : -1), instead of simply
+//
+//     ((expr) ? 1 : -1).
+//
+//   This is to avoid running into a bug in MS VC 7.1, which
+//   causes ((0.0) ? 1 : -1) to incorrectly evaluate to 1.
+
+// StaticAssertTypeEqHelper is used by StaticAssertTypeEq defined in gtest.h.
+//
+// This template is declared, but intentionally undefined.
+        template <typename T1, typename T2>
+        struct StaticAssertTypeEqHelper;
+
+        template <typename T>
+        struct StaticAssertTypeEqHelper<T, T> {
+            enum { value = true };
+        };
+
+// Evaluates to the number of elements in 'array'.
+#define GTEST_ARRAY_SIZE_(array) (sizeof(array) / sizeof(array[0]))
+
+#if GTEST_HAS_GLOBAL_STRING
+        typedef ::string string;
+#else
+        typedef ::std::string string;
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+        typedef ::wstring wstring;
+#elif GTEST_HAS_STD_WSTRING
+        typedef ::std::wstring wstring;
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+// A helper for suppressing warnings on constant condition.  It just
+// returns 'condition'.
+        GTEST_API_ bool IsTrue(bool condition);
+
+// Defines scoped_ptr.
+
+// This implementation of scoped_ptr is PARTIAL - it only contains
+// enough stuff to satisfy Google Test's need.
+        template <typename T>
+        class scoped_ptr
+        {
+        public:
+            typedef T element_type;
+
+            explicit scoped_ptr(T *p = NULL) : ptr_(p) {}
+            ~scoped_ptr()
+            {
+                reset();
+            }
+
+            T &operator*() const
+            {
+                return *ptr_;
+            }
+            T *operator->() const
+            {
+                return ptr_;
+            }
+            T *get() const
+            {
+                return ptr_;
+            }
+
+            T *release()
+            {
+                T *const ptr = ptr_;
+                ptr_ = NULL;
+                return ptr;
+            }
+
+            void reset(T *p = NULL)
+            {
+                if (p != ptr_) {
+                    if (IsTrue(sizeof(T) > 0)) {  // Makes sure T is a complete type.
+                        delete ptr_;
+                    }
+                    ptr_ = p;
+                }
+            }
+
+            friend void swap(scoped_ptr &a, scoped_ptr &b)
+            {
+                using std::swap;
+                swap(a.ptr_, b.ptr_);
+            }
+
+        private:
+            T *ptr_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(scoped_ptr);
+        };
+
+// Defines RE.
+
+// A simple C++ wrapper for <regex.h>.  It uses the POSIX Extended
+// Regular Expression syntax.
+        class GTEST_API_ RE
+        {
+        public:
+            // A copy constructor is required by the Standard to initialize object
+            // references from r-values.
+            RE(const RE &other)
+            {
+                Init(other.pattern());
+            }
+
+            // Constructs an RE from a string.
+            RE(const ::std::string &regex)
+            {
+                Init(regex.c_str());    // NOLINT
+            }
+
+#if GTEST_HAS_GLOBAL_STRING
+
+            RE(const ::string &regex)
+            {
+                Init(regex.c_str());    // NOLINT
+            }
+
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+            RE(const char *regex)
+            {
+                Init(regex);    // NOLINT
+            }
+            ~RE();
+
+            // Returns the string representation of the regex.
+            const char *pattern() const
+            {
+                return pattern_;
+            }
+
+            // FullMatch(str, re) returns true iff regular expression re matches
+            // the entire str.
+            // PartialMatch(str, re) returns true iff regular expression re
+            // matches a substring of str (including str itself).
+            //
+            // TODO(wan@google.com): make FullMatch() and PartialMatch() work
+            // when str contains NUL characters.
+            static bool FullMatch(const ::std::string &str, const RE &re)
+            {
+                return FullMatch(str.c_str(), re);
+            }
+            static bool PartialMatch(const ::std::string &str, const RE &re)
+            {
+                return PartialMatch(str.c_str(), re);
+            }
+
+#if GTEST_HAS_GLOBAL_STRING
+
+            static bool FullMatch(const ::string &str, const RE &re)
+            {
+                return FullMatch(str.c_str(), re);
+            }
+            static bool PartialMatch(const ::string &str, const RE &re)
+            {
+                return PartialMatch(str.c_str(), re);
+            }
+
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+            static bool FullMatch(const char *str, const RE &re);
+            static bool PartialMatch(const char *str, const RE &re);
+
+        private:
+            void Init(const char *regex);
+
+            // We use a const char* instead of an std::string, as Google Test used to be
+            // used where std::string is not available.  TODO(wan@google.com): change to
+            // std::string.
+            const char *pattern_;
+            bool is_valid_;
+
+#if GTEST_USES_POSIX_RE
+
+            regex_t full_regex_;     // For FullMatch().
+            regex_t partial_regex_;  // For PartialMatch().
+
+#else  // GTEST_USES_SIMPLE_RE
+
+            const char *full_pattern_;  // For FullMatch();
+
+#endif
+
+            GTEST_DISALLOW_ASSIGN_(RE);
+        };
+
+// Formats a source file path and a line number as they would appear
+// in an error message from the compiler used to compile this code.
+        GTEST_API_ ::std::string FormatFileLocation(const char *file, int line);
+
+// Formats a file location for compiler-independent XML output.
+// Although this function is not platform dependent, we put it next to
+// FormatFileLocation in order to contrast the two functions.
+        GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(const char *file,
+                                                                       int line);
+
+// Defines logging utilities:
+//   GTEST_LOG_(severity) - logs messages at the specified severity level. The
+//                          message itself is streamed into the macro.
+//   LogToStderr()  - directs all log messages to stderr.
+//   FlushInfoLog() - flushes informational log messages.
+
+        enum GTestLogSeverity {
+            GTEST_INFO,
+            GTEST_WARNING,
+            GTEST_ERROR,
+            GTEST_FATAL
+        };
+
+// Formats log entry severity, provides a stream object for streaming the
+// log message, and terminates the message with a newline when going out of
+// scope.
+        class GTEST_API_ GTestLog
+        {
+        public:
+            GTestLog(GTestLogSeverity severity, const char *file, int line);
+
+            // Flushes the buffers and, if severity is GTEST_FATAL, aborts the program.
+            ~GTestLog();
+
+            ::std::ostream &GetStream()
+            {
+                return ::std::cerr;
+            }
+
+        private:
+            const GTestLogSeverity severity_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestLog);
+        };
+
+#if !defined(GTEST_LOG_)
+
+# define GTEST_LOG_(severity) \
+    ::testing::internal::GTestLog(::testing::internal::GTEST_##severity, \
+                                  __FILE__, __LINE__).GetStream()
+
+        inline void LogToStderr() {}
+        inline void FlushInfoLog()
+        {
+            fflush(NULL);
+        }
+
+#endif  // !defined(GTEST_LOG_)
+
+#if !defined(GTEST_CHECK_)
+// INTERNAL IMPLEMENTATION - DO NOT USE.
+//
+// GTEST_CHECK_ is an all-mode assert. It aborts the program if the condition
+// is not satisfied.
+//  Synopsis:
+//    GTEST_CHECK_(boolean_condition);
+//     or
+//    GTEST_CHECK_(boolean_condition) << "Additional message";
+//
+//    This checks the condition and if the condition is not satisfied
+//    it prints message about the condition violation, including the
+//    condition itself, plus additional message streamed into it, if any,
+//    and then it aborts the program. It aborts the program irrespective of
+//    whether it is built in the debug mode or not.
+# define GTEST_CHECK_(condition) \
+    GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+    if (::testing::internal::IsTrue(condition)) \
+      ; \
+    else \
+      GTEST_LOG_(FATAL) << "Condition " #condition " failed. "
+#endif  // !defined(GTEST_CHECK_)
+
+// An all-mode assert to verify that the given POSIX-style function
+// call returns 0 (indicating success).  Known limitation: this
+// doesn't expand to a balanced 'if' statement, so enclose the macro
+// in {} if you need to use it as the only statement in an 'if'
+// branch.
+#define GTEST_CHECK_POSIX_SUCCESS_(posix_call) \
+  if (const int gtest_error = (posix_call)) \
+    GTEST_LOG_(FATAL) << #posix_call << "failed with error " \
+                      << gtest_error
+
+#if GTEST_HAS_STD_MOVE_
+        using std::move;
+#else  // GTEST_HAS_STD_MOVE_
+        template <typename T>
+        const T &move(const T &t)
+        {
+            return t;
+        }
+#endif  // GTEST_HAS_STD_MOVE_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Use ImplicitCast_ as a safe version of static_cast for upcasting in
+// the type hierarchy (e.g. casting a Foo* to a SuperclassOfFoo* or a
+// const Foo*).  When you use ImplicitCast_, the compiler checks that
+// the cast is safe.  Such explicit ImplicitCast_s are necessary in
+// surprisingly many situations where C++ demands an exact type match
+// instead of an argument type convertible to a target type.
+//
+// The syntax for using ImplicitCast_ is the same as for static_cast:
+//
+//   ImplicitCast_<ToType>(expr)
+//
+// ImplicitCast_ would have been part of the C++ standard library,
+// but the proposal was submitted too late.  It will probably make
+// its way into the language in the future.
+//
+// This relatively ugly name is intentional. It prevents clashes with
+// similar functions users may have (e.g., implicit_cast). The internal
+// namespace alone is not enough because the function can be found by ADL.
+        template<typename To>
+        inline To ImplicitCast_(To x)
+        {
+            return x;
+        }
+
+// When you upcast (that is, cast a pointer from type Foo to type
+// SuperclassOfFoo), it's fine to use ImplicitCast_<>, since upcasts
+// always succeed.  When you downcast (that is, cast a pointer from
+// type Foo to type SubclassOfFoo), static_cast<> isn't safe, because
+// how do you know the pointer is really of type SubclassOfFoo?  It
+// could be a bare Foo, or of type DifferentSubclassOfFoo.  Thus,
+// when you downcast, you should use this macro.  In debug mode, we
+// use dynamic_cast<> to double-check the downcast is legal (we die
+// if it's not).  In normal mode, we do the efficient static_cast<>
+// instead.  Thus, it's important to test in debug mode to make sure
+// the cast is legal!
+//    This is the only place in the code we should use dynamic_cast<>.
+// In particular, you SHOULDN'T be using dynamic_cast<> in order to
+// do RTTI (eg code like this:
+//    if (dynamic_cast<Subclass1>(foo)) HandleASubclass1Object(foo);
+//    if (dynamic_cast<Subclass2>(foo)) HandleASubclass2Object(foo);
+// You should design the code some other way not to need this.
+//
+// This relatively ugly name is intentional. It prevents clashes with
+// similar functions users may have (e.g., down_cast). The internal
+// namespace alone is not enough because the function can be found by ADL.
+        template<typename To, typename From>  // use like this: DownCast_<T*>(foo);
+        inline To DownCast_(From *f)    // so we only accept pointers
+        {
+            // Ensures that To is a sub-type of From *.  This test is here only
+            // for compile-time type checking, and has no overhead in an
+            // optimized build at run-time, as it will be optimized away
+            // completely.
+            GTEST_INTENTIONAL_CONST_COND_PUSH_()
+            if (false) {
+                GTEST_INTENTIONAL_CONST_COND_POP_()
+                const To to = NULL;
+                ::testing::internal::ImplicitCast_<From *>(to);
+            }
+
+#if GTEST_HAS_RTTI
+            // RTTI: debug mode only!
+            GTEST_CHECK_(f == NULL || dynamic_cast<To>(f) != NULL);
+#endif
+            return static_cast<To>(f);
+        }
+
+// Downcasts the pointer of type Base to Derived.
+// Derived must be a subclass of Base. The parameter MUST
+// point to a class of type Derived, not any subclass of it.
+// When RTTI is available, the function performs a runtime
+// check to enforce this.
+        template <class Derived, class Base>
+        Derived *CheckedDowncastToActualType(Base *base)
+        {
+#if GTEST_HAS_RTTI
+            GTEST_CHECK_(typeid(*base) == typeid(Derived));
+#endif
+
+#if GTEST_HAS_DOWNCAST_
+            return ::down_cast<Derived *>(base);
+#elif GTEST_HAS_RTTI
+            return dynamic_cast<Derived *>(base); // NOLINT
+#else
+            return static_cast<Derived *>(base); // Poor man's downcast.
+#endif
+        }
+
+#if GTEST_HAS_STREAM_REDIRECTION
+
+// Defines the stderr capturer:
+//   CaptureStdout     - starts capturing stdout.
+//   GetCapturedStdout - stops capturing stdout and returns the captured string.
+//   CaptureStderr     - starts capturing stderr.
+//   GetCapturedStderr - stops capturing stderr and returns the captured string.
+//
+        GTEST_API_ void CaptureStdout();
+        GTEST_API_ std::string GetCapturedStdout();
+        GTEST_API_ void CaptureStderr();
+        GTEST_API_ std::string GetCapturedStderr();
+
+#endif  // GTEST_HAS_STREAM_REDIRECTION
+
+// Returns the size (in bytes) of a file.
+        GTEST_API_ size_t GetFileSize(FILE *file);
+
+// Reads the entire content of a file as a string.
+        GTEST_API_ std::string ReadEntireFile(FILE *file);
+
+// All command line arguments.
+        GTEST_API_ const ::std::vector<testing::internal::string> &GetArgvs();
+
+#if GTEST_HAS_DEATH_TEST
+
+        const ::std::vector<testing::internal::string> &GetInjectableArgvs();
+        void SetInjectableArgvs(const ::std::vector<testing::internal::string> *
+                                new_argvs);
+
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+// Defines synchronization primitives.
+#if GTEST_IS_THREADSAFE
+# if GTEST_HAS_PTHREAD
+// Sleeps for (roughly) n milliseconds.  This function is only for testing
+// Google Test's own constructs.  Don't use it in user tests, either
+// directly or indirectly.
+        inline void SleepMilliseconds(int n)
+        {
+            const timespec time = {
+                0,                  // 0 seconds.
+                n * 1000L * 1000L,  // And n ms.
+            };
+            nanosleep(&time, NULL);
+        }
+# endif  // GTEST_HAS_PTHREAD
+
+# if GTEST_HAS_NOTIFICATION_
+// Notification has already been imported into the namespace.
+// Nothing to do here.
+
+# elif GTEST_HAS_PTHREAD
+// Allows a controller thread to pause execution of newly created
+// threads until notified.  Instances of this class must be created
+// and destroyed in the controller thread.
+//
+// This class is only for testing Google Test's own constructs. Do not
+// use it in user tests, either directly or indirectly.
+        class Notification
+        {
+        public:
+            Notification() : notified_(false)
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, NULL));
+            }
+            ~Notification()
+            {
+                pthread_mutex_destroy(&mutex_);
+            }
+
+            // Notifies all threads created with this notification to start. Must
+            // be called from the controller thread.
+            void Notify()
+            {
+                pthread_mutex_lock(&mutex_);
+                notified_ = true;
+                pthread_mutex_unlock(&mutex_);
+            }
+
+            // Blocks until the controller thread notifies. Must be called from a test
+            // thread.
+            void WaitForNotification()
+            {
+                for (;;) {
+                    pthread_mutex_lock(&mutex_);
+                    const bool notified = notified_;
+                    pthread_mutex_unlock(&mutex_);
+                    if (notified)
+                        break;
+                    SleepMilliseconds(10);
+                }
+            }
+
+        private:
+            pthread_mutex_t mutex_;
+            bool notified_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification);
+        };
+
+# elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+
+        GTEST_API_ void SleepMilliseconds(int n);
+
+// Provides leak-safe Windows kernel handle ownership.
+// Used in death tests and in threading support.
+        class GTEST_API_ AutoHandle
+        {
+        public:
+            // Assume that Win32 HANDLE type is equivalent to void*. Doing so allows us to
+            // avoid including <windows.h> in this header file. Including <windows.h> is
+            // undesirable because it defines a lot of symbols and macros that tend to
+            // conflict with client code. This assumption is verified by
+            // WindowsTypesTest.HANDLEIsVoidStar.
+            typedef void *Handle;
+            AutoHandle();
+            explicit AutoHandle(Handle handle);
+
+            ~AutoHandle();
+
+            Handle Get() const;
+            void Reset();
+            void Reset(Handle handle);
+
+        private:
+            // Returns true iff the handle is a valid handle object that can be closed.
+            bool IsCloseable() const;
+
+            Handle handle_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(AutoHandle);
+        };
+
+// Allows a controller thread to pause execution of newly created
+// threads until notified.  Instances of this class must be created
+// and destroyed in the controller thread.
+//
+// This class is only for testing Google Test's own constructs. Do not
+// use it in user tests, either directly or indirectly.
+        class GTEST_API_ Notification
+        {
+        public:
+            Notification();
+            void Notify();
+            void WaitForNotification();
+
+        private:
+            AutoHandle event_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification);
+        };
+# endif  // GTEST_HAS_NOTIFICATION_
+
+// On MinGW, we can have both GTEST_OS_WINDOWS and GTEST_HAS_PTHREAD
+// defined, but we don't want to use MinGW's pthreads implementation, which
+// has conformance problems with some versions of the POSIX standard.
+# if GTEST_HAS_PTHREAD && !GTEST_OS_WINDOWS_MINGW
+
+// As a C-function, ThreadFuncWithCLinkage cannot be templated itself.
+// Consequently, it cannot select a correct instantiation of ThreadWithParam
+// in order to call its Run(). Introducing ThreadWithParamBase as a
+// non-templated base class for ThreadWithParam allows us to bypass this
+// problem.
+        class ThreadWithParamBase
+        {
+        public:
+            virtual ~ThreadWithParamBase() {}
+            virtual void Run() = 0;
+        };
+
+// pthread_create() accepts a pointer to a function type with the C linkage.
+// According to the Standard (7.5/1), function types with different linkages
+// are different even if they are otherwise identical.  Some compilers (for
+// example, SunStudio) treat them as different types.  Since class methods
+// cannot be defined with C-linkage we need to define a free C-function to
+// pass into pthread_create().
+        extern "C" inline void *ThreadFuncWithCLinkage(void *thread)
+        {
+            static_cast<ThreadWithParamBase *>(thread)->Run();
+            return NULL;
+        }
+
+// Helper class for testing Google Test's multi-threading constructs.
+// To use it, write:
+//
+//   void ThreadFunc(int param) { /* Do things with param */ }
+//   Notification thread_can_start;
+//   ...
+//   // The thread_can_start parameter is optional; you can supply NULL.
+//   ThreadWithParam<int> thread(&ThreadFunc, 5, &thread_can_start);
+//   thread_can_start.Notify();
+//
+// These classes are only for testing Google Test's own constructs. Do
+// not use them in user tests, either directly or indirectly.
+        template <typename T>
+        class ThreadWithParam : public ThreadWithParamBase
+        {
+        public:
+            typedef void UserThreadFunc(T);
+
+            ThreadWithParam(UserThreadFunc *func, T param, Notification *thread_can_start)
+                : func_(func),
+                  param_(param),
+                  thread_can_start_(thread_can_start),
+                  finished_(false)
+            {
+                ThreadWithParamBase *const base = this;
+                // The thread can be created only after all fields except thread_
+                // have been initialized.
+                GTEST_CHECK_POSIX_SUCCESS_(
+                    pthread_create(&thread_, 0, &ThreadFuncWithCLinkage, base));
+            }
+            ~ThreadWithParam()
+            {
+                Join();
+            }
+
+            void Join()
+            {
+                if (!finished_) {
+                    GTEST_CHECK_POSIX_SUCCESS_(pthread_join(thread_, 0));
+                    finished_ = true;
+                }
+            }
+
+            virtual void Run()
+            {
+                if (thread_can_start_ != NULL)
+                    thread_can_start_->WaitForNotification();
+                func_(param_);
+            }
+
+        private:
+            UserThreadFunc *const func_;  // User-supplied thread function.
+            const T param_;  // User-supplied parameter to the thread function.
+            // When non-NULL, used to block execution until the controller thread
+            // notifies.
+            Notification *const thread_can_start_;
+            bool finished_;  // true iff we know that the thread function has finished.
+            pthread_t thread_;  // The native thread object.
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam);
+        };
+# endif  // !GTEST_OS_WINDOWS && GTEST_HAS_PTHREAD ||
+        // GTEST_HAS_MUTEX_AND_THREAD_LOCAL_
+
+# if GTEST_HAS_MUTEX_AND_THREAD_LOCAL_
+// Mutex and ThreadLocal have already been imported into the namespace.
+// Nothing to do here.
+
+# elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+
+// Mutex implements mutex on Windows platforms.  It is used in conjunction
+// with class MutexLock:
+//
+//   Mutex mutex;
+//   ...
+//   MutexLock lock(&mutex);  // Acquires the mutex and releases it at the
+//                            // end of the current scope.
+//
+// A static Mutex *must* be defined or declared using one of the following
+// macros:
+//   GTEST_DEFINE_STATIC_MUTEX_(g_some_mutex);
+//   GTEST_DECLARE_STATIC_MUTEX_(g_some_mutex);
+//
+// (A non-static Mutex is defined/declared in the usual way).
+        class GTEST_API_ Mutex
+        {
+        public:
+            enum MutexType { kStatic = 0, kDynamic = 1 };
+            // We rely on kStaticMutex being 0 as it is to what the linker initializes
+            // type_ in static mutexes.  critical_section_ will be initialized lazily
+            // in ThreadSafeLazyInit().
+            enum StaticConstructorSelector { kStaticMutex = 0 };
+
+            // This constructor intentionally does nothing.  It relies on type_ being
+            // statically initialized to 0 (effectively setting it to kStatic) and on
+            // ThreadSafeLazyInit() to lazily initialize the rest of the members.
+            explicit Mutex(StaticConstructorSelector /*dummy*/) {}
+
+            Mutex();
+            ~Mutex();
+
+            void Lock();
+
+            void Unlock();
+
+            // Does nothing if the current thread holds the mutex. Otherwise, crashes
+            // with high probability.
+            void AssertHeld();
+
+        private:
+            // Initializes owner_thread_id_ and critical_section_ in static mutexes.
+            void ThreadSafeLazyInit();
+
+            // Per http://blogs.msdn.com/b/oldnewthing/archive/2004/02/23/78395.aspx,
+            // we assume that 0 is an invalid value for thread IDs.
+            unsigned int owner_thread_id_;
+
+            // For static mutexes, we rely on these members being initialized to zeros
+            // by the linker.
+            MutexType type_;
+            long critical_section_init_phase_;  // NOLINT
+            GTEST_CRITICAL_SECTION *critical_section_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex);
+        };
+
+# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+    extern ::testing::internal::Mutex mutex
+
+# define GTEST_DEFINE_STATIC_MUTEX_(mutex) \
+    ::testing::internal::Mutex mutex(::testing::internal::Mutex::kStaticMutex)
+
+// We cannot name this class MutexLock because the ctor declaration would
+// conflict with a macro named MutexLock, which is defined on some
+// platforms. That macro is used as a defensive measure to prevent against
+// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than
+// "MutexLock l(&mu)".  Hence the typedef trick below.
+        class GTestMutexLock
+        {
+        public:
+            explicit GTestMutexLock(Mutex *mutex)
+                : mutex_(mutex)
+            {
+                mutex_->Lock();
+            }
+
+            ~GTestMutexLock()
+            {
+                mutex_->Unlock();
+            }
+
+        private:
+            Mutex *const mutex_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock);
+        };
+
+        typedef GTestMutexLock MutexLock;
+
+// Base class for ValueHolder<T>.  Allows a caller to hold and delete a value
+// without knowing its type.
+        class ThreadLocalValueHolderBase
+        {
+        public:
+            virtual ~ThreadLocalValueHolderBase() {}
+        };
+
+// Provides a way for a thread to send notifications to a ThreadLocal
+// regardless of its parameter type.
+        class ThreadLocalBase
+        {
+        public:
+            // Creates a new ValueHolder<T> object holding a default value passed to
+            // this ThreadLocal<T>'s constructor and returns it.  It is the caller's
+            // responsibility not to call this when the ThreadLocal<T> instance already
+            // has a value on the current thread.
+            virtual ThreadLocalValueHolderBase *NewValueForCurrentThread() const = 0;
+
+        protected:
+            ThreadLocalBase() {}
+            virtual ~ThreadLocalBase() {}
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocalBase);
+        };
+
+// Maps a thread to a set of ThreadLocals that have values instantiated on that
+// thread and notifies them when the thread exits.  A ThreadLocal instance is
+// expected to persist until all threads it has values on have terminated.
+        class GTEST_API_ ThreadLocalRegistry
+        {
+        public:
+            // Registers thread_local_instance as having value on the current thread.
+            // Returns a value that can be used to identify the thread from other threads.
+            static ThreadLocalValueHolderBase *GetValueOnCurrentThread(
+                const ThreadLocalBase *thread_local_instance);
+
+            // Invoked when a ThreadLocal instance is destroyed.
+            static void OnThreadLocalDestroyed(
+                const ThreadLocalBase *thread_local_instance);
+        };
+
+        class GTEST_API_ ThreadWithParamBase
+        {
+        public:
+            void Join();
+
+        protected:
+            class Runnable
+            {
+            public:
+                virtual ~Runnable() {}
+                virtual void Run() = 0;
+            };
+
+            ThreadWithParamBase(Runnable *runnable, Notification *thread_can_start);
+            virtual ~ThreadWithParamBase();
+
+        private:
+            AutoHandle thread_;
+        };
+
+// Helper class for testing Google Test's multi-threading constructs.
+        template <typename T>
+        class ThreadWithParam : public ThreadWithParamBase
+        {
+        public:
+            typedef void UserThreadFunc(T);
+
+            ThreadWithParam(UserThreadFunc *func, T param, Notification *thread_can_start)
+                : ThreadWithParamBase(new RunnableImpl(func, param), thread_can_start)
+            {
+            }
+            virtual ~ThreadWithParam() {}
+
+        private:
+            class RunnableImpl : public Runnable
+            {
+            public:
+                RunnableImpl(UserThreadFunc *func, T param)
+                    : func_(func),
+                      param_(param)
+                {
+                }
+                virtual ~RunnableImpl() {}
+                virtual void Run()
+                {
+                    func_(param_);
+                }
+
+            private:
+                UserThreadFunc *const func_;
+                const T param_;
+
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(RunnableImpl);
+            };
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam);
+        };
+
+// Implements thread-local storage on Windows systems.
+//
+//   // Thread 1
+//   ThreadLocal<int> tl(100);  // 100 is the default value for each thread.
+//
+//   // Thread 2
+//   tl.set(150);  // Changes the value for thread 2 only.
+//   EXPECT_EQ(150, tl.get());
+//
+//   // Thread 1
+//   EXPECT_EQ(100, tl.get());  // In thread 1, tl has the original value.
+//   tl.set(200);
+//   EXPECT_EQ(200, tl.get());
+//
+// The template type argument T must have a public copy constructor.
+// In addition, the default ThreadLocal constructor requires T to have
+// a public default constructor.
+//
+// The users of a TheadLocal instance have to make sure that all but one
+// threads (including the main one) using that instance have exited before
+// destroying it. Otherwise, the per-thread objects managed for them by the
+// ThreadLocal instance are not guaranteed to be destroyed on all platforms.
+//
+// Google Test only uses global ThreadLocal objects.  That means they
+// will die after main() has returned.  Therefore, no per-thread
+// object managed by Google Test will be leaked as long as all threads
+// using Google Test have exited when main() returns.
+        template <typename T>
+        class ThreadLocal : public ThreadLocalBase
+        {
+        public:
+            ThreadLocal() : default_factory_(new DefaultValueHolderFactory()) {}
+            explicit ThreadLocal(const T &value)
+                : default_factory_(new InstanceValueHolderFactory(value)) {}
+
+            ~ThreadLocal()
+            {
+                ThreadLocalRegistry::OnThreadLocalDestroyed(this);
+            }
+
+            T *pointer()
+            {
+                return GetOrCreateValue();
+            }
+            const T *pointer() const
+            {
+                return GetOrCreateValue();
+            }
+            const T &get() const
+            {
+                return *pointer();
+            }
+            void set(const T &value)
+            {
+                *pointer() = value;
+            }
+
+        private:
+            // Holds a value of T.  Can be deleted via its base class without the caller
+            // knowing the type of T.
+            class ValueHolder : public ThreadLocalValueHolderBase
+            {
+            public:
+                ValueHolder() : value_() {}
+                explicit ValueHolder(const T &value) : value_(value) {}
+
+                T *pointer()
+                {
+                    return &value_;
+                }
+
+            private:
+                T value_;
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder);
+            };
+
+
+            T *GetOrCreateValue() const
+            {
+                return static_cast<ValueHolder *>(
+                           ThreadLocalRegistry::GetValueOnCurrentThread(this))->pointer();
+            }
+
+            virtual ThreadLocalValueHolderBase *NewValueForCurrentThread() const
+            {
+                return default_factory_->MakeNewHolder();
+            }
+
+            class ValueHolderFactory
+            {
+            public:
+                ValueHolderFactory() {}
+                virtual ~ValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const = 0;
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolderFactory);
+            };
+
+            class DefaultValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                DefaultValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder();
+                }
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultValueHolderFactory);
+            };
+
+            class InstanceValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                explicit InstanceValueHolderFactory(const T &value) : value_(value) {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder(value_);
+                }
+
+            private:
+                const T value_;  // The value for each thread.
+
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(InstanceValueHolderFactory);
+            };
+
+            scoped_ptr<ValueHolderFactory> default_factory_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal);
+        };
+
+# elif GTEST_HAS_PTHREAD
+
+// MutexBase and Mutex implement mutex on pthreads-based platforms.
+        class MutexBase
+        {
+        public:
+            // Acquires this mutex.
+            void Lock()
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_lock(&mutex_));
+                owner_ = pthread_self();
+                has_owner_ = true;
+            }
+
+            // Releases this mutex.
+            void Unlock()
+            {
+                // Since the lock is being released the owner_ field should no longer be
+                // considered valid. We don't protect writing to has_owner_ here, as it's
+                // the caller's responsibility to ensure that the current thread holds the
+                // mutex when this is called.
+                has_owner_ = false;
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_unlock(&mutex_));
+            }
+
+            // Does nothing if the current thread holds the mutex. Otherwise, crashes
+            // with high probability.
+            void AssertHeld() const
+            {
+                GTEST_CHECK_(has_owner_ && pthread_equal(owner_, pthread_self()))
+                        << "The current thread is not holding the mutex @" << this;
+            }
+
+            // A static mutex may be used before main() is entered.  It may even
+            // be used before the dynamic initialization stage.  Therefore we
+            // must be able to initialize a static mutex object at link time.
+            // This means MutexBase has to be a POD and its member variables
+            // have to be public.
+        public:
+            pthread_mutex_t mutex_;  // The underlying pthread mutex.
+            // has_owner_ indicates whether the owner_ field below contains a valid thread
+            // ID and is therefore safe to inspect (e.g., to use in pthread_equal()). All
+            // accesses to the owner_ field should be protected by a check of this field.
+            // An alternative might be to memset() owner_ to all zeros, but there's no
+            // guarantee that a zero'd pthread_t is necessarily invalid or even different
+            // from pthread_self().
+            bool has_owner_;
+            pthread_t owner_;  // The thread holding the mutex.
+        };
+
+// Forward-declares a static mutex.
+#  define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+     extern ::testing::internal::MutexBase mutex
+
+// Defines and statically (i.e. at link time) initializes a static mutex.
+#  define GTEST_DEFINE_STATIC_MUTEX_(mutex) \
+     ::testing::internal::MutexBase mutex = { PTHREAD_MUTEX_INITIALIZER, false, pthread_t() }
+
+// The Mutex class can only be used for mutexes created at runtime. It
+// shares its API with MutexBase otherwise.
+        class Mutex : public MutexBase
+        {
+        public:
+            Mutex()
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, NULL));
+                has_owner_ = false;
+            }
+            ~Mutex()
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_destroy(&mutex_));
+            }
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex);
+        };
+
+// We cannot name this class MutexLock because the ctor declaration would
+// conflict with a macro named MutexLock, which is defined on some
+// platforms. That macro is used as a defensive measure to prevent against
+// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than
+// "MutexLock l(&mu)".  Hence the typedef trick below.
+        class GTestMutexLock
+        {
+        public:
+            explicit GTestMutexLock(MutexBase *mutex)
+                : mutex_(mutex)
+            {
+                mutex_->Lock();
+            }
+
+            ~GTestMutexLock()
+            {
+                mutex_->Unlock();
+            }
+
+        private:
+            MutexBase *const mutex_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock);
+        };
+
+        typedef GTestMutexLock MutexLock;
+
+// Helpers for ThreadLocal.
+
+// pthread_key_create() requires DeleteThreadLocalValue() to have
+// C-linkage.  Therefore it cannot be templatized to access
+// ThreadLocal<T>.  Hence the need for class
+// ThreadLocalValueHolderBase.
+        class ThreadLocalValueHolderBase
+        {
+        public:
+            virtual ~ThreadLocalValueHolderBase() {}
+        };
+
+// Called by pthread to delete thread-local data stored by
+// pthread_setspecific().
+        extern "C" inline void DeleteThreadLocalValue(void *value_holder)
+        {
+            delete static_cast<ThreadLocalValueHolderBase *>(value_holder);
+        }
+
+// Implements thread-local storage on pthreads-based systems.
+        template <typename T>
+        class ThreadLocal
+        {
+        public:
+            ThreadLocal()
+                : key_(CreateKey()), default_factory_(new DefaultValueHolderFactory()) {}
+            explicit ThreadLocal(const T &value)
+                : key_(CreateKey()),
+                  default_factory_(new InstanceValueHolderFactory(value)) {}
+
+            ~ThreadLocal()
+            {
+                // Destroys the managed object for the current thread, if any.
+                DeleteThreadLocalValue(pthread_getspecific(key_));
+
+                // Releases resources associated with the key.  This will *not*
+                // delete managed objects for other threads.
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_key_delete(key_));
+            }
+
+            T *pointer()
+            {
+                return GetOrCreateValue();
+            }
+            const T *pointer() const
+            {
+                return GetOrCreateValue();
+            }
+            const T &get() const
+            {
+                return *pointer();
+            }
+            void set(const T &value)
+            {
+                *pointer() = value;
+            }
+
+        private:
+            // Holds a value of type T.
+            class ValueHolder : public ThreadLocalValueHolderBase
+            {
+            public:
+                ValueHolder() : value_() {}
+                explicit ValueHolder(const T &value) : value_(value) {}
+
+                T *pointer()
+                {
+                    return &value_;
+                }
+
+            private:
+                T value_;
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder);
+            };
+
+            static pthread_key_t CreateKey()
+            {
+                pthread_key_t key;
+                // When a thread exits, DeleteThreadLocalValue() will be called on
+                // the object managed for that thread.
+                GTEST_CHECK_POSIX_SUCCESS_(
+                    pthread_key_create(&key, &DeleteThreadLocalValue));
+                return key;
+            }
+
+            T *GetOrCreateValue() const
+            {
+                ThreadLocalValueHolderBase *const holder =
+                    static_cast<ThreadLocalValueHolderBase *>(pthread_getspecific(key_));
+                if (holder != NULL) {
+                    return CheckedDowncastToActualType<ValueHolder>(holder)->pointer();
+                }
+
+                ValueHolder *const new_holder = default_factory_->MakeNewHolder();
+                ThreadLocalValueHolderBase *const holder_base = new_holder;
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_setspecific(key_, holder_base));
+                return new_holder->pointer();
+            }
+
+            class ValueHolderFactory
+            {
+            public:
+                ValueHolderFactory() {}
+                virtual ~ValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const = 0;
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolderFactory);
+            };
+
+            class DefaultValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                DefaultValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder();
+                }
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultValueHolderFactory);
+            };
+
+            class InstanceValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                explicit InstanceValueHolderFactory(const T &value) : value_(value) {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder(value_);
+                }
+
+            private:
+                const T value_;  // The value for each thread.
+
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(InstanceValueHolderFactory);
+            };
+
+            // A key pthreads uses for looking up per-thread values.
+            const pthread_key_t key_;
+            scoped_ptr<ValueHolderFactory> default_factory_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal);
+        };
+
+# endif  // GTEST_HAS_MUTEX_AND_THREAD_LOCAL_
+
+#else  // GTEST_IS_THREADSAFE
+
+// A dummy implementation of synchronization primitives (mutex, lock,
+// and thread-local variable).  Necessary for compiling Google Test where
+// mutex is not supported - using Google Test in multiple threads is not
+// supported on such platforms.
+
+        class Mutex
+        {
+        public:
+            Mutex() {}
+            void Lock() {}
+            void Unlock() {}
+            void AssertHeld() const {}
+        };
+
+# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+  extern ::testing::internal::Mutex mutex
+
+# define GTEST_DEFINE_STATIC_MUTEX_(mutex) ::testing::internal::Mutex mutex
+
+// We cannot name this class MutexLock because the ctor declaration would
+// conflict with a macro named MutexLock, which is defined on some
+// platforms. That macro is used as a defensive measure to prevent against
+// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than
+// "MutexLock l(&mu)".  Hence the typedef trick below.
+        class GTestMutexLock
+        {
+        public:
+            explicit GTestMutexLock(Mutex *) {} // NOLINT
+        };
+
+        typedef GTestMutexLock MutexLock;
+
+        template <typename T>
+        class ThreadLocal
+        {
+        public:
+            ThreadLocal() : value_() {}
+            explicit ThreadLocal(const T &value) : value_(value) {}
+            T *pointer()
+            {
+                return &value_;
+            }
+            const T *pointer() const
+            {
+                return &value_;
+            }
+            const T &get() const
+            {
+                return value_;
+            }
+            void set(const T &value)
+            {
+                value_ = value;
+            }
+        private:
+            T value_;
+        };
+
+#endif  // GTEST_IS_THREADSAFE
+
+// Returns the number of threads running in the process, or 0 to indicate that
+// we cannot detect it.
+        GTEST_API_ size_t GetThreadCount();
+
+// Passing non-POD classes through ellipsis (...) crashes the ARM
+// compiler and generates a warning in Sun Studio.  The Nokia Symbian
+// and the IBM XL C/C++ compiler try to instantiate a copy constructor
+// for objects passed through ellipsis (...), failing for uncopyable
+// objects.  We define this to ensure that only POD is passed through
+// ellipsis on these systems.
+#if defined(__SYMBIAN32__) || defined(__IBMCPP__) || defined(__SUNPRO_CC)
+// We lose support for NULL detection where the compiler doesn't like
+// passing non-POD classes through ellipsis (...).
+# define GTEST_ELLIPSIS_NEEDS_POD_ 1
+#else
+# define GTEST_CAN_COMPARE_NULL 1
+#endif
+
+// The Nokia Symbian and IBM XL C/C++ compilers cannot decide between
+// const T& and const T* in a function template.  These compilers
+// _can_ decide between class template specializations for T and T*,
+// so a tr1::type_traits-like is_pointer works.
+#if defined(__SYMBIAN32__) || defined(__IBMCPP__)
+# define GTEST_NEEDS_IS_POINTER_ 1
+#endif
+
+        template <bool bool_value>
+        struct bool_constant {
+            typedef bool_constant<bool_value> type;
+            static const bool value = bool_value;
+        };
+        template <bool bool_value> const bool bool_constant<bool_value>::value;
+
+        typedef bool_constant<false> false_type;
+        typedef bool_constant<true> true_type;
+
+        template <typename T>
+        struct is_pointer : public false_type {};
+
+        template <typename T>
+        struct is_pointer<T *> : public true_type {};
+
+        template <typename Iterator>
+        struct IteratorTraits {
+            typedef typename Iterator::value_type value_type;
+        };
+
+        template <typename T>
+        struct IteratorTraits<T *> {
+            typedef T value_type;
+        };
+
+        template <typename T>
+        struct IteratorTraits<const T *> {
+            typedef T value_type;
+        };
+
+#if GTEST_OS_WINDOWS
+# define GTEST_PATH_SEP_ "\\"
+# define GTEST_HAS_ALT_PATH_SEP_ 1
+// The biggest signed integer type the compiler supports.
+        typedef __int64 BiggestInt;
+#else
+# define GTEST_PATH_SEP_ "/"
+# define GTEST_HAS_ALT_PATH_SEP_ 0
+        typedef long long BiggestInt;  // NOLINT
+#endif  // GTEST_OS_WINDOWS
+
+// Utilities for char.
+
+// isspace(int ch) and friends accept an unsigned char or EOF.  char
+// may be signed, depending on the compiler (or compiler flags).
+// Therefore we need to cast a char to unsigned char before calling
+// isspace(), etc.
+
+        inline bool IsAlpha(char ch)
+        {
+            return isalpha(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsAlNum(char ch)
+        {
+            return isalnum(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsDigit(char ch)
+        {
+            return isdigit(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsLower(char ch)
+        {
+            return islower(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsSpace(char ch)
+        {
+            return isspace(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsUpper(char ch)
+        {
+            return isupper(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsXDigit(char ch)
+        {
+            return isxdigit(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsXDigit(wchar_t ch)
+        {
+            const unsigned char low_byte = static_cast<unsigned char>(ch);
+            return ch == low_byte && isxdigit(low_byte) != 0;
+        }
+
+        inline char ToLower(char ch)
+        {
+            return static_cast<char>(tolower(static_cast<unsigned char>(ch)));
+        }
+        inline char ToUpper(char ch)
+        {
+            return static_cast<char>(toupper(static_cast<unsigned char>(ch)));
+        }
+
+        inline std::string StripTrailingSpaces(std::string str)
+        {
+            std::string::iterator it = str.end();
+            while (it != str.begin() && IsSpace(*--it))
+                it = str.erase(it);
+            return str;
+        }
+
+// The testing::internal::posix namespace holds wrappers for common
+// POSIX functions.  These wrappers hide the differences between
+// Windows/MSVC and POSIX systems.  Since some compilers define these
+// standard functions as macros, the wrapper cannot have the same name
+// as the wrapped function.
+
+        namespace posix
+        {
+
+// Functions with a different name on Windows.
+
+#if GTEST_OS_WINDOWS
+
+            typedef struct _stat StatStruct;
+
+# ifdef __BORLANDC__
+            inline int IsATTY(int fd)
+            {
+                return isatty(fd);
+            }
+            inline int StrCaseCmp(const char *s1, const char *s2)
+            {
+                return stricmp(s1, s2);
+            }
+            inline char *StrDup(const char *src)
+            {
+                return strdup(src);
+            }
+# else  // !__BORLANDC__
+#  if GTEST_OS_WINDOWS_MOBILE
+            inline int IsATTY(int /* fd */)
+            {
+                return 0;
+            }
+#  else
+            inline int IsATTY(int fd)
+            {
+                return _isatty(fd);
+            }
+#  endif  // GTEST_OS_WINDOWS_MOBILE
+            inline int StrCaseCmp(const char *s1, const char *s2)
+            {
+                return _stricmp(s1, s2);
+            }
+            inline char *StrDup(const char *src)
+            {
+                return _strdup(src);
+            }
+# endif  // __BORLANDC__
+
+# if GTEST_OS_WINDOWS_MOBILE
+            inline int FileNo(FILE *file)
+            {
+                return reinterpret_cast<int>(_fileno(file));
+            }
+// Stat(), RmDir(), and IsDir() are not needed on Windows CE at this
+// time and thus not defined there.
+# else
+            inline int FileNo(FILE *file)
+            {
+                return _fileno(file);
+            }
+            inline int Stat(const char *path, StatStruct *buf)
+            {
+                return _stat(path, buf);
+            }
+            inline int RmDir(const char *dir)
+            {
+                return _rmdir(dir);
+            }
+            inline bool IsDir(const StatStruct &st)
+            {
+                return (_S_IFDIR & st.st_mode) != 0;
+            }
+# endif  // GTEST_OS_WINDOWS_MOBILE
+
+#else
+
+            typedef struct stat StatStruct;
+
+            inline int FileNo(FILE *file)
+            {
+                return fileno(file);
+            }
+            inline int IsATTY(int fd)
+            {
+                return isatty(fd);
+            }
+            inline int Stat(const char *path, StatStruct *buf)
+            {
+                return stat(path, buf);
+            }
+            inline int StrCaseCmp(const char *s1, const char *s2)
+            {
+                return strcasecmp(s1, s2);
+            }
+            inline char *StrDup(const char *src)
+            {
+                return strdup(src);
+            }
+            inline int RmDir(const char *dir)
+            {
+                return rmdir(dir);
+            }
+            inline bool IsDir(const StatStruct &st)
+            {
+                return S_ISDIR(st.st_mode);
+            }
+
+#endif  // GTEST_OS_WINDOWS
+
+// Functions deprecated by MSVC 8.0.
+
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996 /* deprecated function */)
+
+            inline const char *StrNCpy(char *dest, const char *src, size_t n)
+            {
+                return strncpy(dest, src, n);
+            }
+
+// ChDir(), FReopen(), FDOpen(), Read(), Write(), Close(), and
+// StrError() aren't needed on Windows CE at this time and thus not
+// defined there.
+
+#if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+            inline int ChDir(const char *dir)
+            {
+                return chdir(dir);
+            }
+#endif
+            inline FILE *FOpen(const char *path, const char *mode)
+            {
+                return fopen(path, mode);
+            }
+#if !GTEST_OS_WINDOWS_MOBILE
+            inline FILE *FReopen(const char *path, const char *mode, FILE *stream)
+            {
+                return freopen(path, mode, stream);
+            }
+            inline FILE *FDOpen(int fd, const char *mode)
+            {
+                return fdopen(fd, mode);
+            }
+#endif
+            inline int FClose(FILE *fp)
+            {
+                return fclose(fp);
+            }
+#if !GTEST_OS_WINDOWS_MOBILE
+            inline int Read(int fd, void *buf, unsigned int count)
+            {
+                return static_cast<int>(read(fd, buf, count));
+            }
+            inline int Write(int fd, const void *buf, unsigned int count)
+            {
+                return static_cast<int>(write(fd, buf, count));
+            }
+            inline int Close(int fd)
+            {
+                return close(fd);
+            }
+            inline const char *StrError(int errnum)
+            {
+                return strerror(errnum);
+            }
+#endif
+            inline const char *GetEnv(const char *name)
+            {
+#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE | GTEST_OS_WINDOWS_RT
+                // We are on Windows CE, which has no environment variables.
+                static_cast<void>(name);  // To prevent 'unused argument' warning.
+                return NULL;
+#elif defined(__BORLANDC__) || defined(__SunOS_5_8) || defined(__SunOS_5_9)
+                // Environment variables which we programmatically clear will be set to the
+                // empty string rather than unset (NULL).  Handle that case.
+                const char *const env = getenv(name);
+                return (env != NULL && env[0] != '\0') ? env : NULL;
+#else
+                return getenv(name);
+#endif
+            }
+
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+#if GTEST_OS_WINDOWS_MOBILE
+// Windows CE has no C library. The abort() function is used in
+// several places in Google Test. This implementation provides a reasonable
+// imitation of standard behaviour.
+            void Abort();
+#else
+            inline void Abort()
+            {
+                abort();
+            }
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+        }  // namespace posix
+
+// MSVC "deprecates" snprintf and issues warnings wherever it is used.  In
+// order to avoid these warnings, we need to use _snprintf or _snprintf_s on
+// MSVC-based platforms.  We map the GTEST_SNPRINTF_ macro to the appropriate
+// function in order to achieve that.  We use macro definition here because
+// snprintf is a variadic function.
+#if _MSC_VER >= 1400 && !GTEST_OS_WINDOWS_MOBILE
+// MSVC 2005 and above support variadic macros.
+# define GTEST_SNPRINTF_(buffer, size, format, ...) \
+     _snprintf_s(buffer, size, size, format, __VA_ARGS__)
+#elif defined(_MSC_VER)
+// Windows CE does not define _snprintf_s and MSVC prior to 2005 doesn't
+// complain about _snprintf.
+# define GTEST_SNPRINTF_ _snprintf
+#else
+# define GTEST_SNPRINTF_ snprintf
+#endif
+
+// The maximum number a BiggestInt can represent.  This definition
+// works no matter BiggestInt is represented in one's complement or
+// two's complement.
+//
+// We cannot rely on numeric_limits in STL, as __int64 and long long
+// are not part of standard C++ and numeric_limits doesn't need to be
+// defined for them.
+        const BiggestInt kMaxBiggestInt =
+            ~(static_cast<BiggestInt>(1) << (8*sizeof(BiggestInt) - 1));
+
+// This template class serves as a compile-time function from size to
+// type.  It maps a size in bytes to a primitive type with that
+// size. e.g.
+//
+//   TypeWithSize<4>::UInt
+//
+// is typedef-ed to be unsigned int (unsigned integer made up of 4
+// bytes).
+//
+// Such functionality should belong to STL, but I cannot find it
+// there.
+//
+// Google Test uses this class in the implementation of floating-point
+// comparison.
+//
+// For now it only handles UInt (unsigned int) as that's all Google Test
+// needs.  Other types can be easily added in the future if need
+// arises.
+        template <size_t size>
+        class TypeWithSize
+        {
+        public:
+            // This prevents the user from using TypeWithSize<N> with incorrect
+            // values of N.
+            typedef void UInt;
+        };
+
+// The specialization for size 4.
+        template <>
+        class TypeWithSize<4>
+        {
+        public:
+            // unsigned int has size 4 in both gcc and MSVC.
+            //
+            // As base/basictypes.h doesn't compile on Windows, we cannot use
+            // uint32, uint64, and etc here.
+            typedef int Int;
+            typedef unsigned int UInt;
+        };
+
+// The specialization for size 8.
+        template <>
+        class TypeWithSize<8>
+        {
+        public:
+#if GTEST_OS_WINDOWS
+            typedef __int64 Int;
+            typedef unsigned __int64 UInt;
+#else
+            typedef long long Int;  // NOLINT
+            typedef unsigned long long UInt;  // NOLINT
+#endif  // GTEST_OS_WINDOWS
+        };
+
+// Integer types of known sizes.
+        typedef TypeWithSize<4>::Int Int32;
+        typedef TypeWithSize<4>::UInt UInt32;
+        typedef TypeWithSize<8>::Int Int64;
+        typedef TypeWithSize<8>::UInt UInt64;
+        typedef TypeWithSize<8>::Int TimeInMillis;  // Represents time in milliseconds.
+
+// Utilities for command line flags and environment variables.
+
+// Macro for referencing flags.
+#if !defined(GTEST_FLAG)
+# define GTEST_FLAG(name) FLAGS_gtest_##name
+#endif  // !defined(GTEST_FLAG)
+
+#if !defined(GTEST_USE_OWN_FLAGFILE_FLAG_)
+# define GTEST_USE_OWN_FLAGFILE_FLAG_ 1
+#endif  // !defined(GTEST_USE_OWN_FLAGFILE_FLAG_)
+
+#if !defined(GTEST_DECLARE_bool_)
+# define GTEST_FLAG_SAVER_ ::testing::internal::GTestFlagSaver
+
+// Macros for declaring flags.
+# define GTEST_DECLARE_bool_(name) GTEST_API_ extern bool GTEST_FLAG(name)
+# define GTEST_DECLARE_int32_(name) \
+    GTEST_API_ extern ::testing::internal::Int32 GTEST_FLAG(name)
+#define GTEST_DECLARE_string_(name) \
+    GTEST_API_ extern ::std::string GTEST_FLAG(name)
+
+// Macros for defining flags.
+#define GTEST_DEFINE_bool_(name, default_val, doc) \
+    GTEST_API_ bool GTEST_FLAG(name) = (default_val)
+#define GTEST_DEFINE_int32_(name, default_val, doc) \
+    GTEST_API_ ::testing::internal::Int32 GTEST_FLAG(name) = (default_val)
+#define GTEST_DEFINE_string_(name, default_val, doc) \
+    GTEST_API_ ::std::string GTEST_FLAG(name) = (default_val)
+
+#endif  // !defined(GTEST_DECLARE_bool_)
+
+// Thread annotations
+#if !defined(GTEST_EXCLUSIVE_LOCK_REQUIRED_)
+# define GTEST_EXCLUSIVE_LOCK_REQUIRED_(locks)
+# define GTEST_LOCK_EXCLUDED_(locks)
+#endif  // !defined(GTEST_EXCLUSIVE_LOCK_REQUIRED_)
+
+// Parses 'str' for a 32-bit signed integer.  If successful, writes the result
+// to *value and returns true; otherwise leaves *value unchanged and returns
+// false.
+// TODO(chandlerc): Find a better way to refactor flag and environment parsing
+// out of both gtest-port.cc and gtest.cc to avoid exporting this utility
+// function.
+        bool ParseInt32(const Message &src_text, const char *str, Int32 *value);
+
+// Parses a bool/Int32/string from the environment variable
+// corresponding to the given Google Test flag.
+        bool BoolFromGTestEnv(const char *flag, bool default_val);
+        GTEST_API_ Int32 Int32FromGTestEnv(const char *flag, Int32 default_val);
+        std::string StringFromGTestEnv(const char *flag, const char *default_val);
+
+    }  // namespace internal
+
+// Returns a path to temporary directory.
+// Tries to determine an appropriate directory for the platform.
+    GTEST_API_ std::string TempDir();
+
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+
+#if GTEST_OS_LINUX
+# include <stdlib.h>
+# include <sys/types.h>
+# include <sys/wait.h>
+# include <unistd.h>
+#endif  // GTEST_OS_LINUX
+
+#if GTEST_HAS_EXCEPTIONS
+# include <stdexcept>
+#endif
+
+#include <ctype.h>
+#include <float.h>
+#include <string.h>
+#include <iomanip>
+#include <limits>
+#include <map>
+#include <set>
+#include <string>
+#include <vector>
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the Message class.
+//
+// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
+// leave some internal implementation details in this header file.
+// They are clearly marked by comments like this:
+//
+//   // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+//
+// Such code is NOT meant to be used by a user directly, and is subject
+// to CHANGE WITHOUT NOTICE.  Therefore DO NOT DEPEND ON IT in a user
+// program!
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+#define GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+
+#include <limits>
+
+
+// Ensures that there is at least one operator<< in the global namespace.
+// See Message& operator<<(...) below for why.
+void operator<<(const testing::internal::Secret &, int);
+
+namespace testing
+{
+
+// The Message class works like an ostream repeater.
+//
+// Typical usage:
+//
+//   1. You stream a bunch of values to a Message object.
+//      It will remember the text in a stringstream.
+//   2. Then you stream the Message object to an ostream.
+//      This causes the text in the Message to be streamed
+//      to the ostream.
+//
+// For example;
+//
+//   testing::Message foo;
+//   foo << 1 << " != " << 2;
+//   std::cout << foo;
+//
+// will print "1 != 2".
+//
+// Message is not intended to be inherited from.  In particular, its
+// destructor is not virtual.
+//
+// Note that stringstream behaves differently in gcc and in MSVC.  You
+// can stream a NULL char pointer to it in the former, but not in the
+// latter (it causes an access violation if you do).  The Message
+// class hides this difference by treating a NULL char pointer as
+// "(null)".
+    class GTEST_API_ Message
+    {
+    private:
+        // The type of basic IO manipulators (endl, ends, and flush) for
+        // narrow streams.
+        typedef std::ostream &(*BasicNarrowIoManip)(std::ostream &);
+
+    public:
+        // Constructs an empty Message.
+        Message();
+
+        // Copy constructor.
+        Message(const Message &msg) : ss_(new ::std::stringstream)    // NOLINT
+        {
+            *ss_ << msg.GetString();
+        }
+
+        // Constructs a Message from a C-string.
+        explicit Message(const char *str) : ss_(new ::std::stringstream)
+        {
+            *ss_ << str;
+        }
+
+#if GTEST_OS_SYMBIAN
+        // Streams a value (either a pointer or not) to this object.
+        template <typename T>
+        inline Message &operator <<(const T &value)
+        {
+            StreamHelper(typename internal::is_pointer<T>::type(), value);
+            return *this;
+        }
+#else
+        // Streams a non-pointer value to this object.
+        template <typename T>
+        inline Message &operator <<(const T &val)
+        {
+            // Some libraries overload << for STL containers.  These
+            // overloads are defined in the global namespace instead of ::std.
+            //
+            // C++'s symbol lookup rule (i.e. Koenig lookup) says that these
+            // overloads are visible in either the std namespace or the global
+            // namespace, but not other namespaces, including the testing
+            // namespace which Google Test's Message class is in.
+            //
+            // To allow STL containers (and other types that has a << operator
+            // defined in the global namespace) to be used in Google Test
+            // assertions, testing::Message must access the custom << operator
+            // from the global namespace.  With this using declaration,
+            // overloads of << defined in the global namespace and those
+            // visible via Koenig lookup are both exposed in this function.
+            using ::operator <<;
+            *ss_ << val;
+            return *this;
+        }
+
+        // Streams a pointer value to this object.
+        //
+        // This function is an overload of the previous one.  When you
+        // stream a pointer to a Message, this definition will be used as it
+        // is more specialized.  (The C++ Standard, section
+        // [temp.func.order].)  If you stream a non-pointer, then the
+        // previous definition will be used.
+        //
+        // The reason for this overload is that streaming a NULL pointer to
+        // ostream is undefined behavior.  Depending on the compiler, you
+        // may get "0", "(nil)", "(null)", or an access violation.  To
+        // ensure consistent result across compilers, we always treat NULL
+        // as "(null)".
+        template <typename T>
+        inline Message &operator <<(T *const &pointer)    // NOLINT
+        {
+            if (pointer == NULL) {
+                *ss_ << "(null)";
+            } else {
+                *ss_ << pointer;
+            }
+            return *this;
+        }
+#endif  // GTEST_OS_SYMBIAN
+
+        // Since the basic IO manipulators are overloaded for both narrow
+        // and wide streams, we have to provide this specialized definition
+        // of operator <<, even though its body is the same as the
+        // templatized version above.  Without this definition, streaming
+        // endl or other basic IO manipulators to Message will confuse the
+        // compiler.
+        Message &operator <<(BasicNarrowIoManip val)
+        {
+            *ss_ << val;
+            return *this;
+        }
+
+        // Instead of 1/0, we want to see true/false for bool values.
+        Message &operator <<(bool b)
+        {
+            return *this << (b ? "true" : "false");
+        }
+
+        // These two overloads allow streaming a wide C string to a Message
+        // using the UTF-8 encoding.
+        Message &operator <<(const wchar_t *wide_c_str);
+        Message &operator <<(wchar_t *wide_c_str);
+
+#if GTEST_HAS_STD_WSTRING
+        // Converts the given wide string to a narrow string using the UTF-8
+        // encoding, and streams the result to this Message object.
+        Message &operator <<(const ::std::wstring &wstr);
+#endif  // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+        // Converts the given wide string to a narrow string using the UTF-8
+        // encoding, and streams the result to this Message object.
+        Message &operator <<(const ::wstring &wstr);
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+        // Gets the text streamed to this object so far as an std::string.
+        // Each '\0' character in the buffer is replaced with "\\0".
+        //
+        // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        std::string GetString() const;
+
+    private:
+
+#if GTEST_OS_SYMBIAN
+        // These are needed as the Nokia Symbian Compiler cannot decide between
+        // const T& and const T* in a function template. The Nokia compiler _can_
+        // decide between class template specializations for T and T*, so a
+        // tr1::type_traits-like is_pointer works, and we can overload on that.
+        template <typename T>
+        inline void StreamHelper(internal::true_type /*is_pointer*/, T *pointer)
+        {
+            if (pointer == NULL) {
+                *ss_ << "(null)";
+            } else {
+                *ss_ << pointer;
+            }
+        }
+        template <typename T>
+        inline void StreamHelper(internal::false_type /*is_pointer*/,
+                                 const T &value)
+        {
+            // See the comments in Message& operator <<(const T&) above for why
+            // we need this using statement.
+            using ::operator <<;
+            *ss_ << value;
+        }
+#endif  // GTEST_OS_SYMBIAN
+
+        // We'll hold the text streamed to this object here.
+        const internal::scoped_ptr< ::std::stringstream> ss_;
+
+        // We declare (but don't implement) this to prevent the compiler
+        // from implementing the assignment operator.
+        void operator=(const Message &);
+    };
+
+// Streams a Message to an ostream.
+    inline std::ostream &operator <<(std::ostream &os, const Message &sb)
+    {
+        return os << sb.GetString();
+    }
+
+    namespace internal
+    {
+
+// Converts a streamable value to an std::string.  A NULL pointer is
+// converted to "(null)".  When the input value is a ::string,
+// ::std::string, ::wstring, or ::std::wstring object, each NUL
+// character in it is replaced with "\\0".
+        template <typename T>
+        std::string StreamableToString(const T &streamable)
+        {
+            return (Message() << streamable).GetString();
+        }
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file declares the String class and functions used internally by
+// Google Test.  They are subject to change without notice. They should not used
+// by code external to Google Test.
+//
+// This header file is #included by <gtest/internal/gtest-internal.h>.
+// It should not be #included by other files.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+
+#ifdef __BORLANDC__
+// string.h is not guaranteed to provide strcpy on C++ Builder.
+# include <mem.h>
+#endif
+
+#include <string.h>
+#include <string>
+
+
+namespace testing
+{
+    namespace internal
+    {
+
+// String - an abstract class holding static string utilities.
+        class GTEST_API_ String
+        {
+        public:
+            // Static utility methods
+
+            // Clones a 0-terminated C string, allocating memory using new.  The
+            // caller is responsible for deleting the return value using
+            // delete[].  Returns the cloned string, or NULL if the input is
+            // NULL.
+            //
+            // This is different from strdup() in string.h, which allocates
+            // memory using malloc().
+            static const char *CloneCString(const char *c_str);
+
+#if GTEST_OS_WINDOWS_MOBILE
+            // Windows CE does not have the 'ANSI' versions of Win32 APIs. To be
+            // able to pass strings to Win32 APIs on CE we need to convert them
+            // to 'Unicode', UTF-16.
+
+            // Creates a UTF-16 wide string from the given ANSI string, allocating
+            // memory using new. The caller is responsible for deleting the return
+            // value using delete[]. Returns the wide string, or NULL if the
+            // input is NULL.
+            //
+            // The wide string is created using the ANSI codepage (CP_ACP) to
+            // match the behaviour of the ANSI versions of Win32 calls and the
+            // C runtime.
+            static LPCWSTR AnsiToUtf16(const char *c_str);
+
+            // Creates an ANSI string from the given wide string, allocating
+            // memory using new. The caller is responsible for deleting the return
+            // value using delete[]. Returns the ANSI string, or NULL if the
+            // input is NULL.
+            //
+            // The returned string is created using the ANSI codepage (CP_ACP) to
+            // match the behaviour of the ANSI versions of Win32 calls and the
+            // C runtime.
+            static const char *Utf16ToAnsi(LPCWSTR utf16_str);
+#endif
+
+            // Compares two C strings.  Returns true iff they have the same content.
+            //
+            // Unlike strcmp(), this function can handle NULL argument(s).  A
+            // NULL C string is considered different to any non-NULL C string,
+            // including the empty string.
+            static bool CStringEquals(const char *lhs, const char *rhs);
+
+            // Converts a wide C string to a String using the UTF-8 encoding.
+            // NULL will be converted to "(null)".  If an error occurred during
+            // the conversion, "(failed to convert from wide string)" is
+            // returned.
+            static std::string ShowWideCString(const wchar_t *wide_c_str);
+
+            // Compares two wide C strings.  Returns true iff they have the same
+            // content.
+            //
+            // Unlike wcscmp(), this function can handle NULL argument(s).  A
+            // NULL C string is considered different to any non-NULL C string,
+            // including the empty string.
+            static bool WideCStringEquals(const wchar_t *lhs, const wchar_t *rhs);
+
+            // Compares two C strings, ignoring case.  Returns true iff they
+            // have the same content.
+            //
+            // Unlike strcasecmp(), this function can handle NULL argument(s).
+            // A NULL C string is considered different to any non-NULL C string,
+            // including the empty string.
+            static bool CaseInsensitiveCStringEquals(const char *lhs,
+                                                     const char *rhs);
+
+            // Compares two wide C strings, ignoring case.  Returns true iff they
+            // have the same content.
+            //
+            // Unlike wcscasecmp(), this function can handle NULL argument(s).
+            // A NULL C string is considered different to any non-NULL wide C string,
+            // including the empty string.
+            // NB: The implementations on different platforms slightly differ.
+            // On windows, this method uses _wcsicmp which compares according to LC_CTYPE
+            // environment variable. On GNU platform this method uses wcscasecmp
+            // which compares according to LC_CTYPE category of the current locale.
+            // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the
+            // current locale.
+            static bool CaseInsensitiveWideCStringEquals(const wchar_t *lhs,
+                                                         const wchar_t *rhs);
+
+            // Returns true iff the given string ends with the given suffix, ignoring
+            // case. Any string is considered to end with an empty suffix.
+            static bool EndsWithCaseInsensitive(
+                const std::string &str, const std::string &suffix);
+
+            // Formats an int value as "%02d".
+            static std::string FormatIntWidth2(int value);  // "%02d" for width == 2
+
+            // Formats an int value as "%X".
+            static std::string FormatHexInt(int value);
+
+            // Formats a byte as "%02X".
+            static std::string FormatByte(unsigned char value);
+
+        private:
+            String();  // Not meant to be instantiated.
+        };  // class String
+
+// Gets the content of the stringstream's buffer as an std::string.  Each '\0'
+// character in the buffer is replaced with "\\0".
+        GTEST_API_ std::string StringStreamToString(::std::stringstream *stream);
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: keith.ray@gmail.com (Keith Ray)
+//
+// Google Test filepath utilities
+//
+// This header file declares classes and functions used internally by
+// Google Test.  They are subject to change without notice.
+//
+// This file is #included in <gtest/internal/gtest-internal.h>.
+// Do not include this header file separately!
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+
+
+namespace testing
+{
+    namespace internal
+    {
+
+// FilePath - a class for file and directory pathname manipulation which
+// handles platform-specific conventions (like the pathname separator).
+// Used for helper functions for naming files in a directory for xml output.
+// Except for Set methods, all methods are const or static, which provides an
+// "immutable value object" -- useful for peace of mind.
+// A FilePath with a value ending in a path separator ("like/this/") represents
+// a directory, otherwise it is assumed to represent a file. In either case,
+// it may or may not represent an actual file or directory in the file system.
+// Names are NOT checked for syntax correctness -- no checking for illegal
+// characters, malformed paths, etc.
+
+        class GTEST_API_ FilePath
+        {
+        public:
+            FilePath() : pathname_("") { }
+            FilePath(const FilePath &rhs) : pathname_(rhs.pathname_) { }
+
+            explicit FilePath(const std::string &pathname) : pathname_(pathname)
+            {
+                Normalize();
+            }
+
+            FilePath &operator=(const FilePath &rhs)
+            {
+                Set(rhs);
+                return *this;
+            }
+
+            void Set(const FilePath &rhs)
+            {
+                pathname_ = rhs.pathname_;
+            }
+
+            const std::string &string() const
+            {
+                return pathname_;
+            }
+            const char *c_str() const
+            {
+                return pathname_.c_str();
+            }
+
+            // Returns the current working directory, or "" if unsuccessful.
+            static FilePath GetCurrentDir();
+
+            // Given directory = "dir", base_name = "test", number = 0,
+            // extension = "xml", returns "dir/test.xml". If number is greater
+            // than zero (e.g., 12), returns "dir/test_12.xml".
+            // On Windows platform, uses \ as the separator rather than /.
+            static FilePath MakeFileName(const FilePath &directory,
+                                         const FilePath &base_name,
+                                         int number,
+                                         const char *extension);
+
+            // Given directory = "dir", relative_path = "test.xml",
+            // returns "dir/test.xml".
+            // On Windows, uses \ as the separator rather than /.
+            static FilePath ConcatPaths(const FilePath &directory,
+                                        const FilePath &relative_path);
+
+            // Returns a pathname for a file that does not currently exist. The pathname
+            // will be directory/base_name.extension or
+            // directory/base_name_<number>.extension if directory/base_name.extension
+            // already exists. The number will be incremented until a pathname is found
+            // that does not already exist.
+            // Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'.
+            // There could be a race condition if two or more processes are calling this
+            // function at the same time -- they could both pick the same filename.
+            static FilePath GenerateUniqueFileName(const FilePath &directory,
+                                                   const FilePath &base_name,
+                                                   const char *extension);
+
+            // Returns true iff the path is "".
+            bool IsEmpty() const
+            {
+                return pathname_.empty();
+            }
+
+            // If input name has a trailing separator character, removes it and returns
+            // the name, otherwise return the name string unmodified.
+            // On Windows platform, uses \ as the separator, other platforms use /.
+            FilePath RemoveTrailingPathSeparator() const;
+
+            // Returns a copy of the FilePath with the directory part removed.
+            // Example: FilePath("path/to/file").RemoveDirectoryName() returns
+            // FilePath("file"). If there is no directory part ("just_a_file"), it returns
+            // the FilePath unmodified. If there is no file part ("just_a_dir/") it
+            // returns an empty FilePath ("").
+            // On Windows platform, '\' is the path separator, otherwise it is '/'.
+            FilePath RemoveDirectoryName() const;
+
+            // RemoveFileName returns the directory path with the filename removed.
+            // Example: FilePath("path/to/file").RemoveFileName() returns "path/to/".
+            // If the FilePath is "a_file" or "/a_file", RemoveFileName returns
+            // FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does
+            // not have a file, like "just/a/dir/", it returns the FilePath unmodified.
+            // On Windows platform, '\' is the path separator, otherwise it is '/'.
+            FilePath RemoveFileName() const;
+
+            // Returns a copy of the FilePath with the case-insensitive extension removed.
+            // Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns
+            // FilePath("dir/file"). If a case-insensitive extension is not
+            // found, returns a copy of the original FilePath.
+            FilePath RemoveExtension(const char *extension) const;
+
+            // Creates directories so that path exists. Returns true if successful or if
+            // the directories already exist; returns false if unable to create
+            // directories for any reason. Will also return false if the FilePath does
+            // not represent a directory (that is, it doesn't end with a path separator).
+            bool CreateDirectoriesRecursively() const;
+
+            // Create the directory so that path exists. Returns true if successful or
+            // if the directory already exists; returns false if unable to create the
+            // directory for any reason, including if the parent directory does not
+            // exist. Not named "CreateDirectory" because that's a macro on Windows.
+            bool CreateFolder() const;
+
+            // Returns true if FilePath describes something in the file-system,
+            // either a file, directory, or whatever, and that something exists.
+            bool FileOrDirectoryExists() const;
+
+            // Returns true if pathname describes a directory in the file-system
+            // that exists.
+            bool DirectoryExists() const;
+
+            // Returns true if FilePath ends with a path separator, which indicates that
+            // it is intended to represent a directory. Returns false otherwise.
+            // This does NOT check that a directory (or file) actually exists.
+            bool IsDirectory() const;
+
+            // Returns true if pathname describes a root directory. (Windows has one
+            // root directory per disk drive.)
+            bool IsRootDirectory() const;
+
+            // Returns true if pathname describes an absolute path.
+            bool IsAbsolutePath() const;
+
+        private:
+            // Replaces multiple consecutive separators with a single separator.
+            // For example, "bar///foo" becomes "bar/foo". Does not eliminate other
+            // redundancies that might be in a pathname involving "." or "..".
+            //
+            // A pathname with multiple consecutive separators may occur either through
+            // user error or as a result of some scripts or APIs that generate a pathname
+            // with a trailing separator. On other platforms the same API or script
+            // may NOT generate a pathname with a trailing "/". Then elsewhere that
+            // pathname may have another "/" and pathname components added to it,
+            // without checking for the separator already being there.
+            // The script language and operating system may allow paths like "foo//bar"
+            // but some of the functions in FilePath will not handle that correctly. In
+            // particular, RemoveTrailingPathSeparator() only removes one separator, and
+            // it is called in CreateDirectoriesRecursively() assuming that it will change
+            // a pathname from directory syntax (trailing separator) to filename syntax.
+            //
+            // On Windows this method also replaces the alternate path separator '/' with
+            // the primary path separator '\\', so that for example "bar\\/\\foo" becomes
+            // "bar\\foo".
+
+            void Normalize();
+
+            // Returns a pointer to the last occurrence of a valid path separator in
+            // the FilePath. On Windows, for example, both '/' and '\' are valid path
+            // separators. Returns NULL if no path separator was found.
+            const char *FindLastPathSeparator() const;
+
+            std::string pathname_;
+        };  // class FilePath
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+// This file was GENERATED by command:
+//     pump.py gtest-type-util.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Type utilities needed for implementing typed and type-parameterized
+// tests.  This file is generated by a SCRIPT.  DO NOT EDIT BY HAND!
+//
+// Currently we support at most 50 types in a list, and at most 50
+// type-parameterized tests in one type-parameterized test case.
+// Please contact googletestframework@googlegroups.com if you need
+// more.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+
+
+// #ifdef __GNUC__ is too general here.  It is possible to use gcc without using
+// libstdc++ (which is where cxxabi.h comes from).
+# if GTEST_HAS_CXXABI_H_
+#  include <cxxabi.h>
+# elif defined(__HP_aCC)
+#  include <acxx_demangle.h>
+# endif  // GTEST_HASH_CXXABI_H_
+
+namespace testing
+{
+    namespace internal
+    {
+
+// GetTypeName<T>() returns a human-readable name of type T.
+// NB: This function is also used in Google Mock, so don't move it inside of
+// the typed-test-only section below.
+        template <typename T>
+        std::string GetTypeName()
+        {
+# if GTEST_HAS_RTTI
+
+            const char *const name = typeid(T).name();
+#  if GTEST_HAS_CXXABI_H_ || defined(__HP_aCC)
+            int status = 0;
+            // gcc's implementation of typeid(T).name() mangles the type name,
+            // so we have to demangle it.
+#   if GTEST_HAS_CXXABI_H_
+            using abi::__cxa_demangle;
+#   endif  // GTEST_HAS_CXXABI_H_
+            char *const readable_name = __cxa_demangle(name, 0, 0, &status);
+            const std::string name_str(status == 0 ? readable_name : name);
+            free(readable_name);
+            return name_str;
+#  else
+            return name;
+#  endif  // GTEST_HAS_CXXABI_H_ || __HP_aCC
+
+# else
+
+            return "<type>";
+
+# endif  // GTEST_HAS_RTTI
+        }
+
+#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// AssertyTypeEq<T1, T2>::type is defined iff T1 and T2 are the same
+// type.  This can be used as a compile-time assertion to ensure that
+// two types are equal.
+
+        template <typename T1, typename T2>
+        struct AssertTypeEq;
+
+        template <typename T>
+        struct AssertTypeEq<T, T> {
+            typedef bool type;
+        };
+
+// A unique type used as the default value for the arguments of class
+// template Types.  This allows us to simulate variadic templates
+// (e.g. Types<int>, Type<int, double>, and etc), which C++ doesn't
+// support directly.
+        struct None {};
+
+// The following family of struct and struct templates are used to
+// represent type lists.  In particular, TypesN<T1, T2, ..., TN>
+// represents a type list with N types (T1, T2, ..., and TN) in it.
+// Except for Types0, every struct in the family has two member types:
+// Head for the first type in the list, and Tail for the rest of the
+// list.
+
+// The empty type list.
+        struct Types0 {};
+
+// Type lists of length 1, 2, 3, and so on.
+
+        template <typename T1>
+        struct Types1 {
+            typedef T1 Head;
+            typedef Types0 Tail;
+        };
+        template <typename T1, typename T2>
+        struct Types2 {
+            typedef T1 Head;
+            typedef Types1<T2> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3>
+        struct Types3 {
+            typedef T1 Head;
+            typedef Types2<T2, T3> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        struct Types4 {
+            typedef T1 Head;
+            typedef Types3<T2, T3, T4> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        struct Types5 {
+            typedef T1 Head;
+            typedef Types4<T2, T3, T4, T5> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        struct Types6 {
+            typedef T1 Head;
+            typedef Types5<T2, T3, T4, T5, T6> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        struct Types7 {
+            typedef T1 Head;
+            typedef Types6<T2, T3, T4, T5, T6, T7> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        struct Types8 {
+            typedef T1 Head;
+            typedef Types7<T2, T3, T4, T5, T6, T7, T8> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        struct Types9 {
+            typedef T1 Head;
+            typedef Types8<T2, T3, T4, T5, T6, T7, T8, T9> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        struct Types10 {
+            typedef T1 Head;
+            typedef Types9<T2, T3, T4, T5, T6, T7, T8, T9, T10> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11>
+        struct Types11 {
+            typedef T1 Head;
+            typedef Types10<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12>
+        struct Types12 {
+            typedef T1 Head;
+            typedef Types11<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13>
+        struct Types13 {
+            typedef T1 Head;
+            typedef Types12<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14>
+        struct Types14 {
+            typedef T1 Head;
+            typedef Types13<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15>
+        struct Types15 {
+            typedef T1 Head;
+            typedef Types14<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16>
+        struct Types16 {
+            typedef T1 Head;
+            typedef Types15<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17>
+        struct Types17 {
+            typedef T1 Head;
+            typedef Types16<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18>
+        struct Types18 {
+            typedef T1 Head;
+            typedef Types17<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19>
+        struct Types19 {
+            typedef T1 Head;
+            typedef Types18<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20>
+        struct Types20 {
+            typedef T1 Head;
+            typedef Types19<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21>
+        struct Types21 {
+            typedef T1 Head;
+            typedef Types20<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22>
+        struct Types22 {
+            typedef T1 Head;
+            typedef Types21<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23>
+        struct Types23 {
+            typedef T1 Head;
+            typedef Types22<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24>
+        struct Types24 {
+            typedef T1 Head;
+            typedef Types23<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25>
+        struct Types25 {
+            typedef T1 Head;
+            typedef Types24<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26>
+        struct Types26 {
+            typedef T1 Head;
+            typedef Types25<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27>
+        struct Types27 {
+            typedef T1 Head;
+            typedef Types26<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28>
+        struct Types28 {
+            typedef T1 Head;
+            typedef Types27<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29>
+        struct Types29 {
+            typedef T1 Head;
+            typedef Types28<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30>
+        struct Types30 {
+            typedef T1 Head;
+            typedef Types29<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31>
+        struct Types31 {
+            typedef T1 Head;
+            typedef Types30<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32>
+        struct Types32 {
+            typedef T1 Head;
+            typedef Types31<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33>
+        struct Types33 {
+            typedef T1 Head;
+            typedef Types32<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34>
+        struct Types34 {
+            typedef T1 Head;
+            typedef Types33<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35>
+        struct Types35 {
+            typedef T1 Head;
+            typedef Types34<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36>
+        struct Types36 {
+            typedef T1 Head;
+            typedef Types35<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37>
+        struct Types37 {
+            typedef T1 Head;
+            typedef Types36<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38>
+        struct Types38 {
+            typedef T1 Head;
+            typedef Types37<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39>
+        struct Types39 {
+            typedef T1 Head;
+            typedef Types38<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40>
+        struct Types40 {
+            typedef T1 Head;
+            typedef Types39<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41>
+        struct Types41 {
+            typedef T1 Head;
+            typedef Types40<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42>
+        struct Types42 {
+            typedef T1 Head;
+            typedef Types41<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43>
+        struct Types43 {
+            typedef T1 Head;
+            typedef Types42<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44>
+        struct Types44 {
+            typedef T1 Head;
+            typedef Types43<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45>
+        struct Types45 {
+            typedef T1 Head;
+            typedef Types44<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46>
+        struct Types46 {
+            typedef T1 Head;
+            typedef Types45<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47>
+        struct Types47 {
+            typedef T1 Head;
+            typedef Types46<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48>
+        struct Types48 {
+            typedef T1 Head;
+            typedef Types47<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47, T48> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49>
+        struct Types49 {
+            typedef T1 Head;
+            typedef Types48<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47, T48, T49> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49, typename T50>
+        struct Types50 {
+            typedef T1 Head;
+            typedef Types49<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47, T48, T49, T50> Tail;
+        };
+
+
+    }  // namespace internal
+
+// We don't want to require the users to write TypesN<...> directly,
+// as that would require them to count the length.  Types<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Types<int>
+// will appear as Types<int, None, None, ..., None> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Types<T1, ..., TN>, and Google Test will translate
+// that to TypesN<T1, ..., TN> internally to make error messages
+// readable.  The translation is done by the 'type' member of the
+// Types template.
+    template <typename T1 = internal::None, typename T2 = internal::None,
+              typename T3 = internal::None, typename T4 = internal::None,
+              typename T5 = internal::None, typename T6 = internal::None,
+              typename T7 = internal::None, typename T8 = internal::None,
+              typename T9 = internal::None, typename T10 = internal::None,
+              typename T11 = internal::None, typename T12 = internal::None,
+              typename T13 = internal::None, typename T14 = internal::None,
+              typename T15 = internal::None, typename T16 = internal::None,
+              typename T17 = internal::None, typename T18 = internal::None,
+              typename T19 = internal::None, typename T20 = internal::None,
+              typename T21 = internal::None, typename T22 = internal::None,
+              typename T23 = internal::None, typename T24 = internal::None,
+              typename T25 = internal::None, typename T26 = internal::None,
+              typename T27 = internal::None, typename T28 = internal::None,
+              typename T29 = internal::None, typename T30 = internal::None,
+              typename T31 = internal::None, typename T32 = internal::None,
+              typename T33 = internal::None, typename T34 = internal::None,
+              typename T35 = internal::None, typename T36 = internal::None,
+              typename T37 = internal::None, typename T38 = internal::None,
+              typename T39 = internal::None, typename T40 = internal::None,
+              typename T41 = internal::None, typename T42 = internal::None,
+              typename T43 = internal::None, typename T44 = internal::None,
+              typename T45 = internal::None, typename T46 = internal::None,
+              typename T47 = internal::None, typename T48 = internal::None,
+              typename T49 = internal::None, typename T50 = internal::None>
+    struct Types {
+        typedef internal::Types50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47, T48, T49, T50> type;
+    };
+
+    template <>
+    struct Types<internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types0 type;
+    };
+    template <typename T1>
+    struct Types<T1, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types1<T1> type;
+    };
+    template <typename T1, typename T2>
+    struct Types<T1, T2, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types2<T1, T2> type;
+    };
+    template <typename T1, typename T2, typename T3>
+    struct Types<T1, T2, T3, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types3<T1, T2, T3> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4>
+    struct Types<T1, T2, T3, T4, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types4<T1, T2, T3, T4> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5>
+    struct Types<T1, T2, T3, T4, T5, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types5<T1, T2, T3, T4, T5> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6>
+    struct Types<T1, T2, T3, T4, T5, T6, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types6<T1, T2, T3, T4, T5, T6> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types7<T1, T2, T3, T4, T5, T6, T7> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types8<T1, T2, T3, T4, T5, T6, T7, T8> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types9<T1, T2, T3, T4, T5, T6, T7, T8, T9> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+                T12> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+                T26> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+                T40> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, T47, internal::None, internal::None, internal::None> {
+        typedef internal::Types47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, T47, T48, internal::None, internal::None> {
+        typedef internal::Types48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47, T48> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48, typename T49>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, T47, T48, T49, internal::None> {
+        typedef internal::Types49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47, T48, T49> type;
+    };
+
+    namespace internal
+    {
+
+# define GTEST_TEMPLATE_ template <typename T> class
+
+// The template "selector" struct TemplateSel<Tmpl> is used to
+// represent Tmpl, which must be a class template with one type
+// parameter, as a type.  TemplateSel<Tmpl>::Bind<T>::type is defined
+// as the type Tmpl<T>.  This allows us to actually instantiate the
+// template "selected" by TemplateSel<Tmpl>.
+//
+// This trick is necessary for simulating typedef for class templates,
+// which C++ doesn't support directly.
+        template <GTEST_TEMPLATE_ Tmpl>
+        struct TemplateSel {
+            template <typename T>
+            struct Bind {
+                typedef Tmpl<T> type;
+            };
+        };
+
+# define GTEST_BIND_(TmplSel, T) \
+  TmplSel::template Bind<T>::type
+
+// A unique struct template used as the default value for the
+// arguments of class template Templates.  This allows us to simulate
+// variadic templates (e.g. Templates<int>, Templates<int, double>,
+// and etc), which C++ doesn't support directly.
+        template <typename T>
+        struct NoneT {};
+
+// The following family of struct and struct templates are used to
+// represent template lists.  In particular, TemplatesN<T1, T2, ...,
+// TN> represents a list of N templates (T1, T2, ..., and TN).  Except
+// for Templates0, every struct in the family has two member types:
+// Head for the selector of the first template in the list, and Tail
+// for the rest of the list.
+
+// The empty template list.
+        struct Templates0 {};
+
+// Template lists of length 1, 2, 3, and so on.
+
+        template <GTEST_TEMPLATE_ T1>
+        struct Templates1 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates0 Tail;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2>
+        struct Templates2 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates1<T2> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3>
+        struct Templates3 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates2<T2, T3> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4>
+        struct Templates4 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates3<T2, T3, T4> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5>
+        struct Templates5 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates4<T2, T3, T4, T5> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6>
+        struct Templates6 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates5<T2, T3, T4, T5, T6> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7>
+        struct Templates7 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates6<T2, T3, T4, T5, T6, T7> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8>
+        struct Templates8 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates7<T2, T3, T4, T5, T6, T7, T8> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9>
+        struct Templates9 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates8<T2, T3, T4, T5, T6, T7, T8, T9> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10>
+        struct Templates10 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates9<T2, T3, T4, T5, T6, T7, T8, T9, T10> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11>
+        struct Templates11 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates10<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12>
+        struct Templates12 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates11<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13>
+        struct Templates13 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates12<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14>
+        struct Templates14 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates13<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15>
+        struct Templates15 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates14<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16>
+        struct Templates16 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates15<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17>
+        struct Templates17 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates16<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18>
+        struct Templates18 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates17<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19>
+        struct Templates19 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates18<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20>
+        struct Templates20 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates19<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21>
+        struct Templates21 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates20<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22>
+        struct Templates22 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates21<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23>
+        struct Templates23 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates22<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24>
+        struct Templates24 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates23<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25>
+        struct Templates25 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates24<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26>
+        struct Templates26 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates25<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27>
+        struct Templates27 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates26<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28>
+        struct Templates28 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates27<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29>
+        struct Templates29 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates28<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30>
+        struct Templates30 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates29<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31>
+        struct Templates31 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates30<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32>
+        struct Templates32 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates31<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33>
+        struct Templates33 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates32<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34>
+        struct Templates34 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates33<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35>
+        struct Templates35 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates34<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36>
+        struct Templates36 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates35<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37>
+        struct Templates37 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates36<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38>
+        struct Templates38 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates37<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39>
+        struct Templates39 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates38<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40>
+        struct Templates40 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates39<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41>
+        struct Templates41 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates40<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42>
+        struct Templates42 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates41<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43>
+        struct Templates43 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates42<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44>
+        struct Templates44 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates43<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45>
+        struct Templates45 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates44<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46>
+        struct Templates46 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates45<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47>
+        struct Templates47 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates46<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48>
+        struct Templates48 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates47<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47, T48> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+                  GTEST_TEMPLATE_ T49>
+        struct Templates49 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates48<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47, T48, T49> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+                  GTEST_TEMPLATE_ T49, GTEST_TEMPLATE_ T50>
+        struct Templates50 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates49<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47, T48, T49, T50> Tail;
+        };
+
+
+// We don't want to require the users to write TemplatesN<...> directly,
+// as that would require them to count the length.  Templates<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Templates<list>
+// will appear as Templates<list, NoneT, NoneT, ..., NoneT> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Templates<T1, ..., TN>, and Google Test will translate
+// that to TemplatesN<T1, ..., TN> internally to make error messages
+// readable.  The translation is done by the 'type' member of the
+// Templates template.
+        template <GTEST_TEMPLATE_ T1 = NoneT, GTEST_TEMPLATE_ T2 = NoneT,
+                  GTEST_TEMPLATE_ T3 = NoneT, GTEST_TEMPLATE_ T4 = NoneT,
+                  GTEST_TEMPLATE_ T5 = NoneT, GTEST_TEMPLATE_ T6 = NoneT,
+                  GTEST_TEMPLATE_ T7 = NoneT, GTEST_TEMPLATE_ T8 = NoneT,
+                  GTEST_TEMPLATE_ T9 = NoneT, GTEST_TEMPLATE_ T10 = NoneT,
+                  GTEST_TEMPLATE_ T11 = NoneT, GTEST_TEMPLATE_ T12 = NoneT,
+                  GTEST_TEMPLATE_ T13 = NoneT, GTEST_TEMPLATE_ T14 = NoneT,
+                  GTEST_TEMPLATE_ T15 = NoneT, GTEST_TEMPLATE_ T16 = NoneT,
+                  GTEST_TEMPLATE_ T17 = NoneT, GTEST_TEMPLATE_ T18 = NoneT,
+                  GTEST_TEMPLATE_ T19 = NoneT, GTEST_TEMPLATE_ T20 = NoneT,
+                  GTEST_TEMPLATE_ T21 = NoneT, GTEST_TEMPLATE_ T22 = NoneT,
+                  GTEST_TEMPLATE_ T23 = NoneT, GTEST_TEMPLATE_ T24 = NoneT,
+                  GTEST_TEMPLATE_ T25 = NoneT, GTEST_TEMPLATE_ T26 = NoneT,
+                  GTEST_TEMPLATE_ T27 = NoneT, GTEST_TEMPLATE_ T28 = NoneT,
+                  GTEST_TEMPLATE_ T29 = NoneT, GTEST_TEMPLATE_ T30 = NoneT,
+                  GTEST_TEMPLATE_ T31 = NoneT, GTEST_TEMPLATE_ T32 = NoneT,
+                  GTEST_TEMPLATE_ T33 = NoneT, GTEST_TEMPLATE_ T34 = NoneT,
+                  GTEST_TEMPLATE_ T35 = NoneT, GTEST_TEMPLATE_ T36 = NoneT,
+                  GTEST_TEMPLATE_ T37 = NoneT, GTEST_TEMPLATE_ T38 = NoneT,
+                  GTEST_TEMPLATE_ T39 = NoneT, GTEST_TEMPLATE_ T40 = NoneT,
+                  GTEST_TEMPLATE_ T41 = NoneT, GTEST_TEMPLATE_ T42 = NoneT,
+                  GTEST_TEMPLATE_ T43 = NoneT, GTEST_TEMPLATE_ T44 = NoneT,
+                  GTEST_TEMPLATE_ T45 = NoneT, GTEST_TEMPLATE_ T46 = NoneT,
+                  GTEST_TEMPLATE_ T47 = NoneT, GTEST_TEMPLATE_ T48 = NoneT,
+                  GTEST_TEMPLATE_ T49 = NoneT, GTEST_TEMPLATE_ T50 = NoneT>
+        struct Templates {
+            typedef Templates50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47, T48, T49, T50> type;
+        };
+
+        template <>
+        struct Templates<NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates0 type;
+        };
+        template <GTEST_TEMPLATE_ T1>
+        struct Templates<T1, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates1<T1> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2>
+        struct Templates<T1, T2, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates2<T1, T2> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3>
+        struct Templates<T1, T2, T3, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates3<T1, T2, T3> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4>
+        struct Templates<T1, T2, T3, T4, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates4<T1, T2, T3, T4> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5>
+        struct Templates<T1, T2, T3, T4, T5, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates5<T1, T2, T3, T4, T5> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6>
+        struct Templates<T1, T2, T3, T4, T5, T6, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates6<T1, T2, T3, T4, T5, T6> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates7<T1, T2, T3, T4, T5, T6, T7> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates8<T1, T2, T3, T4, T5, T6, T7, T8> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates9<T1, T2, T3, T4, T5, T6, T7, T8, T9> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                    T13> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT> {
+            typedef Templates22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT> {
+            typedef Templates23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT> {
+            typedef Templates24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                    T27> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                    T41> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, T47, NoneT, NoneT, NoneT> {
+            typedef Templates47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, T47, T48, NoneT, NoneT> {
+            typedef Templates48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47, T48> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+                  GTEST_TEMPLATE_ T49>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, T47, T48, T49, NoneT> {
+            typedef Templates49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47, T48, T49> type;
+        };
+
+// The TypeList template makes it possible to use either a single type
+// or a Types<...> list in TYPED_TEST_CASE() and
+// INSTANTIATE_TYPED_TEST_CASE_P().
+
+        template <typename T>
+        struct TypeList {
+            typedef Types1<T> type;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49, typename T50>
+        struct TypeList<Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                   T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                   T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                   T44, T45, T46, T47, T48, T49, T50> > {
+            typedef typename Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                    T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                    T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                    T41, T42, T43, T44, T45, T46, T47, T48, T49, T50>::type type;
+        };
+
+#endif  // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+
+// Due to C++ preprocessor weirdness, we need double indirection to
+// concatenate two tokens when one of them is __LINE__.  Writing
+//
+//   foo ## __LINE__
+//
+// will result in the token foo__LINE__, instead of foo followed by
+// the current line number.  For more details, see
+// http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.6
+#define GTEST_CONCAT_TOKEN_(foo, bar) GTEST_CONCAT_TOKEN_IMPL_(foo, bar)
+#define GTEST_CONCAT_TOKEN_IMPL_(foo, bar) foo ## bar
+
+class ProtocolMessage;
+namespace proto2
+{
+    class Message;
+}
+
+namespace testing
+{
+
+// Forward declarations.
+
+    class AssertionResult;                 // Result of an assertion.
+    class Message;                         // Represents a failure message.
+    class Test;                            // Represents a test.
+    class TestInfo;                        // Information about a test.
+    class TestPartResult;                  // Result of a test part.
+    class UnitTest;                        // A collection of test cases.
+
+    template <typename T>
+    ::std::string PrintToString(const T &value);
+
+    namespace internal
+    {
+
+        struct TraceInfo;                      // Information about a trace point.
+        class ScopedTrace;                     // Implements scoped trace.
+        class TestInfoImpl;                    // Opaque implementation of TestInfo
+        class UnitTestImpl;                    // Opaque implementation of UnitTest
+
+// The text used in failure messages to indicate the start of the
+// stack trace.
+        GTEST_API_ extern const char kStackTraceMarker[];
+
+// Two overloaded helpers for checking at compile time whether an
+// expression is a null pointer literal (i.e. NULL or any 0-valued
+// compile-time integral constant).  Their return values have
+// different sizes, so we can use sizeof() to test which version is
+// picked by the compiler.  These helpers have no implementations, as
+// we only need their signatures.
+//
+// Given IsNullLiteralHelper(x), the compiler will pick the first
+// version if x can be implicitly converted to Secret*, and pick the
+// second version otherwise.  Since Secret is a secret and incomplete
+// type, the only expression a user can write that has type Secret* is
+// a null pointer literal.  Therefore, we know that x is a null
+// pointer literal if and only if the first version is picked by the
+// compiler.
+        char IsNullLiteralHelper(Secret *p);
+        char (&IsNullLiteralHelper(...))[2];  // NOLINT
+
+// A compile-time bool constant that is true if and only if x is a
+// null pointer literal (i.e. NULL or any 0-valued compile-time
+// integral constant).
+#ifdef GTEST_ELLIPSIS_NEEDS_POD_
+// We lose support for NULL detection where the compiler doesn't like
+// passing non-POD classes through ellipsis (...).
+# define GTEST_IS_NULL_LITERAL_(x) false
+#else
+# define GTEST_IS_NULL_LITERAL_(x) \
+    (sizeof(::testing::internal::IsNullLiteralHelper(x)) == 1)
+#endif  // GTEST_ELLIPSIS_NEEDS_POD_
+
+// Appends the user-supplied message to the Google-Test-generated message.
+        GTEST_API_ std::string AppendUserMessage(
+            const std::string &gtest_msg, const Message &user_msg);
+
+#if GTEST_HAS_EXCEPTIONS
+
+// This exception is thrown by (and only by) a failed Google Test
+// assertion when GTEST_FLAG(throw_on_failure) is true (if exceptions
+// are enabled).  We derive it from std::runtime_error, which is for
+// errors presumably detectable only at run time.  Since
+// std::runtime_error inherits from std::exception, many testing
+// frameworks know how to extract and print the message inside it.
+        class GTEST_API_ GoogleTestFailureException : public ::std::runtime_error
+        {
+        public:
+            explicit GoogleTestFailureException(const TestPartResult &failure);
+        };
+
+#endif  // GTEST_HAS_EXCEPTIONS
+
+// A helper class for creating scoped traces in user programs.
+        class GTEST_API_ ScopedTrace
+        {
+        public:
+            // The c'tor pushes the given source file location and message onto
+            // a trace stack maintained by Google Test.
+            ScopedTrace(const char *file, int line, const Message &message);
+
+            // The d'tor pops the info pushed by the c'tor.
+            //
+            // Note that the d'tor is not virtual in order to be efficient.
+            // Don't inherit from ScopedTrace!
+            ~ScopedTrace();
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedTrace);
+        } GTEST_ATTRIBUTE_UNUSED_;  // A ScopedTrace object does its job in its
+        // c'tor and d'tor.  Therefore it doesn't
+        // need to be used otherwise.
+
+        namespace edit_distance
+        {
+// Returns the optimal edits to go from 'left' to 'right'.
+// All edits cost the same, with replace having lower priority than
+// add/remove.
+// Simple implementation of the Wagner-Fischer algorithm.
+// See http://en.wikipedia.org/wiki/Wagner-Fischer_algorithm
+            enum EditType { kMatch, kAdd, kRemove, kReplace };
+            GTEST_API_ std::vector<EditType> CalculateOptimalEdits(
+                const std::vector<size_t> &left, const std::vector<size_t> &right);
+
+// Same as above, but the input is represented as strings.
+            GTEST_API_ std::vector<EditType> CalculateOptimalEdits(
+                const std::vector<std::string> &left,
+                const std::vector<std::string> &right);
+
+// Create a diff of the input strings in Unified diff format.
+            GTEST_API_ std::string CreateUnifiedDiff(const std::vector<std::string> &left,
+                                                     const std::vector<std::string> &right,
+                                                     size_t context = 2);
+
+        }  // namespace edit_distance
+
+// Calculate the diff between 'left' and 'right' and return it in unified diff
+// format.
+// If not null, stores in 'total_line_count' the total number of lines found
+// in left + right.
+        GTEST_API_ std::string DiffStrings(const std::string &left,
+                                           const std::string &right,
+                                           size_t *total_line_count);
+
+// Constructs and returns the message for an equality assertion
+// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure.
+//
+// The first four parameters are the expressions used in the assertion
+// and their values, as strings.  For example, for ASSERT_EQ(foo, bar)
+// where foo is 5 and bar is 6, we have:
+//
+//   expected_expression: "foo"
+//   actual_expression:   "bar"
+//   expected_value:      "5"
+//   actual_value:        "6"
+//
+// The ignoring_case parameter is true iff the assertion is a
+// *_STRCASEEQ*.  When it's true, the string " (ignoring case)" will
+// be inserted into the message.
+        GTEST_API_ AssertionResult EqFailure(const char *expected_expression,
+                                             const char *actual_expression,
+                                             const std::string &expected_value,
+                                             const std::string &actual_value,
+                                             bool ignoring_case);
+
+// Constructs a failure message for Boolean assertions such as EXPECT_TRUE.
+        GTEST_API_ std::string GetBoolAssertionFailureMessage(
+            const AssertionResult &assertion_result,
+            const char *expression_text,
+            const char *actual_predicate_value,
+            const char *expected_predicate_value);
+
+// This template class represents an IEEE floating-point number
+// (either single-precision or double-precision, depending on the
+// template parameters).
+//
+// The purpose of this class is to do more sophisticated number
+// comparison.  (Due to round-off error, etc, it's very unlikely that
+// two floating-points will be equal exactly.  Hence a naive
+// comparison by the == operation often doesn't work.)
+//
+// Format of IEEE floating-point:
+//
+//   The most-significant bit being the leftmost, an IEEE
+//   floating-point looks like
+//
+//     sign_bit exponent_bits fraction_bits
+//
+//   Here, sign_bit is a single bit that designates the sign of the
+//   number.
+//
+//   For float, there are 8 exponent bits and 23 fraction bits.
+//
+//   For double, there are 11 exponent bits and 52 fraction bits.
+//
+//   More details can be found at
+//   http://en.wikipedia.org/wiki/IEEE_floating-point_standard.
+//
+// Template parameter:
+//
+//   RawType: the raw floating-point type (either float or double)
+        template <typename RawType>
+        class FloatingPoint
+        {
+        public:
+            // Defines the unsigned integer type that has the same size as the
+            // floating point number.
+            typedef typename TypeWithSize<sizeof(RawType)>::UInt Bits;
+
+            // Constants.
+
+            // # of bits in a number.
+            static const size_t kBitCount = 8*sizeof(RawType);
+
+            // # of fraction bits in a number.
+            static const size_t kFractionBitCount =
+                std::numeric_limits<RawType>::digits - 1;
+
+            // # of exponent bits in a number.
+            static const size_t kExponentBitCount = kBitCount - 1 - kFractionBitCount;
+
+            // The mask for the sign bit.
+            static const Bits kSignBitMask = static_cast<Bits>(1) << (kBitCount - 1);
+
+            // The mask for the fraction bits.
+            static const Bits kFractionBitMask =
+                ~static_cast<Bits>(0) >> (kExponentBitCount + 1);
+
+            // The mask for the exponent bits.
+            static const Bits kExponentBitMask = ~(kSignBitMask | kFractionBitMask);
+
+            // How many ULP's (Units in the Last Place) we want to tolerate when
+            // comparing two numbers.  The larger the value, the more error we
+            // allow.  A 0 value means that two numbers must be exactly the same
+            // to be considered equal.
+            //
+            // The maximum error of a single floating-point operation is 0.5
+            // units in the last place.  On Intel CPU's, all floating-point
+            // calculations are done with 80-bit precision, while double has 64
+            // bits.  Therefore, 4 should be enough for ordinary use.
+            //
+            // See the following article for more details on ULP:
+            // http://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/
+            static const size_t kMaxUlps = 4;
+
+            // Constructs a FloatingPoint from a raw floating-point number.
+            //
+            // On an Intel CPU, passing a non-normalized NAN (Not a Number)
+            // around may change its bits, although the new value is guaranteed
+            // to be also a NAN.  Therefore, don't expect this constructor to
+            // preserve the bits in x when x is a NAN.
+            explicit FloatingPoint(const RawType &x)
+            {
+                u_.value_ = x;
+            }
+
+            // Static methods
+
+            // Reinterprets a bit pattern as a floating-point number.
+            //
+            // This function is needed to test the AlmostEquals() method.
+            static RawType ReinterpretBits(const Bits bits)
+            {
+                FloatingPoint fp(0);
+                fp.u_.bits_ = bits;
+                return fp.u_.value_;
+            }
+
+            // Returns the floating-point number that represent positive infinity.
+            static RawType Infinity()
+            {
+                return ReinterpretBits(kExponentBitMask);
+            }
+
+            // Returns the maximum representable finite floating-point number.
+            static RawType Max();
+
+            // Non-static methods
+
+            // Returns the bits that represents this number.
+            const Bits &bits() const
+            {
+                return u_.bits_;
+            }
+
+            // Returns the exponent bits of this number.
+            Bits exponent_bits() const
+            {
+                return kExponentBitMask & u_.bits_;
+            }
+
+            // Returns the fraction bits of this number.
+            Bits fraction_bits() const
+            {
+                return kFractionBitMask & u_.bits_;
+            }
+
+            // Returns the sign bit of this number.
+            Bits sign_bit() const
+            {
+                return kSignBitMask & u_.bits_;
+            }
+
+            // Returns true iff this is NAN (not a number).
+            bool is_nan() const
+            {
+                // It's a NAN if the exponent bits are all ones and the fraction
+                // bits are not entirely zeros.
+                return (exponent_bits() == kExponentBitMask) && (fraction_bits() != 0);
+            }
+
+            // Returns true iff this number is at most kMaxUlps ULP's away from
+            // rhs.  In particular, this function:
+            //
+            //   - returns false if either number is (or both are) NAN.
+            //   - treats really large numbers as almost equal to infinity.
+            //   - thinks +0.0 and -0.0 are 0 DLP's apart.
+            bool AlmostEquals(const FloatingPoint &rhs) const
+            {
+                // The IEEE standard says that any comparison operation involving
+                // a NAN must return false.
+                if (is_nan() || rhs.is_nan()) return false;
+
+                return DistanceBetweenSignAndMagnitudeNumbers(u_.bits_, rhs.u_.bits_)
+                       <= kMaxUlps;
+            }
+
+        private:
+            // The data type used to store the actual floating-point number.
+            union FloatingPointUnion {
+                RawType value_;  // The raw floating-point number.
+                Bits bits_;      // The bits that represent the number.
+            };
+
+            // Converts an integer from the sign-and-magnitude representation to
+            // the biased representation.  More precisely, let N be 2 to the
+            // power of (kBitCount - 1), an integer x is represented by the
+            // unsigned number x + N.
+            //
+            // For instance,
+            //
+            //   -N + 1 (the most negative number representable using
+            //          sign-and-magnitude) is represented by 1;
+            //   0      is represented by N; and
+            //   N - 1  (the biggest number representable using
+            //          sign-and-magnitude) is represented by 2N - 1.
+            //
+            // Read http://en.wikipedia.org/wiki/Signed_number_representations
+            // for more details on signed number representations.
+            static Bits SignAndMagnitudeToBiased(const Bits &sam)
+            {
+                if (kSignBitMask & sam) {
+                    // sam represents a negative number.
+                    return ~sam + 1;
+                } else {
+                    // sam represents a positive number.
+                    return kSignBitMask | sam;
+                }
+            }
+
+            // Given two numbers in the sign-and-magnitude representation,
+            // returns the distance between them as an unsigned number.
+            static Bits DistanceBetweenSignAndMagnitudeNumbers(const Bits &sam1,
+                                                               const Bits &sam2)
+            {
+                const Bits biased1 = SignAndMagnitudeToBiased(sam1);
+                const Bits biased2 = SignAndMagnitudeToBiased(sam2);
+                return (biased1 >= biased2) ? (biased1 - biased2) : (biased2 - biased1);
+            }
+
+            FloatingPointUnion u_;
+        };
+
+// We cannot use std::numeric_limits<T>::max() as it clashes with the max()
+// macro defined by <windows.h>.
+        template <>
+        inline float FloatingPoint<float>::Max()
+        {
+            return FLT_MAX;
+        }
+        template <>
+        inline double FloatingPoint<double>::Max()
+        {
+            return DBL_MAX;
+        }
+
+// Typedefs the instances of the FloatingPoint template class that we
+// care to use.
+        typedef FloatingPoint<float> Float;
+        typedef FloatingPoint<double> Double;
+
+// In order to catch the mistake of putting tests that use different
+// test fixture classes in the same test case, we need to assign
+// unique IDs to fixture classes and compare them.  The TypeId type is
+// used to hold such IDs.  The user should treat TypeId as an opaque
+// type: the only operation allowed on TypeId values is to compare
+// them for equality using the == operator.
+        typedef const void *TypeId;
+
+        template <typename T>
+        class TypeIdHelper
+        {
+        public:
+            // dummy_ must not have a const type.  Otherwise an overly eager
+            // compiler (e.g. MSVC 7.1 & 8.0) may try to merge
+            // TypeIdHelper<T>::dummy_ for different Ts as an "optimization".
+            static bool dummy_;
+        };
+
+        template <typename T>
+        bool TypeIdHelper<T>::dummy_ = false;
+
+// GetTypeId<T>() returns the ID of type T.  Different values will be
+// returned for different types.  Calling the function twice with the
+// same type argument is guaranteed to return the same ID.
+        template <typename T>
+        TypeId GetTypeId()
+        {
+            // The compiler is required to allocate a different
+            // TypeIdHelper<T>::dummy_ variable for each T used to instantiate
+            // the template.  Therefore, the address of dummy_ is guaranteed to
+            // be unique.
+            return &(TypeIdHelper<T>::dummy_);
+        }
+
+// Returns the type ID of ::testing::Test.  Always call this instead
+// of GetTypeId< ::testing::Test>() to get the type ID of
+// ::testing::Test, as the latter may give the wrong result due to a
+// suspected linker bug when compiling Google Test as a Mac OS X
+// framework.
+        GTEST_API_ TypeId GetTestTypeId();
+
+// Defines the abstract factory interface that creates instances
+// of a Test object.
+        class TestFactoryBase
+        {
+        public:
+            virtual ~TestFactoryBase() {}
+
+            // Creates a test instance to run. The instance is both created and destroyed
+            // within TestInfoImpl::Run()
+            virtual Test *CreateTest() = 0;
+
+        protected:
+            TestFactoryBase() {}
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(TestFactoryBase);
+        };
+
+// This class provides implementation of TeastFactoryBase interface.
+// It is used in TEST and TEST_F macros.
+        template <class TestClass>
+        class TestFactoryImpl : public TestFactoryBase
+        {
+        public:
+            virtual Test *CreateTest()
+            {
+                return new TestClass;
+            }
+        };
+
+#if GTEST_OS_WINDOWS
+
+// Predicate-formatters for implementing the HRESULT checking macros
+// {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}
+// We pass a long instead of HRESULT to avoid causing an
+// include dependency for the HRESULT type.
+        GTEST_API_ AssertionResult IsHRESULTSuccess(const char *expr,
+                                                    long hr);  // NOLINT
+        GTEST_API_ AssertionResult IsHRESULTFailure(const char *expr,
+                                                    long hr);  // NOLINT
+
+#endif  // GTEST_OS_WINDOWS
+
+// Types of SetUpTestCase() and TearDownTestCase() functions.
+        typedef void (*SetUpTestCaseFunc)();
+        typedef void (*TearDownTestCaseFunc)();
+
+        struct CodeLocation {
+            CodeLocation(const std::string &a_file, int a_line)
+                : file(a_file), line(a_line) {}
+
+            std::string file;
+            int line;
+        };
+
+// Creates a new TestInfo object and registers it with Google Test;
+// returns the created object.
+//
+// Arguments:
+//
+//   test_case_name:   name of the test case
+//   name:             name of the test
+//   type_param        the name of the test's type parameter, or NULL if
+//                     this is not a typed or a type-parameterized test.
+//   value_param       text representation of the test's value parameter,
+//                     or NULL if this is not a type-parameterized test.
+//   code_location:    code location where the test is defined
+//   fixture_class_id: ID of the test fixture class
+//   set_up_tc:        pointer to the function that sets up the test case
+//   tear_down_tc:     pointer to the function that tears down the test case
+//   factory:          pointer to the factory that creates a test object.
+//                     The newly created TestInfo instance will assume
+//                     ownership of the factory object.
+        GTEST_API_ TestInfo *MakeAndRegisterTestInfo(
+            const char *test_case_name,
+            const char *name,
+            const char *type_param,
+            const char *value_param,
+            CodeLocation code_location,
+            TypeId fixture_class_id,
+            SetUpTestCaseFunc set_up_tc,
+            TearDownTestCaseFunc tear_down_tc,
+            TestFactoryBase *factory);
+
+// If *pstr starts with the given prefix, modifies *pstr to be right
+// past the prefix and returns true; otherwise leaves *pstr unchanged
+// and returns false.  None of pstr, *pstr, and prefix can be NULL.
+        GTEST_API_ bool SkipPrefix(const char *prefix, const char **pstr);
+
+#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// State of the definition of a type-parameterized test case.
+        class GTEST_API_ TypedTestCasePState
+        {
+        public:
+            TypedTestCasePState() : registered_(false) {}
+
+            // Adds the given test name to defined_test_names_ and return true
+            // if the test case hasn't been registered; otherwise aborts the
+            // program.
+            bool AddTestName(const char *file, int line, const char *case_name,
+                             const char *test_name)
+            {
+                if (registered_) {
+                    fprintf(stderr, "%s Test %s must be defined before "
+                            "REGISTER_TYPED_TEST_CASE_P(%s, ...).\n",
+                            FormatFileLocation(file, line).c_str(), test_name, case_name);
+                    fflush(stderr);
+                    posix::Abort();
+                }
+                registered_tests_.insert(
+                    ::std::make_pair(test_name, CodeLocation(file, line)));
+                return true;
+            }
+
+            bool TestExists(const std::string &test_name) const
+            {
+                return registered_tests_.count(test_name) > 0;
+            }
+
+            const CodeLocation &GetCodeLocation(const std::string &test_name) const
+            {
+                RegisteredTestsMap::const_iterator it = registered_tests_.find(test_name);
+                GTEST_CHECK_(it != registered_tests_.end());
+                return it->second;
+            }
+
+            // Verifies that registered_tests match the test names in
+            // defined_test_names_; returns registered_tests if successful, or
+            // aborts the program otherwise.
+            const char *VerifyRegisteredTestNames(
+                const char *file, int line, const char *registered_tests);
+
+        private:
+            typedef ::std::map<std::string, CodeLocation> RegisteredTestsMap;
+
+            bool registered_;
+            RegisteredTestsMap registered_tests_;
+        };
+
+// Skips to the first non-space char after the first comma in 'str';
+// returns NULL if no comma is found in 'str'.
+        inline const char *SkipComma(const char *str)
+        {
+            const char *comma = strchr(str, ',');
+            if (comma == NULL) {
+                return NULL;
+            }
+            while (IsSpace(*(++comma))) {}
+            return comma;
+        }
+
+// Returns the prefix of 'str' before the first comma in it; returns
+// the entire string if it contains no comma.
+        inline std::string GetPrefixUntilComma(const char *str)
+        {
+            const char *comma = strchr(str, ',');
+            return comma == NULL ? str : std::string(str, comma);
+        }
+
+// Splits a given string on a given delimiter, populating a given
+// vector with the fields.
+        void SplitString(const ::std::string &str, char delimiter,
+                         ::std::vector< ::std::string> *dest);
+
+// TypeParameterizedTest<Fixture, TestSel, Types>::Register()
+// registers a list of type-parameterized tests with Google Test.  The
+// return value is insignificant - we just need to return something
+// such that we can call this function in a namespace scope.
+//
+// Implementation note: The GTEST_TEMPLATE_ macro declares a template
+// template parameter.  It's defined in gtest-type-util.h.
+        template <GTEST_TEMPLATE_ Fixture, class TestSel, typename Types>
+        class TypeParameterizedTest
+        {
+        public:
+            // 'index' is the index of the test in the type list 'Types'
+            // specified in INSTANTIATE_TYPED_TEST_CASE_P(Prefix, TestCase,
+            // Types).  Valid values for 'index' are [0, N - 1] where N is the
+            // length of Types.
+            static bool Register(const char *prefix,
+                                 CodeLocation code_location,
+                                 const char *case_name, const char *test_names,
+                                 int index)
+            {
+                typedef typename Types::Head Type;
+                typedef Fixture<Type> FixtureClass;
+                typedef typename GTEST_BIND_(TestSel, Type) TestClass;
+
+                // First, registers the first type-parameterized test in the type
+                // list.
+                MakeAndRegisterTestInfo(
+                    (std::string(prefix) + (prefix[0] == '\0' ? "" : "/") + case_name + "/"
+                     + StreamableToString(index)).c_str(),
+                    StripTrailingSpaces(GetPrefixUntilComma(test_names)).c_str(),
+                    GetTypeName<Type>().c_str(),
+                    NULL,  // No value parameter.
+                    code_location,
+                    GetTypeId<FixtureClass>(),
+                    TestClass::SetUpTestCase,
+                    TestClass::TearDownTestCase,
+                    new TestFactoryImpl<TestClass>);
+
+                // Next, recurses (at compile time) with the tail of the type list.
+                return TypeParameterizedTest<Fixture, TestSel, typename Types::Tail>
+                       ::Register(prefix, code_location, case_name, test_names, index + 1);
+            }
+        };
+
+// The base case for the compile time recursion.
+        template <GTEST_TEMPLATE_ Fixture, class TestSel>
+        class TypeParameterizedTest<Fixture, TestSel, Types0>
+        {
+        public:
+            static bool Register(const char * /*prefix*/, CodeLocation,
+                                 const char * /*case_name*/, const char * /*test_names*/,
+                                 int /*index*/)
+            {
+                return true;
+            }
+        };
+
+// TypeParameterizedTestCase<Fixture, Tests, Types>::Register()
+// registers *all combinations* of 'Tests' and 'Types' with Google
+// Test.  The return value is insignificant - we just need to return
+// something such that we can call this function in a namespace scope.
+        template <GTEST_TEMPLATE_ Fixture, typename Tests, typename Types>
+        class TypeParameterizedTestCase
+        {
+        public:
+            static bool Register(const char *prefix, CodeLocation code_location,
+                                 const TypedTestCasePState *state,
+                                 const char *case_name, const char *test_names)
+            {
+                std::string test_name = StripTrailingSpaces(
+                                            GetPrefixUntilComma(test_names));
+                if (!state->TestExists(test_name)) {
+                    fprintf(stderr, "Failed to get code location for test %s.%s at %s.",
+                            case_name, test_name.c_str(),
+                            FormatFileLocation(code_location.file.c_str(),
+                                               code_location.line).c_str());
+                    fflush(stderr);
+                    posix::Abort();
+                }
+                const CodeLocation &test_location = state->GetCodeLocation(test_name);
+
+                typedef typename Tests::Head Head;
+
+                // First, register the first test in 'Test' for each type in 'Types'.
+                TypeParameterizedTest<Fixture, Head, Types>::Register(
+                    prefix, test_location, case_name, test_names, 0);
+
+                // Next, recurses (at compile time) with the tail of the test list.
+                return TypeParameterizedTestCase<Fixture, typename Tests::Tail, Types>
+                       ::Register(prefix, code_location, state,
+                                  case_name, SkipComma(test_names));
+            }
+        };
+
+// The base case for the compile time recursion.
+        template <GTEST_TEMPLATE_ Fixture, typename Types>
+        class TypeParameterizedTestCase<Fixture, Templates0, Types>
+        {
+        public:
+            static bool Register(const char * /*prefix*/, CodeLocation,
+                                 const TypedTestCasePState * /*state*/,
+                                 const char * /*case_name*/, const char * /*test_names*/)
+            {
+                return true;
+            }
+        };
+
+#endif  // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// Returns the current OS stack trace as an std::string.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag.  The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in
+// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't.
+        GTEST_API_ std::string GetCurrentOsStackTraceExceptTop(
+            UnitTest *unit_test, int skip_count);
+
+// Helpers for suppressing warnings on unreachable code or constant
+// condition.
+
+// Always returns true.
+        GTEST_API_ bool AlwaysTrue();
+
+// Always returns false.
+        inline bool AlwaysFalse()
+        {
+            return !AlwaysTrue();
+        }
+
+// Helper for suppressing false warning from Clang on a const char*
+// variable declared in a conditional expression always being NULL in
+// the else branch.
+        struct GTEST_API_ ConstCharPtr {
+            ConstCharPtr(const char *str) : value(str) {}
+            operator bool() const
+            {
+                return true;
+            }
+            const char *value;
+        };
+
+// A simple Linear Congruential Generator for generating random
+// numbers with a uniform distribution.  Unlike rand() and srand(), it
+// doesn't use global state (and therefore can't interfere with user
+// code).  Unlike rand_r(), it's portable.  An LCG isn't very random,
+// but it's good enough for our purposes.
+        class GTEST_API_ Random
+        {
+        public:
+            static const UInt32 kMaxRange = 1u << 31;
+
+            explicit Random(UInt32 seed) : state_(seed) {}
+
+            void Reseed(UInt32 seed)
+            {
+                state_ = seed;
+            }
+
+            // Generates a random number from [0, range).  Crashes if 'range' is
+            // 0 or greater than kMaxRange.
+            UInt32 Generate(UInt32 range);
+
+        private:
+            UInt32 state_;
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Random);
+        };
+
+// Defining a variable of type CompileAssertTypesEqual<T1, T2> will cause a
+// compiler error iff T1 and T2 are different types.
+        template <typename T1, typename T2>
+        struct CompileAssertTypesEqual;
+
+        template <typename T>
+        struct CompileAssertTypesEqual<T, T> {
+        };
+
+// Removes the reference from a type if it is a reference type,
+// otherwise leaves it unchanged.  This is the same as
+// tr1::remove_reference, which is not widely available yet.
+        template <typename T>
+        struct RemoveReference {
+            typedef T type;
+        };  // NOLINT
+        template <typename T>
+        struct RemoveReference<T &> {
+            typedef T type;
+        };  // NOLINT
+
+// A handy wrapper around RemoveReference that works when the argument
+// T depends on template parameters.
+#define GTEST_REMOVE_REFERENCE_(T) \
+    typename ::testing::internal::RemoveReference<T>::type
+
+// Removes const from a type if it is a const type, otherwise leaves
+// it unchanged.  This is the same as tr1::remove_const, which is not
+// widely available yet.
+        template <typename T>
+        struct RemoveConst {
+            typedef T type;
+        };  // NOLINT
+        template <typename T>
+        struct RemoveConst<const T> {
+            typedef T type;
+        };  // NOLINT
+
+// MSVC 8.0, Sun C++, and IBM XL C++ have a bug which causes the above
+// definition to fail to remove the const in 'const int[3]' and 'const
+// char[3][4]'.  The following specialization works around the bug.
+        template <typename T, size_t N>
+        struct RemoveConst<const T[N]> {
+            typedef typename RemoveConst<T>::type type[N];
+        };
+
+#if defined(_MSC_VER) && _MSC_VER < 1400
+// This is the only specialization that allows VC++ 7.1 to remove const in
+// 'const int[3] and 'const int[3][4]'.  However, it causes trouble with GCC
+// and thus needs to be conditionally compiled.
+        template <typename T, size_t N>
+        struct RemoveConst<T[N]> {
+            typedef typename RemoveConst<T>::type type[N];
+        };
+#endif
+
+// A handy wrapper around RemoveConst that works when the argument
+// T depends on template parameters.
+#define GTEST_REMOVE_CONST_(T) \
+    typename ::testing::internal::RemoveConst<T>::type
+
+// Turns const U&, U&, const U, and U all into U.
+#define GTEST_REMOVE_REFERENCE_AND_CONST_(T) \
+    GTEST_REMOVE_CONST_(GTEST_REMOVE_REFERENCE_(T))
+
+// Adds reference to a type if it is not a reference type,
+// otherwise leaves it unchanged.  This is the same as
+// tr1::add_reference, which is not widely available yet.
+        template <typename T>
+        struct AddReference {
+            typedef T &type;
+        };  // NOLINT
+        template <typename T>
+        struct AddReference<T &> {
+            typedef T &type;
+        };  // NOLINT
+
+// A handy wrapper around AddReference that works when the argument T
+// depends on template parameters.
+#define GTEST_ADD_REFERENCE_(T) \
+    typename ::testing::internal::AddReference<T>::type
+
+// Adds a reference to const on top of T as necessary.  For example,
+// it transforms
+//
+//   char         ==> const char&
+//   const char   ==> const char&
+//   char&        ==> const char&
+//   const char&  ==> const char&
+//
+// The argument T must depend on some template parameters.
+#define GTEST_REFERENCE_TO_CONST_(T) \
+    GTEST_ADD_REFERENCE_(const GTEST_REMOVE_REFERENCE_(T))
+
+// ImplicitlyConvertible<From, To>::value is a compile-time bool
+// constant that's true iff type From can be implicitly converted to
+// type To.
+        template <typename From, typename To>
+        class ImplicitlyConvertible
+        {
+        private:
+            // We need the following helper functions only for their types.
+            // They have no implementations.
+
+            // MakeFrom() is an expression whose type is From.  We cannot simply
+            // use From(), as the type From may not have a public default
+            // constructor.
+            static typename AddReference<From>::type MakeFrom();
+
+            // These two functions are overloaded.  Given an expression
+            // Helper(x), the compiler will pick the first version if x can be
+            // implicitly converted to type To; otherwise it will pick the
+            // second version.
+            //
+            // The first version returns a value of size 1, and the second
+            // version returns a value of size 2.  Therefore, by checking the
+            // size of Helper(x), which can be done at compile time, we can tell
+            // which version of Helper() is used, and hence whether x can be
+            // implicitly converted to type To.
+            static char Helper(To);
+            static char (&Helper(...))[2];  // NOLINT
+
+            // We have to put the 'public' section after the 'private' section,
+            // or MSVC refuses to compile the code.
+        public:
+#if defined(__BORLANDC__)
+            // C++Builder cannot use member overload resolution during template
+            // instantiation.  The simplest workaround is to use its C++0x type traits
+            // functions (C++Builder 2009 and above only).
+            static const bool value = __is_convertible(From, To);
+#else
+            // MSVC warns about implicitly converting from double to int for
+            // possible loss of data, so we need to temporarily disable the
+            // warning.
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4244)
+            static const bool value =
+                sizeof(Helper(ImplicitlyConvertible::MakeFrom())) == 1;
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+#endif  // __BORLANDC__
+        };
+        template <typename From, typename To>
+        const bool ImplicitlyConvertible<From, To>::value;
+
+// IsAProtocolMessage<T>::value is a compile-time bool constant that's
+// true iff T is type ProtocolMessage, proto2::Message, or a subclass
+// of those.
+        template <typename T>
+        struct IsAProtocolMessage
+            : public bool_constant<
+              ImplicitlyConvertible<const T *, const ::ProtocolMessage *>::value ||
+              ImplicitlyConvertible<const T *, const ::proto2::Message *>::value> {
+        };
+
+// When the compiler sees expression IsContainerTest<C>(0), if C is an
+// STL-style container class, the first overload of IsContainerTest
+// will be viable (since both C::iterator* and C::const_iterator* are
+// valid types and NULL can be implicitly converted to them).  It will
+// be picked over the second overload as 'int' is a perfect match for
+// the type of argument 0.  If C::iterator or C::const_iterator is not
+// a valid type, the first overload is not viable, and the second
+// overload will be picked.  Therefore, we can determine whether C is
+// a container class by checking the type of IsContainerTest<C>(0).
+// The value of the expression is insignificant.
+//
+// Note that we look for both C::iterator and C::const_iterator.  The
+// reason is that C++ injects the name of a class as a member of the
+// class itself (e.g. you can refer to class iterator as either
+// 'iterator' or 'iterator::iterator').  If we look for C::iterator
+// only, for example, we would mistakenly think that a class named
+// iterator is an STL container.
+//
+// Also note that the simpler approach of overloading
+// IsContainerTest(typename C::const_iterator*) and
+// IsContainerTest(...) doesn't work with Visual Age C++ and Sun C++.
+        typedef int IsContainer;
+        template <class C>
+        IsContainer IsContainerTest(int /* dummy */,
+                                    typename C::iterator * /* it */ = NULL,
+                                    typename C::const_iterator * /* const_it */ = NULL)
+        {
+            return 0;
+        }
+
+        typedef char IsNotContainer;
+        template <class C>
+        IsNotContainer IsContainerTest(long /* dummy */)
+        {
+            return '\0';
+        }
+
+// EnableIf<condition>::type is void when 'Cond' is true, and
+// undefined when 'Cond' is false.  To use SFINAE to make a function
+// overload only apply when a particular expression is true, add
+// "typename EnableIf<expression>::type* = 0" as the last parameter.
+        template<bool> struct EnableIf;
+        template<> struct EnableIf<true> {
+            typedef void type;
+        };  // NOLINT
+
+// Utilities for native arrays.
+
+// ArrayEq() compares two k-dimensional native arrays using the
+// elements' operator==, where k can be any integer >= 0.  When k is
+// 0, ArrayEq() degenerates into comparing a single pair of values.
+
+        template <typename T, typename U>
+        bool ArrayEq(const T *lhs, size_t size, const U *rhs);
+
+// This generic version is used when k is 0.
+        template <typename T, typename U>
+        inline bool ArrayEq(const T &lhs, const U &rhs)
+        {
+            return lhs == rhs;
+        }
+
+// This overload is used when k >= 1.
+        template <typename T, typename U, size_t N>
+        inline bool ArrayEq(const T(&lhs)[N], const U(&rhs)[N])
+        {
+            return internal::ArrayEq(lhs, N, rhs);
+        }
+
+// This helper reduces code bloat.  If we instead put its logic inside
+// the previous ArrayEq() function, arrays with different sizes would
+// lead to different copies of the template code.
+        template <typename T, typename U>
+        bool ArrayEq(const T *lhs, size_t size, const U *rhs)
+        {
+            for (size_t i = 0; i != size; i++) {
+                if (!internal::ArrayEq(lhs[i], rhs[i]))
+                    return false;
+            }
+            return true;
+        }
+
+// Finds the first element in the iterator range [begin, end) that
+// equals elem.  Element may be a native array type itself.
+        template <typename Iter, typename Element>
+        Iter ArrayAwareFind(Iter begin, Iter end, const Element &elem)
+        {
+            for (Iter it = begin; it != end; ++it) {
+                if (internal::ArrayEq(*it, elem))
+                    return it;
+            }
+            return end;
+        }
+
+// CopyArray() copies a k-dimensional native array using the elements'
+// operator=, where k can be any integer >= 0.  When k is 0,
+// CopyArray() degenerates into copying a single value.
+
+        template <typename T, typename U>
+        void CopyArray(const T *from, size_t size, U *to);
+
+// This generic version is used when k is 0.
+        template <typename T, typename U>
+        inline void CopyArray(const T &from, U *to)
+        {
+            *to = from;
+        }
+
+// This overload is used when k >= 1.
+        template <typename T, typename U, size_t N>
+        inline void CopyArray(const T(&from)[N], U(*to)[N])
+        {
+            internal::CopyArray(from, N, *to);
+        }
+
+// This helper reduces code bloat.  If we instead put its logic inside
+// the previous CopyArray() function, arrays with different sizes
+// would lead to different copies of the template code.
+        template <typename T, typename U>
+        void CopyArray(const T *from, size_t size, U *to)
+        {
+            for (size_t i = 0; i != size; i++) {
+                internal::CopyArray(from[i], to + i);
+            }
+        }
+
+// The relation between an NativeArray object (see below) and the
+// native array it represents.
+// We use 2 different structs to allow non-copyable types to be used, as long
+// as RelationToSourceReference() is passed.
+        struct RelationToSourceReference {};
+        struct RelationToSourceCopy {};
+
+// Adapts a native array to a read-only STL-style container.  Instead
+// of the complete STL container concept, this adaptor only implements
+// members useful for Google Mock's container matchers.  New members
+// should be added as needed.  To simplify the implementation, we only
+// support Element being a raw type (i.e. having no top-level const or
+// reference modifier).  It's the client's responsibility to satisfy
+// this requirement.  Element can be an array type itself (hence
+// multi-dimensional arrays are supported).
+        template <typename Element>
+        class NativeArray
+        {
+        public:
+            // STL-style container typedefs.
+            typedef Element value_type;
+            typedef Element *iterator;
+            typedef const Element *const_iterator;
+
+            // Constructs from a native array. References the source.
+            NativeArray(const Element *array, size_t count, RelationToSourceReference)
+            {
+                InitRef(array, count);
+            }
+
+            // Constructs from a native array. Copies the source.
+            NativeArray(const Element *array, size_t count, RelationToSourceCopy)
+            {
+                InitCopy(array, count);
+            }
+
+            // Copy constructor.
+            NativeArray(const NativeArray &rhs)
+            {
+                (this->*rhs.clone_)(rhs.array_, rhs.size_);
+            }
+
+            ~NativeArray()
+            {
+                if (clone_ != &NativeArray::InitRef)
+                    delete[] array_;
+            }
+
+            // STL-style container methods.
+            size_t size() const
+            {
+                return size_;
+            }
+            const_iterator begin() const
+            {
+                return array_;
+            }
+            const_iterator end() const
+            {
+                return array_ + size_;
+            }
+            bool operator==(const NativeArray &rhs) const
+            {
+                return size() == rhs.size() &&
+                       ArrayEq(begin(), size(), rhs.begin());
+            }
+
+        private:
+            enum {
+                kCheckTypeIsNotConstOrAReference = StaticAssertTypeEqHelper<
+                                                   Element, GTEST_REMOVE_REFERENCE_AND_CONST_(Element)>::value,
+            };
+
+            // Initializes this object with a copy of the input.
+            void InitCopy(const Element *array, size_t a_size)
+            {
+                Element *const copy = new Element[a_size];
+                CopyArray(array, a_size, copy);
+                array_ = copy;
+                size_ = a_size;
+                clone_ = &NativeArray::InitCopy;
+            }
+
+            // Initializes this object with a reference of the input.
+            void InitRef(const Element *array, size_t a_size)
+            {
+                array_ = array;
+                size_ = a_size;
+                clone_ = &NativeArray::InitRef;
+            }
+
+            const Element *array_;
+            size_t size_;
+            void (NativeArray::*clone_)(const Element *, size_t);
+
+            GTEST_DISALLOW_ASSIGN_(NativeArray);
+        };
+
+    }  // namespace internal
+}  // namespace testing
+
+#define GTEST_MESSAGE_AT_(file, line, message, result_type) \
+  ::testing::internal::AssertHelper(result_type, file, line, message) \
+    = ::testing::Message()
+
+#define GTEST_MESSAGE_(message, result_type) \
+  GTEST_MESSAGE_AT_(__FILE__, __LINE__, message, result_type)
+
+#define GTEST_FATAL_FAILURE_(message) \
+  return GTEST_MESSAGE_(message, ::testing::TestPartResult::kFatalFailure)
+
+#define GTEST_NONFATAL_FAILURE_(message) \
+  GTEST_MESSAGE_(message, ::testing::TestPartResult::kNonFatalFailure)
+
+#define GTEST_SUCCESS_(message) \
+  GTEST_MESSAGE_(message, ::testing::TestPartResult::kSuccess)
+
+#define GTEST_SKIP_(message) \
+  return GTEST_MESSAGE_(message, ::testing::TestPartResult::kSkip)
+
+// Suppresses MSVC warnings 4072 (unreachable code) for the code following
+// statement if it returns or throws (or doesn't return or throw in some
+// situations).
+#define GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) \
+  if (::testing::internal::AlwaysTrue()) { statement; }
+
+#define GTEST_TEST_THROW_(statement, expected_exception, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::ConstCharPtr gtest_msg = "") { \
+    bool gtest_caught_expected = false; \
+    try { \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    } \
+    catch (expected_exception const&) { \
+      gtest_caught_expected = true; \
+    } \
+    catch (...) { \
+      gtest_msg.value = \
+          "Expected: " #statement " throws an exception of type " \
+          #expected_exception ".\n  Actual: it throws a different type."; \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \
+    } \
+    if (!gtest_caught_expected) { \
+      gtest_msg.value = \
+          "Expected: " #statement " throws an exception of type " \
+          #expected_exception ".\n  Actual: it throws nothing."; \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__): \
+      fail(gtest_msg.value)
+
+#define GTEST_TEST_NO_THROW_(statement, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    try { \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    } \
+    catch (...) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__): \
+      fail("Expected: " #statement " doesn't throw an exception.\n" \
+           "  Actual: it throws.")
+
+#define GTEST_TEST_ANY_THROW_(statement, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    bool gtest_caught_any = false; \
+    try { \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    } \
+    catch (...) { \
+      gtest_caught_any = true; \
+    } \
+    if (!gtest_caught_any) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__): \
+      fail("Expected: " #statement " throws an exception.\n" \
+           "  Actual: it doesn't.")
+
+
+// Implements Boolean test assertions such as EXPECT_TRUE. expression can be
+// either a boolean expression or an AssertionResult. text is a textual
+// representation of expression as it was passed into the EXPECT_TRUE.
+#define GTEST_TEST_BOOLEAN_(expression, text, actual, expected, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (const ::testing::AssertionResult gtest_ar_ = \
+      ::testing::AssertionResult(expression)) \
+    ; \
+  else \
+    fail(::testing::internal::GetBoolAssertionFailureMessage(\
+        gtest_ar_, text, #actual, #expected).c_str())
+
+#define GTEST_TEST_NO_FATAL_FAILURE_(statement, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    ::testing::internal::HasNewFatalFailureHelper gtest_fatal_failure_checker; \
+    GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    if (gtest_fatal_failure_checker.has_new_fatal_failure()) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__): \
+      fail("Expected: " #statement " doesn't generate new fatal " \
+           "failures in the current thread.\n" \
+           "  Actual: it does.")
+
+// Expands to the name of the class that implements the given test.
+#define GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \
+  test_case_name##_##test_name##_Test
+
+// Helper macro for defining tests.
+#define GTEST_TEST_(test_case_name, test_name, parent_class, parent_id)\
+class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) : public parent_class {\
+ public:\
+  GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {}\
+ private:\
+  virtual void TestBody();\
+  static ::testing::TestInfo* const test_info_ GTEST_ATTRIBUTE_UNUSED_;\
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(\
+      GTEST_TEST_CLASS_NAME_(test_case_name, test_name));\
+};\
+\
+::testing::TestInfo* const GTEST_TEST_CLASS_NAME_(test_case_name, test_name)\
+  ::test_info_ =\
+    ::testing::internal::MakeAndRegisterTestInfo(\
+        #test_case_name, #test_name, NULL, NULL, \
+        ::testing::internal::CodeLocation(__FILE__, __LINE__), \
+        (parent_id), \
+        parent_class::SetUpTestCase, \
+        parent_class::TearDownTestCase, \
+        new ::testing::internal::TestFactoryImpl<\
+            GTEST_TEST_CLASS_NAME_(test_case_name, test_name)>);\
+void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody()
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the public API for death tests.  It is
+// #included by gtest.h so a user doesn't need to include this
+// directly.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines internal utilities needed for implementing
+// death tests.  They are subject to change without notice.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+
+
+#include <stdio.h>
+
+namespace testing
+{
+    namespace internal
+    {
+
+        GTEST_DECLARE_string_(internal_run_death_test);
+
+// Names of the flags (needed for parsing Google Test flags).
+        const char kDeathTestStyleFlag[] = "death_test_style";
+        const char kDeathTestUseFork[] = "death_test_use_fork";
+        const char kInternalRunDeathTestFlag[] = "internal_run_death_test";
+
+#if GTEST_HAS_DEATH_TEST
+
+// DeathTest is a class that hides much of the complexity of the
+// GTEST_DEATH_TEST_ macro.  It is abstract; its static Create method
+// returns a concrete class that depends on the prevailing death test
+// style, as defined by the --gtest_death_test_style and/or
+// --gtest_internal_run_death_test flags.
+
+// In describing the results of death tests, these terms are used with
+// the corresponding definitions:
+//
+// exit status:  The integer exit information in the format specified
+//               by wait(2)
+// exit code:    The integer code passed to exit(3), _exit(2), or
+//               returned from main()
+        class GTEST_API_ DeathTest
+        {
+        public:
+            // Create returns false if there was an error determining the
+            // appropriate action to take for the current death test; for example,
+            // if the gtest_death_test_style flag is set to an invalid value.
+            // The LastMessage method will return a more detailed message in that
+            // case.  Otherwise, the DeathTest pointer pointed to by the "test"
+            // argument is set.  If the death test should be skipped, the pointer
+            // is set to NULL; otherwise, it is set to the address of a new concrete
+            // DeathTest object that controls the execution of the current test.
+            static bool Create(const char *statement, const RE *regex,
+                               const char *file, int line, DeathTest **test);
+            DeathTest();
+            virtual ~DeathTest() { }
+
+            // A helper class that aborts a death test when it's deleted.
+            class ReturnSentinel
+            {
+            public:
+                explicit ReturnSentinel(DeathTest *test) : test_(test) { }
+                ~ReturnSentinel()
+                {
+                    test_->Abort(TEST_ENCOUNTERED_RETURN_STATEMENT);
+                }
+            private:
+                DeathTest *const test_;
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ReturnSentinel);
+            } GTEST_ATTRIBUTE_UNUSED_;
+
+            // An enumeration of possible roles that may be taken when a death
+            // test is encountered.  EXECUTE means that the death test logic should
+            // be executed immediately.  OVERSEE means that the program should prepare
+            // the appropriate environment for a child process to execute the death
+            // test, then wait for it to complete.
+            enum TestRole { OVERSEE_TEST, EXECUTE_TEST };
+
+            // An enumeration of the three reasons that a test might be aborted.
+            enum AbortReason {
+                TEST_ENCOUNTERED_RETURN_STATEMENT,
+                TEST_THREW_EXCEPTION,
+                TEST_DID_NOT_DIE
+            };
+
+            // Assumes one of the above roles.
+            virtual TestRole AssumeRole() = 0;
+
+            // Waits for the death test to finish and returns its status.
+            virtual int Wait() = 0;
+
+            // Returns true if the death test passed; that is, the test process
+            // exited during the test, its exit status matches a user-supplied
+            // predicate, and its stderr output matches a user-supplied regular
+            // expression.
+            // The user-supplied predicate may be a macro expression rather
+            // than a function pointer or functor, or else Wait and Passed could
+            // be combined.
+            virtual bool Passed(bool exit_status_ok) = 0;
+
+            // Signals that the death test did not die as expected.
+            virtual void Abort(AbortReason reason) = 0;
+
+            // Returns a human-readable outcome message regarding the outcome of
+            // the last death test.
+            static const char *LastMessage();
+
+            static void set_last_death_test_message(const std::string &message);
+
+        private:
+            // A string containing a description of the outcome of the last death test.
+            static std::string last_death_test_message_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(DeathTest);
+        };
+
+// Factory interface for death tests.  May be mocked out for testing.
+        class DeathTestFactory
+        {
+        public:
+            virtual ~DeathTestFactory() { }
+            virtual bool Create(const char *statement, const RE *regex,
+                                const char *file, int line, DeathTest **test) = 0;
+        };
+
+// A concrete DeathTestFactory implementation for normal use.
+        class DefaultDeathTestFactory : public DeathTestFactory
+        {
+        public:
+            virtual bool Create(const char *statement, const RE *regex,
+                                const char *file, int line, DeathTest **test);
+        };
+
+// Returns true if exit_status describes a process that was terminated
+// by a signal, or exited normally with a nonzero exit code.
+        GTEST_API_ bool ExitedUnsuccessfully(int exit_status);
+
+// Traps C++ exceptions escaping statement and reports them as test
+// failures. Note that trapping SEH exceptions is not implemented here.
+# if GTEST_HAS_EXCEPTIONS
+#  define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \
+  try { \
+    GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+  } catch (const ::std::exception& gtest_exception) { \
+    fprintf(\
+        stderr, \
+        "\n%s: Caught std::exception-derived exception escaping the " \
+        "death test statement. Exception message: %s\n", \
+        ::testing::internal::FormatFileLocation(__FILE__, __LINE__).c_str(), \
+        gtest_exception.what()); \
+    fflush(stderr); \
+    death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \
+  } catch (...) { \
+    death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \
+  }
+
+# else
+#  define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \
+  GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement)
+
+# endif
+
+// This macro is for implementing ASSERT_DEATH*, EXPECT_DEATH*,
+// ASSERT_EXIT*, and EXPECT_EXIT*.
+# define GTEST_DEATH_TEST_(statement, predicate, regex, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    const ::testing::internal::RE& gtest_regex = (regex); \
+    ::testing::internal::DeathTest* gtest_dt; \
+    if (!::testing::internal::DeathTest::Create(#statement, &gtest_regex, \
+        __FILE__, __LINE__, &gtest_dt)) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \
+    } \
+    if (gtest_dt != NULL) { \
+      ::testing::internal::scoped_ptr< ::testing::internal::DeathTest> \
+          gtest_dt_ptr(gtest_dt); \
+      switch (gtest_dt->AssumeRole()) { \
+        case ::testing::internal::DeathTest::OVERSEE_TEST: \
+          if (!gtest_dt->Passed(predicate(gtest_dt->Wait()))) { \
+            goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \
+          } \
+          break; \
+        case ::testing::internal::DeathTest::EXECUTE_TEST: { \
+          ::testing::internal::DeathTest::ReturnSentinel \
+              gtest_sentinel(gtest_dt); \
+          GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, gtest_dt); \
+          gtest_dt->Abort(::testing::internal::DeathTest::TEST_DID_NOT_DIE); \
+          break; \
+        } \
+        default: \
+          break; \
+      } \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__): \
+      fail(::testing::internal::DeathTest::LastMessage())
+// The symbol "fail" here expands to something into which a message
+// can be streamed.
+
+// This macro is for implementing ASSERT/EXPECT_DEBUG_DEATH when compiled in
+// NDEBUG mode. In this case we need the statements to be executed, the regex is
+// ignored, and the macro must accept a streamed message even though the message
+// is never printed.
+# define GTEST_EXECUTE_STATEMENT_(statement, regex) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+     GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+  } else \
+    ::testing::Message()
+
+// A class representing the parsed contents of the
+// --gtest_internal_run_death_test flag, as it existed when
+// RUN_ALL_TESTS was called.
+        class InternalRunDeathTestFlag
+        {
+        public:
+            InternalRunDeathTestFlag(const std::string &a_file,
+                                     int a_line,
+                                     int an_index,
+                                     int a_write_fd)
+                : file_(a_file), line_(a_line), index_(an_index),
+                  write_fd_(a_write_fd) {}
+
+            ~InternalRunDeathTestFlag()
+            {
+                if (write_fd_ >= 0)
+                    posix::Close(write_fd_);
+            }
+
+            const std::string &file() const
+            {
+                return file_;
+            }
+            int line() const
+            {
+                return line_;
+            }
+            int index() const
+            {
+                return index_;
+            }
+            int write_fd() const
+            {
+                return write_fd_;
+            }
+
+        private:
+            std::string file_;
+            int line_;
+            int index_;
+            int write_fd_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(InternalRunDeathTestFlag);
+        };
+
+// Returns a newly created InternalRunDeathTestFlag object with fields
+// initialized from the GTEST_FLAG(internal_run_death_test) flag if
+// the flag is specified; otherwise returns NULL.
+        InternalRunDeathTestFlag *ParseInternalRunDeathTestFlag();
+
+#else  // GTEST_HAS_DEATH_TEST
+
+// This macro is used for implementing macros such as
+// EXPECT_DEATH_IF_SUPPORTED and ASSERT_DEATH_IF_SUPPORTED on systems where
+// death tests are not supported. Those macros must compile on such systems
+// iff EXPECT_DEATH and ASSERT_DEATH compile with the same parameters on
+// systems that support death tests. This allows one to write such a macro
+// on a system that does not support death tests and be sure that it will
+// compile on a death-test supporting system.
+//
+// Parameters:
+//   statement -  A statement that a macro such as EXPECT_DEATH would test
+//                for program termination. This macro has to make sure this
+//                statement is compiled but not executed, to ensure that
+//                EXPECT_DEATH_IF_SUPPORTED compiles with a certain
+//                parameter iff EXPECT_DEATH compiles with it.
+//   regex     -  A regex that a macro such as EXPECT_DEATH would use to test
+//                the output of statement.  This parameter has to be
+//                compiled but not evaluated by this macro, to ensure that
+//                this macro only accepts expressions that a macro such as
+//                EXPECT_DEATH would accept.
+//   terminator - Must be an empty statement for EXPECT_DEATH_IF_SUPPORTED
+//                and a return statement for ASSERT_DEATH_IF_SUPPORTED.
+//                This ensures that ASSERT_DEATH_IF_SUPPORTED will not
+//                compile inside functions where ASSERT_DEATH doesn't
+//                compile.
+//
+//  The branch that has an always false condition is used to ensure that
+//  statement and regex are compiled (and thus syntactically correct) but
+//  never executed. The unreachable code macro protects the terminator
+//  statement from generating an 'unreachable code' warning in case
+//  statement unconditionally returns or throws. The Message constructor at
+//  the end allows the syntax of streaming additional messages into the
+//  macro, for compilational compatibility with EXPECT_DEATH/ASSERT_DEATH.
+# define GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, terminator) \
+    GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+    if (::testing::internal::AlwaysTrue()) { \
+      GTEST_LOG_(WARNING) \
+          << "Death tests are not supported on this platform.\n" \
+          << "Statement '" #statement "' cannot be verified."; \
+    } else if (::testing::internal::AlwaysFalse()) { \
+      ::testing::internal::RE::PartialMatch(".*", (regex)); \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+      terminator; \
+    } else \
+      ::testing::Message()
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+
+namespace testing
+{
+
+// This flag controls the style of death tests.  Valid values are "threadsafe",
+// meaning that the death test child process will re-execute the test binary
+// from the start, running only a single death test, or "fast",
+// meaning that the child process will execute the test logic immediately
+// after forking.
+    GTEST_DECLARE_string_(death_test_style);
+
+#if GTEST_HAS_DEATH_TEST
+
+    namespace internal
+    {
+
+// Returns a Boolean value indicating whether the caller is currently
+// executing in the context of the death test child process.  Tools such as
+// Valgrind heap checkers may need this to modify their behavior in death
+// tests.  IMPORTANT: This is an internal utility.  Using it may break the
+// implementation of death tests.  User code MUST NOT use it.
+        GTEST_API_ bool InDeathTestChild();
+
+    }  // namespace internal
+
+// The following macros are useful for writing death tests.
+
+// Here's what happens when an ASSERT_DEATH* or EXPECT_DEATH* is
+// executed:
+//
+//   1. It generates a warning if there is more than one active
+//   thread.  This is because it's safe to fork() or clone() only
+//   when there is a single thread.
+//
+//   2. The parent process clone()s a sub-process and runs the death
+//   test in it; the sub-process exits with code 0 at the end of the
+//   death test, if it hasn't exited already.
+//
+//   3. The parent process waits for the sub-process to terminate.
+//
+//   4. The parent process checks the exit code and error message of
+//   the sub-process.
+//
+// Examples:
+//
+//   ASSERT_DEATH(server.SendMessage(56, "Hello"), "Invalid port number");
+//   for (int i = 0; i < 5; i++) {
+//     EXPECT_DEATH(server.ProcessRequest(i),
+//                  "Invalid request .* in ProcessRequest()")
+//                  << "Failed to die on request " << i;
+//   }
+//
+//   ASSERT_EXIT(server.ExitNow(), ::testing::ExitedWithCode(0), "Exiting");
+//
+//   bool KilledBySIGHUP(int exit_code) {
+//     return WIFSIGNALED(exit_code) && WTERMSIG(exit_code) == SIGHUP;
+//   }
+//
+//   ASSERT_EXIT(client.HangUpServer(), KilledBySIGHUP, "Hanging up!");
+//
+// On the regular expressions used in death tests:
+//
+//   On POSIX-compliant systems (*nix), we use the <regex.h> library,
+//   which uses the POSIX extended regex syntax.
+//
+//   On other platforms (e.g. Windows), we only support a simple regex
+//   syntax implemented as part of Google Test.  This limited
+//   implementation should be enough most of the time when writing
+//   death tests; though it lacks many features you can find in PCRE
+//   or POSIX extended regex syntax.  For example, we don't support
+//   union ("x|y"), grouping ("(xy)"), brackets ("[xy]"), and
+//   repetition count ("x{5,7}"), among others.
+//
+//   Below is the syntax that we do support.  We chose it to be a
+//   subset of both PCRE and POSIX extended regex, so it's easy to
+//   learn wherever you come from.  In the following: 'A' denotes a
+//   literal character, period (.), or a single \\ escape sequence;
+//   'x' and 'y' denote regular expressions; 'm' and 'n' are for
+//   natural numbers.
+//
+//     c     matches any literal character c
+//     \\d   matches any decimal digit
+//     \\D   matches any character that's not a decimal digit
+//     \\f   matches \f
+//     \\n   matches \n
+//     \\r   matches \r
+//     \\s   matches any ASCII whitespace, including \n
+//     \\S   matches any character that's not a whitespace
+//     \\t   matches \t
+//     \\v   matches \v
+//     \\w   matches any letter, _, or decimal digit
+//     \\W   matches any character that \\w doesn't match
+//     \\c   matches any literal character c, which must be a punctuation
+//     .     matches any single character except \n
+//     A?    matches 0 or 1 occurrences of A
+//     A*    matches 0 or many occurrences of A
+//     A+    matches 1 or many occurrences of A
+//     ^     matches the beginning of a string (not that of each line)
+//     $     matches the end of a string (not that of each line)
+//     xy    matches x followed by y
+//
+//   If you accidentally use PCRE or POSIX extended regex features
+//   not implemented by us, you will get a run-time failure.  In that
+//   case, please try to rewrite your regular expression within the
+//   above syntax.
+//
+//   This implementation is *not* meant to be as highly tuned or robust
+//   as a compiled regex library, but should perform well enough for a
+//   death test, which already incurs significant overhead by launching
+//   a child process.
+//
+// Known caveats:
+//
+//   A "threadsafe" style death test obtains the path to the test
+//   program from argv[0] and re-executes it in the sub-process.  For
+//   simplicity, the current implementation doesn't search the PATH
+//   when launching the sub-process.  This means that the user must
+//   invoke the test program via a path that contains at least one
+//   path separator (e.g. path/to/foo_test and
+//   /absolute/path/to/bar_test are fine, but foo_test is not).  This
+//   is rarely a problem as people usually don't put the test binary
+//   directory in PATH.
+//
+// TODO(wan@google.com): make thread-safe death tests search the PATH.
+
+// Asserts that a given statement causes the program to exit, with an
+// integer exit status that satisfies predicate, and emitting error output
+// that matches regex.
+# define ASSERT_EXIT(statement, predicate, regex) \
+    GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_FATAL_FAILURE_)
+
+// Like ASSERT_EXIT, but continues on to successive tests in the
+// test case, if any:
+# define EXPECT_EXIT(statement, predicate, regex) \
+    GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_NONFATAL_FAILURE_)
+
+// Asserts that a given statement causes the program to exit, either by
+// explicitly exiting with a nonzero exit code or being killed by a
+// signal, and emitting error output that matches regex.
+# define ASSERT_DEATH(statement, regex) \
+    ASSERT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex)
+
+// Like ASSERT_DEATH, but continues on to successive tests in the
+// test case, if any:
+# define EXPECT_DEATH(statement, regex) \
+    EXPECT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex)
+
+// Two predicate classes that can be used in {ASSERT,EXPECT}_EXIT*:
+
+// Tests that an exit code describes a normal exit with a given exit code.
+    class GTEST_API_ ExitedWithCode
+    {
+    public:
+        explicit ExitedWithCode(int exit_code);
+        bool operator()(int exit_status) const;
+    private:
+        // No implementation - assignment is unsupported.
+        void operator=(const ExitedWithCode &other);
+
+        const int exit_code_;
+    };
+
+# if !GTEST_OS_WINDOWS
+// Tests that an exit code describes an exit due to termination by a
+// given signal.
+    class GTEST_API_ KilledBySignal
+    {
+    public:
+        explicit KilledBySignal(int signum);
+        bool operator()(int exit_status) const;
+    private:
+        const int signum_;
+    };
+# endif  // !GTEST_OS_WINDOWS
+
+// EXPECT_DEBUG_DEATH asserts that the given statements die in debug mode.
+// The death testing framework causes this to have interesting semantics,
+// since the sideeffects of the call are only visible in opt mode, and not
+// in debug mode.
+//
+// In practice, this can be used to test functions that utilize the
+// LOG(DFATAL) macro using the following style:
+//
+// int DieInDebugOr12(int* sideeffect) {
+//   if (sideeffect) {
+//     *sideeffect = 12;
+//   }
+//   LOG(DFATAL) << "death";
+//   return 12;
+// }
+//
+// TEST(TestCase, TestDieOr12WorksInDgbAndOpt) {
+//   int sideeffect = 0;
+//   // Only asserts in dbg.
+//   EXPECT_DEBUG_DEATH(DieInDebugOr12(&sideeffect), "death");
+//
+// #ifdef NDEBUG
+//   // opt-mode has sideeffect visible.
+//   EXPECT_EQ(12, sideeffect);
+// #else
+//   // dbg-mode no visible sideeffect.
+//   EXPECT_EQ(0, sideeffect);
+// #endif
+// }
+//
+// This will assert that DieInDebugReturn12InOpt() crashes in debug
+// mode, usually due to a DCHECK or LOG(DFATAL), but returns the
+// appropriate fallback value (12 in this case) in opt mode. If you
+// need to test that a function has appropriate side-effects in opt
+// mode, include assertions against the side-effects.  A general
+// pattern for this is:
+//
+// EXPECT_DEBUG_DEATH({
+//   // Side-effects here will have an effect after this statement in
+//   // opt mode, but none in debug mode.
+//   EXPECT_EQ(12, DieInDebugOr12(&sideeffect));
+// }, "death");
+//
+# ifdef NDEBUG
+
+#  define EXPECT_DEBUG_DEATH(statement, regex) \
+  GTEST_EXECUTE_STATEMENT_(statement, regex)
+
+#  define ASSERT_DEBUG_DEATH(statement, regex) \
+  GTEST_EXECUTE_STATEMENT_(statement, regex)
+
+# else
+
+#  define EXPECT_DEBUG_DEATH(statement, regex) \
+  EXPECT_DEATH(statement, regex)
+
+#  define ASSERT_DEBUG_DEATH(statement, regex) \
+  ASSERT_DEATH(statement, regex)
+
+# endif  // NDEBUG for EXPECT_DEBUG_DEATH
+#endif  // GTEST_HAS_DEATH_TEST
+
+// EXPECT_DEATH_IF_SUPPORTED(statement, regex) and
+// ASSERT_DEATH_IF_SUPPORTED(statement, regex) expand to real death tests if
+// death tests are supported; otherwise they just issue a warning.  This is
+// useful when you are combining death test assertions with normal test
+// assertions in one test.
+#if GTEST_HAS_DEATH_TEST
+# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \
+    EXPECT_DEATH(statement, regex)
+# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \
+    ASSERT_DEATH(statement, regex)
+#else
+# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \
+    GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, )
+# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \
+    GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, return)
+#endif
+
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+// This file was GENERATED by command:
+//     pump.py gtest-param-test.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: vladl@google.com (Vlad Losev)
+//
+// Macros and functions for implementing parameterized tests
+// in Google C++ Testing Framework (Google Test)
+//
+// This file is generated by a SCRIPT.  DO NOT EDIT BY HAND!
+//
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+
+
+// Value-parameterized tests allow you to test your code with different
+// parameters without writing multiple copies of the same test.
+//
+// Here is how you use value-parameterized tests:
+
+#if 0
+
+// To write value-parameterized tests, first you should define a fixture
+// class. It is usually derived from testing::TestWithParam<T> (see below for
+// another inheritance scheme that's sometimes useful in more complicated
+// class hierarchies), where the type of your parameter values.
+// TestWithParam<T> is itself derived from testing::Test. T can be any
+// copyable type. If it's a raw pointer, you are responsible for managing the
+// lifespan of the pointed values.
+
+class FooTest : public ::testing::TestWithParam<const char *>
+{
+    // You can implement all the usual class fixture members here.
+};
+
+// Then, use the TEST_P macro to define as many parameterized tests
+// for this fixture as you want. The _P suffix is for "parameterized"
+// or "pattern", whichever you prefer to think.
+
+TEST_P(FooTest, DoesBlah)
+{
+    // Inside a test, access the test parameter with the GetParam() method
+    // of the TestWithParam<T> class:
+    EXPECT_TRUE(foo.Blah(GetParam()));
+    ...
+}
+
+TEST_P(FooTest, HasBlahBlah)
+{
+    ...
+}
+
+// Finally, you can use INSTANTIATE_TEST_CASE_P to instantiate the test
+// case with any set of parameters you want. Google Test defines a number
+// of functions for generating test parameters. They return what we call
+// (surprise!) parameter generators. Here is a  summary of them, which
+// are all in the testing namespace:
+//
+//
+//  Range(begin, end [, step]) - Yields values {begin, begin+step,
+//                               begin+step+step, ...}. The values do not
+//                               include end. step defaults to 1.
+//  Values(v1, v2, ..., vN)    - Yields values {v1, v2, ..., vN}.
+//  ValuesIn(container)        - Yields values from a C-style array, an STL
+//  ValuesIn(begin,end)          container, or an iterator range [begin, end).
+//  Bool()                     - Yields sequence {false, true}.
+//  Combine(g1, g2, ..., gN)   - Yields all combinations (the Cartesian product
+//                               for the math savvy) of the values generated
+//                               by the N generators.
+//
+// For more details, see comments at the definitions of these functions below
+// in this file.
+//
+// The following statement will instantiate tests from the FooTest test case
+// each with parameter values "meeny", "miny", and "moe".
+
+INSTANTIATE_TEST_CASE_P(InstantiationName,
+                        FooTest,
+                        Values("meeny", "miny", "moe"));
+
+// To distinguish different instances of the pattern, (yes, you
+// can instantiate it more then once) the first argument to the
+// INSTANTIATE_TEST_CASE_P macro is a prefix that will be added to the
+// actual test case name. Remember to pick unique prefixes for different
+// instantiations. The tests from the instantiation above will have
+// these names:
+//
+//    * InstantiationName/FooTest.DoesBlah/0 for "meeny"
+//    * InstantiationName/FooTest.DoesBlah/1 for "miny"
+//    * InstantiationName/FooTest.DoesBlah/2 for "moe"
+//    * InstantiationName/FooTest.HasBlahBlah/0 for "meeny"
+//    * InstantiationName/FooTest.HasBlahBlah/1 for "miny"
+//    * InstantiationName/FooTest.HasBlahBlah/2 for "moe"
+//
+// You can use these names in --gtest_filter.
+//
+// This statement will instantiate all tests from FooTest again, each
+// with parameter values "cat" and "dog":
+
+const char *pets[] = {"cat", "dog"};
+INSTANTIATE_TEST_CASE_P(AnotherInstantiationName, FooTest, ValuesIn(pets));
+
+// The tests from the instantiation above will have these names:
+//
+//    * AnotherInstantiationName/FooTest.DoesBlah/0 for "cat"
+//    * AnotherInstantiationName/FooTest.DoesBlah/1 for "dog"
+//    * AnotherInstantiationName/FooTest.HasBlahBlah/0 for "cat"
+//    * AnotherInstantiationName/FooTest.HasBlahBlah/1 for "dog"
+//
+// Please note that INSTANTIATE_TEST_CASE_P will instantiate all tests
+// in the given test case, whether their definitions come before or
+// AFTER the INSTANTIATE_TEST_CASE_P statement.
+//
+// Please also note that generator expressions (including parameters to the
+// generators) are evaluated in InitGoogleTest(), after main() has started.
+// This allows the user on one hand, to adjust generator parameters in order
+// to dynamically determine a set of tests to run and on the other hand,
+// give the user a chance to inspect the generated tests with Google Test
+// reflection API before RUN_ALL_TESTS() is executed.
+//
+// You can see samples/sample7_unittest.cc and samples/sample8_unittest.cc
+// for more examples.
+//
+// In the future, we plan to publish the API for defining new parameter
+// generators. But for now this interface remains part of the internal
+// implementation and is subject to change.
+//
+//
+// A parameterized test fixture must be derived from testing::Test and from
+// testing::WithParamInterface<T>, where T is the type of the parameter
+// values. Inheriting from TestWithParam<T> satisfies that requirement because
+// TestWithParam<T> inherits from both Test and WithParamInterface. In more
+// complicated hierarchies, however, it is occasionally useful to inherit
+// separately from Test and WithParamInterface. For example:
+
+class BaseTest : public ::testing::Test
+{
+    // You can inherit all the usual members for a non-parameterized test
+    // fixture here.
+};
+
+class DerivedTest : public BaseTest, public ::testing::WithParamInterface<int>
+{
+    // The usual test fixture members go here too.
+};
+
+TEST_F(BaseTest, HasFoo)
+{
+    // This is an ordinary non-parameterized test.
+}
+
+TEST_P(DerivedTest, DoesBlah)
+{
+    // GetParam works just the same here as if you inherit from TestWithParam.
+    EXPECT_TRUE(foo.Blah(GetParam()));
+}
+
+#endif  // 0
+
+
+#if !GTEST_OS_SYMBIAN
+# include <utility>
+#endif
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*.  Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// Type and function utilities for implementing parameterized tests.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+
+#include <ctype.h>
+
+#include <iterator>
+#include <set>
+#include <utility>
+#include <vector>
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*.  Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+// Copyright 2003 Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: Dan Egnor (egnor@google.com)
+//
+// A "smart" pointer type with reference tracking.  Every pointer to a
+// particular object is kept on a circular linked list.  When the last pointer
+// to an object is destroyed or reassigned, the object is deleted.
+//
+// Used properly, this deletes the object when the last reference goes away.
+// There are several caveats:
+// - Like all reference counting schemes, cycles lead to leaks.
+// - Each smart pointer is actually two pointers (8 bytes instead of 4).
+// - Every time a pointer is assigned, the entire list of pointers to that
+//   object is traversed.  This class is therefore NOT SUITABLE when there
+//   will often be more than two or three pointers to a particular object.
+// - References are only tracked as long as linked_ptr<> objects are copied.
+//   If a linked_ptr<> is converted to a raw pointer and back, BAD THINGS
+//   will happen (double deletion).
+//
+// A good use of this class is storing object references in STL containers.
+// You can safely put linked_ptr<> in a vector<>.
+// Other uses may not be as good.
+//
+// Note: If you use an incomplete type with linked_ptr<>, the class
+// *containing* linked_ptr<> must have a constructor and destructor (even
+// if they do nothing!).
+//
+// Bill Gibbons suggested we use something like this.
+//
+// Thread Safety:
+//   Unlike other linked_ptr implementations, in this implementation
+//   a linked_ptr object is thread-safe in the sense that:
+//     - it's safe to copy linked_ptr objects concurrently,
+//     - it's safe to copy *from* a linked_ptr and read its underlying
+//       raw pointer (e.g. via get()) concurrently, and
+//     - it's safe to write to two linked_ptrs that point to the same
+//       shared object concurrently.
+// TODO(wan@google.com): rename this to safe_linked_ptr to avoid
+// confusion with normal linked_ptr.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+
+#include <stdlib.h>
+#include <assert.h>
+
+
+namespace testing
+{
+    namespace internal
+    {
+
+// Protects copying of all linked_ptr objects.
+        GTEST_API_ GTEST_DECLARE_STATIC_MUTEX_(g_linked_ptr_mutex);
+
+// This is used internally by all instances of linked_ptr<>.  It needs to be
+// a non-template class because different types of linked_ptr<> can refer to
+// the same object (linked_ptr<Superclass>(obj) vs linked_ptr<Subclass>(obj)).
+// So, it needs to be possible for different types of linked_ptr to participate
+// in the same circular linked list, so we need a single class type here.
+//
+// DO NOT USE THIS CLASS DIRECTLY YOURSELF.  Use linked_ptr<T>.
+        class linked_ptr_internal
+        {
+        public:
+            // Create a new circle that includes only this instance.
+            void join_new()
+            {
+                next_ = this;
+            }
+
+            // Many linked_ptr operations may change p.link_ for some linked_ptr
+            // variable p in the same circle as this object.  Therefore we need
+            // to prevent two such operations from occurring concurrently.
+            //
+            // Note that different types of linked_ptr objects can coexist in a
+            // circle (e.g. linked_ptr<Base>, linked_ptr<Derived1>, and
+            // linked_ptr<Derived2>).  Therefore we must use a single mutex to
+            // protect all linked_ptr objects.  This can create serious
+            // contention in production code, but is acceptable in a testing
+            // framework.
+
+            // Join an existing circle.
+            void join(linked_ptr_internal const *ptr)
+            GTEST_LOCK_EXCLUDED_(g_linked_ptr_mutex)
+            {
+                MutexLock lock(&g_linked_ptr_mutex);
+
+                linked_ptr_internal const *p = ptr;
+                while (p->next_ != ptr) {
+                    assert(p->next_ != this &&
+                           "Trying to join() a linked ring we are already in. "
+                           "Is GMock thread safety enabled?");
+                    p = p->next_;
+                }
+                p->next_ = this;
+                next_ = ptr;
+            }
+
+            // Leave whatever circle we're part of.  Returns true if we were the
+            // last member of the circle.  Once this is done, you can join() another.
+            bool depart()
+            GTEST_LOCK_EXCLUDED_(g_linked_ptr_mutex)
+            {
+                MutexLock lock(&g_linked_ptr_mutex);
+
+                if (next_ == this) return true;
+                linked_ptr_internal const *p = next_;
+                while (p->next_ != this) {
+                    assert(p->next_ != next_ &&
+                           "Trying to depart() a linked ring we are not in. "
+                           "Is GMock thread safety enabled?");
+                    p = p->next_;
+                }
+                p->next_ = next_;
+                return false;
+            }
+
+        private:
+            mutable linked_ptr_internal const *next_;
+        };
+
+        template <typename T>
+        class linked_ptr
+        {
+        public:
+            typedef T element_type;
+
+            // Take over ownership of a raw pointer.  This should happen as soon as
+            // possible after the object is created.
+            explicit linked_ptr(T *ptr = NULL)
+            {
+                capture(ptr);
+            }
+            ~linked_ptr()
+            {
+                depart();
+            }
+
+            // Copy an existing linked_ptr<>, adding ourselves to the list of references.
+            template <typename U> linked_ptr(linked_ptr<U> const &ptr)
+            {
+                copy(&ptr);
+            }
+            linked_ptr(linked_ptr const &ptr)    // NOLINT
+            {
+                assert(&ptr != this);
+                copy(&ptr);
+            }
+
+            // Assignment releases the old value and acquires the new.
+            template <typename U> linked_ptr &operator=(linked_ptr<U> const &ptr)
+            {
+                depart();
+                copy(&ptr);
+                return *this;
+            }
+
+            linked_ptr &operator=(linked_ptr const &ptr)
+            {
+                if (&ptr != this) {
+                    depart();
+                    copy(&ptr);
+                }
+                return *this;
+            }
+
+            // Smart pointer members.
+            void reset(T *ptr = NULL)
+            {
+                depart();
+                capture(ptr);
+            }
+            T *get() const
+            {
+                return value_;
+            }
+            T *operator->() const
+            {
+                return value_;
+            }
+            T &operator*() const
+            {
+                return *value_;
+            }
+
+            bool operator==(T *p) const
+            {
+                return value_ == p;
+            }
+            bool operator!=(T *p) const
+            {
+                return value_ != p;
+            }
+            template <typename U>
+            bool operator==(linked_ptr<U> const &ptr) const
+            {
+                return value_ == ptr.get();
+            }
+            template <typename U>
+            bool operator!=(linked_ptr<U> const &ptr) const
+            {
+                return value_ != ptr.get();
+            }
+
+        private:
+            template <typename U>
+            friend class linked_ptr;
+
+            T *value_;
+            linked_ptr_internal link_;
+
+            void depart()
+            {
+                if (link_.depart()) delete value_;
+            }
+
+            void capture(T *ptr)
+            {
+                value_ = ptr;
+                link_.join_new();
+            }
+
+            template <typename U> void copy(linked_ptr<U> const *ptr)
+            {
+                value_ = ptr->get();
+                if (value_)
+                    link_.join(&ptr->link_);
+                else
+                    link_.join_new();
+            }
+        };
+
+        template<typename T> inline
+        bool operator==(T *ptr, const linked_ptr<T> &x)
+        {
+            return ptr == x.get();
+        }
+
+        template<typename T> inline
+        bool operator!=(T *ptr, const linked_ptr<T> &x)
+        {
+            return ptr != x.get();
+        }
+
+// A function to convert T* into linked_ptr<T>
+// Doing e.g. make_linked_ptr(new FooBarBaz<type>(arg)) is a shorter notation
+// for linked_ptr<FooBarBaz<type> >(new FooBarBaz<type>(arg))
+        template <typename T>
+        linked_ptr<T> make_linked_ptr(T *ptr)
+        {
+            return linked_ptr<T>(ptr);
+        }
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Google Test - The Google C++ Testing Framework
+//
+// This file implements a universal value printer that can print a
+// value of any type T:
+//
+//   void ::testing::internal::UniversalPrinter<T>::Print(value, ostream_ptr);
+//
+// A user can teach this function how to print a class type T by
+// defining either operator<<() or PrintTo() in the namespace that
+// defines T.  More specifically, the FIRST defined function in the
+// following list will be used (assuming T is defined in namespace
+// foo):
+//
+//   1. foo::PrintTo(const T&, ostream*)
+//   2. operator<<(ostream&, const T&) defined in either foo or the
+//      global namespace.
+//
+// If none of the above is defined, it will print the debug string of
+// the value if it is a protocol buffer, or print the raw bytes in the
+// value otherwise.
+//
+// To aid debugging: when T is a reference type, the address of the
+// value is also printed; when T is a (const) char pointer, both the
+// pointer value and the NUL-terminated string it points to are
+// printed.
+//
+// We also provide some convenient wrappers:
+//
+//   // Prints a value to a string.  For a (const or not) char
+//   // pointer, the NUL-terminated string (but not the pointer) is
+//   // printed.
+//   std::string ::testing::PrintToString(const T& value);
+//
+//   // Prints a value tersely: for a reference type, the referenced
+//   // value (but not the address) is printed; for a (const or not) char
+//   // pointer, the NUL-terminated string (but not the pointer) is
+//   // printed.
+//   void ::testing::internal::UniversalTersePrint(const T& value, ostream*);
+//
+//   // Prints value using the type inferred by the compiler.  The difference
+//   // from UniversalTersePrint() is that this function prints both the
+//   // pointer and the NUL-terminated string for a (const or not) char pointer.
+//   void ::testing::internal::UniversalPrint(const T& value, ostream*);
+//
+//   // Prints the fields of a tuple tersely to a string vector, one
+//   // element for each field. Tuple support must be enabled in
+//   // gtest-port.h.
+//   std::vector<string> UniversalTersePrintTupleFieldsToStrings(
+//       const Tuple& value);
+//
+// Known limitation:
+//
+// The print primitives print the elements of an STL-style container
+// using the compiler-inferred type of *iter where iter is a
+// const_iterator of the container.  When const_iterator is an input
+// iterator but not a forward iterator, this inferred type may not
+// match value_type, and the print output may be incorrect.  In
+// practice, this is rarely a problem as for most containers
+// const_iterator is a forward iterator.  We'll fix this if there's an
+// actual need for it.  Note that this fix cannot rely on value_type
+// being defined as many user-defined container types don't have
+// value_type.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+
+#include <ostream>  // NOLINT
+#include <sstream>
+#include <string>
+#include <utility>
+#include <vector>
+
+#if GTEST_HAS_STD_TUPLE_
+# include <tuple>
+#endif
+
+namespace testing
+{
+
+// Definitions in the 'internal' and 'internal2' name spaces are
+// subject to change without notice.  DO NOT USE THEM IN USER CODE!
+    namespace internal2
+    {
+
+// Prints the given number of bytes in the given object to the given
+// ostream.
+        GTEST_API_ void PrintBytesInObjectTo(const unsigned char *obj_bytes,
+                                             size_t count,
+                                             ::std::ostream *os);
+
+// For selecting which printer to use when a given type has neither <<
+// nor PrintTo().
+        enum TypeKind {
+            kProtobuf,              // a protobuf type
+            kConvertibleToInteger,  // a type implicitly convertible to BiggestInt
+            // (e.g. a named or unnamed enum type)
+            kOtherType              // anything else
+        };
+
+// TypeWithoutFormatter<T, kTypeKind>::PrintValue(value, os) is called
+// by the universal printer to print a value of type T when neither
+// operator<< nor PrintTo() is defined for T, where kTypeKind is the
+// "kind" of T as defined by enum TypeKind.
+        template <typename T, TypeKind kTypeKind>
+        class TypeWithoutFormatter
+        {
+        public:
+            // This default version is called when kTypeKind is kOtherType.
+            static void PrintValue(const T &value, ::std::ostream *os)
+            {
+                PrintBytesInObjectTo(reinterpret_cast<const unsigned char *>(&value),
+                                     sizeof(value), os);
+            }
+        };
+
+// We print a protobuf using its ShortDebugString() when the string
+// doesn't exceed this many characters; otherwise we print it using
+// DebugString() for better readability.
+        const size_t kProtobufOneLinerMaxLength = 50;
+
+        template <typename T>
+        class TypeWithoutFormatter<T, kProtobuf>
+        {
+        public:
+            static void PrintValue(const T &value, ::std::ostream *os)
+            {
+                std::string pretty_str = value.ShortDebugString();
+                if (pretty_str.length() > kProtobufOneLinerMaxLength) {
+                    pretty_str = "\n" + value.DebugString();
+                }
+                *os << ("<" + pretty_str + ">");
+            }
+        };
+
+        template <typename T>
+        class TypeWithoutFormatter<T, kConvertibleToInteger>
+        {
+        public:
+            // Since T has no << operator or PrintTo() but can be implicitly
+            // converted to BiggestInt, we print it as a BiggestInt.
+            //
+            // Most likely T is an enum type (either named or unnamed), in which
+            // case printing it as an integer is the desired behavior.  In case
+            // T is not an enum, printing it as an integer is the best we can do
+            // given that it has no user-defined printer.
+            static void PrintValue(const T &value, ::std::ostream *os)
+            {
+                const internal::BiggestInt kBigInt = value;
+                *os << kBigInt;
+            }
+        };
+
+// Prints the given value to the given ostream.  If the value is a
+// protocol message, its debug string is printed; if it's an enum or
+// of a type implicitly convertible to BiggestInt, it's printed as an
+// integer; otherwise the bytes in the value are printed.  This is
+// what UniversalPrinter<T>::Print() does when it knows nothing about
+// type T and T has neither << operator nor PrintTo().
+//
+// A user can override this behavior for a class type Foo by defining
+// a << operator in the namespace where Foo is defined.
+//
+// We put this operator in namespace 'internal2' instead of 'internal'
+// to simplify the implementation, as much code in 'internal' needs to
+// use << in STL, which would conflict with our own << were it defined
+// in 'internal'.
+//
+// Note that this operator<< takes a generic std::basic_ostream<Char,
+// CharTraits> type instead of the more restricted std::ostream.  If
+// we define it to take an std::ostream instead, we'll get an
+// "ambiguous overloads" compiler error when trying to print a type
+// Foo that supports streaming to std::basic_ostream<Char,
+// CharTraits>, as the compiler cannot tell whether
+// operator<<(std::ostream&, const T&) or
+// operator<<(std::basic_stream<Char, CharTraits>, const Foo&) is more
+// specific.
+        template <typename Char, typename CharTraits, typename T>
+        ::std::basic_ostream<Char, CharTraits> &operator<<(
+            ::std::basic_ostream<Char, CharTraits> &os, const T &x)
+        {
+            TypeWithoutFormatter<T,
+                                 (internal::IsAProtocolMessage<T>::value ? kProtobuf :
+                                  internal::ImplicitlyConvertible<const T &, internal::BiggestInt>::value ?
+                                  kConvertibleToInteger : kOtherType)>::PrintValue(x, &os);
+            return os;
+        }
+
+    }  // namespace internal2
+}  // namespace testing
+
+// This namespace MUST NOT BE NESTED IN ::testing, or the name look-up
+// magic needed for implementing UniversalPrinter won't work.
+namespace testing_internal
+{
+
+// Used to print a value that is not an STL-style container when the
+// user doesn't define PrintTo() for it.
+    template <typename T>
+    void DefaultPrintNonContainerTo(const T &value, ::std::ostream *os)
+    {
+        // With the following statement, during unqualified name lookup,
+        // testing::internal2::operator<< appears as if it was declared in
+        // the nearest enclosing namespace that contains both
+        // ::testing_internal and ::testing::internal2, i.e. the global
+        // namespace.  For more details, refer to the C++ Standard section
+        // 7.3.4-1 [namespace.udir].  This allows us to fall back onto
+        // testing::internal2::operator<< in case T doesn't come with a <<
+        // operator.
+        //
+        // We cannot write 'using ::testing::internal2::operator<<;', which
+        // gcc 3.3 fails to compile due to a compiler bug.
+        using namespace ::testing::internal2;  // NOLINT
+
+        // Assuming T is defined in namespace foo, in the next statement,
+        // the compiler will consider all of:
+        //
+        //   1. foo::operator<< (thanks to Koenig look-up),
+        //   2. ::operator<< (as the current namespace is enclosed in ::),
+        //   3. testing::internal2::operator<< (thanks to the using statement above).
+        //
+        // The operator<< whose type matches T best will be picked.
+        //
+        // We deliberately allow #2 to be a candidate, as sometimes it's
+        // impossible to define #1 (e.g. when foo is ::std, defining
+        // anything in it is undefined behavior unless you are a compiler
+        // vendor.).
+        *os << value;
+    }
+
+}  // namespace testing_internal
+
+namespace testing
+{
+    namespace internal
+    {
+
+// FormatForComparison<ToPrint, OtherOperand>::Format(value) formats a
+// value of type ToPrint that is an operand of a comparison assertion
+// (e.g. ASSERT_EQ).  OtherOperand is the type of the other operand in
+// the comparison, and is used to help determine the best way to
+// format the value.  In particular, when the value is a C string
+// (char pointer) and the other operand is an STL string object, we
+// want to format the C string as a string, since we know it is
+// compared by value with the string object.  If the value is a char
+// pointer but the other operand is not an STL string object, we don't
+// know whether the pointer is supposed to point to a NUL-terminated
+// string, and thus want to print it as a pointer to be safe.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+
+// The default case.
+        template <typename ToPrint, typename OtherOperand>
+        class FormatForComparison
+        {
+        public:
+            static ::std::string Format(const ToPrint &value)
+            {
+                return ::testing::PrintToString(value);
+            }
+        };
+
+// Array.
+        template <typename ToPrint, size_t N, typename OtherOperand>
+        class FormatForComparison<ToPrint[N], OtherOperand>
+        {
+        public:
+            static ::std::string Format(const ToPrint *value)
+            {
+                return FormatForComparison<const ToPrint *, OtherOperand>::Format(value);
+            }
+        };
+
+// By default, print C string as pointers to be safe, as we don't know
+// whether they actually point to a NUL-terminated string.
+
+#define GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(CharType)                \
+  template <typename OtherOperand>                                      \
+  class FormatForComparison<CharType*, OtherOperand> {                  \
+   public:                                                              \
+    static ::std::string Format(CharType* value) {                      \
+      return ::testing::PrintToString(static_cast<const void*>(value)); \
+    }                                                                   \
+  }
+
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(char);
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const char);
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(wchar_t);
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const wchar_t);
+
+#undef GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_
+
+// If a C string is compared with an STL string object, we know it's meant
+// to point to a NUL-terminated string, and thus can print it as a string.
+
+#define GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(CharType, OtherStringType) \
+  template <>                                                           \
+  class FormatForComparison<CharType*, OtherStringType> {               \
+   public:                                                              \
+    static ::std::string Format(CharType* value) {                      \
+      return ::testing::PrintToString(value);                           \
+    }                                                                   \
+  }
+
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(char, ::std::string);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const char, ::std::string);
+
+#if GTEST_HAS_GLOBAL_STRING
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(char, ::string);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const char, ::string);
+#endif
+
+#if GTEST_HAS_GLOBAL_WSTRING
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(wchar_t, ::wstring);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const wchar_t, ::wstring);
+#endif
+
+#if GTEST_HAS_STD_WSTRING
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(wchar_t, ::std::wstring);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const wchar_t, ::std::wstring);
+#endif
+
+#undef GTEST_IMPL_FORMAT_C_STRING_AS_STRING_
+
+// Formats a comparison assertion (e.g. ASSERT_EQ, EXPECT_LT, and etc)
+// operand to be used in a failure message.  The type (but not value)
+// of the other operand may affect the format.  This allows us to
+// print a char* as a raw pointer when it is compared against another
+// char* or void*, and print it as a C string when it is compared
+// against an std::string object, for example.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        template <typename T1, typename T2>
+        std::string FormatForComparisonFailureMessage(
+            const T1 &value, const T2 & /* other_operand */)
+        {
+            return FormatForComparison<T1, T2>::Format(value);
+        }
+
+// UniversalPrinter<T>::Print(value, ostream_ptr) prints the given
+// value to the given ostream.  The caller must ensure that
+// 'ostream_ptr' is not NULL, or the behavior is undefined.
+//
+// We define UniversalPrinter as a class template (as opposed to a
+// function template), as we need to partially specialize it for
+// reference types, which cannot be done with function templates.
+        template <typename T>
+        class UniversalPrinter;
+
+        template <typename T>
+        void UniversalPrint(const T &value, ::std::ostream *os);
+
+        enum DefaultPrinterType {
+            kPrintContainer,
+            kPrintPointer,
+            kPrintFunctionPointer,
+            kPrintOther,
+        };
+        template <DefaultPrinterType type> struct WrapPrinterType {};
+
+// Used to print an STL-style container when the user doesn't define
+// a PrintTo() for it.
+        template <typename C>
+        void DefaultPrintTo(WrapPrinterType<kPrintContainer> /* dummy */,
+                            const C &container, ::std::ostream *os)
+        {
+            const size_t kMaxCount = 32;  // The maximum number of elements to print.
+            *os << '{';
+            size_t count = 0;
+            for (typename C::const_iterator it = container.begin();
+                 it != container.end(); ++it, ++count) {
+                if (count > 0) {
+                    *os << ',';
+                    if (count == kMaxCount) {  // Enough has been printed.
+                        *os << " ...";
+                        break;
+                    }
+                }
+                *os << ' ';
+                // We cannot call PrintTo(*it, os) here as PrintTo() doesn't
+                // handle *it being a native array.
+                internal::UniversalPrint(*it, os);
+            }
+
+            if (count > 0) {
+                *os << ' ';
+            }
+            *os << '}';
+        }
+
+// Used to print a pointer that is neither a char pointer nor a member
+// pointer, when the user doesn't define PrintTo() for it.  (A member
+// variable pointer or member function pointer doesn't really point to
+// a location in the address space.  Their representation is
+// implementation-defined.  Therefore they will be printed as raw
+// bytes.)
+        template <typename T>
+        void DefaultPrintTo(WrapPrinterType<kPrintPointer> /* dummy */,
+                            T *p, ::std::ostream *os)
+        {
+            if (p == NULL) {
+                *os << "NULL";
+            } else {
+                // T is not a function type.  We just call << to print p,
+                // relying on ADL to pick up user-defined << for their pointer
+                // types, if any.
+                *os << p;
+            }
+        }
+        template <typename T>
+        void DefaultPrintTo(WrapPrinterType<kPrintFunctionPointer> /* dummy */,
+                            T *p, ::std::ostream *os)
+        {
+            if (p == NULL) {
+                *os << "NULL";
+            } else {
+                // T is a function type, so '*os << p' doesn't do what we want
+                // (it just prints p as bool).  We want to print p as a const
+                // void*.  However, we cannot cast it to const void* directly,
+                // even using reinterpret_cast, as earlier versions of gcc
+                // (e.g. 3.4.5) cannot compile the cast when p is a function
+                // pointer.  Casting to UInt64 first solves the problem.
+                *os << reinterpret_cast<const void *>(
+                        reinterpret_cast<internal::UInt64>(p));
+            }
+        }
+
+// Used to print a non-container, non-pointer value when the user
+// doesn't define PrintTo() for it.
+        template <typename T>
+        void DefaultPrintTo(WrapPrinterType<kPrintOther> /* dummy */,
+                            const T &value, ::std::ostream *os)
+        {
+            ::testing_internal::DefaultPrintNonContainerTo(value, os);
+        }
+
+// Prints the given value using the << operator if it has one;
+// otherwise prints the bytes in it.  This is what
+// UniversalPrinter<T>::Print() does when PrintTo() is not specialized
+// or overloaded for type T.
+//
+// A user can override this behavior for a class type Foo by defining
+// an overload of PrintTo() in the namespace where Foo is defined.  We
+// give the user this option as sometimes defining a << operator for
+// Foo is not desirable (e.g. the coding style may prevent doing it,
+// or there is already a << operator but it doesn't do what the user
+// wants).
+        template <typename T>
+        void PrintTo(const T &value, ::std::ostream *os)
+        {
+            // DefaultPrintTo() is overloaded.  The type of its first argument
+            // determines which version will be picked.
+            //
+            // Note that we check for container types here, prior to we check
+            // for protocol message types in our operator<<.  The rationale is:
+            //
+            // For protocol messages, we want to give people a chance to
+            // override Google Mock's format by defining a PrintTo() or
+            // operator<<.  For STL containers, other formats can be
+            // incompatible with Google Mock's format for the container
+            // elements; therefore we check for container types here to ensure
+            // that our format is used.
+            //
+            // Note that MSVC and clang-cl do allow an implicit conversion from
+            // pointer-to-function to pointer-to-object, but clang-cl warns on it.
+            // So don't use ImplicitlyConvertible if it can be helped since it will
+            // cause this warning, and use a separate overload of DefaultPrintTo for
+            // function pointers so that the `*os << p` in the object pointer overload
+            // doesn't cause that warning either.
+            DefaultPrintTo(
+                WrapPrinterType<sizeof(IsContainerTest<T>(0)) == sizeof(IsContainer)
+                ? kPrintContainer : !is_pointer<T>::value
+                ? kPrintOther
+#if GTEST_LANG_CXX11
+                : std::is_function<typename std::remove_pointer<T>::type>::value
+#else
+                : !internal::ImplicitlyConvertible<T, const void *>::value
+#endif
+                ? kPrintFunctionPointer
+                : kPrintPointer>(),
+                value, os);
+        }
+
+// The following list of PrintTo() overloads tells
+// UniversalPrinter<T>::Print() how to print standard types (built-in
+// types, strings, plain arrays, and pointers).
+
+// Overloads for various char types.
+        GTEST_API_ void PrintTo(unsigned char c, ::std::ostream *os);
+        GTEST_API_ void PrintTo(signed char c, ::std::ostream *os);
+        inline void PrintTo(char c, ::std::ostream *os)
+        {
+            // When printing a plain char, we always treat it as unsigned.  This
+            // way, the output won't be affected by whether the compiler thinks
+            // char is signed or not.
+            PrintTo(static_cast<unsigned char>(c), os);
+        }
+
+// Overloads for other simple built-in types.
+        inline void PrintTo(bool x, ::std::ostream *os)
+        {
+            *os << (x ? "true" : "false");
+        }
+
+// Overload for wchar_t type.
+// Prints a wchar_t as a symbol if it is printable or as its internal
+// code otherwise and also as its decimal code (except for L'\0').
+// The L'\0' char is printed as "L'\\0'". The decimal code is printed
+// as signed integer when wchar_t is implemented by the compiler
+// as a signed type and is printed as an unsigned integer when wchar_t
+// is implemented as an unsigned type.
+        GTEST_API_ void PrintTo(wchar_t wc, ::std::ostream *os);
+
+// Overloads for C strings.
+        GTEST_API_ void PrintTo(const char *s, ::std::ostream *os);
+        inline void PrintTo(char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const char *>(s), os);
+        }
+
+// signed/unsigned char is often used for representing binary data, so
+// we print pointers to it as void* to be safe.
+        inline void PrintTo(const signed char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+        inline void PrintTo(signed char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+        inline void PrintTo(const unsigned char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+        inline void PrintTo(unsigned char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+
+// MSVC can be configured to define wchar_t as a typedef of unsigned
+// short.  It defines _NATIVE_WCHAR_T_DEFINED when wchar_t is a native
+// type.  When wchar_t is a typedef, defining an overload for const
+// wchar_t* would cause unsigned short* be printed as a wide string,
+// possibly causing invalid memory accesses.
+#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED)
+// Overloads for wide C strings
+        GTEST_API_ void PrintTo(const wchar_t *s, ::std::ostream *os);
+        inline void PrintTo(wchar_t *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const wchar_t *>(s), os);
+        }
+#endif
+
+// Overload for C arrays.  Multi-dimensional arrays are printed
+// properly.
+
+// Prints the given number of elements in an array, without printing
+// the curly braces.
+        template <typename T>
+        void PrintRawArrayTo(const T a[], size_t count, ::std::ostream *os)
+        {
+            UniversalPrint(a[0], os);
+            for (size_t i = 1; i != count; i++) {
+                *os << ", ";
+                UniversalPrint(a[i], os);
+            }
+        }
+
+// Overloads for ::string and ::std::string.
+#if GTEST_HAS_GLOBAL_STRING
+        GTEST_API_ void PrintStringTo(const ::string &s, ::std::ostream *os);
+        inline void PrintTo(const ::string &s, ::std::ostream *os)
+        {
+            PrintStringTo(s, os);
+        }
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+        GTEST_API_ void PrintStringTo(const ::std::string &s, ::std::ostream *os);
+        inline void PrintTo(const ::std::string &s, ::std::ostream *os)
+        {
+            PrintStringTo(s, os);
+        }
+
+// Overloads for ::wstring and ::std::wstring.
+#if GTEST_HAS_GLOBAL_WSTRING
+        GTEST_API_ void PrintWideStringTo(const ::wstring &s, ::std::ostream *os);
+        inline void PrintTo(const ::wstring &s, ::std::ostream *os)
+        {
+            PrintWideStringTo(s, os);
+        }
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+#if GTEST_HAS_STD_WSTRING
+        GTEST_API_ void PrintWideStringTo(const ::std::wstring &s, ::std::ostream *os);
+        inline void PrintTo(const ::std::wstring &s, ::std::ostream *os)
+        {
+            PrintWideStringTo(s, os);
+        }
+#endif  // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+// Helper function for printing a tuple.  T must be instantiated with
+// a tuple type.
+        template <typename T>
+        void PrintTupleTo(const T &t, ::std::ostream *os);
+#endif  // GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+
+#if GTEST_HAS_TR1_TUPLE
+// Overload for ::std::tr1::tuple.  Needed for printing function arguments,
+// which are packed as tuples.
+
+// Overloaded PrintTo() for tuples of various arities.  We support
+// tuples of up-to 10 fields.  The following implementation works
+// regardless of whether tr1::tuple is implemented using the
+// non-standard variadic template feature or not.
+
+        inline void PrintTo(const ::std::tr1::tuple<> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1>
+        void PrintTo(const ::std::tr1::tuple<T1> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2>
+        void PrintTo(const ::std::tr1::tuple<T1, T2> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        void PrintTo(
+            const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> &t,
+            ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+#endif  // GTEST_HAS_TR1_TUPLE
+
+#if GTEST_HAS_STD_TUPLE_
+        template <typename... Types>
+        void PrintTo(const ::std::tuple<Types...> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+#endif  // GTEST_HAS_STD_TUPLE_
+
+// Overload for std::pair.
+        template <typename T1, typename T2>
+        void PrintTo(const ::std::pair<T1, T2> &value, ::std::ostream *os)
+        {
+            *os << '(';
+            // We cannot use UniversalPrint(value.first, os) here, as T1 may be
+            // a reference type.  The same for printing value.second.
+            UniversalPrinter<T1>::Print(value.first, os);
+            *os << ", ";
+            UniversalPrinter<T2>::Print(value.second, os);
+            *os << ')';
+        }
+
+// Implements printing a non-reference type T by letting the compiler
+// pick the right overload of PrintTo() for T.
+        template <typename T>
+        class UniversalPrinter
+        {
+        public:
+            // MSVC warns about adding const to a function type, so we want to
+            // disable the warning.
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180)
+
+            // Note: we deliberately don't call this PrintTo(), as that name
+            // conflicts with ::testing::internal::PrintTo in the body of the
+            // function.
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                // By default, ::testing::internal::PrintTo() is used for printing
+                // the value.
+                //
+                // Thanks to Koenig look-up, if T is a class and has its own
+                // PrintTo() function defined in its namespace, that function will
+                // be visible here.  Since it is more specific than the generic ones
+                // in ::testing::internal, it will be picked by the compiler in the
+                // following statement - exactly what we want.
+                PrintTo(value, os);
+            }
+
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+        };
+
+// UniversalPrintArray(begin, len, os) prints an array of 'len'
+// elements, starting at address 'begin'.
+        template <typename T>
+        void UniversalPrintArray(const T *begin, size_t len, ::std::ostream *os)
+        {
+            if (len == 0) {
+                *os << "{}";
+            } else {
+                *os << "{ ";
+                const size_t kThreshold = 18;
+                const size_t kChunkSize = 8;
+                // If the array has more than kThreshold elements, we'll have to
+                // omit some details by printing only the first and the last
+                // kChunkSize elements.
+                // TODO(wan@google.com): let the user control the threshold using a flag.
+                if (len <= kThreshold) {
+                    PrintRawArrayTo(begin, len, os);
+                } else {
+                    PrintRawArrayTo(begin, kChunkSize, os);
+                    *os << ", ..., ";
+                    PrintRawArrayTo(begin + len - kChunkSize, kChunkSize, os);
+                }
+                *os << " }";
+            }
+        }
+// This overload prints a (const) char array compactly.
+        GTEST_API_ void UniversalPrintArray(
+            const char *begin, size_t len, ::std::ostream *os);
+
+// This overload prints a (const) wchar_t array compactly.
+        GTEST_API_ void UniversalPrintArray(
+            const wchar_t *begin, size_t len, ::std::ostream *os);
+
+// Implements printing an array type T[N].
+        template <typename T, size_t N>
+        class UniversalPrinter<T[N]>
+        {
+        public:
+            // Prints the given array, omitting some elements when there are too
+            // many.
+            static void Print(const T (&a)[N], ::std::ostream *os)
+            {
+                UniversalPrintArray(a, N, os);
+            }
+        };
+
+// Implements printing a reference type T&.
+        template <typename T>
+        class UniversalPrinter<T &>
+        {
+        public:
+            // MSVC warns about adding const to a function type, so we want to
+            // disable the warning.
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180)
+
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                // Prints the address of the value.  We use reinterpret_cast here
+                // as static_cast doesn't compile when T is a function type.
+                *os << "@" << reinterpret_cast<const void *>(&value) << " ";
+
+                // Then prints the value itself.
+                UniversalPrint(value, os);
+            }
+
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+        };
+
+// Prints a value tersely: for a reference type, the referenced value
+// (but not the address) is printed; for a (const) char pointer, the
+// NUL-terminated string (but not the pointer) is printed.
+
+        template <typename T>
+        class UniversalTersePrinter
+        {
+        public:
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                UniversalPrint(value, os);
+            }
+        };
+        template <typename T>
+        class UniversalTersePrinter<T &>
+        {
+        public:
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                UniversalPrint(value, os);
+            }
+        };
+        template <typename T, size_t N>
+        class UniversalTersePrinter<T[N]>
+        {
+        public:
+            static void Print(const T (&value)[N], ::std::ostream *os)
+            {
+                UniversalPrinter<T[N]>::Print(value, os);
+            }
+        };
+        template <>
+        class UniversalTersePrinter<const char *>
+        {
+        public:
+            static void Print(const char *str, ::std::ostream *os)
+            {
+                if (str == NULL) {
+                    *os << "NULL";
+                } else {
+                    UniversalPrint(std::string(str), os);
+                }
+            }
+        };
+        template <>
+        class UniversalTersePrinter<char *>
+        {
+        public:
+            static void Print(char *str, ::std::ostream *os)
+            {
+                UniversalTersePrinter<const char *>::Print(str, os);
+            }
+        };
+
+#if GTEST_HAS_STD_WSTRING
+        template <>
+        class UniversalTersePrinter<const wchar_t *>
+        {
+        public:
+            static void Print(const wchar_t *str, ::std::ostream *os)
+            {
+                if (str == NULL) {
+                    *os << "NULL";
+                } else {
+                    UniversalPrint(::std::wstring(str), os);
+                }
+            }
+        };
+#endif
+
+        template <>
+        class UniversalTersePrinter<wchar_t *>
+        {
+        public:
+            static void Print(wchar_t *str, ::std::ostream *os)
+            {
+                UniversalTersePrinter<const wchar_t *>::Print(str, os);
+            }
+        };
+
+        template <typename T>
+        void UniversalTersePrint(const T &value, ::std::ostream *os)
+        {
+            UniversalTersePrinter<T>::Print(value, os);
+        }
+
+// Prints a value using the type inferred by the compiler.  The
+// difference between this and UniversalTersePrint() is that for a
+// (const) char pointer, this prints both the pointer and the
+// NUL-terminated string.
+        template <typename T>
+        void UniversalPrint(const T &value, ::std::ostream *os)
+        {
+            // A workarond for the bug in VC++ 7.1 that prevents us from instantiating
+            // UniversalPrinter with T directly.
+            typedef T T1;
+            UniversalPrinter<T1>::Print(value, os);
+        }
+
+        typedef ::std::vector<string> Strings;
+
+// TuplePolicy<TupleT> must provide:
+// - tuple_size
+//     size of tuple TupleT.
+// - get<size_t I>(const TupleT& t)
+//     static function extracting element I of tuple TupleT.
+// - tuple_element<size_t I>::type
+//     type of element I of tuple TupleT.
+        template <typename TupleT>
+        struct TuplePolicy;
+
+#if GTEST_HAS_TR1_TUPLE
+        template <typename TupleT>
+        struct TuplePolicy {
+            typedef TupleT Tuple;
+            static const size_t tuple_size = ::std::tr1::tuple_size<Tuple>::value;
+
+            template <size_t I>
+            struct tuple_element : ::std::tr1::tuple_element<I, Tuple> {};
+
+            template <size_t I>
+            static typename AddReference<
+            const typename ::std::tr1::tuple_element<I, Tuple>::type>::type get(
+                const Tuple &tuple)
+            {
+                return ::std::tr1::get<I>(tuple);
+            }
+        };
+        template <typename TupleT>
+        const size_t TuplePolicy<TupleT>::tuple_size;
+#endif  // GTEST_HAS_TR1_TUPLE
+
+#if GTEST_HAS_STD_TUPLE_
+        template <typename... Types>
+        struct TuplePolicy< ::std::tuple<Types...> > {
+            typedef ::std::tuple<Types...> Tuple;
+            static const size_t tuple_size = ::std::tuple_size<Tuple>::value;
+
+            template <size_t I>
+            struct tuple_element : ::std::tuple_element<I, Tuple> {};
+
+            template <size_t I>
+            static const typename ::std::tuple_element<I, Tuple>::type &get(
+                const Tuple &tuple)
+            {
+                return ::std::get<I>(tuple);
+            }
+        };
+        template <typename... Types>
+        const size_t TuplePolicy< ::std::tuple<Types...> >::tuple_size;
+#endif  // GTEST_HAS_STD_TUPLE_
+
+#if GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+// This helper template allows PrintTo() for tuples and
+// UniversalTersePrintTupleFieldsToStrings() to be defined by
+// induction on the number of tuple fields.  The idea is that
+// TuplePrefixPrinter<N>::PrintPrefixTo(t, os) prints the first N
+// fields in tuple t, and can be defined in terms of
+// TuplePrefixPrinter<N - 1>.
+//
+// The inductive case.
+        template <size_t N>
+        struct TuplePrefixPrinter {
+            // Prints the first N fields of a tuple.
+            template <typename Tuple>
+            static void PrintPrefixTo(const Tuple &t, ::std::ostream *os)
+            {
+                TuplePrefixPrinter<N - 1>::PrintPrefixTo(t, os);
+                GTEST_INTENTIONAL_CONST_COND_PUSH_()
+                if (N > 1) {
+                    GTEST_INTENTIONAL_CONST_COND_POP_()
+                    *os << ", ";
+                }
+                UniversalPrinter<
+                typename TuplePolicy<Tuple>::template tuple_element<N - 1>::type>
+                ::Print(TuplePolicy<Tuple>::template get<N - 1>(t), os);
+            }
+
+            // Tersely prints the first N fields of a tuple to a string vector,
+            // one element for each field.
+            template <typename Tuple>
+            static void TersePrintPrefixToStrings(const Tuple &t, Strings *strings)
+            {
+                TuplePrefixPrinter<N - 1>::TersePrintPrefixToStrings(t, strings);
+                ::std::stringstream ss;
+                UniversalTersePrint(TuplePolicy<Tuple>::template get<N - 1>(t), &ss);
+                strings->push_back(ss.str());
+            }
+        };
+
+// Base case.
+        template <>
+        struct TuplePrefixPrinter<0> {
+            template <typename Tuple>
+            static void PrintPrefixTo(const Tuple &, ::std::ostream *) {}
+
+            template <typename Tuple>
+            static void TersePrintPrefixToStrings(const Tuple &, Strings *) {}
+        };
+
+// Helper function for printing a tuple.
+// Tuple must be either std::tr1::tuple or std::tuple type.
+        template <typename Tuple>
+        void PrintTupleTo(const Tuple &t, ::std::ostream *os)
+        {
+            *os << "(";
+            TuplePrefixPrinter<TuplePolicy<Tuple>::tuple_size>::PrintPrefixTo(t, os);
+            *os << ")";
+        }
+
+// Prints the fields of a tuple tersely to a string vector, one
+// element for each field.  See the comment before
+// UniversalTersePrint() for how we define "tersely".
+        template <typename Tuple>
+        Strings UniversalTersePrintTupleFieldsToStrings(const Tuple &value)
+        {
+            Strings result;
+            TuplePrefixPrinter<TuplePolicy<Tuple>::tuple_size>::
+            TersePrintPrefixToStrings(value, &result);
+            return result;
+        }
+#endif  // GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+
+    }  // namespace internal
+
+    template <typename T>
+    ::std::string PrintToString(const T &value)
+    {
+        ::std::stringstream ss;
+        internal::UniversalTersePrinter<T>::Print(value, &ss);
+        return ss.str();
+    }
+
+}  // namespace testing
+
+// Include any custom printer added by the local installation.
+// We must include this header at the end to make sure it can use the
+// declarations from this file.
+// Copyright 2015, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// This file provides an injection point for custom printers in a local
+// installation of gTest.
+// It will be included from gtest-printers.h and the overrides in this file
+// will be visible to everyone.
+// See documentation at gtest/gtest-printers.h for details on how to define a
+// custom printer.
+//
+// ** Custom implementation starts here **
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing
+{
+
+// Input to a parameterized test name generator, describing a test parameter.
+// Consists of the parameter value and the integer parameter index.
+    template <class ParamType>
+    struct TestParamInfo {
+        TestParamInfo(const ParamType &a_param, size_t an_index) :
+            param(a_param),
+            index(an_index) {}
+        ParamType param;
+        size_t index;
+    };
+
+// A builtin parameterized test name generator which returns the result of
+// testing::PrintToString.
+    struct PrintToStringParamName {
+        template <class ParamType>
+        std::string operator()(const TestParamInfo<ParamType> &info) const
+        {
+            return PrintToString(info.param);
+        }
+    };
+
+    namespace internal
+    {
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Outputs a message explaining invalid registration of different
+// fixture class for the same test case. This may happen when
+// TEST_P macro is used to define two tests with the same name
+// but in different namespaces.
+        GTEST_API_ void ReportInvalidTestCaseType(const char *test_case_name,
+                                                  CodeLocation code_location);
+
+        template <typename> class ParamGeneratorInterface;
+        template <typename> class ParamGenerator;
+
+// Interface for iterating over elements provided by an implementation
+// of ParamGeneratorInterface<T>.
+        template <typename T>
+        class ParamIteratorInterface
+        {
+        public:
+            virtual ~ParamIteratorInterface() {}
+            // A pointer to the base generator instance.
+            // Used only for the purposes of iterator comparison
+            // to make sure that two iterators belong to the same generator.
+            virtual const ParamGeneratorInterface<T> *BaseGenerator() const = 0;
+            // Advances iterator to point to the next element
+            // provided by the generator. The caller is responsible
+            // for not calling Advance() on an iterator equal to
+            // BaseGenerator()->End().
+            virtual void Advance() = 0;
+            // Clones the iterator object. Used for implementing copy semantics
+            // of ParamIterator<T>.
+            virtual ParamIteratorInterface *Clone() const = 0;
+            // Dereferences the current iterator and provides (read-only) access
+            // to the pointed value. It is the caller's responsibility not to call
+            // Current() on an iterator equal to BaseGenerator()->End().
+            // Used for implementing ParamGenerator<T>::operator*().
+            virtual const T *Current() const = 0;
+            // Determines whether the given iterator and other point to the same
+            // element in the sequence generated by the generator.
+            // Used for implementing ParamGenerator<T>::operator==().
+            virtual bool Equals(const ParamIteratorInterface &other) const = 0;
+        };
+
+// Class iterating over elements provided by an implementation of
+// ParamGeneratorInterface<T>. It wraps ParamIteratorInterface<T>
+// and implements the const forward iterator concept.
+        template <typename T>
+        class ParamIterator
+        {
+        public:
+            typedef T value_type;
+            typedef const T &reference;
+            typedef ptrdiff_t difference_type;
+
+            // ParamIterator assumes ownership of the impl_ pointer.
+            ParamIterator(const ParamIterator &other) : impl_(other.impl_->Clone()) {}
+            ParamIterator &operator=(const ParamIterator &other)
+            {
+                if (this != &other)
+                    impl_.reset(other.impl_->Clone());
+                return *this;
+            }
+
+            const T &operator*() const
+            {
+                return *impl_->Current();
+            }
+            const T *operator->() const
+            {
+                return impl_->Current();
+            }
+            // Prefix version of operator++.
+            ParamIterator &operator++()
+            {
+                impl_->Advance();
+                return *this;
+            }
+            // Postfix version of operator++.
+            ParamIterator operator++(int /*unused*/)
+            {
+                ParamIteratorInterface<T> *clone = impl_->Clone();
+                impl_->Advance();
+                return ParamIterator(clone);
+            }
+            bool operator==(const ParamIterator &other) const
+            {
+                return impl_.get() == other.impl_.get() || impl_->Equals(*other.impl_);
+            }
+            bool operator!=(const ParamIterator &other) const
+            {
+                return !(*this == other);
+            }
+
+        private:
+            friend class ParamGenerator<T>;
+            explicit ParamIterator(ParamIteratorInterface<T> *impl) : impl_(impl) {}
+            scoped_ptr<ParamIteratorInterface<T> > impl_;
+        };
+
+// ParamGeneratorInterface<T> is the binary interface to access generators
+// defined in other translation units.
+        template <typename T>
+        class ParamGeneratorInterface
+        {
+        public:
+            typedef T ParamType;
+
+            virtual ~ParamGeneratorInterface() {}
+
+            // Generator interface definition
+            virtual ParamIteratorInterface<T> *Begin() const = 0;
+            virtual ParamIteratorInterface<T> *End() const = 0;
+        };
+
+// Wraps ParamGeneratorInterface<T> and provides general generator syntax
+// compatible with the STL Container concept.
+// This class implements copy initialization semantics and the contained
+// ParamGeneratorInterface<T> instance is shared among all copies
+// of the original object. This is possible because that instance is immutable.
+        template<typename T>
+        class ParamGenerator
+        {
+        public:
+            typedef ParamIterator<T> iterator;
+
+            explicit ParamGenerator(ParamGeneratorInterface<T> *impl) : impl_(impl) {}
+            ParamGenerator(const ParamGenerator &other) : impl_(other.impl_) {}
+
+            ParamGenerator &operator=(const ParamGenerator &other)
+            {
+                impl_ = other.impl_;
+                return *this;
+            }
+
+            iterator begin() const
+            {
+                return iterator(impl_->Begin());
+            }
+            iterator end() const
+            {
+                return iterator(impl_->End());
+            }
+
+        private:
+            linked_ptr<const ParamGeneratorInterface<T> > impl_;
+        };
+
+// Generates values from a range of two comparable values. Can be used to
+// generate sequences of user-defined types that implement operator+() and
+// operator<().
+// This class is used in the Range() function.
+        template <typename T, typename IncrementT>
+        class RangeGenerator : public ParamGeneratorInterface<T>
+        {
+        public:
+            RangeGenerator(T begin, T end, IncrementT step)
+                : begin_(begin), end_(end),
+                  step_(step), end_index_(CalculateEndIndex(begin, end, step)) {}
+            virtual ~RangeGenerator() {}
+
+            virtual ParamIteratorInterface<T> *Begin() const
+            {
+                return new Iterator(this, begin_, 0, step_);
+            }
+            virtual ParamIteratorInterface<T> *End() const
+            {
+                return new Iterator(this, end_, end_index_, step_);
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<T>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<T> *base, T value, int index,
+                         IncrementT step)
+                    : base_(base), value_(value), index_(index), step_(step) {}
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<T> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                virtual void Advance()
+                {
+                    value_ = static_cast<T>(value_ + step_);
+                    index_++;
+                }
+                virtual ParamIteratorInterface<T> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const T *Current() const
+                {
+                    return &value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<T> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const int other_index =
+                        CheckedDowncastToActualType<const Iterator>(&other)->index_;
+                    return index_ == other_index;
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : ParamIteratorInterface<T>(),
+                      base_(other.base_), value_(other.value_), index_(other.index_),
+                      step_(other.step_) {}
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<T> *const base_;
+                T value_;
+                int index_;
+                const IncrementT step_;
+            };  // class RangeGenerator::Iterator
+
+            static int CalculateEndIndex(const T &begin,
+                                         const T &end,
+                                         const IncrementT &step)
+            {
+                int end_index = 0;
+                for (T i = begin; i < end; i = static_cast<T>(i + step))
+                    end_index++;
+                return end_index;
+            }
+
+            // No implementation - assignment is unsupported.
+            void operator=(const RangeGenerator &other);
+
+            const T begin_;
+            const T end_;
+            const IncrementT step_;
+            // The index for the end() iterator. All the elements in the generated
+            // sequence are indexed (0-based) to aid iterator comparison.
+            const int end_index_;
+        };  // class RangeGenerator
+
+
+// Generates values from a pair of STL-style iterators. Used in the
+// ValuesIn() function. The elements are copied from the source range
+// since the source can be located on the stack, and the generator
+// is likely to persist beyond that stack frame.
+        template <typename T>
+        class ValuesInIteratorRangeGenerator : public ParamGeneratorInterface<T>
+        {
+        public:
+            template <typename ForwardIterator>
+            ValuesInIteratorRangeGenerator(ForwardIterator begin, ForwardIterator end)
+                : container_(begin, end) {}
+            virtual ~ValuesInIteratorRangeGenerator() {}
+
+            virtual ParamIteratorInterface<T> *Begin() const
+            {
+                return new Iterator(this, container_.begin());
+            }
+            virtual ParamIteratorInterface<T> *End() const
+            {
+                return new Iterator(this, container_.end());
+            }
+
+        private:
+            typedef typename ::std::vector<T> ContainerType;
+
+            class Iterator : public ParamIteratorInterface<T>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<T> *base,
+                         typename ContainerType::const_iterator iterator)
+                    : base_(base), iterator_(iterator) {}
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<T> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                virtual void Advance()
+                {
+                    ++iterator_;
+                    value_.reset();
+                }
+                virtual ParamIteratorInterface<T> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                // We need to use cached value referenced by iterator_ because *iterator_
+                // can return a temporary object (and of type other then T), so just
+                // having "return &*iterator_;" doesn't work.
+                // value_ is updated here and not in Advance() because Advance()
+                // can advance iterator_ beyond the end of the range, and we cannot
+                // detect that fact. The client code, on the other hand, is
+                // responsible for not calling Current() on an out-of-range iterator.
+                virtual const T *Current() const
+                {
+                    if (value_.get() == NULL)
+                        value_.reset(new T(*iterator_));
+                    return value_.get();
+                }
+                virtual bool Equals(const ParamIteratorInterface<T> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    return iterator_ ==
+                           CheckedDowncastToActualType<const Iterator>(&other)->iterator_;
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                // The explicit constructor call suppresses a false warning
+                // emitted by gcc when supplied with the -Wextra option.
+                    : ParamIteratorInterface<T>(),
+                      base_(other.base_),
+                      iterator_(other.iterator_) {}
+
+                const ParamGeneratorInterface<T> *const base_;
+                typename ContainerType::const_iterator iterator_;
+                // A cached value of *iterator_. We keep it here to allow access by
+                // pointer in the wrapping iterator's operator->().
+                // value_ needs to be mutable to be accessed in Current().
+                // Use of scoped_ptr helps manage cached value's lifetime,
+                // which is bound by the lifespan of the iterator itself.
+                mutable scoped_ptr<const T> value_;
+            };  // class ValuesInIteratorRangeGenerator::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const ValuesInIteratorRangeGenerator &other);
+
+            const ContainerType container_;
+        };  // class ValuesInIteratorRangeGenerator
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Default parameterized test name generator, returns a string containing the
+// integer test parameter index.
+        template <class ParamType>
+        std::string DefaultParamName(const TestParamInfo<ParamType> &info)
+        {
+            Message name_stream;
+            name_stream << info.index;
+            return name_stream.GetString();
+        }
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Parameterized test name overload helpers, which help the
+// INSTANTIATE_TEST_CASE_P macro choose between the default parameterized
+// test name generator and user param name generator.
+        template <class ParamType, class ParamNameGenFunctor>
+        ParamNameGenFunctor GetParamNameGen(ParamNameGenFunctor func)
+        {
+            return func;
+        }
+
+        template <class ParamType>
+        struct ParamNameGenFunc {
+            typedef std::string Type(const TestParamInfo<ParamType> &);
+        };
+
+        template <class ParamType>
+        typename ParamNameGenFunc<ParamType>::Type *GetParamNameGen()
+        {
+            return DefaultParamName;
+        }
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Stores a parameter value and later creates tests parameterized with that
+// value.
+        template <class TestClass>
+        class ParameterizedTestFactory : public TestFactoryBase
+        {
+        public:
+            typedef typename TestClass::ParamType ParamType;
+            explicit ParameterizedTestFactory(ParamType parameter) :
+                parameter_(parameter) {}
+            virtual Test *CreateTest()
+            {
+                TestClass::SetParam(&parameter_);
+                return new TestClass();
+            }
+
+        private:
+            const ParamType parameter_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestFactory);
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// TestMetaFactoryBase is a base class for meta-factories that create
+// test factories for passing into MakeAndRegisterTestInfo function.
+        template <class ParamType>
+        class TestMetaFactoryBase
+        {
+        public:
+            virtual ~TestMetaFactoryBase() {}
+
+            virtual TestFactoryBase *CreateTestFactory(ParamType parameter) = 0;
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// TestMetaFactory creates test factories for passing into
+// MakeAndRegisterTestInfo function. Since MakeAndRegisterTestInfo receives
+// ownership of test factory pointer, same factory object cannot be passed
+// into that method twice. But ParameterizedTestCaseInfo is going to call
+// it for each Test/Parameter value combination. Thus it needs meta factory
+// creator class.
+        template <class TestCase>
+        class TestMetaFactory
+            : public TestMetaFactoryBase<typename TestCase::ParamType>
+        {
+        public:
+            typedef typename TestCase::ParamType ParamType;
+
+            TestMetaFactory() {}
+
+            virtual TestFactoryBase *CreateTestFactory(ParamType parameter)
+            {
+                return new ParameterizedTestFactory<TestCase>(parameter);
+            }
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(TestMetaFactory);
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseInfoBase is a generic interface
+// to ParameterizedTestCaseInfo classes. ParameterizedTestCaseInfoBase
+// accumulates test information provided by TEST_P macro invocations
+// and generators provided by INSTANTIATE_TEST_CASE_P macro invocations
+// and uses that information to register all resulting test instances
+// in RegisterTests method. The ParameterizeTestCaseRegistry class holds
+// a collection of pointers to the ParameterizedTestCaseInfo objects
+// and calls RegisterTests() on each of them when asked.
+        class ParameterizedTestCaseInfoBase
+        {
+        public:
+            virtual ~ParameterizedTestCaseInfoBase() {}
+
+            // Base part of test case name for display purposes.
+            virtual const std::string &GetTestCaseName() const = 0;
+            // Test case id to verify identity.
+            virtual TypeId GetTestCaseTypeId() const = 0;
+            // UnitTest class invokes this method to register tests in this
+            // test case right before running them in RUN_ALL_TESTS macro.
+            // This method should not be called more then once on any single
+            // instance of a ParameterizedTestCaseInfoBase derived class.
+            virtual void RegisterTests() = 0;
+
+        protected:
+            ParameterizedTestCaseInfoBase() {}
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfoBase);
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseInfo accumulates tests obtained from TEST_P
+// macro invocations for a particular test case and generators
+// obtained from INSTANTIATE_TEST_CASE_P macro invocations for that
+// test case. It registers tests with all values generated by all
+// generators when asked.
+        template <class TestCase>
+        class ParameterizedTestCaseInfo : public ParameterizedTestCaseInfoBase
+        {
+        public:
+            // ParamType and GeneratorCreationFunc are private types but are required
+            // for declarations of public methods AddTestPattern() and
+            // AddTestCaseInstantiation().
+            typedef typename TestCase::ParamType ParamType;
+            // A function that returns an instance of appropriate generator type.
+            typedef ParamGenerator<ParamType>(GeneratorCreationFunc)();
+            typedef typename ParamNameGenFunc<ParamType>::Type ParamNameGeneratorFunc;
+
+            explicit ParameterizedTestCaseInfo(
+                const char *name, CodeLocation code_location)
+                : test_case_name_(name), code_location_(code_location) {}
+
+            // Test case base name for display purposes.
+            virtual const std::string &GetTestCaseName() const
+            {
+                return test_case_name_;
+            }
+            // Test case id to verify identity.
+            virtual TypeId GetTestCaseTypeId() const
+            {
+                return GetTypeId<TestCase>();
+            }
+            // TEST_P macro uses AddTestPattern() to record information
+            // about a single test in a LocalTestInfo structure.
+            // test_case_name is the base name of the test case (without invocation
+            // prefix). test_base_name is the name of an individual test without
+            // parameter index. For the test SequenceA/FooTest.DoBar/1 FooTest is
+            // test case base name and DoBar is test base name.
+            void AddTestPattern(const char *test_case_name,
+                                const char *test_base_name,
+                                TestMetaFactoryBase<ParamType> *meta_factory)
+            {
+                tests_.push_back(linked_ptr<TestInfo>(new TestInfo(test_case_name,
+                                                                   test_base_name,
+                                                                   meta_factory)));
+            }
+            // INSTANTIATE_TEST_CASE_P macro uses AddGenerator() to record information
+            // about a generator.
+            int AddTestCaseInstantiation(const std::string &instantiation_name,
+                                         GeneratorCreationFunc *func,
+                                         ParamNameGeneratorFunc *name_func,
+                                         const char *file, int line)
+            {
+                instantiations_.push_back(
+                    InstantiationInfo(instantiation_name, func, name_func, file, line));
+                return 0;  // Return value used only to run this method in namespace scope.
+            }
+            // UnitTest class invokes this method to register tests in this test case
+            // test cases right before running tests in RUN_ALL_TESTS macro.
+            // This method should not be called more then once on any single
+            // instance of a ParameterizedTestCaseInfoBase derived class.
+            // UnitTest has a guard to prevent from calling this method more then once.
+            virtual void RegisterTests()
+            {
+                for (typename TestInfoContainer::iterator test_it = tests_.begin();
+                     test_it != tests_.end(); ++test_it) {
+                    linked_ptr<TestInfo> test_info = *test_it;
+                    for (typename InstantiationContainer::iterator gen_it =
+                             instantiations_.begin(); gen_it != instantiations_.end();
+                         ++gen_it) {
+                        const std::string &instantiation_name = gen_it->name;
+                        ParamGenerator<ParamType> generator((*gen_it->generator)());
+                        ParamNameGeneratorFunc *name_func = gen_it->name_func;
+                        const char *file = gen_it->file;
+                        int line = gen_it->line;
+
+                        std::string test_case_name;
+                        if ( !instantiation_name.empty() )
+                            test_case_name = instantiation_name + "/";
+                        test_case_name += test_info->test_case_base_name;
+
+                        size_t i = 0;
+                        std::set<std::string> test_param_names;
+                        for (typename ParamGenerator<ParamType>::iterator param_it =
+                                 generator.begin();
+                             param_it != generator.end(); ++param_it, ++i) {
+                            Message test_name_stream;
+
+                            std::string param_name = name_func(
+                                                         TestParamInfo<ParamType>(*param_it, i));
+
+                            GTEST_CHECK_(IsValidParamName(param_name))
+                                    << "Parameterized test name '" << param_name
+                                    << "' is invalid, in " << file
+                                    << " line " << line << std::endl;
+
+                            GTEST_CHECK_(test_param_names.count(param_name) == 0)
+                                    << "Duplicate parameterized test name '" << param_name
+                                    << "', in " << file << " line " << line << std::endl;
+
+                            test_param_names.insert(param_name);
+
+                            test_name_stream << test_info->test_base_name << "/" << param_name;
+                            MakeAndRegisterTestInfo(
+                                test_case_name.c_str(),
+                                test_name_stream.GetString().c_str(),
+                                NULL,  // No type parameter.
+                                PrintToString(*param_it).c_str(),
+                                code_location_,
+                                GetTestCaseTypeId(),
+                                TestCase::SetUpTestCase,
+                                TestCase::TearDownTestCase,
+                                test_info->test_meta_factory->CreateTestFactory(*param_it));
+                        }  // for param_it
+                    }  // for gen_it
+                }  // for test_it
+            }  // RegisterTests
+
+        private:
+            // LocalTestInfo structure keeps information about a single test registered
+            // with TEST_P macro.
+            struct TestInfo {
+                TestInfo(const char *a_test_case_base_name,
+                         const char *a_test_base_name,
+                         TestMetaFactoryBase<ParamType> *a_test_meta_factory) :
+                    test_case_base_name(a_test_case_base_name),
+                    test_base_name(a_test_base_name),
+                    test_meta_factory(a_test_meta_factory) {}
+
+                const std::string test_case_base_name;
+                const std::string test_base_name;
+                const scoped_ptr<TestMetaFactoryBase<ParamType> > test_meta_factory;
+            };
+            typedef ::std::vector<linked_ptr<TestInfo> > TestInfoContainer;
+            // Records data received from INSTANTIATE_TEST_CASE_P macros:
+            //  <Instantiation name, Sequence generator creation function,
+            //     Name generator function, Source file, Source line>
+            struct InstantiationInfo {
+                InstantiationInfo(const std::string &name_in,
+                                  GeneratorCreationFunc *generator_in,
+                                  ParamNameGeneratorFunc *name_func_in,
+                                  const char *file_in,
+                                  int line_in)
+                    : name(name_in),
+                      generator(generator_in),
+                      name_func(name_func_in),
+                      file(file_in),
+                      line(line_in) {}
+
+                std::string name;
+                GeneratorCreationFunc *generator;
+                ParamNameGeneratorFunc *name_func;
+                const char *file;
+                int line;
+            };
+            typedef ::std::vector<InstantiationInfo> InstantiationContainer;
+
+            static bool IsValidParamName(const std::string &name)
+            {
+                // Check for empty string
+                if (name.empty())
+                    return false;
+
+                // Check for invalid characters
+                for (std::string::size_type index = 0; index < name.size(); ++index) {
+                    if (!isalnum(name[index]) && name[index] != '_')
+                        return false;
+                }
+
+                return true;
+            }
+
+            const std::string test_case_name_;
+            CodeLocation code_location_;
+            TestInfoContainer tests_;
+            InstantiationContainer instantiations_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfo);
+        };  // class ParameterizedTestCaseInfo
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseRegistry contains a map of ParameterizedTestCaseInfoBase
+// classes accessed by test case names. TEST_P and INSTANTIATE_TEST_CASE_P
+// macros use it to locate their corresponding ParameterizedTestCaseInfo
+// descriptors.
+        class ParameterizedTestCaseRegistry
+        {
+        public:
+            ParameterizedTestCaseRegistry() {}
+            ~ParameterizedTestCaseRegistry()
+            {
+                for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+                     it != test_case_infos_.end(); ++it) {
+                    delete *it;
+                }
+            }
+
+            // Looks up or creates and returns a structure containing information about
+            // tests and instantiations of a particular test case.
+            template <class TestCase>
+            ParameterizedTestCaseInfo<TestCase> *GetTestCasePatternHolder(
+                const char *test_case_name,
+                CodeLocation code_location)
+            {
+                ParameterizedTestCaseInfo<TestCase> *typed_test_info = NULL;
+                for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+                     it != test_case_infos_.end(); ++it) {
+                    if ((*it)->GetTestCaseName() == test_case_name) {
+                        if ((*it)->GetTestCaseTypeId() != GetTypeId<TestCase>()) {
+                            // Complain about incorrect usage of Google Test facilities
+                            // and terminate the program since we cannot guaranty correct
+                            // test case setup and tear-down in this case.
+                            ReportInvalidTestCaseType(test_case_name, code_location);
+                            posix::Abort();
+                        } else {
+                            // At this point we are sure that the object we found is of the same
+                            // type we are looking for, so we downcast it to that type
+                            // without further checks.
+                            typed_test_info = CheckedDowncastToActualType<
+                                              ParameterizedTestCaseInfo<TestCase> >(*it);
+                        }
+                        break;
+                    }
+                }
+                if (typed_test_info == NULL) {
+                    typed_test_info = new ParameterizedTestCaseInfo<TestCase>(
+                        test_case_name, code_location);
+                    test_case_infos_.push_back(typed_test_info);
+                }
+                return typed_test_info;
+            }
+            void RegisterTests()
+            {
+                for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+                     it != test_case_infos_.end(); ++it) {
+                    (*it)->RegisterTests();
+                }
+            }
+
+        private:
+            typedef ::std::vector<ParameterizedTestCaseInfoBase *> TestCaseInfoContainer;
+
+            TestCaseInfoContainer test_case_infos_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseRegistry);
+        };
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  //  GTEST_HAS_PARAM_TEST
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+// This file was GENERATED by command:
+//     pump.py gtest-param-util-generated.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// Type and function utilities for implementing parameterized tests.
+// This file is generated by a SCRIPT.  DO NOT EDIT BY HAND!
+//
+// Currently Google Test supports at most 50 arguments in Values,
+// and at most 10 arguments in Combine. Please contact
+// googletestframework@googlegroups.com if you need more.
+// Please note that the number of arguments to Combine is limited
+// by the maximum arity of the implementation of tuple which is
+// currently set at 10.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*.  Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing
+{
+
+// Forward declarations of ValuesIn(), which is implemented in
+// include/gtest/gtest-param-test.h.
+    template <typename ForwardIterator>
+    internal::ParamGenerator<
+    typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+    ValuesIn(ForwardIterator begin, ForwardIterator end);
+
+    template <typename T, size_t N>
+    internal::ParamGenerator<T> ValuesIn(const T (&array)[N]);
+
+    template <class Container>
+    internal::ParamGenerator<typename Container::value_type> ValuesIn(
+        const Container &container);
+
+    namespace internal
+    {
+
+// Used in the Values() function to provide polymorphic capabilities.
+        template <typename T1>
+        class ValueArray1
+        {
+        public:
+            explicit ValueArray1(T1 v1) : v1_(v1) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_)};
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray1 &other);
+
+            const T1 v1_;
+        };
+
+        template <typename T1, typename T2>
+        class ValueArray2
+        {
+        public:
+            ValueArray2(T1 v1, T2 v2) : v1_(v1), v2_(v2) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_)};
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray2 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+        };
+
+        template <typename T1, typename T2, typename T3>
+        class ValueArray3
+        {
+        public:
+            ValueArray3(T1 v1, T2 v2, T3 v3) : v1_(v1), v2_(v2), v3_(v3) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray3 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        class ValueArray4
+        {
+        public:
+            ValueArray4(T1 v1, T2 v2, T3 v3, T4 v4) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray4 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        class ValueArray5
+        {
+        public:
+            ValueArray5(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray5 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        class ValueArray6
+        {
+        public:
+            ValueArray6(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray6 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        class ValueArray7
+        {
+        public:
+            ValueArray7(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray7 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        class ValueArray8
+        {
+        public:
+            ValueArray8(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                        T8 v8) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray8 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        class ValueArray9
+        {
+        public:
+            ValueArray9(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                        T9 v9) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray9 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        class ValueArray10
+        {
+        public:
+            ValueArray10(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray10 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11>
+        class ValueArray11
+        {
+        public:
+            ValueArray11(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray11 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12>
+        class ValueArray12
+        {
+        public:
+            ValueArray12(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray12 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13>
+        class ValueArray13
+        {
+        public:
+            ValueArray13(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray13 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14>
+        class ValueArray14
+        {
+        public:
+            ValueArray14(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray14 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15>
+        class ValueArray15
+        {
+        public:
+            ValueArray15(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray15 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16>
+        class ValueArray16
+        {
+        public:
+            ValueArray16(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray16 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17>
+        class ValueArray17
+        {
+        public:
+            ValueArray17(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+                         T17 v17) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray17 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18>
+        class ValueArray18
+        {
+        public:
+            ValueArray18(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray18 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19>
+        class ValueArray19
+        {
+        public:
+            ValueArray19(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray19 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20>
+        class ValueArray20
+        {
+        public:
+            ValueArray20(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray20 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21>
+        class ValueArray21
+        {
+        public:
+            ValueArray21(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray21 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22>
+        class ValueArray22
+        {
+        public:
+            ValueArray22(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray22 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23>
+        class ValueArray23
+        {
+        public:
+            ValueArray23(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray23 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24>
+        class ValueArray24
+        {
+        public:
+            ValueArray24(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray24 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25>
+        class ValueArray25
+        {
+        public:
+            ValueArray25(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+                         T25 v25) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray25 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26>
+        class ValueArray26
+        {
+        public:
+            ValueArray26(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray26 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27>
+        class ValueArray27
+        {
+        public:
+            ValueArray27(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+                v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+                v26_(v26), v27_(v27) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray27 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28>
+        class ValueArray28
+        {
+        public:
+            ValueArray28(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+                v25_(v25), v26_(v26), v27_(v27), v28_(v28) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray28 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29>
+        class ValueArray29
+        {
+        public:
+            ValueArray29(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+                v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray29 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30>
+        class ValueArray30
+        {
+        public:
+            ValueArray30(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray30 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31>
+        class ValueArray31
+        {
+        public:
+            ValueArray31(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray31 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32>
+        class ValueArray32
+        {
+        public:
+            ValueArray32(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+                v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray32 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33>
+        class ValueArray33
+        {
+        public:
+            ValueArray33(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32,
+                         T33 v33) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray33 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34>
+        class ValueArray34
+        {
+        public:
+            ValueArray34(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray34 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35>
+        class ValueArray35
+        {
+        public:
+            ValueArray35(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+                v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+                v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31),
+                v32_(v32), v33_(v33), v34_(v34), v35_(v35) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray35 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36>
+        class ValueArray36
+        {
+        public:
+            ValueArray36(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+                v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30),
+                v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray36 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37>
+        class ValueArray37
+        {
+        public:
+            ValueArray37(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+                v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29),
+                v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35),
+                v36_(v36), v37_(v37) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray37 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38>
+        class ValueArray38
+        {
+        public:
+            ValueArray38(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray38 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39>
+        class ValueArray39
+        {
+        public:
+            ValueArray39(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray39 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40>
+        class ValueArray40
+        {
+        public:
+            ValueArray40(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+                v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33),
+                v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39),
+                v40_(v40) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray40 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41>
+        class ValueArray41
+        {
+        public:
+            ValueArray41(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40,
+                         T41 v41) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray41 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42>
+        class ValueArray42
+        {
+        public:
+            ValueArray42(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41), v42_(v42) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray42 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43>
+        class ValueArray43
+        {
+        public:
+            ValueArray43(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+                v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+                v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31),
+                v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37),
+                v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray43 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44>
+        class ValueArray44
+        {
+        public:
+            ValueArray44(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+                v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30),
+                v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36),
+                v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42),
+                v43_(v43), v44_(v44) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray44 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45>
+        class ValueArray45
+        {
+        public:
+            ValueArray45(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+                v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29),
+                v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35),
+                v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41),
+                v42_(v42), v43_(v43), v44_(v44), v45_(v45) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray45 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46>
+        class ValueArray46
+        {
+        public:
+            ValueArray46(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40),
+                v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray46 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47>
+        class ValueArray47
+        {
+        public:
+            ValueArray47(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40),
+                v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46),
+                v47_(v47) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray47 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48>
+        class ValueArray48
+        {
+        public:
+            ValueArray48(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+                v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33),
+                v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39),
+                v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45),
+                v46_(v46), v47_(v47), v48_(v48) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_),
+                                   static_cast<T>(v48_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray48 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+            const T48 v48_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49>
+        class ValueArray49
+        {
+        public:
+            ValueArray49(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48,
+                         T49 v49) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44),
+                v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_),
+                                   static_cast<T>(v48_), static_cast<T>(v49_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray49 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+            const T48 v48_;
+            const T49 v49_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49, typename T50>
+        class ValueArray50
+        {
+        public:
+            ValueArray50(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48, T49 v49,
+                         T50 v50) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44),
+                v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49), v50_(v50) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_),
+                                   static_cast<T>(v48_), static_cast<T>(v49_), static_cast<T>(v50_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray50 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+            const T48 v48_;
+            const T49 v49_;
+            const T50 v50_;
+        };
+
+# if GTEST_HAS_COMBINE
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Generates values from the Cartesian product of values produced
+// by the argument generators.
+//
+        template <typename T1, typename T2>
+        class CartesianProductGenerator2
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2> ParamType;
+
+            CartesianProductGenerator2(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2)
+                : g1_(g1), g2_(g2) {}
+            virtual ~CartesianProductGenerator2() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current2_;
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator2::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator2 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+        };  // class CartesianProductGenerator2
+
+
+        template <typename T1, typename T2, typename T3>
+        class CartesianProductGenerator3
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3> ParamType;
+
+            CartesianProductGenerator3(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3)
+                : g1_(g1), g2_(g2), g3_(g3) {}
+            virtual ~CartesianProductGenerator3() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current3_;
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator3::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator3 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+        };  // class CartesianProductGenerator3
+
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        class CartesianProductGenerator4
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4> ParamType;
+
+            CartesianProductGenerator4(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {}
+            virtual ~CartesianProductGenerator4() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current4_;
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator4::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator4 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+        };  // class CartesianProductGenerator4
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        class CartesianProductGenerator5
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5> ParamType;
+
+            CartesianProductGenerator5(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {}
+            virtual ~CartesianProductGenerator5() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current5_;
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator5::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator5 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+        };  // class CartesianProductGenerator5
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        class CartesianProductGenerator6
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5,
+              T6> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6> ParamType;
+
+            CartesianProductGenerator6(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {}
+            virtual ~CartesianProductGenerator6() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current6_;
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator6::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator6 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+        };  // class CartesianProductGenerator6
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        class CartesianProductGenerator7
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7> ParamType;
+
+            CartesianProductGenerator7(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {}
+            virtual ~CartesianProductGenerator7() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current7_;
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator7::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator7 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+        };  // class CartesianProductGenerator7
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        class CartesianProductGenerator8
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7, T8> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8> ParamType;
+
+            CartesianProductGenerator8(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7,
+                                       const ParamGenerator<T8> &g8)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7),
+                  g8_(g8) {}
+            virtual ~CartesianProductGenerator8() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin(), g8_, g8_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+                                    g8_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7,
+                         const ParamGenerator<T8> &g8,
+                         const typename ParamGenerator<T8>::iterator &current8)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+                      begin8_(g8.begin()), end8_(g8.end()), current8_(current8)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current8_;
+                    if (current8_ == end8_) {
+                        current8_ = begin8_;
+                        ++current7_;
+                    }
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_ &&
+                               current8_ == typed_other->current8_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_),
+                      begin8_(other.begin8_),
+                      end8_(other.end8_),
+                      current8_(other.current8_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_, *current8_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_ ||
+                        current8_ == end8_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                const typename ParamGenerator<T8>::iterator begin8_;
+                const typename ParamGenerator<T8>::iterator end8_;
+                typename ParamGenerator<T8>::iterator current8_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator8::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator8 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+            const ParamGenerator<T8> g8_;
+        };  // class CartesianProductGenerator8
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        class CartesianProductGenerator9
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7, T8, T9> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9> ParamType;
+
+            CartesianProductGenerator9(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7,
+                                       const ParamGenerator<T8> &g8, const ParamGenerator<T9> &g9)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9) {}
+            virtual ~CartesianProductGenerator9() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+                                    g8_.end(), g9_, g9_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7,
+                         const ParamGenerator<T8> &g8,
+                         const typename ParamGenerator<T8>::iterator &current8,
+                         const ParamGenerator<T9> &g9,
+                         const typename ParamGenerator<T9>::iterator &current9)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+                      begin8_(g8.begin()), end8_(g8.end()), current8_(current8),
+                      begin9_(g9.begin()), end9_(g9.end()), current9_(current9)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current9_;
+                    if (current9_ == end9_) {
+                        current9_ = begin9_;
+                        ++current8_;
+                    }
+                    if (current8_ == end8_) {
+                        current8_ = begin8_;
+                        ++current7_;
+                    }
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_ &&
+                               current8_ == typed_other->current8_ &&
+                               current9_ == typed_other->current9_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_),
+                      begin8_(other.begin8_),
+                      end8_(other.end8_),
+                      current8_(other.current8_),
+                      begin9_(other.begin9_),
+                      end9_(other.end9_),
+                      current9_(other.current9_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_, *current8_,
+                                                   *current9_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_ ||
+                        current8_ == end8_ ||
+                        current9_ == end9_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                const typename ParamGenerator<T8>::iterator begin8_;
+                const typename ParamGenerator<T8>::iterator end8_;
+                typename ParamGenerator<T8>::iterator current8_;
+                const typename ParamGenerator<T9>::iterator begin9_;
+                const typename ParamGenerator<T9>::iterator end9_;
+                typename ParamGenerator<T9>::iterator current9_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator9::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator9 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+            const ParamGenerator<T8> g8_;
+            const ParamGenerator<T9> g9_;
+        };  // class CartesianProductGenerator9
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        class CartesianProductGenerator10
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7, T8, T9, T10> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> ParamType;
+
+            CartesianProductGenerator10(const ParamGenerator<T1> &g1,
+                                        const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                        const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                        const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7,
+                                        const ParamGenerator<T8> &g8, const ParamGenerator<T9> &g9,
+                                        const ParamGenerator<T10> &g10)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9), g10_(g10) {}
+            virtual ~CartesianProductGenerator10() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin(), g10_, g10_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+                                    g8_.end(), g9_, g9_.end(), g10_, g10_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7,
+                         const ParamGenerator<T8> &g8,
+                         const typename ParamGenerator<T8>::iterator &current8,
+                         const ParamGenerator<T9> &g9,
+                         const typename ParamGenerator<T9>::iterator &current9,
+                         const ParamGenerator<T10> &g10,
+                         const typename ParamGenerator<T10>::iterator &current10)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+                      begin8_(g8.begin()), end8_(g8.end()), current8_(current8),
+                      begin9_(g9.begin()), end9_(g9.end()), current9_(current9),
+                      begin10_(g10.begin()), end10_(g10.end()), current10_(current10)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current10_;
+                    if (current10_ == end10_) {
+                        current10_ = begin10_;
+                        ++current9_;
+                    }
+                    if (current9_ == end9_) {
+                        current9_ = begin9_;
+                        ++current8_;
+                    }
+                    if (current8_ == end8_) {
+                        current8_ = begin8_;
+                        ++current7_;
+                    }
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_ &&
+                               current8_ == typed_other->current8_ &&
+                               current9_ == typed_other->current9_ &&
+                               current10_ == typed_other->current10_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_),
+                      begin8_(other.begin8_),
+                      end8_(other.end8_),
+                      current8_(other.current8_),
+                      begin9_(other.begin9_),
+                      end9_(other.end9_),
+                      current9_(other.current9_),
+                      begin10_(other.begin10_),
+                      end10_(other.end10_),
+                      current10_(other.current10_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_, *current8_,
+                                                   *current9_, *current10_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_ ||
+                        current8_ == end8_ ||
+                        current9_ == end9_ ||
+                        current10_ == end10_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                const typename ParamGenerator<T8>::iterator begin8_;
+                const typename ParamGenerator<T8>::iterator end8_;
+                typename ParamGenerator<T8>::iterator current8_;
+                const typename ParamGenerator<T9>::iterator begin9_;
+                const typename ParamGenerator<T9>::iterator end9_;
+                typename ParamGenerator<T9>::iterator current9_;
+                const typename ParamGenerator<T10>::iterator begin10_;
+                const typename ParamGenerator<T10>::iterator end10_;
+                typename ParamGenerator<T10>::iterator current10_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator10::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator10 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+            const ParamGenerator<T8> g8_;
+            const ParamGenerator<T9> g9_;
+            const ParamGenerator<T10> g10_;
+        };  // class CartesianProductGenerator10
+
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Helper classes providing Combine() with polymorphic features. They allow
+// casting CartesianProductGeneratorN<T> to ParamGenerator<U> if T is
+// convertible to U.
+//
+        template <class Generator1, class Generator2>
+        class CartesianProductHolder2
+        {
+        public:
+            CartesianProductHolder2(const Generator1 &g1, const Generator2 &g2)
+                : g1_(g1), g2_(g2) {}
+            template <typename T1, typename T2>
+            operator ParamGenerator< ::testing::tuple<T1, T2> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2> >(
+                           new CartesianProductGenerator2<T1, T2>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder2 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+        };  // class CartesianProductHolder2
+
+        template <class Generator1, class Generator2, class Generator3>
+        class CartesianProductHolder3
+        {
+        public:
+            CartesianProductHolder3(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3)
+                : g1_(g1), g2_(g2), g3_(g3) {}
+            template <typename T1, typename T2, typename T3>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3> >(
+                           new CartesianProductGenerator3<T1, T2, T3>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder3 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+        };  // class CartesianProductHolder3
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4>
+        class CartesianProductHolder4
+        {
+        public:
+            CartesianProductHolder4(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {}
+            template <typename T1, typename T2, typename T3, typename T4>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4> >(
+                           new CartesianProductGenerator4<T1, T2, T3, T4>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder4 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+        };  // class CartesianProductHolder4
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5>
+        class CartesianProductHolder5
+        {
+        public:
+            CartesianProductHolder5(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5> >(
+                           new CartesianProductGenerator5<T1, T2, T3, T4, T5>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder5 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+        };  // class CartesianProductHolder5
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6>
+        class CartesianProductHolder6
+        {
+        public:
+            CartesianProductHolder6(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6> >(
+                           new CartesianProductGenerator6<T1, T2, T3, T4, T5, T6>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder6 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+        };  // class CartesianProductHolder6
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7>
+        class CartesianProductHolder7
+        {
+        public:
+            CartesianProductHolder7(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6, const Generator7 &g7)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+                     T7> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7> >(
+                           new CartesianProductGenerator7<T1, T2, T3, T4, T5, T6, T7>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder7 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+        };  // class CartesianProductHolder7
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7,
+                  class Generator8>
+        class CartesianProductHolder8
+        {
+        public:
+            CartesianProductHolder8(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6, const Generator7 &g7, const Generator8 &g8)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7),
+                  g8_(g8) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7, typename T8>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7,
+                     T8> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8> >(
+                           new CartesianProductGenerator8<T1, T2, T3, T4, T5, T6, T7, T8>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_),
+                               static_cast<ParamGenerator<T8> >(g8_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder8 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+            const Generator8 g8_;
+        };  // class CartesianProductHolder8
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7,
+                  class Generator8, class Generator9>
+        class CartesianProductHolder9
+        {
+        public:
+            CartesianProductHolder9(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6, const Generator7 &g7, const Generator8 &g8,
+                                    const Generator9 &g9)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7, typename T8, typename T9>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+                     T9> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+                       T9> >(
+                           new CartesianProductGenerator9<T1, T2, T3, T4, T5, T6, T7, T8, T9>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_),
+                               static_cast<ParamGenerator<T8> >(g8_),
+                               static_cast<ParamGenerator<T9> >(g9_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder9 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+            const Generator8 g8_;
+            const Generator9 g9_;
+        };  // class CartesianProductHolder9
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7,
+                  class Generator8, class Generator9, class Generator10>
+        class CartesianProductHolder10
+        {
+        public:
+            CartesianProductHolder10(const Generator1 &g1, const Generator2 &g2,
+                                     const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                     const Generator6 &g6, const Generator7 &g7, const Generator8 &g8,
+                                     const Generator9 &g9, const Generator10 &g10)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9), g10_(g10) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7, typename T8, typename T9, typename T10>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9,
+                     T10> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9,
+                       T10> >(
+                           new CartesianProductGenerator10<T1, T2, T3, T4, T5, T6, T7, T8, T9,
+                           T10>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_),
+                               static_cast<ParamGenerator<T8> >(g8_),
+                               static_cast<ParamGenerator<T9> >(g9_),
+                               static_cast<ParamGenerator<T10> >(g10_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder10 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+            const Generator8 g8_;
+            const Generator9 g9_;
+            const Generator10 g10_;
+        };  // class CartesianProductHolder10
+
+# endif  // GTEST_HAS_COMBINE
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  //  GTEST_HAS_PARAM_TEST
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing
+{
+
+// Functions producing parameter generators.
+//
+// Google Test uses these generators to produce parameters for value-
+// parameterized tests. When a parameterized test case is instantiated
+// with a particular generator, Google Test creates and runs tests
+// for each element in the sequence produced by the generator.
+//
+// In the following sample, tests from test case FooTest are instantiated
+// each three times with parameter values 3, 5, and 8:
+//
+// class FooTest : public TestWithParam<int> { ... };
+//
+// TEST_P(FooTest, TestThis) {
+// }
+// TEST_P(FooTest, TestThat) {
+// }
+// INSTANTIATE_TEST_CASE_P(TestSequence, FooTest, Values(3, 5, 8));
+//
+
+// Range() returns generators providing sequences of values in a range.
+//
+// Synopsis:
+// Range(start, end)
+//   - returns a generator producing a sequence of values {start, start+1,
+//     start+2, ..., }.
+// Range(start, end, step)
+//   - returns a generator producing a sequence of values {start, start+step,
+//     start+step+step, ..., }.
+// Notes:
+//   * The generated sequences never include end. For example, Range(1, 5)
+//     returns a generator producing a sequence {1, 2, 3, 4}. Range(1, 9, 2)
+//     returns a generator producing {1, 3, 5, 7}.
+//   * start and end must have the same type. That type may be any integral or
+//     floating-point type or a user defined type satisfying these conditions:
+//     * It must be assignable (have operator=() defined).
+//     * It must have operator+() (operator+(int-compatible type) for
+//       two-operand version).
+//     * It must have operator<() defined.
+//     Elements in the resulting sequences will also have that type.
+//   * Condition start < end must be satisfied in order for resulting sequences
+//     to contain any elements.
+//
+    template <typename T, typename IncrementT>
+    internal::ParamGenerator<T> Range(T start, T end, IncrementT step)
+    {
+        return internal::ParamGenerator<T>(
+                   new internal::RangeGenerator<T, IncrementT>(start, end, step));
+    }
+
+    template <typename T>
+    internal::ParamGenerator<T> Range(T start, T end)
+    {
+        return Range(start, end, 1);
+    }
+
+// ValuesIn() function allows generation of tests with parameters coming from
+// a container.
+//
+// Synopsis:
+// ValuesIn(const T (&array)[N])
+//   - returns a generator producing sequences with elements from
+//     a C-style array.
+// ValuesIn(const Container& container)
+//   - returns a generator producing sequences with elements from
+//     an STL-style container.
+// ValuesIn(Iterator begin, Iterator end)
+//   - returns a generator producing sequences with elements from
+//     a range [begin, end) defined by a pair of STL-style iterators. These
+//     iterators can also be plain C pointers.
+//
+// Please note that ValuesIn copies the values from the containers
+// passed in and keeps them to generate tests in RUN_ALL_TESTS().
+//
+// Examples:
+//
+// This instantiates tests from test case StringTest
+// each with C-string values of "foo", "bar", and "baz":
+//
+// const char* strings[] = {"foo", "bar", "baz"};
+// INSTANTIATE_TEST_CASE_P(StringSequence, SrtingTest, ValuesIn(strings));
+//
+// This instantiates tests from test case StlStringTest
+// each with STL strings with values "a" and "b":
+//
+// ::std::vector< ::std::string> GetParameterStrings() {
+//   ::std::vector< ::std::string> v;
+//   v.push_back("a");
+//   v.push_back("b");
+//   return v;
+// }
+//
+// INSTANTIATE_TEST_CASE_P(CharSequence,
+//                         StlStringTest,
+//                         ValuesIn(GetParameterStrings()));
+//
+//
+// This will also instantiate tests from CharTest
+// each with parameter values 'a' and 'b':
+//
+// ::std::list<char> GetParameterChars() {
+//   ::std::list<char> list;
+//   list.push_back('a');
+//   list.push_back('b');
+//   return list;
+// }
+// ::std::list<char> l = GetParameterChars();
+// INSTANTIATE_TEST_CASE_P(CharSequence2,
+//                         CharTest,
+//                         ValuesIn(l.begin(), l.end()));
+//
+    template <typename ForwardIterator>
+    internal::ParamGenerator<
+    typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+    ValuesIn(ForwardIterator begin, ForwardIterator end)
+    {
+        typedef typename ::testing::internal::IteratorTraits<ForwardIterator>
+        ::value_type ParamType;
+        return internal::ParamGenerator<ParamType>(
+                   new internal::ValuesInIteratorRangeGenerator<ParamType>(begin, end));
+    }
+
+    template <typename T, size_t N>
+    internal::ParamGenerator<T> ValuesIn(const T (&array)[N])
+    {
+        return ValuesIn(array, array + N);
+    }
+
+    template <class Container>
+    internal::ParamGenerator<typename Container::value_type> ValuesIn(
+        const Container &container)
+    {
+        return ValuesIn(container.begin(), container.end());
+    }
+
+// Values() allows generating tests from explicitly specified list of
+// parameters.
+//
+// Synopsis:
+// Values(T v1, T v2, ..., T vN)
+//   - returns a generator producing sequences with elements v1, v2, ..., vN.
+//
+// For example, this instantiates tests from test case BarTest each
+// with values "one", "two", and "three":
+//
+// INSTANTIATE_TEST_CASE_P(NumSequence, BarTest, Values("one", "two", "three"));
+//
+// This instantiates tests from test case BazTest each with values 1, 2, 3.5.
+// The exact type of values will depend on the type of parameter in BazTest.
+//
+// INSTANTIATE_TEST_CASE_P(FloatingNumbers, BazTest, Values(1, 2, 3.5));
+//
+// Currently, Values() supports from 1 to 50 parameters.
+//
+    template <typename T1>
+    internal::ValueArray1<T1> Values(T1 v1)
+    {
+        return internal::ValueArray1<T1>(v1);
+    }
+
+    template <typename T1, typename T2>
+    internal::ValueArray2<T1, T2> Values(T1 v1, T2 v2)
+    {
+        return internal::ValueArray2<T1, T2>(v1, v2);
+    }
+
+    template <typename T1, typename T2, typename T3>
+    internal::ValueArray3<T1, T2, T3> Values(T1 v1, T2 v2, T3 v3)
+    {
+        return internal::ValueArray3<T1, T2, T3>(v1, v2, v3);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4>
+    internal::ValueArray4<T1, T2, T3, T4> Values(T1 v1, T2 v2, T3 v3, T4 v4)
+    {
+        return internal::ValueArray4<T1, T2, T3, T4>(v1, v2, v3, v4);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5>
+    internal::ValueArray5<T1, T2, T3, T4, T5> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                     T5 v5)
+    {
+        return internal::ValueArray5<T1, T2, T3, T4, T5>(v1, v2, v3, v4, v5);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6>
+    internal::ValueArray6<T1, T2, T3, T4, T5, T6> Values(T1 v1, T2 v2, T3 v3,
+                                                         T4 v4, T5 v5, T6 v6)
+    {
+        return internal::ValueArray6<T1, T2, T3, T4, T5, T6>(v1, v2, v3, v4, v5, v6);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7>
+    internal::ValueArray7<T1, T2, T3, T4, T5, T6, T7> Values(T1 v1, T2 v2, T3 v3,
+                                                             T4 v4, T5 v5, T6 v6, T7 v7)
+    {
+        return internal::ValueArray7<T1, T2, T3, T4, T5, T6, T7>(v1, v2, v3, v4, v5,
+                                                                 v6, v7);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8>
+    internal::ValueArray8<T1, T2, T3, T4, T5, T6, T7, T8> Values(T1 v1, T2 v2,
+                                                                 T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8)
+    {
+        return internal::ValueArray8<T1, T2, T3, T4, T5, T6, T7, T8>(v1, v2, v3, v4,
+                                                                     v5, v6, v7, v8);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9>
+    internal::ValueArray9<T1, T2, T3, T4, T5, T6, T7, T8, T9> Values(T1 v1, T2 v2,
+                                                                     T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9)
+    {
+        return internal::ValueArray9<T1, T2, T3, T4, T5, T6, T7, T8, T9>(v1, v2, v3,
+                                                                         v4, v5, v6, v7, v8, v9);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10>
+    internal::ValueArray10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> Values(T1 v1,
+                                                                           T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10)
+    {
+        return internal::ValueArray10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10>(v1,
+                                                                               v2, v3, v4, v5, v6, v7, v8, v9, v10);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11>
+    internal::ValueArray11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10,
+             T11> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11)
+    {
+        return internal::ValueArray11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10,
+               T11>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12>
+    internal::ValueArray12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+             T12> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12)
+    {
+        return internal::ValueArray12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13>
+    internal::ValueArray13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+             T13> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13)
+    {
+        return internal::ValueArray13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14>
+    internal::ValueArray14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14)
+    {
+        return internal::ValueArray14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+                              v14);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15>
+    internal::ValueArray15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                              T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15)
+    {
+        return internal::ValueArray15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+                                   v13, v14, v15);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16>
+    internal::ValueArray16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                   T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                   T16 v16)
+    {
+        return internal::ValueArray16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+                                        v12, v13, v14, v15, v16);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17>
+    internal::ValueArray17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                        T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                        T16 v16, T17 v17)
+    {
+        return internal::ValueArray17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+                                             v11, v12, v13, v14, v15, v16, v17);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18>
+    internal::ValueArray18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+                                             T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                             T16 v16, T17 v17, T18 v18)
+    {
+        return internal::ValueArray18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+                                                  v10, v11, v12, v13, v14, v15, v16, v17, v18);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19>
+    internal::ValueArray19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+                                                  T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+                                                  T15 v15, T16 v16, T17 v17, T18 v18, T19 v19)
+    {
+        return internal::ValueArray19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19>(v1, v2, v3, v4, v5, v6, v7, v8,
+                                                       v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20>
+    internal::ValueArray20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                       T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                       T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20)
+    {
+        return internal::ValueArray20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20>(v1, v2, v3, v4, v5, v6, v7,
+                                                            v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21>
+    internal::ValueArray21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                            T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                            T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21)
+    {
+        return internal::ValueArray21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21>(v1, v2, v3, v4, v5, v6,
+                                                                 v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22>
+    internal::ValueArray22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22> Values(T1 v1, T2 v2, T3 v3,
+                                                                 T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                 T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                 T21 v21, T22 v22)
+    {
+        return internal::ValueArray22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22>(v1, v2, v3, v4,
+                                                                      v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                      v20, v21, v22);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23>
+    internal::ValueArray23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> Values(T1 v1, T2 v2,
+                                                                      T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                      T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                      T21 v21, T22 v22, T23 v23)
+    {
+        return internal::ValueArray23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23>(v1, v2, v3,
+                                                                           v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                           v20, v21, v22, v23);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24>
+    internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> Values(T1 v1, T2 v2,
+                                                                           T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                           T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                           T21 v21, T22 v22, T23 v23, T24 v24)
+    {
+        return internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24>(v1, v2,
+                                                                                v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
+                                                                                v19, v20, v21, v22, v23, v24);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25>
+    internal::ValueArray25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Values(T1 v1,
+                                                                                T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11,
+                                                                                T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19,
+                                                                                T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25)
+    {
+        return internal::ValueArray25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25>(v1,
+                                                                                     v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17,
+                                                                                     v18, v19, v20, v21, v22, v23, v24, v25);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26>
+    internal::ValueArray26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+             T26> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26)
+    {
+        return internal::ValueArray26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15,
+                    v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27>
+    internal::ValueArray27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+             T27> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27)
+    {
+        return internal::ValueArray27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14,
+                         v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28>
+    internal::ValueArray28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+             T28> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28)
+    {
+        return internal::ValueArray28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+                              v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27,
+                              v28);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29>
+    internal::ValueArray29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29)
+    {
+        return internal::ValueArray29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+                                   v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26,
+                                   v27, v28, v29);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30>
+    internal::ValueArray30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                              T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+                              T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+                              T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30)
+    {
+        return internal::ValueArray30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+                                        v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25,
+                                        v26, v27, v28, v29, v30);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31>
+    internal::ValueArray31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                   T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                   T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                   T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31)
+    {
+        return internal::ValueArray31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+                                             v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24,
+                                             v25, v26, v27, v28, v29, v30, v31);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32>
+    internal::ValueArray32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                        T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                        T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                        T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                        T32 v32)
+    {
+        return internal::ValueArray32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+                                                  v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                  v24, v25, v26, v27, v28, v29, v30, v31, v32);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33>
+    internal::ValueArray33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+                                             T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                             T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                             T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                             T32 v32, T33 v33)
+    {
+        return internal::ValueArray33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33>(v1, v2, v3, v4, v5, v6, v7, v8,
+                                                       v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                       v24, v25, v26, v27, v28, v29, v30, v31, v32, v33);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34>
+    internal::ValueArray34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+                                                  T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+                                                  T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22,
+                                                  T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30,
+                                                  T31 v31, T32 v32, T33 v33, T34 v34)
+    {
+        return internal::ValueArray34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34>(v1, v2, v3, v4, v5, v6, v7,
+                                                            v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22,
+                                                            v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35>
+    internal::ValueArray35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                       T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                       T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+                                                       T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+                                                       T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35)
+    {
+        return internal::ValueArray35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35>(v1, v2, v3, v4, v5, v6,
+                                                                 v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21,
+                                                                 v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36>
+    internal::ValueArray36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                            T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                            T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+                                                            T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+                                                            T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36)
+    {
+        return internal::ValueArray36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36>(v1, v2, v3, v4,
+                                                                      v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                      v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+                                                                      v34, v35, v36);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37>
+    internal::ValueArray37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37> Values(T1 v1, T2 v2, T3 v3,
+                                                                 T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                 T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                 T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+                                                                 T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+                                                                 T37 v37)
+    {
+        return internal::ValueArray37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37>(v1, v2, v3,
+                                                                           v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                           v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+                                                                           v34, v35, v36, v37);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38>
+    internal::ValueArray38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> Values(T1 v1, T2 v2,
+                                                                      T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                      T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                      T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+                                                                      T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+                                                                      T37 v37, T38 v38)
+    {
+        return internal::ValueArray38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38>(v1, v2,
+                                                                                v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
+                                                                                v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32,
+                                                                                v33, v34, v35, v36, v37, v38);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39>
+    internal::ValueArray39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Values(T1 v1, T2 v2,
+                                                                           T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                           T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                           T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+                                                                           T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+                                                                           T37 v37, T38 v38, T39 v39)
+    {
+        return internal::ValueArray39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39>(v1,
+                                                                                     v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17,
+                                                                                     v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31,
+                                                                                     v32, v33, v34, v35, v36, v37, v38, v39);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40>
+    internal::ValueArray40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Values(T1 v1,
+                                                                                T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11,
+                                                                                T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19,
+                                                                                T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27,
+                                                                                T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35,
+                                                                                T36 v36, T37 v37, T38 v38, T39 v39, T40 v40)
+    {
+        return internal::ValueArray40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15,
+                    v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29,
+                    v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41>
+    internal::ValueArray41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+             T41> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41)
+    {
+        return internal::ValueArray41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14,
+                         v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28,
+                         v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42>
+    internal::ValueArray42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+             T42> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42)
+    {
+        return internal::ValueArray42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+                              v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27,
+                              v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41,
+                              v42);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43>
+    internal::ValueArray43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+             T43> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43)
+    {
+        return internal::ValueArray43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+                                   v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26,
+                                   v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40,
+                                   v41, v42, v43);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44>
+    internal::ValueArray44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44)
+    {
+        return internal::ValueArray44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+                                        v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25,
+                                        v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39,
+                                        v40, v41, v42, v43, v44);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45>
+    internal::ValueArray45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                              T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+                              T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+                              T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32,
+                              T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40,
+                              T41 v41, T42 v42, T43 v43, T44 v44, T45 v45)
+    {
+        return internal::ValueArray45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+                                             v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24,
+                                             v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38,
+                                             v39, v40, v41, v42, v43, v44, v45);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46>
+    internal::ValueArray46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                   T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                   T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                   T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                   T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+                                   T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46)
+    {
+        return internal::ValueArray46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+                                                  v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                  v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37,
+                                                  v38, v39, v40, v41, v42, v43, v44, v45, v46);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47>
+    internal::ValueArray47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                        T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                        T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                        T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                        T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+                                        T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47)
+    {
+        return internal::ValueArray47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47>(v1, v2, v3, v4, v5, v6, v7, v8,
+                                                       v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                       v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37,
+                                                       v38, v39, v40, v41, v42, v43, v44, v45, v46, v47);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48>
+    internal::ValueArray48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47, T48> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+                                             T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                             T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                             T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                             T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+                                             T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47,
+                                             T48 v48)
+    {
+        return internal::ValueArray48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47, T48>(v1, v2, v3, v4, v5, v6, v7,
+                                                            v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22,
+                                                            v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36,
+                                                            v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48, typename T49>
+    internal::ValueArray49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47, T48, T49> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+                                                  T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+                                                  T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22,
+                                                  T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30,
+                                                  T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38,
+                                                  T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46,
+                                                  T47 v47, T48 v48, T49 v49)
+    {
+        return internal::ValueArray49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47, T48, T49>(v1, v2, v3, v4, v5, v6,
+                                                                 v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21,
+                                                                 v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35,
+                                                                 v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48, v49);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48, typename T49, typename T50>
+    internal::ValueArray50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47, T48, T49, T50> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                       T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                       T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+                                                       T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+                                                       T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37,
+                                                       T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45,
+                                                       T46 v46, T47 v47, T48 v48, T49 v49, T50 v50)
+    {
+        return internal::ValueArray50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47, T48, T49, T50>(v1, v2, v3, v4,
+                                                                      v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                      v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+                                                                      v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47,
+                                                                      v48, v49, v50);
+    }
+
+// Bool() allows generating tests with parameters in a set of (false, true).
+//
+// Synopsis:
+// Bool()
+//   - returns a generator producing sequences with elements {false, true}.
+//
+// It is useful when testing code that depends on Boolean flags. Combinations
+// of multiple flags can be tested when several Bool()'s are combined using
+// Combine() function.
+//
+// In the following example all tests in the test case FlagDependentTest
+// will be instantiated twice with parameters false and true.
+//
+// class FlagDependentTest : public testing::TestWithParam<bool> {
+//   virtual void SetUp() {
+//     external_flag = GetParam();
+//   }
+// }
+// INSTANTIATE_TEST_CASE_P(BoolSequence, FlagDependentTest, Bool());
+//
+    inline internal::ParamGenerator<bool> Bool()
+    {
+        return Values(false, true);
+    }
+
+# if GTEST_HAS_COMBINE
+// Combine() allows the user to combine two or more sequences to produce
+// values of a Cartesian product of those sequences' elements.
+//
+// Synopsis:
+// Combine(gen1, gen2, ..., genN)
+//   - returns a generator producing sequences with elements coming from
+//     the Cartesian product of elements from the sequences generated by
+//     gen1, gen2, ..., genN. The sequence elements will have a type of
+//     tuple<T1, T2, ..., TN> where T1, T2, ..., TN are the types
+//     of elements from sequences produces by gen1, gen2, ..., genN.
+//
+// Combine can have up to 10 arguments. This number is currently limited
+// by the maximum number of elements in the tuple implementation used by Google
+// Test.
+//
+// Example:
+//
+// This will instantiate tests in test case AnimalTest each one with
+// the parameter values tuple("cat", BLACK), tuple("cat", WHITE),
+// tuple("dog", BLACK), and tuple("dog", WHITE):
+//
+// enum Color { BLACK, GRAY, WHITE };
+// class AnimalTest
+//     : public testing::TestWithParam<tuple<const char*, Color> > {...};
+//
+// TEST_P(AnimalTest, AnimalLooksNice) {...}
+//
+// INSTANTIATE_TEST_CASE_P(AnimalVariations, AnimalTest,
+//                         Combine(Values("cat", "dog"),
+//                                 Values(BLACK, WHITE)));
+//
+// This will instantiate tests in FlagDependentTest with all variations of two
+// Boolean flags:
+//
+// class FlagDependentTest
+//     : public testing::TestWithParam<tuple<bool, bool> > {
+//   virtual void SetUp() {
+//     // Assigns external_flag_1 and external_flag_2 values from the tuple.
+//     tie(external_flag_1, external_flag_2) = GetParam();
+//   }
+// };
+//
+// TEST_P(FlagDependentTest, TestFeature1) {
+//   // Test your code using external_flag_1 and external_flag_2 here.
+// }
+// INSTANTIATE_TEST_CASE_P(TwoBoolSequence, FlagDependentTest,
+//                         Combine(Bool(), Bool()));
+//
+    template <typename Generator1, typename Generator2>
+    internal::CartesianProductHolder2<Generator1, Generator2> Combine(
+        const Generator1 &g1, const Generator2 &g2)
+    {
+        return internal::CartesianProductHolder2<Generator1, Generator2>(
+                   g1, g2);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3>
+    internal::CartesianProductHolder3<Generator1, Generator2, Generator3> Combine(
+        const Generator1 &g1, const Generator2 &g2, const Generator3 &g3)
+    {
+        return internal::CartesianProductHolder3<Generator1, Generator2, Generator3>(
+                   g1, g2, g3);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4>
+    internal::CartesianProductHolder4<Generator1, Generator2, Generator3,
+             Generator4> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4)
+    {
+        return internal::CartesianProductHolder4<Generator1, Generator2, Generator3,
+               Generator4>(
+                   g1, g2, g3, g4);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5>
+    internal::CartesianProductHolder5<Generator1, Generator2, Generator3,
+             Generator4, Generator5> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5)
+    {
+        return internal::CartesianProductHolder5<Generator1, Generator2, Generator3,
+               Generator4, Generator5>(
+                   g1, g2, g3, g4, g5);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6>
+    internal::CartesianProductHolder6<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6)
+    {
+        return internal::CartesianProductHolder6<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6>(
+                   g1, g2, g3, g4, g5, g6);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7>
+    internal::CartesianProductHolder7<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7)
+    {
+        return internal::CartesianProductHolder7<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7>(
+                   g1, g2, g3, g4, g5, g6, g7);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7, typename Generator8>
+    internal::CartesianProductHolder8<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7, Generator8> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7, const Generator8 &g8)
+    {
+        return internal::CartesianProductHolder8<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7, Generator8>(
+                   g1, g2, g3, g4, g5, g6, g7, g8);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7, typename Generator8, typename Generator9>
+    internal::CartesianProductHolder9<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7, Generator8,
+             Generator9> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7, const Generator8 &g8, const Generator9 &g9)
+    {
+        return internal::CartesianProductHolder9<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7, Generator8, Generator9>(
+                   g1, g2, g3, g4, g5, g6, g7, g8, g9);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7, typename Generator8, typename Generator9,
+              typename Generator10>
+    internal::CartesianProductHolder10<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7, Generator8, Generator9,
+             Generator10> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7, const Generator8 &g8, const Generator9 &g9,
+                 const Generator10 &g10)
+    {
+        return internal::CartesianProductHolder10<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7, Generator8, Generator9,
+               Generator10>(
+                   g1, g2, g3, g4, g5, g6, g7, g8, g9, g10);
+    }
+# endif  // GTEST_HAS_COMBINE
+
+
+
+# define TEST_P(test_case_name, test_name) \
+  class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \
+      : public test_case_name { \
+   public: \
+    GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {} \
+    virtual void TestBody(); \
+   private: \
+    static int AddToRegistry() { \
+      ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+          GetTestCasePatternHolder<test_case_name>(\
+              #test_case_name, \
+              ::testing::internal::CodeLocation(\
+                  __FILE__, __LINE__))->AddTestPattern(\
+                      #test_case_name, \
+                      #test_name, \
+                      new ::testing::internal::TestMetaFactory< \
+                          GTEST_TEST_CLASS_NAME_(\
+                              test_case_name, test_name)>()); \
+      return 0; \
+    } \
+    static int gtest_registering_dummy_ GTEST_ATTRIBUTE_UNUSED_; \
+    GTEST_DISALLOW_COPY_AND_ASSIGN_(\
+        GTEST_TEST_CLASS_NAME_(test_case_name, test_name)); \
+  }; \
+  int GTEST_TEST_CLASS_NAME_(test_case_name, \
+                             test_name)::gtest_registering_dummy_ = \
+      GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::AddToRegistry(); \
+  void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody()
+
+// The optional last argument to INSTANTIATE_TEST_CASE_P allows the user
+// to specify a function or functor that generates custom test name suffixes
+// based on the test parameters. The function should accept one argument of
+// type testing::TestParamInfo<class ParamType>, and return std::string.
+//
+// testing::PrintToStringParamName is a builtin test suffix generator that
+// returns the value of testing::PrintToString(GetParam()). It does not work
+// for std::string or C strings.
+//
+// Note: test names must be non-empty, unique, and may only contain ASCII
+// alphanumeric characters or underscore.
+
+# define INSTANTIATE_TEST_CASE_P(prefix, test_case_name, generator, ...) \
+  ::testing::internal::ParamGenerator<test_case_name::ParamType> \
+      gtest_##prefix##test_case_name##_EvalGenerator_() { return generator; } \
+  ::std::string gtest_##prefix##test_case_name##_EvalGenerateName_( \
+      const ::testing::TestParamInfo<test_case_name::ParamType>& info) { \
+    return ::testing::internal::GetParamNameGen<test_case_name::ParamType> \
+        (__VA_ARGS__)(info); \
+  } \
+  int gtest_##prefix##test_case_name##_dummy_ GTEST_ATTRIBUTE_UNUSED_ = \
+      ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+          GetTestCasePatternHolder<test_case_name>(\
+              #test_case_name, \
+              ::testing::internal::CodeLocation(\
+                  __FILE__, __LINE__))->AddTestCaseInstantiation(\
+                      #prefix, \
+                      &gtest_##prefix##test_case_name##_EvalGenerator_, \
+                      &gtest_##prefix##test_case_name##_EvalGenerateName_, \
+                      __FILE__, __LINE__)
+
+}  // namespace testing
+
+#endif  // GTEST_HAS_PARAM_TEST
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Google C++ Testing Framework definitions useful in production code.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+
+// When you need to test the private or protected members of a class,
+// use the FRIEND_TEST macro to declare your tests as friends of the
+// class.  For example:
+//
+// class MyClass {
+//  private:
+//   void MyMethod();
+//   FRIEND_TEST(MyClassTest, MyMethod);
+// };
+//
+// class MyClassTest : public testing::Test {
+//   // ...
+// };
+//
+// TEST_F(MyClassTest, MyMethod) {
+//   // Can call MyClass::MyMethod() here.
+// }
+
+#define FRIEND_TEST(test_case_name, test_name)\
+friend class test_case_name##_##test_name##_Test
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+#define GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+
+#include <iosfwd>
+#include <vector>
+
+namespace testing
+{
+
+// A copyable object representing the result of a test part (i.e. an
+// assertion or an explicit FAIL(), ADD_FAILURE(), or SUCCESS()).
+//
+// Don't inherit from TestPartResult as its destructor is not virtual.
+    class GTEST_API_ TestPartResult
+    {
+    public:
+        // The possible outcomes of a test part (i.e. an assertion or an
+        // explicit SUCCEED(), FAIL(), or ADD_FAILURE()).
+        enum Type {
+            kSuccess,          // Succeeded.
+            kNonFatalFailure,  // Failed but the test can continue.
+            kFatalFailure,     // Failed and the test should be terminated.
+            kSkip              // Skipped.
+        };
+
+        // C'tor.  TestPartResult does NOT have a default constructor.
+        // Always use this constructor (with parameters) to create a
+        // TestPartResult object.
+        TestPartResult(Type a_type,
+                       const char *a_file_name,
+                       int a_line_number,
+                       const char *a_message)
+            : type_(a_type),
+              file_name_(a_file_name == NULL ? "" : a_file_name),
+              line_number_(a_line_number),
+              summary_(ExtractSummary(a_message)),
+              message_(a_message)
+        {
+        }
+
+        // Gets the outcome of the test part.
+        Type type() const
+        {
+            return type_;
+        }
+
+        // Gets the name of the source file where the test part took place, or
+        // NULL if it's unknown.
+        const char *file_name() const
+        {
+            return file_name_.empty() ? NULL : file_name_.c_str();
+        }
+
+        // Gets the line in the source file where the test part took place,
+        // or -1 if it's unknown.
+        int line_number() const
+        {
+            return line_number_;
+        }
+
+        // Gets the summary of the failure message.
+        const char *summary() const
+        {
+            return summary_.c_str();
+        }
+
+        // Gets the message associated with the test part.
+        const char *message() const
+        {
+            return message_.c_str();
+        }
+
+        // Returns true if and only if the test part was skipped.
+        bool skipped() const
+        {
+            return type_ == kSkip;
+        }
+
+        // Returns true iff the test part passed.
+        bool passed() const
+        {
+            return type_ == kSuccess;
+        }
+
+        // Returns true iff the test part failed.
+        bool failed() const
+        {
+            return fatally_failed() || nonfatally_failed();
+        }
+
+        // Returns true iff the test part non-fatally failed.
+        bool nonfatally_failed() const
+        {
+            return type_ == kNonFatalFailure;
+        }
+
+        // Returns true iff the test part fatally failed.
+        bool fatally_failed() const
+        {
+            return type_ == kFatalFailure;
+        }
+
+    private:
+        Type type_;
+
+        // Gets the summary of the failure message by omitting the stack
+        // trace in it.
+        static std::string ExtractSummary(const char *message);
+
+        // The name of the source file where the test part took place, or
+        // "" if the source file is unknown.
+        std::string file_name_;
+        // The line in the source file where the test part took place, or -1
+        // if the line number is unknown.
+        int line_number_;
+        std::string summary_;  // The test failure summary.
+        std::string message_;  // The test failure message.
+    };
+
+// Prints a TestPartResult object.
+    std::ostream &operator<<(std::ostream &os, const TestPartResult &result);
+
+// An array of TestPartResult objects.
+//
+// Don't inherit from TestPartResultArray as its destructor is not
+// virtual.
+    class GTEST_API_ TestPartResultArray
+    {
+    public:
+        TestPartResultArray() {}
+
+        // Appends the given TestPartResult to the array.
+        void Append(const TestPartResult &result);
+
+        // Returns the TestPartResult at the given index (0-based).
+        const TestPartResult &GetTestPartResult(int index) const;
+
+        // Returns the number of TestPartResult objects in the array.
+        int size() const;
+
+    private:
+        std::vector<TestPartResult> array_;
+
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestPartResultArray);
+    };
+
+// This interface knows how to report a test part result.
+    class TestPartResultReporterInterface
+    {
+    public:
+        virtual ~TestPartResultReporterInterface() {}
+
+        virtual void ReportTestPartResult(const TestPartResult &result) = 0;
+    };
+
+    namespace internal
+    {
+
+// This helper class is used by {ASSERT|EXPECT}_NO_FATAL_FAILURE to check if a
+// statement generates new fatal failures. To do so it registers itself as the
+// current test part result reporter. Besides checking if fatal failures were
+// reported, it only delegates the reporting to the former result reporter.
+// The original result reporter is restored in the destructor.
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        class GTEST_API_ HasNewFatalFailureHelper
+            : public TestPartResultReporterInterface
+        {
+        public:
+            HasNewFatalFailureHelper();
+            virtual ~HasNewFatalFailureHelper();
+            virtual void ReportTestPartResult(const TestPartResult &result);
+            bool has_new_fatal_failure() const
+            {
+                return has_new_fatal_failure_;
+            }
+        private:
+            bool has_new_fatal_failure_;
+            TestPartResultReporterInterface *original_reporter_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(HasNewFatalFailureHelper);
+        };
+
+    }  // namespace internal
+
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+
+// This header implements typed tests and type-parameterized tests.
+
+// Typed (aka type-driven) tests repeat the same test for types in a
+// list.  You must know which types you want to test with when writing
+// typed tests. Here's how you do it:
+
+#if 0
+
+// First, define a fixture class template.  It should be parameterized
+// by a type.  Remember to derive it from testing::Test.
+template <typename T>
+class FooTest : public testing::Test
+{
+public:
+    ...
+    typedef std::list<T> List;
+    static T shared_;
+    T value_;
+};
+
+// Next, associate a list of types with the test case, which will be
+// repeated for each type in the list.  The typedef is necessary for
+// the macro to parse correctly.
+typedef testing::Types<char, int, unsigned int> MyTypes;
+TYPED_TEST_CASE(FooTest, MyTypes);
+
+// If the type list contains only one type, you can write that type
+// directly without Types<...>:
+//   TYPED_TEST_CASE(FooTest, int);
+
+// Then, use TYPED_TEST() instead of TEST_F() to define as many typed
+// tests for this test case as you want.
+TYPED_TEST(FooTest, DoesBlah)
+{
+    // Inside a test, refer to TypeParam to get the type parameter.
+    // Since we are inside a derived class template, C++ requires use to
+    // visit the members of FooTest via 'this'.
+    TypeParam n = this->value_;
+
+    // To visit static members of the fixture, add the TestFixture::
+    // prefix.
+    n += TestFixture::shared_;
+
+    // To refer to typedefs in the fixture, add the "typename
+    // TestFixture::" prefix.
+    typename TestFixture::List values;
+    values.push_back(n);
+    ...
+}
+
+TYPED_TEST(FooTest, HasPropertyA)
+{
+    ...
+}
+
+#endif  // 0
+
+// Type-parameterized tests are abstract test patterns parameterized
+// by a type.  Compared with typed tests, type-parameterized tests
+// allow you to define the test pattern without knowing what the type
+// parameters are.  The defined pattern can be instantiated with
+// different types any number of times, in any number of translation
+// units.
+//
+// If you are designing an interface or concept, you can define a
+// suite of type-parameterized tests to verify properties that any
+// valid implementation of the interface/concept should have.  Then,
+// each implementation can easily instantiate the test suite to verify
+// that it conforms to the requirements, without having to write
+// similar tests repeatedly.  Here's an example:
+
+#if 0
+
+// First, define a fixture class template.  It should be parameterized
+// by a type.  Remember to derive it from testing::Test.
+template <typename T>
+class FooTest : public testing::Test
+{
+    ...
+};
+
+// Next, declare that you will define a type-parameterized test case
+// (the _P suffix is for "parameterized" or "pattern", whichever you
+// prefer):
+TYPED_TEST_CASE_P(FooTest);
+
+// Then, use TYPED_TEST_P() to define as many type-parameterized tests
+// for this type-parameterized test case as you want.
+TYPED_TEST_P(FooTest, DoesBlah)
+{
+    // Inside a test, refer to TypeParam to get the type parameter.
+    TypeParam n = 0;
+    ...
+}
+
+TYPED_TEST_P(FooTest, HasPropertyA)
+{
+    ...
+}
+
+// Now the tricky part: you need to register all test patterns before
+// you can instantiate them.  The first argument of the macro is the
+// test case name; the rest are the names of the tests in this test
+// case.
+REGISTER_TYPED_TEST_CASE_P(FooTest,
+                           DoesBlah, HasPropertyA);
+
+// Finally, you are free to instantiate the pattern with the types you
+// want.  If you put the above code in a header file, you can #include
+// it in multiple C++ source files and instantiate it multiple times.
+//
+// To distinguish different instances of the pattern, the first
+// argument to the INSTANTIATE_* macro is a prefix that will be added
+// to the actual test case name.  Remember to pick unique prefixes for
+// different instances.
+typedef testing::Types<char, int, unsigned int> MyTypes;
+INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, MyTypes);
+
+// If the type list contains only one type, you can write that type
+// directly without Types<...>:
+//   INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, int);
+
+#endif  // 0
+
+
+// Implements typed tests.
+
+#if GTEST_HAS_TYPED_TEST
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the name of the typedef for the type parameters of the
+// given test case.
+# define GTEST_TYPE_PARAMS_(TestCaseName) gtest_type_params_##TestCaseName##_
+
+// The 'Types' template argument below must have spaces around it
+// since some compilers may choke on '>>' when passing a template
+// instance (e.g. Types<int>)
+# define TYPED_TEST_CASE(CaseName, Types) \
+  typedef ::testing::internal::TypeList< Types >::type \
+      GTEST_TYPE_PARAMS_(CaseName)
+
+# define TYPED_TEST(CaseName, TestName) \
+  template <typename gtest_TypeParam_> \
+  class GTEST_TEST_CLASS_NAME_(CaseName, TestName) \
+      : public CaseName<gtest_TypeParam_> { \
+   private: \
+    typedef CaseName<gtest_TypeParam_> TestFixture; \
+    typedef gtest_TypeParam_ TypeParam; \
+    virtual void TestBody(); \
+  }; \
+  bool gtest_##CaseName##_##TestName##_registered_ GTEST_ATTRIBUTE_UNUSED_ = \
+      ::testing::internal::TypeParameterizedTest< \
+          CaseName, \
+          ::testing::internal::TemplateSel< \
+              GTEST_TEST_CLASS_NAME_(CaseName, TestName)>, \
+          GTEST_TYPE_PARAMS_(CaseName)>::Register(\
+              "", ::testing::internal::CodeLocation(__FILE__, __LINE__), \
+              #CaseName, #TestName, 0); \
+  template <typename gtest_TypeParam_> \
+  void GTEST_TEST_CLASS_NAME_(CaseName, TestName)<gtest_TypeParam_>::TestBody()
+
+#endif  // GTEST_HAS_TYPED_TEST
+
+// Implements type-parameterized tests.
+
+#if GTEST_HAS_TYPED_TEST_P
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the namespace name that the type-parameterized tests for
+// the given type-parameterized test case are defined in.  The exact
+// name of the namespace is subject to change without notice.
+# define GTEST_CASE_NAMESPACE_(TestCaseName) \
+  gtest_case_##TestCaseName##_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the name of the variable used to remember the names of
+// the defined tests in the given test case.
+# define GTEST_TYPED_TEST_CASE_P_STATE_(TestCaseName) \
+  gtest_typed_test_case_p_state_##TestCaseName##_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE DIRECTLY.
+//
+// Expands to the name of the variable used to remember the names of
+// the registered tests in the given test case.
+# define GTEST_REGISTERED_TEST_NAMES_(TestCaseName) \
+  gtest_registered_test_names_##TestCaseName##_
+
+// The variables defined in the type-parameterized test macros are
+// static as typically these macros are used in a .h file that can be
+// #included in multiple translation units linked together.
+# define TYPED_TEST_CASE_P(CaseName) \
+  static ::testing::internal::TypedTestCasePState \
+      GTEST_TYPED_TEST_CASE_P_STATE_(CaseName)
+
+# define TYPED_TEST_P(CaseName, TestName) \
+  namespace GTEST_CASE_NAMESPACE_(CaseName) { \
+  template <typename gtest_TypeParam_> \
+  class TestName : public CaseName<gtest_TypeParam_> { \
+   private: \
+    typedef CaseName<gtest_TypeParam_> TestFixture; \
+    typedef gtest_TypeParam_ TypeParam; \
+    virtual void TestBody(); \
+  }; \
+  static bool gtest_##TestName##_defined_ GTEST_ATTRIBUTE_UNUSED_ = \
+      GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).AddTestName(\
+          __FILE__, __LINE__, #CaseName, #TestName); \
+  } \
+  template <typename gtest_TypeParam_> \
+  void GTEST_CASE_NAMESPACE_(CaseName)::TestName<gtest_TypeParam_>::TestBody()
+
+# define REGISTER_TYPED_TEST_CASE_P(CaseName, ...) \
+  namespace GTEST_CASE_NAMESPACE_(CaseName) { \
+  typedef ::testing::internal::Templates<__VA_ARGS__>::type gtest_AllTests_; \
+  } \
+  static const char* const GTEST_REGISTERED_TEST_NAMES_(CaseName) \
+      GTEST_ATTRIBUTE_UNUSED_ = \
+          GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).VerifyRegisteredTestNames(\
+              __FILE__, __LINE__, #__VA_ARGS__)
+
+// The 'Types' template argument below must have spaces around it
+// since some compilers may choke on '>>' when passing a template
+// instance (e.g. Types<int>)
+# define INSTANTIATE_TYPED_TEST_CASE_P(Prefix, CaseName, Types) \
+  bool gtest_##Prefix##_##CaseName GTEST_ATTRIBUTE_UNUSED_ = \
+      ::testing::internal::TypeParameterizedTestCase<CaseName, \
+          GTEST_CASE_NAMESPACE_(CaseName)::gtest_AllTests_, \
+          ::testing::internal::TypeList< Types >::type>::Register(\
+              #Prefix, \
+              ::testing::internal::CodeLocation(__FILE__, __LINE__), \
+              &GTEST_TYPED_TEST_CASE_P_STATE_(CaseName), \
+              #CaseName, GTEST_REGISTERED_TEST_NAMES_(CaseName))
+
+#endif  // GTEST_HAS_TYPED_TEST_P
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+
+// Depending on the platform, different string classes are available.
+// On Linux, in addition to ::std::string, Google also makes use of
+// class ::string, which has the same interface as ::std::string, but
+// has a different implementation.
+//
+// You can define GTEST_HAS_GLOBAL_STRING to 1 to indicate that
+// ::string is available AND is a distinct type to ::std::string, or
+// define it to 0 to indicate otherwise.
+//
+// If ::std::string and ::string are the same class on your platform
+// due to aliasing, you should define GTEST_HAS_GLOBAL_STRING to 0.
+//
+// If you do not define GTEST_HAS_GLOBAL_STRING, it is defined
+// heuristically.
+
+namespace testing
+{
+
+// Declares the flags.
+
+// This flag temporary enables the disabled tests.
+    GTEST_DECLARE_bool_(also_run_disabled_tests);
+
+// This flag brings the debugger on an assertion failure.
+    GTEST_DECLARE_bool_(break_on_failure);
+
+// This flag controls whether Google Test catches all test-thrown exceptions
+// and logs them as failures.
+    GTEST_DECLARE_bool_(catch_exceptions);
+
+// This flag enables using colors in terminal output. Available values are
+// "yes" to enable colors, "no" (disable colors), or "auto" (the default)
+// to let Google Test decide.
+    GTEST_DECLARE_string_(color);
+
+// This flag sets up the filter to select by name using a glob pattern
+// the tests to run. If the filter is not given all tests are executed.
+    GTEST_DECLARE_string_(filter);
+
+// This flag causes the Google Test to list tests. None of the tests listed
+// are actually run if the flag is provided.
+    GTEST_DECLARE_bool_(list_tests);
+
+// This flag controls whether Google Test emits a detailed XML report to a file
+// in addition to its normal textual output.
+    GTEST_DECLARE_string_(output);
+
+// This flags control whether Google Test prints the elapsed time for each
+// test.
+    GTEST_DECLARE_bool_(print_time);
+
+// This flag specifies the random number seed.
+    GTEST_DECLARE_int32_(random_seed);
+
+// This flag sets how many times the tests are repeated. The default value
+// is 1. If the value is -1 the tests are repeating forever.
+    GTEST_DECLARE_int32_(repeat);
+
+// This flag controls whether Google Test includes Google Test internal
+// stack frames in failure stack traces.
+    GTEST_DECLARE_bool_(show_internal_stack_frames);
+
+// When this flag is specified, tests' order is randomized on every iteration.
+    GTEST_DECLARE_bool_(shuffle);
+
+// This flag specifies the maximum number of stack frames to be
+// printed in a failure message.
+    GTEST_DECLARE_int32_(stack_trace_depth);
+
+// When this flag is specified, a failed assertion will throw an
+// exception if exceptions are enabled, or exit the program with a
+// non-zero code otherwise.
+    GTEST_DECLARE_bool_(throw_on_failure);
+
+// When this flag is set with a "host:port" string, on supported
+// platforms test results are streamed to the specified port on
+// the specified host machine.
+    GTEST_DECLARE_string_(stream_result_to);
+
+// The upper limit for valid stack trace depths.
+    const int kMaxStackTraceDepth = 100;
+
+    namespace internal
+    {
+
+        class AssertHelper;
+        class DefaultGlobalTestPartResultReporter;
+        class ExecDeathTest;
+        class NoExecDeathTest;
+        class FinalSuccessChecker;
+        class GTestFlagSaver;
+        class StreamingListenerTest;
+        class TestResultAccessor;
+        class TestEventListenersAccessor;
+        class TestEventRepeater;
+        class UnitTestRecordPropertyTestHelper;
+        class WindowsDeathTest;
+        class UnitTestImpl *GetUnitTestImpl();
+        void ReportFailureInUnknownLocation(TestPartResult::Type result_type,
+                                            const std::string &message);
+
+    }  // namespace internal
+
+// The friend relationship of some of these classes is cyclic.
+// If we don't forward declare them the compiler might confuse the classes
+// in friendship clauses with same named classes on the scope.
+    class Test;
+    class TestCase;
+    class TestInfo;
+    class UnitTest;
+
+// A class for indicating whether an assertion was successful.  When
+// the assertion wasn't successful, the AssertionResult object
+// remembers a non-empty message that describes how it failed.
+//
+// To create an instance of this class, use one of the factory functions
+// (AssertionSuccess() and AssertionFailure()).
+//
+// This class is useful for two purposes:
+//   1. Defining predicate functions to be used with Boolean test assertions
+//      EXPECT_TRUE/EXPECT_FALSE and their ASSERT_ counterparts
+//   2. Defining predicate-format functions to be
+//      used with predicate assertions (ASSERT_PRED_FORMAT*, etc).
+//
+// For example, if you define IsEven predicate:
+//
+//   testing::AssertionResult IsEven(int n) {
+//     if ((n % 2) == 0)
+//       return testing::AssertionSuccess();
+//     else
+//       return testing::AssertionFailure() << n << " is odd";
+//   }
+//
+// Then the failed expectation EXPECT_TRUE(IsEven(Fib(5)))
+// will print the message
+//
+//   Value of: IsEven(Fib(5))
+//     Actual: false (5 is odd)
+//   Expected: true
+//
+// instead of a more opaque
+//
+//   Value of: IsEven(Fib(5))
+//     Actual: false
+//   Expected: true
+//
+// in case IsEven is a simple Boolean predicate.
+//
+// If you expect your predicate to be reused and want to support informative
+// messages in EXPECT_FALSE and ASSERT_FALSE (negative assertions show up
+// about half as often as positive ones in our tests), supply messages for
+// both success and failure cases:
+//
+//   testing::AssertionResult IsEven(int n) {
+//     if ((n % 2) == 0)
+//       return testing::AssertionSuccess() << n << " is even";
+//     else
+//       return testing::AssertionFailure() << n << " is odd";
+//   }
+//
+// Then a statement EXPECT_FALSE(IsEven(Fib(6))) will print
+//
+//   Value of: IsEven(Fib(6))
+//     Actual: true (8 is even)
+//   Expected: false
+//
+// NB: Predicates that support negative Boolean assertions have reduced
+// performance in positive ones so be careful not to use them in tests
+// that have lots (tens of thousands) of positive Boolean assertions.
+//
+// To use this class with EXPECT_PRED_FORMAT assertions such as:
+//
+//   // Verifies that Foo() returns an even number.
+//   EXPECT_PRED_FORMAT1(IsEven, Foo());
+//
+// you need to define:
+//
+//   testing::AssertionResult IsEven(const char* expr, int n) {
+//     if ((n % 2) == 0)
+//       return testing::AssertionSuccess();
+//     else
+//       return testing::AssertionFailure()
+//         << "Expected: " << expr << " is even\n  Actual: it's " << n;
+//   }
+//
+// If Foo() returns 5, you will see the following message:
+//
+//   Expected: Foo() is even
+//     Actual: it's 5
+//
+    class GTEST_API_ AssertionResult
+    {
+    public:
+        // Copy constructor.
+        // Used in EXPECT_TRUE/FALSE(assertion_result).
+        AssertionResult(const AssertionResult &other);
+
+        GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 /* forcing value to bool */)
+
+        // Used in the EXPECT_TRUE/FALSE(bool_expression).
+        //
+        // T must be contextually convertible to bool.
+        //
+        // The second parameter prevents this overload from being considered if
+        // the argument is implicitly convertible to AssertionResult. In that case
+        // we want AssertionResult's copy constructor to be used.
+        template <typename T>
+        explicit AssertionResult(
+            const T &success,
+            typename internal::EnableIf<
+            !internal::ImplicitlyConvertible<T, AssertionResult>::value>::type *
+            /*enabler*/ = NULL)
+            : success_(success) {}
+
+        GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+        // Assignment operator.
+        AssertionResult &operator=(AssertionResult other)
+        {
+            swap(other);
+            return *this;
+        }
+
+        // Returns true iff the assertion succeeded.
+        operator bool() const
+        {
+            return success_;    // NOLINT
+        }
+
+        // Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE.
+        AssertionResult operator!() const;
+
+        // Returns the text streamed into this AssertionResult. Test assertions
+        // use it when they fail (i.e., the predicate's outcome doesn't match the
+        // assertion's expectation). When nothing has been streamed into the
+        // object, returns an empty string.
+        const char *message() const
+        {
+            return message_.get() != NULL ?  message_->c_str() : "";
+        }
+        // TODO(vladl@google.com): Remove this after making sure no clients use it.
+        // Deprecated; please use message() instead.
+        const char *failure_message() const
+        {
+            return message();
+        }
+
+        // Streams a custom failure message into this object.
+        template <typename T> AssertionResult &operator<<(const T &value)
+        {
+            AppendMessage(Message() << value);
+            return *this;
+        }
+
+        // Allows streaming basic output manipulators such as endl or flush into
+        // this object.
+        AssertionResult &operator<<(
+            ::std::ostream& (*basic_manipulator)(::std::ostream &stream))
+        {
+            AppendMessage(Message() << basic_manipulator);
+            return *this;
+        }
+
+    private:
+        // Appends the contents of message to message_.
+        void AppendMessage(const Message &a_message)
+        {
+            if (message_.get() == NULL)
+                message_.reset(new ::std::string);
+            message_->append(a_message.GetString().c_str());
+        }
+
+        // Swap the contents of this AssertionResult with other.
+        void swap(AssertionResult &other);
+
+        // Stores result of the assertion predicate.
+        bool success_;
+        // Stores the message describing the condition in case the expectation
+        // construct is not satisfied with the predicate's outcome.
+        // Referenced via a pointer to avoid taking too much stack frame space
+        // with test assertions.
+        internal::scoped_ptr< ::std::string> message_;
+    };
+
+// Makes a successful assertion result.
+    GTEST_API_ AssertionResult AssertionSuccess();
+
+// Makes a failed assertion result.
+    GTEST_API_ AssertionResult AssertionFailure();
+
+// Makes a failed assertion result with the given failure message.
+// Deprecated; use AssertionFailure() << msg.
+    GTEST_API_ AssertionResult AssertionFailure(const Message &msg);
+
+// The abstract class that all tests inherit from.
+//
+// In Google Test, a unit test program contains one or many TestCases, and
+// each TestCase contains one or many Tests.
+//
+// When you define a test using the TEST macro, you don't need to
+// explicitly derive from Test - the TEST macro automatically does
+// this for you.
+//
+// The only time you derive from Test is when defining a test fixture
+// to be used a TEST_F.  For example:
+//
+//   class FooTest : public testing::Test {
+//    protected:
+//     void SetUp() override { ... }
+//     void TearDown() override { ... }
+//     ...
+//   };
+//
+//   TEST_F(FooTest, Bar) { ... }
+//   TEST_F(FooTest, Baz) { ... }
+//
+// Test is not copyable.
+    class GTEST_API_ Test
+    {
+    public:
+        friend class TestInfo;
+
+        // Defines types for pointers to functions that set up and tear down
+        // a test case.
+        typedef internal::SetUpTestCaseFunc SetUpTestCaseFunc;
+        typedef internal::TearDownTestCaseFunc TearDownTestCaseFunc;
+
+        // The d'tor is virtual as we intend to inherit from Test.
+        virtual ~Test();
+
+        // Sets up the stuff shared by all tests in this test case.
+        //
+        // Google Test will call Foo::SetUpTestCase() before running the first
+        // test in test case Foo.  Hence a sub-class can define its own
+        // SetUpTestCase() method to shadow the one defined in the super
+        // class.
+        static void SetUpTestCase() {}
+
+        // Tears down the stuff shared by all tests in this test case.
+        //
+        // Google Test will call Foo::TearDownTestCase() after running the last
+        // test in test case Foo.  Hence a sub-class can define its own
+        // TearDownTestCase() method to shadow the one defined in the super
+        // class.
+        static void TearDownTestCase() {}
+
+        // Returns true iff the current test has a fatal failure.
+        static bool HasFatalFailure();
+
+        // Returns true iff the current test has a non-fatal failure.
+        static bool HasNonfatalFailure();
+
+        // Returns true if and only if the current test was skipped.
+        static bool IsSkipped();
+
+        // Returns true iff the current test has a (either fatal or
+        // non-fatal) failure.
+        static bool HasFailure()
+        {
+            return HasFatalFailure() || HasNonfatalFailure();
+        }
+
+        // Logs a property for the current test, test case, or for the entire
+        // invocation of the test program when used outside of the context of a
+        // test case.  Only the last value for a given key is remembered.  These
+        // are public static so they can be called from utility functions that are
+        // not members of the test fixture.  Calls to RecordProperty made during
+        // lifespan of the test (from the moment its constructor starts to the
+        // moment its destructor finishes) will be output in XML as attributes of
+        // the <testcase> element.  Properties recorded from fixture's
+        // SetUpTestCase or TearDownTestCase are logged as attributes of the
+        // corresponding <testsuite> element.  Calls to RecordProperty made in the
+        // global context (before or after invocation of RUN_ALL_TESTS and from
+        // SetUp/TearDown method of Environment objects registered with Google
+        // Test) will be output as attributes of the <testsuites> element.
+        static void RecordProperty(const std::string &key, const std::string &value);
+        static void RecordProperty(const std::string &key, int value);
+
+    protected:
+        // Creates a Test object.
+        Test();
+
+        // Sets up the test fixture.
+        virtual void SetUp();
+
+        // Tears down the test fixture.
+        virtual void TearDown();
+
+    private:
+        // Returns true iff the current test has the same fixture class as
+        // the first test in the current test case.
+        static bool HasSameFixtureClass();
+
+        // Runs the test after the test fixture has been set up.
+        //
+        // A sub-class must implement this to define the test logic.
+        //
+        // DO NOT OVERRIDE THIS FUNCTION DIRECTLY IN A USER PROGRAM.
+        // Instead, use the TEST or TEST_F macro.
+        virtual void TestBody() = 0;
+
+        // Sets up, executes, and tears down the test.
+        void Run();
+
+        // Deletes self.  We deliberately pick an unusual name for this
+        // internal method to avoid clashing with names used in user TESTs.
+        void DeleteSelf_()
+        {
+            delete this;
+        }
+
+        const internal::scoped_ptr< GTEST_FLAG_SAVER_ > gtest_flag_saver_;
+
+        // Often a user misspells SetUp() as Setup() and spends a long time
+        // wondering why it is never called by Google Test.  The declaration of
+        // the following method is solely for catching such an error at
+        // compile time:
+        //
+        //   - The return type is deliberately chosen to be not void, so it
+        //   will be a conflict if void Setup() is declared in the user's
+        //   test fixture.
+        //
+        //   - This method is private, so it will be another compiler error
+        //   if the method is called from the user's test fixture.
+        //
+        // DO NOT OVERRIDE THIS FUNCTION.
+        //
+        // If you see an error about overriding the following function or
+        // about it being private, you have mis-spelled SetUp() as Setup().
+        struct Setup_should_be_spelled_SetUp {};
+        virtual Setup_should_be_spelled_SetUp *Setup()
+        {
+            return NULL;
+        }
+
+        // We disallow copying Tests.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(Test);
+    };
+
+    typedef internal::TimeInMillis TimeInMillis;
+
+// A copyable object representing a user specified test property which can be
+// output as a key/value string pair.
+//
+// Don't inherit from TestProperty as its destructor is not virtual.
+    class TestProperty
+    {
+    public:
+        // C'tor.  TestProperty does NOT have a default constructor.
+        // Always use this constructor (with parameters) to create a
+        // TestProperty object.
+        TestProperty(const std::string &a_key, const std::string &a_value) :
+            key_(a_key), value_(a_value)
+        {
+        }
+
+        // Gets the user supplied key.
+        const char *key() const
+        {
+            return key_.c_str();
+        }
+
+        // Gets the user supplied value.
+        const char *value() const
+        {
+            return value_.c_str();
+        }
+
+        // Sets a new value, overriding the one supplied in the constructor.
+        void SetValue(const std::string &new_value)
+        {
+            value_ = new_value;
+        }
+
+    private:
+        // The key supplied by the user.
+        std::string key_;
+        // The value supplied by the user.
+        std::string value_;
+    };
+
+// The result of a single Test.  This includes a list of
+// TestPartResults, a list of TestProperties, a count of how many
+// death tests there are in the Test, and how much time it took to run
+// the Test.
+//
+// TestResult is not copyable.
+    class GTEST_API_ TestResult
+    {
+    public:
+        // Creates an empty TestResult.
+        TestResult();
+
+        // D'tor.  Do not inherit from TestResult.
+        ~TestResult();
+
+        // Gets the number of all test parts.  This is the sum of the number
+        // of successful test parts and the number of failed test parts.
+        int total_part_count() const;
+
+        // Returns the number of the test properties.
+        int test_property_count() const;
+
+        // Returns true if the test passed (i.e. no test part failed).
+        bool Passed() const
+        {
+            return !Skipped() && !Failed();
+        }
+
+        // Returns true if and only if the test was skipped.
+        bool Skipped() const;
+
+        // Returns true iff the test failed.
+        bool Failed() const;
+
+        // Returns true iff the test fatally failed.
+        bool HasFatalFailure() const;
+
+        // Returns true iff the test has a non-fatal failure.
+        bool HasNonfatalFailure() const;
+
+        // Returns true if and only if the current test was skipped.
+        static bool IsSkipped();
+
+        // Returns the elapsed time, in milliseconds.
+        TimeInMillis elapsed_time() const
+        {
+            return elapsed_time_;
+        }
+
+        // Returns the i-th test part result among all the results. i can range
+        // from 0 to test_property_count() - 1. If i is not in that range, aborts
+        // the program.
+        const TestPartResult &GetTestPartResult(int i) const;
+
+        // Returns the i-th test property. i can range from 0 to
+        // test_property_count() - 1. If i is not in that range, aborts the
+        // program.
+        const TestProperty &GetTestProperty(int i) const;
+
+    private:
+        friend class TestInfo;
+        friend class TestCase;
+        friend class UnitTest;
+        friend class internal::DefaultGlobalTestPartResultReporter;
+        friend class internal::ExecDeathTest;
+        friend class internal::TestResultAccessor;
+        friend class internal::UnitTestImpl;
+        friend class internal::WindowsDeathTest;
+
+        // Gets the vector of TestPartResults.
+        const std::vector<TestPartResult> &test_part_results() const
+        {
+            return test_part_results_;
+        }
+
+        // Gets the vector of TestProperties.
+        const std::vector<TestProperty> &test_properties() const
+        {
+            return test_properties_;
+        }
+
+        // Sets the elapsed time.
+        void set_elapsed_time(TimeInMillis elapsed)
+        {
+            elapsed_time_ = elapsed;
+        }
+
+        // Adds a test property to the list. The property is validated and may add
+        // a non-fatal failure if invalid (e.g., if it conflicts with reserved
+        // key names). If a property is already recorded for the same key, the
+        // value will be updated, rather than storing multiple values for the same
+        // key.  xml_element specifies the element for which the property is being
+        // recorded and is used for validation.
+        void RecordProperty(const std::string &xml_element,
+                            const TestProperty &test_property);
+
+        // Adds a failure if the key is a reserved attribute of Google Test
+        // testcase tags.  Returns true if the property is valid.
+        // TODO(russr): Validate attribute names are legal and human readable.
+        static bool ValidateTestProperty(const std::string &xml_element,
+                                         const TestProperty &test_property);
+
+        // Adds a test part result to the list.
+        void AddTestPartResult(const TestPartResult &test_part_result);
+
+        // Returns the death test count.
+        int death_test_count() const
+        {
+            return death_test_count_;
+        }
+
+        // Increments the death test count, returning the new count.
+        int increment_death_test_count()
+        {
+            return ++death_test_count_;
+        }
+
+        // Clears the test part results.
+        void ClearTestPartResults();
+
+        // Clears the object.
+        void Clear();
+
+        // Protects mutable state of the property vector and of owned
+        // properties, whose values may be updated.
+        internal::Mutex test_properites_mutex_;
+
+        // The vector of TestPartResults
+        std::vector<TestPartResult> test_part_results_;
+        // The vector of TestProperties
+        std::vector<TestProperty> test_properties_;
+        // Running count of death tests.
+        int death_test_count_;
+        // The elapsed time, in milliseconds.
+        TimeInMillis elapsed_time_;
+
+        // We disallow copying TestResult.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestResult);
+    };  // class TestResult
+
+// A TestInfo object stores the following information about a test:
+//
+//   Test case name
+//   Test name
+//   Whether the test should be run
+//   A function pointer that creates the test object when invoked
+//   Test result
+//
+// The constructor of TestInfo registers itself with the UnitTest
+// singleton such that the RUN_ALL_TESTS() macro knows which tests to
+// run.
+    class GTEST_API_ TestInfo
+    {
+    public:
+        // Destructs a TestInfo object.  This function is not virtual, so
+        // don't inherit from TestInfo.
+        ~TestInfo();
+
+        // Returns the test case name.
+        const char *test_case_name() const
+        {
+            return test_case_name_.c_str();
+        }
+
+        // Returns the test name.
+        const char *name() const
+        {
+            return name_.c_str();
+        }
+
+        // Returns the name of the parameter type, or NULL if this is not a typed
+        // or a type-parameterized test.
+        const char *type_param() const
+        {
+            if (type_param_.get() != NULL)
+                return type_param_->c_str();
+            return NULL;
+        }
+
+        // Returns the text representation of the value parameter, or NULL if this
+        // is not a value-parameterized test.
+        const char *value_param() const
+        {
+            if (value_param_.get() != NULL)
+                return value_param_->c_str();
+            return NULL;
+        }
+
+        // Returns the file name where this test is defined.
+        const char *file() const
+        {
+            return location_.file.c_str();
+        }
+
+        // Returns the line where this test is defined.
+        int line() const
+        {
+            return location_.line;
+        }
+
+        // Returns true if this test should run, that is if the test is not
+        // disabled (or it is disabled but the also_run_disabled_tests flag has
+        // been specified) and its full name matches the user-specified filter.
+        //
+        // Google Test allows the user to filter the tests by their full names.
+        // The full name of a test Bar in test case Foo is defined as
+        // "Foo.Bar".  Only the tests that match the filter will run.
+        //
+        // A filter is a colon-separated list of glob (not regex) patterns,
+        // optionally followed by a '-' and a colon-separated list of
+        // negative patterns (tests to exclude).  A test is run if it
+        // matches one of the positive patterns and does not match any of
+        // the negative patterns.
+        //
+        // For example, *A*:Foo.* is a filter that matches any string that
+        // contains the character 'A' or starts with "Foo.".
+        bool should_run() const
+        {
+            return should_run_;
+        }
+
+        // Returns true iff this test will appear in the XML report.
+        bool is_reportable() const
+        {
+            // For now, the XML report includes all tests matching the filter.
+            // In the future, we may trim tests that are excluded because of
+            // sharding.
+            return matches_filter_;
+        }
+
+        // Returns the result of the test.
+        const TestResult *result() const
+        {
+            return &result_;
+        }
+
+    private:
+#if GTEST_HAS_DEATH_TEST
+        friend class internal::DefaultDeathTestFactory;
+#endif  // GTEST_HAS_DEATH_TEST
+        friend class Test;
+        friend class TestCase;
+        friend class internal::UnitTestImpl;
+        friend class internal::StreamingListenerTest;
+        friend TestInfo *internal::MakeAndRegisterTestInfo(
+            const char *test_case_name,
+            const char *name,
+            const char *type_param,
+            const char *value_param,
+            internal::CodeLocation code_location,
+            internal::TypeId fixture_class_id,
+            Test::SetUpTestCaseFunc set_up_tc,
+            Test::TearDownTestCaseFunc tear_down_tc,
+            internal::TestFactoryBase *factory);
+
+        // Constructs a TestInfo object. The newly constructed instance assumes
+        // ownership of the factory object.
+        TestInfo(const std::string &test_case_name,
+                 const std::string &name,
+                 const char *a_type_param,   // NULL if not a type-parameterized test
+                 const char *a_value_param,  // NULL if not a value-parameterized test
+                 internal::CodeLocation a_code_location,
+                 internal::TypeId fixture_class_id,
+                 internal::TestFactoryBase *factory);
+
+        // Increments the number of death tests encountered in this test so
+        // far.
+        int increment_death_test_count()
+        {
+            return result_.increment_death_test_count();
+        }
+
+        // Creates the test object, runs it, records its result, and then
+        // deletes it.
+        void Run();
+
+        static void ClearTestResult(TestInfo *test_info)
+        {
+            test_info->result_.Clear();
+        }
+
+        // These fields are immutable properties of the test.
+        const std::string test_case_name_;     // Test case name
+        const std::string name_;               // Test name
+        // Name of the parameter type, or NULL if this is not a typed or a
+        // type-parameterized test.
+        const internal::scoped_ptr<const ::std::string> type_param_;
+        // Text representation of the value parameter, or NULL if this is not a
+        // value-parameterized test.
+        const internal::scoped_ptr<const ::std::string> value_param_;
+        internal::CodeLocation location_;
+        const internal::TypeId fixture_class_id_;   // ID of the test fixture class
+        bool should_run_;                 // True iff this test should run
+        bool is_disabled_;                // True iff this test is disabled
+        bool matches_filter_;             // True if this test matches the
+        // user-specified filter.
+        internal::TestFactoryBase *const factory_;  // The factory that creates
+        // the test object
+
+        // This field is mutable and needs to be reset before running the
+        // test for the second time.
+        TestResult result_;
+
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestInfo);
+    };
+
+// A test case, which consists of a vector of TestInfos.
+//
+// TestCase is not copyable.
+    class GTEST_API_ TestCase
+    {
+    public:
+        // Creates a TestCase with the given name.
+        //
+        // TestCase does NOT have a default constructor.  Always use this
+        // constructor to create a TestCase object.
+        //
+        // Arguments:
+        //
+        //   name:         name of the test case
+        //   a_type_param: the name of the test's type parameter, or NULL if
+        //                 this is not a type-parameterized test.
+        //   set_up_tc:    pointer to the function that sets up the test case
+        //   tear_down_tc: pointer to the function that tears down the test case
+        TestCase(const char *name, const char *a_type_param,
+                 Test::SetUpTestCaseFunc set_up_tc,
+                 Test::TearDownTestCaseFunc tear_down_tc);
+
+        // Destructor of TestCase.
+        virtual ~TestCase();
+
+        // Gets the name of the TestCase.
+        const char *name() const
+        {
+            return name_.c_str();
+        }
+
+        // Returns the name of the parameter type, or NULL if this is not a
+        // type-parameterized test case.
+        const char *type_param() const
+        {
+            if (type_param_.get() != NULL)
+                return type_param_->c_str();
+            return NULL;
+        }
+
+        // Returns true if any test in this test case should run.
+        bool should_run() const
+        {
+            return should_run_;
+        }
+
+        // Gets the number of successful tests in this test case.
+        int successful_test_count() const;
+
+        // Gets the number of skipped tests in this test suite.
+        int skipped_test_count() const;
+
+        // Gets the number of failed tests in this test case.
+        int failed_test_count() const;
+
+        // Gets the number of disabled tests that will be reported in the XML report.
+        int reportable_disabled_test_count() const;
+
+        // Gets the number of disabled tests in this test case.
+        int disabled_test_count() const;
+
+        // Gets the number of tests to be printed in the XML report.
+        int reportable_test_count() const;
+
+        // Get the number of tests in this test case that should run.
+        int test_to_run_count() const;
+
+        // Gets the number of all tests in this test case.
+        int total_test_count() const;
+
+        // Returns true iff the test case passed.
+        bool Passed() const
+        {
+            return !Failed();
+        }
+
+        // Returns true iff the test case failed.
+        bool Failed() const
+        {
+            return failed_test_count() > 0;
+        }
+
+        // Returns the elapsed time, in milliseconds.
+        TimeInMillis elapsed_time() const
+        {
+            return elapsed_time_;
+        }
+
+        // Returns the i-th test among all the tests. i can range from 0 to
+        // total_test_count() - 1. If i is not in that range, returns NULL.
+        const TestInfo *GetTestInfo(int i) const;
+
+        // Returns the TestResult that holds test properties recorded during
+        // execution of SetUpTestCase and TearDownTestCase.
+        const TestResult &ad_hoc_test_result() const
+        {
+            return ad_hoc_test_result_;
+        }
+
+    private:
+        friend class Test;
+        friend class internal::UnitTestImpl;
+
+        // Gets the (mutable) vector of TestInfos in this TestCase.
+        std::vector<TestInfo *> &test_info_list()
+        {
+            return test_info_list_;
+        }
+
+        // Gets the (immutable) vector of TestInfos in this TestCase.
+        const std::vector<TestInfo *> &test_info_list() const
+        {
+            return test_info_list_;
+        }
+
+        // Returns the i-th test among all the tests. i can range from 0 to
+        // total_test_count() - 1. If i is not in that range, returns NULL.
+        TestInfo *GetMutableTestInfo(int i);
+
+        // Sets the should_run member.
+        void set_should_run(bool should)
+        {
+            should_run_ = should;
+        }
+
+        // Adds a TestInfo to this test case.  Will delete the TestInfo upon
+        // destruction of the TestCase object.
+        void AddTestInfo(TestInfo *test_info);
+
+        // Clears the results of all tests in this test case.
+        void ClearResult();
+
+        // Clears the results of all tests in the given test case.
+        static void ClearTestCaseResult(TestCase *test_case)
+        {
+            test_case->ClearResult();
+        }
+
+        // Runs every test in this TestCase.
+        void Run();
+
+        // Runs SetUpTestCase() for this TestCase.  This wrapper is needed
+        // for catching exceptions thrown from SetUpTestCase().
+        void RunSetUpTestCase()
+        {
+            (*set_up_tc_)();
+        }
+
+        // Runs TearDownTestCase() for this TestCase.  This wrapper is
+        // needed for catching exceptions thrown from TearDownTestCase().
+        void RunTearDownTestCase()
+        {
+            (*tear_down_tc_)();
+        }
+
+        // Returns true iff test passed.
+        static bool TestPassed(const TestInfo *test_info)
+        {
+            return test_info->should_run() && test_info->result()->Passed();
+        }
+
+        // Returns true if and only if test skipped.
+        static bool TestSkipped(const TestInfo *test_info)
+        {
+            return test_info->should_run() && test_info->result()->Skipped();
+        }
+
+        // Returns true iff test failed.
+        static bool TestFailed(const TestInfo *test_info)
+        {
+            return test_info->should_run() && test_info->result()->Failed();
+        }
+
+        // Returns true iff the test is disabled and will be reported in the XML
+        // report.
+        static bool TestReportableDisabled(const TestInfo *test_info)
+        {
+            return test_info->is_reportable() && test_info->is_disabled_;
+        }
+
+        // Returns true iff test is disabled.
+        static bool TestDisabled(const TestInfo *test_info)
+        {
+            return test_info->is_disabled_;
+        }
+
+        // Returns true iff this test will appear in the XML report.
+        static bool TestReportable(const TestInfo *test_info)
+        {
+            return test_info->is_reportable();
+        }
+
+        // Returns true if the given test should run.
+        static bool ShouldRunTest(const TestInfo *test_info)
+        {
+            return test_info->should_run();
+        }
+
+        // Shuffles the tests in this test case.
+        void ShuffleTests(internal::Random *random);
+
+        // Restores the test order to before the first shuffle.
+        void UnshuffleTests();
+
+        // Name of the test case.
+        std::string name_;
+        // Name of the parameter type, or NULL if this is not a typed or a
+        // type-parameterized test.
+        const internal::scoped_ptr<const ::std::string> type_param_;
+        // The vector of TestInfos in their original order.  It owns the
+        // elements in the vector.
+        std::vector<TestInfo *> test_info_list_;
+        // Provides a level of indirection for the test list to allow easy
+        // shuffling and restoring the test order.  The i-th element in this
+        // vector is the index of the i-th test in the shuffled test list.
+        std::vector<int> test_indices_;
+        // Pointer to the function that sets up the test case.
+        Test::SetUpTestCaseFunc set_up_tc_;
+        // Pointer to the function that tears down the test case.
+        Test::TearDownTestCaseFunc tear_down_tc_;
+        // True iff any test in this test case should run.
+        bool should_run_;
+        // Elapsed time, in milliseconds.
+        TimeInMillis elapsed_time_;
+        // Holds test properties recorded during execution of SetUpTestCase and
+        // TearDownTestCase.
+        TestResult ad_hoc_test_result_;
+
+        // We disallow copying TestCases.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestCase);
+    };
+
+// An Environment object is capable of setting up and tearing down an
+// environment.  You should subclass this to define your own
+// environment(s).
+//
+// An Environment object does the set-up and tear-down in virtual
+// methods SetUp() and TearDown() instead of the constructor and the
+// destructor, as:
+//
+//   1. You cannot safely throw from a destructor.  This is a problem
+//      as in some cases Google Test is used where exceptions are enabled, and
+//      we may want to implement ASSERT_* using exceptions where they are
+//      available.
+//   2. You cannot use ASSERT_* directly in a constructor or
+//      destructor.
+    class Environment
+    {
+    public:
+        // The d'tor is virtual as we need to subclass Environment.
+        virtual ~Environment() {}
+
+        // Override this to define how to set up the environment.
+        virtual void SetUp() {}
+
+        // Override this to define how to tear down the environment.
+        virtual void TearDown() {}
+    private:
+        // If you see an error about overriding the following function or
+        // about it being private, you have mis-spelled SetUp() as Setup().
+        struct Setup_should_be_spelled_SetUp {};
+        virtual Setup_should_be_spelled_SetUp *Setup()
+        {
+            return NULL;
+        }
+    };
+
+// The interface for tracing execution of tests. The methods are organized in
+// the order the corresponding events are fired.
+    class TestEventListener
+    {
+    public:
+        virtual ~TestEventListener() {}
+
+        // Fired before any test activity starts.
+        virtual void OnTestProgramStart(const UnitTest &unit_test) = 0;
+
+        // Fired before each iteration of tests starts.  There may be more than
+        // one iteration if GTEST_FLAG(repeat) is set. iteration is the iteration
+        // index, starting from 0.
+        virtual void OnTestIterationStart(const UnitTest &unit_test,
+                                          int iteration) = 0;
+
+        // Fired before environment set-up for each iteration of tests starts.
+        virtual void OnEnvironmentsSetUpStart(const UnitTest &unit_test) = 0;
+
+        // Fired after environment set-up for each iteration of tests ends.
+        virtual void OnEnvironmentsSetUpEnd(const UnitTest &unit_test) = 0;
+
+        // Fired before the test case starts.
+        virtual void OnTestCaseStart(const TestCase &test_case) = 0;
+
+        // Fired before the test starts.
+        virtual void OnTestStart(const TestInfo &test_info) = 0;
+
+        // Fired after a failed assertion or a SUCCEED() invocation.
+        virtual void OnTestPartResult(const TestPartResult &test_part_result) = 0;
+
+        // Fired after the test ends.
+        virtual void OnTestEnd(const TestInfo &test_info) = 0;
+
+        // Fired after the test case ends.
+        virtual void OnTestCaseEnd(const TestCase &test_case) = 0;
+
+        // Fired before environment tear-down for each iteration of tests starts.
+        virtual void OnEnvironmentsTearDownStart(const UnitTest &unit_test) = 0;
+
+        // Fired after environment tear-down for each iteration of tests ends.
+        virtual void OnEnvironmentsTearDownEnd(const UnitTest &unit_test) = 0;
+
+        // Fired after each iteration of tests finishes.
+        virtual void OnTestIterationEnd(const UnitTest &unit_test,
+                                        int iteration) = 0;
+
+        // Fired after all test activities have ended.
+        virtual void OnTestProgramEnd(const UnitTest &unit_test) = 0;
+    };
+
+// The convenience class for users who need to override just one or two
+// methods and are not concerned that a possible change to a signature of
+// the methods they override will not be caught during the build.  For
+// comments about each method please see the definition of TestEventListener
+// above.
+    class EmptyTestEventListener : public TestEventListener
+    {
+    public:
+        virtual void OnTestProgramStart(const UnitTest & /*unit_test*/) {}
+        virtual void OnTestIterationStart(const UnitTest & /*unit_test*/,
+                                          int /*iteration*/) {}
+        virtual void OnEnvironmentsSetUpStart(const UnitTest & /*unit_test*/) {}
+        virtual void OnEnvironmentsSetUpEnd(const UnitTest & /*unit_test*/) {}
+        virtual void OnTestCaseStart(const TestCase & /*test_case*/) {}
+        virtual void OnTestStart(const TestInfo & /*test_info*/) {}
+        virtual void OnTestPartResult(const TestPartResult & /*test_part_result*/) {}
+        virtual void OnTestEnd(const TestInfo & /*test_info*/) {}
+        virtual void OnTestCaseEnd(const TestCase & /*test_case*/) {}
+        virtual void OnEnvironmentsTearDownStart(const UnitTest & /*unit_test*/) {}
+        virtual void OnEnvironmentsTearDownEnd(const UnitTest & /*unit_test*/) {}
+        virtual void OnTestIterationEnd(const UnitTest & /*unit_test*/,
+                                        int /*iteration*/) {}
+        virtual void OnTestProgramEnd(const UnitTest & /*unit_test*/) {}
+    };
+
+// TestEventListeners lets users add listeners to track events in Google Test.
+    class GTEST_API_ TestEventListeners
+    {
+    public:
+        TestEventListeners();
+        ~TestEventListeners();
+
+        // Appends an event listener to the end of the list. Google Test assumes
+        // the ownership of the listener (i.e. it will delete the listener when
+        // the test program finishes).
+        void Append(TestEventListener *listener);
+
+        // Removes the given event listener from the list and returns it.  It then
+        // becomes the caller's responsibility to delete the listener. Returns
+        // NULL if the listener is not found in the list.
+        TestEventListener *Release(TestEventListener *listener);
+
+        // Returns the standard listener responsible for the default console
+        // output.  Can be removed from the listeners list to shut down default
+        // console output.  Note that removing this object from the listener list
+        // with Release transfers its ownership to the caller and makes this
+        // function return NULL the next time.
+        TestEventListener *default_result_printer() const
+        {
+            return default_result_printer_;
+        }
+
+        // Returns the standard listener responsible for the default XML output
+        // controlled by the --gtest_output=xml flag.  Can be removed from the
+        // listeners list by users who want to shut down the default XML output
+        // controlled by this flag and substitute it with custom one.  Note that
+        // removing this object from the listener list with Release transfers its
+        // ownership to the caller and makes this function return NULL the next
+        // time.
+        TestEventListener *default_xml_generator() const
+        {
+            return default_xml_generator_;
+        }
+
+    private:
+        friend class TestCase;
+        friend class TestInfo;
+        friend class internal::DefaultGlobalTestPartResultReporter;
+        friend class internal::NoExecDeathTest;
+        friend class internal::TestEventListenersAccessor;
+        friend class internal::UnitTestImpl;
+
+        // Returns repeater that broadcasts the TestEventListener events to all
+        // subscribers.
+        TestEventListener *repeater();
+
+        // Sets the default_result_printer attribute to the provided listener.
+        // The listener is also added to the listener list and previous
+        // default_result_printer is removed from it and deleted. The listener can
+        // also be NULL in which case it will not be added to the list. Does
+        // nothing if the previous and the current listener objects are the same.
+        void SetDefaultResultPrinter(TestEventListener *listener);
+
+        // Sets the default_xml_generator attribute to the provided listener.  The
+        // listener is also added to the listener list and previous
+        // default_xml_generator is removed from it and deleted. The listener can
+        // also be NULL in which case it will not be added to the list. Does
+        // nothing if the previous and the current listener objects are the same.
+        void SetDefaultXmlGenerator(TestEventListener *listener);
+
+        // Controls whether events will be forwarded by the repeater to the
+        // listeners in the list.
+        bool EventForwardingEnabled() const;
+        void SuppressEventForwarding();
+
+        // The actual list of listeners.
+        internal::TestEventRepeater *repeater_;
+        // Listener responsible for the standard result output.
+        TestEventListener *default_result_printer_;
+        // Listener responsible for the creation of the XML output file.
+        TestEventListener *default_xml_generator_;
+
+        // We disallow copying TestEventListeners.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventListeners);
+    };
+
+// A UnitTest consists of a vector of TestCases.
+//
+// This is a singleton class.  The only instance of UnitTest is
+// created when UnitTest::GetInstance() is first called.  This
+// instance is never deleted.
+//
+// UnitTest is not copyable.
+//
+// This class is thread-safe as long as the methods are called
+// according to their specification.
+    class GTEST_API_ UnitTest
+    {
+    public:
+        // Gets the singleton UnitTest object.  The first time this method
+        // is called, a UnitTest object is constructed and returned.
+        // Consecutive calls will return the same object.
+        static UnitTest *GetInstance();
+
+        // Runs all tests in this UnitTest object and prints the result.
+        // Returns 0 if successful, or 1 otherwise.
+        //
+        // This method can only be called from the main thread.
+        //
+        // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        int Run() GTEST_MUST_USE_RESULT_;
+
+        // Returns the working directory when the first TEST() or TEST_F()
+        // was executed.  The UnitTest object owns the string.
+        const char *original_working_dir() const;
+
+        // Returns the TestCase object for the test that's currently running,
+        // or NULL if no test is running.
+        const TestCase *current_test_case() const
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Returns the TestInfo object for the test that's currently running,
+        // or NULL if no test is running.
+        const TestInfo *current_test_info() const
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Returns the random seed used at the start of the current test run.
+        int random_seed() const;
+
+#if GTEST_HAS_PARAM_TEST
+        // Returns the ParameterizedTestCaseRegistry object used to keep track of
+        // value-parameterized tests and instantiate and register them.
+        //
+        // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        internal::ParameterizedTestCaseRegistry &parameterized_test_registry()
+        GTEST_LOCK_EXCLUDED_(mutex_);
+#endif  // GTEST_HAS_PARAM_TEST
+
+        // Gets the number of successful test cases.
+        int successful_test_case_count() const;
+
+        // Gets the number of failed test cases.
+        int failed_test_case_count() const;
+
+        // Gets the number of all test cases.
+        int total_test_case_count() const;
+
+        // Gets the number of all test cases that contain at least one test
+        // that should run.
+        int test_case_to_run_count() const;
+
+        // Gets the number of successful tests.
+        int successful_test_count() const;
+
+        // Gets the number of skipped tests.
+        int skipped_test_count() const;
+
+        // Gets the number of failed tests.
+        int failed_test_count() const;
+
+        // Gets the number of disabled tests that will be reported in the XML report.
+        int reportable_disabled_test_count() const;
+
+        // Gets the number of disabled tests.
+        int disabled_test_count() const;
+
+        // Gets the number of tests to be printed in the XML report.
+        int reportable_test_count() const;
+
+        // Gets the number of all tests.
+        int total_test_count() const;
+
+        // Gets the number of tests that should run.
+        int test_to_run_count() const;
+
+        // Gets the time of the test program start, in ms from the start of the
+        // UNIX epoch.
+        TimeInMillis start_timestamp() const;
+
+        // Gets the elapsed time, in milliseconds.
+        TimeInMillis elapsed_time() const;
+
+        // Returns true iff the unit test passed (i.e. all test cases passed).
+        bool Passed() const;
+
+        // Returns true iff the unit test failed (i.e. some test case failed
+        // or something outside of all tests failed).
+        bool Failed() const;
+
+        // Gets the i-th test case among all the test cases. i can range from 0 to
+        // total_test_case_count() - 1. If i is not in that range, returns NULL.
+        const TestCase *GetTestCase(int i) const;
+
+        // Returns the TestResult containing information on test failures and
+        // properties logged outside of individual test cases.
+        const TestResult &ad_hoc_test_result() const;
+
+        // Returns the list of event listeners that can be used to track events
+        // inside Google Test.
+        TestEventListeners &listeners();
+
+    private:
+        // Registers and returns a global test environment.  When a test
+        // program is run, all global test environments will be set-up in
+        // the order they were registered.  After all tests in the program
+        // have finished, all global test environments will be torn-down in
+        // the *reverse* order they were registered.
+        //
+        // The UnitTest object takes ownership of the given environment.
+        //
+        // This method can only be called from the main thread.
+        Environment *AddEnvironment(Environment *env);
+
+        // Adds a TestPartResult to the current TestResult object.  All
+        // Google Test assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc)
+        // eventually call this to report their results.  The user code
+        // should use the assertion macros instead of calling this directly.
+        void AddTestPartResult(TestPartResult::Type result_type,
+                               const char *file_name,
+                               int line_number,
+                               const std::string &message,
+                               const std::string &os_stack_trace)
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Adds a TestProperty to the current TestResult object when invoked from
+        // inside a test, to current TestCase's ad_hoc_test_result_ when invoked
+        // from SetUpTestCase or TearDownTestCase, or to the global property set
+        // when invoked elsewhere.  If the result already contains a property with
+        // the same key, the value will be updated.
+        void RecordProperty(const std::string &key, const std::string &value);
+
+        // Gets the i-th test case among all the test cases. i can range from 0 to
+        // total_test_case_count() - 1. If i is not in that range, returns NULL.
+        TestCase *GetMutableTestCase(int i);
+
+        // Accessors for the implementation object.
+        internal::UnitTestImpl *impl()
+        {
+            return impl_;
+        }
+        const internal::UnitTestImpl *impl() const
+        {
+            return impl_;
+        }
+
+        // These classes and functions are friends as they need to access private
+        // members of UnitTest.
+        friend class Test;
+        friend class internal::AssertHelper;
+        friend class internal::ScopedTrace;
+        friend class internal::StreamingListenerTest;
+        friend class internal::UnitTestRecordPropertyTestHelper;
+        friend Environment *AddGlobalTestEnvironment(Environment *env);
+        friend internal::UnitTestImpl *internal::GetUnitTestImpl();
+        friend void internal::ReportFailureInUnknownLocation(
+            TestPartResult::Type result_type,
+            const std::string &message);
+
+        // Creates an empty UnitTest.
+        UnitTest();
+
+        // D'tor
+        virtual ~UnitTest();
+
+        // Pushes a trace defined by SCOPED_TRACE() on to the per-thread
+        // Google Test trace stack.
+        void PushGTestTrace(const internal::TraceInfo &trace)
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Pops a trace from the per-thread Google Test trace stack.
+        void PopGTestTrace()
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Protects mutable state in *impl_.  This is mutable as some const
+        // methods need to lock it too.
+        mutable internal::Mutex mutex_;
+
+        // Opaque implementation object.  This field is never changed once
+        // the object is constructed.  We don't mark it as const here, as
+        // doing so will cause a warning in the constructor of UnitTest.
+        // Mutable state in *impl_ is protected by mutex_.
+        internal::UnitTestImpl *impl_;
+
+        // We disallow copying UnitTest.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTest);
+    };
+
+// A convenient wrapper for adding an environment for the test
+// program.
+//
+// You should call this before RUN_ALL_TESTS() is called, probably in
+// main().  If you use gtest_main, you need to call this before main()
+// starts for it to take effect.  For example, you can define a global
+// variable like this:
+//
+//   testing::Environment* const foo_env =
+//       testing::AddGlobalTestEnvironment(new FooEnvironment);
+//
+// However, we strongly recommend you to write your own main() and
+// call AddGlobalTestEnvironment() there, as relying on initialization
+// of global variables makes the code harder to read and may cause
+// problems when you register multiple environments from different
+// translation units and the environments have dependencies among them
+// (remember that the compiler doesn't guarantee the order in which
+// global variables from different translation units are initialized).
+    inline Environment *AddGlobalTestEnvironment(Environment *env)
+    {
+        return UnitTest::GetInstance()->AddEnvironment(env);
+    }
+
+// Initializes Google Test.  This must be called before calling
+// RUN_ALL_TESTS().  In particular, it parses a command line for the
+// flags that Google Test recognizes.  Whenever a Google Test flag is
+// seen, it is removed from argv, and *argc is decremented.
+//
+// No value is returned.  Instead, the Google Test flag variables are
+// updated.
+//
+// Calling the function for the second time has no user-visible effect.
+    GTEST_API_ void InitGoogleTest(int *argc, char **argv);
+
+// This overloaded version can be used in Windows programs compiled in
+// UNICODE mode.
+    GTEST_API_ void InitGoogleTest(int *argc, wchar_t **argv);
+    GTEST_API_ bool GetGtestHelpFlag();
+
+    namespace internal
+    {
+
+// Separate the error generating code from the code path to reduce the stack
+// frame size of CmpHelperEQ. This helps reduce the overhead of some sanitizers
+// when calling EXPECT_* in a tight loop.
+        template <typename T1, typename T2>
+        AssertionResult CmpHelperEQFailure(const char *lhs_expression,
+                                           const char *rhs_expression,
+                                           const T1 &lhs, const T2 &rhs)
+        {
+            return EqFailure(lhs_expression,
+                             rhs_expression,
+                             FormatForComparisonFailureMessage(lhs, rhs),
+                             FormatForComparisonFailureMessage(rhs, lhs),
+                             false);
+        }
+
+// The helper function for {ASSERT|EXPECT}_EQ.
+        template <typename T1, typename T2>
+        AssertionResult CmpHelperEQ(const char *lhs_expression,
+                                    const char *rhs_expression,
+                                    const T1 &lhs,
+                                    const T2 &rhs)
+        {
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4389 /* signed/unsigned mismatch */)
+            if (lhs == rhs) {
+                return AssertionSuccess();
+            }
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+            return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+        }
+
+// With this overloaded version, we allow anonymous enums to be used
+// in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous enums
+// can be implicitly cast to BiggestInt.
+        GTEST_API_ AssertionResult CmpHelperEQ(const char *lhs_expression,
+                                               const char *rhs_expression,
+                                               BiggestInt lhs,
+                                               BiggestInt rhs);
+
+// The helper class for {ASSERT|EXPECT}_EQ.  The template argument
+// lhs_is_null_literal is true iff the first argument to ASSERT_EQ()
+// is a null pointer literal.  The following default implementation is
+// for lhs_is_null_literal being false.
+        template <bool lhs_is_null_literal>
+        class EqHelper
+        {
+        public:
+            // This templatized version is for the general case.
+            template <typename T1, typename T2>
+            static AssertionResult Compare(const char *lhs_expression,
+                                           const char *rhs_expression,
+                                           const T1 &lhs,
+                                           const T2 &rhs)
+            {
+                return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs);
+            }
+
+            // With this overloaded version, we allow anonymous enums to be used
+            // in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous
+            // enums can be implicitly cast to BiggestInt.
+            //
+            // Even though its body looks the same as the above version, we
+            // cannot merge the two, as it will make anonymous enums unhappy.
+            static AssertionResult Compare(const char *lhs_expression,
+                                           const char *rhs_expression,
+                                           BiggestInt lhs,
+                                           BiggestInt rhs)
+            {
+                return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs);
+            }
+        };
+
+// This specialization is used when the first argument to ASSERT_EQ()
+// is a null pointer literal, like NULL, false, or 0.
+        template <>
+        class EqHelper<true>
+        {
+        public:
+            // We define two overloaded versions of Compare().  The first
+            // version will be picked when the second argument to ASSERT_EQ() is
+            // NOT a pointer, e.g. ASSERT_EQ(0, AnIntFunction()) or
+            // EXPECT_EQ(false, a_bool).
+            template <typename T1, typename T2>
+            static AssertionResult Compare(
+                const char *lhs_expression,
+                const char *rhs_expression,
+                const T1 &lhs,
+                const T2 &rhs,
+                // The following line prevents this overload from being considered if T2
+                // is not a pointer type.  We need this because ASSERT_EQ(NULL, my_ptr)
+                // expands to Compare("", "", NULL, my_ptr), which requires a conversion
+                // to match the Secret* in the other overload, which would otherwise make
+                // this template match better.
+                typename EnableIf<!is_pointer<T2>::value>::type * = 0)
+            {
+                return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs);
+            }
+
+            // This version will be picked when the second argument to ASSERT_EQ() is a
+            // pointer, e.g. ASSERT_EQ(NULL, a_pointer).
+            template <typename T>
+            static AssertionResult Compare(
+                const char *lhs_expression,
+                const char *rhs_expression,
+                // We used to have a second template parameter instead of Secret*.  That
+                // template parameter would deduce to 'long', making this a better match
+                // than the first overload even without the first overload's EnableIf.
+                // Unfortunately, gcc with -Wconversion-null warns when "passing NULL to
+                // non-pointer argument" (even a deduced integral argument), so the old
+                // implementation caused warnings in user code.
+                Secret * /* lhs (NULL) */,
+                T *rhs)
+            {
+                // We already know that 'lhs' is a null pointer.
+                return CmpHelperEQ(lhs_expression, rhs_expression,
+                                   static_cast<T *>(NULL), rhs);
+            }
+        };
+
+// Separate the error generating code from the code path to reduce the stack
+// frame size of CmpHelperOP. This helps reduce the overhead of some sanitizers
+// when calling EXPECT_OP in a tight loop.
+        template <typename T1, typename T2>
+        AssertionResult CmpHelperOpFailure(const char *expr1, const char *expr2,
+                                           const T1 &val1, const T2 &val2,
+                                           const char *op)
+        {
+            return AssertionFailure()
+                   << "Expected: (" << expr1 << ") " << op << " (" << expr2
+                   << "), actual: " << FormatForComparisonFailureMessage(val1, val2)
+                   << " vs " << FormatForComparisonFailureMessage(val2, val1);
+        }
+
+// A macro for implementing the helper functions needed to implement
+// ASSERT_?? and EXPECT_??.  It is here just to avoid copy-and-paste
+// of similar code.
+//
+// For each templatized helper function, we also define an overloaded
+// version for BiggestInt in order to reduce code bloat and allow
+// anonymous enums to be used with {ASSERT|EXPECT}_?? when compiled
+// with gcc 4.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+
+#define GTEST_IMPL_CMP_HELPER_(op_name, op)\
+template <typename T1, typename T2>\
+AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \
+                                   const T1& val1, const T2& val2) {\
+  if (val1 op val2) {\
+    return AssertionSuccess();\
+  } else {\
+    return CmpHelperOpFailure(expr1, expr2, val1, val2, #op);\
+  }\
+}\
+GTEST_API_ AssertionResult CmpHelper##op_name(\
+    const char* expr1, const char* expr2, BiggestInt val1, BiggestInt val2)
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+
+// Implements the helper function for {ASSERT|EXPECT}_NE
+        GTEST_IMPL_CMP_HELPER_(NE, !=);
+// Implements the helper function for {ASSERT|EXPECT}_LE
+        GTEST_IMPL_CMP_HELPER_(LE, <=);
+// Implements the helper function for {ASSERT|EXPECT}_LT
+        GTEST_IMPL_CMP_HELPER_(LT, <);
+// Implements the helper function for {ASSERT|EXPECT}_GE
+        GTEST_IMPL_CMP_HELPER_(GE, >=);
+// Implements the helper function for {ASSERT|EXPECT}_GT
+        GTEST_IMPL_CMP_HELPER_(GT, >);
+
+#undef GTEST_IMPL_CMP_HELPER_
+
+// The helper function for {ASSERT|EXPECT}_STREQ.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTREQ(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const char *s1,
+                                                  const char *s2);
+
+// The helper function for {ASSERT|EXPECT}_STRCASEEQ.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRCASEEQ(const char *s1_expression,
+                                                      const char *s2_expression,
+                                                      const char *s1,
+                                                      const char *s2);
+
+// The helper function for {ASSERT|EXPECT}_STRNE.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRNE(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const char *s1,
+                                                  const char *s2);
+
+// The helper function for {ASSERT|EXPECT}_STRCASENE.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRCASENE(const char *s1_expression,
+                                                      const char *s2_expression,
+                                                      const char *s1,
+                                                      const char *s2);
+
+
+// Helper function for *_STREQ on wide strings.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTREQ(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const wchar_t *s1,
+                                                  const wchar_t *s2);
+
+// Helper function for *_STRNE on wide strings.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRNE(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const wchar_t *s1,
+                                                  const wchar_t *s2);
+
+    }  // namespace internal
+
+// IsSubstring() and IsNotSubstring() are intended to be used as the
+// first argument to {EXPECT,ASSERT}_PRED_FORMAT2(), not by
+// themselves.  They check whether needle is a substring of haystack
+// (NULL is considered a substring of itself only), and return an
+// appropriate error message when they fail.
+//
+// The {needle,haystack}_expr arguments are the stringified
+// expressions that generated the two real arguments.
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const char *needle, const char *haystack);
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const wchar_t *needle, const wchar_t *haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const char *needle, const char *haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const wchar_t *needle, const wchar_t *haystack);
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::string &needle, const ::std::string &haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::string &needle, const ::std::string &haystack);
+
+#if GTEST_HAS_STD_WSTRING
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::wstring &needle, const ::std::wstring &haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::wstring &needle, const ::std::wstring &haystack);
+#endif  // GTEST_HAS_STD_WSTRING
+
+    namespace internal
+    {
+
+// Helper template function for comparing floating-points.
+//
+// Template parameter:
+//
+//   RawType: the raw floating-point type (either float or double)
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        template <typename RawType>
+        AssertionResult CmpHelperFloatingPointEQ(const char *lhs_expression,
+                                                 const char *rhs_expression,
+                                                 RawType lhs_value,
+                                                 RawType rhs_value)
+        {
+            const FloatingPoint<RawType> lhs(lhs_value), rhs(rhs_value);
+
+            if (lhs.AlmostEquals(rhs)) {
+                return AssertionSuccess();
+            }
+
+            ::std::stringstream lhs_ss;
+            lhs_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+                   << lhs_value;
+
+            ::std::stringstream rhs_ss;
+            rhs_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+                   << rhs_value;
+
+            return EqFailure(lhs_expression,
+                             rhs_expression,
+                             StringStreamToString(&lhs_ss),
+                             StringStreamToString(&rhs_ss),
+                             false);
+        }
+
+// Helper function for implementing ASSERT_NEAR.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult DoubleNearPredFormat(const char *expr1,
+                                                        const char *expr2,
+                                                        const char *abs_error_expr,
+                                                        double val1,
+                                                        double val2,
+                                                        double abs_error);
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+// A class that enables one to stream messages to assertion macros
+        class GTEST_API_ AssertHelper
+        {
+        public:
+            // Constructor.
+            AssertHelper(TestPartResult::Type type,
+                         const char *file,
+                         int line,
+                         const char *message);
+            ~AssertHelper();
+
+            // Message assignment is a semantic trick to enable assertion
+            // streaming; see the GTEST_MESSAGE_ macro below.
+            void operator=(const Message &message) const;
+
+        private:
+            // We put our data in a struct so that the size of the AssertHelper class can
+            // be as small as possible.  This is important because gcc is incapable of
+            // re-using stack space even for temporary variables, so every EXPECT_EQ
+            // reserves stack space for another AssertHelper.
+            struct AssertHelperData {
+                AssertHelperData(TestPartResult::Type t,
+                                 const char *srcfile,
+                                 int line_num,
+                                 const char *msg)
+                    : type(t), file(srcfile), line(line_num), message(msg) { }
+
+                TestPartResult::Type const type;
+                const char *const file;
+                int const line;
+                std::string const message;
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelperData);
+            };
+
+            AssertHelperData *const data_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelper);
+        };
+
+    }  // namespace internal
+
+#if GTEST_HAS_PARAM_TEST
+// The pure interface class that all value-parameterized tests inherit from.
+// A value-parameterized class must inherit from both ::testing::Test and
+// ::testing::WithParamInterface. In most cases that just means inheriting
+// from ::testing::TestWithParam, but more complicated test hierarchies
+// may need to inherit from Test and WithParamInterface at different levels.
+//
+// This interface has support for accessing the test parameter value via
+// the GetParam() method.
+//
+// Use it with one of the parameter generator defining functions, like Range(),
+// Values(), ValuesIn(), Bool(), and Combine().
+//
+// class FooTest : public ::testing::TestWithParam<int> {
+//  protected:
+//   FooTest() {
+//     // Can use GetParam() here.
+//   }
+//   virtual ~FooTest() {
+//     // Can use GetParam() here.
+//   }
+//   virtual void SetUp() {
+//     // Can use GetParam() here.
+//   }
+//   virtual void TearDown {
+//     // Can use GetParam() here.
+//   }
+// };
+// TEST_P(FooTest, DoesBar) {
+//   // Can use GetParam() method here.
+//   Foo foo;
+//   ASSERT_TRUE(foo.DoesBar(GetParam()));
+// }
+// INSTANTIATE_TEST_CASE_P(OneToTenRange, FooTest, ::testing::Range(1, 10));
+
+    template <typename T>
+    class WithParamInterface
+    {
+    public:
+        typedef T ParamType;
+        virtual ~WithParamInterface() {}
+
+        // The current parameter value. Is also available in the test fixture's
+        // constructor. This member function is non-static, even though it only
+        // references static data, to reduce the opportunity for incorrect uses
+        // like writing 'WithParamInterface<bool>::GetParam()' for a test that
+        // uses a fixture whose parameter type is int.
+        const ParamType &GetParam() const
+        {
+            GTEST_CHECK_(parameter_ != NULL)
+                    << "GetParam() can only be called inside a value-parameterized test "
+                    << "-- did you intend to write TEST_P instead of TEST_F?";
+            return *parameter_;
+        }
+
+    private:
+        // Sets parameter value. The caller is responsible for making sure the value
+        // remains alive and unchanged throughout the current test.
+        static void SetParam(const ParamType *parameter)
+        {
+            parameter_ = parameter;
+        }
+
+        // Static value used for accessing parameter during a test lifetime.
+        static const ParamType *parameter_;
+
+        // TestClass must be a subclass of WithParamInterface<T> and Test.
+        template <class TestClass> friend class internal::ParameterizedTestFactory;
+    };
+
+    template <typename T>
+    const T *WithParamInterface<T>::parameter_ = NULL;
+
+// Most value-parameterized classes can ignore the existence of
+// WithParamInterface, and can just inherit from ::testing::TestWithParam.
+
+    template <typename T>
+    class TestWithParam : public Test, public WithParamInterface<T>
+    {
+    };
+
+#endif  // GTEST_HAS_PARAM_TEST
+
+// Macros for indicating success/failure in test code.
+
+// Skips test in runtime.
+// Skipping test aborts current function.
+// Skipped tests are neither successful nor failed.
+#define GTEST_SKIP() GTEST_SKIP_("")
+
+// ADD_FAILURE unconditionally adds a failure to the current test.
+// SUCCEED generates a success - it doesn't automatically make the
+// current test successful, as a test is only successful when it has
+// no failure.
+//
+// EXPECT_* verifies that a certain condition is satisfied.  If not,
+// it behaves like ADD_FAILURE.  In particular:
+//
+//   EXPECT_TRUE  verifies that a Boolean condition is true.
+//   EXPECT_FALSE verifies that a Boolean condition is false.
+//
+// FAIL and ASSERT_* are similar to ADD_FAILURE and EXPECT_*, except
+// that they will also abort the current function on failure.  People
+// usually want the fail-fast behavior of FAIL and ASSERT_*, but those
+// writing data-driven tests often find themselves using ADD_FAILURE
+// and EXPECT_* more.
+
+// Generates a nonfatal failure with a generic message.
+#define ADD_FAILURE() GTEST_NONFATAL_FAILURE_("Failed")
+
+// Generates a nonfatal failure at the given source file location with
+// a generic message.
+#define ADD_FAILURE_AT(file, line) \
+  GTEST_MESSAGE_AT_(file, line, "Failed", \
+                    ::testing::TestPartResult::kNonFatalFailure)
+
+// Generates a fatal failure with a generic message.
+#define GTEST_FAIL() GTEST_FATAL_FAILURE_("Failed")
+
+// Define this macro to 1 to omit the definition of FAIL(), which is a
+// generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_FAIL
+# define FAIL() GTEST_FAIL()
+#endif
+
+// Generates a success with a generic message.
+#define GTEST_SUCCEED() GTEST_SUCCESS_("Succeeded")
+
+// Define this macro to 1 to omit the definition of SUCCEED(), which
+// is a generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_SUCCEED
+# define SUCCEED() GTEST_SUCCEED()
+#endif
+
+// Macros for testing exceptions.
+//
+//    * {ASSERT|EXPECT}_THROW(statement, expected_exception):
+//         Tests that the statement throws the expected exception.
+//    * {ASSERT|EXPECT}_NO_THROW(statement):
+//         Tests that the statement doesn't throw any exception.
+//    * {ASSERT|EXPECT}_ANY_THROW(statement):
+//         Tests that the statement throws an exception.
+
+#define EXPECT_THROW(statement, expected_exception) \
+  GTEST_TEST_THROW_(statement, expected_exception, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_NO_THROW(statement) \
+  GTEST_TEST_NO_THROW_(statement, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_ANY_THROW(statement) \
+  GTEST_TEST_ANY_THROW_(statement, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_THROW(statement, expected_exception) \
+  GTEST_TEST_THROW_(statement, expected_exception, GTEST_FATAL_FAILURE_)
+#define ASSERT_NO_THROW(statement) \
+  GTEST_TEST_NO_THROW_(statement, GTEST_FATAL_FAILURE_)
+#define ASSERT_ANY_THROW(statement) \
+  GTEST_TEST_ANY_THROW_(statement, GTEST_FATAL_FAILURE_)
+
+// Boolean assertions. Condition can be either a Boolean expression or an
+// AssertionResult. For more information on how to use AssertionResult with
+// these macros see comments on that class.
+#define EXPECT_TRUE(condition) \
+  GTEST_TEST_BOOLEAN_((condition), #condition, false, true, \
+                      GTEST_NONFATAL_FAILURE_)
+#define EXPECT_FALSE(condition) \
+  GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \
+                      GTEST_NONFATAL_FAILURE_)
+#define ASSERT_TRUE(condition) \
+  GTEST_TEST_BOOLEAN_((condition), #condition, false, true, \
+                      GTEST_FATAL_FAILURE_)
+#define ASSERT_FALSE(condition) \
+  GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \
+                      GTEST_FATAL_FAILURE_)
+
+// Includes the auto-generated header that implements a family of
+// generic predicate assertion macros.
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// This file is AUTOMATICALLY GENERATED on 10/31/2011 by command
+// 'gen_gtest_pred_impl.py 5'.  DO NOT EDIT BY HAND!
+//
+// Implements a family of generic predicate assertion macros.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+
+// Makes sure this header is not included before gtest.h.
+#ifndef GTEST_INCLUDE_GTEST_GTEST_H_
+# error Do not include gtest_pred_impl.h directly.  Include gtest.h instead.
+#endif  // GTEST_INCLUDE_GTEST_GTEST_H_
+
+// This header implements a family of generic predicate assertion
+// macros:
+//
+//   ASSERT_PRED_FORMAT1(pred_format, v1)
+//   ASSERT_PRED_FORMAT2(pred_format, v1, v2)
+//   ...
+//
+// where pred_format is a function or functor that takes n (in the
+// case of ASSERT_PRED_FORMATn) values and their source expression
+// text, and returns a testing::AssertionResult.  See the definition
+// of ASSERT_EQ in gtest.h for an example.
+//
+// If you don't care about formatting, you can use the more
+// restrictive version:
+//
+//   ASSERT_PRED1(pred, v1)
+//   ASSERT_PRED2(pred, v1, v2)
+//   ...
+//
+// where pred is an n-ary function or functor that returns bool,
+// and the values v1, v2, ..., must support the << operator for
+// streaming to std::ostream.
+//
+// We also define the EXPECT_* variations.
+//
+// For now we only support predicates whose arity is at most 5.
+// Please email googletestframework@googlegroups.com if you need
+// support for higher arities.
+
+// GTEST_ASSERT_ is the basic statement to which all of the assertions
+// in this file reduce.  Don't use this in your code.
+
+#define GTEST_ASSERT_(expression, on_failure) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (const ::testing::AssertionResult gtest_ar = (expression)) \
+    ; \
+  else \
+    on_failure(gtest_ar.failure_message())
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED1.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1>
+    AssertionResult AssertPred1Helper(const char *pred_text,
+                                      const char *e1,
+                                      Pred pred,
+                                      const T1 &v1)
+    {
+        if (pred(v1)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT1.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT1_(pred_format, v1, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, v1), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED1.  Don't use
+// this in your code.
+#define GTEST_PRED1_(pred, v1, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred1Helper(#pred, \
+                                             #v1, \
+                                             pred, \
+                                             v1), on_failure)
+
+// Unary predicate assertion macros.
+#define EXPECT_PRED_FORMAT1(pred_format, v1) \
+  GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED1(pred, v1) \
+  GTEST_PRED1_(pred, v1, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT1(pred_format, v1) \
+  GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED1(pred, v1) \
+  GTEST_PRED1_(pred, v1, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED2.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2>
+    AssertionResult AssertPred2Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2)
+    {
+        if (pred(v1, v2)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT2.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT2_(pred_format, v1, v2, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, v1, v2), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED2.  Don't use
+// this in your code.
+#define GTEST_PRED2_(pred, v1, v2, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred2Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             pred, \
+                                             v1, \
+                                             v2), on_failure)
+
+// Binary predicate assertion macros.
+#define EXPECT_PRED_FORMAT2(pred_format, v1, v2) \
+  GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED2(pred, v1, v2) \
+  GTEST_PRED2_(pred, v1, v2, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT2(pred_format, v1, v2) \
+  GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED2(pred, v1, v2) \
+  GTEST_PRED2_(pred, v1, v2, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED3.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2,
+              typename T3>
+    AssertionResult AssertPred3Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      const char *e3,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2,
+                                      const T3 &v3)
+    {
+        if (pred(v1, v2, v3)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ", "
+               << e3 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2
+               << "\n" << e3 << " evaluates to " << v3;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT3.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, #v3, v1, v2, v3), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED3.  Don't use
+// this in your code.
+#define GTEST_PRED3_(pred, v1, v2, v3, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred3Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             #v3, \
+                                             pred, \
+                                             v1, \
+                                             v2, \
+                                             v3), on_failure)
+
+// Ternary predicate assertion macros.
+#define EXPECT_PRED_FORMAT3(pred_format, v1, v2, v3) \
+  GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED3(pred, v1, v2, v3) \
+  GTEST_PRED3_(pred, v1, v2, v3, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT3(pred_format, v1, v2, v3) \
+  GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED3(pred, v1, v2, v3) \
+  GTEST_PRED3_(pred, v1, v2, v3, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED4.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2,
+              typename T3,
+              typename T4>
+    AssertionResult AssertPred4Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      const char *e3,
+                                      const char *e4,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2,
+                                      const T3 &v3,
+                                      const T4 &v4)
+    {
+        if (pred(v1, v2, v3, v4)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ", "
+               << e3 << ", "
+               << e4 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2
+               << "\n" << e3 << " evaluates to " << v3
+               << "\n" << e4 << " evaluates to " << v4;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT4.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, v1, v2, v3, v4), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED4.  Don't use
+// this in your code.
+#define GTEST_PRED4_(pred, v1, v2, v3, v4, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred4Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             #v3, \
+                                             #v4, \
+                                             pred, \
+                                             v1, \
+                                             v2, \
+                                             v3, \
+                                             v4), on_failure)
+
+// 4-ary predicate assertion macros.
+#define EXPECT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \
+  GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED4(pred, v1, v2, v3, v4) \
+  GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \
+  GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED4(pred, v1, v2, v3, v4) \
+  GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED5.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2,
+              typename T3,
+              typename T4,
+              typename T5>
+    AssertionResult AssertPred5Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      const char *e3,
+                                      const char *e4,
+                                      const char *e5,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2,
+                                      const T3 &v3,
+                                      const T4 &v4,
+                                      const T5 &v5)
+    {
+        if (pred(v1, v2, v3, v4, v5)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ", "
+               << e3 << ", "
+               << e4 << ", "
+               << e5 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2
+               << "\n" << e3 << " evaluates to " << v3
+               << "\n" << e4 << " evaluates to " << v4
+               << "\n" << e5 << " evaluates to " << v5;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT5.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, #v5, v1, v2, v3, v4, v5), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED5.  Don't use
+// this in your code.
+#define GTEST_PRED5_(pred, v1, v2, v3, v4, v5, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred5Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             #v3, \
+                                             #v4, \
+                                             #v5, \
+                                             pred, \
+                                             v1, \
+                                             v2, \
+                                             v3, \
+                                             v4, \
+                                             v5), on_failure)
+
+// 5-ary predicate assertion macros.
+#define EXPECT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \
+  GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED5(pred, v1, v2, v3, v4, v5) \
+  GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \
+  GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED5(pred, v1, v2, v3, v4, v5) \
+  GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_)
+
+
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+
+// Macros for testing equalities and inequalities.
+//
+//    * {ASSERT|EXPECT}_EQ(v1, v2): Tests that v1 == v2
+//    * {ASSERT|EXPECT}_NE(v1, v2): Tests that v1 != v2
+//    * {ASSERT|EXPECT}_LT(v1, v2): Tests that v1 < v2
+//    * {ASSERT|EXPECT}_LE(v1, v2): Tests that v1 <= v2
+//    * {ASSERT|EXPECT}_GT(v1, v2): Tests that v1 > v2
+//    * {ASSERT|EXPECT}_GE(v1, v2): Tests that v1 >= v2
+//
+// When they are not, Google Test prints both the tested expressions and
+// their actual values.  The values must be compatible built-in types,
+// or you will get a compiler error.  By "compatible" we mean that the
+// values can be compared by the respective operator.
+//
+// Note:
+//
+//   1. It is possible to make a user-defined type work with
+//   {ASSERT|EXPECT}_??(), but that requires overloading the
+//   comparison operators and is thus discouraged by the Google C++
+//   Usage Guide.  Therefore, you are advised to use the
+//   {ASSERT|EXPECT}_TRUE() macro to assert that two objects are
+//   equal.
+//
+//   2. The {ASSERT|EXPECT}_??() macros do pointer comparisons on
+//   pointers (in particular, C strings).  Therefore, if you use it
+//   with two C strings, you are testing how their locations in memory
+//   are related, not how their content is related.  To compare two C
+//   strings by content, use {ASSERT|EXPECT}_STR*().
+//
+//   3. {ASSERT|EXPECT}_EQ(v1, v2) is preferred to
+//   {ASSERT|EXPECT}_TRUE(v1 == v2), as the former tells you
+//   what the actual value is when it fails, and similarly for the
+//   other comparisons.
+//
+//   4. Do not depend on the order in which {ASSERT|EXPECT}_??()
+//   evaluate their arguments, which is undefined.
+//
+//   5. These macros evaluate their arguments exactly once.
+//
+// Examples:
+//
+//   EXPECT_NE(5, Foo());
+//   EXPECT_EQ(NULL, a_pointer);
+//   ASSERT_LT(i, array_size);
+//   ASSERT_GT(records.size(), 0) << "There is no record left.";
+
+#define EXPECT_EQ(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal:: \
+                      EqHelper<GTEST_IS_NULL_LITERAL_(val1)>::Compare, \
+                      val1, val2)
+#define EXPECT_NE(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2)
+#define EXPECT_LE(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2)
+#define EXPECT_LT(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2)
+#define EXPECT_GE(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2)
+#define EXPECT_GT(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2)
+
+#define GTEST_ASSERT_EQ(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal:: \
+                      EqHelper<GTEST_IS_NULL_LITERAL_(val1)>::Compare, \
+                      val1, val2)
+#define GTEST_ASSERT_NE(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2)
+#define GTEST_ASSERT_LE(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2)
+#define GTEST_ASSERT_LT(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2)
+#define GTEST_ASSERT_GE(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2)
+#define GTEST_ASSERT_GT(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2)
+
+// Define macro GTEST_DONT_DEFINE_ASSERT_XY to 1 to omit the definition of
+// ASSERT_XY(), which clashes with some users' own code.
+
+#if !GTEST_DONT_DEFINE_ASSERT_EQ
+# define ASSERT_EQ(val1, val2) GTEST_ASSERT_EQ(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_NE
+# define ASSERT_NE(val1, val2) GTEST_ASSERT_NE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_LE
+# define ASSERT_LE(val1, val2) GTEST_ASSERT_LE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_LT
+# define ASSERT_LT(val1, val2) GTEST_ASSERT_LT(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_GE
+# define ASSERT_GE(val1, val2) GTEST_ASSERT_GE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_GT
+# define ASSERT_GT(val1, val2) GTEST_ASSERT_GT(val1, val2)
+#endif
+
+// C-string Comparisons.  All tests treat NULL and any non-NULL string
+// as different.  Two NULLs are equal.
+//
+//    * {ASSERT|EXPECT}_STREQ(s1, s2):     Tests that s1 == s2
+//    * {ASSERT|EXPECT}_STRNE(s1, s2):     Tests that s1 != s2
+//    * {ASSERT|EXPECT}_STRCASEEQ(s1, s2): Tests that s1 == s2, ignoring case
+//    * {ASSERT|EXPECT}_STRCASENE(s1, s2): Tests that s1 != s2, ignoring case
+//
+// For wide or narrow string objects, you can use the
+// {ASSERT|EXPECT}_??() macros.
+//
+// Don't depend on the order in which the arguments are evaluated,
+// which is undefined.
+//
+// These macros evaluate their arguments exactly once.
+
+#define EXPECT_STREQ(s1, s2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, s1, s2)
+#define EXPECT_STRNE(s1, s2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2)
+#define EXPECT_STRCASEEQ(s1, s2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, s1, s2)
+#define EXPECT_STRCASENE(s1, s2)\
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2)
+
+#define ASSERT_STREQ(s1, s2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, s1, s2)
+#define ASSERT_STRNE(s1, s2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2)
+#define ASSERT_STRCASEEQ(s1, s2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, s1, s2)
+#define ASSERT_STRCASENE(s1, s2)\
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2)
+
+// Macros for comparing floating-point numbers.
+//
+//    * {ASSERT|EXPECT}_FLOAT_EQ(val1, val2):
+//         Tests that two float values are almost equal.
+//    * {ASSERT|EXPECT}_DOUBLE_EQ(val1, val2):
+//         Tests that two double values are almost equal.
+//    * {ASSERT|EXPECT}_NEAR(v1, v2, abs_error):
+//         Tests that v1 and v2 are within the given distance to each other.
+//
+// Google Test uses ULP-based comparison to automatically pick a default
+// error bound that is appropriate for the operands.  See the
+// FloatingPoint template class in gtest-internal.h if you are
+// interested in the implementation details.
+
+#define EXPECT_FLOAT_EQ(val1, val2)\
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<float>, \
+                      val1, val2)
+
+#define EXPECT_DOUBLE_EQ(val1, val2)\
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<double>, \
+                      val1, val2)
+
+#define ASSERT_FLOAT_EQ(val1, val2)\
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<float>, \
+                      val1, val2)
+
+#define ASSERT_DOUBLE_EQ(val1, val2)\
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<double>, \
+                      val1, val2)
+
+#define EXPECT_NEAR(val1, val2, abs_error)\
+  EXPECT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \
+                      val1, val2, abs_error)
+
+#define ASSERT_NEAR(val1, val2, abs_error)\
+  ASSERT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \
+                      val1, val2, abs_error)
+
+// These predicate format functions work on floating-point values, and
+// can be used in {ASSERT|EXPECT}_PRED_FORMAT2*(), e.g.
+//
+//   EXPECT_PRED_FORMAT2(testing::DoubleLE, Foo(), 5.0);
+
+// Asserts that val1 is less than, or almost equal to, val2.  Fails
+// otherwise.  In particular, it fails if either val1 or val2 is NaN.
+    GTEST_API_ AssertionResult FloatLE(const char *expr1, const char *expr2,
+                                       float val1, float val2);
+    GTEST_API_ AssertionResult DoubleLE(const char *expr1, const char *expr2,
+                                        double val1, double val2);
+
+
+#if GTEST_OS_WINDOWS
+
+// Macros that test for HRESULT failure and success, these are only useful
+// on Windows, and rely on Windows SDK macros and APIs to compile.
+//
+//    * {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}(expr)
+//
+// When expr unexpectedly fails or succeeds, Google Test prints the
+// expected result and the actual result with both a human-readable
+// string representation of the error, if available, as well as the
+// hex result code.
+# define EXPECT_HRESULT_SUCCEEDED(expr) \
+    EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr))
+
+# define ASSERT_HRESULT_SUCCEEDED(expr) \
+    ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr))
+
+# define EXPECT_HRESULT_FAILED(expr) \
+    EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr))
+
+# define ASSERT_HRESULT_FAILED(expr) \
+    ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr))
+
+#endif  // GTEST_OS_WINDOWS
+
+// Macros that execute statement and check that it doesn't generate new fatal
+// failures in the current thread.
+//
+//   * {ASSERT|EXPECT}_NO_FATAL_FAILURE(statement);
+//
+// Examples:
+//
+//   EXPECT_NO_FATAL_FAILURE(Process());
+//   ASSERT_NO_FATAL_FAILURE(Process()) << "Process() failed";
+//
+#define ASSERT_NO_FATAL_FAILURE(statement) \
+    GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_FATAL_FAILURE_)
+#define EXPECT_NO_FATAL_FAILURE(statement) \
+    GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_NONFATAL_FAILURE_)
+
+// Causes a trace (including the source file path, the current line
+// number, and the given message) to be included in every test failure
+// message generated by code in the current scope.  The effect is
+// undone when the control leaves the current scope.
+//
+// The message argument can be anything streamable to std::ostream.
+//
+// In the implementation, we include the current line number as part
+// of the dummy variable name, thus allowing multiple SCOPED_TRACE()s
+// to appear in the same block - as long as they are on different
+// lines.
+#define SCOPED_TRACE(message) \
+  ::testing::internal::ScopedTrace GTEST_CONCAT_TOKEN_(gtest_trace_, __LINE__)(\
+    __FILE__, __LINE__, ::testing::Message() << (message))
+
+// Compile-time assertion for type equality.
+// StaticAssertTypeEq<type1, type2>() compiles iff type1 and type2 are
+// the same type.  The value it returns is not interesting.
+//
+// Instead of making StaticAssertTypeEq a class template, we make it a
+// function template that invokes a helper class template.  This
+// prevents a user from misusing StaticAssertTypeEq<T1, T2> by
+// defining objects of that type.
+//
+// CAVEAT:
+//
+// When used inside a method of a class template,
+// StaticAssertTypeEq<T1, T2>() is effective ONLY IF the method is
+// instantiated.  For example, given:
+//
+//   template <typename T> class Foo {
+//    public:
+//     void Bar() { testing::StaticAssertTypeEq<int, T>(); }
+//   };
+//
+// the code:
+//
+//   void Test1() { Foo<bool> foo; }
+//
+// will NOT generate a compiler error, as Foo<bool>::Bar() is never
+// actually instantiated.  Instead, you need:
+//
+//   void Test2() { Foo<bool> foo; foo.Bar(); }
+//
+// to cause a compiler error.
+    template <typename T1, typename T2>
+    bool StaticAssertTypeEq()
+    {
+        (void)internal::StaticAssertTypeEqHelper<T1, T2>();
+        return true;
+    }
+
+// Defines a test.
+//
+// The first parameter is the name of the test case, and the second
+// parameter is the name of the test within the test case.
+//
+// The convention is to end the test case name with "Test".  For
+// example, a test case for the Foo class can be named FooTest.
+//
+// Test code should appear between braces after an invocation of
+// this macro.  Example:
+//
+//   TEST(FooTest, InitializesCorrectly) {
+//     Foo foo;
+//     EXPECT_TRUE(foo.StatusIsOK());
+//   }
+
+// Note that we call GetTestTypeId() instead of GetTypeId<
+// ::testing::Test>() here to get the type ID of testing::Test.  This
+// is to work around a suspected linker bug when using Google Test as
+// a framework on Mac OS X.  The bug causes GetTypeId<
+// ::testing::Test>() to return different values depending on whether
+// the call is from the Google Test framework itself or from user test
+// code.  GetTestTypeId() is guaranteed to always return the same
+// value, as it always calls GetTypeId<>() from the Google Test
+// framework.
+#define GTEST_TEST(test_case_name, test_name)\
+  GTEST_TEST_(test_case_name, test_name, \
+              ::testing::Test, ::testing::internal::GetTestTypeId())
+
+// Define this macro to 1 to omit the definition of TEST(), which
+// is a generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_TEST
+# define TEST(test_case_name, test_name) GTEST_TEST(test_case_name, test_name)
+#endif
+
+// Defines a test that uses a test fixture.
+//
+// The first parameter is the name of the test fixture class, which
+// also doubles as the test case name.  The second parameter is the
+// name of the test within the test case.
+//
+// A test fixture class must be declared earlier.  The user should put
+// the test code between braces after using this macro.  Example:
+//
+//   class FooTest : public testing::Test {
+//    protected:
+//     virtual void SetUp() { b_.AddElement(3); }
+//
+//     Foo a_;
+//     Foo b_;
+//   };
+//
+//   TEST_F(FooTest, InitializesCorrectly) {
+//     EXPECT_TRUE(a_.StatusIsOK());
+//   }
+//
+//   TEST_F(FooTest, ReturnsElementCountCorrectly) {
+//     EXPECT_EQ(0, a_.size());
+//     EXPECT_EQ(1, b_.size());
+//   }
+
+#define TEST_F(test_fixture, test_name)\
+  GTEST_TEST_(test_fixture, test_name, test_fixture, \
+              ::testing::internal::GetTypeId<test_fixture>())
+
+// Returns a path to temporary directory.
+// Tries to determine an appropriate directory for the platform.
+    GTEST_API_ std::string TempDir();
+
+}  // namespace testing
+
+// Use this function in main() to run all tests.  It returns 0 if all
+// tests are successful, or 1 otherwise.
+//
+// RUN_ALL_TESTS() should be invoked after the command line has been
+// parsed by InitGoogleTest().
+//
+// This function was formerly a macro; thus, it is in the global
+// namespace and has an all-caps name.
+int RUN_ALL_TESTS() GTEST_MUST_USE_RESULT_;
+
+inline int RUN_ALL_TESTS()
+{
+    return ::testing::UnitTest::GetInstance()->Run();
+}
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_H_
diff --git a/memkind-1.10.0/test/hbw_allocator_performance_tests.cpp b/memkind-1.10.0/test/hbw_allocator_performance_tests.cpp
new file mode 100644
index 0000000..4fd05b9
--- /dev/null
+++ b/memkind-1.10.0/test/hbw_allocator_performance_tests.cpp
@@ -0,0 +1,421 @@
+/*
+* Copyright (C) 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/TaskFactory.hpp"
+#include "allocator_perf_tool/Stats.hpp"
+#include "allocator_perf_tool/Thread.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+
+class HbwAllocatorPerformanceTest: public :: testing::Test
+{
+private:
+    AllocatorFactory allocator_factory;
+
+protected:
+    void SetUp()
+    {
+        allocator_factory.initialize_allocator(AllocatorTypes::STANDARD_ALLOCATOR);
+    }
+
+    void TearDown()
+    {}
+
+    float run(unsigned kind, unsigned call, size_t threads_number,
+              size_t alloc_size, unsigned mem_operations_num)
+    {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+        TypesConf func_calls;
+        TypesConf allocator_types;
+
+        func_calls.enable_type(FunctionCalls::FREE);
+        func_calls.enable_type(call);
+        allocator_types.enable_type(kind);
+
+        TaskConf conf = {
+            mem_operations_num, //number of memory operations
+            {
+                mem_operations_num, //number of memory operations
+                alloc_size, //min. size of single allocation
+                alloc_size //max. size of single allocatioion
+            },
+            func_calls, //enable function calls
+            allocator_types, //enable allocators
+            11, //random seed
+            false, //do not log memory operations and statistics to csv file
+        };
+
+        for (int i=0; i<threads_number; i++) {
+            Task *task = task_factory.create(conf);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats time_stats;
+        for (int i=0; i<tasks.size(); i++) {
+            time_stats += tasks[i]->get_results();
+        }
+
+        return time_stats.stats[kind][call].total_time;
+    }
+
+    void run_test(unsigned kind, unsigned call, size_t threads_number,
+                  size_t alloc_size, unsigned mem_operations_num)
+    {
+        allocator_factory.initialize_allocator(kind);
+        float ref_time = run(AllocatorTypes::STANDARD_ALLOCATOR, call, threads_number,
+                             alloc_size, mem_operations_num);
+        float perf_time = run(kind, call, threads_number, alloc_size,
+                              mem_operations_num);
+        float ref_delta_time_percent = allocator_factory.calc_ref_delta(ref_time,
+                                                                        perf_time);
+
+        GTestAdapter::RecordProperty("total_time_spend_on_alloc", perf_time);
+        GTestAdapter::RecordProperty("alloc_operations_per_thread", mem_operations_num);
+        GTestAdapter::RecordProperty("ref_delta_time_percent", ref_delta_time_percent);
+    }
+
+};
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
diff --git a/memkind-1.10.0/test/hbw_allocator_tests.cpp b/memkind-1.10.0/test/hbw_allocator_tests.cpp
new file mode 100644
index 0000000..751f905
--- /dev/null
+++ b/memkind-1.10.0/test/hbw_allocator_tests.cpp
@@ -0,0 +1,135 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbw_allocator.h>
+
+#include "common.h"
+#include <vector>
+
+// Tests for hbw::allocator class.
+class HbwAllocatorTests: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+};
+
+//Test standard memory allocation and deallocation.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_DetaultAllocatorTest)
+{
+    const size_t size = 512;
+    hbw::allocator<size_t> allocator;
+
+    hbw::allocator<size_t>::pointer ptr = allocator.allocate(size);
+
+    ASSERT_TRUE(NULL != ptr);
+
+    //Do the actually memory writing
+    for (size_t i=0; i<size; i++) {
+        ptr[i] = i;
+    }
+
+    allocator.deallocate(ptr, size);
+}
+
+//Test address conversion functionality.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_AddressConversion)
+{
+    const size_t size = 512;
+    hbw::allocator<int> allocator;
+
+    int *ptr = allocator.allocate(size);
+
+    ASSERT_TRUE(NULL != ptr);
+
+    const int excpected_val = 4;
+    ptr[0] = excpected_val;
+
+    hbw::allocator<int>::reference reference = *ptr;
+    hbw::allocator<int>::const_reference const_reference = *ptr;
+
+    hbw::allocator<int>::pointer test_ptr = allocator.address(reference);
+    hbw::allocator<int>::const_pointer test_const_ptr = allocator.address(
+                                                            const_reference);
+
+    ASSERT_TRUE(NULL != test_ptr);
+    ASSERT_TRUE(NULL != test_const_ptr);
+
+    EXPECT_EQ(excpected_val, test_ptr[0]);
+    EXPECT_EQ(excpected_val, test_const_ptr[0]);
+
+    allocator.deallocate(ptr, size);
+}
+
+//Test for boundaries of allocation sizes.
+//We expect to catch allocation exceptions caused by out of bound sizes.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_AllocationSizeOutOfBounds)
+{
+    hbw::allocator<size_t> allocator;
+
+    const size_t over_size = allocator.max_size() + 1;
+
+    ASSERT_THROW(allocator.allocate(over_size), std::bad_alloc);
+
+    const size_t max_size = -1; //This will give maximum value of size_t.
+
+    ASSERT_THROW(allocator.allocate(max_size), std::bad_alloc);
+
+    ASSERT_THROW(allocator.allocate(0), std::bad_alloc);
+}
+
+//Test if variable will be constructed.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_AllocatorConstruct)
+{
+    hbw::allocator<int> allocator;
+
+    int x = 0;
+    int expect_val = 4;
+    allocator.construct(&x, expect_val);
+
+    EXPECT_EQ(expect_val, x);
+}
+
+//Test the integration with std::vector.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_StandardVector)
+{
+    std::vector<int, hbw::allocator<int> > vec;
+    const size_t size = 10000;
+
+    for (size_t i=0; i<size; i++) {
+        vec.push_back(i);
+    }
+
+    EXPECT_EQ(size, vec.size());
+    EXPECT_EQ(1, vec[1]);
+
+    vec.clear();
+
+    EXPECT_EQ((size_t)0, vec.size());
+}
diff --git a/memkind-1.10.0/test/hbw_detection_test.py b/memkind-1.10.0/test/hbw_detection_test.py
new file mode 100755
index 0000000..5fd34a5
--- /dev/null
+++ b/memkind-1.10.0/test/hbw_detection_test.py
@@ -0,0 +1,60 @@
+#
+#  Copyright (C) 2017 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+from distutils.spawn import find_executable
+from python_framework import CMD_helper
+import os
+
+class Test_hbw_detection(object):
+    os.environ["PATH"] += os.pathsep + os.path.dirname(os.path.dirname(__file__))
+    binary_path = find_executable("memkind-hbw-nodes")
+    environ_err_test = "../environ_err_hbw_malloc_test"
+    expected_libnuma_warning = "libnuma: Warning: node argument -1 is out of range\n\n"
+    fail_msg = "Test failed with:\n {0}"
+    cmd_helper = CMD_helper()
+
+    def get_hbw_nodes(self, nodemask=None):
+        """ This function executes memkind function 'get_mbind_nodemask' and returns its output - comma-separated HBW nodes """
+        command = self.binary_path
+        if (nodemask):
+            command = "MEMKIND_HBW_NODES={}".format(nodemask) + command
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: Execution of \'{0}\' returns {1}, \noutput: {2}".format(command, retcode, output))
+        print("\nExecution of {} returns output {}".format(command, output))
+        return output
+
+    def test_TC_MEMKIND_hbw_detection_compare_nodemask_default_and_env_variable(self):
+        """ This test checks whether hbw_nodemask_default and hbw_nodemask_env_variable has the same value """
+        hbw_nodemask_default = self.get_hbw_nodes()
+        hbw_nodemask_env_variable = self.get_hbw_nodes(hbw_nodemask_default)
+        assert hbw_nodemask_default == hbw_nodemask_env_variable, self.fail_msg.format("Error: Nodemask hbw_nodemask_default ({0}) " \
+               "is not the same as nodemask hbw_nodemask_env_variable ({1})".format(hbw_nodemask_default, hbw_nodemask_env_variable))
+
+    def test_TC_MEMKIND_hbw_detection_negative_hbw_malloc(self):
+        """ This test sets usupported value of MEMKIND_HBW_NODES, then try to perform a successful allocation from DRAM using hbw_malloc()
+        thanks to default HBW_POLICY_PREFERRED policy """
+        command = "MEMKIND_HBW_NODES=-1 " + self.cmd_helper.get_command_path(self.environ_err_test)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode != 0, self.fail_msg.format("\nError: Execution of: \'{0}\' returns: {1} \noutput: {2}".format(command, retcode, output))
+        assert self.expected_libnuma_warning == output, self.fail_msg.format("Error: expected libnuma warning ({0}) " \
+               "was not found (output: {1})").format(self.expected_libnuma_warning, output)
diff --git a/memkind-1.10.0/test/hbw_verify_function_test.cpp b/memkind-1.10.0/test/hbw_verify_function_test.cpp
new file mode 100644
index 0000000..9bdd36c
--- /dev/null
+++ b/memkind-1.10.0/test/hbw_verify_function_test.cpp
@@ -0,0 +1,396 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "memkind.h"
+#include "memkind/internal/memkind_hbw.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+#include "hbwmalloc.h"
+#include "common.h"
+
+#include <sys/mman.h>
+#include <numaif.h>
+#include <numa.h>
+#include <stdint.h>
+
+#define SHIFT_BYTES(ptr, bytes) ((char*)(ptr)+(bytes))
+
+/*
+ * Set of tests for hbw_verify_memory_region() function,
+ * which intend to check if allocated memory fully fall into high bandwidth memory.
+ * Note: in this tests we are using internal function memkind_hbw_all_get_mbind_nodemask().
+ * In future we intend to rewrite this function when we will have replacement in new API.
+ */
+class HbwVerifyFunctionTest: public :: testing::Test
+{
+protected:
+    const size_t BLOCK_SIZE = 64;
+    const size_t page_size = sysconf(_SC_PAGESIZE);
+    const int flags = MAP_ANONYMOUS | MAP_PRIVATE;
+};
+
+/*
+ * Group of basic tests for different sizes of allocations
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_page_size_not_round)
+{
+    size_t size = page_size * 1024 + 5;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_page_size_round)
+{
+    size_t size = page_size * 1024;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_iterate_1_byte_to_8194_bytes)
+{
+    for (size_t size = 1; size <= (page_size * 2 + 2);
+         size++) { //iterate through 2 pages and 2 bytes
+        char *ptr = (char *) hbw_malloc(size);
+        ASSERT_FALSE(ptr == NULL);
+        EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+        hbw_free(ptr);
+    }
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_ext_5GB)
+{
+    size_t size = 5ull*(1<<30); //5GB - big allocation
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+/*
+ * Test setting memory without HBW_TOUCH_PAGES flag
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_setting_memory_without_flag)
+{
+    size_t size = page_size;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    memset(ptr, '.', size);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, 0), 0);
+    for (size_t i = 0; i < size; i++) { //check that content is unchanged
+        EXPECT_TRUE(ptr[i] == '.');
+    }
+    hbw_free(ptr);
+}
+
+/*
+ * Test HBW_TOUCH_PAGES flag
+ */
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_HBW_TOUCH_PAGES_check_overwritten_content)
+{
+    size_t size = 5 * page_size;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    memset(ptr, '.', size);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    for (size_t i = 0; i < size;
+         i++) { //check that content of all pages doesn't change
+        EXPECT_TRUE(ptr[i] == '.');
+    }
+    hbw_free(ptr);
+}
+
+/*
+ * Tests check if number of 64-page blocks are working correctly
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_many_blocks_round)
+{
+    size_t size = 16 * (BLOCK_SIZE * page_size); //exactly 16 * 64-page blocks
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_many_blocks_not_round)
+{
+    size_t size = (16 * BLOCK_SIZE * page_size) + (8 *
+                                                   page_size); //16 * 64-page blocks and 1 * 8-paged block
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+/*
+ * Check if 2 blocks and parts of the third block are working correctly.
+ * With this test we can be sure that different sizes of allocation (not only whole blocks) are ok.
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_iterate_not_round)
+{
+    for (size_t i = 1; i < BLOCK_SIZE; i++) {
+        size_t size = (2 * BLOCK_SIZE * page_size) + (i *
+                                                      page_size); // 2 * 64-paged blocks and iterate through whole 3rd 64-paged block
+        char *ptr = (char *) hbw_malloc(size);
+        ASSERT_FALSE(ptr == NULL);
+        EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+        hbw_free(ptr);
+    }
+}
+
+/*
+ * Tests for other kinds and malloc
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_2MBPages_HBW_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    size_t size = 2 * 1024 * 1024 * 10; //10 * 2MB pages
+    char *ptr = (char *) memkind_malloc(MEMKIND_HBW_HUGETLB, size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    memkind_free(MEMKIND_HBW_HUGETLB, ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_DEFAULT)
+{
+    size_t size = page_size * 1024;
+    char *ptr = (char *) memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), -1);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_malloc)
+{
+    size_t size = page_size * 1024;
+    char *ptr = (char *) malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), -1);
+    free(ptr);
+}
+
+/*
+ * Group of negative tests
+ */
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_Negative_size_0_and_SET_MEMORY_flag)
+{
+    void *ptr = hbw_malloc(page_size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, 0, HBW_TOUCH_PAGES), EINVAL);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Negative_size_0_without_flag)
+{
+    void *ptr = hbw_malloc(page_size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, 0, 0), EINVAL);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_Negative_Uninitialized_Memory_without_flag)
+{
+    void *ptr = NULL;
+    EXPECT_EQ(hbw_verify_memory_region(ptr, page_size * 1024, 0), EINVAL);
+}
+
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_Negative_Uninitialized_Memory_and_SET_MEMORY_flag)
+{
+    void *ptr = NULL;
+    EXPECT_EQ(hbw_verify_memory_region(ptr, page_size * 1024, HBW_TOUCH_PAGES),
+              EINVAL);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Negative_Without_Memset)
+{
+    size_t size = page_size * 1024;
+    void *ptr = hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, 0), -1);
+    hbw_free(ptr);
+}
+
+/*
+ * Corner cases: tests for half of pages
+ * + HBM memory
+ * # HBM memory and verified
+ * - not HBM memory, but allocated
+ * = not HBM memory and verified
+ */
+
+/* 3 pages:
+ * |++##|####|##++|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //all pages should fall on HBM
+    mbind(ptr, size, MPOL_BIND, hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //verified are: half of the first page, second page and half of the third page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), 0);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 3 pages:
+ * |####|####|----|
+ * |++##|####|==--|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages_1_and_2_page)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //first and second page should fall on HBM
+    mbind(ptr, size - page_size, MPOL_BIND, hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //check if mbind is successful |####|####|----|
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size - page_size, HBW_TOUCH_PAGES), 0);
+
+    //verified are: half of the first page, second page and half of the third page |++##|####|==--|
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), -1);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 3 pages:
+ * |----|####|####|
+ * |--==|####|##++|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages_2_and_3_page)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //second and third page should fall on HBM
+    mbind(SHIFT_BYTES(ptr, page_size), size - page_size, MPOL_BIND,
+          hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //check if mbind is successful |----|####|####|
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size),
+                                       size - page_size, HBW_TOUCH_PAGES), 0);
+
+    //verified are: half of the first page, second page and half of the third page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), -1);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 3 pages:
+ * |####|----|++++| and  |++++|----|####|
+ * |++##|====|##++|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages_1_and_3_page)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //first and third page should fall on HBM
+    mbind(ptr, page_size, MPOL_BIND, hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+    mbind(SHIFT_BYTES(ptr, 2 * page_size), page_size, MPOL_BIND, hbw_nodemask.maskp,
+          NUMA_NUM_NODES, 0);
+
+    //check if mbind is successful |####|----|++++| and  |++++|----|####|
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size - 2*page_size, HBW_TOUCH_PAGES),
+              0);
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, 2*page_size),
+                                       size - 2*page_size, HBW_TOUCH_PAGES), 0);
+
+    //verified are: half of the first page, second page and half of the third page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), -1);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 5 pages:
+ * |----|+###|####|###+|----|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Boundaries_CornerCase)
+{
+    size_t size = 5 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    size_t tested_pages_size = size - 2 * page_size;
+
+    //second, third and fourth page should fall on HBM
+    mbind(SHIFT_BYTES(ptr, page_size), tested_pages_size, MPOL_BIND,
+          hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //verified are: second byte of the second page to last byte -1 of fourth page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size + 1),
+                                       tested_pages_size - 2, HBW_TOUCH_PAGES), 0);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 5 pages:
+ * |-###|####|####|####|###-|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_partial_verification)
+{
+    size_t size = 5 * page_size;
+
+    //all pages should fall on HBM
+    void *ptr = hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+
+    //but only second byte of the first page to last byte -1 of the last page are touched
+    memset(SHIFT_BYTES(ptr, 1), 0, size - 2);
+
+    //verified are: second byte of the first page to last byte -1 of last page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, 1), size - 2, 0), 0);
+    hbw_free(ptr);
+}
diff --git a/memkind-1.10.0/test/heap_manager_init_perf_test.cpp b/memkind-1.10.0/test/heap_manager_init_perf_test.cpp
new file mode 100644
index 0000000..a48aece
--- /dev/null
+++ b/memkind-1.10.0/test/heap_manager_init_perf_test.cpp
@@ -0,0 +1,167 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include <sstream>
+#include <vector>
+
+#include "common.h"
+#include "allocator_perf_tool/Configuration.hpp"
+#include "allocator_perf_tool/AllocatorFactory.hpp"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+//Test heap managers initialization performance.
+class HeapManagerInitPerfTest: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {
+        //Calculate reference statistics.
+        ref_time = allocator_factory.initialize_allocator(
+                       AllocatorTypes::STANDARD_ALLOCATOR).total_time;
+    }
+
+    void TearDown()
+    {}
+
+    void run_test(unsigned allocator_type)
+    {
+        AllocatorFactory::initialization_stat stat =
+            allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_DEFAULT);
+
+        post_test(stat);
+    }
+
+    void post_test(AllocatorFactory::initialization_stat &stat)
+    {
+        //Calculate (%) distance to the reference time for function calls.
+        stat.ref_delta_time = allocator_factory.calc_ref_delta(ref_time,
+                                                               stat.total_time);
+
+        std::stringstream elapsed_time;
+        elapsed_time << stat.total_time;
+
+        std::stringstream ref_delta_time;
+        ref_delta_time << std::fixed << stat.ref_delta_time << std::endl;
+
+        RecordProperty("elapsed_time", elapsed_time.str());
+        RecordProperty("ref_delta_time_percent_rate", ref_delta_time.str());
+
+        for (int i=0; i<stat.memory_overhead.size(); i++) {
+            std::stringstream node;
+            node << "memory_overhad_node_" << i;
+            std::stringstream memory_overhead;
+            memory_overhead << stat.memory_overhead[i];
+            RecordProperty(node.str(), memory_overhead.str());
+        }
+    }
+
+    AllocatorFactory allocator_factory;
+    float ref_time;
+};
+
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_DEFAULT)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_DEFAULT);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_INTERLEAVE)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_INTERLEAVE);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW_INTERLEAVE)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_INTERLEAVE);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW_PREFERRED)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_PREFERRED);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HUGETLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_GBTLB)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_GBTLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_HUGETLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest,
+       test_TC_MEMKIND_perf_libinit_HBW_PREFERRED_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(
+            AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_ext_libinit_HBW_GBTLB)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_GBTLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest,
+       test_TC_MEMKIND_perf_libinit_HBW_PREFERRED_GBTLB)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(
+            AllocatorTypes::MEMKIND_HBW_PREFERRED_GBTLB);
+    post_test(stat);
+}
+
diff --git a/memkind-1.10.0/test/huge_page_test.cpp b/memkind-1.10.0/test/huge_page_test.cpp
new file mode 100644
index 0000000..f6555de
--- /dev/null
+++ b/memkind-1.10.0/test/huge_page_test.cpp
@@ -0,0 +1,90 @@
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/HugePageUnmap.hpp"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+#include "TimerSysTime.hpp"
+#include "Thread.hpp"
+
+/*
+* This test was created because of the munmap() fail in jemalloc.
+* There are two root causes of the error:
+* - kernel bug (munmap() fails when the size is not aligned)
+* - heap Manager doesn’t provide size aligned to 2MB pages for munmap()
+* Test allocates 2000MB using Huge Pages (50threads*10operations*4MBalloc_size),
+* but it needs extra Huge Pages due to overhead caused by heap management.
+*/
+class  HugePageTest: public :: testing::Test
+{
+protected:
+    void run()
+    {
+        unsigned mem_operations_num = 10;
+        size_t threads_number = 50;
+        bool touch_memory = true;
+        size_t size_1MB = 1024*1024;
+        size_t alignment = 2*size_1MB;
+        size_t alloc_size = 4*size_1MB;
+
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+
+        TimerSysTime timer;
+        timer.start();
+
+        // This bug occurs more frequently under stress of multithreaded allocations.
+        for (int i=0; i<threads_number; i++) {
+            Task *task = new HugePageUnmap(mem_operations_num, touch_memory, alignment,
+                                           alloc_size, HBW_PAGESIZE_2MB);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+        }
+
+        float elapsed_time = timer.getElapsedTime();
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+        threads_manager.release();
+
+        //task release
+        for (int i=0; i<tasks.size(); i++) {
+            delete tasks[i];
+        }
+
+        RecordProperty("threads_number", threads_number);
+        RecordProperty("memory_operations_per_thread", mem_operations_num);
+        RecordProperty("elapsed_time", elapsed_time);
+    }
+};
+
+
+
+// Test passes when there is no crash.
+TEST_F(HugePageTest, test_TC_MEMKIND_ext_UNMAP_HUGE_PAGE)
+{
+    HugePageOrganizer huge_page_organizer(1024);
+    run();
+}
diff --git a/memkind-1.10.0/test/load_tbbmalloc_symbols.c b/memkind-1.10.0/test/load_tbbmalloc_symbols.c
new file mode 100644
index 0000000..fd186a4
--- /dev/null
+++ b/memkind-1.10.0/test/load_tbbmalloc_symbols.c
@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "tbbmalloc.h"
+
+int load_tbbmalloc_symbols()
+{
+    const char so_name[]="libtbbmalloc.so.2";
+    void *tbb_handle = dlopen(so_name, RTLD_LAZY);
+    if(!tbb_handle) {
+        printf("Cannot load %s\n", so_name);
+        return -1;
+    }
+
+    scalable_malloc = dlsym(tbb_handle, "scalable_malloc");
+    if(!scalable_malloc) {
+        printf("Cannot load scalable_malloc symbol from %s\n", so_name);
+        return -1;
+    }
+
+    scalable_realloc = dlsym(tbb_handle, "scalable_realloc");
+    if(!scalable_realloc) {
+        printf("Cannot load scalable_realloc symbol from %s\n", so_name);
+        return -1;
+    }
+
+    scalable_calloc = dlsym(tbb_handle, "scalable_calloc");
+    if(!scalable_calloc) {
+        printf("Cannot load scalable_calloc symbol from %s\n", so_name);
+        return -1;
+    }
+
+    scalable_free = dlsym(tbb_handle, "scalable_free");
+    if(!scalable_free) {
+        printf("Cannot load scalable_free symbol from %s\n", so_name);
+        return -1;
+    }
+
+    return 0;
+}
diff --git a/memkind-1.10.0/test/locality_test.cpp b/memkind-1.10.0/test/locality_test.cpp
new file mode 100644
index 0000000..65fe49c
--- /dev/null
+++ b/memkind-1.10.0/test/locality_test.cpp
@@ -0,0 +1,155 @@
+/*
+* Copyright (C) 2017 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <stdio.h>
+#include <numa.h>
+#include <hbwmalloc.h>
+#include <memkind.h>
+#include <vector>
+#include <memory>
+#include <gtest/gtest.h>
+#include "allocator_perf_tool/Allocation_info.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+
+typedef std::unique_ptr<void, void(*)(void *)> hbw_mem_ptr;
+
+
+class HBWPreferredLocalityTest: public ::testing::Test
+{
+private:
+    int find_closest_node(int node, const std::vector<int> &nodes)
+    {
+        int min_distance = 0;
+        int closest_node = -1;
+        for (int i = 0; i < nodes.size(); i++) {
+            int distance = numa_distance(node, nodes[i]);
+            if (distance && (distance < min_distance || min_distance == 0)) {
+                min_distance = distance;
+                closest_node = nodes[i];
+            }
+        }
+        return closest_node;
+    }
+
+    bool pin_to_cpu(int cpu_id)
+    {
+        cpu_set_t cpu_set;
+        CPU_ZERO(&cpu_set);
+        CPU_SET(cpu_id, &cpu_set);
+        return sched_setaffinity(0, sizeof(cpu_set_t), &cpu_set) != -1;
+    }
+
+    void check_ptr_numa(void *ptr, int expected_numa_id, int cpu_id, size_t size)
+    {
+        memset(ptr, 1, size);
+
+        int numa_id = get_numa_node_id(ptr);
+        EXPECT_EQ(numa_id, expected_numa_id);
+
+        char property_name[50];
+        snprintf(property_name, 50, "actual_numa_for_cpu_%d_expected_numa_%d", cpu_id,
+                 expected_numa_id);
+        GTestAdapter::RecordProperty(property_name, numa_id);
+    }
+
+public:
+    void pin_memory_in_requesting_mem_thread(size_t size,
+                                             const std::vector<int> &cpu_ids,
+                                             const std::vector<int> &mcdram_nodes)
+    {
+#ifdef _OPENMP
+        int threads_num = cpu_ids.size();
+        int ret = hbw_set_policy(HBW_POLICY_PREFERRED);
+        ASSERT_EQ(ret, 0);
+
+        #pragma omp parallel for num_threads(threads_num)
+        for (int i = 0; i < threads_num; i++) {
+            if (!pin_to_cpu(cpu_ids[i])) {
+                ADD_FAILURE();
+                continue;
+            }
+
+            void *internal_ptr = hbw_malloc(size);
+            if (!internal_ptr) {
+                ADD_FAILURE();
+                continue;
+            }
+            hbw_mem_ptr ptr(internal_ptr, hbw_free);
+            int expected_numa_id = find_closest_node(numa_node_of_cpu(cpu_ids[i]),
+                                                     mcdram_nodes);
+            check_ptr_numa(ptr.get(), expected_numa_id, cpu_ids[i], size);
+        }
+#else
+        std::cout << "[ SKIPPED ] Feature OPENMP not supported" << std::endl;
+#endif
+    }
+
+    void pin_memory_in_other_thread_than_requesting_mem(size_t size,
+                                                        const std::vector<int> &cpu_ids,
+                                                        const std::vector<int> &mcdram_nodes)
+    {
+#ifdef _OPENMP
+        int threads_num = cpu_ids.size();
+        int ret = hbw_set_policy(HBW_POLICY_PREFERRED);
+        ASSERT_EQ(ret, 0);
+
+        int main_thread_cpu_id = 0;
+        int expected_numa_id = find_closest_node(main_thread_cpu_id, mcdram_nodes);
+        ASSERT_TRUE(pin_to_cpu(main_thread_cpu_id));
+
+        std::vector<hbw_mem_ptr> ptrs;
+        for (int i = 0; i < threads_num; i++) {
+            void *internal_ptr = hbw_malloc(size);
+            ASSERT_TRUE(internal_ptr);
+            ptrs.emplace_back(internal_ptr, hbw_free);
+        }
+
+        #pragma omp parallel for num_threads(threads_num)
+        for (int i = 0; i < threads_num; i++) {
+            if (!pin_to_cpu(cpu_ids[i])) {
+                ADD_FAILURE();
+                continue;
+            }
+            check_ptr_numa(ptrs[i].get(), expected_numa_id, cpu_ids[i], size);
+        }
+#else
+        std::cout << "[ SKIPPED ] Feature OPENMP not supported" << std::endl;
+#endif
+    }
+};
+
+TEST_F(HBWPreferredLocalityTest,
+       test_TC_MEMKIND_KNL_SNC4_pin_memory_in_requesting_mem_thread_4_threads_100_bytes)
+{
+    pin_memory_in_requesting_mem_thread(100u, std::vector<int> {0, 18, 36, 54},
+                                        std::vector<int> {4, 5, 6, 7});
+}
+
+TEST_F(HBWPreferredLocalityTest,
+       test_TC_MEMKIND_KNL_SNC4_pin_memory_in_other_thread_than_requesting_mem_4_threads_100_bytes)
+{
+    pin_memory_in_other_thread_than_requesting_mem(100u, std::vector<int> {0, 18, 36, 54},
+                                                   std::vector<int> {4, 5, 6, 7});
+}
+
diff --git a/memkind-1.10.0/test/main.cpp b/memkind-1.10.0/test/main.cpp
new file mode 100644
index 0000000..e1cdd05
--- /dev/null
+++ b/memkind-1.10.0/test/main.cpp
@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "common.h"
+
+char *PMEM_DIR = const_cast<char *>("/tmp/");
+
+int main(int argc, char **argv)
+{
+    int opt = 0;
+    struct stat st;
+
+    testing::InitGoogleTest(&argc, argv);
+
+    if (testing::GetGtestHelpFlag()) {
+        printf("\nMemkind options:\n"
+               "-d <directory_path>   change directory on which PMEM kinds\n"
+               "                      are created (default /tmp/)\n");
+        return 0;
+    }
+
+    while ((opt = getopt(argc, argv, "d:")) != -1) {
+        switch (opt) {
+            case 'd':
+                PMEM_DIR = optarg;
+
+                if (stat(PMEM_DIR, &st) != 0 || !S_ISDIR(st.st_mode)) {
+                    printf("%s : Error in getting path status or"
+                           " invalid or non-existent directory\n", PMEM_DIR);
+                    return -1;
+                }
+
+                break;
+            default:
+                return -1;
+        }
+    }
+
+    return RUN_ALL_TESTS();
+}
diff --git a/memkind-1.10.0/test/memkind-afts-ext.ts b/memkind-1.10.0/test/memkind-afts-ext.ts
new file mode 100644
index 0000000..ea8ed6d
--- /dev/null
+++ b/memkind-1.10.0/test/memkind-afts-ext.ts
@@ -0,0 +1,3 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=-PerformanceTest.*
+/usr/share/mpss/test/memkind-dt/decorator_test -a
+/usr/share/mpss/test/memkind-dt/gb_page_tests_bind_policy -a
diff --git a/memkind-1.10.0/test/memkind-afts.ts b/memkind-1.10.0/test/memkind-afts.ts
new file mode 100644
index 0000000..738741e
--- /dev/null
+++ b/memkind-1.10.0/test/memkind-afts.ts
@@ -0,0 +1,4 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=-PerformanceTest.*:*.*ext*
+/usr/share/mpss/test/memkind-dt/decorator_test -a
+/usr/share/mpss/test/memkind-dt/gb_page_tests_bind_policy -a --gtest_filter=-*.*ext*
+/usr/share/mpss/test/memkind-dt/gb_page_tests_bind_policy -a --gtest_filter=-*.*ext*
diff --git a/memkind-1.10.0/test/memkind-perf-ext.ts b/memkind-1.10.0/test/memkind-perf-ext.ts
new file mode 100644
index 0000000..a9eaf58
--- /dev/null
+++ b/memkind-1.10.0/test/memkind-perf-ext.ts
@@ -0,0 +1,2 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=PerformanceTest.*
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=HeapManagerInitPerfTest*:AllocPerformanceTest*
diff --git a/memkind-1.10.0/test/memkind-perf.ts b/memkind-1.10.0/test/memkind-perf.ts
new file mode 100644
index 0000000..ab3b530
--- /dev/null
+++ b/memkind-1.10.0/test/memkind-perf.ts
@@ -0,0 +1,2 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=PerformanceTest.*
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=HeapManagerInitPerfTest*:AllocPerformanceTest*:-*.*ext*
diff --git a/memkind-1.10.0/test/memkind-pytests.ts b/memkind-1.10.0/test/memkind-pytests.ts
new file mode 100644
index 0000000..008c4a9
--- /dev/null
+++ b/memkind-1.10.0/test/memkind-pytests.ts
@@ -0,0 +1 @@
+-k "test_TC_MEMKIND_"
diff --git a/memkind-1.10.0/test/memkind-slts.ts b/memkind-1.10.0/test/memkind-slts.ts
new file mode 100644
index 0000000..9139ed2
--- /dev/null
+++ b/memkind-1.10.0/test/memkind-slts.ts
@@ -0,0 +1,2 @@
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=AllocateToMaxStressTests*:-HugePageTest*
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=HugePageTest* --gtest_repeat=10 --gtest_break_on_failure
diff --git a/memkind-1.10.0/test/memkind_allocator_tests.cpp b/memkind-1.10.0/test/memkind_allocator_tests.cpp
new file mode 100644
index 0000000..8b08260
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_allocator_tests.cpp
@@ -0,0 +1,309 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind_allocator.h"
+#include "common.h"
+#include <vector>
+#include <memory>
+#include <array>
+#include <list>
+#include <map>
+#include <utility>
+#include <string>
+#include <algorithm>
+#include <scoped_allocator>
+#include <thread>
+
+// Tests for memkind::allocator class.
+class MemkindAllocatorTests: public ::testing::Test
+{
+public:
+    libmemkind::static_kind::allocator<int> int_alc_regular { libmemkind::kinds::REGULAR };
+    libmemkind::static_kind::allocator<int> int_alc_default { libmemkind::kinds::DEFAULT };
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+
+//Compare two same kind different type allocators
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindDifferentType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<char> alc2 { int_alc_regular };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+//Compare two same kind same type
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindSameType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+////Compare two different kind different type
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindDifferentType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<char> alc2 { int_alc_default };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+////Compare two different kind same type allocators
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindSameType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+////Copy assignment test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_CopyAssignment_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc1 != alc2);
+    alc1 = alc2;
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+////Move constructor test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_MoveConstructor_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc2 == alc1);
+    libmemkind::static_kind::allocator<int> alc3 = std::move(alc1);
+    ASSERT_TRUE(alc2 == alc3);
+}
+
+////Move assignment test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_MoveAssignment_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc1 != alc2);
+
+    {
+        libmemkind::static_kind::allocator<int> alc3 { alc2 };
+        alc1 = std::move(alc3);
+    }
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+////Single allocation-deallocation test
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_Allocator_SingleAllocationDeallocation_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    int *created_object = alc1.allocate(1);
+    alc1.deallocate(created_object, 1);
+}
+
+////Shared cross-allocation-deallocation test
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_Allocator_SharedAllocationDeallocation_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    int *created_object = nullptr;
+    int *created_object_2 = nullptr;
+    created_object = alc1.allocate(1);
+    ASSERT_TRUE(alc1 == alc2);
+    alc2.deallocate(created_object, 1);
+    created_object = alc2.allocate(1);
+    alc1.deallocate(created_object, 1);
+
+    created_object = alc1.allocate(1);
+    created_object_2 = alc2.allocate(1);
+    alc2.deallocate(created_object, 1);
+    alc1.deallocate(created_object_2, 1);
+}
+
+////Construct-destroy test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_ConstructDestroy_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    int *created_object = alc1.allocate(1);
+    alc1.construct(created_object);
+    alc1.destroy(created_object);
+}
+
+////Testing thread-safety support
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_Allocator_MultithreadingSupport_Test)
+{
+    const size_t num_threads = 1000;
+    const size_t iteration_count = 5000;
+
+    std::vector<std::thread> thds;
+    for (size_t i = 0; i < num_threads; ++i) {
+
+        thds.push_back(std::thread( [&]() {
+            std::vector<libmemkind::static_kind::allocator<int>> allocators_local;
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.push_back(libmemkind::static_kind::allocator<int>
+                                           ( int_alc_regular ));
+            }
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.pop_back();
+            }
+        }));
+    }
+
+    for (size_t i = 0; i < num_threads; ++i) {
+        thds[i].join();
+    }
+}
+
+////Test multiple memkind allocator usage
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_MultipleAllocatorUsage_Test)
+{
+    {
+        libmemkind::static_kind::allocator<int> alloc { libmemkind::kinds::HBW } ;
+    }
+
+    {
+        libmemkind::static_kind::allocator<int> alloc { libmemkind::kinds::DAX_KMEM } ;
+    }
+}
+
+////Test vector
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Vector_Test)
+{
+    libmemkind::static_kind::allocator<int> alc{ int_alc_default };
+    std::vector<int, libmemkind::static_kind::allocator<int>> vector{ alc };
+
+    const int num = 20;
+
+    for (int i = 0; i < num; ++i) {
+        vector.push_back(0xDEAD + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        ASSERT_TRUE(vector[i] == 0xDEAD + i);
+    }
+}
+
+////Test list
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_AllocatorUsage_List_Test)
+{
+    libmemkind::static_kind::allocator<int> alc{ int_alc_regular };
+
+    std::list<int, libmemkind::static_kind::allocator<int>> list{ alc };
+
+    const int num = 4;
+
+    for (int i = 0; i < num; ++i) {
+        list.emplace_back(0xBEAC011 + i);
+        ASSERT_TRUE(list.back() == 0xBEAC011 + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        list.pop_back();
+    }
+}
+
+////Test map
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Map_Test)
+{
+    typedef std::basic_string<std::pair<const std::string, std::string>, libmemkind::static_kind::allocator<char>>
+            memkind_string;
+    libmemkind::static_kind::allocator<memkind_string> alc{ libmemkind::kinds::REGULAR };
+
+    std::map<std::string, std::string, std::less<std::string>, libmemkind::static_kind::allocator<std::pair<const std::string, std::string>>>
+    map{ std::less<std::string>(), alc };
+
+    const int num = 10;
+
+    for (int i = 0; i < num; ++i) {
+        map[std::to_string(i)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string(i)] == std::to_string(0x0CEA11 + i));
+        map[std::to_string((i * 997 + 83) % 113)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string((i * 997 + 83) % 113)] == std::to_string(
+                        0x0CEA11 + i));
+    }
+}
+
+#if _GLIBCXX_USE_CXX11_ABI
+//Test vector of strings
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorUsage_VectorOfString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::static_kind::allocator<char>>
+            memkind_string;
+    typedef libmemkind::static_kind::allocator<memkind_string> memkind_alloc;
+
+    memkind_alloc alc{ libmemkind::kinds::DEFAULT };
+    libmemkind::static_kind::allocator<char> st_alc{alc};
+
+    std::vector<memkind_string,std::scoped_allocator_adaptor<memkind_alloc>> vec{ alc };
+    memkind_string arg{ "Very very loooong striiiing", st_alc };
+
+    vec.push_back(arg);
+    ASSERT_TRUE(vec.back() == arg);
+}
+
+////Test map of int strings
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorScopedUsage_MapOfIntString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::static_kind::allocator<char>>
+            memkind_string;
+    typedef int key_t;
+    typedef memkind_string value_t;
+    typedef std::pair<const key_t, value_t> target_pair;
+    typedef libmemkind::static_kind::allocator<target_pair> pmem_alloc;
+    typedef libmemkind::static_kind::allocator<char> str_allocator_t;
+    typedef std::map<key_t, value_t, std::less<key_t>, std::scoped_allocator_adaptor<pmem_alloc>>
+            map_t;
+
+    pmem_alloc map_allocator( libmemkind::kinds::REGULAR );
+
+    str_allocator_t str_allocator( map_allocator );
+
+    value_t source_str1("Lorem ipsum dolor ", str_allocator);
+    value_t source_str2("sit amet consectetuer adipiscing elit", str_allocator );
+
+    map_t target_map{ std::scoped_allocator_adaptor<pmem_alloc>(map_allocator) };
+
+    target_map[key_t(165)] = source_str1;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str1);
+    target_map[key_t(165)] = source_str2;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str2);
+}
+#endif // _GLIBCXX_USE_CXX11_ABI
diff --git a/memkind-1.10.0/test/memkind_dax_kmem_test.cpp b/memkind-1.10.0/test/memkind_dax_kmem_test.cpp
new file mode 100644
index 0000000..381b389
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_dax_kmem_test.cpp
@@ -0,0 +1,447 @@
+/*
+ * Copyright (C) 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <algorithm>
+#include <numaif.h>
+#include <string>
+#include <unistd.h>
+
+#include "common.h"
+#include "dax_kmem_nodes.h"
+#include "proc_stat.h"
+#include "sys/types.h"
+#include "sys/sysinfo.h"
+#include "TestPolicy.hpp"
+
+class MemkindDaxKmemFunctionalTests: public ::testing::Test
+{
+protected:
+    DaxKmem dax_kmem_nodes;
+
+    void SetUp()
+    {
+        if (dax_kmem_nodes.size() < 1) {
+            GTEST_SKIP() << "Minimum 1 PMEM NUMA required." << std::endl;
+        }
+    }
+};
+
+class MemkindDaxKmemTestsParam: public ::testing::Test,
+    public ::testing::WithParamInterface<memkind_t>
+{
+protected:
+    DaxKmem dax_kmem_nodes;
+    memkind_t kind;
+    void SetUp()
+    {
+        kind = GetParam();
+        if (kind == MEMKIND_DAX_KMEM_PREFERRED) {
+            std::set<int> regular_nodes = TestPolicy::get_regular_numa_nodes();
+            for (auto const &node: regular_nodes) {
+                auto closest_dax_kmem_nodes = dax_kmem_nodes.get_closest_numa_nodes(node);
+                if (closest_dax_kmem_nodes.size() > 1)
+                    GTEST_SKIP() << "Skip test for MEMKIND_DAX_KMEM_PREFFERED - "
+                                 "more than one PMEM NUMA node is closest to node: "
+                                 << node << std::endl;
+            }
+        }
+    }
+};
+
+INSTANTIATE_TEST_CASE_P(
+    KindParam, MemkindDaxKmemTestsParam,
+    ::testing::Values(MEMKIND_DAX_KMEM, MEMKIND_DAX_KMEM_ALL,
+                      MEMKIND_DAX_KMEM_PREFERRED));
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_free_with_NULL_kind_4096_bytes)
+{
+    void *ptr = memkind_malloc(kind, 4 * KB);
+    ASSERT_NE(nullptr, ptr);
+    memkind_free(nullptr, ptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_alloc_zero)
+{
+    void *test1 = memkind_malloc(kind, 0);
+    ASSERT_EQ(test1, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_alloc_size_max)
+{
+    errno = 0;
+    void *test1 = memkind_malloc(kind, SIZE_MAX);
+    ASSERT_EQ(test1, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_calloc_zero)
+{
+    void *test = memkind_calloc(kind, 0, 100);
+    ASSERT_EQ(test, nullptr);
+
+    test = memkind_calloc(kind, 100, 0);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_calloc_size_max)
+{
+    const size_t size = SIZE_MAX;
+    const size_t num = 1;
+    errno = 0;
+
+    void *test = memkind_calloc(kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_calloc_num_max)
+{
+    const size_t size = 10;
+    const size_t num = SIZE_MAX;
+    errno = 0;
+
+    void *test = memkind_calloc(kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc)
+{
+    const size_t size1 = 512;
+    const size_t size2 = 1 * KB;
+    char *default_str = nullptr;
+
+    default_str = (char *)memkind_realloc(kind, default_str, size1);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc with size %zu\n", size1);
+    printf("%s", default_str);
+
+    default_str = (char *)memkind_realloc(kind, default_str, size2);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc with size %zu\n", size2);
+    printf("%s", default_str);
+
+    memkind_free(kind, default_str);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_zero)
+{
+    const size_t size = 1 * KB;
+
+    void *test = memkind_malloc(kind, size);
+    ASSERT_NE(test, nullptr);
+
+    void *new_test = memkind_realloc(kind, test, 0);
+    ASSERT_EQ(new_test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_size_max)
+{
+    const size_t size = 1 * KB;
+
+    void *test = memkind_malloc(kind, size);
+    ASSERT_NE(test, nullptr);
+
+    errno = 0;
+    void *new_test = memkind_realloc(kind, test, SIZE_MAX);
+    ASSERT_EQ(new_test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+
+    memkind_free(kind, test);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_size_zero)
+{
+    const size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+    const size_t iteration = 100;
+
+    for (unsigned i = 0; i < iteration; ++i) {
+        test = memkind_malloc(kind, size);
+        ASSERT_NE(test, nullptr);
+        errno = 0;
+        new_test = memkind_realloc(kind, test, 0);
+        ASSERT_EQ(new_test, nullptr);
+        ASSERT_EQ(errno, 0);
+    }
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_nullptr)
+{
+    const size_t size = 1 * KB;
+    void *test = nullptr;
+
+    test = memkind_realloc(kind, test, size);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(kind, test);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_increase_decrease_size)
+{
+    size_t size = 1 * KB;
+    const char val[] = "test_TC_MEMKIND_PmemReallocIncreaseSizeNullKindVariant";
+    int status;
+
+    char *test1 = (char *)memkind_malloc(kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 2 * KB;
+    char *test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    size = 512;
+    char *test3 = (char *)memkind_realloc(nullptr, test2, size);
+    ASSERT_NE(test3, nullptr);
+    status = memcmp(val, test3, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(kind, test3);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_free_nullptr)
+{
+    const int n_iter = 5000;
+
+    for (int i = 0; i < n_iter; ++i) {
+        memkind_free(kind, nullptr);
+    }
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_posix_memalign_alignment_less_than_void)
+{
+    void *test = nullptr;
+    const size_t size = 32;
+    const size_t wrong_alignment = 4;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_posix_memalign_alignment_not_pow2)
+{
+    void *test = nullptr;
+    const size_t size = 32;
+    const size_t wrong_alignment = sizeof(void *)+1;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_posix_memalign_correct_alignment)
+{
+    void *test = nullptr;
+    const size_t size = 32;
+    const size_t alignment = sizeof(void *);
+
+    errno = 0;
+    int ret = memkind_posix_memalign(kind, &test, alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, 0);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(nullptr, test);
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_alloc_until_full_numa)
+{
+    ProcStat stat;
+    const size_t alloc_size = 100 * MB;
+    std::set<void *> allocations;
+    size_t numa_size;
+    int numa_id = -1;
+    const int n_swap_alloc = 20;
+
+    void *ptr = memkind_malloc(MEMKIND_DAX_KMEM, alloc_size);
+    ASSERT_NE(nullptr, ptr);
+    memset(ptr, 'a', alloc_size);
+    allocations.insert(ptr);
+
+    get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+    numa_size = numa_node_size64(numa_id, nullptr);
+
+    while (numa_size > alloc_size * allocations.size()) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    size_t init_swap = stat.get_used_swap_space_size_bytes();
+    for(int i = 0; i < n_swap_alloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    ASSERT_GE(stat.get_used_swap_space_size_bytes(), init_swap);
+
+    for (auto const &ptr: allocations) {
+        memkind_free(MEMKIND_DAX_KMEM, ptr);
+    }
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_ALL_alloc_until_full_numa)
+{
+    if (dax_kmem_nodes.size() < 2)
+        GTEST_SKIP() << "This test requires minimum 2 kmem dax nodes";
+
+    ProcStat stat;
+    void *ptr;
+    const size_t alloc_size = 100 * MB;
+    std::set<void *> allocations;
+    size_t sum_of_dax_kmem_free_space = dax_kmem_nodes.get_free_space();
+    int numa_id = -1;
+    const int n_swap_alloc = 20;
+
+    while (sum_of_dax_kmem_free_space > alloc_size * allocations.size()) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_ALL, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+
+        get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+        ASSERT_TRUE(dax_kmem_nodes.contains(numa_id));
+    }
+
+    size_t init_swap = stat.get_used_swap_space_size_bytes();
+    for(int i = 0; i < n_swap_alloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_ALL, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    ASSERT_GE(stat.get_used_swap_space_size_bytes(), init_swap);
+
+    for (auto const &ptr: allocations) {
+        memkind_free(MEMKIND_DAX_KMEM_ALL, ptr);
+    }
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_PREFFERED_alloc_until_full_numa)
+{
+    std::set<int> regular_nodes = TestPolicy::get_regular_numa_nodes();
+    for (auto const &node: regular_nodes) {
+        auto closest_dax_kmem_nodes = dax_kmem_nodes.get_closest_numa_nodes(node);
+        if (closest_dax_kmem_nodes.size() > 1)
+            GTEST_SKIP() << "Skip test for MEMKIND_DAX_KMEM_PREFFERED - "
+                         "more than one PMEM NUMA node is closest to node: "
+                         << node << std::endl;
+    }
+
+    ProcStat stat;
+    const size_t alloc_size = 100 * MB;
+    std::set<void *> allocations;
+    size_t numa_size;
+    int numa_id = -1;
+    const int n_swap_alloc = 20;
+
+    void *ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+    ASSERT_NE(nullptr, ptr);
+    memset(ptr, 'a', alloc_size);
+    allocations.insert(ptr);
+
+    get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+    int process_cpu = sched_getcpu();
+    int process_node = numa_node_of_cpu(process_cpu);
+    std::set<int> closest_numa_ids = dax_kmem_nodes.get_closest_numa_nodes(
+                                         process_node);
+    numa_size = numa_node_size64(numa_id, nullptr);
+
+    while (0.99 * numa_size > alloc_size * allocations.size()) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+
+        get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+        ASSERT_TRUE(closest_numa_ids.find(numa_id) != closest_numa_ids.end());
+    }
+
+    for (int i = 0; i < n_swap_alloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    ASSERT_EQ(stat.get_used_swap_space_size_bytes(), 0U);
+
+    for (auto const &ptr: allocations) {
+        memkind_free(MEMKIND_DAX_KMEM_ALL, ptr);
+    }
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_PREFFERED_check_prerequisities)
+{
+    bool can_run = false;
+    std::set<int> regular_nodes = TestPolicy::get_regular_numa_nodes();
+
+    for (auto const &node: regular_nodes) {
+        auto closest_dax_kmem_nodes = dax_kmem_nodes.get_closest_numa_nodes(node);
+        if (closest_dax_kmem_nodes.size() > 1)
+            can_run = true;
+    }
+
+    if (!can_run)
+        GTEST_SKIP() <<
+                     "Skip test for MEMKIND_DAX_KMEM_PREFFERED checking prerequisities - "
+                     "none of the nodes have more than one closest PMEM NUMA nodes";
+
+    const size_t alloc_size = 100 * MB;
+    void *ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+    ASSERT_EQ(nullptr, ptr);
+}
diff --git a/memkind-1.10.0/test/memkind_defrag_reallocate.cpp b/memkind-1.10.0/test/memkind_defrag_reallocate.cpp
new file mode 100644
index 0000000..2b685e8
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_defrag_reallocate.cpp
@@ -0,0 +1,140 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "common.h"
+#include "vector"
+#include <stdio.h>
+
+class MemkindDefragReallocateTests: public ::testing::Test
+{
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+TEST_F(MemkindDefragReallocateTests, test_TC_MEMKIND_NULLTransfer)
+{
+    void *ptr = memkind_defrag_reallocate(nullptr, nullptr);
+    ASSERT_EQ(ptr, nullptr);
+    ptr = memkind_defrag_reallocate(MEMKIND_REGULAR, nullptr);
+    ASSERT_EQ(ptr, nullptr);
+}
+
+TEST_F(MemkindDefragReallocateTests, test_TC_MEMKIND_Negative_test)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 512);
+    ASSERT_NE(ptr, nullptr);
+    void *new_ptr = memkind_defrag_reallocate(MEMKIND_DEFAULT, ptr);
+    ASSERT_EQ(new_ptr, nullptr);
+}
+
+TEST_F(MemkindDefragReallocateTests,
+       test_TC_MEMKIND_TCACHE_Default_kind_success)
+{
+    std::vector<void *> alloc_vec;
+    const size_t number_of_malloc = 10000000;
+    alloc_vec.reserve(2*number_of_malloc);
+    size_t count_mem_transfer = 0;
+    void *ptr;
+    size_t i;
+
+    for (i = 0; i < number_of_malloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DEFAULT, 150);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 150);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < number_of_malloc; ++i) {
+        if (i % 2 == 0) {
+            memkind_free(MEMKIND_DEFAULT,alloc_vec.at(i));
+            alloc_vec.at(i) = nullptr;
+        }
+        ptr = memkind_malloc(MEMKIND_DEFAULT, 300);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 300);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < alloc_vec.size(); ++i) {
+        void *new_ptr = memkind_defrag_reallocate(MEMKIND_DEFAULT, alloc_vec.at(i));
+        if (new_ptr) {
+            alloc_vec.at(i) = new_ptr;
+            count_mem_transfer++;
+        }
+    }
+
+    ASSERT_NE(count_mem_transfer, 0U);
+
+    for(auto const &val: alloc_vec) {
+        memkind_free(MEMKIND_DEFAULT, val);
+    }
+}
+
+TEST_F(MemkindDefragReallocateTests,
+       test_TC_MEMKIND_TCACHE_REGULAR_kind_success)
+{
+    std::vector<void *> alloc_vec;
+    const size_t number_of_malloc = 10000000;
+    alloc_vec.reserve(2*number_of_malloc);
+    size_t count_mem_transfer = 0;
+    void *ptr;
+    size_t i;
+
+    for (i = 0; i < number_of_malloc ; ++i) {
+        ptr = memkind_malloc(MEMKIND_REGULAR, 150);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 150);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < number_of_malloc; ++i) {
+        if (i % 2 == 0) {
+            memkind_free(MEMKIND_REGULAR,alloc_vec.at(i));
+            alloc_vec.at(i) = nullptr;
+        }
+        ptr = memkind_malloc(MEMKIND_REGULAR, 300);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 300);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < alloc_vec.size(); ++i) {
+        void *new_ptr = memkind_defrag_reallocate(nullptr, alloc_vec.at(i));
+        if (new_ptr) {
+            alloc_vec.at(i) = new_ptr;
+            count_mem_transfer++;
+        }
+    }
+
+    ASSERT_NE(count_mem_transfer, 0U);
+
+    for(auto const &val: alloc_vec) {
+        memkind_free(MEMKIND_REGULAR, val);
+    }
+}
diff --git a/memkind-1.10.0/test/memkind_detect_kind_tests.cpp b/memkind-1.10.0/test/memkind_detect_kind_tests.cpp
new file mode 100644
index 0000000..e2fdb43
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_detect_kind_tests.cpp
@@ -0,0 +1,145 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "common.h"
+
+extern const char *PMEM_DIR;
+
+class MemkindDetectKindTests: public ::testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindNullPtr)
+{
+    struct memkind *temp_kind = memkind_detect_kind(nullptr);
+    ASSERT_EQ(nullptr, temp_kind);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindPmemKind)
+{
+    memkind_t pmem_kind_temp = nullptr;
+    int err = memkind_create_pmem(PMEM_DIR, 0, &pmem_kind_temp);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_temp);
+
+    void *ptr = memkind_malloc(pmem_kind_temp, 512);
+    ASSERT_NE(nullptr, ptr);
+
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(pmem_kind_temp, temp_kind);
+
+    memkind_free(pmem_kind_temp, ptr);
+
+    err = memkind_destroy_kind(pmem_kind_temp);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindRegularKind)
+{
+    void *ptr = memkind_malloc(MEMKIND_REGULAR, 512);
+    ASSERT_NE(nullptr, ptr);
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(MEMKIND_REGULAR, temp_kind);
+    memkind_free(MEMKIND_REGULAR, ptr);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectDefaultSmallSizeKind)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 512);
+    ASSERT_NE(nullptr, ptr);
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(MEMKIND_DEFAULT, temp_kind);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectDefaultBigSizeKind)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 8 * MB);
+    ASSERT_NE(nullptr, ptr);
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(MEMKIND_DEFAULT, temp_kind);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindMixKind)
+{
+    const size_t alloc_size = 512;
+    memkind_t pmem_kind_temp_1 = nullptr;
+    memkind_t pmem_kind_temp_2 = nullptr;
+    struct memkind *temp_kind = nullptr;
+    void *ptr_pmem_1 = nullptr;
+    void *ptr_pmem_2 = nullptr;
+    void *ptr_regular = nullptr;
+    void *ptr_default = nullptr;
+
+    int err = memkind_create_pmem(PMEM_DIR, 0, &pmem_kind_temp_1);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_temp_1);
+
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_kind_temp_2);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_temp_2);
+
+    ptr_pmem_1 = memkind_malloc(pmem_kind_temp_1, alloc_size);
+    ASSERT_NE(nullptr, ptr_pmem_1);
+
+    ptr_pmem_2 = memkind_malloc(pmem_kind_temp_2, alloc_size);
+    ASSERT_NE(nullptr, ptr_pmem_2);
+
+    ptr_default = memkind_malloc(MEMKIND_DEFAULT, alloc_size);
+    ASSERT_NE(nullptr, ptr_default);
+
+    ptr_regular = memkind_malloc(MEMKIND_REGULAR, alloc_size);
+    ASSERT_NE(nullptr, ptr_regular);
+
+    temp_kind = memkind_detect_kind(ptr_default);
+    ASSERT_EQ(MEMKIND_DEFAULT, temp_kind);
+
+    temp_kind = memkind_detect_kind(ptr_pmem_1);
+    ASSERT_EQ(pmem_kind_temp_1, temp_kind);
+
+    temp_kind = memkind_detect_kind(ptr_regular);
+    ASSERT_EQ(MEMKIND_REGULAR, temp_kind);
+
+    temp_kind = memkind_detect_kind(ptr_pmem_2);
+    ASSERT_EQ(pmem_kind_temp_2, temp_kind);
+
+    memkind_free(pmem_kind_temp_2, ptr_pmem_2);
+    memkind_free(pmem_kind_temp_1, ptr_pmem_1);
+    memkind_free(MEMKIND_DEFAULT, ptr_default);
+    memkind_free(MEMKIND_REGULAR, ptr_regular);
+
+    err = memkind_destroy_kind(pmem_kind_temp_2);
+    ASSERT_EQ(0, err);
+    err = memkind_destroy_kind(pmem_kind_temp_1);
+    ASSERT_EQ(0, err);
+}
diff --git a/memkind-1.10.0/test/memkind_null_kind_test.cpp b/memkind-1.10.0/test/memkind_null_kind_test.cpp
new file mode 100644
index 0000000..7d7d052
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_null_kind_test.cpp
@@ -0,0 +1,229 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include "allocator_perf_tool/TimerSysTime.hpp"
+#include "common.h"
+
+class MemkindNullKindTests: public ::testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultKindFreeNullPtr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_free(MEMKIND_DEFAULT, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_NullKindFreeNullPtr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_free(nullptr, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_NullKindUsableSizeNullPtr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_malloc_usable_size(nullptr, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultRegularKindFreeNullPtr)
+{
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = 1 * MB;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+    }
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultReallocNullptrSizeZero)
+{
+    const double test_time = 5;
+    void *test_nullptr = nullptr;
+    void *test_ptr_malloc = nullptr;
+    void *test_ptr_realloc = nullptr;
+    TimerSysTime timer;
+    timer.start();
+    do {
+        test_ptr_malloc = memkind_malloc(MEMKIND_DEFAULT, 5 * MB);
+        errno = 0;
+        test_ptr_realloc = memkind_realloc(MEMKIND_DEFAULT, test_ptr_malloc, 0);
+        ASSERT_EQ(test_ptr_realloc, nullptr);
+        ASSERT_EQ(errno, 0);
+
+        errno = 0;
+        //equivalent to memkind_malloc(MEMKIND_DEFAULT,0)
+        test_nullptr = memkind_realloc(MEMKIND_DEFAULT, nullptr, 0);
+        ASSERT_EQ(test_nullptr, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests,
+       test_TC_MEMKIND_DefaultKindReallocNullKindSizeZeroImplicitFree)
+{
+    const double test_time = 5;
+    size_t size = 1 * KB;
+    void *test_ptr_malloc = nullptr;
+    void *test_ptr_realloc = nullptr;
+    TimerSysTime timer;
+    timer.start();
+    do {
+        errno = 0;
+        test_ptr_malloc = memkind_malloc(MEMKIND_DEFAULT, size);
+        ASSERT_NE(test_ptr_malloc, nullptr);
+        ASSERT_EQ(errno, 0);
+
+        errno = 0;
+        test_ptr_realloc = memkind_realloc(nullptr, test_ptr_malloc, 0);
+        ASSERT_EQ(test_ptr_realloc, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultKindReallocNullKindSizeMax)
+{
+    size_t size = 1 * KB;
+    void *test_ptr_malloc = nullptr;
+    void *test_ptr_realloc = nullptr;
+
+    test_ptr_malloc = memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_NE(test_ptr_malloc, nullptr);
+
+    errno = 0;
+    test_ptr_realloc = memkind_realloc(nullptr, test_ptr_malloc, SIZE_MAX);
+    ASSERT_EQ(test_ptr_realloc, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+
+    memkind_free(nullptr, test_ptr_malloc);
+}
+
+TEST_F(MemkindNullKindTests,
+       test_TC_MEMKIND_DefaultReallocIncreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_DefaultReallocIncreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size *= 2;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(MEMKIND_DEFAULT, test2);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_ReallocNullptrNullKind)
+{
+    size_t size = 1 * KB;
+
+    errno = 0;
+    void *test = memkind_realloc(nullptr, nullptr, size);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, EINVAL);
+}
+
+TEST_F(MemkindNullKindTests,
+       test_TC_MEMKIND_DefaultReallocDecreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_DefaultReallocDecreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 4;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, size);
+    ASSERT_EQ(status, 0);
+
+    memkind_free(MEMKIND_DEFAULT, test2);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_ReallocNullptrNullKindSizeMax)
+{
+    errno = 0;
+    void *test = memkind_realloc(nullptr, nullptr, SIZE_MAX);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, EINVAL);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_ReallocNullptrNullKindSizeZero)
+{
+    errno = 0;
+    void *test = memkind_realloc(nullptr, nullptr, 0);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, EINVAL);
+}
diff --git a/memkind-1.10.0/test/memkind_pmem_config_tests.cpp b/memkind-1.10.0/test/memkind_pmem_config_tests.cpp
new file mode 100644
index 0000000..4657dfe
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_pmem_config_tests.cpp
@@ -0,0 +1,173 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "include/memkind/internal/memkind_private.h"
+#include "include/memkind/internal/memkind_pmem.h"
+
+#include "common.h"
+
+#include <sys/statfs.h>
+
+extern const char  *PMEM_DIR;
+
+class MemkindConfigTests: public ::testing::Test
+{
+
+protected:
+    struct memkind_config *global_test_cfg;
+    void SetUp()
+    {
+        global_test_cfg = memkind_config_new();
+        ASSERT_NE(nullptr, global_test_cfg);
+    }
+
+    void TearDown()
+    {
+        memkind_config_delete(global_test_cfg);
+    }
+};
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_CreateAndDestroyEmptyConfig)
+{
+    struct memkind_config *tmp_cfg = memkind_config_new();
+    ASSERT_NE(nullptr, tmp_cfg);
+    memkind_config_delete(tmp_cfg);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemConSetConfigSizeZero)
+{
+    memkind_config_set_size(global_test_cfg, 0U);
+    ASSERT_EQ(global_test_cfg->pmem_size, 0U);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemSetConfigSizeNotZero)
+{
+    memkind_config_set_size(global_test_cfg, 100 * MB);
+    ASSERT_EQ(global_test_cfg->pmem_size, 100 * MB);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemSetConfigPath)
+{
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+
+    ASSERT_STREQ(global_test_cfg->pmem_dir, PMEM_DIR);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemSetDefaultMemoryUsagePolicy)
+{
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+
+    ASSERT_EQ(global_test_cfg->policy, MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_SetConservativeMemoryUsagePolicy)
+{
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+
+    ASSERT_EQ(global_test_cfg->policy, MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsFailWrongDirectory)
+{
+    memkind_t pmem_kind = nullptr;
+
+    memkind_config_set_path(global_test_cfg, "/temp/non_exisitng_directory");
+    memkind_config_set_size(global_test_cfg, 0U);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+
+    int err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsFailWrongSize)
+{
+    memkind_t pmem_kind = nullptr;
+
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+    memkind_config_set_size(global_test_cfg, MEMKIND_PMEM_MIN_SIZE -1);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+
+    int err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsFailWrongPolicy)
+{
+    memkind_t pmem_kind = nullptr;
+
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+    memkind_config_set_size(global_test_cfg, MEMKIND_PMEM_MIN_SIZE -1);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_MAX_VALUE);
+
+    int err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsSuccessCheckConservativePolicy)
+{
+    memkind_t pmem_kind = nullptr;
+    int err = 0;
+    double blocksBeforeFree = 0;
+    double blocksAfterFree = 0;
+    struct memkind_pmem *priv = nullptr;
+    void *temp_ptr = nullptr;
+    struct stat st;
+
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+    memkind_config_set_size(global_test_cfg, 0U);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+    err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, 0);
+
+    priv = static_cast<memkind_pmem *>(pmem_kind->priv);
+
+    temp_ptr = memkind_malloc(pmem_kind, 1 * KB);
+
+    err = fstat(priv->fd, &st);
+    ASSERT_EQ(err, 0);
+    blocksBeforeFree = st.st_blocks;
+
+    memkind_free(pmem_kind, temp_ptr);
+
+    err = fstat(priv->fd, &st);
+    ASSERT_EQ(err, 0);
+    blocksAfterFree = st.st_blocks;
+
+    //if blocksAfterFree is less than blocksBeforeFree, extent was called.
+    ASSERT_LT(blocksAfterFree, blocksBeforeFree);
+
+    err = memkind_destroy_kind(pmem_kind);
+    ASSERT_EQ(err, 0);
+}
diff --git a/memkind-1.10.0/test/memkind_pmem_long_time_tests.cpp b/memkind-1.10.0/test/memkind_pmem_long_time_tests.cpp
new file mode 100644
index 0000000..ce7cd49
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_pmem_long_time_tests.cpp
@@ -0,0 +1,120 @@
+/*
+ * Copyright (C) 2018 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "allocator_perf_tool/TimerSysTime.hpp"
+#include "common.h"
+
+#define STRESS_TIME (3*24*60*60)
+
+extern const char  *PMEM_DIR;
+
+static const size_t small_size[] = {8, 16, 32, 48, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384,
+                                    448, 512, 640, 768, 896, 1 * KB, 1280, 1536, 1792, 2 * KB, 2560, 3 * KB,
+                                    3584, 4 * KB, 5 * KB, 6 * KB, 7 * KB, 8 * KB, 10 * KB, 12 * KB, 14 * KB
+                                   };
+
+static const size_t large_size[] = {16 * KB, 32 * KB, 20 * KB, 24 * KB, 28 * KB, 32 * KB, 40 * KB, 48 * KB,
+                                    56 * KB, 64 * KB, 80 * KB, 96 * KB, 112 * KB, 128 * KB, 160 * KB, 192 * KB,
+                                    224 * KB, 256 * KB, 320 * KB, 384 * KB, 448 * KB, 512 * KB, 640 * KB,
+                                    768 * KB, 896 * KB, 1 * MB, 1280 * KB, 1536 * KB, 1792 * KB, 2 * MB,
+                                    2560 * KB, 3 * MB, 3584 * KB, 4 * MB, 5 * MB, 6 * MB, 7 * MB, 8 * MB
+                                   };
+
+class MemkindPmemLongTimeStress: public :: testing::Test
+{
+
+protected:
+    memkind_t pmem_kind;
+    void SetUp()
+    {
+        int err = memkind_create_pmem(PMEM_DIR, 0, &pmem_kind);
+        ASSERT_EQ(0, err);
+        ASSERT_NE(nullptr, pmem_kind);
+    }
+
+    void TearDown()
+    {
+        int err = memkind_destroy_kind(pmem_kind);
+        ASSERT_EQ(0, err);
+    }
+};
+
+TEST_F(MemkindPmemLongTimeStress, DISABLED_test_TC_MEMKIND_PmemStressSmallSize)
+{
+    void *test = nullptr;
+    TimerSysTime timer;
+    timer.start();
+
+    do {
+        for (size_t i = 0; i < ARRAY_SIZE(small_size); i++) {
+            test = memkind_malloc(pmem_kind, small_size[i]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+        }
+    } while (timer.getElapsedTime() < STRESS_TIME);
+}
+
+TEST_F(MemkindPmemLongTimeStress, DISABLED_test_TC_MEMKIND_PmemStressLargeSize)
+{
+    void *test = nullptr;
+    TimerSysTime timer;
+    timer.start();
+
+    do {
+        for (size_t i = 0; i < ARRAY_SIZE(large_size); i++) {
+            test = memkind_malloc(pmem_kind, large_size[i]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+        }
+    } while (timer.getElapsedTime() < STRESS_TIME);
+}
+
+TEST_F(MemkindPmemLongTimeStress,
+       DISABLED_test_TC_MEMKIND_PmemStressSmallAndLargeSize)
+{
+    void *test = nullptr;
+    size_t i = 0, j = 0;
+    TimerSysTime timer;
+    timer.start();
+
+    do {
+        if (i < ARRAY_SIZE(small_size)) {
+            test = memkind_malloc(pmem_kind, small_size[i]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+            i++;
+        } else
+            i = 0;
+
+        if (j < ARRAY_SIZE(large_size)) {
+            test = memkind_malloc(pmem_kind, large_size[j]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+            j++;
+        } else
+            j = 0;
+
+    } while (timer.getElapsedTime() < STRESS_TIME);
+}
diff --git a/memkind-1.10.0/test/memkind_pmem_tests.cpp b/memkind-1.10.0/test/memkind_pmem_tests.cpp
new file mode 100644
index 0000000..f51f5f5
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_pmem_tests.cpp
@@ -0,0 +1,1695 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_pmem.h>
+#include <memkind/internal/memkind_private.h>
+#include "allocator_perf_tool/TimerSysTime.hpp"
+
+#include <sys/param.h>
+#include <sys/mman.h>
+#include <sys/statfs.h>
+#include <stdio.h>
+#include <pthread.h>
+#include <vector>
+#include "common.h"
+
+static const size_t PMEM_PART_SIZE = MEMKIND_PMEM_MIN_SIZE + 4 * KB;
+static const size_t PMEM_NO_LIMIT = 0;
+extern const char  *PMEM_DIR;
+static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
+static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+
+class MemkindPmemTests: public ::testing::Test
+{
+
+protected:
+    memkind_t pmem_kind;
+    void SetUp()
+    {
+        // create PMEM partition
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind);
+        ASSERT_EQ(0, err);
+        ASSERT_NE(pmem_kind, nullptr);
+    }
+
+    void TearDown()
+    {
+        int err = memkind_destroy_kind(pmem_kind);
+        ASSERT_EQ(0, err);
+    }
+};
+
+class MemkindPmemTestsCalloc : public MemkindPmemTests,
+    public ::testing::WithParamInterface<std::tuple<int, int>>
+{
+};
+
+class MemkindPmemTestsMalloc : public ::testing::Test,
+    public ::testing::WithParamInterface<size_t>
+{
+};
+
+static void pmem_get_size(struct memkind *kind, size_t &total, size_t &free)
+{
+    struct memkind_pmem *priv = reinterpret_cast<struct memkind_pmem *>(kind->priv);
+
+    total = priv->max_size;
+    free = priv->max_size - priv->current_size; /* rough estimation */
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPriv)
+{
+    size_t total_mem = 0;
+    size_t free_mem = 0;
+
+    pmem_get_size(pmem_kind, total_mem, free_mem);
+
+    ASSERT_NE(total_mem, 0U);
+    ASSERT_NE(free_mem, 0U);
+
+    ASSERT_EQ(total_mem, roundup(PMEM_PART_SIZE, MEMKIND_PMEM_CHUNK_SIZE));
+
+    size_t offset = total_mem - free_mem;
+    ASSERT_LT(offset, MEMKIND_PMEM_CHUNK_SIZE);
+    ASSERT_LT(offset, total_mem);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCreatePmemFailNonExistingDirectory)
+{
+    const char *non_existing_directory = "/temp/non_exisitng_directory";
+    struct memkind *pmem_temp = nullptr;
+    errno = 0;
+    int err = memkind_create_pmem(non_existing_directory, MEMKIND_PMEM_MIN_SIZE,
+                                  &pmem_temp);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    ASSERT_EQ(nullptr, pmem_temp);
+    ASSERT_EQ(ENOENT, errno);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocFragmentation)
+{
+    const size_t size_array[] = {
+        10,
+        100,
+        200,
+        500,
+        1000,
+        2000,
+        3000,
+        1 * MB,
+        2 * MB,
+        3 * MB,
+        4 * MB
+    };
+
+    const size_t iteration = 1000;
+
+    struct memkind *pmem_temp = nullptr;
+
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_temp);
+
+    //check malloc-free possibility with same sizes and 1000 iterations
+    for ( unsigned j = 0; j < iteration; ++j) {
+        for (unsigned i = 0 ; i < ARRAY_SIZE(size_array); ++i) {
+            void *alloc = memkind_malloc(pmem_temp, size_array[i]);
+            ASSERT_NE(nullptr, alloc);
+            memkind_free(pmem_temp, alloc);
+            alloc = nullptr;
+        }
+    }
+    err = memkind_destroy_kind(pmem_temp);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCreatePmemFailWritePermissionIssue)
+{
+    struct stat path_stat;
+    struct memkind *pmem_temp = nullptr;
+    char temp_dir[] = "/tmp/tmpdir.XXXXXX";
+
+    char *dir_name = mkdtemp(temp_dir);
+    ASSERT_NE(nullptr, dir_name);
+
+    int err = stat(dir_name, &path_stat);
+    ASSERT_EQ(0, err);
+
+    err = chmod(dir_name, path_stat.st_mode & ~S_IWUSR);
+    ASSERT_EQ(0, err);
+
+    errno = 0;
+    err = memkind_create_pmem(dir_name, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    ASSERT_EQ(nullptr, pmem_temp);
+    ASSERT_EQ(EACCES, errno);
+
+    err = rmdir(temp_dir);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMalloc)
+{
+    const size_t size = 1 * KB;
+    char *default_str = nullptr;
+
+    default_str = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_malloc MEMKIND_PMEM\n");
+    printf("%s", default_str);
+
+    memkind_free(pmem_kind, default_str);
+
+    // Out of memory
+    default_str = (char *)memkind_malloc(pmem_kind, 2 * PMEM_PART_SIZE);
+    ASSERT_EQ(nullptr, default_str);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocZero)
+{
+    void *test1 = nullptr;
+
+    test1 = memkind_malloc(pmem_kind, 0);
+    ASSERT_EQ(test1, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocSizeMax)
+{
+    void *test1 = nullptr;
+
+    errno = 0;
+    test1 = memkind_malloc(pmem_kind, SIZE_MAX);
+    ASSERT_EQ(test1, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCallocSmallClassMultipleTimes)
+{
+    const size_t size = 1 * KB;
+    const size_t num = 1;
+    const size_t iteration = 100;
+    char *default_str = nullptr;
+
+    for (size_t i = 0; i < iteration; ++i) {
+        default_str = (char *)memkind_calloc(pmem_kind, num, size);
+        ASSERT_NE(nullptr, default_str);
+        ASSERT_EQ(*default_str, 0);
+        ASSERT_EQ(memcmp(default_str, default_str + 1, size - 1), 0);
+
+        sprintf(default_str, "memkind_calloc MEMKIND_PMEM\n");
+
+        memkind_free(pmem_kind, default_str);
+    }
+}
+
+/*
+ * Test will check if it is not possible to allocate memory
+ * with calloc arguments size or num equal to zero
+ */
+TEST_P(MemkindPmemTestsCalloc, test_TC_MEMKIND_PmemCallocZero)
+{
+    void *test = nullptr;
+    size_t size = std::get<0>(GetParam());
+    size_t num = std::get<1>(GetParam());
+
+    test = memkind_calloc(pmem_kind, size, num);
+    ASSERT_EQ(test, nullptr);
+}
+
+INSTANTIATE_TEST_CASE_P(
+    CallocParam, MemkindPmemTestsCalloc,
+    ::testing::Values(std::make_tuple(10, 0),
+                      std::make_tuple(0, 0),
+                      std::make_tuple(0, 10)));
+
+TEST_P(MemkindPmemTestsCalloc, test_TC_MEMKIND_PmemCallocSizeMax)
+{
+    void *test = nullptr;
+    size_t size = SIZE_MAX;
+    size_t num = 1;
+    errno = 0;
+
+    test = memkind_calloc(pmem_kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindPmemTestsCalloc, test_TC_MEMKIND_PmemCallocNumMax)
+{
+    void *test = nullptr;
+    size_t size = 10;
+    size_t num = SIZE_MAX;
+    errno = 0;
+
+    test = memkind_calloc(pmem_kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCallocHugeClassMultipleTimes)
+{
+    const size_t size = MEMKIND_PMEM_CHUNK_SIZE;
+    const size_t num = 1;
+    const size_t iteration = 100;
+    char *default_str = nullptr;
+
+    for (size_t i = 0; i < iteration; ++i) {
+        default_str = (char *)memkind_calloc(pmem_kind, num, size);
+        ASSERT_NE(nullptr, default_str);
+        ASSERT_EQ(*default_str, 0);
+        ASSERT_EQ(memcmp(default_str, default_str + 1, size - 1), 0);
+
+        sprintf(default_str, "memkind_calloc MEMKIND_PMEM\n");
+
+        memkind_free(pmem_kind, default_str);
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeMemoryAfterDestroyLargeClass)
+{
+    memkind_t pmem_kind_test;
+    struct statfs st;
+    double blocksAvailable;
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_test);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_test);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    blocksAvailable = st.f_bfree;
+
+    while(1) {
+        if (memkind_malloc(pmem_kind_test, 16 * KB) == nullptr)
+            break;
+    }
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_GT(blocksAvailable, st.f_bfree);
+
+    err = memkind_destroy_kind(pmem_kind_test);
+    ASSERT_EQ(0, err);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_NEAR(blocksAvailable, st.f_bfree, 1);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeMemoryAfterDestroySmallClass)
+{
+    memkind_t pmem_kind_test;
+    struct statfs st;
+    double blocksAvailable;
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_test);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_test);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    blocksAvailable = st.f_bfree;
+
+    for(int i = 0; i < 100; ++i) {
+        ASSERT_NE(memkind_malloc(pmem_kind_test, 32), nullptr);
+    }
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_GT(blocksAvailable, st.f_bfree);
+
+    err = memkind_destroy_kind(pmem_kind_test);
+    ASSERT_EQ(0, err);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_NEAR(blocksAvailable, st.f_bfree, 1);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemRealloc)
+{
+    const size_t size1 = 512;
+    const size_t size2 = 1 * KB;
+    char *default_str = nullptr;
+
+    default_str = (char *)memkind_realloc(pmem_kind, default_str, size1);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc MEMKIND_PMEM with size %zu\n", size1);
+    printf("%s", default_str);
+
+    default_str = (char *)memkind_realloc(pmem_kind, default_str, size2);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc MEMKIND_PMEM with size %zu\n", size2);
+    printf("%s", default_str);
+
+    memkind_free(pmem_kind, default_str);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocUsableSize)
+{
+    const struct {
+        size_t size;
+        size_t spacing;
+    } check_sizes[] = {
+        {.size = 10, .spacing = 8},
+        {.size = 100, .spacing = 16},
+        {.size = 200, .spacing = 32},
+        {.size = 500, .spacing = 64},
+        {.size = 1000, .spacing = 128},
+        {.size = 2000, .spacing = 256},
+        {.size = 3000, .spacing = 512},
+        {.size = 1 * MB, .spacing = 4 * MB},
+        {.size = 2 * MB, .spacing = 4 * MB},
+        {.size = 3 * MB, .spacing = 4 * MB},
+        {.size = 4 * MB, .spacing = 4 * MB}
+    };
+    struct memkind *pmem_temp = nullptr;
+
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_temp);
+    size_t usable_size = memkind_malloc_usable_size(pmem_temp, nullptr);
+    size_t nullkind_usable_size = memkind_malloc_usable_size(nullptr, nullptr);
+    ASSERT_EQ(0u, usable_size);
+    ASSERT_EQ(nullkind_usable_size, usable_size);
+    for (unsigned int i = 0; i < ARRAY_SIZE(check_sizes); ++i) {
+        size_t size = check_sizes[i].size;
+        void *alloc = memkind_malloc(pmem_temp, size);
+        ASSERT_NE(nullptr, alloc);
+
+        usable_size = memkind_malloc_usable_size(pmem_temp, alloc);
+        nullkind_usable_size = memkind_malloc_usable_size(nullptr, alloc);
+        ASSERT_EQ(nullkind_usable_size, usable_size);
+        size_t diff = usable_size - size;
+        ASSERT_GE(usable_size, size);
+        ASSERT_LE(diff, check_sizes[i].spacing);
+
+        memkind_free(pmem_temp, alloc);
+    }
+    err = memkind_destroy_kind(pmem_temp);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemResize)
+{
+    const size_t size = MEMKIND_PMEM_CHUNK_SIZE;
+    char *pmem_str10 = nullptr;
+    char *pmem_strX = nullptr;
+    memkind_t pmem_kind_no_limit = nullptr;
+    memkind_t pmem_kind_not_possible = nullptr;
+    int err = 0;
+
+    pmem_str10 = (char *)memkind_malloc(pmem_kind, MEMKIND_PMEM_MIN_SIZE);
+    ASSERT_NE(nullptr, pmem_str10);
+
+    // Out of memory
+    pmem_strX = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_EQ(nullptr, pmem_strX);
+
+    memkind_free(pmem_kind, pmem_str10);
+    memkind_free(pmem_kind, pmem_strX);
+
+    err = memkind_create_pmem(PMEM_DIR, PMEM_NO_LIMIT, &pmem_kind_no_limit);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_no_limit);
+
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_no_limit, MEMKIND_PMEM_MIN_SIZE);
+    ASSERT_NE(nullptr, pmem_str10);
+
+    pmem_strX = (char *)memkind_malloc(pmem_kind_no_limit, size);
+    ASSERT_NE(nullptr, pmem_strX);
+
+    memkind_free(pmem_kind_no_limit, pmem_str10);
+    memkind_free(pmem_kind_no_limit, pmem_strX);
+
+    err = memkind_destroy_kind(pmem_kind_no_limit);
+    ASSERT_EQ(0, err);
+
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE-1,
+                              &pmem_kind_not_possible);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    ASSERT_EQ(nullptr, pmem_kind_not_possible);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocZero)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+
+    test = memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test, nullptr);
+
+    new_test = memkind_realloc(pmem_kind, test, 0);
+    ASSERT_EQ(new_test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocSizeMax)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+
+    test = memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test, nullptr);
+    errno = 0;
+    new_test = memkind_realloc(pmem_kind, test, SIZE_MAX);
+    ASSERT_EQ(new_test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+
+    memkind_free(pmem_kind, test);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocSizeZero)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+    const size_t iteration = 100;
+
+    for (unsigned i = 0; i < iteration; ++i) {
+        test = memkind_malloc(pmem_kind, size);
+        ASSERT_NE(test, nullptr);
+        errno = 0;
+        new_test = memkind_realloc(pmem_kind, test, 0);
+        ASSERT_EQ(new_test, nullptr);
+        ASSERT_EQ(errno, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocPtrCheck)
+{
+    size_t size = 1 * KB;
+    void *ptr_malloc = nullptr;
+    void *ptr_malloc_copy = nullptr;
+    void *ptr_realloc = nullptr;
+
+    ptr_malloc = memkind_malloc(pmem_kind, size);
+    ASSERT_NE(ptr_malloc, nullptr);
+
+    ptr_malloc_copy = ptr_malloc;
+
+    ptr_realloc = memkind_realloc(pmem_kind, ptr_malloc, PMEM_PART_SIZE);
+    ASSERT_EQ(ptr_realloc, nullptr);
+    ASSERT_EQ(ptr_malloc, ptr_malloc_copy);
+
+    memkind_free(pmem_kind, ptr_malloc);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptr)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+
+    test = memkind_realloc(pmem_kind, test, size);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(pmem_kind, test);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptrSizeMax)
+{
+    void *test = nullptr;
+
+    test = memkind_realloc(pmem_kind, test, SIZE_MAX);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptrZero)
+{
+    void *test = nullptr;
+
+    test = memkind_realloc(pmem_kind, test, 0);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocIncreaseSize)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocIncreaseSize";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size *= 2;
+    test2 = (char *)memkind_realloc(pmem_kind, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocIncreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocIncreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size *= 2;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocDecreaseSize)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocDecreaseSize";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 4;
+    test2 = (char *)memkind_realloc(pmem_kind, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, size);
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocDecreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocDecreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 4;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, size);
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+/*
+ * This test shows realloc "in-place" mechanism.
+ * In some cases like allocation shrinking within the same size class
+ * realloc will shrink allocation "in-place",
+ * but in others not (like changing size class).
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocInPlace)
+{
+    void *test1 = memkind_malloc(pmem_kind, 10 * MB);
+    ASSERT_NE(test1, nullptr);
+
+    /* Several reallocations within the same jemalloc size class*/
+    void *test1r = memkind_realloc(pmem_kind, test1, 6 * MB);
+    ASSERT_EQ(test1r, test1);
+
+    test1r = memkind_realloc(pmem_kind, test1, 10 * MB);
+    ASSERT_EQ(test1r, test1);
+
+    test1r = memkind_realloc(pmem_kind, test1, 8 * MB);
+    ASSERT_EQ(test1r, test1);
+
+    void *test2 = memkind_malloc(pmem_kind, 4 * MB);
+    ASSERT_NE(test2, nullptr);
+
+    /* 4MB => 16B (changing size class) */
+    void *test2r = memkind_realloc(pmem_kind, test2, 16);
+    ASSERT_NE(test2r, nullptr);
+
+    /* 8MB => 16B */
+    test1r = memkind_realloc(pmem_kind, test1, 16);
+
+    /*
+     * If the old size of the allocation is larger than
+     * the chunk size (4MB), we can reallocate it to 4MB first (in place),
+     * releasing some space, which makes it possible to do the actual
+     * shrinking...
+     */
+    ASSERT_NE(test1r, nullptr);
+    ASSERT_NE(test1r, test1);
+
+    /* ... and leaves some memory for new allocations. */
+    void *test3 = memkind_malloc(pmem_kind, 5 * MB);
+    ASSERT_NE(test3, nullptr);
+
+    memkind_free(pmem_kind, test1r);
+    memkind_free(pmem_kind, test2r);
+    memkind_free(pmem_kind, test3);
+}
+
+/*
+ * This test shows that we can make a single highest possible allocation
+ * and there still will be a place for another allocations.
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMaxFill)
+{
+    const int possible_alloc_max = 4;
+    void *test[possible_alloc_max+1] = {nullptr};
+    size_t total_mem = 0;
+    size_t free_mem = 0;
+    int i, j;
+
+    pmem_get_size(pmem_kind, total_mem, free_mem);
+
+    for (i = 0; i < possible_alloc_max; i++) {
+        for (j = total_mem; j > 0; --j) {
+            test[i] = memkind_malloc(pmem_kind, j);
+            if(test[i] != nullptr)
+                break;
+        }
+        ASSERT_NE(j, 0);
+    }
+
+    for (j = total_mem; j > 0; --j) {
+        test[possible_alloc_max] = memkind_malloc(pmem_kind, j);
+        if(test[possible_alloc_max] != nullptr)
+            break;
+    }
+    //Ensure there is no more space available on kind
+    ASSERT_EQ(j, 0);
+
+    for (i = 0; i < possible_alloc_max; i++) {
+        memkind_free(pmem_kind, test[i]);
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeNullptr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_free(pmem_kind, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptrSizeZero)
+{
+    const double test_time = 5;
+    void *test_nullptr = nullptr;
+    TimerSysTime timer;
+    timer.start();
+    do {
+        errno = 0;
+        //equivalent to memkind_malloc(pmem_kind,0)
+        test_nullptr = memkind_realloc(pmem_kind, nullptr, 0);
+        ASSERT_EQ(test_nullptr, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullKindSizeZero)
+{
+    const double test_time = 5;
+    void *test_ptr_malloc = nullptr;
+    void *test_nullptr = nullptr;
+    size_t size = 1 * KB;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        errno = 0;
+        test_ptr_malloc = memkind_malloc(pmem_kind, size);
+        ASSERT_NE(test_ptr_malloc, nullptr);
+        ASSERT_EQ(errno, 0);
+
+        errno = 0;
+        test_nullptr = memkind_realloc(nullptr, test_ptr_malloc, 0);
+        ASSERT_EQ(test_nullptr, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+/*
+ * This test will stress pmem kind with malloc-free loop
+ * with various sizes for malloc
+ */
+TEST_P(MemkindPmemTestsMalloc, test_TC_MEMKIND_PmemMallocSizeConservative)
+{
+    const int loop_limit = 10;
+    int initial_alloc_limit = 0;
+    std::vector<void *> pmem_vec;
+    void *ptr;
+    size_t alloc_size = GetParam();
+    int i, err;
+    memkind_t kind = nullptr;
+
+    memkind_config *test_cfg = memkind_config_new();
+    ASSERT_NE(nullptr, test_cfg);
+    memkind_config_set_path(test_cfg, PMEM_DIR);
+    memkind_config_set_size(test_cfg, PMEM_PART_SIZE);
+    memkind_config_set_memory_usage_policy(test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+    err =  memkind_create_pmem_with_config(test_cfg, &kind);
+    ASSERT_EQ(err, 0);
+    memkind_config_delete(test_cfg);
+
+    //check maximum number of allocations right after creating the kind
+    while ((ptr = memkind_malloc(kind, alloc_size)) != nullptr) {
+        pmem_vec.push_back(ptr);
+    }
+
+    initial_alloc_limit = pmem_vec.size();
+
+    for(auto const &val: pmem_vec) {
+        memkind_free(kind, val);
+    }
+    pmem_vec.clear();
+
+    //check number of allocations in consecutive iterations of malloc-free loop
+    for (i = 0; i < loop_limit; i++) {
+
+        while ((ptr = memkind_malloc(kind, alloc_size)) != nullptr) {
+            pmem_vec.push_back(ptr);
+        }
+        int temp_limit_of_allocations = pmem_vec.size();
+
+        for(auto const &val: pmem_vec) {
+            memkind_free(kind, val);
+        }
+        pmem_vec.clear();
+
+        ASSERT_GE(temp_limit_of_allocations, 0.98 * initial_alloc_limit);
+    }
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(0, err);
+}
+
+INSTANTIATE_TEST_CASE_P(
+    MallocParam, MemkindPmemTestsMalloc,
+    ::testing::Values(32, 60, 80, 100, 128, 150, 160, 250, 256, 300, 320,
+                      500, 512, 800, 896, 3000, 4 * KB, 6000, 10000, 60000,
+                      96 * KB, 112 * KB, 128 * KB, 160 * KB, 192 * KB, 500000,
+                      2 * MB, 5 * MB));
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocSmallSizeFillConservative)
+{
+    const size_t small_size[] = {8, 16, 32, 48, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384,
+                                 448, 512, 640, 768, 896, 1 * KB, 1280, 1536, 1792, 2 * KB, 2560,
+                                 3 * KB, 3584, 4 * KB, 5 * KB, 6 * KB, 7 * KB, 8 * KB, 10 * KB,
+                                 12 * KB, 14 * KB
+                                };
+    const int loop_limit = 100;
+    int initial_alloc_limit = 0;
+    int err;
+    std::vector<void *> pmem_vec;
+    void *ptr;
+    unsigned i, j;
+    memkind_t kind = nullptr;
+
+    memkind_config *test_cfg = memkind_config_new();
+    ASSERT_NE(nullptr, test_cfg);
+
+    memkind_config_set_path(test_cfg, PMEM_DIR);
+    memkind_config_set_size(test_cfg, PMEM_PART_SIZE);
+    memkind_config_set_memory_usage_policy(test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+    err =  memkind_create_pmem_with_config(test_cfg, &kind);
+    ASSERT_EQ(err, 0);
+    memkind_config_delete(test_cfg);
+
+    //check maximum number of allocations right after creating the kind
+    j = 0;
+    while ((ptr = memkind_malloc(kind, small_size[j++])) != nullptr) {
+        pmem_vec.push_back(ptr);
+        if (j == ARRAY_SIZE(small_size)) j = 0;
+    }
+
+    initial_alloc_limit = pmem_vec.size();
+
+    for(auto const &val: pmem_vec) {
+        memkind_free(kind, val);
+    }
+    pmem_vec.clear();
+
+    //check number of allocations in consecutive iterations of malloc-free loop
+    for (i = 0; i < loop_limit; i++) {
+
+        j = 0;
+        while ((ptr = memkind_malloc(kind, small_size[j++])) != nullptr) {
+            pmem_vec.push_back(ptr);
+            if (j == ARRAY_SIZE(small_size)) j = 0;
+        }
+
+        int temp_limit_of_allocations = pmem_vec.size();
+
+        for(auto const &val: pmem_vec) {
+            memkind_free(kind, val);
+        }
+        pmem_vec.clear();
+
+        ASSERT_GE(temp_limit_of_allocations, 0.98 * initial_alloc_limit);
+    }
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignWrongAlignmentLessThanVoidAndNotPowerOfTwo)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t wrong_alignment = 3;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignWrongAlignmentLessThanVoidAndPowerOfTwo)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t wrong_alignment = sizeof(void *)/2;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignWrongAlignmentNotPowerOfTwo)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t wrong_alignment = sizeof(void *)+1;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignLowestCorrectAlignment)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t alignment = sizeof(void *);
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, 0);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(pmem_kind, test);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPosixMemalignSizeZero)
+{
+    void *test = nullptr;
+    size_t alignment = sizeof(void *);
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, alignment, 0);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, 0);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPosixMemalignSizeMax)
+{
+    void *test = nullptr;
+    size_t alignment = 64;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, alignment, SIZE_MAX);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, ENOMEM);
+    ASSERT_EQ(test, nullptr);
+}
+
+/*
+ * This is a basic alignment test which will make alignment allocations,
+ * check pointers, write and read values from allocated memory
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPosixMemalign)
+{
+    const int test_value = 123456;
+    const int test_loop = 10;
+    uintptr_t alignment;
+    unsigned i = 0;
+    void *test = nullptr;
+    std::vector<void *> pmem_vec;
+    int ret;
+
+    for(alignment =  1 * KB; alignment <= 128 * KB; alignment *= 2) {
+        for (i = 0; i < test_loop; i++) {
+            errno = 0;
+            while((ret = memkind_posix_memalign(pmem_kind, &test, alignment,
+                                                sizeof(int *))) == 0) {
+                ASSERT_EQ(errno, 0);
+
+                pmem_vec.push_back(test);
+
+                //test pointer should be usable
+                *(int *)test = test_value;
+                ASSERT_EQ(*(int *)test, test_value);
+
+                //check for correct address alignment
+                ASSERT_EQ((uintptr_t)(test) & (alignment - 1), (uintptr_t)0);
+                errno = 0;
+            }
+
+            for(auto const &val: pmem_vec) {
+                memkind_free(pmem_kind, val);
+            }
+            pmem_vec.clear();
+        }
+    }
+}
+
+static memkind_t *pools;
+static int npools = 3;
+static void *thread_func(void *arg)
+{
+    int start_idx = *(int *)arg;
+    int err = 0;
+    for (int idx = 0; idx < npools; ++idx) {
+        int pool_id = start_idx + idx;
+
+        if (pools[pool_id] == nullptr) {
+            err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pools[pool_id]);
+            EXPECT_EQ(0, err);
+        }
+
+        if (err == 0) {
+            void *test = memkind_malloc(pools[pool_id], sizeof(void *));
+            EXPECT_TRUE(test != nullptr);
+            memkind_free(pools[pool_id], test);
+            memkind_destroy_kind(pools[pool_id]);
+        }
+    }
+
+    return nullptr;
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMultithreads)
+{
+    int nthreads = 10, status = 0;
+    pthread_t *threads = (pthread_t *)calloc(nthreads, sizeof(pthread_t));
+    ASSERT_NE(threads, nullptr);
+    int *pool_idx = (int *)calloc(nthreads, sizeof(int));
+    ASSERT_NE(pool_idx, nullptr);
+    pools = (memkind_t *)calloc(npools * nthreads, sizeof(memkind_t));
+    ASSERT_NE(pools, nullptr);
+
+    for (int t = 0; t < nthreads; t++) {
+        pool_idx[t] = npools * t;
+        status = pthread_create(&threads[t], nullptr, thread_func, &pool_idx[t]);
+        ASSERT_EQ(0, status);
+    }
+
+    for (int t = 0; t < nthreads; t++) {
+        status = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, status);
+    }
+
+    free(pools);
+    free(threads);
+    free(pool_idx);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemDestroyKind)
+{
+    const size_t pmem_array_size = 10;
+    struct memkind *pmem_kind_array[pmem_array_size] = {nullptr};
+
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE,
+                                  &pmem_kind_array[0]);
+    ASSERT_EQ(err, 0);
+
+    err = memkind_destroy_kind(pmem_kind_array[0]);
+    ASSERT_EQ(err, 0);
+
+    for (unsigned int i = 0; i < pmem_array_size; ++i) {
+        err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_kind_array[i]);
+        ASSERT_EQ(err, 0);
+    }
+
+    char *pmem_middle_name = pmem_kind_array[5]->name;
+    err = memkind_destroy_kind(pmem_kind_array[5]);
+    ASSERT_EQ(err, 0);
+
+    err = memkind_destroy_kind(pmem_kind_array[6]);
+    ASSERT_EQ(err, 0);
+
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_kind_array[5]);
+    ASSERT_EQ(err, 0);
+
+    char *pmem_new_middle_name = pmem_kind_array[5]->name;
+
+    ASSERT_STREQ(pmem_middle_name, pmem_new_middle_name);
+
+    for (unsigned int i = 0; i < pmem_array_size; ++i) {
+        if (i != 6) {
+            err = memkind_destroy_kind(pmem_kind_array[i]);
+            ASSERT_EQ(err, 0);
+        }
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemDestroyKindArenaZero)
+{
+    struct memkind *pmem_temp_1 = nullptr;
+    struct memkind *pmem_temp_2 = nullptr;
+    unsigned int arena_zero = 0;
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp_1);
+    ASSERT_EQ(err, 0);
+
+    arena_zero = pmem_temp_1->arena_zero;
+    err = memkind_destroy_kind(pmem_temp_1);
+    ASSERT_EQ(err, 0);
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp_2);
+    ASSERT_EQ(err, 0);
+
+    ASSERT_EQ(arena_zero,pmem_temp_2->arena_zero);
+
+    err = memkind_destroy_kind(pmem_temp_2);
+    ASSERT_EQ(err, 0);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCreateDestroyKindEmptyLoop)
+{
+    struct memkind *pmem_temp = nullptr;
+
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocSmallSizeFreeDefinedPmemKind)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size = 1 * KB;
+
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+        void *ptr = memkind_malloc(pmem_temp, size);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocDifferentSizesDifferentKindsDefinedFreeForAllKinds)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = MEMKIND_PMEM_CHUNK_SIZE;
+    void *ptr = nullptr;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+
+        ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr = memkind_malloc(pmem_temp, size_2);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocDifferentSizesDifferentKindsDefinedFreeForNotPmemKinds)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = MEMKIND_PMEM_CHUNK_SIZE;
+    void *ptr = nullptr;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+
+        ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr = memkind_malloc(pmem_temp, size_2);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(nullptr, ptr);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocDifferentSizesDifferentKindsNullKindFree)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = MEMKIND_PMEM_CHUNK_SIZE;
+    void *ptr = nullptr;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+
+        ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr = memkind_malloc(pmem_temp, size_2);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(nullptr, ptr);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopWithRealloc)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 512;
+    const size_t size_2 = 1 * KB;
+
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+        void *ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        void *ptr_2 = memkind_realloc(pmem_temp, ptr, size_2);
+        ASSERT_NE(nullptr, ptr_2);
+        memkind_free(pmem_temp, ptr_2);
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateCheckErrorCodeArenaCreate)
+{
+    struct memkind *pmem_temp[MEMKIND_MAX_KIND] = { nullptr };
+    unsigned i = 0, j = 0;
+    int err = 0;
+
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp[i]);
+        if (err) {
+            ASSERT_EQ(err, MEMKIND_ERROR_ARENAS_CREATE);
+            break;
+        }
+    }
+    for (j = 0; j < i; ++j) {
+        err = memkind_destroy_kind(pmem_temp[j]);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+static void *thread_func_kinds(void *arg)
+{
+    memkind_t pmem_thread_kind;
+    int err = 0;
+
+    EXPECT_TRUE(pthread_mutex_lock(&mutex) == 0);
+    EXPECT_TRUE(pthread_cond_wait(&cond, &mutex) == 0);
+    EXPECT_TRUE(pthread_mutex_unlock(&mutex) == 0);
+
+    err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_thread_kind);
+
+    if(err == 0) {
+        void *test = memkind_malloc(pmem_thread_kind, 32);
+        EXPECT_TRUE(test != nullptr);
+
+        memkind_free(pmem_thread_kind, test);
+        err = memkind_destroy_kind(pmem_thread_kind);
+        EXPECT_EQ(0, err);
+    }
+
+    return nullptr;
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMultithreadsStressKindsCreate)
+{
+    const int nthreads = 50;
+    int i, t, err;
+    int max_possible_kind = 0;
+    memkind_t pmem_kinds[MEMKIND_MAX_KIND] = {nullptr};
+    pthread_t *threads = (pthread_t *)calloc(nthreads, sizeof(pthread_t));
+    ASSERT_NE(threads, nullptr);
+
+    // This loop will create as many kinds as possible
+    // to obtain a real kind limit
+    for (i = 0; i < MEMKIND_MAX_KIND; i++) {
+        err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kinds[i]);
+        if(err != 0) {
+            ASSERT_GT(i, 0);
+            max_possible_kind = i;
+            --i;
+            break;
+        }
+        ASSERT_NE(nullptr, pmem_kinds[i]);
+    }
+
+    // destroy last kind so it will be possible
+    // to create only one pmem kind in threads
+    err = memkind_destroy_kind(pmem_kinds[i]);
+    ASSERT_EQ(0, err);
+
+    for (t = 0; t < nthreads; t++) {
+        err = pthread_create(&threads[t], nullptr, thread_func_kinds, nullptr);
+        ASSERT_EQ(0, err);
+    }
+
+    sleep(1);
+    err = pthread_cond_broadcast(&cond);
+    ASSERT_EQ(0, err);
+
+    for (t = 0; t < nthreads; t++) {
+        err = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, err);
+    }
+
+    for (i = 0; i < max_possible_kind - 1; i++) {
+        err = memkind_destroy_kind(pmem_kinds[i]);
+        ASSERT_EQ(0, err);
+    }
+
+    free(threads);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemKindFreeBenchmarkOneThread)
+{
+    const size_t pmem_array_size = 10;
+    const size_t alloc_size = 512;
+    struct memkind *pmem_kind_array[pmem_array_size] = { nullptr };
+    std::vector<void *> pmem_vec_exp[pmem_array_size];
+    std::vector<void *> pmem_vec_imp[pmem_array_size];
+    void *ptr;
+    TimerSysTime timer;
+    double test1Time, test2Time;
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        while ((ptr = memkind_malloc(pmem_kind_array[i], alloc_size)) != nullptr) {
+            pmem_vec_exp[i].push_back(ptr);
+        }
+    }
+
+    timer.start();
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        for(auto const &val: pmem_vec_exp[i]) {
+            memkind_free(pmem_kind_array[i], val);
+        }
+    }
+
+    test1Time = timer.getElapsedTime();
+    printf("Free time with explicitly kind: %f\n", test1Time);
+
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        while ((ptr = memkind_malloc(pmem_kind_array[i], alloc_size)) != nullptr) {
+            pmem_vec_imp[i].push_back(ptr);
+        }
+    }
+
+    timer.start();
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        for(auto const &val: pmem_vec_imp[i]) {
+            memkind_free(nullptr, val);
+        }
+    }
+    test2Time = timer.getElapsedTime();
+    printf("Free time with implicitly kind: %f\n", test2Time);
+
+    ASSERT_LT(test1Time, test2Time);
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        int err = memkind_destroy_kind(pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+}
+
+static const int threadsNum = 10;
+static memkind *testKind[threadsNum] = { nullptr };
+static const int mallocCount = 100000;
+static void *ptr[threadsNum][mallocCount] = { { nullptr } };
+
+static void *thread_func_FreeWithNullptr(void *arg)
+{
+    int kindIndex = *(int *)arg;
+    EXPECT_TRUE(pthread_mutex_lock(&mutex) == 0);
+    EXPECT_TRUE(pthread_cond_wait(&cond, &mutex) == 0);
+    EXPECT_TRUE(pthread_mutex_unlock(&mutex) == 0);
+
+    for (int j = 0; j < mallocCount; ++j) {
+        if (ptr[kindIndex][j] == nullptr) {
+            break;
+        }
+        memkind_free(nullptr, ptr[kindIndex][j]);
+        ptr[kindIndex][j] = nullptr;
+    }
+
+    return nullptr;
+}
+
+static void *thread_func_FreeWithKind(void *arg)
+{
+    int kindIndex = *(int *)arg;
+    EXPECT_TRUE(pthread_mutex_lock(&mutex) == 0);
+    EXPECT_TRUE(pthread_cond_wait(&cond, &mutex) == 0);
+    EXPECT_TRUE(pthread_mutex_unlock(&mutex) == 0);
+
+    for (int j = 0; j < mallocCount; ++j) {
+        if (ptr[kindIndex][j] == nullptr) {
+            break;
+        }
+        memkind_free(testKind[kindIndex], ptr[kindIndex][j]);
+    }
+
+    return nullptr;
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemKindFreeBenchmarkWithThreads)
+{
+    const size_t allocSize = 512;
+    int err;
+    TimerSysTime timer;
+    double duration;
+    pthread_t *threads = (pthread_t *)calloc(threadsNum, sizeof(pthread_t));
+    ASSERT_NE(threads, nullptr);
+
+    for (int i = 0; i < threadsNum; ++i) {
+        err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &testKind[i]);
+        ASSERT_EQ(err, 0);
+        ASSERT_NE(nullptr, testKind[i]);
+        int j = 0;
+        for (j = 0; j < mallocCount; ++j) {
+            ptr[i][j] = memkind_malloc(testKind[i], allocSize);
+            if (ptr[i][j] == nullptr) {
+                break;
+            }
+        }
+        ASSERT_NE(j, mallocCount);
+    }
+
+    int threadIndex[threadsNum];
+    for (int i = 0; i < threadsNum; ++i) {
+        threadIndex[i] = i;
+    }
+
+    for (int t = 0; t < threadsNum; t++) {
+        err = pthread_create(&threads[t], nullptr, thread_func_FreeWithKind,
+                             &threadIndex[t]);
+        ASSERT_EQ(0, err);
+    }
+
+    // sleep is here to ensure that all threads start at one the same time
+    sleep(1);
+    timer.start();
+    err = pthread_cond_broadcast(&cond);
+    ASSERT_EQ(0, err);
+    for (int t = 0; t < threadsNum; ++t) {
+        int err = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, err);
+    }
+    duration = timer.getElapsedTime();
+    printf("Free time with explicitly kind: %f\n", duration);
+
+    for (int i = 0; i < threadsNum; i++) {
+        int j = 0;
+        for (j = 0; j < mallocCount; ++j) {
+            ptr[i][j] = memkind_malloc(testKind[i], allocSize);
+            if (ptr[i][j] == nullptr) {
+                break;
+            }
+        }
+        ASSERT_NE(j, mallocCount);
+    }
+
+    for (int t = 0; t < threadsNum; ++t) {
+        err = pthread_create(&threads[t], nullptr, thread_func_FreeWithNullptr,
+                             &threadIndex[t]);
+        ASSERT_EQ(0, err);
+    }
+
+    // sleep is here to ensure that all threads start at one the same time
+    sleep(1);
+    timer.start();
+    err = pthread_cond_broadcast(&cond);
+    ASSERT_EQ(0, err);
+    for (int t = 0; t < threadsNum; t++) {
+        err = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, err);
+    }
+
+    duration = timer.getElapsedTime();
+    printf("Free time with implicitly kind: %f\n", duration);
+
+    for (int i = 0; i < threadsNum; ++i) {
+        err = memkind_destroy_kind(testKind[i]);
+        ASSERT_EQ(0, err);
+    }
+    free(threads);
+}
+
+/*
+ * Test will create array of kinds, fill them and then attempt to free one ptr
+ * passing nullptr instead of kind. Then it will try to malloc again and if it will be
+ * successful the test passes (memkind_free(nullptr,...) was successful)
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeUsingNullptrInsteadOfKind)
+{
+    const size_t pmem_array_size = 10;
+    const size_t alloc_size = 1 * KB;
+    struct memkind *pmem_kind_array[pmem_array_size] = { nullptr };
+    std::vector<void *> pmem_vec[pmem_array_size];
+    void *testPtr = nullptr;
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        while ((testPtr = memkind_malloc(pmem_kind_array[i], alloc_size)) != nullptr) {
+            pmem_vec[i].push_back(testPtr);
+        }
+    }
+
+    memkind_free(nullptr, pmem_vec[5].at(5));
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        // attempt to alloc memory to the kinds
+        testPtr = memkind_malloc(pmem_kind_array[i], alloc_size);
+        if (i == 5) {
+            // allocation should be successful - confirmation that memkind_free(nullptr,...) works fine
+            ASSERT_NE(testPtr, nullptr);
+            pmem_vec[i].at(5) = testPtr;
+        } else {
+            // There is no more free space in other kinds
+            ASSERT_EQ(testPtr, nullptr);
+        }
+    }
+
+    // free the rest of the space and destroy kinds.
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        for(auto const &val: pmem_vec[i]) {
+            memkind_free(pmem_kind_array[i], val);
+        }
+        int err = memkind_destroy_kind(pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+}
+
+/*
+ * This is a test which confirms that extent deallocation function ( pmem_extent_dalloc )
+ * was called correctly for pmem allocation.
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCheckExtentDalloc)
+{
+    struct memkind *kind = nullptr;
+    const int mallocLimit = 10000;
+    void *ptr[mallocLimit] = { nullptr };
+    struct stat st;
+    double initialBlocks;
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &kind);
+    ASSERT_EQ(err, 0);
+
+    struct memkind_pmem *priv = (memkind_pmem *)kind->priv;
+
+    for (int x = 0; x < 10; ++x) {
+        // sleep is here to help trigger dalloc extent
+        sleep(2);
+        int allocCount = 0;
+        for (int i = 0; i < mallocLimit; ++i) {
+            ptr[i] = memkind_malloc(kind, 32);
+            if (ptr[i] == nullptr)
+                break;
+
+            allocCount = i;
+        }
+
+        // store initial amount of allocated blocks
+        if (x == 0) {
+            ASSERT_EQ(0, fstat(priv->fd, &st));
+            initialBlocks = st.st_blocks;
+        }
+
+        for (int i = 0; i < allocCount; ++i)
+            memkind_free(kind, ptr[i]);
+
+        ASSERT_EQ(0, fstat(priv->fd, &st));
+        // if amount of blocks is less than initial, extent was called.
+        if (initialBlocks > st.st_blocks)
+            break;
+    }
+    ASSERT_GT(initialBlocks, st.st_blocks);
+
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKINDPmemDefragreallocate_success)
+{
+    struct memkind *kind = nullptr;
+    unsigned i;
+    unsigned count_mem_transfer = 0;
+    const unsigned alloc = 50000;
+    void *nptr;
+    std::vector<void *> pmem_vec;
+    pmem_vec.reserve(alloc);
+    int err = memkind_create_pmem(PMEM_DIR, 0, &kind);
+    ASSERT_EQ(err, 0);
+
+    for (i = 0; i < alloc; ++i) {
+        void *ptr =  memkind_malloc(kind, 1 * KB);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 1 * KB);
+        pmem_vec.push_back(ptr);
+    }
+
+    for (i = 1; i < alloc;) {
+        memkind_free(kind, pmem_vec.at(i));
+        pmem_vec.at(i) = nullptr;
+        // Free memory with irregular pattern
+        if (i % 2 == 0)
+            i += 3;
+        else
+            i += 5;
+    }
+
+    for (i = 0; i < pmem_vec.size(); ++i) {
+        nptr = memkind_defrag_reallocate(kind, pmem_vec.at(i));
+        if (nptr) {
+            pmem_vec.at(i) = nptr;
+            count_mem_transfer++;
+        }
+    }
+    ASSERT_NE(count_mem_transfer, 0U);
+
+    for(auto const &val: pmem_vec) {
+        memkind_free(kind, val);
+    }
+
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(err, 0);
+}
diff --git a/memkind-1.10.0/test/memkind_stat_test.cpp b/memkind-1.10.0/test/memkind_stat_test.cpp
new file mode 100644
index 0000000..42f5075
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_stat_test.cpp
@@ -0,0 +1,85 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include "common.h"
+
+class MemkindStatTests: public ::testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_StatUnknownType)
+{
+    size_t value;
+    int err = memkind_get_stat(nullptr, MEMKIND_STAT_TYPE_MAX_VALUE, &value);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    err = memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_MAX_VALUE, &value);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+}
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_DefaultKind)
+{
+    size_t value;
+    int err = memkind_update_cached_stats();
+    ASSERT_EQ(MEMKIND_SUCCESS, err);
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 100);
+    for(int i = 0; i < MEMKIND_STAT_TYPE_MAX_VALUE; ++i) {
+        err = memkind_get_stat(MEMKIND_DEFAULT, static_cast<memkind_stat_type>(i),
+                               &value);
+        ASSERT_EQ(MEMKIND_SUCCESS, err);
+    }
+    memkind_free(nullptr, ptr);
+}
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_EmptyRegularKind)
+{
+    size_t stat;
+    int err = memkind_update_cached_stats();
+    ASSERT_EQ(MEMKIND_SUCCESS, err);
+    for(int i = 0; i < MEMKIND_STAT_TYPE_MAX_VALUE; ++i) {
+        err = memkind_get_stat(MEMKIND_REGULAR, static_cast<memkind_stat_type>(i),
+                               &stat);
+        ASSERT_EQ(MEMKIND_SUCCESS, err);
+        ASSERT_EQ(stat, 0U);
+    }
+}
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_GetGlobal)
+{
+    size_t value;
+    int err = memkind_update_cached_stats();
+    ASSERT_EQ(MEMKIND_SUCCESS, err);
+    for(int i = 0; i < MEMKIND_STAT_TYPE_MAX_VALUE; ++i) {
+        int err = memkind_get_stat(nullptr, static_cast<memkind_stat_type>(i), &value);
+        ASSERT_EQ(MEMKIND_SUCCESS, err);
+    }
+}
diff --git a/memkind-1.10.0/test/memkind_versioning_tests.cpp b/memkind-1.10.0/test/memkind_versioning_tests.cpp
new file mode 100644
index 0000000..12e8000
--- /dev/null
+++ b/memkind-1.10.0/test/memkind_versioning_tests.cpp
@@ -0,0 +1,60 @@
+/*
+ * Copyright (C) 2015 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#include "memkind.h"
+
+#include "common.h"
+
+/*
+ * memkind versioning tests.
+ */
+class MemkindVersioningTests: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+    const int max_version_value = 999;
+
+};
+
+//Test memkind_get_version().
+TEST_F(MemkindVersioningTests, test_TC_MEMKIND_GetVersionFunc)
+{
+    int max_return_val =
+        1000000 * max_version_value
+        + 1000 * max_version_value
+        + max_version_value;
+
+    //version number > 0
+    EXPECT_GT(memkind_get_version(), 0);
+
+    //version number <= max_return_val
+    EXPECT_LE(memkind_get_version(), max_return_val);
+}
diff --git a/memkind-1.10.0/test/memory_footprint_test.cpp b/memkind-1.10.0/test/memory_footprint_test.cpp
new file mode 100644
index 0000000..feaf0e8
--- /dev/null
+++ b/memkind-1.10.0/test/memory_footprint_test.cpp
@@ -0,0 +1,354 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "common.h"
+#include "random_sizes_allocator.h"
+#include "proc_stat.h"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+#include "allocator_perf_tool/Allocation_info.hpp"
+
+#include <memkind.h>
+
+#include <condition_variable>
+#include <functional>
+#include <mutex>
+#include <random>
+#include <thread>
+
+class Worker
+{
+public:
+    Worker(RandomSizesAllocator &&allocator, double malloc_probability)
+        : allocator(std::move(allocator)), malloc_probability(malloc_probability)  {}
+
+    void work()
+    {
+        if (allocator.empty() || get_random_bool(malloc_probability)) {
+            requested_memory_sum += allocator.malloc_random_memory();
+        } else {
+            requested_memory_sum -= allocator.free_random_memory();
+        }
+    }
+
+    size_t get_requested_memory_sum_bytes() const
+    {
+        return requested_memory_sum;
+    }
+
+private:
+    bool get_random_bool(double probability)
+    {
+        std::bernoulli_distribution distribution(probability);
+        return distribution(generator);
+    }
+
+    std::default_random_engine generator;
+    RandomSizesAllocator allocator;
+    size_t requested_memory_sum = 0;
+    double malloc_probability;
+};
+
+
+class MemoryFootprintStats
+{
+public:
+    void reset()
+    {
+        std::lock_guard<std::mutex> lk(sample_guard);
+        initial_virtual_memory = proc_stat.get_virtual_memory_size_bytes();
+        initial_physical_memory = proc_stat.get_physical_memory_size_bytes();
+        vm_overhead_sum = 0;
+        current_vm_overhead = 0;
+        max_vm_overhead = 0;
+
+        current_phys_overhead = 0;
+        phys_overhead_sum = 0;
+        max_phys_overhead = 0;
+
+        requested_memory = 0;
+        sample_count = 0;
+    }
+
+    void sample(long long requested_memory_bytes)
+    {
+        std::lock_guard<std::mutex> lk(sample_guard);
+        current_vm = proc_stat.get_virtual_memory_size_bytes();
+        current_phys = proc_stat.get_physical_memory_size_bytes();
+
+        sample_count++;
+
+        requested_memory = requested_memory_bytes;
+
+        current_vm_overhead = current_vm - initial_virtual_memory - requested_memory;
+        vm_overhead_sum += current_vm_overhead;
+        max_vm_overhead = std::max(max_vm_overhead, current_vm_overhead);
+
+        current_phys_overhead = current_phys - initial_physical_memory;
+        phys_overhead_sum += current_phys_overhead;
+        max_phys_overhead = std::max(max_phys_overhead, current_phys_overhead);
+    }
+
+    void log_data() const
+    {
+        std::lock_guard<std::mutex> lk(sample_guard);
+        GTestAdapter::RecordProperty("avg_vm_overhead_per_operation_mb",
+                                     convert_bytes_to_mb(vm_overhead_sum) / sample_count);
+        GTestAdapter::RecordProperty("avg_vm_overhead_growth_per_operation_mb",
+                                     convert_bytes_to_mb(current_vm_overhead) / sample_count);
+        GTestAdapter::RecordProperty("max_vm_overhead_mb",
+                                     convert_bytes_to_mb(max_vm_overhead));
+        GTestAdapter::RecordProperty("avg_phys_overhead_per_operation_mb",
+                                     convert_bytes_to_mb(phys_overhead_sum) / sample_count);
+        GTestAdapter::RecordProperty("max_phys_overhead_mb",
+                                     convert_bytes_to_mb(max_phys_overhead));
+        GTestAdapter::RecordProperty("overhead_to_requested_memory_ratio_percent",
+                                     100.f * current_vm_overhead / requested_memory);
+        GTestAdapter::RecordProperty("requested_memory_mb",
+                                     convert_bytes_to_mb(requested_memory));
+    }
+private:
+    long long initial_virtual_memory;
+    long long vm_overhead_sum = 0;
+    long long current_vm_overhead = 0;
+    long long max_vm_overhead = 0;
+
+    long long initial_physical_memory;
+    long long current_phys_overhead = 0;
+    long long phys_overhead_sum = 0;
+    long long max_phys_overhead = 0;
+
+    long long requested_memory = 0;
+    long long sample_count = 0;
+
+    long long current_vm;
+    long long current_phys;
+
+    ProcStat proc_stat;
+    mutable std::mutex sample_guard;
+};
+
+/* Execute func calling it n_calls times in n_threads threads.
+ * The execution is multithreaded but the func calls order is sequential.
+ * func takes thread id, and operation id as an argument,
+ * and must return thread id of the next thread, where thread ids are in range <0, n_threads-1>.
+ * init_thread_id specify the initial thread id.
+ */
+void run_multithreaded_seq_exec(unsigned n_threads, unsigned init_thread_id,
+                                unsigned n_calls, std::function<unsigned(unsigned, unsigned)> func)
+{
+    std::vector<std::thread> threads;
+    std::mutex mutex;
+    std::condition_variable turns_holder;
+    unsigned current_call = 0;
+    unsigned current_tid = init_thread_id;
+
+    threads.reserve(n_threads);
+
+    mutex.lock();
+
+    for(int tid=0; tid<n_threads; ++tid) {
+        threads.emplace_back([ &, tid]() {
+            while(current_call < n_calls) {
+                std::unique_lock<std::mutex> lk(mutex);
+                turns_holder.wait(lk, [ &,tid] {return current_tid == tid || current_call == n_calls;});
+                if(current_call == n_calls) {
+                    return;
+                }
+                current_tid = func(tid, current_call);
+                ASSERT_LT(current_tid, n_threads) << "Incorrect thread id!";
+                current_call++;
+                lk.unlock();
+                turns_holder.notify_all();
+            }
+        });
+    }
+
+    mutex.unlock();
+
+    for(int i=0; i<threads.size(); i++) {
+        threads[i].join();
+    }
+}
+
+/*
+ * Create threads and measure the cost of maintaining allocations from threads.
+ * Allocations order is sequential (otherwise the results might be very nondeterministic).
+ */
+void run_test(memkind_t kind, size_t min_size, size_t max_size,
+              unsigned n_threads, double malloc_probability=1.0, unsigned n_calls=1000)
+{
+    Worker worker(RandomSizesAllocator(kind, min_size, max_size, n_calls),
+                  malloc_probability);
+
+    MemoryFootprintStats mem_footprint_stats;
+
+    auto func = [&](unsigned tid, unsigned id) -> unsigned {
+        if(id == 0)
+        {
+            mem_footprint_stats.reset();
+        }
+
+        worker.work();
+        mem_footprint_stats.sample(worker.get_requested_memory_sum_bytes());
+
+        return (tid + 1) % n_threads; //next thread id
+    };
+    run_multithreaded_seq_exec(n_threads, 0, n_calls, func);
+    mem_footprint_stats.log_data();
+}
+
+class MemoryFootprintTest: public :: testing::Test
+{};
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_small_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_small_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 16 * KB, 1 * MB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_medium_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 16 * KB, 1 * MB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_large_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 2 * MB, 10 * MB, 1, 1.0, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_large_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 2 * MB, 10 * MB, 10, 1.0, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_small_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_small_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 10, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 16 * KB, 1 * MB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_large_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 2 * MB, 10 * MB,  10, 0.8, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_small_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_small_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_medium_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_large_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 1, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_large_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 10, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_small_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_small_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 10, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_medium_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 10, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_large_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 1, 0.8, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_large_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 10, 0.8, 100);
+}
diff --git a/memkind-1.10.0/test/memory_manager.h b/memkind-1.10.0/test/memory_manager.h
new file mode 100644
index 0000000..e44dd8d
--- /dev/null
+++ b/memkind-1.10.0/test/memory_manager.h
@@ -0,0 +1,81 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include <memkind.h>
+#include <new>
+
+class MemoryManager
+{
+private:
+    memkind_t kind;
+    size_t memory_size;
+    void *memory_pointer;
+
+    void move(MemoryManager &&other)
+    {
+        kind = other.kind;
+        memory_size = other.memory_size;
+        if (memory_pointer)
+            memkind_free(kind, memory_pointer);
+        memory_pointer = std::move(other.memory_pointer);
+        other.memory_pointer = nullptr;
+    }
+
+public:
+    MemoryManager(memkind_t kind, size_t size) :
+        kind(kind),
+        memory_size(size),
+        memory_pointer(memkind_malloc(kind, size))
+    {
+        if(!memory_pointer) {
+            throw std::bad_alloc();
+        }
+    }
+
+    size_t size()
+    {
+        return memory_size;
+    }
+
+    MemoryManager(const MemoryManager &) = delete;
+
+    MemoryManager(MemoryManager &&other)
+    {
+        move(std::move(other));
+    }
+
+    MemoryManager &operator=(MemoryManager &&other)
+    {
+        move(std::move(other));
+        return *this;
+    }
+
+    ~MemoryManager()
+    {
+        if (memory_pointer)
+            memkind_free(kind, memory_pointer);
+    }
+};
+
diff --git a/memkind-1.10.0/test/multithreaded_tests.cpp b/memkind-1.10.0/test/multithreaded_tests.cpp
new file mode 100644
index 0000000..8c6b8a8
--- /dev/null
+++ b/memkind-1.10.0/test/multithreaded_tests.cpp
@@ -0,0 +1,87 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include <fstream>
+#include <algorithm>
+
+#include "common.h"
+#include "check.h"
+#include "omp.h"
+#include "trial_generator.h"
+
+#define NTHREADS 2
+
+/* Multithread test content which will also use the trial_benerator to call
+ * malloc, calloc, memalign, realloc, with different memory sizes but this time
+ * it will use Open MP threads with NTHREADS set to 2
+ */
+class MultithreadedTest : public TGTest
+{
+};
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_Malloc_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_MALLOC);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_Calloc_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_CALLOC);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_Memalign_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_MEMALIGN);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_MemalignPsize_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_MEMALIGN_PSIZE);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
diff --git a/memkind-1.10.0/test/negative_tests.cpp b/memkind-1.10.0/test/negative_tests.cpp
new file mode 100644
index 0000000..02ea1ab
--- /dev/null
+++ b/memkind-1.10.0/test/negative_tests.cpp
@@ -0,0 +1,593 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <fstream>
+#include <algorithm>
+#include <numa.h>
+#include <errno.h>
+#include <limits.h>
+#include <sys/sysinfo.h>
+
+#include "common.h"
+#include "check.h"
+#include "omp.h"
+#include "trial_generator.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+/* Set of negative test cases for memkind, its main goal are to verify that the
+ * library behaves accordingly to documentation when calling an API with
+ * invalid inputs, incorrect usage, NULL pointers.
+ */
+class NegativeTest: public ::testing::Test
+{};
+
+class NegativeTestHuge: public ::testing::Test
+{
+private:
+    //Enable huge pages to avoid false positive test result.
+    HugePageOrganizer huge_page_organizer = HugePageOrganizer(8);
+};
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_zero_memtype)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  memkind_memtype_t(), //Set incorrect value.
+                  MEMKIND_POLICY_PREFERRED_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_incorrect_memtype)
+{
+    memkind_memtype_t memtype_flags;
+    //Set incorrect value.
+    memset(&memtype_flags, -1, sizeof(memtype_flags));
+
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  memtype_flags,
+                  MEMKIND_POLICY_PREFERRED_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_incorrect_policy)
+{
+    memkind_policy_t policy;
+    //Set incorrect value.
+    memset(&policy, -1, sizeof(policy));
+
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  policy,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_incorrect_mask)
+{
+    memkind_bits_t flags;
+
+    //Set incorrect value.
+    memset(&flags, 255, sizeof(flags));
+
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_PREFERRED_LOCAL,
+                  flags,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_LOCAL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_LOCAL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_LOCAL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_ALL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_ALL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_LOCAL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_LOCAL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_ALL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_HIGH_BANDWIDTH_INTERLEAVE_LOCAL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_HIGH_BANDWIDTH,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_HIGH_BANDWIDTH_INTERLEAVE_LOCAL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_HIGH_BANDWIDTH,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_HIGH_BANDWIDTH_INTERLEAVE_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_HIGH_BANDWIDTH,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_HIGH_BANDWIDTH_BIND_ALL)
+{
+    memkind_t kind;
+    int flags_tmp = MEMKIND_MEMTYPE_DEFAULT | MEMKIND_MEMTYPE_HIGH_BANDWIDTH;
+    memkind_memtype_t memtype_flags;
+    memcpy(&memtype_flags, &flags_tmp, sizeof(memtype_flags));
+
+    int ret = memkind_create_kind(
+                  memtype_flags,
+                  MEMKIND_POLICY_BIND_ALL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_HIGH_BANDWIDTH_INTERLEAVE_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int flags_tmp = MEMKIND_MEMTYPE_DEFAULT | MEMKIND_MEMTYPE_HIGH_BANDWIDTH;
+    memkind_memtype_t memtype_flags;
+    memcpy(&memtype_flags, &flags_tmp, sizeof(memtype_flags));
+
+    int ret = memkind_create_kind(
+                  memtype_flags,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_ErrorMemAlign)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = EINVAL;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_DEFAULT,
+                                 &ptr,
+                                 5,
+                                 100);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_DefaultSizeZero)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = 0;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_DEFAULT,
+                                 &ptr,
+                                 16,
+                                 0);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_ErrorAlignment)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = EINVAL;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_HBW,
+                                 &ptr,
+                                 5,
+                                 100);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_HBWSizeZero)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = 0;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_HBW,
+                                 &ptr,
+                                 16,
+                                 0);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_ErrorAllocM)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = ENOMEM;
+    struct sysinfo info;
+    unsigned long long MemTotal = 0;
+
+    ret = sysinfo(&info);
+    EXPECT_EQ(ret, 0);
+
+    //Determine total memory size as totalram (total usable main memory size)
+    //multiplied by mem_unit (memory unit size in bytes). This value is equal
+    //to MemTotal field in /proc/meminfo.
+    MemTotal = info.totalram * info.mem_unit;
+
+    RecordProperty("MemTotal_kB", MemTotal/KB);
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_HBW,
+                                 &ptr,
+                                 16,
+                                 2*MemTotal);
+    EXPECT_EQ(err, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_malloc_over_size)
+{
+    void *ptr = hbw_malloc(SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_malloc_size_zero)
+{
+    void *ptr = hbw_malloc(0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_malloc_over_size)
+{
+    void *ptr = memkind_malloc(MEMKIND_HBW, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_malloc_size_zero)
+{
+    void *ptr = memkind_malloc(MEMKIND_HBW, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_calloc_over_size)
+{
+    void *ptr = hbw_calloc(1, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_calloc_size_zero)
+{
+    void *ptr = hbw_calloc(1, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_calloc_over_size)
+{
+    void *ptr = memkind_calloc(MEMKIND_HBW, 1, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_calloc_size_zero)
+{
+    void *ptr = memkind_calloc(MEMKIND_HBW, 1, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_realloc_over_size)
+{
+    void *ptr = hbw_realloc(NULL, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_realloc_over_size)
+{
+    void *ptr = memkind_realloc(MEMKIND_HBW, NULL, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_posix_memalign_over_size)
+{
+    void *ptr = NULL;
+    int ret = hbw_posix_memalign(&ptr, 4096, SIZE_MAX);
+    EXPECT_TRUE(ptr == NULL);
+    EXPECT_EQ(ENOMEM, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_posix_memalign_size_zero)
+{
+    void *ptr = NULL;
+    errno = 0;
+
+    int ret = hbw_posix_memalign(&ptr, 4096, 0);
+    EXPECT_TRUE(ptr == NULL);
+    EXPECT_EQ(0, ret);
+    EXPECT_EQ(0, errno);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_posix_memalign_over_size)
+{
+    void *ptr = NULL;
+    int ret = memkind_posix_memalign(MEMKIND_HBW, &ptr, 4096, SIZE_MAX);
+    EXPECT_TRUE(ptr == NULL);
+    EXPECT_EQ(ENOMEM, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBFailureMemalign)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = EINVAL;
+
+    ret = hbw_posix_memalign_psize(&ptr,
+                                   1073741824,
+                                   1073741826,
+                                   HBW_PAGESIZE_1GB_STRICT);
+    EXPECT_EQ(ret, err);
+    EXPECT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_RegularReallocWithMemAllign)
+{
+    int ret = 0;
+    void *ptr = NULL;
+
+    ret = hbw_posix_memalign_psize(&ptr,
+                                   4096,
+                                   4096,
+                                   HBW_PAGESIZE_4KB);
+    EXPECT_EQ(ret, 0);
+    ASSERT_TRUE(ptr != NULL);
+    memset(ptr, 0, 4096);
+    ptr = hbw_realloc(ptr, 8192);
+    memset(ptr, 0, 8192);
+    hbw_free(ptr);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicy)
+{
+    // First call should be successful, consequent should generate a warning
+    // and be ignored
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_PREFERRED), 0);
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_INTERLEAVE), EPERM);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+    EXPECT_EQ(hbw_set_policy((hbw_policy_t)0xFF), EINVAL);
+}
+
+//Check if hbw_set_policy() will be ignored after malloc.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterMalloc)
+{
+    void *ptr = hbw_malloc(512);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after calloc.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterCalloc)
+{
+    void *ptr = hbw_calloc(512, 1);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after realloc.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterRealloc)
+{
+    void *ptr = hbw_malloc(512);
+    EXPECT_TRUE(ptr != NULL);
+
+    hbw_realloc(ptr, 512);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after hbw_posix_memalign.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterHbwPosixMemalign)
+{
+    void *ptr = NULL;
+
+    hbw_posix_memalign(&ptr, 2048, 2048);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after hbw_posix_memalign_psize.
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_SetPolicyAfterHbwPosixMemalignPsize)
+{
+    void *ptr = NULL;
+
+    hbw_posix_memalign_psize(&ptr, 2048, 2048, HBW_PAGESIZE_4KB);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBMemalignPsizeAllign)
+{
+    void *ptr = NULL;
+    int ret = 0;
+    int err = EINVAL;
+
+    ret = hbw_posix_memalign_psize(&ptr, -1, 1024, HBW_PAGESIZE_1GB);
+    EXPECT_EQ(err, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBNullRealloc)
+{
+    void *ptr = NULL;
+    ptr = memkind_realloc(MEMKIND_HBW_GBTLB, NULL, -1);
+    EXPECT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBNullFree)
+{
+    memkind_free(MEMKIND_GBTLB, NULL);
+}
+
+TEST_F(NegativeTestHuge,
+       test_TC_MEMKIND_hbwmalloc_memalign_psize_Interleave_Policy_PAGE_SIZE_2MB)
+{
+    void *ptr = NULL;
+    hbw_set_policy(HBW_POLICY_INTERLEAVE);
+    int ret = hbw_posix_memalign_psize(&ptr, 4096, 4096, HBW_PAGESIZE_2MB);
+    ASSERT_EQ(EINVAL, ret);
+}
diff --git a/memkind-1.10.0/test/performance/framework.cpp b/memkind-1.10.0/test/performance/framework.cpp
new file mode 100644
index 0000000..aca24f0
--- /dev/null
+++ b/memkind-1.10.0/test/performance/framework.cpp
@@ -0,0 +1,357 @@
+/*
+* Copyright (C) 2014 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <algorithm> // sort
+#include <math.h> // log2
+#include <cassert>
+#include <iostream>
+#include "framework.hpp"
+
+namespace performance_tests
+{
+    namespace ch = std::chrono;
+    using std::cout;
+    using std::endl;
+    using std::unique_lock;
+    using std::mutex;
+
+
+#ifdef __DEBUG
+    mutex g_coutMutex;
+    int g_msgLevel = 1;
+#endif
+
+    void Barrier::wait()
+    {
+        unique_lock<mutex> lock(m_barrierMutex);
+        // Decrement number of threads awaited at the barrier
+        m_waiting--;
+        if (m_waiting == 0) {
+            // Called by the last expected thread - notify all waiting threads and exit
+            m_cVar.notify_all();
+            // Store the time when barrier was released
+            if (m_releasedAt.tv_sec == 0 && m_releasedAt.tv_nsec == 0) {
+                clock_gettime(CLOCK_MONOTONIC, &m_releasedAt);
+            }
+            return;
+        }
+        // Wait unitl the last expected thread calls wait() on Barrier instance, or timeout occurs
+        m_cVar.wait_until(lock, ch::system_clock::now() + ch::seconds(10), []() {
+            return GetInstance().m_waiting == 0;
+        });
+    }
+
+    // Worker class
+    Worker::Worker(
+        uint32_t actionsCount,
+        const vector<size_t> &allocationSizes,
+        Operation *freeOperation,
+        memkind_t kind)
+        : m_actionsCount(actionsCount)
+        , m_allocationSizes(allocationSizes)
+        , m_actions(vector<Action*>(actionsCount, nullptr))
+        , m_kind(kind)
+    {
+        assert(freeOperation->getName() == OperationName::Free);
+    }
+
+    Worker::~Worker()
+    {
+        for (Action *action : m_actions) { //each action
+            delete action;
+        }
+
+    }
+
+    void Worker::init(const vector<Operation *> &testOperations,
+                      Operation *&freeOperation)
+    {
+        for(uint32_t i = 0 ; i < m_actionsCount ; i++) {
+            int bucketSize = rand() % Operation::MaxBucketSize;
+
+            for (Operation *operation : testOperations) { //each operation
+                if (operation->checkCondition(bucketSize)) {
+                    size_t size = m_allocationSizes[m_allocationSizes.size() > 1 ? rand() %
+                                                                             m_allocationSizes.size() : 0];
+                    m_actions[i] = new Action(
+                        operation,
+                        freeOperation,
+                        m_kind,
+                        size,
+                        log2(rand() % size),
+                        sizeof(void *) * (1 << ((rand() % Operation::MemalignMaxMultiplier))));
+                    break;
+                }
+            }
+        }
+    }
+
+    void Worker::run()
+    {
+        m_thread = new thread(&Worker::work, this);
+    }
+
+#ifdef __DEBUG
+    uint16_t Worker::getId()
+    {
+        return m_threadId;
+    }
+    void Worker::setId(uint16_t threadId)
+    {
+        m_threadId = threadId;
+    }
+#endif
+
+    void Worker::finish()
+    {
+        if (m_thread != nullptr) {
+            m_thread->join();
+            delete m_thread;
+        }
+    }
+
+    void Worker::work()
+    {
+        EMIT(1, "Entering barrier " << m_threadId)
+        Barrier::GetInstance().wait();
+        EMIT(1, "Starting thread " << m_threadId)
+        for (Action *action : m_actions) {
+            action->alloc();
+        }
+    }
+
+    void Worker::clean()
+    {
+        EMIT(2, "Cleaning thread " << m_threadId)
+        for (Action *action : m_actions) {
+            action->free();
+        }
+        EMIT(1, "Thread " << m_threadId << " finished")
+    }
+
+    // PerformanceTest class
+    PerformanceTest::PerformanceTest(
+        size_t repeatsCount,
+        size_t threadsCount,
+        size_t operationsCount)
+        : m_repeatsCount(repeatsCount)
+        , m_discardCount(repeatsCount * (distardPercent / 100.0))
+        , m_threadsCount(threadsCount)
+        , m_operationsCount(operationsCount)
+        , m_executionMode(ExecutionMode::SingleInteration)
+    {
+    }
+
+    void PerformanceTest::setAllocationSizes(const vector<size_t> &allocationSizes)
+    {
+        m_allocationSizes = allocationSizes;
+    }
+
+    void PerformanceTest::setOperations(const vector<vector<Operation *>>
+                                        &testOperations, Operation *freeOperation)
+    {
+        m_testOperations = testOperations;
+        m_freeOperation = freeOperation;
+    }
+
+    void PerformanceTest::setExecutionMode(ExecutionMode executionMode)
+    {
+        m_executionMode = executionMode;
+    }
+
+    void PerformanceTest::setKind(const vector<memkind_t> &kinds)
+    {
+        m_kinds = kinds;
+    }
+
+    inline void PerformanceTest::runIteration()
+    {
+        timespec iterationStop, iterationStart;
+
+        Barrier::GetInstance().reset(m_threadsCount);
+        for (Worker *worker : m_workers) {
+            worker->run();
+        }
+        for (Worker *worker : m_workers) {
+            worker->finish();
+        }
+        EMIT(1, "Alloc completed");
+        clock_gettime(CLOCK_MONOTONIC, &iterationStop);
+        iterationStart = Barrier::GetInstance().releasedAt();
+        m_durations.push_back(
+            (iterationStop.tv_sec  * NanoSecInSec + iterationStop.tv_nsec) -
+            (iterationStart.tv_sec * NanoSecInSec + iterationStart.tv_nsec)
+        );
+        for (Worker *worker : m_workers) {
+            worker->clean();
+        }
+    }
+
+    void PerformanceTest::prepareWorkers()
+    {
+        for (size_t threadId = 0; threadId < m_threadsCount; threadId++) {
+            m_workers.push_back(
+                new Worker(
+                    m_operationsCount,
+                    m_allocationSizes,
+                    m_freeOperation,
+                    m_kinds.size() > 0 ? m_kinds[threadId % m_kinds.size()] : nullptr)
+            );
+#ifdef __DEBUG
+            m_workers.back()->setId(threadId);
+#endif
+            if (m_executionMode == ExecutionMode::SingleInteration) {
+                // In ManyIterations mode, operations will be set for each thread at the beginning of each iteration
+                m_workers.back()->init(m_testOperations[threadId % m_testOperations.size()],
+                                       m_freeOperation);
+            }
+        }
+    }
+
+    Metrics PerformanceTest::getMetrics()
+    {
+        uint64_t totalDuration = 0;
+
+        std::sort(m_durations.begin(), m_durations.end());
+
+        m_durations.erase(m_durations.end() - m_discardCount, m_durations.end());
+        for (uint64_t &duration : m_durations) {
+            totalDuration += duration;
+        }
+
+        Metrics metrics;
+
+        metrics.executedOperations   = m_durations.size() * m_threadsCount *
+                                       m_operationsCount;
+        metrics.totalDuration        = totalDuration;
+        metrics.repeatDuration       = (double) totalDuration /
+                                       ((uint64_t)m_durations.size() * NanoSecInSec);
+        metrics.iterationDuration    = metrics.repeatDuration;
+        if (m_executionMode == ExecutionMode::ManyIterations) {
+            metrics.executedOperations *= m_testOperations.size();
+            metrics.iterationDuration  /= m_testOperations.size();
+        }
+        metrics.operationsPerSecond  = (double) metrics.executedOperations *
+                                       NanoSecInSec / totalDuration;
+        metrics.avgOperationDuration = (double) totalDuration /
+                                       metrics.executedOperations;
+        assert(metrics.iterationDuration != 0.0);
+        return metrics;
+    }
+
+    void PerformanceTest::writeMetrics(const string &suiteName,
+                                       const string &caseName, const string &fileName)
+    {
+        Metrics metrics = getMetrics();
+
+        // For thousands separation
+        setlocale(LC_ALL, "");
+        if (!fileName.empty()) {
+            FILE *f;
+            if((f = fopen(fileName.c_str(), "a+"))) {
+                fprintf(f,
+                        "%s;%s;%zu;%zu;%lu;%f;%f;%f;%f\n",
+                        suiteName.c_str(),
+                        caseName.c_str(),
+                        m_repeatsCount,
+                        m_threadsCount,
+                        metrics.executedOperations,
+                        metrics.operationsPerSecond,
+                        metrics.avgOperationDuration,
+                        metrics.iterationDuration,
+                        metrics.repeatDuration);
+                fclose(f);
+            }
+
+        }
+        printf("Operations/sec:\t\t\t%'f\n"
+               "Avg. operation duration:\t%f nsec\n"
+               "Iteration duration:\t\t%f sec\n"
+               "Repeat duration:\t\t%f sec\n",
+               metrics.operationsPerSecond,
+               metrics.avgOperationDuration,
+               metrics.iterationDuration,
+               metrics.repeatDuration);
+    }
+
+    int PerformanceTest::run()
+    {
+        if (m_testOperations.empty() ||
+            m_allocationSizes.empty() ||
+            m_freeOperation == nullptr) {
+            cout << "ERROR: Test not initialized" << endl;
+            return 1;
+        }
+        // Create threads
+        prepareWorkers();
+        //warmup kinds
+        void *alloc = nullptr;
+
+        for (const memkind_t &kind : m_kinds) {
+            m_testOperations[0][0]->perform(kind, alloc, 1e6);
+            m_freeOperation->perform(kind, alloc);
+        }
+        for (size_t repeat = 0; repeat < m_repeatsCount; repeat++) {
+            EMIT(1, "Test run #" << repeat)
+            if (m_executionMode == ExecutionMode::SingleInteration) {
+                runIteration();
+            } else {
+                // Perform each operations list in separate iteration, for each thread
+                for (vector<Operation *> &ops : m_testOperations) {
+                    for (Worker *worker : m_workers) {
+                        worker->init(ops, m_freeOperation);
+                    }
+                    runIteration();
+                }
+            }
+        }
+        return 0;
+    }
+
+    void PerformanceTest::showInfo()
+    {
+        printf("Test parameters: %lu repeats, %lu threads, %d operations per thread\n",
+               m_repeatsCount,
+               m_threadsCount,
+               m_operationsCount);
+        printf("Thread memory allocation operations:\n");
+        for (unsigned long i = 0; i < m_testOperations.size(); i++) {
+            if (m_executionMode == ExecutionMode::SingleInteration) {
+                printf("\tThread %lu,%lu,...\n", i, i + (m_testOperations.size()));
+            } else {
+                printf("\tIteration %lu\n", i);
+            }
+            for (const Operation *op : m_testOperations[i]) {
+                printf("\t\t %s (bucket size: %d)\n", op->getNameStr().c_str(),
+                       op->getBucketSize());
+            }
+        }
+        printf("Memory free operation:\n\t\t%s\n",
+               m_freeOperation->getNameStr().c_str());
+        printf("Allocation sizes:\n");
+        for (size_t size : m_allocationSizes) {
+            printf("\t\t%lu bytes\n", size);
+        }
+    }
+}
diff --git a/memkind-1.10.0/test/performance/framework.hpp b/memkind-1.10.0/test/performance/framework.hpp
new file mode 100644
index 0000000..a7c2d08
--- /dev/null
+++ b/memkind-1.10.0/test/performance/framework.hpp
@@ -0,0 +1,280 @@
+/*
+* Copyright (C) 2014 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <memkind.h>
+
+#include <thread>
+#include <vector>
+#include <mutex>
+#include <condition_variable>
+// Malloc, jemalloc, memkind jemalloc and memkind memory operations definitions
+#include "operations.hpp"
+
+/* Framework for testing memory allocators pefromance */
+namespace performance_tests
+{
+// Nanoseconds in second
+    const uint32_t NanoSecInSec = 1e9;
+
+    // Simple barrier implementation
+    class Barrier
+    {
+        // Barrier mutex
+        std::mutex m_barrierMutex;
+        // Contitional variable
+        std::condition_variable m_cVar;
+        // Number of threads expected to enter the barrier
+        size_t m_waiting;
+        timespec m_releasedAt;
+
+    public:
+        // Called by each thread entering the barrier; returns control to caller
+        // only after been called from the last thread expected at the barrier
+        void wait();
+
+        // (Re)Initializes the barrier
+        void reset(unsigned waiting)
+        {
+            m_releasedAt.tv_sec = m_releasedAt.tv_nsec = 0;
+            m_waiting = waiting;
+        }
+
+        // Get time when barrier was released
+        timespec &releasedAt()
+        {
+            return m_releasedAt;
+        }
+
+        // Singleton
+        static Barrier &GetInstance()
+        {
+            // Automatically created and deleted one and only instance
+            static Barrier instance;
+            return instance;
+        }
+    private:
+        Barrier()
+        {
+            reset(0);
+        }
+        // Cannot be used with singleton, so prevent compiler from creating them automatically
+        Barrier(Barrier const &)        = delete;
+        void operator=(Barrier const &) = delete;
+    };
+
+    // Data of a single test action, that is, memory operation (malloc, calloc, etc.)
+    // to perform and its parameters (size, alignment etc.)
+    class Action
+    {
+    protected:
+        Operation *m_operation;
+        Operation *m_freeOperation;
+        const memkind_t m_kind;
+        void *m_allocation;
+        const size_t m_size;
+        const size_t m_offset;
+        const size_t m_alignment;
+
+    public:
+        Action(
+            Operation *operation,
+            Operation *freeOperation,
+            const memkind_t kind,
+            const size_t size,
+            const size_t offset,
+            const size_t alignment)
+            : m_operation(operation)
+            , m_freeOperation(freeOperation)
+            , m_kind(kind)
+            , m_allocation(nullptr)
+            , m_size(size)
+            , m_offset(offset)
+            , m_alignment(alignment)
+        {}
+
+        Action(
+            Operation *operation,
+            Operation *freeOperation,
+            const memkind_t kind)
+            : Action(operation, freeOperation, kind, 0, 0, 0)
+        {
+        }
+
+        void alloc()
+        {
+            m_operation->perform(m_kind, m_allocation, m_size, m_offset, m_alignment);
+        }
+
+        void free()
+        {
+            m_freeOperation->perform(m_kind, m_allocation);
+        }
+
+    };
+
+    // Performs and tracks requested memory operations in a separate thread
+    class Worker
+    {
+    protected:
+#ifdef __DEBUG
+        uint16_t m_threadId;
+#endif
+        // Requested number of test actions
+        const uint32_t m_actionsCount;
+        // List of memory block sizes - for each memory allocation operation actual value is chosen randomly
+        const vector<size_t> &m_allocationSizes;
+        // List of test actions
+        vector<Action *> m_actions;
+        // Memory free action
+        Action *m_freeAction;
+        // Operation kind (useful for memkind only)
+        memkind_t m_kind;
+        // Working thread
+        thread *m_thread;
+
+    public:
+        Worker(
+            uint32_t actionsCount,
+            const vector<size_t> &allocationSizes,
+            Operation *freeOperation,
+            memkind_t kind);
+
+        ~Worker();
+
+        // Set operations list for the worker
+        void init(const vector<Operation *> &testOperations, Operation *&freeOperation);
+
+        // Create & start thread
+        void run();
+
+#ifdef __DEBUG
+        // Get thread id
+        uint16_t getId();
+
+        // Set thread id
+        void setId(uint16_t threadId);
+#endif
+
+        // Finish thread and free all allocations made
+        void finish();
+
+        // Free allocated memory
+        virtual void clean();
+
+    private:
+        // Actual thread function (allow inheritance)
+        virtual void work();
+    };
+
+    enum ExecutionMode {
+        SingleInteration, // Single iteration, operations listS will be distributed among threads sequentially
+        ManyIterations    // Each operations list will be run in separate iteration by each thread
+    };
+
+    struct Metrics {
+        uint64_t executedOperations;
+        uint64_t totalDuration;
+        double operationsPerSecond;
+        double avgOperationDuration;
+        double iterationDuration;
+        double repeatDuration;
+    };
+
+    // Performance test parameters class
+    class PerformanceTest
+    {
+    protected:
+        // empirically determined % of worst results needed to be discarded
+        // to eliminate malloc() performance results skewness
+        static constexpr double distardPercent = 20.0;
+    protected:
+        // Number of test repeats
+        size_t                      m_repeatsCount;
+        // Number of test repeats with worst results to be discarded
+        size_t                      m_discardCount;
+        // Number of threads
+        size_t                      m_threadsCount;
+        // Number of memory operations in each thread
+        uint32_t                    m_operationsCount;
+        // List of allocation sizes
+        vector<size_t>              m_allocationSizes;
+        // List of list of allocation operations, utlization depends on execution mode
+        vector<vector<Operation *>> m_testOperations;
+        // Free operation
+        Operation                  *m_freeOperation;
+        // List of memory kinds (for memkind allocation only)
+        // distributed among threads sequentially
+        vector<memkind_t>           m_kinds;
+        // List of thread workers
+        vector<Worker *>            m_workers;
+        // Time measurement
+        vector<uint64_t>            m_durations;
+        // Execution mode
+        ExecutionMode               m_executionMode;
+
+    public:
+        // Create test
+        PerformanceTest(
+            size_t repeatsCount,
+            size_t threadsCount,
+            size_t operationsCount);
+
+        virtual ~PerformanceTest() {}
+
+        // Set list of block sizes
+        void setAllocationSizes(const vector<size_t> &allocationSizes);
+
+        // Set list of operations per thread/per iteration (depending on execution mode)
+        void setOperations(const vector<vector<Operation *>> &testOperations,
+                           Operation *freeOperation);
+
+        // Set per-thread list of memory kinds
+        void setKind(const vector<memkind_t> &kinds);
+
+        // Set execution mode (different operations per each thread/same operations for each thread, but many iterations)
+        void setExecutionMode(ExecutionMode operationMode);
+
+        // Execute test
+        int run();
+
+        // Print test parameters
+        virtual void showInfo();
+
+        // Write test metrics
+        void writeMetrics(const string &suiteName, const string &caseName,
+                          const string &fileName = "");
+
+        Metrics getMetrics();
+    private:
+
+        // Run single iteration
+        void runIteration();
+
+        // Setup thread workers
+        void prepareWorkers();
+
+    };
+}
diff --git a/memkind-1.10.0/test/performance/operations.hpp b/memkind-1.10.0/test/performance/operations.hpp
new file mode 100644
index 0000000..7a5d6b6
--- /dev/null
+++ b/memkind-1.10.0/test/performance/operations.hpp
@@ -0,0 +1,477 @@
+/*
+* Copyright (C) 2014 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <memkind.h>
+
+#include <malloc.h>
+#include <cassert>
+#include "jemalloc/jemalloc.h"
+
+// Malloc, jemalloc, memkind jemalloc and memkind memory operations definitions
+namespace performance_tests
+{
+    using std::vector;
+    using std::string;
+    using std::thread;
+
+#ifdef __DEBUG
+#include <mutex>
+    // Write entire text at once, avoiding switching to another thread
+    extern int g_msgLevel;
+    extern std::mutex g_coutMutex;
+#define EMIT(LEVEL, TEXT) \
+    if (g_msgLevel >= LEVEL) \
+    { \
+        g_coutMutex.lock(); std::cout << TEXT << std::endl; g_coutMutex.unlock(); \
+    }
+#else
+#define EMIT(LEVEL, TEXT)
+#endif
+
+    // Use jemalloc, compiled with unique prefix (--with-jemalloc-prefix= configure option)
+#ifdef SYSTEM_JEMALLOC_PREFIX
+#define TOKENPASTE(x, y) x ## y
+#define JE(x, y) TOKENPASTE(x, y)
+#define jexx_malloc JE(SYSTEM_JEMALLOC_PREFIX, malloc)
+#define jexx_calloc JE(SYSTEM_JEMALLOC_PREFIX, calloc)
+#define jexx_memalign JE(SYSTEM_JEMALLOC_PREFIX, memalign)
+#define jexx_realloc JE(SYSTEM_JEMALLOC_PREFIX, realloc)
+#define jexx_free JE(SYSTEM_JEMALLOC_PREFIX, free)
+    extern "C" {
+        // jemalloc function prototypes
+        // full header cannot be include due to conflict with memkind jemalloc
+        extern void *jexx_malloc(size_t size);
+        extern void *jexx_calloc(size_t num, size_t size);
+        extern void *jexx_memalign(size_t alignment, size_t size);
+        extern void *jexx_realloc(void *ptr, size_t size);
+        extern void  jexx_free(void *ptr);
+    }
+#endif
+
+    // Available memory operations
+    enum OperationName {
+        Malloc,
+        Calloc,
+        Realloc,
+        Align,
+        Free,
+        Invalid
+    };
+
+    // Reprents a memory operation
+    class Operation
+    {
+    public:
+        // Each operation is assigned a bucket size from range (0, MaxBucketSize)
+        static const unsigned MaxBucketSize = 100;
+        // For memalign operation, alignment parameter will be a random value
+        // from range (sizeof(void*), sizeof(void*) * MemalignMaxMultiplier)
+        static const unsigned MemalignMaxMultiplier = 4;
+
+    protected:
+        OperationName m_name;
+        // If random number from range (0, MaxBucketSize) is lower than m_bucketSize, an operation will be performed
+        unsigned m_bucketSize;
+
+    public:
+        Operation()
+        {
+        }
+
+        // Creates an operation with given name and bucket size
+        // If no bucket size is given, operation will be always performed
+        Operation(
+            OperationName name,
+            unsigned bucketSize = MaxBucketSize)
+            : m_name(name)
+            , m_bucketSize(bucketSize)
+        {
+            assert(bucketSize <= MaxBucketSize);
+        }
+
+        virtual ~Operation() {}
+        ;
+
+        // Check if operation should be performed (currently drawn random number lower than bucket size)
+        bool checkCondition(unsigned ballSize) const
+        {
+            return (ballSize < m_bucketSize);
+        }
+
+        OperationName getName() const
+        {
+            return m_name;
+        }
+
+        string getNameStr() const
+        {
+            switch (m_name) {
+                case OperationName::Malloc:
+                    return "malloc";
+
+                case OperationName::Calloc:
+                    return "calloc";
+
+                case OperationName::Realloc:
+                    return "realloc";
+
+                case OperationName::Align:
+                    return "align";
+
+                case OperationName::Free:
+                    return "free";
+
+                default:
+                    return "<unknown>";
+            }
+        }
+
+        // Get operation bucket size
+        unsigned getBucketSize() const
+        {
+            return m_bucketSize;
+        }
+
+        // perform memory operation
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size = 0,
+                             size_t offset=0,
+                             size_t alignment=0)
+        const = 0;
+
+    };
+
+    // Malloc memory operations
+    class MallocOperation : public Operation
+    {
+    public:
+        MallocOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        MallocOperation(
+            OperationName name,
+            unsigned bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch(m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        free(mem);
+                    }
+                    mem = malloc(size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        free(mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = calloc((1 << offset), (size >> offset));
+                    break;
+                }
+                case Realloc: {
+                    mem = realloc(mem, size);
+                    break;
+                }
+                case Align: {
+                    if (mem != nullptr) {
+                        free(mem);
+                    }
+                    // randomly choose alignment from (8, 8 * MemalignMaxMultiplie)
+                    mem = memalign(alignment, size);
+                    break;
+                }
+                case Free: {
+                    free(mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    throw "Not implemented";
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+        }
+    };
+
+#ifdef SYSTEM_JEMALLOC_PREFIX
+    // Jemalloc memory operations
+    class JemallocOperation : public Operation
+    {
+    public:
+        JemallocOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        JemallocOperation(
+            OperationName name,
+            unsigned bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch(m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        jexx_free(mem);
+                    }
+                    mem = jexx_malloc(size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        jexx_free(mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = jexx_calloc((1 <<, offset), (size >>, offset));
+                    break;
+                }
+                case Realloc: {
+                    mem = jexx_realloc(mem, size);
+                    break;
+                }
+                case Align: {
+                    if (mem != nullptr) {
+                        jexx_free(mem);
+                    }
+                    // randomly choose alignment from (8, 8 * MemalignMaxMultiplie)
+                    mem = jexx_memalign(alignment, size);
+                    break;
+                }
+                case Free: {
+                    jexx_free(mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    throw "Not implemented";
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+        }
+    };
+#endif
+    // Jemkmalloc memory operations
+    class JemkmallocOperation : public Operation
+    {
+    public:
+        JemkmallocOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        JemkmallocOperation(
+            OperationName name,
+            unsigned bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+#ifdef JEMK
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch(m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        jemk_free(mem);
+                    }
+                    mem = jemk_malloc(size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        jemk_free(mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = jemk_calloc((1 << offset), (size >> offset));
+                    break;
+                }
+                case Realloc: {
+                    mem = jemk_realloc(mem, size);
+                    break;
+                }
+                case Align: {
+                    if (mem != nullptr) {
+                        jemk_free(mem);
+                    }
+                    // randomly choose alignment from (8, 8 * MemalignMaxMultiplie)
+                    mem = jemk_memalign(alignment, size);
+                    break;
+                }
+                case Free: {
+                    jemk_free(mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    throw "Not implemented";
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+#endif // JE_MK
+        }
+    };
+
+    // Memkind memory operations
+    class MemkindOperation : public Operation
+    {
+    public:
+        MemkindOperation()
+        {}
+
+        MemkindOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        MemkindOperation(
+            OperationName name,
+            size_t bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch (m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        memkind_free(kind, mem);
+                    }
+                    mem = memkind_malloc(kind, size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        memkind_free(kind, mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = memkind_calloc(kind, 1 << offset, size >> offset);
+                    break;
+                }
+                case Realloc: {
+                    mem = memkind_realloc(kind, mem, size);
+                    break;
+                }
+
+                case Align: {
+                    if (mem != nullptr) {
+                        memkind_free(kind, mem);
+                        mem = nullptr;
+                    }
+                    // randomly choose alignment from (sizeof(void*), sizeof(void*) * MemalignMaxMultiplie)
+                    if (memkind_posix_memalign(kind, &mem, alignment, size) != 0) {
+                        // failure
+                        mem = nullptr;
+                    }
+                    break;
+                }
+                case Free: {
+                    memkind_free(kind, mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+        }
+    };
+}
diff --git a/memkind-1.10.0/test/performance/perf_tests.cpp b/memkind-1.10.0/test/performance/perf_tests.cpp
new file mode 100644
index 0000000..6018337
--- /dev/null
+++ b/memkind-1.10.0/test/performance/perf_tests.cpp
@@ -0,0 +1,162 @@
+/*
+* Copyright (C) 2014 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "perf_tests.hpp"
+#include <iostream>
+#include <cmath>
+#include <gtest/gtest.h>
+
+// Memkind performance tests
+using std::cout;
+using std::endl;
+using std::abs;
+using std::ostringstream;
+
+// Memkind tests
+class PerformanceTest : public testing::Test
+{
+protected:
+    const double Tolerance = 0.15;
+    const double Confidence = 0.10;
+
+    Metrics referenceMetrics;
+    Metrics performanceMetrics;
+    PerfTestCase<performance_tests::MallocOperation> referenceTest;
+    PerfTestCase<performance_tests::MemkindOperation> performanceTest;
+
+    template<class T>
+    static void RecordProperty(const string &key, const T value)
+    {
+        ostringstream values;
+        values << value;
+        testing::Test::RecordProperty(key, values.str());
+    }
+
+    void SetUp()
+    {
+    }
+
+    void TearDown()
+    {
+    }
+
+    bool checkDelta(double value, double reference, const string &info,
+                    double delta, bool higherIsBetter = false)
+    {
+        // If higherIsBetter is true, current value should be >= (reference - delta); otherwise, it should be <= (reference + delta)
+        double threshold = higherIsBetter ? reference * (1 - delta) : reference *
+                           (1 + delta);
+        cout <<
+             "Metric: " << info << ". Reference value: " << reference << ". "
+             "Expected: " << (higherIsBetter ? ">= " : "<= ") << threshold << " (delta = " <<
+             delta << ")."
+             "Actual: " << value << " (delta = " <<
+             ((value > reference ? (value / reference) : (reference / value)) - 1.0) /
+             ((higherIsBetter != (value > reference)) ? 1.0 : -1.0)
+             << ")."
+             << endl ;
+        if (higherIsBetter ? (value <= threshold) : (value >= threshold)) {
+            cout << "WARNING : Value of '" << info << "' outside expected bounds!" << endl;
+            return false;
+        }
+        return true;
+    }
+
+    bool compareMetrics(Metrics metrics, Metrics reference, double delta)
+    {
+        bool result = true;
+        result &= checkDelta(metrics.operationsPerSecond, reference.operationsPerSecond,
+                             "operationsPerSecond", delta, true);
+        result &= checkDelta(metrics.avgOperationDuration,
+                             reference.avgOperationDuration, "avgOperationDuration", delta);
+        return result;
+    }
+
+    void writeMetrics(Metrics metrics, Metrics reference)
+    {
+        RecordProperty("ops_per_sec", metrics.operationsPerSecond);
+        RecordProperty("ops_per_sec_vs_ref",
+                       (reference.operationsPerSecond - metrics.operationsPerSecond) * 100.0f
+                       / reference.operationsPerSecond);
+        RecordProperty("avg_op_time_nsec", metrics.avgOperationDuration);
+        RecordProperty("avg_op_time_nsec_vs_ref",
+                       (metrics.avgOperationDuration - reference.avgOperationDuration) * 100.0f
+                       / reference.avgOperationDuration);
+    }
+
+    void run()
+    {
+        cout << "Running reference std::malloc test" << endl;
+        referenceMetrics = referenceTest.runTest();
+
+        cout << "Running memkind test" << endl;
+        performanceMetrics = performanceTest.runTest();
+
+        writeMetrics(performanceMetrics, referenceMetrics);
+        EXPECT_TRUE(compareMetrics(performanceMetrics, referenceMetrics,
+                                   Tolerance + Confidence));
+    }
+};
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_single_op_single_iter)
+{
+    referenceTest.setupTest_singleOpSingleIter();
+    performanceTest.setupTest_singleOpSingleIter();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_single_iter)
+{
+    referenceTest.setupTest_manyOpsSingleIter();
+    performanceTest.setupTest_manyOpsSingleIter();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_single_iter_huge_alloc)
+{
+    referenceTest.setupTest_manyOpsSingleIterHugeAlloc();
+    performanceTest.setupTest_manyOpsSingleIterHugeAlloc();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_single_op_many_iters)
+{
+    referenceTest.setupTest_singleOpManyIters();
+    performanceTest.setupTest_singleOpManyIters();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_many_iters)
+{
+    referenceTest.setupTest_manyOpsManyIters();
+    performanceTest.setupTest_manyOpsManyIters();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_many_iters_many_kinds)
+{
+    referenceTest.setupTest_manyOpsManyIters();
+    performanceTest.setupTest_manyOpsManyIters();
+    run();
+}
diff --git a/memkind-1.10.0/test/performance/perf_tests.hpp b/memkind-1.10.0/test/performance/perf_tests.hpp
new file mode 100644
index 0000000..8cb5bd5
--- /dev/null
+++ b/memkind-1.10.0/test/performance/perf_tests.hpp
@@ -0,0 +1,190 @@
+/*
+* Copyright (C) 2014 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include "framework.hpp"
+#include "operations.hpp"
+
+using std::vector;
+using std::string;
+
+using performance_tests::Operation;
+using performance_tests::OperationName;
+using performance_tests::Metrics;
+using performance_tests::ExecutionMode;
+
+template <class T>
+class PerfTestCase
+{
+
+private:
+    vector<vector<Operation *>> m_testOperations;
+    Operation *m_freeOperation;
+    performance_tests::PerformanceTest *m_test;
+    const unsigned m_seed = 1297654;
+    const unsigned m_repeats = 50;
+    const unsigned m_threads = 64;
+    const unsigned m_iterations = 100;
+
+public:
+    PerfTestCase()
+    {
+        m_freeOperation = new T(OperationName::Free);
+        srand(m_seed);
+        m_test = nullptr;
+    }
+
+    ~PerfTestCase()
+    {
+        delete m_test;
+        for (vector<Operation *> ops : m_testOperations) {
+            for (Operation *op : ops) {
+                delete op;
+            }
+        }
+    }
+
+    // Perform common actions for all test cases
+    Metrics runTest(vector<memkind_t> kinds = { MEMKIND_DEFAULT })
+    {
+        m_test->setKind(kinds);
+        m_test->showInfo();
+        m_test->run();
+        return m_test->getMetrics();
+    }
+
+    // malloc only
+    // 128 bytes
+
+    void setupTest_singleOpSingleIter()
+    {
+
+        m_testOperations = { { new T(OperationName::Malloc) } };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 128 });
+    }
+
+    // malloc, calloc, realloc and align (equal probability)
+    // 120, 521, 1200 and 4099 bytes
+    void setupTest_manyOpsSingleIter()
+    {
+        m_testOperations = { {
+                new T(OperationName::Malloc, 25),
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 120, 521, 1200, 4099 });
+    }
+
+    // malloc, calloc, realloc and align (equal probability)
+    // 500000, 1000000, 2000000 and 4000000 bytes
+    void setupTest_manyOpsSingleIterHugeAlloc()
+    {
+        m_testOperations = { {
+                new T(OperationName::Malloc, 25),
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 500000, 1000000, 2000000, 4000000 });
+    }
+
+    // 4 iterations of each thread (first malloc, then calloc, realloc and align)
+    // 120, 521, 1200 and 4099 bytes
+    void setupTest_singleOpManyIters()
+    {
+        m_testOperations = {
+            { new T(OperationName::Malloc, 100) },
+            { new T(OperationName::Calloc, 100) },
+            { new T(OperationName::Realloc, 100) },
+            { new T(OperationName::Align, 100) }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 120, 521, 1200, 4099 });
+        m_test->setExecutionMode(ExecutionMode::ManyIterations);
+    }
+
+    // 4 iterations of each thread, all with same set of operations (malloc, then calloc, realloc and align),
+    // but different operation probability
+    // 120, 521, 1200 and 4099 bytes
+    void setupTest_manyOpsManyIters()
+    {
+        m_testOperations = {
+            {
+                new T(OperationName::Malloc, 25),
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Malloc, 50),
+                new T(OperationName::Calloc, 70),
+                new T(OperationName::Realloc, 80),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Malloc, 60),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Realloc, 60),
+                new T(OperationName::Malloc, 80),
+                new T(OperationName::Calloc, 90),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Realloc, 40),
+                new T(OperationName::Malloc, 55),
+                new T(OperationName::Calloc, 70),
+                new T(OperationName::Align, 100)
+            }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 120, 521, 1200, 4099 });
+        m_test->setExecutionMode(ExecutionMode::ManyIterations);
+    }
+};
diff --git a/memkind-1.10.0/test/pmem_alloc_performance_tests.cpp b/memkind-1.10.0/test/pmem_alloc_performance_tests.cpp
new file mode 100644
index 0000000..f6ee32d
--- /dev/null
+++ b/memkind-1.10.0/test/pmem_alloc_performance_tests.cpp
@@ -0,0 +1,430 @@
+/*
+* Copyright (C) 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/TaskFactory.hpp"
+#include "allocator_perf_tool/Stats.hpp"
+#include "allocator_perf_tool/Thread.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+#include "allocator_perf_tool/PmemMockup.hpp"
+
+static const size_t PMEM_PART_SIZE = 0;
+extern const char  *PMEM_DIR;
+
+class PmemAllocPerformanceTest: public :: testing::Test
+{
+private:
+    AllocatorFactory allocator_factory;
+
+protected:
+    void SetUp()
+    {
+        allocator_factory.initialize_allocator(AllocatorTypes::STANDARD_ALLOCATOR);
+
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &MEMKIND_PMEM_MOCKUP);
+        ASSERT_EQ(0, err);
+        ASSERT_NE(nullptr, MEMKIND_PMEM_MOCKUP);
+    }
+
+    void TearDown()
+    {
+        int err = memkind_destroy_kind(MEMKIND_PMEM_MOCKUP);
+        ASSERT_EQ(0, err);
+    }
+
+    float run(unsigned kind, unsigned call, size_t threads_number,
+              size_t alloc_size, unsigned mem_operations_num)
+    {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+        TypesConf func_calls;
+        TypesConf allocator_types;
+
+        func_calls.enable_type(FunctionCalls::FREE);
+        func_calls.enable_type(call);
+        allocator_types.enable_type(kind);
+
+        TaskConf conf = {
+            mem_operations_num, //number of memory operations
+            {
+                mem_operations_num, //number of memory operations
+                alloc_size, //min. size of single allocation
+                alloc_size //max. size of single allocatioion
+            },
+            func_calls, //enable function calls
+            allocator_types, //enable allocators
+            11, //random seed
+            false, //do not log memory operations and statistics to csv file
+        };
+
+        for (int i=0; i<threads_number; i++) {
+            Task *task = task_factory.create(conf);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats time_stats;
+        for (int i=0; i<tasks.size(); i++) {
+            time_stats += tasks[i]->get_results();
+        }
+
+        return time_stats.stats[kind][call].total_time;
+    }
+
+    void run_test(unsigned kind, unsigned call, size_t threads_number,
+                  size_t alloc_size, unsigned mem_operations_num)
+    {
+        allocator_factory.initialize_allocator(kind);
+        float ref_time = run(AllocatorTypes::STANDARD_ALLOCATOR, call, threads_number,
+                             alloc_size, mem_operations_num);
+        float perf_time = run(kind, call, threads_number, alloc_size,
+                              mem_operations_num);
+        float ref_delta_time_percent = allocator_factory.calc_ref_delta(ref_time,
+                                                                        perf_time);
+
+        GTestAdapter::RecordProperty("total_time_spend_on_alloc", perf_time);
+        GTestAdapter::RecordProperty("alloc_operations_per_thread", mem_operations_num);
+        GTestAdapter::RecordProperty("ref_delta_time_percent", ref_delta_time_percent);
+    }
+
+};
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 100, 10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 1572864,
+             1000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 100, 10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 1572864,
+             1000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 1572864,
+             1000);
+}
diff --git a/memkind-1.10.0/test/pmem_allocator_tests.cpp b/memkind-1.10.0/test/pmem_allocator_tests.cpp
new file mode 100644
index 0000000..54a601e
--- /dev/null
+++ b/memkind-1.10.0/test/pmem_allocator_tests.cpp
@@ -0,0 +1,322 @@
+/*
+ * Copyright (C) 2018 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "memkind.h"
+#include "pmem_allocator.h"
+#include "common.h"
+#include <vector>
+#include <memory>
+#include <array>
+#include <list>
+#include <map>
+#include <utility>
+#include <string>
+#include <algorithm>
+#include <scoped_allocator>
+#include <thread>
+
+extern const char *PMEM_DIR;
+
+// Tests for pmem::allocator class.
+class PmemAllocatorTests: public ::testing::Test
+{
+public:
+    const size_t pmem_max_size = 1024*1024*1024;
+    static_assert(MEMKIND_MEM_USAGE_POLICY_MAX_VALUE == static_cast<int>
+                  (libmemkind::allocation_policy::MAX),
+                  "Allocation policy is not inline with memory usage policy");
+    libmemkind::pmem::allocator<int> alloc_source { std::string(PMEM_DIR), pmem_max_size };
+    libmemkind::pmem::allocator<int> alloc_conservative  { std::string(PMEM_DIR), pmem_max_size, libmemkind::allocation_policy::CONSERVATIVE };
+
+    libmemkind::pmem::allocator<int> alloc_source_f1 { std::string(PMEM_DIR), pmem_max_size };
+    libmemkind::pmem::allocator<int> alloc_source_f2 { std::string(PMEM_DIR), pmem_max_size };
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+
+//Compare two same kind different type allocators
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindDifferentType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<char> alc2 { alloc_source_f1 };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+//Compare two same kind same type
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindSameType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f1 };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+//Compare two different kind different type
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindDifferentType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<char> alc2 { alloc_source_f2 };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+//Compare two different kind same type allocators
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindSameType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+//Copy assignment test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_CopyAssignment_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_TRUE(alc1 != alc2);
+    alc1 = alc2;
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+//Move constructor test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_MoveConstructor_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f2 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_TRUE(alc2 == alc1);
+    libmemkind::pmem::allocator<int> alc3 = std::move(alc1);
+    ASSERT_TRUE(alc2 == alc3);
+}
+
+//Move assignment test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_MoveAssignment_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_TRUE(alc1 != alc2);
+
+    {
+        libmemkind::pmem::allocator<int> alc3 { alc2 };
+        alc1 = std::move(alc3);
+    }
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+//Single allocation-deallocation test
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_Allocator_SingleAllocationDeallocation_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    int *created_object = alc1.allocate(1);
+    alc1.deallocate(created_object, 1);
+}
+
+//Single allocation-deallocation test conservative policy
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_Allocator_SingleAllocationDeallocation_TestConservative)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_conservative };
+    int *created_object = alc1.allocate(1);
+    alc1.deallocate(created_object, 1);
+}
+
+//Shared cross-allocation-deallocation test
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_Allocator_SharedAllocationDeallocation_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f1 };
+    int *created_object = nullptr;
+    int *created_object_2 = nullptr;
+    created_object = alc1.allocate(1);
+    ASSERT_TRUE(alc1 == alc2);
+    alc2.deallocate(created_object, 1);
+    created_object = alc2.allocate(1);
+    alc1.deallocate(created_object, 1);
+
+    created_object = alc1.allocate(1);
+    created_object_2 = alc2.allocate(1);
+    alc2.deallocate(created_object, 1);
+    alc1.deallocate(created_object_2, 1);
+}
+
+//Construct-destroy test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_ConstructDestroy_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    int *created_object = alc1.allocate(1);
+    alc1.construct(created_object);
+    alc1.destroy(created_object);
+}
+
+//Testing thread-safety support
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_MultithreadingSupport_Test)
+{
+    const size_t num_threads = 1000;
+    const size_t iteration_count = 5000;
+
+    std::vector<std::thread> thds;
+    for (size_t i = 0; i < num_threads; ++i) {
+
+        thds.push_back(std::thread( [&]() {
+            std::vector<libmemkind::pmem::allocator<int>> allocators_local;
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.push_back(libmemkind::pmem::allocator<int> ( alloc_source ));
+            }
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.pop_back();
+            }
+        }));
+    }
+
+    for (size_t i = 0; i < num_threads; ++i) {
+        thds[i].join();
+    }
+}
+
+//Test multiple memkind allocator usage
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_MultipleAllocatorUsage_Test)
+{
+    {
+        libmemkind::pmem::allocator<int> alloc { PMEM_DIR, pmem_max_size } ;
+    }
+
+    {
+        libmemkind::pmem::allocator<int> alloc { PMEM_DIR, pmem_max_size } ;
+    }
+}
+
+//Test vector
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Vector_Test)
+{
+    libmemkind::pmem::allocator<int> alc{ alloc_source  };
+    std::vector<int, libmemkind::pmem::allocator<int>> vector{ alc };
+
+    const int num = 20;
+
+    for (int i = 0; i < num; ++i) {
+        vector.push_back(0xDEAD + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        ASSERT_TRUE(vector[i] == 0xDEAD + i);
+    }
+}
+
+//Test list
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_List_Test)
+{
+    libmemkind::pmem::allocator<int> alc{ alloc_source };
+
+    std::list<int, libmemkind::pmem::allocator<int>> list{ alc };
+
+    const int num = 4;
+
+    for (int i = 0; i < num; ++i) {
+        list.emplace_back(0xBEAC011 + i);
+        ASSERT_TRUE(list.back() == 0xBEAC011 + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        list.pop_back();
+    }
+}
+
+//Test map
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Map_Test)
+{
+    libmemkind::pmem::allocator<std::pair<const std::string, std::string>> alc{ alloc_source };
+
+    std::map<std::string, std::string, std::less<std::string>, libmemkind::pmem::allocator<std::pair<const std::string, std::string>>>
+    map{ std::less<std::string>(), alc };
+
+    const int num = 10;
+
+    for (int i = 0; i < num; ++i) {
+        map[std::to_string(i)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string(i)] == std::to_string(0x0CEA11 + i));
+        map[std::to_string((i * 997 + 83) % 113)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string((i * 997 + 83) % 113)] == std::to_string(
+                        0x0CEA11 + i));
+    }
+}
+
+#if _GLIBCXX_USE_CXX11_ABI
+//Test vector of strings
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_VectorOfString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::pmem::allocator<char>>
+            pmem_string;
+    typedef libmemkind::pmem::allocator<pmem_string> pmem_alloc;
+
+    pmem_alloc alc{ alloc_source };
+    libmemkind::pmem::allocator<char> st_alc{alc};
+
+    std::vector<pmem_string,std::scoped_allocator_adaptor<pmem_alloc>> vec{ alc };
+    pmem_string arg{ "Very very loooong striiiing", st_alc };
+
+    vec.push_back(arg);
+    ASSERT_TRUE(vec.back() == arg);
+}
+
+//Test map of int strings
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorScopedUsage_MapOfIntString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::pmem::allocator<char>>
+            pmem_string;
+    typedef int key_t;
+    typedef pmem_string value_t;
+    typedef std::pair<const key_t, value_t> target_pair;
+    typedef libmemkind::pmem::allocator<target_pair> pmem_alloc;
+    typedef libmemkind::pmem::allocator<char> str_allocator_t;
+    typedef std::map<key_t, value_t, std::less<key_t>, std::scoped_allocator_adaptor<pmem_alloc>>
+            map_t;
+
+    pmem_alloc map_allocator( alloc_source );
+
+    str_allocator_t str_allocator( map_allocator );
+
+    value_t source_str1("Lorem ipsum dolor ", str_allocator);
+    value_t source_str2("sit amet consectetuer adipiscing elit", str_allocator );
+
+    map_t target_map{ std::scoped_allocator_adaptor<pmem_alloc>(map_allocator) };
+
+    target_map[key_t(165)] = source_str1;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str1);
+    target_map[key_t(165)] = source_str2;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str2);
+}
+#endif // _GLIBCXX_USE_CXX11_ABI
diff --git a/memkind-1.10.0/test/proc_stat.h b/memkind-1.10.0/test/proc_stat.h
new file mode 100644
index 0000000..79f8530
--- /dev/null
+++ b/memkind-1.10.0/test/proc_stat.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (C) 2017 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include <cstring>
+#include <fstream>
+
+class ProcStat
+{
+public:
+    size_t get_virtual_memory_size_bytes()
+    {
+        get_stat("VmSize", str_value);
+        return strtol(str_value, NULL, 10) * 1024;
+    }
+
+    size_t get_physical_memory_size_bytes()
+    {
+        get_stat("VmRSS", str_value);
+        return strtol(str_value, NULL, 10) * 1024;
+    }
+
+    size_t get_used_swap_space_size_bytes(void)
+    {
+        get_stat("VmSwap", str_value);
+        return strtol(str_value, NULL, 10) * 1024;
+    }
+private:
+    /* We are avoiding to allocate local buffers,
+     * since it can produce noise in memory footprint tests.
+     */
+    char line[1024];
+    char current_entry_name[1024];
+    char str_value[1024];
+
+    // Note: this function is not thread-safe.
+    void get_stat(const char *field_name, char *value)
+    {
+        char *pos = nullptr;
+        std::ifstream file("/proc/self/status", std::ifstream::in);
+        if (file.is_open()) {
+            while (file.getline(line, sizeof(line))) {
+                pos = strstr(line, field_name);
+                if (pos) {
+                    sscanf(pos, "%64[a-zA-Z_0-9()]: %s", current_entry_name, value);
+                    break;
+                }
+            }
+            file.close();
+        }
+    }
+};
+
diff --git a/memkind-1.10.0/test/python_framework/__init__.py b/memkind-1.10.0/test/python_framework/__init__.py
new file mode 100644
index 0000000..bf66a99
--- /dev/null
+++ b/memkind-1.10.0/test/python_framework/__init__.py
@@ -0,0 +1,26 @@
+#
+#  Copyright (C) 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+from python_framework.cmd_helper import CMD_helper
+from python_framework.huge_page_organizer import Huge_page_organizer
diff --git a/memkind-1.10.0/test/python_framework/cmd_helper.py b/memkind-1.10.0/test/python_framework/cmd_helper.py
new file mode 100644
index 0000000..68a869d
--- /dev/null
+++ b/memkind-1.10.0/test/python_framework/cmd_helper.py
@@ -0,0 +1,55 @@
+#
+#  Copyright (C) 2017 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+import pytest
+import os
+import tempfile
+import subprocess
+
+class CMD_helper(object):
+
+    def execute_cmd(self, command, sudo=False):
+        if sudo:
+            command = "sudo {0}".format(command)
+        #Initialize temp file for stdout. Will be removed when closed.
+        outfile = tempfile.SpooledTemporaryFile()
+        try:
+            #Invoke process
+            p = subprocess.Popen(command, stdout=outfile, stderr=subprocess.STDOUT, shell=True)
+            p.communicate()
+            #Read stdout from file
+            outfile.seek(0)
+            stdout = outfile.read()
+        except:
+            raise
+        finally:
+            #Make sure the file is closed
+            outfile.close()
+        retcode = p.returncode
+        return stdout, retcode
+
+    def get_command_path(self, binary):
+        """Get the path to the binary."""
+        path = os.path.dirname(os.path.abspath(__file__))
+        return os.path.join(path, binary)
diff --git a/memkind-1.10.0/test/python_framework/huge_page_organizer.py b/memkind-1.10.0/test/python_framework/huge_page_organizer.py
new file mode 100644
index 0000000..6603551
--- /dev/null
+++ b/memkind-1.10.0/test/python_framework/huge_page_organizer.py
@@ -0,0 +1,74 @@
+#
+#  Copyright (C) 2017 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+from cmd_helper import CMD_helper
+import os
+import itertools
+
+"""
+Huge_page_organizer sets hugepages per NUMA node and restores initial setup of hugepages.
+It writes and reads from the same file, so using Huge_page_organizer with parallel
+execution may cause undefined behaviour.
+"""
+class Huge_page_organizer(object):
+
+    cmd_helper = CMD_helper()
+    path = "/sys/devices/system/node/node{0}/hugepages/hugepages-2048kB/nr_hugepages"
+    restore_values = []
+
+    def __restore(self):
+        """Restore initial setup of hugepages."""
+        for node_id, restore_value in enumerate(self.restore_values):
+            self.__set_nr_hugepages(node_id, restore_value)
+
+    def __get_nr_hugepages(self, node_id):
+        """Return number of hugepages on given node or None if given node is not configured."""
+        if not os.path.isfile(self.path.format(node_id)):
+            return None
+        with open(self.path.format(node_id), "r") as f:
+            return int(f.read())
+
+    def __set_nr_hugepages(self, node_id, nr_hugepages):
+        """Set hugepages on given node to given number. Return True on success, False otherwise."""
+        command = 'sh -c "echo {0} >> {1}"'.format(nr_hugepages, self.path.format(node_id))
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=True)
+        if retcode:
+            return False
+        return self.__get_nr_hugepages(node_id) is nr_hugepages
+
+    def __init__(self, nr_hugepages_per_node):
+        """Save current hugepages setup and set hugepages per NUMA node."""
+        for node_id in itertools.count():
+            nr_hugepages = self.__get_nr_hugepages(node_id)
+            if nr_hugepages is None:
+                break
+            self.restore_values.append(nr_hugepages)
+            retcode = self.__set_nr_hugepages(node_id, nr_hugepages_per_node)
+            if not retcode:
+               self.__restore()
+            assert retcode, "Error: Could not set the requested amount of hugepages."
+
+    def __del__(self):
+        """Call __restore() function."""
+        self.__restore()
diff --git a/memkind-1.10.0/test/random_sizes_allocator.h b/memkind-1.10.0/test/random_sizes_allocator.h
new file mode 100644
index 0000000..0c96fb9
--- /dev/null
+++ b/memkind-1.10.0/test/random_sizes_allocator.h
@@ -0,0 +1,76 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include "memory_manager.h"
+#include <vector>
+#include <random>
+
+class RandomSizesAllocator
+{
+private:
+    std::vector<MemoryManager> allocated_memory;
+    memkind_t kind;
+    std::default_random_engine generator;
+    std::uniform_int_distribution<int> memory_distribution;
+
+    size_t get_random_size()
+    {
+        return memory_distribution(generator);
+    }
+
+public:
+    RandomSizesAllocator(memkind_t kind, size_t min_size, size_t max_size,
+                         int max_allocations_number) :
+        kind(kind),
+        memory_distribution(min_size, max_size)
+    {
+        allocated_memory.reserve(max_allocations_number);
+    }
+
+    size_t malloc_random_memory()
+    {
+        size_t size = get_random_size();
+        allocated_memory.emplace_back(kind, size);
+        return size;
+    }
+
+    size_t free_random_memory()
+    {
+        if (empty())
+            return 0;
+        std::uniform_int_distribution<int> distribution(0, allocated_memory.size() - 1);
+        int random_index = distribution(generator);
+        auto it = std::begin(allocated_memory) + random_index;
+        size_t size = it->size();
+        allocated_memory.erase(it);
+        return size;
+    }
+
+    bool empty()
+    {
+        return allocated_memory.empty();
+    }
+};
+
diff --git a/memkind-1.10.0/test/run_alloc_benchmark.sh b/memkind-1.10.0/test/run_alloc_benchmark.sh
new file mode 100644
index 0000000..ef85a9e
--- /dev/null
+++ b/memkind-1.10.0/test/run_alloc_benchmark.sh
@@ -0,0 +1,60 @@
+#!/bin/bash
+#
+#  Copyright (C) 2014 - 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+# Allocator
+ALLOCATOR="glibc tbb hbw pmem"
+# Thread configuration
+THREADS=(1 2 4 8 16 18 36)
+# Memory configuration (in kB) / iterations
+MEMORY=(1 4 16 64 256 1024 4096 16384)
+# Iterations
+ITERS=1000
+
+export KMP_AFFINITY=scatter,granularity=fine
+
+# For each algorithm
+for alloc in $ALLOCATOR
+do
+    rm -f alloctest_$alloc.txt
+    echo "# Number of threads, allocation size [kB], average malloc and free time [ms], average allocation time [ms], \
+average free time [ms], first allocation time [ms], first free time [ms]" >> alloctest_$alloc.txt
+    # For each number of threads
+    for nthr in ${THREADS[*]}
+    do
+        # For each amount of memory
+        for mem in ${MEMORY[*]}
+        do
+            echo "OMP_NUM_THREADS=$nthr ./alloc_benchmark_$alloc $ITERS $mem >> alloctest_$alloc.txt"
+            OMP_NUM_THREADS=$nthr ./alloc_benchmark_$alloc $ITERS $mem >> alloctest_$alloc.txt
+            ret=$?
+            if [ $ret -ne 0 ]; then
+                echo "Error: alloc_benchmark_$alloc returned $ret"
+                exit
+            fi
+        done
+    done
+done
+
+echo "Data collected. You can draw performance plots using python draw_plots.py."
diff --git a/memkind-1.10.0/test/static_kinds_list.h b/memkind-1.10.0/test/static_kinds_list.h
new file mode 100644
index 0000000..f43b7e2
--- /dev/null
+++ b/memkind-1.10.0/test/static_kinds_list.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <memkind.h>
+
+
+static memkind_t static_kinds_list[] = {
+    MEMKIND_DEFAULT,
+    MEMKIND_HBW,
+    MEMKIND_HBW_HUGETLB,
+    MEMKIND_HBW_PREFERRED,
+    MEMKIND_HBW_PREFERRED_HUGETLB,
+    MEMKIND_HUGETLB,
+    MEMKIND_HBW_GBTLB,
+    MEMKIND_HBW_PREFERRED_GBTLB,
+    MEMKIND_GBTLB,
+    MEMKIND_HBW_INTERLEAVE,
+    MEMKIND_INTERLEAVE,
+    MEMKIND_DAX_KMEM,
+    MEMKIND_DAX_KMEM_ALL,
+    MEMKIND_DAX_KMEM_PREFERRED
+};
+
+
diff --git a/memkind-1.10.0/test/static_kinds_tests.cpp b/memkind-1.10.0/test/static_kinds_tests.cpp
new file mode 100644
index 0000000..947aabd
--- /dev/null
+++ b/memkind-1.10.0/test/static_kinds_tests.cpp
@@ -0,0 +1,59 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "memkind/internal/memkind_private.h"
+
+#include "common.h"
+#include "static_kinds_list.h"
+
+/*
+ * Set of tests for checking if static kinds meets non-trivial assumptions
+ */
+
+class StaticKindsTest: public :: testing::Test
+{
+public:
+
+protected:
+    void SetUp()
+    {
+    }
+};
+
+
+/*
+ * Assumption: all static kinds should implement init_once operation
+ * Reason:  init_once should perform memkind_register (and other initialization if needed)
+ *          we are also using that fact to optimize initialization on first use (in memkind_malloc etc.)
+ */
+TEST_F(StaticKindsTest, test_TC_MEMKIND_STATIC_KINDS_INIT_ONCE)
+{
+    for(size_t i=0; i<(sizeof(static_kinds_list)/sizeof(static_kinds_list[0]));
+        i++) {
+        ASSERT_TRUE(static_kinds_list[i]->ops->init_once != NULL) <<
+                                                                  static_kinds_list[i]->name << " does not implement init_once operation!";
+    }
+}
+
diff --git a/memkind-1.10.0/test/tbbmalloc.h b/memkind-1.10.0/test/tbbmalloc.h
new file mode 100644
index 0000000..f52530b
--- /dev/null
+++ b/memkind-1.10.0/test/tbbmalloc.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright (C) 2016 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <dlfcn.h>
+
+extern void *(*scalable_malloc)(size_t);
+extern void *(*scalable_realloc)(void *, size_t);
+extern void *(*scalable_calloc)(size_t, size_t);
+extern void  (*scalable_free)(void *);
+
+int load_tbbmalloc_symbols();
diff --git a/memkind-1.10.0/test/test.sh b/memkind-1.10.0/test/test.sh
new file mode 100755
index 0000000..4c8e052
--- /dev/null
+++ b/memkind-1.10.0/test/test.sh
@@ -0,0 +1,384 @@
+#!/bin/bash
+#
+#  Copyright (C) 2014 - 2020 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+basedir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+PROGNAME=`basename $0`
+
+# Default path (to installation dir of memkind-tests RPM)
+TEST_PATH="$basedir/"
+
+# Gtest binaries executed by Berta
+GTEST_BINARIES=(all_tests decorator_test allocator_perf_tool_tests gb_page_tests_bind_policy memkind_stat_test defrag_allocate)
+
+# Pytest files executed by Berta
+PYTEST_FILES=(hbw_detection_test.py autohbw_test.py trace_mechanism_test.py)
+
+red=`tput setaf 1`
+green=`tput setaf 2`
+yellow=`tput setaf 3`
+default=`tput sgr0`
+
+err=0
+
+function usage () {
+   cat <<EOF
+
+Usage: $PROGNAME [-c csv_file] [-l log_file] [-f test_filter] [-T tests_dir] [-d] [-m] [-p] [-e] [-x] [-s] [-h]
+
+OPTIONS
+    -c,
+        path to CSV file
+    -l,
+        path to LOG file
+    -f,
+        test filter
+    -T,
+        path to tests directory
+    -d,
+        skip high bandwidth memory nodes detection tests
+    -m,
+        skip tests that require 2MB pages configured on the machine
+    -p,
+        skip python tests
+    -e,
+        run only HBW performance operation tests
+    -x,
+        skip tests that are passed as value
+    -s,
+        run disabled test (e.g. pmem long time stress test)
+    -h,
+        parameter added to display script usage
+EOF
+}
+
+function emit() {
+    if [ "$LOG_FILE" != "" ]; then
+        echo "$@" 2>&1 | tee -a $LOG_FILE;
+    else
+        echo "$@"
+    fi
+}
+
+function normalize_path {
+    local PATH=$1
+    if [ ! -d $PATH ];
+    then
+        echo "Not a directory: '$PATH'"
+        usage
+        exit 1
+    fi
+    if [[ $PATH != /* ]]; then
+        PATH=`pwd`/$PATH
+    fi
+    echo $PATH
+}
+
+function show_skipped_tests()
+{
+    SKIP_PATTERN=$1
+    DEFAULT_IFS=$IFS
+
+    # Search for gtests that match given pattern
+    for i in ${!GTEST_BINARIES[*]}; do
+        GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+        for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+            if [[ $LINE == *. ]]; then
+                TEST_SUITE=$LINE;
+            else
+                if [[ "$TEST_SUITE$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    # Search for pytests that match given pattern
+    for i in ${!PYTEST_FILES[*]}; do
+        PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+        IFS=$'\n'
+        for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+            if [[ $LINE == *"<Class "* ]]; then
+                TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            elif [[ $LINE == *"<Function "* ]]; then
+                LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+                if [[ "$TEST_SUITE.$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE.$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    IFS=$DEFAULT_IFS
+    emit ""
+}
+
+function execute_gtest()
+{
+    ret_val=1
+    TESTCMD=$1
+    TEST=$2
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST""$SKIPPED_GTESTS""$RUN_DISABLED_GTEST")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*OK ].*'
+    PATFAILED='.*FAILED  ].*'
+        PATSKIPPED='.*PASSED  ] 0.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST,${green}PASSED${default}"
+        ret_val=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        RESULT="$TEST,${yellow}SKIPPED${default}"
+        ret_val=0
+    else
+        RESULT="$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret_val
+}
+
+function execute_pytest()
+{
+    ret=1
+    TESTCMD=$1
+    TEST_SUITE=$2
+    TEST=$3
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST_SUITE.$TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST$SKIPPED_PYTESTS")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*1 passed.*'
+    PATFAILED='.*1 failed.*'
+    PATSKIPPED='.*deselected.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST_SUITE.$TEST,${green}PASSED${default}"
+        ret=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST_SUITE.$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        return 0
+    else
+        RESULT="$TEST_SUITE.$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret
+}
+
+#Check support for numa nodes (at least two)
+function check_numa()
+{
+    numactl --hardware | grep "^node 1" > /dev/null
+    if [ $? -ne 0 ]; then
+        echo "ERROR: $0 requires a NUMA enabled system with more than one node."
+        exit 1
+    fi
+}
+
+#Check support for High Bandwidth Memory - simulate one if no one was found
+function check_hbw_nodes()
+{
+    if [ ! -f /usr/bin/memkind-hbw-nodes ]; then
+        if [ -x ./memkind-hbw-nodes ]; then
+            export PATH=$PATH:$PWD
+        else
+            echo "Cannot find 'memkind-hbw-nodes' in $PWD. Did you run 'make'?"
+            exit 1
+        fi
+    fi
+
+    ret=$(memkind-hbw-nodes)
+    if [[ $ret == "" ]]; then
+        export MEMKIND_HBW_NODES=1
+        TEST_PREFIX="numactl --membind=0 --cpunodebind=$MEMKIND_HBW_NODES %s"
+    fi
+}
+
+#Check automatic support for persistent memory NUMA node - simulate one if no one was found
+function check_auto_dax_kmem_nodes()
+{
+    if [ ! -f /usr/bin/memkind-auto-dax-kmem-nodes ]; then
+        if [ -x ./memkind-auto-dax-kmem-nodes ]; then
+            export PATH=$PATH:$PWD
+        else
+            echo "Cannot find 'memkind-auto-dax-kmem-nodes' in $PWD. Did you run 'make'?"
+            exit 1
+        fi
+    fi
+
+    ret=$(memkind-auto-dax-kmem-nodes)
+    if [[ $ret == "" ]]; then
+        export MEMKIND_DAX_KMEM_NODES=1
+    fi
+}
+
+#begin of main script
+
+check_numa
+
+check_hbw_nodes
+check_auto_dax_kmem_nodes
+
+OPTIND=1
+
+while getopts "T:c:f:l:hdemsx:p:" opt; do
+    case "$opt" in
+        T)
+            TEST_PATH=$OPTARG;
+            ;;
+        c)
+            CSV=$OPTARG;
+            ;;
+        f)
+            TEST_FILTER=$OPTARG;
+            ;;
+        l)
+            LOG_FILE=$OPTARG;
+            ;;
+        m)
+            echo "Skipping tests that require 2MB pages due to unsatisfactory system conditions"
+            if [[ "$SKIPPED_GTESTS" == "" ]]; then
+                SKIPPED_GTESTS=":-*test_TC_MEMKIND_2MBPages_*"
+            else
+                SKIPPED_GTESTS=$SKIPPED_GTESTS":*test_TC_MEMKIND_2MBPages_*"
+            fi
+            if [[ "$SKIPPED_PYTESTS" == "" ]]; then
+                SKIPPED_PYTESTS=" and not test_TC_MEMKIND_2MBPages_"
+            else
+                SKIPPED_PYTESTS=$SKIPPED_PYTESTS" and not test_TC_MEMKIND_2MBPages_"
+            fi
+            show_skipped_tests "test_TC_MEMKIND_2MBPages_"
+            ;;
+        d)
+            echo "Skipping tests that detect high bandwidth memory nodes due to unsatisfactory system conditions"
+            if [[ $SKIPPED_PYTESTS == "" ]]; then
+                SKIPPED_PYTESTS=" and not hbw_detection"
+            else
+                SKIPPED_PYTESTS=$SKIPPED_PYTESTS" and not hbw_detection"
+            fi
+            show_skipped_tests "test_TC_MEMKIND_hbw_detection"
+            ;;
+        e)
+            GTEST_BINARIES=(performance_test)
+            SKIPPED_PYTESTS=$SKIPPED_PYTESTS$PYTEST_FILES
+            break;
+            ;;
+        p)
+            SKIPPED_PYTESTS=$SKIPPED_PYTESTS$OPTARG
+            show_skipped_tests "$OPTARG"
+            break;
+            ;;
+        x)
+            echo "Skipping some tests on demand '$OPTARG'"
+            if [[ $SKIPPED_GTESTS == "" ]]; then
+                SKIPPED_GTESTS=":-"$OPTARG
+            else
+                SKIPPED_GTESTS=$SKIPPED_GTESTS":"$OPTARG
+            fi
+            show_skipped_tests "$OPTARG"
+            ;;
+        s)
+            echo "Run also disabled tests"
+            RUN_DISABLED_GTEST=" --gtest_also_run_disabled_tests"
+            ;;
+        h)
+            usage;
+            exit 0;
+            ;;
+    esac
+done
+
+TEST_PATH=`normalize_path "$TEST_PATH"`
+
+# Clear any remnants of previous execution(s)
+rm -rf $CSV
+rm -rf $LOG_FILE
+
+# Run tests written in gtest
+for i in ${!GTEST_BINARIES[*]}; do
+    GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+    emit
+    emit "### Processing gtest binary '$GTEST_BINARY_PATH' ###"
+    for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+        if [[ $LINE == *. ]]; then
+            TEST_SUITE=$LINE;
+        else
+            TEST_CMD="$GTEST_BINARY_PATH --gtest_filter=%s 2>&1"
+            execute_gtest "$TEST_CMD" "$TEST_SUITE$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+# Run tests written in pytest
+for i in ${!PYTEST_FILES[*]}; do
+    PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+    emit
+    emit "### Processing pytest file '$PTEST_BINARY_PATH' ###"
+    IFS=$'\n'
+    for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+        if [[ $LINE == *"<Class "* ]]; then
+            TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+        elif [[ $LINE == *"<Function "* ]]; then
+            LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            TEST_CMD="py.test $PTEST_BINARY_PATH -k='%s' 2>&1"
+            execute_pytest "$TEST_CMD" "$TEST_SUITE" "$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+exit $err
diff --git a/memkind-1.10.0/test/test_dax_kmem.sh b/memkind-1.10.0/test/test_dax_kmem.sh
new file mode 100755
index 0000000..a2e2455
--- /dev/null
+++ b/memkind-1.10.0/test/test_dax_kmem.sh
@@ -0,0 +1,287 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 - 2020 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+basedir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+PROGNAME=`basename $0`
+
+# Default path (to installation dir of memkind-tests RPM)
+TEST_PATH="$basedir/"
+
+# Gtest binaries
+GTEST_BINARIES=(dax_kmem_test)
+
+# Pytest files
+PYTEST_FILES=(dax_kmem_env_var_test.py)
+
+red=`tput setaf 1`
+green=`tput setaf 2`
+yellow=`tput setaf 3`
+default=`tput sgr0`
+
+err=0
+
+function usage () {
+   cat <<EOF
+
+Usage: $PROGNAME [-h]
+
+OPTIONS
+    -h,
+        display script usage
+EOF
+}
+
+function emit() {
+    if [ "$LOG_FILE" != "" ]; then
+        echo "$@" 2>&1 | tee -a $LOG_FILE;
+    else
+        echo "$@"
+    fi
+}
+
+function normalize_path {
+    local PATH=$1
+    if [ ! -d $PATH ];
+    then
+        echo "Not a directory: '$PATH'"
+        usage
+        exit 1
+    fi
+    if [[ $PATH != /* ]]; then
+        PATH=`pwd`/$PATH
+    fi
+    echo $PATH
+}
+
+function show_skipped_tests()
+{
+    SKIP_PATTERN=$1
+    DEFAULT_IFS=$IFS
+
+    # Search for gtests that match given pattern
+    for i in ${!GTEST_BINARIES[*]}; do
+        GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+        for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+            if [[ $LINE == *. ]]; then
+                TEST_SUITE=$LINE;
+            else
+                if [[ "$TEST_SUITE$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    # Search for pytests that match given pattern
+    for i in ${!PYTEST_FILES[*]}; do
+        PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+        IFS=$'\n'
+        for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+            if [[ $LINE == *"<Class "* ]]; then
+                TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            elif [[ $LINE == *"<Function "* ]]; then
+                LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+                if [[ "$TEST_SUITE.$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE.$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    IFS=$DEFAULT_IFS
+    emit ""
+}
+
+function execute_gtest()
+{
+    ret_val=1
+    TESTCMD=$1
+    TEST=$2
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST""$SKIPPED_GTESTS""$RUN_DISABLED_GTEST")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*OK ].*'
+    PATFAILED='.*FAILED  ].*'
+        PATSKIPPED='.*PASSED  ] 0.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST,${green}PASSED${default}"
+        ret_val=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        RESULT="$TEST,${yellow}SKIPPED${default}"
+        ret_val=0
+    else
+        RESULT="$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret_val
+}
+
+function execute_pytest()
+{
+    ret=1
+    TESTCMD=$1
+    TEST_SUITE=$2
+    TEST=$3
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST_SUITE.$TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST$SKIPPED_PYTESTS")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*1 passed.*'
+    PATFAILED='.*1 failed.*'
+    PATSKIPPED='.*deselected.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST_SUITE.$TEST,${green}PASSED${default}"
+        ret=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST_SUITE.$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        return 0
+    else
+        RESULT="$TEST_SUITE.$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret
+}
+
+#Check support for numa nodes (at least two)
+function check_numa()
+{
+    numactl --hardware | grep "^node 1" > /dev/null
+    if [ $? -ne 0 ]; then
+        echo "ERROR: $0 requires a NUMA enabled system with more than one node."
+        exit 1
+    fi
+}
+
+#Check automatic support for persistent memory NUMA node - simulate one if no one was found
+function check_auto_dax_kmem_nodes()
+{
+    if [ ! -f /usr/bin/memkind-auto-dax-kmem-nodes ]; then
+        if [ -x ./memkind-auto-dax-kmem-nodes ]; then
+            export PATH=$PATH:$PWD
+        else
+            echo "Cannot find 'memkind-auto-dax-kmem-nodes' in $PWD. Did you run 'make'?"
+            exit 1
+        fi
+    fi
+
+    ret=$(memkind-auto-dax-kmem-nodes)
+    emit "The binary memkind-auto-dax-kmem-nodes returned code $ret."
+    if [[ $ret == "" ]]; then
+        export MEMKIND_DAX_KMEM_NODES=1
+    fi
+}
+
+#begin of main script
+
+check_numa
+
+check_auto_dax_kmem_nodes
+
+OPTIND=1
+
+while getopts ":h" opt; do
+    case "$opt" in
+        h)
+            usage;
+            exit 0;
+            ;;
+    esac
+done
+
+TEST_PATH=`normalize_path "$TEST_PATH"`
+
+# Clear any remnants of previous execution(s)
+rm -rf $CSV
+rm -rf $LOG_FILE
+
+# Run tests written in gtest
+for i in ${!GTEST_BINARIES[*]}; do
+    GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+    emit
+    emit "### Processing gtest binary '$GTEST_BINARY_PATH' ###"
+    for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+        if [[ $LINE == *. ]]; then
+            TEST_SUITE=$LINE;
+        else
+            TEST_CMD="$GTEST_BINARY_PATH --gtest_filter=%s 2>&1"
+            execute_gtest "$TEST_CMD" "$TEST_SUITE$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+# Run tests written in pytest
+for i in ${!PYTEST_FILES[*]}; do
+    PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+    emit
+    emit "### Processing pytest file '$PTEST_BINARY_PATH' ###"
+    IFS=$'\n'
+    for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+        if [[ $LINE == *"<Class "* ]]; then
+            TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+        elif [[ $LINE == *"<Function "* ]]; then
+            LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            TEST_CMD="py.test $PTEST_BINARY_PATH -k='%s' 2>&1"
+            execute_pytest "$TEST_CMD" "$TEST_SUITE" "$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+exit $err
diff --git a/memkind-1.10.0/test/trace_mechanism_test.py b/memkind-1.10.0/test/trace_mechanism_test.py
new file mode 100755
index 0000000..dba7ffd
--- /dev/null
+++ b/memkind-1.10.0/test/trace_mechanism_test.py
@@ -0,0 +1,62 @@
+#
+#  Copyright (C) 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+import pytest
+import os
+import tempfile
+import subprocess
+from python_framework import CMD_helper
+from python_framework import Huge_page_organizer
+
+class Test_trace_mechanism(object):
+    binary = "../trace_mechanism_test_helper"
+    fail_msg = "Test failed with:\n {0}"
+    debug_env = "MEMKIND_DEBUG=1 "
+    cmd_helper = CMD_helper()
+
+    def test_TC_MEMKIND_logging_MEMKIND_HBW(self):
+        #This test executes trace_mechanism_test_helper and test if MEMKIND_INFO message occurs while calling MEMKIND_HBW
+        command = self.debug_env + self.cmd_helper.get_command_path(self.binary) + " MEMKIND_HBW"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: trace_mechanism_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert "MEMKIND_INFO: NUMA node" in output, self.fail_msg.format("\nError: trace mechanism in memkind doesn't show MEMKIND_INFO message \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_2MBPages_logging_MEMKIND_HUGETLB(self):
+        huge_page_organizer = Huge_page_organizer(8)
+        #This test executes trace_mechanism_test_helper and test if MEMKIND_INFO message occurs while calling MEMKIND_HUGETLB
+        command = self.debug_env + self.cmd_helper.get_command_path(self.binary) + " MEMKIND_HUGETLB"
+        print "Executing command: {0}".format(command)
+        output, retcode= self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: trace_mechanism_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert "MEMKIND_INFO: Number of" in output, self.fail_msg.format("\nError: trace mechanism in memkind doesn't show MEMKIND_INFO message \noutput: {0}").format(output)
+        assert "MEMKIND_INFO: Overcommit limit for" in output, self.fail_msg.format("\nError: trace mechanism in memkind doesn't show MEMKIND_INFO message \n output: {0}").format(output)
+
+    def test_TC_MEMKIND_logging_negative_MEMKIND_DEBUG_env(self):
+        #This test executes trace_mechanism_test_helper and test if setting MEMKIND_DEBUG to wrong value causes MEMKIND_WARNING message
+        command = "MEMKIND_DEBUG=-1 " + self.cmd_helper.get_command_path(self.binary) + " MEMKIND_HBW"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: trace_mechanism_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert "MEMKIND_WARNING: debug option" in output, self.fail_msg.format("\nError: setting wrong MEMKIND_DEBUG environment variable doesn't show MEMKIND_WARNING \noutput: {0})").format(output)
diff --git a/memkind-1.10.0/test/trace_mechanism_test_helper.c b/memkind-1.10.0/test/trace_mechanism_test_helper.c
new file mode 100644
index 0000000..bbbe9d8
--- /dev/null
+++ b/memkind-1.10.0/test/trace_mechanism_test_helper.c
@@ -0,0 +1,63 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "memkind.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+int main(int argc, char *argv[])
+{
+    int err = 0;
+    const size_t size = 1024 * 1024;
+    void *buf = NULL;
+
+    //It is expected that "MEMKIND_HBW" or "MEMKIND_HUGETLB" argument is passed
+    if (argc != 2) {
+        printf("Error: Wrong number of parameters\n");
+        err = -1;
+        return err;
+    }
+
+    if (strcmp(argv[1], "MEMKIND_HBW") == 0) {
+        buf = memkind_malloc(MEMKIND_HBW, size);
+        if (buf == NULL) {
+            printf("Allocation of MEMKIND_HBW failed\n");
+            err = -1;
+        }
+        memkind_free(MEMKIND_HBW, buf);
+    } else if (strcmp(argv[1], "MEMKIND_HUGETLB") == 0) {
+        buf = memkind_malloc(MEMKIND_HUGETLB, size);
+        if (buf == NULL) {
+            printf("Allocation of MEMKIND_HUGETLB failed\n");
+            err = -1;
+        }
+        memkind_free(MEMKIND_HUGETLB, buf);
+    } else {
+        printf("Error: unknown parameter\n");
+        err = -1;
+    }
+
+    return err;
+}
diff --git a/memkind-1.10.0/test/trial_generator.cpp b/memkind-1.10.0/test/trial_generator.cpp
new file mode 100644
index 0000000..0a4a349
--- /dev/null
+++ b/memkind-1.10.0/test/trial_generator.cpp
@@ -0,0 +1,325 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+#include "trial_generator.h"
+#include "check.h"
+#include <vector>
+#include <numa.h>
+#include <numaif.h>
+
+trial_t TrialGenerator :: create_trial_tuple(alloc_api_t api,
+                                             size_t size,
+                                             size_t alignment,
+                                             int page_size,
+                                             memkind_t memkind,
+                                             int free_index)
+{
+    trial_t ltrial;
+    ltrial.api = api;
+    ltrial.size = size;
+    ltrial.alignment = alignment;
+    ltrial.page_size = page_size;
+    ltrial.memkind = memkind;
+    ltrial.free_index = free_index;
+    return ltrial;
+}
+
+
+void TrialGenerator :: generate_gb (alloc_api_t api, int number_of_gb_pages,
+                                    memkind_t memkind, alloc_api_t api_free, bool psize_strict, size_t align)
+{
+    std::vector<size_t> sizes_to_alloc;
+    //When API = HBW_MEMALIGN_PSIZE: psize is set to HBW_PAGESIZE_1GB_STRICT when allocation is a multiple of 1GB. Otherwise it is set to HBW_PAGESIZE_1GB.
+    for (int i=1; i <= number_of_gb_pages; i++) {
+        if (psize_strict || api!=HBW_MEMALIGN_PSIZE)
+            sizes_to_alloc.push_back(i*GB);
+        else
+            sizes_to_alloc.push_back(i*GB+1);
+    }
+    int k = 0;
+    trial_vec.clear();
+
+    for (int i = 0; i< (int)sizes_to_alloc.size(); i++) {
+        trial_vec.push_back(create_trial_tuple(api, sizes_to_alloc[i],
+                                               align, 2*MB,
+                                               memkind,
+                                               -1));
+        if (i > 0)
+            k++;
+        trial_vec.push_back(create_trial_tuple(api_free,0,0,0,
+                                               memkind,
+                                               k++));
+    }
+}
+
+int n_random(int i)
+{
+    return random() % i;
+}
+
+void TrialGenerator :: generate_size_2bytes_2KB_2MB(alloc_api_t api)
+{
+    size_t size[] = {2, 2*KB, 2*MB};
+
+    int k = 0;
+    trial_vec.clear();
+    for (unsigned int i = 0; i < (int)(sizeof(size)/sizeof(size[0]));
+         i++) {
+        trial_vec.push_back(
+            create_trial_tuple(
+                api,size[i],
+                32,
+                4096,
+                MEMKIND_HBW,
+                -1
+            )
+        );
+
+        if (i > 0) k++;
+        trial_vec.push_back(create_trial_tuple(HBW_FREE, 0, 0, 0,
+                                               MEMKIND_HBW, k));
+        k++;
+    }
+}
+
+void TrialGenerator :: print()
+{
+
+    std::vector<trial_t>:: iterator it;
+
+    std::cout <<"*********** Size: "<< trial_vec.size()
+              <<"********\n";
+    std::cout << "SIZE PSIZE ALIGN FREE KIND"<<std::endl;
+
+    for (it = trial_vec.begin();
+         it != trial_vec.end();
+         it++) {
+        std::cout << it->size <<" "
+                  << it->page_size <<" "
+                  << it->alignment <<" "
+                  << it->free_index <<" "
+                  << it->memkind <<" "
+                  <<std::endl;
+    }
+
+}
+
+
+void TrialGenerator :: run(int num_bandwidth, std::vector<int> &bandwidth)
+{
+
+    int num_trial = trial_vec.size();
+    int i, ret = 0;
+    void **ptr_vec = NULL;
+
+    ptr_vec = (void **) malloc (num_trial *
+                                sizeof (void *));
+    if (NULL == ptr_vec) {
+        fprintf (stderr, "Error in allocating ptr array\n");
+        exit(-1);
+    }
+
+    for (i = 0; i < num_trial; ++i) {
+        ptr_vec[i] = NULL;
+    }
+    for (i = 0; i < num_trial; ++i) {
+        switch(trial_vec[i].api) {
+            case HBW_FREE:
+                if (i == num_trial - 1 || trial_vec[i + 1].api != HBW_REALLOC) {
+                    hbw_free(ptr_vec[trial_vec[i].free_index]);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                    ptr_vec[i] = NULL;
+                } else {
+                    ptr_vec[i + 1] = hbw_realloc(ptr_vec[trial_vec[i].free_index],
+                                                 trial_vec[i + 1].size);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                }
+                break;
+            case MEMKIND_FREE:
+                if (i == num_trial - 1 || trial_vec[i + 1].api != MEMKIND_REALLOC) {
+                    memkind_free(trial_vec[i].memkind,
+                                 ptr_vec[trial_vec[i].free_index]);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                    ptr_vec[i] = NULL;
+                } else {
+                    ptr_vec[i + 1] = memkind_realloc(trial_vec[i].memkind,
+                                                     ptr_vec[trial_vec[i].free_index],
+                                                     trial_vec[i + 1].size);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                }
+                break;
+            case HBW_MALLOC:
+                fprintf (stdout,"Allocating %zd bytes using hbw_malloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = hbw_malloc(trial_vec[i].size);
+                break;
+            case HBW_CALLOC:
+                fprintf (stdout,"Allocating %zd bytes using hbw_calloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = hbw_calloc(trial_vec[i].size, 1);
+                break;
+            case HBW_REALLOC:
+                fprintf (stdout,"Allocating %zd bytes using hbw_realloc\n",
+                         trial_vec[i].size);
+                fflush(stdout);
+                if (NULL == ptr_vec[i]) {
+                    ptr_vec[i] = hbw_realloc(NULL, trial_vec[i].size);
+                }
+                break;
+            case HBW_MEMALIGN:
+                fprintf (stdout,"Allocating %zd bytes using hbw_memalign\n",
+                         trial_vec[i].size);
+                ret =  hbw_posix_memalign(&ptr_vec[i],
+                                          trial_vec[i].alignment,
+                                          trial_vec[i].size);
+                break;
+            case HBW_MEMALIGN_PSIZE:
+                fprintf (stdout,"Allocating %zd bytes using hbw_memalign_psize\n",
+                         trial_vec[i].size);
+                hbw_pagesize_t psize;
+                if (trial_vec[i].page_size == 4096)
+                    psize = HBW_PAGESIZE_4KB;
+                else if (trial_vec[i].page_size == 2097152)
+                    psize = HBW_PAGESIZE_2MB;
+                else if (trial_vec[i].size %
+                         trial_vec[i].page_size > 0)
+                    psize = HBW_PAGESIZE_1GB;
+                else
+                    psize = HBW_PAGESIZE_1GB_STRICT;
+
+                ret = hbw_posix_memalign_psize(&ptr_vec[i],
+                                               trial_vec[i].alignment,
+                                               trial_vec[i].size,
+                                               psize);
+
+                break;
+            case MEMKIND_MALLOC:
+                fprintf (stdout,"Allocating %zd bytes using memkind_malloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = memkind_malloc(trial_vec[i].memkind,
+                                            trial_vec[i].size);
+                break;
+            case MEMKIND_CALLOC:
+                fprintf (stdout,"Allocating %zd bytes using memkind_calloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = memkind_calloc(trial_vec[i].memkind,
+                                            trial_vec[i].size, 1);
+                break;
+            case MEMKIND_REALLOC:
+                fprintf (stdout,"Allocating %zd bytes using memkind_realloc\n",
+                         trial_vec[i].size);
+                if (NULL == ptr_vec[i]) {
+                    ptr_vec[i] = memkind_realloc(trial_vec[i].memkind,
+                                                 ptr_vec[i],
+                                                 trial_vec[i].size);
+                }
+                break;
+            case MEMKIND_POSIX_MEMALIGN:
+                fprintf (stdout,
+                         "Allocating %zd bytes using memkind_posix_memalign\n",
+                         trial_vec[i].size);
+
+                ret = memkind_posix_memalign(trial_vec[i].memkind,
+                                             &ptr_vec[i],
+                                             trial_vec[i].alignment,
+                                             trial_vec[i].size);
+                break;
+        }
+        if (trial_vec[i].api != HBW_FREE &&
+            trial_vec[i].api != MEMKIND_FREE &&
+            trial_vec[i].memkind != MEMKIND_DEFAULT) {
+            ASSERT_TRUE(ptr_vec[i] != NULL);
+            memset(ptr_vec[i], 0, trial_vec[i].size);
+            Check check(ptr_vec[i], trial_vec[i]);
+            if (trial_vec[i].api == HBW_CALLOC) {
+                EXPECT_EQ(0, check.check_zero());
+            }
+            if (trial_vec[i].api == HBW_MEMALIGN ||
+                trial_vec[i].api == HBW_MEMALIGN_PSIZE ||
+                trial_vec[i].api == MEMKIND_POSIX_MEMALIGN) {
+                EXPECT_EQ(0, check.check_align(trial_vec[i].alignment));
+                EXPECT_EQ(0, ret);
+            }
+            if (trial_vec[i].api == HBW_MEMALIGN_PSIZE ||
+                (trial_vec[i].api == MEMKIND_MALLOC &&
+                 (trial_vec[i].memkind == MEMKIND_HBW_HUGETLB ||
+                  trial_vec[i].memkind == MEMKIND_HBW_PREFERRED_HUGETLB))) {
+                EXPECT_EQ(0, check.check_page_size(trial_vec[i].page_size));
+            }
+        }
+    }
+    for (i = 0; i < num_trial; ++i) {
+        if (ptr_vec[i]) {
+            hbw_free(ptr_vec[i]);
+        }
+    }
+}
+
+void TGTest :: SetUp()
+{
+    size_t node;
+    char *hbw_nodes_env, *endptr;
+    tgen = std::unique_ptr<TrialGenerator>(new TrialGenerator());
+
+    hbw_nodes_env = secure_getenv("MEMKIND_HBW_NODES");
+    if (hbw_nodes_env) {
+        num_bandwidth = 128;
+        for (node = 0; node < num_bandwidth; node++) {
+            bandwidth.push_back(1);
+        }
+        node = strtol(hbw_nodes_env, &endptr, 10);
+        bandwidth.push_back(2);
+        while (*endptr == ':') {
+            hbw_nodes_env = endptr + 1;
+            node = strtol(hbw_nodes_env, &endptr, 10);
+            if (endptr != hbw_nodes_env && node >= 0 && node < num_bandwidth) {
+                bandwidth.push_back(2);
+            }
+        }
+    } else {
+        num_bandwidth = NUMA_NUM_NODES;
+        nodemask_t nodemask;
+        struct bitmask nodemask_bm = {NUMA_NUM_NODES, nodemask.n};
+        numa_bitmask_clearall(&nodemask_bm);
+
+        memkind_hbw_all_get_mbind_nodemask(NULL, nodemask.n, NUMA_NUM_NODES);
+
+        int i, nodes_num = numa_num_configured_nodes();
+        for (i=0; i<NUMA_NUM_NODES; i++) {
+            if (i >= nodes_num) {
+                bandwidth.push_back(0);
+            } else if (numa_bitmask_isbitset(&nodemask_bm, i)) {
+                bandwidth.push_back(2);
+            } else {
+                bandwidth.push_back(1);
+            }
+        }
+    }
+}
+
+void TGTest :: TearDown()
+{}
diff --git a/memkind-1.10.0/test/trial_generator.h b/memkind-1.10.0/test/trial_generator.h
new file mode 100644
index 0000000..fd05954
--- /dev/null
+++ b/memkind-1.10.0/test/trial_generator.h
@@ -0,0 +1,97 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef execute_trials_include_h
+#define execute_trials_include_h
+
+#include "hbwmalloc.h"
+#include "memkind.h"
+
+#include <vector>
+#include <stdlib.h>
+#include <vector>
+#include <memory>
+#include <algorithm>
+#include <fstream>
+#include <iostream>
+
+#include "common.h"
+
+
+typedef enum {
+    HBW_MALLOC,
+    HBW_CALLOC,
+    HBW_REALLOC,
+    HBW_MEMALIGN,
+    HBW_MEMALIGN_PSIZE,
+    HBW_FREE,
+    MEMKIND_MALLOC,
+    MEMKIND_CALLOC,
+    MEMKIND_REALLOC,
+    MEMKIND_POSIX_MEMALIGN,
+    MEMKIND_FREE
+} alloc_api_t;
+
+typedef struct {
+    alloc_api_t api;
+    size_t size;
+    size_t alignment;
+    size_t page_size;
+    memkind_t memkind;
+    int free_index;
+} trial_t;
+
+class TrialGenerator
+{
+public:
+    TrialGenerator() {}
+    void generate_gb(alloc_api_t api, int number_of_gb_pages, memkind_t memkind,
+                     alloc_api_t api_free, bool psize_strict=false, size_t align = GB);
+    void run(int num_bandwidth, std::vector<int> &bandwidths);
+    void generate_size_2bytes_2KB_2MB(alloc_api_t api);
+    /*For debugging purposes*/
+    void print();
+private:
+    std::vector<trial_t> trial_vec;
+    trial_t create_trial_tuple(alloc_api_t api,
+                               size_t size,
+                               size_t alignment,
+                               int page_size,
+                               memkind_t memkind,
+                               int free_index);
+
+
+};
+
+class TGTest : public::testing::Test
+{
+protected:
+    size_t num_bandwidth;
+    std::vector<int> bandwidth;
+    std::unique_ptr<TrialGenerator> tgen;
+    void SetUp();
+    void TearDown();
+};
+
+#endif
diff --git a/memkind-1.10.0/utils/docker/Dockerfile.fedora-31 b/memkind-1.10.0/utils/docker/Dockerfile.fedora-31
new file mode 100644
index 0000000..e7d97dd
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/Dockerfile.fedora-31
@@ -0,0 +1,75 @@
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Pull base image
+FROM fedora:31
+
+LABEL maintainer="katarzyna.wasiuta@intel.com"
+
+# Update the dnf cache and install basic tools
+RUN dnf update -y && dnf install -y \
+    asciidoctor \
+    automake \
+    bash-completion \
+    ca-certificates \
+    daxctl-devel \
+    devscripts \
+    expect \
+    g++ \
+    git \
+    json-c-devel \
+    keyutils-libs-devel \
+    kmod-devel \
+    libtool \
+    libudev-devel \
+    libuuid-devel \
+    numactl \
+    numactl-devel \
+    pkgconfig \
+    python-pip \
+    rpmdevtools \
+    sudo \
+    systemd \
+    which \
+    whois \
+ && dnf clean all
+
+# Install packages required by python tests
+RUN pip install wheel
+RUN pip install setuptools==3.4.1
+RUN pip install pytest==3.9.2
+
+# Add user
+ENV USER memkinduser
+ENV USERPASS memkindpass
+RUN useradd -m $USER -p `mkpasswd $USERPASS`
+RUN gpasswd wheel -a $USER
+RUN echo '%wheel ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers
+
+# Create directory for memkind repository
+WORKDIR /home/$USER/memkind
+
+# Allow user to create files in the home directory
+RUN chown -R $USER:wheel /home/$USER
+
+# Change user to $USER
+USER $USER
diff --git a/memkind-1.10.0/utils/docker/Dockerfile.ubuntu-18.04 b/memkind-1.10.0/utils/docker/Dockerfile.ubuntu-18.04
new file mode 100644
index 0000000..42cf238
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/Dockerfile.ubuntu-18.04
@@ -0,0 +1,71 @@
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Pull base image
+FROM ubuntu:18.04
+
+LABEL maintainer="katarzyna.wasiuta@intel.com"
+
+# Update the Apt cache and install basic tools
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends \
+    asciidoctor \
+    automake \
+    bash-completion \
+    ca-certificates \
+    curl \
+    devscripts \
+    g++ \
+    git \
+    libjson-c-dev \
+    libkeyutils-dev \
+    libkmod-dev \
+    libnuma-dev \
+    libtool \
+    libudev-dev \
+    numactl \
+    pkg-config \
+    python-pip \
+    sudo \
+    systemd \
+    uuid-dev \
+    whois \
+ && rm -rf /var/lib/apt/lists/*
+
+# Install packages required by python tests
+RUN pip install wheel
+RUN pip install setuptools==3.4.1
+RUN pip install pytest==3.9.2
+
+# Add user
+ENV USER memkinduser
+ENV USERPASS memkindpass
+RUN useradd -m $USER -g sudo -p `mkpasswd $USERPASS`
+RUN echo '%sudo ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers
+
+# Create directory for memkind repository
+WORKDIR /home/$USER/memkind
+
+# Allow user to create files in the home directory
+RUN chown -R $USER:sudo /home/$USER
+
+# Change user to $USER
+USER $USER
diff --git a/memkind-1.10.0/utils/docker/README.md b/memkind-1.10.0/utils/docker/README.md
new file mode 100644
index 0000000..32d285e
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/README.md
@@ -0,0 +1,51 @@
+# **README**
+
+This is utils/docker/README.
+
+Scripts in this directory let run a Docker container with Ubuntu 18.04 and Fedora 31 environment
+to build, test and optionally measure test coverage of any pull request to memkind project, inside it.
+
+# Environment variables
+
+* **MEMKIND_HOST_WORKDIR** - Defines the absolute path to memkind project root directory on host environment.
+    **This variable must be defined.**
+
+* **CODECOV_TOKEN** - Codecov token for memkind repository to upload the coverage results.
+
+* **TEST_SUITE_NAME** - Name of test suite (possible values are HBW/PMEM/DAX_KMEM).
+
+* **PMEM_HOST_PATH** - PMEM mount device path on host environment (useful for testing PMEM configuration). Default value of **PMEM_HOST_PATH** is
+*'/tmp/'*.
+
+* **NDCTL_LIBRARY_VERSION** - ndctl library version.
+For fully testing MEMKIND_DAX_KMEM, ndctl library version tag must be passed as parameter,
+see https://github.com/pmem/ndctl/tags.
+
+* **TBB_LIBRARY_VERSION** - Intel Threading Building Blocks library version.
+For testing Threading Building Blocks, TBB library version tag must be passed as parameter,
+see https://github.com/01org/tbb/tags.
+
+# Files
+*'docker_run_build.sh'*  is used to build of memkind.
+
+*'docker_run_coverage.sh'*  is used for uploading coverage report on [Codecov.io](Codecov.io)
+
+*'docker_run_test.sh'*  is used to run tests of memkind.
+
+*'docker_install_ndctl.sh'*  is used to install ndctl library.
+
+*'docker_install_tbb.sh'*  is used to install Intel Threading Building Blocks library.
+
+*'set_host_configuration.sh'*  is used to set hugepages configuration on host machine.
+
+# Building and running the container
+
+One can use the script *run_local.sh* just to build and run tests, e.g.:
+
+```
+$ ./run_local.sh Dockerfile.ubuntu-18.04
+```
+
+**Note:**
+
+Others class group must have write permission to MEMKIND_HOST_WORKDIR to use the run_local script.
diff --git a/memkind-1.10.0/utils/docker/docker_install_ndctl.sh b/memkind-1.10.0/utils/docker/docker_install_ndctl.sh
new file mode 100755
index 0000000..9b8c04a
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/docker_install_ndctl.sh
@@ -0,0 +1,63 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# docker_install_ndctl.sh - is called inside a Docker container;
+# installs ndctl library
+#
+
+set -e
+
+NDCTL_RELEASE_URL=https://github.com/pmem/ndctl/releases/tag/"$NDCTL_LIBRARY_VERSION"
+NDCTL_TAR_GZ="$NDCTL_LIBRARY_VERSION".tar.gz
+NDCTL_TARBALL_URL=https://github.com/pmem/ndctl/archive/"$NDCTL_TAR_GZ"
+
+# check if specified ndctl release exists
+if curl --output /dev/null --silent --head --fail "$NDCTL_RELEASE_URL"; then
+  echo "ndctl version ${NDCTL_LIBRARY_VERSION} exist."
+else
+  echo "ndctl url: ${NDCTL_RELEASE_URL} is not valid. ndctl version: ${NDCTL_LIBRARY_VERSION} doesn't exist."
+  exit 1
+fi
+
+NDCTL_LOCAL_DIR="$HOME"/ndctl/"$NDCTL_LIBRARY_VERSION"
+NDCTL_LOCAL_TAR_GZ="$NDCTL_LOCAL_DIR"/"$NDCTL_TAR_GZ"
+
+# create ndctl directory in home directory
+mkdir -p "$NDCTL_LOCAL_DIR"
+
+# download and untar ndctl library to ndctl directory
+curl -L "$NDCTL_TARBALL_URL" -o "$NDCTL_LOCAL_TAR_GZ"
+tar -xzf "$NDCTL_LOCAL_TAR_GZ" -C "$NDCTL_LOCAL_DIR" --strip-components=1
+
+# go to ndctl directory and build ndctl library
+cd "$NDCTL_LOCAL_DIR"
+./autogen.sh
+./configure --prefix=/usr --sysconfdir=/etc --libdir=/usr/lib
+make -j "$(nproc --all)"
+sudo make -j "$(nproc --all)" install
+
+# update shared library cache
+sudo ldconfig
+# return to previous directory
+cd -
diff --git a/memkind-1.10.0/utils/docker/docker_install_tbb.sh b/memkind-1.10.0/utils/docker/docker_install_tbb.sh
new file mode 100755
index 0000000..eb2b0ef
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/docker_install_tbb.sh
@@ -0,0 +1,59 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# docker_install_tbb.sh - is called inside a Docker container;
+# installs Threading Building Block library
+#
+
+set -e
+
+TBB_RELEASE_URL=https://github.com/01org/tbb/releases/tag/"$TBB_LIBRARY_VERSION"
+TBB_TAR_GZ="$TBB_LIBRARY_VERSION".tar.gz
+TBB_TARBALL_URL=https://github.com/01org/tbb/archive/"$TBB_TAR_GZ"
+TBB_VARS_SH="tbbvars.sh"
+
+# check if specified TBB release exists
+if curl --output /dev/null --silent --head --fail "$TBB_RELEASE_URL"; then
+  echo "TBB version ${TBB_LIBRARY_VERSION} exist."
+else
+  echo "TBB url: ${TBB_RELEASE_URL} is not valid. TBB version: ${TBB_LIBRARY_VERSION} doesn't exist."
+  exit 1
+fi
+
+TBB_LOCAL_DIR="$HOME"/tbb/"$TBB_LIBRARY_VERSION"
+TBB_LOCAL_TAR_GZ="$TBB_LOCAL_DIR"/"$TBB_TAR_GZ"
+
+# create TBB directory in home directory
+mkdir -p "$TBB_LOCAL_DIR"
+
+# download and untar TBB library to TBB directory
+curl -L "$TBB_TARBALL_URL" -o "$TBB_LOCAL_TAR_GZ"
+tar -xzf "$TBB_LOCAL_TAR_GZ" -C "$TBB_LOCAL_DIR" --strip-components=1
+
+# build TBB library
+make -j "$(nproc --all)" --directory="$TBB_LOCAL_DIR"
+TBB_RELEASE_DIR=$(ls -d "$TBB_LOCAL_DIR"/build/*release)
+
+# set environment variables regarding TBB package
+source "$TBB_RELEASE_DIR"/"$TBB_VARS_SH"
diff --git a/memkind-1.10.0/utils/docker/docker_run_build.sh b/memkind-1.10.0/utils/docker/docker_run_build.sh
new file mode 100755
index 0000000..a8ed07b
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/docker_run_build.sh
@@ -0,0 +1,64 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# docker_run_build.sh - is called inside a Docker container;
+# prepares and runs memkind build for specified pull request number
+#
+set -e
+
+export UTILS_PREFIX=utils/docker
+
+if [ -n "$CODECOV_TOKEN" ]; then
+    GCOV_OPTION="--enable-gcov"
+fi
+
+# if ndctl library version is specified install library
+if [ -n "$NDCTL_LIBRARY_VERSION" ]; then
+    "$UTILS_PREFIX"/docker_install_ndctl.sh
+fi
+
+# building memkind sources and tests
+./autogen.sh
+./configure --prefix=/usr $GCOV_OPTION
+make -j "$(nproc --all)"
+make -j "$(nproc --all)" checkprogs
+
+# building RPM package
+if [[ $(cat /etc/os-release) = *"Fedora"* ]]; then
+    make -j "$(nproc --all)" rpm
+fi
+
+# installing memkind
+sudo make install
+
+# if TBB library version is specified install library and use it
+# as MEMKIND_HEAP_MANAGER
+if [ -n "$TBB_LIBRARY_VERSION" ]; then
+    source "$UTILS_PREFIX"/docker_install_tbb.sh
+    HEAP_MANAGER="TBB"
+fi
+if [ -n "$TEST_SUITE_NAME" ]; then
+    "$UTILS_PREFIX"/docker_run_test.sh "$HEAP_MANAGER"
+fi
diff --git a/memkind-1.10.0/utils/docker/docker_run_coverage.sh b/memkind-1.10.0/utils/docker/docker_run_coverage.sh
new file mode 100755
index 0000000..c51452d
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/docker_run_coverage.sh
@@ -0,0 +1,35 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# docker_run_coverage.sh - is called inside a Docker container;
+# calculates coverage and creates online report
+#
+# Parameters:
+# -memkind repository directory
+# Options used:
+# -p dir       Project root directory
+# -Z           Exit with 1 if not successful. Default will Exit with 0
+
+MEMKIND_REPO_PATH=$1
+bash <(curl -s https://codecov.io/bash) -t "$CODECOV_TOKEN" -cF "$TEST_SUITE_NAME" -p "$MEMKIND_REPO_PATH" -Z
diff --git a/memkind-1.10.0/utils/docker/docker_run_test.sh b/memkind-1.10.0/utils/docker/docker_run_test.sh
new file mode 100755
index 0000000..4d1da2e
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/docker_run_test.sh
@@ -0,0 +1,51 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# docker_run_test.sh - is called inside a Docker container;
+# runs specified memkind tests
+#
+# Parameters:
+# -heap manager
+set -e
+
+HEAP_MANAGER=$1
+
+if [ "$TEST_SUITE_NAME" = "HBW" ]; then
+    # running tests and display output in case of failure
+    make check || { cat test-suite.log; exit 1; }
+elif [ "$TEST_SUITE_NAME" = "PMEM" ]; then
+    make PMEM_PATH="$PMEM_CONTAINER_PATH" unit_tests_pmem
+    # running pmem examples
+    find examples/.libs -name "pmem*" -executable -type f -exec sh -c "MEMKIND_HEAP_MANAGER=$HEAP_MANAGER "{}" $PMEM_CONTAINER_PATH" \;
+elif [ "$TEST_SUITE_NAME" = "DAX_KMEM" ]; then
+    make unit_tests_dax_kmem
+else
+    echo "Unknown Test suite ${TEST_SUITE_NAME}"
+fi;
+
+# executing coverage script if codecov token is set
+if [ -n "$CODECOV_TOKEN" ]; then
+    "$UTILS_PREFIX"/docker_run_coverage.sh "$PWD"
+fi
diff --git a/memkind-1.10.0/utils/docker/run_local.sh b/memkind-1.10.0/utils/docker/run_local.sh
new file mode 100755
index 0000000..40d04a2
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/run_local.sh
@@ -0,0 +1,68 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# run_local.sh - builds a container with Docker image
+# and runs docker_run_build.sh script inside it
+#
+# Parameters:
+# -name of the Docker image file
+
+set -e
+
+DOCKER_IMAGE_NAME="$1"
+export PMEM_HOST_PATH=${PMEM_HOST_PATH:-/tmp/}
+
+if [[ ! -f "$DOCKER_IMAGE_NAME" ]]; then
+    echo "Docker image "$DOCKER_IMAGE_NAME" does not exist."
+    exit 1
+fi
+
+if [[ -z "$MEMKIND_HOST_WORKDIR" ]]; then
+    echo "MEMKIND_HOST_WORKDIR has to be set."
+    exit 1
+fi
+
+# need to be inline with Dockerfile WORKDIR
+MEMKIND_CONTAINER_WORKDIR=/home/memkinduser/memkind/
+PMEM_CONTAINER_PATH=/home/memkinduser/mnt_pmem/
+
+docker build --tag memkind_cont \
+             --file "$DOCKER_IMAGE_NAME" \
+             --build-arg http_proxy=$http_proxy \
+             --build-arg https_proxy=$https_proxy \
+             .
+docker run --rm \
+           --privileged=true \
+           --tty=true \
+           --env http_proxy=$http_proxy \
+           --env https_proxy=$https_proxy \
+           --env CODECOV_TOKEN="$CODECOV_TOKEN" \
+           --env TEST_SUITE_NAME="$TEST_SUITE_NAME" \
+           --env TBB_LIBRARY_VERSION="$TBB_LIBRARY_VERSION" \
+           --env NDCTL_LIBRARY_VERSION="$NDCTL_LIBRARY_VERSION" \
+           --env PMEM_CONTAINER_PATH="$PMEM_CONTAINER_PATH" \
+           --mount type=bind,source="$MEMKIND_HOST_WORKDIR",target="$MEMKIND_CONTAINER_WORKDIR" \
+           --mount type=bind,source="$PMEM_HOST_PATH",target="$PMEM_CONTAINER_PATH" \
+           memkind_cont utils/docker/docker_run_build.sh
diff --git a/memkind-1.10.0/utils/docker/set_host_configuration.sh b/memkind-1.10.0/utils/docker/set_host_configuration.sh
new file mode 100755
index 0000000..3cc027b
--- /dev/null
+++ b/memkind-1.10.0/utils/docker/set_host_configuration.sh
@@ -0,0 +1,44 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# set_host_configuration.sh - set Huge Pages parameters required for memkind tests
+set -e
+
+MEMKIND_HUGE_PAGES_NO=3000
+MEMKIND_OVERCOMMIT_HUGEPAGES_NO=128
+
+# find out current configuration
+MEMKIND_HUGE_PAGES_FOUND=$(cat /proc/sys/vm/nr_hugepages)
+MEMKIND_OVERCOMMIT_HUGEPAGES_FOUND=$(cat /proc/sys/vm/nr_overcommit_hugepages)
+
+# set expected configuration
+if [ "$MEMKIND_HUGE_PAGES_FOUND" != "$MEMKIND_HUGE_PAGES_NO" ]; then
+    echo "Setting number of hugepages to ${MEMKIND_HUGE_PAGES_NO}."
+    sudo sysctl vm.nr_hugepages="$MEMKIND_HUGE_PAGES_NO"
+fi
+if [ "$MEMKIND_OVERCOMMIT_HUGEPAGES_FOUND" != "$MEMKIND_OVERCOMMIT_HUGEPAGES_NO" ]; then
+    echo "Setting number of overcommit hugepages to ${MEMKIND_OVERCOMMIT_HUGEPAGES_NO}."
+    sudo sysctl vm.nr_overcommit_hugepages="$MEMKIND_OVERCOMMIT_HUGEPAGES_NO"
+fi
diff --git a/memkind.pc.in b/memkind.pc.in
new file mode 100644
index 0000000..4441acc
--- /dev/null
+++ b/memkind.pc.in
@@ -0,0 +1,13 @@
+version=@VERSION@
+prefix=@prefix@
+exec_prefix=@exec_prefix@
+libdir=@libdir@
+includedir=@includedir@
+
+Name: memkind
+Description: User Extensible Heap Manager
+URL: http://memkind.github.io/memkind
+Version: @VERSION@
+Cflags: -I${includedir}
+Libs: -L${libdir} -lmemkind
+Libs.private: -ldl -lnuma -lpthread @DAXCTL_LIBS@
diff --git a/memkind.spec.mk b/memkind.spec.mk
new file mode 100644
index 0000000..49a29e1
--- /dev/null
+++ b/memkind.spec.mk
@@ -0,0 +1,262 @@
+#
+#  Copyright (C) 2014 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+# targets for building rpm
+version ?= 0.0.0
+release ?= 1
+arch = $(shell uname -p)
+name ?= memkind
+
+rpm: $(name)-$(version).tar.gz
+	rpmbuild $(rpmbuild_flags) $^ -ta
+
+memkind-$(version).spec:
+	@echo "$$memkind_spec" > $@
+	cat ChangeLog >> $@
+
+.PHONY: rpm
+
+define memkind_spec
+Summary: User Extensible Heap Manager
+Name: $(name)
+Version: $(version)
+Release: $(release)
+License: BSD-2-Clause
+Group: System Environment/Libraries
+URL: http://memkind.github.io/memkind
+BuildRoot: %{_tmppath}/%{name}-%{version}-%{release}-root-%(%{__id_u} -n)
+BuildRequires: automake libtool gcc-c++ unzip
+%if %{defined suse_version}
+BuildRequires: libnuma-devel
+%else
+BuildRequires: numactl-devel
+%endif
+
+%define daxctl_min_version 66
+%if %{defined suse_version}
+BuildRequires: libdaxctl-devel >= %{daxctl_min_version}
+%else
+BuildRequires: daxctl-devel >= %{daxctl_min_version}
+%endif
+
+Prefix: %{_prefix}
+Prefix: %{_unitdir}
+%if %{undefined suse_version}
+Obsoletes: memkind
+Provides: memkind libmemkind0
+%endif
+
+%define namespace memkind
+
+%if %{defined suse_version}
+%define docdir %{_defaultdocdir}/%{namespace}
+%else
+%define docdir %{_defaultdocdir}/%{namespace}-%{version}
+%endif
+
+# Upstream testing of memkind is done exclusively on x86_64; other archs
+# are unsupported but may work.
+ExclusiveArch: x86_64 ppc64 ppc64le s390x aarch64
+
+# default values if version is a tagged release on github
+%{!?commit: %define commit %{version}}
+%{!?buildsubdir: %define buildsubdir %{namespace}-%{commit}}
+Source0: https://github.com/%{namespace}/%{namespace}/archive/v%{commit}/%{buildsubdir}.tar.gz
+
+%description
+The memkind library is an user extensible heap manager built on top of
+jemalloc which enables control of memory characteristics and a
+partitioning of the heap between kinds of memory. The kinds of memory
+are defined by operating system memory policies that have been applied
+to virtual address ranges. Memory characteristics supported by
+memkind without user extension include control of NUMA and page size
+features. The jemalloc non-standard interface has been extended to
+enable specialized arenas to make requests for virtual memory from the
+operating system through the memkind partition interface. Through the
+other memkind interfaces the user can control and extend memory
+partition features and allocate memory while selecting enabled
+features. This software is being made available for early evaluation.
+Feedback on design or implementation is greatly appreciated.
+
+%package devel
+Summary: Memkind User Extensible Heap Manager development lib and tools
+Group: Development/Libraries
+Requires: %{name} = %{version}-%{release}
+%if %{undefined suse_version}
+Obsoletes: memkind-devel
+Provides: memkind-devel
+%endif
+
+%description devel
+Install header files and development aids to link memkind library into
+applications.
+
+%package tests
+Summary: Extension to libnuma for kinds of memory - validation
+Group: Validation/Libraries
+Requires: %{name} = %{version}-%{release}
+
+%description tests
+memkind functional tests
+
+%prep
+%setup -q -a 0 -n $(name)-%{version}
+
+%build
+# It is required that we configure and build the jemalloc subdirectory
+# before we configure and start building the top level memkind directory.
+# To ensure the memkind build step is able to discover the output
+# of the jemalloc build we must create an 'obj' directory, and build
+# from within that directory.
+
+cd %{_builddir}/%{buildsubdir}
+echo %{version} > %{_builddir}/%{buildsubdir}/VERSION
+./build.sh --prefix=%{_prefix} --includedir=%{_includedir} --libdir=%{_libdir} \
+           --bindir=%{_bindir} --docdir=%{_docdir}/%{namespace} --mandir=%{_mandir} --sbindir=%{_sbindir}
+
+%install
+cd %{_builddir}/%{buildsubdir}
+%{__make} DESTDIR=%{buildroot} install
+%{__install} -d %{buildroot}$(memkind_test_dir)
+%{__install} -d %{buildroot}/%{_unitdir}
+%{__install} -d %{buildroot}/$(memkind_test_dir)/python_framework
+%{__install} test/.libs/* test/*.sh test/*.ts test/*.py %{buildroot}$(memkind_test_dir)
+%{__install} test/python_framework/*.py %{buildroot}/$(memkind_test_dir)/python_framework
+rm -f %{buildroot}$(memkind_test_dir)/libautohbw.*
+rm -f %{buildroot}/%{_libdir}/lib%{namespace}.{l,}a
+rm -f %{buildroot}/%{_libdir}/libautohbw.{l,}a
+
+%pre
+
+%post
+/sbin/ldconfig
+
+%preun
+
+%postun
+/sbin/ldconfig
+
+%files
+%defattr(-,root,root,-)
+%license %{_docdir}/%{namespace}/COPYING
+%doc %{_docdir}/%{namespace}/README
+%doc %{_docdir}/%{namespace}/VERSION
+%dir %{_docdir}/%{namespace}
+%{_libdir}/lib%{namespace}.so.*
+%{_libdir}/libautohbw.so.*
+%{_bindir}/%{namespace}-auto-dax-kmem-nodes
+%{_bindir}/%{namespace}-hbw-nodes
+
+%files devel
+%defattr(-,root,root,-)
+%{_includedir}
+%{_includedir}/hbwmalloc.h
+%{_includedir}/hbw_allocator.h
+%{_includedir}/memkind_allocator.h
+%{_includedir}/pmem_allocator.h
+%{_libdir}/lib%{namespace}.so
+%{_libdir}/libautohbw.so
+%{_libdir}/pkgconfig/memkind.pc
+%{_includedir}/%{namespace}.h
+%{_mandir}/man1/memkind-auto-dax-kmem-nodes.1.*
+%{_mandir}/man1/memkind-hbw-nodes.1.*
+%{_mandir}/man3/hbwmalloc.3.*
+%{_mandir}/man3/hbwallocator.3.*
+%{_mandir}/man3/pmemallocator.3.*
+%{_mandir}/man3/%{namespace}*.3.*
+%{_mandir}/man7/autohbw.7.*
+
+%files tests
+%defattr(-,root,root,-)
+$(memkind_test_dir)/all_tests
+${memkind_test_dir}/environ_err_dax_kmem_malloc_positive_test
+${memkind_test_dir}/environ_err_dax_kmem_malloc_test
+$(memkind_test_dir)/environ_err_hbw_malloc_test
+${memkind_test_dir}/dax_kmem_test
+$(memkind_test_dir)/decorator_test
+$(memkind_test_dir)/locality_test
+$(memkind_test_dir)/freeing_memory_segfault_test
+$(memkind_test_dir)/gb_page_tests_bind_policy
+$(memkind_test_dir)/filter_memkind
+$(memkind_test_dir)/hello_hbw
+$(memkind_test_dir)/hello_memkind
+$(memkind_test_dir)/hello_memkind_debug
+$(memkind_test_dir)/memkind_allocated
+$(memkind_test_dir)/memkind_cpp_allocator
+$(memkind_test_dir)/memkind_get_stat
+$(memkind_test_dir)/memkind_stat_test
+$(memkind_test_dir)/autohbw_candidates
+${memkind_test_dir}/pmem_kinds
+${memkind_test_dir}/pmem_malloc
+${memkind_test_dir}/pmem_malloc_unlimited
+${memkind_test_dir}/pmem_usable_size
+${memkind_test_dir}/pmem_alignment
+${memkind_test_dir}/pmem_and_dax_kmem_kind
+${memkind_test_dir}/pmem_and_default_kind
+${memkind_test_dir}/pmem_config
+$(memkind_test_dir)/pmem_detect_kind
+${memkind_test_dir}/pmem_multithreads
+${memkind_test_dir}/pmem_multithreads_onekind
+$(memkind_test_dir)/pmem_free_with_unknown_kind
+${memkind_test_dir}/pmem_cpp_allocator
+${memkind_test_dir}/pmem_test
+${memkind_test_dir}/allocator_perf_tool_tests
+${memkind_test_dir}/perf_tool
+${memkind_test_dir}/autohbw_test_helper
+${memkind_test_dir}/trace_mechanism_test_helper
+$(memkind_test_dir)/memkind-afts.ts
+$(memkind_test_dir)/memkind-afts-ext.ts
+$(memkind_test_dir)/memkind-slts.ts
+$(memkind_test_dir)/memkind-perf.ts
+$(memkind_test_dir)/memkind-perf-ext.ts
+$(memkind_test_dir)/memkind-pytests.ts
+$(memkind_test_dir)/performance_test
+$(memkind_test_dir)/test.sh
+$(memkind_test_dir)/test_dax_kmem.sh
+$(memkind_test_dir)/hbw_detection_test.py
+$(memkind_test_dir)/dax_kmem_env_var_test.py
+$(memkind_test_dir)/autohbw_test.py
+$(memkind_test_dir)/trace_mechanism_test.py
+$(memkind_test_dir)/python_framework
+$(memkind_test_dir)/python_framework/cmd_helper.py
+$(memkind_test_dir)/python_framework/huge_page_organizer.py
+$(memkind_test_dir)/python_framework/__init__.py
+$(memkind_test_dir)/draw_plots.py
+$(memkind_test_dir)/run_alloc_benchmark.sh
+$(memkind_test_dir)/alloc_benchmark_hbw
+$(memkind_test_dir)/alloc_benchmark_glibc
+$(memkind_test_dir)/alloc_benchmark_tbb
+$(memkind_test_dir)/alloc_benchmark_pmem
+$(memkind_test_dir)/fragmentation_benchmark_pmem
+$(memkind_test_dir)/defrag_reallocate
+
+%exclude $(memkind_test_dir)/*.pyo
+%exclude $(memkind_test_dir)/*.pyc
+%exclude $(memkind_test_dir)/python_framework/*.pyo
+%exclude $(memkind_test_dir)/python_framework/*.pyc
+
+%changelog
+endef
+
+export memkind_spec
diff --git a/src/Makefile.mk b/src/Makefile.mk
new file mode 100644
index 0000000..714a8dc
--- /dev/null
+++ b/src/Makefile.mk
@@ -0,0 +1,28 @@
+#
+#  Copyright (C) 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+clean-local: src-clean
+
+src-clean:
+	rm -f src/*.gcno
diff --git a/src/hbwmalloc.c b/src/hbwmalloc.c
new file mode 100644
index 0000000..16d8c29
--- /dev/null
+++ b/src/hbwmalloc.c
@@ -0,0 +1,302 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbwmalloc.h>
+#include <memkind.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <pthread.h>
+#include <errno.h>
+#include <numa.h>
+#include <numaif.h>
+#include <unistd.h>
+#include <stdint.h>
+
+static hbw_policy_t hbw_policy_g = HBW_POLICY_PREFERRED;
+static pthread_once_t hbw_policy_once_g = PTHREAD_ONCE_INIT;
+
+static void hbw_policy_bind_init(void)
+{
+    hbw_policy_g = HBW_POLICY_BIND;
+}
+
+static void hbw_policy_bind_all_init(void)
+{
+    hbw_policy_g = HBW_POLICY_BIND_ALL;
+}
+
+static void hbw_policy_preferred_init(void)
+{
+    hbw_policy_g = HBW_POLICY_PREFERRED;
+}
+
+static void hbw_policy_interleave_init(void)
+{
+    hbw_policy_g = HBW_POLICY_INTERLEAVE;
+}
+
+// This function is intended to be called once per pagesize
+// Getting kind should be done using hbw_get_kind() defined below
+static memkind_t hbw_choose_kind(hbw_pagesize_t pagesize)
+{
+    memkind_t result = NULL;
+
+    hbw_set_policy(hbw_policy_g);
+
+    int policy = hbw_get_policy();
+
+    // PREFERRED policy have separate handling cause it can fallback
+    // to non-HBW kinds in case of HBW absence
+    if (policy != HBW_POLICY_PREFERRED ) {
+        switch (pagesize) {
+            case HBW_PAGESIZE_2MB:
+                if(policy == HBW_POLICY_BIND_ALL) {
+                    result = MEMKIND_HBW_ALL_HUGETLB;
+                } else {
+                    result = MEMKIND_HBW_HUGETLB;
+                }
+                break;
+            case HBW_PAGESIZE_1GB:
+            case HBW_PAGESIZE_1GB_STRICT:
+                result = MEMKIND_HBW_GBTLB;
+                break;
+            default:
+                if (policy == HBW_POLICY_BIND) {
+                    result = MEMKIND_HBW;
+                } else if (policy == HBW_POLICY_BIND_ALL) {
+                    result = MEMKIND_HBW_ALL;
+                } else {
+                    result = MEMKIND_HBW_INTERLEAVE;
+                }
+                break;
+        }
+    } else if (memkind_check_available(MEMKIND_HBW) == 0) {
+        switch (pagesize) {
+            case HBW_PAGESIZE_2MB:
+                result = MEMKIND_HBW_PREFERRED_HUGETLB;
+                break;
+            case HBW_PAGESIZE_1GB:
+            case HBW_PAGESIZE_1GB_STRICT:
+                result = MEMKIND_HBW_PREFERRED_GBTLB;
+                break;
+            default:
+                result = MEMKIND_HBW_PREFERRED;
+                break;
+        }
+    } else {
+        switch (pagesize) {
+            case HBW_PAGESIZE_2MB:
+                result = MEMKIND_HUGETLB;
+                break;
+            case HBW_PAGESIZE_1GB:
+            case HBW_PAGESIZE_1GB_STRICT:
+                result = MEMKIND_GBTLB;
+                break;
+            default:
+                result = MEMKIND_DEFAULT;
+                break;
+        }
+    }
+    return result;
+}
+
+static memkind_t pagesize_kind[HBW_PAGESIZE_MAX_VALUE];
+static inline memkind_t hbw_get_kind(hbw_pagesize_t pagesize)
+{
+    if(pagesize_kind[pagesize] == NULL) {
+        pagesize_kind[pagesize] = hbw_choose_kind(pagesize);
+    }
+    return pagesize_kind[pagesize];
+}
+
+
+MEMKIND_EXPORT hbw_policy_t hbw_get_policy(void)
+{
+    return hbw_policy_g;
+}
+
+MEMKIND_EXPORT int hbw_set_policy(hbw_policy_t mode)
+{
+    switch(mode) {
+        case HBW_POLICY_PREFERRED:
+            pthread_once(&hbw_policy_once_g, hbw_policy_preferred_init);
+            break;
+        case HBW_POLICY_BIND:
+            pthread_once(&hbw_policy_once_g, hbw_policy_bind_init);
+            break;
+        case HBW_POLICY_BIND_ALL:
+            pthread_once(&hbw_policy_once_g, hbw_policy_bind_all_init);
+            break;
+        case HBW_POLICY_INTERLEAVE:
+            pthread_once(&hbw_policy_once_g, hbw_policy_interleave_init);
+            break;
+        default:
+            return EINVAL;
+    }
+
+    if (mode != hbw_policy_g) {
+        return EPERM;
+    }
+
+    return 0;
+}
+
+MEMKIND_EXPORT int hbw_check_available(void)
+{
+    return  (memkind_check_available(MEMKIND_HBW) == 0) ? 0 : ENODEV;
+}
+
+static inline void hbw_touch_page(void *addr)
+{
+    volatile char *temp_ptr = (volatile char *) addr;
+    char value = temp_ptr[0];
+    temp_ptr[0] = value;
+}
+
+MEMKIND_EXPORT int hbw_verify_memory_region(void *addr, size_t size, int flags)
+{
+    /*
+     * if size is invalid, flags have unsupported bit set or if addr is NULL.
+     */
+    if (addr == NULL || size == 0 || flags & ~HBW_TOUCH_PAGES) {
+        return EINVAL;
+    }
+
+    /*
+     * 4KB is the smallest pagesize. When pagesize is bigger, pages are verified more than once
+     */
+    const size_t page_size = sysconf(_SC_PAGESIZE);
+    const size_t page_mask = ~(page_size-1);
+
+    /*
+     * block size should be power of two to enable compiler optimizations
+     */
+    const unsigned block_size = 64;
+
+    char *end = addr + size;
+    char *aligned_beg = (char *)((uintptr_t)addr & page_mask);
+    nodemask_t nodemask;
+    struct bitmask expected_nodemask = {NUMA_NUM_NODES, nodemask.n};
+
+    memkind_hbw_all_get_mbind_nodemask(NULL, expected_nodemask.maskp,
+                                       expected_nodemask.size);
+
+    while(aligned_beg < end) {
+        int nodes[block_size];
+        void *pages[block_size];
+        int i = 0, page_count = 0;
+        char *iter_end = aligned_beg + block_size*page_size;
+
+        if (iter_end > end) {
+            iter_end = end;
+        }
+
+        while (aligned_beg < iter_end) {
+            if (flags & HBW_TOUCH_PAGES) {
+                hbw_touch_page(aligned_beg);
+            }
+            pages[page_count++] = aligned_beg;
+            aligned_beg += page_size;
+        }
+
+        if (move_pages(0, page_count, pages, NULL, nodes, MPOL_MF_MOVE)) {
+            return EFAULT;
+        }
+
+        for (i = 0; i < page_count; i++) {
+            /*
+             * negative value of nodes[i] indicates that move_pages could not establish
+             * page location, e.g. addr is not pointing to valid virtual mapping
+             */
+            if(nodes[i] < 0) {
+                return -1;
+            }
+            /*
+             * if nodes[i] is not present in expected_nodemask then
+             * physical memory backing page is not hbw
+             */
+            if (!numa_bitmask_isbitset(&expected_nodemask, nodes[i])) {
+                return -1;
+            }
+        }
+    }
+
+    return 0;
+}
+
+MEMKIND_EXPORT void *hbw_malloc(size_t size)
+{
+    return memkind_malloc(hbw_get_kind(HBW_PAGESIZE_4KB), size);
+}
+
+MEMKIND_EXPORT void *hbw_calloc(size_t num, size_t size)
+{
+    return memkind_calloc(hbw_get_kind(HBW_PAGESIZE_4KB), num, size);
+}
+
+MEMKIND_EXPORT int hbw_posix_memalign(void **memptr, size_t alignment,
+                                      size_t size)
+{
+    return memkind_posix_memalign(hbw_get_kind(HBW_PAGESIZE_4KB), memptr, alignment,
+                                  size);
+}
+
+MEMKIND_EXPORT int hbw_posix_memalign_psize(void **memptr, size_t alignment,
+                                            size_t size,
+                                            hbw_pagesize_t pagesize)
+{
+    if (pagesize == HBW_PAGESIZE_1GB_STRICT && size % (1 << 30)) {
+        return EINVAL;
+    }
+
+    if((pagesize == HBW_PAGESIZE_2MB ||
+        pagesize == HBW_PAGESIZE_1GB_STRICT ||
+        pagesize == HBW_PAGESIZE_1GB) &&
+       hbw_get_policy() == HBW_POLICY_INTERLEAVE) {
+
+        log_err("HBW_POLICY_INTERLEAVE is unsupported with used page size!");
+        return EINVAL;
+    }
+
+    return memkind_posix_memalign(hbw_get_kind(pagesize), memptr, alignment, size);
+}
+
+MEMKIND_EXPORT void *hbw_realloc(void *ptr, size_t size)
+{
+    return memkind_realloc(hbw_get_kind(HBW_PAGESIZE_4KB), ptr, size);
+}
+
+MEMKIND_EXPORT void hbw_free(void *ptr)
+{
+    memkind_free(0, ptr);
+}
+
+MEMKIND_EXPORT size_t hbw_malloc_usable_size(void *ptr)
+{
+    return memkind_malloc_usable_size(0, ptr);
+}
diff --git a/src/heap_manager.c b/src/heap_manager.c
new file mode 100644
index 0000000..f6c8678
--- /dev/null
+++ b/src/heap_manager.c
@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/heap_manager.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/memkind_arena.h>
+
+#include <stdio.h>
+#include <pthread.h>
+#include <string.h>
+
+static struct heap_manager_ops *heap_manager_g;
+
+static pthread_once_t heap_manager_init_once_g = PTHREAD_ONCE_INIT;
+
+struct heap_manager_ops {
+    void (*init)(struct memkind *kind);
+    size_t (*heap_manager_malloc_usable_size)(void *ptr);
+    void (*heap_manager_free)(void *ptr);
+    void *(*heap_manager_realloc)(void *ptr, size_t size);
+    struct memkind *(*heap_manager_detect_kind)(void *ptr);
+    int (*heap_manager_update_cached_stats)(void);
+    int (*heap_manager_get_stat)(memkind_stat_type stat, size_t *value);
+    void *(*heap_manager_defrag_reallocate)(void *ptr);
+};
+
+static struct heap_manager_ops arena_heap_manager_g = {
+    .init = memkind_arena_init,
+    .heap_manager_malloc_usable_size = memkind_arena_malloc_usable_size,
+    .heap_manager_free = memkind_arena_free_with_kind_detect,
+    .heap_manager_realloc = memkind_arena_realloc_with_kind_detect,
+    .heap_manager_detect_kind = memkind_arena_detect_kind,
+    .heap_manager_update_cached_stats = memkind_arena_update_cached_stats,
+    .heap_manager_get_stat = memkind_arena_get_global_stat,
+    .heap_manager_defrag_reallocate = memkind_arena_defrag_reallocate_with_kind_detect
+};
+
+static struct heap_manager_ops tbb_heap_manager_g = {
+    .init = tbb_initialize,
+    .heap_manager_malloc_usable_size = tbb_pool_malloc_usable_size_with_kind_detect,
+    .heap_manager_free = tbb_pool_free_with_kind_detect,
+    .heap_manager_realloc = tbb_pool_realloc_with_kind_detect,
+    .heap_manager_detect_kind = tbb_detect_kind,
+    .heap_manager_update_cached_stats = tbb_update_cached_stats,
+    .heap_manager_get_stat = tbb_get_global_stat,
+    .heap_manager_defrag_reallocate = tbb_pool_defrag_reallocate_with_kind_detect
+};
+
+static void set_heap_manager()
+{
+    heap_manager_g = &arena_heap_manager_g;
+    const char *env = secure_getenv("MEMKIND_HEAP_MANAGER");
+    if(env && strcmp(env, "TBB") == 0) {
+        heap_manager_g = &tbb_heap_manager_g;
+    }
+}
+
+static inline struct heap_manager_ops *get_heap_manager()
+{
+    pthread_once(&heap_manager_init_once_g, set_heap_manager);
+    return heap_manager_g;
+}
+
+void heap_manager_init(struct memkind *kind)
+{
+    get_heap_manager()->init(kind);
+}
+
+size_t heap_manager_malloc_usable_size(void *ptr)
+{
+    return get_heap_manager()->heap_manager_malloc_usable_size(ptr);
+}
+
+void heap_manager_free(void *ptr)
+{
+    get_heap_manager()->heap_manager_free(ptr);
+}
+
+void *heap_manager_realloc(void *ptr, size_t size)
+{
+    return get_heap_manager()->heap_manager_realloc(ptr, size);
+}
+
+struct memkind *heap_manager_detect_kind(void *ptr)
+{
+    return get_heap_manager()->heap_manager_detect_kind(ptr);
+}
+
+int heap_manager_update_cached_stats(void)
+{
+    return get_heap_manager()->heap_manager_update_cached_stats();
+}
+
+int heap_manager_get_stat(memkind_stat_type stat, size_t *value)
+{
+    return get_heap_manager()->heap_manager_get_stat(stat, value);
+}
+
+void *heap_manager_defrag_reallocate(void *ptr)
+{
+    return get_heap_manager()->heap_manager_defrag_reallocate(ptr);
+}
diff --git a/src/memkind-auto-dax-kmem-nodes.c b/src/memkind-auto-dax-kmem-nodes.c
new file mode 100644
index 0000000..06c30dc
--- /dev/null
+++ b/src/memkind-auto-dax-kmem-nodes.c
@@ -0,0 +1,108 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_dax_kmem.h>
+#include <memkind.h>
+#include <numa.h>
+#include <stdio.h>
+
+#define MAX_ARG_LEN 8
+
+static const char *const help_message =
+    "\n"
+    "NAME\n"
+    "    memkind-auto-dax-kmem-nodes - prints comma-separated list of automatically detected persistent memory NUMA nodes.\n"
+    "\n"
+    "SYNOPSIS\n"
+    "    memkind-auto-dax-kmem-nodes -h | --help\n"
+    "        Prints this help message.\n"
+    "\n"
+    "DESCRIPTION\n"
+    "    Prints a comma-separated list of automatically detected persistent memory NUMA nodes\n"
+    "    that can be used with the numactl --membind option.\n"
+    "\n"
+    "EXIT STATUS\n"
+    "    Return code is :\n"
+    "        0 on success\n"
+    "        1 if automatic detection is disabled (memkind wasn't build with suitable libdaxctl-devel version)\n"
+    "        2 on invalid argument\n"
+    "        3 on other failure\n"
+    "\n"
+    "COPYRIGHT\n"
+    "    Copyright 2019 Intel Corporation All Rights Reserved.\n"
+    "\n"
+    "AUTHORS\n"
+    "    Michal Biesek\n"
+    "\n"
+    "SEE ALSO\n"
+    "    memkind(3)\n"
+    "\n";
+
+
+static int print_dax_kmem_nodes()
+{
+    unsigned i, j = 0;
+
+    nodemask_t nodemask;
+    struct bitmask nodemask_dax_kmem = {NUMA_NUM_NODES, nodemask.n};
+
+    // ensuring functions in numa library are defined
+    if (numa_available() == -1) {
+        return 3;
+    }
+    numa_bitmask_clearall(&nodemask_dax_kmem);
+
+    //WARNING: code below is usage of memkind experimental API which may be changed in future
+    int status = memkind_dax_kmem_all_get_mbind_nodemask(NULL, nodemask.n,
+                                                         NUMA_NUM_NODES);
+
+    if (status == MEMKIND_ERROR_OPERATION_FAILED ) {
+        return 1;
+    } else if (status != MEMKIND_SUCCESS ) {
+        return 3;
+    }
+
+    for(i = 0; i<NUMA_NUM_NODES; i++) {
+        if(numa_bitmask_isbitset(&nodemask_dax_kmem, i)) {
+            printf("%u%s", i, (++j == numa_bitmask_weight(&nodemask_dax_kmem)) ? "" : ",");
+        }
+    }
+    printf("\n");
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+    if(argc == 1) {
+        return print_dax_kmem_nodes();
+    } else if ((argc == 2) && (strncmp(argv[1], "-h", MAX_ARG_LEN) == 0 ||
+                               strncmp(argv[1], "--help", MAX_ARG_LEN) == 0)) {
+        printf("%s", help_message);
+        return 0;
+    }
+
+    printf("ERROR: Unknown option %s. More info with \"%s --help\".\n", argv[1],
+           argv[0]);
+    return 2;
+}
diff --git a/src/memkind-hbw-nodes.c b/src/memkind-hbw-nodes.c
new file mode 100644
index 0000000..9b3fac4
--- /dev/null
+++ b/src/memkind-hbw-nodes.c
@@ -0,0 +1,103 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+
+#include <numa.h>
+#include <stdio.h>
+
+#define MAX_ARG_LEN 8
+
+const char *help_message =
+    "\n"
+    "NAME\n"
+    "    memkind-hbw-nodes - Prints comma-separated list of high bandwidth nodes.\n"
+    "\n"
+    "SYNOPSIS\n"
+    "    memkind-hbw-nodes -h | --help\n"
+    "        Prints this help message.\n"
+    "\n"
+    "DESCRIPTION\n"
+    "    Prints a comma-separated list of high bandwidth NUMA nodes\n"
+    "    that can be used with the numactl --membind option.\n"
+    "\n"
+    "EXIT STATUS\n"
+    "    Return code is :\n"
+    "        0 on success\n"
+    "        1 on failure\n"
+    "        2 on invalid argument\n"
+    "\n"
+    "COPYRIGHT\n"
+    "    Copyright 2016 Intel Corporation All Rights Reserved.\n"
+    "\n"
+    "AUTHORS\n"
+    "    Krzysztof Kulakowski\n"
+    "\n"
+    "SEE ALSO\n"
+    "    hbwmalloc(3), memkind(3)\n"
+    "\n";
+
+extern unsigned int numa_bitmask_weight(const struct bitmask *bmp );
+
+int print_hbw_nodes()
+{
+    int i, j = 0;
+
+    nodemask_t nodemask;
+    struct bitmask nodemask_bm = {NUMA_NUM_NODES, nodemask.n};
+
+    // ensuring functions in numa library are defined
+    if (numa_available() == -1) {
+        return 3;
+    }
+    numa_bitmask_clearall(&nodemask_bm);
+
+    //WARNING: code below is usage of memkind experimental API which may be changed in future
+    if(memkind_hbw_all_get_mbind_nodemask(NULL, nodemask.n, NUMA_NUM_NODES) != 0) {
+        return 1;
+    }
+
+    for(i=0; i<NUMA_NUM_NODES; i++) {
+        if(numa_bitmask_isbitset(&nodemask_bm, i)) {
+            printf("%d%s", i, (++j == numa_bitmask_weight(&nodemask_bm)) ? "" : ",");
+        }
+    }
+    printf("\n");
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+    if(argc == 1) {
+        return print_hbw_nodes();
+    } else if ((argc == 2) && (strncmp(argv[1], "-h", MAX_ARG_LEN) == 0 ||
+                               strncmp(argv[1], "--help", MAX_ARG_LEN) == 0)) {
+        printf("%s", help_message);
+        return 2;
+    }
+
+    printf("ERROR: Unknown option %s. More info with \"%s --help\".\n", argv[1],
+           argv[0]);
+    return 2;
+}
diff --git a/src/memkind.c b/src/memkind.c
new file mode 100644
index 0000000..d3f169d
--- /dev/null
+++ b/src/memkind.c
@@ -0,0 +1,924 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#define MEMKIND_VERSION_MAJOR 1
+#define MEMKIND_VERSION_MINOR 10
+#define MEMKIND_VERSION_PATCH 0
+
+#include <memkind.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_regular.h>
+#include <memkind/internal/memkind_gbtlb.h>
+#include <memkind/internal/memkind_dax_kmem.h>
+#include <memkind/internal/memkind_pmem.h>
+#include <memkind/internal/memkind_interleave.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/heap_manager.h>
+
+#include "config.h"
+
+#include <numa.h>
+#include <sys/param.h>
+#include <sys/mman.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <assert.h>
+#include <pthread.h>
+#include <errno.h>
+#include <signal.h>
+#include <fcntl.h>
+#include <unistd.h>
+
+/* Clear bits in x, but only this specified in mask. */
+#define CLEAR_BIT(x, mask) ((x) &= (~(mask)))
+
+extern struct memkind_ops MEMKIND_HBW_GBTLB_OPS;
+extern struct memkind_ops MEMKIND_HBW_PREFERRED_GBTLB_OPS;
+extern struct memkind_ops MEMKIND_GBTLB_OPS;
+
+static struct memkind MEMKIND_DEFAULT_STATIC = {
+    .ops =  &MEMKIND_DEFAULT_OPS,
+    .partition = MEMKIND_PARTITION_DEFAULT,
+    .name = "memkind_default",
+    .init_once = PTHREAD_ONCE_INIT,
+    .arena_zero = 0,
+    .arena_map_len = ARENA_LIMIT_DEFAULT_KIND - 1,
+};
+
+static struct memkind MEMKIND_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HUGETLB,
+    .name = "memkind_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_INTERLEAVE_STATIC = {
+    .ops = &MEMKIND_INTERLEAVE_OPS,
+    .partition = MEMKIND_PARTITION_INTERLEAVE,
+    .name = "memkind_interleave",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_STATIC = {
+    .ops = &MEMKIND_HBW_OPS,
+    .partition = MEMKIND_PARTITION_HBW,
+    .name = "memkind_hbw",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_ALL_STATIC = {
+    .ops = &MEMKIND_HBW_ALL_OPS,
+    .partition = MEMKIND_PARTITION_HBW_ALL,
+    .name = "memkind_hbw_all",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_PREFERRED_STATIC = {
+    .ops = &MEMKIND_HBW_PREFERRED_OPS,
+    .partition = MEMKIND_PARTITION_HBW_PREFERRED,
+    .name = "memkind_hbw_preferred",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HBW_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_HUGETLB,
+    .name = "memkind_hbw_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_ALL_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HBW_ALL_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_ALL_HUGETLB,
+    .name = "memkind_hbw_all_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_PREFERRED_HUGETLB_STATIC = {
+    .ops = &MEMKIND_HBW_PREFERRED_HUGETLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_PREFERRED_HUGETLB,
+    .name = "memkind_hbw_preferred_hugetlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_GBTLB_STATIC = {
+    .ops = &MEMKIND_HBW_GBTLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_GBTLB,
+    .name = "memkind_hbw_gbtlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_PREFERRED_GBTLB_STATIC = {
+    .ops = &MEMKIND_HBW_PREFERRED_GBTLB_OPS,
+    .partition = MEMKIND_PARTITION_HBW_PREFERRED_GBTLB,
+    .name = "memkind_hbw_preferred_gbtlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_GBTLB_STATIC = {
+    .ops = &MEMKIND_GBTLB_OPS,
+    .partition = MEMKIND_PARTITION_GBTLB,
+    .name = "memkind_gbtlb",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_HBW_INTERLEAVE_STATIC = {
+    .ops = &MEMKIND_HBW_INTERLEAVE_OPS,
+    .partition = MEMKIND_PARTITION_HBW_INTERLEAVE,
+    .name = "memkind_hbw_interleave",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_REGULAR_STATIC = {
+    .ops = &MEMKIND_REGULAR_OPS,
+    .partition = MEMKIND_PARTITION_REGULAR,
+    .name = "memkind_regular",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_DAX_KMEM_STATIC = {
+    .ops = &MEMKIND_DAX_KMEM_OPS,
+    .partition = MEMKIND_PARTITION_DAX_KMEM,
+    .name = "memkind_dax_kmem",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_DAX_KMEM_ALL_STATIC = {
+    .ops = &MEMKIND_DAX_KMEM_ALL_OPS,
+    .partition = MEMKIND_PARTITION_DAX_KMEM_ALL,
+    .name = "memkind_dax_kmem_all",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+static struct memkind MEMKIND_DAX_KMEM_PREFERRED_STATIC = {
+    .ops = &MEMKIND_DAX_KMEM_PREFERRED_OPS,
+    .partition = MEMKIND_PARTITION_DAX_KMEM_PREFERRED,
+    .name = "memkind_dax_kmem_preferred",
+    .init_once = PTHREAD_ONCE_INIT,
+};
+
+MEMKIND_EXPORT struct memkind *MEMKIND_DEFAULT = &MEMKIND_DEFAULT_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HUGETLB = &MEMKIND_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_INTERLEAVE = &MEMKIND_INTERLEAVE_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW = &MEMKIND_HBW_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_ALL = &MEMKIND_HBW_ALL_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_PREFERRED =
+        &MEMKIND_HBW_PREFERRED_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_HUGETLB =
+        &MEMKIND_HBW_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_ALL_HUGETLB =
+        &MEMKIND_HBW_ALL_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_PREFERRED_HUGETLB =
+        &MEMKIND_HBW_PREFERRED_HUGETLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_GBTLB = &MEMKIND_HBW_GBTLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_PREFERRED_GBTLB =
+        &MEMKIND_HBW_PREFERRED_GBTLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_GBTLB = &MEMKIND_GBTLB_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_HBW_INTERLEAVE =
+        &MEMKIND_HBW_INTERLEAVE_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_REGULAR = &MEMKIND_REGULAR_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_DAX_KMEM = &MEMKIND_DAX_KMEM_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_DAX_KMEM_ALL =
+        &MEMKIND_DAX_KMEM_ALL_STATIC;
+MEMKIND_EXPORT struct memkind *MEMKIND_DAX_KMEM_PREFERRED =
+        &MEMKIND_DAX_KMEM_PREFERRED_STATIC;
+
+struct memkind_registry {
+    struct memkind *partition_map[MEMKIND_MAX_KIND];
+    int num_kind;
+    pthread_mutex_t lock;
+};
+
+static struct memkind_registry memkind_registry_g = {
+    {
+        [MEMKIND_PARTITION_DEFAULT] = &MEMKIND_DEFAULT_STATIC,
+        [MEMKIND_PARTITION_HBW] = &MEMKIND_HBW_STATIC,
+        [MEMKIND_PARTITION_HBW_PREFERRED] = &MEMKIND_HBW_PREFERRED_STATIC,
+        [MEMKIND_PARTITION_HBW_HUGETLB] = &MEMKIND_HBW_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_HBW_PREFERRED_HUGETLB] = &MEMKIND_HBW_PREFERRED_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_HUGETLB] = &MEMKIND_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_HBW_GBTLB] = &MEMKIND_HBW_GBTLB_STATIC,
+        [MEMKIND_PARTITION_HBW_PREFERRED_GBTLB] = &MEMKIND_HBW_PREFERRED_GBTLB_STATIC,
+        [MEMKIND_PARTITION_GBTLB] = &MEMKIND_GBTLB_STATIC,
+        [MEMKIND_PARTITION_HBW_INTERLEAVE] = &MEMKIND_HBW_INTERLEAVE_STATIC,
+        [MEMKIND_PARTITION_INTERLEAVE] = &MEMKIND_INTERLEAVE_STATIC,
+        [MEMKIND_PARTITION_REGULAR] = &MEMKIND_REGULAR_STATIC,
+        [MEMKIND_PARTITION_HBW_ALL] = &MEMKIND_HBW_ALL_STATIC,
+        [MEMKIND_PARTITION_HBW_ALL_HUGETLB] = &MEMKIND_HBW_ALL_HUGETLB_STATIC,
+        [MEMKIND_PARTITION_DAX_KMEM] = &MEMKIND_DAX_KMEM_STATIC,
+        [MEMKIND_PARTITION_DAX_KMEM_ALL] = &MEMKIND_DAX_KMEM_ALL_STATIC,
+        [MEMKIND_PARTITION_DAX_KMEM_PREFERRED] = &MEMKIND_DAX_KMEM_PREFERRED_STATIC,
+    },
+    MEMKIND_NUM_BASE_KIND,
+    PTHREAD_MUTEX_INITIALIZER
+};
+
+void *kind_mmap(struct memkind *kind, void *addr, size_t size)
+{
+    if (MEMKIND_LIKELY(kind->ops->mmap == NULL)) {
+        return memkind_default_mmap(kind, addr, size);
+    } else {
+        return kind->ops->mmap(kind, addr, size);
+    }
+}
+
+
+static int validate_memtype_bits(memkind_memtype_t memtype)
+{
+    if(memtype == 0) return -1;
+
+    CLEAR_BIT(memtype, MEMKIND_MEMTYPE_DEFAULT);
+    CLEAR_BIT(memtype, MEMKIND_MEMTYPE_HIGH_BANDWIDTH);
+
+    if(memtype != 0) return -1;
+    return 0;
+}
+
+static int validate_flags_bits(memkind_bits_t flags)
+{
+    CLEAR_BIT(flags, MEMKIND_MASK_PAGE_SIZE_2MB);
+
+    if(flags != 0) return -1;
+    return 0;
+}
+
+static int validate_policy(memkind_policy_t policy)
+{
+    if((policy >= 0) && (policy < MEMKIND_POLICY_MAX_VALUE)) return 0;
+    return -1;
+}
+
+struct create_args {
+    memkind_t kind;
+    memkind_policy_t policy;
+    memkind_bits_t flags;
+    memkind_memtype_t memtype_flags;
+};
+
+static struct create_args supported_args[] = {
+
+    {&MEMKIND_HBW_STATIC,                    MEMKIND_POLICY_BIND_LOCAL,      0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_HUGETLB_STATIC,            MEMKIND_POLICY_BIND_LOCAL,      MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_ALL_STATIC,                MEMKIND_POLICY_BIND_ALL,        0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_ALL_HUGETLB_STATIC,        MEMKIND_POLICY_BIND_ALL,        MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_PREFERRED_STATIC,          MEMKIND_POLICY_PREFERRED_LOCAL, 0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_PREFERRED_HUGETLB_STATIC,  MEMKIND_POLICY_PREFERRED_LOCAL, MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_HBW_INTERLEAVE_STATIC,         MEMKIND_POLICY_INTERLEAVE_ALL,  0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH},
+    {&MEMKIND_DEFAULT_STATIC,                MEMKIND_POLICY_PREFERRED_LOCAL, 0,                          MEMKIND_MEMTYPE_DEFAULT},
+    {&MEMKIND_HUGETLB_STATIC,                MEMKIND_POLICY_PREFERRED_LOCAL, MEMKIND_MASK_PAGE_SIZE_2MB, MEMKIND_MEMTYPE_DEFAULT},
+    {&MEMKIND_INTERLEAVE_STATIC,             MEMKIND_POLICY_INTERLEAVE_ALL,  0,                          MEMKIND_MEMTYPE_HIGH_BANDWIDTH | MEMKIND_MEMTYPE_DEFAULT},
+};
+
+/* Kind creation */
+MEMKIND_EXPORT int memkind_create_kind(memkind_memtype_t memtype_flags,
+                                       memkind_policy_t policy,
+                                       memkind_bits_t flags,
+                                       memkind_t *kind)
+{
+    if(validate_memtype_bits(memtype_flags) != 0) {
+        log_err("Cannot create kind: incorrect memtype_flags.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if(validate_flags_bits(flags) != 0) {
+        log_err("Cannot create kind: incorrect flags.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if(validate_policy(policy) != 0) {
+        log_err("Cannot create kind: incorrect policy.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if(kind == NULL) {
+        log_err("Cannot create kind: 'kind' is NULL pointer.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    int i, num_supported_args = sizeof(supported_args) / sizeof(struct create_args);
+    for(i = 0; i < num_supported_args; i++) {
+        if((supported_args[i].memtype_flags == memtype_flags) &&
+           (supported_args[i].policy == policy) &&
+           (supported_args[i].flags == flags)) {
+
+            if(memkind_check_available(supported_args[i].kind) == 0) {
+                *kind = supported_args[i].kind;
+                return MEMKIND_SUCCESS;
+            } else if(policy == MEMKIND_POLICY_PREFERRED_LOCAL) {
+                *kind = MEMKIND_DEFAULT;
+                return MEMKIND_SUCCESS;
+            }
+            log_err("Cannot create kind: requested memory type is not available.");
+            return MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE;
+        }
+    }
+
+    log_err("Cannot create kind: unsupported set of capabilities.");
+    return MEMKIND_ERROR_INVALID;
+}
+
+static void memkind_destroy_dynamic_kind_from_register(unsigned int i,
+                                                       memkind_t kind)
+{
+    if (i >= MEMKIND_NUM_BASE_KIND) {
+        memkind_registry_g.partition_map[i] = NULL;
+        --memkind_registry_g.num_kind;
+        free(kind);
+    }
+}
+
+/* Kind destruction. */
+MEMKIND_EXPORT int memkind_destroy_kind(memkind_t kind)
+{
+    if (pthread_mutex_lock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to acquire mutex");
+    unsigned i;
+    int err = kind->ops->destroy(kind);
+    for (i = MEMKIND_NUM_BASE_KIND; i < MEMKIND_MAX_KIND; ++i) {
+        if (memkind_registry_g.partition_map[i] &&
+            strcmp(kind->name, memkind_registry_g.partition_map[i]->name) == 0) {
+            memkind_destroy_dynamic_kind_from_register(i, kind);
+            break;
+        }
+    }
+    if (pthread_mutex_unlock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to release mutex");
+    return err;
+}
+
+MEMKIND_EXPORT memkind_t memkind_detect_kind(void *ptr)
+{
+    return heap_manager_detect_kind(ptr);
+}
+
+/* Declare weak symbols for allocator decorators */
+extern void memkind_malloc_pre(struct memkind **,
+                               size_t *) __attribute__((weak));
+extern void memkind_malloc_post(struct memkind *, size_t,
+                                void **) __attribute__((weak));
+extern void memkind_calloc_pre(struct memkind **, size_t *,
+                               size_t *) __attribute__((weak));
+extern void memkind_calloc_post(struct memkind *, size_t, size_t,
+                                void **) __attribute__((weak));
+extern void memkind_posix_memalign_pre(struct memkind **, void **, size_t *,
+                                       size_t *) __attribute__((weak));
+extern void memkind_posix_memalign_post(struct memkind *, void **, size_t,
+                                        size_t, int *) __attribute__((weak));
+extern void memkind_realloc_pre(struct memkind **, void **,
+                                size_t *) __attribute__((weak));
+extern void memkind_realloc_post(struct memkind *, void *, size_t,
+                                 void **) __attribute__((weak));
+extern void memkind_free_pre(struct memkind **, void **) __attribute__((weak));
+extern void memkind_free_post(struct memkind *, void *) __attribute__((weak));
+
+MEMKIND_EXPORT int memkind_get_version()
+{
+    return MEMKIND_VERSION_MAJOR * 1000000 + MEMKIND_VERSION_MINOR * 1000 +
+           MEMKIND_VERSION_PATCH;
+}
+
+MEMKIND_EXPORT void memkind_error_message(int err, char *msg, size_t size)
+{
+    switch (err) {
+        case MEMKIND_ERROR_UNAVAILABLE:
+            strncpy(msg, "<memkind> Requested memory kind is not available", size);
+            break;
+        case MEMKIND_ERROR_MBIND:
+            strncpy(msg, "<memkind> Call to mbind() failed", size);
+            break;
+        case MEMKIND_ERROR_MMAP:
+            strncpy(msg, "<memkind> Call to mmap() failed", size);
+            break;
+        case MEMKIND_ERROR_MALLOC:
+            strncpy(msg, "<memkind> Call to malloc() failed", size);
+            break;
+        case MEMKIND_ERROR_ENVIRON:
+            strncpy(msg, "<memkind> Error parsing environment variable (MEMKIND_*)", size);
+            break;
+        case MEMKIND_ERROR_INVALID:
+            strncpy(msg, "<memkind> Invalid input arguments to memkind routine", size);
+            break;
+        case MEMKIND_ERROR_TOOMANY:
+            snprintf(msg, size,
+                     "<memkind> Attempted to initialize more than maximum (%i) number of kinds",
+                     MEMKIND_MAX_KIND);
+            break;
+        case MEMKIND_ERROR_RUNTIME:
+            strncpy(msg, "<memkind> Unspecified run-time error", size);
+            break;
+        case EINVAL:
+            strncpy(msg,
+                    "<memkind> Alignment must be a power of two and larger than sizeof(void *)",
+                    size);
+            break;
+        case ENOMEM:
+            strncpy(msg, "<memkind> Call to jemk_mallocx() failed", size);
+            break;
+        case MEMKIND_ERROR_HUGETLB:
+            strncpy(msg, "<memkind> unable to allocate huge pages", size);
+            break;
+        case MEMKIND_ERROR_BADOPS:
+            strncpy(msg,
+                    "<memkind> memkind_ops structure is poorly formed (missing or incorrect functions)",
+                    size);
+            break;
+        case MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE:
+            strncpy(msg, "<memkind> Requested memory type is not available", size);
+            break;
+        case MEMKIND_ERROR_OPERATION_FAILED:
+            strncpy(msg, "<memkind> Operation failed", size);
+            break;
+        case MEMKIND_ERROR_ARENAS_CREATE:
+            strncpy(msg, "<memkind> Call to jemalloc's arenas.create () failed", size);
+            break;
+        default:
+            snprintf(msg, size, "<memkind> Undefined error number: %i", err);
+            break;
+    }
+    if (size > 0) {
+        msg[size-1] = '\0';
+    }
+}
+
+void memkind_init(memkind_t kind, bool check_numa)
+{
+    log_info("Initializing kind %s.", kind->name);
+    heap_manager_init(kind);
+    if (check_numa) {
+        int err = numa_available();
+        if (err) {
+            log_fatal("[%s] NUMA not available (error code:%d).", kind->name, err);
+            abort();
+        }
+    }
+}
+
+static void nop(void) {}
+
+static int memkind_create(struct memkind_ops *ops, const char *name,
+                          struct memkind **kind)
+{
+    int err;
+    unsigned i;
+    unsigned id_kind = 0;
+
+    *kind = NULL;
+    if (pthread_mutex_lock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to acquire mutex");
+
+    if (memkind_registry_g.num_kind == MEMKIND_MAX_KIND) {
+        log_err("Attempted to initialize more than maximum (%i) number of kinds.",
+                MEMKIND_MAX_KIND);
+        err = MEMKIND_ERROR_TOOMANY;
+        goto exit;
+    }
+    if (ops->create == NULL ||
+        ops->destroy == NULL ||
+        ops->malloc == NULL ||
+        ops->calloc == NULL ||
+        ops->realloc == NULL ||
+        ops->posix_memalign == NULL ||
+        ops->free == NULL ||
+        ops->init_once != NULL) {
+        err = MEMKIND_ERROR_BADOPS;
+        goto exit;
+    }
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        if (memkind_registry_g.partition_map[i] == NULL) {
+            id_kind = i;
+            break;
+        } else if (strcmp(name, memkind_registry_g.partition_map[i]->name) == 0) {
+            log_err("Kind with the name %s already exists", name);
+            err = MEMKIND_ERROR_INVALID;
+            goto exit;
+        }
+    }
+    *kind = (struct memkind *)calloc(1, sizeof(struct memkind));
+    if (!*kind) {
+        err = MEMKIND_ERROR_MALLOC;
+        log_err("calloc() failed.");
+        goto exit;
+    }
+
+    (*kind)->partition = memkind_registry_g.num_kind;
+    err = ops->create(*kind, ops, name);
+    if (err) {
+        free(*kind);
+        goto exit;
+    }
+    memkind_registry_g.partition_map[id_kind] = *kind;
+    ++memkind_registry_g.num_kind;
+
+    (*kind)->init_once = PTHREAD_ONCE_INIT;
+    pthread_once(&(*kind)->init_once,
+                 nop); //this is done to avoid init_once for dynamic kinds
+exit:
+    if (pthread_mutex_unlock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to release mutex");
+
+    return err;
+}
+
+#ifdef __GNUC__
+__attribute__((constructor))
+#endif
+static void memkind_construct(void)
+{
+    const char *env = secure_getenv("MEMKIND_HEAP_MANAGER");
+    if (env && strcmp(env, "TBB") == 0) {
+        load_tbb_symbols();
+    } else {
+        env = secure_getenv("MEMKIND_BACKGROUND_THREAD_LIMIT");
+        if (env) {
+            char *end;
+            errno = 0;
+            size_t thread_limit = strtoull(env, &end, 10);
+            if (*end != '\0' || errno != 0 ) {
+                log_err("Error on parsing value MEMKIND_BACKGROUND_THREAD_LIMIT");
+                return;
+            }
+            memkind_arena_enable_background_threads(thread_limit);
+        }
+    }
+}
+
+#ifdef __GNUC__
+__attribute__((destructor))
+#endif
+static int memkind_finalize(void)
+{
+    struct memkind *kind;
+    unsigned i;
+    int err = MEMKIND_SUCCESS;
+
+    if (pthread_mutex_lock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to acquire mutex");
+
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        kind = memkind_registry_g.partition_map[i];
+        if (kind && kind->ops->finalize) {
+            err = kind->ops->finalize(kind);
+            if (err) {
+                goto exit;
+            }
+            memkind_destroy_dynamic_kind_from_register(i, kind);
+        }
+    }
+    assert(memkind_registry_g.num_kind == MEMKIND_NUM_BASE_KIND);
+
+exit:
+    if (pthread_mutex_unlock(&memkind_registry_g.lock) != 0)
+        assert(0 && "failed to release mutex");
+
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_check_available(struct memkind *kind)
+{
+    int err = MEMKIND_SUCCESS;
+
+    if (MEMKIND_LIKELY(kind->ops->check_available)) {
+        err = kind->ops->check_available(kind);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT size_t memkind_malloc_usable_size(struct memkind *kind,
+                                                 void *ptr)
+{
+    if (!kind) {
+        return heap_manager_malloc_usable_size(ptr);
+    } else {
+        return kind->ops->malloc_usable_size(kind, ptr);
+    }
+}
+
+MEMKIND_EXPORT void *memkind_malloc(struct memkind *kind, size_t size)
+{
+    void *result;
+
+    pthread_once(&kind->init_once, kind->ops->init_once);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_malloc_pre) {
+        memkind_malloc_pre(&kind, &size);
+    }
+#endif
+
+    result = kind->ops->malloc(kind, size);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_malloc_post) {
+        memkind_malloc_post(kind, size, &result);
+    }
+#endif
+
+    return result;
+}
+
+MEMKIND_EXPORT void *memkind_calloc(struct memkind *kind, size_t num,
+                                    size_t size)
+{
+    void *result;
+
+    pthread_once(&kind->init_once, kind->ops->init_once);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_calloc_pre) {
+        memkind_calloc_pre(&kind, &num, &size);
+    }
+#endif
+
+    result = kind->ops->calloc(kind, num, size);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_calloc_post) {
+        memkind_calloc_post(kind, num, size, &result);
+    }
+#endif
+
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_posix_memalign(struct memkind *kind, void **memptr,
+                                          size_t alignment,
+                                          size_t size)
+{
+    int err;
+
+    pthread_once(&kind->init_once, kind->ops->init_once);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_posix_memalign_pre) {
+        memkind_posix_memalign_pre(&kind, memptr, &alignment, &size);
+    }
+#endif
+
+    err = kind->ops->posix_memalign(kind, memptr, alignment, size);
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_posix_memalign_post) {
+        memkind_posix_memalign_post(kind, memptr, alignment, size, &err);
+    }
+#endif
+
+    return err;
+}
+
+MEMKIND_EXPORT void *memkind_realloc(struct memkind *kind, void *ptr,
+                                     size_t size)
+{
+    void *result;
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_realloc_pre) {
+        memkind_realloc_pre(&kind, &ptr, &size);
+    }
+#endif
+
+    if (!kind) {
+        result = heap_manager_realloc(ptr, size);
+    } else {
+        pthread_once(&kind->init_once, kind->ops->init_once);
+        result = kind->ops->realloc(kind, ptr, size);
+    }
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_realloc_post) {
+        memkind_realloc_post(kind, ptr, size, &result);
+    }
+#endif
+
+    return result;
+}
+
+MEMKIND_EXPORT void memkind_free(struct memkind *kind, void *ptr)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_free_pre) {
+        memkind_free_pre(&kind, &ptr);
+    }
+#endif
+    if (!kind) {
+        heap_manager_free(ptr);
+    } else {
+        pthread_once(&kind->init_once, kind->ops->init_once);
+        kind->ops->free(kind, ptr);
+    }
+
+#ifdef MEMKIND_DECORATION_ENABLED
+    if (memkind_free_post) {
+        memkind_free_post(kind, ptr);
+    }
+#endif
+}
+
+static int memkind_tmpfile(const char *dir, int *fd)
+{
+    static char template[] = "/memkind.XXXXXX";
+    int err = MEMKIND_SUCCESS;
+    int oerrno;
+    int dir_len = strlen(dir);
+
+    if (dir_len > PATH_MAX) {
+        log_err("Could not create temporary file: too long path.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    char fullname[dir_len + sizeof (template)];
+    (void) strcpy(fullname, dir);
+    (void) strcat(fullname, template);
+
+    sigset_t set, oldset;
+    sigfillset(&set);
+    (void) sigprocmask(SIG_BLOCK, &set, &oldset);
+
+    if ((*fd = mkstemp(fullname)) < 0) {
+        log_err("Could not create temporary file: errno=%d.", errno);
+        err = MEMKIND_ERROR_INVALID;
+        goto exit;
+    }
+
+    (void) unlink(fullname);
+    (void) sigprocmask(SIG_SETMASK, &oldset, NULL);
+
+    return err;
+
+exit:
+    oerrno = errno;
+    (void) sigprocmask(SIG_SETMASK, &oldset, NULL);
+    if (*fd != -1) {
+        (void) close(*fd);
+    }
+    *fd = -1;
+    errno = oerrno;
+    return err;
+}
+
+MEMKIND_EXPORT struct memkind_config *memkind_config_new(void)
+{
+    struct memkind_config *cfg = (struct memkind_config *)malloc(sizeof(
+                                                                     struct memkind_config));
+    return cfg;
+}
+
+MEMKIND_EXPORT void memkind_config_delete(struct memkind_config *cfg)
+{
+    free(cfg);
+}
+
+MEMKIND_EXPORT void memkind_config_set_path(struct memkind_config *cfg,
+                                            const char *pmem_dir)
+{
+    cfg->pmem_dir = pmem_dir;
+}
+
+MEMKIND_EXPORT void memkind_config_set_size(struct memkind_config *cfg,
+                                            size_t pmem_size)
+{
+    cfg->pmem_size = pmem_size;
+}
+
+MEMKIND_EXPORT void memkind_config_set_memory_usage_policy(
+    struct memkind_config *cfg, memkind_mem_usage_policy policy)
+{
+    cfg->policy = policy;
+}
+
+MEMKIND_EXPORT int memkind_create_pmem(const char *dir, size_t max_size,
+                                       struct memkind **kind)
+{
+    int oerrno;
+
+    if (max_size && max_size < MEMKIND_PMEM_MIN_SIZE) {
+        log_err("Cannot create pmem: invalid size.");
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if (max_size) {
+        /* round up to a multiple of jemalloc chunk size */
+        max_size = roundup(max_size, MEMKIND_PMEM_CHUNK_SIZE);
+    }
+
+    int fd = -1;
+    char name[16];
+
+    int err = memkind_tmpfile(dir, &fd);
+    if (err) {
+        goto exit;
+    }
+
+    snprintf(name, sizeof (name), "pmem%08x", fd);
+
+    err = memkind_create(&MEMKIND_PMEM_OPS, name, kind);
+    if (err) {
+        goto exit;
+    }
+
+    struct memkind_pmem *priv = (*kind)->priv;
+
+    priv->fd = fd;
+    priv->offset = 0;
+    priv->current_size = 0;
+    priv->max_size = max_size;
+
+    return err;
+
+exit:
+    oerrno = errno;
+    if (fd != -1) {
+        (void) close(fd);
+    }
+    errno = oerrno;
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_create_pmem_with_config(struct memkind_config *cfg,
+                                                   struct memkind **kind)
+{
+    int status = memkind_create_pmem(cfg->pmem_dir, cfg->pmem_size, kind);
+    if (MEMKIND_LIKELY(!status)) {
+        status = (*kind)->ops->update_memory_usage_policy(*kind, cfg->policy);
+    }
+
+    return status;
+}
+
+
+static int memkind_get_kind_by_partition_internal(int partition,
+                                                  struct memkind **kind)
+{
+    int err = MEMKIND_SUCCESS;
+
+    if (MEMKIND_LIKELY(partition >= 0 &&
+                       partition < MEMKIND_MAX_KIND &&
+                       memkind_registry_g.partition_map[partition] != NULL)) {
+        *kind = memkind_registry_g.partition_map[partition];
+    } else {
+        *kind = NULL;
+        err = MEMKIND_ERROR_UNAVAILABLE;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_get_kind_by_partition(int partition,
+                                                 struct memkind **kind)
+{
+    return memkind_get_kind_by_partition_internal(partition, kind);
+}
+
+MEMKIND_EXPORT int memkind_update_cached_stats(void)
+{
+    return heap_manager_update_cached_stats();
+}
+
+MEMKIND_EXPORT void *memkind_defrag_reallocate(memkind_t kind, void *ptr)
+{
+    if (!kind) {
+        return heap_manager_defrag_reallocate(ptr);
+    } else {
+        return kind->ops->defrag_reallocate(kind, ptr);
+    }
+}
+
+MEMKIND_EXPORT int memkind_get_stat(memkind_t kind, memkind_stat_type stat,
+                                    size_t *value)
+{
+    if (MEMKIND_UNLIKELY(stat >= MEMKIND_STAT_TYPE_MAX_VALUE)) {
+        log_err("Unrecognized type of memory statistic %d", stat);
+        return MEMKIND_ERROR_INVALID;
+    }
+
+    if (!kind) {
+        return heap_manager_get_stat(stat, value);
+    } else {
+        return kind->ops->get_stat(kind, stat, value);
+    }
+}
diff --git a/src/memkind_arena.c b/src/memkind_arena.c
new file mode 100644
index 0000000..8b525cc
--- /dev/null
+++ b/src/memkind_arena.c
@@ -0,0 +1,938 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <limits.h>
+#include <stdint.h>
+#include <unistd.h>
+#include <pthread.h>
+#include <errno.h>
+#include <numa.h>
+#include <numaif.h>
+#include <jemalloc/jemalloc.h>
+#include <utmpx.h>
+#include <sched.h>
+#include <limits.h>
+#include <sys/mman.h>
+#include <sys/param.h>
+#include <assert.h>
+
+#include "config.h"
+
+#define HUGE_PAGE_SIZE (1ull << MEMKIND_MASK_PAGE_SIZE_2MB)
+#define PAGE_2_BYTES(x) ((x) << 12)
+
+static const char *const global_stats[MEMKIND_STAT_TYPE_MAX_VALUE] = {
+    "stats.resident",
+    "stats.active",
+    "stats.allocated"
+};
+
+#define ARENA_STAT_MAX 2
+
+struct stats_arena {
+    const char *stats[ARENA_STAT_MAX];
+    unsigned stats_no;
+};
+
+static const struct stats_arena arena_stats [MEMKIND_STAT_TYPE_MAX_VALUE] = {
+    { .stats = {"stats.arenas.%u.resident", NULL}, .stats_no = 1},
+    { .stats = {"stats.arenas.%u.pactive", NULL}, .stats_no = 1},
+    { .stats = {"stats.arenas.%u.small.allocated", "stats.arenas.%u.large.allocated"}, .stats_no = 2},
+};
+
+static void *jemk_mallocx_check(size_t size, int flags);
+static void *jemk_rallocx_check(void *ptr, size_t size, int flags);
+static void tcache_finalize(void *args);
+
+static unsigned integer_log2(unsigned v)
+{
+    return (sizeof(unsigned) * 8) - (__builtin_clz(v) + 1);
+}
+
+static unsigned round_pow2_up(unsigned v)
+{
+    unsigned v_log2 = integer_log2(v);
+
+    if (v != 1 << v_log2) {
+        v = 1 << (v_log2 + 1);
+    }
+    return v;
+}
+
+MEMKIND_EXPORT int memkind_set_arena_map_len(struct memkind *kind)
+{
+    if (kind->ops->get_arena == memkind_bijective_get_arena) {
+        kind->arena_map_len = 1;
+    } else if (kind->ops->get_arena == memkind_thread_get_arena) {
+        char *arena_num_env = secure_getenv("MEMKIND_ARENA_NUM_PER_KIND");
+
+        if (arena_num_env) {
+            unsigned long int arena_num_value = strtoul(arena_num_env, NULL, 10);
+
+            if ((arena_num_value == 0) || (arena_num_value > INT_MAX)) {
+                log_err("Wrong MEMKIND_ARENA_NUM_PER_KIND environment value: %lu.",
+                        arena_num_value);
+                return MEMKIND_ERROR_ENVIRON;
+            }
+
+            kind->arena_map_len = arena_num_value;
+        } else {
+            int calculated_arena_num = numa_num_configured_cpus() * 4;
+
+#if ARENA_LIMIT_PER_KIND != 0
+            calculated_arena_num = MIN((ARENA_LIMIT_PER_KIND), calculated_arena_num);
+#endif
+            kind->arena_map_len = calculated_arena_num;
+        }
+
+        kind->arena_map_len = round_pow2_up(kind->arena_map_len);
+    }
+
+    kind->arena_map_mask = kind->arena_map_len - 1;
+    return 0;
+}
+
+static pthread_once_t arena_config_once = PTHREAD_ONCE_INIT;
+static int arena_init_status;
+
+static pthread_key_t tcache_key;
+static bool memkind_hog_memory;
+
+static void arena_config_init()
+{
+    const char *str = secure_getenv("MEMKIND_HOG_MEMORY");
+    memkind_hog_memory = str && str[0] == '1';
+
+    arena_init_status = pthread_key_create(&tcache_key, tcache_finalize);
+}
+
+#define MALLOCX_ARENA_MAX           0xffe        // copy-pasted from jemalloc/internal/jemalloc_internal.h
+#define DIRTY_DECAY_MS_DEFAULT      10000        // copy-pasted from jemalloc/internal/arena_types.h DIRTY_DECAY_MS_DEFAULT
+#define DIRTY_DECAY_MS_CONSERVATIVE     0
+
+static struct memkind *arena_registry_g[MALLOCX_ARENA_MAX];
+static pthread_mutex_t arena_registry_write_lock;
+
+struct memkind *get_kind_by_arena(unsigned arena_ind)
+{
+    // there is no way to obtain MALLOCX_ARENA_MAX from jemalloc
+    // so this checks if arena_ind does not exceed assumed range
+    assert(arena_ind < MALLOCX_ARENA_MAX);
+
+    return arena_registry_g[arena_ind];
+}
+
+// Allocates size bytes aligned to alignment. Returns NULL if allocation fails.
+static void *alloc_aligned_slow(size_t size, size_t alignment,
+                                struct memkind *kind)
+{
+    size_t extended_size = size + alignment;
+    void *ptr;
+
+    ptr = kind_mmap(kind,  NULL, extended_size);
+
+    if(ptr == MAP_FAILED) {
+        return NULL;
+    }
+
+    uintptr_t addr = (uintptr_t)ptr;
+    uintptr_t aligned_addr = (addr + alignment) & ~(alignment - 1);
+
+    size_t head_len = aligned_addr - addr;
+    if (head_len > 0) {
+        munmap(ptr, head_len);
+    }
+
+    uintptr_t tail = aligned_addr + size;
+    size_t tail_len = (addr + extended_size) - (aligned_addr + size);
+    if (tail_len > 0) {
+        munmap((void *)tail, tail_len);
+    }
+
+    return (void *)aligned_addr;
+}
+
+
+void *arena_extent_alloc(extent_hooks_t *extent_hooks,
+                         void *new_addr,
+                         size_t size,
+                         size_t alignment,
+                         bool *zero,
+                         bool *commit,
+                         unsigned arena_ind)
+{
+    struct memkind *kind = get_kind_by_arena(arena_ind);
+
+    int err = memkind_check_available(kind);
+    if (err) {
+        return NULL;
+    }
+
+    void *addr = kind_mmap(kind, new_addr, size);
+    if (addr == MAP_FAILED) {
+        return NULL;
+    }
+
+    if (new_addr != NULL && addr != new_addr) {
+        /* wrong place */
+        munmap(addr, size);
+        return NULL;
+    }
+
+    if ((uintptr_t)addr & (alignment-1)) {
+        munmap(addr, size);
+        addr = alloc_aligned_slow(size, alignment, kind);
+        if(addr == NULL) {
+            return NULL;
+        }
+    }
+
+    *zero = true;
+    *commit = true;
+
+    return addr;
+}
+
+void *arena_extent_alloc_hugetlb(extent_hooks_t *extent_hooks,
+                                 void *new_addr,
+                                 size_t size,
+                                 size_t alignment,
+                                 bool *zero,
+                                 bool *commit,
+                                 unsigned arena_ind)
+{
+    //round up to huge page size
+    size = (size + (HUGE_PAGE_SIZE - 1)) & ~(HUGE_PAGE_SIZE - 1);
+    return arena_extent_alloc(extent_hooks, new_addr, size, alignment, zero, commit,
+                              arena_ind);
+}
+
+bool arena_extent_dalloc(extent_hooks_t *extent_hooks,
+                         void *addr,
+                         size_t size,
+                         bool committed,
+                         unsigned arena_ind)
+{
+    return true;
+}
+
+bool arena_extent_commit(extent_hooks_t *extent_hooks,
+                         void *addr,
+                         size_t size,
+                         size_t offset,
+                         size_t length,
+                         unsigned arena_ind)
+{
+    return false;
+}
+
+bool arena_extent_decommit(extent_hooks_t *extent_hooks,
+                           void *addr,
+                           size_t size,
+                           size_t offset,
+                           size_t length,
+                           unsigned arena_ind)
+{
+    return true;
+}
+
+bool arena_extent_purge(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        size_t offset,
+                        size_t length,
+                        unsigned arena_ind)
+{
+    if (memkind_hog_memory) {
+        return true;
+    }
+
+    int err = madvise(addr + offset, length, MADV_DONTNEED);
+    return (err != 0);
+}
+
+bool arena_extent_split(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        size_t size_a,
+                        size_t size_b,
+                        bool committed,
+                        unsigned arena_ind)
+{
+    return false;
+}
+
+bool arena_extent_merge(extent_hooks_t *extent_hooks,
+                        void *addr_a,
+                        size_t size_a,
+                        void *addr_b,
+                        size_t size_b,
+                        bool committed,
+                        unsigned arena_ind)
+{
+    return false;
+}
+
+extent_hooks_t arena_extent_hooks = {
+    .alloc = arena_extent_alloc,
+    .dalloc = arena_extent_dalloc,
+    .commit = arena_extent_commit,
+    .decommit = arena_extent_decommit,
+    .purge_lazy = arena_extent_purge,
+    .split = arena_extent_split,
+    .merge = arena_extent_merge
+};
+
+extent_hooks_t arena_extent_hooks_hugetlb = {
+    .alloc = arena_extent_alloc_hugetlb,
+    .dalloc = arena_extent_dalloc,
+    .commit = arena_extent_commit,
+    .decommit = arena_extent_decommit,
+    .purge_lazy = arena_extent_purge,
+    .split = arena_extent_split,
+    .merge = arena_extent_merge
+};
+
+extent_hooks_t *get_extent_hooks_by_kind(struct memkind *kind)
+{
+    if (kind == MEMKIND_HUGETLB
+        || kind == MEMKIND_HBW_HUGETLB
+        || kind == MEMKIND_HBW_ALL_HUGETLB
+        || kind == MEMKIND_HBW_PREFERRED_HUGETLB) {
+        return &arena_extent_hooks_hugetlb;
+    } else {
+        return &arena_extent_hooks;
+    }
+}
+
+MEMKIND_EXPORT int memkind_arena_create_map(struct memkind *kind,
+                                            extent_hooks_t *hooks)
+{
+    int err;
+
+    pthread_once(&arena_config_once, arena_config_init);
+    if(arena_init_status) {
+        return arena_init_status;
+    }
+
+    err = memkind_set_arena_map_len(kind);
+    if(err) {
+        return err;
+    }
+#ifdef MEMKIND_TLS
+    if (kind->ops->get_arena == memkind_thread_get_arena) {
+        pthread_key_create(&(kind->arena_key), free);
+    }
+#endif
+
+    if (pthread_mutex_lock(&arena_registry_write_lock) != 0)
+        assert(0 && "failed to acquire mutex");
+    unsigned i;
+    size_t unsigned_size = sizeof(unsigned int);
+    kind->arena_zero = UINT_MAX;
+    for(i = 0; i<kind->arena_map_len; i++) {
+        unsigned arena_index;
+        //create new arena with consecutive index
+        err = jemk_mallctl("arenas.create", (void *)&arena_index, &unsigned_size, NULL,
+                           0);
+        if(err) {
+            log_err("Could not create arena.");
+            err = MEMKIND_ERROR_ARENAS_CREATE;
+            goto exit;
+        }
+        //store arena with lowest index (arenas could be created in descending/ascending order)
+        if(kind->arena_zero > arena_index) {
+            kind->arena_zero = arena_index;
+        }
+        //setup extent_hooks for newly created arena
+        char cmd[64];
+        snprintf(cmd, sizeof(cmd), "arena.%u.extent_hooks", arena_index);
+        err = jemk_mallctl(cmd, NULL, NULL, (void *)&hooks, sizeof(extent_hooks_t *));
+        if(err) {
+            goto exit;
+        }
+        arena_registry_g[arena_index] = kind;
+    }
+
+exit:
+    if (pthread_mutex_unlock(&arena_registry_write_lock) != 0)
+        assert(0 && "failed to release mutex");
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_arena_create(struct memkind *kind,
+                                        struct memkind_ops *ops, const char *name)
+{
+    int err = memkind_default_create(kind, ops, name);
+    if (!err) {
+        err = memkind_arena_create_map(kind, get_extent_hooks_by_kind(kind));
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_arena_destroy(struct memkind *kind)
+{
+    if (kind->arena_map_len) {
+        char cmd[128];
+        unsigned i;
+
+        if (pthread_mutex_lock(&arena_registry_write_lock) != 0)
+            assert(0 && "failed to acquire mutex");
+
+        for (i = 0; i < kind->arena_map_len; ++i) {
+            snprintf(cmd, 128, "arena.%u.destroy", kind->arena_zero + i);
+            jemk_mallctl(cmd, NULL, NULL, NULL, 0);
+            arena_registry_g[kind->arena_zero + i] = NULL;
+        }
+
+        if (pthread_mutex_unlock(&arena_registry_write_lock) != 0)
+            assert(0 && "failed to release mutex");
+
+#ifdef MEMKIND_TLS
+        if (kind->ops->get_arena == memkind_thread_get_arena) {
+            pthread_key_delete(kind->arena_key);
+        }
+#endif
+    }
+
+    memkind_default_destroy(kind);
+    return 0;
+}
+
+int memkind_arena_finalize(struct memkind *kind)
+{
+    return memkind_arena_destroy(kind);
+}
+
+// max allocation size to be cached by tcache mechanism
+#define TCACHE_MAX (1<<(JEMALLOC_TCACHE_CLASS))
+
+static void tcache_finalize(void *args)
+{
+    int i;
+    unsigned *tcache_map = args;
+    for(i = 0; i<MEMKIND_NUM_BASE_KIND; i++) {
+        if(tcache_map[i] != 0) {
+            jemk_mallctl("tcache.destroy", NULL, NULL, (void *)&tcache_map[i],
+                         sizeof(unsigned));
+        }
+    }
+}
+
+static inline int memkind_lookup_arena(void *ptr, unsigned int *arena)
+{
+    size_t sz = sizeof(unsigned);
+    unsigned temp_arena;
+    int err = jemk_mallctl("arenas.lookup", &temp_arena, &sz, &ptr, sizeof(ptr));
+
+    if (err) {
+        log_err("Could not found arena, err=%d", err);
+        return 1;
+    }
+
+    *arena = temp_arena;
+    return 0;
+}
+
+MEMKIND_EXPORT struct memkind *memkind_arena_detect_kind(void *ptr)
+{
+    if (!ptr) {
+        return NULL;
+    }
+    struct memkind *kind = NULL;
+    unsigned arena;
+    int err = memkind_lookup_arena(ptr, &arena);
+    if (MEMKIND_LIKELY(!err)) {
+        kind = get_kind_by_arena(arena);
+    }
+
+    /* if no kind was associated with arena it means that allocation doesn't come from
+       jemk_*allocx API - it is jemk_*alloc API (MEMKIND_DEFAULT) */
+
+    return (kind) ? kind : MEMKIND_DEFAULT;
+}
+
+static inline int get_tcache_flag(unsigned partition, size_t size)
+{
+
+    // do not cache allocation larger than tcache_max nor those coming from non-static kinds
+    if(size > TCACHE_MAX || partition >= MEMKIND_NUM_BASE_KIND) {
+        return MALLOCX_TCACHE_NONE;
+    }
+
+    unsigned *tcache_map = pthread_getspecific(tcache_key);
+    if(tcache_map == NULL) {
+        tcache_map = calloc(MEMKIND_NUM_BASE_KIND, sizeof(unsigned));
+        if(tcache_map == NULL) {
+            return MALLOCX_TCACHE_NONE;
+        }
+        pthread_setspecific(tcache_key, (void *)tcache_map);
+    }
+
+    if(MEMKIND_UNLIKELY(tcache_map[partition] == 0)) {
+        size_t unsigned_size = sizeof(unsigned);
+        int err = jemk_mallctl("tcache.create", (void *)&tcache_map[partition],
+                               &unsigned_size, NULL, 0);
+        if(err) {
+            log_err("Could not acquire tcache, err=%d", err);
+            return MALLOCX_TCACHE_NONE;
+        }
+    }
+    return MALLOCX_TCACHE(tcache_map[partition]);
+}
+
+MEMKIND_EXPORT void *memkind_arena_malloc(struct memkind *kind, size_t size)
+{
+    void *result = NULL;
+    unsigned arena;
+
+    int err = kind->ops->get_arena(kind, &arena, size);
+    if (MEMKIND_LIKELY(!err)) {
+        result = jemk_mallocx_check(size,
+                                    MALLOCX_ARENA(arena) | get_tcache_flag(kind->partition, size));
+    }
+    return result;
+}
+
+static void *memkind_arena_malloc_no_tcache(struct memkind *kind, size_t size)
+{
+    void *result = NULL;
+    if (kind == MEMKIND_DEFAULT) {
+        result = jemk_mallocx_check(size, MALLOCX_TCACHE_NONE);
+    } else {
+        unsigned arena;
+        int err = kind->ops->get_arena(kind, &arena, size);
+        if (MEMKIND_LIKELY(!err)) {
+            result = jemk_mallocx_check(size, MALLOCX_ARENA(arena) | MALLOCX_TCACHE_NONE);
+        }
+    }
+    return result;
+}
+
+MEMKIND_EXPORT void memkind_arena_free(struct memkind *kind, void *ptr)
+{
+    if (kind == MEMKIND_DEFAULT) {
+        jemk_free(ptr);
+    } else if (ptr != NULL) {
+        jemk_dallocx(ptr, get_tcache_flag(kind->partition, 0));
+    }
+}
+
+MEMKIND_EXPORT void memkind_arena_free_with_kind_detect(void *ptr)
+{
+    struct memkind *kind = memkind_arena_detect_kind(ptr);
+
+    memkind_arena_free(kind, ptr);
+}
+
+MEMKIND_EXPORT size_t memkind_arena_malloc_usable_size(void *ptr)
+{
+    return memkind_default_malloc_usable_size(NULL, ptr);
+}
+
+MEMKIND_EXPORT void *memkind_arena_realloc(struct memkind *kind, void *ptr,
+                                           size_t size)
+{
+    unsigned arena;
+
+    if (size == 0 && ptr != NULL) {
+        memkind_arena_free(kind, ptr);
+        ptr = NULL;
+    } else {
+        int err = kind->ops->get_arena(kind, &arena, size);
+        if (MEMKIND_LIKELY(!err)) {
+            if (ptr == NULL) {
+                ptr = jemk_mallocx_check(size,
+                                         MALLOCX_ARENA(arena) | get_tcache_flag(kind->partition, size));
+            } else {
+                ptr = jemk_rallocx_check(ptr, size,
+                                         MALLOCX_ARENA(arena) | get_tcache_flag(kind->partition, size));
+            }
+        }
+    }
+    return ptr;
+}
+
+int memkind_arena_update_cached_stats(void)
+{
+    uint64_t epoch = 1;
+    return jemk_mallctl("epoch", NULL, NULL, &epoch, sizeof(epoch));
+}
+
+MEMKIND_EXPORT void *memkind_arena_realloc_with_kind_detect(void *ptr,
+                                                            size_t size)
+{
+    if (!ptr) {
+        errno = EINVAL;
+        return NULL;
+    }
+    struct memkind *kind = memkind_arena_detect_kind(ptr);
+    if (kind == MEMKIND_DEFAULT) {
+        return memkind_default_realloc(kind, ptr, size);
+    } else {
+        return memkind_arena_realloc(kind, ptr, size);
+    }
+}
+
+MEMKIND_EXPORT int memkind_arena_update_memory_usage_policy(
+    struct memkind *kind, memkind_mem_usage_policy policy)
+{
+    int err = MEMKIND_SUCCESS;
+    unsigned i;
+    ssize_t dirty_decay_val = 0;
+    switch ( policy ) {
+        case MEMKIND_MEM_USAGE_POLICY_DEFAULT:
+            dirty_decay_val = DIRTY_DECAY_MS_DEFAULT;
+            break;
+        case MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE:
+            dirty_decay_val = DIRTY_DECAY_MS_CONSERVATIVE;
+            break;
+        default:
+            log_err("Unrecognized memory policy %d", policy);
+            return MEMKIND_ERROR_INVALID;
+    }
+
+    for (i = 0; i < kind->arena_map_len; ++i) {
+        char cmd[64];
+
+        snprintf(cmd, sizeof(cmd), "arena.%u.dirty_decay_ms", kind->arena_zero + i);
+        err = jemk_mallctl(cmd, NULL, NULL, (void *)&dirty_decay_val, sizeof(ssize_t));
+        if ( err ) {
+            log_err("Incorrect dirty_decay_ms value %zu", dirty_decay_val);
+            return MEMKIND_ERROR_INVALID;
+        }
+    }
+
+    return err;
+}
+
+MEMKIND_EXPORT void *memkind_arena_calloc(struct memkind *kind, size_t num,
+                                          size_t size)
+{
+    void *result = NULL;
+    unsigned arena;
+
+    int err = kind->ops->get_arena(kind, &arena, size);
+    if (MEMKIND_LIKELY(!err)) {
+        result = jemk_mallocx_check(num * size,
+                                    MALLOCX_ARENA(arena) | MALLOCX_ZERO | get_tcache_flag(kind->partition, size));
+    }
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_arena_posix_memalign(struct memkind *kind,
+                                                void **memptr, size_t alignment,
+                                                size_t size)
+{
+    int err;
+    unsigned arena;
+
+    *memptr = NULL;
+    err = kind->ops->get_arena(kind, &arena, size);
+    if (MEMKIND_LIKELY(!err)) {
+        err = memkind_posix_check_alignment(kind, alignment);
+    }
+    if (MEMKIND_LIKELY(!err)) {
+        if (MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+            return 0;
+        }
+        /* posix_memalign should not change errno.
+           Set it to its previous value after calling jemalloc */
+        int errno_before = errno;
+        *memptr = jemk_mallocx_check(size,
+                                     MALLOCX_ALIGN(alignment) | MALLOCX_ARENA(arena) | get_tcache_flag(
+                                         kind->partition, size));
+        errno = errno_before;
+        err = *memptr ? 0 : ENOMEM;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_bijective_get_arena(struct memkind *kind,
+                                               unsigned int *arena, size_t size)
+{
+    *arena = kind->arena_zero;
+    return 0;
+}
+
+#ifdef MEMKIND_TLS
+MEMKIND_EXPORT int memkind_thread_get_arena(struct memkind *kind,
+                                            unsigned int *arena, size_t size)
+{
+    int err = 0;
+    unsigned int *arena_tsd;
+    arena_tsd = pthread_getspecific(kind->arena_key);
+
+    if (MEMKIND_UNLIKELY(arena_tsd == NULL)) {
+        arena_tsd = malloc(sizeof(unsigned int));
+        if (arena_tsd == NULL) {
+            err = MEMKIND_ERROR_MALLOC;
+            log_err("malloc() failed.");
+        }
+        if (!err) {
+            // On glibc pthread_self() is incremented by 0x801000 for every
+            // thread (no matter the arch's word width).  This might change
+            // in the future, but even in the worst case the hash will
+            // degenerate to a single bucket with no loss of correctness.
+            *arena_tsd = ((uint64_t)pthread_self() >> 12) %
+                         kind->arena_map_len;
+            err = pthread_setspecific(kind->arena_key, arena_tsd) ?
+                  MEMKIND_ERROR_RUNTIME : 0;
+        }
+    }
+    *arena = kind->arena_zero + *arena_tsd;
+    return err;
+}
+
+#else
+
+/*
+ *
+ * We use thread control block as unique thread identifier
+ * For more read: https://www.akkadia.org/drepper/tls.pdf
+ * We could consider using rdfsbase when it will arrive to linux kernel
+ *
+ * This approach works only on glibc (and possibly similar implementations)
+ * but that covers our current needs.
+ *
+ */
+static uintptr_t get_fs_base()
+{
+    return (uintptr_t)pthread_self();
+}
+
+MEMKIND_EXPORT int memkind_thread_get_arena(struct memkind *kind,
+                                            unsigned int *arena, size_t size)
+{
+    unsigned int arena_idx;
+    // it's likely that each thread control block lies on different page
+    // so we extracting page number with >> 12 to improve hashing
+    arena_idx = (get_fs_base() >> 12) & kind->arena_map_mask;
+    *arena = kind->arena_zero + arena_idx;
+    return 0;
+}
+#endif //MEMKIND_TLS
+
+static void *jemk_mallocx_check(size_t size, int flags)
+{
+    /*
+     * Checking for out of range size due to unhandled error in
+     * jemk_mallocx().  Size invalid for the range
+     * LLONG_MAX <= size <= ULLONG_MAX
+     * which is the result of passing a negative signed number as size
+     */
+    void *result = NULL;
+
+    if (MEMKIND_UNLIKELY(size >= LLONG_MAX)) {
+        errno = ENOMEM;
+    } else if (size != 0) {
+        result = jemk_mallocx(size, flags);
+    }
+    return result;
+}
+
+static void *jemk_rallocx_check(void *ptr, size_t size, int flags)
+{
+    /*
+     * Checking for out of range size due to unhandled error in
+     * jemk_mallocx().  Size invalid for the range
+     * LLONG_MAX <= size <= ULLONG_MAX
+     * which is the result of passing a negative signed number as size
+     */
+    void *result = NULL;
+
+    if (MEMKIND_UNLIKELY(size >= LLONG_MAX)) {
+        errno = ENOMEM;
+    } else {
+        result = jemk_rallocx(ptr, size, flags);
+    }
+    return result;
+
+}
+
+void memkind_arena_init(struct memkind *kind)
+{
+    if (kind != MEMKIND_DEFAULT) {
+        int err = memkind_arena_create_map(kind, get_extent_hooks_by_kind(kind));
+        if (err) {
+            log_fatal("[%s] Failed to create arena map (error code:%d).", kind->name, err);
+            abort();
+        }
+    }
+}
+
+static int memkind_arena_get_stat(struct memkind *kind, memkind_stat_type stat,
+                                  bool check_init, size_t *value)
+{
+    size_t sz = sizeof(size_t);
+    size_t temp_stat;
+    int err = MEMKIND_SUCCESS;
+    unsigned i, j;
+    char cmd[128];
+
+    *value = 0;
+    for (i = 0; i < kind->arena_map_len; ++i) {
+        if (check_init) {
+            bool is_init;
+            size_t sz_b_state = sizeof(is_init);
+            snprintf(cmd, 128, "arena.%u.initialized", kind->arena_zero + i);
+            err =  jemk_mallctl(cmd, (void *)&is_init, &sz_b_state, NULL, 0);
+            if (err) {
+                log_err("Error on getting initialized state of arena.");
+                return MEMKIND_ERROR_INVALID;
+            }
+            if (!is_init) {
+                continue;
+            }
+        }
+
+        for (j = 0; j < arena_stats[stat].stats_no; ++j) {
+            snprintf(cmd, 128, arena_stats[stat].stats[j], kind->arena_zero + i);
+            err = jemk_mallctl(cmd, &temp_stat, &sz, NULL, 0);
+            if (err) {
+                log_err("Error on getting arena statistic.");
+                return MEMKIND_ERROR_INVALID;
+            }
+            *value += temp_stat;
+        }
+    }
+    return err;
+}
+
+int memkind_arena_get_stat_with_check_init(struct memkind *kind,
+                                           memkind_stat_type stat, bool check_init, size_t *value)
+{
+    int status;
+    switch (stat) {
+        case MEMKIND_STAT_TYPE_RESIDENT:
+        case MEMKIND_STAT_TYPE_ALLOCATED:
+            status = memkind_arena_get_stat(kind, stat, check_init, value);
+            break;
+        case MEMKIND_STAT_TYPE_ACTIVE:
+            status = memkind_arena_get_stat(kind, stat, check_init, value);
+            *value = PAGE_2_BYTES(*value);
+            break;
+        default:
+            //not reached
+            return MEMKIND_ERROR_INVALID;
+            break;
+    }
+    return status;
+}
+
+int memkind_arena_get_kind_stat(struct memkind *kind, memkind_stat_type stat,
+                                size_t *value)
+{
+    return memkind_arena_get_stat_with_check_init(kind, stat, false, value);
+}
+
+int memkind_arena_get_global_stat(memkind_stat_type stat, size_t *value)
+{
+    size_t sz = sizeof(size_t);
+    int err = jemk_mallctl(global_stats[stat], value, &sz, NULL, 0);
+    if (err) {
+        log_err("Error on getting global statistic.");
+        return MEMKIND_ERROR_INVALID;
+    }
+    return err;
+}
+
+int memkind_arena_enable_background_threads(size_t threads_limit)
+{
+    bool background_thread_val = true;
+    int err;
+
+    if (threads_limit) {
+        err = jemk_mallctl("max_background_threads", NULL, NULL, &threads_limit,
+                           sizeof(size_t));
+        if (err) {
+            log_err("Error on setting threads limit");
+            return MEMKIND_ERROR_INVALID;
+        }
+    }
+    err = jemk_mallctl("background_thread", NULL, NULL, &background_thread_val,
+                       sizeof(bool));
+    if (err) {
+        log_err("Error on activating background thread");
+        return MEMKIND_ERROR_INVALID;
+    }
+    return err;
+}
+
+#define DEST_SLAB_END(begin, size) ((uintptr_t)begin+ (uintptr_t)size)
+
+struct mem_util_stats {
+    void *target_slab;      // address of the slab of a potential realloaction would go to ( NULL in case of large/huge allocation)
+    size_t nfree;           // number of free regions in the slab
+    size_t nregs;           // number of regions in the slab
+    size_t slab_size;       // size of the slab in bytes
+    size_t bin_nfree;       // total number of free regions in the bin the slab belongs to
+    size_t bin_nregs;       // total number of regions in the bin the slab belongs to
+};
+
+void *memkind_arena_defrag_reallocate_with_kind_detect (void *ptr)
+{
+    return memkind_arena_defrag_reallocate(memkind_detect_kind(ptr), ptr);
+}
+
+void *memkind_arena_defrag_reallocate(struct memkind *kind, void *ptr)
+{
+    if (!ptr) {
+        return NULL;
+    }
+
+    size_t out_sz = sizeof(struct mem_util_stats);
+    struct mem_util_stats out;
+    int err = jemk_mallctl("experimental.utilization.query", &out, &out_sz, &ptr,
+                           sizeof(ptr));
+    if (err) {
+        log_err("Error on get utilization query");
+        return NULL;
+    }
+
+    // Check if input pointer resides outside of potential reallocation slab
+    // Check if occupied regions inside the slab are below average occupied regions inside bin
+    // Check if there are some free regions in the destination slab
+    if (out.target_slab &&
+        ((ptr < out.target_slab) ||
+         (uintptr_t)ptr > DEST_SLAB_END(out.target_slab, out.slab_size)) &&
+        out.nfree * out.bin_nregs >= out.nregs * out.bin_nfree &&
+        out.nfree != 0) {
+        size_t size = memkind_malloc_usable_size(kind, ptr);
+        void *ptr_new = memkind_arena_malloc_no_tcache(kind, size);
+        if (MEMKIND_UNLIKELY(!ptr_new)) return NULL;
+        memcpy(ptr_new, ptr, size);
+        memkind_free(kind, ptr);
+        return ptr_new;
+    }
+    return NULL;
+}
diff --git a/src/memkind_bandwidth.c b/src/memkind_bandwidth.c
new file mode 100644
index 0000000..8c99dae
--- /dev/null
+++ b/src/memkind_bandwidth.c
@@ -0,0 +1,372 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <errno.h>
+#include <limits.h>
+#include <stdint.h>
+
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_bandwidth.h>
+
+struct numanode_bandwidth_t {
+    int numanode;
+    int bandwidth;
+};
+
+struct bandwidth_nodes_t {
+    int bandwidth;
+    int num_numanodes;
+    int *numanodes;
+};
+
+VEC(vec_cpu_node, int);
+
+#define BANDWIDTH_NODE_HIGH_VAL       2
+#define BANDWIDTH_NODE_LOW_VAL        1
+#define BANDWIDTH_NODE_NOT_PRESENT    0
+
+static void bandwidth_assign_arbitrary_values(int *bandwidth,
+                                              struct bitmask *numa_nodes_bm)
+{
+    unsigned i;
+    int nodes_num = numa_num_configured_nodes();
+
+    for (i = 0; i<NUMA_NUM_NODES; i++) {
+        if (numa_bitmask_isbitset(numa_nodes_bm, i)) {
+            bandwidth[i] = BANDWIDTH_NODE_HIGH_VAL;
+            nodes_num--;
+        }
+    }
+
+    for (i = 0; i<NUMA_NUM_NODES; i++) {
+        if (!numa_bitmask_isbitset(numa_nodes_bm, i)) {
+            bandwidth[i] = BANDWIDTH_NODE_LOW_VAL;
+            nodes_num--;
+        }
+        if (nodes_num==0) break;
+    }
+}
+
+int bandwidth_fill(int *bandwidth, get_node_bitmask get_bitmask)
+{
+    struct bitmask *high_val_node_mask = numa_allocate_nodemask();
+    int ret = get_bitmask(high_val_node_mask);
+    if(ret == 0) {
+        bandwidth_assign_arbitrary_values(bandwidth, high_val_node_mask);
+    }
+    numa_bitmask_free(high_val_node_mask);
+
+    return ret;
+}
+
+static int bandwidth_fill_values_from_enviroment(int *bandwidth,
+                                                 char *nodes_env)
+{
+    struct bitmask *numa_nodes_bm = numa_parse_nodestring(nodes_env);
+
+    if (!numa_nodes_bm) {
+        log_err("Invalid value of environment variable.");
+        return MEMKIND_ERROR_ENVIRON;
+    } else {
+        bandwidth_assign_arbitrary_values(bandwidth, numa_nodes_bm);
+        numa_bitmask_free(numa_nodes_bm);
+    }
+
+    return 0;
+}
+
+static int bandwidth_fill_nodes(int *bandwidth, fill_bandwidth_values fill,
+                                const char *env)
+{
+    char *high_value_nodes_env = secure_getenv(env);
+    if (high_value_nodes_env) {
+        log_info("Environment variable %s detected: %s.", env, high_value_nodes_env);
+        return bandwidth_fill_values_from_enviroment(bandwidth, high_value_nodes_env);
+    }
+    return fill(bandwidth);
+}
+
+static int bandwidth_compare_numanode(const void *a, const void *b)
+{
+    /***************************************************************************
+    *  qsort comparison function for numa_node_bandwidth structures.  Sorts in *
+    *  order of bandwidth and then numanode.                                   *
+    ***************************************************************************/
+    int result;
+    struct numanode_bandwidth_t *aa = (struct numanode_bandwidth_t *)(a);
+    struct numanode_bandwidth_t *bb = (struct numanode_bandwidth_t *)(b);
+    result = (aa->bandwidth > bb->bandwidth) - (aa->bandwidth < bb->bandwidth);
+    if (result == 0) {
+        result = (aa->numanode > bb->numanode) - (aa->numanode < bb->numanode);
+    }
+    return result;
+}
+
+static int bandwidth_create_nodes(const int *bandwidth, int *num_unique,
+                                  struct bandwidth_nodes_t **bandwidth_nodes)
+{
+    /***************************************************************************
+    *   bandwidth (IN):                                                        *
+    *       A vector of length equal NUMA_NUM_NODES that gives bandwidth for   *
+    *       each numa node, zero if numa node has unknown bandwidth.           *
+    *   num_unique (OUT):                                                      *
+    *       number of unique non-zero bandwidth values in bandwidth            *
+    *       vector.                                                            *
+    *   bandwidth_nodes (OUT):                                                 *
+    *       A list of length num_unique sorted by bandwidth value where        *
+    *       each element gives a list of the numa nodes that have the          *
+    *       given bandwidth.                                                   *
+    *   RETURNS MEMKIND_SUCCESS on success, error code on failure              *
+    ***************************************************************************/
+    int err = MEMKIND_SUCCESS;
+    int i, last_bandwidth;
+    *bandwidth_nodes = NULL;
+    /* allocate space for sorting array */
+    int nodes_num = numa_num_configured_nodes();
+    struct numanode_bandwidth_t *numanode_bandwidth = malloc(sizeof(
+                                                                 struct numanode_bandwidth_t) * nodes_num);
+    if (!numanode_bandwidth) {
+        err = MEMKIND_ERROR_MALLOC;
+        log_err("malloc() failed.");
+    }
+    if (!err) {
+        /* set sorting array */
+        int j = 0;
+        for (i = 0; i < NUMA_NUM_NODES; ++i) {
+            if (bandwidth[i] != BANDWIDTH_NODE_NOT_PRESENT) {
+                numanode_bandwidth[j].numanode = i;
+                numanode_bandwidth[j].bandwidth = bandwidth[i];
+                ++j;
+            }
+        }
+
+        if (j == 0) {
+            err = MEMKIND_ERROR_UNAVAILABLE;
+        }
+    }
+    if (!err) {
+        qsort(numanode_bandwidth, nodes_num, sizeof(struct numanode_bandwidth_t),
+              bandwidth_compare_numanode);
+        /* calculate the number of unique bandwidths */
+        *num_unique = 1;
+        last_bandwidth = numanode_bandwidth[0].bandwidth;
+        for (i = 1; i < nodes_num; ++i) {
+            if (numanode_bandwidth[i].bandwidth != last_bandwidth) {
+                last_bandwidth = numanode_bandwidth[i].bandwidth;
+                ++*num_unique;
+            }
+        }
+        /* allocate output array */
+        *bandwidth_nodes = (struct bandwidth_nodes_t *)malloc(sizeof(
+                                                                  struct bandwidth_nodes_t) * (*num_unique) + sizeof(int) * nodes_num);
+        if (!*bandwidth_nodes) {
+            err = MEMKIND_ERROR_MALLOC;
+            log_err("malloc() failed.");
+        }
+    }
+    if (!err) {
+        /* populate output */
+        (*bandwidth_nodes)[0].numanodes = (int *)(*bandwidth_nodes + *num_unique);
+        last_bandwidth = numanode_bandwidth[0].bandwidth;
+        int k = 0;
+        int l = 0;
+        for (i = 0; i < nodes_num; ++i, ++l) {
+            (*bandwidth_nodes)[0].numanodes[i] = numanode_bandwidth[i].numanode;
+            if (numanode_bandwidth[i].bandwidth != last_bandwidth) {
+                (*bandwidth_nodes)[k].num_numanodes = l;
+                (*bandwidth_nodes)[k].bandwidth = last_bandwidth;
+                l = 0;
+                ++k;
+                (*bandwidth_nodes)[k].numanodes = (*bandwidth_nodes)[0].numanodes + i;
+                last_bandwidth = numanode_bandwidth[i].bandwidth;
+            }
+        }
+        (*bandwidth_nodes)[k].num_numanodes = l;
+        (*bandwidth_nodes)[k].bandwidth = last_bandwidth;
+    }
+    if (numanode_bandwidth) {
+        free(numanode_bandwidth);
+    }
+    if (err) {
+        if (*bandwidth_nodes) {
+            free(*bandwidth_nodes);
+        }
+    }
+    return err;
+}
+
+static int bandwidth_set_closest_numanode(int num_unique,
+                                          const struct bandwidth_nodes_t *bandwidth_nodes, bool is_single_node,
+                                          int num_cpunode, struct vec_cpu_node **closest_numanode)
+{
+    /***************************************************************************
+    *   num_unique (IN):                                                       *
+    *       Length of bandwidth_nodes vector.                                  *
+    *   bandwidth_nodes (IN):                                                  *
+    *       Output vector from bandwidth_create_nodes().                       *
+    *   is_single_node (IN):                                                   *
+    *       Flag used to mark is only single closest numanode                  *
+    *       is a valid configuration                                           *
+    *   num_cpunode (IN):                                                      *
+    *       Number of cpu's and length of closest_numanode.                    *
+    *   closest_numanode (OUT):                                                *
+    *       Vector that maps cpu index to closest numa node(s)                 *
+    *       of the specified bandwidth.                                        *
+    *   RETURNS zero on success, error code on failure                         *
+    ***************************************************************************/
+    int err = MEMKIND_SUCCESS;
+    int min_distance, distance, i, j, old_errno;
+    struct bandwidth_nodes_t match;
+    match.bandwidth = -1;
+    int target_bandwidth = bandwidth_nodes[num_unique-1].bandwidth;
+
+    struct vec_cpu_node *node_arr = (struct vec_cpu_node *) calloc(num_cpunode,
+                                                                   sizeof(struct vec_cpu_node));
+
+    if (!node_arr) {
+        log_err("calloc() failed.");
+        return MEMKIND_ERROR_MALLOC;
+    }
+
+    for (i = 0; i < num_unique; ++i) {
+        if (bandwidth_nodes[i].bandwidth == target_bandwidth) {
+            match = bandwidth_nodes[i];
+            break;
+        }
+    }
+    if (match.bandwidth == -1) {
+        err = MEMKIND_ERROR_UNAVAILABLE;
+    } else {
+        for (i = 0; i < num_cpunode; ++i) {
+            min_distance = INT_MAX;
+            for (j = 0; j < match.num_numanodes; ++j) {
+                old_errno = errno;
+                distance = numa_distance(numa_node_of_cpu(i), match.numanodes[j]);
+                errno = old_errno;
+                if (distance < min_distance) {
+                    min_distance = distance;
+                    VEC_CLEAR(&node_arr[i]);
+                    VEC_PUSH_BACK(&node_arr[i], match.numanodes[j]);
+                } else if (distance == min_distance) {
+                    VEC_PUSH_BACK(&node_arr[i], match.numanodes[j]);
+                }
+            }
+            if (VEC_SIZE(&node_arr[i]) > 1 && is_single_node &&
+                numa_bitmask_isbitset(numa_all_cpus_ptr, i)) {
+                log_err("Invalid Numa Configuration for cpu %d", i);
+                int node = -1;
+                VEC_FOREACH(node, &node_arr[i]) {
+                    log_err("Node %d", node);
+                }
+                err = MEMKIND_ERROR_RUNTIME;
+            }
+        }
+    }
+
+    if (err) {
+        for (i = 0; i < num_cpunode; ++i) {
+            VEC_DELETE(&node_arr[i]);
+        }
+        free(node_arr);
+        node_arr = NULL;
+    }
+
+    *closest_numanode = node_arr;
+
+    return err;
+}
+
+int set_closest_numanode(fill_bandwidth_values fill_values, const char *env,
+                         struct vec_cpu_node **closest_numanode, int num_cpu, bool is_single_node)
+{
+    int status;
+    int num_unique = 0;
+    int i;
+    struct bandwidth_nodes_t *bandwidth_nodes = NULL;
+    int *bandwidth = (int *)calloc(NUMA_NUM_NODES, sizeof(int));
+
+    if (!bandwidth) {
+        log_err("calloc() failed.");
+        return MEMKIND_ERROR_MALLOC;
+    }
+
+    status = bandwidth_fill_nodes(bandwidth, fill_values, env);
+    if (status)
+        goto exit;
+
+    status = bandwidth_create_nodes(bandwidth, &num_unique, &bandwidth_nodes);
+    if (status)
+        goto exit;
+
+    status = bandwidth_set_closest_numanode(num_unique, bandwidth_nodes,
+                                            is_single_node, num_cpu, closest_numanode);
+
+    for(i = 0; i < bandwidth_nodes[num_unique-1].num_numanodes; ++i) {
+        log_info("NUMA node %d is high-bandwidth/dax-kmem memory.",
+                 bandwidth_nodes[num_unique-1].numanodes[i]);
+    }
+
+exit:
+
+    free(bandwidth_nodes);
+    free(bandwidth);
+
+    return status;
+}
+
+void set_bitmask_for_all_closest_numanodes(unsigned long *nodemask,
+                                           unsigned long maxnode, const struct vec_cpu_node *closest_numanode, int num_cpu)
+{
+    if (MEMKIND_LIKELY(nodemask)) {
+        struct bitmask nodemask_bm = {maxnode, nodemask};
+        int cpu;
+        int node = -1;
+        numa_bitmask_clearall(&nodemask_bm);
+        for (cpu = 0; cpu < num_cpu; ++cpu) {
+            VEC_FOREACH(node, &closest_numanode[cpu]) {
+                numa_bitmask_setbit(&nodemask_bm, node);
+            }
+        }
+    }
+}
+
+int set_bitmask_for_current_closest_numanode(unsigned long *nodemask,
+                                             unsigned long maxnode, const struct vec_cpu_node *closest_numanode, int num_cpu)
+{
+    if (MEMKIND_LIKELY(nodemask)) {
+        struct bitmask nodemask_bm = {maxnode, nodemask};
+        numa_bitmask_clearall(&nodemask_bm);
+        int cpu = sched_getcpu();
+        if (MEMKIND_LIKELY(cpu < num_cpu)) {
+            int node = -1;
+            VEC_FOREACH(node, &closest_numanode[cpu]) {
+                numa_bitmask_setbit(&nodemask_bm, node);
+            }
+        } else {
+            return MEMKIND_ERROR_RUNTIME;
+        }
+    }
+    return MEMKIND_SUCCESS;
+}
diff --git a/src/memkind_dax_kmem.c b/src/memkind_dax_kmem.c
new file mode 100644
index 0000000..cf9b46a
--- /dev/null
+++ b/src/memkind_dax_kmem.c
@@ -0,0 +1,257 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_bandwidth.h>
+#include <memkind/internal/memkind_dax_kmem.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/heap_manager.h>
+
+#include "config.h"
+#include <pthread.h>
+#include <numa.h>
+#include <errno.h>
+
+struct dax_closest_numanode_t {
+    int init_err;
+    int num_cpu;
+    struct vec_cpu_node *closest_numanode;
+};
+
+#define NODE_VARIANT_MULTIPLE 0
+#define NODE_VARIANT_SINGLE   1
+#define NODE_VARIANT_MAX      2
+
+static struct dax_closest_numanode_t
+    memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MAX];
+static pthread_once_t memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_MAX]
+    = {PTHREAD_ONCE_INIT};
+
+static void memkind_dax_kmem_closest_numanode_init(void);
+static void memkind_dax_kmem_preferred_closest_numanode_init(void);
+#ifdef MEMKIND_DAXCTL_KMEM
+#include <daxctl/libdaxctl.h>
+#ifndef daxctl_region_foreach_safe
+#define daxctl_region_foreach_safe(ctx, region, _region) \
+    for (region = daxctl_region_get_first(ctx), \
+         _region = region ? daxctl_region_get_next(region) : NULL; \
+         region != NULL; \
+         region = _region, \
+        _region = _region ? daxctl_region_get_next(_region) : NULL)
+#endif
+
+#ifndef daxctl_dev_foreach_safe
+#define daxctl_dev_foreach_safe(region, dev, _dev) \
+    for (dev = daxctl_dev_get_first(region), \
+         _dev = dev ? daxctl_dev_get_next(dev) : NULL; \
+         dev != NULL; \
+         dev = _dev, \
+        _dev = _dev ? daxctl_dev_get_next(_dev) : NULL)
+#endif
+
+static int get_dax_kmem_nodes(struct bitmask *dax_kmem_node_mask)
+{
+    struct daxctl_region *region, *_region;
+    struct daxctl_dev *dev, *_dev;
+    struct daxctl_ctx *ctx;
+
+    int rc = daxctl_new(&ctx);
+    if (rc < 0)
+        return MEMKIND_ERROR_UNAVAILABLE;
+
+    daxctl_region_foreach_safe(ctx, region, _region) {
+        daxctl_dev_foreach_safe(region, dev, _dev) {
+            struct daxctl_memory *mem = daxctl_dev_get_memory(dev);
+            if (mem) {
+                numa_bitmask_setbit(dax_kmem_node_mask,
+                                    (unsigned)daxctl_dev_get_target_node(dev));
+            }
+        }
+    }
+
+    daxctl_unref(ctx);
+
+    return (numa_bitmask_weight(dax_kmem_node_mask) != 0) ? MEMKIND_SUCCESS :
+           MEMKIND_ERROR_UNAVAILABLE;
+}
+
+static int fill_dax_kmem_values_automatic(int *bandwidth)
+{
+    return bandwidth_fill(bandwidth, get_dax_kmem_nodes);
+}
+#else
+static int fill_dax_kmem_values_automatic(int *bandwidth)
+{
+    log_err("DAX KMEM nodes cannot be automatically detected.");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+#endif
+
+static int memkind_dax_kmem_check_available(struct memkind *kind)
+{
+    return kind->ops->get_mbind_nodemask(kind, NULL, 0);
+}
+
+static int memkind_dax_kmem_get_mbind_nodemask(struct memkind *kind,
+                                               unsigned long *nodemask, unsigned long maxnode)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MULTIPLE];
+    pthread_once(&memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_MULTIPLE],
+                 memkind_dax_kmem_closest_numanode_init);
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        g->init_err = set_bitmask_for_current_closest_numanode(nodemask, maxnode,
+                                                               g->closest_numanode, g->num_cpu);
+    }
+    return g->init_err;
+}
+
+static int memkind_dax_kmem_get_preferred_mbind_nodemask(struct memkind *kind,
+                                                         unsigned long *nodemask, unsigned long maxnode)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_SINGLE];
+    pthread_once(&memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_SINGLE],
+                 memkind_dax_kmem_preferred_closest_numanode_init);
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        g->init_err = set_bitmask_for_current_closest_numanode(nodemask, maxnode,
+                                                               g->closest_numanode, g->num_cpu);
+    }
+    return g->init_err;
+}
+
+MEMKIND_EXPORT int memkind_dax_kmem_all_get_mbind_nodemask(struct memkind *kind,
+                                                           unsigned long *nodemask, unsigned long maxnode)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MULTIPLE];
+    pthread_once(&memkind_dax_kmem_closest_numanode_once_g[NODE_VARIANT_MULTIPLE],
+                 memkind_dax_kmem_closest_numanode_init);
+
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        set_bitmask_for_all_closest_numanodes(nodemask, maxnode, g->closest_numanode,
+                                              g->num_cpu);
+    }
+    return g->init_err;
+}
+
+static void memkind_dax_kmem_closest_numanode_init(void)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_MULTIPLE];
+    g->num_cpu = numa_num_configured_cpus();
+    g->closest_numanode = NULL;
+    g->init_err = set_closest_numanode(fill_dax_kmem_values_automatic,
+                                       "MEMKIND_DAX_KMEM_NODES", &g->closest_numanode, g->num_cpu, false);
+}
+
+static void memkind_dax_kmem_preferred_closest_numanode_init(void)
+{
+    struct dax_closest_numanode_t *g =
+            &memkind_dax_kmem_closest_numanode_g[NODE_VARIANT_SINGLE];
+    g->num_cpu = numa_num_configured_cpus();
+    g->closest_numanode = NULL;
+    g->init_err = set_closest_numanode(fill_dax_kmem_values_automatic,
+                                       "MEMKIND_DAX_KMEM_NODES", &g->closest_numanode, g->num_cpu, true);
+}
+
+static void memkind_dax_kmem_init_once(void)
+{
+    memkind_init(MEMKIND_DAX_KMEM, true);
+}
+
+static void memkind_dax_kmem_all_init_once(void)
+{
+    memkind_init(MEMKIND_DAX_KMEM_ALL, true);
+}
+
+static void memkind_dax_kmem_preferred_init_once(void)
+{
+    memkind_init(MEMKIND_DAX_KMEM_PREFERRED, true);
+}
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DAX_KMEM_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_dax_kmem_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_dax_kmem_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_dax_kmem_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DAX_KMEM_ALL_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_dax_kmem_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_dax_kmem_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_dax_kmem_all_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DAX_KMEM_PREFERRED_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_dax_kmem_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_dax_kmem_get_preferred_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_dax_kmem_preferred_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
diff --git a/src/memkind_default.c b/src/memkind_default.c
new file mode 100644
index 0000000..9a3b96a
--- /dev/null
+++ b/src/memkind_default.c
@@ -0,0 +1,265 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/heap_manager.h>
+
+#include <numa.h>
+#include <numaif.h>
+#include <sys/mman.h>
+#include <errno.h>
+#include <jemalloc/jemalloc.h>
+#include <stdint.h>
+
+#include "config.h"
+
+#ifndef MADV_NOHUGEPAGE
+#define MADV_NOHUGEPAGE 15
+#endif
+
+static int memkind_default_get_kind_stat(struct memkind *kind,
+                                         memkind_stat_type stat, size_t *value)
+{
+    return memkind_arena_get_stat_with_check_init(kind, stat, true, value);
+}
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_DEFAULT_OPS = {
+    .create = memkind_default_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_default_malloc,
+    .calloc = memkind_default_calloc,
+    .posix_memalign = memkind_default_posix_memalign,
+    .realloc = memkind_default_realloc,
+    .free = memkind_default_free,
+    .init_once = memkind_default_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_default_destroy,
+    .get_stat = memkind_default_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT int memkind_default_create(struct memkind *kind,
+                                          struct memkind_ops *ops, const char *name)
+{
+    int err = 0;
+
+    kind->ops = ops;
+    if (strlen(name) >= MEMKIND_NAME_LENGTH_PRIV) {
+        kind->name[0] = '\0';
+        err = MEMKIND_ERROR_INVALID;
+    } else {
+        strcpy(kind->name, name);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_default_destroy(struct memkind *kind)
+{
+    return 0;
+}
+
+MEMKIND_EXPORT void *memkind_default_malloc(struct memkind *kind, size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+        return NULL;
+    }
+    return jemk_malloc(size);
+}
+
+MEMKIND_EXPORT void *memkind_default_calloc(struct memkind *kind, size_t num,
+                                            size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(num) || size_out_of_bounds(size))) {
+        return NULL;
+    }
+    return jemk_calloc(num, size);
+}
+
+MEMKIND_EXPORT int memkind_default_posix_memalign(struct memkind *kind,
+                                                  void **memptr, size_t alignment, size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+        *memptr = NULL;
+        return 0;
+    }
+    return jemk_posix_memalign(memptr, alignment, size);
+}
+
+MEMKIND_EXPORT void *memkind_default_realloc(struct memkind *kind, void *ptr,
+                                             size_t size)
+{
+    if(MEMKIND_UNLIKELY(size_out_of_bounds(size))) {
+        jemk_free(ptr);
+        return NULL;
+    }
+    return jemk_realloc(ptr, size);
+}
+
+MEMKIND_EXPORT void memkind_default_free(struct memkind *kind, void *ptr)
+{
+    jemk_free(ptr);
+}
+
+MEMKIND_EXPORT size_t memkind_default_malloc_usable_size(struct memkind *kind,
+                                                         void *ptr)
+{
+    return jemk_malloc_usable_size(ptr);
+}
+
+MEMKIND_EXPORT void *memkind_default_mmap(struct memkind *kind, void *addr,
+                                          size_t size)
+{
+    void *result = MAP_FAILED;
+    int err = 0;
+    int flags;
+
+    if (kind->ops->get_mmap_flags) {
+        err = kind->ops->get_mmap_flags(kind, &flags);
+    } else {
+        err = memkind_default_get_mmap_flags(kind, &flags);
+    }
+    if (MEMKIND_LIKELY(!err)) {
+        result = mmap(addr, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+        if (result == MAP_FAILED) {
+            log_err("syscall mmap() returned: %p", result);
+            return result;
+        }
+    }
+    if (kind->ops->mbind) {
+        err = kind->ops->mbind(kind, result, size);
+        if (err) {
+            munmap(result, size);
+            result = MAP_FAILED;
+        }
+    }
+    if (kind->ops->madvise) {
+        err = kind->ops->madvise(kind, result, size);
+        if (err) {
+            munmap(result, size);
+            result = MAP_FAILED;
+        }
+    }
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_nohugepage_madvise(struct memkind *kind, void *addr,
+                                              size_t size)
+{
+    int err = madvise(addr, size, MADV_NOHUGEPAGE);
+
+    //checking if EINVAL was returned due to lack of THP support in kernel
+    if ((err == EINVAL) && (((uintptr_t) addr & (uintptr_t) 0xfff) == 0) &&
+        (size > 0)) {
+        return 0;
+    }
+    if (MEMKIND_UNLIKELY(err)) {
+        log_err("syscall madvise() returned: %d", err);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_default_mbind(struct memkind *kind, void *ptr,
+                                         size_t size)
+{
+    nodemask_t nodemask;
+    int err = 0;
+    int mode;
+
+    if (MEMKIND_UNLIKELY(kind->ops->get_mbind_nodemask == NULL ||
+                         kind->ops->get_mbind_mode == NULL)) {
+        log_err("memkind_ops->mbind_mode or memkind_ops->bind_nodemask is NULL.");
+        return MEMKIND_ERROR_BADOPS;
+    }
+    err = kind->ops->get_mbind_nodemask(kind, nodemask.n, NUMA_NUM_NODES);
+    if (MEMKIND_UNLIKELY(err)) {
+        return err;
+    }
+    err = kind->ops->get_mbind_mode(kind, &mode);
+    if (MEMKIND_UNLIKELY(err)) {
+        return err;
+    }
+    err = mbind(ptr, size, mode, nodemask.n, NUMA_NUM_NODES, 0);
+    if (MEMKIND_UNLIKELY(err)) {
+        log_err("syscall mbind() returned: %d", err);
+        return MEMKIND_ERROR_MBIND;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_default_get_mmap_flags(struct memkind *kind,
+                                                  int *flags)
+{
+    *flags = MAP_PRIVATE | MAP_ANONYMOUS;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_default_get_mbind_nodemask(struct memkind *kind,
+                                                      unsigned long *nodemask,
+                                                      unsigned long maxnode)
+{
+    struct bitmask nodemask_bm = {maxnode, nodemask};
+    copy_bitmask_to_bitmask(numa_all_nodes_ptr, &nodemask_bm);
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_default_get_mbind_mode(struct memkind *kind,
+                                                  int *mode)
+{
+    *mode = MPOL_BIND;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_preferred_get_mbind_mode(struct memkind *kind,
+                                                    int *mode)
+{
+    *mode = MPOL_PREFERRED;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_interleave_get_mbind_mode(struct memkind *kind,
+                                                     int *mode)
+{
+    *mode = MPOL_INTERLEAVE;
+    return 0;
+}
+
+MEMKIND_EXPORT int memkind_posix_check_alignment(struct memkind *kind,
+                                                 size_t alignment)
+{
+    int err = 0;
+    if ((alignment < sizeof(void *)) ||
+        (((alignment - 1) & alignment) != 0)) {
+        err = EINVAL;
+    }
+    return err;
+}
+
+MEMKIND_EXPORT void memkind_default_init_once(void)
+{
+    heap_manager_init(MEMKIND_DEFAULT);
+}
diff --git a/src/memkind_gbtlb.c b/src/memkind_gbtlb.c
new file mode 100644
index 0000000..0470517
--- /dev/null
+++ b/src/memkind_gbtlb.c
@@ -0,0 +1,128 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_gbtlb.h>
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/memkind_arena.h>
+
+static void memkind_hbw_gbtlb_init_once(void);
+static void memkind_hbw_preferred_gbtlb_init_once(void);
+static void memkind_gbtlb_init_once(void);
+
+static void *gbtlb_mmap(struct memkind *kind, void *addr, size_t size);
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_GBTLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = gbtlb_mmap,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_gbtlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_PREFERRED_GBTLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = gbtlb_mmap,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_preferred_gbtlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_GBTLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = gbtlb_mmap,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_gbtlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+#define ONE_GB 1073741824ULL
+
+static void memkind_gbtlb_ceil_size(size_t *size)
+{
+    *size = (*size % ONE_GB) ? ((*size / ONE_GB) + 1) * ONE_GB : *size;
+}
+
+static void *gbtlb_mmap(struct memkind *kind, void *addr, size_t size)
+{
+    memkind_gbtlb_ceil_size(&size);
+    return memkind_default_mmap(kind, addr, size);
+}
+
+static void memkind_hbw_gbtlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_GBTLB, false);
+}
+
+static void memkind_hbw_preferred_gbtlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_PREFERRED_GBTLB, false);
+}
+
+static void memkind_gbtlb_init_once(void)
+{
+    memkind_init(MEMKIND_GBTLB, false);
+}
diff --git a/src/memkind_hbw.c b/src/memkind_hbw.c
new file mode 100644
index 0000000..93d0da6
--- /dev/null
+++ b/src/memkind_hbw.c
@@ -0,0 +1,404 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_bandwidth.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/heap_manager.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <unistd.h>
+#include <limits.h>
+#include <pthread.h>
+#include <numa.h>
+#include <numaif.h>
+#include <errno.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <utmpx.h>
+#include <sched.h>
+#include <stdint.h>
+#include <assert.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_ALL_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_all_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_hugetlb_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_ALL_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_hugetlb_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_all_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_PREFERRED_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_preferred_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_PREFERRED_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_hugetlb_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_mbind_mode = memkind_preferred_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_preferred_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HBW_INTERLEAVE_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hbw_check_available,
+    .mbind = memkind_default_mbind,
+    .madvise = memkind_nohugepage_madvise,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_interleave_get_mbind_mode,
+    .get_mbind_nodemask = memkind_hbw_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hbw_interleave_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+struct hbw_closest_numanode_t {
+    int init_err;
+    int num_cpu;
+    struct vec_cpu_node *closest_numanode;
+};
+
+static struct hbw_closest_numanode_t memkind_hbw_closest_numanode_g;
+static pthread_once_t memkind_hbw_closest_numanode_once_g = PTHREAD_ONCE_INIT;
+
+static void memkind_hbw_closest_numanode_init(void);
+
+// This declaration is necessary, cause it's missing in headers from libnuma 2.0.8
+extern unsigned int numa_bitmask_weight(const struct bitmask *bmp );
+
+static int fill_bandwidth_values_heuristically (int *bandwidth);
+
+MEMKIND_EXPORT int memkind_hbw_check_available(struct memkind *kind)
+{
+    return kind->ops->get_mbind_nodemask(kind, NULL, 0);
+}
+
+MEMKIND_EXPORT int memkind_hbw_hugetlb_check_available(struct memkind *kind)
+{
+    int err = memkind_hbw_check_available(kind);
+    if (!err) {
+        err = memkind_hugetlb_check_available_2mb(kind);
+    }
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_hbw_get_mbind_nodemask(struct memkind *kind,
+                                                  unsigned long *nodemask,
+                                                  unsigned long maxnode)
+{
+    struct hbw_closest_numanode_t *g = &memkind_hbw_closest_numanode_g;
+    pthread_once(&memkind_hbw_closest_numanode_once_g,
+                 memkind_hbw_closest_numanode_init);
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        g->init_err = set_bitmask_for_current_closest_numanode(nodemask, maxnode,
+                                                               g->closest_numanode, g->num_cpu);
+    }
+    return g->init_err;
+}
+
+MEMKIND_EXPORT int memkind_hbw_all_get_mbind_nodemask(struct memkind *kind,
+                                                      unsigned long *nodemask,
+                                                      unsigned long maxnode)
+{
+    struct hbw_closest_numanode_t *g = &memkind_hbw_closest_numanode_g;
+    pthread_once(&memkind_hbw_closest_numanode_once_g,
+                 memkind_hbw_closest_numanode_init);
+
+    if (MEMKIND_LIKELY(!g->init_err)) {
+        set_bitmask_for_all_closest_numanodes(nodemask, maxnode, g->closest_numanode,
+                                              g->num_cpu);
+    }
+    return g->init_err;
+}
+
+typedef struct registers_t {
+    uint32_t eax;
+    uint32_t ebx;
+    uint32_t ecx;
+    uint32_t edx;
+} registers_t;
+
+inline static void cpuid_asm(int leaf, int subleaf, registers_t *registers)
+{
+#ifdef __x86_64__
+    asm volatile("cpuid":"=a"(registers->eax),
+                 "=b"(registers->ebx),
+                 "=c"(registers->ecx),
+                 "=d"(registers->edx):"0"(leaf), "2"(subleaf));
+#else
+    // Non-x86 can't possibly be Knight's Landing nor Knight's Mill.
+    registers->eax = 0;
+#endif
+}
+
+#define CPUID_MODEL_SHIFT       (4)
+#define CPUID_MODEL_MASK        (0xf)
+#define CPUID_EXT_MODEL_MASK    (0xf)
+#define CPUID_EXT_MODEL_SHIFT   (16)
+#define CPUID_FAMILY_MASK       (0xf)
+#define CPUID_FAMILY_SHIFT      (8)
+#define CPU_MODEL_KNL           (0x57)
+#define CPU_MODEL_KNM           (0x85)
+#define CPU_FAMILY_INTEL        (0x06)
+
+typedef struct {
+    uint32_t model;
+    uint32_t family;
+} cpu_model_data_t;
+
+static cpu_model_data_t get_cpu_model_data()
+{
+    registers_t registers;
+    cpuid_asm(1, 0, &registers);
+    uint32_t model = (registers.eax >> CPUID_MODEL_SHIFT) & CPUID_MODEL_MASK;
+    uint32_t model_ext = (registers.eax >> CPUID_EXT_MODEL_SHIFT) &
+                         CPUID_EXT_MODEL_MASK;
+
+    cpu_model_data_t data;
+    data.model = model | (model_ext << 4);
+    data.family = (registers.eax >> CPUID_FAMILY_SHIFT) & CPUID_FAMILY_MASK;
+    return data;
+}
+
+static bool is_hbm_supported(cpu_model_data_t cpu)
+{
+    return cpu.family == CPU_FAMILY_INTEL &&
+           (cpu.model == CPU_MODEL_KNL || cpu.model == CPU_MODEL_KNM);
+}
+
+static int get_high_bandwidth_nodes(struct bitmask *hbw_node_mask)
+{
+    int nodes_num = numa_num_configured_nodes();
+    // Check if NUMA configuration is supported.
+    if(nodes_num == 2 || nodes_num == 4 || nodes_num == 8) {
+        struct bitmask *node_cpus = numa_allocate_cpumask();
+
+        assert(hbw_node_mask->size >= nodes_num);
+        assert(node_cpus->size >= nodes_num);
+        int i;
+        for(i=0; i<nodes_num; i++) {
+            numa_node_to_cpus(i, node_cpus);
+            if(numa_bitmask_weight(node_cpus) == 0) {
+                //NUMA nodes without CPU are HBW nodes.
+                numa_bitmask_setbit(hbw_node_mask, i);
+            }
+        }
+
+        numa_bitmask_free(node_cpus);
+
+        if(2*numa_bitmask_weight(hbw_node_mask) == nodes_num) {
+            return 0;
+        }
+    }
+
+    return MEMKIND_ERROR_UNAVAILABLE;
+}
+
+///This function tries to fill bandwidth array based on knowledge about known CPU models
+static int fill_bandwidth_values_heuristically(int *bandwidth)
+{
+    cpu_model_data_t cpu = get_cpu_model_data();
+
+    if(!is_hbm_supported(cpu)) {
+        log_err("High Bandwidth Memory is not supported by this CPU.");
+        return MEMKIND_ERROR_UNAVAILABLE;
+    }
+
+    switch(cpu.model) {
+        case CPU_MODEL_KNL:
+        case CPU_MODEL_KNM: {
+            int ret = bandwidth_fill(bandwidth, get_high_bandwidth_nodes);
+            if(ret == 0) {
+                log_info("Detected High Bandwidth Memory.");
+            }
+            return ret;
+        }
+        default:
+            return MEMKIND_ERROR_UNAVAILABLE;
+    }
+}
+static void memkind_hbw_closest_numanode_init(void)
+{
+    struct hbw_closest_numanode_t *g = &memkind_hbw_closest_numanode_g;
+    g->num_cpu = numa_num_configured_cpus();
+    g->closest_numanode = NULL;
+    g->init_err = set_closest_numanode(fill_bandwidth_values_heuristically,
+                                       "MEMKIND_HBW_NODES", &g->closest_numanode, g->num_cpu, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_init_once(void)
+{
+    memkind_init(MEMKIND_HBW, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_all_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_ALL, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_HUGETLB, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_all_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_ALL_HUGETLB, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_preferred_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_PREFERRED, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_preferred_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_PREFERRED_HUGETLB, true);
+}
+
+MEMKIND_EXPORT void memkind_hbw_interleave_init_once(void)
+{
+    memkind_init(MEMKIND_HBW_INTERLEAVE, true);
+}
diff --git a/src/memkind_hugetlb.c b/src/memkind_hugetlb.c
new file mode 100644
index 0000000..dab2baa
--- /dev/null
+++ b/src/memkind_hugetlb.c
@@ -0,0 +1,379 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hugetlb.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <sys/mman.h>
+#ifndef MAP_HUGETLB
+#define MAP_HUGETLB 0x40000
+#endif
+#ifndef MAP_HUGE_2MB
+#define MAP_HUGE_2MB (21 << 26)
+#endif
+
+#include <stdio.h>
+#include <errno.h>
+#include <numa.h>
+#include <pthread.h>
+#include <dirent.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_HUGETLB_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_hugetlb_check_available_2mb,
+    .get_mmap_flags = memkind_hugetlb_get_mmap_flags,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_hugetlb_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+static int get_nr_overcommit_hugepages_cached(size_t pagesize, size_t *out);
+static int get_nr_hugepages_cached(size_t pagesize, struct bitmask *nodemask,
+                                   size_t *out);
+
+static int memkind_hugetlb_check_available(struct memkind *kind,
+                                           size_t huge_size);
+
+MEMKIND_EXPORT int memkind_hugetlb_get_mmap_flags(struct memkind *kind,
+                                                  int *flags)
+{
+    *flags = MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB | MAP_HUGE_2MB;
+    return 0;
+}
+
+MEMKIND_EXPORT void memkind_hugetlb_init_once(void)
+{
+    memkind_init(MEMKIND_HUGETLB, true);
+}
+
+MEMKIND_EXPORT int memkind_hugetlb_check_available_2mb(struct memkind *kind)
+{
+    return memkind_hugetlb_check_available(kind, 2097152);
+}
+
+/* huge_size: the huge page size in bytes */
+static int memkind_hugetlb_check_available(struct memkind *kind,
+                                           size_t huge_size)
+{
+    int err = 0;
+    nodemask_t nodemask;
+    struct bitmask nodemask_bm = {NUMA_NUM_NODES, nodemask.n};
+
+    /* on x86_64 default huge page size is 2MB */
+    if (huge_size == 0) {
+        huge_size = 2097152;
+    }
+
+    if (kind->ops->get_mbind_nodemask) {
+        err = kind->ops->get_mbind_nodemask(kind, nodemask.n, NUMA_NUM_NODES);
+    } else {
+        numa_bitmask_setall(&nodemask_bm);
+    }
+
+    size_t nr_persistent_hugepages, nr_overcommit_hugepages;
+
+    err = get_nr_hugepages_cached(huge_size, &nodemask_bm,
+                                  &nr_persistent_hugepages);
+    if(err) {
+        return err;
+    }
+
+    err = get_nr_overcommit_hugepages_cached(huge_size, &nr_overcommit_hugepages);
+    if(err) {
+        return err;
+    }
+
+    if (!nr_overcommit_hugepages && !nr_persistent_hugepages) {
+        log_err("Persistent hugepages and overcommit hugepages are not available.");
+        return MEMKIND_ERROR_HUGETLB;
+    }
+
+    return err;
+}
+
+struct hugepage_size_info {
+    size_t size;
+    size_t *nr_hugepages_per_node_array;
+    size_t  nr_overcommit;
+};
+
+struct memkind_hugepages_config_t {
+    struct hugepage_size_info **hugepages_info_array;
+    int hugepages_info_array_len;
+    int err; // 0 if sysfs parsing successful, appropriate memkind_error otherwise
+} memkind_hugepages_config;
+
+static pthread_once_t memkind_hugepages_config_once_g = PTHREAD_ONCE_INIT;
+
+static struct hugepage_size_info *allocate_hugepage_size_info()
+{
+    struct hugepage_size_info *newInfo = malloc(sizeof(struct hugepage_size_info));
+    if(newInfo == NULL) {
+        log_err("malloc() failed.");
+        return NULL;
+    }
+
+    newInfo->nr_hugepages_per_node_array = calloc(NUMA_NUM_NODES, sizeof(size_t));
+    if(newInfo->nr_hugepages_per_node_array == NULL) {
+        free(newInfo);
+        log_err("calloc() failed.");
+        return NULL;
+    }
+
+    return newInfo;
+}
+
+static size_t get_sysfs_entry_value(const char *entry_path)
+{
+    int errno_before;
+    FILE *fid;
+    int num_read;
+    size_t value_read, ret = 0;
+
+    errno_before = errno;
+    fid = fopen(entry_path, "r");
+    if (fid) {
+        num_read = fscanf(fid, "%zud", &value_read);
+        if(num_read) {
+            ret  = value_read;
+        }
+        fclose(fid);
+    } else {
+        errno = errno_before;
+    }
+    return ret;
+}
+
+// construct hugepage_size_info object and fill it with data for provided pagesize
+static void init_hugepage_size_info(size_t pagesize,
+                                    struct hugepage_size_info *newInfo)
+{
+    char formatted_path[128];
+    const char *nr_path_fmt =
+        "/sys/devices/system/node/node%u/hugepages/hugepages-%zukB/nr_hugepages";
+    const char *nr_overcommit_path_fmt =
+        "/sys/kernel/mm/hugepages/hugepages-%zukB/nr_overcommit_hugepages";
+    int snprintf_ret = 0;
+    size_t node;
+
+    size_t pagesize_kb = pagesize >> 10;
+
+    newInfo->size = pagesize;
+
+    //read overcommit hugepages limit for this pagesize
+    snprintf_ret = snprintf(formatted_path, sizeof(formatted_path),
+                            nr_overcommit_path_fmt, pagesize_kb);
+    if (snprintf_ret > 0 && snprintf_ret < sizeof(formatted_path)) {
+        newInfo->nr_overcommit = get_sysfs_entry_value(formatted_path);
+        log_info("Overcommit limit for %zu kB hugepages is %zu.", pagesize,
+                 newInfo->nr_overcommit);
+    }
+
+    //read every node nr_hugepages for this pagesize
+    for (node = 0; node < NUMA_NUM_NODES; ++node) {
+        snprintf_ret = snprintf(formatted_path, sizeof(formatted_path), nr_path_fmt,
+                                node, pagesize_kb);
+        if(snprintf_ret > 0 && snprintf_ret < sizeof(formatted_path)) {
+            newInfo->nr_hugepages_per_node_array[node] = get_sysfs_entry_value(
+                                                             formatted_path);
+            if(node < numa_num_configured_nodes()) {
+                log_info("Number of %zu kB hugepages on node %zu equals %zu.", pagesize, node,
+                         newInfo->nr_hugepages_per_node_array[node]);
+            }
+        }
+    }
+}
+
+// get hugepage size in bytes out of sysfs dir name
+static int parse_pagesize_from_sysfs_entry(const char *entry, size_t *out)
+{
+    size_t pagesize;
+    int ret = sscanf(entry, "hugepages-%zukB", &pagesize);
+
+    if(ret == 1) {
+        *out = pagesize << 10; //we are using bytes but kernel is using kB
+        return 0;
+    }
+
+    return -1;
+}
+
+
+static void hugepages_config_init_once()
+{
+    unsigned j, i = 0;
+    size_t pagesize;
+    struct hugepage_size_info **hugepages_info_array = NULL;
+    struct dirent *dir;
+    DIR *hugepages_sysfs = opendir("/sys/kernel/mm/hugepages");
+    if(hugepages_sysfs == NULL) {
+        memkind_hugepages_config.err = MEMKIND_ERROR_HUGETLB;
+        log_err("/sys/kernel/mm/hugepages directory is not available.");
+        return;
+    }
+
+    unsigned hugepages_info_array_len = 2; //initial size of array
+    hugepages_info_array = malloc(hugepages_info_array_len * sizeof(
+                                      struct hugepage_size_info *));
+    if (hugepages_info_array == NULL) {
+        memkind_hugepages_config.err = MEMKIND_ERROR_MALLOC;
+        closedir(hugepages_sysfs);
+        log_err("malloc() failed.");
+        return;
+    }
+
+    while ((dir = readdir(hugepages_sysfs)) != NULL) {
+        if(dir->d_type == DT_DIR &&
+           parse_pagesize_from_sysfs_entry(dir->d_name, &pagesize) == 0) {
+            struct hugepage_size_info *new_hugepage_info = allocate_hugepage_size_info();
+            if(new_hugepage_info == NULL) {
+                memkind_hugepages_config.err = MEMKIND_ERROR_MALLOC;
+                break;
+            }
+
+            init_hugepage_size_info(pagesize, new_hugepage_info);
+
+            //there is more hugepage sizes than expected, reallocation of array needed
+            if(i == hugepages_info_array_len) {
+                hugepages_info_array_len *= 2;
+                struct hugepage_size_info **swap_tmp = realloc(hugepages_info_array,
+                                                               hugepages_info_array_len * sizeof(struct hugepage_size_info *));
+                if(swap_tmp == NULL) {
+                    free(new_hugepage_info);
+                    memkind_hugepages_config.err = MEMKIND_ERROR_MALLOC;
+                    log_err("realloc() failed.");
+                    break;
+                }
+                hugepages_info_array = swap_tmp;
+
+            }
+            hugepages_info_array[i] = new_hugepage_info;
+            i++;
+        }
+    }
+
+    closedir(hugepages_sysfs);
+
+    if(memkind_hugepages_config.err == 0) {
+        memkind_hugepages_config.hugepages_info_array = hugepages_info_array;
+        memkind_hugepages_config.hugepages_info_array_len = i;
+    } else {
+        for(j=0; j<i; j++) {
+            free(hugepages_info_array[i]);
+        }
+        free(hugepages_info_array);
+    }
+
+    return;
+}
+
+#ifdef __GNUC__
+__attribute__((destructor))
+#endif
+static void destroy_hugepages_per_node()
+{
+    int i;
+    for(i=0; i<memkind_hugepages_config.hugepages_info_array_len; i++) {
+        free(memkind_hugepages_config.hugepages_info_array[i]);
+    }
+    free(memkind_hugepages_config.hugepages_info_array);
+}
+
+// helper function that find and return hugepage_size_info object for specified pagesize
+static struct hugepage_size_info *get_hugepage_info_for_pagesize(
+    size_t pagesize)
+{
+    int i;
+
+    for(i=0; i<memkind_hugepages_config.hugepages_info_array_len; i++) {
+        if(memkind_hugepages_config.hugepages_info_array[i]->size == pagesize) {
+            return memkind_hugepages_config.hugepages_info_array[i];
+        }
+    }
+    return NULL;
+}
+
+// returns sum of pre-allocated hugepage for specified pagesize and set of nodes
+static int get_nr_hugepages_cached(size_t pagesize, struct bitmask *nodemask,
+                                   size_t *out)
+{
+    int i;
+    size_t nr_hugepages = 0;
+    int num_node = numa_num_configured_nodes();
+    pthread_once(&memkind_hugepages_config_once_g,
+                 hugepages_config_init_once);
+
+
+    if(memkind_hugepages_config.err != 0) {
+        return memkind_hugepages_config.err;
+    }
+
+    struct hugepage_size_info *info = get_hugepage_info_for_pagesize(pagesize);
+    if(info == NULL) {
+        log_err("Unable to allocate hugepages, because info about pre-allocated hugepages is not available.");
+        return MEMKIND_ERROR_HUGETLB;
+    }
+
+    for(i=0; i<num_node; i++) {
+        if(numa_bitmask_isbitset(nodemask, i)) {
+            nr_hugepages += info->nr_hugepages_per_node_array[i];
+        }
+    }
+
+    *out = nr_hugepages;
+    return 0;
+}
+
+// returns hugepages overcommit limit for specified pagesize
+static int get_nr_overcommit_hugepages_cached(size_t pagesize, size_t *out)
+{
+    pthread_once(&memkind_hugepages_config_once_g,
+                 hugepages_config_init_once);
+
+    if(memkind_hugepages_config.err != 0) {
+        return memkind_hugepages_config.err;
+    }
+
+    struct hugepage_size_info *info = get_hugepage_info_for_pagesize(pagesize);
+    if(info == NULL) {
+        log_err("Unable to allocate hugepages, because info about overcommit hugepages is not available.");
+        return MEMKIND_ERROR_HUGETLB;
+    }
+
+    *out = info->nr_overcommit;
+    return 0;
+}
+
diff --git a/src/memkind_interleave.c b/src/memkind_interleave.c
new file mode 100644
index 0000000..79d56f8
--- /dev/null
+++ b/src/memkind_interleave.c
@@ -0,0 +1,54 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_interleave.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_private.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_INTERLEAVE_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mbind = memkind_default_mbind,
+    .madvise = memkind_nohugepage_madvise,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_interleave_get_mbind_mode,
+    .get_mbind_nodemask = memkind_default_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_interleave_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_arena_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+MEMKIND_EXPORT void memkind_interleave_init_once(void)
+{
+    memkind_init(MEMKIND_INTERLEAVE, true);
+}
diff --git a/src/memkind_log.c b/src/memkind_log.c
new file mode 100644
index 0000000..86aa8c1
--- /dev/null
+++ b/src/memkind_log.c
@@ -0,0 +1,104 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_log.h>
+
+#include <assert.h>
+#include <stdio.h>
+#include <pthread.h>
+#include <stdarg.h>
+#include <stdlib.h>
+#include <string.h>
+#include <stdbool.h>
+
+typedef enum {
+    MESSAGE_TYPE_INFO,
+    MESSAGE_TYPE_ERROR,
+    MESSAGE_TYPE_FATAL,
+    MESSAGE_TYPE_MAX_VALUE,
+} message_type_t;
+
+static char *message_prefixes[MESSAGE_TYPE_MAX_VALUE] = {
+    [MESSAGE_TYPE_INFO]     = "MEMKIND_INFO",
+    [MESSAGE_TYPE_ERROR]    = "MEMKIND_ERROR",
+    [MESSAGE_TYPE_FATAL]    = "MEMKIND_FATAL",
+};
+
+static bool log_enabled;
+
+static pthread_once_t init_once = PTHREAD_ONCE_INIT;
+static void log_init_once(void)
+{
+    char *memkind_debug_env = secure_getenv("MEMKIND_DEBUG");
+
+    if (memkind_debug_env) {
+        if(strcmp(memkind_debug_env, "1") == 0) {
+            log_enabled = true;
+        } else {
+            fprintf(stderr,
+                    "MEMKIND_WARNING: debug option \"%s\" unknown; Try man memkind for available options.\n",
+                    memkind_debug_env);
+        }
+    }
+}
+
+static pthread_mutex_t log_lock = PTHREAD_MUTEX_INITIALIZER;
+static void log_generic(message_type_t type, const char *format, va_list args)
+{
+    pthread_once(&init_once, log_init_once);
+    if(log_enabled || (type == MESSAGE_TYPE_FATAL)) {
+        if (pthread_mutex_lock(&log_lock) != 0)
+            assert(0 && "failed to acquire mutex");
+        fprintf(stderr, "%s: ", message_prefixes[type]);
+        vfprintf(stderr, format, args);
+        fprintf(stderr, "\n");
+        if (pthread_mutex_unlock(&log_lock) != 0)
+            assert(0 && "failed to release mutex");
+    }
+}
+
+void log_info(const char *format, ...)
+{
+    va_list args;
+    va_start(args, format);
+    log_generic(MESSAGE_TYPE_INFO, format, args);
+    va_end(args);
+}
+
+void log_err(const char *format, ...)
+{
+    va_list args;
+    va_start(args, format);
+    log_generic(MESSAGE_TYPE_ERROR, format, args);
+    va_end(args);
+}
+
+void log_fatal(const char *format, ...)
+{
+    va_list args;
+    va_start(args, format);
+    log_generic(MESSAGE_TYPE_FATAL, format, args);
+    va_end(args);
+}
+
diff --git a/src/memkind_pmem.c b/src/memkind_pmem.c
new file mode 100644
index 0000000..deeef18
--- /dev/null
+++ b/src/memkind_pmem.c
@@ -0,0 +1,292 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_pmem.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/memkind_log.h>
+
+#include <sys/mman.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <errno.h>
+#include <assert.h>
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_PMEM_OPS = {
+    .create = memkind_pmem_create,
+    .destroy = memkind_pmem_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .mmap = memkind_pmem_mmap,
+    .get_mmap_flags = memkind_pmem_get_mmap_flags,
+    .get_arena = memkind_thread_get_arena,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_pmem_destroy,
+    .update_memory_usage_policy = memkind_arena_update_memory_usage_policy,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+void *pmem_extent_alloc(extent_hooks_t *extent_hooks,
+                        void *new_addr,
+                        size_t size,
+                        size_t alignment,
+                        bool *zero,
+                        bool *commit,
+                        unsigned arena_ind)
+{
+    int err;
+    void *addr = NULL;
+
+    if (new_addr != NULL) {
+        /* not supported */
+        goto exit;
+    }
+
+    struct memkind *kind = get_kind_by_arena(arena_ind);
+    if (kind == NULL) {
+        return NULL;
+    }
+
+    err = memkind_check_available(kind);
+    if (err) {
+        goto exit;
+    }
+
+    addr = memkind_pmem_mmap(kind, new_addr, size);
+
+    if (addr != MAP_FAILED) {
+        *zero = true;
+        *commit = true;
+
+        /* XXX - check alignment */
+    } else {
+        addr = NULL;
+    }
+exit:
+    return addr;
+}
+
+bool pmem_extent_dalloc(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        bool committed,
+                        unsigned arena_ind)
+{
+    // if madvise fail, it means that addr isn't mapped shared (doesn't come from pmem)
+    // and it should be unmapped to avoid space exhaustion when calling large number of
+    // operations like memkind_create_pmem and memkind_destroy_kind
+    errno = 0;
+    int status = madvise(addr, size, MADV_REMOVE);
+    if (!status) {
+        struct memkind *kind = get_kind_by_arena(arena_ind);
+        struct memkind_pmem *priv = kind->priv;
+        if (pthread_mutex_lock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to acquire mutex");
+        priv->current_size -= size;
+        if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to release mutex");
+    } else {
+        if (errno == EOPNOTSUPP) {
+            log_fatal("Filesystem doesn't support FALLOC_FL_PUNCH_HOLE.");
+            abort();
+        }
+        if (munmap(addr, size) == -1) {
+            log_err("munmap failed!");
+        }
+    }
+    return true;
+}
+
+bool pmem_extent_commit(extent_hooks_t *extent_hooks,
+                        void *addr,
+                        size_t size,
+                        size_t offset,
+                        size_t length,
+                        unsigned arena_ind)
+{
+    /* do nothing - report success */
+    return false;
+}
+
+bool pmem_extent_decommit(extent_hooks_t *extent_hooks,
+                          void *addr,
+                          size_t size,
+                          size_t offset,
+                          size_t length,
+                          unsigned arena_ind)
+{
+    /* do nothing - report failure (opt-out) */
+    return true;
+}
+
+bool pmem_extent_purge(extent_hooks_t *extent_hooks,
+                       void *addr,
+                       size_t size,
+                       size_t offset,
+                       size_t length,
+                       unsigned arena_ind)
+{
+    /* do nothing - report failure (opt-out) */
+    return true;
+}
+
+bool pmem_extent_split(extent_hooks_t *extent_hooks,
+                       void *addr,
+                       size_t size,
+                       size_t size_a,
+                       size_t size_b,
+                       bool committed,
+                       unsigned arena_ind)
+{
+    /* do nothing - report success */
+    return false;
+}
+
+bool pmem_extent_merge(extent_hooks_t *extent_hooks,
+                       void *addr_a,
+                       size_t size_a,
+                       void *addr_b,
+                       size_t size_b,
+                       bool committed,
+                       unsigned arena_ind)
+{
+    /* do nothing - report success */
+    return false;
+}
+
+void pmem_extent_destroy(extent_hooks_t *extent_hooks,
+                         void *addr,
+                         size_t size,
+                         bool committed,
+                         unsigned arena_ind)
+{
+    if (munmap(addr, size) == -1) {
+        log_err("munmap failed!");
+    }
+}
+
+static extent_hooks_t pmem_extent_hooks = {
+    .alloc = pmem_extent_alloc,
+    .dalloc = pmem_extent_dalloc,
+    .commit = pmem_extent_commit,
+    .decommit = pmem_extent_decommit,
+    .purge_lazy = pmem_extent_purge,
+    .split = pmem_extent_split,
+    .merge = pmem_extent_merge,
+    .destroy = pmem_extent_destroy
+};
+
+MEMKIND_EXPORT int memkind_pmem_create(struct memkind *kind,
+                                       struct memkind_ops *ops, const char *name)
+{
+    struct memkind_pmem *priv;
+    int err;
+
+    priv = (struct memkind_pmem *)malloc(sizeof(struct memkind_pmem));
+    if (!priv) {
+        log_err("malloc() failed.");
+        return MEMKIND_ERROR_MALLOC;
+    }
+
+    if (pthread_mutex_init(&priv->pmem_lock, NULL) != 0) {
+        err = MEMKIND_ERROR_RUNTIME;
+        goto exit;
+    }
+
+    err = memkind_default_create(kind, ops, name);
+    if (err) {
+        goto exit;
+    }
+
+    err = memkind_arena_create_map(kind, &pmem_extent_hooks);
+    if (err) {
+        goto exit;
+    }
+
+    kind->priv = priv;
+    return 0;
+
+exit:
+    /* err is set, please don't overwrite it with result of pthread_mutex_destroy */
+    pthread_mutex_destroy(&priv->pmem_lock);
+    free(priv);
+    return err;
+}
+
+MEMKIND_EXPORT int memkind_pmem_destroy(struct memkind *kind)
+{
+    struct memkind_pmem *priv = kind->priv;
+
+    memkind_arena_destroy(kind);
+
+    pthread_mutex_destroy(&priv->pmem_lock);
+
+    (void) close(priv->fd);
+    free(priv);
+
+    return 0;
+}
+
+MEMKIND_EXPORT void *memkind_pmem_mmap(struct memkind *kind, void *addr,
+                                       size_t size)
+{
+    struct memkind_pmem *priv = kind->priv;
+    void *result;
+
+    if (pthread_mutex_lock(&priv->pmem_lock) != 0)
+        assert(0 && "failed to acquire mutex");
+
+    if (priv->max_size != 0 && priv->current_size + size > priv->max_size) {
+        if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to release mutex");
+        return MAP_FAILED;
+    }
+
+    if ((errno = posix_fallocate(priv->fd, priv->offset, (off_t)size)) != 0) {
+        if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+            assert(0 && "failed to release mutex");
+        return MAP_FAILED;
+    }
+
+    if ((result = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, priv->fd,
+                       priv->offset)) != MAP_FAILED) {
+        priv->offset += size;
+        priv->current_size += size;
+    }
+
+    if (pthread_mutex_unlock(&priv->pmem_lock) != 0)
+        assert(0 && "failed to release mutex");
+
+    return result;
+}
+
+MEMKIND_EXPORT int memkind_pmem_get_mmap_flags(struct memkind *kind, int *flags)
+{
+    *flags = MAP_SHARED;
+    return 0;
+}
diff --git a/src/memkind_regular.c b/src/memkind_regular.c
new file mode 100644
index 0000000..4cd1887
--- /dev/null
+++ b/src/memkind_regular.c
@@ -0,0 +1,109 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include <memkind/internal/memkind_arena.h>
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/heap_manager.h>
+
+#include <numa.h>
+
+static struct bitmask *regular_nodes_mask = NULL;
+
+static void regular_nodes_init(void)
+{
+    unsigned i;
+    unsigned nodes_num = (unsigned)numa_num_configured_nodes();
+    struct bitmask *node_cpus = numa_allocate_cpumask();
+
+    regular_nodes_mask = numa_allocate_nodemask();
+
+    for (i = 0; i < nodes_num; i++) {
+        numa_node_to_cpus(i, node_cpus);
+        if (numa_bitmask_weight(node_cpus))
+            numa_bitmask_setbit(regular_nodes_mask, i);
+    }
+    numa_bitmask_free(node_cpus);
+}
+
+static void memkind_regular_init_once(void)
+{
+    regular_nodes_init();
+    memkind_init(MEMKIND_REGULAR, true);
+}
+
+static int memkind_regular_check_available(struct memkind *kind)
+{
+    /* init_once method is called in memkind_malloc function
+     * when memkind malloc is not called this function will fail.
+     * Call pthread_once to be sure that situation mentioned
+     * above will never happen */
+    pthread_once(&kind->init_once, kind->ops->init_once);
+    return regular_nodes_mask != NULL ? MEMKIND_SUCCESS : MEMKIND_ERROR_UNAVAILABLE;
+}
+
+MEMKIND_EXPORT int memkind_regular_all_get_mbind_nodemask(struct memkind *kind,
+                                                          unsigned long *nodemask,
+                                                          unsigned long maxnode)
+{
+    struct bitmask nodemask_bm = {maxnode, nodemask};
+
+    if(!regular_nodes_mask) {
+        return MEMKIND_ERROR_UNAVAILABLE;
+    }
+
+    copy_bitmask_to_bitmask(regular_nodes_mask, &nodemask_bm);
+    return MEMKIND_SUCCESS;
+}
+
+static int memkind_regular_finalize(memkind_t kind)
+{
+    if(regular_nodes_mask)
+        numa_bitmask_free(regular_nodes_mask);
+
+    return memkind_arena_finalize(kind);
+}
+
+MEMKIND_EXPORT struct memkind_ops MEMKIND_REGULAR_OPS = {
+    .create = memkind_arena_create,
+    .destroy = memkind_default_destroy,
+    .malloc = memkind_arena_malloc,
+    .calloc = memkind_arena_calloc,
+    .posix_memalign = memkind_arena_posix_memalign,
+    .realloc = memkind_arena_realloc,
+    .free = memkind_arena_free,
+    .check_available = memkind_regular_check_available,
+    .mbind = memkind_default_mbind,
+    .get_mmap_flags = memkind_default_get_mmap_flags,
+    .get_mbind_mode = memkind_default_get_mbind_mode,
+    .get_mbind_nodemask = memkind_regular_all_get_mbind_nodemask,
+    .get_arena = memkind_thread_get_arena,
+    .init_once = memkind_regular_init_once,
+    .malloc_usable_size = memkind_default_malloc_usable_size,
+    .finalize = memkind_regular_finalize,
+    .get_stat = memkind_arena_get_kind_stat,
+    .defrag_reallocate = memkind_arena_defrag_reallocate
+};
+
+
diff --git a/src/tbb_wrapper.c b/src/tbb_wrapper.c
new file mode 100644
index 0000000..869a5d4
--- /dev/null
+++ b/src/tbb_wrapper.c
@@ -0,0 +1,308 @@
+/*
+ * Copyright (C) 2017 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_default.h>
+#include <memkind/internal/memkind_log.h>
+#include <memkind/internal/memkind_private.h>
+#include <memkind/internal/tbb_wrapper.h>
+#include <memkind/internal/tbb_mem_pool_policy.h>
+#include <limits.h>
+
+#include <stdint.h>
+#include <errno.h>
+#include <stdio.h>
+#include <sys/mman.h>
+#include <stdlib.h>
+#include <dlfcn.h>
+#include <string.h>
+
+void *(*pool_malloc)(void *, size_t);
+void *(*pool_realloc)(void *, void *, size_t);
+void *(*pool_aligned_malloc)(void *, size_t, size_t);
+bool (*pool_free)(void *, void *);
+int (*pool_create_v1)(intptr_t, const struct MemPoolPolicy *, void **);
+bool (*pool_destroy)(void *);
+void *(*pool_identify)(void *object);
+size_t (*pool_msize)(void *, void *);
+
+static void *tbb_handle = NULL;
+static bool TBBInitDone = false;
+
+void load_tbb_symbols(void)
+{
+    const char so_name[]="libtbbmalloc.so.2";
+    tbb_handle = dlopen(so_name, RTLD_LAZY);
+    if(!tbb_handle) {
+        log_fatal("%s not found.", so_name);
+        abort();
+    }
+
+    pool_malloc = dlsym(tbb_handle, "_ZN3rml11pool_mallocEPNS_10MemoryPoolEm");
+    pool_realloc = dlsym(tbb_handle, "_ZN3rml12pool_reallocEPNS_10MemoryPoolEPvm");
+    pool_aligned_malloc = dlsym(tbb_handle,
+                                "_ZN3rml19pool_aligned_mallocEPNS_10MemoryPoolEmm");
+    pool_free = dlsym(tbb_handle, "_ZN3rml9pool_freeEPNS_10MemoryPoolEPv");
+    pool_create_v1 = dlsym(tbb_handle,
+                           "_ZN3rml14pool_create_v1ElPKNS_13MemPoolPolicyEPPNS_10MemoryPoolE");
+    pool_destroy = dlsym(tbb_handle, "_ZN3rml12pool_destroyEPNS_10MemoryPoolE");
+    pool_identify = dlsym(tbb_handle, "_ZN3rml13pool_identifyEPv");
+    pool_msize = dlsym(tbb_handle, "_ZN3rml10pool_msizeEPNS_10MemoryPoolEPv");
+
+    if(!pool_malloc ||
+       !pool_realloc ||
+       !pool_aligned_malloc ||
+       !pool_free ||
+       !pool_create_v1 ||
+       !pool_destroy ||
+       !pool_identify ) {
+        log_fatal("Could not find symbols in %s.", so_name);
+        dlclose(tbb_handle);
+        abort();
+    }
+    TBBInitDone = true;
+}
+
+//Granularity of raw_alloc allocations
+#define GRANULARITY 2*1024*1024
+static void *raw_alloc(intptr_t pool_id, size_t *bytes/*=n*GRANULARITY*/)
+{
+    void *ptr = kind_mmap((struct memkind *)pool_id, NULL, *bytes);
+    return (ptr==MAP_FAILED) ? NULL : ptr;
+}
+
+static int raw_free(intptr_t pool_id, void *raw_ptr, size_t raw_bytes)
+{
+    return munmap(raw_ptr, raw_bytes);
+}
+
+static void *tbb_pool_malloc(struct memkind *kind, size_t size)
+{
+    if(size_out_of_bounds(size)) return NULL;
+    void *result = pool_malloc(kind->priv, size);
+    if (!result)
+        errno = ENOMEM;
+    return result;
+}
+
+static void *tbb_pool_calloc(struct memkind *kind, size_t num, size_t size)
+{
+    if (size_out_of_bounds(num) || size_out_of_bounds(size)) return NULL;
+
+    const size_t array_size = num*size;
+    if (array_size/num != size) {
+        errno = ENOMEM;
+        return NULL;
+    }
+    void *result = pool_malloc(kind->priv, array_size);
+    if (result) {
+        memset(result, 0, array_size);
+    } else {
+        errno = ENOMEM;
+    }
+    return result;
+}
+
+static void *tbb_pool_common_realloc(void *pool, void *ptr, size_t size)
+{
+    if (size_out_of_bounds(size)) {
+        pool_free(pool, ptr);
+        return NULL;
+    }
+    void *result = pool_realloc(pool, ptr, size);
+    if (!result)
+        errno = ENOMEM;
+    return result;
+}
+
+static void *tbb_pool_realloc(struct memkind *kind, void *ptr, size_t size)
+{
+    return tbb_pool_common_realloc(kind->priv, ptr, size);
+}
+
+void *tbb_pool_realloc_with_kind_detect(void *ptr, size_t size)
+{
+    if (!ptr) {
+        errno = EINVAL;
+        return NULL;
+    }
+    return tbb_pool_common_realloc(pool_identify(ptr), ptr, size);
+}
+
+int tbb_get_global_stat(memkind_stat_type stat, size_t *value)
+{
+    log_err("Get allocator statistic is not supported by TBB");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+
+static int tbb_get_kind_stat(memkind_t kind, memkind_stat_type stat,
+                             size_t *value)
+{
+    log_err("Get kind statistic is not supported by TBB");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+
+int tbb_update_cached_stats(void)
+{
+    log_err("Update cached statistics is not supported by TBB");
+    return MEMKIND_ERROR_OPERATION_FAILED;
+}
+
+void *tbb_pool_defrag_reallocate_with_kind_detect(void *ptr)
+{
+    log_err("Defrag reallocate method is not supported by TBB");
+    return NULL;
+}
+
+static void *tbb_defrag_reallocate(struct memkind *kind, void *ptr)
+{
+    log_err("Defrag reallocate method is not supported by TBB");
+    return NULL;
+}
+
+struct memkind *tbb_detect_kind(void *ptr)
+{
+    if (!ptr) {
+        return NULL;
+    }
+    struct memkind *kind = NULL;
+    unsigned i;
+    void *pool = pool_identify(ptr);
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_get_kind_by_partition(i, &kind);
+        if (!err && kind->priv == pool) {
+            break;
+        }
+    }
+
+    return kind;
+}
+
+static int tbb_pool_posix_memalign(struct memkind *kind, void **memptr,
+                                   size_t alignment, size_t size)
+{
+    //Check if alignment is "at least as large as sizeof(void *)".
+    if(!alignment && (0 != (alignment & (alignment-sizeof(void *))))) return EINVAL;
+    //Check if alignment is "a power of 2".
+    if(alignment & (alignment-1)) return EINVAL;
+    if(size_out_of_bounds(size)) {
+        *memptr = NULL;
+        return 0;
+    }
+    void *result = pool_aligned_malloc(kind->priv, size, alignment);
+    if (!result) {
+        return ENOMEM;
+    }
+    *memptr = result;
+    return 0;
+}
+
+void tbb_pool_free_with_kind_detect(void *ptr)
+{
+    if (ptr) {
+        pool_free(pool_identify(ptr), ptr);
+    }
+}
+
+static void tbb_pool_free(struct memkind *kind, void *ptr)
+{
+    pool_free(kind->priv, ptr);
+}
+
+static size_t tbb_pool_common_malloc_usable_size(void *pool, void *ptr)
+{
+    if (pool_msize) {
+        return pool_msize(pool, ptr);
+    } else {
+        log_err("memkind_malloc_usable_size() is not supported by this TBB version.");
+        return 0;
+    }
+}
+
+static size_t tbb_pool_malloc_usable_size(struct memkind *kind, void *ptr)
+{
+    return tbb_pool_common_malloc_usable_size(kind->priv, ptr);
+}
+
+size_t tbb_pool_malloc_usable_size_with_kind_detect(void *ptr)
+{
+    size_t size = 0;
+    if (ptr) {
+        size = tbb_pool_common_malloc_usable_size(pool_identify(ptr), ptr);
+    }
+    return size;
+}
+
+static int tbb_update_memory_usage_policy(struct memkind *kind,
+                                          memkind_mem_usage_policy policy)
+{
+    log_info("Memkind memory usage policies have no effect in TBB manager.");
+    return MEMKIND_SUCCESS;
+}
+
+static int tbb_destroy(struct memkind *kind)
+{
+    bool pool_destroy_ret = pool_destroy(kind->priv);
+    dlclose(tbb_handle);
+
+    if(!pool_destroy_ret) {
+        log_err("TBB pool destroy failure.");
+        return MEMKIND_ERROR_OPERATION_FAILED;
+    }
+    return MEMKIND_SUCCESS;
+}
+
+void tbb_initialize(struct memkind *kind)
+{
+    if(!kind || !TBBInitDone) {
+        log_fatal("Failed to initialize TBB.");
+        abort();
+    }
+
+    struct MemPoolPolicy policy = {
+        .pAlloc = raw_alloc,
+        .pFree = raw_free,
+        .granularity = GRANULARITY,
+        .version = 1,
+        .fixedPool = false,
+        .keepAllMemory = false,
+        .reserved = 0
+    };
+
+    pool_create_v1((intptr_t)kind, &policy, &kind->priv);
+    if (!kind->priv) {
+        log_fatal("Unable to create TBB memory pool.");
+        abort();
+    }
+
+    kind->ops->malloc = tbb_pool_malloc;
+    kind->ops->calloc = tbb_pool_calloc;
+    kind->ops->posix_memalign = tbb_pool_posix_memalign;
+    kind->ops->realloc = tbb_pool_realloc;
+    kind->ops->free = tbb_pool_free;
+    kind->ops->finalize = tbb_destroy;
+    kind->ops->malloc_usable_size = tbb_pool_malloc_usable_size;
+    kind->ops->update_memory_usage_policy = tbb_update_memory_usage_policy;
+    kind->ops->get_stat = tbb_get_kind_stat;
+    kind->ops->defrag_reallocate = tbb_defrag_reallocate;
+}
diff --git a/test/Allocator.hpp b/test/Allocator.hpp
new file mode 100644
index 0000000..5156efc
--- /dev/null
+++ b/test/Allocator.hpp
@@ -0,0 +1,222 @@
+/*
+* Copyright (C) 2017 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <stdio.h>
+#include <assert.h>
+#include <map>
+#include <iterator>
+
+class Allocator
+{
+public:
+
+    virtual void *malloc(size_t size) = 0;
+    virtual void *calloc(size_t num, size_t size) = 0;
+    virtual void *realloc(void *ptr, size_t size) = 0;
+    virtual int memalign(void **ptr, size_t alignment, size_t size) = 0;
+    virtual void free(void *ptr) = 0;
+
+    /* get_numa_policy() returns MPOL_INTERLEAVE, MPOL_BIND, MPOL_PREFERRED
+     or -1 when the allocator is not providing NUMA policy */
+    virtual int get_numa_policy() = 0;
+    virtual bool is_high_bandwidth() = 0;
+    virtual size_t get_page_size() = 0;
+
+    virtual ~Allocator(void) {}
+};
+
+class MemkindAllocator : public Allocator
+{
+public:
+
+    MemkindAllocator(memkind_memtype_t memtype, memkind_policy_t policy,
+                     memkind_bits_t flags) :
+        memtype(memtype),
+        flags(flags)
+    {
+        int ret = memkind_create_kind(memtype, policy, flags, &kind);
+        assert(ret == MEMKIND_SUCCESS);
+        assert(kind != NULL);
+    }
+
+    MemkindAllocator(memkind_t kind) : kind(kind)
+    {
+        assert(kind != NULL);
+    }
+
+    virtual void *malloc(size_t size)
+    {
+        return memkind_malloc(kind, size);
+    }
+
+    virtual void *calloc(size_t num, size_t size)
+    {
+        return memkind_calloc(kind, num, size);
+    }
+
+    virtual void *realloc(void *ptr, size_t size)
+    {
+        return memkind_realloc(kind, ptr, size);
+    }
+
+    virtual int memalign(void **ptr, size_t alignment, size_t size)
+    {
+        return memkind_posix_memalign(kind, ptr, alignment, size);
+    }
+
+    virtual void free(void *ptr)
+    {
+        memkind_free(kind, ptr);
+    }
+
+    virtual int get_numa_policy()
+    {
+        static std::map<memkind_t, int> kind_policy = {
+            std::make_pair(MEMKIND_HBW_INTERLEAVE, MPOL_INTERLEAVE),
+            std::make_pair(MEMKIND_INTERLEAVE, MPOL_INTERLEAVE),
+            std::make_pair(MEMKIND_HBW_PREFERRED, MPOL_PREFERRED),
+            std::make_pair(MEMKIND_HBW_PREFERRED_HUGETLB, MPOL_PREFERRED),
+            std::make_pair(MEMKIND_HBW, MPOL_BIND),
+            std::make_pair(MEMKIND_HBW_HUGETLB, MPOL_BIND),
+            std::make_pair(MEMKIND_REGULAR, MPOL_BIND),
+            std::make_pair(MEMKIND_DEFAULT, MPOL_DEFAULT),
+            std::make_pair(MEMKIND_HUGETLB, MPOL_DEFAULT),
+            std::make_pair(MEMKIND_HBW_ALL_HUGETLB, MPOL_BIND),
+            std::make_pair(MEMKIND_HBW_ALL, MPOL_BIND)
+        };
+        auto it = kind_policy.find(kind);
+
+        if(it != std::end(kind_policy)) {
+            return it->second;
+        }
+        return -1;
+    }
+
+    virtual bool is_high_bandwidth()
+    {
+        return  memtype == MEMKIND_MEMTYPE_HIGH_BANDWIDTH ||
+                kind == MEMKIND_HBW ||
+                kind == MEMKIND_HBW_HUGETLB ||
+                kind == MEMKIND_HBW_PREFERRED ||
+                kind == MEMKIND_HBW_INTERLEAVE;
+    }
+
+    virtual size_t get_page_size()
+    {
+        if (kind == MEMKIND_HUGETLB ||
+            kind == MEMKIND_HBW_HUGETLB ||
+            kind == MEMKIND_HBW_PREFERRED_HUGETLB ||
+            (flags & MEMKIND_MASK_PAGE_SIZE_2MB)) {
+            return 2*MB;
+        }
+        return 4*KB;
+    }
+
+    virtual ~MemkindAllocator()
+    {
+        int ret = memkind_destroy_kind(kind);
+        assert(ret == MEMKIND_SUCCESS);
+        kind = NULL;
+    }
+
+private:
+    memkind_t kind = NULL;
+    memkind_memtype_t memtype = memkind_memtype_t();
+    memkind_bits_t flags = memkind_bits_t();
+};
+
+class HbwmallocAllocator : public Allocator
+{
+public:
+
+    HbwmallocAllocator(hbw_policy_t hbw_policy)
+    {
+        hbw_set_policy(hbw_policy);
+    }
+
+    virtual void *malloc(size_t size)
+    {
+        return hbw_malloc(size);
+    }
+
+    virtual void *calloc(size_t num, size_t size)
+    {
+        return hbw_calloc(num, size);
+    }
+
+    virtual void *realloc(void *ptr, size_t size)
+    {
+        return hbw_realloc(ptr, size);
+    }
+
+    virtual void set_memalign_page_size(hbw_pagesize_t psize)
+    {
+        page_size = psize;
+    }
+
+    virtual int memalign(void **ptr, size_t alignment, size_t size)
+    {
+        if (page_size == HBW_PAGESIZE_4KB) {
+            return hbw_posix_memalign(ptr, alignment, size);
+        } else {
+            return hbw_posix_memalign_psize(ptr, alignment, size, page_size);
+        }
+    }
+
+    virtual void free(void *ptr)
+    {
+        hbw_free(ptr);
+    }
+
+    virtual int get_numa_policy()
+    {
+        if (hbw_get_policy() == HBW_POLICY_INTERLEAVE) {
+            return MPOL_INTERLEAVE;
+        } else if (hbw_get_policy() == HBW_POLICY_PREFERRED) {
+            return MPOL_PREFERRED;
+        } else if (hbw_get_policy() == HBW_POLICY_BIND ||
+                   hbw_get_policy() == HBW_POLICY_BIND_ALL) {
+            return MPOL_BIND;
+        }
+        return -1;
+    }
+
+    virtual bool is_high_bandwidth()
+    {
+        return  hbw_check_available() == 0;
+    }
+
+    virtual size_t get_page_size()
+    {
+        if (page_size == HBW_PAGESIZE_2MB) {
+            return 2*MB;
+        }
+        return 4*KB;
+    }
+
+private:
+    hbw_pagesize_t page_size = HBW_PAGESIZE_4KB;
+};
+
diff --git a/test/Makefile.mk b/test/Makefile.mk
new file mode 100644
index 0000000..03a5621
--- /dev/null
+++ b/test/Makefile.mk
@@ -0,0 +1,341 @@
+#
+#  Copyright (C) 2014 - 2020 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+AM_CPPFLAGS += -Itest/gtest_fused -DMEMKIND_DEPRECATED\(x\)=x
+
+check_PROGRAMS += test/all_tests \
+                  test/allocator_perf_tool_tests \
+                  test/autohbw_test_helper \
+                  test/dax_kmem_test \
+                  test/decorator_test \
+                  test/environ_err_dax_kmem_malloc_test \
+                  test/environ_err_dax_kmem_malloc_positive_test \
+                  test/environ_err_hbw_malloc_test \
+                  test/freeing_memory_segfault_test \
+                  test/gb_page_tests_bind_policy \
+                  test/locality_test \
+                  test/memkind_stat_test \
+                  test/performance_test \
+                  test/trace_mechanism_test_helper \
+                  # end
+if HAVE_CXX11
+check_PROGRAMS += test/pmem_test \
+                  test/defrag_reallocate
+endif
+
+TESTS += test/test.sh
+
+EXTRA_DIST += test/autohbw_test.py \
+              test/draw_plots.py \
+              test/gtest_fused/gtest/gtest-all.cc \
+              test/gtest_fused/gtest/gtest.h \
+              test/hbw_detection_test.py \
+              test/dax_kmem_env_var_test.py \
+              test/memkind-afts-ext.ts \
+              test/memkind-afts.ts \
+              test/memkind-perf-ext.ts \
+              test/memkind-perf.ts \
+              test/memkind-pytests.ts \
+              test/memkind-slts.ts \
+              test/python_framework/__init__.py \
+              test/python_framework/cmd_helper.py \
+              test/python_framework/huge_page_organizer.py \
+              test/run_alloc_benchmark.sh \
+              test/trace_mechanism_test.py \
+              # end
+
+test_all_tests_LDADD = libmemkind.la
+test_allocator_perf_tool_tests_LDADD = libmemkind.la
+test_autohbw_test_helper_LDADD = libmemkind.la
+test_dax_kmem_test_LDADD = libmemkind.la
+test_decorator_test_LDADD = libmemkind.la
+test_environ_err_hbw_malloc_test_LDADD = libmemkind.la
+test_environ_err_dax_kmem_malloc_test_LDADD = libmemkind.la
+test_environ_err_dax_kmem_malloc_positive_test_LDADD = libmemkind.la
+test_freeing_memory_segfault_test_LDADD = libmemkind.la
+test_gb_page_tests_bind_policy_LDADD = libmemkind.la
+test_memkind_stat_test_LDADD = libmemkind.la
+test_trace_mechanism_test_helper_LDADD = libmemkind.la
+
+if HAVE_CXX11
+test_pmem_test_SOURCES = $(fused_gtest) test/memkind_pmem_config_tests.cpp test/memkind_pmem_long_time_tests.cpp test/memkind_pmem_tests.cpp
+test_pmem_test_LDADD = libmemkind.la
+test_defrag_reallocate_SOURCES = $(fused_gtest) test/memkind_defrag_reallocate.cpp
+test_defrag_reallocate_LDADD = libmemkind.la
+endif
+
+fused_gtest = test/gtest_fused/gtest/gtest-all.cc \
+              test/main.cpp \
+              # end
+
+test_all_tests_SOURCES = $(fused_gtest) \
+                         test/Allocator.hpp \
+                         test/TestPolicy.hpp \
+                         test/bat_tests.cpp \
+                         test/check.cpp \
+                         test/check.h \
+                         test/common.h \
+                         test/dlopen_test.cpp \
+                         test/error_message_tests.cpp \
+                         test/get_arena_test.cpp \
+                         test/hbw_allocator_tests.cpp \
+                         test/hbw_verify_function_test.cpp \
+                         test/memkind_allocator_tests.cpp \
+                         test/memkind_detect_kind_tests.cpp \
+                         test/memkind_null_kind_test.cpp \
+                         test/memkind_versioning_tests.cpp \
+                         test/multithreaded_tests.cpp \
+                         test/negative_tests.cpp \
+                         test/pmem_allocator_tests.cpp \
+                         test/static_kinds_list.h \
+                         test/static_kinds_tests.cpp \
+                         test/trial_generator.cpp \
+                         test/trial_generator.h \
+                         #end
+
+test_performance_test_SOURCES = $(fused_gtest) test/performance/perf_tests.cpp \
+                                test/performance/perf_tests.hpp \
+                                test/performance/framework.cpp \
+                                test/performance/framework.hpp \
+                                test/performance/operations.hpp \
+                                test/performance/perf_tests.hpp
+
+test_performance_test_LDADD = libmemkind.la
+
+test_locality_test_SOURCES = $(fused_gtest) test/allocator_perf_tool/Allocation_info.cpp test/locality_test.cpp
+test_locality_test_LDADD = libmemkind.la
+
+test_locality_test_CPPFLAGS = $(OPENMP_CFLAGS) -O0 -Wno-error $(AM_CPPFLAGS)
+test_locality_test_CXXFLAGS = $(OPENMP_CFLAGS) -O0 -Wno-error $(AM_CPPFLAGS)
+
+test_autohbw_test_helper_SOURCES = test/autohbw_test_helper.c
+test_decorator_test_SOURCES = $(fused_gtest) test/decorator_test.cpp test/decorator_test.h
+test_dax_kmem_test_SOURCES = $(fused_gtest) test/dax_kmem_nodes.h test/dax_kmem_nodes.cpp test/memkind_dax_kmem_test.cpp
+test_environ_err_hbw_malloc_test_SOURCES = test/environ_err_hbw_malloc_test.cpp
+test_environ_err_dax_kmem_malloc_test_SOURCES = test/environ_err_dax_kmem_malloc_test.cpp
+test_environ_err_dax_kmem_malloc_positive_test_SOURCES = test/environ_err_dax_kmem_malloc_positive_test.cpp
+test_freeing_memory_segfault_test_SOURCES = $(fused_gtest) test/freeing_memory_segfault_test.cpp
+test_gb_page_tests_bind_policy_SOURCES = $(fused_gtest) test/gb_page_tests_bind_policy.cpp test/trial_generator.cpp test/check.cpp
+test_memkind_stat_test_SOURCES = $(fused_gtest) test/memkind_stat_test.cpp
+test_trace_mechanism_test_helper_SOURCES = test/trace_mechanism_test_helper.c
+
+#Tests based on Allocator Perf Tool
+allocator_perf_tool_library_sources = test/allocator_perf_tool/AllocationSizes.hpp \
+                                      test/allocator_perf_tool/Allocation_info.hpp \
+                                      test/allocator_perf_tool/Allocation_info.cpp \
+                                      test/allocator_perf_tool/Allocator.hpp \
+                                      test/allocator_perf_tool/AllocatorFactory.hpp \
+                                      test/allocator_perf_tool/CSVLogger.hpp \
+                                      test/allocator_perf_tool/CommandLine.hpp \
+                                      test/allocator_perf_tool/Configuration.hpp \
+                                      test/allocator_perf_tool/ConsoleLog.hpp \
+                                      test/allocator_perf_tool/FunctionCalls.hpp \
+                                      test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp \
+                                      test/allocator_perf_tool/FunctionCallsPerformanceTask.h \
+                                      test/allocator_perf_tool/GTestAdapter.hpp \
+                                      test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/HugePageOrganizer.hpp \
+                                      test/allocator_perf_tool/HugePageUnmap.hpp \
+                                      test/allocator_perf_tool/Iterator.hpp \
+                                      test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/Numastat.hpp \
+                                      test/allocator_perf_tool/PmemMockup.cpp \
+                                      test/allocator_perf_tool/PmemMockup.hpp \
+                                      test/allocator_perf_tool/Runnable.hpp \
+                                      test/allocator_perf_tool/ScenarioWorkload.cpp \
+                                      test/allocator_perf_tool/ScenarioWorkload.h \
+                                      test/allocator_perf_tool/StandardAllocatorWithTimer.hpp \
+                                      test/allocator_perf_tool/Stats.hpp \
+                                      test/allocator_perf_tool/StressIncreaseToMax.cpp \
+                                      test/allocator_perf_tool/StressIncreaseToMax.h \
+                                      test/allocator_perf_tool/Task.hpp \
+                                      test/allocator_perf_tool/TaskFactory.hpp \
+                                      test/allocator_perf_tool/Tests.hpp \
+                                      test/allocator_perf_tool/Thread.hpp \
+                                      test/allocator_perf_tool/TimerSysTime.hpp \
+                                      test/allocator_perf_tool/VectorIterator.hpp \
+                                      test/allocator_perf_tool/Workload.hpp \
+                                      test/allocator_perf_tool/WrappersMacros.hpp \
+                                      test/memory_manager.h \
+                                      test/proc_stat.h \
+                                      test/random_sizes_allocator.h \
+                                      # end
+
+
+test_allocator_perf_tool_tests_SOURCES = $(allocator_perf_tool_library_sources) \
+                                         $(fused_gtest) \
+                                         test/alloc_performance_tests.cpp \
+                                         test/allocate_to_max_stress_test.cpp \
+                                         test/hbw_allocator_performance_tests.cpp \
+                                         test/heap_manager_init_perf_test.cpp \
+                                         test/huge_page_test.cpp \
+                                         test/memory_footprint_test.cpp \
+                                         test/pmem_alloc_performance_tests.cpp \
+                                         # end
+
+
+test_allocator_perf_tool_tests_CPPFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_allocator_perf_tool_tests_CXXFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_allocator_perf_tool_tests_LDFLAGS = -lpthread -lnuma
+
+NUMAKIND_MAX = 2048
+test_all_tests_CXXFLAGS = $(AM_CXXFLAGS) $(CXXFLAGS) $(OPENMP_CFLAGS) -DNUMAKIND_MAX=$(NUMAKIND_MAX)
+test_all_tests_LDFLAGS = -ldl
+
+#Allocator Perf Tool stand alone app
+check_PROGRAMS += test/perf_tool
+test_perf_tool_LDADD = libmemkind.la
+test_perf_tool_SOURCES = $(allocator_perf_tool_library_sources) \
+                         test/allocator_perf_tool/main.cpp \
+                         # end
+
+
+test_perf_tool_CPPFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_perf_tool_CXXFLAGS = -Itest/allocator_perf_tool/ -O0 -Wno-error $(AM_CPPFLAGS)
+test_perf_tool_LDFLAGS = -lpthread -lnuma
+if HAVE_CXX11
+test_perf_tool_CPPFLAGS += -std=c++11
+test_perf_tool_CXXFLAGS += -std=c++11
+endif
+
+#Alloc benchmark
+check_PROGRAMS += test/alloc_benchmark_glibc \
+                  test/alloc_benchmark_hbw \
+                  test/alloc_benchmark_pmem \
+                  test/alloc_benchmark_tbb \
+                  # end
+
+test_alloc_benchmark_glibc_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall
+test_alloc_benchmark_glibc_LDADD = libmemkind.la
+test_alloc_benchmark_glibc_SOURCES = test/alloc_benchmark.c
+
+test_alloc_benchmark_hbw_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall -DHBWMALLOC
+test_alloc_benchmark_hbw_LDADD = libmemkind.la
+test_alloc_benchmark_hbw_LDFLAGS = -lmemkind
+test_alloc_benchmark_hbw_SOURCES = test/alloc_benchmark.c
+
+test_alloc_benchmark_pmem_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall -DPMEMMALLOC
+test_alloc_benchmark_pmem_LDADD = libmemkind.la
+test_alloc_benchmark_pmem_LDFLAGS = -ldl
+test_alloc_benchmark_pmem_SOURCES = test/alloc_benchmark.c
+
+test_alloc_benchmark_tbb_CFLAGS = -O0 -g $(OPENMP_CFLAGS) -Wall -DTBBMALLOC
+test_alloc_benchmark_tbb_LDADD = libmemkind.la
+test_alloc_benchmark_tbb_LDFLAGS = -ldl
+test_alloc_benchmark_tbb_SOURCES = test/alloc_benchmark.c \
+                                   test/load_tbbmalloc_symbols.c \
+                                   test/tbbmalloc.h \
+                                   # end
+
+# Pmem fragmentation benchmark
+check_PROGRAMS += test/fragmentation_benchmark_pmem
+test_fragmentation_benchmark_pmem_LDADD = libmemkind.la
+test_fragmentation_benchmark_pmem_SOURCES = test/fragmentation_benchmark_pmem.cpp
+test_fragmentation_benchmark_pmem_CXXFLAGS = -O0 -Wall
+if HAVE_CXX11
+test_fragmentation_benchmark_pmem_CXXFLAGS += -std=c++11
+endif
+
+# Examples as tests
+check_PROGRAMS += test/autohbw_candidates \
+                  test/filter_memkind \
+                  test/hello_hbw \
+                  test/hello_memkind \
+                  test/hello_memkind_debug \
+                  test/memkind_get_stat \
+                  test/pmem_alignment \
+                  test/pmem_and_dax_kmem_kind \
+                  test/pmem_and_default_kind \
+                  test/pmem_config \
+                  test/pmem_detect_kind \
+                  test/pmem_free_with_unknown_kind \
+                  test/pmem_kinds \
+                  test/pmem_malloc \
+                  test/pmem_malloc_unlimited \
+                  test/pmem_multithreads \
+                  test/pmem_multithreads_onekind \
+                  test/pmem_usable_size \
+                  # end
+if HAVE_CXX11
+check_PROGRAMS += test/memkind_allocated \
+                  test/memkind_cpp_allocator \
+                  test/pmem_cpp_allocator
+endif
+
+test_autohbw_candidates_LDADD = libmemkind.la
+test_filter_memkind_LDADD = libmemkind.la
+test_hello_hbw_LDADD = libmemkind.la
+test_hello_memkind_LDADD = libmemkind.la
+test_hello_memkind_debug_LDADD = libmemkind.la
+test_memkind_get_stat_LDADD = libmemkind.la
+test_pmem_alignment_LDADD = libmemkind.la
+test_pmem_and_dax_kmem_kind_LDADD = libmemkind.la
+test_pmem_and_default_kind_LDADD = libmemkind.la
+test_pmem_config_LDADD = libmemkind.la
+test_pmem_detect_kind_LDADD = libmemkind.la
+test_pmem_free_with_unknown_kind_LDADD = libmemkind.la
+test_pmem_kinds_LDADD = libmemkind.la
+test_pmem_malloc_LDADD = libmemkind.la
+test_pmem_malloc_unlimited_LDADD = libmemkind.la
+test_pmem_multithreads_LDADD = libmemkind.la
+test_pmem_multithreads_onekind_LDADD = libmemkind.la
+test_pmem_usable_size_LDADD = libmemkind.la
+if HAVE_CXX11
+test_memkind_allocated_LDADD = libmemkind.la
+test_memkind_cpp_allocator_LDADD = libmemkind.la
+test_pmem_cpp_allocator_LDADD = libmemkind.la
+endif
+
+test_autohbw_candidates_SOURCES = examples/autohbw_candidates.c
+test_filter_memkind_SOURCES = examples/filter_example.c
+test_hello_hbw_SOURCES = examples/hello_hbw_example.c
+test_hello_memkind_SOURCES = examples/hello_memkind_example.c
+test_hello_memkind_debug_SOURCES = examples/hello_memkind_example.c examples/memkind_decorator_debug.c
+test_memkind_get_stat_SOURCES = examples/memkind_get_stat.c
+test_pmem_alignment_SOURCES = examples/pmem_alignment.c
+test_pmem_and_dax_kmem_kind_SOURCES = examples/pmem_and_dax_kmem_kind.c
+test_pmem_and_default_kind_SOURCES = examples/pmem_and_default_kind.c
+test_pmem_config_SOURCES = examples/pmem_config.c
+test_pmem_detect_kind_SOURCES = examples/pmem_detect_kind.c
+test_pmem_free_with_unknown_kind_SOURCES = examples/pmem_free_with_unknown_kind.c
+test_pmem_kinds_SOURCES = examples/pmem_kinds.c
+test_pmem_malloc_SOURCES = examples/pmem_malloc.c
+test_pmem_malloc_unlimited_SOURCES = examples/pmem_malloc_unlimited.c
+test_pmem_multithreads_SOURCES = examples/pmem_multithreads.c
+test_pmem_multithreads_onekind_SOURCES = examples/pmem_multithreads_onekind.c
+test_pmem_usable_size_SOURCES = examples/pmem_usable_size.c
+
+test_libautohbw_la_SOURCES = autohbw/autohbw.c
+noinst_LTLIBRARIES += test/libautohbw.la
+if HAVE_CXX11
+test_memkind_allocated_SOURCES = examples/memkind_allocated_example.cpp examples/memkind_allocated.hpp
+test_memkind_cpp_allocator_SOURCES = examples/memkind_cpp_allocator.cpp
+test_pmem_cpp_allocator_SOURCES = examples/pmem_cpp_allocator.cpp
+endif
+
+clean-local: test-clean
+
+test-clean:
+	find test \( -name "*.gcda" -o -name "*.gcno" \) -type f -delete
diff --git a/test/README b/test/README
new file mode 100644
index 0000000..2f7bbcd
--- /dev/null
+++ b/test/README
@@ -0,0 +1,85 @@
+MEMKIND TESTS
+=============
+
+
+DISCLAIMER
+==========
+SEE COPYING FILE FOR LICENSE INFORMATION.
+
+THIS SOFTWARE IS PROVIDED AS A DEVELOPMENT SNAPSHOT TO AID
+COLLABORATION AND WAS NOT ISSUED AS A RELEASED PRODUCT BY INTEL.
+
+LAST UPDATE
+===========
+Agata Wozniak <agata.wozniak@intel.com>
+Wednesday May 11 2016
+
+TESTS SCENARIOS
+===============
+
+memkind-afts
+------------
+This is scenario for basic functional tests. Maximum number of allocations per numa node:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+
+memkind-afts-ext
+----------------
+The same tests as in memkind-afts, however this tests are more ravenous and needs more memory
+available. Special test cases prepared only for this scenario are described as ravenous
+and needs at least:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+memory available per numa node.
+
+memkind-perf
+------------
+This is scenario for performance tests, excluding tests that allocate more than:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+
+memkind-perf-ext
+----------------
+This test scenario extend memkind-perf for tests that allocate more than:
+- 512 x 4KB pages (summary: 2GB)
+- 16 x 2MB pages
+AllocPerformanceTest is a group of performance tests for *alloc() execution. TC properties:
+- 6 kinds (MEMKIND_DEFAULT, MEMKIND_HBW, MEMKIND_INTERLEAVE, MEMKIND_HBW_INTERLEAVE, MEMKIND_HBW_PREFERRED, PMEM kind)
+- 3 calls (malloc, calloc, realloc)
+- 5 allocation sizes (100bytes, 4096bytes, 1000bytes, 1001bytes, 1.5MB - 1572864bytes)
+- single thread and multithreaded (10threads and 72threads)
+All tests record three values:
+- total_time_spend_on_alloc (total time spend on *alloc() call during the test - less is better)
+- alloc_operations_per_thread (number of done allocations per thread - this value is constant)
+- ref_delta_time_percent (percent delta between standard allocator and current tested kind - less is better)
+
+memkind-slts
+------------
+This is scenario for all stress and longevity tests.
+
+BENCHMARKS
+===============
+
+fragmentation_benchmark_pmem
+------------
+The program logging memory fragmentation of specific workload in file
+with date and timestamps ("bench_single_thread_yyyymmdd-hhmm_mem_<pmem_memory_usage_policy>.log").
+
+The program takes, as arguments:
+* pmem_path - path where file-backed memory kind should be created
+* pmem_size - file-backed memory kind size limit, given in megabytes
+* pmem_memory_usage_policy - 0 - MEMKIND_MEM_USAGE_POLICY_DEFAULT, 1 - MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE
+* test_time - time how long program runs, given in seconds
+
+For example:
+
+    ./fragmentation_benchmark_pmem /mnt/pmem 17 0 60
+
+This will create PMEM kind limited with 17 MB in /mnt/pmem/ with default memory usage policy
+and save results in log file after 60 seconds.
+
+MANUAL TESTING
+==============
+In order to run tests on local machine, test.sh should be executed.
+When test.sh is run, all tests are executed as long as -d (skip MCDRAM nodes detection), -m (skip
+tests that require 2MB pages) parameters are not used.
diff --git a/test/TestPolicy.hpp b/test/TestPolicy.hpp
new file mode 100644
index 0000000..91d534f
--- /dev/null
+++ b/test/TestPolicy.hpp
@@ -0,0 +1,177 @@
+/*
+ * Copyright (C) 2017 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include <memkind/internal/memkind_hbw.h>
+
+#include "common.h"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+#include "numa.h"
+
+#include <memory>
+
+namespace TestPolicy
+{
+    typedef std::unique_ptr<struct bitmask, decltype(&numa_free_nodemask)>
+    unique_bitmask_ptr;
+
+    unique_bitmask_ptr make_nodemask_ptr()
+    {
+        return unique_bitmask_ptr(numa_allocate_nodemask(), numa_free_nodemask);
+    }
+
+    unique_bitmask_ptr make_cpumask_ptr()
+    {
+        return unique_bitmask_ptr(numa_allocate_cpumask(), numa_free_nodemask);
+
+    }
+
+    int get_num_of_pages(const size_t size, const size_t page_size)
+    {
+        size_t pages_number = size / page_size;
+        pages_number += size % page_size ? 1 : 0;
+        return pages_number;
+    }
+
+    std::vector<void *> get_address_of_pages(const void *ptr,
+                                             const size_t pages_number, const size_t page_size)
+    {
+        std::vector<void *> address(pages_number);
+        const size_t page_mask = ~(page_size-1);
+        address[0] = (void *)((uintptr_t)ptr &
+                              page_mask); //aligned address of first page
+        for (size_t page_num = 1; page_num < pages_number; page_num++) {
+            address[page_num] = (char *)address[page_num-1] + page_size;
+        }
+        return address;
+    }
+
+    void record_page_association(const void *ptr, const size_t size,
+                                 const size_t page_size)
+    {
+        size_t pages_number = get_num_of_pages(size, page_size);
+        std::vector<void *> address = get_address_of_pages(ptr, pages_number,
+                                                           page_size);
+
+        int max_node_id = numa_max_node();
+        std::vector<int> nodes(pages_number);
+        std::vector<int> pages_on_node(max_node_id+1);
+
+        if (move_pages(0, pages_number, address.data(), NULL, nodes.data(),
+                       MPOL_MF_MOVE)) {
+            fprintf(stderr, "Error: move_pages() returned %s\n", strerror(errno));
+            return;
+        }
+
+        for (size_t i = 0; i < pages_number; i++) {
+            if (nodes[i] < 0) {
+                fprintf(stderr,"Error: status of page %p is %d\n", address[i], nodes[i]);
+                return;
+            } else {
+                pages_on_node[nodes[i]]++;
+            }
+        }
+
+        for (size_t i = 0; i < (size_t)max_node_id + 1; i++) {
+            if (pages_on_node[i] > 0) {
+                char buffer[1024];
+                snprintf(buffer, sizeof(buffer), "Node%zd", i);
+                GTestAdapter::RecordProperty(buffer, pages_on_node[i]);
+            }
+        }
+    }
+
+    void check_numa_nodes(unique_bitmask_ptr &expected_bitmask, int policy,
+                          void *ptr, size_t size)
+    {
+
+        const size_t page_size = sysconf(_SC_PAGESIZE);
+        size_t pages_number = get_num_of_pages(size, page_size);
+        std::vector<void *> address = get_address_of_pages(ptr, pages_number,
+                                                           page_size);
+        unique_bitmask_ptr returned_bitmask = make_nodemask_ptr();
+        int status = -1;
+
+        for (size_t page_num = 0; page_num < address.size(); page_num++) {
+            ASSERT_EQ(0, get_mempolicy(&status, returned_bitmask->maskp,
+                                       returned_bitmask->size, address[page_num], MPOL_F_ADDR));
+            ASSERT_EQ(policy, status);
+            switch(policy) {
+                case MPOL_INTERLEAVE:
+                    EXPECT_TRUE(numa_bitmask_equal(expected_bitmask.get(), returned_bitmask.get()));
+                    break;
+                case MPOL_DEFAULT:
+                    break;
+                case MPOL_BIND:
+                case MPOL_PREFERRED:
+                    for(int i=0; i < numa_num_possible_nodes(); i++) {
+                        if(numa_bitmask_isbitset(returned_bitmask.get(), i)) {
+                            EXPECT_TRUE(numa_bitmask_isbitset(expected_bitmask.get(), i));
+                        }
+                    }
+                    break;
+                default:
+                    assert(!"Unknown policy\n");
+            }
+        }
+    }
+
+    void check_hbw_numa_nodes(int policy, void *ptr, size_t size)
+    {
+        unique_bitmask_ptr expected_bitmask = make_nodemask_ptr();
+
+        memkind_hbw_all_get_mbind_nodemask(NULL, expected_bitmask->maskp,
+                                           expected_bitmask->size);
+        check_numa_nodes(expected_bitmask, policy, ptr, size);
+    }
+
+    void check_all_numa_nodes(int policy, void *ptr, size_t size)
+    {
+        if (policy != MPOL_INTERLEAVE && policy != MPOL_DEFAULT) return;
+
+        unique_bitmask_ptr expected_bitmask = make_nodemask_ptr();
+
+        for(int i=0; i < numa_num_configured_nodes(); i++) {
+            numa_bitmask_setbit(expected_bitmask.get(), i);
+        }
+
+        check_numa_nodes(expected_bitmask, policy, ptr, size);
+    }
+
+    std::set<int> get_regular_numa_nodes(void)
+    {
+        struct bitmask *cpu_mask = numa_allocate_cpumask();
+        std::set<int> regular_nodes;
+
+        const int MAXNODE_ID = numa_max_node();
+        for (int id = 0; id <= MAXNODE_ID; ++id) {
+            numa_node_to_cpus(id, cpu_mask);
+
+            if (numa_bitmask_weight(cpu_mask) != 0) {
+                regular_nodes.insert(id);
+            }
+        }
+        numa_free_cpumask(cpu_mask);
+
+        return regular_nodes;
+    }
+}
diff --git a/test/alloc_benchmark.c b/test/alloc_benchmark.c
new file mode 100644
index 0000000..698974c
--- /dev/null
+++ b/test/alloc_benchmark.c
@@ -0,0 +1,213 @@
+/*
+ * Copyright (C) 2016 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <sys/time.h>
+#include <unistd.h>
+#include <stdint.h>
+#include <limits.h>
+#ifdef _OPENMP
+#include <omp.h>
+#endif
+#if defined(HBWMALLOC)
+#include <hbwmalloc.h>
+#define MALLOC_FN hbw_malloc
+#define FREE_FN hbw_free
+#elif defined (TBBMALLOC)
+#include "tbbmalloc.h"
+void *(*scalable_malloc)(size_t);
+void *(*scalable_realloc)(void *, size_t);
+void *(*scalable_calloc)(size_t, size_t);
+void  (*scalable_free)(void *);
+#define MALLOC_FN scalable_malloc
+#define FREE_FN scalable_free
+#elif defined (PMEMMALLOC)
+#include <sys/stat.h>
+#include "memkind.h"
+#define MALLOC_FN(x) memkind_malloc(pmem_bench_kind, (x))
+#define FREE_FN(x) memkind_free(pmem_bench_kind, (x))
+
+static const size_t PMEM_PART_SIZE = 0;
+static const char *PMEM_DIR = "/tmp/";
+static memkind_t pmem_bench_kind;
+#else
+#define MALLOC_FN malloc
+#define FREE_FN free
+#endif
+
+double ctimer(void);
+void usage(char *name);
+
+int main(int argc, char *argv[])
+{
+#ifdef _OPENMP
+    int nthr = omp_get_max_threads();
+#else
+    int nthr = 1;
+#endif
+    long n, size;
+    size_t alloc_size;
+    unsigned long i;
+    double dt, t_start, t_end, t_malloc, t_free, t_first_malloc, t_first_free,
+           malloc_time = 0.0, free_time = 0.0, first_malloc_time, first_free_time;
+    void *ptr;
+#ifdef TBBMALLOC
+    int ret;
+
+    ret = load_tbbmalloc_symbols();
+    if (ret) {
+        printf("Error: TBB symbols not loaded (ret: %d)\n", ret);
+        return EXIT_FAILURE;
+    }
+#endif
+#ifdef PMEMMALLOC
+    struct stat st;
+
+    /* Handle command line arguments */
+    if (argc == 3 || argc == 4) {
+        n = atol(argv[1]);
+        size = atol(argv[2]);
+
+        if (argc == 4) {
+            if (stat(argv[3], &st) != 0 || !S_ISDIR(st.st_mode)) {
+                usage(argv[0]);
+                return EXIT_FAILURE;
+            } else {
+                PMEM_DIR = argv[3];
+            }
+        }
+    }
+    if ((argc != 3 && argc != 4) || n < 0 || size < 0 || size > (LONG_MAX >> 10)) {
+        usage(argv[0]);
+        return EXIT_FAILURE;
+    }
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_bench_kind);
+    if (err) {
+        printf("Error: memkind_create_pmem failed %d\n", err);
+        return EXIT_FAILURE;
+    }
+#else
+    /* Handle command line arguments */
+    if (argc == 3) {
+        n = atol(argv[1]);
+        size = atol(argv[2]);
+    }
+    if (argc != 3 || n < 0 || size < 0 || size > (LONG_MAX >> 10)) {
+        usage(argv[0]);
+        return EXIT_FAILURE;
+    }
+#endif
+    alloc_size = (size_t) size * 1024;
+
+    /* Get pagesize and compute page_mask */
+    const size_t page_size = sysconf(_SC_PAGESIZE);
+    const size_t page_mask = ~(page_size-1);
+
+    /* Warm up */
+    t_first_malloc = ctimer();
+    ptr = MALLOC_FN(alloc_size);
+    first_malloc_time = ctimer() - t_first_malloc;
+    if (ptr == NULL) {
+        printf("Error: first allocation failed\n");
+        return EXIT_FAILURE;
+    }
+    t_first_free = ctimer();
+    FREE_FN(ptr);
+    first_free_time = ctimer() - t_first_free;
+    ptr = NULL;
+
+    t_start = ctimer();
+    #pragma omp parallel private(i,t_malloc,t_free,ptr) reduction(max:malloc_time,free_time)
+    {
+        malloc_time = 0.0;
+        free_time = 0.0;
+        for (i=0; i<n; i++) {
+            t_malloc = ctimer();
+            ptr = (void *) MALLOC_FN(alloc_size);
+            malloc_time += ctimer() - t_malloc;
+            #pragma omp critical
+            {
+                if (ptr == NULL) {
+                    printf("Error: allocation failed\n");
+                    exit(EXIT_FAILURE);
+                }
+            }
+
+            /* Make sure to touch every page */
+            char *end = ptr + alloc_size;
+            char *aligned_beg = (char *)((uintptr_t)ptr & page_mask);
+            while(aligned_beg < end) {
+                char *temp_ptr = (char *) aligned_beg;
+                char value = temp_ptr[0];
+                temp_ptr[0] = value;
+                aligned_beg += page_size;
+            }
+
+            t_free = ctimer();
+            FREE_FN(ptr);
+            free_time += ctimer() - t_free;
+            ptr = NULL;
+        }
+    }
+    t_end = ctimer();
+    dt = t_end - t_start;
+
+    printf("%d %lu %8.6f %8.6f  %8.6f  %8.6f  %8.6f\n",
+           nthr, size, dt/n, malloc_time/n, free_time/n, first_malloc_time,
+           first_free_time);
+#ifdef PMEMMALLOC
+    err = memkind_destroy_kind(pmem_bench_kind);
+    if (err) {
+        printf("Error: memkind_destroy_kind failed %d\n", err);
+        return EXIT_FAILURE;
+    }
+#endif
+    return EXIT_SUCCESS;
+}
+
+void usage(char *name)
+{
+#ifdef PMEMMALLOC
+    printf("Usage: %s <N> <SIZE> [DIR], where \n"
+           "N is an number of repetitions \n"
+           "SIZE is an allocation size in kbytes\n"
+           "DIR is a custom path for PMEM kind, (default: \"/tmp/\")\n",
+           name);
+#else
+    printf("Usage: %s <N> <SIZE>, where \n"
+           "N is an number of repetitions \n"
+           "SIZE is an allocation size in kbytes\n", name);
+#endif
+}
+
+inline double ctimer()
+{
+    struct timeval tmr;
+    gettimeofday(&tmr, NULL);
+    /* Return time in ms */
+    return (tmr.tv_sec + tmr.tv_usec/1000000.0)*1000;
+}
diff --git a/test/alloc_performance_tests.cpp b/test/alloc_performance_tests.cpp
new file mode 100644
index 0000000..35e9cdc
--- /dev/null
+++ b/test/alloc_performance_tests.cpp
@@ -0,0 +1,1641 @@
+/*
+* Copyright (C) 2016 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/TaskFactory.hpp"
+#include "allocator_perf_tool/Stats.hpp"
+#include "allocator_perf_tool/Thread.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+
+class AllocPerformanceTest: public :: testing::Test
+{
+private:
+    AllocatorFactory allocator_factory;
+
+protected:
+    void SetUp()
+    {
+        allocator_factory.initialize_allocator(AllocatorTypes::STANDARD_ALLOCATOR);
+    }
+
+    void TearDown()
+    {}
+
+    float run(unsigned kind, unsigned call, size_t threads_number,
+              size_t alloc_size, unsigned mem_operations_num)
+    {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+        TypesConf func_calls;
+        TypesConf allocator_types;
+
+        func_calls.enable_type(FunctionCalls::FREE);
+        func_calls.enable_type(call);
+        allocator_types.enable_type(kind);
+
+        TaskConf conf = {
+            mem_operations_num, //number of memory operations
+            {
+                mem_operations_num, //number of memory operations
+                alloc_size, //min. size of single allocation
+                alloc_size //max. size of single allocatioion
+            },
+            func_calls, //enable function calls
+            allocator_types, //enable allocators
+            11, //random seed
+            false, //do not log memory operations and statistics to csv file
+        };
+
+        for (int i=0; i<threads_number; i++) {
+            Task *task = task_factory.create(conf);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats time_stats;
+        for (int i=0; i<tasks.size(); i++) {
+            time_stats += tasks[i]->get_results();
+        }
+
+        return time_stats.stats[kind][call].total_time;
+    }
+
+    void run_test(unsigned kind, unsigned call, size_t threads_number,
+                  size_t alloc_size, unsigned mem_operations_num)
+    {
+        allocator_factory.initialize_allocator(kind);
+        float ref_time = run(AllocatorTypes::STANDARD_ALLOCATOR, call, threads_number,
+                             alloc_size, mem_operations_num);
+        float perf_time = run(kind, call, threads_number, alloc_size,
+                              mem_operations_num);
+        float ref_delta_time_percent = allocator_factory.calc_ref_delta(ref_time,
+                                                                        perf_time);
+
+        GTestAdapter::RecordProperty("total_time_spend_on_alloc", perf_time);
+        GTestAdapter::RecordProperty("alloc_operations_per_thread", mem_operations_num);
+        GTestAdapter::RecordProperty("ref_delta_time_percent", ref_delta_time_percent);
+    }
+
+};
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::MALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::CALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_DEFAULT_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_DEFAULT, FunctionCalls::REALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 1, 1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::MALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 1, 1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_ext_MEMKIND_HBW_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::CALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW, FunctionCalls::REALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::MALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::CALLOC, 72, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_INTERLEAVE_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::MALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::CALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             100, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_INTERLEAVE_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_INTERLEAVE, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::MALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::CALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             4096, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             1000, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             1001, 10000);
+}
+
+TEST_F(AllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_HBW_PREFERRED_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_HBW_PREFERRED, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
diff --git a/test/allocate_to_max_stress_test.cpp b/test/allocate_to_max_stress_test.cpp
new file mode 100644
index 0000000..4895ab7
--- /dev/null
+++ b/test/allocate_to_max_stress_test.cpp
@@ -0,0 +1,176 @@
+/*
+ * Copyright (C) 2015 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <chrono>
+#include "common.h"
+#include "allocator_perf_tool/Configuration.hpp"
+#include "allocator_perf_tool/StressIncreaseToMax.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+//memkind stress and longevity tests using Allocatr Perf Tool.
+class AllocateToMaxStressTests: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+    //Allocates memory up to 'memory_request_limit'.
+    void run(TypesConf kinds, TypesConf func_calls, unsigned operations,
+             size_t size_from, size_t size_to, size_t memory_request_limit,
+             bool touch_memory)
+    {
+        RecordProperty("memory_operations", operations);
+        RecordProperty("size_from", size_from);
+        RecordProperty("size_to", size_to);
+
+        TaskConf task_conf = {
+            .n = operations, //number of memory operations
+            .allocation_sizes_conf = {
+                operations, //number of memory operations
+                size_from, //no random sizes.
+                size_to
+            },
+            .func_calls = func_calls, //enable allocator function call
+            .allocators_types = kinds, //enable allocator
+            .seed = 11, //random seed
+            .touch_memory = touch_memory //enable or disable touching memory
+        };
+
+        std::chrono::time_point<std::chrono::system_clock> start, end;
+        start = std::chrono::system_clock::now();
+
+        //Execute test iterations.
+        std::vector<iteration_result> results =
+            StressIncreaseToMax::execute_test_iterations(task_conf, 120,
+                                                         memory_request_limit);
+
+        end = std::chrono::system_clock::now();
+
+        std::chrono::duration<double> elapsed_time = end - start;
+
+        RecordProperty("elapsed_time", elapsed_time.count());
+
+        //Check finish status.
+        EXPECT_EQ(check_allocation_errors(results, task_conf), 0);
+    }
+
+    //Check true allocation errors over all iterations.
+    //Return iteration number (>0) when error occurs, or zero
+    int check_allocation_errors(std::vector<iteration_result> &results,
+                                const TaskConf &task_conf)
+    {
+        for (size_t i=0; i<results.size(); i++) {
+            //Check if test ends with allocation error.
+            if(results[i].is_allocation_error) {
+                return i+1;
+            }
+        }
+
+        return 0;
+    }
+};
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_MEMKIND_HBW)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW), TypesConf(FunctionCalls::MALLOC),
+        1024, MB, MB, GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_MEMKIND_INTERLEAVE)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_INTERLEAVE),
+        TypesConf(FunctionCalls::MALLOC), 4096, MB, MB, 4*GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_MEMKIND_HBW_PREFERRED)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW_PREFERRED),
+        TypesConf(FunctionCalls::MALLOC), 17408, MB, MB, 17*GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_2MBPages_slts_ext_ALLOCATE_TO_MAX_MEMKIND_HBW_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(2250);
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW_HUGETLB),
+        TypesConf(FunctionCalls::MALLOC), 1024, 4*MB, 4*MB, GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_DIFFERENT_SIZES)
+{
+    run(TypesConf(AllocatorTypes::MEMKIND_HBW), TypesConf(FunctionCalls::MALLOC),
+        2500, 1, 8*MB, GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_AND_FREE_MEMKIND_DEFAULT)
+{
+    TypesConf func_calls;
+    func_calls.enable_type(FunctionCalls::MALLOC);
+    func_calls.enable_type(FunctionCalls::FREE);
+    run(TypesConf(AllocatorTypes::MEMKIND_DEFAULT), func_calls, 2500, 500*MB, 8*GB,
+        16*GB, false);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_AND_FREE_MEMKIND_REGULAR)
+{
+    TypesConf func_calls;
+    func_calls.enable_type(FunctionCalls::MALLOC);
+    func_calls.enable_type(FunctionCalls::FREE);
+    run(TypesConf(AllocatorTypes::MEMKIND_REGULAR), func_calls, 2500, 500*MB, 8*GB,
+        16*GB, false);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ALLOCATE_TO_MAX_DIFFERENT_KINDS)
+{
+    TypesConf kinds;
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_PREFERRED);
+    kinds.enable_type(AllocatorTypes::MEMKIND_DEFAULT);
+    kinds.enable_type(AllocatorTypes::MEMKIND_INTERLEAVE);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_INTERLEAVE);
+    kinds.enable_type(AllocatorTypes::MEMKIND_REGULAR);
+    run(kinds, TypesConf(FunctionCalls::MALLOC), 2048, MB, MB, 2*GB, true);
+}
+
+TEST_F(AllocateToMaxStressTests,
+       test_TC_MEMKIND_slts_ext_ALLOCATE_TO_MAX_DIFFERENT_KINDS_WITH_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(2250);
+    TypesConf kinds;
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_HUGETLB);
+    kinds.enable_type(AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB);
+    run(kinds, TypesConf(FunctionCalls::MALLOC), 2048, MB, MB, 2*GB, true);
+}
diff --git a/test/allocator_perf_tool/AllocationSizes.hpp b/test/allocator_perf_tool/AllocationSizes.hpp
new file mode 100644
index 0000000..59b780d
--- /dev/null
+++ b/test/allocator_perf_tool/AllocationSizes.hpp
@@ -0,0 +1,60 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <vector>
+#include <cstdlib>
+
+#include "VectorIterator.hpp"
+
+class AllocationSizes
+{
+public:
+
+    static VectorIterator<size_t> generate_random_sizes(int sizes_num, size_t from,
+                                                        size_t to, int seed)
+    {
+        srand(seed);
+        std::vector<size_t> sizes;
+        size_t range = to - from;
+
+        if(from == to) {
+            for (int i=0; i<sizes_num; i++)
+                sizes.push_back(from);
+        } else {
+            for (int i=0; i<sizes_num; i++)
+                sizes.push_back(rand() % (range-1)+from);
+        }
+
+        return VectorIterator<size_t>::create(sizes);
+    }
+
+    static VectorIterator<size_t> generate_random_sizes(AllocationSizesConf conf,
+                                                        int seed)
+    {
+        return generate_random_sizes(conf.n, conf.size_from, conf.size_to, seed);
+    }
+
+};
diff --git a/test/allocator_perf_tool/Allocation_info.cpp b/test/allocator_perf_tool/Allocation_info.cpp
new file mode 100644
index 0000000..8f345c5
--- /dev/null
+++ b/test/allocator_perf_tool/Allocation_info.cpp
@@ -0,0 +1,39 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "Allocation_info.hpp"
+#include <numaif.h>
+
+double convert_bytes_to_mb(uint64_t bytes)
+{
+    return bytes / (1024.0 * 1024.0);
+}
+
+int get_numa_node_id(void *ptr)
+{
+    int status = -1;
+
+    get_mempolicy(&status, NULL, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+    return status;
+}
diff --git a/test/allocator_perf_tool/Allocation_info.hpp b/test/allocator_perf_tool/Allocation_info.hpp
new file mode 100644
index 0000000..c33c49a
--- /dev/null
+++ b/test/allocator_perf_tool/Allocation_info.hpp
@@ -0,0 +1,44 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <stdint.h>
+#include <stddef.h>
+
+
+
+//This structure is responsible to store information about single memory operation.
+struct memory_operation {
+    void *ptr;
+    double total_time;
+    size_t size_of_allocation;
+    unsigned allocator_type;
+    unsigned allocation_method;
+    bool is_allocated;
+    int error_code;
+};
+
+double convert_bytes_to_mb(uint64_t bytes);
+
+int get_numa_node_id(void *ptr);
diff --git a/test/allocator_perf_tool/Allocator.hpp b/test/allocator_perf_tool/Allocator.hpp
new file mode 100644
index 0000000..41eaf99
--- /dev/null
+++ b/test/allocator_perf_tool/Allocator.hpp
@@ -0,0 +1,41 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <stdio.h>
+#include "Allocation_info.hpp"
+
+class Allocator
+{
+public:
+
+    virtual memory_operation wrapped_malloc(size_t size) = 0;
+    virtual memory_operation wrapped_calloc(size_t num, size_t size) = 0;
+    virtual memory_operation wrapped_realloc(void *ptr, size_t size) = 0;
+    virtual void wrapped_free(void *ptr) = 0;
+    virtual unsigned type() = 0;
+
+    virtual ~Allocator(void) {}
+};
+
diff --git a/test/allocator_perf_tool/AllocatorFactory.hpp b/test/allocator_perf_tool/AllocatorFactory.hpp
new file mode 100644
index 0000000..4dcd81b
--- /dev/null
+++ b/test/allocator_perf_tool/AllocatorFactory.hpp
@@ -0,0 +1,212 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "memkind.h"
+
+#include "Allocator.hpp"
+#include "StandardAllocatorWithTimer.hpp"
+#include "VectorIterator.hpp"
+#include "Configuration.hpp"
+#include "JemallocAllocatorWithTimer.hpp"
+#include "MemkindAllocatorWithTimer.hpp"
+#include "HBWmallocAllocatorWithTimer.hpp"
+#include "Numastat.hpp"
+#include "PmemMockup.hpp"
+
+#include <vector>
+#include <assert.h>
+#include <string.h>
+#include <map>
+
+
+class AllocatorFactory
+{
+public:
+    //Allocator initialization statistics
+    struct initialization_stat {
+        float total_time; //Total time of initialization.
+        float ref_delta_time; //Delta Time.
+        unsigned allocator_type;
+        std::vector<float> memory_overhead; //Memory overhead per numa node.
+    };
+
+    AllocatorFactory()
+    {
+        memkind_allocators[AllocatorTypes::MEMKIND_DEFAULT] = MemkindAllocatorWithTimer(
+                                                                  MEMKIND_DEFAULT, AllocatorTypes::MEMKIND_DEFAULT);
+        memkind_allocators[AllocatorTypes::MEMKIND_REGULAR] = MemkindAllocatorWithTimer(
+                                                                  MEMKIND_REGULAR, AllocatorTypes::MEMKIND_REGULAR);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW] = MemkindAllocatorWithTimer(
+                                                              MEMKIND_HBW, AllocatorTypes::MEMKIND_HBW);
+        memkind_allocators[AllocatorTypes::MEMKIND_INTERLEAVE] =
+            MemkindAllocatorWithTimer(MEMKIND_INTERLEAVE,
+                                      AllocatorTypes::MEMKIND_INTERLEAVE);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_INTERLEAVE] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_INTERLEAVE,
+                                      AllocatorTypes::MEMKIND_HBW_INTERLEAVE);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_PREFERRED] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_PREFERRED,
+                                      AllocatorTypes::MEMKIND_HBW_PREFERRED);
+        memkind_allocators[AllocatorTypes::MEMKIND_HUGETLB] = MemkindAllocatorWithTimer(
+                                                                  MEMKIND_HUGETLB, AllocatorTypes::MEMKIND_HUGETLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_GBTLB] = MemkindAllocatorWithTimer(
+                                                                MEMKIND_GBTLB, AllocatorTypes::MEMKIND_GBTLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_HUGETLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_HUGETLB,
+                                      AllocatorTypes::MEMKIND_HBW_HUGETLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_PREFERRED_HUGETLB,
+                                      AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_GBTLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_GBTLB, AllocatorTypes::MEMKIND_HBW_GBTLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_HBW_PREFERRED_GBTLB] =
+            MemkindAllocatorWithTimer(MEMKIND_HBW_PREFERRED_GBTLB,
+                                      AllocatorTypes::MEMKIND_HBW_PREFERRED_GBTLB);
+        memkind_allocators[AllocatorTypes::MEMKIND_PMEM] = MemkindAllocatorWithTimer(
+                                                               MEMKIND_PMEM_MOCKUP, AllocatorTypes::MEMKIND_PMEM);
+    }
+
+    //Get existing allocator without creating new.
+    //The owner of existing allocator is AllocatorFactory object.
+    Allocator *get_existing(unsigned type)
+    {
+        switch(type) {
+            case AllocatorTypes::STANDARD_ALLOCATOR:
+                return &standard_allocator;
+
+            case AllocatorTypes::JEMALLOC:
+                return &jemalloc;
+
+            case AllocatorTypes::HBWMALLOC_ALLOCATOR:
+                return &hbwmalloc_allocator;
+
+            default: {
+                if(memkind_allocators.count(type))
+                    return &memkind_allocators[type];
+
+                assert(!"'type' out of range!");
+            }
+        }
+    }
+
+    initialization_stat initialize_allocator(Allocator &allocator)
+    {
+        size_t initial_size = 512;
+        float before_node1 = Numastat::get_total_memory(0);
+        float before_node2 = Numastat::get_total_memory(1);
+        initialization_stat stat = {0};
+
+        //malloc
+        memory_operation malloc_data = allocator.wrapped_malloc(initial_size);
+        stat.total_time += malloc_data.total_time;
+
+        //realloc
+        memory_operation realloc_data = allocator.wrapped_realloc(malloc_data.ptr, 256);
+        allocator.wrapped_free(realloc_data.ptr);
+
+        stat.total_time += realloc_data.total_time;
+
+        //calloc
+        memory_operation calloc_data = allocator.wrapped_calloc(initial_size, 1);
+        allocator.wrapped_free(calloc_data.ptr);
+
+        stat.total_time += calloc_data.total_time;
+
+        stat.allocator_type = allocator.type();
+
+        //calc memory overhead
+        stat.memory_overhead.push_back(Numastat::get_total_memory(0) - before_node1);
+        stat.memory_overhead.push_back(Numastat::get_total_memory(1) - before_node2);
+
+        return stat;
+    }
+
+    initialization_stat initialize_allocator(unsigned type)
+    {
+        return initialize_allocator(*get_existing(type));
+    }
+
+    //Calc percent delta between reference value and current value.
+    float calc_ref_delta(float ref_value, float value)
+    {
+        return ((value / ref_value)  - 1.0) * 100.0;
+    }
+
+    //Test initialization performance over available allocators.
+    //Return statistics.
+    std::vector<initialization_stat> initialization_test()
+    {
+        std::vector<initialization_stat> stats;
+        initialization_stat stat;
+
+        stat = initialize_allocator(standard_allocator);
+        float ref_time = stat.total_time;
+        stats.push_back(stat);
+
+        //Loop over available allocators to call initializer and compute stats.
+        for (unsigned i=1; i<AllocatorTypes::NUM_OF_ALLOCATOR_TYPES; i++) {
+            stat = initialize_allocator(*get_existing(i));
+            stat.ref_delta_time = calc_ref_delta(ref_time, stat.total_time);
+            stats.push_back(stat);
+        }
+
+        return stats;
+    }
+
+    VectorIterator<Allocator *> generate_random_allocator_calls(int num, int seed,
+                                                                TypesConf allocator_calls)
+    {
+        srand(seed);
+        std::vector<Allocator *> allocators_calls;
+
+        for (int i=0; i<num; i++) {
+            int index;
+
+            do {
+                index = (rand() % (AllocatorTypes::NUM_OF_ALLOCATOR_TYPES));
+            } while(!allocator_calls.is_enabled(index));
+
+            allocators_calls.push_back(get_existing(index));
+        }
+
+        return VectorIterator<Allocator *>::create(allocators_calls);
+    }
+
+    //Return kind to the corresponding AllocatorTypes enum specified in argument.
+    memkind_t get_kind_by_type(unsigned type)
+    {
+        if(memkind_allocators.count(type))
+            return memkind_allocators[type].get_kind();
+
+        assert(!"'type' out of range!");
+    }
+
+private:
+    StandardAllocatorWithTimer standard_allocator;
+    JemallocAllocatorWithTimer jemalloc;
+    HBWmallocAllocatorWithTimer hbwmalloc_allocator;
+
+    std::map<unsigned, MemkindAllocatorWithTimer> memkind_allocators;
+};
diff --git a/test/allocator_perf_tool/CSVLogger.hpp b/test/allocator_perf_tool/CSVLogger.hpp
new file mode 100644
index 0000000..fd60731
--- /dev/null
+++ b/test/allocator_perf_tool/CSVLogger.hpp
@@ -0,0 +1,61 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <iostream>
+#include <string>
+#include <sstream>
+#include <fstream>
+#include <ios>
+
+namespace csv
+{
+
+    class Row
+    {
+    public:
+        Row()
+        {
+            row << std::fixed;
+            row.precision(6);
+        }
+
+        template<class T>
+        void append(const T &e)
+        {
+            row << "," << e;
+        }
+
+        std::string export_row() const
+        {
+            std::stringstream ss(row.str());
+            ss << std::endl;
+            return ss.str();
+        }
+
+    private:
+        std::stringstream row;
+    };
+
+}
diff --git a/test/allocator_perf_tool/CommandLine.hpp b/test/allocator_perf_tool/CommandLine.hpp
new file mode 100644
index 0000000..1d04f6b
--- /dev/null
+++ b/test/allocator_perf_tool/CommandLine.hpp
@@ -0,0 +1,85 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <map>
+#include <vector>
+#include <string>
+
+class CommandLine
+{
+public:
+
+    //Parse and write to val when option exist and strtol(...) > 0, otherwise val is not changed.
+    //T should be an integer type.
+    template<class T>
+    void parse_with_strtol(const std::string &option, T &val)
+    {
+        if(args.count(option)) {
+            T tmp = static_cast<T>(strtol(args[option].c_str(), NULL, 10));
+            if(tmp > 0)
+                val = tmp;
+            else
+                printf("Warning! Option '%s' may not be set.\n", option.c_str());
+        } else {
+            //Do not modify val.
+            printf("Warning! Option '%s' is not present.\n", option.c_str());
+        }
+    }
+
+    bool is_option_present(const std::string &option) const
+    {
+        return args.count(option);
+    }
+
+    bool is_option_set(const std::string option, std::string val)
+    {
+        if(is_option_present(option))
+            return (args[option] == val);
+        return false;
+    }
+
+    const std::string &get_option_value(const std::string &option)
+    {
+        return args[option];
+    }
+
+    CommandLine(int argc, char *argv[])
+    {
+        for (int i=0; i<argc; i++) {
+            std::string arg(argv[i]);
+            size_t found = arg.find("=");
+
+            if(found != std::string::npos) {
+                std::string option = arg.substr(0, found);
+                std::string val = arg.substr(found+1, arg.length());
+
+                args[option] = val;
+            }
+        }
+    }
+
+private:
+    std::map<std::string, std::string> args;
+};
diff --git a/test/allocator_perf_tool/Configuration.hpp b/test/allocator_perf_tool/Configuration.hpp
new file mode 100644
index 0000000..c3ce905
--- /dev/null
+++ b/test/allocator_perf_tool/Configuration.hpp
@@ -0,0 +1,155 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <string>
+#include <map>
+#include <assert.h>
+
+//AllocatorTypes class represent allocator types and names related to this types.
+class AllocatorTypes
+{
+public:
+    enum {
+        STANDARD_ALLOCATOR,
+        JEMALLOC,
+        MEMKIND_DEFAULT,
+        MEMKIND_REGULAR,
+        MEMKIND_HBW,
+        MEMKIND_INTERLEAVE,
+        MEMKIND_HBW_INTERLEAVE,
+        MEMKIND_HBW_PREFERRED,
+        MEMKIND_HUGETLB,
+        MEMKIND_GBTLB,
+        MEMKIND_HBW_HUGETLB,
+        MEMKIND_HBW_PREFERRED_HUGETLB,
+        MEMKIND_HBW_GBTLB,
+        MEMKIND_HBW_PREFERRED_GBTLB,
+        HBWMALLOC_ALLOCATOR,
+        MEMKIND_PMEM,
+        NUM_OF_ALLOCATOR_TYPES
+    };
+
+    static const std::string &allocator_name(unsigned type)
+    {
+        static const std::string names[] = {
+            "STANDARD_ALLOCATOR",
+            "JEMALLOC",
+            "MEMKIND_DEFAULT",
+            "MEMKIND_REGULAR",
+            "MEMKIND_HBW",
+            "MEMKIND_INTERLEAVE",
+            "MEMKIND_HBW_INTERLEAVE",
+            "MEMKIND_HBW_PREFERRED",
+            "MEMKIND_HUGETLB",
+            "MEMKIND_GBTLB",
+            "MEMKIND_HBW_HUGETLB",
+            "MEMKIND_HBW_PREFERRED_HUGETLB",
+            "MEMKIND_HBW_GBTLB",
+            "MEMKIND_HBW_PREFERRED_GBTLB",
+            "HBWMALLOC_ALLOCATOR",
+            "MEMKIND_PMEM"
+        };
+
+        if(type >= NUM_OF_ALLOCATOR_TYPES) assert(!"Invalid input argument!");
+
+        return names[type];
+    }
+
+    static unsigned allocator_type(const std::string &name)
+    {
+        for (unsigned i=0; i<NUM_OF_ALLOCATOR_TYPES; i++) {
+            if(allocator_name(i) == name)
+                return i;
+        }
+
+        assert(!"Invalid input argument!");
+    }
+
+    static bool is_valid_memkind(unsigned type)
+    {
+        return (type >= MEMKIND_DEFAULT) && (type < NUM_OF_ALLOCATOR_TYPES);
+    }
+};
+
+//Enable or disable enum values (types).
+class TypesConf
+{
+public:
+    TypesConf() {}
+
+    TypesConf(unsigned type)
+    {
+        enable_type(type);
+    }
+
+    void enable_type(unsigned type)
+    {
+        types[type] = true;
+    }
+
+    void disable_type(unsigned type)
+    {
+        if(types.count(type))
+            types[type] = false;
+    }
+
+    bool is_enabled(unsigned type) const
+    {
+        return (types.count(type) ? types.find(type)->second : false);
+    }
+
+private:
+    std::map<unsigned, bool> types;
+};
+
+//AllocationSizesConf class represents allocation sizes configuration.
+//This data is needed to generate "n" sizes in range from "size_from" to "size_to".
+class AllocationSizesConf
+{
+public:
+    unsigned n;
+    size_t size_from;
+    size_t size_to;
+};
+
+//TaskConf class contain configuration data for task,
+//where:
+// - "n" - number of iterations,
+// - "allocation_sizes_conf" - allocation sizes configuration,
+// - "func_calls" - enabled or disabled function calls,
+// - "allocators_types" - enable allocators,
+// - "seed" - random seed.
+// - "touch_memory" - enable or disable touching memory
+class TaskConf
+{
+public:
+    unsigned n;
+    AllocationSizesConf allocation_sizes_conf;
+    TypesConf func_calls;
+    TypesConf allocators_types;
+    unsigned seed;
+    bool touch_memory;
+    bool is_csv_log_enabled;
+};
diff --git a/test/allocator_perf_tool/ConsoleLog.hpp b/test/allocator_perf_tool/ConsoleLog.hpp
new file mode 100644
index 0000000..396bcca
--- /dev/null
+++ b/test/allocator_perf_tool/ConsoleLog.hpp
@@ -0,0 +1,77 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include "Configuration.hpp"
+#include "Stats.hpp"
+#include "FunctionCalls.hpp"
+
+#include <stdio.h>
+
+class ConsoleLog
+{
+public:
+    static void print_stats(TimeStats &stats, unsigned allocator_type,
+                            unsigned func_calls)
+    {
+        if(stats.stats.count(allocator_type)) {
+            if(stats.stats[allocator_type].count(func_calls)) {
+                MethodStats method_stats = stats.stats[allocator_type][func_calls];
+                printf(" %20s (%u) | %7s | %10f.s | %10f.s  | %zu bytes/%f MB \n",
+                       AllocatorTypes::allocator_name(allocator_type).c_str(),
+                       allocator_type,
+                       FunctionCalls::function_name(func_calls).c_str(),
+                       method_stats.total_time,
+                       method_stats.average_time,
+                       method_stats.allocation_size,
+                       convert_bytes_to_mb(method_stats.allocation_size)
+                      );
+            }
+        }
+    }
+
+    static void print_table(TimeStats &stats)
+    {
+        printf("\n====== Allocators function calls performance =================================================\n");
+        printf(" %20s Id:   Method:    Total time:    Average time:  Allocated memory bytes/MB: \n",
+               "Allocator:");
+        for (unsigned i=0; i<=AllocatorTypes::MEMKIND_HBW_PREFERRED; i++) {
+            for (unsigned func_call=FunctionCalls::FREE+1;
+                 func_call<FunctionCalls::NUM_OF_FUNCTIONS; func_call++) {
+                print_stats(stats, i, func_call);
+            }
+        }
+        printf("==============================================================================================\n");
+    }
+
+    static void print_requested_memory(TimeStats &stats, std::string test_name)
+    {
+        printf("\n====== Requested memory stats for %s =================\n",
+               test_name.c_str());
+        printf("Total requested allocations: %zu bytes/%f MB. \n",stats.get_allocated(),
+               convert_bytes_to_mb(stats.get_allocated()));
+        printf("Total requested deallocations: %zu bytes/%f MB. \n",
+               stats.get_deallocated(), convert_bytes_to_mb(stats.get_deallocated()));
+        printf("=====================================================================\n");
+    }
+};
diff --git a/test/allocator_perf_tool/FunctionCalls.hpp b/test/allocator_perf_tool/FunctionCalls.hpp
new file mode 100644
index 0000000..ef2d7da
--- /dev/null
+++ b/test/allocator_perf_tool/FunctionCalls.hpp
@@ -0,0 +1,75 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "VectorIterator.hpp"
+#include "Configuration.hpp"
+#include <vector>
+#include <cstdlib>
+
+class FunctionCalls
+{
+public:
+    enum {FREE, MALLOC, CALLOC, REALLOC, NUM_OF_FUNCTIONS};
+
+    static const std::string function_name(unsigned type)
+    {
+        static std::string names[] = {"free", "malloc", "calloc", "realloc"};
+
+        if(type >= NUM_OF_FUNCTIONS) assert(!"Invalidate input argument!");
+
+        return names[type];
+    }
+
+    static unsigned function_type(const std::string &name)
+    {
+        for (unsigned i=0; i<NUM_OF_FUNCTIONS; i++) {
+            if(function_name(i) == name)
+                return i;
+        }
+
+        assert(!"Invalid input argument!");
+    }
+
+    static VectorIterator<int> generate_random_allocator_func_calls(int call_num,
+                                                                    int seed, TypesConf func_calls)
+    {
+        std::vector<unsigned> avail_types;
+
+        std::srand(seed);
+        std::vector<int> calls;
+
+        for (int i=0; i<call_num; i++) {
+            int index;
+            do {
+                index = rand() % (NUM_OF_FUNCTIONS);
+            } while(!func_calls.is_enabled(index));
+
+            calls.push_back(index);
+        }
+
+        return VectorIterator<int>::create(calls);
+    }
+
+};
diff --git a/test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp b/test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
new file mode 100644
index 0000000..bdfe5c3
--- /dev/null
+++ b/test/allocator_perf_tool/FunctionCallsPerformanceTask.cpp
@@ -0,0 +1,51 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#include "FunctionCallsPerformanceTask.h"
+
+
+void FunctionCallsPerformanceTask::run()
+{
+    VectorIterator<size_t> allocation_sizes =
+        AllocationSizes::generate_random_sizes(task_conf.allocation_sizes_conf,
+                                               task_conf.seed);
+
+    VectorIterator<int> func_calls =
+        FunctionCalls::generate_random_allocator_func_calls(task_conf.n, task_conf.seed,
+                                                            task_conf.func_calls);
+
+    AllocatorFactory allocator_types;
+    VectorIterator<Allocator *> allocators_calls =
+        allocator_types.generate_random_allocator_calls(task_conf.n, task_conf.seed,
+                                                        task_conf.allocators_types);
+
+    ScenarioWorkload scenario_workload = ScenarioWorkload(
+                                             &allocators_calls,
+                                             &allocation_sizes,
+                                             &func_calls
+                                         );
+
+    while (scenario_workload.run());
+
+    results = scenario_workload.get_allocations_info();
+}
diff --git a/test/allocator_perf_tool/FunctionCallsPerformanceTask.h b/test/allocator_perf_tool/FunctionCallsPerformanceTask.h
new file mode 100644
index 0000000..1add585
--- /dev/null
+++ b/test/allocator_perf_tool/FunctionCallsPerformanceTask.h
@@ -0,0 +1,54 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Task.hpp"
+#include "AllocatorFactory.hpp"
+#include "FunctionCalls.hpp"
+#include "AllocationSizes.hpp"
+#include "VectorIterator.hpp"
+#include "ScenarioWorkload.h"
+#include "Configuration.hpp"
+
+
+class FunctionCallsPerformanceTask :
+    public Task
+{
+public:
+    FunctionCallsPerformanceTask(TaskConf conf)
+    {
+        task_conf = conf;
+    }
+
+    void run();
+
+    std::vector<memory_operation> get_results()
+    {
+        return results;
+    }
+
+private:
+    TaskConf task_conf;
+    std::vector<memory_operation> results;
+};
diff --git a/test/allocator_perf_tool/GTestAdapter.hpp b/test/allocator_perf_tool/GTestAdapter.hpp
new file mode 100644
index 0000000..30dc9ce
--- /dev/null
+++ b/test/allocator_perf_tool/GTestAdapter.hpp
@@ -0,0 +1,49 @@
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <iostream>
+#include <sstream>
+#include <gtest/gtest.h>
+
+/*
+ * The GTestAdapter class is an adapter for GTest framework.
+ * All methods from GTest framework that need to be adapted should be located here.
+ */
+
+class GTestAdapter
+{
+public:
+    template<class T>
+    static void RecordProperty(const std::string &key, const T &value)
+    {
+        std::ostringstream tmp_value;
+        tmp_value << value;
+        testing::Test::RecordProperty(key, tmp_value.str());
+        std::ios::fmtflags flags(std::cout.flags());
+        std::cout << key << ": " << std::fixed  <<  std::setprecision(
+                      5) << value << std::endl;
+        std::cout.flags(flags);
+    }
+};
diff --git a/test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp b/test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
new file mode 100644
index 0000000..4ef7463
--- /dev/null
+++ b/test/allocator_perf_tool/HBWmallocAllocatorWithTimer.hpp
@@ -0,0 +1,74 @@
+/*
+* Copyright (C) 2017 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <hbwmalloc.h>
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+#include <cerrno>
+
+#include <stdlib.h>
+
+
+class HBWmallocAllocatorWithTimer
+    : public Allocator
+{
+public:
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC)
+        data.ptr = hbw_malloc(size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC)
+        data.ptr = hbw_calloc(num, size);
+        END_TEST
+    }
+
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC)
+        data.ptr = hbw_realloc(ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        hbw_free(ptr);
+    }
+
+    unsigned type()
+    {
+        return AllocatorTypes::HBWMALLOC_ALLOCATOR;
+    }
+
+};
diff --git a/test/allocator_perf_tool/HugePageOrganizer.hpp b/test/allocator_perf_tool/HugePageOrganizer.hpp
new file mode 100644
index 0000000..64180c9
--- /dev/null
+++ b/test/allocator_perf_tool/HugePageOrganizer.hpp
@@ -0,0 +1,94 @@
+
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <fstream>
+#include <string.h>
+#include <numa.h>
+#include <vector>
+#include <iostream>
+#include <stdexcept>
+
+/*
+ * HugePageOrganizer sets hugepages per NUMA node and restore initial setup of hugepages.
+ * It writes and reads from the same file, so using HugePageOrganizer with parallel execution may cause undefined behaviour.
+ */
+
+class HugePageOrganizer
+{
+public:
+    HugePageOrganizer(int nr_hugepages_per_node)
+    {
+        for (int node_id = 0; node_id < numa_num_configured_nodes(); node_id++) {
+            initial_nr_hugepages_per_nodes.push_back(get_nr_hugepages(node_id));
+            if (set_nr_hugepages(nr_hugepages_per_node, node_id)) {
+                restore();
+                throw std::runtime_error("Error: Could not set the requested amount of huge pages.");
+            }
+        }
+    }
+
+    ~HugePageOrganizer()
+    {
+        restore();
+    }
+private:
+    std::vector<int> initial_nr_hugepages_per_nodes;
+
+    int get_nr_hugepages(int node_number)
+    {
+        std::string line;
+        char path[128];
+        sprintf(path,
+                "/sys/devices/system/node/node%d/hugepages/hugepages-2048kB/nr_hugepages",
+                node_number);
+        std::ifstream file(path);
+        if (!file) {
+            return -1;
+        }
+        std::getline(file, line);
+        return strtol(line.c_str(), 0, 10);
+    }
+
+    int set_nr_hugepages(int nr_hugepages, int node_number)
+    {
+        char cmd[128];
+        sprintf(cmd, "sudo sh -c \"echo %d > \
+            /sys/devices/system/node/node%d/hugepages/hugepages-2048kB/nr_hugepages\"",
+                nr_hugepages, node_number);
+        if (system(cmd) || (get_nr_hugepages(node_number) != nr_hugepages)) {
+            return -1;
+        }
+        return 0;
+    }
+
+    void restore()
+    {
+        for(size_t node_id = 0; node_id < initial_nr_hugepages_per_nodes.size();
+            node_id++) {
+            set_nr_hugepages(initial_nr_hugepages_per_nodes[node_id], node_id);
+        }
+    }
+};
diff --git a/test/allocator_perf_tool/HugePageUnmap.hpp b/test/allocator_perf_tool/HugePageUnmap.hpp
new file mode 100644
index 0000000..237c357
--- /dev/null
+++ b/test/allocator_perf_tool/HugePageUnmap.hpp
@@ -0,0 +1,82 @@
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <vector>
+#include "Task.hpp"
+
+class HugePageUnmap: public Task
+{
+
+public:
+    HugePageUnmap(int operations, bool touch_memory, size_t alignment_size,
+                  size_t alloc_size, hbw_pagesize_t page_size) :
+        mem_operations_num(operations),
+        touch_memory(touch_memory),
+        alignment_size(alignment_size),
+        alloc_size(alloc_size),
+        page_size(page_size)
+    {}
+
+    ~HugePageUnmap()
+    {
+        for(int i=0; i<mem_operations_num; i++) {
+            hbw_free(results[i].ptr);
+        }
+    };
+
+    void run()
+    {
+        void *ptr = NULL;
+
+        for(int i=0; i<mem_operations_num; i++) {
+            int ret = hbw_posix_memalign_psize(&ptr, alignment_size, alloc_size, page_size);
+
+            ASSERT_EQ(ret, 0);
+            ASSERT_FALSE(ptr == NULL);
+
+            if(touch_memory) {
+                memset(ptr, alignment_size, alignment_size);
+            }
+
+            memory_operation data;
+            data.ptr = ptr;
+            results.push_back(data);
+        }
+    }
+
+    std::vector<memory_operation> get_results()
+    {
+        return results;
+    }
+
+private:
+    int mem_operations_num;
+    std::vector<memory_operation> results;
+
+    bool touch_memory;
+    size_t alignment_size;
+    size_t alloc_size;
+    hbw_pagesize_t page_size;
+};
+
diff --git a/test/allocator_perf_tool/Iterator.hpp b/test/allocator_perf_tool/Iterator.hpp
new file mode 100644
index 0000000..1891ef4
--- /dev/null
+++ b/test/allocator_perf_tool/Iterator.hpp
@@ -0,0 +1,34 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+template<class T>
+class Iterator
+{
+public:
+    virtual bool has_next() const = 0;
+    virtual T next() = 0;
+    virtual size_t size() const = 0;
+};
+
diff --git a/test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp b/test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
new file mode 100644
index 0000000..466af7b
--- /dev/null
+++ b/test/allocator_perf_tool/JemallocAllocatorWithTimer.hpp
@@ -0,0 +1,68 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+
+#include <jemalloc/jemalloc.h>
+
+class JemallocAllocatorWithTimer
+    : public Allocator
+{
+public:
+
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(AllocatorTypes::JEMALLOC, FunctionCalls::MALLOC)
+        data.ptr = malloc(size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(AllocatorTypes::JEMALLOC, FunctionCalls::CALLOC)
+        data.ptr = calloc(num, size);
+        END_TEST
+    }
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(AllocatorTypes::JEMALLOC, FunctionCalls::REALLOC)
+        data.ptr = realloc(ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        free(ptr);
+    }
+
+    unsigned type()
+    {
+        return AllocatorTypes::JEMALLOC;
+    }
+};
diff --git a/test/allocator_perf_tool/Makefile b/test/allocator_perf_tool/Makefile
new file mode 100644
index 0000000..e53248e
--- /dev/null
+++ b/test/allocator_perf_tool/Makefile
@@ -0,0 +1,26 @@
+#
+#  Copyright (C) 2015 - 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+perf_tool: main.cpp ScenarioWorkload.cpp FunctionCallsPerformanceTask.cpp StressIncreaseToMax.cpp Allocation_info.cpp PmemMockup.cpp
+	g++ -o perf_tool $^ -O0 -lmemkind -std=c++11 -lpthread -lnuma -g
diff --git a/test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp b/test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
new file mode 100644
index 0000000..7265b18
--- /dev/null
+++ b/test/allocator_perf_tool/MemkindAllocatorWithTimer.hpp
@@ -0,0 +1,97 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <memkind.h>
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+#include <cerrno>
+
+#include <stdlib.h>
+
+
+class MemkindAllocatorWithTimer
+    : public Allocator
+{
+public:
+    MemkindAllocatorWithTimer() : kind(&MEMKIND_DEFAULT) {}
+
+    MemkindAllocatorWithTimer(memkind_t &memory_kind, unsigned kind_type_id)
+    {
+        kind = &memory_kind;
+        type_id = kind_type_id;
+    }
+    ~MemkindAllocatorWithTimer(void) {}
+
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(type_id, FunctionCalls::MALLOC)
+        data.ptr = memkind_malloc(*kind, size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(type_id, FunctionCalls::CALLOC)
+        data.ptr = memkind_calloc(*kind, num, size);
+        END_TEST
+    }
+
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(type_id, FunctionCalls::REALLOC)
+        data.ptr = memkind_realloc(*kind, ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        memkind_free(*kind, ptr);
+    }
+
+    void change_kind(memkind_t &memory_kind, unsigned kind_type_id)
+    {
+        kind = &memory_kind;
+        type_id = kind_type_id;
+    }
+
+    unsigned type()
+    {
+        return type_id;
+    }
+    memkind_t get_kind()
+    {
+        return *kind;
+    }
+
+private:
+    memkind_t *kind;
+    unsigned type_id;
+
+};
diff --git a/test/allocator_perf_tool/Numastat.hpp b/test/allocator_perf_tool/Numastat.hpp
new file mode 100644
index 0000000..d8ecb19
--- /dev/null
+++ b/test/allocator_perf_tool/Numastat.hpp
@@ -0,0 +1,85 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include <assert.h>
+
+#include <sys/types.h>
+#include <unistd.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <iostream>
+#include <string>
+#include <sstream>
+
+
+class Numastat
+{
+public:
+    //Returns numastat memory usage per node.
+    static float get_total_memory(unsigned node)
+    {
+        static pid_t pid = getpid();
+
+        std::stringstream cmd;
+        cmd << "numastat " << pid;
+
+        FILE *file;
+        char buff[256];
+        float result = -1.0;
+
+        if((file = popen(cmd.str().c_str(), "r"))) {
+            while(fgets(buff, 256, file));
+            pclose(file);
+
+            //We got: "Total                     1181.90            2.00         1183.90".
+            //2.00 is our memory usage from Node 1.
+            std::string last_line(buff);
+
+            size_t dot_pos = 0;
+            //Now we search in: "           2.00         1183.90".
+            for(unsigned i=0; i<=node; i++) {
+                dot_pos = last_line.find(".", dot_pos+1);
+                assert(dot_pos != std::string::npos);
+            }
+
+            //We are at: " 2.00         1183.90".
+            size_t number_begin = last_line.rfind(" ", dot_pos);
+            assert(number_begin != std::string::npos);
+
+            number_begin += 1;
+            buff[dot_pos+3] = '\0';
+
+#if __cplusplus > 201100L
+            result = strtod(&buff[number_begin], NULL);
+#else
+            result = atof(&buff[number_begin]);
+#endif
+        }
+
+        assert(result > -0.01);
+
+        return result;
+    }
+
+};
diff --git a/test/allocator_perf_tool/PmemMockup.cpp b/test/allocator_perf_tool/PmemMockup.cpp
new file mode 100644
index 0000000..619af5d
--- /dev/null
+++ b/test/allocator_perf_tool/PmemMockup.cpp
@@ -0,0 +1,26 @@
+/*
+* Copyright (C) 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#include "PmemMockup.hpp"
+
+struct memkind *MEMKIND_PMEM_MOCKUP;
diff --git a/test/allocator_perf_tool/PmemMockup.hpp b/test/allocator_perf_tool/PmemMockup.hpp
new file mode 100644
index 0000000..a096ebe
--- /dev/null
+++ b/test/allocator_perf_tool/PmemMockup.hpp
@@ -0,0 +1,30 @@
+/*
+* Copyright (C) 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include "memkind.h"
+
+///
+/// \brief Mockup structure for PMEM kind to use allocator_perf_tool engine
+///
+extern memkind_t MEMKIND_PMEM_MOCKUP;
diff --git a/test/allocator_perf_tool/Runnable.hpp b/test/allocator_perf_tool/Runnable.hpp
new file mode 100644
index 0000000..57455cc
--- /dev/null
+++ b/test/allocator_perf_tool/Runnable.hpp
@@ -0,0 +1,31 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+class Runnable
+{
+public:
+    virtual void run() = 0;
+    virtual ~Runnable() {};
+};
diff --git a/test/allocator_perf_tool/ScenarioWorkload.cpp b/test/allocator_perf_tool/ScenarioWorkload.cpp
new file mode 100644
index 0000000..e4ad796
--- /dev/null
+++ b/test/allocator_perf_tool/ScenarioWorkload.cpp
@@ -0,0 +1,113 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#include "ScenarioWorkload.h"
+
+ScenarioWorkload::ScenarioWorkload(VectorIterator<Allocator *> *a,
+                                   VectorIterator<size_t> *as,  VectorIterator<int> *fc)
+{
+    allocators = a;
+    func_calls = fc;
+    alloc_sizes = as;
+}
+
+bool ScenarioWorkload::run()
+{
+    if(func_calls->has_next() && allocators->has_next() &&
+       alloc_sizes->has_next()) {
+        switch(func_calls->next()) {
+            case FunctionCalls::MALLOC: {
+                memory_operation data = allocators->next()->wrapped_malloc(alloc_sizes->next());
+                post_allocation_check(data);
+                break;
+            }
+            case FunctionCalls::CALLOC: {
+                memory_operation data = allocators->next()->wrapped_calloc(1,
+                                                                           alloc_sizes->next());
+                post_allocation_check(data);
+                break;
+            }
+            case FunctionCalls::REALLOC: {
+                //Guarantee the memory for realloc.
+                Allocator *allocator = allocators->next();
+                memory_operation to_realloc = allocator->wrapped_malloc(512);
+
+                memory_operation data = allocator->wrapped_realloc(to_realloc.ptr,
+                                                                   alloc_sizes->next());
+                post_allocation_check(data);
+                break;
+            }
+            case FunctionCalls::FREE: {
+                memory_operation *data = get_allocated_memory();
+
+                if(!allocations.empty() && (data != NULL)) {
+                    allocator_factory.get_existing(data->allocator_type)->wrapped_free(data->ptr);
+                    data->is_allocated = false;
+
+                    memory_operation free_op = *data;
+                    free_op.allocation_method = FunctionCalls::FREE;
+                    allocations.push_back(free_op);
+                }
+
+                break;
+            }
+            default:
+                assert(!"Function call identifier out of range.");
+                break;
+        }
+
+        return true;
+    }
+
+    return false;
+}
+
+ScenarioWorkload::~ScenarioWorkload(void)
+{
+    for (int i=0; i<allocations.size(); i++) {
+        memory_operation data = allocations[i];
+        if(data.is_allocated && (data.allocation_method != FunctionCalls::FREE))
+            allocator_factory.get_existing(data.allocator_type)->wrapped_free(data.ptr);
+    }
+}
+
+memory_operation *ScenarioWorkload::get_allocated_memory()
+{
+    for (int i=allocations.size()-1; i>=0; i--) {
+        memory_operation *data = &allocations[i];
+        if(data->is_allocated)
+            return data;
+    }
+
+    return NULL;
+}
+
+void ScenarioWorkload::post_allocation_check(const memory_operation &data)
+{
+    allocations.push_back(data);
+    if(touch_memory_on_allocation && (data.ptr != NULL) &&
+       (data.error_code != ENOMEM)) {
+        //Write memory to ensure physical allocation.
+        memset(data.ptr, 1, data.size_of_allocation);
+    }
+}
diff --git a/test/allocator_perf_tool/ScenarioWorkload.h b/test/allocator_perf_tool/ScenarioWorkload.h
new file mode 100644
index 0000000..1d80072
--- /dev/null
+++ b/test/allocator_perf_tool/ScenarioWorkload.h
@@ -0,0 +1,68 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include "Workload.hpp"
+#include "FunctionCalls.hpp"
+#include "Allocator.hpp"
+#include "AllocatorFactory.hpp"
+
+#include <string.h>
+
+class ScenarioWorkload :
+    public Workload
+{
+public:
+    ScenarioWorkload(VectorIterator<Allocator *> *a, VectorIterator<size_t> *as,
+                     VectorIterator<int> *fc);
+    ~ScenarioWorkload(void);
+
+    double get_time_costs();
+
+    const std::vector<memory_operation> &get_allocations_info() const
+    {
+        return allocations;
+    }
+
+    bool run();
+
+    memory_operation *get_allocated_memory();
+
+    void enable_touch_memory_on_allocation(bool enable)
+    {
+        touch_memory_on_allocation = enable;
+    }
+
+    void post_allocation_check(const memory_operation &data);
+
+private:
+    AllocatorFactory allocator_factory;
+    std::vector<memory_operation> allocations;
+
+    bool touch_memory_on_allocation;
+
+    VectorIterator<int> *func_calls;
+    VectorIterator<size_t> *alloc_sizes;
+    VectorIterator<Allocator *> *allocators;
+};
+
diff --git a/test/allocator_perf_tool/StandardAllocatorWithTimer.hpp b/test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
new file mode 100644
index 0000000..0c53b95
--- /dev/null
+++ b/test/allocator_perf_tool/StandardAllocatorWithTimer.hpp
@@ -0,0 +1,68 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Allocator.hpp"
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "WrappersMacros.hpp"
+#include "FunctionCalls.hpp"
+
+#include <stdlib.h>
+
+class StandardAllocatorWithTimer
+    : public Allocator
+{
+public:
+
+    memory_operation wrapped_malloc(size_t size)
+    {
+        START_TEST(AllocatorTypes::STANDARD_ALLOCATOR, FunctionCalls::MALLOC)
+        data.ptr = malloc(size);
+        END_TEST
+    }
+
+    memory_operation wrapped_calloc(size_t num, size_t size)
+    {
+        START_TEST(AllocatorTypes::STANDARD_ALLOCATOR, FunctionCalls::CALLOC)
+        data.ptr = calloc(num, size);
+        END_TEST
+    }
+    memory_operation wrapped_realloc(void *ptr, size_t size)
+    {
+        START_TEST(AllocatorTypes::STANDARD_ALLOCATOR, FunctionCalls::REALLOC)
+        data.ptr = realloc(ptr, size);
+        END_TEST
+    }
+
+    void wrapped_free(void *ptr)
+    {
+        free(ptr);
+    }
+
+    unsigned type()
+    {
+        return AllocatorTypes::STANDARD_ALLOCATOR;
+    }
+};
diff --git a/test/allocator_perf_tool/Stats.hpp b/test/allocator_perf_tool/Stats.hpp
new file mode 100644
index 0000000..c474f04
--- /dev/null
+++ b/test/allocator_perf_tool/Stats.hpp
@@ -0,0 +1,101 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <map>
+#include <vector>
+
+#include "Allocation_info.hpp"
+#include "Configuration.hpp"
+#include "FunctionCalls.hpp"
+
+class MethodStats
+{
+public:
+    MethodStats()
+    {
+        total_time = 0.0;
+        average_time = 0;
+        allocation_size = 0;
+        samples_num = 0;
+    }
+
+    double total_time;
+    double average_time;
+    unsigned samples_num;
+    size_t allocation_size;
+};
+
+
+class TimeStats
+{
+public:
+
+    TimeStats()
+    {
+        allocated = 0;
+        deallocated = 0;
+    }
+
+    std::map<unsigned, std::map<unsigned, MethodStats> > stats;
+
+    TimeStats &operator+=(const std::vector<memory_operation> &data)
+    {
+        for (size_t i=0; i<data.size(); i++) {
+            memory_operation tmp = data[i];
+            MethodStats &method_stats = stats[tmp.allocator_type][tmp.allocation_method];
+            method_stats.allocation_size += tmp.size_of_allocation;
+            method_stats.total_time += tmp.total_time;
+            method_stats.samples_num++;
+
+            //Update average.
+            double total_time = method_stats.total_time;
+            double samples_num = method_stats.samples_num;
+            double avg = total_time/samples_num;
+            method_stats.average_time = avg;
+
+            if(tmp.allocation_method != FunctionCalls::FREE) {
+                allocated += tmp.size_of_allocation;
+            } else {
+                deallocated += tmp.size_of_allocation;
+            }
+        }
+
+        return *this;
+    }
+
+    size_t get_allocated() const
+    {
+        return allocated;
+    }
+    size_t get_deallocated() const
+    {
+        return deallocated;
+    }
+
+private:
+    size_t allocated;
+    size_t deallocated;
+};
+
diff --git a/test/allocator_perf_tool/StressIncreaseToMax.cpp b/test/allocator_perf_tool/StressIncreaseToMax.cpp
new file mode 100644
index 0000000..908eef5
--- /dev/null
+++ b/test/allocator_perf_tool/StressIncreaseToMax.cpp
@@ -0,0 +1,138 @@
+/*
+* Copyright (C) 2015 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "StressIncreaseToMax.h"
+
+
+void StressIncreaseToMax::run()
+{
+    //Generate constant allocation sizes.
+    VectorIterator<size_t> allocation_sizes =
+        AllocationSizes::generate_random_sizes(task_conf.allocation_sizes_conf,
+                                               task_conf.seed);
+    //Generate only mallocs.
+    VectorIterator<int> func_calls =
+        FunctionCalls::generate_random_allocator_func_calls(task_conf.n, task_conf.seed,
+                                                            task_conf.func_calls);
+
+    unsigned type;
+    for (type = 0; type < AllocatorTypes::NUM_OF_ALLOCATOR_TYPES; type++) {
+        if(task_conf.allocators_types.is_enabled(type))
+            break; //Assume that there is only one type.
+    }
+
+
+    AllocatorFactory allocator_factory;
+    VectorIterator<Allocator *> allocators_calls =
+        allocator_factory.generate_random_allocator_calls(task_conf.n, task_conf.seed,
+                                                          task_conf.allocators_types);
+
+    ScenarioWorkload scenario_workload(
+        &allocators_calls,
+        &allocation_sizes,
+        &func_calls
+    );
+
+    scenario_workload.enable_touch_memory_on_allocation(task_conf.touch_memory);
+
+    test_status.is_allocation_error = false;
+
+    size_t requested_memory = 0;
+    bool has_reach_memory_request_limit = false;
+
+    while (!has_reach_memory_request_limit &&
+           !test_status.is_allocation_error &&
+           (test_status.has_next_memory_operation = scenario_workload.run())) {
+        memory_operation data = scenario_workload.get_allocations_info().back();
+        test_status.is_allocation_error = (data.error_code == ENOMEM) ||
+                                          (data.ptr == NULL);
+
+        if(data.allocation_method != FunctionCalls::FREE) {
+            requested_memory += data.size_of_allocation;
+            has_reach_memory_request_limit = requested_memory >= req_mem_limit;
+        }
+    }
+
+    if(!(scenario_workload.get_allocations_info().size() < task_conf.n) &&
+       !has_reach_memory_request_limit)
+        printf("\nWARNING: Too few memory operations to reach the limit.\n");
+    if(test_status.is_allocation_error) printf("\nWARNING: Allocation error. \n");
+
+    results = scenario_workload.get_allocations_info();
+}
+
+std::vector<iteration_result> StressIncreaseToMax::execute_test_iterations(
+    const TaskConf &task_conf,
+    unsigned time,
+    size_t requested_memory_limit)
+{
+    TimerSysTime timer;
+    unsigned itr = 0;
+    std::vector<iteration_result> results;
+
+    std::ofstream csv_file;
+
+    csv::Row row;
+    row.append("Iteration");
+    row.append("Allocated memory (MB)");
+    row.append("Elapsed time (seconds)");
+    if(task_conf.is_csv_log_enabled) {
+        csv_file.open("stress_test_increase_to_max.csv");
+        csv_file << row.export_row();
+    }
+    printf("%s", row.export_row().c_str());
+
+    timer.start();
+
+    while (timer.getElapsedTime() < time) {
+        StressIncreaseToMax stress_test(task_conf, requested_memory_limit);
+        stress_test.run();
+        float elapsed_time = timer.getElapsedTime();
+
+        TimeStats stats;
+        stats += stress_test.get_results();
+
+        results.push_back(stress_test.get_test_status());
+
+        //Log every iteration of StressIncreaseToMax test.
+        csv::Row row;
+        row.append(itr);
+        row.append(convert_bytes_to_mb(stats.get_allocated()));
+        row.append(elapsed_time);
+        if(task_conf.is_csv_log_enabled) {
+            csv_file << row.export_row();
+        }
+        printf("%s", row.export_row().c_str());
+        fflush(stdout);
+
+        itr++;
+    }
+
+    printf("\nStress test (StressIncreaseToMax) finish in time %f.\n",
+           timer.getElapsedTime());
+
+    csv_file.close();
+
+    return results;
+}
diff --git a/test/allocator_perf_tool/StressIncreaseToMax.h b/test/allocator_perf_tool/StressIncreaseToMax.h
new file mode 100644
index 0000000..867d8fa
--- /dev/null
+++ b/test/allocator_perf_tool/StressIncreaseToMax.h
@@ -0,0 +1,78 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Configuration.hpp"
+#include "ScenarioWorkload.h"
+#include "FunctionCalls.hpp"
+#include "AllocationSizes.hpp"
+#include "VectorIterator.hpp"
+#include "AllocatorFactory.hpp"
+#include "Task.hpp"
+#include "Numastat.hpp"
+#include "CSVLogger.hpp"
+#include "TimerSysTime.hpp"
+#include "Stats.hpp"
+
+#include <vector>
+
+struct iteration_result {
+    bool has_next_memory_operation;
+    bool is_allocation_error;
+};
+
+class StressIncreaseToMax
+    : public Task
+{
+public:
+    StressIncreaseToMax(const TaskConf &conf, size_t requested_memory_limit)
+        : task_conf(conf),
+          req_mem_limit(requested_memory_limit)
+    {}
+
+    void run();
+
+    //Return memory operations from the last run.
+    std::vector<memory_operation> get_results()
+    {
+        return results;
+    }
+    iteration_result get_test_status()
+    {
+        return test_status;
+    }
+
+    static std::vector<iteration_result> execute_test_iterations(
+        const TaskConf &task_conf, unsigned time, size_t requested_memory_limit);
+
+private:
+    size_t req_mem_limit;
+    ScenarioWorkload *scenario_workload;
+    std::vector<memory_operation> results;
+    const TaskConf &task_conf;
+
+    //Test status
+    iteration_result test_status;
+};
+
diff --git a/test/allocator_perf_tool/Task.hpp b/test/allocator_perf_tool/Task.hpp
new file mode 100644
index 0000000..0631669
--- /dev/null
+++ b/test/allocator_perf_tool/Task.hpp
@@ -0,0 +1,39 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <vector>
+
+#include "Runnable.hpp"
+#include "Allocation_info.hpp"
+
+class Task
+    : public Runnable
+{
+public:
+    virtual ~Task() {}
+
+    virtual std::vector<memory_operation> get_results() = 0;
+};
+
diff --git a/test/allocator_perf_tool/TaskFactory.hpp b/test/allocator_perf_tool/TaskFactory.hpp
new file mode 100644
index 0000000..ea9ff28
--- /dev/null
+++ b/test/allocator_perf_tool/TaskFactory.hpp
@@ -0,0 +1,58 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <cstdlib>
+#include <vector>
+#include <assert.h>
+
+#include "Task.hpp"
+#include "FunctionCallsPerformanceTask.h"
+#include "Configuration.hpp"
+
+
+class TaskFactory
+{
+public:
+
+    Task *create(TaskConf conf)
+    {
+        Task *task = NULL;
+        task = new FunctionCallsPerformanceTask(conf);
+
+        tasks.push_back(task);
+
+        return task;
+    }
+
+    ~TaskFactory()
+    {
+        for (int i=0; i<tasks.size(); i++) {
+            delete tasks[i];
+        }
+    }
+
+private:
+    std::vector<Task *> tasks;
+};
diff --git a/test/allocator_perf_tool/Tests.hpp b/test/allocator_perf_tool/Tests.hpp
new file mode 100644
index 0000000..d7f20d4
--- /dev/null
+++ b/test/allocator_perf_tool/Tests.hpp
@@ -0,0 +1,191 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+/*
+This file contain self-tests for Allocator Perf Tool.
+*/
+#pragma once
+
+#include <unistd.h>
+
+#include "Configuration.hpp"
+#include "AllocatorFactory.hpp"
+#include "AllocationSizes.hpp"
+#include "VectorIterator.hpp"
+#include "StandardAllocatorWithTimer.hpp"
+#include "ScenarioWorkload.h"
+#include "TaskFactory.hpp"
+#include "Task.hpp"
+
+//Basic test for allocators. Allocate 32 bytes by malloc, then deallocate it by free.
+void test_allocator(Allocator &allocator, const char *allocator_name)
+{
+    printf("\n============= Allocator (%s) test ============= \n",allocator_name);
+
+    memory_operation data = allocator.wrapped_malloc(32);
+    printf("malloc: %p \n",data.ptr);
+
+    assert(data.ptr!=NULL);
+
+    allocator.wrapped_free(data.ptr);
+    printf("free: %p \n",data.ptr);
+
+    printf("\n==================================================== \n");
+}
+
+//Test for the workload classes. This test count and validate the number or runs.
+void test_workload(Workload &workload, const int N, const char *workload_name)
+{
+    printf("\n============= Work load (%s) test ============= \n",workload_name);
+
+    assert(workload.run()); // Do "single" memory operation (allocate or deallocate).
+
+    int i;
+    for (i=1; workload.run(); i++); // Do the rest of operations.
+    assert(i == N);
+
+    printf("\n==================================================== \n");
+}
+
+//Simulate time consuming memory operation.
+memory_operation time_consuming_memory_operation(unsigned int seconds)
+{
+    size_t size;
+
+    START_TEST(0,0)
+    sleep(seconds);
+    END_TEST
+}
+
+//This test check if timer has measured time.
+void test_timer()
+{
+    float total_time = time_consuming_memory_operation(2).total_time;
+
+    bool test_res = (total_time >=  2.0) && (total_time < 2.2);
+    if(!test_res)
+        printf("test_timer(): unexpected timing %f", total_time);
+    assert(test_res);
+}
+
+//Test for iterators. Check the number of iterations, such as:
+//N == number of iterations.
+template<class T>
+void test_iterator(Iterator<T> &it, const int N)
+{
+    printf("\n================= Iteartor test ==================== \n");
+
+    assert(it.size() == N);
+    int i;
+    for (i=0; it.has_next(); i++) {
+        it.next();
+    }
+    printf("iterations=%d/%d\n",i,N);
+    assert(i == N);
+
+    printf("\n==================================================== \n");
+}
+
+//This test validate range of values generated by random generators.
+template<class T, class C>
+void test_iterator_values(VectorIterator<T> &it, const C from, const C to)
+{
+    for (int i=0; it.has_next(); i++) {
+        T val = it.next();
+
+        if(val < from) std::cout << "ivalid value: actual=" << val << ", expected= >="
+                                     << from << std::endl;
+
+        if(val > to) std::cout << "ivalid value: actual=" << val << ", expected= <=" <<
+                                   to << std::endl;
+    }
+}
+
+//Test time counting instrumentation in StandardAllocatorWithTimer.
+//The instrumentation is made with START_TEST and END_TEST macros from WrapperMacros.h.
+void test_allocator_with_timer(int N, int seed)
+{
+    printf("\n============= Allocator with timer test ============= \n");
+
+    StandardAllocatorWithTimer allocator;
+    VectorIterator<size_t> allocation_sizes =
+        AllocationSizes::generate_random_sizes(N, 32, 2048, seed);
+    memory_operation data;
+
+    double elaspsed_time = 0;
+    for (int i=0; i<N; i++) {
+        data = allocator.wrapped_malloc(allocation_sizes.next());
+        elaspsed_time += data.total_time;
+        allocator.wrapped_free(data.ptr);
+    }
+
+    printf("%d allocations and frees done in time: %f \n", N, elaspsed_time);
+
+    printf("\n==================================================== \n");
+}
+
+//Test behavior of TypesConfiguration class, by enabling and disabling types.
+void test_types_conf()
+{
+    TypesConf types;
+
+    for (unsigned i=0; i<FunctionCalls::NUM_OF_FUNCTIONS; i++) {
+        types.enable_type(i);
+        assert(types.is_enabled(i));
+        types.disable_type(i);
+        assert(!types.is_enabled(i));
+    }
+}
+
+void execute_self_tests()
+{
+    const int N = 10000;
+    const size_t size_from = 32, size_to = 2048;
+    const unsigned seed = 11;
+
+    test_types_conf();
+
+    {
+        VectorIterator<size_t> it = AllocationSizes::generate_random_sizes(N, size_from,
+                                                                           size_to,seed);
+        test_iterator_values(it, size_from, size_to);
+    }
+
+    {
+        TypesConf enable_func_calls;
+        enable_func_calls.enable_type(FunctionCalls::MALLOC);
+
+        VectorIterator<int> it = FunctionCalls::generate_random_allocator_func_calls(N,
+                                                                                     seed, enable_func_calls);
+        test_iterator(it, N);
+    }
+
+//Timer implementation __cplusplus < 201100L is based on CPU clocks counting and it will fail on this test.
+#if __cplusplus > 201100L
+    test_timer();
+#endif
+
+    printf("Test completed! (press ENTER to continue)\n");
+}
+
diff --git a/test/allocator_perf_tool/Thread.hpp b/test/allocator_perf_tool/Thread.hpp
new file mode 100644
index 0000000..3b83ccc
--- /dev/null
+++ b/test/allocator_perf_tool/Thread.hpp
@@ -0,0 +1,99 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+#include <pthread.h>
+#include <assert.h>
+
+#include "Runnable.hpp"
+
+class Thread
+{
+public:
+    Thread(Runnable *runnable) : runnable_task(runnable) {}
+
+    void start()
+    {
+        int err = pthread_create(&thread_handle, NULL, execute_thread,
+                                 static_cast<void *>(runnable_task));
+        assert(!err);
+    };
+
+    void wait()
+    {
+        pthread_join(thread_handle, NULL);
+    };
+
+    Runnable *get_runnable_task()
+    {
+        return runnable_task;
+    }
+
+private:
+    static void *execute_thread(void *ptr)
+    {
+        Runnable *runnable = static_cast<Runnable *>(ptr);
+        assert(runnable);
+        runnable->run();
+        pthread_exit(NULL);
+    }
+
+    pthread_t thread_handle;
+    Runnable *runnable_task;
+};
+
+
+class ThreadsManager
+{
+public:
+    ThreadsManager(std::vector<Thread *> &threads_vec) : threads(threads_vec) {}
+    ~ThreadsManager()
+    {
+        release();
+    }
+
+    void start()
+    {
+        for (int i=0; i<threads.size(); i++) {
+            threads[i]->start();
+        }
+    }
+
+    void barrier()
+    {
+        for (int i=0; i<threads.size(); i++) {
+            threads[i]->wait();
+        }
+    }
+
+    void release()
+    {
+        for (int i=0; i<threads.size(); i++) {
+            delete threads[i];
+        }
+        threads.clear();
+    }
+
+private:
+    std::vector<Thread *> &threads;
+};
diff --git a/test/allocator_perf_tool/TimerSysTime.hpp b/test/allocator_perf_tool/TimerSysTime.hpp
new file mode 100644
index 0000000..55e69dd
--- /dev/null
+++ b/test/allocator_perf_tool/TimerSysTime.hpp
@@ -0,0 +1,48 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <sys/time.h>
+
+class TimerSysTime
+{
+public:
+    void start()
+    {
+        gettimeofday(&last, 0);
+    }
+
+    double getElapsedTime() const
+    {
+        struct timeval now;
+        gettimeofday(&now, 0);
+        double time_delta_sec = ((double)now.tv_sec - (double)last.tv_sec);
+        double time_delta_usec = ((double)now.tv_usec - (double)last.tv_usec) /
+                                 1000000.0;
+        return time_delta_sec + time_delta_usec;
+    }
+
+private:
+    struct timeval last;
+};
diff --git a/test/allocator_perf_tool/VectorIterator.hpp b/test/allocator_perf_tool/VectorIterator.hpp
new file mode 100644
index 0000000..3bb1863
--- /dev/null
+++ b/test/allocator_perf_tool/VectorIterator.hpp
@@ -0,0 +1,69 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Iterator.hpp"
+
+#include <vector>
+
+template<class T>
+class VectorIterator
+    : public Iterator<T>
+{
+public:
+
+    static VectorIterator create(int init_size)
+    {
+        return VectorIterator(init_size);
+    }
+    static VectorIterator create(std::vector<T> v)
+    {
+        return VectorIterator(v);
+    }
+
+    bool has_next() const
+    {
+        return !vec.empty();
+    }
+
+    T next()
+    {
+        T val = vec.back();
+        vec.pop_back();
+        return val;
+    }
+
+    size_t size() const
+    {
+        return vec.size();
+    }
+
+private:
+    VectorIterator(std::vector<T> v)
+    {
+        vec = v;
+    }
+
+    std::vector<T> vec;
+};
diff --git a/test/allocator_perf_tool/Workload.hpp b/test/allocator_perf_tool/Workload.hpp
new file mode 100644
index 0000000..96de5a3
--- /dev/null
+++ b/test/allocator_perf_tool/Workload.hpp
@@ -0,0 +1,37 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include "Allocator.hpp"
+
+class Workload
+{
+public:
+    virtual bool run() = 0;
+    virtual ~Workload(void) {}
+
+protected:
+    Allocator *allocator;
+};
+
diff --git a/test/allocator_perf_tool/WrappersMacros.hpp b/test/allocator_perf_tool/WrappersMacros.hpp
new file mode 100644
index 0000000..7069f59
--- /dev/null
+++ b/test/allocator_perf_tool/WrappersMacros.hpp
@@ -0,0 +1,43 @@
+/*
+* Copyright (C) 2015 - 2017 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+#pragma once
+
+#include <ctime>
+#include <sys/time.h>
+
+#include <chrono>
+#define START_TEST(ALLOCATOR, METHOD) \
+	memory_operation data; \
+	data.allocator_type = ALLOCATOR; \
+	data.allocation_method = METHOD;\
+	data.error_code = 0; \
+	std::chrono::system_clock::time_point last = std::chrono::high_resolution_clock::now();
+#define END_TEST \
+	std::chrono::system_clock::time_point now = std::chrono::high_resolution_clock::now(); \
+	std::chrono::duration<double,std::milli> elapsedTime(now - last); \
+	data.total_time = elapsedTime.count() / 1000.0; \
+	data.size_of_allocation = size; \
+	data.error_code = errno; \
+	data.is_allocated = true; \
+	return data;
diff --git a/test/allocator_perf_tool/main.cpp b/test/allocator_perf_tool/main.cpp
new file mode 100644
index 0000000..3aa3057
--- /dev/null
+++ b/test/allocator_perf_tool/main.cpp
@@ -0,0 +1,231 @@
+/*
+* Copyright (C) 2015 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <stdio.h>
+#include <assert.h>
+#include <iostream>
+#include <vector>
+
+#include "Configuration.hpp"
+#include "AllocatorFactory.hpp"
+#include "TaskFactory.hpp"
+#include "Task.hpp"
+#include "ConsoleLog.hpp"
+#include "Stats.hpp"
+#include "Thread.hpp"
+#include "Tests.hpp"
+#include "CommandLine.hpp"
+#include "FunctionCallsPerformanceTask.h"
+#include "StressIncreaseToMax.h"
+
+/*
+Command line description.
+Syntax:
+	key=value
+Options:
+	- 'test' - specify the test case. This option can be used with the following values: 'calls', 'all' or 'self',
+	where:
+		'calls' - function calls performance test,
+		'all' - execute both above ('footprint' and 'calls') tests,
+		'self' - execute self tests
+		's1' - stress tests
+		(perform allocations until the maximum amount of allocated memory has been reached, than frees allocated memory.
+		If the time interval has not been exceed, than repeat the test),
+	- 'operations' - the number of memory operations per thread
+	- 'size_from' - lower bound for the random sizes of allocation
+	- 'size_to' - upper bound for the random sizes of allocation
+	- 'seed' - random seed
+	- 'threads_num' - the number of threads per test case
+	- 'time' - minimum execution time interval
+	- 'kind' - the kind to test
+	- 'csv_log' - if 'true' then log to csv file memory operations and statistics
+	- 'call' specify the allocation function call. This option can be used with the following values: 'malloc' (default), 'calloc', 'realloc',
+	- 'requested_memory_limit' test stops when the requested memory limit has been reached
+* - maximum of available memory in OS, or maximum memory based 'operations' parameter
+Example:
+1. Performance test:
+./perf_tool test=all operations=1000 size_from=32 size_to=20480 seed=11 threads_num=200
+2. Stress test
+./perf_tool test=s1 time=120 kind=MEMKIND_HBW size_from=1048576 csv_log=true requested_memory_limit=1048576
+*/
+
+int main(int argc, char *argv[])
+{
+    unsigned mem_operations_num = 1000;
+    size_t size_from = 32, size_to = 2048*1024;
+    unsigned seed = 11;
+    //should be at least one
+    size_t threads_number = 10;
+
+    CommandLine cmd_line(argc, argv);
+
+    if((argc >= 1) && cmd_line.is_option_set("test", "self")) {
+        execute_self_tests();
+        getchar();
+    }
+
+    cmd_line.parse_with_strtol("operations", mem_operations_num);
+    cmd_line.parse_with_strtol("size_from", size_from);
+    cmd_line.parse_with_strtol("size_to", size_to);
+    cmd_line.parse_with_strtol("seed", seed);
+    cmd_line.parse_with_strtol("threads_num", threads_number);
+
+    bool is_csv_log_enabled = cmd_line.is_option_set("csv_log", "true");
+
+    //Heap Manager initialization
+    std::vector<AllocatorFactory::initialization_stat> stats =
+        AllocatorFactory().initialization_test();
+
+    if(!cmd_line.is_option_set("print_init_stats", "false")) {
+        printf("\nInitialization overhead:\n");
+        for (int i=0; i<stats.size(); i++) {
+            AllocatorFactory::initialization_stat stat = stats[i];
+            printf("%32s : time=%7.7f.s, ref_delta_time=%15f, node0=%10fMB, node1=%7.7fMB\n",
+                   AllocatorTypes::allocator_name(stat.allocator_type).c_str(),
+                   stat.total_time,
+                   stat.ref_delta_time,
+                   stat.memory_overhead[0],
+                   stat.memory_overhead[1]);
+        }
+    }
+
+    //Stress test by repeatedly increasing memory (to maximum), until given time interval has been exceed.
+    if(cmd_line.is_option_set("test", "s1")) {
+        printf("Stress test (StressIncreaseToMax) start. \n");
+
+        if(!cmd_line.is_option_present("operations"))
+            mem_operations_num = 1000000;
+
+        unsigned time = 120; //Default time interval.
+        cmd_line.parse_with_strtol("time", time);
+
+        size_t requested_memory_limit = 1024*1024;
+        cmd_line.parse_with_strtol("requested_memory_limit", requested_memory_limit);
+
+        unsigned allocator = AllocatorTypes::MEMKIND_HBW;
+        if(cmd_line.is_option_present("kind")) {
+            //Enable memkind allocator and specify kind.
+            allocator = AllocatorTypes::allocator_type(cmd_line.get_option_value("kind"));
+        }
+        TypesConf allocator_types;
+        allocator_types.enable_type(allocator);
+
+        TypesConf enable_func_calls;
+        enable_func_calls.enable_type(FunctionCalls::MALLOC);
+
+        TaskConf task_conf = {
+            mem_operations_num,
+            {
+                mem_operations_num,
+                size_from, //No random sizes.
+                size_from
+            },
+            enable_func_calls,
+            allocator_types,
+            11,
+            is_csv_log_enabled,
+        };
+
+        StressIncreaseToMax::execute_test_iterations(task_conf, time,
+                                                     requested_memory_limit);
+        return 0;
+    }
+
+    printf("\nTest configuration: \n");
+    printf("\t memory operations per thread = %u \n", mem_operations_num);
+    printf("\t seed = %d\n", seed);
+    printf("\t number of threads = %zu\n", threads_number);
+    printf("\t size from-to = %zu-%zu\n\n", size_from, size_to);
+
+    assert(size_from <= size_to);
+
+
+    TypesConf func_calls;
+    func_calls.enable_type(FunctionCalls::FREE);
+
+    if(cmd_line.is_option_present("call")) {
+        //Enable heap manager function call.
+        func_calls.enable_type(FunctionCalls::function_type(
+                                   cmd_line.get_option_value("call")));
+    } else {
+        func_calls.enable_type(FunctionCalls::MALLOC);
+    }
+
+    TypesConf allocator_types;
+    if(cmd_line.is_option_present("allocator")) {
+        allocator_types.enable_type(AllocatorTypes::allocator_type(
+                                        cmd_line.get_option_value("allocator")));
+    } else {
+        for(unsigned i = 0; i <= AllocatorTypes::MEMKIND_HBW_PREFERRED; i++) {
+            allocator_types.enable_type(i);
+        }
+    }
+
+    TaskConf conf = {
+        mem_operations_num, //number memory operations
+        {
+            mem_operations_num, //number of memory operations
+            size_from, //min. size of single allocation
+            size_to //max. size of single allocatioion
+        },
+        func_calls, //enable function calls
+        allocator_types, //enable allocators
+        seed, //random seed
+        is_csv_log_enabled,
+    };
+
+    //Function calls test
+    if(cmd_line.is_option_set("test", "calls") ||
+       cmd_line.is_option_set("test", "all")) {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+
+        for (int i=0; i<threads_number; i++) {
+            FunctionCallsPerformanceTask *task =
+                static_cast<FunctionCallsPerformanceTask *>(
+                    task_factory.create(conf)
+                );
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats stats;
+        for (int i=0; i<tasks.size(); i++) {
+            stats += tasks[i]->get_results();
+        }
+
+        ConsoleLog::print_table(stats);
+        ConsoleLog::print_requested_memory(stats, "func. calls test");
+
+        threads_manager.release();
+    }
+
+    return 0;
+}
diff --git a/test/autohbw_test.py b/test/autohbw_test.py
new file mode 100755
index 0000000..661ec62
--- /dev/null
+++ b/test/autohbw_test.py
@@ -0,0 +1,80 @@
+#
+#  Copyright (C) 2016 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+import pytest
+import os
+from python_framework import CMD_helper
+
+def _get_lib_path():
+    for p in ["/usr/lib64", "/usr/lib"]:
+        if os.path.isdir(p):
+            return p
+    raise Exception("Cannot find library path in OS")
+
+class Test_autohbw(object):
+    binary = "../autohbw_test_helper"
+    fail_msg = "Test failed with:\n {0}"
+    test_prefix = "AUTO_HBW_LOG=2 LD_PRELOAD=%s/libautohbw.so.0 " % _get_lib_path()
+    memkind_malloc_log = "In my memkind malloc"
+    memkind_calloc_log = "In my memkind calloc"
+    memkind_realloc_log = "In my memkind realloc"
+    memkind_posix_memalign_log = "In my memkind align"
+    memkind_free_log = "In my memkind free"
+    cmd_helper = CMD_helper()
+
+    def test_TC_MEMKIND_autohbw_malloc_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding malloc() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " malloc"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_malloc_log in output, self.fail_msg.format("\nError: malloc was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("\nError: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_autohbw_calloc_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding calloc() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " calloc"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_calloc_log in output, self.fail_msg.format("\nError: calloc was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("Error: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_autohbw_realloc_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding realloc() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " realloc"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_realloc_log in output, self.fail_msg.format("\nError: realloc was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("\nError: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_autohbw_posix_memalign_and_free(self):
+        """ This test executes ./autohbw_test_helper with LD_PRELOAD that is overriding posix_memalign() and free() to equivalent autohbw functions"""
+        command = self.test_prefix + self.cmd_helper.get_command_path(self.binary) + " posix_memalign"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: autohbw_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert self.memkind_posix_memalign_log in output, self.fail_msg.format("\nError: posix_memalign was not overridden by autohbw equivalent \noutput: {0}").format(output)
+        assert self.memkind_free_log in output, self.fail_msg.format("\nError: free was not overridden by autohbw equivalent \noutput: {0}").format(output)
diff --git a/test/autohbw_test_helper.c b/test/autohbw_test_helper.c
new file mode 100644
index 0000000..c04d9ef
--- /dev/null
+++ b/test/autohbw_test_helper.c
@@ -0,0 +1,80 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+int main(int argc, char *argv[])
+{
+    int err = 0;
+    const size_t size = 1024 * 1024;
+    void *buf = NULL;
+
+    //It is expected that "malloc", "calloc", "realloc" or "posix_memalign" argument is passed
+    if (argc != 2) {
+        printf("Error: Wrong number of parameters\n");
+        err = -1;
+        return err;
+    }
+
+    if (strcmp(argv[1], "malloc") == 0) {
+        buf = malloc(size);
+        if (buf == NULL) {
+            printf("Error: malloc returned NULL\n");
+            err = -1;
+        }
+        free(buf);
+    } else if (strcmp(argv[1], "calloc") == 0) {
+        buf = calloc(size, 1);
+        if (buf == NULL) {
+            printf("Error: calloc returned NULL\n");
+            err = -1;
+        }
+        free(buf);
+    } else if (strcmp(argv[1], "realloc") == 0) {
+        buf = malloc(size);
+        if (buf == NULL) {
+            printf("Error: malloc before realloc returned NULL\n");
+            err = -1;
+        }
+        buf = realloc(buf, size * 2);
+        if (buf == NULL) {
+            printf("Error: realloc returned NULL\n");
+            err = -1;
+        }
+        free(buf);
+    } else if (strcmp(argv[1], "posix_memalign") == 0) {
+        err = posix_memalign(&buf, 64, size);
+        if (err != 0) {
+            printf("Error: posix_memalign returned %d\n", err);
+        }
+        free(buf);
+    } else {
+        printf("Error: unknown parameter\n");
+        err = -1;
+    }
+
+    return err;
+}
diff --git a/test/bat_tests.cpp b/test/bat_tests.cpp
new file mode 100644
index 0000000..9d4d318
--- /dev/null
+++ b/test/bat_tests.cpp
@@ -0,0 +1,532 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include <fstream>
+#include <algorithm>
+#include <numaif.h>
+#include <memkind/internal/memkind_regular.h>
+
+#include "common.h"
+#include "check.h"
+#include "omp.h"
+#include "trial_generator.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+#include "Allocator.hpp"
+#include "TestPolicy.hpp"
+
+#define TEST_PREFIX "test_TC_MEMKIND_BAT_"
+
+typedef void(*test_function)(Allocator *, size_t);
+
+typedef std::tuple<test_function, memkind_memtype_t, memkind_policy_t, memkind_bits_t, size_t>
+memtype_policy_test_params;
+
+test_function get_function(memtype_policy_test_params params)
+{
+    return std::get<0>(params);
+}
+
+memkind_memtype_t get_memtype(memtype_policy_test_params params)
+{
+    return std::get<1>(params);
+}
+
+memkind_policy_t get_policy(memtype_policy_test_params params)
+{
+    return std::get<2>(params);
+}
+
+memkind_bits_t get_flags(memtype_policy_test_params params)
+{
+    return std::get<3>(params);
+}
+
+size_t get_size(memtype_policy_test_params params)
+{
+    return std::get<4>(params);
+}
+
+typedef std::tuple<test_function, hbw_policy_t, size_t> hbw_policy_test_params;
+
+test_function get_function(hbw_policy_test_params params)
+{
+    return std::get<0>(params);
+}
+
+hbw_policy_t get_hbw_policy(hbw_policy_test_params params)
+{
+    return std::get<1>(params);
+}
+
+size_t get_size(hbw_policy_test_params params)
+{
+    return std::get<2>(params);
+}
+
+typedef std::tuple<hbw_policy_t, size_t> hbw_policy_huge_page_test_params;
+
+hbw_policy_t get_hbw_policy(hbw_policy_huge_page_test_params params)
+{
+    return std::get<0>(params);
+}
+
+size_t get_size(hbw_policy_huge_page_test_params params)
+{
+    return std::get<1>(params);
+}
+
+
+
+/*
+ * Set of basic acceptance tests.
+ */
+class BATest: public TGTest {};
+class BasicAllocTest
+{
+public:
+
+    BasicAllocTest(Allocator *allocator) : allocator(allocator) {}
+
+    void check_policy_and_numa_node(void *ptr, size_t size)
+    {
+        int policy = allocator->get_numa_policy();
+
+        EXPECT_NE(-1, policy);
+
+        if (allocator->is_high_bandwidth()) {
+            TestPolicy::check_hbw_numa_nodes(policy, ptr, size);
+        } else {
+            TestPolicy::check_all_numa_nodes(policy, ptr, size);
+        }
+    }
+
+    void record_page_association(void *ptr, size_t size)
+    {
+        TestPolicy::record_page_association(ptr, size, allocator->get_page_size());
+    }
+
+    void malloc(size_t size)
+    {
+        void *ptr = allocator->malloc(size);
+        ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+        void *memset_ret = memset(ptr, 3, size);
+        EXPECT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void calloc(size_t num, size_t size)
+    {
+        void *ptr = allocator->calloc(num, size);
+        ASSERT_TRUE(ptr != NULL) << "calloc() returns NULL";
+        void *memset_ret = memset(ptr, 3, size);
+        ASSERT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void realloc(size_t size)
+    {
+        void *ptr = allocator->malloc(size);
+        ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+        size_t realloc_size = size+128;
+        ptr = allocator->realloc(ptr, realloc_size);
+        ASSERT_TRUE(ptr != NULL) << "realloc() returns NULL";
+        void *memset_ret = memset(ptr, 3, realloc_size);
+        ASSERT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void memalign(size_t alignment, size_t size)
+    {
+        void *ptr = NULL;
+        int ret = allocator->memalign(&ptr, alignment, size);
+        ASSERT_EQ(0, ret) << "posix_memalign() != 0";
+        ASSERT_TRUE(ptr != NULL) << "posix_memalign() returns NULL pointer";
+        void *memset_ret = memset(ptr, 3, size);
+        ASSERT_TRUE(memset_ret != NULL);
+        record_page_association(ptr, size);
+        check_policy_and_numa_node(ptr, size);
+        allocator->free(ptr);
+    }
+
+    void free(size_t size)
+    {
+        void *ptr = allocator->malloc(size);
+        ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+        allocator->free(ptr);
+    }
+
+    virtual ~BasicAllocTest() {}
+private:
+    Allocator *allocator;
+};
+
+
+static void test_malloc(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).malloc(size);
+}
+
+static void test_calloc(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).calloc(1, size);
+}
+
+static void test_realloc(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).realloc(size);
+}
+
+static void test_memalign(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).memalign(4096, size);
+}
+
+static void test_free(Allocator *allocator, size_t size)
+{
+    BasicAllocTest(allocator).free(size);
+}
+
+
+template <typename T>
+std::vector<T> GetKeys(std::map<T, std::string> dict)
+{
+    std::vector<T> keys;
+    for (auto const &item: dict) {
+        keys.push_back(item.first);
+    }
+    return keys;
+}
+
+struct TestParameters {
+    static const std::map<test_function, std::string> functions;
+    static const std::map<memkind_memtype_t, std::string> memtypes;
+    static const std::map<memkind_policy_t, std::string> policies;
+    static const std::map<hbw_policy_t, std::string> hbw_policies;
+    static const std::vector<size_t> sizes;
+};
+
+const map<test_function, std::string> TestParameters::functions = {
+    {&test_malloc, "malloc"},
+    {&test_calloc, "calloc"},
+    {&test_realloc, "realloc"},
+    {&test_memalign, "memalign"},
+    {&test_free, "free"}
+};
+const std::map<memkind_memtype_t, std::string> TestParameters::memtypes = {{MEMKIND_MEMTYPE_DEFAULT, "DEFAULT"},
+    {MEMKIND_MEMTYPE_HIGH_BANDWIDTH, "HIGH_BANDWIDTH"}
+};
+const std::map<memkind_policy_t, std::string> TestParameters::policies = {{MEMKIND_POLICY_PREFERRED_LOCAL, "PREFERRED_LOCAL"},
+    {MEMKIND_POLICY_BIND_LOCAL, "BIND_LOCAL"},
+    {MEMKIND_POLICY_BIND_ALL, "BIND_ALL"},
+    {MEMKIND_POLICY_INTERLEAVE_ALL, "MEMKIND_POLICY_INTERLEAVE_ALL"}
+};
+const std::map<hbw_policy_t, std::string> TestParameters::hbw_policies = {{HBW_POLICY_PREFERRED, "HBW_POLICY_PREFERRED"},
+    {HBW_POLICY_BIND, "HBW_POLICY_BIND"},
+    {HBW_POLICY_BIND_ALL, "HBW_POLICY_BIND_ALL"},
+    {HBW_POLICY_INTERLEAVE, "HBW_POLICY_INTERLEAVE"}
+};
+const std::vector<size_t> TestParameters::sizes = {4096, 4194305};
+
+
+class MemtypePolicyTest: public TGTest,
+    public ::testing::WithParamInterface<memtype_policy_test_params>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<memtype_policy_test_params> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::functions.at(get_function(info.param)) + "_" +
+               TestParameters::memtypes.at(get_memtype(info.param)) + "_" +
+               TestParameters::policies.at(get_policy(info.param)) + "_" +
+               (get_flags(info.param) == MEMKIND_MASK_PAGE_SIZE_2MB ? "PAGE_SIZE_2MB_" : "") +
+               std::to_string(get_size(info.param)) + "_bytes";
+    }
+};
+
+TEST_P(MemtypePolicyTest, memkind_heap_mgmt)
+{
+    HugePageOrganizer huge_page_organizer(8);
+    auto params = GetParam();
+    auto flags = get_flags(params);
+    MemkindAllocator allocator(get_memtype(GetParam()), get_policy(GetParam()),
+                               flags);
+    get_function(GetParam())(&allocator, get_size(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(DEFAULT,
+                        MemtypePolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::Values(MEMKIND_MEMTYPE_DEFAULT),
+                                           ::testing::Values(MEMKIND_POLICY_PREFERRED_LOCAL),
+                                           ::testing::Values(MEMKIND_MASK_PAGE_SIZE_2MB, memkind_bits_t()),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemtypePolicyTest::get_test_name_suffix
+                       );
+
+INSTANTIATE_TEST_CASE_P(HBW,
+                        MemtypePolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::Values(MEMKIND_MEMTYPE_HIGH_BANDWIDTH),
+                                           ::testing::ValuesIn(GetKeys(TestParameters::policies)),
+                                           ::testing::Values(memkind_bits_t()),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemtypePolicyTest::get_test_name_suffix
+                       );
+
+INSTANTIATE_TEST_CASE_P(HBW_HUGE,
+                        MemtypePolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::Values(MEMKIND_MEMTYPE_HIGH_BANDWIDTH),
+                                           ::testing::Values(MEMKIND_POLICY_PREFERRED_LOCAL,
+                                                             MEMKIND_POLICY_BIND_LOCAL,
+                                                             MEMKIND_POLICY_BIND_ALL),
+                                           ::testing::Values(MEMKIND_MASK_PAGE_SIZE_2MB),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemtypePolicyTest::get_test_name_suffix
+                       );
+
+
+class HBWPolicyHugePageTest: public TGTest,
+    public ::testing::WithParamInterface<hbw_policy_huge_page_test_params>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<hbw_policy_huge_page_test_params> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::hbw_policies.at(get_hbw_policy(info.param)) + "_" +
+               std::to_string(get_size(info.param)) + "_bytes";
+    }
+};
+
+TEST_P(HBWPolicyHugePageTest, memalign)
+{
+    HugePageOrganizer huge_page_organizer(8);
+    HbwmallocAllocator hbwmalloc_allocator(get_hbw_policy(GetParam()));
+    hbwmalloc_allocator.set_memalign_page_size(HBW_PAGESIZE_2MB);
+    BasicAllocTest(&hbwmalloc_allocator).memalign(4096, get_size(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(HBW_memalign,
+                        HBWPolicyHugePageTest,
+                        ::testing::Combine(::testing::Values(HBW_POLICY_PREFERRED, HBW_POLICY_BIND,
+                                                             HBW_POLICY_BIND_ALL),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        HBWPolicyHugePageTest::get_test_name_suffix
+                       );
+
+
+class HBWPolicyTest: public TGTest,
+    public ::testing::WithParamInterface<hbw_policy_test_params>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<hbw_policy_test_params> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::functions.at(get_function(info.param)) + "_" +
+               TestParameters::hbw_policies.at(get_hbw_policy(info.param)) + "_" +
+               std::to_string(get_size(info.param)) + "_bytes";
+    }
+};
+
+TEST_P(HBWPolicyTest, memkind_heap_mgmt)
+{
+    HbwmallocAllocator allocator(get_hbw_policy(GetParam()));
+    get_function(GetParam())(&allocator, get_size(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(HBWPolicy,
+                        HBWPolicyTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::ValuesIn(GetKeys(TestParameters::hbw_policies)),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        HBWPolicyTest::get_test_name_suffix
+                       );
+
+
+class MemkindRegularTest: public TGTest,
+    public ::testing::WithParamInterface<tuple<test_function, size_t>>
+{
+public:
+    static std::string get_test_name_suffix(
+        testing::TestParamInfo<std::tuple<test_function, size_t>> info)
+    {
+        return TEST_PREFIX +
+               TestParameters::functions.at(std::get<0>(info.param)) + "_" +
+               std::to_string(std::get<1>(info.param)) + "_bytes";
+    }
+};
+
+
+TEST_P(MemkindRegularTest, memkind_heap_mgmt)
+{
+    MemkindAllocator allocator(MEMKIND_REGULAR);
+    std::get<0>(GetParam())(&allocator, std::get<1>(GetParam()));
+}
+
+INSTANTIATE_TEST_CASE_P(Regular,
+                        MemkindRegularTest,
+                        ::testing::Combine(::testing::ValuesIn(GetKeys(TestParameters::functions)),
+                                           ::testing::ValuesIn(TestParameters::sizes)),
+                        MemkindRegularTest::get_test_name_suffix
+                       );
+
+
+TEST_F(BATest,
+       test_TC_MEMKIND_malloc_DEFAULT_HIGH_BANDWIDTH_INTERLEAVE_ALL_4194305_bytes)
+{
+    MemkindAllocator allocator((memkind_memtype_t)(MEMKIND_MEMTYPE_DEFAULT |
+                                                   MEMKIND_MEMTYPE_HIGH_BANDWIDTH), MEMKIND_POLICY_INTERLEAVE_ALL,
+                               memkind_bits_t());
+    test_malloc(&allocator, 4194305);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_free_MEMKIND_DEFAULT_free_with_NULL_kind_4096_bytes)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 4096);
+    ASSERT_TRUE(ptr != NULL) << "malloc() returns NULL";
+    memkind_free(0, ptr);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_free_ext_MEMKIND_GBTLB_4096_bytes)
+{
+    HugePageOrganizer huge_page_organizer(1000);
+    MemkindAllocator memkind_allocator(MEMKIND_GBTLB);
+    BasicAllocTest(&memkind_allocator).free(4096);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbwmalloc_Pref_CheckAvailable)
+{
+    ASSERT_EQ(0, hbw_check_available());
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbw_malloc_usable_size_nullptr_0bytes)
+{
+    ASSERT_EQ(0U, hbw_malloc_usable_size(nullptr));
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbw_malloc_usable_size_hbw_malloc_16bytes)
+{
+    void *ptr = hbw_malloc(16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 16U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_malloc_16bytes_def_kind)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 16U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_hbw_malloc_usable_size_hbw_calloc_16bytes_16bytes)
+{
+    void *ptr = hbw_calloc(16, 16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 16U*16U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_calloc_16bytes_16bytes_def_kind)
+{
+    void *ptr = memkind_calloc(MEMKIND_DEFAULT, 16, 16);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 16U*16U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbw_malloc_usable_size_hbw_realloc_1024bytes)
+{
+    void *ptr = hbw_realloc(nullptr, 1024);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 1024U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_realloc_1024bytes_def_kind)
+{
+    void *ptr = memkind_realloc(MEMKIND_DEFAULT, nullptr, 1024);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 1024U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_hbw_malloc_usable_size_hbw_posix_memalign_32bytes)
+{
+    void *ptr = nullptr;
+    int res = hbw_posix_memalign(&ptr, 64, 32);
+    ASSERT_EQ(0, res);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(hbw_malloc_usable_size(ptr), 32U);
+    hbw_free(ptr);
+}
+
+TEST_F(BATest,
+       test_TC_MEMKIND_memkind_malloc_usable_size_memkind_posix_memalign_32bytes_def_kind)
+{
+    void *ptr = nullptr;
+    int res = memkind_posix_memalign(MEMKIND_DEFAULT, &ptr, 64, 32);
+    ASSERT_EQ(0, res);
+    ASSERT_NE(nullptr, ptr);
+    ASSERT_GE(memkind_malloc_usable_size(MEMKIND_DEFAULT, ptr), 32U);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(BATest, test_TC_MEMKIND_hbwmalloc_Pref_Policy)
+{
+    hbw_set_policy(HBW_POLICY_PREFERRED);
+    EXPECT_EQ(HBW_POLICY_PREFERRED, hbw_get_policy());
+}
+
+TEST_F(BATest, test_TC_MEMKIND_REGULAR_nodemask)
+{
+    using namespace TestPolicy;
+    void *mem = memkind_malloc(MEMKIND_REGULAR, 1234567);
+    unique_bitmask_ptr kind_nodemask = make_nodemask_ptr();
+    ASSERT_EQ(0, memkind_regular_all_get_mbind_nodemask(MEMKIND_REGULAR,
+                                                        kind_nodemask->maskp,
+                                                        kind_nodemask->size));
+
+    check_numa_nodes(kind_nodemask, MPOL_BIND, mem, 1234567);
+    memkind_free(MEMKIND_REGULAR, mem);
+}
+
diff --git a/test/check.cpp b/test/check.cpp
new file mode 100644
index 0000000..a038d1e
--- /dev/null
+++ b/test/check.cpp
@@ -0,0 +1,269 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <numaif.h>
+#include <errno.h>
+#include <fstream>
+#include <sstream>
+#include <iostream>
+#include <fstream>
+#include <string>
+#include <cstring>
+#include <numa.h>
+#include <errno.h>
+
+#include "check.h"
+
+
+using namespace std;
+
+
+/*Check each page between the start
+and the end address additionally also
+check the end address for pagesize*/
+Check::Check(const void *p, const trial_t &trial): Check(p, trial.size,
+                                                             trial.page_size)
+{
+}
+
+Check::Check(const void *p, const size_t size, const size_t page_size)
+{
+    const size_t min_page_size = 4096;
+    this->ptr = p;
+    this->size = size;
+    size_t psize = (page_size >= min_page_size ? page_size : min_page_size);
+    if (p && size && psize) {
+        num_address = size / psize;
+        num_address += size % psize ? 1 : 0;
+
+        address = new void *[num_address];
+        size_t i;
+        for (i = 0; i < num_address - 1; ++i) {
+            address[i] = (char *)ptr + i * psize;
+        }
+        address[i] = (char *)p + size - 1;
+    } else {
+        address = NULL;
+        num_address = 0;
+    }
+}
+
+Check::~Check()
+{
+    delete[] address;
+}
+
+Check::Check(const Check &other)
+{
+    num_address = other.num_address;
+
+    address = new void *[num_address];
+    for (size_t i = 0; i < num_address; ++i) {
+        address[i] = other.address[i];
+    }
+}
+
+int Check::check_zero(void)
+{
+    size_t i;
+    const char *cptr = (char *)ptr;
+    for (i = 0; i < size; ++i) {
+        if (cptr[i] != '\0') {
+            return -1;
+        }
+    }
+    return 0;
+}
+
+int Check::check_align(size_t align)
+{
+    return (size_t)ptr % align;
+}
+
+string Check::skip_to_next_entry (ifstream &ip)
+{
+    string temp, token;
+    size_t found = 0;
+    string empty ="";
+
+    while (!ip.eof()) {
+        getline (ip, temp);
+        found = temp.find("-");
+        if (found != string::npos) {
+            istringstream iss(temp);
+            getline(iss, token, ' ');
+            return token;
+        }
+    }
+    return empty;
+}
+
+string Check::skip_to_next_kpage(ifstream &ip)
+{
+    string temp, token;
+    size_t found = 0;
+    string empty ="";
+
+    while (!ip.eof()) {
+        getline (ip, temp);
+        found = temp.find("KernelPageSize:");
+        if (found != string::npos) {
+            return temp;
+        }
+    }
+    return empty;
+}
+
+
+void Check::get_address_range(string &line,
+                              unsigned long long *start_addr,
+                              unsigned long long *end_addr)
+{
+    stringstream ss(line);
+    string token;
+
+    getline(ss, token, '-');
+    *start_addr = strtoul(token.c_str(),
+                          NULL,
+                          16);
+    getline(ss, token, '-');
+    *end_addr = strtoul(token.c_str(),
+                        NULL,
+                        16);
+}
+
+size_t Check::get_kpagesize(string line)
+{
+    stringstream ss(line);
+    string token;
+    size_t pagesize;
+
+
+    ss  >> token;
+    ss  >> token;
+
+    pagesize = atol(token.c_str());
+
+    return (size_t)pagesize;
+}
+
+int Check::check_page_size(size_t page_size)
+{
+    int err = 0;
+    size_t i;
+
+    ip.open ("/proc/self/smaps");
+
+    populate_smaps_table();
+    if (check_page_size(page_size, address[0])) {
+        err = -1;
+    }
+    for (i = 1; i < num_address && !err; ++i) {
+        if (check_page_size(page_size, address[i])) {
+            err = i;
+        }
+    }
+    return err;
+}
+
+int Check::populate_smaps_table ()
+{
+    string read;
+    size_t lpagesize;
+    smaps_entry_t lentry;
+    unsigned long long start_addr;
+    unsigned long long end_addr;
+
+    ip >> read;
+    while (!ip.eof()) {
+
+        start_addr = end_addr = 0;
+        get_address_range (read,
+                           &start_addr,
+                           &end_addr);
+        read = skip_to_next_kpage(ip);
+        getline(ip, read);
+        lpagesize = get_kpagesize(read);
+        lpagesize *= 1024;
+        lentry.start_addr = start_addr;
+        lentry.end_addr = end_addr;
+        lentry.pagesize = lpagesize;
+        smaps_table.push_back(lentry);
+        read = skip_to_next_entry(ip);
+        if (read.empty()) {
+            break;
+        }
+    }
+
+    if (0 == smaps_table.size()) {
+        fprintf(stderr,"Empty smaps table\n");
+        return -1;
+    } else {
+        return 0;
+    }
+
+}
+
+int Check::check_page_size(size_t page_size, void *vaddr)
+{
+    string read;
+    unsigned long long virt_addr;
+    size_t lpagesize;
+    list<smaps_entry_t>::iterator it;
+    unsigned long long start_addr;
+    unsigned long long end_addr;
+
+    virt_addr = (unsigned long long)(vaddr);
+
+    for (it = smaps_table.begin();
+         it != smaps_table.end();
+         it++) {
+
+        start_addr = it->start_addr;
+        end_addr = it->end_addr;
+
+        if ((virt_addr >= start_addr) &&
+            (virt_addr < end_addr)) {
+            lpagesize = it->pagesize;
+            if (lpagesize == page_size) {
+                return 0;
+            } else {
+                /*The pagesize of allocation and req don't match*/
+                fprintf(stderr,"%zd does not match entry in SMAPS (%zd)\n",
+                        page_size, lpagesize);
+                return -1;
+            }
+        }
+    }
+    /*Never found a match!*/
+    return 1;
+}
diff --git a/test/check.h b/test/check.h
new file mode 100644
index 0000000..f9935f3
--- /dev/null
+++ b/test/check.h
@@ -0,0 +1,68 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef check_include_h
+#define check_include_h
+
+#include <list>
+#include "trial_generator.h"
+
+
+typedef struct {
+    unsigned long start_addr;
+    unsigned long end_addr;
+    size_t pagesize;
+} smaps_entry_t;
+
+
+
+using namespace std;
+
+class Check
+{
+public:
+    Check(const void *p, const trial_t &trial);
+    Check(const void *p, const size_t size, const size_t page_size);
+    Check(const Check &);
+    ~Check();
+    int check_page_size(size_t page_size);
+    int check_zero(void);
+    int check_align(size_t align);
+private:
+    const void *ptr;
+    size_t size;
+    void **address;
+    size_t num_address;
+    list<smaps_entry_t>smaps_table;
+    ifstream ip;
+    string skip_to_next_entry(ifstream &);
+    string skip_to_next_kpage(ifstream &);
+    void get_address_range(string &line, unsigned long long *start_addr,
+                           unsigned long long *end_addr);
+    size_t get_kpagesize(string line);
+    int check_page_size(size_t page_size, void *vaddr);
+    int populate_smaps_table();
+};
+
+#endif
diff --git a/test/common.h b/test/common.h
new file mode 100644
index 0000000..bca4ff6
--- /dev/null
+++ b/test/common.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef COMMON_H
+#define COMMON_H
+
+#include <hbwmalloc.h>
+
+#include <iostream>
+#include <stdint.h>
+#include <stdlib.h>
+#include <gtest/gtest.h>
+
+#define MB 1048576ULL
+#define GB 1073741824ULL
+#define KB 1024ULL
+
+#define HBW_SUCCESS 0
+#define HBW_ERROR -1
+
+#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
+
+#endif
diff --git a/test/dax_kmem_env_var_test.py b/test/dax_kmem_env_var_test.py
new file mode 100644
index 0000000..ef314fc
--- /dev/null
+++ b/test/dax_kmem_env_var_test.py
@@ -0,0 +1,67 @@
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+from distutils.spawn import find_executable
+from python_framework import CMD_helper
+import os
+
+class Test_dax_kmem_env_var(object):
+    os.environ["PATH"] += os.pathsep + os.path.dirname(os.path.dirname(__file__))
+    binary_path = find_executable("memkind-auto-dax-kmem-nodes")
+    environ_err_test = "../environ_err_dax_kmem_malloc_test"
+    environ_err_positive_test = "../environ_err_dax_kmem_malloc_positive_test"
+    expected_libnuma_warning = "libnuma: Warning: node argument -1 is out of range\n\n"
+    fail_msg = "Test failed with:\n {0}"
+    cmd_helper = CMD_helper()
+
+    def get_dax_kmem_nodes(self, nodemask=None):
+        """ This function executes memkind function 'get_mbind_nodemask' and returns its output - comma-separated DAX_KMEM nodes """
+        command = self.binary_path
+        if (nodemask):
+            command = "MEMKIND_DAX_KMEM_NODES={}".format(nodemask) + command
+        output, retcode = self.cmd_helper.execute_cmd(command)
+        assert retcode == 0, self.fail_msg.format("\nError: Execution of \'{0}\' returns {1}, \noutput: {2}".format(command, retcode, output))
+        print("\nExecution of {} returns output {}".format(command, output))
+        return output
+
+    def test_TC_MEMKIND_dax_kmem_env_var_compare_nodemask_default_and_env_variable(self):
+        """ This test checks whether dax_kmem_nodemask_default and dax_kmem_nodemask_env_variable have the same value """
+        dax_kmem_nodemask_default = self.get_dax_kmem_nodes()
+        dax_kmem_nodemask_env_variable = self.get_dax_kmem_nodes(dax_kmem_nodemask_default)
+        assert dax_kmem_nodemask_default == dax_kmem_nodemask_env_variable, self.fail_msg.format("Error: Nodemask dax_kmem_nodemask_default ({0}) " \
+               "is not the same as nodemask dax_kmem_nodemask_env_variable ({1})".format(dax_kmem_nodemask_default, dax_kmem_nodemask_env_variable))
+
+    def test_TC_MEMKIND_dax_kmem_env_var_negative_memkind_malloc(self):
+        """ This test sets usupported value of MEMKIND_DAX_KMEM_NODES, then tries to perform a successful allocation from DRAM using memkind_malloc() """
+        command = "MEMKIND_DAX_KMEM_NODES=-1 " + self.cmd_helper.get_command_path(self.environ_err_test)
+        output, retcode = self.cmd_helper.execute_cmd(command)
+        assert retcode != 0, self.fail_msg.format("\nError: Execution of: \'{0}\' returns: {1} \noutput: {2}".format(command, retcode, output))
+        assert self.expected_libnuma_warning == output, self.fail_msg.format("Error: expected libnuma warning ({0}) " \
+               "was not found (output: {1})").format(self.expected_libnuma_warning, output)
+
+    def test_TC_MEMKIND_dax_kmem_env_var_proper_memkind_malloc(self):
+        """ This test checks if allocation is performed on persistent memory correctly """
+        dax_kmem_nodemask_default = self.get_dax_kmem_nodes().strip()
+        command = "MEMKIND_DAX_KMEM_NODES={} ".format(dax_kmem_nodemask_default) + self.cmd_helper.get_command_path(self.environ_err_positive_test)
+        output, retcode = self.cmd_helper.execute_cmd(command)
+        assert retcode == 0, self.fail_msg.format("\nError: Execution of: \'{0}\' returns: {1} \noutput: {2}".format(command, retcode, output))
diff --git a/test/dax_kmem_nodes.cpp b/test/dax_kmem_nodes.cpp
new file mode 100644
index 0000000..f810d77
--- /dev/null
+++ b/test/dax_kmem_nodes.cpp
@@ -0,0 +1,97 @@
+/*
+* Copyright (C) 2020 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <climits>
+#include <numa.h>
+#include <numaif.h>
+
+#include "dax_kmem_nodes.h"
+
+DaxKmem::DaxKmem()
+{
+    dax_kmem_nodes = get_dax_kmem_nodes();
+}
+
+size_t DaxKmem::get_free_space(void)
+{
+    size_t sum_of_dax_kmem_free_space = 0;
+    long long free_space;
+
+    for(auto const &node: dax_kmem_nodes) {
+        numa_node_size64(node, &free_space);
+        sum_of_dax_kmem_free_space += free_space;
+    }
+
+    return sum_of_dax_kmem_free_space;
+}
+
+std::set<int> DaxKmem::get_closest_numa_nodes(int regular_node)
+{
+    int min_distance = INT_MAX;
+    std::set<int> closest_numa_ids;
+
+    for (auto const &node: dax_kmem_nodes) {
+        int distance_to_i_node = numa_distance(regular_node, node);
+
+        if (distance_to_i_node < min_distance) {
+            min_distance = distance_to_i_node;
+            closest_numa_ids.clear();
+            closest_numa_ids.insert(node);
+        } else if (distance_to_i_node == min_distance) {
+            closest_numa_ids.insert(node);
+        }
+    }
+
+    return closest_numa_ids;
+}
+
+std::set<int> DaxKmem::get_dax_kmem_nodes(void)
+{
+    struct bitmask *cpu_mask = numa_allocate_cpumask();
+    std::set<int> dax_kmem_nodes;
+
+    const int MAXNODE_ID = numa_max_node();
+    for (int id = 0; id <= MAXNODE_ID; ++id) {
+        numa_node_to_cpus(id, cpu_mask);
+
+        // Check if NUMA node exists and if it is NUMA node created from persistent memory
+        if (numa_node_size64(id, nullptr) > 0 &&
+            numa_bitmask_weight(cpu_mask) == 0) {
+            dax_kmem_nodes.insert(id);
+        }
+    }
+    numa_free_cpumask(cpu_mask);
+
+    return dax_kmem_nodes;
+}
+
+size_t DaxKmem::size(void)
+{
+    return dax_kmem_nodes.size();
+}
+
+bool DaxKmem::contains(int node)
+{
+    return (dax_kmem_nodes.find(node) != dax_kmem_nodes.end());
+}
diff --git a/test/dax_kmem_nodes.h b/test/dax_kmem_nodes.h
new file mode 100644
index 0000000..058e0bc
--- /dev/null
+++ b/test/dax_kmem_nodes.h
@@ -0,0 +1,39 @@
+/*
+ * Copyright (C) 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <set>
+
+class DaxKmem
+{
+public:
+    DaxKmem();
+    size_t get_free_space(void);
+    std::set<int> get_closest_numa_nodes(int regular_node);
+    size_t size(void);
+    bool contains(int node);
+
+private:
+    std::set<int> dax_kmem_nodes;
+    std::set<int> get_dax_kmem_nodes(void);
+}; // DaxKmem
diff --git a/test/decorator_test.cpp b/test/decorator_test.cpp
new file mode 100644
index 0000000..152111a
--- /dev/null
+++ b/test/decorator_test.cpp
@@ -0,0 +1,116 @@
+/*
+ * Copyright (C) 2014 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include "common.h"
+#include "decorator_test.h"
+#include "config.h"
+
+size_t size = 16;
+memkind_t kind = MEMKIND_DEFAULT;
+
+class DecoratorTest: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {
+        decorators_state = new decorators_flags();
+    }
+
+    void TearDown()
+    {
+        free(decorators_state);
+    }
+};
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_malloc)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_malloc(kind, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    EXPECT_EQ(1, decorators_state->malloc_pre);
+    EXPECT_EQ(1, decorators_state->malloc_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_calloc)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_calloc(kind, 1, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    EXPECT_EQ(1, decorators_state->calloc_pre);
+    EXPECT_EQ(1, decorators_state->calloc_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_posix_memalign)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer;
+
+    int res = memkind_posix_memalign(kind, &buffer, 8, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    ASSERT_EQ(0, res);
+    EXPECT_EQ(1, decorators_state->posix_memalign_pre);
+    EXPECT_EQ(1, decorators_state->posix_memalign_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_realloc)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_realloc(kind, NULL, size);
+
+    ASSERT_TRUE(buffer != NULL);
+    EXPECT_EQ(1, decorators_state->realloc_pre);
+    EXPECT_EQ(1, decorators_state->realloc_post);
+
+    memkind_free(0, buffer);
+#endif
+}
+
+TEST_F(DecoratorTest, test_TC_MEMKIND_DT_free)
+{
+#ifdef MEMKIND_DECORATION_ENABLED
+    void *buffer = memkind_malloc(kind, size);
+
+    ASSERT_TRUE(buffer != NULL);
+
+    memkind_free(0, buffer);
+
+    EXPECT_EQ(1, decorators_state->free_pre);
+    EXPECT_EQ(1, decorators_state->free_post);
+#endif
+}
\ No newline at end of file
diff --git a/test/decorator_test.h b/test/decorator_test.h
new file mode 100644
index 0000000..d713b73
--- /dev/null
+++ b/test/decorator_test.h
@@ -0,0 +1,100 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdlib.h>
+#include <stdio.h>
+#include <memkind.h>
+
+struct decorators_flags {
+    int malloc_pre;
+    int malloc_post;
+    int calloc_pre;
+    int calloc_post;
+    int posix_memalign_pre;
+    int posix_memalign_post;
+    int realloc_pre;
+    int realloc_post;
+    int free_pre;
+    int free_post;
+};
+
+struct decorators_flags *decorators_state;
+
+extern "C" {
+
+    void memkind_malloc_pre(struct memkind *kind, size_t size)
+    {
+        decorators_state->malloc_pre++;
+    }
+
+    void memkind_malloc_post(struct memkind *kind, size_t size, void **result)
+    {
+        decorators_state->malloc_post++;
+    }
+
+    void memkind_calloc_pre(struct memkind *kind, size_t nmemb, size_t size)
+    {
+        decorators_state->calloc_pre++;
+    }
+
+    void memkind_calloc_post(struct memkind *kind, size_t nmemb, size_t size,
+                             void **result)
+    {
+        decorators_state->calloc_post++;
+    }
+
+    void memkind_posix_memalign_pre(struct memkind *kind, void **memptr,
+                                    size_t alignment, size_t size)
+    {
+        decorators_state->posix_memalign_pre++;
+    }
+
+    void memkind_posix_memalign_post(struct memkind *kind, void **memptr,
+                                     size_t alignment, size_t size, int *err)
+    {
+        decorators_state->posix_memalign_post++;
+    }
+
+    void memkind_realloc_pre(struct memkind *kind, void *ptr, size_t size)
+    {
+        decorators_state->realloc_pre++;
+    }
+
+    void memkind_realloc_post(struct memkind *kind, void *ptr, size_t size,
+                              void **result)
+    {
+        decorators_state->realloc_post++;
+    }
+
+    void memkind_free_pre(struct memkind **kind, void **ptr)
+    {
+        decorators_state->free_pre++;
+    }
+
+    void memkind_free_post(struct memkind **kind, void **ptr)
+    {
+        decorators_state->free_post++;
+    }
+
+}
diff --git a/test/dlopen_test.cpp b/test/dlopen_test.cpp
new file mode 100644
index 0000000..8968f13
--- /dev/null
+++ b/test/dlopen_test.cpp
@@ -0,0 +1,108 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+#include <dlfcn.h>
+
+#include "common.h"
+
+class DlopenTest: public :: testing::Test
+{
+protected:
+    DlopenTest()
+    {
+        const char *path = "/usr/lib64/libmemkind.so";
+        if (!pathExists(path)) {
+            path = "/usr/lib/libmemkind.so";
+        }
+        dlerror();
+        handle = dlopen(path, RTLD_LAZY);
+        assert((handle != NULL && dlerror() == NULL) && "Couldn't open libmemkind.so");
+        memkind_malloc = (memkind_malloc_t)dlsym(handle, "memkind_malloc");
+        assert(dlerror() == NULL && "Couldn't get memkind_malloc from memkind library");
+        memkind_free = (memkind_free_t)dlsym(handle, "memkind_free");
+        assert(dlerror() == NULL && "Couldn't get memkind_free from memkind library");
+    }
+
+    ~DlopenTest()
+    {
+        dlclose(handle);
+    }
+
+    void test(const char *kind_name, size_t alloc_size)
+    {
+        void **kind_ptr = (void **)dlsym(handle, kind_name);
+        EXPECT_TRUE(dlerror() == NULL) << "Couldn't get kind from memkind library";
+        EXPECT_TRUE(kind_ptr != NULL) << "Kind ptr to memkind library is NULL";
+
+        void *allocation_ptr = memkind_malloc((*kind_ptr), alloc_size);
+        EXPECT_TRUE(allocation_ptr != NULL) << "Allocation with memkind_malloc failed";
+
+        memset(allocation_ptr, 0, alloc_size);
+
+        memkind_free((*kind_ptr), allocation_ptr);
+    }
+
+    bool pathExists(const char *p)
+    {
+        struct stat info;
+        if (0 != stat(p, &info)) {
+            return false;
+        }
+        return true;
+    }
+
+private:
+    void *handle;
+    typedef void *(*memkind_malloc_t)(void *, size_t);
+    typedef void (*memkind_free_t)(void *, void *);
+    memkind_malloc_t memkind_malloc;
+    memkind_free_t memkind_free;
+};
+
+TEST_F(DlopenTest, test_TC_MEMKIND_DEFAULT_4194305_bytes)
+{
+    test("MEMKIND_DEFAULT", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_4194305_bytes)
+{
+    test("MEMKIND_HBW", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_HUGETLB_4194305_bytes)
+{
+    HugePageOrganizer huge_page_organizer(8);
+    test("MEMKIND_HBW_HUGETLB", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_PREFERRED_4194305_bytes)
+{
+    test("MEMKIND_HBW_PREFERRED", 4194305);
+}
+
+TEST_F(DlopenTest, test_TC_MEMKIND_HBW_INTERLEAVE_4194305_bytes)
+{
+    test("MEMKIND_HBW_INTERLEAVE", 4194305);
+}
diff --git a/test/draw_plots.py b/test/draw_plots.py
new file mode 100644
index 0000000..1d26f60
--- /dev/null
+++ b/test/draw_plots.py
@@ -0,0 +1,271 @@
+#
+#  Copyright (C) 2016 - 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+from mpl_toolkits.mplot3d import Axes3D
+import matplotlib.pyplot as plt
+import numpy as np
+import numpy.ma as ma
+import os
+from shutil import rmtree
+
+files = ('alloctest_hbw.txt', 'alloctest_glibc.txt', 'alloctest_tbb.txt', 'alloctest_pmem.txt')
+legend = ('avg hbw', 'avg glibc', 'avg tbb', 'avg pmem', 'first operation')
+colors = ('red', 'green', 'blue', 'black')
+first_operation_color = 'yellow'
+
+threads_values = ('1', '2', '4', '8', '16', '18', '36')
+
+small_sizes_values = ('1', '4', '16')
+medium_sizes_values = ('64', '256')
+big_sizes_values = ('1024', '4096', '16384')
+sizes_values = (small_sizes_values, medium_sizes_values, big_sizes_values)
+sizes_display = ('small', 'medium', 'big')
+
+operations = ('Allocation', 'Free', 'Total')
+
+iterations = 1000
+output_directory = './plots/'
+
+threads_index      = np.arange(len(threads_values))
+small_sizes_index  = np.arange(len(small_sizes_values))
+medium_sizes_index = np.arange(len(medium_sizes_values))
+big_sizes_index    = np.arange(len(big_sizes_values))
+sizes_index = (small_sizes_index, medium_sizes_index, big_sizes_index)
+
+# 3D and 2D plots needs different width, so that bars do not cover each other in 3D
+bar_width_2D = 0.2
+bar_width_3D = 0.1
+
+# return times (total, allocation, free, first allocation, first free) as columns 2 - 6
+def return_times(entry):
+    return entry[:,2], entry[:,3], entry[:,4], entry[:,5], entry[:,6]
+
+# initialize axis with labels, ticks and values
+def init_axis(fig, suptitle, subplot_projection, x_label, x_ticks, x_values, y_label, y_ticks, y_values, z_label):
+    assert fig is not None
+    fig.clf()
+    fig.suptitle(suptitle)
+    if subplot_projection:
+        ax = fig.add_subplot(111, projection=subplot_projection)
+    else:
+        ax = fig.add_subplot(111)
+    assert ax is not None
+    if x_label:
+        ax.set_xlabel(x_label)
+        if x_ticks is not None and x_values is not None:
+            ax.set_xticks(x_ticks)
+            ax.set_xticklabels(x_values)
+    if y_label:
+        ax.set_ylabel(y_label)
+        if y_ticks is not None and y_values is not None:
+            ax.set_yticks(y_ticks)
+            ax.set_yticklabels(y_values)
+    if z_label:
+        ax.set_zlabel(z_label)
+    return ax
+
+def draw_bar(ax, show_first_operation, x, y, time, init_time, draw_color):
+    assert ax is not None
+    if y is None:
+        if show_first_operation is True:
+            mask_time = ma.where(time>=init_time)
+            mask_init_time = ma.where(init_time>=time)
+            ax.bar(x[mask_time], time[mask_time], width=bar_width, color=draw_color)
+            ax.bar(x, init_time, width=bar_width, color=first_operation_color)
+            ax.bar(x[mask_init_time], time[mask_init_time], width=bar_width, color=draw_color)
+        else:
+            ax.bar(x, time, width=bar_width, color=draw_color)
+    else:
+        if show_first_operation is True:
+            mask_time = ma.where(time>=init_time)
+            mask_init_time = ma.where(init_time>=time)
+            ax.bar(x[mask_time], (time - init_time)[mask_time], y[mask_time], bottom=init_time[mask_time], zdir='y', width=bar_width, color=draw_color)
+            ax.bar(x[mask_init_time], (init_time - time)[mask_init_time], y[mask_init_time], bottom=time[mask_init_time], zdir='y', width=bar_width, color=first_operation_color)
+            ax.bar(x[mask_init_time], time[mask_init_time], y[mask_init_time], zdir='y', width=bar_width, color=draw_color)
+            ax.bar(x[mask_time], init_time[mask_time], y[mask_time], zdir='y', width=bar_width, color=first_operation_color)
+        else:
+            ax.bar(x, time, y, zdir='y', width=bar_width, color=draw_color)
+
+def save_plot(show_first_operation, subfolder, filename):
+    if show_first_operation:
+        path = os.path.join(output_directory, "with_first_operation")
+    else:
+        path = os.path.join(output_directory, "without_first_operation")
+    if subfolder:
+        path = os.path.join(path, subfolder)
+    if not os.path.exists(path):
+        os.mkdir(path)
+    plt.savefig(os.path.join(path, filename))
+    print "Saved file %s" % filename
+
+def load_data_from_files():
+    data = []
+    # load all files into data
+    for f in files:
+        assert os.path.isfile(f) is True
+        data.append(np.loadtxt(f, comments='#'))
+    return data
+
+def set_bar_width_and_offsets(requested_width):
+    bar_width = requested_width
+    offsets = (0, bar_width, 2*bar_width, 3*bar_width)
+    return bar_width, offsets
+
+if os.path.exists(output_directory):
+    rmtree(output_directory)
+os.mkdir(output_directory)
+
+data = load_data_from_files()
+
+fig = plt.figure()
+
+########################################
+# Draw 3D plots (time, sizes, threads) #
+########################################
+
+bar_width, offsets = set_bar_width_and_offsets(bar_width_3D)
+
+# for each size range (small, medium, big)
+for size_values, size_index, size_display in zip(sizes_values, sizes_index, sizes_display):
+    # show plots with and without first operation
+    for show_first_operation in (True, False):
+        # for each operation (allocation, free, total)
+        for operation in operations:
+            # add bar_width to each element of size_index
+            ax = init_axis(fig, "%s time of %s sizes (%s iterations)" % (operation, size_display, iterations), '3d',
+                           'size [kB]', size_index + (bar_width,) * len(size_index), size_values,
+                           'threads', threads_index, threads_values,
+                           'time [ms]')
+            legend_data = []
+            # for each allocator (hbw, glibc, tbb, pmem)
+            for entry, offset, draw_color in zip(data, offsets, colors):
+                # remove all rows where column 1 (size) is not in size_values (current size range)
+                entry = entry[np.in1d(entry[:,1], np.array(size_values).astype(np.int))]
+                # convert column 0 (threads values) to thread index
+                threads_col = np.array([threads_values.index(str(n)) for n in map(int, entry[:,0])])
+                # convert column 1 (sizes values) to size index
+                size_col = [size_values.index(str(n)) for n in map(int, entry[:,1])]
+                # add offset to size index so that bars display near each other
+                size_col = np.array(size_col) + offset
+                total_time_col, alloc_time_col, free_time_col, first_alloc_time_col, first_free_time_col = return_times(entry)
+                if operation == 'Allocation':
+                    draw_bar(ax, show_first_operation, size_col, threads_col, alloc_time_col, first_alloc_time_col, draw_color)
+                elif operation == 'Free':
+                    draw_bar(ax, show_first_operation, size_col, threads_col, free_time_col, first_free_time_col, draw_color)
+                elif operation == 'Total':
+                    draw_bar(ax, show_first_operation, size_col, threads_col, total_time_col, first_alloc_time_col+first_free_time_col, draw_color)
+                legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=draw_color))
+            if show_first_operation is True:
+                legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=first_operation_color))
+            ax.legend(legend_data,legend,loc='best')
+            plt.grid()
+            save_plot(show_first_operation, None, "%s_time_of_%s_sizes_%s_iterations.png" % (operation, size_display, iterations))
+
+################################################
+# Draw 2D plots (time, sizes, constant thread) #
+################################################
+
+bar_width, offsets = set_bar_width_and_offsets(bar_width_2D)
+
+# for each size range (small, medium, big)
+for size_values, size_index, size_display in zip(sizes_values, sizes_index, sizes_display):
+    # show plots with and without first operation
+    for show_first_operation in (True, False):
+        for thread in threads_values:
+            # for each operation (allocation, free, total)
+            for operation in operations:
+                # add bar_width to each element of size_index
+                ax = init_axis(fig, "%s time of %s sizes with %s threads (%s operations)" % (operation, size_display, thread, iterations), None,
+                               'size [kB]', size_index + (bar_width,) * len(size_index), size_values,
+                               'time [ms]', None, None,
+                               None)
+                legend_data = []
+                # for each allocator (hbw, glibc, tbb, pmem)
+                for entry, offset, draw_color in zip(data, offsets, colors):
+                    # remove all rows where column 1 (size) is not in size_values (current size range)
+                    entry = entry[np.in1d(entry[:,1], np.array(size_values).astype(np.int))]
+                    # remove all rows where column 0 (threads) is not equal to currently analyzed thread value
+                    entry = entry[entry[:,0] == int(thread)]
+                    # convert column 0 (threads values) to thread index
+                    threads_col = np.array([threads_values.index(str(n)) for n in map(int, entry[:,0])])
+                    # convert column 1 (size values) to size index
+                    size_col = [size_values.index(str(n)) for n in map(int, entry[:,1])]
+                    # add offset to size index so that bars display near each other
+                    size_col = np.array(size_col) + offset
+                    total_time_col, alloc_time_col, free_time_col, first_alloc_time_col, first_free_time_col = return_times(entry)
+                    if operation == 'Allocation':
+                        draw_bar(ax, show_first_operation, size_col, None, alloc_time_col, first_alloc_time_col, draw_color)
+                    elif operation == 'Free':
+                        draw_bar(ax, show_first_operation, size_col, None, free_time_col, first_free_time_col, draw_color)
+                    elif operation == 'Total':
+                        draw_bar(ax, show_first_operation, size_col, None, total_time_col, first_alloc_time_col+first_free_time_col, draw_color)
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=draw_color))
+                if show_first_operation is True:
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=first_operation_color))
+                ax.legend(legend_data,legend,loc='best')
+                plt.grid()
+                save_plot(show_first_operation, "constant_thread", "%s_time_of_%s_sizes_with_%s_threads_%s_iterations.png" % (operation, size_display, thread, iterations))
+
+################################################
+# Draw 2D plots (time, threads, constant size) #
+################################################
+
+# for each size range (small, medium, big)
+for size_values, size_index, size_display in zip(sizes_values, sizes_index, sizes_display):
+    # show plots with and without first operation
+    for show_first_operation in (True, False):
+        for size in size_values:
+            # for each operation (allocation, free, total)
+            for operation in operations:
+                # add bar_width to each element of threads_index
+                ax = init_axis(fig, "%s time of %s kB (%s operations)" % (operation, size, iterations), None,
+                               'threads', threads_index + (bar_width,) * len(threads_index), threads_values,
+                               'time [ms]', None, None,
+                               None)
+                legend_data = []
+                # for each allocator (hbw, glibc, tbb, pmem)
+                for entry, offset, draw_color in zip(data, offsets, colors):
+                    # remove all rows where column 1 (size) is not in size_values (current size range)
+                    entry = entry[np.in1d(entry[:,1], np.array(size_values).astype(np.int))]
+                    # remove all rows where column 1 (size) is not equal to currently analyzed size value
+                    entry = entry[entry[:,1] == int(size)]
+                    # convert column 0 (threads values) to thread index
+                    threads_col = np.array([threads_values.index(str(n)) for n in map(int, entry[:,0])])
+                    # add offset to thread index so that bars display near each other
+                    threads_col = np.array(threads_col) + offset
+                    # convert column 1 (size values) to size index
+                    size_col = [size_values.index(str(n)) for n in map(int, entry[:,1])]
+                    total_time_col, alloc_time_col, free_time_col, first_alloc_time_col, first_free_time_col = return_times(entry)
+                    if operation == 'Allocation':
+                        draw_bar(ax, show_first_operation, threads_col, None, alloc_time_col, first_alloc_time_col, draw_color)
+                    elif operation == 'Free':
+                        draw_bar(ax, show_first_operation, threads_col, None, free_time_col, first_free_time_col, draw_color)
+                    elif operation == 'Total':
+                        draw_bar(ax, show_first_operation, threads_col, None, total_time_col, first_alloc_time_col+first_free_time_col, draw_color)
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=draw_color))
+                if show_first_operation is True:
+                    legend_data.append(plt.Rectangle((0, 0), 1, 1, fc=first_operation_color))
+                ax.legend(legend_data,legend,loc='best')
+                plt.grid()
+                save_plot(show_first_operation, "constant_size", "%s_time_of_%s_kB_%s_operations.png" % (operation, size, iterations))
diff --git a/test/environ_err_dax_kmem_malloc_positive_test.cpp b/test/environ_err_dax_kmem_malloc_positive_test.cpp
new file mode 100644
index 0000000..ca17f04
--- /dev/null
+++ b/test/environ_err_dax_kmem_malloc_positive_test.cpp
@@ -0,0 +1,103 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "memkind/internal/memkind_dax_kmem.h"
+
+#include <numa.h>
+#include <numaif.h>
+
+#include "common.h"
+
+#include <climits>
+#include <vector>
+
+/* This test is run with overridden MEMKIND_DAX_KMEM_NODES environment variable
+ * and tries to perform allocation on persistent memory using memkind_malloc()
+ */
+int main()
+{
+    struct bitmask *all_dax_kmem_nodes_nodemask = nullptr;
+    void *ptr = nullptr;
+    int ret = 1;
+    std::vector<int> closest_numa_ids;
+    int expected_numa_id = -1;
+    int returned_numa_id = -1;
+    char *env_value_str = secure_getenv("MEMKIND_DAX_KMEM_NODES");
+    int process_cpu = sched_getcpu();
+    int process_node = numa_node_of_cpu(process_cpu);
+    unsigned numa_max_id = (unsigned) numa_max_node();
+    int min_distance;
+
+    if (!env_value_str) {
+        printf("Error: MEMKIND_DAX_KMEM environment variable is empty.\n");
+        goto exit;
+    }
+
+    ptr = memkind_malloc(MEMKIND_DAX_KMEM, KB);
+    if (ptr == nullptr) {
+        printf("Error: allocation failed\n");
+        goto exit;
+    }
+    memset(ptr, 'A', KB/2);
+
+    all_dax_kmem_nodes_nodemask = numa_parse_nodestring(env_value_str);
+    min_distance = INT_MAX;
+    for (unsigned i = 0; i <= numa_max_id; ++i) {
+        if (numa_bitmask_isbitset(all_dax_kmem_nodes_nodemask, i)) {
+            int distance_to_i_node = numa_distance(process_node, i);
+            if (distance_to_i_node < min_distance) {
+                min_distance = distance_to_i_node;
+                closest_numa_ids.clear();
+                closest_numa_ids.push_back(i);
+            } else if (distance_to_i_node == min_distance) {
+                closest_numa_ids.push_back(i);
+            }
+        }
+    }
+
+    get_mempolicy(&returned_numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+
+    for (auto const &closest_numa : closest_numa_ids) {
+        if (closest_numa == returned_numa_id) {
+            expected_numa_id = closest_numa;
+            ret = 0;
+            break;
+        }
+    }
+    if (ret) {
+        printf("Error: Memkind dax kmem and allocated pointer node id's are not equal: %d!=%d\n",
+               expected_numa_id, returned_numa_id);
+    }
+
+exit:
+    if (all_dax_kmem_nodes_nodemask) {
+        numa_free_nodemask(all_dax_kmem_nodes_nodemask);
+    }
+    if (ptr) {
+        memkind_free(MEMKIND_DAX_KMEM, ptr);
+    }
+
+    return ret;
+}
diff --git a/test/environ_err_dax_kmem_malloc_test.cpp b/test/environ_err_dax_kmem_malloc_test.cpp
new file mode 100644
index 0000000..e5c709f
--- /dev/null
+++ b/test/environ_err_dax_kmem_malloc_test.cpp
@@ -0,0 +1,93 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "memkind/internal/memkind_dax_kmem.h"
+
+#include <numa.h>
+#include <numaif.h>
+
+#include "common.h"
+
+/* This test is run with overridden MEMKIND_DAX_KMEM_NODES environment variable
+ * and tries to perform allocation on DRAM using memkind_malloc()
+ */
+int main()
+{
+    struct bitmask *expected_nodemask = nullptr;
+    struct bitmask *returned_nodemask = nullptr;
+    void *ptr = nullptr;
+    int ret = 0;
+    int status = 0;
+
+    ptr = memkind_malloc(MEMKIND_DAX_KMEM, KB);
+    if (ptr != nullptr) {
+        printf("Error: allocation to DAX_KMEM should fail\n");
+        goto exit;
+    }
+
+    ptr = memkind_malloc(MEMKIND_DEFAULT, KB);
+    if (ptr == nullptr) {
+        printf("Error: allocation failed\n");
+        goto exit;
+    }
+
+    expected_nodemask = numa_allocate_nodemask();
+    status = memkind_dax_kmem_all_get_mbind_nodemask(nullptr,
+                                                     expected_nodemask->maskp,
+                                                     expected_nodemask->size);
+    if (status != MEMKIND_ERROR_ENVIRON) {
+        printf("Error: wrong return value from memkind_dax_kmem_all_get_mbind_nodemask()\n");
+        printf("Expected: %d\n", MEMKIND_ERROR_ENVIRON);
+        printf("Actual: %d\n", status);
+        goto exit;
+    }
+
+    returned_nodemask = numa_allocate_nodemask();
+    status = get_mempolicy(nullptr, returned_nodemask->maskp,
+                           returned_nodemask->size,
+                           ptr, MPOL_F_ADDR);
+    if (status) {
+        printf("Error: get_mempolicy() returned %d\n", status);
+        goto exit;
+    }
+
+    ret = numa_bitmask_equal(returned_nodemask, expected_nodemask);
+    if (!ret) {
+        printf("Error: Memkind dax kmem and allocated pointer nodemasks are not equal\n");
+    }
+
+exit:
+    if (expected_nodemask) {
+        numa_free_nodemask(expected_nodemask);
+    }
+    if (returned_nodemask) {
+        numa_free_nodemask(returned_nodemask);
+    }
+    if (ptr) {
+        memkind_free(nullptr, ptr);
+    }
+
+    return ret;
+}
diff --git a/test/environ_err_hbw_malloc_test.cpp b/test/environ_err_hbw_malloc_test.cpp
new file mode 100644
index 0000000..9a954cc
--- /dev/null
+++ b/test/environ_err_hbw_malloc_test.cpp
@@ -0,0 +1,87 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbwmalloc.h>
+#include <memkind.h>
+#include "memkind/internal/memkind_hbw.h"
+
+#include <numa.h>
+#include <numaif.h>
+
+#include "common.h"
+
+/* This test is run with overridden MEMKIND_HBW_NODES environment variable
+ * and tries to perform allocation from DRAM using hbw_malloc() using
+ * default HBW_POLICY_PREFERRED policy.
+ */
+int main()
+{
+    struct bitmask *expected_nodemask = NULL;
+    struct bitmask *returned_nodemask = NULL;
+    void *ptr = NULL;
+    int ret = 0;
+    int status = 0;
+
+    ptr = hbw_malloc(KB);
+    if (ptr == NULL) {
+        printf("Error: allocation failed\n");
+        goto exit;
+    }
+
+    expected_nodemask = numa_allocate_nodemask();
+    status = memkind_hbw_all_get_mbind_nodemask(NULL, expected_nodemask->maskp,
+                                                expected_nodemask->size);
+    if (status != MEMKIND_ERROR_ENVIRON) {
+        printf("Error: wrong return value from memkind_hbw_all_get_mbind_nodemask()\n");
+        printf("Expected: %d\n", MEMKIND_ERROR_ENVIRON);
+        printf("Actual: %d\n", status);
+        goto exit;
+    }
+
+    returned_nodemask = numa_allocate_nodemask();
+    status = get_mempolicy(NULL, returned_nodemask->maskp, returned_nodemask->size,
+                           ptr, MPOL_F_ADDR);
+    if (status) {
+        printf("Error: get_mempolicy() returned %d\n", status);
+        goto exit;
+    }
+
+    ret = numa_bitmask_equal(returned_nodemask, expected_nodemask);
+    if (!ret) {
+        printf("Error: Memkind hbw and allocated pointer nodemasks are not equal\n");
+    }
+
+exit:
+    if (expected_nodemask) {
+        numa_free_nodemask(expected_nodemask);
+    }
+    if (returned_nodemask) {
+        numa_free_nodemask(returned_nodemask);
+    }
+    if (ptr) {
+        hbw_free(ptr);
+    }
+
+    return ret;
+}
diff --git a/test/error_message_tests.cpp b/test/error_message_tests.cpp
new file mode 100644
index 0000000..72f8acc
--- /dev/null
+++ b/test/error_message_tests.cpp
@@ -0,0 +1,94 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <errno.h>
+#include <gtest/gtest.h>
+
+/* Tests which calls APIS in wrong ways to generate Error Messages thrown by the
+ * the memkind library
+ */
+
+const int all_error_code[] = {
+    MEMKIND_ERROR_UNAVAILABLE,
+    MEMKIND_ERROR_MBIND,
+    MEMKIND_ERROR_MMAP,
+    MEMKIND_ERROR_MALLOC,
+    MEMKIND_ERROR_RUNTIME,
+    MEMKIND_ERROR_ENVIRON,
+    MEMKIND_ERROR_INVALID,
+    MEMKIND_ERROR_TOOMANY,
+    MEMKIND_ERROR_BADOPS,
+    MEMKIND_ERROR_HUGETLB,
+    MEMKIND_ERROR_ARENAS_CREATE,
+    MEMKIND_ERROR_OPERATION_FAILED,
+    MEMKIND_ERROR_MEMTYPE_NOT_AVAILABLE,
+    EINVAL,
+    ENOMEM
+};
+
+class ErrorMessage: public :: testing :: Test
+{
+protected:
+
+    void SetUp()
+    {
+    }
+    void TearDown()
+    {
+    }
+};
+
+TEST_F(ErrorMessage, test_TC_MEMKIND_ErrorMsgLength)
+{
+    size_t i;
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    for (i = 0; i < sizeof(all_error_code)/sizeof(all_error_code[0]); ++i) {
+        memkind_error_message(all_error_code[i], error_message,
+                              MEMKIND_ERROR_MESSAGE_SIZE);
+        EXPECT_TRUE(strlen(error_message) < MEMKIND_ERROR_MESSAGE_SIZE - 1);
+    }
+    memkind_error_message(MEMKIND_ERROR_UNAVAILABLE, NULL, 0);
+}
+
+TEST_F(ErrorMessage, test_TC_MEMKIND_ErrorMsgFormat)
+{
+    size_t i;
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    for (i = 0; i < sizeof(all_error_code)/sizeof(all_error_code[0]); ++i) {
+        memkind_error_message(all_error_code[i], error_message,
+                              MEMKIND_ERROR_MESSAGE_SIZE);
+        EXPECT_TRUE(strncmp(error_message, "<memkind>", strlen("<memkind>")) == 0);
+    }
+}
+
+TEST_F(ErrorMessage, test_TC_MEMKIND_ErrorMsgUndefMesg)
+{
+    char error_message[MEMKIND_ERROR_MESSAGE_SIZE];
+    memkind_error_message(-0xdeadbeef, error_message, MEMKIND_ERROR_MESSAGE_SIZE);
+    EXPECT_TRUE(strncmp(error_message, "<memkind> Undefined error number:",
+                        strlen("<memkind> Undefined error number:")) == 0);
+}
+
diff --git a/test/fragmentation_benchmark_pmem.cpp b/test/fragmentation_benchmark_pmem.cpp
new file mode 100644
index 0000000..a0e9a4d
--- /dev/null
+++ b/test/fragmentation_benchmark_pmem.cpp
@@ -0,0 +1,199 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <iostream>
+#include <chrono>
+#include <cstdio>
+#include <cstring>
+#include <ctime>
+#include <random>
+#include <vector>
+
+#define MB (1024 * 1024)
+#define PRINT_FREQ 100000
+#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
+#define BENCHMARK_LOG "bench_single_thread_%Y%m%d-%H%M"
+
+static size_t block_size [] =
+{8519680, 4325376, 8519680, 4325376, 8519680, 4325376, 8519680, 4325376, 432517, 608478};
+
+static struct memkind *pmem_kind;
+static FILE *log_file;
+
+static const char *const log_tag[] = {
+    "_mem_default.log",
+    "_mem_conservative.log"
+};
+
+static void usage(char *name)
+{
+    fprintf(stderr,
+            "Usage: %s pmem_kind_dir_path pmem_size pmem_policy test_time_limit_in_sec\n",
+            name);
+}
+
+static int fragmentatation_test(size_t pmem_max_size, double test_time)
+{
+    std::random_device rd;
+    std::mt19937 mt(rd());
+    std::uniform_int_distribution<> m_size(0, ARRAY_SIZE(block_size) - 1);
+    char *pmem_str = nullptr;
+    std::vector<char *> pmem_strs;
+    size_t total_size = pmem_max_size;
+    size_t print_iter = 0;
+    size_t total_allocated = 0;
+    double elapsed_seconds;
+    auto start = std::chrono::steady_clock::now();
+
+    do {
+        print_iter++;
+        int index = m_size(mt);
+        size_t size = block_size[index];
+        size_t length = pmem_strs.size() / 2;
+        std::uniform_int_distribution<size_t> m_index(0, length - 1);
+
+        while ((pmem_str = static_cast<char *>(memkind_malloc(pmem_kind,
+                                                              size))) == nullptr) {
+            size_t to_evict = m_index(mt);
+            char *str_to_evict = pmem_strs[to_evict];
+            size_t evict_size = memkind_malloc_usable_size(pmem_kind, str_to_evict);
+            total_allocated -= evict_size;
+
+            if (total_allocated < total_size * 0.1) {
+                fprintf(stderr,"allocated less than 10 percent.\n");
+                return 1;
+            }
+            memkind_free(pmem_kind, str_to_evict);
+            pmem_strs.erase(pmem_strs.begin() + to_evict);
+        }
+        pmem_strs.push_back(pmem_str);
+        total_allocated += memkind_malloc_usable_size(pmem_kind, pmem_str);
+
+        if (print_iter % PRINT_FREQ == 0) {
+            fprintf(log_file,"%f\n", static_cast<double>(total_allocated) / total_size);
+            fflush(stdout);
+        }
+
+        auto finish = std::chrono::steady_clock::now();
+        elapsed_seconds = std::chrono::duration_cast<std::chrono::duration<double>>
+                          (finish - start).count();
+    } while (elapsed_seconds < test_time);
+    return 0;
+}
+
+static int create_pmem(const char *pmem_dir, size_t pmem_size,
+                       memkind_mem_usage_policy policy)
+{
+    int err = 0;
+    if (pmem_size == 0 ) {
+        fprintf(stderr, "Invalid size to pmem kind must be not equal zero.\n");
+        return 1;
+    }
+
+    if (policy > MEMKIND_MEM_USAGE_POLICY_MAX_VALUE) {
+        fprintf(stderr, "Invalid memory usage policy param %u.\n", policy);
+        return 1;
+    }
+
+    memkind_config *pmem_cfg = memkind_config_new();
+    if (!pmem_cfg) {
+        fprintf(stderr, "Unable to create pmem configuration.\n");
+        return 1;
+    }
+
+    memkind_config_set_path(pmem_cfg, pmem_dir);
+    memkind_config_set_size(pmem_cfg, pmem_size);
+    memkind_config_set_memory_usage_policy(pmem_cfg, policy);
+
+    err = memkind_create_pmem_with_config(pmem_cfg, &pmem_kind);
+    memkind_config_delete(pmem_cfg);
+    if (err) {
+        fprintf(stderr, "Unable to create pmem kind.\n");
+        return 1;
+    }
+
+    return 0;
+}
+
+static int create_log_file(memkind_mem_usage_policy policy)
+{
+    char file_name[100] = {'\0'};
+    auto result = std::time(nullptr);
+
+    strftime(file_name, 100, BENCHMARK_LOG, std::localtime(&result));
+    std::strcat(file_name, log_tag[policy]);
+
+    if ((log_file = fopen(file_name, "w+")) == nullptr) {
+        fprintf(stderr, "Cannot create output file %s.\n", file_name);
+        return 1;
+    }
+
+    return 0;
+}
+
+int main(int argc, char *argv[])
+{
+    char *pmem_dir;
+    size_t pmem_size;
+    memkind_mem_usage_policy pmem_policy;
+    double test_time_limit_in_sec;
+    int status = 0;
+    int err = 0;
+
+    if (argc != 5) {
+        usage(argv[0]);
+        return 1;
+    } else {
+        pmem_dir = argv[1];
+        pmem_size = std::stoull(argv[2]) * MB;
+        pmem_policy = static_cast<memkind_mem_usage_policy>(std::stoul(argv[3]));
+        test_time_limit_in_sec = std::stod(argv[4]);
+    }
+
+    err = create_pmem(pmem_dir, pmem_size, pmem_policy);
+    if (err) {
+        fprintf(stderr, "Cannot create pmem.\n");
+        return 1;
+    }
+
+    err = create_log_file(pmem_policy);
+    if (err) {
+        memkind_destroy_kind(pmem_kind);
+        fprintf(stderr, "Cannot create log file.\n");
+        return 1;
+    }
+
+    status = fragmentatation_test(pmem_size, test_time_limit_in_sec);
+
+    err = memkind_destroy_kind(pmem_kind);
+    if (err) {
+        fprintf(stderr, "Unable to destroy pmem kind.\n");
+    }
+
+    fclose(log_file);
+    return status;
+
+}
diff --git a/test/freeing_memory_segfault_test.cpp b/test/freeing_memory_segfault_test.cpp
new file mode 100644
index 0000000..a40cba8
--- /dev/null
+++ b/test/freeing_memory_segfault_test.cpp
@@ -0,0 +1,50 @@
+/*
+* Copyright (C) 2017 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "memkind.h"
+#include "common.h"
+#include <thread>
+
+//This test reproduce segfault when using TBB library.
+class FreeingMemorySegfault: public :: testing::Test
+{
+protected:
+    void SetUp() {}
+    void TearDown() {}
+};
+
+TEST_F(FreeingMemorySegfault,
+       test_TC_MEMKIND_freeing_memory_after_thread_finish)
+{
+    void *ptr = nullptr;
+
+    std::thread t([&] {
+        ptr = memkind_malloc(MEMKIND_DEFAULT, 32);
+        ASSERT_TRUE(ptr != NULL);
+    });
+    t.join();
+
+    memkind_free(0, ptr);
+    SUCCEED();
+}
diff --git a/test/gb_page_tests_bind_policy.cpp b/test/gb_page_tests_bind_policy.cpp
new file mode 100644
index 0000000..9af41bb
--- /dev/null
+++ b/test/gb_page_tests_bind_policy.cpp
@@ -0,0 +1,92 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "hbwmalloc.h"
+#include "common.h"
+#include "trial_generator.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+/* This set of tests are intended to test HBW_PAGESIZE_1GB and
+ * HBW_PAGESIZE_1GB_STRICT flags.
+ */
+class GBPagesTestBindPolicy : public TGTest
+{
+protected:
+    void run(int iterations, size_t alignment, bool psize_strict)
+    {
+        std::vector<void *> addr_to_free;
+        hbw_set_policy(HBW_POLICY_BIND);
+        EXPECT_EQ(HBW_POLICY_BIND, hbw_get_policy());
+
+        for (int i = 0; i < iterations; i++) {
+            void *ptr = NULL;
+            int ret;
+            if (psize_strict) {
+                ret = hbw_posix_memalign_psize(&ptr, alignment, GB, HBW_PAGESIZE_1GB_STRICT);
+            } else {
+                ret = hbw_posix_memalign_psize(&ptr, alignment, GB, HBW_PAGESIZE_1GB);
+            }
+            ASSERT_EQ(0, ret);
+            ASSERT_FALSE(ptr == NULL);
+            ASSERT_EQ(0, hbw_verify_memory_region(ptr, GB, HBW_TOUCH_PAGES));
+            addr_to_free.push_back(ptr);
+        }
+
+        for (int i = 0; i < iterations; i++) {
+            hbw_free(addr_to_free[i]);
+        }
+    }
+
+private:
+    HugePageOrganizer huge_page_organizer = HugePageOrganizer(2250);
+};
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Misalign_Preferred_Bind_Strict_1GB)
+{
+    run(1, 2*GB, true);
+}
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Memalign_Psize_Bind_1GB)
+{
+    run(1, GB, true);
+}
+
+/*
+ * Below tests allocate GB pages incrementally.
+*/
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Memalign_Psize_Bind_2GB)
+{
+    run(2, GB, false);
+}
+
+TEST_F(GBPagesTestBindPolicy,
+       test_TC_MEMKIND_GBPages_ext_HBW_Memalign_Psize_Strict_Bind_3GB)
+{
+    run(3, GB, true);
+}
diff --git a/test/get_arena_test.cpp b/test/get_arena_test.cpp
new file mode 100644
index 0000000..e1c125b
--- /dev/null
+++ b/test/get_arena_test.cpp
@@ -0,0 +1,91 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_arena.h>
+
+#include <algorithm>
+#include <vector>
+#include <gtest/gtest.h>
+#include <omp.h>
+#include <pthread.h>
+
+class GetArenaTest: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+};
+
+bool uint_comp(unsigned int a, unsigned int b)
+{
+    return (a < b);
+}
+
+TEST_F(GetArenaTest, test_TC_MEMKIND_ThreadHash)
+{
+#ifdef _OPENMP
+    int num_threads = omp_get_max_threads();
+    std::vector<unsigned int> arena_idx(num_threads);
+    unsigned int thread_idx, idx;
+    int err = 0;
+    size_t size = 0;
+    int i;
+    unsigned max_collisions, collisions;
+    const unsigned collisions_limit = 5;
+
+    //Initialize kind
+    memkind_malloc(MEMKIND_HBW, 0);
+
+    #pragma omp parallel shared(arena_idx) private(thread_idx)
+    {
+        thread_idx = omp_get_thread_num();
+        err = memkind_thread_get_arena(MEMKIND_HBW, &(arena_idx[thread_idx]), size);
+    }
+    ASSERT_TRUE(err == 0);
+    std::sort(arena_idx.begin(), arena_idx.end(), uint_comp);
+    idx = arena_idx[0];
+    collisions = 0;
+    max_collisions = 0;
+    for (i = 1; i < num_threads; ++i) {
+        if (arena_idx[i] == idx) {
+            collisions++;
+        } else {
+            if (collisions > max_collisions) {
+                max_collisions = collisions;
+            }
+            idx = arena_idx[i];
+            collisions = 0;
+        }
+    }
+    EXPECT_LE(max_collisions, collisions_limit);
+    RecordProperty("max_collisions", max_collisions);
+#else
+    std::cout << "[ SKIPPED ] Feature OPENMP not supported" << std::endl;
+#endif
+}
diff --git a/test/gtest_fused/gtest/gtest-all.cc b/test/gtest_fused/gtest/gtest-all.cc
new file mode 100644
index 0000000..1120ce8
--- /dev/null
+++ b/test/gtest_fused/gtest/gtest-all.cc
@@ -0,0 +1,10495 @@
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+// Google C++ Testing Framework (Google Test)
+//
+// Sometimes it's desirable to build Google Test by compiling a single file.
+// This file serves this purpose.
+
+// This line ensures that gtest.h can be compiled on its own, even
+// when it's fused.
+#include "gtest.h"
+
+// The following lines pull in the real gtest *.cc files.
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Utilities for testing Google Test itself and code that uses Google Test
+// (e.g. frameworks built on top of Google Test).
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+#define GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+
+
+namespace testing {
+
+// This helper class can be used to mock out Google Test failure reporting
+// so that we can test Google Test or code that builds on Google Test.
+//
+// An object of this class appends a TestPartResult object to the
+// TestPartResultArray object given in the constructor whenever a Google Test
+// failure is reported. It can either intercept only failures that are
+// generated in the same thread that created this object or it can intercept
+// all generated failures. The scope of this mock object can be controlled with
+// the second argument to the two arguments constructor.
+class GTEST_API_ ScopedFakeTestPartResultReporter
+    : public TestPartResultReporterInterface {
+ public:
+  // The two possible mocking modes of this object.
+  enum InterceptMode {
+    INTERCEPT_ONLY_CURRENT_THREAD,  // Intercepts only thread local failures.
+    INTERCEPT_ALL_THREADS           // Intercepts all failures.
+  };
+
+  // The c'tor sets this object as the test part result reporter used
+  // by Google Test.  The 'result' parameter specifies where to report the
+  // results. This reporter will only catch failures generated in the current
+  // thread. DEPRECATED
+  explicit ScopedFakeTestPartResultReporter(TestPartResultArray* result);
+
+  // Same as above, but you can choose the interception scope of this object.
+  ScopedFakeTestPartResultReporter(InterceptMode intercept_mode,
+                                   TestPartResultArray* result);
+
+  // The d'tor restores the previous test part result reporter.
+  virtual ~ScopedFakeTestPartResultReporter();
+
+  // Appends the TestPartResult object to the TestPartResultArray
+  // received in the constructor.
+  //
+  // This method is from the TestPartResultReporterInterface
+  // interface.
+  virtual void ReportTestPartResult(const TestPartResult& result);
+ private:
+  void Init();
+
+  const InterceptMode intercept_mode_;
+  TestPartResultReporterInterface* old_reporter_;
+  TestPartResultArray* const result_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedFakeTestPartResultReporter);
+};
+
+namespace internal {
+
+// A helper class for implementing EXPECT_FATAL_FAILURE() and
+// EXPECT_NONFATAL_FAILURE().  Its destructor verifies that the given
+// TestPartResultArray contains exactly one failure that has the given
+// type and contains the given substring.  If that's not the case, a
+// non-fatal failure will be generated.
+class GTEST_API_ SingleFailureChecker {
+ public:
+  // The constructor remembers the arguments.
+  SingleFailureChecker(const TestPartResultArray* results,
+                       TestPartResult::Type type, const std::string& substr);
+  ~SingleFailureChecker();
+ private:
+  const TestPartResultArray* const results_;
+  const TestPartResult::Type type_;
+  const std::string substr_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(SingleFailureChecker);
+};
+
+}  // namespace internal
+
+}  // namespace testing
+
+// A set of macros for testing Google Test assertions or code that's expected
+// to generate Google Test fatal failures.  It verifies that the given
+// statement will cause exactly one fatal Google Test failure with 'substr'
+// being part of the failure message.
+//
+// There are two different versions of this macro. EXPECT_FATAL_FAILURE only
+// affects and considers failures generated in the current thread and
+// EXPECT_FATAL_FAILURE_ON_ALL_THREADS does the same but for all threads.
+//
+// The verification of the assertion is done correctly even when the statement
+// throws an exception or aborts the current function.
+//
+// Known restrictions:
+//   - 'statement' cannot reference local non-static variables or
+//     non-static members of the current object.
+//   - 'statement' cannot return a value.
+//   - You cannot stream a failure message to this macro.
+//
+// Note that even though the implementations of the following two
+// macros are much alike, we cannot refactor them to use a common
+// helper macro, due to some peculiarity in how the preprocessor
+// works.  The AcceptsMacroThatExpandsToUnprotectedComma test in
+// gtest_unittest.cc will fail to compile if we do that.
+#define EXPECT_FATAL_FAILURE(statement, substr) \
+  do { \
+    class GTestExpectFatalFailureHelper {\
+     public:\
+      static void Execute() { statement; }\
+    };\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kFatalFailure, (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter:: \
+          INTERCEPT_ONLY_CURRENT_THREAD, &gtest_failures);\
+      GTestExpectFatalFailureHelper::Execute();\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+#define EXPECT_FATAL_FAILURE_ON_ALL_THREADS(statement, substr) \
+  do { \
+    class GTestExpectFatalFailureHelper {\
+     public:\
+      static void Execute() { statement; }\
+    };\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kFatalFailure, (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter:: \
+          INTERCEPT_ALL_THREADS, &gtest_failures);\
+      GTestExpectFatalFailureHelper::Execute();\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+// A macro for testing Google Test assertions or code that's expected to
+// generate Google Test non-fatal failures.  It asserts that the given
+// statement will cause exactly one non-fatal Google Test failure with 'substr'
+// being part of the failure message.
+//
+// There are two different versions of this macro. EXPECT_NONFATAL_FAILURE only
+// affects and considers failures generated in the current thread and
+// EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS does the same but for all threads.
+//
+// 'statement' is allowed to reference local variables and members of
+// the current object.
+//
+// The verification of the assertion is done correctly even when the statement
+// throws an exception or aborts the current function.
+//
+// Known restrictions:
+//   - You cannot stream a failure message to this macro.
+//
+// Note that even though the implementations of the following two
+// macros are much alike, we cannot refactor them to use a common
+// helper macro, due to some peculiarity in how the preprocessor
+// works.  If we do that, the code won't compile when the user gives
+// EXPECT_NONFATAL_FAILURE() a statement that contains a macro that
+// expands to code containing an unprotected comma.  The
+// AcceptsMacroThatExpandsToUnprotectedComma test in gtest_unittest.cc
+// catches that.
+//
+// For the same reason, we have to write
+//   if (::testing::internal::AlwaysTrue()) { statement; }
+// instead of
+//   GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement)
+// to avoid an MSVC warning on unreachable code.
+#define EXPECT_NONFATAL_FAILURE(statement, substr) \
+  do {\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kNonFatalFailure, \
+        (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter:: \
+          INTERCEPT_ONLY_CURRENT_THREAD, &gtest_failures);\
+      if (::testing::internal::AlwaysTrue()) { statement; }\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+#define EXPECT_NONFATAL_FAILURE_ON_ALL_THREADS(statement, substr) \
+  do {\
+    ::testing::TestPartResultArray gtest_failures;\
+    ::testing::internal::SingleFailureChecker gtest_checker(\
+        &gtest_failures, ::testing::TestPartResult::kNonFatalFailure, \
+        (substr));\
+    {\
+      ::testing::ScopedFakeTestPartResultReporter gtest_reporter(\
+          ::testing::ScopedFakeTestPartResultReporter::INTERCEPT_ALL_THREADS, \
+          &gtest_failures);\
+      if (::testing::internal::AlwaysTrue()) { statement; }\
+    }\
+  } while (::testing::internal::AlwaysFalse())
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_SPI_H_
+
+#include <ctype.h>
+#include <math.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <time.h>
+#include <wchar.h>
+#include <wctype.h>
+
+#include <algorithm>
+#include <iomanip>
+#include <limits>
+#include <list>
+#include <map>
+#include <ostream>  // NOLINT
+#include <sstream>
+#include <vector>
+
+#if GTEST_OS_LINUX
+
+// TODO(kenton@google.com): Use autoconf to detect availability of
+// gettimeofday().
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+
+# include <fcntl.h>  // NOLINT
+# include <limits.h>  // NOLINT
+# include <sched.h>  // NOLINT
+// Declares vsnprintf().  This header is not available on Windows.
+# include <strings.h>  // NOLINT
+# include <sys/mman.h>  // NOLINT
+# include <sys/time.h>  // NOLINT
+# include <unistd.h>  // NOLINT
+# include <string>
+
+#elif GTEST_OS_SYMBIAN
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+# include <sys/time.h>  // NOLINT
+
+#elif GTEST_OS_ZOS
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+# include <sys/time.h>  // NOLINT
+
+// On z/OS we additionally need strings.h for strcasecmp.
+# include <strings.h>  // NOLINT
+
+#elif GTEST_OS_WINDOWS_MOBILE  // We are on Windows CE.
+
+# include <windows.h>  // NOLINT
+# undef min
+
+#elif GTEST_OS_WINDOWS  // We are on Windows proper.
+
+# include <io.h>  // NOLINT
+# include <sys/timeb.h>  // NOLINT
+# include <sys/types.h>  // NOLINT
+# include <sys/stat.h>  // NOLINT
+
+# if GTEST_OS_WINDOWS_MINGW
+// MinGW has gettimeofday() but not _ftime64().
+// TODO(kenton@google.com): Use autoconf to detect availability of
+//   gettimeofday().
+// TODO(kenton@google.com): There are other ways to get the time on
+//   Windows, like GetTickCount() or GetSystemTimeAsFileTime().  MinGW
+//   supports these.  consider using them instead.
+#  define GTEST_HAS_GETTIMEOFDAY_ 1
+#  include <sys/time.h>  // NOLINT
+# endif  // GTEST_OS_WINDOWS_MINGW
+
+// cpplint thinks that the header is already included, so we want to
+// silence it.
+# include <windows.h>  // NOLINT
+# undef min
+
+#else
+
+// Assume other platforms have gettimeofday().
+// TODO(kenton@google.com): Use autoconf to detect availability of
+//   gettimeofday().
+# define GTEST_HAS_GETTIMEOFDAY_ 1
+
+// cpplint thinks that the header is already included, so we want to
+// silence it.
+# include <sys/time.h>  // NOLINT
+# include <unistd.h>  // NOLINT
+
+#endif  // GTEST_OS_LINUX
+
+#if GTEST_HAS_EXCEPTIONS
+# include <stdexcept>
+#endif
+
+#if GTEST_CAN_STREAM_RESULTS_
+# include <arpa/inet.h>  // NOLINT
+# include <netdb.h>  // NOLINT
+# include <sys/socket.h>  // NOLINT
+# include <sys/types.h>  // NOLINT
+#endif
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick is to
+// prevent a user from accidentally including gtest-internal-inl.h in
+// his code.
+#define GTEST_IMPLEMENTATION_ 1
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// Utility functions and classes used by the Google C++ testing framework.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// This file contains purely Google Test's internal implementation.  Please
+// DO NOT #INCLUDE IT IN A USER PROGRAM.
+
+#ifndef GTEST_SRC_GTEST_INTERNAL_INL_H_
+#define GTEST_SRC_GTEST_INTERNAL_INL_H_
+
+// GTEST_IMPLEMENTATION_ is defined to 1 iff the current translation unit is
+// part of Google Test's implementation; otherwise it's undefined.
+#if !GTEST_IMPLEMENTATION_
+// If this file is included from the user's code, just say no.
+# error "gtest-internal-inl.h is part of Google Test's internal implementation."
+# error "It must not be included except by Google Test itself."
+#endif  // GTEST_IMPLEMENTATION_
+
+#ifndef _WIN32_WCE
+# include <errno.h>
+#endif  // !_WIN32_WCE
+#include <stddef.h>
+#include <stdlib.h>  // For strtoll/_strtoul64/malloc/free.
+#include <string.h>  // For memmove.
+
+#include <algorithm>
+#include <string>
+#include <vector>
+
+
+#if GTEST_CAN_STREAM_RESULTS_
+# include <arpa/inet.h>  // NOLINT
+# include <netdb.h>  // NOLINT
+#endif
+
+#if GTEST_OS_WINDOWS
+# include <windows.h>  // NOLINT
+#endif  // GTEST_OS_WINDOWS
+
+
+namespace testing {
+
+// Declares the flags.
+//
+// We don't want the users to modify this flag in the code, but want
+// Google Test's own unit tests to be able to access it. Therefore we
+// declare it here as opposed to in gtest.h.
+GTEST_DECLARE_bool_(death_test_use_fork);
+
+namespace internal {
+
+// The value of GetTestTypeId() as seen from within the Google Test
+// library.  This is solely for testing GetTestTypeId().
+GTEST_API_ extern const TypeId kTestTypeIdInGoogleTest;
+
+// Names of the flags (needed for parsing Google Test flags).
+const char kAlsoRunDisabledTestsFlag[] = "also_run_disabled_tests";
+const char kBreakOnFailureFlag[] = "break_on_failure";
+const char kCatchExceptionsFlag[] = "catch_exceptions";
+const char kColorFlag[] = "color";
+const char kFilterFlag[] = "filter";
+const char kListTestsFlag[] = "list_tests";
+const char kOutputFlag[] = "output";
+const char kPrintTimeFlag[] = "print_time";
+const char kRandomSeedFlag[] = "random_seed";
+const char kRepeatFlag[] = "repeat";
+const char kShuffleFlag[] = "shuffle";
+const char kStackTraceDepthFlag[] = "stack_trace_depth";
+const char kStreamResultToFlag[] = "stream_result_to";
+const char kThrowOnFailureFlag[] = "throw_on_failure";
+const char kFlagfileFlag[] = "flagfile";
+
+// A valid random seed must be in [1, kMaxRandomSeed].
+const int kMaxRandomSeed = 99999;
+
+// g_help_flag is true iff the --help flag or an equivalent form is
+// specified on the command line.
+GTEST_API_ extern bool g_help_flag;
+
+// Returns the current time in milliseconds.
+GTEST_API_ TimeInMillis GetTimeInMillis();
+
+// Returns true iff Google Test should use colors in the output.
+GTEST_API_ bool ShouldUseColor(bool stdout_is_tty);
+
+// Formats the given time in milliseconds as seconds.
+GTEST_API_ std::string FormatTimeInMillisAsSeconds(TimeInMillis ms);
+
+// Converts the given time in milliseconds to a date string in the ISO 8601
+// format, without the timezone information.  N.B.: due to the use the
+// non-reentrant localtime() function, this function is not thread safe.  Do
+// not use it in any code that can be called from multiple threads.
+GTEST_API_ std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms);
+
+// Parses a string for an Int32 flag, in the form of "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+GTEST_API_ bool ParseInt32Flag(
+    const char* str, const char* flag, Int32* value);
+
+// Returns a random seed in range [1, kMaxRandomSeed] based on the
+// given --gtest_random_seed flag value.
+inline int GetRandomSeedFromFlag(Int32 random_seed_flag) {
+  const unsigned int raw_seed = (random_seed_flag == 0) ?
+      static_cast<unsigned int>(GetTimeInMillis()) :
+      static_cast<unsigned int>(random_seed_flag);
+
+  // Normalizes the actual seed to range [1, kMaxRandomSeed] such that
+  // it's easy to type.
+  const int normalized_seed =
+      static_cast<int>((raw_seed - 1U) %
+                       static_cast<unsigned int>(kMaxRandomSeed)) + 1;
+  return normalized_seed;
+}
+
+// Returns the first valid random seed after 'seed'.  The behavior is
+// undefined if 'seed' is invalid.  The seed after kMaxRandomSeed is
+// considered to be 1.
+inline int GetNextRandomSeed(int seed) {
+  GTEST_CHECK_(1 <= seed && seed <= kMaxRandomSeed)
+      << "Invalid random seed " << seed << " - must be in [1, "
+      << kMaxRandomSeed << "].";
+  const int next_seed = seed + 1;
+  return (next_seed > kMaxRandomSeed) ? 1 : next_seed;
+}
+
+// This class saves the values of all Google Test flags in its c'tor, and
+// restores them in its d'tor.
+class GTestFlagSaver {
+ public:
+  // The c'tor.
+  GTestFlagSaver() {
+    also_run_disabled_tests_ = GTEST_FLAG(also_run_disabled_tests);
+    break_on_failure_ = GTEST_FLAG(break_on_failure);
+    catch_exceptions_ = GTEST_FLAG(catch_exceptions);
+    color_ = GTEST_FLAG(color);
+    death_test_style_ = GTEST_FLAG(death_test_style);
+    death_test_use_fork_ = GTEST_FLAG(death_test_use_fork);
+    filter_ = GTEST_FLAG(filter);
+    internal_run_death_test_ = GTEST_FLAG(internal_run_death_test);
+    list_tests_ = GTEST_FLAG(list_tests);
+    output_ = GTEST_FLAG(output);
+    print_time_ = GTEST_FLAG(print_time);
+    random_seed_ = GTEST_FLAG(random_seed);
+    repeat_ = GTEST_FLAG(repeat);
+    shuffle_ = GTEST_FLAG(shuffle);
+    stack_trace_depth_ = GTEST_FLAG(stack_trace_depth);
+    stream_result_to_ = GTEST_FLAG(stream_result_to);
+    throw_on_failure_ = GTEST_FLAG(throw_on_failure);
+  }
+
+  // The d'tor is not virtual.  DO NOT INHERIT FROM THIS CLASS.
+  ~GTestFlagSaver() {
+    GTEST_FLAG(also_run_disabled_tests) = also_run_disabled_tests_;
+    GTEST_FLAG(break_on_failure) = break_on_failure_;
+    GTEST_FLAG(catch_exceptions) = catch_exceptions_;
+    GTEST_FLAG(color) = color_;
+    GTEST_FLAG(death_test_style) = death_test_style_;
+    GTEST_FLAG(death_test_use_fork) = death_test_use_fork_;
+    GTEST_FLAG(filter) = filter_;
+    GTEST_FLAG(internal_run_death_test) = internal_run_death_test_;
+    GTEST_FLAG(list_tests) = list_tests_;
+    GTEST_FLAG(output) = output_;
+    GTEST_FLAG(print_time) = print_time_;
+    GTEST_FLAG(random_seed) = random_seed_;
+    GTEST_FLAG(repeat) = repeat_;
+    GTEST_FLAG(shuffle) = shuffle_;
+    GTEST_FLAG(stack_trace_depth) = stack_trace_depth_;
+    GTEST_FLAG(stream_result_to) = stream_result_to_;
+    GTEST_FLAG(throw_on_failure) = throw_on_failure_;
+  }
+
+ private:
+  // Fields for saving the original values of flags.
+  bool also_run_disabled_tests_;
+  bool break_on_failure_;
+  bool catch_exceptions_;
+  std::string color_;
+  std::string death_test_style_;
+  bool death_test_use_fork_;
+  std::string filter_;
+  std::string internal_run_death_test_;
+  bool list_tests_;
+  std::string output_;
+  bool print_time_;
+  internal::Int32 random_seed_;
+  internal::Int32 repeat_;
+  bool shuffle_;
+  internal::Int32 stack_trace_depth_;
+  std::string stream_result_to_;
+  bool throw_on_failure_;
+} GTEST_ATTRIBUTE_UNUSED_;
+
+// Converts a Unicode code point to a narrow string in UTF-8 encoding.
+// code_point parameter is of type UInt32 because wchar_t may not be
+// wide enough to contain a code point.
+// If the code_point is not a valid Unicode code point
+// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted
+// to "(Invalid Unicode 0xXXXXXXXX)".
+GTEST_API_ std::string CodePointToUtf8(UInt32 code_point);
+
+// Converts a wide string to a narrow string in UTF-8 encoding.
+// The wide string is assumed to have the following encoding:
+//   UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS)
+//   UTF-32 if sizeof(wchar_t) == 4 (on Linux)
+// Parameter str points to a null-terminated wide string.
+// Parameter num_chars may additionally limit the number
+// of wchar_t characters processed. -1 is used when the entire string
+// should be processed.
+// If the string contains code points that are not valid Unicode code points
+// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding
+// and contains invalid UTF-16 surrogate pairs, values in those pairs
+// will be encoded as individual Unicode characters from Basic Normal Plane.
+GTEST_API_ std::string WideStringToUtf8(const wchar_t* str, int num_chars);
+
+// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file
+// if the variable is present. If a file already exists at this location, this
+// function will write over it. If the variable is present, but the file cannot
+// be created, prints an error and exits.
+void WriteToShardStatusFileIfNeeded();
+
+// Checks whether sharding is enabled by examining the relevant
+// environment variable values. If the variables are present,
+// but inconsistent (e.g., shard_index >= total_shards), prints
+// an error and exits. If in_subprocess_for_death_test, sharding is
+// disabled because it must only be applied to the original test
+// process. Otherwise, we could filter out death tests we intended to execute.
+GTEST_API_ bool ShouldShard(const char* total_shards_str,
+                            const char* shard_index_str,
+                            bool in_subprocess_for_death_test);
+
+// Parses the environment variable var as an Int32. If it is unset,
+// returns default_val. If it is not an Int32, prints an error and
+// and aborts.
+GTEST_API_ Int32 Int32FromEnvOrDie(const char* env_var, Int32 default_val);
+
+// Given the total number of shards, the shard index, and the test id,
+// returns true iff the test should be run on this shard. The test id is
+// some arbitrary but unique non-negative integer assigned to each test
+// method. Assumes that 0 <= shard_index < total_shards.
+GTEST_API_ bool ShouldRunTestOnShard(
+    int total_shards, int shard_index, int test_id);
+
+// STL container utilities.
+
+// Returns the number of elements in the given container that satisfy
+// the given predicate.
+template <class Container, typename Predicate>
+inline int CountIf(const Container& c, Predicate predicate) {
+  // Implemented as an explicit loop since std::count_if() in libCstd on
+  // Solaris has a non-standard signature.
+  int count = 0;
+  for (typename Container::const_iterator it = c.begin(); it != c.end(); ++it) {
+    if (predicate(*it))
+      ++count;
+  }
+  return count;
+}
+
+// Applies a function/functor to each element in the container.
+template <class Container, typename Functor>
+void ForEach(const Container& c, Functor functor) {
+  std::for_each(c.begin(), c.end(), functor);
+}
+
+// Returns the i-th element of the vector, or default_value if i is not
+// in range [0, v.size()).
+template <typename E>
+inline E GetElementOr(const std::vector<E>& v, int i, E default_value) {
+  return (i < 0 || i >= static_cast<int>(v.size())) ? default_value : v[i];
+}
+
+// Performs an in-place shuffle of a range of the vector's elements.
+// 'begin' and 'end' are element indices as an STL-style range;
+// i.e. [begin, end) are shuffled, where 'end' == size() means to
+// shuffle to the end of the vector.
+template <typename E>
+void ShuffleRange(internal::Random* random, int begin, int end,
+                  std::vector<E>* v) {
+  const int size = static_cast<int>(v->size());
+  GTEST_CHECK_(0 <= begin && begin <= size)
+      << "Invalid shuffle range start " << begin << ": must be in range [0, "
+      << size << "].";
+  GTEST_CHECK_(begin <= end && end <= size)
+      << "Invalid shuffle range finish " << end << ": must be in range ["
+      << begin << ", " << size << "].";
+
+  // Fisher-Yates shuffle, from
+  // http://en.wikipedia.org/wiki/Fisher-Yates_shuffle
+  for (int range_width = end - begin; range_width >= 2; range_width--) {
+    const int last_in_range = begin + range_width - 1;
+    const int selected = begin + random->Generate(range_width);
+    std::swap((*v)[selected], (*v)[last_in_range]);
+  }
+}
+
+// Performs an in-place shuffle of the vector's elements.
+template <typename E>
+inline void Shuffle(internal::Random* random, std::vector<E>* v) {
+  ShuffleRange(random, 0, static_cast<int>(v->size()), v);
+}
+
+// A function for deleting an object.  Handy for being used as a
+// functor.
+template <typename T>
+static void Delete(T* x) {
+  delete x;
+}
+
+// A predicate that checks the key of a TestProperty against a known key.
+//
+// TestPropertyKeyIs is copyable.
+class TestPropertyKeyIs {
+ public:
+  // Constructor.
+  //
+  // TestPropertyKeyIs has NO default constructor.
+  explicit TestPropertyKeyIs(const std::string& key) : key_(key) {}
+
+  // Returns true iff the test name of test property matches on key_.
+  bool operator()(const TestProperty& test_property) const {
+    return test_property.key() == key_;
+  }
+
+ private:
+  std::string key_;
+};
+
+// Class UnitTestOptions.
+//
+// This class contains functions for processing options the user
+// specifies when running the tests.  It has only static members.
+//
+// In most cases, the user can specify an option using either an
+// environment variable or a command line flag.  E.g. you can set the
+// test filter using either GTEST_FILTER or --gtest_filter.  If both
+// the variable and the flag are present, the latter overrides the
+// former.
+class GTEST_API_ UnitTestOptions {
+ public:
+  // Functions for processing the gtest_output flag.
+
+  // Returns the output format, or "" for normal printed output.
+  static std::string GetOutputFormat();
+
+  // Returns the absolute path of the requested output file, or the
+  // default (test_detail.xml in the original working directory) if
+  // none was explicitly specified.
+  static std::string GetAbsolutePathToOutputFile();
+
+  // Functions for processing the gtest_filter flag.
+
+  // Returns true iff the wildcard pattern matches the string.  The
+  // first ':' or '\0' character in pattern marks the end of it.
+  //
+  // This recursive algorithm isn't very efficient, but is clear and
+  // works well enough for matching test names, which are short.
+  static bool PatternMatchesString(const char *pattern, const char *str);
+
+  // Returns true iff the user-specified filter matches the test case
+  // name and the test name.
+  static bool FilterMatchesTest(const std::string &test_case_name,
+                                const std::string &test_name);
+
+#if GTEST_OS_WINDOWS
+  // Function for supporting the gtest_catch_exception flag.
+
+  // Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the
+  // given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise.
+  // This function is useful as an __except condition.
+  static int GTestShouldProcessSEH(DWORD exception_code);
+#endif  // GTEST_OS_WINDOWS
+
+  // Returns true if "name" matches the ':' separated list of glob-style
+  // filters in "filter".
+  static bool MatchesFilter(const std::string& name, const char* filter);
+};
+
+// Returns the current application's name, removing directory path if that
+// is present.  Used by UnitTestOptions::GetOutputFile.
+GTEST_API_ FilePath GetCurrentExecutableName();
+
+// The role interface for getting the OS stack trace as a string.
+class OsStackTraceGetterInterface {
+ public:
+  OsStackTraceGetterInterface() {}
+  virtual ~OsStackTraceGetterInterface() {}
+
+  // Returns the current OS stack trace as an std::string.  Parameters:
+  //
+  //   max_depth  - the maximum number of stack frames to be included
+  //                in the trace.
+  //   skip_count - the number of top frames to be skipped; doesn't count
+  //                against max_depth.
+  virtual std::string CurrentStackTrace(int max_depth, int skip_count) = 0;
+
+  // UponLeavingGTest() should be called immediately before Google Test calls
+  // user code. It saves some information about the current stack that
+  // CurrentStackTrace() will use to find and hide Google Test stack frames.
+  virtual void UponLeavingGTest() = 0;
+
+  // This string is inserted in place of stack frames that are part of
+  // Google Test's implementation.
+  static const char* const kElidedFramesMarker;
+
+ private:
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetterInterface);
+};
+
+// A working implementation of the OsStackTraceGetterInterface interface.
+class OsStackTraceGetter : public OsStackTraceGetterInterface {
+ public:
+  OsStackTraceGetter() {}
+
+  virtual std::string CurrentStackTrace(int max_depth, int skip_count);
+  virtual void UponLeavingGTest();
+
+ private:
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(OsStackTraceGetter);
+};
+
+// Information about a Google Test trace point.
+struct TraceInfo {
+  const char* file;
+  int line;
+  std::string message;
+};
+
+// This is the default global test part result reporter used in UnitTestImpl.
+// This class should only be used by UnitTestImpl.
+class DefaultGlobalTestPartResultReporter
+  : public TestPartResultReporterInterface {
+ public:
+  explicit DefaultGlobalTestPartResultReporter(UnitTestImpl* unit_test);
+  // Implements the TestPartResultReporterInterface. Reports the test part
+  // result in the current test.
+  virtual void ReportTestPartResult(const TestPartResult& result);
+
+ private:
+  UnitTestImpl* const unit_test_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultGlobalTestPartResultReporter);
+};
+
+// This is the default per thread test part result reporter used in
+// UnitTestImpl. This class should only be used by UnitTestImpl.
+class DefaultPerThreadTestPartResultReporter
+    : public TestPartResultReporterInterface {
+ public:
+  explicit DefaultPerThreadTestPartResultReporter(UnitTestImpl* unit_test);
+  // Implements the TestPartResultReporterInterface. The implementation just
+  // delegates to the current global test part result reporter of *unit_test_.
+  virtual void ReportTestPartResult(const TestPartResult& result);
+
+ private:
+  UnitTestImpl* const unit_test_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultPerThreadTestPartResultReporter);
+};
+
+// The private implementation of the UnitTest class.  We don't protect
+// the methods under a mutex, as this class is not accessible by a
+// user and the UnitTest class that delegates work to this class does
+// proper locking.
+class GTEST_API_ UnitTestImpl {
+ public:
+  explicit UnitTestImpl(UnitTest* parent);
+  virtual ~UnitTestImpl();
+
+  // There are two different ways to register your own TestPartResultReporter.
+  // You can register your own repoter to listen either only for test results
+  // from the current thread or for results from all threads.
+  // By default, each per-thread test result repoter just passes a new
+  // TestPartResult to the global test result reporter, which registers the
+  // test part result for the currently running test.
+
+  // Returns the global test part result reporter.
+  TestPartResultReporterInterface* GetGlobalTestPartResultReporter();
+
+  // Sets the global test part result reporter.
+  void SetGlobalTestPartResultReporter(
+      TestPartResultReporterInterface* reporter);
+
+  // Returns the test part result reporter for the current thread.
+  TestPartResultReporterInterface* GetTestPartResultReporterForCurrentThread();
+
+  // Sets the test part result reporter for the current thread.
+  void SetTestPartResultReporterForCurrentThread(
+      TestPartResultReporterInterface* reporter);
+
+  // Gets the number of successful test cases.
+  int successful_test_case_count() const;
+
+  // Gets the number of failed test cases.
+  int failed_test_case_count() const;
+
+  // Gets the number of all test cases.
+  int total_test_case_count() const;
+
+  // Gets the number of all test cases that contain at least one test
+  // that should run.
+  int test_case_to_run_count() const;
+
+  // Gets the number of successful tests.
+  int successful_test_count() const;
+
+  // Gets the number of skipped tests.
+  int skipped_test_count() const;
+
+  // Gets the number of failed tests.
+  int failed_test_count() const;
+
+  // Gets the number of disabled tests that will be reported in the XML report.
+  int reportable_disabled_test_count() const;
+
+  // Gets the number of disabled tests.
+  int disabled_test_count() const;
+
+  // Gets the number of tests to be printed in the XML report.
+  int reportable_test_count() const;
+
+  // Gets the number of all tests.
+  int total_test_count() const;
+
+  // Gets the number of tests that should run.
+  int test_to_run_count() const;
+
+  // Gets the time of the test program start, in ms from the start of the
+  // UNIX epoch.
+  TimeInMillis start_timestamp() const { return start_timestamp_; }
+
+  // Gets the elapsed time, in milliseconds.
+  TimeInMillis elapsed_time() const { return elapsed_time_; }
+
+  // Returns true iff the unit test passed (i.e. all test cases passed).
+  bool Passed() const { return !Failed(); }
+
+  // Returns true iff the unit test failed (i.e. some test case failed
+  // or something outside of all tests failed).
+  bool Failed() const {
+    return failed_test_case_count() > 0 || ad_hoc_test_result()->Failed();
+  }
+
+  // Gets the i-th test case among all the test cases. i can range from 0 to
+  // total_test_case_count() - 1. If i is not in that range, returns NULL.
+  const TestCase* GetTestCase(int i) const {
+    const int index = GetElementOr(test_case_indices_, i, -1);
+    return index < 0 ? NULL : test_cases_[i];
+  }
+
+  // Gets the i-th test case among all the test cases. i can range from 0 to
+  // total_test_case_count() - 1. If i is not in that range, returns NULL.
+  TestCase* GetMutableTestCase(int i) {
+    const int index = GetElementOr(test_case_indices_, i, -1);
+    return index < 0 ? NULL : test_cases_[index];
+  }
+
+  // Provides access to the event listener list.
+  TestEventListeners* listeners() { return &listeners_; }
+
+  // Returns the TestResult for the test that's currently running, or
+  // the TestResult for the ad hoc test if no test is running.
+  TestResult* current_test_result();
+
+  // Returns the TestResult for the ad hoc test.
+  const TestResult* ad_hoc_test_result() const { return &ad_hoc_test_result_; }
+
+  // Sets the OS stack trace getter.
+  //
+  // Does nothing if the input and the current OS stack trace getter
+  // are the same; otherwise, deletes the old getter and makes the
+  // input the current getter.
+  void set_os_stack_trace_getter(OsStackTraceGetterInterface* getter);
+
+  // Returns the current OS stack trace getter if it is not NULL;
+  // otherwise, creates an OsStackTraceGetter, makes it the current
+  // getter, and returns it.
+  OsStackTraceGetterInterface* os_stack_trace_getter();
+
+  // Returns the current OS stack trace as an std::string.
+  //
+  // The maximum number of stack frames to be included is specified by
+  // the gtest_stack_trace_depth flag.  The skip_count parameter
+  // specifies the number of top frames to be skipped, which doesn't
+  // count against the number of frames to be included.
+  //
+  // For example, if Foo() calls Bar(), which in turn calls
+  // CurrentOsStackTraceExceptTop(1), Foo() will be included in the
+  // trace but Bar() and CurrentOsStackTraceExceptTop() won't.
+  std::string CurrentOsStackTraceExceptTop(int skip_count) GTEST_NO_INLINE_;
+
+  // Finds and returns a TestCase with the given name.  If one doesn't
+  // exist, creates one and returns it.
+  //
+  // Arguments:
+  //
+  //   test_case_name: name of the test case
+  //   type_param:     the name of the test's type parameter, or NULL if
+  //                   this is not a typed or a type-parameterized test.
+  //   set_up_tc:      pointer to the function that sets up the test case
+  //   tear_down_tc:   pointer to the function that tears down the test case
+  TestCase* GetTestCase(const char* test_case_name,
+                        const char* type_param,
+                        Test::SetUpTestCaseFunc set_up_tc,
+                        Test::TearDownTestCaseFunc tear_down_tc);
+
+  // Adds a TestInfo to the unit test.
+  //
+  // Arguments:
+  //
+  //   set_up_tc:    pointer to the function that sets up the test case
+  //   tear_down_tc: pointer to the function that tears down the test case
+  //   test_info:    the TestInfo object
+  void AddTestInfo(Test::SetUpTestCaseFunc set_up_tc,
+                   Test::TearDownTestCaseFunc tear_down_tc,
+                   TestInfo* test_info) {
+    // In order to support thread-safe death tests, we need to
+    // remember the original working directory when the test program
+    // was first invoked.  We cannot do this in RUN_ALL_TESTS(), as
+    // the user may have changed the current directory before calling
+    // RUN_ALL_TESTS().  Therefore we capture the current directory in
+    // AddTestInfo(), which is called to register a TEST or TEST_F
+    // before main() is reached.
+    if (original_working_dir_.IsEmpty()) {
+      original_working_dir_.Set(FilePath::GetCurrentDir());
+      GTEST_CHECK_(!original_working_dir_.IsEmpty())
+          << "Failed to get the current working directory.";
+    }
+
+    GetTestCase(test_info->test_case_name(),
+                test_info->type_param(),
+                set_up_tc,
+                tear_down_tc)->AddTestInfo(test_info);
+  }
+
+#if GTEST_HAS_PARAM_TEST
+  // Returns ParameterizedTestCaseRegistry object used to keep track of
+  // value-parameterized tests and instantiate and register them.
+  internal::ParameterizedTestCaseRegistry& parameterized_test_registry() {
+    return parameterized_test_registry_;
+  }
+#endif  // GTEST_HAS_PARAM_TEST
+
+  // Sets the TestCase object for the test that's currently running.
+  void set_current_test_case(TestCase* a_current_test_case) {
+    current_test_case_ = a_current_test_case;
+  }
+
+  // Sets the TestInfo object for the test that's currently running.  If
+  // current_test_info is NULL, the assertion results will be stored in
+  // ad_hoc_test_result_.
+  void set_current_test_info(TestInfo* a_current_test_info) {
+    current_test_info_ = a_current_test_info;
+  }
+
+  // Registers all parameterized tests defined using TEST_P and
+  // INSTANTIATE_TEST_CASE_P, creating regular tests for each test/parameter
+  // combination. This method can be called more then once; it has guards
+  // protecting from registering the tests more then once.  If
+  // value-parameterized tests are disabled, RegisterParameterizedTests is
+  // present but does nothing.
+  void RegisterParameterizedTests();
+
+  // Runs all tests in this UnitTest object, prints the result, and
+  // returns true if all tests are successful.  If any exception is
+  // thrown during a test, this test is considered to be failed, but
+  // the rest of the tests will still be run.
+  bool RunAllTests();
+
+  // Clears the results of all tests, except the ad hoc tests.
+  void ClearNonAdHocTestResult() {
+    ForEach(test_cases_, TestCase::ClearTestCaseResult);
+  }
+
+  // Clears the results of ad-hoc test assertions.
+  void ClearAdHocTestResult() {
+    ad_hoc_test_result_.Clear();
+  }
+
+  // Adds a TestProperty to the current TestResult object when invoked in a
+  // context of a test or a test case, or to the global property set. If the
+  // result already contains a property with the same key, the value will be
+  // updated.
+  void RecordProperty(const TestProperty& test_property);
+
+  enum ReactionToSharding {
+    HONOR_SHARDING_PROTOCOL,
+    IGNORE_SHARDING_PROTOCOL
+  };
+
+  // Matches the full name of each test against the user-specified
+  // filter to decide whether the test should run, then records the
+  // result in each TestCase and TestInfo object.
+  // If shard_tests == HONOR_SHARDING_PROTOCOL, further filters tests
+  // based on sharding variables in the environment.
+  // Returns the number of tests that should run.
+  int FilterTests(ReactionToSharding shard_tests);
+
+  // Prints the names of the tests matching the user-specified filter flag.
+  void ListTestsMatchingFilter();
+
+  const TestCase* current_test_case() const { return current_test_case_; }
+  TestInfo* current_test_info() { return current_test_info_; }
+  const TestInfo* current_test_info() const { return current_test_info_; }
+
+  // Returns the vector of environments that need to be set-up/torn-down
+  // before/after the tests are run.
+  std::vector<Environment*>& environments() { return environments_; }
+
+  // Getters for the per-thread Google Test trace stack.
+  std::vector<TraceInfo>& gtest_trace_stack() {
+    return *(gtest_trace_stack_.pointer());
+  }
+  const std::vector<TraceInfo>& gtest_trace_stack() const {
+    return gtest_trace_stack_.get();
+  }
+
+#if GTEST_HAS_DEATH_TEST
+  void InitDeathTestSubprocessControlInfo() {
+    internal_run_death_test_flag_.reset(ParseInternalRunDeathTestFlag());
+  }
+  // Returns a pointer to the parsed --gtest_internal_run_death_test
+  // flag, or NULL if that flag was not specified.
+  // This information is useful only in a death test child process.
+  // Must not be called before a call to InitGoogleTest.
+  const InternalRunDeathTestFlag* internal_run_death_test_flag() const {
+    return internal_run_death_test_flag_.get();
+  }
+
+  // Returns a pointer to the current death test factory.
+  internal::DeathTestFactory* death_test_factory() {
+    return death_test_factory_.get();
+  }
+
+  void SuppressTestEventsIfInSubprocess();
+
+  friend class ReplaceDeathTestFactory;
+#endif  // GTEST_HAS_DEATH_TEST
+
+  // Initializes the event listener performing XML output as specified by
+  // UnitTestOptions. Must not be called before InitGoogleTest.
+  void ConfigureXmlOutput();
+
+#if GTEST_CAN_STREAM_RESULTS_
+  // Initializes the event listener for streaming test results to a socket.
+  // Must not be called before InitGoogleTest.
+  void ConfigureStreamingOutput();
+#endif
+
+  // Performs initialization dependent upon flag values obtained in
+  // ParseGoogleTestFlagsOnly.  Is called from InitGoogleTest after the call to
+  // ParseGoogleTestFlagsOnly.  In case a user neglects to call InitGoogleTest
+  // this function is also called from RunAllTests.  Since this function can be
+  // called more than once, it has to be idempotent.
+  void PostFlagParsingInit();
+
+  // Gets the random seed used at the start of the current test iteration.
+  int random_seed() const { return random_seed_; }
+
+  // Gets the random number generator.
+  internal::Random* random() { return &random_; }
+
+  // Shuffles all test cases, and the tests within each test case,
+  // making sure that death tests are still run first.
+  void ShuffleTests();
+
+  // Restores the test cases and tests to their order before the first shuffle.
+  void UnshuffleTests();
+
+  // Returns the value of GTEST_FLAG(catch_exceptions) at the moment
+  // UnitTest::Run() starts.
+  bool catch_exceptions() const { return catch_exceptions_; }
+
+ private:
+  friend class ::testing::UnitTest;
+
+  // Used by UnitTest::Run() to capture the state of
+  // GTEST_FLAG(catch_exceptions) at the moment it starts.
+  void set_catch_exceptions(bool value) { catch_exceptions_ = value; }
+
+  // The UnitTest object that owns this implementation object.
+  UnitTest* const parent_;
+
+  // The working directory when the first TEST() or TEST_F() was
+  // executed.
+  internal::FilePath original_working_dir_;
+
+  // The default test part result reporters.
+  DefaultGlobalTestPartResultReporter default_global_test_part_result_reporter_;
+  DefaultPerThreadTestPartResultReporter
+      default_per_thread_test_part_result_reporter_;
+
+  // Points to (but doesn't own) the global test part result reporter.
+  TestPartResultReporterInterface* global_test_part_result_repoter_;
+
+  // Protects read and write access to global_test_part_result_reporter_.
+  internal::Mutex global_test_part_result_reporter_mutex_;
+
+  // Points to (but doesn't own) the per-thread test part result reporter.
+  internal::ThreadLocal<TestPartResultReporterInterface*>
+      per_thread_test_part_result_reporter_;
+
+  // The vector of environments that need to be set-up/torn-down
+  // before/after the tests are run.
+  std::vector<Environment*> environments_;
+
+  // The vector of TestCases in their original order.  It owns the
+  // elements in the vector.
+  std::vector<TestCase*> test_cases_;
+
+  // Provides a level of indirection for the test case list to allow
+  // easy shuffling and restoring the test case order.  The i-th
+  // element of this vector is the index of the i-th test case in the
+  // shuffled order.
+  std::vector<int> test_case_indices_;
+
+#if GTEST_HAS_PARAM_TEST
+  // ParameterizedTestRegistry object used to register value-parameterized
+  // tests.
+  internal::ParameterizedTestCaseRegistry parameterized_test_registry_;
+
+  // Indicates whether RegisterParameterizedTests() has been called already.
+  bool parameterized_tests_registered_;
+#endif  // GTEST_HAS_PARAM_TEST
+
+  // Index of the last death test case registered.  Initially -1.
+  int last_death_test_case_;
+
+  // This points to the TestCase for the currently running test.  It
+  // changes as Google Test goes through one test case after another.
+  // When no test is running, this is set to NULL and Google Test
+  // stores assertion results in ad_hoc_test_result_.  Initially NULL.
+  TestCase* current_test_case_;
+
+  // This points to the TestInfo for the currently running test.  It
+  // changes as Google Test goes through one test after another.  When
+  // no test is running, this is set to NULL and Google Test stores
+  // assertion results in ad_hoc_test_result_.  Initially NULL.
+  TestInfo* current_test_info_;
+
+  // Normally, a user only writes assertions inside a TEST or TEST_F,
+  // or inside a function called by a TEST or TEST_F.  Since Google
+  // Test keeps track of which test is current running, it can
+  // associate such an assertion with the test it belongs to.
+  //
+  // If an assertion is encountered when no TEST or TEST_F is running,
+  // Google Test attributes the assertion result to an imaginary "ad hoc"
+  // test, and records the result in ad_hoc_test_result_.
+  TestResult ad_hoc_test_result_;
+
+  // The list of event listeners that can be used to track events inside
+  // Google Test.
+  TestEventListeners listeners_;
+
+  // The OS stack trace getter.  Will be deleted when the UnitTest
+  // object is destructed.  By default, an OsStackTraceGetter is used,
+  // but the user can set this field to use a custom getter if that is
+  // desired.
+  OsStackTraceGetterInterface* os_stack_trace_getter_;
+
+  // True iff PostFlagParsingInit() has been called.
+  bool post_flag_parse_init_performed_;
+
+  // The random number seed used at the beginning of the test run.
+  int random_seed_;
+
+  // Our random number generator.
+  internal::Random random_;
+
+  // The time of the test program start, in ms from the start of the
+  // UNIX epoch.
+  TimeInMillis start_timestamp_;
+
+  // How long the test took to run, in milliseconds.
+  TimeInMillis elapsed_time_;
+
+#if GTEST_HAS_DEATH_TEST
+  // The decomposed components of the gtest_internal_run_death_test flag,
+  // parsed when RUN_ALL_TESTS is called.
+  internal::scoped_ptr<InternalRunDeathTestFlag> internal_run_death_test_flag_;
+  internal::scoped_ptr<internal::DeathTestFactory> death_test_factory_;
+#endif  // GTEST_HAS_DEATH_TEST
+
+  // A per-thread stack of traces created by the SCOPED_TRACE() macro.
+  internal::ThreadLocal<std::vector<TraceInfo> > gtest_trace_stack_;
+
+  // The value of GTEST_FLAG(catch_exceptions) at the moment RunAllTests()
+  // starts.
+  bool catch_exceptions_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTestImpl);
+};  // class UnitTestImpl
+
+// Convenience function for accessing the global UnitTest
+// implementation object.
+inline UnitTestImpl* GetUnitTestImpl() {
+  return UnitTest::GetInstance()->impl();
+}
+
+#if GTEST_USES_SIMPLE_RE
+
+// Internal helper functions for implementing the simple regular
+// expression matcher.
+GTEST_API_ bool IsInSet(char ch, const char* str);
+GTEST_API_ bool IsAsciiDigit(char ch);
+GTEST_API_ bool IsAsciiPunct(char ch);
+GTEST_API_ bool IsRepeat(char ch);
+GTEST_API_ bool IsAsciiWhiteSpace(char ch);
+GTEST_API_ bool IsAsciiWordChar(char ch);
+GTEST_API_ bool IsValidEscape(char ch);
+GTEST_API_ bool AtomMatchesChar(bool escaped, char pattern, char ch);
+GTEST_API_ bool ValidateRegex(const char* regex);
+GTEST_API_ bool MatchRegexAtHead(const char* regex, const char* str);
+GTEST_API_ bool MatchRepetitionAndRegexAtHead(
+    bool escaped, char ch, char repeat, const char* regex, const char* str);
+GTEST_API_ bool MatchRegexAnywhere(const char* regex, const char* str);
+
+#endif  // GTEST_USES_SIMPLE_RE
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.
+GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, char** argv);
+GTEST_API_ void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv);
+
+#if GTEST_HAS_DEATH_TEST
+
+// Returns the message describing the last system error, regardless of the
+// platform.
+GTEST_API_ std::string GetLastErrnoDescription();
+
+// Attempts to parse a string into a positive integer pointed to by the
+// number parameter.  Returns true if that is possible.
+// GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we can use
+// it here.
+template <typename Integer>
+bool ParseNaturalNumber(const ::std::string& str, Integer* number) {
+  // Fail fast if the given string does not begin with a digit;
+  // this bypasses strtoXXX's "optional leading whitespace and plus
+  // or minus sign" semantics, which are undesirable here.
+  if (str.empty() || !IsDigit(str[0])) {
+    return false;
+  }
+  errno = 0;
+
+  char* end;
+  // BiggestConvertible is the largest integer type that system-provided
+  // string-to-number conversion routines can return.
+
+# if GTEST_OS_WINDOWS && !defined(__GNUC__)
+
+  // MSVC and C++ Builder define __int64 instead of the standard long long.
+  typedef unsigned __int64 BiggestConvertible;
+  const BiggestConvertible parsed = _strtoui64(str.c_str(), &end, 10);
+
+# else
+
+  typedef unsigned long long BiggestConvertible;  // NOLINT
+  const BiggestConvertible parsed = strtoull(str.c_str(), &end, 10);
+
+# endif  // GTEST_OS_WINDOWS && !defined(__GNUC__)
+
+  const bool parse_success = *end == '\0' && errno == 0;
+
+  // TODO(vladl@google.com): Convert this to compile time assertion when it is
+  // available.
+  GTEST_CHECK_(sizeof(Integer) <= sizeof(parsed));
+
+  const Integer result = static_cast<Integer>(parsed);
+  if (parse_success && static_cast<BiggestConvertible>(result) == parsed) {
+    *number = result;
+    return true;
+  }
+  return false;
+}
+#endif  // GTEST_HAS_DEATH_TEST
+
+// TestResult contains some private methods that should be hidden from
+// Google Test user but are required for testing. This class allow our tests
+// to access them.
+//
+// This class is supplied only for the purpose of testing Google Test's own
+// constructs. Do not use it in user tests, either directly or indirectly.
+class TestResultAccessor {
+ public:
+  static void RecordProperty(TestResult* test_result,
+                             const std::string& xml_element,
+                             const TestProperty& property) {
+    test_result->RecordProperty(xml_element, property);
+  }
+
+  static void ClearTestPartResults(TestResult* test_result) {
+    test_result->ClearTestPartResults();
+  }
+
+  static const std::vector<testing::TestPartResult>& test_part_results(
+      const TestResult& test_result) {
+    return test_result.test_part_results();
+  }
+};
+
+#if GTEST_CAN_STREAM_RESULTS_
+
+// Streams test results to the given port on the given host machine.
+class GTEST_API_ StreamingListener : public EmptyTestEventListener {
+ public:
+  // Abstract base class for writing strings to a socket.
+  class AbstractSocketWriter {
+   public:
+    virtual ~AbstractSocketWriter() {}
+
+    // Sends a string to the socket.
+    virtual void Send(const std::string& message) = 0;
+
+    // Closes the socket.
+    virtual void CloseConnection() {}
+
+    // Sends a string and a newline to the socket.
+    void SendLn(const std::string& message) { Send(message + "\n"); }
+  };
+
+  // Concrete class for actually writing strings to a socket.
+  class SocketWriter : public AbstractSocketWriter {
+   public:
+    SocketWriter(const std::string& host, const std::string& port)
+        : sockfd_(-1), host_name_(host), port_num_(port) {
+      MakeConnection();
+    }
+
+    virtual ~SocketWriter() {
+      if (sockfd_ != -1)
+        CloseConnection();
+    }
+
+    // Sends a string to the socket.
+    virtual void Send(const std::string& message) {
+      GTEST_CHECK_(sockfd_ != -1)
+          << "Send() can be called only when there is a connection.";
+
+      const int len = static_cast<int>(message.length());
+      if (write(sockfd_, message.c_str(), len) != len) {
+        GTEST_LOG_(WARNING)
+            << "stream_result_to: failed to stream to "
+            << host_name_ << ":" << port_num_;
+      }
+    }
+
+   private:
+    // Creates a client socket and connects to the server.
+    void MakeConnection();
+
+    // Closes the socket.
+    void CloseConnection() {
+      GTEST_CHECK_(sockfd_ != -1)
+          << "CloseConnection() can be called only when there is a connection.";
+
+      close(sockfd_);
+      sockfd_ = -1;
+    }
+
+    int sockfd_;  // socket file descriptor
+    const std::string host_name_;
+    const std::string port_num_;
+
+    GTEST_DISALLOW_COPY_AND_ASSIGN_(SocketWriter);
+  };  // class SocketWriter
+
+  // Escapes '=', '&', '%', and '\n' characters in str as "%xx".
+  static std::string UrlEncode(const char* str);
+
+  StreamingListener(const std::string& host, const std::string& port)
+      : socket_writer_(new SocketWriter(host, port)) {
+    Start();
+  }
+
+  explicit StreamingListener(AbstractSocketWriter* socket_writer)
+      : socket_writer_(socket_writer) { Start(); }
+
+  void OnTestProgramStart(const UnitTest& /* unit_test */) {
+    SendLn("event=TestProgramStart");
+  }
+
+  void OnTestProgramEnd(const UnitTest& unit_test) {
+    // Note that Google Test current only report elapsed time for each
+    // test iteration, not for the entire test program.
+    SendLn("event=TestProgramEnd&passed=" + FormatBool(unit_test.Passed()));
+
+    // Notify the streaming server to stop.
+    socket_writer_->CloseConnection();
+  }
+
+  void OnTestIterationStart(const UnitTest& /* unit_test */, int iteration) {
+    SendLn("event=TestIterationStart&iteration=" +
+           StreamableToString(iteration));
+  }
+
+  void OnTestIterationEnd(const UnitTest& unit_test, int /* iteration */) {
+    SendLn("event=TestIterationEnd&passed=" +
+           FormatBool(unit_test.Passed()) + "&elapsed_time=" +
+           StreamableToString(unit_test.elapsed_time()) + "ms");
+  }
+
+  void OnTestCaseStart(const TestCase& test_case) {
+    SendLn(std::string("event=TestCaseStart&name=") + test_case.name());
+  }
+
+  void OnTestCaseEnd(const TestCase& test_case) {
+    SendLn("event=TestCaseEnd&passed=" + FormatBool(test_case.Passed())
+           + "&elapsed_time=" + StreamableToString(test_case.elapsed_time())
+           + "ms");
+  }
+
+  void OnTestStart(const TestInfo& test_info) {
+    SendLn(std::string("event=TestStart&name=") + test_info.name());
+  }
+
+  void OnTestEnd(const TestInfo& test_info) {
+    SendLn("event=TestEnd&passed=" +
+           FormatBool((test_info.result())->Passed()) +
+           "&elapsed_time=" +
+           StreamableToString((test_info.result())->elapsed_time()) + "ms");
+  }
+
+  void OnTestPartResult(const TestPartResult& test_part_result) {
+    const char* file_name = test_part_result.file_name();
+    if (file_name == NULL)
+      file_name = "";
+    SendLn("event=TestPartResult&file=" + UrlEncode(file_name) +
+           "&line=" + StreamableToString(test_part_result.line_number()) +
+           "&message=" + UrlEncode(test_part_result.message()));
+  }
+
+ private:
+  // Sends the given message and a newline to the socket.
+  void SendLn(const std::string& message) { socket_writer_->SendLn(message); }
+
+  // Called at the start of streaming to notify the receiver what
+  // protocol we are using.
+  void Start() { SendLn("gtest_streaming_protocol_version=1.0"); }
+
+  std::string FormatBool(bool value) { return value ? "1" : "0"; }
+
+  const scoped_ptr<AbstractSocketWriter> socket_writer_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(StreamingListener);
+};  // class StreamingListener
+
+#endif  // GTEST_CAN_STREAM_RESULTS_
+
+}  // namespace internal
+
+// Returns the value of g_help_flag.
+bool GetGtestHelpFlag() {
+  return internal::g_help_flag;
+}
+}  // namespace testing
+
+#endif  // GTEST_SRC_GTEST_INTERNAL_INL_H_
+#undef GTEST_IMPLEMENTATION_
+
+#if GTEST_OS_WINDOWS
+# define vsnprintf _vsnprintf
+#endif  // GTEST_OS_WINDOWS
+
+namespace testing {
+
+using internal::CountIf;
+using internal::ForEach;
+using internal::GetElementOr;
+using internal::Shuffle;
+
+// Constants.
+
+// A test whose test case name or test name matches this filter is
+// disabled and not run.
+static const char kDisableTestFilter[] = "DISABLED_*:*/DISABLED_*";
+
+// A test case whose name matches this filter is considered a death
+// test case and will be run before test cases whose name doesn't
+// match this filter.
+static const char kDeathTestCaseFilter[] = "*DeathTest:*DeathTest/*";
+
+// A test filter that matches everything.
+static const char kUniversalFilter[] = "*";
+
+// The default output file for XML output.
+static const char kDefaultOutputFile[] = "test_detail.xml";
+
+// The environment variable name for the test shard index.
+static const char kTestShardIndex[] = "GTEST_SHARD_INDEX";
+// The environment variable name for the total number of test shards.
+static const char kTestTotalShards[] = "GTEST_TOTAL_SHARDS";
+// The environment variable name for the test shard status file.
+static const char kTestShardStatusFile[] = "GTEST_SHARD_STATUS_FILE";
+
+namespace internal {
+
+// The text used in failure messages to indicate the start of the
+// stack trace.
+const char kStackTraceMarker[] = "\nStack trace:\n";
+
+// g_help_flag is true iff the --help flag or an equivalent form is
+// specified on the command line.
+bool g_help_flag = false;
+
+}  // namespace internal
+
+static const char* GetDefaultFilter() {
+#ifdef GTEST_TEST_FILTER_ENV_VAR_
+  const char* const testbridge_test_only = getenv(GTEST_TEST_FILTER_ENV_VAR_);
+  if (testbridge_test_only != NULL) {
+    return testbridge_test_only;
+  }
+#endif  // GTEST_TEST_FILTER_ENV_VAR_
+  return kUniversalFilter;
+}
+
+GTEST_DEFINE_bool_(
+    also_run_disabled_tests,
+    internal::BoolFromGTestEnv("also_run_disabled_tests", false),
+    "Run disabled tests too, in addition to the tests normally being run.");
+
+GTEST_DEFINE_bool_(
+    break_on_failure,
+    internal::BoolFromGTestEnv("break_on_failure", false),
+    "True iff a failed assertion should be a debugger break-point.");
+
+GTEST_DEFINE_bool_(
+    catch_exceptions,
+    internal::BoolFromGTestEnv("catch_exceptions", true),
+    "True iff " GTEST_NAME_
+    " should catch exceptions and treat them as test failures.");
+
+GTEST_DEFINE_string_(
+    color,
+    internal::StringFromGTestEnv("color", "auto"),
+    "Whether to use colors in the output.  Valid values: yes, no, "
+    "and auto.  'auto' means to use colors if the output is "
+    "being sent to a terminal and the TERM environment variable "
+    "is set to a terminal type that supports colors.");
+
+GTEST_DEFINE_string_(
+    filter,
+    internal::StringFromGTestEnv("filter", GetDefaultFilter()),
+    "A colon-separated list of glob (not regex) patterns "
+    "for filtering the tests to run, optionally followed by a "
+    "'-' and a : separated list of negative patterns (tests to "
+    "exclude).  A test is run if it matches one of the positive "
+    "patterns and does not match any of the negative patterns.");
+
+GTEST_DEFINE_bool_(list_tests, false,
+                   "List all tests without running them.");
+
+GTEST_DEFINE_string_(
+    output,
+    internal::StringFromGTestEnv("output", ""),
+    "A format (currently must be \"xml\"), optionally followed "
+    "by a colon and an output file name or directory. A directory "
+    "is indicated by a trailing pathname separator. "
+    "Examples: \"xml:filename.xml\", \"xml::directoryname/\". "
+    "If a directory is specified, output files will be created "
+    "within that directory, with file-names based on the test "
+    "executable's name and, if necessary, made unique by adding "
+    "digits.");
+
+GTEST_DEFINE_bool_(
+    print_time,
+    internal::BoolFromGTestEnv("print_time", true),
+    "True iff " GTEST_NAME_
+    " should display elapsed time in text output.");
+
+GTEST_DEFINE_int32_(
+    random_seed,
+    internal::Int32FromGTestEnv("random_seed", 0),
+    "Random number seed to use when shuffling test orders.  Must be in range "
+    "[1, 99999], or 0 to use a seed based on the current time.");
+
+GTEST_DEFINE_int32_(
+    repeat,
+    internal::Int32FromGTestEnv("repeat", 1),
+    "How many times to repeat each test.  Specify a negative number "
+    "for repeating forever.  Useful for shaking out flaky tests.");
+
+GTEST_DEFINE_bool_(
+    show_internal_stack_frames, false,
+    "True iff " GTEST_NAME_ " should include internal stack frames when "
+    "printing test failure stack traces.");
+
+GTEST_DEFINE_bool_(
+    shuffle,
+    internal::BoolFromGTestEnv("shuffle", false),
+    "True iff " GTEST_NAME_
+    " should randomize tests' order on every run.");
+
+GTEST_DEFINE_int32_(
+    stack_trace_depth,
+    internal::Int32FromGTestEnv("stack_trace_depth", kMaxStackTraceDepth),
+    "The maximum number of stack frames to print when an "
+    "assertion fails.  The valid range is 0 through 100, inclusive.");
+
+GTEST_DEFINE_string_(
+    stream_result_to,
+    internal::StringFromGTestEnv("stream_result_to", ""),
+    "This flag specifies the host name and the port number on which to stream "
+    "test results. Example: \"localhost:555\". The flag is effective only on "
+    "Linux.");
+
+GTEST_DEFINE_bool_(
+    throw_on_failure,
+    internal::BoolFromGTestEnv("throw_on_failure", false),
+    "When this flag is specified, a failed assertion will throw an exception "
+    "if exceptions are enabled or exit the program with a non-zero code "
+    "otherwise.");
+
+#if GTEST_USE_OWN_FLAGFILE_FLAG_
+GTEST_DEFINE_string_(
+    flagfile,
+    internal::StringFromGTestEnv("flagfile", ""),
+    "This flag specifies the flagfile to read command-line flags from.");
+#endif  // GTEST_USE_OWN_FLAGFILE_FLAG_
+
+namespace internal {
+
+// Generates a random number from [0, range), using a Linear
+// Congruential Generator (LCG).  Crashes if 'range' is 0 or greater
+// than kMaxRange.
+UInt32 Random::Generate(UInt32 range) {
+  // These constants are the same as are used in glibc's rand(3).
+  state_ = (1103515245U*state_ + 12345U) % kMaxRange;
+
+  GTEST_CHECK_(range > 0)
+      << "Cannot generate a number in the range [0, 0).";
+  GTEST_CHECK_(range <= kMaxRange)
+      << "Generation of a number in [0, " << range << ") was requested, "
+      << "but this can only generate numbers in [0, " << kMaxRange << ").";
+
+  // Converting via modulus introduces a bit of downward bias, but
+  // it's simple, and a linear congruential generator isn't too good
+  // to begin with.
+  return state_ % range;
+}
+
+// GTestIsInitialized() returns true iff the user has initialized
+// Google Test.  Useful for catching the user mistake of not initializing
+// Google Test before calling RUN_ALL_TESTS().
+static bool GTestIsInitialized() { return GetArgvs().size() > 0; }
+
+// Iterates over a vector of TestCases, keeping a running sum of the
+// results of calling a given int-returning method on each.
+// Returns the sum.
+static int SumOverTestCaseList(const std::vector<TestCase*>& case_list,
+                               int (TestCase::*method)() const) {
+  int sum = 0;
+  for (size_t i = 0; i < case_list.size(); i++) {
+    sum += (case_list[i]->*method)();
+  }
+  return sum;
+}
+
+// Returns true iff the test case passed.
+static bool TestCasePassed(const TestCase* test_case) {
+  return test_case->should_run() && test_case->Passed();
+}
+
+// Returns true iff the test case failed.
+static bool TestCaseFailed(const TestCase* test_case) {
+  return test_case->should_run() && test_case->Failed();
+}
+
+// Returns true iff test_case contains at least one test that should
+// run.
+static bool ShouldRunTestCase(const TestCase* test_case) {
+  return test_case->should_run();
+}
+
+// AssertHelper constructor.
+AssertHelper::AssertHelper(TestPartResult::Type type,
+                           const char* file,
+                           int line,
+                           const char* message)
+    : data_(new AssertHelperData(type, file, line, message)) {
+}
+
+AssertHelper::~AssertHelper() {
+  delete data_;
+}
+
+// Message assignment, for assertion streaming support.
+void AssertHelper::operator=(const Message& message) const {
+  UnitTest::GetInstance()->
+    AddTestPartResult(data_->type, data_->file, data_->line,
+                      AppendUserMessage(data_->message, message),
+                      UnitTest::GetInstance()->impl()
+                      ->CurrentOsStackTraceExceptTop(1)
+                      // Skips the stack frame for this function itself.
+                      );  // NOLINT
+}
+
+// Mutex for linked pointers.
+GTEST_API_ GTEST_DEFINE_STATIC_MUTEX_(g_linked_ptr_mutex);
+
+// A copy of all command line arguments.  Set by InitGoogleTest().
+::std::vector<testing::internal::string> g_argvs;
+
+const ::std::vector<testing::internal::string>& GetArgvs() {
+#if defined(GTEST_CUSTOM_GET_ARGVS_)
+  return GTEST_CUSTOM_GET_ARGVS_();
+#else  // defined(GTEST_CUSTOM_GET_ARGVS_)
+  return g_argvs;
+#endif  // defined(GTEST_CUSTOM_GET_ARGVS_)
+}
+
+// Returns the current application's name, removing directory path if that
+// is present.
+FilePath GetCurrentExecutableName() {
+  FilePath result;
+
+#if GTEST_OS_WINDOWS
+  result.Set(FilePath(GetArgvs()[0]).RemoveExtension("exe"));
+#else
+  result.Set(FilePath(GetArgvs()[0]));
+#endif  // GTEST_OS_WINDOWS
+
+  return result.RemoveDirectoryName();
+}
+
+// Functions for processing the gtest_output flag.
+
+// Returns the output format, or "" for normal printed output.
+std::string UnitTestOptions::GetOutputFormat() {
+  const char* const gtest_output_flag = GTEST_FLAG(output).c_str();
+  if (gtest_output_flag == NULL) return std::string("");
+
+  const char* const colon = strchr(gtest_output_flag, ':');
+  return (colon == NULL) ?
+      std::string(gtest_output_flag) :
+      std::string(gtest_output_flag, colon - gtest_output_flag);
+}
+
+// Returns the name of the requested output file, or the default if none
+// was explicitly specified.
+std::string UnitTestOptions::GetAbsolutePathToOutputFile() {
+  const char* const gtest_output_flag = GTEST_FLAG(output).c_str();
+  if (gtest_output_flag == NULL)
+    return "";
+
+  const char* const colon = strchr(gtest_output_flag, ':');
+  if (colon == NULL)
+    return internal::FilePath::ConcatPaths(
+        internal::FilePath(
+            UnitTest::GetInstance()->original_working_dir()),
+        internal::FilePath(kDefaultOutputFile)).string();
+
+  internal::FilePath output_name(colon + 1);
+  if (!output_name.IsAbsolutePath())
+    // TODO(wan@google.com): on Windows \some\path is not an absolute
+    // path (as its meaning depends on the current drive), yet the
+    // following logic for turning it into an absolute path is wrong.
+    // Fix it.
+    output_name = internal::FilePath::ConcatPaths(
+        internal::FilePath(UnitTest::GetInstance()->original_working_dir()),
+        internal::FilePath(colon + 1));
+
+  if (!output_name.IsDirectory())
+    return output_name.string();
+
+  internal::FilePath result(internal::FilePath::GenerateUniqueFileName(
+      output_name, internal::GetCurrentExecutableName(),
+      GetOutputFormat().c_str()));
+  return result.string();
+}
+
+// Returns true iff the wildcard pattern matches the string.  The
+// first ':' or '\0' character in pattern marks the end of it.
+//
+// This recursive algorithm isn't very efficient, but is clear and
+// works well enough for matching test names, which are short.
+bool UnitTestOptions::PatternMatchesString(const char *pattern,
+                                           const char *str) {
+  switch (*pattern) {
+    case '\0':
+    case ':':  // Either ':' or '\0' marks the end of the pattern.
+      return *str == '\0';
+    case '?':  // Matches any single character.
+      return *str != '\0' && PatternMatchesString(pattern + 1, str + 1);
+    case '*':  // Matches any string (possibly empty) of characters.
+      return (*str != '\0' && PatternMatchesString(pattern, str + 1)) ||
+          PatternMatchesString(pattern + 1, str);
+    default:  // Non-special character.  Matches itself.
+      return *pattern == *str &&
+          PatternMatchesString(pattern + 1, str + 1);
+  }
+}
+
+bool UnitTestOptions::MatchesFilter(
+    const std::string& name, const char* filter) {
+  const char *cur_pattern = filter;
+  for (;;) {
+    if (PatternMatchesString(cur_pattern, name.c_str())) {
+      return true;
+    }
+
+    // Finds the next pattern in the filter.
+    cur_pattern = strchr(cur_pattern, ':');
+
+    // Returns if no more pattern can be found.
+    if (cur_pattern == NULL) {
+      return false;
+    }
+
+    // Skips the pattern separator (the ':' character).
+    cur_pattern++;
+  }
+}
+
+// Returns true iff the user-specified filter matches the test case
+// name and the test name.
+bool UnitTestOptions::FilterMatchesTest(const std::string &test_case_name,
+                                        const std::string &test_name) {
+  const std::string& full_name = test_case_name + "." + test_name.c_str();
+
+  // Split --gtest_filter at '-', if there is one, to separate into
+  // positive filter and negative filter portions
+  const char* const p = GTEST_FLAG(filter).c_str();
+  const char* const dash = strchr(p, '-');
+  std::string positive;
+  std::string negative;
+  if (dash == NULL) {
+    positive = GTEST_FLAG(filter).c_str();  // Whole string is a positive filter
+    negative = "";
+  } else {
+    positive = std::string(p, dash);   // Everything up to the dash
+    negative = std::string(dash + 1);  // Everything after the dash
+    if (positive.empty()) {
+      // Treat '-test1' as the same as '*-test1'
+      positive = kUniversalFilter;
+    }
+  }
+
+  // A filter is a colon-separated list of patterns.  It matches a
+  // test if any pattern in it matches the test.
+  return (MatchesFilter(full_name, positive.c_str()) &&
+          !MatchesFilter(full_name, negative.c_str()));
+}
+
+#if GTEST_HAS_SEH
+// Returns EXCEPTION_EXECUTE_HANDLER if Google Test should handle the
+// given SEH exception, or EXCEPTION_CONTINUE_SEARCH otherwise.
+// This function is useful as an __except condition.
+int UnitTestOptions::GTestShouldProcessSEH(DWORD exception_code) {
+  // Google Test should handle a SEH exception if:
+  //   1. the user wants it to, AND
+  //   2. this is not a breakpoint exception, AND
+  //   3. this is not a C++ exception (VC++ implements them via SEH,
+  //      apparently).
+  //
+  // SEH exception code for C++ exceptions.
+  // (see http://support.microsoft.com/kb/185294 for more information).
+  const DWORD kCxxExceptionCode = 0xe06d7363;
+
+  bool should_handle = true;
+
+  if (!GTEST_FLAG(catch_exceptions))
+    should_handle = false;
+  else if (exception_code == EXCEPTION_BREAKPOINT)
+    should_handle = false;
+  else if (exception_code == kCxxExceptionCode)
+    should_handle = false;
+
+  return should_handle ? EXCEPTION_EXECUTE_HANDLER : EXCEPTION_CONTINUE_SEARCH;
+}
+#endif  // GTEST_HAS_SEH
+
+}  // namespace internal
+
+// The c'tor sets this object as the test part result reporter used by
+// Google Test.  The 'result' parameter specifies where to report the
+// results. Intercepts only failures from the current thread.
+ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter(
+    TestPartResultArray* result)
+    : intercept_mode_(INTERCEPT_ONLY_CURRENT_THREAD),
+      result_(result) {
+  Init();
+}
+
+// The c'tor sets this object as the test part result reporter used by
+// Google Test.  The 'result' parameter specifies where to report the
+// results.
+ScopedFakeTestPartResultReporter::ScopedFakeTestPartResultReporter(
+    InterceptMode intercept_mode, TestPartResultArray* result)
+    : intercept_mode_(intercept_mode),
+      result_(result) {
+  Init();
+}
+
+void ScopedFakeTestPartResultReporter::Init() {
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  if (intercept_mode_ == INTERCEPT_ALL_THREADS) {
+    old_reporter_ = impl->GetGlobalTestPartResultReporter();
+    impl->SetGlobalTestPartResultReporter(this);
+  } else {
+    old_reporter_ = impl->GetTestPartResultReporterForCurrentThread();
+    impl->SetTestPartResultReporterForCurrentThread(this);
+  }
+}
+
+// The d'tor restores the test part result reporter used by Google Test
+// before.
+ScopedFakeTestPartResultReporter::~ScopedFakeTestPartResultReporter() {
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  if (intercept_mode_ == INTERCEPT_ALL_THREADS) {
+    impl->SetGlobalTestPartResultReporter(old_reporter_);
+  } else {
+    impl->SetTestPartResultReporterForCurrentThread(old_reporter_);
+  }
+}
+
+// Increments the test part result count and remembers the result.
+// This method is from the TestPartResultReporterInterface interface.
+void ScopedFakeTestPartResultReporter::ReportTestPartResult(
+    const TestPartResult& result) {
+  result_->Append(result);
+}
+
+namespace internal {
+
+// Returns the type ID of ::testing::Test.  We should always call this
+// instead of GetTypeId< ::testing::Test>() to get the type ID of
+// testing::Test.  This is to work around a suspected linker bug when
+// using Google Test as a framework on Mac OS X.  The bug causes
+// GetTypeId< ::testing::Test>() to return different values depending
+// on whether the call is from the Google Test framework itself or
+// from user test code.  GetTestTypeId() is guaranteed to always
+// return the same value, as it always calls GetTypeId<>() from the
+// gtest.cc, which is within the Google Test framework.
+TypeId GetTestTypeId() {
+  return GetTypeId<Test>();
+}
+
+// The value of GetTestTypeId() as seen from within the Google Test
+// library.  This is solely for testing GetTestTypeId().
+extern const TypeId kTestTypeIdInGoogleTest = GetTestTypeId();
+
+// This predicate-formatter checks that 'results' contains a test part
+// failure of the given type and that the failure message contains the
+// given substring.
+AssertionResult HasOneFailure(const char* /* results_expr */,
+                              const char* /* type_expr */,
+                              const char* /* substr_expr */,
+                              const TestPartResultArray& results,
+                              TestPartResult::Type type,
+                              const std::string& substr) {
+  const std::string expected(type == TestPartResult::kFatalFailure ?
+                        "1 fatal failure" :
+                        "1 non-fatal failure");
+  Message msg;
+  if (results.size() != 1) {
+    msg << "Expected: " << expected << "\n"
+        << "  Actual: " << results.size() << " failures";
+    for (int i = 0; i < results.size(); i++) {
+      msg << "\n" << results.GetTestPartResult(i);
+    }
+    return AssertionFailure() << msg;
+  }
+
+  const TestPartResult& r = results.GetTestPartResult(0);
+  if (r.type() != type) {
+    return AssertionFailure() << "Expected: " << expected << "\n"
+                              << "  Actual:\n"
+                              << r;
+  }
+
+  if (strstr(r.message(), substr.c_str()) == NULL) {
+    return AssertionFailure() << "Expected: " << expected << " containing \""
+                              << substr << "\"\n"
+                              << "  Actual:\n"
+                              << r;
+  }
+
+  return AssertionSuccess();
+}
+
+// The constructor of SingleFailureChecker remembers where to look up
+// test part results, what type of failure we expect, and what
+// substring the failure message should contain.
+SingleFailureChecker::SingleFailureChecker(const TestPartResultArray* results,
+                                           TestPartResult::Type type,
+                                           const std::string& substr)
+    : results_(results), type_(type), substr_(substr) {}
+
+// The destructor of SingleFailureChecker verifies that the given
+// TestPartResultArray contains exactly one failure that has the given
+// type and contains the given substring.  If that's not the case, a
+// non-fatal failure will be generated.
+SingleFailureChecker::~SingleFailureChecker() {
+  EXPECT_PRED_FORMAT3(HasOneFailure, *results_, type_, substr_);
+}
+
+DefaultGlobalTestPartResultReporter::DefaultGlobalTestPartResultReporter(
+    UnitTestImpl* unit_test) : unit_test_(unit_test) {}
+
+void DefaultGlobalTestPartResultReporter::ReportTestPartResult(
+    const TestPartResult& result) {
+  unit_test_->current_test_result()->AddTestPartResult(result);
+  unit_test_->listeners()->repeater()->OnTestPartResult(result);
+}
+
+DefaultPerThreadTestPartResultReporter::DefaultPerThreadTestPartResultReporter(
+    UnitTestImpl* unit_test) : unit_test_(unit_test) {}
+
+void DefaultPerThreadTestPartResultReporter::ReportTestPartResult(
+    const TestPartResult& result) {
+  unit_test_->GetGlobalTestPartResultReporter()->ReportTestPartResult(result);
+}
+
+// Returns the global test part result reporter.
+TestPartResultReporterInterface*
+UnitTestImpl::GetGlobalTestPartResultReporter() {
+  internal::MutexLock lock(&global_test_part_result_reporter_mutex_);
+  return global_test_part_result_repoter_;
+}
+
+// Sets the global test part result reporter.
+void UnitTestImpl::SetGlobalTestPartResultReporter(
+    TestPartResultReporterInterface* reporter) {
+  internal::MutexLock lock(&global_test_part_result_reporter_mutex_);
+  global_test_part_result_repoter_ = reporter;
+}
+
+// Returns the test part result reporter for the current thread.
+TestPartResultReporterInterface*
+UnitTestImpl::GetTestPartResultReporterForCurrentThread() {
+  return per_thread_test_part_result_reporter_.get();
+}
+
+// Sets the test part result reporter for the current thread.
+void UnitTestImpl::SetTestPartResultReporterForCurrentThread(
+    TestPartResultReporterInterface* reporter) {
+  per_thread_test_part_result_reporter_.set(reporter);
+}
+
+// Gets the number of successful test cases.
+int UnitTestImpl::successful_test_case_count() const {
+  return CountIf(test_cases_, TestCasePassed);
+}
+
+// Gets the number of failed test cases.
+int UnitTestImpl::failed_test_case_count() const {
+  return CountIf(test_cases_, TestCaseFailed);
+}
+
+// Gets the number of all test cases.
+int UnitTestImpl::total_test_case_count() const {
+  return static_cast<int>(test_cases_.size());
+}
+
+// Gets the number of all test cases that contain at least one test
+// that should run.
+int UnitTestImpl::test_case_to_run_count() const {
+  return CountIf(test_cases_, ShouldRunTestCase);
+}
+
+// Gets the number of successful tests.
+int UnitTestImpl::successful_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::successful_test_count);
+}
+
+// Gets the number of skipped tests.
+int UnitTestImpl::skipped_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::skipped_test_count);
+}
+
+// Gets the number of failed tests.
+int UnitTestImpl::failed_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::failed_test_count);
+}
+
+// Gets the number of disabled tests that will be reported in the XML report.
+int UnitTestImpl::reportable_disabled_test_count() const {
+  return SumOverTestCaseList(test_cases_,
+                             &TestCase::reportable_disabled_test_count);
+}
+
+// Gets the number of disabled tests.
+int UnitTestImpl::disabled_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::disabled_test_count);
+}
+
+// Gets the number of tests to be printed in the XML report.
+int UnitTestImpl::reportable_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::reportable_test_count);
+}
+
+// Gets the number of all tests.
+int UnitTestImpl::total_test_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::total_test_count);
+}
+
+// Gets the number of tests that should run.
+int UnitTestImpl::test_to_run_count() const {
+  return SumOverTestCaseList(test_cases_, &TestCase::test_to_run_count);
+}
+
+// Returns the current OS stack trace as an std::string.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag.  The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// CurrentOsStackTraceExceptTop(1), Foo() will be included in the
+// trace but Bar() and CurrentOsStackTraceExceptTop() won't.
+std::string UnitTestImpl::CurrentOsStackTraceExceptTop(int skip_count) {
+  return os_stack_trace_getter()->CurrentStackTrace(
+      static_cast<int>(GTEST_FLAG(stack_trace_depth)),
+      skip_count + 1
+      // Skips the user-specified number of frames plus this function
+      // itself.
+      );  // NOLINT
+}
+
+// Returns the current time in milliseconds.
+TimeInMillis GetTimeInMillis() {
+#if GTEST_OS_WINDOWS_MOBILE || defined(__BORLANDC__)
+  // Difference between 1970-01-01 and 1601-01-01 in milliseconds.
+  // http://analogous.blogspot.com/2005/04/epoch.html
+  const TimeInMillis kJavaEpochToWinFileTimeDelta =
+    static_cast<TimeInMillis>(116444736UL) * 100000UL;
+  const DWORD kTenthMicrosInMilliSecond = 10000;
+
+  SYSTEMTIME now_systime;
+  FILETIME now_filetime;
+  ULARGE_INTEGER now_int64;
+  // TODO(kenton@google.com): Shouldn't this just use
+  //   GetSystemTimeAsFileTime()?
+  GetSystemTime(&now_systime);
+  if (SystemTimeToFileTime(&now_systime, &now_filetime)) {
+    now_int64.LowPart = now_filetime.dwLowDateTime;
+    now_int64.HighPart = now_filetime.dwHighDateTime;
+    now_int64.QuadPart = (now_int64.QuadPart / kTenthMicrosInMilliSecond) -
+      kJavaEpochToWinFileTimeDelta;
+    return now_int64.QuadPart;
+  }
+  return 0;
+#elif GTEST_OS_WINDOWS && !GTEST_HAS_GETTIMEOFDAY_
+  __timeb64 now;
+
+  // MSVC 8 deprecates _ftime64(), so we want to suppress warning 4996
+  // (deprecated function) there.
+  // TODO(kenton@google.com): Use GetTickCount()?  Or use
+  //   SystemTimeToFileTime()
+  GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996)
+  _ftime64(&now);
+  GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+  return static_cast<TimeInMillis>(now.time) * 1000 + now.millitm;
+#elif GTEST_HAS_GETTIMEOFDAY_
+  struct timeval now;
+  gettimeofday(&now, NULL);
+  return static_cast<TimeInMillis>(now.tv_sec) * 1000 + now.tv_usec / 1000;
+#else
+# error "Don't know how to get the current time on your system."
+#endif
+}
+
+// Utilities
+
+// class String.
+
+#if GTEST_OS_WINDOWS_MOBILE
+// Creates a UTF-16 wide string from the given ANSI string, allocating
+// memory using new. The caller is responsible for deleting the return
+// value using delete[]. Returns the wide string, or NULL if the
+// input is NULL.
+LPCWSTR String::AnsiToUtf16(const char* ansi) {
+  if (!ansi) return NULL;
+  const int length = strlen(ansi);
+  const int unicode_length =
+      MultiByteToWideChar(CP_ACP, 0, ansi, length,
+                          NULL, 0);
+  WCHAR* unicode = new WCHAR[unicode_length + 1];
+  MultiByteToWideChar(CP_ACP, 0, ansi, length,
+                      unicode, unicode_length);
+  unicode[unicode_length] = 0;
+  return unicode;
+}
+
+// Creates an ANSI string from the given wide string, allocating
+// memory using new. The caller is responsible for deleting the return
+// value using delete[]. Returns the ANSI string, or NULL if the
+// input is NULL.
+const char* String::Utf16ToAnsi(LPCWSTR utf16_str)  {
+  if (!utf16_str) return NULL;
+  const int ansi_length =
+      WideCharToMultiByte(CP_ACP, 0, utf16_str, -1,
+                          NULL, 0, NULL, NULL);
+  char* ansi = new char[ansi_length + 1];
+  WideCharToMultiByte(CP_ACP, 0, utf16_str, -1,
+                      ansi, ansi_length, NULL, NULL);
+  ansi[ansi_length] = 0;
+  return ansi;
+}
+
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+// Compares two C strings.  Returns true iff they have the same content.
+//
+// Unlike strcmp(), this function can handle NULL argument(s).  A NULL
+// C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::CStringEquals(const char * lhs, const char * rhs) {
+  if ( lhs == NULL ) return rhs == NULL;
+
+  if ( rhs == NULL ) return false;
+
+  return strcmp(lhs, rhs) == 0;
+}
+
+#if GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING
+
+// Converts an array of wide chars to a narrow string using the UTF-8
+// encoding, and streams the result to the given Message object.
+static void StreamWideCharsToMessage(const wchar_t* wstr, size_t length,
+                                     Message* msg) {
+  for (size_t i = 0; i != length; ) {  // NOLINT
+    if (wstr[i] != L'\0') {
+      *msg << WideStringToUtf8(wstr + i, static_cast<int>(length - i));
+      while (i != length && wstr[i] != L'\0')
+        i++;
+    } else {
+      *msg << '\0';
+      i++;
+    }
+  }
+}
+
+#endif  // GTEST_HAS_STD_WSTRING || GTEST_HAS_GLOBAL_WSTRING
+
+void SplitString(const ::std::string& str, char delimiter,
+                 ::std::vector< ::std::string>* dest) {
+  ::std::vector< ::std::string> parsed;
+  ::std::string::size_type pos = 0;
+  while (::testing::internal::AlwaysTrue()) {
+    const ::std::string::size_type colon = str.find(delimiter, pos);
+    if (colon == ::std::string::npos) {
+      parsed.push_back(str.substr(pos));
+      break;
+    } else {
+      parsed.push_back(str.substr(pos, colon - pos));
+      pos = colon + 1;
+    }
+  }
+  dest->swap(parsed);
+}
+
+}  // namespace internal
+
+// Constructs an empty Message.
+// We allocate the stringstream separately because otherwise each use of
+// ASSERT/EXPECT in a procedure adds over 200 bytes to the procedure's
+// stack frame leading to huge stack frames in some cases; gcc does not reuse
+// the stack space.
+Message::Message() : ss_(new ::std::stringstream) {
+  // By default, we want there to be enough precision when printing
+  // a double to a Message.
+  *ss_ << std::setprecision(std::numeric_limits<double>::digits10 + 2);
+}
+
+// These two overloads allow streaming a wide C string to a Message
+// using the UTF-8 encoding.
+Message& Message::operator <<(const wchar_t* wide_c_str) {
+  return *this << internal::String::ShowWideCString(wide_c_str);
+}
+Message& Message::operator <<(wchar_t* wide_c_str) {
+  return *this << internal::String::ShowWideCString(wide_c_str);
+}
+
+#if GTEST_HAS_STD_WSTRING
+// Converts the given wide string to a narrow string using the UTF-8
+// encoding, and streams the result to this Message object.
+Message& Message::operator <<(const ::std::wstring& wstr) {
+  internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this);
+  return *this;
+}
+#endif  // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+// Converts the given wide string to a narrow string using the UTF-8
+// encoding, and streams the result to this Message object.
+Message& Message::operator <<(const ::wstring& wstr) {
+  internal::StreamWideCharsToMessage(wstr.c_str(), wstr.length(), this);
+  return *this;
+}
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+// Gets the text streamed to this object so far as an std::string.
+// Each '\0' character in the buffer is replaced with "\\0".
+std::string Message::GetString() const {
+  return internal::StringStreamToString(ss_.get());
+}
+
+// AssertionResult constructors.
+// Used in EXPECT_TRUE/FALSE(assertion_result).
+AssertionResult::AssertionResult(const AssertionResult& other)
+    : success_(other.success_),
+      message_(other.message_.get() != NULL ?
+               new ::std::string(*other.message_) :
+               static_cast< ::std::string*>(NULL)) {
+}
+
+// Swaps two AssertionResults.
+void AssertionResult::swap(AssertionResult& other) {
+  using std::swap;
+  swap(success_, other.success_);
+  swap(message_, other.message_);
+}
+
+// Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE.
+AssertionResult AssertionResult::operator!() const {
+  AssertionResult negation(!success_);
+  if (message_.get() != NULL)
+    negation << *message_;
+  return negation;
+}
+
+// Makes a successful assertion result.
+AssertionResult AssertionSuccess() {
+  return AssertionResult(true);
+}
+
+// Makes a failed assertion result.
+AssertionResult AssertionFailure() {
+  return AssertionResult(false);
+}
+
+// Makes a failed assertion result with the given failure message.
+// Deprecated; use AssertionFailure() << message.
+AssertionResult AssertionFailure(const Message& message) {
+  return AssertionFailure() << message;
+}
+
+namespace internal {
+
+namespace edit_distance {
+std::vector<EditType> CalculateOptimalEdits(const std::vector<size_t>& left,
+                                            const std::vector<size_t>& right) {
+  std::vector<std::vector<double> > costs(
+      left.size() + 1, std::vector<double>(right.size() + 1));
+  std::vector<std::vector<EditType> > best_move(
+      left.size() + 1, std::vector<EditType>(right.size() + 1));
+
+  // Populate for empty right.
+  for (size_t l_i = 0; l_i < costs.size(); ++l_i) {
+    costs[l_i][0] = static_cast<double>(l_i);
+    best_move[l_i][0] = kRemove;
+  }
+  // Populate for empty left.
+  for (size_t r_i = 1; r_i < costs[0].size(); ++r_i) {
+    costs[0][r_i] = static_cast<double>(r_i);
+    best_move[0][r_i] = kAdd;
+  }
+
+  for (size_t l_i = 0; l_i < left.size(); ++l_i) {
+    for (size_t r_i = 0; r_i < right.size(); ++r_i) {
+      if (left[l_i] == right[r_i]) {
+        // Found a match. Consume it.
+        costs[l_i + 1][r_i + 1] = costs[l_i][r_i];
+        best_move[l_i + 1][r_i + 1] = kMatch;
+        continue;
+      }
+
+      const double add = costs[l_i + 1][r_i];
+      const double remove = costs[l_i][r_i + 1];
+      const double replace = costs[l_i][r_i];
+      if (add < remove && add < replace) {
+        costs[l_i + 1][r_i + 1] = add + 1;
+        best_move[l_i + 1][r_i + 1] = kAdd;
+      } else if (remove < add && remove < replace) {
+        costs[l_i + 1][r_i + 1] = remove + 1;
+        best_move[l_i + 1][r_i + 1] = kRemove;
+      } else {
+        // We make replace a little more expensive than add/remove to lower
+        // their priority.
+        costs[l_i + 1][r_i + 1] = replace + 1.00001;
+        best_move[l_i + 1][r_i + 1] = kReplace;
+      }
+    }
+  }
+
+  // Reconstruct the best path. We do it in reverse order.
+  std::vector<EditType> best_path;
+  for (size_t l_i = left.size(), r_i = right.size(); l_i > 0 || r_i > 0;) {
+    EditType move = best_move[l_i][r_i];
+    best_path.push_back(move);
+    l_i -= move != kAdd;
+    r_i -= move != kRemove;
+  }
+  std::reverse(best_path.begin(), best_path.end());
+  return best_path;
+}
+
+namespace {
+
+// Helper class to convert string into ids with deduplication.
+class InternalStrings {
+ public:
+  size_t GetId(const std::string& str) {
+    IdMap::iterator it = ids_.find(str);
+    if (it != ids_.end()) return it->second;
+    size_t id = ids_.size();
+    return ids_[str] = id;
+  }
+
+ private:
+  typedef std::map<std::string, size_t> IdMap;
+  IdMap ids_;
+};
+
+}  // namespace
+
+std::vector<EditType> CalculateOptimalEdits(
+    const std::vector<std::string>& left,
+    const std::vector<std::string>& right) {
+  std::vector<size_t> left_ids, right_ids;
+  {
+    InternalStrings intern_table;
+    for (size_t i = 0; i < left.size(); ++i) {
+      left_ids.push_back(intern_table.GetId(left[i]));
+    }
+    for (size_t i = 0; i < right.size(); ++i) {
+      right_ids.push_back(intern_table.GetId(right[i]));
+    }
+  }
+  return CalculateOptimalEdits(left_ids, right_ids);
+}
+
+namespace {
+
+// Helper class that holds the state for one hunk and prints it out to the
+// stream.
+// It reorders adds/removes when possible to group all removes before all
+// adds. It also adds the hunk header before printint into the stream.
+class Hunk {
+ public:
+  Hunk(size_t left_start, size_t right_start)
+      : left_start_(left_start),
+        right_start_(right_start),
+        adds_(),
+        removes_(),
+        common_() {}
+
+  void PushLine(char edit, const char* line) {
+    switch (edit) {
+      case ' ':
+        ++common_;
+        FlushEdits();
+        hunk_.push_back(std::make_pair(' ', line));
+        break;
+      case '-':
+        ++removes_;
+        hunk_removes_.push_back(std::make_pair('-', line));
+        break;
+      case '+':
+        ++adds_;
+        hunk_adds_.push_back(std::make_pair('+', line));
+        break;
+    }
+  }
+
+  void PrintTo(std::ostream* os) {
+    PrintHeader(os);
+    FlushEdits();
+    for (std::list<std::pair<char, const char*> >::const_iterator it =
+             hunk_.begin();
+         it != hunk_.end(); ++it) {
+      *os << it->first << it->second << "\n";
+    }
+  }
+
+  bool has_edits() const { return adds_ || removes_; }
+
+ private:
+  void FlushEdits() {
+    hunk_.splice(hunk_.end(), hunk_removes_);
+    hunk_.splice(hunk_.end(), hunk_adds_);
+  }
+
+  // Print a unified diff header for one hunk.
+  // The format is
+  //   "@@ -<left_start>,<left_length> +<right_start>,<right_length> @@"
+  // where the left/right parts are omitted if unnecessary.
+  void PrintHeader(std::ostream* ss) const {
+    *ss << "@@ ";
+    if (removes_) {
+      *ss << "-" << left_start_ << "," << (removes_ + common_);
+    }
+    if (removes_ && adds_) {
+      *ss << " ";
+    }
+    if (adds_) {
+      *ss << "+" << right_start_ << "," << (adds_ + common_);
+    }
+    *ss << " @@\n";
+  }
+
+  size_t left_start_, right_start_;
+  size_t adds_, removes_, common_;
+  std::list<std::pair<char, const char*> > hunk_, hunk_adds_, hunk_removes_;
+};
+
+}  // namespace
+
+// Create a list of diff hunks in Unified diff format.
+// Each hunk has a header generated by PrintHeader above plus a body with
+// lines prefixed with ' ' for no change, '-' for deletion and '+' for
+// addition.
+// 'context' represents the desired unchanged prefix/suffix around the diff.
+// If two hunks are close enough that their contexts overlap, then they are
+// joined into one hunk.
+std::string CreateUnifiedDiff(const std::vector<std::string>& left,
+                              const std::vector<std::string>& right,
+                              size_t context) {
+  const std::vector<EditType> edits = CalculateOptimalEdits(left, right);
+
+  size_t l_i = 0, r_i = 0, edit_i = 0;
+  std::stringstream ss;
+  while (edit_i < edits.size()) {
+    // Find first edit.
+    while (edit_i < edits.size() && edits[edit_i] == kMatch) {
+      ++l_i;
+      ++r_i;
+      ++edit_i;
+    }
+
+    // Find the first line to include in the hunk.
+    const size_t prefix_context = std::min(l_i, context);
+    Hunk hunk(l_i - prefix_context + 1, r_i - prefix_context + 1);
+    for (size_t i = prefix_context; i > 0; --i) {
+      hunk.PushLine(' ', left[l_i - i].c_str());
+    }
+
+    // Iterate the edits until we found enough suffix for the hunk or the input
+    // is over.
+    size_t n_suffix = 0;
+    for (; edit_i < edits.size(); ++edit_i) {
+      if (n_suffix >= context) {
+        // Continue only if the next hunk is very close.
+        std::vector<EditType>::const_iterator it = edits.begin() + edit_i;
+        while (it != edits.end() && *it == kMatch) ++it;
+        if (it == edits.end() || (it - edits.begin()) - edit_i >= context) {
+          // There is no next edit or it is too far away.
+          break;
+        }
+      }
+
+      EditType edit = edits[edit_i];
+      // Reset count when a non match is found.
+      n_suffix = edit == kMatch ? n_suffix + 1 : 0;
+
+      if (edit == kMatch || edit == kRemove || edit == kReplace) {
+        hunk.PushLine(edit == kMatch ? ' ' : '-', left[l_i].c_str());
+      }
+      if (edit == kAdd || edit == kReplace) {
+        hunk.PushLine('+', right[r_i].c_str());
+      }
+
+      // Advance indices, depending on edit type.
+      l_i += edit != kAdd;
+      r_i += edit != kRemove;
+    }
+
+    if (!hunk.has_edits()) {
+      // We are done. We don't want this hunk.
+      break;
+    }
+
+    hunk.PrintTo(&ss);
+  }
+  return ss.str();
+}
+
+}  // namespace edit_distance
+
+namespace {
+
+// The string representation of the values received in EqFailure() are already
+// escaped. Split them on escaped '\n' boundaries. Leave all other escaped
+// characters the same.
+std::vector<std::string> SplitEscapedString(const std::string& str) {
+  std::vector<std::string> lines;
+  size_t start = 0, end = str.size();
+  if (end > 2 && str[0] == '"' && str[end - 1] == '"') {
+    ++start;
+    --end;
+  }
+  bool escaped = false;
+  for (size_t i = start; i + 1 < end; ++i) {
+    if (escaped) {
+      escaped = false;
+      if (str[i] == 'n') {
+        lines.push_back(str.substr(start, i - start - 1));
+        start = i + 1;
+      }
+    } else {
+      escaped = str[i] == '\\';
+    }
+  }
+  lines.push_back(str.substr(start, end - start));
+  return lines;
+}
+
+}  // namespace
+
+// Constructs and returns the message for an equality assertion
+// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure.
+//
+// The first four parameters are the expressions used in the assertion
+// and their values, as strings.  For example, for ASSERT_EQ(foo, bar)
+// where foo is 5 and bar is 6, we have:
+//
+//   lhs_expression: "foo"
+//   rhs_expression: "bar"
+//   lhs_value:      "5"
+//   rhs_value:      "6"
+//
+// The ignoring_case parameter is true iff the assertion is a
+// *_STRCASEEQ*.  When it's true, the string "Ignoring case" will
+// be inserted into the message.
+AssertionResult EqFailure(const char* lhs_expression,
+                          const char* rhs_expression,
+                          const std::string& lhs_value,
+                          const std::string& rhs_value,
+                          bool ignoring_case) {
+  Message msg;
+  msg << "      Expected: " << lhs_expression;
+  if (lhs_value != lhs_expression) {
+    msg << "\n      Which is: " << lhs_value;
+  }
+  msg << "\nTo be equal to: " << rhs_expression;
+  if (rhs_value != rhs_expression) {
+    msg << "\n      Which is: " << rhs_value;
+  }
+
+  if (ignoring_case) {
+    msg << "\nIgnoring case";
+  }
+
+  if (!lhs_value.empty() && !rhs_value.empty()) {
+    const std::vector<std::string> lhs_lines =
+        SplitEscapedString(lhs_value);
+    const std::vector<std::string> rhs_lines =
+        SplitEscapedString(rhs_value);
+    if (lhs_lines.size() > 1 || rhs_lines.size() > 1) {
+      msg << "\nWith diff:\n"
+          << edit_distance::CreateUnifiedDiff(lhs_lines, rhs_lines);
+    }
+  }
+
+  return AssertionFailure() << msg;
+}
+
+// Constructs a failure message for Boolean assertions such as EXPECT_TRUE.
+std::string GetBoolAssertionFailureMessage(
+    const AssertionResult& assertion_result,
+    const char* expression_text,
+    const char* actual_predicate_value,
+    const char* expected_predicate_value) {
+  const char* actual_message = assertion_result.message();
+  Message msg;
+  msg << "Value of: " << expression_text
+      << "\n  Actual: " << actual_predicate_value;
+  if (actual_message[0] != '\0')
+    msg << " (" << actual_message << ")";
+  msg << "\nExpected: " << expected_predicate_value;
+  return msg.GetString();
+}
+
+// Helper function for implementing ASSERT_NEAR.
+AssertionResult DoubleNearPredFormat(const char* expr1,
+                                     const char* expr2,
+                                     const char* abs_error_expr,
+                                     double val1,
+                                     double val2,
+                                     double abs_error) {
+  const double diff = fabs(val1 - val2);
+  if (diff <= abs_error) return AssertionSuccess();
+
+  // TODO(wan): do not print the value of an expression if it's
+  // already a literal.
+  return AssertionFailure()
+      << "The difference between " << expr1 << " and " << expr2
+      << " is " << diff << ", which exceeds " << abs_error_expr << ", where\n"
+      << expr1 << " evaluates to " << val1 << ",\n"
+      << expr2 << " evaluates to " << val2 << ", and\n"
+      << abs_error_expr << " evaluates to " << abs_error << ".";
+}
+
+
+// Helper template for implementing FloatLE() and DoubleLE().
+template <typename RawType>
+AssertionResult FloatingPointLE(const char* expr1,
+                                const char* expr2,
+                                RawType val1,
+                                RawType val2) {
+  // Returns success if val1 is less than val2,
+  if (val1 < val2) {
+    return AssertionSuccess();
+  }
+
+  // or if val1 is almost equal to val2.
+  const FloatingPoint<RawType> lhs(val1), rhs(val2);
+  if (lhs.AlmostEquals(rhs)) {
+    return AssertionSuccess();
+  }
+
+  // Note that the above two checks will both fail if either val1 or
+  // val2 is NaN, as the IEEE floating-point standard requires that
+  // any predicate involving a NaN must return false.
+
+  ::std::stringstream val1_ss;
+  val1_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+          << val1;
+
+  ::std::stringstream val2_ss;
+  val2_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+          << val2;
+
+  return AssertionFailure()
+      << "Expected: (" << expr1 << ") <= (" << expr2 << ")\n"
+      << "  Actual: " << StringStreamToString(&val1_ss) << " vs "
+      << StringStreamToString(&val2_ss);
+}
+
+}  // namespace internal
+
+// Asserts that val1 is less than, or almost equal to, val2.  Fails
+// otherwise.  In particular, it fails if either val1 or val2 is NaN.
+AssertionResult FloatLE(const char* expr1, const char* expr2,
+                        float val1, float val2) {
+  return internal::FloatingPointLE<float>(expr1, expr2, val1, val2);
+}
+
+// Asserts that val1 is less than, or almost equal to, val2.  Fails
+// otherwise.  In particular, it fails if either val1 or val2 is NaN.
+AssertionResult DoubleLE(const char* expr1, const char* expr2,
+                         double val1, double val2) {
+  return internal::FloatingPointLE<double>(expr1, expr2, val1, val2);
+}
+
+namespace internal {
+
+// The helper function for {ASSERT|EXPECT}_EQ with int or enum
+// arguments.
+AssertionResult CmpHelperEQ(const char* lhs_expression,
+                            const char* rhs_expression,
+                            BiggestInt lhs,
+                            BiggestInt rhs) {
+  if (lhs == rhs) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   FormatForComparisonFailureMessage(lhs, rhs),
+                   FormatForComparisonFailureMessage(rhs, lhs),
+                   false);
+}
+
+// A macro for implementing the helper functions needed to implement
+// ASSERT_?? and EXPECT_?? with integer or enum arguments.  It is here
+// just to avoid copy-and-paste of similar code.
+#define GTEST_IMPL_CMP_HELPER_(op_name, op)\
+AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \
+                                   BiggestInt val1, BiggestInt val2) {\
+  if (val1 op val2) {\
+    return AssertionSuccess();\
+  } else {\
+    return AssertionFailure() \
+        << "Expected: (" << expr1 << ") " #op " (" << expr2\
+        << "), actual: " << FormatForComparisonFailureMessage(val1, val2)\
+        << " vs " << FormatForComparisonFailureMessage(val2, val1);\
+  }\
+}
+
+// Implements the helper function for {ASSERT|EXPECT}_NE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(NE, !=)
+// Implements the helper function for {ASSERT|EXPECT}_LE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(LE, <=)
+// Implements the helper function for {ASSERT|EXPECT}_LT with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(LT, < )
+// Implements the helper function for {ASSERT|EXPECT}_GE with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(GE, >=)
+// Implements the helper function for {ASSERT|EXPECT}_GT with int or
+// enum arguments.
+GTEST_IMPL_CMP_HELPER_(GT, > )
+
+#undef GTEST_IMPL_CMP_HELPER_
+
+// The helper function for {ASSERT|EXPECT}_STREQ.
+AssertionResult CmpHelperSTREQ(const char* lhs_expression,
+                               const char* rhs_expression,
+                               const char* lhs,
+                               const char* rhs) {
+  if (String::CStringEquals(lhs, rhs)) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   PrintToString(lhs),
+                   PrintToString(rhs),
+                   false);
+}
+
+// The helper function for {ASSERT|EXPECT}_STRCASEEQ.
+AssertionResult CmpHelperSTRCASEEQ(const char* lhs_expression,
+                                   const char* rhs_expression,
+                                   const char* lhs,
+                                   const char* rhs) {
+  if (String::CaseInsensitiveCStringEquals(lhs, rhs)) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   PrintToString(lhs),
+                   PrintToString(rhs),
+                   true);
+}
+
+// The helper function for {ASSERT|EXPECT}_STRNE.
+AssertionResult CmpHelperSTRNE(const char* s1_expression,
+                               const char* s2_expression,
+                               const char* s1,
+                               const char* s2) {
+  if (!String::CStringEquals(s1, s2)) {
+    return AssertionSuccess();
+  } else {
+    return AssertionFailure() << "Expected: (" << s1_expression << ") != ("
+                              << s2_expression << "), actual: \""
+                              << s1 << "\" vs \"" << s2 << "\"";
+  }
+}
+
+// The helper function for {ASSERT|EXPECT}_STRCASENE.
+AssertionResult CmpHelperSTRCASENE(const char* s1_expression,
+                                   const char* s2_expression,
+                                   const char* s1,
+                                   const char* s2) {
+  if (!String::CaseInsensitiveCStringEquals(s1, s2)) {
+    return AssertionSuccess();
+  } else {
+    return AssertionFailure()
+        << "Expected: (" << s1_expression << ") != ("
+        << s2_expression << ") (ignoring case), actual: \""
+        << s1 << "\" vs \"" << s2 << "\"";
+  }
+}
+
+}  // namespace internal
+
+namespace {
+
+// Helper functions for implementing IsSubString() and IsNotSubstring().
+
+// This group of overloaded functions return true iff needle is a
+// substring of haystack.  NULL is considered a substring of itself
+// only.
+
+bool IsSubstringPred(const char* needle, const char* haystack) {
+  if (needle == NULL || haystack == NULL)
+    return needle == haystack;
+
+  return strstr(haystack, needle) != NULL;
+}
+
+bool IsSubstringPred(const wchar_t* needle, const wchar_t* haystack) {
+  if (needle == NULL || haystack == NULL)
+    return needle == haystack;
+
+  return wcsstr(haystack, needle) != NULL;
+}
+
+// StringType here can be either ::std::string or ::std::wstring.
+template <typename StringType>
+bool IsSubstringPred(const StringType& needle,
+                     const StringType& haystack) {
+  return haystack.find(needle) != StringType::npos;
+}
+
+// This function implements either IsSubstring() or IsNotSubstring(),
+// depending on the value of the expected_to_be_substring parameter.
+// StringType here can be const char*, const wchar_t*, ::std::string,
+// or ::std::wstring.
+template <typename StringType>
+AssertionResult IsSubstringImpl(
+    bool expected_to_be_substring,
+    const char* needle_expr, const char* haystack_expr,
+    const StringType& needle, const StringType& haystack) {
+  if (IsSubstringPred(needle, haystack) == expected_to_be_substring)
+    return AssertionSuccess();
+
+  const bool is_wide_string = sizeof(needle[0]) > 1;
+  const char* const begin_string_quote = is_wide_string ? "L\"" : "\"";
+  return AssertionFailure()
+      << "Value of: " << needle_expr << "\n"
+      << "  Actual: " << begin_string_quote << needle << "\"\n"
+      << "Expected: " << (expected_to_be_substring ? "" : "not ")
+      << "a substring of " << haystack_expr << "\n"
+      << "Which is: " << begin_string_quote << haystack << "\"";
+}
+
+}  // namespace
+
+// IsSubstring() and IsNotSubstring() check whether needle is a
+// substring of haystack (NULL is considered a substring of itself
+// only), and return an appropriate error message when they fail.
+
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const char* needle, const char* haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const wchar_t* needle, const wchar_t* haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const char* needle, const char* haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const wchar_t* needle, const wchar_t* haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::string& needle, const ::std::string& haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::string& needle, const ::std::string& haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+
+#if GTEST_HAS_STD_WSTRING
+AssertionResult IsSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::wstring& needle, const ::std::wstring& haystack) {
+  return IsSubstringImpl(true, needle_expr, haystack_expr, needle, haystack);
+}
+
+AssertionResult IsNotSubstring(
+    const char* needle_expr, const char* haystack_expr,
+    const ::std::wstring& needle, const ::std::wstring& haystack) {
+  return IsSubstringImpl(false, needle_expr, haystack_expr, needle, haystack);
+}
+#endif  // GTEST_HAS_STD_WSTRING
+
+namespace internal {
+
+#if GTEST_OS_WINDOWS
+
+namespace {
+
+// Helper function for IsHRESULT{SuccessFailure} predicates
+AssertionResult HRESULTFailureHelper(const char* expr,
+                                     const char* expected,
+                                     long hr) {  // NOLINT
+# if GTEST_OS_WINDOWS_MOBILE
+
+  // Windows CE doesn't support FormatMessage.
+  const char error_text[] = "";
+
+# else
+
+  // Looks up the human-readable system message for the HRESULT code
+  // and since we're not passing any params to FormatMessage, we don't
+  // want inserts expanded.
+  const DWORD kFlags = FORMAT_MESSAGE_FROM_SYSTEM |
+                       FORMAT_MESSAGE_IGNORE_INSERTS;
+  const DWORD kBufSize = 4096;
+  // Gets the system's human readable message string for this HRESULT.
+  char error_text[kBufSize] = { '\0' };
+  DWORD message_length = ::FormatMessageA(kFlags,
+                                          0,  // no source, we're asking system
+                                          hr,  // the error
+                                          0,  // no line width restrictions
+                                          error_text,  // output buffer
+                                          kBufSize,  // buf size
+                                          NULL);  // no arguments for inserts
+  // Trims tailing white space (FormatMessage leaves a trailing CR-LF)
+  for (; message_length && IsSpace(error_text[message_length - 1]);
+          --message_length) {
+    error_text[message_length - 1] = '\0';
+  }
+
+# endif  // GTEST_OS_WINDOWS_MOBILE
+
+  const std::string error_hex("0x" + String::FormatHexInt(hr));
+  return ::testing::AssertionFailure()
+      << "Expected: " << expr << " " << expected << ".\n"
+      << "  Actual: " << error_hex << " " << error_text << "\n";
+}
+
+}  // namespace
+
+AssertionResult IsHRESULTSuccess(const char* expr, long hr) {  // NOLINT
+  if (SUCCEEDED(hr)) {
+    return AssertionSuccess();
+  }
+  return HRESULTFailureHelper(expr, "succeeds", hr);
+}
+
+AssertionResult IsHRESULTFailure(const char* expr, long hr) {  // NOLINT
+  if (FAILED(hr)) {
+    return AssertionSuccess();
+  }
+  return HRESULTFailureHelper(expr, "fails", hr);
+}
+
+#endif  // GTEST_OS_WINDOWS
+
+// Utility functions for encoding Unicode text (wide strings) in
+// UTF-8.
+
+// A Unicode code-point can have up to 21 bits, and is encoded in UTF-8
+// like this:
+//
+// Code-point length   Encoding
+//   0 -  7 bits       0xxxxxxx
+//   8 - 11 bits       110xxxxx 10xxxxxx
+//  12 - 16 bits       1110xxxx 10xxxxxx 10xxxxxx
+//  17 - 21 bits       11110xxx 10xxxxxx 10xxxxxx 10xxxxxx
+
+// The maximum code-point a one-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint1 = (static_cast<UInt32>(1) <<  7) - 1;
+
+// The maximum code-point a two-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint2 = (static_cast<UInt32>(1) << (5 + 6)) - 1;
+
+// The maximum code-point a three-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint3 = (static_cast<UInt32>(1) << (4 + 2*6)) - 1;
+
+// The maximum code-point a four-byte UTF-8 sequence can represent.
+const UInt32 kMaxCodePoint4 = (static_cast<UInt32>(1) << (3 + 3*6)) - 1;
+
+// Chops off the n lowest bits from a bit pattern.  Returns the n
+// lowest bits.  As a side effect, the original bit pattern will be
+// shifted to the right by n bits.
+inline UInt32 ChopLowBits(UInt32* bits, int n) {
+  const UInt32 low_bits = *bits & ((static_cast<UInt32>(1) << n) - 1);
+  *bits >>= n;
+  return low_bits;
+}
+
+// Converts a Unicode code point to a narrow string in UTF-8 encoding.
+// code_point parameter is of type UInt32 because wchar_t may not be
+// wide enough to contain a code point.
+// If the code_point is not a valid Unicode code point
+// (i.e. outside of Unicode range U+0 to U+10FFFF) it will be converted
+// to "(Invalid Unicode 0xXXXXXXXX)".
+std::string CodePointToUtf8(UInt32 code_point) {
+  if (code_point > kMaxCodePoint4) {
+    return "(Invalid Unicode 0x" + String::FormatHexInt(code_point) + ")";
+  }
+
+  char str[5];  // Big enough for the largest valid code point.
+  if (code_point <= kMaxCodePoint1) {
+    str[1] = '\0';
+    str[0] = static_cast<char>(code_point);                          // 0xxxxxxx
+  } else if (code_point <= kMaxCodePoint2) {
+    str[2] = '\0';
+    str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[0] = static_cast<char>(0xC0 | code_point);                   // 110xxxxx
+  } else if (code_point <= kMaxCodePoint3) {
+    str[3] = '\0';
+    str[2] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[0] = static_cast<char>(0xE0 | code_point);                   // 1110xxxx
+  } else {  // code_point <= kMaxCodePoint4
+    str[4] = '\0';
+    str[3] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[2] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[1] = static_cast<char>(0x80 | ChopLowBits(&code_point, 6));  // 10xxxxxx
+    str[0] = static_cast<char>(0xF0 | code_point);                   // 11110xxx
+  }
+  return str;
+}
+
+// The following two functions only make sense if the the system
+// uses UTF-16 for wide string encoding. All supported systems
+// with 16 bit wchar_t (Windows, Cygwin, Symbian OS) do use UTF-16.
+
+// Determines if the arguments constitute UTF-16 surrogate pair
+// and thus should be combined into a single Unicode code point
+// using CreateCodePointFromUtf16SurrogatePair.
+inline bool IsUtf16SurrogatePair(wchar_t first, wchar_t second) {
+  return sizeof(wchar_t) == 2 &&
+      (first & 0xFC00) == 0xD800 && (second & 0xFC00) == 0xDC00;
+}
+
+// Creates a Unicode code point from UTF16 surrogate pair.
+inline UInt32 CreateCodePointFromUtf16SurrogatePair(wchar_t first,
+                                                    wchar_t second) {
+  const UInt32 mask = (1 << 10) - 1;
+  return (sizeof(wchar_t) == 2) ?
+      (((first & mask) << 10) | (second & mask)) + 0x10000 :
+      // This function should not be called when the condition is
+      // false, but we provide a sensible default in case it is.
+      static_cast<UInt32>(first);
+}
+
+// Converts a wide string to a narrow string in UTF-8 encoding.
+// The wide string is assumed to have the following encoding:
+//   UTF-16 if sizeof(wchar_t) == 2 (on Windows, Cygwin, Symbian OS)
+//   UTF-32 if sizeof(wchar_t) == 4 (on Linux)
+// Parameter str points to a null-terminated wide string.
+// Parameter num_chars may additionally limit the number
+// of wchar_t characters processed. -1 is used when the entire string
+// should be processed.
+// If the string contains code points that are not valid Unicode code points
+// (i.e. outside of Unicode range U+0 to U+10FFFF) they will be output
+// as '(Invalid Unicode 0xXXXXXXXX)'. If the string is in UTF16 encoding
+// and contains invalid UTF-16 surrogate pairs, values in those pairs
+// will be encoded as individual Unicode characters from Basic Normal Plane.
+std::string WideStringToUtf8(const wchar_t* str, int num_chars) {
+  if (num_chars == -1)
+    num_chars = static_cast<int>(wcslen(str));
+
+  ::std::stringstream stream;
+  for (int i = 0; i < num_chars; ++i) {
+    UInt32 unicode_code_point;
+
+    if (str[i] == L'\0') {
+      break;
+    } else if (i + 1 < num_chars && IsUtf16SurrogatePair(str[i], str[i + 1])) {
+      unicode_code_point = CreateCodePointFromUtf16SurrogatePair(str[i],
+                                                                 str[i + 1]);
+      i++;
+    } else {
+      unicode_code_point = static_cast<UInt32>(str[i]);
+    }
+
+    stream << CodePointToUtf8(unicode_code_point);
+  }
+  return StringStreamToString(&stream);
+}
+
+// Converts a wide C string to an std::string using the UTF-8 encoding.
+// NULL will be converted to "(null)".
+std::string String::ShowWideCString(const wchar_t * wide_c_str) {
+  if (wide_c_str == NULL)  return "(null)";
+
+  return internal::WideStringToUtf8(wide_c_str, -1);
+}
+
+// Compares two wide C strings.  Returns true iff they have the same
+// content.
+//
+// Unlike wcscmp(), this function can handle NULL argument(s).  A NULL
+// C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::WideCStringEquals(const wchar_t * lhs, const wchar_t * rhs) {
+  if (lhs == NULL) return rhs == NULL;
+
+  if (rhs == NULL) return false;
+
+  return wcscmp(lhs, rhs) == 0;
+}
+
+// Helper function for *_STREQ on wide strings.
+AssertionResult CmpHelperSTREQ(const char* lhs_expression,
+                               const char* rhs_expression,
+                               const wchar_t* lhs,
+                               const wchar_t* rhs) {
+  if (String::WideCStringEquals(lhs, rhs)) {
+    return AssertionSuccess();
+  }
+
+  return EqFailure(lhs_expression,
+                   rhs_expression,
+                   PrintToString(lhs),
+                   PrintToString(rhs),
+                   false);
+}
+
+// Helper function for *_STRNE on wide strings.
+AssertionResult CmpHelperSTRNE(const char* s1_expression,
+                               const char* s2_expression,
+                               const wchar_t* s1,
+                               const wchar_t* s2) {
+  if (!String::WideCStringEquals(s1, s2)) {
+    return AssertionSuccess();
+  }
+
+  return AssertionFailure() << "Expected: (" << s1_expression << ") != ("
+                            << s2_expression << "), actual: "
+                            << PrintToString(s1)
+                            << " vs " << PrintToString(s2);
+}
+
+// Compares two C strings, ignoring case.  Returns true iff they have
+// the same content.
+//
+// Unlike strcasecmp(), this function can handle NULL argument(s).  A
+// NULL C string is considered different to any non-NULL C string,
+// including the empty string.
+bool String::CaseInsensitiveCStringEquals(const char * lhs, const char * rhs) {
+  if (lhs == NULL)
+    return rhs == NULL;
+  if (rhs == NULL)
+    return false;
+  return posix::StrCaseCmp(lhs, rhs) == 0;
+}
+
+  // Compares two wide C strings, ignoring case.  Returns true iff they
+  // have the same content.
+  //
+  // Unlike wcscasecmp(), this function can handle NULL argument(s).
+  // A NULL C string is considered different to any non-NULL wide C string,
+  // including the empty string.
+  // NB: The implementations on different platforms slightly differ.
+  // On windows, this method uses _wcsicmp which compares according to LC_CTYPE
+  // environment variable. On GNU platform this method uses wcscasecmp
+  // which compares according to LC_CTYPE category of the current locale.
+  // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the
+  // current locale.
+bool String::CaseInsensitiveWideCStringEquals(const wchar_t* lhs,
+                                              const wchar_t* rhs) {
+  if (lhs == NULL) return rhs == NULL;
+
+  if (rhs == NULL) return false;
+
+#if GTEST_OS_WINDOWS
+  return _wcsicmp(lhs, rhs) == 0;
+#elif GTEST_OS_LINUX && !GTEST_OS_LINUX_ANDROID
+  return wcscasecmp(lhs, rhs) == 0;
+#else
+  // Android, Mac OS X and Cygwin don't define wcscasecmp.
+  // Other unknown OSes may not define it either.
+  wint_t left, right;
+  do {
+    left = towlower(*lhs++);
+    right = towlower(*rhs++);
+  } while (left && left == right);
+  return left == right;
+#endif  // OS selector
+}
+
+// Returns true iff str ends with the given suffix, ignoring case.
+// Any string is considered to end with an empty suffix.
+bool String::EndsWithCaseInsensitive(
+    const std::string& str, const std::string& suffix) {
+  const size_t str_len = str.length();
+  const size_t suffix_len = suffix.length();
+  return (str_len >= suffix_len) &&
+         CaseInsensitiveCStringEquals(str.c_str() + str_len - suffix_len,
+                                      suffix.c_str());
+}
+
+// Formats an int value as "%02d".
+std::string String::FormatIntWidth2(int value) {
+  std::stringstream ss;
+  ss << std::setfill('0') << std::setw(2) << value;
+  return ss.str();
+}
+
+// Formats an int value as "%X".
+std::string String::FormatHexInt(int value) {
+  std::stringstream ss;
+  ss << std::hex << std::uppercase << value;
+  return ss.str();
+}
+
+// Formats a byte as "%02X".
+std::string String::FormatByte(unsigned char value) {
+  std::stringstream ss;
+  ss << std::setfill('0') << std::setw(2) << std::hex << std::uppercase
+     << static_cast<unsigned int>(value);
+  return ss.str();
+}
+
+// Converts the buffer in a stringstream to an std::string, converting NUL
+// bytes to "\\0" along the way.
+std::string StringStreamToString(::std::stringstream* ss) {
+  const ::std::string& str = ss->str();
+  const char* const start = str.c_str();
+  const char* const end = start + str.length();
+
+  std::string result;
+  result.reserve(2 * (end - start));
+  for (const char* ch = start; ch != end; ++ch) {
+    if (*ch == '\0') {
+      result += "\\0";  // Replaces NUL with "\\0";
+    } else {
+      result += *ch;
+    }
+  }
+
+  return result;
+}
+
+// Appends the user-supplied message to the Google-Test-generated message.
+std::string AppendUserMessage(const std::string& gtest_msg,
+                              const Message& user_msg) {
+  // Appends the user message if it's non-empty.
+  const std::string user_msg_string = user_msg.GetString();
+  if (user_msg_string.empty()) {
+    return gtest_msg;
+  }
+
+  return gtest_msg + "\n" + user_msg_string;
+}
+
+}  // namespace internal
+
+// class TestResult
+
+// Creates an empty TestResult.
+TestResult::TestResult()
+    : death_test_count_(0),
+      elapsed_time_(0) {
+}
+
+// D'tor.
+TestResult::~TestResult() {
+}
+
+// Returns the i-th test part result among all the results. i can
+// range from 0 to total_part_count() - 1. If i is not in that range,
+// aborts the program.
+const TestPartResult& TestResult::GetTestPartResult(int i) const {
+  if (i < 0 || i >= total_part_count())
+    internal::posix::Abort();
+  return test_part_results_.at(i);
+}
+
+// Returns the i-th test property. i can range from 0 to
+// test_property_count() - 1. If i is not in that range, aborts the
+// program.
+const TestProperty& TestResult::GetTestProperty(int i) const {
+  if (i < 0 || i >= test_property_count())
+    internal::posix::Abort();
+  return test_properties_.at(i);
+}
+
+// Clears the test part results.
+void TestResult::ClearTestPartResults() {
+  test_part_results_.clear();
+}
+
+// Adds a test part result to the list.
+void TestResult::AddTestPartResult(const TestPartResult& test_part_result) {
+  test_part_results_.push_back(test_part_result);
+}
+
+// Adds a test property to the list. If a property with the same key as the
+// supplied property is already represented, the value of this test_property
+// replaces the old value for that key.
+void TestResult::RecordProperty(const std::string& xml_element,
+                                const TestProperty& test_property) {
+  if (!ValidateTestProperty(xml_element, test_property)) {
+    return;
+  }
+  internal::MutexLock lock(&test_properites_mutex_);
+  const std::vector<TestProperty>::iterator property_with_matching_key =
+      std::find_if(test_properties_.begin(), test_properties_.end(),
+                   internal::TestPropertyKeyIs(test_property.key()));
+  if (property_with_matching_key == test_properties_.end()) {
+    test_properties_.push_back(test_property);
+    return;
+  }
+  property_with_matching_key->SetValue(test_property.value());
+}
+
+// The list of reserved attributes used in the <testsuites> element of XML
+// output.
+static const char* const kReservedTestSuitesAttributes[] = {
+  "disabled",
+  "errors",
+  "failures",
+  "name",
+  "random_seed",
+  "tests",
+  "time",
+  "timestamp"
+};
+
+// The list of reserved attributes used in the <testsuite> element of XML
+// output.
+static const char* const kReservedTestSuiteAttributes[] = {
+  "disabled",
+  "errors",
+  "failures",
+  "name",
+  "tests",
+  "time"
+};
+
+// The list of reserved attributes used in the <testcase> element of XML output.
+static const char* const kReservedTestCaseAttributes[] = {
+  "classname",
+  "name",
+  "status",
+  "time",
+  "type_param",
+  "value_param"
+};
+
+template <int kSize>
+std::vector<std::string> ArrayAsVector(const char* const (&array)[kSize]) {
+  return std::vector<std::string>(array, array + kSize);
+}
+
+static std::vector<std::string> GetReservedAttributesForElement(
+    const std::string& xml_element) {
+  if (xml_element == "testsuites") {
+    return ArrayAsVector(kReservedTestSuitesAttributes);
+  } else if (xml_element == "testsuite") {
+    return ArrayAsVector(kReservedTestSuiteAttributes);
+  } else if (xml_element == "testcase") {
+    return ArrayAsVector(kReservedTestCaseAttributes);
+  } else {
+    GTEST_CHECK_(false) << "Unrecognized xml_element provided: " << xml_element;
+  }
+  // This code is unreachable but some compilers may not realizes that.
+  return std::vector<std::string>();
+}
+
+static std::string FormatWordList(const std::vector<std::string>& words) {
+  Message word_list;
+  for (size_t i = 0; i < words.size(); ++i) {
+    if (i > 0 && words.size() > 2) {
+      word_list << ", ";
+    }
+    if (i == words.size() - 1) {
+      word_list << "and ";
+    }
+    word_list << "'" << words[i] << "'";
+  }
+  return word_list.GetString();
+}
+
+bool ValidateTestPropertyName(const std::string& property_name,
+                              const std::vector<std::string>& reserved_names) {
+  if (std::find(reserved_names.begin(), reserved_names.end(), property_name) !=
+          reserved_names.end()) {
+    ADD_FAILURE() << "Reserved key used in RecordProperty(): " << property_name
+                  << " (" << FormatWordList(reserved_names)
+                  << " are reserved by " << GTEST_NAME_ << ")";
+    return false;
+  }
+  return true;
+}
+
+// Adds a failure if the key is a reserved attribute of the element named
+// xml_element.  Returns true if the property is valid.
+bool TestResult::ValidateTestProperty(const std::string& xml_element,
+                                      const TestProperty& test_property) {
+  return ValidateTestPropertyName(test_property.key(),
+                                  GetReservedAttributesForElement(xml_element));
+}
+
+// Clears the object.
+void TestResult::Clear() {
+  test_part_results_.clear();
+  test_properties_.clear();
+  death_test_count_ = 0;
+  elapsed_time_ = 0;
+}
+
+// Returns true off the test part was skipped.
+static bool TestPartSkipped(const TestPartResult& result) {
+  return result.skipped();
+}
+
+// Returns true if and only if the test was skipped.
+bool TestResult::Skipped() const {
+  return !Failed() && CountIf(test_part_results_, TestPartSkipped) > 0;
+}
+
+// Returns true iff the test failed.
+bool TestResult::Failed() const {
+  for (int i = 0; i < total_part_count(); ++i) {
+    if (GetTestPartResult(i).failed())
+      return true;
+  }
+  return false;
+}
+
+// Returns true iff the test part fatally failed.
+static bool TestPartFatallyFailed(const TestPartResult& result) {
+  return result.fatally_failed();
+}
+
+// Returns true iff the test fatally failed.
+bool TestResult::HasFatalFailure() const {
+  return CountIf(test_part_results_, TestPartFatallyFailed) > 0;
+}
+
+// Returns true iff the test part non-fatally failed.
+static bool TestPartNonfatallyFailed(const TestPartResult& result) {
+  return result.nonfatally_failed();
+}
+
+// Returns true iff the test has a non-fatal failure.
+bool TestResult::HasNonfatalFailure() const {
+  return CountIf(test_part_results_, TestPartNonfatallyFailed) > 0;
+}
+
+// Gets the number of all test parts.  This is the sum of the number
+// of successful test parts and the number of failed test parts.
+int TestResult::total_part_count() const {
+  return static_cast<int>(test_part_results_.size());
+}
+
+// Returns the number of the test properties.
+int TestResult::test_property_count() const {
+  return static_cast<int>(test_properties_.size());
+}
+
+// class Test
+
+// Creates a Test object.
+
+// The c'tor saves the states of all flags.
+Test::Test()
+    : gtest_flag_saver_(new GTEST_FLAG_SAVER_) {
+}
+
+// The d'tor restores the states of all flags.  The actual work is
+// done by the d'tor of the gtest_flag_saver_ field, and thus not
+// visible here.
+Test::~Test() {
+}
+
+// Sets up the test fixture.
+//
+// A sub-class may override this.
+void Test::SetUp() {
+}
+
+// Tears down the test fixture.
+//
+// A sub-class may override this.
+void Test::TearDown() {
+}
+
+// Allows user supplied key value pairs to be recorded for later output.
+void Test::RecordProperty(const std::string& key, const std::string& value) {
+  UnitTest::GetInstance()->RecordProperty(key, value);
+}
+
+// Allows user supplied key value pairs to be recorded for later output.
+void Test::RecordProperty(const std::string& key, int value) {
+  Message value_message;
+  value_message << value;
+  RecordProperty(key, value_message.GetString().c_str());
+}
+
+namespace internal {
+
+void ReportFailureInUnknownLocation(TestPartResult::Type result_type,
+                                    const std::string& message) {
+  // This function is a friend of UnitTest and as such has access to
+  // AddTestPartResult.
+  UnitTest::GetInstance()->AddTestPartResult(
+      result_type,
+      NULL,  // No info about the source file where the exception occurred.
+      -1,    // We have no info on which line caused the exception.
+      message,
+      "");   // No stack trace, either.
+}
+
+}  // namespace internal
+
+// Google Test requires all tests in the same test case to use the same test
+// fixture class.  This function checks if the current test has the
+// same fixture class as the first test in the current test case.  If
+// yes, it returns true; otherwise it generates a Google Test failure and
+// returns false.
+bool Test::HasSameFixtureClass() {
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  const TestCase* const test_case = impl->current_test_case();
+
+  // Info about the first test in the current test case.
+  const TestInfo* const first_test_info = test_case->test_info_list()[0];
+  const internal::TypeId first_fixture_id = first_test_info->fixture_class_id_;
+  const char* const first_test_name = first_test_info->name();
+
+  // Info about the current test.
+  const TestInfo* const this_test_info = impl->current_test_info();
+  const internal::TypeId this_fixture_id = this_test_info->fixture_class_id_;
+  const char* const this_test_name = this_test_info->name();
+
+  if (this_fixture_id != first_fixture_id) {
+    // Is the first test defined using TEST?
+    const bool first_is_TEST = first_fixture_id == internal::GetTestTypeId();
+    // Is this test defined using TEST?
+    const bool this_is_TEST = this_fixture_id == internal::GetTestTypeId();
+
+    if (first_is_TEST || this_is_TEST) {
+      // Both TEST and TEST_F appear in same test case, which is incorrect.
+      // Tell the user how to fix this.
+
+      // Gets the name of the TEST and the name of the TEST_F.  Note
+      // that first_is_TEST and this_is_TEST cannot both be true, as
+      // the fixture IDs are different for the two tests.
+      const char* const TEST_name =
+          first_is_TEST ? first_test_name : this_test_name;
+      const char* const TEST_F_name =
+          first_is_TEST ? this_test_name : first_test_name;
+
+      ADD_FAILURE()
+          << "All tests in the same test case must use the same test fixture\n"
+          << "class, so mixing TEST_F and TEST in the same test case is\n"
+          << "illegal.  In test case " << this_test_info->test_case_name()
+          << ",\n"
+          << "test " << TEST_F_name << " is defined using TEST_F but\n"
+          << "test " << TEST_name << " is defined using TEST.  You probably\n"
+          << "want to change the TEST to TEST_F or move it to another test\n"
+          << "case.";
+    } else {
+      // Two fixture classes with the same name appear in two different
+      // namespaces, which is not allowed. Tell the user how to fix this.
+      ADD_FAILURE()
+          << "All tests in the same test case must use the same test fixture\n"
+          << "class.  However, in test case "
+          << this_test_info->test_case_name() << ",\n"
+          << "you defined test " << first_test_name
+          << " and test " << this_test_name << "\n"
+          << "using two different test fixture classes.  This can happen if\n"
+          << "the two classes are from different namespaces or translation\n"
+          << "units and have the same name.  You should probably rename one\n"
+          << "of the classes to put the tests into different test cases.";
+    }
+    return false;
+  }
+
+  return true;
+}
+
+#if GTEST_HAS_SEH
+
+// Adds an "exception thrown" fatal failure to the current test.  This
+// function returns its result via an output parameter pointer because VC++
+// prohibits creation of objects with destructors on stack in functions
+// using __try (see error C2712).
+static std::string* FormatSehExceptionMessage(DWORD exception_code,
+                                              const char* location) {
+  Message message;
+  message << "SEH exception with code 0x" << std::setbase(16) <<
+    exception_code << std::setbase(10) << " thrown in " << location << ".";
+
+  return new std::string(message.GetString());
+}
+
+#endif  // GTEST_HAS_SEH
+
+namespace internal {
+
+#if GTEST_HAS_EXCEPTIONS
+
+// Adds an "exception thrown" fatal failure to the current test.
+static std::string FormatCxxExceptionMessage(const char* description,
+                                             const char* location) {
+  Message message;
+  if (description != NULL) {
+    message << "C++ exception with description \"" << description << "\"";
+  } else {
+    message << "Unknown C++ exception";
+  }
+  message << " thrown in " << location << ".";
+
+  return message.GetString();
+}
+
+static std::string PrintTestPartResultToString(
+    const TestPartResult& test_part_result);
+
+GoogleTestFailureException::GoogleTestFailureException(
+    const TestPartResult& failure)
+    : ::std::runtime_error(PrintTestPartResultToString(failure).c_str()) {}
+
+#endif  // GTEST_HAS_EXCEPTIONS
+
+// We put these helper functions in the internal namespace as IBM's xlC
+// compiler rejects the code if they were declared static.
+
+// Runs the given method and handles SEH exceptions it throws, when
+// SEH is supported; returns the 0-value for type Result in case of an
+// SEH exception.  (Microsoft compilers cannot handle SEH and C++
+// exceptions in the same function.  Therefore, we provide a separate
+// wrapper function for handling SEH exceptions.)
+template <class T, typename Result>
+Result HandleSehExceptionsInMethodIfSupported(
+    T* object, Result (T::*method)(), const char* location) {
+#if GTEST_HAS_SEH
+  __try {
+    return (object->*method)();
+  } __except (internal::UnitTestOptions::GTestShouldProcessSEH(  // NOLINT
+      GetExceptionCode())) {
+    // We create the exception message on the heap because VC++ prohibits
+    // creation of objects with destructors on stack in functions using __try
+    // (see error C2712).
+    std::string* exception_message = FormatSehExceptionMessage(
+        GetExceptionCode(), location);
+    internal::ReportFailureInUnknownLocation(TestPartResult::kFatalFailure,
+                                             *exception_message);
+    delete exception_message;
+    return static_cast<Result>(0);
+  }
+#else
+  (void)location;
+  return (object->*method)();
+#endif  // GTEST_HAS_SEH
+}
+
+// Runs the given method and catches and reports C++ and/or SEH-style
+// exceptions, if they are supported; returns the 0-value for type
+// Result in case of an SEH exception.
+template <class T, typename Result>
+Result HandleExceptionsInMethodIfSupported(
+    T* object, Result (T::*method)(), const char* location) {
+  // NOTE: The user code can affect the way in which Google Test handles
+  // exceptions by setting GTEST_FLAG(catch_exceptions), but only before
+  // RUN_ALL_TESTS() starts. It is technically possible to check the flag
+  // after the exception is caught and either report or re-throw the
+  // exception based on the flag's value:
+  //
+  // try {
+  //   // Perform the test method.
+  // } catch (...) {
+  //   if (GTEST_FLAG(catch_exceptions))
+  //     // Report the exception as failure.
+  //   else
+  //     throw;  // Re-throws the original exception.
+  // }
+  //
+  // However, the purpose of this flag is to allow the program to drop into
+  // the debugger when the exception is thrown. On most platforms, once the
+  // control enters the catch block, the exception origin information is
+  // lost and the debugger will stop the program at the point of the
+  // re-throw in this function -- instead of at the point of the original
+  // throw statement in the code under test.  For this reason, we perform
+  // the check early, sacrificing the ability to affect Google Test's
+  // exception handling in the method where the exception is thrown.
+  if (internal::GetUnitTestImpl()->catch_exceptions()) {
+#if GTEST_HAS_EXCEPTIONS
+    try {
+      return HandleSehExceptionsInMethodIfSupported(object, method, location);
+    } catch (const internal::GoogleTestFailureException&) {  // NOLINT
+      // This exception type can only be thrown by a failed Google
+      // Test assertion with the intention of letting another testing
+      // framework catch it.  Therefore we just re-throw it.
+      throw;
+    } catch (const std::exception& e) {  // NOLINT
+      internal::ReportFailureInUnknownLocation(
+          TestPartResult::kFatalFailure,
+          FormatCxxExceptionMessage(e.what(), location));
+    } catch (...) {  // NOLINT
+      internal::ReportFailureInUnknownLocation(
+          TestPartResult::kFatalFailure,
+          FormatCxxExceptionMessage(NULL, location));
+    }
+    return static_cast<Result>(0);
+#else
+    return HandleSehExceptionsInMethodIfSupported(object, method, location);
+#endif  // GTEST_HAS_EXCEPTIONS
+  } else {
+    return (object->*method)();
+  }
+}
+
+}  // namespace internal
+
+// Runs the test and updates the test result.
+void Test::Run() {
+  if (!HasSameFixtureClass()) return;
+
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(this, &Test::SetUp, "SetUp()");
+  // We will run the test only if SetUp() was successful and didn't call
+  // GTEST_SKIP.
+  if (!HasFatalFailure() && !IsSkipped()) {
+    impl->os_stack_trace_getter()->UponLeavingGTest();
+    internal::HandleExceptionsInMethodIfSupported(
+        this, &Test::TestBody, "the test body");
+  }
+
+  // However, we want to clean up as much as possible.  Hence we will
+  // always call TearDown(), even if SetUp() or the test body has
+  // failed.
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      this, &Test::TearDown, "TearDown()");
+}
+
+// Returns true iff the current test has a fatal failure.
+bool Test::HasFatalFailure() {
+  return internal::GetUnitTestImpl()->current_test_result()->HasFatalFailure();
+}
+
+// Returns true iff the current test has a non-fatal failure.
+bool Test::HasNonfatalFailure() {
+  return internal::GetUnitTestImpl()->current_test_result()->
+      HasNonfatalFailure();
+}
+
+// Returns true if and only if the current test was skipped.
+bool Test::IsSkipped() {
+  return internal::GetUnitTestImpl()->current_test_result()->Skipped();
+}
+
+// class TestInfo
+
+// Constructs a TestInfo object. It assumes ownership of the test factory
+// object.
+TestInfo::TestInfo(const std::string& a_test_case_name,
+                   const std::string& a_name,
+                   const char* a_type_param,
+                   const char* a_value_param,
+                   internal::CodeLocation a_code_location,
+                   internal::TypeId fixture_class_id,
+                   internal::TestFactoryBase* factory)
+    : test_case_name_(a_test_case_name),
+      name_(a_name),
+      type_param_(a_type_param ? new std::string(a_type_param) : NULL),
+      value_param_(a_value_param ? new std::string(a_value_param) : NULL),
+      location_(a_code_location),
+      fixture_class_id_(fixture_class_id),
+      should_run_(false),
+      is_disabled_(false),
+      matches_filter_(false),
+      factory_(factory),
+      result_() {}
+
+// Destructs a TestInfo object.
+TestInfo::~TestInfo() { delete factory_; }
+
+namespace internal {
+
+// Creates a new TestInfo object and registers it with Google Test;
+// returns the created object.
+//
+// Arguments:
+//
+//   test_case_name:   name of the test case
+//   name:             name of the test
+//   type_param:       the name of the test's type parameter, or NULL if
+//                     this is not a typed or a type-parameterized test.
+//   value_param:      text representation of the test's value parameter,
+//                     or NULL if this is not a value-parameterized test.
+//   code_location:    code location where the test is defined
+//   fixture_class_id: ID of the test fixture class
+//   set_up_tc:        pointer to the function that sets up the test case
+//   tear_down_tc:     pointer to the function that tears down the test case
+//   factory:          pointer to the factory that creates a test object.
+//                     The newly created TestInfo instance will assume
+//                     ownership of the factory object.
+TestInfo* MakeAndRegisterTestInfo(
+    const char* test_case_name,
+    const char* name,
+    const char* type_param,
+    const char* value_param,
+    CodeLocation code_location,
+    TypeId fixture_class_id,
+    SetUpTestCaseFunc set_up_tc,
+    TearDownTestCaseFunc tear_down_tc,
+    TestFactoryBase* factory) {
+  TestInfo* const test_info =
+      new TestInfo(test_case_name, name, type_param, value_param,
+                   code_location, fixture_class_id, factory);
+  GetUnitTestImpl()->AddTestInfo(set_up_tc, tear_down_tc, test_info);
+  return test_info;
+}
+
+#if GTEST_HAS_PARAM_TEST
+void ReportInvalidTestCaseType(const char* test_case_name,
+                               CodeLocation code_location) {
+  Message errors;
+  errors
+      << "Attempted redefinition of test case " << test_case_name << ".\n"
+      << "All tests in the same test case must use the same test fixture\n"
+      << "class.  However, in test case " << test_case_name << ", you tried\n"
+      << "to define a test using a fixture class different from the one\n"
+      << "used earlier. This can happen if the two fixture classes are\n"
+      << "from different namespaces and have the same name. You should\n"
+      << "probably rename one of the classes to put the tests into different\n"
+      << "test cases.";
+
+  fprintf(stderr, "%s %s",
+          FormatFileLocation(code_location.file.c_str(),
+                             code_location.line).c_str(),
+          errors.GetString().c_str());
+}
+#endif  // GTEST_HAS_PARAM_TEST
+
+}  // namespace internal
+
+namespace {
+
+// A predicate that checks the test name of a TestInfo against a known
+// value.
+//
+// This is used for implementation of the TestCase class only.  We put
+// it in the anonymous namespace to prevent polluting the outer
+// namespace.
+//
+// TestNameIs is copyable.
+class TestNameIs {
+ public:
+  // Constructor.
+  //
+  // TestNameIs has NO default constructor.
+  explicit TestNameIs(const char* name)
+      : name_(name) {}
+
+  // Returns true iff the test name of test_info matches name_.
+  bool operator()(const TestInfo * test_info) const {
+    return test_info && test_info->name() == name_;
+  }
+
+ private:
+  std::string name_;
+};
+
+}  // namespace
+
+namespace internal {
+
+// This method expands all parameterized tests registered with macros TEST_P
+// and INSTANTIATE_TEST_CASE_P into regular tests and registers those.
+// This will be done just once during the program runtime.
+void UnitTestImpl::RegisterParameterizedTests() {
+#if GTEST_HAS_PARAM_TEST
+  if (!parameterized_tests_registered_) {
+    parameterized_test_registry_.RegisterTests();
+    parameterized_tests_registered_ = true;
+  }
+#endif
+}
+
+}  // namespace internal
+
+// Creates the test object, runs it, records its result, and then
+// deletes it.
+void TestInfo::Run() {
+  if (!should_run_) return;
+
+  // Tells UnitTest where to store test result.
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  impl->set_current_test_info(this);
+
+  TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater();
+
+  // Notifies the unit test event listeners that a test is about to start.
+  repeater->OnTestStart(*this);
+
+  const TimeInMillis start = internal::GetTimeInMillis();
+
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+
+  // Creates the test object.
+  Test* const test = internal::HandleExceptionsInMethodIfSupported(
+      factory_, &internal::TestFactoryBase::CreateTest,
+      "the test fixture's constructor");
+
+  // Runs the test only if the test object was created and its
+  // constructor didn't generate a fatal failure or invoke GTEST_SKIP().
+  if ((test != NULL) && !Test::HasFatalFailure() && !Test::IsSkipped()) {
+    // This doesn't throw as all user code that can throw are wrapped into
+    // exception handling code.
+    test->Run();
+  }
+
+  // Deletes the test object.
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      test, &Test::DeleteSelf_, "the test fixture's destructor");
+
+  result_.set_elapsed_time(internal::GetTimeInMillis() - start);
+
+  // Notifies the unit test event listener that a test has just finished.
+  repeater->OnTestEnd(*this);
+
+  // Tells UnitTest to stop associating assertion results to this
+  // test.
+  impl->set_current_test_info(NULL);
+}
+
+// class TestCase
+
+// Gets the number of successful tests in this test case.
+int TestCase::successful_test_count() const {
+  return CountIf(test_info_list_, TestPassed);
+}
+
+// Gets the number of successful tests in this test suite.
+int TestCase::skipped_test_count() const {
+  return CountIf(test_info_list_, TestSkipped);
+}
+
+// Gets the number of failed tests in this test case.
+int TestCase::failed_test_count() const {
+  return CountIf(test_info_list_, TestFailed);
+}
+
+// Gets the number of disabled tests that will be reported in the XML report.
+int TestCase::reportable_disabled_test_count() const {
+  return CountIf(test_info_list_, TestReportableDisabled);
+}
+
+// Gets the number of disabled tests in this test case.
+int TestCase::disabled_test_count() const {
+  return CountIf(test_info_list_, TestDisabled);
+}
+
+// Gets the number of tests to be printed in the XML report.
+int TestCase::reportable_test_count() const {
+  return CountIf(test_info_list_, TestReportable);
+}
+
+// Get the number of tests in this test case that should run.
+int TestCase::test_to_run_count() const {
+  return CountIf(test_info_list_, ShouldRunTest);
+}
+
+// Gets the number of all tests.
+int TestCase::total_test_count() const {
+  return static_cast<int>(test_info_list_.size());
+}
+
+// Creates a TestCase with the given name.
+//
+// Arguments:
+//
+//   name:         name of the test case
+//   a_type_param: the name of the test case's type parameter, or NULL if
+//                 this is not a typed or a type-parameterized test case.
+//   set_up_tc:    pointer to the function that sets up the test case
+//   tear_down_tc: pointer to the function that tears down the test case
+TestCase::TestCase(const char* a_name, const char* a_type_param,
+                   Test::SetUpTestCaseFunc set_up_tc,
+                   Test::TearDownTestCaseFunc tear_down_tc)
+    : name_(a_name),
+      type_param_(a_type_param ? new std::string(a_type_param) : NULL),
+      set_up_tc_(set_up_tc),
+      tear_down_tc_(tear_down_tc),
+      should_run_(false),
+      elapsed_time_(0) {
+}
+
+// Destructor of TestCase.
+TestCase::~TestCase() {
+  // Deletes every Test in the collection.
+  ForEach(test_info_list_, internal::Delete<TestInfo>);
+}
+
+// Returns the i-th test among all the tests. i can range from 0 to
+// total_test_count() - 1. If i is not in that range, returns NULL.
+const TestInfo* TestCase::GetTestInfo(int i) const {
+  const int index = GetElementOr(test_indices_, i, -1);
+  return index < 0 ? NULL : test_info_list_[index];
+}
+
+// Returns the i-th test among all the tests. i can range from 0 to
+// total_test_count() - 1. If i is not in that range, returns NULL.
+TestInfo* TestCase::GetMutableTestInfo(int i) {
+  const int index = GetElementOr(test_indices_, i, -1);
+  return index < 0 ? NULL : test_info_list_[index];
+}
+
+// Adds a test to this test case.  Will delete the test upon
+// destruction of the TestCase object.
+void TestCase::AddTestInfo(TestInfo * test_info) {
+  test_info_list_.push_back(test_info);
+  test_indices_.push_back(static_cast<int>(test_indices_.size()));
+}
+
+// Runs every test in this TestCase.
+void TestCase::Run() {
+  if (!should_run_) return;
+
+  internal::UnitTestImpl* const impl = internal::GetUnitTestImpl();
+  impl->set_current_test_case(this);
+
+  TestEventListener* repeater = UnitTest::GetInstance()->listeners().repeater();
+
+  repeater->OnTestCaseStart(*this);
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      this, &TestCase::RunSetUpTestCase, "SetUpTestCase()");
+
+  const internal::TimeInMillis start = internal::GetTimeInMillis();
+  for (int i = 0; i < total_test_count(); i++) {
+    GetMutableTestInfo(i)->Run();
+  }
+  elapsed_time_ = internal::GetTimeInMillis() - start;
+
+  impl->os_stack_trace_getter()->UponLeavingGTest();
+  internal::HandleExceptionsInMethodIfSupported(
+      this, &TestCase::RunTearDownTestCase, "TearDownTestCase()");
+
+  repeater->OnTestCaseEnd(*this);
+  impl->set_current_test_case(NULL);
+}
+
+// Clears the results of all tests in this test case.
+void TestCase::ClearResult() {
+  ad_hoc_test_result_.Clear();
+  ForEach(test_info_list_, TestInfo::ClearTestResult);
+}
+
+// Shuffles the tests in this test case.
+void TestCase::ShuffleTests(internal::Random* random) {
+  Shuffle(random, &test_indices_);
+}
+
+// Restores the test order to before the first shuffle.
+void TestCase::UnshuffleTests() {
+  for (size_t i = 0; i < test_indices_.size(); i++) {
+    test_indices_[i] = static_cast<int>(i);
+  }
+}
+
+// Formats a countable noun.  Depending on its quantity, either the
+// singular form or the plural form is used. e.g.
+//
+// FormatCountableNoun(1, "formula", "formuli") returns "1 formula".
+// FormatCountableNoun(5, "book", "books") returns "5 books".
+static std::string FormatCountableNoun(int count,
+                                       const char * singular_form,
+                                       const char * plural_form) {
+  return internal::StreamableToString(count) + " " +
+      (count == 1 ? singular_form : plural_form);
+}
+
+// Formats the count of tests.
+static std::string FormatTestCount(int test_count) {
+  return FormatCountableNoun(test_count, "test", "tests");
+}
+
+// Formats the count of test cases.
+static std::string FormatTestCaseCount(int test_case_count) {
+  return FormatCountableNoun(test_case_count, "test case", "test cases");
+}
+
+// Converts a TestPartResult::Type enum to human-friendly string
+// representation.  Both kNonFatalFailure and kFatalFailure are translated
+// to "Failure", as the user usually doesn't care about the difference
+// between the two when viewing the test result.
+static const char * TestPartResultTypeToString(TestPartResult::Type type) {
+  switch (type) {
+    case TestPartResult::kSkip:
+      return "Skipped";
+    case TestPartResult::kSuccess:
+      return "Success";
+
+    case TestPartResult::kNonFatalFailure:
+    case TestPartResult::kFatalFailure:
+#ifdef _MSC_VER
+      return "error: ";
+#else
+      return "Failure\n";
+#endif
+    default:
+      return "Unknown result type";
+  }
+}
+
+namespace internal {
+
+// Prints a TestPartResult to an std::string.
+static std::string PrintTestPartResultToString(
+    const TestPartResult& test_part_result) {
+  return (Message()
+          << internal::FormatFileLocation(test_part_result.file_name(),
+                                          test_part_result.line_number())
+          << " " << TestPartResultTypeToString(test_part_result.type())
+          << test_part_result.message()).GetString();
+}
+
+// Prints a TestPartResult.
+static void PrintTestPartResult(const TestPartResult& test_part_result) {
+  const std::string& result =
+      PrintTestPartResultToString(test_part_result);
+  printf("%s\n", result.c_str());
+  fflush(stdout);
+  // If the test program runs in Visual Studio or a debugger, the
+  // following statements add the test part result message to the Output
+  // window such that the user can double-click on it to jump to the
+  // corresponding source code location; otherwise they do nothing.
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+  // We don't call OutputDebugString*() on Windows Mobile, as printing
+  // to stdout is done by OutputDebugString() there already - we don't
+  // want the same message printed twice.
+  ::OutputDebugStringA(result.c_str());
+  ::OutputDebugStringA("\n");
+#endif
+}
+
+// class PrettyUnitTestResultPrinter
+
+enum GTestColor {
+  COLOR_DEFAULT,
+  COLOR_RED,
+  COLOR_GREEN,
+  COLOR_YELLOW
+};
+
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \
+    !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+
+// Returns the character attribute for the given color.
+WORD GetColorAttribute(GTestColor color) {
+  switch (color) {
+    case COLOR_RED:    return FOREGROUND_RED;
+    case COLOR_GREEN:  return FOREGROUND_GREEN;
+    case COLOR_YELLOW: return FOREGROUND_RED | FOREGROUND_GREEN;
+    default:           return 0;
+  }
+}
+
+#else
+
+// Returns the ANSI color code for the given color.  COLOR_DEFAULT is
+// an invalid input.
+const char* GetAnsiColorCode(GTestColor color) {
+  switch (color) {
+    case COLOR_RED:     return "1";
+    case COLOR_GREEN:   return "2";
+    case COLOR_YELLOW:  return "3";
+    default:            return NULL;
+  };
+}
+
+#endif  // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+
+// Returns true iff Google Test should use colors in the output.
+bool ShouldUseColor(bool stdout_is_tty) {
+  const char* const gtest_color = GTEST_FLAG(color).c_str();
+
+  if (String::CaseInsensitiveCStringEquals(gtest_color, "auto")) {
+#if GTEST_OS_WINDOWS
+    // On Windows the TERM variable is usually not set, but the
+    // console there does support colors.
+    return stdout_is_tty;
+#else
+    // On non-Windows platforms, we rely on the TERM variable.
+    const char* const term = posix::GetEnv("TERM");
+    const bool term_supports_color =
+        String::CStringEquals(term, "xterm") ||
+        String::CStringEquals(term, "xterm-color") ||
+        String::CStringEquals(term, "xterm-256color") ||
+        String::CStringEquals(term, "screen") ||
+        String::CStringEquals(term, "screen-256color") ||
+        String::CStringEquals(term, "tmux") ||
+        String::CStringEquals(term, "tmux-256color") ||
+        String::CStringEquals(term, "rxvt-unicode") ||
+        String::CStringEquals(term, "rxvt-unicode-256color") ||
+        String::CStringEquals(term, "linux") ||
+        String::CStringEquals(term, "cygwin");
+    return stdout_is_tty && term_supports_color;
+#endif  // GTEST_OS_WINDOWS
+  }
+
+  return String::CaseInsensitiveCStringEquals(gtest_color, "yes") ||
+      String::CaseInsensitiveCStringEquals(gtest_color, "true") ||
+      String::CaseInsensitiveCStringEquals(gtest_color, "t") ||
+      String::CStringEquals(gtest_color, "1");
+  // We take "yes", "true", "t", and "1" as meaning "yes".  If the
+  // value is neither one of these nor "auto", we treat it as "no" to
+  // be conservative.
+}
+
+// Helpers for printing colored strings to stdout. Note that on Windows, we
+// cannot simply emit special characters and have the terminal change colors.
+// This routine must actually emit the characters rather than return a string
+// that would be colored when printed, as can be done on Linux.
+GTEST_ATTRIBUTE_PRINTF_(2, 3)
+void ColoredPrintf(GTestColor color, const char* fmt, ...) {
+  va_list args;
+  va_start(args, fmt);
+
+#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS || \
+    GTEST_OS_IOS || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT
+  const bool use_color = AlwaysFalse();
+#else
+  static const bool in_color_mode =
+      ShouldUseColor(posix::IsATTY(posix::FileNo(stdout)) != 0);
+  const bool use_color = in_color_mode && (color != COLOR_DEFAULT);
+#endif  // GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || GTEST_OS_ZOS
+  // The '!= 0' comparison is necessary to satisfy MSVC 7.1.
+
+  if (!use_color) {
+    vprintf(fmt, args);
+    va_end(args);
+    return;
+  }
+
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE && \
+    !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+  const HANDLE stdout_handle = GetStdHandle(STD_OUTPUT_HANDLE);
+
+  // Gets the current text color.
+  CONSOLE_SCREEN_BUFFER_INFO buffer_info;
+  GetConsoleScreenBufferInfo(stdout_handle, &buffer_info);
+  const WORD old_color_attrs = buffer_info.wAttributes;
+
+  // We need to flush the stream buffers into the console before each
+  // SetConsoleTextAttribute call lest it affect the text that is already
+  // printed but has not yet reached the console.
+  fflush(stdout);
+  SetConsoleTextAttribute(stdout_handle,
+                          GetColorAttribute(color) | FOREGROUND_INTENSITY);
+  vprintf(fmt, args);
+
+  fflush(stdout);
+  // Restores the text color.
+  SetConsoleTextAttribute(stdout_handle, old_color_attrs);
+#else
+  printf("\033[0;3%sm", GetAnsiColorCode(color));
+  vprintf(fmt, args);
+  printf("\033[m");  // Resets the terminal to default.
+#endif  // GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_MOBILE
+  va_end(args);
+}
+
+// Text printed in Google Test's text output and --gunit_list_tests
+// output to label the type parameter and value parameter for a test.
+static const char kTypeParamLabel[] = "TypeParam";
+static const char kValueParamLabel[] = "GetParam()";
+
+void PrintFullTestCommentIfPresent(const TestInfo& test_info) {
+  const char* const type_param = test_info.type_param();
+  const char* const value_param = test_info.value_param();
+
+  if (type_param != NULL || value_param != NULL) {
+    printf(", where ");
+    if (type_param != NULL) {
+      printf("%s = %s", kTypeParamLabel, type_param);
+      if (value_param != NULL)
+        printf(" and ");
+    }
+    if (value_param != NULL) {
+      printf("%s = %s", kValueParamLabel, value_param);
+    }
+  }
+}
+
+// This class implements the TestEventListener interface.
+//
+// Class PrettyUnitTestResultPrinter is copyable.
+class PrettyUnitTestResultPrinter : public TestEventListener {
+ public:
+  PrettyUnitTestResultPrinter() {}
+  static void PrintTestName(const char * test_case, const char * test) {
+    printf("%s.%s", test_case, test);
+  }
+
+  // The following methods override what's in the TestEventListener class.
+  virtual void OnTestProgramStart(const UnitTest& /*unit_test*/) {}
+  virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration);
+  virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsSetUpEnd(const UnitTest& /*unit_test*/) {}
+  virtual void OnTestCaseStart(const TestCase& test_case);
+  virtual void OnTestStart(const TestInfo& test_info);
+  virtual void OnTestPartResult(const TestPartResult& result);
+  virtual void OnTestEnd(const TestInfo& test_info);
+  virtual void OnTestCaseEnd(const TestCase& test_case);
+  virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsTearDownEnd(const UnitTest& /*unit_test*/) {}
+  virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+  virtual void OnTestProgramEnd(const UnitTest& /*unit_test*/) {}
+
+ private:
+  static void PrintFailedTests(const UnitTest& unit_test);
+  static void PrintSkippedTests(const UnitTest& unit_test);
+};
+
+  // Fired before each iteration of tests starts.
+void PrettyUnitTestResultPrinter::OnTestIterationStart(
+    const UnitTest& unit_test, int iteration) {
+  if (GTEST_FLAG(repeat) != 1)
+    printf("\nRepeating all tests (iteration %d) . . .\n\n", iteration + 1);
+
+  const char* const filter = GTEST_FLAG(filter).c_str();
+
+  // Prints the filter if it's not *.  This reminds the user that some
+  // tests may be skipped.
+  if (!String::CStringEquals(filter, kUniversalFilter)) {
+    ColoredPrintf(COLOR_YELLOW,
+                  "Note: %s filter = %s\n", GTEST_NAME_, filter);
+  }
+
+  if (internal::ShouldShard(kTestTotalShards, kTestShardIndex, false)) {
+    const Int32 shard_index = Int32FromEnvOrDie(kTestShardIndex, -1);
+    ColoredPrintf(COLOR_YELLOW,
+                  "Note: This is test shard %d of %s.\n",
+                  static_cast<int>(shard_index) + 1,
+                  internal::posix::GetEnv(kTestTotalShards));
+  }
+
+  if (GTEST_FLAG(shuffle)) {
+    ColoredPrintf(COLOR_YELLOW,
+                  "Note: Randomizing tests' orders with a seed of %d .\n",
+                  unit_test.random_seed());
+  }
+
+  ColoredPrintf(COLOR_GREEN,  "[==========] ");
+  printf("Running %s from %s.\n",
+         FormatTestCount(unit_test.test_to_run_count()).c_str(),
+         FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str());
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnEnvironmentsSetUpStart(
+    const UnitTest& /*unit_test*/) {
+  ColoredPrintf(COLOR_GREEN,  "[----------] ");
+  printf("Global test environment set-up.\n");
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestCaseStart(const TestCase& test_case) {
+  const std::string counts =
+      FormatCountableNoun(test_case.test_to_run_count(), "test", "tests");
+  ColoredPrintf(COLOR_GREEN, "[----------] ");
+  printf("%s from %s", counts.c_str(), test_case.name());
+  if (test_case.type_param() == NULL) {
+    printf("\n");
+  } else {
+    printf(", where %s = %s\n", kTypeParamLabel, test_case.type_param());
+  }
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestStart(const TestInfo& test_info) {
+  ColoredPrintf(COLOR_GREEN,  "[ RUN      ] ");
+  PrintTestName(test_info.test_case_name(), test_info.name());
+  printf("\n");
+  fflush(stdout);
+}
+
+// Called after an assertion failure.
+void PrettyUnitTestResultPrinter::OnTestPartResult(
+    const TestPartResult& result) {
+  switch(result.type()) {
+    // If the test part succeeded,
+    // we don't need to do anything.
+    case TestPartResult::kSuccess:
+      return;
+    default:
+      // Print failure message from the assertion
+      // (e.g. expected this and got that).
+      PrintTestPartResult(result);
+      fflush(stdout);
+  }
+}
+
+void PrettyUnitTestResultPrinter::OnTestEnd(const TestInfo& test_info) {
+  if (test_info.result()->Passed()) {
+    ColoredPrintf(COLOR_GREEN, "[       OK ] ");
+  } else if (test_info.result()->Skipped()) {
+    ColoredPrintf(COLOR_GREEN, "[  SKIPPED ] ");
+  } else {
+    ColoredPrintf(COLOR_RED, "[  FAILED  ] ");
+  }
+  PrintTestName(test_info.test_case_name(), test_info.name());
+  if (test_info.result()->Failed())
+    PrintFullTestCommentIfPresent(test_info);
+
+  if (GTEST_FLAG(print_time)) {
+    printf(" (%s ms)\n", internal::StreamableToString(
+           test_info.result()->elapsed_time()).c_str());
+  } else {
+    printf("\n");
+  }
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnTestCaseEnd(const TestCase& test_case) {
+  if (!GTEST_FLAG(print_time)) return;
+
+  const std::string counts =
+      FormatCountableNoun(test_case.test_to_run_count(), "test", "tests");
+  ColoredPrintf(COLOR_GREEN, "[----------] ");
+  printf("%s from %s (%s ms total)\n\n",
+         counts.c_str(), test_case.name(),
+         internal::StreamableToString(test_case.elapsed_time()).c_str());
+  fflush(stdout);
+}
+
+void PrettyUnitTestResultPrinter::OnEnvironmentsTearDownStart(
+    const UnitTest& /*unit_test*/) {
+  ColoredPrintf(COLOR_GREEN,  "[----------] ");
+  printf("Global test environment tear-down\n");
+  fflush(stdout);
+}
+
+// Internal helper for printing the list of failed tests.
+void PrettyUnitTestResultPrinter::PrintFailedTests(const UnitTest& unit_test) {
+  const int failed_test_count = unit_test.failed_test_count();
+  if (failed_test_count == 0) {
+    return;
+  }
+
+  for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+    const TestCase& test_case = *unit_test.GetTestCase(i);
+    if (!test_case.should_run() || (test_case.failed_test_count() == 0)) {
+      continue;
+    }
+    for (int j = 0; j < test_case.total_test_count(); ++j) {
+      const TestInfo& test_info = *test_case.GetTestInfo(j);
+      if (!test_info.should_run() || !test_info.result()->Failed()) {
+        continue;
+      }
+      ColoredPrintf(COLOR_RED, "[  FAILED  ] ");
+      printf("%s.%s", test_case.name(), test_info.name());
+      PrintFullTestCommentIfPresent(test_info);
+      printf("\n");
+    }
+  }
+}
+
+// Internal helper for printing the list of skipped tests.
+void PrettyUnitTestResultPrinter::PrintSkippedTests(const UnitTest& unit_test) {
+  const int skipped_test_count = unit_test.skipped_test_count();
+  if (skipped_test_count == 0) {
+    return;
+  }
+
+  for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+    const TestCase& test_suite = *unit_test.GetTestCase(i);
+    if (!test_suite.should_run() || (test_suite.skipped_test_count() == 0)) {
+      continue;
+    }
+    for (int j = 0; j < test_suite.total_test_count(); ++j) {
+      const TestInfo& test_info = *test_suite.GetTestInfo(j);
+      if (!test_info.should_run() || !test_info.result()->Skipped()) {
+        continue;
+      }
+      ColoredPrintf(COLOR_GREEN, "[  SKIPPED ] ");
+      printf("%s.%s", test_suite.name(), test_info.name());
+      printf("\n");
+    }
+  }
+}
+
+void PrettyUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test,
+                                                     int /*iteration*/) {
+  ColoredPrintf(COLOR_GREEN,  "[==========] ");
+  printf("%s from %s ran.",
+         FormatTestCount(unit_test.test_to_run_count()).c_str(),
+         FormatTestCaseCount(unit_test.test_case_to_run_count()).c_str());
+  if (GTEST_FLAG(print_time)) {
+    printf(" (%s ms total)",
+           internal::StreamableToString(unit_test.elapsed_time()).c_str());
+  }
+  printf("\n");
+  ColoredPrintf(COLOR_GREEN,  "[  PASSED  ] ");
+  printf("%s.\n", FormatTestCount(unit_test.successful_test_count()).c_str());
+
+  const int skipped_test_count = unit_test.skipped_test_count();
+  if (skipped_test_count > 0) {
+    ColoredPrintf(COLOR_GREEN, "[  SKIPPED ] ");
+    printf("%s, listed below:\n", FormatTestCount(skipped_test_count).c_str());
+    PrintSkippedTests(unit_test);
+  }
+
+  int num_failures = unit_test.failed_test_count();
+  if (!unit_test.Passed()) {
+    const int failed_test_count = unit_test.failed_test_count();
+    ColoredPrintf(COLOR_RED,  "[  FAILED  ] ");
+    printf("%s, listed below:\n", FormatTestCount(failed_test_count).c_str());
+    PrintFailedTests(unit_test);
+    printf("\n%2d FAILED %s\n", num_failures,
+                        num_failures == 1 ? "TEST" : "TESTS");
+  }
+
+  int num_disabled = unit_test.reportable_disabled_test_count();
+  if (num_disabled && !GTEST_FLAG(also_run_disabled_tests)) {
+    if (!num_failures) {
+      printf("\n");  // Add a spacer if no FAILURE banner is displayed.
+    }
+    ColoredPrintf(COLOR_YELLOW,
+                  "  YOU HAVE %d DISABLED %s\n\n",
+                  num_disabled,
+                  num_disabled == 1 ? "TEST" : "TESTS");
+  }
+  // Ensure that Google Test output is printed before, e.g., heapchecker output.
+  fflush(stdout);
+}
+
+// End PrettyUnitTestResultPrinter
+
+// class TestEventRepeater
+//
+// This class forwards events to other event listeners.
+class TestEventRepeater : public TestEventListener {
+ public:
+  TestEventRepeater() : forwarding_enabled_(true) {}
+  virtual ~TestEventRepeater();
+  void Append(TestEventListener *listener);
+  TestEventListener* Release(TestEventListener* listener);
+
+  // Controls whether events will be forwarded to listeners_. Set to false
+  // in death test child processes.
+  bool forwarding_enabled() const { return forwarding_enabled_; }
+  void set_forwarding_enabled(bool enable) { forwarding_enabled_ = enable; }
+
+  virtual void OnTestProgramStart(const UnitTest& unit_test);
+  virtual void OnTestIterationStart(const UnitTest& unit_test, int iteration);
+  virtual void OnEnvironmentsSetUpStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsSetUpEnd(const UnitTest& unit_test);
+  virtual void OnTestCaseStart(const TestCase& test_case);
+  virtual void OnTestStart(const TestInfo& test_info);
+  virtual void OnTestPartResult(const TestPartResult& result);
+  virtual void OnTestEnd(const TestInfo& test_info);
+  virtual void OnTestCaseEnd(const TestCase& test_case);
+  virtual void OnEnvironmentsTearDownStart(const UnitTest& unit_test);
+  virtual void OnEnvironmentsTearDownEnd(const UnitTest& unit_test);
+  virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+  virtual void OnTestProgramEnd(const UnitTest& unit_test);
+
+ private:
+  // Controls whether events will be forwarded to listeners_. Set to false
+  // in death test child processes.
+  bool forwarding_enabled_;
+  // The list of listeners that receive events.
+  std::vector<TestEventListener*> listeners_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventRepeater);
+};
+
+TestEventRepeater::~TestEventRepeater() {
+  ForEach(listeners_, Delete<TestEventListener>);
+}
+
+void TestEventRepeater::Append(TestEventListener *listener) {
+  listeners_.push_back(listener);
+}
+
+// TODO(vladl@google.com): Factor the search functionality into Vector::Find.
+TestEventListener* TestEventRepeater::Release(TestEventListener *listener) {
+  for (size_t i = 0; i < listeners_.size(); ++i) {
+    if (listeners_[i] == listener) {
+      listeners_.erase(listeners_.begin() + i);
+      return listener;
+    }
+  }
+
+  return NULL;
+}
+
+// Since most methods are very similar, use macros to reduce boilerplate.
+// This defines a member that forwards the call to all listeners.
+#define GTEST_REPEATER_METHOD_(Name, Type) \
+void TestEventRepeater::Name(const Type& parameter) { \
+  if (forwarding_enabled_) { \
+    for (size_t i = 0; i < listeners_.size(); i++) { \
+      listeners_[i]->Name(parameter); \
+    } \
+  } \
+}
+// This defines a member that forwards the call to all listeners in reverse
+// order.
+#define GTEST_REVERSE_REPEATER_METHOD_(Name, Type) \
+void TestEventRepeater::Name(const Type& parameter) { \
+  if (forwarding_enabled_) { \
+    for (int i = static_cast<int>(listeners_.size()) - 1; i >= 0; i--) { \
+      listeners_[i]->Name(parameter); \
+    } \
+  } \
+}
+
+GTEST_REPEATER_METHOD_(OnTestProgramStart, UnitTest)
+GTEST_REPEATER_METHOD_(OnEnvironmentsSetUpStart, UnitTest)
+GTEST_REPEATER_METHOD_(OnTestCaseStart, TestCase)
+GTEST_REPEATER_METHOD_(OnTestStart, TestInfo)
+GTEST_REPEATER_METHOD_(OnTestPartResult, TestPartResult)
+GTEST_REPEATER_METHOD_(OnEnvironmentsTearDownStart, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsSetUpEnd, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnEnvironmentsTearDownEnd, UnitTest)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestEnd, TestInfo)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestCaseEnd, TestCase)
+GTEST_REVERSE_REPEATER_METHOD_(OnTestProgramEnd, UnitTest)
+
+#undef GTEST_REPEATER_METHOD_
+#undef GTEST_REVERSE_REPEATER_METHOD_
+
+void TestEventRepeater::OnTestIterationStart(const UnitTest& unit_test,
+                                             int iteration) {
+  if (forwarding_enabled_) {
+    for (size_t i = 0; i < listeners_.size(); i++) {
+      listeners_[i]->OnTestIterationStart(unit_test, iteration);
+    }
+  }
+}
+
+void TestEventRepeater::OnTestIterationEnd(const UnitTest& unit_test,
+                                           int iteration) {
+  if (forwarding_enabled_) {
+    for (int i = static_cast<int>(listeners_.size()) - 1; i >= 0; i--) {
+      listeners_[i]->OnTestIterationEnd(unit_test, iteration);
+    }
+  }
+}
+
+// End TestEventRepeater
+
+// This class generates an XML output file.
+class XmlUnitTestResultPrinter : public EmptyTestEventListener {
+ public:
+  explicit XmlUnitTestResultPrinter(const char* output_file);
+
+  virtual void OnTestIterationEnd(const UnitTest& unit_test, int iteration);
+
+ private:
+  // Is c a whitespace character that is normalized to a space character
+  // when it appears in an XML attribute value?
+  static bool IsNormalizableWhitespace(char c) {
+    return c == 0x9 || c == 0xA || c == 0xD;
+  }
+
+  // May c appear in a well-formed XML document?
+  static bool IsValidXmlCharacter(char c) {
+    return IsNormalizableWhitespace(c) || c >= 0x20;
+  }
+
+  // Returns an XML-escaped copy of the input string str.  If
+  // is_attribute is true, the text is meant to appear as an attribute
+  // value, and normalizable whitespace is preserved by replacing it
+  // with character references.
+  static std::string EscapeXml(const std::string& str, bool is_attribute);
+
+  // Returns the given string with all characters invalid in XML removed.
+  static std::string RemoveInvalidXmlCharacters(const std::string& str);
+
+  // Convenience wrapper around EscapeXml when str is an attribute value.
+  static std::string EscapeXmlAttribute(const std::string& str) {
+    return EscapeXml(str, true);
+  }
+
+  // Convenience wrapper around EscapeXml when str is not an attribute value.
+  static std::string EscapeXmlText(const char* str) {
+    return EscapeXml(str, false);
+  }
+
+  // Verifies that the given attribute belongs to the given element and
+  // streams the attribute as XML.
+  static void OutputXmlAttribute(std::ostream* stream,
+                                 const std::string& element_name,
+                                 const std::string& name,
+                                 const std::string& value);
+
+  // Streams an XML CDATA section, escaping invalid CDATA sequences as needed.
+  static void OutputXmlCDataSection(::std::ostream* stream, const char* data);
+
+  // Streams an XML representation of a TestInfo object.
+  static void OutputXmlTestInfo(::std::ostream* stream,
+                                const char* test_case_name,
+                                const TestInfo& test_info);
+
+  // Prints an XML representation of a TestCase object
+  static void PrintXmlTestCase(::std::ostream* stream,
+                               const TestCase& test_case);
+
+  // Prints an XML summary of unit_test to output stream out.
+  static void PrintXmlUnitTest(::std::ostream* stream,
+                               const UnitTest& unit_test);
+
+  // Produces a string representing the test properties in a result as space
+  // delimited XML attributes based on the property key="value" pairs.
+  // When the std::string is not empty, it includes a space at the beginning,
+  // to delimit this attribute from prior attributes.
+  static std::string TestPropertiesAsXmlAttributes(const TestResult& result);
+
+  // The output file.
+  const std::string output_file_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(XmlUnitTestResultPrinter);
+};
+
+// Creates a new XmlUnitTestResultPrinter.
+XmlUnitTestResultPrinter::XmlUnitTestResultPrinter(const char* output_file)
+    : output_file_(output_file) {
+  if (output_file_.c_str() == NULL || output_file_.empty()) {
+    fprintf(stderr, "XML output file may not be null\n");
+    fflush(stderr);
+    exit(EXIT_FAILURE);
+  }
+}
+
+// Called after the unit test ends.
+void XmlUnitTestResultPrinter::OnTestIterationEnd(const UnitTest& unit_test,
+                                                  int /*iteration*/) {
+  FILE* xmlout = NULL;
+  FilePath output_file(output_file_);
+  FilePath output_dir(output_file.RemoveFileName());
+
+  if (output_dir.CreateDirectoriesRecursively()) {
+    xmlout = posix::FOpen(output_file_.c_str(), "w");
+  }
+  if (xmlout == NULL) {
+    // TODO(wan): report the reason of the failure.
+    //
+    // We don't do it for now as:
+    //
+    //   1. There is no urgent need for it.
+    //   2. It's a bit involved to make the errno variable thread-safe on
+    //      all three operating systems (Linux, Windows, and Mac OS).
+    //   3. To interpret the meaning of errno in a thread-safe way,
+    //      we need the strerror_r() function, which is not available on
+    //      Windows.
+    fprintf(stderr,
+            "Unable to open file \"%s\"\n",
+            output_file_.c_str());
+    fflush(stderr);
+    exit(EXIT_FAILURE);
+  }
+  std::stringstream stream;
+  PrintXmlUnitTest(&stream, unit_test);
+  fprintf(xmlout, "%s", StringStreamToString(&stream).c_str());
+  fclose(xmlout);
+}
+
+// Returns an XML-escaped copy of the input string str.  If is_attribute
+// is true, the text is meant to appear as an attribute value, and
+// normalizable whitespace is preserved by replacing it with character
+// references.
+//
+// Invalid XML characters in str, if any, are stripped from the output.
+// It is expected that most, if not all, of the text processed by this
+// module will consist of ordinary English text.
+// If this module is ever modified to produce version 1.1 XML output,
+// most invalid characters can be retained using character references.
+// TODO(wan): It might be nice to have a minimally invasive, human-readable
+// escaping scheme for invalid characters, rather than dropping them.
+std::string XmlUnitTestResultPrinter::EscapeXml(
+    const std::string& str, bool is_attribute) {
+  Message m;
+
+  for (size_t i = 0; i < str.size(); ++i) {
+    const char ch = str[i];
+    switch (ch) {
+      case '<':
+        m << "&lt;";
+        break;
+      case '>':
+        m << "&gt;";
+        break;
+      case '&':
+        m << "&amp;";
+        break;
+      case '\'':
+        if (is_attribute)
+          m << "&apos;";
+        else
+          m << '\'';
+        break;
+      case '"':
+        if (is_attribute)
+          m << "&quot;";
+        else
+          m << '"';
+        break;
+      default:
+        if (IsValidXmlCharacter(ch)) {
+          if (is_attribute && IsNormalizableWhitespace(ch))
+            m << "&#x" << String::FormatByte(static_cast<unsigned char>(ch))
+              << ";";
+          else
+            m << ch;
+        }
+        break;
+    }
+  }
+
+  return m.GetString();
+}
+
+// Returns the given string with all characters invalid in XML removed.
+// Currently invalid characters are dropped from the string. An
+// alternative is to replace them with certain characters such as . or ?.
+std::string XmlUnitTestResultPrinter::RemoveInvalidXmlCharacters(
+    const std::string& str) {
+  std::string output;
+  output.reserve(str.size());
+  for (std::string::const_iterator it = str.begin(); it != str.end(); ++it)
+    if (IsValidXmlCharacter(*it))
+      output.push_back(*it);
+
+  return output;
+}
+
+// The following routines generate an XML representation of a UnitTest
+// object.
+//
+// This is how Google Test concepts map to the DTD:
+//
+// <testsuites name="AllTests">        <-- corresponds to a UnitTest object
+//   <testsuite name="testcase-name">  <-- corresponds to a TestCase object
+//     <testcase name="test-name">     <-- corresponds to a TestInfo object
+//       <failure message="...">...</failure>
+//       <failure message="...">...</failure>
+//       <failure message="...">...</failure>
+//                                     <-- individual assertion failures
+//     </testcase>
+//   </testsuite>
+// </testsuites>
+
+// Formats the given time in milliseconds as seconds.
+std::string FormatTimeInMillisAsSeconds(TimeInMillis ms) {
+  ::std::stringstream ss;
+  ss << (static_cast<double>(ms) * 1e-3);
+  return ss.str();
+}
+
+static bool PortableLocaltime(time_t seconds, struct tm* out) {
+#if defined(_MSC_VER)
+  return localtime_s(out, &seconds) == 0;
+#elif defined(__MINGW32__) || defined(__MINGW64__)
+  // MINGW <time.h> provides neither localtime_r nor localtime_s, but uses
+  // Windows' localtime(), which has a thread-local tm buffer.
+  struct tm* tm_ptr = localtime(&seconds);  // NOLINT
+  if (tm_ptr == NULL)
+    return false;
+  *out = *tm_ptr;
+  return true;
+#else
+  return localtime_r(&seconds, out) != NULL;
+#endif
+}
+
+// Converts the given epoch time in milliseconds to a date string in the ISO
+// 8601 format, without the timezone information.
+std::string FormatEpochTimeInMillisAsIso8601(TimeInMillis ms) {
+  struct tm time_struct;
+  if (!PortableLocaltime(static_cast<time_t>(ms / 1000), &time_struct))
+    return "";
+  // YYYY-MM-DDThh:mm:ss
+  return StreamableToString(time_struct.tm_year + 1900) + "-" +
+      String::FormatIntWidth2(time_struct.tm_mon + 1) + "-" +
+      String::FormatIntWidth2(time_struct.tm_mday) + "T" +
+      String::FormatIntWidth2(time_struct.tm_hour) + ":" +
+      String::FormatIntWidth2(time_struct.tm_min) + ":" +
+      String::FormatIntWidth2(time_struct.tm_sec);
+}
+
+// Streams an XML CDATA section, escaping invalid CDATA sequences as needed.
+void XmlUnitTestResultPrinter::OutputXmlCDataSection(::std::ostream* stream,
+                                                     const char* data) {
+  const char* segment = data;
+  *stream << "<![CDATA[";
+  for (;;) {
+    const char* const next_segment = strstr(segment, "]]>");
+    if (next_segment != NULL) {
+      stream->write(
+          segment, static_cast<std::streamsize>(next_segment - segment));
+      *stream << "]]>]]&gt;<![CDATA[";
+      segment = next_segment + strlen("]]>");
+    } else {
+      *stream << segment;
+      break;
+    }
+  }
+  *stream << "]]>";
+}
+
+void XmlUnitTestResultPrinter::OutputXmlAttribute(
+    std::ostream* stream,
+    const std::string& element_name,
+    const std::string& name,
+    const std::string& value) {
+  const std::vector<std::string>& allowed_names =
+      GetReservedAttributesForElement(element_name);
+
+  GTEST_CHECK_(std::find(allowed_names.begin(), allowed_names.end(), name) !=
+                   allowed_names.end())
+      << "Attribute " << name << " is not allowed for element <" << element_name
+      << ">.";
+
+  *stream << " " << name << "=\"" << EscapeXmlAttribute(value) << "\"";
+}
+
+// Prints an XML representation of a TestInfo object.
+// TODO(wan): There is also value in printing properties with the plain printer.
+void XmlUnitTestResultPrinter::OutputXmlTestInfo(::std::ostream* stream,
+                                                 const char* test_case_name,
+                                                 const TestInfo& test_info) {
+  const TestResult& result = *test_info.result();
+  const std::string kTestcase = "testcase";
+
+  *stream << "    <testcase";
+  OutputXmlAttribute(stream, kTestcase, "name", test_info.name());
+
+  if (test_info.value_param() != NULL) {
+    OutputXmlAttribute(stream, kTestcase, "value_param",
+                       test_info.value_param());
+  }
+  if (test_info.type_param() != NULL) {
+    OutputXmlAttribute(stream, kTestcase, "type_param", test_info.type_param());
+  }
+
+  OutputXmlAttribute(stream, kTestcase, "status",
+                     test_info.should_run() ? "run" : "notrun");
+  OutputXmlAttribute(stream, kTestcase, "result",
+                     test_info.should_run()
+                         ? (result.Skipped() ? "skipped" : "completed")
+                         : "suppressed");
+  OutputXmlAttribute(stream, kTestcase, "time",
+                     FormatTimeInMillisAsSeconds(result.elapsed_time()));
+  OutputXmlAttribute(stream, kTestcase, "classname", test_case_name);
+  *stream << TestPropertiesAsXmlAttributes(result);
+
+  int failures = 0;
+  for (int i = 0; i < result.total_part_count(); ++i) {
+    const TestPartResult& part = result.GetTestPartResult(i);
+    if (part.failed()) {
+      if (++failures == 1) {
+        *stream << ">\n";
+      }
+      const std::string location =
+          internal::FormatCompilerIndependentFileLocation(part.file_name(),
+                                                          part.line_number());
+      const std::string summary = location + "\n" + part.summary();
+      *stream << "      <failure message=\""
+              << EscapeXmlAttribute(summary.c_str())
+              << "\" type=\"\">";
+      const std::string detail = location + "\n" + part.message();
+      OutputXmlCDataSection(stream, RemoveInvalidXmlCharacters(detail).c_str());
+      *stream << "</failure>\n";
+    }
+  }
+
+  if (failures == 0)
+    *stream << " />\n";
+  else
+    *stream << "    </testcase>\n";
+}
+
+// Prints an XML representation of a TestCase object
+void XmlUnitTestResultPrinter::PrintXmlTestCase(std::ostream* stream,
+                                                const TestCase& test_case) {
+  const std::string kTestsuite = "testsuite";
+  *stream << "  <" << kTestsuite;
+  OutputXmlAttribute(stream, kTestsuite, "name", test_case.name());
+  OutputXmlAttribute(stream, kTestsuite, "tests",
+                     StreamableToString(test_case.reportable_test_count()));
+  OutputXmlAttribute(stream, kTestsuite, "failures",
+                     StreamableToString(test_case.failed_test_count()));
+  OutputXmlAttribute(
+      stream, kTestsuite, "disabled",
+      StreamableToString(test_case.reportable_disabled_test_count()));
+  OutputXmlAttribute(stream, kTestsuite, "errors", "0");
+  OutputXmlAttribute(stream, kTestsuite, "time",
+                     FormatTimeInMillisAsSeconds(test_case.elapsed_time()));
+  *stream << TestPropertiesAsXmlAttributes(test_case.ad_hoc_test_result())
+          << ">\n";
+
+  for (int i = 0; i < test_case.total_test_count(); ++i) {
+    if (test_case.GetTestInfo(i)->is_reportable())
+      OutputXmlTestInfo(stream, test_case.name(), *test_case.GetTestInfo(i));
+  }
+  *stream << "  </" << kTestsuite << ">\n";
+}
+
+// Prints an XML summary of unit_test to output stream out.
+void XmlUnitTestResultPrinter::PrintXmlUnitTest(std::ostream* stream,
+                                                const UnitTest& unit_test) {
+  const std::string kTestsuites = "testsuites";
+
+  *stream << "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n";
+  *stream << "<" << kTestsuites;
+
+  OutputXmlAttribute(stream, kTestsuites, "tests",
+                     StreamableToString(unit_test.reportable_test_count()));
+  OutputXmlAttribute(stream, kTestsuites, "failures",
+                     StreamableToString(unit_test.failed_test_count()));
+  OutputXmlAttribute(
+      stream, kTestsuites, "disabled",
+      StreamableToString(unit_test.reportable_disabled_test_count()));
+  OutputXmlAttribute(stream, kTestsuites, "errors", "0");
+  OutputXmlAttribute(
+      stream, kTestsuites, "timestamp",
+      FormatEpochTimeInMillisAsIso8601(unit_test.start_timestamp()));
+  OutputXmlAttribute(stream, kTestsuites, "time",
+                     FormatTimeInMillisAsSeconds(unit_test.elapsed_time()));
+
+  if (GTEST_FLAG(shuffle)) {
+    OutputXmlAttribute(stream, kTestsuites, "random_seed",
+                       StreamableToString(unit_test.random_seed()));
+  }
+
+  *stream << TestPropertiesAsXmlAttributes(unit_test.ad_hoc_test_result());
+
+  OutputXmlAttribute(stream, kTestsuites, "name", "AllTests");
+  *stream << ">\n";
+
+  for (int i = 0; i < unit_test.total_test_case_count(); ++i) {
+    if (unit_test.GetTestCase(i)->reportable_test_count() > 0)
+      PrintXmlTestCase(stream, *unit_test.GetTestCase(i));
+  }
+  *stream << "</" << kTestsuites << ">\n";
+}
+
+// Produces a string representing the test properties in a result as space
+// delimited XML attributes based on the property key="value" pairs.
+std::string XmlUnitTestResultPrinter::TestPropertiesAsXmlAttributes(
+    const TestResult& result) {
+  Message attributes;
+  for (int i = 0; i < result.test_property_count(); ++i) {
+    const TestProperty& property = result.GetTestProperty(i);
+    attributes << " " << property.key() << "="
+        << "\"" << EscapeXmlAttribute(property.value()) << "\"";
+  }
+  return attributes.GetString();
+}
+
+// End XmlUnitTestResultPrinter
+
+#if GTEST_CAN_STREAM_RESULTS_
+
+// Checks if str contains '=', '&', '%' or '\n' characters. If yes,
+// replaces them by "%xx" where xx is their hexadecimal value. For
+// example, replaces "=" with "%3D".  This algorithm is O(strlen(str))
+// in both time and space -- important as the input str may contain an
+// arbitrarily long test failure message and stack trace.
+std::string StreamingListener::UrlEncode(const char* str) {
+  std::string result;
+  result.reserve(strlen(str) + 1);
+  for (char ch = *str; ch != '\0'; ch = *++str) {
+    switch (ch) {
+      case '%':
+      case '=':
+      case '&':
+      case '\n':
+        result.append("%" + String::FormatByte(static_cast<unsigned char>(ch)));
+        break;
+      default:
+        result.push_back(ch);
+        break;
+    }
+  }
+  return result;
+}
+
+void StreamingListener::SocketWriter::MakeConnection() {
+  GTEST_CHECK_(sockfd_ == -1)
+      << "MakeConnection() can't be called when there is already a connection.";
+
+  addrinfo hints;
+  memset(&hints, 0, sizeof(hints));
+  hints.ai_family = AF_UNSPEC;    // To allow both IPv4 and IPv6 addresses.
+  hints.ai_socktype = SOCK_STREAM;
+  addrinfo* servinfo = NULL;
+
+  // Use the getaddrinfo() to get a linked list of IP addresses for
+  // the given host name.
+  const int error_num = getaddrinfo(
+      host_name_.c_str(), port_num_.c_str(), &hints, &servinfo);
+  if (error_num != 0) {
+    GTEST_LOG_(WARNING) << "stream_result_to: getaddrinfo() failed: "
+                        << gai_strerror(error_num);
+  }
+
+  // Loop through all the results and connect to the first we can.
+  for (addrinfo* cur_addr = servinfo; sockfd_ == -1 && cur_addr != NULL;
+       cur_addr = cur_addr->ai_next) {
+    sockfd_ = socket(
+        cur_addr->ai_family, cur_addr->ai_socktype, cur_addr->ai_protocol);
+    if (sockfd_ != -1) {
+      // Connect the client socket to the server socket.
+      if (connect(sockfd_, cur_addr->ai_addr, cur_addr->ai_addrlen) == -1) {
+        close(sockfd_);
+        sockfd_ = -1;
+      }
+    }
+  }
+
+  freeaddrinfo(servinfo);  // all done with this structure
+
+  if (sockfd_ == -1) {
+    GTEST_LOG_(WARNING) << "stream_result_to: failed to connect to "
+                        << host_name_ << ":" << port_num_;
+  }
+}
+
+// End of class Streaming Listener
+#endif  // GTEST_CAN_STREAM_RESULTS__
+
+// Class ScopedTrace
+
+// Pushes the given source file location and message onto a per-thread
+// trace stack maintained by Google Test.
+ScopedTrace::ScopedTrace(const char* file, int line, const Message& message)
+    GTEST_LOCK_EXCLUDED_(&UnitTest::mutex_) {
+  TraceInfo trace;
+  trace.file = file;
+  trace.line = line;
+  trace.message = message.GetString();
+
+  UnitTest::GetInstance()->PushGTestTrace(trace);
+}
+
+// Pops the info pushed by the c'tor.
+ScopedTrace::~ScopedTrace()
+    GTEST_LOCK_EXCLUDED_(&UnitTest::mutex_) {
+  UnitTest::GetInstance()->PopGTestTrace();
+}
+
+
+// class OsStackTraceGetter
+
+const char* const OsStackTraceGetterInterface::kElidedFramesMarker =
+    "... " GTEST_NAME_ " internal frames ...";
+
+std::string OsStackTraceGetter::CurrentStackTrace(int /*max_depth*/,
+                                                  int /*skip_count*/) {
+  return "";
+}
+
+void OsStackTraceGetter::UponLeavingGTest() {}
+
+// A helper class that creates the premature-exit file in its
+// constructor and deletes the file in its destructor.
+class ScopedPrematureExitFile {
+ public:
+  explicit ScopedPrematureExitFile(const char* premature_exit_filepath)
+      : premature_exit_filepath_(premature_exit_filepath) {
+    // If a path to the premature-exit file is specified...
+    if (premature_exit_filepath != NULL && *premature_exit_filepath != '\0') {
+      // create the file with a single "0" character in it.  I/O
+      // errors are ignored as there's nothing better we can do and we
+      // don't want to fail the test because of this.
+      FILE* pfile = posix::FOpen(premature_exit_filepath, "w");
+      fwrite("0", 1, 1, pfile);
+      fclose(pfile);
+    }
+  }
+
+  ~ScopedPrematureExitFile() {
+    if (premature_exit_filepath_ != NULL && *premature_exit_filepath_ != '\0') {
+      remove(premature_exit_filepath_);
+    }
+  }
+
+ private:
+  const char* const premature_exit_filepath_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedPrematureExitFile);
+};
+
+}  // namespace internal
+
+// class TestEventListeners
+
+TestEventListeners::TestEventListeners()
+    : repeater_(new internal::TestEventRepeater()),
+      default_result_printer_(NULL),
+      default_xml_generator_(NULL) {
+}
+
+TestEventListeners::~TestEventListeners() { delete repeater_; }
+
+// Returns the standard listener responsible for the default console
+// output.  Can be removed from the listeners list to shut down default
+// console output.  Note that removing this object from the listener list
+// with Release transfers its ownership to the user.
+void TestEventListeners::Append(TestEventListener* listener) {
+  repeater_->Append(listener);
+}
+
+// Removes the given event listener from the list and returns it.  It then
+// becomes the caller's responsibility to delete the listener. Returns
+// NULL if the listener is not found in the list.
+TestEventListener* TestEventListeners::Release(TestEventListener* listener) {
+  if (listener == default_result_printer_)
+    default_result_printer_ = NULL;
+  else if (listener == default_xml_generator_)
+    default_xml_generator_ = NULL;
+  return repeater_->Release(listener);
+}
+
+// Returns repeater that broadcasts the TestEventListener events to all
+// subscribers.
+TestEventListener* TestEventListeners::repeater() { return repeater_; }
+
+// Sets the default_result_printer attribute to the provided listener.
+// The listener is also added to the listener list and previous
+// default_result_printer is removed from it and deleted. The listener can
+// also be NULL in which case it will not be added to the list. Does
+// nothing if the previous and the current listener objects are the same.
+void TestEventListeners::SetDefaultResultPrinter(TestEventListener* listener) {
+  if (default_result_printer_ != listener) {
+    // It is an error to pass this method a listener that is already in the
+    // list.
+    delete Release(default_result_printer_);
+    default_result_printer_ = listener;
+    if (listener != NULL)
+      Append(listener);
+  }
+}
+
+// Sets the default_xml_generator attribute to the provided listener.  The
+// listener is also added to the listener list and previous
+// default_xml_generator is removed from it and deleted. The listener can
+// also be NULL in which case it will not be added to the list. Does
+// nothing if the previous and the current listener objects are the same.
+void TestEventListeners::SetDefaultXmlGenerator(TestEventListener* listener) {
+  if (default_xml_generator_ != listener) {
+    // It is an error to pass this method a listener that is already in the
+    // list.
+    delete Release(default_xml_generator_);
+    default_xml_generator_ = listener;
+    if (listener != NULL)
+      Append(listener);
+  }
+}
+
+// Controls whether events will be forwarded by the repeater to the
+// listeners in the list.
+bool TestEventListeners::EventForwardingEnabled() const {
+  return repeater_->forwarding_enabled();
+}
+
+void TestEventListeners::SuppressEventForwarding() {
+  repeater_->set_forwarding_enabled(false);
+}
+
+// class UnitTest
+
+// Gets the singleton UnitTest object.  The first time this method is
+// called, a UnitTest object is constructed and returned.  Consecutive
+// calls will return the same object.
+//
+// We don't protect this under mutex_ as a user is not supposed to
+// call this before main() starts, from which point on the return
+// value will never change.
+UnitTest* UnitTest::GetInstance() {
+  // When compiled with MSVC 7.1 in optimized mode, destroying the
+  // UnitTest object upon exiting the program messes up the exit code,
+  // causing successful tests to appear failed.  We have to use a
+  // different implementation in this case to bypass the compiler bug.
+  // This implementation makes the compiler happy, at the cost of
+  // leaking the UnitTest object.
+
+  // CodeGear C++Builder insists on a public destructor for the
+  // default implementation.  Use this implementation to keep good OO
+  // design with private destructor.
+
+#if (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__)
+  static UnitTest* const instance = new UnitTest;
+  return instance;
+#else
+  static UnitTest instance;
+  return &instance;
+#endif  // (_MSC_VER == 1310 && !defined(_DEBUG)) || defined(__BORLANDC__)
+}
+
+// Gets the number of successful test cases.
+int UnitTest::successful_test_case_count() const {
+  return impl()->successful_test_case_count();
+}
+
+// Gets the number of failed test cases.
+int UnitTest::failed_test_case_count() const {
+  return impl()->failed_test_case_count();
+}
+
+// Gets the number of all test cases.
+int UnitTest::total_test_case_count() const {
+  return impl()->total_test_case_count();
+}
+
+// Gets the number of all test cases that contain at least one test
+// that should run.
+int UnitTest::test_case_to_run_count() const {
+  return impl()->test_case_to_run_count();
+}
+
+// Gets the number of successful tests.
+int UnitTest::successful_test_count() const {
+  return impl()->successful_test_count();
+}
+
+// Gets the number of skipped tests.
+int UnitTest::skipped_test_count() const {
+  return impl()->skipped_test_count();
+}
+
+// Gets the number of failed tests.
+int UnitTest::failed_test_count() const { return impl()->failed_test_count(); }
+
+// Gets the number of disabled tests that will be reported in the XML report.
+int UnitTest::reportable_disabled_test_count() const {
+  return impl()->reportable_disabled_test_count();
+}
+
+// Gets the number of disabled tests.
+int UnitTest::disabled_test_count() const {
+  return impl()->disabled_test_count();
+}
+
+// Gets the number of tests to be printed in the XML report.
+int UnitTest::reportable_test_count() const {
+  return impl()->reportable_test_count();
+}
+
+// Gets the number of all tests.
+int UnitTest::total_test_count() const { return impl()->total_test_count(); }
+
+// Gets the number of tests that should run.
+int UnitTest::test_to_run_count() const { return impl()->test_to_run_count(); }
+
+// Gets the time of the test program start, in ms from the start of the
+// UNIX epoch.
+internal::TimeInMillis UnitTest::start_timestamp() const {
+    return impl()->start_timestamp();
+}
+
+// Gets the elapsed time, in milliseconds.
+internal::TimeInMillis UnitTest::elapsed_time() const {
+  return impl()->elapsed_time();
+}
+
+// Returns true iff the unit test passed (i.e. all test cases passed).
+bool UnitTest::Passed() const { return impl()->Passed(); }
+
+// Returns true iff the unit test failed (i.e. some test case failed
+// or something outside of all tests failed).
+bool UnitTest::Failed() const { return impl()->Failed(); }
+
+// Gets the i-th test case among all the test cases. i can range from 0 to
+// total_test_case_count() - 1. If i is not in that range, returns NULL.
+const TestCase* UnitTest::GetTestCase(int i) const {
+  return impl()->GetTestCase(i);
+}
+
+// Returns the TestResult containing information on test failures and
+// properties logged outside of individual test cases.
+const TestResult& UnitTest::ad_hoc_test_result() const {
+  return *impl()->ad_hoc_test_result();
+}
+
+// Gets the i-th test case among all the test cases. i can range from 0 to
+// total_test_case_count() - 1. If i is not in that range, returns NULL.
+TestCase* UnitTest::GetMutableTestCase(int i) {
+  return impl()->GetMutableTestCase(i);
+}
+
+// Returns the list of event listeners that can be used to track events
+// inside Google Test.
+TestEventListeners& UnitTest::listeners() {
+  return *impl()->listeners();
+}
+
+// Registers and returns a global test environment.  When a test
+// program is run, all global test environments will be set-up in the
+// order they were registered.  After all tests in the program have
+// finished, all global test environments will be torn-down in the
+// *reverse* order they were registered.
+//
+// The UnitTest object takes ownership of the given environment.
+//
+// We don't protect this under mutex_, as we only support calling it
+// from the main thread.
+Environment* UnitTest::AddEnvironment(Environment* env) {
+  if (env == NULL) {
+    return NULL;
+  }
+
+  impl_->environments().push_back(env);
+  return env;
+}
+
+// Adds a TestPartResult to the current TestResult object.  All Google Test
+// assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc) eventually call
+// this to report their results.  The user code should use the
+// assertion macros instead of calling this directly.
+void UnitTest::AddTestPartResult(
+    TestPartResult::Type result_type,
+    const char* file_name,
+    int line_number,
+    const std::string& message,
+    const std::string& os_stack_trace) GTEST_LOCK_EXCLUDED_(mutex_) {
+  Message msg;
+  msg << message;
+
+  internal::MutexLock lock(&mutex_);
+  if (impl_->gtest_trace_stack().size() > 0) {
+    msg << "\n" << GTEST_NAME_ << " trace:";
+
+    for (int i = static_cast<int>(impl_->gtest_trace_stack().size());
+         i > 0; --i) {
+      const internal::TraceInfo& trace = impl_->gtest_trace_stack()[i - 1];
+      msg << "\n" << internal::FormatFileLocation(trace.file, trace.line)
+          << " " << trace.message;
+    }
+  }
+
+  if (os_stack_trace.c_str() != NULL && !os_stack_trace.empty()) {
+    msg << internal::kStackTraceMarker << os_stack_trace;
+  }
+
+  const TestPartResult result =
+    TestPartResult(result_type, file_name, line_number,
+                   msg.GetString().c_str());
+  impl_->GetTestPartResultReporterForCurrentThread()->
+      ReportTestPartResult(result);
+
+  if (result_type != TestPartResult::kSuccess &&
+      result_type != TestPartResult::kSkip) {
+    // gtest_break_on_failure takes precedence over
+    // gtest_throw_on_failure.  This allows a user to set the latter
+    // in the code (perhaps in order to use Google Test assertions
+    // with another testing framework) and specify the former on the
+    // command line for debugging.
+    if (GTEST_FLAG(break_on_failure)) {
+#if GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+      // Using DebugBreak on Windows allows gtest to still break into a debugger
+      // when a failure happens and both the --gtest_break_on_failure and
+      // the --gtest_catch_exceptions flags are specified.
+      DebugBreak();
+#else
+      // Dereference NULL through a volatile pointer to prevent the compiler
+      // from removing. We use this rather than abort() or __builtin_trap() for
+      // portability: Symbian doesn't implement abort() well, and some debuggers
+      // don't correctly trap abort().
+      *static_cast<volatile int*>(NULL) = 1;
+#endif  // GTEST_OS_WINDOWS
+    } else if (GTEST_FLAG(throw_on_failure)) {
+#if GTEST_HAS_EXCEPTIONS
+      throw internal::GoogleTestFailureException(result);
+#else
+      // We cannot call abort() as it generates a pop-up in debug mode
+      // that cannot be suppressed in VC 7.1 or below.
+      exit(1);
+#endif
+    }
+  }
+}
+
+// Adds a TestProperty to the current TestResult object when invoked from
+// inside a test, to current TestCase's ad_hoc_test_result_ when invoked
+// from SetUpTestCase or TearDownTestCase, or to the global property set
+// when invoked elsewhere.  If the result already contains a property with
+// the same key, the value will be updated.
+void UnitTest::RecordProperty(const std::string& key,
+                              const std::string& value) {
+  impl_->RecordProperty(TestProperty(key, value));
+}
+
+// Runs all tests in this UnitTest object and prints the result.
+// Returns 0 if successful, or 1 otherwise.
+//
+// We don't protect this under mutex_, as we only support calling it
+// from the main thread.
+int UnitTest::Run() {
+  const bool in_death_test_child_process =
+      internal::GTEST_FLAG(internal_run_death_test).length() > 0;
+
+  // Google Test implements this protocol for catching that a test
+  // program exits before returning control to Google Test:
+  //
+  //   1. Upon start, Google Test creates a file whose absolute path
+  //      is specified by the environment variable
+  //      TEST_PREMATURE_EXIT_FILE.
+  //   2. When Google Test has finished its work, it deletes the file.
+  //
+  // This allows a test runner to set TEST_PREMATURE_EXIT_FILE before
+  // running a Google-Test-based test program and check the existence
+  // of the file at the end of the test execution to see if it has
+  // exited prematurely.
+
+  // If we are in the child process of a death test, don't
+  // create/delete the premature exit file, as doing so is unnecessary
+  // and will confuse the parent process.  Otherwise, create/delete
+  // the file upon entering/leaving this function.  If the program
+  // somehow exits before this function has a chance to return, the
+  // premature-exit file will be left undeleted, causing a test runner
+  // that understands the premature-exit-file protocol to report the
+  // test as having failed.
+  const internal::ScopedPrematureExitFile premature_exit_file(
+      in_death_test_child_process ?
+      NULL : internal::posix::GetEnv("TEST_PREMATURE_EXIT_FILE"));
+
+  // Captures the value of GTEST_FLAG(catch_exceptions).  This value will be
+  // used for the duration of the program.
+  impl()->set_catch_exceptions(GTEST_FLAG(catch_exceptions));
+
+#if GTEST_HAS_SEH
+  // Either the user wants Google Test to catch exceptions thrown by the
+  // tests or this is executing in the context of death test child
+  // process. In either case the user does not want to see pop-up dialogs
+  // about crashes - they are expected.
+  if (impl()->catch_exceptions() || in_death_test_child_process) {
+# if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+    // SetErrorMode doesn't exist on CE.
+    SetErrorMode(SEM_FAILCRITICALERRORS | SEM_NOALIGNMENTFAULTEXCEPT |
+                 SEM_NOGPFAULTERRORBOX | SEM_NOOPENFILEERRORBOX);
+# endif  // !GTEST_OS_WINDOWS_MOBILE
+
+# if (defined(_MSC_VER) || GTEST_OS_WINDOWS_MINGW) && !GTEST_OS_WINDOWS_MOBILE
+    // Death test children can be terminated with _abort().  On Windows,
+    // _abort() can show a dialog with a warning message.  This forces the
+    // abort message to go to stderr instead.
+    _set_error_mode(_OUT_TO_STDERR);
+# endif
+
+# if _MSC_VER >= 1400 && !GTEST_OS_WINDOWS_MOBILE
+    // In the debug version, Visual Studio pops up a separate dialog
+    // offering a choice to debug the aborted program. We need to suppress
+    // this dialog or it will pop up for every EXPECT/ASSERT_DEATH statement
+    // executed. Google Test will notify the user of any unexpected
+    // failure via stderr.
+    //
+    // VC++ doesn't define _set_abort_behavior() prior to the version 8.0.
+    // Users of prior VC versions shall suffer the agony and pain of
+    // clicking through the countless debug dialogs.
+    // TODO(vladl@google.com): find a way to suppress the abort dialog() in the
+    // debug mode when compiled with VC 7.1 or lower.
+    if (!GTEST_FLAG(break_on_failure))
+      _set_abort_behavior(
+          0x0,                                    // Clear the following flags:
+          _WRITE_ABORT_MSG | _CALL_REPORTFAULT);  // pop-up window, core dump.
+# endif
+  }
+#endif  // GTEST_HAS_SEH
+
+  return internal::HandleExceptionsInMethodIfSupported(
+      impl(),
+      &internal::UnitTestImpl::RunAllTests,
+      "auxiliary test code (environments or event listeners)") ? 0 : 1;
+}
+
+// Returns the working directory when the first TEST() or TEST_F() was
+// executed.
+const char* UnitTest::original_working_dir() const {
+  return impl_->original_working_dir_.c_str();
+}
+
+// Returns the TestCase object for the test that's currently running,
+// or NULL if no test is running.
+const TestCase* UnitTest::current_test_case() const
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  return impl_->current_test_case();
+}
+
+// Returns the TestInfo object for the test that's currently running,
+// or NULL if no test is running.
+const TestInfo* UnitTest::current_test_info() const
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  return impl_->current_test_info();
+}
+
+// Returns the random seed used at the start of the current test run.
+int UnitTest::random_seed() const { return impl_->random_seed(); }
+
+#if GTEST_HAS_PARAM_TEST
+// Returns ParameterizedTestCaseRegistry object used to keep track of
+// value-parameterized tests and instantiate and register them.
+internal::ParameterizedTestCaseRegistry&
+    UnitTest::parameterized_test_registry()
+        GTEST_LOCK_EXCLUDED_(mutex_) {
+  return impl_->parameterized_test_registry();
+}
+#endif  // GTEST_HAS_PARAM_TEST
+
+// Creates an empty UnitTest.
+UnitTest::UnitTest() {
+  impl_ = new internal::UnitTestImpl(this);
+}
+
+// Destructor of UnitTest.
+UnitTest::~UnitTest() {
+  delete impl_;
+}
+
+// Pushes a trace defined by SCOPED_TRACE() on to the per-thread
+// Google Test trace stack.
+void UnitTest::PushGTestTrace(const internal::TraceInfo& trace)
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  impl_->gtest_trace_stack().push_back(trace);
+}
+
+// Pops a trace from the per-thread Google Test trace stack.
+void UnitTest::PopGTestTrace()
+    GTEST_LOCK_EXCLUDED_(mutex_) {
+  internal::MutexLock lock(&mutex_);
+  impl_->gtest_trace_stack().pop_back();
+}
+
+namespace internal {
+
+UnitTestImpl::UnitTestImpl(UnitTest* parent)
+    : parent_(parent),
+      GTEST_DISABLE_MSC_WARNINGS_PUSH_(4355 /* using this in initializer */)
+      default_global_test_part_result_reporter_(this),
+      default_per_thread_test_part_result_reporter_(this),
+      GTEST_DISABLE_MSC_WARNINGS_POP_()
+      global_test_part_result_repoter_(
+          &default_global_test_part_result_reporter_),
+      per_thread_test_part_result_reporter_(
+          &default_per_thread_test_part_result_reporter_),
+#if GTEST_HAS_PARAM_TEST
+      parameterized_test_registry_(),
+      parameterized_tests_registered_(false),
+#endif  // GTEST_HAS_PARAM_TEST
+      last_death_test_case_(-1),
+      current_test_case_(NULL),
+      current_test_info_(NULL),
+      ad_hoc_test_result_(),
+      os_stack_trace_getter_(NULL),
+      post_flag_parse_init_performed_(false),
+      random_seed_(0),  // Will be overridden by the flag before first use.
+      random_(0),  // Will be reseeded before first use.
+      start_timestamp_(0),
+      elapsed_time_(0),
+#if GTEST_HAS_DEATH_TEST
+      death_test_factory_(new DefaultDeathTestFactory),
+#endif
+      // Will be overridden by the flag before first use.
+      catch_exceptions_(false) {
+  listeners()->SetDefaultResultPrinter(new PrettyUnitTestResultPrinter);
+}
+
+UnitTestImpl::~UnitTestImpl() {
+  // Deletes every TestCase.
+  ForEach(test_cases_, internal::Delete<TestCase>);
+
+  // Deletes every Environment.
+  ForEach(environments_, internal::Delete<Environment>);
+
+  delete os_stack_trace_getter_;
+}
+
+// Adds a TestProperty to the current TestResult object when invoked in a
+// context of a test, to current test case's ad_hoc_test_result when invoke
+// from SetUpTestCase/TearDownTestCase, or to the global property set
+// otherwise.  If the result already contains a property with the same key,
+// the value will be updated.
+void UnitTestImpl::RecordProperty(const TestProperty& test_property) {
+  std::string xml_element;
+  TestResult* test_result;  // TestResult appropriate for property recording.
+
+  if (current_test_info_ != NULL) {
+    xml_element = "testcase";
+    test_result = &(current_test_info_->result_);
+  } else if (current_test_case_ != NULL) {
+    xml_element = "testsuite";
+    test_result = &(current_test_case_->ad_hoc_test_result_);
+  } else {
+    xml_element = "testsuites";
+    test_result = &ad_hoc_test_result_;
+  }
+  test_result->RecordProperty(xml_element, test_property);
+}
+
+#if GTEST_HAS_DEATH_TEST
+// Disables event forwarding if the control is currently in a death test
+// subprocess. Must not be called before InitGoogleTest.
+void UnitTestImpl::SuppressTestEventsIfInSubprocess() {
+  if (internal_run_death_test_flag_.get() != NULL)
+    listeners()->SuppressEventForwarding();
+}
+#endif  // GTEST_HAS_DEATH_TEST
+
+// Initializes event listeners performing XML output as specified by
+// UnitTestOptions. Must not be called before InitGoogleTest.
+void UnitTestImpl::ConfigureXmlOutput() {
+  const std::string& output_format = UnitTestOptions::GetOutputFormat();
+  if (output_format == "xml") {
+    listeners()->SetDefaultXmlGenerator(new XmlUnitTestResultPrinter(
+        UnitTestOptions::GetAbsolutePathToOutputFile().c_str()));
+  } else if (output_format != "") {
+    printf("WARNING: unrecognized output format \"%s\" ignored.\n",
+           output_format.c_str());
+    fflush(stdout);
+  }
+}
+
+#if GTEST_CAN_STREAM_RESULTS_
+// Initializes event listeners for streaming test results in string form.
+// Must not be called before InitGoogleTest.
+void UnitTestImpl::ConfigureStreamingOutput() {
+  const std::string& target = GTEST_FLAG(stream_result_to);
+  if (!target.empty()) {
+    const size_t pos = target.find(':');
+    if (pos != std::string::npos) {
+      listeners()->Append(new StreamingListener(target.substr(0, pos),
+                                                target.substr(pos+1)));
+    } else {
+      printf("WARNING: unrecognized streaming target \"%s\" ignored.\n",
+             target.c_str());
+      fflush(stdout);
+    }
+  }
+}
+#endif  // GTEST_CAN_STREAM_RESULTS_
+
+// Performs initialization dependent upon flag values obtained in
+// ParseGoogleTestFlagsOnly.  Is called from InitGoogleTest after the call to
+// ParseGoogleTestFlagsOnly.  In case a user neglects to call InitGoogleTest
+// this function is also called from RunAllTests.  Since this function can be
+// called more than once, it has to be idempotent.
+void UnitTestImpl::PostFlagParsingInit() {
+  // Ensures that this function does not execute more than once.
+  if (!post_flag_parse_init_performed_) {
+    post_flag_parse_init_performed_ = true;
+
+#if defined(GTEST_CUSTOM_TEST_EVENT_LISTENER_)
+    // Register to send notifications about key process state changes.
+    listeners()->Append(new GTEST_CUSTOM_TEST_EVENT_LISTENER_());
+#endif  // defined(GTEST_CUSTOM_TEST_EVENT_LISTENER_)
+
+#if GTEST_HAS_DEATH_TEST
+    InitDeathTestSubprocessControlInfo();
+    SuppressTestEventsIfInSubprocess();
+#endif  // GTEST_HAS_DEATH_TEST
+
+    // Registers parameterized tests. This makes parameterized tests
+    // available to the UnitTest reflection API without running
+    // RUN_ALL_TESTS.
+    RegisterParameterizedTests();
+
+    // Configures listeners for XML output. This makes it possible for users
+    // to shut down the default XML output before invoking RUN_ALL_TESTS.
+    ConfigureXmlOutput();
+
+#if GTEST_CAN_STREAM_RESULTS_
+    // Configures listeners for streaming test results to the specified server.
+    ConfigureStreamingOutput();
+#endif  // GTEST_CAN_STREAM_RESULTS_
+  }
+}
+
+// A predicate that checks the name of a TestCase against a known
+// value.
+//
+// This is used for implementation of the UnitTest class only.  We put
+// it in the anonymous namespace to prevent polluting the outer
+// namespace.
+//
+// TestCaseNameIs is copyable.
+class TestCaseNameIs {
+ public:
+  // Constructor.
+  explicit TestCaseNameIs(const std::string& name)
+      : name_(name) {}
+
+  // Returns true iff the name of test_case matches name_.
+  bool operator()(const TestCase* test_case) const {
+    return test_case != NULL && strcmp(test_case->name(), name_.c_str()) == 0;
+  }
+
+ private:
+  std::string name_;
+};
+
+// Finds and returns a TestCase with the given name.  If one doesn't
+// exist, creates one and returns it.  It's the CALLER'S
+// RESPONSIBILITY to ensure that this function is only called WHEN THE
+// TESTS ARE NOT SHUFFLED.
+//
+// Arguments:
+//
+//   test_case_name: name of the test case
+//   type_param:     the name of the test case's type parameter, or NULL if
+//                   this is not a typed or a type-parameterized test case.
+//   set_up_tc:      pointer to the function that sets up the test case
+//   tear_down_tc:   pointer to the function that tears down the test case
+TestCase* UnitTestImpl::GetTestCase(const char* test_case_name,
+                                    const char* type_param,
+                                    Test::SetUpTestCaseFunc set_up_tc,
+                                    Test::TearDownTestCaseFunc tear_down_tc) {
+  // Can we find a TestCase with the given name?
+  const std::vector<TestCase*>::const_iterator test_case =
+      std::find_if(test_cases_.begin(), test_cases_.end(),
+                   TestCaseNameIs(test_case_name));
+
+  if (test_case != test_cases_.end())
+    return *test_case;
+
+  // No.  Let's create one.
+  TestCase* const new_test_case =
+      new TestCase(test_case_name, type_param, set_up_tc, tear_down_tc);
+
+  // Is this a death test case?
+  if (internal::UnitTestOptions::MatchesFilter(test_case_name,
+                                               kDeathTestCaseFilter)) {
+    // Yes.  Inserts the test case after the last death test case
+    // defined so far.  This only works when the test cases haven't
+    // been shuffled.  Otherwise we may end up running a death test
+    // after a non-death test.
+    ++last_death_test_case_;
+    test_cases_.insert(test_cases_.begin() + last_death_test_case_,
+                       new_test_case);
+  } else {
+    // No.  Appends to the end of the list.
+    test_cases_.push_back(new_test_case);
+  }
+
+  test_case_indices_.push_back(static_cast<int>(test_case_indices_.size()));
+  return new_test_case;
+}
+
+// Helpers for setting up / tearing down the given environment.  They
+// are for use in the ForEach() function.
+static void SetUpEnvironment(Environment* env) { env->SetUp(); }
+static void TearDownEnvironment(Environment* env) { env->TearDown(); }
+
+// Runs all tests in this UnitTest object, prints the result, and
+// returns true if all tests are successful.  If any exception is
+// thrown during a test, the test is considered to be failed, but the
+// rest of the tests will still be run.
+//
+// When parameterized tests are enabled, it expands and registers
+// parameterized tests first in RegisterParameterizedTests().
+// All other functions called from RunAllTests() may safely assume that
+// parameterized tests are ready to be counted and run.
+bool UnitTestImpl::RunAllTests() {
+  // Makes sure InitGoogleTest() was called.
+  if (!GTestIsInitialized()) {
+    printf("%s",
+           "\nThis test program did NOT call ::testing::InitGoogleTest "
+           "before calling RUN_ALL_TESTS().  Please fix it.\n");
+    return false;
+  }
+
+  // Do not run any test if the --help flag was specified.
+  if (g_help_flag)
+    return true;
+
+  // Repeats the call to the post-flag parsing initialization in case the
+  // user didn't call InitGoogleTest.
+  PostFlagParsingInit();
+
+  // Even if sharding is not on, test runners may want to use the
+  // GTEST_SHARD_STATUS_FILE to query whether the test supports the sharding
+  // protocol.
+  internal::WriteToShardStatusFileIfNeeded();
+
+  // True iff we are in a subprocess for running a thread-safe-style
+  // death test.
+  bool in_subprocess_for_death_test = false;
+
+#if GTEST_HAS_DEATH_TEST
+  in_subprocess_for_death_test = (internal_run_death_test_flag_.get() != NULL);
+# if defined(GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_)
+  if (in_subprocess_for_death_test) {
+    GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_();
+  }
+# endif  // defined(GTEST_EXTRA_DEATH_TEST_CHILD_SETUP_)
+#endif  // GTEST_HAS_DEATH_TEST
+
+  const bool should_shard = ShouldShard(kTestTotalShards, kTestShardIndex,
+                                        in_subprocess_for_death_test);
+
+  // Compares the full test names with the filter to decide which
+  // tests to run.
+  const bool has_tests_to_run = FilterTests(should_shard
+                                              ? HONOR_SHARDING_PROTOCOL
+                                              : IGNORE_SHARDING_PROTOCOL) > 0;
+
+  // Lists the tests and exits if the --gtest_list_tests flag was specified.
+  if (GTEST_FLAG(list_tests)) {
+    // This must be called *after* FilterTests() has been called.
+    ListTestsMatchingFilter();
+    return true;
+  }
+
+  random_seed_ = GTEST_FLAG(shuffle) ?
+      GetRandomSeedFromFlag(GTEST_FLAG(random_seed)) : 0;
+
+  // True iff at least one test has failed.
+  bool failed = false;
+
+  TestEventListener* repeater = listeners()->repeater();
+
+  start_timestamp_ = GetTimeInMillis();
+  repeater->OnTestProgramStart(*parent_);
+
+  // How many times to repeat the tests?  We don't want to repeat them
+  // when we are inside the subprocess of a death test.
+  const int repeat = in_subprocess_for_death_test ? 1 : GTEST_FLAG(repeat);
+  // Repeats forever if the repeat count is negative.
+  const bool forever = repeat < 0;
+  for (int i = 0; forever || i != repeat; i++) {
+    // We want to preserve failures generated by ad-hoc test
+    // assertions executed before RUN_ALL_TESTS().
+    ClearNonAdHocTestResult();
+
+    const TimeInMillis start = GetTimeInMillis();
+
+    // Shuffles test cases and tests if requested.
+    if (has_tests_to_run && GTEST_FLAG(shuffle)) {
+      random()->Reseed(random_seed_);
+      // This should be done before calling OnTestIterationStart(),
+      // such that a test event listener can see the actual test order
+      // in the event.
+      ShuffleTests();
+    }
+
+    // Tells the unit test event listeners that the tests are about to start.
+    repeater->OnTestIterationStart(*parent_, i);
+
+    // Runs each test case if there is at least one test to run.
+    if (has_tests_to_run) {
+      // Sets up all environments beforehand.
+      repeater->OnEnvironmentsSetUpStart(*parent_);
+      ForEach(environments_, SetUpEnvironment);
+      repeater->OnEnvironmentsSetUpEnd(*parent_);
+
+      // Runs the tests only if there was no fatal failure or skip triggered
+      //during global set-up.
+      if (Test::IsSkipped()) {
+        // Emit diagnostics when global set-up calls skip, as it will not be
+        // emitted by default.
+        TestResult& test_result =
+            *internal::GetUnitTestImpl()->current_test_result();
+        for (int j = 0; j < test_result.total_part_count(); ++j) {
+          const TestPartResult& test_part_result =
+              test_result.GetTestPartResult(j);
+          if (test_part_result.type() == TestPartResult::kSkip) {
+            const std::string& result = test_part_result.message();
+            printf("%s\n", result.c_str());
+          }
+        }
+        fflush(stdout);
+      } else if (!Test::HasFatalFailure()) {
+        for (int test_index = 0; test_index < total_test_case_count();
+             test_index++) {
+          GetMutableTestCase(test_index)->Run();
+        }
+      }
+
+      // Tears down all environments in reverse order afterwards.
+      repeater->OnEnvironmentsTearDownStart(*parent_);
+      std::for_each(environments_.rbegin(), environments_.rend(),
+                    TearDownEnvironment);
+      repeater->OnEnvironmentsTearDownEnd(*parent_);
+    }
+
+    elapsed_time_ = GetTimeInMillis() - start;
+
+    // Tells the unit test event listener that the tests have just finished.
+    repeater->OnTestIterationEnd(*parent_, i);
+
+    // Gets the result and clears it.
+    if (!Passed()) {
+      failed = true;
+    }
+
+    // Restores the original test order after the iteration.  This
+    // allows the user to quickly repro a failure that happens in the
+    // N-th iteration without repeating the first (N - 1) iterations.
+    // This is not enclosed in "if (GTEST_FLAG(shuffle)) { ... }", in
+    // case the user somehow changes the value of the flag somewhere
+    // (it's always safe to unshuffle the tests).
+    UnshuffleTests();
+
+    if (GTEST_FLAG(shuffle)) {
+      // Picks a new random seed for each iteration.
+      random_seed_ = GetNextRandomSeed(random_seed_);
+    }
+  }
+
+  repeater->OnTestProgramEnd(*parent_);
+
+  return !failed;
+}
+
+// Reads the GTEST_SHARD_STATUS_FILE environment variable, and creates the file
+// if the variable is present. If a file already exists at this location, this
+// function will write over it. If the variable is present, but the file cannot
+// be created, prints an error and exits.
+void WriteToShardStatusFileIfNeeded() {
+  const char* const test_shard_file = posix::GetEnv(kTestShardStatusFile);
+  if (test_shard_file != NULL) {
+    FILE* const file = posix::FOpen(test_shard_file, "w");
+    if (file == NULL) {
+      ColoredPrintf(COLOR_RED,
+                    "Could not write to the test shard status file \"%s\" "
+                    "specified by the %s environment variable.\n",
+                    test_shard_file, kTestShardStatusFile);
+      fflush(stdout);
+      exit(EXIT_FAILURE);
+    }
+    fclose(file);
+  }
+}
+
+// Checks whether sharding is enabled by examining the relevant
+// environment variable values. If the variables are present,
+// but inconsistent (i.e., shard_index >= total_shards), prints
+// an error and exits. If in_subprocess_for_death_test, sharding is
+// disabled because it must only be applied to the original test
+// process. Otherwise, we could filter out death tests we intended to execute.
+bool ShouldShard(const char* total_shards_env,
+                 const char* shard_index_env,
+                 bool in_subprocess_for_death_test) {
+  if (in_subprocess_for_death_test) {
+    return false;
+  }
+
+  const Int32 total_shards = Int32FromEnvOrDie(total_shards_env, -1);
+  const Int32 shard_index = Int32FromEnvOrDie(shard_index_env, -1);
+
+  if (total_shards == -1 && shard_index == -1) {
+    return false;
+  } else if (total_shards == -1 && shard_index != -1) {
+    const Message msg = Message()
+      << "Invalid environment variables: you have "
+      << kTestShardIndex << " = " << shard_index
+      << ", but have left " << kTestTotalShards << " unset.\n";
+    ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str());
+    fflush(stdout);
+    exit(EXIT_FAILURE);
+  } else if (total_shards != -1 && shard_index == -1) {
+    const Message msg = Message()
+      << "Invalid environment variables: you have "
+      << kTestTotalShards << " = " << total_shards
+      << ", but have left " << kTestShardIndex << " unset.\n";
+    ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str());
+    fflush(stdout);
+    exit(EXIT_FAILURE);
+  } else if (shard_index < 0 || shard_index >= total_shards) {
+    const Message msg = Message()
+      << "Invalid environment variables: we require 0 <= "
+      << kTestShardIndex << " < " << kTestTotalShards
+      << ", but you have " << kTestShardIndex << "=" << shard_index
+      << ", " << kTestTotalShards << "=" << total_shards << ".\n";
+    ColoredPrintf(COLOR_RED, "%s", msg.GetString().c_str());
+    fflush(stdout);
+    exit(EXIT_FAILURE);
+  }
+
+  return total_shards > 1;
+}
+
+// Parses the environment variable var as an Int32. If it is unset,
+// returns default_val. If it is not an Int32, prints an error
+// and aborts.
+Int32 Int32FromEnvOrDie(const char* var, Int32 default_val) {
+  const char* str_val = posix::GetEnv(var);
+  if (str_val == NULL) {
+    return default_val;
+  }
+
+  Int32 result;
+  if (!ParseInt32(Message() << "The value of environment variable " << var,
+                  str_val, &result)) {
+    exit(EXIT_FAILURE);
+  }
+  return result;
+}
+
+// Given the total number of shards, the shard index, and the test id,
+// returns true iff the test should be run on this shard. The test id is
+// some arbitrary but unique non-negative integer assigned to each test
+// method. Assumes that 0 <= shard_index < total_shards.
+bool ShouldRunTestOnShard(int total_shards, int shard_index, int test_id) {
+  return (test_id % total_shards) == shard_index;
+}
+
+// Compares the name of each test with the user-specified filter to
+// decide whether the test should be run, then records the result in
+// each TestCase and TestInfo object.
+// If shard_tests == true, further filters tests based on sharding
+// variables in the environment - see
+// http://code.google.com/p/googletest/wiki/GoogleTestAdvancedGuide.
+// Returns the number of tests that should run.
+int UnitTestImpl::FilterTests(ReactionToSharding shard_tests) {
+  const Int32 total_shards = shard_tests == HONOR_SHARDING_PROTOCOL ?
+      Int32FromEnvOrDie(kTestTotalShards, -1) : -1;
+  const Int32 shard_index = shard_tests == HONOR_SHARDING_PROTOCOL ?
+      Int32FromEnvOrDie(kTestShardIndex, -1) : -1;
+
+  // num_runnable_tests are the number of tests that will
+  // run across all shards (i.e., match filter and are not disabled).
+  // num_selected_tests are the number of tests to be run on
+  // this shard.
+  int num_runnable_tests = 0;
+  int num_selected_tests = 0;
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    TestCase* const test_case = test_cases_[i];
+    const std::string &test_case_name = test_case->name();
+    test_case->set_should_run(false);
+
+    for (size_t j = 0; j < test_case->test_info_list().size(); j++) {
+      TestInfo* const test_info = test_case->test_info_list()[j];
+      const std::string test_name(test_info->name());
+      // A test is disabled if test case name or test name matches
+      // kDisableTestFilter.
+      const bool is_disabled =
+          internal::UnitTestOptions::MatchesFilter(test_case_name,
+                                                   kDisableTestFilter) ||
+          internal::UnitTestOptions::MatchesFilter(test_name,
+                                                   kDisableTestFilter);
+      test_info->is_disabled_ = is_disabled;
+
+      const bool matches_filter =
+          internal::UnitTestOptions::FilterMatchesTest(test_case_name,
+                                                       test_name);
+      test_info->matches_filter_ = matches_filter;
+
+      const bool is_runnable =
+          (GTEST_FLAG(also_run_disabled_tests) || !is_disabled) &&
+          matches_filter;
+
+      const bool is_selected = is_runnable &&
+          (shard_tests == IGNORE_SHARDING_PROTOCOL ||
+           ShouldRunTestOnShard(total_shards, shard_index,
+                                num_runnable_tests));
+
+      num_runnable_tests += is_runnable;
+      num_selected_tests += is_selected;
+
+      test_info->should_run_ = is_selected;
+      test_case->set_should_run(test_case->should_run() || is_selected);
+    }
+  }
+  return num_selected_tests;
+}
+
+// Prints the given C-string on a single line by replacing all '\n'
+// characters with string "\\n".  If the output takes more than
+// max_length characters, only prints the first max_length characters
+// and "...".
+static void PrintOnOneLine(const char* str, int max_length) {
+  if (str != NULL) {
+    for (int i = 0; *str != '\0'; ++str) {
+      if (i >= max_length) {
+        printf("...");
+        break;
+      }
+      if (*str == '\n') {
+        printf("\\n");
+        i += 2;
+      } else {
+        printf("%c", *str);
+        ++i;
+      }
+    }
+  }
+}
+
+// Prints the names of the tests matching the user-specified filter flag.
+void UnitTestImpl::ListTestsMatchingFilter() {
+  // Print at most this many characters for each type/value parameter.
+  const int kMaxParamLength = 250;
+
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    const TestCase* const test_case = test_cases_[i];
+    bool printed_test_case_name = false;
+
+    for (size_t j = 0; j < test_case->test_info_list().size(); j++) {
+      const TestInfo* const test_info =
+          test_case->test_info_list()[j];
+      if (test_info->matches_filter_) {
+        if (!printed_test_case_name) {
+          printed_test_case_name = true;
+          printf("%s.", test_case->name());
+          if (test_case->type_param() != NULL) {
+            printf("  # %s = ", kTypeParamLabel);
+            // We print the type parameter on a single line to make
+            // the output easy to parse by a program.
+            PrintOnOneLine(test_case->type_param(), kMaxParamLength);
+          }
+          printf("\n");
+        }
+        printf("  %s", test_info->name());
+        printf("\n");
+      }
+    }
+  }
+  fflush(stdout);
+}
+
+// Sets the OS stack trace getter.
+//
+// Does nothing if the input and the current OS stack trace getter are
+// the same; otherwise, deletes the old getter and makes the input the
+// current getter.
+void UnitTestImpl::set_os_stack_trace_getter(
+    OsStackTraceGetterInterface* getter) {
+  if (os_stack_trace_getter_ != getter) {
+    delete os_stack_trace_getter_;
+    os_stack_trace_getter_ = getter;
+  }
+}
+
+// Returns the current OS stack trace getter if it is not NULL;
+// otherwise, creates an OsStackTraceGetter, makes it the current
+// getter, and returns it.
+OsStackTraceGetterInterface* UnitTestImpl::os_stack_trace_getter() {
+  if (os_stack_trace_getter_ == NULL) {
+#ifdef GTEST_OS_STACK_TRACE_GETTER_
+    os_stack_trace_getter_ = new GTEST_OS_STACK_TRACE_GETTER_;
+#else
+    os_stack_trace_getter_ = new OsStackTraceGetter;
+#endif  // GTEST_OS_STACK_TRACE_GETTER_
+  }
+
+  return os_stack_trace_getter_;
+}
+
+// Returns the TestResult for the test that's currently running, or
+// the TestResult for the ad hoc test if no test is running.
+TestResult* UnitTestImpl::current_test_result() {
+  return current_test_info_ ?
+      &(current_test_info_->result_) : &ad_hoc_test_result_;
+}
+
+// Shuffles all test cases, and the tests within each test case,
+// making sure that death tests are still run first.
+void UnitTestImpl::ShuffleTests() {
+  // Shuffles the death test cases.
+  ShuffleRange(random(), 0, last_death_test_case_ + 1, &test_case_indices_);
+
+  // Shuffles the non-death test cases.
+  ShuffleRange(random(), last_death_test_case_ + 1,
+               static_cast<int>(test_cases_.size()), &test_case_indices_);
+
+  // Shuffles the tests inside each test case.
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    test_cases_[i]->ShuffleTests(random());
+  }
+}
+
+// Restores the test cases and tests to their order before the first shuffle.
+void UnitTestImpl::UnshuffleTests() {
+  for (size_t i = 0; i < test_cases_.size(); i++) {
+    // Unshuffles the tests in each test case.
+    test_cases_[i]->UnshuffleTests();
+    // Resets the index of each test case.
+    test_case_indices_[i] = static_cast<int>(i);
+  }
+}
+
+// Returns the current OS stack trace as an std::string.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag.  The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in
+// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't.
+std::string GetCurrentOsStackTraceExceptTop(UnitTest* /*unit_test*/,
+                                            int skip_count) {
+  // We pass skip_count + 1 to skip this wrapper function in addition
+  // to what the user really wants to skip.
+  return GetUnitTestImpl()->CurrentOsStackTraceExceptTop(skip_count + 1);
+}
+
+// Used by the GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_ macro to
+// suppress unreachable code warnings.
+namespace {
+class ClassUniqueToAlwaysTrue {};
+}
+
+bool IsTrue(bool condition) { return condition; }
+
+bool AlwaysTrue() {
+#if GTEST_HAS_EXCEPTIONS
+  // This condition is always false so AlwaysTrue() never actually throws,
+  // but it makes the compiler think that it may throw.
+  if (IsTrue(false))
+    throw ClassUniqueToAlwaysTrue();
+#endif  // GTEST_HAS_EXCEPTIONS
+  return true;
+}
+
+// If *pstr starts with the given prefix, modifies *pstr to be right
+// past the prefix and returns true; otherwise leaves *pstr unchanged
+// and returns false.  None of pstr, *pstr, and prefix can be NULL.
+bool SkipPrefix(const char* prefix, const char** pstr) {
+  const size_t prefix_len = strlen(prefix);
+  if (strncmp(*pstr, prefix, prefix_len) == 0) {
+    *pstr += prefix_len;
+    return true;
+  }
+  return false;
+}
+
+// Parses a string as a command line flag.  The string should have
+// the format "--flag=value".  When def_optional is true, the "=value"
+// part can be omitted.
+//
+// Returns the value of the flag, or NULL if the parsing failed.
+const char* ParseFlagValue(const char* str,
+                           const char* flag,
+                           bool def_optional) {
+  // str and flag must not be NULL.
+  if (str == NULL || flag == NULL) return NULL;
+
+  // The flag must start with "--" followed by GTEST_FLAG_PREFIX_.
+  const std::string flag_str = std::string("--") + GTEST_FLAG_PREFIX_ + flag;
+  const size_t flag_len = flag_str.length();
+  if (strncmp(str, flag_str.c_str(), flag_len) != 0) return NULL;
+
+  // Skips the flag name.
+  const char* flag_end = str + flag_len;
+
+  // When def_optional is true, it's OK to not have a "=value" part.
+  if (def_optional && (flag_end[0] == '\0')) {
+    return flag_end;
+  }
+
+  // If def_optional is true and there are more characters after the
+  // flag name, or if def_optional is false, there must be a '=' after
+  // the flag name.
+  if (flag_end[0] != '=') return NULL;
+
+  // Returns the string after "=".
+  return flag_end + 1;
+}
+
+// Parses a string for a bool flag, in the form of either
+// "--flag=value" or "--flag".
+//
+// In the former case, the value is taken as true as long as it does
+// not start with '0', 'f', or 'F'.
+//
+// In the latter case, the value is taken as true.
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+bool ParseBoolFlag(const char* str, const char* flag, bool* value) {
+  // Gets the value of the flag as a string.
+  const char* const value_str = ParseFlagValue(str, flag, true);
+
+  // Aborts if the parsing failed.
+  if (value_str == NULL) return false;
+
+  // Converts the string value to a bool.
+  *value = !(*value_str == '0' || *value_str == 'f' || *value_str == 'F');
+  return true;
+}
+
+// Parses a string for an Int32 flag, in the form of
+// "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+bool ParseInt32Flag(const char* str, const char* flag, Int32* value) {
+  // Gets the value of the flag as a string.
+  const char* const value_str = ParseFlagValue(str, flag, false);
+
+  // Aborts if the parsing failed.
+  if (value_str == NULL) return false;
+
+  // Sets *value to the value of the flag.
+  return ParseInt32(Message() << "The value of flag --" << flag,
+                    value_str, value);
+}
+
+// Parses a string for a string flag, in the form of
+// "--flag=value".
+//
+// On success, stores the value of the flag in *value, and returns
+// true.  On failure, returns false without changing *value.
+bool ParseStringFlag(const char* str, const char* flag, std::string* value) {
+  // Gets the value of the flag as a string.
+  const char* const value_str = ParseFlagValue(str, flag, false);
+
+  // Aborts if the parsing failed.
+  if (value_str == NULL) return false;
+
+  // Sets *value to the value of the flag.
+  *value = value_str;
+  return true;
+}
+
+// Determines whether a string has a prefix that Google Test uses for its
+// flags, i.e., starts with GTEST_FLAG_PREFIX_ or GTEST_FLAG_PREFIX_DASH_.
+// If Google Test detects that a command line flag has its prefix but is not
+// recognized, it will print its help message. Flags starting with
+// GTEST_INTERNAL_PREFIX_ followed by "internal_" are considered Google Test
+// internal flags and do not trigger the help message.
+static bool HasGoogleTestFlagPrefix(const char* str) {
+  return (SkipPrefix("--", &str) ||
+          SkipPrefix("-", &str) ||
+          SkipPrefix("/", &str)) &&
+         !SkipPrefix(GTEST_FLAG_PREFIX_ "internal_", &str) &&
+         (SkipPrefix(GTEST_FLAG_PREFIX_, &str) ||
+          SkipPrefix(GTEST_FLAG_PREFIX_DASH_, &str));
+}
+
+// Prints a string containing code-encoded text.  The following escape
+// sequences can be used in the string to control the text color:
+//
+//   @@    prints a single '@' character.
+//   @R    changes the color to red.
+//   @G    changes the color to green.
+//   @Y    changes the color to yellow.
+//   @D    changes to the default terminal text color.
+//
+// TODO(wan@google.com): Write tests for this once we add stdout
+// capturing to Google Test.
+static void PrintColorEncoded(const char* str) {
+  GTestColor color = COLOR_DEFAULT;  // The current color.
+
+  // Conceptually, we split the string into segments divided by escape
+  // sequences.  Then we print one segment at a time.  At the end of
+  // each iteration, the str pointer advances to the beginning of the
+  // next segment.
+  for (;;) {
+    const char* p = strchr(str, '@');
+    if (p == NULL) {
+      ColoredPrintf(color, "%s", str);
+      return;
+    }
+
+    ColoredPrintf(color, "%s", std::string(str, p).c_str());
+
+    const char ch = p[1];
+    str = p + 2;
+    if (ch == '@') {
+      ColoredPrintf(color, "@");
+    } else if (ch == 'D') {
+      color = COLOR_DEFAULT;
+    } else if (ch == 'R') {
+      color = COLOR_RED;
+    } else if (ch == 'G') {
+      color = COLOR_GREEN;
+    } else if (ch == 'Y') {
+      color = COLOR_YELLOW;
+    } else {
+      --str;
+    }
+  }
+}
+
+static const char kColorEncodedHelpMessage[] =
+"This program contains tests written using " GTEST_NAME_ ". You can use the\n"
+"following command line flags to control its behavior:\n"
+"\n"
+"Test Selection:\n"
+"  @G--" GTEST_FLAG_PREFIX_ "list_tests@D\n"
+"      List the names of all tests instead of running them. The name of\n"
+"      TEST(Foo, Bar) is \"Foo.Bar\".\n"
+"  @G--" GTEST_FLAG_PREFIX_ "filter=@YPOSTIVE_PATTERNS"
+    "[@G-@YNEGATIVE_PATTERNS]@D\n"
+"      Run only the tests whose name matches one of the positive patterns but\n"
+"      none of the negative patterns. '?' matches any single character; '*'\n"
+"      matches any substring; ':' separates two patterns.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "also_run_disabled_tests@D\n"
+"      Run all disabled tests too.\n"
+"\n"
+"Test Execution:\n"
+"  @G--" GTEST_FLAG_PREFIX_ "repeat=@Y[COUNT]@D\n"
+"      Run the tests repeatedly; use a negative count to repeat forever.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "shuffle@D\n"
+"      Randomize tests' orders on every iteration.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "random_seed=@Y[NUMBER]@D\n"
+"      Random number seed to use for shuffling test orders (between 1 and\n"
+"      99999, or 0 to use a seed based on the current time).\n"
+"\n"
+"Test Output:\n"
+"  @G--" GTEST_FLAG_PREFIX_ "color=@Y(@Gyes@Y|@Gno@Y|@Gauto@Y)@D\n"
+"      Enable/disable colored output. The default is @Gauto@D.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "print_time=0@D\n"
+"      Don't print the elapsed time of each test.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "output=xml@Y[@G:@YDIRECTORY_PATH@G"
+    GTEST_PATH_SEP_ "@Y|@G:@YFILE_PATH]@D\n"
+"      Generate an XML report in the given directory or with the given file\n"
+"      name. @YFILE_PATH@D defaults to @Gtest_details.xml@D.\n"
+#if GTEST_CAN_STREAM_RESULTS_
+"  @G--" GTEST_FLAG_PREFIX_ "stream_result_to=@YHOST@G:@YPORT@D\n"
+"      Stream test results to the given server.\n"
+#endif  // GTEST_CAN_STREAM_RESULTS_
+"\n"
+"Assertion Behavior:\n"
+#if GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS
+"  @G--" GTEST_FLAG_PREFIX_ "death_test_style=@Y(@Gfast@Y|@Gthreadsafe@Y)@D\n"
+"      Set the default death test style.\n"
+#endif  // GTEST_HAS_DEATH_TEST && !GTEST_OS_WINDOWS
+"  @G--" GTEST_FLAG_PREFIX_ "break_on_failure@D\n"
+"      Turn assertion failures into debugger break-points.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "throw_on_failure@D\n"
+"      Turn assertion failures into C++ exceptions.\n"
+"  @G--" GTEST_FLAG_PREFIX_ "catch_exceptions=0@D\n"
+"      Do not report exceptions as test failures. Instead, allow them\n"
+"      to crash the program or throw a pop-up (on Windows).\n"
+"\n"
+"Except for @G--" GTEST_FLAG_PREFIX_ "list_tests@D, you can alternatively set "
+    "the corresponding\n"
+"environment variable of a flag (all letters in upper-case). For example, to\n"
+"disable colored text output, you can either specify @G--" GTEST_FLAG_PREFIX_
+    "color=no@D or set\n"
+"the @G" GTEST_FLAG_PREFIX_UPPER_ "COLOR@D environment variable to @Gno@D.\n"
+"\n"
+"For more information, please read the " GTEST_NAME_ " documentation at\n"
+"@G" GTEST_PROJECT_URL_ "@D. If you find a bug in " GTEST_NAME_ "\n"
+"(not one in your own code or tests), please report it to\n"
+"@G<" GTEST_DEV_EMAIL_ ">@D.\n";
+
+bool ParseGoogleTestFlag(const char* const arg) {
+  return ParseBoolFlag(arg, kAlsoRunDisabledTestsFlag,
+                       &GTEST_FLAG(also_run_disabled_tests)) ||
+      ParseBoolFlag(arg, kBreakOnFailureFlag,
+                    &GTEST_FLAG(break_on_failure)) ||
+      ParseBoolFlag(arg, kCatchExceptionsFlag,
+                    &GTEST_FLAG(catch_exceptions)) ||
+      ParseStringFlag(arg, kColorFlag, &GTEST_FLAG(color)) ||
+      ParseStringFlag(arg, kDeathTestStyleFlag,
+                      &GTEST_FLAG(death_test_style)) ||
+      ParseBoolFlag(arg, kDeathTestUseFork,
+                    &GTEST_FLAG(death_test_use_fork)) ||
+      ParseStringFlag(arg, kFilterFlag, &GTEST_FLAG(filter)) ||
+      ParseStringFlag(arg, kInternalRunDeathTestFlag,
+                      &GTEST_FLAG(internal_run_death_test)) ||
+      ParseBoolFlag(arg, kListTestsFlag, &GTEST_FLAG(list_tests)) ||
+      ParseStringFlag(arg, kOutputFlag, &GTEST_FLAG(output)) ||
+      ParseBoolFlag(arg, kPrintTimeFlag, &GTEST_FLAG(print_time)) ||
+      ParseInt32Flag(arg, kRandomSeedFlag, &GTEST_FLAG(random_seed)) ||
+      ParseInt32Flag(arg, kRepeatFlag, &GTEST_FLAG(repeat)) ||
+      ParseBoolFlag(arg, kShuffleFlag, &GTEST_FLAG(shuffle)) ||
+      ParseInt32Flag(arg, kStackTraceDepthFlag,
+                     &GTEST_FLAG(stack_trace_depth)) ||
+      ParseStringFlag(arg, kStreamResultToFlag,
+                      &GTEST_FLAG(stream_result_to)) ||
+      ParseBoolFlag(arg, kThrowOnFailureFlag,
+                    &GTEST_FLAG(throw_on_failure));
+}
+
+#if GTEST_USE_OWN_FLAGFILE_FLAG_
+void LoadFlagsFromFile(const std::string& path) {
+  FILE* flagfile = posix::FOpen(path.c_str(), "r");
+  if (!flagfile) {
+    fprintf(stderr,
+            "Unable to open file \"%s\"\n",
+            GTEST_FLAG(flagfile).c_str());
+    fflush(stderr);
+    exit(EXIT_FAILURE);
+  }
+  std::string contents(ReadEntireFile(flagfile));
+  posix::FClose(flagfile);
+  std::vector<std::string> lines;
+  SplitString(contents, '\n', &lines);
+  for (size_t i = 0; i < lines.size(); ++i) {
+    if (lines[i].empty())
+      continue;
+    if (!ParseGoogleTestFlag(lines[i].c_str()))
+      g_help_flag = true;
+  }
+}
+#endif  // GTEST_USE_OWN_FLAGFILE_FLAG_
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.  The type parameter CharType can be
+// instantiated to either char or wchar_t.
+template <typename CharType>
+void ParseGoogleTestFlagsOnlyImpl(int* argc, CharType** argv) {
+  for (int i = 1; i < *argc; i++) {
+    const std::string arg_string = StreamableToString(argv[i]);
+    const char* const arg = arg_string.c_str();
+
+    using internal::ParseBoolFlag;
+    using internal::ParseInt32Flag;
+    using internal::ParseStringFlag;
+
+    bool remove_flag = false;
+    if (ParseGoogleTestFlag(arg)) {
+      remove_flag = true;
+#if GTEST_USE_OWN_FLAGFILE_FLAG_
+    } else if (ParseStringFlag(arg, kFlagfileFlag, &GTEST_FLAG(flagfile))) {
+      LoadFlagsFromFile(GTEST_FLAG(flagfile));
+      remove_flag = true;
+#endif  // GTEST_USE_OWN_FLAGFILE_FLAG_
+    } else if (arg_string == "--help" || arg_string == "-h" ||
+               arg_string == "-?" || arg_string == "/?" ||
+               HasGoogleTestFlagPrefix(arg)) {
+      // Both help flag and unrecognized Google Test flags (excluding
+      // internal ones) trigger help display.
+      g_help_flag = true;
+    }
+
+    if (remove_flag) {
+      // Shift the remainder of the argv list left by one.  Note
+      // that argv has (*argc + 1) elements, the last one always being
+      // NULL.  The following loop moves the trailing NULL element as
+      // well.
+      for (int j = i; j != *argc; j++) {
+        argv[j] = argv[j + 1];
+      }
+
+      // Decrements the argument count.
+      (*argc)--;
+
+      // We also need to decrement the iterator as we just removed
+      // an element.
+      i--;
+    }
+  }
+
+  if (g_help_flag) {
+    // We print the help here instead of in RUN_ALL_TESTS(), as the
+    // latter may not be called at all if the user is using Google
+    // Test with another testing framework.
+    PrintColorEncoded(kColorEncodedHelpMessage);
+  }
+}
+
+// Parses the command line for Google Test flags, without initializing
+// other parts of Google Test.
+void ParseGoogleTestFlagsOnly(int* argc, char** argv) {
+  ParseGoogleTestFlagsOnlyImpl(argc, argv);
+}
+void ParseGoogleTestFlagsOnly(int* argc, wchar_t** argv) {
+  ParseGoogleTestFlagsOnlyImpl(argc, argv);
+}
+
+// The internal implementation of InitGoogleTest().
+//
+// The type parameter CharType can be instantiated to either char or
+// wchar_t.
+template <typename CharType>
+void InitGoogleTestImpl(int* argc, CharType** argv) {
+  // We don't want to run the initialization code twice.
+  if (GTestIsInitialized()) return;
+
+  if (*argc <= 0) return;
+
+  g_argvs.clear();
+  for (int i = 0; i != *argc; i++) {
+    g_argvs.push_back(StreamableToString(argv[i]));
+  }
+
+  ParseGoogleTestFlagsOnly(argc, argv);
+  GetUnitTestImpl()->PostFlagParsingInit();
+}
+
+}  // namespace internal
+
+// Initializes Google Test.  This must be called before calling
+// RUN_ALL_TESTS().  In particular, it parses a command line for the
+// flags that Google Test recognizes.  Whenever a Google Test flag is
+// seen, it is removed from argv, and *argc is decremented.
+//
+// No value is returned.  Instead, the Google Test flag variables are
+// updated.
+//
+// Calling the function for the second time has no user-visible effect.
+void InitGoogleTest(int* argc, char** argv) {
+#if defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_(argc, argv);
+#else  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  internal::InitGoogleTestImpl(argc, argv);
+#endif  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+}
+
+// This overloaded version can be used in Windows programs compiled in
+// UNICODE mode.
+void InitGoogleTest(int* argc, wchar_t** argv) {
+#if defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_(argc, argv);
+#else  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+  internal::InitGoogleTestImpl(argc, argv);
+#endif  // defined(GTEST_CUSTOM_INIT_GOOGLE_TEST_FUNCTION_)
+}
+
+std::string TempDir() {
+#if defined(GTEST_CUSTOM_TEMPDIR_FUNCTION_)
+    return GTEST_CUSTOM_TEMPDIR_FUNCTION_();
+#endif
+#if GTEST_OS_WINDOWS_MOBILE
+  return "\\temp\\";
+#elif GTEST_OS_WINDOWS
+  const char* temp_dir = internal::posix::GetEnv("TEMP");
+  if (temp_dir == NULL || temp_dir[0] == '\0')
+    return "\\temp\\";
+  else if (temp_dir[strlen(temp_dir) - 1] == '\\')
+    return temp_dir;
+  else
+    return std::string(temp_dir) + "\\";
+#elif GTEST_OS_LINUX_ANDROID
+  return "/sdcard/";
+#else
+  return "/tmp/";
+#endif  // GTEST_OS_WINDOWS_MOBILE
+}
+
+}  // namespace testing
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan), vladl@google.com (Vlad Losev)
+//
+// This file implements death tests.
+
+
+#if GTEST_HAS_DEATH_TEST
+
+# if GTEST_OS_MAC
+#  include <crt_externs.h>
+# endif  // GTEST_OS_MAC
+
+# include <errno.h>
+# include <fcntl.h>
+# include <limits.h>
+
+# if GTEST_OS_LINUX
+#  include <signal.h>
+# endif  // GTEST_OS_LINUX
+
+# include <stdarg.h>
+
+# if GTEST_OS_WINDOWS
+#  include <windows.h>
+# else
+#  include <sys/mman.h>
+#  include <sys/wait.h>
+# endif  // GTEST_OS_WINDOWS
+
+# if GTEST_OS_QNX
+#  include <spawn.h>
+# endif  // GTEST_OS_QNX
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick exists to
+// prevent the accidental inclusion of gtest-internal-inl.h in the
+// user's code.
+#define GTEST_IMPLEMENTATION_ 1
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+
+// Constants.
+
+// The default death test style.
+static const char kDefaultDeathTestStyle[] = "fast";
+
+GTEST_DEFINE_string_(
+    death_test_style,
+    internal::StringFromGTestEnv("death_test_style", kDefaultDeathTestStyle),
+    "Indicates how to run a death test in a forked child process: "
+    "\"threadsafe\" (child process re-executes the test binary "
+    "from the beginning, running only the specific death test) or "
+    "\"fast\" (child process runs the death test immediately "
+    "after forking).");
+
+GTEST_DEFINE_bool_(
+    death_test_use_fork,
+    internal::BoolFromGTestEnv("death_test_use_fork", false),
+    "Instructs to use fork()/_exit() instead of clone() in death tests. "
+    "Ignored and always uses fork() on POSIX systems where clone() is not "
+    "implemented. Useful when running under valgrind or similar tools if "
+    "those do not support clone(). Valgrind 3.3.1 will just fail if "
+    "it sees an unsupported combination of clone() flags. "
+    "It is not recommended to use this flag w/o valgrind though it will "
+    "work in 99% of the cases. Once valgrind is fixed, this flag will "
+    "most likely be removed.");
+
+namespace internal {
+GTEST_DEFINE_string_(
+    internal_run_death_test, "",
+    "Indicates the file, line number, temporal index of "
+    "the single death test to run, and a file descriptor to "
+    "which a success code may be sent, all separated by "
+    "the '|' characters.  This flag is specified if and only if the current "
+    "process is a sub-process launched for running a thread-safe "
+    "death test.  FOR INTERNAL USE ONLY.");
+}  // namespace internal
+
+#if GTEST_HAS_DEATH_TEST
+
+namespace internal {
+
+// Valid only for fast death tests. Indicates the code is running in the
+// child process of a fast style death test.
+# if !GTEST_OS_WINDOWS
+static bool g_in_fast_death_test_child = false;
+# endif
+
+// Returns a Boolean value indicating whether the caller is currently
+// executing in the context of the death test child process.  Tools such as
+// Valgrind heap checkers may need this to modify their behavior in death
+// tests.  IMPORTANT: This is an internal utility.  Using it may break the
+// implementation of death tests.  User code MUST NOT use it.
+bool InDeathTestChild() {
+# if GTEST_OS_WINDOWS
+
+  // On Windows, death tests are thread-safe regardless of the value of the
+  // death_test_style flag.
+  return !GTEST_FLAG(internal_run_death_test).empty();
+
+# else
+
+  if (GTEST_FLAG(death_test_style) == "threadsafe")
+    return !GTEST_FLAG(internal_run_death_test).empty();
+  else
+    return g_in_fast_death_test_child;
+#endif
+}
+
+}  // namespace internal
+
+// ExitedWithCode constructor.
+ExitedWithCode::ExitedWithCode(int exit_code) : exit_code_(exit_code) {
+}
+
+// ExitedWithCode function-call operator.
+bool ExitedWithCode::operator()(int exit_status) const {
+# if GTEST_OS_WINDOWS
+
+  return exit_status == exit_code_;
+
+# else
+
+  return WIFEXITED(exit_status) && WEXITSTATUS(exit_status) == exit_code_;
+
+# endif  // GTEST_OS_WINDOWS
+}
+
+# if !GTEST_OS_WINDOWS
+// KilledBySignal constructor.
+KilledBySignal::KilledBySignal(int signum) : signum_(signum) {
+}
+
+// KilledBySignal function-call operator.
+bool KilledBySignal::operator()(int exit_status) const {
+#  if defined(GTEST_KILLED_BY_SIGNAL_OVERRIDE_)
+  {
+    bool result;
+    if (GTEST_KILLED_BY_SIGNAL_OVERRIDE_(signum_, exit_status, &result)) {
+      return result;
+    }
+  }
+#  endif  // defined(GTEST_KILLED_BY_SIGNAL_OVERRIDE_)
+  return WIFSIGNALED(exit_status) && WTERMSIG(exit_status) == signum_;
+}
+# endif  // !GTEST_OS_WINDOWS
+
+namespace internal {
+
+// Utilities needed for death tests.
+
+// Generates a textual description of a given exit code, in the format
+// specified by wait(2).
+static std::string ExitSummary(int exit_code) {
+  Message m;
+
+# if GTEST_OS_WINDOWS
+
+  m << "Exited with exit status " << exit_code;
+
+# else
+
+  if (WIFEXITED(exit_code)) {
+    m << "Exited with exit status " << WEXITSTATUS(exit_code);
+  } else if (WIFSIGNALED(exit_code)) {
+    m << "Terminated by signal " << WTERMSIG(exit_code);
+  }
+#  ifdef WCOREDUMP
+  if (WCOREDUMP(exit_code)) {
+    m << " (core dumped)";
+  }
+#  endif
+# endif  // GTEST_OS_WINDOWS
+
+  return m.GetString();
+}
+
+// Returns true if exit_status describes a process that was terminated
+// by a signal, or exited normally with a nonzero exit code.
+bool ExitedUnsuccessfully(int exit_status) {
+  return !ExitedWithCode(0)(exit_status);
+}
+
+# if !GTEST_OS_WINDOWS
+// Generates a textual failure message when a death test finds more than
+// one thread running, or cannot determine the number of threads, prior
+// to executing the given statement.  It is the responsibility of the
+// caller not to pass a thread_count of 1.
+static std::string DeathTestThreadWarning(size_t thread_count) {
+  Message msg;
+  msg << "Death tests use fork(), which is unsafe particularly"
+      << " in a threaded context. For this test, " << GTEST_NAME_ << " ";
+  if (thread_count == 0)
+    msg << "couldn't detect the number of threads.";
+  else
+    msg << "detected " << thread_count << " threads.";
+  return msg.GetString();
+}
+# endif  // !GTEST_OS_WINDOWS
+
+// Flag characters for reporting a death test that did not die.
+static const char kDeathTestLived = 'L';
+static const char kDeathTestReturned = 'R';
+static const char kDeathTestThrew = 'T';
+static const char kDeathTestInternalError = 'I';
+
+// An enumeration describing all of the possible ways that a death test can
+// conclude.  DIED means that the process died while executing the test
+// code; LIVED means that process lived beyond the end of the test code;
+// RETURNED means that the test statement attempted to execute a return
+// statement, which is not allowed; THREW means that the test statement
+// returned control by throwing an exception.  IN_PROGRESS means the test
+// has not yet concluded.
+// TODO(vladl@google.com): Unify names and possibly values for
+// AbortReason, DeathTestOutcome, and flag characters above.
+enum DeathTestOutcome { IN_PROGRESS, DIED, LIVED, RETURNED, THREW };
+
+// Routine for aborting the program which is safe to call from an
+// exec-style death test child process, in which case the error
+// message is propagated back to the parent process.  Otherwise, the
+// message is simply printed to stderr.  In either case, the program
+// then exits with status 1.
+void DeathTestAbort(const std::string& message) {
+  // On a POSIX system, this function may be called from a threadsafe-style
+  // death test child process, which operates on a very small stack.  Use
+  // the heap for any additional non-minuscule memory requirements.
+  const InternalRunDeathTestFlag* const flag =
+      GetUnitTestImpl()->internal_run_death_test_flag();
+  if (flag != NULL) {
+    FILE* parent = posix::FDOpen(flag->write_fd(), "w");
+    fputc(kDeathTestInternalError, parent);
+    fprintf(parent, "%s", message.c_str());
+    fflush(parent);
+    _exit(1);
+  } else {
+    fprintf(stderr, "%s", message.c_str());
+    fflush(stderr);
+    posix::Abort();
+  }
+}
+
+// A replacement for CHECK that calls DeathTestAbort if the assertion
+// fails.
+# define GTEST_DEATH_TEST_CHECK_(expression) \
+  do { \
+    if (!::testing::internal::IsTrue(expression)) { \
+      DeathTestAbort( \
+          ::std::string("CHECK failed: File ") + __FILE__ +  ", line " \
+          + ::testing::internal::StreamableToString(__LINE__) + ": " \
+          + #expression); \
+    } \
+  } while (::testing::internal::AlwaysFalse())
+
+// This macro is similar to GTEST_DEATH_TEST_CHECK_, but it is meant for
+// evaluating any system call that fulfills two conditions: it must return
+// -1 on failure, and set errno to EINTR when it is interrupted and
+// should be tried again.  The macro expands to a loop that repeatedly
+// evaluates the expression as long as it evaluates to -1 and sets
+// errno to EINTR.  If the expression evaluates to -1 but errno is
+// something other than EINTR, DeathTestAbort is called.
+# define GTEST_DEATH_TEST_CHECK_SYSCALL_(expression) \
+  do { \
+    int gtest_retval; \
+    do { \
+      gtest_retval = (expression); \
+    } while (gtest_retval == -1 && errno == EINTR); \
+    if (gtest_retval == -1) { \
+      DeathTestAbort( \
+          ::std::string("CHECK failed: File ") + __FILE__ + ", line " \
+          + ::testing::internal::StreamableToString(__LINE__) + ": " \
+          + #expression + " != -1"); \
+    } \
+  } while (::testing::internal::AlwaysFalse())
+
+// Returns the message describing the last system error in errno.
+std::string GetLastErrnoDescription() {
+    return errno == 0 ? "" : posix::StrError(errno);
+}
+
+// This is called from a death test parent process to read a failure
+// message from the death test child process and log it with the FATAL
+// severity. On Windows, the message is read from a pipe handle. On other
+// platforms, it is read from a file descriptor.
+static void FailFromInternalError(int fd) {
+  Message error;
+  char buffer[256];
+  int num_read;
+
+  do {
+    while ((num_read = posix::Read(fd, buffer, 255)) > 0) {
+      buffer[num_read] = '\0';
+      error << buffer;
+    }
+  } while (num_read == -1 && errno == EINTR);
+
+  if (num_read == 0) {
+    GTEST_LOG_(FATAL) << error.GetString();
+  } else {
+    const int last_error = errno;
+    GTEST_LOG_(FATAL) << "Error while reading death test internal: "
+                      << GetLastErrnoDescription() << " [" << last_error << "]";
+  }
+}
+
+// Death test constructor.  Increments the running death test count
+// for the current test.
+DeathTest::DeathTest() {
+  TestInfo* const info = GetUnitTestImpl()->current_test_info();
+  if (info == NULL) {
+    DeathTestAbort("Cannot run a death test outside of a TEST or "
+                   "TEST_F construct");
+  }
+}
+
+// Creates and returns a death test by dispatching to the current
+// death test factory.
+bool DeathTest::Create(const char* statement, const RE* regex,
+                       const char* file, int line, DeathTest** test) {
+  return GetUnitTestImpl()->death_test_factory()->Create(
+      statement, regex, file, line, test);
+}
+
+const char* DeathTest::LastMessage() {
+  return last_death_test_message_.c_str();
+}
+
+void DeathTest::set_last_death_test_message(const std::string& message) {
+  last_death_test_message_ = message;
+}
+
+std::string DeathTest::last_death_test_message_;
+
+// Provides cross platform implementation for some death functionality.
+class DeathTestImpl : public DeathTest {
+ protected:
+  DeathTestImpl(const char* a_statement, const RE* a_regex)
+      : statement_(a_statement),
+        regex_(a_regex),
+        spawned_(false),
+        status_(-1),
+        outcome_(IN_PROGRESS),
+        read_fd_(-1),
+        write_fd_(-1) {}
+
+  // read_fd_ is expected to be closed and cleared by a derived class.
+  ~DeathTestImpl() { GTEST_DEATH_TEST_CHECK_(read_fd_ == -1); }
+
+  void Abort(AbortReason reason);
+  virtual bool Passed(bool status_ok);
+
+  const char* statement() const { return statement_; }
+  const RE* regex() const { return regex_; }
+  bool spawned() const { return spawned_; }
+  void set_spawned(bool is_spawned) { spawned_ = is_spawned; }
+  int status() const { return status_; }
+  void set_status(int a_status) { status_ = a_status; }
+  DeathTestOutcome outcome() const { return outcome_; }
+  void set_outcome(DeathTestOutcome an_outcome) { outcome_ = an_outcome; }
+  int read_fd() const { return read_fd_; }
+  void set_read_fd(int fd) { read_fd_ = fd; }
+  int write_fd() const { return write_fd_; }
+  void set_write_fd(int fd) { write_fd_ = fd; }
+
+  // Called in the parent process only. Reads the result code of the death
+  // test child process via a pipe, interprets it to set the outcome_
+  // member, and closes read_fd_.  Outputs diagnostics and terminates in
+  // case of unexpected codes.
+  void ReadAndInterpretStatusByte();
+
+ private:
+  // The textual content of the code this object is testing.  This class
+  // doesn't own this string and should not attempt to delete it.
+  const char* const statement_;
+  // The regular expression which test output must match.  DeathTestImpl
+  // doesn't own this object and should not attempt to delete it.
+  const RE* const regex_;
+  // True if the death test child process has been successfully spawned.
+  bool spawned_;
+  // The exit status of the child process.
+  int status_;
+  // How the death test concluded.
+  DeathTestOutcome outcome_;
+  // Descriptor to the read end of the pipe to the child process.  It is
+  // always -1 in the child process.  The child keeps its write end of the
+  // pipe in write_fd_.
+  int read_fd_;
+  // Descriptor to the child's write end of the pipe to the parent process.
+  // It is always -1 in the parent process.  The parent keeps its end of the
+  // pipe in read_fd_.
+  int write_fd_;
+};
+
+// Called in the parent process only. Reads the result code of the death
+// test child process via a pipe, interprets it to set the outcome_
+// member, and closes read_fd_.  Outputs diagnostics and terminates in
+// case of unexpected codes.
+void DeathTestImpl::ReadAndInterpretStatusByte() {
+  char flag;
+  int bytes_read;
+
+  // The read() here blocks until data is available (signifying the
+  // failure of the death test) or until the pipe is closed (signifying
+  // its success), so it's okay to call this in the parent before
+  // the child process has exited.
+  do {
+    bytes_read = posix::Read(read_fd(), &flag, 1);
+  } while (bytes_read == -1 && errno == EINTR);
+
+  if (bytes_read == 0) {
+    set_outcome(DIED);
+  } else if (bytes_read == 1) {
+    switch (flag) {
+      case kDeathTestReturned:
+        set_outcome(RETURNED);
+        break;
+      case kDeathTestThrew:
+        set_outcome(THREW);
+        break;
+      case kDeathTestLived:
+        set_outcome(LIVED);
+        break;
+      case kDeathTestInternalError:
+        FailFromInternalError(read_fd());  // Does not return.
+        break;
+      default:
+        GTEST_LOG_(FATAL) << "Death test child process reported "
+                          << "unexpected status byte ("
+                          << static_cast<unsigned int>(flag) << ")";
+    }
+  } else {
+    GTEST_LOG_(FATAL) << "Read from death test child process failed: "
+                      << GetLastErrnoDescription();
+  }
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Close(read_fd()));
+  set_read_fd(-1);
+}
+
+// Signals that the death test code which should have exited, didn't.
+// Should be called only in a death test child process.
+// Writes a status byte to the child's status file descriptor, then
+// calls _exit(1).
+void DeathTestImpl::Abort(AbortReason reason) {
+  // The parent process considers the death test to be a failure if
+  // it finds any data in our pipe.  So, here we write a single flag byte
+  // to the pipe, then exit.
+  const char status_ch =
+      reason == TEST_DID_NOT_DIE ? kDeathTestLived :
+      reason == TEST_THREW_EXCEPTION ? kDeathTestThrew : kDeathTestReturned;
+
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(posix::Write(write_fd(), &status_ch, 1));
+  // We are leaking the descriptor here because on some platforms (i.e.,
+  // when built as Windows DLL), destructors of global objects will still
+  // run after calling _exit(). On such systems, write_fd_ will be
+  // indirectly closed from the destructor of UnitTestImpl, causing double
+  // close if it is also closed here. On debug configurations, double close
+  // may assert. As there are no in-process buffers to flush here, we are
+  // relying on the OS to close the descriptor after the process terminates
+  // when the destructors are not run.
+  _exit(1);  // Exits w/o any normal exit hooks (we were supposed to crash)
+}
+
+// Returns an indented copy of stderr output for a death test.
+// This makes distinguishing death test output lines from regular log lines
+// much easier.
+static ::std::string FormatDeathTestOutput(const ::std::string& output) {
+  ::std::string ret;
+  for (size_t at = 0; ; ) {
+    const size_t line_end = output.find('\n', at);
+    ret += "[  DEATH   ] ";
+    if (line_end == ::std::string::npos) {
+      ret += output.substr(at);
+      break;
+    }
+    ret += output.substr(at, line_end + 1 - at);
+    at = line_end + 1;
+  }
+  return ret;
+}
+
+// Assesses the success or failure of a death test, using both private
+// members which have previously been set, and one argument:
+//
+// Private data members:
+//   outcome:  An enumeration describing how the death test
+//             concluded: DIED, LIVED, THREW, or RETURNED.  The death test
+//             fails in the latter three cases.
+//   status:   The exit status of the child process. On *nix, it is in the
+//             in the format specified by wait(2). On Windows, this is the
+//             value supplied to the ExitProcess() API or a numeric code
+//             of the exception that terminated the program.
+//   regex:    A regular expression object to be applied to
+//             the test's captured standard error output; the death test
+//             fails if it does not match.
+//
+// Argument:
+//   status_ok: true if exit_status is acceptable in the context of
+//              this particular death test, which fails if it is false
+//
+// Returns true iff all of the above conditions are met.  Otherwise, the
+// first failing condition, in the order given above, is the one that is
+// reported. Also sets the last death test message string.
+bool DeathTestImpl::Passed(bool status_ok) {
+  if (!spawned())
+    return false;
+
+  const std::string error_message = GetCapturedStderr();
+
+  bool success = false;
+  Message buffer;
+
+  buffer << "Death test: " << statement() << "\n";
+  switch (outcome()) {
+    case LIVED:
+      buffer << "    Result: failed to die.\n"
+             << " Error msg:\n" << FormatDeathTestOutput(error_message);
+      break;
+    case THREW:
+      buffer << "    Result: threw an exception.\n"
+             << " Error msg:\n" << FormatDeathTestOutput(error_message);
+      break;
+    case RETURNED:
+      buffer << "    Result: illegal return in test statement.\n"
+             << " Error msg:\n" << FormatDeathTestOutput(error_message);
+      break;
+    case DIED:
+      if (status_ok) {
+        const bool matched = RE::PartialMatch(error_message.c_str(), *regex());
+        if (matched) {
+          success = true;
+        } else {
+          buffer << "    Result: died but not with expected error.\n"
+                 << "  Expected: " << regex()->pattern() << "\n"
+                 << "Actual msg:\n" << FormatDeathTestOutput(error_message);
+        }
+      } else {
+        buffer << "    Result: died but not with expected exit code:\n"
+               << "            " << ExitSummary(status()) << "\n"
+               << "Actual msg:\n" << FormatDeathTestOutput(error_message);
+      }
+      break;
+    case IN_PROGRESS:
+    default:
+      GTEST_LOG_(FATAL)
+          << "DeathTest::Passed somehow called before conclusion of test";
+  }
+
+  DeathTest::set_last_death_test_message(buffer.GetString());
+  return success;
+}
+
+# if GTEST_OS_WINDOWS
+// WindowsDeathTest implements death tests on Windows. Due to the
+// specifics of starting new processes on Windows, death tests there are
+// always threadsafe, and Google Test considers the
+// --gtest_death_test_style=fast setting to be equivalent to
+// --gtest_death_test_style=threadsafe there.
+//
+// A few implementation notes:  Like the Linux version, the Windows
+// implementation uses pipes for child-to-parent communication. But due to
+// the specifics of pipes on Windows, some extra steps are required:
+//
+// 1. The parent creates a communication pipe and stores handles to both
+//    ends of it.
+// 2. The parent starts the child and provides it with the information
+//    necessary to acquire the handle to the write end of the pipe.
+// 3. The child acquires the write end of the pipe and signals the parent
+//    using a Windows event.
+// 4. Now the parent can release the write end of the pipe on its side. If
+//    this is done before step 3, the object's reference count goes down to
+//    0 and it is destroyed, preventing the child from acquiring it. The
+//    parent now has to release it, or read operations on the read end of
+//    the pipe will not return when the child terminates.
+// 5. The parent reads child's output through the pipe (outcome code and
+//    any possible error messages) from the pipe, and its stderr and then
+//    determines whether to fail the test.
+//
+// Note: to distinguish Win32 API calls from the local method and function
+// calls, the former are explicitly resolved in the global namespace.
+//
+class WindowsDeathTest : public DeathTestImpl {
+ public:
+  WindowsDeathTest(const char* a_statement,
+                   const RE* a_regex,
+                   const char* file,
+                   int line)
+      : DeathTestImpl(a_statement, a_regex), file_(file), line_(line) {}
+
+  // All of these virtual functions are inherited from DeathTest.
+  virtual int Wait();
+  virtual TestRole AssumeRole();
+
+ private:
+  // The name of the file in which the death test is located.
+  const char* const file_;
+  // The line number on which the death test is located.
+  const int line_;
+  // Handle to the write end of the pipe to the child process.
+  AutoHandle write_handle_;
+  // Child process handle.
+  AutoHandle child_handle_;
+  // Event the child process uses to signal the parent that it has
+  // acquired the handle to the write end of the pipe. After seeing this
+  // event the parent can release its own handles to make sure its
+  // ReadFile() calls return when the child terminates.
+  AutoHandle event_handle_;
+};
+
+// Waits for the child in a death test to exit, returning its exit
+// status, or 0 if no child process exists.  As a side effect, sets the
+// outcome data member.
+int WindowsDeathTest::Wait() {
+  if (!spawned())
+    return 0;
+
+  // Wait until the child either signals that it has acquired the write end
+  // of the pipe or it dies.
+  const HANDLE wait_handles[2] = { child_handle_.Get(), event_handle_.Get() };
+  switch (::WaitForMultipleObjects(2,
+                                   wait_handles,
+                                   FALSE,  // Waits for any of the handles.
+                                   INFINITE)) {
+    case WAIT_OBJECT_0:
+    case WAIT_OBJECT_0 + 1:
+      break;
+    default:
+      GTEST_DEATH_TEST_CHECK_(false);  // Should not get here.
+  }
+
+  // The child has acquired the write end of the pipe or exited.
+  // We release the handle on our side and continue.
+  write_handle_.Reset();
+  event_handle_.Reset();
+
+  ReadAndInterpretStatusByte();
+
+  // Waits for the child process to exit if it haven't already. This
+  // returns immediately if the child has already exited, regardless of
+  // whether previous calls to WaitForMultipleObjects synchronized on this
+  // handle or not.
+  GTEST_DEATH_TEST_CHECK_(
+      WAIT_OBJECT_0 == ::WaitForSingleObject(child_handle_.Get(),
+                                             INFINITE));
+  DWORD status_code;
+  GTEST_DEATH_TEST_CHECK_(
+      ::GetExitCodeProcess(child_handle_.Get(), &status_code) != FALSE);
+  child_handle_.Reset();
+  set_status(static_cast<int>(status_code));
+  return status();
+}
+
+// The AssumeRole process for a Windows death test.  It creates a child
+// process with the same executable as the current process to run the
+// death test.  The child process is given the --gtest_filter and
+// --gtest_internal_run_death_test flags such that it knows to run the
+// current death test only.
+DeathTest::TestRole WindowsDeathTest::AssumeRole() {
+  const UnitTestImpl* const impl = GetUnitTestImpl();
+  const InternalRunDeathTestFlag* const flag =
+      impl->internal_run_death_test_flag();
+  const TestInfo* const info = impl->current_test_info();
+  const int death_test_index = info->result()->death_test_count();
+
+  if (flag != NULL) {
+    // ParseInternalRunDeathTestFlag() has performed all the necessary
+    // processing.
+    set_write_fd(flag->write_fd());
+    return EXECUTE_TEST;
+  }
+
+  // WindowsDeathTest uses an anonymous pipe to communicate results of
+  // a death test.
+  SECURITY_ATTRIBUTES handles_are_inheritable = {
+    sizeof(SECURITY_ATTRIBUTES), NULL, TRUE };
+  HANDLE read_handle, write_handle;
+  GTEST_DEATH_TEST_CHECK_(
+      ::CreatePipe(&read_handle, &write_handle, &handles_are_inheritable,
+                   0)  // Default buffer size.
+      != FALSE);
+  set_read_fd(::_open_osfhandle(reinterpret_cast<intptr_t>(read_handle),
+                                O_RDONLY));
+  write_handle_.Reset(write_handle);
+  event_handle_.Reset(::CreateEvent(
+      &handles_are_inheritable,
+      TRUE,    // The event will automatically reset to non-signaled state.
+      FALSE,   // The initial state is non-signalled.
+      NULL));  // The even is unnamed.
+  GTEST_DEATH_TEST_CHECK_(event_handle_.Get() != NULL);
+  const std::string filter_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kFilterFlag + "=" +
+      info->test_case_name() + "." + info->name();
+  const std::string internal_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag +
+      "=" + file_ + "|" + StreamableToString(line_) + "|" +
+      StreamableToString(death_test_index) + "|" +
+      StreamableToString(static_cast<unsigned int>(::GetCurrentProcessId())) +
+      // size_t has the same width as pointers on both 32-bit and 64-bit
+      // Windows platforms.
+      // See http://msdn.microsoft.com/en-us/library/tcxf1dw6.aspx.
+      "|" + StreamableToString(reinterpret_cast<size_t>(write_handle)) +
+      "|" + StreamableToString(reinterpret_cast<size_t>(event_handle_.Get()));
+
+  char executable_path[_MAX_PATH + 1];  // NOLINT
+  GTEST_DEATH_TEST_CHECK_(
+      _MAX_PATH + 1 != ::GetModuleFileNameA(NULL,
+                                            executable_path,
+                                            _MAX_PATH));
+
+  std::string command_line =
+      std::string(::GetCommandLineA()) + " " + filter_flag + " \"" +
+      internal_flag + "\"";
+
+  DeathTest::set_last_death_test_message("");
+
+  CaptureStderr();
+  // Flush the log buffers since the log streams are shared with the child.
+  FlushInfoLog();
+
+  // The child process will share the standard handles with the parent.
+  STARTUPINFOA startup_info;
+  memset(&startup_info, 0, sizeof(STARTUPINFO));
+  startup_info.dwFlags = STARTF_USESTDHANDLES;
+  startup_info.hStdInput = ::GetStdHandle(STD_INPUT_HANDLE);
+  startup_info.hStdOutput = ::GetStdHandle(STD_OUTPUT_HANDLE);
+  startup_info.hStdError = ::GetStdHandle(STD_ERROR_HANDLE);
+
+  PROCESS_INFORMATION process_info;
+  GTEST_DEATH_TEST_CHECK_(::CreateProcessA(
+      executable_path,
+      const_cast<char*>(command_line.c_str()),
+      NULL,   // Returned process handle is not inheritable.
+      NULL,   // Returned thread handle is not inheritable.
+      TRUE,   // Child inherits all inheritable handles (for write_handle_).
+      0x0,    // Default creation flags.
+      NULL,   // Inherit the parent's environment.
+      UnitTest::GetInstance()->original_working_dir(),
+      &startup_info,
+      &process_info) != FALSE);
+  child_handle_.Reset(process_info.hProcess);
+  ::CloseHandle(process_info.hThread);
+  set_spawned(true);
+  return OVERSEE_TEST;
+}
+# else  // We are not on Windows.
+
+// ForkingDeathTest provides implementations for most of the abstract
+// methods of the DeathTest interface.  Only the AssumeRole method is
+// left undefined.
+class ForkingDeathTest : public DeathTestImpl {
+ public:
+  ForkingDeathTest(const char* statement, const RE* regex);
+
+  // All of these virtual functions are inherited from DeathTest.
+  virtual int Wait();
+
+ protected:
+  void set_child_pid(pid_t child_pid) { child_pid_ = child_pid; }
+
+ private:
+  // PID of child process during death test; 0 in the child process itself.
+  pid_t child_pid_;
+};
+
+// Constructs a ForkingDeathTest.
+ForkingDeathTest::ForkingDeathTest(const char* a_statement, const RE* a_regex)
+    : DeathTestImpl(a_statement, a_regex),
+      child_pid_(-1) {}
+
+// Waits for the child in a death test to exit, returning its exit
+// status, or 0 if no child process exists.  As a side effect, sets the
+// outcome data member.
+int ForkingDeathTest::Wait() {
+  if (!spawned())
+    return 0;
+
+  ReadAndInterpretStatusByte();
+
+  int status_value;
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(waitpid(child_pid_, &status_value, 0));
+  set_status(status_value);
+  return status_value;
+}
+
+// A concrete death test class that forks, then immediately runs the test
+// in the child process.
+class NoExecDeathTest : public ForkingDeathTest {
+ public:
+  NoExecDeathTest(const char* a_statement, const RE* a_regex) :
+      ForkingDeathTest(a_statement, a_regex) { }
+  virtual TestRole AssumeRole();
+};
+
+// The AssumeRole process for a fork-and-run death test.  It implements a
+// straightforward fork, with a simple pipe to transmit the status byte.
+DeathTest::TestRole NoExecDeathTest::AssumeRole() {
+  const size_t thread_count = GetThreadCount();
+  if (thread_count != 1) {
+    GTEST_LOG_(WARNING) << DeathTestThreadWarning(thread_count);
+  }
+
+  int pipe_fd[2];
+  GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1);
+
+  DeathTest::set_last_death_test_message("");
+  CaptureStderr();
+  // When we fork the process below, the log file buffers are copied, but the
+  // file descriptors are shared.  We flush all log files here so that closing
+  // the file descriptors in the child process doesn't throw off the
+  // synchronization between descriptors and buffers in the parent process.
+  // This is as close to the fork as possible to avoid a race condition in case
+  // there are multiple threads running before the death test, and another
+  // thread writes to the log file.
+  FlushInfoLog();
+
+  const pid_t child_pid = fork();
+  GTEST_DEATH_TEST_CHECK_(child_pid != -1);
+  set_child_pid(child_pid);
+  if (child_pid == 0) {
+    GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[0]));
+    set_write_fd(pipe_fd[1]);
+    // Redirects all logging to stderr in the child process to prevent
+    // concurrent writes to the log files.  We capture stderr in the parent
+    // process and append the child process' output to a log.
+    LogToStderr();
+    // Event forwarding to the listeners of event listener API mush be shut
+    // down in death test subprocesses.
+    GetUnitTestImpl()->listeners()->SuppressEventForwarding();
+    g_in_fast_death_test_child = true;
+    return EXECUTE_TEST;
+  } else {
+    GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1]));
+    set_read_fd(pipe_fd[0]);
+    set_spawned(true);
+    return OVERSEE_TEST;
+  }
+}
+
+// A concrete death test class that forks and re-executes the main
+// program from the beginning, with command-line flags set that cause
+// only this specific death test to be run.
+class ExecDeathTest : public ForkingDeathTest {
+ public:
+  ExecDeathTest(const char* a_statement, const RE* a_regex,
+                const char* file, int line) :
+      ForkingDeathTest(a_statement, a_regex), file_(file), line_(line) { }
+  virtual TestRole AssumeRole();
+ private:
+  static ::std::vector<std::string> GetArgvsForDeathTestChildProcess() {
+    ::std::vector<std::string> args = GetInjectableArgvs();
+#  if defined(GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_)
+    ::std::vector<std::string> extra_args =
+        GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_();
+    args.insert(args.end(), extra_args.begin(), extra_args.end());
+#  endif  // defined(GTEST_EXTRA_DEATH_TEST_COMMAND_LINE_ARGS_)
+    return args;
+  }
+  // The name of the file in which the death test is located.
+  const char* const file_;
+  // The line number on which the death test is located.
+  const int line_;
+};
+
+// Utility class for accumulating command-line arguments.
+class Arguments {
+ public:
+  Arguments() {
+    args_.push_back(NULL);
+  }
+
+  ~Arguments() {
+    for (std::vector<char*>::iterator i = args_.begin(); i != args_.end();
+         ++i) {
+      free(*i);
+    }
+  }
+  void AddArgument(const char* argument) {
+    args_.insert(args_.end() - 1, posix::StrDup(argument));
+  }
+
+  template <typename Str>
+  void AddArguments(const ::std::vector<Str>& arguments) {
+    for (typename ::std::vector<Str>::const_iterator i = arguments.begin();
+         i != arguments.end();
+         ++i) {
+      args_.insert(args_.end() - 1, posix::StrDup(i->c_str()));
+    }
+  }
+  char* const* Argv() {
+    return &args_[0];
+  }
+
+ private:
+  std::vector<char*> args_;
+};
+
+// A struct that encompasses the arguments to the child process of a
+// threadsafe-style death test process.
+struct ExecDeathTestArgs {
+  char* const* argv;  // Command-line arguments for the child's call to exec
+  int close_fd;       // File descriptor to close; the read end of a pipe
+};
+
+#  if GTEST_OS_MAC
+inline char** GetEnviron() {
+  // When Google Test is built as a framework on MacOS X, the environ variable
+  // is unavailable. Apple's documentation (man environ) recommends using
+  // _NSGetEnviron() instead.
+  return *_NSGetEnviron();
+}
+#  else
+// Some POSIX platforms expect you to declare environ. extern "C" makes
+// it reside in the global namespace.
+extern "C" char** environ;
+inline char** GetEnviron() { return environ; }
+#  endif  // GTEST_OS_MAC
+
+#  if !GTEST_OS_QNX
+// The main function for a threadsafe-style death test child process.
+// This function is called in a clone()-ed process and thus must avoid
+// any potentially unsafe operations like malloc or libc functions.
+static int ExecDeathTestChildMain(void* child_arg) {
+  ExecDeathTestArgs* const args = static_cast<ExecDeathTestArgs*>(child_arg);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(close(args->close_fd));
+
+  // We need to execute the test program in the same environment where
+  // it was originally invoked.  Therefore we change to the original
+  // working directory first.
+  const char* const original_dir =
+      UnitTest::GetInstance()->original_working_dir();
+  // We can safely call chdir() as it's a direct system call.
+  if (chdir(original_dir) != 0) {
+    DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " +
+                   GetLastErrnoDescription());
+    return EXIT_FAILURE;
+  }
+
+  // We can safely call execve() as it's a direct system call.  We
+  // cannot use execvp() as it's a libc function and thus potentially
+  // unsafe.  Since execve() doesn't search the PATH, the user must
+  // invoke the test program via a valid path that contains at least
+  // one path separator.
+  execve(args->argv[0], args->argv, GetEnviron());
+  DeathTestAbort(std::string("execve(") + args->argv[0] + ", ...) in " +
+                 original_dir + " failed: " +
+                 GetLastErrnoDescription());
+  return EXIT_FAILURE;
+}
+#  endif  // !GTEST_OS_QNX
+
+// Two utility routines that together determine the direction the stack
+// grows.
+// This could be accomplished more elegantly by a single recursive
+// function, but we want to guard against the unlikely possibility of
+// a smart compiler optimizing the recursion away.
+//
+// GTEST_NO_INLINE_ is required to prevent GCC 4.6 from inlining
+// StackLowerThanAddress into StackGrowsDown, which then doesn't give
+// correct answer.
+void StackLowerThanAddress(const void* ptr, bool* result) GTEST_NO_INLINE_;
+void StackLowerThanAddress(const void* ptr, bool* result) {
+  int dummy;
+  *result = (&dummy < ptr);
+}
+
+// Make sure AddressSanitizer does not tamper with the stack here.
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+bool StackGrowsDown() {
+  int dummy;
+  bool result;
+  StackLowerThanAddress(&dummy, &result);
+  return result;
+}
+
+// Spawns a child process with the same executable as the current process in
+// a thread-safe manner and instructs it to run the death test.  The
+// implementation uses fork(2) + exec.  On systems where clone(2) is
+// available, it is used instead, being slightly more thread-safe.  On QNX,
+// fork supports only single-threaded environments, so this function uses
+// spawn(2) there instead.  The function dies with an error message if
+// anything goes wrong.
+static pid_t ExecDeathTestSpawnChild(char* const* argv, int close_fd) {
+  ExecDeathTestArgs args = { argv, close_fd };
+  pid_t child_pid = -1;
+
+#  if GTEST_OS_QNX
+  // Obtains the current directory and sets it to be closed in the child
+  // process.
+  const int cwd_fd = open(".", O_RDONLY);
+  GTEST_DEATH_TEST_CHECK_(cwd_fd != -1);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(cwd_fd, F_SETFD, FD_CLOEXEC));
+  // We need to execute the test program in the same environment where
+  // it was originally invoked.  Therefore we change to the original
+  // working directory first.
+  const char* const original_dir =
+      UnitTest::GetInstance()->original_working_dir();
+  // We can safely call chdir() as it's a direct system call.
+  if (chdir(original_dir) != 0) {
+    DeathTestAbort(std::string("chdir(\"") + original_dir + "\") failed: " +
+                   GetLastErrnoDescription());
+    return EXIT_FAILURE;
+  }
+
+  int fd_flags;
+  // Set close_fd to be closed after spawn.
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(fd_flags = fcntl(close_fd, F_GETFD));
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(fcntl(close_fd, F_SETFD,
+                                        fd_flags | FD_CLOEXEC));
+  struct inheritance inherit = {0};
+  // spawn is a system call.
+  child_pid = spawn(args.argv[0], 0, NULL, &inherit, args.argv, GetEnviron());
+  // Restores the current working directory.
+  GTEST_DEATH_TEST_CHECK_(fchdir(cwd_fd) != -1);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(close(cwd_fd));
+
+#  else   // GTEST_OS_QNX
+#   if GTEST_OS_LINUX
+  // When a SIGPROF signal is received while fork() or clone() are executing,
+  // the process may hang. To avoid this, we ignore SIGPROF here and re-enable
+  // it after the call to fork()/clone() is complete.
+  struct sigaction saved_sigprof_action;
+  struct sigaction ignore_sigprof_action;
+  memset(&ignore_sigprof_action, 0, sizeof(ignore_sigprof_action));
+  sigemptyset(&ignore_sigprof_action.sa_mask);
+  ignore_sigprof_action.sa_handler = SIG_IGN;
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(sigaction(
+      SIGPROF, &ignore_sigprof_action, &saved_sigprof_action));
+#   endif  // GTEST_OS_LINUX
+
+#   if GTEST_HAS_CLONE
+  const bool use_fork = GTEST_FLAG(death_test_use_fork);
+
+  if (!use_fork) {
+    static const bool stack_grows_down = StackGrowsDown();
+    const size_t stack_size = getpagesize();
+    // MMAP_ANONYMOUS is not defined on Mac, so we use MAP_ANON instead.
+    void* const stack = mmap(NULL, stack_size, PROT_READ | PROT_WRITE,
+                             MAP_ANON | MAP_PRIVATE, -1, 0);
+    GTEST_DEATH_TEST_CHECK_(stack != MAP_FAILED);
+
+    // Maximum stack alignment in bytes:  For a downward-growing stack, this
+    // amount is subtracted from size of the stack space to get an address
+    // that is within the stack space and is aligned on all systems we care
+    // about.  As far as I know there is no ABI with stack alignment greater
+    // than 64.  We assume stack and stack_size already have alignment of
+    // kMaxStackAlignment.
+    const size_t kMaxStackAlignment = 64;
+    void* const stack_top =
+        static_cast<char*>(stack) +
+            (stack_grows_down ? stack_size - kMaxStackAlignment : 0);
+    GTEST_DEATH_TEST_CHECK_(stack_size > kMaxStackAlignment &&
+        reinterpret_cast<intptr_t>(stack_top) % kMaxStackAlignment == 0);
+
+    child_pid = clone(&ExecDeathTestChildMain, stack_top, SIGCHLD, &args);
+
+    GTEST_DEATH_TEST_CHECK_(munmap(stack, stack_size) != -1);
+  }
+#   else
+  const bool use_fork = true;
+#   endif  // GTEST_HAS_CLONE
+
+  if (use_fork && (child_pid = fork()) == 0) {
+      ExecDeathTestChildMain(&args);
+      _exit(0);
+  }
+#  endif  // GTEST_OS_QNX
+#  if GTEST_OS_LINUX
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(
+      sigaction(SIGPROF, &saved_sigprof_action, NULL));
+#  endif  // GTEST_OS_LINUX
+
+  GTEST_DEATH_TEST_CHECK_(child_pid != -1);
+  return child_pid;
+}
+
+// The AssumeRole process for a fork-and-exec death test.  It re-executes the
+// main program from the beginning, setting the --gtest_filter
+// and --gtest_internal_run_death_test flags to cause only the current
+// death test to be re-run.
+DeathTest::TestRole ExecDeathTest::AssumeRole() {
+  const UnitTestImpl* const impl = GetUnitTestImpl();
+  const InternalRunDeathTestFlag* const flag =
+      impl->internal_run_death_test_flag();
+  const TestInfo* const info = impl->current_test_info();
+  const int death_test_index = info->result()->death_test_count();
+
+  if (flag != NULL) {
+    set_write_fd(flag->write_fd());
+    return EXECUTE_TEST;
+  }
+
+  int pipe_fd[2];
+  GTEST_DEATH_TEST_CHECK_(pipe(pipe_fd) != -1);
+  // Clear the close-on-exec flag on the write end of the pipe, lest
+  // it be closed when the child process does an exec:
+  GTEST_DEATH_TEST_CHECK_(fcntl(pipe_fd[1], F_SETFD, 0) != -1);
+
+  const std::string filter_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kFilterFlag + "="
+      + info->test_case_name() + "." + info->name();
+  const std::string internal_flag =
+      std::string("--") + GTEST_FLAG_PREFIX_ + kInternalRunDeathTestFlag + "="
+      + file_ + "|" + StreamableToString(line_) + "|"
+      + StreamableToString(death_test_index) + "|"
+      + StreamableToString(pipe_fd[1]);
+  Arguments args;
+  args.AddArguments(GetArgvsForDeathTestChildProcess());
+  args.AddArgument(filter_flag.c_str());
+  args.AddArgument(internal_flag.c_str());
+
+  DeathTest::set_last_death_test_message("");
+
+  CaptureStderr();
+  // See the comment in NoExecDeathTest::AssumeRole for why the next line
+  // is necessary.
+  FlushInfoLog();
+
+  const pid_t child_pid = ExecDeathTestSpawnChild(args.Argv(), pipe_fd[0]);
+  GTEST_DEATH_TEST_CHECK_SYSCALL_(close(pipe_fd[1]));
+  set_child_pid(child_pid);
+  set_read_fd(pipe_fd[0]);
+  set_spawned(true);
+  return OVERSEE_TEST;
+}
+
+# endif  // !GTEST_OS_WINDOWS
+
+// Creates a concrete DeathTest-derived class that depends on the
+// --gtest_death_test_style flag, and sets the pointer pointed to
+// by the "test" argument to its address.  If the test should be
+// skipped, sets that pointer to NULL.  Returns true, unless the
+// flag is set to an invalid value.
+bool DefaultDeathTestFactory::Create(const char* statement, const RE* regex,
+                                     const char* file, int line,
+                                     DeathTest** test) {
+  UnitTestImpl* const impl = GetUnitTestImpl();
+  const InternalRunDeathTestFlag* const flag =
+      impl->internal_run_death_test_flag();
+  const int death_test_index = impl->current_test_info()
+      ->increment_death_test_count();
+
+  if (flag != NULL) {
+    if (death_test_index > flag->index()) {
+      DeathTest::set_last_death_test_message(
+          "Death test count (" + StreamableToString(death_test_index)
+          + ") somehow exceeded expected maximum ("
+          + StreamableToString(flag->index()) + ")");
+      return false;
+    }
+
+    if (!(flag->file() == file && flag->line() == line &&
+          flag->index() == death_test_index)) {
+      *test = NULL;
+      return true;
+    }
+  }
+
+# if GTEST_OS_WINDOWS
+
+  if (GTEST_FLAG(death_test_style) == "threadsafe" ||
+      GTEST_FLAG(death_test_style) == "fast") {
+    *test = new WindowsDeathTest(statement, regex, file, line);
+  }
+
+# else
+
+  if (GTEST_FLAG(death_test_style) == "threadsafe") {
+    *test = new ExecDeathTest(statement, regex, file, line);
+  } else if (GTEST_FLAG(death_test_style) == "fast") {
+    *test = new NoExecDeathTest(statement, regex);
+  }
+
+# endif  // GTEST_OS_WINDOWS
+
+  else {  // NOLINT - this is more readable than unbalanced brackets inside #if.
+    DeathTest::set_last_death_test_message(
+        "Unknown death test style \"" + GTEST_FLAG(death_test_style)
+        + "\" encountered");
+    return false;
+  }
+
+  return true;
+}
+
+# if GTEST_OS_WINDOWS
+// Recreates the pipe and event handles from the provided parameters,
+// signals the event, and returns a file descriptor wrapped around the pipe
+// handle. This function is called in the child process only.
+int GetStatusFileDescriptor(unsigned int parent_process_id,
+                            size_t write_handle_as_size_t,
+                            size_t event_handle_as_size_t) {
+  AutoHandle parent_process_handle(::OpenProcess(PROCESS_DUP_HANDLE,
+                                                   FALSE,  // Non-inheritable.
+                                                   parent_process_id));
+  if (parent_process_handle.Get() == INVALID_HANDLE_VALUE) {
+    DeathTestAbort("Unable to open parent process " +
+                   StreamableToString(parent_process_id));
+  }
+
+  // TODO(vladl@google.com): Replace the following check with a
+  // compile-time assertion when available.
+  GTEST_CHECK_(sizeof(HANDLE) <= sizeof(size_t));
+
+  const HANDLE write_handle =
+      reinterpret_cast<HANDLE>(write_handle_as_size_t);
+  HANDLE dup_write_handle;
+
+  // The newly initialized handle is accessible only in in the parent
+  // process. To obtain one accessible within the child, we need to use
+  // DuplicateHandle.
+  if (!::DuplicateHandle(parent_process_handle.Get(), write_handle,
+                         ::GetCurrentProcess(), &dup_write_handle,
+                         0x0,    // Requested privileges ignored since
+                                 // DUPLICATE_SAME_ACCESS is used.
+                         FALSE,  // Request non-inheritable handler.
+                         DUPLICATE_SAME_ACCESS)) {
+    DeathTestAbort("Unable to duplicate the pipe handle " +
+                   StreamableToString(write_handle_as_size_t) +
+                   " from the parent process " +
+                   StreamableToString(parent_process_id));
+  }
+
+  const HANDLE event_handle = reinterpret_cast<HANDLE>(event_handle_as_size_t);
+  HANDLE dup_event_handle;
+
+  if (!::DuplicateHandle(parent_process_handle.Get(), event_handle,
+                         ::GetCurrentProcess(), &dup_event_handle,
+                         0x0,
+                         FALSE,
+                         DUPLICATE_SAME_ACCESS)) {
+    DeathTestAbort("Unable to duplicate the event handle " +
+                   StreamableToString(event_handle_as_size_t) +
+                   " from the parent process " +
+                   StreamableToString(parent_process_id));
+  }
+
+  const int write_fd =
+      ::_open_osfhandle(reinterpret_cast<intptr_t>(dup_write_handle), O_APPEND);
+  if (write_fd == -1) {
+    DeathTestAbort("Unable to convert pipe handle " +
+                   StreamableToString(write_handle_as_size_t) +
+                   " to a file descriptor");
+  }
+
+  // Signals the parent that the write end of the pipe has been acquired
+  // so the parent can release its own write end.
+  ::SetEvent(dup_event_handle);
+
+  return write_fd;
+}
+# endif  // GTEST_OS_WINDOWS
+
+// Returns a newly created InternalRunDeathTestFlag object with fields
+// initialized from the GTEST_FLAG(internal_run_death_test) flag if
+// the flag is specified; otherwise returns NULL.
+InternalRunDeathTestFlag* ParseInternalRunDeathTestFlag() {
+  if (GTEST_FLAG(internal_run_death_test) == "") return NULL;
+
+  // GTEST_HAS_DEATH_TEST implies that we have ::std::string, so we
+  // can use it here.
+  int line = -1;
+  int index = -1;
+  ::std::vector< ::std::string> fields;
+  SplitString(GTEST_FLAG(internal_run_death_test).c_str(), '|', &fields);
+  int write_fd = -1;
+
+# if GTEST_OS_WINDOWS
+
+  unsigned int parent_process_id = 0;
+  size_t write_handle_as_size_t = 0;
+  size_t event_handle_as_size_t = 0;
+
+  if (fields.size() != 6
+      || !ParseNaturalNumber(fields[1], &line)
+      || !ParseNaturalNumber(fields[2], &index)
+      || !ParseNaturalNumber(fields[3], &parent_process_id)
+      || !ParseNaturalNumber(fields[4], &write_handle_as_size_t)
+      || !ParseNaturalNumber(fields[5], &event_handle_as_size_t)) {
+    DeathTestAbort("Bad --gtest_internal_run_death_test flag: " +
+                   GTEST_FLAG(internal_run_death_test));
+  }
+  write_fd = GetStatusFileDescriptor(parent_process_id,
+                                     write_handle_as_size_t,
+                                     event_handle_as_size_t);
+# else
+
+  if (fields.size() != 4
+      || !ParseNaturalNumber(fields[1], &line)
+      || !ParseNaturalNumber(fields[2], &index)
+      || !ParseNaturalNumber(fields[3], &write_fd)) {
+    DeathTestAbort("Bad --gtest_internal_run_death_test flag: "
+        + GTEST_FLAG(internal_run_death_test));
+  }
+
+# endif  // GTEST_OS_WINDOWS
+
+  return new InternalRunDeathTestFlag(fields[0], line, index, write_fd);
+}
+
+}  // namespace internal
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+}  // namespace testing
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: keith.ray@gmail.com (Keith Ray)
+
+
+#include <stdlib.h>
+
+#if GTEST_OS_WINDOWS_MOBILE
+# include <windows.h>
+#elif GTEST_OS_WINDOWS
+# include <direct.h>
+# include <io.h>
+#elif GTEST_OS_SYMBIAN
+// Symbian OpenC has PATH_MAX in sys/syslimits.h
+# include <sys/syslimits.h>
+#else
+# include <limits.h>
+# include <climits>  // Some Linux distributions define PATH_MAX here.
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+#if GTEST_OS_WINDOWS
+# define GTEST_PATH_MAX_ _MAX_PATH
+#elif defined(PATH_MAX)
+# define GTEST_PATH_MAX_ PATH_MAX
+#elif defined(_XOPEN_PATH_MAX)
+# define GTEST_PATH_MAX_ _XOPEN_PATH_MAX
+#else
+# define GTEST_PATH_MAX_ _POSIX_PATH_MAX
+#endif  // GTEST_OS_WINDOWS
+
+
+namespace testing {
+namespace internal {
+
+#if GTEST_OS_WINDOWS
+// On Windows, '\\' is the standard path separator, but many tools and the
+// Windows API also accept '/' as an alternate path separator. Unless otherwise
+// noted, a file path can contain either kind of path separators, or a mixture
+// of them.
+const char kPathSeparator = '\\';
+const char kAlternatePathSeparator = '/';
+const char kAlternatePathSeparatorString[] = "/";
+# if GTEST_OS_WINDOWS_MOBILE
+// Windows CE doesn't have a current directory. You should not use
+// the current directory in tests on Windows CE, but this at least
+// provides a reasonable fallback.
+const char kCurrentDirectoryString[] = "\\";
+// Windows CE doesn't define INVALID_FILE_ATTRIBUTES
+const DWORD kInvalidFileAttributes = 0xffffffff;
+# else
+const char kCurrentDirectoryString[] = ".\\";
+# endif  // GTEST_OS_WINDOWS_MOBILE
+#else
+const char kPathSeparator = '/';
+const char kCurrentDirectoryString[] = "./";
+#endif  // GTEST_OS_WINDOWS
+
+// Returns whether the given character is a valid path separator.
+static bool IsPathSeparator(char c) {
+#if GTEST_HAS_ALT_PATH_SEP_
+  return (c == kPathSeparator) || (c == kAlternatePathSeparator);
+#else
+  return c == kPathSeparator;
+#endif
+}
+
+// Returns the current working directory, or "" if unsuccessful.
+FilePath FilePath::GetCurrentDir() {
+#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT
+  // Windows CE doesn't have a current directory, so we just return
+  // something reasonable.
+  return FilePath(kCurrentDirectoryString);
+#elif GTEST_OS_WINDOWS
+  char cwd[GTEST_PATH_MAX_ + 1] = { '\0' };
+  return FilePath(_getcwd(cwd, sizeof(cwd)) == NULL ? "" : cwd);
+#else
+  char cwd[GTEST_PATH_MAX_ + 1] = { '\0' };
+  char* result = getcwd(cwd, sizeof(cwd));
+# if GTEST_OS_NACL
+  // getcwd will likely fail in NaCl due to the sandbox, so return something
+  // reasonable. The user may have provided a shim implementation for getcwd,
+  // however, so fallback only when failure is detected.
+  return FilePath(result == NULL ? kCurrentDirectoryString : cwd);
+# endif  // GTEST_OS_NACL
+  return FilePath(result == NULL ? "" : cwd);
+#endif  // GTEST_OS_WINDOWS_MOBILE
+}
+
+// Returns a copy of the FilePath with the case-insensitive extension removed.
+// Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns
+// FilePath("dir/file"). If a case-insensitive extension is not
+// found, returns a copy of the original FilePath.
+FilePath FilePath::RemoveExtension(const char* extension) const {
+  const std::string dot_extension = std::string(".") + extension;
+  if (String::EndsWithCaseInsensitive(pathname_, dot_extension)) {
+    return FilePath(pathname_.substr(
+        0, pathname_.length() - dot_extension.length()));
+  }
+  return *this;
+}
+
+// Returns a pointer to the last occurrence of a valid path separator in
+// the FilePath. On Windows, for example, both '/' and '\' are valid path
+// separators. Returns NULL if no path separator was found.
+const char* FilePath::FindLastPathSeparator() const {
+  const char* const last_sep = strrchr(c_str(), kPathSeparator);
+#if GTEST_HAS_ALT_PATH_SEP_
+  const char* const last_alt_sep = strrchr(c_str(), kAlternatePathSeparator);
+  // Comparing two pointers of which only one is NULL is undefined.
+  if (last_alt_sep != NULL &&
+      (last_sep == NULL || last_alt_sep > last_sep)) {
+    return last_alt_sep;
+  }
+#endif
+  return last_sep;
+}
+
+// Returns a copy of the FilePath with the directory part removed.
+// Example: FilePath("path/to/file").RemoveDirectoryName() returns
+// FilePath("file"). If there is no directory part ("just_a_file"), it returns
+// the FilePath unmodified. If there is no file part ("just_a_dir/") it
+// returns an empty FilePath ("").
+// On Windows platform, '\' is the path separator, otherwise it is '/'.
+FilePath FilePath::RemoveDirectoryName() const {
+  const char* const last_sep = FindLastPathSeparator();
+  return last_sep ? FilePath(last_sep + 1) : *this;
+}
+
+// RemoveFileName returns the directory path with the filename removed.
+// Example: FilePath("path/to/file").RemoveFileName() returns "path/to/".
+// If the FilePath is "a_file" or "/a_file", RemoveFileName returns
+// FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does
+// not have a file, like "just/a/dir/", it returns the FilePath unmodified.
+// On Windows platform, '\' is the path separator, otherwise it is '/'.
+FilePath FilePath::RemoveFileName() const {
+  const char* const last_sep = FindLastPathSeparator();
+  std::string dir;
+  if (last_sep) {
+    dir = std::string(c_str(), last_sep + 1 - c_str());
+  } else {
+    dir = kCurrentDirectoryString;
+  }
+  return FilePath(dir);
+}
+
+// Helper functions for naming files in a directory for xml output.
+
+// Given directory = "dir", base_name = "test", number = 0,
+// extension = "xml", returns "dir/test.xml". If number is greater
+// than zero (e.g., 12), returns "dir/test_12.xml".
+// On Windows platform, uses \ as the separator rather than /.
+FilePath FilePath::MakeFileName(const FilePath& directory,
+                                const FilePath& base_name,
+                                int number,
+                                const char* extension) {
+  std::string file;
+  if (number == 0) {
+    file = base_name.string() + "." + extension;
+  } else {
+    file = base_name.string() + "_" + StreamableToString(number)
+        + "." + extension;
+  }
+  return ConcatPaths(directory, FilePath(file));
+}
+
+// Given directory = "dir", relative_path = "test.xml", returns "dir/test.xml".
+// On Windows, uses \ as the separator rather than /.
+FilePath FilePath::ConcatPaths(const FilePath& directory,
+                               const FilePath& relative_path) {
+  if (directory.IsEmpty())
+    return relative_path;
+  const FilePath dir(directory.RemoveTrailingPathSeparator());
+  return FilePath(dir.string() + kPathSeparator + relative_path.string());
+}
+
+// Returns true if pathname describes something findable in the file-system,
+// either a file, directory, or whatever.
+bool FilePath::FileOrDirectoryExists() const {
+#if GTEST_OS_WINDOWS_MOBILE
+  LPCWSTR unicode = String::AnsiToUtf16(pathname_.c_str());
+  const DWORD attributes = GetFileAttributes(unicode);
+  delete [] unicode;
+  return attributes != kInvalidFileAttributes;
+#else
+  posix::StatStruct file_stat;
+  return posix::Stat(pathname_.c_str(), &file_stat) == 0;
+#endif  // GTEST_OS_WINDOWS_MOBILE
+}
+
+// Returns true if pathname describes a directory in the file-system
+// that exists.
+bool FilePath::DirectoryExists() const {
+  bool result = false;
+#if GTEST_OS_WINDOWS
+  // Don't strip off trailing separator if path is a root directory on
+  // Windows (like "C:\\").
+  const FilePath& path(IsRootDirectory() ? *this :
+                                           RemoveTrailingPathSeparator());
+#else
+  const FilePath& path(*this);
+#endif
+
+#if GTEST_OS_WINDOWS_MOBILE
+  LPCWSTR unicode = String::AnsiToUtf16(path.c_str());
+  const DWORD attributes = GetFileAttributes(unicode);
+  delete [] unicode;
+  if ((attributes != kInvalidFileAttributes) &&
+      (attributes & FILE_ATTRIBUTE_DIRECTORY)) {
+    result = true;
+  }
+#else
+  posix::StatStruct file_stat;
+  result = posix::Stat(path.c_str(), &file_stat) == 0 &&
+      posix::IsDir(file_stat);
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+  return result;
+}
+
+// Returns true if pathname describes a root directory. (Windows has one
+// root directory per disk drive.)
+bool FilePath::IsRootDirectory() const {
+#if GTEST_OS_WINDOWS
+  // TODO(wan@google.com): on Windows a network share like
+  // \\server\share can be a root directory, although it cannot be the
+  // current directory.  Handle this properly.
+  return pathname_.length() == 3 && IsAbsolutePath();
+#else
+  return pathname_.length() == 1 && IsPathSeparator(pathname_.c_str()[0]);
+#endif
+}
+
+// Returns true if pathname describes an absolute path.
+bool FilePath::IsAbsolutePath() const {
+  const char* const name = pathname_.c_str();
+#if GTEST_OS_WINDOWS
+  return pathname_.length() >= 3 &&
+     ((name[0] >= 'a' && name[0] <= 'z') ||
+      (name[0] >= 'A' && name[0] <= 'Z')) &&
+     name[1] == ':' &&
+     IsPathSeparator(name[2]);
+#else
+  return IsPathSeparator(name[0]);
+#endif
+}
+
+// Returns a pathname for a file that does not currently exist. The pathname
+// will be directory/base_name.extension or
+// directory/base_name_<number>.extension if directory/base_name.extension
+// already exists. The number will be incremented until a pathname is found
+// that does not already exist.
+// Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'.
+// There could be a race condition if two or more processes are calling this
+// function at the same time -- they could both pick the same filename.
+FilePath FilePath::GenerateUniqueFileName(const FilePath& directory,
+                                          const FilePath& base_name,
+                                          const char* extension) {
+  FilePath full_pathname;
+  int number = 0;
+  do {
+    full_pathname.Set(MakeFileName(directory, base_name, number++, extension));
+  } while (full_pathname.FileOrDirectoryExists());
+  return full_pathname;
+}
+
+// Returns true if FilePath ends with a path separator, which indicates that
+// it is intended to represent a directory. Returns false otherwise.
+// This does NOT check that a directory (or file) actually exists.
+bool FilePath::IsDirectory() const {
+  return !pathname_.empty() &&
+         IsPathSeparator(pathname_.c_str()[pathname_.length() - 1]);
+}
+
+// Create directories so that path exists. Returns true if successful or if
+// the directories already exist; returns false if unable to create directories
+// for any reason.
+bool FilePath::CreateDirectoriesRecursively() const {
+  if (!this->IsDirectory()) {
+    return false;
+  }
+
+  if (pathname_.length() == 0 || this->DirectoryExists()) {
+    return true;
+  }
+
+  const FilePath parent(this->RemoveTrailingPathSeparator().RemoveFileName());
+  return parent.CreateDirectoriesRecursively() && this->CreateFolder();
+}
+
+// Create the directory so that path exists. Returns true if successful or
+// if the directory already exists; returns false if unable to create the
+// directory for any reason, including if the parent directory does not
+// exist. Not named "CreateDirectory" because that's a macro on Windows.
+bool FilePath::CreateFolder() const {
+#if GTEST_OS_WINDOWS_MOBILE
+  FilePath removed_sep(this->RemoveTrailingPathSeparator());
+  LPCWSTR unicode = String::AnsiToUtf16(removed_sep.c_str());
+  int result = CreateDirectory(unicode, NULL) ? 0 : -1;
+  delete [] unicode;
+#elif GTEST_OS_WINDOWS
+  int result = _mkdir(pathname_.c_str());
+#else
+  int result = mkdir(pathname_.c_str(), 0777);
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+  if (result == -1) {
+    return this->DirectoryExists();  // An error is OK if the directory exists.
+  }
+  return true;  // No error.
+}
+
+// If input name has a trailing separator character, remove it and return the
+// name, otherwise return the name string unmodified.
+// On Windows platform, uses \ as the separator, other platforms use /.
+FilePath FilePath::RemoveTrailingPathSeparator() const {
+  return IsDirectory()
+      ? FilePath(pathname_.substr(0, pathname_.length() - 1))
+      : *this;
+}
+
+// Removes any redundant separators that might be in the pathname.
+// For example, "bar///foo" becomes "bar/foo". Does not eliminate other
+// redundancies that might be in a pathname involving "." or "..".
+// TODO(wan@google.com): handle Windows network shares (e.g. \\server\share).
+void FilePath::Normalize() {
+  if (pathname_.c_str() == NULL) {
+    pathname_ = "";
+    return;
+  }
+  const char* src = pathname_.c_str();
+  char* const dest = new char[pathname_.length() + 1];
+  char* dest_ptr = dest;
+  memset(dest_ptr, 0, pathname_.length() + 1);
+
+  while (*src != '\0') {
+    *dest_ptr = *src;
+    if (!IsPathSeparator(*src)) {
+      src++;
+    } else {
+#if GTEST_HAS_ALT_PATH_SEP_
+      if (*dest_ptr == kAlternatePathSeparator) {
+        *dest_ptr = kPathSeparator;
+      }
+#endif
+      while (IsPathSeparator(*src))
+        src++;
+    }
+    dest_ptr++;
+  }
+  *dest_ptr = '\0';
+  pathname_ = dest;
+  delete[] dest;
+}
+
+}  // namespace internal
+}  // namespace testing
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+
+#include <limits.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#include <fstream>
+
+#if GTEST_OS_WINDOWS
+# include <windows.h>
+# include <io.h>
+# include <sys/stat.h>
+# include <map>  // Used in ThreadLocal.
+#else
+# include <unistd.h>
+#endif  // GTEST_OS_WINDOWS
+
+#if GTEST_OS_MAC
+# include <mach/mach_init.h>
+# include <mach/task.h>
+# include <mach/vm_map.h>
+#endif  // GTEST_OS_MAC
+
+#if GTEST_OS_QNX
+# include <devctl.h>
+# include <fcntl.h>
+# include <sys/procfs.h>
+#endif  // GTEST_OS_QNX
+
+#if GTEST_OS_AIX
+# include <procinfo.h>
+# include <sys/types.h>
+#endif  // GTEST_OS_AIX
+
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick exists to
+// prevent the accidental inclusion of gtest-internal-inl.h in the
+// user's code.
+#define GTEST_IMPLEMENTATION_ 1
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+namespace internal {
+
+#if defined(_MSC_VER) || defined(__BORLANDC__)
+// MSVC and C++Builder do not provide a definition of STDERR_FILENO.
+const int kStdOutFileno = 1;
+const int kStdErrFileno = 2;
+#else
+const int kStdOutFileno = STDOUT_FILENO;
+const int kStdErrFileno = STDERR_FILENO;
+#endif  // _MSC_VER
+
+#if GTEST_OS_LINUX
+
+namespace {
+template <typename T>
+T ReadProcFileField(const std::string& filename, int field) {
+  std::string dummy;
+  std::ifstream file(filename.c_str());
+  while (field-- > 0) {
+    file >> dummy;
+  }
+  T output = 0;
+  file >> output;
+  return output;
+}
+}  // namespace
+
+// Returns the number of active threads, or 0 when there is an error.
+size_t GetThreadCount() {
+  const std::string filename =
+      (Message() << "/proc/" << getpid() << "/stat").GetString();
+  return ReadProcFileField<int>(filename, 19);
+}
+
+#elif GTEST_OS_MAC
+
+size_t GetThreadCount() {
+  const task_t task = mach_task_self();
+  mach_msg_type_number_t thread_count;
+  thread_act_array_t thread_list;
+  const kern_return_t status = task_threads(task, &thread_list, &thread_count);
+  if (status == KERN_SUCCESS) {
+    // task_threads allocates resources in thread_list and we need to free them
+    // to avoid leaks.
+    vm_deallocate(task,
+                  reinterpret_cast<vm_address_t>(thread_list),
+                  sizeof(thread_t) * thread_count);
+    return static_cast<size_t>(thread_count);
+  } else {
+    return 0;
+  }
+}
+
+#elif GTEST_OS_QNX
+
+// Returns the number of threads running in the process, or 0 to indicate that
+// we cannot detect it.
+size_t GetThreadCount() {
+  const int fd = open("/proc/self/as", O_RDONLY);
+  if (fd < 0) {
+    return 0;
+  }
+  procfs_info process_info;
+  const int status =
+      devctl(fd, DCMD_PROC_INFO, &process_info, sizeof(process_info), NULL);
+  close(fd);
+  if (status == EOK) {
+    return static_cast<size_t>(process_info.num_threads);
+  } else {
+    return 0;
+  }
+}
+
+#elif GTEST_OS_AIX
+
+size_t GetThreadCount() {
+  struct procentry64 entry;
+  pid_t pid = getpid();
+  int status = getprocs64(&entry, sizeof(entry), NULL, 0, &pid, 1);
+  if (status == 1) {
+    return entry.pi_thcount;
+  } else {
+    return 0;
+  }
+}
+
+#else
+
+size_t GetThreadCount() {
+  // There's no portable way to detect the number of threads, so we just
+  // return 0 to indicate that we cannot detect it.
+  return 0;
+}
+
+#endif  // GTEST_OS_LINUX
+
+#if GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS
+
+void SleepMilliseconds(int n) {
+  ::Sleep(n);
+}
+
+AutoHandle::AutoHandle()
+    : handle_(INVALID_HANDLE_VALUE) {}
+
+AutoHandle::AutoHandle(Handle handle)
+    : handle_(handle) {}
+
+AutoHandle::~AutoHandle() {
+  Reset();
+}
+
+AutoHandle::Handle AutoHandle::Get() const {
+  return handle_;
+}
+
+void AutoHandle::Reset() {
+  Reset(INVALID_HANDLE_VALUE);
+}
+
+void AutoHandle::Reset(HANDLE handle) {
+  // Resetting with the same handle we already own is invalid.
+  if (handle_ != handle) {
+    if (IsCloseable()) {
+      ::CloseHandle(handle_);
+    }
+    handle_ = handle;
+  } else {
+    GTEST_CHECK_(!IsCloseable())
+        << "Resetting a valid handle to itself is likely a programmer error "
+            "and thus not allowed.";
+  }
+}
+
+bool AutoHandle::IsCloseable() const {
+  // Different Windows APIs may use either of these values to represent an
+  // invalid handle.
+  return handle_ != NULL && handle_ != INVALID_HANDLE_VALUE;
+}
+
+Notification::Notification()
+    : event_(::CreateEvent(NULL,   // Default security attributes.
+                           TRUE,   // Do not reset automatically.
+                           FALSE,  // Initially unset.
+                           NULL)) {  // Anonymous event.
+  GTEST_CHECK_(event_.Get() != NULL);
+}
+
+void Notification::Notify() {
+  GTEST_CHECK_(::SetEvent(event_.Get()) != FALSE);
+}
+
+void Notification::WaitForNotification() {
+  GTEST_CHECK_(
+      ::WaitForSingleObject(event_.Get(), INFINITE) == WAIT_OBJECT_0);
+}
+
+Mutex::Mutex()
+    : owner_thread_id_(0),
+      type_(kDynamic),
+      critical_section_init_phase_(0),
+      critical_section_(new CRITICAL_SECTION) {
+  ::InitializeCriticalSection(critical_section_);
+}
+
+Mutex::~Mutex() {
+  // Static mutexes are leaked intentionally. It is not thread-safe to try
+  // to clean them up.
+  // TODO(yukawa): Switch to Slim Reader/Writer (SRW) Locks, which requires
+  // nothing to clean it up but is available only on Vista and later.
+  // http://msdn.microsoft.com/en-us/library/windows/desktop/aa904937.aspx
+  if (type_ == kDynamic) {
+    ::DeleteCriticalSection(critical_section_);
+    delete critical_section_;
+    critical_section_ = NULL;
+  }
+}
+
+void Mutex::Lock() {
+  ThreadSafeLazyInit();
+  ::EnterCriticalSection(critical_section_);
+  owner_thread_id_ = ::GetCurrentThreadId();
+}
+
+void Mutex::Unlock() {
+  ThreadSafeLazyInit();
+  // We don't protect writing to owner_thread_id_ here, as it's the
+  // caller's responsibility to ensure that the current thread holds the
+  // mutex when this is called.
+  owner_thread_id_ = 0;
+  ::LeaveCriticalSection(critical_section_);
+}
+
+// Does nothing if the current thread holds the mutex. Otherwise, crashes
+// with high probability.
+void Mutex::AssertHeld() {
+  ThreadSafeLazyInit();
+  GTEST_CHECK_(owner_thread_id_ == ::GetCurrentThreadId())
+      << "The current thread is not holding the mutex @" << this;
+}
+
+// Initializes owner_thread_id_ and critical_section_ in static mutexes.
+void Mutex::ThreadSafeLazyInit() {
+  // Dynamic mutexes are initialized in the constructor.
+  if (type_ == kStatic) {
+    switch (
+        ::InterlockedCompareExchange(&critical_section_init_phase_, 1L, 0L)) {
+      case 0:
+        // If critical_section_init_phase_ was 0 before the exchange, we
+        // are the first to test it and need to perform the initialization.
+        owner_thread_id_ = 0;
+        critical_section_ = new CRITICAL_SECTION;
+        ::InitializeCriticalSection(critical_section_);
+        // Updates the critical_section_init_phase_ to 2 to signal
+        // initialization complete.
+        GTEST_CHECK_(::InterlockedCompareExchange(
+                          &critical_section_init_phase_, 2L, 1L) ==
+                      1L);
+        break;
+      case 1:
+        // Somebody else is already initializing the mutex; spin until they
+        // are done.
+        while (::InterlockedCompareExchange(&critical_section_init_phase_,
+                                            2L,
+                                            2L) != 2L) {
+          // Possibly yields the rest of the thread's time slice to other
+          // threads.
+          ::Sleep(0);
+        }
+        break;
+
+      case 2:
+        break;  // The mutex is already initialized and ready for use.
+
+      default:
+        GTEST_CHECK_(false)
+            << "Unexpected value of critical_section_init_phase_ "
+            << "while initializing a static mutex.";
+    }
+  }
+}
+
+namespace {
+
+class ThreadWithParamSupport : public ThreadWithParamBase {
+ public:
+  static HANDLE CreateThread(Runnable* runnable,
+                             Notification* thread_can_start) {
+    ThreadMainParam* param = new ThreadMainParam(runnable, thread_can_start);
+    DWORD thread_id;
+    // TODO(yukawa): Consider to use _beginthreadex instead.
+    HANDLE thread_handle = ::CreateThread(
+        NULL,    // Default security.
+        0,       // Default stack size.
+        &ThreadWithParamSupport::ThreadMain,
+        param,   // Parameter to ThreadMainStatic
+        0x0,     // Default creation flags.
+        &thread_id);  // Need a valid pointer for the call to work under Win98.
+    GTEST_CHECK_(thread_handle != NULL) << "CreateThread failed with error "
+                                        << ::GetLastError() << ".";
+    if (thread_handle == NULL) {
+      delete param;
+    }
+    return thread_handle;
+  }
+
+ private:
+  struct ThreadMainParam {
+    ThreadMainParam(Runnable* runnable, Notification* thread_can_start)
+        : runnable_(runnable),
+          thread_can_start_(thread_can_start) {
+    }
+    scoped_ptr<Runnable> runnable_;
+    // Does not own.
+    Notification* thread_can_start_;
+  };
+
+  static DWORD WINAPI ThreadMain(void* ptr) {
+    // Transfers ownership.
+    scoped_ptr<ThreadMainParam> param(static_cast<ThreadMainParam*>(ptr));
+    if (param->thread_can_start_ != NULL)
+      param->thread_can_start_->WaitForNotification();
+    param->runnable_->Run();
+    return 0;
+  }
+
+  // Prohibit instantiation.
+  ThreadWithParamSupport();
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParamSupport);
+};
+
+}  // namespace
+
+ThreadWithParamBase::ThreadWithParamBase(Runnable *runnable,
+                                         Notification* thread_can_start)
+      : thread_(ThreadWithParamSupport::CreateThread(runnable,
+                                                     thread_can_start)) {
+}
+
+ThreadWithParamBase::~ThreadWithParamBase() {
+  Join();
+}
+
+void ThreadWithParamBase::Join() {
+  GTEST_CHECK_(::WaitForSingleObject(thread_.Get(), INFINITE) == WAIT_OBJECT_0)
+      << "Failed to join the thread with error " << ::GetLastError() << ".";
+}
+
+// Maps a thread to a set of ThreadIdToThreadLocals that have values
+// instantiated on that thread and notifies them when the thread exits.  A
+// ThreadLocal instance is expected to persist until all threads it has
+// values on have terminated.
+class ThreadLocalRegistryImpl {
+ public:
+  // Registers thread_local_instance as having value on the current thread.
+  // Returns a value that can be used to identify the thread from other threads.
+  static ThreadLocalValueHolderBase* GetValueOnCurrentThread(
+      const ThreadLocalBase* thread_local_instance) {
+    DWORD current_thread = ::GetCurrentThreadId();
+    MutexLock lock(&mutex_);
+    ThreadIdToThreadLocals* const thread_to_thread_locals =
+        GetThreadLocalsMapLocked();
+    ThreadIdToThreadLocals::iterator thread_local_pos =
+        thread_to_thread_locals->find(current_thread);
+    if (thread_local_pos == thread_to_thread_locals->end()) {
+      thread_local_pos = thread_to_thread_locals->insert(
+          std::make_pair(current_thread, ThreadLocalValues())).first;
+      StartWatcherThreadFor(current_thread);
+    }
+    ThreadLocalValues& thread_local_values = thread_local_pos->second;
+    ThreadLocalValues::iterator value_pos =
+        thread_local_values.find(thread_local_instance);
+    if (value_pos == thread_local_values.end()) {
+      value_pos =
+          thread_local_values
+              .insert(std::make_pair(
+                  thread_local_instance,
+                  linked_ptr<ThreadLocalValueHolderBase>(
+                      thread_local_instance->NewValueForCurrentThread())))
+              .first;
+    }
+    return value_pos->second.get();
+  }
+
+  static void OnThreadLocalDestroyed(
+      const ThreadLocalBase* thread_local_instance) {
+    std::vector<linked_ptr<ThreadLocalValueHolderBase> > value_holders;
+    // Clean up the ThreadLocalValues data structure while holding the lock, but
+    // defer the destruction of the ThreadLocalValueHolderBases.
+    {
+      MutexLock lock(&mutex_);
+      ThreadIdToThreadLocals* const thread_to_thread_locals =
+          GetThreadLocalsMapLocked();
+      for (ThreadIdToThreadLocals::iterator it =
+          thread_to_thread_locals->begin();
+          it != thread_to_thread_locals->end();
+          ++it) {
+        ThreadLocalValues& thread_local_values = it->second;
+        ThreadLocalValues::iterator value_pos =
+            thread_local_values.find(thread_local_instance);
+        if (value_pos != thread_local_values.end()) {
+          value_holders.push_back(value_pos->second);
+          thread_local_values.erase(value_pos);
+          // This 'if' can only be successful at most once, so theoretically we
+          // could break out of the loop here, but we don't bother doing so.
+        }
+      }
+    }
+    // Outside the lock, let the destructor for 'value_holders' deallocate the
+    // ThreadLocalValueHolderBases.
+  }
+
+  static void OnThreadExit(DWORD thread_id) {
+    GTEST_CHECK_(thread_id != 0) << ::GetLastError();
+    std::vector<linked_ptr<ThreadLocalValueHolderBase> > value_holders;
+    // Clean up the ThreadIdToThreadLocals data structure while holding the
+    // lock, but defer the destruction of the ThreadLocalValueHolderBases.
+    {
+      MutexLock lock(&mutex_);
+      ThreadIdToThreadLocals* const thread_to_thread_locals =
+          GetThreadLocalsMapLocked();
+      ThreadIdToThreadLocals::iterator thread_local_pos =
+          thread_to_thread_locals->find(thread_id);
+      if (thread_local_pos != thread_to_thread_locals->end()) {
+        ThreadLocalValues& thread_local_values = thread_local_pos->second;
+        for (ThreadLocalValues::iterator value_pos =
+            thread_local_values.begin();
+            value_pos != thread_local_values.end();
+            ++value_pos) {
+          value_holders.push_back(value_pos->second);
+        }
+        thread_to_thread_locals->erase(thread_local_pos);
+      }
+    }
+    // Outside the lock, let the destructor for 'value_holders' deallocate the
+    // ThreadLocalValueHolderBases.
+  }
+
+ private:
+  // In a particular thread, maps a ThreadLocal object to its value.
+  typedef std::map<const ThreadLocalBase*,
+                   linked_ptr<ThreadLocalValueHolderBase> > ThreadLocalValues;
+  // Stores all ThreadIdToThreadLocals having values in a thread, indexed by
+  // thread's ID.
+  typedef std::map<DWORD, ThreadLocalValues> ThreadIdToThreadLocals;
+
+  // Holds the thread id and thread handle that we pass from
+  // StartWatcherThreadFor to WatcherThreadFunc.
+  typedef std::pair<DWORD, HANDLE> ThreadIdAndHandle;
+
+  static void StartWatcherThreadFor(DWORD thread_id) {
+    // The returned handle will be kept in thread_map and closed by
+    // watcher_thread in WatcherThreadFunc.
+    HANDLE thread = ::OpenThread(SYNCHRONIZE | THREAD_QUERY_INFORMATION,
+                                 FALSE,
+                                 thread_id);
+    GTEST_CHECK_(thread != NULL);
+    // We need to to pass a valid thread ID pointer into CreateThread for it
+    // to work correctly under Win98.
+    DWORD watcher_thread_id;
+    HANDLE watcher_thread = ::CreateThread(
+        NULL,   // Default security.
+        0,      // Default stack size
+        &ThreadLocalRegistryImpl::WatcherThreadFunc,
+        reinterpret_cast<LPVOID>(new ThreadIdAndHandle(thread_id, thread)),
+        CREATE_SUSPENDED,
+        &watcher_thread_id);
+    GTEST_CHECK_(watcher_thread != NULL);
+    // Give the watcher thread the same priority as ours to avoid being
+    // blocked by it.
+    ::SetThreadPriority(watcher_thread,
+                        ::GetThreadPriority(::GetCurrentThread()));
+    ::ResumeThread(watcher_thread);
+    ::CloseHandle(watcher_thread);
+  }
+
+  // Monitors exit from a given thread and notifies those
+  // ThreadIdToThreadLocals about thread termination.
+  static DWORD WINAPI WatcherThreadFunc(LPVOID param) {
+    const ThreadIdAndHandle* tah =
+        reinterpret_cast<const ThreadIdAndHandle*>(param);
+    GTEST_CHECK_(
+        ::WaitForSingleObject(tah->second, INFINITE) == WAIT_OBJECT_0);
+    OnThreadExit(tah->first);
+    ::CloseHandle(tah->second);
+    delete tah;
+    return 0;
+  }
+
+  // Returns map of thread local instances.
+  static ThreadIdToThreadLocals* GetThreadLocalsMapLocked() {
+    mutex_.AssertHeld();
+    static ThreadIdToThreadLocals* map = new ThreadIdToThreadLocals;
+    return map;
+  }
+
+  // Protects access to GetThreadLocalsMapLocked() and its return value.
+  static Mutex mutex_;
+  // Protects access to GetThreadMapLocked() and its return value.
+  static Mutex thread_map_mutex_;
+};
+
+Mutex ThreadLocalRegistryImpl::mutex_(Mutex::kStaticMutex);
+Mutex ThreadLocalRegistryImpl::thread_map_mutex_(Mutex::kStaticMutex);
+
+ThreadLocalValueHolderBase* ThreadLocalRegistry::GetValueOnCurrentThread(
+      const ThreadLocalBase* thread_local_instance) {
+  return ThreadLocalRegistryImpl::GetValueOnCurrentThread(
+      thread_local_instance);
+}
+
+void ThreadLocalRegistry::OnThreadLocalDestroyed(
+      const ThreadLocalBase* thread_local_instance) {
+  ThreadLocalRegistryImpl::OnThreadLocalDestroyed(thread_local_instance);
+}
+
+#endif  // GTEST_IS_THREADSAFE && GTEST_OS_WINDOWS
+
+#if GTEST_USES_POSIX_RE
+
+// Implements RE.  Currently only needed for death tests.
+
+RE::~RE() {
+  if (is_valid_) {
+    // regfree'ing an invalid regex might crash because the content
+    // of the regex is undefined. Since the regex's are essentially
+    // the same, one cannot be valid (or invalid) without the other
+    // being so too.
+    regfree(&partial_regex_);
+    regfree(&full_regex_);
+  }
+  free(const_cast<char*>(pattern_));
+}
+
+// Returns true iff regular expression re matches the entire str.
+bool RE::FullMatch(const char* str, const RE& re) {
+  if (!re.is_valid_) return false;
+
+  regmatch_t match;
+  return regexec(&re.full_regex_, str, 1, &match, 0) == 0;
+}
+
+// Returns true iff regular expression re matches a substring of str
+// (including str itself).
+bool RE::PartialMatch(const char* str, const RE& re) {
+  if (!re.is_valid_) return false;
+
+  regmatch_t match;
+  return regexec(&re.partial_regex_, str, 1, &match, 0) == 0;
+}
+
+// Initializes an RE from its string representation.
+void RE::Init(const char* regex) {
+  pattern_ = posix::StrDup(regex);
+
+  // Reserves enough bytes to hold the regular expression used for a
+  // full match.
+  const size_t full_regex_len = strlen(regex) + 10;
+  char* const full_pattern = new char[full_regex_len];
+
+  snprintf(full_pattern, full_regex_len, "^(%s)$", regex);
+  is_valid_ = regcomp(&full_regex_, full_pattern, REG_EXTENDED) == 0;
+  // We want to call regcomp(&partial_regex_, ...) even if the
+  // previous expression returns false.  Otherwise partial_regex_ may
+  // not be properly initialized can may cause trouble when it's
+  // freed.
+  //
+  // Some implementation of POSIX regex (e.g. on at least some
+  // versions of Cygwin) doesn't accept the empty string as a valid
+  // regex.  We change it to an equivalent form "()" to be safe.
+  if (is_valid_) {
+    const char* const partial_regex = (*regex == '\0') ? "()" : regex;
+    is_valid_ = regcomp(&partial_regex_, partial_regex, REG_EXTENDED) == 0;
+  }
+  EXPECT_TRUE(is_valid_)
+      << "Regular expression \"" << regex
+      << "\" is not a valid POSIX Extended regular expression.";
+
+  delete[] full_pattern;
+}
+
+#elif GTEST_USES_SIMPLE_RE
+
+// Returns true iff ch appears anywhere in str (excluding the
+// terminating '\0' character).
+bool IsInSet(char ch, const char* str) {
+  return ch != '\0' && strchr(str, ch) != NULL;
+}
+
+// Returns true iff ch belongs to the given classification.  Unlike
+// similar functions in <ctype.h>, these aren't affected by the
+// current locale.
+bool IsAsciiDigit(char ch) { return '0' <= ch && ch <= '9'; }
+bool IsAsciiPunct(char ch) {
+  return IsInSet(ch, "^-!\"#$%&'()*+,./:;<=>?@[\\]_`{|}~");
+}
+bool IsRepeat(char ch) { return IsInSet(ch, "?*+"); }
+bool IsAsciiWhiteSpace(char ch) { return IsInSet(ch, " \f\n\r\t\v"); }
+bool IsAsciiWordChar(char ch) {
+  return ('a' <= ch && ch <= 'z') || ('A' <= ch && ch <= 'Z') ||
+      ('0' <= ch && ch <= '9') || ch == '_';
+}
+
+// Returns true iff "\\c" is a supported escape sequence.
+bool IsValidEscape(char c) {
+  return (IsAsciiPunct(c) || IsInSet(c, "dDfnrsStvwW"));
+}
+
+// Returns true iff the given atom (specified by escaped and pattern)
+// matches ch.  The result is undefined if the atom is invalid.
+bool AtomMatchesChar(bool escaped, char pattern_char, char ch) {
+  if (escaped) {  // "\\p" where p is pattern_char.
+    switch (pattern_char) {
+      case 'd': return IsAsciiDigit(ch);
+      case 'D': return !IsAsciiDigit(ch);
+      case 'f': return ch == '\f';
+      case 'n': return ch == '\n';
+      case 'r': return ch == '\r';
+      case 's': return IsAsciiWhiteSpace(ch);
+      case 'S': return !IsAsciiWhiteSpace(ch);
+      case 't': return ch == '\t';
+      case 'v': return ch == '\v';
+      case 'w': return IsAsciiWordChar(ch);
+      case 'W': return !IsAsciiWordChar(ch);
+    }
+    return IsAsciiPunct(pattern_char) && pattern_char == ch;
+  }
+
+  return (pattern_char == '.' && ch != '\n') || pattern_char == ch;
+}
+
+// Helper function used by ValidateRegex() to format error messages.
+std::string FormatRegexSyntaxError(const char* regex, int index) {
+  return (Message() << "Syntax error at index " << index
+          << " in simple regular expression \"" << regex << "\": ").GetString();
+}
+
+// Generates non-fatal failures and returns false if regex is invalid;
+// otherwise returns true.
+bool ValidateRegex(const char* regex) {
+  if (regex == NULL) {
+    // TODO(wan@google.com): fix the source file location in the
+    // assertion failures to match where the regex is used in user
+    // code.
+    ADD_FAILURE() << "NULL is not a valid simple regular expression.";
+    return false;
+  }
+
+  bool is_valid = true;
+
+  // True iff ?, *, or + can follow the previous atom.
+  bool prev_repeatable = false;
+  for (int i = 0; regex[i]; i++) {
+    if (regex[i] == '\\') {  // An escape sequence
+      i++;
+      if (regex[i] == '\0') {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
+                      << "'\\' cannot appear at the end.";
+        return false;
+      }
+
+      if (!IsValidEscape(regex[i])) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i - 1)
+                      << "invalid escape sequence \"\\" << regex[i] << "\".";
+        is_valid = false;
+      }
+      prev_repeatable = true;
+    } else {  // Not an escape sequence.
+      const char ch = regex[i];
+
+      if (ch == '^' && i > 0) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'^' can only appear at the beginning.";
+        is_valid = false;
+      } else if (ch == '$' && regex[i + 1] != '\0') {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'$' can only appear at the end.";
+        is_valid = false;
+      } else if (IsInSet(ch, "()[]{}|")) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'" << ch << "' is unsupported.";
+        is_valid = false;
+      } else if (IsRepeat(ch) && !prev_repeatable) {
+        ADD_FAILURE() << FormatRegexSyntaxError(regex, i)
+                      << "'" << ch << "' can only follow a repeatable token.";
+        is_valid = false;
+      }
+
+      prev_repeatable = !IsInSet(ch, "^$?*+");
+    }
+  }
+
+  return is_valid;
+}
+
+// Matches a repeated regex atom followed by a valid simple regular
+// expression.  The regex atom is defined as c if escaped is false,
+// or \c otherwise.  repeat is the repetition meta character (?, *,
+// or +).  The behavior is undefined if str contains too many
+// characters to be indexable by size_t, in which case the test will
+// probably time out anyway.  We are fine with this limitation as
+// std::string has it too.
+bool MatchRepetitionAndRegexAtHead(
+    bool escaped, char c, char repeat, const char* regex,
+    const char* str) {
+  const size_t min_count = (repeat == '+') ? 1 : 0;
+  const size_t max_count = (repeat == '?') ? 1 :
+      static_cast<size_t>(-1) - 1;
+  // We cannot call numeric_limits::max() as it conflicts with the
+  // max() macro on Windows.
+
+  for (size_t i = 0; i <= max_count; ++i) {
+    // We know that the atom matches each of the first i characters in str.
+    if (i >= min_count && MatchRegexAtHead(regex, str + i)) {
+      // We have enough matches at the head, and the tail matches too.
+      // Since we only care about *whether* the pattern matches str
+      // (as opposed to *how* it matches), there is no need to find a
+      // greedy match.
+      return true;
+    }
+    if (str[i] == '\0' || !AtomMatchesChar(escaped, c, str[i]))
+      return false;
+  }
+  return false;
+}
+
+// Returns true iff regex matches a prefix of str.  regex must be a
+// valid simple regular expression and not start with "^", or the
+// result is undefined.
+bool MatchRegexAtHead(const char* regex, const char* str) {
+  if (*regex == '\0')  // An empty regex matches a prefix of anything.
+    return true;
+
+  // "$" only matches the end of a string.  Note that regex being
+  // valid guarantees that there's nothing after "$" in it.
+  if (*regex == '$')
+    return *str == '\0';
+
+  // Is the first thing in regex an escape sequence?
+  const bool escaped = *regex == '\\';
+  if (escaped)
+    ++regex;
+  if (IsRepeat(regex[1])) {
+    // MatchRepetitionAndRegexAtHead() calls MatchRegexAtHead(), so
+    // here's an indirect recursion.  It terminates as the regex gets
+    // shorter in each recursion.
+    return MatchRepetitionAndRegexAtHead(
+        escaped, regex[0], regex[1], regex + 2, str);
+  } else {
+    // regex isn't empty, isn't "$", and doesn't start with a
+    // repetition.  We match the first atom of regex with the first
+    // character of str and recurse.
+    return (*str != '\0') && AtomMatchesChar(escaped, *regex, *str) &&
+        MatchRegexAtHead(regex + 1, str + 1);
+  }
+}
+
+// Returns true iff regex matches any substring of str.  regex must be
+// a valid simple regular expression, or the result is undefined.
+//
+// The algorithm is recursive, but the recursion depth doesn't exceed
+// the regex length, so we won't need to worry about running out of
+// stack space normally.  In rare cases the time complexity can be
+// exponential with respect to the regex length + the string length,
+// but usually it's must faster (often close to linear).
+bool MatchRegexAnywhere(const char* regex, const char* str) {
+  if (regex == NULL || str == NULL)
+    return false;
+
+  if (*regex == '^')
+    return MatchRegexAtHead(regex + 1, str);
+
+  // A successful match can be anywhere in str.
+  do {
+    if (MatchRegexAtHead(regex, str))
+      return true;
+  } while (*str++ != '\0');
+  return false;
+}
+
+// Implements the RE class.
+
+RE::~RE() {
+  free(const_cast<char*>(pattern_));
+  free(const_cast<char*>(full_pattern_));
+}
+
+// Returns true iff regular expression re matches the entire str.
+bool RE::FullMatch(const char* str, const RE& re) {
+  return re.is_valid_ && MatchRegexAnywhere(re.full_pattern_, str);
+}
+
+// Returns true iff regular expression re matches a substring of str
+// (including str itself).
+bool RE::PartialMatch(const char* str, const RE& re) {
+  return re.is_valid_ && MatchRegexAnywhere(re.pattern_, str);
+}
+
+// Initializes an RE from its string representation.
+void RE::Init(const char* regex) {
+  pattern_ = full_pattern_ = NULL;
+  if (regex != NULL) {
+    pattern_ = posix::StrDup(regex);
+  }
+
+  is_valid_ = ValidateRegex(regex);
+  if (!is_valid_) {
+    // No need to calculate the full pattern when the regex is invalid.
+    return;
+  }
+
+  const size_t len = strlen(regex);
+  // Reserves enough bytes to hold the regular expression used for a
+  // full match: we need space to prepend a '^', append a '$', and
+  // terminate the string with '\0'.
+  char* buffer = static_cast<char*>(malloc(len + 3));
+  full_pattern_ = buffer;
+
+  if (*regex != '^')
+    *buffer++ = '^';  // Makes sure full_pattern_ starts with '^'.
+
+  // We don't use snprintf or strncpy, as they trigger a warning when
+  // compiled with VC++ 8.0.
+  memcpy(buffer, regex, len);
+  buffer += len;
+
+  if (len == 0 || regex[len - 1] != '$')
+    *buffer++ = '$';  // Makes sure full_pattern_ ends with '$'.
+
+  *buffer = '\0';
+}
+
+#endif  // GTEST_USES_POSIX_RE
+
+const char kUnknownFile[] = "unknown file";
+
+// Formats a source file path and a line number as they would appear
+// in an error message from the compiler used to compile this code.
+GTEST_API_ ::std::string FormatFileLocation(const char* file, int line) {
+  const std::string file_name(file == NULL ? kUnknownFile : file);
+
+  if (line < 0) {
+    return file_name + ":";
+  }
+#ifdef _MSC_VER
+  return file_name + "(" + StreamableToString(line) + "):";
+#else
+  return file_name + ":" + StreamableToString(line) + ":";
+#endif  // _MSC_VER
+}
+
+// Formats a file location for compiler-independent XML output.
+// Although this function is not platform dependent, we put it next to
+// FormatFileLocation in order to contrast the two functions.
+// Note that FormatCompilerIndependentFileLocation() does NOT append colon
+// to the file location it produces, unlike FormatFileLocation().
+GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(
+    const char* file, int line) {
+  const std::string file_name(file == NULL ? kUnknownFile : file);
+
+  if (line < 0)
+    return file_name;
+  else
+    return file_name + ":" + StreamableToString(line);
+}
+
+GTestLog::GTestLog(GTestLogSeverity severity, const char* file, int line)
+    : severity_(severity) {
+  const char* const marker =
+      severity == GTEST_INFO ?    "[  INFO ]" :
+      severity == GTEST_WARNING ? "[WARNING]" :
+      severity == GTEST_ERROR ?   "[ ERROR ]" : "[ FATAL ]";
+  GetStream() << ::std::endl << marker << " "
+              << FormatFileLocation(file, line).c_str() << ": ";
+}
+
+// Flushes the buffers and, if severity is GTEST_FATAL, aborts the program.
+GTestLog::~GTestLog() {
+  GetStream() << ::std::endl;
+  if (severity_ == GTEST_FATAL) {
+    fflush(stderr);
+    posix::Abort();
+  }
+}
+// Disable Microsoft deprecation warnings for POSIX functions called from
+// this class (creat, dup, dup2, and close)
+GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996)
+
+#if GTEST_HAS_STREAM_REDIRECTION
+
+// Object that captures an output stream (stdout/stderr).
+class CapturedStream {
+ public:
+  // The ctor redirects the stream to a temporary file.
+  explicit CapturedStream(int fd) : fd_(fd), uncaptured_fd_(dup(fd)) {
+# if GTEST_OS_WINDOWS
+    char temp_dir_path[MAX_PATH + 1] = { '\0' };  // NOLINT
+    char temp_file_path[MAX_PATH + 1] = { '\0' };  // NOLINT
+
+    ::GetTempPathA(sizeof(temp_dir_path), temp_dir_path);
+    const UINT success = ::GetTempFileNameA(temp_dir_path,
+                                            "gtest_redir",
+                                            0,  // Generate unique file name.
+                                            temp_file_path);
+    GTEST_CHECK_(success != 0)
+        << "Unable to create a temporary file in " << temp_dir_path;
+    const int captured_fd = creat(temp_file_path, _S_IREAD | _S_IWRITE);
+    GTEST_CHECK_(captured_fd != -1) << "Unable to open temporary file "
+                                    << temp_file_path;
+    filename_ = temp_file_path;
+# else
+    // There's no guarantee that a test has write access to the current
+    // directory, so we create the temporary file in the /tmp directory
+    // instead. We use /tmp on most systems, and /sdcard on Android.
+    // That's because Android doesn't have /tmp.
+#  if GTEST_OS_LINUX_ANDROID
+    // Note: Android applications are expected to call the framework's
+    // Context.getExternalStorageDirectory() method through JNI to get
+    // the location of the world-writable SD Card directory. However,
+    // this requires a Context handle, which cannot be retrieved
+    // globally from native code. Doing so also precludes running the
+    // code as part of a regular standalone executable, which doesn't
+    // run in a Dalvik process (e.g. when running it through 'adb shell').
+    //
+    // The location /sdcard is directly accessible from native code
+    // and is the only location (unofficially) supported by the Android
+    // team. It's generally a symlink to the real SD Card mount point
+    // which can be /mnt/sdcard, /mnt/sdcard0, /system/media/sdcard, or
+    // other OEM-customized locations. Never rely on these, and always
+    // use /sdcard.
+    char name_template[] = "/sdcard/gtest_captured_stream.XXXXXX";
+#  else
+    char name_template[] = "/tmp/captured_stream.XXXXXX";
+#  endif  // GTEST_OS_LINUX_ANDROID
+    const int captured_fd = mkstemp(name_template);
+    filename_ = name_template;
+# endif  // GTEST_OS_WINDOWS
+    fflush(NULL);
+    dup2(captured_fd, fd_);
+    close(captured_fd);
+  }
+
+  ~CapturedStream() {
+    remove(filename_.c_str());
+  }
+
+  std::string GetCapturedString() {
+    if (uncaptured_fd_ != -1) {
+      // Restores the original stream.
+      fflush(NULL);
+      dup2(uncaptured_fd_, fd_);
+      close(uncaptured_fd_);
+      uncaptured_fd_ = -1;
+    }
+
+    FILE* const file = posix::FOpen(filename_.c_str(), "r");
+    const std::string content = ReadEntireFile(file);
+    posix::FClose(file);
+    return content;
+  }
+
+ private:
+  const int fd_;  // A stream to capture.
+  int uncaptured_fd_;
+  // Name of the temporary file holding the stderr output.
+  ::std::string filename_;
+
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(CapturedStream);
+};
+
+GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+static CapturedStream* g_captured_stderr = NULL;
+static CapturedStream* g_captured_stdout = NULL;
+
+// Starts capturing an output stream (stdout/stderr).
+void CaptureStream(int fd, const char* stream_name, CapturedStream** stream) {
+  if (*stream != NULL) {
+    GTEST_LOG_(FATAL) << "Only one " << stream_name
+                      << " capturer can exist at a time.";
+  }
+  *stream = new CapturedStream(fd);
+}
+
+// Stops capturing the output stream and returns the captured string.
+std::string GetCapturedStream(CapturedStream** captured_stream) {
+  const std::string content = (*captured_stream)->GetCapturedString();
+
+  delete *captured_stream;
+  *captured_stream = NULL;
+
+  return content;
+}
+
+// Starts capturing stdout.
+void CaptureStdout() {
+  CaptureStream(kStdOutFileno, "stdout", &g_captured_stdout);
+}
+
+// Starts capturing stderr.
+void CaptureStderr() {
+  CaptureStream(kStdErrFileno, "stderr", &g_captured_stderr);
+}
+
+// Stops capturing stdout and returns the captured string.
+std::string GetCapturedStdout() {
+  return GetCapturedStream(&g_captured_stdout);
+}
+
+// Stops capturing stderr and returns the captured string.
+std::string GetCapturedStderr() {
+  return GetCapturedStream(&g_captured_stderr);
+}
+
+#endif  // GTEST_HAS_STREAM_REDIRECTION
+
+size_t GetFileSize(FILE* file) {
+  fseek(file, 0, SEEK_END);
+  return static_cast<size_t>(ftell(file));
+}
+
+std::string ReadEntireFile(FILE* file) {
+  const size_t file_size = GetFileSize(file);
+  char* const buffer = new char[file_size];
+
+  size_t bytes_last_read = 0;  // # of bytes read in the last fread()
+  size_t bytes_read = 0;       // # of bytes read so far
+
+  fseek(file, 0, SEEK_SET);
+
+  // Keeps reading the file until we cannot read further or the
+  // pre-determined file size is reached.
+  do {
+    bytes_last_read = fread(buffer+bytes_read, 1, file_size-bytes_read, file);
+    bytes_read += bytes_last_read;
+  } while (bytes_last_read > 0 && bytes_read < file_size);
+
+  const std::string content(buffer, bytes_read);
+  delete[] buffer;
+
+  return content;
+}
+
+#if GTEST_HAS_DEATH_TEST
+
+static const ::std::vector<testing::internal::string>* g_injected_test_argvs =
+                                        NULL;  // Owned.
+
+void SetInjectableArgvs(const ::std::vector<testing::internal::string>* argvs) {
+  if (g_injected_test_argvs != argvs)
+    delete g_injected_test_argvs;
+  g_injected_test_argvs = argvs;
+}
+
+const ::std::vector<testing::internal::string>& GetInjectableArgvs() {
+  if (g_injected_test_argvs != NULL) {
+    return *g_injected_test_argvs;
+  }
+  return GetArgvs();
+}
+#endif  // GTEST_HAS_DEATH_TEST
+
+#if GTEST_OS_WINDOWS_MOBILE
+namespace posix {
+void Abort() {
+  DebugBreak();
+  TerminateProcess(GetCurrentProcess(), 1);
+}
+}  // namespace posix
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+// Returns the name of the environment variable corresponding to the
+// given flag.  For example, FlagToEnvVar("foo") will return
+// "GTEST_FOO" in the open-source version.
+static std::string FlagToEnvVar(const char* flag) {
+  const std::string full_flag =
+      (Message() << GTEST_FLAG_PREFIX_ << flag).GetString();
+
+  Message env_var;
+  for (size_t i = 0; i != full_flag.length(); i++) {
+    env_var << ToUpper(full_flag.c_str()[i]);
+  }
+
+  return env_var.GetString();
+}
+
+// Parses 'str' for a 32-bit signed integer.  If successful, writes
+// the result to *value and returns true; otherwise leaves *value
+// unchanged and returns false.
+bool ParseInt32(const Message& src_text, const char* str, Int32* value) {
+  // Parses the environment variable as a decimal integer.
+  char* end = NULL;
+  const long long_value = strtol(str, &end, 10);  // NOLINT
+
+  // Has strtol() consumed all characters in the string?
+  if (*end != '\0') {
+    // No - an invalid character was encountered.
+    Message msg;
+    msg << "WARNING: " << src_text
+        << " is expected to be a 32-bit integer, but actually"
+        << " has value \"" << str << "\".\n";
+    printf("%s", msg.GetString().c_str());
+    fflush(stdout);
+    return false;
+  }
+
+  // Is the parsed value in the range of an Int32?
+  const Int32 result = static_cast<Int32>(long_value);
+  if (long_value == LONG_MAX || long_value == LONG_MIN ||
+      // The parsed value overflows as a long.  (strtol() returns
+      // LONG_MAX or LONG_MIN when the input overflows.)
+      result != long_value
+      // The parsed value overflows as an Int32.
+      ) {
+    Message msg;
+    msg << "WARNING: " << src_text
+        << " is expected to be a 32-bit integer, but actually"
+        << " has value " << str << ", which overflows.\n";
+    printf("%s", msg.GetString().c_str());
+    fflush(stdout);
+    return false;
+  }
+
+  *value = result;
+  return true;
+}
+
+// Reads and returns the Boolean environment variable corresponding to
+// the given flag; if it's not set, returns default_value.
+//
+// The value is considered true iff it's not "0".
+bool BoolFromGTestEnv(const char* flag, bool default_value) {
+#if defined(GTEST_GET_BOOL_FROM_ENV_)
+  return GTEST_GET_BOOL_FROM_ENV_(flag, default_value);
+#endif  // defined(GTEST_GET_BOOL_FROM_ENV_)
+  const std::string env_var = FlagToEnvVar(flag);
+  const char* const string_value = posix::GetEnv(env_var.c_str());
+  return string_value == NULL ?
+      default_value : strcmp(string_value, "0") != 0;
+}
+
+// Reads and returns a 32-bit integer stored in the environment
+// variable corresponding to the given flag; if it isn't set or
+// doesn't represent a valid 32-bit integer, returns default_value.
+Int32 Int32FromGTestEnv(const char* flag, Int32 default_value) {
+#if defined(GTEST_GET_INT32_FROM_ENV_)
+  return GTEST_GET_INT32_FROM_ENV_(flag, default_value);
+#endif  // defined(GTEST_GET_INT32_FROM_ENV_)
+  const std::string env_var = FlagToEnvVar(flag);
+  const char* const string_value = posix::GetEnv(env_var.c_str());
+  if (string_value == NULL) {
+    // The environment variable is not set.
+    return default_value;
+  }
+
+  Int32 result = default_value;
+  if (!ParseInt32(Message() << "Environment variable " << env_var,
+                  string_value, &result)) {
+    printf("The default value %s is used.\n",
+           (Message() << default_value).GetString().c_str());
+    fflush(stdout);
+    return default_value;
+  }
+
+  return result;
+}
+
+// Reads and returns the string environment variable corresponding to
+// the given flag; if it's not set, returns default_value.
+std::string StringFromGTestEnv(const char* flag, const char* default_value) {
+#if defined(GTEST_GET_STRING_FROM_ENV_)
+  return GTEST_GET_STRING_FROM_ENV_(flag, default_value);
+#endif  // defined(GTEST_GET_STRING_FROM_ENV_)
+  const std::string env_var = FlagToEnvVar(flag);
+  const char* value = posix::GetEnv(env_var.c_str());
+  if (value != NULL) {
+    return value;
+  }
+
+  // As a special case for the 'output' flag, if GTEST_OUTPUT is not
+  // set, we look for XML_OUTPUT_FILE, which is set by the Bazel build
+  // system.  The value of XML_OUTPUT_FILE is a filename without the
+  // "xml:" prefix of GTEST_OUTPUT.
+  //
+  // The net priority order after flag processing is thus:
+  //   --gtest_output command line flag
+  //   GTEST_OUTPUT environment variable
+  //   XML_OUTPUT_FILE environment variable
+  //   'default_value'
+  if (strcmp(flag, "output") == 0) {
+    value = posix::GetEnv("XML_OUTPUT_FILE");
+    if (value != NULL) {
+      return std::string("xml:") + value;
+    }
+  }
+  return default_value;
+}
+
+}  // namespace internal
+}  // namespace testing
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Google Test - The Google C++ Testing Framework
+//
+// This file implements a universal value printer that can print a
+// value of any type T:
+//
+//   void ::testing::internal::UniversalPrinter<T>::Print(value, ostream_ptr);
+//
+// It uses the << operator when possible, and prints the bytes in the
+// object otherwise.  A user can override its behavior for a class
+// type Foo by defining either operator<<(::std::ostream&, const Foo&)
+// or void PrintTo(const Foo&, ::std::ostream*) in the namespace that
+// defines Foo.
+
+#include <ctype.h>
+#include <stdio.h>
+#include <cwchar>
+#include <ostream>  // NOLINT
+#include <string>
+
+namespace testing {
+
+namespace {
+
+using ::std::ostream;
+
+// Prints a segment of bytes in the given object.
+GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+void PrintByteSegmentInObjectTo(const unsigned char* obj_bytes, size_t start,
+                                size_t count, ostream* os) {
+  char text[5] = "";
+  for (size_t i = 0; i != count; i++) {
+    const size_t j = start + i;
+    if (i != 0) {
+      // Organizes the bytes into groups of 2 for easy parsing by
+      // human.
+      if ((j % 2) == 0)
+        *os << ' ';
+      else
+        *os << '-';
+    }
+    GTEST_SNPRINTF_(text, sizeof(text), "%02X", obj_bytes[j]);
+    *os << text;
+  }
+}
+
+// Prints the bytes in the given value to the given ostream.
+void PrintBytesInObjectToImpl(const unsigned char* obj_bytes, size_t count,
+                              ostream* os) {
+  // Tells the user how big the object is.
+  *os << count << "-byte object <";
+
+  const size_t kThreshold = 132;
+  const size_t kChunkSize = 64;
+  // If the object size is bigger than kThreshold, we'll have to omit
+  // some details by printing only the first and the last kChunkSize
+  // bytes.
+  // TODO(wan): let the user control the threshold using a flag.
+  if (count < kThreshold) {
+    PrintByteSegmentInObjectTo(obj_bytes, 0, count, os);
+  } else {
+    PrintByteSegmentInObjectTo(obj_bytes, 0, kChunkSize, os);
+    *os << " ... ";
+    // Rounds up to 2-byte boundary.
+    const size_t resume_pos = (count - kChunkSize + 1)/2*2;
+    PrintByteSegmentInObjectTo(obj_bytes, resume_pos, count - resume_pos, os);
+  }
+  *os << ">";
+}
+
+}  // namespace
+
+namespace internal2 {
+
+// Delegates to PrintBytesInObjectToImpl() to print the bytes in the
+// given object.  The delegation simplifies the implementation, which
+// uses the << operator and thus is easier done outside of the
+// ::testing::internal namespace, which contains a << operator that
+// sometimes conflicts with the one in STL.
+void PrintBytesInObjectTo(const unsigned char* obj_bytes, size_t count,
+                          ostream* os) {
+  PrintBytesInObjectToImpl(obj_bytes, count, os);
+}
+
+}  // namespace internal2
+
+namespace internal {
+
+// Depending on the value of a char (or wchar_t), we print it in one
+// of three formats:
+//   - as is if it's a printable ASCII (e.g. 'a', '2', ' '),
+//   - as a hexadecimal escape sequence (e.g. '\x7F'), or
+//   - as a special escape sequence (e.g. '\r', '\n').
+enum CharFormat {
+  kAsIs,
+  kHexEscape,
+  kSpecialEscape
+};
+
+// Returns true if c is a printable ASCII character.  We test the
+// value of c directly instead of calling isprint(), which is buggy on
+// Windows Mobile.
+inline bool IsPrintableAscii(wchar_t c) {
+  return 0x20 <= c && c <= 0x7E;
+}
+
+// Prints a wide or narrow char c as a character literal without the
+// quotes, escaping it when necessary; returns how c was formatted.
+// The template argument UnsignedChar is the unsigned version of Char,
+// which is the type of c.
+template <typename UnsignedChar, typename Char>
+static CharFormat PrintAsCharLiteralTo(Char c, ostream* os) {
+  switch (static_cast<wchar_t>(c)) {
+    case L'\0':
+      *os << "\\0";
+      break;
+    case L'\'':
+      *os << "\\'";
+      break;
+    case L'\\':
+      *os << "\\\\";
+      break;
+    case L'\a':
+      *os << "\\a";
+      break;
+    case L'\b':
+      *os << "\\b";
+      break;
+    case L'\f':
+      *os << "\\f";
+      break;
+    case L'\n':
+      *os << "\\n";
+      break;
+    case L'\r':
+      *os << "\\r";
+      break;
+    case L'\t':
+      *os << "\\t";
+      break;
+    case L'\v':
+      *os << "\\v";
+      break;
+    default:
+      if (IsPrintableAscii(c)) {
+        *os << static_cast<char>(c);
+        return kAsIs;
+      } else {
+        *os << "\\x" + String::FormatHexInt(static_cast<UnsignedChar>(c));
+        return kHexEscape;
+      }
+  }
+  return kSpecialEscape;
+}
+
+// Prints a wchar_t c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsStringLiteralTo(wchar_t c, ostream* os) {
+  switch (c) {
+    case L'\'':
+      *os << "'";
+      return kAsIs;
+    case L'"':
+      *os << "\\\"";
+      return kSpecialEscape;
+    default:
+      return PrintAsCharLiteralTo<wchar_t>(c, os);
+  }
+}
+
+// Prints a char c as if it's part of a string literal, escaping it when
+// necessary; returns how c was formatted.
+static CharFormat PrintAsStringLiteralTo(char c, ostream* os) {
+  return PrintAsStringLiteralTo(
+      static_cast<wchar_t>(static_cast<unsigned char>(c)), os);
+}
+
+// Prints a wide or narrow character c and its code.  '\0' is printed
+// as "'\\0'", other unprintable characters are also properly escaped
+// using the standard C++ escape sequence.  The template argument
+// UnsignedChar is the unsigned version of Char, which is the type of c.
+template <typename UnsignedChar, typename Char>
+void PrintCharAndCodeTo(Char c, ostream* os) {
+  // First, print c as a literal in the most readable form we can find.
+  *os << ((sizeof(c) > 1) ? "L'" : "'");
+  const CharFormat format = PrintAsCharLiteralTo<UnsignedChar>(c, os);
+  *os << "'";
+
+  // To aid user debugging, we also print c's code in decimal, unless
+  // it's 0 (in which case c was printed as '\\0', making the code
+  // obvious).
+  if (c == 0)
+    return;
+  *os << " (" << static_cast<int>(c);
+
+  // For more convenience, we print c's code again in hexadecimal,
+  // unless c was already printed in the form '\x##' or the code is in
+  // [1, 9].
+  if (format == kHexEscape || (1 <= c && c <= 9)) {
+    // Do nothing.
+  } else {
+    *os << ", 0x" << String::FormatHexInt(static_cast<UnsignedChar>(c));
+  }
+  *os << ")";
+}
+
+void PrintTo(unsigned char c, ::std::ostream* os) {
+  PrintCharAndCodeTo<unsigned char>(c, os);
+}
+void PrintTo(signed char c, ::std::ostream* os) {
+  PrintCharAndCodeTo<unsigned char>(c, os);
+}
+
+// Prints a wchar_t as a symbol if it is printable or as its internal
+// code otherwise and also as its code.  L'\0' is printed as "L'\\0'".
+void PrintTo(wchar_t wc, ostream* os) {
+  PrintCharAndCodeTo<wchar_t>(wc, os);
+}
+
+// Prints the given array of characters to the ostream.  CharType must be either
+// char or wchar_t.
+// The array starts at begin, the length is len, it may include '\0' characters
+// and may not be NUL-terminated.
+template <typename CharType>
+GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+static void PrintCharsAsStringTo(
+    const CharType* begin, size_t len, ostream* os) {
+  const char* const kQuoteBegin = sizeof(CharType) == 1 ? "\"" : "L\"";
+  *os << kQuoteBegin;
+  bool is_previous_hex = false;
+  for (size_t index = 0; index < len; ++index) {
+    const CharType cur = begin[index];
+    if (is_previous_hex && IsXDigit(cur)) {
+      // Previous character is of '\x..' form and this character can be
+      // interpreted as another hexadecimal digit in its number. Break string to
+      // disambiguate.
+      *os << "\" " << kQuoteBegin;
+    }
+    is_previous_hex = PrintAsStringLiteralTo(cur, os) == kHexEscape;
+  }
+  *os << "\"";
+}
+
+// Prints a (const) char/wchar_t array of 'len' elements, starting at address
+// 'begin'.  CharType must be either char or wchar_t.
+template <typename CharType>
+GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+static void UniversalPrintCharArray(
+    const CharType* begin, size_t len, ostream* os) {
+  // The code
+  //   const char kFoo[] = "foo";
+  // generates an array of 4, not 3, elements, with the last one being '\0'.
+  //
+  // Therefore when printing a char array, we don't print the last element if
+  // it's '\0', such that the output matches the string literal as it's
+  // written in the source code.
+  if (len > 0 && begin[len - 1] == '\0') {
+    PrintCharsAsStringTo(begin, len - 1, os);
+    return;
+  }
+
+  // If, however, the last element in the array is not '\0', e.g.
+  //    const char kFoo[] = { 'f', 'o', 'o' };
+  // we must print the entire array.  We also print a message to indicate
+  // that the array is not NUL-terminated.
+  PrintCharsAsStringTo(begin, len, os);
+  *os << " (no terminating NUL)";
+}
+
+// Prints a (const) char array of 'len' elements, starting at address 'begin'.
+void UniversalPrintArray(const char* begin, size_t len, ostream* os) {
+  UniversalPrintCharArray(begin, len, os);
+}
+
+// Prints a (const) wchar_t array of 'len' elements, starting at address
+// 'begin'.
+void UniversalPrintArray(const wchar_t* begin, size_t len, ostream* os) {
+  UniversalPrintCharArray(begin, len, os);
+}
+
+// Prints the given C string to the ostream.
+void PrintTo(const char* s, ostream* os) {
+  if (s == NULL) {
+    *os << "NULL";
+  } else {
+    *os << ImplicitCast_<const void*>(s) << " pointing to ";
+    PrintCharsAsStringTo(s, strlen(s), os);
+  }
+}
+
+// MSVC compiler can be configured to define whar_t as a typedef
+// of unsigned short. Defining an overload for const wchar_t* in that case
+// would cause pointers to unsigned shorts be printed as wide strings,
+// possibly accessing more memory than intended and causing invalid
+// memory accesses. MSVC defines _NATIVE_WCHAR_T_DEFINED symbol when
+// wchar_t is implemented as a native type.
+#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED)
+// Prints the given wide C string to the ostream.
+void PrintTo(const wchar_t* s, ostream* os) {
+  if (s == NULL) {
+    *os << "NULL";
+  } else {
+    *os << ImplicitCast_<const void*>(s) << " pointing to ";
+    PrintCharsAsStringTo(s, std::wcslen(s), os);
+  }
+}
+#endif  // wchar_t is native
+
+// Prints a ::string object.
+#if GTEST_HAS_GLOBAL_STRING
+void PrintStringTo(const ::string& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+void PrintStringTo(const ::std::string& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+
+// Prints a ::wstring object.
+#if GTEST_HAS_GLOBAL_WSTRING
+void PrintWideStringTo(const ::wstring& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+#if GTEST_HAS_STD_WSTRING
+void PrintWideStringTo(const ::std::wstring& s, ostream* os) {
+  PrintCharsAsStringTo(s.data(), s.size(), os);
+}
+#endif  // GTEST_HAS_STD_WSTRING
+
+}  // namespace internal
+
+}  // namespace testing
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+// The Google C++ Testing Framework (Google Test)
+
+
+// Indicates that this translation unit is part of Google Test's
+// implementation.  It must come before gtest-internal-inl.h is
+// included, or there will be a compiler error.  This trick exists to
+// prevent the accidental inclusion of gtest-internal-inl.h in the
+// user's code.
+#define GTEST_IMPLEMENTATION_ 1
+#undef GTEST_IMPLEMENTATION_
+
+namespace testing {
+
+using internal::GetUnitTestImpl;
+
+// Gets the summary of the failure message by omitting the stack trace
+// in it.
+std::string TestPartResult::ExtractSummary(const char* message) {
+  const char* const stack_trace = strstr(message, internal::kStackTraceMarker);
+  return stack_trace == NULL ? message :
+      std::string(message, stack_trace);
+}
+
+// Prints a TestPartResult object.
+std::ostream& operator<<(std::ostream& os, const TestPartResult& result) {
+  return os
+      << result.file_name() << ":" << result.line_number() << ": "
+      << (result.type() == TestPartResult::kSuccess ? "Success" :
+          result.type() == TestPartResult::kFatalFailure ? "Fatal failure" :
+          "Non-fatal failure") << ":\n"
+      << result.message() << std::endl;
+}
+
+// Appends a TestPartResult to the array.
+void TestPartResultArray::Append(const TestPartResult& result) {
+  array_.push_back(result);
+}
+
+// Returns the TestPartResult at the given index (0-based).
+const TestPartResult& TestPartResultArray::GetTestPartResult(int index) const {
+  if (index < 0 || index >= size()) {
+    printf("\nInvalid index (%d) into TestPartResultArray.\n", index);
+    internal::posix::Abort();
+  }
+
+  return array_[index];
+}
+
+// Returns the number of TestPartResult objects in the array.
+int TestPartResultArray::size() const {
+  return static_cast<int>(array_.size());
+}
+
+namespace internal {
+
+HasNewFatalFailureHelper::HasNewFatalFailureHelper()
+    : has_new_fatal_failure_(false),
+      original_reporter_(GetUnitTestImpl()->
+                         GetTestPartResultReporterForCurrentThread()) {
+  GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(this);
+}
+
+HasNewFatalFailureHelper::~HasNewFatalFailureHelper() {
+  GetUnitTestImpl()->SetTestPartResultReporterForCurrentThread(
+      original_reporter_);
+}
+
+void HasNewFatalFailureHelper::ReportTestPartResult(
+    const TestPartResult& result) {
+  if (result.fatally_failed())
+    has_new_fatal_failure_ = true;
+  original_reporter_->ReportTestPartResult(result);
+}
+
+}  // namespace internal
+
+}  // namespace testing
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+
+namespace testing {
+namespace internal {
+
+#if GTEST_HAS_TYPED_TEST_P
+
+// Skips to the first non-space char in str. Returns an empty string if str
+// contains only whitespace characters.
+static const char* SkipSpaces(const char* str) {
+  while (IsSpace(*str))
+    str++;
+  return str;
+}
+
+static std::vector<std::string> SplitIntoTestNames(const char* src) {
+  std::vector<std::string> name_vec;
+  src = SkipSpaces(src);
+  for (; src != NULL; src = SkipComma(src)) {
+    name_vec.push_back(StripTrailingSpaces(GetPrefixUntilComma(src)));
+  }
+  return name_vec;
+}
+
+// Verifies that registered_tests match the test names in
+// registered_tests_; returns registered_tests if successful, or
+// aborts the program otherwise.
+const char* TypedTestCasePState::VerifyRegisteredTestNames(
+    const char* file, int line, const char* registered_tests) {
+  typedef RegisteredTestsMap::const_iterator RegisteredTestIter;
+  registered_ = true;
+
+  std::vector<std::string> name_vec = SplitIntoTestNames(registered_tests);
+
+  Message errors;
+
+  std::set<std::string> tests;
+  for (std::vector<std::string>::const_iterator name_it = name_vec.begin();
+       name_it != name_vec.end(); ++name_it) {
+    const std::string& name = *name_it;
+    if (tests.count(name) != 0) {
+      errors << "Test " << name << " is listed more than once.\n";
+      continue;
+    }
+
+    bool found = false;
+    for (RegisteredTestIter it = registered_tests_.begin();
+         it != registered_tests_.end();
+         ++it) {
+      if (name == it->first) {
+        found = true;
+        break;
+      }
+    }
+
+    if (found) {
+      tests.insert(name);
+    } else {
+      errors << "No test named " << name
+             << " can be found in this test case.\n";
+    }
+  }
+
+  for (RegisteredTestIter it = registered_tests_.begin();
+       it != registered_tests_.end();
+       ++it) {
+    if (tests.count(it->first) == 0) {
+      errors << "You forgot to list test " << it->first << ".\n";
+    }
+  }
+
+  const std::string& errors_str = errors.GetString();
+  if (errors_str != "") {
+    fprintf(stderr, "%s %s", FormatFileLocation(file, line).c_str(),
+            errors_str.c_str());
+    fflush(stderr);
+    posix::Abort();
+  }
+
+  return registered_tests;
+}
+
+#endif  // GTEST_HAS_TYPED_TEST_P
+
+}  // namespace internal
+}  // namespace testing
diff --git a/test/gtest_fused/gtest/gtest.h b/test/gtest_fused/gtest/gtest.h
new file mode 100644
index 0000000..55c43d6
--- /dev/null
+++ b/test/gtest_fused/gtest/gtest.h
@@ -0,0 +1,22635 @@
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the public API for Google Test.  It should be
+// included by any test program that uses Google Test.
+//
+// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
+// leave some internal implementation details in this header file.
+// They are clearly marked by comments like this:
+//
+//   // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+//
+// Such code is NOT meant to be used by a user directly, and is subject
+// to CHANGE WITHOUT NOTICE.  Therefore DO NOT DEPEND ON IT in a user
+// program!
+//
+// Acknowledgment: Google Test borrowed the idea of automatic test
+// registration from Barthelemy Dagenais' (barthelemy@prologique.com)
+// easyUnit framework.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_H_
+
+#include <limits>
+#include <ostream>
+#include <vector>
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file declares functions and macros used internally by
+// Google Test.  They are subject to change without notice.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan)
+//
+// Low-level types and utilities for porting Google Test to various
+// platforms.  All macros ending with _ and symbols defined in an
+// internal namespace are subject to change without notice.  Code
+// outside Google Test MUST NOT USE THEM DIRECTLY.  Macros that don't
+// end with _ are part of Google Test's public API and can be used by
+// code outside Google Test.
+//
+// This file is fundamental to Google Test.  All other Google Test source
+// files are expected to #include this.  Therefore, it cannot #include
+// any other Google Test header.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+
+// Environment-describing macros
+// -----------------------------
+//
+// Google Test can be used in many different environments.  Macros in
+// this section tell Google Test what kind of environment it is being
+// used in, such that Google Test can provide environment-specific
+// features and implementations.
+//
+// Google Test tries to automatically detect the properties of its
+// environment, so users usually don't need to worry about these
+// macros.  However, the automatic detection is not perfect.
+// Sometimes it's necessary for a user to define some of the following
+// macros in the build script to override Google Test's decisions.
+//
+// If the user doesn't define a macro in the list, Google Test will
+// provide a default definition.  After this header is #included, all
+// macros in this list will be defined to either 1 or 0.
+//
+// Notes to maintainers:
+//   - Each macro here is a user-tweakable knob; do not grow the list
+//     lightly.
+//   - Use #if to key off these macros.  Don't use #ifdef or "#if
+//     defined(...)", which will not work as these macros are ALWAYS
+//     defined.
+//
+//   GTEST_HAS_CLONE          - Define it to 1/0 to indicate that clone(2)
+//                              is/isn't available.
+//   GTEST_HAS_EXCEPTIONS     - Define it to 1/0 to indicate that exceptions
+//                              are enabled.
+//   GTEST_HAS_GLOBAL_STRING  - Define it to 1/0 to indicate that ::string
+//                              is/isn't available (some systems define
+//                              ::string, which is different to std::string).
+//   GTEST_HAS_GLOBAL_WSTRING - Define it to 1/0 to indicate that ::string
+//                              is/isn't available (some systems define
+//                              ::wstring, which is different to std::wstring).
+//   GTEST_HAS_POSIX_RE       - Define it to 1/0 to indicate that POSIX regular
+//                              expressions are/aren't available.
+//   GTEST_HAS_PTHREAD        - Define it to 1/0 to indicate that <pthread.h>
+//                              is/isn't available.
+//   GTEST_HAS_RTTI           - Define it to 1/0 to indicate that RTTI is/isn't
+//                              enabled.
+//   GTEST_HAS_STD_WSTRING    - Define it to 1/0 to indicate that
+//                              std::wstring does/doesn't work (Google Test can
+//                              be used where std::wstring is unavailable).
+//   GTEST_HAS_TR1_TUPLE      - Define it to 1/0 to indicate tr1::tuple
+//                              is/isn't available.
+//   GTEST_HAS_SEH            - Define it to 1/0 to indicate whether the
+//                              compiler supports Microsoft's "Structured
+//                              Exception Handling".
+//   GTEST_HAS_STREAM_REDIRECTION
+//                            - Define it to 1/0 to indicate whether the
+//                              platform supports I/O stream redirection using
+//                              dup() and dup2().
+//   GTEST_USE_OWN_TR1_TUPLE  - Define it to 1/0 to indicate whether Google
+//                              Test's own tr1 tuple implementation should be
+//                              used.  Unused when the user sets
+//                              GTEST_HAS_TR1_TUPLE to 0.
+//   GTEST_LANG_CXX11         - Define it to 1/0 to indicate that Google Test
+//                              is building in C++11/C++98 mode.
+//   GTEST_LINKED_AS_SHARED_LIBRARY
+//                            - Define to 1 when compiling tests that use
+//                              Google Test as a shared library (known as
+//                              DLL on Windows).
+//   GTEST_CREATE_SHARED_LIBRARY
+//                            - Define to 1 when compiling Google Test itself
+//                              as a shared library.
+
+// Platform-indicating macros
+// --------------------------
+//
+// Macros indicating the platform on which Google Test is being used
+// (a macro is defined to 1 if compiled on the given platform;
+// otherwise UNDEFINED -- it's never defined to 0.).  Google Test
+// defines these macros automatically.  Code outside Google Test MUST
+// NOT define them.
+//
+//   GTEST_OS_AIX      - IBM AIX
+//   GTEST_OS_CYGWIN   - Cygwin
+//   GTEST_OS_FREEBSD  - FreeBSD
+//   GTEST_OS_HPUX     - HP-UX
+//   GTEST_OS_LINUX    - Linux
+//     GTEST_OS_LINUX_ANDROID - Google Android
+//   GTEST_OS_MAC      - Mac OS X
+//     GTEST_OS_IOS    - iOS
+//   GTEST_OS_NACL     - Google Native Client (NaCl)
+//   GTEST_OS_NETBSD   - NetBSD
+//   GTEST_OS_OPENBSD  - OpenBSD
+//   GTEST_OS_QNX      - QNX
+//   GTEST_OS_SOLARIS  - Sun Solaris
+//   GTEST_OS_SYMBIAN  - Symbian
+//   GTEST_OS_WINDOWS  - Windows (Desktop, MinGW, or Mobile)
+//     GTEST_OS_WINDOWS_DESKTOP  - Windows Desktop
+//     GTEST_OS_WINDOWS_MINGW    - MinGW
+//     GTEST_OS_WINDOWS_MOBILE   - Windows Mobile
+//     GTEST_OS_WINDOWS_PHONE    - Windows Phone
+//     GTEST_OS_WINDOWS_RT       - Windows Store App/WinRT
+//   GTEST_OS_ZOS      - z/OS
+//
+// Among the platforms, Cygwin, Linux, Max OS X, and Windows have the
+// most stable support.  Since core members of the Google Test project
+// don't have access to other platforms, support for them may be less
+// stable.  If you notice any problems on your platform, please notify
+// googletestframework@googlegroups.com (patches for fixing them are
+// even more welcome!).
+//
+// It is possible that none of the GTEST_OS_* macros are defined.
+
+// Feature-indicating macros
+// -------------------------
+//
+// Macros indicating which Google Test features are available (a macro
+// is defined to 1 if the corresponding feature is supported;
+// otherwise UNDEFINED -- it's never defined to 0.).  Google Test
+// defines these macros automatically.  Code outside Google Test MUST
+// NOT define them.
+//
+// These macros are public so that portable tests can be written.
+// Such tests typically surround code using a feature with an #if
+// which controls that code.  For example:
+//
+// #if GTEST_HAS_DEATH_TEST
+//   EXPECT_DEATH(DoSomethingDeadly());
+// #endif
+//
+//   GTEST_HAS_COMBINE      - the Combine() function (for value-parameterized
+//                            tests)
+//   GTEST_HAS_DEATH_TEST   - death tests
+//   GTEST_HAS_PARAM_TEST   - value-parameterized tests
+//   GTEST_HAS_TYPED_TEST   - typed tests
+//   GTEST_HAS_TYPED_TEST_P - type-parameterized tests
+//   GTEST_IS_THREADSAFE    - Google Test is thread-safe.
+//   GTEST_USES_POSIX_RE    - enhanced POSIX regex is used. Do not confuse with
+//                            GTEST_HAS_POSIX_RE (see above) which users can
+//                            define themselves.
+//   GTEST_USES_SIMPLE_RE   - our own simple regex is used;
+//                            the above two are mutually exclusive.
+//   GTEST_CAN_COMPARE_NULL - accepts untyped NULL in EXPECT_EQ().
+
+// Misc public macros
+// ------------------
+//
+//   GTEST_FLAG(flag_name)  - references the variable corresponding to
+//                            the given Google Test flag.
+
+// Internal utilities
+// ------------------
+//
+// The following macros and utilities are for Google Test's INTERNAL
+// use only.  Code outside Google Test MUST NOT USE THEM DIRECTLY.
+//
+// Macros for basic C++ coding:
+//   GTEST_AMBIGUOUS_ELSE_BLOCKER_ - for disabling a gcc warning.
+//   GTEST_ATTRIBUTE_UNUSED_  - declares that a class' instances or a
+//                              variable don't have to be used.
+//   GTEST_DISALLOW_ASSIGN_   - disables operator=.
+//   GTEST_DISALLOW_COPY_AND_ASSIGN_ - disables copy ctor and operator=.
+//   GTEST_MUST_USE_RESULT_   - declares that a function's result must be used.
+//   GTEST_INTENTIONAL_CONST_COND_PUSH_ - start code section where MSVC C4127 is
+//                                        suppressed (constant conditional).
+//   GTEST_INTENTIONAL_CONST_COND_POP_  - finish code section where MSVC C4127
+//                                        is suppressed.
+//
+// C++11 feature wrappers:
+//
+//   testing::internal::move  - portability wrapper for std::move.
+//
+// Synchronization:
+//   Mutex, MutexLock, ThreadLocal, GetThreadCount()
+//                            - synchronization primitives.
+//
+// Template meta programming:
+//   is_pointer     - as in TR1; needed on Symbian and IBM XL C/C++ only.
+//   IteratorTraits - partial implementation of std::iterator_traits, which
+//                    is not available in libCstd when compiled with Sun C++.
+//
+// Smart pointers:
+//   scoped_ptr     - as in TR2.
+//
+// Regular expressions:
+//   RE             - a simple regular expression class using the POSIX
+//                    Extended Regular Expression syntax on UNIX-like
+//                    platforms, or a reduced regular exception syntax on
+//                    other platforms, including Windows.
+//
+// Logging:
+//   GTEST_LOG_()   - logs messages at the specified severity level.
+//   LogToStderr()  - directs all log messages to stderr.
+//   FlushInfoLog() - flushes informational log messages.
+//
+// Stdout and stderr capturing:
+//   CaptureStdout()     - starts capturing stdout.
+//   GetCapturedStdout() - stops capturing stdout and returns the captured
+//                         string.
+//   CaptureStderr()     - starts capturing stderr.
+//   GetCapturedStderr() - stops capturing stderr and returns the captured
+//                         string.
+//
+// Integer types:
+//   TypeWithSize   - maps an integer to a int type.
+//   Int32, UInt32, Int64, UInt64, TimeInMillis
+//                  - integers of known sizes.
+//   BiggestInt     - the biggest signed integer type.
+//
+// Command-line utilities:
+//   GTEST_DECLARE_*()  - declares a flag.
+//   GTEST_DEFINE_*()   - defines a flag.
+//   GetInjectableArgvs() - returns the command line as a vector of strings.
+//
+// Environment variable utilities:
+//   GetEnv()             - gets the value of an environment variable.
+//   BoolFromGTestEnv()   - parses a bool environment variable.
+//   Int32FromGTestEnv()  - parses an Int32 environment variable.
+//   StringFromGTestEnv() - parses a string environment variable.
+
+#include <ctype.h>   // for isspace, etc
+#include <stddef.h>  // for ptrdiff_t
+#include <stdlib.h>
+#include <stdio.h>
+#include <string.h>
+#ifndef _WIN32_WCE
+# include <sys/types.h>
+# include <sys/stat.h>
+#endif  // !_WIN32_WCE
+
+#if defined __APPLE__
+# include <AvailabilityMacros.h>
+# include <TargetConditionals.h>
+#endif
+
+#include <algorithm>  // NOLINT
+#include <iostream>  // NOLINT
+#include <sstream>  // NOLINT
+#include <string>  // NOLINT
+#include <utility>
+#include <vector>  // NOLINT
+
+// Copyright 2015, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the GTEST_OS_* macro.
+// It is separate from gtest-port.h so that custom/gtest-port.h can include it.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_
+
+// Determines the platform on which Google Test is compiled.
+#ifdef __CYGWIN__
+# define GTEST_OS_CYGWIN 1
+#elif defined __SYMBIAN32__
+# define GTEST_OS_SYMBIAN 1
+#elif defined _WIN32
+# define GTEST_OS_WINDOWS 1
+# ifdef _WIN32_WCE
+#  define GTEST_OS_WINDOWS_MOBILE 1
+# elif defined(__MINGW__) || defined(__MINGW32__)
+#  define GTEST_OS_WINDOWS_MINGW 1
+# elif defined(WINAPI_FAMILY)
+#  include <winapifamily.h>
+#  if WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_DESKTOP)
+#   define GTEST_OS_WINDOWS_DESKTOP 1
+#  elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_PHONE_APP)
+#   define GTEST_OS_WINDOWS_PHONE 1
+#  elif WINAPI_FAMILY_PARTITION(WINAPI_PARTITION_APP)
+#   define GTEST_OS_WINDOWS_RT 1
+#  else
+// WINAPI_FAMILY defined but no known partition matched.
+// Default to desktop.
+#   define GTEST_OS_WINDOWS_DESKTOP 1
+#  endif
+# else
+#  define GTEST_OS_WINDOWS_DESKTOP 1
+# endif  // _WIN32_WCE
+#elif defined __APPLE__
+# define GTEST_OS_MAC 1
+# if TARGET_OS_IPHONE
+#  define GTEST_OS_IOS 1
+# endif
+#elif defined __FreeBSD__
+# define GTEST_OS_FREEBSD 1
+#elif defined __linux__
+# define GTEST_OS_LINUX 1
+# if defined __ANDROID__
+#  define GTEST_OS_LINUX_ANDROID 1
+# endif
+#elif defined __MVS__
+# define GTEST_OS_ZOS 1
+#elif defined(__sun) && defined(__SVR4)
+# define GTEST_OS_SOLARIS 1
+#elif defined(_AIX)
+# define GTEST_OS_AIX 1
+#elif defined(__hpux)
+# define GTEST_OS_HPUX 1
+#elif defined __native_client__
+# define GTEST_OS_NACL 1
+#elif defined __NetBSD__
+# define GTEST_OS_NETBSD 1
+#elif defined __OpenBSD__
+# define GTEST_OS_OPENBSD 1
+#elif defined __QNX__
+# define GTEST_OS_QNX 1
+#endif  // __CYGWIN__
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_ARCH_H_
+// Copyright 2015, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Injection point for custom user configurations.
+// The following macros can be defined:
+//
+//   Flag related macros:
+//     GTEST_FLAG(flag_name)
+//     GTEST_USE_OWN_FLAGFILE_FLAG_  - Define to 0 when the system provides its
+//                                     own flagfile flag parsing.
+//     GTEST_DECLARE_bool_(name)
+//     GTEST_DECLARE_int32_(name)
+//     GTEST_DECLARE_string_(name)
+//     GTEST_DEFINE_bool_(name, default_val, doc)
+//     GTEST_DEFINE_int32_(name, default_val, doc)
+//     GTEST_DEFINE_string_(name, default_val, doc)
+//
+//   Test filtering:
+//     GTEST_TEST_FILTER_ENV_VAR_ - The name of an environment variable that
+//                                  will be used if --GTEST_FLAG(test_filter)
+//                                  is not provided.
+//
+//   Logging:
+//     GTEST_LOG_(severity)
+//     GTEST_CHECK_(condition)
+//     Functions LogToStderr() and FlushInfoLog() have to be provided too.
+//
+//   Threading:
+//     GTEST_HAS_NOTIFICATION_ - Enabled if Notification is already provided.
+//     GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ - Enabled if Mutex and ThreadLocal are
+//                                         already provided.
+//     Must also provide GTEST_DECLARE_STATIC_MUTEX_(mutex) and
+//     GTEST_DEFINE_STATIC_MUTEX_(mutex)
+//
+//     GTEST_EXCLUSIVE_LOCK_REQUIRED_(locks)
+//     GTEST_LOCK_EXCLUDED_(locks)
+//
+//   Exporting API symbols:
+//     GTEST_API_ - Specifier for exported symbols.
+//
+// ** Custom implementation starts here **
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PORT_H_
+
+#if !defined(GTEST_DEV_EMAIL_)
+# define GTEST_DEV_EMAIL_ "googletestframework@@googlegroups.com"
+# define GTEST_FLAG_PREFIX_ "gtest_"
+# define GTEST_FLAG_PREFIX_DASH_ "gtest-"
+# define GTEST_FLAG_PREFIX_UPPER_ "GTEST_"
+# define GTEST_NAME_ "Google Test"
+# define GTEST_PROJECT_URL_ "https://github.com/google/googletest/"
+#endif  // !defined(GTEST_DEV_EMAIL_)
+
+#if !defined(GTEST_INIT_GOOGLE_TEST_NAME_)
+# define GTEST_INIT_GOOGLE_TEST_NAME_ "testing::InitGoogleTest"
+#endif  // !defined(GTEST_INIT_GOOGLE_TEST_NAME_)
+
+// Determines the version of gcc that is used to compile this.
+#ifdef __GNUC__
+// 40302 means version 4.3.2.
+# define GTEST_GCC_VER_ \
+    (__GNUC__*10000 + __GNUC_MINOR__*100 + __GNUC_PATCHLEVEL__)
+#endif  // __GNUC__
+
+// Macros for disabling Microsoft Visual C++ warnings.
+//
+//   GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 4385)
+//   /* code that triggers warnings C4800 and C4385 */
+//   GTEST_DISABLE_MSC_WARNINGS_POP_()
+#if _MSC_VER >= 1500
+# define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings) \
+    __pragma(warning(push))                        \
+    __pragma(warning(disable: warnings))
+# define GTEST_DISABLE_MSC_WARNINGS_POP_()          \
+    __pragma(warning(pop))
+#else
+// Older versions of MSVC don't have __pragma.
+# define GTEST_DISABLE_MSC_WARNINGS_PUSH_(warnings)
+# define GTEST_DISABLE_MSC_WARNINGS_POP_()
+#endif
+
+#ifndef GTEST_LANG_CXX11
+// gcc and clang define __GXX_EXPERIMENTAL_CXX0X__ when
+// -std={c,gnu}++{0x,11} is passed.  The C++11 standard specifies a
+// value for __cplusplus, and recent versions of clang, gcc, and
+// probably other compilers set that too in C++11 mode.
+# if __GXX_EXPERIMENTAL_CXX0X__ || __cplusplus >= 201103L
+// Compiling in at least C++11 mode.
+#  define GTEST_LANG_CXX11 1
+# else
+#  define GTEST_LANG_CXX11 0
+# endif
+#endif
+
+// Distinct from C++11 language support, some environments don't provide
+// proper C++11 library support. Notably, it's possible to build in
+// C++11 mode when targeting Mac OS X 10.6, which has an old libstdc++
+// with no C++11 support.
+//
+// libstdc++ has sufficient C++11 support as of GCC 4.6.0, __GLIBCXX__
+// 20110325, but maintenance releases in the 4.4 and 4.5 series followed
+// this date, so check for those versions by their date stamps.
+// https://gcc.gnu.org/onlinedocs/libstdc++/manual/abi.html#abi.versioning
+#if GTEST_LANG_CXX11 && \
+    (!defined(__GLIBCXX__) || ( \
+        __GLIBCXX__ >= 20110325ul &&  /* GCC >= 4.6.0 */ \
+        /* Blacklist of patch releases of older branches: */ \
+        __GLIBCXX__ != 20110416ul &&  /* GCC 4.4.6 */ \
+        __GLIBCXX__ != 20120313ul &&  /* GCC 4.4.7 */ \
+        __GLIBCXX__ != 20110428ul &&  /* GCC 4.5.3 */ \
+        __GLIBCXX__ != 20120702ul))   /* GCC 4.5.4 */
+# define GTEST_STDLIB_CXX11 1
+#endif
+
+// Only use C++11 library features if the library provides them.
+#if GTEST_STDLIB_CXX11
+# define GTEST_HAS_STD_BEGIN_AND_END_ 1
+# define GTEST_HAS_STD_FORWARD_LIST_ 1
+# define GTEST_HAS_STD_FUNCTION_ 1
+# define GTEST_HAS_STD_INITIALIZER_LIST_ 1
+# define GTEST_HAS_STD_MOVE_ 1
+# define GTEST_HAS_STD_SHARED_PTR_ 1
+# define GTEST_HAS_STD_TYPE_TRAITS_ 1
+# define GTEST_HAS_STD_UNIQUE_PTR_ 1
+#endif
+
+// C++11 specifies that <tuple> provides std::tuple.
+// Some platforms still might not have it, however.
+#if GTEST_LANG_CXX11
+# define GTEST_HAS_STD_TUPLE_ 1
+# if defined(__clang__)
+// Inspired by http://clang.llvm.org/docs/LanguageExtensions.html#__has_include
+#  if defined(__has_include) && !__has_include(<tuple>)
+#   undef GTEST_HAS_STD_TUPLE_
+#  endif
+# elif defined(_MSC_VER)
+// Inspired by boost/config/stdlib/dinkumware.hpp
+#  if defined(_CPPLIB_VER) && _CPPLIB_VER < 520
+#   undef GTEST_HAS_STD_TUPLE_
+#  endif
+# elif defined(__GLIBCXX__)
+// Inspired by boost/config/stdlib/libstdcpp3.hpp,
+// http://gcc.gnu.org/gcc-4.2/changes.html and
+// http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#manual.intro.status.standard.200x
+#  if __GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 2)
+#   undef GTEST_HAS_STD_TUPLE_
+#  endif
+# endif
+#endif
+
+// Brings in definitions for functions used in the testing::internal::posix
+// namespace (read, write, close, chdir, isatty, stat). We do not currently
+// use them on Windows Mobile.
+#if GTEST_OS_WINDOWS
+# if !GTEST_OS_WINDOWS_MOBILE
+#  include <direct.h>
+#  include <io.h>
+# endif
+// In order to avoid having to include <windows.h>, use forward declaration
+#if GTEST_OS_WINDOWS_MINGW && !defined(__MINGW64_VERSION_MAJOR)
+// MinGW defined _CRITICAL_SECTION and _RTL_CRITICAL_SECTION as two
+// separate (equivalent) structs, instead of using typedef
+typedef struct _CRITICAL_SECTION GTEST_CRITICAL_SECTION;
+#else
+// Assume CRITICAL_SECTION is a typedef of _RTL_CRITICAL_SECTION.
+// This assumption is verified by
+// WindowsTypesTest.CRITICAL_SECTIONIs_RTL_CRITICAL_SECTION.
+typedef struct _RTL_CRITICAL_SECTION GTEST_CRITICAL_SECTION;
+#endif
+#else
+// This assumes that non-Windows OSes provide unistd.h. For OSes where this
+// is not the case, we need to include headers that provide the functions
+// mentioned above.
+# include <unistd.h>
+# include <strings.h>
+#endif  // GTEST_OS_WINDOWS
+
+#if GTEST_OS_LINUX_ANDROID
+// Used to define __ANDROID_API__ matching the target NDK API level.
+#  include <android/api-level.h>  // NOLINT
+#endif
+
+// Defines this to true iff Google Test can use POSIX regular expressions.
+#ifndef GTEST_HAS_POSIX_RE
+# if GTEST_OS_LINUX_ANDROID
+// On Android, <regex.h> is only available starting with Gingerbread.
+#  define GTEST_HAS_POSIX_RE (__ANDROID_API__ >= 9)
+# else
+#  define GTEST_HAS_POSIX_RE (!GTEST_OS_WINDOWS)
+# endif
+#endif
+
+#if GTEST_USES_PCRE
+// The appropriate headers have already been included.
+
+#elif GTEST_HAS_POSIX_RE
+
+// On some platforms, <regex.h> needs someone to define size_t, and
+// won't compile otherwise.  We can #include it here as we already
+// included <stdlib.h>, which is guaranteed to define size_t through
+// <stddef.h>.
+# include <regex.h>  // NOLINT
+
+# define GTEST_USES_POSIX_RE 1
+
+#elif GTEST_OS_WINDOWS
+
+// <regex.h> is not available on Windows.  Use our own simple regex
+// implementation instead.
+# define GTEST_USES_SIMPLE_RE 1
+
+#else
+
+// <regex.h> may not be available on this platform.  Use our own
+// simple regex implementation instead.
+# define GTEST_USES_SIMPLE_RE 1
+
+#endif  // GTEST_USES_PCRE
+
+#ifndef GTEST_HAS_EXCEPTIONS
+// The user didn't tell us whether exceptions are enabled, so we need
+// to figure it out.
+# if defined(_MSC_VER) || defined(__BORLANDC__)
+// MSVC's and C++Builder's implementations of the STL use the _HAS_EXCEPTIONS
+// macro to enable exceptions, so we'll do the same.
+// Assumes that exceptions are enabled by default.
+#  ifndef _HAS_EXCEPTIONS
+#   define _HAS_EXCEPTIONS 1
+#  endif  // _HAS_EXCEPTIONS
+#  define GTEST_HAS_EXCEPTIONS _HAS_EXCEPTIONS
+# elif defined(__clang__)
+// clang defines __EXCEPTIONS iff exceptions are enabled before clang 220714,
+// but iff cleanups are enabled after that. In Obj-C++ files, there can be
+// cleanups for ObjC exceptions which also need cleanups, even if C++ exceptions
+// are disabled. clang has __has_feature(cxx_exceptions) which checks for C++
+// exceptions starting at clang r206352, but which checked for cleanups prior to
+// that. To reliably check for C++ exception availability with clang, check for
+// __EXCEPTIONS && __has_feature(cxx_exceptions).
+#  define GTEST_HAS_EXCEPTIONS (__EXCEPTIONS && __has_feature(cxx_exceptions))
+# elif defined(__GNUC__) && __EXCEPTIONS
+// gcc defines __EXCEPTIONS to 1 iff exceptions are enabled.
+#  define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__SUNPRO_CC)
+// Sun Pro CC supports exceptions.  However, there is no compile-time way of
+// detecting whether they are enabled or not.  Therefore, we assume that
+// they are enabled unless the user tells us otherwise.
+#  define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__IBMCPP__) && __EXCEPTIONS
+// xlC defines __EXCEPTIONS to 1 iff exceptions are enabled.
+#  define GTEST_HAS_EXCEPTIONS 1
+# elif defined(__HP_aCC)
+// Exception handling is in effect by default in HP aCC compiler. It has to
+// be turned of by +noeh compiler option if desired.
+#  define GTEST_HAS_EXCEPTIONS 1
+# else
+// For other compilers, we assume exceptions are disabled to be
+// conservative.
+#  define GTEST_HAS_EXCEPTIONS 0
+# endif  // defined(_MSC_VER) || defined(__BORLANDC__)
+#endif  // GTEST_HAS_EXCEPTIONS
+
+#if !defined(GTEST_HAS_STD_STRING)
+// Even though we don't use this macro any longer, we keep it in case
+// some clients still depend on it.
+# define GTEST_HAS_STD_STRING 1
+#elif !GTEST_HAS_STD_STRING
+// The user told us that ::std::string isn't available.
+# error "Google Test cannot be used where ::std::string isn't available."
+#endif  // !defined(GTEST_HAS_STD_STRING)
+
+#ifndef GTEST_HAS_GLOBAL_STRING
+// The user didn't tell us whether ::string is available, so we need
+// to figure it out.
+
+# define GTEST_HAS_GLOBAL_STRING 0
+
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+#ifndef GTEST_HAS_STD_WSTRING
+// The user didn't tell us whether ::std::wstring is available, so we need
+// to figure it out.
+// TODO(wan@google.com): uses autoconf to detect whether ::std::wstring
+//   is available.
+
+// Cygwin 1.7 and below doesn't support ::std::wstring.
+// Solaris' libc++ doesn't support it either.  Android has
+// no support for it at least as recent as Froyo (2.2).
+# define GTEST_HAS_STD_WSTRING \
+    (!(GTEST_OS_LINUX_ANDROID || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS))
+
+#endif  // GTEST_HAS_STD_WSTRING
+
+#ifndef GTEST_HAS_GLOBAL_WSTRING
+// The user didn't tell us whether ::wstring is available, so we need
+// to figure it out.
+# define GTEST_HAS_GLOBAL_WSTRING \
+    (GTEST_HAS_STD_WSTRING && GTEST_HAS_GLOBAL_STRING)
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+// Determines whether RTTI is available.
+#ifndef GTEST_HAS_RTTI
+// The user didn't tell us whether RTTI is enabled, so we need to
+// figure it out.
+
+# ifdef _MSC_VER
+
+#  ifdef _CPPRTTI  // MSVC defines this macro iff RTTI is enabled.
+#   define GTEST_HAS_RTTI 1
+#  else
+#   define GTEST_HAS_RTTI 0
+#  endif
+
+// Starting with version 4.3.2, gcc defines __GXX_RTTI iff RTTI is enabled.
+# elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40302)
+
+#  ifdef __GXX_RTTI
+// When building against STLport with the Android NDK and with
+// -frtti -fno-exceptions, the build fails at link time with undefined
+// references to __cxa_bad_typeid. Note sure if STL or toolchain bug,
+// so disable RTTI when detected.
+#   if GTEST_OS_LINUX_ANDROID && defined(_STLPORT_MAJOR) && \
+       !defined(__EXCEPTIONS)
+#    define GTEST_HAS_RTTI 0
+#   else
+#    define GTEST_HAS_RTTI 1
+#   endif  // GTEST_OS_LINUX_ANDROID && __STLPORT_MAJOR && !__EXCEPTIONS
+#  else
+#   define GTEST_HAS_RTTI 0
+#  endif  // __GXX_RTTI
+
+// Clang defines __GXX_RTTI starting with version 3.0, but its manual recommends
+// using has_feature instead. has_feature(cxx_rtti) is supported since 2.7, the
+// first version with C++ support.
+# elif defined(__clang__)
+
+#  define GTEST_HAS_RTTI __has_feature(cxx_rtti)
+
+// Starting with version 9.0 IBM Visual Age defines __RTTI_ALL__ to 1 if
+// both the typeid and dynamic_cast features are present.
+# elif defined(__IBMCPP__) && (__IBMCPP__ >= 900)
+
+#  ifdef __RTTI_ALL__
+#   define GTEST_HAS_RTTI 1
+#  else
+#   define GTEST_HAS_RTTI 0
+#  endif
+
+# else
+
+// For all other compilers, we assume RTTI is enabled.
+#  define GTEST_HAS_RTTI 1
+
+# endif  // _MSC_VER
+
+#endif  // GTEST_HAS_RTTI
+
+// It's this header's responsibility to #include <typeinfo> when RTTI
+// is enabled.
+#if GTEST_HAS_RTTI
+# include <typeinfo>
+#endif
+
+// Determines whether Google Test can use the pthreads library.
+#ifndef GTEST_HAS_PTHREAD
+// The user didn't tell us explicitly, so we make reasonable assumptions about
+// which platforms have pthreads support.
+//
+// To disable threading support in Google Test, add -DGTEST_HAS_PTHREAD=0
+// to your compiler flags.
+# define GTEST_HAS_PTHREAD (GTEST_OS_LINUX || GTEST_OS_MAC || GTEST_OS_HPUX \
+    || GTEST_OS_QNX || GTEST_OS_FREEBSD || GTEST_OS_NACL || GTEST_OS_NETBSD)
+#endif  // GTEST_HAS_PTHREAD
+
+#if GTEST_HAS_PTHREAD
+// gtest-port.h guarantees to #include <pthread.h> when GTEST_HAS_PTHREAD is
+// true.
+# include <pthread.h>  // NOLINT
+
+// For timespec and nanosleep, used below.
+# include <time.h>  // NOLINT
+#endif
+
+// Determines if hash_map/hash_set are available.
+// Only used for testing against those containers.
+#if !defined(GTEST_HAS_HASH_MAP_)
+# if defined(_MSC_VER) && (_MSC_VER < 1900)
+#  define GTEST_HAS_HASH_MAP_ 1  // Indicates that hash_map is available.
+#  define GTEST_HAS_HASH_SET_ 1  // Indicates that hash_set is available.
+# endif  // _MSC_VER
+#endif  // !defined(GTEST_HAS_HASH_MAP_)
+
+// Determines whether Google Test can use tr1/tuple.  You can define
+// this macro to 0 to prevent Google Test from using tuple (any
+// feature depending on tuple with be disabled in this mode).
+#ifndef GTEST_HAS_TR1_TUPLE
+# if GTEST_OS_LINUX_ANDROID && defined(_STLPORT_MAJOR)
+// STLport, provided with the Android NDK, has neither <tr1/tuple> or <tuple>.
+#  define GTEST_HAS_TR1_TUPLE 0
+# else
+// The user didn't tell us not to do it, so we assume it's OK.
+#  define GTEST_HAS_TR1_TUPLE 1
+# endif
+#endif  // GTEST_HAS_TR1_TUPLE
+
+// Determines whether Google Test's own tr1 tuple implementation
+// should be used.
+#ifndef GTEST_USE_OWN_TR1_TUPLE
+// The user didn't tell us, so we need to figure it out.
+
+// We use our own TR1 tuple if we aren't sure the user has an
+// implementation of it already.  At this time, libstdc++ 4.0.0+ and
+// MSVC 2010 are the only mainstream standard libraries that come
+// with a TR1 tuple implementation.  NVIDIA's CUDA NVCC compiler
+// pretends to be GCC by defining __GNUC__ and friends, but cannot
+// compile GCC's tuple implementation.  MSVC 2008 (9.0) provides TR1
+// tuple in a 323 MB Feature Pack download, which we cannot assume the
+// user has.  QNX's QCC compiler is a modified GCC but it doesn't
+// support TR1 tuple.  libc++ only provides std::tuple, in C++11 mode,
+// and it can be used with some compilers that define __GNUC__.
+# if (defined(__GNUC__) && !defined(__CUDACC__) && (GTEST_GCC_VER_ >= 40000) \
+      && !GTEST_OS_QNX && !defined(_LIBCPP_VERSION)) || _MSC_VER >= 1600
+#  define GTEST_ENV_HAS_TR1_TUPLE_ 1
+# endif
+
+// C++11 specifies that <tuple> provides std::tuple. Use that if gtest is used
+// in C++11 mode and libstdc++ isn't very old (binaries targeting OS X 10.6
+// can build with clang but need to use gcc4.2's libstdc++).
+# if GTEST_LANG_CXX11 && (!defined(__GLIBCXX__) || __GLIBCXX__ > 20110325)
+#  define GTEST_ENV_HAS_STD_TUPLE_ 1
+# endif
+
+# if GTEST_ENV_HAS_TR1_TUPLE_ || GTEST_ENV_HAS_STD_TUPLE_
+#  define GTEST_USE_OWN_TR1_TUPLE 0
+# else
+#  define GTEST_USE_OWN_TR1_TUPLE 1
+# endif
+
+#endif  // GTEST_USE_OWN_TR1_TUPLE
+
+// To avoid conditional compilation everywhere, we make it
+// gtest-port.h's responsibility to #include the header implementing
+// tuple.
+#if GTEST_HAS_STD_TUPLE_
+# include <tuple>  // IWYU pragma: export
+# define GTEST_TUPLE_NAMESPACE_ ::std
+#endif  // GTEST_HAS_STD_TUPLE_
+
+// We include tr1::tuple even if std::tuple is available to define printers for
+// them.
+#if GTEST_HAS_TR1_TUPLE
+# ifndef GTEST_TUPLE_NAMESPACE_
+#  define GTEST_TUPLE_NAMESPACE_ ::std::tr1
+# endif  // GTEST_TUPLE_NAMESPACE_
+
+# if GTEST_USE_OWN_TR1_TUPLE
+// This file was GENERATED by command:
+//     pump.py gtest-tuple.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2009 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Implements a subset of TR1 tuple needed by Google Test and Google Mock.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+
+#include <utility>  // For ::std::pair.
+
+// The compiler used in Symbian has a bug that prevents us from declaring the
+// tuple template as a friend (it complains that tuple is redefined).  This
+// hack bypasses the bug by declaring the members that should otherwise be
+// private as public.
+// Sun Studio versions < 12 also have the above bug.
+#if defined(__SYMBIAN32__) || (defined(__SUNPRO_CC) && __SUNPRO_CC < 0x590)
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ public:
+#else
+# define GTEST_DECLARE_TUPLE_AS_FRIEND_ \
+    template <GTEST_10_TYPENAMES_(U)> friend class tuple; \
+   private:
+#endif
+
+// Visual Studio 2010, 2012, and 2013 define symbols in std::tr1 that conflict
+// with our own definitions. Therefore using our own tuple does not work on
+// those compilers.
+#if defined(_MSC_VER) && _MSC_VER >= 1600  /* 1600 is Visual Studio 2010 */
+# error "gtest's tuple doesn't compile on Visual Studio 2010 or later. \
+GTEST_USE_OWN_TR1_TUPLE must be set to 0 on those compilers."
+#endif
+
+// GTEST_n_TUPLE_(T) is the type of an n-tuple.
+#define GTEST_0_TUPLE_(T) tuple<>
+#define GTEST_1_TUPLE_(T) tuple<T##0, void, void, void, void, void, void, \
+    void, void, void>
+#define GTEST_2_TUPLE_(T) tuple<T##0, T##1, void, void, void, void, void, \
+    void, void, void>
+#define GTEST_3_TUPLE_(T) tuple<T##0, T##1, T##2, void, void, void, void, \
+    void, void, void>
+#define GTEST_4_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, void, void, void, \
+    void, void, void>
+#define GTEST_5_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, void, void, \
+    void, void, void>
+#define GTEST_6_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, void, \
+    void, void, void>
+#define GTEST_7_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    void, void, void>
+#define GTEST_8_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    T##7, void, void>
+#define GTEST_9_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    T##7, T##8, void>
+#define GTEST_10_TUPLE_(T) tuple<T##0, T##1, T##2, T##3, T##4, T##5, T##6, \
+    T##7, T##8, T##9>
+
+// GTEST_n_TYPENAMES_(T) declares a list of n typenames.
+#define GTEST_0_TYPENAMES_(T)
+#define GTEST_1_TYPENAMES_(T) typename T##0
+#define GTEST_2_TYPENAMES_(T) typename T##0, typename T##1
+#define GTEST_3_TYPENAMES_(T) typename T##0, typename T##1, typename T##2
+#define GTEST_4_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3
+#define GTEST_5_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4
+#define GTEST_6_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5
+#define GTEST_7_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6
+#define GTEST_8_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6, typename T##7
+#define GTEST_9_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6, \
+    typename T##7, typename T##8
+#define GTEST_10_TYPENAMES_(T) typename T##0, typename T##1, typename T##2, \
+    typename T##3, typename T##4, typename T##5, typename T##6, \
+    typename T##7, typename T##8, typename T##9
+
+// In theory, defining stuff in the ::std namespace is undefined
+// behavior.  We can do this as we are playing the role of a standard
+// library vendor.
+namespace std
+{
+    namespace tr1
+    {
+
+        template <typename T0 = void, typename T1 = void, typename T2 = void,
+                  typename T3 = void, typename T4 = void, typename T5 = void,
+                  typename T6 = void, typename T7 = void, typename T8 = void,
+                  typename T9 = void>
+        class tuple;
+
+// Anything in namespace gtest_internal is Google Test's INTERNAL
+// IMPLEMENTATION DETAIL and MUST NOT BE USED DIRECTLY in user code.
+        namespace gtest_internal
+        {
+
+// ByRef<T>::type is T if T is a reference; otherwise it's const T&.
+            template <typename T>
+            struct ByRef {
+                typedef const T &type;
+            };  // NOLINT
+            template <typename T>
+            struct ByRef<T &> {
+                typedef T &type;
+            };  // NOLINT
+
+// A handy wrapper for ByRef.
+#define GTEST_BY_REF_(T) typename ::std::tr1::gtest_internal::ByRef<T>::type
+
+// AddRef<T>::type is T if T is a reference; otherwise it's T&.  This
+// is the same as tr1::add_reference<T>::type.
+            template <typename T>
+            struct AddRef {
+                typedef T &type;
+            };  // NOLINT
+            template <typename T>
+            struct AddRef<T &> {
+                typedef T &type;
+            };  // NOLINT
+
+// A handy wrapper for AddRef.
+#define GTEST_ADD_REF_(T) typename ::std::tr1::gtest_internal::AddRef<T>::type
+
+// A helper for implementing get<k>().
+            template <int k> class Get;
+
+// A helper for implementing tuple_element<k, T>.  kIndexValid is true
+// iff k < the number of fields in tuple type T.
+            template <bool kIndexValid, int kIndex, class Tuple>
+            struct TupleElement;
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 0, GTEST_10_TUPLE_(T) > {
+                typedef T0 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 1, GTEST_10_TUPLE_(T) > {
+                typedef T1 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 2, GTEST_10_TUPLE_(T) > {
+                typedef T2 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 3, GTEST_10_TUPLE_(T) > {
+                typedef T3 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 4, GTEST_10_TUPLE_(T) > {
+                typedef T4 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 5, GTEST_10_TUPLE_(T) > {
+                typedef T5 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 6, GTEST_10_TUPLE_(T) > {
+                typedef T6 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 7, GTEST_10_TUPLE_(T) > {
+                typedef T7 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 8, GTEST_10_TUPLE_(T) > {
+                typedef T8 type;
+            };
+
+            template <GTEST_10_TYPENAMES_(T)>
+            struct TupleElement<true, 9, GTEST_10_TUPLE_(T) > {
+                typedef T9 type;
+            };
+
+        }  // namespace gtest_internal
+
+        template <>
+        class tuple<>
+        {
+        public:
+            tuple() {}
+            tuple(const tuple & /* t */)  {}
+            tuple &operator=(const tuple & /* t */)
+            {
+                return *this;
+            }
+        };
+
+        template <GTEST_1_TYPENAMES_(T)>
+        class GTEST_1_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0) : f0_(f0) {}
+
+            tuple(const tuple& t) : f0_(t.f0_) {}
+
+            template <GTEST_1_TYPENAMES_(U)>
+            tuple(const GTEST_1_TUPLE_(U)& t) : f0_(t.f0_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_1_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_1_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_1_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_1_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                return *this;
+            }
+
+            T0 f0_;
+        };
+
+        template <GTEST_2_TYPENAMES_(T)>
+        class GTEST_2_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1) : f0_(f0),
+                f1_(f1) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_) {}
+
+            template <GTEST_2_TYPENAMES_(U)>
+            tuple(const GTEST_2_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_) {}
+            template <typename U0, typename U1>
+            tuple(const ::std::pair<U0, U1> &p) : f0_(p.first), f1_(p.second) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_2_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_2_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+            template <typename U0, typename U1>
+            tuple &operator=(const ::std::pair<U0, U1> &p) {
+                f0_ = p.first;
+                f1_ = p.second;
+                return *this;
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_2_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_2_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+        };
+
+        template <GTEST_3_TYPENAMES_(T)>
+        class GTEST_3_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2) : f0_(f0), f1_(f1), f2_(f2) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {}
+
+            template <GTEST_3_TYPENAMES_(U)>
+            tuple(const GTEST_3_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_3_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_3_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_3_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_3_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+        };
+
+        template <GTEST_4_TYPENAMES_(T)>
+        class GTEST_4_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3) : f0_(f0), f1_(f1), f2_(f2),
+                f3_(f3) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_) {}
+
+            template <GTEST_4_TYPENAMES_(U)>
+            tuple(const GTEST_4_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_4_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_4_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_4_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_4_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+        };
+
+        template <GTEST_5_TYPENAMES_(T)>
+        class GTEST_5_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3,
+                           GTEST_BY_REF_(T4) f4) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_) {}
+
+            template <GTEST_5_TYPENAMES_(U)>
+            tuple(const GTEST_5_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_5_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_5_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_5_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_5_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+        };
+
+        template <GTEST_6_TYPENAMES_(T)>
+        class GTEST_6_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+                f5_(f5) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_) {}
+
+            template <GTEST_6_TYPENAMES_(U)>
+            tuple(const GTEST_6_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_6_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_6_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_6_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_6_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+        };
+
+        template <GTEST_7_TYPENAMES_(T)>
+        class GTEST_7_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6) : f0_(f0), f1_(f1), f2_(f2),
+                f3_(f3), f4_(f4), f5_(f5), f6_(f6) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {}
+
+            template <GTEST_7_TYPENAMES_(U)>
+            tuple(const GTEST_7_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_7_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_7_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_7_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_7_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+        };
+
+        template <GTEST_8_TYPENAMES_(T)>
+        class GTEST_8_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6,
+                           GTEST_BY_REF_(T7) f7) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+                f5_(f5), f6_(f6), f7_(f7) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {}
+
+            template <GTEST_8_TYPENAMES_(U)>
+            tuple(const GTEST_8_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_8_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_8_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_8_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_8_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                f7_ = t.f7_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+            T7 f7_;
+        };
+
+        template <GTEST_9_TYPENAMES_(T)>
+        class GTEST_9_TUPLE_(T)
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7,
+                           GTEST_BY_REF_(T8) f8) : f0_(f0), f1_(f1), f2_(f2), f3_(f3), f4_(f4),
+                f5_(f5), f6_(f6), f7_(f7), f8_(f8) {}
+
+            tuple(const tuple& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {}
+
+            template <GTEST_9_TYPENAMES_(U)>
+            tuple(const GTEST_9_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_) {}
+
+            tuple &operator=(const tuple& t) {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_9_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_9_TUPLE_(U)& t) {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_9_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_9_TUPLE_(U)& t) {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                f7_ = t.f7_;
+                f8_ = t.f8_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+            T7 f7_;
+            T8 f8_;
+        };
+
+        template <GTEST_10_TYPENAMES_(T)>
+        class tuple
+        {
+        public:
+            template <int k> friend class gtest_internal::Get;
+
+            tuple() : f0_(), f1_(), f2_(), f3_(), f4_(), f5_(), f6_(), f7_(), f8_(),
+                f9_() {}
+
+            explicit tuple(GTEST_BY_REF_(T0) f0, GTEST_BY_REF_(T1) f1,
+                           GTEST_BY_REF_(T2) f2, GTEST_BY_REF_(T3) f3, GTEST_BY_REF_(T4) f4,
+                           GTEST_BY_REF_(T5) f5, GTEST_BY_REF_(T6) f6, GTEST_BY_REF_(T7) f7,
+                           GTEST_BY_REF_(T8) f8, GTEST_BY_REF_(T9) f9) : f0_(f0), f1_(f1), f2_(f2),
+                f3_(f3), f4_(f4), f5_(f5), f6_(f6), f7_(f7), f8_(f8), f9_(f9) {}
+
+            tuple(const tuple &t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_), f3_(t.f3_),
+                f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_), f9_(t.f9_) {}
+
+            template <GTEST_10_TYPENAMES_(U)>
+            tuple(const GTEST_10_TUPLE_(U)& t) : f0_(t.f0_), f1_(t.f1_), f2_(t.f2_),
+                f3_(t.f3_), f4_(t.f4_), f5_(t.f5_), f6_(t.f6_), f7_(t.f7_), f8_(t.f8_),
+                f9_(t.f9_) {}
+
+            tuple &operator=(const tuple &t)
+            {
+                return CopyFrom(t);
+            }
+
+            template <GTEST_10_TYPENAMES_(U)>
+            tuple &operator=(const GTEST_10_TUPLE_(U)& t)
+            {
+                return CopyFrom(t);
+            }
+
+            GTEST_DECLARE_TUPLE_AS_FRIEND_
+
+            template <GTEST_10_TYPENAMES_(U)>
+            tuple &CopyFrom(const GTEST_10_TUPLE_(U)& t)
+            {
+                f0_ = t.f0_;
+                f1_ = t.f1_;
+                f2_ = t.f2_;
+                f3_ = t.f3_;
+                f4_ = t.f4_;
+                f5_ = t.f5_;
+                f6_ = t.f6_;
+                f7_ = t.f7_;
+                f8_ = t.f8_;
+                f9_ = t.f9_;
+                return *this;
+            }
+
+            T0 f0_;
+            T1 f1_;
+            T2 f2_;
+            T3 f3_;
+            T4 f4_;
+            T5 f5_;
+            T6 f6_;
+            T7 f7_;
+            T8 f8_;
+            T9 f9_;
+        };
+
+// 6.1.3.2 Tuple creation functions.
+
+// Known limitations: we don't support passing an
+// std::tr1::reference_wrapper<T> to make_tuple().  And we don't
+// implement tie().
+
+        inline tuple<> make_tuple()
+        {
+            return tuple<>();
+        }
+
+        template <GTEST_1_TYPENAMES_(T)>
+        inline GTEST_1_TUPLE_(T) make_tuple(const T0 &f0)
+        {
+            return GTEST_1_TUPLE_(T)(f0);
+        }
+
+        template <GTEST_2_TYPENAMES_(T)>
+        inline GTEST_2_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1)
+        {
+            return GTEST_2_TUPLE_(T)(f0, f1);
+        }
+
+        template <GTEST_3_TYPENAMES_(T)>
+        inline GTEST_3_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2)
+        {
+            return GTEST_3_TUPLE_(T)(f0, f1, f2);
+        }
+
+        template <GTEST_4_TYPENAMES_(T)>
+        inline GTEST_4_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3)
+        {
+            return GTEST_4_TUPLE_(T)(f0, f1, f2, f3);
+        }
+
+        template <GTEST_5_TYPENAMES_(T)>
+        inline GTEST_5_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4)
+        {
+            return GTEST_5_TUPLE_(T)(f0, f1, f2, f3, f4);
+        }
+
+        template <GTEST_6_TYPENAMES_(T)>
+        inline GTEST_6_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5)
+        {
+            return GTEST_6_TUPLE_(T)(f0, f1, f2, f3, f4, f5);
+        }
+
+        template <GTEST_7_TYPENAMES_(T)>
+        inline GTEST_7_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6)
+        {
+            return GTEST_7_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6);
+        }
+
+        template <GTEST_8_TYPENAMES_(T)>
+        inline GTEST_8_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6, const T7 &f7)
+        {
+            return GTEST_8_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7);
+        }
+
+        template <GTEST_9_TYPENAMES_(T)>
+        inline GTEST_9_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                            const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6, const T7 &f7,
+                                            const T8 &f8)
+        {
+            return GTEST_9_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8);
+        }
+
+        template <GTEST_10_TYPENAMES_(T)>
+        inline GTEST_10_TUPLE_(T) make_tuple(const T0 &f0, const T1 &f1, const T2 &f2,
+                                             const T3 &f3, const T4 &f4, const T5 &f5, const T6 &f6, const T7 &f7,
+                                             const T8 &f8, const T9 &f9)
+        {
+            return GTEST_10_TUPLE_(T)(f0, f1, f2, f3, f4, f5, f6, f7, f8, f9);
+        }
+
+// 6.1.3.3 Tuple helper classes.
+
+        template <typename Tuple> struct tuple_size;
+
+        template <GTEST_0_TYPENAMES_(T)>
+        struct tuple_size<GTEST_0_TUPLE_(T) > {
+            static const int value = 0;
+        };
+
+        template <GTEST_1_TYPENAMES_(T)>
+        struct tuple_size<GTEST_1_TUPLE_(T) > {
+            static const int value = 1;
+        };
+
+        template <GTEST_2_TYPENAMES_(T)>
+        struct tuple_size<GTEST_2_TUPLE_(T) > {
+            static const int value = 2;
+        };
+
+        template <GTEST_3_TYPENAMES_(T)>
+        struct tuple_size<GTEST_3_TUPLE_(T) > {
+            static const int value = 3;
+        };
+
+        template <GTEST_4_TYPENAMES_(T)>
+        struct tuple_size<GTEST_4_TUPLE_(T) > {
+            static const int value = 4;
+        };
+
+        template <GTEST_5_TYPENAMES_(T)>
+        struct tuple_size<GTEST_5_TUPLE_(T) > {
+            static const int value = 5;
+        };
+
+        template <GTEST_6_TYPENAMES_(T)>
+        struct tuple_size<GTEST_6_TUPLE_(T) > {
+            static const int value = 6;
+        };
+
+        template <GTEST_7_TYPENAMES_(T)>
+        struct tuple_size<GTEST_7_TUPLE_(T) > {
+            static const int value = 7;
+        };
+
+        template <GTEST_8_TYPENAMES_(T)>
+        struct tuple_size<GTEST_8_TUPLE_(T) > {
+            static const int value = 8;
+        };
+
+        template <GTEST_9_TYPENAMES_(T)>
+        struct tuple_size<GTEST_9_TUPLE_(T) > {
+            static const int value = 9;
+        };
+
+        template <GTEST_10_TYPENAMES_(T)>
+        struct tuple_size<GTEST_10_TUPLE_(T) > {
+            static const int value = 10;
+        };
+
+        template <int k, class Tuple>
+        struct tuple_element {
+            typedef typename gtest_internal::TupleElement<
+            k < (tuple_size<Tuple>::value), k, Tuple>::type type;
+        };
+
+#define GTEST_TUPLE_ELEMENT_(k, Tuple) typename tuple_element<k, Tuple >::type
+
+// 6.1.3.4 Element access.
+
+        namespace gtest_internal
+        {
+
+            template <>
+            class Get<0>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f0_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(0, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f0_;
+                }
+            };
+
+            template <>
+            class Get<1>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f1_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(1, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f1_;
+                }
+            };
+
+            template <>
+            class Get<2>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f2_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(2, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f2_;
+                }
+            };
+
+            template <>
+            class Get<3>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f3_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(3, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f3_;
+                }
+            };
+
+            template <>
+            class Get<4>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f4_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(4, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f4_;
+                }
+            };
+
+            template <>
+            class Get<5>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f5_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(5, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f5_;
+                }
+            };
+
+            template <>
+            class Get<6>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f6_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(6, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f6_;
+                }
+            };
+
+            template <>
+            class Get<7>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f7_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(7, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f7_;
+                }
+            };
+
+            template <>
+            class Get<8>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f8_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(8, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f8_;
+                }
+            };
+
+            template <>
+            class Get<9>
+            {
+            public:
+                template <class Tuple>
+                static GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple))
+                Field(Tuple &t)
+                {
+                    return t.f9_;    // NOLINT
+                }
+
+                template <class Tuple>
+                static GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(9, Tuple))
+                ConstField(const Tuple &t)
+                {
+                    return t.f9_;
+                }
+            };
+
+        }  // namespace gtest_internal
+
+        template <int k, GTEST_10_TYPENAMES_(T)>
+        GTEST_ADD_REF_(GTEST_TUPLE_ELEMENT_(k, GTEST_10_TUPLE_(T)))
+        get(GTEST_10_TUPLE_(T)& t)
+        {
+            return gtest_internal::Get<k>::Field(t);
+        }
+
+        template <int k, GTEST_10_TYPENAMES_(T)>
+        GTEST_BY_REF_(GTEST_TUPLE_ELEMENT_(k,  GTEST_10_TUPLE_(T)))
+        get(const GTEST_10_TUPLE_(T)& t)
+        {
+            return gtest_internal::Get<k>::ConstField(t);
+        }
+
+// 6.1.3.5 Relational operators
+
+// We only implement == and !=, as we don't have a need for the rest yet.
+
+        namespace gtest_internal
+        {
+
+// SameSizeTuplePrefixComparator<k, k>::Eq(t1, t2) returns true if the
+// first k fields of t1 equals the first k fields of t2.
+// SameSizeTuplePrefixComparator(k1, k2) would be a compiler error if
+// k1 != k2.
+            template <int kSize1, int kSize2>
+            struct SameSizeTuplePrefixComparator;
+
+            template <>
+            struct SameSizeTuplePrefixComparator<0, 0> {
+                template <class Tuple1, class Tuple2>
+                static bool Eq(const Tuple1 & /* t1 */, const Tuple2 & /* t2 */)
+                {
+                    return true;
+                }
+            };
+
+            template <int k>
+            struct SameSizeTuplePrefixComparator<k, k> {
+                template <class Tuple1, class Tuple2>
+                static bool Eq(const Tuple1 &t1, const Tuple2 &t2)
+                {
+                    return SameSizeTuplePrefixComparator<k - 1, k - 1>::Eq(t1, t2) &&
+                           ::std::tr1::get<k - 1>(t1) == ::std::tr1::get<k - 1>(t2);
+                }
+            };
+
+        }  // namespace gtest_internal
+
+        template <GTEST_10_TYPENAMES_(T), GTEST_10_TYPENAMES_(U)>
+        inline bool operator==(const GTEST_10_TUPLE_(T)& t,
+                               const GTEST_10_TUPLE_(U)& u)
+        {
+            return gtest_internal::SameSizeTuplePrefixComparator<
+                   tuple_size<GTEST_10_TUPLE_(T) >::value,
+                   tuple_size<GTEST_10_TUPLE_(U) >::value>::Eq(t, u);
+        }
+
+        template <GTEST_10_TYPENAMES_(T), GTEST_10_TYPENAMES_(U)>
+        inline bool operator!=(const GTEST_10_TUPLE_(T)& t,
+                               const GTEST_10_TUPLE_(U)& u)
+        {
+            return !(t == u);
+        }
+
+// 6.1.4 Pairs.
+// Unimplemented.
+
+    }  // namespace tr1
+}  // namespace std
+
+#undef GTEST_0_TUPLE_
+#undef GTEST_1_TUPLE_
+#undef GTEST_2_TUPLE_
+#undef GTEST_3_TUPLE_
+#undef GTEST_4_TUPLE_
+#undef GTEST_5_TUPLE_
+#undef GTEST_6_TUPLE_
+#undef GTEST_7_TUPLE_
+#undef GTEST_8_TUPLE_
+#undef GTEST_9_TUPLE_
+#undef GTEST_10_TUPLE_
+
+#undef GTEST_0_TYPENAMES_
+#undef GTEST_1_TYPENAMES_
+#undef GTEST_2_TYPENAMES_
+#undef GTEST_3_TYPENAMES_
+#undef GTEST_4_TYPENAMES_
+#undef GTEST_5_TYPENAMES_
+#undef GTEST_6_TYPENAMES_
+#undef GTEST_7_TYPENAMES_
+#undef GTEST_8_TYPENAMES_
+#undef GTEST_9_TYPENAMES_
+#undef GTEST_10_TYPENAMES_
+
+#undef GTEST_DECLARE_TUPLE_AS_FRIEND_
+#undef GTEST_BY_REF_
+#undef GTEST_ADD_REF_
+#undef GTEST_TUPLE_ELEMENT_
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TUPLE_H_
+# elif GTEST_ENV_HAS_STD_TUPLE_
+#  include <tuple>
+// C++11 puts its tuple into the ::std namespace rather than
+// ::std::tr1.  gtest expects tuple to live in ::std::tr1, so put it there.
+// This causes undefined behavior, but supported compilers react in
+// the way we intend.
+namespace std
+{
+    namespace tr1
+    {
+        using ::std::get;
+        using ::std::make_tuple;
+        using ::std::tuple;
+        using ::std::tuple_element;
+        using ::std::tuple_size;
+    }
+}
+
+# elif GTEST_OS_SYMBIAN
+
+// On Symbian, BOOST_HAS_TR1_TUPLE causes Boost's TR1 tuple library to
+// use STLport's tuple implementation, which unfortunately doesn't
+// work as the copy of STLport distributed with Symbian is incomplete.
+// By making sure BOOST_HAS_TR1_TUPLE is undefined, we force Boost to
+// use its own tuple implementation.
+#  ifdef BOOST_HAS_TR1_TUPLE
+#   undef BOOST_HAS_TR1_TUPLE
+#  endif  // BOOST_HAS_TR1_TUPLE
+
+// This prevents <boost/tr1/detail/config.hpp>, which defines
+// BOOST_HAS_TR1_TUPLE, from being #included by Boost's <tuple>.
+#  define BOOST_TR1_DETAIL_CONFIG_HPP_INCLUDED
+#  include <tuple>  // IWYU pragma: export  // NOLINT
+
+# elif defined(__GNUC__) && (GTEST_GCC_VER_ >= 40000)
+// GCC 4.0+ implements tr1/tuple in the <tr1/tuple> header.  This does
+// not conform to the TR1 spec, which requires the header to be <tuple>.
+
+#  if !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302
+// Until version 4.3.2, gcc has a bug that causes <tr1/functional>,
+// which is #included by <tr1/tuple>, to not compile when RTTI is
+// disabled.  _TR1_FUNCTIONAL is the header guard for
+// <tr1/functional>.  Hence the following #define is a hack to prevent
+// <tr1/functional> from being included.
+#   define _TR1_FUNCTIONAL 1
+#   include <tr1/tuple>
+#   undef _TR1_FUNCTIONAL  // Allows the user to #include
+// <tr1/functional> if he chooses to.
+#  else
+#   include <tr1/tuple>  // NOLINT
+#  endif  // !GTEST_HAS_RTTI && GTEST_GCC_VER_ < 40302
+
+# else
+// If the compiler is not GCC 4.0+, we assume the user is using a
+// spec-conforming TR1 implementation.
+#  include <tuple>  // IWYU pragma: export  // NOLINT
+# endif  // GTEST_USE_OWN_TR1_TUPLE
+
+#endif  // GTEST_HAS_TR1_TUPLE
+
+// Determines whether clone(2) is supported.
+// Usually it will only be available on Linux, excluding
+// Linux on the Itanium architecture.
+// Also see http://linux.die.net/man/2/clone.
+#ifndef GTEST_HAS_CLONE
+// The user didn't tell us, so we need to figure it out.
+
+# if GTEST_OS_LINUX && !defined(__ia64__)
+#  if GTEST_OS_LINUX_ANDROID
+// On Android, clone() became available at different API levels for each 32-bit
+// architecture.
+#    if defined(__LP64__) || \
+        (defined(__arm__) && __ANDROID_API__ >= 9) || \
+        (defined(__mips__) && __ANDROID_API__ >= 12) || \
+        (defined(__i386__) && __ANDROID_API__ >= 17)
+#     define GTEST_HAS_CLONE 1
+#    else
+#     define GTEST_HAS_CLONE 0
+#    endif
+#  else
+#   define GTEST_HAS_CLONE 1
+#  endif
+# else
+#  define GTEST_HAS_CLONE 0
+# endif  // GTEST_OS_LINUX && !defined(__ia64__)
+
+#endif  // GTEST_HAS_CLONE
+
+// Determines whether to support stream redirection. This is used to test
+// output correctness and to implement death tests.
+#ifndef GTEST_HAS_STREAM_REDIRECTION
+// By default, we assume that stream redirection is supported on all
+// platforms except known mobile ones.
+# if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_SYMBIAN || \
+    GTEST_OS_WINDOWS_PHONE || GTEST_OS_WINDOWS_RT
+#  define GTEST_HAS_STREAM_REDIRECTION 0
+# else
+#  define GTEST_HAS_STREAM_REDIRECTION 1
+# endif  // !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_SYMBIAN
+#endif  // GTEST_HAS_STREAM_REDIRECTION
+
+// Determines whether to support death tests.
+// Google Test does not support death tests for VC 7.1 and earlier as
+// abort() in a VC 7.1 application compiled as GUI in debug config
+// pops up a dialog window that cannot be suppressed programmatically.
+#if (GTEST_OS_LINUX || GTEST_OS_CYGWIN || GTEST_OS_SOLARIS || \
+     (GTEST_OS_MAC && !GTEST_OS_IOS) || \
+     (GTEST_OS_WINDOWS_DESKTOP && _MSC_VER >= 1400) || \
+     GTEST_OS_WINDOWS_MINGW || GTEST_OS_AIX || GTEST_OS_HPUX || \
+     GTEST_OS_OPENBSD || GTEST_OS_QNX || GTEST_OS_FREEBSD || GTEST_OS_NETBSD)
+# define GTEST_HAS_DEATH_TEST 1
+#endif
+
+// We don't support MSVC 7.1 with exceptions disabled now.  Therefore
+// all the compilers we care about are adequate for supporting
+// value-parameterized tests.
+#define GTEST_HAS_PARAM_TEST 1
+
+// Determines whether to support type-driven tests.
+
+// Typed tests need <typeinfo> and variadic macros, which GCC, VC++ 8.0,
+// Sun Pro CC, IBM Visual Age, and HP aCC support.
+#if defined(__GNUC__) || (_MSC_VER >= 1400) || defined(__SUNPRO_CC) || \
+    defined(__IBMCPP__) || defined(__HP_aCC)
+# define GTEST_HAS_TYPED_TEST 1
+# define GTEST_HAS_TYPED_TEST_P 1
+#endif
+
+// Determines whether to support Combine(). This only makes sense when
+// value-parameterized tests are enabled.  The implementation doesn't
+// work on Sun Studio since it doesn't understand templated conversion
+// operators.
+#if GTEST_HAS_PARAM_TEST && GTEST_HAS_TR1_TUPLE && !defined(__SUNPRO_CC)
+# define GTEST_HAS_COMBINE 1
+#endif
+
+// Determines whether the system compiler uses UTF-16 for encoding wide strings.
+#define GTEST_WIDE_STRING_USES_UTF16_ \
+    (GTEST_OS_WINDOWS || GTEST_OS_CYGWIN || GTEST_OS_SYMBIAN || GTEST_OS_AIX)
+
+// Determines whether test results can be streamed to a socket.
+#if GTEST_OS_LINUX
+# define GTEST_CAN_STREAM_RESULTS_ 1
+#endif
+
+// Defines some utility macros.
+
+// The GNU compiler emits a warning if nested "if" statements are followed by
+// an "else" statement and braces are not used to explicitly disambiguate the
+// "else" binding.  This leads to problems with code like:
+//
+//   if (gate)
+//     ASSERT_*(condition) << "Some message";
+//
+// The "switch (0) case 0:" idiom is used to suppress this.
+#ifdef __INTEL_COMPILER
+# define GTEST_AMBIGUOUS_ELSE_BLOCKER_
+#else
+# define GTEST_AMBIGUOUS_ELSE_BLOCKER_ switch (0) case 0: default:  // NOLINT
+#endif
+
+// Use this annotation at the end of a struct/class definition to
+// prevent the compiler from optimizing away instances that are never
+// used.  This is useful when all interesting logic happens inside the
+// c'tor and / or d'tor.  Example:
+//
+//   struct Foo {
+//     Foo() { ... }
+//   } GTEST_ATTRIBUTE_UNUSED_;
+//
+// Also use it after a variable or parameter declaration to tell the
+// compiler the variable/parameter does not have to be used.
+#if defined(__GNUC__) && !defined(COMPILER_ICC)
+# define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused))
+#elif defined(__clang__)
+# if __has_attribute(unused)
+#  define GTEST_ATTRIBUTE_UNUSED_ __attribute__ ((unused))
+# endif
+#endif
+#ifndef GTEST_ATTRIBUTE_UNUSED_
+# define GTEST_ATTRIBUTE_UNUSED_
+#endif
+
+// Use this annotation before a function that takes a printf format string.
+#if defined(__GNUC__) && !defined(COMPILER_ICC)
+# if defined(__MINGW_PRINTF_FORMAT)
+// MinGW has two different printf implementations. Ensure the format macro
+// matches the selected implementation. See
+// https://sourceforge.net/p/mingw-w64/wiki2/gnu%20printf/.
+#  define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check) \
+       __attribute__((__format__(__MINGW_PRINTF_FORMAT, string_index, \
+                                 first_to_check)))
+# else
+#  define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check) \
+       __attribute__((__format__(__printf__, string_index, first_to_check)))
+# endif
+#else
+# define GTEST_ATTRIBUTE_PRINTF_(string_index, first_to_check)
+#endif
+
+// A macro to disallow operator=
+// This should be used in the private: declarations for a class.
+#define GTEST_DISALLOW_ASSIGN_(type)\
+  void operator=(type const &)
+
+// A macro to disallow copy constructor and operator=
+// This should be used in the private: declarations for a class.
+#define GTEST_DISALLOW_COPY_AND_ASSIGN_(type)\
+  type(type const &);\
+  GTEST_DISALLOW_ASSIGN_(type)
+
+// Tell the compiler to warn about unused return values for functions declared
+// with this macro.  The macro should be used on function declarations
+// following the argument list:
+//
+//   Sprocket* AllocateSprocket() GTEST_MUST_USE_RESULT_;
+#if defined(__GNUC__) && (GTEST_GCC_VER_ >= 30400) && !defined(COMPILER_ICC)
+# define GTEST_MUST_USE_RESULT_ __attribute__ ((warn_unused_result))
+#else
+# define GTEST_MUST_USE_RESULT_
+#endif  // __GNUC__ && (GTEST_GCC_VER_ >= 30400) && !COMPILER_ICC
+
+// MS C++ compiler emits warning when a conditional expression is compile time
+// constant. In some contexts this warning is false positive and needs to be
+// suppressed. Use the following two macros in such cases:
+//
+// GTEST_INTENTIONAL_CONST_COND_PUSH_()
+// while (true) {
+// GTEST_INTENTIONAL_CONST_COND_POP_()
+// }
+# define GTEST_INTENTIONAL_CONST_COND_PUSH_() \
+    GTEST_DISABLE_MSC_WARNINGS_PUSH_(4127)
+# define GTEST_INTENTIONAL_CONST_COND_POP_() \
+    GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+// Determine whether the compiler supports Microsoft's Structured Exception
+// Handling.  This is supported by several Windows compilers but generally
+// does not exist on any other system.
+#ifndef GTEST_HAS_SEH
+// The user didn't tell us, so we need to figure it out.
+
+# if defined(_MSC_VER) || defined(__BORLANDC__)
+// These two compilers are known to support SEH.
+#  define GTEST_HAS_SEH 1
+# else
+// Assume no SEH.
+#  define GTEST_HAS_SEH 0
+# endif
+
+#define GTEST_IS_THREADSAFE \
+    (GTEST_HAS_MUTEX_AND_THREAD_LOCAL_ \
+     || (GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT) \
+     || GTEST_HAS_PTHREAD)
+
+#endif  // GTEST_HAS_SEH
+
+// GTEST_API_ qualifies all symbols that must be exported. The definitions below
+// are guarded by #ifndef to give embedders a chance to define GTEST_API_ in
+// gtest/internal/custom/gtest-port.h
+#ifndef GTEST_API_
+
+#ifdef _MSC_VER
+# if GTEST_LINKED_AS_SHARED_LIBRARY
+#  define GTEST_API_ __declspec(dllimport)
+# elif GTEST_CREATE_SHARED_LIBRARY
+#  define GTEST_API_ __declspec(dllexport)
+# endif
+#elif __GNUC__ >= 4 || defined(__clang__)
+# define GTEST_API_ __attribute__((visibility ("default")))
+#endif // _MSC_VER
+
+#endif // GTEST_API_
+
+#ifndef GTEST_API_
+# define GTEST_API_
+#endif // GTEST_API_
+
+#ifdef __GNUC__
+// Ask the compiler to never inline a given function.
+# define GTEST_NO_INLINE_ __attribute__((noinline))
+#else
+# define GTEST_NO_INLINE_
+#endif
+
+// _LIBCPP_VERSION is defined by the libc++ library from the LLVM project.
+#if defined(__GLIBCXX__) || defined(_LIBCPP_VERSION)
+# define GTEST_HAS_CXXABI_H_ 1
+#else
+# define GTEST_HAS_CXXABI_H_ 0
+#endif
+
+// A function level attribute to disable checking for use of uninitialized
+// memory when built with MemorySanitizer.
+#if defined(__clang__)
+# if __has_feature(memory_sanitizer)
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_ \
+       __attribute__((no_sanitize_memory))
+# else
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+# endif  // __has_feature(memory_sanitizer)
+#else
+# define GTEST_ATTRIBUTE_NO_SANITIZE_MEMORY_
+#endif  // __clang__
+
+// A function level attribute to disable AddressSanitizer instrumentation.
+#if defined(__clang__)
+# if __has_feature(address_sanitizer)
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_ \
+       __attribute__((no_sanitize_address))
+# else
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+# endif  // __has_feature(address_sanitizer)
+#else
+# define GTEST_ATTRIBUTE_NO_SANITIZE_ADDRESS_
+#endif  // __clang__
+
+// A function level attribute to disable ThreadSanitizer instrumentation.
+#if defined(__clang__)
+# if __has_feature(thread_sanitizer)
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_ \
+       __attribute__((no_sanitize_thread))
+# else
+#  define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+# endif  // __has_feature(thread_sanitizer)
+#else
+# define GTEST_ATTRIBUTE_NO_SANITIZE_THREAD_
+#endif  // __clang__
+
+namespace testing
+{
+
+    class Message;
+
+#if defined(GTEST_TUPLE_NAMESPACE_)
+// Import tuple and friends into the ::testing namespace.
+// It is part of our interface, having them in ::testing allows us to change
+// their types as needed.
+    using GTEST_TUPLE_NAMESPACE_::get;
+    using GTEST_TUPLE_NAMESPACE_::make_tuple;
+    using GTEST_TUPLE_NAMESPACE_::tuple;
+    using GTEST_TUPLE_NAMESPACE_::tuple_size;
+    using GTEST_TUPLE_NAMESPACE_::tuple_element;
+#endif  // defined(GTEST_TUPLE_NAMESPACE_)
+
+    namespace internal
+    {
+
+// A secret type that Google Test users don't know about.  It has no
+// definition on purpose.  Therefore it's impossible to create a
+// Secret object, which is what we want.
+        class Secret;
+
+// The GTEST_COMPILE_ASSERT_ macro can be used to verify that a compile time
+// expression is true. For example, you could use it to verify the
+// size of a static array:
+//
+//   GTEST_COMPILE_ASSERT_(GTEST_ARRAY_SIZE_(names) == NUM_NAMES,
+//                         names_incorrect_size);
+//
+// or to make sure a struct is smaller than a certain size:
+//
+//   GTEST_COMPILE_ASSERT_(sizeof(foo) < 128, foo_too_large);
+//
+// The second argument to the macro is the name of the variable. If
+// the expression is false, most compilers will issue a warning/error
+// containing the name of the variable.
+
+#if GTEST_LANG_CXX11
+# define GTEST_COMPILE_ASSERT_(expr, msg) static_assert(expr, #msg)
+#else  // !GTEST_LANG_CXX11
+        template <bool>
+        struct CompileAssert {
+        };
+
+# define GTEST_COMPILE_ASSERT_(expr, msg) \
+  typedef ::testing::internal::CompileAssert<(static_cast<bool>(expr))> \
+      msg[static_cast<bool>(expr) ? 1 : -1] GTEST_ATTRIBUTE_UNUSED_
+#endif  // !GTEST_LANG_CXX11
+
+// Implementation details of GTEST_COMPILE_ASSERT_:
+//
+// (In C++11, we simply use static_assert instead of the following)
+//
+// - GTEST_COMPILE_ASSERT_ works by defining an array type that has -1
+//   elements (and thus is invalid) when the expression is false.
+//
+// - The simpler definition
+//
+//    #define GTEST_COMPILE_ASSERT_(expr, msg) typedef char msg[(expr) ? 1 : -1]
+//
+//   does not work, as gcc supports variable-length arrays whose sizes
+//   are determined at run-time (this is gcc's extension and not part
+//   of the C++ standard).  As a result, gcc fails to reject the
+//   following code with the simple definition:
+//
+//     int foo;
+//     GTEST_COMPILE_ASSERT_(foo, msg); // not supposed to compile as foo is
+//                                      // not a compile-time constant.
+//
+// - By using the type CompileAssert<(bool(expr))>, we ensures that
+//   expr is a compile-time constant.  (Template arguments must be
+//   determined at compile-time.)
+//
+// - The outter parentheses in CompileAssert<(bool(expr))> are necessary
+//   to work around a bug in gcc 3.4.4 and 4.0.1.  If we had written
+//
+//     CompileAssert<bool(expr)>
+//
+//   instead, these compilers will refuse to compile
+//
+//     GTEST_COMPILE_ASSERT_(5 > 0, some_message);
+//
+//   (They seem to think the ">" in "5 > 0" marks the end of the
+//   template argument list.)
+//
+// - The array size is (bool(expr) ? 1 : -1), instead of simply
+//
+//     ((expr) ? 1 : -1).
+//
+//   This is to avoid running into a bug in MS VC 7.1, which
+//   causes ((0.0) ? 1 : -1) to incorrectly evaluate to 1.
+
+// StaticAssertTypeEqHelper is used by StaticAssertTypeEq defined in gtest.h.
+//
+// This template is declared, but intentionally undefined.
+        template <typename T1, typename T2>
+        struct StaticAssertTypeEqHelper;
+
+        template <typename T>
+        struct StaticAssertTypeEqHelper<T, T> {
+            enum { value = true };
+        };
+
+// Evaluates to the number of elements in 'array'.
+#define GTEST_ARRAY_SIZE_(array) (sizeof(array) / sizeof(array[0]))
+
+#if GTEST_HAS_GLOBAL_STRING
+        typedef ::string string;
+#else
+        typedef ::std::string string;
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+        typedef ::wstring wstring;
+#elif GTEST_HAS_STD_WSTRING
+        typedef ::std::wstring wstring;
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+// A helper for suppressing warnings on constant condition.  It just
+// returns 'condition'.
+        GTEST_API_ bool IsTrue(bool condition);
+
+// Defines scoped_ptr.
+
+// This implementation of scoped_ptr is PARTIAL - it only contains
+// enough stuff to satisfy Google Test's need.
+        template <typename T>
+        class scoped_ptr
+        {
+        public:
+            typedef T element_type;
+
+            explicit scoped_ptr(T *p = NULL) : ptr_(p) {}
+            ~scoped_ptr()
+            {
+                reset();
+            }
+
+            T &operator*() const
+            {
+                return *ptr_;
+            }
+            T *operator->() const
+            {
+                return ptr_;
+            }
+            T *get() const
+            {
+                return ptr_;
+            }
+
+            T *release()
+            {
+                T *const ptr = ptr_;
+                ptr_ = NULL;
+                return ptr;
+            }
+
+            void reset(T *p = NULL)
+            {
+                if (p != ptr_) {
+                    if (IsTrue(sizeof(T) > 0)) {  // Makes sure T is a complete type.
+                        delete ptr_;
+                    }
+                    ptr_ = p;
+                }
+            }
+
+            friend void swap(scoped_ptr &a, scoped_ptr &b)
+            {
+                using std::swap;
+                swap(a.ptr_, b.ptr_);
+            }
+
+        private:
+            T *ptr_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(scoped_ptr);
+        };
+
+// Defines RE.
+
+// A simple C++ wrapper for <regex.h>.  It uses the POSIX Extended
+// Regular Expression syntax.
+        class GTEST_API_ RE
+        {
+        public:
+            // A copy constructor is required by the Standard to initialize object
+            // references from r-values.
+            RE(const RE &other)
+            {
+                Init(other.pattern());
+            }
+
+            // Constructs an RE from a string.
+            RE(const ::std::string &regex)
+            {
+                Init(regex.c_str());    // NOLINT
+            }
+
+#if GTEST_HAS_GLOBAL_STRING
+
+            RE(const ::string &regex)
+            {
+                Init(regex.c_str());    // NOLINT
+            }
+
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+            RE(const char *regex)
+            {
+                Init(regex);    // NOLINT
+            }
+            ~RE();
+
+            // Returns the string representation of the regex.
+            const char *pattern() const
+            {
+                return pattern_;
+            }
+
+            // FullMatch(str, re) returns true iff regular expression re matches
+            // the entire str.
+            // PartialMatch(str, re) returns true iff regular expression re
+            // matches a substring of str (including str itself).
+            //
+            // TODO(wan@google.com): make FullMatch() and PartialMatch() work
+            // when str contains NUL characters.
+            static bool FullMatch(const ::std::string &str, const RE &re)
+            {
+                return FullMatch(str.c_str(), re);
+            }
+            static bool PartialMatch(const ::std::string &str, const RE &re)
+            {
+                return PartialMatch(str.c_str(), re);
+            }
+
+#if GTEST_HAS_GLOBAL_STRING
+
+            static bool FullMatch(const ::string &str, const RE &re)
+            {
+                return FullMatch(str.c_str(), re);
+            }
+            static bool PartialMatch(const ::string &str, const RE &re)
+            {
+                return PartialMatch(str.c_str(), re);
+            }
+
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+            static bool FullMatch(const char *str, const RE &re);
+            static bool PartialMatch(const char *str, const RE &re);
+
+        private:
+            void Init(const char *regex);
+
+            // We use a const char* instead of an std::string, as Google Test used to be
+            // used where std::string is not available.  TODO(wan@google.com): change to
+            // std::string.
+            const char *pattern_;
+            bool is_valid_;
+
+#if GTEST_USES_POSIX_RE
+
+            regex_t full_regex_;     // For FullMatch().
+            regex_t partial_regex_;  // For PartialMatch().
+
+#else  // GTEST_USES_SIMPLE_RE
+
+            const char *full_pattern_;  // For FullMatch();
+
+#endif
+
+            GTEST_DISALLOW_ASSIGN_(RE);
+        };
+
+// Formats a source file path and a line number as they would appear
+// in an error message from the compiler used to compile this code.
+        GTEST_API_ ::std::string FormatFileLocation(const char *file, int line);
+
+// Formats a file location for compiler-independent XML output.
+// Although this function is not platform dependent, we put it next to
+// FormatFileLocation in order to contrast the two functions.
+        GTEST_API_ ::std::string FormatCompilerIndependentFileLocation(const char *file,
+                                                                       int line);
+
+// Defines logging utilities:
+//   GTEST_LOG_(severity) - logs messages at the specified severity level. The
+//                          message itself is streamed into the macro.
+//   LogToStderr()  - directs all log messages to stderr.
+//   FlushInfoLog() - flushes informational log messages.
+
+        enum GTestLogSeverity {
+            GTEST_INFO,
+            GTEST_WARNING,
+            GTEST_ERROR,
+            GTEST_FATAL
+        };
+
+// Formats log entry severity, provides a stream object for streaming the
+// log message, and terminates the message with a newline when going out of
+// scope.
+        class GTEST_API_ GTestLog
+        {
+        public:
+            GTestLog(GTestLogSeverity severity, const char *file, int line);
+
+            // Flushes the buffers and, if severity is GTEST_FATAL, aborts the program.
+            ~GTestLog();
+
+            ::std::ostream &GetStream()
+            {
+                return ::std::cerr;
+            }
+
+        private:
+            const GTestLogSeverity severity_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestLog);
+        };
+
+#if !defined(GTEST_LOG_)
+
+# define GTEST_LOG_(severity) \
+    ::testing::internal::GTestLog(::testing::internal::GTEST_##severity, \
+                                  __FILE__, __LINE__).GetStream()
+
+        inline void LogToStderr() {}
+        inline void FlushInfoLog()
+        {
+            fflush(NULL);
+        }
+
+#endif  // !defined(GTEST_LOG_)
+
+#if !defined(GTEST_CHECK_)
+// INTERNAL IMPLEMENTATION - DO NOT USE.
+//
+// GTEST_CHECK_ is an all-mode assert. It aborts the program if the condition
+// is not satisfied.
+//  Synopsis:
+//    GTEST_CHECK_(boolean_condition);
+//     or
+//    GTEST_CHECK_(boolean_condition) << "Additional message";
+//
+//    This checks the condition and if the condition is not satisfied
+//    it prints message about the condition violation, including the
+//    condition itself, plus additional message streamed into it, if any,
+//    and then it aborts the program. It aborts the program irrespective of
+//    whether it is built in the debug mode or not.
+# define GTEST_CHECK_(condition) \
+    GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+    if (::testing::internal::IsTrue(condition)) \
+      ; \
+    else \
+      GTEST_LOG_(FATAL) << "Condition " #condition " failed. "
+#endif  // !defined(GTEST_CHECK_)
+
+// An all-mode assert to verify that the given POSIX-style function
+// call returns 0 (indicating success).  Known limitation: this
+// doesn't expand to a balanced 'if' statement, so enclose the macro
+// in {} if you need to use it as the only statement in an 'if'
+// branch.
+#define GTEST_CHECK_POSIX_SUCCESS_(posix_call) \
+  if (const int gtest_error = (posix_call)) \
+    GTEST_LOG_(FATAL) << #posix_call << "failed with error " \
+                      << gtest_error
+
+#if GTEST_HAS_STD_MOVE_
+        using std::move;
+#else  // GTEST_HAS_STD_MOVE_
+        template <typename T>
+        const T &move(const T &t)
+        {
+            return t;
+        }
+#endif  // GTEST_HAS_STD_MOVE_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Use ImplicitCast_ as a safe version of static_cast for upcasting in
+// the type hierarchy (e.g. casting a Foo* to a SuperclassOfFoo* or a
+// const Foo*).  When you use ImplicitCast_, the compiler checks that
+// the cast is safe.  Such explicit ImplicitCast_s are necessary in
+// surprisingly many situations where C++ demands an exact type match
+// instead of an argument type convertible to a target type.
+//
+// The syntax for using ImplicitCast_ is the same as for static_cast:
+//
+//   ImplicitCast_<ToType>(expr)
+//
+// ImplicitCast_ would have been part of the C++ standard library,
+// but the proposal was submitted too late.  It will probably make
+// its way into the language in the future.
+//
+// This relatively ugly name is intentional. It prevents clashes with
+// similar functions users may have (e.g., implicit_cast). The internal
+// namespace alone is not enough because the function can be found by ADL.
+        template<typename To>
+        inline To ImplicitCast_(To x)
+        {
+            return x;
+        }
+
+// When you upcast (that is, cast a pointer from type Foo to type
+// SuperclassOfFoo), it's fine to use ImplicitCast_<>, since upcasts
+// always succeed.  When you downcast (that is, cast a pointer from
+// type Foo to type SubclassOfFoo), static_cast<> isn't safe, because
+// how do you know the pointer is really of type SubclassOfFoo?  It
+// could be a bare Foo, or of type DifferentSubclassOfFoo.  Thus,
+// when you downcast, you should use this macro.  In debug mode, we
+// use dynamic_cast<> to double-check the downcast is legal (we die
+// if it's not).  In normal mode, we do the efficient static_cast<>
+// instead.  Thus, it's important to test in debug mode to make sure
+// the cast is legal!
+//    This is the only place in the code we should use dynamic_cast<>.
+// In particular, you SHOULDN'T be using dynamic_cast<> in order to
+// do RTTI (eg code like this:
+//    if (dynamic_cast<Subclass1>(foo)) HandleASubclass1Object(foo);
+//    if (dynamic_cast<Subclass2>(foo)) HandleASubclass2Object(foo);
+// You should design the code some other way not to need this.
+//
+// This relatively ugly name is intentional. It prevents clashes with
+// similar functions users may have (e.g., down_cast). The internal
+// namespace alone is not enough because the function can be found by ADL.
+        template<typename To, typename From>  // use like this: DownCast_<T*>(foo);
+        inline To DownCast_(From *f)    // so we only accept pointers
+        {
+            // Ensures that To is a sub-type of From *.  This test is here only
+            // for compile-time type checking, and has no overhead in an
+            // optimized build at run-time, as it will be optimized away
+            // completely.
+            GTEST_INTENTIONAL_CONST_COND_PUSH_()
+            if (false) {
+                GTEST_INTENTIONAL_CONST_COND_POP_()
+                const To to = NULL;
+                ::testing::internal::ImplicitCast_<From *>(to);
+            }
+
+#if GTEST_HAS_RTTI
+            // RTTI: debug mode only!
+            GTEST_CHECK_(f == NULL || dynamic_cast<To>(f) != NULL);
+#endif
+            return static_cast<To>(f);
+        }
+
+// Downcasts the pointer of type Base to Derived.
+// Derived must be a subclass of Base. The parameter MUST
+// point to a class of type Derived, not any subclass of it.
+// When RTTI is available, the function performs a runtime
+// check to enforce this.
+        template <class Derived, class Base>
+        Derived *CheckedDowncastToActualType(Base *base)
+        {
+#if GTEST_HAS_RTTI
+            GTEST_CHECK_(typeid(*base) == typeid(Derived));
+#endif
+
+#if GTEST_HAS_DOWNCAST_
+            return ::down_cast<Derived *>(base);
+#elif GTEST_HAS_RTTI
+            return dynamic_cast<Derived *>(base); // NOLINT
+#else
+            return static_cast<Derived *>(base); // Poor man's downcast.
+#endif
+        }
+
+#if GTEST_HAS_STREAM_REDIRECTION
+
+// Defines the stderr capturer:
+//   CaptureStdout     - starts capturing stdout.
+//   GetCapturedStdout - stops capturing stdout and returns the captured string.
+//   CaptureStderr     - starts capturing stderr.
+//   GetCapturedStderr - stops capturing stderr and returns the captured string.
+//
+        GTEST_API_ void CaptureStdout();
+        GTEST_API_ std::string GetCapturedStdout();
+        GTEST_API_ void CaptureStderr();
+        GTEST_API_ std::string GetCapturedStderr();
+
+#endif  // GTEST_HAS_STREAM_REDIRECTION
+
+// Returns the size (in bytes) of a file.
+        GTEST_API_ size_t GetFileSize(FILE *file);
+
+// Reads the entire content of a file as a string.
+        GTEST_API_ std::string ReadEntireFile(FILE *file);
+
+// All command line arguments.
+        GTEST_API_ const ::std::vector<testing::internal::string> &GetArgvs();
+
+#if GTEST_HAS_DEATH_TEST
+
+        const ::std::vector<testing::internal::string> &GetInjectableArgvs();
+        void SetInjectableArgvs(const ::std::vector<testing::internal::string> *
+                                new_argvs);
+
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+// Defines synchronization primitives.
+#if GTEST_IS_THREADSAFE
+# if GTEST_HAS_PTHREAD
+// Sleeps for (roughly) n milliseconds.  This function is only for testing
+// Google Test's own constructs.  Don't use it in user tests, either
+// directly or indirectly.
+        inline void SleepMilliseconds(int n)
+        {
+            const timespec time = {
+                0,                  // 0 seconds.
+                n * 1000L * 1000L,  // And n ms.
+            };
+            nanosleep(&time, NULL);
+        }
+# endif  // GTEST_HAS_PTHREAD
+
+# if GTEST_HAS_NOTIFICATION_
+// Notification has already been imported into the namespace.
+// Nothing to do here.
+
+# elif GTEST_HAS_PTHREAD
+// Allows a controller thread to pause execution of newly created
+// threads until notified.  Instances of this class must be created
+// and destroyed in the controller thread.
+//
+// This class is only for testing Google Test's own constructs. Do not
+// use it in user tests, either directly or indirectly.
+        class Notification
+        {
+        public:
+            Notification() : notified_(false)
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, NULL));
+            }
+            ~Notification()
+            {
+                pthread_mutex_destroy(&mutex_);
+            }
+
+            // Notifies all threads created with this notification to start. Must
+            // be called from the controller thread.
+            void Notify()
+            {
+                pthread_mutex_lock(&mutex_);
+                notified_ = true;
+                pthread_mutex_unlock(&mutex_);
+            }
+
+            // Blocks until the controller thread notifies. Must be called from a test
+            // thread.
+            void WaitForNotification()
+            {
+                for (;;) {
+                    pthread_mutex_lock(&mutex_);
+                    const bool notified = notified_;
+                    pthread_mutex_unlock(&mutex_);
+                    if (notified)
+                        break;
+                    SleepMilliseconds(10);
+                }
+            }
+
+        private:
+            pthread_mutex_t mutex_;
+            bool notified_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification);
+        };
+
+# elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+
+        GTEST_API_ void SleepMilliseconds(int n);
+
+// Provides leak-safe Windows kernel handle ownership.
+// Used in death tests and in threading support.
+        class GTEST_API_ AutoHandle
+        {
+        public:
+            // Assume that Win32 HANDLE type is equivalent to void*. Doing so allows us to
+            // avoid including <windows.h> in this header file. Including <windows.h> is
+            // undesirable because it defines a lot of symbols and macros that tend to
+            // conflict with client code. This assumption is verified by
+            // WindowsTypesTest.HANDLEIsVoidStar.
+            typedef void *Handle;
+            AutoHandle();
+            explicit AutoHandle(Handle handle);
+
+            ~AutoHandle();
+
+            Handle Get() const;
+            void Reset();
+            void Reset(Handle handle);
+
+        private:
+            // Returns true iff the handle is a valid handle object that can be closed.
+            bool IsCloseable() const;
+
+            Handle handle_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(AutoHandle);
+        };
+
+// Allows a controller thread to pause execution of newly created
+// threads until notified.  Instances of this class must be created
+// and destroyed in the controller thread.
+//
+// This class is only for testing Google Test's own constructs. Do not
+// use it in user tests, either directly or indirectly.
+        class GTEST_API_ Notification
+        {
+        public:
+            Notification();
+            void Notify();
+            void WaitForNotification();
+
+        private:
+            AutoHandle event_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Notification);
+        };
+# endif  // GTEST_HAS_NOTIFICATION_
+
+// On MinGW, we can have both GTEST_OS_WINDOWS and GTEST_HAS_PTHREAD
+// defined, but we don't want to use MinGW's pthreads implementation, which
+// has conformance problems with some versions of the POSIX standard.
+# if GTEST_HAS_PTHREAD && !GTEST_OS_WINDOWS_MINGW
+
+// As a C-function, ThreadFuncWithCLinkage cannot be templated itself.
+// Consequently, it cannot select a correct instantiation of ThreadWithParam
+// in order to call its Run(). Introducing ThreadWithParamBase as a
+// non-templated base class for ThreadWithParam allows us to bypass this
+// problem.
+        class ThreadWithParamBase
+        {
+        public:
+            virtual ~ThreadWithParamBase() {}
+            virtual void Run() = 0;
+        };
+
+// pthread_create() accepts a pointer to a function type with the C linkage.
+// According to the Standard (7.5/1), function types with different linkages
+// are different even if they are otherwise identical.  Some compilers (for
+// example, SunStudio) treat them as different types.  Since class methods
+// cannot be defined with C-linkage we need to define a free C-function to
+// pass into pthread_create().
+        extern "C" inline void *ThreadFuncWithCLinkage(void *thread)
+        {
+            static_cast<ThreadWithParamBase *>(thread)->Run();
+            return NULL;
+        }
+
+// Helper class for testing Google Test's multi-threading constructs.
+// To use it, write:
+//
+//   void ThreadFunc(int param) { /* Do things with param */ }
+//   Notification thread_can_start;
+//   ...
+//   // The thread_can_start parameter is optional; you can supply NULL.
+//   ThreadWithParam<int> thread(&ThreadFunc, 5, &thread_can_start);
+//   thread_can_start.Notify();
+//
+// These classes are only for testing Google Test's own constructs. Do
+// not use them in user tests, either directly or indirectly.
+        template <typename T>
+        class ThreadWithParam : public ThreadWithParamBase
+        {
+        public:
+            typedef void UserThreadFunc(T);
+
+            ThreadWithParam(UserThreadFunc *func, T param, Notification *thread_can_start)
+                : func_(func),
+                  param_(param),
+                  thread_can_start_(thread_can_start),
+                  finished_(false)
+            {
+                ThreadWithParamBase *const base = this;
+                // The thread can be created only after all fields except thread_
+                // have been initialized.
+                GTEST_CHECK_POSIX_SUCCESS_(
+                    pthread_create(&thread_, 0, &ThreadFuncWithCLinkage, base));
+            }
+            ~ThreadWithParam()
+            {
+                Join();
+            }
+
+            void Join()
+            {
+                if (!finished_) {
+                    GTEST_CHECK_POSIX_SUCCESS_(pthread_join(thread_, 0));
+                    finished_ = true;
+                }
+            }
+
+            virtual void Run()
+            {
+                if (thread_can_start_ != NULL)
+                    thread_can_start_->WaitForNotification();
+                func_(param_);
+            }
+
+        private:
+            UserThreadFunc *const func_;  // User-supplied thread function.
+            const T param_;  // User-supplied parameter to the thread function.
+            // When non-NULL, used to block execution until the controller thread
+            // notifies.
+            Notification *const thread_can_start_;
+            bool finished_;  // true iff we know that the thread function has finished.
+            pthread_t thread_;  // The native thread object.
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam);
+        };
+# endif  // !GTEST_OS_WINDOWS && GTEST_HAS_PTHREAD ||
+        // GTEST_HAS_MUTEX_AND_THREAD_LOCAL_
+
+# if GTEST_HAS_MUTEX_AND_THREAD_LOCAL_
+// Mutex and ThreadLocal have already been imported into the namespace.
+// Nothing to do here.
+
+# elif GTEST_OS_WINDOWS && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+
+// Mutex implements mutex on Windows platforms.  It is used in conjunction
+// with class MutexLock:
+//
+//   Mutex mutex;
+//   ...
+//   MutexLock lock(&mutex);  // Acquires the mutex and releases it at the
+//                            // end of the current scope.
+//
+// A static Mutex *must* be defined or declared using one of the following
+// macros:
+//   GTEST_DEFINE_STATIC_MUTEX_(g_some_mutex);
+//   GTEST_DECLARE_STATIC_MUTEX_(g_some_mutex);
+//
+// (A non-static Mutex is defined/declared in the usual way).
+        class GTEST_API_ Mutex
+        {
+        public:
+            enum MutexType { kStatic = 0, kDynamic = 1 };
+            // We rely on kStaticMutex being 0 as it is to what the linker initializes
+            // type_ in static mutexes.  critical_section_ will be initialized lazily
+            // in ThreadSafeLazyInit().
+            enum StaticConstructorSelector { kStaticMutex = 0 };
+
+            // This constructor intentionally does nothing.  It relies on type_ being
+            // statically initialized to 0 (effectively setting it to kStatic) and on
+            // ThreadSafeLazyInit() to lazily initialize the rest of the members.
+            explicit Mutex(StaticConstructorSelector /*dummy*/) {}
+
+            Mutex();
+            ~Mutex();
+
+            void Lock();
+
+            void Unlock();
+
+            // Does nothing if the current thread holds the mutex. Otherwise, crashes
+            // with high probability.
+            void AssertHeld();
+
+        private:
+            // Initializes owner_thread_id_ and critical_section_ in static mutexes.
+            void ThreadSafeLazyInit();
+
+            // Per http://blogs.msdn.com/b/oldnewthing/archive/2004/02/23/78395.aspx,
+            // we assume that 0 is an invalid value for thread IDs.
+            unsigned int owner_thread_id_;
+
+            // For static mutexes, we rely on these members being initialized to zeros
+            // by the linker.
+            MutexType type_;
+            long critical_section_init_phase_;  // NOLINT
+            GTEST_CRITICAL_SECTION *critical_section_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex);
+        };
+
+# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+    extern ::testing::internal::Mutex mutex
+
+# define GTEST_DEFINE_STATIC_MUTEX_(mutex) \
+    ::testing::internal::Mutex mutex(::testing::internal::Mutex::kStaticMutex)
+
+// We cannot name this class MutexLock because the ctor declaration would
+// conflict with a macro named MutexLock, which is defined on some
+// platforms. That macro is used as a defensive measure to prevent against
+// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than
+// "MutexLock l(&mu)".  Hence the typedef trick below.
+        class GTestMutexLock
+        {
+        public:
+            explicit GTestMutexLock(Mutex *mutex)
+                : mutex_(mutex)
+            {
+                mutex_->Lock();
+            }
+
+            ~GTestMutexLock()
+            {
+                mutex_->Unlock();
+            }
+
+        private:
+            Mutex *const mutex_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock);
+        };
+
+        typedef GTestMutexLock MutexLock;
+
+// Base class for ValueHolder<T>.  Allows a caller to hold and delete a value
+// without knowing its type.
+        class ThreadLocalValueHolderBase
+        {
+        public:
+            virtual ~ThreadLocalValueHolderBase() {}
+        };
+
+// Provides a way for a thread to send notifications to a ThreadLocal
+// regardless of its parameter type.
+        class ThreadLocalBase
+        {
+        public:
+            // Creates a new ValueHolder<T> object holding a default value passed to
+            // this ThreadLocal<T>'s constructor and returns it.  It is the caller's
+            // responsibility not to call this when the ThreadLocal<T> instance already
+            // has a value on the current thread.
+            virtual ThreadLocalValueHolderBase *NewValueForCurrentThread() const = 0;
+
+        protected:
+            ThreadLocalBase() {}
+            virtual ~ThreadLocalBase() {}
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocalBase);
+        };
+
+// Maps a thread to a set of ThreadLocals that have values instantiated on that
+// thread and notifies them when the thread exits.  A ThreadLocal instance is
+// expected to persist until all threads it has values on have terminated.
+        class GTEST_API_ ThreadLocalRegistry
+        {
+        public:
+            // Registers thread_local_instance as having value on the current thread.
+            // Returns a value that can be used to identify the thread from other threads.
+            static ThreadLocalValueHolderBase *GetValueOnCurrentThread(
+                const ThreadLocalBase *thread_local_instance);
+
+            // Invoked when a ThreadLocal instance is destroyed.
+            static void OnThreadLocalDestroyed(
+                const ThreadLocalBase *thread_local_instance);
+        };
+
+        class GTEST_API_ ThreadWithParamBase
+        {
+        public:
+            void Join();
+
+        protected:
+            class Runnable
+            {
+            public:
+                virtual ~Runnable() {}
+                virtual void Run() = 0;
+            };
+
+            ThreadWithParamBase(Runnable *runnable, Notification *thread_can_start);
+            virtual ~ThreadWithParamBase();
+
+        private:
+            AutoHandle thread_;
+        };
+
+// Helper class for testing Google Test's multi-threading constructs.
+        template <typename T>
+        class ThreadWithParam : public ThreadWithParamBase
+        {
+        public:
+            typedef void UserThreadFunc(T);
+
+            ThreadWithParam(UserThreadFunc *func, T param, Notification *thread_can_start)
+                : ThreadWithParamBase(new RunnableImpl(func, param), thread_can_start)
+            {
+            }
+            virtual ~ThreadWithParam() {}
+
+        private:
+            class RunnableImpl : public Runnable
+            {
+            public:
+                RunnableImpl(UserThreadFunc *func, T param)
+                    : func_(func),
+                      param_(param)
+                {
+                }
+                virtual ~RunnableImpl() {}
+                virtual void Run()
+                {
+                    func_(param_);
+                }
+
+            private:
+                UserThreadFunc *const func_;
+                const T param_;
+
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(RunnableImpl);
+            };
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadWithParam);
+        };
+
+// Implements thread-local storage on Windows systems.
+//
+//   // Thread 1
+//   ThreadLocal<int> tl(100);  // 100 is the default value for each thread.
+//
+//   // Thread 2
+//   tl.set(150);  // Changes the value for thread 2 only.
+//   EXPECT_EQ(150, tl.get());
+//
+//   // Thread 1
+//   EXPECT_EQ(100, tl.get());  // In thread 1, tl has the original value.
+//   tl.set(200);
+//   EXPECT_EQ(200, tl.get());
+//
+// The template type argument T must have a public copy constructor.
+// In addition, the default ThreadLocal constructor requires T to have
+// a public default constructor.
+//
+// The users of a TheadLocal instance have to make sure that all but one
+// threads (including the main one) using that instance have exited before
+// destroying it. Otherwise, the per-thread objects managed for them by the
+// ThreadLocal instance are not guaranteed to be destroyed on all platforms.
+//
+// Google Test only uses global ThreadLocal objects.  That means they
+// will die after main() has returned.  Therefore, no per-thread
+// object managed by Google Test will be leaked as long as all threads
+// using Google Test have exited when main() returns.
+        template <typename T>
+        class ThreadLocal : public ThreadLocalBase
+        {
+        public:
+            ThreadLocal() : default_factory_(new DefaultValueHolderFactory()) {}
+            explicit ThreadLocal(const T &value)
+                : default_factory_(new InstanceValueHolderFactory(value)) {}
+
+            ~ThreadLocal()
+            {
+                ThreadLocalRegistry::OnThreadLocalDestroyed(this);
+            }
+
+            T *pointer()
+            {
+                return GetOrCreateValue();
+            }
+            const T *pointer() const
+            {
+                return GetOrCreateValue();
+            }
+            const T &get() const
+            {
+                return *pointer();
+            }
+            void set(const T &value)
+            {
+                *pointer() = value;
+            }
+
+        private:
+            // Holds a value of T.  Can be deleted via its base class without the caller
+            // knowing the type of T.
+            class ValueHolder : public ThreadLocalValueHolderBase
+            {
+            public:
+                ValueHolder() : value_() {}
+                explicit ValueHolder(const T &value) : value_(value) {}
+
+                T *pointer()
+                {
+                    return &value_;
+                }
+
+            private:
+                T value_;
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder);
+            };
+
+
+            T *GetOrCreateValue() const
+            {
+                return static_cast<ValueHolder *>(
+                           ThreadLocalRegistry::GetValueOnCurrentThread(this))->pointer();
+            }
+
+            virtual ThreadLocalValueHolderBase *NewValueForCurrentThread() const
+            {
+                return default_factory_->MakeNewHolder();
+            }
+
+            class ValueHolderFactory
+            {
+            public:
+                ValueHolderFactory() {}
+                virtual ~ValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const = 0;
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolderFactory);
+            };
+
+            class DefaultValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                DefaultValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder();
+                }
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultValueHolderFactory);
+            };
+
+            class InstanceValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                explicit InstanceValueHolderFactory(const T &value) : value_(value) {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder(value_);
+                }
+
+            private:
+                const T value_;  // The value for each thread.
+
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(InstanceValueHolderFactory);
+            };
+
+            scoped_ptr<ValueHolderFactory> default_factory_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal);
+        };
+
+# elif GTEST_HAS_PTHREAD
+
+// MutexBase and Mutex implement mutex on pthreads-based platforms.
+        class MutexBase
+        {
+        public:
+            // Acquires this mutex.
+            void Lock()
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_lock(&mutex_));
+                owner_ = pthread_self();
+                has_owner_ = true;
+            }
+
+            // Releases this mutex.
+            void Unlock()
+            {
+                // Since the lock is being released the owner_ field should no longer be
+                // considered valid. We don't protect writing to has_owner_ here, as it's
+                // the caller's responsibility to ensure that the current thread holds the
+                // mutex when this is called.
+                has_owner_ = false;
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_unlock(&mutex_));
+            }
+
+            // Does nothing if the current thread holds the mutex. Otherwise, crashes
+            // with high probability.
+            void AssertHeld() const
+            {
+                GTEST_CHECK_(has_owner_ && pthread_equal(owner_, pthread_self()))
+                        << "The current thread is not holding the mutex @" << this;
+            }
+
+            // A static mutex may be used before main() is entered.  It may even
+            // be used before the dynamic initialization stage.  Therefore we
+            // must be able to initialize a static mutex object at link time.
+            // This means MutexBase has to be a POD and its member variables
+            // have to be public.
+        public:
+            pthread_mutex_t mutex_;  // The underlying pthread mutex.
+            // has_owner_ indicates whether the owner_ field below contains a valid thread
+            // ID and is therefore safe to inspect (e.g., to use in pthread_equal()). All
+            // accesses to the owner_ field should be protected by a check of this field.
+            // An alternative might be to memset() owner_ to all zeros, but there's no
+            // guarantee that a zero'd pthread_t is necessarily invalid or even different
+            // from pthread_self().
+            bool has_owner_;
+            pthread_t owner_;  // The thread holding the mutex.
+        };
+
+// Forward-declares a static mutex.
+#  define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+     extern ::testing::internal::MutexBase mutex
+
+// Defines and statically (i.e. at link time) initializes a static mutex.
+#  define GTEST_DEFINE_STATIC_MUTEX_(mutex) \
+     ::testing::internal::MutexBase mutex = { PTHREAD_MUTEX_INITIALIZER, false, pthread_t() }
+
+// The Mutex class can only be used for mutexes created at runtime. It
+// shares its API with MutexBase otherwise.
+        class Mutex : public MutexBase
+        {
+        public:
+            Mutex()
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_init(&mutex_, NULL));
+                has_owner_ = false;
+            }
+            ~Mutex()
+            {
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_mutex_destroy(&mutex_));
+            }
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Mutex);
+        };
+
+// We cannot name this class MutexLock because the ctor declaration would
+// conflict with a macro named MutexLock, which is defined on some
+// platforms. That macro is used as a defensive measure to prevent against
+// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than
+// "MutexLock l(&mu)".  Hence the typedef trick below.
+        class GTestMutexLock
+        {
+        public:
+            explicit GTestMutexLock(MutexBase *mutex)
+                : mutex_(mutex)
+            {
+                mutex_->Lock();
+            }
+
+            ~GTestMutexLock()
+            {
+                mutex_->Unlock();
+            }
+
+        private:
+            MutexBase *const mutex_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(GTestMutexLock);
+        };
+
+        typedef GTestMutexLock MutexLock;
+
+// Helpers for ThreadLocal.
+
+// pthread_key_create() requires DeleteThreadLocalValue() to have
+// C-linkage.  Therefore it cannot be templatized to access
+// ThreadLocal<T>.  Hence the need for class
+// ThreadLocalValueHolderBase.
+        class ThreadLocalValueHolderBase
+        {
+        public:
+            virtual ~ThreadLocalValueHolderBase() {}
+        };
+
+// Called by pthread to delete thread-local data stored by
+// pthread_setspecific().
+        extern "C" inline void DeleteThreadLocalValue(void *value_holder)
+        {
+            delete static_cast<ThreadLocalValueHolderBase *>(value_holder);
+        }
+
+// Implements thread-local storage on pthreads-based systems.
+        template <typename T>
+        class ThreadLocal
+        {
+        public:
+            ThreadLocal()
+                : key_(CreateKey()), default_factory_(new DefaultValueHolderFactory()) {}
+            explicit ThreadLocal(const T &value)
+                : key_(CreateKey()),
+                  default_factory_(new InstanceValueHolderFactory(value)) {}
+
+            ~ThreadLocal()
+            {
+                // Destroys the managed object for the current thread, if any.
+                DeleteThreadLocalValue(pthread_getspecific(key_));
+
+                // Releases resources associated with the key.  This will *not*
+                // delete managed objects for other threads.
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_key_delete(key_));
+            }
+
+            T *pointer()
+            {
+                return GetOrCreateValue();
+            }
+            const T *pointer() const
+            {
+                return GetOrCreateValue();
+            }
+            const T &get() const
+            {
+                return *pointer();
+            }
+            void set(const T &value)
+            {
+                *pointer() = value;
+            }
+
+        private:
+            // Holds a value of type T.
+            class ValueHolder : public ThreadLocalValueHolderBase
+            {
+            public:
+                ValueHolder() : value_() {}
+                explicit ValueHolder(const T &value) : value_(value) {}
+
+                T *pointer()
+                {
+                    return &value_;
+                }
+
+            private:
+                T value_;
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolder);
+            };
+
+            static pthread_key_t CreateKey()
+            {
+                pthread_key_t key;
+                // When a thread exits, DeleteThreadLocalValue() will be called on
+                // the object managed for that thread.
+                GTEST_CHECK_POSIX_SUCCESS_(
+                    pthread_key_create(&key, &DeleteThreadLocalValue));
+                return key;
+            }
+
+            T *GetOrCreateValue() const
+            {
+                ThreadLocalValueHolderBase *const holder =
+                    static_cast<ThreadLocalValueHolderBase *>(pthread_getspecific(key_));
+                if (holder != NULL) {
+                    return CheckedDowncastToActualType<ValueHolder>(holder)->pointer();
+                }
+
+                ValueHolder *const new_holder = default_factory_->MakeNewHolder();
+                ThreadLocalValueHolderBase *const holder_base = new_holder;
+                GTEST_CHECK_POSIX_SUCCESS_(pthread_setspecific(key_, holder_base));
+                return new_holder->pointer();
+            }
+
+            class ValueHolderFactory
+            {
+            public:
+                ValueHolderFactory() {}
+                virtual ~ValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const = 0;
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ValueHolderFactory);
+            };
+
+            class DefaultValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                DefaultValueHolderFactory() {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder();
+                }
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(DefaultValueHolderFactory);
+            };
+
+            class InstanceValueHolderFactory : public ValueHolderFactory
+            {
+            public:
+                explicit InstanceValueHolderFactory(const T &value) : value_(value) {}
+                virtual ValueHolder *MakeNewHolder() const
+                {
+                    return new ValueHolder(value_);
+                }
+
+            private:
+                const T value_;  // The value for each thread.
+
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(InstanceValueHolderFactory);
+            };
+
+            // A key pthreads uses for looking up per-thread values.
+            const pthread_key_t key_;
+            scoped_ptr<ValueHolderFactory> default_factory_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ThreadLocal);
+        };
+
+# endif  // GTEST_HAS_MUTEX_AND_THREAD_LOCAL_
+
+#else  // GTEST_IS_THREADSAFE
+
+// A dummy implementation of synchronization primitives (mutex, lock,
+// and thread-local variable).  Necessary for compiling Google Test where
+// mutex is not supported - using Google Test in multiple threads is not
+// supported on such platforms.
+
+        class Mutex
+        {
+        public:
+            Mutex() {}
+            void Lock() {}
+            void Unlock() {}
+            void AssertHeld() const {}
+        };
+
+# define GTEST_DECLARE_STATIC_MUTEX_(mutex) \
+  extern ::testing::internal::Mutex mutex
+
+# define GTEST_DEFINE_STATIC_MUTEX_(mutex) ::testing::internal::Mutex mutex
+
+// We cannot name this class MutexLock because the ctor declaration would
+// conflict with a macro named MutexLock, which is defined on some
+// platforms. That macro is used as a defensive measure to prevent against
+// inadvertent misuses of MutexLock like "MutexLock(&mu)" rather than
+// "MutexLock l(&mu)".  Hence the typedef trick below.
+        class GTestMutexLock
+        {
+        public:
+            explicit GTestMutexLock(Mutex *) {} // NOLINT
+        };
+
+        typedef GTestMutexLock MutexLock;
+
+        template <typename T>
+        class ThreadLocal
+        {
+        public:
+            ThreadLocal() : value_() {}
+            explicit ThreadLocal(const T &value) : value_(value) {}
+            T *pointer()
+            {
+                return &value_;
+            }
+            const T *pointer() const
+            {
+                return &value_;
+            }
+            const T &get() const
+            {
+                return value_;
+            }
+            void set(const T &value)
+            {
+                value_ = value;
+            }
+        private:
+            T value_;
+        };
+
+#endif  // GTEST_IS_THREADSAFE
+
+// Returns the number of threads running in the process, or 0 to indicate that
+// we cannot detect it.
+        GTEST_API_ size_t GetThreadCount();
+
+// Passing non-POD classes through ellipsis (...) crashes the ARM
+// compiler and generates a warning in Sun Studio.  The Nokia Symbian
+// and the IBM XL C/C++ compiler try to instantiate a copy constructor
+// for objects passed through ellipsis (...), failing for uncopyable
+// objects.  We define this to ensure that only POD is passed through
+// ellipsis on these systems.
+#if defined(__SYMBIAN32__) || defined(__IBMCPP__) || defined(__SUNPRO_CC)
+// We lose support for NULL detection where the compiler doesn't like
+// passing non-POD classes through ellipsis (...).
+# define GTEST_ELLIPSIS_NEEDS_POD_ 1
+#else
+# define GTEST_CAN_COMPARE_NULL 1
+#endif
+
+// The Nokia Symbian and IBM XL C/C++ compilers cannot decide between
+// const T& and const T* in a function template.  These compilers
+// _can_ decide between class template specializations for T and T*,
+// so a tr1::type_traits-like is_pointer works.
+#if defined(__SYMBIAN32__) || defined(__IBMCPP__)
+# define GTEST_NEEDS_IS_POINTER_ 1
+#endif
+
+        template <bool bool_value>
+        struct bool_constant {
+            typedef bool_constant<bool_value> type;
+            static const bool value = bool_value;
+        };
+        template <bool bool_value> const bool bool_constant<bool_value>::value;
+
+        typedef bool_constant<false> false_type;
+        typedef bool_constant<true> true_type;
+
+        template <typename T>
+        struct is_pointer : public false_type {};
+
+        template <typename T>
+        struct is_pointer<T *> : public true_type {};
+
+        template <typename Iterator>
+        struct IteratorTraits {
+            typedef typename Iterator::value_type value_type;
+        };
+
+        template <typename T>
+        struct IteratorTraits<T *> {
+            typedef T value_type;
+        };
+
+        template <typename T>
+        struct IteratorTraits<const T *> {
+            typedef T value_type;
+        };
+
+#if GTEST_OS_WINDOWS
+# define GTEST_PATH_SEP_ "\\"
+# define GTEST_HAS_ALT_PATH_SEP_ 1
+// The biggest signed integer type the compiler supports.
+        typedef __int64 BiggestInt;
+#else
+# define GTEST_PATH_SEP_ "/"
+# define GTEST_HAS_ALT_PATH_SEP_ 0
+        typedef long long BiggestInt;  // NOLINT
+#endif  // GTEST_OS_WINDOWS
+
+// Utilities for char.
+
+// isspace(int ch) and friends accept an unsigned char or EOF.  char
+// may be signed, depending on the compiler (or compiler flags).
+// Therefore we need to cast a char to unsigned char before calling
+// isspace(), etc.
+
+        inline bool IsAlpha(char ch)
+        {
+            return isalpha(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsAlNum(char ch)
+        {
+            return isalnum(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsDigit(char ch)
+        {
+            return isdigit(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsLower(char ch)
+        {
+            return islower(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsSpace(char ch)
+        {
+            return isspace(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsUpper(char ch)
+        {
+            return isupper(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsXDigit(char ch)
+        {
+            return isxdigit(static_cast<unsigned char>(ch)) != 0;
+        }
+        inline bool IsXDigit(wchar_t ch)
+        {
+            const unsigned char low_byte = static_cast<unsigned char>(ch);
+            return ch == low_byte && isxdigit(low_byte) != 0;
+        }
+
+        inline char ToLower(char ch)
+        {
+            return static_cast<char>(tolower(static_cast<unsigned char>(ch)));
+        }
+        inline char ToUpper(char ch)
+        {
+            return static_cast<char>(toupper(static_cast<unsigned char>(ch)));
+        }
+
+        inline std::string StripTrailingSpaces(std::string str)
+        {
+            std::string::iterator it = str.end();
+            while (it != str.begin() && IsSpace(*--it))
+                it = str.erase(it);
+            return str;
+        }
+
+// The testing::internal::posix namespace holds wrappers for common
+// POSIX functions.  These wrappers hide the differences between
+// Windows/MSVC and POSIX systems.  Since some compilers define these
+// standard functions as macros, the wrapper cannot have the same name
+// as the wrapped function.
+
+        namespace posix
+        {
+
+// Functions with a different name on Windows.
+
+#if GTEST_OS_WINDOWS
+
+            typedef struct _stat StatStruct;
+
+# ifdef __BORLANDC__
+            inline int IsATTY(int fd)
+            {
+                return isatty(fd);
+            }
+            inline int StrCaseCmp(const char *s1, const char *s2)
+            {
+                return stricmp(s1, s2);
+            }
+            inline char *StrDup(const char *src)
+            {
+                return strdup(src);
+            }
+# else  // !__BORLANDC__
+#  if GTEST_OS_WINDOWS_MOBILE
+            inline int IsATTY(int /* fd */)
+            {
+                return 0;
+            }
+#  else
+            inline int IsATTY(int fd)
+            {
+                return _isatty(fd);
+            }
+#  endif  // GTEST_OS_WINDOWS_MOBILE
+            inline int StrCaseCmp(const char *s1, const char *s2)
+            {
+                return _stricmp(s1, s2);
+            }
+            inline char *StrDup(const char *src)
+            {
+                return _strdup(src);
+            }
+# endif  // __BORLANDC__
+
+# if GTEST_OS_WINDOWS_MOBILE
+            inline int FileNo(FILE *file)
+            {
+                return reinterpret_cast<int>(_fileno(file));
+            }
+// Stat(), RmDir(), and IsDir() are not needed on Windows CE at this
+// time and thus not defined there.
+# else
+            inline int FileNo(FILE *file)
+            {
+                return _fileno(file);
+            }
+            inline int Stat(const char *path, StatStruct *buf)
+            {
+                return _stat(path, buf);
+            }
+            inline int RmDir(const char *dir)
+            {
+                return _rmdir(dir);
+            }
+            inline bool IsDir(const StatStruct &st)
+            {
+                return (_S_IFDIR & st.st_mode) != 0;
+            }
+# endif  // GTEST_OS_WINDOWS_MOBILE
+
+#else
+
+            typedef struct stat StatStruct;
+
+            inline int FileNo(FILE *file)
+            {
+                return fileno(file);
+            }
+            inline int IsATTY(int fd)
+            {
+                return isatty(fd);
+            }
+            inline int Stat(const char *path, StatStruct *buf)
+            {
+                return stat(path, buf);
+            }
+            inline int StrCaseCmp(const char *s1, const char *s2)
+            {
+                return strcasecmp(s1, s2);
+            }
+            inline char *StrDup(const char *src)
+            {
+                return strdup(src);
+            }
+            inline int RmDir(const char *dir)
+            {
+                return rmdir(dir);
+            }
+            inline bool IsDir(const StatStruct &st)
+            {
+                return S_ISDIR(st.st_mode);
+            }
+
+#endif  // GTEST_OS_WINDOWS
+
+// Functions deprecated by MSVC 8.0.
+
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4996 /* deprecated function */)
+
+            inline const char *StrNCpy(char *dest, const char *src, size_t n)
+            {
+                return strncpy(dest, src, n);
+            }
+
+// ChDir(), FReopen(), FDOpen(), Read(), Write(), Close(), and
+// StrError() aren't needed on Windows CE at this time and thus not
+// defined there.
+
+#if !GTEST_OS_WINDOWS_MOBILE && !GTEST_OS_WINDOWS_PHONE && !GTEST_OS_WINDOWS_RT
+            inline int ChDir(const char *dir)
+            {
+                return chdir(dir);
+            }
+#endif
+            inline FILE *FOpen(const char *path, const char *mode)
+            {
+                return fopen(path, mode);
+            }
+#if !GTEST_OS_WINDOWS_MOBILE
+            inline FILE *FReopen(const char *path, const char *mode, FILE *stream)
+            {
+                return freopen(path, mode, stream);
+            }
+            inline FILE *FDOpen(int fd, const char *mode)
+            {
+                return fdopen(fd, mode);
+            }
+#endif
+            inline int FClose(FILE *fp)
+            {
+                return fclose(fp);
+            }
+#if !GTEST_OS_WINDOWS_MOBILE
+            inline int Read(int fd, void *buf, unsigned int count)
+            {
+                return static_cast<int>(read(fd, buf, count));
+            }
+            inline int Write(int fd, const void *buf, unsigned int count)
+            {
+                return static_cast<int>(write(fd, buf, count));
+            }
+            inline int Close(int fd)
+            {
+                return close(fd);
+            }
+            inline const char *StrError(int errnum)
+            {
+                return strerror(errnum);
+            }
+#endif
+            inline const char *GetEnv(const char *name)
+            {
+#if GTEST_OS_WINDOWS_MOBILE || GTEST_OS_WINDOWS_PHONE | GTEST_OS_WINDOWS_RT
+                // We are on Windows CE, which has no environment variables.
+                static_cast<void>(name);  // To prevent 'unused argument' warning.
+                return NULL;
+#elif defined(__BORLANDC__) || defined(__SunOS_5_8) || defined(__SunOS_5_9)
+                // Environment variables which we programmatically clear will be set to the
+                // empty string rather than unset (NULL).  Handle that case.
+                const char *const env = getenv(name);
+                return (env != NULL && env[0] != '\0') ? env : NULL;
+#else
+                return getenv(name);
+#endif
+            }
+
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+#if GTEST_OS_WINDOWS_MOBILE
+// Windows CE has no C library. The abort() function is used in
+// several places in Google Test. This implementation provides a reasonable
+// imitation of standard behaviour.
+            void Abort();
+#else
+            inline void Abort()
+            {
+                abort();
+            }
+#endif  // GTEST_OS_WINDOWS_MOBILE
+
+        }  // namespace posix
+
+// MSVC "deprecates" snprintf and issues warnings wherever it is used.  In
+// order to avoid these warnings, we need to use _snprintf or _snprintf_s on
+// MSVC-based platforms.  We map the GTEST_SNPRINTF_ macro to the appropriate
+// function in order to achieve that.  We use macro definition here because
+// snprintf is a variadic function.
+#if _MSC_VER >= 1400 && !GTEST_OS_WINDOWS_MOBILE
+// MSVC 2005 and above support variadic macros.
+# define GTEST_SNPRINTF_(buffer, size, format, ...) \
+     _snprintf_s(buffer, size, size, format, __VA_ARGS__)
+#elif defined(_MSC_VER)
+// Windows CE does not define _snprintf_s and MSVC prior to 2005 doesn't
+// complain about _snprintf.
+# define GTEST_SNPRINTF_ _snprintf
+#else
+# define GTEST_SNPRINTF_ snprintf
+#endif
+
+// The maximum number a BiggestInt can represent.  This definition
+// works no matter BiggestInt is represented in one's complement or
+// two's complement.
+//
+// We cannot rely on numeric_limits in STL, as __int64 and long long
+// are not part of standard C++ and numeric_limits doesn't need to be
+// defined for them.
+        const BiggestInt kMaxBiggestInt =
+            ~(static_cast<BiggestInt>(1) << (8*sizeof(BiggestInt) - 1));
+
+// This template class serves as a compile-time function from size to
+// type.  It maps a size in bytes to a primitive type with that
+// size. e.g.
+//
+//   TypeWithSize<4>::UInt
+//
+// is typedef-ed to be unsigned int (unsigned integer made up of 4
+// bytes).
+//
+// Such functionality should belong to STL, but I cannot find it
+// there.
+//
+// Google Test uses this class in the implementation of floating-point
+// comparison.
+//
+// For now it only handles UInt (unsigned int) as that's all Google Test
+// needs.  Other types can be easily added in the future if need
+// arises.
+        template <size_t size>
+        class TypeWithSize
+        {
+        public:
+            // This prevents the user from using TypeWithSize<N> with incorrect
+            // values of N.
+            typedef void UInt;
+        };
+
+// The specialization for size 4.
+        template <>
+        class TypeWithSize<4>
+        {
+        public:
+            // unsigned int has size 4 in both gcc and MSVC.
+            //
+            // As base/basictypes.h doesn't compile on Windows, we cannot use
+            // uint32, uint64, and etc here.
+            typedef int Int;
+            typedef unsigned int UInt;
+        };
+
+// The specialization for size 8.
+        template <>
+        class TypeWithSize<8>
+        {
+        public:
+#if GTEST_OS_WINDOWS
+            typedef __int64 Int;
+            typedef unsigned __int64 UInt;
+#else
+            typedef long long Int;  // NOLINT
+            typedef unsigned long long UInt;  // NOLINT
+#endif  // GTEST_OS_WINDOWS
+        };
+
+// Integer types of known sizes.
+        typedef TypeWithSize<4>::Int Int32;
+        typedef TypeWithSize<4>::UInt UInt32;
+        typedef TypeWithSize<8>::Int Int64;
+        typedef TypeWithSize<8>::UInt UInt64;
+        typedef TypeWithSize<8>::Int TimeInMillis;  // Represents time in milliseconds.
+
+// Utilities for command line flags and environment variables.
+
+// Macro for referencing flags.
+#if !defined(GTEST_FLAG)
+# define GTEST_FLAG(name) FLAGS_gtest_##name
+#endif  // !defined(GTEST_FLAG)
+
+#if !defined(GTEST_USE_OWN_FLAGFILE_FLAG_)
+# define GTEST_USE_OWN_FLAGFILE_FLAG_ 1
+#endif  // !defined(GTEST_USE_OWN_FLAGFILE_FLAG_)
+
+#if !defined(GTEST_DECLARE_bool_)
+# define GTEST_FLAG_SAVER_ ::testing::internal::GTestFlagSaver
+
+// Macros for declaring flags.
+# define GTEST_DECLARE_bool_(name) GTEST_API_ extern bool GTEST_FLAG(name)
+# define GTEST_DECLARE_int32_(name) \
+    GTEST_API_ extern ::testing::internal::Int32 GTEST_FLAG(name)
+#define GTEST_DECLARE_string_(name) \
+    GTEST_API_ extern ::std::string GTEST_FLAG(name)
+
+// Macros for defining flags.
+#define GTEST_DEFINE_bool_(name, default_val, doc) \
+    GTEST_API_ bool GTEST_FLAG(name) = (default_val)
+#define GTEST_DEFINE_int32_(name, default_val, doc) \
+    GTEST_API_ ::testing::internal::Int32 GTEST_FLAG(name) = (default_val)
+#define GTEST_DEFINE_string_(name, default_val, doc) \
+    GTEST_API_ ::std::string GTEST_FLAG(name) = (default_val)
+
+#endif  // !defined(GTEST_DECLARE_bool_)
+
+// Thread annotations
+#if !defined(GTEST_EXCLUSIVE_LOCK_REQUIRED_)
+# define GTEST_EXCLUSIVE_LOCK_REQUIRED_(locks)
+# define GTEST_LOCK_EXCLUDED_(locks)
+#endif  // !defined(GTEST_EXCLUSIVE_LOCK_REQUIRED_)
+
+// Parses 'str' for a 32-bit signed integer.  If successful, writes the result
+// to *value and returns true; otherwise leaves *value unchanged and returns
+// false.
+// TODO(chandlerc): Find a better way to refactor flag and environment parsing
+// out of both gtest-port.cc and gtest.cc to avoid exporting this utility
+// function.
+        bool ParseInt32(const Message &src_text, const char *str, Int32 *value);
+
+// Parses a bool/Int32/string from the environment variable
+// corresponding to the given Google Test flag.
+        bool BoolFromGTestEnv(const char *flag, bool default_val);
+        GTEST_API_ Int32 Int32FromGTestEnv(const char *flag, Int32 default_val);
+        std::string StringFromGTestEnv(const char *flag, const char *default_val);
+
+    }  // namespace internal
+
+// Returns a path to temporary directory.
+// Tries to determine an appropriate directory for the platform.
+    GTEST_API_ std::string TempDir();
+
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PORT_H_
+
+#if GTEST_OS_LINUX
+# include <stdlib.h>
+# include <sys/types.h>
+# include <sys/wait.h>
+# include <unistd.h>
+#endif  // GTEST_OS_LINUX
+
+#if GTEST_HAS_EXCEPTIONS
+# include <stdexcept>
+#endif
+
+#include <ctype.h>
+#include <float.h>
+#include <string.h>
+#include <iomanip>
+#include <limits>
+#include <map>
+#include <set>
+#include <string>
+#include <vector>
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the Message class.
+//
+// IMPORTANT NOTE: Due to limitation of the C++ language, we have to
+// leave some internal implementation details in this header file.
+// They are clearly marked by comments like this:
+//
+//   // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+//
+// Such code is NOT meant to be used by a user directly, and is subject
+// to CHANGE WITHOUT NOTICE.  Therefore DO NOT DEPEND ON IT in a user
+// program!
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+#define GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+
+#include <limits>
+
+
+// Ensures that there is at least one operator<< in the global namespace.
+// See Message& operator<<(...) below for why.
+void operator<<(const testing::internal::Secret &, int);
+
+namespace testing
+{
+
+// The Message class works like an ostream repeater.
+//
+// Typical usage:
+//
+//   1. You stream a bunch of values to a Message object.
+//      It will remember the text in a stringstream.
+//   2. Then you stream the Message object to an ostream.
+//      This causes the text in the Message to be streamed
+//      to the ostream.
+//
+// For example;
+//
+//   testing::Message foo;
+//   foo << 1 << " != " << 2;
+//   std::cout << foo;
+//
+// will print "1 != 2".
+//
+// Message is not intended to be inherited from.  In particular, its
+// destructor is not virtual.
+//
+// Note that stringstream behaves differently in gcc and in MSVC.  You
+// can stream a NULL char pointer to it in the former, but not in the
+// latter (it causes an access violation if you do).  The Message
+// class hides this difference by treating a NULL char pointer as
+// "(null)".
+    class GTEST_API_ Message
+    {
+    private:
+        // The type of basic IO manipulators (endl, ends, and flush) for
+        // narrow streams.
+        typedef std::ostream &(*BasicNarrowIoManip)(std::ostream &);
+
+    public:
+        // Constructs an empty Message.
+        Message();
+
+        // Copy constructor.
+        Message(const Message &msg) : ss_(new ::std::stringstream)    // NOLINT
+        {
+            *ss_ << msg.GetString();
+        }
+
+        // Constructs a Message from a C-string.
+        explicit Message(const char *str) : ss_(new ::std::stringstream)
+        {
+            *ss_ << str;
+        }
+
+#if GTEST_OS_SYMBIAN
+        // Streams a value (either a pointer or not) to this object.
+        template <typename T>
+        inline Message &operator <<(const T &value)
+        {
+            StreamHelper(typename internal::is_pointer<T>::type(), value);
+            return *this;
+        }
+#else
+        // Streams a non-pointer value to this object.
+        template <typename T>
+        inline Message &operator <<(const T &val)
+        {
+            // Some libraries overload << for STL containers.  These
+            // overloads are defined in the global namespace instead of ::std.
+            //
+            // C++'s symbol lookup rule (i.e. Koenig lookup) says that these
+            // overloads are visible in either the std namespace or the global
+            // namespace, but not other namespaces, including the testing
+            // namespace which Google Test's Message class is in.
+            //
+            // To allow STL containers (and other types that has a << operator
+            // defined in the global namespace) to be used in Google Test
+            // assertions, testing::Message must access the custom << operator
+            // from the global namespace.  With this using declaration,
+            // overloads of << defined in the global namespace and those
+            // visible via Koenig lookup are both exposed in this function.
+            using ::operator <<;
+            *ss_ << val;
+            return *this;
+        }
+
+        // Streams a pointer value to this object.
+        //
+        // This function is an overload of the previous one.  When you
+        // stream a pointer to a Message, this definition will be used as it
+        // is more specialized.  (The C++ Standard, section
+        // [temp.func.order].)  If you stream a non-pointer, then the
+        // previous definition will be used.
+        //
+        // The reason for this overload is that streaming a NULL pointer to
+        // ostream is undefined behavior.  Depending on the compiler, you
+        // may get "0", "(nil)", "(null)", or an access violation.  To
+        // ensure consistent result across compilers, we always treat NULL
+        // as "(null)".
+        template <typename T>
+        inline Message &operator <<(T *const &pointer)    // NOLINT
+        {
+            if (pointer == NULL) {
+                *ss_ << "(null)";
+            } else {
+                *ss_ << pointer;
+            }
+            return *this;
+        }
+#endif  // GTEST_OS_SYMBIAN
+
+        // Since the basic IO manipulators are overloaded for both narrow
+        // and wide streams, we have to provide this specialized definition
+        // of operator <<, even though its body is the same as the
+        // templatized version above.  Without this definition, streaming
+        // endl or other basic IO manipulators to Message will confuse the
+        // compiler.
+        Message &operator <<(BasicNarrowIoManip val)
+        {
+            *ss_ << val;
+            return *this;
+        }
+
+        // Instead of 1/0, we want to see true/false for bool values.
+        Message &operator <<(bool b)
+        {
+            return *this << (b ? "true" : "false");
+        }
+
+        // These two overloads allow streaming a wide C string to a Message
+        // using the UTF-8 encoding.
+        Message &operator <<(const wchar_t *wide_c_str);
+        Message &operator <<(wchar_t *wide_c_str);
+
+#if GTEST_HAS_STD_WSTRING
+        // Converts the given wide string to a narrow string using the UTF-8
+        // encoding, and streams the result to this Message object.
+        Message &operator <<(const ::std::wstring &wstr);
+#endif  // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_GLOBAL_WSTRING
+        // Converts the given wide string to a narrow string using the UTF-8
+        // encoding, and streams the result to this Message object.
+        Message &operator <<(const ::wstring &wstr);
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+        // Gets the text streamed to this object so far as an std::string.
+        // Each '\0' character in the buffer is replaced with "\\0".
+        //
+        // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        std::string GetString() const;
+
+    private:
+
+#if GTEST_OS_SYMBIAN
+        // These are needed as the Nokia Symbian Compiler cannot decide between
+        // const T& and const T* in a function template. The Nokia compiler _can_
+        // decide between class template specializations for T and T*, so a
+        // tr1::type_traits-like is_pointer works, and we can overload on that.
+        template <typename T>
+        inline void StreamHelper(internal::true_type /*is_pointer*/, T *pointer)
+        {
+            if (pointer == NULL) {
+                *ss_ << "(null)";
+            } else {
+                *ss_ << pointer;
+            }
+        }
+        template <typename T>
+        inline void StreamHelper(internal::false_type /*is_pointer*/,
+                                 const T &value)
+        {
+            // See the comments in Message& operator <<(const T&) above for why
+            // we need this using statement.
+            using ::operator <<;
+            *ss_ << value;
+        }
+#endif  // GTEST_OS_SYMBIAN
+
+        // We'll hold the text streamed to this object here.
+        const internal::scoped_ptr< ::std::stringstream> ss_;
+
+        // We declare (but don't implement) this to prevent the compiler
+        // from implementing the assignment operator.
+        void operator=(const Message &);
+    };
+
+// Streams a Message to an ostream.
+    inline std::ostream &operator <<(std::ostream &os, const Message &sb)
+    {
+        return os << sb.GetString();
+    }
+
+    namespace internal
+    {
+
+// Converts a streamable value to an std::string.  A NULL pointer is
+// converted to "(null)".  When the input value is a ::string,
+// ::std::string, ::wstring, or ::std::wstring object, each NUL
+// character in it is replaced with "\\0".
+        template <typename T>
+        std::string StreamableToString(const T &streamable)
+        {
+            return (Message() << streamable).GetString();
+        }
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_MESSAGE_H_
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file declares the String class and functions used internally by
+// Google Test.  They are subject to change without notice. They should not used
+// by code external to Google Test.
+//
+// This header file is #included by <gtest/internal/gtest-internal.h>.
+// It should not be #included by other files.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+
+#ifdef __BORLANDC__
+// string.h is not guaranteed to provide strcpy on C++ Builder.
+# include <mem.h>
+#endif
+
+#include <string.h>
+#include <string>
+
+
+namespace testing
+{
+    namespace internal
+    {
+
+// String - an abstract class holding static string utilities.
+        class GTEST_API_ String
+        {
+        public:
+            // Static utility methods
+
+            // Clones a 0-terminated C string, allocating memory using new.  The
+            // caller is responsible for deleting the return value using
+            // delete[].  Returns the cloned string, or NULL if the input is
+            // NULL.
+            //
+            // This is different from strdup() in string.h, which allocates
+            // memory using malloc().
+            static const char *CloneCString(const char *c_str);
+
+#if GTEST_OS_WINDOWS_MOBILE
+            // Windows CE does not have the 'ANSI' versions of Win32 APIs. To be
+            // able to pass strings to Win32 APIs on CE we need to convert them
+            // to 'Unicode', UTF-16.
+
+            // Creates a UTF-16 wide string from the given ANSI string, allocating
+            // memory using new. The caller is responsible for deleting the return
+            // value using delete[]. Returns the wide string, or NULL if the
+            // input is NULL.
+            //
+            // The wide string is created using the ANSI codepage (CP_ACP) to
+            // match the behaviour of the ANSI versions of Win32 calls and the
+            // C runtime.
+            static LPCWSTR AnsiToUtf16(const char *c_str);
+
+            // Creates an ANSI string from the given wide string, allocating
+            // memory using new. The caller is responsible for deleting the return
+            // value using delete[]. Returns the ANSI string, or NULL if the
+            // input is NULL.
+            //
+            // The returned string is created using the ANSI codepage (CP_ACP) to
+            // match the behaviour of the ANSI versions of Win32 calls and the
+            // C runtime.
+            static const char *Utf16ToAnsi(LPCWSTR utf16_str);
+#endif
+
+            // Compares two C strings.  Returns true iff they have the same content.
+            //
+            // Unlike strcmp(), this function can handle NULL argument(s).  A
+            // NULL C string is considered different to any non-NULL C string,
+            // including the empty string.
+            static bool CStringEquals(const char *lhs, const char *rhs);
+
+            // Converts a wide C string to a String using the UTF-8 encoding.
+            // NULL will be converted to "(null)".  If an error occurred during
+            // the conversion, "(failed to convert from wide string)" is
+            // returned.
+            static std::string ShowWideCString(const wchar_t *wide_c_str);
+
+            // Compares two wide C strings.  Returns true iff they have the same
+            // content.
+            //
+            // Unlike wcscmp(), this function can handle NULL argument(s).  A
+            // NULL C string is considered different to any non-NULL C string,
+            // including the empty string.
+            static bool WideCStringEquals(const wchar_t *lhs, const wchar_t *rhs);
+
+            // Compares two C strings, ignoring case.  Returns true iff they
+            // have the same content.
+            //
+            // Unlike strcasecmp(), this function can handle NULL argument(s).
+            // A NULL C string is considered different to any non-NULL C string,
+            // including the empty string.
+            static bool CaseInsensitiveCStringEquals(const char *lhs,
+                                                     const char *rhs);
+
+            // Compares two wide C strings, ignoring case.  Returns true iff they
+            // have the same content.
+            //
+            // Unlike wcscasecmp(), this function can handle NULL argument(s).
+            // A NULL C string is considered different to any non-NULL wide C string,
+            // including the empty string.
+            // NB: The implementations on different platforms slightly differ.
+            // On windows, this method uses _wcsicmp which compares according to LC_CTYPE
+            // environment variable. On GNU platform this method uses wcscasecmp
+            // which compares according to LC_CTYPE category of the current locale.
+            // On MacOS X, it uses towlower, which also uses LC_CTYPE category of the
+            // current locale.
+            static bool CaseInsensitiveWideCStringEquals(const wchar_t *lhs,
+                                                         const wchar_t *rhs);
+
+            // Returns true iff the given string ends with the given suffix, ignoring
+            // case. Any string is considered to end with an empty suffix.
+            static bool EndsWithCaseInsensitive(
+                const std::string &str, const std::string &suffix);
+
+            // Formats an int value as "%02d".
+            static std::string FormatIntWidth2(int value);  // "%02d" for width == 2
+
+            // Formats an int value as "%X".
+            static std::string FormatHexInt(int value);
+
+            // Formats a byte as "%02X".
+            static std::string FormatByte(unsigned char value);
+
+        private:
+            String();  // Not meant to be instantiated.
+        };  // class String
+
+// Gets the content of the stringstream's buffer as an std::string.  Each '\0'
+// character in the buffer is replaced with "\\0".
+        GTEST_API_ std::string StringStreamToString(::std::stringstream *stream);
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_STRING_H_
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: keith.ray@gmail.com (Keith Ray)
+//
+// Google Test filepath utilities
+//
+// This header file declares classes and functions used internally by
+// Google Test.  They are subject to change without notice.
+//
+// This file is #included in <gtest/internal/gtest-internal.h>.
+// Do not include this header file separately!
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+
+
+namespace testing
+{
+    namespace internal
+    {
+
+// FilePath - a class for file and directory pathname manipulation which
+// handles platform-specific conventions (like the pathname separator).
+// Used for helper functions for naming files in a directory for xml output.
+// Except for Set methods, all methods are const or static, which provides an
+// "immutable value object" -- useful for peace of mind.
+// A FilePath with a value ending in a path separator ("like/this/") represents
+// a directory, otherwise it is assumed to represent a file. In either case,
+// it may or may not represent an actual file or directory in the file system.
+// Names are NOT checked for syntax correctness -- no checking for illegal
+// characters, malformed paths, etc.
+
+        class GTEST_API_ FilePath
+        {
+        public:
+            FilePath() : pathname_("") { }
+            FilePath(const FilePath &rhs) : pathname_(rhs.pathname_) { }
+
+            explicit FilePath(const std::string &pathname) : pathname_(pathname)
+            {
+                Normalize();
+            }
+
+            FilePath &operator=(const FilePath &rhs)
+            {
+                Set(rhs);
+                return *this;
+            }
+
+            void Set(const FilePath &rhs)
+            {
+                pathname_ = rhs.pathname_;
+            }
+
+            const std::string &string() const
+            {
+                return pathname_;
+            }
+            const char *c_str() const
+            {
+                return pathname_.c_str();
+            }
+
+            // Returns the current working directory, or "" if unsuccessful.
+            static FilePath GetCurrentDir();
+
+            // Given directory = "dir", base_name = "test", number = 0,
+            // extension = "xml", returns "dir/test.xml". If number is greater
+            // than zero (e.g., 12), returns "dir/test_12.xml".
+            // On Windows platform, uses \ as the separator rather than /.
+            static FilePath MakeFileName(const FilePath &directory,
+                                         const FilePath &base_name,
+                                         int number,
+                                         const char *extension);
+
+            // Given directory = "dir", relative_path = "test.xml",
+            // returns "dir/test.xml".
+            // On Windows, uses \ as the separator rather than /.
+            static FilePath ConcatPaths(const FilePath &directory,
+                                        const FilePath &relative_path);
+
+            // Returns a pathname for a file that does not currently exist. The pathname
+            // will be directory/base_name.extension or
+            // directory/base_name_<number>.extension if directory/base_name.extension
+            // already exists. The number will be incremented until a pathname is found
+            // that does not already exist.
+            // Examples: 'dir/foo_test.xml' or 'dir/foo_test_1.xml'.
+            // There could be a race condition if two or more processes are calling this
+            // function at the same time -- they could both pick the same filename.
+            static FilePath GenerateUniqueFileName(const FilePath &directory,
+                                                   const FilePath &base_name,
+                                                   const char *extension);
+
+            // Returns true iff the path is "".
+            bool IsEmpty() const
+            {
+                return pathname_.empty();
+            }
+
+            // If input name has a trailing separator character, removes it and returns
+            // the name, otherwise return the name string unmodified.
+            // On Windows platform, uses \ as the separator, other platforms use /.
+            FilePath RemoveTrailingPathSeparator() const;
+
+            // Returns a copy of the FilePath with the directory part removed.
+            // Example: FilePath("path/to/file").RemoveDirectoryName() returns
+            // FilePath("file"). If there is no directory part ("just_a_file"), it returns
+            // the FilePath unmodified. If there is no file part ("just_a_dir/") it
+            // returns an empty FilePath ("").
+            // On Windows platform, '\' is the path separator, otherwise it is '/'.
+            FilePath RemoveDirectoryName() const;
+
+            // RemoveFileName returns the directory path with the filename removed.
+            // Example: FilePath("path/to/file").RemoveFileName() returns "path/to/".
+            // If the FilePath is "a_file" or "/a_file", RemoveFileName returns
+            // FilePath("./") or, on Windows, FilePath(".\\"). If the filepath does
+            // not have a file, like "just/a/dir/", it returns the FilePath unmodified.
+            // On Windows platform, '\' is the path separator, otherwise it is '/'.
+            FilePath RemoveFileName() const;
+
+            // Returns a copy of the FilePath with the case-insensitive extension removed.
+            // Example: FilePath("dir/file.exe").RemoveExtension("EXE") returns
+            // FilePath("dir/file"). If a case-insensitive extension is not
+            // found, returns a copy of the original FilePath.
+            FilePath RemoveExtension(const char *extension) const;
+
+            // Creates directories so that path exists. Returns true if successful or if
+            // the directories already exist; returns false if unable to create
+            // directories for any reason. Will also return false if the FilePath does
+            // not represent a directory (that is, it doesn't end with a path separator).
+            bool CreateDirectoriesRecursively() const;
+
+            // Create the directory so that path exists. Returns true if successful or
+            // if the directory already exists; returns false if unable to create the
+            // directory for any reason, including if the parent directory does not
+            // exist. Not named "CreateDirectory" because that's a macro on Windows.
+            bool CreateFolder() const;
+
+            // Returns true if FilePath describes something in the file-system,
+            // either a file, directory, or whatever, and that something exists.
+            bool FileOrDirectoryExists() const;
+
+            // Returns true if pathname describes a directory in the file-system
+            // that exists.
+            bool DirectoryExists() const;
+
+            // Returns true if FilePath ends with a path separator, which indicates that
+            // it is intended to represent a directory. Returns false otherwise.
+            // This does NOT check that a directory (or file) actually exists.
+            bool IsDirectory() const;
+
+            // Returns true if pathname describes a root directory. (Windows has one
+            // root directory per disk drive.)
+            bool IsRootDirectory() const;
+
+            // Returns true if pathname describes an absolute path.
+            bool IsAbsolutePath() const;
+
+        private:
+            // Replaces multiple consecutive separators with a single separator.
+            // For example, "bar///foo" becomes "bar/foo". Does not eliminate other
+            // redundancies that might be in a pathname involving "." or "..".
+            //
+            // A pathname with multiple consecutive separators may occur either through
+            // user error or as a result of some scripts or APIs that generate a pathname
+            // with a trailing separator. On other platforms the same API or script
+            // may NOT generate a pathname with a trailing "/". Then elsewhere that
+            // pathname may have another "/" and pathname components added to it,
+            // without checking for the separator already being there.
+            // The script language and operating system may allow paths like "foo//bar"
+            // but some of the functions in FilePath will not handle that correctly. In
+            // particular, RemoveTrailingPathSeparator() only removes one separator, and
+            // it is called in CreateDirectoriesRecursively() assuming that it will change
+            // a pathname from directory syntax (trailing separator) to filename syntax.
+            //
+            // On Windows this method also replaces the alternate path separator '/' with
+            // the primary path separator '\\', so that for example "bar\\/\\foo" becomes
+            // "bar\\foo".
+
+            void Normalize();
+
+            // Returns a pointer to the last occurrence of a valid path separator in
+            // the FilePath. On Windows, for example, both '/' and '\' are valid path
+            // separators. Returns NULL if no path separator was found.
+            const char *FindLastPathSeparator() const;
+
+            std::string pathname_;
+        };  // class FilePath
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_FILEPATH_H_
+// This file was GENERATED by command:
+//     pump.py gtest-type-util.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Type utilities needed for implementing typed and type-parameterized
+// tests.  This file is generated by a SCRIPT.  DO NOT EDIT BY HAND!
+//
+// Currently we support at most 50 types in a list, and at most 50
+// type-parameterized tests in one type-parameterized test case.
+// Please contact googletestframework@googlegroups.com if you need
+// more.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+
+
+// #ifdef __GNUC__ is too general here.  It is possible to use gcc without using
+// libstdc++ (which is where cxxabi.h comes from).
+# if GTEST_HAS_CXXABI_H_
+#  include <cxxabi.h>
+# elif defined(__HP_aCC)
+#  include <acxx_demangle.h>
+# endif  // GTEST_HASH_CXXABI_H_
+
+namespace testing
+{
+    namespace internal
+    {
+
+// GetTypeName<T>() returns a human-readable name of type T.
+// NB: This function is also used in Google Mock, so don't move it inside of
+// the typed-test-only section below.
+        template <typename T>
+        std::string GetTypeName()
+        {
+# if GTEST_HAS_RTTI
+
+            const char *const name = typeid(T).name();
+#  if GTEST_HAS_CXXABI_H_ || defined(__HP_aCC)
+            int status = 0;
+            // gcc's implementation of typeid(T).name() mangles the type name,
+            // so we have to demangle it.
+#   if GTEST_HAS_CXXABI_H_
+            using abi::__cxa_demangle;
+#   endif  // GTEST_HAS_CXXABI_H_
+            char *const readable_name = __cxa_demangle(name, 0, 0, &status);
+            const std::string name_str(status == 0 ? readable_name : name);
+            free(readable_name);
+            return name_str;
+#  else
+            return name;
+#  endif  // GTEST_HAS_CXXABI_H_ || __HP_aCC
+
+# else
+
+            return "<type>";
+
+# endif  // GTEST_HAS_RTTI
+        }
+
+#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// AssertyTypeEq<T1, T2>::type is defined iff T1 and T2 are the same
+// type.  This can be used as a compile-time assertion to ensure that
+// two types are equal.
+
+        template <typename T1, typename T2>
+        struct AssertTypeEq;
+
+        template <typename T>
+        struct AssertTypeEq<T, T> {
+            typedef bool type;
+        };
+
+// A unique type used as the default value for the arguments of class
+// template Types.  This allows us to simulate variadic templates
+// (e.g. Types<int>, Type<int, double>, and etc), which C++ doesn't
+// support directly.
+        struct None {};
+
+// The following family of struct and struct templates are used to
+// represent type lists.  In particular, TypesN<T1, T2, ..., TN>
+// represents a type list with N types (T1, T2, ..., and TN) in it.
+// Except for Types0, every struct in the family has two member types:
+// Head for the first type in the list, and Tail for the rest of the
+// list.
+
+// The empty type list.
+        struct Types0 {};
+
+// Type lists of length 1, 2, 3, and so on.
+
+        template <typename T1>
+        struct Types1 {
+            typedef T1 Head;
+            typedef Types0 Tail;
+        };
+        template <typename T1, typename T2>
+        struct Types2 {
+            typedef T1 Head;
+            typedef Types1<T2> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3>
+        struct Types3 {
+            typedef T1 Head;
+            typedef Types2<T2, T3> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        struct Types4 {
+            typedef T1 Head;
+            typedef Types3<T2, T3, T4> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        struct Types5 {
+            typedef T1 Head;
+            typedef Types4<T2, T3, T4, T5> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        struct Types6 {
+            typedef T1 Head;
+            typedef Types5<T2, T3, T4, T5, T6> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        struct Types7 {
+            typedef T1 Head;
+            typedef Types6<T2, T3, T4, T5, T6, T7> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        struct Types8 {
+            typedef T1 Head;
+            typedef Types7<T2, T3, T4, T5, T6, T7, T8> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        struct Types9 {
+            typedef T1 Head;
+            typedef Types8<T2, T3, T4, T5, T6, T7, T8, T9> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        struct Types10 {
+            typedef T1 Head;
+            typedef Types9<T2, T3, T4, T5, T6, T7, T8, T9, T10> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11>
+        struct Types11 {
+            typedef T1 Head;
+            typedef Types10<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12>
+        struct Types12 {
+            typedef T1 Head;
+            typedef Types11<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13>
+        struct Types13 {
+            typedef T1 Head;
+            typedef Types12<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14>
+        struct Types14 {
+            typedef T1 Head;
+            typedef Types13<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15>
+        struct Types15 {
+            typedef T1 Head;
+            typedef Types14<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16>
+        struct Types16 {
+            typedef T1 Head;
+            typedef Types15<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17>
+        struct Types17 {
+            typedef T1 Head;
+            typedef Types16<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18>
+        struct Types18 {
+            typedef T1 Head;
+            typedef Types17<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19>
+        struct Types19 {
+            typedef T1 Head;
+            typedef Types18<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20>
+        struct Types20 {
+            typedef T1 Head;
+            typedef Types19<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21>
+        struct Types21 {
+            typedef T1 Head;
+            typedef Types20<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22>
+        struct Types22 {
+            typedef T1 Head;
+            typedef Types21<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23>
+        struct Types23 {
+            typedef T1 Head;
+            typedef Types22<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24>
+        struct Types24 {
+            typedef T1 Head;
+            typedef Types23<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25>
+        struct Types25 {
+            typedef T1 Head;
+            typedef Types24<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26>
+        struct Types26 {
+            typedef T1 Head;
+            typedef Types25<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27>
+        struct Types27 {
+            typedef T1 Head;
+            typedef Types26<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28>
+        struct Types28 {
+            typedef T1 Head;
+            typedef Types27<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29>
+        struct Types29 {
+            typedef T1 Head;
+            typedef Types28<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30>
+        struct Types30 {
+            typedef T1 Head;
+            typedef Types29<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31>
+        struct Types31 {
+            typedef T1 Head;
+            typedef Types30<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32>
+        struct Types32 {
+            typedef T1 Head;
+            typedef Types31<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33>
+        struct Types33 {
+            typedef T1 Head;
+            typedef Types32<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34>
+        struct Types34 {
+            typedef T1 Head;
+            typedef Types33<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35>
+        struct Types35 {
+            typedef T1 Head;
+            typedef Types34<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36>
+        struct Types36 {
+            typedef T1 Head;
+            typedef Types35<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37>
+        struct Types37 {
+            typedef T1 Head;
+            typedef Types36<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38>
+        struct Types38 {
+            typedef T1 Head;
+            typedef Types37<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39>
+        struct Types39 {
+            typedef T1 Head;
+            typedef Types38<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40>
+        struct Types40 {
+            typedef T1 Head;
+            typedef Types39<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41>
+        struct Types41 {
+            typedef T1 Head;
+            typedef Types40<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42>
+        struct Types42 {
+            typedef T1 Head;
+            typedef Types41<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43>
+        struct Types43 {
+            typedef T1 Head;
+            typedef Types42<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44>
+        struct Types44 {
+            typedef T1 Head;
+            typedef Types43<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45>
+        struct Types45 {
+            typedef T1 Head;
+            typedef Types44<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46>
+        struct Types46 {
+            typedef T1 Head;
+            typedef Types45<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47>
+        struct Types47 {
+            typedef T1 Head;
+            typedef Types46<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48>
+        struct Types48 {
+            typedef T1 Head;
+            typedef Types47<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47, T48> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49>
+        struct Types49 {
+            typedef T1 Head;
+            typedef Types48<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47, T48, T49> Tail;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49, typename T50>
+        struct Types50 {
+            typedef T1 Head;
+            typedef Types49<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+                    T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                    T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                    T44, T45, T46, T47, T48, T49, T50> Tail;
+        };
+
+
+    }  // namespace internal
+
+// We don't want to require the users to write TypesN<...> directly,
+// as that would require them to count the length.  Types<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Types<int>
+// will appear as Types<int, None, None, ..., None> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Types<T1, ..., TN>, and Google Test will translate
+// that to TypesN<T1, ..., TN> internally to make error messages
+// readable.  The translation is done by the 'type' member of the
+// Types template.
+    template <typename T1 = internal::None, typename T2 = internal::None,
+              typename T3 = internal::None, typename T4 = internal::None,
+              typename T5 = internal::None, typename T6 = internal::None,
+              typename T7 = internal::None, typename T8 = internal::None,
+              typename T9 = internal::None, typename T10 = internal::None,
+              typename T11 = internal::None, typename T12 = internal::None,
+              typename T13 = internal::None, typename T14 = internal::None,
+              typename T15 = internal::None, typename T16 = internal::None,
+              typename T17 = internal::None, typename T18 = internal::None,
+              typename T19 = internal::None, typename T20 = internal::None,
+              typename T21 = internal::None, typename T22 = internal::None,
+              typename T23 = internal::None, typename T24 = internal::None,
+              typename T25 = internal::None, typename T26 = internal::None,
+              typename T27 = internal::None, typename T28 = internal::None,
+              typename T29 = internal::None, typename T30 = internal::None,
+              typename T31 = internal::None, typename T32 = internal::None,
+              typename T33 = internal::None, typename T34 = internal::None,
+              typename T35 = internal::None, typename T36 = internal::None,
+              typename T37 = internal::None, typename T38 = internal::None,
+              typename T39 = internal::None, typename T40 = internal::None,
+              typename T41 = internal::None, typename T42 = internal::None,
+              typename T43 = internal::None, typename T44 = internal::None,
+              typename T45 = internal::None, typename T46 = internal::None,
+              typename T47 = internal::None, typename T48 = internal::None,
+              typename T49 = internal::None, typename T50 = internal::None>
+    struct Types {
+        typedef internal::Types50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47, T48, T49, T50> type;
+    };
+
+    template <>
+    struct Types<internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types0 type;
+    };
+    template <typename T1>
+    struct Types<T1, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types1<T1> type;
+    };
+    template <typename T1, typename T2>
+    struct Types<T1, T2, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types2<T1, T2> type;
+    };
+    template <typename T1, typename T2, typename T3>
+    struct Types<T1, T2, T3, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types3<T1, T2, T3> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4>
+    struct Types<T1, T2, T3, T4, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types4<T1, T2, T3, T4> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5>
+    struct Types<T1, T2, T3, T4, T5, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types5<T1, T2, T3, T4, T5> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6>
+    struct Types<T1, T2, T3, T4, T5, T6, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types6<T1, T2, T3, T4, T5, T6> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types7<T1, T2, T3, T4, T5, T6, T7> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types8<T1, T2, T3, T4, T5, T6, T7, T8> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types9<T1, T2, T3, T4, T5, T6, T7, T8, T9> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+                T12> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+                T26> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+                T40> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, internal::None,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None, internal::None> {
+        typedef internal::Types43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None, internal::None> {
+        typedef internal::Types44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               internal::None, internal::None, internal::None, internal::None,
+               internal::None> {
+        typedef internal::Types45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, internal::None, internal::None, internal::None, internal::None> {
+        typedef internal::Types46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, T47, internal::None, internal::None, internal::None> {
+        typedef internal::Types47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, T47, T48, internal::None, internal::None> {
+        typedef internal::Types48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47, T48> type;
+    };
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48, typename T49>
+    struct Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15,
+               T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29, T30,
+               T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44, T45,
+               T46, T47, T48, T49, internal::None> {
+        typedef internal::Types49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                T41, T42, T43, T44, T45, T46, T47, T48, T49> type;
+    };
+
+    namespace internal
+    {
+
+# define GTEST_TEMPLATE_ template <typename T> class
+
+// The template "selector" struct TemplateSel<Tmpl> is used to
+// represent Tmpl, which must be a class template with one type
+// parameter, as a type.  TemplateSel<Tmpl>::Bind<T>::type is defined
+// as the type Tmpl<T>.  This allows us to actually instantiate the
+// template "selected" by TemplateSel<Tmpl>.
+//
+// This trick is necessary for simulating typedef for class templates,
+// which C++ doesn't support directly.
+        template <GTEST_TEMPLATE_ Tmpl>
+        struct TemplateSel {
+            template <typename T>
+            struct Bind {
+                typedef Tmpl<T> type;
+            };
+        };
+
+# define GTEST_BIND_(TmplSel, T) \
+  TmplSel::template Bind<T>::type
+
+// A unique struct template used as the default value for the
+// arguments of class template Templates.  This allows us to simulate
+// variadic templates (e.g. Templates<int>, Templates<int, double>,
+// and etc), which C++ doesn't support directly.
+        template <typename T>
+        struct NoneT {};
+
+// The following family of struct and struct templates are used to
+// represent template lists.  In particular, TemplatesN<T1, T2, ...,
+// TN> represents a list of N templates (T1, T2, ..., and TN).  Except
+// for Templates0, every struct in the family has two member types:
+// Head for the selector of the first template in the list, and Tail
+// for the rest of the list.
+
+// The empty template list.
+        struct Templates0 {};
+
+// Template lists of length 1, 2, 3, and so on.
+
+        template <GTEST_TEMPLATE_ T1>
+        struct Templates1 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates0 Tail;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2>
+        struct Templates2 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates1<T2> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3>
+        struct Templates3 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates2<T2, T3> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4>
+        struct Templates4 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates3<T2, T3, T4> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5>
+        struct Templates5 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates4<T2, T3, T4, T5> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6>
+        struct Templates6 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates5<T2, T3, T4, T5, T6> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7>
+        struct Templates7 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates6<T2, T3, T4, T5, T6, T7> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8>
+        struct Templates8 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates7<T2, T3, T4, T5, T6, T7, T8> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9>
+        struct Templates9 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates8<T2, T3, T4, T5, T6, T7, T8, T9> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10>
+        struct Templates10 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates9<T2, T3, T4, T5, T6, T7, T8, T9, T10> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11>
+        struct Templates11 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates10<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12>
+        struct Templates12 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates11<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13>
+        struct Templates13 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates12<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14>
+        struct Templates14 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates13<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15>
+        struct Templates15 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates14<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16>
+        struct Templates16 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates15<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17>
+        struct Templates17 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates16<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18>
+        struct Templates18 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates17<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19>
+        struct Templates19 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates18<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20>
+        struct Templates20 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates19<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21>
+        struct Templates21 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates20<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22>
+        struct Templates22 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates21<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23>
+        struct Templates23 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates22<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24>
+        struct Templates24 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates23<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25>
+        struct Templates25 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates24<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26>
+        struct Templates26 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates25<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27>
+        struct Templates27 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates26<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28>
+        struct Templates28 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates27<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29>
+        struct Templates29 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates28<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30>
+        struct Templates30 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates29<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31>
+        struct Templates31 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates30<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32>
+        struct Templates32 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates31<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33>
+        struct Templates33 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates32<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34>
+        struct Templates34 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates33<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35>
+        struct Templates35 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates34<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36>
+        struct Templates36 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates35<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37>
+        struct Templates37 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates36<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38>
+        struct Templates38 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates37<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39>
+        struct Templates39 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates38<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40>
+        struct Templates40 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates39<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41>
+        struct Templates41 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates40<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42>
+        struct Templates42 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates41<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43>
+        struct Templates43 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates42<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44>
+        struct Templates44 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates43<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45>
+        struct Templates45 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates44<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46>
+        struct Templates46 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates45<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47>
+        struct Templates47 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates46<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48>
+        struct Templates48 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates47<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47, T48> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+                  GTEST_TEMPLATE_ T49>
+        struct Templates49 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates48<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47, T48, T49> Tail;
+        };
+
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+                  GTEST_TEMPLATE_ T49, GTEST_TEMPLATE_ T50>
+        struct Templates50 {
+            typedef TemplateSel<T1> Head;
+            typedef Templates49<T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                    T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                    T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+                    T43, T44, T45, T46, T47, T48, T49, T50> Tail;
+        };
+
+
+// We don't want to require the users to write TemplatesN<...> directly,
+// as that would require them to count the length.  Templates<...> is much
+// easier to write, but generates horrible messages when there is a
+// compiler error, as gcc insists on printing out each template
+// argument, even if it has the default value (this means Templates<list>
+// will appear as Templates<list, NoneT, NoneT, ..., NoneT> in the compiler
+// errors).
+//
+// Our solution is to combine the best part of the two approaches: a
+// user would write Templates<T1, ..., TN>, and Google Test will translate
+// that to TemplatesN<T1, ..., TN> internally to make error messages
+// readable.  The translation is done by the 'type' member of the
+// Templates template.
+        template <GTEST_TEMPLATE_ T1 = NoneT, GTEST_TEMPLATE_ T2 = NoneT,
+                  GTEST_TEMPLATE_ T3 = NoneT, GTEST_TEMPLATE_ T4 = NoneT,
+                  GTEST_TEMPLATE_ T5 = NoneT, GTEST_TEMPLATE_ T6 = NoneT,
+                  GTEST_TEMPLATE_ T7 = NoneT, GTEST_TEMPLATE_ T8 = NoneT,
+                  GTEST_TEMPLATE_ T9 = NoneT, GTEST_TEMPLATE_ T10 = NoneT,
+                  GTEST_TEMPLATE_ T11 = NoneT, GTEST_TEMPLATE_ T12 = NoneT,
+                  GTEST_TEMPLATE_ T13 = NoneT, GTEST_TEMPLATE_ T14 = NoneT,
+                  GTEST_TEMPLATE_ T15 = NoneT, GTEST_TEMPLATE_ T16 = NoneT,
+                  GTEST_TEMPLATE_ T17 = NoneT, GTEST_TEMPLATE_ T18 = NoneT,
+                  GTEST_TEMPLATE_ T19 = NoneT, GTEST_TEMPLATE_ T20 = NoneT,
+                  GTEST_TEMPLATE_ T21 = NoneT, GTEST_TEMPLATE_ T22 = NoneT,
+                  GTEST_TEMPLATE_ T23 = NoneT, GTEST_TEMPLATE_ T24 = NoneT,
+                  GTEST_TEMPLATE_ T25 = NoneT, GTEST_TEMPLATE_ T26 = NoneT,
+                  GTEST_TEMPLATE_ T27 = NoneT, GTEST_TEMPLATE_ T28 = NoneT,
+                  GTEST_TEMPLATE_ T29 = NoneT, GTEST_TEMPLATE_ T30 = NoneT,
+                  GTEST_TEMPLATE_ T31 = NoneT, GTEST_TEMPLATE_ T32 = NoneT,
+                  GTEST_TEMPLATE_ T33 = NoneT, GTEST_TEMPLATE_ T34 = NoneT,
+                  GTEST_TEMPLATE_ T35 = NoneT, GTEST_TEMPLATE_ T36 = NoneT,
+                  GTEST_TEMPLATE_ T37 = NoneT, GTEST_TEMPLATE_ T38 = NoneT,
+                  GTEST_TEMPLATE_ T39 = NoneT, GTEST_TEMPLATE_ T40 = NoneT,
+                  GTEST_TEMPLATE_ T41 = NoneT, GTEST_TEMPLATE_ T42 = NoneT,
+                  GTEST_TEMPLATE_ T43 = NoneT, GTEST_TEMPLATE_ T44 = NoneT,
+                  GTEST_TEMPLATE_ T45 = NoneT, GTEST_TEMPLATE_ T46 = NoneT,
+                  GTEST_TEMPLATE_ T47 = NoneT, GTEST_TEMPLATE_ T48 = NoneT,
+                  GTEST_TEMPLATE_ T49 = NoneT, GTEST_TEMPLATE_ T50 = NoneT>
+        struct Templates {
+            typedef Templates50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47, T48, T49, T50> type;
+        };
+
+        template <>
+        struct Templates<NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates0 type;
+        };
+        template <GTEST_TEMPLATE_ T1>
+        struct Templates<T1, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates1<T1> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2>
+        struct Templates<T1, T2, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates2<T1, T2> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3>
+        struct Templates<T1, T2, T3, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates3<T1, T2, T3> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4>
+        struct Templates<T1, T2, T3, T4, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates4<T1, T2, T3, T4> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5>
+        struct Templates<T1, T2, T3, T4, T5, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates5<T1, T2, T3, T4, T5> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6>
+        struct Templates<T1, T2, T3, T4, T5, T6, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates6<T1, T2, T3, T4, T5, T6> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates7<T1, T2, T3, T4, T5, T6, T7> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates8<T1, T2, T3, T4, T5, T6, T7, T8> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates9<T1, T2, T3, T4, T5, T6, T7, T8, T9> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                    T13> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT> {
+            typedef Templates22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT> {
+            typedef Templates23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT> {
+            typedef Templates24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                    T27> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT> {
+            typedef Templates28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT> {
+            typedef Templates29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, NoneT, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, NoneT, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, NoneT, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, NoneT, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                    T41> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, NoneT,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   NoneT, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, NoneT, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, NoneT, NoneT, NoneT, NoneT> {
+            typedef Templates46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, T47, NoneT, NoneT, NoneT> {
+            typedef Templates47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, T47, T48, NoneT, NoneT> {
+            typedef Templates48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47, T48> type;
+        };
+        template <GTEST_TEMPLATE_ T1, GTEST_TEMPLATE_ T2, GTEST_TEMPLATE_ T3,
+                  GTEST_TEMPLATE_ T4, GTEST_TEMPLATE_ T5, GTEST_TEMPLATE_ T6,
+                  GTEST_TEMPLATE_ T7, GTEST_TEMPLATE_ T8, GTEST_TEMPLATE_ T9,
+                  GTEST_TEMPLATE_ T10, GTEST_TEMPLATE_ T11, GTEST_TEMPLATE_ T12,
+                  GTEST_TEMPLATE_ T13, GTEST_TEMPLATE_ T14, GTEST_TEMPLATE_ T15,
+                  GTEST_TEMPLATE_ T16, GTEST_TEMPLATE_ T17, GTEST_TEMPLATE_ T18,
+                  GTEST_TEMPLATE_ T19, GTEST_TEMPLATE_ T20, GTEST_TEMPLATE_ T21,
+                  GTEST_TEMPLATE_ T22, GTEST_TEMPLATE_ T23, GTEST_TEMPLATE_ T24,
+                  GTEST_TEMPLATE_ T25, GTEST_TEMPLATE_ T26, GTEST_TEMPLATE_ T27,
+                  GTEST_TEMPLATE_ T28, GTEST_TEMPLATE_ T29, GTEST_TEMPLATE_ T30,
+                  GTEST_TEMPLATE_ T31, GTEST_TEMPLATE_ T32, GTEST_TEMPLATE_ T33,
+                  GTEST_TEMPLATE_ T34, GTEST_TEMPLATE_ T35, GTEST_TEMPLATE_ T36,
+                  GTEST_TEMPLATE_ T37, GTEST_TEMPLATE_ T38, GTEST_TEMPLATE_ T39,
+                  GTEST_TEMPLATE_ T40, GTEST_TEMPLATE_ T41, GTEST_TEMPLATE_ T42,
+                  GTEST_TEMPLATE_ T43, GTEST_TEMPLATE_ T44, GTEST_TEMPLATE_ T45,
+                  GTEST_TEMPLATE_ T46, GTEST_TEMPLATE_ T47, GTEST_TEMPLATE_ T48,
+                  GTEST_TEMPLATE_ T49>
+        struct Templates<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14,
+                   T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28, T29,
+                   T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43, T44,
+                   T45, T46, T47, T48, T49, NoneT> {
+            typedef Templates49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                    T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+                    T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+                    T42, T43, T44, T45, T46, T47, T48, T49> type;
+        };
+
+// The TypeList template makes it possible to use either a single type
+// or a Types<...> list in TYPED_TEST_CASE() and
+// INSTANTIATE_TYPED_TEST_CASE_P().
+
+        template <typename T>
+        struct TypeList {
+            typedef Types1<T> type;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49, typename T50>
+        struct TypeList<Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+                   T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+                   T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+                   T44, T45, T46, T47, T48, T49, T50> > {
+            typedef typename Types<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+                    T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+                    T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+                    T41, T42, T43, T44, T45, T46, T47, T48, T49, T50>::type type;
+        };
+
+#endif  // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_TYPE_UTIL_H_
+
+// Due to C++ preprocessor weirdness, we need double indirection to
+// concatenate two tokens when one of them is __LINE__.  Writing
+//
+//   foo ## __LINE__
+//
+// will result in the token foo__LINE__, instead of foo followed by
+// the current line number.  For more details, see
+// http://www.parashift.com/c++-faq-lite/misc-technical-issues.html#faq-39.6
+#define GTEST_CONCAT_TOKEN_(foo, bar) GTEST_CONCAT_TOKEN_IMPL_(foo, bar)
+#define GTEST_CONCAT_TOKEN_IMPL_(foo, bar) foo ## bar
+
+class ProtocolMessage;
+namespace proto2
+{
+    class Message;
+}
+
+namespace testing
+{
+
+// Forward declarations.
+
+    class AssertionResult;                 // Result of an assertion.
+    class Message;                         // Represents a failure message.
+    class Test;                            // Represents a test.
+    class TestInfo;                        // Information about a test.
+    class TestPartResult;                  // Result of a test part.
+    class UnitTest;                        // A collection of test cases.
+
+    template <typename T>
+    ::std::string PrintToString(const T &value);
+
+    namespace internal
+    {
+
+        struct TraceInfo;                      // Information about a trace point.
+        class ScopedTrace;                     // Implements scoped trace.
+        class TestInfoImpl;                    // Opaque implementation of TestInfo
+        class UnitTestImpl;                    // Opaque implementation of UnitTest
+
+// The text used in failure messages to indicate the start of the
+// stack trace.
+        GTEST_API_ extern const char kStackTraceMarker[];
+
+// Two overloaded helpers for checking at compile time whether an
+// expression is a null pointer literal (i.e. NULL or any 0-valued
+// compile-time integral constant).  Their return values have
+// different sizes, so we can use sizeof() to test which version is
+// picked by the compiler.  These helpers have no implementations, as
+// we only need their signatures.
+//
+// Given IsNullLiteralHelper(x), the compiler will pick the first
+// version if x can be implicitly converted to Secret*, and pick the
+// second version otherwise.  Since Secret is a secret and incomplete
+// type, the only expression a user can write that has type Secret* is
+// a null pointer literal.  Therefore, we know that x is a null
+// pointer literal if and only if the first version is picked by the
+// compiler.
+        char IsNullLiteralHelper(Secret *p);
+        char (&IsNullLiteralHelper(...))[2];  // NOLINT
+
+// A compile-time bool constant that is true if and only if x is a
+// null pointer literal (i.e. NULL or any 0-valued compile-time
+// integral constant).
+#ifdef GTEST_ELLIPSIS_NEEDS_POD_
+// We lose support for NULL detection where the compiler doesn't like
+// passing non-POD classes through ellipsis (...).
+# define GTEST_IS_NULL_LITERAL_(x) false
+#else
+# define GTEST_IS_NULL_LITERAL_(x) \
+    (sizeof(::testing::internal::IsNullLiteralHelper(x)) == 1)
+#endif  // GTEST_ELLIPSIS_NEEDS_POD_
+
+// Appends the user-supplied message to the Google-Test-generated message.
+        GTEST_API_ std::string AppendUserMessage(
+            const std::string &gtest_msg, const Message &user_msg);
+
+#if GTEST_HAS_EXCEPTIONS
+
+// This exception is thrown by (and only by) a failed Google Test
+// assertion when GTEST_FLAG(throw_on_failure) is true (if exceptions
+// are enabled).  We derive it from std::runtime_error, which is for
+// errors presumably detectable only at run time.  Since
+// std::runtime_error inherits from std::exception, many testing
+// frameworks know how to extract and print the message inside it.
+        class GTEST_API_ GoogleTestFailureException : public ::std::runtime_error
+        {
+        public:
+            explicit GoogleTestFailureException(const TestPartResult &failure);
+        };
+
+#endif  // GTEST_HAS_EXCEPTIONS
+
+// A helper class for creating scoped traces in user programs.
+        class GTEST_API_ ScopedTrace
+        {
+        public:
+            // The c'tor pushes the given source file location and message onto
+            // a trace stack maintained by Google Test.
+            ScopedTrace(const char *file, int line, const Message &message);
+
+            // The d'tor pops the info pushed by the c'tor.
+            //
+            // Note that the d'tor is not virtual in order to be efficient.
+            // Don't inherit from ScopedTrace!
+            ~ScopedTrace();
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ScopedTrace);
+        } GTEST_ATTRIBUTE_UNUSED_;  // A ScopedTrace object does its job in its
+        // c'tor and d'tor.  Therefore it doesn't
+        // need to be used otherwise.
+
+        namespace edit_distance
+        {
+// Returns the optimal edits to go from 'left' to 'right'.
+// All edits cost the same, with replace having lower priority than
+// add/remove.
+// Simple implementation of the Wagner-Fischer algorithm.
+// See http://en.wikipedia.org/wiki/Wagner-Fischer_algorithm
+            enum EditType { kMatch, kAdd, kRemove, kReplace };
+            GTEST_API_ std::vector<EditType> CalculateOptimalEdits(
+                const std::vector<size_t> &left, const std::vector<size_t> &right);
+
+// Same as above, but the input is represented as strings.
+            GTEST_API_ std::vector<EditType> CalculateOptimalEdits(
+                const std::vector<std::string> &left,
+                const std::vector<std::string> &right);
+
+// Create a diff of the input strings in Unified diff format.
+            GTEST_API_ std::string CreateUnifiedDiff(const std::vector<std::string> &left,
+                                                     const std::vector<std::string> &right,
+                                                     size_t context = 2);
+
+        }  // namespace edit_distance
+
+// Calculate the diff between 'left' and 'right' and return it in unified diff
+// format.
+// If not null, stores in 'total_line_count' the total number of lines found
+// in left + right.
+        GTEST_API_ std::string DiffStrings(const std::string &left,
+                                           const std::string &right,
+                                           size_t *total_line_count);
+
+// Constructs and returns the message for an equality assertion
+// (e.g. ASSERT_EQ, EXPECT_STREQ, etc) failure.
+//
+// The first four parameters are the expressions used in the assertion
+// and their values, as strings.  For example, for ASSERT_EQ(foo, bar)
+// where foo is 5 and bar is 6, we have:
+//
+//   expected_expression: "foo"
+//   actual_expression:   "bar"
+//   expected_value:      "5"
+//   actual_value:        "6"
+//
+// The ignoring_case parameter is true iff the assertion is a
+// *_STRCASEEQ*.  When it's true, the string " (ignoring case)" will
+// be inserted into the message.
+        GTEST_API_ AssertionResult EqFailure(const char *expected_expression,
+                                             const char *actual_expression,
+                                             const std::string &expected_value,
+                                             const std::string &actual_value,
+                                             bool ignoring_case);
+
+// Constructs a failure message for Boolean assertions such as EXPECT_TRUE.
+        GTEST_API_ std::string GetBoolAssertionFailureMessage(
+            const AssertionResult &assertion_result,
+            const char *expression_text,
+            const char *actual_predicate_value,
+            const char *expected_predicate_value);
+
+// This template class represents an IEEE floating-point number
+// (either single-precision or double-precision, depending on the
+// template parameters).
+//
+// The purpose of this class is to do more sophisticated number
+// comparison.  (Due to round-off error, etc, it's very unlikely that
+// two floating-points will be equal exactly.  Hence a naive
+// comparison by the == operation often doesn't work.)
+//
+// Format of IEEE floating-point:
+//
+//   The most-significant bit being the leftmost, an IEEE
+//   floating-point looks like
+//
+//     sign_bit exponent_bits fraction_bits
+//
+//   Here, sign_bit is a single bit that designates the sign of the
+//   number.
+//
+//   For float, there are 8 exponent bits and 23 fraction bits.
+//
+//   For double, there are 11 exponent bits and 52 fraction bits.
+//
+//   More details can be found at
+//   http://en.wikipedia.org/wiki/IEEE_floating-point_standard.
+//
+// Template parameter:
+//
+//   RawType: the raw floating-point type (either float or double)
+        template <typename RawType>
+        class FloatingPoint
+        {
+        public:
+            // Defines the unsigned integer type that has the same size as the
+            // floating point number.
+            typedef typename TypeWithSize<sizeof(RawType)>::UInt Bits;
+
+            // Constants.
+
+            // # of bits in a number.
+            static const size_t kBitCount = 8*sizeof(RawType);
+
+            // # of fraction bits in a number.
+            static const size_t kFractionBitCount =
+                std::numeric_limits<RawType>::digits - 1;
+
+            // # of exponent bits in a number.
+            static const size_t kExponentBitCount = kBitCount - 1 - kFractionBitCount;
+
+            // The mask for the sign bit.
+            static const Bits kSignBitMask = static_cast<Bits>(1) << (kBitCount - 1);
+
+            // The mask for the fraction bits.
+            static const Bits kFractionBitMask =
+                ~static_cast<Bits>(0) >> (kExponentBitCount + 1);
+
+            // The mask for the exponent bits.
+            static const Bits kExponentBitMask = ~(kSignBitMask | kFractionBitMask);
+
+            // How many ULP's (Units in the Last Place) we want to tolerate when
+            // comparing two numbers.  The larger the value, the more error we
+            // allow.  A 0 value means that two numbers must be exactly the same
+            // to be considered equal.
+            //
+            // The maximum error of a single floating-point operation is 0.5
+            // units in the last place.  On Intel CPU's, all floating-point
+            // calculations are done with 80-bit precision, while double has 64
+            // bits.  Therefore, 4 should be enough for ordinary use.
+            //
+            // See the following article for more details on ULP:
+            // http://randomascii.wordpress.com/2012/02/25/comparing-floating-point-numbers-2012-edition/
+            static const size_t kMaxUlps = 4;
+
+            // Constructs a FloatingPoint from a raw floating-point number.
+            //
+            // On an Intel CPU, passing a non-normalized NAN (Not a Number)
+            // around may change its bits, although the new value is guaranteed
+            // to be also a NAN.  Therefore, don't expect this constructor to
+            // preserve the bits in x when x is a NAN.
+            explicit FloatingPoint(const RawType &x)
+            {
+                u_.value_ = x;
+            }
+
+            // Static methods
+
+            // Reinterprets a bit pattern as a floating-point number.
+            //
+            // This function is needed to test the AlmostEquals() method.
+            static RawType ReinterpretBits(const Bits bits)
+            {
+                FloatingPoint fp(0);
+                fp.u_.bits_ = bits;
+                return fp.u_.value_;
+            }
+
+            // Returns the floating-point number that represent positive infinity.
+            static RawType Infinity()
+            {
+                return ReinterpretBits(kExponentBitMask);
+            }
+
+            // Returns the maximum representable finite floating-point number.
+            static RawType Max();
+
+            // Non-static methods
+
+            // Returns the bits that represents this number.
+            const Bits &bits() const
+            {
+                return u_.bits_;
+            }
+
+            // Returns the exponent bits of this number.
+            Bits exponent_bits() const
+            {
+                return kExponentBitMask & u_.bits_;
+            }
+
+            // Returns the fraction bits of this number.
+            Bits fraction_bits() const
+            {
+                return kFractionBitMask & u_.bits_;
+            }
+
+            // Returns the sign bit of this number.
+            Bits sign_bit() const
+            {
+                return kSignBitMask & u_.bits_;
+            }
+
+            // Returns true iff this is NAN (not a number).
+            bool is_nan() const
+            {
+                // It's a NAN if the exponent bits are all ones and the fraction
+                // bits are not entirely zeros.
+                return (exponent_bits() == kExponentBitMask) && (fraction_bits() != 0);
+            }
+
+            // Returns true iff this number is at most kMaxUlps ULP's away from
+            // rhs.  In particular, this function:
+            //
+            //   - returns false if either number is (or both are) NAN.
+            //   - treats really large numbers as almost equal to infinity.
+            //   - thinks +0.0 and -0.0 are 0 DLP's apart.
+            bool AlmostEquals(const FloatingPoint &rhs) const
+            {
+                // The IEEE standard says that any comparison operation involving
+                // a NAN must return false.
+                if (is_nan() || rhs.is_nan()) return false;
+
+                return DistanceBetweenSignAndMagnitudeNumbers(u_.bits_, rhs.u_.bits_)
+                       <= kMaxUlps;
+            }
+
+        private:
+            // The data type used to store the actual floating-point number.
+            union FloatingPointUnion {
+                RawType value_;  // The raw floating-point number.
+                Bits bits_;      // The bits that represent the number.
+            };
+
+            // Converts an integer from the sign-and-magnitude representation to
+            // the biased representation.  More precisely, let N be 2 to the
+            // power of (kBitCount - 1), an integer x is represented by the
+            // unsigned number x + N.
+            //
+            // For instance,
+            //
+            //   -N + 1 (the most negative number representable using
+            //          sign-and-magnitude) is represented by 1;
+            //   0      is represented by N; and
+            //   N - 1  (the biggest number representable using
+            //          sign-and-magnitude) is represented by 2N - 1.
+            //
+            // Read http://en.wikipedia.org/wiki/Signed_number_representations
+            // for more details on signed number representations.
+            static Bits SignAndMagnitudeToBiased(const Bits &sam)
+            {
+                if (kSignBitMask & sam) {
+                    // sam represents a negative number.
+                    return ~sam + 1;
+                } else {
+                    // sam represents a positive number.
+                    return kSignBitMask | sam;
+                }
+            }
+
+            // Given two numbers in the sign-and-magnitude representation,
+            // returns the distance between them as an unsigned number.
+            static Bits DistanceBetweenSignAndMagnitudeNumbers(const Bits &sam1,
+                                                               const Bits &sam2)
+            {
+                const Bits biased1 = SignAndMagnitudeToBiased(sam1);
+                const Bits biased2 = SignAndMagnitudeToBiased(sam2);
+                return (biased1 >= biased2) ? (biased1 - biased2) : (biased2 - biased1);
+            }
+
+            FloatingPointUnion u_;
+        };
+
+// We cannot use std::numeric_limits<T>::max() as it clashes with the max()
+// macro defined by <windows.h>.
+        template <>
+        inline float FloatingPoint<float>::Max()
+        {
+            return FLT_MAX;
+        }
+        template <>
+        inline double FloatingPoint<double>::Max()
+        {
+            return DBL_MAX;
+        }
+
+// Typedefs the instances of the FloatingPoint template class that we
+// care to use.
+        typedef FloatingPoint<float> Float;
+        typedef FloatingPoint<double> Double;
+
+// In order to catch the mistake of putting tests that use different
+// test fixture classes in the same test case, we need to assign
+// unique IDs to fixture classes and compare them.  The TypeId type is
+// used to hold such IDs.  The user should treat TypeId as an opaque
+// type: the only operation allowed on TypeId values is to compare
+// them for equality using the == operator.
+        typedef const void *TypeId;
+
+        template <typename T>
+        class TypeIdHelper
+        {
+        public:
+            // dummy_ must not have a const type.  Otherwise an overly eager
+            // compiler (e.g. MSVC 7.1 & 8.0) may try to merge
+            // TypeIdHelper<T>::dummy_ for different Ts as an "optimization".
+            static bool dummy_;
+        };
+
+        template <typename T>
+        bool TypeIdHelper<T>::dummy_ = false;
+
+// GetTypeId<T>() returns the ID of type T.  Different values will be
+// returned for different types.  Calling the function twice with the
+// same type argument is guaranteed to return the same ID.
+        template <typename T>
+        TypeId GetTypeId()
+        {
+            // The compiler is required to allocate a different
+            // TypeIdHelper<T>::dummy_ variable for each T used to instantiate
+            // the template.  Therefore, the address of dummy_ is guaranteed to
+            // be unique.
+            return &(TypeIdHelper<T>::dummy_);
+        }
+
+// Returns the type ID of ::testing::Test.  Always call this instead
+// of GetTypeId< ::testing::Test>() to get the type ID of
+// ::testing::Test, as the latter may give the wrong result due to a
+// suspected linker bug when compiling Google Test as a Mac OS X
+// framework.
+        GTEST_API_ TypeId GetTestTypeId();
+
+// Defines the abstract factory interface that creates instances
+// of a Test object.
+        class TestFactoryBase
+        {
+        public:
+            virtual ~TestFactoryBase() {}
+
+            // Creates a test instance to run. The instance is both created and destroyed
+            // within TestInfoImpl::Run()
+            virtual Test *CreateTest() = 0;
+
+        protected:
+            TestFactoryBase() {}
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(TestFactoryBase);
+        };
+
+// This class provides implementation of TeastFactoryBase interface.
+// It is used in TEST and TEST_F macros.
+        template <class TestClass>
+        class TestFactoryImpl : public TestFactoryBase
+        {
+        public:
+            virtual Test *CreateTest()
+            {
+                return new TestClass;
+            }
+        };
+
+#if GTEST_OS_WINDOWS
+
+// Predicate-formatters for implementing the HRESULT checking macros
+// {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}
+// We pass a long instead of HRESULT to avoid causing an
+// include dependency for the HRESULT type.
+        GTEST_API_ AssertionResult IsHRESULTSuccess(const char *expr,
+                                                    long hr);  // NOLINT
+        GTEST_API_ AssertionResult IsHRESULTFailure(const char *expr,
+                                                    long hr);  // NOLINT
+
+#endif  // GTEST_OS_WINDOWS
+
+// Types of SetUpTestCase() and TearDownTestCase() functions.
+        typedef void (*SetUpTestCaseFunc)();
+        typedef void (*TearDownTestCaseFunc)();
+
+        struct CodeLocation {
+            CodeLocation(const std::string &a_file, int a_line)
+                : file(a_file), line(a_line) {}
+
+            std::string file;
+            int line;
+        };
+
+// Creates a new TestInfo object and registers it with Google Test;
+// returns the created object.
+//
+// Arguments:
+//
+//   test_case_name:   name of the test case
+//   name:             name of the test
+//   type_param        the name of the test's type parameter, or NULL if
+//                     this is not a typed or a type-parameterized test.
+//   value_param       text representation of the test's value parameter,
+//                     or NULL if this is not a type-parameterized test.
+//   code_location:    code location where the test is defined
+//   fixture_class_id: ID of the test fixture class
+//   set_up_tc:        pointer to the function that sets up the test case
+//   tear_down_tc:     pointer to the function that tears down the test case
+//   factory:          pointer to the factory that creates a test object.
+//                     The newly created TestInfo instance will assume
+//                     ownership of the factory object.
+        GTEST_API_ TestInfo *MakeAndRegisterTestInfo(
+            const char *test_case_name,
+            const char *name,
+            const char *type_param,
+            const char *value_param,
+            CodeLocation code_location,
+            TypeId fixture_class_id,
+            SetUpTestCaseFunc set_up_tc,
+            TearDownTestCaseFunc tear_down_tc,
+            TestFactoryBase *factory);
+
+// If *pstr starts with the given prefix, modifies *pstr to be right
+// past the prefix and returns true; otherwise leaves *pstr unchanged
+// and returns false.  None of pstr, *pstr, and prefix can be NULL.
+        GTEST_API_ bool SkipPrefix(const char *prefix, const char **pstr);
+
+#if GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// State of the definition of a type-parameterized test case.
+        class GTEST_API_ TypedTestCasePState
+        {
+        public:
+            TypedTestCasePState() : registered_(false) {}
+
+            // Adds the given test name to defined_test_names_ and return true
+            // if the test case hasn't been registered; otherwise aborts the
+            // program.
+            bool AddTestName(const char *file, int line, const char *case_name,
+                             const char *test_name)
+            {
+                if (registered_) {
+                    fprintf(stderr, "%s Test %s must be defined before "
+                            "REGISTER_TYPED_TEST_CASE_P(%s, ...).\n",
+                            FormatFileLocation(file, line).c_str(), test_name, case_name);
+                    fflush(stderr);
+                    posix::Abort();
+                }
+                registered_tests_.insert(
+                    ::std::make_pair(test_name, CodeLocation(file, line)));
+                return true;
+            }
+
+            bool TestExists(const std::string &test_name) const
+            {
+                return registered_tests_.count(test_name) > 0;
+            }
+
+            const CodeLocation &GetCodeLocation(const std::string &test_name) const
+            {
+                RegisteredTestsMap::const_iterator it = registered_tests_.find(test_name);
+                GTEST_CHECK_(it != registered_tests_.end());
+                return it->second;
+            }
+
+            // Verifies that registered_tests match the test names in
+            // defined_test_names_; returns registered_tests if successful, or
+            // aborts the program otherwise.
+            const char *VerifyRegisteredTestNames(
+                const char *file, int line, const char *registered_tests);
+
+        private:
+            typedef ::std::map<std::string, CodeLocation> RegisteredTestsMap;
+
+            bool registered_;
+            RegisteredTestsMap registered_tests_;
+        };
+
+// Skips to the first non-space char after the first comma in 'str';
+// returns NULL if no comma is found in 'str'.
+        inline const char *SkipComma(const char *str)
+        {
+            const char *comma = strchr(str, ',');
+            if (comma == NULL) {
+                return NULL;
+            }
+            while (IsSpace(*(++comma))) {}
+            return comma;
+        }
+
+// Returns the prefix of 'str' before the first comma in it; returns
+// the entire string if it contains no comma.
+        inline std::string GetPrefixUntilComma(const char *str)
+        {
+            const char *comma = strchr(str, ',');
+            return comma == NULL ? str : std::string(str, comma);
+        }
+
+// Splits a given string on a given delimiter, populating a given
+// vector with the fields.
+        void SplitString(const ::std::string &str, char delimiter,
+                         ::std::vector< ::std::string> *dest);
+
+// TypeParameterizedTest<Fixture, TestSel, Types>::Register()
+// registers a list of type-parameterized tests with Google Test.  The
+// return value is insignificant - we just need to return something
+// such that we can call this function in a namespace scope.
+//
+// Implementation note: The GTEST_TEMPLATE_ macro declares a template
+// template parameter.  It's defined in gtest-type-util.h.
+        template <GTEST_TEMPLATE_ Fixture, class TestSel, typename Types>
+        class TypeParameterizedTest
+        {
+        public:
+            // 'index' is the index of the test in the type list 'Types'
+            // specified in INSTANTIATE_TYPED_TEST_CASE_P(Prefix, TestCase,
+            // Types).  Valid values for 'index' are [0, N - 1] where N is the
+            // length of Types.
+            static bool Register(const char *prefix,
+                                 CodeLocation code_location,
+                                 const char *case_name, const char *test_names,
+                                 int index)
+            {
+                typedef typename Types::Head Type;
+                typedef Fixture<Type> FixtureClass;
+                typedef typename GTEST_BIND_(TestSel, Type) TestClass;
+
+                // First, registers the first type-parameterized test in the type
+                // list.
+                MakeAndRegisterTestInfo(
+                    (std::string(prefix) + (prefix[0] == '\0' ? "" : "/") + case_name + "/"
+                     + StreamableToString(index)).c_str(),
+                    StripTrailingSpaces(GetPrefixUntilComma(test_names)).c_str(),
+                    GetTypeName<Type>().c_str(),
+                    NULL,  // No value parameter.
+                    code_location,
+                    GetTypeId<FixtureClass>(),
+                    TestClass::SetUpTestCase,
+                    TestClass::TearDownTestCase,
+                    new TestFactoryImpl<TestClass>);
+
+                // Next, recurses (at compile time) with the tail of the type list.
+                return TypeParameterizedTest<Fixture, TestSel, typename Types::Tail>
+                       ::Register(prefix, code_location, case_name, test_names, index + 1);
+            }
+        };
+
+// The base case for the compile time recursion.
+        template <GTEST_TEMPLATE_ Fixture, class TestSel>
+        class TypeParameterizedTest<Fixture, TestSel, Types0>
+        {
+        public:
+            static bool Register(const char * /*prefix*/, CodeLocation,
+                                 const char * /*case_name*/, const char * /*test_names*/,
+                                 int /*index*/)
+            {
+                return true;
+            }
+        };
+
+// TypeParameterizedTestCase<Fixture, Tests, Types>::Register()
+// registers *all combinations* of 'Tests' and 'Types' with Google
+// Test.  The return value is insignificant - we just need to return
+// something such that we can call this function in a namespace scope.
+        template <GTEST_TEMPLATE_ Fixture, typename Tests, typename Types>
+        class TypeParameterizedTestCase
+        {
+        public:
+            static bool Register(const char *prefix, CodeLocation code_location,
+                                 const TypedTestCasePState *state,
+                                 const char *case_name, const char *test_names)
+            {
+                std::string test_name = StripTrailingSpaces(
+                                            GetPrefixUntilComma(test_names));
+                if (!state->TestExists(test_name)) {
+                    fprintf(stderr, "Failed to get code location for test %s.%s at %s.",
+                            case_name, test_name.c_str(),
+                            FormatFileLocation(code_location.file.c_str(),
+                                               code_location.line).c_str());
+                    fflush(stderr);
+                    posix::Abort();
+                }
+                const CodeLocation &test_location = state->GetCodeLocation(test_name);
+
+                typedef typename Tests::Head Head;
+
+                // First, register the first test in 'Test' for each type in 'Types'.
+                TypeParameterizedTest<Fixture, Head, Types>::Register(
+                    prefix, test_location, case_name, test_names, 0);
+
+                // Next, recurses (at compile time) with the tail of the test list.
+                return TypeParameterizedTestCase<Fixture, typename Tests::Tail, Types>
+                       ::Register(prefix, code_location, state,
+                                  case_name, SkipComma(test_names));
+            }
+        };
+
+// The base case for the compile time recursion.
+        template <GTEST_TEMPLATE_ Fixture, typename Types>
+        class TypeParameterizedTestCase<Fixture, Templates0, Types>
+        {
+        public:
+            static bool Register(const char * /*prefix*/, CodeLocation,
+                                 const TypedTestCasePState * /*state*/,
+                                 const char * /*case_name*/, const char * /*test_names*/)
+            {
+                return true;
+            }
+        };
+
+#endif  // GTEST_HAS_TYPED_TEST || GTEST_HAS_TYPED_TEST_P
+
+// Returns the current OS stack trace as an std::string.
+//
+// The maximum number of stack frames to be included is specified by
+// the gtest_stack_trace_depth flag.  The skip_count parameter
+// specifies the number of top frames to be skipped, which doesn't
+// count against the number of frames to be included.
+//
+// For example, if Foo() calls Bar(), which in turn calls
+// GetCurrentOsStackTraceExceptTop(..., 1), Foo() will be included in
+// the trace but Bar() and GetCurrentOsStackTraceExceptTop() won't.
+        GTEST_API_ std::string GetCurrentOsStackTraceExceptTop(
+            UnitTest *unit_test, int skip_count);
+
+// Helpers for suppressing warnings on unreachable code or constant
+// condition.
+
+// Always returns true.
+        GTEST_API_ bool AlwaysTrue();
+
+// Always returns false.
+        inline bool AlwaysFalse()
+        {
+            return !AlwaysTrue();
+        }
+
+// Helper for suppressing false warning from Clang on a const char*
+// variable declared in a conditional expression always being NULL in
+// the else branch.
+        struct GTEST_API_ ConstCharPtr {
+            ConstCharPtr(const char *str) : value(str) {}
+            operator bool() const
+            {
+                return true;
+            }
+            const char *value;
+        };
+
+// A simple Linear Congruential Generator for generating random
+// numbers with a uniform distribution.  Unlike rand() and srand(), it
+// doesn't use global state (and therefore can't interfere with user
+// code).  Unlike rand_r(), it's portable.  An LCG isn't very random,
+// but it's good enough for our purposes.
+        class GTEST_API_ Random
+        {
+        public:
+            static const UInt32 kMaxRange = 1u << 31;
+
+            explicit Random(UInt32 seed) : state_(seed) {}
+
+            void Reseed(UInt32 seed)
+            {
+                state_ = seed;
+            }
+
+            // Generates a random number from [0, range).  Crashes if 'range' is
+            // 0 or greater than kMaxRange.
+            UInt32 Generate(UInt32 range);
+
+        private:
+            UInt32 state_;
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(Random);
+        };
+
+// Defining a variable of type CompileAssertTypesEqual<T1, T2> will cause a
+// compiler error iff T1 and T2 are different types.
+        template <typename T1, typename T2>
+        struct CompileAssertTypesEqual;
+
+        template <typename T>
+        struct CompileAssertTypesEqual<T, T> {
+        };
+
+// Removes the reference from a type if it is a reference type,
+// otherwise leaves it unchanged.  This is the same as
+// tr1::remove_reference, which is not widely available yet.
+        template <typename T>
+        struct RemoveReference {
+            typedef T type;
+        };  // NOLINT
+        template <typename T>
+        struct RemoveReference<T &> {
+            typedef T type;
+        };  // NOLINT
+
+// A handy wrapper around RemoveReference that works when the argument
+// T depends on template parameters.
+#define GTEST_REMOVE_REFERENCE_(T) \
+    typename ::testing::internal::RemoveReference<T>::type
+
+// Removes const from a type if it is a const type, otherwise leaves
+// it unchanged.  This is the same as tr1::remove_const, which is not
+// widely available yet.
+        template <typename T>
+        struct RemoveConst {
+            typedef T type;
+        };  // NOLINT
+        template <typename T>
+        struct RemoveConst<const T> {
+            typedef T type;
+        };  // NOLINT
+
+// MSVC 8.0, Sun C++, and IBM XL C++ have a bug which causes the above
+// definition to fail to remove the const in 'const int[3]' and 'const
+// char[3][4]'.  The following specialization works around the bug.
+        template <typename T, size_t N>
+        struct RemoveConst<const T[N]> {
+            typedef typename RemoveConst<T>::type type[N];
+        };
+
+#if defined(_MSC_VER) && _MSC_VER < 1400
+// This is the only specialization that allows VC++ 7.1 to remove const in
+// 'const int[3] and 'const int[3][4]'.  However, it causes trouble with GCC
+// and thus needs to be conditionally compiled.
+        template <typename T, size_t N>
+        struct RemoveConst<T[N]> {
+            typedef typename RemoveConst<T>::type type[N];
+        };
+#endif
+
+// A handy wrapper around RemoveConst that works when the argument
+// T depends on template parameters.
+#define GTEST_REMOVE_CONST_(T) \
+    typename ::testing::internal::RemoveConst<T>::type
+
+// Turns const U&, U&, const U, and U all into U.
+#define GTEST_REMOVE_REFERENCE_AND_CONST_(T) \
+    GTEST_REMOVE_CONST_(GTEST_REMOVE_REFERENCE_(T))
+
+// Adds reference to a type if it is not a reference type,
+// otherwise leaves it unchanged.  This is the same as
+// tr1::add_reference, which is not widely available yet.
+        template <typename T>
+        struct AddReference {
+            typedef T &type;
+        };  // NOLINT
+        template <typename T>
+        struct AddReference<T &> {
+            typedef T &type;
+        };  // NOLINT
+
+// A handy wrapper around AddReference that works when the argument T
+// depends on template parameters.
+#define GTEST_ADD_REFERENCE_(T) \
+    typename ::testing::internal::AddReference<T>::type
+
+// Adds a reference to const on top of T as necessary.  For example,
+// it transforms
+//
+//   char         ==> const char&
+//   const char   ==> const char&
+//   char&        ==> const char&
+//   const char&  ==> const char&
+//
+// The argument T must depend on some template parameters.
+#define GTEST_REFERENCE_TO_CONST_(T) \
+    GTEST_ADD_REFERENCE_(const GTEST_REMOVE_REFERENCE_(T))
+
+// ImplicitlyConvertible<From, To>::value is a compile-time bool
+// constant that's true iff type From can be implicitly converted to
+// type To.
+        template <typename From, typename To>
+        class ImplicitlyConvertible
+        {
+        private:
+            // We need the following helper functions only for their types.
+            // They have no implementations.
+
+            // MakeFrom() is an expression whose type is From.  We cannot simply
+            // use From(), as the type From may not have a public default
+            // constructor.
+            static typename AddReference<From>::type MakeFrom();
+
+            // These two functions are overloaded.  Given an expression
+            // Helper(x), the compiler will pick the first version if x can be
+            // implicitly converted to type To; otherwise it will pick the
+            // second version.
+            //
+            // The first version returns a value of size 1, and the second
+            // version returns a value of size 2.  Therefore, by checking the
+            // size of Helper(x), which can be done at compile time, we can tell
+            // which version of Helper() is used, and hence whether x can be
+            // implicitly converted to type To.
+            static char Helper(To);
+            static char (&Helper(...))[2];  // NOLINT
+
+            // We have to put the 'public' section after the 'private' section,
+            // or MSVC refuses to compile the code.
+        public:
+#if defined(__BORLANDC__)
+            // C++Builder cannot use member overload resolution during template
+            // instantiation.  The simplest workaround is to use its C++0x type traits
+            // functions (C++Builder 2009 and above only).
+            static const bool value = __is_convertible(From, To);
+#else
+            // MSVC warns about implicitly converting from double to int for
+            // possible loss of data, so we need to temporarily disable the
+            // warning.
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4244)
+            static const bool value =
+                sizeof(Helper(ImplicitlyConvertible::MakeFrom())) == 1;
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+#endif  // __BORLANDC__
+        };
+        template <typename From, typename To>
+        const bool ImplicitlyConvertible<From, To>::value;
+
+// IsAProtocolMessage<T>::value is a compile-time bool constant that's
+// true iff T is type ProtocolMessage, proto2::Message, or a subclass
+// of those.
+        template <typename T>
+        struct IsAProtocolMessage
+            : public bool_constant<
+              ImplicitlyConvertible<const T *, const ::ProtocolMessage *>::value ||
+              ImplicitlyConvertible<const T *, const ::proto2::Message *>::value> {
+        };
+
+// When the compiler sees expression IsContainerTest<C>(0), if C is an
+// STL-style container class, the first overload of IsContainerTest
+// will be viable (since both C::iterator* and C::const_iterator* are
+// valid types and NULL can be implicitly converted to them).  It will
+// be picked over the second overload as 'int' is a perfect match for
+// the type of argument 0.  If C::iterator or C::const_iterator is not
+// a valid type, the first overload is not viable, and the second
+// overload will be picked.  Therefore, we can determine whether C is
+// a container class by checking the type of IsContainerTest<C>(0).
+// The value of the expression is insignificant.
+//
+// Note that we look for both C::iterator and C::const_iterator.  The
+// reason is that C++ injects the name of a class as a member of the
+// class itself (e.g. you can refer to class iterator as either
+// 'iterator' or 'iterator::iterator').  If we look for C::iterator
+// only, for example, we would mistakenly think that a class named
+// iterator is an STL container.
+//
+// Also note that the simpler approach of overloading
+// IsContainerTest(typename C::const_iterator*) and
+// IsContainerTest(...) doesn't work with Visual Age C++ and Sun C++.
+        typedef int IsContainer;
+        template <class C>
+        IsContainer IsContainerTest(int /* dummy */,
+                                    typename C::iterator * /* it */ = NULL,
+                                    typename C::const_iterator * /* const_it */ = NULL)
+        {
+            return 0;
+        }
+
+        typedef char IsNotContainer;
+        template <class C>
+        IsNotContainer IsContainerTest(long /* dummy */)
+        {
+            return '\0';
+        }
+
+// EnableIf<condition>::type is void when 'Cond' is true, and
+// undefined when 'Cond' is false.  To use SFINAE to make a function
+// overload only apply when a particular expression is true, add
+// "typename EnableIf<expression>::type* = 0" as the last parameter.
+        template<bool> struct EnableIf;
+        template<> struct EnableIf<true> {
+            typedef void type;
+        };  // NOLINT
+
+// Utilities for native arrays.
+
+// ArrayEq() compares two k-dimensional native arrays using the
+// elements' operator==, where k can be any integer >= 0.  When k is
+// 0, ArrayEq() degenerates into comparing a single pair of values.
+
+        template <typename T, typename U>
+        bool ArrayEq(const T *lhs, size_t size, const U *rhs);
+
+// This generic version is used when k is 0.
+        template <typename T, typename U>
+        inline bool ArrayEq(const T &lhs, const U &rhs)
+        {
+            return lhs == rhs;
+        }
+
+// This overload is used when k >= 1.
+        template <typename T, typename U, size_t N>
+        inline bool ArrayEq(const T(&lhs)[N], const U(&rhs)[N])
+        {
+            return internal::ArrayEq(lhs, N, rhs);
+        }
+
+// This helper reduces code bloat.  If we instead put its logic inside
+// the previous ArrayEq() function, arrays with different sizes would
+// lead to different copies of the template code.
+        template <typename T, typename U>
+        bool ArrayEq(const T *lhs, size_t size, const U *rhs)
+        {
+            for (size_t i = 0; i != size; i++) {
+                if (!internal::ArrayEq(lhs[i], rhs[i]))
+                    return false;
+            }
+            return true;
+        }
+
+// Finds the first element in the iterator range [begin, end) that
+// equals elem.  Element may be a native array type itself.
+        template <typename Iter, typename Element>
+        Iter ArrayAwareFind(Iter begin, Iter end, const Element &elem)
+        {
+            for (Iter it = begin; it != end; ++it) {
+                if (internal::ArrayEq(*it, elem))
+                    return it;
+            }
+            return end;
+        }
+
+// CopyArray() copies a k-dimensional native array using the elements'
+// operator=, where k can be any integer >= 0.  When k is 0,
+// CopyArray() degenerates into copying a single value.
+
+        template <typename T, typename U>
+        void CopyArray(const T *from, size_t size, U *to);
+
+// This generic version is used when k is 0.
+        template <typename T, typename U>
+        inline void CopyArray(const T &from, U *to)
+        {
+            *to = from;
+        }
+
+// This overload is used when k >= 1.
+        template <typename T, typename U, size_t N>
+        inline void CopyArray(const T(&from)[N], U(*to)[N])
+        {
+            internal::CopyArray(from, N, *to);
+        }
+
+// This helper reduces code bloat.  If we instead put its logic inside
+// the previous CopyArray() function, arrays with different sizes
+// would lead to different copies of the template code.
+        template <typename T, typename U>
+        void CopyArray(const T *from, size_t size, U *to)
+        {
+            for (size_t i = 0; i != size; i++) {
+                internal::CopyArray(from[i], to + i);
+            }
+        }
+
+// The relation between an NativeArray object (see below) and the
+// native array it represents.
+// We use 2 different structs to allow non-copyable types to be used, as long
+// as RelationToSourceReference() is passed.
+        struct RelationToSourceReference {};
+        struct RelationToSourceCopy {};
+
+// Adapts a native array to a read-only STL-style container.  Instead
+// of the complete STL container concept, this adaptor only implements
+// members useful for Google Mock's container matchers.  New members
+// should be added as needed.  To simplify the implementation, we only
+// support Element being a raw type (i.e. having no top-level const or
+// reference modifier).  It's the client's responsibility to satisfy
+// this requirement.  Element can be an array type itself (hence
+// multi-dimensional arrays are supported).
+        template <typename Element>
+        class NativeArray
+        {
+        public:
+            // STL-style container typedefs.
+            typedef Element value_type;
+            typedef Element *iterator;
+            typedef const Element *const_iterator;
+
+            // Constructs from a native array. References the source.
+            NativeArray(const Element *array, size_t count, RelationToSourceReference)
+            {
+                InitRef(array, count);
+            }
+
+            // Constructs from a native array. Copies the source.
+            NativeArray(const Element *array, size_t count, RelationToSourceCopy)
+            {
+                InitCopy(array, count);
+            }
+
+            // Copy constructor.
+            NativeArray(const NativeArray &rhs)
+            {
+                (this->*rhs.clone_)(rhs.array_, rhs.size_);
+            }
+
+            ~NativeArray()
+            {
+                if (clone_ != &NativeArray::InitRef)
+                    delete[] array_;
+            }
+
+            // STL-style container methods.
+            size_t size() const
+            {
+                return size_;
+            }
+            const_iterator begin() const
+            {
+                return array_;
+            }
+            const_iterator end() const
+            {
+                return array_ + size_;
+            }
+            bool operator==(const NativeArray &rhs) const
+            {
+                return size() == rhs.size() &&
+                       ArrayEq(begin(), size(), rhs.begin());
+            }
+
+        private:
+            enum {
+                kCheckTypeIsNotConstOrAReference = StaticAssertTypeEqHelper<
+                                                   Element, GTEST_REMOVE_REFERENCE_AND_CONST_(Element)>::value,
+            };
+
+            // Initializes this object with a copy of the input.
+            void InitCopy(const Element *array, size_t a_size)
+            {
+                Element *const copy = new Element[a_size];
+                CopyArray(array, a_size, copy);
+                array_ = copy;
+                size_ = a_size;
+                clone_ = &NativeArray::InitCopy;
+            }
+
+            // Initializes this object with a reference of the input.
+            void InitRef(const Element *array, size_t a_size)
+            {
+                array_ = array;
+                size_ = a_size;
+                clone_ = &NativeArray::InitRef;
+            }
+
+            const Element *array_;
+            size_t size_;
+            void (NativeArray::*clone_)(const Element *, size_t);
+
+            GTEST_DISALLOW_ASSIGN_(NativeArray);
+        };
+
+    }  // namespace internal
+}  // namespace testing
+
+#define GTEST_MESSAGE_AT_(file, line, message, result_type) \
+  ::testing::internal::AssertHelper(result_type, file, line, message) \
+    = ::testing::Message()
+
+#define GTEST_MESSAGE_(message, result_type) \
+  GTEST_MESSAGE_AT_(__FILE__, __LINE__, message, result_type)
+
+#define GTEST_FATAL_FAILURE_(message) \
+  return GTEST_MESSAGE_(message, ::testing::TestPartResult::kFatalFailure)
+
+#define GTEST_NONFATAL_FAILURE_(message) \
+  GTEST_MESSAGE_(message, ::testing::TestPartResult::kNonFatalFailure)
+
+#define GTEST_SUCCESS_(message) \
+  GTEST_MESSAGE_(message, ::testing::TestPartResult::kSuccess)
+
+#define GTEST_SKIP_(message) \
+  return GTEST_MESSAGE_(message, ::testing::TestPartResult::kSkip)
+
+// Suppresses MSVC warnings 4072 (unreachable code) for the code following
+// statement if it returns or throws (or doesn't return or throw in some
+// situations).
+#define GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement) \
+  if (::testing::internal::AlwaysTrue()) { statement; }
+
+#define GTEST_TEST_THROW_(statement, expected_exception, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::ConstCharPtr gtest_msg = "") { \
+    bool gtest_caught_expected = false; \
+    try { \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    } \
+    catch (expected_exception const&) { \
+      gtest_caught_expected = true; \
+    } \
+    catch (...) { \
+      gtest_msg.value = \
+          "Expected: " #statement " throws an exception of type " \
+          #expected_exception ".\n  Actual: it throws a different type."; \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \
+    } \
+    if (!gtest_caught_expected) { \
+      gtest_msg.value = \
+          "Expected: " #statement " throws an exception of type " \
+          #expected_exception ".\n  Actual: it throws nothing."; \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testthrow_, __LINE__): \
+      fail(gtest_msg.value)
+
+#define GTEST_TEST_NO_THROW_(statement, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    try { \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    } \
+    catch (...) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testnothrow_, __LINE__): \
+      fail("Expected: " #statement " doesn't throw an exception.\n" \
+           "  Actual: it throws.")
+
+#define GTEST_TEST_ANY_THROW_(statement, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    bool gtest_caught_any = false; \
+    try { \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    } \
+    catch (...) { \
+      gtest_caught_any = true; \
+    } \
+    if (!gtest_caught_any) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testanythrow_, __LINE__): \
+      fail("Expected: " #statement " throws an exception.\n" \
+           "  Actual: it doesn't.")
+
+
+// Implements Boolean test assertions such as EXPECT_TRUE. expression can be
+// either a boolean expression or an AssertionResult. text is a textual
+// representation of expression as it was passed into the EXPECT_TRUE.
+#define GTEST_TEST_BOOLEAN_(expression, text, actual, expected, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (const ::testing::AssertionResult gtest_ar_ = \
+      ::testing::AssertionResult(expression)) \
+    ; \
+  else \
+    fail(::testing::internal::GetBoolAssertionFailureMessage(\
+        gtest_ar_, text, #actual, #expected).c_str())
+
+#define GTEST_TEST_NO_FATAL_FAILURE_(statement, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    ::testing::internal::HasNewFatalFailureHelper gtest_fatal_failure_checker; \
+    GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+    if (gtest_fatal_failure_checker.has_new_fatal_failure()) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__); \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_testnofatal_, __LINE__): \
+      fail("Expected: " #statement " doesn't generate new fatal " \
+           "failures in the current thread.\n" \
+           "  Actual: it does.")
+
+// Expands to the name of the class that implements the given test.
+#define GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \
+  test_case_name##_##test_name##_Test
+
+// Helper macro for defining tests.
+#define GTEST_TEST_(test_case_name, test_name, parent_class, parent_id)\
+class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) : public parent_class {\
+ public:\
+  GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {}\
+ private:\
+  virtual void TestBody();\
+  static ::testing::TestInfo* const test_info_ GTEST_ATTRIBUTE_UNUSED_;\
+  GTEST_DISALLOW_COPY_AND_ASSIGN_(\
+      GTEST_TEST_CLASS_NAME_(test_case_name, test_name));\
+};\
+\
+::testing::TestInfo* const GTEST_TEST_CLASS_NAME_(test_case_name, test_name)\
+  ::test_info_ =\
+    ::testing::internal::MakeAndRegisterTestInfo(\
+        #test_case_name, #test_name, NULL, NULL, \
+        ::testing::internal::CodeLocation(__FILE__, __LINE__), \
+        (parent_id), \
+        parent_class::SetUpTestCase, \
+        parent_class::TearDownTestCase, \
+        new ::testing::internal::TestFactoryImpl<\
+            GTEST_TEST_CLASS_NAME_(test_case_name, test_name)>);\
+void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody()
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_INTERNAL_H_
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines the public API for death tests.  It is
+// #included by gtest.h so a user doesn't need to include this
+// directly.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+
+// Copyright 2005, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: wan@google.com (Zhanyong Wan), eefacm@gmail.com (Sean Mcafee)
+//
+// The Google C++ Testing Framework (Google Test)
+//
+// This header file defines internal utilities needed for implementing
+// death tests.  They are subject to change without notice.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+
+
+#include <stdio.h>
+
+namespace testing
+{
+    namespace internal
+    {
+
+        GTEST_DECLARE_string_(internal_run_death_test);
+
+// Names of the flags (needed for parsing Google Test flags).
+        const char kDeathTestStyleFlag[] = "death_test_style";
+        const char kDeathTestUseFork[] = "death_test_use_fork";
+        const char kInternalRunDeathTestFlag[] = "internal_run_death_test";
+
+#if GTEST_HAS_DEATH_TEST
+
+// DeathTest is a class that hides much of the complexity of the
+// GTEST_DEATH_TEST_ macro.  It is abstract; its static Create method
+// returns a concrete class that depends on the prevailing death test
+// style, as defined by the --gtest_death_test_style and/or
+// --gtest_internal_run_death_test flags.
+
+// In describing the results of death tests, these terms are used with
+// the corresponding definitions:
+//
+// exit status:  The integer exit information in the format specified
+//               by wait(2)
+// exit code:    The integer code passed to exit(3), _exit(2), or
+//               returned from main()
+        class GTEST_API_ DeathTest
+        {
+        public:
+            // Create returns false if there was an error determining the
+            // appropriate action to take for the current death test; for example,
+            // if the gtest_death_test_style flag is set to an invalid value.
+            // The LastMessage method will return a more detailed message in that
+            // case.  Otherwise, the DeathTest pointer pointed to by the "test"
+            // argument is set.  If the death test should be skipped, the pointer
+            // is set to NULL; otherwise, it is set to the address of a new concrete
+            // DeathTest object that controls the execution of the current test.
+            static bool Create(const char *statement, const RE *regex,
+                               const char *file, int line, DeathTest **test);
+            DeathTest();
+            virtual ~DeathTest() { }
+
+            // A helper class that aborts a death test when it's deleted.
+            class ReturnSentinel
+            {
+            public:
+                explicit ReturnSentinel(DeathTest *test) : test_(test) { }
+                ~ReturnSentinel()
+                {
+                    test_->Abort(TEST_ENCOUNTERED_RETURN_STATEMENT);
+                }
+            private:
+                DeathTest *const test_;
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(ReturnSentinel);
+            } GTEST_ATTRIBUTE_UNUSED_;
+
+            // An enumeration of possible roles that may be taken when a death
+            // test is encountered.  EXECUTE means that the death test logic should
+            // be executed immediately.  OVERSEE means that the program should prepare
+            // the appropriate environment for a child process to execute the death
+            // test, then wait for it to complete.
+            enum TestRole { OVERSEE_TEST, EXECUTE_TEST };
+
+            // An enumeration of the three reasons that a test might be aborted.
+            enum AbortReason {
+                TEST_ENCOUNTERED_RETURN_STATEMENT,
+                TEST_THREW_EXCEPTION,
+                TEST_DID_NOT_DIE
+            };
+
+            // Assumes one of the above roles.
+            virtual TestRole AssumeRole() = 0;
+
+            // Waits for the death test to finish and returns its status.
+            virtual int Wait() = 0;
+
+            // Returns true if the death test passed; that is, the test process
+            // exited during the test, its exit status matches a user-supplied
+            // predicate, and its stderr output matches a user-supplied regular
+            // expression.
+            // The user-supplied predicate may be a macro expression rather
+            // than a function pointer or functor, or else Wait and Passed could
+            // be combined.
+            virtual bool Passed(bool exit_status_ok) = 0;
+
+            // Signals that the death test did not die as expected.
+            virtual void Abort(AbortReason reason) = 0;
+
+            // Returns a human-readable outcome message regarding the outcome of
+            // the last death test.
+            static const char *LastMessage();
+
+            static void set_last_death_test_message(const std::string &message);
+
+        private:
+            // A string containing a description of the outcome of the last death test.
+            static std::string last_death_test_message_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(DeathTest);
+        };
+
+// Factory interface for death tests.  May be mocked out for testing.
+        class DeathTestFactory
+        {
+        public:
+            virtual ~DeathTestFactory() { }
+            virtual bool Create(const char *statement, const RE *regex,
+                                const char *file, int line, DeathTest **test) = 0;
+        };
+
+// A concrete DeathTestFactory implementation for normal use.
+        class DefaultDeathTestFactory : public DeathTestFactory
+        {
+        public:
+            virtual bool Create(const char *statement, const RE *regex,
+                                const char *file, int line, DeathTest **test);
+        };
+
+// Returns true if exit_status describes a process that was terminated
+// by a signal, or exited normally with a nonzero exit code.
+        GTEST_API_ bool ExitedUnsuccessfully(int exit_status);
+
+// Traps C++ exceptions escaping statement and reports them as test
+// failures. Note that trapping SEH exceptions is not implemented here.
+# if GTEST_HAS_EXCEPTIONS
+#  define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \
+  try { \
+    GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+  } catch (const ::std::exception& gtest_exception) { \
+    fprintf(\
+        stderr, \
+        "\n%s: Caught std::exception-derived exception escaping the " \
+        "death test statement. Exception message: %s\n", \
+        ::testing::internal::FormatFileLocation(__FILE__, __LINE__).c_str(), \
+        gtest_exception.what()); \
+    fflush(stderr); \
+    death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \
+  } catch (...) { \
+    death_test->Abort(::testing::internal::DeathTest::TEST_THREW_EXCEPTION); \
+  }
+
+# else
+#  define GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, death_test) \
+  GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement)
+
+# endif
+
+// This macro is for implementing ASSERT_DEATH*, EXPECT_DEATH*,
+// ASSERT_EXIT*, and EXPECT_EXIT*.
+# define GTEST_DEATH_TEST_(statement, predicate, regex, fail) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+    const ::testing::internal::RE& gtest_regex = (regex); \
+    ::testing::internal::DeathTest* gtest_dt; \
+    if (!::testing::internal::DeathTest::Create(#statement, &gtest_regex, \
+        __FILE__, __LINE__, &gtest_dt)) { \
+      goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \
+    } \
+    if (gtest_dt != NULL) { \
+      ::testing::internal::scoped_ptr< ::testing::internal::DeathTest> \
+          gtest_dt_ptr(gtest_dt); \
+      switch (gtest_dt->AssumeRole()) { \
+        case ::testing::internal::DeathTest::OVERSEE_TEST: \
+          if (!gtest_dt->Passed(predicate(gtest_dt->Wait()))) { \
+            goto GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__); \
+          } \
+          break; \
+        case ::testing::internal::DeathTest::EXECUTE_TEST: { \
+          ::testing::internal::DeathTest::ReturnSentinel \
+              gtest_sentinel(gtest_dt); \
+          GTEST_EXECUTE_DEATH_TEST_STATEMENT_(statement, gtest_dt); \
+          gtest_dt->Abort(::testing::internal::DeathTest::TEST_DID_NOT_DIE); \
+          break; \
+        } \
+        default: \
+          break; \
+      } \
+    } \
+  } else \
+    GTEST_CONCAT_TOKEN_(gtest_label_, __LINE__): \
+      fail(::testing::internal::DeathTest::LastMessage())
+// The symbol "fail" here expands to something into which a message
+// can be streamed.
+
+// This macro is for implementing ASSERT/EXPECT_DEBUG_DEATH when compiled in
+// NDEBUG mode. In this case we need the statements to be executed, the regex is
+// ignored, and the macro must accept a streamed message even though the message
+// is never printed.
+# define GTEST_EXECUTE_STATEMENT_(statement, regex) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (::testing::internal::AlwaysTrue()) { \
+     GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+  } else \
+    ::testing::Message()
+
+// A class representing the parsed contents of the
+// --gtest_internal_run_death_test flag, as it existed when
+// RUN_ALL_TESTS was called.
+        class InternalRunDeathTestFlag
+        {
+        public:
+            InternalRunDeathTestFlag(const std::string &a_file,
+                                     int a_line,
+                                     int an_index,
+                                     int a_write_fd)
+                : file_(a_file), line_(a_line), index_(an_index),
+                  write_fd_(a_write_fd) {}
+
+            ~InternalRunDeathTestFlag()
+            {
+                if (write_fd_ >= 0)
+                    posix::Close(write_fd_);
+            }
+
+            const std::string &file() const
+            {
+                return file_;
+            }
+            int line() const
+            {
+                return line_;
+            }
+            int index() const
+            {
+                return index_;
+            }
+            int write_fd() const
+            {
+                return write_fd_;
+            }
+
+        private:
+            std::string file_;
+            int line_;
+            int index_;
+            int write_fd_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(InternalRunDeathTestFlag);
+        };
+
+// Returns a newly created InternalRunDeathTestFlag object with fields
+// initialized from the GTEST_FLAG(internal_run_death_test) flag if
+// the flag is specified; otherwise returns NULL.
+        InternalRunDeathTestFlag *ParseInternalRunDeathTestFlag();
+
+#else  // GTEST_HAS_DEATH_TEST
+
+// This macro is used for implementing macros such as
+// EXPECT_DEATH_IF_SUPPORTED and ASSERT_DEATH_IF_SUPPORTED on systems where
+// death tests are not supported. Those macros must compile on such systems
+// iff EXPECT_DEATH and ASSERT_DEATH compile with the same parameters on
+// systems that support death tests. This allows one to write such a macro
+// on a system that does not support death tests and be sure that it will
+// compile on a death-test supporting system.
+//
+// Parameters:
+//   statement -  A statement that a macro such as EXPECT_DEATH would test
+//                for program termination. This macro has to make sure this
+//                statement is compiled but not executed, to ensure that
+//                EXPECT_DEATH_IF_SUPPORTED compiles with a certain
+//                parameter iff EXPECT_DEATH compiles with it.
+//   regex     -  A regex that a macro such as EXPECT_DEATH would use to test
+//                the output of statement.  This parameter has to be
+//                compiled but not evaluated by this macro, to ensure that
+//                this macro only accepts expressions that a macro such as
+//                EXPECT_DEATH would accept.
+//   terminator - Must be an empty statement for EXPECT_DEATH_IF_SUPPORTED
+//                and a return statement for ASSERT_DEATH_IF_SUPPORTED.
+//                This ensures that ASSERT_DEATH_IF_SUPPORTED will not
+//                compile inside functions where ASSERT_DEATH doesn't
+//                compile.
+//
+//  The branch that has an always false condition is used to ensure that
+//  statement and regex are compiled (and thus syntactically correct) but
+//  never executed. The unreachable code macro protects the terminator
+//  statement from generating an 'unreachable code' warning in case
+//  statement unconditionally returns or throws. The Message constructor at
+//  the end allows the syntax of streaming additional messages into the
+//  macro, for compilational compatibility with EXPECT_DEATH/ASSERT_DEATH.
+# define GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, terminator) \
+    GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+    if (::testing::internal::AlwaysTrue()) { \
+      GTEST_LOG_(WARNING) \
+          << "Death tests are not supported on this platform.\n" \
+          << "Statement '" #statement "' cannot be verified."; \
+    } else if (::testing::internal::AlwaysFalse()) { \
+      ::testing::internal::RE::PartialMatch(".*", (regex)); \
+      GTEST_SUPPRESS_UNREACHABLE_CODE_WARNING_BELOW_(statement); \
+      terminator; \
+    } else \
+      ::testing::Message()
+
+#endif  // GTEST_HAS_DEATH_TEST
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_DEATH_TEST_INTERNAL_H_
+
+namespace testing
+{
+
+// This flag controls the style of death tests.  Valid values are "threadsafe",
+// meaning that the death test child process will re-execute the test binary
+// from the start, running only a single death test, or "fast",
+// meaning that the child process will execute the test logic immediately
+// after forking.
+    GTEST_DECLARE_string_(death_test_style);
+
+#if GTEST_HAS_DEATH_TEST
+
+    namespace internal
+    {
+
+// Returns a Boolean value indicating whether the caller is currently
+// executing in the context of the death test child process.  Tools such as
+// Valgrind heap checkers may need this to modify their behavior in death
+// tests.  IMPORTANT: This is an internal utility.  Using it may break the
+// implementation of death tests.  User code MUST NOT use it.
+        GTEST_API_ bool InDeathTestChild();
+
+    }  // namespace internal
+
+// The following macros are useful for writing death tests.
+
+// Here's what happens when an ASSERT_DEATH* or EXPECT_DEATH* is
+// executed:
+//
+//   1. It generates a warning if there is more than one active
+//   thread.  This is because it's safe to fork() or clone() only
+//   when there is a single thread.
+//
+//   2. The parent process clone()s a sub-process and runs the death
+//   test in it; the sub-process exits with code 0 at the end of the
+//   death test, if it hasn't exited already.
+//
+//   3. The parent process waits for the sub-process to terminate.
+//
+//   4. The parent process checks the exit code and error message of
+//   the sub-process.
+//
+// Examples:
+//
+//   ASSERT_DEATH(server.SendMessage(56, "Hello"), "Invalid port number");
+//   for (int i = 0; i < 5; i++) {
+//     EXPECT_DEATH(server.ProcessRequest(i),
+//                  "Invalid request .* in ProcessRequest()")
+//                  << "Failed to die on request " << i;
+//   }
+//
+//   ASSERT_EXIT(server.ExitNow(), ::testing::ExitedWithCode(0), "Exiting");
+//
+//   bool KilledBySIGHUP(int exit_code) {
+//     return WIFSIGNALED(exit_code) && WTERMSIG(exit_code) == SIGHUP;
+//   }
+//
+//   ASSERT_EXIT(client.HangUpServer(), KilledBySIGHUP, "Hanging up!");
+//
+// On the regular expressions used in death tests:
+//
+//   On POSIX-compliant systems (*nix), we use the <regex.h> library,
+//   which uses the POSIX extended regex syntax.
+//
+//   On other platforms (e.g. Windows), we only support a simple regex
+//   syntax implemented as part of Google Test.  This limited
+//   implementation should be enough most of the time when writing
+//   death tests; though it lacks many features you can find in PCRE
+//   or POSIX extended regex syntax.  For example, we don't support
+//   union ("x|y"), grouping ("(xy)"), brackets ("[xy]"), and
+//   repetition count ("x{5,7}"), among others.
+//
+//   Below is the syntax that we do support.  We chose it to be a
+//   subset of both PCRE and POSIX extended regex, so it's easy to
+//   learn wherever you come from.  In the following: 'A' denotes a
+//   literal character, period (.), or a single \\ escape sequence;
+//   'x' and 'y' denote regular expressions; 'm' and 'n' are for
+//   natural numbers.
+//
+//     c     matches any literal character c
+//     \\d   matches any decimal digit
+//     \\D   matches any character that's not a decimal digit
+//     \\f   matches \f
+//     \\n   matches \n
+//     \\r   matches \r
+//     \\s   matches any ASCII whitespace, including \n
+//     \\S   matches any character that's not a whitespace
+//     \\t   matches \t
+//     \\v   matches \v
+//     \\w   matches any letter, _, or decimal digit
+//     \\W   matches any character that \\w doesn't match
+//     \\c   matches any literal character c, which must be a punctuation
+//     .     matches any single character except \n
+//     A?    matches 0 or 1 occurrences of A
+//     A*    matches 0 or many occurrences of A
+//     A+    matches 1 or many occurrences of A
+//     ^     matches the beginning of a string (not that of each line)
+//     $     matches the end of a string (not that of each line)
+//     xy    matches x followed by y
+//
+//   If you accidentally use PCRE or POSIX extended regex features
+//   not implemented by us, you will get a run-time failure.  In that
+//   case, please try to rewrite your regular expression within the
+//   above syntax.
+//
+//   This implementation is *not* meant to be as highly tuned or robust
+//   as a compiled regex library, but should perform well enough for a
+//   death test, which already incurs significant overhead by launching
+//   a child process.
+//
+// Known caveats:
+//
+//   A "threadsafe" style death test obtains the path to the test
+//   program from argv[0] and re-executes it in the sub-process.  For
+//   simplicity, the current implementation doesn't search the PATH
+//   when launching the sub-process.  This means that the user must
+//   invoke the test program via a path that contains at least one
+//   path separator (e.g. path/to/foo_test and
+//   /absolute/path/to/bar_test are fine, but foo_test is not).  This
+//   is rarely a problem as people usually don't put the test binary
+//   directory in PATH.
+//
+// TODO(wan@google.com): make thread-safe death tests search the PATH.
+
+// Asserts that a given statement causes the program to exit, with an
+// integer exit status that satisfies predicate, and emitting error output
+// that matches regex.
+# define ASSERT_EXIT(statement, predicate, regex) \
+    GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_FATAL_FAILURE_)
+
+// Like ASSERT_EXIT, but continues on to successive tests in the
+// test case, if any:
+# define EXPECT_EXIT(statement, predicate, regex) \
+    GTEST_DEATH_TEST_(statement, predicate, regex, GTEST_NONFATAL_FAILURE_)
+
+// Asserts that a given statement causes the program to exit, either by
+// explicitly exiting with a nonzero exit code or being killed by a
+// signal, and emitting error output that matches regex.
+# define ASSERT_DEATH(statement, regex) \
+    ASSERT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex)
+
+// Like ASSERT_DEATH, but continues on to successive tests in the
+// test case, if any:
+# define EXPECT_DEATH(statement, regex) \
+    EXPECT_EXIT(statement, ::testing::internal::ExitedUnsuccessfully, regex)
+
+// Two predicate classes that can be used in {ASSERT,EXPECT}_EXIT*:
+
+// Tests that an exit code describes a normal exit with a given exit code.
+    class GTEST_API_ ExitedWithCode
+    {
+    public:
+        explicit ExitedWithCode(int exit_code);
+        bool operator()(int exit_status) const;
+    private:
+        // No implementation - assignment is unsupported.
+        void operator=(const ExitedWithCode &other);
+
+        const int exit_code_;
+    };
+
+# if !GTEST_OS_WINDOWS
+// Tests that an exit code describes an exit due to termination by a
+// given signal.
+    class GTEST_API_ KilledBySignal
+    {
+    public:
+        explicit KilledBySignal(int signum);
+        bool operator()(int exit_status) const;
+    private:
+        const int signum_;
+    };
+# endif  // !GTEST_OS_WINDOWS
+
+// EXPECT_DEBUG_DEATH asserts that the given statements die in debug mode.
+// The death testing framework causes this to have interesting semantics,
+// since the sideeffects of the call are only visible in opt mode, and not
+// in debug mode.
+//
+// In practice, this can be used to test functions that utilize the
+// LOG(DFATAL) macro using the following style:
+//
+// int DieInDebugOr12(int* sideeffect) {
+//   if (sideeffect) {
+//     *sideeffect = 12;
+//   }
+//   LOG(DFATAL) << "death";
+//   return 12;
+// }
+//
+// TEST(TestCase, TestDieOr12WorksInDgbAndOpt) {
+//   int sideeffect = 0;
+//   // Only asserts in dbg.
+//   EXPECT_DEBUG_DEATH(DieInDebugOr12(&sideeffect), "death");
+//
+// #ifdef NDEBUG
+//   // opt-mode has sideeffect visible.
+//   EXPECT_EQ(12, sideeffect);
+// #else
+//   // dbg-mode no visible sideeffect.
+//   EXPECT_EQ(0, sideeffect);
+// #endif
+// }
+//
+// This will assert that DieInDebugReturn12InOpt() crashes in debug
+// mode, usually due to a DCHECK or LOG(DFATAL), but returns the
+// appropriate fallback value (12 in this case) in opt mode. If you
+// need to test that a function has appropriate side-effects in opt
+// mode, include assertions against the side-effects.  A general
+// pattern for this is:
+//
+// EXPECT_DEBUG_DEATH({
+//   // Side-effects here will have an effect after this statement in
+//   // opt mode, but none in debug mode.
+//   EXPECT_EQ(12, DieInDebugOr12(&sideeffect));
+// }, "death");
+//
+# ifdef NDEBUG
+
+#  define EXPECT_DEBUG_DEATH(statement, regex) \
+  GTEST_EXECUTE_STATEMENT_(statement, regex)
+
+#  define ASSERT_DEBUG_DEATH(statement, regex) \
+  GTEST_EXECUTE_STATEMENT_(statement, regex)
+
+# else
+
+#  define EXPECT_DEBUG_DEATH(statement, regex) \
+  EXPECT_DEATH(statement, regex)
+
+#  define ASSERT_DEBUG_DEATH(statement, regex) \
+  ASSERT_DEATH(statement, regex)
+
+# endif  // NDEBUG for EXPECT_DEBUG_DEATH
+#endif  // GTEST_HAS_DEATH_TEST
+
+// EXPECT_DEATH_IF_SUPPORTED(statement, regex) and
+// ASSERT_DEATH_IF_SUPPORTED(statement, regex) expand to real death tests if
+// death tests are supported; otherwise they just issue a warning.  This is
+// useful when you are combining death test assertions with normal test
+// assertions in one test.
+#if GTEST_HAS_DEATH_TEST
+# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \
+    EXPECT_DEATH(statement, regex)
+# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \
+    ASSERT_DEATH(statement, regex)
+#else
+# define EXPECT_DEATH_IF_SUPPORTED(statement, regex) \
+    GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, )
+# define ASSERT_DEATH_IF_SUPPORTED(statement, regex) \
+    GTEST_UNSUPPORTED_DEATH_TEST_(statement, regex, return)
+#endif
+
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_DEATH_TEST_H_
+// This file was GENERATED by command:
+//     pump.py gtest-param-test.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: vladl@google.com (Vlad Losev)
+//
+// Macros and functions for implementing parameterized tests
+// in Google C++ Testing Framework (Google Test)
+//
+// This file is generated by a SCRIPT.  DO NOT EDIT BY HAND!
+//
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+
+
+// Value-parameterized tests allow you to test your code with different
+// parameters without writing multiple copies of the same test.
+//
+// Here is how you use value-parameterized tests:
+
+#if 0
+
+// To write value-parameterized tests, first you should define a fixture
+// class. It is usually derived from testing::TestWithParam<T> (see below for
+// another inheritance scheme that's sometimes useful in more complicated
+// class hierarchies), where the type of your parameter values.
+// TestWithParam<T> is itself derived from testing::Test. T can be any
+// copyable type. If it's a raw pointer, you are responsible for managing the
+// lifespan of the pointed values.
+
+class FooTest : public ::testing::TestWithParam<const char *>
+{
+    // You can implement all the usual class fixture members here.
+};
+
+// Then, use the TEST_P macro to define as many parameterized tests
+// for this fixture as you want. The _P suffix is for "parameterized"
+// or "pattern", whichever you prefer to think.
+
+TEST_P(FooTest, DoesBlah)
+{
+    // Inside a test, access the test parameter with the GetParam() method
+    // of the TestWithParam<T> class:
+    EXPECT_TRUE(foo.Blah(GetParam()));
+    ...
+}
+
+TEST_P(FooTest, HasBlahBlah)
+{
+    ...
+}
+
+// Finally, you can use INSTANTIATE_TEST_CASE_P to instantiate the test
+// case with any set of parameters you want. Google Test defines a number
+// of functions for generating test parameters. They return what we call
+// (surprise!) parameter generators. Here is a  summary of them, which
+// are all in the testing namespace:
+//
+//
+//  Range(begin, end [, step]) - Yields values {begin, begin+step,
+//                               begin+step+step, ...}. The values do not
+//                               include end. step defaults to 1.
+//  Values(v1, v2, ..., vN)    - Yields values {v1, v2, ..., vN}.
+//  ValuesIn(container)        - Yields values from a C-style array, an STL
+//  ValuesIn(begin,end)          container, or an iterator range [begin, end).
+//  Bool()                     - Yields sequence {false, true}.
+//  Combine(g1, g2, ..., gN)   - Yields all combinations (the Cartesian product
+//                               for the math savvy) of the values generated
+//                               by the N generators.
+//
+// For more details, see comments at the definitions of these functions below
+// in this file.
+//
+// The following statement will instantiate tests from the FooTest test case
+// each with parameter values "meeny", "miny", and "moe".
+
+INSTANTIATE_TEST_CASE_P(InstantiationName,
+                        FooTest,
+                        Values("meeny", "miny", "moe"));
+
+// To distinguish different instances of the pattern, (yes, you
+// can instantiate it more then once) the first argument to the
+// INSTANTIATE_TEST_CASE_P macro is a prefix that will be added to the
+// actual test case name. Remember to pick unique prefixes for different
+// instantiations. The tests from the instantiation above will have
+// these names:
+//
+//    * InstantiationName/FooTest.DoesBlah/0 for "meeny"
+//    * InstantiationName/FooTest.DoesBlah/1 for "miny"
+//    * InstantiationName/FooTest.DoesBlah/2 for "moe"
+//    * InstantiationName/FooTest.HasBlahBlah/0 for "meeny"
+//    * InstantiationName/FooTest.HasBlahBlah/1 for "miny"
+//    * InstantiationName/FooTest.HasBlahBlah/2 for "moe"
+//
+// You can use these names in --gtest_filter.
+//
+// This statement will instantiate all tests from FooTest again, each
+// with parameter values "cat" and "dog":
+
+const char *pets[] = {"cat", "dog"};
+INSTANTIATE_TEST_CASE_P(AnotherInstantiationName, FooTest, ValuesIn(pets));
+
+// The tests from the instantiation above will have these names:
+//
+//    * AnotherInstantiationName/FooTest.DoesBlah/0 for "cat"
+//    * AnotherInstantiationName/FooTest.DoesBlah/1 for "dog"
+//    * AnotherInstantiationName/FooTest.HasBlahBlah/0 for "cat"
+//    * AnotherInstantiationName/FooTest.HasBlahBlah/1 for "dog"
+//
+// Please note that INSTANTIATE_TEST_CASE_P will instantiate all tests
+// in the given test case, whether their definitions come before or
+// AFTER the INSTANTIATE_TEST_CASE_P statement.
+//
+// Please also note that generator expressions (including parameters to the
+// generators) are evaluated in InitGoogleTest(), after main() has started.
+// This allows the user on one hand, to adjust generator parameters in order
+// to dynamically determine a set of tests to run and on the other hand,
+// give the user a chance to inspect the generated tests with Google Test
+// reflection API before RUN_ALL_TESTS() is executed.
+//
+// You can see samples/sample7_unittest.cc and samples/sample8_unittest.cc
+// for more examples.
+//
+// In the future, we plan to publish the API for defining new parameter
+// generators. But for now this interface remains part of the internal
+// implementation and is subject to change.
+//
+//
+// A parameterized test fixture must be derived from testing::Test and from
+// testing::WithParamInterface<T>, where T is the type of the parameter
+// values. Inheriting from TestWithParam<T> satisfies that requirement because
+// TestWithParam<T> inherits from both Test and WithParamInterface. In more
+// complicated hierarchies, however, it is occasionally useful to inherit
+// separately from Test and WithParamInterface. For example:
+
+class BaseTest : public ::testing::Test
+{
+    // You can inherit all the usual members for a non-parameterized test
+    // fixture here.
+};
+
+class DerivedTest : public BaseTest, public ::testing::WithParamInterface<int>
+{
+    // The usual test fixture members go here too.
+};
+
+TEST_F(BaseTest, HasFoo)
+{
+    // This is an ordinary non-parameterized test.
+}
+
+TEST_P(DerivedTest, DoesBlah)
+{
+    // GetParam works just the same here as if you inherit from TestWithParam.
+    EXPECT_TRUE(foo.Blah(GetParam()));
+}
+
+#endif  // 0
+
+
+#if !GTEST_OS_SYMBIAN
+# include <utility>
+#endif
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*.  Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// Type and function utilities for implementing parameterized tests.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+
+#include <ctype.h>
+
+#include <iterator>
+#include <set>
+#include <utility>
+#include <vector>
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*.  Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+// Copyright 2003 Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Authors: Dan Egnor (egnor@google.com)
+//
+// A "smart" pointer type with reference tracking.  Every pointer to a
+// particular object is kept on a circular linked list.  When the last pointer
+// to an object is destroyed or reassigned, the object is deleted.
+//
+// Used properly, this deletes the object when the last reference goes away.
+// There are several caveats:
+// - Like all reference counting schemes, cycles lead to leaks.
+// - Each smart pointer is actually two pointers (8 bytes instead of 4).
+// - Every time a pointer is assigned, the entire list of pointers to that
+//   object is traversed.  This class is therefore NOT SUITABLE when there
+//   will often be more than two or three pointers to a particular object.
+// - References are only tracked as long as linked_ptr<> objects are copied.
+//   If a linked_ptr<> is converted to a raw pointer and back, BAD THINGS
+//   will happen (double deletion).
+//
+// A good use of this class is storing object references in STL containers.
+// You can safely put linked_ptr<> in a vector<>.
+// Other uses may not be as good.
+//
+// Note: If you use an incomplete type with linked_ptr<>, the class
+// *containing* linked_ptr<> must have a constructor and destructor (even
+// if they do nothing!).
+//
+// Bill Gibbons suggested we use something like this.
+//
+// Thread Safety:
+//   Unlike other linked_ptr implementations, in this implementation
+//   a linked_ptr object is thread-safe in the sense that:
+//     - it's safe to copy linked_ptr objects concurrently,
+//     - it's safe to copy *from* a linked_ptr and read its underlying
+//       raw pointer (e.g. via get()) concurrently, and
+//     - it's safe to write to two linked_ptrs that point to the same
+//       shared object concurrently.
+// TODO(wan@google.com): rename this to safe_linked_ptr to avoid
+// confusion with normal linked_ptr.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+
+#include <stdlib.h>
+#include <assert.h>
+
+
+namespace testing
+{
+    namespace internal
+    {
+
+// Protects copying of all linked_ptr objects.
+        GTEST_API_ GTEST_DECLARE_STATIC_MUTEX_(g_linked_ptr_mutex);
+
+// This is used internally by all instances of linked_ptr<>.  It needs to be
+// a non-template class because different types of linked_ptr<> can refer to
+// the same object (linked_ptr<Superclass>(obj) vs linked_ptr<Subclass>(obj)).
+// So, it needs to be possible for different types of linked_ptr to participate
+// in the same circular linked list, so we need a single class type here.
+//
+// DO NOT USE THIS CLASS DIRECTLY YOURSELF.  Use linked_ptr<T>.
+        class linked_ptr_internal
+        {
+        public:
+            // Create a new circle that includes only this instance.
+            void join_new()
+            {
+                next_ = this;
+            }
+
+            // Many linked_ptr operations may change p.link_ for some linked_ptr
+            // variable p in the same circle as this object.  Therefore we need
+            // to prevent two such operations from occurring concurrently.
+            //
+            // Note that different types of linked_ptr objects can coexist in a
+            // circle (e.g. linked_ptr<Base>, linked_ptr<Derived1>, and
+            // linked_ptr<Derived2>).  Therefore we must use a single mutex to
+            // protect all linked_ptr objects.  This can create serious
+            // contention in production code, but is acceptable in a testing
+            // framework.
+
+            // Join an existing circle.
+            void join(linked_ptr_internal const *ptr)
+            GTEST_LOCK_EXCLUDED_(g_linked_ptr_mutex)
+            {
+                MutexLock lock(&g_linked_ptr_mutex);
+
+                linked_ptr_internal const *p = ptr;
+                while (p->next_ != ptr) {
+                    assert(p->next_ != this &&
+                           "Trying to join() a linked ring we are already in. "
+                           "Is GMock thread safety enabled?");
+                    p = p->next_;
+                }
+                p->next_ = this;
+                next_ = ptr;
+            }
+
+            // Leave whatever circle we're part of.  Returns true if we were the
+            // last member of the circle.  Once this is done, you can join() another.
+            bool depart()
+            GTEST_LOCK_EXCLUDED_(g_linked_ptr_mutex)
+            {
+                MutexLock lock(&g_linked_ptr_mutex);
+
+                if (next_ == this) return true;
+                linked_ptr_internal const *p = next_;
+                while (p->next_ != this) {
+                    assert(p->next_ != next_ &&
+                           "Trying to depart() a linked ring we are not in. "
+                           "Is GMock thread safety enabled?");
+                    p = p->next_;
+                }
+                p->next_ = next_;
+                return false;
+            }
+
+        private:
+            mutable linked_ptr_internal const *next_;
+        };
+
+        template <typename T>
+        class linked_ptr
+        {
+        public:
+            typedef T element_type;
+
+            // Take over ownership of a raw pointer.  This should happen as soon as
+            // possible after the object is created.
+            explicit linked_ptr(T *ptr = NULL)
+            {
+                capture(ptr);
+            }
+            ~linked_ptr()
+            {
+                depart();
+            }
+
+            // Copy an existing linked_ptr<>, adding ourselves to the list of references.
+            template <typename U> linked_ptr(linked_ptr<U> const &ptr)
+            {
+                copy(&ptr);
+            }
+            linked_ptr(linked_ptr const &ptr)    // NOLINT
+            {
+                assert(&ptr != this);
+                copy(&ptr);
+            }
+
+            // Assignment releases the old value and acquires the new.
+            template <typename U> linked_ptr &operator=(linked_ptr<U> const &ptr)
+            {
+                depart();
+                copy(&ptr);
+                return *this;
+            }
+
+            linked_ptr &operator=(linked_ptr const &ptr)
+            {
+                if (&ptr != this) {
+                    depart();
+                    copy(&ptr);
+                }
+                return *this;
+            }
+
+            // Smart pointer members.
+            void reset(T *ptr = NULL)
+            {
+                depart();
+                capture(ptr);
+            }
+            T *get() const
+            {
+                return value_;
+            }
+            T *operator->() const
+            {
+                return value_;
+            }
+            T &operator*() const
+            {
+                return *value_;
+            }
+
+            bool operator==(T *p) const
+            {
+                return value_ == p;
+            }
+            bool operator!=(T *p) const
+            {
+                return value_ != p;
+            }
+            template <typename U>
+            bool operator==(linked_ptr<U> const &ptr) const
+            {
+                return value_ == ptr.get();
+            }
+            template <typename U>
+            bool operator!=(linked_ptr<U> const &ptr) const
+            {
+                return value_ != ptr.get();
+            }
+
+        private:
+            template <typename U>
+            friend class linked_ptr;
+
+            T *value_;
+            linked_ptr_internal link_;
+
+            void depart()
+            {
+                if (link_.depart()) delete value_;
+            }
+
+            void capture(T *ptr)
+            {
+                value_ = ptr;
+                link_.join_new();
+            }
+
+            template <typename U> void copy(linked_ptr<U> const *ptr)
+            {
+                value_ = ptr->get();
+                if (value_)
+                    link_.join(&ptr->link_);
+                else
+                    link_.join_new();
+            }
+        };
+
+        template<typename T> inline
+        bool operator==(T *ptr, const linked_ptr<T> &x)
+        {
+            return ptr == x.get();
+        }
+
+        template<typename T> inline
+        bool operator!=(T *ptr, const linked_ptr<T> &x)
+        {
+            return ptr != x.get();
+        }
+
+// A function to convert T* into linked_ptr<T>
+// Doing e.g. make_linked_ptr(new FooBarBaz<type>(arg)) is a shorter notation
+// for linked_ptr<FooBarBaz<type> >(new FooBarBaz<type>(arg))
+        template <typename T>
+        linked_ptr<T> make_linked_ptr(T *ptr)
+        {
+            return linked_ptr<T>(ptr);
+        }
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_LINKED_PTR_H_
+// Copyright 2007, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+// Google Test - The Google C++ Testing Framework
+//
+// This file implements a universal value printer that can print a
+// value of any type T:
+//
+//   void ::testing::internal::UniversalPrinter<T>::Print(value, ostream_ptr);
+//
+// A user can teach this function how to print a class type T by
+// defining either operator<<() or PrintTo() in the namespace that
+// defines T.  More specifically, the FIRST defined function in the
+// following list will be used (assuming T is defined in namespace
+// foo):
+//
+//   1. foo::PrintTo(const T&, ostream*)
+//   2. operator<<(ostream&, const T&) defined in either foo or the
+//      global namespace.
+//
+// If none of the above is defined, it will print the debug string of
+// the value if it is a protocol buffer, or print the raw bytes in the
+// value otherwise.
+//
+// To aid debugging: when T is a reference type, the address of the
+// value is also printed; when T is a (const) char pointer, both the
+// pointer value and the NUL-terminated string it points to are
+// printed.
+//
+// We also provide some convenient wrappers:
+//
+//   // Prints a value to a string.  For a (const or not) char
+//   // pointer, the NUL-terminated string (but not the pointer) is
+//   // printed.
+//   std::string ::testing::PrintToString(const T& value);
+//
+//   // Prints a value tersely: for a reference type, the referenced
+//   // value (but not the address) is printed; for a (const or not) char
+//   // pointer, the NUL-terminated string (but not the pointer) is
+//   // printed.
+//   void ::testing::internal::UniversalTersePrint(const T& value, ostream*);
+//
+//   // Prints value using the type inferred by the compiler.  The difference
+//   // from UniversalTersePrint() is that this function prints both the
+//   // pointer and the NUL-terminated string for a (const or not) char pointer.
+//   void ::testing::internal::UniversalPrint(const T& value, ostream*);
+//
+//   // Prints the fields of a tuple tersely to a string vector, one
+//   // element for each field. Tuple support must be enabled in
+//   // gtest-port.h.
+//   std::vector<string> UniversalTersePrintTupleFieldsToStrings(
+//       const Tuple& value);
+//
+// Known limitation:
+//
+// The print primitives print the elements of an STL-style container
+// using the compiler-inferred type of *iter where iter is a
+// const_iterator of the container.  When const_iterator is an input
+// iterator but not a forward iterator, this inferred type may not
+// match value_type, and the print output may be incorrect.  In
+// practice, this is rarely a problem as for most containers
+// const_iterator is a forward iterator.  We'll fix this if there's an
+// actual need for it.  Note that this fix cannot rely on value_type
+// being defined as many user-defined container types don't have
+// value_type.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+
+#include <ostream>  // NOLINT
+#include <sstream>
+#include <string>
+#include <utility>
+#include <vector>
+
+#if GTEST_HAS_STD_TUPLE_
+# include <tuple>
+#endif
+
+namespace testing
+{
+
+// Definitions in the 'internal' and 'internal2' name spaces are
+// subject to change without notice.  DO NOT USE THEM IN USER CODE!
+    namespace internal2
+    {
+
+// Prints the given number of bytes in the given object to the given
+// ostream.
+        GTEST_API_ void PrintBytesInObjectTo(const unsigned char *obj_bytes,
+                                             size_t count,
+                                             ::std::ostream *os);
+
+// For selecting which printer to use when a given type has neither <<
+// nor PrintTo().
+        enum TypeKind {
+            kProtobuf,              // a protobuf type
+            kConvertibleToInteger,  // a type implicitly convertible to BiggestInt
+            // (e.g. a named or unnamed enum type)
+            kOtherType              // anything else
+        };
+
+// TypeWithoutFormatter<T, kTypeKind>::PrintValue(value, os) is called
+// by the universal printer to print a value of type T when neither
+// operator<< nor PrintTo() is defined for T, where kTypeKind is the
+// "kind" of T as defined by enum TypeKind.
+        template <typename T, TypeKind kTypeKind>
+        class TypeWithoutFormatter
+        {
+        public:
+            // This default version is called when kTypeKind is kOtherType.
+            static void PrintValue(const T &value, ::std::ostream *os)
+            {
+                PrintBytesInObjectTo(reinterpret_cast<const unsigned char *>(&value),
+                                     sizeof(value), os);
+            }
+        };
+
+// We print a protobuf using its ShortDebugString() when the string
+// doesn't exceed this many characters; otherwise we print it using
+// DebugString() for better readability.
+        const size_t kProtobufOneLinerMaxLength = 50;
+
+        template <typename T>
+        class TypeWithoutFormatter<T, kProtobuf>
+        {
+        public:
+            static void PrintValue(const T &value, ::std::ostream *os)
+            {
+                std::string pretty_str = value.ShortDebugString();
+                if (pretty_str.length() > kProtobufOneLinerMaxLength) {
+                    pretty_str = "\n" + value.DebugString();
+                }
+                *os << ("<" + pretty_str + ">");
+            }
+        };
+
+        template <typename T>
+        class TypeWithoutFormatter<T, kConvertibleToInteger>
+        {
+        public:
+            // Since T has no << operator or PrintTo() but can be implicitly
+            // converted to BiggestInt, we print it as a BiggestInt.
+            //
+            // Most likely T is an enum type (either named or unnamed), in which
+            // case printing it as an integer is the desired behavior.  In case
+            // T is not an enum, printing it as an integer is the best we can do
+            // given that it has no user-defined printer.
+            static void PrintValue(const T &value, ::std::ostream *os)
+            {
+                const internal::BiggestInt kBigInt = value;
+                *os << kBigInt;
+            }
+        };
+
+// Prints the given value to the given ostream.  If the value is a
+// protocol message, its debug string is printed; if it's an enum or
+// of a type implicitly convertible to BiggestInt, it's printed as an
+// integer; otherwise the bytes in the value are printed.  This is
+// what UniversalPrinter<T>::Print() does when it knows nothing about
+// type T and T has neither << operator nor PrintTo().
+//
+// A user can override this behavior for a class type Foo by defining
+// a << operator in the namespace where Foo is defined.
+//
+// We put this operator in namespace 'internal2' instead of 'internal'
+// to simplify the implementation, as much code in 'internal' needs to
+// use << in STL, which would conflict with our own << were it defined
+// in 'internal'.
+//
+// Note that this operator<< takes a generic std::basic_ostream<Char,
+// CharTraits> type instead of the more restricted std::ostream.  If
+// we define it to take an std::ostream instead, we'll get an
+// "ambiguous overloads" compiler error when trying to print a type
+// Foo that supports streaming to std::basic_ostream<Char,
+// CharTraits>, as the compiler cannot tell whether
+// operator<<(std::ostream&, const T&) or
+// operator<<(std::basic_stream<Char, CharTraits>, const Foo&) is more
+// specific.
+        template <typename Char, typename CharTraits, typename T>
+        ::std::basic_ostream<Char, CharTraits> &operator<<(
+            ::std::basic_ostream<Char, CharTraits> &os, const T &x)
+        {
+            TypeWithoutFormatter<T,
+                                 (internal::IsAProtocolMessage<T>::value ? kProtobuf :
+                                  internal::ImplicitlyConvertible<const T &, internal::BiggestInt>::value ?
+                                  kConvertibleToInteger : kOtherType)>::PrintValue(x, &os);
+            return os;
+        }
+
+    }  // namespace internal2
+}  // namespace testing
+
+// This namespace MUST NOT BE NESTED IN ::testing, or the name look-up
+// magic needed for implementing UniversalPrinter won't work.
+namespace testing_internal
+{
+
+// Used to print a value that is not an STL-style container when the
+// user doesn't define PrintTo() for it.
+    template <typename T>
+    void DefaultPrintNonContainerTo(const T &value, ::std::ostream *os)
+    {
+        // With the following statement, during unqualified name lookup,
+        // testing::internal2::operator<< appears as if it was declared in
+        // the nearest enclosing namespace that contains both
+        // ::testing_internal and ::testing::internal2, i.e. the global
+        // namespace.  For more details, refer to the C++ Standard section
+        // 7.3.4-1 [namespace.udir].  This allows us to fall back onto
+        // testing::internal2::operator<< in case T doesn't come with a <<
+        // operator.
+        //
+        // We cannot write 'using ::testing::internal2::operator<<;', which
+        // gcc 3.3 fails to compile due to a compiler bug.
+        using namespace ::testing::internal2;  // NOLINT
+
+        // Assuming T is defined in namespace foo, in the next statement,
+        // the compiler will consider all of:
+        //
+        //   1. foo::operator<< (thanks to Koenig look-up),
+        //   2. ::operator<< (as the current namespace is enclosed in ::),
+        //   3. testing::internal2::operator<< (thanks to the using statement above).
+        //
+        // The operator<< whose type matches T best will be picked.
+        //
+        // We deliberately allow #2 to be a candidate, as sometimes it's
+        // impossible to define #1 (e.g. when foo is ::std, defining
+        // anything in it is undefined behavior unless you are a compiler
+        // vendor.).
+        *os << value;
+    }
+
+}  // namespace testing_internal
+
+namespace testing
+{
+    namespace internal
+    {
+
+// FormatForComparison<ToPrint, OtherOperand>::Format(value) formats a
+// value of type ToPrint that is an operand of a comparison assertion
+// (e.g. ASSERT_EQ).  OtherOperand is the type of the other operand in
+// the comparison, and is used to help determine the best way to
+// format the value.  In particular, when the value is a C string
+// (char pointer) and the other operand is an STL string object, we
+// want to format the C string as a string, since we know it is
+// compared by value with the string object.  If the value is a char
+// pointer but the other operand is not an STL string object, we don't
+// know whether the pointer is supposed to point to a NUL-terminated
+// string, and thus want to print it as a pointer to be safe.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+
+// The default case.
+        template <typename ToPrint, typename OtherOperand>
+        class FormatForComparison
+        {
+        public:
+            static ::std::string Format(const ToPrint &value)
+            {
+                return ::testing::PrintToString(value);
+            }
+        };
+
+// Array.
+        template <typename ToPrint, size_t N, typename OtherOperand>
+        class FormatForComparison<ToPrint[N], OtherOperand>
+        {
+        public:
+            static ::std::string Format(const ToPrint *value)
+            {
+                return FormatForComparison<const ToPrint *, OtherOperand>::Format(value);
+            }
+        };
+
+// By default, print C string as pointers to be safe, as we don't know
+// whether they actually point to a NUL-terminated string.
+
+#define GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(CharType)                \
+  template <typename OtherOperand>                                      \
+  class FormatForComparison<CharType*, OtherOperand> {                  \
+   public:                                                              \
+    static ::std::string Format(CharType* value) {                      \
+      return ::testing::PrintToString(static_cast<const void*>(value)); \
+    }                                                                   \
+  }
+
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(char);
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const char);
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(wchar_t);
+        GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_(const wchar_t);
+
+#undef GTEST_IMPL_FORMAT_C_STRING_AS_POINTER_
+
+// If a C string is compared with an STL string object, we know it's meant
+// to point to a NUL-terminated string, and thus can print it as a string.
+
+#define GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(CharType, OtherStringType) \
+  template <>                                                           \
+  class FormatForComparison<CharType*, OtherStringType> {               \
+   public:                                                              \
+    static ::std::string Format(CharType* value) {                      \
+      return ::testing::PrintToString(value);                           \
+    }                                                                   \
+  }
+
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(char, ::std::string);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const char, ::std::string);
+
+#if GTEST_HAS_GLOBAL_STRING
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(char, ::string);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const char, ::string);
+#endif
+
+#if GTEST_HAS_GLOBAL_WSTRING
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(wchar_t, ::wstring);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const wchar_t, ::wstring);
+#endif
+
+#if GTEST_HAS_STD_WSTRING
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(wchar_t, ::std::wstring);
+        GTEST_IMPL_FORMAT_C_STRING_AS_STRING_(const wchar_t, ::std::wstring);
+#endif
+
+#undef GTEST_IMPL_FORMAT_C_STRING_AS_STRING_
+
+// Formats a comparison assertion (e.g. ASSERT_EQ, EXPECT_LT, and etc)
+// operand to be used in a failure message.  The type (but not value)
+// of the other operand may affect the format.  This allows us to
+// print a char* as a raw pointer when it is compared against another
+// char* or void*, and print it as a C string when it is compared
+// against an std::string object, for example.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        template <typename T1, typename T2>
+        std::string FormatForComparisonFailureMessage(
+            const T1 &value, const T2 & /* other_operand */)
+        {
+            return FormatForComparison<T1, T2>::Format(value);
+        }
+
+// UniversalPrinter<T>::Print(value, ostream_ptr) prints the given
+// value to the given ostream.  The caller must ensure that
+// 'ostream_ptr' is not NULL, or the behavior is undefined.
+//
+// We define UniversalPrinter as a class template (as opposed to a
+// function template), as we need to partially specialize it for
+// reference types, which cannot be done with function templates.
+        template <typename T>
+        class UniversalPrinter;
+
+        template <typename T>
+        void UniversalPrint(const T &value, ::std::ostream *os);
+
+        enum DefaultPrinterType {
+            kPrintContainer,
+            kPrintPointer,
+            kPrintFunctionPointer,
+            kPrintOther,
+        };
+        template <DefaultPrinterType type> struct WrapPrinterType {};
+
+// Used to print an STL-style container when the user doesn't define
+// a PrintTo() for it.
+        template <typename C>
+        void DefaultPrintTo(WrapPrinterType<kPrintContainer> /* dummy */,
+                            const C &container, ::std::ostream *os)
+        {
+            const size_t kMaxCount = 32;  // The maximum number of elements to print.
+            *os << '{';
+            size_t count = 0;
+            for (typename C::const_iterator it = container.begin();
+                 it != container.end(); ++it, ++count) {
+                if (count > 0) {
+                    *os << ',';
+                    if (count == kMaxCount) {  // Enough has been printed.
+                        *os << " ...";
+                        break;
+                    }
+                }
+                *os << ' ';
+                // We cannot call PrintTo(*it, os) here as PrintTo() doesn't
+                // handle *it being a native array.
+                internal::UniversalPrint(*it, os);
+            }
+
+            if (count > 0) {
+                *os << ' ';
+            }
+            *os << '}';
+        }
+
+// Used to print a pointer that is neither a char pointer nor a member
+// pointer, when the user doesn't define PrintTo() for it.  (A member
+// variable pointer or member function pointer doesn't really point to
+// a location in the address space.  Their representation is
+// implementation-defined.  Therefore they will be printed as raw
+// bytes.)
+        template <typename T>
+        void DefaultPrintTo(WrapPrinterType<kPrintPointer> /* dummy */,
+                            T *p, ::std::ostream *os)
+        {
+            if (p == NULL) {
+                *os << "NULL";
+            } else {
+                // T is not a function type.  We just call << to print p,
+                // relying on ADL to pick up user-defined << for their pointer
+                // types, if any.
+                *os << p;
+            }
+        }
+        template <typename T>
+        void DefaultPrintTo(WrapPrinterType<kPrintFunctionPointer> /* dummy */,
+                            T *p, ::std::ostream *os)
+        {
+            if (p == NULL) {
+                *os << "NULL";
+            } else {
+                // T is a function type, so '*os << p' doesn't do what we want
+                // (it just prints p as bool).  We want to print p as a const
+                // void*.  However, we cannot cast it to const void* directly,
+                // even using reinterpret_cast, as earlier versions of gcc
+                // (e.g. 3.4.5) cannot compile the cast when p is a function
+                // pointer.  Casting to UInt64 first solves the problem.
+                *os << reinterpret_cast<const void *>(
+                        reinterpret_cast<internal::UInt64>(p));
+            }
+        }
+
+// Used to print a non-container, non-pointer value when the user
+// doesn't define PrintTo() for it.
+        template <typename T>
+        void DefaultPrintTo(WrapPrinterType<kPrintOther> /* dummy */,
+                            const T &value, ::std::ostream *os)
+        {
+            ::testing_internal::DefaultPrintNonContainerTo(value, os);
+        }
+
+// Prints the given value using the << operator if it has one;
+// otherwise prints the bytes in it.  This is what
+// UniversalPrinter<T>::Print() does when PrintTo() is not specialized
+// or overloaded for type T.
+//
+// A user can override this behavior for a class type Foo by defining
+// an overload of PrintTo() in the namespace where Foo is defined.  We
+// give the user this option as sometimes defining a << operator for
+// Foo is not desirable (e.g. the coding style may prevent doing it,
+// or there is already a << operator but it doesn't do what the user
+// wants).
+        template <typename T>
+        void PrintTo(const T &value, ::std::ostream *os)
+        {
+            // DefaultPrintTo() is overloaded.  The type of its first argument
+            // determines which version will be picked.
+            //
+            // Note that we check for container types here, prior to we check
+            // for protocol message types in our operator<<.  The rationale is:
+            //
+            // For protocol messages, we want to give people a chance to
+            // override Google Mock's format by defining a PrintTo() or
+            // operator<<.  For STL containers, other formats can be
+            // incompatible with Google Mock's format for the container
+            // elements; therefore we check for container types here to ensure
+            // that our format is used.
+            //
+            // Note that MSVC and clang-cl do allow an implicit conversion from
+            // pointer-to-function to pointer-to-object, but clang-cl warns on it.
+            // So don't use ImplicitlyConvertible if it can be helped since it will
+            // cause this warning, and use a separate overload of DefaultPrintTo for
+            // function pointers so that the `*os << p` in the object pointer overload
+            // doesn't cause that warning either.
+            DefaultPrintTo(
+                WrapPrinterType<sizeof(IsContainerTest<T>(0)) == sizeof(IsContainer)
+                ? kPrintContainer : !is_pointer<T>::value
+                ? kPrintOther
+#if GTEST_LANG_CXX11
+                : std::is_function<typename std::remove_pointer<T>::type>::value
+#else
+                : !internal::ImplicitlyConvertible<T, const void *>::value
+#endif
+                ? kPrintFunctionPointer
+                : kPrintPointer>(),
+                value, os);
+        }
+
+// The following list of PrintTo() overloads tells
+// UniversalPrinter<T>::Print() how to print standard types (built-in
+// types, strings, plain arrays, and pointers).
+
+// Overloads for various char types.
+        GTEST_API_ void PrintTo(unsigned char c, ::std::ostream *os);
+        GTEST_API_ void PrintTo(signed char c, ::std::ostream *os);
+        inline void PrintTo(char c, ::std::ostream *os)
+        {
+            // When printing a plain char, we always treat it as unsigned.  This
+            // way, the output won't be affected by whether the compiler thinks
+            // char is signed or not.
+            PrintTo(static_cast<unsigned char>(c), os);
+        }
+
+// Overloads for other simple built-in types.
+        inline void PrintTo(bool x, ::std::ostream *os)
+        {
+            *os << (x ? "true" : "false");
+        }
+
+// Overload for wchar_t type.
+// Prints a wchar_t as a symbol if it is printable or as its internal
+// code otherwise and also as its decimal code (except for L'\0').
+// The L'\0' char is printed as "L'\\0'". The decimal code is printed
+// as signed integer when wchar_t is implemented by the compiler
+// as a signed type and is printed as an unsigned integer when wchar_t
+// is implemented as an unsigned type.
+        GTEST_API_ void PrintTo(wchar_t wc, ::std::ostream *os);
+
+// Overloads for C strings.
+        GTEST_API_ void PrintTo(const char *s, ::std::ostream *os);
+        inline void PrintTo(char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const char *>(s), os);
+        }
+
+// signed/unsigned char is often used for representing binary data, so
+// we print pointers to it as void* to be safe.
+        inline void PrintTo(const signed char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+        inline void PrintTo(signed char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+        inline void PrintTo(const unsigned char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+        inline void PrintTo(unsigned char *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const void *>(s), os);
+        }
+
+// MSVC can be configured to define wchar_t as a typedef of unsigned
+// short.  It defines _NATIVE_WCHAR_T_DEFINED when wchar_t is a native
+// type.  When wchar_t is a typedef, defining an overload for const
+// wchar_t* would cause unsigned short* be printed as a wide string,
+// possibly causing invalid memory accesses.
+#if !defined(_MSC_VER) || defined(_NATIVE_WCHAR_T_DEFINED)
+// Overloads for wide C strings
+        GTEST_API_ void PrintTo(const wchar_t *s, ::std::ostream *os);
+        inline void PrintTo(wchar_t *s, ::std::ostream *os)
+        {
+            PrintTo(ImplicitCast_<const wchar_t *>(s), os);
+        }
+#endif
+
+// Overload for C arrays.  Multi-dimensional arrays are printed
+// properly.
+
+// Prints the given number of elements in an array, without printing
+// the curly braces.
+        template <typename T>
+        void PrintRawArrayTo(const T a[], size_t count, ::std::ostream *os)
+        {
+            UniversalPrint(a[0], os);
+            for (size_t i = 1; i != count; i++) {
+                *os << ", ";
+                UniversalPrint(a[i], os);
+            }
+        }
+
+// Overloads for ::string and ::std::string.
+#if GTEST_HAS_GLOBAL_STRING
+        GTEST_API_ void PrintStringTo(const ::string &s, ::std::ostream *os);
+        inline void PrintTo(const ::string &s, ::std::ostream *os)
+        {
+            PrintStringTo(s, os);
+        }
+#endif  // GTEST_HAS_GLOBAL_STRING
+
+        GTEST_API_ void PrintStringTo(const ::std::string &s, ::std::ostream *os);
+        inline void PrintTo(const ::std::string &s, ::std::ostream *os)
+        {
+            PrintStringTo(s, os);
+        }
+
+// Overloads for ::wstring and ::std::wstring.
+#if GTEST_HAS_GLOBAL_WSTRING
+        GTEST_API_ void PrintWideStringTo(const ::wstring &s, ::std::ostream *os);
+        inline void PrintTo(const ::wstring &s, ::std::ostream *os)
+        {
+            PrintWideStringTo(s, os);
+        }
+#endif  // GTEST_HAS_GLOBAL_WSTRING
+
+#if GTEST_HAS_STD_WSTRING
+        GTEST_API_ void PrintWideStringTo(const ::std::wstring &s, ::std::ostream *os);
+        inline void PrintTo(const ::std::wstring &s, ::std::ostream *os)
+        {
+            PrintWideStringTo(s, os);
+        }
+#endif  // GTEST_HAS_STD_WSTRING
+
+#if GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+// Helper function for printing a tuple.  T must be instantiated with
+// a tuple type.
+        template <typename T>
+        void PrintTupleTo(const T &t, ::std::ostream *os);
+#endif  // GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+
+#if GTEST_HAS_TR1_TUPLE
+// Overload for ::std::tr1::tuple.  Needed for printing function arguments,
+// which are packed as tuples.
+
+// Overloaded PrintTo() for tuples of various arities.  We support
+// tuples of up-to 10 fields.  The following implementation works
+// regardless of whether tr1::tuple is implemented using the
+// non-standard variadic template feature or not.
+
+        inline void PrintTo(const ::std::tr1::tuple<> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1>
+        void PrintTo(const ::std::tr1::tuple<T1> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2>
+        void PrintTo(const ::std::tr1::tuple<T1, T2> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        void PrintTo(const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9> &t,
+                     ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        void PrintTo(
+            const ::std::tr1::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> &t,
+            ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+#endif  // GTEST_HAS_TR1_TUPLE
+
+#if GTEST_HAS_STD_TUPLE_
+        template <typename... Types>
+        void PrintTo(const ::std::tuple<Types...> &t, ::std::ostream *os)
+        {
+            PrintTupleTo(t, os);
+        }
+#endif  // GTEST_HAS_STD_TUPLE_
+
+// Overload for std::pair.
+        template <typename T1, typename T2>
+        void PrintTo(const ::std::pair<T1, T2> &value, ::std::ostream *os)
+        {
+            *os << '(';
+            // We cannot use UniversalPrint(value.first, os) here, as T1 may be
+            // a reference type.  The same for printing value.second.
+            UniversalPrinter<T1>::Print(value.first, os);
+            *os << ", ";
+            UniversalPrinter<T2>::Print(value.second, os);
+            *os << ')';
+        }
+
+// Implements printing a non-reference type T by letting the compiler
+// pick the right overload of PrintTo() for T.
+        template <typename T>
+        class UniversalPrinter
+        {
+        public:
+            // MSVC warns about adding const to a function type, so we want to
+            // disable the warning.
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180)
+
+            // Note: we deliberately don't call this PrintTo(), as that name
+            // conflicts with ::testing::internal::PrintTo in the body of the
+            // function.
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                // By default, ::testing::internal::PrintTo() is used for printing
+                // the value.
+                //
+                // Thanks to Koenig look-up, if T is a class and has its own
+                // PrintTo() function defined in its namespace, that function will
+                // be visible here.  Since it is more specific than the generic ones
+                // in ::testing::internal, it will be picked by the compiler in the
+                // following statement - exactly what we want.
+                PrintTo(value, os);
+            }
+
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+        };
+
+// UniversalPrintArray(begin, len, os) prints an array of 'len'
+// elements, starting at address 'begin'.
+        template <typename T>
+        void UniversalPrintArray(const T *begin, size_t len, ::std::ostream *os)
+        {
+            if (len == 0) {
+                *os << "{}";
+            } else {
+                *os << "{ ";
+                const size_t kThreshold = 18;
+                const size_t kChunkSize = 8;
+                // If the array has more than kThreshold elements, we'll have to
+                // omit some details by printing only the first and the last
+                // kChunkSize elements.
+                // TODO(wan@google.com): let the user control the threshold using a flag.
+                if (len <= kThreshold) {
+                    PrintRawArrayTo(begin, len, os);
+                } else {
+                    PrintRawArrayTo(begin, kChunkSize, os);
+                    *os << ", ..., ";
+                    PrintRawArrayTo(begin + len - kChunkSize, kChunkSize, os);
+                }
+                *os << " }";
+            }
+        }
+// This overload prints a (const) char array compactly.
+        GTEST_API_ void UniversalPrintArray(
+            const char *begin, size_t len, ::std::ostream *os);
+
+// This overload prints a (const) wchar_t array compactly.
+        GTEST_API_ void UniversalPrintArray(
+            const wchar_t *begin, size_t len, ::std::ostream *os);
+
+// Implements printing an array type T[N].
+        template <typename T, size_t N>
+        class UniversalPrinter<T[N]>
+        {
+        public:
+            // Prints the given array, omitting some elements when there are too
+            // many.
+            static void Print(const T (&a)[N], ::std::ostream *os)
+            {
+                UniversalPrintArray(a, N, os);
+            }
+        };
+
+// Implements printing a reference type T&.
+        template <typename T>
+        class UniversalPrinter<T &>
+        {
+        public:
+            // MSVC warns about adding const to a function type, so we want to
+            // disable the warning.
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4180)
+
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                // Prints the address of the value.  We use reinterpret_cast here
+                // as static_cast doesn't compile when T is a function type.
+                *os << "@" << reinterpret_cast<const void *>(&value) << " ";
+
+                // Then prints the value itself.
+                UniversalPrint(value, os);
+            }
+
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+        };
+
+// Prints a value tersely: for a reference type, the referenced value
+// (but not the address) is printed; for a (const) char pointer, the
+// NUL-terminated string (but not the pointer) is printed.
+
+        template <typename T>
+        class UniversalTersePrinter
+        {
+        public:
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                UniversalPrint(value, os);
+            }
+        };
+        template <typename T>
+        class UniversalTersePrinter<T &>
+        {
+        public:
+            static void Print(const T &value, ::std::ostream *os)
+            {
+                UniversalPrint(value, os);
+            }
+        };
+        template <typename T, size_t N>
+        class UniversalTersePrinter<T[N]>
+        {
+        public:
+            static void Print(const T (&value)[N], ::std::ostream *os)
+            {
+                UniversalPrinter<T[N]>::Print(value, os);
+            }
+        };
+        template <>
+        class UniversalTersePrinter<const char *>
+        {
+        public:
+            static void Print(const char *str, ::std::ostream *os)
+            {
+                if (str == NULL) {
+                    *os << "NULL";
+                } else {
+                    UniversalPrint(std::string(str), os);
+                }
+            }
+        };
+        template <>
+        class UniversalTersePrinter<char *>
+        {
+        public:
+            static void Print(char *str, ::std::ostream *os)
+            {
+                UniversalTersePrinter<const char *>::Print(str, os);
+            }
+        };
+
+#if GTEST_HAS_STD_WSTRING
+        template <>
+        class UniversalTersePrinter<const wchar_t *>
+        {
+        public:
+            static void Print(const wchar_t *str, ::std::ostream *os)
+            {
+                if (str == NULL) {
+                    *os << "NULL";
+                } else {
+                    UniversalPrint(::std::wstring(str), os);
+                }
+            }
+        };
+#endif
+
+        template <>
+        class UniversalTersePrinter<wchar_t *>
+        {
+        public:
+            static void Print(wchar_t *str, ::std::ostream *os)
+            {
+                UniversalTersePrinter<const wchar_t *>::Print(str, os);
+            }
+        };
+
+        template <typename T>
+        void UniversalTersePrint(const T &value, ::std::ostream *os)
+        {
+            UniversalTersePrinter<T>::Print(value, os);
+        }
+
+// Prints a value using the type inferred by the compiler.  The
+// difference between this and UniversalTersePrint() is that for a
+// (const) char pointer, this prints both the pointer and the
+// NUL-terminated string.
+        template <typename T>
+        void UniversalPrint(const T &value, ::std::ostream *os)
+        {
+            // A workarond for the bug in VC++ 7.1 that prevents us from instantiating
+            // UniversalPrinter with T directly.
+            typedef T T1;
+            UniversalPrinter<T1>::Print(value, os);
+        }
+
+        typedef ::std::vector<string> Strings;
+
+// TuplePolicy<TupleT> must provide:
+// - tuple_size
+//     size of tuple TupleT.
+// - get<size_t I>(const TupleT& t)
+//     static function extracting element I of tuple TupleT.
+// - tuple_element<size_t I>::type
+//     type of element I of tuple TupleT.
+        template <typename TupleT>
+        struct TuplePolicy;
+
+#if GTEST_HAS_TR1_TUPLE
+        template <typename TupleT>
+        struct TuplePolicy {
+            typedef TupleT Tuple;
+            static const size_t tuple_size = ::std::tr1::tuple_size<Tuple>::value;
+
+            template <size_t I>
+            struct tuple_element : ::std::tr1::tuple_element<I, Tuple> {};
+
+            template <size_t I>
+            static typename AddReference<
+            const typename ::std::tr1::tuple_element<I, Tuple>::type>::type get(
+                const Tuple &tuple)
+            {
+                return ::std::tr1::get<I>(tuple);
+            }
+        };
+        template <typename TupleT>
+        const size_t TuplePolicy<TupleT>::tuple_size;
+#endif  // GTEST_HAS_TR1_TUPLE
+
+#if GTEST_HAS_STD_TUPLE_
+        template <typename... Types>
+        struct TuplePolicy< ::std::tuple<Types...> > {
+            typedef ::std::tuple<Types...> Tuple;
+            static const size_t tuple_size = ::std::tuple_size<Tuple>::value;
+
+            template <size_t I>
+            struct tuple_element : ::std::tuple_element<I, Tuple> {};
+
+            template <size_t I>
+            static const typename ::std::tuple_element<I, Tuple>::type &get(
+                const Tuple &tuple)
+            {
+                return ::std::get<I>(tuple);
+            }
+        };
+        template <typename... Types>
+        const size_t TuplePolicy< ::std::tuple<Types...> >::tuple_size;
+#endif  // GTEST_HAS_STD_TUPLE_
+
+#if GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+// This helper template allows PrintTo() for tuples and
+// UniversalTersePrintTupleFieldsToStrings() to be defined by
+// induction on the number of tuple fields.  The idea is that
+// TuplePrefixPrinter<N>::PrintPrefixTo(t, os) prints the first N
+// fields in tuple t, and can be defined in terms of
+// TuplePrefixPrinter<N - 1>.
+//
+// The inductive case.
+        template <size_t N>
+        struct TuplePrefixPrinter {
+            // Prints the first N fields of a tuple.
+            template <typename Tuple>
+            static void PrintPrefixTo(const Tuple &t, ::std::ostream *os)
+            {
+                TuplePrefixPrinter<N - 1>::PrintPrefixTo(t, os);
+                GTEST_INTENTIONAL_CONST_COND_PUSH_()
+                if (N > 1) {
+                    GTEST_INTENTIONAL_CONST_COND_POP_()
+                    *os << ", ";
+                }
+                UniversalPrinter<
+                typename TuplePolicy<Tuple>::template tuple_element<N - 1>::type>
+                ::Print(TuplePolicy<Tuple>::template get<N - 1>(t), os);
+            }
+
+            // Tersely prints the first N fields of a tuple to a string vector,
+            // one element for each field.
+            template <typename Tuple>
+            static void TersePrintPrefixToStrings(const Tuple &t, Strings *strings)
+            {
+                TuplePrefixPrinter<N - 1>::TersePrintPrefixToStrings(t, strings);
+                ::std::stringstream ss;
+                UniversalTersePrint(TuplePolicy<Tuple>::template get<N - 1>(t), &ss);
+                strings->push_back(ss.str());
+            }
+        };
+
+// Base case.
+        template <>
+        struct TuplePrefixPrinter<0> {
+            template <typename Tuple>
+            static void PrintPrefixTo(const Tuple &, ::std::ostream *) {}
+
+            template <typename Tuple>
+            static void TersePrintPrefixToStrings(const Tuple &, Strings *) {}
+        };
+
+// Helper function for printing a tuple.
+// Tuple must be either std::tr1::tuple or std::tuple type.
+        template <typename Tuple>
+        void PrintTupleTo(const Tuple &t, ::std::ostream *os)
+        {
+            *os << "(";
+            TuplePrefixPrinter<TuplePolicy<Tuple>::tuple_size>::PrintPrefixTo(t, os);
+            *os << ")";
+        }
+
+// Prints the fields of a tuple tersely to a string vector, one
+// element for each field.  See the comment before
+// UniversalTersePrint() for how we define "tersely".
+        template <typename Tuple>
+        Strings UniversalTersePrintTupleFieldsToStrings(const Tuple &value)
+        {
+            Strings result;
+            TuplePrefixPrinter<TuplePolicy<Tuple>::tuple_size>::
+            TersePrintPrefixToStrings(value, &result);
+            return result;
+        }
+#endif  // GTEST_HAS_TR1_TUPLE || GTEST_HAS_STD_TUPLE_
+
+    }  // namespace internal
+
+    template <typename T>
+    ::std::string PrintToString(const T &value)
+    {
+        ::std::stringstream ss;
+        internal::UniversalTersePrinter<T>::Print(value, &ss);
+        return ss.str();
+    }
+
+}  // namespace testing
+
+// Include any custom printer added by the local installation.
+// We must include this header at the end to make sure it can use the
+// declarations from this file.
+// Copyright 2015, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// This file provides an injection point for custom printers in a local
+// installation of gTest.
+// It will be included from gtest-printers.h and the overrides in this file
+// will be visible to everyone.
+// See documentation at gtest/gtest-printers.h for details on how to define a
+// custom printer.
+//
+// ** Custom implementation starts here **
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_CUSTOM_GTEST_PRINTERS_H_
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PRINTERS_H_
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing
+{
+
+// Input to a parameterized test name generator, describing a test parameter.
+// Consists of the parameter value and the integer parameter index.
+    template <class ParamType>
+    struct TestParamInfo {
+        TestParamInfo(const ParamType &a_param, size_t an_index) :
+            param(a_param),
+            index(an_index) {}
+        ParamType param;
+        size_t index;
+    };
+
+// A builtin parameterized test name generator which returns the result of
+// testing::PrintToString.
+    struct PrintToStringParamName {
+        template <class ParamType>
+        std::string operator()(const TestParamInfo<ParamType> &info) const
+        {
+            return PrintToString(info.param);
+        }
+    };
+
+    namespace internal
+    {
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Outputs a message explaining invalid registration of different
+// fixture class for the same test case. This may happen when
+// TEST_P macro is used to define two tests with the same name
+// but in different namespaces.
+        GTEST_API_ void ReportInvalidTestCaseType(const char *test_case_name,
+                                                  CodeLocation code_location);
+
+        template <typename> class ParamGeneratorInterface;
+        template <typename> class ParamGenerator;
+
+// Interface for iterating over elements provided by an implementation
+// of ParamGeneratorInterface<T>.
+        template <typename T>
+        class ParamIteratorInterface
+        {
+        public:
+            virtual ~ParamIteratorInterface() {}
+            // A pointer to the base generator instance.
+            // Used only for the purposes of iterator comparison
+            // to make sure that two iterators belong to the same generator.
+            virtual const ParamGeneratorInterface<T> *BaseGenerator() const = 0;
+            // Advances iterator to point to the next element
+            // provided by the generator. The caller is responsible
+            // for not calling Advance() on an iterator equal to
+            // BaseGenerator()->End().
+            virtual void Advance() = 0;
+            // Clones the iterator object. Used for implementing copy semantics
+            // of ParamIterator<T>.
+            virtual ParamIteratorInterface *Clone() const = 0;
+            // Dereferences the current iterator and provides (read-only) access
+            // to the pointed value. It is the caller's responsibility not to call
+            // Current() on an iterator equal to BaseGenerator()->End().
+            // Used for implementing ParamGenerator<T>::operator*().
+            virtual const T *Current() const = 0;
+            // Determines whether the given iterator and other point to the same
+            // element in the sequence generated by the generator.
+            // Used for implementing ParamGenerator<T>::operator==().
+            virtual bool Equals(const ParamIteratorInterface &other) const = 0;
+        };
+
+// Class iterating over elements provided by an implementation of
+// ParamGeneratorInterface<T>. It wraps ParamIteratorInterface<T>
+// and implements the const forward iterator concept.
+        template <typename T>
+        class ParamIterator
+        {
+        public:
+            typedef T value_type;
+            typedef const T &reference;
+            typedef ptrdiff_t difference_type;
+
+            // ParamIterator assumes ownership of the impl_ pointer.
+            ParamIterator(const ParamIterator &other) : impl_(other.impl_->Clone()) {}
+            ParamIterator &operator=(const ParamIterator &other)
+            {
+                if (this != &other)
+                    impl_.reset(other.impl_->Clone());
+                return *this;
+            }
+
+            const T &operator*() const
+            {
+                return *impl_->Current();
+            }
+            const T *operator->() const
+            {
+                return impl_->Current();
+            }
+            // Prefix version of operator++.
+            ParamIterator &operator++()
+            {
+                impl_->Advance();
+                return *this;
+            }
+            // Postfix version of operator++.
+            ParamIterator operator++(int /*unused*/)
+            {
+                ParamIteratorInterface<T> *clone = impl_->Clone();
+                impl_->Advance();
+                return ParamIterator(clone);
+            }
+            bool operator==(const ParamIterator &other) const
+            {
+                return impl_.get() == other.impl_.get() || impl_->Equals(*other.impl_);
+            }
+            bool operator!=(const ParamIterator &other) const
+            {
+                return !(*this == other);
+            }
+
+        private:
+            friend class ParamGenerator<T>;
+            explicit ParamIterator(ParamIteratorInterface<T> *impl) : impl_(impl) {}
+            scoped_ptr<ParamIteratorInterface<T> > impl_;
+        };
+
+// ParamGeneratorInterface<T> is the binary interface to access generators
+// defined in other translation units.
+        template <typename T>
+        class ParamGeneratorInterface
+        {
+        public:
+            typedef T ParamType;
+
+            virtual ~ParamGeneratorInterface() {}
+
+            // Generator interface definition
+            virtual ParamIteratorInterface<T> *Begin() const = 0;
+            virtual ParamIteratorInterface<T> *End() const = 0;
+        };
+
+// Wraps ParamGeneratorInterface<T> and provides general generator syntax
+// compatible with the STL Container concept.
+// This class implements copy initialization semantics and the contained
+// ParamGeneratorInterface<T> instance is shared among all copies
+// of the original object. This is possible because that instance is immutable.
+        template<typename T>
+        class ParamGenerator
+        {
+        public:
+            typedef ParamIterator<T> iterator;
+
+            explicit ParamGenerator(ParamGeneratorInterface<T> *impl) : impl_(impl) {}
+            ParamGenerator(const ParamGenerator &other) : impl_(other.impl_) {}
+
+            ParamGenerator &operator=(const ParamGenerator &other)
+            {
+                impl_ = other.impl_;
+                return *this;
+            }
+
+            iterator begin() const
+            {
+                return iterator(impl_->Begin());
+            }
+            iterator end() const
+            {
+                return iterator(impl_->End());
+            }
+
+        private:
+            linked_ptr<const ParamGeneratorInterface<T> > impl_;
+        };
+
+// Generates values from a range of two comparable values. Can be used to
+// generate sequences of user-defined types that implement operator+() and
+// operator<().
+// This class is used in the Range() function.
+        template <typename T, typename IncrementT>
+        class RangeGenerator : public ParamGeneratorInterface<T>
+        {
+        public:
+            RangeGenerator(T begin, T end, IncrementT step)
+                : begin_(begin), end_(end),
+                  step_(step), end_index_(CalculateEndIndex(begin, end, step)) {}
+            virtual ~RangeGenerator() {}
+
+            virtual ParamIteratorInterface<T> *Begin() const
+            {
+                return new Iterator(this, begin_, 0, step_);
+            }
+            virtual ParamIteratorInterface<T> *End() const
+            {
+                return new Iterator(this, end_, end_index_, step_);
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<T>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<T> *base, T value, int index,
+                         IncrementT step)
+                    : base_(base), value_(value), index_(index), step_(step) {}
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<T> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                virtual void Advance()
+                {
+                    value_ = static_cast<T>(value_ + step_);
+                    index_++;
+                }
+                virtual ParamIteratorInterface<T> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const T *Current() const
+                {
+                    return &value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<T> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const int other_index =
+                        CheckedDowncastToActualType<const Iterator>(&other)->index_;
+                    return index_ == other_index;
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : ParamIteratorInterface<T>(),
+                      base_(other.base_), value_(other.value_), index_(other.index_),
+                      step_(other.step_) {}
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<T> *const base_;
+                T value_;
+                int index_;
+                const IncrementT step_;
+            };  // class RangeGenerator::Iterator
+
+            static int CalculateEndIndex(const T &begin,
+                                         const T &end,
+                                         const IncrementT &step)
+            {
+                int end_index = 0;
+                for (T i = begin; i < end; i = static_cast<T>(i + step))
+                    end_index++;
+                return end_index;
+            }
+
+            // No implementation - assignment is unsupported.
+            void operator=(const RangeGenerator &other);
+
+            const T begin_;
+            const T end_;
+            const IncrementT step_;
+            // The index for the end() iterator. All the elements in the generated
+            // sequence are indexed (0-based) to aid iterator comparison.
+            const int end_index_;
+        };  // class RangeGenerator
+
+
+// Generates values from a pair of STL-style iterators. Used in the
+// ValuesIn() function. The elements are copied from the source range
+// since the source can be located on the stack, and the generator
+// is likely to persist beyond that stack frame.
+        template <typename T>
+        class ValuesInIteratorRangeGenerator : public ParamGeneratorInterface<T>
+        {
+        public:
+            template <typename ForwardIterator>
+            ValuesInIteratorRangeGenerator(ForwardIterator begin, ForwardIterator end)
+                : container_(begin, end) {}
+            virtual ~ValuesInIteratorRangeGenerator() {}
+
+            virtual ParamIteratorInterface<T> *Begin() const
+            {
+                return new Iterator(this, container_.begin());
+            }
+            virtual ParamIteratorInterface<T> *End() const
+            {
+                return new Iterator(this, container_.end());
+            }
+
+        private:
+            typedef typename ::std::vector<T> ContainerType;
+
+            class Iterator : public ParamIteratorInterface<T>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<T> *base,
+                         typename ContainerType::const_iterator iterator)
+                    : base_(base), iterator_(iterator) {}
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<T> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                virtual void Advance()
+                {
+                    ++iterator_;
+                    value_.reset();
+                }
+                virtual ParamIteratorInterface<T> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                // We need to use cached value referenced by iterator_ because *iterator_
+                // can return a temporary object (and of type other then T), so just
+                // having "return &*iterator_;" doesn't work.
+                // value_ is updated here and not in Advance() because Advance()
+                // can advance iterator_ beyond the end of the range, and we cannot
+                // detect that fact. The client code, on the other hand, is
+                // responsible for not calling Current() on an out-of-range iterator.
+                virtual const T *Current() const
+                {
+                    if (value_.get() == NULL)
+                        value_.reset(new T(*iterator_));
+                    return value_.get();
+                }
+                virtual bool Equals(const ParamIteratorInterface<T> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    return iterator_ ==
+                           CheckedDowncastToActualType<const Iterator>(&other)->iterator_;
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                // The explicit constructor call suppresses a false warning
+                // emitted by gcc when supplied with the -Wextra option.
+                    : ParamIteratorInterface<T>(),
+                      base_(other.base_),
+                      iterator_(other.iterator_) {}
+
+                const ParamGeneratorInterface<T> *const base_;
+                typename ContainerType::const_iterator iterator_;
+                // A cached value of *iterator_. We keep it here to allow access by
+                // pointer in the wrapping iterator's operator->().
+                // value_ needs to be mutable to be accessed in Current().
+                // Use of scoped_ptr helps manage cached value's lifetime,
+                // which is bound by the lifespan of the iterator itself.
+                mutable scoped_ptr<const T> value_;
+            };  // class ValuesInIteratorRangeGenerator::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const ValuesInIteratorRangeGenerator &other);
+
+            const ContainerType container_;
+        };  // class ValuesInIteratorRangeGenerator
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Default parameterized test name generator, returns a string containing the
+// integer test parameter index.
+        template <class ParamType>
+        std::string DefaultParamName(const TestParamInfo<ParamType> &info)
+        {
+            Message name_stream;
+            name_stream << info.index;
+            return name_stream.GetString();
+        }
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Parameterized test name overload helpers, which help the
+// INSTANTIATE_TEST_CASE_P macro choose between the default parameterized
+// test name generator and user param name generator.
+        template <class ParamType, class ParamNameGenFunctor>
+        ParamNameGenFunctor GetParamNameGen(ParamNameGenFunctor func)
+        {
+            return func;
+        }
+
+        template <class ParamType>
+        struct ParamNameGenFunc {
+            typedef std::string Type(const TestParamInfo<ParamType> &);
+        };
+
+        template <class ParamType>
+        typename ParamNameGenFunc<ParamType>::Type *GetParamNameGen()
+        {
+            return DefaultParamName;
+        }
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Stores a parameter value and later creates tests parameterized with that
+// value.
+        template <class TestClass>
+        class ParameterizedTestFactory : public TestFactoryBase
+        {
+        public:
+            typedef typename TestClass::ParamType ParamType;
+            explicit ParameterizedTestFactory(ParamType parameter) :
+                parameter_(parameter) {}
+            virtual Test *CreateTest()
+            {
+                TestClass::SetParam(&parameter_);
+                return new TestClass();
+            }
+
+        private:
+            const ParamType parameter_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestFactory);
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// TestMetaFactoryBase is a base class for meta-factories that create
+// test factories for passing into MakeAndRegisterTestInfo function.
+        template <class ParamType>
+        class TestMetaFactoryBase
+        {
+        public:
+            virtual ~TestMetaFactoryBase() {}
+
+            virtual TestFactoryBase *CreateTestFactory(ParamType parameter) = 0;
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// TestMetaFactory creates test factories for passing into
+// MakeAndRegisterTestInfo function. Since MakeAndRegisterTestInfo receives
+// ownership of test factory pointer, same factory object cannot be passed
+// into that method twice. But ParameterizedTestCaseInfo is going to call
+// it for each Test/Parameter value combination. Thus it needs meta factory
+// creator class.
+        template <class TestCase>
+        class TestMetaFactory
+            : public TestMetaFactoryBase<typename TestCase::ParamType>
+        {
+        public:
+            typedef typename TestCase::ParamType ParamType;
+
+            TestMetaFactory() {}
+
+            virtual TestFactoryBase *CreateTestFactory(ParamType parameter)
+            {
+                return new ParameterizedTestFactory<TestCase>(parameter);
+            }
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(TestMetaFactory);
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseInfoBase is a generic interface
+// to ParameterizedTestCaseInfo classes. ParameterizedTestCaseInfoBase
+// accumulates test information provided by TEST_P macro invocations
+// and generators provided by INSTANTIATE_TEST_CASE_P macro invocations
+// and uses that information to register all resulting test instances
+// in RegisterTests method. The ParameterizeTestCaseRegistry class holds
+// a collection of pointers to the ParameterizedTestCaseInfo objects
+// and calls RegisterTests() on each of them when asked.
+        class ParameterizedTestCaseInfoBase
+        {
+        public:
+            virtual ~ParameterizedTestCaseInfoBase() {}
+
+            // Base part of test case name for display purposes.
+            virtual const std::string &GetTestCaseName() const = 0;
+            // Test case id to verify identity.
+            virtual TypeId GetTestCaseTypeId() const = 0;
+            // UnitTest class invokes this method to register tests in this
+            // test case right before running them in RUN_ALL_TESTS macro.
+            // This method should not be called more then once on any single
+            // instance of a ParameterizedTestCaseInfoBase derived class.
+            virtual void RegisterTests() = 0;
+
+        protected:
+            ParameterizedTestCaseInfoBase() {}
+
+        private:
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfoBase);
+        };
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseInfo accumulates tests obtained from TEST_P
+// macro invocations for a particular test case and generators
+// obtained from INSTANTIATE_TEST_CASE_P macro invocations for that
+// test case. It registers tests with all values generated by all
+// generators when asked.
+        template <class TestCase>
+        class ParameterizedTestCaseInfo : public ParameterizedTestCaseInfoBase
+        {
+        public:
+            // ParamType and GeneratorCreationFunc are private types but are required
+            // for declarations of public methods AddTestPattern() and
+            // AddTestCaseInstantiation().
+            typedef typename TestCase::ParamType ParamType;
+            // A function that returns an instance of appropriate generator type.
+            typedef ParamGenerator<ParamType>(GeneratorCreationFunc)();
+            typedef typename ParamNameGenFunc<ParamType>::Type ParamNameGeneratorFunc;
+
+            explicit ParameterizedTestCaseInfo(
+                const char *name, CodeLocation code_location)
+                : test_case_name_(name), code_location_(code_location) {}
+
+            // Test case base name for display purposes.
+            virtual const std::string &GetTestCaseName() const
+            {
+                return test_case_name_;
+            }
+            // Test case id to verify identity.
+            virtual TypeId GetTestCaseTypeId() const
+            {
+                return GetTypeId<TestCase>();
+            }
+            // TEST_P macro uses AddTestPattern() to record information
+            // about a single test in a LocalTestInfo structure.
+            // test_case_name is the base name of the test case (without invocation
+            // prefix). test_base_name is the name of an individual test without
+            // parameter index. For the test SequenceA/FooTest.DoBar/1 FooTest is
+            // test case base name and DoBar is test base name.
+            void AddTestPattern(const char *test_case_name,
+                                const char *test_base_name,
+                                TestMetaFactoryBase<ParamType> *meta_factory)
+            {
+                tests_.push_back(linked_ptr<TestInfo>(new TestInfo(test_case_name,
+                                                                   test_base_name,
+                                                                   meta_factory)));
+            }
+            // INSTANTIATE_TEST_CASE_P macro uses AddGenerator() to record information
+            // about a generator.
+            int AddTestCaseInstantiation(const std::string &instantiation_name,
+                                         GeneratorCreationFunc *func,
+                                         ParamNameGeneratorFunc *name_func,
+                                         const char *file, int line)
+            {
+                instantiations_.push_back(
+                    InstantiationInfo(instantiation_name, func, name_func, file, line));
+                return 0;  // Return value used only to run this method in namespace scope.
+            }
+            // UnitTest class invokes this method to register tests in this test case
+            // test cases right before running tests in RUN_ALL_TESTS macro.
+            // This method should not be called more then once on any single
+            // instance of a ParameterizedTestCaseInfoBase derived class.
+            // UnitTest has a guard to prevent from calling this method more then once.
+            virtual void RegisterTests()
+            {
+                for (typename TestInfoContainer::iterator test_it = tests_.begin();
+                     test_it != tests_.end(); ++test_it) {
+                    linked_ptr<TestInfo> test_info = *test_it;
+                    for (typename InstantiationContainer::iterator gen_it =
+                             instantiations_.begin(); gen_it != instantiations_.end();
+                         ++gen_it) {
+                        const std::string &instantiation_name = gen_it->name;
+                        ParamGenerator<ParamType> generator((*gen_it->generator)());
+                        ParamNameGeneratorFunc *name_func = gen_it->name_func;
+                        const char *file = gen_it->file;
+                        int line = gen_it->line;
+
+                        std::string test_case_name;
+                        if ( !instantiation_name.empty() )
+                            test_case_name = instantiation_name + "/";
+                        test_case_name += test_info->test_case_base_name;
+
+                        size_t i = 0;
+                        std::set<std::string> test_param_names;
+                        for (typename ParamGenerator<ParamType>::iterator param_it =
+                                 generator.begin();
+                             param_it != generator.end(); ++param_it, ++i) {
+                            Message test_name_stream;
+
+                            std::string param_name = name_func(
+                                                         TestParamInfo<ParamType>(*param_it, i));
+
+                            GTEST_CHECK_(IsValidParamName(param_name))
+                                    << "Parameterized test name '" << param_name
+                                    << "' is invalid, in " << file
+                                    << " line " << line << std::endl;
+
+                            GTEST_CHECK_(test_param_names.count(param_name) == 0)
+                                    << "Duplicate parameterized test name '" << param_name
+                                    << "', in " << file << " line " << line << std::endl;
+
+                            test_param_names.insert(param_name);
+
+                            test_name_stream << test_info->test_base_name << "/" << param_name;
+                            MakeAndRegisterTestInfo(
+                                test_case_name.c_str(),
+                                test_name_stream.GetString().c_str(),
+                                NULL,  // No type parameter.
+                                PrintToString(*param_it).c_str(),
+                                code_location_,
+                                GetTestCaseTypeId(),
+                                TestCase::SetUpTestCase,
+                                TestCase::TearDownTestCase,
+                                test_info->test_meta_factory->CreateTestFactory(*param_it));
+                        }  // for param_it
+                    }  // for gen_it
+                }  // for test_it
+            }  // RegisterTests
+
+        private:
+            // LocalTestInfo structure keeps information about a single test registered
+            // with TEST_P macro.
+            struct TestInfo {
+                TestInfo(const char *a_test_case_base_name,
+                         const char *a_test_base_name,
+                         TestMetaFactoryBase<ParamType> *a_test_meta_factory) :
+                    test_case_base_name(a_test_case_base_name),
+                    test_base_name(a_test_base_name),
+                    test_meta_factory(a_test_meta_factory) {}
+
+                const std::string test_case_base_name;
+                const std::string test_base_name;
+                const scoped_ptr<TestMetaFactoryBase<ParamType> > test_meta_factory;
+            };
+            typedef ::std::vector<linked_ptr<TestInfo> > TestInfoContainer;
+            // Records data received from INSTANTIATE_TEST_CASE_P macros:
+            //  <Instantiation name, Sequence generator creation function,
+            //     Name generator function, Source file, Source line>
+            struct InstantiationInfo {
+                InstantiationInfo(const std::string &name_in,
+                                  GeneratorCreationFunc *generator_in,
+                                  ParamNameGeneratorFunc *name_func_in,
+                                  const char *file_in,
+                                  int line_in)
+                    : name(name_in),
+                      generator(generator_in),
+                      name_func(name_func_in),
+                      file(file_in),
+                      line(line_in) {}
+
+                std::string name;
+                GeneratorCreationFunc *generator;
+                ParamNameGeneratorFunc *name_func;
+                const char *file;
+                int line;
+            };
+            typedef ::std::vector<InstantiationInfo> InstantiationContainer;
+
+            static bool IsValidParamName(const std::string &name)
+            {
+                // Check for empty string
+                if (name.empty())
+                    return false;
+
+                // Check for invalid characters
+                for (std::string::size_type index = 0; index < name.size(); ++index) {
+                    if (!isalnum(name[index]) && name[index] != '_')
+                        return false;
+                }
+
+                return true;
+            }
+
+            const std::string test_case_name_;
+            CodeLocation code_location_;
+            TestInfoContainer tests_;
+            InstantiationContainer instantiations_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseInfo);
+        };  // class ParameterizedTestCaseInfo
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// ParameterizedTestCaseRegistry contains a map of ParameterizedTestCaseInfoBase
+// classes accessed by test case names. TEST_P and INSTANTIATE_TEST_CASE_P
+// macros use it to locate their corresponding ParameterizedTestCaseInfo
+// descriptors.
+        class ParameterizedTestCaseRegistry
+        {
+        public:
+            ParameterizedTestCaseRegistry() {}
+            ~ParameterizedTestCaseRegistry()
+            {
+                for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+                     it != test_case_infos_.end(); ++it) {
+                    delete *it;
+                }
+            }
+
+            // Looks up or creates and returns a structure containing information about
+            // tests and instantiations of a particular test case.
+            template <class TestCase>
+            ParameterizedTestCaseInfo<TestCase> *GetTestCasePatternHolder(
+                const char *test_case_name,
+                CodeLocation code_location)
+            {
+                ParameterizedTestCaseInfo<TestCase> *typed_test_info = NULL;
+                for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+                     it != test_case_infos_.end(); ++it) {
+                    if ((*it)->GetTestCaseName() == test_case_name) {
+                        if ((*it)->GetTestCaseTypeId() != GetTypeId<TestCase>()) {
+                            // Complain about incorrect usage of Google Test facilities
+                            // and terminate the program since we cannot guaranty correct
+                            // test case setup and tear-down in this case.
+                            ReportInvalidTestCaseType(test_case_name, code_location);
+                            posix::Abort();
+                        } else {
+                            // At this point we are sure that the object we found is of the same
+                            // type we are looking for, so we downcast it to that type
+                            // without further checks.
+                            typed_test_info = CheckedDowncastToActualType<
+                                              ParameterizedTestCaseInfo<TestCase> >(*it);
+                        }
+                        break;
+                    }
+                }
+                if (typed_test_info == NULL) {
+                    typed_test_info = new ParameterizedTestCaseInfo<TestCase>(
+                        test_case_name, code_location);
+                    test_case_infos_.push_back(typed_test_info);
+                }
+                return typed_test_info;
+            }
+            void RegisterTests()
+            {
+                for (TestCaseInfoContainer::iterator it = test_case_infos_.begin();
+                     it != test_case_infos_.end(); ++it) {
+                    (*it)->RegisterTests();
+                }
+            }
+
+        private:
+            typedef ::std::vector<ParameterizedTestCaseInfoBase *> TestCaseInfoContainer;
+
+            TestCaseInfoContainer test_case_infos_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(ParameterizedTestCaseRegistry);
+        };
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  //  GTEST_HAS_PARAM_TEST
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_H_
+// This file was GENERATED by command:
+//     pump.py gtest-param-util-generated.h.pump
+// DO NOT EDIT BY HAND!!!
+
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: vladl@google.com (Vlad Losev)
+
+// Type and function utilities for implementing parameterized tests.
+// This file is generated by a SCRIPT.  DO NOT EDIT BY HAND!
+//
+// Currently Google Test supports at most 50 arguments in Values,
+// and at most 10 arguments in Combine. Please contact
+// googletestframework@googlegroups.com if you need more.
+// Please note that the number of arguments to Combine is limited
+// by the maximum arity of the implementation of tuple which is
+// currently set at 10.
+
+#ifndef GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+#define GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+
+// scripts/fuse_gtest.py depends on gtest's own header being #included
+// *unconditionally*.  Therefore these #includes cannot be moved
+// inside #if GTEST_HAS_PARAM_TEST.
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing
+{
+
+// Forward declarations of ValuesIn(), which is implemented in
+// include/gtest/gtest-param-test.h.
+    template <typename ForwardIterator>
+    internal::ParamGenerator<
+    typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+    ValuesIn(ForwardIterator begin, ForwardIterator end);
+
+    template <typename T, size_t N>
+    internal::ParamGenerator<T> ValuesIn(const T (&array)[N]);
+
+    template <class Container>
+    internal::ParamGenerator<typename Container::value_type> ValuesIn(
+        const Container &container);
+
+    namespace internal
+    {
+
+// Used in the Values() function to provide polymorphic capabilities.
+        template <typename T1>
+        class ValueArray1
+        {
+        public:
+            explicit ValueArray1(T1 v1) : v1_(v1) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_)};
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray1 &other);
+
+            const T1 v1_;
+        };
+
+        template <typename T1, typename T2>
+        class ValueArray2
+        {
+        public:
+            ValueArray2(T1 v1, T2 v2) : v1_(v1), v2_(v2) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_)};
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray2 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+        };
+
+        template <typename T1, typename T2, typename T3>
+        class ValueArray3
+        {
+        public:
+            ValueArray3(T1 v1, T2 v2, T3 v3) : v1_(v1), v2_(v2), v3_(v3) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray3 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        class ValueArray4
+        {
+        public:
+            ValueArray4(T1 v1, T2 v2, T3 v3, T4 v4) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray4 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        class ValueArray5
+        {
+        public:
+            ValueArray5(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray5 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        class ValueArray6
+        {
+        public:
+            ValueArray6(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray6 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        class ValueArray7
+        {
+        public:
+            ValueArray7(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray7 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        class ValueArray8
+        {
+        public:
+            ValueArray8(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                        T8 v8) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray8 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        class ValueArray9
+        {
+        public:
+            ValueArray9(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                        T9 v9) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray9 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        class ValueArray10
+        {
+        public:
+            ValueArray10(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray10 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11>
+        class ValueArray11
+        {
+        public:
+            ValueArray11(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray11 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12>
+        class ValueArray12
+        {
+        public:
+            ValueArray12(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray12 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13>
+        class ValueArray13
+        {
+        public:
+            ValueArray13(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray13 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14>
+        class ValueArray14
+        {
+        public:
+            ValueArray14(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray14 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15>
+        class ValueArray15
+        {
+        public:
+            ValueArray15(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray15 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16>
+        class ValueArray16
+        {
+        public:
+            ValueArray16(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray16 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17>
+        class ValueArray17
+        {
+        public:
+            ValueArray17(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+                         T17 v17) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray17 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18>
+        class ValueArray18
+        {
+        public:
+            ValueArray18(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray18 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19>
+        class ValueArray19
+        {
+        public:
+            ValueArray19(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray19 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20>
+        class ValueArray20
+        {
+        public:
+            ValueArray20(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray20 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21>
+        class ValueArray21
+        {
+        public:
+            ValueArray21(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray21 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22>
+        class ValueArray22
+        {
+        public:
+            ValueArray22(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray22 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23>
+        class ValueArray23
+        {
+        public:
+            ValueArray23(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray23 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24>
+        class ValueArray24
+        {
+        public:
+            ValueArray24(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray24 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25>
+        class ValueArray25
+        {
+        public:
+            ValueArray25(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+                         T25 v25) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray25 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26>
+        class ValueArray26
+        {
+        public:
+            ValueArray26(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray26 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27>
+        class ValueArray27
+        {
+        public:
+            ValueArray27(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+                v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+                v26_(v26), v27_(v27) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray27 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28>
+        class ValueArray28
+        {
+        public:
+            ValueArray28(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+                v25_(v25), v26_(v26), v27_(v27), v28_(v28) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray28 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29>
+        class ValueArray29
+        {
+        public:
+            ValueArray29(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+                v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray29 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30>
+        class ValueArray30
+        {
+        public:
+            ValueArray30(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray30 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31>
+        class ValueArray31
+        {
+        public:
+            ValueArray31(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray31 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32>
+        class ValueArray32
+        {
+        public:
+            ValueArray32(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+                v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray32 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33>
+        class ValueArray33
+        {
+        public:
+            ValueArray33(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32,
+                         T33 v33) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray33 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34>
+        class ValueArray34
+        {
+        public:
+            ValueArray34(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray34 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35>
+        class ValueArray35
+        {
+        public:
+            ValueArray35(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+                v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+                v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31),
+                v32_(v32), v33_(v33), v34_(v34), v35_(v35) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray35 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36>
+        class ValueArray36
+        {
+        public:
+            ValueArray36(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+                v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30),
+                v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray36 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37>
+        class ValueArray37
+        {
+        public:
+            ValueArray37(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+                v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29),
+                v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35),
+                v36_(v36), v37_(v37) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray37 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38>
+        class ValueArray38
+        {
+        public:
+            ValueArray38(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray38 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39>
+        class ValueArray39
+        {
+        public:
+            ValueArray39(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray39 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40>
+        class ValueArray40
+        {
+        public:
+            ValueArray40(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+                v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33),
+                v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39),
+                v40_(v40) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray40 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41>
+        class ValueArray41
+        {
+        public:
+            ValueArray41(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40,
+                         T41 v41) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray41 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42>
+        class ValueArray42
+        {
+        public:
+            ValueArray42(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41), v42_(v42) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray42 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43>
+        class ValueArray43
+        {
+        public:
+            ValueArray43(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6),
+                v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13),
+                v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19),
+                v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25),
+                v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31),
+                v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37),
+                v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray43 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44>
+        class ValueArray44
+        {
+        public:
+            ValueArray44(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5),
+                v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12),
+                v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17), v18_(v18),
+                v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23), v24_(v24),
+                v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29), v30_(v30),
+                v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35), v36_(v36),
+                v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41), v42_(v42),
+                v43_(v43), v44_(v44) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray44 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45>
+        class ValueArray45
+        {
+        public:
+            ValueArray45(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45) : v1_(v1), v2_(v2), v3_(v3), v4_(v4),
+                v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10), v11_(v11),
+                v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16), v17_(v17),
+                v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22), v23_(v23),
+                v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28), v29_(v29),
+                v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34), v35_(v35),
+                v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40), v41_(v41),
+                v42_(v42), v43_(v43), v44_(v44), v45_(v45) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray45 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46>
+        class ValueArray46
+        {
+        public:
+            ValueArray46(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46) : v1_(v1), v2_(v2), v3_(v3),
+                v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40),
+                v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray46 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47>
+        class ValueArray47
+        {
+        public:
+            ValueArray47(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47) : v1_(v1), v2_(v2),
+                v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9), v10_(v10),
+                v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15), v16_(v16),
+                v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21), v22_(v22),
+                v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27), v28_(v28),
+                v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33), v34_(v34),
+                v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39), v40_(v40),
+                v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45), v46_(v46),
+                v47_(v47) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray47 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48>
+        class ValueArray48
+        {
+        public:
+            ValueArray48(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48) : v1_(v1),
+                v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7), v8_(v8), v9_(v9),
+                v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14), v15_(v15),
+                v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20), v21_(v21),
+                v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26), v27_(v27),
+                v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32), v33_(v33),
+                v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38), v39_(v39),
+                v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44), v45_(v45),
+                v46_(v46), v47_(v47), v48_(v48) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_),
+                                   static_cast<T>(v48_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray48 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+            const T48 v48_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49>
+        class ValueArray49
+        {
+        public:
+            ValueArray49(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48,
+                         T49 v49) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44),
+                v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_),
+                                   static_cast<T>(v48_), static_cast<T>(v49_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray49 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+            const T48 v48_;
+            const T49 v49_;
+        };
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10,
+                  typename T11, typename T12, typename T13, typename T14, typename T15,
+                  typename T16, typename T17, typename T18, typename T19, typename T20,
+                  typename T21, typename T22, typename T23, typename T24, typename T25,
+                  typename T26, typename T27, typename T28, typename T29, typename T30,
+                  typename T31, typename T32, typename T33, typename T34, typename T35,
+                  typename T36, typename T37, typename T38, typename T39, typename T40,
+                  typename T41, typename T42, typename T43, typename T44, typename T45,
+                  typename T46, typename T47, typename T48, typename T49, typename T50>
+        class ValueArray50
+        {
+        public:
+            ValueArray50(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47, T48 v48, T49 v49,
+                         T50 v50) : v1_(v1), v2_(v2), v3_(v3), v4_(v4), v5_(v5), v6_(v6), v7_(v7),
+                v8_(v8), v9_(v9), v10_(v10), v11_(v11), v12_(v12), v13_(v13), v14_(v14),
+                v15_(v15), v16_(v16), v17_(v17), v18_(v18), v19_(v19), v20_(v20),
+                v21_(v21), v22_(v22), v23_(v23), v24_(v24), v25_(v25), v26_(v26),
+                v27_(v27), v28_(v28), v29_(v29), v30_(v30), v31_(v31), v32_(v32),
+                v33_(v33), v34_(v34), v35_(v35), v36_(v36), v37_(v37), v38_(v38),
+                v39_(v39), v40_(v40), v41_(v41), v42_(v42), v43_(v43), v44_(v44),
+                v45_(v45), v46_(v46), v47_(v47), v48_(v48), v49_(v49), v50_(v50) {}
+
+            template <typename T>
+            operator ParamGenerator<T>() const
+            {
+                const T array[] = {static_cast<T>(v1_), static_cast<T>(v2_),
+                                   static_cast<T>(v3_), static_cast<T>(v4_), static_cast<T>(v5_),
+                                   static_cast<T>(v6_), static_cast<T>(v7_), static_cast<T>(v8_),
+                                   static_cast<T>(v9_), static_cast<T>(v10_), static_cast<T>(v11_),
+                                   static_cast<T>(v12_), static_cast<T>(v13_), static_cast<T>(v14_),
+                                   static_cast<T>(v15_), static_cast<T>(v16_), static_cast<T>(v17_),
+                                   static_cast<T>(v18_), static_cast<T>(v19_), static_cast<T>(v20_),
+                                   static_cast<T>(v21_), static_cast<T>(v22_), static_cast<T>(v23_),
+                                   static_cast<T>(v24_), static_cast<T>(v25_), static_cast<T>(v26_),
+                                   static_cast<T>(v27_), static_cast<T>(v28_), static_cast<T>(v29_),
+                                   static_cast<T>(v30_), static_cast<T>(v31_), static_cast<T>(v32_),
+                                   static_cast<T>(v33_), static_cast<T>(v34_), static_cast<T>(v35_),
+                                   static_cast<T>(v36_), static_cast<T>(v37_), static_cast<T>(v38_),
+                                   static_cast<T>(v39_), static_cast<T>(v40_), static_cast<T>(v41_),
+                                   static_cast<T>(v42_), static_cast<T>(v43_), static_cast<T>(v44_),
+                                   static_cast<T>(v45_), static_cast<T>(v46_), static_cast<T>(v47_),
+                                   static_cast<T>(v48_), static_cast<T>(v49_), static_cast<T>(v50_)
+                                  };
+                return ValuesIn(array);
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const ValueArray50 &other);
+
+            const T1 v1_;
+            const T2 v2_;
+            const T3 v3_;
+            const T4 v4_;
+            const T5 v5_;
+            const T6 v6_;
+            const T7 v7_;
+            const T8 v8_;
+            const T9 v9_;
+            const T10 v10_;
+            const T11 v11_;
+            const T12 v12_;
+            const T13 v13_;
+            const T14 v14_;
+            const T15 v15_;
+            const T16 v16_;
+            const T17 v17_;
+            const T18 v18_;
+            const T19 v19_;
+            const T20 v20_;
+            const T21 v21_;
+            const T22 v22_;
+            const T23 v23_;
+            const T24 v24_;
+            const T25 v25_;
+            const T26 v26_;
+            const T27 v27_;
+            const T28 v28_;
+            const T29 v29_;
+            const T30 v30_;
+            const T31 v31_;
+            const T32 v32_;
+            const T33 v33_;
+            const T34 v34_;
+            const T35 v35_;
+            const T36 v36_;
+            const T37 v37_;
+            const T38 v38_;
+            const T39 v39_;
+            const T40 v40_;
+            const T41 v41_;
+            const T42 v42_;
+            const T43 v43_;
+            const T44 v44_;
+            const T45 v45_;
+            const T46 v46_;
+            const T47 v47_;
+            const T48 v48_;
+            const T49 v49_;
+            const T50 v50_;
+        };
+
+# if GTEST_HAS_COMBINE
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Generates values from the Cartesian product of values produced
+// by the argument generators.
+//
+        template <typename T1, typename T2>
+        class CartesianProductGenerator2
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2> ParamType;
+
+            CartesianProductGenerator2(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2)
+                : g1_(g1), g2_(g2) {}
+            virtual ~CartesianProductGenerator2() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current2_;
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator2::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator2 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+        };  // class CartesianProductGenerator2
+
+
+        template <typename T1, typename T2, typename T3>
+        class CartesianProductGenerator3
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3> ParamType;
+
+            CartesianProductGenerator3(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3)
+                : g1_(g1), g2_(g2), g3_(g3) {}
+            virtual ~CartesianProductGenerator3() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current3_;
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator3::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator3 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+        };  // class CartesianProductGenerator3
+
+
+        template <typename T1, typename T2, typename T3, typename T4>
+        class CartesianProductGenerator4
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4> ParamType;
+
+            CartesianProductGenerator4(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {}
+            virtual ~CartesianProductGenerator4() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current4_;
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator4::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator4 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+        };  // class CartesianProductGenerator4
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5>
+        class CartesianProductGenerator5
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5> ParamType;
+
+            CartesianProductGenerator5(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {}
+            virtual ~CartesianProductGenerator5() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current5_;
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator5::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator5 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+        };  // class CartesianProductGenerator5
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6>
+        class CartesianProductGenerator6
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5,
+              T6> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6> ParamType;
+
+            CartesianProductGenerator6(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {}
+            virtual ~CartesianProductGenerator6() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current6_;
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator6::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator6 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+        };  // class CartesianProductGenerator6
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7>
+        class CartesianProductGenerator7
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7> ParamType;
+
+            CartesianProductGenerator7(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {}
+            virtual ~CartesianProductGenerator7() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current7_;
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator7::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator7 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+        };  // class CartesianProductGenerator7
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8>
+        class CartesianProductGenerator8
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7, T8> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8> ParamType;
+
+            CartesianProductGenerator8(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7,
+                                       const ParamGenerator<T8> &g8)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7),
+                  g8_(g8) {}
+            virtual ~CartesianProductGenerator8() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin(), g8_, g8_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+                                    g8_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7,
+                         const ParamGenerator<T8> &g8,
+                         const typename ParamGenerator<T8>::iterator &current8)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+                      begin8_(g8.begin()), end8_(g8.end()), current8_(current8)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current8_;
+                    if (current8_ == end8_) {
+                        current8_ = begin8_;
+                        ++current7_;
+                    }
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_ &&
+                               current8_ == typed_other->current8_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_),
+                      begin8_(other.begin8_),
+                      end8_(other.end8_),
+                      current8_(other.current8_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_, *current8_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_ ||
+                        current8_ == end8_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                const typename ParamGenerator<T8>::iterator begin8_;
+                const typename ParamGenerator<T8>::iterator end8_;
+                typename ParamGenerator<T8>::iterator current8_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator8::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator8 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+            const ParamGenerator<T8> g8_;
+        };  // class CartesianProductGenerator8
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9>
+        class CartesianProductGenerator9
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7, T8, T9> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9> ParamType;
+
+            CartesianProductGenerator9(const ParamGenerator<T1> &g1,
+                                       const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                       const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                       const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7,
+                                       const ParamGenerator<T8> &g8, const ParamGenerator<T9> &g9)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9) {}
+            virtual ~CartesianProductGenerator9() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+                                    g8_.end(), g9_, g9_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7,
+                         const ParamGenerator<T8> &g8,
+                         const typename ParamGenerator<T8>::iterator &current8,
+                         const ParamGenerator<T9> &g9,
+                         const typename ParamGenerator<T9>::iterator &current9)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+                      begin8_(g8.begin()), end8_(g8.end()), current8_(current8),
+                      begin9_(g9.begin()), end9_(g9.end()), current9_(current9)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current9_;
+                    if (current9_ == end9_) {
+                        current9_ = begin9_;
+                        ++current8_;
+                    }
+                    if (current8_ == end8_) {
+                        current8_ = begin8_;
+                        ++current7_;
+                    }
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_ &&
+                               current8_ == typed_other->current8_ &&
+                               current9_ == typed_other->current9_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_),
+                      begin8_(other.begin8_),
+                      end8_(other.end8_),
+                      current8_(other.current8_),
+                      begin9_(other.begin9_),
+                      end9_(other.end9_),
+                      current9_(other.current9_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_, *current8_,
+                                                   *current9_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_ ||
+                        current8_ == end8_ ||
+                        current9_ == end9_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                const typename ParamGenerator<T8>::iterator begin8_;
+                const typename ParamGenerator<T8>::iterator end8_;
+                typename ParamGenerator<T8>::iterator current8_;
+                const typename ParamGenerator<T9>::iterator begin9_;
+                const typename ParamGenerator<T9>::iterator end9_;
+                typename ParamGenerator<T9>::iterator current9_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator9::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator9 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+            const ParamGenerator<T8> g8_;
+            const ParamGenerator<T9> g9_;
+        };  // class CartesianProductGenerator9
+
+
+        template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                  typename T6, typename T7, typename T8, typename T9, typename T10>
+        class CartesianProductGenerator10
+            : public ParamGeneratorInterface< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+              T7, T8, T9, T10> >
+        {
+        public:
+            typedef ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> ParamType;
+
+            CartesianProductGenerator10(const ParamGenerator<T1> &g1,
+                                        const ParamGenerator<T2> &g2, const ParamGenerator<T3> &g3,
+                                        const ParamGenerator<T4> &g4, const ParamGenerator<T5> &g5,
+                                        const ParamGenerator<T6> &g6, const ParamGenerator<T7> &g7,
+                                        const ParamGenerator<T8> &g8, const ParamGenerator<T9> &g9,
+                                        const ParamGenerator<T10> &g10)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9), g10_(g10) {}
+            virtual ~CartesianProductGenerator10() {}
+
+            virtual ParamIteratorInterface<ParamType> *Begin() const
+            {
+                return new Iterator(this, g1_, g1_.begin(), g2_, g2_.begin(), g3_,
+                                    g3_.begin(), g4_, g4_.begin(), g5_, g5_.begin(), g6_, g6_.begin(), g7_,
+                                    g7_.begin(), g8_, g8_.begin(), g9_, g9_.begin(), g10_, g10_.begin());
+            }
+            virtual ParamIteratorInterface<ParamType> *End() const
+            {
+                return new Iterator(this, g1_, g1_.end(), g2_, g2_.end(), g3_, g3_.end(),
+                                    g4_, g4_.end(), g5_, g5_.end(), g6_, g6_.end(), g7_, g7_.end(), g8_,
+                                    g8_.end(), g9_, g9_.end(), g10_, g10_.end());
+            }
+
+        private:
+            class Iterator : public ParamIteratorInterface<ParamType>
+            {
+            public:
+                Iterator(const ParamGeneratorInterface<ParamType> *base,
+                         const ParamGenerator<T1> &g1,
+                         const typename ParamGenerator<T1>::iterator &current1,
+                         const ParamGenerator<T2> &g2,
+                         const typename ParamGenerator<T2>::iterator &current2,
+                         const ParamGenerator<T3> &g3,
+                         const typename ParamGenerator<T3>::iterator &current3,
+                         const ParamGenerator<T4> &g4,
+                         const typename ParamGenerator<T4>::iterator &current4,
+                         const ParamGenerator<T5> &g5,
+                         const typename ParamGenerator<T5>::iterator &current5,
+                         const ParamGenerator<T6> &g6,
+                         const typename ParamGenerator<T6>::iterator &current6,
+                         const ParamGenerator<T7> &g7,
+                         const typename ParamGenerator<T7>::iterator &current7,
+                         const ParamGenerator<T8> &g8,
+                         const typename ParamGenerator<T8>::iterator &current8,
+                         const ParamGenerator<T9> &g9,
+                         const typename ParamGenerator<T9>::iterator &current9,
+                         const ParamGenerator<T10> &g10,
+                         const typename ParamGenerator<T10>::iterator &current10)
+                    : base_(base),
+                      begin1_(g1.begin()), end1_(g1.end()), current1_(current1),
+                      begin2_(g2.begin()), end2_(g2.end()), current2_(current2),
+                      begin3_(g3.begin()), end3_(g3.end()), current3_(current3),
+                      begin4_(g4.begin()), end4_(g4.end()), current4_(current4),
+                      begin5_(g5.begin()), end5_(g5.end()), current5_(current5),
+                      begin6_(g6.begin()), end6_(g6.end()), current6_(current6),
+                      begin7_(g7.begin()), end7_(g7.end()), current7_(current7),
+                      begin8_(g8.begin()), end8_(g8.end()), current8_(current8),
+                      begin9_(g9.begin()), end9_(g9.end()), current9_(current9),
+                      begin10_(g10.begin()), end10_(g10.end()), current10_(current10)
+                {
+                    ComputeCurrentValue();
+                }
+                virtual ~Iterator() {}
+
+                virtual const ParamGeneratorInterface<ParamType> *BaseGenerator() const
+                {
+                    return base_;
+                }
+                // Advance should not be called on beyond-of-range iterators
+                // so no component iterators must be beyond end of range, either.
+                virtual void Advance()
+                {
+                    assert(!AtEnd());
+                    ++current10_;
+                    if (current10_ == end10_) {
+                        current10_ = begin10_;
+                        ++current9_;
+                    }
+                    if (current9_ == end9_) {
+                        current9_ = begin9_;
+                        ++current8_;
+                    }
+                    if (current8_ == end8_) {
+                        current8_ = begin8_;
+                        ++current7_;
+                    }
+                    if (current7_ == end7_) {
+                        current7_ = begin7_;
+                        ++current6_;
+                    }
+                    if (current6_ == end6_) {
+                        current6_ = begin6_;
+                        ++current5_;
+                    }
+                    if (current5_ == end5_) {
+                        current5_ = begin5_;
+                        ++current4_;
+                    }
+                    if (current4_ == end4_) {
+                        current4_ = begin4_;
+                        ++current3_;
+                    }
+                    if (current3_ == end3_) {
+                        current3_ = begin3_;
+                        ++current2_;
+                    }
+                    if (current2_ == end2_) {
+                        current2_ = begin2_;
+                        ++current1_;
+                    }
+                    ComputeCurrentValue();
+                }
+                virtual ParamIteratorInterface<ParamType> *Clone() const
+                {
+                    return new Iterator(*this);
+                }
+                virtual const ParamType *Current() const
+                {
+                    return &current_value_;
+                }
+                virtual bool Equals(const ParamIteratorInterface<ParamType> &other) const
+                {
+                    // Having the same base generator guarantees that the other
+                    // iterator is of the same type and we can downcast.
+                    GTEST_CHECK_(BaseGenerator() == other.BaseGenerator())
+                            << "The program attempted to compare iterators "
+                            << "from different generators." << std::endl;
+                    const Iterator *typed_other =
+                        CheckedDowncastToActualType<const Iterator>(&other);
+                    // We must report iterators equal if they both point beyond their
+                    // respective ranges. That can happen in a variety of fashions,
+                    // so we have to consult AtEnd().
+                    return (AtEnd() && typed_other->AtEnd()) ||
+                           (
+                               current1_ == typed_other->current1_ &&
+                               current2_ == typed_other->current2_ &&
+                               current3_ == typed_other->current3_ &&
+                               current4_ == typed_other->current4_ &&
+                               current5_ == typed_other->current5_ &&
+                               current6_ == typed_other->current6_ &&
+                               current7_ == typed_other->current7_ &&
+                               current8_ == typed_other->current8_ &&
+                               current9_ == typed_other->current9_ &&
+                               current10_ == typed_other->current10_);
+                }
+
+            private:
+                Iterator(const Iterator &other)
+                    : base_(other.base_),
+                      begin1_(other.begin1_),
+                      end1_(other.end1_),
+                      current1_(other.current1_),
+                      begin2_(other.begin2_),
+                      end2_(other.end2_),
+                      current2_(other.current2_),
+                      begin3_(other.begin3_),
+                      end3_(other.end3_),
+                      current3_(other.current3_),
+                      begin4_(other.begin4_),
+                      end4_(other.end4_),
+                      current4_(other.current4_),
+                      begin5_(other.begin5_),
+                      end5_(other.end5_),
+                      current5_(other.current5_),
+                      begin6_(other.begin6_),
+                      end6_(other.end6_),
+                      current6_(other.current6_),
+                      begin7_(other.begin7_),
+                      end7_(other.end7_),
+                      current7_(other.current7_),
+                      begin8_(other.begin8_),
+                      end8_(other.end8_),
+                      current8_(other.current8_),
+                      begin9_(other.begin9_),
+                      end9_(other.end9_),
+                      current9_(other.current9_),
+                      begin10_(other.begin10_),
+                      end10_(other.end10_),
+                      current10_(other.current10_)
+                {
+                    ComputeCurrentValue();
+                }
+
+                void ComputeCurrentValue()
+                {
+                    if (!AtEnd())
+                        current_value_ = ParamType(*current1_, *current2_, *current3_,
+                                                   *current4_, *current5_, *current6_, *current7_, *current8_,
+                                                   *current9_, *current10_);
+                }
+                bool AtEnd() const
+                {
+                    // We must report iterator past the end of the range when either of the
+                    // component iterators has reached the end of its range.
+                    return
+                        current1_ == end1_ ||
+                        current2_ == end2_ ||
+                        current3_ == end3_ ||
+                        current4_ == end4_ ||
+                        current5_ == end5_ ||
+                        current6_ == end6_ ||
+                        current7_ == end7_ ||
+                        current8_ == end8_ ||
+                        current9_ == end9_ ||
+                        current10_ == end10_;
+                }
+
+                // No implementation - assignment is unsupported.
+                void operator=(const Iterator &other);
+
+                const ParamGeneratorInterface<ParamType> *const base_;
+                // begin[i]_ and end[i]_ define the i-th range that Iterator traverses.
+                // current[i]_ is the actual traversing iterator.
+                const typename ParamGenerator<T1>::iterator begin1_;
+                const typename ParamGenerator<T1>::iterator end1_;
+                typename ParamGenerator<T1>::iterator current1_;
+                const typename ParamGenerator<T2>::iterator begin2_;
+                const typename ParamGenerator<T2>::iterator end2_;
+                typename ParamGenerator<T2>::iterator current2_;
+                const typename ParamGenerator<T3>::iterator begin3_;
+                const typename ParamGenerator<T3>::iterator end3_;
+                typename ParamGenerator<T3>::iterator current3_;
+                const typename ParamGenerator<T4>::iterator begin4_;
+                const typename ParamGenerator<T4>::iterator end4_;
+                typename ParamGenerator<T4>::iterator current4_;
+                const typename ParamGenerator<T5>::iterator begin5_;
+                const typename ParamGenerator<T5>::iterator end5_;
+                typename ParamGenerator<T5>::iterator current5_;
+                const typename ParamGenerator<T6>::iterator begin6_;
+                const typename ParamGenerator<T6>::iterator end6_;
+                typename ParamGenerator<T6>::iterator current6_;
+                const typename ParamGenerator<T7>::iterator begin7_;
+                const typename ParamGenerator<T7>::iterator end7_;
+                typename ParamGenerator<T7>::iterator current7_;
+                const typename ParamGenerator<T8>::iterator begin8_;
+                const typename ParamGenerator<T8>::iterator end8_;
+                typename ParamGenerator<T8>::iterator current8_;
+                const typename ParamGenerator<T9>::iterator begin9_;
+                const typename ParamGenerator<T9>::iterator end9_;
+                typename ParamGenerator<T9>::iterator current9_;
+                const typename ParamGenerator<T10>::iterator begin10_;
+                const typename ParamGenerator<T10>::iterator end10_;
+                typename ParamGenerator<T10>::iterator current10_;
+                ParamType current_value_;
+            };  // class CartesianProductGenerator10::Iterator
+
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductGenerator10 &other);
+
+            const ParamGenerator<T1> g1_;
+            const ParamGenerator<T2> g2_;
+            const ParamGenerator<T3> g3_;
+            const ParamGenerator<T4> g4_;
+            const ParamGenerator<T5> g5_;
+            const ParamGenerator<T6> g6_;
+            const ParamGenerator<T7> g7_;
+            const ParamGenerator<T8> g8_;
+            const ParamGenerator<T9> g9_;
+            const ParamGenerator<T10> g10_;
+        };  // class CartesianProductGenerator10
+
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Helper classes providing Combine() with polymorphic features. They allow
+// casting CartesianProductGeneratorN<T> to ParamGenerator<U> if T is
+// convertible to U.
+//
+        template <class Generator1, class Generator2>
+        class CartesianProductHolder2
+        {
+        public:
+            CartesianProductHolder2(const Generator1 &g1, const Generator2 &g2)
+                : g1_(g1), g2_(g2) {}
+            template <typename T1, typename T2>
+            operator ParamGenerator< ::testing::tuple<T1, T2> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2> >(
+                           new CartesianProductGenerator2<T1, T2>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder2 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+        };  // class CartesianProductHolder2
+
+        template <class Generator1, class Generator2, class Generator3>
+        class CartesianProductHolder3
+        {
+        public:
+            CartesianProductHolder3(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3)
+                : g1_(g1), g2_(g2), g3_(g3) {}
+            template <typename T1, typename T2, typename T3>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3> >(
+                           new CartesianProductGenerator3<T1, T2, T3>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder3 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+        };  // class CartesianProductHolder3
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4>
+        class CartesianProductHolder4
+        {
+        public:
+            CartesianProductHolder4(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4) {}
+            template <typename T1, typename T2, typename T3, typename T4>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4> >(
+                           new CartesianProductGenerator4<T1, T2, T3, T4>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder4 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+        };  // class CartesianProductHolder4
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5>
+        class CartesianProductHolder5
+        {
+        public:
+            CartesianProductHolder5(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5> >(
+                           new CartesianProductGenerator5<T1, T2, T3, T4, T5>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder5 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+        };  // class CartesianProductHolder5
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6>
+        class CartesianProductHolder6
+        {
+        public:
+            CartesianProductHolder6(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6> >(
+                           new CartesianProductGenerator6<T1, T2, T3, T4, T5, T6>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder6 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+        };  // class CartesianProductHolder6
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7>
+        class CartesianProductHolder7
+        {
+        public:
+            CartesianProductHolder7(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6, const Generator7 &g7)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6,
+                     T7> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7> >(
+                           new CartesianProductGenerator7<T1, T2, T3, T4, T5, T6, T7>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder7 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+        };  // class CartesianProductHolder7
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7,
+                  class Generator8>
+        class CartesianProductHolder8
+        {
+        public:
+            CartesianProductHolder8(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6, const Generator7 &g7, const Generator8 &g8)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7),
+                  g8_(g8) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7, typename T8>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7,
+                     T8> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8> >(
+                           new CartesianProductGenerator8<T1, T2, T3, T4, T5, T6, T7, T8>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_),
+                               static_cast<ParamGenerator<T8> >(g8_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder8 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+            const Generator8 g8_;
+        };  // class CartesianProductHolder8
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7,
+                  class Generator8, class Generator9>
+        class CartesianProductHolder9
+        {
+        public:
+            CartesianProductHolder9(const Generator1 &g1, const Generator2 &g2,
+                                    const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                    const Generator6 &g6, const Generator7 &g7, const Generator8 &g8,
+                                    const Generator9 &g9)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7, typename T8, typename T9>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+                     T9> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8,
+                       T9> >(
+                           new CartesianProductGenerator9<T1, T2, T3, T4, T5, T6, T7, T8, T9>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_),
+                               static_cast<ParamGenerator<T8> >(g8_),
+                               static_cast<ParamGenerator<T9> >(g9_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder9 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+            const Generator8 g8_;
+            const Generator9 g9_;
+        };  // class CartesianProductHolder9
+
+        template <class Generator1, class Generator2, class Generator3,
+                  class Generator4, class Generator5, class Generator6, class Generator7,
+                  class Generator8, class Generator9, class Generator10>
+        class CartesianProductHolder10
+        {
+        public:
+            CartesianProductHolder10(const Generator1 &g1, const Generator2 &g2,
+                                     const Generator3 &g3, const Generator4 &g4, const Generator5 &g5,
+                                     const Generator6 &g6, const Generator7 &g7, const Generator8 &g8,
+                                     const Generator9 &g9, const Generator10 &g10)
+                : g1_(g1), g2_(g2), g3_(g3), g4_(g4), g5_(g5), g6_(g6), g7_(g7), g8_(g8),
+                  g9_(g9), g10_(g10) {}
+            template <typename T1, typename T2, typename T3, typename T4, typename T5,
+                      typename T6, typename T7, typename T8, typename T9, typename T10>
+            operator ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9,
+                     T10> >() const
+            {
+                return ParamGenerator< ::testing::tuple<T1, T2, T3, T4, T5, T6, T7, T8, T9,
+                       T10> >(
+                           new CartesianProductGenerator10<T1, T2, T3, T4, T5, T6, T7, T8, T9,
+                           T10>(
+                               static_cast<ParamGenerator<T1> >(g1_),
+                               static_cast<ParamGenerator<T2> >(g2_),
+                               static_cast<ParamGenerator<T3> >(g3_),
+                               static_cast<ParamGenerator<T4> >(g4_),
+                               static_cast<ParamGenerator<T5> >(g5_),
+                               static_cast<ParamGenerator<T6> >(g6_),
+                               static_cast<ParamGenerator<T7> >(g7_),
+                               static_cast<ParamGenerator<T8> >(g8_),
+                               static_cast<ParamGenerator<T9> >(g9_),
+                               static_cast<ParamGenerator<T10> >(g10_)));
+            }
+
+        private:
+            // No implementation - assignment is unsupported.
+            void operator=(const CartesianProductHolder10 &other);
+
+            const Generator1 g1_;
+            const Generator2 g2_;
+            const Generator3 g3_;
+            const Generator4 g4_;
+            const Generator5 g5_;
+            const Generator6 g6_;
+            const Generator7 g7_;
+            const Generator8 g8_;
+            const Generator9 g9_;
+            const Generator10 g10_;
+        };  // class CartesianProductHolder10
+
+# endif  // GTEST_HAS_COMBINE
+
+    }  // namespace internal
+}  // namespace testing
+
+#endif  //  GTEST_HAS_PARAM_TEST
+
+#endif  // GTEST_INCLUDE_GTEST_INTERNAL_GTEST_PARAM_UTIL_GENERATED_H_
+
+#if GTEST_HAS_PARAM_TEST
+
+namespace testing
+{
+
+// Functions producing parameter generators.
+//
+// Google Test uses these generators to produce parameters for value-
+// parameterized tests. When a parameterized test case is instantiated
+// with a particular generator, Google Test creates and runs tests
+// for each element in the sequence produced by the generator.
+//
+// In the following sample, tests from test case FooTest are instantiated
+// each three times with parameter values 3, 5, and 8:
+//
+// class FooTest : public TestWithParam<int> { ... };
+//
+// TEST_P(FooTest, TestThis) {
+// }
+// TEST_P(FooTest, TestThat) {
+// }
+// INSTANTIATE_TEST_CASE_P(TestSequence, FooTest, Values(3, 5, 8));
+//
+
+// Range() returns generators providing sequences of values in a range.
+//
+// Synopsis:
+// Range(start, end)
+//   - returns a generator producing a sequence of values {start, start+1,
+//     start+2, ..., }.
+// Range(start, end, step)
+//   - returns a generator producing a sequence of values {start, start+step,
+//     start+step+step, ..., }.
+// Notes:
+//   * The generated sequences never include end. For example, Range(1, 5)
+//     returns a generator producing a sequence {1, 2, 3, 4}. Range(1, 9, 2)
+//     returns a generator producing {1, 3, 5, 7}.
+//   * start and end must have the same type. That type may be any integral or
+//     floating-point type or a user defined type satisfying these conditions:
+//     * It must be assignable (have operator=() defined).
+//     * It must have operator+() (operator+(int-compatible type) for
+//       two-operand version).
+//     * It must have operator<() defined.
+//     Elements in the resulting sequences will also have that type.
+//   * Condition start < end must be satisfied in order for resulting sequences
+//     to contain any elements.
+//
+    template <typename T, typename IncrementT>
+    internal::ParamGenerator<T> Range(T start, T end, IncrementT step)
+    {
+        return internal::ParamGenerator<T>(
+                   new internal::RangeGenerator<T, IncrementT>(start, end, step));
+    }
+
+    template <typename T>
+    internal::ParamGenerator<T> Range(T start, T end)
+    {
+        return Range(start, end, 1);
+    }
+
+// ValuesIn() function allows generation of tests with parameters coming from
+// a container.
+//
+// Synopsis:
+// ValuesIn(const T (&array)[N])
+//   - returns a generator producing sequences with elements from
+//     a C-style array.
+// ValuesIn(const Container& container)
+//   - returns a generator producing sequences with elements from
+//     an STL-style container.
+// ValuesIn(Iterator begin, Iterator end)
+//   - returns a generator producing sequences with elements from
+//     a range [begin, end) defined by a pair of STL-style iterators. These
+//     iterators can also be plain C pointers.
+//
+// Please note that ValuesIn copies the values from the containers
+// passed in and keeps them to generate tests in RUN_ALL_TESTS().
+//
+// Examples:
+//
+// This instantiates tests from test case StringTest
+// each with C-string values of "foo", "bar", and "baz":
+//
+// const char* strings[] = {"foo", "bar", "baz"};
+// INSTANTIATE_TEST_CASE_P(StringSequence, SrtingTest, ValuesIn(strings));
+//
+// This instantiates tests from test case StlStringTest
+// each with STL strings with values "a" and "b":
+//
+// ::std::vector< ::std::string> GetParameterStrings() {
+//   ::std::vector< ::std::string> v;
+//   v.push_back("a");
+//   v.push_back("b");
+//   return v;
+// }
+//
+// INSTANTIATE_TEST_CASE_P(CharSequence,
+//                         StlStringTest,
+//                         ValuesIn(GetParameterStrings()));
+//
+//
+// This will also instantiate tests from CharTest
+// each with parameter values 'a' and 'b':
+//
+// ::std::list<char> GetParameterChars() {
+//   ::std::list<char> list;
+//   list.push_back('a');
+//   list.push_back('b');
+//   return list;
+// }
+// ::std::list<char> l = GetParameterChars();
+// INSTANTIATE_TEST_CASE_P(CharSequence2,
+//                         CharTest,
+//                         ValuesIn(l.begin(), l.end()));
+//
+    template <typename ForwardIterator>
+    internal::ParamGenerator<
+    typename ::testing::internal::IteratorTraits<ForwardIterator>::value_type>
+    ValuesIn(ForwardIterator begin, ForwardIterator end)
+    {
+        typedef typename ::testing::internal::IteratorTraits<ForwardIterator>
+        ::value_type ParamType;
+        return internal::ParamGenerator<ParamType>(
+                   new internal::ValuesInIteratorRangeGenerator<ParamType>(begin, end));
+    }
+
+    template <typename T, size_t N>
+    internal::ParamGenerator<T> ValuesIn(const T (&array)[N])
+    {
+        return ValuesIn(array, array + N);
+    }
+
+    template <class Container>
+    internal::ParamGenerator<typename Container::value_type> ValuesIn(
+        const Container &container)
+    {
+        return ValuesIn(container.begin(), container.end());
+    }
+
+// Values() allows generating tests from explicitly specified list of
+// parameters.
+//
+// Synopsis:
+// Values(T v1, T v2, ..., T vN)
+//   - returns a generator producing sequences with elements v1, v2, ..., vN.
+//
+// For example, this instantiates tests from test case BarTest each
+// with values "one", "two", and "three":
+//
+// INSTANTIATE_TEST_CASE_P(NumSequence, BarTest, Values("one", "two", "three"));
+//
+// This instantiates tests from test case BazTest each with values 1, 2, 3.5.
+// The exact type of values will depend on the type of parameter in BazTest.
+//
+// INSTANTIATE_TEST_CASE_P(FloatingNumbers, BazTest, Values(1, 2, 3.5));
+//
+// Currently, Values() supports from 1 to 50 parameters.
+//
+    template <typename T1>
+    internal::ValueArray1<T1> Values(T1 v1)
+    {
+        return internal::ValueArray1<T1>(v1);
+    }
+
+    template <typename T1, typename T2>
+    internal::ValueArray2<T1, T2> Values(T1 v1, T2 v2)
+    {
+        return internal::ValueArray2<T1, T2>(v1, v2);
+    }
+
+    template <typename T1, typename T2, typename T3>
+    internal::ValueArray3<T1, T2, T3> Values(T1 v1, T2 v2, T3 v3)
+    {
+        return internal::ValueArray3<T1, T2, T3>(v1, v2, v3);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4>
+    internal::ValueArray4<T1, T2, T3, T4> Values(T1 v1, T2 v2, T3 v3, T4 v4)
+    {
+        return internal::ValueArray4<T1, T2, T3, T4>(v1, v2, v3, v4);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5>
+    internal::ValueArray5<T1, T2, T3, T4, T5> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                     T5 v5)
+    {
+        return internal::ValueArray5<T1, T2, T3, T4, T5>(v1, v2, v3, v4, v5);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6>
+    internal::ValueArray6<T1, T2, T3, T4, T5, T6> Values(T1 v1, T2 v2, T3 v3,
+                                                         T4 v4, T5 v5, T6 v6)
+    {
+        return internal::ValueArray6<T1, T2, T3, T4, T5, T6>(v1, v2, v3, v4, v5, v6);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7>
+    internal::ValueArray7<T1, T2, T3, T4, T5, T6, T7> Values(T1 v1, T2 v2, T3 v3,
+                                                             T4 v4, T5 v5, T6 v6, T7 v7)
+    {
+        return internal::ValueArray7<T1, T2, T3, T4, T5, T6, T7>(v1, v2, v3, v4, v5,
+                                                                 v6, v7);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8>
+    internal::ValueArray8<T1, T2, T3, T4, T5, T6, T7, T8> Values(T1 v1, T2 v2,
+                                                                 T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8)
+    {
+        return internal::ValueArray8<T1, T2, T3, T4, T5, T6, T7, T8>(v1, v2, v3, v4,
+                                                                     v5, v6, v7, v8);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9>
+    internal::ValueArray9<T1, T2, T3, T4, T5, T6, T7, T8, T9> Values(T1 v1, T2 v2,
+                                                                     T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9)
+    {
+        return internal::ValueArray9<T1, T2, T3, T4, T5, T6, T7, T8, T9>(v1, v2, v3,
+                                                                         v4, v5, v6, v7, v8, v9);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10>
+    internal::ValueArray10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10> Values(T1 v1,
+                                                                           T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10)
+    {
+        return internal::ValueArray10<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10>(v1,
+                                                                               v2, v3, v4, v5, v6, v7, v8, v9, v10);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11>
+    internal::ValueArray11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10,
+             T11> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11)
+    {
+        return internal::ValueArray11<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10,
+               T11>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12>
+    internal::ValueArray12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+             T12> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12)
+    {
+        return internal::ValueArray12<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13>
+    internal::ValueArray13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12,
+             T13> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13)
+    {
+        return internal::ValueArray13<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14>
+    internal::ValueArray14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14)
+    {
+        return internal::ValueArray14<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+                              v14);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15>
+    internal::ValueArray15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                              T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15)
+    {
+        return internal::ValueArray15<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+                                   v13, v14, v15);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16>
+    internal::ValueArray16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                   T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                   T16 v16)
+    {
+        return internal::ValueArray16<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+                                        v12, v13, v14, v15, v16);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17>
+    internal::ValueArray17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                        T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                        T16 v16, T17 v17)
+    {
+        return internal::ValueArray17<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+                                             v11, v12, v13, v14, v15, v16, v17);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18>
+    internal::ValueArray18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+                                             T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                             T16 v16, T17 v17, T18 v18)
+    {
+        return internal::ValueArray18<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+                                                  v10, v11, v12, v13, v14, v15, v16, v17, v18);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19>
+    internal::ValueArray19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+                                                  T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+                                                  T15 v15, T16 v16, T17 v17, T18 v18, T19 v19)
+    {
+        return internal::ValueArray19<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19>(v1, v2, v3, v4, v5, v6, v7, v8,
+                                                       v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20>
+    internal::ValueArray20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                       T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                       T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20)
+    {
+        return internal::ValueArray20<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20>(v1, v2, v3, v4, v5, v6, v7,
+                                                            v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21>
+    internal::ValueArray21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                            T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                            T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21)
+    {
+        return internal::ValueArray21<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21>(v1, v2, v3, v4, v5, v6,
+                                                                 v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22>
+    internal::ValueArray22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22> Values(T1 v1, T2 v2, T3 v3,
+                                                                 T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                 T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                 T21 v21, T22 v22)
+    {
+        return internal::ValueArray22<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22>(v1, v2, v3, v4,
+                                                                      v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                      v20, v21, v22);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23>
+    internal::ValueArray23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23> Values(T1 v1, T2 v2,
+                                                                      T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                      T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                      T21 v21, T22 v22, T23 v23)
+    {
+        return internal::ValueArray23<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23>(v1, v2, v3,
+                                                                           v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                           v20, v21, v22, v23);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24>
+    internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24> Values(T1 v1, T2 v2,
+                                                                           T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                           T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                           T21 v21, T22 v22, T23 v23, T24 v24)
+    {
+        return internal::ValueArray24<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24>(v1, v2,
+                                                                                v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
+                                                                                v19, v20, v21, v22, v23, v24);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25>
+    internal::ValueArray25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25> Values(T1 v1,
+                                                                                T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11,
+                                                                                T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19,
+                                                                                T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25)
+    {
+        return internal::ValueArray25<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25>(v1,
+                                                                                     v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17,
+                                                                                     v18, v19, v20, v21, v22, v23, v24, v25);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26>
+    internal::ValueArray26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+             T26> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26)
+    {
+        return internal::ValueArray26<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15,
+                    v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27>
+    internal::ValueArray27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26,
+             T27> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27)
+    {
+        return internal::ValueArray27<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14,
+                         v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28>
+    internal::ValueArray28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27,
+             T28> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28)
+    {
+        return internal::ValueArray28<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+                              v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27,
+                              v28);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29>
+    internal::ValueArray29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29)
+    {
+        return internal::ValueArray29<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+                                   v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26,
+                                   v27, v28, v29);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30>
+    internal::ValueArray30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                              T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+                              T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+                              T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30)
+    {
+        return internal::ValueArray30<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+                                        v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25,
+                                        v26, v27, v28, v29, v30);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31>
+    internal::ValueArray31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                   T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                   T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                   T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31)
+    {
+        return internal::ValueArray31<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+                                             v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24,
+                                             v25, v26, v27, v28, v29, v30, v31);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32>
+    internal::ValueArray32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                        T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                        T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                        T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                        T32 v32)
+    {
+        return internal::ValueArray32<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+                                                  v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                  v24, v25, v26, v27, v28, v29, v30, v31, v32);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33>
+    internal::ValueArray33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+                                             T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                             T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                             T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                             T32 v32, T33 v33)
+    {
+        return internal::ValueArray33<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33>(v1, v2, v3, v4, v5, v6, v7, v8,
+                                                       v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                       v24, v25, v26, v27, v28, v29, v30, v31, v32, v33);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34>
+    internal::ValueArray34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+                                                  T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+                                                  T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22,
+                                                  T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30,
+                                                  T31 v31, T32 v32, T33 v33, T34 v34)
+    {
+        return internal::ValueArray34<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34>(v1, v2, v3, v4, v5, v6, v7,
+                                                            v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22,
+                                                            v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35>
+    internal::ValueArray35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                       T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                       T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+                                                       T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+                                                       T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35)
+    {
+        return internal::ValueArray35<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35>(v1, v2, v3, v4, v5, v6,
+                                                                 v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21,
+                                                                 v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36>
+    internal::ValueArray36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                            T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                            T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+                                                            T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+                                                            T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36)
+    {
+        return internal::ValueArray36<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36>(v1, v2, v3, v4,
+                                                                      v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                      v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+                                                                      v34, v35, v36);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37>
+    internal::ValueArray37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37> Values(T1 v1, T2 v2, T3 v3,
+                                                                 T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                 T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                 T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+                                                                 T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+                                                                 T37 v37)
+    {
+        return internal::ValueArray37<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37>(v1, v2, v3,
+                                                                           v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                           v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+                                                                           v34, v35, v36, v37);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38>
+    internal::ValueArray38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38> Values(T1 v1, T2 v2,
+                                                                      T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                      T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                      T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+                                                                      T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+                                                                      T37 v37, T38 v38)
+    {
+        return internal::ValueArray38<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38>(v1, v2,
+                                                                                v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18,
+                                                                                v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32,
+                                                                                v33, v34, v35, v36, v37, v38);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39>
+    internal::ValueArray39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39> Values(T1 v1, T2 v2,
+                                                                           T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12,
+                                                                           T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20,
+                                                                           T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28,
+                                                                           T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36,
+                                                                           T37 v37, T38 v38, T39 v39)
+    {
+        return internal::ValueArray39<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39>(v1,
+                                                                                     v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17,
+                                                                                     v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31,
+                                                                                     v32, v33, v34, v35, v36, v37, v38, v39);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40>
+    internal::ValueArray40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40> Values(T1 v1,
+                                                                                T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11,
+                                                                                T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19,
+                                                                                T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27,
+                                                                                T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35,
+                                                                                T36 v36, T37 v37, T38 v38, T39 v39, T40 v40)
+    {
+        return internal::ValueArray40<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15,
+                    v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28, v29,
+                    v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41>
+    internal::ValueArray41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40,
+             T41> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41)
+    {
+        return internal::ValueArray41<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13, v14,
+                         v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27, v28,
+                         v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42>
+    internal::ValueArray42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41,
+             T42> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42)
+    {
+        return internal::ValueArray42<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12, v13,
+                              v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26, v27,
+                              v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40, v41,
+                              v42);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43>
+    internal::ValueArray43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42,
+             T43> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43)
+    {
+        return internal::ValueArray43<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11, v12,
+                                   v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25, v26,
+                                   v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39, v40,
+                                   v41, v42, v43);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44>
+    internal::ValueArray44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8, T9 v9,
+                         T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16, T17 v17,
+                         T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24, T25 v25,
+                         T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32, T33 v33,
+                         T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40, T41 v41,
+                         T42 v42, T43 v43, T44 v44)
+    {
+        return internal::ValueArray44<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10, v11,
+                                        v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24, v25,
+                                        v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38, v39,
+                                        v40, v41, v42, v43, v44);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45>
+    internal::ValueArray45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7, T8 v8,
+                              T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15, T16 v16,
+                              T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23, T24 v24,
+                              T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31, T32 v32,
+                              T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39, T40 v40,
+                              T41 v41, T42 v42, T43 v43, T44 v44, T45 v45)
+    {
+        return internal::ValueArray45<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45>(v1, v2, v3, v4, v5, v6, v7, v8, v9, v10,
+                                             v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23, v24,
+                                             v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37, v38,
+                                             v39, v40, v41, v42, v43, v44, v45);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46>
+    internal::ValueArray46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                   T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                   T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                   T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                   T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+                                   T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46)
+    {
+        return internal::ValueArray46<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46>(v1, v2, v3, v4, v5, v6, v7, v8, v9,
+                                                  v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                  v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37,
+                                                  v38, v39, v40, v41, v42, v43, v44, v45, v46);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47>
+    internal::ValueArray47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6, T7 v7,
+                                        T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                        T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                        T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                        T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+                                        T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47)
+    {
+        return internal::ValueArray47<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47>(v1, v2, v3, v4, v5, v6, v7, v8,
+                                                       v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22, v23,
+                                                       v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36, v37,
+                                                       v38, v39, v40, v41, v42, v43, v44, v45, v46, v47);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48>
+    internal::ValueArray48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47, T48> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5, T6 v6,
+                                             T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14, T15 v15,
+                                             T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22, T23 v23,
+                                             T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30, T31 v31,
+                                             T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38, T39 v39,
+                                             T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46, T47 v47,
+                                             T48 v48)
+    {
+        return internal::ValueArray48<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47, T48>(v1, v2, v3, v4, v5, v6, v7,
+                                                            v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21, v22,
+                                                            v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35, v36,
+                                                            v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48, typename T49>
+    internal::ValueArray49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47, T48, T49> Values(T1 v1, T2 v2, T3 v3, T4 v4, T5 v5,
+                                                  T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13, T14 v14,
+                                                  T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21, T22 v22,
+                                                  T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29, T30 v30,
+                                                  T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37, T38 v38,
+                                                  T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45, T46 v46,
+                                                  T47 v47, T48 v48, T49 v49)
+    {
+        return internal::ValueArray49<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47, T48, T49>(v1, v2, v3, v4, v5, v6,
+                                                                 v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19, v20, v21,
+                                                                 v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33, v34, v35,
+                                                                 v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47, v48, v49);
+    }
+
+    template <typename T1, typename T2, typename T3, typename T4, typename T5,
+              typename T6, typename T7, typename T8, typename T9, typename T10,
+              typename T11, typename T12, typename T13, typename T14, typename T15,
+              typename T16, typename T17, typename T18, typename T19, typename T20,
+              typename T21, typename T22, typename T23, typename T24, typename T25,
+              typename T26, typename T27, typename T28, typename T29, typename T30,
+              typename T31, typename T32, typename T33, typename T34, typename T35,
+              typename T36, typename T37, typename T38, typename T39, typename T40,
+              typename T41, typename T42, typename T43, typename T44, typename T45,
+              typename T46, typename T47, typename T48, typename T49, typename T50>
+    internal::ValueArray50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13,
+             T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25, T26, T27, T28,
+             T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39, T40, T41, T42, T43,
+             T44, T45, T46, T47, T48, T49, T50> Values(T1 v1, T2 v2, T3 v3, T4 v4,
+                                                       T5 v5, T6 v6, T7 v7, T8 v8, T9 v9, T10 v10, T11 v11, T12 v12, T13 v13,
+                                                       T14 v14, T15 v15, T16 v16, T17 v17, T18 v18, T19 v19, T20 v20, T21 v21,
+                                                       T22 v22, T23 v23, T24 v24, T25 v25, T26 v26, T27 v27, T28 v28, T29 v29,
+                                                       T30 v30, T31 v31, T32 v32, T33 v33, T34 v34, T35 v35, T36 v36, T37 v37,
+                                                       T38 v38, T39 v39, T40 v40, T41 v41, T42 v42, T43 v43, T44 v44, T45 v45,
+                                                       T46 v46, T47 v47, T48 v48, T49 v49, T50 v50)
+    {
+        return internal::ValueArray50<T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11,
+               T12, T13, T14, T15, T16, T17, T18, T19, T20, T21, T22, T23, T24, T25,
+               T26, T27, T28, T29, T30, T31, T32, T33, T34, T35, T36, T37, T38, T39,
+               T40, T41, T42, T43, T44, T45, T46, T47, T48, T49, T50>(v1, v2, v3, v4,
+                                                                      v5, v6, v7, v8, v9, v10, v11, v12, v13, v14, v15, v16, v17, v18, v19,
+                                                                      v20, v21, v22, v23, v24, v25, v26, v27, v28, v29, v30, v31, v32, v33,
+                                                                      v34, v35, v36, v37, v38, v39, v40, v41, v42, v43, v44, v45, v46, v47,
+                                                                      v48, v49, v50);
+    }
+
+// Bool() allows generating tests with parameters in a set of (false, true).
+//
+// Synopsis:
+// Bool()
+//   - returns a generator producing sequences with elements {false, true}.
+//
+// It is useful when testing code that depends on Boolean flags. Combinations
+// of multiple flags can be tested when several Bool()'s are combined using
+// Combine() function.
+//
+// In the following example all tests in the test case FlagDependentTest
+// will be instantiated twice with parameters false and true.
+//
+// class FlagDependentTest : public testing::TestWithParam<bool> {
+//   virtual void SetUp() {
+//     external_flag = GetParam();
+//   }
+// }
+// INSTANTIATE_TEST_CASE_P(BoolSequence, FlagDependentTest, Bool());
+//
+    inline internal::ParamGenerator<bool> Bool()
+    {
+        return Values(false, true);
+    }
+
+# if GTEST_HAS_COMBINE
+// Combine() allows the user to combine two or more sequences to produce
+// values of a Cartesian product of those sequences' elements.
+//
+// Synopsis:
+// Combine(gen1, gen2, ..., genN)
+//   - returns a generator producing sequences with elements coming from
+//     the Cartesian product of elements from the sequences generated by
+//     gen1, gen2, ..., genN. The sequence elements will have a type of
+//     tuple<T1, T2, ..., TN> where T1, T2, ..., TN are the types
+//     of elements from sequences produces by gen1, gen2, ..., genN.
+//
+// Combine can have up to 10 arguments. This number is currently limited
+// by the maximum number of elements in the tuple implementation used by Google
+// Test.
+//
+// Example:
+//
+// This will instantiate tests in test case AnimalTest each one with
+// the parameter values tuple("cat", BLACK), tuple("cat", WHITE),
+// tuple("dog", BLACK), and tuple("dog", WHITE):
+//
+// enum Color { BLACK, GRAY, WHITE };
+// class AnimalTest
+//     : public testing::TestWithParam<tuple<const char*, Color> > {...};
+//
+// TEST_P(AnimalTest, AnimalLooksNice) {...}
+//
+// INSTANTIATE_TEST_CASE_P(AnimalVariations, AnimalTest,
+//                         Combine(Values("cat", "dog"),
+//                                 Values(BLACK, WHITE)));
+//
+// This will instantiate tests in FlagDependentTest with all variations of two
+// Boolean flags:
+//
+// class FlagDependentTest
+//     : public testing::TestWithParam<tuple<bool, bool> > {
+//   virtual void SetUp() {
+//     // Assigns external_flag_1 and external_flag_2 values from the tuple.
+//     tie(external_flag_1, external_flag_2) = GetParam();
+//   }
+// };
+//
+// TEST_P(FlagDependentTest, TestFeature1) {
+//   // Test your code using external_flag_1 and external_flag_2 here.
+// }
+// INSTANTIATE_TEST_CASE_P(TwoBoolSequence, FlagDependentTest,
+//                         Combine(Bool(), Bool()));
+//
+    template <typename Generator1, typename Generator2>
+    internal::CartesianProductHolder2<Generator1, Generator2> Combine(
+        const Generator1 &g1, const Generator2 &g2)
+    {
+        return internal::CartesianProductHolder2<Generator1, Generator2>(
+                   g1, g2);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3>
+    internal::CartesianProductHolder3<Generator1, Generator2, Generator3> Combine(
+        const Generator1 &g1, const Generator2 &g2, const Generator3 &g3)
+    {
+        return internal::CartesianProductHolder3<Generator1, Generator2, Generator3>(
+                   g1, g2, g3);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4>
+    internal::CartesianProductHolder4<Generator1, Generator2, Generator3,
+             Generator4> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4)
+    {
+        return internal::CartesianProductHolder4<Generator1, Generator2, Generator3,
+               Generator4>(
+                   g1, g2, g3, g4);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5>
+    internal::CartesianProductHolder5<Generator1, Generator2, Generator3,
+             Generator4, Generator5> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5)
+    {
+        return internal::CartesianProductHolder5<Generator1, Generator2, Generator3,
+               Generator4, Generator5>(
+                   g1, g2, g3, g4, g5);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6>
+    internal::CartesianProductHolder6<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6)
+    {
+        return internal::CartesianProductHolder6<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6>(
+                   g1, g2, g3, g4, g5, g6);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7>
+    internal::CartesianProductHolder7<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7)
+    {
+        return internal::CartesianProductHolder7<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7>(
+                   g1, g2, g3, g4, g5, g6, g7);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7, typename Generator8>
+    internal::CartesianProductHolder8<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7, Generator8> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7, const Generator8 &g8)
+    {
+        return internal::CartesianProductHolder8<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7, Generator8>(
+                   g1, g2, g3, g4, g5, g6, g7, g8);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7, typename Generator8, typename Generator9>
+    internal::CartesianProductHolder9<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7, Generator8,
+             Generator9> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7, const Generator8 &g8, const Generator9 &g9)
+    {
+        return internal::CartesianProductHolder9<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7, Generator8, Generator9>(
+                   g1, g2, g3, g4, g5, g6, g7, g8, g9);
+    }
+
+    template <typename Generator1, typename Generator2, typename Generator3,
+              typename Generator4, typename Generator5, typename Generator6,
+              typename Generator7, typename Generator8, typename Generator9,
+              typename Generator10>
+    internal::CartesianProductHolder10<Generator1, Generator2, Generator3,
+             Generator4, Generator5, Generator6, Generator7, Generator8, Generator9,
+             Generator10> Combine(
+                 const Generator1 &g1, const Generator2 &g2, const Generator3 &g3,
+                 const Generator4 &g4, const Generator5 &g5, const Generator6 &g6,
+                 const Generator7 &g7, const Generator8 &g8, const Generator9 &g9,
+                 const Generator10 &g10)
+    {
+        return internal::CartesianProductHolder10<Generator1, Generator2, Generator3,
+               Generator4, Generator5, Generator6, Generator7, Generator8, Generator9,
+               Generator10>(
+                   g1, g2, g3, g4, g5, g6, g7, g8, g9, g10);
+    }
+# endif  // GTEST_HAS_COMBINE
+
+
+
+# define TEST_P(test_case_name, test_name) \
+  class GTEST_TEST_CLASS_NAME_(test_case_name, test_name) \
+      : public test_case_name { \
+   public: \
+    GTEST_TEST_CLASS_NAME_(test_case_name, test_name)() {} \
+    virtual void TestBody(); \
+   private: \
+    static int AddToRegistry() { \
+      ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+          GetTestCasePatternHolder<test_case_name>(\
+              #test_case_name, \
+              ::testing::internal::CodeLocation(\
+                  __FILE__, __LINE__))->AddTestPattern(\
+                      #test_case_name, \
+                      #test_name, \
+                      new ::testing::internal::TestMetaFactory< \
+                          GTEST_TEST_CLASS_NAME_(\
+                              test_case_name, test_name)>()); \
+      return 0; \
+    } \
+    static int gtest_registering_dummy_ GTEST_ATTRIBUTE_UNUSED_; \
+    GTEST_DISALLOW_COPY_AND_ASSIGN_(\
+        GTEST_TEST_CLASS_NAME_(test_case_name, test_name)); \
+  }; \
+  int GTEST_TEST_CLASS_NAME_(test_case_name, \
+                             test_name)::gtest_registering_dummy_ = \
+      GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::AddToRegistry(); \
+  void GTEST_TEST_CLASS_NAME_(test_case_name, test_name)::TestBody()
+
+// The optional last argument to INSTANTIATE_TEST_CASE_P allows the user
+// to specify a function or functor that generates custom test name suffixes
+// based on the test parameters. The function should accept one argument of
+// type testing::TestParamInfo<class ParamType>, and return std::string.
+//
+// testing::PrintToStringParamName is a builtin test suffix generator that
+// returns the value of testing::PrintToString(GetParam()). It does not work
+// for std::string or C strings.
+//
+// Note: test names must be non-empty, unique, and may only contain ASCII
+// alphanumeric characters or underscore.
+
+# define INSTANTIATE_TEST_CASE_P(prefix, test_case_name, generator, ...) \
+  ::testing::internal::ParamGenerator<test_case_name::ParamType> \
+      gtest_##prefix##test_case_name##_EvalGenerator_() { return generator; } \
+  ::std::string gtest_##prefix##test_case_name##_EvalGenerateName_( \
+      const ::testing::TestParamInfo<test_case_name::ParamType>& info) { \
+    return ::testing::internal::GetParamNameGen<test_case_name::ParamType> \
+        (__VA_ARGS__)(info); \
+  } \
+  int gtest_##prefix##test_case_name##_dummy_ GTEST_ATTRIBUTE_UNUSED_ = \
+      ::testing::UnitTest::GetInstance()->parameterized_test_registry(). \
+          GetTestCasePatternHolder<test_case_name>(\
+              #test_case_name, \
+              ::testing::internal::CodeLocation(\
+                  __FILE__, __LINE__))->AddTestCaseInstantiation(\
+                      #prefix, \
+                      &gtest_##prefix##test_case_name##_EvalGenerator_, \
+                      &gtest_##prefix##test_case_name##_EvalGenerateName_, \
+                      __FILE__, __LINE__)
+
+}  // namespace testing
+
+#endif  // GTEST_HAS_PARAM_TEST
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PARAM_TEST_H_
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+//
+// Google C++ Testing Framework definitions useful in production code.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+
+// When you need to test the private or protected members of a class,
+// use the FRIEND_TEST macro to declare your tests as friends of the
+// class.  For example:
+//
+// class MyClass {
+//  private:
+//   void MyMethod();
+//   FRIEND_TEST(MyClassTest, MyMethod);
+// };
+//
+// class MyClassTest : public testing::Test {
+//   // ...
+// };
+//
+// TEST_F(MyClassTest, MyMethod) {
+//   // Can call MyClass::MyMethod() here.
+// }
+
+#define FRIEND_TEST(test_case_name, test_name)\
+friend class test_case_name##_##test_name##_Test
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PROD_H_
+// Copyright 2008, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: mheule@google.com (Markus Heule)
+//
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+#define GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+
+#include <iosfwd>
+#include <vector>
+
+namespace testing
+{
+
+// A copyable object representing the result of a test part (i.e. an
+// assertion or an explicit FAIL(), ADD_FAILURE(), or SUCCESS()).
+//
+// Don't inherit from TestPartResult as its destructor is not virtual.
+    class GTEST_API_ TestPartResult
+    {
+    public:
+        // The possible outcomes of a test part (i.e. an assertion or an
+        // explicit SUCCEED(), FAIL(), or ADD_FAILURE()).
+        enum Type {
+            kSuccess,          // Succeeded.
+            kNonFatalFailure,  // Failed but the test can continue.
+            kFatalFailure,     // Failed and the test should be terminated.
+            kSkip              // Skipped.
+        };
+
+        // C'tor.  TestPartResult does NOT have a default constructor.
+        // Always use this constructor (with parameters) to create a
+        // TestPartResult object.
+        TestPartResult(Type a_type,
+                       const char *a_file_name,
+                       int a_line_number,
+                       const char *a_message)
+            : type_(a_type),
+              file_name_(a_file_name == NULL ? "" : a_file_name),
+              line_number_(a_line_number),
+              summary_(ExtractSummary(a_message)),
+              message_(a_message)
+        {
+        }
+
+        // Gets the outcome of the test part.
+        Type type() const
+        {
+            return type_;
+        }
+
+        // Gets the name of the source file where the test part took place, or
+        // NULL if it's unknown.
+        const char *file_name() const
+        {
+            return file_name_.empty() ? NULL : file_name_.c_str();
+        }
+
+        // Gets the line in the source file where the test part took place,
+        // or -1 if it's unknown.
+        int line_number() const
+        {
+            return line_number_;
+        }
+
+        // Gets the summary of the failure message.
+        const char *summary() const
+        {
+            return summary_.c_str();
+        }
+
+        // Gets the message associated with the test part.
+        const char *message() const
+        {
+            return message_.c_str();
+        }
+
+        // Returns true if and only if the test part was skipped.
+        bool skipped() const
+        {
+            return type_ == kSkip;
+        }
+
+        // Returns true iff the test part passed.
+        bool passed() const
+        {
+            return type_ == kSuccess;
+        }
+
+        // Returns true iff the test part failed.
+        bool failed() const
+        {
+            return fatally_failed() || nonfatally_failed();
+        }
+
+        // Returns true iff the test part non-fatally failed.
+        bool nonfatally_failed() const
+        {
+            return type_ == kNonFatalFailure;
+        }
+
+        // Returns true iff the test part fatally failed.
+        bool fatally_failed() const
+        {
+            return type_ == kFatalFailure;
+        }
+
+    private:
+        Type type_;
+
+        // Gets the summary of the failure message by omitting the stack
+        // trace in it.
+        static std::string ExtractSummary(const char *message);
+
+        // The name of the source file where the test part took place, or
+        // "" if the source file is unknown.
+        std::string file_name_;
+        // The line in the source file where the test part took place, or -1
+        // if the line number is unknown.
+        int line_number_;
+        std::string summary_;  // The test failure summary.
+        std::string message_;  // The test failure message.
+    };
+
+// Prints a TestPartResult object.
+    std::ostream &operator<<(std::ostream &os, const TestPartResult &result);
+
+// An array of TestPartResult objects.
+//
+// Don't inherit from TestPartResultArray as its destructor is not
+// virtual.
+    class GTEST_API_ TestPartResultArray
+    {
+    public:
+        TestPartResultArray() {}
+
+        // Appends the given TestPartResult to the array.
+        void Append(const TestPartResult &result);
+
+        // Returns the TestPartResult at the given index (0-based).
+        const TestPartResult &GetTestPartResult(int index) const;
+
+        // Returns the number of TestPartResult objects in the array.
+        int size() const;
+
+    private:
+        std::vector<TestPartResult> array_;
+
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestPartResultArray);
+    };
+
+// This interface knows how to report a test part result.
+    class TestPartResultReporterInterface
+    {
+    public:
+        virtual ~TestPartResultReporterInterface() {}
+
+        virtual void ReportTestPartResult(const TestPartResult &result) = 0;
+    };
+
+    namespace internal
+    {
+
+// This helper class is used by {ASSERT|EXPECT}_NO_FATAL_FAILURE to check if a
+// statement generates new fatal failures. To do so it registers itself as the
+// current test part result reporter. Besides checking if fatal failures were
+// reported, it only delegates the reporting to the former result reporter.
+// The original result reporter is restored in the destructor.
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        class GTEST_API_ HasNewFatalFailureHelper
+            : public TestPartResultReporterInterface
+        {
+        public:
+            HasNewFatalFailureHelper();
+            virtual ~HasNewFatalFailureHelper();
+            virtual void ReportTestPartResult(const TestPartResult &result);
+            bool has_new_fatal_failure() const
+            {
+                return has_new_fatal_failure_;
+            }
+        private:
+            bool has_new_fatal_failure_;
+            TestPartResultReporterInterface *original_reporter_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(HasNewFatalFailureHelper);
+        };
+
+    }  // namespace internal
+
+}  // namespace testing
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_TEST_PART_H_
+// Copyright 2008 Google Inc.
+// All Rights Reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+//
+// Author: wan@google.com (Zhanyong Wan)
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+#define GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+
+// This header implements typed tests and type-parameterized tests.
+
+// Typed (aka type-driven) tests repeat the same test for types in a
+// list.  You must know which types you want to test with when writing
+// typed tests. Here's how you do it:
+
+#if 0
+
+// First, define a fixture class template.  It should be parameterized
+// by a type.  Remember to derive it from testing::Test.
+template <typename T>
+class FooTest : public testing::Test
+{
+public:
+    ...
+    typedef std::list<T> List;
+    static T shared_;
+    T value_;
+};
+
+// Next, associate a list of types with the test case, which will be
+// repeated for each type in the list.  The typedef is necessary for
+// the macro to parse correctly.
+typedef testing::Types<char, int, unsigned int> MyTypes;
+TYPED_TEST_CASE(FooTest, MyTypes);
+
+// If the type list contains only one type, you can write that type
+// directly without Types<...>:
+//   TYPED_TEST_CASE(FooTest, int);
+
+// Then, use TYPED_TEST() instead of TEST_F() to define as many typed
+// tests for this test case as you want.
+TYPED_TEST(FooTest, DoesBlah)
+{
+    // Inside a test, refer to TypeParam to get the type parameter.
+    // Since we are inside a derived class template, C++ requires use to
+    // visit the members of FooTest via 'this'.
+    TypeParam n = this->value_;
+
+    // To visit static members of the fixture, add the TestFixture::
+    // prefix.
+    n += TestFixture::shared_;
+
+    // To refer to typedefs in the fixture, add the "typename
+    // TestFixture::" prefix.
+    typename TestFixture::List values;
+    values.push_back(n);
+    ...
+}
+
+TYPED_TEST(FooTest, HasPropertyA)
+{
+    ...
+}
+
+#endif  // 0
+
+// Type-parameterized tests are abstract test patterns parameterized
+// by a type.  Compared with typed tests, type-parameterized tests
+// allow you to define the test pattern without knowing what the type
+// parameters are.  The defined pattern can be instantiated with
+// different types any number of times, in any number of translation
+// units.
+//
+// If you are designing an interface or concept, you can define a
+// suite of type-parameterized tests to verify properties that any
+// valid implementation of the interface/concept should have.  Then,
+// each implementation can easily instantiate the test suite to verify
+// that it conforms to the requirements, without having to write
+// similar tests repeatedly.  Here's an example:
+
+#if 0
+
+// First, define a fixture class template.  It should be parameterized
+// by a type.  Remember to derive it from testing::Test.
+template <typename T>
+class FooTest : public testing::Test
+{
+    ...
+};
+
+// Next, declare that you will define a type-parameterized test case
+// (the _P suffix is for "parameterized" or "pattern", whichever you
+// prefer):
+TYPED_TEST_CASE_P(FooTest);
+
+// Then, use TYPED_TEST_P() to define as many type-parameterized tests
+// for this type-parameterized test case as you want.
+TYPED_TEST_P(FooTest, DoesBlah)
+{
+    // Inside a test, refer to TypeParam to get the type parameter.
+    TypeParam n = 0;
+    ...
+}
+
+TYPED_TEST_P(FooTest, HasPropertyA)
+{
+    ...
+}
+
+// Now the tricky part: you need to register all test patterns before
+// you can instantiate them.  The first argument of the macro is the
+// test case name; the rest are the names of the tests in this test
+// case.
+REGISTER_TYPED_TEST_CASE_P(FooTest,
+                           DoesBlah, HasPropertyA);
+
+// Finally, you are free to instantiate the pattern with the types you
+// want.  If you put the above code in a header file, you can #include
+// it in multiple C++ source files and instantiate it multiple times.
+//
+// To distinguish different instances of the pattern, the first
+// argument to the INSTANTIATE_* macro is a prefix that will be added
+// to the actual test case name.  Remember to pick unique prefixes for
+// different instances.
+typedef testing::Types<char, int, unsigned int> MyTypes;
+INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, MyTypes);
+
+// If the type list contains only one type, you can write that type
+// directly without Types<...>:
+//   INSTANTIATE_TYPED_TEST_CASE_P(My, FooTest, int);
+
+#endif  // 0
+
+
+// Implements typed tests.
+
+#if GTEST_HAS_TYPED_TEST
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the name of the typedef for the type parameters of the
+// given test case.
+# define GTEST_TYPE_PARAMS_(TestCaseName) gtest_type_params_##TestCaseName##_
+
+// The 'Types' template argument below must have spaces around it
+// since some compilers may choke on '>>' when passing a template
+// instance (e.g. Types<int>)
+# define TYPED_TEST_CASE(CaseName, Types) \
+  typedef ::testing::internal::TypeList< Types >::type \
+      GTEST_TYPE_PARAMS_(CaseName)
+
+# define TYPED_TEST(CaseName, TestName) \
+  template <typename gtest_TypeParam_> \
+  class GTEST_TEST_CLASS_NAME_(CaseName, TestName) \
+      : public CaseName<gtest_TypeParam_> { \
+   private: \
+    typedef CaseName<gtest_TypeParam_> TestFixture; \
+    typedef gtest_TypeParam_ TypeParam; \
+    virtual void TestBody(); \
+  }; \
+  bool gtest_##CaseName##_##TestName##_registered_ GTEST_ATTRIBUTE_UNUSED_ = \
+      ::testing::internal::TypeParameterizedTest< \
+          CaseName, \
+          ::testing::internal::TemplateSel< \
+              GTEST_TEST_CLASS_NAME_(CaseName, TestName)>, \
+          GTEST_TYPE_PARAMS_(CaseName)>::Register(\
+              "", ::testing::internal::CodeLocation(__FILE__, __LINE__), \
+              #CaseName, #TestName, 0); \
+  template <typename gtest_TypeParam_> \
+  void GTEST_TEST_CLASS_NAME_(CaseName, TestName)<gtest_TypeParam_>::TestBody()
+
+#endif  // GTEST_HAS_TYPED_TEST
+
+// Implements type-parameterized tests.
+
+#if GTEST_HAS_TYPED_TEST_P
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the namespace name that the type-parameterized tests for
+// the given type-parameterized test case are defined in.  The exact
+// name of the namespace is subject to change without notice.
+# define GTEST_CASE_NAMESPACE_(TestCaseName) \
+  gtest_case_##TestCaseName##_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+//
+// Expands to the name of the variable used to remember the names of
+// the defined tests in the given test case.
+# define GTEST_TYPED_TEST_CASE_P_STATE_(TestCaseName) \
+  gtest_typed_test_case_p_state_##TestCaseName##_
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE DIRECTLY.
+//
+// Expands to the name of the variable used to remember the names of
+// the registered tests in the given test case.
+# define GTEST_REGISTERED_TEST_NAMES_(TestCaseName) \
+  gtest_registered_test_names_##TestCaseName##_
+
+// The variables defined in the type-parameterized test macros are
+// static as typically these macros are used in a .h file that can be
+// #included in multiple translation units linked together.
+# define TYPED_TEST_CASE_P(CaseName) \
+  static ::testing::internal::TypedTestCasePState \
+      GTEST_TYPED_TEST_CASE_P_STATE_(CaseName)
+
+# define TYPED_TEST_P(CaseName, TestName) \
+  namespace GTEST_CASE_NAMESPACE_(CaseName) { \
+  template <typename gtest_TypeParam_> \
+  class TestName : public CaseName<gtest_TypeParam_> { \
+   private: \
+    typedef CaseName<gtest_TypeParam_> TestFixture; \
+    typedef gtest_TypeParam_ TypeParam; \
+    virtual void TestBody(); \
+  }; \
+  static bool gtest_##TestName##_defined_ GTEST_ATTRIBUTE_UNUSED_ = \
+      GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).AddTestName(\
+          __FILE__, __LINE__, #CaseName, #TestName); \
+  } \
+  template <typename gtest_TypeParam_> \
+  void GTEST_CASE_NAMESPACE_(CaseName)::TestName<gtest_TypeParam_>::TestBody()
+
+# define REGISTER_TYPED_TEST_CASE_P(CaseName, ...) \
+  namespace GTEST_CASE_NAMESPACE_(CaseName) { \
+  typedef ::testing::internal::Templates<__VA_ARGS__>::type gtest_AllTests_; \
+  } \
+  static const char* const GTEST_REGISTERED_TEST_NAMES_(CaseName) \
+      GTEST_ATTRIBUTE_UNUSED_ = \
+          GTEST_TYPED_TEST_CASE_P_STATE_(CaseName).VerifyRegisteredTestNames(\
+              __FILE__, __LINE__, #__VA_ARGS__)
+
+// The 'Types' template argument below must have spaces around it
+// since some compilers may choke on '>>' when passing a template
+// instance (e.g. Types<int>)
+# define INSTANTIATE_TYPED_TEST_CASE_P(Prefix, CaseName, Types) \
+  bool gtest_##Prefix##_##CaseName GTEST_ATTRIBUTE_UNUSED_ = \
+      ::testing::internal::TypeParameterizedTestCase<CaseName, \
+          GTEST_CASE_NAMESPACE_(CaseName)::gtest_AllTests_, \
+          ::testing::internal::TypeList< Types >::type>::Register(\
+              #Prefix, \
+              ::testing::internal::CodeLocation(__FILE__, __LINE__), \
+              &GTEST_TYPED_TEST_CASE_P_STATE_(CaseName), \
+              #CaseName, GTEST_REGISTERED_TEST_NAMES_(CaseName))
+
+#endif  // GTEST_HAS_TYPED_TEST_P
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_TYPED_TEST_H_
+
+// Depending on the platform, different string classes are available.
+// On Linux, in addition to ::std::string, Google also makes use of
+// class ::string, which has the same interface as ::std::string, but
+// has a different implementation.
+//
+// You can define GTEST_HAS_GLOBAL_STRING to 1 to indicate that
+// ::string is available AND is a distinct type to ::std::string, or
+// define it to 0 to indicate otherwise.
+//
+// If ::std::string and ::string are the same class on your platform
+// due to aliasing, you should define GTEST_HAS_GLOBAL_STRING to 0.
+//
+// If you do not define GTEST_HAS_GLOBAL_STRING, it is defined
+// heuristically.
+
+namespace testing
+{
+
+// Declares the flags.
+
+// This flag temporary enables the disabled tests.
+    GTEST_DECLARE_bool_(also_run_disabled_tests);
+
+// This flag brings the debugger on an assertion failure.
+    GTEST_DECLARE_bool_(break_on_failure);
+
+// This flag controls whether Google Test catches all test-thrown exceptions
+// and logs them as failures.
+    GTEST_DECLARE_bool_(catch_exceptions);
+
+// This flag enables using colors in terminal output. Available values are
+// "yes" to enable colors, "no" (disable colors), or "auto" (the default)
+// to let Google Test decide.
+    GTEST_DECLARE_string_(color);
+
+// This flag sets up the filter to select by name using a glob pattern
+// the tests to run. If the filter is not given all tests are executed.
+    GTEST_DECLARE_string_(filter);
+
+// This flag causes the Google Test to list tests. None of the tests listed
+// are actually run if the flag is provided.
+    GTEST_DECLARE_bool_(list_tests);
+
+// This flag controls whether Google Test emits a detailed XML report to a file
+// in addition to its normal textual output.
+    GTEST_DECLARE_string_(output);
+
+// This flags control whether Google Test prints the elapsed time for each
+// test.
+    GTEST_DECLARE_bool_(print_time);
+
+// This flag specifies the random number seed.
+    GTEST_DECLARE_int32_(random_seed);
+
+// This flag sets how many times the tests are repeated. The default value
+// is 1. If the value is -1 the tests are repeating forever.
+    GTEST_DECLARE_int32_(repeat);
+
+// This flag controls whether Google Test includes Google Test internal
+// stack frames in failure stack traces.
+    GTEST_DECLARE_bool_(show_internal_stack_frames);
+
+// When this flag is specified, tests' order is randomized on every iteration.
+    GTEST_DECLARE_bool_(shuffle);
+
+// This flag specifies the maximum number of stack frames to be
+// printed in a failure message.
+    GTEST_DECLARE_int32_(stack_trace_depth);
+
+// When this flag is specified, a failed assertion will throw an
+// exception if exceptions are enabled, or exit the program with a
+// non-zero code otherwise.
+    GTEST_DECLARE_bool_(throw_on_failure);
+
+// When this flag is set with a "host:port" string, on supported
+// platforms test results are streamed to the specified port on
+// the specified host machine.
+    GTEST_DECLARE_string_(stream_result_to);
+
+// The upper limit for valid stack trace depths.
+    const int kMaxStackTraceDepth = 100;
+
+    namespace internal
+    {
+
+        class AssertHelper;
+        class DefaultGlobalTestPartResultReporter;
+        class ExecDeathTest;
+        class NoExecDeathTest;
+        class FinalSuccessChecker;
+        class GTestFlagSaver;
+        class StreamingListenerTest;
+        class TestResultAccessor;
+        class TestEventListenersAccessor;
+        class TestEventRepeater;
+        class UnitTestRecordPropertyTestHelper;
+        class WindowsDeathTest;
+        class UnitTestImpl *GetUnitTestImpl();
+        void ReportFailureInUnknownLocation(TestPartResult::Type result_type,
+                                            const std::string &message);
+
+    }  // namespace internal
+
+// The friend relationship of some of these classes is cyclic.
+// If we don't forward declare them the compiler might confuse the classes
+// in friendship clauses with same named classes on the scope.
+    class Test;
+    class TestCase;
+    class TestInfo;
+    class UnitTest;
+
+// A class for indicating whether an assertion was successful.  When
+// the assertion wasn't successful, the AssertionResult object
+// remembers a non-empty message that describes how it failed.
+//
+// To create an instance of this class, use one of the factory functions
+// (AssertionSuccess() and AssertionFailure()).
+//
+// This class is useful for two purposes:
+//   1. Defining predicate functions to be used with Boolean test assertions
+//      EXPECT_TRUE/EXPECT_FALSE and their ASSERT_ counterparts
+//   2. Defining predicate-format functions to be
+//      used with predicate assertions (ASSERT_PRED_FORMAT*, etc).
+//
+// For example, if you define IsEven predicate:
+//
+//   testing::AssertionResult IsEven(int n) {
+//     if ((n % 2) == 0)
+//       return testing::AssertionSuccess();
+//     else
+//       return testing::AssertionFailure() << n << " is odd";
+//   }
+//
+// Then the failed expectation EXPECT_TRUE(IsEven(Fib(5)))
+// will print the message
+//
+//   Value of: IsEven(Fib(5))
+//     Actual: false (5 is odd)
+//   Expected: true
+//
+// instead of a more opaque
+//
+//   Value of: IsEven(Fib(5))
+//     Actual: false
+//   Expected: true
+//
+// in case IsEven is a simple Boolean predicate.
+//
+// If you expect your predicate to be reused and want to support informative
+// messages in EXPECT_FALSE and ASSERT_FALSE (negative assertions show up
+// about half as often as positive ones in our tests), supply messages for
+// both success and failure cases:
+//
+//   testing::AssertionResult IsEven(int n) {
+//     if ((n % 2) == 0)
+//       return testing::AssertionSuccess() << n << " is even";
+//     else
+//       return testing::AssertionFailure() << n << " is odd";
+//   }
+//
+// Then a statement EXPECT_FALSE(IsEven(Fib(6))) will print
+//
+//   Value of: IsEven(Fib(6))
+//     Actual: true (8 is even)
+//   Expected: false
+//
+// NB: Predicates that support negative Boolean assertions have reduced
+// performance in positive ones so be careful not to use them in tests
+// that have lots (tens of thousands) of positive Boolean assertions.
+//
+// To use this class with EXPECT_PRED_FORMAT assertions such as:
+//
+//   // Verifies that Foo() returns an even number.
+//   EXPECT_PRED_FORMAT1(IsEven, Foo());
+//
+// you need to define:
+//
+//   testing::AssertionResult IsEven(const char* expr, int n) {
+//     if ((n % 2) == 0)
+//       return testing::AssertionSuccess();
+//     else
+//       return testing::AssertionFailure()
+//         << "Expected: " << expr << " is even\n  Actual: it's " << n;
+//   }
+//
+// If Foo() returns 5, you will see the following message:
+//
+//   Expected: Foo() is even
+//     Actual: it's 5
+//
+    class GTEST_API_ AssertionResult
+    {
+    public:
+        // Copy constructor.
+        // Used in EXPECT_TRUE/FALSE(assertion_result).
+        AssertionResult(const AssertionResult &other);
+
+        GTEST_DISABLE_MSC_WARNINGS_PUSH_(4800 /* forcing value to bool */)
+
+        // Used in the EXPECT_TRUE/FALSE(bool_expression).
+        //
+        // T must be contextually convertible to bool.
+        //
+        // The second parameter prevents this overload from being considered if
+        // the argument is implicitly convertible to AssertionResult. In that case
+        // we want AssertionResult's copy constructor to be used.
+        template <typename T>
+        explicit AssertionResult(
+            const T &success,
+            typename internal::EnableIf<
+            !internal::ImplicitlyConvertible<T, AssertionResult>::value>::type *
+            /*enabler*/ = NULL)
+            : success_(success) {}
+
+        GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+        // Assignment operator.
+        AssertionResult &operator=(AssertionResult other)
+        {
+            swap(other);
+            return *this;
+        }
+
+        // Returns true iff the assertion succeeded.
+        operator bool() const
+        {
+            return success_;    // NOLINT
+        }
+
+        // Returns the assertion's negation. Used with EXPECT/ASSERT_FALSE.
+        AssertionResult operator!() const;
+
+        // Returns the text streamed into this AssertionResult. Test assertions
+        // use it when they fail (i.e., the predicate's outcome doesn't match the
+        // assertion's expectation). When nothing has been streamed into the
+        // object, returns an empty string.
+        const char *message() const
+        {
+            return message_.get() != NULL ?  message_->c_str() : "";
+        }
+        // TODO(vladl@google.com): Remove this after making sure no clients use it.
+        // Deprecated; please use message() instead.
+        const char *failure_message() const
+        {
+            return message();
+        }
+
+        // Streams a custom failure message into this object.
+        template <typename T> AssertionResult &operator<<(const T &value)
+        {
+            AppendMessage(Message() << value);
+            return *this;
+        }
+
+        // Allows streaming basic output manipulators such as endl or flush into
+        // this object.
+        AssertionResult &operator<<(
+            ::std::ostream& (*basic_manipulator)(::std::ostream &stream))
+        {
+            AppendMessage(Message() << basic_manipulator);
+            return *this;
+        }
+
+    private:
+        // Appends the contents of message to message_.
+        void AppendMessage(const Message &a_message)
+        {
+            if (message_.get() == NULL)
+                message_.reset(new ::std::string);
+            message_->append(a_message.GetString().c_str());
+        }
+
+        // Swap the contents of this AssertionResult with other.
+        void swap(AssertionResult &other);
+
+        // Stores result of the assertion predicate.
+        bool success_;
+        // Stores the message describing the condition in case the expectation
+        // construct is not satisfied with the predicate's outcome.
+        // Referenced via a pointer to avoid taking too much stack frame space
+        // with test assertions.
+        internal::scoped_ptr< ::std::string> message_;
+    };
+
+// Makes a successful assertion result.
+    GTEST_API_ AssertionResult AssertionSuccess();
+
+// Makes a failed assertion result.
+    GTEST_API_ AssertionResult AssertionFailure();
+
+// Makes a failed assertion result with the given failure message.
+// Deprecated; use AssertionFailure() << msg.
+    GTEST_API_ AssertionResult AssertionFailure(const Message &msg);
+
+// The abstract class that all tests inherit from.
+//
+// In Google Test, a unit test program contains one or many TestCases, and
+// each TestCase contains one or many Tests.
+//
+// When you define a test using the TEST macro, you don't need to
+// explicitly derive from Test - the TEST macro automatically does
+// this for you.
+//
+// The only time you derive from Test is when defining a test fixture
+// to be used a TEST_F.  For example:
+//
+//   class FooTest : public testing::Test {
+//    protected:
+//     void SetUp() override { ... }
+//     void TearDown() override { ... }
+//     ...
+//   };
+//
+//   TEST_F(FooTest, Bar) { ... }
+//   TEST_F(FooTest, Baz) { ... }
+//
+// Test is not copyable.
+    class GTEST_API_ Test
+    {
+    public:
+        friend class TestInfo;
+
+        // Defines types for pointers to functions that set up and tear down
+        // a test case.
+        typedef internal::SetUpTestCaseFunc SetUpTestCaseFunc;
+        typedef internal::TearDownTestCaseFunc TearDownTestCaseFunc;
+
+        // The d'tor is virtual as we intend to inherit from Test.
+        virtual ~Test();
+
+        // Sets up the stuff shared by all tests in this test case.
+        //
+        // Google Test will call Foo::SetUpTestCase() before running the first
+        // test in test case Foo.  Hence a sub-class can define its own
+        // SetUpTestCase() method to shadow the one defined in the super
+        // class.
+        static void SetUpTestCase() {}
+
+        // Tears down the stuff shared by all tests in this test case.
+        //
+        // Google Test will call Foo::TearDownTestCase() after running the last
+        // test in test case Foo.  Hence a sub-class can define its own
+        // TearDownTestCase() method to shadow the one defined in the super
+        // class.
+        static void TearDownTestCase() {}
+
+        // Returns true iff the current test has a fatal failure.
+        static bool HasFatalFailure();
+
+        // Returns true iff the current test has a non-fatal failure.
+        static bool HasNonfatalFailure();
+
+        // Returns true if and only if the current test was skipped.
+        static bool IsSkipped();
+
+        // Returns true iff the current test has a (either fatal or
+        // non-fatal) failure.
+        static bool HasFailure()
+        {
+            return HasFatalFailure() || HasNonfatalFailure();
+        }
+
+        // Logs a property for the current test, test case, or for the entire
+        // invocation of the test program when used outside of the context of a
+        // test case.  Only the last value for a given key is remembered.  These
+        // are public static so they can be called from utility functions that are
+        // not members of the test fixture.  Calls to RecordProperty made during
+        // lifespan of the test (from the moment its constructor starts to the
+        // moment its destructor finishes) will be output in XML as attributes of
+        // the <testcase> element.  Properties recorded from fixture's
+        // SetUpTestCase or TearDownTestCase are logged as attributes of the
+        // corresponding <testsuite> element.  Calls to RecordProperty made in the
+        // global context (before or after invocation of RUN_ALL_TESTS and from
+        // SetUp/TearDown method of Environment objects registered with Google
+        // Test) will be output as attributes of the <testsuites> element.
+        static void RecordProperty(const std::string &key, const std::string &value);
+        static void RecordProperty(const std::string &key, int value);
+
+    protected:
+        // Creates a Test object.
+        Test();
+
+        // Sets up the test fixture.
+        virtual void SetUp();
+
+        // Tears down the test fixture.
+        virtual void TearDown();
+
+    private:
+        // Returns true iff the current test has the same fixture class as
+        // the first test in the current test case.
+        static bool HasSameFixtureClass();
+
+        // Runs the test after the test fixture has been set up.
+        //
+        // A sub-class must implement this to define the test logic.
+        //
+        // DO NOT OVERRIDE THIS FUNCTION DIRECTLY IN A USER PROGRAM.
+        // Instead, use the TEST or TEST_F macro.
+        virtual void TestBody() = 0;
+
+        // Sets up, executes, and tears down the test.
+        void Run();
+
+        // Deletes self.  We deliberately pick an unusual name for this
+        // internal method to avoid clashing with names used in user TESTs.
+        void DeleteSelf_()
+        {
+            delete this;
+        }
+
+        const internal::scoped_ptr< GTEST_FLAG_SAVER_ > gtest_flag_saver_;
+
+        // Often a user misspells SetUp() as Setup() and spends a long time
+        // wondering why it is never called by Google Test.  The declaration of
+        // the following method is solely for catching such an error at
+        // compile time:
+        //
+        //   - The return type is deliberately chosen to be not void, so it
+        //   will be a conflict if void Setup() is declared in the user's
+        //   test fixture.
+        //
+        //   - This method is private, so it will be another compiler error
+        //   if the method is called from the user's test fixture.
+        //
+        // DO NOT OVERRIDE THIS FUNCTION.
+        //
+        // If you see an error about overriding the following function or
+        // about it being private, you have mis-spelled SetUp() as Setup().
+        struct Setup_should_be_spelled_SetUp {};
+        virtual Setup_should_be_spelled_SetUp *Setup()
+        {
+            return NULL;
+        }
+
+        // We disallow copying Tests.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(Test);
+    };
+
+    typedef internal::TimeInMillis TimeInMillis;
+
+// A copyable object representing a user specified test property which can be
+// output as a key/value string pair.
+//
+// Don't inherit from TestProperty as its destructor is not virtual.
+    class TestProperty
+    {
+    public:
+        // C'tor.  TestProperty does NOT have a default constructor.
+        // Always use this constructor (with parameters) to create a
+        // TestProperty object.
+        TestProperty(const std::string &a_key, const std::string &a_value) :
+            key_(a_key), value_(a_value)
+        {
+        }
+
+        // Gets the user supplied key.
+        const char *key() const
+        {
+            return key_.c_str();
+        }
+
+        // Gets the user supplied value.
+        const char *value() const
+        {
+            return value_.c_str();
+        }
+
+        // Sets a new value, overriding the one supplied in the constructor.
+        void SetValue(const std::string &new_value)
+        {
+            value_ = new_value;
+        }
+
+    private:
+        // The key supplied by the user.
+        std::string key_;
+        // The value supplied by the user.
+        std::string value_;
+    };
+
+// The result of a single Test.  This includes a list of
+// TestPartResults, a list of TestProperties, a count of how many
+// death tests there are in the Test, and how much time it took to run
+// the Test.
+//
+// TestResult is not copyable.
+    class GTEST_API_ TestResult
+    {
+    public:
+        // Creates an empty TestResult.
+        TestResult();
+
+        // D'tor.  Do not inherit from TestResult.
+        ~TestResult();
+
+        // Gets the number of all test parts.  This is the sum of the number
+        // of successful test parts and the number of failed test parts.
+        int total_part_count() const;
+
+        // Returns the number of the test properties.
+        int test_property_count() const;
+
+        // Returns true if the test passed (i.e. no test part failed).
+        bool Passed() const
+        {
+            return !Skipped() && !Failed();
+        }
+
+        // Returns true if and only if the test was skipped.
+        bool Skipped() const;
+
+        // Returns true iff the test failed.
+        bool Failed() const;
+
+        // Returns true iff the test fatally failed.
+        bool HasFatalFailure() const;
+
+        // Returns true iff the test has a non-fatal failure.
+        bool HasNonfatalFailure() const;
+
+        // Returns true if and only if the current test was skipped.
+        static bool IsSkipped();
+
+        // Returns the elapsed time, in milliseconds.
+        TimeInMillis elapsed_time() const
+        {
+            return elapsed_time_;
+        }
+
+        // Returns the i-th test part result among all the results. i can range
+        // from 0 to test_property_count() - 1. If i is not in that range, aborts
+        // the program.
+        const TestPartResult &GetTestPartResult(int i) const;
+
+        // Returns the i-th test property. i can range from 0 to
+        // test_property_count() - 1. If i is not in that range, aborts the
+        // program.
+        const TestProperty &GetTestProperty(int i) const;
+
+    private:
+        friend class TestInfo;
+        friend class TestCase;
+        friend class UnitTest;
+        friend class internal::DefaultGlobalTestPartResultReporter;
+        friend class internal::ExecDeathTest;
+        friend class internal::TestResultAccessor;
+        friend class internal::UnitTestImpl;
+        friend class internal::WindowsDeathTest;
+
+        // Gets the vector of TestPartResults.
+        const std::vector<TestPartResult> &test_part_results() const
+        {
+            return test_part_results_;
+        }
+
+        // Gets the vector of TestProperties.
+        const std::vector<TestProperty> &test_properties() const
+        {
+            return test_properties_;
+        }
+
+        // Sets the elapsed time.
+        void set_elapsed_time(TimeInMillis elapsed)
+        {
+            elapsed_time_ = elapsed;
+        }
+
+        // Adds a test property to the list. The property is validated and may add
+        // a non-fatal failure if invalid (e.g., if it conflicts with reserved
+        // key names). If a property is already recorded for the same key, the
+        // value will be updated, rather than storing multiple values for the same
+        // key.  xml_element specifies the element for which the property is being
+        // recorded and is used for validation.
+        void RecordProperty(const std::string &xml_element,
+                            const TestProperty &test_property);
+
+        // Adds a failure if the key is a reserved attribute of Google Test
+        // testcase tags.  Returns true if the property is valid.
+        // TODO(russr): Validate attribute names are legal and human readable.
+        static bool ValidateTestProperty(const std::string &xml_element,
+                                         const TestProperty &test_property);
+
+        // Adds a test part result to the list.
+        void AddTestPartResult(const TestPartResult &test_part_result);
+
+        // Returns the death test count.
+        int death_test_count() const
+        {
+            return death_test_count_;
+        }
+
+        // Increments the death test count, returning the new count.
+        int increment_death_test_count()
+        {
+            return ++death_test_count_;
+        }
+
+        // Clears the test part results.
+        void ClearTestPartResults();
+
+        // Clears the object.
+        void Clear();
+
+        // Protects mutable state of the property vector and of owned
+        // properties, whose values may be updated.
+        internal::Mutex test_properites_mutex_;
+
+        // The vector of TestPartResults
+        std::vector<TestPartResult> test_part_results_;
+        // The vector of TestProperties
+        std::vector<TestProperty> test_properties_;
+        // Running count of death tests.
+        int death_test_count_;
+        // The elapsed time, in milliseconds.
+        TimeInMillis elapsed_time_;
+
+        // We disallow copying TestResult.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestResult);
+    };  // class TestResult
+
+// A TestInfo object stores the following information about a test:
+//
+//   Test case name
+//   Test name
+//   Whether the test should be run
+//   A function pointer that creates the test object when invoked
+//   Test result
+//
+// The constructor of TestInfo registers itself with the UnitTest
+// singleton such that the RUN_ALL_TESTS() macro knows which tests to
+// run.
+    class GTEST_API_ TestInfo
+    {
+    public:
+        // Destructs a TestInfo object.  This function is not virtual, so
+        // don't inherit from TestInfo.
+        ~TestInfo();
+
+        // Returns the test case name.
+        const char *test_case_name() const
+        {
+            return test_case_name_.c_str();
+        }
+
+        // Returns the test name.
+        const char *name() const
+        {
+            return name_.c_str();
+        }
+
+        // Returns the name of the parameter type, or NULL if this is not a typed
+        // or a type-parameterized test.
+        const char *type_param() const
+        {
+            if (type_param_.get() != NULL)
+                return type_param_->c_str();
+            return NULL;
+        }
+
+        // Returns the text representation of the value parameter, or NULL if this
+        // is not a value-parameterized test.
+        const char *value_param() const
+        {
+            if (value_param_.get() != NULL)
+                return value_param_->c_str();
+            return NULL;
+        }
+
+        // Returns the file name where this test is defined.
+        const char *file() const
+        {
+            return location_.file.c_str();
+        }
+
+        // Returns the line where this test is defined.
+        int line() const
+        {
+            return location_.line;
+        }
+
+        // Returns true if this test should run, that is if the test is not
+        // disabled (or it is disabled but the also_run_disabled_tests flag has
+        // been specified) and its full name matches the user-specified filter.
+        //
+        // Google Test allows the user to filter the tests by their full names.
+        // The full name of a test Bar in test case Foo is defined as
+        // "Foo.Bar".  Only the tests that match the filter will run.
+        //
+        // A filter is a colon-separated list of glob (not regex) patterns,
+        // optionally followed by a '-' and a colon-separated list of
+        // negative patterns (tests to exclude).  A test is run if it
+        // matches one of the positive patterns and does not match any of
+        // the negative patterns.
+        //
+        // For example, *A*:Foo.* is a filter that matches any string that
+        // contains the character 'A' or starts with "Foo.".
+        bool should_run() const
+        {
+            return should_run_;
+        }
+
+        // Returns true iff this test will appear in the XML report.
+        bool is_reportable() const
+        {
+            // For now, the XML report includes all tests matching the filter.
+            // In the future, we may trim tests that are excluded because of
+            // sharding.
+            return matches_filter_;
+        }
+
+        // Returns the result of the test.
+        const TestResult *result() const
+        {
+            return &result_;
+        }
+
+    private:
+#if GTEST_HAS_DEATH_TEST
+        friend class internal::DefaultDeathTestFactory;
+#endif  // GTEST_HAS_DEATH_TEST
+        friend class Test;
+        friend class TestCase;
+        friend class internal::UnitTestImpl;
+        friend class internal::StreamingListenerTest;
+        friend TestInfo *internal::MakeAndRegisterTestInfo(
+            const char *test_case_name,
+            const char *name,
+            const char *type_param,
+            const char *value_param,
+            internal::CodeLocation code_location,
+            internal::TypeId fixture_class_id,
+            Test::SetUpTestCaseFunc set_up_tc,
+            Test::TearDownTestCaseFunc tear_down_tc,
+            internal::TestFactoryBase *factory);
+
+        // Constructs a TestInfo object. The newly constructed instance assumes
+        // ownership of the factory object.
+        TestInfo(const std::string &test_case_name,
+                 const std::string &name,
+                 const char *a_type_param,   // NULL if not a type-parameterized test
+                 const char *a_value_param,  // NULL if not a value-parameterized test
+                 internal::CodeLocation a_code_location,
+                 internal::TypeId fixture_class_id,
+                 internal::TestFactoryBase *factory);
+
+        // Increments the number of death tests encountered in this test so
+        // far.
+        int increment_death_test_count()
+        {
+            return result_.increment_death_test_count();
+        }
+
+        // Creates the test object, runs it, records its result, and then
+        // deletes it.
+        void Run();
+
+        static void ClearTestResult(TestInfo *test_info)
+        {
+            test_info->result_.Clear();
+        }
+
+        // These fields are immutable properties of the test.
+        const std::string test_case_name_;     // Test case name
+        const std::string name_;               // Test name
+        // Name of the parameter type, or NULL if this is not a typed or a
+        // type-parameterized test.
+        const internal::scoped_ptr<const ::std::string> type_param_;
+        // Text representation of the value parameter, or NULL if this is not a
+        // value-parameterized test.
+        const internal::scoped_ptr<const ::std::string> value_param_;
+        internal::CodeLocation location_;
+        const internal::TypeId fixture_class_id_;   // ID of the test fixture class
+        bool should_run_;                 // True iff this test should run
+        bool is_disabled_;                // True iff this test is disabled
+        bool matches_filter_;             // True if this test matches the
+        // user-specified filter.
+        internal::TestFactoryBase *const factory_;  // The factory that creates
+        // the test object
+
+        // This field is mutable and needs to be reset before running the
+        // test for the second time.
+        TestResult result_;
+
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestInfo);
+    };
+
+// A test case, which consists of a vector of TestInfos.
+//
+// TestCase is not copyable.
+    class GTEST_API_ TestCase
+    {
+    public:
+        // Creates a TestCase with the given name.
+        //
+        // TestCase does NOT have a default constructor.  Always use this
+        // constructor to create a TestCase object.
+        //
+        // Arguments:
+        //
+        //   name:         name of the test case
+        //   a_type_param: the name of the test's type parameter, or NULL if
+        //                 this is not a type-parameterized test.
+        //   set_up_tc:    pointer to the function that sets up the test case
+        //   tear_down_tc: pointer to the function that tears down the test case
+        TestCase(const char *name, const char *a_type_param,
+                 Test::SetUpTestCaseFunc set_up_tc,
+                 Test::TearDownTestCaseFunc tear_down_tc);
+
+        // Destructor of TestCase.
+        virtual ~TestCase();
+
+        // Gets the name of the TestCase.
+        const char *name() const
+        {
+            return name_.c_str();
+        }
+
+        // Returns the name of the parameter type, or NULL if this is not a
+        // type-parameterized test case.
+        const char *type_param() const
+        {
+            if (type_param_.get() != NULL)
+                return type_param_->c_str();
+            return NULL;
+        }
+
+        // Returns true if any test in this test case should run.
+        bool should_run() const
+        {
+            return should_run_;
+        }
+
+        // Gets the number of successful tests in this test case.
+        int successful_test_count() const;
+
+        // Gets the number of skipped tests in this test suite.
+        int skipped_test_count() const;
+
+        // Gets the number of failed tests in this test case.
+        int failed_test_count() const;
+
+        // Gets the number of disabled tests that will be reported in the XML report.
+        int reportable_disabled_test_count() const;
+
+        // Gets the number of disabled tests in this test case.
+        int disabled_test_count() const;
+
+        // Gets the number of tests to be printed in the XML report.
+        int reportable_test_count() const;
+
+        // Get the number of tests in this test case that should run.
+        int test_to_run_count() const;
+
+        // Gets the number of all tests in this test case.
+        int total_test_count() const;
+
+        // Returns true iff the test case passed.
+        bool Passed() const
+        {
+            return !Failed();
+        }
+
+        // Returns true iff the test case failed.
+        bool Failed() const
+        {
+            return failed_test_count() > 0;
+        }
+
+        // Returns the elapsed time, in milliseconds.
+        TimeInMillis elapsed_time() const
+        {
+            return elapsed_time_;
+        }
+
+        // Returns the i-th test among all the tests. i can range from 0 to
+        // total_test_count() - 1. If i is not in that range, returns NULL.
+        const TestInfo *GetTestInfo(int i) const;
+
+        // Returns the TestResult that holds test properties recorded during
+        // execution of SetUpTestCase and TearDownTestCase.
+        const TestResult &ad_hoc_test_result() const
+        {
+            return ad_hoc_test_result_;
+        }
+
+    private:
+        friend class Test;
+        friend class internal::UnitTestImpl;
+
+        // Gets the (mutable) vector of TestInfos in this TestCase.
+        std::vector<TestInfo *> &test_info_list()
+        {
+            return test_info_list_;
+        }
+
+        // Gets the (immutable) vector of TestInfos in this TestCase.
+        const std::vector<TestInfo *> &test_info_list() const
+        {
+            return test_info_list_;
+        }
+
+        // Returns the i-th test among all the tests. i can range from 0 to
+        // total_test_count() - 1. If i is not in that range, returns NULL.
+        TestInfo *GetMutableTestInfo(int i);
+
+        // Sets the should_run member.
+        void set_should_run(bool should)
+        {
+            should_run_ = should;
+        }
+
+        // Adds a TestInfo to this test case.  Will delete the TestInfo upon
+        // destruction of the TestCase object.
+        void AddTestInfo(TestInfo *test_info);
+
+        // Clears the results of all tests in this test case.
+        void ClearResult();
+
+        // Clears the results of all tests in the given test case.
+        static void ClearTestCaseResult(TestCase *test_case)
+        {
+            test_case->ClearResult();
+        }
+
+        // Runs every test in this TestCase.
+        void Run();
+
+        // Runs SetUpTestCase() for this TestCase.  This wrapper is needed
+        // for catching exceptions thrown from SetUpTestCase().
+        void RunSetUpTestCase()
+        {
+            (*set_up_tc_)();
+        }
+
+        // Runs TearDownTestCase() for this TestCase.  This wrapper is
+        // needed for catching exceptions thrown from TearDownTestCase().
+        void RunTearDownTestCase()
+        {
+            (*tear_down_tc_)();
+        }
+
+        // Returns true iff test passed.
+        static bool TestPassed(const TestInfo *test_info)
+        {
+            return test_info->should_run() && test_info->result()->Passed();
+        }
+
+        // Returns true if and only if test skipped.
+        static bool TestSkipped(const TestInfo *test_info)
+        {
+            return test_info->should_run() && test_info->result()->Skipped();
+        }
+
+        // Returns true iff test failed.
+        static bool TestFailed(const TestInfo *test_info)
+        {
+            return test_info->should_run() && test_info->result()->Failed();
+        }
+
+        // Returns true iff the test is disabled and will be reported in the XML
+        // report.
+        static bool TestReportableDisabled(const TestInfo *test_info)
+        {
+            return test_info->is_reportable() && test_info->is_disabled_;
+        }
+
+        // Returns true iff test is disabled.
+        static bool TestDisabled(const TestInfo *test_info)
+        {
+            return test_info->is_disabled_;
+        }
+
+        // Returns true iff this test will appear in the XML report.
+        static bool TestReportable(const TestInfo *test_info)
+        {
+            return test_info->is_reportable();
+        }
+
+        // Returns true if the given test should run.
+        static bool ShouldRunTest(const TestInfo *test_info)
+        {
+            return test_info->should_run();
+        }
+
+        // Shuffles the tests in this test case.
+        void ShuffleTests(internal::Random *random);
+
+        // Restores the test order to before the first shuffle.
+        void UnshuffleTests();
+
+        // Name of the test case.
+        std::string name_;
+        // Name of the parameter type, or NULL if this is not a typed or a
+        // type-parameterized test.
+        const internal::scoped_ptr<const ::std::string> type_param_;
+        // The vector of TestInfos in their original order.  It owns the
+        // elements in the vector.
+        std::vector<TestInfo *> test_info_list_;
+        // Provides a level of indirection for the test list to allow easy
+        // shuffling and restoring the test order.  The i-th element in this
+        // vector is the index of the i-th test in the shuffled test list.
+        std::vector<int> test_indices_;
+        // Pointer to the function that sets up the test case.
+        Test::SetUpTestCaseFunc set_up_tc_;
+        // Pointer to the function that tears down the test case.
+        Test::TearDownTestCaseFunc tear_down_tc_;
+        // True iff any test in this test case should run.
+        bool should_run_;
+        // Elapsed time, in milliseconds.
+        TimeInMillis elapsed_time_;
+        // Holds test properties recorded during execution of SetUpTestCase and
+        // TearDownTestCase.
+        TestResult ad_hoc_test_result_;
+
+        // We disallow copying TestCases.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestCase);
+    };
+
+// An Environment object is capable of setting up and tearing down an
+// environment.  You should subclass this to define your own
+// environment(s).
+//
+// An Environment object does the set-up and tear-down in virtual
+// methods SetUp() and TearDown() instead of the constructor and the
+// destructor, as:
+//
+//   1. You cannot safely throw from a destructor.  This is a problem
+//      as in some cases Google Test is used where exceptions are enabled, and
+//      we may want to implement ASSERT_* using exceptions where they are
+//      available.
+//   2. You cannot use ASSERT_* directly in a constructor or
+//      destructor.
+    class Environment
+    {
+    public:
+        // The d'tor is virtual as we need to subclass Environment.
+        virtual ~Environment() {}
+
+        // Override this to define how to set up the environment.
+        virtual void SetUp() {}
+
+        // Override this to define how to tear down the environment.
+        virtual void TearDown() {}
+    private:
+        // If you see an error about overriding the following function or
+        // about it being private, you have mis-spelled SetUp() as Setup().
+        struct Setup_should_be_spelled_SetUp {};
+        virtual Setup_should_be_spelled_SetUp *Setup()
+        {
+            return NULL;
+        }
+    };
+
+// The interface for tracing execution of tests. The methods are organized in
+// the order the corresponding events are fired.
+    class TestEventListener
+    {
+    public:
+        virtual ~TestEventListener() {}
+
+        // Fired before any test activity starts.
+        virtual void OnTestProgramStart(const UnitTest &unit_test) = 0;
+
+        // Fired before each iteration of tests starts.  There may be more than
+        // one iteration if GTEST_FLAG(repeat) is set. iteration is the iteration
+        // index, starting from 0.
+        virtual void OnTestIterationStart(const UnitTest &unit_test,
+                                          int iteration) = 0;
+
+        // Fired before environment set-up for each iteration of tests starts.
+        virtual void OnEnvironmentsSetUpStart(const UnitTest &unit_test) = 0;
+
+        // Fired after environment set-up for each iteration of tests ends.
+        virtual void OnEnvironmentsSetUpEnd(const UnitTest &unit_test) = 0;
+
+        // Fired before the test case starts.
+        virtual void OnTestCaseStart(const TestCase &test_case) = 0;
+
+        // Fired before the test starts.
+        virtual void OnTestStart(const TestInfo &test_info) = 0;
+
+        // Fired after a failed assertion or a SUCCEED() invocation.
+        virtual void OnTestPartResult(const TestPartResult &test_part_result) = 0;
+
+        // Fired after the test ends.
+        virtual void OnTestEnd(const TestInfo &test_info) = 0;
+
+        // Fired after the test case ends.
+        virtual void OnTestCaseEnd(const TestCase &test_case) = 0;
+
+        // Fired before environment tear-down for each iteration of tests starts.
+        virtual void OnEnvironmentsTearDownStart(const UnitTest &unit_test) = 0;
+
+        // Fired after environment tear-down for each iteration of tests ends.
+        virtual void OnEnvironmentsTearDownEnd(const UnitTest &unit_test) = 0;
+
+        // Fired after each iteration of tests finishes.
+        virtual void OnTestIterationEnd(const UnitTest &unit_test,
+                                        int iteration) = 0;
+
+        // Fired after all test activities have ended.
+        virtual void OnTestProgramEnd(const UnitTest &unit_test) = 0;
+    };
+
+// The convenience class for users who need to override just one or two
+// methods and are not concerned that a possible change to a signature of
+// the methods they override will not be caught during the build.  For
+// comments about each method please see the definition of TestEventListener
+// above.
+    class EmptyTestEventListener : public TestEventListener
+    {
+    public:
+        virtual void OnTestProgramStart(const UnitTest & /*unit_test*/) {}
+        virtual void OnTestIterationStart(const UnitTest & /*unit_test*/,
+                                          int /*iteration*/) {}
+        virtual void OnEnvironmentsSetUpStart(const UnitTest & /*unit_test*/) {}
+        virtual void OnEnvironmentsSetUpEnd(const UnitTest & /*unit_test*/) {}
+        virtual void OnTestCaseStart(const TestCase & /*test_case*/) {}
+        virtual void OnTestStart(const TestInfo & /*test_info*/) {}
+        virtual void OnTestPartResult(const TestPartResult & /*test_part_result*/) {}
+        virtual void OnTestEnd(const TestInfo & /*test_info*/) {}
+        virtual void OnTestCaseEnd(const TestCase & /*test_case*/) {}
+        virtual void OnEnvironmentsTearDownStart(const UnitTest & /*unit_test*/) {}
+        virtual void OnEnvironmentsTearDownEnd(const UnitTest & /*unit_test*/) {}
+        virtual void OnTestIterationEnd(const UnitTest & /*unit_test*/,
+                                        int /*iteration*/) {}
+        virtual void OnTestProgramEnd(const UnitTest & /*unit_test*/) {}
+    };
+
+// TestEventListeners lets users add listeners to track events in Google Test.
+    class GTEST_API_ TestEventListeners
+    {
+    public:
+        TestEventListeners();
+        ~TestEventListeners();
+
+        // Appends an event listener to the end of the list. Google Test assumes
+        // the ownership of the listener (i.e. it will delete the listener when
+        // the test program finishes).
+        void Append(TestEventListener *listener);
+
+        // Removes the given event listener from the list and returns it.  It then
+        // becomes the caller's responsibility to delete the listener. Returns
+        // NULL if the listener is not found in the list.
+        TestEventListener *Release(TestEventListener *listener);
+
+        // Returns the standard listener responsible for the default console
+        // output.  Can be removed from the listeners list to shut down default
+        // console output.  Note that removing this object from the listener list
+        // with Release transfers its ownership to the caller and makes this
+        // function return NULL the next time.
+        TestEventListener *default_result_printer() const
+        {
+            return default_result_printer_;
+        }
+
+        // Returns the standard listener responsible for the default XML output
+        // controlled by the --gtest_output=xml flag.  Can be removed from the
+        // listeners list by users who want to shut down the default XML output
+        // controlled by this flag and substitute it with custom one.  Note that
+        // removing this object from the listener list with Release transfers its
+        // ownership to the caller and makes this function return NULL the next
+        // time.
+        TestEventListener *default_xml_generator() const
+        {
+            return default_xml_generator_;
+        }
+
+    private:
+        friend class TestCase;
+        friend class TestInfo;
+        friend class internal::DefaultGlobalTestPartResultReporter;
+        friend class internal::NoExecDeathTest;
+        friend class internal::TestEventListenersAccessor;
+        friend class internal::UnitTestImpl;
+
+        // Returns repeater that broadcasts the TestEventListener events to all
+        // subscribers.
+        TestEventListener *repeater();
+
+        // Sets the default_result_printer attribute to the provided listener.
+        // The listener is also added to the listener list and previous
+        // default_result_printer is removed from it and deleted. The listener can
+        // also be NULL in which case it will not be added to the list. Does
+        // nothing if the previous and the current listener objects are the same.
+        void SetDefaultResultPrinter(TestEventListener *listener);
+
+        // Sets the default_xml_generator attribute to the provided listener.  The
+        // listener is also added to the listener list and previous
+        // default_xml_generator is removed from it and deleted. The listener can
+        // also be NULL in which case it will not be added to the list. Does
+        // nothing if the previous and the current listener objects are the same.
+        void SetDefaultXmlGenerator(TestEventListener *listener);
+
+        // Controls whether events will be forwarded by the repeater to the
+        // listeners in the list.
+        bool EventForwardingEnabled() const;
+        void SuppressEventForwarding();
+
+        // The actual list of listeners.
+        internal::TestEventRepeater *repeater_;
+        // Listener responsible for the standard result output.
+        TestEventListener *default_result_printer_;
+        // Listener responsible for the creation of the XML output file.
+        TestEventListener *default_xml_generator_;
+
+        // We disallow copying TestEventListeners.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(TestEventListeners);
+    };
+
+// A UnitTest consists of a vector of TestCases.
+//
+// This is a singleton class.  The only instance of UnitTest is
+// created when UnitTest::GetInstance() is first called.  This
+// instance is never deleted.
+//
+// UnitTest is not copyable.
+//
+// This class is thread-safe as long as the methods are called
+// according to their specification.
+    class GTEST_API_ UnitTest
+    {
+    public:
+        // Gets the singleton UnitTest object.  The first time this method
+        // is called, a UnitTest object is constructed and returned.
+        // Consecutive calls will return the same object.
+        static UnitTest *GetInstance();
+
+        // Runs all tests in this UnitTest object and prints the result.
+        // Returns 0 if successful, or 1 otherwise.
+        //
+        // This method can only be called from the main thread.
+        //
+        // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        int Run() GTEST_MUST_USE_RESULT_;
+
+        // Returns the working directory when the first TEST() or TEST_F()
+        // was executed.  The UnitTest object owns the string.
+        const char *original_working_dir() const;
+
+        // Returns the TestCase object for the test that's currently running,
+        // or NULL if no test is running.
+        const TestCase *current_test_case() const
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Returns the TestInfo object for the test that's currently running,
+        // or NULL if no test is running.
+        const TestInfo *current_test_info() const
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Returns the random seed used at the start of the current test run.
+        int random_seed() const;
+
+#if GTEST_HAS_PARAM_TEST
+        // Returns the ParameterizedTestCaseRegistry object used to keep track of
+        // value-parameterized tests and instantiate and register them.
+        //
+        // INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        internal::ParameterizedTestCaseRegistry &parameterized_test_registry()
+        GTEST_LOCK_EXCLUDED_(mutex_);
+#endif  // GTEST_HAS_PARAM_TEST
+
+        // Gets the number of successful test cases.
+        int successful_test_case_count() const;
+
+        // Gets the number of failed test cases.
+        int failed_test_case_count() const;
+
+        // Gets the number of all test cases.
+        int total_test_case_count() const;
+
+        // Gets the number of all test cases that contain at least one test
+        // that should run.
+        int test_case_to_run_count() const;
+
+        // Gets the number of successful tests.
+        int successful_test_count() const;
+
+        // Gets the number of skipped tests.
+        int skipped_test_count() const;
+
+        // Gets the number of failed tests.
+        int failed_test_count() const;
+
+        // Gets the number of disabled tests that will be reported in the XML report.
+        int reportable_disabled_test_count() const;
+
+        // Gets the number of disabled tests.
+        int disabled_test_count() const;
+
+        // Gets the number of tests to be printed in the XML report.
+        int reportable_test_count() const;
+
+        // Gets the number of all tests.
+        int total_test_count() const;
+
+        // Gets the number of tests that should run.
+        int test_to_run_count() const;
+
+        // Gets the time of the test program start, in ms from the start of the
+        // UNIX epoch.
+        TimeInMillis start_timestamp() const;
+
+        // Gets the elapsed time, in milliseconds.
+        TimeInMillis elapsed_time() const;
+
+        // Returns true iff the unit test passed (i.e. all test cases passed).
+        bool Passed() const;
+
+        // Returns true iff the unit test failed (i.e. some test case failed
+        // or something outside of all tests failed).
+        bool Failed() const;
+
+        // Gets the i-th test case among all the test cases. i can range from 0 to
+        // total_test_case_count() - 1. If i is not in that range, returns NULL.
+        const TestCase *GetTestCase(int i) const;
+
+        // Returns the TestResult containing information on test failures and
+        // properties logged outside of individual test cases.
+        const TestResult &ad_hoc_test_result() const;
+
+        // Returns the list of event listeners that can be used to track events
+        // inside Google Test.
+        TestEventListeners &listeners();
+
+    private:
+        // Registers and returns a global test environment.  When a test
+        // program is run, all global test environments will be set-up in
+        // the order they were registered.  After all tests in the program
+        // have finished, all global test environments will be torn-down in
+        // the *reverse* order they were registered.
+        //
+        // The UnitTest object takes ownership of the given environment.
+        //
+        // This method can only be called from the main thread.
+        Environment *AddEnvironment(Environment *env);
+
+        // Adds a TestPartResult to the current TestResult object.  All
+        // Google Test assertion macros (e.g. ASSERT_TRUE, EXPECT_EQ, etc)
+        // eventually call this to report their results.  The user code
+        // should use the assertion macros instead of calling this directly.
+        void AddTestPartResult(TestPartResult::Type result_type,
+                               const char *file_name,
+                               int line_number,
+                               const std::string &message,
+                               const std::string &os_stack_trace)
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Adds a TestProperty to the current TestResult object when invoked from
+        // inside a test, to current TestCase's ad_hoc_test_result_ when invoked
+        // from SetUpTestCase or TearDownTestCase, or to the global property set
+        // when invoked elsewhere.  If the result already contains a property with
+        // the same key, the value will be updated.
+        void RecordProperty(const std::string &key, const std::string &value);
+
+        // Gets the i-th test case among all the test cases. i can range from 0 to
+        // total_test_case_count() - 1. If i is not in that range, returns NULL.
+        TestCase *GetMutableTestCase(int i);
+
+        // Accessors for the implementation object.
+        internal::UnitTestImpl *impl()
+        {
+            return impl_;
+        }
+        const internal::UnitTestImpl *impl() const
+        {
+            return impl_;
+        }
+
+        // These classes and functions are friends as they need to access private
+        // members of UnitTest.
+        friend class Test;
+        friend class internal::AssertHelper;
+        friend class internal::ScopedTrace;
+        friend class internal::StreamingListenerTest;
+        friend class internal::UnitTestRecordPropertyTestHelper;
+        friend Environment *AddGlobalTestEnvironment(Environment *env);
+        friend internal::UnitTestImpl *internal::GetUnitTestImpl();
+        friend void internal::ReportFailureInUnknownLocation(
+            TestPartResult::Type result_type,
+            const std::string &message);
+
+        // Creates an empty UnitTest.
+        UnitTest();
+
+        // D'tor
+        virtual ~UnitTest();
+
+        // Pushes a trace defined by SCOPED_TRACE() on to the per-thread
+        // Google Test trace stack.
+        void PushGTestTrace(const internal::TraceInfo &trace)
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Pops a trace from the per-thread Google Test trace stack.
+        void PopGTestTrace()
+        GTEST_LOCK_EXCLUDED_(mutex_);
+
+        // Protects mutable state in *impl_.  This is mutable as some const
+        // methods need to lock it too.
+        mutable internal::Mutex mutex_;
+
+        // Opaque implementation object.  This field is never changed once
+        // the object is constructed.  We don't mark it as const here, as
+        // doing so will cause a warning in the constructor of UnitTest.
+        // Mutable state in *impl_ is protected by mutex_.
+        internal::UnitTestImpl *impl_;
+
+        // We disallow copying UnitTest.
+        GTEST_DISALLOW_COPY_AND_ASSIGN_(UnitTest);
+    };
+
+// A convenient wrapper for adding an environment for the test
+// program.
+//
+// You should call this before RUN_ALL_TESTS() is called, probably in
+// main().  If you use gtest_main, you need to call this before main()
+// starts for it to take effect.  For example, you can define a global
+// variable like this:
+//
+//   testing::Environment* const foo_env =
+//       testing::AddGlobalTestEnvironment(new FooEnvironment);
+//
+// However, we strongly recommend you to write your own main() and
+// call AddGlobalTestEnvironment() there, as relying on initialization
+// of global variables makes the code harder to read and may cause
+// problems when you register multiple environments from different
+// translation units and the environments have dependencies among them
+// (remember that the compiler doesn't guarantee the order in which
+// global variables from different translation units are initialized).
+    inline Environment *AddGlobalTestEnvironment(Environment *env)
+    {
+        return UnitTest::GetInstance()->AddEnvironment(env);
+    }
+
+// Initializes Google Test.  This must be called before calling
+// RUN_ALL_TESTS().  In particular, it parses a command line for the
+// flags that Google Test recognizes.  Whenever a Google Test flag is
+// seen, it is removed from argv, and *argc is decremented.
+//
+// No value is returned.  Instead, the Google Test flag variables are
+// updated.
+//
+// Calling the function for the second time has no user-visible effect.
+    GTEST_API_ void InitGoogleTest(int *argc, char **argv);
+
+// This overloaded version can be used in Windows programs compiled in
+// UNICODE mode.
+    GTEST_API_ void InitGoogleTest(int *argc, wchar_t **argv);
+    GTEST_API_ bool GetGtestHelpFlag();
+
+    namespace internal
+    {
+
+// Separate the error generating code from the code path to reduce the stack
+// frame size of CmpHelperEQ. This helps reduce the overhead of some sanitizers
+// when calling EXPECT_* in a tight loop.
+        template <typename T1, typename T2>
+        AssertionResult CmpHelperEQFailure(const char *lhs_expression,
+                                           const char *rhs_expression,
+                                           const T1 &lhs, const T2 &rhs)
+        {
+            return EqFailure(lhs_expression,
+                             rhs_expression,
+                             FormatForComparisonFailureMessage(lhs, rhs),
+                             FormatForComparisonFailureMessage(rhs, lhs),
+                             false);
+        }
+
+// The helper function for {ASSERT|EXPECT}_EQ.
+        template <typename T1, typename T2>
+        AssertionResult CmpHelperEQ(const char *lhs_expression,
+                                    const char *rhs_expression,
+                                    const T1 &lhs,
+                                    const T2 &rhs)
+        {
+            GTEST_DISABLE_MSC_WARNINGS_PUSH_(4389 /* signed/unsigned mismatch */)
+            if (lhs == rhs) {
+                return AssertionSuccess();
+            }
+            GTEST_DISABLE_MSC_WARNINGS_POP_()
+
+            return CmpHelperEQFailure(lhs_expression, rhs_expression, lhs, rhs);
+        }
+
+// With this overloaded version, we allow anonymous enums to be used
+// in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous enums
+// can be implicitly cast to BiggestInt.
+        GTEST_API_ AssertionResult CmpHelperEQ(const char *lhs_expression,
+                                               const char *rhs_expression,
+                                               BiggestInt lhs,
+                                               BiggestInt rhs);
+
+// The helper class for {ASSERT|EXPECT}_EQ.  The template argument
+// lhs_is_null_literal is true iff the first argument to ASSERT_EQ()
+// is a null pointer literal.  The following default implementation is
+// for lhs_is_null_literal being false.
+        template <bool lhs_is_null_literal>
+        class EqHelper
+        {
+        public:
+            // This templatized version is for the general case.
+            template <typename T1, typename T2>
+            static AssertionResult Compare(const char *lhs_expression,
+                                           const char *rhs_expression,
+                                           const T1 &lhs,
+                                           const T2 &rhs)
+            {
+                return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs);
+            }
+
+            // With this overloaded version, we allow anonymous enums to be used
+            // in {ASSERT|EXPECT}_EQ when compiled with gcc 4, as anonymous
+            // enums can be implicitly cast to BiggestInt.
+            //
+            // Even though its body looks the same as the above version, we
+            // cannot merge the two, as it will make anonymous enums unhappy.
+            static AssertionResult Compare(const char *lhs_expression,
+                                           const char *rhs_expression,
+                                           BiggestInt lhs,
+                                           BiggestInt rhs)
+            {
+                return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs);
+            }
+        };
+
+// This specialization is used when the first argument to ASSERT_EQ()
+// is a null pointer literal, like NULL, false, or 0.
+        template <>
+        class EqHelper<true>
+        {
+        public:
+            // We define two overloaded versions of Compare().  The first
+            // version will be picked when the second argument to ASSERT_EQ() is
+            // NOT a pointer, e.g. ASSERT_EQ(0, AnIntFunction()) or
+            // EXPECT_EQ(false, a_bool).
+            template <typename T1, typename T2>
+            static AssertionResult Compare(
+                const char *lhs_expression,
+                const char *rhs_expression,
+                const T1 &lhs,
+                const T2 &rhs,
+                // The following line prevents this overload from being considered if T2
+                // is not a pointer type.  We need this because ASSERT_EQ(NULL, my_ptr)
+                // expands to Compare("", "", NULL, my_ptr), which requires a conversion
+                // to match the Secret* in the other overload, which would otherwise make
+                // this template match better.
+                typename EnableIf<!is_pointer<T2>::value>::type * = 0)
+            {
+                return CmpHelperEQ(lhs_expression, rhs_expression, lhs, rhs);
+            }
+
+            // This version will be picked when the second argument to ASSERT_EQ() is a
+            // pointer, e.g. ASSERT_EQ(NULL, a_pointer).
+            template <typename T>
+            static AssertionResult Compare(
+                const char *lhs_expression,
+                const char *rhs_expression,
+                // We used to have a second template parameter instead of Secret*.  That
+                // template parameter would deduce to 'long', making this a better match
+                // than the first overload even without the first overload's EnableIf.
+                // Unfortunately, gcc with -Wconversion-null warns when "passing NULL to
+                // non-pointer argument" (even a deduced integral argument), so the old
+                // implementation caused warnings in user code.
+                Secret * /* lhs (NULL) */,
+                T *rhs)
+            {
+                // We already know that 'lhs' is a null pointer.
+                return CmpHelperEQ(lhs_expression, rhs_expression,
+                                   static_cast<T *>(NULL), rhs);
+            }
+        };
+
+// Separate the error generating code from the code path to reduce the stack
+// frame size of CmpHelperOP. This helps reduce the overhead of some sanitizers
+// when calling EXPECT_OP in a tight loop.
+        template <typename T1, typename T2>
+        AssertionResult CmpHelperOpFailure(const char *expr1, const char *expr2,
+                                           const T1 &val1, const T2 &val2,
+                                           const char *op)
+        {
+            return AssertionFailure()
+                   << "Expected: (" << expr1 << ") " << op << " (" << expr2
+                   << "), actual: " << FormatForComparisonFailureMessage(val1, val2)
+                   << " vs " << FormatForComparisonFailureMessage(val2, val1);
+        }
+
+// A macro for implementing the helper functions needed to implement
+// ASSERT_?? and EXPECT_??.  It is here just to avoid copy-and-paste
+// of similar code.
+//
+// For each templatized helper function, we also define an overloaded
+// version for BiggestInt in order to reduce code bloat and allow
+// anonymous enums to be used with {ASSERT|EXPECT}_?? when compiled
+// with gcc 4.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+
+#define GTEST_IMPL_CMP_HELPER_(op_name, op)\
+template <typename T1, typename T2>\
+AssertionResult CmpHelper##op_name(const char* expr1, const char* expr2, \
+                                   const T1& val1, const T2& val2) {\
+  if (val1 op val2) {\
+    return AssertionSuccess();\
+  } else {\
+    return CmpHelperOpFailure(expr1, expr2, val1, val2, #op);\
+  }\
+}\
+GTEST_API_ AssertionResult CmpHelper##op_name(\
+    const char* expr1, const char* expr2, BiggestInt val1, BiggestInt val2)
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+
+// Implements the helper function for {ASSERT|EXPECT}_NE
+        GTEST_IMPL_CMP_HELPER_(NE, !=);
+// Implements the helper function for {ASSERT|EXPECT}_LE
+        GTEST_IMPL_CMP_HELPER_(LE, <=);
+// Implements the helper function for {ASSERT|EXPECT}_LT
+        GTEST_IMPL_CMP_HELPER_(LT, <);
+// Implements the helper function for {ASSERT|EXPECT}_GE
+        GTEST_IMPL_CMP_HELPER_(GE, >=);
+// Implements the helper function for {ASSERT|EXPECT}_GT
+        GTEST_IMPL_CMP_HELPER_(GT, >);
+
+#undef GTEST_IMPL_CMP_HELPER_
+
+// The helper function for {ASSERT|EXPECT}_STREQ.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTREQ(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const char *s1,
+                                                  const char *s2);
+
+// The helper function for {ASSERT|EXPECT}_STRCASEEQ.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRCASEEQ(const char *s1_expression,
+                                                      const char *s2_expression,
+                                                      const char *s1,
+                                                      const char *s2);
+
+// The helper function for {ASSERT|EXPECT}_STRNE.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRNE(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const char *s1,
+                                                  const char *s2);
+
+// The helper function for {ASSERT|EXPECT}_STRCASENE.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRCASENE(const char *s1_expression,
+                                                      const char *s2_expression,
+                                                      const char *s1,
+                                                      const char *s2);
+
+
+// Helper function for *_STREQ on wide strings.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTREQ(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const wchar_t *s1,
+                                                  const wchar_t *s2);
+
+// Helper function for *_STRNE on wide strings.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult CmpHelperSTRNE(const char *s1_expression,
+                                                  const char *s2_expression,
+                                                  const wchar_t *s1,
+                                                  const wchar_t *s2);
+
+    }  // namespace internal
+
+// IsSubstring() and IsNotSubstring() are intended to be used as the
+// first argument to {EXPECT,ASSERT}_PRED_FORMAT2(), not by
+// themselves.  They check whether needle is a substring of haystack
+// (NULL is considered a substring of itself only), and return an
+// appropriate error message when they fail.
+//
+// The {needle,haystack}_expr arguments are the stringified
+// expressions that generated the two real arguments.
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const char *needle, const char *haystack);
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const wchar_t *needle, const wchar_t *haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const char *needle, const char *haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const wchar_t *needle, const wchar_t *haystack);
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::string &needle, const ::std::string &haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::string &needle, const ::std::string &haystack);
+
+#if GTEST_HAS_STD_WSTRING
+    GTEST_API_ AssertionResult IsSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::wstring &needle, const ::std::wstring &haystack);
+    GTEST_API_ AssertionResult IsNotSubstring(
+        const char *needle_expr, const char *haystack_expr,
+        const ::std::wstring &needle, const ::std::wstring &haystack);
+#endif  // GTEST_HAS_STD_WSTRING
+
+    namespace internal
+    {
+
+// Helper template function for comparing floating-points.
+//
+// Template parameter:
+//
+//   RawType: the raw floating-point type (either float or double)
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        template <typename RawType>
+        AssertionResult CmpHelperFloatingPointEQ(const char *lhs_expression,
+                                                 const char *rhs_expression,
+                                                 RawType lhs_value,
+                                                 RawType rhs_value)
+        {
+            const FloatingPoint<RawType> lhs(lhs_value), rhs(rhs_value);
+
+            if (lhs.AlmostEquals(rhs)) {
+                return AssertionSuccess();
+            }
+
+            ::std::stringstream lhs_ss;
+            lhs_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+                   << lhs_value;
+
+            ::std::stringstream rhs_ss;
+            rhs_ss << std::setprecision(std::numeric_limits<RawType>::digits10 + 2)
+                   << rhs_value;
+
+            return EqFailure(lhs_expression,
+                             rhs_expression,
+                             StringStreamToString(&lhs_ss),
+                             StringStreamToString(&rhs_ss),
+                             false);
+        }
+
+// Helper function for implementing ASSERT_NEAR.
+//
+// INTERNAL IMPLEMENTATION - DO NOT USE IN A USER PROGRAM.
+        GTEST_API_ AssertionResult DoubleNearPredFormat(const char *expr1,
+                                                        const char *expr2,
+                                                        const char *abs_error_expr,
+                                                        double val1,
+                                                        double val2,
+                                                        double abs_error);
+
+// INTERNAL IMPLEMENTATION - DO NOT USE IN USER CODE.
+// A class that enables one to stream messages to assertion macros
+        class GTEST_API_ AssertHelper
+        {
+        public:
+            // Constructor.
+            AssertHelper(TestPartResult::Type type,
+                         const char *file,
+                         int line,
+                         const char *message);
+            ~AssertHelper();
+
+            // Message assignment is a semantic trick to enable assertion
+            // streaming; see the GTEST_MESSAGE_ macro below.
+            void operator=(const Message &message) const;
+
+        private:
+            // We put our data in a struct so that the size of the AssertHelper class can
+            // be as small as possible.  This is important because gcc is incapable of
+            // re-using stack space even for temporary variables, so every EXPECT_EQ
+            // reserves stack space for another AssertHelper.
+            struct AssertHelperData {
+                AssertHelperData(TestPartResult::Type t,
+                                 const char *srcfile,
+                                 int line_num,
+                                 const char *msg)
+                    : type(t), file(srcfile), line(line_num), message(msg) { }
+
+                TestPartResult::Type const type;
+                const char *const file;
+                int const line;
+                std::string const message;
+
+            private:
+                GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelperData);
+            };
+
+            AssertHelperData *const data_;
+
+            GTEST_DISALLOW_COPY_AND_ASSIGN_(AssertHelper);
+        };
+
+    }  // namespace internal
+
+#if GTEST_HAS_PARAM_TEST
+// The pure interface class that all value-parameterized tests inherit from.
+// A value-parameterized class must inherit from both ::testing::Test and
+// ::testing::WithParamInterface. In most cases that just means inheriting
+// from ::testing::TestWithParam, but more complicated test hierarchies
+// may need to inherit from Test and WithParamInterface at different levels.
+//
+// This interface has support for accessing the test parameter value via
+// the GetParam() method.
+//
+// Use it with one of the parameter generator defining functions, like Range(),
+// Values(), ValuesIn(), Bool(), and Combine().
+//
+// class FooTest : public ::testing::TestWithParam<int> {
+//  protected:
+//   FooTest() {
+//     // Can use GetParam() here.
+//   }
+//   virtual ~FooTest() {
+//     // Can use GetParam() here.
+//   }
+//   virtual void SetUp() {
+//     // Can use GetParam() here.
+//   }
+//   virtual void TearDown {
+//     // Can use GetParam() here.
+//   }
+// };
+// TEST_P(FooTest, DoesBar) {
+//   // Can use GetParam() method here.
+//   Foo foo;
+//   ASSERT_TRUE(foo.DoesBar(GetParam()));
+// }
+// INSTANTIATE_TEST_CASE_P(OneToTenRange, FooTest, ::testing::Range(1, 10));
+
+    template <typename T>
+    class WithParamInterface
+    {
+    public:
+        typedef T ParamType;
+        virtual ~WithParamInterface() {}
+
+        // The current parameter value. Is also available in the test fixture's
+        // constructor. This member function is non-static, even though it only
+        // references static data, to reduce the opportunity for incorrect uses
+        // like writing 'WithParamInterface<bool>::GetParam()' for a test that
+        // uses a fixture whose parameter type is int.
+        const ParamType &GetParam() const
+        {
+            GTEST_CHECK_(parameter_ != NULL)
+                    << "GetParam() can only be called inside a value-parameterized test "
+                    << "-- did you intend to write TEST_P instead of TEST_F?";
+            return *parameter_;
+        }
+
+    private:
+        // Sets parameter value. The caller is responsible for making sure the value
+        // remains alive and unchanged throughout the current test.
+        static void SetParam(const ParamType *parameter)
+        {
+            parameter_ = parameter;
+        }
+
+        // Static value used for accessing parameter during a test lifetime.
+        static const ParamType *parameter_;
+
+        // TestClass must be a subclass of WithParamInterface<T> and Test.
+        template <class TestClass> friend class internal::ParameterizedTestFactory;
+    };
+
+    template <typename T>
+    const T *WithParamInterface<T>::parameter_ = NULL;
+
+// Most value-parameterized classes can ignore the existence of
+// WithParamInterface, and can just inherit from ::testing::TestWithParam.
+
+    template <typename T>
+    class TestWithParam : public Test, public WithParamInterface<T>
+    {
+    };
+
+#endif  // GTEST_HAS_PARAM_TEST
+
+// Macros for indicating success/failure in test code.
+
+// Skips test in runtime.
+// Skipping test aborts current function.
+// Skipped tests are neither successful nor failed.
+#define GTEST_SKIP() GTEST_SKIP_("")
+
+// ADD_FAILURE unconditionally adds a failure to the current test.
+// SUCCEED generates a success - it doesn't automatically make the
+// current test successful, as a test is only successful when it has
+// no failure.
+//
+// EXPECT_* verifies that a certain condition is satisfied.  If not,
+// it behaves like ADD_FAILURE.  In particular:
+//
+//   EXPECT_TRUE  verifies that a Boolean condition is true.
+//   EXPECT_FALSE verifies that a Boolean condition is false.
+//
+// FAIL and ASSERT_* are similar to ADD_FAILURE and EXPECT_*, except
+// that they will also abort the current function on failure.  People
+// usually want the fail-fast behavior of FAIL and ASSERT_*, but those
+// writing data-driven tests often find themselves using ADD_FAILURE
+// and EXPECT_* more.
+
+// Generates a nonfatal failure with a generic message.
+#define ADD_FAILURE() GTEST_NONFATAL_FAILURE_("Failed")
+
+// Generates a nonfatal failure at the given source file location with
+// a generic message.
+#define ADD_FAILURE_AT(file, line) \
+  GTEST_MESSAGE_AT_(file, line, "Failed", \
+                    ::testing::TestPartResult::kNonFatalFailure)
+
+// Generates a fatal failure with a generic message.
+#define GTEST_FAIL() GTEST_FATAL_FAILURE_("Failed")
+
+// Define this macro to 1 to omit the definition of FAIL(), which is a
+// generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_FAIL
+# define FAIL() GTEST_FAIL()
+#endif
+
+// Generates a success with a generic message.
+#define GTEST_SUCCEED() GTEST_SUCCESS_("Succeeded")
+
+// Define this macro to 1 to omit the definition of SUCCEED(), which
+// is a generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_SUCCEED
+# define SUCCEED() GTEST_SUCCEED()
+#endif
+
+// Macros for testing exceptions.
+//
+//    * {ASSERT|EXPECT}_THROW(statement, expected_exception):
+//         Tests that the statement throws the expected exception.
+//    * {ASSERT|EXPECT}_NO_THROW(statement):
+//         Tests that the statement doesn't throw any exception.
+//    * {ASSERT|EXPECT}_ANY_THROW(statement):
+//         Tests that the statement throws an exception.
+
+#define EXPECT_THROW(statement, expected_exception) \
+  GTEST_TEST_THROW_(statement, expected_exception, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_NO_THROW(statement) \
+  GTEST_TEST_NO_THROW_(statement, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_ANY_THROW(statement) \
+  GTEST_TEST_ANY_THROW_(statement, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_THROW(statement, expected_exception) \
+  GTEST_TEST_THROW_(statement, expected_exception, GTEST_FATAL_FAILURE_)
+#define ASSERT_NO_THROW(statement) \
+  GTEST_TEST_NO_THROW_(statement, GTEST_FATAL_FAILURE_)
+#define ASSERT_ANY_THROW(statement) \
+  GTEST_TEST_ANY_THROW_(statement, GTEST_FATAL_FAILURE_)
+
+// Boolean assertions. Condition can be either a Boolean expression or an
+// AssertionResult. For more information on how to use AssertionResult with
+// these macros see comments on that class.
+#define EXPECT_TRUE(condition) \
+  GTEST_TEST_BOOLEAN_((condition), #condition, false, true, \
+                      GTEST_NONFATAL_FAILURE_)
+#define EXPECT_FALSE(condition) \
+  GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \
+                      GTEST_NONFATAL_FAILURE_)
+#define ASSERT_TRUE(condition) \
+  GTEST_TEST_BOOLEAN_((condition), #condition, false, true, \
+                      GTEST_FATAL_FAILURE_)
+#define ASSERT_FALSE(condition) \
+  GTEST_TEST_BOOLEAN_(!(condition), #condition, true, false, \
+                      GTEST_FATAL_FAILURE_)
+
+// Includes the auto-generated header that implements a family of
+// generic predicate assertion macros.
+// Copyright 2006, Google Inc.
+// All rights reserved.
+//
+// Redistribution and use in source and binary forms, with or without
+// modification, are permitted provided that the following conditions are
+// met:
+//
+//     * Redistributions of source code must retain the above copyright
+// notice, this list of conditions and the following disclaimer.
+//     * Redistributions in binary form must reproduce the above
+// copyright notice, this list of conditions and the following disclaimer
+// in the documentation and/or other materials provided with the
+// distribution.
+//     * Neither the name of Google Inc. nor the names of its
+// contributors may be used to endorse or promote products derived from
+// this software without specific prior written permission.
+//
+// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+// This file is AUTOMATICALLY GENERATED on 10/31/2011 by command
+// 'gen_gtest_pred_impl.py 5'.  DO NOT EDIT BY HAND!
+//
+// Implements a family of generic predicate assertion macros.
+
+#ifndef GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+#define GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+
+// Makes sure this header is not included before gtest.h.
+#ifndef GTEST_INCLUDE_GTEST_GTEST_H_
+# error Do not include gtest_pred_impl.h directly.  Include gtest.h instead.
+#endif  // GTEST_INCLUDE_GTEST_GTEST_H_
+
+// This header implements a family of generic predicate assertion
+// macros:
+//
+//   ASSERT_PRED_FORMAT1(pred_format, v1)
+//   ASSERT_PRED_FORMAT2(pred_format, v1, v2)
+//   ...
+//
+// where pred_format is a function or functor that takes n (in the
+// case of ASSERT_PRED_FORMATn) values and their source expression
+// text, and returns a testing::AssertionResult.  See the definition
+// of ASSERT_EQ in gtest.h for an example.
+//
+// If you don't care about formatting, you can use the more
+// restrictive version:
+//
+//   ASSERT_PRED1(pred, v1)
+//   ASSERT_PRED2(pred, v1, v2)
+//   ...
+//
+// where pred is an n-ary function or functor that returns bool,
+// and the values v1, v2, ..., must support the << operator for
+// streaming to std::ostream.
+//
+// We also define the EXPECT_* variations.
+//
+// For now we only support predicates whose arity is at most 5.
+// Please email googletestframework@googlegroups.com if you need
+// support for higher arities.
+
+// GTEST_ASSERT_ is the basic statement to which all of the assertions
+// in this file reduce.  Don't use this in your code.
+
+#define GTEST_ASSERT_(expression, on_failure) \
+  GTEST_AMBIGUOUS_ELSE_BLOCKER_ \
+  if (const ::testing::AssertionResult gtest_ar = (expression)) \
+    ; \
+  else \
+    on_failure(gtest_ar.failure_message())
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED1.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1>
+    AssertionResult AssertPred1Helper(const char *pred_text,
+                                      const char *e1,
+                                      Pred pred,
+                                      const T1 &v1)
+    {
+        if (pred(v1)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT1.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT1_(pred_format, v1, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, v1), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED1.  Don't use
+// this in your code.
+#define GTEST_PRED1_(pred, v1, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred1Helper(#pred, \
+                                             #v1, \
+                                             pred, \
+                                             v1), on_failure)
+
+// Unary predicate assertion macros.
+#define EXPECT_PRED_FORMAT1(pred_format, v1) \
+  GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED1(pred, v1) \
+  GTEST_PRED1_(pred, v1, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT1(pred_format, v1) \
+  GTEST_PRED_FORMAT1_(pred_format, v1, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED1(pred, v1) \
+  GTEST_PRED1_(pred, v1, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED2.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2>
+    AssertionResult AssertPred2Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2)
+    {
+        if (pred(v1, v2)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT2.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT2_(pred_format, v1, v2, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, v1, v2), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED2.  Don't use
+// this in your code.
+#define GTEST_PRED2_(pred, v1, v2, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred2Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             pred, \
+                                             v1, \
+                                             v2), on_failure)
+
+// Binary predicate assertion macros.
+#define EXPECT_PRED_FORMAT2(pred_format, v1, v2) \
+  GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED2(pred, v1, v2) \
+  GTEST_PRED2_(pred, v1, v2, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT2(pred_format, v1, v2) \
+  GTEST_PRED_FORMAT2_(pred_format, v1, v2, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED2(pred, v1, v2) \
+  GTEST_PRED2_(pred, v1, v2, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED3.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2,
+              typename T3>
+    AssertionResult AssertPred3Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      const char *e3,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2,
+                                      const T3 &v3)
+    {
+        if (pred(v1, v2, v3)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ", "
+               << e3 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2
+               << "\n" << e3 << " evaluates to " << v3;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT3.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, #v3, v1, v2, v3), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED3.  Don't use
+// this in your code.
+#define GTEST_PRED3_(pred, v1, v2, v3, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred3Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             #v3, \
+                                             pred, \
+                                             v1, \
+                                             v2, \
+                                             v3), on_failure)
+
+// Ternary predicate assertion macros.
+#define EXPECT_PRED_FORMAT3(pred_format, v1, v2, v3) \
+  GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED3(pred, v1, v2, v3) \
+  GTEST_PRED3_(pred, v1, v2, v3, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT3(pred_format, v1, v2, v3) \
+  GTEST_PRED_FORMAT3_(pred_format, v1, v2, v3, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED3(pred, v1, v2, v3) \
+  GTEST_PRED3_(pred, v1, v2, v3, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED4.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2,
+              typename T3,
+              typename T4>
+    AssertionResult AssertPred4Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      const char *e3,
+                                      const char *e4,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2,
+                                      const T3 &v3,
+                                      const T4 &v4)
+    {
+        if (pred(v1, v2, v3, v4)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ", "
+               << e3 << ", "
+               << e4 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2
+               << "\n" << e3 << " evaluates to " << v3
+               << "\n" << e4 << " evaluates to " << v4;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT4.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, v1, v2, v3, v4), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED4.  Don't use
+// this in your code.
+#define GTEST_PRED4_(pred, v1, v2, v3, v4, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred4Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             #v3, \
+                                             #v4, \
+                                             pred, \
+                                             v1, \
+                                             v2, \
+                                             v3, \
+                                             v4), on_failure)
+
+// 4-ary predicate assertion macros.
+#define EXPECT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \
+  GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED4(pred, v1, v2, v3, v4) \
+  GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT4(pred_format, v1, v2, v3, v4) \
+  GTEST_PRED_FORMAT4_(pred_format, v1, v2, v3, v4, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED4(pred, v1, v2, v3, v4) \
+  GTEST_PRED4_(pred, v1, v2, v3, v4, GTEST_FATAL_FAILURE_)
+
+
+
+// Helper function for implementing {EXPECT|ASSERT}_PRED5.  Don't use
+// this in your code.
+    template <typename Pred,
+              typename T1,
+              typename T2,
+              typename T3,
+              typename T4,
+              typename T5>
+    AssertionResult AssertPred5Helper(const char *pred_text,
+                                      const char *e1,
+                                      const char *e2,
+                                      const char *e3,
+                                      const char *e4,
+                                      const char *e5,
+                                      Pred pred,
+                                      const T1 &v1,
+                                      const T2 &v2,
+                                      const T3 &v3,
+                                      const T4 &v4,
+                                      const T5 &v5)
+    {
+        if (pred(v1, v2, v3, v4, v5)) return AssertionSuccess();
+
+        return AssertionFailure() << pred_text << "("
+               << e1 << ", "
+               << e2 << ", "
+               << e3 << ", "
+               << e4 << ", "
+               << e5 << ") evaluates to false, where"
+               << "\n" << e1 << " evaluates to " << v1
+               << "\n" << e2 << " evaluates to " << v2
+               << "\n" << e3 << " evaluates to " << v3
+               << "\n" << e4 << " evaluates to " << v4
+               << "\n" << e5 << " evaluates to " << v5;
+    }
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED_FORMAT5.
+// Don't use this in your code.
+#define GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, on_failure)\
+  GTEST_ASSERT_(pred_format(#v1, #v2, #v3, #v4, #v5, v1, v2, v3, v4, v5), \
+                on_failure)
+
+// Internal macro for implementing {EXPECT|ASSERT}_PRED5.  Don't use
+// this in your code.
+#define GTEST_PRED5_(pred, v1, v2, v3, v4, v5, on_failure)\
+  GTEST_ASSERT_(::testing::AssertPred5Helper(#pred, \
+                                             #v1, \
+                                             #v2, \
+                                             #v3, \
+                                             #v4, \
+                                             #v5, \
+                                             pred, \
+                                             v1, \
+                                             v2, \
+                                             v3, \
+                                             v4, \
+                                             v5), on_failure)
+
+// 5-ary predicate assertion macros.
+#define EXPECT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \
+  GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_)
+#define EXPECT_PRED5(pred, v1, v2, v3, v4, v5) \
+  GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_NONFATAL_FAILURE_)
+#define ASSERT_PRED_FORMAT5(pred_format, v1, v2, v3, v4, v5) \
+  GTEST_PRED_FORMAT5_(pred_format, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_)
+#define ASSERT_PRED5(pred, v1, v2, v3, v4, v5) \
+  GTEST_PRED5_(pred, v1, v2, v3, v4, v5, GTEST_FATAL_FAILURE_)
+
+
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_PRED_IMPL_H_
+
+// Macros for testing equalities and inequalities.
+//
+//    * {ASSERT|EXPECT}_EQ(v1, v2): Tests that v1 == v2
+//    * {ASSERT|EXPECT}_NE(v1, v2): Tests that v1 != v2
+//    * {ASSERT|EXPECT}_LT(v1, v2): Tests that v1 < v2
+//    * {ASSERT|EXPECT}_LE(v1, v2): Tests that v1 <= v2
+//    * {ASSERT|EXPECT}_GT(v1, v2): Tests that v1 > v2
+//    * {ASSERT|EXPECT}_GE(v1, v2): Tests that v1 >= v2
+//
+// When they are not, Google Test prints both the tested expressions and
+// their actual values.  The values must be compatible built-in types,
+// or you will get a compiler error.  By "compatible" we mean that the
+// values can be compared by the respective operator.
+//
+// Note:
+//
+//   1. It is possible to make a user-defined type work with
+//   {ASSERT|EXPECT}_??(), but that requires overloading the
+//   comparison operators and is thus discouraged by the Google C++
+//   Usage Guide.  Therefore, you are advised to use the
+//   {ASSERT|EXPECT}_TRUE() macro to assert that two objects are
+//   equal.
+//
+//   2. The {ASSERT|EXPECT}_??() macros do pointer comparisons on
+//   pointers (in particular, C strings).  Therefore, if you use it
+//   with two C strings, you are testing how their locations in memory
+//   are related, not how their content is related.  To compare two C
+//   strings by content, use {ASSERT|EXPECT}_STR*().
+//
+//   3. {ASSERT|EXPECT}_EQ(v1, v2) is preferred to
+//   {ASSERT|EXPECT}_TRUE(v1 == v2), as the former tells you
+//   what the actual value is when it fails, and similarly for the
+//   other comparisons.
+//
+//   4. Do not depend on the order in which {ASSERT|EXPECT}_??()
+//   evaluate their arguments, which is undefined.
+//
+//   5. These macros evaluate their arguments exactly once.
+//
+// Examples:
+//
+//   EXPECT_NE(5, Foo());
+//   EXPECT_EQ(NULL, a_pointer);
+//   ASSERT_LT(i, array_size);
+//   ASSERT_GT(records.size(), 0) << "There is no record left.";
+
+#define EXPECT_EQ(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal:: \
+                      EqHelper<GTEST_IS_NULL_LITERAL_(val1)>::Compare, \
+                      val1, val2)
+#define EXPECT_NE(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2)
+#define EXPECT_LE(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2)
+#define EXPECT_LT(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2)
+#define EXPECT_GE(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2)
+#define EXPECT_GT(val1, val2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2)
+
+#define GTEST_ASSERT_EQ(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal:: \
+                      EqHelper<GTEST_IS_NULL_LITERAL_(val1)>::Compare, \
+                      val1, val2)
+#define GTEST_ASSERT_NE(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperNE, val1, val2)
+#define GTEST_ASSERT_LE(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLE, val1, val2)
+#define GTEST_ASSERT_LT(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperLT, val1, val2)
+#define GTEST_ASSERT_GE(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGE, val1, val2)
+#define GTEST_ASSERT_GT(val1, val2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperGT, val1, val2)
+
+// Define macro GTEST_DONT_DEFINE_ASSERT_XY to 1 to omit the definition of
+// ASSERT_XY(), which clashes with some users' own code.
+
+#if !GTEST_DONT_DEFINE_ASSERT_EQ
+# define ASSERT_EQ(val1, val2) GTEST_ASSERT_EQ(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_NE
+# define ASSERT_NE(val1, val2) GTEST_ASSERT_NE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_LE
+# define ASSERT_LE(val1, val2) GTEST_ASSERT_LE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_LT
+# define ASSERT_LT(val1, val2) GTEST_ASSERT_LT(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_GE
+# define ASSERT_GE(val1, val2) GTEST_ASSERT_GE(val1, val2)
+#endif
+
+#if !GTEST_DONT_DEFINE_ASSERT_GT
+# define ASSERT_GT(val1, val2) GTEST_ASSERT_GT(val1, val2)
+#endif
+
+// C-string Comparisons.  All tests treat NULL and any non-NULL string
+// as different.  Two NULLs are equal.
+//
+//    * {ASSERT|EXPECT}_STREQ(s1, s2):     Tests that s1 == s2
+//    * {ASSERT|EXPECT}_STRNE(s1, s2):     Tests that s1 != s2
+//    * {ASSERT|EXPECT}_STRCASEEQ(s1, s2): Tests that s1 == s2, ignoring case
+//    * {ASSERT|EXPECT}_STRCASENE(s1, s2): Tests that s1 != s2, ignoring case
+//
+// For wide or narrow string objects, you can use the
+// {ASSERT|EXPECT}_??() macros.
+//
+// Don't depend on the order in which the arguments are evaluated,
+// which is undefined.
+//
+// These macros evaluate their arguments exactly once.
+
+#define EXPECT_STREQ(s1, s2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, s1, s2)
+#define EXPECT_STRNE(s1, s2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2)
+#define EXPECT_STRCASEEQ(s1, s2) \
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, s1, s2)
+#define EXPECT_STRCASENE(s1, s2)\
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2)
+
+#define ASSERT_STREQ(s1, s2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTREQ, s1, s2)
+#define ASSERT_STRNE(s1, s2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRNE, s1, s2)
+#define ASSERT_STRCASEEQ(s1, s2) \
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASEEQ, s1, s2)
+#define ASSERT_STRCASENE(s1, s2)\
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperSTRCASENE, s1, s2)
+
+// Macros for comparing floating-point numbers.
+//
+//    * {ASSERT|EXPECT}_FLOAT_EQ(val1, val2):
+//         Tests that two float values are almost equal.
+//    * {ASSERT|EXPECT}_DOUBLE_EQ(val1, val2):
+//         Tests that two double values are almost equal.
+//    * {ASSERT|EXPECT}_NEAR(v1, v2, abs_error):
+//         Tests that v1 and v2 are within the given distance to each other.
+//
+// Google Test uses ULP-based comparison to automatically pick a default
+// error bound that is appropriate for the operands.  See the
+// FloatingPoint template class in gtest-internal.h if you are
+// interested in the implementation details.
+
+#define EXPECT_FLOAT_EQ(val1, val2)\
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<float>, \
+                      val1, val2)
+
+#define EXPECT_DOUBLE_EQ(val1, val2)\
+  EXPECT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<double>, \
+                      val1, val2)
+
+#define ASSERT_FLOAT_EQ(val1, val2)\
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<float>, \
+                      val1, val2)
+
+#define ASSERT_DOUBLE_EQ(val1, val2)\
+  ASSERT_PRED_FORMAT2(::testing::internal::CmpHelperFloatingPointEQ<double>, \
+                      val1, val2)
+
+#define EXPECT_NEAR(val1, val2, abs_error)\
+  EXPECT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \
+                      val1, val2, abs_error)
+
+#define ASSERT_NEAR(val1, val2, abs_error)\
+  ASSERT_PRED_FORMAT3(::testing::internal::DoubleNearPredFormat, \
+                      val1, val2, abs_error)
+
+// These predicate format functions work on floating-point values, and
+// can be used in {ASSERT|EXPECT}_PRED_FORMAT2*(), e.g.
+//
+//   EXPECT_PRED_FORMAT2(testing::DoubleLE, Foo(), 5.0);
+
+// Asserts that val1 is less than, or almost equal to, val2.  Fails
+// otherwise.  In particular, it fails if either val1 or val2 is NaN.
+    GTEST_API_ AssertionResult FloatLE(const char *expr1, const char *expr2,
+                                       float val1, float val2);
+    GTEST_API_ AssertionResult DoubleLE(const char *expr1, const char *expr2,
+                                        double val1, double val2);
+
+
+#if GTEST_OS_WINDOWS
+
+// Macros that test for HRESULT failure and success, these are only useful
+// on Windows, and rely on Windows SDK macros and APIs to compile.
+//
+//    * {ASSERT|EXPECT}_HRESULT_{SUCCEEDED|FAILED}(expr)
+//
+// When expr unexpectedly fails or succeeds, Google Test prints the
+// expected result and the actual result with both a human-readable
+// string representation of the error, if available, as well as the
+// hex result code.
+# define EXPECT_HRESULT_SUCCEEDED(expr) \
+    EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr))
+
+# define ASSERT_HRESULT_SUCCEEDED(expr) \
+    ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTSuccess, (expr))
+
+# define EXPECT_HRESULT_FAILED(expr) \
+    EXPECT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr))
+
+# define ASSERT_HRESULT_FAILED(expr) \
+    ASSERT_PRED_FORMAT1(::testing::internal::IsHRESULTFailure, (expr))
+
+#endif  // GTEST_OS_WINDOWS
+
+// Macros that execute statement and check that it doesn't generate new fatal
+// failures in the current thread.
+//
+//   * {ASSERT|EXPECT}_NO_FATAL_FAILURE(statement);
+//
+// Examples:
+//
+//   EXPECT_NO_FATAL_FAILURE(Process());
+//   ASSERT_NO_FATAL_FAILURE(Process()) << "Process() failed";
+//
+#define ASSERT_NO_FATAL_FAILURE(statement) \
+    GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_FATAL_FAILURE_)
+#define EXPECT_NO_FATAL_FAILURE(statement) \
+    GTEST_TEST_NO_FATAL_FAILURE_(statement, GTEST_NONFATAL_FAILURE_)
+
+// Causes a trace (including the source file path, the current line
+// number, and the given message) to be included in every test failure
+// message generated by code in the current scope.  The effect is
+// undone when the control leaves the current scope.
+//
+// The message argument can be anything streamable to std::ostream.
+//
+// In the implementation, we include the current line number as part
+// of the dummy variable name, thus allowing multiple SCOPED_TRACE()s
+// to appear in the same block - as long as they are on different
+// lines.
+#define SCOPED_TRACE(message) \
+  ::testing::internal::ScopedTrace GTEST_CONCAT_TOKEN_(gtest_trace_, __LINE__)(\
+    __FILE__, __LINE__, ::testing::Message() << (message))
+
+// Compile-time assertion for type equality.
+// StaticAssertTypeEq<type1, type2>() compiles iff type1 and type2 are
+// the same type.  The value it returns is not interesting.
+//
+// Instead of making StaticAssertTypeEq a class template, we make it a
+// function template that invokes a helper class template.  This
+// prevents a user from misusing StaticAssertTypeEq<T1, T2> by
+// defining objects of that type.
+//
+// CAVEAT:
+//
+// When used inside a method of a class template,
+// StaticAssertTypeEq<T1, T2>() is effective ONLY IF the method is
+// instantiated.  For example, given:
+//
+//   template <typename T> class Foo {
+//    public:
+//     void Bar() { testing::StaticAssertTypeEq<int, T>(); }
+//   };
+//
+// the code:
+//
+//   void Test1() { Foo<bool> foo; }
+//
+// will NOT generate a compiler error, as Foo<bool>::Bar() is never
+// actually instantiated.  Instead, you need:
+//
+//   void Test2() { Foo<bool> foo; foo.Bar(); }
+//
+// to cause a compiler error.
+    template <typename T1, typename T2>
+    bool StaticAssertTypeEq()
+    {
+        (void)internal::StaticAssertTypeEqHelper<T1, T2>();
+        return true;
+    }
+
+// Defines a test.
+//
+// The first parameter is the name of the test case, and the second
+// parameter is the name of the test within the test case.
+//
+// The convention is to end the test case name with "Test".  For
+// example, a test case for the Foo class can be named FooTest.
+//
+// Test code should appear between braces after an invocation of
+// this macro.  Example:
+//
+//   TEST(FooTest, InitializesCorrectly) {
+//     Foo foo;
+//     EXPECT_TRUE(foo.StatusIsOK());
+//   }
+
+// Note that we call GetTestTypeId() instead of GetTypeId<
+// ::testing::Test>() here to get the type ID of testing::Test.  This
+// is to work around a suspected linker bug when using Google Test as
+// a framework on Mac OS X.  The bug causes GetTypeId<
+// ::testing::Test>() to return different values depending on whether
+// the call is from the Google Test framework itself or from user test
+// code.  GetTestTypeId() is guaranteed to always return the same
+// value, as it always calls GetTypeId<>() from the Google Test
+// framework.
+#define GTEST_TEST(test_case_name, test_name)\
+  GTEST_TEST_(test_case_name, test_name, \
+              ::testing::Test, ::testing::internal::GetTestTypeId())
+
+// Define this macro to 1 to omit the definition of TEST(), which
+// is a generic name and clashes with some other libraries.
+#if !GTEST_DONT_DEFINE_TEST
+# define TEST(test_case_name, test_name) GTEST_TEST(test_case_name, test_name)
+#endif
+
+// Defines a test that uses a test fixture.
+//
+// The first parameter is the name of the test fixture class, which
+// also doubles as the test case name.  The second parameter is the
+// name of the test within the test case.
+//
+// A test fixture class must be declared earlier.  The user should put
+// the test code between braces after using this macro.  Example:
+//
+//   class FooTest : public testing::Test {
+//    protected:
+//     virtual void SetUp() { b_.AddElement(3); }
+//
+//     Foo a_;
+//     Foo b_;
+//   };
+//
+//   TEST_F(FooTest, InitializesCorrectly) {
+//     EXPECT_TRUE(a_.StatusIsOK());
+//   }
+//
+//   TEST_F(FooTest, ReturnsElementCountCorrectly) {
+//     EXPECT_EQ(0, a_.size());
+//     EXPECT_EQ(1, b_.size());
+//   }
+
+#define TEST_F(test_fixture, test_name)\
+  GTEST_TEST_(test_fixture, test_name, test_fixture, \
+              ::testing::internal::GetTypeId<test_fixture>())
+
+// Returns a path to temporary directory.
+// Tries to determine an appropriate directory for the platform.
+    GTEST_API_ std::string TempDir();
+
+}  // namespace testing
+
+// Use this function in main() to run all tests.  It returns 0 if all
+// tests are successful, or 1 otherwise.
+//
+// RUN_ALL_TESTS() should be invoked after the command line has been
+// parsed by InitGoogleTest().
+//
+// This function was formerly a macro; thus, it is in the global
+// namespace and has an all-caps name.
+int RUN_ALL_TESTS() GTEST_MUST_USE_RESULT_;
+
+inline int RUN_ALL_TESTS()
+{
+    return ::testing::UnitTest::GetInstance()->Run();
+}
+
+#endif  // GTEST_INCLUDE_GTEST_GTEST_H_
diff --git a/test/hbw_allocator_performance_tests.cpp b/test/hbw_allocator_performance_tests.cpp
new file mode 100644
index 0000000..4fd05b9
--- /dev/null
+++ b/test/hbw_allocator_performance_tests.cpp
@@ -0,0 +1,421 @@
+/*
+* Copyright (C) 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/TaskFactory.hpp"
+#include "allocator_perf_tool/Stats.hpp"
+#include "allocator_perf_tool/Thread.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+
+class HbwAllocatorPerformanceTest: public :: testing::Test
+{
+private:
+    AllocatorFactory allocator_factory;
+
+protected:
+    void SetUp()
+    {
+        allocator_factory.initialize_allocator(AllocatorTypes::STANDARD_ALLOCATOR);
+    }
+
+    void TearDown()
+    {}
+
+    float run(unsigned kind, unsigned call, size_t threads_number,
+              size_t alloc_size, unsigned mem_operations_num)
+    {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+        TypesConf func_calls;
+        TypesConf allocator_types;
+
+        func_calls.enable_type(FunctionCalls::FREE);
+        func_calls.enable_type(call);
+        allocator_types.enable_type(kind);
+
+        TaskConf conf = {
+            mem_operations_num, //number of memory operations
+            {
+                mem_operations_num, //number of memory operations
+                alloc_size, //min. size of single allocation
+                alloc_size //max. size of single allocatioion
+            },
+            func_calls, //enable function calls
+            allocator_types, //enable allocators
+            11, //random seed
+            false, //do not log memory operations and statistics to csv file
+        };
+
+        for (int i=0; i<threads_number; i++) {
+            Task *task = task_factory.create(conf);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats time_stats;
+        for (int i=0; i<tasks.size(); i++) {
+            time_stats += tasks[i]->get_results();
+        }
+
+        return time_stats.stats[kind][call].total_time;
+    }
+
+    void run_test(unsigned kind, unsigned call, size_t threads_number,
+                  size_t alloc_size, unsigned mem_operations_num)
+    {
+        allocator_factory.initialize_allocator(kind);
+        float ref_time = run(AllocatorTypes::STANDARD_ALLOCATOR, call, threads_number,
+                             alloc_size, mem_operations_num);
+        float perf_time = run(kind, call, threads_number, alloc_size,
+                              mem_operations_num);
+        float ref_delta_time_percent = allocator_factory.calc_ref_delta(ref_time,
+                                                                        perf_time);
+
+        GTestAdapter::RecordProperty("total_time_spend_on_alloc", perf_time);
+        GTestAdapter::RecordProperty("alloc_operations_per_thread", mem_operations_num);
+        GTestAdapter::RecordProperty("ref_delta_time_percent", ref_delta_time_percent);
+    }
+
+};
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::MALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::CALLOC, 72,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 1,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 10,
+             1572864, 10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(HbwAllocatorPerformanceTest,
+       test_TC_MEMKIND_HBWMALLOC_ALLOCATOR_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::HBWMALLOC_ALLOCATOR, FunctionCalls::REALLOC, 72,
+             1572864, 10000);
+}
diff --git a/test/hbw_allocator_tests.cpp b/test/hbw_allocator_tests.cpp
new file mode 100644
index 0000000..751f905
--- /dev/null
+++ b/test/hbw_allocator_tests.cpp
@@ -0,0 +1,135 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <hbw_allocator.h>
+
+#include "common.h"
+#include <vector>
+
+// Tests for hbw::allocator class.
+class HbwAllocatorTests: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+};
+
+//Test standard memory allocation and deallocation.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_DetaultAllocatorTest)
+{
+    const size_t size = 512;
+    hbw::allocator<size_t> allocator;
+
+    hbw::allocator<size_t>::pointer ptr = allocator.allocate(size);
+
+    ASSERT_TRUE(NULL != ptr);
+
+    //Do the actually memory writing
+    for (size_t i=0; i<size; i++) {
+        ptr[i] = i;
+    }
+
+    allocator.deallocate(ptr, size);
+}
+
+//Test address conversion functionality.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_AddressConversion)
+{
+    const size_t size = 512;
+    hbw::allocator<int> allocator;
+
+    int *ptr = allocator.allocate(size);
+
+    ASSERT_TRUE(NULL != ptr);
+
+    const int excpected_val = 4;
+    ptr[0] = excpected_val;
+
+    hbw::allocator<int>::reference reference = *ptr;
+    hbw::allocator<int>::const_reference const_reference = *ptr;
+
+    hbw::allocator<int>::pointer test_ptr = allocator.address(reference);
+    hbw::allocator<int>::const_pointer test_const_ptr = allocator.address(
+                                                            const_reference);
+
+    ASSERT_TRUE(NULL != test_ptr);
+    ASSERT_TRUE(NULL != test_const_ptr);
+
+    EXPECT_EQ(excpected_val, test_ptr[0]);
+    EXPECT_EQ(excpected_val, test_const_ptr[0]);
+
+    allocator.deallocate(ptr, size);
+}
+
+//Test for boundaries of allocation sizes.
+//We expect to catch allocation exceptions caused by out of bound sizes.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_AllocationSizeOutOfBounds)
+{
+    hbw::allocator<size_t> allocator;
+
+    const size_t over_size = allocator.max_size() + 1;
+
+    ASSERT_THROW(allocator.allocate(over_size), std::bad_alloc);
+
+    const size_t max_size = -1; //This will give maximum value of size_t.
+
+    ASSERT_THROW(allocator.allocate(max_size), std::bad_alloc);
+
+    ASSERT_THROW(allocator.allocate(0), std::bad_alloc);
+}
+
+//Test if variable will be constructed.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_AllocatorConstruct)
+{
+    hbw::allocator<int> allocator;
+
+    int x = 0;
+    int expect_val = 4;
+    allocator.construct(&x, expect_val);
+
+    EXPECT_EQ(expect_val, x);
+}
+
+//Test the integration with std::vector.
+TEST_F(HbwAllocatorTests, test_TC_MEMKIND_StandardVector)
+{
+    std::vector<int, hbw::allocator<int> > vec;
+    const size_t size = 10000;
+
+    for (size_t i=0; i<size; i++) {
+        vec.push_back(i);
+    }
+
+    EXPECT_EQ(size, vec.size());
+    EXPECT_EQ(1, vec[1]);
+
+    vec.clear();
+
+    EXPECT_EQ((size_t)0, vec.size());
+}
diff --git a/test/hbw_detection_test.py b/test/hbw_detection_test.py
new file mode 100755
index 0000000..5fd34a5
--- /dev/null
+++ b/test/hbw_detection_test.py
@@ -0,0 +1,60 @@
+#
+#  Copyright (C) 2017 - 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+from distutils.spawn import find_executable
+from python_framework import CMD_helper
+import os
+
+class Test_hbw_detection(object):
+    os.environ["PATH"] += os.pathsep + os.path.dirname(os.path.dirname(__file__))
+    binary_path = find_executable("memkind-hbw-nodes")
+    environ_err_test = "../environ_err_hbw_malloc_test"
+    expected_libnuma_warning = "libnuma: Warning: node argument -1 is out of range\n\n"
+    fail_msg = "Test failed with:\n {0}"
+    cmd_helper = CMD_helper()
+
+    def get_hbw_nodes(self, nodemask=None):
+        """ This function executes memkind function 'get_mbind_nodemask' and returns its output - comma-separated HBW nodes """
+        command = self.binary_path
+        if (nodemask):
+            command = "MEMKIND_HBW_NODES={}".format(nodemask) + command
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: Execution of \'{0}\' returns {1}, \noutput: {2}".format(command, retcode, output))
+        print("\nExecution of {} returns output {}".format(command, output))
+        return output
+
+    def test_TC_MEMKIND_hbw_detection_compare_nodemask_default_and_env_variable(self):
+        """ This test checks whether hbw_nodemask_default and hbw_nodemask_env_variable has the same value """
+        hbw_nodemask_default = self.get_hbw_nodes()
+        hbw_nodemask_env_variable = self.get_hbw_nodes(hbw_nodemask_default)
+        assert hbw_nodemask_default == hbw_nodemask_env_variable, self.fail_msg.format("Error: Nodemask hbw_nodemask_default ({0}) " \
+               "is not the same as nodemask hbw_nodemask_env_variable ({1})".format(hbw_nodemask_default, hbw_nodemask_env_variable))
+
+    def test_TC_MEMKIND_hbw_detection_negative_hbw_malloc(self):
+        """ This test sets usupported value of MEMKIND_HBW_NODES, then try to perform a successful allocation from DRAM using hbw_malloc()
+        thanks to default HBW_POLICY_PREFERRED policy """
+        command = "MEMKIND_HBW_NODES=-1 " + self.cmd_helper.get_command_path(self.environ_err_test)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode != 0, self.fail_msg.format("\nError: Execution of: \'{0}\' returns: {1} \noutput: {2}".format(command, retcode, output))
+        assert self.expected_libnuma_warning == output, self.fail_msg.format("Error: expected libnuma warning ({0}) " \
+               "was not found (output: {1})").format(self.expected_libnuma_warning, output)
diff --git a/test/hbw_verify_function_test.cpp b/test/hbw_verify_function_test.cpp
new file mode 100644
index 0000000..9bdd36c
--- /dev/null
+++ b/test/hbw_verify_function_test.cpp
@@ -0,0 +1,396 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "memkind.h"
+#include "memkind/internal/memkind_hbw.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+#include "hbwmalloc.h"
+#include "common.h"
+
+#include <sys/mman.h>
+#include <numaif.h>
+#include <numa.h>
+#include <stdint.h>
+
+#define SHIFT_BYTES(ptr, bytes) ((char*)(ptr)+(bytes))
+
+/*
+ * Set of tests for hbw_verify_memory_region() function,
+ * which intend to check if allocated memory fully fall into high bandwidth memory.
+ * Note: in this tests we are using internal function memkind_hbw_all_get_mbind_nodemask().
+ * In future we intend to rewrite this function when we will have replacement in new API.
+ */
+class HbwVerifyFunctionTest: public :: testing::Test
+{
+protected:
+    const size_t BLOCK_SIZE = 64;
+    const size_t page_size = sysconf(_SC_PAGESIZE);
+    const int flags = MAP_ANONYMOUS | MAP_PRIVATE;
+};
+
+/*
+ * Group of basic tests for different sizes of allocations
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_page_size_not_round)
+{
+    size_t size = page_size * 1024 + 5;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_page_size_round)
+{
+    size_t size = page_size * 1024;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_iterate_1_byte_to_8194_bytes)
+{
+    for (size_t size = 1; size <= (page_size * 2 + 2);
+         size++) { //iterate through 2 pages and 2 bytes
+        char *ptr = (char *) hbw_malloc(size);
+        ASSERT_FALSE(ptr == NULL);
+        EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+        hbw_free(ptr);
+    }
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_ext_5GB)
+{
+    size_t size = 5ull*(1<<30); //5GB - big allocation
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+/*
+ * Test setting memory without HBW_TOUCH_PAGES flag
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_setting_memory_without_flag)
+{
+    size_t size = page_size;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    memset(ptr, '.', size);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, 0), 0);
+    for (size_t i = 0; i < size; i++) { //check that content is unchanged
+        EXPECT_TRUE(ptr[i] == '.');
+    }
+    hbw_free(ptr);
+}
+
+/*
+ * Test HBW_TOUCH_PAGES flag
+ */
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_HBW_TOUCH_PAGES_check_overwritten_content)
+{
+    size_t size = 5 * page_size;
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    memset(ptr, '.', size);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    for (size_t i = 0; i < size;
+         i++) { //check that content of all pages doesn't change
+        EXPECT_TRUE(ptr[i] == '.');
+    }
+    hbw_free(ptr);
+}
+
+/*
+ * Tests check if number of 64-page blocks are working correctly
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_many_blocks_round)
+{
+    size_t size = 16 * (BLOCK_SIZE * page_size); //exactly 16 * 64-page blocks
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_many_blocks_not_round)
+{
+    size_t size = (16 * BLOCK_SIZE * page_size) + (8 *
+                                                   page_size); //16 * 64-page blocks and 1 * 8-paged block
+    char *ptr = (char *) hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    hbw_free(ptr);
+}
+
+/*
+ * Check if 2 blocks and parts of the third block are working correctly.
+ * With this test we can be sure that different sizes of allocation (not only whole blocks) are ok.
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_iterate_not_round)
+{
+    for (size_t i = 1; i < BLOCK_SIZE; i++) {
+        size_t size = (2 * BLOCK_SIZE * page_size) + (i *
+                                                      page_size); // 2 * 64-paged blocks and iterate through whole 3rd 64-paged block
+        char *ptr = (char *) hbw_malloc(size);
+        ASSERT_FALSE(ptr == NULL);
+        EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+        hbw_free(ptr);
+    }
+}
+
+/*
+ * Tests for other kinds and malloc
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_2MBPages_HBW_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    size_t size = 2 * 1024 * 1024 * 10; //10 * 2MB pages
+    char *ptr = (char *) memkind_malloc(MEMKIND_HBW_HUGETLB, size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), 0);
+    memkind_free(MEMKIND_HBW_HUGETLB, ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_DEFAULT)
+{
+    size_t size = page_size * 1024;
+    char *ptr = (char *) memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), -1);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_malloc)
+{
+    size_t size = page_size * 1024;
+    char *ptr = (char *) malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, HBW_TOUCH_PAGES), -1);
+    free(ptr);
+}
+
+/*
+ * Group of negative tests
+ */
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_Negative_size_0_and_SET_MEMORY_flag)
+{
+    void *ptr = hbw_malloc(page_size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, 0, HBW_TOUCH_PAGES), EINVAL);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Negative_size_0_without_flag)
+{
+    void *ptr = hbw_malloc(page_size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, 0, 0), EINVAL);
+    hbw_free(ptr);
+}
+
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_Negative_Uninitialized_Memory_without_flag)
+{
+    void *ptr = NULL;
+    EXPECT_EQ(hbw_verify_memory_region(ptr, page_size * 1024, 0), EINVAL);
+}
+
+TEST_F(HbwVerifyFunctionTest,
+       test_TC_MEMKIND_Negative_Uninitialized_Memory_and_SET_MEMORY_flag)
+{
+    void *ptr = NULL;
+    EXPECT_EQ(hbw_verify_memory_region(ptr, page_size * 1024, HBW_TOUCH_PAGES),
+              EINVAL);
+}
+
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Negative_Without_Memset)
+{
+    size_t size = page_size * 1024;
+    void *ptr = hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size, 0), -1);
+    hbw_free(ptr);
+}
+
+/*
+ * Corner cases: tests for half of pages
+ * + HBM memory
+ * # HBM memory and verified
+ * - not HBM memory, but allocated
+ * = not HBM memory and verified
+ */
+
+/* 3 pages:
+ * |++##|####|##++|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //all pages should fall on HBM
+    mbind(ptr, size, MPOL_BIND, hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //verified are: half of the first page, second page and half of the third page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), 0);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 3 pages:
+ * |####|####|----|
+ * |++##|####|==--|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages_1_and_2_page)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //first and second page should fall on HBM
+    mbind(ptr, size - page_size, MPOL_BIND, hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //check if mbind is successful |####|####|----|
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size - page_size, HBW_TOUCH_PAGES), 0);
+
+    //verified are: half of the first page, second page and half of the third page |++##|####|==--|
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), -1);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 3 pages:
+ * |----|####|####|
+ * |--==|####|##++|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages_2_and_3_page)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //second and third page should fall on HBM
+    mbind(SHIFT_BYTES(ptr, page_size), size - page_size, MPOL_BIND,
+          hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //check if mbind is successful |----|####|####|
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size),
+                                       size - page_size, HBW_TOUCH_PAGES), 0);
+
+    //verified are: half of the first page, second page and half of the third page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), -1);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 3 pages:
+ * |####|----|++++| and  |++++|----|####|
+ * |++##|====|##++|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Half_Pages_1_and_3_page)
+{
+    size_t size = 3 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    //first and third page should fall on HBM
+    mbind(ptr, page_size, MPOL_BIND, hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+    mbind(SHIFT_BYTES(ptr, 2 * page_size), page_size, MPOL_BIND, hbw_nodemask.maskp,
+          NUMA_NUM_NODES, 0);
+
+    //check if mbind is successful |####|----|++++| and  |++++|----|####|
+    EXPECT_EQ(hbw_verify_memory_region(ptr, size - 2*page_size, HBW_TOUCH_PAGES),
+              0);
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, 2*page_size),
+                                       size - 2*page_size, HBW_TOUCH_PAGES), 0);
+
+    //verified are: half of the first page, second page and half of the third page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size/2),
+                                       size - page_size, HBW_TOUCH_PAGES), -1);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 5 pages:
+ * |----|+###|####|###+|----|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_Boundaries_CornerCase)
+{
+    size_t size = 5 * page_size;
+    nodemask_t nodemask;
+    struct bitmask hbw_nodemask = {NUMA_NUM_NODES, nodemask.n};
+    //function memkind_hbw_all_get_mbind_nodemask() has to be rewritten, when we will have replacement in new API.
+    memkind_hbw_all_get_mbind_nodemask(NULL, hbw_nodemask.maskp, hbw_nodemask.size);
+    char *ptr = (char *) mmap(NULL, size, PROT_READ | PROT_WRITE, flags, -1, 0);
+    ASSERT_FALSE(ptr == NULL);
+
+    size_t tested_pages_size = size - 2 * page_size;
+
+    //second, third and fourth page should fall on HBM
+    mbind(SHIFT_BYTES(ptr, page_size), tested_pages_size, MPOL_BIND,
+          hbw_nodemask.maskp, NUMA_NUM_NODES, 0);
+
+    //verified are: second byte of the second page to last byte -1 of fourth page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, page_size + 1),
+                                       tested_pages_size - 2, HBW_TOUCH_PAGES), 0);
+    EXPECT_EQ(munmap(ptr, size), 0);
+}
+
+/* 5 pages:
+ * |-###|####|####|####|###-|
+ */
+TEST_F(HbwVerifyFunctionTest, test_TC_MEMKIND_HBW_partial_verification)
+{
+    size_t size = 5 * page_size;
+
+    //all pages should fall on HBM
+    void *ptr = hbw_malloc(size);
+    ASSERT_FALSE(ptr == NULL);
+
+    //but only second byte of the first page to last byte -1 of the last page are touched
+    memset(SHIFT_BYTES(ptr, 1), 0, size - 2);
+
+    //verified are: second byte of the first page to last byte -1 of last page
+    EXPECT_EQ(hbw_verify_memory_region(SHIFT_BYTES(ptr, 1), size - 2, 0), 0);
+    hbw_free(ptr);
+}
diff --git a/test/heap_manager_init_perf_test.cpp b/test/heap_manager_init_perf_test.cpp
new file mode 100644
index 0000000..a48aece
--- /dev/null
+++ b/test/heap_manager_init_perf_test.cpp
@@ -0,0 +1,167 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include <sstream>
+#include <vector>
+
+#include "common.h"
+#include "allocator_perf_tool/Configuration.hpp"
+#include "allocator_perf_tool/AllocatorFactory.hpp"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+//Test heap managers initialization performance.
+class HeapManagerInitPerfTest: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {
+        //Calculate reference statistics.
+        ref_time = allocator_factory.initialize_allocator(
+                       AllocatorTypes::STANDARD_ALLOCATOR).total_time;
+    }
+
+    void TearDown()
+    {}
+
+    void run_test(unsigned allocator_type)
+    {
+        AllocatorFactory::initialization_stat stat =
+            allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_DEFAULT);
+
+        post_test(stat);
+    }
+
+    void post_test(AllocatorFactory::initialization_stat &stat)
+    {
+        //Calculate (%) distance to the reference time for function calls.
+        stat.ref_delta_time = allocator_factory.calc_ref_delta(ref_time,
+                                                               stat.total_time);
+
+        std::stringstream elapsed_time;
+        elapsed_time << stat.total_time;
+
+        std::stringstream ref_delta_time;
+        ref_delta_time << std::fixed << stat.ref_delta_time << std::endl;
+
+        RecordProperty("elapsed_time", elapsed_time.str());
+        RecordProperty("ref_delta_time_percent_rate", ref_delta_time.str());
+
+        for (int i=0; i<stat.memory_overhead.size(); i++) {
+            std::stringstream node;
+            node << "memory_overhad_node_" << i;
+            std::stringstream memory_overhead;
+            memory_overhead << stat.memory_overhead[i];
+            RecordProperty(node.str(), memory_overhead.str());
+        }
+    }
+
+    AllocatorFactory allocator_factory;
+    float ref_time;
+};
+
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_DEFAULT)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_DEFAULT);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_INTERLEAVE)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_INTERLEAVE);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW_INTERLEAVE)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_INTERLEAVE);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW_PREFERRED)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_PREFERRED);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HUGETLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_GBTLB)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_GBTLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_libinit_HBW_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_HUGETLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest,
+       test_TC_MEMKIND_perf_libinit_HBW_PREFERRED_HUGETLB)
+{
+    HugePageOrganizer huge_page_organizer(16);
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(
+            AllocatorTypes::MEMKIND_HBW_PREFERRED_HUGETLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest, test_TC_MEMKIND_perf_ext_libinit_HBW_GBTLB)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(AllocatorTypes::MEMKIND_HBW_GBTLB);
+    post_test(stat);
+}
+
+TEST_F(HeapManagerInitPerfTest,
+       test_TC_MEMKIND_perf_libinit_HBW_PREFERRED_GBTLB)
+{
+    AllocatorFactory::initialization_stat stat =
+        allocator_factory.initialize_allocator(
+            AllocatorTypes::MEMKIND_HBW_PREFERRED_GBTLB);
+    post_test(stat);
+}
+
diff --git a/test/huge_page_test.cpp b/test/huge_page_test.cpp
new file mode 100644
index 0000000..f6555de
--- /dev/null
+++ b/test/huge_page_test.cpp
@@ -0,0 +1,90 @@
+/*
+* Copyright (C) 2016 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/HugePageUnmap.hpp"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+#include "TimerSysTime.hpp"
+#include "Thread.hpp"
+
+/*
+* This test was created because of the munmap() fail in jemalloc.
+* There are two root causes of the error:
+* - kernel bug (munmap() fails when the size is not aligned)
+* - heap Manager doesn’t provide size aligned to 2MB pages for munmap()
+* Test allocates 2000MB using Huge Pages (50threads*10operations*4MBalloc_size),
+* but it needs extra Huge Pages due to overhead caused by heap management.
+*/
+class  HugePageTest: public :: testing::Test
+{
+protected:
+    void run()
+    {
+        unsigned mem_operations_num = 10;
+        size_t threads_number = 50;
+        bool touch_memory = true;
+        size_t size_1MB = 1024*1024;
+        size_t alignment = 2*size_1MB;
+        size_t alloc_size = 4*size_1MB;
+
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+
+        TimerSysTime timer;
+        timer.start();
+
+        // This bug occurs more frequently under stress of multithreaded allocations.
+        for (int i=0; i<threads_number; i++) {
+            Task *task = new HugePageUnmap(mem_operations_num, touch_memory, alignment,
+                                           alloc_size, HBW_PAGESIZE_2MB);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+        }
+
+        float elapsed_time = timer.getElapsedTime();
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+        threads_manager.release();
+
+        //task release
+        for (int i=0; i<tasks.size(); i++) {
+            delete tasks[i];
+        }
+
+        RecordProperty("threads_number", threads_number);
+        RecordProperty("memory_operations_per_thread", mem_operations_num);
+        RecordProperty("elapsed_time", elapsed_time);
+    }
+};
+
+
+
+// Test passes when there is no crash.
+TEST_F(HugePageTest, test_TC_MEMKIND_ext_UNMAP_HUGE_PAGE)
+{
+    HugePageOrganizer huge_page_organizer(1024);
+    run();
+}
diff --git a/test/load_tbbmalloc_symbols.c b/test/load_tbbmalloc_symbols.c
new file mode 100644
index 0000000..fd186a4
--- /dev/null
+++ b/test/load_tbbmalloc_symbols.c
@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "tbbmalloc.h"
+
+int load_tbbmalloc_symbols()
+{
+    const char so_name[]="libtbbmalloc.so.2";
+    void *tbb_handle = dlopen(so_name, RTLD_LAZY);
+    if(!tbb_handle) {
+        printf("Cannot load %s\n", so_name);
+        return -1;
+    }
+
+    scalable_malloc = dlsym(tbb_handle, "scalable_malloc");
+    if(!scalable_malloc) {
+        printf("Cannot load scalable_malloc symbol from %s\n", so_name);
+        return -1;
+    }
+
+    scalable_realloc = dlsym(tbb_handle, "scalable_realloc");
+    if(!scalable_realloc) {
+        printf("Cannot load scalable_realloc symbol from %s\n", so_name);
+        return -1;
+    }
+
+    scalable_calloc = dlsym(tbb_handle, "scalable_calloc");
+    if(!scalable_calloc) {
+        printf("Cannot load scalable_calloc symbol from %s\n", so_name);
+        return -1;
+    }
+
+    scalable_free = dlsym(tbb_handle, "scalable_free");
+    if(!scalable_free) {
+        printf("Cannot load scalable_free symbol from %s\n", so_name);
+        return -1;
+    }
+
+    return 0;
+}
diff --git a/test/locality_test.cpp b/test/locality_test.cpp
new file mode 100644
index 0000000..65fe49c
--- /dev/null
+++ b/test/locality_test.cpp
@@ -0,0 +1,155 @@
+/*
+* Copyright (C) 2017 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <stdio.h>
+#include <numa.h>
+#include <hbwmalloc.h>
+#include <memkind.h>
+#include <vector>
+#include <memory>
+#include <gtest/gtest.h>
+#include "allocator_perf_tool/Allocation_info.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+
+typedef std::unique_ptr<void, void(*)(void *)> hbw_mem_ptr;
+
+
+class HBWPreferredLocalityTest: public ::testing::Test
+{
+private:
+    int find_closest_node(int node, const std::vector<int> &nodes)
+    {
+        int min_distance = 0;
+        int closest_node = -1;
+        for (int i = 0; i < nodes.size(); i++) {
+            int distance = numa_distance(node, nodes[i]);
+            if (distance && (distance < min_distance || min_distance == 0)) {
+                min_distance = distance;
+                closest_node = nodes[i];
+            }
+        }
+        return closest_node;
+    }
+
+    bool pin_to_cpu(int cpu_id)
+    {
+        cpu_set_t cpu_set;
+        CPU_ZERO(&cpu_set);
+        CPU_SET(cpu_id, &cpu_set);
+        return sched_setaffinity(0, sizeof(cpu_set_t), &cpu_set) != -1;
+    }
+
+    void check_ptr_numa(void *ptr, int expected_numa_id, int cpu_id, size_t size)
+    {
+        memset(ptr, 1, size);
+
+        int numa_id = get_numa_node_id(ptr);
+        EXPECT_EQ(numa_id, expected_numa_id);
+
+        char property_name[50];
+        snprintf(property_name, 50, "actual_numa_for_cpu_%d_expected_numa_%d", cpu_id,
+                 expected_numa_id);
+        GTestAdapter::RecordProperty(property_name, numa_id);
+    }
+
+public:
+    void pin_memory_in_requesting_mem_thread(size_t size,
+                                             const std::vector<int> &cpu_ids,
+                                             const std::vector<int> &mcdram_nodes)
+    {
+#ifdef _OPENMP
+        int threads_num = cpu_ids.size();
+        int ret = hbw_set_policy(HBW_POLICY_PREFERRED);
+        ASSERT_EQ(ret, 0);
+
+        #pragma omp parallel for num_threads(threads_num)
+        for (int i = 0; i < threads_num; i++) {
+            if (!pin_to_cpu(cpu_ids[i])) {
+                ADD_FAILURE();
+                continue;
+            }
+
+            void *internal_ptr = hbw_malloc(size);
+            if (!internal_ptr) {
+                ADD_FAILURE();
+                continue;
+            }
+            hbw_mem_ptr ptr(internal_ptr, hbw_free);
+            int expected_numa_id = find_closest_node(numa_node_of_cpu(cpu_ids[i]),
+                                                     mcdram_nodes);
+            check_ptr_numa(ptr.get(), expected_numa_id, cpu_ids[i], size);
+        }
+#else
+        std::cout << "[ SKIPPED ] Feature OPENMP not supported" << std::endl;
+#endif
+    }
+
+    void pin_memory_in_other_thread_than_requesting_mem(size_t size,
+                                                        const std::vector<int> &cpu_ids,
+                                                        const std::vector<int> &mcdram_nodes)
+    {
+#ifdef _OPENMP
+        int threads_num = cpu_ids.size();
+        int ret = hbw_set_policy(HBW_POLICY_PREFERRED);
+        ASSERT_EQ(ret, 0);
+
+        int main_thread_cpu_id = 0;
+        int expected_numa_id = find_closest_node(main_thread_cpu_id, mcdram_nodes);
+        ASSERT_TRUE(pin_to_cpu(main_thread_cpu_id));
+
+        std::vector<hbw_mem_ptr> ptrs;
+        for (int i = 0; i < threads_num; i++) {
+            void *internal_ptr = hbw_malloc(size);
+            ASSERT_TRUE(internal_ptr);
+            ptrs.emplace_back(internal_ptr, hbw_free);
+        }
+
+        #pragma omp parallel for num_threads(threads_num)
+        for (int i = 0; i < threads_num; i++) {
+            if (!pin_to_cpu(cpu_ids[i])) {
+                ADD_FAILURE();
+                continue;
+            }
+            check_ptr_numa(ptrs[i].get(), expected_numa_id, cpu_ids[i], size);
+        }
+#else
+        std::cout << "[ SKIPPED ] Feature OPENMP not supported" << std::endl;
+#endif
+    }
+};
+
+TEST_F(HBWPreferredLocalityTest,
+       test_TC_MEMKIND_KNL_SNC4_pin_memory_in_requesting_mem_thread_4_threads_100_bytes)
+{
+    pin_memory_in_requesting_mem_thread(100u, std::vector<int> {0, 18, 36, 54},
+                                        std::vector<int> {4, 5, 6, 7});
+}
+
+TEST_F(HBWPreferredLocalityTest,
+       test_TC_MEMKIND_KNL_SNC4_pin_memory_in_other_thread_than_requesting_mem_4_threads_100_bytes)
+{
+    pin_memory_in_other_thread_than_requesting_mem(100u, std::vector<int> {0, 18, 36, 54},
+                                                   std::vector<int> {4, 5, 6, 7});
+}
+
diff --git a/test/main.cpp b/test/main.cpp
new file mode 100644
index 0000000..e1cdd05
--- /dev/null
+++ b/test/main.cpp
@@ -0,0 +1,61 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "common.h"
+
+char *PMEM_DIR = const_cast<char *>("/tmp/");
+
+int main(int argc, char **argv)
+{
+    int opt = 0;
+    struct stat st;
+
+    testing::InitGoogleTest(&argc, argv);
+
+    if (testing::GetGtestHelpFlag()) {
+        printf("\nMemkind options:\n"
+               "-d <directory_path>   change directory on which PMEM kinds\n"
+               "                      are created (default /tmp/)\n");
+        return 0;
+    }
+
+    while ((opt = getopt(argc, argv, "d:")) != -1) {
+        switch (opt) {
+            case 'd':
+                PMEM_DIR = optarg;
+
+                if (stat(PMEM_DIR, &st) != 0 || !S_ISDIR(st.st_mode)) {
+                    printf("%s : Error in getting path status or"
+                           " invalid or non-existent directory\n", PMEM_DIR);
+                    return -1;
+                }
+
+                break;
+            default:
+                return -1;
+        }
+    }
+
+    return RUN_ALL_TESTS();
+}
diff --git a/test/memkind-afts-ext.ts b/test/memkind-afts-ext.ts
new file mode 100644
index 0000000..ea8ed6d
--- /dev/null
+++ b/test/memkind-afts-ext.ts
@@ -0,0 +1,3 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=-PerformanceTest.*
+/usr/share/mpss/test/memkind-dt/decorator_test -a
+/usr/share/mpss/test/memkind-dt/gb_page_tests_bind_policy -a
diff --git a/test/memkind-afts.ts b/test/memkind-afts.ts
new file mode 100644
index 0000000..738741e
--- /dev/null
+++ b/test/memkind-afts.ts
@@ -0,0 +1,4 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=-PerformanceTest.*:*.*ext*
+/usr/share/mpss/test/memkind-dt/decorator_test -a
+/usr/share/mpss/test/memkind-dt/gb_page_tests_bind_policy -a --gtest_filter=-*.*ext*
+/usr/share/mpss/test/memkind-dt/gb_page_tests_bind_policy -a --gtest_filter=-*.*ext*
diff --git a/test/memkind-perf-ext.ts b/test/memkind-perf-ext.ts
new file mode 100644
index 0000000..a9eaf58
--- /dev/null
+++ b/test/memkind-perf-ext.ts
@@ -0,0 +1,2 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=PerformanceTest.*
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=HeapManagerInitPerfTest*:AllocPerformanceTest*
diff --git a/test/memkind-perf.ts b/test/memkind-perf.ts
new file mode 100644
index 0000000..ab3b530
--- /dev/null
+++ b/test/memkind-perf.ts
@@ -0,0 +1,2 @@
+/usr/share/mpss/test/memkind-dt/all_tests -a --gtest_filter=PerformanceTest.*
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=HeapManagerInitPerfTest*:AllocPerformanceTest*:-*.*ext*
diff --git a/test/memkind-pytests.ts b/test/memkind-pytests.ts
new file mode 100644
index 0000000..008c4a9
--- /dev/null
+++ b/test/memkind-pytests.ts
@@ -0,0 +1 @@
+-k "test_TC_MEMKIND_"
diff --git a/test/memkind-slts.ts b/test/memkind-slts.ts
new file mode 100644
index 0000000..9139ed2
--- /dev/null
+++ b/test/memkind-slts.ts
@@ -0,0 +1,2 @@
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=AllocateToMaxStressTests*:-HugePageTest*
+/usr/share/mpss/test/memkind-dt/allocator_perf_tool_tests -a --gtest_filter=HugePageTest* --gtest_repeat=10 --gtest_break_on_failure
diff --git a/test/memkind_allocator_tests.cpp b/test/memkind_allocator_tests.cpp
new file mode 100644
index 0000000..8b08260
--- /dev/null
+++ b/test/memkind_allocator_tests.cpp
@@ -0,0 +1,309 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind_allocator.h"
+#include "common.h"
+#include <vector>
+#include <memory>
+#include <array>
+#include <list>
+#include <map>
+#include <utility>
+#include <string>
+#include <algorithm>
+#include <scoped_allocator>
+#include <thread>
+
+// Tests for memkind::allocator class.
+class MemkindAllocatorTests: public ::testing::Test
+{
+public:
+    libmemkind::static_kind::allocator<int> int_alc_regular { libmemkind::kinds::REGULAR };
+    libmemkind::static_kind::allocator<int> int_alc_default { libmemkind::kinds::DEFAULT };
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+
+//Compare two same kind different type allocators
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindDifferentType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<char> alc2 { int_alc_regular };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+//Compare two same kind same type
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindSameType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+////Compare two different kind different type
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindDifferentType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<char> alc2 { int_alc_default };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+////Compare two different kind same type allocators
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindSameType_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+////Copy assignment test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_CopyAssignment_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc1 != alc2);
+    alc1 = alc2;
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+////Move constructor test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_MoveConstructor_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc2 == alc1);
+    libmemkind::static_kind::allocator<int> alc3 = std::move(alc1);
+    ASSERT_TRUE(alc2 == alc3);
+}
+
+////Move assignment test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_MoveAssignment_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    ASSERT_TRUE(alc1 != alc2);
+
+    {
+        libmemkind::static_kind::allocator<int> alc3 { alc2 };
+        alc1 = std::move(alc3);
+    }
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+////Single allocation-deallocation test
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_Allocator_SingleAllocationDeallocation_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    int *created_object = alc1.allocate(1);
+    alc1.deallocate(created_object, 1);
+}
+
+////Shared cross-allocation-deallocation test
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_Allocator_SharedAllocationDeallocation_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_default };
+    libmemkind::static_kind::allocator<int> alc2 { int_alc_default };
+    int *created_object = nullptr;
+    int *created_object_2 = nullptr;
+    created_object = alc1.allocate(1);
+    ASSERT_TRUE(alc1 == alc2);
+    alc2.deallocate(created_object, 1);
+    created_object = alc2.allocate(1);
+    alc1.deallocate(created_object, 1);
+
+    created_object = alc1.allocate(1);
+    created_object_2 = alc2.allocate(1);
+    alc2.deallocate(created_object, 1);
+    alc1.deallocate(created_object_2, 1);
+}
+
+////Construct-destroy test
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_Allocator_ConstructDestroy_Test)
+{
+    libmemkind::static_kind::allocator<int> alc1 { int_alc_regular };
+    int *created_object = alc1.allocate(1);
+    alc1.construct(created_object);
+    alc1.destroy(created_object);
+}
+
+////Testing thread-safety support
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_Allocator_MultithreadingSupport_Test)
+{
+    const size_t num_threads = 1000;
+    const size_t iteration_count = 5000;
+
+    std::vector<std::thread> thds;
+    for (size_t i = 0; i < num_threads; ++i) {
+
+        thds.push_back(std::thread( [&]() {
+            std::vector<libmemkind::static_kind::allocator<int>> allocators_local;
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.push_back(libmemkind::static_kind::allocator<int>
+                                           ( int_alc_regular ));
+            }
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.pop_back();
+            }
+        }));
+    }
+
+    for (size_t i = 0; i < num_threads; ++i) {
+        thds[i].join();
+    }
+}
+
+////Test multiple memkind allocator usage
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_MultipleAllocatorUsage_Test)
+{
+    {
+        libmemkind::static_kind::allocator<int> alloc { libmemkind::kinds::HBW } ;
+    }
+
+    {
+        libmemkind::static_kind::allocator<int> alloc { libmemkind::kinds::DAX_KMEM } ;
+    }
+}
+
+////Test vector
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Vector_Test)
+{
+    libmemkind::static_kind::allocator<int> alc{ int_alc_default };
+    std::vector<int, libmemkind::static_kind::allocator<int>> vector{ alc };
+
+    const int num = 20;
+
+    for (int i = 0; i < num; ++i) {
+        vector.push_back(0xDEAD + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        ASSERT_TRUE(vector[i] == 0xDEAD + i);
+    }
+}
+
+////Test list
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_AllocatorUsage_List_Test)
+{
+    libmemkind::static_kind::allocator<int> alc{ int_alc_regular };
+
+    std::list<int, libmemkind::static_kind::allocator<int>> list{ alc };
+
+    const int num = 4;
+
+    for (int i = 0; i < num; ++i) {
+        list.emplace_back(0xBEAC011 + i);
+        ASSERT_TRUE(list.back() == 0xBEAC011 + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        list.pop_back();
+    }
+}
+
+////Test map
+TEST_F(MemkindAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Map_Test)
+{
+    typedef std::basic_string<std::pair<const std::string, std::string>, libmemkind::static_kind::allocator<char>>
+            memkind_string;
+    libmemkind::static_kind::allocator<memkind_string> alc{ libmemkind::kinds::REGULAR };
+
+    std::map<std::string, std::string, std::less<std::string>, libmemkind::static_kind::allocator<std::pair<const std::string, std::string>>>
+    map{ std::less<std::string>(), alc };
+
+    const int num = 10;
+
+    for (int i = 0; i < num; ++i) {
+        map[std::to_string(i)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string(i)] == std::to_string(0x0CEA11 + i));
+        map[std::to_string((i * 997 + 83) % 113)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string((i * 997 + 83) % 113)] == std::to_string(
+                        0x0CEA11 + i));
+    }
+}
+
+#if _GLIBCXX_USE_CXX11_ABI
+//Test vector of strings
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorUsage_VectorOfString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::static_kind::allocator<char>>
+            memkind_string;
+    typedef libmemkind::static_kind::allocator<memkind_string> memkind_alloc;
+
+    memkind_alloc alc{ libmemkind::kinds::DEFAULT };
+    libmemkind::static_kind::allocator<char> st_alc{alc};
+
+    std::vector<memkind_string,std::scoped_allocator_adaptor<memkind_alloc>> vec{ alc };
+    memkind_string arg{ "Very very loooong striiiing", st_alc };
+
+    vec.push_back(arg);
+    ASSERT_TRUE(vec.back() == arg);
+}
+
+////Test map of int strings
+TEST_F(MemkindAllocatorTests,
+       test_TC_MEMKIND_AllocatorScopedUsage_MapOfIntString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::static_kind::allocator<char>>
+            memkind_string;
+    typedef int key_t;
+    typedef memkind_string value_t;
+    typedef std::pair<const key_t, value_t> target_pair;
+    typedef libmemkind::static_kind::allocator<target_pair> pmem_alloc;
+    typedef libmemkind::static_kind::allocator<char> str_allocator_t;
+    typedef std::map<key_t, value_t, std::less<key_t>, std::scoped_allocator_adaptor<pmem_alloc>>
+            map_t;
+
+    pmem_alloc map_allocator( libmemkind::kinds::REGULAR );
+
+    str_allocator_t str_allocator( map_allocator );
+
+    value_t source_str1("Lorem ipsum dolor ", str_allocator);
+    value_t source_str2("sit amet consectetuer adipiscing elit", str_allocator );
+
+    map_t target_map{ std::scoped_allocator_adaptor<pmem_alloc>(map_allocator) };
+
+    target_map[key_t(165)] = source_str1;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str1);
+    target_map[key_t(165)] = source_str2;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str2);
+}
+#endif // _GLIBCXX_USE_CXX11_ABI
diff --git a/test/memkind_dax_kmem_test.cpp b/test/memkind_dax_kmem_test.cpp
new file mode 100644
index 0000000..381b389
--- /dev/null
+++ b/test/memkind_dax_kmem_test.cpp
@@ -0,0 +1,447 @@
+/*
+ * Copyright (C) 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <algorithm>
+#include <numaif.h>
+#include <string>
+#include <unistd.h>
+
+#include "common.h"
+#include "dax_kmem_nodes.h"
+#include "proc_stat.h"
+#include "sys/types.h"
+#include "sys/sysinfo.h"
+#include "TestPolicy.hpp"
+
+class MemkindDaxKmemFunctionalTests: public ::testing::Test
+{
+protected:
+    DaxKmem dax_kmem_nodes;
+
+    void SetUp()
+    {
+        if (dax_kmem_nodes.size() < 1) {
+            GTEST_SKIP() << "Minimum 1 PMEM NUMA required." << std::endl;
+        }
+    }
+};
+
+class MemkindDaxKmemTestsParam: public ::testing::Test,
+    public ::testing::WithParamInterface<memkind_t>
+{
+protected:
+    DaxKmem dax_kmem_nodes;
+    memkind_t kind;
+    void SetUp()
+    {
+        kind = GetParam();
+        if (kind == MEMKIND_DAX_KMEM_PREFERRED) {
+            std::set<int> regular_nodes = TestPolicy::get_regular_numa_nodes();
+            for (auto const &node: regular_nodes) {
+                auto closest_dax_kmem_nodes = dax_kmem_nodes.get_closest_numa_nodes(node);
+                if (closest_dax_kmem_nodes.size() > 1)
+                    GTEST_SKIP() << "Skip test for MEMKIND_DAX_KMEM_PREFFERED - "
+                                 "more than one PMEM NUMA node is closest to node: "
+                                 << node << std::endl;
+            }
+        }
+    }
+};
+
+INSTANTIATE_TEST_CASE_P(
+    KindParam, MemkindDaxKmemTestsParam,
+    ::testing::Values(MEMKIND_DAX_KMEM, MEMKIND_DAX_KMEM_ALL,
+                      MEMKIND_DAX_KMEM_PREFERRED));
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_free_with_NULL_kind_4096_bytes)
+{
+    void *ptr = memkind_malloc(kind, 4 * KB);
+    ASSERT_NE(nullptr, ptr);
+    memkind_free(nullptr, ptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_alloc_zero)
+{
+    void *test1 = memkind_malloc(kind, 0);
+    ASSERT_EQ(test1, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_alloc_size_max)
+{
+    errno = 0;
+    void *test1 = memkind_malloc(kind, SIZE_MAX);
+    ASSERT_EQ(test1, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_calloc_zero)
+{
+    void *test = memkind_calloc(kind, 0, 100);
+    ASSERT_EQ(test, nullptr);
+
+    test = memkind_calloc(kind, 100, 0);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_calloc_size_max)
+{
+    const size_t size = SIZE_MAX;
+    const size_t num = 1;
+    errno = 0;
+
+    void *test = memkind_calloc(kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_calloc_num_max)
+{
+    const size_t size = 10;
+    const size_t num = SIZE_MAX;
+    errno = 0;
+
+    void *test = memkind_calloc(kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc)
+{
+    const size_t size1 = 512;
+    const size_t size2 = 1 * KB;
+    char *default_str = nullptr;
+
+    default_str = (char *)memkind_realloc(kind, default_str, size1);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc with size %zu\n", size1);
+    printf("%s", default_str);
+
+    default_str = (char *)memkind_realloc(kind, default_str, size2);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc with size %zu\n", size2);
+    printf("%s", default_str);
+
+    memkind_free(kind, default_str);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_zero)
+{
+    const size_t size = 1 * KB;
+
+    void *test = memkind_malloc(kind, size);
+    ASSERT_NE(test, nullptr);
+
+    void *new_test = memkind_realloc(kind, test, 0);
+    ASSERT_EQ(new_test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_size_max)
+{
+    const size_t size = 1 * KB;
+
+    void *test = memkind_malloc(kind, size);
+    ASSERT_NE(test, nullptr);
+
+    errno = 0;
+    void *new_test = memkind_realloc(kind, test, SIZE_MAX);
+    ASSERT_EQ(new_test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+
+    memkind_free(kind, test);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_size_zero)
+{
+    const size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+    const size_t iteration = 100;
+
+    for (unsigned i = 0; i < iteration; ++i) {
+        test = memkind_malloc(kind, size);
+        ASSERT_NE(test, nullptr);
+        errno = 0;
+        new_test = memkind_realloc(kind, test, 0);
+        ASSERT_EQ(new_test, nullptr);
+        ASSERT_EQ(errno, 0);
+    }
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_nullptr)
+{
+    const size_t size = 1 * KB;
+    void *test = nullptr;
+
+    test = memkind_realloc(kind, test, size);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(kind, test);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_realloc_increase_decrease_size)
+{
+    size_t size = 1 * KB;
+    const char val[] = "test_TC_MEMKIND_PmemReallocIncreaseSizeNullKindVariant";
+    int status;
+
+    char *test1 = (char *)memkind_malloc(kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 2 * KB;
+    char *test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    size = 512;
+    char *test3 = (char *)memkind_realloc(nullptr, test2, size);
+    ASSERT_NE(test3, nullptr);
+    status = memcmp(val, test3, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(kind, test3);
+}
+
+TEST_P(MemkindDaxKmemTestsParam, test_TC_MEMKIND_MEMKIND_DAX_KMEM_free_nullptr)
+{
+    const int n_iter = 5000;
+
+    for (int i = 0; i < n_iter; ++i) {
+        memkind_free(kind, nullptr);
+    }
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_posix_memalign_alignment_less_than_void)
+{
+    void *test = nullptr;
+    const size_t size = 32;
+    const size_t wrong_alignment = 4;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_posix_memalign_alignment_not_pow2)
+{
+    void *test = nullptr;
+    const size_t size = 32;
+    const size_t wrong_alignment = sizeof(void *)+1;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_P(MemkindDaxKmemTestsParam,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_posix_memalign_correct_alignment)
+{
+    void *test = nullptr;
+    const size_t size = 32;
+    const size_t alignment = sizeof(void *);
+
+    errno = 0;
+    int ret = memkind_posix_memalign(kind, &test, alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, 0);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(nullptr, test);
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_alloc_until_full_numa)
+{
+    ProcStat stat;
+    const size_t alloc_size = 100 * MB;
+    std::set<void *> allocations;
+    size_t numa_size;
+    int numa_id = -1;
+    const int n_swap_alloc = 20;
+
+    void *ptr = memkind_malloc(MEMKIND_DAX_KMEM, alloc_size);
+    ASSERT_NE(nullptr, ptr);
+    memset(ptr, 'a', alloc_size);
+    allocations.insert(ptr);
+
+    get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+    numa_size = numa_node_size64(numa_id, nullptr);
+
+    while (numa_size > alloc_size * allocations.size()) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    size_t init_swap = stat.get_used_swap_space_size_bytes();
+    for(int i = 0; i < n_swap_alloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    ASSERT_GE(stat.get_used_swap_space_size_bytes(), init_swap);
+
+    for (auto const &ptr: allocations) {
+        memkind_free(MEMKIND_DAX_KMEM, ptr);
+    }
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_ALL_alloc_until_full_numa)
+{
+    if (dax_kmem_nodes.size() < 2)
+        GTEST_SKIP() << "This test requires minimum 2 kmem dax nodes";
+
+    ProcStat stat;
+    void *ptr;
+    const size_t alloc_size = 100 * MB;
+    std::set<void *> allocations;
+    size_t sum_of_dax_kmem_free_space = dax_kmem_nodes.get_free_space();
+    int numa_id = -1;
+    const int n_swap_alloc = 20;
+
+    while (sum_of_dax_kmem_free_space > alloc_size * allocations.size()) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_ALL, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+
+        get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+        ASSERT_TRUE(dax_kmem_nodes.contains(numa_id));
+    }
+
+    size_t init_swap = stat.get_used_swap_space_size_bytes();
+    for(int i = 0; i < n_swap_alloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_ALL, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    ASSERT_GE(stat.get_used_swap_space_size_bytes(), init_swap);
+
+    for (auto const &ptr: allocations) {
+        memkind_free(MEMKIND_DAX_KMEM_ALL, ptr);
+    }
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_PREFFERED_alloc_until_full_numa)
+{
+    std::set<int> regular_nodes = TestPolicy::get_regular_numa_nodes();
+    for (auto const &node: regular_nodes) {
+        auto closest_dax_kmem_nodes = dax_kmem_nodes.get_closest_numa_nodes(node);
+        if (closest_dax_kmem_nodes.size() > 1)
+            GTEST_SKIP() << "Skip test for MEMKIND_DAX_KMEM_PREFFERED - "
+                         "more than one PMEM NUMA node is closest to node: "
+                         << node << std::endl;
+    }
+
+    ProcStat stat;
+    const size_t alloc_size = 100 * MB;
+    std::set<void *> allocations;
+    size_t numa_size;
+    int numa_id = -1;
+    const int n_swap_alloc = 20;
+
+    void *ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+    ASSERT_NE(nullptr, ptr);
+    memset(ptr, 'a', alloc_size);
+    allocations.insert(ptr);
+
+    get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+    int process_cpu = sched_getcpu();
+    int process_node = numa_node_of_cpu(process_cpu);
+    std::set<int> closest_numa_ids = dax_kmem_nodes.get_closest_numa_nodes(
+                                         process_node);
+    numa_size = numa_node_size64(numa_id, nullptr);
+
+    while (0.99 * numa_size > alloc_size * allocations.size()) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+
+        get_mempolicy(&numa_id, nullptr, 0, ptr, MPOL_F_NODE | MPOL_F_ADDR);
+        ASSERT_TRUE(closest_numa_ids.find(numa_id) != closest_numa_ids.end());
+    }
+
+    for (int i = 0; i < n_swap_alloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+        ASSERT_NE(nullptr, ptr);
+        memset(ptr, 'a', alloc_size);
+        allocations.insert(ptr);
+    }
+
+    ASSERT_EQ(stat.get_used_swap_space_size_bytes(), 0U);
+
+    for (auto const &ptr: allocations) {
+        memkind_free(MEMKIND_DAX_KMEM_ALL, ptr);
+    }
+}
+
+TEST_F(MemkindDaxKmemFunctionalTests,
+       test_TC_MEMKIND_MEMKIND_DAX_KMEM_PREFFERED_check_prerequisities)
+{
+    bool can_run = false;
+    std::set<int> regular_nodes = TestPolicy::get_regular_numa_nodes();
+
+    for (auto const &node: regular_nodes) {
+        auto closest_dax_kmem_nodes = dax_kmem_nodes.get_closest_numa_nodes(node);
+        if (closest_dax_kmem_nodes.size() > 1)
+            can_run = true;
+    }
+
+    if (!can_run)
+        GTEST_SKIP() <<
+                     "Skip test for MEMKIND_DAX_KMEM_PREFFERED checking prerequisities - "
+                     "none of the nodes have more than one closest PMEM NUMA nodes";
+
+    const size_t alloc_size = 100 * MB;
+    void *ptr = memkind_malloc(MEMKIND_DAX_KMEM_PREFERRED, alloc_size);
+    ASSERT_EQ(nullptr, ptr);
+}
diff --git a/test/memkind_defrag_reallocate.cpp b/test/memkind_defrag_reallocate.cpp
new file mode 100644
index 0000000..2b685e8
--- /dev/null
+++ b/test/memkind_defrag_reallocate.cpp
@@ -0,0 +1,140 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "common.h"
+#include "vector"
+#include <stdio.h>
+
+class MemkindDefragReallocateTests: public ::testing::Test
+{
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+TEST_F(MemkindDefragReallocateTests, test_TC_MEMKIND_NULLTransfer)
+{
+    void *ptr = memkind_defrag_reallocate(nullptr, nullptr);
+    ASSERT_EQ(ptr, nullptr);
+    ptr = memkind_defrag_reallocate(MEMKIND_REGULAR, nullptr);
+    ASSERT_EQ(ptr, nullptr);
+}
+
+TEST_F(MemkindDefragReallocateTests, test_TC_MEMKIND_Negative_test)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 512);
+    ASSERT_NE(ptr, nullptr);
+    void *new_ptr = memkind_defrag_reallocate(MEMKIND_DEFAULT, ptr);
+    ASSERT_EQ(new_ptr, nullptr);
+}
+
+TEST_F(MemkindDefragReallocateTests,
+       test_TC_MEMKIND_TCACHE_Default_kind_success)
+{
+    std::vector<void *> alloc_vec;
+    const size_t number_of_malloc = 10000000;
+    alloc_vec.reserve(2*number_of_malloc);
+    size_t count_mem_transfer = 0;
+    void *ptr;
+    size_t i;
+
+    for (i = 0; i < number_of_malloc; ++i) {
+        ptr = memkind_malloc(MEMKIND_DEFAULT, 150);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 150);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < number_of_malloc; ++i) {
+        if (i % 2 == 0) {
+            memkind_free(MEMKIND_DEFAULT,alloc_vec.at(i));
+            alloc_vec.at(i) = nullptr;
+        }
+        ptr = memkind_malloc(MEMKIND_DEFAULT, 300);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 300);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < alloc_vec.size(); ++i) {
+        void *new_ptr = memkind_defrag_reallocate(MEMKIND_DEFAULT, alloc_vec.at(i));
+        if (new_ptr) {
+            alloc_vec.at(i) = new_ptr;
+            count_mem_transfer++;
+        }
+    }
+
+    ASSERT_NE(count_mem_transfer, 0U);
+
+    for(auto const &val: alloc_vec) {
+        memkind_free(MEMKIND_DEFAULT, val);
+    }
+}
+
+TEST_F(MemkindDefragReallocateTests,
+       test_TC_MEMKIND_TCACHE_REGULAR_kind_success)
+{
+    std::vector<void *> alloc_vec;
+    const size_t number_of_malloc = 10000000;
+    alloc_vec.reserve(2*number_of_malloc);
+    size_t count_mem_transfer = 0;
+    void *ptr;
+    size_t i;
+
+    for (i = 0; i < number_of_malloc ; ++i) {
+        ptr = memkind_malloc(MEMKIND_REGULAR, 150);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 150);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < number_of_malloc; ++i) {
+        if (i % 2 == 0) {
+            memkind_free(MEMKIND_REGULAR,alloc_vec.at(i));
+            alloc_vec.at(i) = nullptr;
+        }
+        ptr = memkind_malloc(MEMKIND_REGULAR, 300);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 300);
+        alloc_vec.push_back(ptr);
+    }
+
+    for (i = 0; i < alloc_vec.size(); ++i) {
+        void *new_ptr = memkind_defrag_reallocate(nullptr, alloc_vec.at(i));
+        if (new_ptr) {
+            alloc_vec.at(i) = new_ptr;
+            count_mem_transfer++;
+        }
+    }
+
+    ASSERT_NE(count_mem_transfer, 0U);
+
+    for(auto const &val: alloc_vec) {
+        memkind_free(MEMKIND_REGULAR, val);
+    }
+}
diff --git a/test/memkind_detect_kind_tests.cpp b/test/memkind_detect_kind_tests.cpp
new file mode 100644
index 0000000..e2fdb43
--- /dev/null
+++ b/test/memkind_detect_kind_tests.cpp
@@ -0,0 +1,145 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "common.h"
+
+extern const char *PMEM_DIR;
+
+class MemkindDetectKindTests: public ::testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindNullPtr)
+{
+    struct memkind *temp_kind = memkind_detect_kind(nullptr);
+    ASSERT_EQ(nullptr, temp_kind);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindPmemKind)
+{
+    memkind_t pmem_kind_temp = nullptr;
+    int err = memkind_create_pmem(PMEM_DIR, 0, &pmem_kind_temp);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_temp);
+
+    void *ptr = memkind_malloc(pmem_kind_temp, 512);
+    ASSERT_NE(nullptr, ptr);
+
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(pmem_kind_temp, temp_kind);
+
+    memkind_free(pmem_kind_temp, ptr);
+
+    err = memkind_destroy_kind(pmem_kind_temp);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindRegularKind)
+{
+    void *ptr = memkind_malloc(MEMKIND_REGULAR, 512);
+    ASSERT_NE(nullptr, ptr);
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(MEMKIND_REGULAR, temp_kind);
+    memkind_free(MEMKIND_REGULAR, ptr);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectDefaultSmallSizeKind)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 512);
+    ASSERT_NE(nullptr, ptr);
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(MEMKIND_DEFAULT, temp_kind);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectDefaultBigSizeKind)
+{
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 8 * MB);
+    ASSERT_NE(nullptr, ptr);
+    struct memkind *temp_kind = memkind_detect_kind(ptr);
+    ASSERT_EQ(MEMKIND_DEFAULT, temp_kind);
+    memkind_free(MEMKIND_DEFAULT, ptr);
+}
+
+TEST_F(MemkindDetectKindTests, test_TC_MEMKIND_DetectKindMixKind)
+{
+    const size_t alloc_size = 512;
+    memkind_t pmem_kind_temp_1 = nullptr;
+    memkind_t pmem_kind_temp_2 = nullptr;
+    struct memkind *temp_kind = nullptr;
+    void *ptr_pmem_1 = nullptr;
+    void *ptr_pmem_2 = nullptr;
+    void *ptr_regular = nullptr;
+    void *ptr_default = nullptr;
+
+    int err = memkind_create_pmem(PMEM_DIR, 0, &pmem_kind_temp_1);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_temp_1);
+
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_kind_temp_2);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_temp_2);
+
+    ptr_pmem_1 = memkind_malloc(pmem_kind_temp_1, alloc_size);
+    ASSERT_NE(nullptr, ptr_pmem_1);
+
+    ptr_pmem_2 = memkind_malloc(pmem_kind_temp_2, alloc_size);
+    ASSERT_NE(nullptr, ptr_pmem_2);
+
+    ptr_default = memkind_malloc(MEMKIND_DEFAULT, alloc_size);
+    ASSERT_NE(nullptr, ptr_default);
+
+    ptr_regular = memkind_malloc(MEMKIND_REGULAR, alloc_size);
+    ASSERT_NE(nullptr, ptr_regular);
+
+    temp_kind = memkind_detect_kind(ptr_default);
+    ASSERT_EQ(MEMKIND_DEFAULT, temp_kind);
+
+    temp_kind = memkind_detect_kind(ptr_pmem_1);
+    ASSERT_EQ(pmem_kind_temp_1, temp_kind);
+
+    temp_kind = memkind_detect_kind(ptr_regular);
+    ASSERT_EQ(MEMKIND_REGULAR, temp_kind);
+
+    temp_kind = memkind_detect_kind(ptr_pmem_2);
+    ASSERT_EQ(pmem_kind_temp_2, temp_kind);
+
+    memkind_free(pmem_kind_temp_2, ptr_pmem_2);
+    memkind_free(pmem_kind_temp_1, ptr_pmem_1);
+    memkind_free(MEMKIND_DEFAULT, ptr_default);
+    memkind_free(MEMKIND_REGULAR, ptr_regular);
+
+    err = memkind_destroy_kind(pmem_kind_temp_2);
+    ASSERT_EQ(0, err);
+    err = memkind_destroy_kind(pmem_kind_temp_1);
+    ASSERT_EQ(0, err);
+}
diff --git a/test/memkind_null_kind_test.cpp b/test/memkind_null_kind_test.cpp
new file mode 100644
index 0000000..7d7d052
--- /dev/null
+++ b/test/memkind_null_kind_test.cpp
@@ -0,0 +1,229 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include "allocator_perf_tool/TimerSysTime.hpp"
+#include "common.h"
+
+class MemkindNullKindTests: public ::testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultKindFreeNullPtr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_free(MEMKIND_DEFAULT, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_NullKindFreeNullPtr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_free(nullptr, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_NullKindUsableSizeNullPtr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_malloc_usable_size(nullptr, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultRegularKindFreeNullPtr)
+{
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = 1 * MB;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+    }
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultReallocNullptrSizeZero)
+{
+    const double test_time = 5;
+    void *test_nullptr = nullptr;
+    void *test_ptr_malloc = nullptr;
+    void *test_ptr_realloc = nullptr;
+    TimerSysTime timer;
+    timer.start();
+    do {
+        test_ptr_malloc = memkind_malloc(MEMKIND_DEFAULT, 5 * MB);
+        errno = 0;
+        test_ptr_realloc = memkind_realloc(MEMKIND_DEFAULT, test_ptr_malloc, 0);
+        ASSERT_EQ(test_ptr_realloc, nullptr);
+        ASSERT_EQ(errno, 0);
+
+        errno = 0;
+        //equivalent to memkind_malloc(MEMKIND_DEFAULT,0)
+        test_nullptr = memkind_realloc(MEMKIND_DEFAULT, nullptr, 0);
+        ASSERT_EQ(test_nullptr, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests,
+       test_TC_MEMKIND_DefaultKindReallocNullKindSizeZeroImplicitFree)
+{
+    const double test_time = 5;
+    size_t size = 1 * KB;
+    void *test_ptr_malloc = nullptr;
+    void *test_ptr_realloc = nullptr;
+    TimerSysTime timer;
+    timer.start();
+    do {
+        errno = 0;
+        test_ptr_malloc = memkind_malloc(MEMKIND_DEFAULT, size);
+        ASSERT_NE(test_ptr_malloc, nullptr);
+        ASSERT_EQ(errno, 0);
+
+        errno = 0;
+        test_ptr_realloc = memkind_realloc(nullptr, test_ptr_malloc, 0);
+        ASSERT_EQ(test_ptr_realloc, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_DefaultKindReallocNullKindSizeMax)
+{
+    size_t size = 1 * KB;
+    void *test_ptr_malloc = nullptr;
+    void *test_ptr_realloc = nullptr;
+
+    test_ptr_malloc = memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_NE(test_ptr_malloc, nullptr);
+
+    errno = 0;
+    test_ptr_realloc = memkind_realloc(nullptr, test_ptr_malloc, SIZE_MAX);
+    ASSERT_EQ(test_ptr_realloc, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+
+    memkind_free(nullptr, test_ptr_malloc);
+}
+
+TEST_F(MemkindNullKindTests,
+       test_TC_MEMKIND_DefaultReallocIncreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_DefaultReallocIncreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size *= 2;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(MEMKIND_DEFAULT, test2);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_ReallocNullptrNullKind)
+{
+    size_t size = 1 * KB;
+
+    errno = 0;
+    void *test = memkind_realloc(nullptr, nullptr, size);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, EINVAL);
+}
+
+TEST_F(MemkindNullKindTests,
+       test_TC_MEMKIND_DefaultReallocDecreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_DefaultReallocDecreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(MEMKIND_DEFAULT, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 4;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, size);
+    ASSERT_EQ(status, 0);
+
+    memkind_free(MEMKIND_DEFAULT, test2);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_ReallocNullptrNullKindSizeMax)
+{
+    errno = 0;
+    void *test = memkind_realloc(nullptr, nullptr, SIZE_MAX);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, EINVAL);
+}
+
+TEST_F(MemkindNullKindTests, test_TC_MEMKIND_ReallocNullptrNullKindSizeZero)
+{
+    errno = 0;
+    void *test = memkind_realloc(nullptr, nullptr, 0);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, EINVAL);
+}
diff --git a/test/memkind_pmem_config_tests.cpp b/test/memkind_pmem_config_tests.cpp
new file mode 100644
index 0000000..4657dfe
--- /dev/null
+++ b/test/memkind_pmem_config_tests.cpp
@@ -0,0 +1,173 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "include/memkind/internal/memkind_private.h"
+#include "include/memkind/internal/memkind_pmem.h"
+
+#include "common.h"
+
+#include <sys/statfs.h>
+
+extern const char  *PMEM_DIR;
+
+class MemkindConfigTests: public ::testing::Test
+{
+
+protected:
+    struct memkind_config *global_test_cfg;
+    void SetUp()
+    {
+        global_test_cfg = memkind_config_new();
+        ASSERT_NE(nullptr, global_test_cfg);
+    }
+
+    void TearDown()
+    {
+        memkind_config_delete(global_test_cfg);
+    }
+};
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_CreateAndDestroyEmptyConfig)
+{
+    struct memkind_config *tmp_cfg = memkind_config_new();
+    ASSERT_NE(nullptr, tmp_cfg);
+    memkind_config_delete(tmp_cfg);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemConSetConfigSizeZero)
+{
+    memkind_config_set_size(global_test_cfg, 0U);
+    ASSERT_EQ(global_test_cfg->pmem_size, 0U);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemSetConfigSizeNotZero)
+{
+    memkind_config_set_size(global_test_cfg, 100 * MB);
+    ASSERT_EQ(global_test_cfg->pmem_size, 100 * MB);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemSetConfigPath)
+{
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+
+    ASSERT_STREQ(global_test_cfg->pmem_dir, PMEM_DIR);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_PmemSetDefaultMemoryUsagePolicy)
+{
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+
+    ASSERT_EQ(global_test_cfg->policy, MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+}
+
+TEST_F(MemkindConfigTests, test_TC_MEMKIND_SetConservativeMemoryUsagePolicy)
+{
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+
+    ASSERT_EQ(global_test_cfg->policy, MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsFailWrongDirectory)
+{
+    memkind_t pmem_kind = nullptr;
+
+    memkind_config_set_path(global_test_cfg, "/temp/non_exisitng_directory");
+    memkind_config_set_size(global_test_cfg, 0U);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+
+    int err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsFailWrongSize)
+{
+    memkind_t pmem_kind = nullptr;
+
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+    memkind_config_set_size(global_test_cfg, MEMKIND_PMEM_MIN_SIZE -1);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_DEFAULT);
+
+    int err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsFailWrongPolicy)
+{
+    memkind_t pmem_kind = nullptr;
+
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+    memkind_config_set_size(global_test_cfg, MEMKIND_PMEM_MIN_SIZE -1);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_MAX_VALUE);
+
+    int err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(MemkindConfigTests,
+       test_TC_MEMKIND_PmemCreatePmemWithParamsSuccessCheckConservativePolicy)
+{
+    memkind_t pmem_kind = nullptr;
+    int err = 0;
+    double blocksBeforeFree = 0;
+    double blocksAfterFree = 0;
+    struct memkind_pmem *priv = nullptr;
+    void *temp_ptr = nullptr;
+    struct stat st;
+
+    memkind_config_set_path(global_test_cfg, PMEM_DIR);
+    memkind_config_set_size(global_test_cfg, 0U);
+    memkind_config_set_memory_usage_policy(global_test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+    err = memkind_create_pmem_with_config(global_test_cfg, &pmem_kind);
+    ASSERT_EQ(err, 0);
+
+    priv = static_cast<memkind_pmem *>(pmem_kind->priv);
+
+    temp_ptr = memkind_malloc(pmem_kind, 1 * KB);
+
+    err = fstat(priv->fd, &st);
+    ASSERT_EQ(err, 0);
+    blocksBeforeFree = st.st_blocks;
+
+    memkind_free(pmem_kind, temp_ptr);
+
+    err = fstat(priv->fd, &st);
+    ASSERT_EQ(err, 0);
+    blocksAfterFree = st.st_blocks;
+
+    //if blocksAfterFree is less than blocksBeforeFree, extent was called.
+    ASSERT_LT(blocksAfterFree, blocksBeforeFree);
+
+    err = memkind_destroy_kind(pmem_kind);
+    ASSERT_EQ(err, 0);
+}
diff --git a/test/memkind_pmem_long_time_tests.cpp b/test/memkind_pmem_long_time_tests.cpp
new file mode 100644
index 0000000..ce7cd49
--- /dev/null
+++ b/test/memkind_pmem_long_time_tests.cpp
@@ -0,0 +1,120 @@
+/*
+ * Copyright (C) 2018 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+#include "allocator_perf_tool/TimerSysTime.hpp"
+#include "common.h"
+
+#define STRESS_TIME (3*24*60*60)
+
+extern const char  *PMEM_DIR;
+
+static const size_t small_size[] = {8, 16, 32, 48, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384,
+                                    448, 512, 640, 768, 896, 1 * KB, 1280, 1536, 1792, 2 * KB, 2560, 3 * KB,
+                                    3584, 4 * KB, 5 * KB, 6 * KB, 7 * KB, 8 * KB, 10 * KB, 12 * KB, 14 * KB
+                                   };
+
+static const size_t large_size[] = {16 * KB, 32 * KB, 20 * KB, 24 * KB, 28 * KB, 32 * KB, 40 * KB, 48 * KB,
+                                    56 * KB, 64 * KB, 80 * KB, 96 * KB, 112 * KB, 128 * KB, 160 * KB, 192 * KB,
+                                    224 * KB, 256 * KB, 320 * KB, 384 * KB, 448 * KB, 512 * KB, 640 * KB,
+                                    768 * KB, 896 * KB, 1 * MB, 1280 * KB, 1536 * KB, 1792 * KB, 2 * MB,
+                                    2560 * KB, 3 * MB, 3584 * KB, 4 * MB, 5 * MB, 6 * MB, 7 * MB, 8 * MB
+                                   };
+
+class MemkindPmemLongTimeStress: public :: testing::Test
+{
+
+protected:
+    memkind_t pmem_kind;
+    void SetUp()
+    {
+        int err = memkind_create_pmem(PMEM_DIR, 0, &pmem_kind);
+        ASSERT_EQ(0, err);
+        ASSERT_NE(nullptr, pmem_kind);
+    }
+
+    void TearDown()
+    {
+        int err = memkind_destroy_kind(pmem_kind);
+        ASSERT_EQ(0, err);
+    }
+};
+
+TEST_F(MemkindPmemLongTimeStress, DISABLED_test_TC_MEMKIND_PmemStressSmallSize)
+{
+    void *test = nullptr;
+    TimerSysTime timer;
+    timer.start();
+
+    do {
+        for (size_t i = 0; i < ARRAY_SIZE(small_size); i++) {
+            test = memkind_malloc(pmem_kind, small_size[i]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+        }
+    } while (timer.getElapsedTime() < STRESS_TIME);
+}
+
+TEST_F(MemkindPmemLongTimeStress, DISABLED_test_TC_MEMKIND_PmemStressLargeSize)
+{
+    void *test = nullptr;
+    TimerSysTime timer;
+    timer.start();
+
+    do {
+        for (size_t i = 0; i < ARRAY_SIZE(large_size); i++) {
+            test = memkind_malloc(pmem_kind, large_size[i]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+        }
+    } while (timer.getElapsedTime() < STRESS_TIME);
+}
+
+TEST_F(MemkindPmemLongTimeStress,
+       DISABLED_test_TC_MEMKIND_PmemStressSmallAndLargeSize)
+{
+    void *test = nullptr;
+    size_t i = 0, j = 0;
+    TimerSysTime timer;
+    timer.start();
+
+    do {
+        if (i < ARRAY_SIZE(small_size)) {
+            test = memkind_malloc(pmem_kind, small_size[i]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+            i++;
+        } else
+            i = 0;
+
+        if (j < ARRAY_SIZE(large_size)) {
+            test = memkind_malloc(pmem_kind, large_size[j]);
+            ASSERT_NE(test, nullptr);
+            memkind_free(pmem_kind, test);
+            j++;
+        } else
+            j = 0;
+
+    } while (timer.getElapsedTime() < STRESS_TIME);
+}
diff --git a/test/memkind_pmem_tests.cpp b/test/memkind_pmem_tests.cpp
new file mode 100644
index 0000000..f51f5f5
--- /dev/null
+++ b/test/memkind_pmem_tests.cpp
@@ -0,0 +1,1695 @@
+/*
+ * Copyright (C) 2015 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_pmem.h>
+#include <memkind/internal/memkind_private.h>
+#include "allocator_perf_tool/TimerSysTime.hpp"
+
+#include <sys/param.h>
+#include <sys/mman.h>
+#include <sys/statfs.h>
+#include <stdio.h>
+#include <pthread.h>
+#include <vector>
+#include "common.h"
+
+static const size_t PMEM_PART_SIZE = MEMKIND_PMEM_MIN_SIZE + 4 * KB;
+static const size_t PMEM_NO_LIMIT = 0;
+extern const char  *PMEM_DIR;
+static pthread_mutex_t mutex = PTHREAD_MUTEX_INITIALIZER;
+static pthread_cond_t cond = PTHREAD_COND_INITIALIZER;
+
+class MemkindPmemTests: public ::testing::Test
+{
+
+protected:
+    memkind_t pmem_kind;
+    void SetUp()
+    {
+        // create PMEM partition
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind);
+        ASSERT_EQ(0, err);
+        ASSERT_NE(pmem_kind, nullptr);
+    }
+
+    void TearDown()
+    {
+        int err = memkind_destroy_kind(pmem_kind);
+        ASSERT_EQ(0, err);
+    }
+};
+
+class MemkindPmemTestsCalloc : public MemkindPmemTests,
+    public ::testing::WithParamInterface<std::tuple<int, int>>
+{
+};
+
+class MemkindPmemTestsMalloc : public ::testing::Test,
+    public ::testing::WithParamInterface<size_t>
+{
+};
+
+static void pmem_get_size(struct memkind *kind, size_t &total, size_t &free)
+{
+    struct memkind_pmem *priv = reinterpret_cast<struct memkind_pmem *>(kind->priv);
+
+    total = priv->max_size;
+    free = priv->max_size - priv->current_size; /* rough estimation */
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPriv)
+{
+    size_t total_mem = 0;
+    size_t free_mem = 0;
+
+    pmem_get_size(pmem_kind, total_mem, free_mem);
+
+    ASSERT_NE(total_mem, 0U);
+    ASSERT_NE(free_mem, 0U);
+
+    ASSERT_EQ(total_mem, roundup(PMEM_PART_SIZE, MEMKIND_PMEM_CHUNK_SIZE));
+
+    size_t offset = total_mem - free_mem;
+    ASSERT_LT(offset, MEMKIND_PMEM_CHUNK_SIZE);
+    ASSERT_LT(offset, total_mem);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCreatePmemFailNonExistingDirectory)
+{
+    const char *non_existing_directory = "/temp/non_exisitng_directory";
+    struct memkind *pmem_temp = nullptr;
+    errno = 0;
+    int err = memkind_create_pmem(non_existing_directory, MEMKIND_PMEM_MIN_SIZE,
+                                  &pmem_temp);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    ASSERT_EQ(nullptr, pmem_temp);
+    ASSERT_EQ(ENOENT, errno);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocFragmentation)
+{
+    const size_t size_array[] = {
+        10,
+        100,
+        200,
+        500,
+        1000,
+        2000,
+        3000,
+        1 * MB,
+        2 * MB,
+        3 * MB,
+        4 * MB
+    };
+
+    const size_t iteration = 1000;
+
+    struct memkind *pmem_temp = nullptr;
+
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_temp);
+
+    //check malloc-free possibility with same sizes and 1000 iterations
+    for ( unsigned j = 0; j < iteration; ++j) {
+        for (unsigned i = 0 ; i < ARRAY_SIZE(size_array); ++i) {
+            void *alloc = memkind_malloc(pmem_temp, size_array[i]);
+            ASSERT_NE(nullptr, alloc);
+            memkind_free(pmem_temp, alloc);
+            alloc = nullptr;
+        }
+    }
+    err = memkind_destroy_kind(pmem_temp);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCreatePmemFailWritePermissionIssue)
+{
+    struct stat path_stat;
+    struct memkind *pmem_temp = nullptr;
+    char temp_dir[] = "/tmp/tmpdir.XXXXXX";
+
+    char *dir_name = mkdtemp(temp_dir);
+    ASSERT_NE(nullptr, dir_name);
+
+    int err = stat(dir_name, &path_stat);
+    ASSERT_EQ(0, err);
+
+    err = chmod(dir_name, path_stat.st_mode & ~S_IWUSR);
+    ASSERT_EQ(0, err);
+
+    errno = 0;
+    err = memkind_create_pmem(dir_name, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    ASSERT_EQ(nullptr, pmem_temp);
+    ASSERT_EQ(EACCES, errno);
+
+    err = rmdir(temp_dir);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMalloc)
+{
+    const size_t size = 1 * KB;
+    char *default_str = nullptr;
+
+    default_str = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_malloc MEMKIND_PMEM\n");
+    printf("%s", default_str);
+
+    memkind_free(pmem_kind, default_str);
+
+    // Out of memory
+    default_str = (char *)memkind_malloc(pmem_kind, 2 * PMEM_PART_SIZE);
+    ASSERT_EQ(nullptr, default_str);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocZero)
+{
+    void *test1 = nullptr;
+
+    test1 = memkind_malloc(pmem_kind, 0);
+    ASSERT_EQ(test1, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocSizeMax)
+{
+    void *test1 = nullptr;
+
+    errno = 0;
+    test1 = memkind_malloc(pmem_kind, SIZE_MAX);
+    ASSERT_EQ(test1, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCallocSmallClassMultipleTimes)
+{
+    const size_t size = 1 * KB;
+    const size_t num = 1;
+    const size_t iteration = 100;
+    char *default_str = nullptr;
+
+    for (size_t i = 0; i < iteration; ++i) {
+        default_str = (char *)memkind_calloc(pmem_kind, num, size);
+        ASSERT_NE(nullptr, default_str);
+        ASSERT_EQ(*default_str, 0);
+        ASSERT_EQ(memcmp(default_str, default_str + 1, size - 1), 0);
+
+        sprintf(default_str, "memkind_calloc MEMKIND_PMEM\n");
+
+        memkind_free(pmem_kind, default_str);
+    }
+}
+
+/*
+ * Test will check if it is not possible to allocate memory
+ * with calloc arguments size or num equal to zero
+ */
+TEST_P(MemkindPmemTestsCalloc, test_TC_MEMKIND_PmemCallocZero)
+{
+    void *test = nullptr;
+    size_t size = std::get<0>(GetParam());
+    size_t num = std::get<1>(GetParam());
+
+    test = memkind_calloc(pmem_kind, size, num);
+    ASSERT_EQ(test, nullptr);
+}
+
+INSTANTIATE_TEST_CASE_P(
+    CallocParam, MemkindPmemTestsCalloc,
+    ::testing::Values(std::make_tuple(10, 0),
+                      std::make_tuple(0, 0),
+                      std::make_tuple(0, 10)));
+
+TEST_P(MemkindPmemTestsCalloc, test_TC_MEMKIND_PmemCallocSizeMax)
+{
+    void *test = nullptr;
+    size_t size = SIZE_MAX;
+    size_t num = 1;
+    errno = 0;
+
+    test = memkind_calloc(pmem_kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_P(MemkindPmemTestsCalloc, test_TC_MEMKIND_PmemCallocNumMax)
+{
+    void *test = nullptr;
+    size_t size = 10;
+    size_t num = SIZE_MAX;
+    errno = 0;
+
+    test = memkind_calloc(pmem_kind, size, num);
+    ASSERT_EQ(test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCallocHugeClassMultipleTimes)
+{
+    const size_t size = MEMKIND_PMEM_CHUNK_SIZE;
+    const size_t num = 1;
+    const size_t iteration = 100;
+    char *default_str = nullptr;
+
+    for (size_t i = 0; i < iteration; ++i) {
+        default_str = (char *)memkind_calloc(pmem_kind, num, size);
+        ASSERT_NE(nullptr, default_str);
+        ASSERT_EQ(*default_str, 0);
+        ASSERT_EQ(memcmp(default_str, default_str + 1, size - 1), 0);
+
+        sprintf(default_str, "memkind_calloc MEMKIND_PMEM\n");
+
+        memkind_free(pmem_kind, default_str);
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeMemoryAfterDestroyLargeClass)
+{
+    memkind_t pmem_kind_test;
+    struct statfs st;
+    double blocksAvailable;
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_test);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_test);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    blocksAvailable = st.f_bfree;
+
+    while(1) {
+        if (memkind_malloc(pmem_kind_test, 16 * KB) == nullptr)
+            break;
+    }
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_GT(blocksAvailable, st.f_bfree);
+
+    err = memkind_destroy_kind(pmem_kind_test);
+    ASSERT_EQ(0, err);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_NEAR(blocksAvailable, st.f_bfree, 1);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeMemoryAfterDestroySmallClass)
+{
+    memkind_t pmem_kind_test;
+    struct statfs st;
+    double blocksAvailable;
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_test);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_test);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    blocksAvailable = st.f_bfree;
+
+    for(int i = 0; i < 100; ++i) {
+        ASSERT_NE(memkind_malloc(pmem_kind_test, 32), nullptr);
+    }
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_GT(blocksAvailable, st.f_bfree);
+
+    err = memkind_destroy_kind(pmem_kind_test);
+    ASSERT_EQ(0, err);
+
+    ASSERT_EQ(0, statfs(PMEM_DIR, &st));
+    ASSERT_NEAR(blocksAvailable, st.f_bfree, 1);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemRealloc)
+{
+    const size_t size1 = 512;
+    const size_t size2 = 1 * KB;
+    char *default_str = nullptr;
+
+    default_str = (char *)memkind_realloc(pmem_kind, default_str, size1);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc MEMKIND_PMEM with size %zu\n", size1);
+    printf("%s", default_str);
+
+    default_str = (char *)memkind_realloc(pmem_kind, default_str, size2);
+    ASSERT_NE(nullptr, default_str);
+
+    sprintf(default_str, "memkind_realloc MEMKIND_PMEM with size %zu\n", size2);
+    printf("%s", default_str);
+
+    memkind_free(pmem_kind, default_str);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocUsableSize)
+{
+    const struct {
+        size_t size;
+        size_t spacing;
+    } check_sizes[] = {
+        {.size = 10, .spacing = 8},
+        {.size = 100, .spacing = 16},
+        {.size = 200, .spacing = 32},
+        {.size = 500, .spacing = 64},
+        {.size = 1000, .spacing = 128},
+        {.size = 2000, .spacing = 256},
+        {.size = 3000, .spacing = 512},
+        {.size = 1 * MB, .spacing = 4 * MB},
+        {.size = 2 * MB, .spacing = 4 * MB},
+        {.size = 3 * MB, .spacing = 4 * MB},
+        {.size = 4 * MB, .spacing = 4 * MB}
+    };
+    struct memkind *pmem_temp = nullptr;
+
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_temp);
+    size_t usable_size = memkind_malloc_usable_size(pmem_temp, nullptr);
+    size_t nullkind_usable_size = memkind_malloc_usable_size(nullptr, nullptr);
+    ASSERT_EQ(0u, usable_size);
+    ASSERT_EQ(nullkind_usable_size, usable_size);
+    for (unsigned int i = 0; i < ARRAY_SIZE(check_sizes); ++i) {
+        size_t size = check_sizes[i].size;
+        void *alloc = memkind_malloc(pmem_temp, size);
+        ASSERT_NE(nullptr, alloc);
+
+        usable_size = memkind_malloc_usable_size(pmem_temp, alloc);
+        nullkind_usable_size = memkind_malloc_usable_size(nullptr, alloc);
+        ASSERT_EQ(nullkind_usable_size, usable_size);
+        size_t diff = usable_size - size;
+        ASSERT_GE(usable_size, size);
+        ASSERT_LE(diff, check_sizes[i].spacing);
+
+        memkind_free(pmem_temp, alloc);
+    }
+    err = memkind_destroy_kind(pmem_temp);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemResize)
+{
+    const size_t size = MEMKIND_PMEM_CHUNK_SIZE;
+    char *pmem_str10 = nullptr;
+    char *pmem_strX = nullptr;
+    memkind_t pmem_kind_no_limit = nullptr;
+    memkind_t pmem_kind_not_possible = nullptr;
+    int err = 0;
+
+    pmem_str10 = (char *)memkind_malloc(pmem_kind, MEMKIND_PMEM_MIN_SIZE);
+    ASSERT_NE(nullptr, pmem_str10);
+
+    // Out of memory
+    pmem_strX = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_EQ(nullptr, pmem_strX);
+
+    memkind_free(pmem_kind, pmem_str10);
+    memkind_free(pmem_kind, pmem_strX);
+
+    err = memkind_create_pmem(PMEM_DIR, PMEM_NO_LIMIT, &pmem_kind_no_limit);
+    ASSERT_EQ(0, err);
+    ASSERT_NE(nullptr, pmem_kind_no_limit);
+
+    pmem_str10 = (char *)memkind_malloc(pmem_kind_no_limit, MEMKIND_PMEM_MIN_SIZE);
+    ASSERT_NE(nullptr, pmem_str10);
+
+    pmem_strX = (char *)memkind_malloc(pmem_kind_no_limit, size);
+    ASSERT_NE(nullptr, pmem_strX);
+
+    memkind_free(pmem_kind_no_limit, pmem_str10);
+    memkind_free(pmem_kind_no_limit, pmem_strX);
+
+    err = memkind_destroy_kind(pmem_kind_no_limit);
+    ASSERT_EQ(0, err);
+
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE-1,
+                              &pmem_kind_not_possible);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    ASSERT_EQ(nullptr, pmem_kind_not_possible);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocZero)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+
+    test = memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test, nullptr);
+
+    new_test = memkind_realloc(pmem_kind, test, 0);
+    ASSERT_EQ(new_test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocSizeMax)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+
+    test = memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test, nullptr);
+    errno = 0;
+    new_test = memkind_realloc(pmem_kind, test, SIZE_MAX);
+    ASSERT_EQ(new_test, nullptr);
+    ASSERT_EQ(errno, ENOMEM);
+
+    memkind_free(pmem_kind, test);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocSizeZero)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+    void *new_test = nullptr;
+    const size_t iteration = 100;
+
+    for (unsigned i = 0; i < iteration; ++i) {
+        test = memkind_malloc(pmem_kind, size);
+        ASSERT_NE(test, nullptr);
+        errno = 0;
+        new_test = memkind_realloc(pmem_kind, test, 0);
+        ASSERT_EQ(new_test, nullptr);
+        ASSERT_EQ(errno, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocPtrCheck)
+{
+    size_t size = 1 * KB;
+    void *ptr_malloc = nullptr;
+    void *ptr_malloc_copy = nullptr;
+    void *ptr_realloc = nullptr;
+
+    ptr_malloc = memkind_malloc(pmem_kind, size);
+    ASSERT_NE(ptr_malloc, nullptr);
+
+    ptr_malloc_copy = ptr_malloc;
+
+    ptr_realloc = memkind_realloc(pmem_kind, ptr_malloc, PMEM_PART_SIZE);
+    ASSERT_EQ(ptr_realloc, nullptr);
+    ASSERT_EQ(ptr_malloc, ptr_malloc_copy);
+
+    memkind_free(pmem_kind, ptr_malloc);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptr)
+{
+    size_t size = 1 * KB;
+    void *test = nullptr;
+
+    test = memkind_realloc(pmem_kind, test, size);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(pmem_kind, test);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptrSizeMax)
+{
+    void *test = nullptr;
+
+    test = memkind_realloc(pmem_kind, test, SIZE_MAX);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptrZero)
+{
+    void *test = nullptr;
+
+    test = memkind_realloc(pmem_kind, test, 0);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocIncreaseSize)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocIncreaseSize";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size *= 2;
+    test2 = (char *)memkind_realloc(pmem_kind, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocIncreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocIncreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size *= 2;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, sizeof(val));
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocDecreaseSize)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocDecreaseSize";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 4;
+    test2 = (char *)memkind_realloc(pmem_kind, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, size);
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocDecreaseSizeNullKindVariant)
+{
+    size_t size = 1 * KB;
+    char *test1 = nullptr;
+    char *test2 = nullptr;
+    const char val[] = "test_TC_MEMKIND_PmemReallocDecreaseSizeNullKindVariant";
+    int status;
+
+    test1 = (char *)memkind_malloc(pmem_kind, size);
+    ASSERT_NE(test1, nullptr);
+
+    sprintf(test1, "%s", val);
+
+    size = 4;
+    test2 = (char *)memkind_realloc(nullptr, test1, size);
+    ASSERT_NE(test2, nullptr);
+    status = memcmp(val, test2, size);
+    ASSERT_EQ(status, 0);
+
+    memkind_free(pmem_kind, test2);
+}
+
+/*
+ * This test shows realloc "in-place" mechanism.
+ * In some cases like allocation shrinking within the same size class
+ * realloc will shrink allocation "in-place",
+ * but in others not (like changing size class).
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocInPlace)
+{
+    void *test1 = memkind_malloc(pmem_kind, 10 * MB);
+    ASSERT_NE(test1, nullptr);
+
+    /* Several reallocations within the same jemalloc size class*/
+    void *test1r = memkind_realloc(pmem_kind, test1, 6 * MB);
+    ASSERT_EQ(test1r, test1);
+
+    test1r = memkind_realloc(pmem_kind, test1, 10 * MB);
+    ASSERT_EQ(test1r, test1);
+
+    test1r = memkind_realloc(pmem_kind, test1, 8 * MB);
+    ASSERT_EQ(test1r, test1);
+
+    void *test2 = memkind_malloc(pmem_kind, 4 * MB);
+    ASSERT_NE(test2, nullptr);
+
+    /* 4MB => 16B (changing size class) */
+    void *test2r = memkind_realloc(pmem_kind, test2, 16);
+    ASSERT_NE(test2r, nullptr);
+
+    /* 8MB => 16B */
+    test1r = memkind_realloc(pmem_kind, test1, 16);
+
+    /*
+     * If the old size of the allocation is larger than
+     * the chunk size (4MB), we can reallocate it to 4MB first (in place),
+     * releasing some space, which makes it possible to do the actual
+     * shrinking...
+     */
+    ASSERT_NE(test1r, nullptr);
+    ASSERT_NE(test1r, test1);
+
+    /* ... and leaves some memory for new allocations. */
+    void *test3 = memkind_malloc(pmem_kind, 5 * MB);
+    ASSERT_NE(test3, nullptr);
+
+    memkind_free(pmem_kind, test1r);
+    memkind_free(pmem_kind, test2r);
+    memkind_free(pmem_kind, test3);
+}
+
+/*
+ * This test shows that we can make a single highest possible allocation
+ * and there still will be a place for another allocations.
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMaxFill)
+{
+    const int possible_alloc_max = 4;
+    void *test[possible_alloc_max+1] = {nullptr};
+    size_t total_mem = 0;
+    size_t free_mem = 0;
+    int i, j;
+
+    pmem_get_size(pmem_kind, total_mem, free_mem);
+
+    for (i = 0; i < possible_alloc_max; i++) {
+        for (j = total_mem; j > 0; --j) {
+            test[i] = memkind_malloc(pmem_kind, j);
+            if(test[i] != nullptr)
+                break;
+        }
+        ASSERT_NE(j, 0);
+    }
+
+    for (j = total_mem; j > 0; --j) {
+        test[possible_alloc_max] = memkind_malloc(pmem_kind, j);
+        if(test[possible_alloc_max] != nullptr)
+            break;
+    }
+    //Ensure there is no more space available on kind
+    ASSERT_EQ(j, 0);
+
+    for (i = 0; i < possible_alloc_max; i++) {
+        memkind_free(pmem_kind, test[i]);
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeNullptr)
+{
+    const double test_time = 5;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        memkind_free(pmem_kind, nullptr);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullptrSizeZero)
+{
+    const double test_time = 5;
+    void *test_nullptr = nullptr;
+    TimerSysTime timer;
+    timer.start();
+    do {
+        errno = 0;
+        //equivalent to memkind_malloc(pmem_kind,0)
+        test_nullptr = memkind_realloc(pmem_kind, nullptr, 0);
+        ASSERT_EQ(test_nullptr, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemReallocNullKindSizeZero)
+{
+    const double test_time = 5;
+    void *test_ptr_malloc = nullptr;
+    void *test_nullptr = nullptr;
+    size_t size = 1 * KB;
+
+    TimerSysTime timer;
+    timer.start();
+    do {
+        errno = 0;
+        test_ptr_malloc = memkind_malloc(pmem_kind, size);
+        ASSERT_NE(test_ptr_malloc, nullptr);
+        ASSERT_EQ(errno, 0);
+
+        errno = 0;
+        test_nullptr = memkind_realloc(nullptr, test_ptr_malloc, 0);
+        ASSERT_EQ(test_nullptr, nullptr);
+        ASSERT_EQ(errno, 0);
+    } while (timer.getElapsedTime() < test_time);
+}
+
+/*
+ * This test will stress pmem kind with malloc-free loop
+ * with various sizes for malloc
+ */
+TEST_P(MemkindPmemTestsMalloc, test_TC_MEMKIND_PmemMallocSizeConservative)
+{
+    const int loop_limit = 10;
+    int initial_alloc_limit = 0;
+    std::vector<void *> pmem_vec;
+    void *ptr;
+    size_t alloc_size = GetParam();
+    int i, err;
+    memkind_t kind = nullptr;
+
+    memkind_config *test_cfg = memkind_config_new();
+    ASSERT_NE(nullptr, test_cfg);
+    memkind_config_set_path(test_cfg, PMEM_DIR);
+    memkind_config_set_size(test_cfg, PMEM_PART_SIZE);
+    memkind_config_set_memory_usage_policy(test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+    err =  memkind_create_pmem_with_config(test_cfg, &kind);
+    ASSERT_EQ(err, 0);
+    memkind_config_delete(test_cfg);
+
+    //check maximum number of allocations right after creating the kind
+    while ((ptr = memkind_malloc(kind, alloc_size)) != nullptr) {
+        pmem_vec.push_back(ptr);
+    }
+
+    initial_alloc_limit = pmem_vec.size();
+
+    for(auto const &val: pmem_vec) {
+        memkind_free(kind, val);
+    }
+    pmem_vec.clear();
+
+    //check number of allocations in consecutive iterations of malloc-free loop
+    for (i = 0; i < loop_limit; i++) {
+
+        while ((ptr = memkind_malloc(kind, alloc_size)) != nullptr) {
+            pmem_vec.push_back(ptr);
+        }
+        int temp_limit_of_allocations = pmem_vec.size();
+
+        for(auto const &val: pmem_vec) {
+            memkind_free(kind, val);
+        }
+        pmem_vec.clear();
+
+        ASSERT_GE(temp_limit_of_allocations, 0.98 * initial_alloc_limit);
+    }
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(0, err);
+}
+
+INSTANTIATE_TEST_CASE_P(
+    MallocParam, MemkindPmemTestsMalloc,
+    ::testing::Values(32, 60, 80, 100, 128, 150, 160, 250, 256, 300, 320,
+                      500, 512, 800, 896, 3000, 4 * KB, 6000, 10000, 60000,
+                      96 * KB, 112 * KB, 128 * KB, 160 * KB, 192 * KB, 500000,
+                      2 * MB, 5 * MB));
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMallocSmallSizeFillConservative)
+{
+    const size_t small_size[] = {8, 16, 32, 48, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320, 384,
+                                 448, 512, 640, 768, 896, 1 * KB, 1280, 1536, 1792, 2 * KB, 2560,
+                                 3 * KB, 3584, 4 * KB, 5 * KB, 6 * KB, 7 * KB, 8 * KB, 10 * KB,
+                                 12 * KB, 14 * KB
+                                };
+    const int loop_limit = 100;
+    int initial_alloc_limit = 0;
+    int err;
+    std::vector<void *> pmem_vec;
+    void *ptr;
+    unsigned i, j;
+    memkind_t kind = nullptr;
+
+    memkind_config *test_cfg = memkind_config_new();
+    ASSERT_NE(nullptr, test_cfg);
+
+    memkind_config_set_path(test_cfg, PMEM_DIR);
+    memkind_config_set_size(test_cfg, PMEM_PART_SIZE);
+    memkind_config_set_memory_usage_policy(test_cfg,
+                                           MEMKIND_MEM_USAGE_POLICY_CONSERVATIVE);
+    err =  memkind_create_pmem_with_config(test_cfg, &kind);
+    ASSERT_EQ(err, 0);
+    memkind_config_delete(test_cfg);
+
+    //check maximum number of allocations right after creating the kind
+    j = 0;
+    while ((ptr = memkind_malloc(kind, small_size[j++])) != nullptr) {
+        pmem_vec.push_back(ptr);
+        if (j == ARRAY_SIZE(small_size)) j = 0;
+    }
+
+    initial_alloc_limit = pmem_vec.size();
+
+    for(auto const &val: pmem_vec) {
+        memkind_free(kind, val);
+    }
+    pmem_vec.clear();
+
+    //check number of allocations in consecutive iterations of malloc-free loop
+    for (i = 0; i < loop_limit; i++) {
+
+        j = 0;
+        while ((ptr = memkind_malloc(kind, small_size[j++])) != nullptr) {
+            pmem_vec.push_back(ptr);
+            if (j == ARRAY_SIZE(small_size)) j = 0;
+        }
+
+        int temp_limit_of_allocations = pmem_vec.size();
+
+        for(auto const &val: pmem_vec) {
+            memkind_free(kind, val);
+        }
+        pmem_vec.clear();
+
+        ASSERT_GE(temp_limit_of_allocations, 0.98 * initial_alloc_limit);
+    }
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignWrongAlignmentLessThanVoidAndNotPowerOfTwo)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t wrong_alignment = 3;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignWrongAlignmentLessThanVoidAndPowerOfTwo)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t wrong_alignment = sizeof(void *)/2;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignWrongAlignmentNotPowerOfTwo)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t wrong_alignment = sizeof(void *)+1;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, wrong_alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, EINVAL);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemPosixMemalignLowestCorrectAlignment)
+{
+    void *test = nullptr;
+    size_t size = 32;
+    size_t alignment = sizeof(void *);
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, alignment, size);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, 0);
+    ASSERT_NE(test, nullptr);
+
+    memkind_free(pmem_kind, test);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPosixMemalignSizeZero)
+{
+    void *test = nullptr;
+    size_t alignment = sizeof(void *);
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, alignment, 0);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, 0);
+    ASSERT_EQ(test, nullptr);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPosixMemalignSizeMax)
+{
+    void *test = nullptr;
+    size_t alignment = 64;
+
+    errno = 0;
+    int ret = memkind_posix_memalign(pmem_kind, &test, alignment, SIZE_MAX);
+    ASSERT_EQ(errno, 0);
+    ASSERT_EQ(ret, ENOMEM);
+    ASSERT_EQ(test, nullptr);
+}
+
+/*
+ * This is a basic alignment test which will make alignment allocations,
+ * check pointers, write and read values from allocated memory
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemPosixMemalign)
+{
+    const int test_value = 123456;
+    const int test_loop = 10;
+    uintptr_t alignment;
+    unsigned i = 0;
+    void *test = nullptr;
+    std::vector<void *> pmem_vec;
+    int ret;
+
+    for(alignment =  1 * KB; alignment <= 128 * KB; alignment *= 2) {
+        for (i = 0; i < test_loop; i++) {
+            errno = 0;
+            while((ret = memkind_posix_memalign(pmem_kind, &test, alignment,
+                                                sizeof(int *))) == 0) {
+                ASSERT_EQ(errno, 0);
+
+                pmem_vec.push_back(test);
+
+                //test pointer should be usable
+                *(int *)test = test_value;
+                ASSERT_EQ(*(int *)test, test_value);
+
+                //check for correct address alignment
+                ASSERT_EQ((uintptr_t)(test) & (alignment - 1), (uintptr_t)0);
+                errno = 0;
+            }
+
+            for(auto const &val: pmem_vec) {
+                memkind_free(pmem_kind, val);
+            }
+            pmem_vec.clear();
+        }
+    }
+}
+
+static memkind_t *pools;
+static int npools = 3;
+static void *thread_func(void *arg)
+{
+    int start_idx = *(int *)arg;
+    int err = 0;
+    for (int idx = 0; idx < npools; ++idx) {
+        int pool_id = start_idx + idx;
+
+        if (pools[pool_id] == nullptr) {
+            err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pools[pool_id]);
+            EXPECT_EQ(0, err);
+        }
+
+        if (err == 0) {
+            void *test = memkind_malloc(pools[pool_id], sizeof(void *));
+            EXPECT_TRUE(test != nullptr);
+            memkind_free(pools[pool_id], test);
+            memkind_destroy_kind(pools[pool_id]);
+        }
+    }
+
+    return nullptr;
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMultithreads)
+{
+    int nthreads = 10, status = 0;
+    pthread_t *threads = (pthread_t *)calloc(nthreads, sizeof(pthread_t));
+    ASSERT_NE(threads, nullptr);
+    int *pool_idx = (int *)calloc(nthreads, sizeof(int));
+    ASSERT_NE(pool_idx, nullptr);
+    pools = (memkind_t *)calloc(npools * nthreads, sizeof(memkind_t));
+    ASSERT_NE(pools, nullptr);
+
+    for (int t = 0; t < nthreads; t++) {
+        pool_idx[t] = npools * t;
+        status = pthread_create(&threads[t], nullptr, thread_func, &pool_idx[t]);
+        ASSERT_EQ(0, status);
+    }
+
+    for (int t = 0; t < nthreads; t++) {
+        status = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, status);
+    }
+
+    free(pools);
+    free(threads);
+    free(pool_idx);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemDestroyKind)
+{
+    const size_t pmem_array_size = 10;
+    struct memkind *pmem_kind_array[pmem_array_size] = {nullptr};
+
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE,
+                                  &pmem_kind_array[0]);
+    ASSERT_EQ(err, 0);
+
+    err = memkind_destroy_kind(pmem_kind_array[0]);
+    ASSERT_EQ(err, 0);
+
+    for (unsigned int i = 0; i < pmem_array_size; ++i) {
+        err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_kind_array[i]);
+        ASSERT_EQ(err, 0);
+    }
+
+    char *pmem_middle_name = pmem_kind_array[5]->name;
+    err = memkind_destroy_kind(pmem_kind_array[5]);
+    ASSERT_EQ(err, 0);
+
+    err = memkind_destroy_kind(pmem_kind_array[6]);
+    ASSERT_EQ(err, 0);
+
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_kind_array[5]);
+    ASSERT_EQ(err, 0);
+
+    char *pmem_new_middle_name = pmem_kind_array[5]->name;
+
+    ASSERT_STREQ(pmem_middle_name, pmem_new_middle_name);
+
+    for (unsigned int i = 0; i < pmem_array_size; ++i) {
+        if (i != 6) {
+            err = memkind_destroy_kind(pmem_kind_array[i]);
+            ASSERT_EQ(err, 0);
+        }
+    }
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemDestroyKindArenaZero)
+{
+    struct memkind *pmem_temp_1 = nullptr;
+    struct memkind *pmem_temp_2 = nullptr;
+    unsigned int arena_zero = 0;
+    int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp_1);
+    ASSERT_EQ(err, 0);
+
+    arena_zero = pmem_temp_1->arena_zero;
+    err = memkind_destroy_kind(pmem_temp_1);
+    ASSERT_EQ(err, 0);
+    err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp_2);
+    ASSERT_EQ(err, 0);
+
+    ASSERT_EQ(arena_zero,pmem_temp_2->arena_zero);
+
+    err = memkind_destroy_kind(pmem_temp_2);
+    ASSERT_EQ(err, 0);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCreateDestroyKindEmptyLoop)
+{
+    struct memkind *pmem_temp = nullptr;
+
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocSmallSizeFreeDefinedPmemKind)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size = 1 * KB;
+
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+        void *ptr = memkind_malloc(pmem_temp, size);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocDifferentSizesDifferentKindsDefinedFreeForAllKinds)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = MEMKIND_PMEM_CHUNK_SIZE;
+    void *ptr = nullptr;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+
+        ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr = memkind_malloc(pmem_temp, size_2);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocDifferentSizesDifferentKindsDefinedFreeForNotPmemKinds)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = MEMKIND_PMEM_CHUNK_SIZE;
+    void *ptr = nullptr;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+
+        ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr = memkind_malloc(pmem_temp, size_2);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(nullptr, ptr);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(MEMKIND_DEFAULT, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(MEMKIND_REGULAR, ptr_regular);
+
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopMallocDifferentSizesDifferentKindsNullKindFree)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 1 * KB;
+    const size_t size_2 = MEMKIND_PMEM_CHUNK_SIZE;
+    void *ptr = nullptr;
+    void *ptr_default = nullptr;
+    void *ptr_regular = nullptr;
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+
+        ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(pmem_temp, ptr);
+        ptr = memkind_malloc(pmem_temp, size_2);
+        ASSERT_NE(nullptr, ptr);
+        memkind_free(nullptr, ptr);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_2);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_default = memkind_malloc(MEMKIND_DEFAULT, size_1);
+        ASSERT_NE(nullptr, ptr_default);
+        memkind_free(nullptr, ptr_default);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_2);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+        ptr_regular = memkind_malloc(MEMKIND_REGULAR, size_1);
+        ASSERT_NE(nullptr, ptr_regular);
+        memkind_free(nullptr, ptr_regular);
+
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateDestroyKindLoopWithRealloc)
+{
+    struct memkind *pmem_temp = nullptr;
+    const size_t size_1 = 512;
+    const size_t size_2 = 1 * KB;
+
+    for (unsigned int i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp);
+        ASSERT_EQ(err, 0);
+        void *ptr = memkind_malloc(pmem_temp, size_1);
+        ASSERT_NE(nullptr, ptr);
+        void *ptr_2 = memkind_realloc(pmem_temp, ptr, size_2);
+        ASSERT_NE(nullptr, ptr_2);
+        memkind_free(pmem_temp, ptr_2);
+        err = memkind_destroy_kind(pmem_temp);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+TEST_F(MemkindPmemTests,
+       test_TC_MEMKIND_PmemCreateCheckErrorCodeArenaCreate)
+{
+    struct memkind *pmem_temp[MEMKIND_MAX_KIND] = { nullptr };
+    unsigned i = 0, j = 0;
+    int err = 0;
+
+    for (i = 0; i < MEMKIND_MAX_KIND; ++i) {
+        err = memkind_create_pmem(PMEM_DIR, MEMKIND_PMEM_MIN_SIZE, &pmem_temp[i]);
+        if (err) {
+            ASSERT_EQ(err, MEMKIND_ERROR_ARENAS_CREATE);
+            break;
+        }
+    }
+    for (j = 0; j < i; ++j) {
+        err = memkind_destroy_kind(pmem_temp[j]);
+        ASSERT_EQ(err, 0);
+    }
+}
+
+static void *thread_func_kinds(void *arg)
+{
+    memkind_t pmem_thread_kind;
+    int err = 0;
+
+    EXPECT_TRUE(pthread_mutex_lock(&mutex) == 0);
+    EXPECT_TRUE(pthread_cond_wait(&cond, &mutex) == 0);
+    EXPECT_TRUE(pthread_mutex_unlock(&mutex) == 0);
+
+    err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_thread_kind);
+
+    if(err == 0) {
+        void *test = memkind_malloc(pmem_thread_kind, 32);
+        EXPECT_TRUE(test != nullptr);
+
+        memkind_free(pmem_thread_kind, test);
+        err = memkind_destroy_kind(pmem_thread_kind);
+        EXPECT_EQ(0, err);
+    }
+
+    return nullptr;
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemMultithreadsStressKindsCreate)
+{
+    const int nthreads = 50;
+    int i, t, err;
+    int max_possible_kind = 0;
+    memkind_t pmem_kinds[MEMKIND_MAX_KIND] = {nullptr};
+    pthread_t *threads = (pthread_t *)calloc(nthreads, sizeof(pthread_t));
+    ASSERT_NE(threads, nullptr);
+
+    // This loop will create as many kinds as possible
+    // to obtain a real kind limit
+    for (i = 0; i < MEMKIND_MAX_KIND; i++) {
+        err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kinds[i]);
+        if(err != 0) {
+            ASSERT_GT(i, 0);
+            max_possible_kind = i;
+            --i;
+            break;
+        }
+        ASSERT_NE(nullptr, pmem_kinds[i]);
+    }
+
+    // destroy last kind so it will be possible
+    // to create only one pmem kind in threads
+    err = memkind_destroy_kind(pmem_kinds[i]);
+    ASSERT_EQ(0, err);
+
+    for (t = 0; t < nthreads; t++) {
+        err = pthread_create(&threads[t], nullptr, thread_func_kinds, nullptr);
+        ASSERT_EQ(0, err);
+    }
+
+    sleep(1);
+    err = pthread_cond_broadcast(&cond);
+    ASSERT_EQ(0, err);
+
+    for (t = 0; t < nthreads; t++) {
+        err = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, err);
+    }
+
+    for (i = 0; i < max_possible_kind - 1; i++) {
+        err = memkind_destroy_kind(pmem_kinds[i]);
+        ASSERT_EQ(0, err);
+    }
+
+    free(threads);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemKindFreeBenchmarkOneThread)
+{
+    const size_t pmem_array_size = 10;
+    const size_t alloc_size = 512;
+    struct memkind *pmem_kind_array[pmem_array_size] = { nullptr };
+    std::vector<void *> pmem_vec_exp[pmem_array_size];
+    std::vector<void *> pmem_vec_imp[pmem_array_size];
+    void *ptr;
+    TimerSysTime timer;
+    double test1Time, test2Time;
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        while ((ptr = memkind_malloc(pmem_kind_array[i], alloc_size)) != nullptr) {
+            pmem_vec_exp[i].push_back(ptr);
+        }
+    }
+
+    timer.start();
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        for(auto const &val: pmem_vec_exp[i]) {
+            memkind_free(pmem_kind_array[i], val);
+        }
+    }
+
+    test1Time = timer.getElapsedTime();
+    printf("Free time with explicitly kind: %f\n", test1Time);
+
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        while ((ptr = memkind_malloc(pmem_kind_array[i], alloc_size)) != nullptr) {
+            pmem_vec_imp[i].push_back(ptr);
+        }
+    }
+
+    timer.start();
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        for(auto const &val: pmem_vec_imp[i]) {
+            memkind_free(nullptr, val);
+        }
+    }
+    test2Time = timer.getElapsedTime();
+    printf("Free time with implicitly kind: %f\n", test2Time);
+
+    ASSERT_LT(test1Time, test2Time);
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        int err = memkind_destroy_kind(pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+}
+
+static const int threadsNum = 10;
+static memkind *testKind[threadsNum] = { nullptr };
+static const int mallocCount = 100000;
+static void *ptr[threadsNum][mallocCount] = { { nullptr } };
+
+static void *thread_func_FreeWithNullptr(void *arg)
+{
+    int kindIndex = *(int *)arg;
+    EXPECT_TRUE(pthread_mutex_lock(&mutex) == 0);
+    EXPECT_TRUE(pthread_cond_wait(&cond, &mutex) == 0);
+    EXPECT_TRUE(pthread_mutex_unlock(&mutex) == 0);
+
+    for (int j = 0; j < mallocCount; ++j) {
+        if (ptr[kindIndex][j] == nullptr) {
+            break;
+        }
+        memkind_free(nullptr, ptr[kindIndex][j]);
+        ptr[kindIndex][j] = nullptr;
+    }
+
+    return nullptr;
+}
+
+static void *thread_func_FreeWithKind(void *arg)
+{
+    int kindIndex = *(int *)arg;
+    EXPECT_TRUE(pthread_mutex_lock(&mutex) == 0);
+    EXPECT_TRUE(pthread_cond_wait(&cond, &mutex) == 0);
+    EXPECT_TRUE(pthread_mutex_unlock(&mutex) == 0);
+
+    for (int j = 0; j < mallocCount; ++j) {
+        if (ptr[kindIndex][j] == nullptr) {
+            break;
+        }
+        memkind_free(testKind[kindIndex], ptr[kindIndex][j]);
+    }
+
+    return nullptr;
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemKindFreeBenchmarkWithThreads)
+{
+    const size_t allocSize = 512;
+    int err;
+    TimerSysTime timer;
+    double duration;
+    pthread_t *threads = (pthread_t *)calloc(threadsNum, sizeof(pthread_t));
+    ASSERT_NE(threads, nullptr);
+
+    for (int i = 0; i < threadsNum; ++i) {
+        err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &testKind[i]);
+        ASSERT_EQ(err, 0);
+        ASSERT_NE(nullptr, testKind[i]);
+        int j = 0;
+        for (j = 0; j < mallocCount; ++j) {
+            ptr[i][j] = memkind_malloc(testKind[i], allocSize);
+            if (ptr[i][j] == nullptr) {
+                break;
+            }
+        }
+        ASSERT_NE(j, mallocCount);
+    }
+
+    int threadIndex[threadsNum];
+    for (int i = 0; i < threadsNum; ++i) {
+        threadIndex[i] = i;
+    }
+
+    for (int t = 0; t < threadsNum; t++) {
+        err = pthread_create(&threads[t], nullptr, thread_func_FreeWithKind,
+                             &threadIndex[t]);
+        ASSERT_EQ(0, err);
+    }
+
+    // sleep is here to ensure that all threads start at one the same time
+    sleep(1);
+    timer.start();
+    err = pthread_cond_broadcast(&cond);
+    ASSERT_EQ(0, err);
+    for (int t = 0; t < threadsNum; ++t) {
+        int err = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, err);
+    }
+    duration = timer.getElapsedTime();
+    printf("Free time with explicitly kind: %f\n", duration);
+
+    for (int i = 0; i < threadsNum; i++) {
+        int j = 0;
+        for (j = 0; j < mallocCount; ++j) {
+            ptr[i][j] = memkind_malloc(testKind[i], allocSize);
+            if (ptr[i][j] == nullptr) {
+                break;
+            }
+        }
+        ASSERT_NE(j, mallocCount);
+    }
+
+    for (int t = 0; t < threadsNum; ++t) {
+        err = pthread_create(&threads[t], nullptr, thread_func_FreeWithNullptr,
+                             &threadIndex[t]);
+        ASSERT_EQ(0, err);
+    }
+
+    // sleep is here to ensure that all threads start at one the same time
+    sleep(1);
+    timer.start();
+    err = pthread_cond_broadcast(&cond);
+    ASSERT_EQ(0, err);
+    for (int t = 0; t < threadsNum; t++) {
+        err = pthread_join(threads[t], nullptr);
+        ASSERT_EQ(0, err);
+    }
+
+    duration = timer.getElapsedTime();
+    printf("Free time with implicitly kind: %f\n", duration);
+
+    for (int i = 0; i < threadsNum; ++i) {
+        err = memkind_destroy_kind(testKind[i]);
+        ASSERT_EQ(0, err);
+    }
+    free(threads);
+}
+
+/*
+ * Test will create array of kinds, fill them and then attempt to free one ptr
+ * passing nullptr instead of kind. Then it will try to malloc again and if it will be
+ * successful the test passes (memkind_free(nullptr,...) was successful)
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemFreeUsingNullptrInsteadOfKind)
+{
+    const size_t pmem_array_size = 10;
+    const size_t alloc_size = 1 * KB;
+    struct memkind *pmem_kind_array[pmem_array_size] = { nullptr };
+    std::vector<void *> pmem_vec[pmem_array_size];
+    void *testPtr = nullptr;
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        while ((testPtr = memkind_malloc(pmem_kind_array[i], alloc_size)) != nullptr) {
+            pmem_vec[i].push_back(testPtr);
+        }
+    }
+
+    memkind_free(nullptr, pmem_vec[5].at(5));
+
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        // attempt to alloc memory to the kinds
+        testPtr = memkind_malloc(pmem_kind_array[i], alloc_size);
+        if (i == 5) {
+            // allocation should be successful - confirmation that memkind_free(nullptr,...) works fine
+            ASSERT_NE(testPtr, nullptr);
+            pmem_vec[i].at(5) = testPtr;
+        } else {
+            // There is no more free space in other kinds
+            ASSERT_EQ(testPtr, nullptr);
+        }
+    }
+
+    // free the rest of the space and destroy kinds.
+    for (size_t i = 0; i < pmem_array_size; ++i) {
+        for(auto const &val: pmem_vec[i]) {
+            memkind_free(pmem_kind_array[i], val);
+        }
+        int err = memkind_destroy_kind(pmem_kind_array[i]);
+        ASSERT_EQ(0, err);
+    }
+}
+
+/*
+ * This is a test which confirms that extent deallocation function ( pmem_extent_dalloc )
+ * was called correctly for pmem allocation.
+ */
+TEST_F(MemkindPmemTests, test_TC_MEMKIND_PmemCheckExtentDalloc)
+{
+    struct memkind *kind = nullptr;
+    const int mallocLimit = 10000;
+    void *ptr[mallocLimit] = { nullptr };
+    struct stat st;
+    double initialBlocks;
+
+    int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &kind);
+    ASSERT_EQ(err, 0);
+
+    struct memkind_pmem *priv = (memkind_pmem *)kind->priv;
+
+    for (int x = 0; x < 10; ++x) {
+        // sleep is here to help trigger dalloc extent
+        sleep(2);
+        int allocCount = 0;
+        for (int i = 0; i < mallocLimit; ++i) {
+            ptr[i] = memkind_malloc(kind, 32);
+            if (ptr[i] == nullptr)
+                break;
+
+            allocCount = i;
+        }
+
+        // store initial amount of allocated blocks
+        if (x == 0) {
+            ASSERT_EQ(0, fstat(priv->fd, &st));
+            initialBlocks = st.st_blocks;
+        }
+
+        for (int i = 0; i < allocCount; ++i)
+            memkind_free(kind, ptr[i]);
+
+        ASSERT_EQ(0, fstat(priv->fd, &st));
+        // if amount of blocks is less than initial, extent was called.
+        if (initialBlocks > st.st_blocks)
+            break;
+    }
+    ASSERT_GT(initialBlocks, st.st_blocks);
+
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(0, err);
+}
+
+TEST_F(MemkindPmemTests, test_TC_MEMKINDPmemDefragreallocate_success)
+{
+    struct memkind *kind = nullptr;
+    unsigned i;
+    unsigned count_mem_transfer = 0;
+    const unsigned alloc = 50000;
+    void *nptr;
+    std::vector<void *> pmem_vec;
+    pmem_vec.reserve(alloc);
+    int err = memkind_create_pmem(PMEM_DIR, 0, &kind);
+    ASSERT_EQ(err, 0);
+
+    for (i = 0; i < alloc; ++i) {
+        void *ptr =  memkind_malloc(kind, 1 * KB);
+        ASSERT_NE(ptr, nullptr);
+        memset(ptr, 'a', 1 * KB);
+        pmem_vec.push_back(ptr);
+    }
+
+    for (i = 1; i < alloc;) {
+        memkind_free(kind, pmem_vec.at(i));
+        pmem_vec.at(i) = nullptr;
+        // Free memory with irregular pattern
+        if (i % 2 == 0)
+            i += 3;
+        else
+            i += 5;
+    }
+
+    for (i = 0; i < pmem_vec.size(); ++i) {
+        nptr = memkind_defrag_reallocate(kind, pmem_vec.at(i));
+        if (nptr) {
+            pmem_vec.at(i) = nptr;
+            count_mem_transfer++;
+        }
+    }
+    ASSERT_NE(count_mem_transfer, 0U);
+
+    for(auto const &val: pmem_vec) {
+        memkind_free(kind, val);
+    }
+
+    err = memkind_destroy_kind(kind);
+    ASSERT_EQ(err, 0);
+}
diff --git a/test/memkind_stat_test.cpp b/test/memkind_stat_test.cpp
new file mode 100644
index 0000000..42f5075
--- /dev/null
+++ b/test/memkind_stat_test.cpp
@@ -0,0 +1,85 @@
+/*
+ * Copyright (C) 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include "common.h"
+
+class MemkindStatTests: public ::testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_StatUnknownType)
+{
+    size_t value;
+    int err = memkind_get_stat(nullptr, MEMKIND_STAT_TYPE_MAX_VALUE, &value);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+    err = memkind_get_stat(MEMKIND_REGULAR, MEMKIND_STAT_TYPE_MAX_VALUE, &value);
+    ASSERT_EQ(MEMKIND_ERROR_INVALID, err);
+}
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_DefaultKind)
+{
+    size_t value;
+    int err = memkind_update_cached_stats();
+    ASSERT_EQ(MEMKIND_SUCCESS, err);
+    void *ptr = memkind_malloc(MEMKIND_DEFAULT, 100);
+    for(int i = 0; i < MEMKIND_STAT_TYPE_MAX_VALUE; ++i) {
+        err = memkind_get_stat(MEMKIND_DEFAULT, static_cast<memkind_stat_type>(i),
+                               &value);
+        ASSERT_EQ(MEMKIND_SUCCESS, err);
+    }
+    memkind_free(nullptr, ptr);
+}
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_EmptyRegularKind)
+{
+    size_t stat;
+    int err = memkind_update_cached_stats();
+    ASSERT_EQ(MEMKIND_SUCCESS, err);
+    for(int i = 0; i < MEMKIND_STAT_TYPE_MAX_VALUE; ++i) {
+        err = memkind_get_stat(MEMKIND_REGULAR, static_cast<memkind_stat_type>(i),
+                               &stat);
+        ASSERT_EQ(MEMKIND_SUCCESS, err);
+        ASSERT_EQ(stat, 0U);
+    }
+}
+
+TEST_F(MemkindStatTests, test_TC_MEMKIND_GetGlobal)
+{
+    size_t value;
+    int err = memkind_update_cached_stats();
+    ASSERT_EQ(MEMKIND_SUCCESS, err);
+    for(int i = 0; i < MEMKIND_STAT_TYPE_MAX_VALUE; ++i) {
+        int err = memkind_get_stat(nullptr, static_cast<memkind_stat_type>(i), &value);
+        ASSERT_EQ(MEMKIND_SUCCESS, err);
+    }
+}
diff --git a/test/memkind_versioning_tests.cpp b/test/memkind_versioning_tests.cpp
new file mode 100644
index 0000000..12e8000
--- /dev/null
+++ b/test/memkind_versioning_tests.cpp
@@ -0,0 +1,60 @@
+/*
+ * Copyright (C) 2015 - 2016 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+
+#include "memkind.h"
+
+#include "common.h"
+
+/*
+ * memkind versioning tests.
+ */
+class MemkindVersioningTests: public :: testing::Test
+{
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+
+    const int max_version_value = 999;
+
+};
+
+//Test memkind_get_version().
+TEST_F(MemkindVersioningTests, test_TC_MEMKIND_GetVersionFunc)
+{
+    int max_return_val =
+        1000000 * max_version_value
+        + 1000 * max_version_value
+        + max_version_value;
+
+    //version number > 0
+    EXPECT_GT(memkind_get_version(), 0);
+
+    //version number <= max_return_val
+    EXPECT_LE(memkind_get_version(), max_return_val);
+}
diff --git a/test/memory_footprint_test.cpp b/test/memory_footprint_test.cpp
new file mode 100644
index 0000000..feaf0e8
--- /dev/null
+++ b/test/memory_footprint_test.cpp
@@ -0,0 +1,354 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "common.h"
+#include "random_sizes_allocator.h"
+#include "proc_stat.h"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+#include "allocator_perf_tool/Allocation_info.hpp"
+
+#include <memkind.h>
+
+#include <condition_variable>
+#include <functional>
+#include <mutex>
+#include <random>
+#include <thread>
+
+class Worker
+{
+public:
+    Worker(RandomSizesAllocator &&allocator, double malloc_probability)
+        : allocator(std::move(allocator)), malloc_probability(malloc_probability)  {}
+
+    void work()
+    {
+        if (allocator.empty() || get_random_bool(malloc_probability)) {
+            requested_memory_sum += allocator.malloc_random_memory();
+        } else {
+            requested_memory_sum -= allocator.free_random_memory();
+        }
+    }
+
+    size_t get_requested_memory_sum_bytes() const
+    {
+        return requested_memory_sum;
+    }
+
+private:
+    bool get_random_bool(double probability)
+    {
+        std::bernoulli_distribution distribution(probability);
+        return distribution(generator);
+    }
+
+    std::default_random_engine generator;
+    RandomSizesAllocator allocator;
+    size_t requested_memory_sum = 0;
+    double malloc_probability;
+};
+
+
+class MemoryFootprintStats
+{
+public:
+    void reset()
+    {
+        std::lock_guard<std::mutex> lk(sample_guard);
+        initial_virtual_memory = proc_stat.get_virtual_memory_size_bytes();
+        initial_physical_memory = proc_stat.get_physical_memory_size_bytes();
+        vm_overhead_sum = 0;
+        current_vm_overhead = 0;
+        max_vm_overhead = 0;
+
+        current_phys_overhead = 0;
+        phys_overhead_sum = 0;
+        max_phys_overhead = 0;
+
+        requested_memory = 0;
+        sample_count = 0;
+    }
+
+    void sample(long long requested_memory_bytes)
+    {
+        std::lock_guard<std::mutex> lk(sample_guard);
+        current_vm = proc_stat.get_virtual_memory_size_bytes();
+        current_phys = proc_stat.get_physical_memory_size_bytes();
+
+        sample_count++;
+
+        requested_memory = requested_memory_bytes;
+
+        current_vm_overhead = current_vm - initial_virtual_memory - requested_memory;
+        vm_overhead_sum += current_vm_overhead;
+        max_vm_overhead = std::max(max_vm_overhead, current_vm_overhead);
+
+        current_phys_overhead = current_phys - initial_physical_memory;
+        phys_overhead_sum += current_phys_overhead;
+        max_phys_overhead = std::max(max_phys_overhead, current_phys_overhead);
+    }
+
+    void log_data() const
+    {
+        std::lock_guard<std::mutex> lk(sample_guard);
+        GTestAdapter::RecordProperty("avg_vm_overhead_per_operation_mb",
+                                     convert_bytes_to_mb(vm_overhead_sum) / sample_count);
+        GTestAdapter::RecordProperty("avg_vm_overhead_growth_per_operation_mb",
+                                     convert_bytes_to_mb(current_vm_overhead) / sample_count);
+        GTestAdapter::RecordProperty("max_vm_overhead_mb",
+                                     convert_bytes_to_mb(max_vm_overhead));
+        GTestAdapter::RecordProperty("avg_phys_overhead_per_operation_mb",
+                                     convert_bytes_to_mb(phys_overhead_sum) / sample_count);
+        GTestAdapter::RecordProperty("max_phys_overhead_mb",
+                                     convert_bytes_to_mb(max_phys_overhead));
+        GTestAdapter::RecordProperty("overhead_to_requested_memory_ratio_percent",
+                                     100.f * current_vm_overhead / requested_memory);
+        GTestAdapter::RecordProperty("requested_memory_mb",
+                                     convert_bytes_to_mb(requested_memory));
+    }
+private:
+    long long initial_virtual_memory;
+    long long vm_overhead_sum = 0;
+    long long current_vm_overhead = 0;
+    long long max_vm_overhead = 0;
+
+    long long initial_physical_memory;
+    long long current_phys_overhead = 0;
+    long long phys_overhead_sum = 0;
+    long long max_phys_overhead = 0;
+
+    long long requested_memory = 0;
+    long long sample_count = 0;
+
+    long long current_vm;
+    long long current_phys;
+
+    ProcStat proc_stat;
+    mutable std::mutex sample_guard;
+};
+
+/* Execute func calling it n_calls times in n_threads threads.
+ * The execution is multithreaded but the func calls order is sequential.
+ * func takes thread id, and operation id as an argument,
+ * and must return thread id of the next thread, where thread ids are in range <0, n_threads-1>.
+ * init_thread_id specify the initial thread id.
+ */
+void run_multithreaded_seq_exec(unsigned n_threads, unsigned init_thread_id,
+                                unsigned n_calls, std::function<unsigned(unsigned, unsigned)> func)
+{
+    std::vector<std::thread> threads;
+    std::mutex mutex;
+    std::condition_variable turns_holder;
+    unsigned current_call = 0;
+    unsigned current_tid = init_thread_id;
+
+    threads.reserve(n_threads);
+
+    mutex.lock();
+
+    for(int tid=0; tid<n_threads; ++tid) {
+        threads.emplace_back([ &, tid]() {
+            while(current_call < n_calls) {
+                std::unique_lock<std::mutex> lk(mutex);
+                turns_holder.wait(lk, [ &,tid] {return current_tid == tid || current_call == n_calls;});
+                if(current_call == n_calls) {
+                    return;
+                }
+                current_tid = func(tid, current_call);
+                ASSERT_LT(current_tid, n_threads) << "Incorrect thread id!";
+                current_call++;
+                lk.unlock();
+                turns_holder.notify_all();
+            }
+        });
+    }
+
+    mutex.unlock();
+
+    for(int i=0; i<threads.size(); i++) {
+        threads[i].join();
+    }
+}
+
+/*
+ * Create threads and measure the cost of maintaining allocations from threads.
+ * Allocations order is sequential (otherwise the results might be very nondeterministic).
+ */
+void run_test(memkind_t kind, size_t min_size, size_t max_size,
+              unsigned n_threads, double malloc_probability=1.0, unsigned n_calls=1000)
+{
+    Worker worker(RandomSizesAllocator(kind, min_size, max_size, n_calls),
+                  malloc_probability);
+
+    MemoryFootprintStats mem_footprint_stats;
+
+    auto func = [&](unsigned tid, unsigned id) -> unsigned {
+        if(id == 0)
+        {
+            mem_footprint_stats.reset();
+        }
+
+        worker.work();
+        mem_footprint_stats.sample(worker.get_requested_memory_sum_bytes());
+
+        return (tid + 1) % n_threads; //next thread id
+    };
+    run_multithreaded_seq_exec(n_threads, 0, n_calls, func);
+    mem_footprint_stats.log_data();
+}
+
+class MemoryFootprintTest: public :: testing::Test
+{};
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_small_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_small_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 16 * KB, 1 * MB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_medium_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 16 * KB, 1 * MB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_large_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 2 * MB, 10 * MB, 1, 1.0, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_only_malloc_large_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 2 * MB, 10 * MB, 10, 1.0, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_small_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_small_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 128, 15 * KB, 10, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_DEFAULT, 16 * KB, 1 * MB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_DEFAULT_random_malloc80_free20_random_large_allocations_10_thread)
+{
+    run_test(MEMKIND_DEFAULT, 2 * MB, 10 * MB,  10, 0.8, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_small_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_small_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 1);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_medium_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 10);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_large_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 1, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_only_malloc_large_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 10, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_small_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_small_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 128, 15 * KB, 10, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_medium_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 1, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_medium_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 16 * KB, 1 * MB, 10, 0.8);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_large_allocations_1_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 1, 0.8, 100);
+}
+
+TEST_F(MemoryFootprintTest,
+       test_TC_MEMKIND_HBW_random_malloc80_free20_random_large_allocations_10_thread)
+{
+    run_test(MEMKIND_HBW, 2 * MB, 10 * MB, 10, 0.8, 100);
+}
diff --git a/test/memory_manager.h b/test/memory_manager.h
new file mode 100644
index 0000000..e44dd8d
--- /dev/null
+++ b/test/memory_manager.h
@@ -0,0 +1,81 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include <memkind.h>
+#include <new>
+
+class MemoryManager
+{
+private:
+    memkind_t kind;
+    size_t memory_size;
+    void *memory_pointer;
+
+    void move(MemoryManager &&other)
+    {
+        kind = other.kind;
+        memory_size = other.memory_size;
+        if (memory_pointer)
+            memkind_free(kind, memory_pointer);
+        memory_pointer = std::move(other.memory_pointer);
+        other.memory_pointer = nullptr;
+    }
+
+public:
+    MemoryManager(memkind_t kind, size_t size) :
+        kind(kind),
+        memory_size(size),
+        memory_pointer(memkind_malloc(kind, size))
+    {
+        if(!memory_pointer) {
+            throw std::bad_alloc();
+        }
+    }
+
+    size_t size()
+    {
+        return memory_size;
+    }
+
+    MemoryManager(const MemoryManager &) = delete;
+
+    MemoryManager(MemoryManager &&other)
+    {
+        move(std::move(other));
+    }
+
+    MemoryManager &operator=(MemoryManager &&other)
+    {
+        move(std::move(other));
+        return *this;
+    }
+
+    ~MemoryManager()
+    {
+        if (memory_pointer)
+            memkind_free(kind, memory_pointer);
+    }
+};
+
diff --git a/test/multithreaded_tests.cpp b/test/multithreaded_tests.cpp
new file mode 100644
index 0000000..8c6b8a8
--- /dev/null
+++ b/test/multithreaded_tests.cpp
@@ -0,0 +1,87 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+
+#include <fstream>
+#include <algorithm>
+
+#include "common.h"
+#include "check.h"
+#include "omp.h"
+#include "trial_generator.h"
+
+#define NTHREADS 2
+
+/* Multithread test content which will also use the trial_benerator to call
+ * malloc, calloc, memalign, realloc, with different memory sizes but this time
+ * it will use Open MP threads with NTHREADS set to 2
+ */
+class MultithreadedTest : public TGTest
+{
+};
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_Malloc_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_MALLOC);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_Calloc_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_CALLOC);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_Memalign_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_MEMALIGN);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
+
+TEST_F(MultithreadedTest,
+       test_TC_MEMKIND_Multithread_HBW_MemalignPsize_2bytes_2KB_2MB_sizes)
+{
+    tgen->generate_size_2bytes_2KB_2MB(HBW_MEMALIGN_PSIZE);
+    #pragma omp parallel num_threads(NTHREADS)
+    {
+        tgen->run(num_bandwidth,
+                  bandwidth);
+    }
+}
diff --git a/test/negative_tests.cpp b/test/negative_tests.cpp
new file mode 100644
index 0000000..02ea1ab
--- /dev/null
+++ b/test/negative_tests.cpp
@@ -0,0 +1,593 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind.h>
+
+#include <fstream>
+#include <algorithm>
+#include <numa.h>
+#include <errno.h>
+#include <limits.h>
+#include <sys/sysinfo.h>
+
+#include "common.h"
+#include "check.h"
+#include "omp.h"
+#include "trial_generator.h"
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+/* Set of negative test cases for memkind, its main goal are to verify that the
+ * library behaves accordingly to documentation when calling an API with
+ * invalid inputs, incorrect usage, NULL pointers.
+ */
+class NegativeTest: public ::testing::Test
+{};
+
+class NegativeTestHuge: public ::testing::Test
+{
+private:
+    //Enable huge pages to avoid false positive test result.
+    HugePageOrganizer huge_page_organizer = HugePageOrganizer(8);
+};
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_zero_memtype)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  memkind_memtype_t(), //Set incorrect value.
+                  MEMKIND_POLICY_PREFERRED_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_incorrect_memtype)
+{
+    memkind_memtype_t memtype_flags;
+    //Set incorrect value.
+    memset(&memtype_flags, -1, sizeof(memtype_flags));
+
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  memtype_flags,
+                  MEMKIND_POLICY_PREFERRED_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_incorrect_policy)
+{
+    memkind_policy_t policy;
+    //Set incorrect value.
+    memset(&policy, -1, sizeof(policy));
+
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  policy,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_incorrect_mask)
+{
+    memkind_bits_t flags;
+
+    //Set incorrect value.
+    memset(&flags, 255, sizeof(flags));
+
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_PREFERRED_LOCAL,
+                  flags,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_LOCAL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_LOCAL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_LOCAL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_ALL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_ALL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_BIND_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_BIND_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_LOCAL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_LOCAL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_ALL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_INTERLEAVE_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_DEFAULT,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_HIGH_BANDWIDTH_INTERLEAVE_LOCAL)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_HIGH_BANDWIDTH,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_HIGH_BANDWIDTH_INTERLEAVE_LOCAL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_HIGH_BANDWIDTH,
+                  MEMKIND_POLICY_INTERLEAVE_LOCAL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_HIGH_BANDWIDTH_INTERLEAVE_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int ret = memkind_create_kind(
+                  MEMKIND_MEMTYPE_HIGH_BANDWIDTH,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_HIGH_BANDWIDTH_BIND_ALL)
+{
+    memkind_t kind;
+    int flags_tmp = MEMKIND_MEMTYPE_DEFAULT | MEMKIND_MEMTYPE_HIGH_BANDWIDTH;
+    memkind_memtype_t memtype_flags;
+    memcpy(&memtype_flags, &flags_tmp, sizeof(memtype_flags));
+
+    int ret = memkind_create_kind(
+                  memtype_flags,
+                  MEMKIND_POLICY_BIND_ALL,
+                  memkind_bits_t(),
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_create_kind_DEFAULT_HIGH_BANDWIDTH_INTERLEAVE_ALL_PAGE_SIZE_2MB)
+{
+    memkind_t kind;
+    int flags_tmp = MEMKIND_MEMTYPE_DEFAULT | MEMKIND_MEMTYPE_HIGH_BANDWIDTH;
+    memkind_memtype_t memtype_flags;
+    memcpy(&memtype_flags, &flags_tmp, sizeof(memtype_flags));
+
+    int ret = memkind_create_kind(
+                  memtype_flags,
+                  MEMKIND_POLICY_INTERLEAVE_ALL,
+                  MEMKIND_MASK_PAGE_SIZE_2MB,
+                  &kind);
+    ASSERT_EQ(ret, MEMKIND_ERROR_INVALID);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_ErrorMemAlign)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = EINVAL;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_DEFAULT,
+                                 &ptr,
+                                 5,
+                                 100);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_DefaultSizeZero)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = 0;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_DEFAULT,
+                                 &ptr,
+                                 16,
+                                 0);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_ErrorAlignment)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = EINVAL;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_HBW,
+                                 &ptr,
+                                 5,
+                                 100);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_HBWSizeZero)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = 0;
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_HBW,
+                                 &ptr,
+                                 16,
+                                 0);
+    EXPECT_EQ(err, ret);
+    EXPECT_EQ(errno, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_ErrorAllocM)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = ENOMEM;
+    struct sysinfo info;
+    unsigned long long MemTotal = 0;
+
+    ret = sysinfo(&info);
+    EXPECT_EQ(ret, 0);
+
+    //Determine total memory size as totalram (total usable main memory size)
+    //multiplied by mem_unit (memory unit size in bytes). This value is equal
+    //to MemTotal field in /proc/meminfo.
+    MemTotal = info.totalram * info.mem_unit;
+
+    RecordProperty("MemTotal_kB", MemTotal/KB);
+
+    errno = 0;
+    ret = memkind_posix_memalign(MEMKIND_HBW,
+                                 &ptr,
+                                 16,
+                                 2*MemTotal);
+    EXPECT_EQ(err, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_malloc_over_size)
+{
+    void *ptr = hbw_malloc(SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_malloc_size_zero)
+{
+    void *ptr = hbw_malloc(0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_malloc_over_size)
+{
+    void *ptr = memkind_malloc(MEMKIND_HBW, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_malloc_size_zero)
+{
+    void *ptr = memkind_malloc(MEMKIND_HBW, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_calloc_over_size)
+{
+    void *ptr = hbw_calloc(1, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_calloc_size_zero)
+{
+    void *ptr = hbw_calloc(1, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_calloc_over_size)
+{
+    void *ptr = memkind_calloc(MEMKIND_HBW, 1, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_calloc_size_zero)
+{
+    void *ptr = memkind_calloc(MEMKIND_HBW, 1, 0);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_realloc_over_size)
+{
+    void *ptr = hbw_realloc(NULL, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_realloc_over_size)
+{
+    void *ptr = memkind_realloc(MEMKIND_HBW, NULL, SIZE_MAX);
+    ASSERT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_posix_memalign_over_size)
+{
+    void *ptr = NULL;
+    int ret = hbw_posix_memalign(&ptr, 4096, SIZE_MAX);
+    EXPECT_TRUE(ptr == NULL);
+    EXPECT_EQ(ENOMEM, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_hbw_posix_memalign_size_zero)
+{
+    void *ptr = NULL;
+    errno = 0;
+
+    int ret = hbw_posix_memalign(&ptr, 4096, 0);
+    EXPECT_TRUE(ptr == NULL);
+    EXPECT_EQ(0, ret);
+    EXPECT_EQ(0, errno);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_memkind_posix_memalign_over_size)
+{
+    void *ptr = NULL;
+    int ret = memkind_posix_memalign(MEMKIND_HBW, &ptr, 4096, SIZE_MAX);
+    EXPECT_TRUE(ptr == NULL);
+    EXPECT_EQ(ENOMEM, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBFailureMemalign)
+{
+    int ret = 0;
+    void *ptr = NULL;
+    int err = EINVAL;
+
+    ret = hbw_posix_memalign_psize(&ptr,
+                                   1073741824,
+                                   1073741826,
+                                   HBW_PAGESIZE_1GB_STRICT);
+    EXPECT_EQ(ret, err);
+    EXPECT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_RegularReallocWithMemAllign)
+{
+    int ret = 0;
+    void *ptr = NULL;
+
+    ret = hbw_posix_memalign_psize(&ptr,
+                                   4096,
+                                   4096,
+                                   HBW_PAGESIZE_4KB);
+    EXPECT_EQ(ret, 0);
+    ASSERT_TRUE(ptr != NULL);
+    memset(ptr, 0, 4096);
+    ptr = hbw_realloc(ptr, 8192);
+    memset(ptr, 0, 8192);
+    hbw_free(ptr);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicy)
+{
+    // First call should be successful, consequent should generate a warning
+    // and be ignored
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_PREFERRED), 0);
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_INTERLEAVE), EPERM);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+    EXPECT_EQ(hbw_set_policy((hbw_policy_t)0xFF), EINVAL);
+}
+
+//Check if hbw_set_policy() will be ignored after malloc.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterMalloc)
+{
+    void *ptr = hbw_malloc(512);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after calloc.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterCalloc)
+{
+    void *ptr = hbw_calloc(512, 1);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after realloc.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterRealloc)
+{
+    void *ptr = hbw_malloc(512);
+    EXPECT_TRUE(ptr != NULL);
+
+    hbw_realloc(ptr, 512);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after hbw_posix_memalign.
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_SetPolicyAfterHbwPosixMemalign)
+{
+    void *ptr = NULL;
+
+    hbw_posix_memalign(&ptr, 2048, 2048);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+//Check if hbw_set_policy() will be ignored after hbw_posix_memalign_psize.
+TEST_F(NegativeTest,
+       test_TC_MEMKIND_Negative_SetPolicyAfterHbwPosixMemalignPsize)
+{
+    void *ptr = NULL;
+
+    hbw_posix_memalign_psize(&ptr, 2048, 2048, HBW_PAGESIZE_4KB);
+    EXPECT_TRUE(ptr != NULL);
+
+    EXPECT_EQ(hbw_set_policy(HBW_POLICY_BIND), EPERM);
+    EXPECT_NE(hbw_get_policy(), HBW_POLICY_BIND);
+    EXPECT_EQ(hbw_get_policy(), HBW_POLICY_PREFERRED);
+
+    hbw_free(ptr);
+}
+
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBMemalignPsizeAllign)
+{
+    void *ptr = NULL;
+    int ret = 0;
+    int err = EINVAL;
+
+    ret = hbw_posix_memalign_psize(&ptr, -1, 1024, HBW_PAGESIZE_1GB);
+    EXPECT_EQ(err, ret);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBNullRealloc)
+{
+    void *ptr = NULL;
+    ptr = memkind_realloc(MEMKIND_HBW_GBTLB, NULL, -1);
+    EXPECT_TRUE(ptr == NULL);
+}
+
+TEST_F(NegativeTest, test_TC_MEMKIND_Negative_GBNullFree)
+{
+    memkind_free(MEMKIND_GBTLB, NULL);
+}
+
+TEST_F(NegativeTestHuge,
+       test_TC_MEMKIND_hbwmalloc_memalign_psize_Interleave_Policy_PAGE_SIZE_2MB)
+{
+    void *ptr = NULL;
+    hbw_set_policy(HBW_POLICY_INTERLEAVE);
+    int ret = hbw_posix_memalign_psize(&ptr, 4096, 4096, HBW_PAGESIZE_2MB);
+    ASSERT_EQ(EINVAL, ret);
+}
diff --git a/test/performance/framework.cpp b/test/performance/framework.cpp
new file mode 100644
index 0000000..aca24f0
--- /dev/null
+++ b/test/performance/framework.cpp
@@ -0,0 +1,357 @@
+/*
+* Copyright (C) 2014 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include <algorithm> // sort
+#include <math.h> // log2
+#include <cassert>
+#include <iostream>
+#include "framework.hpp"
+
+namespace performance_tests
+{
+    namespace ch = std::chrono;
+    using std::cout;
+    using std::endl;
+    using std::unique_lock;
+    using std::mutex;
+
+
+#ifdef __DEBUG
+    mutex g_coutMutex;
+    int g_msgLevel = 1;
+#endif
+
+    void Barrier::wait()
+    {
+        unique_lock<mutex> lock(m_barrierMutex);
+        // Decrement number of threads awaited at the barrier
+        m_waiting--;
+        if (m_waiting == 0) {
+            // Called by the last expected thread - notify all waiting threads and exit
+            m_cVar.notify_all();
+            // Store the time when barrier was released
+            if (m_releasedAt.tv_sec == 0 && m_releasedAt.tv_nsec == 0) {
+                clock_gettime(CLOCK_MONOTONIC, &m_releasedAt);
+            }
+            return;
+        }
+        // Wait unitl the last expected thread calls wait() on Barrier instance, or timeout occurs
+        m_cVar.wait_until(lock, ch::system_clock::now() + ch::seconds(10), []() {
+            return GetInstance().m_waiting == 0;
+        });
+    }
+
+    // Worker class
+    Worker::Worker(
+        uint32_t actionsCount,
+        const vector<size_t> &allocationSizes,
+        Operation *freeOperation,
+        memkind_t kind)
+        : m_actionsCount(actionsCount)
+        , m_allocationSizes(allocationSizes)
+        , m_actions(vector<Action*>(actionsCount, nullptr))
+        , m_kind(kind)
+    {
+        assert(freeOperation->getName() == OperationName::Free);
+    }
+
+    Worker::~Worker()
+    {
+        for (Action *action : m_actions) { //each action
+            delete action;
+        }
+
+    }
+
+    void Worker::init(const vector<Operation *> &testOperations,
+                      Operation *&freeOperation)
+    {
+        for(uint32_t i = 0 ; i < m_actionsCount ; i++) {
+            int bucketSize = rand() % Operation::MaxBucketSize;
+
+            for (Operation *operation : testOperations) { //each operation
+                if (operation->checkCondition(bucketSize)) {
+                    size_t size = m_allocationSizes[m_allocationSizes.size() > 1 ? rand() %
+                                                                             m_allocationSizes.size() : 0];
+                    m_actions[i] = new Action(
+                        operation,
+                        freeOperation,
+                        m_kind,
+                        size,
+                        log2(rand() % size),
+                        sizeof(void *) * (1 << ((rand() % Operation::MemalignMaxMultiplier))));
+                    break;
+                }
+            }
+        }
+    }
+
+    void Worker::run()
+    {
+        m_thread = new thread(&Worker::work, this);
+    }
+
+#ifdef __DEBUG
+    uint16_t Worker::getId()
+    {
+        return m_threadId;
+    }
+    void Worker::setId(uint16_t threadId)
+    {
+        m_threadId = threadId;
+    }
+#endif
+
+    void Worker::finish()
+    {
+        if (m_thread != nullptr) {
+            m_thread->join();
+            delete m_thread;
+        }
+    }
+
+    void Worker::work()
+    {
+        EMIT(1, "Entering barrier " << m_threadId)
+        Barrier::GetInstance().wait();
+        EMIT(1, "Starting thread " << m_threadId)
+        for (Action *action : m_actions) {
+            action->alloc();
+        }
+    }
+
+    void Worker::clean()
+    {
+        EMIT(2, "Cleaning thread " << m_threadId)
+        for (Action *action : m_actions) {
+            action->free();
+        }
+        EMIT(1, "Thread " << m_threadId << " finished")
+    }
+
+    // PerformanceTest class
+    PerformanceTest::PerformanceTest(
+        size_t repeatsCount,
+        size_t threadsCount,
+        size_t operationsCount)
+        : m_repeatsCount(repeatsCount)
+        , m_discardCount(repeatsCount * (distardPercent / 100.0))
+        , m_threadsCount(threadsCount)
+        , m_operationsCount(operationsCount)
+        , m_executionMode(ExecutionMode::SingleInteration)
+    {
+    }
+
+    void PerformanceTest::setAllocationSizes(const vector<size_t> &allocationSizes)
+    {
+        m_allocationSizes = allocationSizes;
+    }
+
+    void PerformanceTest::setOperations(const vector<vector<Operation *>>
+                                        &testOperations, Operation *freeOperation)
+    {
+        m_testOperations = testOperations;
+        m_freeOperation = freeOperation;
+    }
+
+    void PerformanceTest::setExecutionMode(ExecutionMode executionMode)
+    {
+        m_executionMode = executionMode;
+    }
+
+    void PerformanceTest::setKind(const vector<memkind_t> &kinds)
+    {
+        m_kinds = kinds;
+    }
+
+    inline void PerformanceTest::runIteration()
+    {
+        timespec iterationStop, iterationStart;
+
+        Barrier::GetInstance().reset(m_threadsCount);
+        for (Worker *worker : m_workers) {
+            worker->run();
+        }
+        for (Worker *worker : m_workers) {
+            worker->finish();
+        }
+        EMIT(1, "Alloc completed");
+        clock_gettime(CLOCK_MONOTONIC, &iterationStop);
+        iterationStart = Barrier::GetInstance().releasedAt();
+        m_durations.push_back(
+            (iterationStop.tv_sec  * NanoSecInSec + iterationStop.tv_nsec) -
+            (iterationStart.tv_sec * NanoSecInSec + iterationStart.tv_nsec)
+        );
+        for (Worker *worker : m_workers) {
+            worker->clean();
+        }
+    }
+
+    void PerformanceTest::prepareWorkers()
+    {
+        for (size_t threadId = 0; threadId < m_threadsCount; threadId++) {
+            m_workers.push_back(
+                new Worker(
+                    m_operationsCount,
+                    m_allocationSizes,
+                    m_freeOperation,
+                    m_kinds.size() > 0 ? m_kinds[threadId % m_kinds.size()] : nullptr)
+            );
+#ifdef __DEBUG
+            m_workers.back()->setId(threadId);
+#endif
+            if (m_executionMode == ExecutionMode::SingleInteration) {
+                // In ManyIterations mode, operations will be set for each thread at the beginning of each iteration
+                m_workers.back()->init(m_testOperations[threadId % m_testOperations.size()],
+                                       m_freeOperation);
+            }
+        }
+    }
+
+    Metrics PerformanceTest::getMetrics()
+    {
+        uint64_t totalDuration = 0;
+
+        std::sort(m_durations.begin(), m_durations.end());
+
+        m_durations.erase(m_durations.end() - m_discardCount, m_durations.end());
+        for (uint64_t &duration : m_durations) {
+            totalDuration += duration;
+        }
+
+        Metrics metrics;
+
+        metrics.executedOperations   = m_durations.size() * m_threadsCount *
+                                       m_operationsCount;
+        metrics.totalDuration        = totalDuration;
+        metrics.repeatDuration       = (double) totalDuration /
+                                       ((uint64_t)m_durations.size() * NanoSecInSec);
+        metrics.iterationDuration    = metrics.repeatDuration;
+        if (m_executionMode == ExecutionMode::ManyIterations) {
+            metrics.executedOperations *= m_testOperations.size();
+            metrics.iterationDuration  /= m_testOperations.size();
+        }
+        metrics.operationsPerSecond  = (double) metrics.executedOperations *
+                                       NanoSecInSec / totalDuration;
+        metrics.avgOperationDuration = (double) totalDuration /
+                                       metrics.executedOperations;
+        assert(metrics.iterationDuration != 0.0);
+        return metrics;
+    }
+
+    void PerformanceTest::writeMetrics(const string &suiteName,
+                                       const string &caseName, const string &fileName)
+    {
+        Metrics metrics = getMetrics();
+
+        // For thousands separation
+        setlocale(LC_ALL, "");
+        if (!fileName.empty()) {
+            FILE *f;
+            if((f = fopen(fileName.c_str(), "a+"))) {
+                fprintf(f,
+                        "%s;%s;%zu;%zu;%lu;%f;%f;%f;%f\n",
+                        suiteName.c_str(),
+                        caseName.c_str(),
+                        m_repeatsCount,
+                        m_threadsCount,
+                        metrics.executedOperations,
+                        metrics.operationsPerSecond,
+                        metrics.avgOperationDuration,
+                        metrics.iterationDuration,
+                        metrics.repeatDuration);
+                fclose(f);
+            }
+
+        }
+        printf("Operations/sec:\t\t\t%'f\n"
+               "Avg. operation duration:\t%f nsec\n"
+               "Iteration duration:\t\t%f sec\n"
+               "Repeat duration:\t\t%f sec\n",
+               metrics.operationsPerSecond,
+               metrics.avgOperationDuration,
+               metrics.iterationDuration,
+               metrics.repeatDuration);
+    }
+
+    int PerformanceTest::run()
+    {
+        if (m_testOperations.empty() ||
+            m_allocationSizes.empty() ||
+            m_freeOperation == nullptr) {
+            cout << "ERROR: Test not initialized" << endl;
+            return 1;
+        }
+        // Create threads
+        prepareWorkers();
+        //warmup kinds
+        void *alloc = nullptr;
+
+        for (const memkind_t &kind : m_kinds) {
+            m_testOperations[0][0]->perform(kind, alloc, 1e6);
+            m_freeOperation->perform(kind, alloc);
+        }
+        for (size_t repeat = 0; repeat < m_repeatsCount; repeat++) {
+            EMIT(1, "Test run #" << repeat)
+            if (m_executionMode == ExecutionMode::SingleInteration) {
+                runIteration();
+            } else {
+                // Perform each operations list in separate iteration, for each thread
+                for (vector<Operation *> &ops : m_testOperations) {
+                    for (Worker *worker : m_workers) {
+                        worker->init(ops, m_freeOperation);
+                    }
+                    runIteration();
+                }
+            }
+        }
+        return 0;
+    }
+
+    void PerformanceTest::showInfo()
+    {
+        printf("Test parameters: %lu repeats, %lu threads, %d operations per thread\n",
+               m_repeatsCount,
+               m_threadsCount,
+               m_operationsCount);
+        printf("Thread memory allocation operations:\n");
+        for (unsigned long i = 0; i < m_testOperations.size(); i++) {
+            if (m_executionMode == ExecutionMode::SingleInteration) {
+                printf("\tThread %lu,%lu,...\n", i, i + (m_testOperations.size()));
+            } else {
+                printf("\tIteration %lu\n", i);
+            }
+            for (const Operation *op : m_testOperations[i]) {
+                printf("\t\t %s (bucket size: %d)\n", op->getNameStr().c_str(),
+                       op->getBucketSize());
+            }
+        }
+        printf("Memory free operation:\n\t\t%s\n",
+               m_freeOperation->getNameStr().c_str());
+        printf("Allocation sizes:\n");
+        for (size_t size : m_allocationSizes) {
+            printf("\t\t%lu bytes\n", size);
+        }
+    }
+}
diff --git a/test/performance/framework.hpp b/test/performance/framework.hpp
new file mode 100644
index 0000000..a7c2d08
--- /dev/null
+++ b/test/performance/framework.hpp
@@ -0,0 +1,280 @@
+/*
+* Copyright (C) 2014 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <memkind.h>
+
+#include <thread>
+#include <vector>
+#include <mutex>
+#include <condition_variable>
+// Malloc, jemalloc, memkind jemalloc and memkind memory operations definitions
+#include "operations.hpp"
+
+/* Framework for testing memory allocators pefromance */
+namespace performance_tests
+{
+// Nanoseconds in second
+    const uint32_t NanoSecInSec = 1e9;
+
+    // Simple barrier implementation
+    class Barrier
+    {
+        // Barrier mutex
+        std::mutex m_barrierMutex;
+        // Contitional variable
+        std::condition_variable m_cVar;
+        // Number of threads expected to enter the barrier
+        size_t m_waiting;
+        timespec m_releasedAt;
+
+    public:
+        // Called by each thread entering the barrier; returns control to caller
+        // only after been called from the last thread expected at the barrier
+        void wait();
+
+        // (Re)Initializes the barrier
+        void reset(unsigned waiting)
+        {
+            m_releasedAt.tv_sec = m_releasedAt.tv_nsec = 0;
+            m_waiting = waiting;
+        }
+
+        // Get time when barrier was released
+        timespec &releasedAt()
+        {
+            return m_releasedAt;
+        }
+
+        // Singleton
+        static Barrier &GetInstance()
+        {
+            // Automatically created and deleted one and only instance
+            static Barrier instance;
+            return instance;
+        }
+    private:
+        Barrier()
+        {
+            reset(0);
+        }
+        // Cannot be used with singleton, so prevent compiler from creating them automatically
+        Barrier(Barrier const &)        = delete;
+        void operator=(Barrier const &) = delete;
+    };
+
+    // Data of a single test action, that is, memory operation (malloc, calloc, etc.)
+    // to perform and its parameters (size, alignment etc.)
+    class Action
+    {
+    protected:
+        Operation *m_operation;
+        Operation *m_freeOperation;
+        const memkind_t m_kind;
+        void *m_allocation;
+        const size_t m_size;
+        const size_t m_offset;
+        const size_t m_alignment;
+
+    public:
+        Action(
+            Operation *operation,
+            Operation *freeOperation,
+            const memkind_t kind,
+            const size_t size,
+            const size_t offset,
+            const size_t alignment)
+            : m_operation(operation)
+            , m_freeOperation(freeOperation)
+            , m_kind(kind)
+            , m_allocation(nullptr)
+            , m_size(size)
+            , m_offset(offset)
+            , m_alignment(alignment)
+        {}
+
+        Action(
+            Operation *operation,
+            Operation *freeOperation,
+            const memkind_t kind)
+            : Action(operation, freeOperation, kind, 0, 0, 0)
+        {
+        }
+
+        void alloc()
+        {
+            m_operation->perform(m_kind, m_allocation, m_size, m_offset, m_alignment);
+        }
+
+        void free()
+        {
+            m_freeOperation->perform(m_kind, m_allocation);
+        }
+
+    };
+
+    // Performs and tracks requested memory operations in a separate thread
+    class Worker
+    {
+    protected:
+#ifdef __DEBUG
+        uint16_t m_threadId;
+#endif
+        // Requested number of test actions
+        const uint32_t m_actionsCount;
+        // List of memory block sizes - for each memory allocation operation actual value is chosen randomly
+        const vector<size_t> &m_allocationSizes;
+        // List of test actions
+        vector<Action *> m_actions;
+        // Memory free action
+        Action *m_freeAction;
+        // Operation kind (useful for memkind only)
+        memkind_t m_kind;
+        // Working thread
+        thread *m_thread;
+
+    public:
+        Worker(
+            uint32_t actionsCount,
+            const vector<size_t> &allocationSizes,
+            Operation *freeOperation,
+            memkind_t kind);
+
+        ~Worker();
+
+        // Set operations list for the worker
+        void init(const vector<Operation *> &testOperations, Operation *&freeOperation);
+
+        // Create & start thread
+        void run();
+
+#ifdef __DEBUG
+        // Get thread id
+        uint16_t getId();
+
+        // Set thread id
+        void setId(uint16_t threadId);
+#endif
+
+        // Finish thread and free all allocations made
+        void finish();
+
+        // Free allocated memory
+        virtual void clean();
+
+    private:
+        // Actual thread function (allow inheritance)
+        virtual void work();
+    };
+
+    enum ExecutionMode {
+        SingleInteration, // Single iteration, operations listS will be distributed among threads sequentially
+        ManyIterations    // Each operations list will be run in separate iteration by each thread
+    };
+
+    struct Metrics {
+        uint64_t executedOperations;
+        uint64_t totalDuration;
+        double operationsPerSecond;
+        double avgOperationDuration;
+        double iterationDuration;
+        double repeatDuration;
+    };
+
+    // Performance test parameters class
+    class PerformanceTest
+    {
+    protected:
+        // empirically determined % of worst results needed to be discarded
+        // to eliminate malloc() performance results skewness
+        static constexpr double distardPercent = 20.0;
+    protected:
+        // Number of test repeats
+        size_t                      m_repeatsCount;
+        // Number of test repeats with worst results to be discarded
+        size_t                      m_discardCount;
+        // Number of threads
+        size_t                      m_threadsCount;
+        // Number of memory operations in each thread
+        uint32_t                    m_operationsCount;
+        // List of allocation sizes
+        vector<size_t>              m_allocationSizes;
+        // List of list of allocation operations, utlization depends on execution mode
+        vector<vector<Operation *>> m_testOperations;
+        // Free operation
+        Operation                  *m_freeOperation;
+        // List of memory kinds (for memkind allocation only)
+        // distributed among threads sequentially
+        vector<memkind_t>           m_kinds;
+        // List of thread workers
+        vector<Worker *>            m_workers;
+        // Time measurement
+        vector<uint64_t>            m_durations;
+        // Execution mode
+        ExecutionMode               m_executionMode;
+
+    public:
+        // Create test
+        PerformanceTest(
+            size_t repeatsCount,
+            size_t threadsCount,
+            size_t operationsCount);
+
+        virtual ~PerformanceTest() {}
+
+        // Set list of block sizes
+        void setAllocationSizes(const vector<size_t> &allocationSizes);
+
+        // Set list of operations per thread/per iteration (depending on execution mode)
+        void setOperations(const vector<vector<Operation *>> &testOperations,
+                           Operation *freeOperation);
+
+        // Set per-thread list of memory kinds
+        void setKind(const vector<memkind_t> &kinds);
+
+        // Set execution mode (different operations per each thread/same operations for each thread, but many iterations)
+        void setExecutionMode(ExecutionMode operationMode);
+
+        // Execute test
+        int run();
+
+        // Print test parameters
+        virtual void showInfo();
+
+        // Write test metrics
+        void writeMetrics(const string &suiteName, const string &caseName,
+                          const string &fileName = "");
+
+        Metrics getMetrics();
+    private:
+
+        // Run single iteration
+        void runIteration();
+
+        // Setup thread workers
+        void prepareWorkers();
+
+    };
+}
diff --git a/test/performance/operations.hpp b/test/performance/operations.hpp
new file mode 100644
index 0000000..7a5d6b6
--- /dev/null
+++ b/test/performance/operations.hpp
@@ -0,0 +1,477 @@
+/*
+* Copyright (C) 2014 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include <memkind.h>
+
+#include <malloc.h>
+#include <cassert>
+#include "jemalloc/jemalloc.h"
+
+// Malloc, jemalloc, memkind jemalloc and memkind memory operations definitions
+namespace performance_tests
+{
+    using std::vector;
+    using std::string;
+    using std::thread;
+
+#ifdef __DEBUG
+#include <mutex>
+    // Write entire text at once, avoiding switching to another thread
+    extern int g_msgLevel;
+    extern std::mutex g_coutMutex;
+#define EMIT(LEVEL, TEXT) \
+    if (g_msgLevel >= LEVEL) \
+    { \
+        g_coutMutex.lock(); std::cout << TEXT << std::endl; g_coutMutex.unlock(); \
+    }
+#else
+#define EMIT(LEVEL, TEXT)
+#endif
+
+    // Use jemalloc, compiled with unique prefix (--with-jemalloc-prefix= configure option)
+#ifdef SYSTEM_JEMALLOC_PREFIX
+#define TOKENPASTE(x, y) x ## y
+#define JE(x, y) TOKENPASTE(x, y)
+#define jexx_malloc JE(SYSTEM_JEMALLOC_PREFIX, malloc)
+#define jexx_calloc JE(SYSTEM_JEMALLOC_PREFIX, calloc)
+#define jexx_memalign JE(SYSTEM_JEMALLOC_PREFIX, memalign)
+#define jexx_realloc JE(SYSTEM_JEMALLOC_PREFIX, realloc)
+#define jexx_free JE(SYSTEM_JEMALLOC_PREFIX, free)
+    extern "C" {
+        // jemalloc function prototypes
+        // full header cannot be include due to conflict with memkind jemalloc
+        extern void *jexx_malloc(size_t size);
+        extern void *jexx_calloc(size_t num, size_t size);
+        extern void *jexx_memalign(size_t alignment, size_t size);
+        extern void *jexx_realloc(void *ptr, size_t size);
+        extern void  jexx_free(void *ptr);
+    }
+#endif
+
+    // Available memory operations
+    enum OperationName {
+        Malloc,
+        Calloc,
+        Realloc,
+        Align,
+        Free,
+        Invalid
+    };
+
+    // Reprents a memory operation
+    class Operation
+    {
+    public:
+        // Each operation is assigned a bucket size from range (0, MaxBucketSize)
+        static const unsigned MaxBucketSize = 100;
+        // For memalign operation, alignment parameter will be a random value
+        // from range (sizeof(void*), sizeof(void*) * MemalignMaxMultiplier)
+        static const unsigned MemalignMaxMultiplier = 4;
+
+    protected:
+        OperationName m_name;
+        // If random number from range (0, MaxBucketSize) is lower than m_bucketSize, an operation will be performed
+        unsigned m_bucketSize;
+
+    public:
+        Operation()
+        {
+        }
+
+        // Creates an operation with given name and bucket size
+        // If no bucket size is given, operation will be always performed
+        Operation(
+            OperationName name,
+            unsigned bucketSize = MaxBucketSize)
+            : m_name(name)
+            , m_bucketSize(bucketSize)
+        {
+            assert(bucketSize <= MaxBucketSize);
+        }
+
+        virtual ~Operation() {}
+        ;
+
+        // Check if operation should be performed (currently drawn random number lower than bucket size)
+        bool checkCondition(unsigned ballSize) const
+        {
+            return (ballSize < m_bucketSize);
+        }
+
+        OperationName getName() const
+        {
+            return m_name;
+        }
+
+        string getNameStr() const
+        {
+            switch (m_name) {
+                case OperationName::Malloc:
+                    return "malloc";
+
+                case OperationName::Calloc:
+                    return "calloc";
+
+                case OperationName::Realloc:
+                    return "realloc";
+
+                case OperationName::Align:
+                    return "align";
+
+                case OperationName::Free:
+                    return "free";
+
+                default:
+                    return "<unknown>";
+            }
+        }
+
+        // Get operation bucket size
+        unsigned getBucketSize() const
+        {
+            return m_bucketSize;
+        }
+
+        // perform memory operation
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size = 0,
+                             size_t offset=0,
+                             size_t alignment=0)
+        const = 0;
+
+    };
+
+    // Malloc memory operations
+    class MallocOperation : public Operation
+    {
+    public:
+        MallocOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        MallocOperation(
+            OperationName name,
+            unsigned bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch(m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        free(mem);
+                    }
+                    mem = malloc(size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        free(mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = calloc((1 << offset), (size >> offset));
+                    break;
+                }
+                case Realloc: {
+                    mem = realloc(mem, size);
+                    break;
+                }
+                case Align: {
+                    if (mem != nullptr) {
+                        free(mem);
+                    }
+                    // randomly choose alignment from (8, 8 * MemalignMaxMultiplie)
+                    mem = memalign(alignment, size);
+                    break;
+                }
+                case Free: {
+                    free(mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    throw "Not implemented";
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+        }
+    };
+
+#ifdef SYSTEM_JEMALLOC_PREFIX
+    // Jemalloc memory operations
+    class JemallocOperation : public Operation
+    {
+    public:
+        JemallocOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        JemallocOperation(
+            OperationName name,
+            unsigned bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch(m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        jexx_free(mem);
+                    }
+                    mem = jexx_malloc(size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        jexx_free(mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = jexx_calloc((1 <<, offset), (size >>, offset));
+                    break;
+                }
+                case Realloc: {
+                    mem = jexx_realloc(mem, size);
+                    break;
+                }
+                case Align: {
+                    if (mem != nullptr) {
+                        jexx_free(mem);
+                    }
+                    // randomly choose alignment from (8, 8 * MemalignMaxMultiplie)
+                    mem = jexx_memalign(alignment, size);
+                    break;
+                }
+                case Free: {
+                    jexx_free(mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    throw "Not implemented";
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+        }
+    };
+#endif
+    // Jemkmalloc memory operations
+    class JemkmallocOperation : public Operation
+    {
+    public:
+        JemkmallocOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        JemkmallocOperation(
+            OperationName name,
+            unsigned bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+#ifdef JEMK
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch(m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        jemk_free(mem);
+                    }
+                    mem = jemk_malloc(size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        jemk_free(mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = jemk_calloc((1 << offset), (size >> offset));
+                    break;
+                }
+                case Realloc: {
+                    mem = jemk_realloc(mem, size);
+                    break;
+                }
+                case Align: {
+                    if (mem != nullptr) {
+                        jemk_free(mem);
+                    }
+                    // randomly choose alignment from (8, 8 * MemalignMaxMultiplie)
+                    mem = jemk_memalign(alignment, size);
+                    break;
+                }
+                case Free: {
+                    jemk_free(mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    throw "Not implemented";
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+#endif // JE_MK
+        }
+    };
+
+    // Memkind memory operations
+    class MemkindOperation : public Operation
+    {
+    public:
+        MemkindOperation()
+        {}
+
+        MemkindOperation(
+            OperationName name)
+            : Operation(name)
+        {
+        }
+
+        MemkindOperation(
+            OperationName name,
+            size_t bucketSize)
+            : Operation(name, bucketSize)
+        {
+        }
+
+        virtual void perform(const memkind_t &kind,
+                             void *&mem,
+                             size_t size,
+                             size_t offset,
+                             size_t alignment) const override
+        {
+            EMIT(2, "Entering Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+            switch (m_name) {
+                case Malloc: {
+                    if (mem != nullptr) {
+                        memkind_free(kind, mem);
+                    }
+                    mem = memkind_malloc(kind, size);
+                    break;
+                }
+                case Calloc: {
+                    if (mem != nullptr) {
+                        memkind_free(kind, mem);
+                    }
+                    // split allocation size randomly
+                    // between number of elements and element size
+                    mem = memkind_calloc(kind, 1 << offset, size >> offset);
+                    break;
+                }
+                case Realloc: {
+                    mem = memkind_realloc(kind, mem, size);
+                    break;
+                }
+
+                case Align: {
+                    if (mem != nullptr) {
+                        memkind_free(kind, mem);
+                        mem = nullptr;
+                    }
+                    // randomly choose alignment from (sizeof(void*), sizeof(void*) * MemalignMaxMultiplie)
+                    if (memkind_posix_memalign(kind, &mem, alignment, size) != 0) {
+                        // failure
+                        mem = nullptr;
+                    }
+                    break;
+                }
+                case Free: {
+                    memkind_free(kind, mem);
+                    mem = nullptr;
+                    break;
+                }
+
+                default:
+                    break;
+            }
+            EMIT(2, "Exiting Operation::" << getNameStr()
+                 <<  ", size=" << size
+                 << ", offset=" << offset
+                 << ", alignment=" << alignment
+                 << ", mem=" << mem)
+        }
+    };
+}
diff --git a/test/performance/perf_tests.cpp b/test/performance/perf_tests.cpp
new file mode 100644
index 0000000..6018337
--- /dev/null
+++ b/test/performance/perf_tests.cpp
@@ -0,0 +1,162 @@
+/*
+* Copyright (C) 2014 - 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "perf_tests.hpp"
+#include <iostream>
+#include <cmath>
+#include <gtest/gtest.h>
+
+// Memkind performance tests
+using std::cout;
+using std::endl;
+using std::abs;
+using std::ostringstream;
+
+// Memkind tests
+class PerformanceTest : public testing::Test
+{
+protected:
+    const double Tolerance = 0.15;
+    const double Confidence = 0.10;
+
+    Metrics referenceMetrics;
+    Metrics performanceMetrics;
+    PerfTestCase<performance_tests::MallocOperation> referenceTest;
+    PerfTestCase<performance_tests::MemkindOperation> performanceTest;
+
+    template<class T>
+    static void RecordProperty(const string &key, const T value)
+    {
+        ostringstream values;
+        values << value;
+        testing::Test::RecordProperty(key, values.str());
+    }
+
+    void SetUp()
+    {
+    }
+
+    void TearDown()
+    {
+    }
+
+    bool checkDelta(double value, double reference, const string &info,
+                    double delta, bool higherIsBetter = false)
+    {
+        // If higherIsBetter is true, current value should be >= (reference - delta); otherwise, it should be <= (reference + delta)
+        double threshold = higherIsBetter ? reference * (1 - delta) : reference *
+                           (1 + delta);
+        cout <<
+             "Metric: " << info << ". Reference value: " << reference << ". "
+             "Expected: " << (higherIsBetter ? ">= " : "<= ") << threshold << " (delta = " <<
+             delta << ")."
+             "Actual: " << value << " (delta = " <<
+             ((value > reference ? (value / reference) : (reference / value)) - 1.0) /
+             ((higherIsBetter != (value > reference)) ? 1.0 : -1.0)
+             << ")."
+             << endl ;
+        if (higherIsBetter ? (value <= threshold) : (value >= threshold)) {
+            cout << "WARNING : Value of '" << info << "' outside expected bounds!" << endl;
+            return false;
+        }
+        return true;
+    }
+
+    bool compareMetrics(Metrics metrics, Metrics reference, double delta)
+    {
+        bool result = true;
+        result &= checkDelta(metrics.operationsPerSecond, reference.operationsPerSecond,
+                             "operationsPerSecond", delta, true);
+        result &= checkDelta(metrics.avgOperationDuration,
+                             reference.avgOperationDuration, "avgOperationDuration", delta);
+        return result;
+    }
+
+    void writeMetrics(Metrics metrics, Metrics reference)
+    {
+        RecordProperty("ops_per_sec", metrics.operationsPerSecond);
+        RecordProperty("ops_per_sec_vs_ref",
+                       (reference.operationsPerSecond - metrics.operationsPerSecond) * 100.0f
+                       / reference.operationsPerSecond);
+        RecordProperty("avg_op_time_nsec", metrics.avgOperationDuration);
+        RecordProperty("avg_op_time_nsec_vs_ref",
+                       (metrics.avgOperationDuration - reference.avgOperationDuration) * 100.0f
+                       / reference.avgOperationDuration);
+    }
+
+    void run()
+    {
+        cout << "Running reference std::malloc test" << endl;
+        referenceMetrics = referenceTest.runTest();
+
+        cout << "Running memkind test" << endl;
+        performanceMetrics = performanceTest.runTest();
+
+        writeMetrics(performanceMetrics, referenceMetrics);
+        EXPECT_TRUE(compareMetrics(performanceMetrics, referenceMetrics,
+                                   Tolerance + Confidence));
+    }
+};
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_single_op_single_iter)
+{
+    referenceTest.setupTest_singleOpSingleIter();
+    performanceTest.setupTest_singleOpSingleIter();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_single_iter)
+{
+    referenceTest.setupTest_manyOpsSingleIter();
+    performanceTest.setupTest_manyOpsSingleIter();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_single_iter_huge_alloc)
+{
+    referenceTest.setupTest_manyOpsSingleIterHugeAlloc();
+    performanceTest.setupTest_manyOpsSingleIterHugeAlloc();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_single_op_many_iters)
+{
+    referenceTest.setupTest_singleOpManyIters();
+    performanceTest.setupTest_singleOpManyIters();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_many_iters)
+{
+    referenceTest.setupTest_manyOpsManyIters();
+    performanceTest.setupTest_manyOpsManyIters();
+    run();
+}
+
+TEST_F(PerformanceTest, test_TC_MEMKIND_perf_many_ops_many_iters_many_kinds)
+{
+    referenceTest.setupTest_manyOpsManyIters();
+    performanceTest.setupTest_manyOpsManyIters();
+    run();
+}
diff --git a/test/performance/perf_tests.hpp b/test/performance/perf_tests.hpp
new file mode 100644
index 0000000..8cb5bd5
--- /dev/null
+++ b/test/performance/perf_tests.hpp
@@ -0,0 +1,190 @@
+/*
+* Copyright (C) 2014 - 2018 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#pragma once
+
+#include "framework.hpp"
+#include "operations.hpp"
+
+using std::vector;
+using std::string;
+
+using performance_tests::Operation;
+using performance_tests::OperationName;
+using performance_tests::Metrics;
+using performance_tests::ExecutionMode;
+
+template <class T>
+class PerfTestCase
+{
+
+private:
+    vector<vector<Operation *>> m_testOperations;
+    Operation *m_freeOperation;
+    performance_tests::PerformanceTest *m_test;
+    const unsigned m_seed = 1297654;
+    const unsigned m_repeats = 50;
+    const unsigned m_threads = 64;
+    const unsigned m_iterations = 100;
+
+public:
+    PerfTestCase()
+    {
+        m_freeOperation = new T(OperationName::Free);
+        srand(m_seed);
+        m_test = nullptr;
+    }
+
+    ~PerfTestCase()
+    {
+        delete m_test;
+        for (vector<Operation *> ops : m_testOperations) {
+            for (Operation *op : ops) {
+                delete op;
+            }
+        }
+    }
+
+    // Perform common actions for all test cases
+    Metrics runTest(vector<memkind_t> kinds = { MEMKIND_DEFAULT })
+    {
+        m_test->setKind(kinds);
+        m_test->showInfo();
+        m_test->run();
+        return m_test->getMetrics();
+    }
+
+    // malloc only
+    // 128 bytes
+
+    void setupTest_singleOpSingleIter()
+    {
+
+        m_testOperations = { { new T(OperationName::Malloc) } };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 128 });
+    }
+
+    // malloc, calloc, realloc and align (equal probability)
+    // 120, 521, 1200 and 4099 bytes
+    void setupTest_manyOpsSingleIter()
+    {
+        m_testOperations = { {
+                new T(OperationName::Malloc, 25),
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 120, 521, 1200, 4099 });
+    }
+
+    // malloc, calloc, realloc and align (equal probability)
+    // 500000, 1000000, 2000000 and 4000000 bytes
+    void setupTest_manyOpsSingleIterHugeAlloc()
+    {
+        m_testOperations = { {
+                new T(OperationName::Malloc, 25),
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 500000, 1000000, 2000000, 4000000 });
+    }
+
+    // 4 iterations of each thread (first malloc, then calloc, realloc and align)
+    // 120, 521, 1200 and 4099 bytes
+    void setupTest_singleOpManyIters()
+    {
+        m_testOperations = {
+            { new T(OperationName::Malloc, 100) },
+            { new T(OperationName::Calloc, 100) },
+            { new T(OperationName::Realloc, 100) },
+            { new T(OperationName::Align, 100) }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 120, 521, 1200, 4099 });
+        m_test->setExecutionMode(ExecutionMode::ManyIterations);
+    }
+
+    // 4 iterations of each thread, all with same set of operations (malloc, then calloc, realloc and align),
+    // but different operation probability
+    // 120, 521, 1200 and 4099 bytes
+    void setupTest_manyOpsManyIters()
+    {
+        m_testOperations = {
+            {
+                new T(OperationName::Malloc, 25),
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Malloc, 50),
+                new T(OperationName::Calloc, 70),
+                new T(OperationName::Realloc, 80),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Calloc, 50),
+                new T(OperationName::Malloc, 60),
+                new T(OperationName::Realloc, 75),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Realloc, 60),
+                new T(OperationName::Malloc, 80),
+                new T(OperationName::Calloc, 90),
+                new T(OperationName::Align, 100)
+            },
+            {
+                new T(OperationName::Realloc, 40),
+                new T(OperationName::Malloc, 55),
+                new T(OperationName::Calloc, 70),
+                new T(OperationName::Align, 100)
+            }
+        };
+
+        m_test = new performance_tests::PerformanceTest(m_repeats, m_threads,
+                                                        m_iterations * 10);
+        m_test->setOperations(m_testOperations, m_freeOperation);
+        m_test->setAllocationSizes({ 120, 521, 1200, 4099 });
+        m_test->setExecutionMode(ExecutionMode::ManyIterations);
+    }
+};
diff --git a/test/pmem_alloc_performance_tests.cpp b/test/pmem_alloc_performance_tests.cpp
new file mode 100644
index 0000000..f6ee32d
--- /dev/null
+++ b/test/pmem_alloc_performance_tests.cpp
@@ -0,0 +1,430 @@
+/*
+* Copyright (C) 2019 Intel Corporation.
+* All rights reserved.
+*
+* Redistribution and use in source and binary forms, with or without
+* modification, are permitted provided that the following conditions are met:
+* 1. Redistributions of source code must retain the above copyright notice(s),
+*    this list of conditions and the following disclaimer.
+* 2. Redistributions in binary form must reproduce the above copyright notice(s),
+*    this list of conditions and the following disclaimer in the documentation
+*    and/or other materials provided with the distribution.
+*
+* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+* OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+* EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+* OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+* ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+*/
+
+#include "common.h"
+#include "allocator_perf_tool/TaskFactory.hpp"
+#include "allocator_perf_tool/Stats.hpp"
+#include "allocator_perf_tool/Thread.hpp"
+#include "allocator_perf_tool/GTestAdapter.hpp"
+#include "allocator_perf_tool/PmemMockup.hpp"
+
+static const size_t PMEM_PART_SIZE = 0;
+extern const char  *PMEM_DIR;
+
+class PmemAllocPerformanceTest: public :: testing::Test
+{
+private:
+    AllocatorFactory allocator_factory;
+
+protected:
+    void SetUp()
+    {
+        allocator_factory.initialize_allocator(AllocatorTypes::STANDARD_ALLOCATOR);
+
+        int err = memkind_create_pmem(PMEM_DIR, PMEM_PART_SIZE, &MEMKIND_PMEM_MOCKUP);
+        ASSERT_EQ(0, err);
+        ASSERT_NE(nullptr, MEMKIND_PMEM_MOCKUP);
+    }
+
+    void TearDown()
+    {
+        int err = memkind_destroy_kind(MEMKIND_PMEM_MOCKUP);
+        ASSERT_EQ(0, err);
+    }
+
+    float run(unsigned kind, unsigned call, size_t threads_number,
+              size_t alloc_size, unsigned mem_operations_num)
+    {
+        TaskFactory task_factory;
+        std::vector<Thread *> threads;
+        std::vector<Task *> tasks;
+        TypesConf func_calls;
+        TypesConf allocator_types;
+
+        func_calls.enable_type(FunctionCalls::FREE);
+        func_calls.enable_type(call);
+        allocator_types.enable_type(kind);
+
+        TaskConf conf = {
+            mem_operations_num, //number of memory operations
+            {
+                mem_operations_num, //number of memory operations
+                alloc_size, //min. size of single allocation
+                alloc_size //max. size of single allocatioion
+            },
+            func_calls, //enable function calls
+            allocator_types, //enable allocators
+            11, //random seed
+            false, //do not log memory operations and statistics to csv file
+        };
+
+        for (int i=0; i<threads_number; i++) {
+            Task *task = task_factory.create(conf);
+            tasks.push_back(task);
+            threads.push_back(new Thread(task));
+            conf.seed += 1;
+        }
+
+        ThreadsManager threads_manager(threads);
+        threads_manager.start();
+        threads_manager.barrier();
+
+        TimeStats time_stats;
+        for (int i=0; i<tasks.size(); i++) {
+            time_stats += tasks[i]->get_results();
+        }
+
+        return time_stats.stats[kind][call].total_time;
+    }
+
+    void run_test(unsigned kind, unsigned call, size_t threads_number,
+                  size_t alloc_size, unsigned mem_operations_num)
+    {
+        allocator_factory.initialize_allocator(kind);
+        float ref_time = run(AllocatorTypes::STANDARD_ALLOCATOR, call, threads_number,
+                             alloc_size, mem_operations_num);
+        float perf_time = run(kind, call, threads_number, alloc_size,
+                              mem_operations_num);
+        float ref_delta_time_percent = allocator_factory.calc_ref_delta(ref_time,
+                                                                        perf_time);
+
+        GTestAdapter::RecordProperty("total_time_spend_on_alloc", perf_time);
+        GTestAdapter::RecordProperty("alloc_operations_per_thread", mem_operations_num);
+        GTestAdapter::RecordProperty("ref_delta_time_percent", ref_delta_time_percent);
+    }
+
+};
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 100, 10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_malloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::MALLOC, 72, 1572864,
+             1000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 100, 10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_calloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::CALLOC, 72, 1572864,
+             1000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_1_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 1, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_10_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 10, 1572864,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_100_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 100,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_4096_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 4096,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_1000_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 1000,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_1001_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 1001,
+             10000);
+}
+
+TEST_F(PmemAllocPerformanceTest,
+       test_TC_MEMKIND_MEMKIND_PMEM_realloc_72_thread_1572864_bytes)
+{
+    run_test(AllocatorTypes::MEMKIND_PMEM, FunctionCalls::REALLOC, 72, 1572864,
+             1000);
+}
diff --git a/test/pmem_allocator_tests.cpp b/test/pmem_allocator_tests.cpp
new file mode 100644
index 0000000..54a601e
--- /dev/null
+++ b/test/pmem_allocator_tests.cpp
@@ -0,0 +1,322 @@
+/*
+ * Copyright (C) 2018 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "memkind.h"
+#include "pmem_allocator.h"
+#include "common.h"
+#include <vector>
+#include <memory>
+#include <array>
+#include <list>
+#include <map>
+#include <utility>
+#include <string>
+#include <algorithm>
+#include <scoped_allocator>
+#include <thread>
+
+extern const char *PMEM_DIR;
+
+// Tests for pmem::allocator class.
+class PmemAllocatorTests: public ::testing::Test
+{
+public:
+    const size_t pmem_max_size = 1024*1024*1024;
+    static_assert(MEMKIND_MEM_USAGE_POLICY_MAX_VALUE == static_cast<int>
+                  (libmemkind::allocation_policy::MAX),
+                  "Allocation policy is not inline with memory usage policy");
+    libmemkind::pmem::allocator<int> alloc_source { std::string(PMEM_DIR), pmem_max_size };
+    libmemkind::pmem::allocator<int> alloc_conservative  { std::string(PMEM_DIR), pmem_max_size, libmemkind::allocation_policy::CONSERVATIVE };
+
+    libmemkind::pmem::allocator<int> alloc_source_f1 { std::string(PMEM_DIR), pmem_max_size };
+    libmemkind::pmem::allocator<int> alloc_source_f2 { std::string(PMEM_DIR), pmem_max_size };
+
+protected:
+    void SetUp()
+    {}
+
+    void TearDown()
+    {}
+};
+
+
+//Compare two same kind different type allocators
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindDifferentType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<char> alc2 { alloc_source_f1 };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+//Compare two same kind same type
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_SameKindSameType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f1 };
+    ASSERT_TRUE(alc1 == alc2);
+    ASSERT_FALSE(alc1 != alc2);
+}
+
+//Compare two different kind different type
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindDifferentType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<char> alc2 { alloc_source_f2 };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+//Compare two different kind same type allocators
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorCompare_DifferentKindSameType_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_FALSE(alc1 == alc2);
+    ASSERT_TRUE(alc1 != alc2);
+}
+
+//Copy assignment test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_CopyAssignment_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_TRUE(alc1 != alc2);
+    alc1 = alc2;
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+//Move constructor test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_MoveConstructor_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f2 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_TRUE(alc2 == alc1);
+    libmemkind::pmem::allocator<int> alc3 = std::move(alc1);
+    ASSERT_TRUE(alc2 == alc3);
+}
+
+//Move assignment test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_MoveAssignment_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f2 };
+    ASSERT_TRUE(alc1 != alc2);
+
+    {
+        libmemkind::pmem::allocator<int> alc3 { alc2 };
+        alc1 = std::move(alc3);
+    }
+    ASSERT_TRUE(alc1 == alc2);
+}
+
+//Single allocation-deallocation test
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_Allocator_SingleAllocationDeallocation_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    int *created_object = alc1.allocate(1);
+    alc1.deallocate(created_object, 1);
+}
+
+//Single allocation-deallocation test conservative policy
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_Allocator_SingleAllocationDeallocation_TestConservative)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_conservative };
+    int *created_object = alc1.allocate(1);
+    alc1.deallocate(created_object, 1);
+}
+
+//Shared cross-allocation-deallocation test
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_Allocator_SharedAllocationDeallocation_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    libmemkind::pmem::allocator<int> alc2 { alloc_source_f1 };
+    int *created_object = nullptr;
+    int *created_object_2 = nullptr;
+    created_object = alc1.allocate(1);
+    ASSERT_TRUE(alc1 == alc2);
+    alc2.deallocate(created_object, 1);
+    created_object = alc2.allocate(1);
+    alc1.deallocate(created_object, 1);
+
+    created_object = alc1.allocate(1);
+    created_object_2 = alc2.allocate(1);
+    alc2.deallocate(created_object, 1);
+    alc1.deallocate(created_object_2, 1);
+}
+
+//Construct-destroy test
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_ConstructDestroy_Test)
+{
+    libmemkind::pmem::allocator<int> alc1 { alloc_source_f1 };
+    int *created_object = alc1.allocate(1);
+    alc1.construct(created_object);
+    alc1.destroy(created_object);
+}
+
+//Testing thread-safety support
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_Allocator_MultithreadingSupport_Test)
+{
+    const size_t num_threads = 1000;
+    const size_t iteration_count = 5000;
+
+    std::vector<std::thread> thds;
+    for (size_t i = 0; i < num_threads; ++i) {
+
+        thds.push_back(std::thread( [&]() {
+            std::vector<libmemkind::pmem::allocator<int>> allocators_local;
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.push_back(libmemkind::pmem::allocator<int> ( alloc_source ));
+            }
+            for (size_t j = 0; j < iteration_count; j++) {
+                allocators_local.pop_back();
+            }
+        }));
+    }
+
+    for (size_t i = 0; i < num_threads; ++i) {
+        thds[i].join();
+    }
+}
+
+//Test multiple memkind allocator usage
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_MultipleAllocatorUsage_Test)
+{
+    {
+        libmemkind::pmem::allocator<int> alloc { PMEM_DIR, pmem_max_size } ;
+    }
+
+    {
+        libmemkind::pmem::allocator<int> alloc { PMEM_DIR, pmem_max_size } ;
+    }
+}
+
+//Test vector
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Vector_Test)
+{
+    libmemkind::pmem::allocator<int> alc{ alloc_source  };
+    std::vector<int, libmemkind::pmem::allocator<int>> vector{ alc };
+
+    const int num = 20;
+
+    for (int i = 0; i < num; ++i) {
+        vector.push_back(0xDEAD + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        ASSERT_TRUE(vector[i] == 0xDEAD + i);
+    }
+}
+
+//Test list
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_List_Test)
+{
+    libmemkind::pmem::allocator<int> alc{ alloc_source };
+
+    std::list<int, libmemkind::pmem::allocator<int>> list{ alc };
+
+    const int num = 4;
+
+    for (int i = 0; i < num; ++i) {
+        list.emplace_back(0xBEAC011 + i);
+        ASSERT_TRUE(list.back() == 0xBEAC011 + i);
+    }
+
+    for (int i = 0; i < num; ++i) {
+        list.pop_back();
+    }
+}
+
+//Test map
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_Map_Test)
+{
+    libmemkind::pmem::allocator<std::pair<const std::string, std::string>> alc{ alloc_source };
+
+    std::map<std::string, std::string, std::less<std::string>, libmemkind::pmem::allocator<std::pair<const std::string, std::string>>>
+    map{ std::less<std::string>(), alc };
+
+    const int num = 10;
+
+    for (int i = 0; i < num; ++i) {
+        map[std::to_string(i)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string(i)] == std::to_string(0x0CEA11 + i));
+        map[std::to_string((i * 997 + 83) % 113)] = std::to_string(0x0CEA11 + i);
+        ASSERT_TRUE(map[std::to_string((i * 997 + 83) % 113)] == std::to_string(
+                        0x0CEA11 + i));
+    }
+}
+
+#if _GLIBCXX_USE_CXX11_ABI
+//Test vector of strings
+TEST_F(PmemAllocatorTests, test_TC_MEMKIND_AllocatorUsage_VectorOfString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::pmem::allocator<char>>
+            pmem_string;
+    typedef libmemkind::pmem::allocator<pmem_string> pmem_alloc;
+
+    pmem_alloc alc{ alloc_source };
+    libmemkind::pmem::allocator<char> st_alc{alc};
+
+    std::vector<pmem_string,std::scoped_allocator_adaptor<pmem_alloc>> vec{ alc };
+    pmem_string arg{ "Very very loooong striiiing", st_alc };
+
+    vec.push_back(arg);
+    ASSERT_TRUE(vec.back() == arg);
+}
+
+//Test map of int strings
+TEST_F(PmemAllocatorTests,
+       test_TC_MEMKIND_AllocatorScopedUsage_MapOfIntString_Test)
+{
+    typedef std::basic_string<char, std::char_traits<char>, libmemkind::pmem::allocator<char>>
+            pmem_string;
+    typedef int key_t;
+    typedef pmem_string value_t;
+    typedef std::pair<const key_t, value_t> target_pair;
+    typedef libmemkind::pmem::allocator<target_pair> pmem_alloc;
+    typedef libmemkind::pmem::allocator<char> str_allocator_t;
+    typedef std::map<key_t, value_t, std::less<key_t>, std::scoped_allocator_adaptor<pmem_alloc>>
+            map_t;
+
+    pmem_alloc map_allocator( alloc_source );
+
+    str_allocator_t str_allocator( map_allocator );
+
+    value_t source_str1("Lorem ipsum dolor ", str_allocator);
+    value_t source_str2("sit amet consectetuer adipiscing elit", str_allocator );
+
+    map_t target_map{ std::scoped_allocator_adaptor<pmem_alloc>(map_allocator) };
+
+    target_map[key_t(165)] = source_str1;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str1);
+    target_map[key_t(165)] = source_str2;
+    ASSERT_TRUE(target_map[key_t(165)] == source_str2);
+}
+#endif // _GLIBCXX_USE_CXX11_ABI
diff --git a/test/proc_stat.h b/test/proc_stat.h
new file mode 100644
index 0000000..79f8530
--- /dev/null
+++ b/test/proc_stat.h
@@ -0,0 +1,74 @@
+/*
+ * Copyright (C) 2017 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include <cstring>
+#include <fstream>
+
+class ProcStat
+{
+public:
+    size_t get_virtual_memory_size_bytes()
+    {
+        get_stat("VmSize", str_value);
+        return strtol(str_value, NULL, 10) * 1024;
+    }
+
+    size_t get_physical_memory_size_bytes()
+    {
+        get_stat("VmRSS", str_value);
+        return strtol(str_value, NULL, 10) * 1024;
+    }
+
+    size_t get_used_swap_space_size_bytes(void)
+    {
+        get_stat("VmSwap", str_value);
+        return strtol(str_value, NULL, 10) * 1024;
+    }
+private:
+    /* We are avoiding to allocate local buffers,
+     * since it can produce noise in memory footprint tests.
+     */
+    char line[1024];
+    char current_entry_name[1024];
+    char str_value[1024];
+
+    // Note: this function is not thread-safe.
+    void get_stat(const char *field_name, char *value)
+    {
+        char *pos = nullptr;
+        std::ifstream file("/proc/self/status", std::ifstream::in);
+        if (file.is_open()) {
+            while (file.getline(line, sizeof(line))) {
+                pos = strstr(line, field_name);
+                if (pos) {
+                    sscanf(pos, "%64[a-zA-Z_0-9()]: %s", current_entry_name, value);
+                    break;
+                }
+            }
+            file.close();
+        }
+    }
+};
+
diff --git a/test/python_framework/__init__.py b/test/python_framework/__init__.py
new file mode 100644
index 0000000..bf66a99
--- /dev/null
+++ b/test/python_framework/__init__.py
@@ -0,0 +1,26 @@
+#
+#  Copyright (C) 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+from python_framework.cmd_helper import CMD_helper
+from python_framework.huge_page_organizer import Huge_page_organizer
diff --git a/test/python_framework/cmd_helper.py b/test/python_framework/cmd_helper.py
new file mode 100644
index 0000000..68a869d
--- /dev/null
+++ b/test/python_framework/cmd_helper.py
@@ -0,0 +1,55 @@
+#
+#  Copyright (C) 2017 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+import pytest
+import os
+import tempfile
+import subprocess
+
+class CMD_helper(object):
+
+    def execute_cmd(self, command, sudo=False):
+        if sudo:
+            command = "sudo {0}".format(command)
+        #Initialize temp file for stdout. Will be removed when closed.
+        outfile = tempfile.SpooledTemporaryFile()
+        try:
+            #Invoke process
+            p = subprocess.Popen(command, stdout=outfile, stderr=subprocess.STDOUT, shell=True)
+            p.communicate()
+            #Read stdout from file
+            outfile.seek(0)
+            stdout = outfile.read()
+        except:
+            raise
+        finally:
+            #Make sure the file is closed
+            outfile.close()
+        retcode = p.returncode
+        return stdout, retcode
+
+    def get_command_path(self, binary):
+        """Get the path to the binary."""
+        path = os.path.dirname(os.path.abspath(__file__))
+        return os.path.join(path, binary)
diff --git a/test/python_framework/huge_page_organizer.py b/test/python_framework/huge_page_organizer.py
new file mode 100644
index 0000000..6603551
--- /dev/null
+++ b/test/python_framework/huge_page_organizer.py
@@ -0,0 +1,74 @@
+#
+#  Copyright (C) 2017 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+from cmd_helper import CMD_helper
+import os
+import itertools
+
+"""
+Huge_page_organizer sets hugepages per NUMA node and restores initial setup of hugepages.
+It writes and reads from the same file, so using Huge_page_organizer with parallel
+execution may cause undefined behaviour.
+"""
+class Huge_page_organizer(object):
+
+    cmd_helper = CMD_helper()
+    path = "/sys/devices/system/node/node{0}/hugepages/hugepages-2048kB/nr_hugepages"
+    restore_values = []
+
+    def __restore(self):
+        """Restore initial setup of hugepages."""
+        for node_id, restore_value in enumerate(self.restore_values):
+            self.__set_nr_hugepages(node_id, restore_value)
+
+    def __get_nr_hugepages(self, node_id):
+        """Return number of hugepages on given node or None if given node is not configured."""
+        if not os.path.isfile(self.path.format(node_id)):
+            return None
+        with open(self.path.format(node_id), "r") as f:
+            return int(f.read())
+
+    def __set_nr_hugepages(self, node_id, nr_hugepages):
+        """Set hugepages on given node to given number. Return True on success, False otherwise."""
+        command = 'sh -c "echo {0} >> {1}"'.format(nr_hugepages, self.path.format(node_id))
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=True)
+        if retcode:
+            return False
+        return self.__get_nr_hugepages(node_id) is nr_hugepages
+
+    def __init__(self, nr_hugepages_per_node):
+        """Save current hugepages setup and set hugepages per NUMA node."""
+        for node_id in itertools.count():
+            nr_hugepages = self.__get_nr_hugepages(node_id)
+            if nr_hugepages is None:
+                break
+            self.restore_values.append(nr_hugepages)
+            retcode = self.__set_nr_hugepages(node_id, nr_hugepages_per_node)
+            if not retcode:
+               self.__restore()
+            assert retcode, "Error: Could not set the requested amount of hugepages."
+
+    def __del__(self):
+        """Call __restore() function."""
+        self.__restore()
diff --git a/test/random_sizes_allocator.h b/test/random_sizes_allocator.h
new file mode 100644
index 0000000..0c96fb9
--- /dev/null
+++ b/test/random_sizes_allocator.h
@@ -0,0 +1,76 @@
+/*
+ * Copyright (C) 2017 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+#include "memory_manager.h"
+#include <vector>
+#include <random>
+
+class RandomSizesAllocator
+{
+private:
+    std::vector<MemoryManager> allocated_memory;
+    memkind_t kind;
+    std::default_random_engine generator;
+    std::uniform_int_distribution<int> memory_distribution;
+
+    size_t get_random_size()
+    {
+        return memory_distribution(generator);
+    }
+
+public:
+    RandomSizesAllocator(memkind_t kind, size_t min_size, size_t max_size,
+                         int max_allocations_number) :
+        kind(kind),
+        memory_distribution(min_size, max_size)
+    {
+        allocated_memory.reserve(max_allocations_number);
+    }
+
+    size_t malloc_random_memory()
+    {
+        size_t size = get_random_size();
+        allocated_memory.emplace_back(kind, size);
+        return size;
+    }
+
+    size_t free_random_memory()
+    {
+        if (empty())
+            return 0;
+        std::uniform_int_distribution<int> distribution(0, allocated_memory.size() - 1);
+        int random_index = distribution(generator);
+        auto it = std::begin(allocated_memory) + random_index;
+        size_t size = it->size();
+        allocated_memory.erase(it);
+        return size;
+    }
+
+    bool empty()
+    {
+        return allocated_memory.empty();
+    }
+};
+
diff --git a/test/run_alloc_benchmark.sh b/test/run_alloc_benchmark.sh
new file mode 100644
index 0000000..ef85a9e
--- /dev/null
+++ b/test/run_alloc_benchmark.sh
@@ -0,0 +1,60 @@
+#!/bin/bash
+#
+#  Copyright (C) 2014 - 2018 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+# Allocator
+ALLOCATOR="glibc tbb hbw pmem"
+# Thread configuration
+THREADS=(1 2 4 8 16 18 36)
+# Memory configuration (in kB) / iterations
+MEMORY=(1 4 16 64 256 1024 4096 16384)
+# Iterations
+ITERS=1000
+
+export KMP_AFFINITY=scatter,granularity=fine
+
+# For each algorithm
+for alloc in $ALLOCATOR
+do
+    rm -f alloctest_$alloc.txt
+    echo "# Number of threads, allocation size [kB], average malloc and free time [ms], average allocation time [ms], \
+average free time [ms], first allocation time [ms], first free time [ms]" >> alloctest_$alloc.txt
+    # For each number of threads
+    for nthr in ${THREADS[*]}
+    do
+        # For each amount of memory
+        for mem in ${MEMORY[*]}
+        do
+            echo "OMP_NUM_THREADS=$nthr ./alloc_benchmark_$alloc $ITERS $mem >> alloctest_$alloc.txt"
+            OMP_NUM_THREADS=$nthr ./alloc_benchmark_$alloc $ITERS $mem >> alloctest_$alloc.txt
+            ret=$?
+            if [ $ret -ne 0 ]; then
+                echo "Error: alloc_benchmark_$alloc returned $ret"
+                exit
+            fi
+        done
+    done
+done
+
+echo "Data collected. You can draw performance plots using python draw_plots.py."
diff --git a/test/static_kinds_list.h b/test/static_kinds_list.h
new file mode 100644
index 0000000..f43b7e2
--- /dev/null
+++ b/test/static_kinds_list.h
@@ -0,0 +1,47 @@
+/*
+ * Copyright (C) 2016 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#pragma once
+
+#include <memkind.h>
+
+
+static memkind_t static_kinds_list[] = {
+    MEMKIND_DEFAULT,
+    MEMKIND_HBW,
+    MEMKIND_HBW_HUGETLB,
+    MEMKIND_HBW_PREFERRED,
+    MEMKIND_HBW_PREFERRED_HUGETLB,
+    MEMKIND_HUGETLB,
+    MEMKIND_HBW_GBTLB,
+    MEMKIND_HBW_PREFERRED_GBTLB,
+    MEMKIND_GBTLB,
+    MEMKIND_HBW_INTERLEAVE,
+    MEMKIND_INTERLEAVE,
+    MEMKIND_DAX_KMEM,
+    MEMKIND_DAX_KMEM_ALL,
+    MEMKIND_DAX_KMEM_PREFERRED
+};
+
+
diff --git a/test/static_kinds_tests.cpp b/test/static_kinds_tests.cpp
new file mode 100644
index 0000000..947aabd
--- /dev/null
+++ b/test/static_kinds_tests.cpp
@@ -0,0 +1,59 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "memkind.h"
+#include "memkind/internal/memkind_private.h"
+
+#include "common.h"
+#include "static_kinds_list.h"
+
+/*
+ * Set of tests for checking if static kinds meets non-trivial assumptions
+ */
+
+class StaticKindsTest: public :: testing::Test
+{
+public:
+
+protected:
+    void SetUp()
+    {
+    }
+};
+
+
+/*
+ * Assumption: all static kinds should implement init_once operation
+ * Reason:  init_once should perform memkind_register (and other initialization if needed)
+ *          we are also using that fact to optimize initialization on first use (in memkind_malloc etc.)
+ */
+TEST_F(StaticKindsTest, test_TC_MEMKIND_STATIC_KINDS_INIT_ONCE)
+{
+    for(size_t i=0; i<(sizeof(static_kinds_list)/sizeof(static_kinds_list[0]));
+        i++) {
+        ASSERT_TRUE(static_kinds_list[i]->ops->init_once != NULL) <<
+                                                                  static_kinds_list[i]->name << " does not implement init_once operation!";
+    }
+}
+
diff --git a/test/tbbmalloc.h b/test/tbbmalloc.h
new file mode 100644
index 0000000..f52530b
--- /dev/null
+++ b/test/tbbmalloc.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright (C) 2016 - 2020 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <stdio.h>
+#include <dlfcn.h>
+
+extern void *(*scalable_malloc)(size_t);
+extern void *(*scalable_realloc)(void *, size_t);
+extern void *(*scalable_calloc)(size_t, size_t);
+extern void  (*scalable_free)(void *);
+
+int load_tbbmalloc_symbols();
diff --git a/test/test.sh b/test/test.sh
new file mode 100755
index 0000000..4c8e052
--- /dev/null
+++ b/test/test.sh
@@ -0,0 +1,384 @@
+#!/bin/bash
+#
+#  Copyright (C) 2014 - 2020 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+basedir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+PROGNAME=`basename $0`
+
+# Default path (to installation dir of memkind-tests RPM)
+TEST_PATH="$basedir/"
+
+# Gtest binaries executed by Berta
+GTEST_BINARIES=(all_tests decorator_test allocator_perf_tool_tests gb_page_tests_bind_policy memkind_stat_test defrag_allocate)
+
+# Pytest files executed by Berta
+PYTEST_FILES=(hbw_detection_test.py autohbw_test.py trace_mechanism_test.py)
+
+red=`tput setaf 1`
+green=`tput setaf 2`
+yellow=`tput setaf 3`
+default=`tput sgr0`
+
+err=0
+
+function usage () {
+   cat <<EOF
+
+Usage: $PROGNAME [-c csv_file] [-l log_file] [-f test_filter] [-T tests_dir] [-d] [-m] [-p] [-e] [-x] [-s] [-h]
+
+OPTIONS
+    -c,
+        path to CSV file
+    -l,
+        path to LOG file
+    -f,
+        test filter
+    -T,
+        path to tests directory
+    -d,
+        skip high bandwidth memory nodes detection tests
+    -m,
+        skip tests that require 2MB pages configured on the machine
+    -p,
+        skip python tests
+    -e,
+        run only HBW performance operation tests
+    -x,
+        skip tests that are passed as value
+    -s,
+        run disabled test (e.g. pmem long time stress test)
+    -h,
+        parameter added to display script usage
+EOF
+}
+
+function emit() {
+    if [ "$LOG_FILE" != "" ]; then
+        echo "$@" 2>&1 | tee -a $LOG_FILE;
+    else
+        echo "$@"
+    fi
+}
+
+function normalize_path {
+    local PATH=$1
+    if [ ! -d $PATH ];
+    then
+        echo "Not a directory: '$PATH'"
+        usage
+        exit 1
+    fi
+    if [[ $PATH != /* ]]; then
+        PATH=`pwd`/$PATH
+    fi
+    echo $PATH
+}
+
+function show_skipped_tests()
+{
+    SKIP_PATTERN=$1
+    DEFAULT_IFS=$IFS
+
+    # Search for gtests that match given pattern
+    for i in ${!GTEST_BINARIES[*]}; do
+        GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+        for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+            if [[ $LINE == *. ]]; then
+                TEST_SUITE=$LINE;
+            else
+                if [[ "$TEST_SUITE$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    # Search for pytests that match given pattern
+    for i in ${!PYTEST_FILES[*]}; do
+        PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+        IFS=$'\n'
+        for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+            if [[ $LINE == *"<Class "* ]]; then
+                TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            elif [[ $LINE == *"<Function "* ]]; then
+                LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+                if [[ "$TEST_SUITE.$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE.$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    IFS=$DEFAULT_IFS
+    emit ""
+}
+
+function execute_gtest()
+{
+    ret_val=1
+    TESTCMD=$1
+    TEST=$2
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST""$SKIPPED_GTESTS""$RUN_DISABLED_GTEST")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*OK ].*'
+    PATFAILED='.*FAILED  ].*'
+        PATSKIPPED='.*PASSED  ] 0.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST,${green}PASSED${default}"
+        ret_val=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        RESULT="$TEST,${yellow}SKIPPED${default}"
+        ret_val=0
+    else
+        RESULT="$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret_val
+}
+
+function execute_pytest()
+{
+    ret=1
+    TESTCMD=$1
+    TEST_SUITE=$2
+    TEST=$3
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST_SUITE.$TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST$SKIPPED_PYTESTS")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*1 passed.*'
+    PATFAILED='.*1 failed.*'
+    PATSKIPPED='.*deselected.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST_SUITE.$TEST,${green}PASSED${default}"
+        ret=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST_SUITE.$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        return 0
+    else
+        RESULT="$TEST_SUITE.$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret
+}
+
+#Check support for numa nodes (at least two)
+function check_numa()
+{
+    numactl --hardware | grep "^node 1" > /dev/null
+    if [ $? -ne 0 ]; then
+        echo "ERROR: $0 requires a NUMA enabled system with more than one node."
+        exit 1
+    fi
+}
+
+#Check support for High Bandwidth Memory - simulate one if no one was found
+function check_hbw_nodes()
+{
+    if [ ! -f /usr/bin/memkind-hbw-nodes ]; then
+        if [ -x ./memkind-hbw-nodes ]; then
+            export PATH=$PATH:$PWD
+        else
+            echo "Cannot find 'memkind-hbw-nodes' in $PWD. Did you run 'make'?"
+            exit 1
+        fi
+    fi
+
+    ret=$(memkind-hbw-nodes)
+    if [[ $ret == "" ]]; then
+        export MEMKIND_HBW_NODES=1
+        TEST_PREFIX="numactl --membind=0 --cpunodebind=$MEMKIND_HBW_NODES %s"
+    fi
+}
+
+#Check automatic support for persistent memory NUMA node - simulate one if no one was found
+function check_auto_dax_kmem_nodes()
+{
+    if [ ! -f /usr/bin/memkind-auto-dax-kmem-nodes ]; then
+        if [ -x ./memkind-auto-dax-kmem-nodes ]; then
+            export PATH=$PATH:$PWD
+        else
+            echo "Cannot find 'memkind-auto-dax-kmem-nodes' in $PWD. Did you run 'make'?"
+            exit 1
+        fi
+    fi
+
+    ret=$(memkind-auto-dax-kmem-nodes)
+    if [[ $ret == "" ]]; then
+        export MEMKIND_DAX_KMEM_NODES=1
+    fi
+}
+
+#begin of main script
+
+check_numa
+
+check_hbw_nodes
+check_auto_dax_kmem_nodes
+
+OPTIND=1
+
+while getopts "T:c:f:l:hdemsx:p:" opt; do
+    case "$opt" in
+        T)
+            TEST_PATH=$OPTARG;
+            ;;
+        c)
+            CSV=$OPTARG;
+            ;;
+        f)
+            TEST_FILTER=$OPTARG;
+            ;;
+        l)
+            LOG_FILE=$OPTARG;
+            ;;
+        m)
+            echo "Skipping tests that require 2MB pages due to unsatisfactory system conditions"
+            if [[ "$SKIPPED_GTESTS" == "" ]]; then
+                SKIPPED_GTESTS=":-*test_TC_MEMKIND_2MBPages_*"
+            else
+                SKIPPED_GTESTS=$SKIPPED_GTESTS":*test_TC_MEMKIND_2MBPages_*"
+            fi
+            if [[ "$SKIPPED_PYTESTS" == "" ]]; then
+                SKIPPED_PYTESTS=" and not test_TC_MEMKIND_2MBPages_"
+            else
+                SKIPPED_PYTESTS=$SKIPPED_PYTESTS" and not test_TC_MEMKIND_2MBPages_"
+            fi
+            show_skipped_tests "test_TC_MEMKIND_2MBPages_"
+            ;;
+        d)
+            echo "Skipping tests that detect high bandwidth memory nodes due to unsatisfactory system conditions"
+            if [[ $SKIPPED_PYTESTS == "" ]]; then
+                SKIPPED_PYTESTS=" and not hbw_detection"
+            else
+                SKIPPED_PYTESTS=$SKIPPED_PYTESTS" and not hbw_detection"
+            fi
+            show_skipped_tests "test_TC_MEMKIND_hbw_detection"
+            ;;
+        e)
+            GTEST_BINARIES=(performance_test)
+            SKIPPED_PYTESTS=$SKIPPED_PYTESTS$PYTEST_FILES
+            break;
+            ;;
+        p)
+            SKIPPED_PYTESTS=$SKIPPED_PYTESTS$OPTARG
+            show_skipped_tests "$OPTARG"
+            break;
+            ;;
+        x)
+            echo "Skipping some tests on demand '$OPTARG'"
+            if [[ $SKIPPED_GTESTS == "" ]]; then
+                SKIPPED_GTESTS=":-"$OPTARG
+            else
+                SKIPPED_GTESTS=$SKIPPED_GTESTS":"$OPTARG
+            fi
+            show_skipped_tests "$OPTARG"
+            ;;
+        s)
+            echo "Run also disabled tests"
+            RUN_DISABLED_GTEST=" --gtest_also_run_disabled_tests"
+            ;;
+        h)
+            usage;
+            exit 0;
+            ;;
+    esac
+done
+
+TEST_PATH=`normalize_path "$TEST_PATH"`
+
+# Clear any remnants of previous execution(s)
+rm -rf $CSV
+rm -rf $LOG_FILE
+
+# Run tests written in gtest
+for i in ${!GTEST_BINARIES[*]}; do
+    GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+    emit
+    emit "### Processing gtest binary '$GTEST_BINARY_PATH' ###"
+    for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+        if [[ $LINE == *. ]]; then
+            TEST_SUITE=$LINE;
+        else
+            TEST_CMD="$GTEST_BINARY_PATH --gtest_filter=%s 2>&1"
+            execute_gtest "$TEST_CMD" "$TEST_SUITE$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+# Run tests written in pytest
+for i in ${!PYTEST_FILES[*]}; do
+    PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+    emit
+    emit "### Processing pytest file '$PTEST_BINARY_PATH' ###"
+    IFS=$'\n'
+    for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+        if [[ $LINE == *"<Class "* ]]; then
+            TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+        elif [[ $LINE == *"<Function "* ]]; then
+            LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            TEST_CMD="py.test $PTEST_BINARY_PATH -k='%s' 2>&1"
+            execute_pytest "$TEST_CMD" "$TEST_SUITE" "$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+exit $err
diff --git a/test/test_dax_kmem.sh b/test/test_dax_kmem.sh
new file mode 100755
index 0000000..a2e2455
--- /dev/null
+++ b/test/test_dax_kmem.sh
@@ -0,0 +1,287 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 - 2020 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+basedir="$( cd "$( dirname "${BASH_SOURCE[0]}" )" && pwd )"
+PROGNAME=`basename $0`
+
+# Default path (to installation dir of memkind-tests RPM)
+TEST_PATH="$basedir/"
+
+# Gtest binaries
+GTEST_BINARIES=(dax_kmem_test)
+
+# Pytest files
+PYTEST_FILES=(dax_kmem_env_var_test.py)
+
+red=`tput setaf 1`
+green=`tput setaf 2`
+yellow=`tput setaf 3`
+default=`tput sgr0`
+
+err=0
+
+function usage () {
+   cat <<EOF
+
+Usage: $PROGNAME [-h]
+
+OPTIONS
+    -h,
+        display script usage
+EOF
+}
+
+function emit() {
+    if [ "$LOG_FILE" != "" ]; then
+        echo "$@" 2>&1 | tee -a $LOG_FILE;
+    else
+        echo "$@"
+    fi
+}
+
+function normalize_path {
+    local PATH=$1
+    if [ ! -d $PATH ];
+    then
+        echo "Not a directory: '$PATH'"
+        usage
+        exit 1
+    fi
+    if [[ $PATH != /* ]]; then
+        PATH=`pwd`/$PATH
+    fi
+    echo $PATH
+}
+
+function show_skipped_tests()
+{
+    SKIP_PATTERN=$1
+    DEFAULT_IFS=$IFS
+
+    # Search for gtests that match given pattern
+    for i in ${!GTEST_BINARIES[*]}; do
+        GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+        for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+            if [[ $LINE == *. ]]; then
+                TEST_SUITE=$LINE;
+            else
+                if [[ "$TEST_SUITE$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    # Search for pytests that match given pattern
+    for i in ${!PYTEST_FILES[*]}; do
+        PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+        IFS=$'\n'
+        for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+            if [[ $LINE == *"<Class "* ]]; then
+                TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            elif [[ $LINE == *"<Function "* ]]; then
+                LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+                if [[ "$TEST_SUITE.$LINE" == *"$SKIP_PATTERN"* ]]; then
+                    emit "$TEST_SUITE.$LINE,${yellow}SKIPPED${default}"
+                fi
+            fi
+        done
+    done
+
+    IFS=$DEFAULT_IFS
+    emit ""
+}
+
+function execute_gtest()
+{
+    ret_val=1
+    TESTCMD=$1
+    TEST=$2
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST""$SKIPPED_GTESTS""$RUN_DISABLED_GTEST")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*OK ].*'
+    PATFAILED='.*FAILED  ].*'
+        PATSKIPPED='.*PASSED  ] 0.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST,${green}PASSED${default}"
+        ret_val=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        RESULT="$TEST,${yellow}SKIPPED${default}"
+        ret_val=0
+    else
+        RESULT="$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret_val
+}
+
+function execute_pytest()
+{
+    ret=1
+    TESTCMD=$1
+    TEST_SUITE=$2
+    TEST=$3
+    # Apply filter (if provided)
+    if [ "$TEST_FILTER" != "" ]; then
+        if [[ $TEST_SUITE.$TEST != $TEST_FILTER ]]; then
+            return
+        fi
+    fi
+    # Concatenate test command
+    TESTCMD=$(printf "$TESTCMD" "$TEST$SKIPPED_PYTESTS")
+    # And test prefix if applicable
+    if [ "$TEST_PREFIX" != "" ]; then
+        TESTCMD=$(printf "$TEST_PREFIX" "$TESTCMD")
+    fi
+    OUTPUT=`eval $TESTCMD`
+    PATOK='.*1 passed.*'
+    PATFAILED='.*1 failed.*'
+    PATSKIPPED='.*deselected.*'
+    if [[ $OUTPUT =~ $PATOK ]]; then
+        RESULT="$TEST_SUITE.$TEST,${green}PASSED${default}"
+        ret=0
+    elif [[ $OUTPUT =~ $PATFAILED ]]; then
+        RESULT="$TEST_SUITE.$TEST,${red}FAILED${default}"
+    elif [[ $OUTPUT =~ $PATSKIPPED ]]; then
+        return 0
+    else
+        RESULT="$TEST_SUITE.$TEST,${red}CRASH${default}"
+    fi
+    if [ "$CSV" != "" ]; then
+        emit "$OUTPUT"
+        echo $RESULT >> $CSV
+    else
+        echo $RESULT
+    fi
+    return $ret
+}
+
+#Check support for numa nodes (at least two)
+function check_numa()
+{
+    numactl --hardware | grep "^node 1" > /dev/null
+    if [ $? -ne 0 ]; then
+        echo "ERROR: $0 requires a NUMA enabled system with more than one node."
+        exit 1
+    fi
+}
+
+#Check automatic support for persistent memory NUMA node - simulate one if no one was found
+function check_auto_dax_kmem_nodes()
+{
+    if [ ! -f /usr/bin/memkind-auto-dax-kmem-nodes ]; then
+        if [ -x ./memkind-auto-dax-kmem-nodes ]; then
+            export PATH=$PATH:$PWD
+        else
+            echo "Cannot find 'memkind-auto-dax-kmem-nodes' in $PWD. Did you run 'make'?"
+            exit 1
+        fi
+    fi
+
+    ret=$(memkind-auto-dax-kmem-nodes)
+    emit "The binary memkind-auto-dax-kmem-nodes returned code $ret."
+    if [[ $ret == "" ]]; then
+        export MEMKIND_DAX_KMEM_NODES=1
+    fi
+}
+
+#begin of main script
+
+check_numa
+
+check_auto_dax_kmem_nodes
+
+OPTIND=1
+
+while getopts ":h" opt; do
+    case "$opt" in
+        h)
+            usage;
+            exit 0;
+            ;;
+    esac
+done
+
+TEST_PATH=`normalize_path "$TEST_PATH"`
+
+# Clear any remnants of previous execution(s)
+rm -rf $CSV
+rm -rf $LOG_FILE
+
+# Run tests written in gtest
+for i in ${!GTEST_BINARIES[*]}; do
+    GTEST_BINARY_PATH=$TEST_PATH${GTEST_BINARIES[$i]}
+    emit
+    emit "### Processing gtest binary '$GTEST_BINARY_PATH' ###"
+    for LINE in $($GTEST_BINARY_PATH --gtest_list_tests); do
+        if [[ $LINE == *. ]]; then
+            TEST_SUITE=$LINE;
+        else
+            TEST_CMD="$GTEST_BINARY_PATH --gtest_filter=%s 2>&1"
+            execute_gtest "$TEST_CMD" "$TEST_SUITE$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+# Run tests written in pytest
+for i in ${!PYTEST_FILES[*]}; do
+    PTEST_BINARY_PATH=$TEST_PATH${PYTEST_FILES[$i]}
+    emit
+    emit "### Processing pytest file '$PTEST_BINARY_PATH' ###"
+    IFS=$'\n'
+    for LINE in $(py.test $PTEST_BINARY_PATH --collect-only); do
+        if [[ $LINE == *"<Class "* ]]; then
+            TEST_SUITE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+        elif [[ $LINE == *"<Function "* ]]; then
+            LINE=$(sed "s/^.*'\(.*\)'.*$/\1/" <<< $LINE)
+            TEST_CMD="py.test $PTEST_BINARY_PATH -k='%s' 2>&1"
+            execute_pytest "$TEST_CMD" "$TEST_SUITE" "$LINE"
+            ret=$?
+            if [ $err -eq 0 ]; then err=$ret; fi
+        fi
+    done
+done
+
+exit $err
diff --git a/test/trace_mechanism_test.py b/test/trace_mechanism_test.py
new file mode 100755
index 0000000..dba7ffd
--- /dev/null
+++ b/test/trace_mechanism_test.py
@@ -0,0 +1,62 @@
+#
+#  Copyright (C) 2016 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+#
+
+import pytest
+import os
+import tempfile
+import subprocess
+from python_framework import CMD_helper
+from python_framework import Huge_page_organizer
+
+class Test_trace_mechanism(object):
+    binary = "../trace_mechanism_test_helper"
+    fail_msg = "Test failed with:\n {0}"
+    debug_env = "MEMKIND_DEBUG=1 "
+    cmd_helper = CMD_helper()
+
+    def test_TC_MEMKIND_logging_MEMKIND_HBW(self):
+        #This test executes trace_mechanism_test_helper and test if MEMKIND_INFO message occurs while calling MEMKIND_HBW
+        command = self.debug_env + self.cmd_helper.get_command_path(self.binary) + " MEMKIND_HBW"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: trace_mechanism_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert "MEMKIND_INFO: NUMA node" in output, self.fail_msg.format("\nError: trace mechanism in memkind doesn't show MEMKIND_INFO message \noutput: {0}").format(output)
+
+    def test_TC_MEMKIND_2MBPages_logging_MEMKIND_HUGETLB(self):
+        huge_page_organizer = Huge_page_organizer(8)
+        #This test executes trace_mechanism_test_helper and test if MEMKIND_INFO message occurs while calling MEMKIND_HUGETLB
+        command = self.debug_env + self.cmd_helper.get_command_path(self.binary) + " MEMKIND_HUGETLB"
+        print "Executing command: {0}".format(command)
+        output, retcode= self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: trace_mechanism_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert "MEMKIND_INFO: Number of" in output, self.fail_msg.format("\nError: trace mechanism in memkind doesn't show MEMKIND_INFO message \noutput: {0}").format(output)
+        assert "MEMKIND_INFO: Overcommit limit for" in output, self.fail_msg.format("\nError: trace mechanism in memkind doesn't show MEMKIND_INFO message \n output: {0}").format(output)
+
+    def test_TC_MEMKIND_logging_negative_MEMKIND_DEBUG_env(self):
+        #This test executes trace_mechanism_test_helper and test if setting MEMKIND_DEBUG to wrong value causes MEMKIND_WARNING message
+        command = "MEMKIND_DEBUG=-1 " + self.cmd_helper.get_command_path(self.binary) + " MEMKIND_HBW"
+        print "Executing command: {0}".format(command)
+        output, retcode = self.cmd_helper.execute_cmd(command, sudo=False)
+        assert retcode == 0, self.fail_msg.format("\nError: trace_mechanism_test_helper returned {0} \noutput: {1}".format(retcode,output))
+        assert "MEMKIND_WARNING: debug option" in output, self.fail_msg.format("\nError: setting wrong MEMKIND_DEBUG environment variable doesn't show MEMKIND_WARNING \noutput: {0})").format(output)
diff --git a/test/trace_mechanism_test_helper.c b/test/trace_mechanism_test_helper.c
new file mode 100644
index 0000000..bbbe9d8
--- /dev/null
+++ b/test/trace_mechanism_test_helper.c
@@ -0,0 +1,63 @@
+/*
+ * Copyright (C) 2016 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+#include "memkind.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+
+int main(int argc, char *argv[])
+{
+    int err = 0;
+    const size_t size = 1024 * 1024;
+    void *buf = NULL;
+
+    //It is expected that "MEMKIND_HBW" or "MEMKIND_HUGETLB" argument is passed
+    if (argc != 2) {
+        printf("Error: Wrong number of parameters\n");
+        err = -1;
+        return err;
+    }
+
+    if (strcmp(argv[1], "MEMKIND_HBW") == 0) {
+        buf = memkind_malloc(MEMKIND_HBW, size);
+        if (buf == NULL) {
+            printf("Allocation of MEMKIND_HBW failed\n");
+            err = -1;
+        }
+        memkind_free(MEMKIND_HBW, buf);
+    } else if (strcmp(argv[1], "MEMKIND_HUGETLB") == 0) {
+        buf = memkind_malloc(MEMKIND_HUGETLB, size);
+        if (buf == NULL) {
+            printf("Allocation of MEMKIND_HUGETLB failed\n");
+            err = -1;
+        }
+        memkind_free(MEMKIND_HUGETLB, buf);
+    } else {
+        printf("Error: unknown parameter\n");
+        err = -1;
+    }
+
+    return err;
+}
diff --git a/test/trial_generator.cpp b/test/trial_generator.cpp
new file mode 100644
index 0000000..0a4a349
--- /dev/null
+++ b/test/trial_generator.cpp
@@ -0,0 +1,325 @@
+/*
+ * Copyright (C) 2014 - 2019 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include <memkind/internal/memkind_hbw.h>
+#include "allocator_perf_tool/HugePageOrganizer.hpp"
+
+#include "trial_generator.h"
+#include "check.h"
+#include <vector>
+#include <numa.h>
+#include <numaif.h>
+
+trial_t TrialGenerator :: create_trial_tuple(alloc_api_t api,
+                                             size_t size,
+                                             size_t alignment,
+                                             int page_size,
+                                             memkind_t memkind,
+                                             int free_index)
+{
+    trial_t ltrial;
+    ltrial.api = api;
+    ltrial.size = size;
+    ltrial.alignment = alignment;
+    ltrial.page_size = page_size;
+    ltrial.memkind = memkind;
+    ltrial.free_index = free_index;
+    return ltrial;
+}
+
+
+void TrialGenerator :: generate_gb (alloc_api_t api, int number_of_gb_pages,
+                                    memkind_t memkind, alloc_api_t api_free, bool psize_strict, size_t align)
+{
+    std::vector<size_t> sizes_to_alloc;
+    //When API = HBW_MEMALIGN_PSIZE: psize is set to HBW_PAGESIZE_1GB_STRICT when allocation is a multiple of 1GB. Otherwise it is set to HBW_PAGESIZE_1GB.
+    for (int i=1; i <= number_of_gb_pages; i++) {
+        if (psize_strict || api!=HBW_MEMALIGN_PSIZE)
+            sizes_to_alloc.push_back(i*GB);
+        else
+            sizes_to_alloc.push_back(i*GB+1);
+    }
+    int k = 0;
+    trial_vec.clear();
+
+    for (int i = 0; i< (int)sizes_to_alloc.size(); i++) {
+        trial_vec.push_back(create_trial_tuple(api, sizes_to_alloc[i],
+                                               align, 2*MB,
+                                               memkind,
+                                               -1));
+        if (i > 0)
+            k++;
+        trial_vec.push_back(create_trial_tuple(api_free,0,0,0,
+                                               memkind,
+                                               k++));
+    }
+}
+
+int n_random(int i)
+{
+    return random() % i;
+}
+
+void TrialGenerator :: generate_size_2bytes_2KB_2MB(alloc_api_t api)
+{
+    size_t size[] = {2, 2*KB, 2*MB};
+
+    int k = 0;
+    trial_vec.clear();
+    for (unsigned int i = 0; i < (int)(sizeof(size)/sizeof(size[0]));
+         i++) {
+        trial_vec.push_back(
+            create_trial_tuple(
+                api,size[i],
+                32,
+                4096,
+                MEMKIND_HBW,
+                -1
+            )
+        );
+
+        if (i > 0) k++;
+        trial_vec.push_back(create_trial_tuple(HBW_FREE, 0, 0, 0,
+                                               MEMKIND_HBW, k));
+        k++;
+    }
+}
+
+void TrialGenerator :: print()
+{
+
+    std::vector<trial_t>:: iterator it;
+
+    std::cout <<"*********** Size: "<< trial_vec.size()
+              <<"********\n";
+    std::cout << "SIZE PSIZE ALIGN FREE KIND"<<std::endl;
+
+    for (it = trial_vec.begin();
+         it != trial_vec.end();
+         it++) {
+        std::cout << it->size <<" "
+                  << it->page_size <<" "
+                  << it->alignment <<" "
+                  << it->free_index <<" "
+                  << it->memkind <<" "
+                  <<std::endl;
+    }
+
+}
+
+
+void TrialGenerator :: run(int num_bandwidth, std::vector<int> &bandwidth)
+{
+
+    int num_trial = trial_vec.size();
+    int i, ret = 0;
+    void **ptr_vec = NULL;
+
+    ptr_vec = (void **) malloc (num_trial *
+                                sizeof (void *));
+    if (NULL == ptr_vec) {
+        fprintf (stderr, "Error in allocating ptr array\n");
+        exit(-1);
+    }
+
+    for (i = 0; i < num_trial; ++i) {
+        ptr_vec[i] = NULL;
+    }
+    for (i = 0; i < num_trial; ++i) {
+        switch(trial_vec[i].api) {
+            case HBW_FREE:
+                if (i == num_trial - 1 || trial_vec[i + 1].api != HBW_REALLOC) {
+                    hbw_free(ptr_vec[trial_vec[i].free_index]);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                    ptr_vec[i] = NULL;
+                } else {
+                    ptr_vec[i + 1] = hbw_realloc(ptr_vec[trial_vec[i].free_index],
+                                                 trial_vec[i + 1].size);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                }
+                break;
+            case MEMKIND_FREE:
+                if (i == num_trial - 1 || trial_vec[i + 1].api != MEMKIND_REALLOC) {
+                    memkind_free(trial_vec[i].memkind,
+                                 ptr_vec[trial_vec[i].free_index]);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                    ptr_vec[i] = NULL;
+                } else {
+                    ptr_vec[i + 1] = memkind_realloc(trial_vec[i].memkind,
+                                                     ptr_vec[trial_vec[i].free_index],
+                                                     trial_vec[i + 1].size);
+                    ptr_vec[trial_vec[i].free_index] = NULL;
+                }
+                break;
+            case HBW_MALLOC:
+                fprintf (stdout,"Allocating %zd bytes using hbw_malloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = hbw_malloc(trial_vec[i].size);
+                break;
+            case HBW_CALLOC:
+                fprintf (stdout,"Allocating %zd bytes using hbw_calloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = hbw_calloc(trial_vec[i].size, 1);
+                break;
+            case HBW_REALLOC:
+                fprintf (stdout,"Allocating %zd bytes using hbw_realloc\n",
+                         trial_vec[i].size);
+                fflush(stdout);
+                if (NULL == ptr_vec[i]) {
+                    ptr_vec[i] = hbw_realloc(NULL, trial_vec[i].size);
+                }
+                break;
+            case HBW_MEMALIGN:
+                fprintf (stdout,"Allocating %zd bytes using hbw_memalign\n",
+                         trial_vec[i].size);
+                ret =  hbw_posix_memalign(&ptr_vec[i],
+                                          trial_vec[i].alignment,
+                                          trial_vec[i].size);
+                break;
+            case HBW_MEMALIGN_PSIZE:
+                fprintf (stdout,"Allocating %zd bytes using hbw_memalign_psize\n",
+                         trial_vec[i].size);
+                hbw_pagesize_t psize;
+                if (trial_vec[i].page_size == 4096)
+                    psize = HBW_PAGESIZE_4KB;
+                else if (trial_vec[i].page_size == 2097152)
+                    psize = HBW_PAGESIZE_2MB;
+                else if (trial_vec[i].size %
+                         trial_vec[i].page_size > 0)
+                    psize = HBW_PAGESIZE_1GB;
+                else
+                    psize = HBW_PAGESIZE_1GB_STRICT;
+
+                ret = hbw_posix_memalign_psize(&ptr_vec[i],
+                                               trial_vec[i].alignment,
+                                               trial_vec[i].size,
+                                               psize);
+
+                break;
+            case MEMKIND_MALLOC:
+                fprintf (stdout,"Allocating %zd bytes using memkind_malloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = memkind_malloc(trial_vec[i].memkind,
+                                            trial_vec[i].size);
+                break;
+            case MEMKIND_CALLOC:
+                fprintf (stdout,"Allocating %zd bytes using memkind_calloc\n",
+                         trial_vec[i].size);
+                ptr_vec[i] = memkind_calloc(trial_vec[i].memkind,
+                                            trial_vec[i].size, 1);
+                break;
+            case MEMKIND_REALLOC:
+                fprintf (stdout,"Allocating %zd bytes using memkind_realloc\n",
+                         trial_vec[i].size);
+                if (NULL == ptr_vec[i]) {
+                    ptr_vec[i] = memkind_realloc(trial_vec[i].memkind,
+                                                 ptr_vec[i],
+                                                 trial_vec[i].size);
+                }
+                break;
+            case MEMKIND_POSIX_MEMALIGN:
+                fprintf (stdout,
+                         "Allocating %zd bytes using memkind_posix_memalign\n",
+                         trial_vec[i].size);
+
+                ret = memkind_posix_memalign(trial_vec[i].memkind,
+                                             &ptr_vec[i],
+                                             trial_vec[i].alignment,
+                                             trial_vec[i].size);
+                break;
+        }
+        if (trial_vec[i].api != HBW_FREE &&
+            trial_vec[i].api != MEMKIND_FREE &&
+            trial_vec[i].memkind != MEMKIND_DEFAULT) {
+            ASSERT_TRUE(ptr_vec[i] != NULL);
+            memset(ptr_vec[i], 0, trial_vec[i].size);
+            Check check(ptr_vec[i], trial_vec[i]);
+            if (trial_vec[i].api == HBW_CALLOC) {
+                EXPECT_EQ(0, check.check_zero());
+            }
+            if (trial_vec[i].api == HBW_MEMALIGN ||
+                trial_vec[i].api == HBW_MEMALIGN_PSIZE ||
+                trial_vec[i].api == MEMKIND_POSIX_MEMALIGN) {
+                EXPECT_EQ(0, check.check_align(trial_vec[i].alignment));
+                EXPECT_EQ(0, ret);
+            }
+            if (trial_vec[i].api == HBW_MEMALIGN_PSIZE ||
+                (trial_vec[i].api == MEMKIND_MALLOC &&
+                 (trial_vec[i].memkind == MEMKIND_HBW_HUGETLB ||
+                  trial_vec[i].memkind == MEMKIND_HBW_PREFERRED_HUGETLB))) {
+                EXPECT_EQ(0, check.check_page_size(trial_vec[i].page_size));
+            }
+        }
+    }
+    for (i = 0; i < num_trial; ++i) {
+        if (ptr_vec[i]) {
+            hbw_free(ptr_vec[i]);
+        }
+    }
+}
+
+void TGTest :: SetUp()
+{
+    size_t node;
+    char *hbw_nodes_env, *endptr;
+    tgen = std::unique_ptr<TrialGenerator>(new TrialGenerator());
+
+    hbw_nodes_env = secure_getenv("MEMKIND_HBW_NODES");
+    if (hbw_nodes_env) {
+        num_bandwidth = 128;
+        for (node = 0; node < num_bandwidth; node++) {
+            bandwidth.push_back(1);
+        }
+        node = strtol(hbw_nodes_env, &endptr, 10);
+        bandwidth.push_back(2);
+        while (*endptr == ':') {
+            hbw_nodes_env = endptr + 1;
+            node = strtol(hbw_nodes_env, &endptr, 10);
+            if (endptr != hbw_nodes_env && node >= 0 && node < num_bandwidth) {
+                bandwidth.push_back(2);
+            }
+        }
+    } else {
+        num_bandwidth = NUMA_NUM_NODES;
+        nodemask_t nodemask;
+        struct bitmask nodemask_bm = {NUMA_NUM_NODES, nodemask.n};
+        numa_bitmask_clearall(&nodemask_bm);
+
+        memkind_hbw_all_get_mbind_nodemask(NULL, nodemask.n, NUMA_NUM_NODES);
+
+        int i, nodes_num = numa_num_configured_nodes();
+        for (i=0; i<NUMA_NUM_NODES; i++) {
+            if (i >= nodes_num) {
+                bandwidth.push_back(0);
+            } else if (numa_bitmask_isbitset(&nodemask_bm, i)) {
+                bandwidth.push_back(2);
+            } else {
+                bandwidth.push_back(1);
+            }
+        }
+    }
+}
+
+void TGTest :: TearDown()
+{}
diff --git a/test/trial_generator.h b/test/trial_generator.h
new file mode 100644
index 0000000..fd05954
--- /dev/null
+++ b/test/trial_generator.h
@@ -0,0 +1,97 @@
+/*
+ * Copyright (C) 2014 - 2018 Intel Corporation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ * 1. Redistributions of source code must retain the above copyright notice(s),
+ *    this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright notice(s),
+ *    this list of conditions and the following disclaimer in the documentation
+ *    and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+ * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+ * EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+ * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef execute_trials_include_h
+#define execute_trials_include_h
+
+#include "hbwmalloc.h"
+#include "memkind.h"
+
+#include <vector>
+#include <stdlib.h>
+#include <vector>
+#include <memory>
+#include <algorithm>
+#include <fstream>
+#include <iostream>
+
+#include "common.h"
+
+
+typedef enum {
+    HBW_MALLOC,
+    HBW_CALLOC,
+    HBW_REALLOC,
+    HBW_MEMALIGN,
+    HBW_MEMALIGN_PSIZE,
+    HBW_FREE,
+    MEMKIND_MALLOC,
+    MEMKIND_CALLOC,
+    MEMKIND_REALLOC,
+    MEMKIND_POSIX_MEMALIGN,
+    MEMKIND_FREE
+} alloc_api_t;
+
+typedef struct {
+    alloc_api_t api;
+    size_t size;
+    size_t alignment;
+    size_t page_size;
+    memkind_t memkind;
+    int free_index;
+} trial_t;
+
+class TrialGenerator
+{
+public:
+    TrialGenerator() {}
+    void generate_gb(alloc_api_t api, int number_of_gb_pages, memkind_t memkind,
+                     alloc_api_t api_free, bool psize_strict=false, size_t align = GB);
+    void run(int num_bandwidth, std::vector<int> &bandwidths);
+    void generate_size_2bytes_2KB_2MB(alloc_api_t api);
+    /*For debugging purposes*/
+    void print();
+private:
+    std::vector<trial_t> trial_vec;
+    trial_t create_trial_tuple(alloc_api_t api,
+                               size_t size,
+                               size_t alignment,
+                               int page_size,
+                               memkind_t memkind,
+                               int free_index);
+
+
+};
+
+class TGTest : public::testing::Test
+{
+protected:
+    size_t num_bandwidth;
+    std::vector<int> bandwidth;
+    std::unique_ptr<TrialGenerator> tgen;
+    void SetUp();
+    void TearDown();
+};
+
+#endif
diff --git a/utils/docker/Dockerfile.fedora-31 b/utils/docker/Dockerfile.fedora-31
new file mode 100644
index 0000000..e7d97dd
--- /dev/null
+++ b/utils/docker/Dockerfile.fedora-31
@@ -0,0 +1,75 @@
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Pull base image
+FROM fedora:31
+
+LABEL maintainer="katarzyna.wasiuta@intel.com"
+
+# Update the dnf cache and install basic tools
+RUN dnf update -y && dnf install -y \
+    asciidoctor \
+    automake \
+    bash-completion \
+    ca-certificates \
+    daxctl-devel \
+    devscripts \
+    expect \
+    g++ \
+    git \
+    json-c-devel \
+    keyutils-libs-devel \
+    kmod-devel \
+    libtool \
+    libudev-devel \
+    libuuid-devel \
+    numactl \
+    numactl-devel \
+    pkgconfig \
+    python-pip \
+    rpmdevtools \
+    sudo \
+    systemd \
+    which \
+    whois \
+ && dnf clean all
+
+# Install packages required by python tests
+RUN pip install wheel
+RUN pip install setuptools==3.4.1
+RUN pip install pytest==3.9.2
+
+# Add user
+ENV USER memkinduser
+ENV USERPASS memkindpass
+RUN useradd -m $USER -p `mkpasswd $USERPASS`
+RUN gpasswd wheel -a $USER
+RUN echo '%wheel ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers
+
+# Create directory for memkind repository
+WORKDIR /home/$USER/memkind
+
+# Allow user to create files in the home directory
+RUN chown -R $USER:wheel /home/$USER
+
+# Change user to $USER
+USER $USER
diff --git a/utils/docker/Dockerfile.ubuntu-18.04 b/utils/docker/Dockerfile.ubuntu-18.04
new file mode 100644
index 0000000..42cf238
--- /dev/null
+++ b/utils/docker/Dockerfile.ubuntu-18.04
@@ -0,0 +1,71 @@
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# Pull base image
+FROM ubuntu:18.04
+
+LABEL maintainer="katarzyna.wasiuta@intel.com"
+
+# Update the Apt cache and install basic tools
+RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y --no-install-recommends \
+    asciidoctor \
+    automake \
+    bash-completion \
+    ca-certificates \
+    curl \
+    devscripts \
+    g++ \
+    git \
+    libjson-c-dev \
+    libkeyutils-dev \
+    libkmod-dev \
+    libnuma-dev \
+    libtool \
+    libudev-dev \
+    numactl \
+    pkg-config \
+    python-pip \
+    sudo \
+    systemd \
+    uuid-dev \
+    whois \
+ && rm -rf /var/lib/apt/lists/*
+
+# Install packages required by python tests
+RUN pip install wheel
+RUN pip install setuptools==3.4.1
+RUN pip install pytest==3.9.2
+
+# Add user
+ENV USER memkinduser
+ENV USERPASS memkindpass
+RUN useradd -m $USER -g sudo -p `mkpasswd $USERPASS`
+RUN echo '%sudo ALL=(ALL) NOPASSWD:ALL' >> /etc/sudoers
+
+# Create directory for memkind repository
+WORKDIR /home/$USER/memkind
+
+# Allow user to create files in the home directory
+RUN chown -R $USER:sudo /home/$USER
+
+# Change user to $USER
+USER $USER
diff --git a/utils/docker/README.md b/utils/docker/README.md
new file mode 100644
index 0000000..32d285e
--- /dev/null
+++ b/utils/docker/README.md
@@ -0,0 +1,51 @@
+# **README**
+
+This is utils/docker/README.
+
+Scripts in this directory let run a Docker container with Ubuntu 18.04 and Fedora 31 environment
+to build, test and optionally measure test coverage of any pull request to memkind project, inside it.
+
+# Environment variables
+
+* **MEMKIND_HOST_WORKDIR** - Defines the absolute path to memkind project root directory on host environment.
+    **This variable must be defined.**
+
+* **CODECOV_TOKEN** - Codecov token for memkind repository to upload the coverage results.
+
+* **TEST_SUITE_NAME** - Name of test suite (possible values are HBW/PMEM/DAX_KMEM).
+
+* **PMEM_HOST_PATH** - PMEM mount device path on host environment (useful for testing PMEM configuration). Default value of **PMEM_HOST_PATH** is
+*'/tmp/'*.
+
+* **NDCTL_LIBRARY_VERSION** - ndctl library version.
+For fully testing MEMKIND_DAX_KMEM, ndctl library version tag must be passed as parameter,
+see https://github.com/pmem/ndctl/tags.
+
+* **TBB_LIBRARY_VERSION** - Intel Threading Building Blocks library version.
+For testing Threading Building Blocks, TBB library version tag must be passed as parameter,
+see https://github.com/01org/tbb/tags.
+
+# Files
+*'docker_run_build.sh'*  is used to build of memkind.
+
+*'docker_run_coverage.sh'*  is used for uploading coverage report on [Codecov.io](Codecov.io)
+
+*'docker_run_test.sh'*  is used to run tests of memkind.
+
+*'docker_install_ndctl.sh'*  is used to install ndctl library.
+
+*'docker_install_tbb.sh'*  is used to install Intel Threading Building Blocks library.
+
+*'set_host_configuration.sh'*  is used to set hugepages configuration on host machine.
+
+# Building and running the container
+
+One can use the script *run_local.sh* just to build and run tests, e.g.:
+
+```
+$ ./run_local.sh Dockerfile.ubuntu-18.04
+```
+
+**Note:**
+
+Others class group must have write permission to MEMKIND_HOST_WORKDIR to use the run_local script.
diff --git a/utils/docker/docker_install_ndctl.sh b/utils/docker/docker_install_ndctl.sh
new file mode 100755
index 0000000..9b8c04a
--- /dev/null
+++ b/utils/docker/docker_install_ndctl.sh
@@ -0,0 +1,63 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# docker_install_ndctl.sh - is called inside a Docker container;
+# installs ndctl library
+#
+
+set -e
+
+NDCTL_RELEASE_URL=https://github.com/pmem/ndctl/releases/tag/"$NDCTL_LIBRARY_VERSION"
+NDCTL_TAR_GZ="$NDCTL_LIBRARY_VERSION".tar.gz
+NDCTL_TARBALL_URL=https://github.com/pmem/ndctl/archive/"$NDCTL_TAR_GZ"
+
+# check if specified ndctl release exists
+if curl --output /dev/null --silent --head --fail "$NDCTL_RELEASE_URL"; then
+  echo "ndctl version ${NDCTL_LIBRARY_VERSION} exist."
+else
+  echo "ndctl url: ${NDCTL_RELEASE_URL} is not valid. ndctl version: ${NDCTL_LIBRARY_VERSION} doesn't exist."
+  exit 1
+fi
+
+NDCTL_LOCAL_DIR="$HOME"/ndctl/"$NDCTL_LIBRARY_VERSION"
+NDCTL_LOCAL_TAR_GZ="$NDCTL_LOCAL_DIR"/"$NDCTL_TAR_GZ"
+
+# create ndctl directory in home directory
+mkdir -p "$NDCTL_LOCAL_DIR"
+
+# download and untar ndctl library to ndctl directory
+curl -L "$NDCTL_TARBALL_URL" -o "$NDCTL_LOCAL_TAR_GZ"
+tar -xzf "$NDCTL_LOCAL_TAR_GZ" -C "$NDCTL_LOCAL_DIR" --strip-components=1
+
+# go to ndctl directory and build ndctl library
+cd "$NDCTL_LOCAL_DIR"
+./autogen.sh
+./configure --prefix=/usr --sysconfdir=/etc --libdir=/usr/lib
+make -j "$(nproc --all)"
+sudo make -j "$(nproc --all)" install
+
+# update shared library cache
+sudo ldconfig
+# return to previous directory
+cd -
diff --git a/utils/docker/docker_install_tbb.sh b/utils/docker/docker_install_tbb.sh
new file mode 100755
index 0000000..eb2b0ef
--- /dev/null
+++ b/utils/docker/docker_install_tbb.sh
@@ -0,0 +1,59 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# docker_install_tbb.sh - is called inside a Docker container;
+# installs Threading Building Block library
+#
+
+set -e
+
+TBB_RELEASE_URL=https://github.com/01org/tbb/releases/tag/"$TBB_LIBRARY_VERSION"
+TBB_TAR_GZ="$TBB_LIBRARY_VERSION".tar.gz
+TBB_TARBALL_URL=https://github.com/01org/tbb/archive/"$TBB_TAR_GZ"
+TBB_VARS_SH="tbbvars.sh"
+
+# check if specified TBB release exists
+if curl --output /dev/null --silent --head --fail "$TBB_RELEASE_URL"; then
+  echo "TBB version ${TBB_LIBRARY_VERSION} exist."
+else
+  echo "TBB url: ${TBB_RELEASE_URL} is not valid. TBB version: ${TBB_LIBRARY_VERSION} doesn't exist."
+  exit 1
+fi
+
+TBB_LOCAL_DIR="$HOME"/tbb/"$TBB_LIBRARY_VERSION"
+TBB_LOCAL_TAR_GZ="$TBB_LOCAL_DIR"/"$TBB_TAR_GZ"
+
+# create TBB directory in home directory
+mkdir -p "$TBB_LOCAL_DIR"
+
+# download and untar TBB library to TBB directory
+curl -L "$TBB_TARBALL_URL" -o "$TBB_LOCAL_TAR_GZ"
+tar -xzf "$TBB_LOCAL_TAR_GZ" -C "$TBB_LOCAL_DIR" --strip-components=1
+
+# build TBB library
+make -j "$(nproc --all)" --directory="$TBB_LOCAL_DIR"
+TBB_RELEASE_DIR=$(ls -d "$TBB_LOCAL_DIR"/build/*release)
+
+# set environment variables regarding TBB package
+source "$TBB_RELEASE_DIR"/"$TBB_VARS_SH"
diff --git a/utils/docker/docker_run_build.sh b/utils/docker/docker_run_build.sh
new file mode 100755
index 0000000..a8ed07b
--- /dev/null
+++ b/utils/docker/docker_run_build.sh
@@ -0,0 +1,64 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# docker_run_build.sh - is called inside a Docker container;
+# prepares and runs memkind build for specified pull request number
+#
+set -e
+
+export UTILS_PREFIX=utils/docker
+
+if [ -n "$CODECOV_TOKEN" ]; then
+    GCOV_OPTION="--enable-gcov"
+fi
+
+# if ndctl library version is specified install library
+if [ -n "$NDCTL_LIBRARY_VERSION" ]; then
+    "$UTILS_PREFIX"/docker_install_ndctl.sh
+fi
+
+# building memkind sources and tests
+./autogen.sh
+./configure --prefix=/usr $GCOV_OPTION
+make -j "$(nproc --all)"
+make -j "$(nproc --all)" checkprogs
+
+# building RPM package
+if [[ $(cat /etc/os-release) = *"Fedora"* ]]; then
+    make -j "$(nproc --all)" rpm
+fi
+
+# installing memkind
+sudo make install
+
+# if TBB library version is specified install library and use it
+# as MEMKIND_HEAP_MANAGER
+if [ -n "$TBB_LIBRARY_VERSION" ]; then
+    source "$UTILS_PREFIX"/docker_install_tbb.sh
+    HEAP_MANAGER="TBB"
+fi
+if [ -n "$TEST_SUITE_NAME" ]; then
+    "$UTILS_PREFIX"/docker_run_test.sh "$HEAP_MANAGER"
+fi
diff --git a/utils/docker/docker_run_coverage.sh b/utils/docker/docker_run_coverage.sh
new file mode 100755
index 0000000..c51452d
--- /dev/null
+++ b/utils/docker/docker_run_coverage.sh
@@ -0,0 +1,35 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+# docker_run_coverage.sh - is called inside a Docker container;
+# calculates coverage and creates online report
+#
+# Parameters:
+# -memkind repository directory
+# Options used:
+# -p dir       Project root directory
+# -Z           Exit with 1 if not successful. Default will Exit with 0
+
+MEMKIND_REPO_PATH=$1
+bash <(curl -s https://codecov.io/bash) -t "$CODECOV_TOKEN" -cF "$TEST_SUITE_NAME" -p "$MEMKIND_REPO_PATH" -Z
diff --git a/utils/docker/docker_run_test.sh b/utils/docker/docker_run_test.sh
new file mode 100755
index 0000000..4d1da2e
--- /dev/null
+++ b/utils/docker/docker_run_test.sh
@@ -0,0 +1,51 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# docker_run_test.sh - is called inside a Docker container;
+# runs specified memkind tests
+#
+# Parameters:
+# -heap manager
+set -e
+
+HEAP_MANAGER=$1
+
+if [ "$TEST_SUITE_NAME" = "HBW" ]; then
+    # running tests and display output in case of failure
+    make check || { cat test-suite.log; exit 1; }
+elif [ "$TEST_SUITE_NAME" = "PMEM" ]; then
+    make PMEM_PATH="$PMEM_CONTAINER_PATH" unit_tests_pmem
+    # running pmem examples
+    find examples/.libs -name "pmem*" -executable -type f -exec sh -c "MEMKIND_HEAP_MANAGER=$HEAP_MANAGER "{}" $PMEM_CONTAINER_PATH" \;
+elif [ "$TEST_SUITE_NAME" = "DAX_KMEM" ]; then
+    make unit_tests_dax_kmem
+else
+    echo "Unknown Test suite ${TEST_SUITE_NAME}"
+fi;
+
+# executing coverage script if codecov token is set
+if [ -n "$CODECOV_TOKEN" ]; then
+    "$UTILS_PREFIX"/docker_run_coverage.sh "$PWD"
+fi
diff --git a/utils/docker/run_local.sh b/utils/docker/run_local.sh
new file mode 100755
index 0000000..40d04a2
--- /dev/null
+++ b/utils/docker/run_local.sh
@@ -0,0 +1,68 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# run_local.sh - builds a container with Docker image
+# and runs docker_run_build.sh script inside it
+#
+# Parameters:
+# -name of the Docker image file
+
+set -e
+
+DOCKER_IMAGE_NAME="$1"
+export PMEM_HOST_PATH=${PMEM_HOST_PATH:-/tmp/}
+
+if [[ ! -f "$DOCKER_IMAGE_NAME" ]]; then
+    echo "Docker image "$DOCKER_IMAGE_NAME" does not exist."
+    exit 1
+fi
+
+if [[ -z "$MEMKIND_HOST_WORKDIR" ]]; then
+    echo "MEMKIND_HOST_WORKDIR has to be set."
+    exit 1
+fi
+
+# need to be inline with Dockerfile WORKDIR
+MEMKIND_CONTAINER_WORKDIR=/home/memkinduser/memkind/
+PMEM_CONTAINER_PATH=/home/memkinduser/mnt_pmem/
+
+docker build --tag memkind_cont \
+             --file "$DOCKER_IMAGE_NAME" \
+             --build-arg http_proxy=$http_proxy \
+             --build-arg https_proxy=$https_proxy \
+             .
+docker run --rm \
+           --privileged=true \
+           --tty=true \
+           --env http_proxy=$http_proxy \
+           --env https_proxy=$https_proxy \
+           --env CODECOV_TOKEN="$CODECOV_TOKEN" \
+           --env TEST_SUITE_NAME="$TEST_SUITE_NAME" \
+           --env TBB_LIBRARY_VERSION="$TBB_LIBRARY_VERSION" \
+           --env NDCTL_LIBRARY_VERSION="$NDCTL_LIBRARY_VERSION" \
+           --env PMEM_CONTAINER_PATH="$PMEM_CONTAINER_PATH" \
+           --mount type=bind,source="$MEMKIND_HOST_WORKDIR",target="$MEMKIND_CONTAINER_WORKDIR" \
+           --mount type=bind,source="$PMEM_HOST_PATH",target="$PMEM_CONTAINER_PATH" \
+           memkind_cont utils/docker/docker_run_build.sh
diff --git a/utils/docker/set_host_configuration.sh b/utils/docker/set_host_configuration.sh
new file mode 100755
index 0000000..3cc027b
--- /dev/null
+++ b/utils/docker/set_host_configuration.sh
@@ -0,0 +1,44 @@
+#!/bin/bash
+#
+#  Copyright (C) 2019 Intel Corporation.
+#  All rights reserved.
+#
+#  Redistribution and use in source and binary forms, with or without
+#  modification, are permitted provided that the following conditions are met:
+#  1. Redistributions of source code must retain the above copyright notice(s),
+#     this list of conditions and the following disclaimer.
+#  2. Redistributions in binary form must reproduce the above copyright notice(s),
+#     this list of conditions and the following disclaimer in the documentation
+#     and/or other materials provided with the distribution.
+#
+#  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+#  OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+#  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+#  EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+#  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+#  LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+#  PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+#  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+#  OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+#  ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+#
+# set_host_configuration.sh - set Huge Pages parameters required for memkind tests
+set -e
+
+MEMKIND_HUGE_PAGES_NO=3000
+MEMKIND_OVERCOMMIT_HUGEPAGES_NO=128
+
+# find out current configuration
+MEMKIND_HUGE_PAGES_FOUND=$(cat /proc/sys/vm/nr_hugepages)
+MEMKIND_OVERCOMMIT_HUGEPAGES_FOUND=$(cat /proc/sys/vm/nr_overcommit_hugepages)
+
+# set expected configuration
+if [ "$MEMKIND_HUGE_PAGES_FOUND" != "$MEMKIND_HUGE_PAGES_NO" ]; then
+    echo "Setting number of hugepages to ${MEMKIND_HUGE_PAGES_NO}."
+    sudo sysctl vm.nr_hugepages="$MEMKIND_HUGE_PAGES_NO"
+fi
+if [ "$MEMKIND_OVERCOMMIT_HUGEPAGES_FOUND" != "$MEMKIND_OVERCOMMIT_HUGEPAGES_NO" ]; then
+    echo "Setting number of overcommit hugepages to ${MEMKIND_OVERCOMMIT_HUGEPAGES_NO}."
+    sudo sysctl vm.nr_overcommit_hugepages="$MEMKIND_OVERCOMMIT_HUGEPAGES_NO"
+fi